{
  "best_metric": null,
  "best_model_checkpoint": null,
  "epoch": 0.9999371266897202,
  "eval_steps": 500,
  "global_step": 5964,
  "is_hyper_param_search": false,
  "is_local_process_zero": true,
  "is_world_process_zero": true,
  "log_history": [
    {
      "epoch": 0.00016766216074609662,
      "grad_norm": 6.196194648742676,
      "learning_rate": 1.11731843575419e-07,
      "loss": 3.11,
      "step": 1
    },
    {
      "epoch": 0.00033532432149219324,
      "grad_norm": 6.196194648742676,
      "learning_rate": 1.11731843575419e-07,
      "loss": 3.0615,
      "step": 2
    },
    {
      "epoch": 0.0005029864822382899,
      "grad_norm": 6.196194648742676,
      "learning_rate": 1.11731843575419e-07,
      "loss": 2.9922,
      "step": 3
    },
    {
      "epoch": 0.0006706486429843865,
      "grad_norm": 6.196194648742676,
      "learning_rate": 1.11731843575419e-07,
      "loss": 2.9874,
      "step": 4
    },
    {
      "epoch": 0.0008383108037304831,
      "grad_norm": 5.285811901092529,
      "learning_rate": 2.23463687150838e-07,
      "loss": 2.7919,
      "step": 5
    },
    {
      "epoch": 0.0010059729644765797,
      "grad_norm": 5.377310752868652,
      "learning_rate": 3.3519553072625703e-07,
      "loss": 2.9085,
      "step": 6
    },
    {
      "epoch": 0.0011736351252226763,
      "grad_norm": 4.72886848449707,
      "learning_rate": 4.46927374301676e-07,
      "loss": 2.6002,
      "step": 7
    },
    {
      "epoch": 0.001341297285968773,
      "grad_norm": 4.93556022644043,
      "learning_rate": 5.58659217877095e-07,
      "loss": 2.4875,
      "step": 8
    },
    {
      "epoch": 0.0015089594467148696,
      "grad_norm": 5.479576587677002,
      "learning_rate": 6.703910614525141e-07,
      "loss": 2.8652,
      "step": 9
    },
    {
      "epoch": 0.0016766216074609662,
      "grad_norm": 5.086491584777832,
      "learning_rate": 7.82122905027933e-07,
      "loss": 2.6703,
      "step": 10
    },
    {
      "epoch": 0.0018442837682070628,
      "grad_norm": 5.002711296081543,
      "learning_rate": 8.93854748603352e-07,
      "loss": 2.6001,
      "step": 11
    },
    {
      "epoch": 0.0020119459289531594,
      "grad_norm": 6.0489115715026855,
      "learning_rate": 1.005586592178771e-06,
      "loss": 2.7088,
      "step": 12
    },
    {
      "epoch": 0.002179608089699256,
      "grad_norm": 6.025905132293701,
      "learning_rate": 1.11731843575419e-06,
      "loss": 2.4601,
      "step": 13
    },
    {
      "epoch": 0.0023472702504453526,
      "grad_norm": 5.1472296714782715,
      "learning_rate": 1.229050279329609e-06,
      "loss": 2.1754,
      "step": 14
    },
    {
      "epoch": 0.002514932411191449,
      "grad_norm": 4.403075695037842,
      "learning_rate": 1.3407821229050281e-06,
      "loss": 1.8755,
      "step": 15
    },
    {
      "epoch": 0.002682594571937546,
      "grad_norm": 3.35719633102417,
      "learning_rate": 1.4525139664804472e-06,
      "loss": 1.7016,
      "step": 16
    },
    {
      "epoch": 0.0028502567326836423,
      "grad_norm": 4.584776878356934,
      "learning_rate": 1.564245810055866e-06,
      "loss": 1.7616,
      "step": 17
    },
    {
      "epoch": 0.003017918893429739,
      "grad_norm": 2.5632901191711426,
      "learning_rate": 1.675977653631285e-06,
      "loss": 1.6525,
      "step": 18
    },
    {
      "epoch": 0.0031855810541758355,
      "grad_norm": 2.6150174140930176,
      "learning_rate": 1.787709497206704e-06,
      "loss": 1.5917,
      "step": 19
    },
    {
      "epoch": 0.0033532432149219324,
      "grad_norm": 2.1641123294830322,
      "learning_rate": 1.899441340782123e-06,
      "loss": 1.4867,
      "step": 20
    },
    {
      "epoch": 0.0035209053756680288,
      "grad_norm": 2.4881510734558105,
      "learning_rate": 2.011173184357542e-06,
      "loss": 1.627,
      "step": 21
    },
    {
      "epoch": 0.0036885675364141256,
      "grad_norm": 1.2478208541870117,
      "learning_rate": 2.1229050279329612e-06,
      "loss": 1.3723,
      "step": 22
    },
    {
      "epoch": 0.003856229697160222,
      "grad_norm": 1.4412287473678589,
      "learning_rate": 2.23463687150838e-06,
      "loss": 1.4862,
      "step": 23
    },
    {
      "epoch": 0.004023891857906319,
      "grad_norm": 1.1230992078781128,
      "learning_rate": 2.3463687150837993e-06,
      "loss": 1.3752,
      "step": 24
    },
    {
      "epoch": 0.004191554018652416,
      "grad_norm": 1.0392589569091797,
      "learning_rate": 2.458100558659218e-06,
      "loss": 1.3647,
      "step": 25
    },
    {
      "epoch": 0.004359216179398512,
      "grad_norm": 0.9552459716796875,
      "learning_rate": 2.569832402234637e-06,
      "loss": 1.3772,
      "step": 26
    },
    {
      "epoch": 0.0045268783401446085,
      "grad_norm": 1.024999737739563,
      "learning_rate": 2.6815642458100562e-06,
      "loss": 1.2639,
      "step": 27
    },
    {
      "epoch": 0.004694540500890705,
      "grad_norm": 0.9447226524353027,
      "learning_rate": 2.793296089385475e-06,
      "loss": 1.4179,
      "step": 28
    },
    {
      "epoch": 0.004862202661636802,
      "grad_norm": 1.181274175643921,
      "learning_rate": 2.9050279329608943e-06,
      "loss": 1.5209,
      "step": 29
    },
    {
      "epoch": 0.005029864822382898,
      "grad_norm": 0.858210563659668,
      "learning_rate": 3.016759776536313e-06,
      "loss": 1.3232,
      "step": 30
    },
    {
      "epoch": 0.005197526983128995,
      "grad_norm": 1.189065933227539,
      "learning_rate": 3.128491620111732e-06,
      "loss": 1.5529,
      "step": 31
    },
    {
      "epoch": 0.005365189143875092,
      "grad_norm": 0.9217438697814941,
      "learning_rate": 3.240223463687151e-06,
      "loss": 1.3698,
      "step": 32
    },
    {
      "epoch": 0.005532851304621189,
      "grad_norm": 1.0276325941085815,
      "learning_rate": 3.35195530726257e-06,
      "loss": 1.4141,
      "step": 33
    },
    {
      "epoch": 0.005700513465367285,
      "grad_norm": 0.8150102496147156,
      "learning_rate": 3.4636871508379893e-06,
      "loss": 1.2984,
      "step": 34
    },
    {
      "epoch": 0.005868175626113381,
      "grad_norm": 0.7057741284370422,
      "learning_rate": 3.575418994413408e-06,
      "loss": 1.3079,
      "step": 35
    },
    {
      "epoch": 0.006035837786859478,
      "grad_norm": 0.8169935345649719,
      "learning_rate": 3.687150837988827e-06,
      "loss": 1.4512,
      "step": 36
    },
    {
      "epoch": 0.006203499947605575,
      "grad_norm": 0.61067134141922,
      "learning_rate": 3.798882681564246e-06,
      "loss": 1.1787,
      "step": 37
    },
    {
      "epoch": 0.006371162108351671,
      "grad_norm": 0.7273681163787842,
      "learning_rate": 3.910614525139665e-06,
      "loss": 1.3199,
      "step": 38
    },
    {
      "epoch": 0.006538824269097768,
      "grad_norm": 0.6396476030349731,
      "learning_rate": 4.022346368715084e-06,
      "loss": 1.1523,
      "step": 39
    },
    {
      "epoch": 0.006706486429843865,
      "grad_norm": 0.5737459659576416,
      "learning_rate": 4.134078212290504e-06,
      "loss": 1.1673,
      "step": 40
    },
    {
      "epoch": 0.0068741485905899615,
      "grad_norm": 0.719789445400238,
      "learning_rate": 4.2458100558659224e-06,
      "loss": 1.3278,
      "step": 41
    },
    {
      "epoch": 0.0070418107513360575,
      "grad_norm": 0.5431291460990906,
      "learning_rate": 4.357541899441341e-06,
      "loss": 1.2429,
      "step": 42
    },
    {
      "epoch": 0.007209472912082154,
      "grad_norm": 0.6092455387115479,
      "learning_rate": 4.46927374301676e-06,
      "loss": 1.1985,
      "step": 43
    },
    {
      "epoch": 0.007377135072828251,
      "grad_norm": 0.5443903803825378,
      "learning_rate": 4.581005586592179e-06,
      "loss": 1.1565,
      "step": 44
    },
    {
      "epoch": 0.007544797233574348,
      "grad_norm": 0.5537682175636292,
      "learning_rate": 4.692737430167599e-06,
      "loss": 1.2249,
      "step": 45
    },
    {
      "epoch": 0.007712459394320444,
      "grad_norm": 0.6771538853645325,
      "learning_rate": 4.8044692737430175e-06,
      "loss": 1.1947,
      "step": 46
    },
    {
      "epoch": 0.007880121555066542,
      "grad_norm": 0.5913676023483276,
      "learning_rate": 4.916201117318436e-06,
      "loss": 1.1952,
      "step": 47
    },
    {
      "epoch": 0.008047783715812638,
      "grad_norm": 0.5791601538658142,
      "learning_rate": 5.027932960893855e-06,
      "loss": 1.2931,
      "step": 48
    },
    {
      "epoch": 0.008215445876558734,
      "grad_norm": 0.6004194617271423,
      "learning_rate": 5.139664804469274e-06,
      "loss": 1.207,
      "step": 49
    },
    {
      "epoch": 0.008383108037304831,
      "grad_norm": 0.6252710223197937,
      "learning_rate": 5.251396648044693e-06,
      "loss": 1.2427,
      "step": 50
    },
    {
      "epoch": 0.008550770198050927,
      "grad_norm": 0.5790568590164185,
      "learning_rate": 5.3631284916201125e-06,
      "loss": 1.2262,
      "step": 51
    },
    {
      "epoch": 0.008718432358797023,
      "grad_norm": 0.5642638802528381,
      "learning_rate": 5.474860335195531e-06,
      "loss": 1.1652,
      "step": 52
    },
    {
      "epoch": 0.008886094519543121,
      "grad_norm": 0.5329627394676208,
      "learning_rate": 5.58659217877095e-06,
      "loss": 1.1425,
      "step": 53
    },
    {
      "epoch": 0.009053756680289217,
      "grad_norm": 0.6454995274543762,
      "learning_rate": 5.698324022346369e-06,
      "loss": 1.2716,
      "step": 54
    },
    {
      "epoch": 0.009221418841035315,
      "grad_norm": 0.6814166307449341,
      "learning_rate": 5.810055865921789e-06,
      "loss": 1.1937,
      "step": 55
    },
    {
      "epoch": 0.00938908100178141,
      "grad_norm": 0.6075778007507324,
      "learning_rate": 5.9217877094972075e-06,
      "loss": 1.2141,
      "step": 56
    },
    {
      "epoch": 0.009556743162527507,
      "grad_norm": 0.53324955701828,
      "learning_rate": 6.033519553072626e-06,
      "loss": 1.1361,
      "step": 57
    },
    {
      "epoch": 0.009724405323273604,
      "grad_norm": 0.5701301693916321,
      "learning_rate": 6.145251396648045e-06,
      "loss": 1.1547,
      "step": 58
    },
    {
      "epoch": 0.0098920674840197,
      "grad_norm": 0.4995803236961365,
      "learning_rate": 6.256983240223464e-06,
      "loss": 1.1433,
      "step": 59
    },
    {
      "epoch": 0.010059729644765796,
      "grad_norm": 0.5348702669143677,
      "learning_rate": 6.368715083798883e-06,
      "loss": 1.1044,
      "step": 60
    },
    {
      "epoch": 0.010227391805511894,
      "grad_norm": 0.5916696190834045,
      "learning_rate": 6.480446927374302e-06,
      "loss": 1.1822,
      "step": 61
    },
    {
      "epoch": 0.01039505396625799,
      "grad_norm": 0.5305972695350647,
      "learning_rate": 6.592178770949721e-06,
      "loss": 1.2197,
      "step": 62
    },
    {
      "epoch": 0.010562716127004088,
      "grad_norm": 0.7340630888938904,
      "learning_rate": 6.70391061452514e-06,
      "loss": 1.2318,
      "step": 63
    },
    {
      "epoch": 0.010730378287750184,
      "grad_norm": 0.5850639939308167,
      "learning_rate": 6.815642458100559e-06,
      "loss": 1.2176,
      "step": 64
    },
    {
      "epoch": 0.01089804044849628,
      "grad_norm": 0.6341410279273987,
      "learning_rate": 6.927374301675979e-06,
      "loss": 1.1692,
      "step": 65
    },
    {
      "epoch": 0.011065702609242377,
      "grad_norm": 0.766492486000061,
      "learning_rate": 7.0391061452513975e-06,
      "loss": 1.147,
      "step": 66
    },
    {
      "epoch": 0.011233364769988473,
      "grad_norm": 0.541501522064209,
      "learning_rate": 7.150837988826816e-06,
      "loss": 1.2227,
      "step": 67
    },
    {
      "epoch": 0.01140102693073457,
      "grad_norm": 0.47469985485076904,
      "learning_rate": 7.262569832402235e-06,
      "loss": 1.1507,
      "step": 68
    },
    {
      "epoch": 0.011568689091480667,
      "grad_norm": 0.6132412552833557,
      "learning_rate": 7.374301675977654e-06,
      "loss": 1.282,
      "step": 69
    },
    {
      "epoch": 0.011736351252226763,
      "grad_norm": 0.4676038920879364,
      "learning_rate": 7.486033519553073e-06,
      "loss": 1.1221,
      "step": 70
    },
    {
      "epoch": 0.01190401341297286,
      "grad_norm": 0.5179970264434814,
      "learning_rate": 7.597765363128492e-06,
      "loss": 1.0446,
      "step": 71
    },
    {
      "epoch": 0.012071675573718956,
      "grad_norm": 0.4430614113807678,
      "learning_rate": 7.709497206703911e-06,
      "loss": 1.1906,
      "step": 72
    },
    {
      "epoch": 0.012239337734465052,
      "grad_norm": 0.4925226867198944,
      "learning_rate": 7.82122905027933e-06,
      "loss": 1.1505,
      "step": 73
    },
    {
      "epoch": 0.01240699989521115,
      "grad_norm": 0.5351929664611816,
      "learning_rate": 7.932960893854749e-06,
      "loss": 1.207,
      "step": 74
    },
    {
      "epoch": 0.012574662055957246,
      "grad_norm": 0.5194462537765503,
      "learning_rate": 8.044692737430168e-06,
      "loss": 1.1883,
      "step": 75
    },
    {
      "epoch": 0.012742324216703342,
      "grad_norm": 0.5400429964065552,
      "learning_rate": 8.156424581005588e-06,
      "loss": 1.0012,
      "step": 76
    },
    {
      "epoch": 0.01290998637744944,
      "grad_norm": 0.47353821992874146,
      "learning_rate": 8.268156424581007e-06,
      "loss": 1.171,
      "step": 77
    },
    {
      "epoch": 0.013077648538195536,
      "grad_norm": 0.6433570384979248,
      "learning_rate": 8.379888268156426e-06,
      "loss": 1.2229,
      "step": 78
    },
    {
      "epoch": 0.013245310698941633,
      "grad_norm": 0.5411203503608704,
      "learning_rate": 8.491620111731845e-06,
      "loss": 1.0776,
      "step": 79
    },
    {
      "epoch": 0.01341297285968773,
      "grad_norm": 0.6546565294265747,
      "learning_rate": 8.603351955307264e-06,
      "loss": 1.1872,
      "step": 80
    },
    {
      "epoch": 0.013580635020433825,
      "grad_norm": 0.6882161498069763,
      "learning_rate": 8.715083798882683e-06,
      "loss": 1.1454,
      "step": 81
    },
    {
      "epoch": 0.013748297181179923,
      "grad_norm": 0.5556113123893738,
      "learning_rate": 8.826815642458101e-06,
      "loss": 1.1364,
      "step": 82
    },
    {
      "epoch": 0.013915959341926019,
      "grad_norm": 0.6302836537361145,
      "learning_rate": 8.93854748603352e-06,
      "loss": 1.2659,
      "step": 83
    },
    {
      "epoch": 0.014083621502672115,
      "grad_norm": 0.8605783581733704,
      "learning_rate": 9.050279329608939e-06,
      "loss": 1.1607,
      "step": 84
    },
    {
      "epoch": 0.014251283663418213,
      "grad_norm": 0.5417123436927795,
      "learning_rate": 9.162011173184358e-06,
      "loss": 1.1393,
      "step": 85
    },
    {
      "epoch": 0.014418945824164309,
      "grad_norm": 0.6310127973556519,
      "learning_rate": 9.273743016759777e-06,
      "loss": 1.3032,
      "step": 86
    },
    {
      "epoch": 0.014586607984910405,
      "grad_norm": 0.5595749020576477,
      "learning_rate": 9.385474860335197e-06,
      "loss": 1.1699,
      "step": 87
    },
    {
      "epoch": 0.014754270145656502,
      "grad_norm": 0.5508741736412048,
      "learning_rate": 9.497206703910616e-06,
      "loss": 1.0422,
      "step": 88
    },
    {
      "epoch": 0.014921932306402598,
      "grad_norm": 0.44198355078697205,
      "learning_rate": 9.608938547486035e-06,
      "loss": 1.006,
      "step": 89
    },
    {
      "epoch": 0.015089594467148696,
      "grad_norm": 0.612446129322052,
      "learning_rate": 9.720670391061454e-06,
      "loss": 1.2126,
      "step": 90
    },
    {
      "epoch": 0.015257256627894792,
      "grad_norm": 0.534264087677002,
      "learning_rate": 9.832402234636873e-06,
      "loss": 1.0954,
      "step": 91
    },
    {
      "epoch": 0.015424918788640888,
      "grad_norm": 0.4617749750614166,
      "learning_rate": 9.944134078212291e-06,
      "loss": 1.0987,
      "step": 92
    },
    {
      "epoch": 0.015592580949386986,
      "grad_norm": 0.5459811687469482,
      "learning_rate": 1.005586592178771e-05,
      "loss": 1.1247,
      "step": 93
    },
    {
      "epoch": 0.015760243110133083,
      "grad_norm": 0.4880772829055786,
      "learning_rate": 1.0167597765363129e-05,
      "loss": 1.0422,
      "step": 94
    },
    {
      "epoch": 0.01592790527087918,
      "grad_norm": 0.5880354046821594,
      "learning_rate": 1.0279329608938548e-05,
      "loss": 1.166,
      "step": 95
    },
    {
      "epoch": 0.016095567431625275,
      "grad_norm": 0.5863614678382874,
      "learning_rate": 1.0391061452513967e-05,
      "loss": 1.1016,
      "step": 96
    },
    {
      "epoch": 0.01626322959237137,
      "grad_norm": 0.5670218467712402,
      "learning_rate": 1.0502793296089386e-05,
      "loss": 0.9635,
      "step": 97
    },
    {
      "epoch": 0.016430891753117467,
      "grad_norm": 0.8734570741653442,
      "learning_rate": 1.0614525139664806e-05,
      "loss": 1.0499,
      "step": 98
    },
    {
      "epoch": 0.016598553913863563,
      "grad_norm": 0.6028108596801758,
      "learning_rate": 1.0726256983240225e-05,
      "loss": 1.2074,
      "step": 99
    },
    {
      "epoch": 0.016766216074609663,
      "grad_norm": 0.5182443857192993,
      "learning_rate": 1.0837988826815644e-05,
      "loss": 1.0133,
      "step": 100
    },
    {
      "epoch": 0.01693387823535576,
      "grad_norm": 0.7909883260726929,
      "learning_rate": 1.0949720670391063e-05,
      "loss": 1.1455,
      "step": 101
    },
    {
      "epoch": 0.017101540396101855,
      "grad_norm": 0.6897411942481995,
      "learning_rate": 1.1061452513966481e-05,
      "loss": 1.0995,
      "step": 102
    },
    {
      "epoch": 0.01726920255684795,
      "grad_norm": 0.6529523134231567,
      "learning_rate": 1.11731843575419e-05,
      "loss": 1.0004,
      "step": 103
    },
    {
      "epoch": 0.017436864717594047,
      "grad_norm": 0.68169766664505,
      "learning_rate": 1.1284916201117319e-05,
      "loss": 1.1065,
      "step": 104
    },
    {
      "epoch": 0.017604526878340146,
      "grad_norm": 0.7525808811187744,
      "learning_rate": 1.1396648044692738e-05,
      "loss": 1.1017,
      "step": 105
    },
    {
      "epoch": 0.017772189039086242,
      "grad_norm": 0.6268847584724426,
      "learning_rate": 1.1508379888268157e-05,
      "loss": 1.0249,
      "step": 106
    },
    {
      "epoch": 0.017939851199832338,
      "grad_norm": 0.6089746952056885,
      "learning_rate": 1.1620111731843577e-05,
      "loss": 1.1194,
      "step": 107
    },
    {
      "epoch": 0.018107513360578434,
      "grad_norm": 0.8341772556304932,
      "learning_rate": 1.1731843575418994e-05,
      "loss": 1.1703,
      "step": 108
    },
    {
      "epoch": 0.01827517552132453,
      "grad_norm": 0.5683283805847168,
      "learning_rate": 1.1843575418994415e-05,
      "loss": 1.0827,
      "step": 109
    },
    {
      "epoch": 0.01844283768207063,
      "grad_norm": 0.6581299304962158,
      "learning_rate": 1.1955307262569834e-05,
      "loss": 1.091,
      "step": 110
    },
    {
      "epoch": 0.018610499842816725,
      "grad_norm": 0.489067018032074,
      "learning_rate": 1.2067039106145253e-05,
      "loss": 1.0926,
      "step": 111
    },
    {
      "epoch": 0.01877816200356282,
      "grad_norm": 0.76224684715271,
      "learning_rate": 1.2178770949720671e-05,
      "loss": 1.0664,
      "step": 112
    },
    {
      "epoch": 0.018945824164308917,
      "grad_norm": 0.5604745149612427,
      "learning_rate": 1.229050279329609e-05,
      "loss": 1.1076,
      "step": 113
    },
    {
      "epoch": 0.019113486325055013,
      "grad_norm": 0.6854090094566345,
      "learning_rate": 1.2402234636871509e-05,
      "loss": 1.128,
      "step": 114
    },
    {
      "epoch": 0.01928114848580111,
      "grad_norm": 0.7920408844947815,
      "learning_rate": 1.2513966480446928e-05,
      "loss": 1.2003,
      "step": 115
    },
    {
      "epoch": 0.01944881064654721,
      "grad_norm": 0.6446617841720581,
      "learning_rate": 1.2625698324022347e-05,
      "loss": 1.1067,
      "step": 116
    },
    {
      "epoch": 0.019616472807293305,
      "grad_norm": 0.5878286361694336,
      "learning_rate": 1.2737430167597766e-05,
      "loss": 1.116,
      "step": 117
    },
    {
      "epoch": 0.0197841349680394,
      "grad_norm": 0.46466970443725586,
      "learning_rate": 1.2849162011173186e-05,
      "loss": 1.0498,
      "step": 118
    },
    {
      "epoch": 0.019951797128785496,
      "grad_norm": 0.6509464383125305,
      "learning_rate": 1.2960893854748603e-05,
      "loss": 1.1395,
      "step": 119
    },
    {
      "epoch": 0.020119459289531592,
      "grad_norm": 0.6141960024833679,
      "learning_rate": 1.3072625698324024e-05,
      "loss": 0.9784,
      "step": 120
    },
    {
      "epoch": 0.020287121450277692,
      "grad_norm": 0.6657926440238953,
      "learning_rate": 1.3184357541899443e-05,
      "loss": 1.0352,
      "step": 121
    },
    {
      "epoch": 0.020454783611023788,
      "grad_norm": 0.5377758741378784,
      "learning_rate": 1.3296089385474861e-05,
      "loss": 1.0195,
      "step": 122
    },
    {
      "epoch": 0.020622445771769884,
      "grad_norm": 0.5579245686531067,
      "learning_rate": 1.340782122905028e-05,
      "loss": 1.0121,
      "step": 123
    },
    {
      "epoch": 0.02079010793251598,
      "grad_norm": 0.5483793020248413,
      "learning_rate": 1.3519553072625699e-05,
      "loss": 1.0246,
      "step": 124
    },
    {
      "epoch": 0.020957770093262076,
      "grad_norm": 0.5603605508804321,
      "learning_rate": 1.3631284916201118e-05,
      "loss": 1.0742,
      "step": 125
    },
    {
      "epoch": 0.021125432254008175,
      "grad_norm": 0.761964738368988,
      "learning_rate": 1.3743016759776537e-05,
      "loss": 1.0959,
      "step": 126
    },
    {
      "epoch": 0.02129309441475427,
      "grad_norm": 0.49273210763931274,
      "learning_rate": 1.3854748603351957e-05,
      "loss": 1.2079,
      "step": 127
    },
    {
      "epoch": 0.021460756575500367,
      "grad_norm": 0.4556741714477539,
      "learning_rate": 1.3966480446927374e-05,
      "loss": 1.0563,
      "step": 128
    },
    {
      "epoch": 0.021628418736246463,
      "grad_norm": 0.6324474811553955,
      "learning_rate": 1.4078212290502795e-05,
      "loss": 1.002,
      "step": 129
    },
    {
      "epoch": 0.02179608089699256,
      "grad_norm": 0.5139722228050232,
      "learning_rate": 1.4189944134078212e-05,
      "loss": 1.0051,
      "step": 130
    },
    {
      "epoch": 0.021963743057738655,
      "grad_norm": 0.5473005175590515,
      "learning_rate": 1.4301675977653633e-05,
      "loss": 1.051,
      "step": 131
    },
    {
      "epoch": 0.022131405218484754,
      "grad_norm": 0.6989631056785583,
      "learning_rate": 1.4413407821229052e-05,
      "loss": 1.0833,
      "step": 132
    },
    {
      "epoch": 0.02229906737923085,
      "grad_norm": 0.6231628656387329,
      "learning_rate": 1.452513966480447e-05,
      "loss": 1.1625,
      "step": 133
    },
    {
      "epoch": 0.022466729539976946,
      "grad_norm": 0.7199480533599854,
      "learning_rate": 1.463687150837989e-05,
      "loss": 1.0238,
      "step": 134
    },
    {
      "epoch": 0.022634391700723042,
      "grad_norm": 0.5824499130249023,
      "learning_rate": 1.4748603351955308e-05,
      "loss": 1.0499,
      "step": 135
    },
    {
      "epoch": 0.02280205386146914,
      "grad_norm": 0.6677570343017578,
      "learning_rate": 1.4860335195530729e-05,
      "loss": 1.1293,
      "step": 136
    },
    {
      "epoch": 0.022969716022215238,
      "grad_norm": 0.5944182276725769,
      "learning_rate": 1.4972067039106146e-05,
      "loss": 1.0386,
      "step": 137
    },
    {
      "epoch": 0.023137378182961334,
      "grad_norm": 1.3920981884002686,
      "learning_rate": 1.5083798882681566e-05,
      "loss": 1.1251,
      "step": 138
    },
    {
      "epoch": 0.02330504034370743,
      "grad_norm": 0.7003018260002136,
      "learning_rate": 1.5195530726256983e-05,
      "loss": 1.0233,
      "step": 139
    },
    {
      "epoch": 0.023472702504453526,
      "grad_norm": 0.6967798471450806,
      "learning_rate": 1.5307262569832404e-05,
      "loss": 1.0562,
      "step": 140
    },
    {
      "epoch": 0.02364036466519962,
      "grad_norm": 0.5572317838668823,
      "learning_rate": 1.5418994413407823e-05,
      "loss": 0.9461,
      "step": 141
    },
    {
      "epoch": 0.02380802682594572,
      "grad_norm": 0.7279749512672424,
      "learning_rate": 1.553072625698324e-05,
      "loss": 1.0506,
      "step": 142
    },
    {
      "epoch": 0.023975688986691817,
      "grad_norm": 0.5048472285270691,
      "learning_rate": 1.564245810055866e-05,
      "loss": 1.048,
      "step": 143
    },
    {
      "epoch": 0.024143351147437913,
      "grad_norm": 0.6082873344421387,
      "learning_rate": 1.575418994413408e-05,
      "loss": 1.0414,
      "step": 144
    },
    {
      "epoch": 0.02431101330818401,
      "grad_norm": 0.5643455386161804,
      "learning_rate": 1.5865921787709498e-05,
      "loss": 1.0404,
      "step": 145
    },
    {
      "epoch": 0.024478675468930105,
      "grad_norm": 0.46841081976890564,
      "learning_rate": 1.5977653631284917e-05,
      "loss": 1.0697,
      "step": 146
    },
    {
      "epoch": 0.0246463376296762,
      "grad_norm": 0.5768006443977356,
      "learning_rate": 1.6089385474860336e-05,
      "loss": 1.1046,
      "step": 147
    },
    {
      "epoch": 0.0248139997904223,
      "grad_norm": 0.6741750836372375,
      "learning_rate": 1.6201117318435755e-05,
      "loss": 1.0681,
      "step": 148
    },
    {
      "epoch": 0.024981661951168396,
      "grad_norm": 0.6723300814628601,
      "learning_rate": 1.6312849162011177e-05,
      "loss": 1.1534,
      "step": 149
    },
    {
      "epoch": 0.025149324111914492,
      "grad_norm": 0.5499408841133118,
      "learning_rate": 1.6424581005586592e-05,
      "loss": 1.0647,
      "step": 150
    },
    {
      "epoch": 0.025316986272660588,
      "grad_norm": 0.5709015727043152,
      "learning_rate": 1.6536312849162014e-05,
      "loss": 0.9977,
      "step": 151
    },
    {
      "epoch": 0.025484648433406684,
      "grad_norm": 0.6413765549659729,
      "learning_rate": 1.664804469273743e-05,
      "loss": 1.1216,
      "step": 152
    },
    {
      "epoch": 0.025652310594152784,
      "grad_norm": 0.566363513469696,
      "learning_rate": 1.6759776536312852e-05,
      "loss": 1.1001,
      "step": 153
    },
    {
      "epoch": 0.02581997275489888,
      "grad_norm": 0.5891416072845459,
      "learning_rate": 1.687150837988827e-05,
      "loss": 0.9475,
      "step": 154
    },
    {
      "epoch": 0.025987634915644976,
      "grad_norm": 0.6749016046524048,
      "learning_rate": 1.698324022346369e-05,
      "loss": 1.1254,
      "step": 155
    },
    {
      "epoch": 0.02615529707639107,
      "grad_norm": 0.5689506530761719,
      "learning_rate": 1.709497206703911e-05,
      "loss": 1.0695,
      "step": 156
    },
    {
      "epoch": 0.026322959237137167,
      "grad_norm": 0.8832252621650696,
      "learning_rate": 1.7206703910614527e-05,
      "loss": 1.0325,
      "step": 157
    },
    {
      "epoch": 0.026490621397883267,
      "grad_norm": 0.7964456081390381,
      "learning_rate": 1.7318435754189946e-05,
      "loss": 1.0879,
      "step": 158
    },
    {
      "epoch": 0.026658283558629363,
      "grad_norm": 0.5532066822052002,
      "learning_rate": 1.7430167597765365e-05,
      "loss": 1.0784,
      "step": 159
    },
    {
      "epoch": 0.02682594571937546,
      "grad_norm": 0.6335783004760742,
      "learning_rate": 1.7541899441340784e-05,
      "loss": 1.0252,
      "step": 160
    },
    {
      "epoch": 0.026993607880121555,
      "grad_norm": 0.5445144176483154,
      "learning_rate": 1.7653631284916203e-05,
      "loss": 1.1504,
      "step": 161
    },
    {
      "epoch": 0.02716127004086765,
      "grad_norm": 0.6088312268257141,
      "learning_rate": 1.776536312849162e-05,
      "loss": 1.0816,
      "step": 162
    },
    {
      "epoch": 0.027328932201613747,
      "grad_norm": 0.6831774711608887,
      "learning_rate": 1.787709497206704e-05,
      "loss": 1.035,
      "step": 163
    },
    {
      "epoch": 0.027496594362359846,
      "grad_norm": 0.4983123242855072,
      "learning_rate": 1.798882681564246e-05,
      "loss": 1.0276,
      "step": 164
    },
    {
      "epoch": 0.027664256523105942,
      "grad_norm": 0.5578044056892395,
      "learning_rate": 1.8100558659217878e-05,
      "loss": 1.0178,
      "step": 165
    },
    {
      "epoch": 0.027831918683852038,
      "grad_norm": 0.7284736633300781,
      "learning_rate": 1.8212290502793297e-05,
      "loss": 0.9363,
      "step": 166
    },
    {
      "epoch": 0.027999580844598134,
      "grad_norm": 0.43957066535949707,
      "learning_rate": 1.8324022346368716e-05,
      "loss": 1.0298,
      "step": 167
    },
    {
      "epoch": 0.02816724300534423,
      "grad_norm": 0.46956202387809753,
      "learning_rate": 1.8435754189944135e-05,
      "loss": 1.0963,
      "step": 168
    },
    {
      "epoch": 0.02833490516609033,
      "grad_norm": 0.6402130126953125,
      "learning_rate": 1.8547486033519553e-05,
      "loss": 1.0679,
      "step": 169
    },
    {
      "epoch": 0.028502567326836425,
      "grad_norm": 0.6925402283668518,
      "learning_rate": 1.8659217877094972e-05,
      "loss": 1.0334,
      "step": 170
    },
    {
      "epoch": 0.02867022948758252,
      "grad_norm": 1.1133993864059448,
      "learning_rate": 1.8770949720670394e-05,
      "loss": 0.9692,
      "step": 171
    },
    {
      "epoch": 0.028837891648328617,
      "grad_norm": 0.5076237916946411,
      "learning_rate": 1.888268156424581e-05,
      "loss": 1.1312,
      "step": 172
    },
    {
      "epoch": 0.029005553809074713,
      "grad_norm": 0.6855918169021606,
      "learning_rate": 1.8994413407821232e-05,
      "loss": 1.0913,
      "step": 173
    },
    {
      "epoch": 0.02917321596982081,
      "grad_norm": 0.5001389384269714,
      "learning_rate": 1.910614525139665e-05,
      "loss": 1.1183,
      "step": 174
    },
    {
      "epoch": 0.02934087813056691,
      "grad_norm": 0.5716369152069092,
      "learning_rate": 1.921787709497207e-05,
      "loss": 0.9813,
      "step": 175
    },
    {
      "epoch": 0.029508540291313005,
      "grad_norm": 0.7187888622283936,
      "learning_rate": 1.932960893854749e-05,
      "loss": 0.9594,
      "step": 176
    },
    {
      "epoch": 0.0296762024520591,
      "grad_norm": 0.5117635726928711,
      "learning_rate": 1.9441340782122907e-05,
      "loss": 1.0199,
      "step": 177
    },
    {
      "epoch": 0.029843864612805197,
      "grad_norm": 0.5199819803237915,
      "learning_rate": 1.9553072625698326e-05,
      "loss": 1.0014,
      "step": 178
    },
    {
      "epoch": 0.030011526773551293,
      "grad_norm": 0.5007014870643616,
      "learning_rate": 1.9664804469273745e-05,
      "loss": 0.9336,
      "step": 179
    },
    {
      "epoch": 0.030179188934297392,
      "grad_norm": 0.5830110907554626,
      "learning_rate": 1.9776536312849164e-05,
      "loss": 1.0689,
      "step": 180
    },
    {
      "epoch": 0.030346851095043488,
      "grad_norm": 0.527374804019928,
      "learning_rate": 1.9888268156424583e-05,
      "loss": 1.0273,
      "step": 181
    },
    {
      "epoch": 0.030514513255789584,
      "grad_norm": 0.5847291350364685,
      "learning_rate": 2e-05,
      "loss": 1.0122,
      "step": 182
    },
    {
      "epoch": 0.03068217541653568,
      "grad_norm": 0.5212209820747375,
      "learning_rate": 1.9999998525438088e-05,
      "loss": 0.8963,
      "step": 183
    },
    {
      "epoch": 0.030849837577281776,
      "grad_norm": 0.49605804681777954,
      "learning_rate": 1.999999410175278e-05,
      "loss": 0.9395,
      "step": 184
    },
    {
      "epoch": 0.031017499738027875,
      "grad_norm": 0.47218137979507446,
      "learning_rate": 1.9999986728945384e-05,
      "loss": 1.0157,
      "step": 185
    },
    {
      "epoch": 0.03118516189877397,
      "grad_norm": 0.7420831322669983,
      "learning_rate": 1.9999976407018074e-05,
      "loss": 1.0173,
      "step": 186
    },
    {
      "epoch": 0.031352824059520064,
      "grad_norm": 0.48486003279685974,
      "learning_rate": 1.9999963135973894e-05,
      "loss": 0.9915,
      "step": 187
    },
    {
      "epoch": 0.03152048622026617,
      "grad_norm": 0.6362403035163879,
      "learning_rate": 1.9999946915816756e-05,
      "loss": 1.0975,
      "step": 188
    },
    {
      "epoch": 0.03168814838101226,
      "grad_norm": 0.5725786685943604,
      "learning_rate": 1.9999927746551447e-05,
      "loss": 1.0043,
      "step": 189
    },
    {
      "epoch": 0.03185581054175836,
      "grad_norm": 0.5909467339515686,
      "learning_rate": 1.9999905628183617e-05,
      "loss": 1.0289,
      "step": 190
    },
    {
      "epoch": 0.032023472702504455,
      "grad_norm": 0.5642744898796082,
      "learning_rate": 1.999988056071979e-05,
      "loss": 1.1165,
      "step": 191
    },
    {
      "epoch": 0.03219113486325055,
      "grad_norm": 0.3727734386920929,
      "learning_rate": 1.999985254416736e-05,
      "loss": 0.9655,
      "step": 192
    },
    {
      "epoch": 0.03235879702399665,
      "grad_norm": 0.48998305201530457,
      "learning_rate": 1.9999821578534588e-05,
      "loss": 1.1287,
      "step": 193
    },
    {
      "epoch": 0.03252645918474274,
      "grad_norm": 0.44185349345207214,
      "learning_rate": 1.9999787663830606e-05,
      "loss": 0.9565,
      "step": 194
    },
    {
      "epoch": 0.03269412134548884,
      "grad_norm": 0.40309658646583557,
      "learning_rate": 1.9999750800065415e-05,
      "loss": 1.0181,
      "step": 195
    },
    {
      "epoch": 0.032861783506234934,
      "grad_norm": 0.5723433494567871,
      "learning_rate": 1.9999710987249892e-05,
      "loss": 1.0667,
      "step": 196
    },
    {
      "epoch": 0.03302944566698103,
      "grad_norm": 0.5408050417900085,
      "learning_rate": 1.9999668225395772e-05,
      "loss": 0.9858,
      "step": 197
    },
    {
      "epoch": 0.033197107827727126,
      "grad_norm": 0.83476322889328,
      "learning_rate": 1.9999622514515667e-05,
      "loss": 1.0893,
      "step": 198
    },
    {
      "epoch": 0.03336476998847323,
      "grad_norm": 0.8037353754043579,
      "learning_rate": 1.999957385462306e-05,
      "loss": 0.9974,
      "step": 199
    },
    {
      "epoch": 0.033532432149219325,
      "grad_norm": 0.7400439381599426,
      "learning_rate": 1.9999522245732302e-05,
      "loss": 1.1179,
      "step": 200
    },
    {
      "epoch": 0.03370009430996542,
      "grad_norm": 0.5144438743591309,
      "learning_rate": 1.999946768785861e-05,
      "loss": 1.0092,
      "step": 201
    },
    {
      "epoch": 0.03386775647071152,
      "grad_norm": 0.5504910349845886,
      "learning_rate": 1.9999410181018074e-05,
      "loss": 0.9403,
      "step": 202
    },
    {
      "epoch": 0.03403541863145761,
      "grad_norm": 0.5855805277824402,
      "learning_rate": 1.999934972522766e-05,
      "loss": 1.0515,
      "step": 203
    },
    {
      "epoch": 0.03420308079220371,
      "grad_norm": 0.7673451900482178,
      "learning_rate": 1.999928632050519e-05,
      "loss": 1.1424,
      "step": 204
    },
    {
      "epoch": 0.034370742952949805,
      "grad_norm": 0.7211079597473145,
      "learning_rate": 1.999921996686936e-05,
      "loss": 0.9997,
      "step": 205
    },
    {
      "epoch": 0.0345384051136959,
      "grad_norm": 0.4259258806705475,
      "learning_rate": 1.9999150664339745e-05,
      "loss": 0.9601,
      "step": 206
    },
    {
      "epoch": 0.034706067274442,
      "grad_norm": 0.4799496531486511,
      "learning_rate": 1.9999078412936784e-05,
      "loss": 1.0027,
      "step": 207
    },
    {
      "epoch": 0.03487372943518809,
      "grad_norm": 0.5094309449195862,
      "learning_rate": 1.999900321268178e-05,
      "loss": 0.9668,
      "step": 208
    },
    {
      "epoch": 0.035041391595934196,
      "grad_norm": 0.49225088953971863,
      "learning_rate": 1.9998925063596915e-05,
      "loss": 1.0153,
      "step": 209
    },
    {
      "epoch": 0.03520905375668029,
      "grad_norm": 0.4400315582752228,
      "learning_rate": 1.9998843965705235e-05,
      "loss": 1.0868,
      "step": 210
    },
    {
      "epoch": 0.03537671591742639,
      "grad_norm": 0.6707110404968262,
      "learning_rate": 1.9998759919030653e-05,
      "loss": 1.0686,
      "step": 211
    },
    {
      "epoch": 0.035544378078172484,
      "grad_norm": 0.4518424868583679,
      "learning_rate": 1.999867292359796e-05,
      "loss": 1.0012,
      "step": 212
    },
    {
      "epoch": 0.03571204023891858,
      "grad_norm": 0.39341041445732117,
      "learning_rate": 1.999858297943281e-05,
      "loss": 0.9752,
      "step": 213
    },
    {
      "epoch": 0.035879702399664676,
      "grad_norm": 0.42496252059936523,
      "learning_rate": 1.9998490086561727e-05,
      "loss": 1.008,
      "step": 214
    },
    {
      "epoch": 0.03604736456041077,
      "grad_norm": 0.5031048059463501,
      "learning_rate": 1.999839424501211e-05,
      "loss": 1.024,
      "step": 215
    },
    {
      "epoch": 0.03621502672115687,
      "grad_norm": 0.4959554970264435,
      "learning_rate": 1.9998295454812218e-05,
      "loss": 0.9674,
      "step": 216
    },
    {
      "epoch": 0.036382688881902964,
      "grad_norm": 0.5379063487052917,
      "learning_rate": 1.9998193715991192e-05,
      "loss": 1.0666,
      "step": 217
    },
    {
      "epoch": 0.03655035104264906,
      "grad_norm": 0.46192410588264465,
      "learning_rate": 1.9998089028579035e-05,
      "loss": 0.9727,
      "step": 218
    },
    {
      "epoch": 0.036718013203395156,
      "grad_norm": 0.5664047002792358,
      "learning_rate": 1.9997981392606617e-05,
      "loss": 1.0543,
      "step": 219
    },
    {
      "epoch": 0.03688567536414126,
      "grad_norm": 0.7212116122245789,
      "learning_rate": 1.9997870808105683e-05,
      "loss": 1.07,
      "step": 220
    },
    {
      "epoch": 0.037053337524887354,
      "grad_norm": 0.712087869644165,
      "learning_rate": 1.9997757275108847e-05,
      "loss": 1.0069,
      "step": 221
    },
    {
      "epoch": 0.03722099968563345,
      "grad_norm": 0.6852487921714783,
      "learning_rate": 1.999764079364959e-05,
      "loss": 1.015,
      "step": 222
    },
    {
      "epoch": 0.037388661846379546,
      "grad_norm": 0.7390941977500916,
      "learning_rate": 1.9997521363762264e-05,
      "loss": 1.0333,
      "step": 223
    },
    {
      "epoch": 0.03755632400712564,
      "grad_norm": 0.5429960489273071,
      "learning_rate": 1.999739898548209e-05,
      "loss": 1.0331,
      "step": 224
    },
    {
      "epoch": 0.03772398616787174,
      "grad_norm": 0.5059285759925842,
      "learning_rate": 1.999727365884516e-05,
      "loss": 1.0684,
      "step": 225
    },
    {
      "epoch": 0.037891648328617834,
      "grad_norm": 0.546054482460022,
      "learning_rate": 1.999714538388843e-05,
      "loss": 1.0447,
      "step": 226
    },
    {
      "epoch": 0.03805931048936393,
      "grad_norm": 0.420764297246933,
      "learning_rate": 1.9997014160649736e-05,
      "loss": 1.033,
      "step": 227
    },
    {
      "epoch": 0.038226972650110026,
      "grad_norm": 0.4457850754261017,
      "learning_rate": 1.9996879989167777e-05,
      "loss": 1.0114,
      "step": 228
    },
    {
      "epoch": 0.03839463481085612,
      "grad_norm": 0.5681815147399902,
      "learning_rate": 1.999674286948212e-05,
      "loss": 1.0674,
      "step": 229
    },
    {
      "epoch": 0.03856229697160222,
      "grad_norm": 0.4577525854110718,
      "learning_rate": 1.9996602801633197e-05,
      "loss": 1.0182,
      "step": 230
    },
    {
      "epoch": 0.03872995913234832,
      "grad_norm": 0.47984087467193604,
      "learning_rate": 1.999645978566233e-05,
      "loss": 1.0901,
      "step": 231
    },
    {
      "epoch": 0.03889762129309442,
      "grad_norm": 0.5177239179611206,
      "learning_rate": 1.9996313821611682e-05,
      "loss": 1.0182,
      "step": 232
    },
    {
      "epoch": 0.03906528345384051,
      "grad_norm": 0.5124174952507019,
      "learning_rate": 1.9996164909524306e-05,
      "loss": 0.9901,
      "step": 233
    },
    {
      "epoch": 0.03923294561458661,
      "grad_norm": 0.5485550761222839,
      "learning_rate": 1.9996013049444117e-05,
      "loss": 0.9861,
      "step": 234
    },
    {
      "epoch": 0.039400607775332705,
      "grad_norm": 1.155792474746704,
      "learning_rate": 1.9995858241415904e-05,
      "loss": 1.0383,
      "step": 235
    },
    {
      "epoch": 0.0395682699360788,
      "grad_norm": 0.46704602241516113,
      "learning_rate": 1.999570048548532e-05,
      "loss": 1.002,
      "step": 236
    },
    {
      "epoch": 0.0397359320968249,
      "grad_norm": 0.5665890574455261,
      "learning_rate": 1.9995539781698882e-05,
      "loss": 1.0785,
      "step": 237
    },
    {
      "epoch": 0.03990359425757099,
      "grad_norm": 0.41134896874427795,
      "learning_rate": 1.9995376130103992e-05,
      "loss": 0.8948,
      "step": 238
    },
    {
      "epoch": 0.04007125641831709,
      "grad_norm": 0.47926095128059387,
      "learning_rate": 1.9995209530748913e-05,
      "loss": 0.9718,
      "step": 239
    },
    {
      "epoch": 0.040238918579063185,
      "grad_norm": 0.4371228814125061,
      "learning_rate": 1.9995039983682774e-05,
      "loss": 1.0359,
      "step": 240
    },
    {
      "epoch": 0.04040658073980929,
      "grad_norm": 0.5196617245674133,
      "learning_rate": 1.9994867488955573e-05,
      "loss": 1.0706,
      "step": 241
    },
    {
      "epoch": 0.040574242900555384,
      "grad_norm": 0.4669308364391327,
      "learning_rate": 1.9994692046618188e-05,
      "loss": 0.9539,
      "step": 242
    },
    {
      "epoch": 0.04074190506130148,
      "grad_norm": 0.5073533654212952,
      "learning_rate": 1.9994513656722356e-05,
      "loss": 1.0063,
      "step": 243
    },
    {
      "epoch": 0.040909567222047576,
      "grad_norm": 0.47835734486579895,
      "learning_rate": 1.9994332319320688e-05,
      "loss": 0.9411,
      "step": 244
    },
    {
      "epoch": 0.04107722938279367,
      "grad_norm": 0.5108445286750793,
      "learning_rate": 1.9994148034466655e-05,
      "loss": 0.9498,
      "step": 245
    },
    {
      "epoch": 0.04124489154353977,
      "grad_norm": 0.45715704560279846,
      "learning_rate": 1.9993960802214617e-05,
      "loss": 1.0476,
      "step": 246
    },
    {
      "epoch": 0.041412553704285864,
      "grad_norm": 0.5002694129943848,
      "learning_rate": 1.9993770622619784e-05,
      "loss": 1.0674,
      "step": 247
    },
    {
      "epoch": 0.04158021586503196,
      "grad_norm": 0.659797728061676,
      "learning_rate": 1.9993577495738245e-05,
      "loss": 1.0106,
      "step": 248
    },
    {
      "epoch": 0.041747878025778055,
      "grad_norm": 0.69830322265625,
      "learning_rate": 1.999338142162695e-05,
      "loss": 1.0719,
      "step": 249
    },
    {
      "epoch": 0.04191554018652415,
      "grad_norm": 0.6348698735237122,
      "learning_rate": 1.9993182400343727e-05,
      "loss": 1.0302,
      "step": 250
    },
    {
      "epoch": 0.04208320234727025,
      "grad_norm": 0.5063202977180481,
      "learning_rate": 1.9992980431947276e-05,
      "loss": 1.115,
      "step": 251
    },
    {
      "epoch": 0.04225086450801635,
      "grad_norm": 0.43055394291877747,
      "learning_rate": 1.9992775516497148e-05,
      "loss": 0.9884,
      "step": 252
    },
    {
      "epoch": 0.042418526668762446,
      "grad_norm": 0.48402297496795654,
      "learning_rate": 1.9992567654053785e-05,
      "loss": 1.0252,
      "step": 253
    },
    {
      "epoch": 0.04258618882950854,
      "grad_norm": 0.48005810379981995,
      "learning_rate": 1.9992356844678485e-05,
      "loss": 0.9396,
      "step": 254
    },
    {
      "epoch": 0.04275385099025464,
      "grad_norm": 0.5253524780273438,
      "learning_rate": 1.9992143088433417e-05,
      "loss": 1.0266,
      "step": 255
    },
    {
      "epoch": 0.042921513151000734,
      "grad_norm": 0.4629131853580475,
      "learning_rate": 1.9991926385381623e-05,
      "loss": 1.0202,
      "step": 256
    },
    {
      "epoch": 0.04308917531174683,
      "grad_norm": 0.675323486328125,
      "learning_rate": 1.9991706735587006e-05,
      "loss": 1.007,
      "step": 257
    },
    {
      "epoch": 0.043256837472492926,
      "grad_norm": 0.49460679292678833,
      "learning_rate": 1.999148413911435e-05,
      "loss": 0.983,
      "step": 258
    },
    {
      "epoch": 0.04342449963323902,
      "grad_norm": 0.5188019871711731,
      "learning_rate": 1.9991258596029297e-05,
      "loss": 1.1205,
      "step": 259
    },
    {
      "epoch": 0.04359216179398512,
      "grad_norm": 0.5086391568183899,
      "learning_rate": 1.9991030106398367e-05,
      "loss": 1.0457,
      "step": 260
    },
    {
      "epoch": 0.043759823954731214,
      "grad_norm": 0.5028776526451111,
      "learning_rate": 1.999079867028894e-05,
      "loss": 0.9295,
      "step": 261
    },
    {
      "epoch": 0.04392748611547731,
      "grad_norm": 0.6068524718284607,
      "learning_rate": 1.999056428776927e-05,
      "loss": 1.1401,
      "step": 262
    },
    {
      "epoch": 0.04409514827622341,
      "grad_norm": 0.49584200978279114,
      "learning_rate": 1.999032695890848e-05,
      "loss": 0.974,
      "step": 263
    },
    {
      "epoch": 0.04426281043696951,
      "grad_norm": 0.47098278999328613,
      "learning_rate": 1.9990086683776564e-05,
      "loss": 0.9915,
      "step": 264
    },
    {
      "epoch": 0.044430472597715605,
      "grad_norm": 0.6156361699104309,
      "learning_rate": 1.9989843462444377e-05,
      "loss": 1.0524,
      "step": 265
    },
    {
      "epoch": 0.0445981347584617,
      "grad_norm": 0.7538841962814331,
      "learning_rate": 1.998959729498365e-05,
      "loss": 1.0742,
      "step": 266
    },
    {
      "epoch": 0.0447657969192078,
      "grad_norm": 0.6614930629730225,
      "learning_rate": 1.9989348181466987e-05,
      "loss": 0.9725,
      "step": 267
    },
    {
      "epoch": 0.04493345907995389,
      "grad_norm": 0.7805516719818115,
      "learning_rate": 1.9989096121967842e-05,
      "loss": 1.0216,
      "step": 268
    },
    {
      "epoch": 0.04510112124069999,
      "grad_norm": 0.6585594415664673,
      "learning_rate": 1.9988841116560566e-05,
      "loss": 0.9367,
      "step": 269
    },
    {
      "epoch": 0.045268783401446085,
      "grad_norm": 0.5541887879371643,
      "learning_rate": 1.9988583165320347e-05,
      "loss": 1.0547,
      "step": 270
    },
    {
      "epoch": 0.04543644556219218,
      "grad_norm": 0.7834230065345764,
      "learning_rate": 1.998832226832327e-05,
      "loss": 0.9246,
      "step": 271
    },
    {
      "epoch": 0.04560410772293828,
      "grad_norm": 0.5578247308731079,
      "learning_rate": 1.9988058425646268e-05,
      "loss": 0.9464,
      "step": 272
    },
    {
      "epoch": 0.04577176988368437,
      "grad_norm": 0.5382641553878784,
      "learning_rate": 1.9987791637367157e-05,
      "loss": 1.0635,
      "step": 273
    },
    {
      "epoch": 0.045939432044430475,
      "grad_norm": 0.7065907716751099,
      "learning_rate": 1.998752190356462e-05,
      "loss": 0.9919,
      "step": 274
    },
    {
      "epoch": 0.04610709420517657,
      "grad_norm": 0.5382146239280701,
      "learning_rate": 1.99872492243182e-05,
      "loss": 1.0498,
      "step": 275
    },
    {
      "epoch": 0.04627475636592267,
      "grad_norm": 0.5285945534706116,
      "learning_rate": 1.9986973599708305e-05,
      "loss": 0.9463,
      "step": 276
    },
    {
      "epoch": 0.04644241852666876,
      "grad_norm": 0.5281641483306885,
      "learning_rate": 1.9986695029816237e-05,
      "loss": 0.984,
      "step": 277
    },
    {
      "epoch": 0.04661008068741486,
      "grad_norm": 0.5835294723510742,
      "learning_rate": 1.9986413514724136e-05,
      "loss": 1.0335,
      "step": 278
    },
    {
      "epoch": 0.046777742848160955,
      "grad_norm": 0.4561820924282074,
      "learning_rate": 1.9986129054515035e-05,
      "loss": 1.0397,
      "step": 279
    },
    {
      "epoch": 0.04694540500890705,
      "grad_norm": 0.5885711312294006,
      "learning_rate": 1.9985841649272814e-05,
      "loss": 1.0512,
      "step": 280
    },
    {
      "epoch": 0.04711306716965315,
      "grad_norm": 0.441677451133728,
      "learning_rate": 1.998555129908224e-05,
      "loss": 1.0904,
      "step": 281
    },
    {
      "epoch": 0.04728072933039924,
      "grad_norm": 0.4709782302379608,
      "learning_rate": 1.9985258004028938e-05,
      "loss": 1.099,
      "step": 282
    },
    {
      "epoch": 0.04744839149114534,
      "grad_norm": 0.5236688256263733,
      "learning_rate": 1.998496176419941e-05,
      "loss": 0.976,
      "step": 283
    },
    {
      "epoch": 0.04761605365189144,
      "grad_norm": 0.4500700831413269,
      "learning_rate": 1.998466257968101e-05,
      "loss": 0.9947,
      "step": 284
    },
    {
      "epoch": 0.04778371581263754,
      "grad_norm": 0.44306880235671997,
      "learning_rate": 1.998436045056198e-05,
      "loss": 1.0322,
      "step": 285
    },
    {
      "epoch": 0.047951377973383634,
      "grad_norm": 0.44918814301490784,
      "learning_rate": 1.9984055376931414e-05,
      "loss": 0.9553,
      "step": 286
    },
    {
      "epoch": 0.04811904013412973,
      "grad_norm": 0.5018520951271057,
      "learning_rate": 1.998374735887929e-05,
      "loss": 1.0741,
      "step": 287
    },
    {
      "epoch": 0.048286702294875826,
      "grad_norm": 0.4231759011745453,
      "learning_rate": 1.9983436396496442e-05,
      "loss": 1.0914,
      "step": 288
    },
    {
      "epoch": 0.04845436445562192,
      "grad_norm": 0.5387656688690186,
      "learning_rate": 1.9983122489874582e-05,
      "loss": 0.9482,
      "step": 289
    },
    {
      "epoch": 0.04862202661636802,
      "grad_norm": 0.460317999124527,
      "learning_rate": 1.9982805639106276e-05,
      "loss": 1.0508,
      "step": 290
    },
    {
      "epoch": 0.048789688777114114,
      "grad_norm": 0.5162121653556824,
      "learning_rate": 1.9982485844284972e-05,
      "loss": 1.0266,
      "step": 291
    },
    {
      "epoch": 0.04895735093786021,
      "grad_norm": 0.7197742462158203,
      "learning_rate": 1.9982163105504983e-05,
      "loss": 0.9918,
      "step": 292
    },
    {
      "epoch": 0.049125013098606306,
      "grad_norm": 0.6081408262252808,
      "learning_rate": 1.9981837422861487e-05,
      "loss": 0.9567,
      "step": 293
    },
    {
      "epoch": 0.0492926752593524,
      "grad_norm": 0.5756159424781799,
      "learning_rate": 1.9981508796450532e-05,
      "loss": 0.9618,
      "step": 294
    },
    {
      "epoch": 0.049460337420098505,
      "grad_norm": 0.43800628185272217,
      "learning_rate": 1.998117722636903e-05,
      "loss": 1.0086,
      "step": 295
    },
    {
      "epoch": 0.0496279995808446,
      "grad_norm": 0.6097040772438049,
      "learning_rate": 1.9980842712714774e-05,
      "loss": 0.9752,
      "step": 296
    },
    {
      "epoch": 0.0497956617415907,
      "grad_norm": 0.5512593984603882,
      "learning_rate": 1.9980505255586408e-05,
      "loss": 1.0035,
      "step": 297
    },
    {
      "epoch": 0.04996332390233679,
      "grad_norm": 0.6284067034721375,
      "learning_rate": 1.9980164855083454e-05,
      "loss": 0.9961,
      "step": 298
    },
    {
      "epoch": 0.05013098606308289,
      "grad_norm": 0.30609074234962463,
      "learning_rate": 1.9979821511306308e-05,
      "loss": 0.9692,
      "step": 299
    },
    {
      "epoch": 0.050298648223828984,
      "grad_norm": 0.58941650390625,
      "learning_rate": 1.9979475224356214e-05,
      "loss": 1.0258,
      "step": 300
    },
    {
      "epoch": 0.05046631038457508,
      "grad_norm": 0.4312822222709656,
      "learning_rate": 1.9979125994335306e-05,
      "loss": 1.1427,
      "step": 301
    },
    {
      "epoch": 0.050633972545321176,
      "grad_norm": 0.4537133276462555,
      "learning_rate": 1.997877382134657e-05,
      "loss": 0.9259,
      "step": 302
    },
    {
      "epoch": 0.05080163470606727,
      "grad_norm": 0.4248863458633423,
      "learning_rate": 1.9978418705493868e-05,
      "loss": 1.036,
      "step": 303
    },
    {
      "epoch": 0.05096929686681337,
      "grad_norm": 0.5460931658744812,
      "learning_rate": 1.9978060646881933e-05,
      "loss": 1.002,
      "step": 304
    },
    {
      "epoch": 0.051136959027559464,
      "grad_norm": 0.5437471866607666,
      "learning_rate": 1.9977699645616353e-05,
      "loss": 1.0424,
      "step": 305
    },
    {
      "epoch": 0.05130462118830557,
      "grad_norm": 0.6186936497688293,
      "learning_rate": 1.9977335701803597e-05,
      "loss": 0.9514,
      "step": 306
    },
    {
      "epoch": 0.05147228334905166,
      "grad_norm": 0.6375221014022827,
      "learning_rate": 1.9976968815550994e-05,
      "loss": 1.0237,
      "step": 307
    },
    {
      "epoch": 0.05163994550979776,
      "grad_norm": 0.5780304670333862,
      "learning_rate": 1.9976598986966746e-05,
      "loss": 1.0126,
      "step": 308
    },
    {
      "epoch": 0.051807607670543855,
      "grad_norm": 0.47898033261299133,
      "learning_rate": 1.9976226216159916e-05,
      "loss": 1.0855,
      "step": 309
    },
    {
      "epoch": 0.05197526983128995,
      "grad_norm": 0.6292722225189209,
      "learning_rate": 1.9975850503240444e-05,
      "loss": 0.9622,
      "step": 310
    },
    {
      "epoch": 0.05214293199203605,
      "grad_norm": 0.6259632110595703,
      "learning_rate": 1.997547184831913e-05,
      "loss": 0.952,
      "step": 311
    },
    {
      "epoch": 0.05231059415278214,
      "grad_norm": 0.4796934425830841,
      "learning_rate": 1.9975090251507637e-05,
      "loss": 0.9743,
      "step": 312
    },
    {
      "epoch": 0.05247825631352824,
      "grad_norm": 0.4715990424156189,
      "learning_rate": 1.9974705712918515e-05,
      "loss": 0.9663,
      "step": 313
    },
    {
      "epoch": 0.052645918474274335,
      "grad_norm": 0.45795300602912903,
      "learning_rate": 1.9974318232665164e-05,
      "loss": 1.0288,
      "step": 314
    },
    {
      "epoch": 0.05281358063502043,
      "grad_norm": 0.4572800099849701,
      "learning_rate": 1.9973927810861855e-05,
      "loss": 0.9768,
      "step": 315
    },
    {
      "epoch": 0.052981242795766534,
      "grad_norm": 0.6992179751396179,
      "learning_rate": 1.9973534447623727e-05,
      "loss": 1.0511,
      "step": 316
    },
    {
      "epoch": 0.05314890495651263,
      "grad_norm": 0.6290198564529419,
      "learning_rate": 1.9973138143066793e-05,
      "loss": 1.0728,
      "step": 317
    },
    {
      "epoch": 0.053316567117258726,
      "grad_norm": 0.4092872142791748,
      "learning_rate": 1.997273889730792e-05,
      "loss": 0.9257,
      "step": 318
    },
    {
      "epoch": 0.05348422927800482,
      "grad_norm": 0.5989061594009399,
      "learning_rate": 1.997233671046486e-05,
      "loss": 0.9965,
      "step": 319
    },
    {
      "epoch": 0.05365189143875092,
      "grad_norm": 0.70281583070755,
      "learning_rate": 1.997193158265622e-05,
      "loss": 1.0463,
      "step": 320
    },
    {
      "epoch": 0.053819553599497014,
      "grad_norm": 0.6267127990722656,
      "learning_rate": 1.9971523514001473e-05,
      "loss": 1.0258,
      "step": 321
    },
    {
      "epoch": 0.05398721576024311,
      "grad_norm": 0.6525517106056213,
      "learning_rate": 1.9971112504620968e-05,
      "loss": 1.0238,
      "step": 322
    },
    {
      "epoch": 0.054154877920989206,
      "grad_norm": 0.5730092525482178,
      "learning_rate": 1.9970698554635913e-05,
      "loss": 0.9462,
      "step": 323
    },
    {
      "epoch": 0.0543225400817353,
      "grad_norm": 0.5112807750701904,
      "learning_rate": 1.997028166416839e-05,
      "loss": 1.0223,
      "step": 324
    },
    {
      "epoch": 0.0544902022424814,
      "grad_norm": 0.6277716755867004,
      "learning_rate": 1.996986183334134e-05,
      "loss": 0.9783,
      "step": 325
    },
    {
      "epoch": 0.054657864403227494,
      "grad_norm": 0.4657522439956665,
      "learning_rate": 1.9969439062278588e-05,
      "loss": 1.0838,
      "step": 326
    },
    {
      "epoch": 0.054825526563973596,
      "grad_norm": 0.5641884803771973,
      "learning_rate": 1.9969013351104803e-05,
      "loss": 0.9719,
      "step": 327
    },
    {
      "epoch": 0.05499318872471969,
      "grad_norm": 0.6853857636451721,
      "learning_rate": 1.9968584699945534e-05,
      "loss": 1.0499,
      "step": 328
    },
    {
      "epoch": 0.05516085088546579,
      "grad_norm": 0.594561755657196,
      "learning_rate": 1.9968153108927198e-05,
      "loss": 0.9929,
      "step": 329
    },
    {
      "epoch": 0.055328513046211884,
      "grad_norm": 0.43680334091186523,
      "learning_rate": 1.996771857817708e-05,
      "loss": 0.9872,
      "step": 330
    },
    {
      "epoch": 0.05549617520695798,
      "grad_norm": 0.37571197748184204,
      "learning_rate": 1.9967281107823326e-05,
      "loss": 0.951,
      "step": 331
    },
    {
      "epoch": 0.055663837367704076,
      "grad_norm": 0.4594593942165375,
      "learning_rate": 1.9966840697994945e-05,
      "loss": 0.9354,
      "step": 332
    },
    {
      "epoch": 0.05583149952845017,
      "grad_norm": 0.48202845454216003,
      "learning_rate": 1.996639734882183e-05,
      "loss": 0.9478,
      "step": 333
    },
    {
      "epoch": 0.05599916168919627,
      "grad_norm": 0.4156875014305115,
      "learning_rate": 1.996595106043472e-05,
      "loss": 0.962,
      "step": 334
    },
    {
      "epoch": 0.056166823849942364,
      "grad_norm": 0.5083615779876709,
      "learning_rate": 1.9965501832965238e-05,
      "loss": 1.0278,
      "step": 335
    },
    {
      "epoch": 0.05633448601068846,
      "grad_norm": 0.5085647702217102,
      "learning_rate": 1.9965049666545863e-05,
      "loss": 0.9764,
      "step": 336
    },
    {
      "epoch": 0.056502148171434556,
      "grad_norm": 0.5676213502883911,
      "learning_rate": 1.9964594561309946e-05,
      "loss": 1.0438,
      "step": 337
    },
    {
      "epoch": 0.05666981033218066,
      "grad_norm": 0.5562122464179993,
      "learning_rate": 1.9964136517391708e-05,
      "loss": 0.9988,
      "step": 338
    },
    {
      "epoch": 0.056837472492926755,
      "grad_norm": 0.5095664858818054,
      "learning_rate": 1.9963675534926224e-05,
      "loss": 1.0729,
      "step": 339
    },
    {
      "epoch": 0.05700513465367285,
      "grad_norm": 0.5423064231872559,
      "learning_rate": 1.9963211614049443e-05,
      "loss": 1.0693,
      "step": 340
    },
    {
      "epoch": 0.05717279681441895,
      "grad_norm": 0.666354775428772,
      "learning_rate": 1.996274475489819e-05,
      "loss": 0.9466,
      "step": 341
    },
    {
      "epoch": 0.05734045897516504,
      "grad_norm": 0.5196070075035095,
      "learning_rate": 1.9962274957610136e-05,
      "loss": 0.9676,
      "step": 342
    },
    {
      "epoch": 0.05750812113591114,
      "grad_norm": 0.6126438975334167,
      "learning_rate": 1.996180222232384e-05,
      "loss": 0.9033,
      "step": 343
    },
    {
      "epoch": 0.057675783296657235,
      "grad_norm": 0.5262427926063538,
      "learning_rate": 1.9961326549178713e-05,
      "loss": 0.9979,
      "step": 344
    },
    {
      "epoch": 0.05784344545740333,
      "grad_norm": 0.5726452469825745,
      "learning_rate": 1.9960847938315037e-05,
      "loss": 0.9771,
      "step": 345
    },
    {
      "epoch": 0.05801110761814943,
      "grad_norm": 0.5337097644805908,
      "learning_rate": 1.996036638987396e-05,
      "loss": 0.9814,
      "step": 346
    },
    {
      "epoch": 0.05817876977889552,
      "grad_norm": 0.6877371072769165,
      "learning_rate": 1.99598819039975e-05,
      "loss": 1.1398,
      "step": 347
    },
    {
      "epoch": 0.05834643193964162,
      "grad_norm": 0.6705992221832275,
      "learning_rate": 1.995939448082853e-05,
      "loss": 1.0732,
      "step": 348
    },
    {
      "epoch": 0.05851409410038772,
      "grad_norm": 0.7965999841690063,
      "learning_rate": 1.9958904120510807e-05,
      "loss": 1.0088,
      "step": 349
    },
    {
      "epoch": 0.05868175626113382,
      "grad_norm": 0.6121879816055298,
      "learning_rate": 1.995841082318894e-05,
      "loss": 1.0457,
      "step": 350
    },
    {
      "epoch": 0.058849418421879914,
      "grad_norm": 0.892564594745636,
      "learning_rate": 1.9957914589008405e-05,
      "loss": 0.9857,
      "step": 351
    },
    {
      "epoch": 0.05901708058262601,
      "grad_norm": 0.4690881669521332,
      "learning_rate": 1.9957415418115552e-05,
      "loss": 0.9926,
      "step": 352
    },
    {
      "epoch": 0.059184742743372105,
      "grad_norm": 0.8930359482765198,
      "learning_rate": 1.995691331065759e-05,
      "loss": 1.0861,
      "step": 353
    },
    {
      "epoch": 0.0593524049041182,
      "grad_norm": 0.6299958825111389,
      "learning_rate": 1.99564082667826e-05,
      "loss": 1.0111,
      "step": 354
    },
    {
      "epoch": 0.0595200670648643,
      "grad_norm": 0.6326963901519775,
      "learning_rate": 1.9955900286639524e-05,
      "loss": 1.0335,
      "step": 355
    },
    {
      "epoch": 0.05968772922561039,
      "grad_norm": 0.5321698188781738,
      "learning_rate": 1.9955389370378172e-05,
      "loss": 0.9373,
      "step": 356
    },
    {
      "epoch": 0.05985539138635649,
      "grad_norm": 0.42701229453086853,
      "learning_rate": 1.9954875518149216e-05,
      "loss": 1.0024,
      "step": 357
    },
    {
      "epoch": 0.060023053547102585,
      "grad_norm": 0.7688174247741699,
      "learning_rate": 1.99543587301042e-05,
      "loss": 1.0012,
      "step": 358
    },
    {
      "epoch": 0.06019071570784869,
      "grad_norm": 0.47423213720321655,
      "learning_rate": 1.9953839006395534e-05,
      "loss": 0.9813,
      "step": 359
    },
    {
      "epoch": 0.060358377868594784,
      "grad_norm": 0.48588648438453674,
      "learning_rate": 1.995331634717649e-05,
      "loss": 1.0102,
      "step": 360
    },
    {
      "epoch": 0.06052604002934088,
      "grad_norm": 0.586348295211792,
      "learning_rate": 1.99527907526012e-05,
      "loss": 1.0298,
      "step": 361
    },
    {
      "epoch": 0.060693702190086976,
      "grad_norm": 0.6227160692214966,
      "learning_rate": 1.995226222282468e-05,
      "loss": 1.0071,
      "step": 362
    },
    {
      "epoch": 0.06086136435083307,
      "grad_norm": 0.6064775586128235,
      "learning_rate": 1.9951730758002787e-05,
      "loss": 1.014,
      "step": 363
    },
    {
      "epoch": 0.06102902651157917,
      "grad_norm": 0.6879925727844238,
      "learning_rate": 1.9951196358292266e-05,
      "loss": 1.0248,
      "step": 364
    },
    {
      "epoch": 0.061196688672325264,
      "grad_norm": 0.4487518072128296,
      "learning_rate": 1.9950659023850716e-05,
      "loss": 1.0143,
      "step": 365
    },
    {
      "epoch": 0.06136435083307136,
      "grad_norm": 0.4098510444164276,
      "learning_rate": 1.99501187548366e-05,
      "loss": 0.9518,
      "step": 366
    },
    {
      "epoch": 0.061532012993817456,
      "grad_norm": 0.49687305092811584,
      "learning_rate": 1.9949575551409258e-05,
      "loss": 0.9737,
      "step": 367
    },
    {
      "epoch": 0.06169967515456355,
      "grad_norm": 0.43834513425827026,
      "learning_rate": 1.9949029413728877e-05,
      "loss": 0.9448,
      "step": 368
    },
    {
      "epoch": 0.06186733731530965,
      "grad_norm": 0.4889039993286133,
      "learning_rate": 1.9948480341956524e-05,
      "loss": 1.0808,
      "step": 369
    },
    {
      "epoch": 0.06203499947605575,
      "grad_norm": 0.4841388165950775,
      "learning_rate": 1.9947928336254132e-05,
      "loss": 1.0362,
      "step": 370
    },
    {
      "epoch": 0.06220266163680185,
      "grad_norm": 0.6277660131454468,
      "learning_rate": 1.9947373396784487e-05,
      "loss": 1.0221,
      "step": 371
    },
    {
      "epoch": 0.06237032379754794,
      "grad_norm": 0.5336534380912781,
      "learning_rate": 1.9946815523711253e-05,
      "loss": 1.0528,
      "step": 372
    },
    {
      "epoch": 0.06253798595829403,
      "grad_norm": 0.999744713306427,
      "learning_rate": 1.9946254717198954e-05,
      "loss": 1.1037,
      "step": 373
    },
    {
      "epoch": 0.06270564811904013,
      "grad_norm": 0.6568665504455566,
      "learning_rate": 1.9945690977412972e-05,
      "loss": 1.0035,
      "step": 374
    },
    {
      "epoch": 0.06287331027978624,
      "grad_norm": 0.4865848422050476,
      "learning_rate": 1.9945124304519568e-05,
      "loss": 1.0624,
      "step": 375
    },
    {
      "epoch": 0.06304097244053233,
      "grad_norm": 0.4973253607749939,
      "learning_rate": 1.9944554698685853e-05,
      "loss": 0.924,
      "step": 376
    },
    {
      "epoch": 0.06320863460127843,
      "grad_norm": 0.7405445575714111,
      "learning_rate": 1.9943982160079823e-05,
      "loss": 1.013,
      "step": 377
    },
    {
      "epoch": 0.06337629676202453,
      "grad_norm": 0.5829219222068787,
      "learning_rate": 1.9943406688870316e-05,
      "loss": 0.9768,
      "step": 378
    },
    {
      "epoch": 0.06354395892277062,
      "grad_norm": 0.4399314820766449,
      "learning_rate": 1.9942828285227047e-05,
      "loss": 1.0749,
      "step": 379
    },
    {
      "epoch": 0.06371162108351672,
      "grad_norm": 0.47130849957466125,
      "learning_rate": 1.9942246949320598e-05,
      "loss": 0.917,
      "step": 380
    },
    {
      "epoch": 0.06387928324426281,
      "grad_norm": 0.46881669759750366,
      "learning_rate": 1.9941662681322412e-05,
      "loss": 0.9698,
      "step": 381
    },
    {
      "epoch": 0.06404694540500891,
      "grad_norm": 0.4304395318031311,
      "learning_rate": 1.9941075481404795e-05,
      "loss": 1.0417,
      "step": 382
    },
    {
      "epoch": 0.064214607565755,
      "grad_norm": 0.6960583925247192,
      "learning_rate": 1.994048534974092e-05,
      "loss": 1.0315,
      "step": 383
    },
    {
      "epoch": 0.0643822697265011,
      "grad_norm": 0.6453326940536499,
      "learning_rate": 1.9939892286504822e-05,
      "loss": 0.9886,
      "step": 384
    },
    {
      "epoch": 0.0645499318872472,
      "grad_norm": 0.4545916020870209,
      "learning_rate": 1.993929629187141e-05,
      "loss": 0.9806,
      "step": 385
    },
    {
      "epoch": 0.0647175940479933,
      "grad_norm": 0.38227584958076477,
      "learning_rate": 1.993869736601644e-05,
      "loss": 0.9928,
      "step": 386
    },
    {
      "epoch": 0.06488525620873939,
      "grad_norm": 0.48465222120285034,
      "learning_rate": 1.9938095509116547e-05,
      "loss": 1.1353,
      "step": 387
    },
    {
      "epoch": 0.06505291836948549,
      "grad_norm": 0.4780482351779938,
      "learning_rate": 1.9937490721349232e-05,
      "loss": 0.9548,
      "step": 388
    },
    {
      "epoch": 0.06522058053023158,
      "grad_norm": 0.5209558010101318,
      "learning_rate": 1.9936883002892843e-05,
      "loss": 1.0472,
      "step": 389
    },
    {
      "epoch": 0.06538824269097768,
      "grad_norm": 0.48167288303375244,
      "learning_rate": 1.9936272353926616e-05,
      "loss": 1.0512,
      "step": 390
    },
    {
      "epoch": 0.06555590485172377,
      "grad_norm": 0.4132109582424164,
      "learning_rate": 1.9935658774630632e-05,
      "loss": 0.9558,
      "step": 391
    },
    {
      "epoch": 0.06572356701246987,
      "grad_norm": 0.562962532043457,
      "learning_rate": 1.993504226518584e-05,
      "loss": 0.9254,
      "step": 392
    },
    {
      "epoch": 0.06589122917321596,
      "grad_norm": 0.4468546211719513,
      "learning_rate": 1.9934422825774064e-05,
      "loss": 0.9164,
      "step": 393
    },
    {
      "epoch": 0.06605889133396206,
      "grad_norm": 0.4178194999694824,
      "learning_rate": 1.9933800456577978e-05,
      "loss": 0.9262,
      "step": 394
    },
    {
      "epoch": 0.06622655349470816,
      "grad_norm": 0.7586342096328735,
      "learning_rate": 1.993317515778113e-05,
      "loss": 0.9454,
      "step": 395
    },
    {
      "epoch": 0.06639421565545425,
      "grad_norm": 0.5593826770782471,
      "learning_rate": 1.9932546929567924e-05,
      "loss": 1.0024,
      "step": 396
    },
    {
      "epoch": 0.06656187781620036,
      "grad_norm": 0.4449445903301239,
      "learning_rate": 1.993191577212364e-05,
      "loss": 0.9994,
      "step": 397
    },
    {
      "epoch": 0.06672953997694646,
      "grad_norm": 0.5252183675765991,
      "learning_rate": 1.993128168563441e-05,
      "loss": 1.0497,
      "step": 398
    },
    {
      "epoch": 0.06689720213769255,
      "grad_norm": 0.45754343271255493,
      "learning_rate": 1.9930644670287233e-05,
      "loss": 1.0029,
      "step": 399
    },
    {
      "epoch": 0.06706486429843865,
      "grad_norm": 0.4065295159816742,
      "learning_rate": 1.993000472626997e-05,
      "loss": 0.8735,
      "step": 400
    },
    {
      "epoch": 0.06723252645918475,
      "grad_norm": 0.4539814591407776,
      "learning_rate": 1.9929361853771355e-05,
      "loss": 1.017,
      "step": 401
    },
    {
      "epoch": 0.06740018861993084,
      "grad_norm": 0.46643343567848206,
      "learning_rate": 1.9928716052980977e-05,
      "loss": 1.0082,
      "step": 402
    },
    {
      "epoch": 0.06756785078067694,
      "grad_norm": 0.48987117409706116,
      "learning_rate": 1.9928067324089286e-05,
      "loss": 0.9141,
      "step": 403
    },
    {
      "epoch": 0.06773551294142303,
      "grad_norm": 0.3898268938064575,
      "learning_rate": 1.9927415667287605e-05,
      "loss": 1.0585,
      "step": 404
    },
    {
      "epoch": 0.06790317510216913,
      "grad_norm": 0.5273199081420898,
      "learning_rate": 1.9926761082768116e-05,
      "loss": 1.0478,
      "step": 405
    },
    {
      "epoch": 0.06807083726291523,
      "grad_norm": 0.5008173584938049,
      "learning_rate": 1.992610357072386e-05,
      "loss": 1.0472,
      "step": 406
    },
    {
      "epoch": 0.06823849942366132,
      "grad_norm": 0.42762380838394165,
      "learning_rate": 1.992544313134875e-05,
      "loss": 0.9769,
      "step": 407
    },
    {
      "epoch": 0.06840616158440742,
      "grad_norm": 0.43059566617012024,
      "learning_rate": 1.9924779764837553e-05,
      "loss": 1.0213,
      "step": 408
    },
    {
      "epoch": 0.06857382374515351,
      "grad_norm": 0.485988050699234,
      "learning_rate": 1.992411347138591e-05,
      "loss": 1.0403,
      "step": 409
    },
    {
      "epoch": 0.06874148590589961,
      "grad_norm": 0.5233361721038818,
      "learning_rate": 1.9923444251190314e-05,
      "loss": 0.9312,
      "step": 410
    },
    {
      "epoch": 0.0689091480666457,
      "grad_norm": 0.7353261709213257,
      "learning_rate": 1.992277210444813e-05,
      "loss": 1.0113,
      "step": 411
    },
    {
      "epoch": 0.0690768102273918,
      "grad_norm": 0.4786561131477356,
      "learning_rate": 1.9922097031357575e-05,
      "loss": 0.8914,
      "step": 412
    },
    {
      "epoch": 0.0692444723881379,
      "grad_norm": 0.6206673383712769,
      "learning_rate": 1.9921419032117752e-05,
      "loss": 1.0199,
      "step": 413
    },
    {
      "epoch": 0.069412134548884,
      "grad_norm": 0.5633043646812439,
      "learning_rate": 1.9920738106928593e-05,
      "loss": 1.0299,
      "step": 414
    },
    {
      "epoch": 0.06957979670963009,
      "grad_norm": 0.6748209595680237,
      "learning_rate": 1.992005425599092e-05,
      "loss": 0.9454,
      "step": 415
    },
    {
      "epoch": 0.06974745887037619,
      "grad_norm": 0.6763722896575928,
      "learning_rate": 1.9919367479506413e-05,
      "loss": 1.1172,
      "step": 416
    },
    {
      "epoch": 0.06991512103112228,
      "grad_norm": 0.6160325407981873,
      "learning_rate": 1.9918677777677605e-05,
      "loss": 1.0262,
      "step": 417
    },
    {
      "epoch": 0.07008278319186839,
      "grad_norm": 0.6006767749786377,
      "learning_rate": 1.99179851507079e-05,
      "loss": 1.0134,
      "step": 418
    },
    {
      "epoch": 0.07025044535261449,
      "grad_norm": 0.49446508288383484,
      "learning_rate": 1.9917289598801556e-05,
      "loss": 1.0113,
      "step": 419
    },
    {
      "epoch": 0.07041810751336058,
      "grad_norm": 0.44796285033226013,
      "learning_rate": 1.991659112216371e-05,
      "loss": 0.9279,
      "step": 420
    },
    {
      "epoch": 0.07058576967410668,
      "grad_norm": 0.4072938561439514,
      "learning_rate": 1.9915889721000347e-05,
      "loss": 0.9871,
      "step": 421
    },
    {
      "epoch": 0.07075343183485278,
      "grad_norm": 0.5549907088279724,
      "learning_rate": 1.991518539551832e-05,
      "loss": 0.9974,
      "step": 422
    },
    {
      "epoch": 0.07092109399559887,
      "grad_norm": 0.4595836102962494,
      "learning_rate": 1.991447814592534e-05,
      "loss": 1.0345,
      "step": 423
    },
    {
      "epoch": 0.07108875615634497,
      "grad_norm": 0.422370582818985,
      "learning_rate": 1.9913767972429984e-05,
      "loss": 0.9588,
      "step": 424
    },
    {
      "epoch": 0.07125641831709106,
      "grad_norm": 0.4242878258228302,
      "learning_rate": 1.9913054875241694e-05,
      "loss": 0.9386,
      "step": 425
    },
    {
      "epoch": 0.07142408047783716,
      "grad_norm": 0.5106265544891357,
      "learning_rate": 1.991233885457077e-05,
      "loss": 0.9666,
      "step": 426
    },
    {
      "epoch": 0.07159174263858326,
      "grad_norm": 0.4680672585964203,
      "learning_rate": 1.9911619910628374e-05,
      "loss": 0.9116,
      "step": 427
    },
    {
      "epoch": 0.07175940479932935,
      "grad_norm": 0.5930494666099548,
      "learning_rate": 1.9910898043626534e-05,
      "loss": 1.035,
      "step": 428
    },
    {
      "epoch": 0.07192706696007545,
      "grad_norm": 0.7918307781219482,
      "learning_rate": 1.9910173253778136e-05,
      "loss": 1.0634,
      "step": 429
    },
    {
      "epoch": 0.07209472912082154,
      "grad_norm": 0.4641052186489105,
      "learning_rate": 1.990944554129693e-05,
      "loss": 0.834,
      "step": 430
    },
    {
      "epoch": 0.07226239128156764,
      "grad_norm": 0.46474114060401917,
      "learning_rate": 1.9908714906397525e-05,
      "loss": 0.8806,
      "step": 431
    },
    {
      "epoch": 0.07243005344231374,
      "grad_norm": 0.4513774812221527,
      "learning_rate": 1.99079813492954e-05,
      "loss": 0.9827,
      "step": 432
    },
    {
      "epoch": 0.07259771560305983,
      "grad_norm": 0.45238038897514343,
      "learning_rate": 1.990724487020688e-05,
      "loss": 0.9823,
      "step": 433
    },
    {
      "epoch": 0.07276537776380593,
      "grad_norm": 0.44659870862960815,
      "learning_rate": 1.9906505469349176e-05,
      "loss": 0.8166,
      "step": 434
    },
    {
      "epoch": 0.07293303992455202,
      "grad_norm": 0.4721660912036896,
      "learning_rate": 1.9905763146940332e-05,
      "loss": 0.947,
      "step": 435
    },
    {
      "epoch": 0.07310070208529812,
      "grad_norm": 0.38887742161750793,
      "learning_rate": 1.990501790319928e-05,
      "loss": 0.935,
      "step": 436
    },
    {
      "epoch": 0.07326836424604422,
      "grad_norm": 0.472125381231308,
      "learning_rate": 1.9904269738345793e-05,
      "loss": 1.0113,
      "step": 437
    },
    {
      "epoch": 0.07343602640679031,
      "grad_norm": 0.4801134467124939,
      "learning_rate": 1.9903518652600518e-05,
      "loss": 0.9506,
      "step": 438
    },
    {
      "epoch": 0.07360368856753641,
      "grad_norm": 0.47314056754112244,
      "learning_rate": 1.990276464618496e-05,
      "loss": 0.916,
      "step": 439
    },
    {
      "epoch": 0.07377135072828252,
      "grad_norm": 0.5533985495567322,
      "learning_rate": 1.9902007719321483e-05,
      "loss": 1.0233,
      "step": 440
    },
    {
      "epoch": 0.07393901288902861,
      "grad_norm": 0.47966158390045166,
      "learning_rate": 1.9901247872233316e-05,
      "loss": 0.9955,
      "step": 441
    },
    {
      "epoch": 0.07410667504977471,
      "grad_norm": 0.5108001232147217,
      "learning_rate": 1.9900485105144544e-05,
      "loss": 0.9569,
      "step": 442
    },
    {
      "epoch": 0.0742743372105208,
      "grad_norm": 0.5768658518791199,
      "learning_rate": 1.989971941828012e-05,
      "loss": 0.9923,
      "step": 443
    },
    {
      "epoch": 0.0744419993712669,
      "grad_norm": 0.5834493041038513,
      "learning_rate": 1.9898950811865854e-05,
      "loss": 0.9778,
      "step": 444
    },
    {
      "epoch": 0.074609661532013,
      "grad_norm": 0.6800981760025024,
      "learning_rate": 1.9898179286128415e-05,
      "loss": 1.0887,
      "step": 445
    },
    {
      "epoch": 0.07477732369275909,
      "grad_norm": 0.4957737624645233,
      "learning_rate": 1.9897404841295337e-05,
      "loss": 0.959,
      "step": 446
    },
    {
      "epoch": 0.07494498585350519,
      "grad_norm": 0.4654386341571808,
      "learning_rate": 1.9896627477595017e-05,
      "loss": 1.0396,
      "step": 447
    },
    {
      "epoch": 0.07511264801425128,
      "grad_norm": 0.4682818353176117,
      "learning_rate": 1.9895847195256702e-05,
      "loss": 0.9909,
      "step": 448
    },
    {
      "epoch": 0.07528031017499738,
      "grad_norm": 0.4938904643058777,
      "learning_rate": 1.9895063994510512e-05,
      "loss": 1.0329,
      "step": 449
    },
    {
      "epoch": 0.07544797233574348,
      "grad_norm": 0.5164246559143066,
      "learning_rate": 1.989427787558742e-05,
      "loss": 1.005,
      "step": 450
    },
    {
      "epoch": 0.07561563449648957,
      "grad_norm": 0.7106466889381409,
      "learning_rate": 1.9893488838719267e-05,
      "loss": 1.0256,
      "step": 451
    },
    {
      "epoch": 0.07578329665723567,
      "grad_norm": 0.6531050205230713,
      "learning_rate": 1.9892696884138743e-05,
      "loss": 1.0237,
      "step": 452
    },
    {
      "epoch": 0.07595095881798176,
      "grad_norm": 0.7170326709747314,
      "learning_rate": 1.989190201207941e-05,
      "loss": 0.9953,
      "step": 453
    },
    {
      "epoch": 0.07611862097872786,
      "grad_norm": 0.5520920753479004,
      "learning_rate": 1.9891104222775683e-05,
      "loss": 1.0664,
      "step": 454
    },
    {
      "epoch": 0.07628628313947396,
      "grad_norm": 0.5298479795455933,
      "learning_rate": 1.9890303516462842e-05,
      "loss": 0.867,
      "step": 455
    },
    {
      "epoch": 0.07645394530022005,
      "grad_norm": 0.7908779978752136,
      "learning_rate": 1.9889499893377024e-05,
      "loss": 0.9758,
      "step": 456
    },
    {
      "epoch": 0.07662160746096615,
      "grad_norm": 0.649056613445282,
      "learning_rate": 1.988869335375523e-05,
      "loss": 1.0339,
      "step": 457
    },
    {
      "epoch": 0.07678926962171224,
      "grad_norm": 0.6440086960792542,
      "learning_rate": 1.988788389783531e-05,
      "loss": 0.8257,
      "step": 458
    },
    {
      "epoch": 0.07695693178245834,
      "grad_norm": 0.498965322971344,
      "learning_rate": 1.9887071525855993e-05,
      "loss": 1.0239,
      "step": 459
    },
    {
      "epoch": 0.07712459394320444,
      "grad_norm": 0.6430321931838989,
      "learning_rate": 1.9886256238056854e-05,
      "loss": 0.9629,
      "step": 460
    },
    {
      "epoch": 0.07729225610395055,
      "grad_norm": 0.5625649094581604,
      "learning_rate": 1.9885438034678327e-05,
      "loss": 0.915,
      "step": 461
    },
    {
      "epoch": 0.07745991826469664,
      "grad_norm": 0.58819580078125,
      "learning_rate": 1.9884616915961717e-05,
      "loss": 1.0797,
      "step": 462
    },
    {
      "epoch": 0.07762758042544274,
      "grad_norm": 0.6026577949523926,
      "learning_rate": 1.9883792882149177e-05,
      "loss": 0.8432,
      "step": 463
    },
    {
      "epoch": 0.07779524258618883,
      "grad_norm": 0.5222564339637756,
      "learning_rate": 1.988296593348373e-05,
      "loss": 0.9467,
      "step": 464
    },
    {
      "epoch": 0.07796290474693493,
      "grad_norm": 0.616072952747345,
      "learning_rate": 1.988213607020925e-05,
      "loss": 1.0728,
      "step": 465
    },
    {
      "epoch": 0.07813056690768103,
      "grad_norm": 0.4086941182613373,
      "learning_rate": 1.988130329257047e-05,
      "loss": 1.0023,
      "step": 466
    },
    {
      "epoch": 0.07829822906842712,
      "grad_norm": 0.60639888048172,
      "learning_rate": 1.9880467600812992e-05,
      "loss": 0.9698,
      "step": 467
    },
    {
      "epoch": 0.07846589122917322,
      "grad_norm": 0.6214800477027893,
      "learning_rate": 1.9879628995183274e-05,
      "loss": 1.0132,
      "step": 468
    },
    {
      "epoch": 0.07863355338991931,
      "grad_norm": 0.4865536689758301,
      "learning_rate": 1.9878787475928624e-05,
      "loss": 0.8728,
      "step": 469
    },
    {
      "epoch": 0.07880121555066541,
      "grad_norm": 0.6627867221832275,
      "learning_rate": 1.9877943043297223e-05,
      "loss": 1.0299,
      "step": 470
    },
    {
      "epoch": 0.0789688777114115,
      "grad_norm": 0.5963670611381531,
      "learning_rate": 1.98770956975381e-05,
      "loss": 1.0561,
      "step": 471
    },
    {
      "epoch": 0.0791365398721576,
      "grad_norm": 0.48536255955696106,
      "learning_rate": 1.987624543890115e-05,
      "loss": 1.0183,
      "step": 472
    },
    {
      "epoch": 0.0793042020329037,
      "grad_norm": 0.4555172622203827,
      "learning_rate": 1.9875392267637123e-05,
      "loss": 0.9699,
      "step": 473
    },
    {
      "epoch": 0.0794718641936498,
      "grad_norm": 0.5513790845870972,
      "learning_rate": 1.9874536183997636e-05,
      "loss": 1.035,
      "step": 474
    },
    {
      "epoch": 0.07963952635439589,
      "grad_norm": 0.5619701743125916,
      "learning_rate": 1.987367718823515e-05,
      "loss": 1.0642,
      "step": 475
    },
    {
      "epoch": 0.07980718851514199,
      "grad_norm": 0.43956467509269714,
      "learning_rate": 1.9872815280602996e-05,
      "loss": 1.0078,
      "step": 476
    },
    {
      "epoch": 0.07997485067588808,
      "grad_norm": 0.40225303173065186,
      "learning_rate": 1.9871950461355365e-05,
      "loss": 0.9142,
      "step": 477
    },
    {
      "epoch": 0.08014251283663418,
      "grad_norm": 0.49438419938087463,
      "learning_rate": 1.9871082730747298e-05,
      "loss": 0.9941,
      "step": 478
    },
    {
      "epoch": 0.08031017499738027,
      "grad_norm": 0.6986750364303589,
      "learning_rate": 1.9870212089034707e-05,
      "loss": 0.9752,
      "step": 479
    },
    {
      "epoch": 0.08047783715812637,
      "grad_norm": 0.5081155300140381,
      "learning_rate": 1.9869338536474345e-05,
      "loss": 0.9865,
      "step": 480
    },
    {
      "epoch": 0.08064549931887247,
      "grad_norm": 0.5153716802597046,
      "learning_rate": 1.986846207332384e-05,
      "loss": 1.1058,
      "step": 481
    },
    {
      "epoch": 0.08081316147961858,
      "grad_norm": 0.5596706867218018,
      "learning_rate": 1.986758269984167e-05,
      "loss": 0.9267,
      "step": 482
    },
    {
      "epoch": 0.08098082364036467,
      "grad_norm": 0.486183762550354,
      "learning_rate": 1.9866700416287174e-05,
      "loss": 1.0091,
      "step": 483
    },
    {
      "epoch": 0.08114848580111077,
      "grad_norm": 0.5393663644790649,
      "learning_rate": 1.986581522292055e-05,
      "loss": 1.0414,
      "step": 484
    },
    {
      "epoch": 0.08131614796185686,
      "grad_norm": 0.46591389179229736,
      "learning_rate": 1.9864927120002845e-05,
      "loss": 0.882,
      "step": 485
    },
    {
      "epoch": 0.08148381012260296,
      "grad_norm": 0.5534307360649109,
      "learning_rate": 1.9864036107795978e-05,
      "loss": 0.8958,
      "step": 486
    },
    {
      "epoch": 0.08165147228334906,
      "grad_norm": 0.4817509353160858,
      "learning_rate": 1.986314218656272e-05,
      "loss": 0.9899,
      "step": 487
    },
    {
      "epoch": 0.08181913444409515,
      "grad_norm": 0.4864092767238617,
      "learning_rate": 1.98622453565667e-05,
      "loss": 0.9151,
      "step": 488
    },
    {
      "epoch": 0.08198679660484125,
      "grad_norm": 0.581265389919281,
      "learning_rate": 1.9861345618072395e-05,
      "loss": 0.9738,
      "step": 489
    },
    {
      "epoch": 0.08215445876558734,
      "grad_norm": 0.37302395701408386,
      "learning_rate": 1.986044297134516e-05,
      "loss": 0.9457,
      "step": 490
    },
    {
      "epoch": 0.08232212092633344,
      "grad_norm": 0.5029395222663879,
      "learning_rate": 1.9859537416651197e-05,
      "loss": 0.9713,
      "step": 491
    },
    {
      "epoch": 0.08248978308707954,
      "grad_norm": 0.7264549136161804,
      "learning_rate": 1.9858628954257554e-05,
      "loss": 1.0336,
      "step": 492
    },
    {
      "epoch": 0.08265744524782563,
      "grad_norm": 0.5837662816047668,
      "learning_rate": 1.985771758443216e-05,
      "loss": 0.9516,
      "step": 493
    },
    {
      "epoch": 0.08282510740857173,
      "grad_norm": 0.4827551245689392,
      "learning_rate": 1.9856803307443782e-05,
      "loss": 0.9188,
      "step": 494
    },
    {
      "epoch": 0.08299276956931782,
      "grad_norm": 0.5622090101242065,
      "learning_rate": 1.9855886123562055e-05,
      "loss": 0.9516,
      "step": 495
    },
    {
      "epoch": 0.08316043173006392,
      "grad_norm": 0.5088374614715576,
      "learning_rate": 1.9854966033057462e-05,
      "loss": 0.9087,
      "step": 496
    },
    {
      "epoch": 0.08332809389081001,
      "grad_norm": 0.5393046140670776,
      "learning_rate": 1.985404303620136e-05,
      "loss": 1.0514,
      "step": 497
    },
    {
      "epoch": 0.08349575605155611,
      "grad_norm": 0.6230491995811462,
      "learning_rate": 1.985311713326594e-05,
      "loss": 1.0803,
      "step": 498
    },
    {
      "epoch": 0.0836634182123022,
      "grad_norm": 0.47609272599220276,
      "learning_rate": 1.9852188324524272e-05,
      "loss": 0.8345,
      "step": 499
    },
    {
      "epoch": 0.0838310803730483,
      "grad_norm": 0.6452209949493408,
      "learning_rate": 1.9851256610250268e-05,
      "loss": 0.9531,
      "step": 500
    },
    {
      "epoch": 0.0839987425337944,
      "grad_norm": 0.4989963173866272,
      "learning_rate": 1.9850321990718703e-05,
      "loss": 1.0362,
      "step": 501
    },
    {
      "epoch": 0.0841664046945405,
      "grad_norm": 0.5229847431182861,
      "learning_rate": 1.984938446620521e-05,
      "loss": 1.0242,
      "step": 502
    },
    {
      "epoch": 0.08433406685528659,
      "grad_norm": 0.4766591489315033,
      "learning_rate": 1.984844403698627e-05,
      "loss": 1.058,
      "step": 503
    },
    {
      "epoch": 0.0845017290160327,
      "grad_norm": 0.4473041594028473,
      "learning_rate": 1.9847500703339238e-05,
      "loss": 1.0586,
      "step": 504
    },
    {
      "epoch": 0.0846693911767788,
      "grad_norm": 0.45140060782432556,
      "learning_rate": 1.9846554465542306e-05,
      "loss": 0.9494,
      "step": 505
    },
    {
      "epoch": 0.08483705333752489,
      "grad_norm": 0.44374144077301025,
      "learning_rate": 1.9845605323874532e-05,
      "loss": 0.947,
      "step": 506
    },
    {
      "epoch": 0.08500471549827099,
      "grad_norm": 0.41401800513267517,
      "learning_rate": 1.9844653278615836e-05,
      "loss": 0.9302,
      "step": 507
    },
    {
      "epoch": 0.08517237765901708,
      "grad_norm": 0.5847159624099731,
      "learning_rate": 1.984369833004698e-05,
      "loss": 0.9366,
      "step": 508
    },
    {
      "epoch": 0.08534003981976318,
      "grad_norm": 0.547697126865387,
      "learning_rate": 1.9842740478449594e-05,
      "loss": 1.0562,
      "step": 509
    },
    {
      "epoch": 0.08550770198050928,
      "grad_norm": 0.5329285860061646,
      "learning_rate": 1.9841779724106165e-05,
      "loss": 0.9999,
      "step": 510
    },
    {
      "epoch": 0.08567536414125537,
      "grad_norm": 0.5062839984893799,
      "learning_rate": 1.9840816067300022e-05,
      "loss": 1.045,
      "step": 511
    },
    {
      "epoch": 0.08584302630200147,
      "grad_norm": 0.457762748003006,
      "learning_rate": 1.9839849508315365e-05,
      "loss": 1.0371,
      "step": 512
    },
    {
      "epoch": 0.08601068846274756,
      "grad_norm": 0.5547719597816467,
      "learning_rate": 1.9838880047437243e-05,
      "loss": 1.0181,
      "step": 513
    },
    {
      "epoch": 0.08617835062349366,
      "grad_norm": 0.5168294310569763,
      "learning_rate": 1.983790768495156e-05,
      "loss": 0.961,
      "step": 514
    },
    {
      "epoch": 0.08634601278423976,
      "grad_norm": 0.6367170214653015,
      "learning_rate": 1.9836932421145084e-05,
      "loss": 0.999,
      "step": 515
    },
    {
      "epoch": 0.08651367494498585,
      "grad_norm": 0.6939343214035034,
      "learning_rate": 1.9835954256305423e-05,
      "loss": 0.9431,
      "step": 516
    },
    {
      "epoch": 0.08668133710573195,
      "grad_norm": 0.503060519695282,
      "learning_rate": 1.9834973190721057e-05,
      "loss": 0.877,
      "step": 517
    },
    {
      "epoch": 0.08684899926647804,
      "grad_norm": 0.6143543124198914,
      "learning_rate": 1.9833989224681314e-05,
      "loss": 1.016,
      "step": 518
    },
    {
      "epoch": 0.08701666142722414,
      "grad_norm": 0.4443577826023102,
      "learning_rate": 1.9833002358476375e-05,
      "loss": 0.9724,
      "step": 519
    },
    {
      "epoch": 0.08718432358797024,
      "grad_norm": 0.5362754464149475,
      "learning_rate": 1.983201259239728e-05,
      "loss": 1.0051,
      "step": 520
    },
    {
      "epoch": 0.08735198574871633,
      "grad_norm": 0.4462275505065918,
      "learning_rate": 1.9831019926735924e-05,
      "loss": 0.9377,
      "step": 521
    },
    {
      "epoch": 0.08751964790946243,
      "grad_norm": 0.8423423767089844,
      "learning_rate": 1.9830024361785055e-05,
      "loss": 0.9989,
      "step": 522
    },
    {
      "epoch": 0.08768731007020852,
      "grad_norm": 0.7546736001968384,
      "learning_rate": 1.982902589783828e-05,
      "loss": 1.0704,
      "step": 523
    },
    {
      "epoch": 0.08785497223095462,
      "grad_norm": 0.536176323890686,
      "learning_rate": 1.9828024535190056e-05,
      "loss": 0.9193,
      "step": 524
    },
    {
      "epoch": 0.08802263439170073,
      "grad_norm": 0.8669360876083374,
      "learning_rate": 1.98270202741357e-05,
      "loss": 0.9705,
      "step": 525
    },
    {
      "epoch": 0.08819029655244683,
      "grad_norm": 0.5635377764701843,
      "learning_rate": 1.9826013114971375e-05,
      "loss": 0.9177,
      "step": 526
    },
    {
      "epoch": 0.08835795871319292,
      "grad_norm": 0.5467398166656494,
      "learning_rate": 1.9825003057994112e-05,
      "loss": 0.9935,
      "step": 527
    },
    {
      "epoch": 0.08852562087393902,
      "grad_norm": 0.45431026816368103,
      "learning_rate": 1.9823990103501783e-05,
      "loss": 1.0419,
      "step": 528
    },
    {
      "epoch": 0.08869328303468511,
      "grad_norm": 0.5135222673416138,
      "learning_rate": 1.9822974251793126e-05,
      "loss": 0.9543,
      "step": 529
    },
    {
      "epoch": 0.08886094519543121,
      "grad_norm": 0.4904458522796631,
      "learning_rate": 1.9821955503167726e-05,
      "loss": 1.0024,
      "step": 530
    },
    {
      "epoch": 0.0890286073561773,
      "grad_norm": 0.3739793002605438,
      "learning_rate": 1.9820933857926024e-05,
      "loss": 0.9023,
      "step": 531
    },
    {
      "epoch": 0.0891962695169234,
      "grad_norm": 0.4209132492542267,
      "learning_rate": 1.9819909316369314e-05,
      "loss": 1.0788,
      "step": 532
    },
    {
      "epoch": 0.0893639316776695,
      "grad_norm": 0.5051009654998779,
      "learning_rate": 1.9818881878799755e-05,
      "loss": 1.0887,
      "step": 533
    },
    {
      "epoch": 0.0895315938384156,
      "grad_norm": 0.43670955300331116,
      "learning_rate": 1.981785154552034e-05,
      "loss": 0.9689,
      "step": 534
    },
    {
      "epoch": 0.08969925599916169,
      "grad_norm": 0.36185288429260254,
      "learning_rate": 1.9816818316834933e-05,
      "loss": 0.9555,
      "step": 535
    },
    {
      "epoch": 0.08986691815990779,
      "grad_norm": 0.565312922000885,
      "learning_rate": 1.981578219304824e-05,
      "loss": 1.1048,
      "step": 536
    },
    {
      "epoch": 0.09003458032065388,
      "grad_norm": 0.37618112564086914,
      "learning_rate": 1.981474317446584e-05,
      "loss": 0.8761,
      "step": 537
    },
    {
      "epoch": 0.09020224248139998,
      "grad_norm": 0.47744205594062805,
      "learning_rate": 1.9813701261394136e-05,
      "loss": 0.9031,
      "step": 538
    },
    {
      "epoch": 0.09036990464214607,
      "grad_norm": 0.3989582359790802,
      "learning_rate": 1.981265645414041e-05,
      "loss": 0.9217,
      "step": 539
    },
    {
      "epoch": 0.09053756680289217,
      "grad_norm": 0.5119574069976807,
      "learning_rate": 1.9811608753012792e-05,
      "loss": 0.9871,
      "step": 540
    },
    {
      "epoch": 0.09070522896363827,
      "grad_norm": 0.6921777725219727,
      "learning_rate": 1.981055815832025e-05,
      "loss": 1.056,
      "step": 541
    },
    {
      "epoch": 0.09087289112438436,
      "grad_norm": 0.4229326546192169,
      "learning_rate": 1.9809504670372634e-05,
      "loss": 0.9696,
      "step": 542
    },
    {
      "epoch": 0.09104055328513046,
      "grad_norm": 0.47192320227622986,
      "learning_rate": 1.9808448289480614e-05,
      "loss": 1.0248,
      "step": 543
    },
    {
      "epoch": 0.09120821544587655,
      "grad_norm": 0.5146031379699707,
      "learning_rate": 1.9807389015955742e-05,
      "loss": 0.9254,
      "step": 544
    },
    {
      "epoch": 0.09137587760662265,
      "grad_norm": 0.48524725437164307,
      "learning_rate": 1.9806326850110406e-05,
      "loss": 0.9548,
      "step": 545
    },
    {
      "epoch": 0.09154353976736875,
      "grad_norm": 0.48073190450668335,
      "learning_rate": 1.980526179225785e-05,
      "loss": 0.9626,
      "step": 546
    },
    {
      "epoch": 0.09171120192811485,
      "grad_norm": 0.6683670878410339,
      "learning_rate": 1.980419384271217e-05,
      "loss": 1.0316,
      "step": 547
    },
    {
      "epoch": 0.09187886408886095,
      "grad_norm": 0.511667788028717,
      "learning_rate": 1.9803123001788327e-05,
      "loss": 0.9347,
      "step": 548
    },
    {
      "epoch": 0.09204652624960705,
      "grad_norm": 0.4302925765514374,
      "learning_rate": 1.980204926980212e-05,
      "loss": 0.9375,
      "step": 549
    },
    {
      "epoch": 0.09221418841035314,
      "grad_norm": 0.6657047271728516,
      "learning_rate": 1.980097264707021e-05,
      "loss": 1.0072,
      "step": 550
    },
    {
      "epoch": 0.09238185057109924,
      "grad_norm": 0.5272813439369202,
      "learning_rate": 1.97998931339101e-05,
      "loss": 0.9524,
      "step": 551
    },
    {
      "epoch": 0.09254951273184533,
      "grad_norm": 0.5764918327331543,
      "learning_rate": 1.9798810730640153e-05,
      "loss": 0.973,
      "step": 552
    },
    {
      "epoch": 0.09271717489259143,
      "grad_norm": 0.41316327452659607,
      "learning_rate": 1.9797725437579586e-05,
      "loss": 0.9984,
      "step": 553
    },
    {
      "epoch": 0.09288483705333753,
      "grad_norm": 0.40722402930259705,
      "learning_rate": 1.9796637255048462e-05,
      "loss": 0.82,
      "step": 554
    },
    {
      "epoch": 0.09305249921408362,
      "grad_norm": 0.3617995083332062,
      "learning_rate": 1.9795546183367708e-05,
      "loss": 0.9146,
      "step": 555
    },
    {
      "epoch": 0.09322016137482972,
      "grad_norm": 0.40822163224220276,
      "learning_rate": 1.979445222285908e-05,
      "loss": 0.963,
      "step": 556
    },
    {
      "epoch": 0.09338782353557581,
      "grad_norm": 0.5854674577713013,
      "learning_rate": 1.9793355373845214e-05,
      "loss": 1.0111,
      "step": 557
    },
    {
      "epoch": 0.09355548569632191,
      "grad_norm": 0.44850680232048035,
      "learning_rate": 1.9792255636649574e-05,
      "loss": 0.9146,
      "step": 558
    },
    {
      "epoch": 0.093723147857068,
      "grad_norm": 0.4834839999675751,
      "learning_rate": 1.9791153011596497e-05,
      "loss": 1.0255,
      "step": 559
    },
    {
      "epoch": 0.0938908100178141,
      "grad_norm": 0.5060991048812866,
      "learning_rate": 1.9790047499011152e-05,
      "loss": 1.0309,
      "step": 560
    },
    {
      "epoch": 0.0940584721785602,
      "grad_norm": 0.5379565954208374,
      "learning_rate": 1.9788939099219572e-05,
      "loss": 0.9369,
      "step": 561
    },
    {
      "epoch": 0.0942261343393063,
      "grad_norm": 0.44498369097709656,
      "learning_rate": 1.9787827812548634e-05,
      "loss": 0.9539,
      "step": 562
    },
    {
      "epoch": 0.09439379650005239,
      "grad_norm": 0.45467501878738403,
      "learning_rate": 1.9786713639326076e-05,
      "loss": 0.9934,
      "step": 563
    },
    {
      "epoch": 0.09456145866079849,
      "grad_norm": 0.454647034406662,
      "learning_rate": 1.978559657988048e-05,
      "loss": 0.8791,
      "step": 564
    },
    {
      "epoch": 0.09472912082154458,
      "grad_norm": 0.4199789762496948,
      "learning_rate": 1.9784476634541276e-05,
      "loss": 0.9348,
      "step": 565
    },
    {
      "epoch": 0.09489678298229068,
      "grad_norm": 0.34197133779525757,
      "learning_rate": 1.9783353803638757e-05,
      "loss": 0.9424,
      "step": 566
    },
    {
      "epoch": 0.09506444514303677,
      "grad_norm": 0.49100300669670105,
      "learning_rate": 1.978222808750405e-05,
      "loss": 0.9444,
      "step": 567
    },
    {
      "epoch": 0.09523210730378288,
      "grad_norm": 0.5817572474479675,
      "learning_rate": 1.978109948646915e-05,
      "loss": 0.9855,
      "step": 568
    },
    {
      "epoch": 0.09539976946452898,
      "grad_norm": 0.6111281514167786,
      "learning_rate": 1.97799680008669e-05,
      "loss": 0.9756,
      "step": 569
    },
    {
      "epoch": 0.09556743162527508,
      "grad_norm": 0.49359261989593506,
      "learning_rate": 1.977883363103098e-05,
      "loss": 0.9946,
      "step": 570
    },
    {
      "epoch": 0.09573509378602117,
      "grad_norm": 0.5452761650085449,
      "learning_rate": 1.9777696377295933e-05,
      "loss": 1.0238,
      "step": 571
    },
    {
      "epoch": 0.09590275594676727,
      "grad_norm": 0.4995993375778198,
      "learning_rate": 1.9776556239997146e-05,
      "loss": 0.9368,
      "step": 572
    },
    {
      "epoch": 0.09607041810751336,
      "grad_norm": 0.5189254283905029,
      "learning_rate": 1.9775413219470864e-05,
      "loss": 0.9531,
      "step": 573
    },
    {
      "epoch": 0.09623808026825946,
      "grad_norm": 0.6281453371047974,
      "learning_rate": 1.9774267316054177e-05,
      "loss": 0.9533,
      "step": 574
    },
    {
      "epoch": 0.09640574242900556,
      "grad_norm": 0.8713141083717346,
      "learning_rate": 1.9773118530085024e-05,
      "loss": 0.8697,
      "step": 575
    },
    {
      "epoch": 0.09657340458975165,
      "grad_norm": 0.6054361462593079,
      "learning_rate": 1.97719668619022e-05,
      "loss": 0.9665,
      "step": 576
    },
    {
      "epoch": 0.09674106675049775,
      "grad_norm": 0.5086394548416138,
      "learning_rate": 1.9770812311845344e-05,
      "loss": 0.9154,
      "step": 577
    },
    {
      "epoch": 0.09690872891124384,
      "grad_norm": 0.6285595297813416,
      "learning_rate": 1.9769654880254944e-05,
      "loss": 0.876,
      "step": 578
    },
    {
      "epoch": 0.09707639107198994,
      "grad_norm": 0.4077214300632477,
      "learning_rate": 1.9768494567472348e-05,
      "loss": 0.8753,
      "step": 579
    },
    {
      "epoch": 0.09724405323273604,
      "grad_norm": 0.47626930475234985,
      "learning_rate": 1.976733137383974e-05,
      "loss": 0.9437,
      "step": 580
    },
    {
      "epoch": 0.09741171539348213,
      "grad_norm": 0.48559632897377014,
      "learning_rate": 1.9766165299700164e-05,
      "loss": 1.0168,
      "step": 581
    },
    {
      "epoch": 0.09757937755422823,
      "grad_norm": 0.582771897315979,
      "learning_rate": 1.9764996345397505e-05,
      "loss": 0.8995,
      "step": 582
    },
    {
      "epoch": 0.09774703971497432,
      "grad_norm": 0.5317685008049011,
      "learning_rate": 1.976382451127651e-05,
      "loss": 0.8447,
      "step": 583
    },
    {
      "epoch": 0.09791470187572042,
      "grad_norm": 0.5318000912666321,
      "learning_rate": 1.976264979768276e-05,
      "loss": 1.0047,
      "step": 584
    },
    {
      "epoch": 0.09808236403646652,
      "grad_norm": 0.6099218130111694,
      "learning_rate": 1.97614722049627e-05,
      "loss": 0.9661,
      "step": 585
    },
    {
      "epoch": 0.09825002619721261,
      "grad_norm": 0.6967294812202454,
      "learning_rate": 1.976029173346361e-05,
      "loss": 1.0169,
      "step": 586
    },
    {
      "epoch": 0.09841768835795871,
      "grad_norm": 0.741862952709198,
      "learning_rate": 1.975910838353363e-05,
      "loss": 1.0529,
      "step": 587
    },
    {
      "epoch": 0.0985853505187048,
      "grad_norm": 0.5964228510856628,
      "learning_rate": 1.9757922155521737e-05,
      "loss": 1.0295,
      "step": 588
    },
    {
      "epoch": 0.09875301267945091,
      "grad_norm": 0.5316452383995056,
      "learning_rate": 1.9756733049777772e-05,
      "loss": 1.0512,
      "step": 589
    },
    {
      "epoch": 0.09892067484019701,
      "grad_norm": 0.533889889717102,
      "learning_rate": 1.9755541066652416e-05,
      "loss": 0.9739,
      "step": 590
    },
    {
      "epoch": 0.0990883370009431,
      "grad_norm": 0.5988102555274963,
      "learning_rate": 1.9754346206497197e-05,
      "loss": 0.8895,
      "step": 591
    },
    {
      "epoch": 0.0992559991616892,
      "grad_norm": 0.38310664892196655,
      "learning_rate": 1.9753148469664496e-05,
      "loss": 1.0405,
      "step": 592
    },
    {
      "epoch": 0.0994236613224353,
      "grad_norm": 0.46506670117378235,
      "learning_rate": 1.975194785650754e-05,
      "loss": 0.975,
      "step": 593
    },
    {
      "epoch": 0.0995913234831814,
      "grad_norm": 0.47853362560272217,
      "learning_rate": 1.975074436738041e-05,
      "loss": 0.9476,
      "step": 594
    },
    {
      "epoch": 0.09975898564392749,
      "grad_norm": 0.5202714204788208,
      "learning_rate": 1.9749538002638015e-05,
      "loss": 0.98,
      "step": 595
    },
    {
      "epoch": 0.09992664780467359,
      "grad_norm": 0.49216657876968384,
      "learning_rate": 1.9748328762636138e-05,
      "loss": 0.919,
      "step": 596
    },
    {
      "epoch": 0.10009430996541968,
      "grad_norm": 0.37900805473327637,
      "learning_rate": 1.9747116647731404e-05,
      "loss": 0.9919,
      "step": 597
    },
    {
      "epoch": 0.10026197212616578,
      "grad_norm": 0.5223008394241333,
      "learning_rate": 1.9745901658281267e-05,
      "loss": 0.9399,
      "step": 598
    },
    {
      "epoch": 0.10042963428691187,
      "grad_norm": 0.6181545257568359,
      "learning_rate": 1.9744683794644052e-05,
      "loss": 1.0172,
      "step": 599
    },
    {
      "epoch": 0.10059729644765797,
      "grad_norm": 0.5866363048553467,
      "learning_rate": 1.974346305717892e-05,
      "loss": 0.9147,
      "step": 600
    },
    {
      "epoch": 0.10076495860840406,
      "grad_norm": 0.5286931991577148,
      "learning_rate": 1.974223944624588e-05,
      "loss": 1.0886,
      "step": 601
    },
    {
      "epoch": 0.10093262076915016,
      "grad_norm": 0.7786131501197815,
      "learning_rate": 1.974101296220579e-05,
      "loss": 1.0236,
      "step": 602
    },
    {
      "epoch": 0.10110028292989626,
      "grad_norm": 0.6727932095527649,
      "learning_rate": 1.9739783605420357e-05,
      "loss": 0.9602,
      "step": 603
    },
    {
      "epoch": 0.10126794509064235,
      "grad_norm": 0.5062850713729858,
      "learning_rate": 1.9738551376252134e-05,
      "loss": 0.9491,
      "step": 604
    },
    {
      "epoch": 0.10143560725138845,
      "grad_norm": 0.6286392211914062,
      "learning_rate": 1.973731627506452e-05,
      "loss": 1.0341,
      "step": 605
    },
    {
      "epoch": 0.10160326941213454,
      "grad_norm": 0.7011929750442505,
      "learning_rate": 1.973607830222176e-05,
      "loss": 0.9357,
      "step": 606
    },
    {
      "epoch": 0.10177093157288064,
      "grad_norm": 0.5614083409309387,
      "learning_rate": 1.9734837458088947e-05,
      "loss": 0.9866,
      "step": 607
    },
    {
      "epoch": 0.10193859373362674,
      "grad_norm": 0.5135806202888489,
      "learning_rate": 1.9733593743032024e-05,
      "loss": 0.9159,
      "step": 608
    },
    {
      "epoch": 0.10210625589437283,
      "grad_norm": 0.47957277297973633,
      "learning_rate": 1.9732347157417776e-05,
      "loss": 0.9339,
      "step": 609
    },
    {
      "epoch": 0.10227391805511893,
      "grad_norm": 0.4980842173099518,
      "learning_rate": 1.9731097701613843e-05,
      "loss": 0.9543,
      "step": 610
    },
    {
      "epoch": 0.10244158021586504,
      "grad_norm": 0.6782700419425964,
      "learning_rate": 1.9729845375988694e-05,
      "loss": 1.0212,
      "step": 611
    },
    {
      "epoch": 0.10260924237661113,
      "grad_norm": 0.47850656509399414,
      "learning_rate": 1.972859018091166e-05,
      "loss": 0.9725,
      "step": 612
    },
    {
      "epoch": 0.10277690453735723,
      "grad_norm": 0.5600951313972473,
      "learning_rate": 1.9727332116752918e-05,
      "loss": 0.9931,
      "step": 613
    },
    {
      "epoch": 0.10294456669810333,
      "grad_norm": 0.445838987827301,
      "learning_rate": 1.9726071183883482e-05,
      "loss": 0.996,
      "step": 614
    },
    {
      "epoch": 0.10311222885884942,
      "grad_norm": 0.6184260845184326,
      "learning_rate": 1.9724807382675215e-05,
      "loss": 0.9265,
      "step": 615
    },
    {
      "epoch": 0.10327989101959552,
      "grad_norm": 0.8456107378005981,
      "learning_rate": 1.9723540713500834e-05,
      "loss": 0.9978,
      "step": 616
    },
    {
      "epoch": 0.10344755318034161,
      "grad_norm": 0.5166464447975159,
      "learning_rate": 1.972227117673389e-05,
      "loss": 0.9357,
      "step": 617
    },
    {
      "epoch": 0.10361521534108771,
      "grad_norm": 0.5970107913017273,
      "learning_rate": 1.972099877274879e-05,
      "loss": 0.9209,
      "step": 618
    },
    {
      "epoch": 0.1037828775018338,
      "grad_norm": 0.4248644709587097,
      "learning_rate": 1.971972350192077e-05,
      "loss": 0.8596,
      "step": 619
    },
    {
      "epoch": 0.1039505396625799,
      "grad_norm": 0.6106349229812622,
      "learning_rate": 1.971844536462594e-05,
      "loss": 0.9357,
      "step": 620
    },
    {
      "epoch": 0.104118201823326,
      "grad_norm": 0.4165198802947998,
      "learning_rate": 1.9717164361241225e-05,
      "loss": 0.9382,
      "step": 621
    },
    {
      "epoch": 0.1042858639840721,
      "grad_norm": 0.5961109399795532,
      "learning_rate": 1.9715880492144415e-05,
      "loss": 1.0102,
      "step": 622
    },
    {
      "epoch": 0.10445352614481819,
      "grad_norm": 0.39271023869514465,
      "learning_rate": 1.9714593757714143e-05,
      "loss": 1.008,
      "step": 623
    },
    {
      "epoch": 0.10462118830556429,
      "grad_norm": 0.4923511743545532,
      "learning_rate": 1.9713304158329873e-05,
      "loss": 1.0765,
      "step": 624
    },
    {
      "epoch": 0.10478885046631038,
      "grad_norm": 0.49212756752967834,
      "learning_rate": 1.9712011694371932e-05,
      "loss": 0.9187,
      "step": 625
    },
    {
      "epoch": 0.10495651262705648,
      "grad_norm": 0.4398774802684784,
      "learning_rate": 1.971071636622148e-05,
      "loss": 0.9804,
      "step": 626
    },
    {
      "epoch": 0.10512417478780257,
      "grad_norm": 0.4320284426212311,
      "learning_rate": 1.9709418174260523e-05,
      "loss": 0.8478,
      "step": 627
    },
    {
      "epoch": 0.10529183694854867,
      "grad_norm": 0.5619915127754211,
      "learning_rate": 1.9708117118871918e-05,
      "loss": 0.96,
      "step": 628
    },
    {
      "epoch": 0.10545949910929477,
      "grad_norm": 0.4957972466945648,
      "learning_rate": 1.9706813200439357e-05,
      "loss": 0.9265,
      "step": 629
    },
    {
      "epoch": 0.10562716127004086,
      "grad_norm": 0.41308099031448364,
      "learning_rate": 1.970550641934739e-05,
      "loss": 0.9377,
      "step": 630
    },
    {
      "epoch": 0.10579482343078696,
      "grad_norm": 0.43232348561286926,
      "learning_rate": 1.9704196775981397e-05,
      "loss": 0.958,
      "step": 631
    },
    {
      "epoch": 0.10596248559153307,
      "grad_norm": 0.5194548964500427,
      "learning_rate": 1.970288427072761e-05,
      "loss": 0.9631,
      "step": 632
    },
    {
      "epoch": 0.10613014775227916,
      "grad_norm": 0.5242116451263428,
      "learning_rate": 1.9701568903973104e-05,
      "loss": 0.9468,
      "step": 633
    },
    {
      "epoch": 0.10629780991302526,
      "grad_norm": 0.6986731886863708,
      "learning_rate": 1.9700250676105793e-05,
      "loss": 0.9996,
      "step": 634
    },
    {
      "epoch": 0.10646547207377136,
      "grad_norm": 0.6663616895675659,
      "learning_rate": 1.9698929587514444e-05,
      "loss": 0.9063,
      "step": 635
    },
    {
      "epoch": 0.10663313423451745,
      "grad_norm": 0.49162620306015015,
      "learning_rate": 1.9697605638588655e-05,
      "loss": 1.01,
      "step": 636
    },
    {
      "epoch": 0.10680079639526355,
      "grad_norm": 0.5097907185554504,
      "learning_rate": 1.9696278829718882e-05,
      "loss": 0.9027,
      "step": 637
    },
    {
      "epoch": 0.10696845855600964,
      "grad_norm": 0.46596378087997437,
      "learning_rate": 1.9694949161296416e-05,
      "loss": 0.9775,
      "step": 638
    },
    {
      "epoch": 0.10713612071675574,
      "grad_norm": 0.4176481068134308,
      "learning_rate": 1.969361663371339e-05,
      "loss": 0.8596,
      "step": 639
    },
    {
      "epoch": 0.10730378287750184,
      "grad_norm": 0.4506107568740845,
      "learning_rate": 1.9692281247362785e-05,
      "loss": 0.9667,
      "step": 640
    },
    {
      "epoch": 0.10747144503824793,
      "grad_norm": 0.38935619592666626,
      "learning_rate": 1.9690943002638425e-05,
      "loss": 1.032,
      "step": 641
    },
    {
      "epoch": 0.10763910719899403,
      "grad_norm": 0.45161691308021545,
      "learning_rate": 1.968960189993497e-05,
      "loss": 0.988,
      "step": 642
    },
    {
      "epoch": 0.10780676935974012,
      "grad_norm": 0.6230657696723938,
      "learning_rate": 1.968825793964793e-05,
      "loss": 0.9048,
      "step": 643
    },
    {
      "epoch": 0.10797443152048622,
      "grad_norm": 0.5937588810920715,
      "learning_rate": 1.9686911122173657e-05,
      "loss": 0.9231,
      "step": 644
    },
    {
      "epoch": 0.10814209368123232,
      "grad_norm": 0.6096629500389099,
      "learning_rate": 1.9685561447909342e-05,
      "loss": 0.92,
      "step": 645
    },
    {
      "epoch": 0.10830975584197841,
      "grad_norm": 0.6745415329933167,
      "learning_rate": 1.9684208917253018e-05,
      "loss": 0.9689,
      "step": 646
    },
    {
      "epoch": 0.10847741800272451,
      "grad_norm": 0.5356251001358032,
      "learning_rate": 1.968285353060357e-05,
      "loss": 0.9586,
      "step": 647
    },
    {
      "epoch": 0.1086450801634706,
      "grad_norm": 0.5918893218040466,
      "learning_rate": 1.9681495288360716e-05,
      "loss": 0.941,
      "step": 648
    },
    {
      "epoch": 0.1088127423242167,
      "grad_norm": 0.46468257904052734,
      "learning_rate": 1.9680134190925015e-05,
      "loss": 0.9289,
      "step": 649
    },
    {
      "epoch": 0.1089804044849628,
      "grad_norm": 0.46366575360298157,
      "learning_rate": 1.9678770238697876e-05,
      "loss": 0.9416,
      "step": 650
    },
    {
      "epoch": 0.10914806664570889,
      "grad_norm": 0.5816429257392883,
      "learning_rate": 1.9677403432081543e-05,
      "loss": 0.9545,
      "step": 651
    },
    {
      "epoch": 0.10931572880645499,
      "grad_norm": 0.9998892545700073,
      "learning_rate": 1.9676033771479102e-05,
      "loss": 0.951,
      "step": 652
    },
    {
      "epoch": 0.10948339096720108,
      "grad_norm": 0.381295382976532,
      "learning_rate": 1.9674661257294487e-05,
      "loss": 1.0326,
      "step": 653
    },
    {
      "epoch": 0.10965105312794719,
      "grad_norm": 0.46346166729927063,
      "learning_rate": 1.9673285889932468e-05,
      "loss": 1.0274,
      "step": 654
    },
    {
      "epoch": 0.10981871528869329,
      "grad_norm": 0.44738730788230896,
      "learning_rate": 1.9671907669798655e-05,
      "loss": 0.8813,
      "step": 655
    },
    {
      "epoch": 0.10998637744943938,
      "grad_norm": 0.5271283388137817,
      "learning_rate": 1.9670526597299505e-05,
      "loss": 0.9463,
      "step": 656
    },
    {
      "epoch": 0.11015403961018548,
      "grad_norm": 0.54680335521698,
      "learning_rate": 1.966914267284232e-05,
      "loss": 0.9622,
      "step": 657
    },
    {
      "epoch": 0.11032170177093158,
      "grad_norm": 0.5334712862968445,
      "learning_rate": 1.966775589683522e-05,
      "loss": 0.9911,
      "step": 658
    },
    {
      "epoch": 0.11048936393167767,
      "grad_norm": 0.5559065341949463,
      "learning_rate": 1.96663662696872e-05,
      "loss": 0.9833,
      "step": 659
    },
    {
      "epoch": 0.11065702609242377,
      "grad_norm": 0.5740633606910706,
      "learning_rate": 1.9664973791808064e-05,
      "loss": 0.8797,
      "step": 660
    },
    {
      "epoch": 0.11082468825316986,
      "grad_norm": 0.4243875741958618,
      "learning_rate": 1.966357846360848e-05,
      "loss": 0.9998,
      "step": 661
    },
    {
      "epoch": 0.11099235041391596,
      "grad_norm": 0.5085628032684326,
      "learning_rate": 1.9662180285499944e-05,
      "loss": 0.9774,
      "step": 662
    },
    {
      "epoch": 0.11116001257466206,
      "grad_norm": 0.54347163438797,
      "learning_rate": 1.96607792578948e-05,
      "loss": 1.0116,
      "step": 663
    },
    {
      "epoch": 0.11132767473540815,
      "grad_norm": 0.5442708134651184,
      "learning_rate": 1.965937538120622e-05,
      "loss": 0.9071,
      "step": 664
    },
    {
      "epoch": 0.11149533689615425,
      "grad_norm": 0.5275413990020752,
      "learning_rate": 1.9657968655848232e-05,
      "loss": 0.9896,
      "step": 665
    },
    {
      "epoch": 0.11166299905690034,
      "grad_norm": 0.3896923065185547,
      "learning_rate": 1.9656559082235697e-05,
      "loss": 0.9375,
      "step": 666
    },
    {
      "epoch": 0.11183066121764644,
      "grad_norm": 0.4918743073940277,
      "learning_rate": 1.965514666078431e-05,
      "loss": 0.9335,
      "step": 667
    },
    {
      "epoch": 0.11199832337839254,
      "grad_norm": 0.5577867031097412,
      "learning_rate": 1.9653731391910614e-05,
      "loss": 0.9044,
      "step": 668
    },
    {
      "epoch": 0.11216598553913863,
      "grad_norm": 0.4111005961894989,
      "learning_rate": 1.9652313276031995e-05,
      "loss": 0.9077,
      "step": 669
    },
    {
      "epoch": 0.11233364769988473,
      "grad_norm": 0.576390266418457,
      "learning_rate": 1.9650892313566667e-05,
      "loss": 1.0825,
      "step": 670
    },
    {
      "epoch": 0.11250130986063082,
      "grad_norm": 0.5298464298248291,
      "learning_rate": 1.9649468504933685e-05,
      "loss": 0.9854,
      "step": 671
    },
    {
      "epoch": 0.11266897202137692,
      "grad_norm": 0.5587629675865173,
      "learning_rate": 1.964804185055296e-05,
      "loss": 0.909,
      "step": 672
    },
    {
      "epoch": 0.11283663418212302,
      "grad_norm": 0.6600515246391296,
      "learning_rate": 1.9646612350845223e-05,
      "loss": 0.9134,
      "step": 673
    },
    {
      "epoch": 0.11300429634286911,
      "grad_norm": 0.5140649676322937,
      "learning_rate": 1.9645180006232046e-05,
      "loss": 1.0089,
      "step": 674
    },
    {
      "epoch": 0.11317195850361522,
      "grad_norm": 0.7896112203598022,
      "learning_rate": 1.9643744817135856e-05,
      "loss": 0.8986,
      "step": 675
    },
    {
      "epoch": 0.11333962066436132,
      "grad_norm": 0.5896207094192505,
      "learning_rate": 1.9642306783979902e-05,
      "loss": 0.9698,
      "step": 676
    },
    {
      "epoch": 0.11350728282510741,
      "grad_norm": 0.5102990865707397,
      "learning_rate": 1.9640865907188273e-05,
      "loss": 0.993,
      "step": 677
    },
    {
      "epoch": 0.11367494498585351,
      "grad_norm": 0.6648029685020447,
      "learning_rate": 1.9639422187185913e-05,
      "loss": 1.0052,
      "step": 678
    },
    {
      "epoch": 0.1138426071465996,
      "grad_norm": 0.5018223524093628,
      "learning_rate": 1.9637975624398583e-05,
      "loss": 0.9741,
      "step": 679
    },
    {
      "epoch": 0.1140102693073457,
      "grad_norm": 0.32673925161361694,
      "learning_rate": 1.9636526219252898e-05,
      "loss": 0.9169,
      "step": 680
    },
    {
      "epoch": 0.1141779314680918,
      "grad_norm": 0.39858177304267883,
      "learning_rate": 1.9635073972176303e-05,
      "loss": 0.9473,
      "step": 681
    },
    {
      "epoch": 0.1143455936288379,
      "grad_norm": 0.4796540439128876,
      "learning_rate": 1.9633618883597087e-05,
      "loss": 1.0437,
      "step": 682
    },
    {
      "epoch": 0.11451325578958399,
      "grad_norm": 0.517875611782074,
      "learning_rate": 1.963216095394437e-05,
      "loss": 0.9306,
      "step": 683
    },
    {
      "epoch": 0.11468091795033009,
      "grad_norm": 0.756864607334137,
      "learning_rate": 1.9630700183648112e-05,
      "loss": 0.8807,
      "step": 684
    },
    {
      "epoch": 0.11484858011107618,
      "grad_norm": 0.4738442897796631,
      "learning_rate": 1.9629236573139118e-05,
      "loss": 0.9638,
      "step": 685
    },
    {
      "epoch": 0.11501624227182228,
      "grad_norm": 0.45979225635528564,
      "learning_rate": 1.9627770122849017e-05,
      "loss": 0.8824,
      "step": 686
    },
    {
      "epoch": 0.11518390443256837,
      "grad_norm": 0.47601425647735596,
      "learning_rate": 1.9626300833210293e-05,
      "loss": 0.9283,
      "step": 687
    },
    {
      "epoch": 0.11535156659331447,
      "grad_norm": 0.49816030263900757,
      "learning_rate": 1.962482870465625e-05,
      "loss": 0.8787,
      "step": 688
    },
    {
      "epoch": 0.11551922875406057,
      "grad_norm": 0.5461615920066833,
      "learning_rate": 1.9623353737621035e-05,
      "loss": 0.9864,
      "step": 689
    },
    {
      "epoch": 0.11568689091480666,
      "grad_norm": 0.5288472771644592,
      "learning_rate": 1.9621875932539643e-05,
      "loss": 0.9758,
      "step": 690
    },
    {
      "epoch": 0.11585455307555276,
      "grad_norm": 0.6684693098068237,
      "learning_rate": 1.9620395289847894e-05,
      "loss": 0.905,
      "step": 691
    },
    {
      "epoch": 0.11602221523629885,
      "grad_norm": 0.46418723464012146,
      "learning_rate": 1.9618911809982445e-05,
      "loss": 0.9914,
      "step": 692
    },
    {
      "epoch": 0.11618987739704495,
      "grad_norm": 0.677675187587738,
      "learning_rate": 1.9617425493380796e-05,
      "loss": 1.0004,
      "step": 693
    },
    {
      "epoch": 0.11635753955779105,
      "grad_norm": 0.646614134311676,
      "learning_rate": 1.9615936340481276e-05,
      "loss": 1.0012,
      "step": 694
    },
    {
      "epoch": 0.11652520171853714,
      "grad_norm": 0.8820503354072571,
      "learning_rate": 1.9614444351723056e-05,
      "loss": 0.8818,
      "step": 695
    },
    {
      "epoch": 0.11669286387928324,
      "grad_norm": 0.6305606961250305,
      "learning_rate": 1.9612949527546147e-05,
      "loss": 0.9922,
      "step": 696
    },
    {
      "epoch": 0.11686052604002935,
      "grad_norm": 0.664298415184021,
      "learning_rate": 1.9611451868391384e-05,
      "loss": 0.86,
      "step": 697
    },
    {
      "epoch": 0.11702818820077544,
      "grad_norm": 0.845890462398529,
      "learning_rate": 1.9609951374700443e-05,
      "loss": 0.9133,
      "step": 698
    },
    {
      "epoch": 0.11719585036152154,
      "grad_norm": 0.5247465372085571,
      "learning_rate": 1.9608448046915852e-05,
      "loss": 0.9349,
      "step": 699
    },
    {
      "epoch": 0.11736351252226764,
      "grad_norm": 0.48808741569519043,
      "learning_rate": 1.9606941885480947e-05,
      "loss": 0.9186,
      "step": 700
    },
    {
      "epoch": 0.11753117468301373,
      "grad_norm": 0.4177757501602173,
      "learning_rate": 1.960543289083992e-05,
      "loss": 0.9658,
      "step": 701
    },
    {
      "epoch": 0.11769883684375983,
      "grad_norm": 0.4671350419521332,
      "learning_rate": 1.9603921063437795e-05,
      "loss": 0.9344,
      "step": 702
    },
    {
      "epoch": 0.11786649900450592,
      "grad_norm": 0.4618290960788727,
      "learning_rate": 1.960240640372042e-05,
      "loss": 0.965,
      "step": 703
    },
    {
      "epoch": 0.11803416116525202,
      "grad_norm": 0.5182496905326843,
      "learning_rate": 1.9600888912134495e-05,
      "loss": 0.9024,
      "step": 704
    },
    {
      "epoch": 0.11820182332599811,
      "grad_norm": 0.5245487689971924,
      "learning_rate": 1.9599368589127542e-05,
      "loss": 0.933,
      "step": 705
    },
    {
      "epoch": 0.11836948548674421,
      "grad_norm": 0.4822204113006592,
      "learning_rate": 1.9597845435147924e-05,
      "loss": 0.9824,
      "step": 706
    },
    {
      "epoch": 0.1185371476474903,
      "grad_norm": 0.7162145972251892,
      "learning_rate": 1.959631945064484e-05,
      "loss": 1.0036,
      "step": 707
    },
    {
      "epoch": 0.1187048098082364,
      "grad_norm": 0.4953795075416565,
      "learning_rate": 1.9594790636068323e-05,
      "loss": 0.897,
      "step": 708
    },
    {
      "epoch": 0.1188724719689825,
      "grad_norm": 0.6701470017433167,
      "learning_rate": 1.9593258991869235e-05,
      "loss": 0.9,
      "step": 709
    },
    {
      "epoch": 0.1190401341297286,
      "grad_norm": 0.72853022813797,
      "learning_rate": 1.959172451849928e-05,
      "loss": 0.9866,
      "step": 710
    },
    {
      "epoch": 0.11920779629047469,
      "grad_norm": 0.5792762637138367,
      "learning_rate": 1.9590187216410987e-05,
      "loss": 0.9393,
      "step": 711
    },
    {
      "epoch": 0.11937545845122079,
      "grad_norm": 0.5983287692070007,
      "learning_rate": 1.9588647086057734e-05,
      "loss": 1.029,
      "step": 712
    },
    {
      "epoch": 0.11954312061196688,
      "grad_norm": 0.7085567712783813,
      "learning_rate": 1.958710412789372e-05,
      "loss": 0.9128,
      "step": 713
    },
    {
      "epoch": 0.11971078277271298,
      "grad_norm": 0.44527891278266907,
      "learning_rate": 1.9585558342373986e-05,
      "loss": 0.9395,
      "step": 714
    },
    {
      "epoch": 0.11987844493345907,
      "grad_norm": 0.5031890273094177,
      "learning_rate": 1.9584009729954395e-05,
      "loss": 1.034,
      "step": 715
    },
    {
      "epoch": 0.12004610709420517,
      "grad_norm": 0.4327860176563263,
      "learning_rate": 1.9582458291091664e-05,
      "loss": 0.9596,
      "step": 716
    },
    {
      "epoch": 0.12021376925495127,
      "grad_norm": 0.642092764377594,
      "learning_rate": 1.9580904026243322e-05,
      "loss": 0.9595,
      "step": 717
    },
    {
      "epoch": 0.12038143141569738,
      "grad_norm": 0.6110759973526001,
      "learning_rate": 1.9579346935867743e-05,
      "loss": 0.9861,
      "step": 718
    },
    {
      "epoch": 0.12054909357644347,
      "grad_norm": 0.46162545680999756,
      "learning_rate": 1.9577787020424135e-05,
      "loss": 0.8916,
      "step": 719
    },
    {
      "epoch": 0.12071675573718957,
      "grad_norm": 0.5048706531524658,
      "learning_rate": 1.9576224280372537e-05,
      "loss": 0.9519,
      "step": 720
    },
    {
      "epoch": 0.12088441789793566,
      "grad_norm": 0.44191446900367737,
      "learning_rate": 1.9574658716173817e-05,
      "loss": 1.0365,
      "step": 721
    },
    {
      "epoch": 0.12105208005868176,
      "grad_norm": 0.7060146331787109,
      "learning_rate": 1.9573090328289675e-05,
      "loss": 0.9745,
      "step": 722
    },
    {
      "epoch": 0.12121974221942786,
      "grad_norm": 0.48769238591194153,
      "learning_rate": 1.957151911718266e-05,
      "loss": 1.0441,
      "step": 723
    },
    {
      "epoch": 0.12138740438017395,
      "grad_norm": 0.4783252775669098,
      "learning_rate": 1.9569945083316133e-05,
      "loss": 0.9536,
      "step": 724
    },
    {
      "epoch": 0.12155506654092005,
      "grad_norm": 0.5853663682937622,
      "learning_rate": 1.9568368227154297e-05,
      "loss": 0.8748,
      "step": 725
    },
    {
      "epoch": 0.12172272870166614,
      "grad_norm": 0.5817577242851257,
      "learning_rate": 1.9566788549162188e-05,
      "loss": 1.0118,
      "step": 726
    },
    {
      "epoch": 0.12189039086241224,
      "grad_norm": 0.44570523500442505,
      "learning_rate": 1.9565206049805673e-05,
      "loss": 0.934,
      "step": 727
    },
    {
      "epoch": 0.12205805302315834,
      "grad_norm": 0.7238737940788269,
      "learning_rate": 1.9563620729551448e-05,
      "loss": 0.9746,
      "step": 728
    },
    {
      "epoch": 0.12222571518390443,
      "grad_norm": 0.43183931708335876,
      "learning_rate": 1.9562032588867047e-05,
      "loss": 0.9409,
      "step": 729
    },
    {
      "epoch": 0.12239337734465053,
      "grad_norm": 0.4638485014438629,
      "learning_rate": 1.956044162822083e-05,
      "loss": 0.8582,
      "step": 730
    },
    {
      "epoch": 0.12256103950539662,
      "grad_norm": 0.5277942419052124,
      "learning_rate": 1.955884784808199e-05,
      "loss": 0.899,
      "step": 731
    },
    {
      "epoch": 0.12272870166614272,
      "grad_norm": 0.5024052858352661,
      "learning_rate": 1.9557251248920556e-05,
      "loss": 1.0152,
      "step": 732
    },
    {
      "epoch": 0.12289636382688882,
      "grad_norm": 0.46881332993507385,
      "learning_rate": 1.9555651831207383e-05,
      "loss": 0.9594,
      "step": 733
    },
    {
      "epoch": 0.12306402598763491,
      "grad_norm": 0.6736809015274048,
      "learning_rate": 1.955404959541416e-05,
      "loss": 1.0201,
      "step": 734
    },
    {
      "epoch": 0.12323168814838101,
      "grad_norm": 1.2368195056915283,
      "learning_rate": 1.95524445420134e-05,
      "loss": 0.8936,
      "step": 735
    },
    {
      "epoch": 0.1233993503091271,
      "grad_norm": 0.5060524344444275,
      "learning_rate": 1.9550836671478465e-05,
      "loss": 0.8747,
      "step": 736
    },
    {
      "epoch": 0.1235670124698732,
      "grad_norm": 0.7231465578079224,
      "learning_rate": 1.9549225984283528e-05,
      "loss": 0.8913,
      "step": 737
    },
    {
      "epoch": 0.1237346746306193,
      "grad_norm": 0.5625017881393433,
      "learning_rate": 1.9547612480903598e-05,
      "loss": 1.0382,
      "step": 738
    },
    {
      "epoch": 0.1239023367913654,
      "grad_norm": 0.6068775057792664,
      "learning_rate": 1.9545996161814526e-05,
      "loss": 0.9309,
      "step": 739
    },
    {
      "epoch": 0.1240699989521115,
      "grad_norm": 0.5323880314826965,
      "learning_rate": 1.9544377027492974e-05,
      "loss": 0.861,
      "step": 740
    },
    {
      "epoch": 0.1242376611128576,
      "grad_norm": 0.5942120552062988,
      "learning_rate": 1.954275507841646e-05,
      "loss": 0.9093,
      "step": 741
    },
    {
      "epoch": 0.1244053232736037,
      "grad_norm": 0.5026261210441589,
      "learning_rate": 1.9541130315063297e-05,
      "loss": 0.9982,
      "step": 742
    },
    {
      "epoch": 0.12457298543434979,
      "grad_norm": 0.5589470863342285,
      "learning_rate": 1.9539502737912664e-05,
      "loss": 0.9119,
      "step": 743
    },
    {
      "epoch": 0.12474064759509589,
      "grad_norm": 0.7846218347549438,
      "learning_rate": 1.9537872347444544e-05,
      "loss": 0.9607,
      "step": 744
    },
    {
      "epoch": 0.12490830975584198,
      "grad_norm": 0.5878773927688599,
      "learning_rate": 1.9536239144139767e-05,
      "loss": 0.9941,
      "step": 745
    },
    {
      "epoch": 0.12507597191658806,
      "grad_norm": 0.5514938235282898,
      "learning_rate": 1.9534603128479976e-05,
      "loss": 0.9315,
      "step": 746
    },
    {
      "epoch": 0.12524363407733416,
      "grad_norm": 0.7384706735610962,
      "learning_rate": 1.953296430094766e-05,
      "loss": 0.9754,
      "step": 747
    },
    {
      "epoch": 0.12541129623808026,
      "grad_norm": 0.45522984862327576,
      "learning_rate": 1.953132266202613e-05,
      "loss": 0.9331,
      "step": 748
    },
    {
      "epoch": 0.12557895839882635,
      "grad_norm": 0.6133016347885132,
      "learning_rate": 1.9529678212199517e-05,
      "loss": 0.9913,
      "step": 749
    },
    {
      "epoch": 0.12574662055957248,
      "grad_norm": 0.6371902227401733,
      "learning_rate": 1.9528030951952796e-05,
      "loss": 1.0075,
      "step": 750
    },
    {
      "epoch": 0.12591428272031857,
      "grad_norm": 0.5542172789573669,
      "learning_rate": 1.9526380881771766e-05,
      "loss": 0.9307,
      "step": 751
    },
    {
      "epoch": 0.12608194488106467,
      "grad_norm": 0.5045576095581055,
      "learning_rate": 1.952472800214305e-05,
      "loss": 0.9441,
      "step": 752
    },
    {
      "epoch": 0.12624960704181076,
      "grad_norm": 0.4427657723426819,
      "learning_rate": 1.95230723135541e-05,
      "loss": 0.8946,
      "step": 753
    },
    {
      "epoch": 0.12641726920255686,
      "grad_norm": 0.4645458161830902,
      "learning_rate": 1.9521413816493206e-05,
      "loss": 0.9801,
      "step": 754
    },
    {
      "epoch": 0.12658493136330295,
      "grad_norm": 0.4842035174369812,
      "learning_rate": 1.9519752511449476e-05,
      "loss": 0.9389,
      "step": 755
    },
    {
      "epoch": 0.12675259352404905,
      "grad_norm": 0.4109826683998108,
      "learning_rate": 1.9518088398912846e-05,
      "loss": 0.9059,
      "step": 756
    },
    {
      "epoch": 0.12692025568479515,
      "grad_norm": 0.4934842586517334,
      "learning_rate": 1.9516421479374088e-05,
      "loss": 0.9971,
      "step": 757
    },
    {
      "epoch": 0.12708791784554124,
      "grad_norm": 0.3934910297393799,
      "learning_rate": 1.9514751753324795e-05,
      "loss": 0.9402,
      "step": 758
    },
    {
      "epoch": 0.12725558000628734,
      "grad_norm": 0.45803117752075195,
      "learning_rate": 1.951307922125739e-05,
      "loss": 0.9056,
      "step": 759
    },
    {
      "epoch": 0.12742324216703343,
      "grad_norm": 0.4900297522544861,
      "learning_rate": 1.9511403883665126e-05,
      "loss": 0.892,
      "step": 760
    },
    {
      "epoch": 0.12759090432777953,
      "grad_norm": 0.5113170146942139,
      "learning_rate": 1.9509725741042077e-05,
      "loss": 0.9232,
      "step": 761
    },
    {
      "epoch": 0.12775856648852563,
      "grad_norm": 0.3888622522354126,
      "learning_rate": 1.9508044793883152e-05,
      "loss": 0.8747,
      "step": 762
    },
    {
      "epoch": 0.12792622864927172,
      "grad_norm": 0.5268617272377014,
      "learning_rate": 1.9506361042684077e-05,
      "loss": 0.8486,
      "step": 763
    },
    {
      "epoch": 0.12809389081001782,
      "grad_norm": 0.49528518319129944,
      "learning_rate": 1.950467448794142e-05,
      "loss": 0.8963,
      "step": 764
    },
    {
      "epoch": 0.12826155297076391,
      "grad_norm": 0.38182708621025085,
      "learning_rate": 1.9502985130152557e-05,
      "loss": 0.9232,
      "step": 765
    },
    {
      "epoch": 0.12842921513151,
      "grad_norm": 0.4788253903388977,
      "learning_rate": 1.950129296981571e-05,
      "loss": 0.9111,
      "step": 766
    },
    {
      "epoch": 0.1285968772922561,
      "grad_norm": 0.49691474437713623,
      "learning_rate": 1.949959800742991e-05,
      "loss": 0.99,
      "step": 767
    },
    {
      "epoch": 0.1287645394530022,
      "grad_norm": 0.7026716470718384,
      "learning_rate": 1.949790024349503e-05,
      "loss": 0.9412,
      "step": 768
    },
    {
      "epoch": 0.1289322016137483,
      "grad_norm": 0.3846409320831299,
      "learning_rate": 1.949619967851175e-05,
      "loss": 0.9823,
      "step": 769
    },
    {
      "epoch": 0.1290998637744944,
      "grad_norm": 0.5087264776229858,
      "learning_rate": 1.94944963129816e-05,
      "loss": 0.9425,
      "step": 770
    },
    {
      "epoch": 0.1292675259352405,
      "grad_norm": 0.42938852310180664,
      "learning_rate": 1.9492790147406918e-05,
      "loss": 0.9304,
      "step": 771
    },
    {
      "epoch": 0.1294351880959866,
      "grad_norm": 0.4374798834323883,
      "learning_rate": 1.9491081182290877e-05,
      "loss": 0.8703,
      "step": 772
    },
    {
      "epoch": 0.12960285025673268,
      "grad_norm": 0.3993895947933197,
      "learning_rate": 1.9489369418137464e-05,
      "loss": 0.8903,
      "step": 773
    },
    {
      "epoch": 0.12977051241747878,
      "grad_norm": 0.4784952700138092,
      "learning_rate": 1.9487654855451507e-05,
      "loss": 1.0,
      "step": 774
    },
    {
      "epoch": 0.12993817457822487,
      "grad_norm": 0.48840826749801636,
      "learning_rate": 1.948593749473865e-05,
      "loss": 0.8849,
      "step": 775
    },
    {
      "epoch": 0.13010583673897097,
      "grad_norm": 0.42742201685905457,
      "learning_rate": 1.9484217336505362e-05,
      "loss": 0.992,
      "step": 776
    },
    {
      "epoch": 0.13027349889971707,
      "grad_norm": 0.4182666838169098,
      "learning_rate": 1.948249438125894e-05,
      "loss": 1.0784,
      "step": 777
    },
    {
      "epoch": 0.13044116106046316,
      "grad_norm": 0.3841330111026764,
      "learning_rate": 1.9480768629507508e-05,
      "loss": 0.9422,
      "step": 778
    },
    {
      "epoch": 0.13060882322120926,
      "grad_norm": 0.3590227961540222,
      "learning_rate": 1.9479040081760005e-05,
      "loss": 0.9938,
      "step": 779
    },
    {
      "epoch": 0.13077648538195535,
      "grad_norm": 0.5238223075866699,
      "learning_rate": 1.9477308738526207e-05,
      "loss": 0.8615,
      "step": 780
    },
    {
      "epoch": 0.13094414754270145,
      "grad_norm": 0.423281192779541,
      "learning_rate": 1.9475574600316704e-05,
      "loss": 0.9103,
      "step": 781
    },
    {
      "epoch": 0.13111180970344755,
      "grad_norm": 0.38820210099220276,
      "learning_rate": 1.947383766764292e-05,
      "loss": 0.8673,
      "step": 782
    },
    {
      "epoch": 0.13127947186419364,
      "grad_norm": 0.6232205033302307,
      "learning_rate": 1.9472097941017092e-05,
      "loss": 0.8361,
      "step": 783
    },
    {
      "epoch": 0.13144713402493974,
      "grad_norm": 0.5535816550254822,
      "learning_rate": 1.947035542095229e-05,
      "loss": 0.9535,
      "step": 784
    },
    {
      "epoch": 0.13161479618568583,
      "grad_norm": 0.4846842885017395,
      "learning_rate": 1.9468610107962405e-05,
      "loss": 0.9282,
      "step": 785
    },
    {
      "epoch": 0.13178245834643193,
      "grad_norm": 0.8760911822319031,
      "learning_rate": 1.9466862002562154e-05,
      "loss": 0.9522,
      "step": 786
    },
    {
      "epoch": 0.13195012050717803,
      "grad_norm": 0.5710975527763367,
      "learning_rate": 1.946511110526707e-05,
      "loss": 1.0379,
      "step": 787
    },
    {
      "epoch": 0.13211778266792412,
      "grad_norm": 0.7084416747093201,
      "learning_rate": 1.9463357416593517e-05,
      "loss": 1.0698,
      "step": 788
    },
    {
      "epoch": 0.13228544482867022,
      "grad_norm": 0.6290233135223389,
      "learning_rate": 1.946160093705868e-05,
      "loss": 0.8534,
      "step": 789
    },
    {
      "epoch": 0.1324531069894163,
      "grad_norm": 0.5018370747566223,
      "learning_rate": 1.945984166718056e-05,
      "loss": 1.0086,
      "step": 790
    },
    {
      "epoch": 0.1326207691501624,
      "grad_norm": 0.46618857979774475,
      "learning_rate": 1.9458079607478e-05,
      "loss": 0.903,
      "step": 791
    },
    {
      "epoch": 0.1327884313109085,
      "grad_norm": 0.40250569581985474,
      "learning_rate": 1.9456314758470642e-05,
      "loss": 0.9104,
      "step": 792
    },
    {
      "epoch": 0.13295609347165463,
      "grad_norm": 0.44152161478996277,
      "learning_rate": 1.9454547120678966e-05,
      "loss": 0.9364,
      "step": 793
    },
    {
      "epoch": 0.13312375563240073,
      "grad_norm": 0.4875357151031494,
      "learning_rate": 1.945277669462427e-05,
      "loss": 0.9933,
      "step": 794
    },
    {
      "epoch": 0.13329141779314682,
      "grad_norm": 0.43597495555877686,
      "learning_rate": 1.9451003480828677e-05,
      "loss": 1.0492,
      "step": 795
    },
    {
      "epoch": 0.13345907995389292,
      "grad_norm": 0.5917737483978271,
      "learning_rate": 1.9449227479815126e-05,
      "loss": 0.9141,
      "step": 796
    },
    {
      "epoch": 0.133626742114639,
      "grad_norm": 0.6467670202255249,
      "learning_rate": 1.9447448692107383e-05,
      "loss": 0.9828,
      "step": 797
    },
    {
      "epoch": 0.1337944042753851,
      "grad_norm": 0.6712422966957092,
      "learning_rate": 1.9445667118230037e-05,
      "loss": 0.9484,
      "step": 798
    },
    {
      "epoch": 0.1339620664361312,
      "grad_norm": 0.4396202564239502,
      "learning_rate": 1.944388275870849e-05,
      "loss": 1.0194,
      "step": 799
    },
    {
      "epoch": 0.1341297285968773,
      "grad_norm": 0.5531654357910156,
      "learning_rate": 1.9442095614068977e-05,
      "loss": 0.9144,
      "step": 800
    },
    {
      "epoch": 0.1342973907576234,
      "grad_norm": 0.5850380063056946,
      "learning_rate": 1.944030568483855e-05,
      "loss": 0.8749,
      "step": 801
    },
    {
      "epoch": 0.1344650529183695,
      "grad_norm": 0.5013009309768677,
      "learning_rate": 1.943851297154508e-05,
      "loss": 0.92,
      "step": 802
    },
    {
      "epoch": 0.1346327150791156,
      "grad_norm": 0.501086413860321,
      "learning_rate": 1.9436717474717253e-05,
      "loss": 0.9618,
      "step": 803
    },
    {
      "epoch": 0.13480037723986169,
      "grad_norm": 0.4982043206691742,
      "learning_rate": 1.9434919194884593e-05,
      "loss": 0.9695,
      "step": 804
    },
    {
      "epoch": 0.13496803940060778,
      "grad_norm": 0.3764241635799408,
      "learning_rate": 1.9433118132577432e-05,
      "loss": 0.8995,
      "step": 805
    },
    {
      "epoch": 0.13513570156135388,
      "grad_norm": 0.533713161945343,
      "learning_rate": 1.9431314288326925e-05,
      "loss": 1.018,
      "step": 806
    },
    {
      "epoch": 0.13530336372209997,
      "grad_norm": 0.5478500127792358,
      "learning_rate": 1.942950766266505e-05,
      "loss": 0.9551,
      "step": 807
    },
    {
      "epoch": 0.13547102588284607,
      "grad_norm": 0.6805524826049805,
      "learning_rate": 1.9427698256124598e-05,
      "loss": 1.0809,
      "step": 808
    },
    {
      "epoch": 0.13563868804359216,
      "grad_norm": 0.48846960067749023,
      "learning_rate": 1.9425886069239192e-05,
      "loss": 0.9434,
      "step": 809
    },
    {
      "epoch": 0.13580635020433826,
      "grad_norm": 0.5114566683769226,
      "learning_rate": 1.9424071102543262e-05,
      "loss": 0.982,
      "step": 810
    },
    {
      "epoch": 0.13597401236508436,
      "grad_norm": 0.5536603927612305,
      "learning_rate": 1.942225335657207e-05,
      "loss": 1.0784,
      "step": 811
    },
    {
      "epoch": 0.13614167452583045,
      "grad_norm": 0.5240349173545837,
      "learning_rate": 1.9420432831861688e-05,
      "loss": 0.8194,
      "step": 812
    },
    {
      "epoch": 0.13630933668657655,
      "grad_norm": 0.42566990852355957,
      "learning_rate": 1.941860952894901e-05,
      "loss": 0.9778,
      "step": 813
    },
    {
      "epoch": 0.13647699884732264,
      "grad_norm": 0.49338826537132263,
      "learning_rate": 1.9416783448371758e-05,
      "loss": 0.8738,
      "step": 814
    },
    {
      "epoch": 0.13664466100806874,
      "grad_norm": 0.5864494442939758,
      "learning_rate": 1.941495459066846e-05,
      "loss": 0.9206,
      "step": 815
    },
    {
      "epoch": 0.13681232316881484,
      "grad_norm": 0.3878153860569,
      "learning_rate": 1.9413122956378466e-05,
      "loss": 0.9598,
      "step": 816
    },
    {
      "epoch": 0.13697998532956093,
      "grad_norm": 0.9809596538543701,
      "learning_rate": 1.941128854604195e-05,
      "loss": 0.9102,
      "step": 817
    },
    {
      "epoch": 0.13714764749030703,
      "grad_norm": 0.5030829906463623,
      "learning_rate": 1.9409451360199907e-05,
      "loss": 0.9994,
      "step": 818
    },
    {
      "epoch": 0.13731530965105312,
      "grad_norm": 0.510537326335907,
      "learning_rate": 1.9407611399394145e-05,
      "loss": 0.8467,
      "step": 819
    },
    {
      "epoch": 0.13748297181179922,
      "grad_norm": 0.6032730937004089,
      "learning_rate": 1.9405768664167284e-05,
      "loss": 0.978,
      "step": 820
    },
    {
      "epoch": 0.13765063397254532,
      "grad_norm": 0.5233103632926941,
      "learning_rate": 1.9403923155062776e-05,
      "loss": 0.9526,
      "step": 821
    },
    {
      "epoch": 0.1378182961332914,
      "grad_norm": 0.6112736463546753,
      "learning_rate": 1.9402074872624883e-05,
      "loss": 0.9314,
      "step": 822
    },
    {
      "epoch": 0.1379859582940375,
      "grad_norm": 0.6209530234336853,
      "learning_rate": 1.9400223817398684e-05,
      "loss": 1.0253,
      "step": 823
    },
    {
      "epoch": 0.1381536204547836,
      "grad_norm": 0.5491402745246887,
      "learning_rate": 1.939836998993008e-05,
      "loss": 0.919,
      "step": 824
    },
    {
      "epoch": 0.1383212826155297,
      "grad_norm": 0.8800770044326782,
      "learning_rate": 1.939651339076579e-05,
      "loss": 0.8918,
      "step": 825
    },
    {
      "epoch": 0.1384889447762758,
      "grad_norm": 0.48260605335235596,
      "learning_rate": 1.939465402045335e-05,
      "loss": 1.0064,
      "step": 826
    },
    {
      "epoch": 0.1386566069370219,
      "grad_norm": 0.7165107131004333,
      "learning_rate": 1.9392791879541096e-05,
      "loss": 0.8179,
      "step": 827
    },
    {
      "epoch": 0.138824269097768,
      "grad_norm": 0.4312284290790558,
      "learning_rate": 1.9390926968578213e-05,
      "loss": 0.9617,
      "step": 828
    },
    {
      "epoch": 0.13899193125851408,
      "grad_norm": 0.4455448091030121,
      "learning_rate": 1.9389059288114684e-05,
      "loss": 0.8876,
      "step": 829
    },
    {
      "epoch": 0.13915959341926018,
      "grad_norm": 0.55988609790802,
      "learning_rate": 1.9387188838701304e-05,
      "loss": 0.9553,
      "step": 830
    },
    {
      "epoch": 0.13932725558000628,
      "grad_norm": 0.5089917778968811,
      "learning_rate": 1.93853156208897e-05,
      "loss": 0.8604,
      "step": 831
    },
    {
      "epoch": 0.13949491774075237,
      "grad_norm": 0.487514466047287,
      "learning_rate": 1.9383439635232296e-05,
      "loss": 0.9055,
      "step": 832
    },
    {
      "epoch": 0.13966257990149847,
      "grad_norm": 0.4499186873435974,
      "learning_rate": 1.9381560882282354e-05,
      "loss": 0.991,
      "step": 833
    },
    {
      "epoch": 0.13983024206224456,
      "grad_norm": 0.43431034684181213,
      "learning_rate": 1.9379679362593937e-05,
      "loss": 0.9932,
      "step": 834
    },
    {
      "epoch": 0.13999790422299066,
      "grad_norm": 0.7048651576042175,
      "learning_rate": 1.937779507672193e-05,
      "loss": 0.9272,
      "step": 835
    },
    {
      "epoch": 0.14016556638373678,
      "grad_norm": 0.9253151416778564,
      "learning_rate": 1.9375908025222033e-05,
      "loss": 0.9153,
      "step": 836
    },
    {
      "epoch": 0.14033322854448288,
      "grad_norm": 0.9569368362426758,
      "learning_rate": 1.9374018208650755e-05,
      "loss": 0.9495,
      "step": 837
    },
    {
      "epoch": 0.14050089070522898,
      "grad_norm": 1.2080894708633423,
      "learning_rate": 1.9372125627565435e-05,
      "loss": 0.9293,
      "step": 838
    },
    {
      "epoch": 0.14066855286597507,
      "grad_norm": 0.41559672355651855,
      "learning_rate": 1.937023028252421e-05,
      "loss": 0.8253,
      "step": 839
    },
    {
      "epoch": 0.14083621502672117,
      "grad_norm": 1.0294303894042969,
      "learning_rate": 1.9368332174086046e-05,
      "loss": 1.0341,
      "step": 840
    },
    {
      "epoch": 0.14100387718746726,
      "grad_norm": 0.47193440794944763,
      "learning_rate": 1.936643130281072e-05,
      "loss": 0.9289,
      "step": 841
    },
    {
      "epoch": 0.14117153934821336,
      "grad_norm": 0.499761164188385,
      "learning_rate": 1.9364527669258815e-05,
      "loss": 0.9112,
      "step": 842
    },
    {
      "epoch": 0.14133920150895946,
      "grad_norm": 0.40940871834754944,
      "learning_rate": 1.9362621273991747e-05,
      "loss": 0.8562,
      "step": 843
    },
    {
      "epoch": 0.14150686366970555,
      "grad_norm": 0.897140622138977,
      "learning_rate": 1.936071211757173e-05,
      "loss": 0.9378,
      "step": 844
    },
    {
      "epoch": 0.14167452583045165,
      "grad_norm": 0.6589300632476807,
      "learning_rate": 1.935880020056179e-05,
      "loss": 0.8975,
      "step": 845
    },
    {
      "epoch": 0.14184218799119774,
      "grad_norm": 0.5597459077835083,
      "learning_rate": 1.935688552352579e-05,
      "loss": 0.9226,
      "step": 846
    },
    {
      "epoch": 0.14200985015194384,
      "grad_norm": 0.7280226945877075,
      "learning_rate": 1.9354968087028383e-05,
      "loss": 0.8995,
      "step": 847
    },
    {
      "epoch": 0.14217751231268994,
      "grad_norm": 0.8181997537612915,
      "learning_rate": 1.9353047891635044e-05,
      "loss": 1.011,
      "step": 848
    },
    {
      "epoch": 0.14234517447343603,
      "grad_norm": 0.5827728509902954,
      "learning_rate": 1.9351124937912065e-05,
      "loss": 1.0122,
      "step": 849
    },
    {
      "epoch": 0.14251283663418213,
      "grad_norm": 0.47748494148254395,
      "learning_rate": 1.9349199226426552e-05,
      "loss": 1.0305,
      "step": 850
    },
    {
      "epoch": 0.14268049879492822,
      "grad_norm": 0.40899279713630676,
      "learning_rate": 1.9347270757746416e-05,
      "loss": 0.9256,
      "step": 851
    },
    {
      "epoch": 0.14284816095567432,
      "grad_norm": 0.4111090302467346,
      "learning_rate": 1.9345339532440387e-05,
      "loss": 0.9297,
      "step": 852
    },
    {
      "epoch": 0.14301582311642042,
      "grad_norm": 0.48824769258499146,
      "learning_rate": 1.9343405551078008e-05,
      "loss": 0.9565,
      "step": 853
    },
    {
      "epoch": 0.1431834852771665,
      "grad_norm": 0.5168846249580383,
      "learning_rate": 1.9341468814229635e-05,
      "loss": 0.9212,
      "step": 854
    },
    {
      "epoch": 0.1433511474379126,
      "grad_norm": 0.5405130386352539,
      "learning_rate": 1.9339529322466432e-05,
      "loss": 1.0312,
      "step": 855
    },
    {
      "epoch": 0.1435188095986587,
      "grad_norm": 0.4492095112800598,
      "learning_rate": 1.9337587076360387e-05,
      "loss": 0.865,
      "step": 856
    },
    {
      "epoch": 0.1436864717594048,
      "grad_norm": 0.562528133392334,
      "learning_rate": 1.9335642076484283e-05,
      "loss": 0.9668,
      "step": 857
    },
    {
      "epoch": 0.1438541339201509,
      "grad_norm": 0.6639419198036194,
      "learning_rate": 1.9333694323411732e-05,
      "loss": 0.9625,
      "step": 858
    },
    {
      "epoch": 0.144021796080897,
      "grad_norm": 0.42684265971183777,
      "learning_rate": 1.933174381771715e-05,
      "loss": 0.855,
      "step": 859
    },
    {
      "epoch": 0.1441894582416431,
      "grad_norm": 0.6701211333274841,
      "learning_rate": 1.932979055997576e-05,
      "loss": 1.0239,
      "step": 860
    },
    {
      "epoch": 0.14435712040238918,
      "grad_norm": 0.619564950466156,
      "learning_rate": 1.9327834550763605e-05,
      "loss": 0.9425,
      "step": 861
    },
    {
      "epoch": 0.14452478256313528,
      "grad_norm": 0.6924566626548767,
      "learning_rate": 1.9325875790657537e-05,
      "loss": 1.0193,
      "step": 862
    },
    {
      "epoch": 0.14469244472388137,
      "grad_norm": 0.4808378219604492,
      "learning_rate": 1.9323914280235218e-05,
      "loss": 0.9406,
      "step": 863
    },
    {
      "epoch": 0.14486010688462747,
      "grad_norm": 0.797642171382904,
      "learning_rate": 1.9321950020075117e-05,
      "loss": 0.9621,
      "step": 864
    },
    {
      "epoch": 0.14502776904537357,
      "grad_norm": 0.5004295110702515,
      "learning_rate": 1.9319983010756527e-05,
      "loss": 0.848,
      "step": 865
    },
    {
      "epoch": 0.14519543120611966,
      "grad_norm": 0.5292854905128479,
      "learning_rate": 1.931801325285954e-05,
      "loss": 0.9103,
      "step": 866
    },
    {
      "epoch": 0.14536309336686576,
      "grad_norm": 0.5634934306144714,
      "learning_rate": 1.931604074696506e-05,
      "loss": 0.9064,
      "step": 867
    },
    {
      "epoch": 0.14553075552761185,
      "grad_norm": 0.5075427889823914,
      "learning_rate": 1.9314065493654806e-05,
      "loss": 1.054,
      "step": 868
    },
    {
      "epoch": 0.14569841768835795,
      "grad_norm": 0.46562308073043823,
      "learning_rate": 1.9312087493511304e-05,
      "loss": 0.8924,
      "step": 869
    },
    {
      "epoch": 0.14586607984910405,
      "grad_norm": 0.6002721786499023,
      "learning_rate": 1.931010674711789e-05,
      "loss": 0.9316,
      "step": 870
    },
    {
      "epoch": 0.14603374200985014,
      "grad_norm": 0.5705669522285461,
      "learning_rate": 1.930812325505871e-05,
      "loss": 0.903,
      "step": 871
    },
    {
      "epoch": 0.14620140417059624,
      "grad_norm": 0.43029841780662537,
      "learning_rate": 1.930613701791872e-05,
      "loss": 0.9611,
      "step": 872
    },
    {
      "epoch": 0.14636906633134233,
      "grad_norm": 0.4147214889526367,
      "learning_rate": 1.930414803628369e-05,
      "loss": 0.8383,
      "step": 873
    },
    {
      "epoch": 0.14653672849208843,
      "grad_norm": 0.6177957653999329,
      "learning_rate": 1.930215631074019e-05,
      "loss": 0.9271,
      "step": 874
    },
    {
      "epoch": 0.14670439065283453,
      "grad_norm": 0.7204621434211731,
      "learning_rate": 1.930016184187561e-05,
      "loss": 0.9782,
      "step": 875
    },
    {
      "epoch": 0.14687205281358062,
      "grad_norm": 0.5664174556732178,
      "learning_rate": 1.9298164630278137e-05,
      "loss": 0.9013,
      "step": 876
    },
    {
      "epoch": 0.14703971497432672,
      "grad_norm": 0.5182133316993713,
      "learning_rate": 1.929616467653678e-05,
      "loss": 0.9522,
      "step": 877
    },
    {
      "epoch": 0.14720737713507281,
      "grad_norm": 0.5646714568138123,
      "learning_rate": 1.929416198124135e-05,
      "loss": 0.8977,
      "step": 878
    },
    {
      "epoch": 0.14737503929581894,
      "grad_norm": 0.4593978524208069,
      "learning_rate": 1.9292156544982456e-05,
      "loss": 0.8403,
      "step": 879
    },
    {
      "epoch": 0.14754270145656503,
      "grad_norm": 0.49368441104888916,
      "learning_rate": 1.929014836835154e-05,
      "loss": 0.957,
      "step": 880
    },
    {
      "epoch": 0.14771036361731113,
      "grad_norm": 0.7045102715492249,
      "learning_rate": 1.928813745194083e-05,
      "loss": 0.9834,
      "step": 881
    },
    {
      "epoch": 0.14787802577805723,
      "grad_norm": 0.5111392140388489,
      "learning_rate": 1.9286123796343373e-05,
      "loss": 0.8888,
      "step": 882
    },
    {
      "epoch": 0.14804568793880332,
      "grad_norm": 0.3912787437438965,
      "learning_rate": 1.9284107402153022e-05,
      "loss": 0.8511,
      "step": 883
    },
    {
      "epoch": 0.14821335009954942,
      "grad_norm": 0.665562093257904,
      "learning_rate": 1.928208826996443e-05,
      "loss": 1.0157,
      "step": 884
    },
    {
      "epoch": 0.14838101226029551,
      "grad_norm": 0.653226375579834,
      "learning_rate": 1.9280066400373073e-05,
      "loss": 0.9063,
      "step": 885
    },
    {
      "epoch": 0.1485486744210416,
      "grad_norm": 0.42588213086128235,
      "learning_rate": 1.9278041793975216e-05,
      "loss": 1.0176,
      "step": 886
    },
    {
      "epoch": 0.1487163365817877,
      "grad_norm": 0.465458482503891,
      "learning_rate": 1.927601445136795e-05,
      "loss": 0.9025,
      "step": 887
    },
    {
      "epoch": 0.1488839987425338,
      "grad_norm": 0.505429208278656,
      "learning_rate": 1.927398437314916e-05,
      "loss": 0.9376,
      "step": 888
    },
    {
      "epoch": 0.1490516609032799,
      "grad_norm": 0.43974682688713074,
      "learning_rate": 1.9271951559917537e-05,
      "loss": 1.0323,
      "step": 889
    },
    {
      "epoch": 0.149219323064026,
      "grad_norm": 0.5050548911094666,
      "learning_rate": 1.9269916012272586e-05,
      "loss": 0.9432,
      "step": 890
    },
    {
      "epoch": 0.1493869852247721,
      "grad_norm": 0.5680622458457947,
      "learning_rate": 1.9267877730814615e-05,
      "loss": 0.9627,
      "step": 891
    },
    {
      "epoch": 0.14955464738551819,
      "grad_norm": 0.43889379501342773,
      "learning_rate": 1.9265836716144744e-05,
      "loss": 0.9195,
      "step": 892
    },
    {
      "epoch": 0.14972230954626428,
      "grad_norm": 0.4966212511062622,
      "learning_rate": 1.9263792968864882e-05,
      "loss": 0.9697,
      "step": 893
    },
    {
      "epoch": 0.14988997170701038,
      "grad_norm": 0.44300034642219543,
      "learning_rate": 1.9261746489577767e-05,
      "loss": 1.0088,
      "step": 894
    },
    {
      "epoch": 0.15005763386775647,
      "grad_norm": 0.6189342141151428,
      "learning_rate": 1.9259697278886926e-05,
      "loss": 0.9452,
      "step": 895
    },
    {
      "epoch": 0.15022529602850257,
      "grad_norm": 0.5580673813819885,
      "learning_rate": 1.9257645337396692e-05,
      "loss": 0.8635,
      "step": 896
    },
    {
      "epoch": 0.15039295818924867,
      "grad_norm": 0.47836774587631226,
      "learning_rate": 1.9255590665712214e-05,
      "loss": 0.9098,
      "step": 897
    },
    {
      "epoch": 0.15056062034999476,
      "grad_norm": 0.5996960997581482,
      "learning_rate": 1.9253533264439436e-05,
      "loss": 0.9834,
      "step": 898
    },
    {
      "epoch": 0.15072828251074086,
      "grad_norm": 0.4515504240989685,
      "learning_rate": 1.9251473134185116e-05,
      "loss": 0.9444,
      "step": 899
    },
    {
      "epoch": 0.15089594467148695,
      "grad_norm": 0.41490310430526733,
      "learning_rate": 1.924941027555681e-05,
      "loss": 1.0415,
      "step": 900
    },
    {
      "epoch": 0.15106360683223305,
      "grad_norm": 0.40346625447273254,
      "learning_rate": 1.924734468916288e-05,
      "loss": 0.903,
      "step": 901
    },
    {
      "epoch": 0.15123126899297915,
      "grad_norm": 0.5119394063949585,
      "learning_rate": 1.9245276375612493e-05,
      "loss": 0.8893,
      "step": 902
    },
    {
      "epoch": 0.15139893115372524,
      "grad_norm": 0.4831201136112213,
      "learning_rate": 1.924320533551562e-05,
      "loss": 1.0228,
      "step": 903
    },
    {
      "epoch": 0.15156659331447134,
      "grad_norm": 0.4201967418193817,
      "learning_rate": 1.9241131569483035e-05,
      "loss": 0.893,
      "step": 904
    },
    {
      "epoch": 0.15173425547521743,
      "grad_norm": 0.5925395488739014,
      "learning_rate": 1.923905507812632e-05,
      "loss": 0.9479,
      "step": 905
    },
    {
      "epoch": 0.15190191763596353,
      "grad_norm": 0.45901793241500854,
      "learning_rate": 1.923697586205786e-05,
      "loss": 0.861,
      "step": 906
    },
    {
      "epoch": 0.15206957979670963,
      "grad_norm": 0.510019063949585,
      "learning_rate": 1.9234893921890834e-05,
      "loss": 0.8774,
      "step": 907
    },
    {
      "epoch": 0.15223724195745572,
      "grad_norm": 0.6654001474380493,
      "learning_rate": 1.923280925823924e-05,
      "loss": 0.9092,
      "step": 908
    },
    {
      "epoch": 0.15240490411820182,
      "grad_norm": 0.49889883399009705,
      "learning_rate": 1.9230721871717862e-05,
      "loss": 0.8707,
      "step": 909
    },
    {
      "epoch": 0.1525725662789479,
      "grad_norm": 0.5877993106842041,
      "learning_rate": 1.9228631762942307e-05,
      "loss": 1.0557,
      "step": 910
    },
    {
      "epoch": 0.152740228439694,
      "grad_norm": 0.3399173319339752,
      "learning_rate": 1.9226538932528967e-05,
      "loss": 0.9035,
      "step": 911
    },
    {
      "epoch": 0.1529078906004401,
      "grad_norm": 0.3397526741027832,
      "learning_rate": 1.9224443381095047e-05,
      "loss": 1.0377,
      "step": 912
    },
    {
      "epoch": 0.1530755527611862,
      "grad_norm": 0.5500420928001404,
      "learning_rate": 1.9222345109258546e-05,
      "loss": 0.9871,
      "step": 913
    },
    {
      "epoch": 0.1532432149219323,
      "grad_norm": 0.6797765493392944,
      "learning_rate": 1.9220244117638277e-05,
      "loss": 0.8941,
      "step": 914
    },
    {
      "epoch": 0.1534108770826784,
      "grad_norm": 0.3991592526435852,
      "learning_rate": 1.9218140406853846e-05,
      "loss": 0.9538,
      "step": 915
    },
    {
      "epoch": 0.1535785392434245,
      "grad_norm": 0.43551596999168396,
      "learning_rate": 1.9216033977525658e-05,
      "loss": 0.9516,
      "step": 916
    },
    {
      "epoch": 0.15374620140417058,
      "grad_norm": 0.4320926070213318,
      "learning_rate": 1.9213924830274934e-05,
      "loss": 0.8031,
      "step": 917
    },
    {
      "epoch": 0.15391386356491668,
      "grad_norm": 0.3962501287460327,
      "learning_rate": 1.921181296572368e-05,
      "loss": 1.0334,
      "step": 918
    },
    {
      "epoch": 0.15408152572566278,
      "grad_norm": 0.484610378742218,
      "learning_rate": 1.9209698384494714e-05,
      "loss": 0.9991,
      "step": 919
    },
    {
      "epoch": 0.15424918788640887,
      "grad_norm": 0.5024212598800659,
      "learning_rate": 1.9207581087211654e-05,
      "loss": 0.8489,
      "step": 920
    },
    {
      "epoch": 0.154416850047155,
      "grad_norm": 0.4835463762283325,
      "learning_rate": 1.9205461074498917e-05,
      "loss": 0.905,
      "step": 921
    },
    {
      "epoch": 0.1545845122079011,
      "grad_norm": 0.3671019673347473,
      "learning_rate": 1.9203338346981718e-05,
      "loss": 0.9507,
      "step": 922
    },
    {
      "epoch": 0.1547521743686472,
      "grad_norm": 0.7113335132598877,
      "learning_rate": 1.9201212905286074e-05,
      "loss": 0.8627,
      "step": 923
    },
    {
      "epoch": 0.15491983652939328,
      "grad_norm": 0.4007314145565033,
      "learning_rate": 1.919908475003881e-05,
      "loss": 0.982,
      "step": 924
    },
    {
      "epoch": 0.15508749869013938,
      "grad_norm": 0.48794025182724,
      "learning_rate": 1.9196953881867544e-05,
      "loss": 1.0267,
      "step": 925
    },
    {
      "epoch": 0.15525516085088548,
      "grad_norm": 0.7895486950874329,
      "learning_rate": 1.9194820301400695e-05,
      "loss": 0.9659,
      "step": 926
    },
    {
      "epoch": 0.15542282301163157,
      "grad_norm": 0.6803207397460938,
      "learning_rate": 1.919268400926748e-05,
      "loss": 0.9211,
      "step": 927
    },
    {
      "epoch": 0.15559048517237767,
      "grad_norm": 0.41591545939445496,
      "learning_rate": 1.9190545006097917e-05,
      "loss": 0.9046,
      "step": 928
    },
    {
      "epoch": 0.15575814733312376,
      "grad_norm": 0.5918329954147339,
      "learning_rate": 1.918840329252283e-05,
      "loss": 0.9703,
      "step": 929
    },
    {
      "epoch": 0.15592580949386986,
      "grad_norm": 0.6053893566131592,
      "learning_rate": 1.9186258869173834e-05,
      "loss": 0.9772,
      "step": 930
    },
    {
      "epoch": 0.15609347165461596,
      "grad_norm": 0.47089946269989014,
      "learning_rate": 1.9184111736683343e-05,
      "loss": 0.9331,
      "step": 931
    },
    {
      "epoch": 0.15626113381536205,
      "grad_norm": 0.4194715619087219,
      "learning_rate": 1.918196189568458e-05,
      "loss": 0.9214,
      "step": 932
    },
    {
      "epoch": 0.15642879597610815,
      "grad_norm": 0.3285825252532959,
      "learning_rate": 1.917980934681155e-05,
      "loss": 0.9413,
      "step": 933
    },
    {
      "epoch": 0.15659645813685424,
      "grad_norm": 0.43966519832611084,
      "learning_rate": 1.9177654090699074e-05,
      "loss": 0.8261,
      "step": 934
    },
    {
      "epoch": 0.15676412029760034,
      "grad_norm": 0.42831242084503174,
      "learning_rate": 1.9175496127982764e-05,
      "loss": 0.9752,
      "step": 935
    },
    {
      "epoch": 0.15693178245834644,
      "grad_norm": 0.34641575813293457,
      "learning_rate": 1.9173335459299025e-05,
      "loss": 0.9415,
      "step": 936
    },
    {
      "epoch": 0.15709944461909253,
      "grad_norm": 0.30951234698295593,
      "learning_rate": 1.9171172085285064e-05,
      "loss": 0.8722,
      "step": 937
    },
    {
      "epoch": 0.15726710677983863,
      "grad_norm": 0.4109880328178406,
      "learning_rate": 1.9169006006578897e-05,
      "loss": 0.942,
      "step": 938
    },
    {
      "epoch": 0.15743476894058472,
      "grad_norm": 0.34938591718673706,
      "learning_rate": 1.9166837223819317e-05,
      "loss": 0.8811,
      "step": 939
    },
    {
      "epoch": 0.15760243110133082,
      "grad_norm": 0.42489537596702576,
      "learning_rate": 1.9164665737645927e-05,
      "loss": 0.9891,
      "step": 940
    },
    {
      "epoch": 0.15777009326207692,
      "grad_norm": 0.43413758277893066,
      "learning_rate": 1.9162491548699128e-05,
      "loss": 0.8942,
      "step": 941
    },
    {
      "epoch": 0.157937755422823,
      "grad_norm": 0.5286901593208313,
      "learning_rate": 1.9160314657620116e-05,
      "loss": 1.0414,
      "step": 942
    },
    {
      "epoch": 0.1581054175835691,
      "grad_norm": 0.4752057194709778,
      "learning_rate": 1.9158135065050878e-05,
      "loss": 0.8457,
      "step": 943
    },
    {
      "epoch": 0.1582730797443152,
      "grad_norm": 0.4286821186542511,
      "learning_rate": 1.9155952771634207e-05,
      "loss": 0.9238,
      "step": 944
    },
    {
      "epoch": 0.1584407419050613,
      "grad_norm": 0.5006504654884338,
      "learning_rate": 1.9153767778013686e-05,
      "loss": 0.9863,
      "step": 945
    },
    {
      "epoch": 0.1586084040658074,
      "grad_norm": 0.9059145450592041,
      "learning_rate": 1.91515800848337e-05,
      "loss": 0.9411,
      "step": 946
    },
    {
      "epoch": 0.1587760662265535,
      "grad_norm": 0.663236141204834,
      "learning_rate": 1.9149389692739424e-05,
      "loss": 0.8345,
      "step": 947
    },
    {
      "epoch": 0.1589437283872996,
      "grad_norm": 0.4279334843158722,
      "learning_rate": 1.9147196602376833e-05,
      "loss": 0.9397,
      "step": 948
    },
    {
      "epoch": 0.15911139054804568,
      "grad_norm": 0.5905395746231079,
      "learning_rate": 1.9145000814392696e-05,
      "loss": 0.9753,
      "step": 949
    },
    {
      "epoch": 0.15927905270879178,
      "grad_norm": 0.5215145945549011,
      "learning_rate": 1.9142802329434574e-05,
      "loss": 0.8985,
      "step": 950
    },
    {
      "epoch": 0.15944671486953788,
      "grad_norm": 0.5803624987602234,
      "learning_rate": 1.9140601148150833e-05,
      "loss": 0.9656,
      "step": 951
    },
    {
      "epoch": 0.15961437703028397,
      "grad_norm": 0.4757272005081177,
      "learning_rate": 1.913839727119063e-05,
      "loss": 0.8953,
      "step": 952
    },
    {
      "epoch": 0.15978203919103007,
      "grad_norm": 0.4840550720691681,
      "learning_rate": 1.9136190699203913e-05,
      "loss": 0.9365,
      "step": 953
    },
    {
      "epoch": 0.15994970135177616,
      "grad_norm": 0.4999079704284668,
      "learning_rate": 1.9133981432841424e-05,
      "loss": 0.9202,
      "step": 954
    },
    {
      "epoch": 0.16011736351252226,
      "grad_norm": 0.5819324851036072,
      "learning_rate": 1.913176947275471e-05,
      "loss": 0.9156,
      "step": 955
    },
    {
      "epoch": 0.16028502567326836,
      "grad_norm": 0.6553800106048584,
      "learning_rate": 1.9129554819596098e-05,
      "loss": 0.9624,
      "step": 956
    },
    {
      "epoch": 0.16045268783401445,
      "grad_norm": 0.4883859157562256,
      "learning_rate": 1.912733747401872e-05,
      "loss": 0.9026,
      "step": 957
    },
    {
      "epoch": 0.16062034999476055,
      "grad_norm": 0.4642502963542938,
      "learning_rate": 1.91251174366765e-05,
      "loss": 0.9512,
      "step": 958
    },
    {
      "epoch": 0.16078801215550664,
      "grad_norm": 0.38176947832107544,
      "learning_rate": 1.9122894708224152e-05,
      "loss": 0.8772,
      "step": 959
    },
    {
      "epoch": 0.16095567431625274,
      "grad_norm": 0.4411686062812805,
      "learning_rate": 1.912066928931719e-05,
      "loss": 0.8728,
      "step": 960
    },
    {
      "epoch": 0.16112333647699884,
      "grad_norm": 0.6528915762901306,
      "learning_rate": 1.9118441180611912e-05,
      "loss": 0.9028,
      "step": 961
    },
    {
      "epoch": 0.16129099863774493,
      "grad_norm": 0.4659099578857422,
      "learning_rate": 1.911621038276542e-05,
      "loss": 0.9866,
      "step": 962
    },
    {
      "epoch": 0.16145866079849103,
      "grad_norm": 0.6679427623748779,
      "learning_rate": 1.9113976896435603e-05,
      "loss": 0.8638,
      "step": 963
    },
    {
      "epoch": 0.16162632295923715,
      "grad_norm": 0.48949748277664185,
      "learning_rate": 1.9111740722281137e-05,
      "loss": 1.0119,
      "step": 964
    },
    {
      "epoch": 0.16179398511998325,
      "grad_norm": 0.5931878685951233,
      "learning_rate": 1.910950186096151e-05,
      "loss": 1.075,
      "step": 965
    },
    {
      "epoch": 0.16196164728072934,
      "grad_norm": 0.6350274085998535,
      "learning_rate": 1.910726031313698e-05,
      "loss": 0.9303,
      "step": 966
    },
    {
      "epoch": 0.16212930944147544,
      "grad_norm": 0.41805300116539,
      "learning_rate": 1.910501607946861e-05,
      "loss": 0.9238,
      "step": 967
    },
    {
      "epoch": 0.16229697160222153,
      "grad_norm": 0.5107613801956177,
      "learning_rate": 1.910276916061825e-05,
      "loss": 0.9229,
      "step": 968
    },
    {
      "epoch": 0.16246463376296763,
      "grad_norm": 0.44543734192848206,
      "learning_rate": 1.910051955724855e-05,
      "loss": 0.9944,
      "step": 969
    },
    {
      "epoch": 0.16263229592371373,
      "grad_norm": 0.5454578399658203,
      "learning_rate": 1.9098267270022943e-05,
      "loss": 0.9248,
      "step": 970
    },
    {
      "epoch": 0.16279995808445982,
      "grad_norm": 0.5694398283958435,
      "learning_rate": 1.9096012299605654e-05,
      "loss": 0.9687,
      "step": 971
    },
    {
      "epoch": 0.16296762024520592,
      "grad_norm": 0.5694398283958435,
      "learning_rate": 1.9096012299605654e-05,
      "loss": 0.9867,
      "step": 972
    },
    {
      "epoch": 0.16313528240595201,
      "grad_norm": 0.49293121695518494,
      "learning_rate": 1.90937546466617e-05,
      "loss": 0.8689,
      "step": 973
    },
    {
      "epoch": 0.1633029445666981,
      "grad_norm": 0.545941948890686,
      "learning_rate": 1.9091494311856903e-05,
      "loss": 0.9003,
      "step": 974
    },
    {
      "epoch": 0.1634706067274442,
      "grad_norm": 0.47112664580345154,
      "learning_rate": 1.9089231295857846e-05,
      "loss": 1.0334,
      "step": 975
    },
    {
      "epoch": 0.1636382688881903,
      "grad_norm": 0.49594390392303467,
      "learning_rate": 1.9086965599331938e-05,
      "loss": 0.8357,
      "step": 976
    },
    {
      "epoch": 0.1638059310489364,
      "grad_norm": 0.655060887336731,
      "learning_rate": 1.9084697222947344e-05,
      "loss": 0.9273,
      "step": 977
    },
    {
      "epoch": 0.1639735932096825,
      "grad_norm": 0.682763934135437,
      "learning_rate": 1.9082426167373053e-05,
      "loss": 0.9753,
      "step": 978
    },
    {
      "epoch": 0.1641412553704286,
      "grad_norm": 0.5177135467529297,
      "learning_rate": 1.9080152433278816e-05,
      "loss": 0.9426,
      "step": 979
    },
    {
      "epoch": 0.1643089175311747,
      "grad_norm": 0.6700261831283569,
      "learning_rate": 1.9077876021335188e-05,
      "loss": 1.0105,
      "step": 980
    },
    {
      "epoch": 0.16447657969192078,
      "grad_norm": 0.4972025752067566,
      "learning_rate": 1.907559693221351e-05,
      "loss": 0.9862,
      "step": 981
    },
    {
      "epoch": 0.16464424185266688,
      "grad_norm": 0.6664692759513855,
      "learning_rate": 1.9073315166585916e-05,
      "loss": 0.9289,
      "step": 982
    },
    {
      "epoch": 0.16481190401341297,
      "grad_norm": 0.6527049541473389,
      "learning_rate": 1.907103072512533e-05,
      "loss": 0.923,
      "step": 983
    },
    {
      "epoch": 0.16497956617415907,
      "grad_norm": 0.4614576995372772,
      "learning_rate": 1.9068743608505454e-05,
      "loss": 0.9671,
      "step": 984
    },
    {
      "epoch": 0.16514722833490517,
      "grad_norm": 0.43237075209617615,
      "learning_rate": 1.9066453817400793e-05,
      "loss": 0.9164,
      "step": 985
    },
    {
      "epoch": 0.16531489049565126,
      "grad_norm": 0.4315905272960663,
      "learning_rate": 1.9064161352486638e-05,
      "loss": 0.9128,
      "step": 986
    },
    {
      "epoch": 0.16548255265639736,
      "grad_norm": 0.5184720158576965,
      "learning_rate": 1.9061866214439054e-05,
      "loss": 0.8276,
      "step": 987
    },
    {
      "epoch": 0.16565021481714345,
      "grad_norm": 0.687303364276886,
      "learning_rate": 1.905956840393491e-05,
      "loss": 0.9414,
      "step": 988
    },
    {
      "epoch": 0.16581787697788955,
      "grad_norm": 0.472570538520813,
      "learning_rate": 1.9057267921651865e-05,
      "loss": 1.0119,
      "step": 989
    },
    {
      "epoch": 0.16598553913863565,
      "grad_norm": 0.5595750212669373,
      "learning_rate": 1.9054964768268354e-05,
      "loss": 0.9516,
      "step": 990
    },
    {
      "epoch": 0.16615320129938174,
      "grad_norm": 0.4138607084751129,
      "learning_rate": 1.905265894446361e-05,
      "loss": 0.9961,
      "step": 991
    },
    {
      "epoch": 0.16632086346012784,
      "grad_norm": 0.5740602016448975,
      "learning_rate": 1.905035045091764e-05,
      "loss": 0.9427,
      "step": 992
    },
    {
      "epoch": 0.16648852562087393,
      "grad_norm": 0.7034726142883301,
      "learning_rate": 1.9048039288311256e-05,
      "loss": 1.0249,
      "step": 993
    },
    {
      "epoch": 0.16665618778162003,
      "grad_norm": 0.567058265209198,
      "learning_rate": 1.904572545732605e-05,
      "loss": 0.9528,
      "step": 994
    },
    {
      "epoch": 0.16682384994236613,
      "grad_norm": 0.43043065071105957,
      "learning_rate": 1.904340895864439e-05,
      "loss": 0.9099,
      "step": 995
    },
    {
      "epoch": 0.16699151210311222,
      "grad_norm": 0.49964940547943115,
      "learning_rate": 1.9041089792949446e-05,
      "loss": 0.9711,
      "step": 996
    },
    {
      "epoch": 0.16715917426385832,
      "grad_norm": 0.7605466246604919,
      "learning_rate": 1.9038767960925168e-05,
      "loss": 0.9072,
      "step": 997
    },
    {
      "epoch": 0.1673268364246044,
      "grad_norm": 0.5706338882446289,
      "learning_rate": 1.9036443463256295e-05,
      "loss": 0.9962,
      "step": 998
    },
    {
      "epoch": 0.1674944985853505,
      "grad_norm": 0.7213522791862488,
      "learning_rate": 1.9034116300628346e-05,
      "loss": 0.9282,
      "step": 999
    },
    {
      "epoch": 0.1676621607460966,
      "grad_norm": 0.5890035629272461,
      "learning_rate": 1.9031786473727633e-05,
      "loss": 0.9342,
      "step": 1000
    },
    {
      "epoch": 0.1678298229068427,
      "grad_norm": 0.5635486841201782,
      "learning_rate": 1.902945398324125e-05,
      "loss": 0.9039,
      "step": 1001
    },
    {
      "epoch": 0.1679974850675888,
      "grad_norm": 0.5275377035140991,
      "learning_rate": 1.902711882985708e-05,
      "loss": 0.9226,
      "step": 1002
    },
    {
      "epoch": 0.1681651472283349,
      "grad_norm": 0.8132424354553223,
      "learning_rate": 1.902478101426378e-05,
      "loss": 0.8846,
      "step": 1003
    },
    {
      "epoch": 0.168332809389081,
      "grad_norm": 0.4180213510990143,
      "learning_rate": 1.902244053715081e-05,
      "loss": 0.9278,
      "step": 1004
    },
    {
      "epoch": 0.16850047154982709,
      "grad_norm": 0.520513117313385,
      "learning_rate": 1.9020097399208403e-05,
      "loss": 1.0251,
      "step": 1005
    },
    {
      "epoch": 0.16866813371057318,
      "grad_norm": 0.6166831851005554,
      "learning_rate": 1.9017751601127575e-05,
      "loss": 0.9955,
      "step": 1006
    },
    {
      "epoch": 0.1688357958713193,
      "grad_norm": 0.6414588689804077,
      "learning_rate": 1.901540314360014e-05,
      "loss": 0.8512,
      "step": 1007
    },
    {
      "epoch": 0.1690034580320654,
      "grad_norm": 0.5270276069641113,
      "learning_rate": 1.9013052027318677e-05,
      "loss": 0.951,
      "step": 1008
    },
    {
      "epoch": 0.1691711201928115,
      "grad_norm": 0.4635254740715027,
      "learning_rate": 1.9010698252976566e-05,
      "loss": 0.8338,
      "step": 1009
    },
    {
      "epoch": 0.1693387823535576,
      "grad_norm": 0.4911115765571594,
      "learning_rate": 1.9008341821267965e-05,
      "loss": 0.8872,
      "step": 1010
    },
    {
      "epoch": 0.1695064445143037,
      "grad_norm": 0.45328113436698914,
      "learning_rate": 1.9005982732887812e-05,
      "loss": 0.8935,
      "step": 1011
    },
    {
      "epoch": 0.16967410667504979,
      "grad_norm": 0.6181873083114624,
      "learning_rate": 1.9003620988531827e-05,
      "loss": 0.9842,
      "step": 1012
    },
    {
      "epoch": 0.16984176883579588,
      "grad_norm": 0.7268185615539551,
      "learning_rate": 1.9001256588896526e-05,
      "loss": 0.8971,
      "step": 1013
    },
    {
      "epoch": 0.17000943099654198,
      "grad_norm": 0.6407479047775269,
      "learning_rate": 1.8998889534679196e-05,
      "loss": 0.8073,
      "step": 1014
    },
    {
      "epoch": 0.17017709315728807,
      "grad_norm": 0.6562261581420898,
      "learning_rate": 1.8996519826577907e-05,
      "loss": 0.9595,
      "step": 1015
    },
    {
      "epoch": 0.17034475531803417,
      "grad_norm": 0.44220224022865295,
      "learning_rate": 1.8994147465291526e-05,
      "loss": 0.9315,
      "step": 1016
    },
    {
      "epoch": 0.17051241747878026,
      "grad_norm": 0.4324677288532257,
      "learning_rate": 1.8991772451519678e-05,
      "loss": 0.8583,
      "step": 1017
    },
    {
      "epoch": 0.17068007963952636,
      "grad_norm": 0.5509170889854431,
      "learning_rate": 1.8989394785962795e-05,
      "loss": 0.9205,
      "step": 1018
    },
    {
      "epoch": 0.17084774180027246,
      "grad_norm": 0.5897940397262573,
      "learning_rate": 1.8987014469322075e-05,
      "loss": 0.8807,
      "step": 1019
    },
    {
      "epoch": 0.17101540396101855,
      "grad_norm": 0.6917343735694885,
      "learning_rate": 1.89846315022995e-05,
      "loss": 1.0349,
      "step": 1020
    },
    {
      "epoch": 0.17118306612176465,
      "grad_norm": 0.44609513878822327,
      "learning_rate": 1.898224588559784e-05,
      "loss": 0.8867,
      "step": 1021
    },
    {
      "epoch": 0.17135072828251074,
      "grad_norm": 0.6210997700691223,
      "learning_rate": 1.8979857619920646e-05,
      "loss": 0.9736,
      "step": 1022
    },
    {
      "epoch": 0.17151839044325684,
      "grad_norm": 0.5329290628433228,
      "learning_rate": 1.8977466705972244e-05,
      "loss": 0.9225,
      "step": 1023
    },
    {
      "epoch": 0.17168605260400294,
      "grad_norm": 0.5414072871208191,
      "learning_rate": 1.8975073144457742e-05,
      "loss": 0.9704,
      "step": 1024
    },
    {
      "epoch": 0.17185371476474903,
      "grad_norm": 0.42674532532691956,
      "learning_rate": 1.8972676936083035e-05,
      "loss": 0.8851,
      "step": 1025
    },
    {
      "epoch": 0.17202137692549513,
      "grad_norm": 0.7599685788154602,
      "learning_rate": 1.8970278081554792e-05,
      "loss": 0.9601,
      "step": 1026
    },
    {
      "epoch": 0.17218903908624122,
      "grad_norm": 0.41676852107048035,
      "learning_rate": 1.8967876581580468e-05,
      "loss": 0.9191,
      "step": 1027
    },
    {
      "epoch": 0.17235670124698732,
      "grad_norm": 0.8608080148696899,
      "learning_rate": 1.8965472436868288e-05,
      "loss": 0.8661,
      "step": 1028
    },
    {
      "epoch": 0.17252436340773342,
      "grad_norm": 0.6654183864593506,
      "learning_rate": 1.896306564812727e-05,
      "loss": 0.9127,
      "step": 1029
    },
    {
      "epoch": 0.1726920255684795,
      "grad_norm": 0.6694945693016052,
      "learning_rate": 1.8960656216067204e-05,
      "loss": 1.0001,
      "step": 1030
    },
    {
      "epoch": 0.1728596877292256,
      "grad_norm": 0.7841224074363708,
      "learning_rate": 1.8958244141398664e-05,
      "loss": 0.9673,
      "step": 1031
    },
    {
      "epoch": 0.1730273498899717,
      "grad_norm": 0.47580161690711975,
      "learning_rate": 1.8955829424832995e-05,
      "loss": 0.928,
      "step": 1032
    },
    {
      "epoch": 0.1731950120507178,
      "grad_norm": 0.7303934097290039,
      "learning_rate": 1.895341206708233e-05,
      "loss": 0.8929,
      "step": 1033
    },
    {
      "epoch": 0.1733626742114639,
      "grad_norm": 0.8481614589691162,
      "learning_rate": 1.8950992068859584e-05,
      "loss": 0.9073,
      "step": 1034
    },
    {
      "epoch": 0.17353033637221,
      "grad_norm": 0.632799506187439,
      "learning_rate": 1.8948569430878435e-05,
      "loss": 1.0129,
      "step": 1035
    },
    {
      "epoch": 0.1736979985329561,
      "grad_norm": 0.7045972347259521,
      "learning_rate": 1.894614415385335e-05,
      "loss": 0.9188,
      "step": 1036
    },
    {
      "epoch": 0.17386566069370218,
      "grad_norm": 0.486174613237381,
      "learning_rate": 1.8943716238499575e-05,
      "loss": 1.0025,
      "step": 1037
    },
    {
      "epoch": 0.17403332285444828,
      "grad_norm": 0.3715253174304962,
      "learning_rate": 1.8941285685533136e-05,
      "loss": 0.9962,
      "step": 1038
    },
    {
      "epoch": 0.17420098501519438,
      "grad_norm": 0.38294509053230286,
      "learning_rate": 1.893885249567083e-05,
      "loss": 0.8867,
      "step": 1039
    },
    {
      "epoch": 0.17436864717594047,
      "grad_norm": 0.5372864603996277,
      "learning_rate": 1.8936416669630236e-05,
      "loss": 0.9098,
      "step": 1040
    },
    {
      "epoch": 0.17453630933668657,
      "grad_norm": 0.42620623111724854,
      "learning_rate": 1.8933978208129705e-05,
      "loss": 0.9706,
      "step": 1041
    },
    {
      "epoch": 0.17470397149743266,
      "grad_norm": 0.4851018488407135,
      "learning_rate": 1.8931537111888378e-05,
      "loss": 1.0786,
      "step": 1042
    },
    {
      "epoch": 0.17487163365817876,
      "grad_norm": 0.4234909117221832,
      "learning_rate": 1.8929093381626153e-05,
      "loss": 1.0391,
      "step": 1043
    },
    {
      "epoch": 0.17503929581892486,
      "grad_norm": 0.5287904739379883,
      "learning_rate": 1.8926647018063724e-05,
      "loss": 0.9263,
      "step": 1044
    },
    {
      "epoch": 0.17520695797967095,
      "grad_norm": 0.5778496861457825,
      "learning_rate": 1.8924198021922553e-05,
      "loss": 0.9885,
      "step": 1045
    },
    {
      "epoch": 0.17537462014041705,
      "grad_norm": 0.4014924466609955,
      "learning_rate": 1.8921746393924877e-05,
      "loss": 0.8806,
      "step": 1046
    },
    {
      "epoch": 0.17554228230116314,
      "grad_norm": 0.5376038551330566,
      "learning_rate": 1.8919292134793718e-05,
      "loss": 0.9444,
      "step": 1047
    },
    {
      "epoch": 0.17570994446190924,
      "grad_norm": 0.43011462688446045,
      "learning_rate": 1.8916835245252854e-05,
      "loss": 0.9794,
      "step": 1048
    },
    {
      "epoch": 0.17587760662265534,
      "grad_norm": 0.38411054015159607,
      "learning_rate": 1.891437572602687e-05,
      "loss": 0.9942,
      "step": 1049
    },
    {
      "epoch": 0.17604526878340146,
      "grad_norm": 0.46081578731536865,
      "learning_rate": 1.8911913577841096e-05,
      "loss": 0.9074,
      "step": 1050
    },
    {
      "epoch": 0.17621293094414756,
      "grad_norm": 0.3966897428035736,
      "learning_rate": 1.8909448801421653e-05,
      "loss": 0.9153,
      "step": 1051
    },
    {
      "epoch": 0.17638059310489365,
      "grad_norm": 0.40592241287231445,
      "learning_rate": 1.8906981397495434e-05,
      "loss": 0.9857,
      "step": 1052
    },
    {
      "epoch": 0.17654825526563975,
      "grad_norm": 0.47412779927253723,
      "learning_rate": 1.8904511366790106e-05,
      "loss": 0.959,
      "step": 1053
    },
    {
      "epoch": 0.17671591742638584,
      "grad_norm": 0.40784403681755066,
      "learning_rate": 1.8902038710034113e-05,
      "loss": 0.9044,
      "step": 1054
    },
    {
      "epoch": 0.17688357958713194,
      "grad_norm": 0.416376531124115,
      "learning_rate": 1.8899563427956674e-05,
      "loss": 0.9814,
      "step": 1055
    },
    {
      "epoch": 0.17705124174787804,
      "grad_norm": 0.5241764187812805,
      "learning_rate": 1.889708552128778e-05,
      "loss": 0.9262,
      "step": 1056
    },
    {
      "epoch": 0.17721890390862413,
      "grad_norm": 0.467724084854126,
      "learning_rate": 1.889460499075819e-05,
      "loss": 0.9341,
      "step": 1057
    },
    {
      "epoch": 0.17738656606937023,
      "grad_norm": 0.4077341854572296,
      "learning_rate": 1.8892121837099452e-05,
      "loss": 0.9989,
      "step": 1058
    },
    {
      "epoch": 0.17755422823011632,
      "grad_norm": 0.7256616950035095,
      "learning_rate": 1.8889636061043874e-05,
      "loss": 0.9983,
      "step": 1059
    },
    {
      "epoch": 0.17772189039086242,
      "grad_norm": 0.49016696214675903,
      "learning_rate": 1.8887147663324545e-05,
      "loss": 0.8761,
      "step": 1060
    },
    {
      "epoch": 0.17788955255160852,
      "grad_norm": 0.6254734396934509,
      "learning_rate": 1.8884656644675318e-05,
      "loss": 1.001,
      "step": 1061
    },
    {
      "epoch": 0.1780572147123546,
      "grad_norm": 0.5470951199531555,
      "learning_rate": 1.888216300583083e-05,
      "loss": 0.8073,
      "step": 1062
    },
    {
      "epoch": 0.1782248768731007,
      "grad_norm": 0.49495676159858704,
      "learning_rate": 1.887966674752649e-05,
      "loss": 0.9055,
      "step": 1063
    },
    {
      "epoch": 0.1783925390338468,
      "grad_norm": 0.4782699644565582,
      "learning_rate": 1.8877167870498466e-05,
      "loss": 1.0146,
      "step": 1064
    },
    {
      "epoch": 0.1785602011945929,
      "grad_norm": 0.8720697164535522,
      "learning_rate": 1.8874666375483713e-05,
      "loss": 0.8812,
      "step": 1065
    },
    {
      "epoch": 0.178727863355339,
      "grad_norm": 0.5451815724372864,
      "learning_rate": 1.8872162263219955e-05,
      "loss": 0.9154,
      "step": 1066
    },
    {
      "epoch": 0.1788955255160851,
      "grad_norm": 0.48717665672302246,
      "learning_rate": 1.886965553444568e-05,
      "loss": 0.9923,
      "step": 1067
    },
    {
      "epoch": 0.1790631876768312,
      "grad_norm": 0.4174070358276367,
      "learning_rate": 1.886714618990016e-05,
      "loss": 0.9578,
      "step": 1068
    },
    {
      "epoch": 0.17923084983757728,
      "grad_norm": 0.38399094343185425,
      "learning_rate": 1.8864634230323426e-05,
      "loss": 0.9129,
      "step": 1069
    },
    {
      "epoch": 0.17939851199832338,
      "grad_norm": 0.6723623871803284,
      "learning_rate": 1.886211965645629e-05,
      "loss": 0.9536,
      "step": 1070
    },
    {
      "epoch": 0.17956617415906947,
      "grad_norm": 0.5474948287010193,
      "learning_rate": 1.8859602469040327e-05,
      "loss": 0.9327,
      "step": 1071
    },
    {
      "epoch": 0.17973383631981557,
      "grad_norm": 0.6205623149871826,
      "learning_rate": 1.885708266881789e-05,
      "loss": 0.9177,
      "step": 1072
    },
    {
      "epoch": 0.17990149848056167,
      "grad_norm": 0.3894779086112976,
      "learning_rate": 1.8854560256532098e-05,
      "loss": 0.9197,
      "step": 1073
    },
    {
      "epoch": 0.18006916064130776,
      "grad_norm": 0.7599223256111145,
      "learning_rate": 1.8852035232926848e-05,
      "loss": 0.8938,
      "step": 1074
    },
    {
      "epoch": 0.18023682280205386,
      "grad_norm": 0.5655592679977417,
      "learning_rate": 1.884950759874679e-05,
      "loss": 0.8273,
      "step": 1075
    },
    {
      "epoch": 0.18040448496279995,
      "grad_norm": 0.5135934948921204,
      "learning_rate": 1.8846977354737356e-05,
      "loss": 0.9494,
      "step": 1076
    },
    {
      "epoch": 0.18057214712354605,
      "grad_norm": 0.7166675925254822,
      "learning_rate": 1.8844444501644755e-05,
      "loss": 0.8252,
      "step": 1077
    },
    {
      "epoch": 0.18073980928429215,
      "grad_norm": 0.7380692958831787,
      "learning_rate": 1.8841909040215947e-05,
      "loss": 0.8823,
      "step": 1078
    },
    {
      "epoch": 0.18090747144503824,
      "grad_norm": 0.6297410130500793,
      "learning_rate": 1.883937097119868e-05,
      "loss": 1.0003,
      "step": 1079
    },
    {
      "epoch": 0.18107513360578434,
      "grad_norm": 0.48514869809150696,
      "learning_rate": 1.883683029534145e-05,
      "loss": 0.9187,
      "step": 1080
    },
    {
      "epoch": 0.18124279576653043,
      "grad_norm": 0.586363673210144,
      "learning_rate": 1.8834287013393547e-05,
      "loss": 0.9495,
      "step": 1081
    },
    {
      "epoch": 0.18141045792727653,
      "grad_norm": 0.5589828491210938,
      "learning_rate": 1.883174112610501e-05,
      "loss": 0.9236,
      "step": 1082
    },
    {
      "epoch": 0.18157812008802263,
      "grad_norm": 0.5361699461936951,
      "learning_rate": 1.882919263422665e-05,
      "loss": 0.9043,
      "step": 1083
    },
    {
      "epoch": 0.18174578224876872,
      "grad_norm": 0.48993849754333496,
      "learning_rate": 1.8826641538510055e-05,
      "loss": 0.9285,
      "step": 1084
    },
    {
      "epoch": 0.18191344440951482,
      "grad_norm": 0.5430667996406555,
      "learning_rate": 1.8824087839707572e-05,
      "loss": 1.0018,
      "step": 1085
    },
    {
      "epoch": 0.18208110657026091,
      "grad_norm": 0.5209212899208069,
      "learning_rate": 1.8821531538572317e-05,
      "loss": 0.9375,
      "step": 1086
    },
    {
      "epoch": 0.182248768731007,
      "grad_norm": 0.4663066267967224,
      "learning_rate": 1.8818972635858174e-05,
      "loss": 0.9695,
      "step": 1087
    },
    {
      "epoch": 0.1824164308917531,
      "grad_norm": 0.4286949634552002,
      "learning_rate": 1.88164111323198e-05,
      "loss": 0.9385,
      "step": 1088
    },
    {
      "epoch": 0.1825840930524992,
      "grad_norm": 0.47834473848342896,
      "learning_rate": 1.8813847028712612e-05,
      "loss": 0.9313,
      "step": 1089
    },
    {
      "epoch": 0.1827517552132453,
      "grad_norm": 0.49863681197166443,
      "learning_rate": 1.8811280325792788e-05,
      "loss": 1.0002,
      "step": 1090
    },
    {
      "epoch": 0.1829194173739914,
      "grad_norm": 0.4201532304286957,
      "learning_rate": 1.8808711024317295e-05,
      "loss": 0.8969,
      "step": 1091
    },
    {
      "epoch": 0.1830870795347375,
      "grad_norm": 0.4612746238708496,
      "learning_rate": 1.880613912504384e-05,
      "loss": 0.9271,
      "step": 1092
    },
    {
      "epoch": 0.18325474169548361,
      "grad_norm": 0.4401395916938782,
      "learning_rate": 1.8803564628730916e-05,
      "loss": 0.8983,
      "step": 1093
    },
    {
      "epoch": 0.1834224038562297,
      "grad_norm": 0.4388650059700012,
      "learning_rate": 1.8800987536137767e-05,
      "loss": 0.8806,
      "step": 1094
    },
    {
      "epoch": 0.1835900660169758,
      "grad_norm": 0.6108338832855225,
      "learning_rate": 1.8798407848024415e-05,
      "loss": 0.9898,
      "step": 1095
    },
    {
      "epoch": 0.1837577281777219,
      "grad_norm": 0.5061869621276855,
      "learning_rate": 1.8795825565151637e-05,
      "loss": 0.8857,
      "step": 1096
    },
    {
      "epoch": 0.183925390338468,
      "grad_norm": 0.44790008664131165,
      "learning_rate": 1.8793240688280986e-05,
      "loss": 0.9221,
      "step": 1097
    },
    {
      "epoch": 0.1840930524992141,
      "grad_norm": 0.3312665820121765,
      "learning_rate": 1.8790653218174766e-05,
      "loss": 0.9456,
      "step": 1098
    },
    {
      "epoch": 0.1842607146599602,
      "grad_norm": 0.44701552391052246,
      "learning_rate": 1.8788063155596062e-05,
      "loss": 0.8792,
      "step": 1099
    },
    {
      "epoch": 0.18442837682070629,
      "grad_norm": 0.4169693887233734,
      "learning_rate": 1.8785470501308714e-05,
      "loss": 0.893,
      "step": 1100
    },
    {
      "epoch": 0.18459603898145238,
      "grad_norm": 0.4645668566226959,
      "learning_rate": 1.8782875256077322e-05,
      "loss": 0.9022,
      "step": 1101
    },
    {
      "epoch": 0.18476370114219848,
      "grad_norm": 0.39160627126693726,
      "learning_rate": 1.8780277420667262e-05,
      "loss": 0.9049,
      "step": 1102
    },
    {
      "epoch": 0.18493136330294457,
      "grad_norm": 0.5148125886917114,
      "learning_rate": 1.877767699584467e-05,
      "loss": 0.9455,
      "step": 1103
    },
    {
      "epoch": 0.18509902546369067,
      "grad_norm": 0.3958742022514343,
      "learning_rate": 1.8775073982376434e-05,
      "loss": 0.8458,
      "step": 1104
    },
    {
      "epoch": 0.18526668762443677,
      "grad_norm": 0.5639967918395996,
      "learning_rate": 1.8772468381030222e-05,
      "loss": 0.8476,
      "step": 1105
    },
    {
      "epoch": 0.18543434978518286,
      "grad_norm": 0.5112168788909912,
      "learning_rate": 1.876986019257446e-05,
      "loss": 0.846,
      "step": 1106
    },
    {
      "epoch": 0.18560201194592896,
      "grad_norm": 0.44852420687675476,
      "learning_rate": 1.876724941777833e-05,
      "loss": 0.974,
      "step": 1107
    },
    {
      "epoch": 0.18576967410667505,
      "grad_norm": 0.46299490332603455,
      "learning_rate": 1.876463605741178e-05,
      "loss": 0.9148,
      "step": 1108
    },
    {
      "epoch": 0.18593733626742115,
      "grad_norm": 0.7084307074546814,
      "learning_rate": 1.876202011224553e-05,
      "loss": 0.9549,
      "step": 1109
    },
    {
      "epoch": 0.18610499842816725,
      "grad_norm": 0.4639042019844055,
      "learning_rate": 1.8759401583051048e-05,
      "loss": 0.9632,
      "step": 1110
    },
    {
      "epoch": 0.18627266058891334,
      "grad_norm": 0.5113834738731384,
      "learning_rate": 1.8756780470600574e-05,
      "loss": 0.9011,
      "step": 1111
    },
    {
      "epoch": 0.18644032274965944,
      "grad_norm": 0.6037055850028992,
      "learning_rate": 1.8754156775667108e-05,
      "loss": 0.9357,
      "step": 1112
    },
    {
      "epoch": 0.18660798491040553,
      "grad_norm": 0.5472172498703003,
      "learning_rate": 1.8751530499024403e-05,
      "loss": 0.9359,
      "step": 1113
    },
    {
      "epoch": 0.18677564707115163,
      "grad_norm": 0.42744895815849304,
      "learning_rate": 1.8748901641446987e-05,
      "loss": 0.9693,
      "step": 1114
    },
    {
      "epoch": 0.18694330923189773,
      "grad_norm": 0.5853201150894165,
      "learning_rate": 1.874627020371014e-05,
      "loss": 0.8754,
      "step": 1115
    },
    {
      "epoch": 0.18711097139264382,
      "grad_norm": 0.5467081069946289,
      "learning_rate": 1.874363618658991e-05,
      "loss": 0.9653,
      "step": 1116
    },
    {
      "epoch": 0.18727863355338992,
      "grad_norm": 0.47121307253837585,
      "learning_rate": 1.87409995908631e-05,
      "loss": 0.9778,
      "step": 1117
    },
    {
      "epoch": 0.187446295714136,
      "grad_norm": 0.5530045628547668,
      "learning_rate": 1.8738360417307265e-05,
      "loss": 0.995,
      "step": 1118
    },
    {
      "epoch": 0.1876139578748821,
      "grad_norm": 0.5050699710845947,
      "learning_rate": 1.873571866670074e-05,
      "loss": 0.9575,
      "step": 1119
    },
    {
      "epoch": 0.1877816200356282,
      "grad_norm": 0.48247194290161133,
      "learning_rate": 1.873307433982261e-05,
      "loss": 0.992,
      "step": 1120
    },
    {
      "epoch": 0.1879492821963743,
      "grad_norm": 0.5888962149620056,
      "learning_rate": 1.8730427437452717e-05,
      "loss": 0.9089,
      "step": 1121
    },
    {
      "epoch": 0.1881169443571204,
      "grad_norm": 0.43505725264549255,
      "learning_rate": 1.8727777960371662e-05,
      "loss": 0.9586,
      "step": 1122
    },
    {
      "epoch": 0.1882846065178665,
      "grad_norm": 0.3793167471885681,
      "learning_rate": 1.8725125909360816e-05,
      "loss": 0.8689,
      "step": 1123
    },
    {
      "epoch": 0.1884522686786126,
      "grad_norm": 0.42412588000297546,
      "learning_rate": 1.8722471285202295e-05,
      "loss": 0.9496,
      "step": 1124
    },
    {
      "epoch": 0.18861993083935868,
      "grad_norm": 0.7520803213119507,
      "learning_rate": 1.871981408867898e-05,
      "loss": 0.9198,
      "step": 1125
    },
    {
      "epoch": 0.18878759300010478,
      "grad_norm": 0.4949362277984619,
      "learning_rate": 1.8717154320574522e-05,
      "loss": 0.8231,
      "step": 1126
    },
    {
      "epoch": 0.18895525516085088,
      "grad_norm": 0.5005122423171997,
      "learning_rate": 1.8714491981673307e-05,
      "loss": 0.9228,
      "step": 1127
    },
    {
      "epoch": 0.18912291732159697,
      "grad_norm": 0.5874606966972351,
      "learning_rate": 1.87118270727605e-05,
      "loss": 0.8301,
      "step": 1128
    },
    {
      "epoch": 0.18929057948234307,
      "grad_norm": 0.5143774747848511,
      "learning_rate": 1.870915959462201e-05,
      "loss": 0.9073,
      "step": 1129
    },
    {
      "epoch": 0.18945824164308916,
      "grad_norm": 0.922320544719696,
      "learning_rate": 1.8706489548044508e-05,
      "loss": 0.9736,
      "step": 1130
    },
    {
      "epoch": 0.18962590380383526,
      "grad_norm": 0.8636284470558167,
      "learning_rate": 1.8703816933815433e-05,
      "loss": 0.9078,
      "step": 1131
    },
    {
      "epoch": 0.18979356596458136,
      "grad_norm": 1.1362345218658447,
      "learning_rate": 1.8701141752722966e-05,
      "loss": 0.9348,
      "step": 1132
    },
    {
      "epoch": 0.18996122812532745,
      "grad_norm": 0.692942202091217,
      "learning_rate": 1.8698464005556047e-05,
      "loss": 0.8462,
      "step": 1133
    },
    {
      "epoch": 0.19012889028607355,
      "grad_norm": 0.6303272843360901,
      "learning_rate": 1.8695783693104386e-05,
      "loss": 0.9073,
      "step": 1134
    },
    {
      "epoch": 0.19029655244681964,
      "grad_norm": 0.502835214138031,
      "learning_rate": 1.8693100816158436e-05,
      "loss": 0.9079,
      "step": 1135
    },
    {
      "epoch": 0.19046421460756577,
      "grad_norm": 0.4576420485973358,
      "learning_rate": 1.869041537550941e-05,
      "loss": 0.9106,
      "step": 1136
    },
    {
      "epoch": 0.19063187676831186,
      "grad_norm": 0.8908449411392212,
      "learning_rate": 1.8687727371949275e-05,
      "loss": 0.9147,
      "step": 1137
    },
    {
      "epoch": 0.19079953892905796,
      "grad_norm": 1.049439787864685,
      "learning_rate": 1.868503680627076e-05,
      "loss": 1.0857,
      "step": 1138
    },
    {
      "epoch": 0.19096720108980406,
      "grad_norm": 0.6737858653068542,
      "learning_rate": 1.868234367926735e-05,
      "loss": 0.918,
      "step": 1139
    },
    {
      "epoch": 0.19113486325055015,
      "grad_norm": 0.5657393932342529,
      "learning_rate": 1.8679647991733268e-05,
      "loss": 0.9149,
      "step": 1140
    },
    {
      "epoch": 0.19130252541129625,
      "grad_norm": 0.5046471953392029,
      "learning_rate": 1.8676949744463523e-05,
      "loss": 0.9214,
      "step": 1141
    },
    {
      "epoch": 0.19147018757204234,
      "grad_norm": 0.5004892349243164,
      "learning_rate": 1.867424893825385e-05,
      "loss": 0.9502,
      "step": 1142
    },
    {
      "epoch": 0.19163784973278844,
      "grad_norm": 0.4683850109577179,
      "learning_rate": 1.8671545573900755e-05,
      "loss": 0.7789,
      "step": 1143
    },
    {
      "epoch": 0.19180551189353454,
      "grad_norm": 0.48335880041122437,
      "learning_rate": 1.866883965220149e-05,
      "loss": 0.8971,
      "step": 1144
    },
    {
      "epoch": 0.19197317405428063,
      "grad_norm": 0.3865186274051666,
      "learning_rate": 1.866613117395407e-05,
      "loss": 0.975,
      "step": 1145
    },
    {
      "epoch": 0.19214083621502673,
      "grad_norm": 0.44036880135536194,
      "learning_rate": 1.8663420139957252e-05,
      "loss": 0.9938,
      "step": 1146
    },
    {
      "epoch": 0.19230849837577282,
      "grad_norm": 0.5232735276222229,
      "learning_rate": 1.866070655101056e-05,
      "loss": 0.9507,
      "step": 1147
    },
    {
      "epoch": 0.19247616053651892,
      "grad_norm": 0.4845692217350006,
      "learning_rate": 1.8657990407914265e-05,
      "loss": 0.9477,
      "step": 1148
    },
    {
      "epoch": 0.19264382269726502,
      "grad_norm": 0.3743135929107666,
      "learning_rate": 1.865527171146938e-05,
      "loss": 0.9608,
      "step": 1149
    },
    {
      "epoch": 0.1928114848580111,
      "grad_norm": 0.7610765695571899,
      "learning_rate": 1.8652550462477697e-05,
      "loss": 0.9364,
      "step": 1150
    },
    {
      "epoch": 0.1929791470187572,
      "grad_norm": 0.35837090015411377,
      "learning_rate": 1.864982666174174e-05,
      "loss": 0.8771,
      "step": 1151
    },
    {
      "epoch": 0.1931468091795033,
      "grad_norm": 0.6423548460006714,
      "learning_rate": 1.864710031006479e-05,
      "loss": 0.9066,
      "step": 1152
    },
    {
      "epoch": 0.1933144713402494,
      "grad_norm": 0.3354455232620239,
      "learning_rate": 1.864437140825088e-05,
      "loss": 0.873,
      "step": 1153
    },
    {
      "epoch": 0.1934821335009955,
      "grad_norm": 0.48024141788482666,
      "learning_rate": 1.8641639957104805e-05,
      "loss": 0.9756,
      "step": 1154
    },
    {
      "epoch": 0.1936497956617416,
      "grad_norm": 0.7103967666625977,
      "learning_rate": 1.8638905957432097e-05,
      "loss": 0.9366,
      "step": 1155
    },
    {
      "epoch": 0.1938174578224877,
      "grad_norm": 0.5881415009498596,
      "learning_rate": 1.8636169410039046e-05,
      "loss": 0.9805,
      "step": 1156
    },
    {
      "epoch": 0.19398511998323378,
      "grad_norm": 0.5831557512283325,
      "learning_rate": 1.86334303157327e-05,
      "loss": 0.8257,
      "step": 1157
    },
    {
      "epoch": 0.19415278214397988,
      "grad_norm": 0.6262730956077576,
      "learning_rate": 1.8630688675320844e-05,
      "loss": 0.8544,
      "step": 1158
    },
    {
      "epoch": 0.19432044430472598,
      "grad_norm": 0.3318983018398285,
      "learning_rate": 1.8627944489612025e-05,
      "loss": 0.8899,
      "step": 1159
    },
    {
      "epoch": 0.19448810646547207,
      "grad_norm": 0.4775957465171814,
      "learning_rate": 1.862519775941554e-05,
      "loss": 0.9246,
      "step": 1160
    },
    {
      "epoch": 0.19465576862621817,
      "grad_norm": 0.5421665906906128,
      "learning_rate": 1.862244848554143e-05,
      "loss": 0.9443,
      "step": 1161
    },
    {
      "epoch": 0.19482343078696426,
      "grad_norm": 0.48920223116874695,
      "learning_rate": 1.8619696668800494e-05,
      "loss": 0.8853,
      "step": 1162
    },
    {
      "epoch": 0.19499109294771036,
      "grad_norm": 0.4863426089286804,
      "learning_rate": 1.8616942310004273e-05,
      "loss": 0.9421,
      "step": 1163
    },
    {
      "epoch": 0.19515875510845646,
      "grad_norm": 0.47465115785598755,
      "learning_rate": 1.8614185409965062e-05,
      "loss": 0.951,
      "step": 1164
    },
    {
      "epoch": 0.19532641726920255,
      "grad_norm": 0.6428936123847961,
      "learning_rate": 1.8611425969495908e-05,
      "loss": 0.8917,
      "step": 1165
    },
    {
      "epoch": 0.19549407942994865,
      "grad_norm": 0.48652878403663635,
      "learning_rate": 1.86086639894106e-05,
      "loss": 1.004,
      "step": 1166
    },
    {
      "epoch": 0.19566174159069474,
      "grad_norm": 0.6992968916893005,
      "learning_rate": 1.8605899470523683e-05,
      "loss": 0.8919,
      "step": 1167
    },
    {
      "epoch": 0.19582940375144084,
      "grad_norm": 0.8496128916740417,
      "learning_rate": 1.8603132413650443e-05,
      "loss": 0.8623,
      "step": 1168
    },
    {
      "epoch": 0.19599706591218694,
      "grad_norm": 0.5947402119636536,
      "learning_rate": 1.8600362819606927e-05,
      "loss": 0.7653,
      "step": 1169
    },
    {
      "epoch": 0.19616472807293303,
      "grad_norm": 0.6545706391334534,
      "learning_rate": 1.859759068920992e-05,
      "loss": 0.8367,
      "step": 1170
    },
    {
      "epoch": 0.19633239023367913,
      "grad_norm": 0.7118186950683594,
      "learning_rate": 1.8594816023276954e-05,
      "loss": 0.9847,
      "step": 1171
    },
    {
      "epoch": 0.19650005239442522,
      "grad_norm": 0.4567353427410126,
      "learning_rate": 1.8592038822626314e-05,
      "loss": 0.9423,
      "step": 1172
    },
    {
      "epoch": 0.19666771455517132,
      "grad_norm": 0.5413332581520081,
      "learning_rate": 1.8589259088077033e-05,
      "loss": 0.9664,
      "step": 1173
    },
    {
      "epoch": 0.19683537671591742,
      "grad_norm": 0.4965800940990448,
      "learning_rate": 1.858647682044889e-05,
      "loss": 0.8317,
      "step": 1174
    },
    {
      "epoch": 0.1970030388766635,
      "grad_norm": 0.4861052930355072,
      "learning_rate": 1.8583692020562404e-05,
      "loss": 0.9442,
      "step": 1175
    },
    {
      "epoch": 0.1971707010374096,
      "grad_norm": 0.8544037342071533,
      "learning_rate": 1.8580904689238855e-05,
      "loss": 0.9905,
      "step": 1176
    },
    {
      "epoch": 0.1973383631981557,
      "grad_norm": 0.49847543239593506,
      "learning_rate": 1.8578114827300253e-05,
      "loss": 0.8249,
      "step": 1177
    },
    {
      "epoch": 0.19750602535890183,
      "grad_norm": 0.3586089611053467,
      "learning_rate": 1.8575322435569368e-05,
      "loss": 0.8953,
      "step": 1178
    },
    {
      "epoch": 0.19767368751964792,
      "grad_norm": 0.4339447021484375,
      "learning_rate": 1.857252751486971e-05,
      "loss": 0.969,
      "step": 1179
    },
    {
      "epoch": 0.19784134968039402,
      "grad_norm": 0.5465442538261414,
      "learning_rate": 1.856973006602554e-05,
      "loss": 1.0242,
      "step": 1180
    },
    {
      "epoch": 0.19800901184114011,
      "grad_norm": 0.45056402683258057,
      "learning_rate": 1.8566930089861853e-05,
      "loss": 0.9243,
      "step": 1181
    },
    {
      "epoch": 0.1981766740018862,
      "grad_norm": 0.5189386010169983,
      "learning_rate": 1.85641275872044e-05,
      "loss": 0.9065,
      "step": 1182
    },
    {
      "epoch": 0.1983443361626323,
      "grad_norm": 0.3600766062736511,
      "learning_rate": 1.8561322558879672e-05,
      "loss": 0.8328,
      "step": 1183
    },
    {
      "epoch": 0.1985119983233784,
      "grad_norm": 0.35876795649528503,
      "learning_rate": 1.855851500571491e-05,
      "loss": 0.9278,
      "step": 1184
    },
    {
      "epoch": 0.1986796604841245,
      "grad_norm": 0.46552401781082153,
      "learning_rate": 1.8555704928538093e-05,
      "loss": 1.0202,
      "step": 1185
    },
    {
      "epoch": 0.1988473226448706,
      "grad_norm": 0.5195321440696716,
      "learning_rate": 1.8552892328177953e-05,
      "loss": 0.962,
      "step": 1186
    },
    {
      "epoch": 0.1990149848056167,
      "grad_norm": 0.4392522871494293,
      "learning_rate": 1.8550077205463953e-05,
      "loss": 0.9341,
      "step": 1187
    },
    {
      "epoch": 0.1991826469663628,
      "grad_norm": 0.44318604469299316,
      "learning_rate": 1.8547259561226313e-05,
      "loss": 1.0245,
      "step": 1188
    },
    {
      "epoch": 0.19935030912710888,
      "grad_norm": 0.4302159249782562,
      "learning_rate": 1.8544439396295987e-05,
      "loss": 0.9287,
      "step": 1189
    },
    {
      "epoch": 0.19951797128785498,
      "grad_norm": 0.4559105634689331,
      "learning_rate": 1.8541616711504676e-05,
      "loss": 1.0109,
      "step": 1190
    },
    {
      "epoch": 0.19968563344860107,
      "grad_norm": 0.49366477131843567,
      "learning_rate": 1.853879150768483e-05,
      "loss": 0.9105,
      "step": 1191
    },
    {
      "epoch": 0.19985329560934717,
      "grad_norm": 0.41833993792533875,
      "learning_rate": 1.8535963785669637e-05,
      "loss": 0.9037,
      "step": 1192
    },
    {
      "epoch": 0.20002095777009327,
      "grad_norm": 0.4915224015712738,
      "learning_rate": 1.8533133546293026e-05,
      "loss": 0.9391,
      "step": 1193
    },
    {
      "epoch": 0.20018861993083936,
      "grad_norm": 0.3822900652885437,
      "learning_rate": 1.8530300790389663e-05,
      "loss": 0.9825,
      "step": 1194
    },
    {
      "epoch": 0.20035628209158546,
      "grad_norm": 0.5990911722183228,
      "learning_rate": 1.8527465518794967e-05,
      "loss": 0.9609,
      "step": 1195
    },
    {
      "epoch": 0.20052394425233155,
      "grad_norm": 0.6043885350227356,
      "learning_rate": 1.85246277323451e-05,
      "loss": 0.9627,
      "step": 1196
    },
    {
      "epoch": 0.20069160641307765,
      "grad_norm": 0.5229475498199463,
      "learning_rate": 1.8521787431876954e-05,
      "loss": 0.9089,
      "step": 1197
    },
    {
      "epoch": 0.20085926857382375,
      "grad_norm": 0.6026935577392578,
      "learning_rate": 1.851894461822817e-05,
      "loss": 0.8736,
      "step": 1198
    },
    {
      "epoch": 0.20102693073456984,
      "grad_norm": 0.5354313850402832,
      "learning_rate": 1.8516099292237132e-05,
      "loss": 0.9127,
      "step": 1199
    },
    {
      "epoch": 0.20119459289531594,
      "grad_norm": 0.5791041254997253,
      "learning_rate": 1.8513251454742957e-05,
      "loss": 0.9913,
      "step": 1200
    },
    {
      "epoch": 0.20136225505606203,
      "grad_norm": 0.5280284285545349,
      "learning_rate": 1.8510401106585508e-05,
      "loss": 0.8859,
      "step": 1201
    },
    {
      "epoch": 0.20152991721680813,
      "grad_norm": 0.43773534893989563,
      "learning_rate": 1.8507548248605393e-05,
      "loss": 0.9555,
      "step": 1202
    },
    {
      "epoch": 0.20169757937755423,
      "grad_norm": 0.49584686756134033,
      "learning_rate": 1.850469288164395e-05,
      "loss": 0.9496,
      "step": 1203
    },
    {
      "epoch": 0.20186524153830032,
      "grad_norm": 0.4935537278652191,
      "learning_rate": 1.8501835006543264e-05,
      "loss": 0.8697,
      "step": 1204
    },
    {
      "epoch": 0.20203290369904642,
      "grad_norm": 0.4067463278770447,
      "learning_rate": 1.8498974624146164e-05,
      "loss": 0.9278,
      "step": 1205
    },
    {
      "epoch": 0.2022005658597925,
      "grad_norm": 0.48257601261138916,
      "learning_rate": 1.84961117352962e-05,
      "loss": 0.8886,
      "step": 1206
    },
    {
      "epoch": 0.2023682280205386,
      "grad_norm": 0.38043829798698425,
      "learning_rate": 1.849324634083768e-05,
      "loss": 0.8953,
      "step": 1207
    },
    {
      "epoch": 0.2025358901812847,
      "grad_norm": 0.5581970810890198,
      "learning_rate": 1.8490378441615645e-05,
      "loss": 0.91,
      "step": 1208
    },
    {
      "epoch": 0.2027035523420308,
      "grad_norm": 0.47461193799972534,
      "learning_rate": 1.8487508038475873e-05,
      "loss": 0.9231,
      "step": 1209
    },
    {
      "epoch": 0.2028712145027769,
      "grad_norm": 0.5092021226882935,
      "learning_rate": 1.848463513226488e-05,
      "loss": 0.9152,
      "step": 1210
    },
    {
      "epoch": 0.203038876663523,
      "grad_norm": 0.48253050446510315,
      "learning_rate": 1.8481759723829922e-05,
      "loss": 0.9013,
      "step": 1211
    },
    {
      "epoch": 0.2032065388242691,
      "grad_norm": 0.49508121609687805,
      "learning_rate": 1.8478881814019e-05,
      "loss": 0.8866,
      "step": 1212
    },
    {
      "epoch": 0.20337420098501519,
      "grad_norm": 0.4377021789550781,
      "learning_rate": 1.8476001403680832e-05,
      "loss": 0.961,
      "step": 1213
    },
    {
      "epoch": 0.20354186314576128,
      "grad_norm": 0.8740084171295166,
      "learning_rate": 1.8473118493664897e-05,
      "loss": 0.9397,
      "step": 1214
    },
    {
      "epoch": 0.20370952530650738,
      "grad_norm": 0.49391335248947144,
      "learning_rate": 1.8470233084821393e-05,
      "loss": 0.8889,
      "step": 1215
    },
    {
      "epoch": 0.20387718746725347,
      "grad_norm": 0.5895149111747742,
      "learning_rate": 1.8467345178001272e-05,
      "loss": 0.8346,
      "step": 1216
    },
    {
      "epoch": 0.20404484962799957,
      "grad_norm": 0.48423007130622864,
      "learning_rate": 1.84644547740562e-05,
      "loss": 0.9015,
      "step": 1217
    },
    {
      "epoch": 0.20421251178874567,
      "grad_norm": 0.34172114729881287,
      "learning_rate": 1.8461561873838612e-05,
      "loss": 0.8922,
      "step": 1218
    },
    {
      "epoch": 0.20438017394949176,
      "grad_norm": 0.40491271018981934,
      "learning_rate": 1.8458666478201643e-05,
      "loss": 0.927,
      "step": 1219
    },
    {
      "epoch": 0.20454783611023786,
      "grad_norm": 0.5788916349411011,
      "learning_rate": 1.845576858799919e-05,
      "loss": 0.9363,
      "step": 1220
    },
    {
      "epoch": 0.20471549827098398,
      "grad_norm": 0.5173137784004211,
      "learning_rate": 1.8452868204085873e-05,
      "loss": 0.8754,
      "step": 1221
    },
    {
      "epoch": 0.20488316043173008,
      "grad_norm": 0.38726314902305603,
      "learning_rate": 1.8449965327317053e-05,
      "loss": 0.8359,
      "step": 1222
    },
    {
      "epoch": 0.20505082259247617,
      "grad_norm": 0.43636006116867065,
      "learning_rate": 1.8447059958548822e-05,
      "loss": 0.9459,
      "step": 1223
    },
    {
      "epoch": 0.20521848475322227,
      "grad_norm": 0.4201851785182953,
      "learning_rate": 1.8444152098638014e-05,
      "loss": 0.92,
      "step": 1224
    },
    {
      "epoch": 0.20538614691396836,
      "grad_norm": 0.541737973690033,
      "learning_rate": 1.8441241748442186e-05,
      "loss": 0.8944,
      "step": 1225
    },
    {
      "epoch": 0.20555380907471446,
      "grad_norm": 0.4758997857570648,
      "learning_rate": 1.8438328908819643e-05,
      "loss": 0.8966,
      "step": 1226
    },
    {
      "epoch": 0.20572147123546056,
      "grad_norm": 0.5539448857307434,
      "learning_rate": 1.8435413580629412e-05,
      "loss": 0.9706,
      "step": 1227
    },
    {
      "epoch": 0.20588913339620665,
      "grad_norm": 0.56991046667099,
      "learning_rate": 1.8432495764731265e-05,
      "loss": 0.976,
      "step": 1228
    },
    {
      "epoch": 0.20605679555695275,
      "grad_norm": 0.4244685173034668,
      "learning_rate": 1.8429575461985693e-05,
      "loss": 0.8839,
      "step": 1229
    },
    {
      "epoch": 0.20622445771769884,
      "grad_norm": 0.6971105337142944,
      "learning_rate": 1.8426652673253943e-05,
      "loss": 0.9441,
      "step": 1230
    },
    {
      "epoch": 0.20639211987844494,
      "grad_norm": 0.5528003573417664,
      "learning_rate": 1.8423727399397968e-05,
      "loss": 0.9105,
      "step": 1231
    },
    {
      "epoch": 0.20655978203919104,
      "grad_norm": 0.3812488913536072,
      "learning_rate": 1.8420799641280478e-05,
      "loss": 0.9143,
      "step": 1232
    },
    {
      "epoch": 0.20672744419993713,
      "grad_norm": 0.423740953207016,
      "learning_rate": 1.8417869399764898e-05,
      "loss": 0.9908,
      "step": 1233
    },
    {
      "epoch": 0.20689510636068323,
      "grad_norm": 0.47226041555404663,
      "learning_rate": 1.8414936675715395e-05,
      "loss": 0.8505,
      "step": 1234
    },
    {
      "epoch": 0.20706276852142932,
      "grad_norm": 0.8200578093528748,
      "learning_rate": 1.8412001469996867e-05,
      "loss": 0.9519,
      "step": 1235
    },
    {
      "epoch": 0.20723043068217542,
      "grad_norm": 0.3981587290763855,
      "learning_rate": 1.840906378347494e-05,
      "loss": 0.8226,
      "step": 1236
    },
    {
      "epoch": 0.20739809284292152,
      "grad_norm": 0.5681799054145813,
      "learning_rate": 1.840612361701598e-05,
      "loss": 0.9721,
      "step": 1237
    },
    {
      "epoch": 0.2075657550036676,
      "grad_norm": 0.617257297039032,
      "learning_rate": 1.8403180971487067e-05,
      "loss": 0.8837,
      "step": 1238
    },
    {
      "epoch": 0.2077334171644137,
      "grad_norm": 0.7258968949317932,
      "learning_rate": 1.8400235847756035e-05,
      "loss": 0.8497,
      "step": 1239
    },
    {
      "epoch": 0.2079010793251598,
      "grad_norm": 0.6802220344543457,
      "learning_rate": 1.839728824669143e-05,
      "loss": 1.0061,
      "step": 1240
    },
    {
      "epoch": 0.2080687414859059,
      "grad_norm": 0.4653562903404236,
      "learning_rate": 1.839433816916254e-05,
      "loss": 0.9083,
      "step": 1241
    },
    {
      "epoch": 0.208236403646652,
      "grad_norm": 0.5622283220291138,
      "learning_rate": 1.839138561603938e-05,
      "loss": 0.891,
      "step": 1242
    },
    {
      "epoch": 0.2084040658073981,
      "grad_norm": 0.4902152121067047,
      "learning_rate": 1.8388430588192693e-05,
      "loss": 0.876,
      "step": 1243
    },
    {
      "epoch": 0.2085717279681442,
      "grad_norm": 0.7380679845809937,
      "learning_rate": 1.838547308649395e-05,
      "loss": 0.9175,
      "step": 1244
    },
    {
      "epoch": 0.20873939012889028,
      "grad_norm": 0.7325884699821472,
      "learning_rate": 1.838251311181536e-05,
      "loss": 0.849,
      "step": 1245
    },
    {
      "epoch": 0.20890705228963638,
      "grad_norm": 0.5089899897575378,
      "learning_rate": 1.8379550665029856e-05,
      "loss": 0.8693,
      "step": 1246
    },
    {
      "epoch": 0.20907471445038248,
      "grad_norm": 0.48446375131607056,
      "learning_rate": 1.8376585747011094e-05,
      "loss": 0.9291,
      "step": 1247
    },
    {
      "epoch": 0.20924237661112857,
      "grad_norm": 0.4304530620574951,
      "learning_rate": 1.837361835863347e-05,
      "loss": 0.9757,
      "step": 1248
    },
    {
      "epoch": 0.20941003877187467,
      "grad_norm": 0.4314531087875366,
      "learning_rate": 1.8370648500772107e-05,
      "loss": 0.9058,
      "step": 1249
    },
    {
      "epoch": 0.20957770093262076,
      "grad_norm": 0.425742506980896,
      "learning_rate": 1.8367676174302848e-05,
      "loss": 0.8532,
      "step": 1250
    },
    {
      "epoch": 0.20974536309336686,
      "grad_norm": 0.5282460451126099,
      "learning_rate": 1.8364701380102267e-05,
      "loss": 0.9287,
      "step": 1251
    },
    {
      "epoch": 0.20991302525411296,
      "grad_norm": 0.43143585324287415,
      "learning_rate": 1.8361724119047675e-05,
      "loss": 0.918,
      "step": 1252
    },
    {
      "epoch": 0.21008068741485905,
      "grad_norm": 0.4454243779182434,
      "learning_rate": 1.8358744392017095e-05,
      "loss": 0.9913,
      "step": 1253
    },
    {
      "epoch": 0.21024834957560515,
      "grad_norm": 0.41314229369163513,
      "learning_rate": 1.835576219988929e-05,
      "loss": 0.9382,
      "step": 1254
    },
    {
      "epoch": 0.21041601173635124,
      "grad_norm": 0.49109718203544617,
      "learning_rate": 1.8352777543543745e-05,
      "loss": 0.937,
      "step": 1255
    },
    {
      "epoch": 0.21058367389709734,
      "grad_norm": 0.4162750244140625,
      "learning_rate": 1.834979042386067e-05,
      "loss": 0.8294,
      "step": 1256
    },
    {
      "epoch": 0.21075133605784344,
      "grad_norm": 0.5138880014419556,
      "learning_rate": 1.8346800841721006e-05,
      "loss": 0.9124,
      "step": 1257
    },
    {
      "epoch": 0.21091899821858953,
      "grad_norm": 0.44537249207496643,
      "learning_rate": 1.8343808798006417e-05,
      "loss": 0.8992,
      "step": 1258
    },
    {
      "epoch": 0.21108666037933563,
      "grad_norm": 0.5039196014404297,
      "learning_rate": 1.8340814293599293e-05,
      "loss": 0.9855,
      "step": 1259
    },
    {
      "epoch": 0.21125432254008172,
      "grad_norm": 0.46184661984443665,
      "learning_rate": 1.8337817329382754e-05,
      "loss": 0.9403,
      "step": 1260
    },
    {
      "epoch": 0.21142198470082782,
      "grad_norm": 0.556057333946228,
      "learning_rate": 1.8334817906240638e-05,
      "loss": 0.8827,
      "step": 1261
    },
    {
      "epoch": 0.21158964686157392,
      "grad_norm": 0.5662396550178528,
      "learning_rate": 1.8331816025057508e-05,
      "loss": 0.9287,
      "step": 1262
    },
    {
      "epoch": 0.21175730902232,
      "grad_norm": 0.4389037489891052,
      "learning_rate": 1.8328811686718667e-05,
      "loss": 0.8198,
      "step": 1263
    },
    {
      "epoch": 0.21192497118306614,
      "grad_norm": 0.4806714951992035,
      "learning_rate": 1.832580489211012e-05,
      "loss": 0.8745,
      "step": 1264
    },
    {
      "epoch": 0.21209263334381223,
      "grad_norm": 0.6693044900894165,
      "learning_rate": 1.8322795642118618e-05,
      "loss": 0.8701,
      "step": 1265
    },
    {
      "epoch": 0.21226029550455833,
      "grad_norm": 0.5517094135284424,
      "learning_rate": 1.8319783937631622e-05,
      "loss": 0.8977,
      "step": 1266
    },
    {
      "epoch": 0.21242795766530442,
      "grad_norm": 0.5779098272323608,
      "learning_rate": 1.8316769779537316e-05,
      "loss": 0.9396,
      "step": 1267
    },
    {
      "epoch": 0.21259561982605052,
      "grad_norm": 0.6421261429786682,
      "learning_rate": 1.831375316872462e-05,
      "loss": 0.9319,
      "step": 1268
    },
    {
      "epoch": 0.21276328198679662,
      "grad_norm": 0.4507690966129303,
      "learning_rate": 1.8310734106083166e-05,
      "loss": 0.9084,
      "step": 1269
    },
    {
      "epoch": 0.2129309441475427,
      "grad_norm": 0.4391637146472931,
      "learning_rate": 1.830771259250331e-05,
      "loss": 0.8856,
      "step": 1270
    },
    {
      "epoch": 0.2130986063082888,
      "grad_norm": 0.7781029939651489,
      "learning_rate": 1.8304688628876147e-05,
      "loss": 0.9225,
      "step": 1271
    },
    {
      "epoch": 0.2132662684690349,
      "grad_norm": 0.5514547824859619,
      "learning_rate": 1.8301662216093464e-05,
      "loss": 0.9248,
      "step": 1272
    },
    {
      "epoch": 0.213433930629781,
      "grad_norm": 0.5290319919586182,
      "learning_rate": 1.82986333550478e-05,
      "loss": 0.8452,
      "step": 1273
    },
    {
      "epoch": 0.2136015927905271,
      "grad_norm": 0.48711273074150085,
      "learning_rate": 1.8295602046632396e-05,
      "loss": 0.9225,
      "step": 1274
    },
    {
      "epoch": 0.2137692549512732,
      "grad_norm": 0.8355472087860107,
      "learning_rate": 1.8292568291741228e-05,
      "loss": 0.9159,
      "step": 1275
    },
    {
      "epoch": 0.2139369171120193,
      "grad_norm": 0.6240776181221008,
      "learning_rate": 1.8289532091268985e-05,
      "loss": 0.8713,
      "step": 1276
    },
    {
      "epoch": 0.21410457927276538,
      "grad_norm": 0.42136117815971375,
      "learning_rate": 1.8286493446111077e-05,
      "loss": 0.8471,
      "step": 1277
    },
    {
      "epoch": 0.21427224143351148,
      "grad_norm": 0.3959744870662689,
      "learning_rate": 1.8283452357163644e-05,
      "loss": 0.8487,
      "step": 1278
    },
    {
      "epoch": 0.21443990359425757,
      "grad_norm": 0.43854543566703796,
      "learning_rate": 1.828040882532354e-05,
      "loss": 0.9884,
      "step": 1279
    },
    {
      "epoch": 0.21460756575500367,
      "grad_norm": 0.5910789966583252,
      "learning_rate": 1.827736285148834e-05,
      "loss": 0.9265,
      "step": 1280
    },
    {
      "epoch": 0.21477522791574977,
      "grad_norm": 0.6404376029968262,
      "learning_rate": 1.8274314436556336e-05,
      "loss": 0.9823,
      "step": 1281
    },
    {
      "epoch": 0.21494289007649586,
      "grad_norm": 0.598853349685669,
      "learning_rate": 1.827126358142654e-05,
      "loss": 0.9459,
      "step": 1282
    },
    {
      "epoch": 0.21511055223724196,
      "grad_norm": 0.4955078363418579,
      "learning_rate": 1.82682102869987e-05,
      "loss": 0.8622,
      "step": 1283
    },
    {
      "epoch": 0.21527821439798805,
      "grad_norm": 0.5622814297676086,
      "learning_rate": 1.8265154554173257e-05,
      "loss": 0.9037,
      "step": 1284
    },
    {
      "epoch": 0.21544587655873415,
      "grad_norm": 0.4194563925266266,
      "learning_rate": 1.8262096383851396e-05,
      "loss": 0.9338,
      "step": 1285
    },
    {
      "epoch": 0.21561353871948025,
      "grad_norm": 0.47616180777549744,
      "learning_rate": 1.8259035776935003e-05,
      "loss": 0.8984,
      "step": 1286
    },
    {
      "epoch": 0.21578120088022634,
      "grad_norm": 0.42187511920928955,
      "learning_rate": 1.8255972734326685e-05,
      "loss": 0.8744,
      "step": 1287
    },
    {
      "epoch": 0.21594886304097244,
      "grad_norm": 0.47959572076797485,
      "learning_rate": 1.8252907256929777e-05,
      "loss": 1.0063,
      "step": 1288
    },
    {
      "epoch": 0.21611652520171853,
      "grad_norm": 0.5419632792472839,
      "learning_rate": 1.8249839345648324e-05,
      "loss": 0.9908,
      "step": 1289
    },
    {
      "epoch": 0.21628418736246463,
      "grad_norm": 0.47485166788101196,
      "learning_rate": 1.8246769001387092e-05,
      "loss": 0.9448,
      "step": 1290
    },
    {
      "epoch": 0.21645184952321073,
      "grad_norm": 0.4475467801094055,
      "learning_rate": 1.8243696225051564e-05,
      "loss": 0.9025,
      "step": 1291
    },
    {
      "epoch": 0.21661951168395682,
      "grad_norm": 0.5631753206253052,
      "learning_rate": 1.8240621017547938e-05,
      "loss": 0.8794,
      "step": 1292
    },
    {
      "epoch": 0.21678717384470292,
      "grad_norm": 0.5549768209457397,
      "learning_rate": 1.8237543379783133e-05,
      "loss": 0.9901,
      "step": 1293
    },
    {
      "epoch": 0.21695483600544901,
      "grad_norm": 0.36228808760643005,
      "learning_rate": 1.8234463312664782e-05,
      "loss": 0.9285,
      "step": 1294
    },
    {
      "epoch": 0.2171224981661951,
      "grad_norm": 0.48294028639793396,
      "learning_rate": 1.8231380817101234e-05,
      "loss": 0.8196,
      "step": 1295
    },
    {
      "epoch": 0.2172901603269412,
      "grad_norm": 0.6337847113609314,
      "learning_rate": 1.8228295894001554e-05,
      "loss": 0.8088,
      "step": 1296
    },
    {
      "epoch": 0.2174578224876873,
      "grad_norm": 0.5867326259613037,
      "learning_rate": 1.8225208544275526e-05,
      "loss": 0.9791,
      "step": 1297
    },
    {
      "epoch": 0.2176254846484334,
      "grad_norm": 0.4840776026248932,
      "learning_rate": 1.822211876883365e-05,
      "loss": 1.0371,
      "step": 1298
    },
    {
      "epoch": 0.2177931468091795,
      "grad_norm": 0.49656978249549866,
      "learning_rate": 1.821902656858713e-05,
      "loss": 0.8567,
      "step": 1299
    },
    {
      "epoch": 0.2179608089699256,
      "grad_norm": 0.5669172406196594,
      "learning_rate": 1.8215931944447904e-05,
      "loss": 0.9545,
      "step": 1300
    },
    {
      "epoch": 0.2181284711306717,
      "grad_norm": 0.4737217128276825,
      "learning_rate": 1.8212834897328614e-05,
      "loss": 0.8801,
      "step": 1301
    },
    {
      "epoch": 0.21829613329141778,
      "grad_norm": 0.49199214577674866,
      "learning_rate": 1.820973542814261e-05,
      "loss": 0.9038,
      "step": 1302
    },
    {
      "epoch": 0.21846379545216388,
      "grad_norm": 0.7255045771598816,
      "learning_rate": 1.820663353780397e-05,
      "loss": 0.8827,
      "step": 1303
    },
    {
      "epoch": 0.21863145761290997,
      "grad_norm": 0.8148097991943359,
      "learning_rate": 1.820352922722748e-05,
      "loss": 0.9634,
      "step": 1304
    },
    {
      "epoch": 0.21879911977365607,
      "grad_norm": 0.6887121200561523,
      "learning_rate": 1.8200422497328635e-05,
      "loss": 0.9764,
      "step": 1305
    },
    {
      "epoch": 0.21896678193440217,
      "grad_norm": 0.6535807251930237,
      "learning_rate": 1.8197313349023655e-05,
      "loss": 0.8518,
      "step": 1306
    },
    {
      "epoch": 0.2191344440951483,
      "grad_norm": 0.4110938012599945,
      "learning_rate": 1.819420178322946e-05,
      "loss": 0.94,
      "step": 1307
    },
    {
      "epoch": 0.21930210625589439,
      "grad_norm": 0.5408376455307007,
      "learning_rate": 1.8191087800863692e-05,
      "loss": 0.9591,
      "step": 1308
    },
    {
      "epoch": 0.21946976841664048,
      "grad_norm": 0.5167989730834961,
      "learning_rate": 1.8187971402844706e-05,
      "loss": 0.8543,
      "step": 1309
    },
    {
      "epoch": 0.21963743057738658,
      "grad_norm": 0.5233778357505798,
      "learning_rate": 1.818485259009156e-05,
      "loss": 0.9628,
      "step": 1310
    },
    {
      "epoch": 0.21980509273813267,
      "grad_norm": 0.5276911854743958,
      "learning_rate": 1.818173136352403e-05,
      "loss": 0.9192,
      "step": 1311
    },
    {
      "epoch": 0.21997275489887877,
      "grad_norm": 0.43043678998947144,
      "learning_rate": 1.8178607724062613e-05,
      "loss": 1.0198,
      "step": 1312
    },
    {
      "epoch": 0.22014041705962487,
      "grad_norm": 0.47249215841293335,
      "learning_rate": 1.8175481672628505e-05,
      "loss": 0.9738,
      "step": 1313
    },
    {
      "epoch": 0.22030807922037096,
      "grad_norm": 0.4545799791812897,
      "learning_rate": 1.8172353210143613e-05,
      "loss": 0.9664,
      "step": 1314
    },
    {
      "epoch": 0.22047574138111706,
      "grad_norm": 0.6046268939971924,
      "learning_rate": 1.8169222337530567e-05,
      "loss": 0.8992,
      "step": 1315
    },
    {
      "epoch": 0.22064340354186315,
      "grad_norm": 0.4744549095630646,
      "learning_rate": 1.816608905571269e-05,
      "loss": 1.0001,
      "step": 1316
    },
    {
      "epoch": 0.22081106570260925,
      "grad_norm": 0.4846677780151367,
      "learning_rate": 1.8162953365614037e-05,
      "loss": 0.9084,
      "step": 1317
    },
    {
      "epoch": 0.22097872786335535,
      "grad_norm": 0.5149449110031128,
      "learning_rate": 1.8159815268159348e-05,
      "loss": 0.8979,
      "step": 1318
    },
    {
      "epoch": 0.22114639002410144,
      "grad_norm": 0.5075791478157043,
      "learning_rate": 1.8156674764274102e-05,
      "loss": 0.8869,
      "step": 1319
    },
    {
      "epoch": 0.22131405218484754,
      "grad_norm": 0.3952072262763977,
      "learning_rate": 1.8153531854884465e-05,
      "loss": 0.8001,
      "step": 1320
    },
    {
      "epoch": 0.22148171434559363,
      "grad_norm": 0.39093104004859924,
      "learning_rate": 1.815038654091732e-05,
      "loss": 0.9095,
      "step": 1321
    },
    {
      "epoch": 0.22164937650633973,
      "grad_norm": 0.40068456530570984,
      "learning_rate": 1.814723882330026e-05,
      "loss": 1.0543,
      "step": 1322
    },
    {
      "epoch": 0.22181703866708583,
      "grad_norm": 0.46475377678871155,
      "learning_rate": 1.814408870296158e-05,
      "loss": 1.0002,
      "step": 1323
    },
    {
      "epoch": 0.22198470082783192,
      "grad_norm": 0.5081574320793152,
      "learning_rate": 1.81409361808303e-05,
      "loss": 0.8861,
      "step": 1324
    },
    {
      "epoch": 0.22215236298857802,
      "grad_norm": 0.4871772229671478,
      "learning_rate": 1.813778125783613e-05,
      "loss": 0.9433,
      "step": 1325
    },
    {
      "epoch": 0.2223200251493241,
      "grad_norm": 0.6324952840805054,
      "learning_rate": 1.81346239349095e-05,
      "loss": 0.9318,
      "step": 1326
    },
    {
      "epoch": 0.2224876873100702,
      "grad_norm": 0.4739683270454407,
      "learning_rate": 1.813146421298154e-05,
      "loss": 0.9027,
      "step": 1327
    },
    {
      "epoch": 0.2226553494708163,
      "grad_norm": 0.44335031509399414,
      "learning_rate": 1.8128302092984095e-05,
      "loss": 0.952,
      "step": 1328
    },
    {
      "epoch": 0.2228230116315624,
      "grad_norm": 0.6963682174682617,
      "learning_rate": 1.812513757584971e-05,
      "loss": 0.9993,
      "step": 1329
    },
    {
      "epoch": 0.2229906737923085,
      "grad_norm": 0.46847036480903625,
      "learning_rate": 1.8121970662511646e-05,
      "loss": 0.9906,
      "step": 1330
    },
    {
      "epoch": 0.2231583359530546,
      "grad_norm": 0.4766705632209778,
      "learning_rate": 1.8118801353903856e-05,
      "loss": 0.9885,
      "step": 1331
    },
    {
      "epoch": 0.2233259981138007,
      "grad_norm": 0.4485360085964203,
      "learning_rate": 1.8115629650961014e-05,
      "loss": 0.7972,
      "step": 1332
    },
    {
      "epoch": 0.22349366027454678,
      "grad_norm": 0.7110408544540405,
      "learning_rate": 1.8112455554618495e-05,
      "loss": 0.9015,
      "step": 1333
    },
    {
      "epoch": 0.22366132243529288,
      "grad_norm": 0.4494708776473999,
      "learning_rate": 1.810927906581238e-05,
      "loss": 0.9505,
      "step": 1334
    },
    {
      "epoch": 0.22382898459603898,
      "grad_norm": 1.1168867349624634,
      "learning_rate": 1.8106100185479446e-05,
      "loss": 1.0235,
      "step": 1335
    },
    {
      "epoch": 0.22399664675678507,
      "grad_norm": 0.4510136842727661,
      "learning_rate": 1.8102918914557196e-05,
      "loss": 0.9523,
      "step": 1336
    },
    {
      "epoch": 0.22416430891753117,
      "grad_norm": 0.5218761563301086,
      "learning_rate": 1.809973525398382e-05,
      "loss": 0.8929,
      "step": 1337
    },
    {
      "epoch": 0.22433197107827726,
      "grad_norm": 0.5287281274795532,
      "learning_rate": 1.8096549204698223e-05,
      "loss": 0.941,
      "step": 1338
    },
    {
      "epoch": 0.22449963323902336,
      "grad_norm": 0.6634778380393982,
      "learning_rate": 1.8093360767640003e-05,
      "loss": 0.9209,
      "step": 1339
    },
    {
      "epoch": 0.22466729539976946,
      "grad_norm": 0.5254518985748291,
      "learning_rate": 1.8090169943749477e-05,
      "loss": 1.0379,
      "step": 1340
    },
    {
      "epoch": 0.22483495756051555,
      "grad_norm": 1.0332894325256348,
      "learning_rate": 1.808697673396765e-05,
      "loss": 0.7821,
      "step": 1341
    },
    {
      "epoch": 0.22500261972126165,
      "grad_norm": 0.6242459416389465,
      "learning_rate": 1.808378113923625e-05,
      "loss": 0.9408,
      "step": 1342
    },
    {
      "epoch": 0.22517028188200774,
      "grad_norm": 0.4452417194843292,
      "learning_rate": 1.8080583160497695e-05,
      "loss": 0.9338,
      "step": 1343
    },
    {
      "epoch": 0.22533794404275384,
      "grad_norm": 0.4866446554660797,
      "learning_rate": 1.80773827986951e-05,
      "loss": 0.9156,
      "step": 1344
    },
    {
      "epoch": 0.22550560620349994,
      "grad_norm": 0.6305269002914429,
      "learning_rate": 1.80741800547723e-05,
      "loss": 0.9429,
      "step": 1345
    },
    {
      "epoch": 0.22567326836424603,
      "grad_norm": 0.4827277362346649,
      "learning_rate": 1.8070974929673818e-05,
      "loss": 0.8953,
      "step": 1346
    },
    {
      "epoch": 0.22584093052499213,
      "grad_norm": 0.7621358036994934,
      "learning_rate": 1.806776742434489e-05,
      "loss": 0.994,
      "step": 1347
    },
    {
      "epoch": 0.22600859268573822,
      "grad_norm": 0.5351371765136719,
      "learning_rate": 1.8064557539731446e-05,
      "loss": 0.8527,
      "step": 1348
    },
    {
      "epoch": 0.22617625484648432,
      "grad_norm": 0.5965827107429504,
      "learning_rate": 1.8061345276780124e-05,
      "loss": 0.942,
      "step": 1349
    },
    {
      "epoch": 0.22634391700723044,
      "grad_norm": 0.5161508321762085,
      "learning_rate": 1.805813063643825e-05,
      "loss": 0.9561,
      "step": 1350
    },
    {
      "epoch": 0.22651157916797654,
      "grad_norm": 0.40499433875083923,
      "learning_rate": 1.805491361965388e-05,
      "loss": 0.8948,
      "step": 1351
    },
    {
      "epoch": 0.22667924132872264,
      "grad_norm": 0.6148918271064758,
      "learning_rate": 1.8051694227375734e-05,
      "loss": 0.8553,
      "step": 1352
    },
    {
      "epoch": 0.22684690348946873,
      "grad_norm": 0.7514647841453552,
      "learning_rate": 1.804847246055326e-05,
      "loss": 0.9548,
      "step": 1353
    },
    {
      "epoch": 0.22701456565021483,
      "grad_norm": 0.6081562638282776,
      "learning_rate": 1.8045248320136593e-05,
      "loss": 0.9345,
      "step": 1354
    },
    {
      "epoch": 0.22718222781096092,
      "grad_norm": 0.6093937158584595,
      "learning_rate": 1.8042021807076575e-05,
      "loss": 0.8476,
      "step": 1355
    },
    {
      "epoch": 0.22734988997170702,
      "grad_norm": 0.7178387641906738,
      "learning_rate": 1.8038792922324743e-05,
      "loss": 0.8914,
      "step": 1356
    },
    {
      "epoch": 0.22751755213245312,
      "grad_norm": 0.6035656332969666,
      "learning_rate": 1.8035561666833336e-05,
      "loss": 0.9777,
      "step": 1357
    },
    {
      "epoch": 0.2276852142931992,
      "grad_norm": 0.7128610014915466,
      "learning_rate": 1.803232804155529e-05,
      "loss": 0.9712,
      "step": 1358
    },
    {
      "epoch": 0.2278528764539453,
      "grad_norm": 0.5069277882575989,
      "learning_rate": 1.8029092047444238e-05,
      "loss": 0.9525,
      "step": 1359
    },
    {
      "epoch": 0.2280205386146914,
      "grad_norm": 0.5084493160247803,
      "learning_rate": 1.8025853685454522e-05,
      "loss": 0.9595,
      "step": 1360
    },
    {
      "epoch": 0.2281882007754375,
      "grad_norm": 0.5395858287811279,
      "learning_rate": 1.8022612956541174e-05,
      "loss": 0.9214,
      "step": 1361
    },
    {
      "epoch": 0.2283558629361836,
      "grad_norm": 0.6916206479072571,
      "learning_rate": 1.801936986165992e-05,
      "loss": 0.9464,
      "step": 1362
    },
    {
      "epoch": 0.2285235250969297,
      "grad_norm": 0.532524585723877,
      "learning_rate": 1.801612440176719e-05,
      "loss": 0.8434,
      "step": 1363
    },
    {
      "epoch": 0.2286911872576758,
      "grad_norm": 0.6037734150886536,
      "learning_rate": 1.8012876577820113e-05,
      "loss": 0.9723,
      "step": 1364
    },
    {
      "epoch": 0.22885884941842188,
      "grad_norm": 0.678534209728241,
      "learning_rate": 1.800962639077651e-05,
      "loss": 0.8821,
      "step": 1365
    },
    {
      "epoch": 0.22902651157916798,
      "grad_norm": 0.48763179779052734,
      "learning_rate": 1.8006373841594905e-05,
      "loss": 0.9745,
      "step": 1366
    },
    {
      "epoch": 0.22919417373991408,
      "grad_norm": 0.4831984341144562,
      "learning_rate": 1.800311893123451e-05,
      "loss": 0.8353,
      "step": 1367
    },
    {
      "epoch": 0.22936183590066017,
      "grad_norm": 0.5547946095466614,
      "learning_rate": 1.7999861660655247e-05,
      "loss": 0.9186,
      "step": 1368
    },
    {
      "epoch": 0.22952949806140627,
      "grad_norm": 0.4802793860435486,
      "learning_rate": 1.7996602030817713e-05,
      "loss": 0.9354,
      "step": 1369
    },
    {
      "epoch": 0.22969716022215236,
      "grad_norm": 0.49225011467933655,
      "learning_rate": 1.7993340042683222e-05,
      "loss": 0.9488,
      "step": 1370
    },
    {
      "epoch": 0.22986482238289846,
      "grad_norm": 0.5383260250091553,
      "learning_rate": 1.7990075697213775e-05,
      "loss": 0.8672,
      "step": 1371
    },
    {
      "epoch": 0.23003248454364456,
      "grad_norm": 0.5191199779510498,
      "learning_rate": 1.798680899537206e-05,
      "loss": 0.974,
      "step": 1372
    },
    {
      "epoch": 0.23020014670439065,
      "grad_norm": 0.45761701464653015,
      "learning_rate": 1.7983539938121476e-05,
      "loss": 0.8985,
      "step": 1373
    },
    {
      "epoch": 0.23036780886513675,
      "grad_norm": 0.4727153778076172,
      "learning_rate": 1.7980268526426105e-05,
      "loss": 0.9525,
      "step": 1374
    },
    {
      "epoch": 0.23053547102588284,
      "grad_norm": 0.4912465512752533,
      "learning_rate": 1.7976994761250727e-05,
      "loss": 1.0002,
      "step": 1375
    },
    {
      "epoch": 0.23070313318662894,
      "grad_norm": 0.41561660170555115,
      "learning_rate": 1.7973718643560817e-05,
      "loss": 0.9725,
      "step": 1376
    },
    {
      "epoch": 0.23087079534737504,
      "grad_norm": 0.4833761155605316,
      "learning_rate": 1.797044017432254e-05,
      "loss": 0.9952,
      "step": 1377
    },
    {
      "epoch": 0.23103845750812113,
      "grad_norm": 0.6341785788536072,
      "learning_rate": 1.7967159354502764e-05,
      "loss": 0.8741,
      "step": 1378
    },
    {
      "epoch": 0.23120611966886723,
      "grad_norm": 0.41993364691734314,
      "learning_rate": 1.7963876185069032e-05,
      "loss": 0.9431,
      "step": 1379
    },
    {
      "epoch": 0.23137378182961332,
      "grad_norm": 0.43000006675720215,
      "learning_rate": 1.7960590666989604e-05,
      "loss": 0.818,
      "step": 1380
    },
    {
      "epoch": 0.23154144399035942,
      "grad_norm": 0.4882097840309143,
      "learning_rate": 1.795730280123341e-05,
      "loss": 0.8718,
      "step": 1381
    },
    {
      "epoch": 0.23170910615110552,
      "grad_norm": 0.7146399617195129,
      "learning_rate": 1.795401258877009e-05,
      "loss": 0.8447,
      "step": 1382
    },
    {
      "epoch": 0.2318767683118516,
      "grad_norm": 0.5772314667701721,
      "learning_rate": 1.795072003056996e-05,
      "loss": 0.7995,
      "step": 1383
    },
    {
      "epoch": 0.2320444304725977,
      "grad_norm": 0.655830442905426,
      "learning_rate": 1.7947425127604045e-05,
      "loss": 0.9063,
      "step": 1384
    },
    {
      "epoch": 0.2322120926333438,
      "grad_norm": 1.0778709650039673,
      "learning_rate": 1.7944127880844043e-05,
      "loss": 0.8961,
      "step": 1385
    },
    {
      "epoch": 0.2323797547940899,
      "grad_norm": 0.36804503202438354,
      "learning_rate": 1.794082829126236e-05,
      "loss": 0.8971,
      "step": 1386
    },
    {
      "epoch": 0.232547416954836,
      "grad_norm": 0.4572069048881531,
      "learning_rate": 1.7937526359832088e-05,
      "loss": 0.8715,
      "step": 1387
    },
    {
      "epoch": 0.2327150791155821,
      "grad_norm": 1.2500030994415283,
      "learning_rate": 1.7934222087526996e-05,
      "loss": 0.8778,
      "step": 1388
    },
    {
      "epoch": 0.2328827412763282,
      "grad_norm": 0.4822962284088135,
      "learning_rate": 1.793091547532157e-05,
      "loss": 0.8366,
      "step": 1389
    },
    {
      "epoch": 0.23305040343707428,
      "grad_norm": 0.36731037497520447,
      "learning_rate": 1.792760652419096e-05,
      "loss": 0.8884,
      "step": 1390
    },
    {
      "epoch": 0.23321806559782038,
      "grad_norm": 0.576995313167572,
      "learning_rate": 1.792429523511102e-05,
      "loss": 0.9172,
      "step": 1391
    },
    {
      "epoch": 0.23338572775856647,
      "grad_norm": 0.406104177236557,
      "learning_rate": 1.792098160905829e-05,
      "loss": 0.9602,
      "step": 1392
    },
    {
      "epoch": 0.2335533899193126,
      "grad_norm": 0.6425048112869263,
      "learning_rate": 1.7917665647010003e-05,
      "loss": 0.9503,
      "step": 1393
    },
    {
      "epoch": 0.2337210520800587,
      "grad_norm": 0.3961374759674072,
      "learning_rate": 1.791434734994407e-05,
      "loss": 0.9535,
      "step": 1394
    },
    {
      "epoch": 0.2338887142408048,
      "grad_norm": 0.47615012526512146,
      "learning_rate": 1.79110267188391e-05,
      "loss": 0.8725,
      "step": 1395
    },
    {
      "epoch": 0.2340563764015509,
      "grad_norm": 0.5663509964942932,
      "learning_rate": 1.7907703754674392e-05,
      "loss": 0.8807,
      "step": 1396
    },
    {
      "epoch": 0.23422403856229698,
      "grad_norm": 0.4356711804866791,
      "learning_rate": 1.790437845842993e-05,
      "loss": 0.9362,
      "step": 1397
    },
    {
      "epoch": 0.23439170072304308,
      "grad_norm": 0.5933541655540466,
      "learning_rate": 1.7901050831086377e-05,
      "loss": 0.9572,
      "step": 1398
    },
    {
      "epoch": 0.23455936288378917,
      "grad_norm": 1.4825639724731445,
      "learning_rate": 1.78977208736251e-05,
      "loss": 0.862,
      "step": 1399
    },
    {
      "epoch": 0.23472702504453527,
      "grad_norm": 0.38333049416542053,
      "learning_rate": 1.789438858702814e-05,
      "loss": 0.9109,
      "step": 1400
    },
    {
      "epoch": 0.23489468720528137,
      "grad_norm": 0.40966543555259705,
      "learning_rate": 1.789105397227823e-05,
      "loss": 0.8819,
      "step": 1401
    },
    {
      "epoch": 0.23506234936602746,
      "grad_norm": 0.5122490525245667,
      "learning_rate": 1.7887717030358794e-05,
      "loss": 0.9017,
      "step": 1402
    },
    {
      "epoch": 0.23523001152677356,
      "grad_norm": 0.6329760551452637,
      "learning_rate": 1.788437776225393e-05,
      "loss": 0.8879,
      "step": 1403
    },
    {
      "epoch": 0.23539767368751965,
      "grad_norm": 0.3476681113243103,
      "learning_rate": 1.7881036168948435e-05,
      "loss": 0.84,
      "step": 1404
    },
    {
      "epoch": 0.23556533584826575,
      "grad_norm": 0.44182735681533813,
      "learning_rate": 1.7877692251427783e-05,
      "loss": 0.7926,
      "step": 1405
    },
    {
      "epoch": 0.23573299800901185,
      "grad_norm": 0.4372468590736389,
      "learning_rate": 1.787434601067814e-05,
      "loss": 0.8847,
      "step": 1406
    },
    {
      "epoch": 0.23590066016975794,
      "grad_norm": 0.4955695867538452,
      "learning_rate": 1.787099744768635e-05,
      "loss": 0.9369,
      "step": 1407
    },
    {
      "epoch": 0.23606832233050404,
      "grad_norm": 0.7181636095046997,
      "learning_rate": 1.7867646563439947e-05,
      "loss": 0.9023,
      "step": 1408
    },
    {
      "epoch": 0.23623598449125013,
      "grad_norm": 0.5251885652542114,
      "learning_rate": 1.786429335892715e-05,
      "loss": 0.8992,
      "step": 1409
    },
    {
      "epoch": 0.23640364665199623,
      "grad_norm": 0.5547168254852295,
      "learning_rate": 1.786093783513686e-05,
      "loss": 0.9733,
      "step": 1410
    },
    {
      "epoch": 0.23657130881274233,
      "grad_norm": 0.6549569964408875,
      "learning_rate": 1.7857579993058662e-05,
      "loss": 0.8548,
      "step": 1411
    },
    {
      "epoch": 0.23673897097348842,
      "grad_norm": 0.6456899046897888,
      "learning_rate": 1.7854219833682825e-05,
      "loss": 0.9847,
      "step": 1412
    },
    {
      "epoch": 0.23690663313423452,
      "grad_norm": 0.3912792503833771,
      "learning_rate": 1.78508573580003e-05,
      "loss": 0.8897,
      "step": 1413
    },
    {
      "epoch": 0.2370742952949806,
      "grad_norm": 0.43298226594924927,
      "learning_rate": 1.7847492567002723e-05,
      "loss": 0.9704,
      "step": 1414
    },
    {
      "epoch": 0.2372419574557267,
      "grad_norm": 0.5298269987106323,
      "learning_rate": 1.784412546168242e-05,
      "loss": 0.9295,
      "step": 1415
    },
    {
      "epoch": 0.2374096196164728,
      "grad_norm": 0.4604285657405853,
      "learning_rate": 1.784075604303238e-05,
      "loss": 0.9533,
      "step": 1416
    },
    {
      "epoch": 0.2375772817772189,
      "grad_norm": 0.6257635354995728,
      "learning_rate": 1.7837384312046297e-05,
      "loss": 0.9461,
      "step": 1417
    },
    {
      "epoch": 0.237744943937965,
      "grad_norm": 0.4810788631439209,
      "learning_rate": 1.7834010269718526e-05,
      "loss": 0.9202,
      "step": 1418
    },
    {
      "epoch": 0.2379126060987111,
      "grad_norm": 0.6605314016342163,
      "learning_rate": 1.7830633917044125e-05,
      "loss": 0.9729,
      "step": 1419
    },
    {
      "epoch": 0.2380802682594572,
      "grad_norm": 0.3528065085411072,
      "learning_rate": 1.7827255255018813e-05,
      "loss": 0.8712,
      "step": 1420
    },
    {
      "epoch": 0.23824793042020329,
      "grad_norm": 0.4199623167514801,
      "learning_rate": 1.7823874284639004e-05,
      "loss": 0.9097,
      "step": 1421
    },
    {
      "epoch": 0.23841559258094938,
      "grad_norm": 0.5196505188941956,
      "learning_rate": 1.782049100690179e-05,
      "loss": 0.9383,
      "step": 1422
    },
    {
      "epoch": 0.23858325474169548,
      "grad_norm": 0.5818756818771362,
      "learning_rate": 1.781710542280493e-05,
      "loss": 0.8194,
      "step": 1423
    },
    {
      "epoch": 0.23875091690244157,
      "grad_norm": 0.43274083733558655,
      "learning_rate": 1.781371753334689e-05,
      "loss": 0.9057,
      "step": 1424
    },
    {
      "epoch": 0.23891857906318767,
      "grad_norm": 0.35609766840934753,
      "learning_rate": 1.7810327339526792e-05,
      "loss": 0.8873,
      "step": 1425
    },
    {
      "epoch": 0.23908624122393377,
      "grad_norm": 0.48200127482414246,
      "learning_rate": 1.7806934842344444e-05,
      "loss": 0.896,
      "step": 1426
    },
    {
      "epoch": 0.23925390338467986,
      "grad_norm": 0.6076840758323669,
      "learning_rate": 1.7803540042800347e-05,
      "loss": 0.838,
      "step": 1427
    },
    {
      "epoch": 0.23942156554542596,
      "grad_norm": 0.6591511964797974,
      "learning_rate": 1.7800142941895658e-05,
      "loss": 0.945,
      "step": 1428
    },
    {
      "epoch": 0.23958922770617205,
      "grad_norm": 0.4139047861099243,
      "learning_rate": 1.7796743540632226e-05,
      "loss": 0.8899,
      "step": 1429
    },
    {
      "epoch": 0.23975688986691815,
      "grad_norm": 0.5683903098106384,
      "learning_rate": 1.779334184001258e-05,
      "loss": 0.892,
      "step": 1430
    },
    {
      "epoch": 0.23992455202766425,
      "grad_norm": 0.4455464482307434,
      "learning_rate": 1.778993784103992e-05,
      "loss": 0.9469,
      "step": 1431
    },
    {
      "epoch": 0.24009221418841034,
      "grad_norm": 0.42767488956451416,
      "learning_rate": 1.778653154471813e-05,
      "loss": 0.887,
      "step": 1432
    },
    {
      "epoch": 0.24025987634915644,
      "grad_norm": 0.6215432286262512,
      "learning_rate": 1.778312295205177e-05,
      "loss": 0.8905,
      "step": 1433
    },
    {
      "epoch": 0.24042753850990253,
      "grad_norm": 0.4767569899559021,
      "learning_rate": 1.7779712064046073e-05,
      "loss": 1.0294,
      "step": 1434
    },
    {
      "epoch": 0.24059520067064866,
      "grad_norm": 0.5109009146690369,
      "learning_rate": 1.7776298881706955e-05,
      "loss": 0.9222,
      "step": 1435
    },
    {
      "epoch": 0.24076286283139475,
      "grad_norm": 0.8203328251838684,
      "learning_rate": 1.7772883406041002e-05,
      "loss": 0.8387,
      "step": 1436
    },
    {
      "epoch": 0.24093052499214085,
      "grad_norm": 0.5343973636627197,
      "learning_rate": 1.7769465638055483e-05,
      "loss": 0.926,
      "step": 1437
    },
    {
      "epoch": 0.24109818715288694,
      "grad_norm": 1.1681573390960693,
      "learning_rate": 1.776604557875834e-05,
      "loss": 0.9941,
      "step": 1438
    },
    {
      "epoch": 0.24126584931363304,
      "grad_norm": 0.5948541164398193,
      "learning_rate": 1.7762623229158192e-05,
      "loss": 0.9241,
      "step": 1439
    },
    {
      "epoch": 0.24143351147437914,
      "grad_norm": 0.5291740894317627,
      "learning_rate": 1.7759198590264327e-05,
      "loss": 0.8958,
      "step": 1440
    },
    {
      "epoch": 0.24160117363512523,
      "grad_norm": 0.5829247832298279,
      "learning_rate": 1.775577166308672e-05,
      "loss": 0.8648,
      "step": 1441
    },
    {
      "epoch": 0.24176883579587133,
      "grad_norm": 0.43644002079963684,
      "learning_rate": 1.775234244863601e-05,
      "loss": 1.0055,
      "step": 1442
    },
    {
      "epoch": 0.24193649795661742,
      "grad_norm": 0.7420687675476074,
      "learning_rate": 1.7748910947923515e-05,
      "loss": 0.8401,
      "step": 1443
    },
    {
      "epoch": 0.24210416011736352,
      "grad_norm": 0.5568789839744568,
      "learning_rate": 1.774547716196123e-05,
      "loss": 0.9209,
      "step": 1444
    },
    {
      "epoch": 0.24227182227810962,
      "grad_norm": 0.47366562485694885,
      "learning_rate": 1.7742041091761823e-05,
      "loss": 1.0516,
      "step": 1445
    },
    {
      "epoch": 0.2424394844388557,
      "grad_norm": 0.37101438641548157,
      "learning_rate": 1.7738602738338625e-05,
      "loss": 0.9446,
      "step": 1446
    },
    {
      "epoch": 0.2426071465996018,
      "grad_norm": 0.34895703196525574,
      "learning_rate": 1.773516210270565e-05,
      "loss": 0.855,
      "step": 1447
    },
    {
      "epoch": 0.2427748087603479,
      "grad_norm": 0.4138221740722656,
      "learning_rate": 1.7731719185877593e-05,
      "loss": 0.8637,
      "step": 1448
    },
    {
      "epoch": 0.242942470921094,
      "grad_norm": 0.44439229369163513,
      "learning_rate": 1.7728273988869807e-05,
      "loss": 0.9166,
      "step": 1449
    },
    {
      "epoch": 0.2431101330818401,
      "grad_norm": 0.6057480573654175,
      "learning_rate": 1.7724826512698324e-05,
      "loss": 0.9553,
      "step": 1450
    },
    {
      "epoch": 0.2432777952425862,
      "grad_norm": 0.5055689811706543,
      "learning_rate": 1.7721376758379847e-05,
      "loss": 0.8872,
      "step": 1451
    },
    {
      "epoch": 0.2434454574033323,
      "grad_norm": 0.4600638151168823,
      "learning_rate": 1.7717924726931754e-05,
      "loss": 0.8982,
      "step": 1452
    },
    {
      "epoch": 0.24361311956407838,
      "grad_norm": 0.4982481002807617,
      "learning_rate": 1.7714470419372086e-05,
      "loss": 0.9145,
      "step": 1453
    },
    {
      "epoch": 0.24378078172482448,
      "grad_norm": 0.47119757533073425,
      "learning_rate": 1.7711013836719565e-05,
      "loss": 0.9157,
      "step": 1454
    },
    {
      "epoch": 0.24394844388557058,
      "grad_norm": 0.5009276270866394,
      "learning_rate": 1.770755497999358e-05,
      "loss": 0.9679,
      "step": 1455
    },
    {
      "epoch": 0.24411610604631667,
      "grad_norm": 0.7232698202133179,
      "learning_rate": 1.770409385021419e-05,
      "loss": 0.8615,
      "step": 1456
    },
    {
      "epoch": 0.24428376820706277,
      "grad_norm": 0.4853869676589966,
      "learning_rate": 1.7700630448402125e-05,
      "loss": 1.0043,
      "step": 1457
    },
    {
      "epoch": 0.24445143036780886,
      "grad_norm": 0.5485565066337585,
      "learning_rate": 1.7697164775578785e-05,
      "loss": 0.8516,
      "step": 1458
    },
    {
      "epoch": 0.24461909252855496,
      "grad_norm": 0.6456785798072815,
      "learning_rate": 1.769369683276624e-05,
      "loss": 0.9454,
      "step": 1459
    },
    {
      "epoch": 0.24478675468930106,
      "grad_norm": 0.5416443347930908,
      "learning_rate": 1.7690226620987233e-05,
      "loss": 0.9289,
      "step": 1460
    },
    {
      "epoch": 0.24495441685004715,
      "grad_norm": 0.5395210981369019,
      "learning_rate": 1.7686754141265162e-05,
      "loss": 0.9846,
      "step": 1461
    },
    {
      "epoch": 0.24512207901079325,
      "grad_norm": 0.4691596031188965,
      "learning_rate": 1.768327939462412e-05,
      "loss": 0.8755,
      "step": 1462
    },
    {
      "epoch": 0.24528974117153934,
      "grad_norm": 0.4843076169490814,
      "learning_rate": 1.7679802382088836e-05,
      "loss": 0.9427,
      "step": 1463
    },
    {
      "epoch": 0.24545740333228544,
      "grad_norm": 0.4803543984889984,
      "learning_rate": 1.7676323104684736e-05,
      "loss": 1.0253,
      "step": 1464
    },
    {
      "epoch": 0.24562506549303154,
      "grad_norm": 0.5563539862632751,
      "learning_rate": 1.7672841563437897e-05,
      "loss": 0.8087,
      "step": 1465
    },
    {
      "epoch": 0.24579272765377763,
      "grad_norm": 0.46070781350135803,
      "learning_rate": 1.766935775937507e-05,
      "loss": 0.8593,
      "step": 1466
    },
    {
      "epoch": 0.24596038981452373,
      "grad_norm": 0.8026439547538757,
      "learning_rate": 1.766587169352367e-05,
      "loss": 0.831,
      "step": 1467
    },
    {
      "epoch": 0.24612805197526982,
      "grad_norm": 0.6293075680732727,
      "learning_rate": 1.7662383366911785e-05,
      "loss": 0.8819,
      "step": 1468
    },
    {
      "epoch": 0.24629571413601592,
      "grad_norm": 0.630046546459198,
      "learning_rate": 1.765889278056816e-05,
      "loss": 0.8985,
      "step": 1469
    },
    {
      "epoch": 0.24646337629676202,
      "grad_norm": 0.5543246865272522,
      "learning_rate": 1.7655399935522216e-05,
      "loss": 0.8879,
      "step": 1470
    },
    {
      "epoch": 0.2466310384575081,
      "grad_norm": 0.440186083316803,
      "learning_rate": 1.7651904832804038e-05,
      "loss": 0.88,
      "step": 1471
    },
    {
      "epoch": 0.2467987006182542,
      "grad_norm": 0.6598295569419861,
      "learning_rate": 1.764840747344437e-05,
      "loss": 0.9657,
      "step": 1472
    },
    {
      "epoch": 0.2469663627790003,
      "grad_norm": 0.8729094862937927,
      "learning_rate": 1.764490785847463e-05,
      "loss": 0.8873,
      "step": 1473
    },
    {
      "epoch": 0.2471340249397464,
      "grad_norm": 0.523439884185791,
      "learning_rate": 1.76414059889269e-05,
      "loss": 0.9577,
      "step": 1474
    },
    {
      "epoch": 0.2473016871004925,
      "grad_norm": 0.41881799697875977,
      "learning_rate": 1.7637901865833923e-05,
      "loss": 0.9237,
      "step": 1475
    },
    {
      "epoch": 0.2474693492612386,
      "grad_norm": 0.5430641770362854,
      "learning_rate": 1.76343954902291e-05,
      "loss": 0.8916,
      "step": 1476
    },
    {
      "epoch": 0.2476370114219847,
      "grad_norm": 0.4627753794193268,
      "learning_rate": 1.763088686314652e-05,
      "loss": 1.0475,
      "step": 1477
    },
    {
      "epoch": 0.2478046735827308,
      "grad_norm": 0.40650758147239685,
      "learning_rate": 1.7627375985620902e-05,
      "loss": 0.8889,
      "step": 1478
    },
    {
      "epoch": 0.2479723357434769,
      "grad_norm": 0.31283724308013916,
      "learning_rate": 1.7623862858687665e-05,
      "loss": 0.8052,
      "step": 1479
    },
    {
      "epoch": 0.248139997904223,
      "grad_norm": 0.4834359288215637,
      "learning_rate": 1.762034748338286e-05,
      "loss": 0.9563,
      "step": 1480
    },
    {
      "epoch": 0.2483076600649691,
      "grad_norm": 0.4991844892501831,
      "learning_rate": 1.761682986074323e-05,
      "loss": 0.9546,
      "step": 1481
    },
    {
      "epoch": 0.2484753222257152,
      "grad_norm": 0.5470111966133118,
      "learning_rate": 1.761330999180615e-05,
      "loss": 0.9472,
      "step": 1482
    },
    {
      "epoch": 0.2486429843864613,
      "grad_norm": 0.4790292978286743,
      "learning_rate": 1.7609787877609678e-05,
      "loss": 0.919,
      "step": 1483
    },
    {
      "epoch": 0.2488106465472074,
      "grad_norm": 0.49103057384490967,
      "learning_rate": 1.7606263519192535e-05,
      "loss": 0.8835,
      "step": 1484
    },
    {
      "epoch": 0.24897830870795348,
      "grad_norm": 0.5295326709747314,
      "learning_rate": 1.760273691759409e-05,
      "loss": 0.9874,
      "step": 1485
    },
    {
      "epoch": 0.24914597086869958,
      "grad_norm": 0.3911377489566803,
      "learning_rate": 1.7599208073854383e-05,
      "loss": 0.8222,
      "step": 1486
    },
    {
      "epoch": 0.24931363302944567,
      "grad_norm": 0.5055460333824158,
      "learning_rate": 1.759567698901412e-05,
      "loss": 0.9445,
      "step": 1487
    },
    {
      "epoch": 0.24948129519019177,
      "grad_norm": 0.6002017855644226,
      "learning_rate": 1.7592143664114656e-05,
      "loss": 0.9085,
      "step": 1488
    },
    {
      "epoch": 0.24964895735093787,
      "grad_norm": 0.4354606568813324,
      "learning_rate": 1.758860810019801e-05,
      "loss": 0.8811,
      "step": 1489
    },
    {
      "epoch": 0.24981661951168396,
      "grad_norm": 0.40272077918052673,
      "learning_rate": 1.758507029830687e-05,
      "loss": 0.8248,
      "step": 1490
    },
    {
      "epoch": 0.24998428167243006,
      "grad_norm": 0.4683881103992462,
      "learning_rate": 1.7581530259484572e-05,
      "loss": 0.9636,
      "step": 1491
    },
    {
      "epoch": 0.2501519438331761,
      "grad_norm": 0.501693606376648,
      "learning_rate": 1.7577987984775125e-05,
      "loss": 0.9065,
      "step": 1492
    },
    {
      "epoch": 0.2503196059939222,
      "grad_norm": 0.5398340821266174,
      "learning_rate": 1.7574443475223183e-05,
      "loss": 0.904,
      "step": 1493
    },
    {
      "epoch": 0.2504872681546683,
      "grad_norm": 0.404601126909256,
      "learning_rate": 1.757089673187406e-05,
      "loss": 0.9034,
      "step": 1494
    },
    {
      "epoch": 0.2506549303154144,
      "grad_norm": 0.5416507124900818,
      "learning_rate": 1.7567347755773752e-05,
      "loss": 0.9015,
      "step": 1495
    },
    {
      "epoch": 0.2508225924761605,
      "grad_norm": 0.611964225769043,
      "learning_rate": 1.756379654796888e-05,
      "loss": 0.8833,
      "step": 1496
    },
    {
      "epoch": 0.2509902546369066,
      "grad_norm": 0.4463139474391937,
      "learning_rate": 1.7560243109506746e-05,
      "loss": 0.9851,
      "step": 1497
    },
    {
      "epoch": 0.2511579167976527,
      "grad_norm": 0.547607958316803,
      "learning_rate": 1.7556687441435302e-05,
      "loss": 0.9473,
      "step": 1498
    },
    {
      "epoch": 0.2513255789583988,
      "grad_norm": 0.5264208912849426,
      "learning_rate": 1.7553129544803158e-05,
      "loss": 0.8994,
      "step": 1499
    },
    {
      "epoch": 0.25149324111914495,
      "grad_norm": 0.4187546372413635,
      "learning_rate": 1.754956942065958e-05,
      "loss": 0.9347,
      "step": 1500
    },
    {
      "epoch": 0.25166090327989105,
      "grad_norm": 0.4960138201713562,
      "learning_rate": 1.7546007070054497e-05,
      "loss": 0.8381,
      "step": 1501
    },
    {
      "epoch": 0.25182856544063714,
      "grad_norm": 0.7610869407653809,
      "learning_rate": 1.754244249403849e-05,
      "loss": 0.9455,
      "step": 1502
    },
    {
      "epoch": 0.25199622760138324,
      "grad_norm": 0.4258002042770386,
      "learning_rate": 1.7538875693662793e-05,
      "loss": 0.9081,
      "step": 1503
    },
    {
      "epoch": 0.25216388976212933,
      "grad_norm": 0.6918498873710632,
      "learning_rate": 1.7535306669979305e-05,
      "loss": 0.8242,
      "step": 1504
    },
    {
      "epoch": 0.25233155192287543,
      "grad_norm": 0.6293070912361145,
      "learning_rate": 1.753173542404057e-05,
      "loss": 0.9185,
      "step": 1505
    },
    {
      "epoch": 0.2524992140836215,
      "grad_norm": 0.5741620063781738,
      "learning_rate": 1.752816195689979e-05,
      "loss": 0.9219,
      "step": 1506
    },
    {
      "epoch": 0.2526668762443676,
      "grad_norm": 0.49497202038764954,
      "learning_rate": 1.752458626961083e-05,
      "loss": 0.9218,
      "step": 1507
    },
    {
      "epoch": 0.2528345384051137,
      "grad_norm": 0.7709310054779053,
      "learning_rate": 1.7521008363228208e-05,
      "loss": 0.9209,
      "step": 1508
    },
    {
      "epoch": 0.2530022005658598,
      "grad_norm": 0.7194563746452332,
      "learning_rate": 1.7517428238807085e-05,
      "loss": 0.8209,
      "step": 1509
    },
    {
      "epoch": 0.2531698627266059,
      "grad_norm": 0.6282023191452026,
      "learning_rate": 1.751384589740329e-05,
      "loss": 0.9013,
      "step": 1510
    },
    {
      "epoch": 0.253337524887352,
      "grad_norm": 0.4894861876964569,
      "learning_rate": 1.7510261340073297e-05,
      "loss": 0.8998,
      "step": 1511
    },
    {
      "epoch": 0.2535051870480981,
      "grad_norm": 0.9297739863395691,
      "learning_rate": 1.7506674567874234e-05,
      "loss": 0.9283,
      "step": 1512
    },
    {
      "epoch": 0.2536728492088442,
      "grad_norm": 0.586509644985199,
      "learning_rate": 1.7503085581863887e-05,
      "loss": 0.9808,
      "step": 1513
    },
    {
      "epoch": 0.2538405113695903,
      "grad_norm": 0.3984715938568115,
      "learning_rate": 1.7499494383100697e-05,
      "loss": 0.8636,
      "step": 1514
    },
    {
      "epoch": 0.2540081735303364,
      "grad_norm": 0.522193431854248,
      "learning_rate": 1.7495900972643745e-05,
      "loss": 0.9593,
      "step": 1515
    },
    {
      "epoch": 0.2541758356910825,
      "grad_norm": 0.5000961422920227,
      "learning_rate": 1.7492305351552777e-05,
      "loss": 1.0069,
      "step": 1516
    },
    {
      "epoch": 0.2543434978518286,
      "grad_norm": 0.42993372678756714,
      "learning_rate": 1.7488707520888183e-05,
      "loss": 0.9476,
      "step": 1517
    },
    {
      "epoch": 0.2545111600125747,
      "grad_norm": 0.5267705917358398,
      "learning_rate": 1.7485107481711014e-05,
      "loss": 0.8771,
      "step": 1518
    },
    {
      "epoch": 0.2546788221733208,
      "grad_norm": 0.4390380084514618,
      "learning_rate": 1.7481505235082958e-05,
      "loss": 0.8721,
      "step": 1519
    },
    {
      "epoch": 0.25484648433406687,
      "grad_norm": 0.5014662742614746,
      "learning_rate": 1.7477900782066365e-05,
      "loss": 0.9384,
      "step": 1520
    },
    {
      "epoch": 0.25501414649481297,
      "grad_norm": 0.42803215980529785,
      "learning_rate": 1.747429412372424e-05,
      "loss": 0.8852,
      "step": 1521
    },
    {
      "epoch": 0.25518180865555906,
      "grad_norm": 0.7259057760238647,
      "learning_rate": 1.747068526112022e-05,
      "loss": 0.9101,
      "step": 1522
    },
    {
      "epoch": 0.25534947081630516,
      "grad_norm": 0.8499658107757568,
      "learning_rate": 1.746707419531861e-05,
      "loss": 0.87,
      "step": 1523
    },
    {
      "epoch": 0.25551713297705125,
      "grad_norm": 0.6882765889167786,
      "learning_rate": 1.7463460927384356e-05,
      "loss": 0.9239,
      "step": 1524
    },
    {
      "epoch": 0.25568479513779735,
      "grad_norm": 0.6137887239456177,
      "learning_rate": 1.745984545838306e-05,
      "loss": 1.059,
      "step": 1525
    },
    {
      "epoch": 0.25585245729854345,
      "grad_norm": 0.5296462774276733,
      "learning_rate": 1.745622778938096e-05,
      "loss": 0.9238,
      "step": 1526
    },
    {
      "epoch": 0.25602011945928954,
      "grad_norm": 0.42972806096076965,
      "learning_rate": 1.7452607921444956e-05,
      "loss": 0.8551,
      "step": 1527
    },
    {
      "epoch": 0.25618778162003564,
      "grad_norm": 0.4224143922328949,
      "learning_rate": 1.744898585564259e-05,
      "loss": 0.8837,
      "step": 1528
    },
    {
      "epoch": 0.25635544378078173,
      "grad_norm": 0.4010677635669708,
      "learning_rate": 1.744536159304206e-05,
      "loss": 0.8855,
      "step": 1529
    },
    {
      "epoch": 0.25652310594152783,
      "grad_norm": 0.6350672841072083,
      "learning_rate": 1.74417351347122e-05,
      "loss": 0.8995,
      "step": 1530
    },
    {
      "epoch": 0.2566907681022739,
      "grad_norm": 0.5510476231575012,
      "learning_rate": 1.7438106481722495e-05,
      "loss": 0.9429,
      "step": 1531
    },
    {
      "epoch": 0.25685843026302,
      "grad_norm": 0.5789257884025574,
      "learning_rate": 1.743447563514309e-05,
      "loss": 0.8655,
      "step": 1532
    },
    {
      "epoch": 0.2570260924237661,
      "grad_norm": 0.5662457346916199,
      "learning_rate": 1.743084259604476e-05,
      "loss": 0.8362,
      "step": 1533
    },
    {
      "epoch": 0.2571937545845122,
      "grad_norm": 0.44558578729629517,
      "learning_rate": 1.742720736549893e-05,
      "loss": 0.8342,
      "step": 1534
    },
    {
      "epoch": 0.2573614167452583,
      "grad_norm": 0.4779844880104065,
      "learning_rate": 1.7423569944577677e-05,
      "loss": 0.8702,
      "step": 1535
    },
    {
      "epoch": 0.2575290789060044,
      "grad_norm": 0.41661155223846436,
      "learning_rate": 1.7419930334353728e-05,
      "loss": 0.947,
      "step": 1536
    },
    {
      "epoch": 0.2576967410667505,
      "grad_norm": 0.5755490064620972,
      "learning_rate": 1.741628853590044e-05,
      "loss": 0.9165,
      "step": 1537
    },
    {
      "epoch": 0.2578644032274966,
      "grad_norm": 0.4679485261440277,
      "learning_rate": 1.741264455029183e-05,
      "loss": 0.847,
      "step": 1538
    },
    {
      "epoch": 0.2580320653882427,
      "grad_norm": 0.5541486144065857,
      "learning_rate": 1.7408998378602555e-05,
      "loss": 0.9664,
      "step": 1539
    },
    {
      "epoch": 0.2581997275489888,
      "grad_norm": 0.45861151814460754,
      "learning_rate": 1.7405350021907912e-05,
      "loss": 0.8682,
      "step": 1540
    },
    {
      "epoch": 0.2583673897097349,
      "grad_norm": 0.3169395923614502,
      "learning_rate": 1.7401699481283847e-05,
      "loss": 0.8557,
      "step": 1541
    },
    {
      "epoch": 0.258535051870481,
      "grad_norm": 0.5471020340919495,
      "learning_rate": 1.739804675780695e-05,
      "loss": 0.9523,
      "step": 1542
    },
    {
      "epoch": 0.2587027140312271,
      "grad_norm": 0.6324790120124817,
      "learning_rate": 1.7394391852554456e-05,
      "loss": 0.9002,
      "step": 1543
    },
    {
      "epoch": 0.2588703761919732,
      "grad_norm": 0.4142170250415802,
      "learning_rate": 1.7390734766604244e-05,
      "loss": 0.938,
      "step": 1544
    },
    {
      "epoch": 0.25903803835271927,
      "grad_norm": 0.633558988571167,
      "learning_rate": 1.738707550103483e-05,
      "loss": 0.8781,
      "step": 1545
    },
    {
      "epoch": 0.25920570051346536,
      "grad_norm": 0.7527374029159546,
      "learning_rate": 1.738341405692538e-05,
      "loss": 0.8925,
      "step": 1546
    },
    {
      "epoch": 0.25937336267421146,
      "grad_norm": 0.6943590044975281,
      "learning_rate": 1.7379750435355696e-05,
      "loss": 0.8406,
      "step": 1547
    },
    {
      "epoch": 0.25954102483495756,
      "grad_norm": 0.5189474821090698,
      "learning_rate": 1.7376084637406222e-05,
      "loss": 0.9907,
      "step": 1548
    },
    {
      "epoch": 0.25970868699570365,
      "grad_norm": 0.4832892417907715,
      "learning_rate": 1.737241666415806e-05,
      "loss": 0.9535,
      "step": 1549
    },
    {
      "epoch": 0.25987634915644975,
      "grad_norm": 0.5611308217048645,
      "learning_rate": 1.7368746516692928e-05,
      "loss": 1.023,
      "step": 1550
    },
    {
      "epoch": 0.26004401131719584,
      "grad_norm": 0.5159692764282227,
      "learning_rate": 1.73650741960932e-05,
      "loss": 0.9357,
      "step": 1551
    },
    {
      "epoch": 0.26021167347794194,
      "grad_norm": 0.5313618183135986,
      "learning_rate": 1.7361399703441893e-05,
      "loss": 1.044,
      "step": 1552
    },
    {
      "epoch": 0.26037933563868804,
      "grad_norm": 0.5279186367988586,
      "learning_rate": 1.735772303982266e-05,
      "loss": 0.9254,
      "step": 1553
    },
    {
      "epoch": 0.26054699779943413,
      "grad_norm": 0.6467424035072327,
      "learning_rate": 1.735404420631979e-05,
      "loss": 0.9476,
      "step": 1554
    },
    {
      "epoch": 0.26071465996018023,
      "grad_norm": 0.48677462339401245,
      "learning_rate": 1.7350363204018227e-05,
      "loss": 0.9348,
      "step": 1555
    },
    {
      "epoch": 0.2608823221209263,
      "grad_norm": 0.7248123288154602,
      "learning_rate": 1.7346680034003528e-05,
      "loss": 0.9353,
      "step": 1556
    },
    {
      "epoch": 0.2610499842816724,
      "grad_norm": 0.41243961453437805,
      "learning_rate": 1.734299469736192e-05,
      "loss": 0.9441,
      "step": 1557
    },
    {
      "epoch": 0.2612176464424185,
      "grad_norm": 0.548994779586792,
      "learning_rate": 1.7339307195180248e-05,
      "loss": 0.9101,
      "step": 1558
    },
    {
      "epoch": 0.2613853086031646,
      "grad_norm": 0.5552588701248169,
      "learning_rate": 1.7335617528546e-05,
      "loss": 0.9204,
      "step": 1559
    },
    {
      "epoch": 0.2615529707639107,
      "grad_norm": 0.6628120541572571,
      "learning_rate": 1.733192569854731e-05,
      "loss": 0.9021,
      "step": 1560
    },
    {
      "epoch": 0.2617206329246568,
      "grad_norm": 0.5505399107933044,
      "learning_rate": 1.732823170627294e-05,
      "loss": 0.8775,
      "step": 1561
    },
    {
      "epoch": 0.2618882950854029,
      "grad_norm": 0.44663092494010925,
      "learning_rate": 1.7324535552812298e-05,
      "loss": 0.8786,
      "step": 1562
    },
    {
      "epoch": 0.262055957246149,
      "grad_norm": 0.47220563888549805,
      "learning_rate": 1.732083723925542e-05,
      "loss": 0.9145,
      "step": 1563
    },
    {
      "epoch": 0.2622236194068951,
      "grad_norm": 0.5293234586715698,
      "learning_rate": 1.731713676669299e-05,
      "loss": 0.9809,
      "step": 1564
    },
    {
      "epoch": 0.2623912815676412,
      "grad_norm": 0.35137319564819336,
      "learning_rate": 1.731343413621632e-05,
      "loss": 0.924,
      "step": 1565
    },
    {
      "epoch": 0.2625589437283873,
      "grad_norm": 0.38129615783691406,
      "learning_rate": 1.7309729348917364e-05,
      "loss": 0.9275,
      "step": 1566
    },
    {
      "epoch": 0.2627266058891334,
      "grad_norm": 0.5752395391464233,
      "learning_rate": 1.7306022405888705e-05,
      "loss": 0.9061,
      "step": 1567
    },
    {
      "epoch": 0.2628942680498795,
      "grad_norm": 0.5390276908874512,
      "learning_rate": 1.7302313308223574e-05,
      "loss": 0.905,
      "step": 1568
    },
    {
      "epoch": 0.26306193021062557,
      "grad_norm": 0.5569885969161987,
      "learning_rate": 1.729860205701582e-05,
      "loss": 0.9214,
      "step": 1569
    },
    {
      "epoch": 0.26322959237137167,
      "grad_norm": 0.5757485032081604,
      "learning_rate": 1.7294888653359945e-05,
      "loss": 0.9978,
      "step": 1570
    },
    {
      "epoch": 0.26339725453211776,
      "grad_norm": 0.5486928224563599,
      "learning_rate": 1.7291173098351077e-05,
      "loss": 0.9747,
      "step": 1571
    },
    {
      "epoch": 0.26356491669286386,
      "grad_norm": 0.43678364157676697,
      "learning_rate": 1.7287455393084972e-05,
      "loss": 0.8731,
      "step": 1572
    },
    {
      "epoch": 0.26373257885360996,
      "grad_norm": 0.47250810265541077,
      "learning_rate": 1.7283735538658034e-05,
      "loss": 0.8917,
      "step": 1573
    },
    {
      "epoch": 0.26390024101435605,
      "grad_norm": 0.4039928615093231,
      "learning_rate": 1.728001353616729e-05,
      "loss": 0.91,
      "step": 1574
    },
    {
      "epoch": 0.26406790317510215,
      "grad_norm": 0.47751352190971375,
      "learning_rate": 1.727628938671041e-05,
      "loss": 0.96,
      "step": 1575
    },
    {
      "epoch": 0.26423556533584824,
      "grad_norm": 0.49584338068962097,
      "learning_rate": 1.7272563091385694e-05,
      "loss": 0.94,
      "step": 1576
    },
    {
      "epoch": 0.26440322749659434,
      "grad_norm": 0.43514296412467957,
      "learning_rate": 1.7268834651292062e-05,
      "loss": 0.9175,
      "step": 1577
    },
    {
      "epoch": 0.26457088965734044,
      "grad_norm": 0.49915921688079834,
      "learning_rate": 1.726510406752908e-05,
      "loss": 0.967,
      "step": 1578
    },
    {
      "epoch": 0.26473855181808653,
      "grad_norm": 0.5309367775917053,
      "learning_rate": 1.726137134119695e-05,
      "loss": 0.8521,
      "step": 1579
    },
    {
      "epoch": 0.2649062139788326,
      "grad_norm": 0.6773521900177002,
      "learning_rate": 1.7257636473396496e-05,
      "loss": 0.8804,
      "step": 1580
    },
    {
      "epoch": 0.2650738761395787,
      "grad_norm": 0.5327140092849731,
      "learning_rate": 1.7253899465229175e-05,
      "loss": 0.8907,
      "step": 1581
    },
    {
      "epoch": 0.2652415383003248,
      "grad_norm": 0.3631095588207245,
      "learning_rate": 1.725016031779708e-05,
      "loss": 0.9123,
      "step": 1582
    },
    {
      "epoch": 0.2654092004610709,
      "grad_norm": 0.47867029905319214,
      "learning_rate": 1.7246419032202926e-05,
      "loss": 0.8792,
      "step": 1583
    },
    {
      "epoch": 0.265576862621817,
      "grad_norm": 0.47533097863197327,
      "learning_rate": 1.7242675609550067e-05,
      "loss": 0.8526,
      "step": 1584
    },
    {
      "epoch": 0.26574452478256316,
      "grad_norm": 0.6036980748176575,
      "learning_rate": 1.723893005094249e-05,
      "loss": 0.8762,
      "step": 1585
    },
    {
      "epoch": 0.26591218694330926,
      "grad_norm": 0.6582639217376709,
      "learning_rate": 1.7235182357484802e-05,
      "loss": 0.779,
      "step": 1586
    },
    {
      "epoch": 0.26607984910405535,
      "grad_norm": 0.8061196208000183,
      "learning_rate": 1.7231432530282246e-05,
      "loss": 0.9484,
      "step": 1587
    },
    {
      "epoch": 0.26624751126480145,
      "grad_norm": 0.4974749982357025,
      "learning_rate": 1.722768057044069e-05,
      "loss": 0.9257,
      "step": 1588
    },
    {
      "epoch": 0.26641517342554755,
      "grad_norm": 0.570415198802948,
      "learning_rate": 1.7223926479066636e-05,
      "loss": 0.9261,
      "step": 1589
    },
    {
      "epoch": 0.26658283558629364,
      "grad_norm": 1.189063549041748,
      "learning_rate": 1.722017025726721e-05,
      "loss": 0.9155,
      "step": 1590
    },
    {
      "epoch": 0.26675049774703974,
      "grad_norm": 0.6250978708267212,
      "learning_rate": 1.721641190615017e-05,
      "loss": 0.8973,
      "step": 1591
    },
    {
      "epoch": 0.26691815990778583,
      "grad_norm": 0.4300852417945862,
      "learning_rate": 1.7212651426823895e-05,
      "loss": 0.8577,
      "step": 1592
    },
    {
      "epoch": 0.26708582206853193,
      "grad_norm": 0.3604830503463745,
      "learning_rate": 1.7208888820397404e-05,
      "loss": 0.8551,
      "step": 1593
    },
    {
      "epoch": 0.267253484229278,
      "grad_norm": 0.5577496886253357,
      "learning_rate": 1.7205124087980334e-05,
      "loss": 1.0528,
      "step": 1594
    },
    {
      "epoch": 0.2674211463900241,
      "grad_norm": 0.5893198251724243,
      "learning_rate": 1.7201357230682952e-05,
      "loss": 0.8834,
      "step": 1595
    },
    {
      "epoch": 0.2675888085507702,
      "grad_norm": 0.48370787501335144,
      "learning_rate": 1.7197588249616147e-05,
      "loss": 0.9746,
      "step": 1596
    },
    {
      "epoch": 0.2677564707115163,
      "grad_norm": 0.8386217355728149,
      "learning_rate": 1.719381714589144e-05,
      "loss": 0.8626,
      "step": 1597
    },
    {
      "epoch": 0.2679241328722624,
      "grad_norm": 0.5362548828125,
      "learning_rate": 1.7190043920620982e-05,
      "loss": 0.9499,
      "step": 1598
    },
    {
      "epoch": 0.2680917950330085,
      "grad_norm": 0.5042610764503479,
      "learning_rate": 1.7186268574917536e-05,
      "loss": 0.9054,
      "step": 1599
    },
    {
      "epoch": 0.2682594571937546,
      "grad_norm": 0.40954241156578064,
      "learning_rate": 1.71824911098945e-05,
      "loss": 0.8414,
      "step": 1600
    },
    {
      "epoch": 0.2684271193545007,
      "grad_norm": 0.8121725916862488,
      "learning_rate": 1.7178711526665895e-05,
      "loss": 0.8461,
      "step": 1601
    },
    {
      "epoch": 0.2685947815152468,
      "grad_norm": 0.4340389370918274,
      "learning_rate": 1.7174929826346373e-05,
      "loss": 0.8011,
      "step": 1602
    },
    {
      "epoch": 0.2687624436759929,
      "grad_norm": 0.7912620902061462,
      "learning_rate": 1.7171146010051196e-05,
      "loss": 0.8803,
      "step": 1603
    },
    {
      "epoch": 0.268930105836739,
      "grad_norm": 0.4555317461490631,
      "learning_rate": 1.716736007889626e-05,
      "loss": 0.895,
      "step": 1604
    },
    {
      "epoch": 0.2690977679974851,
      "grad_norm": 0.5665902495384216,
      "learning_rate": 1.716357203399809e-05,
      "loss": 1.0344,
      "step": 1605
    },
    {
      "epoch": 0.2692654301582312,
      "grad_norm": 0.6170781850814819,
      "learning_rate": 1.7159781876473816e-05,
      "loss": 0.9448,
      "step": 1606
    },
    {
      "epoch": 0.2694330923189773,
      "grad_norm": 0.657977819442749,
      "learning_rate": 1.715598960744121e-05,
      "loss": 0.9308,
      "step": 1607
    },
    {
      "epoch": 0.26960075447972337,
      "grad_norm": 0.6097431778907776,
      "learning_rate": 1.7152195228018662e-05,
      "loss": 0.915,
      "step": 1608
    },
    {
      "epoch": 0.26976841664046947,
      "grad_norm": 0.4871729016304016,
      "learning_rate": 1.714839873932517e-05,
      "loss": 0.8911,
      "step": 1609
    },
    {
      "epoch": 0.26993607880121556,
      "grad_norm": 0.41206493973731995,
      "learning_rate": 1.714460014248038e-05,
      "loss": 0.8257,
      "step": 1610
    },
    {
      "epoch": 0.27010374096196166,
      "grad_norm": 0.451313316822052,
      "learning_rate": 1.7140799438604533e-05,
      "loss": 0.9392,
      "step": 1611
    },
    {
      "epoch": 0.27027140312270775,
      "grad_norm": 0.47345075011253357,
      "learning_rate": 1.713699662881851e-05,
      "loss": 0.8944,
      "step": 1612
    },
    {
      "epoch": 0.27043906528345385,
      "grad_norm": 0.39391496777534485,
      "learning_rate": 1.7133191714243805e-05,
      "loss": 0.9162,
      "step": 1613
    },
    {
      "epoch": 0.27060672744419995,
      "grad_norm": 0.5918540358543396,
      "learning_rate": 1.7129384696002533e-05,
      "loss": 0.9649,
      "step": 1614
    },
    {
      "epoch": 0.27077438960494604,
      "grad_norm": 0.8174153566360474,
      "learning_rate": 1.712557557521743e-05,
      "loss": 0.9903,
      "step": 1615
    },
    {
      "epoch": 0.27094205176569214,
      "grad_norm": 0.5288435220718384,
      "learning_rate": 1.712176435301186e-05,
      "loss": 0.9516,
      "step": 1616
    },
    {
      "epoch": 0.27110971392643823,
      "grad_norm": 0.46600520610809326,
      "learning_rate": 1.7117951030509795e-05,
      "loss": 0.9965,
      "step": 1617
    },
    {
      "epoch": 0.27127737608718433,
      "grad_norm": 0.4489824175834656,
      "learning_rate": 1.7114135608835826e-05,
      "loss": 0.9398,
      "step": 1618
    },
    {
      "epoch": 0.2714450382479304,
      "grad_norm": 0.5334166884422302,
      "learning_rate": 1.7110318089115178e-05,
      "loss": 0.9217,
      "step": 1619
    },
    {
      "epoch": 0.2716127004086765,
      "grad_norm": 0.5517166256904602,
      "learning_rate": 1.7106498472473674e-05,
      "loss": 0.8955,
      "step": 1620
    },
    {
      "epoch": 0.2717803625694226,
      "grad_norm": 0.37127768993377686,
      "learning_rate": 1.710267676003778e-05,
      "loss": 0.8803,
      "step": 1621
    },
    {
      "epoch": 0.2719480247301687,
      "grad_norm": 0.44107526540756226,
      "learning_rate": 1.7098852952934555e-05,
      "loss": 0.9826,
      "step": 1622
    },
    {
      "epoch": 0.2721156868909148,
      "grad_norm": 0.5912244915962219,
      "learning_rate": 1.709502705229169e-05,
      "loss": 0.9638,
      "step": 1623
    },
    {
      "epoch": 0.2722833490516609,
      "grad_norm": 0.4062788486480713,
      "learning_rate": 1.709119905923749e-05,
      "loss": 0.8838,
      "step": 1624
    },
    {
      "epoch": 0.272451011212407,
      "grad_norm": 0.4899587035179138,
      "learning_rate": 1.7087368974900876e-05,
      "loss": 0.9641,
      "step": 1625
    },
    {
      "epoch": 0.2726186733731531,
      "grad_norm": 0.3973456919193268,
      "learning_rate": 1.7083536800411392e-05,
      "loss": 0.919,
      "step": 1626
    },
    {
      "epoch": 0.2727863355338992,
      "grad_norm": 0.5032211542129517,
      "learning_rate": 1.7079702536899193e-05,
      "loss": 0.8664,
      "step": 1627
    },
    {
      "epoch": 0.2729539976946453,
      "grad_norm": 0.8235628604888916,
      "learning_rate": 1.7075866185495046e-05,
      "loss": 0.9643,
      "step": 1628
    },
    {
      "epoch": 0.2731216598553914,
      "grad_norm": 0.4531996548175812,
      "learning_rate": 1.7072027747330348e-05,
      "loss": 0.8904,
      "step": 1629
    },
    {
      "epoch": 0.2732893220161375,
      "grad_norm": 0.44512197375297546,
      "learning_rate": 1.706818722353709e-05,
      "loss": 0.9208,
      "step": 1630
    },
    {
      "epoch": 0.2734569841768836,
      "grad_norm": 0.5449144244194031,
      "learning_rate": 1.70643446152479e-05,
      "loss": 0.92,
      "step": 1631
    },
    {
      "epoch": 0.2736246463376297,
      "grad_norm": 0.3762809932231903,
      "learning_rate": 1.7060499923596002e-05,
      "loss": 0.9175,
      "step": 1632
    },
    {
      "epoch": 0.27379230849837577,
      "grad_norm": 0.5805980563163757,
      "learning_rate": 1.705665314971525e-05,
      "loss": 0.8536,
      "step": 1633
    },
    {
      "epoch": 0.27395997065912187,
      "grad_norm": 0.5688173174858093,
      "learning_rate": 1.70528042947401e-05,
      "loss": 0.914,
      "step": 1634
    },
    {
      "epoch": 0.27412763281986796,
      "grad_norm": 0.6951828598976135,
      "learning_rate": 1.7048953359805637e-05,
      "loss": 0.9056,
      "step": 1635
    },
    {
      "epoch": 0.27429529498061406,
      "grad_norm": 0.5278764367103577,
      "learning_rate": 1.7045100346047535e-05,
      "loss": 0.9174,
      "step": 1636
    },
    {
      "epoch": 0.27446295714136015,
      "grad_norm": 0.42395704984664917,
      "learning_rate": 1.7041245254602104e-05,
      "loss": 1.0499,
      "step": 1637
    },
    {
      "epoch": 0.27463061930210625,
      "grad_norm": 0.7272180318832397,
      "learning_rate": 1.703738808660626e-05,
      "loss": 0.8348,
      "step": 1638
    },
    {
      "epoch": 0.27479828146285235,
      "grad_norm": 0.7003971338272095,
      "learning_rate": 1.7033528843197523e-05,
      "loss": 0.9273,
      "step": 1639
    },
    {
      "epoch": 0.27496594362359844,
      "grad_norm": 0.47901052236557007,
      "learning_rate": 1.702966752551404e-05,
      "loss": 0.839,
      "step": 1640
    },
    {
      "epoch": 0.27513360578434454,
      "grad_norm": 0.3797691762447357,
      "learning_rate": 1.7025804134694552e-05,
      "loss": 0.8958,
      "step": 1641
    },
    {
      "epoch": 0.27530126794509063,
      "grad_norm": 0.4849264323711395,
      "learning_rate": 1.7021938671878427e-05,
      "loss": 0.8831,
      "step": 1642
    },
    {
      "epoch": 0.27546893010583673,
      "grad_norm": 0.3957948088645935,
      "learning_rate": 1.701807113820564e-05,
      "loss": 0.9219,
      "step": 1643
    },
    {
      "epoch": 0.2756365922665828,
      "grad_norm": 0.46925845742225647,
      "learning_rate": 1.7014201534816765e-05,
      "loss": 0.9282,
      "step": 1644
    },
    {
      "epoch": 0.2758042544273289,
      "grad_norm": 0.5414051413536072,
      "learning_rate": 1.7010329862853007e-05,
      "loss": 0.9243,
      "step": 1645
    },
    {
      "epoch": 0.275971916588075,
      "grad_norm": 1.0770642757415771,
      "learning_rate": 1.700645612345616e-05,
      "loss": 0.9393,
      "step": 1646
    },
    {
      "epoch": 0.2761395787488211,
      "grad_norm": 0.4121866226196289,
      "learning_rate": 1.7002580317768646e-05,
      "loss": 0.9172,
      "step": 1647
    },
    {
      "epoch": 0.2763072409095672,
      "grad_norm": 0.6341374516487122,
      "learning_rate": 1.699870244693348e-05,
      "loss": 0.8848,
      "step": 1648
    },
    {
      "epoch": 0.2764749030703133,
      "grad_norm": 0.8026604056358337,
      "learning_rate": 1.6994822512094306e-05,
      "loss": 0.8162,
      "step": 1649
    },
    {
      "epoch": 0.2766425652310594,
      "grad_norm": 0.78305983543396,
      "learning_rate": 1.6990940514395353e-05,
      "loss": 0.8461,
      "step": 1650
    },
    {
      "epoch": 0.2768102273918055,
      "grad_norm": 0.7583958506584167,
      "learning_rate": 1.6987056454981476e-05,
      "loss": 0.9291,
      "step": 1651
    },
    {
      "epoch": 0.2769778895525516,
      "grad_norm": 0.5095685720443726,
      "learning_rate": 1.698317033499813e-05,
      "loss": 1.0257,
      "step": 1652
    },
    {
      "epoch": 0.2771455517132977,
      "grad_norm": 0.8323339223861694,
      "learning_rate": 1.697928215559138e-05,
      "loss": 0.9488,
      "step": 1653
    },
    {
      "epoch": 0.2773132138740438,
      "grad_norm": 0.6618877649307251,
      "learning_rate": 1.6975391917907903e-05,
      "loss": 0.9038,
      "step": 1654
    },
    {
      "epoch": 0.2774808760347899,
      "grad_norm": 0.43667611479759216,
      "learning_rate": 1.6971499623094972e-05,
      "loss": 0.9315,
      "step": 1655
    },
    {
      "epoch": 0.277648538195536,
      "grad_norm": 0.3366505801677704,
      "learning_rate": 1.6967605272300476e-05,
      "loss": 0.8708,
      "step": 1656
    },
    {
      "epoch": 0.2778162003562821,
      "grad_norm": 0.6763900518417358,
      "learning_rate": 1.6963708866672907e-05,
      "loss": 0.863,
      "step": 1657
    },
    {
      "epoch": 0.27798386251702817,
      "grad_norm": 0.41367295384407043,
      "learning_rate": 1.695981040736136e-05,
      "loss": 0.8219,
      "step": 1658
    },
    {
      "epoch": 0.27815152467777426,
      "grad_norm": 0.831977367401123,
      "learning_rate": 1.6955909895515544e-05,
      "loss": 0.9022,
      "step": 1659
    },
    {
      "epoch": 0.27831918683852036,
      "grad_norm": 0.670785129070282,
      "learning_rate": 1.6952007332285766e-05,
      "loss": 0.8842,
      "step": 1660
    },
    {
      "epoch": 0.27848684899926646,
      "grad_norm": 0.759387731552124,
      "learning_rate": 1.694810271882294e-05,
      "loss": 0.8635,
      "step": 1661
    },
    {
      "epoch": 0.27865451116001255,
      "grad_norm": 0.47069522738456726,
      "learning_rate": 1.6944196056278585e-05,
      "loss": 0.8868,
      "step": 1662
    },
    {
      "epoch": 0.27882217332075865,
      "grad_norm": 0.4956364631652832,
      "learning_rate": 1.6940287345804828e-05,
      "loss": 0.9507,
      "step": 1663
    },
    {
      "epoch": 0.27898983548150474,
      "grad_norm": 0.6357293725013733,
      "learning_rate": 1.6936376588554386e-05,
      "loss": 0.842,
      "step": 1664
    },
    {
      "epoch": 0.27915749764225084,
      "grad_norm": 0.48392024636268616,
      "learning_rate": 1.69324637856806e-05,
      "loss": 0.8362,
      "step": 1665
    },
    {
      "epoch": 0.27932515980299694,
      "grad_norm": 0.5613908171653748,
      "learning_rate": 1.69285489383374e-05,
      "loss": 0.9111,
      "step": 1666
    },
    {
      "epoch": 0.27949282196374303,
      "grad_norm": 0.6918414235115051,
      "learning_rate": 1.692463204767932e-05,
      "loss": 0.8713,
      "step": 1667
    },
    {
      "epoch": 0.27966048412448913,
      "grad_norm": 0.5419053435325623,
      "learning_rate": 1.6920713114861507e-05,
      "loss": 0.9079,
      "step": 1668
    },
    {
      "epoch": 0.2798281462852352,
      "grad_norm": 0.4282153248786926,
      "learning_rate": 1.69167921410397e-05,
      "loss": 0.8899,
      "step": 1669
    },
    {
      "epoch": 0.2799958084459813,
      "grad_norm": 0.3876660168170929,
      "learning_rate": 1.691286912737024e-05,
      "loss": 0.8342,
      "step": 1670
    },
    {
      "epoch": 0.28016347060672747,
      "grad_norm": 0.6596119999885559,
      "learning_rate": 1.6908944075010067e-05,
      "loss": 0.8435,
      "step": 1671
    },
    {
      "epoch": 0.28033113276747357,
      "grad_norm": 0.9446216225624084,
      "learning_rate": 1.690501698511674e-05,
      "loss": 0.9391,
      "step": 1672
    },
    {
      "epoch": 0.28049879492821966,
      "grad_norm": 0.44242310523986816,
      "learning_rate": 1.69010878588484e-05,
      "loss": 0.8816,
      "step": 1673
    },
    {
      "epoch": 0.28066645708896576,
      "grad_norm": 0.45864903926849365,
      "learning_rate": 1.6897156697363795e-05,
      "loss": 0.9007,
      "step": 1674
    },
    {
      "epoch": 0.28083411924971186,
      "grad_norm": 0.43198034167289734,
      "learning_rate": 1.689322350182227e-05,
      "loss": 1.0315,
      "step": 1675
    },
    {
      "epoch": 0.28100178141045795,
      "grad_norm": 0.4529312252998352,
      "learning_rate": 1.688928827338378e-05,
      "loss": 0.9204,
      "step": 1676
    },
    {
      "epoch": 0.28116944357120405,
      "grad_norm": 0.3595108985900879,
      "learning_rate": 1.6885351013208874e-05,
      "loss": 0.9069,
      "step": 1677
    },
    {
      "epoch": 0.28133710573195014,
      "grad_norm": 0.40380436182022095,
      "learning_rate": 1.6881411722458688e-05,
      "loss": 0.9084,
      "step": 1678
    },
    {
      "epoch": 0.28150476789269624,
      "grad_norm": 0.3343381881713867,
      "learning_rate": 1.6877470402294978e-05,
      "loss": 0.8491,
      "step": 1679
    },
    {
      "epoch": 0.28167243005344234,
      "grad_norm": 0.47201022505760193,
      "learning_rate": 1.6873527053880082e-05,
      "loss": 0.9348,
      "step": 1680
    },
    {
      "epoch": 0.28184009221418843,
      "grad_norm": 0.4359557628631592,
      "learning_rate": 1.6869581678376944e-05,
      "loss": 0.945,
      "step": 1681
    },
    {
      "epoch": 0.2820077543749345,
      "grad_norm": 0.49326834082603455,
      "learning_rate": 1.6865634276949104e-05,
      "loss": 0.9073,
      "step": 1682
    },
    {
      "epoch": 0.2821754165356806,
      "grad_norm": 0.4376702904701233,
      "learning_rate": 1.68616848507607e-05,
      "loss": 0.9262,
      "step": 1683
    },
    {
      "epoch": 0.2823430786964267,
      "grad_norm": 0.4711952209472656,
      "learning_rate": 1.685773340097647e-05,
      "loss": 0.937,
      "step": 1684
    },
    {
      "epoch": 0.2825107408571728,
      "grad_norm": 0.38133877515792847,
      "learning_rate": 1.6853779928761734e-05,
      "loss": 0.859,
      "step": 1685
    },
    {
      "epoch": 0.2826784030179189,
      "grad_norm": 0.44120439887046814,
      "learning_rate": 1.6849824435282434e-05,
      "loss": 0.9153,
      "step": 1686
    },
    {
      "epoch": 0.282846065178665,
      "grad_norm": 0.5790536403656006,
      "learning_rate": 1.6845866921705086e-05,
      "loss": 0.9605,
      "step": 1687
    },
    {
      "epoch": 0.2830137273394111,
      "grad_norm": 0.40194109082221985,
      "learning_rate": 1.684190738919681e-05,
      "loss": 0.8764,
      "step": 1688
    },
    {
      "epoch": 0.2831813895001572,
      "grad_norm": 0.466818630695343,
      "learning_rate": 1.683794583892533e-05,
      "loss": 0.8306,
      "step": 1689
    },
    {
      "epoch": 0.2833490516609033,
      "grad_norm": 0.677361011505127,
      "learning_rate": 1.6833982272058942e-05,
      "loss": 0.8976,
      "step": 1690
    },
    {
      "epoch": 0.2835167138216494,
      "grad_norm": 0.4113081991672516,
      "learning_rate": 1.683001668976656e-05,
      "loss": 0.9737,
      "step": 1691
    },
    {
      "epoch": 0.2836843759823955,
      "grad_norm": 0.5291762948036194,
      "learning_rate": 1.6826049093217682e-05,
      "loss": 0.855,
      "step": 1692
    },
    {
      "epoch": 0.2838520381431416,
      "grad_norm": 0.5291401147842407,
      "learning_rate": 1.68220794835824e-05,
      "loss": 0.9007,
      "step": 1693
    },
    {
      "epoch": 0.2840197003038877,
      "grad_norm": 0.625732958316803,
      "learning_rate": 1.6818107862031404e-05,
      "loss": 0.8679,
      "step": 1694
    },
    {
      "epoch": 0.2841873624646338,
      "grad_norm": 0.5087775588035583,
      "learning_rate": 1.681413422973597e-05,
      "loss": 0.8911,
      "step": 1695
    },
    {
      "epoch": 0.28435502462537987,
      "grad_norm": 0.5665459632873535,
      "learning_rate": 1.6810158587867973e-05,
      "loss": 0.8136,
      "step": 1696
    },
    {
      "epoch": 0.28452268678612597,
      "grad_norm": 0.5535372495651245,
      "learning_rate": 1.6806180937599882e-05,
      "loss": 0.9728,
      "step": 1697
    },
    {
      "epoch": 0.28469034894687206,
      "grad_norm": 0.5744376182556152,
      "learning_rate": 1.680220128010475e-05,
      "loss": 0.9279,
      "step": 1698
    },
    {
      "epoch": 0.28485801110761816,
      "grad_norm": 0.3833550214767456,
      "learning_rate": 1.6798219616556234e-05,
      "loss": 0.9503,
      "step": 1699
    },
    {
      "epoch": 0.28502567326836425,
      "grad_norm": 0.47775256633758545,
      "learning_rate": 1.6794235948128564e-05,
      "loss": 0.8974,
      "step": 1700
    },
    {
      "epoch": 0.28519333542911035,
      "grad_norm": 0.3792549669742584,
      "learning_rate": 1.679025027599659e-05,
      "loss": 0.8856,
      "step": 1701
    },
    {
      "epoch": 0.28536099758985645,
      "grad_norm": 0.7515558004379272,
      "learning_rate": 1.678626260133572e-05,
      "loss": 0.9097,
      "step": 1702
    },
    {
      "epoch": 0.28552865975060254,
      "grad_norm": 0.4453137218952179,
      "learning_rate": 1.678227292532198e-05,
      "loss": 0.8997,
      "step": 1703
    },
    {
      "epoch": 0.28569632191134864,
      "grad_norm": 0.35799017548561096,
      "learning_rate": 1.6778281249131973e-05,
      "loss": 0.8296,
      "step": 1704
    },
    {
      "epoch": 0.28586398407209473,
      "grad_norm": 0.6710291504859924,
      "learning_rate": 1.6774287573942885e-05,
      "loss": 0.8943,
      "step": 1705
    },
    {
      "epoch": 0.28603164623284083,
      "grad_norm": 0.6353430151939392,
      "learning_rate": 1.6770291900932513e-05,
      "loss": 1.0383,
      "step": 1706
    },
    {
      "epoch": 0.2861993083935869,
      "grad_norm": 0.40030571818351746,
      "learning_rate": 1.6766294231279218e-05,
      "loss": 0.9065,
      "step": 1707
    },
    {
      "epoch": 0.286366970554333,
      "grad_norm": 0.644238293170929,
      "learning_rate": 1.6762294566161976e-05,
      "loss": 0.9221,
      "step": 1708
    },
    {
      "epoch": 0.2865346327150791,
      "grad_norm": 0.528060257434845,
      "learning_rate": 1.6758292906760323e-05,
      "loss": 0.9305,
      "step": 1709
    },
    {
      "epoch": 0.2867022948758252,
      "grad_norm": 0.5396594405174255,
      "learning_rate": 1.675428925425441e-05,
      "loss": 0.8952,
      "step": 1710
    },
    {
      "epoch": 0.2868699570365713,
      "grad_norm": 0.5679185390472412,
      "learning_rate": 1.6750283609824956e-05,
      "loss": 0.9096,
      "step": 1711
    },
    {
      "epoch": 0.2870376191973174,
      "grad_norm": 0.3885863423347473,
      "learning_rate": 1.6746275974653278e-05,
      "loss": 0.8593,
      "step": 1712
    },
    {
      "epoch": 0.2872052813580635,
      "grad_norm": 0.6562624573707581,
      "learning_rate": 1.674226634992128e-05,
      "loss": 0.9612,
      "step": 1713
    },
    {
      "epoch": 0.2873729435188096,
      "grad_norm": 0.46650078892707825,
      "learning_rate": 1.6738254736811446e-05,
      "loss": 0.8997,
      "step": 1714
    },
    {
      "epoch": 0.2875406056795557,
      "grad_norm": 0.4466274380683899,
      "learning_rate": 1.673424113650685e-05,
      "loss": 0.8345,
      "step": 1715
    },
    {
      "epoch": 0.2877082678403018,
      "grad_norm": 0.5457987189292908,
      "learning_rate": 1.673022555019115e-05,
      "loss": 0.9359,
      "step": 1716
    },
    {
      "epoch": 0.2878759300010479,
      "grad_norm": 0.4065357744693756,
      "learning_rate": 1.6726207979048604e-05,
      "loss": 0.9402,
      "step": 1717
    },
    {
      "epoch": 0.288043592161794,
      "grad_norm": 0.4762505292892456,
      "learning_rate": 1.6722188424264036e-05,
      "loss": 0.9184,
      "step": 1718
    },
    {
      "epoch": 0.2882112543225401,
      "grad_norm": 0.40490788221359253,
      "learning_rate": 1.6718166887022856e-05,
      "loss": 0.872,
      "step": 1719
    },
    {
      "epoch": 0.2883789164832862,
      "grad_norm": 0.632838785648346,
      "learning_rate": 1.6714143368511076e-05,
      "loss": 0.9559,
      "step": 1720
    },
    {
      "epoch": 0.28854657864403227,
      "grad_norm": 0.6651273369789124,
      "learning_rate": 1.6710117869915272e-05,
      "loss": 0.8717,
      "step": 1721
    },
    {
      "epoch": 0.28871424080477837,
      "grad_norm": 0.6503715515136719,
      "learning_rate": 1.670609039242262e-05,
      "loss": 0.8381,
      "step": 1722
    },
    {
      "epoch": 0.28888190296552446,
      "grad_norm": 1.0909502506256104,
      "learning_rate": 1.6702060937220873e-05,
      "loss": 0.8412,
      "step": 1723
    },
    {
      "epoch": 0.28904956512627056,
      "grad_norm": 0.9266510605812073,
      "learning_rate": 1.6698029505498364e-05,
      "loss": 0.8889,
      "step": 1724
    },
    {
      "epoch": 0.28921722728701665,
      "grad_norm": 0.591495156288147,
      "learning_rate": 1.669399609844401e-05,
      "loss": 0.9035,
      "step": 1725
    },
    {
      "epoch": 0.28938488944776275,
      "grad_norm": 1.0084071159362793,
      "learning_rate": 1.668996071724732e-05,
      "loss": 0.9163,
      "step": 1726
    },
    {
      "epoch": 0.28955255160850885,
      "grad_norm": 0.6242332458496094,
      "learning_rate": 1.6685923363098374e-05,
      "loss": 0.9056,
      "step": 1727
    },
    {
      "epoch": 0.28972021376925494,
      "grad_norm": 0.39805349707603455,
      "learning_rate": 1.6681884037187836e-05,
      "loss": 0.836,
      "step": 1728
    },
    {
      "epoch": 0.28988787593000104,
      "grad_norm": 0.45632416009902954,
      "learning_rate": 1.6677842740706958e-05,
      "loss": 0.9636,
      "step": 1729
    },
    {
      "epoch": 0.29005553809074713,
      "grad_norm": 0.43758928775787354,
      "learning_rate": 1.667379947484756e-05,
      "loss": 0.8712,
      "step": 1730
    },
    {
      "epoch": 0.29022320025149323,
      "grad_norm": 0.5538960695266724,
      "learning_rate": 1.666975424080206e-05,
      "loss": 0.8779,
      "step": 1731
    },
    {
      "epoch": 0.2903908624122393,
      "grad_norm": 0.48360779881477356,
      "learning_rate": 1.6665707039763445e-05,
      "loss": 0.8818,
      "step": 1732
    },
    {
      "epoch": 0.2905585245729854,
      "grad_norm": 0.5961825251579285,
      "learning_rate": 1.666165787292528e-05,
      "loss": 0.9156,
      "step": 1733
    },
    {
      "epoch": 0.2907261867337315,
      "grad_norm": 0.6016607284545898,
      "learning_rate": 1.6657606741481723e-05,
      "loss": 0.9387,
      "step": 1734
    },
    {
      "epoch": 0.2908938488944776,
      "grad_norm": 0.5556625127792358,
      "learning_rate": 1.6653553646627494e-05,
      "loss": 0.9521,
      "step": 1735
    },
    {
      "epoch": 0.2910615110552237,
      "grad_norm": 0.7546951174736023,
      "learning_rate": 1.6649498589557908e-05,
      "loss": 0.8334,
      "step": 1736
    },
    {
      "epoch": 0.2912291732159698,
      "grad_norm": 0.6190481781959534,
      "learning_rate": 1.6645441571468847e-05,
      "loss": 0.9916,
      "step": 1737
    },
    {
      "epoch": 0.2913968353767159,
      "grad_norm": 0.6533049941062927,
      "learning_rate": 1.6641382593556776e-05,
      "loss": 0.9274,
      "step": 1738
    },
    {
      "epoch": 0.291564497537462,
      "grad_norm": 0.5003221035003662,
      "learning_rate": 1.6637321657018742e-05,
      "loss": 0.9809,
      "step": 1739
    },
    {
      "epoch": 0.2917321596982081,
      "grad_norm": 0.5371973514556885,
      "learning_rate": 1.663325876305236e-05,
      "loss": 0.8339,
      "step": 1740
    },
    {
      "epoch": 0.2918998218589542,
      "grad_norm": 0.4703129231929779,
      "learning_rate": 1.662919391285583e-05,
      "loss": 1.0353,
      "step": 1741
    },
    {
      "epoch": 0.2920674840197003,
      "grad_norm": 0.4482932984828949,
      "learning_rate": 1.6625127107627928e-05,
      "loss": 1.014,
      "step": 1742
    },
    {
      "epoch": 0.2922351461804464,
      "grad_norm": 0.620073139667511,
      "learning_rate": 1.6621058348568008e-05,
      "loss": 0.8419,
      "step": 1743
    },
    {
      "epoch": 0.2924028083411925,
      "grad_norm": 0.4446816146373749,
      "learning_rate": 1.6616987636875987e-05,
      "loss": 0.822,
      "step": 1744
    },
    {
      "epoch": 0.2925704705019386,
      "grad_norm": 0.4312778115272522,
      "learning_rate": 1.661291497375238e-05,
      "loss": 0.909,
      "step": 1745
    },
    {
      "epoch": 0.29273813266268467,
      "grad_norm": 0.4368993639945984,
      "learning_rate": 1.6608840360398257e-05,
      "loss": 0.9185,
      "step": 1746
    },
    {
      "epoch": 0.29290579482343077,
      "grad_norm": 0.43704938888549805,
      "learning_rate": 1.660476379801528e-05,
      "loss": 0.9995,
      "step": 1747
    },
    {
      "epoch": 0.29307345698417686,
      "grad_norm": 0.585857093334198,
      "learning_rate": 1.660068528780567e-05,
      "loss": 0.9022,
      "step": 1748
    },
    {
      "epoch": 0.29324111914492296,
      "grad_norm": 0.4600761830806732,
      "learning_rate": 1.6596604830972234e-05,
      "loss": 0.91,
      "step": 1749
    },
    {
      "epoch": 0.29340878130566905,
      "grad_norm": 0.43956708908081055,
      "learning_rate": 1.6592522428718348e-05,
      "loss": 0.9309,
      "step": 1750
    },
    {
      "epoch": 0.29357644346641515,
      "grad_norm": 0.6964955925941467,
      "learning_rate": 1.6588438082247966e-05,
      "loss": 0.895,
      "step": 1751
    },
    {
      "epoch": 0.29374410562716124,
      "grad_norm": 0.45925039052963257,
      "learning_rate": 1.6584351792765612e-05,
      "loss": 0.8934,
      "step": 1752
    },
    {
      "epoch": 0.29391176778790734,
      "grad_norm": 0.5126753449440002,
      "learning_rate": 1.658026356147638e-05,
      "loss": 0.9728,
      "step": 1753
    },
    {
      "epoch": 0.29407942994865344,
      "grad_norm": 0.4704686999320984,
      "learning_rate": 1.6576173389585938e-05,
      "loss": 0.9259,
      "step": 1754
    },
    {
      "epoch": 0.29424709210939953,
      "grad_norm": 0.46139395236968994,
      "learning_rate": 1.657208127830053e-05,
      "loss": 0.9183,
      "step": 1755
    },
    {
      "epoch": 0.29441475427014563,
      "grad_norm": 0.4636790454387665,
      "learning_rate": 1.656798722882698e-05,
      "loss": 0.8836,
      "step": 1756
    },
    {
      "epoch": 0.2945824164308918,
      "grad_norm": 0.4975752532482147,
      "learning_rate": 1.656389124237266e-05,
      "loss": 0.8966,
      "step": 1757
    },
    {
      "epoch": 0.2947500785916379,
      "grad_norm": 0.5077047348022461,
      "learning_rate": 1.6559793320145536e-05,
      "loss": 0.9251,
      "step": 1758
    },
    {
      "epoch": 0.29491774075238397,
      "grad_norm": 0.544331431388855,
      "learning_rate": 1.6555693463354133e-05,
      "loss": 0.9085,
      "step": 1759
    },
    {
      "epoch": 0.29508540291313007,
      "grad_norm": 0.4727647602558136,
      "learning_rate": 1.655159167320755e-05,
      "loss": 0.8702,
      "step": 1760
    },
    {
      "epoch": 0.29525306507387616,
      "grad_norm": 0.703439474105835,
      "learning_rate": 1.6547487950915458e-05,
      "loss": 0.8626,
      "step": 1761
    },
    {
      "epoch": 0.29542072723462226,
      "grad_norm": 0.49445345997810364,
      "learning_rate": 1.654338229768809e-05,
      "loss": 0.9302,
      "step": 1762
    },
    {
      "epoch": 0.29558838939536836,
      "grad_norm": 0.4604395627975464,
      "learning_rate": 1.6539274714736255e-05,
      "loss": 0.8253,
      "step": 1763
    },
    {
      "epoch": 0.29575605155611445,
      "grad_norm": 0.8136701583862305,
      "learning_rate": 1.6535165203271334e-05,
      "loss": 0.8502,
      "step": 1764
    },
    {
      "epoch": 0.29592371371686055,
      "grad_norm": 0.4312807619571686,
      "learning_rate": 1.653105376450527e-05,
      "loss": 0.9097,
      "step": 1765
    },
    {
      "epoch": 0.29609137587760664,
      "grad_norm": 0.4612438380718231,
      "learning_rate": 1.652694039965058e-05,
      "loss": 0.9087,
      "step": 1766
    },
    {
      "epoch": 0.29625903803835274,
      "grad_norm": 0.4308323860168457,
      "learning_rate": 1.6522825109920342e-05,
      "loss": 0.8786,
      "step": 1767
    },
    {
      "epoch": 0.29642670019909884,
      "grad_norm": 0.7085330486297607,
      "learning_rate": 1.651870789652821e-05,
      "loss": 0.868,
      "step": 1768
    },
    {
      "epoch": 0.29659436235984493,
      "grad_norm": 0.466951459646225,
      "learning_rate": 1.6514588760688397e-05,
      "loss": 0.8911,
      "step": 1769
    },
    {
      "epoch": 0.29676202452059103,
      "grad_norm": 0.864098310470581,
      "learning_rate": 1.6510467703615694e-05,
      "loss": 0.8381,
      "step": 1770
    },
    {
      "epoch": 0.2969296866813371,
      "grad_norm": 0.515808641910553,
      "learning_rate": 1.6506344726525443e-05,
      "loss": 0.8305,
      "step": 1771
    },
    {
      "epoch": 0.2970973488420832,
      "grad_norm": 0.5473104119300842,
      "learning_rate": 1.6502219830633567e-05,
      "loss": 0.8603,
      "step": 1772
    },
    {
      "epoch": 0.2972650110028293,
      "grad_norm": 0.3563365340232849,
      "learning_rate": 1.6498093017156545e-05,
      "loss": 0.8859,
      "step": 1773
    },
    {
      "epoch": 0.2974326731635754,
      "grad_norm": 0.5445033311843872,
      "learning_rate": 1.649396428731143e-05,
      "loss": 0.9604,
      "step": 1774
    },
    {
      "epoch": 0.2976003353243215,
      "grad_norm": 0.40800729393959045,
      "learning_rate": 1.6489833642315834e-05,
      "loss": 0.8372,
      "step": 1775
    },
    {
      "epoch": 0.2977679974850676,
      "grad_norm": 0.4679819345474243,
      "learning_rate": 1.648570108338793e-05,
      "loss": 0.9257,
      "step": 1776
    },
    {
      "epoch": 0.2979356596458137,
      "grad_norm": 0.3677198588848114,
      "learning_rate": 1.648156661174647e-05,
      "loss": 0.8854,
      "step": 1777
    },
    {
      "epoch": 0.2981033218065598,
      "grad_norm": 0.5773995518684387,
      "learning_rate": 1.6477430228610755e-05,
      "loss": 0.9674,
      "step": 1778
    },
    {
      "epoch": 0.2982709839673059,
      "grad_norm": 0.44452473521232605,
      "learning_rate": 1.6473291935200657e-05,
      "loss": 0.9222,
      "step": 1779
    },
    {
      "epoch": 0.298438646128052,
      "grad_norm": 0.4076407849788666,
      "learning_rate": 1.646915173273661e-05,
      "loss": 0.8526,
      "step": 1780
    },
    {
      "epoch": 0.2986063082887981,
      "grad_norm": 0.5405694246292114,
      "learning_rate": 1.6465009622439607e-05,
      "loss": 0.8831,
      "step": 1781
    },
    {
      "epoch": 0.2987739704495442,
      "grad_norm": 0.44485101103782654,
      "learning_rate": 1.6460865605531214e-05,
      "loss": 0.949,
      "step": 1782
    },
    {
      "epoch": 0.2989416326102903,
      "grad_norm": 0.529278039932251,
      "learning_rate": 1.645671968323355e-05,
      "loss": 0.8862,
      "step": 1783
    },
    {
      "epoch": 0.29910929477103637,
      "grad_norm": 0.4891248643398285,
      "learning_rate": 1.64525718567693e-05,
      "loss": 0.8989,
      "step": 1784
    },
    {
      "epoch": 0.29927695693178247,
      "grad_norm": 0.4654734432697296,
      "learning_rate": 1.6448422127361707e-05,
      "loss": 0.8702,
      "step": 1785
    },
    {
      "epoch": 0.29944461909252856,
      "grad_norm": 0.5863080024719238,
      "learning_rate": 1.6444270496234583e-05,
      "loss": 0.9906,
      "step": 1786
    },
    {
      "epoch": 0.29961228125327466,
      "grad_norm": 0.4149811267852783,
      "learning_rate": 1.6440116964612287e-05,
      "loss": 0.9792,
      "step": 1787
    },
    {
      "epoch": 0.29977994341402076,
      "grad_norm": 0.6241120100021362,
      "learning_rate": 1.6435961533719753e-05,
      "loss": 1.0157,
      "step": 1788
    },
    {
      "epoch": 0.29994760557476685,
      "grad_norm": 0.677161455154419,
      "learning_rate": 1.6431804204782468e-05,
      "loss": 0.9572,
      "step": 1789
    },
    {
      "epoch": 0.30011526773551295,
      "grad_norm": 0.5379821062088013,
      "learning_rate": 1.6427644979026473e-05,
      "loss": 0.9378,
      "step": 1790
    },
    {
      "epoch": 0.30028292989625904,
      "grad_norm": 0.40610626339912415,
      "learning_rate": 1.642348385767839e-05,
      "loss": 0.9116,
      "step": 1791
    },
    {
      "epoch": 0.30045059205700514,
      "grad_norm": 0.47444257140159607,
      "learning_rate": 1.641932084196537e-05,
      "loss": 0.7692,
      "step": 1792
    },
    {
      "epoch": 0.30061825421775124,
      "grad_norm": 0.44950932264328003,
      "learning_rate": 1.641515593311515e-05,
      "loss": 0.8458,
      "step": 1793
    },
    {
      "epoch": 0.30078591637849733,
      "grad_norm": 0.5710813999176025,
      "learning_rate": 1.6410989132356005e-05,
      "loss": 0.9122,
      "step": 1794
    },
    {
      "epoch": 0.3009535785392434,
      "grad_norm": 0.45583242177963257,
      "learning_rate": 1.6406820440916778e-05,
      "loss": 0.8478,
      "step": 1795
    },
    {
      "epoch": 0.3011212406999895,
      "grad_norm": 0.5046128630638123,
      "learning_rate": 1.6402649860026868e-05,
      "loss": 0.9029,
      "step": 1796
    },
    {
      "epoch": 0.3012889028607356,
      "grad_norm": 0.5096794962882996,
      "learning_rate": 1.6398477390916235e-05,
      "loss": 0.8417,
      "step": 1797
    },
    {
      "epoch": 0.3014565650214817,
      "grad_norm": 0.510442852973938,
      "learning_rate": 1.6394303034815387e-05,
      "loss": 0.8932,
      "step": 1798
    },
    {
      "epoch": 0.3016242271822278,
      "grad_norm": 0.6885331869125366,
      "learning_rate": 1.639012679295539e-05,
      "loss": 0.9257,
      "step": 1799
    },
    {
      "epoch": 0.3017918893429739,
      "grad_norm": 0.8317268490791321,
      "learning_rate": 1.638594866656788e-05,
      "loss": 0.8742,
      "step": 1800
    },
    {
      "epoch": 0.30195955150372,
      "grad_norm": 0.5611824989318848,
      "learning_rate": 1.638176865688503e-05,
      "loss": 0.8217,
      "step": 1801
    },
    {
      "epoch": 0.3021272136644661,
      "grad_norm": 0.4200962781906128,
      "learning_rate": 1.637758676513958e-05,
      "loss": 0.8437,
      "step": 1802
    },
    {
      "epoch": 0.3022948758252122,
      "grad_norm": 0.6332334876060486,
      "learning_rate": 1.637340299256482e-05,
      "loss": 0.9094,
      "step": 1803
    },
    {
      "epoch": 0.3024625379859583,
      "grad_norm": 0.6723257899284363,
      "learning_rate": 1.6369217340394595e-05,
      "loss": 0.9367,
      "step": 1804
    },
    {
      "epoch": 0.3026302001467044,
      "grad_norm": 0.6654242873191833,
      "learning_rate": 1.6365029809863308e-05,
      "loss": 0.7906,
      "step": 1805
    },
    {
      "epoch": 0.3027978623074505,
      "grad_norm": 0.5226824283599854,
      "learning_rate": 1.6360840402205914e-05,
      "loss": 0.9066,
      "step": 1806
    },
    {
      "epoch": 0.3029655244681966,
      "grad_norm": 0.6385287642478943,
      "learning_rate": 1.635664911865792e-05,
      "loss": 0.9012,
      "step": 1807
    },
    {
      "epoch": 0.3031331866289427,
      "grad_norm": 1.1316041946411133,
      "learning_rate": 1.6352455960455385e-05,
      "loss": 0.9579,
      "step": 1808
    },
    {
      "epoch": 0.30330084878968877,
      "grad_norm": 0.4610840678215027,
      "learning_rate": 1.6348260928834932e-05,
      "loss": 0.9515,
      "step": 1809
    },
    {
      "epoch": 0.30346851095043487,
      "grad_norm": 0.4964912533760071,
      "learning_rate": 1.634406402503372e-05,
      "loss": 0.8182,
      "step": 1810
    },
    {
      "epoch": 0.30363617311118096,
      "grad_norm": 0.6284961700439453,
      "learning_rate": 1.6339865250289468e-05,
      "loss": 0.8412,
      "step": 1811
    },
    {
      "epoch": 0.30380383527192706,
      "grad_norm": 0.5082619786262512,
      "learning_rate": 1.6335664605840446e-05,
      "loss": 0.9138,
      "step": 1812
    },
    {
      "epoch": 0.30397149743267315,
      "grad_norm": 0.5302368402481079,
      "learning_rate": 1.6331462092925485e-05,
      "loss": 0.8233,
      "step": 1813
    },
    {
      "epoch": 0.30413915959341925,
      "grad_norm": 0.4402926564216614,
      "learning_rate": 1.6327257712783945e-05,
      "loss": 0.8033,
      "step": 1814
    },
    {
      "epoch": 0.30430682175416535,
      "grad_norm": 0.5954879522323608,
      "learning_rate": 1.6323051466655756e-05,
      "loss": 0.9123,
      "step": 1815
    },
    {
      "epoch": 0.30447448391491144,
      "grad_norm": 0.5569353699684143,
      "learning_rate": 1.6318843355781395e-05,
      "loss": 0.8772,
      "step": 1816
    },
    {
      "epoch": 0.30464214607565754,
      "grad_norm": 0.5235996246337891,
      "learning_rate": 1.6314633381401882e-05,
      "loss": 0.8687,
      "step": 1817
    },
    {
      "epoch": 0.30480980823640363,
      "grad_norm": 0.4863227307796478,
      "learning_rate": 1.6310421544758794e-05,
      "loss": 0.929,
      "step": 1818
    },
    {
      "epoch": 0.30497747039714973,
      "grad_norm": 0.3899315297603607,
      "learning_rate": 1.630620784709425e-05,
      "loss": 0.8168,
      "step": 1819
    },
    {
      "epoch": 0.3051451325578958,
      "grad_norm": 0.4086456596851349,
      "learning_rate": 1.6301992289650925e-05,
      "loss": 0.9013,
      "step": 1820
    },
    {
      "epoch": 0.3053127947186419,
      "grad_norm": 0.5903594493865967,
      "learning_rate": 1.6297774873672036e-05,
      "loss": 0.9235,
      "step": 1821
    },
    {
      "epoch": 0.305480456879388,
      "grad_norm": 0.5880112648010254,
      "learning_rate": 1.629355560040135e-05,
      "loss": 0.8814,
      "step": 1822
    },
    {
      "epoch": 0.3056481190401341,
      "grad_norm": 0.48934492468833923,
      "learning_rate": 1.6289334471083194e-05,
      "loss": 0.8804,
      "step": 1823
    },
    {
      "epoch": 0.3058157812008802,
      "grad_norm": 0.48269081115722656,
      "learning_rate": 1.6285111486962414e-05,
      "loss": 0.8828,
      "step": 1824
    },
    {
      "epoch": 0.3059834433616263,
      "grad_norm": 0.6265968680381775,
      "learning_rate": 1.6280886649284432e-05,
      "loss": 0.8181,
      "step": 1825
    },
    {
      "epoch": 0.3061511055223724,
      "grad_norm": 0.5036160945892334,
      "learning_rate": 1.62766599592952e-05,
      "loss": 0.7815,
      "step": 1826
    },
    {
      "epoch": 0.3063187676831185,
      "grad_norm": 0.5734903812408447,
      "learning_rate": 1.627243141824123e-05,
      "loss": 0.823,
      "step": 1827
    },
    {
      "epoch": 0.3064864298438646,
      "grad_norm": 0.5602980256080627,
      "learning_rate": 1.626820102736956e-05,
      "loss": 1.0331,
      "step": 1828
    },
    {
      "epoch": 0.3066540920046107,
      "grad_norm": 0.6858996748924255,
      "learning_rate": 1.626396878792779e-05,
      "loss": 0.8685,
      "step": 1829
    },
    {
      "epoch": 0.3068217541653568,
      "grad_norm": 0.6805505752563477,
      "learning_rate": 1.6259734701164058e-05,
      "loss": 0.9075,
      "step": 1830
    },
    {
      "epoch": 0.3069894163261029,
      "grad_norm": 0.5220349431037903,
      "learning_rate": 1.625549876832705e-05,
      "loss": 0.948,
      "step": 1831
    },
    {
      "epoch": 0.307157078486849,
      "grad_norm": 0.4397321939468384,
      "learning_rate": 1.6251260990665993e-05,
      "loss": 0.8982,
      "step": 1832
    },
    {
      "epoch": 0.3073247406475951,
      "grad_norm": 0.542477011680603,
      "learning_rate": 1.624702136943066e-05,
      "loss": 0.9823,
      "step": 1833
    },
    {
      "epoch": 0.30749240280834117,
      "grad_norm": 0.45024630427360535,
      "learning_rate": 1.6242779905871375e-05,
      "loss": 0.901,
      "step": 1834
    },
    {
      "epoch": 0.30766006496908727,
      "grad_norm": 0.6258662939071655,
      "learning_rate": 1.623853660123899e-05,
      "loss": 0.9651,
      "step": 1835
    },
    {
      "epoch": 0.30782772712983336,
      "grad_norm": 0.47294142842292786,
      "learning_rate": 1.6234291456784912e-05,
      "loss": 0.9054,
      "step": 1836
    },
    {
      "epoch": 0.30799538929057946,
      "grad_norm": 0.5752155780792236,
      "learning_rate": 1.623004447376108e-05,
      "loss": 0.9933,
      "step": 1837
    },
    {
      "epoch": 0.30816305145132555,
      "grad_norm": 0.48368701338768005,
      "learning_rate": 1.6225795653419995e-05,
      "loss": 0.9122,
      "step": 1838
    },
    {
      "epoch": 0.30833071361207165,
      "grad_norm": 0.5220962762832642,
      "learning_rate": 1.6221544997014673e-05,
      "loss": 0.9381,
      "step": 1839
    },
    {
      "epoch": 0.30849837577281775,
      "grad_norm": 0.5780762434005737,
      "learning_rate": 1.621729250579869e-05,
      "loss": 1.042,
      "step": 1840
    },
    {
      "epoch": 0.30866603793356384,
      "grad_norm": 0.44534143805503845,
      "learning_rate": 1.6213038181026166e-05,
      "loss": 0.8988,
      "step": 1841
    },
    {
      "epoch": 0.30883370009431,
      "grad_norm": 0.5983442068099976,
      "learning_rate": 1.6208782023951744e-05,
      "loss": 0.8925,
      "step": 1842
    },
    {
      "epoch": 0.3090013622550561,
      "grad_norm": 0.3870874345302582,
      "learning_rate": 1.620452403583062e-05,
      "loss": 0.8957,
      "step": 1843
    },
    {
      "epoch": 0.3091690244158022,
      "grad_norm": 0.46022143959999084,
      "learning_rate": 1.620026421791853e-05,
      "loss": 0.925,
      "step": 1844
    },
    {
      "epoch": 0.3093366865765483,
      "grad_norm": 0.36380746960639954,
      "learning_rate": 1.6196002571471745e-05,
      "loss": 0.8479,
      "step": 1845
    },
    {
      "epoch": 0.3095043487372944,
      "grad_norm": 0.6444778442382812,
      "learning_rate": 1.619173909774708e-05,
      "loss": 0.9413,
      "step": 1846
    },
    {
      "epoch": 0.3096720108980405,
      "grad_norm": 0.36853939294815063,
      "learning_rate": 1.618747379800188e-05,
      "loss": 0.839,
      "step": 1847
    },
    {
      "epoch": 0.30983967305878657,
      "grad_norm": 0.5956276655197144,
      "learning_rate": 1.618320667349404e-05,
      "loss": 0.8552,
      "step": 1848
    },
    {
      "epoch": 0.31000733521953266,
      "grad_norm": 0.7519698143005371,
      "learning_rate": 1.6178937725481984e-05,
      "loss": 0.9885,
      "step": 1849
    },
    {
      "epoch": 0.31017499738027876,
      "grad_norm": 0.7840356826782227,
      "learning_rate": 1.6174666955224686e-05,
      "loss": 0.9566,
      "step": 1850
    },
    {
      "epoch": 0.31034265954102486,
      "grad_norm": 0.4490809738636017,
      "learning_rate": 1.617039436398164e-05,
      "loss": 1.0007,
      "step": 1851
    },
    {
      "epoch": 0.31051032170177095,
      "grad_norm": 0.4416768550872803,
      "learning_rate": 1.616611995301289e-05,
      "loss": 0.8151,
      "step": 1852
    },
    {
      "epoch": 0.31067798386251705,
      "grad_norm": 0.5982283353805542,
      "learning_rate": 1.616184372357901e-05,
      "loss": 0.9484,
      "step": 1853
    },
    {
      "epoch": 0.31084564602326314,
      "grad_norm": 0.5351338386535645,
      "learning_rate": 1.6157565676941115e-05,
      "loss": 0.9318,
      "step": 1854
    },
    {
      "epoch": 0.31101330818400924,
      "grad_norm": 0.3872014582157135,
      "learning_rate": 1.6153285814360856e-05,
      "loss": 1.0033,
      "step": 1855
    },
    {
      "epoch": 0.31118097034475534,
      "grad_norm": 0.48003336787223816,
      "learning_rate": 1.6149004137100412e-05,
      "loss": 0.8638,
      "step": 1856
    },
    {
      "epoch": 0.31134863250550143,
      "grad_norm": 0.5285705924034119,
      "learning_rate": 1.614472064642251e-05,
      "loss": 0.9282,
      "step": 1857
    },
    {
      "epoch": 0.31151629466624753,
      "grad_norm": 0.3734569549560547,
      "learning_rate": 1.61404353435904e-05,
      "loss": 0.8872,
      "step": 1858
    },
    {
      "epoch": 0.3116839568269936,
      "grad_norm": 0.516380786895752,
      "learning_rate": 1.613614822986787e-05,
      "loss": 0.9805,
      "step": 1859
    },
    {
      "epoch": 0.3118516189877397,
      "grad_norm": 0.47568896412849426,
      "learning_rate": 1.6131859306519243e-05,
      "loss": 0.8505,
      "step": 1860
    },
    {
      "epoch": 0.3120192811484858,
      "grad_norm": 0.41954758763313293,
      "learning_rate": 1.612756857480938e-05,
      "loss": 0.7861,
      "step": 1861
    },
    {
      "epoch": 0.3121869433092319,
      "grad_norm": 0.44616973400115967,
      "learning_rate": 1.6123276036003664e-05,
      "loss": 0.8088,
      "step": 1862
    },
    {
      "epoch": 0.312354605469978,
      "grad_norm": 0.5130486488342285,
      "learning_rate": 1.611898169136802e-05,
      "loss": 0.9097,
      "step": 1863
    },
    {
      "epoch": 0.3125222676307241,
      "grad_norm": 0.5425180196762085,
      "learning_rate": 1.6114685542168912e-05,
      "loss": 0.9192,
      "step": 1864
    },
    {
      "epoch": 0.3126899297914702,
      "grad_norm": 0.37737610936164856,
      "learning_rate": 1.6110387589673315e-05,
      "loss": 0.8592,
      "step": 1865
    },
    {
      "epoch": 0.3128575919522163,
      "grad_norm": 0.3993113040924072,
      "learning_rate": 1.610608783514876e-05,
      "loss": 0.9123,
      "step": 1866
    },
    {
      "epoch": 0.3130252541129624,
      "grad_norm": 0.42021462321281433,
      "learning_rate": 1.6101786279863285e-05,
      "loss": 0.9289,
      "step": 1867
    },
    {
      "epoch": 0.3131929162737085,
      "grad_norm": 0.6417856812477112,
      "learning_rate": 1.6097482925085482e-05,
      "loss": 0.9057,
      "step": 1868
    },
    {
      "epoch": 0.3133605784344546,
      "grad_norm": 0.5079712867736816,
      "learning_rate": 1.609317777208446e-05,
      "loss": 1.0042,
      "step": 1869
    },
    {
      "epoch": 0.3135282405952007,
      "grad_norm": 0.4228523075580597,
      "learning_rate": 1.6088870822129862e-05,
      "loss": 0.8974,
      "step": 1870
    },
    {
      "epoch": 0.3136959027559468,
      "grad_norm": 0.487529993057251,
      "learning_rate": 1.608456207649186e-05,
      "loss": 0.8846,
      "step": 1871
    },
    {
      "epoch": 0.31386356491669287,
      "grad_norm": 0.49924778938293457,
      "learning_rate": 1.608025153644116e-05,
      "loss": 0.9101,
      "step": 1872
    },
    {
      "epoch": 0.31403122707743897,
      "grad_norm": 0.43718042969703674,
      "learning_rate": 1.607593920324899e-05,
      "loss": 0.8513,
      "step": 1873
    },
    {
      "epoch": 0.31419888923818506,
      "grad_norm": 0.5099447965621948,
      "learning_rate": 1.6071625078187113e-05,
      "loss": 0.9156,
      "step": 1874
    },
    {
      "epoch": 0.31436655139893116,
      "grad_norm": 0.47883695363998413,
      "learning_rate": 1.6067309162527817e-05,
      "loss": 0.7846,
      "step": 1875
    },
    {
      "epoch": 0.31453421355967726,
      "grad_norm": 0.5323141813278198,
      "learning_rate": 1.606299145754392e-05,
      "loss": 0.9689,
      "step": 1876
    },
    {
      "epoch": 0.31470187572042335,
      "grad_norm": 0.5095372200012207,
      "learning_rate": 1.605867196450876e-05,
      "loss": 0.7851,
      "step": 1877
    },
    {
      "epoch": 0.31486953788116945,
      "grad_norm": 0.4553854763507843,
      "learning_rate": 1.6054350684696216e-05,
      "loss": 0.9793,
      "step": 1878
    },
    {
      "epoch": 0.31503720004191554,
      "grad_norm": 0.5465895533561707,
      "learning_rate": 1.6050027619380687e-05,
      "loss": 0.8846,
      "step": 1879
    },
    {
      "epoch": 0.31520486220266164,
      "grad_norm": 0.42045995593070984,
      "learning_rate": 1.60457027698371e-05,
      "loss": 0.9413,
      "step": 1880
    },
    {
      "epoch": 0.31537252436340774,
      "grad_norm": 0.6263823509216309,
      "learning_rate": 1.6041376137340893e-05,
      "loss": 0.8666,
      "step": 1881
    },
    {
      "epoch": 0.31554018652415383,
      "grad_norm": 0.43639567494392395,
      "learning_rate": 1.603704772316806e-05,
      "loss": 0.8834,
      "step": 1882
    },
    {
      "epoch": 0.3157078486848999,
      "grad_norm": 0.56414794921875,
      "learning_rate": 1.6032717528595098e-05,
      "loss": 0.9547,
      "step": 1883
    },
    {
      "epoch": 0.315875510845646,
      "grad_norm": 0.38132351636886597,
      "learning_rate": 1.602838555489904e-05,
      "loss": 0.9057,
      "step": 1884
    },
    {
      "epoch": 0.3160431730063921,
      "grad_norm": 0.3706132769584656,
      "learning_rate": 1.6024051803357425e-05,
      "loss": 0.8835,
      "step": 1885
    },
    {
      "epoch": 0.3162108351671382,
      "grad_norm": 0.4287024140357971,
      "learning_rate": 1.6019716275248342e-05,
      "loss": 0.9048,
      "step": 1886
    },
    {
      "epoch": 0.3163784973278843,
      "grad_norm": 0.6251978278160095,
      "learning_rate": 1.601537897185039e-05,
      "loss": 0.9559,
      "step": 1887
    },
    {
      "epoch": 0.3165461594886304,
      "grad_norm": 0.590323805809021,
      "learning_rate": 1.6011039894442685e-05,
      "loss": 0.9553,
      "step": 1888
    },
    {
      "epoch": 0.3167138216493765,
      "grad_norm": 0.5125306248664856,
      "learning_rate": 1.600669904430489e-05,
      "loss": 0.8657,
      "step": 1889
    },
    {
      "epoch": 0.3168814838101226,
      "grad_norm": 0.6732698678970337,
      "learning_rate": 1.6002356422717163e-05,
      "loss": 0.8711,
      "step": 1890
    },
    {
      "epoch": 0.3170491459708687,
      "grad_norm": 0.5461569428443909,
      "learning_rate": 1.59980120309602e-05,
      "loss": 0.8593,
      "step": 1891
    },
    {
      "epoch": 0.3172168081316148,
      "grad_norm": 0.5875232219696045,
      "learning_rate": 1.599366587031522e-05,
      "loss": 0.9037,
      "step": 1892
    },
    {
      "epoch": 0.3173844702923609,
      "grad_norm": 0.5603500008583069,
      "learning_rate": 1.5989317942063955e-05,
      "loss": 0.8561,
      "step": 1893
    },
    {
      "epoch": 0.317552132453107,
      "grad_norm": 0.4748394787311554,
      "learning_rate": 1.598496824748866e-05,
      "loss": 0.8718,
      "step": 1894
    },
    {
      "epoch": 0.3177197946138531,
      "grad_norm": 0.43374186754226685,
      "learning_rate": 1.5980616787872122e-05,
      "loss": 0.8226,
      "step": 1895
    },
    {
      "epoch": 0.3178874567745992,
      "grad_norm": 0.5378707647323608,
      "learning_rate": 1.597626356449764e-05,
      "loss": 0.7987,
      "step": 1896
    },
    {
      "epoch": 0.31805511893534527,
      "grad_norm": 0.6311928033828735,
      "learning_rate": 1.5971908578649024e-05,
      "loss": 0.9061,
      "step": 1897
    },
    {
      "epoch": 0.31822278109609137,
      "grad_norm": 0.5331396460533142,
      "learning_rate": 1.5967551831610624e-05,
      "loss": 0.9141,
      "step": 1898
    },
    {
      "epoch": 0.31839044325683746,
      "grad_norm": 0.3972858786582947,
      "learning_rate": 1.596319332466729e-05,
      "loss": 0.8557,
      "step": 1899
    },
    {
      "epoch": 0.31855810541758356,
      "grad_norm": 0.5714462995529175,
      "learning_rate": 1.5958833059104405e-05,
      "loss": 0.9814,
      "step": 1900
    },
    {
      "epoch": 0.31872576757832966,
      "grad_norm": 0.5758419036865234,
      "learning_rate": 1.5954471036207865e-05,
      "loss": 0.8925,
      "step": 1901
    },
    {
      "epoch": 0.31889342973907575,
      "grad_norm": 0.8500440716743469,
      "learning_rate": 1.5950107257264076e-05,
      "loss": 0.9505,
      "step": 1902
    },
    {
      "epoch": 0.31906109189982185,
      "grad_norm": 0.4728148877620697,
      "learning_rate": 1.5945741723559987e-05,
      "loss": 0.8705,
      "step": 1903
    },
    {
      "epoch": 0.31922875406056794,
      "grad_norm": 0.3619634211063385,
      "learning_rate": 1.5941374436383035e-05,
      "loss": 0.9196,
      "step": 1904
    },
    {
      "epoch": 0.31939641622131404,
      "grad_norm": 0.4979732930660248,
      "learning_rate": 1.593700539702119e-05,
      "loss": 0.8234,
      "step": 1905
    },
    {
      "epoch": 0.31956407838206013,
      "grad_norm": 0.5817590951919556,
      "learning_rate": 1.5932634606762935e-05,
      "loss": 0.9044,
      "step": 1906
    },
    {
      "epoch": 0.31973174054280623,
      "grad_norm": 0.516790509223938,
      "learning_rate": 1.5928262066897277e-05,
      "loss": 0.8855,
      "step": 1907
    },
    {
      "epoch": 0.3198994027035523,
      "grad_norm": 0.5449026226997375,
      "learning_rate": 1.5923887778713726e-05,
      "loss": 0.9708,
      "step": 1908
    },
    {
      "epoch": 0.3200670648642984,
      "grad_norm": 0.4417110085487366,
      "learning_rate": 1.591951174350231e-05,
      "loss": 0.957,
      "step": 1909
    },
    {
      "epoch": 0.3202347270250445,
      "grad_norm": 0.5064684748649597,
      "learning_rate": 1.5915133962553584e-05,
      "loss": 0.9166,
      "step": 1910
    },
    {
      "epoch": 0.3204023891857906,
      "grad_norm": 0.4732873737812042,
      "learning_rate": 1.5910754437158608e-05,
      "loss": 0.9244,
      "step": 1911
    },
    {
      "epoch": 0.3205700513465367,
      "grad_norm": 0.7517212629318237,
      "learning_rate": 1.5906373168608952e-05,
      "loss": 0.8813,
      "step": 1912
    },
    {
      "epoch": 0.3207377135072828,
      "grad_norm": 0.44778767228126526,
      "learning_rate": 1.5901990158196712e-05,
      "loss": 0.862,
      "step": 1913
    },
    {
      "epoch": 0.3209053756680289,
      "grad_norm": 0.470090389251709,
      "learning_rate": 1.5897605407214497e-05,
      "loss": 0.919,
      "step": 1914
    },
    {
      "epoch": 0.321073037828775,
      "grad_norm": 0.3986078202724457,
      "learning_rate": 1.589321891695541e-05,
      "loss": 0.9469,
      "step": 1915
    },
    {
      "epoch": 0.3212406999895211,
      "grad_norm": 0.5769183039665222,
      "learning_rate": 1.5888830688713096e-05,
      "loss": 0.915,
      "step": 1916
    },
    {
      "epoch": 0.3214083621502672,
      "grad_norm": 0.42942318320274353,
      "learning_rate": 1.5884440723781687e-05,
      "loss": 0.8832,
      "step": 1917
    },
    {
      "epoch": 0.3215760243110133,
      "grad_norm": 0.6780664920806885,
      "learning_rate": 1.5880049023455848e-05,
      "loss": 0.8358,
      "step": 1918
    },
    {
      "epoch": 0.3217436864717594,
      "grad_norm": 0.3817391097545624,
      "learning_rate": 1.5875655589030738e-05,
      "loss": 0.9046,
      "step": 1919
    },
    {
      "epoch": 0.3219113486325055,
      "grad_norm": 0.700987696647644,
      "learning_rate": 1.5871260421802034e-05,
      "loss": 0.8926,
      "step": 1920
    },
    {
      "epoch": 0.3220790107932516,
      "grad_norm": 0.5530411005020142,
      "learning_rate": 1.5866863523065933e-05,
      "loss": 0.7418,
      "step": 1921
    },
    {
      "epoch": 0.32224667295399767,
      "grad_norm": 0.4336833953857422,
      "learning_rate": 1.586246489411913e-05,
      "loss": 0.9115,
      "step": 1922
    },
    {
      "epoch": 0.32241433511474377,
      "grad_norm": 0.5864166021347046,
      "learning_rate": 1.5858064536258837e-05,
      "loss": 1.0381,
      "step": 1923
    },
    {
      "epoch": 0.32258199727548986,
      "grad_norm": 0.5918449759483337,
      "learning_rate": 1.5853662450782772e-05,
      "loss": 0.889,
      "step": 1924
    },
    {
      "epoch": 0.32274965943623596,
      "grad_norm": 0.43187668919563293,
      "learning_rate": 1.5849258638989166e-05,
      "loss": 0.8609,
      "step": 1925
    },
    {
      "epoch": 0.32291732159698205,
      "grad_norm": 0.5723666548728943,
      "learning_rate": 1.5844853102176757e-05,
      "loss": 0.934,
      "step": 1926
    },
    {
      "epoch": 0.32308498375772815,
      "grad_norm": 0.5808163285255432,
      "learning_rate": 1.5840445841644793e-05,
      "loss": 0.8731,
      "step": 1927
    },
    {
      "epoch": 0.3232526459184743,
      "grad_norm": 0.3876830041408539,
      "learning_rate": 1.5836036858693024e-05,
      "loss": 0.8578,
      "step": 1928
    },
    {
      "epoch": 0.3234203080792204,
      "grad_norm": 0.8593195080757141,
      "learning_rate": 1.583162615462172e-05,
      "loss": 0.9692,
      "step": 1929
    },
    {
      "epoch": 0.3235879702399665,
      "grad_norm": 0.5928516387939453,
      "learning_rate": 1.5827213730731656e-05,
      "loss": 0.9549,
      "step": 1930
    },
    {
      "epoch": 0.3237556324007126,
      "grad_norm": 0.8312473297119141,
      "learning_rate": 1.5822799588324102e-05,
      "loss": 0.8135,
      "step": 1931
    },
    {
      "epoch": 0.3239232945614587,
      "grad_norm": 0.5945383906364441,
      "learning_rate": 1.581838372870085e-05,
      "loss": 0.9046,
      "step": 1932
    },
    {
      "epoch": 0.3240909567222048,
      "grad_norm": 0.4078502953052521,
      "learning_rate": 1.5813966153164186e-05,
      "loss": 0.8857,
      "step": 1933
    },
    {
      "epoch": 0.3242586188829509,
      "grad_norm": 0.38547268509864807,
      "learning_rate": 1.580954686301691e-05,
      "loss": 0.8854,
      "step": 1934
    },
    {
      "epoch": 0.324426281043697,
      "grad_norm": 0.3948238492012024,
      "learning_rate": 1.5805125859562327e-05,
      "loss": 0.8447,
      "step": 1935
    },
    {
      "epoch": 0.32459394320444307,
      "grad_norm": 0.4084761142730713,
      "learning_rate": 1.5800703144104242e-05,
      "loss": 0.8287,
      "step": 1936
    },
    {
      "epoch": 0.32476160536518917,
      "grad_norm": 0.4811871349811554,
      "learning_rate": 1.5796278717946972e-05,
      "loss": 0.944,
      "step": 1937
    },
    {
      "epoch": 0.32492926752593526,
      "grad_norm": 0.3900909721851349,
      "learning_rate": 1.5791852582395334e-05,
      "loss": 0.8761,
      "step": 1938
    },
    {
      "epoch": 0.32509692968668136,
      "grad_norm": 0.432222455739975,
      "learning_rate": 1.5787424738754652e-05,
      "loss": 0.9091,
      "step": 1939
    },
    {
      "epoch": 0.32526459184742745,
      "grad_norm": 0.5341072082519531,
      "learning_rate": 1.5782995188330746e-05,
      "loss": 0.8926,
      "step": 1940
    },
    {
      "epoch": 0.32543225400817355,
      "grad_norm": 0.4855509400367737,
      "learning_rate": 1.5778563932429952e-05,
      "loss": 0.8781,
      "step": 1941
    },
    {
      "epoch": 0.32559991616891965,
      "grad_norm": 0.5989813804626465,
      "learning_rate": 1.57741309723591e-05,
      "loss": 0.8938,
      "step": 1942
    },
    {
      "epoch": 0.32576757832966574,
      "grad_norm": 0.39877378940582275,
      "learning_rate": 1.576969630942552e-05,
      "loss": 0.9231,
      "step": 1943
    },
    {
      "epoch": 0.32593524049041184,
      "grad_norm": 0.5871772766113281,
      "learning_rate": 1.5765259944937058e-05,
      "loss": 0.9229,
      "step": 1944
    },
    {
      "epoch": 0.32610290265115793,
      "grad_norm": 0.7102689743041992,
      "learning_rate": 1.5760821880202047e-05,
      "loss": 0.9258,
      "step": 1945
    },
    {
      "epoch": 0.32627056481190403,
      "grad_norm": 0.5745960474014282,
      "learning_rate": 1.575638211652933e-05,
      "loss": 0.999,
      "step": 1946
    },
    {
      "epoch": 0.3264382269726501,
      "grad_norm": 0.5958930850028992,
      "learning_rate": 1.5751940655228242e-05,
      "loss": 0.9169,
      "step": 1947
    },
    {
      "epoch": 0.3266058891333962,
      "grad_norm": 0.3930474817752838,
      "learning_rate": 1.574749749760863e-05,
      "loss": 0.9572,
      "step": 1948
    },
    {
      "epoch": 0.3267735512941423,
      "grad_norm": 0.9402393698692322,
      "learning_rate": 1.5743052644980838e-05,
      "loss": 1.0284,
      "step": 1949
    },
    {
      "epoch": 0.3269412134548884,
      "grad_norm": 0.7290217280387878,
      "learning_rate": 1.5738606098655708e-05,
      "loss": 0.9351,
      "step": 1950
    },
    {
      "epoch": 0.3271088756156345,
      "grad_norm": 0.7580875158309937,
      "learning_rate": 1.5734157859944574e-05,
      "loss": 0.9613,
      "step": 1951
    },
    {
      "epoch": 0.3272765377763806,
      "grad_norm": 0.5927433371543884,
      "learning_rate": 1.5729707930159282e-05,
      "loss": 0.8718,
      "step": 1952
    },
    {
      "epoch": 0.3274441999371267,
      "grad_norm": 0.5822391510009766,
      "learning_rate": 1.572525631061218e-05,
      "loss": 0.8529,
      "step": 1953
    },
    {
      "epoch": 0.3276118620978728,
      "grad_norm": 0.4330624043941498,
      "learning_rate": 1.572080300261609e-05,
      "loss": 0.8967,
      "step": 1954
    },
    {
      "epoch": 0.3277795242586189,
      "grad_norm": 0.522164523601532,
      "learning_rate": 1.5716348007484358e-05,
      "loss": 0.913,
      "step": 1955
    },
    {
      "epoch": 0.327947186419365,
      "grad_norm": 0.403029203414917,
      "learning_rate": 1.571189132653081e-05,
      "loss": 0.83,
      "step": 1956
    },
    {
      "epoch": 0.3281148485801111,
      "grad_norm": 0.9028471112251282,
      "learning_rate": 1.5707432961069785e-05,
      "loss": 0.9277,
      "step": 1957
    },
    {
      "epoch": 0.3282825107408572,
      "grad_norm": 0.42061564326286316,
      "learning_rate": 1.5702972912416103e-05,
      "loss": 0.829,
      "step": 1958
    },
    {
      "epoch": 0.3284501729016033,
      "grad_norm": 0.5323939323425293,
      "learning_rate": 1.569851118188509e-05,
      "loss": 0.8657,
      "step": 1959
    },
    {
      "epoch": 0.3286178350623494,
      "grad_norm": 0.4818108379840851,
      "learning_rate": 1.569404777079257e-05,
      "loss": 0.9152,
      "step": 1960
    },
    {
      "epoch": 0.32878549722309547,
      "grad_norm": 0.49384453892707825,
      "learning_rate": 1.568958268045485e-05,
      "loss": 0.9482,
      "step": 1961
    },
    {
      "epoch": 0.32895315938384156,
      "grad_norm": 0.6106820702552795,
      "learning_rate": 1.5685115912188745e-05,
      "loss": 0.8776,
      "step": 1962
    },
    {
      "epoch": 0.32912082154458766,
      "grad_norm": 0.6082288026809692,
      "learning_rate": 1.568064746731156e-05,
      "loss": 0.9401,
      "step": 1963
    },
    {
      "epoch": 0.32928848370533376,
      "grad_norm": 0.5787052512168884,
      "learning_rate": 1.5676177347141096e-05,
      "loss": 0.8468,
      "step": 1964
    },
    {
      "epoch": 0.32945614586607985,
      "grad_norm": 0.4497988820075989,
      "learning_rate": 1.5671705552995642e-05,
      "loss": 0.9482,
      "step": 1965
    },
    {
      "epoch": 0.32962380802682595,
      "grad_norm": 0.4412933588027954,
      "learning_rate": 1.5667232086193994e-05,
      "loss": 0.8962,
      "step": 1966
    },
    {
      "epoch": 0.32979147018757204,
      "grad_norm": 0.4735589921474457,
      "learning_rate": 1.566275694805542e-05,
      "loss": 0.8334,
      "step": 1967
    },
    {
      "epoch": 0.32995913234831814,
      "grad_norm": 0.44961315393447876,
      "learning_rate": 1.5658280139899706e-05,
      "loss": 0.9887,
      "step": 1968
    },
    {
      "epoch": 0.33012679450906424,
      "grad_norm": 0.4612094759941101,
      "learning_rate": 1.5653801663047113e-05,
      "loss": 0.9264,
      "step": 1969
    },
    {
      "epoch": 0.33029445666981033,
      "grad_norm": 0.4764600098133087,
      "learning_rate": 1.56493215188184e-05,
      "loss": 0.889,
      "step": 1970
    },
    {
      "epoch": 0.33046211883055643,
      "grad_norm": 0.4745563268661499,
      "learning_rate": 1.5644839708534817e-05,
      "loss": 0.924,
      "step": 1971
    },
    {
      "epoch": 0.3306297809913025,
      "grad_norm": 0.3943979740142822,
      "learning_rate": 1.5640356233518103e-05,
      "loss": 0.9431,
      "step": 1972
    },
    {
      "epoch": 0.3307974431520486,
      "grad_norm": 0.6608482599258423,
      "learning_rate": 1.5635871095090495e-05,
      "loss": 0.9135,
      "step": 1973
    },
    {
      "epoch": 0.3309651053127947,
      "grad_norm": 0.497452974319458,
      "learning_rate": 1.563138429457471e-05,
      "loss": 0.8359,
      "step": 1974
    },
    {
      "epoch": 0.3311327674735408,
      "grad_norm": 0.5150594115257263,
      "learning_rate": 1.5626895833293965e-05,
      "loss": 0.9055,
      "step": 1975
    },
    {
      "epoch": 0.3313004296342869,
      "grad_norm": 0.43873703479766846,
      "learning_rate": 1.562240571257196e-05,
      "loss": 0.8912,
      "step": 1976
    },
    {
      "epoch": 0.331468091795033,
      "grad_norm": 0.5945590138435364,
      "learning_rate": 1.5617913933732892e-05,
      "loss": 0.9009,
      "step": 1977
    },
    {
      "epoch": 0.3316357539557791,
      "grad_norm": 0.5829073190689087,
      "learning_rate": 1.5613420498101438e-05,
      "loss": 0.9412,
      "step": 1978
    },
    {
      "epoch": 0.3318034161165252,
      "grad_norm": 0.6509609818458557,
      "learning_rate": 1.5608925407002774e-05,
      "loss": 0.9201,
      "step": 1979
    },
    {
      "epoch": 0.3319710782772713,
      "grad_norm": 0.48291072249412537,
      "learning_rate": 1.5604428661762548e-05,
      "loss": 0.8516,
      "step": 1980
    },
    {
      "epoch": 0.3321387404380174,
      "grad_norm": 0.5338192582130432,
      "learning_rate": 1.559993026370691e-05,
      "loss": 0.9614,
      "step": 1981
    },
    {
      "epoch": 0.3323064025987635,
      "grad_norm": 0.4796494245529175,
      "learning_rate": 1.5595430214162498e-05,
      "loss": 0.845,
      "step": 1982
    },
    {
      "epoch": 0.3324740647595096,
      "grad_norm": 0.575904905796051,
      "learning_rate": 1.5590928514456428e-05,
      "loss": 0.8648,
      "step": 1983
    },
    {
      "epoch": 0.3326417269202557,
      "grad_norm": 0.42729178071022034,
      "learning_rate": 1.5586425165916308e-05,
      "loss": 0.8632,
      "step": 1984
    },
    {
      "epoch": 0.33280938908100177,
      "grad_norm": 0.3935372233390808,
      "learning_rate": 1.5581920169870227e-05,
      "loss": 0.9263,
      "step": 1985
    },
    {
      "epoch": 0.33297705124174787,
      "grad_norm": 0.49478182196617126,
      "learning_rate": 1.557741352764677e-05,
      "loss": 1.0177,
      "step": 1986
    },
    {
      "epoch": 0.33314471340249396,
      "grad_norm": 0.6155440807342529,
      "learning_rate": 1.5572905240575004e-05,
      "loss": 0.9465,
      "step": 1987
    },
    {
      "epoch": 0.33331237556324006,
      "grad_norm": 0.4374069571495056,
      "learning_rate": 1.556839530998447e-05,
      "loss": 0.9038,
      "step": 1988
    },
    {
      "epoch": 0.33348003772398616,
      "grad_norm": 0.5493287444114685,
      "learning_rate": 1.5563883737205207e-05,
      "loss": 0.8903,
      "step": 1989
    },
    {
      "epoch": 0.33364769988473225,
      "grad_norm": 0.5655654072761536,
      "learning_rate": 1.5559370523567734e-05,
      "loss": 0.8419,
      "step": 1990
    },
    {
      "epoch": 0.33381536204547835,
      "grad_norm": 0.7805976271629333,
      "learning_rate": 1.5554855670403057e-05,
      "loss": 0.8981,
      "step": 1991
    },
    {
      "epoch": 0.33398302420622444,
      "grad_norm": 0.6328807473182678,
      "learning_rate": 1.555033917904265e-05,
      "loss": 0.864,
      "step": 1992
    },
    {
      "epoch": 0.33415068636697054,
      "grad_norm": 0.4354403018951416,
      "learning_rate": 1.5545821050818494e-05,
      "loss": 0.8553,
      "step": 1993
    },
    {
      "epoch": 0.33431834852771664,
      "grad_norm": 0.5558031797409058,
      "learning_rate": 1.5541301287063036e-05,
      "loss": 0.9043,
      "step": 1994
    },
    {
      "epoch": 0.33448601068846273,
      "grad_norm": 0.4632706940174103,
      "learning_rate": 1.5536779889109213e-05,
      "loss": 0.8402,
      "step": 1995
    },
    {
      "epoch": 0.3346536728492088,
      "grad_norm": 0.4619845151901245,
      "learning_rate": 1.5532256858290442e-05,
      "loss": 0.8918,
      "step": 1996
    },
    {
      "epoch": 0.3348213350099549,
      "grad_norm": 0.6596323251724243,
      "learning_rate": 1.5527732195940616e-05,
      "loss": 0.8812,
      "step": 1997
    },
    {
      "epoch": 0.334988997170701,
      "grad_norm": 0.5146951675415039,
      "learning_rate": 1.5523205903394117e-05,
      "loss": 0.93,
      "step": 1998
    },
    {
      "epoch": 0.3351566593314471,
      "grad_norm": 0.4785274565219879,
      "learning_rate": 1.5518677981985802e-05,
      "loss": 0.8647,
      "step": 1999
    },
    {
      "epoch": 0.3353243214921932,
      "grad_norm": 0.4051639139652252,
      "learning_rate": 1.5514148433051018e-05,
      "loss": 0.8242,
      "step": 2000
    },
    {
      "epoch": 0.3354919836529393,
      "grad_norm": 0.5596480965614319,
      "learning_rate": 1.550961725792558e-05,
      "loss": 0.8847,
      "step": 2001
    },
    {
      "epoch": 0.3356596458136854,
      "grad_norm": 0.3232559561729431,
      "learning_rate": 1.5505084457945786e-05,
      "loss": 0.8139,
      "step": 2002
    },
    {
      "epoch": 0.3358273079744315,
      "grad_norm": 0.4186466336250305,
      "learning_rate": 1.5500550034448415e-05,
      "loss": 0.7907,
      "step": 2003
    },
    {
      "epoch": 0.3359949701351776,
      "grad_norm": 0.44500428438186646,
      "learning_rate": 1.549601398877073e-05,
      "loss": 0.9744,
      "step": 2004
    },
    {
      "epoch": 0.3361626322959237,
      "grad_norm": 0.5455708503723145,
      "learning_rate": 1.549147632225046e-05,
      "loss": 0.8561,
      "step": 2005
    },
    {
      "epoch": 0.3363302944566698,
      "grad_norm": 0.43856748938560486,
      "learning_rate": 1.5486937036225824e-05,
      "loss": 0.8707,
      "step": 2006
    },
    {
      "epoch": 0.3364979566174159,
      "grad_norm": 0.5387969017028809,
      "learning_rate": 1.5482396132035515e-05,
      "loss": 0.8936,
      "step": 2007
    },
    {
      "epoch": 0.336665618778162,
      "grad_norm": 0.4660685062408447,
      "learning_rate": 1.5477853611018697e-05,
      "loss": 0.8289,
      "step": 2008
    },
    {
      "epoch": 0.3368332809389081,
      "grad_norm": 0.6854202151298523,
      "learning_rate": 1.5473309474515014e-05,
      "loss": 0.9128,
      "step": 2009
    },
    {
      "epoch": 0.33700094309965417,
      "grad_norm": 0.6004125475883484,
      "learning_rate": 1.5468763723864598e-05,
      "loss": 0.9228,
      "step": 2010
    },
    {
      "epoch": 0.33716860526040027,
      "grad_norm": 0.5102965831756592,
      "learning_rate": 1.546421636040804e-05,
      "loss": 0.855,
      "step": 2011
    },
    {
      "epoch": 0.33733626742114636,
      "grad_norm": 0.38877010345458984,
      "learning_rate": 1.5459667385486416e-05,
      "loss": 0.9162,
      "step": 2012
    },
    {
      "epoch": 0.33750392958189246,
      "grad_norm": 0.5080621242523193,
      "learning_rate": 1.545511680044127e-05,
      "loss": 0.9592,
      "step": 2013
    },
    {
      "epoch": 0.3376715917426386,
      "grad_norm": 0.4342464208602905,
      "learning_rate": 1.5450564606614635e-05,
      "loss": 0.8897,
      "step": 2014
    },
    {
      "epoch": 0.3378392539033847,
      "grad_norm": 0.5745513439178467,
      "learning_rate": 1.5446010805348997e-05,
      "loss": 1.0136,
      "step": 2015
    },
    {
      "epoch": 0.3380069160641308,
      "grad_norm": 0.4189508259296417,
      "learning_rate": 1.5441455397987342e-05,
      "loss": 0.9253,
      "step": 2016
    },
    {
      "epoch": 0.3381745782248769,
      "grad_norm": 0.5819382071495056,
      "learning_rate": 1.5436898385873105e-05,
      "loss": 0.8154,
      "step": 2017
    },
    {
      "epoch": 0.338342240385623,
      "grad_norm": 0.4832051694393158,
      "learning_rate": 1.5432339770350212e-05,
      "loss": 0.8503,
      "step": 2018
    },
    {
      "epoch": 0.3385099025463691,
      "grad_norm": 0.47551408410072327,
      "learning_rate": 1.5427779552763053e-05,
      "loss": 0.8631,
      "step": 2019
    },
    {
      "epoch": 0.3386775647071152,
      "grad_norm": 0.5339730978012085,
      "learning_rate": 1.5423217734456493e-05,
      "loss": 0.9093,
      "step": 2020
    },
    {
      "epoch": 0.3388452268678613,
      "grad_norm": 0.496891051530838,
      "learning_rate": 1.5418654316775868e-05,
      "loss": 0.9656,
      "step": 2021
    },
    {
      "epoch": 0.3390128890286074,
      "grad_norm": 0.5256526470184326,
      "learning_rate": 1.541408930106699e-05,
      "loss": 0.8912,
      "step": 2022
    },
    {
      "epoch": 0.3391805511893535,
      "grad_norm": 0.3820645213127136,
      "learning_rate": 1.5409522688676132e-05,
      "loss": 0.8646,
      "step": 2023
    },
    {
      "epoch": 0.33934821335009957,
      "grad_norm": 0.6550490260124207,
      "learning_rate": 1.5404954480950047e-05,
      "loss": 0.8541,
      "step": 2024
    },
    {
      "epoch": 0.33951587551084567,
      "grad_norm": 0.41333290934562683,
      "learning_rate": 1.540038467923596e-05,
      "loss": 0.8607,
      "step": 2025
    },
    {
      "epoch": 0.33968353767159176,
      "grad_norm": 0.5534454584121704,
      "learning_rate": 1.539581328488156e-05,
      "loss": 0.82,
      "step": 2026
    },
    {
      "epoch": 0.33985119983233786,
      "grad_norm": 0.4226256012916565,
      "learning_rate": 1.5391240299235e-05,
      "loss": 0.8238,
      "step": 2027
    },
    {
      "epoch": 0.34001886199308395,
      "grad_norm": 0.47660452127456665,
      "learning_rate": 1.5386665723644923e-05,
      "loss": 0.8789,
      "step": 2028
    },
    {
      "epoch": 0.34018652415383005,
      "grad_norm": 0.6586236953735352,
      "learning_rate": 1.5382089559460423e-05,
      "loss": 0.9371,
      "step": 2029
    },
    {
      "epoch": 0.34035418631457615,
      "grad_norm": 0.386690229177475,
      "learning_rate": 1.5377511808031064e-05,
      "loss": 0.8813,
      "step": 2030
    },
    {
      "epoch": 0.34052184847532224,
      "grad_norm": 0.4806725084781647,
      "learning_rate": 1.5372932470706885e-05,
      "loss": 0.8278,
      "step": 2031
    },
    {
      "epoch": 0.34068951063606834,
      "grad_norm": 0.6102979183197021,
      "learning_rate": 1.5368351548838386e-05,
      "loss": 0.8384,
      "step": 2032
    },
    {
      "epoch": 0.34085717279681443,
      "grad_norm": 0.5258501172065735,
      "learning_rate": 1.5363769043776545e-05,
      "loss": 0.8905,
      "step": 2033
    },
    {
      "epoch": 0.34102483495756053,
      "grad_norm": 0.4061277508735657,
      "learning_rate": 1.5359184956872785e-05,
      "loss": 0.9178,
      "step": 2034
    },
    {
      "epoch": 0.3411924971183066,
      "grad_norm": 0.43954938650131226,
      "learning_rate": 1.535459928947903e-05,
      "loss": 0.8698,
      "step": 2035
    },
    {
      "epoch": 0.3413601592790527,
      "grad_norm": 0.4769081771373749,
      "learning_rate": 1.5350012042947633e-05,
      "loss": 0.9357,
      "step": 2036
    },
    {
      "epoch": 0.3415278214397988,
      "grad_norm": 0.7200191617012024,
      "learning_rate": 1.5345423218631438e-05,
      "loss": 0.8827,
      "step": 2037
    },
    {
      "epoch": 0.3416954836005449,
      "grad_norm": 0.4411308169364929,
      "learning_rate": 1.5340832817883743e-05,
      "loss": 0.8839,
      "step": 2038
    },
    {
      "epoch": 0.341863145761291,
      "grad_norm": 0.4716808497905731,
      "learning_rate": 1.533624084205832e-05,
      "loss": 0.8936,
      "step": 2039
    },
    {
      "epoch": 0.3420308079220371,
      "grad_norm": 0.38220396637916565,
      "learning_rate": 1.5331647292509394e-05,
      "loss": 0.9168,
      "step": 2040
    },
    {
      "epoch": 0.3421984700827832,
      "grad_norm": 0.3923315107822418,
      "learning_rate": 1.532705217059166e-05,
      "loss": 0.9659,
      "step": 2041
    },
    {
      "epoch": 0.3423661322435293,
      "grad_norm": 0.7197695970535278,
      "learning_rate": 1.5322455477660274e-05,
      "loss": 0.8764,
      "step": 2042
    },
    {
      "epoch": 0.3425337944042754,
      "grad_norm": 0.577632486820221,
      "learning_rate": 1.531785721507086e-05,
      "loss": 0.8478,
      "step": 2043
    },
    {
      "epoch": 0.3427014565650215,
      "grad_norm": 0.4102010428905487,
      "learning_rate": 1.5313257384179508e-05,
      "loss": 0.8483,
      "step": 2044
    },
    {
      "epoch": 0.3428691187257676,
      "grad_norm": 0.37063077092170715,
      "learning_rate": 1.5308655986342757e-05,
      "loss": 0.9228,
      "step": 2045
    },
    {
      "epoch": 0.3430367808865137,
      "grad_norm": 0.6842082142829895,
      "learning_rate": 1.530405302291762e-05,
      "loss": 0.8926,
      "step": 2046
    },
    {
      "epoch": 0.3432044430472598,
      "grad_norm": 0.47118207812309265,
      "learning_rate": 1.529944849526157e-05,
      "loss": 0.9021,
      "step": 2047
    },
    {
      "epoch": 0.3433721052080059,
      "grad_norm": 0.6092045307159424,
      "learning_rate": 1.5294842404732535e-05,
      "loss": 0.9341,
      "step": 2048
    },
    {
      "epoch": 0.34353976736875197,
      "grad_norm": 0.6156114339828491,
      "learning_rate": 1.529023475268891e-05,
      "loss": 0.8798,
      "step": 2049
    },
    {
      "epoch": 0.34370742952949807,
      "grad_norm": 0.5023700594902039,
      "learning_rate": 1.528562554048955e-05,
      "loss": 0.9569,
      "step": 2050
    },
    {
      "epoch": 0.34387509169024416,
      "grad_norm": 0.4121425449848175,
      "learning_rate": 1.5281014769493766e-05,
      "loss": 0.8448,
      "step": 2051
    },
    {
      "epoch": 0.34404275385099026,
      "grad_norm": 0.5774893760681152,
      "learning_rate": 1.527640244106133e-05,
      "loss": 0.8717,
      "step": 2052
    },
    {
      "epoch": 0.34421041601173635,
      "grad_norm": 0.5007169842720032,
      "learning_rate": 1.5271788556552477e-05,
      "loss": 0.9379,
      "step": 2053
    },
    {
      "epoch": 0.34437807817248245,
      "grad_norm": 0.49954840540885925,
      "learning_rate": 1.52671731173279e-05,
      "loss": 0.8518,
      "step": 2054
    },
    {
      "epoch": 0.34454574033322855,
      "grad_norm": 0.7455732822418213,
      "learning_rate": 1.5262556124748754e-05,
      "loss": 0.9908,
      "step": 2055
    },
    {
      "epoch": 0.34471340249397464,
      "grad_norm": 0.43593671917915344,
      "learning_rate": 1.5257937580176636e-05,
      "loss": 0.9346,
      "step": 2056
    },
    {
      "epoch": 0.34488106465472074,
      "grad_norm": 0.4693664014339447,
      "learning_rate": 1.525331748497362e-05,
      "loss": 0.9401,
      "step": 2057
    },
    {
      "epoch": 0.34504872681546683,
      "grad_norm": 0.45605477690696716,
      "learning_rate": 1.5248695840502223e-05,
      "loss": 0.8253,
      "step": 2058
    },
    {
      "epoch": 0.34521638897621293,
      "grad_norm": 1.0499744415283203,
      "learning_rate": 1.5244072648125432e-05,
      "loss": 0.9594,
      "step": 2059
    },
    {
      "epoch": 0.345384051136959,
      "grad_norm": 0.5815116167068481,
      "learning_rate": 1.5239447909206679e-05,
      "loss": 0.8302,
      "step": 2060
    },
    {
      "epoch": 0.3455517132977051,
      "grad_norm": 0.34735602140426636,
      "learning_rate": 1.5234821625109862e-05,
      "loss": 0.922,
      "step": 2061
    },
    {
      "epoch": 0.3457193754584512,
      "grad_norm": 1.402817964553833,
      "learning_rate": 1.5230193797199324e-05,
      "loss": 0.8915,
      "step": 2062
    },
    {
      "epoch": 0.3458870376191973,
      "grad_norm": 0.44540536403656006,
      "learning_rate": 1.5225564426839872e-05,
      "loss": 0.8578,
      "step": 2063
    },
    {
      "epoch": 0.3460546997799434,
      "grad_norm": 0.46437814831733704,
      "learning_rate": 1.522093351539676e-05,
      "loss": 0.8542,
      "step": 2064
    },
    {
      "epoch": 0.3462223619406895,
      "grad_norm": 0.39311614632606506,
      "learning_rate": 1.5216301064235706e-05,
      "loss": 0.7968,
      "step": 2065
    },
    {
      "epoch": 0.3463900241014356,
      "grad_norm": 0.8382595181465149,
      "learning_rate": 1.5211667074722875e-05,
      "loss": 1.0011,
      "step": 2066
    },
    {
      "epoch": 0.3465576862621817,
      "grad_norm": 0.5507627129554749,
      "learning_rate": 1.520703154822489e-05,
      "loss": 0.8895,
      "step": 2067
    },
    {
      "epoch": 0.3467253484229278,
      "grad_norm": 0.5316657423973083,
      "learning_rate": 1.5202394486108823e-05,
      "loss": 0.9189,
      "step": 2068
    },
    {
      "epoch": 0.3468930105836739,
      "grad_norm": 0.6288915276527405,
      "learning_rate": 1.5197755889742202e-05,
      "loss": 0.915,
      "step": 2069
    },
    {
      "epoch": 0.34706067274442,
      "grad_norm": 0.6252545714378357,
      "learning_rate": 1.5193115760493007e-05,
      "loss": 0.8947,
      "step": 2070
    },
    {
      "epoch": 0.3472283349051661,
      "grad_norm": 0.5686506628990173,
      "learning_rate": 1.5188474099729669e-05,
      "loss": 0.8418,
      "step": 2071
    },
    {
      "epoch": 0.3473959970659122,
      "grad_norm": 0.42737486958503723,
      "learning_rate": 1.5183830908821069e-05,
      "loss": 0.8932,
      "step": 2072
    },
    {
      "epoch": 0.3475636592266583,
      "grad_norm": 0.5029768943786621,
      "learning_rate": 1.5179186189136545e-05,
      "loss": 0.9382,
      "step": 2073
    },
    {
      "epoch": 0.34773132138740437,
      "grad_norm": 0.6150388717651367,
      "learning_rate": 1.517453994204588e-05,
      "loss": 0.8753,
      "step": 2074
    },
    {
      "epoch": 0.34789898354815046,
      "grad_norm": 0.40642431378364563,
      "learning_rate": 1.5169892168919312e-05,
      "loss": 0.9051,
      "step": 2075
    },
    {
      "epoch": 0.34806664570889656,
      "grad_norm": 0.5509570240974426,
      "learning_rate": 1.5165242871127523e-05,
      "loss": 0.8797,
      "step": 2076
    },
    {
      "epoch": 0.34823430786964266,
      "grad_norm": 0.5052769184112549,
      "learning_rate": 1.5160592050041654e-05,
      "loss": 0.91,
      "step": 2077
    },
    {
      "epoch": 0.34840197003038875,
      "grad_norm": 0.601817786693573,
      "learning_rate": 1.5155939707033287e-05,
      "loss": 0.9322,
      "step": 2078
    },
    {
      "epoch": 0.34856963219113485,
      "grad_norm": 0.5623035430908203,
      "learning_rate": 1.5151285843474454e-05,
      "loss": 0.8976,
      "step": 2079
    },
    {
      "epoch": 0.34873729435188094,
      "grad_norm": 0.5203773975372314,
      "learning_rate": 1.5146630460737638e-05,
      "loss": 0.8305,
      "step": 2080
    },
    {
      "epoch": 0.34890495651262704,
      "grad_norm": 0.5576394200325012,
      "learning_rate": 1.5141973560195768e-05,
      "loss": 0.9809,
      "step": 2081
    },
    {
      "epoch": 0.34907261867337314,
      "grad_norm": 0.4212713837623596,
      "learning_rate": 1.5137315143222227e-05,
      "loss": 0.8525,
      "step": 2082
    },
    {
      "epoch": 0.34924028083411923,
      "grad_norm": 0.7241677641868591,
      "learning_rate": 1.5132655211190832e-05,
      "loss": 0.8777,
      "step": 2083
    },
    {
      "epoch": 0.34940794299486533,
      "grad_norm": 0.4672906994819641,
      "learning_rate": 1.5127993765475859e-05,
      "loss": 0.894,
      "step": 2084
    },
    {
      "epoch": 0.3495756051556114,
      "grad_norm": 0.38957101106643677,
      "learning_rate": 1.512333080745203e-05,
      "loss": 0.9166,
      "step": 2085
    },
    {
      "epoch": 0.3497432673163575,
      "grad_norm": 0.4429856836795807,
      "learning_rate": 1.5118666338494501e-05,
      "loss": 0.9079,
      "step": 2086
    },
    {
      "epoch": 0.3499109294771036,
      "grad_norm": 0.7306303977966309,
      "learning_rate": 1.5114000359978884e-05,
      "loss": 0.9717,
      "step": 2087
    },
    {
      "epoch": 0.3500785916378497,
      "grad_norm": 0.45758989453315735,
      "learning_rate": 1.5109332873281237e-05,
      "loss": 0.8567,
      "step": 2088
    },
    {
      "epoch": 0.3502462537985958,
      "grad_norm": 0.6235557794570923,
      "learning_rate": 1.510466387977806e-05,
      "loss": 0.8931,
      "step": 2089
    },
    {
      "epoch": 0.3504139159593419,
      "grad_norm": 0.4747142195701599,
      "learning_rate": 1.5099993380846291e-05,
      "loss": 0.8268,
      "step": 2090
    },
    {
      "epoch": 0.350581578120088,
      "grad_norm": 0.505138635635376,
      "learning_rate": 1.509532137786332e-05,
      "loss": 0.8828,
      "step": 2091
    },
    {
      "epoch": 0.3507492402808341,
      "grad_norm": 0.3458777964115143,
      "learning_rate": 1.5090647872206984e-05,
      "loss": 0.8832,
      "step": 2092
    },
    {
      "epoch": 0.3509169024415802,
      "grad_norm": 0.46667036414146423,
      "learning_rate": 1.5085972865255556e-05,
      "loss": 0.9134,
      "step": 2093
    },
    {
      "epoch": 0.3510845646023263,
      "grad_norm": 0.4600054919719696,
      "learning_rate": 1.508129635838775e-05,
      "loss": 0.8623,
      "step": 2094
    },
    {
      "epoch": 0.3512522267630724,
      "grad_norm": 0.4440256953239441,
      "learning_rate": 1.5076618352982724e-05,
      "loss": 0.9679,
      "step": 2095
    },
    {
      "epoch": 0.3514198889238185,
      "grad_norm": 0.48019370436668396,
      "learning_rate": 1.5071938850420087e-05,
      "loss": 0.9159,
      "step": 2096
    },
    {
      "epoch": 0.3515875510845646,
      "grad_norm": 0.5663225650787354,
      "learning_rate": 1.5067257852079876e-05,
      "loss": 0.8541,
      "step": 2097
    },
    {
      "epoch": 0.35175521324531067,
      "grad_norm": 0.4951961040496826,
      "learning_rate": 1.5062575359342575e-05,
      "loss": 0.9002,
      "step": 2098
    },
    {
      "epoch": 0.3519228754060568,
      "grad_norm": 0.4354960024356842,
      "learning_rate": 1.5057891373589113e-05,
      "loss": 0.9126,
      "step": 2099
    },
    {
      "epoch": 0.3520905375668029,
      "grad_norm": 0.5936130285263062,
      "learning_rate": 1.5053205896200856e-05,
      "loss": 0.9094,
      "step": 2100
    },
    {
      "epoch": 0.352258199727549,
      "grad_norm": 0.6447715163230896,
      "learning_rate": 1.5048518928559608e-05,
      "loss": 0.9056,
      "step": 2101
    },
    {
      "epoch": 0.3524258618882951,
      "grad_norm": 0.5037466287612915,
      "learning_rate": 1.504383047204761e-05,
      "loss": 0.8379,
      "step": 2102
    },
    {
      "epoch": 0.3525935240490412,
      "grad_norm": 0.587268590927124,
      "learning_rate": 1.5039140528047548e-05,
      "loss": 0.8673,
      "step": 2103
    },
    {
      "epoch": 0.3527611862097873,
      "grad_norm": 1.0174998044967651,
      "learning_rate": 1.5034449097942549e-05,
      "loss": 0.9398,
      "step": 2104
    },
    {
      "epoch": 0.3529288483705334,
      "grad_norm": 0.7159402966499329,
      "learning_rate": 1.5029756183116171e-05,
      "loss": 0.8737,
      "step": 2105
    },
    {
      "epoch": 0.3530965105312795,
      "grad_norm": 0.46339261531829834,
      "learning_rate": 1.5025061784952408e-05,
      "loss": 0.8339,
      "step": 2106
    },
    {
      "epoch": 0.3532641726920256,
      "grad_norm": 0.810058057308197,
      "learning_rate": 1.50203659048357e-05,
      "loss": 0.8888,
      "step": 2107
    },
    {
      "epoch": 0.3534318348527717,
      "grad_norm": 0.5692547559738159,
      "learning_rate": 1.5015668544150925e-05,
      "loss": 0.9204,
      "step": 2108
    },
    {
      "epoch": 0.3535994970135178,
      "grad_norm": 0.4281153976917267,
      "learning_rate": 1.5010969704283387e-05,
      "loss": 0.7818,
      "step": 2109
    },
    {
      "epoch": 0.3537671591742639,
      "grad_norm": 0.4954259693622589,
      "learning_rate": 1.5006269386618833e-05,
      "loss": 0.9107,
      "step": 2110
    },
    {
      "epoch": 0.35393482133501,
      "grad_norm": 0.5465715527534485,
      "learning_rate": 1.5001567592543445e-05,
      "loss": 0.9045,
      "step": 2111
    },
    {
      "epoch": 0.35410248349575607,
      "grad_norm": 0.5770376920700073,
      "learning_rate": 1.499686432344384e-05,
      "loss": 0.8672,
      "step": 2112
    },
    {
      "epoch": 0.35427014565650217,
      "grad_norm": 0.583249568939209,
      "learning_rate": 1.499215958070707e-05,
      "loss": 0.9409,
      "step": 2113
    },
    {
      "epoch": 0.35443780781724826,
      "grad_norm": 0.5041799545288086,
      "learning_rate": 1.4987453365720624e-05,
      "loss": 0.8529,
      "step": 2114
    },
    {
      "epoch": 0.35460546997799436,
      "grad_norm": 0.4735868573188782,
      "learning_rate": 1.4982745679872421e-05,
      "loss": 0.9399,
      "step": 2115
    },
    {
      "epoch": 0.35477313213874045,
      "grad_norm": 0.4119265079498291,
      "learning_rate": 1.4978036524550816e-05,
      "loss": 0.8372,
      "step": 2116
    },
    {
      "epoch": 0.35494079429948655,
      "grad_norm": 0.4211297929286957,
      "learning_rate": 1.49733259011446e-05,
      "loss": 0.8324,
      "step": 2117
    },
    {
      "epoch": 0.35510845646023265,
      "grad_norm": 0.5091115236282349,
      "learning_rate": 1.496861381104299e-05,
      "loss": 0.8963,
      "step": 2118
    },
    {
      "epoch": 0.35527611862097874,
      "grad_norm": 0.5502423048019409,
      "learning_rate": 1.4963900255635644e-05,
      "loss": 0.8684,
      "step": 2119
    },
    {
      "epoch": 0.35544378078172484,
      "grad_norm": 0.5665629506111145,
      "learning_rate": 1.4959185236312642e-05,
      "loss": 0.9255,
      "step": 2120
    },
    {
      "epoch": 0.35561144294247093,
      "grad_norm": 0.5827038884162903,
      "learning_rate": 1.4954468754464509e-05,
      "loss": 0.8963,
      "step": 2121
    },
    {
      "epoch": 0.35577910510321703,
      "grad_norm": 0.47760000824928284,
      "learning_rate": 1.4949750811482187e-05,
      "loss": 0.9071,
      "step": 2122
    },
    {
      "epoch": 0.3559467672639631,
      "grad_norm": 0.5717841982841492,
      "learning_rate": 1.4945031408757058e-05,
      "loss": 0.864,
      "step": 2123
    },
    {
      "epoch": 0.3561144294247092,
      "grad_norm": 0.5802021622657776,
      "learning_rate": 1.4940310547680938e-05,
      "loss": 0.9182,
      "step": 2124
    },
    {
      "epoch": 0.3562820915854553,
      "grad_norm": 0.632815957069397,
      "learning_rate": 1.4935588229646057e-05,
      "loss": 0.8659,
      "step": 2125
    },
    {
      "epoch": 0.3564497537462014,
      "grad_norm": 1.1413359642028809,
      "learning_rate": 1.4930864456045096e-05,
      "loss": 0.9987,
      "step": 2126
    },
    {
      "epoch": 0.3566174159069475,
      "grad_norm": 0.5720139145851135,
      "learning_rate": 1.4926139228271143e-05,
      "loss": 0.8572,
      "step": 2127
    },
    {
      "epoch": 0.3567850780676936,
      "grad_norm": 0.473939448595047,
      "learning_rate": 1.492141254771774e-05,
      "loss": 0.9373,
      "step": 2128
    },
    {
      "epoch": 0.3569527402284397,
      "grad_norm": 0.6179938316345215,
      "learning_rate": 1.4916684415778828e-05,
      "loss": 0.9507,
      "step": 2129
    },
    {
      "epoch": 0.3571204023891858,
      "grad_norm": 0.48950061202049255,
      "learning_rate": 1.4911954833848805e-05,
      "loss": 0.9136,
      "step": 2130
    },
    {
      "epoch": 0.3572880645499319,
      "grad_norm": 0.46943745017051697,
      "learning_rate": 1.4907223803322474e-05,
      "loss": 0.781,
      "step": 2131
    },
    {
      "epoch": 0.357455726710678,
      "grad_norm": 0.3945772647857666,
      "learning_rate": 1.490249132559508e-05,
      "loss": 0.9011,
      "step": 2132
    },
    {
      "epoch": 0.3576233888714241,
      "grad_norm": 0.4948115944862366,
      "learning_rate": 1.4897757402062285e-05,
      "loss": 0.8314,
      "step": 2133
    },
    {
      "epoch": 0.3577910510321702,
      "grad_norm": 0.4736458957195282,
      "learning_rate": 1.4893022034120186e-05,
      "loss": 0.7921,
      "step": 2134
    },
    {
      "epoch": 0.3579587131929163,
      "grad_norm": 0.618366003036499,
      "learning_rate": 1.4888285223165299e-05,
      "loss": 0.8579,
      "step": 2135
    },
    {
      "epoch": 0.3581263753536624,
      "grad_norm": 0.7641869783401489,
      "learning_rate": 1.4883546970594571e-05,
      "loss": 0.9287,
      "step": 2136
    },
    {
      "epoch": 0.35829403751440847,
      "grad_norm": 0.4503091275691986,
      "learning_rate": 1.4878807277805366e-05,
      "loss": 0.923,
      "step": 2137
    },
    {
      "epoch": 0.35846169967515457,
      "grad_norm": 0.42575526237487793,
      "learning_rate": 1.4874066146195485e-05,
      "loss": 0.9257,
      "step": 2138
    },
    {
      "epoch": 0.35862936183590066,
      "grad_norm": 0.5589778423309326,
      "learning_rate": 1.4869323577163139e-05,
      "loss": 0.918,
      "step": 2139
    },
    {
      "epoch": 0.35879702399664676,
      "grad_norm": 0.5460204482078552,
      "learning_rate": 1.4864579572106977e-05,
      "loss": 0.9419,
      "step": 2140
    },
    {
      "epoch": 0.35896468615739285,
      "grad_norm": 0.4740411937236786,
      "learning_rate": 1.485983413242606e-05,
      "loss": 0.8739,
      "step": 2141
    },
    {
      "epoch": 0.35913234831813895,
      "grad_norm": 0.7383924722671509,
      "learning_rate": 1.4855087259519882e-05,
      "loss": 0.9669,
      "step": 2142
    },
    {
      "epoch": 0.35930001047888505,
      "grad_norm": 0.4387388825416565,
      "learning_rate": 1.485033895478835e-05,
      "loss": 0.91,
      "step": 2143
    },
    {
      "epoch": 0.35946767263963114,
      "grad_norm": 0.41963890194892883,
      "learning_rate": 1.48455892196318e-05,
      "loss": 0.8902,
      "step": 2144
    },
    {
      "epoch": 0.35963533480037724,
      "grad_norm": 0.4966847598552704,
      "learning_rate": 1.4840838055450986e-05,
      "loss": 0.9053,
      "step": 2145
    },
    {
      "epoch": 0.35980299696112333,
      "grad_norm": 0.9311870336532593,
      "learning_rate": 1.4836085463647088e-05,
      "loss": 0.8795,
      "step": 2146
    },
    {
      "epoch": 0.35997065912186943,
      "grad_norm": 0.6284542679786682,
      "learning_rate": 1.4831331445621699e-05,
      "loss": 0.9396,
      "step": 2147
    },
    {
      "epoch": 0.3601383212826155,
      "grad_norm": 0.39609295129776,
      "learning_rate": 1.4826576002776842e-05,
      "loss": 0.8391,
      "step": 2148
    },
    {
      "epoch": 0.3603059834433616,
      "grad_norm": 0.7315961718559265,
      "learning_rate": 1.4821819136514956e-05,
      "loss": 0.9071,
      "step": 2149
    },
    {
      "epoch": 0.3604736456041077,
      "grad_norm": 0.3803544342517853,
      "learning_rate": 1.4817060848238902e-05,
      "loss": 0.9259,
      "step": 2150
    },
    {
      "epoch": 0.3606413077648538,
      "grad_norm": 0.3933790922164917,
      "learning_rate": 1.4812301139351953e-05,
      "loss": 0.9043,
      "step": 2151
    },
    {
      "epoch": 0.3608089699255999,
      "grad_norm": 0.5101569294929504,
      "learning_rate": 1.4807540011257806e-05,
      "loss": 0.9055,
      "step": 2152
    },
    {
      "epoch": 0.360976632086346,
      "grad_norm": 0.4260362982749939,
      "learning_rate": 1.4802777465360581e-05,
      "loss": 0.8096,
      "step": 2153
    },
    {
      "epoch": 0.3611442942470921,
      "grad_norm": 0.4232743978500366,
      "learning_rate": 1.4798013503064809e-05,
      "loss": 1.0025,
      "step": 2154
    },
    {
      "epoch": 0.3613119564078382,
      "grad_norm": 0.6433045864105225,
      "learning_rate": 1.4793248125775442e-05,
      "loss": 0.8771,
      "step": 2155
    },
    {
      "epoch": 0.3614796185685843,
      "grad_norm": 0.42645955085754395,
      "learning_rate": 1.478848133489785e-05,
      "loss": 0.8817,
      "step": 2156
    },
    {
      "epoch": 0.3616472807293304,
      "grad_norm": 0.5128361582756042,
      "learning_rate": 1.4783713131837813e-05,
      "loss": 0.9481,
      "step": 2157
    },
    {
      "epoch": 0.3618149428900765,
      "grad_norm": 0.46490219235420227,
      "learning_rate": 1.4778943518001543e-05,
      "loss": 0.8478,
      "step": 2158
    },
    {
      "epoch": 0.3619826050508226,
      "grad_norm": 0.5935733318328857,
      "learning_rate": 1.4774172494795651e-05,
      "loss": 0.8164,
      "step": 2159
    },
    {
      "epoch": 0.3621502672115687,
      "grad_norm": 0.40863484144210815,
      "learning_rate": 1.476940006362717e-05,
      "loss": 0.8704,
      "step": 2160
    },
    {
      "epoch": 0.3623179293723148,
      "grad_norm": 1.119663119316101,
      "learning_rate": 1.4764626225903555e-05,
      "loss": 0.9581,
      "step": 2161
    },
    {
      "epoch": 0.36248559153306087,
      "grad_norm": 0.5529540181159973,
      "learning_rate": 1.4759850983032666e-05,
      "loss": 0.8493,
      "step": 2162
    },
    {
      "epoch": 0.36265325369380697,
      "grad_norm": 0.6024941205978394,
      "learning_rate": 1.4755074336422778e-05,
      "loss": 0.9546,
      "step": 2163
    },
    {
      "epoch": 0.36282091585455306,
      "grad_norm": 0.34549546241760254,
      "learning_rate": 1.475029628748259e-05,
      "loss": 0.8187,
      "step": 2164
    },
    {
      "epoch": 0.36298857801529916,
      "grad_norm": 0.39582517743110657,
      "learning_rate": 1.4745516837621207e-05,
      "loss": 0.9273,
      "step": 2165
    },
    {
      "epoch": 0.36315624017604525,
      "grad_norm": 0.47179707884788513,
      "learning_rate": 1.4740735988248144e-05,
      "loss": 0.9474,
      "step": 2166
    },
    {
      "epoch": 0.36332390233679135,
      "grad_norm": 0.4781091809272766,
      "learning_rate": 1.4735953740773334e-05,
      "loss": 0.9955,
      "step": 2167
    },
    {
      "epoch": 0.36349156449753744,
      "grad_norm": 0.3955364525318146,
      "learning_rate": 1.4731170096607121e-05,
      "loss": 0.8759,
      "step": 2168
    },
    {
      "epoch": 0.36365922665828354,
      "grad_norm": 0.5289798974990845,
      "learning_rate": 1.4726385057160264e-05,
      "loss": 0.9638,
      "step": 2169
    },
    {
      "epoch": 0.36382688881902964,
      "grad_norm": 0.417681485414505,
      "learning_rate": 1.4721598623843924e-05,
      "loss": 0.8737,
      "step": 2170
    },
    {
      "epoch": 0.36399455097977573,
      "grad_norm": 0.5590136051177979,
      "learning_rate": 1.4716810798069683e-05,
      "loss": 0.9051,
      "step": 2171
    },
    {
      "epoch": 0.36416221314052183,
      "grad_norm": 0.5966497659683228,
      "learning_rate": 1.4712021581249534e-05,
      "loss": 0.8726,
      "step": 2172
    },
    {
      "epoch": 0.3643298753012679,
      "grad_norm": 0.5650334358215332,
      "learning_rate": 1.4707230974795869e-05,
      "loss": 0.9522,
      "step": 2173
    },
    {
      "epoch": 0.364497537462014,
      "grad_norm": 0.46143072843551636,
      "learning_rate": 1.47024389801215e-05,
      "loss": 0.9797,
      "step": 2174
    },
    {
      "epoch": 0.3646651996227601,
      "grad_norm": 0.4127398133277893,
      "learning_rate": 1.4697645598639652e-05,
      "loss": 0.9056,
      "step": 2175
    },
    {
      "epoch": 0.3648328617835062,
      "grad_norm": 0.6077963709831238,
      "learning_rate": 1.4692850831763943e-05,
      "loss": 0.9559,
      "step": 2176
    },
    {
      "epoch": 0.3650005239442523,
      "grad_norm": 0.4098505973815918,
      "learning_rate": 1.4688054680908414e-05,
      "loss": 0.8225,
      "step": 2177
    },
    {
      "epoch": 0.3651681861049984,
      "grad_norm": 0.5953378677368164,
      "learning_rate": 1.4683257147487506e-05,
      "loss": 0.8915,
      "step": 2178
    },
    {
      "epoch": 0.3653358482657445,
      "grad_norm": 0.5840544104576111,
      "learning_rate": 1.4678458232916075e-05,
      "loss": 0.8355,
      "step": 2179
    },
    {
      "epoch": 0.3655035104264906,
      "grad_norm": 0.4457241892814636,
      "learning_rate": 1.467365793860938e-05,
      "loss": 0.8149,
      "step": 2180
    },
    {
      "epoch": 0.3656711725872367,
      "grad_norm": 0.5155458450317383,
      "learning_rate": 1.4668856265983085e-05,
      "loss": 0.8703,
      "step": 2181
    },
    {
      "epoch": 0.3658388347479828,
      "grad_norm": 0.5299484133720398,
      "learning_rate": 1.466405321645326e-05,
      "loss": 0.9161,
      "step": 2182
    },
    {
      "epoch": 0.3660064969087289,
      "grad_norm": 0.44516849517822266,
      "learning_rate": 1.4659248791436393e-05,
      "loss": 0.9276,
      "step": 2183
    },
    {
      "epoch": 0.366174159069475,
      "grad_norm": 0.45529067516326904,
      "learning_rate": 1.465444299234936e-05,
      "loss": 0.8972,
      "step": 2184
    },
    {
      "epoch": 0.36634182123022113,
      "grad_norm": 0.4852927625179291,
      "learning_rate": 1.4649635820609457e-05,
      "loss": 0.8844,
      "step": 2185
    },
    {
      "epoch": 0.36650948339096723,
      "grad_norm": 0.5148529410362244,
      "learning_rate": 1.4644827277634367e-05,
      "loss": 0.9069,
      "step": 2186
    },
    {
      "epoch": 0.3666771455517133,
      "grad_norm": 0.40305307507514954,
      "learning_rate": 1.4640017364842205e-05,
      "loss": 0.8885,
      "step": 2187
    },
    {
      "epoch": 0.3668448077124594,
      "grad_norm": 0.4781140685081482,
      "learning_rate": 1.4635206083651461e-05,
      "loss": 0.8817,
      "step": 2188
    },
    {
      "epoch": 0.3670124698732055,
      "grad_norm": 0.5818206071853638,
      "learning_rate": 1.4630393435481047e-05,
      "loss": 0.8572,
      "step": 2189
    },
    {
      "epoch": 0.3671801320339516,
      "grad_norm": 0.4082125425338745,
      "learning_rate": 1.4625579421750271e-05,
      "loss": 0.8723,
      "step": 2190
    },
    {
      "epoch": 0.3673477941946977,
      "grad_norm": 0.4111286997795105,
      "learning_rate": 1.4620764043878845e-05,
      "loss": 0.8568,
      "step": 2191
    },
    {
      "epoch": 0.3675154563554438,
      "grad_norm": 0.5303593873977661,
      "learning_rate": 1.4615947303286888e-05,
      "loss": 0.8298,
      "step": 2192
    },
    {
      "epoch": 0.3676831185161899,
      "grad_norm": 0.4056621491909027,
      "learning_rate": 1.4611129201394911e-05,
      "loss": 0.9032,
      "step": 2193
    },
    {
      "epoch": 0.367850780676936,
      "grad_norm": 0.4418686628341675,
      "learning_rate": 1.4606309739623834e-05,
      "loss": 0.8864,
      "step": 2194
    },
    {
      "epoch": 0.3680184428376821,
      "grad_norm": 0.4276810586452484,
      "learning_rate": 1.4601488919394975e-05,
      "loss": 0.9574,
      "step": 2195
    },
    {
      "epoch": 0.3681861049984282,
      "grad_norm": 0.8117255568504333,
      "learning_rate": 1.4596666742130055e-05,
      "loss": 0.749,
      "step": 2196
    },
    {
      "epoch": 0.3683537671591743,
      "grad_norm": 1.4907840490341187,
      "learning_rate": 1.4591843209251191e-05,
      "loss": 0.8621,
      "step": 2197
    },
    {
      "epoch": 0.3685214293199204,
      "grad_norm": 0.5102953910827637,
      "learning_rate": 1.4587018322180906e-05,
      "loss": 0.9216,
      "step": 2198
    },
    {
      "epoch": 0.3686890914806665,
      "grad_norm": 1.0061148405075073,
      "learning_rate": 1.458219208234212e-05,
      "loss": 0.9426,
      "step": 2199
    },
    {
      "epoch": 0.36885675364141257,
      "grad_norm": 0.6368162631988525,
      "learning_rate": 1.4577364491158147e-05,
      "loss": 0.9052,
      "step": 2200
    },
    {
      "epoch": 0.36902441580215867,
      "grad_norm": 0.5865437984466553,
      "learning_rate": 1.4572535550052702e-05,
      "loss": 0.9249,
      "step": 2201
    },
    {
      "epoch": 0.36919207796290476,
      "grad_norm": 0.3822292387485504,
      "learning_rate": 1.4567705260449905e-05,
      "loss": 0.8256,
      "step": 2202
    },
    {
      "epoch": 0.36935974012365086,
      "grad_norm": 0.6831461787223816,
      "learning_rate": 1.4562873623774267e-05,
      "loss": 0.851,
      "step": 2203
    },
    {
      "epoch": 0.36952740228439696,
      "grad_norm": 0.6423549652099609,
      "learning_rate": 1.4558040641450695e-05,
      "loss": 0.8969,
      "step": 2204
    },
    {
      "epoch": 0.36969506444514305,
      "grad_norm": 0.583774745464325,
      "learning_rate": 1.4553206314904493e-05,
      "loss": 0.9309,
      "step": 2205
    },
    {
      "epoch": 0.36986272660588915,
      "grad_norm": 0.6059913635253906,
      "learning_rate": 1.4548370645561372e-05,
      "loss": 0.9703,
      "step": 2206
    },
    {
      "epoch": 0.37003038876663524,
      "grad_norm": 0.5601528882980347,
      "learning_rate": 1.4543533634847422e-05,
      "loss": 0.9313,
      "step": 2207
    },
    {
      "epoch": 0.37019805092738134,
      "grad_norm": 0.39669668674468994,
      "learning_rate": 1.4538695284189145e-05,
      "loss": 0.9154,
      "step": 2208
    },
    {
      "epoch": 0.37036571308812744,
      "grad_norm": 0.5013231635093689,
      "learning_rate": 1.4533855595013424e-05,
      "loss": 0.8856,
      "step": 2209
    },
    {
      "epoch": 0.37053337524887353,
      "grad_norm": 0.6057686805725098,
      "learning_rate": 1.4529014568747551e-05,
      "loss": 0.9103,
      "step": 2210
    },
    {
      "epoch": 0.3707010374096196,
      "grad_norm": 0.5593185424804688,
      "learning_rate": 1.4524172206819195e-05,
      "loss": 0.901,
      "step": 2211
    },
    {
      "epoch": 0.3708686995703657,
      "grad_norm": 0.5884758234024048,
      "learning_rate": 1.4519328510656435e-05,
      "loss": 0.9134,
      "step": 2212
    },
    {
      "epoch": 0.3710363617311118,
      "grad_norm": 0.5830038189888,
      "learning_rate": 1.4514483481687733e-05,
      "loss": 0.9205,
      "step": 2213
    },
    {
      "epoch": 0.3712040238918579,
      "grad_norm": 0.5845561027526855,
      "learning_rate": 1.450963712134195e-05,
      "loss": 0.9248,
      "step": 2214
    },
    {
      "epoch": 0.371371686052604,
      "grad_norm": 0.4929487705230713,
      "learning_rate": 1.4504789431048343e-05,
      "loss": 0.8335,
      "step": 2215
    },
    {
      "epoch": 0.3715393482133501,
      "grad_norm": 0.4876137375831604,
      "learning_rate": 1.4499940412236549e-05,
      "loss": 0.8844,
      "step": 2216
    },
    {
      "epoch": 0.3717070103740962,
      "grad_norm": 0.4443703293800354,
      "learning_rate": 1.4495090066336604e-05,
      "loss": 0.9326,
      "step": 2217
    },
    {
      "epoch": 0.3718746725348423,
      "grad_norm": 0.5985068082809448,
      "learning_rate": 1.4490238394778936e-05,
      "loss": 0.9112,
      "step": 2218
    },
    {
      "epoch": 0.3720423346955884,
      "grad_norm": 0.5642960667610168,
      "learning_rate": 1.4485385398994364e-05,
      "loss": 0.9219,
      "step": 2219
    },
    {
      "epoch": 0.3722099968563345,
      "grad_norm": 0.506014883518219,
      "learning_rate": 1.4480531080414096e-05,
      "loss": 0.9068,
      "step": 2220
    },
    {
      "epoch": 0.3723776590170806,
      "grad_norm": 0.5465354919433594,
      "learning_rate": 1.4475675440469729e-05,
      "loss": 0.8949,
      "step": 2221
    },
    {
      "epoch": 0.3725453211778267,
      "grad_norm": 0.4898136854171753,
      "learning_rate": 1.4470818480593258e-05,
      "loss": 0.9157,
      "step": 2222
    },
    {
      "epoch": 0.3727129833385728,
      "grad_norm": 0.4084881544113159,
      "learning_rate": 1.4465960202217053e-05,
      "loss": 0.8657,
      "step": 2223
    },
    {
      "epoch": 0.3728806454993189,
      "grad_norm": 0.5694803595542908,
      "learning_rate": 1.4461100606773884e-05,
      "loss": 0.8568,
      "step": 2224
    },
    {
      "epoch": 0.37304830766006497,
      "grad_norm": 0.41786134243011475,
      "learning_rate": 1.4456239695696903e-05,
      "loss": 0.96,
      "step": 2225
    },
    {
      "epoch": 0.37321596982081107,
      "grad_norm": 0.49354809522628784,
      "learning_rate": 1.4451377470419658e-05,
      "loss": 0.9287,
      "step": 2226
    },
    {
      "epoch": 0.37338363198155716,
      "grad_norm": 0.41685160994529724,
      "learning_rate": 1.4446513932376076e-05,
      "loss": 0.9182,
      "step": 2227
    },
    {
      "epoch": 0.37355129414230326,
      "grad_norm": 0.3885839283466339,
      "learning_rate": 1.444164908300047e-05,
      "loss": 0.8644,
      "step": 2228
    },
    {
      "epoch": 0.37371895630304935,
      "grad_norm": 0.49852511286735535,
      "learning_rate": 1.4436782923727557e-05,
      "loss": 0.9,
      "step": 2229
    },
    {
      "epoch": 0.37388661846379545,
      "grad_norm": 0.5106147527694702,
      "learning_rate": 1.4431915455992416e-05,
      "loss": 0.9022,
      "step": 2230
    },
    {
      "epoch": 0.37405428062454155,
      "grad_norm": 0.5619063973426819,
      "learning_rate": 1.4427046681230525e-05,
      "loss": 0.9078,
      "step": 2231
    },
    {
      "epoch": 0.37422194278528764,
      "grad_norm": 0.4425818622112274,
      "learning_rate": 1.4422176600877751e-05,
      "loss": 0.7656,
      "step": 2232
    },
    {
      "epoch": 0.37438960494603374,
      "grad_norm": 0.5203561186790466,
      "learning_rate": 1.441730521637034e-05,
      "loss": 0.984,
      "step": 2233
    },
    {
      "epoch": 0.37455726710677983,
      "grad_norm": 0.41096168756484985,
      "learning_rate": 1.441243252914492e-05,
      "loss": 0.9238,
      "step": 2234
    },
    {
      "epoch": 0.37472492926752593,
      "grad_norm": 0.47181493043899536,
      "learning_rate": 1.4407558540638509e-05,
      "loss": 0.8651,
      "step": 2235
    },
    {
      "epoch": 0.374892591428272,
      "grad_norm": 0.6022822856903076,
      "learning_rate": 1.4402683252288508e-05,
      "loss": 0.9019,
      "step": 2236
    },
    {
      "epoch": 0.3750602535890181,
      "grad_norm": 0.5729867815971375,
      "learning_rate": 1.4397806665532693e-05,
      "loss": 0.8741,
      "step": 2237
    },
    {
      "epoch": 0.3752279157497642,
      "grad_norm": 0.5488620400428772,
      "learning_rate": 1.4392928781809243e-05,
      "loss": 0.9981,
      "step": 2238
    },
    {
      "epoch": 0.3753955779105103,
      "grad_norm": 0.5091243982315063,
      "learning_rate": 1.4388049602556692e-05,
      "loss": 0.8706,
      "step": 2239
    },
    {
      "epoch": 0.3755632400712564,
      "grad_norm": 0.5202438235282898,
      "learning_rate": 1.4383169129213982e-05,
      "loss": 0.8039,
      "step": 2240
    },
    {
      "epoch": 0.3757309022320025,
      "grad_norm": 0.474795401096344,
      "learning_rate": 1.4378287363220415e-05,
      "loss": 0.902,
      "step": 2241
    },
    {
      "epoch": 0.3758985643927486,
      "grad_norm": 0.43491849303245544,
      "learning_rate": 1.4373404306015695e-05,
      "loss": 0.9261,
      "step": 2242
    },
    {
      "epoch": 0.3760662265534947,
      "grad_norm": 0.6873254776000977,
      "learning_rate": 1.4368519959039882e-05,
      "loss": 0.8585,
      "step": 2243
    },
    {
      "epoch": 0.3762338887142408,
      "grad_norm": 0.4213145971298218,
      "learning_rate": 1.4363634323733442e-05,
      "loss": 0.8277,
      "step": 2244
    },
    {
      "epoch": 0.3764015508749869,
      "grad_norm": 0.7943501472473145,
      "learning_rate": 1.435874740153721e-05,
      "loss": 0.9383,
      "step": 2245
    },
    {
      "epoch": 0.376569213035733,
      "grad_norm": 1.0533539056777954,
      "learning_rate": 1.4353859193892392e-05,
      "loss": 0.8199,
      "step": 2246
    },
    {
      "epoch": 0.3767368751964791,
      "grad_norm": 0.4386700391769409,
      "learning_rate": 1.4348969702240583e-05,
      "loss": 0.8748,
      "step": 2247
    },
    {
      "epoch": 0.3769045373572252,
      "grad_norm": 0.5937883257865906,
      "learning_rate": 1.434407892802376e-05,
      "loss": 0.9957,
      "step": 2248
    },
    {
      "epoch": 0.3770721995179713,
      "grad_norm": 0.6216612458229065,
      "learning_rate": 1.4339186872684268e-05,
      "loss": 0.8387,
      "step": 2249
    },
    {
      "epoch": 0.37723986167871737,
      "grad_norm": 0.6769209504127502,
      "learning_rate": 1.4334293537664836e-05,
      "loss": 0.821,
      "step": 2250
    },
    {
      "epoch": 0.37740752383946347,
      "grad_norm": 0.645228922367096,
      "learning_rate": 1.4329398924408567e-05,
      "loss": 1.0086,
      "step": 2251
    },
    {
      "epoch": 0.37757518600020956,
      "grad_norm": 0.40301021933555603,
      "learning_rate": 1.4324503034358946e-05,
      "loss": 0.8927,
      "step": 2252
    },
    {
      "epoch": 0.37774284816095566,
      "grad_norm": 0.5435569882392883,
      "learning_rate": 1.4319605868959831e-05,
      "loss": 0.8832,
      "step": 2253
    },
    {
      "epoch": 0.37791051032170175,
      "grad_norm": 0.7265287637710571,
      "learning_rate": 1.4314707429655456e-05,
      "loss": 0.891,
      "step": 2254
    },
    {
      "epoch": 0.37807817248244785,
      "grad_norm": 0.4399094879627228,
      "learning_rate": 1.4309807717890432e-05,
      "loss": 0.9177,
      "step": 2255
    },
    {
      "epoch": 0.37824583464319395,
      "grad_norm": 0.3403279185295105,
      "learning_rate": 1.4304906735109746e-05,
      "loss": 0.7757,
      "step": 2256
    },
    {
      "epoch": 0.37841349680394004,
      "grad_norm": 0.45156174898147583,
      "learning_rate": 1.4300004482758755e-05,
      "loss": 0.8753,
      "step": 2257
    },
    {
      "epoch": 0.37858115896468614,
      "grad_norm": 0.780342698097229,
      "learning_rate": 1.4295100962283197e-05,
      "loss": 0.9102,
      "step": 2258
    },
    {
      "epoch": 0.37874882112543223,
      "grad_norm": 0.684647262096405,
      "learning_rate": 1.4290196175129181e-05,
      "loss": 0.8894,
      "step": 2259
    },
    {
      "epoch": 0.37891648328617833,
      "grad_norm": 0.5325799584388733,
      "learning_rate": 1.4285290122743186e-05,
      "loss": 0.9213,
      "step": 2260
    },
    {
      "epoch": 0.3790841454469244,
      "grad_norm": 0.5337144136428833,
      "learning_rate": 1.4280382806572071e-05,
      "loss": 0.909,
      "step": 2261
    },
    {
      "epoch": 0.3792518076076705,
      "grad_norm": 0.4459291398525238,
      "learning_rate": 1.427547422806306e-05,
      "loss": 0.9284,
      "step": 2262
    },
    {
      "epoch": 0.3794194697684166,
      "grad_norm": 0.6480666995048523,
      "learning_rate": 1.4270564388663761e-05,
      "loss": 0.8756,
      "step": 2263
    },
    {
      "epoch": 0.3795871319291627,
      "grad_norm": 0.5705782771110535,
      "learning_rate": 1.4265653289822141e-05,
      "loss": 0.8864,
      "step": 2264
    },
    {
      "epoch": 0.3797547940899088,
      "grad_norm": 0.45472240447998047,
      "learning_rate": 1.4260740932986546e-05,
      "loss": 0.8901,
      "step": 2265
    },
    {
      "epoch": 0.3799224562506549,
      "grad_norm": 0.5093509554862976,
      "learning_rate": 1.4255827319605686e-05,
      "loss": 0.892,
      "step": 2266
    },
    {
      "epoch": 0.380090118411401,
      "grad_norm": 0.522854208946228,
      "learning_rate": 1.4250912451128654e-05,
      "loss": 1.004,
      "step": 2267
    },
    {
      "epoch": 0.3802577805721471,
      "grad_norm": 0.3982647955417633,
      "learning_rate": 1.4245996329004901e-05,
      "loss": 0.862,
      "step": 2268
    },
    {
      "epoch": 0.3804254427328932,
      "grad_norm": 0.48918038606643677,
      "learning_rate": 1.4241078954684253e-05,
      "loss": 0.8579,
      "step": 2269
    },
    {
      "epoch": 0.3805931048936393,
      "grad_norm": 0.3785038888454437,
      "learning_rate": 1.4236160329616903e-05,
      "loss": 0.8844,
      "step": 2270
    },
    {
      "epoch": 0.38076076705438544,
      "grad_norm": 0.5706347227096558,
      "learning_rate": 1.4231240455253419e-05,
      "loss": 0.9279,
      "step": 2271
    },
    {
      "epoch": 0.38092842921513154,
      "grad_norm": 0.6968598365783691,
      "learning_rate": 1.4226319333044729e-05,
      "loss": 0.8614,
      "step": 2272
    },
    {
      "epoch": 0.38109609137587763,
      "grad_norm": 0.5402582287788391,
      "learning_rate": 1.4221396964442135e-05,
      "loss": 0.9304,
      "step": 2273
    },
    {
      "epoch": 0.38126375353662373,
      "grad_norm": 0.4895389974117279,
      "learning_rate": 1.4216473350897301e-05,
      "loss": 0.9808,
      "step": 2274
    },
    {
      "epoch": 0.3814314156973698,
      "grad_norm": 0.35909193754196167,
      "learning_rate": 1.4211548493862266e-05,
      "loss": 0.8544,
      "step": 2275
    },
    {
      "epoch": 0.3815990778581159,
      "grad_norm": 0.5294655561447144,
      "learning_rate": 1.4206622394789432e-05,
      "loss": 0.8938,
      "step": 2276
    },
    {
      "epoch": 0.381766740018862,
      "grad_norm": 0.4982142746448517,
      "learning_rate": 1.4201695055131556e-05,
      "loss": 0.8515,
      "step": 2277
    },
    {
      "epoch": 0.3819344021796081,
      "grad_norm": 0.47081637382507324,
      "learning_rate": 1.4196766476341783e-05,
      "loss": 0.8818,
      "step": 2278
    },
    {
      "epoch": 0.3821020643403542,
      "grad_norm": 0.4863807260990143,
      "learning_rate": 1.4191836659873613e-05,
      "loss": 0.8882,
      "step": 2279
    },
    {
      "epoch": 0.3822697265011003,
      "grad_norm": 0.5890251994132996,
      "learning_rate": 1.4186905607180901e-05,
      "loss": 0.885,
      "step": 2280
    },
    {
      "epoch": 0.3824373886618464,
      "grad_norm": 0.6916296482086182,
      "learning_rate": 1.4181973319717882e-05,
      "loss": 0.8967,
      "step": 2281
    },
    {
      "epoch": 0.3826050508225925,
      "grad_norm": 0.6745121479034424,
      "learning_rate": 1.4177039798939143e-05,
      "loss": 0.9705,
      "step": 2282
    },
    {
      "epoch": 0.3827727129833386,
      "grad_norm": 0.6650843620300293,
      "learning_rate": 1.4172105046299647e-05,
      "loss": 0.8422,
      "step": 2283
    },
    {
      "epoch": 0.3829403751440847,
      "grad_norm": 0.5521268248558044,
      "learning_rate": 1.4167169063254708e-05,
      "loss": 0.8828,
      "step": 2284
    },
    {
      "epoch": 0.3831080373048308,
      "grad_norm": 0.6419956088066101,
      "learning_rate": 1.416223185126001e-05,
      "loss": 0.8358,
      "step": 2285
    },
    {
      "epoch": 0.3832756994655769,
      "grad_norm": 0.535331130027771,
      "learning_rate": 1.4157293411771603e-05,
      "loss": 0.9164,
      "step": 2286
    },
    {
      "epoch": 0.383443361626323,
      "grad_norm": 0.5716285109519958,
      "learning_rate": 1.4152353746245888e-05,
      "loss": 0.9543,
      "step": 2287
    },
    {
      "epoch": 0.38361102378706907,
      "grad_norm": 0.3962705731391907,
      "learning_rate": 1.4147412856139638e-05,
      "loss": 0.9082,
      "step": 2288
    },
    {
      "epoch": 0.38377868594781517,
      "grad_norm": 0.4236718416213989,
      "learning_rate": 1.4142470742909976e-05,
      "loss": 0.8863,
      "step": 2289
    },
    {
      "epoch": 0.38394634810856126,
      "grad_norm": 0.5821020603179932,
      "learning_rate": 1.4137527408014397e-05,
      "loss": 0.9455,
      "step": 2290
    },
    {
      "epoch": 0.38411401026930736,
      "grad_norm": 0.4136497378349304,
      "learning_rate": 1.4132582852910753e-05,
      "loss": 0.8888,
      "step": 2291
    },
    {
      "epoch": 0.38428167243005346,
      "grad_norm": 0.47119471430778503,
      "learning_rate": 1.412763707905725e-05,
      "loss": 0.8335,
      "step": 2292
    },
    {
      "epoch": 0.38444933459079955,
      "grad_norm": 0.421060174703598,
      "learning_rate": 1.4122690087912462e-05,
      "loss": 0.9604,
      "step": 2293
    },
    {
      "epoch": 0.38461699675154565,
      "grad_norm": 0.4163464903831482,
      "learning_rate": 1.4117741880935318e-05,
      "loss": 0.908,
      "step": 2294
    },
    {
      "epoch": 0.38478465891229174,
      "grad_norm": 0.4435379207134247,
      "learning_rate": 1.4112792459585105e-05,
      "loss": 0.9262,
      "step": 2295
    },
    {
      "epoch": 0.38495232107303784,
      "grad_norm": 0.8743917346000671,
      "learning_rate": 1.4107841825321465e-05,
      "loss": 0.8946,
      "step": 2296
    },
    {
      "epoch": 0.38511998323378394,
      "grad_norm": 0.9882781505584717,
      "learning_rate": 1.4102889979604403e-05,
      "loss": 0.9052,
      "step": 2297
    },
    {
      "epoch": 0.38528764539453003,
      "grad_norm": 0.4133789539337158,
      "learning_rate": 1.4097936923894282e-05,
      "loss": 0.8854,
      "step": 2298
    },
    {
      "epoch": 0.3854553075552761,
      "grad_norm": 0.4353831112384796,
      "learning_rate": 1.4092982659651822e-05,
      "loss": 0.8621,
      "step": 2299
    },
    {
      "epoch": 0.3856229697160222,
      "grad_norm": 0.5475447773933411,
      "learning_rate": 1.4088027188338088e-05,
      "loss": 0.8884,
      "step": 2300
    },
    {
      "epoch": 0.3857906318767683,
      "grad_norm": 0.49370449781417847,
      "learning_rate": 1.4083070511414514e-05,
      "loss": 0.9254,
      "step": 2301
    },
    {
      "epoch": 0.3859582940375144,
      "grad_norm": 0.5450944900512695,
      "learning_rate": 1.4078112630342891e-05,
      "loss": 0.8184,
      "step": 2302
    },
    {
      "epoch": 0.3861259561982605,
      "grad_norm": 0.5551537275314331,
      "learning_rate": 1.4073153546585352e-05,
      "loss": 0.8938,
      "step": 2303
    },
    {
      "epoch": 0.3862936183590066,
      "grad_norm": 0.5214157700538635,
      "learning_rate": 1.4068193261604397e-05,
      "loss": 0.8185,
      "step": 2304
    },
    {
      "epoch": 0.3864612805197527,
      "grad_norm": 0.5265510678291321,
      "learning_rate": 1.4063231776862872e-05,
      "loss": 0.9045,
      "step": 2305
    },
    {
      "epoch": 0.3866289426804988,
      "grad_norm": 0.4628002345561981,
      "learning_rate": 1.4058269093823982e-05,
      "loss": 0.8955,
      "step": 2306
    },
    {
      "epoch": 0.3867966048412449,
      "grad_norm": 0.46717819571495056,
      "learning_rate": 1.4053305213951283e-05,
      "loss": 0.807,
      "step": 2307
    },
    {
      "epoch": 0.386964267001991,
      "grad_norm": 0.5428999066352844,
      "learning_rate": 1.4048340138708684e-05,
      "loss": 0.8461,
      "step": 2308
    },
    {
      "epoch": 0.3871319291627371,
      "grad_norm": 0.7314903140068054,
      "learning_rate": 1.404337386956045e-05,
      "loss": 0.7835,
      "step": 2309
    },
    {
      "epoch": 0.3872995913234832,
      "grad_norm": 0.8051000833511353,
      "learning_rate": 1.4038406407971192e-05,
      "loss": 0.8377,
      "step": 2310
    },
    {
      "epoch": 0.3874672534842293,
      "grad_norm": 0.7588483691215515,
      "learning_rate": 1.4033437755405878e-05,
      "loss": 0.9077,
      "step": 2311
    },
    {
      "epoch": 0.3876349156449754,
      "grad_norm": 0.6154654622077942,
      "learning_rate": 1.4028467913329826e-05,
      "loss": 0.8729,
      "step": 2312
    },
    {
      "epoch": 0.38780257780572147,
      "grad_norm": 0.44144371151924133,
      "learning_rate": 1.4023496883208702e-05,
      "loss": 0.9389,
      "step": 2313
    },
    {
      "epoch": 0.38797023996646757,
      "grad_norm": 0.4922724664211273,
      "learning_rate": 1.4018524666508526e-05,
      "loss": 0.8981,
      "step": 2314
    },
    {
      "epoch": 0.38813790212721366,
      "grad_norm": 0.5737454891204834,
      "learning_rate": 1.4013551264695663e-05,
      "loss": 0.9817,
      "step": 2315
    },
    {
      "epoch": 0.38830556428795976,
      "grad_norm": 0.6628448367118835,
      "learning_rate": 1.4008576679236834e-05,
      "loss": 0.9185,
      "step": 2316
    },
    {
      "epoch": 0.38847322644870586,
      "grad_norm": 1.202582597732544,
      "learning_rate": 1.4003600911599105e-05,
      "loss": 0.9302,
      "step": 2317
    },
    {
      "epoch": 0.38864088860945195,
      "grad_norm": 0.4161703586578369,
      "learning_rate": 1.3998623963249894e-05,
      "loss": 0.8186,
      "step": 2318
    },
    {
      "epoch": 0.38880855077019805,
      "grad_norm": 0.658116340637207,
      "learning_rate": 1.3993645835656955e-05,
      "loss": 0.7665,
      "step": 2319
    },
    {
      "epoch": 0.38897621293094414,
      "grad_norm": 0.4697672426700592,
      "learning_rate": 1.3988666530288413e-05,
      "loss": 0.8408,
      "step": 2320
    },
    {
      "epoch": 0.38914387509169024,
      "grad_norm": 0.552117645740509,
      "learning_rate": 1.3983686048612716e-05,
      "loss": 0.7742,
      "step": 2321
    },
    {
      "epoch": 0.38931153725243633,
      "grad_norm": 0.8014466166496277,
      "learning_rate": 1.3978704392098678e-05,
      "loss": 0.8477,
      "step": 2322
    },
    {
      "epoch": 0.38947919941318243,
      "grad_norm": 0.6279250383377075,
      "learning_rate": 1.3973721562215443e-05,
      "loss": 0.8233,
      "step": 2323
    },
    {
      "epoch": 0.3896468615739285,
      "grad_norm": 0.5173192024230957,
      "learning_rate": 1.396873756043252e-05,
      "loss": 0.912,
      "step": 2324
    },
    {
      "epoch": 0.3898145237346746,
      "grad_norm": 0.5528737306594849,
      "learning_rate": 1.396375238821974e-05,
      "loss": 0.926,
      "step": 2325
    },
    {
      "epoch": 0.3899821858954207,
      "grad_norm": 0.6598599553108215,
      "learning_rate": 1.3958766047047298e-05,
      "loss": 0.8262,
      "step": 2326
    },
    {
      "epoch": 0.3901498480561668,
      "grad_norm": 0.5384594798088074,
      "learning_rate": 1.3953778538385732e-05,
      "loss": 0.8689,
      "step": 2327
    },
    {
      "epoch": 0.3903175102169129,
      "grad_norm": 0.47622594237327576,
      "learning_rate": 1.3948789863705914e-05,
      "loss": 0.8962,
      "step": 2328
    },
    {
      "epoch": 0.390485172377659,
      "grad_norm": 0.4942217171192169,
      "learning_rate": 1.3943800024479068e-05,
      "loss": 0.903,
      "step": 2329
    },
    {
      "epoch": 0.3906528345384051,
      "grad_norm": 0.5775742530822754,
      "learning_rate": 1.393880902217676e-05,
      "loss": 0.835,
      "step": 2330
    },
    {
      "epoch": 0.3908204966991512,
      "grad_norm": 0.42464926838874817,
      "learning_rate": 1.3933816858270898e-05,
      "loss": 0.945,
      "step": 2331
    },
    {
      "epoch": 0.3909881588598973,
      "grad_norm": 0.3777066767215729,
      "learning_rate": 1.3928823534233733e-05,
      "loss": 0.8608,
      "step": 2332
    },
    {
      "epoch": 0.3911558210206434,
      "grad_norm": 0.374035507440567,
      "learning_rate": 1.3923829051537858e-05,
      "loss": 0.9155,
      "step": 2333
    },
    {
      "epoch": 0.3913234831813895,
      "grad_norm": 0.5102998614311218,
      "learning_rate": 1.3918833411656205e-05,
      "loss": 0.817,
      "step": 2334
    },
    {
      "epoch": 0.3914911453421356,
      "grad_norm": 0.3295595645904541,
      "learning_rate": 1.3913836616062056e-05,
      "loss": 0.8171,
      "step": 2335
    },
    {
      "epoch": 0.3916588075028817,
      "grad_norm": 0.6316651701927185,
      "learning_rate": 1.3908838666229023e-05,
      "loss": 0.9449,
      "step": 2336
    },
    {
      "epoch": 0.3918264696636278,
      "grad_norm": 0.8228375911712646,
      "learning_rate": 1.3903839563631065e-05,
      "loss": 0.7795,
      "step": 2337
    },
    {
      "epoch": 0.39199413182437387,
      "grad_norm": 0.8097789287567139,
      "learning_rate": 1.389883930974248e-05,
      "loss": 0.89,
      "step": 2338
    },
    {
      "epoch": 0.39216179398511997,
      "grad_norm": 0.4250030517578125,
      "learning_rate": 1.3893837906037903e-05,
      "loss": 0.918,
      "step": 2339
    },
    {
      "epoch": 0.39232945614586606,
      "grad_norm": 0.6134943962097168,
      "learning_rate": 1.3888835353992312e-05,
      "loss": 0.8479,
      "step": 2340
    },
    {
      "epoch": 0.39249711830661216,
      "grad_norm": 0.886298418045044,
      "learning_rate": 1.388383165508102e-05,
      "loss": 0.9473,
      "step": 2341
    },
    {
      "epoch": 0.39266478046735825,
      "grad_norm": 0.378825306892395,
      "learning_rate": 1.3878826810779677e-05,
      "loss": 0.9453,
      "step": 2342
    },
    {
      "epoch": 0.39283244262810435,
      "grad_norm": 0.5482847094535828,
      "learning_rate": 1.3873820822564282e-05,
      "loss": 0.9344,
      "step": 2343
    },
    {
      "epoch": 0.39300010478885045,
      "grad_norm": 0.5132451057434082,
      "learning_rate": 1.3868813691911154e-05,
      "loss": 0.8788,
      "step": 2344
    },
    {
      "epoch": 0.39316776694959654,
      "grad_norm": 0.4483496844768524,
      "learning_rate": 1.3863805420296965e-05,
      "loss": 0.9336,
      "step": 2345
    },
    {
      "epoch": 0.39333542911034264,
      "grad_norm": 0.4734271168708801,
      "learning_rate": 1.3858796009198708e-05,
      "loss": 0.9291,
      "step": 2346
    },
    {
      "epoch": 0.39350309127108873,
      "grad_norm": 0.5556327700614929,
      "learning_rate": 1.385378546009373e-05,
      "loss": 0.8528,
      "step": 2347
    },
    {
      "epoch": 0.39367075343183483,
      "grad_norm": 0.39046409726142883,
      "learning_rate": 1.3848773774459696e-05,
      "loss": 0.952,
      "step": 2348
    },
    {
      "epoch": 0.3938384155925809,
      "grad_norm": 0.6670508980751038,
      "learning_rate": 1.3843760953774616e-05,
      "loss": 0.9058,
      "step": 2349
    },
    {
      "epoch": 0.394006077753327,
      "grad_norm": 0.4425426423549652,
      "learning_rate": 1.3838746999516836e-05,
      "loss": 0.9347,
      "step": 2350
    },
    {
      "epoch": 0.3941737399140731,
      "grad_norm": 0.5918128490447998,
      "learning_rate": 1.3833731913165032e-05,
      "loss": 0.9125,
      "step": 2351
    },
    {
      "epoch": 0.3943414020748192,
      "grad_norm": 0.3698294460773468,
      "learning_rate": 1.3828715696198214e-05,
      "loss": 0.8371,
      "step": 2352
    },
    {
      "epoch": 0.3945090642355653,
      "grad_norm": 0.4565315246582031,
      "learning_rate": 1.3823698350095726e-05,
      "loss": 0.9464,
      "step": 2353
    },
    {
      "epoch": 0.3946767263963114,
      "grad_norm": 0.4425026476383209,
      "learning_rate": 1.381867987633725e-05,
      "loss": 0.9439,
      "step": 2354
    },
    {
      "epoch": 0.3948443885570575,
      "grad_norm": 0.5444761514663696,
      "learning_rate": 1.3813660276402787e-05,
      "loss": 0.8747,
      "step": 2355
    },
    {
      "epoch": 0.39501205071780365,
      "grad_norm": 0.542966902256012,
      "learning_rate": 1.3808639551772688e-05,
      "loss": 0.9186,
      "step": 2356
    },
    {
      "epoch": 0.39517971287854975,
      "grad_norm": 0.42123109102249146,
      "learning_rate": 1.3803617703927621e-05,
      "loss": 0.8864,
      "step": 2357
    },
    {
      "epoch": 0.39534737503929585,
      "grad_norm": 0.4050232768058777,
      "learning_rate": 1.3798594734348596e-05,
      "loss": 0.8296,
      "step": 2358
    },
    {
      "epoch": 0.39551503720004194,
      "grad_norm": 0.536188542842865,
      "learning_rate": 1.3793570644516947e-05,
      "loss": 0.8622,
      "step": 2359
    },
    {
      "epoch": 0.39568269936078804,
      "grad_norm": 0.384074866771698,
      "learning_rate": 1.3788545435914337e-05,
      "loss": 0.8495,
      "step": 2360
    },
    {
      "epoch": 0.39585036152153413,
      "grad_norm": 0.4254181683063507,
      "learning_rate": 1.3783519110022768e-05,
      "loss": 0.8757,
      "step": 2361
    },
    {
      "epoch": 0.39601802368228023,
      "grad_norm": 0.45154184103012085,
      "learning_rate": 1.3778491668324561e-05,
      "loss": 0.8858,
      "step": 2362
    },
    {
      "epoch": 0.3961856858430263,
      "grad_norm": 0.45874765515327454,
      "learning_rate": 1.3773463112302374e-05,
      "loss": 0.8704,
      "step": 2363
    },
    {
      "epoch": 0.3963533480037724,
      "grad_norm": 0.4734143614768982,
      "learning_rate": 1.3768433443439189e-05,
      "loss": 0.8793,
      "step": 2364
    },
    {
      "epoch": 0.3965210101645185,
      "grad_norm": 0.5925267338752747,
      "learning_rate": 1.3763402663218317e-05,
      "loss": 0.8406,
      "step": 2365
    },
    {
      "epoch": 0.3966886723252646,
      "grad_norm": 0.626492440700531,
      "learning_rate": 1.3758370773123397e-05,
      "loss": 0.9716,
      "step": 2366
    },
    {
      "epoch": 0.3968563344860107,
      "grad_norm": 0.6277036666870117,
      "learning_rate": 1.3753337774638397e-05,
      "loss": 0.8995,
      "step": 2367
    },
    {
      "epoch": 0.3970239966467568,
      "grad_norm": 0.5515948534011841,
      "learning_rate": 1.3748303669247611e-05,
      "loss": 0.9342,
      "step": 2368
    },
    {
      "epoch": 0.3971916588075029,
      "grad_norm": 0.8049086928367615,
      "learning_rate": 1.374326845843566e-05,
      "loss": 0.94,
      "step": 2369
    },
    {
      "epoch": 0.397359320968249,
      "grad_norm": 0.45917946100234985,
      "learning_rate": 1.3738232143687489e-05,
      "loss": 0.9634,
      "step": 2370
    },
    {
      "epoch": 0.3975269831289951,
      "grad_norm": 0.5680646896362305,
      "learning_rate": 1.3733194726488366e-05,
      "loss": 0.8937,
      "step": 2371
    },
    {
      "epoch": 0.3976946452897412,
      "grad_norm": 0.5159409046173096,
      "learning_rate": 1.3728156208323892e-05,
      "loss": 0.8145,
      "step": 2372
    },
    {
      "epoch": 0.3978623074504873,
      "grad_norm": 0.45107534527778625,
      "learning_rate": 1.3723116590679986e-05,
      "loss": 0.9567,
      "step": 2373
    },
    {
      "epoch": 0.3980299696112334,
      "grad_norm": 0.374700129032135,
      "learning_rate": 1.3718075875042895e-05,
      "loss": 0.9302,
      "step": 2374
    },
    {
      "epoch": 0.3981976317719795,
      "grad_norm": 0.6136937737464905,
      "learning_rate": 1.3713034062899191e-05,
      "loss": 0.9222,
      "step": 2375
    },
    {
      "epoch": 0.3983652939327256,
      "grad_norm": 0.4689367115497589,
      "learning_rate": 1.3707991155735761e-05,
      "loss": 0.8588,
      "step": 2376
    },
    {
      "epoch": 0.39853295609347167,
      "grad_norm": 0.45876532793045044,
      "learning_rate": 1.3702947155039826e-05,
      "loss": 0.9334,
      "step": 2377
    },
    {
      "epoch": 0.39870061825421776,
      "grad_norm": 0.3927895128726959,
      "learning_rate": 1.369790206229892e-05,
      "loss": 0.8993,
      "step": 2378
    },
    {
      "epoch": 0.39886828041496386,
      "grad_norm": 0.49190419912338257,
      "learning_rate": 1.3692855879000908e-05,
      "loss": 0.8762,
      "step": 2379
    },
    {
      "epoch": 0.39903594257570996,
      "grad_norm": 0.5364404320716858,
      "learning_rate": 1.3687808606633965e-05,
      "loss": 0.9148,
      "step": 2380
    },
    {
      "epoch": 0.39920360473645605,
      "grad_norm": 0.41728052496910095,
      "learning_rate": 1.3682760246686604e-05,
      "loss": 0.8652,
      "step": 2381
    },
    {
      "epoch": 0.39937126689720215,
      "grad_norm": 0.46292200684547424,
      "learning_rate": 1.367771080064764e-05,
      "loss": 0.8716,
      "step": 2382
    },
    {
      "epoch": 0.39953892905794824,
      "grad_norm": 0.5097537040710449,
      "learning_rate": 1.367266027000622e-05,
      "loss": 0.8465,
      "step": 2383
    },
    {
      "epoch": 0.39970659121869434,
      "grad_norm": 0.51020747423172,
      "learning_rate": 1.3667608656251808e-05,
      "loss": 0.8676,
      "step": 2384
    },
    {
      "epoch": 0.39987425337944044,
      "grad_norm": 0.5306868553161621,
      "learning_rate": 1.366255596087419e-05,
      "loss": 0.81,
      "step": 2385
    },
    {
      "epoch": 0.40004191554018653,
      "grad_norm": 0.4405503273010254,
      "learning_rate": 1.3657502185363467e-05,
      "loss": 0.8347,
      "step": 2386
    },
    {
      "epoch": 0.40020957770093263,
      "grad_norm": 0.46693986654281616,
      "learning_rate": 1.3652447331210057e-05,
      "loss": 0.9545,
      "step": 2387
    },
    {
      "epoch": 0.4003772398616787,
      "grad_norm": 0.7226107716560364,
      "learning_rate": 1.3647391399904701e-05,
      "loss": 0.9323,
      "step": 2388
    },
    {
      "epoch": 0.4005449020224248,
      "grad_norm": 0.6572045683860779,
      "learning_rate": 1.3642334392938458e-05,
      "loss": 0.9501,
      "step": 2389
    },
    {
      "epoch": 0.4007125641831709,
      "grad_norm": 0.34955379366874695,
      "learning_rate": 1.3637276311802701e-05,
      "loss": 0.8603,
      "step": 2390
    },
    {
      "epoch": 0.400880226343917,
      "grad_norm": 0.48697224259376526,
      "learning_rate": 1.3632217157989118e-05,
      "loss": 0.8229,
      "step": 2391
    },
    {
      "epoch": 0.4010478885046631,
      "grad_norm": 0.4833279252052307,
      "learning_rate": 1.3627156932989719e-05,
      "loss": 0.9941,
      "step": 2392
    },
    {
      "epoch": 0.4012155506654092,
      "grad_norm": 0.6540115475654602,
      "learning_rate": 1.3622095638296827e-05,
      "loss": 0.8278,
      "step": 2393
    },
    {
      "epoch": 0.4013832128261553,
      "grad_norm": 0.5749005079269409,
      "learning_rate": 1.3617033275403078e-05,
      "loss": 0.9305,
      "step": 2394
    },
    {
      "epoch": 0.4015508749869014,
      "grad_norm": 0.6052101850509644,
      "learning_rate": 1.3611969845801427e-05,
      "loss": 0.8888,
      "step": 2395
    },
    {
      "epoch": 0.4017185371476475,
      "grad_norm": 0.45506125688552856,
      "learning_rate": 1.3606905350985143e-05,
      "loss": 0.8748,
      "step": 2396
    },
    {
      "epoch": 0.4018861993083936,
      "grad_norm": 0.45200562477111816,
      "learning_rate": 1.3601839792447809e-05,
      "loss": 0.8185,
      "step": 2397
    },
    {
      "epoch": 0.4020538614691397,
      "grad_norm": 0.38190123438835144,
      "learning_rate": 1.3596773171683317e-05,
      "loss": 0.827,
      "step": 2398
    },
    {
      "epoch": 0.4022215236298858,
      "grad_norm": 0.4883127510547638,
      "learning_rate": 1.3591705490185875e-05,
      "loss": 0.895,
      "step": 2399
    },
    {
      "epoch": 0.4023891857906319,
      "grad_norm": 0.4430232644081116,
      "learning_rate": 1.3586636749450015e-05,
      "loss": 0.8115,
      "step": 2400
    },
    {
      "epoch": 0.40255684795137797,
      "grad_norm": 0.6474736928939819,
      "learning_rate": 1.3581566950970562e-05,
      "loss": 0.8738,
      "step": 2401
    },
    {
      "epoch": 0.40272451011212407,
      "grad_norm": 0.9033061861991882,
      "learning_rate": 1.3576496096242665e-05,
      "loss": 0.814,
      "step": 2402
    },
    {
      "epoch": 0.40289217227287016,
      "grad_norm": 0.4570147395133972,
      "learning_rate": 1.3571424186761784e-05,
      "loss": 0.7655,
      "step": 2403
    },
    {
      "epoch": 0.40305983443361626,
      "grad_norm": 0.7191325426101685,
      "learning_rate": 1.3566351224023684e-05,
      "loss": 0.9122,
      "step": 2404
    },
    {
      "epoch": 0.40322749659436236,
      "grad_norm": 0.3910098373889923,
      "learning_rate": 1.3561277209524449e-05,
      "loss": 0.8671,
      "step": 2405
    },
    {
      "epoch": 0.40339515875510845,
      "grad_norm": 0.5069496035575867,
      "learning_rate": 1.3556202144760461e-05,
      "loss": 0.8933,
      "step": 2406
    },
    {
      "epoch": 0.40356282091585455,
      "grad_norm": 0.400499552488327,
      "learning_rate": 1.355112603122843e-05,
      "loss": 0.9089,
      "step": 2407
    },
    {
      "epoch": 0.40373048307660064,
      "grad_norm": 0.3381521701812744,
      "learning_rate": 1.3546048870425356e-05,
      "loss": 0.965,
      "step": 2408
    },
    {
      "epoch": 0.40389814523734674,
      "grad_norm": 0.46304380893707275,
      "learning_rate": 1.3540970663848565e-05,
      "loss": 0.8994,
      "step": 2409
    },
    {
      "epoch": 0.40406580739809284,
      "grad_norm": 0.6324905157089233,
      "learning_rate": 1.3535891412995675e-05,
      "loss": 0.8701,
      "step": 2410
    },
    {
      "epoch": 0.40423346955883893,
      "grad_norm": 0.45117953419685364,
      "learning_rate": 1.3530811119364623e-05,
      "loss": 0.8599,
      "step": 2411
    },
    {
      "epoch": 0.404401131719585,
      "grad_norm": 0.5117924809455872,
      "learning_rate": 1.3525729784453648e-05,
      "loss": 0.8782,
      "step": 2412
    },
    {
      "epoch": 0.4045687938803311,
      "grad_norm": 0.452868789434433,
      "learning_rate": 1.3520647409761305e-05,
      "loss": 0.8646,
      "step": 2413
    },
    {
      "epoch": 0.4047364560410772,
      "grad_norm": 0.3938448131084442,
      "learning_rate": 1.3515563996786443e-05,
      "loss": 0.8669,
      "step": 2414
    },
    {
      "epoch": 0.4049041182018233,
      "grad_norm": 0.5183711647987366,
      "learning_rate": 1.3510479547028226e-05,
      "loss": 0.8956,
      "step": 2415
    },
    {
      "epoch": 0.4050717803625694,
      "grad_norm": 0.45423707365989685,
      "learning_rate": 1.3505394061986122e-05,
      "loss": 0.8156,
      "step": 2416
    },
    {
      "epoch": 0.4052394425233155,
      "grad_norm": 0.4229229986667633,
      "learning_rate": 1.3500307543159902e-05,
      "loss": 0.8449,
      "step": 2417
    },
    {
      "epoch": 0.4054071046840616,
      "grad_norm": 0.4813057780265808,
      "learning_rate": 1.3495219992049644e-05,
      "loss": 0.9262,
      "step": 2418
    },
    {
      "epoch": 0.4055747668448077,
      "grad_norm": 0.737172544002533,
      "learning_rate": 1.349013141015573e-05,
      "loss": 0.9281,
      "step": 2419
    },
    {
      "epoch": 0.4057424290055538,
      "grad_norm": 0.5750868320465088,
      "learning_rate": 1.3485041798978846e-05,
      "loss": 0.9335,
      "step": 2420
    },
    {
      "epoch": 0.4059100911662999,
      "grad_norm": 0.42083829641342163,
      "learning_rate": 1.347995116001998e-05,
      "loss": 0.7781,
      "step": 2421
    },
    {
      "epoch": 0.406077753327046,
      "grad_norm": 0.5151885151863098,
      "learning_rate": 1.3474859494780423e-05,
      "loss": 0.909,
      "step": 2422
    },
    {
      "epoch": 0.4062454154877921,
      "grad_norm": 0.4647657871246338,
      "learning_rate": 1.3469766804761774e-05,
      "loss": 0.8934,
      "step": 2423
    },
    {
      "epoch": 0.4064130776485382,
      "grad_norm": 0.5370908975601196,
      "learning_rate": 1.3464673091465929e-05,
      "loss": 0.9387,
      "step": 2424
    },
    {
      "epoch": 0.4065807398092843,
      "grad_norm": 0.4686649441719055,
      "learning_rate": 1.3459578356395088e-05,
      "loss": 0.8882,
      "step": 2425
    },
    {
      "epoch": 0.40674840197003037,
      "grad_norm": 0.6504122018814087,
      "learning_rate": 1.345448260105175e-05,
      "loss": 0.8825,
      "step": 2426
    },
    {
      "epoch": 0.40691606413077647,
      "grad_norm": 0.5108534097671509,
      "learning_rate": 1.344938582693872e-05,
      "loss": 0.8653,
      "step": 2427
    },
    {
      "epoch": 0.40708372629152256,
      "grad_norm": 0.4992177188396454,
      "learning_rate": 1.3444288035559092e-05,
      "loss": 0.8074,
      "step": 2428
    },
    {
      "epoch": 0.40725138845226866,
      "grad_norm": 0.5172570943832397,
      "learning_rate": 1.3439189228416271e-05,
      "loss": 0.8284,
      "step": 2429
    },
    {
      "epoch": 0.40741905061301475,
      "grad_norm": 0.4258668124675751,
      "learning_rate": 1.3434089407013963e-05,
      "loss": 0.8868,
      "step": 2430
    },
    {
      "epoch": 0.40758671277376085,
      "grad_norm": 0.4065842628479004,
      "learning_rate": 1.3428988572856163e-05,
      "loss": 0.908,
      "step": 2431
    },
    {
      "epoch": 0.40775437493450695,
      "grad_norm": 0.6188164353370667,
      "learning_rate": 1.3423886727447176e-05,
      "loss": 0.9106,
      "step": 2432
    },
    {
      "epoch": 0.40792203709525304,
      "grad_norm": 0.549799382686615,
      "learning_rate": 1.3418783872291595e-05,
      "loss": 0.9599,
      "step": 2433
    },
    {
      "epoch": 0.40808969925599914,
      "grad_norm": 0.7364528775215149,
      "learning_rate": 1.3413680008894312e-05,
      "loss": 0.9087,
      "step": 2434
    },
    {
      "epoch": 0.40825736141674523,
      "grad_norm": 0.6526517271995544,
      "learning_rate": 1.3408575138760526e-05,
      "loss": 0.975,
      "step": 2435
    },
    {
      "epoch": 0.40842502357749133,
      "grad_norm": 0.44261178374290466,
      "learning_rate": 1.3403469263395725e-05,
      "loss": 0.9273,
      "step": 2436
    },
    {
      "epoch": 0.4085926857382374,
      "grad_norm": 0.45836639404296875,
      "learning_rate": 1.339836238430569e-05,
      "loss": 0.7632,
      "step": 2437
    },
    {
      "epoch": 0.4087603478989835,
      "grad_norm": 0.39426425099372864,
      "learning_rate": 1.3393254502996511e-05,
      "loss": 1.0172,
      "step": 2438
    },
    {
      "epoch": 0.4089280100597296,
      "grad_norm": 0.47519373893737793,
      "learning_rate": 1.3388145620974558e-05,
      "loss": 0.9009,
      "step": 2439
    },
    {
      "epoch": 0.4090956722204757,
      "grad_norm": 0.4381753206253052,
      "learning_rate": 1.3383035739746507e-05,
      "loss": 0.8718,
      "step": 2440
    },
    {
      "epoch": 0.4092633343812218,
      "grad_norm": 0.5138468742370605,
      "learning_rate": 1.3377924860819325e-05,
      "loss": 0.8698,
      "step": 2441
    },
    {
      "epoch": 0.40943099654196796,
      "grad_norm": 0.40322327613830566,
      "learning_rate": 1.3372812985700274e-05,
      "loss": 0.9485,
      "step": 2442
    },
    {
      "epoch": 0.40959865870271406,
      "grad_norm": 0.5753828883171082,
      "learning_rate": 1.3367700115896908e-05,
      "loss": 0.8841,
      "step": 2443
    },
    {
      "epoch": 0.40976632086346015,
      "grad_norm": 0.3679884970188141,
      "learning_rate": 1.3362586252917079e-05,
      "loss": 0.8343,
      "step": 2444
    },
    {
      "epoch": 0.40993398302420625,
      "grad_norm": 0.4531792104244232,
      "learning_rate": 1.335747139826892e-05,
      "loss": 0.9368,
      "step": 2445
    },
    {
      "epoch": 0.41010164518495235,
      "grad_norm": 0.43219149112701416,
      "learning_rate": 1.3352355553460873e-05,
      "loss": 0.8713,
      "step": 2446
    },
    {
      "epoch": 0.41026930734569844,
      "grad_norm": 0.5404754281044006,
      "learning_rate": 1.3347238720001663e-05,
      "loss": 0.8641,
      "step": 2447
    },
    {
      "epoch": 0.41043696950644454,
      "grad_norm": 0.47969868779182434,
      "learning_rate": 1.3342120899400305e-05,
      "loss": 0.8207,
      "step": 2448
    },
    {
      "epoch": 0.41060463166719063,
      "grad_norm": 0.5258057713508606,
      "learning_rate": 1.333700209316611e-05,
      "loss": 0.9198,
      "step": 2449
    },
    {
      "epoch": 0.41077229382793673,
      "grad_norm": 0.41233959794044495,
      "learning_rate": 1.3331882302808679e-05,
      "loss": 0.9456,
      "step": 2450
    },
    {
      "epoch": 0.4109399559886828,
      "grad_norm": 0.5792062282562256,
      "learning_rate": 1.3326761529837895e-05,
      "loss": 0.818,
      "step": 2451
    },
    {
      "epoch": 0.4111076181494289,
      "grad_norm": 0.5087970495223999,
      "learning_rate": 1.3321639775763942e-05,
      "loss": 0.9785,
      "step": 2452
    },
    {
      "epoch": 0.411275280310175,
      "grad_norm": 0.40101125836372375,
      "learning_rate": 1.3316517042097285e-05,
      "loss": 0.8717,
      "step": 2453
    },
    {
      "epoch": 0.4114429424709211,
      "grad_norm": 0.4285100996494293,
      "learning_rate": 1.3311393330348689e-05,
      "loss": 0.9145,
      "step": 2454
    },
    {
      "epoch": 0.4116106046316672,
      "grad_norm": 0.599571943283081,
      "learning_rate": 1.3306268642029194e-05,
      "loss": 0.9772,
      "step": 2455
    },
    {
      "epoch": 0.4117782667924133,
      "grad_norm": 0.660012423992157,
      "learning_rate": 1.3301142978650133e-05,
      "loss": 0.8471,
      "step": 2456
    },
    {
      "epoch": 0.4119459289531594,
      "grad_norm": 1.0488002300262451,
      "learning_rate": 1.3296016341723135e-05,
      "loss": 0.8742,
      "step": 2457
    },
    {
      "epoch": 0.4121135911139055,
      "grad_norm": 0.5106312036514282,
      "learning_rate": 1.32908887327601e-05,
      "loss": 0.9073,
      "step": 2458
    },
    {
      "epoch": 0.4122812532746516,
      "grad_norm": 0.4380190670490265,
      "learning_rate": 1.328576015327323e-05,
      "loss": 0.8505,
      "step": 2459
    },
    {
      "epoch": 0.4124489154353977,
      "grad_norm": 0.4387100338935852,
      "learning_rate": 1.3280630604775002e-05,
      "loss": 0.9409,
      "step": 2460
    },
    {
      "epoch": 0.4126165775961438,
      "grad_norm": 0.6855369210243225,
      "learning_rate": 1.3275500088778186e-05,
      "loss": 0.8504,
      "step": 2461
    },
    {
      "epoch": 0.4127842397568899,
      "grad_norm": 0.38202857971191406,
      "learning_rate": 1.3270368606795833e-05,
      "loss": 0.9885,
      "step": 2462
    },
    {
      "epoch": 0.412951901917636,
      "grad_norm": 0.6671926379203796,
      "learning_rate": 1.3265236160341282e-05,
      "loss": 0.7529,
      "step": 2463
    },
    {
      "epoch": 0.4131195640783821,
      "grad_norm": 0.6485735774040222,
      "learning_rate": 1.3260102750928154e-05,
      "loss": 0.8242,
      "step": 2464
    },
    {
      "epoch": 0.41328722623912817,
      "grad_norm": 0.3961820602416992,
      "learning_rate": 1.3254968380070356e-05,
      "loss": 0.8644,
      "step": 2465
    },
    {
      "epoch": 0.41345488839987427,
      "grad_norm": 0.7097799777984619,
      "learning_rate": 1.324983304928208e-05,
      "loss": 0.8563,
      "step": 2466
    },
    {
      "epoch": 0.41362255056062036,
      "grad_norm": 0.46577221155166626,
      "learning_rate": 1.3244696760077793e-05,
      "loss": 0.9005,
      "step": 2467
    },
    {
      "epoch": 0.41379021272136646,
      "grad_norm": 0.36700090765953064,
      "learning_rate": 1.3239559513972251e-05,
      "loss": 0.9534,
      "step": 2468
    },
    {
      "epoch": 0.41395787488211255,
      "grad_norm": 0.37399622797966003,
      "learning_rate": 1.3234421312480494e-05,
      "loss": 0.9114,
      "step": 2469
    },
    {
      "epoch": 0.41412553704285865,
      "grad_norm": 0.656875491142273,
      "learning_rate": 1.3229282157117847e-05,
      "loss": 0.9358,
      "step": 2470
    },
    {
      "epoch": 0.41429319920360475,
      "grad_norm": 0.3412625789642334,
      "learning_rate": 1.3224142049399896e-05,
      "loss": 0.7384,
      "step": 2471
    },
    {
      "epoch": 0.41446086136435084,
      "grad_norm": 0.6255205869674683,
      "learning_rate": 1.3219000990842531e-05,
      "loss": 0.859,
      "step": 2472
    },
    {
      "epoch": 0.41462852352509694,
      "grad_norm": 0.809914231300354,
      "learning_rate": 1.321385898296192e-05,
      "loss": 0.7862,
      "step": 2473
    },
    {
      "epoch": 0.41479618568584303,
      "grad_norm": 0.5074305534362793,
      "learning_rate": 1.3208716027274493e-05,
      "loss": 0.894,
      "step": 2474
    },
    {
      "epoch": 0.41496384784658913,
      "grad_norm": 0.5917060971260071,
      "learning_rate": 1.320357212529698e-05,
      "loss": 0.8439,
      "step": 2475
    },
    {
      "epoch": 0.4151315100073352,
      "grad_norm": 0.5297524333000183,
      "learning_rate": 1.3198427278546376e-05,
      "loss": 0.9077,
      "step": 2476
    },
    {
      "epoch": 0.4152991721680813,
      "grad_norm": 0.4922879934310913,
      "learning_rate": 1.3193281488539963e-05,
      "loss": 0.9503,
      "step": 2477
    },
    {
      "epoch": 0.4154668343288274,
      "grad_norm": 0.4003523886203766,
      "learning_rate": 1.3188134756795296e-05,
      "loss": 0.8866,
      "step": 2478
    },
    {
      "epoch": 0.4156344964895735,
      "grad_norm": 0.4242881238460541,
      "learning_rate": 1.3182987084830213e-05,
      "loss": 0.9004,
      "step": 2479
    },
    {
      "epoch": 0.4158021586503196,
      "grad_norm": 0.46138808131217957,
      "learning_rate": 1.317783847416282e-05,
      "loss": 0.8495,
      "step": 2480
    },
    {
      "epoch": 0.4159698208110657,
      "grad_norm": 0.3736347258090973,
      "learning_rate": 1.3172688926311515e-05,
      "loss": 0.798,
      "step": 2481
    },
    {
      "epoch": 0.4161374829718118,
      "grad_norm": 0.587462306022644,
      "learning_rate": 1.316753844279496e-05,
      "loss": 0.8119,
      "step": 2482
    },
    {
      "epoch": 0.4163051451325579,
      "grad_norm": 0.46949222683906555,
      "learning_rate": 1.3162387025132088e-05,
      "loss": 0.8524,
      "step": 2483
    },
    {
      "epoch": 0.416472807293304,
      "grad_norm": 0.419544517993927,
      "learning_rate": 1.315723467484213e-05,
      "loss": 0.8296,
      "step": 2484
    },
    {
      "epoch": 0.4166404694540501,
      "grad_norm": 0.5041922926902771,
      "learning_rate": 1.3152081393444568e-05,
      "loss": 0.9743,
      "step": 2485
    },
    {
      "epoch": 0.4168081316147962,
      "grad_norm": 0.4275999069213867,
      "learning_rate": 1.3146927182459172e-05,
      "loss": 0.8648,
      "step": 2486
    },
    {
      "epoch": 0.4169757937755423,
      "grad_norm": 0.3998974561691284,
      "learning_rate": 1.3141772043405983e-05,
      "loss": 0.8869,
      "step": 2487
    },
    {
      "epoch": 0.4171434559362884,
      "grad_norm": 0.5498875379562378,
      "learning_rate": 1.3136615977805312e-05,
      "loss": 0.8177,
      "step": 2488
    },
    {
      "epoch": 0.4173111180970345,
      "grad_norm": 0.5881977081298828,
      "learning_rate": 1.3131458987177752e-05,
      "loss": 0.9554,
      "step": 2489
    },
    {
      "epoch": 0.41747878025778057,
      "grad_norm": 0.5910879373550415,
      "learning_rate": 1.3126301073044158e-05,
      "loss": 0.8834,
      "step": 2490
    },
    {
      "epoch": 0.41764644241852666,
      "grad_norm": 0.456267386674881,
      "learning_rate": 1.3121142236925666e-05,
      "loss": 0.8794,
      "step": 2491
    },
    {
      "epoch": 0.41781410457927276,
      "grad_norm": 0.3418677747249603,
      "learning_rate": 1.3115982480343679e-05,
      "loss": 0.7878,
      "step": 2492
    },
    {
      "epoch": 0.41798176674001886,
      "grad_norm": 0.44005241990089417,
      "learning_rate": 1.3110821804819877e-05,
      "loss": 0.8275,
      "step": 2493
    },
    {
      "epoch": 0.41814942890076495,
      "grad_norm": 0.4577225148677826,
      "learning_rate": 1.31056602118762e-05,
      "loss": 0.8243,
      "step": 2494
    },
    {
      "epoch": 0.41831709106151105,
      "grad_norm": 0.5882584452629089,
      "learning_rate": 1.3100497703034874e-05,
      "loss": 0.9296,
      "step": 2495
    },
    {
      "epoch": 0.41848475322225714,
      "grad_norm": 0.4153955578804016,
      "learning_rate": 1.3095334279818377e-05,
      "loss": 0.852,
      "step": 2496
    },
    {
      "epoch": 0.41865241538300324,
      "grad_norm": 0.6201179027557373,
      "learning_rate": 1.3090169943749475e-05,
      "loss": 0.8484,
      "step": 2497
    },
    {
      "epoch": 0.41882007754374934,
      "grad_norm": 0.44844475388526917,
      "learning_rate": 1.3085004696351192e-05,
      "loss": 0.9298,
      "step": 2498
    },
    {
      "epoch": 0.41898773970449543,
      "grad_norm": 0.45476582646369934,
      "learning_rate": 1.3079838539146824e-05,
      "loss": 0.9391,
      "step": 2499
    },
    {
      "epoch": 0.41915540186524153,
      "grad_norm": 0.5748103857040405,
      "learning_rate": 1.3074671473659932e-05,
      "loss": 0.8742,
      "step": 2500
    },
    {
      "epoch": 0.4193230640259876,
      "grad_norm": 0.5483225584030151,
      "learning_rate": 1.3069503501414349e-05,
      "loss": 0.8643,
      "step": 2501
    },
    {
      "epoch": 0.4194907261867337,
      "grad_norm": 0.5988959670066833,
      "learning_rate": 1.3064334623934174e-05,
      "loss": 0.9111,
      "step": 2502
    },
    {
      "epoch": 0.4196583883474798,
      "grad_norm": 0.5877673625946045,
      "learning_rate": 1.3059164842743774e-05,
      "loss": 0.9511,
      "step": 2503
    },
    {
      "epoch": 0.4198260505082259,
      "grad_norm": 0.671694278717041,
      "learning_rate": 1.305399415936778e-05,
      "loss": 0.8705,
      "step": 2504
    },
    {
      "epoch": 0.419993712668972,
      "grad_norm": 0.35245388746261597,
      "learning_rate": 1.3048822575331093e-05,
      "loss": 0.8582,
      "step": 2505
    },
    {
      "epoch": 0.4201613748297181,
      "grad_norm": 0.4740123152732849,
      "learning_rate": 1.3043650092158875e-05,
      "loss": 0.8263,
      "step": 2506
    },
    {
      "epoch": 0.4203290369904642,
      "grad_norm": 0.3961458206176758,
      "learning_rate": 1.3038476711376558e-05,
      "loss": 0.8907,
      "step": 2507
    },
    {
      "epoch": 0.4204966991512103,
      "grad_norm": 0.8107627630233765,
      "learning_rate": 1.3033302434509831e-05,
      "loss": 0.8103,
      "step": 2508
    },
    {
      "epoch": 0.4206643613119564,
      "grad_norm": 0.48412755131721497,
      "learning_rate": 1.3028127263084656e-05,
      "loss": 0.8953,
      "step": 2509
    },
    {
      "epoch": 0.4208320234727025,
      "grad_norm": 0.5948150157928467,
      "learning_rate": 1.3022951198627254e-05,
      "loss": 0.871,
      "step": 2510
    },
    {
      "epoch": 0.4209996856334486,
      "grad_norm": 0.4686228334903717,
      "learning_rate": 1.3017774242664112e-05,
      "loss": 0.873,
      "step": 2511
    },
    {
      "epoch": 0.4211673477941947,
      "grad_norm": 0.5049168467521667,
      "learning_rate": 1.3012596396721978e-05,
      "loss": 0.8775,
      "step": 2512
    },
    {
      "epoch": 0.4213350099549408,
      "grad_norm": 0.7438037395477295,
      "learning_rate": 1.3007417662327857e-05,
      "loss": 0.9154,
      "step": 2513
    },
    {
      "epoch": 0.42150267211568687,
      "grad_norm": 0.4495711922645569,
      "learning_rate": 1.3002238041009032e-05,
      "loss": 0.9141,
      "step": 2514
    },
    {
      "epoch": 0.42167033427643297,
      "grad_norm": 0.48235267400741577,
      "learning_rate": 1.2997057534293032e-05,
      "loss": 0.8947,
      "step": 2515
    },
    {
      "epoch": 0.42183799643717906,
      "grad_norm": 0.42142271995544434,
      "learning_rate": 1.2991876143707653e-05,
      "loss": 0.7854,
      "step": 2516
    },
    {
      "epoch": 0.42200565859792516,
      "grad_norm": 0.6156967878341675,
      "learning_rate": 1.2986693870780949e-05,
      "loss": 0.8772,
      "step": 2517
    },
    {
      "epoch": 0.42217332075867126,
      "grad_norm": 0.37790626287460327,
      "learning_rate": 1.2981510717041244e-05,
      "loss": 0.8518,
      "step": 2518
    },
    {
      "epoch": 0.42234098291941735,
      "grad_norm": 0.6372936964035034,
      "learning_rate": 1.2976326684017104e-05,
      "loss": 0.8625,
      "step": 2519
    },
    {
      "epoch": 0.42250864508016345,
      "grad_norm": 0.6741154193878174,
      "learning_rate": 1.297114177323737e-05,
      "loss": 0.8152,
      "step": 2520
    },
    {
      "epoch": 0.42267630724090954,
      "grad_norm": 0.41690656542778015,
      "learning_rate": 1.2965955986231138e-05,
      "loss": 0.8574,
      "step": 2521
    },
    {
      "epoch": 0.42284396940165564,
      "grad_norm": 0.5158107876777649,
      "learning_rate": 1.2960769324527758e-05,
      "loss": 0.8277,
      "step": 2522
    },
    {
      "epoch": 0.42301163156240174,
      "grad_norm": 0.4730595648288727,
      "learning_rate": 1.2955581789656844e-05,
      "loss": 0.9815,
      "step": 2523
    },
    {
      "epoch": 0.42317929372314783,
      "grad_norm": 0.5132426619529724,
      "learning_rate": 1.2950393383148257e-05,
      "loss": 0.8643,
      "step": 2524
    },
    {
      "epoch": 0.4233469558838939,
      "grad_norm": 0.3574404716491699,
      "learning_rate": 1.294520410653213e-05,
      "loss": 1.0134,
      "step": 2525
    },
    {
      "epoch": 0.42351461804464,
      "grad_norm": 0.45536112785339355,
      "learning_rate": 1.2940013961338843e-05,
      "loss": 0.8603,
      "step": 2526
    },
    {
      "epoch": 0.4236822802053861,
      "grad_norm": 0.6392506957054138,
      "learning_rate": 1.2934822949099032e-05,
      "loss": 0.8885,
      "step": 2527
    },
    {
      "epoch": 0.42384994236613227,
      "grad_norm": 0.44098612666130066,
      "learning_rate": 1.2929631071343588e-05,
      "loss": 0.8565,
      "step": 2528
    },
    {
      "epoch": 0.42401760452687837,
      "grad_norm": 0.7202128171920776,
      "learning_rate": 1.2924438329603668e-05,
      "loss": 0.8701,
      "step": 2529
    },
    {
      "epoch": 0.42418526668762446,
      "grad_norm": 0.41382378339767456,
      "learning_rate": 1.2919244725410673e-05,
      "loss": 0.897,
      "step": 2530
    },
    {
      "epoch": 0.42435292884837056,
      "grad_norm": 0.7349418997764587,
      "learning_rate": 1.2914050260296258e-05,
      "loss": 0.8141,
      "step": 2531
    },
    {
      "epoch": 0.42452059100911665,
      "grad_norm": 0.42445793747901917,
      "learning_rate": 1.2908854935792339e-05,
      "loss": 0.94,
      "step": 2532
    },
    {
      "epoch": 0.42468825316986275,
      "grad_norm": 0.7116059064865112,
      "learning_rate": 1.2903658753431077e-05,
      "loss": 0.9679,
      "step": 2533
    },
    {
      "epoch": 0.42485591533060885,
      "grad_norm": 0.4538300633430481,
      "learning_rate": 1.2898461714744897e-05,
      "loss": 0.8899,
      "step": 2534
    },
    {
      "epoch": 0.42502357749135494,
      "grad_norm": 0.3443276882171631,
      "learning_rate": 1.2893263821266462e-05,
      "loss": 0.8228,
      "step": 2535
    },
    {
      "epoch": 0.42519123965210104,
      "grad_norm": 0.44806644320487976,
      "learning_rate": 1.28880650745287e-05,
      "loss": 0.8981,
      "step": 2536
    },
    {
      "epoch": 0.42535890181284713,
      "grad_norm": 0.5360666513442993,
      "learning_rate": 1.2882865476064787e-05,
      "loss": 0.8946,
      "step": 2537
    },
    {
      "epoch": 0.42552656397359323,
      "grad_norm": 0.4778892397880554,
      "learning_rate": 1.2877665027408146e-05,
      "loss": 0.834,
      "step": 2538
    },
    {
      "epoch": 0.4256942261343393,
      "grad_norm": 0.3988834023475647,
      "learning_rate": 1.2872463730092457e-05,
      "loss": 0.9192,
      "step": 2539
    },
    {
      "epoch": 0.4258618882950854,
      "grad_norm": 0.33127453923225403,
      "learning_rate": 1.286726158565164e-05,
      "loss": 0.8632,
      "step": 2540
    },
    {
      "epoch": 0.4260295504558315,
      "grad_norm": 0.5348523259162903,
      "learning_rate": 1.2862058595619882e-05,
      "loss": 0.8979,
      "step": 2541
    },
    {
      "epoch": 0.4261972126165776,
      "grad_norm": 0.4789660573005676,
      "learning_rate": 1.2856854761531601e-05,
      "loss": 0.9459,
      "step": 2542
    },
    {
      "epoch": 0.4263648747773237,
      "grad_norm": 0.45396173000335693,
      "learning_rate": 1.2851650084921474e-05,
      "loss": 0.9301,
      "step": 2543
    },
    {
      "epoch": 0.4265325369380698,
      "grad_norm": 0.9283210635185242,
      "learning_rate": 1.2846444567324428e-05,
      "loss": 0.8131,
      "step": 2544
    },
    {
      "epoch": 0.4267001990988159,
      "grad_norm": 0.4070189595222473,
      "learning_rate": 1.284123821027563e-05,
      "loss": 0.9318,
      "step": 2545
    },
    {
      "epoch": 0.426867861259562,
      "grad_norm": 0.7896355390548706,
      "learning_rate": 1.2836031015310504e-05,
      "loss": 0.9275,
      "step": 2546
    },
    {
      "epoch": 0.4270355234203081,
      "grad_norm": 0.5288454294204712,
      "learning_rate": 1.283082298396471e-05,
      "loss": 0.8689,
      "step": 2547
    },
    {
      "epoch": 0.4272031855810542,
      "grad_norm": 0.6114990711212158,
      "learning_rate": 1.2825614117774166e-05,
      "loss": 0.8482,
      "step": 2548
    },
    {
      "epoch": 0.4273708477418003,
      "grad_norm": 0.4103098213672638,
      "learning_rate": 1.282040441827503e-05,
      "loss": 0.8395,
      "step": 2549
    },
    {
      "epoch": 0.4275385099025464,
      "grad_norm": 0.47522878646850586,
      "learning_rate": 1.2815193887003707e-05,
      "loss": 0.8157,
      "step": 2550
    },
    {
      "epoch": 0.4277061720632925,
      "grad_norm": 0.4092075228691101,
      "learning_rate": 1.2809982525496845e-05,
      "loss": 0.8557,
      "step": 2551
    },
    {
      "epoch": 0.4278738342240386,
      "grad_norm": 0.607429027557373,
      "learning_rate": 1.2804770335291344e-05,
      "loss": 0.8362,
      "step": 2552
    },
    {
      "epoch": 0.42804149638478467,
      "grad_norm": 0.5187923908233643,
      "learning_rate": 1.2799557317924336e-05,
      "loss": 1.0054,
      "step": 2553
    },
    {
      "epoch": 0.42820915854553077,
      "grad_norm": 0.6085100769996643,
      "learning_rate": 1.2794343474933211e-05,
      "loss": 0.8382,
      "step": 2554
    },
    {
      "epoch": 0.42837682070627686,
      "grad_norm": 0.42959046363830566,
      "learning_rate": 1.2789128807855592e-05,
      "loss": 0.8725,
      "step": 2555
    },
    {
      "epoch": 0.42854448286702296,
      "grad_norm": 0.3355506658554077,
      "learning_rate": 1.278391331822935e-05,
      "loss": 0.7844,
      "step": 2556
    },
    {
      "epoch": 0.42871214502776905,
      "grad_norm": 0.5724706649780273,
      "learning_rate": 1.2778697007592602e-05,
      "loss": 0.9136,
      "step": 2557
    },
    {
      "epoch": 0.42887980718851515,
      "grad_norm": 0.4948655366897583,
      "learning_rate": 1.2773479877483695e-05,
      "loss": 0.7616,
      "step": 2558
    },
    {
      "epoch": 0.42904746934926125,
      "grad_norm": 0.44995033740997314,
      "learning_rate": 1.2768261929441228e-05,
      "loss": 0.8336,
      "step": 2559
    },
    {
      "epoch": 0.42921513151000734,
      "grad_norm": 0.32069793343544006,
      "learning_rate": 1.2763043165004038e-05,
      "loss": 0.9137,
      "step": 2560
    },
    {
      "epoch": 0.42938279367075344,
      "grad_norm": 0.4152913987636566,
      "learning_rate": 1.2757823585711209e-05,
      "loss": 0.8715,
      "step": 2561
    },
    {
      "epoch": 0.42955045583149953,
      "grad_norm": 0.8363330364227295,
      "learning_rate": 1.275260319310205e-05,
      "loss": 0.8229,
      "step": 2562
    },
    {
      "epoch": 0.42971811799224563,
      "grad_norm": 0.38372570276260376,
      "learning_rate": 1.2747381988716127e-05,
      "loss": 0.9127,
      "step": 2563
    },
    {
      "epoch": 0.4298857801529917,
      "grad_norm": 0.5683997273445129,
      "learning_rate": 1.2742159974093235e-05,
      "loss": 0.9956,
      "step": 2564
    },
    {
      "epoch": 0.4300534423137378,
      "grad_norm": 0.6207015514373779,
      "learning_rate": 1.2736937150773413e-05,
      "loss": 0.8893,
      "step": 2565
    },
    {
      "epoch": 0.4302211044744839,
      "grad_norm": 0.48093733191490173,
      "learning_rate": 1.2731713520296931e-05,
      "loss": 0.964,
      "step": 2566
    },
    {
      "epoch": 0.43038876663523,
      "grad_norm": 0.5342889428138733,
      "learning_rate": 1.2726489084204308e-05,
      "loss": 0.9238,
      "step": 2567
    },
    {
      "epoch": 0.4305564287959761,
      "grad_norm": 0.36307162046432495,
      "learning_rate": 1.272126384403629e-05,
      "loss": 0.8753,
      "step": 2568
    },
    {
      "epoch": 0.4307240909567222,
      "grad_norm": 0.5717389583587646,
      "learning_rate": 1.2716037801333871e-05,
      "loss": 0.9702,
      "step": 2569
    },
    {
      "epoch": 0.4308917531174683,
      "grad_norm": 0.550253689289093,
      "learning_rate": 1.2710810957638267e-05,
      "loss": 0.873,
      "step": 2570
    },
    {
      "epoch": 0.4310594152782144,
      "grad_norm": 0.5449841618537903,
      "learning_rate": 1.2705583314490947e-05,
      "loss": 0.9191,
      "step": 2571
    },
    {
      "epoch": 0.4312270774389605,
      "grad_norm": 0.34667208790779114,
      "learning_rate": 1.2700354873433608e-05,
      "loss": 0.9353,
      "step": 2572
    },
    {
      "epoch": 0.4313947395997066,
      "grad_norm": 0.4964389204978943,
      "learning_rate": 1.2695125636008176e-05,
      "loss": 0.8727,
      "step": 2573
    },
    {
      "epoch": 0.4315624017604527,
      "grad_norm": 0.44565391540527344,
      "learning_rate": 1.268989560375682e-05,
      "loss": 0.8622,
      "step": 2574
    },
    {
      "epoch": 0.4317300639211988,
      "grad_norm": 0.4834344983100891,
      "learning_rate": 1.2684664778221943e-05,
      "loss": 0.9428,
      "step": 2575
    },
    {
      "epoch": 0.4318977260819449,
      "grad_norm": 0.6531493663787842,
      "learning_rate": 1.2679433160946177e-05,
      "loss": 0.959,
      "step": 2576
    },
    {
      "epoch": 0.432065388242691,
      "grad_norm": 0.445249080657959,
      "learning_rate": 1.2674200753472395e-05,
      "loss": 0.8755,
      "step": 2577
    },
    {
      "epoch": 0.43223305040343707,
      "grad_norm": 0.7106694579124451,
      "learning_rate": 1.2668967557343695e-05,
      "loss": 0.89,
      "step": 2578
    },
    {
      "epoch": 0.43240071256418317,
      "grad_norm": 0.5058774948120117,
      "learning_rate": 1.2663733574103415e-05,
      "loss": 0.9328,
      "step": 2579
    },
    {
      "epoch": 0.43256837472492926,
      "grad_norm": 0.5595027208328247,
      "learning_rate": 1.2658498805295118e-05,
      "loss": 0.8226,
      "step": 2580
    },
    {
      "epoch": 0.43273603688567536,
      "grad_norm": 0.6688955426216125,
      "learning_rate": 1.2653263252462607e-05,
      "loss": 0.9153,
      "step": 2581
    },
    {
      "epoch": 0.43290369904642145,
      "grad_norm": 0.7537813782691956,
      "learning_rate": 1.2648026917149902e-05,
      "loss": 0.8436,
      "step": 2582
    },
    {
      "epoch": 0.43307136120716755,
      "grad_norm": 0.4780575633049011,
      "learning_rate": 1.2642789800901273e-05,
      "loss": 0.8802,
      "step": 2583
    },
    {
      "epoch": 0.43323902336791364,
      "grad_norm": 0.6803731918334961,
      "learning_rate": 1.2637551905261209e-05,
      "loss": 0.8808,
      "step": 2584
    },
    {
      "epoch": 0.43340668552865974,
      "grad_norm": 0.49162110686302185,
      "learning_rate": 1.263231323177442e-05,
      "loss": 0.8593,
      "step": 2585
    },
    {
      "epoch": 0.43357434768940584,
      "grad_norm": 0.45431092381477356,
      "learning_rate": 1.262707378198587e-05,
      "loss": 0.817,
      "step": 2586
    },
    {
      "epoch": 0.43374200985015193,
      "grad_norm": 0.464048832654953,
      "learning_rate": 1.262183355744073e-05,
      "loss": 0.8544,
      "step": 2587
    },
    {
      "epoch": 0.43390967201089803,
      "grad_norm": 0.3684556782245636,
      "learning_rate": 1.2616592559684408e-05,
      "loss": 0.8597,
      "step": 2588
    },
    {
      "epoch": 0.4340773341716441,
      "grad_norm": 0.514134407043457,
      "learning_rate": 1.261135079026254e-05,
      "loss": 0.9222,
      "step": 2589
    },
    {
      "epoch": 0.4342449963323902,
      "grad_norm": 0.37602466344833374,
      "learning_rate": 1.2606108250720985e-05,
      "loss": 0.8417,
      "step": 2590
    },
    {
      "epoch": 0.4344126584931363,
      "grad_norm": 0.5111823081970215,
      "learning_rate": 1.260086494260584e-05,
      "loss": 0.8551,
      "step": 2591
    },
    {
      "epoch": 0.4345803206538824,
      "grad_norm": 0.4155372381210327,
      "learning_rate": 1.2595620867463416e-05,
      "loss": 0.9644,
      "step": 2592
    },
    {
      "epoch": 0.4347479828146285,
      "grad_norm": 0.533291757106781,
      "learning_rate": 1.2590376026840251e-05,
      "loss": 0.8856,
      "step": 2593
    },
    {
      "epoch": 0.4349156449753746,
      "grad_norm": 0.7136791944503784,
      "learning_rate": 1.2585130422283129e-05,
      "loss": 0.8378,
      "step": 2594
    },
    {
      "epoch": 0.4350833071361207,
      "grad_norm": 0.6195120811462402,
      "learning_rate": 1.2579884055339032e-05,
      "loss": 0.8113,
      "step": 2595
    },
    {
      "epoch": 0.4352509692968668,
      "grad_norm": 0.8092840313911438,
      "learning_rate": 1.2574636927555179e-05,
      "loss": 1.0516,
      "step": 2596
    },
    {
      "epoch": 0.4354186314576129,
      "grad_norm": 0.561496376991272,
      "learning_rate": 1.2569389040479012e-05,
      "loss": 0.8378,
      "step": 2597
    },
    {
      "epoch": 0.435586293618359,
      "grad_norm": 0.45630037784576416,
      "learning_rate": 1.2564140395658207e-05,
      "loss": 0.8966,
      "step": 2598
    },
    {
      "epoch": 0.4357539557791051,
      "grad_norm": 0.4840919077396393,
      "learning_rate": 1.2558890994640644e-05,
      "loss": 0.8163,
      "step": 2599
    },
    {
      "epoch": 0.4359216179398512,
      "grad_norm": 0.5131926536560059,
      "learning_rate": 1.2553640838974439e-05,
      "loss": 0.8156,
      "step": 2600
    },
    {
      "epoch": 0.4360892801005973,
      "grad_norm": 0.5673779845237732,
      "learning_rate": 1.2548389930207932e-05,
      "loss": 0.8422,
      "step": 2601
    },
    {
      "epoch": 0.4362569422613434,
      "grad_norm": 0.43523696064949036,
      "learning_rate": 1.2543138269889679e-05,
      "loss": 0.9012,
      "step": 2602
    },
    {
      "epoch": 0.43642460442208947,
      "grad_norm": 0.6371061205863953,
      "learning_rate": 1.253788585956846e-05,
      "loss": 0.9298,
      "step": 2603
    },
    {
      "epoch": 0.43659226658283556,
      "grad_norm": 0.4277431070804596,
      "learning_rate": 1.253263270079327e-05,
      "loss": 0.8382,
      "step": 2604
    },
    {
      "epoch": 0.43675992874358166,
      "grad_norm": 0.49343574047088623,
      "learning_rate": 1.2527378795113339e-05,
      "loss": 0.8996,
      "step": 2605
    },
    {
      "epoch": 0.43692759090432776,
      "grad_norm": 0.3353070616722107,
      "learning_rate": 1.2522124144078103e-05,
      "loss": 0.9396,
      "step": 2606
    },
    {
      "epoch": 0.43709525306507385,
      "grad_norm": 0.37291547656059265,
      "learning_rate": 1.251686874923723e-05,
      "loss": 0.8347,
      "step": 2607
    },
    {
      "epoch": 0.43726291522581995,
      "grad_norm": 0.6479389667510986,
      "learning_rate": 1.251161261214059e-05,
      "loss": 0.7938,
      "step": 2608
    },
    {
      "epoch": 0.43743057738656604,
      "grad_norm": 0.5434373617172241,
      "learning_rate": 1.2506355734338293e-05,
      "loss": 0.8304,
      "step": 2609
    },
    {
      "epoch": 0.43759823954731214,
      "grad_norm": 0.3810223639011383,
      "learning_rate": 1.250109811738066e-05,
      "loss": 0.8859,
      "step": 2610
    },
    {
      "epoch": 0.43776590170805824,
      "grad_norm": 0.4282234311103821,
      "learning_rate": 1.2495839762818216e-05,
      "loss": 0.8614,
      "step": 2611
    },
    {
      "epoch": 0.43793356386880433,
      "grad_norm": 0.4012936055660248,
      "learning_rate": 1.249058067220172e-05,
      "loss": 0.8971,
      "step": 2612
    },
    {
      "epoch": 0.4381012260295505,
      "grad_norm": 0.3515467047691345,
      "learning_rate": 1.2485320847082143e-05,
      "loss": 0.9126,
      "step": 2613
    },
    {
      "epoch": 0.4382688881902966,
      "grad_norm": 0.4747259020805359,
      "learning_rate": 1.2480060289010677e-05,
      "loss": 0.9065,
      "step": 2614
    },
    {
      "epoch": 0.4384365503510427,
      "grad_norm": 0.5856126546859741,
      "learning_rate": 1.247479899953872e-05,
      "loss": 0.8844,
      "step": 2615
    },
    {
      "epoch": 0.43860421251178877,
      "grad_norm": 0.672063946723938,
      "learning_rate": 1.246953698021789e-05,
      "loss": 0.858,
      "step": 2616
    },
    {
      "epoch": 0.43877187467253487,
      "grad_norm": 0.5069361329078674,
      "learning_rate": 1.2464274232600029e-05,
      "loss": 0.8002,
      "step": 2617
    },
    {
      "epoch": 0.43893953683328096,
      "grad_norm": 0.5100276470184326,
      "learning_rate": 1.2459010758237178e-05,
      "loss": 0.8739,
      "step": 2618
    },
    {
      "epoch": 0.43910719899402706,
      "grad_norm": 0.6383470892906189,
      "learning_rate": 1.2453746558681607e-05,
      "loss": 0.9281,
      "step": 2619
    },
    {
      "epoch": 0.43927486115477316,
      "grad_norm": 0.38440290093421936,
      "learning_rate": 1.2448481635485792e-05,
      "loss": 0.8567,
      "step": 2620
    },
    {
      "epoch": 0.43944252331551925,
      "grad_norm": 0.3238489627838135,
      "learning_rate": 1.2443215990202424e-05,
      "loss": 0.8649,
      "step": 2621
    },
    {
      "epoch": 0.43961018547626535,
      "grad_norm": 0.49014559388160706,
      "learning_rate": 1.2437949624384405e-05,
      "loss": 0.8069,
      "step": 2622
    },
    {
      "epoch": 0.43977784763701144,
      "grad_norm": 0.8587152361869812,
      "learning_rate": 1.2432682539584856e-05,
      "loss": 0.8208,
      "step": 2623
    },
    {
      "epoch": 0.43994550979775754,
      "grad_norm": 0.8027268648147583,
      "learning_rate": 1.2427414737357097e-05,
      "loss": 0.9219,
      "step": 2624
    },
    {
      "epoch": 0.44011317195850364,
      "grad_norm": 0.44230809807777405,
      "learning_rate": 1.242214621925468e-05,
      "loss": 0.8819,
      "step": 2625
    },
    {
      "epoch": 0.44028083411924973,
      "grad_norm": 0.3373022973537445,
      "learning_rate": 1.2416876986831348e-05,
      "loss": 0.9194,
      "step": 2626
    },
    {
      "epoch": 0.4404484962799958,
      "grad_norm": 0.613947331905365,
      "learning_rate": 1.2411607041641062e-05,
      "loss": 0.9163,
      "step": 2627
    },
    {
      "epoch": 0.4406161584407419,
      "grad_norm": 0.42350468039512634,
      "learning_rate": 1.2406336385238e-05,
      "loss": 0.8645,
      "step": 2628
    },
    {
      "epoch": 0.440783820601488,
      "grad_norm": 0.4935706853866577,
      "learning_rate": 1.240106501917654e-05,
      "loss": 0.8883,
      "step": 2629
    },
    {
      "epoch": 0.4409514827622341,
      "grad_norm": 0.4629109799861908,
      "learning_rate": 1.2395792945011273e-05,
      "loss": 0.9361,
      "step": 2630
    },
    {
      "epoch": 0.4411191449229802,
      "grad_norm": 0.4974744915962219,
      "learning_rate": 1.2390520164297e-05,
      "loss": 0.9175,
      "step": 2631
    },
    {
      "epoch": 0.4412868070837263,
      "grad_norm": 0.6437855958938599,
      "learning_rate": 1.2385246678588732e-05,
      "loss": 0.899,
      "step": 2632
    },
    {
      "epoch": 0.4414544692444724,
      "grad_norm": 0.4009495675563812,
      "learning_rate": 1.2379972489441678e-05,
      "loss": 0.8831,
      "step": 2633
    },
    {
      "epoch": 0.4416221314052185,
      "grad_norm": 0.41146427392959595,
      "learning_rate": 1.237469759841127e-05,
      "loss": 0.8963,
      "step": 2634
    },
    {
      "epoch": 0.4417897935659646,
      "grad_norm": 0.4736334979534149,
      "learning_rate": 1.236942200705313e-05,
      "loss": 0.915,
      "step": 2635
    },
    {
      "epoch": 0.4419574557267107,
      "grad_norm": 0.3997778296470642,
      "learning_rate": 1.2364145716923103e-05,
      "loss": 0.8388,
      "step": 2636
    },
    {
      "epoch": 0.4421251178874568,
      "grad_norm": 0.38202333450317383,
      "learning_rate": 1.2358868729577232e-05,
      "loss": 0.8934,
      "step": 2637
    },
    {
      "epoch": 0.4422927800482029,
      "grad_norm": 0.5045069456100464,
      "learning_rate": 1.235359104657176e-05,
      "loss": 0.8547,
      "step": 2638
    },
    {
      "epoch": 0.442460442208949,
      "grad_norm": 0.5607749819755554,
      "learning_rate": 1.2348312669463145e-05,
      "loss": 0.8935,
      "step": 2639
    },
    {
      "epoch": 0.4426281043696951,
      "grad_norm": 0.631809413433075,
      "learning_rate": 1.2343033599808044e-05,
      "loss": 0.9419,
      "step": 2640
    },
    {
      "epoch": 0.44279576653044117,
      "grad_norm": 0.827242910861969,
      "learning_rate": 1.2337753839163324e-05,
      "loss": 0.8015,
      "step": 2641
    },
    {
      "epoch": 0.44296342869118727,
      "grad_norm": 0.39556291699409485,
      "learning_rate": 1.2332473389086043e-05,
      "loss": 0.9154,
      "step": 2642
    },
    {
      "epoch": 0.44313109085193336,
      "grad_norm": 1.0048776865005493,
      "learning_rate": 1.2327192251133482e-05,
      "loss": 0.9431,
      "step": 2643
    },
    {
      "epoch": 0.44329875301267946,
      "grad_norm": 0.6327106356620789,
      "learning_rate": 1.2321910426863108e-05,
      "loss": 0.8219,
      "step": 2644
    },
    {
      "epoch": 0.44346641517342555,
      "grad_norm": 0.6693361401557922,
      "learning_rate": 1.2316627917832598e-05,
      "loss": 0.9393,
      "step": 2645
    },
    {
      "epoch": 0.44363407733417165,
      "grad_norm": 0.5261967778205872,
      "learning_rate": 1.2311344725599828e-05,
      "loss": 0.8466,
      "step": 2646
    },
    {
      "epoch": 0.44380173949491775,
      "grad_norm": 0.5540294647216797,
      "learning_rate": 1.2306060851722877e-05,
      "loss": 0.8857,
      "step": 2647
    },
    {
      "epoch": 0.44396940165566384,
      "grad_norm": 0.44563156366348267,
      "learning_rate": 1.2300776297760026e-05,
      "loss": 0.7964,
      "step": 2648
    },
    {
      "epoch": 0.44413706381640994,
      "grad_norm": 1.2173652648925781,
      "learning_rate": 1.2295491065269753e-05,
      "loss": 0.9064,
      "step": 2649
    },
    {
      "epoch": 0.44430472597715603,
      "grad_norm": 0.48927780985832214,
      "learning_rate": 1.229020515581074e-05,
      "loss": 0.9365,
      "step": 2650
    },
    {
      "epoch": 0.44447238813790213,
      "grad_norm": 0.8384155035018921,
      "learning_rate": 1.2284918570941871e-05,
      "loss": 0.925,
      "step": 2651
    },
    {
      "epoch": 0.4446400502986482,
      "grad_norm": 0.5220843553543091,
      "learning_rate": 1.2279631312222218e-05,
      "loss": 0.8733,
      "step": 2652
    },
    {
      "epoch": 0.4448077124593943,
      "grad_norm": 1.3837053775787354,
      "learning_rate": 1.2274343381211067e-05,
      "loss": 0.9314,
      "step": 2653
    },
    {
      "epoch": 0.4449753746201404,
      "grad_norm": 0.49828994274139404,
      "learning_rate": 1.2269054779467884e-05,
      "loss": 0.8824,
      "step": 2654
    },
    {
      "epoch": 0.4451430367808865,
      "grad_norm": 0.4432735741138458,
      "learning_rate": 1.2263765508552354e-05,
      "loss": 1.0045,
      "step": 2655
    },
    {
      "epoch": 0.4453106989416326,
      "grad_norm": 0.4565088748931885,
      "learning_rate": 1.2258475570024343e-05,
      "loss": 0.8646,
      "step": 2656
    },
    {
      "epoch": 0.4454783611023787,
      "grad_norm": 0.587070643901825,
      "learning_rate": 1.2253184965443918e-05,
      "loss": 0.8346,
      "step": 2657
    },
    {
      "epoch": 0.4456460232631248,
      "grad_norm": 0.4128307104110718,
      "learning_rate": 1.2247893696371347e-05,
      "loss": 0.8834,
      "step": 2658
    },
    {
      "epoch": 0.4458136854238709,
      "grad_norm": 0.5621363520622253,
      "learning_rate": 1.224260176436709e-05,
      "loss": 0.8545,
      "step": 2659
    },
    {
      "epoch": 0.445981347584617,
      "grad_norm": 0.46466559171676636,
      "learning_rate": 1.2237309170991805e-05,
      "loss": 0.8809,
      "step": 2660
    },
    {
      "epoch": 0.4461490097453631,
      "grad_norm": 0.5122204422950745,
      "learning_rate": 1.223201591780634e-05,
      "loss": 0.8707,
      "step": 2661
    },
    {
      "epoch": 0.4463166719061092,
      "grad_norm": 0.5194559097290039,
      "learning_rate": 1.222672200637174e-05,
      "loss": 0.9194,
      "step": 2662
    },
    {
      "epoch": 0.4464843340668553,
      "grad_norm": 0.6348057985305786,
      "learning_rate": 1.222142743824925e-05,
      "loss": 0.7764,
      "step": 2663
    },
    {
      "epoch": 0.4466519962276014,
      "grad_norm": 0.4446607530117035,
      "learning_rate": 1.22161322150003e-05,
      "loss": 0.9271,
      "step": 2664
    },
    {
      "epoch": 0.4468196583883475,
      "grad_norm": 0.3675079345703125,
      "learning_rate": 1.2210836338186515e-05,
      "loss": 0.9728,
      "step": 2665
    },
    {
      "epoch": 0.44698732054909357,
      "grad_norm": 0.580569326877594,
      "learning_rate": 1.2205539809369719e-05,
      "loss": 0.8281,
      "step": 2666
    },
    {
      "epoch": 0.44715498270983967,
      "grad_norm": 0.6098851561546326,
      "learning_rate": 1.2200242630111926e-05,
      "loss": 0.9218,
      "step": 2667
    },
    {
      "epoch": 0.44732264487058576,
      "grad_norm": 0.8198440074920654,
      "learning_rate": 1.2194944801975334e-05,
      "loss": 0.8709,
      "step": 2668
    },
    {
      "epoch": 0.44749030703133186,
      "grad_norm": 0.6457405090332031,
      "learning_rate": 1.218964632652234e-05,
      "loss": 0.8942,
      "step": 2669
    },
    {
      "epoch": 0.44765796919207795,
      "grad_norm": 0.6705514788627625,
      "learning_rate": 1.218434720531553e-05,
      "loss": 0.878,
      "step": 2670
    },
    {
      "epoch": 0.44782563135282405,
      "grad_norm": 0.3943946063518524,
      "learning_rate": 1.2179047439917681e-05,
      "loss": 0.8365,
      "step": 2671
    },
    {
      "epoch": 0.44799329351357015,
      "grad_norm": 0.40450936555862427,
      "learning_rate": 1.217374703189176e-05,
      "loss": 0.9172,
      "step": 2672
    },
    {
      "epoch": 0.44816095567431624,
      "grad_norm": 0.5442905426025391,
      "learning_rate": 1.2168445982800923e-05,
      "loss": 0.8982,
      "step": 2673
    },
    {
      "epoch": 0.44832861783506234,
      "grad_norm": 0.3890281319618225,
      "learning_rate": 1.2163144294208511e-05,
      "loss": 0.7854,
      "step": 2674
    },
    {
      "epoch": 0.44849627999580843,
      "grad_norm": 0.41442999243736267,
      "learning_rate": 1.2157841967678064e-05,
      "loss": 0.8477,
      "step": 2675
    },
    {
      "epoch": 0.44866394215655453,
      "grad_norm": 0.6467515230178833,
      "learning_rate": 1.2152539004773298e-05,
      "loss": 0.9126,
      "step": 2676
    },
    {
      "epoch": 0.4488316043173006,
      "grad_norm": 0.6503413915634155,
      "learning_rate": 1.2147235407058124e-05,
      "loss": 0.8516,
      "step": 2677
    },
    {
      "epoch": 0.4489992664780467,
      "grad_norm": 0.4241797924041748,
      "learning_rate": 1.2141931176096643e-05,
      "loss": 0.8361,
      "step": 2678
    },
    {
      "epoch": 0.4491669286387928,
      "grad_norm": 0.4132331609725952,
      "learning_rate": 1.2136626313453136e-05,
      "loss": 0.8998,
      "step": 2679
    },
    {
      "epoch": 0.4493345907995389,
      "grad_norm": 0.5378803610801697,
      "learning_rate": 1.2131320820692067e-05,
      "loss": 0.8882,
      "step": 2680
    },
    {
      "epoch": 0.449502252960285,
      "grad_norm": 0.6029634475708008,
      "learning_rate": 1.21260146993781e-05,
      "loss": 0.8952,
      "step": 2681
    },
    {
      "epoch": 0.4496699151210311,
      "grad_norm": 0.5718348026275635,
      "learning_rate": 1.2120707951076069e-05,
      "loss": 0.9178,
      "step": 2682
    },
    {
      "epoch": 0.4498375772817772,
      "grad_norm": 0.839780330657959,
      "learning_rate": 1.2115400577351009e-05,
      "loss": 0.8232,
      "step": 2683
    },
    {
      "epoch": 0.4500052394425233,
      "grad_norm": 0.4166347086429596,
      "learning_rate": 1.2110092579768117e-05,
      "loss": 0.8531,
      "step": 2684
    },
    {
      "epoch": 0.4501729016032694,
      "grad_norm": 0.6671404242515564,
      "learning_rate": 1.2104783959892798e-05,
      "loss": 0.8739,
      "step": 2685
    },
    {
      "epoch": 0.4503405637640155,
      "grad_norm": 0.4798339307308197,
      "learning_rate": 1.2099474719290625e-05,
      "loss": 0.8764,
      "step": 2686
    },
    {
      "epoch": 0.4505082259247616,
      "grad_norm": 0.40089160203933716,
      "learning_rate": 1.2094164859527363e-05,
      "loss": 0.8791,
      "step": 2687
    },
    {
      "epoch": 0.4506758880855077,
      "grad_norm": 0.36743858456611633,
      "learning_rate": 1.2088854382168946e-05,
      "loss": 0.8389,
      "step": 2688
    },
    {
      "epoch": 0.4508435502462538,
      "grad_norm": 0.882380485534668,
      "learning_rate": 1.2083543288781512e-05,
      "loss": 0.8486,
      "step": 2689
    },
    {
      "epoch": 0.4510112124069999,
      "grad_norm": 0.5218948125839233,
      "learning_rate": 1.2078231580931359e-05,
      "loss": 0.8741,
      "step": 2690
    },
    {
      "epoch": 0.45117887456774597,
      "grad_norm": 0.5964364409446716,
      "learning_rate": 1.2072919260184977e-05,
      "loss": 0.9058,
      "step": 2691
    },
    {
      "epoch": 0.45134653672849206,
      "grad_norm": 0.43583714962005615,
      "learning_rate": 1.2067606328109038e-05,
      "loss": 0.8084,
      "step": 2692
    },
    {
      "epoch": 0.45151419888923816,
      "grad_norm": 0.5721093416213989,
      "learning_rate": 1.2062292786270391e-05,
      "loss": 0.7833,
      "step": 2693
    },
    {
      "epoch": 0.45168186104998426,
      "grad_norm": 0.43429702520370483,
      "learning_rate": 1.2056978636236065e-05,
      "loss": 0.8776,
      "step": 2694
    },
    {
      "epoch": 0.45184952321073035,
      "grad_norm": 0.5420209765434265,
      "learning_rate": 1.2051663879573265e-05,
      "loss": 0.9177,
      "step": 2695
    },
    {
      "epoch": 0.45201718537147645,
      "grad_norm": 0.5308378338813782,
      "learning_rate": 1.2046348517849384e-05,
      "loss": 0.812,
      "step": 2696
    },
    {
      "epoch": 0.45218484753222254,
      "grad_norm": 0.450838565826416,
      "learning_rate": 1.2041032552631984e-05,
      "loss": 0.8564,
      "step": 2697
    },
    {
      "epoch": 0.45235250969296864,
      "grad_norm": 0.5462943315505981,
      "learning_rate": 1.2035715985488814e-05,
      "loss": 0.8829,
      "step": 2698
    },
    {
      "epoch": 0.4525201718537148,
      "grad_norm": 0.41511619091033936,
      "learning_rate": 1.2030398817987785e-05,
      "loss": 0.8796,
      "step": 2699
    },
    {
      "epoch": 0.4526878340144609,
      "grad_norm": 0.4854419529438019,
      "learning_rate": 1.2025081051697007e-05,
      "loss": 0.8153,
      "step": 2700
    },
    {
      "epoch": 0.452855496175207,
      "grad_norm": 0.5956982374191284,
      "learning_rate": 1.2019762688184753e-05,
      "loss": 0.9383,
      "step": 2701
    },
    {
      "epoch": 0.4530231583359531,
      "grad_norm": 0.43768510222435,
      "learning_rate": 1.201444372901947e-05,
      "loss": 0.7995,
      "step": 2702
    },
    {
      "epoch": 0.4531908204966992,
      "grad_norm": 0.3518020212650299,
      "learning_rate": 1.2009124175769784e-05,
      "loss": 0.8851,
      "step": 2703
    },
    {
      "epoch": 0.45335848265744527,
      "grad_norm": 0.632449209690094,
      "learning_rate": 1.2003804030004502e-05,
      "loss": 0.861,
      "step": 2704
    },
    {
      "epoch": 0.45352614481819137,
      "grad_norm": 0.3550005853176117,
      "learning_rate": 1.1998483293292602e-05,
      "loss": 0.8596,
      "step": 2705
    },
    {
      "epoch": 0.45369380697893746,
      "grad_norm": 0.5096614360809326,
      "learning_rate": 1.1993161967203228e-05,
      "loss": 0.8511,
      "step": 2706
    },
    {
      "epoch": 0.45386146913968356,
      "grad_norm": 0.43639296293258667,
      "learning_rate": 1.1987840053305706e-05,
      "loss": 0.8645,
      "step": 2707
    },
    {
      "epoch": 0.45402913130042966,
      "grad_norm": 0.8187398314476013,
      "learning_rate": 1.1982517553169544e-05,
      "loss": 0.9201,
      "step": 2708
    },
    {
      "epoch": 0.45419679346117575,
      "grad_norm": 0.7677807807922363,
      "learning_rate": 1.1977194468364403e-05,
      "loss": 0.8638,
      "step": 2709
    },
    {
      "epoch": 0.45436445562192185,
      "grad_norm": 0.42025095224380493,
      "learning_rate": 1.1971870800460132e-05,
      "loss": 0.802,
      "step": 2710
    },
    {
      "epoch": 0.45453211778266794,
      "grad_norm": 0.43635404109954834,
      "learning_rate": 1.196654655102674e-05,
      "loss": 0.9023,
      "step": 2711
    },
    {
      "epoch": 0.45469977994341404,
      "grad_norm": 0.46622127294540405,
      "learning_rate": 1.1961221721634425e-05,
      "loss": 0.9021,
      "step": 2712
    },
    {
      "epoch": 0.45486744210416014,
      "grad_norm": 0.37527352571487427,
      "learning_rate": 1.1955896313853535e-05,
      "loss": 0.8711,
      "step": 2713
    },
    {
      "epoch": 0.45503510426490623,
      "grad_norm": 0.5640096068382263,
      "learning_rate": 1.1950570329254604e-05,
      "loss": 0.8739,
      "step": 2714
    },
    {
      "epoch": 0.4552027664256523,
      "grad_norm": 0.6676239967346191,
      "learning_rate": 1.1945243769408329e-05,
      "loss": 0.8787,
      "step": 2715
    },
    {
      "epoch": 0.4553704285863984,
      "grad_norm": 0.5405741333961487,
      "learning_rate": 1.1939916635885575e-05,
      "loss": 0.8961,
      "step": 2716
    },
    {
      "epoch": 0.4555380907471445,
      "grad_norm": 0.4485068917274475,
      "learning_rate": 1.193458893025739e-05,
      "loss": 0.8998,
      "step": 2717
    },
    {
      "epoch": 0.4557057529078906,
      "grad_norm": 0.5062191486358643,
      "learning_rate": 1.192926065409497e-05,
      "loss": 0.8388,
      "step": 2718
    },
    {
      "epoch": 0.4558734150686367,
      "grad_norm": 0.6288925409317017,
      "learning_rate": 1.1923931808969693e-05,
      "loss": 0.9418,
      "step": 2719
    },
    {
      "epoch": 0.4560410772293828,
      "grad_norm": 0.47147685289382935,
      "learning_rate": 1.1918602396453102e-05,
      "loss": 0.7866,
      "step": 2720
    },
    {
      "epoch": 0.4562087393901289,
      "grad_norm": 0.3272930383682251,
      "learning_rate": 1.191327241811691e-05,
      "loss": 0.8037,
      "step": 2721
    },
    {
      "epoch": 0.456376401550875,
      "grad_norm": 0.4322454333305359,
      "learning_rate": 1.1907941875532987e-05,
      "loss": 0.8208,
      "step": 2722
    },
    {
      "epoch": 0.4565440637116211,
      "grad_norm": 0.9034909605979919,
      "learning_rate": 1.1902610770273379e-05,
      "loss": 0.8868,
      "step": 2723
    },
    {
      "epoch": 0.4567117258723672,
      "grad_norm": 0.6835198402404785,
      "learning_rate": 1.1897279103910297e-05,
      "loss": 0.8235,
      "step": 2724
    },
    {
      "epoch": 0.4568793880331133,
      "grad_norm": 0.5388729572296143,
      "learning_rate": 1.1891946878016114e-05,
      "loss": 0.8918,
      "step": 2725
    },
    {
      "epoch": 0.4570470501938594,
      "grad_norm": 0.4670707881450653,
      "learning_rate": 1.1886614094163368e-05,
      "loss": 0.8109,
      "step": 2726
    },
    {
      "epoch": 0.4572147123546055,
      "grad_norm": 0.48207026720046997,
      "learning_rate": 1.1881280753924763e-05,
      "loss": 0.9336,
      "step": 2727
    },
    {
      "epoch": 0.4573823745153516,
      "grad_norm": 0.589199423789978,
      "learning_rate": 1.187594685887317e-05,
      "loss": 0.8118,
      "step": 2728
    },
    {
      "epoch": 0.45755003667609767,
      "grad_norm": 0.42270368337631226,
      "learning_rate": 1.1870612410581617e-05,
      "loss": 0.8729,
      "step": 2729
    },
    {
      "epoch": 0.45771769883684377,
      "grad_norm": 0.4935012459754944,
      "learning_rate": 1.18652774106233e-05,
      "loss": 0.8112,
      "step": 2730
    },
    {
      "epoch": 0.45788536099758986,
      "grad_norm": 0.4753749966621399,
      "learning_rate": 1.185994186057158e-05,
      "loss": 0.8847,
      "step": 2731
    },
    {
      "epoch": 0.45805302315833596,
      "grad_norm": 0.48952072858810425,
      "learning_rate": 1.185460576199997e-05,
      "loss": 0.9115,
      "step": 2732
    },
    {
      "epoch": 0.45822068531908206,
      "grad_norm": 0.4920048117637634,
      "learning_rate": 1.1849269116482162e-05,
      "loss": 0.9322,
      "step": 2733
    },
    {
      "epoch": 0.45838834747982815,
      "grad_norm": 0.6082614064216614,
      "learning_rate": 1.1843931925591987e-05,
      "loss": 0.8564,
      "step": 2734
    },
    {
      "epoch": 0.45855600964057425,
      "grad_norm": 0.48946818709373474,
      "learning_rate": 1.1838594190903456e-05,
      "loss": 0.8717,
      "step": 2735
    },
    {
      "epoch": 0.45872367180132034,
      "grad_norm": 0.6212282180786133,
      "learning_rate": 1.1833255913990732e-05,
      "loss": 0.9478,
      "step": 2736
    },
    {
      "epoch": 0.45889133396206644,
      "grad_norm": 0.4374319314956665,
      "learning_rate": 1.1827917096428137e-05,
      "loss": 0.855,
      "step": 2737
    },
    {
      "epoch": 0.45905899612281253,
      "grad_norm": 0.6202477812767029,
      "learning_rate": 1.1822577739790153e-05,
      "loss": 0.8663,
      "step": 2738
    },
    {
      "epoch": 0.45922665828355863,
      "grad_norm": 0.5401442646980286,
      "learning_rate": 1.1817237845651429e-05,
      "loss": 0.8311,
      "step": 2739
    },
    {
      "epoch": 0.4593943204443047,
      "grad_norm": 0.6633997559547424,
      "learning_rate": 1.1811897415586761e-05,
      "loss": 0.8092,
      "step": 2740
    },
    {
      "epoch": 0.4595619826050508,
      "grad_norm": 0.3960711658000946,
      "learning_rate": 1.1806556451171108e-05,
      "loss": 0.8179,
      "step": 2741
    },
    {
      "epoch": 0.4597296447657969,
      "grad_norm": 0.6744838356971741,
      "learning_rate": 1.1801214953979584e-05,
      "loss": 0.7789,
      "step": 2742
    },
    {
      "epoch": 0.459897306926543,
      "grad_norm": 0.45512571930885315,
      "learning_rate": 1.179587292558747e-05,
      "loss": 0.9245,
      "step": 2743
    },
    {
      "epoch": 0.4600649690872891,
      "grad_norm": 0.3947019875049591,
      "learning_rate": 1.1790530367570194e-05,
      "loss": 0.8824,
      "step": 2744
    },
    {
      "epoch": 0.4602326312480352,
      "grad_norm": 0.4425636827945709,
      "learning_rate": 1.1785187281503336e-05,
      "loss": 0.844,
      "step": 2745
    },
    {
      "epoch": 0.4604002934087813,
      "grad_norm": 0.5157086849212646,
      "learning_rate": 1.1779843668962647e-05,
      "loss": 0.9509,
      "step": 2746
    },
    {
      "epoch": 0.4605679555695274,
      "grad_norm": 0.4011687636375427,
      "learning_rate": 1.1774499531524018e-05,
      "loss": 0.9558,
      "step": 2747
    },
    {
      "epoch": 0.4607356177302735,
      "grad_norm": 0.5148869156837463,
      "learning_rate": 1.1769154870763504e-05,
      "loss": 0.8244,
      "step": 2748
    },
    {
      "epoch": 0.4609032798910196,
      "grad_norm": 0.6069275736808777,
      "learning_rate": 1.1763809688257313e-05,
      "loss": 0.9558,
      "step": 2749
    },
    {
      "epoch": 0.4610709420517657,
      "grad_norm": 0.5783160328865051,
      "learning_rate": 1.1758463985581803e-05,
      "loss": 0.8771,
      "step": 2750
    },
    {
      "epoch": 0.4612386042125118,
      "grad_norm": 0.36110037565231323,
      "learning_rate": 1.1753117764313491e-05,
      "loss": 0.8755,
      "step": 2751
    },
    {
      "epoch": 0.4614062663732579,
      "grad_norm": 0.3790871202945709,
      "learning_rate": 1.1747771026029038e-05,
      "loss": 0.7351,
      "step": 2752
    },
    {
      "epoch": 0.461573928534004,
      "grad_norm": 0.6918930411338806,
      "learning_rate": 1.1742423772305269e-05,
      "loss": 0.8708,
      "step": 2753
    },
    {
      "epoch": 0.46174159069475007,
      "grad_norm": 0.48895999789237976,
      "learning_rate": 1.1737076004719155e-05,
      "loss": 0.8933,
      "step": 2754
    },
    {
      "epoch": 0.46190925285549617,
      "grad_norm": 0.5753569006919861,
      "learning_rate": 1.1731727724847816e-05,
      "loss": 0.8738,
      "step": 2755
    },
    {
      "epoch": 0.46207691501624226,
      "grad_norm": 0.418861985206604,
      "learning_rate": 1.1726378934268525e-05,
      "loss": 0.8168,
      "step": 2756
    },
    {
      "epoch": 0.46224457717698836,
      "grad_norm": 0.4654471278190613,
      "learning_rate": 1.172102963455871e-05,
      "loss": 0.8289,
      "step": 2757
    },
    {
      "epoch": 0.46241223933773445,
      "grad_norm": 1.0062460899353027,
      "learning_rate": 1.1715679827295947e-05,
      "loss": 0.9493,
      "step": 2758
    },
    {
      "epoch": 0.46257990149848055,
      "grad_norm": 0.5639843940734863,
      "learning_rate": 1.1710329514057956e-05,
      "loss": 0.902,
      "step": 2759
    },
    {
      "epoch": 0.46274756365922665,
      "grad_norm": 0.5350931882858276,
      "learning_rate": 1.1704978696422614e-05,
      "loss": 0.9216,
      "step": 2760
    },
    {
      "epoch": 0.46291522581997274,
      "grad_norm": 0.489963561296463,
      "learning_rate": 1.169962737596794e-05,
      "loss": 0.7876,
      "step": 2761
    },
    {
      "epoch": 0.46308288798071884,
      "grad_norm": 0.4331090748310089,
      "learning_rate": 1.169427555427211e-05,
      "loss": 0.8466,
      "step": 2762
    },
    {
      "epoch": 0.46325055014146493,
      "grad_norm": 0.48761695623397827,
      "learning_rate": 1.1688923232913436e-05,
      "loss": 0.9871,
      "step": 2763
    },
    {
      "epoch": 0.46341821230221103,
      "grad_norm": 0.3835407793521881,
      "learning_rate": 1.1683570413470384e-05,
      "loss": 0.9213,
      "step": 2764
    },
    {
      "epoch": 0.4635858744629571,
      "grad_norm": 0.4346884787082672,
      "learning_rate": 1.1678217097521576e-05,
      "loss": 0.8505,
      "step": 2765
    },
    {
      "epoch": 0.4637535366237032,
      "grad_norm": 0.5936617255210876,
      "learning_rate": 1.1672863286645763e-05,
      "loss": 0.9202,
      "step": 2766
    },
    {
      "epoch": 0.4639211987844493,
      "grad_norm": 0.5256935954093933,
      "learning_rate": 1.1667508982421853e-05,
      "loss": 0.8315,
      "step": 2767
    },
    {
      "epoch": 0.4640888609451954,
      "grad_norm": 0.5747020244598389,
      "learning_rate": 1.1662154186428891e-05,
      "loss": 0.8775,
      "step": 2768
    },
    {
      "epoch": 0.4642565231059415,
      "grad_norm": 0.5799922347068787,
      "learning_rate": 1.1656798900246082e-05,
      "loss": 0.8667,
      "step": 2769
    },
    {
      "epoch": 0.4644241852666876,
      "grad_norm": 0.4185238778591156,
      "learning_rate": 1.165144312545276e-05,
      "loss": 0.8788,
      "step": 2770
    },
    {
      "epoch": 0.4645918474274337,
      "grad_norm": 0.4439437985420227,
      "learning_rate": 1.164608686362841e-05,
      "loss": 0.8039,
      "step": 2771
    },
    {
      "epoch": 0.4647595095881798,
      "grad_norm": 0.5659282803535461,
      "learning_rate": 1.1640730116352661e-05,
      "loss": 0.8726,
      "step": 2772
    },
    {
      "epoch": 0.4649271717489259,
      "grad_norm": 0.33008360862731934,
      "learning_rate": 1.1635372885205285e-05,
      "loss": 0.8077,
      "step": 2773
    },
    {
      "epoch": 0.465094833909672,
      "grad_norm": 0.4482969343662262,
      "learning_rate": 1.1630015171766196e-05,
      "loss": 0.8443,
      "step": 2774
    },
    {
      "epoch": 0.4652624960704181,
      "grad_norm": 0.42362096905708313,
      "learning_rate": 1.1624656977615445e-05,
      "loss": 0.7937,
      "step": 2775
    },
    {
      "epoch": 0.4654301582311642,
      "grad_norm": 0.8205183148384094,
      "learning_rate": 1.1619298304333236e-05,
      "loss": 0.8775,
      "step": 2776
    },
    {
      "epoch": 0.4655978203919103,
      "grad_norm": 0.46288955211639404,
      "learning_rate": 1.1613939153499903e-05,
      "loss": 0.8718,
      "step": 2777
    },
    {
      "epoch": 0.4657654825526564,
      "grad_norm": 0.6447383761405945,
      "learning_rate": 1.160857952669593e-05,
      "loss": 0.8223,
      "step": 2778
    },
    {
      "epoch": 0.46593314471340247,
      "grad_norm": 0.37369903922080994,
      "learning_rate": 1.1603219425501935e-05,
      "loss": 0.8591,
      "step": 2779
    },
    {
      "epoch": 0.46610080687414857,
      "grad_norm": 0.4112333655357361,
      "learning_rate": 1.159785885149868e-05,
      "loss": 0.9124,
      "step": 2780
    },
    {
      "epoch": 0.46626846903489466,
      "grad_norm": 0.33953040838241577,
      "learning_rate": 1.1592497806267064e-05,
      "loss": 0.8197,
      "step": 2781
    },
    {
      "epoch": 0.46643613119564076,
      "grad_norm": 0.44476771354675293,
      "learning_rate": 1.1587136291388126e-05,
      "loss": 0.9272,
      "step": 2782
    },
    {
      "epoch": 0.46660379335638685,
      "grad_norm": 0.6123186349868774,
      "learning_rate": 1.1581774308443042e-05,
      "loss": 0.8828,
      "step": 2783
    },
    {
      "epoch": 0.46677145551713295,
      "grad_norm": 0.7039437294006348,
      "learning_rate": 1.1576411859013125e-05,
      "loss": 0.8523,
      "step": 2784
    },
    {
      "epoch": 0.4669391176778791,
      "grad_norm": 0.5443616509437561,
      "learning_rate": 1.1571048944679835e-05,
      "loss": 0.8987,
      "step": 2785
    },
    {
      "epoch": 0.4671067798386252,
      "grad_norm": 0.5298369526863098,
      "learning_rate": 1.1565685567024754e-05,
      "loss": 0.8838,
      "step": 2786
    },
    {
      "epoch": 0.4672744419993713,
      "grad_norm": 0.4926570653915405,
      "learning_rate": 1.1560321727629614e-05,
      "loss": 0.8281,
      "step": 2787
    },
    {
      "epoch": 0.4674421041601174,
      "grad_norm": 0.34552261233329773,
      "learning_rate": 1.1554957428076272e-05,
      "loss": 0.8708,
      "step": 2788
    },
    {
      "epoch": 0.4676097663208635,
      "grad_norm": 0.546116054058075,
      "learning_rate": 1.1549592669946732e-05,
      "loss": 0.8396,
      "step": 2789
    },
    {
      "epoch": 0.4677774284816096,
      "grad_norm": 0.5592318773269653,
      "learning_rate": 1.1544227454823124e-05,
      "loss": 0.8501,
      "step": 2790
    },
    {
      "epoch": 0.4679450906423557,
      "grad_norm": 0.3868325352668762,
      "learning_rate": 1.1538861784287718e-05,
      "loss": 0.8726,
      "step": 2791
    },
    {
      "epoch": 0.4681127528031018,
      "grad_norm": 0.5497609972953796,
      "learning_rate": 1.1533495659922917e-05,
      "loss": 0.9093,
      "step": 2792
    },
    {
      "epoch": 0.46828041496384787,
      "grad_norm": 0.592467188835144,
      "learning_rate": 1.1528129083311257e-05,
      "loss": 0.9059,
      "step": 2793
    },
    {
      "epoch": 0.46844807712459396,
      "grad_norm": 0.45810258388519287,
      "learning_rate": 1.1522762056035407e-05,
      "loss": 0.9085,
      "step": 2794
    },
    {
      "epoch": 0.46861573928534006,
      "grad_norm": 0.44684118032455444,
      "learning_rate": 1.1517394579678171e-05,
      "loss": 0.8506,
      "step": 2795
    },
    {
      "epoch": 0.46878340144608616,
      "grad_norm": 0.5014858841896057,
      "learning_rate": 1.1512026655822483e-05,
      "loss": 0.8976,
      "step": 2796
    },
    {
      "epoch": 0.46895106360683225,
      "grad_norm": 0.37813401222229004,
      "learning_rate": 1.1506658286051415e-05,
      "loss": 0.8886,
      "step": 2797
    },
    {
      "epoch": 0.46911872576757835,
      "grad_norm": 0.48495879769325256,
      "learning_rate": 1.1501289471948158e-05,
      "loss": 0.8978,
      "step": 2798
    },
    {
      "epoch": 0.46928638792832444,
      "grad_norm": 0.528350830078125,
      "learning_rate": 1.1495920215096046e-05,
      "loss": 0.878,
      "step": 2799
    },
    {
      "epoch": 0.46945405008907054,
      "grad_norm": 0.7721688151359558,
      "learning_rate": 1.1490550517078538e-05,
      "loss": 0.8757,
      "step": 2800
    },
    {
      "epoch": 0.46962171224981664,
      "grad_norm": 0.5684508085250854,
      "learning_rate": 1.1485180379479229e-05,
      "loss": 0.8388,
      "step": 2801
    },
    {
      "epoch": 0.46978937441056273,
      "grad_norm": 0.5709981322288513,
      "learning_rate": 1.1479809803881829e-05,
      "loss": 1.0181,
      "step": 2802
    },
    {
      "epoch": 0.46995703657130883,
      "grad_norm": 0.4858846068382263,
      "learning_rate": 1.1474438791870201e-05,
      "loss": 0.8368,
      "step": 2803
    },
    {
      "epoch": 0.4701246987320549,
      "grad_norm": 0.4464353919029236,
      "learning_rate": 1.146906734502831e-05,
      "loss": 0.8901,
      "step": 2804
    },
    {
      "epoch": 0.470292360892801,
      "grad_norm": 0.47529369592666626,
      "learning_rate": 1.146369546494027e-05,
      "loss": 0.9403,
      "step": 2805
    },
    {
      "epoch": 0.4704600230535471,
      "grad_norm": 0.4896434545516968,
      "learning_rate": 1.1458323153190314e-05,
      "loss": 0.8692,
      "step": 2806
    },
    {
      "epoch": 0.4706276852142932,
      "grad_norm": 0.44182074069976807,
      "learning_rate": 1.14529504113628e-05,
      "loss": 0.8377,
      "step": 2807
    },
    {
      "epoch": 0.4707953473750393,
      "grad_norm": 0.495044469833374,
      "learning_rate": 1.144757724104222e-05,
      "loss": 0.8722,
      "step": 2808
    },
    {
      "epoch": 0.4709630095357854,
      "grad_norm": 0.6132462620735168,
      "learning_rate": 1.1442203643813184e-05,
      "loss": 0.8313,
      "step": 2809
    },
    {
      "epoch": 0.4711306716965315,
      "grad_norm": 0.726916491985321,
      "learning_rate": 1.1436829621260436e-05,
      "loss": 0.8299,
      "step": 2810
    },
    {
      "epoch": 0.4712983338572776,
      "grad_norm": 0.6654655933380127,
      "learning_rate": 1.1431455174968839e-05,
      "loss": 0.968,
      "step": 2811
    },
    {
      "epoch": 0.4714659960180237,
      "grad_norm": 0.5855336785316467,
      "learning_rate": 1.1426080306523388e-05,
      "loss": 0.8825,
      "step": 2812
    },
    {
      "epoch": 0.4716336581787698,
      "grad_norm": 0.5632380843162537,
      "learning_rate": 1.1420705017509197e-05,
      "loss": 0.8535,
      "step": 2813
    },
    {
      "epoch": 0.4718013203395159,
      "grad_norm": 0.4087710380554199,
      "learning_rate": 1.14153293095115e-05,
      "loss": 0.8571,
      "step": 2814
    },
    {
      "epoch": 0.471968982500262,
      "grad_norm": 0.4733235836029053,
      "learning_rate": 1.1409953184115668e-05,
      "loss": 0.9356,
      "step": 2815
    },
    {
      "epoch": 0.4721366446610081,
      "grad_norm": 0.7453212738037109,
      "learning_rate": 1.1404576642907179e-05,
      "loss": 0.8702,
      "step": 2816
    },
    {
      "epoch": 0.47230430682175417,
      "grad_norm": 0.3844403326511383,
      "learning_rate": 1.1399199687471647e-05,
      "loss": 0.8673,
      "step": 2817
    },
    {
      "epoch": 0.47247196898250027,
      "grad_norm": 0.615859866142273,
      "learning_rate": 1.13938223193948e-05,
      "loss": 0.8172,
      "step": 2818
    },
    {
      "epoch": 0.47263963114324636,
      "grad_norm": 0.43845224380493164,
      "learning_rate": 1.1388444540262496e-05,
      "loss": 0.8666,
      "step": 2819
    },
    {
      "epoch": 0.47280729330399246,
      "grad_norm": 0.38096320629119873,
      "learning_rate": 1.1383066351660698e-05,
      "loss": 0.8549,
      "step": 2820
    },
    {
      "epoch": 0.47297495546473856,
      "grad_norm": 0.974331259727478,
      "learning_rate": 1.1377687755175511e-05,
      "loss": 0.9511,
      "step": 2821
    },
    {
      "epoch": 0.47314261762548465,
      "grad_norm": 0.7325894236564636,
      "learning_rate": 1.1372308752393144e-05,
      "loss": 0.7975,
      "step": 2822
    },
    {
      "epoch": 0.47331027978623075,
      "grad_norm": 0.39797723293304443,
      "learning_rate": 1.1366929344899931e-05,
      "loss": 0.8754,
      "step": 2823
    },
    {
      "epoch": 0.47347794194697684,
      "grad_norm": 0.3541925251483917,
      "learning_rate": 1.1361549534282332e-05,
      "loss": 0.9109,
      "step": 2824
    },
    {
      "epoch": 0.47364560410772294,
      "grad_norm": 0.5293611288070679,
      "learning_rate": 1.1356169322126911e-05,
      "loss": 0.8164,
      "step": 2825
    },
    {
      "epoch": 0.47381326626846904,
      "grad_norm": 0.5768407583236694,
      "learning_rate": 1.135078871002037e-05,
      "loss": 0.9299,
      "step": 2826
    },
    {
      "epoch": 0.47398092842921513,
      "grad_norm": 0.46879512071609497,
      "learning_rate": 1.1345407699549504e-05,
      "loss": 0.8096,
      "step": 2827
    },
    {
      "epoch": 0.4741485905899612,
      "grad_norm": 0.3861348330974579,
      "learning_rate": 1.1340026292301252e-05,
      "loss": 0.8833,
      "step": 2828
    },
    {
      "epoch": 0.4743162527507073,
      "grad_norm": 0.42656075954437256,
      "learning_rate": 1.1334644489862652e-05,
      "loss": 0.8739,
      "step": 2829
    },
    {
      "epoch": 0.4744839149114534,
      "grad_norm": 0.3519597053527832,
      "learning_rate": 1.1329262293820863e-05,
      "loss": 0.839,
      "step": 2830
    },
    {
      "epoch": 0.4746515770721995,
      "grad_norm": 0.48209017515182495,
      "learning_rate": 1.132387970576317e-05,
      "loss": 0.9045,
      "step": 2831
    },
    {
      "epoch": 0.4748192392329456,
      "grad_norm": 0.7903998494148254,
      "learning_rate": 1.1318496727276953e-05,
      "loss": 0.7866,
      "step": 2832
    },
    {
      "epoch": 0.4749869013936917,
      "grad_norm": 0.45061150193214417,
      "learning_rate": 1.1313113359949726e-05,
      "loss": 0.8044,
      "step": 2833
    },
    {
      "epoch": 0.4751545635544378,
      "grad_norm": 0.4440174102783203,
      "learning_rate": 1.1307729605369109e-05,
      "loss": 0.942,
      "step": 2834
    },
    {
      "epoch": 0.4753222257151839,
      "grad_norm": 0.37487006187438965,
      "learning_rate": 1.1302345465122839e-05,
      "loss": 0.9064,
      "step": 2835
    },
    {
      "epoch": 0.47548988787593,
      "grad_norm": 0.6825331449508667,
      "learning_rate": 1.129696094079876e-05,
      "loss": 0.7952,
      "step": 2836
    },
    {
      "epoch": 0.4756575500366761,
      "grad_norm": 0.5929781198501587,
      "learning_rate": 1.1291576033984844e-05,
      "loss": 0.8484,
      "step": 2837
    },
    {
      "epoch": 0.4758252121974222,
      "grad_norm": 0.568448543548584,
      "learning_rate": 1.1286190746269164e-05,
      "loss": 0.8992,
      "step": 2838
    },
    {
      "epoch": 0.4759928743581683,
      "grad_norm": 0.6622653007507324,
      "learning_rate": 1.1280805079239903e-05,
      "loss": 0.9454,
      "step": 2839
    },
    {
      "epoch": 0.4761605365189144,
      "grad_norm": 0.35762348771095276,
      "learning_rate": 1.1275419034485365e-05,
      "loss": 0.7527,
      "step": 2840
    },
    {
      "epoch": 0.4763281986796605,
      "grad_norm": 0.5092772245407104,
      "learning_rate": 1.127003261359396e-05,
      "loss": 0.8905,
      "step": 2841
    },
    {
      "epoch": 0.47649586084040657,
      "grad_norm": 0.5554852485656738,
      "learning_rate": 1.1264645818154211e-05,
      "loss": 0.9333,
      "step": 2842
    },
    {
      "epoch": 0.47666352300115267,
      "grad_norm": 0.40020039677619934,
      "learning_rate": 1.125925864975475e-05,
      "loss": 0.8556,
      "step": 2843
    },
    {
      "epoch": 0.47683118516189876,
      "grad_norm": 0.5516888499259949,
      "learning_rate": 1.125387110998432e-05,
      "loss": 0.8682,
      "step": 2844
    },
    {
      "epoch": 0.47699884732264486,
      "grad_norm": 0.6319754719734192,
      "learning_rate": 1.1248483200431774e-05,
      "loss": 0.8332,
      "step": 2845
    },
    {
      "epoch": 0.47716650948339095,
      "grad_norm": 0.4318239688873291,
      "learning_rate": 1.124309492268607e-05,
      "loss": 0.7839,
      "step": 2846
    },
    {
      "epoch": 0.47733417164413705,
      "grad_norm": 0.3661859929561615,
      "learning_rate": 1.1237706278336284e-05,
      "loss": 0.8549,
      "step": 2847
    },
    {
      "epoch": 0.47750183380488315,
      "grad_norm": 0.3942444622516632,
      "learning_rate": 1.1232317268971586e-05,
      "loss": 0.8146,
      "step": 2848
    },
    {
      "epoch": 0.47766949596562924,
      "grad_norm": 0.570182740688324,
      "learning_rate": 1.1226927896181271e-05,
      "loss": 0.8495,
      "step": 2849
    },
    {
      "epoch": 0.47783715812637534,
      "grad_norm": 0.46788638830184937,
      "learning_rate": 1.1221538161554724e-05,
      "loss": 0.8217,
      "step": 2850
    },
    {
      "epoch": 0.47800482028712143,
      "grad_norm": 0.403129518032074,
      "learning_rate": 1.1216148066681448e-05,
      "loss": 0.8135,
      "step": 2851
    },
    {
      "epoch": 0.47817248244786753,
      "grad_norm": 0.4809132516384125,
      "learning_rate": 1.1210757613151044e-05,
      "loss": 0.7873,
      "step": 2852
    },
    {
      "epoch": 0.4783401446086136,
      "grad_norm": 0.7033389806747437,
      "learning_rate": 1.1205366802553231e-05,
      "loss": 0.8182,
      "step": 2853
    },
    {
      "epoch": 0.4785078067693597,
      "grad_norm": 0.4435984790325165,
      "learning_rate": 1.1199975636477822e-05,
      "loss": 0.9731,
      "step": 2854
    },
    {
      "epoch": 0.4786754689301058,
      "grad_norm": 0.34627559781074524,
      "learning_rate": 1.119458411651474e-05,
      "loss": 0.9428,
      "step": 2855
    },
    {
      "epoch": 0.4788431310908519,
      "grad_norm": 0.4384484887123108,
      "learning_rate": 1.1189192244254006e-05,
      "loss": 0.8891,
      "step": 2856
    },
    {
      "epoch": 0.479010793251598,
      "grad_norm": 0.49134066700935364,
      "learning_rate": 1.1183800021285756e-05,
      "loss": 0.8339,
      "step": 2857
    },
    {
      "epoch": 0.4791784554123441,
      "grad_norm": 0.6400789618492126,
      "learning_rate": 1.1178407449200222e-05,
      "loss": 0.8377,
      "step": 2858
    },
    {
      "epoch": 0.4793461175730902,
      "grad_norm": 0.5748932957649231,
      "learning_rate": 1.1173014529587734e-05,
      "loss": 0.8434,
      "step": 2859
    },
    {
      "epoch": 0.4795137797338363,
      "grad_norm": 0.6503258347511292,
      "learning_rate": 1.1167621264038742e-05,
      "loss": 0.9055,
      "step": 2860
    },
    {
      "epoch": 0.4796814418945824,
      "grad_norm": 0.43417903780937195,
      "learning_rate": 1.1162227654143777e-05,
      "loss": 0.8923,
      "step": 2861
    },
    {
      "epoch": 0.4798491040553285,
      "grad_norm": 0.4653598368167877,
      "learning_rate": 1.1156833701493486e-05,
      "loss": 0.8268,
      "step": 2862
    },
    {
      "epoch": 0.4800167662160746,
      "grad_norm": 0.665627658367157,
      "learning_rate": 1.1151439407678612e-05,
      "loss": 0.8754,
      "step": 2863
    },
    {
      "epoch": 0.4801844283768207,
      "grad_norm": 0.46827465295791626,
      "learning_rate": 1.1146044774289997e-05,
      "loss": 0.8615,
      "step": 2864
    },
    {
      "epoch": 0.4803520905375668,
      "grad_norm": 0.4687343239784241,
      "learning_rate": 1.1140649802918591e-05,
      "loss": 0.9117,
      "step": 2865
    },
    {
      "epoch": 0.4805197526983129,
      "grad_norm": 0.4067821502685547,
      "learning_rate": 1.113525449515543e-05,
      "loss": 0.8897,
      "step": 2866
    },
    {
      "epoch": 0.48068741485905897,
      "grad_norm": 0.5301191210746765,
      "learning_rate": 1.1129858852591662e-05,
      "loss": 0.8381,
      "step": 2867
    },
    {
      "epoch": 0.48085507701980507,
      "grad_norm": 0.5383304953575134,
      "learning_rate": 1.1124462876818526e-05,
      "loss": 0.8658,
      "step": 2868
    },
    {
      "epoch": 0.48102273918055116,
      "grad_norm": 0.34599506855010986,
      "learning_rate": 1.1119066569427366e-05,
      "loss": 0.8396,
      "step": 2869
    },
    {
      "epoch": 0.4811904013412973,
      "grad_norm": 0.40095260739326477,
      "learning_rate": 1.1113669932009619e-05,
      "loss": 0.824,
      "step": 2870
    },
    {
      "epoch": 0.4813580635020434,
      "grad_norm": 0.5407511591911316,
      "learning_rate": 1.1108272966156812e-05,
      "loss": 0.7932,
      "step": 2871
    },
    {
      "epoch": 0.4815257256627895,
      "grad_norm": 0.3525823652744293,
      "learning_rate": 1.1102875673460592e-05,
      "loss": 0.8249,
      "step": 2872
    },
    {
      "epoch": 0.4816933878235356,
      "grad_norm": 0.4650687873363495,
      "learning_rate": 1.1097478055512674e-05,
      "loss": 0.8797,
      "step": 2873
    },
    {
      "epoch": 0.4818610499842817,
      "grad_norm": 0.5316118001937866,
      "learning_rate": 1.1092080113904886e-05,
      "loss": 0.8579,
      "step": 2874
    },
    {
      "epoch": 0.4820287121450278,
      "grad_norm": 0.6918269395828247,
      "learning_rate": 1.1086681850229153e-05,
      "loss": 0.8817,
      "step": 2875
    },
    {
      "epoch": 0.4821963743057739,
      "grad_norm": 0.40541380643844604,
      "learning_rate": 1.1081283266077488e-05,
      "loss": 0.8493,
      "step": 2876
    },
    {
      "epoch": 0.48236403646652,
      "grad_norm": 0.5759273767471313,
      "learning_rate": 1.1075884363041994e-05,
      "loss": 0.9329,
      "step": 2877
    },
    {
      "epoch": 0.4825316986272661,
      "grad_norm": 0.739473283290863,
      "learning_rate": 1.107048514271488e-05,
      "loss": 0.991,
      "step": 2878
    },
    {
      "epoch": 0.4826993607880122,
      "grad_norm": 0.5789638757705688,
      "learning_rate": 1.1065085606688442e-05,
      "loss": 0.8588,
      "step": 2879
    },
    {
      "epoch": 0.4828670229487583,
      "grad_norm": 0.5917166471481323,
      "learning_rate": 1.1059685756555069e-05,
      "loss": 0.901,
      "step": 2880
    },
    {
      "epoch": 0.48303468510950437,
      "grad_norm": 0.42208248376846313,
      "learning_rate": 1.1054285593907248e-05,
      "loss": 0.8927,
      "step": 2881
    },
    {
      "epoch": 0.48320234727025047,
      "grad_norm": 0.6999191641807556,
      "learning_rate": 1.1048885120337546e-05,
      "loss": 0.938,
      "step": 2882
    },
    {
      "epoch": 0.48337000943099656,
      "grad_norm": 0.6355289816856384,
      "learning_rate": 1.1043484337438637e-05,
      "loss": 0.9418,
      "step": 2883
    },
    {
      "epoch": 0.48353767159174266,
      "grad_norm": 0.566191554069519,
      "learning_rate": 1.1038083246803275e-05,
      "loss": 0.9113,
      "step": 2884
    },
    {
      "epoch": 0.48370533375248875,
      "grad_norm": 0.7057574391365051,
      "learning_rate": 1.1032681850024307e-05,
      "loss": 0.7919,
      "step": 2885
    },
    {
      "epoch": 0.48387299591323485,
      "grad_norm": 0.4619996249675751,
      "learning_rate": 1.1027280148694676e-05,
      "loss": 0.8256,
      "step": 2886
    },
    {
      "epoch": 0.48404065807398095,
      "grad_norm": 0.941382646560669,
      "learning_rate": 1.1021878144407408e-05,
      "loss": 0.9087,
      "step": 2887
    },
    {
      "epoch": 0.48420832023472704,
      "grad_norm": 0.39337483048439026,
      "learning_rate": 1.1016475838755623e-05,
      "loss": 0.8315,
      "step": 2888
    },
    {
      "epoch": 0.48437598239547314,
      "grad_norm": 0.40230581164360046,
      "learning_rate": 1.1011073233332525e-05,
      "loss": 0.8307,
      "step": 2889
    },
    {
      "epoch": 0.48454364455621923,
      "grad_norm": 0.6783583760261536,
      "learning_rate": 1.100567032973141e-05,
      "loss": 0.9285,
      "step": 2890
    },
    {
      "epoch": 0.48471130671696533,
      "grad_norm": 0.5566238164901733,
      "learning_rate": 1.1000267129545661e-05,
      "loss": 0.7715,
      "step": 2891
    },
    {
      "epoch": 0.4848789688777114,
      "grad_norm": 0.6129046082496643,
      "learning_rate": 1.0994863634368754e-05,
      "loss": 0.8487,
      "step": 2892
    },
    {
      "epoch": 0.4850466310384575,
      "grad_norm": 0.5205916166305542,
      "learning_rate": 1.0989459845794234e-05,
      "loss": 0.9206,
      "step": 2893
    },
    {
      "epoch": 0.4852142931992036,
      "grad_norm": 0.6074056625366211,
      "learning_rate": 1.0984055765415757e-05,
      "loss": 0.7813,
      "step": 2894
    },
    {
      "epoch": 0.4853819553599497,
      "grad_norm": 1.0004022121429443,
      "learning_rate": 1.0978651394827054e-05,
      "loss": 0.8608,
      "step": 2895
    },
    {
      "epoch": 0.4855496175206958,
      "grad_norm": 0.9308025240898132,
      "learning_rate": 1.0973246735621932e-05,
      "loss": 0.8962,
      "step": 2896
    },
    {
      "epoch": 0.4857172796814419,
      "grad_norm": 0.5531139969825745,
      "learning_rate": 1.0967841789394296e-05,
      "loss": 0.8099,
      "step": 2897
    },
    {
      "epoch": 0.485884941842188,
      "grad_norm": 0.6492610573768616,
      "learning_rate": 1.0962436557738131e-05,
      "loss": 0.79,
      "step": 2898
    },
    {
      "epoch": 0.4860526040029341,
      "grad_norm": 0.557978630065918,
      "learning_rate": 1.095703104224751e-05,
      "loss": 0.7553,
      "step": 2899
    },
    {
      "epoch": 0.4862202661636802,
      "grad_norm": 0.5231502652168274,
      "learning_rate": 1.0951625244516584e-05,
      "loss": 0.898,
      "step": 2900
    },
    {
      "epoch": 0.4863879283244263,
      "grad_norm": 0.8099695444107056,
      "learning_rate": 1.0946219166139587e-05,
      "loss": 0.8512,
      "step": 2901
    },
    {
      "epoch": 0.4865555904851724,
      "grad_norm": 0.5336698293685913,
      "learning_rate": 1.094081280871084e-05,
      "loss": 0.8892,
      "step": 2902
    },
    {
      "epoch": 0.4867232526459185,
      "grad_norm": 0.3855321705341339,
      "learning_rate": 1.0935406173824749e-05,
      "loss": 0.8501,
      "step": 2903
    },
    {
      "epoch": 0.4868909148066646,
      "grad_norm": 0.3807976245880127,
      "learning_rate": 1.0929999263075795e-05,
      "loss": 0.8586,
      "step": 2904
    },
    {
      "epoch": 0.4870585769674107,
      "grad_norm": 1.0085023641586304,
      "learning_rate": 1.0924592078058538e-05,
      "loss": 0.9726,
      "step": 2905
    },
    {
      "epoch": 0.48722623912815677,
      "grad_norm": 0.4902290105819702,
      "learning_rate": 1.0919184620367634e-05,
      "loss": 0.8192,
      "step": 2906
    },
    {
      "epoch": 0.48739390128890286,
      "grad_norm": 0.5184853672981262,
      "learning_rate": 1.0913776891597798e-05,
      "loss": 0.9842,
      "step": 2907
    },
    {
      "epoch": 0.48756156344964896,
      "grad_norm": 0.6011456847190857,
      "learning_rate": 1.0908368893343843e-05,
      "loss": 0.8141,
      "step": 2908
    },
    {
      "epoch": 0.48772922561039506,
      "grad_norm": 0.4382106065750122,
      "learning_rate": 1.0902960627200655e-05,
      "loss": 0.851,
      "step": 2909
    },
    {
      "epoch": 0.48789688777114115,
      "grad_norm": 0.47010084986686707,
      "learning_rate": 1.0897552094763196e-05,
      "loss": 0.9802,
      "step": 2910
    },
    {
      "epoch": 0.48806454993188725,
      "grad_norm": 0.46682846546173096,
      "learning_rate": 1.0892143297626511e-05,
      "loss": 0.8456,
      "step": 2911
    },
    {
      "epoch": 0.48823221209263334,
      "grad_norm": 0.5086984634399414,
      "learning_rate": 1.0886734237385718e-05,
      "loss": 0.8319,
      "step": 2912
    },
    {
      "epoch": 0.48839987425337944,
      "grad_norm": 0.4189518392086029,
      "learning_rate": 1.088132491563602e-05,
      "loss": 0.8569,
      "step": 2913
    },
    {
      "epoch": 0.48856753641412554,
      "grad_norm": 0.4165961742401123,
      "learning_rate": 1.0875915333972689e-05,
      "loss": 0.8158,
      "step": 2914
    },
    {
      "epoch": 0.48873519857487163,
      "grad_norm": 0.6751852035522461,
      "learning_rate": 1.0870505493991084e-05,
      "loss": 0.8983,
      "step": 2915
    },
    {
      "epoch": 0.48890286073561773,
      "grad_norm": 0.4144243001937866,
      "learning_rate": 1.0865095397286623e-05,
      "loss": 0.8411,
      "step": 2916
    },
    {
      "epoch": 0.4890705228963638,
      "grad_norm": 0.6219884753227234,
      "learning_rate": 1.0859685045454821e-05,
      "loss": 0.8788,
      "step": 2917
    },
    {
      "epoch": 0.4892381850571099,
      "grad_norm": 0.4056248366832733,
      "learning_rate": 1.0854274440091254e-05,
      "loss": 0.8773,
      "step": 2918
    },
    {
      "epoch": 0.489405847217856,
      "grad_norm": 0.47027096152305603,
      "learning_rate": 1.0848863582791576e-05,
      "loss": 0.8727,
      "step": 2919
    },
    {
      "epoch": 0.4895735093786021,
      "grad_norm": 0.9668684601783752,
      "learning_rate": 1.0843452475151515e-05,
      "loss": 0.9114,
      "step": 2920
    },
    {
      "epoch": 0.4897411715393482,
      "grad_norm": 0.39250072836875916,
      "learning_rate": 1.0838041118766876e-05,
      "loss": 0.8781,
      "step": 2921
    },
    {
      "epoch": 0.4899088337000943,
      "grad_norm": 0.8956718444824219,
      "learning_rate": 1.0832629515233534e-05,
      "loss": 0.9103,
      "step": 2922
    },
    {
      "epoch": 0.4900764958608404,
      "grad_norm": 0.862220823764801,
      "learning_rate": 1.0827217666147435e-05,
      "loss": 0.8376,
      "step": 2923
    },
    {
      "epoch": 0.4902441580215865,
      "grad_norm": 0.46078476309776306,
      "learning_rate": 1.0821805573104607e-05,
      "loss": 0.9565,
      "step": 2924
    },
    {
      "epoch": 0.4904118201823326,
      "grad_norm": 0.38189762830734253,
      "learning_rate": 1.0816393237701134e-05,
      "loss": 0.8986,
      "step": 2925
    },
    {
      "epoch": 0.4905794823430787,
      "grad_norm": 0.46521735191345215,
      "learning_rate": 1.081098066153319e-05,
      "loss": 0.9397,
      "step": 2926
    },
    {
      "epoch": 0.4907471445038248,
      "grad_norm": 0.41276007890701294,
      "learning_rate": 1.0805567846197007e-05,
      "loss": 0.8809,
      "step": 2927
    },
    {
      "epoch": 0.4909148066645709,
      "grad_norm": 0.36183953285217285,
      "learning_rate": 1.0800154793288888e-05,
      "loss": 0.8985,
      "step": 2928
    },
    {
      "epoch": 0.491082468825317,
      "grad_norm": 0.592648983001709,
      "learning_rate": 1.0794741504405215e-05,
      "loss": 0.921,
      "step": 2929
    },
    {
      "epoch": 0.49125013098606307,
      "grad_norm": 0.7352839708328247,
      "learning_rate": 1.078932798114243e-05,
      "loss": 0.8357,
      "step": 2930
    },
    {
      "epoch": 0.49141779314680917,
      "grad_norm": 0.39497673511505127,
      "learning_rate": 1.078391422509705e-05,
      "loss": 0.7776,
      "step": 2931
    },
    {
      "epoch": 0.49158545530755526,
      "grad_norm": 0.4355677366256714,
      "learning_rate": 1.0778500237865656e-05,
      "loss": 0.87,
      "step": 2932
    },
    {
      "epoch": 0.49175311746830136,
      "grad_norm": 0.403594434261322,
      "learning_rate": 1.0773086021044905e-05,
      "loss": 0.8779,
      "step": 2933
    },
    {
      "epoch": 0.49192077962904746,
      "grad_norm": 0.4891184866428375,
      "learning_rate": 1.0767671576231512e-05,
      "loss": 0.8959,
      "step": 2934
    },
    {
      "epoch": 0.49208844178979355,
      "grad_norm": 0.39034304022789,
      "learning_rate": 1.0762256905022266e-05,
      "loss": 0.8926,
      "step": 2935
    },
    {
      "epoch": 0.49225610395053965,
      "grad_norm": 0.5464780330657959,
      "learning_rate": 1.0756842009014016e-05,
      "loss": 0.8915,
      "step": 2936
    },
    {
      "epoch": 0.49242376611128574,
      "grad_norm": 0.7192713618278503,
      "learning_rate": 1.0751426889803689e-05,
      "loss": 0.8104,
      "step": 2937
    },
    {
      "epoch": 0.49259142827203184,
      "grad_norm": 0.3802322447299957,
      "learning_rate": 1.0746011548988269e-05,
      "loss": 0.853,
      "step": 2938
    },
    {
      "epoch": 0.49275909043277794,
      "grad_norm": 0.6255716681480408,
      "learning_rate": 1.07405959881648e-05,
      "loss": 0.9386,
      "step": 2939
    },
    {
      "epoch": 0.49292675259352403,
      "grad_norm": 0.5813717842102051,
      "learning_rate": 1.073518020893041e-05,
      "loss": 0.9199,
      "step": 2940
    },
    {
      "epoch": 0.4930944147542701,
      "grad_norm": 0.6427274942398071,
      "learning_rate": 1.0729764212882268e-05,
      "loss": 0.696,
      "step": 2941
    },
    {
      "epoch": 0.4932620769150162,
      "grad_norm": 0.34310704469680786,
      "learning_rate": 1.0724348001617626e-05,
      "loss": 0.918,
      "step": 2942
    },
    {
      "epoch": 0.4934297390757623,
      "grad_norm": 0.5016130805015564,
      "learning_rate": 1.0718931576733787e-05,
      "loss": 0.8238,
      "step": 2943
    },
    {
      "epoch": 0.4935974012365084,
      "grad_norm": 0.3720301687717438,
      "learning_rate": 1.0713514939828122e-05,
      "loss": 0.7974,
      "step": 2944
    },
    {
      "epoch": 0.4937650633972545,
      "grad_norm": 0.4041135311126709,
      "learning_rate": 1.0708098092498071e-05,
      "loss": 0.8674,
      "step": 2945
    },
    {
      "epoch": 0.4939327255580006,
      "grad_norm": 0.47650477290153503,
      "learning_rate": 1.070268103634112e-05,
      "loss": 0.8526,
      "step": 2946
    },
    {
      "epoch": 0.4941003877187467,
      "grad_norm": 0.5747629404067993,
      "learning_rate": 1.0697263772954831e-05,
      "loss": 0.8781,
      "step": 2947
    },
    {
      "epoch": 0.4942680498794928,
      "grad_norm": 0.49312344193458557,
      "learning_rate": 1.0691846303936822e-05,
      "loss": 0.8176,
      "step": 2948
    },
    {
      "epoch": 0.4944357120402389,
      "grad_norm": 0.5730398893356323,
      "learning_rate": 1.0686428630884774e-05,
      "loss": 0.8855,
      "step": 2949
    },
    {
      "epoch": 0.494603374200985,
      "grad_norm": 0.4680347144603729,
      "learning_rate": 1.0681010755396417e-05,
      "loss": 0.9677,
      "step": 2950
    },
    {
      "epoch": 0.4947710363617311,
      "grad_norm": 0.4128454327583313,
      "learning_rate": 1.0675592679069558e-05,
      "loss": 0.8525,
      "step": 2951
    },
    {
      "epoch": 0.4949386985224772,
      "grad_norm": 0.4013599157333374,
      "learning_rate": 1.0670174403502051e-05,
      "loss": 0.8549,
      "step": 2952
    },
    {
      "epoch": 0.4951063606832233,
      "grad_norm": 0.4626038670539856,
      "learning_rate": 1.0664755930291817e-05,
      "loss": 0.8399,
      "step": 2953
    },
    {
      "epoch": 0.4952740228439694,
      "grad_norm": 0.3847028911113739,
      "learning_rate": 1.0659337261036824e-05,
      "loss": 0.9009,
      "step": 2954
    },
    {
      "epoch": 0.49544168500471547,
      "grad_norm": 0.6146700382232666,
      "learning_rate": 1.065391839733511e-05,
      "loss": 0.9276,
      "step": 2955
    },
    {
      "epoch": 0.4956093471654616,
      "grad_norm": 0.7219106554985046,
      "learning_rate": 1.0648499340784765e-05,
      "loss": 0.8842,
      "step": 2956
    },
    {
      "epoch": 0.4957770093262077,
      "grad_norm": 0.5040003657341003,
      "learning_rate": 1.0643080092983933e-05,
      "loss": 0.9594,
      "step": 2957
    },
    {
      "epoch": 0.4959446714869538,
      "grad_norm": 0.4865266978740692,
      "learning_rate": 1.0637660655530814e-05,
      "loss": 0.8077,
      "step": 2958
    },
    {
      "epoch": 0.4961123336476999,
      "grad_norm": 0.46663138270378113,
      "learning_rate": 1.0632241030023678e-05,
      "loss": 0.9022,
      "step": 2959
    },
    {
      "epoch": 0.496279995808446,
      "grad_norm": 0.4126185178756714,
      "learning_rate": 1.062682121806083e-05,
      "loss": 0.8231,
      "step": 2960
    },
    {
      "epoch": 0.4964476579691921,
      "grad_norm": 0.6498703956604004,
      "learning_rate": 1.0621401221240646e-05,
      "loss": 0.9935,
      "step": 2961
    },
    {
      "epoch": 0.4966153201299382,
      "grad_norm": 0.44782713055610657,
      "learning_rate": 1.061598104116154e-05,
      "loss": 0.9894,
      "step": 2962
    },
    {
      "epoch": 0.4967829822906843,
      "grad_norm": 0.5257605910301208,
      "learning_rate": 1.0610560679422007e-05,
      "loss": 0.8628,
      "step": 2963
    },
    {
      "epoch": 0.4969506444514304,
      "grad_norm": 0.40351545810699463,
      "learning_rate": 1.0605140137620562e-05,
      "loss": 0.9404,
      "step": 2964
    },
    {
      "epoch": 0.4971183066121765,
      "grad_norm": 0.3297235369682312,
      "learning_rate": 1.0599719417355801e-05,
      "loss": 0.8571,
      "step": 2965
    },
    {
      "epoch": 0.4972859687729226,
      "grad_norm": 0.41913363337516785,
      "learning_rate": 1.0594298520226354e-05,
      "loss": 0.8634,
      "step": 2966
    },
    {
      "epoch": 0.4974536309336687,
      "grad_norm": 0.7649381756782532,
      "learning_rate": 1.0588877447830917e-05,
      "loss": 0.8632,
      "step": 2967
    },
    {
      "epoch": 0.4976212930944148,
      "grad_norm": 0.4903692901134491,
      "learning_rate": 1.058345620176823e-05,
      "loss": 1.0128,
      "step": 2968
    },
    {
      "epoch": 0.49778895525516087,
      "grad_norm": 0.4322664439678192,
      "learning_rate": 1.057803478363708e-05,
      "loss": 0.9595,
      "step": 2969
    },
    {
      "epoch": 0.49795661741590697,
      "grad_norm": 0.4703931510448456,
      "learning_rate": 1.0572613195036319e-05,
      "loss": 0.924,
      "step": 2970
    },
    {
      "epoch": 0.49812427957665306,
      "grad_norm": 0.4785899817943573,
      "learning_rate": 1.0567191437564832e-05,
      "loss": 0.9254,
      "step": 2971
    },
    {
      "epoch": 0.49829194173739916,
      "grad_norm": 0.7141011953353882,
      "learning_rate": 1.0561769512821571e-05,
      "loss": 0.8905,
      "step": 2972
    },
    {
      "epoch": 0.49845960389814525,
      "grad_norm": 0.4536585807800293,
      "learning_rate": 1.055634742240552e-05,
      "loss": 0.779,
      "step": 2973
    },
    {
      "epoch": 0.49862726605889135,
      "grad_norm": 0.478220134973526,
      "learning_rate": 1.055092516791573e-05,
      "loss": 0.8576,
      "step": 2974
    },
    {
      "epoch": 0.49879492821963745,
      "grad_norm": 0.4529150128364563,
      "learning_rate": 1.0545502750951282e-05,
      "loss": 0.8794,
      "step": 2975
    },
    {
      "epoch": 0.49896259038038354,
      "grad_norm": 0.5789543986320496,
      "learning_rate": 1.0540080173111318e-05,
      "loss": 0.9309,
      "step": 2976
    },
    {
      "epoch": 0.49913025254112964,
      "grad_norm": 0.47327640652656555,
      "learning_rate": 1.0534657435995024e-05,
      "loss": 0.9094,
      "step": 2977
    },
    {
      "epoch": 0.49929791470187573,
      "grad_norm": 0.3968491554260254,
      "learning_rate": 1.0529234541201631e-05,
      "loss": 0.8466,
      "step": 2978
    },
    {
      "epoch": 0.49946557686262183,
      "grad_norm": 0.3362426459789276,
      "learning_rate": 1.0523811490330422e-05,
      "loss": 0.8066,
      "step": 2979
    },
    {
      "epoch": 0.4996332390233679,
      "grad_norm": 0.4447023570537567,
      "learning_rate": 1.0518388284980714e-05,
      "loss": 0.8388,
      "step": 2980
    },
    {
      "epoch": 0.499800901184114,
      "grad_norm": 0.594958484172821,
      "learning_rate": 1.0512964926751882e-05,
      "loss": 0.8109,
      "step": 2981
    },
    {
      "epoch": 0.4999685633448601,
      "grad_norm": 0.49405747652053833,
      "learning_rate": 1.0507541417243344e-05,
      "loss": 0.9152,
      "step": 2982
    },
    {
      "epoch": 0.5001362255056062,
      "grad_norm": 0.6013819575309753,
      "learning_rate": 1.0502117758054556e-05,
      "loss": 0.8891,
      "step": 2983
    },
    {
      "epoch": 0.5003038876663523,
      "grad_norm": 0.4540571868419647,
      "learning_rate": 1.0496693950785023e-05,
      "loss": 0.7649,
      "step": 2984
    },
    {
      "epoch": 0.5004715498270984,
      "grad_norm": 0.5885023474693298,
      "learning_rate": 1.0491269997034292e-05,
      "loss": 0.8125,
      "step": 2985
    },
    {
      "epoch": 0.5006392119878444,
      "grad_norm": 0.5920272469520569,
      "learning_rate": 1.0485845898401959e-05,
      "loss": 0.8707,
      "step": 2986
    },
    {
      "epoch": 0.5008068741485906,
      "grad_norm": 0.7279382348060608,
      "learning_rate": 1.048042165648765e-05,
      "loss": 1.001,
      "step": 2987
    },
    {
      "epoch": 0.5009745363093366,
      "grad_norm": 0.4096040427684784,
      "learning_rate": 1.047499727289105e-05,
      "loss": 0.8671,
      "step": 2988
    },
    {
      "epoch": 0.5011421984700828,
      "grad_norm": 0.494907408952713,
      "learning_rate": 1.0469572749211869e-05,
      "loss": 0.8205,
      "step": 2989
    },
    {
      "epoch": 0.5013098606308288,
      "grad_norm": 0.39264318346977234,
      "learning_rate": 1.0464148087049872e-05,
      "loss": 0.9057,
      "step": 2990
    },
    {
      "epoch": 0.501477522791575,
      "grad_norm": 0.5199148654937744,
      "learning_rate": 1.0458723288004858e-05,
      "loss": 0.8148,
      "step": 2991
    },
    {
      "epoch": 0.501645184952321,
      "grad_norm": 0.49908947944641113,
      "learning_rate": 1.0453298353676662e-05,
      "loss": 0.9296,
      "step": 2992
    },
    {
      "epoch": 0.5018128471130672,
      "grad_norm": 0.40984004735946655,
      "learning_rate": 1.0447873285665168e-05,
      "loss": 0.9216,
      "step": 2993
    },
    {
      "epoch": 0.5019805092738132,
      "grad_norm": 0.8072329759597778,
      "learning_rate": 1.04424480855703e-05,
      "loss": 0.9464,
      "step": 2994
    },
    {
      "epoch": 0.5021481714345594,
      "grad_norm": 0.47458645701408386,
      "learning_rate": 1.0437022754992011e-05,
      "loss": 0.9203,
      "step": 2995
    },
    {
      "epoch": 0.5023158335953054,
      "grad_norm": 0.5706420540809631,
      "learning_rate": 1.0431597295530299e-05,
      "loss": 0.9139,
      "step": 2996
    },
    {
      "epoch": 0.5024834957560516,
      "grad_norm": 0.4910517930984497,
      "learning_rate": 1.0426171708785203e-05,
      "loss": 0.9184,
      "step": 2997
    },
    {
      "epoch": 0.5026511579167976,
      "grad_norm": 0.4286358058452606,
      "learning_rate": 1.0420745996356792e-05,
      "loss": 0.8901,
      "step": 2998
    },
    {
      "epoch": 0.5028188200775437,
      "grad_norm": 0.39360329508781433,
      "learning_rate": 1.0415320159845175e-05,
      "loss": 0.8745,
      "step": 2999
    },
    {
      "epoch": 0.5029864822382899,
      "grad_norm": 0.6680741310119629,
      "learning_rate": 1.04098942008505e-05,
      "loss": 0.905,
      "step": 3000
    },
    {
      "epoch": 0.5031541443990359,
      "grad_norm": 0.631578803062439,
      "learning_rate": 1.040446812097295e-05,
      "loss": 0.8432,
      "step": 3001
    },
    {
      "epoch": 0.5033218065597821,
      "grad_norm": 0.6036131381988525,
      "learning_rate": 1.0399041921812745e-05,
      "loss": 0.926,
      "step": 3002
    },
    {
      "epoch": 0.5034894687205281,
      "grad_norm": 0.4073098301887512,
      "learning_rate": 1.0393615604970132e-05,
      "loss": 0.8286,
      "step": 3003
    },
    {
      "epoch": 0.5036571308812743,
      "grad_norm": 0.4562547206878662,
      "learning_rate": 1.0388189172045407e-05,
      "loss": 0.7902,
      "step": 3004
    },
    {
      "epoch": 0.5038247930420203,
      "grad_norm": 0.47056758403778076,
      "learning_rate": 1.0382762624638885e-05,
      "loss": 0.9353,
      "step": 3005
    },
    {
      "epoch": 0.5039924552027665,
      "grad_norm": 0.43075841665267944,
      "learning_rate": 1.0377335964350928e-05,
      "loss": 0.7843,
      "step": 3006
    },
    {
      "epoch": 0.5041601173635125,
      "grad_norm": 0.3730616867542267,
      "learning_rate": 1.0371909192781922e-05,
      "loss": 0.8009,
      "step": 3007
    },
    {
      "epoch": 0.5043277795242587,
      "grad_norm": 0.8039329648017883,
      "learning_rate": 1.0366482311532286e-05,
      "loss": 0.8716,
      "step": 3008
    },
    {
      "epoch": 0.5044954416850047,
      "grad_norm": 0.5027324557304382,
      "learning_rate": 1.0361055322202482e-05,
      "loss": 0.8978,
      "step": 3009
    },
    {
      "epoch": 0.5046631038457509,
      "grad_norm": 0.5500113368034363,
      "learning_rate": 1.0355628226392994e-05,
      "loss": 0.8553,
      "step": 3010
    },
    {
      "epoch": 0.5048307660064969,
      "grad_norm": 0.8844757676124573,
      "learning_rate": 1.0350201025704334e-05,
      "loss": 0.8621,
      "step": 3011
    },
    {
      "epoch": 0.504998428167243,
      "grad_norm": 0.5123084187507629,
      "learning_rate": 1.0344773721737058e-05,
      "loss": 0.8796,
      "step": 3012
    },
    {
      "epoch": 0.5051660903279891,
      "grad_norm": 0.5489059090614319,
      "learning_rate": 1.0339346316091742e-05,
      "loss": 0.911,
      "step": 3013
    },
    {
      "epoch": 0.5053337524887352,
      "grad_norm": 0.7856314778327942,
      "learning_rate": 1.0333918810368994e-05,
      "loss": 0.8677,
      "step": 3014
    },
    {
      "epoch": 0.5055014146494813,
      "grad_norm": 0.44746580719947815,
      "learning_rate": 1.0328491206169456e-05,
      "loss": 0.8347,
      "step": 3015
    },
    {
      "epoch": 0.5056690768102274,
      "grad_norm": 0.7760793566703796,
      "learning_rate": 1.0323063505093791e-05,
      "loss": 0.8911,
      "step": 3016
    },
    {
      "epoch": 0.5058367389709735,
      "grad_norm": 0.7316550612449646,
      "learning_rate": 1.03176357087427e-05,
      "loss": 0.9219,
      "step": 3017
    },
    {
      "epoch": 0.5060044011317196,
      "grad_norm": 0.4846658706665039,
      "learning_rate": 1.0312207818716905e-05,
      "loss": 0.8055,
      "step": 3018
    },
    {
      "epoch": 0.5061720632924657,
      "grad_norm": 0.5303776860237122,
      "learning_rate": 1.0306779836617156e-05,
      "loss": 0.8776,
      "step": 3019
    },
    {
      "epoch": 0.5063397254532118,
      "grad_norm": 0.5125965476036072,
      "learning_rate": 1.0301351764044239e-05,
      "loss": 0.9409,
      "step": 3020
    },
    {
      "epoch": 0.5065073876139579,
      "grad_norm": 0.39778846502304077,
      "learning_rate": 1.0295923602598951e-05,
      "loss": 0.7961,
      "step": 3021
    },
    {
      "epoch": 0.506675049774704,
      "grad_norm": 0.42018333077430725,
      "learning_rate": 1.0290495353882131e-05,
      "loss": 0.8947,
      "step": 3022
    },
    {
      "epoch": 0.50684271193545,
      "grad_norm": 0.5329704284667969,
      "learning_rate": 1.0285067019494632e-05,
      "loss": 0.8357,
      "step": 3023
    },
    {
      "epoch": 0.5070103740961962,
      "grad_norm": 0.6231377720832825,
      "learning_rate": 1.0279638601037338e-05,
      "loss": 0.8307,
      "step": 3024
    },
    {
      "epoch": 0.5071780362569422,
      "grad_norm": 0.5728231072425842,
      "learning_rate": 1.027421010011116e-05,
      "loss": 0.877,
      "step": 3025
    },
    {
      "epoch": 0.5073456984176884,
      "grad_norm": 0.6350905895233154,
      "learning_rate": 1.026878151831703e-05,
      "loss": 0.9096,
      "step": 3026
    },
    {
      "epoch": 0.5075133605784344,
      "grad_norm": 0.6212195754051208,
      "learning_rate": 1.0263352857255899e-05,
      "loss": 0.8157,
      "step": 3027
    },
    {
      "epoch": 0.5076810227391806,
      "grad_norm": 0.40822339057922363,
      "learning_rate": 1.0257924118528749e-05,
      "loss": 0.9032,
      "step": 3028
    },
    {
      "epoch": 0.5078486848999266,
      "grad_norm": 0.364933043718338,
      "learning_rate": 1.0252495303736584e-05,
      "loss": 0.8313,
      "step": 3029
    },
    {
      "epoch": 0.5080163470606728,
      "grad_norm": 0.46916210651397705,
      "learning_rate": 1.0247066414480424e-05,
      "loss": 0.9472,
      "step": 3030
    },
    {
      "epoch": 0.5081840092214188,
      "grad_norm": 0.4093315601348877,
      "learning_rate": 1.0241637452361323e-05,
      "loss": 0.8384,
      "step": 3031
    },
    {
      "epoch": 0.508351671382165,
      "grad_norm": 0.7968176007270813,
      "learning_rate": 1.0236208418980345e-05,
      "loss": 0.857,
      "step": 3032
    },
    {
      "epoch": 0.508519333542911,
      "grad_norm": 0.599753201007843,
      "learning_rate": 1.023077931593858e-05,
      "loss": 0.9746,
      "step": 3033
    },
    {
      "epoch": 0.5086869957036572,
      "grad_norm": 0.6237933039665222,
      "learning_rate": 1.0225350144837134e-05,
      "loss": 0.9419,
      "step": 3034
    },
    {
      "epoch": 0.5088546578644032,
      "grad_norm": 0.41964903473854065,
      "learning_rate": 1.021992090727714e-05,
      "loss": 0.8519,
      "step": 3035
    },
    {
      "epoch": 0.5090223200251494,
      "grad_norm": 0.5063328742980957,
      "learning_rate": 1.021449160485975e-05,
      "loss": 0.9173,
      "step": 3036
    },
    {
      "epoch": 0.5091899821858954,
      "grad_norm": 0.5008317828178406,
      "learning_rate": 1.0209062239186127e-05,
      "loss": 0.8588,
      "step": 3037
    },
    {
      "epoch": 0.5093576443466415,
      "grad_norm": 0.3873981833457947,
      "learning_rate": 1.0203632811857462e-05,
      "loss": 0.85,
      "step": 3038
    },
    {
      "epoch": 0.5095253065073876,
      "grad_norm": 0.6122221946716309,
      "learning_rate": 1.019820332447496e-05,
      "loss": 0.8791,
      "step": 3039
    },
    {
      "epoch": 0.5096929686681337,
      "grad_norm": 0.5497280955314636,
      "learning_rate": 1.0192773778639842e-05,
      "loss": 0.8646,
      "step": 3040
    },
    {
      "epoch": 0.5098606308288798,
      "grad_norm": 0.41970667243003845,
      "learning_rate": 1.0187344175953351e-05,
      "loss": 0.9264,
      "step": 3041
    },
    {
      "epoch": 0.5100282929896259,
      "grad_norm": 0.45841696858406067,
      "learning_rate": 1.018191451801674e-05,
      "loss": 0.8689,
      "step": 3042
    },
    {
      "epoch": 0.510195955150372,
      "grad_norm": 0.5520064234733582,
      "learning_rate": 1.0176484806431288e-05,
      "loss": 0.812,
      "step": 3043
    },
    {
      "epoch": 0.5103636173111181,
      "grad_norm": 0.6513684988021851,
      "learning_rate": 1.017105504279828e-05,
      "loss": 0.8724,
      "step": 3044
    },
    {
      "epoch": 0.5105312794718642,
      "grad_norm": 0.6604621410369873,
      "learning_rate": 1.0165625228719021e-05,
      "loss": 0.8985,
      "step": 3045
    },
    {
      "epoch": 0.5106989416326103,
      "grad_norm": 0.692649781703949,
      "learning_rate": 1.0160195365794832e-05,
      "loss": 0.9538,
      "step": 3046
    },
    {
      "epoch": 0.5108666037933564,
      "grad_norm": 0.5462502837181091,
      "learning_rate": 1.0154765455627048e-05,
      "loss": 0.901,
      "step": 3047
    },
    {
      "epoch": 0.5110342659541025,
      "grad_norm": 0.625444769859314,
      "learning_rate": 1.014933549981701e-05,
      "loss": 0.9138,
      "step": 3048
    },
    {
      "epoch": 0.5112019281148485,
      "grad_norm": 0.4703475534915924,
      "learning_rate": 1.0143905499966086e-05,
      "loss": 0.8508,
      "step": 3049
    },
    {
      "epoch": 0.5113695902755947,
      "grad_norm": 0.522480309009552,
      "learning_rate": 1.013847545767565e-05,
      "loss": 0.8468,
      "step": 3050
    },
    {
      "epoch": 0.5115372524363407,
      "grad_norm": 0.7132927775382996,
      "learning_rate": 1.0133045374547084e-05,
      "loss": 0.8493,
      "step": 3051
    },
    {
      "epoch": 0.5117049145970869,
      "grad_norm": 0.5355094075202942,
      "learning_rate": 1.0127615252181792e-05,
      "loss": 0.8351,
      "step": 3052
    },
    {
      "epoch": 0.5118725767578329,
      "grad_norm": 0.5499727129936218,
      "learning_rate": 1.0122185092181178e-05,
      "loss": 0.9591,
      "step": 3053
    },
    {
      "epoch": 0.5120402389185791,
      "grad_norm": 0.5338245630264282,
      "learning_rate": 1.0116754896146672e-05,
      "loss": 0.8053,
      "step": 3054
    },
    {
      "epoch": 0.5122079010793251,
      "grad_norm": 0.4411284029483795,
      "learning_rate": 1.01113246656797e-05,
      "loss": 0.941,
      "step": 3055
    },
    {
      "epoch": 0.5123755632400713,
      "grad_norm": 0.4244464039802551,
      "learning_rate": 1.0105894402381703e-05,
      "loss": 0.822,
      "step": 3056
    },
    {
      "epoch": 0.5125432254008173,
      "grad_norm": 0.4807767868041992,
      "learning_rate": 1.0100464107854137e-05,
      "loss": 0.8257,
      "step": 3057
    },
    {
      "epoch": 0.5127108875615635,
      "grad_norm": 0.4651215672492981,
      "learning_rate": 1.0095033783698462e-05,
      "loss": 0.8494,
      "step": 3058
    },
    {
      "epoch": 0.5128785497223095,
      "grad_norm": 0.5512310862541199,
      "learning_rate": 1.008960343151615e-05,
      "loss": 0.8645,
      "step": 3059
    },
    {
      "epoch": 0.5130462118830557,
      "grad_norm": 0.4677385091781616,
      "learning_rate": 1.0084173052908675e-05,
      "loss": 0.9542,
      "step": 3060
    },
    {
      "epoch": 0.5132138740438017,
      "grad_norm": 0.4788549542427063,
      "learning_rate": 1.0078742649477522e-05,
      "loss": 0.8827,
      "step": 3061
    },
    {
      "epoch": 0.5133815362045479,
      "grad_norm": 0.5147599577903748,
      "learning_rate": 1.0073312222824187e-05,
      "loss": 0.8223,
      "step": 3062
    },
    {
      "epoch": 0.5135491983652939,
      "grad_norm": 0.47588101029396057,
      "learning_rate": 1.0067881774550175e-05,
      "loss": 0.9135,
      "step": 3063
    },
    {
      "epoch": 0.51371686052604,
      "grad_norm": 0.5515173673629761,
      "learning_rate": 1.0062451306256981e-05,
      "loss": 0.8483,
      "step": 3064
    },
    {
      "epoch": 0.5138845226867861,
      "grad_norm": 0.6544636487960815,
      "learning_rate": 1.0057020819546122e-05,
      "loss": 0.8364,
      "step": 3065
    },
    {
      "epoch": 0.5140521848475322,
      "grad_norm": 0.8613523244857788,
      "learning_rate": 1.0051590316019122e-05,
      "loss": 0.8428,
      "step": 3066
    },
    {
      "epoch": 0.5142198470082783,
      "grad_norm": 0.39275920391082764,
      "learning_rate": 1.0046159797277497e-05,
      "loss": 0.8907,
      "step": 3067
    },
    {
      "epoch": 0.5143875091690244,
      "grad_norm": 0.5242758393287659,
      "learning_rate": 1.0040729264922778e-05,
      "loss": 0.8082,
      "step": 3068
    },
    {
      "epoch": 0.5145551713297705,
      "grad_norm": 0.6845324635505676,
      "learning_rate": 1.0035298720556493e-05,
      "loss": 0.911,
      "step": 3069
    },
    {
      "epoch": 0.5147228334905166,
      "grad_norm": 0.5309914946556091,
      "learning_rate": 1.0029868165780178e-05,
      "loss": 0.9313,
      "step": 3070
    },
    {
      "epoch": 0.5148904956512627,
      "grad_norm": 0.4313521087169647,
      "learning_rate": 1.0024437602195371e-05,
      "loss": 0.8413,
      "step": 3071
    },
    {
      "epoch": 0.5150581578120088,
      "grad_norm": 0.4510025084018707,
      "learning_rate": 1.0019007031403614e-05,
      "loss": 0.8322,
      "step": 3072
    },
    {
      "epoch": 0.5152258199727549,
      "grad_norm": 0.6852996945381165,
      "learning_rate": 1.0013576455006446e-05,
      "loss": 0.9159,
      "step": 3073
    },
    {
      "epoch": 0.515393482133501,
      "grad_norm": 0.5756083726882935,
      "learning_rate": 1.0008145874605414e-05,
      "loss": 0.9771,
      "step": 3074
    },
    {
      "epoch": 0.515561144294247,
      "grad_norm": 0.8508740067481995,
      "learning_rate": 1.0002715291802065e-05,
      "loss": 0.8343,
      "step": 3075
    },
    {
      "epoch": 0.5157288064549932,
      "grad_norm": 0.42136839032173157,
      "learning_rate": 9.997284708197938e-06,
      "loss": 0.8462,
      "step": 3076
    },
    {
      "epoch": 0.5158964686157392,
      "grad_norm": 0.6306905746459961,
      "learning_rate": 9.991854125394587e-06,
      "loss": 0.9103,
      "step": 3077
    },
    {
      "epoch": 0.5160641307764854,
      "grad_norm": 0.5494988560676575,
      "learning_rate": 9.986423544993559e-06,
      "loss": 0.9001,
      "step": 3078
    },
    {
      "epoch": 0.5162317929372314,
      "grad_norm": 0.47950515151023865,
      "learning_rate": 9.98099296859639e-06,
      "loss": 0.8205,
      "step": 3079
    },
    {
      "epoch": 0.5163994550979776,
      "grad_norm": 0.5724183917045593,
      "learning_rate": 9.975562397804632e-06,
      "loss": 0.8277,
      "step": 3080
    },
    {
      "epoch": 0.5165671172587236,
      "grad_norm": 0.29128915071487427,
      "learning_rate": 9.970131834219823e-06,
      "loss": 0.8813,
      "step": 3081
    },
    {
      "epoch": 0.5167347794194698,
      "grad_norm": 0.5448590517044067,
      "learning_rate": 9.964701279443509e-06,
      "loss": 0.8322,
      "step": 3082
    },
    {
      "epoch": 0.5169024415802158,
      "grad_norm": 0.40740934014320374,
      "learning_rate": 9.959270735077226e-06,
      "loss": 0.8544,
      "step": 3083
    },
    {
      "epoch": 0.517070103740962,
      "grad_norm": 0.5573588013648987,
      "learning_rate": 9.953840202722505e-06,
      "loss": 0.9403,
      "step": 3084
    },
    {
      "epoch": 0.5172377659017081,
      "grad_norm": 0.48960620164871216,
      "learning_rate": 9.94840968398088e-06,
      "loss": 0.904,
      "step": 3085
    },
    {
      "epoch": 0.5174054280624542,
      "grad_norm": 0.4078063368797302,
      "learning_rate": 9.942979180453878e-06,
      "loss": 0.9525,
      "step": 3086
    },
    {
      "epoch": 0.5175730902232003,
      "grad_norm": 0.46629518270492554,
      "learning_rate": 9.937548693743024e-06,
      "loss": 0.8169,
      "step": 3087
    },
    {
      "epoch": 0.5177407523839463,
      "grad_norm": 0.5462809801101685,
      "learning_rate": 9.932118225449832e-06,
      "loss": 0.8677,
      "step": 3088
    },
    {
      "epoch": 0.5179084145446925,
      "grad_norm": 0.3972527086734772,
      "learning_rate": 9.926687777175815e-06,
      "loss": 0.8871,
      "step": 3089
    },
    {
      "epoch": 0.5180760767054385,
      "grad_norm": 0.38695618510246277,
      "learning_rate": 9.921257350522481e-06,
      "loss": 0.8438,
      "step": 3090
    },
    {
      "epoch": 0.5182437388661847,
      "grad_norm": 0.41372543573379517,
      "learning_rate": 9.915826947091328e-06,
      "loss": 0.787,
      "step": 3091
    },
    {
      "epoch": 0.5184114010269307,
      "grad_norm": 0.39852261543273926,
      "learning_rate": 9.910396568483855e-06,
      "loss": 0.8882,
      "step": 3092
    },
    {
      "epoch": 0.5185790631876769,
      "grad_norm": 0.6337873935699463,
      "learning_rate": 9.90496621630154e-06,
      "loss": 0.9105,
      "step": 3093
    },
    {
      "epoch": 0.5187467253484229,
      "grad_norm": 0.4452924430370331,
      "learning_rate": 9.899535892145864e-06,
      "loss": 0.8428,
      "step": 3094
    },
    {
      "epoch": 0.5189143875091691,
      "grad_norm": 0.46548423171043396,
      "learning_rate": 9.894105597618297e-06,
      "loss": 0.8208,
      "step": 3095
    },
    {
      "epoch": 0.5190820496699151,
      "grad_norm": 0.6204050779342651,
      "learning_rate": 9.888675334320302e-06,
      "loss": 0.8098,
      "step": 3096
    },
    {
      "epoch": 0.5192497118306613,
      "grad_norm": 0.6824650168418884,
      "learning_rate": 9.883245103853333e-06,
      "loss": 0.9182,
      "step": 3097
    },
    {
      "epoch": 0.5194173739914073,
      "grad_norm": 0.7250338792800903,
      "learning_rate": 9.877814907818824e-06,
      "loss": 0.9327,
      "step": 3098
    },
    {
      "epoch": 0.5195850361521535,
      "grad_norm": 0.6577723622322083,
      "learning_rate": 9.87238474781821e-06,
      "loss": 0.8936,
      "step": 3099
    },
    {
      "epoch": 0.5197526983128995,
      "grad_norm": 0.40129998326301575,
      "learning_rate": 9.866954625452918e-06,
      "loss": 0.8641,
      "step": 3100
    },
    {
      "epoch": 0.5199203604736456,
      "grad_norm": 0.36280500888824463,
      "learning_rate": 9.861524542324355e-06,
      "loss": 0.8502,
      "step": 3101
    },
    {
      "epoch": 0.5200880226343917,
      "grad_norm": 0.44010069966316223,
      "learning_rate": 9.856094500033916e-06,
      "loss": 0.8933,
      "step": 3102
    },
    {
      "epoch": 0.5202556847951378,
      "grad_norm": 0.37167131900787354,
      "learning_rate": 9.850664500182991e-06,
      "loss": 0.8611,
      "step": 3103
    },
    {
      "epoch": 0.5204233469558839,
      "grad_norm": 0.6913161277770996,
      "learning_rate": 9.845234544372956e-06,
      "loss": 0.9099,
      "step": 3104
    },
    {
      "epoch": 0.52059100911663,
      "grad_norm": 0.506624162197113,
      "learning_rate": 9.839804634205168e-06,
      "loss": 0.973,
      "step": 3105
    },
    {
      "epoch": 0.5207586712773761,
      "grad_norm": 0.3480541408061981,
      "learning_rate": 9.834374771280982e-06,
      "loss": 0.8869,
      "step": 3106
    },
    {
      "epoch": 0.5209263334381222,
      "grad_norm": 0.40910738706588745,
      "learning_rate": 9.828944957201724e-06,
      "loss": 0.8806,
      "step": 3107
    },
    {
      "epoch": 0.5210939955988683,
      "grad_norm": 0.47690507769584656,
      "learning_rate": 9.823515193568715e-06,
      "loss": 0.818,
      "step": 3108
    },
    {
      "epoch": 0.5212616577596144,
      "grad_norm": 0.6599212288856506,
      "learning_rate": 9.818085481983262e-06,
      "loss": 0.9207,
      "step": 3109
    },
    {
      "epoch": 0.5214293199203605,
      "grad_norm": 0.47169163823127747,
      "learning_rate": 9.812655824046654e-06,
      "loss": 0.8209,
      "step": 3110
    },
    {
      "epoch": 0.5215969820811066,
      "grad_norm": 0.5017246603965759,
      "learning_rate": 9.807226221360163e-06,
      "loss": 0.8716,
      "step": 3111
    },
    {
      "epoch": 0.5217646442418526,
      "grad_norm": 0.38969993591308594,
      "learning_rate": 9.801796675525044e-06,
      "loss": 0.8849,
      "step": 3112
    },
    {
      "epoch": 0.5219323064025988,
      "grad_norm": 0.6319521069526672,
      "learning_rate": 9.796367188142541e-06,
      "loss": 0.8002,
      "step": 3113
    },
    {
      "epoch": 0.5220999685633448,
      "grad_norm": 0.5612240433692932,
      "learning_rate": 9.790937760813874e-06,
      "loss": 0.8045,
      "step": 3114
    },
    {
      "epoch": 0.522267630724091,
      "grad_norm": 0.5317338705062866,
      "learning_rate": 9.785508395140255e-06,
      "loss": 0.8176,
      "step": 3115
    },
    {
      "epoch": 0.522435292884837,
      "grad_norm": 0.4771307706832886,
      "learning_rate": 9.780079092722863e-06,
      "loss": 0.8691,
      "step": 3116
    },
    {
      "epoch": 0.5226029550455832,
      "grad_norm": 0.8237624764442444,
      "learning_rate": 9.77464985516287e-06,
      "loss": 0.8767,
      "step": 3117
    },
    {
      "epoch": 0.5227706172063292,
      "grad_norm": 0.611888587474823,
      "learning_rate": 9.769220684061422e-06,
      "loss": 0.8261,
      "step": 3118
    },
    {
      "epoch": 0.5229382793670754,
      "grad_norm": 0.7214562296867371,
      "learning_rate": 9.76379158101966e-06,
      "loss": 0.8596,
      "step": 3119
    },
    {
      "epoch": 0.5231059415278214,
      "grad_norm": 0.6046148538589478,
      "learning_rate": 9.75836254763868e-06,
      "loss": 0.9288,
      "step": 3120
    },
    {
      "epoch": 0.5232736036885676,
      "grad_norm": 0.7193319797515869,
      "learning_rate": 9.752933585519578e-06,
      "loss": 0.9434,
      "step": 3121
    },
    {
      "epoch": 0.5234412658493136,
      "grad_norm": 0.539698600769043,
      "learning_rate": 9.747504696263417e-06,
      "loss": 0.8804,
      "step": 3122
    },
    {
      "epoch": 0.5236089280100598,
      "grad_norm": 0.5887028574943542,
      "learning_rate": 9.742075881471251e-06,
      "loss": 0.9242,
      "step": 3123
    },
    {
      "epoch": 0.5237765901708058,
      "grad_norm": 0.6475830674171448,
      "learning_rate": 9.736647142744106e-06,
      "loss": 0.8669,
      "step": 3124
    },
    {
      "epoch": 0.523944252331552,
      "grad_norm": 0.5839582681655884,
      "learning_rate": 9.731218481682973e-06,
      "loss": 0.8992,
      "step": 3125
    },
    {
      "epoch": 0.524111914492298,
      "grad_norm": 0.6649941802024841,
      "learning_rate": 9.72578989988884e-06,
      "loss": 0.8331,
      "step": 3126
    },
    {
      "epoch": 0.5242795766530441,
      "grad_norm": 0.5015733242034912,
      "learning_rate": 9.720361398962661e-06,
      "loss": 0.8447,
      "step": 3127
    },
    {
      "epoch": 0.5244472388137902,
      "grad_norm": 0.45080992579460144,
      "learning_rate": 9.714932980505372e-06,
      "loss": 0.8166,
      "step": 3128
    },
    {
      "epoch": 0.5246149009745363,
      "grad_norm": 0.5960597395896912,
      "learning_rate": 9.709504646117874e-06,
      "loss": 0.8761,
      "step": 3129
    },
    {
      "epoch": 0.5247825631352824,
      "grad_norm": 0.5214238166809082,
      "learning_rate": 9.70407639740105e-06,
      "loss": 0.8754,
      "step": 3130
    },
    {
      "epoch": 0.5249502252960285,
      "grad_norm": 0.418317049741745,
      "learning_rate": 9.698648235955764e-06,
      "loss": 0.9651,
      "step": 3131
    },
    {
      "epoch": 0.5251178874567746,
      "grad_norm": 0.40980616211891174,
      "learning_rate": 9.693220163382844e-06,
      "loss": 0.8648,
      "step": 3132
    },
    {
      "epoch": 0.5252855496175207,
      "grad_norm": 0.5039193034172058,
      "learning_rate": 9.687792181283098e-06,
      "loss": 0.768,
      "step": 3133
    },
    {
      "epoch": 0.5254532117782668,
      "grad_norm": 0.41915470361709595,
      "learning_rate": 9.682364291257304e-06,
      "loss": 0.8895,
      "step": 3134
    },
    {
      "epoch": 0.5256208739390129,
      "grad_norm": 0.5267002582550049,
      "learning_rate": 9.67693649490621e-06,
      "loss": 0.7787,
      "step": 3135
    },
    {
      "epoch": 0.525788536099759,
      "grad_norm": 0.43035128712654114,
      "learning_rate": 9.671508793830548e-06,
      "loss": 0.8968,
      "step": 3136
    },
    {
      "epoch": 0.5259561982605051,
      "grad_norm": 0.4039972722530365,
      "learning_rate": 9.666081189631007e-06,
      "loss": 0.876,
      "step": 3137
    },
    {
      "epoch": 0.5261238604212511,
      "grad_norm": 0.392192006111145,
      "learning_rate": 9.660653683908263e-06,
      "loss": 0.9633,
      "step": 3138
    },
    {
      "epoch": 0.5262915225819973,
      "grad_norm": 0.4463840126991272,
      "learning_rate": 9.655226278262945e-06,
      "loss": 0.8594,
      "step": 3139
    },
    {
      "epoch": 0.5264591847427433,
      "grad_norm": 0.8318145275115967,
      "learning_rate": 9.649798974295668e-06,
      "loss": 0.9015,
      "step": 3140
    },
    {
      "epoch": 0.5266268469034895,
      "grad_norm": 0.42815402150154114,
      "learning_rate": 9.644371773607008e-06,
      "loss": 0.8858,
      "step": 3141
    },
    {
      "epoch": 0.5267945090642355,
      "grad_norm": 0.41154927015304565,
      "learning_rate": 9.638944677797522e-06,
      "loss": 0.8129,
      "step": 3142
    },
    {
      "epoch": 0.5269621712249817,
      "grad_norm": 1.375307559967041,
      "learning_rate": 9.633517688467717e-06,
      "loss": 0.8723,
      "step": 3143
    },
    {
      "epoch": 0.5271298333857277,
      "grad_norm": 0.8266283869743347,
      "learning_rate": 9.628090807218082e-06,
      "loss": 0.9643,
      "step": 3144
    },
    {
      "epoch": 0.5272974955464739,
      "grad_norm": 0.43826937675476074,
      "learning_rate": 9.622664035649074e-06,
      "loss": 1.0305,
      "step": 3145
    },
    {
      "epoch": 0.5274651577072199,
      "grad_norm": 0.48454275727272034,
      "learning_rate": 9.617237375361115e-06,
      "loss": 0.892,
      "step": 3146
    },
    {
      "epoch": 0.5276328198679661,
      "grad_norm": 0.7665025591850281,
      "learning_rate": 9.6118108279546e-06,
      "loss": 0.8997,
      "step": 3147
    },
    {
      "epoch": 0.5278004820287121,
      "grad_norm": 0.5407666563987732,
      "learning_rate": 9.60638439502987e-06,
      "loss": 0.829,
      "step": 3148
    },
    {
      "epoch": 0.5279681441894583,
      "grad_norm": 0.503190815448761,
      "learning_rate": 9.600958078187258e-06,
      "loss": 0.8621,
      "step": 3149
    },
    {
      "epoch": 0.5281358063502043,
      "grad_norm": 0.5572085380554199,
      "learning_rate": 9.595531879027052e-06,
      "loss": 0.8968,
      "step": 3150
    },
    {
      "epoch": 0.5283034685109504,
      "grad_norm": 0.42003175616264343,
      "learning_rate": 9.590105799149502e-06,
      "loss": 0.8576,
      "step": 3151
    },
    {
      "epoch": 0.5284711306716965,
      "grad_norm": 0.44636985659599304,
      "learning_rate": 9.584679840154829e-06,
      "loss": 0.8329,
      "step": 3152
    },
    {
      "epoch": 0.5286387928324426,
      "grad_norm": 0.5694844722747803,
      "learning_rate": 9.579254003643213e-06,
      "loss": 0.8488,
      "step": 3153
    },
    {
      "epoch": 0.5288064549931887,
      "grad_norm": 0.5580739378929138,
      "learning_rate": 9.573828291214799e-06,
      "loss": 0.9743,
      "step": 3154
    },
    {
      "epoch": 0.5289741171539348,
      "grad_norm": 0.4939458966255188,
      "learning_rate": 9.568402704469701e-06,
      "loss": 0.8983,
      "step": 3155
    },
    {
      "epoch": 0.5291417793146809,
      "grad_norm": 0.7728688716888428,
      "learning_rate": 9.562977245007994e-06,
      "loss": 0.8119,
      "step": 3156
    },
    {
      "epoch": 0.529309441475427,
      "grad_norm": 0.5365317463874817,
      "learning_rate": 9.557551914429703e-06,
      "loss": 0.8443,
      "step": 3157
    },
    {
      "epoch": 0.5294771036361731,
      "grad_norm": 0.5773930549621582,
      "learning_rate": 9.552126714334834e-06,
      "loss": 0.8962,
      "step": 3158
    },
    {
      "epoch": 0.5296447657969192,
      "grad_norm": 0.44260895252227783,
      "learning_rate": 9.546701646323341e-06,
      "loss": 0.8515,
      "step": 3159
    },
    {
      "epoch": 0.5298124279576653,
      "grad_norm": 0.5546987652778625,
      "learning_rate": 9.541276711995149e-06,
      "loss": 0.9005,
      "step": 3160
    },
    {
      "epoch": 0.5299800901184114,
      "grad_norm": 0.5294981598854065,
      "learning_rate": 9.535851912950133e-06,
      "loss": 0.8382,
      "step": 3161
    },
    {
      "epoch": 0.5301477522791574,
      "grad_norm": 0.5314891934394836,
      "learning_rate": 9.530427250788134e-06,
      "loss": 0.8224,
      "step": 3162
    },
    {
      "epoch": 0.5303154144399036,
      "grad_norm": 0.4622967839241028,
      "learning_rate": 9.525002727108954e-06,
      "loss": 0.8289,
      "step": 3163
    },
    {
      "epoch": 0.5304830766006496,
      "grad_norm": 0.37820157408714294,
      "learning_rate": 9.51957834351235e-06,
      "loss": 0.8796,
      "step": 3164
    },
    {
      "epoch": 0.5306507387613958,
      "grad_norm": 0.3496609330177307,
      "learning_rate": 9.514154101598046e-06,
      "loss": 0.8418,
      "step": 3165
    },
    {
      "epoch": 0.5308184009221418,
      "grad_norm": 1.156503438949585,
      "learning_rate": 9.508730002965711e-06,
      "loss": 0.9668,
      "step": 3166
    },
    {
      "epoch": 0.530986063082888,
      "grad_norm": 0.39240220189094543,
      "learning_rate": 9.50330604921498e-06,
      "loss": 0.9064,
      "step": 3167
    },
    {
      "epoch": 0.531153725243634,
      "grad_norm": 0.4107801616191864,
      "learning_rate": 9.497882241945446e-06,
      "loss": 0.7483,
      "step": 3168
    },
    {
      "epoch": 0.5313213874043802,
      "grad_norm": 0.3688758313655853,
      "learning_rate": 9.492458582756658e-06,
      "loss": 0.9135,
      "step": 3169
    },
    {
      "epoch": 0.5314890495651263,
      "grad_norm": 0.5714778900146484,
      "learning_rate": 9.487035073248123e-06,
      "loss": 0.8872,
      "step": 3170
    },
    {
      "epoch": 0.5316567117258724,
      "grad_norm": 0.6979575157165527,
      "learning_rate": 9.481611715019287e-06,
      "loss": 0.8417,
      "step": 3171
    },
    {
      "epoch": 0.5318243738866185,
      "grad_norm": 0.41339144110679626,
      "learning_rate": 9.476188509669582e-06,
      "loss": 0.8654,
      "step": 3172
    },
    {
      "epoch": 0.5319920360473646,
      "grad_norm": 0.3709242045879364,
      "learning_rate": 9.470765458798369e-06,
      "loss": 0.8673,
      "step": 3173
    },
    {
      "epoch": 0.5321596982081107,
      "grad_norm": 0.5709183216094971,
      "learning_rate": 9.46534256400498e-06,
      "loss": 0.9746,
      "step": 3174
    },
    {
      "epoch": 0.5323273603688568,
      "grad_norm": 0.9141378998756409,
      "learning_rate": 9.459919826888685e-06,
      "loss": 0.861,
      "step": 3175
    },
    {
      "epoch": 0.5324950225296029,
      "grad_norm": 0.41457533836364746,
      "learning_rate": 9.454497249048723e-06,
      "loss": 0.8245,
      "step": 3176
    },
    {
      "epoch": 0.5326626846903489,
      "grad_norm": 0.48134270310401917,
      "learning_rate": 9.449074832084274e-06,
      "loss": 0.8726,
      "step": 3177
    },
    {
      "epoch": 0.5328303468510951,
      "grad_norm": 0.7749919891357422,
      "learning_rate": 9.443652577594482e-06,
      "loss": 0.7487,
      "step": 3178
    },
    {
      "epoch": 0.5329980090118411,
      "grad_norm": 0.4469508230686188,
      "learning_rate": 9.438230487178434e-06,
      "loss": 0.7934,
      "step": 3179
    },
    {
      "epoch": 0.5331656711725873,
      "grad_norm": 0.4624178111553192,
      "learning_rate": 9.43280856243517e-06,
      "loss": 0.9039,
      "step": 3180
    },
    {
      "epoch": 0.5333333333333333,
      "grad_norm": 0.74716717004776,
      "learning_rate": 9.427386804963684e-06,
      "loss": 0.8049,
      "step": 3181
    },
    {
      "epoch": 0.5335009954940795,
      "grad_norm": 0.8879395723342896,
      "learning_rate": 9.421965216362921e-06,
      "loss": 0.8456,
      "step": 3182
    },
    {
      "epoch": 0.5336686576548255,
      "grad_norm": 0.3464726507663727,
      "learning_rate": 9.416543798231777e-06,
      "loss": 0.8753,
      "step": 3183
    },
    {
      "epoch": 0.5338363198155717,
      "grad_norm": 0.3298185467720032,
      "learning_rate": 9.411122552169086e-06,
      "loss": 0.8975,
      "step": 3184
    },
    {
      "epoch": 0.5340039819763177,
      "grad_norm": 0.5450412034988403,
      "learning_rate": 9.405701479773649e-06,
      "loss": 0.839,
      "step": 3185
    },
    {
      "epoch": 0.5341716441370639,
      "grad_norm": 0.4394127428531647,
      "learning_rate": 9.400280582644204e-06,
      "loss": 0.8743,
      "step": 3186
    },
    {
      "epoch": 0.5343393062978099,
      "grad_norm": 0.4711911976337433,
      "learning_rate": 9.394859862379438e-06,
      "loss": 0.8515,
      "step": 3187
    },
    {
      "epoch": 0.534506968458556,
      "grad_norm": 0.5024020075798035,
      "learning_rate": 9.389439320578e-06,
      "loss": 0.9068,
      "step": 3188
    },
    {
      "epoch": 0.5346746306193021,
      "grad_norm": 0.5180482864379883,
      "learning_rate": 9.384018958838462e-06,
      "loss": 0.884,
      "step": 3189
    },
    {
      "epoch": 0.5348422927800482,
      "grad_norm": 0.3648953437805176,
      "learning_rate": 9.378598778759357e-06,
      "loss": 0.7798,
      "step": 3190
    },
    {
      "epoch": 0.5350099549407943,
      "grad_norm": 0.503474235534668,
      "learning_rate": 9.373178781939172e-06,
      "loss": 0.8051,
      "step": 3191
    },
    {
      "epoch": 0.5351776171015404,
      "grad_norm": 0.393228679895401,
      "learning_rate": 9.367758969976327e-06,
      "loss": 0.8734,
      "step": 3192
    },
    {
      "epoch": 0.5353452792622865,
      "grad_norm": 0.4325978755950928,
      "learning_rate": 9.362339344469189e-06,
      "loss": 0.8687,
      "step": 3193
    },
    {
      "epoch": 0.5355129414230326,
      "grad_norm": 0.5653034448623657,
      "learning_rate": 9.35691990701607e-06,
      "loss": 0.8447,
      "step": 3194
    },
    {
      "epoch": 0.5356806035837787,
      "grad_norm": 0.4211442172527313,
      "learning_rate": 9.351500659215238e-06,
      "loss": 1.029,
      "step": 3195
    },
    {
      "epoch": 0.5358482657445248,
      "grad_norm": 0.5952498912811279,
      "learning_rate": 9.34608160266489e-06,
      "loss": 0.9139,
      "step": 3196
    },
    {
      "epoch": 0.5360159279052709,
      "grad_norm": 0.4283948838710785,
      "learning_rate": 9.34066273896318e-06,
      "loss": 0.8884,
      "step": 3197
    },
    {
      "epoch": 0.536183590066017,
      "grad_norm": 0.4888134300708771,
      "learning_rate": 9.335244069708188e-06,
      "loss": 0.8698,
      "step": 3198
    },
    {
      "epoch": 0.536351252226763,
      "grad_norm": 0.6299174427986145,
      "learning_rate": 9.32982559649795e-06,
      "loss": 0.8223,
      "step": 3199
    },
    {
      "epoch": 0.5365189143875092,
      "grad_norm": 0.5382765531539917,
      "learning_rate": 9.324407320930444e-06,
      "loss": 0.8012,
      "step": 3200
    },
    {
      "epoch": 0.5366865765482552,
      "grad_norm": 1.366763710975647,
      "learning_rate": 9.318989244603586e-06,
      "loss": 0.8934,
      "step": 3201
    },
    {
      "epoch": 0.5368542387090014,
      "grad_norm": 0.5220536589622498,
      "learning_rate": 9.313571369115231e-06,
      "loss": 0.8822,
      "step": 3202
    },
    {
      "epoch": 0.5370219008697474,
      "grad_norm": 0.5266855359077454,
      "learning_rate": 9.30815369606318e-06,
      "loss": 0.8345,
      "step": 3203
    },
    {
      "epoch": 0.5371895630304936,
      "grad_norm": 0.5697585940361023,
      "learning_rate": 9.30273622704517e-06,
      "loss": 0.7852,
      "step": 3204
    },
    {
      "epoch": 0.5373572251912396,
      "grad_norm": 0.5060375928878784,
      "learning_rate": 9.297318963658882e-06,
      "loss": 0.8918,
      "step": 3205
    },
    {
      "epoch": 0.5375248873519858,
      "grad_norm": 0.3974671959877014,
      "learning_rate": 9.291901907501934e-06,
      "loss": 0.8648,
      "step": 3206
    },
    {
      "epoch": 0.5376925495127318,
      "grad_norm": 0.36046454310417175,
      "learning_rate": 9.286485060171881e-06,
      "loss": 0.8569,
      "step": 3207
    },
    {
      "epoch": 0.537860211673478,
      "grad_norm": 0.47915321588516235,
      "learning_rate": 9.281068423266217e-06,
      "loss": 0.9252,
      "step": 3208
    },
    {
      "epoch": 0.538027873834224,
      "grad_norm": 0.44521069526672363,
      "learning_rate": 9.275651998382377e-06,
      "loss": 0.8365,
      "step": 3209
    },
    {
      "epoch": 0.5381955359949702,
      "grad_norm": 1.0060954093933105,
      "learning_rate": 9.270235787117732e-06,
      "loss": 0.8021,
      "step": 3210
    },
    {
      "epoch": 0.5383631981557162,
      "grad_norm": 0.6790063381195068,
      "learning_rate": 9.264819791069596e-06,
      "loss": 0.8377,
      "step": 3211
    },
    {
      "epoch": 0.5385308603164624,
      "grad_norm": 0.45293375849723816,
      "learning_rate": 9.259404011835203e-06,
      "loss": 0.8844,
      "step": 3212
    },
    {
      "epoch": 0.5386985224772084,
      "grad_norm": 0.37161368131637573,
      "learning_rate": 9.253988451011734e-06,
      "loss": 0.8102,
      "step": 3213
    },
    {
      "epoch": 0.5388661846379545,
      "grad_norm": 0.5089200735092163,
      "learning_rate": 9.24857311019631e-06,
      "loss": 0.8451,
      "step": 3214
    },
    {
      "epoch": 0.5390338467987006,
      "grad_norm": 0.4144771695137024,
      "learning_rate": 9.243157990985989e-06,
      "loss": 0.8866,
      "step": 3215
    },
    {
      "epoch": 0.5392015089594467,
      "grad_norm": 0.48967111110687256,
      "learning_rate": 9.237743094977737e-06,
      "loss": 0.9099,
      "step": 3216
    },
    {
      "epoch": 0.5393691711201928,
      "grad_norm": 0.4924265444278717,
      "learning_rate": 9.232328423768491e-06,
      "loss": 0.8574,
      "step": 3217
    },
    {
      "epoch": 0.5395368332809389,
      "grad_norm": 0.6180545687675476,
      "learning_rate": 9.226913978955097e-06,
      "loss": 0.892,
      "step": 3218
    },
    {
      "epoch": 0.539704495441685,
      "grad_norm": 1.3185502290725708,
      "learning_rate": 9.221499762134344e-06,
      "loss": 0.8855,
      "step": 3219
    },
    {
      "epoch": 0.5398721576024311,
      "grad_norm": 0.36848878860473633,
      "learning_rate": 9.216085774902952e-06,
      "loss": 0.7998,
      "step": 3220
    },
    {
      "epoch": 0.5400398197631772,
      "grad_norm": 0.5853696465492249,
      "learning_rate": 9.210672018857572e-06,
      "loss": 0.9138,
      "step": 3221
    },
    {
      "epoch": 0.5402074819239233,
      "grad_norm": 0.46801823377609253,
      "learning_rate": 9.205258495594786e-06,
      "loss": 0.9831,
      "step": 3222
    },
    {
      "epoch": 0.5403751440846694,
      "grad_norm": 0.8978372812271118,
      "learning_rate": 9.199845206711112e-06,
      "loss": 0.8875,
      "step": 3223
    },
    {
      "epoch": 0.5405428062454155,
      "grad_norm": 0.6240953207015991,
      "learning_rate": 9.194432153802996e-06,
      "loss": 0.8235,
      "step": 3224
    },
    {
      "epoch": 0.5407104684061615,
      "grad_norm": 0.7486779689788818,
      "learning_rate": 9.189019338466812e-06,
      "loss": 0.8672,
      "step": 3225
    },
    {
      "epoch": 0.5408781305669077,
      "grad_norm": 0.4113537669181824,
      "learning_rate": 9.183606762298867e-06,
      "loss": 0.8666,
      "step": 3226
    },
    {
      "epoch": 0.5410457927276537,
      "grad_norm": 0.4346449673175812,
      "learning_rate": 9.178194426895396e-06,
      "loss": 0.8889,
      "step": 3227
    },
    {
      "epoch": 0.5412134548883999,
      "grad_norm": 0.6154700517654419,
      "learning_rate": 9.172782333852566e-06,
      "loss": 0.9337,
      "step": 3228
    },
    {
      "epoch": 0.5413811170491459,
      "grad_norm": 0.9208604693412781,
      "learning_rate": 9.16737048476647e-06,
      "loss": 0.9082,
      "step": 3229
    },
    {
      "epoch": 0.5415487792098921,
      "grad_norm": 0.72853684425354,
      "learning_rate": 9.161958881233129e-06,
      "loss": 0.9172,
      "step": 3230
    },
    {
      "epoch": 0.5417164413706381,
      "grad_norm": 0.8726669549942017,
      "learning_rate": 9.156547524848489e-06,
      "loss": 0.7786,
      "step": 3231
    },
    {
      "epoch": 0.5418841035313843,
      "grad_norm": 0.39041054248809814,
      "learning_rate": 9.151136417208424e-06,
      "loss": 0.9218,
      "step": 3232
    },
    {
      "epoch": 0.5420517656921303,
      "grad_norm": 0.3628639280796051,
      "learning_rate": 9.14572555990875e-06,
      "loss": 0.8328,
      "step": 3233
    },
    {
      "epoch": 0.5422194278528765,
      "grad_norm": 0.39143988490104675,
      "learning_rate": 9.140314954545184e-06,
      "loss": 0.874,
      "step": 3234
    },
    {
      "epoch": 0.5423870900136225,
      "grad_norm": 0.8233510255813599,
      "learning_rate": 9.13490460271338e-06,
      "loss": 0.7631,
      "step": 3235
    },
    {
      "epoch": 0.5425547521743687,
      "grad_norm": 0.5175696611404419,
      "learning_rate": 9.12949450600892e-06,
      "loss": 0.8708,
      "step": 3236
    },
    {
      "epoch": 0.5427224143351147,
      "grad_norm": 0.4261394441127777,
      "learning_rate": 9.124084666027311e-06,
      "loss": 0.8978,
      "step": 3237
    },
    {
      "epoch": 0.5428900764958609,
      "grad_norm": 0.48911020159721375,
      "learning_rate": 9.118675084363986e-06,
      "loss": 0.8587,
      "step": 3238
    },
    {
      "epoch": 0.5430577386566069,
      "grad_norm": 0.4675133526325226,
      "learning_rate": 9.113265762614285e-06,
      "loss": 0.8338,
      "step": 3239
    },
    {
      "epoch": 0.543225400817353,
      "grad_norm": 0.7315004467964172,
      "learning_rate": 9.107856702373492e-06,
      "loss": 0.8844,
      "step": 3240
    },
    {
      "epoch": 0.5433930629780991,
      "grad_norm": 0.41969773173332214,
      "learning_rate": 9.102447905236806e-06,
      "loss": 0.8963,
      "step": 3241
    },
    {
      "epoch": 0.5435607251388452,
      "grad_norm": 0.6964146494865417,
      "learning_rate": 9.097039372799345e-06,
      "loss": 0.7791,
      "step": 3242
    },
    {
      "epoch": 0.5437283872995913,
      "grad_norm": 0.6561281681060791,
      "learning_rate": 9.091631106656159e-06,
      "loss": 0.8957,
      "step": 3243
    },
    {
      "epoch": 0.5438960494603374,
      "grad_norm": 0.4261462986469269,
      "learning_rate": 9.086223108402203e-06,
      "loss": 0.7785,
      "step": 3244
    },
    {
      "epoch": 0.5440637116210835,
      "grad_norm": 0.7788665890693665,
      "learning_rate": 9.08081537963237e-06,
      "loss": 0.8151,
      "step": 3245
    },
    {
      "epoch": 0.5442313737818296,
      "grad_norm": 0.44058895111083984,
      "learning_rate": 9.075407921941463e-06,
      "loss": 0.8303,
      "step": 3246
    },
    {
      "epoch": 0.5443990359425757,
      "grad_norm": 0.7544462084770203,
      "learning_rate": 9.07000073692421e-06,
      "loss": 0.8711,
      "step": 3247
    },
    {
      "epoch": 0.5445666981033218,
      "grad_norm": 0.39317142963409424,
      "learning_rate": 9.064593826175254e-06,
      "loss": 0.7691,
      "step": 3248
    },
    {
      "epoch": 0.5447343602640679,
      "grad_norm": 0.48124146461486816,
      "learning_rate": 9.059187191289161e-06,
      "loss": 0.8287,
      "step": 3249
    },
    {
      "epoch": 0.544902022424814,
      "grad_norm": 0.4958220422267914,
      "learning_rate": 9.053780833860416e-06,
      "loss": 0.8171,
      "step": 3250
    },
    {
      "epoch": 0.54506968458556,
      "grad_norm": 0.5220127701759338,
      "learning_rate": 9.04837475548342e-06,
      "loss": 0.857,
      "step": 3251
    },
    {
      "epoch": 0.5452373467463062,
      "grad_norm": 0.37427911162376404,
      "learning_rate": 9.042968957752493e-06,
      "loss": 0.7819,
      "step": 3252
    },
    {
      "epoch": 0.5454050089070522,
      "grad_norm": 0.3906075358390808,
      "learning_rate": 9.037563442261872e-06,
      "loss": 0.8551,
      "step": 3253
    },
    {
      "epoch": 0.5455726710677984,
      "grad_norm": 0.3960884213447571,
      "learning_rate": 9.032158210605708e-06,
      "loss": 0.8849,
      "step": 3254
    },
    {
      "epoch": 0.5457403332285444,
      "grad_norm": 0.651520311832428,
      "learning_rate": 9.02675326437807e-06,
      "loss": 0.8351,
      "step": 3255
    },
    {
      "epoch": 0.5459079953892906,
      "grad_norm": 0.6072437763214111,
      "learning_rate": 9.021348605172951e-06,
      "loss": 0.9324,
      "step": 3256
    },
    {
      "epoch": 0.5460756575500367,
      "grad_norm": 0.39354023337364197,
      "learning_rate": 9.015944234584245e-06,
      "loss": 0.8851,
      "step": 3257
    },
    {
      "epoch": 0.5462433197107828,
      "grad_norm": 1.129841923713684,
      "learning_rate": 9.010540154205767e-06,
      "loss": 0.9725,
      "step": 3258
    },
    {
      "epoch": 0.5464109818715289,
      "grad_norm": 0.7908410429954529,
      "learning_rate": 9.00513636563125e-06,
      "loss": 0.8154,
      "step": 3259
    },
    {
      "epoch": 0.546578644032275,
      "grad_norm": 0.3996315002441406,
      "learning_rate": 8.999732870454339e-06,
      "loss": 0.8057,
      "step": 3260
    },
    {
      "epoch": 0.5467463061930211,
      "grad_norm": 0.5092489719390869,
      "learning_rate": 8.994329670268595e-06,
      "loss": 0.8163,
      "step": 3261
    },
    {
      "epoch": 0.5469139683537672,
      "grad_norm": 0.6034070253372192,
      "learning_rate": 8.988926766667478e-06,
      "loss": 0.8637,
      "step": 3262
    },
    {
      "epoch": 0.5470816305145133,
      "grad_norm": 0.6034070253372192,
      "learning_rate": 8.988926766667478e-06,
      "loss": 0.929,
      "step": 3263
    },
    {
      "epoch": 0.5472492926752593,
      "grad_norm": 0.531434953212738,
      "learning_rate": 8.983524161244379e-06,
      "loss": 0.8021,
      "step": 3264
    },
    {
      "epoch": 0.5474169548360055,
      "grad_norm": 0.4655191898345947,
      "learning_rate": 8.978121855592593e-06,
      "loss": 0.8185,
      "step": 3265
    },
    {
      "epoch": 0.5475846169967515,
      "grad_norm": 0.4762585163116455,
      "learning_rate": 8.972719851305327e-06,
      "loss": 0.8595,
      "step": 3266
    },
    {
      "epoch": 0.5477522791574977,
      "grad_norm": 0.6660287380218506,
      "learning_rate": 8.967318149975695e-06,
      "loss": 0.8693,
      "step": 3267
    },
    {
      "epoch": 0.5479199413182437,
      "grad_norm": 0.40586456656455994,
      "learning_rate": 8.961916753196728e-06,
      "loss": 0.8162,
      "step": 3268
    },
    {
      "epoch": 0.5480876034789899,
      "grad_norm": 0.5270583629608154,
      "learning_rate": 8.956515662561364e-06,
      "loss": 0.9386,
      "step": 3269
    },
    {
      "epoch": 0.5482552656397359,
      "grad_norm": 0.6502846479415894,
      "learning_rate": 8.951114879662455e-06,
      "loss": 0.9053,
      "step": 3270
    },
    {
      "epoch": 0.5484229278004821,
      "grad_norm": 0.3641306757926941,
      "learning_rate": 8.945714406092757e-06,
      "loss": 0.7675,
      "step": 3271
    },
    {
      "epoch": 0.5485905899612281,
      "grad_norm": 0.42551761865615845,
      "learning_rate": 8.940314243444933e-06,
      "loss": 0.8046,
      "step": 3272
    },
    {
      "epoch": 0.5487582521219743,
      "grad_norm": 0.4767121374607086,
      "learning_rate": 8.93491439331156e-06,
      "loss": 0.8365,
      "step": 3273
    },
    {
      "epoch": 0.5489259142827203,
      "grad_norm": 0.4678700864315033,
      "learning_rate": 8.929514857285122e-06,
      "loss": 0.8552,
      "step": 3274
    },
    {
      "epoch": 0.5490935764434665,
      "grad_norm": 0.6308174133300781,
      "learning_rate": 8.924115636958011e-06,
      "loss": 0.7933,
      "step": 3275
    },
    {
      "epoch": 0.5492612386042125,
      "grad_norm": 0.8057112097740173,
      "learning_rate": 8.918716733922519e-06,
      "loss": 0.9257,
      "step": 3276
    },
    {
      "epoch": 0.5494289007649586,
      "grad_norm": 0.7312862277030945,
      "learning_rate": 8.91331814977085e-06,
      "loss": 0.8304,
      "step": 3277
    },
    {
      "epoch": 0.5495965629257047,
      "grad_norm": 0.5104177594184875,
      "learning_rate": 8.907919886095115e-06,
      "loss": 0.8804,
      "step": 3278
    },
    {
      "epoch": 0.5497642250864508,
      "grad_norm": 0.4285842180252075,
      "learning_rate": 8.902521944487328e-06,
      "loss": 0.9018,
      "step": 3279
    },
    {
      "epoch": 0.5499318872471969,
      "grad_norm": 0.49149802327156067,
      "learning_rate": 8.897124326539415e-06,
      "loss": 0.7914,
      "step": 3280
    },
    {
      "epoch": 0.550099549407943,
      "grad_norm": 0.46300008893013,
      "learning_rate": 8.89172703384319e-06,
      "loss": 0.9107,
      "step": 3281
    },
    {
      "epoch": 0.5502672115686891,
      "grad_norm": 0.36543166637420654,
      "learning_rate": 8.886330067990385e-06,
      "loss": 0.8881,
      "step": 3282
    },
    {
      "epoch": 0.5504348737294352,
      "grad_norm": 0.4292105436325073,
      "learning_rate": 8.880933430572634e-06,
      "loss": 0.8234,
      "step": 3283
    },
    {
      "epoch": 0.5506025358901813,
      "grad_norm": 0.7258087396621704,
      "learning_rate": 8.875537123181473e-06,
      "loss": 0.7786,
      "step": 3284
    },
    {
      "epoch": 0.5507701980509274,
      "grad_norm": 0.6154667735099792,
      "learning_rate": 8.870141147408343e-06,
      "loss": 0.8669,
      "step": 3285
    },
    {
      "epoch": 0.5509378602116735,
      "grad_norm": 0.5430100560188293,
      "learning_rate": 8.864745504844572e-06,
      "loss": 0.8951,
      "step": 3286
    },
    {
      "epoch": 0.5511055223724196,
      "grad_norm": 0.6223335862159729,
      "learning_rate": 8.859350197081412e-06,
      "loss": 0.8975,
      "step": 3287
    },
    {
      "epoch": 0.5512731845331656,
      "grad_norm": 0.5001317262649536,
      "learning_rate": 8.853955225710003e-06,
      "loss": 0.8587,
      "step": 3288
    },
    {
      "epoch": 0.5514408466939118,
      "grad_norm": 0.4287770390510559,
      "learning_rate": 8.848560592321391e-06,
      "loss": 0.8479,
      "step": 3289
    },
    {
      "epoch": 0.5516085088546578,
      "grad_norm": 0.4979509711265564,
      "learning_rate": 8.843166298506517e-06,
      "loss": 0.7994,
      "step": 3290
    },
    {
      "epoch": 0.551776171015404,
      "grad_norm": 0.6394463181495667,
      "learning_rate": 8.837772345856226e-06,
      "loss": 0.852,
      "step": 3291
    },
    {
      "epoch": 0.55194383317615,
      "grad_norm": 0.4299682378768921,
      "learning_rate": 8.832378735961263e-06,
      "loss": 0.8318,
      "step": 3292
    },
    {
      "epoch": 0.5521114953368962,
      "grad_norm": 0.4377739727497101,
      "learning_rate": 8.826985470412267e-06,
      "loss": 0.8591,
      "step": 3293
    },
    {
      "epoch": 0.5522791574976422,
      "grad_norm": 0.3901938498020172,
      "learning_rate": 8.821592550799784e-06,
      "loss": 0.8548,
      "step": 3294
    },
    {
      "epoch": 0.5524468196583884,
      "grad_norm": 0.46280205249786377,
      "learning_rate": 8.816199978714249e-06,
      "loss": 0.8837,
      "step": 3295
    },
    {
      "epoch": 0.5526144818191344,
      "grad_norm": 0.44443899393081665,
      "learning_rate": 8.810807755745997e-06,
      "loss": 0.8733,
      "step": 3296
    },
    {
      "epoch": 0.5527821439798806,
      "grad_norm": 0.6398127675056458,
      "learning_rate": 8.805415883485264e-06,
      "loss": 0.8055,
      "step": 3297
    },
    {
      "epoch": 0.5529498061406266,
      "grad_norm": 0.5506210923194885,
      "learning_rate": 8.800024363522181e-06,
      "loss": 0.9219,
      "step": 3298
    },
    {
      "epoch": 0.5531174683013728,
      "grad_norm": 0.4390423595905304,
      "learning_rate": 8.79463319744677e-06,
      "loss": 0.89,
      "step": 3299
    },
    {
      "epoch": 0.5532851304621188,
      "grad_norm": 0.46664154529571533,
      "learning_rate": 8.789242386848958e-06,
      "loss": 0.906,
      "step": 3300
    },
    {
      "epoch": 0.553452792622865,
      "grad_norm": 0.3888428211212158,
      "learning_rate": 8.783851933318555e-06,
      "loss": 0.7911,
      "step": 3301
    },
    {
      "epoch": 0.553620454783611,
      "grad_norm": 0.6484602093696594,
      "learning_rate": 8.778461838445277e-06,
      "loss": 0.9439,
      "step": 3302
    },
    {
      "epoch": 0.5537881169443571,
      "grad_norm": 0.3052983283996582,
      "learning_rate": 8.773072103818734e-06,
      "loss": 0.8427,
      "step": 3303
    },
    {
      "epoch": 0.5539557791051032,
      "grad_norm": 0.42999911308288574,
      "learning_rate": 8.767682731028415e-06,
      "loss": 0.9214,
      "step": 3304
    },
    {
      "epoch": 0.5541234412658493,
      "grad_norm": 0.36539483070373535,
      "learning_rate": 8.762293721663718e-06,
      "loss": 0.9242,
      "step": 3305
    },
    {
      "epoch": 0.5542911034265954,
      "grad_norm": 0.5123153924942017,
      "learning_rate": 8.75690507731393e-06,
      "loss": 0.8203,
      "step": 3306
    },
    {
      "epoch": 0.5544587655873415,
      "grad_norm": 0.38076192140579224,
      "learning_rate": 8.75151679956823e-06,
      "loss": 0.8908,
      "step": 3307
    },
    {
      "epoch": 0.5546264277480876,
      "grad_norm": 0.5812100768089294,
      "learning_rate": 8.746128890015684e-06,
      "loss": 0.908,
      "step": 3308
    },
    {
      "epoch": 0.5547940899088337,
      "grad_norm": 0.3563908636569977,
      "learning_rate": 8.740741350245252e-06,
      "loss": 0.8301,
      "step": 3309
    },
    {
      "epoch": 0.5549617520695798,
      "grad_norm": 0.4139834940433502,
      "learning_rate": 8.73535418184579e-06,
      "loss": 0.8231,
      "step": 3310
    },
    {
      "epoch": 0.5551294142303259,
      "grad_norm": 0.6863387823104858,
      "learning_rate": 8.729967386406042e-06,
      "loss": 0.7782,
      "step": 3311
    },
    {
      "epoch": 0.555297076391072,
      "grad_norm": 0.774410605430603,
      "learning_rate": 8.724580965514638e-06,
      "loss": 0.8801,
      "step": 3312
    },
    {
      "epoch": 0.5554647385518181,
      "grad_norm": 0.37416234612464905,
      "learning_rate": 8.7191949207601e-06,
      "loss": 0.8831,
      "step": 3313
    },
    {
      "epoch": 0.5556324007125641,
      "grad_norm": 0.683139979839325,
      "learning_rate": 8.71380925373084e-06,
      "loss": 0.8318,
      "step": 3314
    },
    {
      "epoch": 0.5558000628733103,
      "grad_norm": 0.41312792897224426,
      "learning_rate": 8.708423966015156e-06,
      "loss": 0.8583,
      "step": 3315
    },
    {
      "epoch": 0.5559677250340563,
      "grad_norm": 0.41628390550613403,
      "learning_rate": 8.70303905920124e-06,
      "loss": 0.8398,
      "step": 3316
    },
    {
      "epoch": 0.5561353871948025,
      "grad_norm": 0.540746808052063,
      "learning_rate": 8.697654534877166e-06,
      "loss": 0.863,
      "step": 3317
    },
    {
      "epoch": 0.5563030493555485,
      "grad_norm": 0.398416668176651,
      "learning_rate": 8.692270394630894e-06,
      "loss": 0.8558,
      "step": 3318
    },
    {
      "epoch": 0.5564707115162947,
      "grad_norm": 0.41070878505706787,
      "learning_rate": 8.686886640050277e-06,
      "loss": 0.9208,
      "step": 3319
    },
    {
      "epoch": 0.5566383736770407,
      "grad_norm": 0.5760504603385925,
      "learning_rate": 8.68150327272305e-06,
      "loss": 0.8301,
      "step": 3320
    },
    {
      "epoch": 0.5568060358377869,
      "grad_norm": 0.4808904230594635,
      "learning_rate": 8.676120294236836e-06,
      "loss": 0.9025,
      "step": 3321
    },
    {
      "epoch": 0.5569736979985329,
      "grad_norm": 0.45318806171417236,
      "learning_rate": 8.670737706179138e-06,
      "loss": 0.8135,
      "step": 3322
    },
    {
      "epoch": 0.5571413601592791,
      "grad_norm": 0.5390448570251465,
      "learning_rate": 8.665355510137351e-06,
      "loss": 0.9409,
      "step": 3323
    },
    {
      "epoch": 0.5573090223200251,
      "grad_norm": 0.4017261266708374,
      "learning_rate": 8.659973707698751e-06,
      "loss": 0.8512,
      "step": 3324
    },
    {
      "epoch": 0.5574766844807713,
      "grad_norm": 0.4574022889137268,
      "learning_rate": 8.654592300450496e-06,
      "loss": 0.8309,
      "step": 3325
    },
    {
      "epoch": 0.5576443466415173,
      "grad_norm": 0.5664092302322388,
      "learning_rate": 8.649211289979638e-06,
      "loss": 0.9195,
      "step": 3326
    },
    {
      "epoch": 0.5578120088022634,
      "grad_norm": 0.41941285133361816,
      "learning_rate": 8.643830677873092e-06,
      "loss": 0.8546,
      "step": 3327
    },
    {
      "epoch": 0.5579796709630095,
      "grad_norm": 0.5523105263710022,
      "learning_rate": 8.638450465717671e-06,
      "loss": 0.82,
      "step": 3328
    },
    {
      "epoch": 0.5581473331237556,
      "grad_norm": 0.4249464273452759,
      "learning_rate": 8.633070655100069e-06,
      "loss": 0.9138,
      "step": 3329
    },
    {
      "epoch": 0.5583149952845017,
      "grad_norm": 0.6280999183654785,
      "learning_rate": 8.627691247606862e-06,
      "loss": 0.8309,
      "step": 3330
    },
    {
      "epoch": 0.5584826574452478,
      "grad_norm": 0.41287532448768616,
      "learning_rate": 8.622312244824492e-06,
      "loss": 0.8317,
      "step": 3331
    },
    {
      "epoch": 0.5586503196059939,
      "grad_norm": 0.5640285015106201,
      "learning_rate": 8.616933648339303e-06,
      "loss": 0.935,
      "step": 3332
    },
    {
      "epoch": 0.55881798176674,
      "grad_norm": 0.38837930560112,
      "learning_rate": 8.61155545973751e-06,
      "loss": 0.8126,
      "step": 3333
    },
    {
      "epoch": 0.5589856439274861,
      "grad_norm": 0.4220353364944458,
      "learning_rate": 8.6061776806052e-06,
      "loss": 0.8232,
      "step": 3334
    },
    {
      "epoch": 0.5591533060882322,
      "grad_norm": 0.4612191617488861,
      "learning_rate": 8.600800312528356e-06,
      "loss": 0.8701,
      "step": 3335
    },
    {
      "epoch": 0.5593209682489783,
      "grad_norm": 1.0260591506958008,
      "learning_rate": 8.595423357092823e-06,
      "loss": 0.9243,
      "step": 3336
    },
    {
      "epoch": 0.5594886304097244,
      "grad_norm": 0.4499152600765228,
      "learning_rate": 8.590046815884337e-06,
      "loss": 0.8106,
      "step": 3337
    },
    {
      "epoch": 0.5596562925704704,
      "grad_norm": 0.6131991147994995,
      "learning_rate": 8.584670690488501e-06,
      "loss": 0.8969,
      "step": 3338
    },
    {
      "epoch": 0.5598239547312166,
      "grad_norm": 0.9590579271316528,
      "learning_rate": 8.579294982490808e-06,
      "loss": 0.9169,
      "step": 3339
    },
    {
      "epoch": 0.5599916168919626,
      "grad_norm": 0.409368097782135,
      "learning_rate": 8.573919693476614e-06,
      "loss": 0.8295,
      "step": 3340
    },
    {
      "epoch": 0.5601592790527088,
      "grad_norm": 0.4588213264942169,
      "learning_rate": 8.568544825031163e-06,
      "loss": 0.8349,
      "step": 3341
    },
    {
      "epoch": 0.5603269412134549,
      "grad_norm": 0.6115896701812744,
      "learning_rate": 8.563170378739567e-06,
      "loss": 0.8692,
      "step": 3342
    },
    {
      "epoch": 0.560494603374201,
      "grad_norm": 0.371785044670105,
      "learning_rate": 8.557796356186818e-06,
      "loss": 0.8572,
      "step": 3343
    },
    {
      "epoch": 0.5606622655349471,
      "grad_norm": 0.38996487855911255,
      "learning_rate": 8.552422758957785e-06,
      "loss": 0.796,
      "step": 3344
    },
    {
      "epoch": 0.5608299276956932,
      "grad_norm": 0.42374661564826965,
      "learning_rate": 8.547049588637204e-06,
      "loss": 0.963,
      "step": 3345
    },
    {
      "epoch": 0.5609975898564393,
      "grad_norm": 0.37725475430488586,
      "learning_rate": 8.54167684680969e-06,
      "loss": 0.8658,
      "step": 3346
    },
    {
      "epoch": 0.5611652520171854,
      "grad_norm": 0.5238369107246399,
      "learning_rate": 8.536304535059731e-06,
      "loss": 0.8589,
      "step": 3347
    },
    {
      "epoch": 0.5613329141779315,
      "grad_norm": 0.4647602140903473,
      "learning_rate": 8.530932654971693e-06,
      "loss": 0.8395,
      "step": 3348
    },
    {
      "epoch": 0.5615005763386776,
      "grad_norm": 0.3794912099838257,
      "learning_rate": 8.525561208129805e-06,
      "loss": 0.8179,
      "step": 3349
    },
    {
      "epoch": 0.5616682384994237,
      "grad_norm": 0.4015001654624939,
      "learning_rate": 8.520190196118173e-06,
      "loss": 0.7695,
      "step": 3350
    },
    {
      "epoch": 0.5618359006601698,
      "grad_norm": 0.4131832420825958,
      "learning_rate": 8.514819620520773e-06,
      "loss": 0.8353,
      "step": 3351
    },
    {
      "epoch": 0.5620035628209159,
      "grad_norm": 0.44537413120269775,
      "learning_rate": 8.509449482921462e-06,
      "loss": 0.8112,
      "step": 3352
    },
    {
      "epoch": 0.5621712249816619,
      "grad_norm": 0.5103172659873962,
      "learning_rate": 8.50407978490396e-06,
      "loss": 0.9033,
      "step": 3353
    },
    {
      "epoch": 0.5623388871424081,
      "grad_norm": 0.4940226674079895,
      "learning_rate": 8.498710528051845e-06,
      "loss": 0.9388,
      "step": 3354
    },
    {
      "epoch": 0.5625065493031541,
      "grad_norm": 0.5164597630500793,
      "learning_rate": 8.493341713948588e-06,
      "loss": 0.9151,
      "step": 3355
    },
    {
      "epoch": 0.5626742114639003,
      "grad_norm": 0.41803959012031555,
      "learning_rate": 8.487973344177517e-06,
      "loss": 0.9144,
      "step": 3356
    },
    {
      "epoch": 0.5628418736246463,
      "grad_norm": 0.4546486735343933,
      "learning_rate": 8.482605420321829e-06,
      "loss": 0.9372,
      "step": 3357
    },
    {
      "epoch": 0.5630095357853925,
      "grad_norm": 0.3524572253227234,
      "learning_rate": 8.477237943964596e-06,
      "loss": 0.8226,
      "step": 3358
    },
    {
      "epoch": 0.5631771979461385,
      "grad_norm": 0.5530046224594116,
      "learning_rate": 8.471870916688745e-06,
      "loss": 0.8778,
      "step": 3359
    },
    {
      "epoch": 0.5633448601068847,
      "grad_norm": 0.45319312810897827,
      "learning_rate": 8.466504340077084e-06,
      "loss": 0.8568,
      "step": 3360
    },
    {
      "epoch": 0.5635125222676307,
      "grad_norm": 0.6729868650436401,
      "learning_rate": 8.461138215712283e-06,
      "loss": 0.917,
      "step": 3361
    },
    {
      "epoch": 0.5636801844283769,
      "grad_norm": 0.461455374956131,
      "learning_rate": 8.45577254517688e-06,
      "loss": 0.7528,
      "step": 3362
    },
    {
      "epoch": 0.5638478465891229,
      "grad_norm": 0.9580578804016113,
      "learning_rate": 8.450407330053271e-06,
      "loss": 0.8159,
      "step": 3363
    },
    {
      "epoch": 0.564015508749869,
      "grad_norm": 0.5438309907913208,
      "learning_rate": 8.44504257192373e-06,
      "loss": 0.7513,
      "step": 3364
    },
    {
      "epoch": 0.5641831709106151,
      "grad_norm": 0.6372382044792175,
      "learning_rate": 8.439678272370391e-06,
      "loss": 0.9019,
      "step": 3365
    },
    {
      "epoch": 0.5643508330713612,
      "grad_norm": 0.6940456628799438,
      "learning_rate": 8.434314432975248e-06,
      "loss": 0.8214,
      "step": 3366
    },
    {
      "epoch": 0.5645184952321073,
      "grad_norm": 0.40165215730667114,
      "learning_rate": 8.428951055320168e-06,
      "loss": 0.8752,
      "step": 3367
    },
    {
      "epoch": 0.5646861573928534,
      "grad_norm": 0.5314671993255615,
      "learning_rate": 8.423588140986876e-06,
      "loss": 0.7734,
      "step": 3368
    },
    {
      "epoch": 0.5648538195535995,
      "grad_norm": 0.3801604211330414,
      "learning_rate": 8.418225691556962e-06,
      "loss": 0.8735,
      "step": 3369
    },
    {
      "epoch": 0.5650214817143456,
      "grad_norm": 0.3730860948562622,
      "learning_rate": 8.412863708611874e-06,
      "loss": 0.8734,
      "step": 3370
    },
    {
      "epoch": 0.5651891438750917,
      "grad_norm": 0.7912408709526062,
      "learning_rate": 8.40750219373294e-06,
      "loss": 0.9414,
      "step": 3371
    },
    {
      "epoch": 0.5653568060358378,
      "grad_norm": 0.46992161870002747,
      "learning_rate": 8.402141148501323e-06,
      "loss": 0.9592,
      "step": 3372
    },
    {
      "epoch": 0.5655244681965839,
      "grad_norm": 0.660154402256012,
      "learning_rate": 8.396780574498066e-06,
      "loss": 0.9717,
      "step": 3373
    },
    {
      "epoch": 0.56569213035733,
      "grad_norm": 0.6320633292198181,
      "learning_rate": 8.391420473304071e-06,
      "loss": 0.8456,
      "step": 3374
    },
    {
      "epoch": 0.565859792518076,
      "grad_norm": 0.5591331720352173,
      "learning_rate": 8.386060846500097e-06,
      "loss": 0.7859,
      "step": 3375
    },
    {
      "epoch": 0.5660274546788222,
      "grad_norm": 0.43963560461997986,
      "learning_rate": 8.38070169566677e-06,
      "loss": 0.887,
      "step": 3376
    },
    {
      "epoch": 0.5661951168395682,
      "grad_norm": 0.3942304849624634,
      "learning_rate": 8.375343022384559e-06,
      "loss": 0.8051,
      "step": 3377
    },
    {
      "epoch": 0.5663627790003144,
      "grad_norm": 0.6130456924438477,
      "learning_rate": 8.369984828233807e-06,
      "loss": 0.8517,
      "step": 3378
    },
    {
      "epoch": 0.5665304411610604,
      "grad_norm": 0.4271846413612366,
      "learning_rate": 8.364627114794716e-06,
      "loss": 0.8557,
      "step": 3379
    },
    {
      "epoch": 0.5666981033218066,
      "grad_norm": 0.44319701194763184,
      "learning_rate": 8.35926988364734e-06,
      "loss": 0.8519,
      "step": 3380
    },
    {
      "epoch": 0.5668657654825526,
      "grad_norm": 0.5144001245498657,
      "learning_rate": 8.353913136371592e-06,
      "loss": 0.9553,
      "step": 3381
    },
    {
      "epoch": 0.5670334276432988,
      "grad_norm": 0.4724206030368805,
      "learning_rate": 8.348556874547242e-06,
      "loss": 0.8115,
      "step": 3382
    },
    {
      "epoch": 0.5672010898040448,
      "grad_norm": 0.5161254405975342,
      "learning_rate": 8.343201099753921e-06,
      "loss": 0.9335,
      "step": 3383
    },
    {
      "epoch": 0.567368751964791,
      "grad_norm": 0.7193112969398499,
      "learning_rate": 8.337845813571109e-06,
      "loss": 0.9,
      "step": 3384
    },
    {
      "epoch": 0.567536414125537,
      "grad_norm": 0.6863871216773987,
      "learning_rate": 8.332491017578152e-06,
      "loss": 0.8844,
      "step": 3385
    },
    {
      "epoch": 0.5677040762862832,
      "grad_norm": 0.5251758098602295,
      "learning_rate": 8.32713671335424e-06,
      "loss": 0.7751,
      "step": 3386
    },
    {
      "epoch": 0.5678717384470292,
      "grad_norm": 0.5140281915664673,
      "learning_rate": 8.321782902478426e-06,
      "loss": 0.8754,
      "step": 3387
    },
    {
      "epoch": 0.5680394006077754,
      "grad_norm": 0.6101822853088379,
      "learning_rate": 8.316429586529616e-06,
      "loss": 0.8659,
      "step": 3388
    },
    {
      "epoch": 0.5682070627685214,
      "grad_norm": 0.4696868062019348,
      "learning_rate": 8.311076767086566e-06,
      "loss": 0.8208,
      "step": 3389
    },
    {
      "epoch": 0.5683747249292675,
      "grad_norm": 0.6828814148902893,
      "learning_rate": 8.305724445727896e-06,
      "loss": 0.8719,
      "step": 3390
    },
    {
      "epoch": 0.5685423870900136,
      "grad_norm": 0.3969288170337677,
      "learning_rate": 8.300372624032062e-06,
      "loss": 0.877,
      "step": 3391
    },
    {
      "epoch": 0.5687100492507597,
      "grad_norm": 0.6055564284324646,
      "learning_rate": 8.29502130357739e-06,
      "loss": 0.8484,
      "step": 3392
    },
    {
      "epoch": 0.5688777114115058,
      "grad_norm": 0.47719958424568176,
      "learning_rate": 8.289670485942044e-06,
      "loss": 1.0181,
      "step": 3393
    },
    {
      "epoch": 0.5690453735722519,
      "grad_norm": 0.43543651700019836,
      "learning_rate": 8.284320172704058e-06,
      "loss": 0.835,
      "step": 3394
    },
    {
      "epoch": 0.569213035732998,
      "grad_norm": 0.6597638130187988,
      "learning_rate": 8.278970365441292e-06,
      "loss": 0.8011,
      "step": 3395
    },
    {
      "epoch": 0.5693806978937441,
      "grad_norm": 0.812634289264679,
      "learning_rate": 8.273621065731479e-06,
      "loss": 0.7814,
      "step": 3396
    },
    {
      "epoch": 0.5695483600544902,
      "grad_norm": 0.45590394735336304,
      "learning_rate": 8.268272275152187e-06,
      "loss": 0.8903,
      "step": 3397
    },
    {
      "epoch": 0.5697160222152363,
      "grad_norm": 0.4711604118347168,
      "learning_rate": 8.262923995280847e-06,
      "loss": 0.8244,
      "step": 3398
    },
    {
      "epoch": 0.5698836843759824,
      "grad_norm": 0.6121546626091003,
      "learning_rate": 8.257576227694736e-06,
      "loss": 0.8225,
      "step": 3399
    },
    {
      "epoch": 0.5700513465367285,
      "grad_norm": 0.9810781478881836,
      "learning_rate": 8.252228973970964e-06,
      "loss": 0.8684,
      "step": 3400
    },
    {
      "epoch": 0.5702190086974745,
      "grad_norm": 0.39338693022727966,
      "learning_rate": 8.246882235686512e-06,
      "loss": 0.9007,
      "step": 3401
    },
    {
      "epoch": 0.5703866708582207,
      "grad_norm": 0.448321133852005,
      "learning_rate": 8.241536014418199e-06,
      "loss": 0.8147,
      "step": 3402
    },
    {
      "epoch": 0.5705543330189667,
      "grad_norm": 0.38478726148605347,
      "learning_rate": 8.23619031174269e-06,
      "loss": 0.793,
      "step": 3403
    },
    {
      "epoch": 0.5707219951797129,
      "grad_norm": 0.4584026038646698,
      "learning_rate": 8.230845129236498e-06,
      "loss": 0.7901,
      "step": 3404
    },
    {
      "epoch": 0.5708896573404589,
      "grad_norm": 0.41249966621398926,
      "learning_rate": 8.225500468475984e-06,
      "loss": 0.8622,
      "step": 3405
    },
    {
      "epoch": 0.5710573195012051,
      "grad_norm": 0.7865109443664551,
      "learning_rate": 8.220156331037355e-06,
      "loss": 0.9252,
      "step": 3406
    },
    {
      "epoch": 0.5712249816619511,
      "grad_norm": 0.4040682315826416,
      "learning_rate": 8.214812718496666e-06,
      "loss": 0.8741,
      "step": 3407
    },
    {
      "epoch": 0.5713926438226973,
      "grad_norm": 0.6797685027122498,
      "learning_rate": 8.209469632429811e-06,
      "loss": 0.8442,
      "step": 3408
    },
    {
      "epoch": 0.5715603059834433,
      "grad_norm": 0.40794631838798523,
      "learning_rate": 8.204127074412531e-06,
      "loss": 0.8695,
      "step": 3409
    },
    {
      "epoch": 0.5717279681441895,
      "grad_norm": 0.3528442680835724,
      "learning_rate": 8.198785046020417e-06,
      "loss": 0.8551,
      "step": 3410
    },
    {
      "epoch": 0.5718956303049355,
      "grad_norm": 0.4704998731613159,
      "learning_rate": 8.193443548828896e-06,
      "loss": 0.7734,
      "step": 3411
    },
    {
      "epoch": 0.5720632924656817,
      "grad_norm": 0.4955574870109558,
      "learning_rate": 8.188102584413244e-06,
      "loss": 0.7795,
      "step": 3412
    },
    {
      "epoch": 0.5722309546264277,
      "grad_norm": 0.5632057189941406,
      "learning_rate": 8.182762154348575e-06,
      "loss": 0.8672,
      "step": 3413
    },
    {
      "epoch": 0.5723986167871739,
      "grad_norm": 0.4081422686576843,
      "learning_rate": 8.177422260209849e-06,
      "loss": 0.8319,
      "step": 3414
    },
    {
      "epoch": 0.5725662789479199,
      "grad_norm": 0.44355064630508423,
      "learning_rate": 8.172082903571868e-06,
      "loss": 0.8316,
      "step": 3415
    },
    {
      "epoch": 0.572733941108666,
      "grad_norm": 0.41933557391166687,
      "learning_rate": 8.16674408600927e-06,
      "loss": 0.904,
      "step": 3416
    },
    {
      "epoch": 0.5729016032694121,
      "grad_norm": 0.47249361872673035,
      "learning_rate": 8.161405809096547e-06,
      "loss": 0.86,
      "step": 3417
    },
    {
      "epoch": 0.5730692654301582,
      "grad_norm": 0.4084494113922119,
      "learning_rate": 8.156068074408016e-06,
      "loss": 0.9057,
      "step": 3418
    },
    {
      "epoch": 0.5732369275909043,
      "grad_norm": 0.45873379707336426,
      "learning_rate": 8.150730883517841e-06,
      "loss": 0.802,
      "step": 3419
    },
    {
      "epoch": 0.5734045897516504,
      "grad_norm": 0.8427038192749023,
      "learning_rate": 8.145394238000028e-06,
      "loss": 0.9047,
      "step": 3420
    },
    {
      "epoch": 0.5735722519123965,
      "grad_norm": 0.33546656370162964,
      "learning_rate": 8.140058139428425e-06,
      "loss": 0.8843,
      "step": 3421
    },
    {
      "epoch": 0.5737399140731426,
      "grad_norm": 0.40231165289878845,
      "learning_rate": 8.134722589376704e-06,
      "loss": 0.8627,
      "step": 3422
    },
    {
      "epoch": 0.5739075762338887,
      "grad_norm": 0.40609556436538696,
      "learning_rate": 8.129387589418385e-06,
      "loss": 0.8711,
      "step": 3423
    },
    {
      "epoch": 0.5740752383946348,
      "grad_norm": 0.4238656163215637,
      "learning_rate": 8.124053141126833e-06,
      "loss": 0.9049,
      "step": 3424
    },
    {
      "epoch": 0.5742429005553809,
      "grad_norm": 0.9477406740188599,
      "learning_rate": 8.118719246075239e-06,
      "loss": 0.8209,
      "step": 3425
    },
    {
      "epoch": 0.574410562716127,
      "grad_norm": 0.44262930750846863,
      "learning_rate": 8.113385905836636e-06,
      "loss": 0.8028,
      "step": 3426
    },
    {
      "epoch": 0.5745782248768732,
      "grad_norm": 0.7265502214431763,
      "learning_rate": 8.10805312198389e-06,
      "loss": 0.8735,
      "step": 3427
    },
    {
      "epoch": 0.5747458870376192,
      "grad_norm": 0.42299774289131165,
      "learning_rate": 8.102720896089705e-06,
      "loss": 0.8704,
      "step": 3428
    },
    {
      "epoch": 0.5749135491983653,
      "grad_norm": 0.663567841053009,
      "learning_rate": 8.097389229726623e-06,
      "loss": 0.944,
      "step": 3429
    },
    {
      "epoch": 0.5750812113591114,
      "grad_norm": 0.9425939917564392,
      "learning_rate": 8.092058124467014e-06,
      "loss": 0.8729,
      "step": 3430
    },
    {
      "epoch": 0.5752488735198575,
      "grad_norm": 0.7026984691619873,
      "learning_rate": 8.086727581883094e-06,
      "loss": 0.8146,
      "step": 3431
    },
    {
      "epoch": 0.5754165356806036,
      "grad_norm": 0.38072091341018677,
      "learning_rate": 8.0813976035469e-06,
      "loss": 0.8646,
      "step": 3432
    },
    {
      "epoch": 0.5755841978413497,
      "grad_norm": 0.3900136947631836,
      "learning_rate": 8.076068191030309e-06,
      "loss": 0.8507,
      "step": 3433
    },
    {
      "epoch": 0.5757518600020958,
      "grad_norm": 0.41140100359916687,
      "learning_rate": 8.070739345905032e-06,
      "loss": 0.8211,
      "step": 3434
    },
    {
      "epoch": 0.5759195221628419,
      "grad_norm": 0.3828379809856415,
      "learning_rate": 8.065411069742615e-06,
      "loss": 0.8792,
      "step": 3435
    },
    {
      "epoch": 0.576087184323588,
      "grad_norm": 0.5758510828018188,
      "learning_rate": 8.060083364114427e-06,
      "loss": 0.9086,
      "step": 3436
    },
    {
      "epoch": 0.5762548464843341,
      "grad_norm": 0.6231465935707092,
      "learning_rate": 8.054756230591676e-06,
      "loss": 1.0062,
      "step": 3437
    },
    {
      "epoch": 0.5764225086450802,
      "grad_norm": 0.42408302426338196,
      "learning_rate": 8.0494296707454e-06,
      "loss": 0.8886,
      "step": 3438
    },
    {
      "epoch": 0.5765901708058263,
      "grad_norm": 0.39178594946861267,
      "learning_rate": 8.044103686146465e-06,
      "loss": 0.7908,
      "step": 3439
    },
    {
      "epoch": 0.5767578329665723,
      "grad_norm": 0.35958993434906006,
      "learning_rate": 8.03877827836558e-06,
      "loss": 0.7988,
      "step": 3440
    },
    {
      "epoch": 0.5769254951273185,
      "grad_norm": 0.5605311393737793,
      "learning_rate": 8.033453448973261e-06,
      "loss": 0.9095,
      "step": 3441
    },
    {
      "epoch": 0.5770931572880645,
      "grad_norm": 0.571583092212677,
      "learning_rate": 8.02812919953987e-06,
      "loss": 0.8901,
      "step": 3442
    },
    {
      "epoch": 0.5772608194488107,
      "grad_norm": 0.630528450012207,
      "learning_rate": 8.022805531635599e-06,
      "loss": 0.8026,
      "step": 3443
    },
    {
      "epoch": 0.5774284816095567,
      "grad_norm": 0.4706905782222748,
      "learning_rate": 8.01748244683046e-06,
      "loss": 0.893,
      "step": 3444
    },
    {
      "epoch": 0.5775961437703029,
      "grad_norm": 0.6315483450889587,
      "learning_rate": 8.012159946694295e-06,
      "loss": 0.8625,
      "step": 3445
    },
    {
      "epoch": 0.5777638059310489,
      "grad_norm": 0.3732752799987793,
      "learning_rate": 8.006838032796775e-06,
      "loss": 0.8929,
      "step": 3446
    },
    {
      "epoch": 0.5779314680917951,
      "grad_norm": 0.4668952524662018,
      "learning_rate": 8.001516706707401e-06,
      "loss": 0.8655,
      "step": 3447
    },
    {
      "epoch": 0.5780991302525411,
      "grad_norm": 0.4400523900985718,
      "learning_rate": 7.996195969995498e-06,
      "loss": 0.8342,
      "step": 3448
    },
    {
      "epoch": 0.5782667924132873,
      "grad_norm": 0.5376847386360168,
      "learning_rate": 7.990875824230217e-06,
      "loss": 0.8972,
      "step": 3449
    },
    {
      "epoch": 0.5784344545740333,
      "grad_norm": 0.5515185594558716,
      "learning_rate": 7.985556270980534e-06,
      "loss": 0.8021,
      "step": 3450
    },
    {
      "epoch": 0.5786021167347795,
      "grad_norm": 0.35805997252464294,
      "learning_rate": 7.98023731181525e-06,
      "loss": 0.8116,
      "step": 3451
    },
    {
      "epoch": 0.5787697788955255,
      "grad_norm": 0.5249135494232178,
      "learning_rate": 7.974918948302993e-06,
      "loss": 0.9464,
      "step": 3452
    },
    {
      "epoch": 0.5789374410562717,
      "grad_norm": 0.5082404017448425,
      "learning_rate": 7.969601182012217e-06,
      "loss": 0.8521,
      "step": 3453
    },
    {
      "epoch": 0.5791051032170177,
      "grad_norm": 0.37490904331207275,
      "learning_rate": 7.964284014511193e-06,
      "loss": 0.7826,
      "step": 3454
    },
    {
      "epoch": 0.5792727653777638,
      "grad_norm": 0.3770364224910736,
      "learning_rate": 7.958967447368018e-06,
      "loss": 0.8681,
      "step": 3455
    },
    {
      "epoch": 0.5794404275385099,
      "grad_norm": 0.3770364224910736,
      "learning_rate": 7.958967447368018e-06,
      "loss": 0.7582,
      "step": 3456
    },
    {
      "epoch": 0.579608089699256,
      "grad_norm": 0.8216555118560791,
      "learning_rate": 7.953651482150619e-06,
      "loss": 0.8643,
      "step": 3457
    },
    {
      "epoch": 0.5797757518600021,
      "grad_norm": 0.4499664604663849,
      "learning_rate": 7.948336120426736e-06,
      "loss": 0.828,
      "step": 3458
    },
    {
      "epoch": 0.5799434140207482,
      "grad_norm": 0.3453725278377533,
      "learning_rate": 7.943021363763941e-06,
      "loss": 0.8029,
      "step": 3459
    },
    {
      "epoch": 0.5801110761814943,
      "grad_norm": 0.5152020454406738,
      "learning_rate": 7.937707213729612e-06,
      "loss": 0.8581,
      "step": 3460
    },
    {
      "epoch": 0.5802787383422404,
      "grad_norm": 0.44359079003334045,
      "learning_rate": 7.932393671890965e-06,
      "loss": 0.847,
      "step": 3461
    },
    {
      "epoch": 0.5804464005029865,
      "grad_norm": 0.6569787859916687,
      "learning_rate": 7.927080739815025e-06,
      "loss": 0.8359,
      "step": 3462
    },
    {
      "epoch": 0.5806140626637326,
      "grad_norm": 0.49237895011901855,
      "learning_rate": 7.921768419068643e-06,
      "loss": 0.8377,
      "step": 3463
    },
    {
      "epoch": 0.5807817248244787,
      "grad_norm": 0.34828248620033264,
      "learning_rate": 7.916456711218494e-06,
      "loss": 0.8696,
      "step": 3464
    },
    {
      "epoch": 0.5809493869852248,
      "grad_norm": 0.6898155808448792,
      "learning_rate": 7.911145617831057e-06,
      "loss": 0.8389,
      "step": 3465
    },
    {
      "epoch": 0.5811170491459708,
      "grad_norm": 0.5621565580368042,
      "learning_rate": 7.90583514047264e-06,
      "loss": 0.878,
      "step": 3466
    },
    {
      "epoch": 0.581284711306717,
      "grad_norm": 0.7397379279136658,
      "learning_rate": 7.900525280709375e-06,
      "loss": 0.875,
      "step": 3467
    },
    {
      "epoch": 0.581452373467463,
      "grad_norm": 0.39626482129096985,
      "learning_rate": 7.895216040107207e-06,
      "loss": 0.7933,
      "step": 3468
    },
    {
      "epoch": 0.5816200356282092,
      "grad_norm": 0.573769211769104,
      "learning_rate": 7.889907420231884e-06,
      "loss": 0.8246,
      "step": 3469
    },
    {
      "epoch": 0.5817876977889552,
      "grad_norm": 0.42165836691856384,
      "learning_rate": 7.884599422648995e-06,
      "loss": 0.8142,
      "step": 3470
    },
    {
      "epoch": 0.5819553599497014,
      "grad_norm": 0.48622065782546997,
      "learning_rate": 7.87929204892393e-06,
      "loss": 0.8257,
      "step": 3471
    },
    {
      "epoch": 0.5821230221104474,
      "grad_norm": 0.5852141380310059,
      "learning_rate": 7.873985300621902e-06,
      "loss": 0.8334,
      "step": 3472
    },
    {
      "epoch": 0.5822906842711936,
      "grad_norm": 0.46930983662605286,
      "learning_rate": 7.868679179307934e-06,
      "loss": 0.7885,
      "step": 3473
    },
    {
      "epoch": 0.5824583464319396,
      "grad_norm": 0.6163612604141235,
      "learning_rate": 7.863373686546868e-06,
      "loss": 0.8423,
      "step": 3474
    },
    {
      "epoch": 0.5826260085926858,
      "grad_norm": 0.6135543584823608,
      "learning_rate": 7.858068823903358e-06,
      "loss": 0.8644,
      "step": 3475
    },
    {
      "epoch": 0.5827936707534318,
      "grad_norm": 0.46974796056747437,
      "learning_rate": 7.852764592941876e-06,
      "loss": 0.8734,
      "step": 3476
    },
    {
      "epoch": 0.582961332914178,
      "grad_norm": 0.4672534465789795,
      "learning_rate": 7.847460995226706e-06,
      "loss": 0.7436,
      "step": 3477
    },
    {
      "epoch": 0.583128995074924,
      "grad_norm": 0.4216882586479187,
      "learning_rate": 7.84215803232194e-06,
      "loss": 0.922,
      "step": 3478
    },
    {
      "epoch": 0.5832966572356701,
      "grad_norm": 0.6704283356666565,
      "learning_rate": 7.836855705791492e-06,
      "loss": 0.8957,
      "step": 3479
    },
    {
      "epoch": 0.5834643193964162,
      "grad_norm": 0.46347349882125854,
      "learning_rate": 7.83155401719908e-06,
      "loss": 0.9365,
      "step": 3480
    },
    {
      "epoch": 0.5836319815571623,
      "grad_norm": 0.32277539372444153,
      "learning_rate": 7.826252968108242e-06,
      "loss": 0.9403,
      "step": 3481
    },
    {
      "epoch": 0.5837996437179084,
      "grad_norm": 0.4305226504802704,
      "learning_rate": 7.820952560082322e-06,
      "loss": 0.8092,
      "step": 3482
    },
    {
      "epoch": 0.5839673058786545,
      "grad_norm": 0.7612534761428833,
      "learning_rate": 7.815652794684472e-06,
      "loss": 0.7739,
      "step": 3483
    },
    {
      "epoch": 0.5841349680394006,
      "grad_norm": 0.5855340361595154,
      "learning_rate": 7.810353673477664e-06,
      "loss": 0.8841,
      "step": 3484
    },
    {
      "epoch": 0.5843026302001467,
      "grad_norm": 0.5339702367782593,
      "learning_rate": 7.805055198024668e-06,
      "loss": 0.9109,
      "step": 3485
    },
    {
      "epoch": 0.5844702923608928,
      "grad_norm": 0.39810431003570557,
      "learning_rate": 7.799757369888079e-06,
      "loss": 0.9111,
      "step": 3486
    },
    {
      "epoch": 0.5846379545216389,
      "grad_norm": 0.46095889806747437,
      "learning_rate": 7.794460190630283e-06,
      "loss": 0.8265,
      "step": 3487
    },
    {
      "epoch": 0.584805616682385,
      "grad_norm": 0.44814544916152954,
      "learning_rate": 7.789163661813486e-06,
      "loss": 0.8003,
      "step": 3488
    },
    {
      "epoch": 0.5849732788431311,
      "grad_norm": 0.4918145537376404,
      "learning_rate": 7.783867784999701e-06,
      "loss": 0.8374,
      "step": 3489
    },
    {
      "epoch": 0.5851409410038771,
      "grad_norm": 0.7340918183326721,
      "learning_rate": 7.778572561750752e-06,
      "loss": 0.8805,
      "step": 3490
    },
    {
      "epoch": 0.5853086031646233,
      "grad_norm": 0.3280177414417267,
      "learning_rate": 7.773277993628265e-06,
      "loss": 0.8806,
      "step": 3491
    },
    {
      "epoch": 0.5854762653253693,
      "grad_norm": 0.3526843786239624,
      "learning_rate": 7.767984082193662e-06,
      "loss": 0.7938,
      "step": 3492
    },
    {
      "epoch": 0.5856439274861155,
      "grad_norm": 0.3464638292789459,
      "learning_rate": 7.762690829008197e-06,
      "loss": 0.9061,
      "step": 3493
    },
    {
      "epoch": 0.5858115896468615,
      "grad_norm": 0.46480873227119446,
      "learning_rate": 7.75739823563291e-06,
      "loss": 0.7788,
      "step": 3494
    },
    {
      "epoch": 0.5859792518076077,
      "grad_norm": 0.411076158285141,
      "learning_rate": 7.752106303628656e-06,
      "loss": 0.8505,
      "step": 3495
    },
    {
      "epoch": 0.5861469139683537,
      "grad_norm": 0.5689971446990967,
      "learning_rate": 7.746815034556084e-06,
      "loss": 0.7777,
      "step": 3496
    },
    {
      "epoch": 0.5863145761290999,
      "grad_norm": 0.9616680145263672,
      "learning_rate": 7.74152442997566e-06,
      "loss": 0.8638,
      "step": 3497
    },
    {
      "epoch": 0.5864822382898459,
      "grad_norm": 1.2664830684661865,
      "learning_rate": 7.736234491447649e-06,
      "loss": 0.8451,
      "step": 3498
    },
    {
      "epoch": 0.5866499004505921,
      "grad_norm": 1.1053142547607422,
      "learning_rate": 7.730945220532116e-06,
      "loss": 0.9524,
      "step": 3499
    },
    {
      "epoch": 0.5868175626113381,
      "grad_norm": 0.40867578983306885,
      "learning_rate": 7.725656618788938e-06,
      "loss": 0.7747,
      "step": 3500
    },
    {
      "epoch": 0.5869852247720843,
      "grad_norm": 0.6661640405654907,
      "learning_rate": 7.720368687777784e-06,
      "loss": 0.8648,
      "step": 3501
    },
    {
      "epoch": 0.5871528869328303,
      "grad_norm": 0.39013051986694336,
      "learning_rate": 7.715081429058132e-06,
      "loss": 0.7754,
      "step": 3502
    },
    {
      "epoch": 0.5873205490935764,
      "grad_norm": 0.47789233922958374,
      "learning_rate": 7.709794844189262e-06,
      "loss": 0.9175,
      "step": 3503
    },
    {
      "epoch": 0.5874882112543225,
      "grad_norm": 0.6550090312957764,
      "learning_rate": 7.70450893473025e-06,
      "loss": 0.8314,
      "step": 3504
    },
    {
      "epoch": 0.5876558734150686,
      "grad_norm": 0.5416508913040161,
      "learning_rate": 7.69922370223998e-06,
      "loss": 0.9176,
      "step": 3505
    },
    {
      "epoch": 0.5878235355758147,
      "grad_norm": 0.5480464100837708,
      "learning_rate": 7.693939148277128e-06,
      "loss": 0.9007,
      "step": 3506
    },
    {
      "epoch": 0.5879911977365608,
      "grad_norm": 0.6138718724250793,
      "learning_rate": 7.688655274400176e-06,
      "loss": 0.8431,
      "step": 3507
    },
    {
      "epoch": 0.5881588598973069,
      "grad_norm": 0.39476701617240906,
      "learning_rate": 7.683372082167404e-06,
      "loss": 0.8014,
      "step": 3508
    },
    {
      "epoch": 0.588326522058053,
      "grad_norm": 0.42859458923339844,
      "learning_rate": 7.678089573136897e-06,
      "loss": 0.876,
      "step": 3509
    },
    {
      "epoch": 0.5884941842187991,
      "grad_norm": 0.5820524096488953,
      "learning_rate": 7.672807748866523e-06,
      "loss": 0.7965,
      "step": 3510
    },
    {
      "epoch": 0.5886618463795452,
      "grad_norm": 0.41693684458732605,
      "learning_rate": 7.667526610913958e-06,
      "loss": 0.8259,
      "step": 3511
    },
    {
      "epoch": 0.5888295085402913,
      "grad_norm": 0.5068554282188416,
      "learning_rate": 7.66224616083668e-06,
      "loss": 0.8,
      "step": 3512
    },
    {
      "epoch": 0.5889971707010374,
      "grad_norm": 0.5913344621658325,
      "learning_rate": 7.656966400191956e-06,
      "loss": 0.7453,
      "step": 3513
    },
    {
      "epoch": 0.5891648328617836,
      "grad_norm": 0.4259031414985657,
      "learning_rate": 7.651687330536862e-06,
      "loss": 0.8127,
      "step": 3514
    },
    {
      "epoch": 0.5893324950225296,
      "grad_norm": 0.6040980815887451,
      "learning_rate": 7.646408953428243e-06,
      "loss": 0.9615,
      "step": 3515
    },
    {
      "epoch": 0.5895001571832758,
      "grad_norm": 0.5717450976371765,
      "learning_rate": 7.641131270422772e-06,
      "loss": 0.8308,
      "step": 3516
    },
    {
      "epoch": 0.5896678193440218,
      "grad_norm": 0.48589468002319336,
      "learning_rate": 7.635854283076897e-06,
      "loss": 0.7764,
      "step": 3517
    },
    {
      "epoch": 0.5898354815047679,
      "grad_norm": 0.5082957744598389,
      "learning_rate": 7.630577992946871e-06,
      "loss": 0.9003,
      "step": 3518
    },
    {
      "epoch": 0.590003143665514,
      "grad_norm": 0.47448721528053284,
      "learning_rate": 7.625302401588735e-06,
      "loss": 0.8517,
      "step": 3519
    },
    {
      "epoch": 0.5901708058262601,
      "grad_norm": 0.5031949877738953,
      "learning_rate": 7.620027510558324e-06,
      "loss": 0.7788,
      "step": 3520
    },
    {
      "epoch": 0.5903384679870062,
      "grad_norm": 0.601178765296936,
      "learning_rate": 7.6147533214112725e-06,
      "loss": 0.8609,
      "step": 3521
    },
    {
      "epoch": 0.5905061301477523,
      "grad_norm": 0.46869635581970215,
      "learning_rate": 7.609479835703002e-06,
      "loss": 0.7583,
      "step": 3522
    },
    {
      "epoch": 0.5906737923084984,
      "grad_norm": 0.6021032929420471,
      "learning_rate": 7.60420705498873e-06,
      "loss": 0.818,
      "step": 3523
    },
    {
      "epoch": 0.5908414544692445,
      "grad_norm": 0.4280416667461395,
      "learning_rate": 7.598934980823465e-06,
      "loss": 0.7912,
      "step": 3524
    },
    {
      "epoch": 0.5910091166299906,
      "grad_norm": 0.9477753639221191,
      "learning_rate": 7.593663614762004e-06,
      "loss": 0.8083,
      "step": 3525
    },
    {
      "epoch": 0.5911767787907367,
      "grad_norm": 0.6156827211380005,
      "learning_rate": 7.58839295835894e-06,
      "loss": 0.8669,
      "step": 3526
    },
    {
      "epoch": 0.5913444409514828,
      "grad_norm": 0.480912446975708,
      "learning_rate": 7.5831230131686585e-06,
      "loss": 0.7717,
      "step": 3527
    },
    {
      "epoch": 0.5915121031122289,
      "grad_norm": 0.7437429428100586,
      "learning_rate": 7.577853780745324e-06,
      "loss": 0.8701,
      "step": 3528
    },
    {
      "epoch": 0.591679765272975,
      "grad_norm": 0.5265527367591858,
      "learning_rate": 7.572585262642904e-06,
      "loss": 0.9543,
      "step": 3529
    },
    {
      "epoch": 0.5918474274337211,
      "grad_norm": 0.5225236415863037,
      "learning_rate": 7.567317460415149e-06,
      "loss": 0.8445,
      "step": 3530
    },
    {
      "epoch": 0.5920150895944671,
      "grad_norm": 0.5767743587493896,
      "learning_rate": 7.562050375615595e-06,
      "loss": 0.9037,
      "step": 3531
    },
    {
      "epoch": 0.5921827517552133,
      "grad_norm": 0.3901894688606262,
      "learning_rate": 7.556784009797581e-06,
      "loss": 0.7714,
      "step": 3532
    },
    {
      "epoch": 0.5923504139159593,
      "grad_norm": 0.39337241649627686,
      "learning_rate": 7.551518364514212e-06,
      "loss": 0.8793,
      "step": 3533
    },
    {
      "epoch": 0.5925180760767055,
      "grad_norm": 0.4687483608722687,
      "learning_rate": 7.546253441318396e-06,
      "loss": 0.9443,
      "step": 3534
    },
    {
      "epoch": 0.5926857382374515,
      "grad_norm": 0.43475142121315,
      "learning_rate": 7.5409892417628215e-06,
      "loss": 0.8109,
      "step": 3535
    },
    {
      "epoch": 0.5928534003981977,
      "grad_norm": 0.40518566966056824,
      "learning_rate": 7.535725767399973e-06,
      "loss": 0.8149,
      "step": 3536
    },
    {
      "epoch": 0.5930210625589437,
      "grad_norm": 0.5907869338989258,
      "learning_rate": 7.530463019782114e-06,
      "loss": 0.8444,
      "step": 3537
    },
    {
      "epoch": 0.5931887247196899,
      "grad_norm": 0.5921319127082825,
      "learning_rate": 7.525201000461284e-06,
      "loss": 0.8426,
      "step": 3538
    },
    {
      "epoch": 0.5933563868804359,
      "grad_norm": 0.5731843709945679,
      "learning_rate": 7.519939710989326e-06,
      "loss": 0.9189,
      "step": 3539
    },
    {
      "epoch": 0.5935240490411821,
      "grad_norm": 0.5038250684738159,
      "learning_rate": 7.514679152917857e-06,
      "loss": 0.7586,
      "step": 3540
    },
    {
      "epoch": 0.5936917112019281,
      "grad_norm": 0.5103881359100342,
      "learning_rate": 7.509419327798283e-06,
      "loss": 0.936,
      "step": 3541
    },
    {
      "epoch": 0.5938593733626742,
      "grad_norm": 0.4073008596897125,
      "learning_rate": 7.504160237181788e-06,
      "loss": 0.8011,
      "step": 3542
    },
    {
      "epoch": 0.5940270355234203,
      "grad_norm": 0.4870217740535736,
      "learning_rate": 7.498901882619345e-06,
      "loss": 0.8778,
      "step": 3543
    },
    {
      "epoch": 0.5941946976841664,
      "grad_norm": 0.4272831380367279,
      "learning_rate": 7.493644265661706e-06,
      "loss": 0.8946,
      "step": 3544
    },
    {
      "epoch": 0.5943623598449125,
      "grad_norm": 0.44826629757881165,
      "learning_rate": 7.48838738785941e-06,
      "loss": 0.8727,
      "step": 3545
    },
    {
      "epoch": 0.5945300220056586,
      "grad_norm": 0.3500494658946991,
      "learning_rate": 7.483131250762776e-06,
      "loss": 0.8728,
      "step": 3546
    },
    {
      "epoch": 0.5946976841664047,
      "grad_norm": 0.5210344195365906,
      "learning_rate": 7.477875855921899e-06,
      "loss": 0.8134,
      "step": 3547
    },
    {
      "epoch": 0.5948653463271508,
      "grad_norm": 0.5424099564552307,
      "learning_rate": 7.472621204886665e-06,
      "loss": 0.9176,
      "step": 3548
    },
    {
      "epoch": 0.5950330084878969,
      "grad_norm": 0.4739905893802643,
      "learning_rate": 7.467367299206733e-06,
      "loss": 0.8229,
      "step": 3549
    },
    {
      "epoch": 0.595200670648643,
      "grad_norm": 0.7030133605003357,
      "learning_rate": 7.4621141404315465e-06,
      "loss": 0.833,
      "step": 3550
    },
    {
      "epoch": 0.5953683328093891,
      "grad_norm": 0.36012545228004456,
      "learning_rate": 7.456861730110326e-06,
      "loss": 0.8498,
      "step": 3551
    },
    {
      "epoch": 0.5955359949701352,
      "grad_norm": 0.4297751486301422,
      "learning_rate": 7.45161006979207e-06,
      "loss": 0.933,
      "step": 3552
    },
    {
      "epoch": 0.5957036571308812,
      "grad_norm": 0.5189722776412964,
      "learning_rate": 7.4463591610255626e-06,
      "loss": 0.7511,
      "step": 3553
    },
    {
      "epoch": 0.5958713192916274,
      "grad_norm": 0.8779076337814331,
      "learning_rate": 7.441109005359357e-06,
      "loss": 0.9207,
      "step": 3554
    },
    {
      "epoch": 0.5960389814523734,
      "grad_norm": 0.35254567861557007,
      "learning_rate": 7.4358596043418e-06,
      "loss": 0.825,
      "step": 3555
    },
    {
      "epoch": 0.5962066436131196,
      "grad_norm": 0.4413334131240845,
      "learning_rate": 7.430610959520991e-06,
      "loss": 0.8796,
      "step": 3556
    },
    {
      "epoch": 0.5963743057738656,
      "grad_norm": 0.6897941827774048,
      "learning_rate": 7.425363072444824e-06,
      "loss": 0.8764,
      "step": 3557
    },
    {
      "epoch": 0.5965419679346118,
      "grad_norm": 0.47298622131347656,
      "learning_rate": 7.420115944660971e-06,
      "loss": 0.8621,
      "step": 3558
    },
    {
      "epoch": 0.5967096300953578,
      "grad_norm": 0.4050973057746887,
      "learning_rate": 7.414869577716877e-06,
      "loss": 0.9009,
      "step": 3559
    },
    {
      "epoch": 0.596877292256104,
      "grad_norm": 0.4958822429180145,
      "learning_rate": 7.409623973159749e-06,
      "loss": 0.8612,
      "step": 3560
    },
    {
      "epoch": 0.59704495441685,
      "grad_norm": 0.38055795431137085,
      "learning_rate": 7.404379132536588e-06,
      "loss": 0.7803,
      "step": 3561
    },
    {
      "epoch": 0.5972126165775962,
      "grad_norm": 0.8478285670280457,
      "learning_rate": 7.399135057394162e-06,
      "loss": 0.8753,
      "step": 3562
    },
    {
      "epoch": 0.5973802787383422,
      "grad_norm": 0.4878985285758972,
      "learning_rate": 7.393891749279015e-06,
      "loss": 0.8193,
      "step": 3563
    },
    {
      "epoch": 0.5975479408990884,
      "grad_norm": 0.5586049556732178,
      "learning_rate": 7.388649209737464e-06,
      "loss": 0.8218,
      "step": 3564
    },
    {
      "epoch": 0.5977156030598344,
      "grad_norm": 0.9109604358673096,
      "learning_rate": 7.383407440315595e-06,
      "loss": 0.8522,
      "step": 3565
    },
    {
      "epoch": 0.5978832652205806,
      "grad_norm": 0.4115822911262512,
      "learning_rate": 7.378166442559271e-06,
      "loss": 0.799,
      "step": 3566
    },
    {
      "epoch": 0.5980509273813266,
      "grad_norm": 0.5169524550437927,
      "learning_rate": 7.372926218014131e-06,
      "loss": 0.8786,
      "step": 3567
    },
    {
      "epoch": 0.5982185895420727,
      "grad_norm": 1.4872629642486572,
      "learning_rate": 7.367686768225582e-06,
      "loss": 0.8903,
      "step": 3568
    },
    {
      "epoch": 0.5983862517028188,
      "grad_norm": 0.5998322367668152,
      "learning_rate": 7.362448094738797e-06,
      "loss": 0.7769,
      "step": 3569
    },
    {
      "epoch": 0.5985539138635649,
      "grad_norm": 0.4917464852333069,
      "learning_rate": 7.35721019909873e-06,
      "loss": 0.7684,
      "step": 3570
    },
    {
      "epoch": 0.598721576024311,
      "grad_norm": 0.4330538213253021,
      "learning_rate": 7.351973082850099e-06,
      "loss": 0.8178,
      "step": 3571
    },
    {
      "epoch": 0.5988892381850571,
      "grad_norm": 0.4778382182121277,
      "learning_rate": 7.346736747537397e-06,
      "loss": 0.9828,
      "step": 3572
    },
    {
      "epoch": 0.5990569003458032,
      "grad_norm": 0.3780740201473236,
      "learning_rate": 7.341501194704885e-06,
      "loss": 0.8749,
      "step": 3573
    },
    {
      "epoch": 0.5992245625065493,
      "grad_norm": 0.6358554363250732,
      "learning_rate": 7.336266425896589e-06,
      "loss": 0.8051,
      "step": 3574
    },
    {
      "epoch": 0.5993922246672954,
      "grad_norm": 0.5639551281929016,
      "learning_rate": 7.331032442656307e-06,
      "loss": 0.8786,
      "step": 3575
    },
    {
      "epoch": 0.5995598868280415,
      "grad_norm": 0.8509765267372131,
      "learning_rate": 7.325799246527609e-06,
      "loss": 0.9342,
      "step": 3576
    },
    {
      "epoch": 0.5997275489887876,
      "grad_norm": 0.4847201406955719,
      "learning_rate": 7.320566839053824e-06,
      "loss": 0.8358,
      "step": 3577
    },
    {
      "epoch": 0.5998952111495337,
      "grad_norm": 1.1400126218795776,
      "learning_rate": 7.315335221778064e-06,
      "loss": 0.9474,
      "step": 3578
    },
    {
      "epoch": 0.6000628733102797,
      "grad_norm": 0.5179569125175476,
      "learning_rate": 7.310104396243185e-06,
      "loss": 0.7884,
      "step": 3579
    },
    {
      "epoch": 0.6002305354710259,
      "grad_norm": 0.457474946975708,
      "learning_rate": 7.304874363991828e-06,
      "loss": 0.858,
      "step": 3580
    },
    {
      "epoch": 0.6003981976317719,
      "grad_norm": 0.6846543550491333,
      "learning_rate": 7.299645126566395e-06,
      "loss": 0.8058,
      "step": 3581
    },
    {
      "epoch": 0.6005658597925181,
      "grad_norm": 0.4435752034187317,
      "learning_rate": 7.294416685509056e-06,
      "loss": 0.8896,
      "step": 3582
    },
    {
      "epoch": 0.6007335219532641,
      "grad_norm": 0.44891929626464844,
      "learning_rate": 7.289189042361736e-06,
      "loss": 0.8745,
      "step": 3583
    },
    {
      "epoch": 0.6009011841140103,
      "grad_norm": 0.4819808602333069,
      "learning_rate": 7.2839621986661325e-06,
      "loss": 0.8753,
      "step": 3584
    },
    {
      "epoch": 0.6010688462747563,
      "grad_norm": 0.5853043794631958,
      "learning_rate": 7.27873615596371e-06,
      "loss": 0.943,
      "step": 3585
    },
    {
      "epoch": 0.6012365084355025,
      "grad_norm": 0.4091236889362335,
      "learning_rate": 7.273510915795694e-06,
      "loss": 0.7616,
      "step": 3586
    },
    {
      "epoch": 0.6014041705962485,
      "grad_norm": 0.47817444801330566,
      "learning_rate": 7.2682864797030705e-06,
      "loss": 0.832,
      "step": 3587
    },
    {
      "epoch": 0.6015718327569947,
      "grad_norm": 0.44186002016067505,
      "learning_rate": 7.26306284922659e-06,
      "loss": 0.8801,
      "step": 3588
    },
    {
      "epoch": 0.6017394949177407,
      "grad_norm": 0.5388672351837158,
      "learning_rate": 7.257840025906766e-06,
      "loss": 0.9696,
      "step": 3589
    },
    {
      "epoch": 0.6019071570784869,
      "grad_norm": 0.4427429139614105,
      "learning_rate": 7.252618011283874e-06,
      "loss": 0.8297,
      "step": 3590
    },
    {
      "epoch": 0.6020748192392329,
      "grad_norm": 0.42489099502563477,
      "learning_rate": 7.247396806897953e-06,
      "loss": 0.8999,
      "step": 3591
    },
    {
      "epoch": 0.602242481399979,
      "grad_norm": 0.41747936606407166,
      "learning_rate": 7.242176414288795e-06,
      "loss": 0.864,
      "step": 3592
    },
    {
      "epoch": 0.6024101435607251,
      "grad_norm": 0.4210547208786011,
      "learning_rate": 7.236956834995963e-06,
      "loss": 0.8419,
      "step": 3593
    },
    {
      "epoch": 0.6025778057214712,
      "grad_norm": 0.37546658515930176,
      "learning_rate": 7.231738070558775e-06,
      "loss": 0.7766,
      "step": 3594
    },
    {
      "epoch": 0.6027454678822173,
      "grad_norm": 0.38674503564834595,
      "learning_rate": 7.226520122516308e-06,
      "loss": 0.9229,
      "step": 3595
    },
    {
      "epoch": 0.6029131300429634,
      "grad_norm": 0.41017404198646545,
      "learning_rate": 7.221302992407404e-06,
      "loss": 0.9184,
      "step": 3596
    },
    {
      "epoch": 0.6030807922037095,
      "grad_norm": 0.5656243562698364,
      "learning_rate": 7.216086681770652e-06,
      "loss": 0.9822,
      "step": 3597
    },
    {
      "epoch": 0.6032484543644556,
      "grad_norm": 0.6094668507575989,
      "learning_rate": 7.210871192144411e-06,
      "loss": 0.8996,
      "step": 3598
    },
    {
      "epoch": 0.6034161165252018,
      "grad_norm": 0.45814836025238037,
      "learning_rate": 7.205656525066793e-06,
      "loss": 0.8522,
      "step": 3599
    },
    {
      "epoch": 0.6035837786859478,
      "grad_norm": 0.4150603115558624,
      "learning_rate": 7.200442682075668e-06,
      "loss": 0.8364,
      "step": 3600
    },
    {
      "epoch": 0.603751440846694,
      "grad_norm": 0.500847578048706,
      "learning_rate": 7.195229664708663e-06,
      "loss": 0.9176,
      "step": 3601
    },
    {
      "epoch": 0.60391910300744,
      "grad_norm": 0.47346231341362,
      "learning_rate": 7.190017474503159e-06,
      "loss": 0.8205,
      "step": 3602
    },
    {
      "epoch": 0.6040867651681862,
      "grad_norm": 0.6502076983451843,
      "learning_rate": 7.184806112996294e-06,
      "loss": 0.9446,
      "step": 3603
    },
    {
      "epoch": 0.6042544273289322,
      "grad_norm": 0.43551987409591675,
      "learning_rate": 7.179595581724971e-06,
      "loss": 0.7325,
      "step": 3604
    },
    {
      "epoch": 0.6044220894896783,
      "grad_norm": 0.8503975868225098,
      "learning_rate": 7.174385882225838e-06,
      "loss": 0.7947,
      "step": 3605
    },
    {
      "epoch": 0.6045897516504244,
      "grad_norm": 0.4926724135875702,
      "learning_rate": 7.169177016035293e-06,
      "loss": 0.8104,
      "step": 3606
    },
    {
      "epoch": 0.6047574138111705,
      "grad_norm": 0.6093217730522156,
      "learning_rate": 7.1639689846895e-06,
      "loss": 0.8905,
      "step": 3607
    },
    {
      "epoch": 0.6049250759719166,
      "grad_norm": 0.5507587194442749,
      "learning_rate": 7.158761789724371e-06,
      "loss": 0.984,
      "step": 3608
    },
    {
      "epoch": 0.6050927381326627,
      "grad_norm": 0.5029594898223877,
      "learning_rate": 7.153555432675574e-06,
      "loss": 0.8459,
      "step": 3609
    },
    {
      "epoch": 0.6052604002934088,
      "grad_norm": 0.4744623899459839,
      "learning_rate": 7.148349915078528e-06,
      "loss": 0.9074,
      "step": 3610
    },
    {
      "epoch": 0.6054280624541549,
      "grad_norm": 0.4237988591194153,
      "learning_rate": 7.143145238468403e-06,
      "loss": 0.8329,
      "step": 3611
    },
    {
      "epoch": 0.605595724614901,
      "grad_norm": 0.3881392776966095,
      "learning_rate": 7.137941404380122e-06,
      "loss": 0.771,
      "step": 3612
    },
    {
      "epoch": 0.6057633867756471,
      "grad_norm": 0.4183195233345032,
      "learning_rate": 7.132738414348361e-06,
      "loss": 0.8318,
      "step": 3613
    },
    {
      "epoch": 0.6059310489363932,
      "grad_norm": 0.5632359981536865,
      "learning_rate": 7.127536269907549e-06,
      "loss": 0.8639,
      "step": 3614
    },
    {
      "epoch": 0.6060987110971393,
      "grad_norm": 0.40016984939575195,
      "learning_rate": 7.122334972591858e-06,
      "loss": 0.9338,
      "step": 3615
    },
    {
      "epoch": 0.6062663732578853,
      "grad_norm": 0.5154279470443726,
      "learning_rate": 7.117134523935217e-06,
      "loss": 0.8404,
      "step": 3616
    },
    {
      "epoch": 0.6064340354186315,
      "grad_norm": 0.5123127102851868,
      "learning_rate": 7.111934925471302e-06,
      "loss": 0.8732,
      "step": 3617
    },
    {
      "epoch": 0.6066016975793775,
      "grad_norm": 0.45707687735557556,
      "learning_rate": 7.10673617873354e-06,
      "loss": 0.7817,
      "step": 3618
    },
    {
      "epoch": 0.6067693597401237,
      "grad_norm": 0.4302932024002075,
      "learning_rate": 7.101538285255108e-06,
      "loss": 0.9229,
      "step": 3619
    },
    {
      "epoch": 0.6069370219008697,
      "grad_norm": 0.4706227481365204,
      "learning_rate": 7.096341246568925e-06,
      "loss": 0.8589,
      "step": 3620
    },
    {
      "epoch": 0.6071046840616159,
      "grad_norm": 0.9093366265296936,
      "learning_rate": 7.091145064207665e-06,
      "loss": 0.8966,
      "step": 3621
    },
    {
      "epoch": 0.6072723462223619,
      "grad_norm": 0.4529713988304138,
      "learning_rate": 7.0859497397037416e-06,
      "loss": 0.7916,
      "step": 3622
    },
    {
      "epoch": 0.6074400083831081,
      "grad_norm": 0.5407692790031433,
      "learning_rate": 7.080755274589332e-06,
      "loss": 0.8121,
      "step": 3623
    },
    {
      "epoch": 0.6076076705438541,
      "grad_norm": 0.3993769884109497,
      "learning_rate": 7.075561670396334e-06,
      "loss": 0.8102,
      "step": 3624
    },
    {
      "epoch": 0.6077753327046003,
      "grad_norm": 0.9855776429176331,
      "learning_rate": 7.070368928656413e-06,
      "loss": 0.9055,
      "step": 3625
    },
    {
      "epoch": 0.6079429948653463,
      "grad_norm": 0.4427030384540558,
      "learning_rate": 7.065177050900971e-06,
      "loss": 0.8445,
      "step": 3626
    },
    {
      "epoch": 0.6081106570260925,
      "grad_norm": 0.42641207575798035,
      "learning_rate": 7.05998603866116e-06,
      "loss": 0.8255,
      "step": 3627
    },
    {
      "epoch": 0.6082783191868385,
      "grad_norm": 0.42815962433815,
      "learning_rate": 7.054795893467875e-06,
      "loss": 0.8354,
      "step": 3628
    },
    {
      "epoch": 0.6084459813475847,
      "grad_norm": 0.4792901575565338,
      "learning_rate": 7.0496066168517455e-06,
      "loss": 0.8226,
      "step": 3629
    },
    {
      "epoch": 0.6086136435083307,
      "grad_norm": 0.5730831027030945,
      "learning_rate": 7.044418210343161e-06,
      "loss": 0.8705,
      "step": 3630
    },
    {
      "epoch": 0.6087813056690768,
      "grad_norm": 0.6771920919418335,
      "learning_rate": 7.039230675472244e-06,
      "loss": 0.8276,
      "step": 3631
    },
    {
      "epoch": 0.6089489678298229,
      "grad_norm": 0.5605460405349731,
      "learning_rate": 7.034044013768868e-06,
      "loss": 0.7989,
      "step": 3632
    },
    {
      "epoch": 0.609116629990569,
      "grad_norm": 0.7433153390884399,
      "learning_rate": 7.028858226762632e-06,
      "loss": 0.9028,
      "step": 3633
    },
    {
      "epoch": 0.6092842921513151,
      "grad_norm": 0.5418516993522644,
      "learning_rate": 7.0236733159829e-06,
      "loss": 0.7459,
      "step": 3634
    },
    {
      "epoch": 0.6094519543120612,
      "grad_norm": 0.5711153745651245,
      "learning_rate": 7.0184892829587605e-06,
      "loss": 0.7997,
      "step": 3635
    },
    {
      "epoch": 0.6096196164728073,
      "grad_norm": 0.5344470143318176,
      "learning_rate": 7.013306129219052e-06,
      "loss": 0.823,
      "step": 3636
    },
    {
      "epoch": 0.6097872786335534,
      "grad_norm": 0.9105053544044495,
      "learning_rate": 7.008123856292352e-06,
      "loss": 0.8645,
      "step": 3637
    },
    {
      "epoch": 0.6099549407942995,
      "grad_norm": 0.33319270610809326,
      "learning_rate": 7.002942465706972e-06,
      "loss": 0.7663,
      "step": 3638
    },
    {
      "epoch": 0.6101226029550456,
      "grad_norm": 0.6968861222267151,
      "learning_rate": 6.9977619589909705e-06,
      "loss": 0.7576,
      "step": 3639
    },
    {
      "epoch": 0.6102902651157917,
      "grad_norm": 0.6255199313163757,
      "learning_rate": 6.992582337672145e-06,
      "loss": 0.886,
      "step": 3640
    },
    {
      "epoch": 0.6104579272765378,
      "grad_norm": 0.6661763787269592,
      "learning_rate": 6.987403603278029e-06,
      "loss": 0.8666,
      "step": 3641
    },
    {
      "epoch": 0.6106255894372838,
      "grad_norm": 0.4097493886947632,
      "learning_rate": 6.982225757335892e-06,
      "loss": 0.7663,
      "step": 3642
    },
    {
      "epoch": 0.61079325159803,
      "grad_norm": 0.48773884773254395,
      "learning_rate": 6.97704880137275e-06,
      "loss": 0.8584,
      "step": 3643
    },
    {
      "epoch": 0.610960913758776,
      "grad_norm": 0.520758867263794,
      "learning_rate": 6.971872736915348e-06,
      "loss": 0.8782,
      "step": 3644
    },
    {
      "epoch": 0.6111285759195222,
      "grad_norm": 0.4427630603313446,
      "learning_rate": 6.9666975654901704e-06,
      "loss": 0.8038,
      "step": 3645
    },
    {
      "epoch": 0.6112962380802682,
      "grad_norm": 0.8011310696601868,
      "learning_rate": 6.9615232886234475e-06,
      "loss": 0.8792,
      "step": 3646
    },
    {
      "epoch": 0.6114639002410144,
      "grad_norm": 0.5017745494842529,
      "learning_rate": 6.956349907841128e-06,
      "loss": 0.9043,
      "step": 3647
    },
    {
      "epoch": 0.6116315624017604,
      "grad_norm": 0.6061432957649231,
      "learning_rate": 6.951177424668909e-06,
      "loss": 0.8709,
      "step": 3648
    },
    {
      "epoch": 0.6117992245625066,
      "grad_norm": 0.4249998927116394,
      "learning_rate": 6.9460058406322195e-06,
      "loss": 0.9202,
      "step": 3649
    },
    {
      "epoch": 0.6119668867232526,
      "grad_norm": 0.47136473655700684,
      "learning_rate": 6.940835157256226e-06,
      "loss": 0.8131,
      "step": 3650
    },
    {
      "epoch": 0.6121345488839988,
      "grad_norm": 0.39274460077285767,
      "learning_rate": 6.93566537606583e-06,
      "loss": 0.9147,
      "step": 3651
    },
    {
      "epoch": 0.6123022110447448,
      "grad_norm": 0.613770067691803,
      "learning_rate": 6.930496498585654e-06,
      "loss": 0.7946,
      "step": 3652
    },
    {
      "epoch": 0.612469873205491,
      "grad_norm": 0.4516802132129669,
      "learning_rate": 6.925328526340072e-06,
      "loss": 0.7832,
      "step": 3653
    },
    {
      "epoch": 0.612637535366237,
      "grad_norm": 0.44192588329315186,
      "learning_rate": 6.920161460853179e-06,
      "loss": 0.9041,
      "step": 3654
    },
    {
      "epoch": 0.6128051975269831,
      "grad_norm": 0.7549390196800232,
      "learning_rate": 6.914995303648811e-06,
      "loss": 0.8354,
      "step": 3655
    },
    {
      "epoch": 0.6129728596877292,
      "grad_norm": 0.49047520756721497,
      "learning_rate": 6.909830056250527e-06,
      "loss": 0.9369,
      "step": 3656
    },
    {
      "epoch": 0.6131405218484753,
      "grad_norm": 0.6662545800209045,
      "learning_rate": 6.9046657201816245e-06,
      "loss": 0.7678,
      "step": 3657
    },
    {
      "epoch": 0.6133081840092214,
      "grad_norm": 0.5436139702796936,
      "learning_rate": 6.89950229696513e-06,
      "loss": 0.8306,
      "step": 3658
    },
    {
      "epoch": 0.6134758461699675,
      "grad_norm": 0.40033626556396484,
      "learning_rate": 6.894339788123801e-06,
      "loss": 0.8623,
      "step": 3659
    },
    {
      "epoch": 0.6136435083307136,
      "grad_norm": 0.43172773718833923,
      "learning_rate": 6.889178195180128e-06,
      "loss": 0.799,
      "step": 3660
    },
    {
      "epoch": 0.6138111704914597,
      "grad_norm": 0.42334967851638794,
      "learning_rate": 6.884017519656323e-06,
      "loss": 0.8647,
      "step": 3661
    },
    {
      "epoch": 0.6139788326522058,
      "grad_norm": 0.37546423077583313,
      "learning_rate": 6.878857763074336e-06,
      "loss": 0.7751,
      "step": 3662
    },
    {
      "epoch": 0.6141464948129519,
      "grad_norm": 0.3862164318561554,
      "learning_rate": 6.873698926955844e-06,
      "loss": 0.8285,
      "step": 3663
    },
    {
      "epoch": 0.614314156973698,
      "grad_norm": 0.5159278512001038,
      "learning_rate": 6.868541012822253e-06,
      "loss": 0.8971,
      "step": 3664
    },
    {
      "epoch": 0.6144818191344441,
      "grad_norm": 0.7204200625419617,
      "learning_rate": 6.86338402219469e-06,
      "loss": 0.8071,
      "step": 3665
    },
    {
      "epoch": 0.6146494812951901,
      "grad_norm": 0.4890683889389038,
      "learning_rate": 6.858227956594022e-06,
      "loss": 0.7644,
      "step": 3666
    },
    {
      "epoch": 0.6148171434559363,
      "grad_norm": 0.544012725353241,
      "learning_rate": 6.853072817540831e-06,
      "loss": 0.8454,
      "step": 3667
    },
    {
      "epoch": 0.6149848056166823,
      "grad_norm": 0.34836164116859436,
      "learning_rate": 6.847918606555433e-06,
      "loss": 0.8329,
      "step": 3668
    },
    {
      "epoch": 0.6151524677774285,
      "grad_norm": 0.5558313727378845,
      "learning_rate": 6.842765325157874e-06,
      "loss": 0.8326,
      "step": 3669
    },
    {
      "epoch": 0.6153201299381745,
      "grad_norm": 0.45246636867523193,
      "learning_rate": 6.837612974867913e-06,
      "loss": 0.849,
      "step": 3670
    },
    {
      "epoch": 0.6154877920989207,
      "grad_norm": 0.6189624667167664,
      "learning_rate": 6.832461557205043e-06,
      "loss": 0.8273,
      "step": 3671
    },
    {
      "epoch": 0.6156554542596667,
      "grad_norm": 0.4349825084209442,
      "learning_rate": 6.827311073688485e-06,
      "loss": 0.853,
      "step": 3672
    },
    {
      "epoch": 0.6158231164204129,
      "grad_norm": 0.5392482876777649,
      "learning_rate": 6.822161525837183e-06,
      "loss": 0.8998,
      "step": 3673
    },
    {
      "epoch": 0.6159907785811589,
      "grad_norm": 0.3823678195476532,
      "learning_rate": 6.8170129151697925e-06,
      "loss": 0.8844,
      "step": 3674
    },
    {
      "epoch": 0.6161584407419051,
      "grad_norm": 0.6585565209388733,
      "learning_rate": 6.811865243204706e-06,
      "loss": 0.7969,
      "step": 3675
    },
    {
      "epoch": 0.6163261029026511,
      "grad_norm": 0.4332069456577301,
      "learning_rate": 6.80671851146004e-06,
      "loss": 0.8772,
      "step": 3676
    },
    {
      "epoch": 0.6164937650633973,
      "grad_norm": 0.5370450615882874,
      "learning_rate": 6.801572721453625e-06,
      "loss": 0.7749,
      "step": 3677
    },
    {
      "epoch": 0.6166614272241433,
      "grad_norm": 0.3939927816390991,
      "learning_rate": 6.796427874703025e-06,
      "loss": 0.9287,
      "step": 3678
    },
    {
      "epoch": 0.6168290893848895,
      "grad_norm": 0.5207778215408325,
      "learning_rate": 6.791283972725509e-06,
      "loss": 0.9161,
      "step": 3679
    },
    {
      "epoch": 0.6169967515456355,
      "grad_norm": 0.8378856182098389,
      "learning_rate": 6.786141017038084e-06,
      "loss": 0.8162,
      "step": 3680
    },
    {
      "epoch": 0.6171644137063816,
      "grad_norm": 0.5842326879501343,
      "learning_rate": 6.780999009157469e-06,
      "loss": 0.7822,
      "step": 3681
    },
    {
      "epoch": 0.6173320758671277,
      "grad_norm": 0.4481719136238098,
      "learning_rate": 6.775857950600107e-06,
      "loss": 0.9661,
      "step": 3682
    },
    {
      "epoch": 0.6174997380278738,
      "grad_norm": 0.37426868081092834,
      "learning_rate": 6.77071784288216e-06,
      "loss": 0.806,
      "step": 3683
    },
    {
      "epoch": 0.61766740018862,
      "grad_norm": 0.48538267612457275,
      "learning_rate": 6.765578687519508e-06,
      "loss": 1.0161,
      "step": 3684
    },
    {
      "epoch": 0.617835062349366,
      "grad_norm": 0.6142372488975525,
      "learning_rate": 6.760440486027751e-06,
      "loss": 0.878,
      "step": 3685
    },
    {
      "epoch": 0.6180027245101122,
      "grad_norm": 0.38488391041755676,
      "learning_rate": 6.7553032399222104e-06,
      "loss": 0.8512,
      "step": 3686
    },
    {
      "epoch": 0.6181703866708582,
      "grad_norm": 0.43986889719963074,
      "learning_rate": 6.750166950717925e-06,
      "loss": 0.8792,
      "step": 3687
    },
    {
      "epoch": 0.6183380488316044,
      "grad_norm": 0.3855423331260681,
      "learning_rate": 6.745031619929646e-06,
      "loss": 0.8564,
      "step": 3688
    },
    {
      "epoch": 0.6185057109923504,
      "grad_norm": 0.446071058511734,
      "learning_rate": 6.739897249071847e-06,
      "loss": 0.8781,
      "step": 3689
    },
    {
      "epoch": 0.6186733731530966,
      "grad_norm": 0.5075111985206604,
      "learning_rate": 6.73476383965872e-06,
      "loss": 0.9628,
      "step": 3690
    },
    {
      "epoch": 0.6188410353138426,
      "grad_norm": 0.3769055902957916,
      "learning_rate": 6.729631393204167e-06,
      "loss": 0.8846,
      "step": 3691
    },
    {
      "epoch": 0.6190086974745888,
      "grad_norm": 0.9417046904563904,
      "learning_rate": 6.7244999112218175e-06,
      "loss": 0.9728,
      "step": 3692
    },
    {
      "epoch": 0.6191763596353348,
      "grad_norm": 0.33916711807250977,
      "learning_rate": 6.719369395225002e-06,
      "loss": 0.7993,
      "step": 3693
    },
    {
      "epoch": 0.619344021796081,
      "grad_norm": 0.5961330533027649,
      "learning_rate": 6.714239846726771e-06,
      "loss": 0.8623,
      "step": 3694
    },
    {
      "epoch": 0.619511683956827,
      "grad_norm": 0.6619173884391785,
      "learning_rate": 6.7091112672399e-06,
      "loss": 0.8258,
      "step": 3695
    },
    {
      "epoch": 0.6196793461175731,
      "grad_norm": 0.45241880416870117,
      "learning_rate": 6.70398365827687e-06,
      "loss": 0.8904,
      "step": 3696
    },
    {
      "epoch": 0.6198470082783192,
      "grad_norm": 0.4222065806388855,
      "learning_rate": 6.69885702134987e-06,
      "loss": 0.9054,
      "step": 3697
    },
    {
      "epoch": 0.6200146704390653,
      "grad_norm": 0.4871988594532013,
      "learning_rate": 6.6937313579708074e-06,
      "loss": 0.9105,
      "step": 3698
    },
    {
      "epoch": 0.6201823325998114,
      "grad_norm": 0.4782107174396515,
      "learning_rate": 6.688606669651312e-06,
      "loss": 0.8084,
      "step": 3699
    },
    {
      "epoch": 0.6203499947605575,
      "grad_norm": 0.3759131133556366,
      "learning_rate": 6.683482957902714e-06,
      "loss": 0.7746,
      "step": 3700
    },
    {
      "epoch": 0.6205176569213036,
      "grad_norm": 0.46672704815864563,
      "learning_rate": 6.678360224236062e-06,
      "loss": 0.8601,
      "step": 3701
    },
    {
      "epoch": 0.6206853190820497,
      "grad_norm": 0.46659931540489197,
      "learning_rate": 6.673238470162109e-06,
      "loss": 0.8369,
      "step": 3702
    },
    {
      "epoch": 0.6208529812427958,
      "grad_norm": 0.5919094681739807,
      "learning_rate": 6.668117697191325e-06,
      "loss": 0.8311,
      "step": 3703
    },
    {
      "epoch": 0.6210206434035419,
      "grad_norm": 0.430733323097229,
      "learning_rate": 6.66299790683389e-06,
      "loss": 0.8729,
      "step": 3704
    },
    {
      "epoch": 0.621188305564288,
      "grad_norm": 0.38417741656303406,
      "learning_rate": 6.657879100599697e-06,
      "loss": 0.832,
      "step": 3705
    },
    {
      "epoch": 0.6213559677250341,
      "grad_norm": 0.611831784248352,
      "learning_rate": 6.652761279998338e-06,
      "loss": 0.853,
      "step": 3706
    },
    {
      "epoch": 0.6215236298857801,
      "grad_norm": 0.48803794384002686,
      "learning_rate": 6.647644446539129e-06,
      "loss": 0.805,
      "step": 3707
    },
    {
      "epoch": 0.6216912920465263,
      "grad_norm": 0.9284753799438477,
      "learning_rate": 6.642528601731082e-06,
      "loss": 0.923,
      "step": 3708
    },
    {
      "epoch": 0.6218589542072723,
      "grad_norm": 0.4466826915740967,
      "learning_rate": 6.6374137470829256e-06,
      "loss": 0.7906,
      "step": 3709
    },
    {
      "epoch": 0.6220266163680185,
      "grad_norm": 0.6097007393836975,
      "learning_rate": 6.632299884103096e-06,
      "loss": 0.8859,
      "step": 3710
    },
    {
      "epoch": 0.6221942785287645,
      "grad_norm": 0.5663206577301025,
      "learning_rate": 6.62718701429973e-06,
      "loss": 0.8152,
      "step": 3711
    },
    {
      "epoch": 0.6223619406895107,
      "grad_norm": 0.42629796266555786,
      "learning_rate": 6.622075139180678e-06,
      "loss": 0.8714,
      "step": 3712
    },
    {
      "epoch": 0.6225296028502567,
      "grad_norm": 0.548659086227417,
      "learning_rate": 6.616964260253496e-06,
      "loss": 0.9259,
      "step": 3713
    },
    {
      "epoch": 0.6226972650110029,
      "grad_norm": 0.8277711272239685,
      "learning_rate": 6.6118543790254465e-06,
      "loss": 1.0303,
      "step": 3714
    },
    {
      "epoch": 0.6228649271717489,
      "grad_norm": 0.3924505412578583,
      "learning_rate": 6.606745497003495e-06,
      "loss": 0.8514,
      "step": 3715
    },
    {
      "epoch": 0.6230325893324951,
      "grad_norm": 0.4219715893268585,
      "learning_rate": 6.601637615694313e-06,
      "loss": 0.7878,
      "step": 3716
    },
    {
      "epoch": 0.6232002514932411,
      "grad_norm": 0.4186965823173523,
      "learning_rate": 6.596530736604278e-06,
      "loss": 0.8878,
      "step": 3717
    },
    {
      "epoch": 0.6233679136539872,
      "grad_norm": 1.0616917610168457,
      "learning_rate": 6.591424861239475e-06,
      "loss": 0.7697,
      "step": 3718
    },
    {
      "epoch": 0.6235355758147333,
      "grad_norm": 0.6864936947822571,
      "learning_rate": 6.586319991105692e-06,
      "loss": 0.914,
      "step": 3719
    },
    {
      "epoch": 0.6237032379754794,
      "grad_norm": 0.5054576992988586,
      "learning_rate": 6.58121612770841e-06,
      "loss": 0.8723,
      "step": 3720
    },
    {
      "epoch": 0.6238709001362255,
      "grad_norm": 0.6317015886306763,
      "learning_rate": 6.5761132725528265e-06,
      "loss": 0.8447,
      "step": 3721
    },
    {
      "epoch": 0.6240385622969716,
      "grad_norm": 0.5034641623497009,
      "learning_rate": 6.571011427143837e-06,
      "loss": 0.7845,
      "step": 3722
    },
    {
      "epoch": 0.6242062244577177,
      "grad_norm": 0.5249319672584534,
      "learning_rate": 6.565910592986038e-06,
      "loss": 0.8479,
      "step": 3723
    },
    {
      "epoch": 0.6243738866184638,
      "grad_norm": 0.47136643528938293,
      "learning_rate": 6.5608107715837295e-06,
      "loss": 0.8429,
      "step": 3724
    },
    {
      "epoch": 0.6245415487792099,
      "grad_norm": 0.7187997698783875,
      "learning_rate": 6.5557119644409115e-06,
      "loss": 0.8155,
      "step": 3725
    },
    {
      "epoch": 0.624709210939956,
      "grad_norm": 0.32335126399993896,
      "learning_rate": 6.550614173061285e-06,
      "loss": 0.7963,
      "step": 3726
    },
    {
      "epoch": 0.6248768731007021,
      "grad_norm": 0.4105938673019409,
      "learning_rate": 6.54551739894825e-06,
      "loss": 0.8302,
      "step": 3727
    },
    {
      "epoch": 0.6250445352614482,
      "grad_norm": 0.5323469042778015,
      "learning_rate": 6.540421643604914e-06,
      "loss": 0.8728,
      "step": 3728
    },
    {
      "epoch": 0.6252121974221942,
      "grad_norm": 0.9208778142929077,
      "learning_rate": 6.535326908534072e-06,
      "loss": 0.838,
      "step": 3729
    },
    {
      "epoch": 0.6253798595829404,
      "grad_norm": 0.5084686875343323,
      "learning_rate": 6.530233195238227e-06,
      "loss": 0.8912,
      "step": 3730
    },
    {
      "epoch": 0.6255475217436864,
      "grad_norm": 0.5037923455238342,
      "learning_rate": 6.525140505219578e-06,
      "loss": 0.8224,
      "step": 3731
    },
    {
      "epoch": 0.6257151839044326,
      "grad_norm": 0.6251398921012878,
      "learning_rate": 6.520048839980023e-06,
      "loss": 0.8781,
      "step": 3732
    },
    {
      "epoch": 0.6258828460651786,
      "grad_norm": 0.4075562059879303,
      "learning_rate": 6.5149582010211595e-06,
      "loss": 0.9049,
      "step": 3733
    },
    {
      "epoch": 0.6260505082259248,
      "grad_norm": 0.6155499219894409,
      "learning_rate": 6.509868589844274e-06,
      "loss": 0.8914,
      "step": 3734
    },
    {
      "epoch": 0.6262181703866708,
      "grad_norm": 0.465683251619339,
      "learning_rate": 6.50478000795036e-06,
      "loss": 0.7889,
      "step": 3735
    },
    {
      "epoch": 0.626385832547417,
      "grad_norm": 0.4047546684741974,
      "learning_rate": 6.499692456840101e-06,
      "loss": 0.8472,
      "step": 3736
    },
    {
      "epoch": 0.626553494708163,
      "grad_norm": 0.5044263005256653,
      "learning_rate": 6.494605938013883e-06,
      "loss": 0.7842,
      "step": 3737
    },
    {
      "epoch": 0.6267211568689092,
      "grad_norm": 0.39084577560424805,
      "learning_rate": 6.489520452971778e-06,
      "loss": 0.8353,
      "step": 3738
    },
    {
      "epoch": 0.6268888190296552,
      "grad_norm": 0.495937705039978,
      "learning_rate": 6.484436003213561e-06,
      "loss": 0.7824,
      "step": 3739
    },
    {
      "epoch": 0.6270564811904014,
      "grad_norm": 0.37145090103149414,
      "learning_rate": 6.479352590238697e-06,
      "loss": 0.8204,
      "step": 3740
    },
    {
      "epoch": 0.6272241433511474,
      "grad_norm": 0.5771486163139343,
      "learning_rate": 6.474270215546352e-06,
      "loss": 0.9038,
      "step": 3741
    },
    {
      "epoch": 0.6273918055118936,
      "grad_norm": 0.49100837111473083,
      "learning_rate": 6.469188880635384e-06,
      "loss": 0.8524,
      "step": 3742
    },
    {
      "epoch": 0.6275594676726396,
      "grad_norm": 0.368540495634079,
      "learning_rate": 6.46410858700433e-06,
      "loss": 0.8622,
      "step": 3743
    },
    {
      "epoch": 0.6277271298333857,
      "grad_norm": 0.5739620923995972,
      "learning_rate": 6.4590293361514385e-06,
      "loss": 0.7969,
      "step": 3744
    },
    {
      "epoch": 0.6278947919941318,
      "grad_norm": 0.39057913422584534,
      "learning_rate": 6.453951129574644e-06,
      "loss": 0.8991,
      "step": 3745
    },
    {
      "epoch": 0.6280624541548779,
      "grad_norm": 0.5076261162757874,
      "learning_rate": 6.448873968771575e-06,
      "loss": 0.8836,
      "step": 3746
    },
    {
      "epoch": 0.628230116315624,
      "grad_norm": 0.5153020620346069,
      "learning_rate": 6.44379785523954e-06,
      "loss": 0.8959,
      "step": 3747
    },
    {
      "epoch": 0.6283977784763701,
      "grad_norm": 0.4407910704612732,
      "learning_rate": 6.438722790475555e-06,
      "loss": 0.8904,
      "step": 3748
    },
    {
      "epoch": 0.6285654406371162,
      "grad_norm": 0.40577349066734314,
      "learning_rate": 6.433648775976317e-06,
      "loss": 0.9416,
      "step": 3749
    },
    {
      "epoch": 0.6287331027978623,
      "grad_norm": 0.7215719819068909,
      "learning_rate": 6.428575813238218e-06,
      "loss": 0.8255,
      "step": 3750
    },
    {
      "epoch": 0.6289007649586084,
      "grad_norm": 0.5980567932128906,
      "learning_rate": 6.423503903757338e-06,
      "loss": 0.7864,
      "step": 3751
    },
    {
      "epoch": 0.6290684271193545,
      "grad_norm": 0.9818123579025269,
      "learning_rate": 6.418433049029441e-06,
      "loss": 0.9186,
      "step": 3752
    },
    {
      "epoch": 0.6292360892801006,
      "grad_norm": 0.935444712638855,
      "learning_rate": 6.4133632505499876e-06,
      "loss": 0.7366,
      "step": 3753
    },
    {
      "epoch": 0.6294037514408467,
      "grad_norm": 0.4127931594848633,
      "learning_rate": 6.4082945098141244e-06,
      "loss": 0.8591,
      "step": 3754
    },
    {
      "epoch": 0.6295714136015927,
      "grad_norm": 0.38492462038993835,
      "learning_rate": 6.403226828316686e-06,
      "loss": 0.8389,
      "step": 3755
    },
    {
      "epoch": 0.6297390757623389,
      "grad_norm": 0.4961259365081787,
      "learning_rate": 6.398160207552196e-06,
      "loss": 0.8869,
      "step": 3756
    },
    {
      "epoch": 0.6299067379230849,
      "grad_norm": 0.4814993739128113,
      "learning_rate": 6.39309464901486e-06,
      "loss": 0.8744,
      "step": 3757
    },
    {
      "epoch": 0.6300744000838311,
      "grad_norm": 0.6155486702919006,
      "learning_rate": 6.3880301541985746e-06,
      "loss": 0.915,
      "step": 3758
    },
    {
      "epoch": 0.6302420622445771,
      "grad_norm": 0.4084378778934479,
      "learning_rate": 6.382966724596922e-06,
      "loss": 0.8946,
      "step": 3759
    },
    {
      "epoch": 0.6304097244053233,
      "grad_norm": 0.5810097455978394,
      "learning_rate": 6.3779043617031775e-06,
      "loss": 0.7778,
      "step": 3760
    },
    {
      "epoch": 0.6305773865660693,
      "grad_norm": 0.6305485963821411,
      "learning_rate": 6.372843067010284e-06,
      "loss": 0.9236,
      "step": 3761
    },
    {
      "epoch": 0.6307450487268155,
      "grad_norm": 0.7881393432617188,
      "learning_rate": 6.367782842010885e-06,
      "loss": 0.8842,
      "step": 3762
    },
    {
      "epoch": 0.6309127108875615,
      "grad_norm": 0.6298486590385437,
      "learning_rate": 6.3627236881973e-06,
      "loss": 0.9203,
      "step": 3763
    },
    {
      "epoch": 0.6310803730483077,
      "grad_norm": 0.5325604677200317,
      "learning_rate": 6.357665607061542e-06,
      "loss": 0.7429,
      "step": 3764
    },
    {
      "epoch": 0.6312480352090537,
      "grad_norm": 0.4600087106227875,
      "learning_rate": 6.352608600095302e-06,
      "loss": 0.8481,
      "step": 3765
    },
    {
      "epoch": 0.6314156973697999,
      "grad_norm": 0.5543720722198486,
      "learning_rate": 6.3475526687899445e-06,
      "loss": 0.9321,
      "step": 3766
    },
    {
      "epoch": 0.6315833595305459,
      "grad_norm": 0.4753662049770355,
      "learning_rate": 6.342497814636536e-06,
      "loss": 0.8124,
      "step": 3767
    },
    {
      "epoch": 0.631751021691292,
      "grad_norm": 0.42197152972221375,
      "learning_rate": 6.3374440391258105e-06,
      "loss": 0.8414,
      "step": 3768
    },
    {
      "epoch": 0.6319186838520381,
      "grad_norm": 0.45187443494796753,
      "learning_rate": 6.332391343748196e-06,
      "loss": 0.8752,
      "step": 3769
    },
    {
      "epoch": 0.6320863460127842,
      "grad_norm": 0.41934412717819214,
      "learning_rate": 6.327339729993783e-06,
      "loss": 0.9122,
      "step": 3770
    },
    {
      "epoch": 0.6322540081735304,
      "grad_norm": 0.7022708654403687,
      "learning_rate": 6.322289199352364e-06,
      "loss": 0.918,
      "step": 3771
    },
    {
      "epoch": 0.6324216703342764,
      "grad_norm": 0.5350022315979004,
      "learning_rate": 6.3172397533134e-06,
      "loss": 0.846,
      "step": 3772
    },
    {
      "epoch": 0.6325893324950226,
      "grad_norm": 0.4491613805294037,
      "learning_rate": 6.312191393366036e-06,
      "loss": 0.8767,
      "step": 3773
    },
    {
      "epoch": 0.6327569946557686,
      "grad_norm": 0.3387396037578583,
      "learning_rate": 6.307144120999097e-06,
      "loss": 0.7956,
      "step": 3774
    },
    {
      "epoch": 0.6329246568165148,
      "grad_norm": 0.6551179885864258,
      "learning_rate": 6.302097937701083e-06,
      "loss": 0.867,
      "step": 3775
    },
    {
      "epoch": 0.6330923189772608,
      "grad_norm": 0.467792809009552,
      "learning_rate": 6.297052844960178e-06,
      "loss": 0.8182,
      "step": 3776
    },
    {
      "epoch": 0.633259981138007,
      "grad_norm": 0.5014797449111938,
      "learning_rate": 6.292008844264242e-06,
      "loss": 0.7121,
      "step": 3777
    },
    {
      "epoch": 0.633427643298753,
      "grad_norm": 0.39362525939941406,
      "learning_rate": 6.286965937100813e-06,
      "loss": 0.8279,
      "step": 3778
    },
    {
      "epoch": 0.6335953054594992,
      "grad_norm": 0.3900938630104065,
      "learning_rate": 6.281924124957107e-06,
      "loss": 0.7894,
      "step": 3779
    },
    {
      "epoch": 0.6337629676202452,
      "grad_norm": 1.166131854057312,
      "learning_rate": 6.276883409320016e-06,
      "loss": 0.7686,
      "step": 3780
    },
    {
      "epoch": 0.6339306297809914,
      "grad_norm": 0.668857753276825,
      "learning_rate": 6.271843791676111e-06,
      "loss": 0.8604,
      "step": 3781
    },
    {
      "epoch": 0.6340982919417374,
      "grad_norm": 0.37729379534721375,
      "learning_rate": 6.266805273511635e-06,
      "loss": 0.9143,
      "step": 3782
    },
    {
      "epoch": 0.6342659541024835,
      "grad_norm": 0.39083096385002136,
      "learning_rate": 6.261767856312517e-06,
      "loss": 0.9144,
      "step": 3783
    },
    {
      "epoch": 0.6344336162632296,
      "grad_norm": 0.46704787015914917,
      "learning_rate": 6.256731541564344e-06,
      "loss": 0.9181,
      "step": 3784
    },
    {
      "epoch": 0.6346012784239757,
      "grad_norm": 0.3930618166923523,
      "learning_rate": 6.25169633075239e-06,
      "loss": 0.7963,
      "step": 3785
    },
    {
      "epoch": 0.6347689405847218,
      "grad_norm": 0.6432356834411621,
      "learning_rate": 6.246662225361603e-06,
      "loss": 0.9242,
      "step": 3786
    },
    {
      "epoch": 0.6349366027454679,
      "grad_norm": 0.4379623830318451,
      "learning_rate": 6.2416292268766075e-06,
      "loss": 0.7977,
      "step": 3787
    },
    {
      "epoch": 0.635104264906214,
      "grad_norm": 0.5398340225219727,
      "learning_rate": 6.236597336781689e-06,
      "loss": 0.8101,
      "step": 3788
    },
    {
      "epoch": 0.6352719270669601,
      "grad_norm": 0.4199172854423523,
      "learning_rate": 6.231566556560814e-06,
      "loss": 0.7766,
      "step": 3789
    },
    {
      "epoch": 0.6354395892277062,
      "grad_norm": 0.47107189893722534,
      "learning_rate": 6.226536887697628e-06,
      "loss": 0.8549,
      "step": 3790
    },
    {
      "epoch": 0.6356072513884523,
      "grad_norm": 0.3926093578338623,
      "learning_rate": 6.221508331675441e-06,
      "loss": 0.8375,
      "step": 3791
    },
    {
      "epoch": 0.6357749135491984,
      "grad_norm": 0.7772989869117737,
      "learning_rate": 6.216480889977236e-06,
      "loss": 0.8615,
      "step": 3792
    },
    {
      "epoch": 0.6359425757099445,
      "grad_norm": 0.7368307709693909,
      "learning_rate": 6.211454564085665e-06,
      "loss": 0.9141,
      "step": 3793
    },
    {
      "epoch": 0.6361102378706905,
      "grad_norm": 0.6034929752349854,
      "learning_rate": 6.206429355483056e-06,
      "loss": 0.871,
      "step": 3794
    },
    {
      "epoch": 0.6362779000314367,
      "grad_norm": 0.6859830021858215,
      "learning_rate": 6.201405265651406e-06,
      "loss": 0.8354,
      "step": 3795
    },
    {
      "epoch": 0.6364455621921827,
      "grad_norm": 0.5110700726509094,
      "learning_rate": 6.1963822960723805e-06,
      "loss": 0.8309,
      "step": 3796
    },
    {
      "epoch": 0.6366132243529289,
      "grad_norm": 0.43675047159194946,
      "learning_rate": 6.191360448227317e-06,
      "loss": 0.7561,
      "step": 3797
    },
    {
      "epoch": 0.6367808865136749,
      "grad_norm": 0.8488208055496216,
      "learning_rate": 6.186339723597216e-06,
      "loss": 0.8132,
      "step": 3798
    },
    {
      "epoch": 0.6369485486744211,
      "grad_norm": 0.754406213760376,
      "learning_rate": 6.181320123662755e-06,
      "loss": 0.8205,
      "step": 3799
    },
    {
      "epoch": 0.6371162108351671,
      "grad_norm": 0.6370114088058472,
      "learning_rate": 6.176301649904276e-06,
      "loss": 0.8229,
      "step": 3800
    },
    {
      "epoch": 0.6372838729959133,
      "grad_norm": 0.6114148497581482,
      "learning_rate": 6.17128430380179e-06,
      "loss": 0.8499,
      "step": 3801
    },
    {
      "epoch": 0.6374515351566593,
      "grad_norm": 0.4482009708881378,
      "learning_rate": 6.166268086834972e-06,
      "loss": 0.8606,
      "step": 3802
    },
    {
      "epoch": 0.6376191973174055,
      "grad_norm": 0.4591228663921356,
      "learning_rate": 6.1612530004831675e-06,
      "loss": 0.87,
      "step": 3803
    },
    {
      "epoch": 0.6377868594781515,
      "grad_norm": 0.37000200152397156,
      "learning_rate": 6.156239046225386e-06,
      "loss": 0.8392,
      "step": 3804
    },
    {
      "epoch": 0.6379545216388977,
      "grad_norm": 0.417057067155838,
      "learning_rate": 6.151226225540306e-06,
      "loss": 0.8257,
      "step": 3805
    },
    {
      "epoch": 0.6381221837996437,
      "grad_norm": 0.5155734419822693,
      "learning_rate": 6.146214539906277e-06,
      "loss": 0.8176,
      "step": 3806
    },
    {
      "epoch": 0.6382898459603898,
      "grad_norm": 0.7226120829582214,
      "learning_rate": 6.141203990801296e-06,
      "loss": 0.8362,
      "step": 3807
    },
    {
      "epoch": 0.6384575081211359,
      "grad_norm": 0.43286317586898804,
      "learning_rate": 6.136194579703039e-06,
      "loss": 0.8771,
      "step": 3808
    },
    {
      "epoch": 0.638625170281882,
      "grad_norm": 0.6042872667312622,
      "learning_rate": 6.131186308088846e-06,
      "loss": 0.91,
      "step": 3809
    },
    {
      "epoch": 0.6387928324426281,
      "grad_norm": 0.5188295841217041,
      "learning_rate": 6.126179177435722e-06,
      "loss": 0.8385,
      "step": 3810
    },
    {
      "epoch": 0.6389604946033742,
      "grad_norm": 0.45364266633987427,
      "learning_rate": 6.121173189220325e-06,
      "loss": 0.7914,
      "step": 3811
    },
    {
      "epoch": 0.6391281567641203,
      "grad_norm": 0.3534972369670868,
      "learning_rate": 6.116168344918982e-06,
      "loss": 0.857,
      "step": 3812
    },
    {
      "epoch": 0.6392958189248664,
      "grad_norm": 0.8765437602996826,
      "learning_rate": 6.11116464600769e-06,
      "loss": 0.8746,
      "step": 3813
    },
    {
      "epoch": 0.6394634810856125,
      "grad_norm": 0.4524255692958832,
      "learning_rate": 6.106162093962098e-06,
      "loss": 0.8807,
      "step": 3814
    },
    {
      "epoch": 0.6396311432463586,
      "grad_norm": 0.5076004862785339,
      "learning_rate": 6.101160690257523e-06,
      "loss": 0.826,
      "step": 3815
    },
    {
      "epoch": 0.6397988054071047,
      "grad_norm": 0.4958697557449341,
      "learning_rate": 6.0961604363689364e-06,
      "loss": 0.7956,
      "step": 3816
    },
    {
      "epoch": 0.6399664675678508,
      "grad_norm": 0.4035363793373108,
      "learning_rate": 6.091161333770978e-06,
      "loss": 0.8481,
      "step": 3817
    },
    {
      "epoch": 0.6401341297285968,
      "grad_norm": 0.5481789708137512,
      "learning_rate": 6.086163383937946e-06,
      "loss": 0.8822,
      "step": 3818
    },
    {
      "epoch": 0.640301791889343,
      "grad_norm": 0.60627281665802,
      "learning_rate": 6.081166588343798e-06,
      "loss": 0.8073,
      "step": 3819
    },
    {
      "epoch": 0.640469454050089,
      "grad_norm": 0.798624575138092,
      "learning_rate": 6.076170948462146e-06,
      "loss": 0.9205,
      "step": 3820
    },
    {
      "epoch": 0.6406371162108352,
      "grad_norm": 0.4653504192829132,
      "learning_rate": 6.07117646576627e-06,
      "loss": 0.7914,
      "step": 3821
    },
    {
      "epoch": 0.6408047783715812,
      "grad_norm": 0.7155263423919678,
      "learning_rate": 6.066183141729104e-06,
      "loss": 0.7911,
      "step": 3822
    },
    {
      "epoch": 0.6409724405323274,
      "grad_norm": 1.396693468093872,
      "learning_rate": 6.061190977823242e-06,
      "loss": 0.8241,
      "step": 3823
    },
    {
      "epoch": 0.6411401026930734,
      "grad_norm": 0.4433426558971405,
      "learning_rate": 6.056199975520936e-06,
      "loss": 0.8162,
      "step": 3824
    },
    {
      "epoch": 0.6413077648538196,
      "grad_norm": 0.4936043620109558,
      "learning_rate": 6.051210136294089e-06,
      "loss": 0.8271,
      "step": 3825
    },
    {
      "epoch": 0.6414754270145656,
      "grad_norm": 0.754987359046936,
      "learning_rate": 6.046221461614271e-06,
      "loss": 0.8933,
      "step": 3826
    },
    {
      "epoch": 0.6416430891753118,
      "grad_norm": 0.6823899745941162,
      "learning_rate": 6.041233952952703e-06,
      "loss": 0.9588,
      "step": 3827
    },
    {
      "epoch": 0.6418107513360578,
      "grad_norm": 0.3952237665653229,
      "learning_rate": 6.036247611780265e-06,
      "loss": 0.8251,
      "step": 3828
    },
    {
      "epoch": 0.641978413496804,
      "grad_norm": 0.505317211151123,
      "learning_rate": 6.031262439567487e-06,
      "loss": 0.921,
      "step": 3829
    },
    {
      "epoch": 0.64214607565755,
      "grad_norm": 0.3997824192047119,
      "learning_rate": 6.026278437784558e-06,
      "loss": 0.8279,
      "step": 3830
    },
    {
      "epoch": 0.6423137378182961,
      "grad_norm": 0.39301612973213196,
      "learning_rate": 6.021295607901323e-06,
      "loss": 0.8972,
      "step": 3831
    },
    {
      "epoch": 0.6424813999790422,
      "grad_norm": 0.3877316415309906,
      "learning_rate": 6.016313951387283e-06,
      "loss": 0.8126,
      "step": 3832
    },
    {
      "epoch": 0.6426490621397883,
      "grad_norm": 0.8617894053459167,
      "learning_rate": 6.011333469711592e-06,
      "loss": 0.7885,
      "step": 3833
    },
    {
      "epoch": 0.6428167243005344,
      "grad_norm": 0.39734694361686707,
      "learning_rate": 6.006354164343047e-06,
      "loss": 0.8481,
      "step": 3834
    },
    {
      "epoch": 0.6429843864612805,
      "grad_norm": 0.529140293598175,
      "learning_rate": 6.001376036750111e-06,
      "loss": 0.8495,
      "step": 3835
    },
    {
      "epoch": 0.6431520486220266,
      "grad_norm": 0.4331158399581909,
      "learning_rate": 5.996399088400895e-06,
      "loss": 0.8104,
      "step": 3836
    },
    {
      "epoch": 0.6433197107827727,
      "grad_norm": 1.3587584495544434,
      "learning_rate": 5.9914233207631655e-06,
      "loss": 0.8862,
      "step": 3837
    },
    {
      "epoch": 0.6434873729435188,
      "grad_norm": 0.4152337312698364,
      "learning_rate": 5.986448735304339e-06,
      "loss": 0.9072,
      "step": 3838
    },
    {
      "epoch": 0.6436550351042649,
      "grad_norm": 0.4645063877105713,
      "learning_rate": 5.981475333491477e-06,
      "loss": 0.9365,
      "step": 3839
    },
    {
      "epoch": 0.643822697265011,
      "grad_norm": 0.3764033019542694,
      "learning_rate": 5.976503116791299e-06,
      "loss": 0.78,
      "step": 3840
    },
    {
      "epoch": 0.6439903594257571,
      "grad_norm": 0.5858711004257202,
      "learning_rate": 5.971532086670174e-06,
      "loss": 0.8986,
      "step": 3841
    },
    {
      "epoch": 0.6441580215865031,
      "grad_norm": 0.5570114254951477,
      "learning_rate": 5.966562244594123e-06,
      "loss": 0.9575,
      "step": 3842
    },
    {
      "epoch": 0.6443256837472493,
      "grad_norm": 0.3770409822463989,
      "learning_rate": 5.9615935920288096e-06,
      "loss": 0.7898,
      "step": 3843
    },
    {
      "epoch": 0.6444933459079953,
      "grad_norm": 0.4317591190338135,
      "learning_rate": 5.956626130439552e-06,
      "loss": 0.8362,
      "step": 3844
    },
    {
      "epoch": 0.6446610080687415,
      "grad_norm": 0.4174540042877197,
      "learning_rate": 5.951659861291317e-06,
      "loss": 0.9139,
      "step": 3845
    },
    {
      "epoch": 0.6448286702294875,
      "grad_norm": 0.6102603673934937,
      "learning_rate": 5.946694786048719e-06,
      "loss": 0.7852,
      "step": 3846
    },
    {
      "epoch": 0.6449963323902337,
      "grad_norm": 0.4747770428657532,
      "learning_rate": 5.941730906176022e-06,
      "loss": 0.7641,
      "step": 3847
    },
    {
      "epoch": 0.6451639945509797,
      "grad_norm": 0.7990584373474121,
      "learning_rate": 5.936768223137133e-06,
      "loss": 0.7503,
      "step": 3848
    },
    {
      "epoch": 0.6453316567117259,
      "grad_norm": 0.7170785665512085,
      "learning_rate": 5.931806738395607e-06,
      "loss": 0.8904,
      "step": 3849
    },
    {
      "epoch": 0.6454993188724719,
      "grad_norm": 0.36233919858932495,
      "learning_rate": 5.926846453414649e-06,
      "loss": 0.7869,
      "step": 3850
    },
    {
      "epoch": 0.6456669810332181,
      "grad_norm": 0.46254944801330566,
      "learning_rate": 5.921887369657113e-06,
      "loss": 0.8671,
      "step": 3851
    },
    {
      "epoch": 0.6458346431939641,
      "grad_norm": 0.3812108039855957,
      "learning_rate": 5.916929488585488e-06,
      "loss": 0.8109,
      "step": 3852
    },
    {
      "epoch": 0.6460023053547103,
      "grad_norm": 0.42729702591896057,
      "learning_rate": 5.9119728116619165e-06,
      "loss": 0.8988,
      "step": 3853
    },
    {
      "epoch": 0.6461699675154563,
      "grad_norm": 0.40828776359558105,
      "learning_rate": 5.907017340348182e-06,
      "loss": 0.9616,
      "step": 3854
    },
    {
      "epoch": 0.6463376296762025,
      "grad_norm": 0.7432594299316406,
      "learning_rate": 5.9020630761057174e-06,
      "loss": 0.9197,
      "step": 3855
    },
    {
      "epoch": 0.6465052918369486,
      "grad_norm": 0.4486801028251648,
      "learning_rate": 5.897110020395601e-06,
      "loss": 0.8346,
      "step": 3856
    },
    {
      "epoch": 0.6466729539976946,
      "grad_norm": 0.8108722567558289,
      "learning_rate": 5.892158174678538e-06,
      "loss": 0.841,
      "step": 3857
    },
    {
      "epoch": 0.6468406161584408,
      "grad_norm": 0.5092005133628845,
      "learning_rate": 5.887207540414899e-06,
      "loss": 0.9681,
      "step": 3858
    },
    {
      "epoch": 0.6470082783191868,
      "grad_norm": 0.4808082580566406,
      "learning_rate": 5.8822581190646835e-06,
      "loss": 0.8459,
      "step": 3859
    },
    {
      "epoch": 0.647175940479933,
      "grad_norm": 0.5065873861312866,
      "learning_rate": 5.877309912087542e-06,
      "loss": 0.8047,
      "step": 3860
    },
    {
      "epoch": 0.647343602640679,
      "grad_norm": 0.8511838316917419,
      "learning_rate": 5.872362920942751e-06,
      "loss": 0.861,
      "step": 3861
    },
    {
      "epoch": 0.6475112648014252,
      "grad_norm": 0.4112839996814728,
      "learning_rate": 5.8674171470892506e-06,
      "loss": 0.8408,
      "step": 3862
    },
    {
      "epoch": 0.6476789269621712,
      "grad_norm": 0.6125646829605103,
      "learning_rate": 5.862472591985605e-06,
      "loss": 0.833,
      "step": 3863
    },
    {
      "epoch": 0.6478465891229174,
      "grad_norm": 0.378488689661026,
      "learning_rate": 5.857529257090027e-06,
      "loss": 0.844,
      "step": 3864
    },
    {
      "epoch": 0.6480142512836634,
      "grad_norm": 0.8819833993911743,
      "learning_rate": 5.852587143860368e-06,
      "loss": 0.9019,
      "step": 3865
    },
    {
      "epoch": 0.6481819134444096,
      "grad_norm": 0.43267491459846497,
      "learning_rate": 5.847646253754115e-06,
      "loss": 0.8093,
      "step": 3866
    },
    {
      "epoch": 0.6483495756051556,
      "grad_norm": 0.5931116938591003,
      "learning_rate": 5.8427065882284e-06,
      "loss": 0.8261,
      "step": 3867
    },
    {
      "epoch": 0.6485172377659018,
      "grad_norm": 0.6502438187599182,
      "learning_rate": 5.837768148739989e-06,
      "loss": 0.8822,
      "step": 3868
    },
    {
      "epoch": 0.6486848999266478,
      "grad_norm": 0.4747580289840698,
      "learning_rate": 5.832830936745294e-06,
      "loss": 0.9101,
      "step": 3869
    },
    {
      "epoch": 0.648852562087394,
      "grad_norm": 0.5370930433273315,
      "learning_rate": 5.827894953700357e-06,
      "loss": 0.9073,
      "step": 3870
    },
    {
      "epoch": 0.64902022424814,
      "grad_norm": 0.6176109313964844,
      "learning_rate": 5.822960201060858e-06,
      "loss": 0.815,
      "step": 3871
    },
    {
      "epoch": 0.6491878864088861,
      "grad_norm": 0.4676288068294525,
      "learning_rate": 5.818026680282123e-06,
      "loss": 0.8403,
      "step": 3872
    },
    {
      "epoch": 0.6493555485696322,
      "grad_norm": 0.45306962728500366,
      "learning_rate": 5.8130943928191e-06,
      "loss": 0.7817,
      "step": 3873
    },
    {
      "epoch": 0.6495232107303783,
      "grad_norm": 0.6352096199989319,
      "learning_rate": 5.80816334012639e-06,
      "loss": 0.7436,
      "step": 3874
    },
    {
      "epoch": 0.6496908728911244,
      "grad_norm": 0.6790918111801147,
      "learning_rate": 5.803233523658219e-06,
      "loss": 0.847,
      "step": 3875
    },
    {
      "epoch": 0.6498585350518705,
      "grad_norm": 0.43328049778938293,
      "learning_rate": 5.7983049448684444e-06,
      "loss": 0.8527,
      "step": 3876
    },
    {
      "epoch": 0.6500261972126166,
      "grad_norm": 0.4411693513393402,
      "learning_rate": 5.793377605210575e-06,
      "loss": 0.7853,
      "step": 3877
    },
    {
      "epoch": 0.6501938593733627,
      "grad_norm": 0.46323803067207336,
      "learning_rate": 5.788451506137736e-06,
      "loss": 0.8508,
      "step": 3878
    },
    {
      "epoch": 0.6503615215341088,
      "grad_norm": 0.7709136009216309,
      "learning_rate": 5.783526649102701e-06,
      "loss": 0.8091,
      "step": 3879
    },
    {
      "epoch": 0.6505291836948549,
      "grad_norm": 0.44342300295829773,
      "learning_rate": 5.7786030355578704e-06,
      "loss": 0.8796,
      "step": 3880
    },
    {
      "epoch": 0.650696845855601,
      "grad_norm": 0.7442541718482971,
      "learning_rate": 5.773680666955272e-06,
      "loss": 0.8136,
      "step": 3881
    },
    {
      "epoch": 0.6508645080163471,
      "grad_norm": 0.6671350002288818,
      "learning_rate": 5.768759544746585e-06,
      "loss": 0.8501,
      "step": 3882
    },
    {
      "epoch": 0.6510321701770931,
      "grad_norm": 0.4241235554218292,
      "learning_rate": 5.7638396703831e-06,
      "loss": 0.8782,
      "step": 3883
    },
    {
      "epoch": 0.6511998323378393,
      "grad_norm": 0.6073123812675476,
      "learning_rate": 5.758921045315751e-06,
      "loss": 0.7927,
      "step": 3884
    },
    {
      "epoch": 0.6513674944985853,
      "grad_norm": 0.48959341645240784,
      "learning_rate": 5.754003670995104e-06,
      "loss": 0.8589,
      "step": 3885
    },
    {
      "epoch": 0.6515351566593315,
      "grad_norm": 0.3820074200630188,
      "learning_rate": 5.749087548871349e-06,
      "loss": 0.8576,
      "step": 3886
    },
    {
      "epoch": 0.6517028188200775,
      "grad_norm": 0.473310649394989,
      "learning_rate": 5.744172680394313e-06,
      "loss": 0.7948,
      "step": 3887
    },
    {
      "epoch": 0.6518704809808237,
      "grad_norm": 0.4938386082649231,
      "learning_rate": 5.73925906701346e-06,
      "loss": 0.8689,
      "step": 3888
    },
    {
      "epoch": 0.6520381431415697,
      "grad_norm": 0.46223780512809753,
      "learning_rate": 5.734346710177863e-06,
      "loss": 0.8193,
      "step": 3889
    },
    {
      "epoch": 0.6522058053023159,
      "grad_norm": 0.34344786405563354,
      "learning_rate": 5.729435611336239e-06,
      "loss": 0.8681,
      "step": 3890
    },
    {
      "epoch": 0.6523734674630619,
      "grad_norm": 0.4444260895252228,
      "learning_rate": 5.724525771936941e-06,
      "loss": 0.8012,
      "step": 3891
    },
    {
      "epoch": 0.6525411296238081,
      "grad_norm": 0.4069541096687317,
      "learning_rate": 5.719617193427933e-06,
      "loss": 0.8493,
      "step": 3892
    },
    {
      "epoch": 0.6527087917845541,
      "grad_norm": 0.3641412854194641,
      "learning_rate": 5.714709877256817e-06,
      "loss": 0.8897,
      "step": 3893
    },
    {
      "epoch": 0.6528764539453003,
      "grad_norm": 0.5475788712501526,
      "learning_rate": 5.709803824870823e-06,
      "loss": 0.8944,
      "step": 3894
    },
    {
      "epoch": 0.6530441161060463,
      "grad_norm": 0.8999494314193726,
      "learning_rate": 5.704899037716803e-06,
      "loss": 0.8227,
      "step": 3895
    },
    {
      "epoch": 0.6532117782667924,
      "grad_norm": 0.5007481575012207,
      "learning_rate": 5.699995517241245e-06,
      "loss": 0.9822,
      "step": 3896
    },
    {
      "epoch": 0.6533794404275385,
      "grad_norm": 0.46740853786468506,
      "learning_rate": 5.695093264890258e-06,
      "loss": 0.8919,
      "step": 3897
    },
    {
      "epoch": 0.6535471025882846,
      "grad_norm": 0.4734141230583191,
      "learning_rate": 5.690192282109568e-06,
      "loss": 0.9623,
      "step": 3898
    },
    {
      "epoch": 0.6537147647490307,
      "grad_norm": 0.4099631905555725,
      "learning_rate": 5.685292570344547e-06,
      "loss": 0.9021,
      "step": 3899
    },
    {
      "epoch": 0.6538824269097768,
      "grad_norm": 0.44652417302131653,
      "learning_rate": 5.68039413104017e-06,
      "loss": 0.8015,
      "step": 3900
    },
    {
      "epoch": 0.6540500890705229,
      "grad_norm": 0.3861241638660431,
      "learning_rate": 5.675496965641057e-06,
      "loss": 0.8545,
      "step": 3901
    },
    {
      "epoch": 0.654217751231269,
      "grad_norm": 0.32507607340812683,
      "learning_rate": 5.6706010755914375e-06,
      "loss": 0.7932,
      "step": 3902
    },
    {
      "epoch": 0.6543854133920151,
      "grad_norm": 0.403301864862442,
      "learning_rate": 5.665706462335167e-06,
      "loss": 0.864,
      "step": 3903
    },
    {
      "epoch": 0.6545530755527612,
      "grad_norm": 0.34610313177108765,
      "learning_rate": 5.6608131273157366e-06,
      "loss": 0.8423,
      "step": 3904
    },
    {
      "epoch": 0.6547207377135073,
      "grad_norm": 0.33153748512268066,
      "learning_rate": 5.655921071976242e-06,
      "loss": 0.8157,
      "step": 3905
    },
    {
      "epoch": 0.6548883998742534,
      "grad_norm": 0.5666701793670654,
      "learning_rate": 5.651030297759419e-06,
      "loss": 0.8519,
      "step": 3906
    },
    {
      "epoch": 0.6550560620349994,
      "grad_norm": 0.658186137676239,
      "learning_rate": 5.646140806107614e-06,
      "loss": 0.9748,
      "step": 3907
    },
    {
      "epoch": 0.6552237241957456,
      "grad_norm": 0.5197761654853821,
      "learning_rate": 5.641252598462794e-06,
      "loss": 0.8331,
      "step": 3908
    },
    {
      "epoch": 0.6553913863564916,
      "grad_norm": 0.38345766067504883,
      "learning_rate": 5.636365676266556e-06,
      "loss": 0.8573,
      "step": 3909
    },
    {
      "epoch": 0.6555590485172378,
      "grad_norm": 0.5214496850967407,
      "learning_rate": 5.6314800409601176e-06,
      "loss": 0.7158,
      "step": 3910
    },
    {
      "epoch": 0.6557267106779838,
      "grad_norm": 0.43563833832740784,
      "learning_rate": 5.6265956939843116e-06,
      "loss": 0.7852,
      "step": 3911
    },
    {
      "epoch": 0.65589437283873,
      "grad_norm": 0.5575563907623291,
      "learning_rate": 5.621712636779589e-06,
      "loss": 0.8713,
      "step": 3912
    },
    {
      "epoch": 0.656062034999476,
      "grad_norm": 0.6472848057746887,
      "learning_rate": 5.616830870786022e-06,
      "loss": 0.8153,
      "step": 3913
    },
    {
      "epoch": 0.6562296971602222,
      "grad_norm": 0.49268069863319397,
      "learning_rate": 5.611950397443307e-06,
      "loss": 0.7555,
      "step": 3914
    },
    {
      "epoch": 0.6563973593209682,
      "grad_norm": 0.4301689863204956,
      "learning_rate": 5.607071218190764e-06,
      "loss": 0.9521,
      "step": 3915
    },
    {
      "epoch": 0.6565650214817144,
      "grad_norm": 0.3915613889694214,
      "learning_rate": 5.602193334467307e-06,
      "loss": 0.9636,
      "step": 3916
    },
    {
      "epoch": 0.6567326836424604,
      "grad_norm": 0.8180561661720276,
      "learning_rate": 5.597316747711495e-06,
      "loss": 0.7843,
      "step": 3917
    },
    {
      "epoch": 0.6569003458032066,
      "grad_norm": 0.4958726465702057,
      "learning_rate": 5.592441459361493e-06,
      "loss": 0.8674,
      "step": 3918
    },
    {
      "epoch": 0.6570680079639526,
      "grad_norm": 0.3985406458377838,
      "learning_rate": 5.587567470855079e-06,
      "loss": 0.8227,
      "step": 3919
    },
    {
      "epoch": 0.6572356701246987,
      "grad_norm": 0.6212599277496338,
      "learning_rate": 5.582694783629663e-06,
      "loss": 0.8025,
      "step": 3920
    },
    {
      "epoch": 0.6574033322854448,
      "grad_norm": 0.5638007521629333,
      "learning_rate": 5.577823399122251e-06,
      "loss": 0.7337,
      "step": 3921
    },
    {
      "epoch": 0.6575709944461909,
      "grad_norm": 0.3695148229598999,
      "learning_rate": 5.572953318769475e-06,
      "loss": 0.7863,
      "step": 3922
    },
    {
      "epoch": 0.657738656606937,
      "grad_norm": 0.40442386269569397,
      "learning_rate": 5.5680845440075885e-06,
      "loss": 0.8887,
      "step": 3923
    },
    {
      "epoch": 0.6579063187676831,
      "grad_norm": 0.8187882304191589,
      "learning_rate": 5.563217076272449e-06,
      "loss": 0.9607,
      "step": 3924
    },
    {
      "epoch": 0.6580739809284292,
      "grad_norm": 0.4451625347137451,
      "learning_rate": 5.55835091699953e-06,
      "loss": 0.8935,
      "step": 3925
    },
    {
      "epoch": 0.6582416430891753,
      "grad_norm": 0.6073238849639893,
      "learning_rate": 5.55348606762393e-06,
      "loss": 0.9452,
      "step": 3926
    },
    {
      "epoch": 0.6584093052499214,
      "grad_norm": 0.8138495087623596,
      "learning_rate": 5.548622529580343e-06,
      "loss": 0.7612,
      "step": 3927
    },
    {
      "epoch": 0.6585769674106675,
      "grad_norm": 0.3803512454032898,
      "learning_rate": 5.543760304303098e-06,
      "loss": 0.8742,
      "step": 3928
    },
    {
      "epoch": 0.6587446295714136,
      "grad_norm": 0.4340585470199585,
      "learning_rate": 5.538899393226122e-06,
      "loss": 0.8225,
      "step": 3929
    },
    {
      "epoch": 0.6589122917321597,
      "grad_norm": 0.35208043456077576,
      "learning_rate": 5.534039797782948e-06,
      "loss": 0.8354,
      "step": 3930
    },
    {
      "epoch": 0.6590799538929057,
      "grad_norm": 0.41237229108810425,
      "learning_rate": 5.529181519406745e-06,
      "loss": 0.8451,
      "step": 3931
    },
    {
      "epoch": 0.6592476160536519,
      "grad_norm": 0.7041605710983276,
      "learning_rate": 5.524324559530269e-06,
      "loss": 0.7869,
      "step": 3932
    },
    {
      "epoch": 0.6594152782143979,
      "grad_norm": 0.416987806558609,
      "learning_rate": 5.519468919585908e-06,
      "loss": 0.8778,
      "step": 3933
    },
    {
      "epoch": 0.6595829403751441,
      "grad_norm": 0.4939228594303131,
      "learning_rate": 5.514614601005641e-06,
      "loss": 0.8206,
      "step": 3934
    },
    {
      "epoch": 0.6597506025358901,
      "grad_norm": 0.4769986867904663,
      "learning_rate": 5.509761605221067e-06,
      "loss": 0.8511,
      "step": 3935
    },
    {
      "epoch": 0.6599182646966363,
      "grad_norm": 0.3630758225917816,
      "learning_rate": 5.504909933663398e-06,
      "loss": 0.8545,
      "step": 3936
    },
    {
      "epoch": 0.6600859268573823,
      "grad_norm": 1.8517969846725464,
      "learning_rate": 5.5000595877634555e-06,
      "loss": 0.9223,
      "step": 3937
    },
    {
      "epoch": 0.6602535890181285,
      "grad_norm": 0.6270850300788879,
      "learning_rate": 5.495210568951661e-06,
      "loss": 0.8476,
      "step": 3938
    },
    {
      "epoch": 0.6604212511788745,
      "grad_norm": 0.7449013590812683,
      "learning_rate": 5.490362878658049e-06,
      "loss": 0.884,
      "step": 3939
    },
    {
      "epoch": 0.6605889133396207,
      "grad_norm": 0.3724488317966461,
      "learning_rate": 5.485516518312269e-06,
      "loss": 0.8121,
      "step": 3940
    },
    {
      "epoch": 0.6607565755003668,
      "grad_norm": 0.6408987641334534,
      "learning_rate": 5.480671489343568e-06,
      "loss": 0.8105,
      "step": 3941
    },
    {
      "epoch": 0.6609242376611129,
      "grad_norm": 0.466304749250412,
      "learning_rate": 5.475827793180808e-06,
      "loss": 0.9077,
      "step": 3942
    },
    {
      "epoch": 0.661091899821859,
      "grad_norm": 0.46313750743865967,
      "learning_rate": 5.470985431252455e-06,
      "loss": 0.8159,
      "step": 3943
    },
    {
      "epoch": 0.661259561982605,
      "grad_norm": 0.723298192024231,
      "learning_rate": 5.4661444049865755e-06,
      "loss": 0.8625,
      "step": 3944
    },
    {
      "epoch": 0.6614272241433512,
      "grad_norm": 0.4626912772655487,
      "learning_rate": 5.461304715810858e-06,
      "loss": 0.7983,
      "step": 3945
    },
    {
      "epoch": 0.6615948863040972,
      "grad_norm": 0.626879870891571,
      "learning_rate": 5.456466365152577e-06,
      "loss": 0.7478,
      "step": 3946
    },
    {
      "epoch": 0.6617625484648434,
      "grad_norm": 0.6091106534004211,
      "learning_rate": 5.451629354438631e-06,
      "loss": 0.7975,
      "step": 3947
    },
    {
      "epoch": 0.6619302106255894,
      "grad_norm": 0.34452587366104126,
      "learning_rate": 5.4467936850955105e-06,
      "loss": 0.8352,
      "step": 3948
    },
    {
      "epoch": 0.6620978727863356,
      "grad_norm": 0.43186381459236145,
      "learning_rate": 5.441959358549309e-06,
      "loss": 0.9035,
      "step": 3949
    },
    {
      "epoch": 0.6622655349470816,
      "grad_norm": 0.46442583203315735,
      "learning_rate": 5.437126376225738e-06,
      "loss": 0.7824,
      "step": 3950
    },
    {
      "epoch": 0.6624331971078278,
      "grad_norm": 0.40834471583366394,
      "learning_rate": 5.4322947395500955e-06,
      "loss": 0.8286,
      "step": 3951
    },
    {
      "epoch": 0.6626008592685738,
      "grad_norm": 0.40990063548088074,
      "learning_rate": 5.4274644499473e-06,
      "loss": 0.8926,
      "step": 3952
    },
    {
      "epoch": 0.66276852142932,
      "grad_norm": 0.335954487323761,
      "learning_rate": 5.422635508841859e-06,
      "loss": 0.9434,
      "step": 3953
    },
    {
      "epoch": 0.662936183590066,
      "grad_norm": 0.3645780384540558,
      "learning_rate": 5.4178079176578836e-06,
      "loss": 0.9416,
      "step": 3954
    },
    {
      "epoch": 0.6631038457508122,
      "grad_norm": 0.46499937772750854,
      "learning_rate": 5.412981677819094e-06,
      "loss": 0.7645,
      "step": 3955
    },
    {
      "epoch": 0.6632715079115582,
      "grad_norm": 0.3888450264930725,
      "learning_rate": 5.408156790748813e-06,
      "loss": 0.8416,
      "step": 3956
    },
    {
      "epoch": 0.6634391700723044,
      "grad_norm": 0.5346174240112305,
      "learning_rate": 5.40333325786995e-06,
      "loss": 0.8997,
      "step": 3957
    },
    {
      "epoch": 0.6636068322330504,
      "grad_norm": 0.424798846244812,
      "learning_rate": 5.398511080605027e-06,
      "loss": 0.7671,
      "step": 3958
    },
    {
      "epoch": 0.6637744943937965,
      "grad_norm": 0.45263102650642395,
      "learning_rate": 5.39369026037617e-06,
      "loss": 0.7754,
      "step": 3959
    },
    {
      "epoch": 0.6639421565545426,
      "grad_norm": 0.6530085206031799,
      "learning_rate": 5.388870798605089e-06,
      "loss": 0.7891,
      "step": 3960
    },
    {
      "epoch": 0.6641098187152887,
      "grad_norm": 0.6561893820762634,
      "learning_rate": 5.384052696713118e-06,
      "loss": 0.8694,
      "step": 3961
    },
    {
      "epoch": 0.6642774808760348,
      "grad_norm": 0.594525158405304,
      "learning_rate": 5.379235956121156e-06,
      "loss": 0.7908,
      "step": 3962
    },
    {
      "epoch": 0.6644451430367809,
      "grad_norm": 0.4702717661857605,
      "learning_rate": 5.37442057824973e-06,
      "loss": 0.8084,
      "step": 3963
    },
    {
      "epoch": 0.664612805197527,
      "grad_norm": 0.5309458374977112,
      "learning_rate": 5.369606564518957e-06,
      "loss": 0.7848,
      "step": 3964
    },
    {
      "epoch": 0.6647804673582731,
      "grad_norm": 0.38657906651496887,
      "learning_rate": 5.364793916348544e-06,
      "loss": 0.8487,
      "step": 3965
    },
    {
      "epoch": 0.6649481295190192,
      "grad_norm": 0.560144305229187,
      "learning_rate": 5.359982635157799e-06,
      "loss": 0.8556,
      "step": 3966
    },
    {
      "epoch": 0.6651157916797653,
      "grad_norm": 0.47209808230400085,
      "learning_rate": 5.355172722365634e-06,
      "loss": 0.8144,
      "step": 3967
    },
    {
      "epoch": 0.6652834538405114,
      "grad_norm": 0.5120728015899658,
      "learning_rate": 5.3503641793905485e-06,
      "loss": 0.9506,
      "step": 3968
    },
    {
      "epoch": 0.6654511160012575,
      "grad_norm": 0.38565343618392944,
      "learning_rate": 5.345557007650642e-06,
      "loss": 0.8712,
      "step": 3969
    },
    {
      "epoch": 0.6656187781620035,
      "grad_norm": 0.5157579183578491,
      "learning_rate": 5.340751208563611e-06,
      "loss": 0.8789,
      "step": 3970
    },
    {
      "epoch": 0.6657864403227497,
      "grad_norm": 0.5449783205986023,
      "learning_rate": 5.3359467835467394e-06,
      "loss": 0.809,
      "step": 3971
    },
    {
      "epoch": 0.6659541024834957,
      "grad_norm": 0.4267489016056061,
      "learning_rate": 5.33114373401692e-06,
      "loss": 0.8642,
      "step": 3972
    },
    {
      "epoch": 0.6661217646442419,
      "grad_norm": 0.38633015751838684,
      "learning_rate": 5.326342061390621e-06,
      "loss": 0.8654,
      "step": 3973
    },
    {
      "epoch": 0.6662894268049879,
      "grad_norm": 0.5656725168228149,
      "learning_rate": 5.321541767083927e-06,
      "loss": 0.888,
      "step": 3974
    },
    {
      "epoch": 0.6664570889657341,
      "grad_norm": 0.4438325762748718,
      "learning_rate": 5.3167428525124976e-06,
      "loss": 0.8804,
      "step": 3975
    },
    {
      "epoch": 0.6666247511264801,
      "grad_norm": 0.4937484860420227,
      "learning_rate": 5.311945319091589e-06,
      "loss": 0.9184,
      "step": 3976
    },
    {
      "epoch": 0.6667924132872263,
      "grad_norm": 0.4187779724597931,
      "learning_rate": 5.307149168236061e-06,
      "loss": 1.0089,
      "step": 3977
    },
    {
      "epoch": 0.6669600754479723,
      "grad_norm": 0.4334414601325989,
      "learning_rate": 5.30235440136035e-06,
      "loss": 0.8644,
      "step": 3978
    },
    {
      "epoch": 0.6671277376087185,
      "grad_norm": 0.6198210716247559,
      "learning_rate": 5.297561019878501e-06,
      "loss": 0.8508,
      "step": 3979
    },
    {
      "epoch": 0.6672953997694645,
      "grad_norm": 0.6033726930618286,
      "learning_rate": 5.292769025204137e-06,
      "loss": 0.7934,
      "step": 3980
    },
    {
      "epoch": 0.6674630619302107,
      "grad_norm": 0.4969884753227234,
      "learning_rate": 5.28797841875047e-06,
      "loss": 0.8088,
      "step": 3981
    },
    {
      "epoch": 0.6676307240909567,
      "grad_norm": 0.35007086396217346,
      "learning_rate": 5.283189201930316e-06,
      "loss": 0.7962,
      "step": 3982
    },
    {
      "epoch": 0.6677983862517028,
      "grad_norm": 0.4103473126888275,
      "learning_rate": 5.2784013761560794e-06,
      "loss": 0.8526,
      "step": 3983
    },
    {
      "epoch": 0.6679660484124489,
      "grad_norm": 0.5877014398574829,
      "learning_rate": 5.273614942839742e-06,
      "loss": 0.8872,
      "step": 3984
    },
    {
      "epoch": 0.668133710573195,
      "grad_norm": 0.39554399251937866,
      "learning_rate": 5.26882990339288e-06,
      "loss": 0.8546,
      "step": 3985
    },
    {
      "epoch": 0.6683013727339411,
      "grad_norm": 0.8084924817085266,
      "learning_rate": 5.2640462592266695e-06,
      "loss": 0.7945,
      "step": 3986
    },
    {
      "epoch": 0.6684690348946872,
      "grad_norm": 0.5444297790527344,
      "learning_rate": 5.259264011751857e-06,
      "loss": 0.9109,
      "step": 3987
    },
    {
      "epoch": 0.6686366970554333,
      "grad_norm": 0.4030567705631256,
      "learning_rate": 5.254483162378795e-06,
      "loss": 0.8065,
      "step": 3988
    },
    {
      "epoch": 0.6688043592161794,
      "grad_norm": 0.8205330967903137,
      "learning_rate": 5.249703712517411e-06,
      "loss": 0.7667,
      "step": 3989
    },
    {
      "epoch": 0.6689720213769255,
      "grad_norm": 0.4742152690887451,
      "learning_rate": 5.244925663577221e-06,
      "loss": 0.8684,
      "step": 3990
    },
    {
      "epoch": 0.6691396835376716,
      "grad_norm": 0.48614606261253357,
      "learning_rate": 5.240149016967338e-06,
      "loss": 0.8188,
      "step": 3991
    },
    {
      "epoch": 0.6693073456984177,
      "grad_norm": 0.5022182464599609,
      "learning_rate": 5.235373774096446e-06,
      "loss": 0.8433,
      "step": 3992
    },
    {
      "epoch": 0.6694750078591638,
      "grad_norm": 0.5748201608657837,
      "learning_rate": 5.230599936372831e-06,
      "loss": 0.9058,
      "step": 3993
    },
    {
      "epoch": 0.6696426700199098,
      "grad_norm": 0.49480316042900085,
      "learning_rate": 5.225827505204355e-06,
      "loss": 0.8084,
      "step": 3994
    },
    {
      "epoch": 0.669810332180656,
      "grad_norm": 0.4499543309211731,
      "learning_rate": 5.221056481998459e-06,
      "loss": 0.8476,
      "step": 3995
    },
    {
      "epoch": 0.669977994341402,
      "grad_norm": 0.6074202656745911,
      "learning_rate": 5.21628686816219e-06,
      "loss": 0.8081,
      "step": 3996
    },
    {
      "epoch": 0.6701456565021482,
      "grad_norm": 0.388506144285202,
      "learning_rate": 5.2115186651021575e-06,
      "loss": 0.8465,
      "step": 3997
    },
    {
      "epoch": 0.6703133186628942,
      "grad_norm": 0.40389159321784973,
      "learning_rate": 5.2067518742245615e-06,
      "loss": 0.8688,
      "step": 3998
    },
    {
      "epoch": 0.6704809808236404,
      "grad_norm": 0.5128644108772278,
      "learning_rate": 5.201986496935196e-06,
      "loss": 0.8765,
      "step": 3999
    },
    {
      "epoch": 0.6706486429843864,
      "grad_norm": 0.38812342286109924,
      "learning_rate": 5.1972225346394216e-06,
      "loss": 0.8367,
      "step": 4000
    },
    {
      "epoch": 0.6708163051451326,
      "grad_norm": 0.6672189831733704,
      "learning_rate": 5.192459988742193e-06,
      "loss": 0.841,
      "step": 4001
    },
    {
      "epoch": 0.6709839673058786,
      "grad_norm": 0.39312562346458435,
      "learning_rate": 5.187698860648054e-06,
      "loss": 0.8424,
      "step": 4002
    },
    {
      "epoch": 0.6711516294666248,
      "grad_norm": 0.7264593839645386,
      "learning_rate": 5.182939151761102e-06,
      "loss": 0.7477,
      "step": 4003
    },
    {
      "epoch": 0.6713192916273708,
      "grad_norm": 0.5246336460113525,
      "learning_rate": 5.178180863485043e-06,
      "loss": 0.858,
      "step": 4004
    },
    {
      "epoch": 0.671486953788117,
      "grad_norm": 0.6613538265228271,
      "learning_rate": 5.173423997223159e-06,
      "loss": 0.8167,
      "step": 4005
    },
    {
      "epoch": 0.671654615948863,
      "grad_norm": 0.5246715545654297,
      "learning_rate": 5.168668554378305e-06,
      "loss": 0.8786,
      "step": 4006
    },
    {
      "epoch": 0.6718222781096092,
      "grad_norm": 0.4578244686126709,
      "learning_rate": 5.163914536352919e-06,
      "loss": 0.7704,
      "step": 4007
    },
    {
      "epoch": 0.6719899402703552,
      "grad_norm": 1.217862606048584,
      "learning_rate": 5.159161944549017e-06,
      "loss": 0.8864,
      "step": 4008
    },
    {
      "epoch": 0.6721576024311013,
      "grad_norm": 0.46265730261802673,
      "learning_rate": 5.154410780368202e-06,
      "loss": 0.8769,
      "step": 4009
    },
    {
      "epoch": 0.6723252645918474,
      "grad_norm": 0.49288302659988403,
      "learning_rate": 5.1496610452116516e-06,
      "loss": 0.886,
      "step": 4010
    },
    {
      "epoch": 0.6724929267525935,
      "grad_norm": 0.4994318187236786,
      "learning_rate": 5.144912740480122e-06,
      "loss": 0.7578,
      "step": 4011
    },
    {
      "epoch": 0.6726605889133396,
      "grad_norm": 0.6784866452217102,
      "learning_rate": 5.14016586757394e-06,
      "loss": 0.7887,
      "step": 4012
    },
    {
      "epoch": 0.6728282510740857,
      "grad_norm": 0.40225499868392944,
      "learning_rate": 5.135420427893026e-06,
      "loss": 0.8414,
      "step": 4013
    },
    {
      "epoch": 0.6729959132348318,
      "grad_norm": 0.4047643840312958,
      "learning_rate": 5.13067642283686e-06,
      "loss": 0.9051,
      "step": 4014
    },
    {
      "epoch": 0.6731635753955779,
      "grad_norm": 0.3079306483268738,
      "learning_rate": 5.125933853804519e-06,
      "loss": 0.835,
      "step": 4015
    },
    {
      "epoch": 0.673331237556324,
      "grad_norm": 0.41882896423339844,
      "learning_rate": 5.121192722194637e-06,
      "loss": 0.8247,
      "step": 4016
    },
    {
      "epoch": 0.6734988997170701,
      "grad_norm": 0.8016254901885986,
      "learning_rate": 5.116453029405431e-06,
      "loss": 0.7974,
      "step": 4017
    },
    {
      "epoch": 0.6736665618778162,
      "grad_norm": 0.4218415915966034,
      "learning_rate": 5.111714776834704e-06,
      "loss": 0.899,
      "step": 4018
    },
    {
      "epoch": 0.6738342240385623,
      "grad_norm": 0.4429084062576294,
      "learning_rate": 5.106977965879816e-06,
      "loss": 0.824,
      "step": 4019
    },
    {
      "epoch": 0.6740018861993083,
      "grad_norm": 0.5343514084815979,
      "learning_rate": 5.1022425979377174e-06,
      "loss": 0.874,
      "step": 4020
    },
    {
      "epoch": 0.6741695483600545,
      "grad_norm": 0.37769246101379395,
      "learning_rate": 5.0975086744049265e-06,
      "loss": 0.8485,
      "step": 4021
    },
    {
      "epoch": 0.6743372105208005,
      "grad_norm": 0.6687188148498535,
      "learning_rate": 5.092776196677529e-06,
      "loss": 0.8538,
      "step": 4022
    },
    {
      "epoch": 0.6745048726815467,
      "grad_norm": 0.5082492232322693,
      "learning_rate": 5.088045166151201e-06,
      "loss": 0.8869,
      "step": 4023
    },
    {
      "epoch": 0.6746725348422927,
      "grad_norm": 0.5272005200386047,
      "learning_rate": 5.083315584221174e-06,
      "loss": 0.7843,
      "step": 4024
    },
    {
      "epoch": 0.6748401970030389,
      "grad_norm": 0.506908118724823,
      "learning_rate": 5.078587452282267e-06,
      "loss": 0.901,
      "step": 4025
    },
    {
      "epoch": 0.6750078591637849,
      "grad_norm": 0.46326175332069397,
      "learning_rate": 5.07386077172886e-06,
      "loss": 0.8058,
      "step": 4026
    },
    {
      "epoch": 0.6751755213245311,
      "grad_norm": 0.7596170902252197,
      "learning_rate": 5.069135543954909e-06,
      "loss": 0.8804,
      "step": 4027
    },
    {
      "epoch": 0.6753431834852772,
      "grad_norm": 0.7367517352104187,
      "learning_rate": 5.064411770353941e-06,
      "loss": 0.9244,
      "step": 4028
    },
    {
      "epoch": 0.6755108456460233,
      "grad_norm": 0.5703383684158325,
      "learning_rate": 5.059689452319068e-06,
      "loss": 0.9342,
      "step": 4029
    },
    {
      "epoch": 0.6756785078067694,
      "grad_norm": 0.5530908703804016,
      "learning_rate": 5.0549685912429434e-06,
      "loss": 0.8298,
      "step": 4030
    },
    {
      "epoch": 0.6758461699675155,
      "grad_norm": 0.530229389667511,
      "learning_rate": 5.050249188517814e-06,
      "loss": 0.8117,
      "step": 4031
    },
    {
      "epoch": 0.6760138321282616,
      "grad_norm": 0.4607187807559967,
      "learning_rate": 5.045531245535495e-06,
      "loss": 0.9085,
      "step": 4032
    },
    {
      "epoch": 0.6761814942890076,
      "grad_norm": 0.4605083167552948,
      "learning_rate": 5.040814763687358e-06,
      "loss": 0.8637,
      "step": 4033
    },
    {
      "epoch": 0.6763491564497538,
      "grad_norm": 0.38844844698905945,
      "learning_rate": 5.03609974436436e-06,
      "loss": 0.8292,
      "step": 4034
    },
    {
      "epoch": 0.6765168186104998,
      "grad_norm": 0.46526795625686646,
      "learning_rate": 5.031386188957013e-06,
      "loss": 0.8265,
      "step": 4035
    },
    {
      "epoch": 0.676684480771246,
      "grad_norm": 0.39576253294944763,
      "learning_rate": 5.026674098855402e-06,
      "loss": 0.7683,
      "step": 4036
    },
    {
      "epoch": 0.676852142931992,
      "grad_norm": 0.5031571388244629,
      "learning_rate": 5.021963475449185e-06,
      "loss": 0.9241,
      "step": 4037
    },
    {
      "epoch": 0.6770198050927382,
      "grad_norm": 0.39039212465286255,
      "learning_rate": 5.017254320127584e-06,
      "loss": 0.8534,
      "step": 4038
    },
    {
      "epoch": 0.6771874672534842,
      "grad_norm": 0.4956362545490265,
      "learning_rate": 5.012546634279378e-06,
      "loss": 0.7214,
      "step": 4039
    },
    {
      "epoch": 0.6773551294142304,
      "grad_norm": 0.4101143181324005,
      "learning_rate": 5.007840419292933e-06,
      "loss": 0.8868,
      "step": 4040
    },
    {
      "epoch": 0.6775227915749764,
      "grad_norm": 0.3842748701572418,
      "learning_rate": 5.0031356765561625e-06,
      "loss": 0.7652,
      "step": 4041
    },
    {
      "epoch": 0.6776904537357226,
      "grad_norm": 0.43894824385643005,
      "learning_rate": 4.998432407456559e-06,
      "loss": 0.7629,
      "step": 4042
    },
    {
      "epoch": 0.6778581158964686,
      "grad_norm": 0.6596818566322327,
      "learning_rate": 4.993730613381173e-06,
      "loss": 0.9031,
      "step": 4043
    },
    {
      "epoch": 0.6780257780572148,
      "grad_norm": 0.3763595223426819,
      "learning_rate": 4.989030295716615e-06,
      "loss": 0.8784,
      "step": 4044
    },
    {
      "epoch": 0.6781934402179608,
      "grad_norm": 0.4675236642360687,
      "learning_rate": 4.984331455849077e-06,
      "loss": 0.7289,
      "step": 4045
    },
    {
      "epoch": 0.678361102378707,
      "grad_norm": 0.5311740040779114,
      "learning_rate": 4.979634095164298e-06,
      "loss": 0.8654,
      "step": 4046
    },
    {
      "epoch": 0.678528764539453,
      "grad_norm": 0.3968749940395355,
      "learning_rate": 4.974938215047595e-06,
      "loss": 0.8619,
      "step": 4047
    },
    {
      "epoch": 0.6786964267001991,
      "grad_norm": 0.4369450509548187,
      "learning_rate": 4.970243816883835e-06,
      "loss": 0.8361,
      "step": 4048
    },
    {
      "epoch": 0.6788640888609452,
      "grad_norm": 0.723330020904541,
      "learning_rate": 4.965550902057453e-06,
      "loss": 0.9035,
      "step": 4049
    },
    {
      "epoch": 0.6790317510216913,
      "grad_norm": 0.9839751720428467,
      "learning_rate": 4.960859471952451e-06,
      "loss": 0.8032,
      "step": 4050
    },
    {
      "epoch": 0.6791994131824374,
      "grad_norm": 0.3737811744213104,
      "learning_rate": 4.956169527952392e-06,
      "loss": 0.8817,
      "step": 4051
    },
    {
      "epoch": 0.6793670753431835,
      "grad_norm": 0.5440712571144104,
      "learning_rate": 4.951481071440398e-06,
      "loss": 0.7694,
      "step": 4052
    },
    {
      "epoch": 0.6795347375039296,
      "grad_norm": 0.5463842749595642,
      "learning_rate": 4.946794103799146e-06,
      "loss": 0.8517,
      "step": 4053
    },
    {
      "epoch": 0.6797023996646757,
      "grad_norm": 0.5319167375564575,
      "learning_rate": 4.94210862641089e-06,
      "loss": 0.8257,
      "step": 4054
    },
    {
      "epoch": 0.6798700618254218,
      "grad_norm": 0.7051066756248474,
      "learning_rate": 4.9374246406574254e-06,
      "loss": 0.8603,
      "step": 4055
    },
    {
      "epoch": 0.6800377239861679,
      "grad_norm": 0.4666949212551117,
      "learning_rate": 4.932742147920128e-06,
      "loss": 0.8018,
      "step": 4056
    },
    {
      "epoch": 0.680205386146914,
      "grad_norm": 0.38911348581314087,
      "learning_rate": 4.928061149579919e-06,
      "loss": 0.8546,
      "step": 4057
    },
    {
      "epoch": 0.6803730483076601,
      "grad_norm": 0.39490512013435364,
      "learning_rate": 4.923381647017278e-06,
      "loss": 0.8482,
      "step": 4058
    },
    {
      "epoch": 0.6805407104684061,
      "grad_norm": 0.4537765085697174,
      "learning_rate": 4.918703641612255e-06,
      "loss": 0.7966,
      "step": 4059
    },
    {
      "epoch": 0.6807083726291523,
      "grad_norm": 0.43742606043815613,
      "learning_rate": 4.914027134744446e-06,
      "loss": 0.778,
      "step": 4060
    },
    {
      "epoch": 0.6808760347898983,
      "grad_norm": 0.4226575195789337,
      "learning_rate": 4.909352127793016e-06,
      "loss": 0.8533,
      "step": 4061
    },
    {
      "epoch": 0.6810436969506445,
      "grad_norm": 0.479563444852829,
      "learning_rate": 4.904678622136682e-06,
      "loss": 0.8895,
      "step": 4062
    },
    {
      "epoch": 0.6812113591113905,
      "grad_norm": 0.5353012084960938,
      "learning_rate": 4.900006619153711e-06,
      "loss": 0.8008,
      "step": 4063
    },
    {
      "epoch": 0.6813790212721367,
      "grad_norm": 0.4960290193557739,
      "learning_rate": 4.895336120221945e-06,
      "loss": 0.8606,
      "step": 4064
    },
    {
      "epoch": 0.6815466834328827,
      "grad_norm": 0.3402813971042633,
      "learning_rate": 4.890667126718763e-06,
      "loss": 0.7323,
      "step": 4065
    },
    {
      "epoch": 0.6817143455936289,
      "grad_norm": 0.4431888163089752,
      "learning_rate": 4.885999640021117e-06,
      "loss": 0.8602,
      "step": 4066
    },
    {
      "epoch": 0.6818820077543749,
      "grad_norm": 0.4127519726753235,
      "learning_rate": 4.881333661505504e-06,
      "loss": 0.8083,
      "step": 4067
    },
    {
      "epoch": 0.6820496699151211,
      "grad_norm": 0.36416029930114746,
      "learning_rate": 4.876669192547972e-06,
      "loss": 0.8568,
      "step": 4068
    },
    {
      "epoch": 0.6822173320758671,
      "grad_norm": 0.7066261768341064,
      "learning_rate": 4.872006234524138e-06,
      "loss": 0.957,
      "step": 4069
    },
    {
      "epoch": 0.6823849942366133,
      "grad_norm": 0.5453369617462158,
      "learning_rate": 4.867344788809172e-06,
      "loss": 0.9183,
      "step": 4070
    },
    {
      "epoch": 0.6825526563973593,
      "grad_norm": 0.5057035088539124,
      "learning_rate": 4.862684856777776e-06,
      "loss": 0.831,
      "step": 4071
    },
    {
      "epoch": 0.6827203185581054,
      "grad_norm": 0.6210739016532898,
      "learning_rate": 4.8580264398042355e-06,
      "loss": 0.7656,
      "step": 4072
    },
    {
      "epoch": 0.6828879807188515,
      "grad_norm": 0.39194726943969727,
      "learning_rate": 4.853369539262365e-06,
      "loss": 0.8204,
      "step": 4073
    },
    {
      "epoch": 0.6830556428795976,
      "grad_norm": 0.39771005511283875,
      "learning_rate": 4.848714156525547e-06,
      "loss": 0.8624,
      "step": 4074
    },
    {
      "epoch": 0.6832233050403437,
      "grad_norm": 0.5039075613021851,
      "learning_rate": 4.844060292966719e-06,
      "loss": 0.8323,
      "step": 4075
    },
    {
      "epoch": 0.6833909672010898,
      "grad_norm": 0.5027855634689331,
      "learning_rate": 4.839407949958349e-06,
      "loss": 0.8624,
      "step": 4076
    },
    {
      "epoch": 0.6835586293618359,
      "grad_norm": 0.46842116117477417,
      "learning_rate": 4.834757128872477e-06,
      "loss": 0.9685,
      "step": 4077
    },
    {
      "epoch": 0.683726291522582,
      "grad_norm": 0.5276632308959961,
      "learning_rate": 4.830107831080692e-06,
      "loss": 0.8333,
      "step": 4078
    },
    {
      "epoch": 0.6838939536833281,
      "grad_norm": 0.40876996517181396,
      "learning_rate": 4.825460057954124e-06,
      "loss": 0.8593,
      "step": 4079
    },
    {
      "epoch": 0.6840616158440742,
      "grad_norm": 0.47190171480178833,
      "learning_rate": 4.820813810863457e-06,
      "loss": 0.8892,
      "step": 4080
    },
    {
      "epoch": 0.6842292780048203,
      "grad_norm": 0.3707180321216583,
      "learning_rate": 4.816169091178935e-06,
      "loss": 0.8239,
      "step": 4081
    },
    {
      "epoch": 0.6843969401655664,
      "grad_norm": 0.3884958326816559,
      "learning_rate": 4.811525900270334e-06,
      "loss": 0.8317,
      "step": 4082
    },
    {
      "epoch": 0.6845646023263124,
      "grad_norm": 0.4536154866218567,
      "learning_rate": 4.806884239506996e-06,
      "loss": 0.7875,
      "step": 4083
    },
    {
      "epoch": 0.6847322644870586,
      "grad_norm": 0.8998631238937378,
      "learning_rate": 4.802244110257802e-06,
      "loss": 0.9078,
      "step": 4084
    },
    {
      "epoch": 0.6848999266478046,
      "grad_norm": 0.6511355042457581,
      "learning_rate": 4.797605513891179e-06,
      "loss": 0.8234,
      "step": 4085
    },
    {
      "epoch": 0.6850675888085508,
      "grad_norm": 0.4585486054420471,
      "learning_rate": 4.792968451775113e-06,
      "loss": 0.8162,
      "step": 4086
    },
    {
      "epoch": 0.6852352509692968,
      "grad_norm": 0.5549787282943726,
      "learning_rate": 4.788332925277126e-06,
      "loss": 0.8883,
      "step": 4087
    },
    {
      "epoch": 0.685402913130043,
      "grad_norm": 0.4339296519756317,
      "learning_rate": 4.783698935764297e-06,
      "loss": 0.8856,
      "step": 4088
    },
    {
      "epoch": 0.685570575290789,
      "grad_norm": 0.44029632210731506,
      "learning_rate": 4.7790664846032455e-06,
      "loss": 0.8069,
      "step": 4089
    },
    {
      "epoch": 0.6857382374515352,
      "grad_norm": 0.4796565771102905,
      "learning_rate": 4.774435573160132e-06,
      "loss": 0.8275,
      "step": 4090
    },
    {
      "epoch": 0.6859058996122812,
      "grad_norm": 0.6569730639457703,
      "learning_rate": 4.769806202800681e-06,
      "loss": 0.8657,
      "step": 4091
    },
    {
      "epoch": 0.6860735617730274,
      "grad_norm": 0.43796613812446594,
      "learning_rate": 4.76517837489014e-06,
      "loss": 0.8583,
      "step": 4092
    },
    {
      "epoch": 0.6862412239337734,
      "grad_norm": 0.41398483514785767,
      "learning_rate": 4.760552090793322e-06,
      "loss": 0.8822,
      "step": 4093
    },
    {
      "epoch": 0.6864088860945196,
      "grad_norm": 0.5004035234451294,
      "learning_rate": 4.755927351874572e-06,
      "loss": 0.8834,
      "step": 4094
    },
    {
      "epoch": 0.6865765482552656,
      "grad_norm": 0.5711424350738525,
      "learning_rate": 4.7513041594977785e-06,
      "loss": 0.9124,
      "step": 4095
    },
    {
      "epoch": 0.6867442104160117,
      "grad_norm": 0.6565632820129395,
      "learning_rate": 4.746682515026382e-06,
      "loss": 0.8686,
      "step": 4096
    },
    {
      "epoch": 0.6869118725767578,
      "grad_norm": 0.7613095641136169,
      "learning_rate": 4.742062419823367e-06,
      "loss": 0.8704,
      "step": 4097
    },
    {
      "epoch": 0.6870795347375039,
      "grad_norm": 0.4163006842136383,
      "learning_rate": 4.737443875251251e-06,
      "loss": 0.7994,
      "step": 4098
    },
    {
      "epoch": 0.68724719689825,
      "grad_norm": 0.4178316593170166,
      "learning_rate": 4.732826882672099e-06,
      "loss": 0.8177,
      "step": 4099
    },
    {
      "epoch": 0.6874148590589961,
      "grad_norm": 0.6707078814506531,
      "learning_rate": 4.728211443447525e-06,
      "loss": 0.8434,
      "step": 4100
    },
    {
      "epoch": 0.6875825212197422,
      "grad_norm": 0.8010030388832092,
      "learning_rate": 4.7235975589386715e-06,
      "loss": 0.9166,
      "step": 4101
    },
    {
      "epoch": 0.6877501833804883,
      "grad_norm": 0.5210207104682922,
      "learning_rate": 4.718985230506239e-06,
      "loss": 0.9055,
      "step": 4102
    },
    {
      "epoch": 0.6879178455412344,
      "grad_norm": 0.42187198996543884,
      "learning_rate": 4.714374459510456e-06,
      "loss": 0.8316,
      "step": 4103
    },
    {
      "epoch": 0.6880855077019805,
      "grad_norm": 0.43324875831604004,
      "learning_rate": 4.709765247311092e-06,
      "loss": 0.8918,
      "step": 4104
    },
    {
      "epoch": 0.6882531698627266,
      "grad_norm": 0.4681563973426819,
      "learning_rate": 4.7051575952674675e-06,
      "loss": 0.7918,
      "step": 4105
    },
    {
      "epoch": 0.6884208320234727,
      "grad_norm": 0.7020363807678223,
      "learning_rate": 4.700551504738431e-06,
      "loss": 0.8408,
      "step": 4106
    },
    {
      "epoch": 0.6885884941842187,
      "grad_norm": 0.4496000409126282,
      "learning_rate": 4.695946977082381e-06,
      "loss": 0.8581,
      "step": 4107
    },
    {
      "epoch": 0.6887561563449649,
      "grad_norm": 0.4724990129470825,
      "learning_rate": 4.691344013657245e-06,
      "loss": 0.8753,
      "step": 4108
    },
    {
      "epoch": 0.6889238185057109,
      "grad_norm": 0.39639273285865784,
      "learning_rate": 4.686742615820493e-06,
      "loss": 0.897,
      "step": 4109
    },
    {
      "epoch": 0.6890914806664571,
      "grad_norm": 0.46177640557289124,
      "learning_rate": 4.682142784929142e-06,
      "loss": 0.7487,
      "step": 4110
    },
    {
      "epoch": 0.6892591428272031,
      "grad_norm": 0.4963343143463135,
      "learning_rate": 4.6775445223397306e-06,
      "loss": 0.9372,
      "step": 4111
    },
    {
      "epoch": 0.6894268049879493,
      "grad_norm": 0.4017559587955475,
      "learning_rate": 4.672947829408346e-06,
      "loss": 0.8145,
      "step": 4112
    },
    {
      "epoch": 0.6895944671486954,
      "grad_norm": 0.5806766152381897,
      "learning_rate": 4.668352707490611e-06,
      "loss": 0.8044,
      "step": 4113
    },
    {
      "epoch": 0.6897621293094415,
      "grad_norm": 0.4939943552017212,
      "learning_rate": 4.663759157941682e-06,
      "loss": 0.9683,
      "step": 4114
    },
    {
      "epoch": 0.6899297914701876,
      "grad_norm": 0.3326122462749481,
      "learning_rate": 4.6591671821162535e-06,
      "loss": 0.7993,
      "step": 4115
    },
    {
      "epoch": 0.6900974536309337,
      "grad_norm": 0.44053709506988525,
      "learning_rate": 4.654576781368566e-06,
      "loss": 0.81,
      "step": 4116
    },
    {
      "epoch": 0.6902651157916798,
      "grad_norm": 0.6346563100814819,
      "learning_rate": 4.6499879570523684e-06,
      "loss": 0.8571,
      "step": 4117
    },
    {
      "epoch": 0.6904327779524259,
      "grad_norm": 0.4332069754600525,
      "learning_rate": 4.645400710520972e-06,
      "loss": 0.7902,
      "step": 4118
    },
    {
      "epoch": 0.690600440113172,
      "grad_norm": 0.4327547252178192,
      "learning_rate": 4.6408150431272135e-06,
      "loss": 0.8567,
      "step": 4119
    },
    {
      "epoch": 0.690768102273918,
      "grad_norm": 0.5181637406349182,
      "learning_rate": 4.636230956223462e-06,
      "loss": 0.7896,
      "step": 4120
    },
    {
      "epoch": 0.6909357644346642,
      "grad_norm": 0.5253118872642517,
      "learning_rate": 4.6316484511616185e-06,
      "loss": 0.818,
      "step": 4121
    },
    {
      "epoch": 0.6911034265954102,
      "grad_norm": 0.3494114875793457,
      "learning_rate": 4.627067529293119e-06,
      "loss": 0.8766,
      "step": 4122
    },
    {
      "epoch": 0.6912710887561564,
      "grad_norm": 0.45745769143104553,
      "learning_rate": 4.622488191968938e-06,
      "loss": 0.8666,
      "step": 4123
    },
    {
      "epoch": 0.6914387509169024,
      "grad_norm": 0.5724239349365234,
      "learning_rate": 4.61791044053958e-06,
      "loss": 0.7707,
      "step": 4124
    },
    {
      "epoch": 0.6916064130776486,
      "grad_norm": 0.41080933809280396,
      "learning_rate": 4.613334276355079e-06,
      "loss": 0.8641,
      "step": 4125
    },
    {
      "epoch": 0.6917740752383946,
      "grad_norm": 0.4222384989261627,
      "learning_rate": 4.608759700764998e-06,
      "loss": 0.7909,
      "step": 4126
    },
    {
      "epoch": 0.6919417373991408,
      "grad_norm": 0.4619828462600708,
      "learning_rate": 4.604186715118445e-06,
      "loss": 0.9484,
      "step": 4127
    },
    {
      "epoch": 0.6921093995598868,
      "grad_norm": 0.36951783299446106,
      "learning_rate": 4.599615320764041e-06,
      "loss": 0.7432,
      "step": 4128
    },
    {
      "epoch": 0.692277061720633,
      "grad_norm": 0.571994960308075,
      "learning_rate": 4.595045519049954e-06,
      "loss": 0.8783,
      "step": 4129
    },
    {
      "epoch": 0.692444723881379,
      "grad_norm": 0.550298273563385,
      "learning_rate": 4.590477311323872e-06,
      "loss": 0.9082,
      "step": 4130
    },
    {
      "epoch": 0.6926123860421252,
      "grad_norm": 0.39157381653785706,
      "learning_rate": 4.585910698933013e-06,
      "loss": 0.8105,
      "step": 4131
    },
    {
      "epoch": 0.6927800482028712,
      "grad_norm": 0.7182436585426331,
      "learning_rate": 4.581345683224133e-06,
      "loss": 0.8682,
      "step": 4132
    },
    {
      "epoch": 0.6929477103636174,
      "grad_norm": 0.5301082730293274,
      "learning_rate": 4.576782265543507e-06,
      "loss": 0.7852,
      "step": 4133
    },
    {
      "epoch": 0.6931153725243634,
      "grad_norm": 0.6268473863601685,
      "learning_rate": 4.572220447236948e-06,
      "loss": 0.9147,
      "step": 4134
    },
    {
      "epoch": 0.6932830346851095,
      "grad_norm": 0.41989865899086,
      "learning_rate": 4.567660229649791e-06,
      "loss": 0.832,
      "step": 4135
    },
    {
      "epoch": 0.6934506968458556,
      "grad_norm": 0.4412514567375183,
      "learning_rate": 4.563101614126895e-06,
      "loss": 0.8326,
      "step": 4136
    },
    {
      "epoch": 0.6936183590066017,
      "grad_norm": 0.6761359572410583,
      "learning_rate": 4.5585446020126634e-06,
      "loss": 0.8856,
      "step": 4137
    },
    {
      "epoch": 0.6937860211673478,
      "grad_norm": 0.5193226933479309,
      "learning_rate": 4.553989194651003e-06,
      "loss": 0.8256,
      "step": 4138
    },
    {
      "epoch": 0.6939536833280939,
      "grad_norm": 0.5478662252426147,
      "learning_rate": 4.549435393385371e-06,
      "loss": 0.8706,
      "step": 4139
    },
    {
      "epoch": 0.69412134548884,
      "grad_norm": 0.559935450553894,
      "learning_rate": 4.544883199558735e-06,
      "loss": 0.8608,
      "step": 4140
    },
    {
      "epoch": 0.6942890076495861,
      "grad_norm": 0.5725769996643066,
      "learning_rate": 4.540332614513589e-06,
      "loss": 0.7859,
      "step": 4141
    },
    {
      "epoch": 0.6944566698103322,
      "grad_norm": 0.5817712545394897,
      "learning_rate": 4.535783639591961e-06,
      "loss": 0.9089,
      "step": 4142
    },
    {
      "epoch": 0.6946243319710783,
      "grad_norm": 1.1822705268859863,
      "learning_rate": 4.5312362761354065e-06,
      "loss": 0.7498,
      "step": 4143
    },
    {
      "epoch": 0.6947919941318244,
      "grad_norm": 0.6353855133056641,
      "learning_rate": 4.526690525484987e-06,
      "loss": 0.8337,
      "step": 4144
    },
    {
      "epoch": 0.6949596562925705,
      "grad_norm": 0.3704085946083069,
      "learning_rate": 4.522146388981306e-06,
      "loss": 0.8336,
      "step": 4145
    },
    {
      "epoch": 0.6951273184533165,
      "grad_norm": 0.5411508679389954,
      "learning_rate": 4.51760386796449e-06,
      "loss": 0.882,
      "step": 4146
    },
    {
      "epoch": 0.6952949806140627,
      "grad_norm": 0.35425445437431335,
      "learning_rate": 4.513062963774176e-06,
      "loss": 0.8379,
      "step": 4147
    },
    {
      "epoch": 0.6954626427748087,
      "grad_norm": 0.44852039217948914,
      "learning_rate": 4.508523677749542e-06,
      "loss": 0.8871,
      "step": 4148
    },
    {
      "epoch": 0.6956303049355549,
      "grad_norm": 0.5101825594902039,
      "learning_rate": 4.5039860112292755e-06,
      "loss": 0.857,
      "step": 4149
    },
    {
      "epoch": 0.6957979670963009,
      "grad_norm": 0.40634283423423767,
      "learning_rate": 4.4994499655515865e-06,
      "loss": 0.8711,
      "step": 4150
    },
    {
      "epoch": 0.6959656292570471,
      "grad_norm": 0.7540550827980042,
      "learning_rate": 4.494915542054219e-06,
      "loss": 0.908,
      "step": 4151
    },
    {
      "epoch": 0.6961332914177931,
      "grad_norm": 1.2536613941192627,
      "learning_rate": 4.490382742074428e-06,
      "loss": 0.7994,
      "step": 4152
    },
    {
      "epoch": 0.6963009535785393,
      "grad_norm": 0.3635319471359253,
      "learning_rate": 4.485851566948985e-06,
      "loss": 0.7661,
      "step": 4153
    },
    {
      "epoch": 0.6964686157392853,
      "grad_norm": 0.37484169006347656,
      "learning_rate": 4.4813220180141995e-06,
      "loss": 0.8714,
      "step": 4154
    },
    {
      "epoch": 0.6966362779000315,
      "grad_norm": 0.3904739320278168,
      "learning_rate": 4.476794096605884e-06,
      "loss": 0.8176,
      "step": 4155
    },
    {
      "epoch": 0.6968039400607775,
      "grad_norm": 0.4750972092151642,
      "learning_rate": 4.472267804059387e-06,
      "loss": 0.925,
      "step": 4156
    },
    {
      "epoch": 0.6969716022215237,
      "grad_norm": 0.6135466694831848,
      "learning_rate": 4.467743141709563e-06,
      "loss": 0.781,
      "step": 4157
    },
    {
      "epoch": 0.6971392643822697,
      "grad_norm": 0.38343942165374756,
      "learning_rate": 4.463220110890787e-06,
      "loss": 0.8202,
      "step": 4158
    },
    {
      "epoch": 0.6973069265430158,
      "grad_norm": 1.104854702949524,
      "learning_rate": 4.458698712936966e-06,
      "loss": 0.8498,
      "step": 4159
    },
    {
      "epoch": 0.6974745887037619,
      "grad_norm": 0.3883248269557953,
      "learning_rate": 4.454178949181507e-06,
      "loss": 0.9099,
      "step": 4160
    },
    {
      "epoch": 0.697642250864508,
      "grad_norm": 1.0206170082092285,
      "learning_rate": 4.4496608209573485e-06,
      "loss": 0.8711,
      "step": 4161
    },
    {
      "epoch": 0.6978099130252541,
      "grad_norm": 0.5924220085144043,
      "learning_rate": 4.445144329596951e-06,
      "loss": 0.828,
      "step": 4162
    },
    {
      "epoch": 0.6979775751860002,
      "grad_norm": 0.349104642868042,
      "learning_rate": 4.440629476432268e-06,
      "loss": 0.7629,
      "step": 4163
    },
    {
      "epoch": 0.6981452373467463,
      "grad_norm": 0.37381574511528015,
      "learning_rate": 4.436116262794793e-06,
      "loss": 0.9125,
      "step": 4164
    },
    {
      "epoch": 0.6983128995074924,
      "grad_norm": 0.40002623200416565,
      "learning_rate": 4.431604690015532e-06,
      "loss": 0.8435,
      "step": 4165
    },
    {
      "epoch": 0.6984805616682385,
      "grad_norm": 0.37091362476348877,
      "learning_rate": 4.427094759425e-06,
      "loss": 0.7839,
      "step": 4166
    },
    {
      "epoch": 0.6986482238289846,
      "grad_norm": 0.6612892150878906,
      "learning_rate": 4.422586472353229e-06,
      "loss": 0.8337,
      "step": 4167
    },
    {
      "epoch": 0.6988158859897307,
      "grad_norm": 0.4454594850540161,
      "learning_rate": 4.418079830129773e-06,
      "loss": 0.8311,
      "step": 4168
    },
    {
      "epoch": 0.6989835481504768,
      "grad_norm": 0.6289452314376831,
      "learning_rate": 4.413574834083694e-06,
      "loss": 0.7649,
      "step": 4169
    },
    {
      "epoch": 0.6991512103112228,
      "grad_norm": 0.39444994926452637,
      "learning_rate": 4.4090714855435746e-06,
      "loss": 0.7753,
      "step": 4170
    },
    {
      "epoch": 0.699318872471969,
      "grad_norm": 0.7020822763442993,
      "learning_rate": 4.404569785837505e-06,
      "loss": 0.9139,
      "step": 4171
    },
    {
      "epoch": 0.699486534632715,
      "grad_norm": 0.47531023621559143,
      "learning_rate": 4.40006973629309e-06,
      "loss": 0.8804,
      "step": 4172
    },
    {
      "epoch": 0.6996541967934612,
      "grad_norm": 0.46519336104393005,
      "learning_rate": 4.395571338237455e-06,
      "loss": 0.837,
      "step": 4173
    },
    {
      "epoch": 0.6998218589542072,
      "grad_norm": 0.5111847519874573,
      "learning_rate": 4.391074592997228e-06,
      "loss": 0.8368,
      "step": 4174
    },
    {
      "epoch": 0.6999895211149534,
      "grad_norm": 0.40244993567466736,
      "learning_rate": 4.386579501898563e-06,
      "loss": 0.843,
      "step": 4175
    },
    {
      "epoch": 0.7001571832756994,
      "grad_norm": 0.5221668481826782,
      "learning_rate": 4.38208606626711e-06,
      "loss": 0.8128,
      "step": 4176
    },
    {
      "epoch": 0.7003248454364456,
      "grad_norm": 0.34330761432647705,
      "learning_rate": 4.37759428742804e-06,
      "loss": 0.8586,
      "step": 4177
    },
    {
      "epoch": 0.7004925075971916,
      "grad_norm": 0.3652341365814209,
      "learning_rate": 4.373104166706039e-06,
      "loss": 0.8355,
      "step": 4178
    },
    {
      "epoch": 0.7006601697579378,
      "grad_norm": 0.675369143486023,
      "learning_rate": 4.368615705425292e-06,
      "loss": 0.9091,
      "step": 4179
    },
    {
      "epoch": 0.7008278319186838,
      "grad_norm": 0.40616875886917114,
      "learning_rate": 4.364128904909509e-06,
      "loss": 0.7958,
      "step": 4180
    },
    {
      "epoch": 0.70099549407943,
      "grad_norm": 0.5879313945770264,
      "learning_rate": 4.359643766481901e-06,
      "loss": 0.8831,
      "step": 4181
    },
    {
      "epoch": 0.701163156240176,
      "grad_norm": 0.49311167001724243,
      "learning_rate": 4.355160291465186e-06,
      "loss": 0.8027,
      "step": 4182
    },
    {
      "epoch": 0.7013308184009222,
      "grad_norm": 0.38633251190185547,
      "learning_rate": 4.3506784811815995e-06,
      "loss": 0.8575,
      "step": 4183
    },
    {
      "epoch": 0.7014984805616682,
      "grad_norm": 0.4957720637321472,
      "learning_rate": 4.346198336952892e-06,
      "loss": 0.801,
      "step": 4184
    },
    {
      "epoch": 0.7016661427224143,
      "grad_norm": 0.368158221244812,
      "learning_rate": 4.341719860100296e-06,
      "loss": 0.9135,
      "step": 4185
    },
    {
      "epoch": 0.7018338048831604,
      "grad_norm": 0.7882813215255737,
      "learning_rate": 4.337243051944583e-06,
      "loss": 0.7851,
      "step": 4186
    },
    {
      "epoch": 0.7020014670439065,
      "grad_norm": 0.38310936093330383,
      "learning_rate": 4.3327679138060105e-06,
      "loss": 0.85,
      "step": 4187
    },
    {
      "epoch": 0.7021691292046526,
      "grad_norm": 0.42763540148735046,
      "learning_rate": 4.3282944470043576e-06,
      "loss": 0.8013,
      "step": 4188
    },
    {
      "epoch": 0.7023367913653987,
      "grad_norm": 0.359748899936676,
      "learning_rate": 4.323822652858911e-06,
      "loss": 0.7907,
      "step": 4189
    },
    {
      "epoch": 0.7025044535261448,
      "grad_norm": 0.5136620998382568,
      "learning_rate": 4.319352532688444e-06,
      "loss": 0.8221,
      "step": 4190
    },
    {
      "epoch": 0.7026721156868909,
      "grad_norm": 0.5542488098144531,
      "learning_rate": 4.314884087811256e-06,
      "loss": 0.8735,
      "step": 4191
    },
    {
      "epoch": 0.702839777847637,
      "grad_norm": 0.5811720490455627,
      "learning_rate": 4.310417319545154e-06,
      "loss": 0.7878,
      "step": 4192
    },
    {
      "epoch": 0.7030074400083831,
      "grad_norm": 0.5262954831123352,
      "learning_rate": 4.305952229207436e-06,
      "loss": 0.8474,
      "step": 4193
    },
    {
      "epoch": 0.7031751021691292,
      "grad_norm": 0.459160178899765,
      "learning_rate": 4.30148881811491e-06,
      "loss": 0.9203,
      "step": 4194
    },
    {
      "epoch": 0.7033427643298753,
      "grad_norm": 0.5392619967460632,
      "learning_rate": 4.297027087583901e-06,
      "loss": 0.8551,
      "step": 4195
    },
    {
      "epoch": 0.7035104264906213,
      "grad_norm": 0.9503580927848816,
      "learning_rate": 4.292567038930218e-06,
      "loss": 0.8948,
      "step": 4196
    },
    {
      "epoch": 0.7036780886513675,
      "grad_norm": 0.4359428584575653,
      "learning_rate": 4.288108673469192e-06,
      "loss": 0.8506,
      "step": 4197
    },
    {
      "epoch": 0.7038457508121136,
      "grad_norm": 0.6888027191162109,
      "learning_rate": 4.283651992515649e-06,
      "loss": 0.8487,
      "step": 4198
    },
    {
      "epoch": 0.7040134129728597,
      "grad_norm": 0.41162729263305664,
      "learning_rate": 4.279196997383912e-06,
      "loss": 0.8153,
      "step": 4199
    },
    {
      "epoch": 0.7041810751336058,
      "grad_norm": 0.6298072934150696,
      "learning_rate": 4.274743689387826e-06,
      "loss": 0.8535,
      "step": 4200
    },
    {
      "epoch": 0.7043487372943519,
      "grad_norm": 0.3720015585422516,
      "learning_rate": 4.270292069840716e-06,
      "loss": 0.9058,
      "step": 4201
    },
    {
      "epoch": 0.704516399455098,
      "grad_norm": 0.4745865762233734,
      "learning_rate": 4.265842140055428e-06,
      "loss": 0.8507,
      "step": 4202
    },
    {
      "epoch": 0.7046840616158441,
      "grad_norm": 0.6363827586174011,
      "learning_rate": 4.261393901344299e-06,
      "loss": 0.8739,
      "step": 4203
    },
    {
      "epoch": 0.7048517237765902,
      "grad_norm": 0.695592999458313,
      "learning_rate": 4.256947355019163e-06,
      "loss": 0.9261,
      "step": 4204
    },
    {
      "epoch": 0.7050193859373363,
      "grad_norm": 0.4004266560077667,
      "learning_rate": 4.252502502391373e-06,
      "loss": 0.8323,
      "step": 4205
    },
    {
      "epoch": 0.7051870480980824,
      "grad_norm": 0.35616323351860046,
      "learning_rate": 4.24805934477176e-06,
      "loss": 0.8492,
      "step": 4206
    },
    {
      "epoch": 0.7053547102588285,
      "grad_norm": 0.5288553237915039,
      "learning_rate": 4.243617883470675e-06,
      "loss": 0.8501,
      "step": 4207
    },
    {
      "epoch": 0.7055223724195746,
      "grad_norm": 1.0049248933792114,
      "learning_rate": 4.2391781197979575e-06,
      "loss": 0.7618,
      "step": 4208
    },
    {
      "epoch": 0.7056900345803206,
      "grad_norm": 1.5840026140213013,
      "learning_rate": 4.2347400550629435e-06,
      "loss": 0.8275,
      "step": 4209
    },
    {
      "epoch": 0.7058576967410668,
      "grad_norm": 0.4762451648712158,
      "learning_rate": 4.230303690574478e-06,
      "loss": 0.8096,
      "step": 4210
    },
    {
      "epoch": 0.7060253589018128,
      "grad_norm": 0.6156831383705139,
      "learning_rate": 4.225869027640902e-06,
      "loss": 0.9071,
      "step": 4211
    },
    {
      "epoch": 0.706193021062559,
      "grad_norm": 0.5486253499984741,
      "learning_rate": 4.22143606757005e-06,
      "loss": 0.7992,
      "step": 4212
    },
    {
      "epoch": 0.706360683223305,
      "grad_norm": 0.6755399703979492,
      "learning_rate": 4.217004811669253e-06,
      "loss": 1.0019,
      "step": 4213
    },
    {
      "epoch": 0.7065283453840512,
      "grad_norm": 0.47275254130363464,
      "learning_rate": 4.212575261245351e-06,
      "loss": 0.8909,
      "step": 4214
    },
    {
      "epoch": 0.7066960075447972,
      "grad_norm": 0.4999307692050934,
      "learning_rate": 4.208147417604665e-06,
      "loss": 0.9178,
      "step": 4215
    },
    {
      "epoch": 0.7068636697055434,
      "grad_norm": 0.3870341181755066,
      "learning_rate": 4.2037212820530285e-06,
      "loss": 0.8844,
      "step": 4216
    },
    {
      "epoch": 0.7070313318662894,
      "grad_norm": 0.45955073833465576,
      "learning_rate": 4.19929685589576e-06,
      "loss": 0.8724,
      "step": 4217
    },
    {
      "epoch": 0.7071989940270356,
      "grad_norm": 0.4071888327598572,
      "learning_rate": 4.194874140437675e-06,
      "loss": 0.8095,
      "step": 4218
    },
    {
      "epoch": 0.7073666561877816,
      "grad_norm": 0.4482811391353607,
      "learning_rate": 4.190453136983091e-06,
      "loss": 0.8302,
      "step": 4219
    },
    {
      "epoch": 0.7075343183485278,
      "grad_norm": 0.46475750207901,
      "learning_rate": 4.186033846835814e-06,
      "loss": 0.8576,
      "step": 4220
    },
    {
      "epoch": 0.7077019805092738,
      "grad_norm": 0.35893514752388,
      "learning_rate": 4.181616271299153e-06,
      "loss": 0.8915,
      "step": 4221
    },
    {
      "epoch": 0.70786964267002,
      "grad_norm": 0.3959294557571411,
      "learning_rate": 4.1772004116759004e-06,
      "loss": 0.8427,
      "step": 4222
    },
    {
      "epoch": 0.708037304830766,
      "grad_norm": 0.45966199040412903,
      "learning_rate": 4.172786269268345e-06,
      "loss": 0.8358,
      "step": 4223
    },
    {
      "epoch": 0.7082049669915121,
      "grad_norm": 0.39270827174186707,
      "learning_rate": 4.1683738453782795e-06,
      "loss": 0.8227,
      "step": 4224
    },
    {
      "epoch": 0.7083726291522582,
      "grad_norm": 0.3983052372932434,
      "learning_rate": 4.16396314130698e-06,
      "loss": 0.8065,
      "step": 4225
    },
    {
      "epoch": 0.7085402913130043,
      "grad_norm": 0.5825944542884827,
      "learning_rate": 4.1595541583552125e-06,
      "loss": 0.8186,
      "step": 4226
    },
    {
      "epoch": 0.7087079534737504,
      "grad_norm": 0.390227735042572,
      "learning_rate": 4.155146897823249e-06,
      "loss": 0.8417,
      "step": 4227
    },
    {
      "epoch": 0.7088756156344965,
      "grad_norm": 0.4769403636455536,
      "learning_rate": 4.150741361010837e-06,
      "loss": 0.8028,
      "step": 4228
    },
    {
      "epoch": 0.7090432777952426,
      "grad_norm": 0.6111634969711304,
      "learning_rate": 4.1463375492172285e-06,
      "loss": 0.8451,
      "step": 4229
    },
    {
      "epoch": 0.7092109399559887,
      "grad_norm": 0.5125744938850403,
      "learning_rate": 4.141935463741168e-06,
      "loss": 0.892,
      "step": 4230
    },
    {
      "epoch": 0.7093786021167348,
      "grad_norm": 0.37312835454940796,
      "learning_rate": 4.137535105880872e-06,
      "loss": 0.8185,
      "step": 4231
    },
    {
      "epoch": 0.7095462642774809,
      "grad_norm": 0.4071962237358093,
      "learning_rate": 4.1331364769340665e-06,
      "loss": 0.8193,
      "step": 4232
    },
    {
      "epoch": 0.709713926438227,
      "grad_norm": 0.3986750543117523,
      "learning_rate": 4.128739578197965e-06,
      "loss": 0.7764,
      "step": 4233
    },
    {
      "epoch": 0.7098815885989731,
      "grad_norm": 0.3785371482372284,
      "learning_rate": 4.124344410969263e-06,
      "loss": 0.83,
      "step": 4234
    },
    {
      "epoch": 0.7100492507597191,
      "grad_norm": 0.41927018761634827,
      "learning_rate": 4.119950976544158e-06,
      "loss": 0.8113,
      "step": 4235
    },
    {
      "epoch": 0.7102169129204653,
      "grad_norm": 0.3834112584590912,
      "learning_rate": 4.1155592762183135e-06,
      "loss": 0.8667,
      "step": 4236
    },
    {
      "epoch": 0.7103845750812113,
      "grad_norm": 0.5270161032676697,
      "learning_rate": 4.111169311286905e-06,
      "loss": 0.8546,
      "step": 4237
    },
    {
      "epoch": 0.7105522372419575,
      "grad_norm": 0.3906877934932709,
      "learning_rate": 4.106781083044591e-06,
      "loss": 0.829,
      "step": 4238
    },
    {
      "epoch": 0.7107198994027035,
      "grad_norm": 0.457753986120224,
      "learning_rate": 4.102394592785509e-06,
      "loss": 0.9014,
      "step": 4239
    },
    {
      "epoch": 0.7108875615634497,
      "grad_norm": 0.6644868850708008,
      "learning_rate": 4.098009841803288e-06,
      "loss": 0.8609,
      "step": 4240
    },
    {
      "epoch": 0.7110552237241957,
      "grad_norm": 0.3552396893501282,
      "learning_rate": 4.093626831391051e-06,
      "loss": 0.8647,
      "step": 4241
    },
    {
      "epoch": 0.7112228858849419,
      "grad_norm": 0.4755293130874634,
      "learning_rate": 4.089245562841396e-06,
      "loss": 0.8338,
      "step": 4242
    },
    {
      "epoch": 0.7113905480456879,
      "grad_norm": 0.7194167375564575,
      "learning_rate": 4.0848660374464185e-06,
      "loss": 0.8174,
      "step": 4243
    },
    {
      "epoch": 0.7115582102064341,
      "grad_norm": 0.5331094264984131,
      "learning_rate": 4.080488256497694e-06,
      "loss": 0.9453,
      "step": 4244
    },
    {
      "epoch": 0.7117258723671801,
      "grad_norm": 0.4913577139377594,
      "learning_rate": 4.076112221286278e-06,
      "loss": 0.9477,
      "step": 4245
    },
    {
      "epoch": 0.7118935345279263,
      "grad_norm": 0.41921189427375793,
      "learning_rate": 4.071737933102726e-06,
      "loss": 0.9137,
      "step": 4246
    },
    {
      "epoch": 0.7120611966886723,
      "grad_norm": 0.3868524432182312,
      "learning_rate": 4.067365393237064e-06,
      "loss": 0.7741,
      "step": 4247
    },
    {
      "epoch": 0.7122288588494184,
      "grad_norm": 0.4459342658519745,
      "learning_rate": 4.062994602978812e-06,
      "loss": 0.8389,
      "step": 4248
    },
    {
      "epoch": 0.7123965210101645,
      "grad_norm": 0.6608433127403259,
      "learning_rate": 4.05862556361697e-06,
      "loss": 0.8265,
      "step": 4249
    },
    {
      "epoch": 0.7125641831709106,
      "grad_norm": 0.8078179955482483,
      "learning_rate": 4.054258276440016e-06,
      "loss": 0.893,
      "step": 4250
    },
    {
      "epoch": 0.7127318453316567,
      "grad_norm": 0.3659352660179138,
      "learning_rate": 4.049892742735924e-06,
      "loss": 0.7951,
      "step": 4251
    },
    {
      "epoch": 0.7128995074924028,
      "grad_norm": 0.6276419162750244,
      "learning_rate": 4.045528963792139e-06,
      "loss": 0.8599,
      "step": 4252
    },
    {
      "epoch": 0.7130671696531489,
      "grad_norm": 0.4345807731151581,
      "learning_rate": 4.041166940895599e-06,
      "loss": 0.863,
      "step": 4253
    },
    {
      "epoch": 0.713234831813895,
      "grad_norm": 0.5194419622421265,
      "learning_rate": 4.036806675332715e-06,
      "loss": 0.8563,
      "step": 4254
    },
    {
      "epoch": 0.7134024939746411,
      "grad_norm": 0.44164738059043884,
      "learning_rate": 4.03244816838938e-06,
      "loss": 0.8806,
      "step": 4255
    },
    {
      "epoch": 0.7135701561353872,
      "grad_norm": 0.40137559175491333,
      "learning_rate": 4.028091421350975e-06,
      "loss": 0.8699,
      "step": 4256
    },
    {
      "epoch": 0.7137378182961333,
      "grad_norm": 0.4443762004375458,
      "learning_rate": 4.023736435502367e-06,
      "loss": 0.7674,
      "step": 4257
    },
    {
      "epoch": 0.7139054804568794,
      "grad_norm": 0.5507704019546509,
      "learning_rate": 4.019383212127879e-06,
      "loss": 0.811,
      "step": 4258
    },
    {
      "epoch": 0.7140731426176254,
      "grad_norm": 0.44957438111305237,
      "learning_rate": 4.015031752511338e-06,
      "loss": 0.8692,
      "step": 4259
    },
    {
      "epoch": 0.7142408047783716,
      "grad_norm": 0.41376200318336487,
      "learning_rate": 4.0106820579360485e-06,
      "loss": 0.81,
      "step": 4260
    },
    {
      "epoch": 0.7144084669391176,
      "grad_norm": 0.4778636693954468,
      "learning_rate": 4.006334129684781e-06,
      "loss": 0.7534,
      "step": 4261
    },
    {
      "epoch": 0.7145761290998638,
      "grad_norm": 0.42495572566986084,
      "learning_rate": 4.001987969039801e-06,
      "loss": 0.8306,
      "step": 4262
    },
    {
      "epoch": 0.7147437912606098,
      "grad_norm": 0.4066849946975708,
      "learning_rate": 3.99764357728284e-06,
      "loss": 0.8782,
      "step": 4263
    },
    {
      "epoch": 0.714911453421356,
      "grad_norm": 0.5115940570831299,
      "learning_rate": 3.993300955695112e-06,
      "loss": 0.7987,
      "step": 4264
    },
    {
      "epoch": 0.715079115582102,
      "grad_norm": 0.3866828680038452,
      "learning_rate": 3.9889601055573145e-06,
      "loss": 0.823,
      "step": 4265
    },
    {
      "epoch": 0.7152467777428482,
      "grad_norm": 0.3616901636123657,
      "learning_rate": 3.984621028149617e-06,
      "loss": 0.9494,
      "step": 4266
    },
    {
      "epoch": 0.7154144399035942,
      "grad_norm": 0.668507993221283,
      "learning_rate": 3.98028372475166e-06,
      "loss": 0.8685,
      "step": 4267
    },
    {
      "epoch": 0.7155821020643404,
      "grad_norm": 0.5150197744369507,
      "learning_rate": 3.975948196642578e-06,
      "loss": 0.8761,
      "step": 4268
    },
    {
      "epoch": 0.7157497642250864,
      "grad_norm": 0.3876942992210388,
      "learning_rate": 3.9716144451009655e-06,
      "loss": 0.8787,
      "step": 4269
    },
    {
      "epoch": 0.7159174263858326,
      "grad_norm": 0.5321659445762634,
      "learning_rate": 3.967282471404903e-06,
      "loss": 0.8081,
      "step": 4270
    },
    {
      "epoch": 0.7160850885465786,
      "grad_norm": 0.657336413860321,
      "learning_rate": 3.962952276831943e-06,
      "loss": 0.691,
      "step": 4271
    },
    {
      "epoch": 0.7162527507073247,
      "grad_norm": 0.6628797650337219,
      "learning_rate": 3.958623862659108e-06,
      "loss": 0.8417,
      "step": 4272
    },
    {
      "epoch": 0.7164204128680708,
      "grad_norm": 0.6671704649925232,
      "learning_rate": 3.954297230162908e-06,
      "loss": 0.7801,
      "step": 4273
    },
    {
      "epoch": 0.7165880750288169,
      "grad_norm": 0.4789895713329315,
      "learning_rate": 3.9499723806193145e-06,
      "loss": 0.8519,
      "step": 4274
    },
    {
      "epoch": 0.716755737189563,
      "grad_norm": 0.2626591920852661,
      "learning_rate": 3.9456493153037835e-06,
      "loss": 0.8363,
      "step": 4275
    },
    {
      "epoch": 0.7169233993503091,
      "grad_norm": 0.5861395001411438,
      "learning_rate": 3.941328035491245e-06,
      "loss": 0.9238,
      "step": 4276
    },
    {
      "epoch": 0.7170910615110552,
      "grad_norm": 0.40450188517570496,
      "learning_rate": 3.937008542456085e-06,
      "loss": 0.8712,
      "step": 4277
    },
    {
      "epoch": 0.7172587236718013,
      "grad_norm": 0.41760146617889404,
      "learning_rate": 3.932690837472184e-06,
      "loss": 0.9621,
      "step": 4278
    },
    {
      "epoch": 0.7174263858325474,
      "grad_norm": 0.42822742462158203,
      "learning_rate": 3.9283749218128885e-06,
      "loss": 0.8108,
      "step": 4279
    },
    {
      "epoch": 0.7175940479932935,
      "grad_norm": 0.5657880902290344,
      "learning_rate": 3.924060796751012e-06,
      "loss": 0.7965,
      "step": 4280
    },
    {
      "epoch": 0.7177617101540396,
      "grad_norm": 0.5241937637329102,
      "learning_rate": 3.91974846355884e-06,
      "loss": 0.8407,
      "step": 4281
    },
    {
      "epoch": 0.7179293723147857,
      "grad_norm": 0.5473652482032776,
      "learning_rate": 3.91543792350814e-06,
      "loss": 0.9083,
      "step": 4282
    },
    {
      "epoch": 0.7180970344755317,
      "grad_norm": 0.5616039633750916,
      "learning_rate": 3.911129177870139e-06,
      "loss": 0.8463,
      "step": 4283
    },
    {
      "epoch": 0.7182646966362779,
      "grad_norm": 1.1539825201034546,
      "learning_rate": 3.9068222279155425e-06,
      "loss": 0.8511,
      "step": 4284
    },
    {
      "epoch": 0.718432358797024,
      "grad_norm": 0.4508279860019684,
      "learning_rate": 3.902517074914522e-06,
      "loss": 0.8484,
      "step": 4285
    },
    {
      "epoch": 0.7186000209577701,
      "grad_norm": 0.4086759388446808,
      "learning_rate": 3.898213720136716e-06,
      "loss": 0.8517,
      "step": 4286
    },
    {
      "epoch": 0.7187676831185162,
      "grad_norm": 0.6924920678138733,
      "learning_rate": 3.893912164851246e-06,
      "loss": 0.8152,
      "step": 4287
    },
    {
      "epoch": 0.7189353452792623,
      "grad_norm": 0.43707549571990967,
      "learning_rate": 3.889612410326685e-06,
      "loss": 0.8664,
      "step": 4288
    },
    {
      "epoch": 0.7191030074400084,
      "grad_norm": 0.3894766569137573,
      "learning_rate": 3.885314457831091e-06,
      "loss": 0.8669,
      "step": 4289
    },
    {
      "epoch": 0.7192706696007545,
      "grad_norm": 0.38495904207229614,
      "learning_rate": 3.881018308631981e-06,
      "loss": 0.8742,
      "step": 4290
    },
    {
      "epoch": 0.7194383317615006,
      "grad_norm": 0.3835095167160034,
      "learning_rate": 3.876723963996338e-06,
      "loss": 0.9261,
      "step": 4291
    },
    {
      "epoch": 0.7196059939222467,
      "grad_norm": 0.6569795608520508,
      "learning_rate": 3.872431425190626e-06,
      "loss": 0.9282,
      "step": 4292
    },
    {
      "epoch": 0.7197736560829928,
      "grad_norm": 0.37913012504577637,
      "learning_rate": 3.8681406934807585e-06,
      "loss": 0.7464,
      "step": 4293
    },
    {
      "epoch": 0.7199413182437389,
      "grad_norm": 0.44137638807296753,
      "learning_rate": 3.863851770132134e-06,
      "loss": 0.8153,
      "step": 4294
    },
    {
      "epoch": 0.720108980404485,
      "grad_norm": 0.45716938376426697,
      "learning_rate": 3.859564656409606e-06,
      "loss": 0.899,
      "step": 4295
    },
    {
      "epoch": 0.720276642565231,
      "grad_norm": 0.5398461818695068,
      "learning_rate": 3.855279353577493e-06,
      "loss": 0.9182,
      "step": 4296
    },
    {
      "epoch": 0.7204443047259772,
      "grad_norm": 0.45532530546188354,
      "learning_rate": 3.850995862899587e-06,
      "loss": 0.8556,
      "step": 4297
    },
    {
      "epoch": 0.7206119668867232,
      "grad_norm": 0.3884469270706177,
      "learning_rate": 3.84671418563915e-06,
      "loss": 0.9219,
      "step": 4298
    },
    {
      "epoch": 0.7207796290474694,
      "grad_norm": 0.5447671413421631,
      "learning_rate": 3.842434323058888e-06,
      "loss": 0.8337,
      "step": 4299
    },
    {
      "epoch": 0.7209472912082154,
      "grad_norm": 0.3661268949508667,
      "learning_rate": 3.838156276420994e-06,
      "loss": 0.8208,
      "step": 4300
    },
    {
      "epoch": 0.7211149533689616,
      "grad_norm": 0.4238070249557495,
      "learning_rate": 3.833880046987114e-06,
      "loss": 0.797,
      "step": 4301
    },
    {
      "epoch": 0.7212826155297076,
      "grad_norm": 0.6158653497695923,
      "learning_rate": 3.8296056360183605e-06,
      "loss": 0.8568,
      "step": 4302
    },
    {
      "epoch": 0.7214502776904538,
      "grad_norm": 0.5459095239639282,
      "learning_rate": 3.82533304477532e-06,
      "loss": 0.8678,
      "step": 4303
    },
    {
      "epoch": 0.7216179398511998,
      "grad_norm": 0.46029114723205566,
      "learning_rate": 3.821062274518017e-06,
      "loss": 0.7086,
      "step": 4304
    },
    {
      "epoch": 0.721785602011946,
      "grad_norm": 0.6634854674339294,
      "learning_rate": 3.816793326505961e-06,
      "loss": 0.8317,
      "step": 4305
    },
    {
      "epoch": 0.721953264172692,
      "grad_norm": 0.5513525605201721,
      "learning_rate": 3.8125262019981224e-06,
      "loss": 0.9066,
      "step": 4306
    },
    {
      "epoch": 0.7221209263334382,
      "grad_norm": 0.9436096549034119,
      "learning_rate": 3.8082609022529226e-06,
      "loss": 0.8321,
      "step": 4307
    },
    {
      "epoch": 0.7222885884941842,
      "grad_norm": 0.3836483359336853,
      "learning_rate": 3.8039974285282567e-06,
      "loss": 0.8915,
      "step": 4308
    },
    {
      "epoch": 0.7224562506549304,
      "grad_norm": 0.6042637825012207,
      "learning_rate": 3.799735782081473e-06,
      "loss": 0.8074,
      "step": 4309
    },
    {
      "epoch": 0.7226239128156764,
      "grad_norm": 0.795918345451355,
      "learning_rate": 3.7954759641693795e-06,
      "loss": 0.8419,
      "step": 4310
    },
    {
      "epoch": 0.7227915749764225,
      "grad_norm": 0.49923935532569885,
      "learning_rate": 3.7912179760482582e-06,
      "loss": 0.9364,
      "step": 4311
    },
    {
      "epoch": 0.7229592371371686,
      "grad_norm": 1.1601747274398804,
      "learning_rate": 3.7869618189738377e-06,
      "loss": 0.9209,
      "step": 4312
    },
    {
      "epoch": 0.7231268992979147,
      "grad_norm": 0.46306419372558594,
      "learning_rate": 3.7827074942013085e-06,
      "loss": 0.8514,
      "step": 4313
    },
    {
      "epoch": 0.7232945614586608,
      "grad_norm": 0.43269678950309753,
      "learning_rate": 3.7784550029853297e-06,
      "loss": 0.9167,
      "step": 4314
    },
    {
      "epoch": 0.7234622236194069,
      "grad_norm": 0.36526837944984436,
      "learning_rate": 3.774204346580008e-06,
      "loss": 0.8636,
      "step": 4315
    },
    {
      "epoch": 0.723629885780153,
      "grad_norm": 0.39668720960617065,
      "learning_rate": 3.769955526238921e-06,
      "loss": 0.8775,
      "step": 4316
    },
    {
      "epoch": 0.7237975479408991,
      "grad_norm": 0.5399099588394165,
      "learning_rate": 3.7657085432150942e-06,
      "loss": 0.8603,
      "step": 4317
    },
    {
      "epoch": 0.7239652101016452,
      "grad_norm": 0.4076732397079468,
      "learning_rate": 3.7614633987610116e-06,
      "loss": 0.8239,
      "step": 4318
    },
    {
      "epoch": 0.7241328722623913,
      "grad_norm": 0.5806723237037659,
      "learning_rate": 3.7572200941286284e-06,
      "loss": 0.7681,
      "step": 4319
    },
    {
      "epoch": 0.7243005344231374,
      "grad_norm": 0.38126981258392334,
      "learning_rate": 3.7529786305693384e-06,
      "loss": 0.7744,
      "step": 4320
    },
    {
      "epoch": 0.7244681965838835,
      "grad_norm": 0.7048682570457458,
      "learning_rate": 3.7487390093340102e-06,
      "loss": 0.8332,
      "step": 4321
    },
    {
      "epoch": 0.7246358587446295,
      "grad_norm": 0.37661606073379517,
      "learning_rate": 3.7445012316729556e-06,
      "loss": 0.8292,
      "step": 4322
    },
    {
      "epoch": 0.7248035209053757,
      "grad_norm": 0.9588567018508911,
      "learning_rate": 3.7402652988359454e-06,
      "loss": 0.8884,
      "step": 4323
    },
    {
      "epoch": 0.7249711830661217,
      "grad_norm": 0.4693271815776825,
      "learning_rate": 3.736031212072211e-06,
      "loss": 0.8441,
      "step": 4324
    },
    {
      "epoch": 0.7251388452268679,
      "grad_norm": 0.45215103030204773,
      "learning_rate": 3.7317989726304416e-06,
      "loss": 0.8707,
      "step": 4325
    },
    {
      "epoch": 0.7253065073876139,
      "grad_norm": 0.4870089292526245,
      "learning_rate": 3.7275685817587738e-06,
      "loss": 0.7248,
      "step": 4326
    },
    {
      "epoch": 0.7254741695483601,
      "grad_norm": 0.49108240008354187,
      "learning_rate": 3.723340040704797e-06,
      "loss": 0.7689,
      "step": 4327
    },
    {
      "epoch": 0.7256418317091061,
      "grad_norm": 0.4583317041397095,
      "learning_rate": 3.7191133507155697e-06,
      "loss": 0.8313,
      "step": 4328
    },
    {
      "epoch": 0.7258094938698523,
      "grad_norm": 0.4599210321903229,
      "learning_rate": 3.7148885130375858e-06,
      "loss": 0.835,
      "step": 4329
    },
    {
      "epoch": 0.7259771560305983,
      "grad_norm": 0.4726078510284424,
      "learning_rate": 3.7106655289168105e-06,
      "loss": 0.8012,
      "step": 4330
    },
    {
      "epoch": 0.7261448181913445,
      "grad_norm": 0.507954478263855,
      "learning_rate": 3.70644439959865e-06,
      "loss": 0.8109,
      "step": 4331
    },
    {
      "epoch": 0.7263124803520905,
      "grad_norm": 0.557049572467804,
      "learning_rate": 3.702225126327965e-06,
      "loss": 0.8935,
      "step": 4332
    },
    {
      "epoch": 0.7264801425128367,
      "grad_norm": 0.3942612409591675,
      "learning_rate": 3.6980077103490773e-06,
      "loss": 0.7554,
      "step": 4333
    },
    {
      "epoch": 0.7266478046735827,
      "grad_norm": 0.41343289613723755,
      "learning_rate": 3.6937921529057497e-06,
      "loss": 0.8367,
      "step": 4334
    },
    {
      "epoch": 0.7268154668343288,
      "grad_norm": 0.36411136388778687,
      "learning_rate": 3.6895784552412084e-06,
      "loss": 0.7969,
      "step": 4335
    },
    {
      "epoch": 0.7269831289950749,
      "grad_norm": 0.47542253136634827,
      "learning_rate": 3.6853666185981206e-06,
      "loss": 0.84,
      "step": 4336
    },
    {
      "epoch": 0.727150791155821,
      "grad_norm": 0.37301474809646606,
      "learning_rate": 3.6811566442186064e-06,
      "loss": 0.8679,
      "step": 4337
    },
    {
      "epoch": 0.7273184533165671,
      "grad_norm": 0.542221188545227,
      "learning_rate": 3.6769485333442466e-06,
      "loss": 0.9051,
      "step": 4338
    },
    {
      "epoch": 0.7274861154773132,
      "grad_norm": 0.4234585464000702,
      "learning_rate": 3.6727422872160614e-06,
      "loss": 0.8635,
      "step": 4339
    },
    {
      "epoch": 0.7276537776380593,
      "grad_norm": 0.31147825717926025,
      "learning_rate": 3.668537907074522e-06,
      "loss": 0.8151,
      "step": 4340
    },
    {
      "epoch": 0.7278214397988054,
      "grad_norm": 0.7354328632354736,
      "learning_rate": 3.6643353941595582e-06,
      "loss": 0.8694,
      "step": 4341
    },
    {
      "epoch": 0.7279891019595515,
      "grad_norm": 0.4674883782863617,
      "learning_rate": 3.6601347497105366e-06,
      "loss": 0.9029,
      "step": 4342
    },
    {
      "epoch": 0.7281567641202976,
      "grad_norm": 0.4117675721645355,
      "learning_rate": 3.6559359749662828e-06,
      "loss": 0.8143,
      "step": 4343
    },
    {
      "epoch": 0.7283244262810437,
      "grad_norm": 0.5582890510559082,
      "learning_rate": 3.6517390711650737e-06,
      "loss": 0.8927,
      "step": 4344
    },
    {
      "epoch": 0.7284920884417898,
      "grad_norm": 0.5436880588531494,
      "learning_rate": 3.647544039544615e-06,
      "loss": 0.8263,
      "step": 4345
    },
    {
      "epoch": 0.7286597506025358,
      "grad_norm": 0.4514172673225403,
      "learning_rate": 3.643350881342085e-06,
      "loss": 0.8709,
      "step": 4346
    },
    {
      "epoch": 0.728827412763282,
      "grad_norm": 0.5195479989051819,
      "learning_rate": 3.6391595977940884e-06,
      "loss": 0.802,
      "step": 4347
    },
    {
      "epoch": 0.728995074924028,
      "grad_norm": 0.40355563163757324,
      "learning_rate": 3.6349701901366916e-06,
      "loss": 0.8583,
      "step": 4348
    },
    {
      "epoch": 0.7291627370847742,
      "grad_norm": 0.39967405796051025,
      "learning_rate": 3.6307826596054108e-06,
      "loss": 0.7492,
      "step": 4349
    },
    {
      "epoch": 0.7293303992455202,
      "grad_norm": 0.37564578652381897,
      "learning_rate": 3.6265970074351843e-06,
      "loss": 0.8525,
      "step": 4350
    },
    {
      "epoch": 0.7294980614062664,
      "grad_norm": 0.5045909285545349,
      "learning_rate": 3.6224132348604213e-06,
      "loss": 0.8313,
      "step": 4351
    },
    {
      "epoch": 0.7296657235670124,
      "grad_norm": 0.35503095388412476,
      "learning_rate": 3.618231343114972e-06,
      "loss": 0.7584,
      "step": 4352
    },
    {
      "epoch": 0.7298333857277586,
      "grad_norm": 0.3874315619468689,
      "learning_rate": 3.614051333432124e-06,
      "loss": 0.77,
      "step": 4353
    },
    {
      "epoch": 0.7300010478885046,
      "grad_norm": 0.4769584834575653,
      "learning_rate": 3.6098732070446084e-06,
      "loss": 0.8299,
      "step": 4354
    },
    {
      "epoch": 0.7301687100492508,
      "grad_norm": 0.39705613255500793,
      "learning_rate": 3.6056969651846176e-06,
      "loss": 0.8454,
      "step": 4355
    },
    {
      "epoch": 0.7303363722099968,
      "grad_norm": 0.6885538697242737,
      "learning_rate": 3.6015226090837664e-06,
      "loss": 0.841,
      "step": 4356
    },
    {
      "epoch": 0.730504034370743,
      "grad_norm": 0.6540099382400513,
      "learning_rate": 3.5973501399731326e-06,
      "loss": 0.8297,
      "step": 4357
    },
    {
      "epoch": 0.730671696531489,
      "grad_norm": 0.4100639522075653,
      "learning_rate": 3.593179559083225e-06,
      "loss": 0.8448,
      "step": 4358
    },
    {
      "epoch": 0.7308393586922352,
      "grad_norm": 0.6150904893875122,
      "learning_rate": 3.589010867643997e-06,
      "loss": 0.7983,
      "step": 4359
    },
    {
      "epoch": 0.7310070208529812,
      "grad_norm": 0.4606917202472687,
      "learning_rate": 3.5848440668848538e-06,
      "loss": 0.8698,
      "step": 4360
    },
    {
      "epoch": 0.7311746830137273,
      "grad_norm": 0.5340200066566467,
      "learning_rate": 3.580679158034629e-06,
      "loss": 0.788,
      "step": 4361
    },
    {
      "epoch": 0.7313423451744734,
      "grad_norm": 0.5118436217308044,
      "learning_rate": 3.5765161423216145e-06,
      "loss": 0.8007,
      "step": 4362
    },
    {
      "epoch": 0.7315100073352195,
      "grad_norm": 0.3920898139476776,
      "learning_rate": 3.572355020973529e-06,
      "loss": 0.8444,
      "step": 4363
    },
    {
      "epoch": 0.7316776694959656,
      "grad_norm": 0.3829520344734192,
      "learning_rate": 3.568195795217537e-06,
      "loss": 0.8475,
      "step": 4364
    },
    {
      "epoch": 0.7318453316567117,
      "grad_norm": 0.48329228162765503,
      "learning_rate": 3.564038466280252e-06,
      "loss": 0.9246,
      "step": 4365
    },
    {
      "epoch": 0.7320129938174578,
      "grad_norm": 0.467337965965271,
      "learning_rate": 3.5598830353877158e-06,
      "loss": 0.8442,
      "step": 4366
    },
    {
      "epoch": 0.7321806559782039,
      "grad_norm": 0.543903112411499,
      "learning_rate": 3.5557295037654227e-06,
      "loss": 0.8777,
      "step": 4367
    },
    {
      "epoch": 0.73234831813895,
      "grad_norm": 0.4708911180496216,
      "learning_rate": 3.5515778726382967e-06,
      "loss": 0.8563,
      "step": 4368
    },
    {
      "epoch": 0.7325159802996961,
      "grad_norm": 0.44531819224357605,
      "learning_rate": 3.5474281432307024e-06,
      "loss": 0.7835,
      "step": 4369
    },
    {
      "epoch": 0.7326836424604423,
      "grad_norm": 0.9017184972763062,
      "learning_rate": 3.54328031676645e-06,
      "loss": 0.8309,
      "step": 4370
    },
    {
      "epoch": 0.7328513046211883,
      "grad_norm": 0.39330801367759705,
      "learning_rate": 3.5391343944687906e-06,
      "loss": 0.8291,
      "step": 4371
    },
    {
      "epoch": 0.7330189667819345,
      "grad_norm": 0.4601340889930725,
      "learning_rate": 3.5349903775603955e-06,
      "loss": 0.9007,
      "step": 4372
    },
    {
      "epoch": 0.7331866289426805,
      "grad_norm": 0.3983735144138336,
      "learning_rate": 3.530848267263394e-06,
      "loss": 0.8141,
      "step": 4373
    },
    {
      "epoch": 0.7333542911034266,
      "grad_norm": 0.3980402648448944,
      "learning_rate": 3.5267080647993468e-06,
      "loss": 0.8734,
      "step": 4374
    },
    {
      "epoch": 0.7335219532641727,
      "grad_norm": 0.43856269121170044,
      "learning_rate": 3.522569771389246e-06,
      "loss": 0.8857,
      "step": 4375
    },
    {
      "epoch": 0.7336896154249188,
      "grad_norm": 0.53123539686203,
      "learning_rate": 3.5184333882535314e-06,
      "loss": 0.8811,
      "step": 4376
    },
    {
      "epoch": 0.7338572775856649,
      "grad_norm": 0.38238826394081116,
      "learning_rate": 3.514298916612071e-06,
      "loss": 0.8164,
      "step": 4377
    },
    {
      "epoch": 0.734024939746411,
      "grad_norm": 0.551599383354187,
      "learning_rate": 3.510166357684168e-06,
      "loss": 0.896,
      "step": 4378
    },
    {
      "epoch": 0.7341926019071571,
      "grad_norm": 1.1665607690811157,
      "learning_rate": 3.5060357126885725e-06,
      "loss": 0.8109,
      "step": 4379
    },
    {
      "epoch": 0.7343602640679032,
      "grad_norm": 0.4617224633693695,
      "learning_rate": 3.5019069828434547e-06,
      "loss": 0.8853,
      "step": 4380
    },
    {
      "epoch": 0.7345279262286493,
      "grad_norm": 0.4224250614643097,
      "learning_rate": 3.497780169366436e-06,
      "loss": 0.9396,
      "step": 4381
    },
    {
      "epoch": 0.7346955883893954,
      "grad_norm": 0.36724886298179626,
      "learning_rate": 3.493655273474561e-06,
      "loss": 0.8428,
      "step": 4382
    },
    {
      "epoch": 0.7348632505501415,
      "grad_norm": 0.47203654050827026,
      "learning_rate": 3.489532296384309e-06,
      "loss": 0.8778,
      "step": 4383
    },
    {
      "epoch": 0.7350309127108876,
      "grad_norm": 0.4965214133262634,
      "learning_rate": 3.4854112393116047e-06,
      "loss": 0.8247,
      "step": 4384
    },
    {
      "epoch": 0.7351985748716336,
      "grad_norm": 0.5500672459602356,
      "learning_rate": 3.481292103471794e-06,
      "loss": 0.9078,
      "step": 4385
    },
    {
      "epoch": 0.7353662370323798,
      "grad_norm": 0.6066471338272095,
      "learning_rate": 3.4771748900796597e-06,
      "loss": 0.867,
      "step": 4386
    },
    {
      "epoch": 0.7355338991931258,
      "grad_norm": 0.3829139173030853,
      "learning_rate": 3.473059600349424e-06,
      "loss": 0.8491,
      "step": 4387
    },
    {
      "epoch": 0.735701561353872,
      "grad_norm": 0.77088463306427,
      "learning_rate": 3.4689462354947312e-06,
      "loss": 0.8976,
      "step": 4388
    },
    {
      "epoch": 0.735869223514618,
      "grad_norm": 0.39039182662963867,
      "learning_rate": 3.4648347967286655e-06,
      "loss": 0.798,
      "step": 4389
    },
    {
      "epoch": 0.7360368856753642,
      "grad_norm": 0.6210637092590332,
      "learning_rate": 3.4607252852637494e-06,
      "loss": 0.7954,
      "step": 4390
    },
    {
      "epoch": 0.7362045478361102,
      "grad_norm": 0.3575737476348877,
      "learning_rate": 3.456617702311914e-06,
      "loss": 0.8076,
      "step": 4391
    },
    {
      "epoch": 0.7363722099968564,
      "grad_norm": 0.41606733202934265,
      "learning_rate": 3.4525120490845444e-06,
      "loss": 0.9159,
      "step": 4392
    },
    {
      "epoch": 0.7365398721576024,
      "grad_norm": 0.3249188959598541,
      "learning_rate": 3.448408326792451e-06,
      "loss": 0.7628,
      "step": 4393
    },
    {
      "epoch": 0.7367075343183486,
      "grad_norm": 0.572081983089447,
      "learning_rate": 3.4443065366458683e-06,
      "loss": 0.8155,
      "step": 4394
    },
    {
      "epoch": 0.7368751964790946,
      "grad_norm": 0.36839038133621216,
      "learning_rate": 3.4402066798544664e-06,
      "loss": 0.82,
      "step": 4395
    },
    {
      "epoch": 0.7370428586398408,
      "grad_norm": 0.4063968360424042,
      "learning_rate": 3.43610875762734e-06,
      "loss": 0.8493,
      "step": 4396
    },
    {
      "epoch": 0.7372105208005868,
      "grad_norm": 0.3296797573566437,
      "learning_rate": 3.432012771173021e-06,
      "loss": 0.8519,
      "step": 4397
    },
    {
      "epoch": 0.737378182961333,
      "grad_norm": 0.39701390266418457,
      "learning_rate": 3.427918721699469e-06,
      "loss": 0.8686,
      "step": 4398
    },
    {
      "epoch": 0.737545845122079,
      "grad_norm": 0.4578464925289154,
      "learning_rate": 3.423826610414066e-06,
      "loss": 0.7675,
      "step": 4399
    },
    {
      "epoch": 0.7377135072828251,
      "grad_norm": 0.4097997546195984,
      "learning_rate": 3.4197364385236255e-06,
      "loss": 0.8792,
      "step": 4400
    },
    {
      "epoch": 0.7378811694435712,
      "grad_norm": 0.36022815108299255,
      "learning_rate": 3.415648207234393e-06,
      "loss": 0.6904,
      "step": 4401
    },
    {
      "epoch": 0.7380488316043173,
      "grad_norm": 0.5350514054298401,
      "learning_rate": 3.4115619177520343e-06,
      "loss": 0.8796,
      "step": 4402
    },
    {
      "epoch": 0.7382164937650634,
      "grad_norm": 0.5072664618492126,
      "learning_rate": 3.4074775712816533e-06,
      "loss": 0.9138,
      "step": 4403
    },
    {
      "epoch": 0.7383841559258095,
      "grad_norm": 0.36521509289741516,
      "learning_rate": 3.40339516902777e-06,
      "loss": 0.7506,
      "step": 4404
    },
    {
      "epoch": 0.7385518180865556,
      "grad_norm": 0.3442452847957611,
      "learning_rate": 3.3993147121943324e-06,
      "loss": 0.8652,
      "step": 4405
    },
    {
      "epoch": 0.7387194802473017,
      "grad_norm": 0.399734228849411,
      "learning_rate": 3.395236201984725e-06,
      "loss": 0.7461,
      "step": 4406
    },
    {
      "epoch": 0.7388871424080478,
      "grad_norm": 0.35992172360420227,
      "learning_rate": 3.3911596396017433e-06,
      "loss": 0.8682,
      "step": 4407
    },
    {
      "epoch": 0.7390548045687939,
      "grad_norm": 0.4022121727466583,
      "learning_rate": 3.387085026247624e-06,
      "loss": 0.8703,
      "step": 4408
    },
    {
      "epoch": 0.73922246672954,
      "grad_norm": 0.4650803506374359,
      "learning_rate": 3.3830123631240163e-06,
      "loss": 0.817,
      "step": 4409
    },
    {
      "epoch": 0.7393901288902861,
      "grad_norm": 0.4014977514743805,
      "learning_rate": 3.378941651431996e-06,
      "loss": 0.7755,
      "step": 4410
    },
    {
      "epoch": 0.7395577910510321,
      "grad_norm": 0.5557853579521179,
      "learning_rate": 3.374872892372074e-06,
      "loss": 0.8462,
      "step": 4411
    },
    {
      "epoch": 0.7397254532117783,
      "grad_norm": 0.380247563123703,
      "learning_rate": 3.370806087144174e-06,
      "loss": 0.8086,
      "step": 4412
    },
    {
      "epoch": 0.7398931153725243,
      "grad_norm": 0.5533876419067383,
      "learning_rate": 3.366741236947644e-06,
      "loss": 0.814,
      "step": 4413
    },
    {
      "epoch": 0.7400607775332705,
      "grad_norm": 0.8545756936073303,
      "learning_rate": 3.3626783429812635e-06,
      "loss": 0.8038,
      "step": 4414
    },
    {
      "epoch": 0.7402284396940165,
      "grad_norm": 0.5220400094985962,
      "learning_rate": 3.3586174064432262e-06,
      "loss": 0.9231,
      "step": 4415
    },
    {
      "epoch": 0.7403961018547627,
      "grad_norm": 0.4009403586387634,
      "learning_rate": 3.354558428531154e-06,
      "loss": 0.9235,
      "step": 4416
    },
    {
      "epoch": 0.7405637640155087,
      "grad_norm": 0.426805704832077,
      "learning_rate": 3.3505014104420976e-06,
      "loss": 0.7895,
      "step": 4417
    },
    {
      "epoch": 0.7407314261762549,
      "grad_norm": 0.3550259470939636,
      "learning_rate": 3.346446353372508e-06,
      "loss": 0.8136,
      "step": 4418
    },
    {
      "epoch": 0.7408990883370009,
      "grad_norm": 0.38828107714653015,
      "learning_rate": 3.342393258518278e-06,
      "loss": 0.8093,
      "step": 4419
    },
    {
      "epoch": 0.7410667504977471,
      "grad_norm": 1.0287100076675415,
      "learning_rate": 3.338342127074721e-06,
      "loss": 0.839,
      "step": 4420
    },
    {
      "epoch": 0.7412344126584931,
      "grad_norm": 0.45076051354408264,
      "learning_rate": 3.334292960236556e-06,
      "loss": 0.8088,
      "step": 4421
    },
    {
      "epoch": 0.7414020748192393,
      "grad_norm": 0.4348759949207306,
      "learning_rate": 3.330245759197941e-06,
      "loss": 0.8436,
      "step": 4422
    },
    {
      "epoch": 0.7415697369799853,
      "grad_norm": 0.4137447774410248,
      "learning_rate": 3.326200525152441e-06,
      "loss": 0.8775,
      "step": 4423
    },
    {
      "epoch": 0.7417373991407314,
      "grad_norm": 0.5322988629341125,
      "learning_rate": 3.322157259293044e-06,
      "loss": 0.8508,
      "step": 4424
    },
    {
      "epoch": 0.7419050613014775,
      "grad_norm": 0.5858178734779358,
      "learning_rate": 3.318115962812165e-06,
      "loss": 0.8505,
      "step": 4425
    },
    {
      "epoch": 0.7420727234622236,
      "grad_norm": 0.44442102313041687,
      "learning_rate": 3.31407663690163e-06,
      "loss": 0.7975,
      "step": 4426
    },
    {
      "epoch": 0.7422403856229697,
      "grad_norm": 0.3603506088256836,
      "learning_rate": 3.3100392827526807e-06,
      "loss": 0.8318,
      "step": 4427
    },
    {
      "epoch": 0.7424080477837158,
      "grad_norm": 0.4573531448841095,
      "learning_rate": 3.306003901555991e-06,
      "loss": 0.892,
      "step": 4428
    },
    {
      "epoch": 0.7425757099444619,
      "grad_norm": 0.4490502178668976,
      "learning_rate": 3.3019704945016385e-06,
      "loss": 0.8911,
      "step": 4429
    },
    {
      "epoch": 0.742743372105208,
      "grad_norm": 0.6446260809898376,
      "learning_rate": 3.297939062779131e-06,
      "loss": 0.9423,
      "step": 4430
    },
    {
      "epoch": 0.7429110342659541,
      "grad_norm": 0.4458027780056,
      "learning_rate": 3.2939096075773835e-06,
      "loss": 0.8041,
      "step": 4431
    },
    {
      "epoch": 0.7430786964267002,
      "grad_norm": 0.40491101145744324,
      "learning_rate": 3.2898821300847294e-06,
      "loss": 0.7675,
      "step": 4432
    },
    {
      "epoch": 0.7432463585874463,
      "grad_norm": 0.650396466255188,
      "learning_rate": 3.2858566314889295e-06,
      "loss": 0.8963,
      "step": 4433
    },
    {
      "epoch": 0.7434140207481924,
      "grad_norm": 0.4394979476928711,
      "learning_rate": 3.281833112977145e-06,
      "loss": 0.7744,
      "step": 4434
    },
    {
      "epoch": 0.7435816829089384,
      "grad_norm": 0.488955557346344,
      "learning_rate": 3.2778115757359695e-06,
      "loss": 0.9099,
      "step": 4435
    },
    {
      "epoch": 0.7437493450696846,
      "grad_norm": 0.4294459819793701,
      "learning_rate": 3.2737920209513994e-06,
      "loss": 0.7751,
      "step": 4436
    },
    {
      "epoch": 0.7439170072304306,
      "grad_norm": 0.3901923596858978,
      "learning_rate": 3.269774449808849e-06,
      "loss": 0.8058,
      "step": 4437
    },
    {
      "epoch": 0.7440846693911768,
      "grad_norm": 0.4745529294013977,
      "learning_rate": 3.2657588634931516e-06,
      "loss": 0.8483,
      "step": 4438
    },
    {
      "epoch": 0.7442523315519228,
      "grad_norm": 0.42034587264060974,
      "learning_rate": 3.261745263188557e-06,
      "loss": 0.8383,
      "step": 4439
    },
    {
      "epoch": 0.744419993712669,
      "grad_norm": 0.3854556977748871,
      "learning_rate": 3.2577336500787237e-06,
      "loss": 0.7848,
      "step": 4440
    },
    {
      "epoch": 0.744587655873415,
      "grad_norm": 0.35779571533203125,
      "learning_rate": 3.253724025346723e-06,
      "loss": 0.8042,
      "step": 4441
    },
    {
      "epoch": 0.7447553180341612,
      "grad_norm": 0.48553743958473206,
      "learning_rate": 3.249716390175046e-06,
      "loss": 0.8119,
      "step": 4442
    },
    {
      "epoch": 0.7449229801949072,
      "grad_norm": 0.4442885220050812,
      "learning_rate": 3.2457107457455915e-06,
      "loss": 0.8189,
      "step": 4443
    },
    {
      "epoch": 0.7450906423556534,
      "grad_norm": 0.3811306059360504,
      "learning_rate": 3.241707093239681e-06,
      "loss": 0.8197,
      "step": 4444
    },
    {
      "epoch": 0.7452583045163994,
      "grad_norm": 0.5182856321334839,
      "learning_rate": 3.23770543383803e-06,
      "loss": 0.8627,
      "step": 4445
    },
    {
      "epoch": 0.7454259666771456,
      "grad_norm": 0.5060182809829712,
      "learning_rate": 3.2337057687207816e-06,
      "loss": 0.8235,
      "step": 4446
    },
    {
      "epoch": 0.7455936288378916,
      "grad_norm": 0.5154516100883484,
      "learning_rate": 3.229708099067491e-06,
      "loss": 0.893,
      "step": 4447
    },
    {
      "epoch": 0.7457612909986377,
      "grad_norm": 0.40294402837753296,
      "learning_rate": 3.225712426057115e-06,
      "loss": 0.948,
      "step": 4448
    },
    {
      "epoch": 0.7459289531593838,
      "grad_norm": 0.36278417706489563,
      "learning_rate": 3.2217187508680314e-06,
      "loss": 0.7185,
      "step": 4449
    },
    {
      "epoch": 0.7460966153201299,
      "grad_norm": 0.4962402582168579,
      "learning_rate": 3.217727074678023e-06,
      "loss": 0.8488,
      "step": 4450
    },
    {
      "epoch": 0.746264277480876,
      "grad_norm": 0.4561363160610199,
      "learning_rate": 3.2137373986642807e-06,
      "loss": 0.8673,
      "step": 4451
    },
    {
      "epoch": 0.7464319396416221,
      "grad_norm": 0.5583760142326355,
      "learning_rate": 3.2097497240034147e-06,
      "loss": 0.841,
      "step": 4452
    },
    {
      "epoch": 0.7465996018023682,
      "grad_norm": 0.4732387065887451,
      "learning_rate": 3.205764051871435e-06,
      "loss": 0.8238,
      "step": 4453
    },
    {
      "epoch": 0.7467672639631143,
      "grad_norm": 0.4361114799976349,
      "learning_rate": 3.201780383443772e-06,
      "loss": 0.8675,
      "step": 4454
    },
    {
      "epoch": 0.7469349261238605,
      "grad_norm": 0.6804105043411255,
      "learning_rate": 3.1977987198952543e-06,
      "loss": 0.8752,
      "step": 4455
    },
    {
      "epoch": 0.7471025882846065,
      "grad_norm": 0.6557078957557678,
      "learning_rate": 3.193819062400122e-06,
      "loss": 0.7997,
      "step": 4456
    },
    {
      "epoch": 0.7472702504453527,
      "grad_norm": 0.5991407036781311,
      "learning_rate": 3.1898414121320277e-06,
      "loss": 0.8506,
      "step": 4457
    },
    {
      "epoch": 0.7474379126060987,
      "grad_norm": 0.5412923097610474,
      "learning_rate": 3.185865770264035e-06,
      "loss": 0.9054,
      "step": 4458
    },
    {
      "epoch": 0.7476055747668449,
      "grad_norm": 0.48456665873527527,
      "learning_rate": 3.1818921379685997e-06,
      "loss": 0.9034,
      "step": 4459
    },
    {
      "epoch": 0.7477732369275909,
      "grad_norm": 0.4137357771396637,
      "learning_rate": 3.177920516417603e-06,
      "loss": 0.8505,
      "step": 4460
    },
    {
      "epoch": 0.747940899088337,
      "grad_norm": 0.48029109835624695,
      "learning_rate": 3.1739509067823192e-06,
      "loss": 0.8321,
      "step": 4461
    },
    {
      "epoch": 0.7481085612490831,
      "grad_norm": 0.8238043785095215,
      "learning_rate": 3.1699833102334397e-06,
      "loss": 0.8374,
      "step": 4462
    },
    {
      "epoch": 0.7482762234098292,
      "grad_norm": 0.42662978172302246,
      "learning_rate": 3.166017727941062e-06,
      "loss": 0.7012,
      "step": 4463
    },
    {
      "epoch": 0.7484438855705753,
      "grad_norm": 0.42485249042510986,
      "learning_rate": 3.1620541610746747e-06,
      "loss": 0.8489,
      "step": 4464
    },
    {
      "epoch": 0.7486115477313214,
      "grad_norm": 0.35580894351005554,
      "learning_rate": 3.1580926108031884e-06,
      "loss": 0.8409,
      "step": 4465
    },
    {
      "epoch": 0.7487792098920675,
      "grad_norm": 0.3833909332752228,
      "learning_rate": 3.154133078294915e-06,
      "loss": 0.8809,
      "step": 4466
    },
    {
      "epoch": 0.7489468720528136,
      "grad_norm": 0.4267159402370453,
      "learning_rate": 3.1501755647175693e-06,
      "loss": 0.7405,
      "step": 4467
    },
    {
      "epoch": 0.7491145342135597,
      "grad_norm": 0.5588497519493103,
      "learning_rate": 3.1462200712382662e-06,
      "loss": 0.8341,
      "step": 4468
    },
    {
      "epoch": 0.7492821963743058,
      "grad_norm": 0.48638758063316345,
      "learning_rate": 3.142266599023537e-06,
      "loss": 0.8086,
      "step": 4469
    },
    {
      "epoch": 0.7494498585350519,
      "grad_norm": 0.37331661581993103,
      "learning_rate": 3.1383151492393015e-06,
      "loss": 0.8458,
      "step": 4470
    },
    {
      "epoch": 0.749617520695798,
      "grad_norm": 0.5566949844360352,
      "learning_rate": 3.1343657230508985e-06,
      "loss": 0.8528,
      "step": 4471
    },
    {
      "epoch": 0.749785182856544,
      "grad_norm": 0.38358446955680847,
      "learning_rate": 3.130418321623061e-06,
      "loss": 0.8063,
      "step": 4472
    },
    {
      "epoch": 0.7499528450172902,
      "grad_norm": 0.5433733463287354,
      "learning_rate": 3.1264729461199208e-06,
      "loss": 0.845,
      "step": 4473
    },
    {
      "epoch": 0.7501205071780362,
      "grad_norm": 0.46310821175575256,
      "learning_rate": 3.1225295977050265e-06,
      "loss": 0.8762,
      "step": 4474
    },
    {
      "epoch": 0.7502881693387824,
      "grad_norm": 0.3718121647834778,
      "learning_rate": 3.1185882775413123e-06,
      "loss": 0.7871,
      "step": 4475
    },
    {
      "epoch": 0.7504558314995284,
      "grad_norm": 0.44388100504875183,
      "learning_rate": 3.114648986791129e-06,
      "loss": 0.8149,
      "step": 4476
    },
    {
      "epoch": 0.7506234936602746,
      "grad_norm": 0.4701133668422699,
      "learning_rate": 3.11071172661622e-06,
      "loss": 0.8766,
      "step": 4477
    },
    {
      "epoch": 0.7507911558210206,
      "grad_norm": 0.4463254511356354,
      "learning_rate": 3.1067764981777292e-06,
      "loss": 0.8164,
      "step": 4478
    },
    {
      "epoch": 0.7509588179817668,
      "grad_norm": 0.5636247396469116,
      "learning_rate": 3.1028433026362093e-06,
      "loss": 0.8465,
      "step": 4479
    },
    {
      "epoch": 0.7511264801425128,
      "grad_norm": 1.0990864038467407,
      "learning_rate": 3.098912141151602e-06,
      "loss": 0.8136,
      "step": 4480
    },
    {
      "epoch": 0.751294142303259,
      "grad_norm": 0.4750821888446808,
      "learning_rate": 3.0949830148832627e-06,
      "loss": 0.8361,
      "step": 4481
    },
    {
      "epoch": 0.751461804464005,
      "grad_norm": 0.4538353383541107,
      "learning_rate": 3.0910559249899364e-06,
      "loss": 0.9258,
      "step": 4482
    },
    {
      "epoch": 0.7516294666247512,
      "grad_norm": 0.408823162317276,
      "learning_rate": 3.0871308726297667e-06,
      "loss": 0.758,
      "step": 4483
    },
    {
      "epoch": 0.7517971287854972,
      "grad_norm": 0.6528570055961609,
      "learning_rate": 3.083207858960302e-06,
      "loss": 0.8146,
      "step": 4484
    },
    {
      "epoch": 0.7519647909462434,
      "grad_norm": 0.42876380681991577,
      "learning_rate": 3.079286885138497e-06,
      "loss": 0.8435,
      "step": 4485
    },
    {
      "epoch": 0.7521324531069894,
      "grad_norm": 0.7172784805297852,
      "learning_rate": 3.0753679523206803e-06,
      "loss": 0.8946,
      "step": 4486
    },
    {
      "epoch": 0.7523001152677355,
      "grad_norm": 0.5575166344642639,
      "learning_rate": 3.0714510616626012e-06,
      "loss": 0.8673,
      "step": 4487
    },
    {
      "epoch": 0.7524677774284816,
      "grad_norm": 0.439159631729126,
      "learning_rate": 3.067536214319402e-06,
      "loss": 0.7876,
      "step": 4488
    },
    {
      "epoch": 0.7526354395892277,
      "grad_norm": 0.3544710874557495,
      "learning_rate": 3.0636234114456144e-06,
      "loss": 0.7859,
      "step": 4489
    },
    {
      "epoch": 0.7528031017499738,
      "grad_norm": 0.3452700972557068,
      "learning_rate": 3.059712654195176e-06,
      "loss": 0.8961,
      "step": 4490
    },
    {
      "epoch": 0.7529707639107199,
      "grad_norm": 0.3834369480609894,
      "learning_rate": 3.055803943721417e-06,
      "loss": 0.8511,
      "step": 4491
    },
    {
      "epoch": 0.753138426071466,
      "grad_norm": 0.3766850531101227,
      "learning_rate": 3.051897281177061e-06,
      "loss": 0.79,
      "step": 4492
    },
    {
      "epoch": 0.7533060882322121,
      "grad_norm": 0.3521707057952881,
      "learning_rate": 3.047992667714237e-06,
      "loss": 0.6716,
      "step": 4493
    },
    {
      "epoch": 0.7534737503929582,
      "grad_norm": 0.4038065969944,
      "learning_rate": 3.044090104484456e-06,
      "loss": 0.7564,
      "step": 4494
    },
    {
      "epoch": 0.7536414125537043,
      "grad_norm": 0.4788320064544678,
      "learning_rate": 3.0401895926386417e-06,
      "loss": 0.8327,
      "step": 4495
    },
    {
      "epoch": 0.7538090747144504,
      "grad_norm": 0.5993980765342712,
      "learning_rate": 3.0362911333270983e-06,
      "loss": 0.8236,
      "step": 4496
    },
    {
      "epoch": 0.7539767368751965,
      "grad_norm": 0.538329005241394,
      "learning_rate": 3.032394727699527e-06,
      "loss": 0.8922,
      "step": 4497
    },
    {
      "epoch": 0.7541443990359425,
      "grad_norm": 0.3130503296852112,
      "learning_rate": 3.0285003769050314e-06,
      "loss": 0.7434,
      "step": 4498
    },
    {
      "epoch": 0.7543120611966887,
      "grad_norm": 0.4620181918144226,
      "learning_rate": 3.024608082092102e-06,
      "loss": 0.8075,
      "step": 4499
    },
    {
      "epoch": 0.7544797233574347,
      "grad_norm": 0.40521252155303955,
      "learning_rate": 3.0207178444086206e-06,
      "loss": 0.881,
      "step": 4500
    },
    {
      "epoch": 0.7546473855181809,
      "grad_norm": 0.39265090227127075,
      "learning_rate": 3.0168296650018736e-06,
      "loss": 0.8214,
      "step": 4501
    },
    {
      "epoch": 0.7548150476789269,
      "grad_norm": 0.48276379704475403,
      "learning_rate": 3.012943545018526e-06,
      "loss": 0.8697,
      "step": 4502
    },
    {
      "epoch": 0.7549827098396731,
      "grad_norm": 0.38085293769836426,
      "learning_rate": 3.0090594856046473e-06,
      "loss": 0.8828,
      "step": 4503
    },
    {
      "epoch": 0.7551503720004191,
      "grad_norm": 0.3857152462005615,
      "learning_rate": 3.0051774879056995e-06,
      "loss": 0.8339,
      "step": 4504
    },
    {
      "epoch": 0.7553180341611653,
      "grad_norm": 0.6330828070640564,
      "learning_rate": 3.00129755306652e-06,
      "loss": 0.6985,
      "step": 4505
    },
    {
      "epoch": 0.7554856963219113,
      "grad_norm": 0.3679242432117462,
      "learning_rate": 2.997419682231355e-06,
      "loss": 0.8357,
      "step": 4506
    },
    {
      "epoch": 0.7556533584826575,
      "grad_norm": 0.5782294273376465,
      "learning_rate": 2.993543876543842e-06,
      "loss": 0.8307,
      "step": 4507
    },
    {
      "epoch": 0.7558210206434035,
      "grad_norm": 0.8186786770820618,
      "learning_rate": 2.9896701371469983e-06,
      "loss": 0.7709,
      "step": 4508
    },
    {
      "epoch": 0.7559886828041497,
      "grad_norm": 0.46019449830055237,
      "learning_rate": 2.9857984651832396e-06,
      "loss": 0.8934,
      "step": 4509
    },
    {
      "epoch": 0.7561563449648957,
      "grad_norm": 0.39549770951271057,
      "learning_rate": 2.981928861794364e-06,
      "loss": 0.8112,
      "step": 4510
    },
    {
      "epoch": 0.7563240071256419,
      "grad_norm": 0.4370620548725128,
      "learning_rate": 2.978061328121573e-06,
      "loss": 0.8993,
      "step": 4511
    },
    {
      "epoch": 0.7564916692863879,
      "grad_norm": 0.421627014875412,
      "learning_rate": 2.9741958653054494e-06,
      "loss": 0.841,
      "step": 4512
    },
    {
      "epoch": 0.756659331447134,
      "grad_norm": 0.36653319001197815,
      "learning_rate": 2.9703324744859653e-06,
      "loss": 0.8663,
      "step": 4513
    },
    {
      "epoch": 0.7568269936078801,
      "grad_norm": 0.38584211468696594,
      "learning_rate": 2.966471156802477e-06,
      "loss": 0.875,
      "step": 4514
    },
    {
      "epoch": 0.7569946557686262,
      "grad_norm": 0.4006466269493103,
      "learning_rate": 2.962611913393744e-06,
      "loss": 0.8039,
      "step": 4515
    },
    {
      "epoch": 0.7571623179293723,
      "grad_norm": 0.45578905940055847,
      "learning_rate": 2.958754745397896e-06,
      "loss": 0.7878,
      "step": 4516
    },
    {
      "epoch": 0.7573299800901184,
      "grad_norm": 0.7102530598640442,
      "learning_rate": 2.9548996539524676e-06,
      "loss": 0.8346,
      "step": 4517
    },
    {
      "epoch": 0.7574976422508645,
      "grad_norm": 0.4307156205177307,
      "learning_rate": 2.9510466401943703e-06,
      "loss": 0.8183,
      "step": 4518
    },
    {
      "epoch": 0.7576653044116106,
      "grad_norm": 0.5517482161521912,
      "learning_rate": 2.947195705259901e-06,
      "loss": 0.8893,
      "step": 4519
    },
    {
      "epoch": 0.7578329665723567,
      "grad_norm": 0.4402889311313629,
      "learning_rate": 2.943346850284754e-06,
      "loss": 0.7601,
      "step": 4520
    },
    {
      "epoch": 0.7580006287331028,
      "grad_norm": 0.6082161068916321,
      "learning_rate": 2.9395000764040007e-06,
      "loss": 0.8256,
      "step": 4521
    },
    {
      "epoch": 0.7581682908938489,
      "grad_norm": 0.43149957060813904,
      "learning_rate": 2.935655384752105e-06,
      "loss": 0.8834,
      "step": 4522
    },
    {
      "epoch": 0.758335953054595,
      "grad_norm": 0.3590730130672455,
      "learning_rate": 2.9318127764629135e-06,
      "loss": 0.7763,
      "step": 4523
    },
    {
      "epoch": 0.758503615215341,
      "grad_norm": 0.5444132089614868,
      "learning_rate": 2.9279722526696552e-06,
      "loss": 0.904,
      "step": 4524
    },
    {
      "epoch": 0.7586712773760872,
      "grad_norm": 0.38523122668266296,
      "learning_rate": 2.9241338145049546e-06,
      "loss": 0.7918,
      "step": 4525
    },
    {
      "epoch": 0.7588389395368332,
      "grad_norm": 0.9955673217773438,
      "learning_rate": 2.9202974631008084e-06,
      "loss": 0.8853,
      "step": 4526
    },
    {
      "epoch": 0.7590066016975794,
      "grad_norm": 0.4135846197605133,
      "learning_rate": 2.9164631995886095e-06,
      "loss": 0.7754,
      "step": 4527
    },
    {
      "epoch": 0.7591742638583254,
      "grad_norm": 0.5456740260124207,
      "learning_rate": 2.9126310250991275e-06,
      "loss": 0.8535,
      "step": 4528
    },
    {
      "epoch": 0.7593419260190716,
      "grad_norm": 1.1154274940490723,
      "learning_rate": 2.908800940762514e-06,
      "loss": 0.9219,
      "step": 4529
    },
    {
      "epoch": 0.7595095881798176,
      "grad_norm": 0.40234553813934326,
      "learning_rate": 2.904972947708312e-06,
      "loss": 0.8607,
      "step": 4530
    },
    {
      "epoch": 0.7596772503405638,
      "grad_norm": 0.33538851141929626,
      "learning_rate": 2.9011470470654512e-06,
      "loss": 0.8112,
      "step": 4531
    },
    {
      "epoch": 0.7598449125013098,
      "grad_norm": 0.42388421297073364,
      "learning_rate": 2.8973232399622242e-06,
      "loss": 0.8061,
      "step": 4532
    },
    {
      "epoch": 0.760012574662056,
      "grad_norm": 0.6167278289794922,
      "learning_rate": 2.8935015275263235e-06,
      "loss": 0.8011,
      "step": 4533
    },
    {
      "epoch": 0.760180236822802,
      "grad_norm": 0.473337858915329,
      "learning_rate": 2.889681910884825e-06,
      "loss": 0.9186,
      "step": 4534
    },
    {
      "epoch": 0.7603478989835482,
      "grad_norm": 0.4435788691043854,
      "learning_rate": 2.8858643911641726e-06,
      "loss": 0.831,
      "step": 4535
    },
    {
      "epoch": 0.7605155611442942,
      "grad_norm": 0.381301611661911,
      "learning_rate": 2.882048969490208e-06,
      "loss": 0.887,
      "step": 4536
    },
    {
      "epoch": 0.7606832233050403,
      "grad_norm": 0.5315547585487366,
      "learning_rate": 2.8782356469881424e-06,
      "loss": 0.9597,
      "step": 4537
    },
    {
      "epoch": 0.7608508854657864,
      "grad_norm": 0.43942558765411377,
      "learning_rate": 2.874424424782569e-06,
      "loss": 0.8002,
      "step": 4538
    },
    {
      "epoch": 0.7610185476265325,
      "grad_norm": 0.35768264532089233,
      "learning_rate": 2.8706153039974705e-06,
      "loss": 0.8481,
      "step": 4539
    },
    {
      "epoch": 0.7611862097872786,
      "grad_norm": 0.405945748090744,
      "learning_rate": 2.8668082857562006e-06,
      "loss": 0.7964,
      "step": 4540
    },
    {
      "epoch": 0.7613538719480247,
      "grad_norm": 0.3525673449039459,
      "learning_rate": 2.863003371181493e-06,
      "loss": 0.8708,
      "step": 4541
    },
    {
      "epoch": 0.7615215341087709,
      "grad_norm": 0.5565782189369202,
      "learning_rate": 2.85920056139547e-06,
      "loss": 0.8222,
      "step": 4542
    },
    {
      "epoch": 0.7616891962695169,
      "grad_norm": 0.36413130164146423,
      "learning_rate": 2.855399857519623e-06,
      "loss": 0.894,
      "step": 4543
    },
    {
      "epoch": 0.7618568584302631,
      "grad_norm": 0.4103677570819855,
      "learning_rate": 2.85160126067483e-06,
      "loss": 0.9292,
      "step": 4544
    },
    {
      "epoch": 0.7620245205910091,
      "grad_norm": 0.4916219711303711,
      "learning_rate": 2.847804771981343e-06,
      "loss": 0.8147,
      "step": 4545
    },
    {
      "epoch": 0.7621921827517553,
      "grad_norm": 0.4736553430557251,
      "learning_rate": 2.8440103925587904e-06,
      "loss": 0.7978,
      "step": 4546
    },
    {
      "epoch": 0.7623598449125013,
      "grad_norm": 0.38820213079452515,
      "learning_rate": 2.840218123526186e-06,
      "loss": 0.9285,
      "step": 4547
    },
    {
      "epoch": 0.7625275070732475,
      "grad_norm": 0.4562012553215027,
      "learning_rate": 2.8364279660019135e-06,
      "loss": 0.8501,
      "step": 4548
    },
    {
      "epoch": 0.7626951692339935,
      "grad_norm": 0.4280639588832855,
      "learning_rate": 2.8326399211037413e-06,
      "loss": 0.8802,
      "step": 4549
    },
    {
      "epoch": 0.7628628313947396,
      "grad_norm": 0.7219691872596741,
      "learning_rate": 2.8288539899488086e-06,
      "loss": 0.8217,
      "step": 4550
    },
    {
      "epoch": 0.7630304935554857,
      "grad_norm": 0.4507511258125305,
      "learning_rate": 2.825070173653631e-06,
      "loss": 0.9204,
      "step": 4551
    },
    {
      "epoch": 0.7631981557162318,
      "grad_norm": 0.47085505723953247,
      "learning_rate": 2.821288473334104e-06,
      "loss": 0.8355,
      "step": 4552
    },
    {
      "epoch": 0.7633658178769779,
      "grad_norm": 0.38516542315483093,
      "learning_rate": 2.8175088901055026e-06,
      "loss": 0.83,
      "step": 4553
    },
    {
      "epoch": 0.763533480037724,
      "grad_norm": 0.5478811264038086,
      "learning_rate": 2.8137314250824686e-06,
      "loss": 0.8329,
      "step": 4554
    },
    {
      "epoch": 0.7637011421984701,
      "grad_norm": 0.4037216305732727,
      "learning_rate": 2.8099560793790204e-06,
      "loss": 0.9106,
      "step": 4555
    },
    {
      "epoch": 0.7638688043592162,
      "grad_norm": 0.3449244499206543,
      "learning_rate": 2.806182854108561e-06,
      "loss": 0.7537,
      "step": 4556
    },
    {
      "epoch": 0.7640364665199623,
      "grad_norm": 0.43161898851394653,
      "learning_rate": 2.8024117503838543e-06,
      "loss": 0.8963,
      "step": 4557
    },
    {
      "epoch": 0.7642041286807084,
      "grad_norm": 0.3741050064563751,
      "learning_rate": 2.798642769317055e-06,
      "loss": 0.7326,
      "step": 4558
    },
    {
      "epoch": 0.7643717908414545,
      "grad_norm": 0.4074038565158844,
      "learning_rate": 2.794875912019669e-06,
      "loss": 0.9326,
      "step": 4559
    },
    {
      "epoch": 0.7645394530022006,
      "grad_norm": 0.5346018671989441,
      "learning_rate": 2.791111179602597e-06,
      "loss": 0.8441,
      "step": 4560
    },
    {
      "epoch": 0.7647071151629466,
      "grad_norm": 0.2614295780658722,
      "learning_rate": 2.7873485731761074e-06,
      "loss": 0.7906,
      "step": 4561
    },
    {
      "epoch": 0.7648747773236928,
      "grad_norm": 0.3567403256893158,
      "learning_rate": 2.783588093849834e-06,
      "loss": 0.8092,
      "step": 4562
    },
    {
      "epoch": 0.7650424394844388,
      "grad_norm": 0.46598732471466064,
      "learning_rate": 2.7798297427327938e-06,
      "loss": 0.9027,
      "step": 4563
    },
    {
      "epoch": 0.765210101645185,
      "grad_norm": 0.7750427722930908,
      "learning_rate": 2.776073520933369e-06,
      "loss": 0.8619,
      "step": 4564
    },
    {
      "epoch": 0.765377763805931,
      "grad_norm": 0.3993176817893982,
      "learning_rate": 2.7723194295593113e-06,
      "loss": 0.854,
      "step": 4565
    },
    {
      "epoch": 0.7655454259666772,
      "grad_norm": 0.40991339087486267,
      "learning_rate": 2.7685674697177568e-06,
      "loss": 0.9568,
      "step": 4566
    },
    {
      "epoch": 0.7657130881274232,
      "grad_norm": 0.4112975001335144,
      "learning_rate": 2.7648176425151972e-06,
      "loss": 0.8523,
      "step": 4567
    },
    {
      "epoch": 0.7658807502881694,
      "grad_norm": 0.35256874561309814,
      "learning_rate": 2.76106994905751e-06,
      "loss": 0.8159,
      "step": 4568
    },
    {
      "epoch": 0.7660484124489154,
      "grad_norm": 0.3873903155326843,
      "learning_rate": 2.7573243904499338e-06,
      "loss": 0.8696,
      "step": 4569
    },
    {
      "epoch": 0.7662160746096616,
      "grad_norm": 0.6696532368659973,
      "learning_rate": 2.753580967797076e-06,
      "loss": 0.8513,
      "step": 4570
    },
    {
      "epoch": 0.7663837367704076,
      "grad_norm": 0.6126317977905273,
      "learning_rate": 2.7498396822029216e-06,
      "loss": 0.7612,
      "step": 4571
    },
    {
      "epoch": 0.7665513989311538,
      "grad_norm": 0.5345373749732971,
      "learning_rate": 2.746100534770829e-06,
      "loss": 0.8084,
      "step": 4572
    },
    {
      "epoch": 0.7667190610918998,
      "grad_norm": 0.3670309782028198,
      "learning_rate": 2.7423635266035053e-06,
      "loss": 0.7996,
      "step": 4573
    },
    {
      "epoch": 0.766886723252646,
      "grad_norm": 0.5064563155174255,
      "learning_rate": 2.738628658803052e-06,
      "loss": 0.8099,
      "step": 4574
    },
    {
      "epoch": 0.767054385413392,
      "grad_norm": 0.44875088334083557,
      "learning_rate": 2.734895932470919e-06,
      "loss": 0.8375,
      "step": 4575
    },
    {
      "epoch": 0.7672220475741381,
      "grad_norm": 0.4273213744163513,
      "learning_rate": 2.7311653487079394e-06,
      "loss": 0.8003,
      "step": 4576
    },
    {
      "epoch": 0.7673897097348842,
      "grad_norm": 0.42424747347831726,
      "learning_rate": 2.7274369086143127e-06,
      "loss": 0.8833,
      "step": 4577
    },
    {
      "epoch": 0.7675573718956303,
      "grad_norm": 0.3960532546043396,
      "learning_rate": 2.7237106132895897e-06,
      "loss": 0.8208,
      "step": 4578
    },
    {
      "epoch": 0.7677250340563764,
      "grad_norm": 0.38370004296302795,
      "learning_rate": 2.719986463832708e-06,
      "loss": 0.8362,
      "step": 4579
    },
    {
      "epoch": 0.7678926962171225,
      "grad_norm": 0.478929728269577,
      "learning_rate": 2.7162644613419687e-06,
      "loss": 0.8642,
      "step": 4580
    },
    {
      "epoch": 0.7680603583778686,
      "grad_norm": 0.806081235408783,
      "learning_rate": 2.712544606915032e-06,
      "loss": 0.8639,
      "step": 4581
    },
    {
      "epoch": 0.7682280205386147,
      "grad_norm": 0.8366953134536743,
      "learning_rate": 2.708826901648928e-06,
      "loss": 0.8082,
      "step": 4582
    },
    {
      "epoch": 0.7683956826993608,
      "grad_norm": 0.6663612723350525,
      "learning_rate": 2.705111346640058e-06,
      "loss": 0.8111,
      "step": 4583
    },
    {
      "epoch": 0.7685633448601069,
      "grad_norm": 0.39448583126068115,
      "learning_rate": 2.7013979429841807e-06,
      "loss": 0.8399,
      "step": 4584
    },
    {
      "epoch": 0.768731007020853,
      "grad_norm": 0.40021032094955444,
      "learning_rate": 2.69768669177643e-06,
      "loss": 0.7316,
      "step": 4585
    },
    {
      "epoch": 0.7688986691815991,
      "grad_norm": 0.378376841545105,
      "learning_rate": 2.6939775941112965e-06,
      "loss": 0.7462,
      "step": 4586
    },
    {
      "epoch": 0.7690663313423451,
      "grad_norm": 0.5967662334442139,
      "learning_rate": 2.690270651082638e-06,
      "loss": 0.8688,
      "step": 4587
    },
    {
      "epoch": 0.7692339935030913,
      "grad_norm": 0.47656574845314026,
      "learning_rate": 2.686565863783681e-06,
      "loss": 0.8774,
      "step": 4588
    },
    {
      "epoch": 0.7694016556638373,
      "grad_norm": 0.506759762763977,
      "learning_rate": 2.68286323330701e-06,
      "loss": 0.934,
      "step": 4589
    },
    {
      "epoch": 0.7695693178245835,
      "grad_norm": 0.42268508672714233,
      "learning_rate": 2.67916276074458e-06,
      "loss": 0.8518,
      "step": 4590
    },
    {
      "epoch": 0.7697369799853295,
      "grad_norm": 0.7668575048446655,
      "learning_rate": 2.6754644471877054e-06,
      "loss": 0.8149,
      "step": 4591
    },
    {
      "epoch": 0.7699046421460757,
      "grad_norm": 0.6320164799690247,
      "learning_rate": 2.6717682937270605e-06,
      "loss": 0.8799,
      "step": 4592
    },
    {
      "epoch": 0.7700723043068217,
      "grad_norm": 0.7159294486045837,
      "learning_rate": 2.668074301452693e-06,
      "loss": 0.8879,
      "step": 4593
    },
    {
      "epoch": 0.7702399664675679,
      "grad_norm": 0.45723140239715576,
      "learning_rate": 2.6643824714540002e-06,
      "loss": 0.8416,
      "step": 4594
    },
    {
      "epoch": 0.7704076286283139,
      "grad_norm": 0.3924510180950165,
      "learning_rate": 2.6606928048197555e-06,
      "loss": 0.8389,
      "step": 4595
    },
    {
      "epoch": 0.7705752907890601,
      "grad_norm": 0.4730377197265625,
      "learning_rate": 2.657005302638085e-06,
      "loss": 0.8408,
      "step": 4596
    },
    {
      "epoch": 0.7707429529498061,
      "grad_norm": 0.4658554196357727,
      "learning_rate": 2.6533199659964737e-06,
      "loss": 0.8246,
      "step": 4597
    },
    {
      "epoch": 0.7709106151105523,
      "grad_norm": 0.7776197195053101,
      "learning_rate": 2.649636795981777e-06,
      "loss": 0.8719,
      "step": 4598
    },
    {
      "epoch": 0.7710782772712983,
      "grad_norm": 0.5712535381317139,
      "learning_rate": 2.6459557936802104e-06,
      "loss": 0.8904,
      "step": 4599
    },
    {
      "epoch": 0.7712459394320444,
      "grad_norm": 0.665768027305603,
      "learning_rate": 2.6422769601773437e-06,
      "loss": 0.7471,
      "step": 4600
    },
    {
      "epoch": 0.7714136015927905,
      "grad_norm": 0.49118784070014954,
      "learning_rate": 2.638600296558108e-06,
      "loss": 0.8099,
      "step": 4601
    },
    {
      "epoch": 0.7715812637535366,
      "grad_norm": 0.4034315347671509,
      "learning_rate": 2.6349258039068026e-06,
      "loss": 0.7578,
      "step": 4602
    },
    {
      "epoch": 0.7717489259142827,
      "grad_norm": 0.6553141474723816,
      "learning_rate": 2.6312534833070745e-06,
      "loss": 0.8429,
      "step": 4603
    },
    {
      "epoch": 0.7719165880750288,
      "grad_norm": 0.4088524878025055,
      "learning_rate": 2.6275833358419447e-06,
      "loss": 0.8106,
      "step": 4604
    },
    {
      "epoch": 0.7720842502357749,
      "grad_norm": 0.533658504486084,
      "learning_rate": 2.6239153625937786e-06,
      "loss": 0.8721,
      "step": 4605
    },
    {
      "epoch": 0.772251912396521,
      "grad_norm": 0.46169909834861755,
      "learning_rate": 2.620249564644307e-06,
      "loss": 0.8185,
      "step": 4606
    },
    {
      "epoch": 0.7724195745572671,
      "grad_norm": 0.5614885687828064,
      "learning_rate": 2.616585943074623e-06,
      "loss": 0.8561,
      "step": 4607
    },
    {
      "epoch": 0.7725872367180132,
      "grad_norm": 0.36912161111831665,
      "learning_rate": 2.612924498965169e-06,
      "loss": 0.9491,
      "step": 4608
    },
    {
      "epoch": 0.7727548988787593,
      "grad_norm": 0.7435131072998047,
      "learning_rate": 2.609265233395757e-06,
      "loss": 0.778,
      "step": 4609
    },
    {
      "epoch": 0.7729225610395054,
      "grad_norm": 0.4584323465824127,
      "learning_rate": 2.605608147445544e-06,
      "loss": 0.8414,
      "step": 4610
    },
    {
      "epoch": 0.7730902232002514,
      "grad_norm": 0.5063695311546326,
      "learning_rate": 2.6019532421930505e-06,
      "loss": 0.747,
      "step": 4611
    },
    {
      "epoch": 0.7732578853609976,
      "grad_norm": 0.40801185369491577,
      "learning_rate": 2.598300518716156e-06,
      "loss": 0.9518,
      "step": 4612
    },
    {
      "epoch": 0.7734255475217436,
      "grad_norm": 0.4017718732357025,
      "learning_rate": 2.594649978092093e-06,
      "loss": 0.8631,
      "step": 4613
    },
    {
      "epoch": 0.7735932096824898,
      "grad_norm": 0.37917450070381165,
      "learning_rate": 2.5910016213974475e-06,
      "loss": 0.8841,
      "step": 4614
    },
    {
      "epoch": 0.7737608718432358,
      "grad_norm": 0.5232954621315002,
      "learning_rate": 2.587355449708171e-06,
      "loss": 0.7955,
      "step": 4615
    },
    {
      "epoch": 0.773928534003982,
      "grad_norm": 0.45333942770957947,
      "learning_rate": 2.58371146409956e-06,
      "loss": 0.7909,
      "step": 4616
    },
    {
      "epoch": 0.774096196164728,
      "grad_norm": 0.34033146500587463,
      "learning_rate": 2.580069665646271e-06,
      "loss": 0.7876,
      "step": 4617
    },
    {
      "epoch": 0.7742638583254742,
      "grad_norm": 0.40777915716171265,
      "learning_rate": 2.576430055422324e-06,
      "loss": 0.934,
      "step": 4618
    },
    {
      "epoch": 0.7744315204862202,
      "grad_norm": 0.3873259425163269,
      "learning_rate": 2.5727926345010724e-06,
      "loss": 0.8527,
      "step": 4619
    },
    {
      "epoch": 0.7745991826469664,
      "grad_norm": 0.4247393012046814,
      "learning_rate": 2.5691574039552423e-06,
      "loss": 0.7864,
      "step": 4620
    },
    {
      "epoch": 0.7747668448077124,
      "grad_norm": 0.9930641055107117,
      "learning_rate": 2.5655243648569116e-06,
      "loss": 0.7773,
      "step": 4621
    },
    {
      "epoch": 0.7749345069684586,
      "grad_norm": 0.48081302642822266,
      "learning_rate": 2.561893518277506e-06,
      "loss": 0.8871,
      "step": 4622
    },
    {
      "epoch": 0.7751021691292046,
      "grad_norm": 1.0073636770248413,
      "learning_rate": 2.558264865287806e-06,
      "loss": 0.9174,
      "step": 4623
    },
    {
      "epoch": 0.7752698312899508,
      "grad_norm": 0.6616734266281128,
      "learning_rate": 2.5546384069579446e-06,
      "loss": 0.8713,
      "step": 4624
    },
    {
      "epoch": 0.7754374934506968,
      "grad_norm": 0.43992239236831665,
      "learning_rate": 2.55101414435741e-06,
      "loss": 0.86,
      "step": 4625
    },
    {
      "epoch": 0.7756051556114429,
      "grad_norm": 0.40123677253723145,
      "learning_rate": 2.547392078555048e-06,
      "loss": 0.8038,
      "step": 4626
    },
    {
      "epoch": 0.7757728177721891,
      "grad_norm": 0.4202883243560791,
      "learning_rate": 2.5437722106190454e-06,
      "loss": 0.7753,
      "step": 4627
    },
    {
      "epoch": 0.7759404799329351,
      "grad_norm": 0.6338961720466614,
      "learning_rate": 2.540154541616945e-06,
      "loss": 0.8622,
      "step": 4628
    },
    {
      "epoch": 0.7761081420936813,
      "grad_norm": 0.7529990077018738,
      "learning_rate": 2.5365390726156456e-06,
      "loss": 0.9001,
      "step": 4629
    },
    {
      "epoch": 0.7762758042544273,
      "grad_norm": 0.517359733581543,
      "learning_rate": 2.532925804681391e-06,
      "loss": 0.849,
      "step": 4630
    },
    {
      "epoch": 0.7764434664151735,
      "grad_norm": 0.4139972925186157,
      "learning_rate": 2.5293147388797813e-06,
      "loss": 0.787,
      "step": 4631
    },
    {
      "epoch": 0.7766111285759195,
      "grad_norm": 0.40291252732276917,
      "learning_rate": 2.5257058762757647e-06,
      "loss": 0.782,
      "step": 4632
    },
    {
      "epoch": 0.7767787907366657,
      "grad_norm": 0.4477655589580536,
      "learning_rate": 2.5220992179336344e-06,
      "loss": 0.8255,
      "step": 4633
    },
    {
      "epoch": 0.7769464528974117,
      "grad_norm": 0.5491387248039246,
      "learning_rate": 2.5184947649170455e-06,
      "loss": 0.7886,
      "step": 4634
    },
    {
      "epoch": 0.7771141150581579,
      "grad_norm": 0.40593987703323364,
      "learning_rate": 2.514892518288988e-06,
      "loss": 0.8754,
      "step": 4635
    },
    {
      "epoch": 0.7772817772189039,
      "grad_norm": 0.5853688716888428,
      "learning_rate": 2.511292479111819e-06,
      "loss": 0.8264,
      "step": 4636
    },
    {
      "epoch": 0.77744943937965,
      "grad_norm": 0.3623380959033966,
      "learning_rate": 2.5076946484472277e-06,
      "loss": 0.8025,
      "step": 4637
    },
    {
      "epoch": 0.7776171015403961,
      "grad_norm": 0.37506017088890076,
      "learning_rate": 2.5040990273562572e-06,
      "loss": 0.7539,
      "step": 4638
    },
    {
      "epoch": 0.7777847637011422,
      "grad_norm": 0.36650562286376953,
      "learning_rate": 2.500505616899307e-06,
      "loss": 0.8193,
      "step": 4639
    },
    {
      "epoch": 0.7779524258618883,
      "grad_norm": 0.5781775116920471,
      "learning_rate": 2.496914418136113e-06,
      "loss": 0.9064,
      "step": 4640
    },
    {
      "epoch": 0.7781200880226344,
      "grad_norm": 0.46714088320732117,
      "learning_rate": 2.4933254321257694e-06,
      "loss": 0.8181,
      "step": 4641
    },
    {
      "epoch": 0.7782877501833805,
      "grad_norm": 0.5534514784812927,
      "learning_rate": 2.489738659926708e-06,
      "loss": 0.9225,
      "step": 4642
    },
    {
      "epoch": 0.7784554123441266,
      "grad_norm": 0.4399142861366272,
      "learning_rate": 2.486154102596713e-06,
      "loss": 0.8527,
      "step": 4643
    },
    {
      "epoch": 0.7786230745048727,
      "grad_norm": 0.4053286910057068,
      "learning_rate": 2.4825717611929144e-06,
      "loss": 0.8012,
      "step": 4644
    },
    {
      "epoch": 0.7787907366656188,
      "grad_norm": 0.5702162981033325,
      "learning_rate": 2.478991636771797e-06,
      "loss": 0.8388,
      "step": 4645
    },
    {
      "epoch": 0.7789583988263649,
      "grad_norm": 0.45069995522499084,
      "learning_rate": 2.475413730389171e-06,
      "loss": 0.9109,
      "step": 4646
    },
    {
      "epoch": 0.779126060987111,
      "grad_norm": 0.4615761935710907,
      "learning_rate": 2.471838043100211e-06,
      "loss": 0.8515,
      "step": 4647
    },
    {
      "epoch": 0.779293723147857,
      "grad_norm": 0.44282227754592896,
      "learning_rate": 2.468264575959436e-06,
      "loss": 0.9149,
      "step": 4648
    },
    {
      "epoch": 0.7794613853086032,
      "grad_norm": 0.45196306705474854,
      "learning_rate": 2.4646933300206977e-06,
      "loss": 0.8429,
      "step": 4649
    },
    {
      "epoch": 0.7796290474693492,
      "grad_norm": 0.39488324522972107,
      "learning_rate": 2.461124306337207e-06,
      "loss": 0.8723,
      "step": 4650
    },
    {
      "epoch": 0.7797967096300954,
      "grad_norm": 0.5315574407577515,
      "learning_rate": 2.457557505961512e-06,
      "loss": 0.9138,
      "step": 4651
    },
    {
      "epoch": 0.7799643717908414,
      "grad_norm": 0.5758631229400635,
      "learning_rate": 2.4539929299455024e-06,
      "loss": 0.8382,
      "step": 4652
    },
    {
      "epoch": 0.7801320339515876,
      "grad_norm": 0.6532036662101746,
      "learning_rate": 2.4504305793404204e-06,
      "loss": 0.8348,
      "step": 4653
    },
    {
      "epoch": 0.7802996961123336,
      "grad_norm": 0.4632951021194458,
      "learning_rate": 2.446870455196847e-06,
      "loss": 0.9538,
      "step": 4654
    },
    {
      "epoch": 0.7804673582730798,
      "grad_norm": 0.5184860825538635,
      "learning_rate": 2.443312558564701e-06,
      "loss": 0.8301,
      "step": 4655
    },
    {
      "epoch": 0.7806350204338258,
      "grad_norm": 0.42855778336524963,
      "learning_rate": 2.439756890493258e-06,
      "loss": 0.9034,
      "step": 4656
    },
    {
      "epoch": 0.780802682594572,
      "grad_norm": 0.6073377132415771,
      "learning_rate": 2.4362034520311216e-06,
      "loss": 0.8989,
      "step": 4657
    },
    {
      "epoch": 0.780970344755318,
      "grad_norm": 0.5297231078147888,
      "learning_rate": 2.4326522442262522e-06,
      "loss": 0.9485,
      "step": 4658
    },
    {
      "epoch": 0.7811380069160642,
      "grad_norm": 0.45657971501350403,
      "learning_rate": 2.429103268125941e-06,
      "loss": 0.8522,
      "step": 4659
    },
    {
      "epoch": 0.7813056690768102,
      "grad_norm": 0.3763165771961212,
      "learning_rate": 2.425556524776821e-06,
      "loss": 0.8281,
      "step": 4660
    },
    {
      "epoch": 0.7814733312375564,
      "grad_norm": 0.35260340571403503,
      "learning_rate": 2.4220120152248783e-06,
      "loss": 0.7833,
      "step": 4661
    },
    {
      "epoch": 0.7816409933983024,
      "grad_norm": 0.8353078961372375,
      "learning_rate": 2.418469740515427e-06,
      "loss": 0.8843,
      "step": 4662
    },
    {
      "epoch": 0.7818086555590485,
      "grad_norm": 0.4255785644054413,
      "learning_rate": 2.4149297016931317e-06,
      "loss": 0.8756,
      "step": 4663
    },
    {
      "epoch": 0.7819763177197946,
      "grad_norm": 0.446634978055954,
      "learning_rate": 2.4113918998019927e-06,
      "loss": 0.8451,
      "step": 4664
    },
    {
      "epoch": 0.7821439798805407,
      "grad_norm": 1.0861172676086426,
      "learning_rate": 2.4078563358853467e-06,
      "loss": 0.879,
      "step": 4665
    },
    {
      "epoch": 0.7823116420412868,
      "grad_norm": 0.6283324360847473,
      "learning_rate": 2.4043230109858805e-06,
      "loss": 0.8991,
      "step": 4666
    },
    {
      "epoch": 0.7824793042020329,
      "grad_norm": 0.3723627030849457,
      "learning_rate": 2.400791926145617e-06,
      "loss": 0.8549,
      "step": 4667
    },
    {
      "epoch": 0.782646966362779,
      "grad_norm": 0.37841206789016724,
      "learning_rate": 2.397263082405914e-06,
      "loss": 0.74,
      "step": 4668
    },
    {
      "epoch": 0.7828146285235251,
      "grad_norm": 0.41133737564086914,
      "learning_rate": 2.3937364808074683e-06,
      "loss": 0.8694,
      "step": 4669
    },
    {
      "epoch": 0.7829822906842712,
      "grad_norm": 0.4194008409976959,
      "learning_rate": 2.390212122390323e-06,
      "loss": 0.8062,
      "step": 4670
    },
    {
      "epoch": 0.7831499528450173,
      "grad_norm": 0.5233654379844666,
      "learning_rate": 2.3866900081938517e-06,
      "loss": 0.7899,
      "step": 4671
    },
    {
      "epoch": 0.7833176150057634,
      "grad_norm": 0.48317602276802063,
      "learning_rate": 2.383170139256774e-06,
      "loss": 0.8615,
      "step": 4672
    },
    {
      "epoch": 0.7834852771665095,
      "grad_norm": 0.7062588334083557,
      "learning_rate": 2.37965251661714e-06,
      "loss": 0.817,
      "step": 4673
    },
    {
      "epoch": 0.7836529393272555,
      "grad_norm": 0.42143264412879944,
      "learning_rate": 2.376137141312338e-06,
      "loss": 0.7995,
      "step": 4674
    },
    {
      "epoch": 0.7838206014880017,
      "grad_norm": 0.42928212881088257,
      "learning_rate": 2.372624014379099e-06,
      "loss": 0.8712,
      "step": 4675
    },
    {
      "epoch": 0.7839882636487477,
      "grad_norm": 0.3896276652812958,
      "learning_rate": 2.3691131368534858e-06,
      "loss": 0.8254,
      "step": 4676
    },
    {
      "epoch": 0.7841559258094939,
      "grad_norm": 0.4230552911758423,
      "learning_rate": 2.365604509770901e-06,
      "loss": 0.7947,
      "step": 4677
    },
    {
      "epoch": 0.7843235879702399,
      "grad_norm": 0.41597506403923035,
      "learning_rate": 2.3620981341660833e-06,
      "loss": 0.8989,
      "step": 4678
    },
    {
      "epoch": 0.7844912501309861,
      "grad_norm": 0.680502712726593,
      "learning_rate": 2.3585940110731008e-06,
      "loss": 0.8588,
      "step": 4679
    },
    {
      "epoch": 0.7846589122917321,
      "grad_norm": 0.4649544954299927,
      "learning_rate": 2.3550921415253704e-06,
      "loss": 0.9314,
      "step": 4680
    },
    {
      "epoch": 0.7848265744524783,
      "grad_norm": 0.4067036807537079,
      "learning_rate": 2.3515925265556293e-06,
      "loss": 0.7538,
      "step": 4681
    },
    {
      "epoch": 0.7849942366132243,
      "grad_norm": 0.3940337002277374,
      "learning_rate": 2.348095167195964e-06,
      "loss": 0.83,
      "step": 4682
    },
    {
      "epoch": 0.7851618987739705,
      "grad_norm": 0.502957820892334,
      "learning_rate": 2.3446000644777856e-06,
      "loss": 0.8394,
      "step": 4683
    },
    {
      "epoch": 0.7853295609347165,
      "grad_norm": 0.5205345749855042,
      "learning_rate": 2.341107219431842e-06,
      "loss": 0.8014,
      "step": 4684
    },
    {
      "epoch": 0.7854972230954627,
      "grad_norm": 0.4484470784664154,
      "learning_rate": 2.337616633088219e-06,
      "loss": 0.8719,
      "step": 4685
    },
    {
      "epoch": 0.7856648852562087,
      "grad_norm": 0.4280824661254883,
      "learning_rate": 2.3341283064763343e-06,
      "loss": 0.8764,
      "step": 4686
    },
    {
      "epoch": 0.7858325474169549,
      "grad_norm": 0.4743936061859131,
      "learning_rate": 2.330642240624933e-06,
      "loss": 0.8052,
      "step": 4687
    },
    {
      "epoch": 0.7860002095777009,
      "grad_norm": 0.5149183869361877,
      "learning_rate": 2.327158436562107e-06,
      "loss": 0.826,
      "step": 4688
    },
    {
      "epoch": 0.786167871738447,
      "grad_norm": 0.42814549803733826,
      "learning_rate": 2.3236768953152665e-06,
      "loss": 0.8015,
      "step": 4689
    },
    {
      "epoch": 0.7863355338991931,
      "grad_norm": 0.49067017436027527,
      "learning_rate": 2.320197617911163e-06,
      "loss": 0.7903,
      "step": 4690
    },
    {
      "epoch": 0.7865031960599392,
      "grad_norm": 0.3798750936985016,
      "learning_rate": 2.316720605375886e-06,
      "loss": 0.8719,
      "step": 4691
    },
    {
      "epoch": 0.7866708582206853,
      "grad_norm": 1.520694613456726,
      "learning_rate": 2.313245858734838e-06,
      "loss": 0.8271,
      "step": 4692
    },
    {
      "epoch": 0.7868385203814314,
      "grad_norm": 0.452315092086792,
      "learning_rate": 2.30977337901277e-06,
      "loss": 0.8206,
      "step": 4693
    },
    {
      "epoch": 0.7870061825421775,
      "grad_norm": 0.3753838837146759,
      "learning_rate": 2.306303167233761e-06,
      "loss": 0.8666,
      "step": 4694
    },
    {
      "epoch": 0.7871738447029236,
      "grad_norm": 0.4556758403778076,
      "learning_rate": 2.3028352244212173e-06,
      "loss": 0.8371,
      "step": 4695
    },
    {
      "epoch": 0.7873415068636697,
      "grad_norm": 0.34695783257484436,
      "learning_rate": 2.2993695515978767e-06,
      "loss": 0.8478,
      "step": 4696
    },
    {
      "epoch": 0.7875091690244158,
      "grad_norm": 0.3780613839626312,
      "learning_rate": 2.2959061497858125e-06,
      "loss": 0.7419,
      "step": 4697
    },
    {
      "epoch": 0.7876768311851619,
      "grad_norm": 0.413053959608078,
      "learning_rate": 2.292445020006422e-06,
      "loss": 0.7941,
      "step": 4698
    },
    {
      "epoch": 0.787844493345908,
      "grad_norm": 0.5531803965568542,
      "learning_rate": 2.2889861632804377e-06,
      "loss": 0.8418,
      "step": 4699
    },
    {
      "epoch": 0.788012155506654,
      "grad_norm": 0.4788338541984558,
      "learning_rate": 2.2855295806279188e-06,
      "loss": 0.9246,
      "step": 4700
    },
    {
      "epoch": 0.7881798176674002,
      "grad_norm": 0.42819738388061523,
      "learning_rate": 2.2820752730682494e-06,
      "loss": 0.8493,
      "step": 4701
    },
    {
      "epoch": 0.7883474798281462,
      "grad_norm": 0.504607081413269,
      "learning_rate": 2.2786232416201548e-06,
      "loss": 0.8319,
      "step": 4702
    },
    {
      "epoch": 0.7885151419888924,
      "grad_norm": 0.3961004912853241,
      "learning_rate": 2.275173487301676e-06,
      "loss": 0.7408,
      "step": 4703
    },
    {
      "epoch": 0.7886828041496384,
      "grad_norm": 0.42878851294517517,
      "learning_rate": 2.2717260111301943e-06,
      "loss": 0.7563,
      "step": 4704
    },
    {
      "epoch": 0.7888504663103846,
      "grad_norm": 0.5974430441856384,
      "learning_rate": 2.268280814122409e-06,
      "loss": 0.8772,
      "step": 4705
    },
    {
      "epoch": 0.7890181284711306,
      "grad_norm": 0.3592509925365448,
      "learning_rate": 2.2648378972943495e-06,
      "loss": 0.8229,
      "step": 4706
    },
    {
      "epoch": 0.7891857906318768,
      "grad_norm": 0.9069346189498901,
      "learning_rate": 2.26139726166138e-06,
      "loss": 0.9157,
      "step": 4707
    },
    {
      "epoch": 0.7893534527926228,
      "grad_norm": 0.3409862220287323,
      "learning_rate": 2.2579589082381813e-06,
      "loss": 0.8402,
      "step": 4708
    },
    {
      "epoch": 0.789521114953369,
      "grad_norm": 0.9539614915847778,
      "learning_rate": 2.2545228380387706e-06,
      "loss": 0.8716,
      "step": 4709
    },
    {
      "epoch": 0.789688777114115,
      "grad_norm": 0.9350457787513733,
      "learning_rate": 2.251089052076487e-06,
      "loss": 0.822,
      "step": 4710
    },
    {
      "epoch": 0.7898564392748612,
      "grad_norm": 0.6432152390480042,
      "learning_rate": 2.247657551363992e-06,
      "loss": 0.8366,
      "step": 4711
    },
    {
      "epoch": 0.7900241014356073,
      "grad_norm": 0.6478852033615112,
      "learning_rate": 2.24422833691328e-06,
      "loss": 0.8435,
      "step": 4712
    },
    {
      "epoch": 0.7901917635963533,
      "grad_norm": 0.4530707597732544,
      "learning_rate": 2.2408014097356732e-06,
      "loss": 0.8305,
      "step": 4713
    },
    {
      "epoch": 0.7903594257570995,
      "grad_norm": 0.7520490288734436,
      "learning_rate": 2.2373767708418116e-06,
      "loss": 0.9746,
      "step": 4714
    },
    {
      "epoch": 0.7905270879178455,
      "grad_norm": 0.40117794275283813,
      "learning_rate": 2.233954421241661e-06,
      "loss": 0.8815,
      "step": 4715
    },
    {
      "epoch": 0.7906947500785917,
      "grad_norm": 0.5994216799736023,
      "learning_rate": 2.2305343619445185e-06,
      "loss": 0.9301,
      "step": 4716
    },
    {
      "epoch": 0.7908624122393377,
      "grad_norm": 0.4644392430782318,
      "learning_rate": 2.227116593958999e-06,
      "loss": 0.7981,
      "step": 4717
    },
    {
      "epoch": 0.7910300744000839,
      "grad_norm": 0.4093725085258484,
      "learning_rate": 2.223701118293048e-06,
      "loss": 0.8982,
      "step": 4718
    },
    {
      "epoch": 0.7911977365608299,
      "grad_norm": 0.8823078870773315,
      "learning_rate": 2.22028793595393e-06,
      "loss": 0.9033,
      "step": 4719
    },
    {
      "epoch": 0.7913653987215761,
      "grad_norm": 0.3445841372013092,
      "learning_rate": 2.2168770479482315e-06,
      "loss": 0.8072,
      "step": 4720
    },
    {
      "epoch": 0.7915330608823221,
      "grad_norm": 0.3272508680820465,
      "learning_rate": 2.213468455281872e-06,
      "loss": 0.7164,
      "step": 4721
    },
    {
      "epoch": 0.7917007230430683,
      "grad_norm": 0.411796510219574,
      "learning_rate": 2.2100621589600813e-06,
      "loss": 0.948,
      "step": 4722
    },
    {
      "epoch": 0.7918683852038143,
      "grad_norm": 0.45838311314582825,
      "learning_rate": 2.206658159987424e-06,
      "loss": 0.8372,
      "step": 4723
    },
    {
      "epoch": 0.7920360473645605,
      "grad_norm": 0.7174457907676697,
      "learning_rate": 2.2032564593677773e-06,
      "loss": 0.7418,
      "step": 4724
    },
    {
      "epoch": 0.7922037095253065,
      "grad_norm": 0.8203630447387695,
      "learning_rate": 2.1998570581043453e-06,
      "loss": 0.8728,
      "step": 4725
    },
    {
      "epoch": 0.7923713716860527,
      "grad_norm": 0.6109965443611145,
      "learning_rate": 2.196459957199657e-06,
      "loss": 0.7454,
      "step": 4726
    },
    {
      "epoch": 0.7925390338467987,
      "grad_norm": 0.6313849687576294,
      "learning_rate": 2.193065157655556e-06,
      "loss": 0.8987,
      "step": 4727
    },
    {
      "epoch": 0.7927066960075448,
      "grad_norm": 0.7500846982002258,
      "learning_rate": 2.189672660473211e-06,
      "loss": 0.8722,
      "step": 4728
    },
    {
      "epoch": 0.7928743581682909,
      "grad_norm": 0.4414680302143097,
      "learning_rate": 2.1862824666531135e-06,
      "loss": 0.7453,
      "step": 4729
    },
    {
      "epoch": 0.793042020329037,
      "grad_norm": 0.4681912958621979,
      "learning_rate": 2.1828945771950704e-06,
      "loss": 0.7655,
      "step": 4730
    },
    {
      "epoch": 0.7932096824897831,
      "grad_norm": 0.4023488163948059,
      "learning_rate": 2.179508993098214e-06,
      "loss": 0.861,
      "step": 4731
    },
    {
      "epoch": 0.7933773446505292,
      "grad_norm": 0.47425827383995056,
      "learning_rate": 2.1761257153610004e-06,
      "loss": 0.9054,
      "step": 4732
    },
    {
      "epoch": 0.7935450068112753,
      "grad_norm": 0.3306763470172882,
      "learning_rate": 2.17274474498119e-06,
      "loss": 0.8637,
      "step": 4733
    },
    {
      "epoch": 0.7937126689720214,
      "grad_norm": 0.3470577001571655,
      "learning_rate": 2.1693660829558804e-06,
      "loss": 0.819,
      "step": 4734
    },
    {
      "epoch": 0.7938803311327675,
      "grad_norm": 0.4094432294368744,
      "learning_rate": 2.165989730281475e-06,
      "loss": 0.7996,
      "step": 4735
    },
    {
      "epoch": 0.7940479932935136,
      "grad_norm": 0.4102506935596466,
      "learning_rate": 2.1626156879537084e-06,
      "loss": 0.7098,
      "step": 4736
    },
    {
      "epoch": 0.7942156554542597,
      "grad_norm": 0.5132640600204468,
      "learning_rate": 2.159243956967624e-06,
      "loss": 0.8522,
      "step": 4737
    },
    {
      "epoch": 0.7943833176150058,
      "grad_norm": 0.755061686038971,
      "learning_rate": 2.155874538317585e-06,
      "loss": 0.7889,
      "step": 4738
    },
    {
      "epoch": 0.7945509797757518,
      "grad_norm": 0.6049810647964478,
      "learning_rate": 2.152507432997277e-06,
      "loss": 0.8063,
      "step": 4739
    },
    {
      "epoch": 0.794718641936498,
      "grad_norm": 0.5153598785400391,
      "learning_rate": 2.149142641999703e-06,
      "loss": 0.7808,
      "step": 4740
    },
    {
      "epoch": 0.794886304097244,
      "grad_norm": 0.3766288757324219,
      "learning_rate": 2.1457801663171805e-06,
      "loss": 0.827,
      "step": 4741
    },
    {
      "epoch": 0.7950539662579902,
      "grad_norm": 0.6483139991760254,
      "learning_rate": 2.1424200069413416e-06,
      "loss": 0.8916,
      "step": 4742
    },
    {
      "epoch": 0.7952216284187362,
      "grad_norm": 0.38120341300964355,
      "learning_rate": 2.139062164863144e-06,
      "loss": 0.768,
      "step": 4743
    },
    {
      "epoch": 0.7953892905794824,
      "grad_norm": 0.5252415537834167,
      "learning_rate": 2.135706641072851e-06,
      "loss": 0.7788,
      "step": 4744
    },
    {
      "epoch": 0.7955569527402284,
      "grad_norm": 0.3720047175884247,
      "learning_rate": 2.132353436560055e-06,
      "loss": 0.8189,
      "step": 4745
    },
    {
      "epoch": 0.7957246149009746,
      "grad_norm": 0.329374760389328,
      "learning_rate": 2.129002552313655e-06,
      "loss": 0.7773,
      "step": 4746
    },
    {
      "epoch": 0.7958922770617206,
      "grad_norm": 0.5027045011520386,
      "learning_rate": 2.125653989321863e-06,
      "loss": 0.8635,
      "step": 4747
    },
    {
      "epoch": 0.7960599392224668,
      "grad_norm": 0.6417919993400574,
      "learning_rate": 2.12230774857222e-06,
      "loss": 0.8453,
      "step": 4748
    },
    {
      "epoch": 0.7962276013832128,
      "grad_norm": 0.4022006690502167,
      "learning_rate": 2.118963831051567e-06,
      "loss": 0.8061,
      "step": 4749
    },
    {
      "epoch": 0.796395263543959,
      "grad_norm": 0.4296365976333618,
      "learning_rate": 2.1156222377460723e-06,
      "loss": 0.8482,
      "step": 4750
    },
    {
      "epoch": 0.796562925704705,
      "grad_norm": 0.28435319662094116,
      "learning_rate": 2.1122829696412108e-06,
      "loss": 0.8003,
      "step": 4751
    },
    {
      "epoch": 0.7967305878654511,
      "grad_norm": 0.4156350791454315,
      "learning_rate": 2.10894602772177e-06,
      "loss": 0.7549,
      "step": 4752
    },
    {
      "epoch": 0.7968982500261972,
      "grad_norm": 0.49609625339508057,
      "learning_rate": 2.105611412971863e-06,
      "loss": 0.8254,
      "step": 4753
    },
    {
      "epoch": 0.7970659121869433,
      "grad_norm": 0.4338054358959198,
      "learning_rate": 2.102279126374902e-06,
      "loss": 0.8666,
      "step": 4754
    },
    {
      "epoch": 0.7972335743476894,
      "grad_norm": 0.4368712604045868,
      "learning_rate": 2.098949168913624e-06,
      "loss": 0.8679,
      "step": 4755
    },
    {
      "epoch": 0.7974012365084355,
      "grad_norm": 0.4860408306121826,
      "learning_rate": 2.095621541570075e-06,
      "loss": 0.8792,
      "step": 4756
    },
    {
      "epoch": 0.7975688986691816,
      "grad_norm": 0.45234760642051697,
      "learning_rate": 2.092296245325609e-06,
      "loss": 0.7176,
      "step": 4757
    },
    {
      "epoch": 0.7977365608299277,
      "grad_norm": 0.487379789352417,
      "learning_rate": 2.0889732811608996e-06,
      "loss": 0.8306,
      "step": 4758
    },
    {
      "epoch": 0.7979042229906738,
      "grad_norm": 0.3871273100376129,
      "learning_rate": 2.085652650055935e-06,
      "loss": 0.7539,
      "step": 4759
    },
    {
      "epoch": 0.7980718851514199,
      "grad_norm": 0.47201892733573914,
      "learning_rate": 2.0823343529900007e-06,
      "loss": 0.78,
      "step": 4760
    },
    {
      "epoch": 0.798239547312166,
      "grad_norm": 0.5183556079864502,
      "learning_rate": 2.0790183909417096e-06,
      "loss": 0.9249,
      "step": 4761
    },
    {
      "epoch": 0.7984072094729121,
      "grad_norm": 0.4703180491924286,
      "learning_rate": 2.0757047648889816e-06,
      "loss": 0.8947,
      "step": 4762
    },
    {
      "epoch": 0.7985748716336581,
      "grad_norm": 0.3326199948787689,
      "learning_rate": 2.07239347580904e-06,
      "loss": 0.8183,
      "step": 4763
    },
    {
      "epoch": 0.7987425337944043,
      "grad_norm": 0.4855404496192932,
      "learning_rate": 2.069084524678432e-06,
      "loss": 0.7859,
      "step": 4764
    },
    {
      "epoch": 0.7989101959551503,
      "grad_norm": 0.5343421101570129,
      "learning_rate": 2.0657779124730036e-06,
      "loss": 0.8775,
      "step": 4765
    },
    {
      "epoch": 0.7990778581158965,
      "grad_norm": 0.43089139461517334,
      "learning_rate": 2.062473640167915e-06,
      "loss": 0.8366,
      "step": 4766
    },
    {
      "epoch": 0.7992455202766425,
      "grad_norm": 0.3698073923587799,
      "learning_rate": 2.0591717087376416e-06,
      "loss": 0.8894,
      "step": 4767
    },
    {
      "epoch": 0.7994131824373887,
      "grad_norm": 0.48625025153160095,
      "learning_rate": 2.0558721191559606e-06,
      "loss": 0.854,
      "step": 4768
    },
    {
      "epoch": 0.7995808445981347,
      "grad_norm": 0.6302156448364258,
      "learning_rate": 2.0525748723959595e-06,
      "loss": 0.826,
      "step": 4769
    },
    {
      "epoch": 0.7997485067588809,
      "grad_norm": 0.4241481125354767,
      "learning_rate": 2.049279969430044e-06,
      "loss": 0.8336,
      "step": 4770
    },
    {
      "epoch": 0.7999161689196269,
      "grad_norm": 0.3742135167121887,
      "learning_rate": 2.045987411229913e-06,
      "loss": 0.784,
      "step": 4771
    },
    {
      "epoch": 0.8000838310803731,
      "grad_norm": 0.6190921664237976,
      "learning_rate": 2.0426971987665912e-06,
      "loss": 0.8335,
      "step": 4772
    },
    {
      "epoch": 0.8002514932411191,
      "grad_norm": 0.8990462422370911,
      "learning_rate": 2.0394093330103992e-06,
      "loss": 0.9597,
      "step": 4773
    },
    {
      "epoch": 0.8004191554018653,
      "grad_norm": 0.4658777117729187,
      "learning_rate": 2.036123814930967e-06,
      "loss": 0.8922,
      "step": 4774
    },
    {
      "epoch": 0.8005868175626113,
      "grad_norm": 0.5698621869087219,
      "learning_rate": 2.0328406454972395e-06,
      "loss": 0.7925,
      "step": 4775
    },
    {
      "epoch": 0.8007544797233574,
      "grad_norm": 0.49836301803588867,
      "learning_rate": 2.0295598256774598e-06,
      "loss": 0.8592,
      "step": 4776
    },
    {
      "epoch": 0.8009221418841035,
      "grad_norm": 0.6130602359771729,
      "learning_rate": 2.0262813564391846e-06,
      "loss": 0.8279,
      "step": 4777
    },
    {
      "epoch": 0.8010898040448496,
      "grad_norm": 0.39174386858940125,
      "learning_rate": 2.023005238749276e-06,
      "loss": 0.7482,
      "step": 4778
    },
    {
      "epoch": 0.8012574662055957,
      "grad_norm": 0.5928496718406677,
      "learning_rate": 2.019731473573897e-06,
      "loss": 0.8215,
      "step": 4779
    },
    {
      "epoch": 0.8014251283663418,
      "grad_norm": 0.38793522119522095,
      "learning_rate": 2.0164600618785247e-06,
      "loss": 0.7668,
      "step": 4780
    },
    {
      "epoch": 0.8015927905270879,
      "grad_norm": 0.5155482888221741,
      "learning_rate": 2.013191004627941e-06,
      "loss": 0.9221,
      "step": 4781
    },
    {
      "epoch": 0.801760452687834,
      "grad_norm": 0.5527405738830566,
      "learning_rate": 2.0099243027862303e-06,
      "loss": 0.8017,
      "step": 4782
    },
    {
      "epoch": 0.8019281148485801,
      "grad_norm": 0.4220729470252991,
      "learning_rate": 2.0066599573167824e-06,
      "loss": 0.8613,
      "step": 4783
    },
    {
      "epoch": 0.8020957770093262,
      "grad_norm": 0.37018314003944397,
      "learning_rate": 2.003397969182289e-06,
      "loss": 0.8766,
      "step": 4784
    },
    {
      "epoch": 0.8022634391700723,
      "grad_norm": 0.5372514128684998,
      "learning_rate": 2.000138339344756e-06,
      "loss": 0.8321,
      "step": 4785
    },
    {
      "epoch": 0.8024311013308184,
      "grad_norm": 0.4727790951728821,
      "learning_rate": 1.9968810687654893e-06,
      "loss": 0.8363,
      "step": 4786
    },
    {
      "epoch": 0.8025987634915644,
      "grad_norm": 0.3786468207836151,
      "learning_rate": 1.9936261584050974e-06,
      "loss": 0.8643,
      "step": 4787
    },
    {
      "epoch": 0.8027664256523106,
      "grad_norm": 0.5220190286636353,
      "learning_rate": 1.9903736092234904e-06,
      "loss": 0.8608,
      "step": 4788
    },
    {
      "epoch": 0.8029340878130566,
      "grad_norm": 0.3876339793205261,
      "learning_rate": 1.9871234221798895e-06,
      "loss": 0.7836,
      "step": 4789
    },
    {
      "epoch": 0.8031017499738028,
      "grad_norm": 0.37965089082717896,
      "learning_rate": 1.9838755982328108e-06,
      "loss": 0.81,
      "step": 4790
    },
    {
      "epoch": 0.8032694121345488,
      "grad_norm": 0.7638852596282959,
      "learning_rate": 1.9806301383400837e-06,
      "loss": 0.9099,
      "step": 4791
    },
    {
      "epoch": 0.803437074295295,
      "grad_norm": 0.9383553862571716,
      "learning_rate": 1.9773870434588306e-06,
      "loss": 0.8597,
      "step": 4792
    },
    {
      "epoch": 0.803604736456041,
      "grad_norm": 0.48865485191345215,
      "learning_rate": 1.9741463145454786e-06,
      "loss": 0.8666,
      "step": 4793
    },
    {
      "epoch": 0.8037723986167872,
      "grad_norm": 0.36044055223464966,
      "learning_rate": 1.970907952555764e-06,
      "loss": 0.8973,
      "step": 4794
    },
    {
      "epoch": 0.8039400607775332,
      "grad_norm": 0.4043533504009247,
      "learning_rate": 1.9676719584447134e-06,
      "loss": 0.8613,
      "step": 4795
    },
    {
      "epoch": 0.8041077229382794,
      "grad_norm": 0.3273265063762665,
      "learning_rate": 1.9644383331666684e-06,
      "loss": 0.8955,
      "step": 4796
    },
    {
      "epoch": 0.8042753850990254,
      "grad_norm": 0.3665297329425812,
      "learning_rate": 1.961207077675261e-06,
      "loss": 0.8371,
      "step": 4797
    },
    {
      "epoch": 0.8044430472597716,
      "grad_norm": 0.385436087846756,
      "learning_rate": 1.9579781929234275e-06,
      "loss": 0.8152,
      "step": 4798
    },
    {
      "epoch": 0.8046107094205177,
      "grad_norm": 0.5350459814071655,
      "learning_rate": 1.9547516798634104e-06,
      "loss": 0.8141,
      "step": 4799
    },
    {
      "epoch": 0.8047783715812638,
      "grad_norm": 0.8556716442108154,
      "learning_rate": 1.9515275394467446e-06,
      "loss": 0.825,
      "step": 4800
    },
    {
      "epoch": 0.8049460337420099,
      "grad_norm": 0.3931393623352051,
      "learning_rate": 1.9483057726242694e-06,
      "loss": 0.8643,
      "step": 4801
    },
    {
      "epoch": 0.8051136959027559,
      "grad_norm": 0.4177151024341583,
      "learning_rate": 1.945086380346126e-06,
      "loss": 0.8337,
      "step": 4802
    },
    {
      "epoch": 0.8052813580635021,
      "grad_norm": 0.4200320839881897,
      "learning_rate": 1.941869363561749e-06,
      "loss": 0.8256,
      "step": 4803
    },
    {
      "epoch": 0.8054490202242481,
      "grad_norm": 0.40648573637008667,
      "learning_rate": 1.9386547232198795e-06,
      "loss": 0.9559,
      "step": 4804
    },
    {
      "epoch": 0.8056166823849943,
      "grad_norm": 0.3557724356651306,
      "learning_rate": 1.9354424602685586e-06,
      "loss": 0.8458,
      "step": 4805
    },
    {
      "epoch": 0.8057843445457403,
      "grad_norm": 0.4115389883518219,
      "learning_rate": 1.932232575655113e-06,
      "loss": 0.8754,
      "step": 4806
    },
    {
      "epoch": 0.8059520067064865,
      "grad_norm": 0.4639323353767395,
      "learning_rate": 1.9290250703261824e-06,
      "loss": 0.9602,
      "step": 4807
    },
    {
      "epoch": 0.8061196688672325,
      "grad_norm": 0.42797258496284485,
      "learning_rate": 1.9258199452277037e-06,
      "loss": 0.8681,
      "step": 4808
    },
    {
      "epoch": 0.8062873310279787,
      "grad_norm": 0.49095794558525085,
      "learning_rate": 1.9226172013049027e-06,
      "loss": 0.8429,
      "step": 4809
    },
    {
      "epoch": 0.8064549931887247,
      "grad_norm": 0.4635695219039917,
      "learning_rate": 1.9194168395023083e-06,
      "loss": 0.7287,
      "step": 4810
    },
    {
      "epoch": 0.8066226553494709,
      "grad_norm": 0.45930927991867065,
      "learning_rate": 1.91621886076375e-06,
      "loss": 0.8617,
      "step": 4811
    },
    {
      "epoch": 0.8067903175102169,
      "grad_norm": 0.4171750247478485,
      "learning_rate": 1.913023266032348e-06,
      "loss": 0.8432,
      "step": 4812
    },
    {
      "epoch": 0.806957979670963,
      "grad_norm": 0.6069956421852112,
      "learning_rate": 1.9098300562505266e-06,
      "loss": 0.8255,
      "step": 4813
    },
    {
      "epoch": 0.8071256418317091,
      "grad_norm": 0.44642847776412964,
      "learning_rate": 1.9066392323600003e-06,
      "loss": 0.7873,
      "step": 4814
    },
    {
      "epoch": 0.8072933039924552,
      "grad_norm": 0.37293142080307007,
      "learning_rate": 1.903450795301781e-06,
      "loss": 0.7853,
      "step": 4815
    },
    {
      "epoch": 0.8074609661532013,
      "grad_norm": 0.5118458271026611,
      "learning_rate": 1.9002647460161826e-06,
      "loss": 0.8282,
      "step": 4816
    },
    {
      "epoch": 0.8076286283139474,
      "grad_norm": 0.4068104326725006,
      "learning_rate": 1.8970810854428046e-06,
      "loss": 0.8237,
      "step": 4817
    },
    {
      "epoch": 0.8077962904746935,
      "grad_norm": 0.4325851500034332,
      "learning_rate": 1.8938998145205557e-06,
      "loss": 0.8656,
      "step": 4818
    },
    {
      "epoch": 0.8079639526354396,
      "grad_norm": 0.39056482911109924,
      "learning_rate": 1.8907209341876266e-06,
      "loss": 0.8158,
      "step": 4819
    },
    {
      "epoch": 0.8081316147961857,
      "grad_norm": 0.4574761986732483,
      "learning_rate": 1.8875444453815073e-06,
      "loss": 0.7855,
      "step": 4820
    },
    {
      "epoch": 0.8082992769569318,
      "grad_norm": 0.776763379573822,
      "learning_rate": 1.8843703490389885e-06,
      "loss": 0.8233,
      "step": 4821
    },
    {
      "epoch": 0.8084669391176779,
      "grad_norm": 0.41815385222435,
      "learning_rate": 1.8811986460961461e-06,
      "loss": 0.7794,
      "step": 4822
    },
    {
      "epoch": 0.808634601278424,
      "grad_norm": 0.3735414743423462,
      "learning_rate": 1.8780293374883584e-06,
      "loss": 0.7637,
      "step": 4823
    },
    {
      "epoch": 0.80880226343917,
      "grad_norm": 0.39200180768966675,
      "learning_rate": 1.8748624241502934e-06,
      "loss": 0.7985,
      "step": 4824
    },
    {
      "epoch": 0.8089699255999162,
      "grad_norm": 0.4573250114917755,
      "learning_rate": 1.871697907015907e-06,
      "loss": 0.7891,
      "step": 4825
    },
    {
      "epoch": 0.8091375877606622,
      "grad_norm": 0.6612563729286194,
      "learning_rate": 1.8685357870184605e-06,
      "loss": 0.8541,
      "step": 4826
    },
    {
      "epoch": 0.8093052499214084,
      "grad_norm": 0.4323849081993103,
      "learning_rate": 1.8653760650905028e-06,
      "loss": 0.8265,
      "step": 4827
    },
    {
      "epoch": 0.8094729120821544,
      "grad_norm": 0.4029139280319214,
      "learning_rate": 1.8622187421638737e-06,
      "loss": 0.8441,
      "step": 4828
    },
    {
      "epoch": 0.8096405742429006,
      "grad_norm": 0.8146257400512695,
      "learning_rate": 1.8590638191697031e-06,
      "loss": 0.8966,
      "step": 4829
    },
    {
      "epoch": 0.8098082364036466,
      "grad_norm": 0.5496284365653992,
      "learning_rate": 1.855911297038422e-06,
      "loss": 0.8491,
      "step": 4830
    },
    {
      "epoch": 0.8099758985643928,
      "grad_norm": 0.5226850509643555,
      "learning_rate": 1.8527611766997444e-06,
      "loss": 0.7946,
      "step": 4831
    },
    {
      "epoch": 0.8101435607251388,
      "grad_norm": 0.4379395544528961,
      "learning_rate": 1.8496134590826864e-06,
      "loss": 0.8962,
      "step": 4832
    },
    {
      "epoch": 0.810311222885885,
      "grad_norm": 0.5736410617828369,
      "learning_rate": 1.8464681451155387e-06,
      "loss": 0.89,
      "step": 4833
    },
    {
      "epoch": 0.810478885046631,
      "grad_norm": 0.39776611328125,
      "learning_rate": 1.8433252357258991e-06,
      "loss": 0.8144,
      "step": 4834
    },
    {
      "epoch": 0.8106465472073772,
      "grad_norm": 0.4503728151321411,
      "learning_rate": 1.8401847318406518e-06,
      "loss": 0.8192,
      "step": 4835
    },
    {
      "epoch": 0.8108142093681232,
      "grad_norm": 0.42749375104904175,
      "learning_rate": 1.8370466343859672e-06,
      "loss": 0.8494,
      "step": 4836
    },
    {
      "epoch": 0.8109818715288694,
      "grad_norm": 0.42171910405158997,
      "learning_rate": 1.8339109442873115e-06,
      "loss": 0.8293,
      "step": 4837
    },
    {
      "epoch": 0.8111495336896154,
      "grad_norm": 0.49447721242904663,
      "learning_rate": 1.8307776624694383e-06,
      "loss": 0.8529,
      "step": 4838
    },
    {
      "epoch": 0.8113171958503615,
      "grad_norm": 0.4726193845272064,
      "learning_rate": 1.8276467898563887e-06,
      "loss": 0.8647,
      "step": 4839
    },
    {
      "epoch": 0.8114848580111076,
      "grad_norm": 0.9029093384742737,
      "learning_rate": 1.8245183273714984e-06,
      "loss": 0.8707,
      "step": 4840
    },
    {
      "epoch": 0.8116525201718537,
      "grad_norm": 0.528593897819519,
      "learning_rate": 1.8213922759373903e-06,
      "loss": 0.8982,
      "step": 4841
    },
    {
      "epoch": 0.8118201823325998,
      "grad_norm": 0.48823660612106323,
      "learning_rate": 1.8182686364759704e-06,
      "loss": 0.8779,
      "step": 4842
    },
    {
      "epoch": 0.8119878444933459,
      "grad_norm": 0.7718713879585266,
      "learning_rate": 1.8151474099084466e-06,
      "loss": 0.8608,
      "step": 4843
    },
    {
      "epoch": 0.812155506654092,
      "grad_norm": 0.4108054041862488,
      "learning_rate": 1.8120285971552987e-06,
      "loss": 0.8489,
      "step": 4844
    },
    {
      "epoch": 0.8123231688148381,
      "grad_norm": 0.5947515368461609,
      "learning_rate": 1.808912199136309e-06,
      "loss": 0.8154,
      "step": 4845
    },
    {
      "epoch": 0.8124908309755842,
      "grad_norm": 0.4202979505062103,
      "learning_rate": 1.8057982167705446e-06,
      "loss": 0.9065,
      "step": 4846
    },
    {
      "epoch": 0.8126584931363303,
      "grad_norm": 0.3972373902797699,
      "learning_rate": 1.8026866509763497e-06,
      "loss": 0.7442,
      "step": 4847
    },
    {
      "epoch": 0.8128261552970764,
      "grad_norm": 0.3535272479057312,
      "learning_rate": 1.799577502671368e-06,
      "loss": 0.8012,
      "step": 4848
    },
    {
      "epoch": 0.8129938174578225,
      "grad_norm": 0.3755205571651459,
      "learning_rate": 1.7964707727725238e-06,
      "loss": 0.7669,
      "step": 4849
    },
    {
      "epoch": 0.8131614796185686,
      "grad_norm": 0.6204691529273987,
      "learning_rate": 1.7933664621960334e-06,
      "loss": 0.8054,
      "step": 4850
    },
    {
      "epoch": 0.8133291417793147,
      "grad_norm": 0.45454415678977966,
      "learning_rate": 1.7902645718573953e-06,
      "loss": 0.7948,
      "step": 4851
    },
    {
      "epoch": 0.8134968039400607,
      "grad_norm": 0.6720665693283081,
      "learning_rate": 1.787165102671391e-06,
      "loss": 0.9038,
      "step": 4852
    },
    {
      "epoch": 0.8136644661008069,
      "grad_norm": 0.34496569633483887,
      "learning_rate": 1.7840680555520961e-06,
      "loss": 0.8538,
      "step": 4853
    },
    {
      "epoch": 0.8138321282615529,
      "grad_norm": 0.47709912061691284,
      "learning_rate": 1.7809734314128712e-06,
      "loss": 0.7695,
      "step": 4854
    },
    {
      "epoch": 0.8139997904222991,
      "grad_norm": 0.4746073782444,
      "learning_rate": 1.777881231166355e-06,
      "loss": 0.8022,
      "step": 4855
    },
    {
      "epoch": 0.8141674525830451,
      "grad_norm": 0.5646691918373108,
      "learning_rate": 1.774791455724475e-06,
      "loss": 0.7788,
      "step": 4856
    },
    {
      "epoch": 0.8143351147437913,
      "grad_norm": 0.4032534062862396,
      "learning_rate": 1.7717041059984486e-06,
      "loss": 0.7922,
      "step": 4857
    },
    {
      "epoch": 0.8145027769045373,
      "grad_norm": 1.140046238899231,
      "learning_rate": 1.7686191828987676e-06,
      "loss": 0.8125,
      "step": 4858
    },
    {
      "epoch": 0.8146704390652835,
      "grad_norm": 0.3570913076400757,
      "learning_rate": 1.7655366873352197e-06,
      "loss": 0.816,
      "step": 4859
    },
    {
      "epoch": 0.8148381012260295,
      "grad_norm": 0.36539798974990845,
      "learning_rate": 1.762456620216869e-06,
      "loss": 0.7864,
      "step": 4860
    },
    {
      "epoch": 0.8150057633867757,
      "grad_norm": 0.3946821987628937,
      "learning_rate": 1.7593789824520624e-06,
      "loss": 0.8419,
      "step": 4861
    },
    {
      "epoch": 0.8151734255475217,
      "grad_norm": 0.33802446722984314,
      "learning_rate": 1.756303774948439e-06,
      "loss": 0.8817,
      "step": 4862
    },
    {
      "epoch": 0.8153410877082679,
      "grad_norm": 0.41984841227531433,
      "learning_rate": 1.7532309986129092e-06,
      "loss": 0.7901,
      "step": 4863
    },
    {
      "epoch": 0.8155087498690139,
      "grad_norm": 0.416354775428772,
      "learning_rate": 1.7501606543516792e-06,
      "loss": 0.7435,
      "step": 4864
    },
    {
      "epoch": 0.81567641202976,
      "grad_norm": 0.4285713732242584,
      "learning_rate": 1.7470927430702277e-06,
      "loss": 0.873,
      "step": 4865
    },
    {
      "epoch": 0.8158440741905061,
      "grad_norm": 0.45582979917526245,
      "learning_rate": 1.7440272656733182e-06,
      "loss": 0.8108,
      "step": 4866
    },
    {
      "epoch": 0.8160117363512522,
      "grad_norm": 0.42091938853263855,
      "learning_rate": 1.7409642230650037e-06,
      "loss": 0.79,
      "step": 4867
    },
    {
      "epoch": 0.8161793985119983,
      "grad_norm": 0.6437473297119141,
      "learning_rate": 1.7379036161486063e-06,
      "loss": 0.8313,
      "step": 4868
    },
    {
      "epoch": 0.8163470606727444,
      "grad_norm": 0.38362085819244385,
      "learning_rate": 1.7348454458267428e-06,
      "loss": 0.7879,
      "step": 4869
    },
    {
      "epoch": 0.8165147228334905,
      "grad_norm": 0.4212976396083832,
      "learning_rate": 1.731789713001304e-06,
      "loss": 0.8341,
      "step": 4870
    },
    {
      "epoch": 0.8166823849942366,
      "grad_norm": 0.47702792286872864,
      "learning_rate": 1.7287364185734602e-06,
      "loss": 0.9573,
      "step": 4871
    },
    {
      "epoch": 0.8168500471549827,
      "grad_norm": 1.012007474899292,
      "learning_rate": 1.7256855634436675e-06,
      "loss": 0.7798,
      "step": 4872
    },
    {
      "epoch": 0.8170177093157288,
      "grad_norm": 0.432649165391922,
      "learning_rate": 1.7226371485116667e-06,
      "loss": 0.8771,
      "step": 4873
    },
    {
      "epoch": 0.8171853714764749,
      "grad_norm": 0.3531188368797302,
      "learning_rate": 1.7195911746764627e-06,
      "loss": 0.8903,
      "step": 4874
    },
    {
      "epoch": 0.817353033637221,
      "grad_norm": 0.39750978350639343,
      "learning_rate": 1.716547642836356e-06,
      "loss": 0.8644,
      "step": 4875
    },
    {
      "epoch": 0.817520695797967,
      "grad_norm": 0.4370575249195099,
      "learning_rate": 1.713506553888924e-06,
      "loss": 0.8534,
      "step": 4876
    },
    {
      "epoch": 0.8176883579587132,
      "grad_norm": 1.1015609502792358,
      "learning_rate": 1.7104679087310172e-06,
      "loss": 0.8118,
      "step": 4877
    },
    {
      "epoch": 0.8178560201194592,
      "grad_norm": 0.3910379707813263,
      "learning_rate": 1.7074317082587755e-06,
      "loss": 0.8986,
      "step": 4878
    },
    {
      "epoch": 0.8180236822802054,
      "grad_norm": 0.3427804708480835,
      "learning_rate": 1.704397953367607e-06,
      "loss": 0.8055,
      "step": 4879
    },
    {
      "epoch": 0.8181913444409514,
      "grad_norm": 0.42192915081977844,
      "learning_rate": 1.7013666449522025e-06,
      "loss": 0.8696,
      "step": 4880
    },
    {
      "epoch": 0.8183590066016976,
      "grad_norm": 0.518771767616272,
      "learning_rate": 1.6983377839065373e-06,
      "loss": 0.8898,
      "step": 4881
    },
    {
      "epoch": 0.8185266687624436,
      "grad_norm": 0.3843253254890442,
      "learning_rate": 1.6953113711238578e-06,
      "loss": 0.8434,
      "step": 4882
    },
    {
      "epoch": 0.8186943309231898,
      "grad_norm": 0.38649222254753113,
      "learning_rate": 1.692287407496689e-06,
      "loss": 0.8963,
      "step": 4883
    },
    {
      "epoch": 0.8188619930839359,
      "grad_norm": 0.5236232876777649,
      "learning_rate": 1.689265893916837e-06,
      "loss": 0.8112,
      "step": 4884
    },
    {
      "epoch": 0.819029655244682,
      "grad_norm": 0.434641033411026,
      "learning_rate": 1.686246831275382e-06,
      "loss": 0.8081,
      "step": 4885
    },
    {
      "epoch": 0.8191973174054281,
      "grad_norm": 0.7000330686569214,
      "learning_rate": 1.683230220462686e-06,
      "loss": 0.9179,
      "step": 4886
    },
    {
      "epoch": 0.8193649795661742,
      "grad_norm": 0.4006652534008026,
      "learning_rate": 1.6802160623683838e-06,
      "loss": 0.7776,
      "step": 4887
    },
    {
      "epoch": 0.8195326417269203,
      "grad_norm": 0.3367191553115845,
      "learning_rate": 1.6772043578813846e-06,
      "loss": 0.7792,
      "step": 4888
    },
    {
      "epoch": 0.8197003038876663,
      "grad_norm": 0.8314791321754456,
      "learning_rate": 1.674195107889881e-06,
      "loss": 0.9016,
      "step": 4889
    },
    {
      "epoch": 0.8198679660484125,
      "grad_norm": 0.6004918217658997,
      "learning_rate": 1.6711883132813355e-06,
      "loss": 0.87,
      "step": 4890
    },
    {
      "epoch": 0.8200356282091585,
      "grad_norm": 0.5409256815910339,
      "learning_rate": 1.668183974942491e-06,
      "loss": 0.872,
      "step": 4891
    },
    {
      "epoch": 0.8202032903699047,
      "grad_norm": 0.7900858521461487,
      "learning_rate": 1.6651820937593677e-06,
      "loss": 0.8623,
      "step": 4892
    },
    {
      "epoch": 0.8203709525306507,
      "grad_norm": 0.3986113667488098,
      "learning_rate": 1.6621826706172494e-06,
      "loss": 0.8713,
      "step": 4893
    },
    {
      "epoch": 0.8205386146913969,
      "grad_norm": 0.35505178570747375,
      "learning_rate": 1.6591857064007067e-06,
      "loss": 0.8332,
      "step": 4894
    },
    {
      "epoch": 0.8207062768521429,
      "grad_norm": 0.3726397752761841,
      "learning_rate": 1.6561912019935843e-06,
      "loss": 0.7985,
      "step": 4895
    },
    {
      "epoch": 0.8208739390128891,
      "grad_norm": 0.3969880938529968,
      "learning_rate": 1.6531991582789964e-06,
      "loss": 0.8429,
      "step": 4896
    },
    {
      "epoch": 0.8210416011736351,
      "grad_norm": 0.4723566472530365,
      "learning_rate": 1.6502095761393332e-06,
      "loss": 0.7666,
      "step": 4897
    },
    {
      "epoch": 0.8212092633343813,
      "grad_norm": 0.8912177085876465,
      "learning_rate": 1.6472224564562577e-06,
      "loss": 0.7355,
      "step": 4898
    },
    {
      "epoch": 0.8213769254951273,
      "grad_norm": 0.3965393602848053,
      "learning_rate": 1.6442378001107106e-06,
      "loss": 0.8948,
      "step": 4899
    },
    {
      "epoch": 0.8215445876558735,
      "grad_norm": 0.45103752613067627,
      "learning_rate": 1.6412556079829066e-06,
      "loss": 0.7278,
      "step": 4900
    },
    {
      "epoch": 0.8217122498166195,
      "grad_norm": 0.4880988895893097,
      "learning_rate": 1.638275880952328e-06,
      "loss": 0.7972,
      "step": 4901
    },
    {
      "epoch": 0.8218799119773657,
      "grad_norm": 0.4846517741680145,
      "learning_rate": 1.6352986198977327e-06,
      "loss": 0.8895,
      "step": 4902
    },
    {
      "epoch": 0.8220475741381117,
      "grad_norm": 0.6601195335388184,
      "learning_rate": 1.632323825697154e-06,
      "loss": 0.8519,
      "step": 4903
    },
    {
      "epoch": 0.8222152362988578,
      "grad_norm": 0.4242129921913147,
      "learning_rate": 1.6293514992278935e-06,
      "loss": 0.8136,
      "step": 4904
    },
    {
      "epoch": 0.8223828984596039,
      "grad_norm": 0.3869190514087677,
      "learning_rate": 1.626381641366529e-06,
      "loss": 0.8755,
      "step": 4905
    },
    {
      "epoch": 0.82255056062035,
      "grad_norm": 0.393439918756485,
      "learning_rate": 1.6234142529889084e-06,
      "loss": 0.8926,
      "step": 4906
    },
    {
      "epoch": 0.8227182227810961,
      "grad_norm": 0.42509788274765015,
      "learning_rate": 1.6204493349701477e-06,
      "loss": 0.8543,
      "step": 4907
    },
    {
      "epoch": 0.8228858849418422,
      "grad_norm": 0.6783158779144287,
      "learning_rate": 1.6174868881846429e-06,
      "loss": 0.7661,
      "step": 4908
    },
    {
      "epoch": 0.8230535471025883,
      "grad_norm": 0.41522926092147827,
      "learning_rate": 1.6145269135060514e-06,
      "loss": 0.9209,
      "step": 4909
    },
    {
      "epoch": 0.8232212092633344,
      "grad_norm": 0.5016360282897949,
      "learning_rate": 1.6115694118073112e-06,
      "loss": 0.7873,
      "step": 4910
    },
    {
      "epoch": 0.8233888714240805,
      "grad_norm": 0.6372166872024536,
      "learning_rate": 1.6086143839606238e-06,
      "loss": 0.8311,
      "step": 4911
    },
    {
      "epoch": 0.8235565335848266,
      "grad_norm": 0.44918274879455566,
      "learning_rate": 1.6056618308374606e-06,
      "loss": 0.8858,
      "step": 4912
    },
    {
      "epoch": 0.8237241957455727,
      "grad_norm": 0.36326393485069275,
      "learning_rate": 1.6027117533085723e-06,
      "loss": 0.7643,
      "step": 4913
    },
    {
      "epoch": 0.8238918579063188,
      "grad_norm": 0.49712425470352173,
      "learning_rate": 1.59976415224397e-06,
      "loss": 0.8362,
      "step": 4914
    },
    {
      "epoch": 0.8240595200670648,
      "grad_norm": 0.5948045253753662,
      "learning_rate": 1.5968190285129349e-06,
      "loss": 0.8402,
      "step": 4915
    },
    {
      "epoch": 0.824227182227811,
      "grad_norm": 0.39514023065567017,
      "learning_rate": 1.5938763829840265e-06,
      "loss": 0.8359,
      "step": 4916
    },
    {
      "epoch": 0.824394844388557,
      "grad_norm": 0.5166681408882141,
      "learning_rate": 1.5909362165250609e-06,
      "loss": 0.7518,
      "step": 4917
    },
    {
      "epoch": 0.8245625065493032,
      "grad_norm": 1.076876163482666,
      "learning_rate": 1.587998530003133e-06,
      "loss": 0.8548,
      "step": 4918
    },
    {
      "epoch": 0.8247301687100492,
      "grad_norm": 0.4052722454071045,
      "learning_rate": 1.5850633242846082e-06,
      "loss": 0.7875,
      "step": 4919
    },
    {
      "epoch": 0.8248978308707954,
      "grad_norm": 0.4035169184207916,
      "learning_rate": 1.5821306002351045e-06,
      "loss": 0.8195,
      "step": 4920
    },
    {
      "epoch": 0.8250654930315414,
      "grad_norm": 0.49509620666503906,
      "learning_rate": 1.5792003587195237e-06,
      "loss": 0.8904,
      "step": 4921
    },
    {
      "epoch": 0.8252331551922876,
      "grad_norm": 0.5392580628395081,
      "learning_rate": 1.5762726006020325e-06,
      "loss": 0.8599,
      "step": 4922
    },
    {
      "epoch": 0.8254008173530336,
      "grad_norm": 0.5295587182044983,
      "learning_rate": 1.5733473267460609e-06,
      "loss": 0.8348,
      "step": 4923
    },
    {
      "epoch": 0.8255684795137798,
      "grad_norm": 0.625406801700592,
      "learning_rate": 1.570424538014307e-06,
      "loss": 0.873,
      "step": 4924
    },
    {
      "epoch": 0.8257361416745258,
      "grad_norm": 0.7624419331550598,
      "learning_rate": 1.5675042352687397e-06,
      "loss": 0.8689,
      "step": 4925
    },
    {
      "epoch": 0.825903803835272,
      "grad_norm": 0.3165587782859802,
      "learning_rate": 1.5645864193705896e-06,
      "loss": 0.8504,
      "step": 4926
    },
    {
      "epoch": 0.826071465996018,
      "grad_norm": 0.377168744802475,
      "learning_rate": 1.5616710911803611e-06,
      "loss": 0.7757,
      "step": 4927
    },
    {
      "epoch": 0.8262391281567641,
      "grad_norm": 0.35758042335510254,
      "learning_rate": 1.558758251557817e-06,
      "loss": 0.8374,
      "step": 4928
    },
    {
      "epoch": 0.8264067903175102,
      "grad_norm": 0.3596714735031128,
      "learning_rate": 1.55584790136199e-06,
      "loss": 0.8467,
      "step": 4929
    },
    {
      "epoch": 0.8265744524782563,
      "grad_norm": 0.3461354970932007,
      "learning_rate": 1.5529400414511809e-06,
      "loss": 0.7623,
      "step": 4930
    },
    {
      "epoch": 0.8267421146390024,
      "grad_norm": 0.4955442547798157,
      "learning_rate": 1.5500346726829495e-06,
      "loss": 0.7475,
      "step": 4931
    },
    {
      "epoch": 0.8269097767997485,
      "grad_norm": 0.3839476406574249,
      "learning_rate": 1.5471317959141296e-06,
      "loss": 0.7481,
      "step": 4932
    },
    {
      "epoch": 0.8270774389604946,
      "grad_norm": 0.5946776866912842,
      "learning_rate": 1.544231412000814e-06,
      "loss": 0.806,
      "step": 4933
    },
    {
      "epoch": 0.8272451011212407,
      "grad_norm": 0.48449745774269104,
      "learning_rate": 1.5413335217983594e-06,
      "loss": 0.7931,
      "step": 4934
    },
    {
      "epoch": 0.8274127632819868,
      "grad_norm": 0.36142489314079285,
      "learning_rate": 1.5384381261613924e-06,
      "loss": 0.8365,
      "step": 4935
    },
    {
      "epoch": 0.8275804254427329,
      "grad_norm": 0.5161846876144409,
      "learning_rate": 1.5355452259437986e-06,
      "loss": 0.9174,
      "step": 4936
    },
    {
      "epoch": 0.827748087603479,
      "grad_norm": 0.5528901815414429,
      "learning_rate": 1.5326548219987326e-06,
      "loss": 0.8181,
      "step": 4937
    },
    {
      "epoch": 0.8279157497642251,
      "grad_norm": 0.45687246322631836,
      "learning_rate": 1.52976691517861e-06,
      "loss": 0.8257,
      "step": 4938
    },
    {
      "epoch": 0.8280834119249711,
      "grad_norm": 0.3636484146118164,
      "learning_rate": 1.5268815063351072e-06,
      "loss": 0.7596,
      "step": 4939
    },
    {
      "epoch": 0.8282510740857173,
      "grad_norm": 0.36633291840553284,
      "learning_rate": 1.5239985963191683e-06,
      "loss": 0.7881,
      "step": 4940
    },
    {
      "epoch": 0.8284187362464633,
      "grad_norm": 0.41070568561553955,
      "learning_rate": 1.5211181859810032e-06,
      "loss": 0.8486,
      "step": 4941
    },
    {
      "epoch": 0.8285863984072095,
      "grad_norm": 0.8401875495910645,
      "learning_rate": 1.5182402761700776e-06,
      "loss": 0.8714,
      "step": 4942
    },
    {
      "epoch": 0.8287540605679555,
      "grad_norm": 0.3784967064857483,
      "learning_rate": 1.5153648677351196e-06,
      "loss": 0.7949,
      "step": 4943
    },
    {
      "epoch": 0.8289217227287017,
      "grad_norm": 0.40644749999046326,
      "learning_rate": 1.5124919615241284e-06,
      "loss": 0.7011,
      "step": 4944
    },
    {
      "epoch": 0.8290893848894477,
      "grad_norm": 0.4508950412273407,
      "learning_rate": 1.5096215583843554e-06,
      "loss": 0.8217,
      "step": 4945
    },
    {
      "epoch": 0.8292570470501939,
      "grad_norm": 0.3945903182029724,
      "learning_rate": 1.5067536591623233e-06,
      "loss": 0.74,
      "step": 4946
    },
    {
      "epoch": 0.8294247092109399,
      "grad_norm": 0.3296574652194977,
      "learning_rate": 1.5038882647038034e-06,
      "loss": 0.8235,
      "step": 4947
    },
    {
      "epoch": 0.8295923713716861,
      "grad_norm": 0.3594135344028473,
      "learning_rate": 1.5010253758538396e-06,
      "loss": 0.8322,
      "step": 4948
    },
    {
      "epoch": 0.8297600335324321,
      "grad_norm": 0.6399166584014893,
      "learning_rate": 1.4981649934567365e-06,
      "loss": 0.8605,
      "step": 4949
    },
    {
      "epoch": 0.8299276956931783,
      "grad_norm": 0.3736465275287628,
      "learning_rate": 1.4953071183560508e-06,
      "loss": 0.7765,
      "step": 4950
    },
    {
      "epoch": 0.8300953578539243,
      "grad_norm": 0.37793681025505066,
      "learning_rate": 1.49245175139461e-06,
      "loss": 0.8175,
      "step": 4951
    },
    {
      "epoch": 0.8302630200146704,
      "grad_norm": 0.5853179693222046,
      "learning_rate": 1.4895988934144956e-06,
      "loss": 0.7904,
      "step": 4952
    },
    {
      "epoch": 0.8304306821754165,
      "grad_norm": 0.5771636962890625,
      "learning_rate": 1.4867485452570473e-06,
      "loss": 0.7803,
      "step": 4953
    },
    {
      "epoch": 0.8305983443361626,
      "grad_norm": 0.35129693150520325,
      "learning_rate": 1.483900707762873e-06,
      "loss": 0.8745,
      "step": 4954
    },
    {
      "epoch": 0.8307660064969087,
      "grad_norm": 0.5694350600242615,
      "learning_rate": 1.4810553817718343e-06,
      "loss": 0.8235,
      "step": 4955
    },
    {
      "epoch": 0.8309336686576548,
      "grad_norm": 0.41765889525413513,
      "learning_rate": 1.4782125681230497e-06,
      "loss": 0.9147,
      "step": 4956
    },
    {
      "epoch": 0.8311013308184009,
      "grad_norm": 0.6250484585762024,
      "learning_rate": 1.4753722676549042e-06,
      "loss": 0.8673,
      "step": 4957
    },
    {
      "epoch": 0.831268992979147,
      "grad_norm": 0.33489978313446045,
      "learning_rate": 1.4725344812050336e-06,
      "loss": 0.7793,
      "step": 4958
    },
    {
      "epoch": 0.8314366551398931,
      "grad_norm": 0.39565497636795044,
      "learning_rate": 1.469699209610338e-06,
      "loss": 0.8873,
      "step": 4959
    },
    {
      "epoch": 0.8316043173006392,
      "grad_norm": 0.41611769795417786,
      "learning_rate": 1.46686645370698e-06,
      "loss": 0.7699,
      "step": 4960
    },
    {
      "epoch": 0.8317719794613853,
      "grad_norm": 0.3994881212711334,
      "learning_rate": 1.4640362143303644e-06,
      "loss": 0.935,
      "step": 4961
    },
    {
      "epoch": 0.8319396416221314,
      "grad_norm": 0.7628393173217773,
      "learning_rate": 1.4612084923151703e-06,
      "loss": 0.8898,
      "step": 4962
    },
    {
      "epoch": 0.8321073037828774,
      "grad_norm": 0.445016473531723,
      "learning_rate": 1.4583832884953241e-06,
      "loss": 0.8472,
      "step": 4963
    },
    {
      "epoch": 0.8322749659436236,
      "grad_norm": 0.46770772337913513,
      "learning_rate": 1.4555606037040159e-06,
      "loss": 0.8566,
      "step": 4964
    },
    {
      "epoch": 0.8324426281043696,
      "grad_norm": 0.35536736249923706,
      "learning_rate": 1.4527404387736943e-06,
      "loss": 0.8784,
      "step": 4965
    },
    {
      "epoch": 0.8326102902651158,
      "grad_norm": 0.3618987202644348,
      "learning_rate": 1.4499227945360505e-06,
      "loss": 0.834,
      "step": 4966
    },
    {
      "epoch": 0.8327779524258618,
      "grad_norm": 0.46326589584350586,
      "learning_rate": 1.44710767182205e-06,
      "loss": 0.8092,
      "step": 4967
    },
    {
      "epoch": 0.832945614586608,
      "grad_norm": 0.3517701029777527,
      "learning_rate": 1.4442950714619075e-06,
      "loss": 0.7963,
      "step": 4968
    },
    {
      "epoch": 0.8331132767473541,
      "grad_norm": 0.5149338245391846,
      "learning_rate": 1.4414849942850927e-06,
      "loss": 0.8521,
      "step": 4969
    },
    {
      "epoch": 0.8332809389081002,
      "grad_norm": 0.5947842597961426,
      "learning_rate": 1.4386774411203285e-06,
      "loss": 0.8438,
      "step": 4970
    },
    {
      "epoch": 0.8334486010688463,
      "grad_norm": 0.46693262457847595,
      "learning_rate": 1.4358724127956036e-06,
      "loss": 0.9143,
      "step": 4971
    },
    {
      "epoch": 0.8336162632295924,
      "grad_norm": 0.3993178904056549,
      "learning_rate": 1.4330699101381496e-06,
      "loss": 0.8364,
      "step": 4972
    },
    {
      "epoch": 0.8337839253903385,
      "grad_norm": 0.44750991463661194,
      "learning_rate": 1.430269933974463e-06,
      "loss": 0.7832,
      "step": 4973
    },
    {
      "epoch": 0.8339515875510846,
      "grad_norm": 0.40281417965888977,
      "learning_rate": 1.4274724851302912e-06,
      "loss": 0.9152,
      "step": 4974
    },
    {
      "epoch": 0.8341192497118307,
      "grad_norm": 0.34603404998779297,
      "learning_rate": 1.4246775644306333e-06,
      "loss": 0.7777,
      "step": 4975
    },
    {
      "epoch": 0.8342869118725768,
      "grad_norm": 0.4452677071094513,
      "learning_rate": 1.4218851726997507e-06,
      "loss": 0.8059,
      "step": 4976
    },
    {
      "epoch": 0.8344545740333229,
      "grad_norm": 0.3128507137298584,
      "learning_rate": 1.4190953107611483e-06,
      "loss": 0.8199,
      "step": 4977
    },
    {
      "epoch": 0.834622236194069,
      "grad_norm": 0.40192288160324097,
      "learning_rate": 1.4163079794375977e-06,
      "loss": 0.7547,
      "step": 4978
    },
    {
      "epoch": 0.8347898983548151,
      "grad_norm": 0.3185981214046478,
      "learning_rate": 1.413523179551115e-06,
      "loss": 0.7361,
      "step": 4979
    },
    {
      "epoch": 0.8349575605155611,
      "grad_norm": 0.5001709461212158,
      "learning_rate": 1.4107409119229675e-06,
      "loss": 0.8404,
      "step": 4980
    },
    {
      "epoch": 0.8351252226763073,
      "grad_norm": 0.5474051237106323,
      "learning_rate": 1.4079611773736879e-06,
      "loss": 0.8478,
      "step": 4981
    },
    {
      "epoch": 0.8352928848370533,
      "grad_norm": 0.4860477149486542,
      "learning_rate": 1.4051839767230479e-06,
      "loss": 0.787,
      "step": 4982
    },
    {
      "epoch": 0.8354605469977995,
      "grad_norm": 0.3890592157840729,
      "learning_rate": 1.4024093107900838e-06,
      "loss": 0.7679,
      "step": 4983
    },
    {
      "epoch": 0.8356282091585455,
      "grad_norm": 0.4080706536769867,
      "learning_rate": 1.3996371803930753e-06,
      "loss": 0.9019,
      "step": 4984
    },
    {
      "epoch": 0.8357958713192917,
      "grad_norm": 0.36382532119750977,
      "learning_rate": 1.3968675863495572e-06,
      "loss": 0.878,
      "step": 4985
    },
    {
      "epoch": 0.8359635334800377,
      "grad_norm": 0.40647241473197937,
      "learning_rate": 1.39410052947632e-06,
      "loss": 0.8264,
      "step": 4986
    },
    {
      "epoch": 0.8361311956407839,
      "grad_norm": 0.36876314878463745,
      "learning_rate": 1.3913360105894048e-06,
      "loss": 0.8902,
      "step": 4987
    },
    {
      "epoch": 0.8362988578015299,
      "grad_norm": 0.402743935585022,
      "learning_rate": 1.3885740305040962e-06,
      "loss": 0.9009,
      "step": 4988
    },
    {
      "epoch": 0.8364665199622761,
      "grad_norm": 0.4157020151615143,
      "learning_rate": 1.3858145900349385e-06,
      "loss": 0.7778,
      "step": 4989
    },
    {
      "epoch": 0.8366341821230221,
      "grad_norm": 0.9957639575004578,
      "learning_rate": 1.3830576899957292e-06,
      "loss": 0.8508,
      "step": 4990
    },
    {
      "epoch": 0.8368018442837682,
      "grad_norm": 0.4136386215686798,
      "learning_rate": 1.3803033311995072e-06,
      "loss": 0.8265,
      "step": 4991
    },
    {
      "epoch": 0.8369695064445143,
      "grad_norm": 0.4569907486438751,
      "learning_rate": 1.3775515144585705e-06,
      "loss": 0.8268,
      "step": 4992
    },
    {
      "epoch": 0.8371371686052604,
      "grad_norm": 0.35812056064605713,
      "learning_rate": 1.3748022405844619e-06,
      "loss": 0.807,
      "step": 4993
    },
    {
      "epoch": 0.8373048307660065,
      "grad_norm": 0.4523765444755554,
      "learning_rate": 1.3720555103879751e-06,
      "loss": 0.9474,
      "step": 4994
    },
    {
      "epoch": 0.8374724929267526,
      "grad_norm": 0.3831980228424072,
      "learning_rate": 1.369311324679159e-06,
      "loss": 0.8371,
      "step": 4995
    },
    {
      "epoch": 0.8376401550874987,
      "grad_norm": 0.4144824147224426,
      "learning_rate": 1.366569684267306e-06,
      "loss": 0.7924,
      "step": 4996
    },
    {
      "epoch": 0.8378078172482448,
      "grad_norm": 0.40071797370910645,
      "learning_rate": 1.363830589960956e-06,
      "loss": 0.7691,
      "step": 4997
    },
    {
      "epoch": 0.8379754794089909,
      "grad_norm": 0.46362945437431335,
      "learning_rate": 1.3610940425679076e-06,
      "loss": 0.8519,
      "step": 4998
    },
    {
      "epoch": 0.838143141569737,
      "grad_norm": 0.3925900161266327,
      "learning_rate": 1.358360042895198e-06,
      "loss": 0.795,
      "step": 4999
    },
    {
      "epoch": 0.8383108037304831,
      "grad_norm": 0.5459030270576477,
      "learning_rate": 1.355628591749122e-06,
      "loss": 0.8003,
      "step": 5000
    },
    {
      "epoch": 0.8384784658912292,
      "grad_norm": 0.3537822663784027,
      "learning_rate": 1.3528996899352155e-06,
      "loss": 0.849,
      "step": 5001
    },
    {
      "epoch": 0.8386461280519752,
      "grad_norm": 0.42230555415153503,
      "learning_rate": 1.3501733382582627e-06,
      "loss": 0.771,
      "step": 5002
    },
    {
      "epoch": 0.8388137902127214,
      "grad_norm": 0.36975836753845215,
      "learning_rate": 1.347449537522305e-06,
      "loss": 0.8506,
      "step": 5003
    },
    {
      "epoch": 0.8389814523734674,
      "grad_norm": 0.4725952446460724,
      "learning_rate": 1.3447282885306201e-06,
      "loss": 0.7526,
      "step": 5004
    },
    {
      "epoch": 0.8391491145342136,
      "grad_norm": 0.6361253261566162,
      "learning_rate": 1.3420095920857378e-06,
      "loss": 0.7488,
      "step": 5005
    },
    {
      "epoch": 0.8393167766949596,
      "grad_norm": 0.3656197786331177,
      "learning_rate": 1.3392934489894427e-06,
      "loss": 0.7931,
      "step": 5006
    },
    {
      "epoch": 0.8394844388557058,
      "grad_norm": 0.3905761241912842,
      "learning_rate": 1.336579860042748e-06,
      "loss": 0.7973,
      "step": 5007
    },
    {
      "epoch": 0.8396521010164518,
      "grad_norm": 0.35232487320899963,
      "learning_rate": 1.333868826045932e-06,
      "loss": 0.7842,
      "step": 5008
    },
    {
      "epoch": 0.839819763177198,
      "grad_norm": 0.5195075273513794,
      "learning_rate": 1.3311603477985101e-06,
      "loss": 0.8098,
      "step": 5009
    },
    {
      "epoch": 0.839987425337944,
      "grad_norm": 0.5655362606048584,
      "learning_rate": 1.3284544260992483e-06,
      "loss": 0.7666,
      "step": 5010
    },
    {
      "epoch": 0.8401550874986902,
      "grad_norm": 0.38470450043678284,
      "learning_rate": 1.3257510617461533e-06,
      "loss": 0.7918,
      "step": 5011
    },
    {
      "epoch": 0.8403227496594362,
      "grad_norm": 0.43328046798706055,
      "learning_rate": 1.32305025553648e-06,
      "loss": 0.7772,
      "step": 5012
    },
    {
      "epoch": 0.8404904118201824,
      "grad_norm": 0.38678938150405884,
      "learning_rate": 1.320352008266731e-06,
      "loss": 0.7945,
      "step": 5013
    },
    {
      "epoch": 0.8406580739809284,
      "grad_norm": 0.4375285506248474,
      "learning_rate": 1.3176563207326554e-06,
      "loss": 0.8519,
      "step": 5014
    },
    {
      "epoch": 0.8408257361416746,
      "grad_norm": 0.5520859360694885,
      "learning_rate": 1.314963193729243e-06,
      "loss": 0.7942,
      "step": 5015
    },
    {
      "epoch": 0.8409933983024206,
      "grad_norm": 0.5742729902267456,
      "learning_rate": 1.312272628050728e-06,
      "loss": 0.7296,
      "step": 5016
    },
    {
      "epoch": 0.8411610604631667,
      "grad_norm": 0.35533037781715393,
      "learning_rate": 1.3095846244905952e-06,
      "loss": 0.8134,
      "step": 5017
    },
    {
      "epoch": 0.8413287226239128,
      "grad_norm": 0.4646261930465698,
      "learning_rate": 1.3068991838415669e-06,
      "loss": 0.8755,
      "step": 5018
    },
    {
      "epoch": 0.8414963847846589,
      "grad_norm": 0.56504225730896,
      "learning_rate": 1.3042163068956147e-06,
      "loss": 0.745,
      "step": 5019
    },
    {
      "epoch": 0.841664046945405,
      "grad_norm": 0.38503456115722656,
      "learning_rate": 1.3015359944439544e-06,
      "loss": 0.8063,
      "step": 5020
    },
    {
      "epoch": 0.8418317091061511,
      "grad_norm": 0.3094675540924072,
      "learning_rate": 1.2988582472770372e-06,
      "loss": 0.806,
      "step": 5021
    },
    {
      "epoch": 0.8419993712668972,
      "grad_norm": 0.4315580129623413,
      "learning_rate": 1.2961830661845697e-06,
      "loss": 0.8175,
      "step": 5022
    },
    {
      "epoch": 0.8421670334276433,
      "grad_norm": 0.3926454484462738,
      "learning_rate": 1.293510451955492e-06,
      "loss": 0.8225,
      "step": 5023
    },
    {
      "epoch": 0.8423346955883894,
      "grad_norm": 0.7259623408317566,
      "learning_rate": 1.2908404053779943e-06,
      "loss": 0.8553,
      "step": 5024
    },
    {
      "epoch": 0.8425023577491355,
      "grad_norm": 0.3977755606174469,
      "learning_rate": 1.2881729272395049e-06,
      "loss": 0.7511,
      "step": 5025
    },
    {
      "epoch": 0.8426700199098816,
      "grad_norm": 0.37917256355285645,
      "learning_rate": 1.2855080183266954e-06,
      "loss": 0.8226,
      "step": 5026
    },
    {
      "epoch": 0.8428376820706277,
      "grad_norm": 0.3832605481147766,
      "learning_rate": 1.2828456794254818e-06,
      "loss": 0.8326,
      "step": 5027
    },
    {
      "epoch": 0.8430053442313737,
      "grad_norm": 0.3946949243545532,
      "learning_rate": 1.2801859113210213e-06,
      "loss": 0.7871,
      "step": 5028
    },
    {
      "epoch": 0.8431730063921199,
      "grad_norm": 0.38526833057403564,
      "learning_rate": 1.277528714797709e-06,
      "loss": 0.8081,
      "step": 5029
    },
    {
      "epoch": 0.8433406685528659,
      "grad_norm": 0.32758602499961853,
      "learning_rate": 1.2748740906391887e-06,
      "loss": 0.7913,
      "step": 5030
    },
    {
      "epoch": 0.8435083307136121,
      "grad_norm": 0.5229128003120422,
      "learning_rate": 1.2722220396283401e-06,
      "loss": 0.7908,
      "step": 5031
    },
    {
      "epoch": 0.8436759928743581,
      "grad_norm": 0.46804341673851013,
      "learning_rate": 1.2695725625472854e-06,
      "loss": 0.7986,
      "step": 5032
    },
    {
      "epoch": 0.8438436550351043,
      "grad_norm": 0.5670343637466431,
      "learning_rate": 1.266925660177394e-06,
      "loss": 0.8722,
      "step": 5033
    },
    {
      "epoch": 0.8440113171958503,
      "grad_norm": 0.3946131765842438,
      "learning_rate": 1.264281333299261e-06,
      "loss": 0.7333,
      "step": 5034
    },
    {
      "epoch": 0.8441789793565965,
      "grad_norm": 1.2765463590621948,
      "learning_rate": 1.2616395826927352e-06,
      "loss": 0.9144,
      "step": 5035
    },
    {
      "epoch": 0.8443466415173425,
      "grad_norm": 0.39245614409446716,
      "learning_rate": 1.2590004091369045e-06,
      "loss": 0.8414,
      "step": 5036
    },
    {
      "epoch": 0.8445143036780887,
      "grad_norm": 0.3988438844680786,
      "learning_rate": 1.2563638134100898e-06,
      "loss": 0.8746,
      "step": 5037
    },
    {
      "epoch": 0.8446819658388347,
      "grad_norm": 0.7531177997589111,
      "learning_rate": 1.2537297962898587e-06,
      "loss": 0.8994,
      "step": 5038
    },
    {
      "epoch": 0.8448496279995809,
      "grad_norm": 0.3667586147785187,
      "learning_rate": 1.2510983585530134e-06,
      "loss": 0.8386,
      "step": 5039
    },
    {
      "epoch": 0.8450172901603269,
      "grad_norm": 0.4034627377986908,
      "learning_rate": 1.2484695009755976e-06,
      "loss": 0.8495,
      "step": 5040
    },
    {
      "epoch": 0.845184952321073,
      "grad_norm": 0.39673033356666565,
      "learning_rate": 1.2458432243328955e-06,
      "loss": 0.8627,
      "step": 5041
    },
    {
      "epoch": 0.8453526144818191,
      "grad_norm": 0.4349612295627594,
      "learning_rate": 1.2432195293994288e-06,
      "loss": 0.8247,
      "step": 5042
    },
    {
      "epoch": 0.8455202766425652,
      "grad_norm": 0.34705671668052673,
      "learning_rate": 1.2405984169489526e-06,
      "loss": 0.8052,
      "step": 5043
    },
    {
      "epoch": 0.8456879388033113,
      "grad_norm": 0.39677438139915466,
      "learning_rate": 1.2379798877544735e-06,
      "loss": 0.8462,
      "step": 5044
    },
    {
      "epoch": 0.8458556009640574,
      "grad_norm": 0.42162394523620605,
      "learning_rate": 1.2353639425882203e-06,
      "loss": 0.8864,
      "step": 5045
    },
    {
      "epoch": 0.8460232631248035,
      "grad_norm": 0.40522149205207825,
      "learning_rate": 1.2327505822216734e-06,
      "loss": 0.8179,
      "step": 5046
    },
    {
      "epoch": 0.8461909252855496,
      "grad_norm": 0.45286887884140015,
      "learning_rate": 1.2301398074255444e-06,
      "loss": 0.7911,
      "step": 5047
    },
    {
      "epoch": 0.8463585874462957,
      "grad_norm": 0.5825637578964233,
      "learning_rate": 1.2275316189697784e-06,
      "loss": 0.8652,
      "step": 5048
    },
    {
      "epoch": 0.8465262496070418,
      "grad_norm": 0.49502861499786377,
      "learning_rate": 1.2249260176235678e-06,
      "loss": 0.8698,
      "step": 5049
    },
    {
      "epoch": 0.8466939117677879,
      "grad_norm": 0.41778823733329773,
      "learning_rate": 1.2223230041553335e-06,
      "loss": 0.7378,
      "step": 5050
    },
    {
      "epoch": 0.846861573928534,
      "grad_norm": 0.38942602276802063,
      "learning_rate": 1.2197225793327393e-06,
      "loss": 0.8643,
      "step": 5051
    },
    {
      "epoch": 0.84702923608928,
      "grad_norm": 0.4558364748954773,
      "learning_rate": 1.2171247439226809e-06,
      "loss": 0.8634,
      "step": 5052
    },
    {
      "epoch": 0.8471968982500262,
      "grad_norm": 0.7153915762901306,
      "learning_rate": 1.2145294986912903e-06,
      "loss": 0.7978,
      "step": 5053
    },
    {
      "epoch": 0.8473645604107722,
      "grad_norm": 0.3906574547290802,
      "learning_rate": 1.2119368444039392e-06,
      "loss": 0.765,
      "step": 5054
    },
    {
      "epoch": 0.8475322225715184,
      "grad_norm": 0.412419855594635,
      "learning_rate": 1.2093467818252358e-06,
      "loss": 0.8181,
      "step": 5055
    },
    {
      "epoch": 0.8476998847322645,
      "grad_norm": 0.3883296549320221,
      "learning_rate": 1.206759311719019e-06,
      "loss": 0.8006,
      "step": 5056
    },
    {
      "epoch": 0.8478675468930106,
      "grad_norm": 0.39480140805244446,
      "learning_rate": 1.2041744348483652e-06,
      "loss": 0.8763,
      "step": 5057
    },
    {
      "epoch": 0.8480352090537567,
      "grad_norm": 0.6128484606742859,
      "learning_rate": 1.2015921519755891e-06,
      "loss": 0.7826,
      "step": 5058
    },
    {
      "epoch": 0.8482028712145028,
      "grad_norm": 0.6531548500061035,
      "learning_rate": 1.1990124638622337e-06,
      "loss": 0.8052,
      "step": 5059
    },
    {
      "epoch": 0.8483705333752489,
      "grad_norm": 0.3930145502090454,
      "learning_rate": 1.196435371269089e-06,
      "loss": 0.8634,
      "step": 5060
    },
    {
      "epoch": 0.848538195535995,
      "grad_norm": 0.5058462619781494,
      "learning_rate": 1.193860874956162e-06,
      "loss": 0.8377,
      "step": 5061
    },
    {
      "epoch": 0.8487058576967411,
      "grad_norm": 0.3774513602256775,
      "learning_rate": 1.1912889756827073e-06,
      "loss": 0.7747,
      "step": 5062
    },
    {
      "epoch": 0.8488735198574872,
      "grad_norm": 0.5487313270568848,
      "learning_rate": 1.1887196742072115e-06,
      "loss": 0.9396,
      "step": 5063
    },
    {
      "epoch": 0.8490411820182333,
      "grad_norm": 0.40379512310028076,
      "learning_rate": 1.186152971287392e-06,
      "loss": 0.875,
      "step": 5064
    },
    {
      "epoch": 0.8492088441789793,
      "grad_norm": 0.45634934306144714,
      "learning_rate": 1.1835888676802021e-06,
      "loss": 0.8128,
      "step": 5065
    },
    {
      "epoch": 0.8493765063397255,
      "grad_norm": 0.4219779074192047,
      "learning_rate": 1.181027364141828e-06,
      "loss": 0.854,
      "step": 5066
    },
    {
      "epoch": 0.8495441685004715,
      "grad_norm": 0.4084298014640808,
      "learning_rate": 1.1784684614276854e-06,
      "loss": 0.8018,
      "step": 5067
    },
    {
      "epoch": 0.8497118306612177,
      "grad_norm": 0.40283524990081787,
      "learning_rate": 1.1759121602924306e-06,
      "loss": 0.8024,
      "step": 5068
    },
    {
      "epoch": 0.8498794928219637,
      "grad_norm": 0.718620240688324,
      "learning_rate": 1.1733584614899474e-06,
      "loss": 0.8947,
      "step": 5069
    },
    {
      "epoch": 0.8500471549827099,
      "grad_norm": 0.4366646111011505,
      "learning_rate": 1.1708073657733498e-06,
      "loss": 0.9105,
      "step": 5070
    },
    {
      "epoch": 0.8502148171434559,
      "grad_norm": 0.6225143074989319,
      "learning_rate": 1.168258873894993e-06,
      "loss": 0.8594,
      "step": 5071
    },
    {
      "epoch": 0.8503824793042021,
      "grad_norm": 0.3575076758861542,
      "learning_rate": 1.1657129866064542e-06,
      "loss": 0.7517,
      "step": 5072
    },
    {
      "epoch": 0.8505501414649481,
      "grad_norm": 0.38253408670425415,
      "learning_rate": 1.1631697046585511e-06,
      "loss": 0.8772,
      "step": 5073
    },
    {
      "epoch": 0.8507178036256943,
      "grad_norm": 0.4497597813606262,
      "learning_rate": 1.1606290288013266e-06,
      "loss": 0.8285,
      "step": 5074
    },
    {
      "epoch": 0.8508854657864403,
      "grad_norm": 0.45971205830574036,
      "learning_rate": 1.158090959784055e-06,
      "loss": 0.8196,
      "step": 5075
    },
    {
      "epoch": 0.8510531279471865,
      "grad_norm": 0.3878073990345001,
      "learning_rate": 1.1555554983552498e-06,
      "loss": 0.8788,
      "step": 5076
    },
    {
      "epoch": 0.8512207901079325,
      "grad_norm": 0.7124265432357788,
      "learning_rate": 1.1530226452626457e-06,
      "loss": 0.7709,
      "step": 5077
    },
    {
      "epoch": 0.8513884522686787,
      "grad_norm": 0.7837972044944763,
      "learning_rate": 1.1504924012532136e-06,
      "loss": 0.8354,
      "step": 5078
    },
    {
      "epoch": 0.8515561144294247,
      "grad_norm": 0.4087355136871338,
      "learning_rate": 1.1479647670731574e-06,
      "loss": 0.7846,
      "step": 5079
    },
    {
      "epoch": 0.8517237765901708,
      "grad_norm": 0.3593689799308777,
      "learning_rate": 1.1454397434679022e-06,
      "loss": 0.786,
      "step": 5080
    },
    {
      "epoch": 0.8518914387509169,
      "grad_norm": 0.836773693561554,
      "learning_rate": 1.1429173311821095e-06,
      "loss": 0.9083,
      "step": 5081
    },
    {
      "epoch": 0.852059100911663,
      "grad_norm": 0.3732576370239258,
      "learning_rate": 1.1403975309596738e-06,
      "loss": 0.8702,
      "step": 5082
    },
    {
      "epoch": 0.8522267630724091,
      "grad_norm": 0.6807574033737183,
      "learning_rate": 1.1378803435437135e-06,
      "loss": 0.8596,
      "step": 5083
    },
    {
      "epoch": 0.8523944252331552,
      "grad_norm": 0.4082207977771759,
      "learning_rate": 1.1353657696765752e-06,
      "loss": 0.8888,
      "step": 5084
    },
    {
      "epoch": 0.8525620873939013,
      "grad_norm": 0.406039297580719,
      "learning_rate": 1.132853810099843e-06,
      "loss": 0.7492,
      "step": 5085
    },
    {
      "epoch": 0.8527297495546474,
      "grad_norm": 0.3803439736366272,
      "learning_rate": 1.1303444655543206e-06,
      "loss": 0.8707,
      "step": 5086
    },
    {
      "epoch": 0.8528974117153935,
      "grad_norm": 0.5561349987983704,
      "learning_rate": 1.1278377367800486e-06,
      "loss": 0.8445,
      "step": 5087
    },
    {
      "epoch": 0.8530650738761396,
      "grad_norm": 0.5199757218360901,
      "learning_rate": 1.1253336245162893e-06,
      "loss": 0.8566,
      "step": 5088
    },
    {
      "epoch": 0.8532327360368857,
      "grad_norm": 0.39440688490867615,
      "learning_rate": 1.1228321295015365e-06,
      "loss": 0.7599,
      "step": 5089
    },
    {
      "epoch": 0.8534003981976318,
      "grad_norm": 0.6887969374656677,
      "learning_rate": 1.1203332524735145e-06,
      "loss": 0.816,
      "step": 5090
    },
    {
      "epoch": 0.8535680603583778,
      "grad_norm": 0.3930720090866089,
      "learning_rate": 1.1178369941691703e-06,
      "loss": 0.7775,
      "step": 5091
    },
    {
      "epoch": 0.853735722519124,
      "grad_norm": 0.5686238408088684,
      "learning_rate": 1.115343355324684e-06,
      "loss": 0.7328,
      "step": 5092
    },
    {
      "epoch": 0.85390338467987,
      "grad_norm": 0.4349891245365143,
      "learning_rate": 1.1128523366754595e-06,
      "loss": 0.8757,
      "step": 5093
    },
    {
      "epoch": 0.8540710468406162,
      "grad_norm": 0.5157162547111511,
      "learning_rate": 1.1103639389561272e-06,
      "loss": 0.8234,
      "step": 5094
    },
    {
      "epoch": 0.8542387090013622,
      "grad_norm": 0.45852211117744446,
      "learning_rate": 1.1078781629005498e-06,
      "loss": 0.7912,
      "step": 5095
    },
    {
      "epoch": 0.8544063711621084,
      "grad_norm": 0.3618316054344177,
      "learning_rate": 1.1053950092418097e-06,
      "loss": 0.8335,
      "step": 5096
    },
    {
      "epoch": 0.8545740333228544,
      "grad_norm": 0.35856497287750244,
      "learning_rate": 1.1029144787122226e-06,
      "loss": 0.8641,
      "step": 5097
    },
    {
      "epoch": 0.8547416954836006,
      "grad_norm": 0.4254955053329468,
      "learning_rate": 1.1004365720433274e-06,
      "loss": 0.8358,
      "step": 5098
    },
    {
      "epoch": 0.8549093576443466,
      "grad_norm": 0.4587903320789337,
      "learning_rate": 1.0979612899658875e-06,
      "loss": 0.8598,
      "step": 5099
    },
    {
      "epoch": 0.8550770198050928,
      "grad_norm": 0.5157368779182434,
      "learning_rate": 1.0954886332098946e-06,
      "loss": 0.7303,
      "step": 5100
    },
    {
      "epoch": 0.8552446819658388,
      "grad_norm": 0.6669063568115234,
      "learning_rate": 1.0930186025045708e-06,
      "loss": 0.8335,
      "step": 5101
    },
    {
      "epoch": 0.855412344126585,
      "grad_norm": 0.39919060468673706,
      "learning_rate": 1.0905511985783513e-06,
      "loss": 0.8401,
      "step": 5102
    },
    {
      "epoch": 0.855580006287331,
      "grad_norm": 0.46347081661224365,
      "learning_rate": 1.088086422158906e-06,
      "loss": 0.8404,
      "step": 5103
    },
    {
      "epoch": 0.8557476684480771,
      "grad_norm": 0.466889888048172,
      "learning_rate": 1.0856242739731325e-06,
      "loss": 0.9507,
      "step": 5104
    },
    {
      "epoch": 0.8559153306088232,
      "grad_norm": 0.39966902136802673,
      "learning_rate": 1.0831647547471435e-06,
      "loss": 0.8386,
      "step": 5105
    },
    {
      "epoch": 0.8560829927695693,
      "grad_norm": 0.4732997715473175,
      "learning_rate": 1.0807078652062874e-06,
      "loss": 0.8487,
      "step": 5106
    },
    {
      "epoch": 0.8562506549303154,
      "grad_norm": 0.5787397027015686,
      "learning_rate": 1.0782536060751237e-06,
      "loss": 0.8471,
      "step": 5107
    },
    {
      "epoch": 0.8564183170910615,
      "grad_norm": 0.40982586145401,
      "learning_rate": 1.0758019780774488e-06,
      "loss": 0.7979,
      "step": 5108
    },
    {
      "epoch": 0.8565859792518076,
      "grad_norm": 0.43968313932418823,
      "learning_rate": 1.0733529819362786e-06,
      "loss": 0.7994,
      "step": 5109
    },
    {
      "epoch": 0.8567536414125537,
      "grad_norm": 0.4219721853733063,
      "learning_rate": 1.0709066183738493e-06,
      "loss": 0.9183,
      "step": 5110
    },
    {
      "epoch": 0.8569213035732998,
      "grad_norm": 0.3750663101673126,
      "learning_rate": 1.0684628881116265e-06,
      "loss": 0.9232,
      "step": 5111
    },
    {
      "epoch": 0.8570889657340459,
      "grad_norm": 0.4194062054157257,
      "learning_rate": 1.0660217918702965e-06,
      "loss": 0.8087,
      "step": 5112
    },
    {
      "epoch": 0.857256627894792,
      "grad_norm": 0.38401052355766296,
      "learning_rate": 1.0635833303697663e-06,
      "loss": 0.8462,
      "step": 5113
    },
    {
      "epoch": 0.8574242900555381,
      "grad_norm": 0.38535383343696594,
      "learning_rate": 1.0611475043291708e-06,
      "loss": 0.8383,
      "step": 5114
    },
    {
      "epoch": 0.8575919522162841,
      "grad_norm": 0.4055083394050598,
      "learning_rate": 1.0587143144668655e-06,
      "loss": 0.8271,
      "step": 5115
    },
    {
      "epoch": 0.8577596143770303,
      "grad_norm": 0.5805023908615112,
      "learning_rate": 1.0562837615004252e-06,
      "loss": 0.8211,
      "step": 5116
    },
    {
      "epoch": 0.8579272765377763,
      "grad_norm": 0.44198617339134216,
      "learning_rate": 1.053855846146653e-06,
      "loss": 0.9089,
      "step": 5117
    },
    {
      "epoch": 0.8580949386985225,
      "grad_norm": 0.6872033476829529,
      "learning_rate": 1.0514305691215687e-06,
      "loss": 0.8235,
      "step": 5118
    },
    {
      "epoch": 0.8582626008592685,
      "grad_norm": 0.3935192823410034,
      "learning_rate": 1.0490079311404178e-06,
      "loss": 0.9665,
      "step": 5119
    },
    {
      "epoch": 0.8584302630200147,
      "grad_norm": 0.42438197135925293,
      "learning_rate": 1.046587932917671e-06,
      "loss": 0.8355,
      "step": 5120
    },
    {
      "epoch": 0.8585979251807607,
      "grad_norm": 0.48464399576187134,
      "learning_rate": 1.0441705751670073e-06,
      "loss": 0.8473,
      "step": 5121
    },
    {
      "epoch": 0.8587655873415069,
      "grad_norm": 0.6867997646331787,
      "learning_rate": 1.041755858601341e-06,
      "loss": 0.933,
      "step": 5122
    },
    {
      "epoch": 0.8589332495022529,
      "grad_norm": 0.36188092827796936,
      "learning_rate": 1.039343783932798e-06,
      "loss": 0.8585,
      "step": 5123
    },
    {
      "epoch": 0.8591009116629991,
      "grad_norm": 0.45864078402519226,
      "learning_rate": 1.0369343518727338e-06,
      "loss": 0.8093,
      "step": 5124
    },
    {
      "epoch": 0.8592685738237451,
      "grad_norm": 0.5093013644218445,
      "learning_rate": 1.0345275631317165e-06,
      "loss": 0.8529,
      "step": 5125
    },
    {
      "epoch": 0.8594362359844913,
      "grad_norm": 0.4407593309879303,
      "learning_rate": 1.0321234184195372e-06,
      "loss": 0.859,
      "step": 5126
    },
    {
      "epoch": 0.8596038981452373,
      "grad_norm": 0.6633378267288208,
      "learning_rate": 1.029721918445209e-06,
      "loss": 0.9038,
      "step": 5127
    },
    {
      "epoch": 0.8597715603059835,
      "grad_norm": 0.38820528984069824,
      "learning_rate": 1.0273230639169663e-06,
      "loss": 0.8278,
      "step": 5128
    },
    {
      "epoch": 0.8599392224667295,
      "grad_norm": 0.3281702399253845,
      "learning_rate": 1.0249268555422597e-06,
      "loss": 0.8391,
      "step": 5129
    },
    {
      "epoch": 0.8601068846274756,
      "grad_norm": 0.3887026906013489,
      "learning_rate": 1.0225332940277577e-06,
      "loss": 0.7809,
      "step": 5130
    },
    {
      "epoch": 0.8602745467882217,
      "grad_norm": 0.40416014194488525,
      "learning_rate": 1.0201423800793553e-06,
      "loss": 0.8301,
      "step": 5131
    },
    {
      "epoch": 0.8604422089489678,
      "grad_norm": 0.5501821637153625,
      "learning_rate": 1.0177541144021596e-06,
      "loss": 0.9014,
      "step": 5132
    },
    {
      "epoch": 0.8606098711097139,
      "grad_norm": 0.4721451997756958,
      "learning_rate": 1.0153684977005018e-06,
      "loss": 0.8609,
      "step": 5133
    },
    {
      "epoch": 0.86077753327046,
      "grad_norm": 0.5385779738426208,
      "learning_rate": 1.0129855306779301e-06,
      "loss": 0.7811,
      "step": 5134
    },
    {
      "epoch": 0.8609451954312061,
      "grad_norm": 0.39139533042907715,
      "learning_rate": 1.010605214037208e-06,
      "loss": 0.8845,
      "step": 5135
    },
    {
      "epoch": 0.8611128575919522,
      "grad_norm": 0.34860363602638245,
      "learning_rate": 1.008227548480324e-06,
      "loss": 0.8886,
      "step": 5136
    },
    {
      "epoch": 0.8612805197526983,
      "grad_norm": 0.3745764195919037,
      "learning_rate": 1.005852534708477e-06,
      "loss": 0.7613,
      "step": 5137
    },
    {
      "epoch": 0.8614481819134444,
      "grad_norm": 0.33172425627708435,
      "learning_rate": 1.0034801734220922e-06,
      "loss": 0.7259,
      "step": 5138
    },
    {
      "epoch": 0.8616158440741905,
      "grad_norm": 0.4795641005039215,
      "learning_rate": 1.001110465320807e-06,
      "loss": 0.8696,
      "step": 5139
    },
    {
      "epoch": 0.8617835062349366,
      "grad_norm": 0.363203227519989,
      "learning_rate": 9.987434111034756e-07,
      "loss": 0.8035,
      "step": 5140
    },
    {
      "epoch": 0.8619511683956828,
      "grad_norm": 0.44719621539115906,
      "learning_rate": 9.96379011468175e-07,
      "loss": 0.8841,
      "step": 5141
    },
    {
      "epoch": 0.8621188305564288,
      "grad_norm": 0.3977639079093933,
      "learning_rate": 9.940172671121929e-07,
      "loss": 0.8901,
      "step": 5142
    },
    {
      "epoch": 0.862286492717175,
      "grad_norm": 0.4082188010215759,
      "learning_rate": 9.91658178732038e-07,
      "loss": 0.7791,
      "step": 5143
    },
    {
      "epoch": 0.862454154877921,
      "grad_norm": 0.4015042185783386,
      "learning_rate": 9.893017470234356e-07,
      "loss": 0.9356,
      "step": 5144
    },
    {
      "epoch": 0.8626218170386671,
      "grad_norm": 0.5281703472137451,
      "learning_rate": 9.869479726813236e-07,
      "loss": 0.7711,
      "step": 5145
    },
    {
      "epoch": 0.8627894791994132,
      "grad_norm": 0.8039502501487732,
      "learning_rate": 9.845968563998621e-07,
      "loss": 0.8713,
      "step": 5146
    },
    {
      "epoch": 0.8629571413601593,
      "grad_norm": 0.4611329436302185,
      "learning_rate": 9.822483988724273e-07,
      "loss": 0.8137,
      "step": 5147
    },
    {
      "epoch": 0.8631248035209054,
      "grad_norm": 0.5646185278892517,
      "learning_rate": 9.799026007916012e-07,
      "loss": 0.7827,
      "step": 5148
    },
    {
      "epoch": 0.8632924656816515,
      "grad_norm": 0.5700774788856506,
      "learning_rate": 9.775594628491913e-07,
      "loss": 0.904,
      "step": 5149
    },
    {
      "epoch": 0.8634601278423976,
      "grad_norm": 0.3906707167625427,
      "learning_rate": 9.752189857362226e-07,
      "loss": 0.8337,
      "step": 5150
    },
    {
      "epoch": 0.8636277900031437,
      "grad_norm": 0.41302984952926636,
      "learning_rate": 9.728811701429242e-07,
      "loss": 0.8204,
      "step": 5151
    },
    {
      "epoch": 0.8637954521638898,
      "grad_norm": 0.49759745597839355,
      "learning_rate": 9.705460167587521e-07,
      "loss": 0.7336,
      "step": 5152
    },
    {
      "epoch": 0.8639631143246359,
      "grad_norm": 0.39098086953163147,
      "learning_rate": 9.68213526272369e-07,
      "loss": 0.8366,
      "step": 5153
    },
    {
      "epoch": 0.864130776485382,
      "grad_norm": 0.3385332226753235,
      "learning_rate": 9.658836993716548e-07,
      "loss": 0.7663,
      "step": 5154
    },
    {
      "epoch": 0.8642984386461281,
      "grad_norm": 0.3982413709163666,
      "learning_rate": 9.635565367437073e-07,
      "loss": 0.9586,
      "step": 5155
    },
    {
      "epoch": 0.8644661008068741,
      "grad_norm": 0.4666019082069397,
      "learning_rate": 9.612320390748331e-07,
      "loss": 0.7692,
      "step": 5156
    },
    {
      "epoch": 0.8646337629676203,
      "grad_norm": 0.9053720235824585,
      "learning_rate": 9.58910207050555e-07,
      "loss": 0.8789,
      "step": 5157
    },
    {
      "epoch": 0.8648014251283663,
      "grad_norm": 0.3826894760131836,
      "learning_rate": 9.565910413556123e-07,
      "loss": 0.848,
      "step": 5158
    },
    {
      "epoch": 0.8649690872891125,
      "grad_norm": 0.3934033513069153,
      "learning_rate": 9.542745426739518e-07,
      "loss": 0.8458,
      "step": 5159
    },
    {
      "epoch": 0.8651367494498585,
      "grad_norm": 0.4008258581161499,
      "learning_rate": 9.519607116887431e-07,
      "loss": 0.8578,
      "step": 5160
    },
    {
      "epoch": 0.8653044116106047,
      "grad_norm": 0.44706180691719055,
      "learning_rate": 9.49649549082361e-07,
      "loss": 0.8339,
      "step": 5161
    },
    {
      "epoch": 0.8654720737713507,
      "grad_norm": 0.7951359748840332,
      "learning_rate": 9.473410555363927e-07,
      "loss": 0.8574,
      "step": 5162
    },
    {
      "epoch": 0.8656397359320969,
      "grad_norm": 0.4611322581768036,
      "learning_rate": 9.450352317316469e-07,
      "loss": 0.8483,
      "step": 5163
    },
    {
      "epoch": 0.8658073980928429,
      "grad_norm": 0.4625532925128937,
      "learning_rate": 9.427320783481353e-07,
      "loss": 0.8936,
      "step": 5164
    },
    {
      "epoch": 0.8659750602535891,
      "grad_norm": 0.37857306003570557,
      "learning_rate": 9.4043159606509e-07,
      "loss": 0.8119,
      "step": 5165
    },
    {
      "epoch": 0.8661427224143351,
      "grad_norm": 0.5109477639198303,
      "learning_rate": 9.381337855609507e-07,
      "loss": 0.8802,
      "step": 5166
    },
    {
      "epoch": 0.8663103845750812,
      "grad_norm": 1.7759164571762085,
      "learning_rate": 9.358386475133674e-07,
      "loss": 0.9155,
      "step": 5167
    },
    {
      "epoch": 0.8664780467358273,
      "grad_norm": 0.3659452497959137,
      "learning_rate": 9.335461825992054e-07,
      "loss": 0.8481,
      "step": 5168
    },
    {
      "epoch": 0.8666457088965734,
      "grad_norm": 0.4770681858062744,
      "learning_rate": 9.312563914945461e-07,
      "loss": 0.7682,
      "step": 5169
    },
    {
      "epoch": 0.8668133710573195,
      "grad_norm": 0.8329793214797974,
      "learning_rate": 9.289692748746726e-07,
      "loss": 0.9175,
      "step": 5170
    },
    {
      "epoch": 0.8669810332180656,
      "grad_norm": 0.441036581993103,
      "learning_rate": 9.266848334140843e-07,
      "loss": 0.8101,
      "step": 5171
    },
    {
      "epoch": 0.8671486953788117,
      "grad_norm": 0.4591298997402191,
      "learning_rate": 9.244030677864901e-07,
      "loss": 0.9234,
      "step": 5172
    },
    {
      "epoch": 0.8673163575395578,
      "grad_norm": 0.39525070786476135,
      "learning_rate": 9.221239786648139e-07,
      "loss": 0.7818,
      "step": 5173
    },
    {
      "epoch": 0.8674840197003039,
      "grad_norm": 0.3860989511013031,
      "learning_rate": 9.198475667211881e-07,
      "loss": 0.851,
      "step": 5174
    },
    {
      "epoch": 0.86765168186105,
      "grad_norm": 0.4074042737483978,
      "learning_rate": 9.175738326269501e-07,
      "loss": 0.8905,
      "step": 5175
    },
    {
      "epoch": 0.8678193440217961,
      "grad_norm": 0.3486952483654022,
      "learning_rate": 9.153027770526546e-07,
      "loss": 0.8109,
      "step": 5176
    },
    {
      "epoch": 0.8679870061825422,
      "grad_norm": 0.3882848918437958,
      "learning_rate": 9.130344006680658e-07,
      "loss": 0.7736,
      "step": 5177
    },
    {
      "epoch": 0.8681546683432882,
      "grad_norm": 0.3613647520542145,
      "learning_rate": 9.107687041421531e-07,
      "loss": 0.765,
      "step": 5178
    },
    {
      "epoch": 0.8683223305040344,
      "grad_norm": 0.3715355396270752,
      "learning_rate": 9.085056881431009e-07,
      "loss": 0.9118,
      "step": 5179
    },
    {
      "epoch": 0.8684899926647804,
      "grad_norm": 0.5110697150230408,
      "learning_rate": 9.062453533383009e-07,
      "loss": 0.8445,
      "step": 5180
    },
    {
      "epoch": 0.8686576548255266,
      "grad_norm": 0.8367433547973633,
      "learning_rate": 9.039877003943497e-07,
      "loss": 0.8523,
      "step": 5181
    },
    {
      "epoch": 0.8688253169862726,
      "grad_norm": 0.4580248296260834,
      "learning_rate": 9.017327299770617e-07,
      "loss": 0.868,
      "step": 5182
    },
    {
      "epoch": 0.8689929791470188,
      "grad_norm": 0.9230180978775024,
      "learning_rate": 8.994804427514514e-07,
      "loss": 0.7957,
      "step": 5183
    },
    {
      "epoch": 0.8691606413077648,
      "grad_norm": 0.37561511993408203,
      "learning_rate": 8.972308393817519e-07,
      "loss": 0.9475,
      "step": 5184
    },
    {
      "epoch": 0.869328303468511,
      "grad_norm": 0.35868212580680847,
      "learning_rate": 8.949839205313948e-07,
      "loss": 0.7533,
      "step": 5185
    },
    {
      "epoch": 0.869495965629257,
      "grad_norm": 0.6453197598457336,
      "learning_rate": 8.927396868630234e-07,
      "loss": 0.9276,
      "step": 5186
    },
    {
      "epoch": 0.8696636277900032,
      "grad_norm": 0.5974776744842529,
      "learning_rate": 8.904981390384937e-07,
      "loss": 0.8797,
      "step": 5187
    },
    {
      "epoch": 0.8698312899507492,
      "grad_norm": 0.7530732154846191,
      "learning_rate": 8.882592777188636e-07,
      "loss": 0.7779,
      "step": 5188
    },
    {
      "epoch": 0.8699989521114954,
      "grad_norm": 0.4910556674003601,
      "learning_rate": 8.860231035644007e-07,
      "loss": 0.8568,
      "step": 5189
    },
    {
      "epoch": 0.8701666142722414,
      "grad_norm": 0.39665699005126953,
      "learning_rate": 8.837896172345827e-07,
      "loss": 0.8578,
      "step": 5190
    },
    {
      "epoch": 0.8703342764329876,
      "grad_norm": 0.3718913793563843,
      "learning_rate": 8.815588193880887e-07,
      "loss": 0.7943,
      "step": 5191
    },
    {
      "epoch": 0.8705019385937336,
      "grad_norm": 0.3665577471256256,
      "learning_rate": 8.793307106828119e-07,
      "loss": 0.903,
      "step": 5192
    },
    {
      "epoch": 0.8706696007544797,
      "grad_norm": 0.38223451375961304,
      "learning_rate": 8.771052917758505e-07,
      "loss": 0.8133,
      "step": 5193
    },
    {
      "epoch": 0.8708372629152258,
      "grad_norm": 0.45462703704833984,
      "learning_rate": 8.748825633235026e-07,
      "loss": 0.8094,
      "step": 5194
    },
    {
      "epoch": 0.8710049250759719,
      "grad_norm": 0.6003639101982117,
      "learning_rate": 8.726625259812816e-07,
      "loss": 0.8195,
      "step": 5195
    },
    {
      "epoch": 0.871172587236718,
      "grad_norm": 0.39782842993736267,
      "learning_rate": 8.70445180403905e-07,
      "loss": 0.8417,
      "step": 5196
    },
    {
      "epoch": 0.8713402493974641,
      "grad_norm": 0.399222731590271,
      "learning_rate": 8.68230527245294e-07,
      "loss": 0.844,
      "step": 5197
    },
    {
      "epoch": 0.8715079115582102,
      "grad_norm": 0.4528278112411499,
      "learning_rate": 8.660185671585764e-07,
      "loss": 0.8549,
      "step": 5198
    },
    {
      "epoch": 0.8716755737189563,
      "grad_norm": 0.3690025210380554,
      "learning_rate": 8.638093007960891e-07,
      "loss": 0.8686,
      "step": 5199
    },
    {
      "epoch": 0.8718432358797024,
      "grad_norm": 0.43686869740486145,
      "learning_rate": 8.616027288093698e-07,
      "loss": 0.7774,
      "step": 5200
    },
    {
      "epoch": 0.8720108980404485,
      "grad_norm": 0.6555036902427673,
      "learning_rate": 8.593988518491658e-07,
      "loss": 0.9075,
      "step": 5201
    },
    {
      "epoch": 0.8721785602011946,
      "grad_norm": 0.32533329725265503,
      "learning_rate": 8.571976705654273e-07,
      "loss": 0.7884,
      "step": 5202
    },
    {
      "epoch": 0.8723462223619407,
      "grad_norm": 0.7068662643432617,
      "learning_rate": 8.54999185607307e-07,
      "loss": 0.7727,
      "step": 5203
    },
    {
      "epoch": 0.8725138845226867,
      "grad_norm": 0.6542850732803345,
      "learning_rate": 8.528033976231698e-07,
      "loss": 0.8242,
      "step": 5204
    },
    {
      "epoch": 0.8726815466834329,
      "grad_norm": 0.411618709564209,
      "learning_rate": 8.506103072605776e-07,
      "loss": 0.998,
      "step": 5205
    },
    {
      "epoch": 0.8728492088441789,
      "grad_norm": 0.3620193600654602,
      "learning_rate": 8.484199151663019e-07,
      "loss": 0.7989,
      "step": 5206
    },
    {
      "epoch": 0.8730168710049251,
      "grad_norm": 0.5024863481521606,
      "learning_rate": 8.462322219863151e-07,
      "loss": 0.8511,
      "step": 5207
    },
    {
      "epoch": 0.8731845331656711,
      "grad_norm": 0.4129505157470703,
      "learning_rate": 8.440472283657941e-07,
      "loss": 0.715,
      "step": 5208
    },
    {
      "epoch": 0.8733521953264173,
      "grad_norm": 0.3400903642177582,
      "learning_rate": 8.418649349491248e-07,
      "loss": 0.852,
      "step": 5209
    },
    {
      "epoch": 0.8735198574871633,
      "grad_norm": 0.607019305229187,
      "learning_rate": 8.396853423798867e-07,
      "loss": 0.7937,
      "step": 5210
    },
    {
      "epoch": 0.8736875196479095,
      "grad_norm": 0.3749978840351105,
      "learning_rate": 8.375084513008736e-07,
      "loss": 0.8734,
      "step": 5211
    },
    {
      "epoch": 0.8738551818086555,
      "grad_norm": 0.3694119155406952,
      "learning_rate": 8.35334262354075e-07,
      "loss": 0.8821,
      "step": 5212
    },
    {
      "epoch": 0.8740228439694017,
      "grad_norm": 0.5530465841293335,
      "learning_rate": 8.33162776180686e-07,
      "loss": 0.7952,
      "step": 5213
    },
    {
      "epoch": 0.8741905061301477,
      "grad_norm": 0.3724326491355896,
      "learning_rate": 8.309939934211053e-07,
      "loss": 0.8398,
      "step": 5214
    },
    {
      "epoch": 0.8743581682908939,
      "grad_norm": 0.4362427890300751,
      "learning_rate": 8.288279147149369e-07,
      "loss": 0.8927,
      "step": 5215
    },
    {
      "epoch": 0.8745258304516399,
      "grad_norm": 0.42852744460105896,
      "learning_rate": 8.266645407009788e-07,
      "loss": 0.9007,
      "step": 5216
    },
    {
      "epoch": 0.874693492612386,
      "grad_norm": 0.43490931391716003,
      "learning_rate": 8.245038720172393e-07,
      "loss": 0.8715,
      "step": 5217
    },
    {
      "epoch": 0.8748611547731321,
      "grad_norm": 0.8107596039772034,
      "learning_rate": 8.223459093009279e-07,
      "loss": 0.8452,
      "step": 5218
    },
    {
      "epoch": 0.8750288169338782,
      "grad_norm": 0.42709115147590637,
      "learning_rate": 8.201906531884507e-07,
      "loss": 0.8322,
      "step": 5219
    },
    {
      "epoch": 0.8751964790946243,
      "grad_norm": 0.4896807074546814,
      "learning_rate": 8.180381043154262e-07,
      "loss": 0.9062,
      "step": 5220
    },
    {
      "epoch": 0.8753641412553704,
      "grad_norm": 0.4519088566303253,
      "learning_rate": 8.158882633166587e-07,
      "loss": 0.7793,
      "step": 5221
    },
    {
      "epoch": 0.8755318034161165,
      "grad_norm": 0.36894258856773376,
      "learning_rate": 8.137411308261689e-07,
      "loss": 0.8589,
      "step": 5222
    },
    {
      "epoch": 0.8756994655768626,
      "grad_norm": 0.41007664799690247,
      "learning_rate": 8.115967074771714e-07,
      "loss": 0.8599,
      "step": 5223
    },
    {
      "epoch": 0.8758671277376087,
      "grad_norm": 0.366462767124176,
      "learning_rate": 8.094549939020824e-07,
      "loss": 0.8493,
      "step": 5224
    },
    {
      "epoch": 0.8760347898983548,
      "grad_norm": 0.3870672583580017,
      "learning_rate": 8.073159907325223e-07,
      "loss": 0.8376,
      "step": 5225
    },
    {
      "epoch": 0.876202452059101,
      "grad_norm": 0.3918464779853821,
      "learning_rate": 8.051796985993077e-07,
      "loss": 0.8635,
      "step": 5226
    },
    {
      "epoch": 0.876370114219847,
      "grad_norm": 0.39704617857933044,
      "learning_rate": 8.030461181324567e-07,
      "loss": 0.8127,
      "step": 5227
    },
    {
      "epoch": 0.8765377763805932,
      "grad_norm": 0.9166821241378784,
      "learning_rate": 8.009152499611905e-07,
      "loss": 0.892,
      "step": 5228
    },
    {
      "epoch": 0.8767054385413392,
      "grad_norm": 0.6163578629493713,
      "learning_rate": 7.987870947139276e-07,
      "loss": 0.8442,
      "step": 5229
    },
    {
      "epoch": 0.8768731007020854,
      "grad_norm": 0.8241729140281677,
      "learning_rate": 7.966616530182858e-07,
      "loss": 0.9117,
      "step": 5230
    },
    {
      "epoch": 0.8770407628628314,
      "grad_norm": 0.7489495277404785,
      "learning_rate": 7.945389255010872e-07,
      "loss": 0.9063,
      "step": 5231
    },
    {
      "epoch": 0.8772084250235775,
      "grad_norm": 1.046480655670166,
      "learning_rate": 7.924189127883475e-07,
      "loss": 0.8623,
      "step": 5232
    },
    {
      "epoch": 0.8773760871843236,
      "grad_norm": 0.3621228039264679,
      "learning_rate": 7.903016155052867e-07,
      "loss": 0.8169,
      "step": 5233
    },
    {
      "epoch": 0.8775437493450697,
      "grad_norm": 0.36051076650619507,
      "learning_rate": 7.881870342763243e-07,
      "loss": 0.8182,
      "step": 5234
    },
    {
      "epoch": 0.8777114115058158,
      "grad_norm": 0.3524509072303772,
      "learning_rate": 7.860751697250702e-07,
      "loss": 0.7873,
      "step": 5235
    },
    {
      "epoch": 0.8778790736665619,
      "grad_norm": 0.3441102206707001,
      "learning_rate": 7.83966022474345e-07,
      "loss": 0.7736,
      "step": 5236
    },
    {
      "epoch": 0.878046735827308,
      "grad_norm": 0.3441102206707001,
      "learning_rate": 7.83966022474345e-07,
      "loss": 0.859,
      "step": 5237
    },
    {
      "epoch": 0.8782143979880541,
      "grad_norm": 0.5010706186294556,
      "learning_rate": 7.818595931461581e-07,
      "loss": 0.8187,
      "step": 5238
    },
    {
      "epoch": 0.8783820601488002,
      "grad_norm": 0.40332692861557007,
      "learning_rate": 7.797558823617257e-07,
      "loss": 0.9172,
      "step": 5239
    },
    {
      "epoch": 0.8785497223095463,
      "grad_norm": 0.476007342338562,
      "learning_rate": 7.77654890741456e-07,
      "loss": 0.7967,
      "step": 5240
    },
    {
      "epoch": 0.8787173844702924,
      "grad_norm": 0.3826432526111603,
      "learning_rate": 7.755566189049557e-07,
      "loss": 0.859,
      "step": 5241
    },
    {
      "epoch": 0.8788850466310385,
      "grad_norm": 0.35931214690208435,
      "learning_rate": 7.73461067471033e-07,
      "loss": 0.7708,
      "step": 5242
    },
    {
      "epoch": 0.8790527087917845,
      "grad_norm": 0.3573903441429138,
      "learning_rate": 7.713682370576947e-07,
      "loss": 0.9159,
      "step": 5243
    },
    {
      "epoch": 0.8792203709525307,
      "grad_norm": 0.4550726115703583,
      "learning_rate": 7.692781282821383e-07,
      "loss": 0.8461,
      "step": 5244
    },
    {
      "epoch": 0.8793880331132767,
      "grad_norm": 0.6071157455444336,
      "learning_rate": 7.671907417607638e-07,
      "loss": 0.8955,
      "step": 5245
    },
    {
      "epoch": 0.8795556952740229,
      "grad_norm": 0.4022810757160187,
      "learning_rate": 7.65106078109168e-07,
      "loss": 0.8044,
      "step": 5246
    },
    {
      "epoch": 0.8797233574347689,
      "grad_norm": 0.37888088822364807,
      "learning_rate": 7.630241379421432e-07,
      "loss": 0.8642,
      "step": 5247
    },
    {
      "epoch": 0.8798910195955151,
      "grad_norm": 0.3679821789264679,
      "learning_rate": 7.609449218736808e-07,
      "loss": 0.8851,
      "step": 5248
    },
    {
      "epoch": 0.8800586817562611,
      "grad_norm": 0.47906655073165894,
      "learning_rate": 7.588684305169669e-07,
      "loss": 0.9026,
      "step": 5249
    },
    {
      "epoch": 0.8802263439170073,
      "grad_norm": 0.4536413848400116,
      "learning_rate": 7.567946644843827e-07,
      "loss": 0.9075,
      "step": 5250
    },
    {
      "epoch": 0.8803940060777533,
      "grad_norm": 0.3367442190647125,
      "learning_rate": 7.5472362438751e-07,
      "loss": 0.7654,
      "step": 5251
    },
    {
      "epoch": 0.8805616682384995,
      "grad_norm": 0.43579649925231934,
      "learning_rate": 7.526553108371215e-07,
      "loss": 0.8099,
      "step": 5252
    },
    {
      "epoch": 0.8807293303992455,
      "grad_norm": 0.8711832165718079,
      "learning_rate": 7.505897244431914e-07,
      "loss": 0.9072,
      "step": 5253
    },
    {
      "epoch": 0.8808969925599917,
      "grad_norm": 0.38806793093681335,
      "learning_rate": 7.485268658148859e-07,
      "loss": 0.7709,
      "step": 5254
    },
    {
      "epoch": 0.8810646547207377,
      "grad_norm": 0.4234929382801056,
      "learning_rate": 7.464667355605648e-07,
      "loss": 0.8378,
      "step": 5255
    },
    {
      "epoch": 0.8812323168814838,
      "grad_norm": 0.5764008164405823,
      "learning_rate": 7.4440933428779e-07,
      "loss": 0.7818,
      "step": 5256
    },
    {
      "epoch": 0.8813999790422299,
      "grad_norm": 0.4254293143749237,
      "learning_rate": 7.423546626033129e-07,
      "loss": 0.837,
      "step": 5257
    },
    {
      "epoch": 0.881567641202976,
      "grad_norm": 0.336610347032547,
      "learning_rate": 7.403027211130786e-07,
      "loss": 0.7539,
      "step": 5258
    },
    {
      "epoch": 0.8817353033637221,
      "grad_norm": 0.377946674823761,
      "learning_rate": 7.382535104222366e-07,
      "loss": 0.8741,
      "step": 5259
    },
    {
      "epoch": 0.8819029655244682,
      "grad_norm": 0.42889639735221863,
      "learning_rate": 7.362070311351177e-07,
      "loss": 0.8826,
      "step": 5260
    },
    {
      "epoch": 0.8820706276852143,
      "grad_norm": 0.5174009203910828,
      "learning_rate": 7.341632838552582e-07,
      "loss": 0.8708,
      "step": 5261
    },
    {
      "epoch": 0.8822382898459604,
      "grad_norm": 0.4477638900279999,
      "learning_rate": 7.321222691853858e-07,
      "loss": 0.872,
      "step": 5262
    },
    {
      "epoch": 0.8824059520067065,
      "grad_norm": 0.3257504999637604,
      "learning_rate": 7.300839877274157e-07,
      "loss": 0.8529,
      "step": 5263
    },
    {
      "epoch": 0.8825736141674526,
      "grad_norm": 0.3661295175552368,
      "learning_rate": 7.28048440082465e-07,
      "loss": 0.8091,
      "step": 5264
    },
    {
      "epoch": 0.8827412763281987,
      "grad_norm": 0.49933600425720215,
      "learning_rate": 7.260156268508444e-07,
      "loss": 0.8484,
      "step": 5265
    },
    {
      "epoch": 0.8829089384889448,
      "grad_norm": 0.33709481358528137,
      "learning_rate": 7.239855486320513e-07,
      "loss": 0.8874,
      "step": 5266
    },
    {
      "epoch": 0.8830766006496908,
      "grad_norm": 0.5059892535209656,
      "learning_rate": 7.219582060247854e-07,
      "loss": 0.8608,
      "step": 5267
    },
    {
      "epoch": 0.883244262810437,
      "grad_norm": 0.43796518445014954,
      "learning_rate": 7.199335996269319e-07,
      "loss": 0.9711,
      "step": 5268
    },
    {
      "epoch": 0.883411924971183,
      "grad_norm": 0.47821682691574097,
      "learning_rate": 7.17911730035572e-07,
      "loss": 0.8459,
      "step": 5269
    },
    {
      "epoch": 0.8835795871319292,
      "grad_norm": 0.34628474712371826,
      "learning_rate": 7.158925978469833e-07,
      "loss": 0.8287,
      "step": 5270
    },
    {
      "epoch": 0.8837472492926752,
      "grad_norm": 0.429559588432312,
      "learning_rate": 7.138762036566305e-07,
      "loss": 0.8746,
      "step": 5271
    },
    {
      "epoch": 0.8839149114534214,
      "grad_norm": 0.35698437690734863,
      "learning_rate": 7.118625480591712e-07,
      "loss": 0.8658,
      "step": 5272
    },
    {
      "epoch": 0.8840825736141674,
      "grad_norm": 0.37137287855148315,
      "learning_rate": 7.098516316484628e-07,
      "loss": 0.8086,
      "step": 5273
    },
    {
      "epoch": 0.8842502357749136,
      "grad_norm": 0.6553245186805725,
      "learning_rate": 7.07843455017544e-07,
      "loss": 0.8455,
      "step": 5274
    },
    {
      "epoch": 0.8844178979356596,
      "grad_norm": 0.4681013226509094,
      "learning_rate": 7.058380187586544e-07,
      "loss": 0.7681,
      "step": 5275
    },
    {
      "epoch": 0.8845855600964058,
      "grad_norm": 0.3756442666053772,
      "learning_rate": 7.038353234632223e-07,
      "loss": 0.7792,
      "step": 5276
    },
    {
      "epoch": 0.8847532222571518,
      "grad_norm": 0.49385350942611694,
      "learning_rate": 7.018353697218638e-07,
      "loss": 0.8486,
      "step": 5277
    },
    {
      "epoch": 0.884920884417898,
      "grad_norm": 0.6206707954406738,
      "learning_rate": 6.998381581243929e-07,
      "loss": 0.8994,
      "step": 5278
    },
    {
      "epoch": 0.885088546578644,
      "grad_norm": 0.31013917922973633,
      "learning_rate": 6.978436892598106e-07,
      "loss": 0.8784,
      "step": 5279
    },
    {
      "epoch": 0.8852562087393901,
      "grad_norm": 0.34381526708602905,
      "learning_rate": 6.95851963716313e-07,
      "loss": 0.7866,
      "step": 5280
    },
    {
      "epoch": 0.8854238709001362,
      "grad_norm": 0.45185163617134094,
      "learning_rate": 6.938629820812826e-07,
      "loss": 0.7698,
      "step": 5281
    },
    {
      "epoch": 0.8855915330608823,
      "grad_norm": 0.4260341227054596,
      "learning_rate": 6.918767449412933e-07,
      "loss": 0.8308,
      "step": 5282
    },
    {
      "epoch": 0.8857591952216284,
      "grad_norm": 0.8337491154670715,
      "learning_rate": 6.898932528821123e-07,
      "loss": 0.9048,
      "step": 5283
    },
    {
      "epoch": 0.8859268573823745,
      "grad_norm": 0.43806082010269165,
      "learning_rate": 6.879125064886993e-07,
      "loss": 0.8681,
      "step": 5284
    },
    {
      "epoch": 0.8860945195431206,
      "grad_norm": 0.3748376667499542,
      "learning_rate": 6.859345063451961e-07,
      "loss": 0.8701,
      "step": 5285
    },
    {
      "epoch": 0.8862621817038667,
      "grad_norm": 0.3831605613231659,
      "learning_rate": 6.839592530349426e-07,
      "loss": 0.8497,
      "step": 5286
    },
    {
      "epoch": 0.8864298438646128,
      "grad_norm": 0.41587305068969727,
      "learning_rate": 6.819867471404628e-07,
      "loss": 0.7885,
      "step": 5287
    },
    {
      "epoch": 0.8865975060253589,
      "grad_norm": 0.43317797780036926,
      "learning_rate": 6.800169892434739e-07,
      "loss": 0.7672,
      "step": 5288
    },
    {
      "epoch": 0.886765168186105,
      "grad_norm": 0.399836003780365,
      "learning_rate": 6.780499799248852e-07,
      "loss": 0.7892,
      "step": 5289
    },
    {
      "epoch": 0.8869328303468511,
      "grad_norm": 0.3666892349720001,
      "learning_rate": 6.760857197647863e-07,
      "loss": 0.8582,
      "step": 5290
    },
    {
      "epoch": 0.8871004925075971,
      "grad_norm": 0.681106448173523,
      "learning_rate": 6.74124209342466e-07,
      "loss": 0.8437,
      "step": 5291
    },
    {
      "epoch": 0.8872681546683433,
      "grad_norm": 0.3994339108467102,
      "learning_rate": 6.721654492363972e-07,
      "loss": 0.8039,
      "step": 5292
    },
    {
      "epoch": 0.8874358168290893,
      "grad_norm": 0.4830002188682556,
      "learning_rate": 6.702094400242421e-07,
      "loss": 0.7977,
      "step": 5293
    },
    {
      "epoch": 0.8876034789898355,
      "grad_norm": 0.6147891879081726,
      "learning_rate": 6.682561822828526e-07,
      "loss": 0.8956,
      "step": 5294
    },
    {
      "epoch": 0.8877711411505815,
      "grad_norm": 0.5718827843666077,
      "learning_rate": 6.663056765882692e-07,
      "loss": 0.8331,
      "step": 5295
    },
    {
      "epoch": 0.8879388033113277,
      "grad_norm": 0.8253036737442017,
      "learning_rate": 6.64357923515716e-07,
      "loss": 0.8295,
      "step": 5296
    },
    {
      "epoch": 0.8881064654720737,
      "grad_norm": 0.36979398131370544,
      "learning_rate": 6.624129236396159e-07,
      "loss": 0.7818,
      "step": 5297
    },
    {
      "epoch": 0.8882741276328199,
      "grad_norm": 0.36641526222229004,
      "learning_rate": 6.604706775335679e-07,
      "loss": 0.7572,
      "step": 5298
    },
    {
      "epoch": 0.8884417897935659,
      "grad_norm": 0.4561346471309662,
      "learning_rate": 6.58531185770368e-07,
      "loss": 0.84,
      "step": 5299
    },
    {
      "epoch": 0.8886094519543121,
      "grad_norm": 0.3623506426811218,
      "learning_rate": 6.565944489219955e-07,
      "loss": 0.884,
      "step": 5300
    },
    {
      "epoch": 0.8887771141150581,
      "grad_norm": 0.4001149833202362,
      "learning_rate": 6.546604675596169e-07,
      "loss": 0.8588,
      "step": 5301
    },
    {
      "epoch": 0.8889447762758043,
      "grad_norm": 0.38912656903266907,
      "learning_rate": 6.52729242253588e-07,
      "loss": 0.8358,
      "step": 5302
    },
    {
      "epoch": 0.8891124384365503,
      "grad_norm": 0.4054897129535675,
      "learning_rate": 6.508007735734522e-07,
      "loss": 0.7658,
      "step": 5303
    },
    {
      "epoch": 0.8892801005972965,
      "grad_norm": 0.4189501106739044,
      "learning_rate": 6.488750620879358e-07,
      "loss": 0.883,
      "step": 5304
    },
    {
      "epoch": 0.8894477627580425,
      "grad_norm": 0.36771664023399353,
      "learning_rate": 6.469521083649588e-07,
      "loss": 0.8027,
      "step": 5305
    },
    {
      "epoch": 0.8896154249187886,
      "grad_norm": 0.37720581889152527,
      "learning_rate": 6.450319129716198e-07,
      "loss": 0.7863,
      "step": 5306
    },
    {
      "epoch": 0.8897830870795347,
      "grad_norm": 0.35849684476852417,
      "learning_rate": 6.431144764742104e-07,
      "loss": 0.825,
      "step": 5307
    },
    {
      "epoch": 0.8899507492402808,
      "grad_norm": 0.7659775614738464,
      "learning_rate": 6.411997994382102e-07,
      "loss": 0.8178,
      "step": 5308
    },
    {
      "epoch": 0.8901184114010269,
      "grad_norm": 0.41936665773391724,
      "learning_rate": 6.392878824282744e-07,
      "loss": 0.8437,
      "step": 5309
    },
    {
      "epoch": 0.890286073561773,
      "grad_norm": 0.5967978835105896,
      "learning_rate": 6.373787260082531e-07,
      "loss": 0.864,
      "step": 5310
    },
    {
      "epoch": 0.8904537357225191,
      "grad_norm": 0.4905378520488739,
      "learning_rate": 6.354723307411837e-07,
      "loss": 0.7812,
      "step": 5311
    },
    {
      "epoch": 0.8906213978832652,
      "grad_norm": 0.34936320781707764,
      "learning_rate": 6.33568697189283e-07,
      "loss": 0.8333,
      "step": 5312
    },
    {
      "epoch": 0.8907890600440114,
      "grad_norm": 0.760906994342804,
      "learning_rate": 6.316678259139542e-07,
      "loss": 0.8622,
      "step": 5313
    },
    {
      "epoch": 0.8909567222047574,
      "grad_norm": 0.44719940423965454,
      "learning_rate": 6.29769717475791e-07,
      "loss": 0.8796,
      "step": 5314
    },
    {
      "epoch": 0.8911243843655036,
      "grad_norm": 0.4683433771133423,
      "learning_rate": 6.278743724345682e-07,
      "loss": 0.8847,
      "step": 5315
    },
    {
      "epoch": 0.8912920465262496,
      "grad_norm": 0.40432867407798767,
      "learning_rate": 6.259817913492461e-07,
      "loss": 0.8508,
      "step": 5316
    },
    {
      "epoch": 0.8914597086869958,
      "grad_norm": 0.4725228548049927,
      "learning_rate": 6.240919747779705e-07,
      "loss": 0.8185,
      "step": 5317
    },
    {
      "epoch": 0.8916273708477418,
      "grad_norm": 0.3801216185092926,
      "learning_rate": 6.222049232780714e-07,
      "loss": 0.7974,
      "step": 5318
    },
    {
      "epoch": 0.891795033008488,
      "grad_norm": 0.3568783402442932,
      "learning_rate": 6.203206374060644e-07,
      "loss": 0.8628,
      "step": 5319
    },
    {
      "epoch": 0.891962695169234,
      "grad_norm": 0.33219611644744873,
      "learning_rate": 6.184391177176474e-07,
      "loss": 0.866,
      "step": 5320
    },
    {
      "epoch": 0.8921303573299801,
      "grad_norm": 0.4283650517463684,
      "learning_rate": 6.165603647677054e-07,
      "loss": 0.7798,
      "step": 5321
    },
    {
      "epoch": 0.8922980194907262,
      "grad_norm": 0.43948283791542053,
      "learning_rate": 6.146843791103052e-07,
      "loss": 0.7352,
      "step": 5322
    },
    {
      "epoch": 0.8924656816514723,
      "grad_norm": 0.39937087893486023,
      "learning_rate": 6.128111612986975e-07,
      "loss": 0.8576,
      "step": 5323
    },
    {
      "epoch": 0.8926333438122184,
      "grad_norm": 0.4143848717212677,
      "learning_rate": 6.109407118853195e-07,
      "loss": 0.8693,
      "step": 5324
    },
    {
      "epoch": 0.8928010059729645,
      "grad_norm": 1.0609921216964722,
      "learning_rate": 6.090730314217863e-07,
      "loss": 0.872,
      "step": 5325
    },
    {
      "epoch": 0.8929686681337106,
      "grad_norm": 0.4229665994644165,
      "learning_rate": 6.072081204589042e-07,
      "loss": 0.828,
      "step": 5326
    },
    {
      "epoch": 0.8931363302944567,
      "grad_norm": 0.3355676233768463,
      "learning_rate": 6.053459795466565e-07,
      "loss": 0.8271,
      "step": 5327
    },
    {
      "epoch": 0.8933039924552028,
      "grad_norm": 0.48312902450561523,
      "learning_rate": 6.034866092342107e-07,
      "loss": 0.7578,
      "step": 5328
    },
    {
      "epoch": 0.8934716546159489,
      "grad_norm": 0.46982505917549133,
      "learning_rate": 6.016300100699191e-07,
      "loss": 0.8265,
      "step": 5329
    },
    {
      "epoch": 0.893639316776695,
      "grad_norm": 0.40215837955474854,
      "learning_rate": 5.997761826013182e-07,
      "loss": 0.8408,
      "step": 5330
    },
    {
      "epoch": 0.8938069789374411,
      "grad_norm": 0.48121073842048645,
      "learning_rate": 5.979251273751197e-07,
      "loss": 0.8354,
      "step": 5331
    },
    {
      "epoch": 0.8939746410981871,
      "grad_norm": 0.3733910918235779,
      "learning_rate": 5.960768449372246e-07,
      "loss": 0.8275,
      "step": 5332
    },
    {
      "epoch": 0.8941423032589333,
      "grad_norm": 0.5000764727592468,
      "learning_rate": 5.942313358327167e-07,
      "loss": 0.7803,
      "step": 5333
    },
    {
      "epoch": 0.8943099654196793,
      "grad_norm": 0.4887655973434448,
      "learning_rate": 5.923886006058566e-07,
      "loss": 0.8851,
      "step": 5334
    },
    {
      "epoch": 0.8944776275804255,
      "grad_norm": 0.3644387722015381,
      "learning_rate": 5.905486398000937e-07,
      "loss": 0.8665,
      "step": 5335
    },
    {
      "epoch": 0.8946452897411715,
      "grad_norm": 0.412713885307312,
      "learning_rate": 5.887114539580497e-07,
      "loss": 0.7574,
      "step": 5336
    },
    {
      "epoch": 0.8948129519019177,
      "grad_norm": 0.3497019112110138,
      "learning_rate": 5.868770436215366e-07,
      "loss": 0.8163,
      "step": 5337
    },
    {
      "epoch": 0.8949806140626637,
      "grad_norm": 0.35792189836502075,
      "learning_rate": 5.85045409331545e-07,
      "loss": 0.8441,
      "step": 5338
    },
    {
      "epoch": 0.8951482762234099,
      "grad_norm": 0.4269263446331024,
      "learning_rate": 5.83216551628244e-07,
      "loss": 0.8261,
      "step": 5339
    },
    {
      "epoch": 0.8953159383841559,
      "grad_norm": 0.4859539270401001,
      "learning_rate": 5.813904710509899e-07,
      "loss": 0.8215,
      "step": 5340
    },
    {
      "epoch": 0.8954836005449021,
      "grad_norm": 0.36638912558555603,
      "learning_rate": 5.795671681383152e-07,
      "loss": 0.8916,
      "step": 5341
    },
    {
      "epoch": 0.8956512627056481,
      "grad_norm": 0.4190085530281067,
      "learning_rate": 5.777466434279322e-07,
      "loss": 0.8154,
      "step": 5342
    },
    {
      "epoch": 0.8958189248663943,
      "grad_norm": 0.41331014037132263,
      "learning_rate": 5.7592889745674e-07,
      "loss": 0.8592,
      "step": 5343
    },
    {
      "epoch": 0.8959865870271403,
      "grad_norm": 0.36141493916511536,
      "learning_rate": 5.741139307608124e-07,
      "loss": 0.8003,
      "step": 5344
    },
    {
      "epoch": 0.8961542491878864,
      "grad_norm": 0.5100393891334534,
      "learning_rate": 5.723017438754042e-07,
      "loss": 0.8431,
      "step": 5345
    },
    {
      "epoch": 0.8963219113486325,
      "grad_norm": 0.3829497694969177,
      "learning_rate": 5.704923373349536e-07,
      "loss": 0.8047,
      "step": 5346
    },
    {
      "epoch": 0.8964895735093786,
      "grad_norm": 0.5301305651664734,
      "learning_rate": 5.68685711673076e-07,
      "loss": 0.7256,
      "step": 5347
    },
    {
      "epoch": 0.8966572356701247,
      "grad_norm": 0.3523816168308258,
      "learning_rate": 5.668818674225684e-07,
      "loss": 0.8074,
      "step": 5348
    },
    {
      "epoch": 0.8968248978308708,
      "grad_norm": 0.41455909609794617,
      "learning_rate": 5.650808051154089e-07,
      "loss": 0.8822,
      "step": 5349
    },
    {
      "epoch": 0.8969925599916169,
      "grad_norm": 0.4645654559135437,
      "learning_rate": 5.632825252827479e-07,
      "loss": 0.934,
      "step": 5350
    },
    {
      "epoch": 0.897160222152363,
      "grad_norm": 0.5317776799201965,
      "learning_rate": 5.614870284549256e-07,
      "loss": 0.8398,
      "step": 5351
    },
    {
      "epoch": 0.8973278843131091,
      "grad_norm": 0.5853843092918396,
      "learning_rate": 5.59694315161452e-07,
      "loss": 0.8806,
      "step": 5352
    },
    {
      "epoch": 0.8974955464738552,
      "grad_norm": 0.39755651354789734,
      "learning_rate": 5.57904385931024e-07,
      "loss": 0.8073,
      "step": 5353
    },
    {
      "epoch": 0.8976632086346013,
      "grad_norm": 0.3885393440723419,
      "learning_rate": 5.561172412915117e-07,
      "loss": 0.8263,
      "step": 5354
    },
    {
      "epoch": 0.8978308707953474,
      "grad_norm": 0.6345336437225342,
      "learning_rate": 5.543328817699667e-07,
      "loss": 0.8009,
      "step": 5355
    },
    {
      "epoch": 0.8979985329560934,
      "grad_norm": 0.5248864889144897,
      "learning_rate": 5.525513078926171e-07,
      "loss": 0.8164,
      "step": 5356
    },
    {
      "epoch": 0.8981661951168396,
      "grad_norm": 0.49112460017204285,
      "learning_rate": 5.507725201848757e-07,
      "loss": 0.7499,
      "step": 5357
    },
    {
      "epoch": 0.8983338572775856,
      "grad_norm": 0.4366603493690491,
      "learning_rate": 5.489965191713254e-07,
      "loss": 0.8424,
      "step": 5358
    },
    {
      "epoch": 0.8985015194383318,
      "grad_norm": 0.4593455195426941,
      "learning_rate": 5.472233053757303e-07,
      "loss": 0.8159,
      "step": 5359
    },
    {
      "epoch": 0.8986691815990778,
      "grad_norm": 0.4607357680797577,
      "learning_rate": 5.454528793210356e-07,
      "loss": 0.8151,
      "step": 5360
    },
    {
      "epoch": 0.898836843759824,
      "grad_norm": 0.4741382598876953,
      "learning_rate": 5.436852415293592e-07,
      "loss": 0.8323,
      "step": 5361
    },
    {
      "epoch": 0.89900450592057,
      "grad_norm": 0.5529037714004517,
      "learning_rate": 5.419203925220029e-07,
      "loss": 0.9415,
      "step": 5362
    },
    {
      "epoch": 0.8991721680813162,
      "grad_norm": 0.5256397128105164,
      "learning_rate": 5.401583328194404e-07,
      "loss": 0.8095,
      "step": 5363
    },
    {
      "epoch": 0.8993398302420622,
      "grad_norm": 0.3243965804576874,
      "learning_rate": 5.383990629413227e-07,
      "loss": 0.8096,
      "step": 5364
    },
    {
      "epoch": 0.8995074924028084,
      "grad_norm": 1.1402196884155273,
      "learning_rate": 5.366425834064848e-07,
      "loss": 0.7679,
      "step": 5365
    },
    {
      "epoch": 0.8996751545635544,
      "grad_norm": 0.37329888343811035,
      "learning_rate": 5.348888947329312e-07,
      "loss": 0.8054,
      "step": 5366
    },
    {
      "epoch": 0.8998428167243006,
      "grad_norm": 0.4582318961620331,
      "learning_rate": 5.33137997437847e-07,
      "loss": 0.8313,
      "step": 5367
    },
    {
      "epoch": 0.9000104788850466,
      "grad_norm": 0.39775338768959045,
      "learning_rate": 5.313898920375948e-07,
      "loss": 0.8463,
      "step": 5368
    },
    {
      "epoch": 0.9001781410457927,
      "grad_norm": 0.4152122437953949,
      "learning_rate": 5.296445790477101e-07,
      "loss": 0.8729,
      "step": 5369
    },
    {
      "epoch": 0.9003458032065388,
      "grad_norm": 0.7189455628395081,
      "learning_rate": 5.279020589829099e-07,
      "loss": 0.8718,
      "step": 5370
    },
    {
      "epoch": 0.9005134653672849,
      "grad_norm": 0.3881506621837616,
      "learning_rate": 5.26162332357083e-07,
      "loss": 0.7515,
      "step": 5371
    },
    {
      "epoch": 0.900681127528031,
      "grad_norm": 0.407922625541687,
      "learning_rate": 5.244253996832971e-07,
      "loss": 0.9079,
      "step": 5372
    },
    {
      "epoch": 0.9008487896887771,
      "grad_norm": 0.3899659812450409,
      "learning_rate": 5.226912614737956e-07,
      "loss": 0.8661,
      "step": 5373
    },
    {
      "epoch": 0.9010164518495232,
      "grad_norm": 0.42218416929244995,
      "learning_rate": 5.20959918239996e-07,
      "loss": 0.7457,
      "step": 5374
    },
    {
      "epoch": 0.9011841140102693,
      "grad_norm": 0.5291997790336609,
      "learning_rate": 5.192313704924933e-07,
      "loss": 0.8461,
      "step": 5375
    },
    {
      "epoch": 0.9013517761710154,
      "grad_norm": 0.43528470396995544,
      "learning_rate": 5.175056187410609e-07,
      "loss": 0.8246,
      "step": 5376
    },
    {
      "epoch": 0.9015194383317615,
      "grad_norm": 0.42045772075653076,
      "learning_rate": 5.157826634946395e-07,
      "loss": 0.8547,
      "step": 5377
    },
    {
      "epoch": 0.9016871004925076,
      "grad_norm": 0.40543699264526367,
      "learning_rate": 5.140625052613501e-07,
      "loss": 0.8254,
      "step": 5378
    },
    {
      "epoch": 0.9018547626532537,
      "grad_norm": 0.3896552622318268,
      "learning_rate": 5.12345144548494e-07,
      "loss": 0.8132,
      "step": 5379
    },
    {
      "epoch": 0.9020224248139997,
      "grad_norm": 0.47298580408096313,
      "learning_rate": 5.106305818625357e-07,
      "loss": 0.8961,
      "step": 5380
    },
    {
      "epoch": 0.9021900869747459,
      "grad_norm": 0.6200173497200012,
      "learning_rate": 5.089188177091254e-07,
      "loss": 0.921,
      "step": 5381
    },
    {
      "epoch": 0.9023577491354919,
      "grad_norm": 0.5027073621749878,
      "learning_rate": 5.072098525930825e-07,
      "loss": 0.774,
      "step": 5382
    },
    {
      "epoch": 0.9025254112962381,
      "grad_norm": 0.49361899495124817,
      "learning_rate": 5.055036870184005e-07,
      "loss": 0.8256,
      "step": 5383
    },
    {
      "epoch": 0.9026930734569841,
      "grad_norm": 0.390128493309021,
      "learning_rate": 5.038003214882503e-07,
      "loss": 0.8341,
      "step": 5384
    },
    {
      "epoch": 0.9028607356177303,
      "grad_norm": 0.550040602684021,
      "learning_rate": 5.020997565049756e-07,
      "loss": 0.7736,
      "step": 5385
    },
    {
      "epoch": 0.9030283977784763,
      "grad_norm": 0.3496764898300171,
      "learning_rate": 5.004019925700921e-07,
      "loss": 0.8466,
      "step": 5386
    },
    {
      "epoch": 0.9031960599392225,
      "grad_norm": 0.3841930031776428,
      "learning_rate": 4.987070301842934e-07,
      "loss": 0.8168,
      "step": 5387
    },
    {
      "epoch": 0.9033637220999685,
      "grad_norm": 0.4360392987728119,
      "learning_rate": 4.97014869847443e-07,
      "loss": 0.9347,
      "step": 5388
    },
    {
      "epoch": 0.9035313842607147,
      "grad_norm": 1.048956036567688,
      "learning_rate": 4.953255120585832e-07,
      "loss": 0.8688,
      "step": 5389
    },
    {
      "epoch": 0.9036990464214607,
      "grad_norm": 0.3388780951499939,
      "learning_rate": 4.936389573159239e-07,
      "loss": 0.8564,
      "step": 5390
    },
    {
      "epoch": 0.9038667085822069,
      "grad_norm": 0.5225685238838196,
      "learning_rate": 4.919552061168509e-07,
      "loss": 0.7863,
      "step": 5391
    },
    {
      "epoch": 0.9040343707429529,
      "grad_norm": 0.5430662631988525,
      "learning_rate": 4.902742589579246e-07,
      "loss": 0.8832,
      "step": 5392
    },
    {
      "epoch": 0.904202032903699,
      "grad_norm": 0.44941407442092896,
      "learning_rate": 4.885961163348762e-07,
      "loss": 0.776,
      "step": 5393
    },
    {
      "epoch": 0.9043696950644451,
      "grad_norm": 0.46763765811920166,
      "learning_rate": 4.869207787426111e-07,
      "loss": 0.938,
      "step": 5394
    },
    {
      "epoch": 0.9045373572251912,
      "grad_norm": 0.3220028579235077,
      "learning_rate": 4.852482466752074e-07,
      "loss": 0.7516,
      "step": 5395
    },
    {
      "epoch": 0.9047050193859373,
      "grad_norm": 1.0635429620742798,
      "learning_rate": 4.835785206259147e-07,
      "loss": 0.8318,
      "step": 5396
    },
    {
      "epoch": 0.9048726815466834,
      "grad_norm": 0.43439677357673645,
      "learning_rate": 4.81911601087155e-07,
      "loss": 0.8004,
      "step": 5397
    },
    {
      "epoch": 0.9050403437074296,
      "grad_norm": 0.4261639714241028,
      "learning_rate": 4.80247488550527e-07,
      "loss": 0.7668,
      "step": 5398
    },
    {
      "epoch": 0.9052080058681756,
      "grad_norm": 0.46629682183265686,
      "learning_rate": 4.785861835067962e-07,
      "loss": 0.8571,
      "step": 5399
    },
    {
      "epoch": 0.9053756680289218,
      "grad_norm": 0.3818049728870392,
      "learning_rate": 4.769276864459016e-07,
      "loss": 0.8136,
      "step": 5400
    },
    {
      "epoch": 0.9055433301896678,
      "grad_norm": 0.39135274291038513,
      "learning_rate": 4.752719978569531e-07,
      "loss": 0.7031,
      "step": 5401
    },
    {
      "epoch": 0.905710992350414,
      "grad_norm": 0.40635475516319275,
      "learning_rate": 4.7361911822823457e-07,
      "loss": 0.8831,
      "step": 5402
    },
    {
      "epoch": 0.90587865451116,
      "grad_norm": 0.5607383847236633,
      "learning_rate": 4.71969048047205e-07,
      "loss": 0.8562,
      "step": 5403
    },
    {
      "epoch": 0.9060463166719062,
      "grad_norm": 0.7370794415473938,
      "learning_rate": 4.703217878004851e-07,
      "loss": 0.8472,
      "step": 5404
    },
    {
      "epoch": 0.9062139788326522,
      "grad_norm": 0.5216053128242493,
      "learning_rate": 4.686773379738729e-07,
      "loss": 0.8061,
      "step": 5405
    },
    {
      "epoch": 0.9063816409933984,
      "grad_norm": 0.43987199664115906,
      "learning_rate": 4.670356990523406e-07,
      "loss": 0.803,
      "step": 5406
    },
    {
      "epoch": 0.9065493031541444,
      "grad_norm": 0.4011021554470062,
      "learning_rate": 4.65396871520023e-07,
      "loss": 0.8302,
      "step": 5407
    },
    {
      "epoch": 0.9067169653148905,
      "grad_norm": 0.40155351161956787,
      "learning_rate": 4.637608558602358e-07,
      "loss": 0.8402,
      "step": 5408
    },
    {
      "epoch": 0.9068846274756366,
      "grad_norm": 0.4337129592895508,
      "learning_rate": 4.6212765255545745e-07,
      "loss": 0.8696,
      "step": 5409
    },
    {
      "epoch": 0.9070522896363827,
      "grad_norm": 0.390887975692749,
      "learning_rate": 4.6049726208733825e-07,
      "loss": 0.7668,
      "step": 5410
    },
    {
      "epoch": 0.9072199517971288,
      "grad_norm": 0.3842809498310089,
      "learning_rate": 4.588696849367036e-07,
      "loss": 0.8776,
      "step": 5411
    },
    {
      "epoch": 0.9073876139578749,
      "grad_norm": 0.43689432740211487,
      "learning_rate": 4.5724492158354397e-07,
      "loss": 0.8334,
      "step": 5412
    },
    {
      "epoch": 0.907555276118621,
      "grad_norm": 0.4424873888492584,
      "learning_rate": 4.556229725070249e-07,
      "loss": 0.8537,
      "step": 5413
    },
    {
      "epoch": 0.9077229382793671,
      "grad_norm": 0.6077720522880554,
      "learning_rate": 4.540038381854772e-07,
      "loss": 0.8586,
      "step": 5414
    },
    {
      "epoch": 0.9078906004401132,
      "grad_norm": 0.4896509349346161,
      "learning_rate": 4.5238751909640224e-07,
      "loss": 0.8448,
      "step": 5415
    },
    {
      "epoch": 0.9080582626008593,
      "grad_norm": 0.440701425075531,
      "learning_rate": 4.5077401571647527e-07,
      "loss": 0.8318,
      "step": 5416
    },
    {
      "epoch": 0.9082259247616054,
      "grad_norm": 0.5645205974578857,
      "learning_rate": 4.4916332852153797e-07,
      "loss": 0.948,
      "step": 5417
    },
    {
      "epoch": 0.9083935869223515,
      "grad_norm": 0.4756021499633789,
      "learning_rate": 4.4755545798659926e-07,
      "loss": 0.8636,
      "step": 5418
    },
    {
      "epoch": 0.9085612490830975,
      "grad_norm": 0.5430362820625305,
      "learning_rate": 4.459504045858443e-07,
      "loss": 0.8831,
      "step": 5419
    },
    {
      "epoch": 0.9087289112438437,
      "grad_norm": 0.3706998825073242,
      "learning_rate": 4.4434816879261987e-07,
      "loss": 0.7866,
      "step": 5420
    },
    {
      "epoch": 0.9088965734045897,
      "grad_norm": 0.3644903302192688,
      "learning_rate": 4.427487510794459e-07,
      "loss": 0.8777,
      "step": 5421
    },
    {
      "epoch": 0.9090642355653359,
      "grad_norm": 0.38833266496658325,
      "learning_rate": 4.4115215191801285e-07,
      "loss": 0.8398,
      "step": 5422
    },
    {
      "epoch": 0.9092318977260819,
      "grad_norm": 0.46523725986480713,
      "learning_rate": 4.3955837177917403e-07,
      "loss": 0.8199,
      "step": 5423
    },
    {
      "epoch": 0.9093995598868281,
      "grad_norm": 0.7484700083732605,
      "learning_rate": 4.379674111329557e-07,
      "loss": 0.81,
      "step": 5424
    },
    {
      "epoch": 0.9095672220475741,
      "grad_norm": 0.4971903860569,
      "learning_rate": 4.3637927044855476e-07,
      "loss": 0.901,
      "step": 5425
    },
    {
      "epoch": 0.9097348842083203,
      "grad_norm": 0.4184390604496002,
      "learning_rate": 4.3479395019433104e-07,
      "loss": 0.8638,
      "step": 5426
    },
    {
      "epoch": 0.9099025463690663,
      "grad_norm": 0.6759586334228516,
      "learning_rate": 4.332114508378138e-07,
      "loss": 0.768,
      "step": 5427
    },
    {
      "epoch": 0.9100702085298125,
      "grad_norm": 0.4398367404937744,
      "learning_rate": 4.3163177284570536e-07,
      "loss": 0.8337,
      "step": 5428
    },
    {
      "epoch": 0.9102378706905585,
      "grad_norm": 0.6298412084579468,
      "learning_rate": 4.300549166838697e-07,
      "loss": 0.8829,
      "step": 5429
    },
    {
      "epoch": 0.9104055328513047,
      "grad_norm": 0.5195366144180298,
      "learning_rate": 4.284808828173426e-07,
      "loss": 0.8517,
      "step": 5430
    },
    {
      "epoch": 0.9105731950120507,
      "grad_norm": 0.44466158747673035,
      "learning_rate": 4.2690967171032604e-07,
      "loss": 0.8467,
      "step": 5431
    },
    {
      "epoch": 0.9107408571727968,
      "grad_norm": 0.4637994170188904,
      "learning_rate": 4.253412838261872e-07,
      "loss": 0.8496,
      "step": 5432
    },
    {
      "epoch": 0.9109085193335429,
      "grad_norm": 0.42862823605537415,
      "learning_rate": 4.237757196274661e-07,
      "loss": 0.7244,
      "step": 5433
    },
    {
      "epoch": 0.911076181494289,
      "grad_norm": 0.479573518037796,
      "learning_rate": 4.222129795758645e-07,
      "loss": 0.9259,
      "step": 5434
    },
    {
      "epoch": 0.9112438436550351,
      "grad_norm": 0.47090259194374084,
      "learning_rate": 4.2065306413225704e-07,
      "loss": 0.9028,
      "step": 5435
    },
    {
      "epoch": 0.9114115058157812,
      "grad_norm": 0.4738175570964813,
      "learning_rate": 4.190959737566802e-07,
      "loss": 0.8315,
      "step": 5436
    },
    {
      "epoch": 0.9115791679765273,
      "grad_norm": 0.6840828657150269,
      "learning_rate": 4.1754170890833777e-07,
      "loss": 0.8972,
      "step": 5437
    },
    {
      "epoch": 0.9117468301372734,
      "grad_norm": 0.4690623879432678,
      "learning_rate": 4.1599027004560535e-07,
      "loss": 0.8862,
      "step": 5438
    },
    {
      "epoch": 0.9119144922980195,
      "grad_norm": 0.4256584346294403,
      "learning_rate": 4.144416576260168e-07,
      "loss": 0.8003,
      "step": 5439
    },
    {
      "epoch": 0.9120821544587656,
      "grad_norm": 0.4423993229866028,
      "learning_rate": 4.1289587210628146e-07,
      "loss": 0.8348,
      "step": 5440
    },
    {
      "epoch": 0.9122498166195117,
      "grad_norm": 1.1761881113052368,
      "learning_rate": 4.1135291394226785e-07,
      "loss": 0.8472,
      "step": 5441
    },
    {
      "epoch": 0.9124174787802578,
      "grad_norm": 0.41688793897628784,
      "learning_rate": 4.0981278358901424e-07,
      "loss": 0.891,
      "step": 5442
    },
    {
      "epoch": 0.9125851409410038,
      "grad_norm": 0.43712949752807617,
      "learning_rate": 4.082754815007239e-07,
      "loss": 0.8237,
      "step": 5443
    },
    {
      "epoch": 0.91275280310175,
      "grad_norm": 0.4862794578075409,
      "learning_rate": 4.0674100813076765e-07,
      "loss": 0.8626,
      "step": 5444
    },
    {
      "epoch": 0.912920465262496,
      "grad_norm": 0.42175430059432983,
      "learning_rate": 4.052093639316801e-07,
      "loss": 0.7836,
      "step": 5445
    },
    {
      "epoch": 0.9130881274232422,
      "grad_norm": 0.7146934866905212,
      "learning_rate": 4.0368054935516097e-07,
      "loss": 0.8276,
      "step": 5446
    },
    {
      "epoch": 0.9132557895839882,
      "grad_norm": 0.5035149455070496,
      "learning_rate": 4.0215456485207636e-07,
      "loss": 0.8425,
      "step": 5447
    },
    {
      "epoch": 0.9134234517447344,
      "grad_norm": 0.48977479338645935,
      "learning_rate": 4.006314108724585e-07,
      "loss": 0.8014,
      "step": 5448
    },
    {
      "epoch": 0.9135911139054804,
      "grad_norm": 0.3573587238788605,
      "learning_rate": 3.99111087865508e-07,
      "loss": 0.8272,
      "step": 5449
    },
    {
      "epoch": 0.9137587760662266,
      "grad_norm": 0.3598082363605499,
      "learning_rate": 3.975935962795807e-07,
      "loss": 0.8045,
      "step": 5450
    },
    {
      "epoch": 0.9139264382269726,
      "grad_norm": 0.4267180860042572,
      "learning_rate": 3.960789365622075e-07,
      "loss": 0.7992,
      "step": 5451
    },
    {
      "epoch": 0.9140941003877188,
      "grad_norm": 0.5257976651191711,
      "learning_rate": 3.9456710916008003e-07,
      "loss": 0.7921,
      "step": 5452
    },
    {
      "epoch": 0.9142617625484648,
      "grad_norm": 0.3591015934944153,
      "learning_rate": 3.930581145190526e-07,
      "loss": 0.8088,
      "step": 5453
    },
    {
      "epoch": 0.914429424709211,
      "grad_norm": 0.4040519893169403,
      "learning_rate": 3.915519530841505e-07,
      "loss": 0.9061,
      "step": 5454
    },
    {
      "epoch": 0.914597086869957,
      "grad_norm": 0.41480520367622375,
      "learning_rate": 3.9004862529955613e-07,
      "loss": 0.7608,
      "step": 5455
    },
    {
      "epoch": 0.9147647490307032,
      "grad_norm": 0.4387241303920746,
      "learning_rate": 3.8854813160861924e-07,
      "loss": 0.8797,
      "step": 5456
    },
    {
      "epoch": 0.9149324111914492,
      "grad_norm": 0.48141932487487793,
      "learning_rate": 3.8705047245385707e-07,
      "loss": 0.8078,
      "step": 5457
    },
    {
      "epoch": 0.9151000733521953,
      "grad_norm": 0.518697202205658,
      "learning_rate": 3.8555564827694627e-07,
      "loss": 0.7451,
      "step": 5458
    },
    {
      "epoch": 0.9152677355129414,
      "grad_norm": 0.4883868396282196,
      "learning_rate": 3.8406365951872636e-07,
      "loss": 0.8408,
      "step": 5459
    },
    {
      "epoch": 0.9154353976736875,
      "grad_norm": 0.5068694949150085,
      "learning_rate": 3.825745066192066e-07,
      "loss": 0.8275,
      "step": 5460
    },
    {
      "epoch": 0.9156030598344336,
      "grad_norm": 0.4290688633918762,
      "learning_rate": 3.810881900175556e-07,
      "loss": 0.8499,
      "step": 5461
    },
    {
      "epoch": 0.9157707219951797,
      "grad_norm": 0.5387886762619019,
      "learning_rate": 3.796047101521072e-07,
      "loss": 0.8061,
      "step": 5462
    },
    {
      "epoch": 0.9159383841559258,
      "grad_norm": 0.5438194870948792,
      "learning_rate": 3.78124067460357e-07,
      "loss": 0.8132,
      "step": 5463
    },
    {
      "epoch": 0.9161060463166719,
      "grad_norm": 0.33834096789360046,
      "learning_rate": 3.766462623789646e-07,
      "loss": 0.7614,
      "step": 5464
    },
    {
      "epoch": 0.916273708477418,
      "grad_norm": 0.47196847200393677,
      "learning_rate": 3.7517129534375475e-07,
      "loss": 0.9023,
      "step": 5465
    },
    {
      "epoch": 0.9164413706381641,
      "grad_norm": 0.4197845160961151,
      "learning_rate": 3.7369916678971053e-07,
      "loss": 0.8842,
      "step": 5466
    },
    {
      "epoch": 0.9166090327989102,
      "grad_norm": 0.43537554144859314,
      "learning_rate": 3.722298771509847e-07,
      "loss": 0.8246,
      "step": 5467
    },
    {
      "epoch": 0.9167766949596563,
      "grad_norm": 0.4930468201637268,
      "learning_rate": 3.707634268608862e-07,
      "loss": 0.9056,
      "step": 5468
    },
    {
      "epoch": 0.9169443571204023,
      "grad_norm": 0.4002600908279419,
      "learning_rate": 3.692998163518902e-07,
      "loss": 0.8087,
      "step": 5469
    },
    {
      "epoch": 0.9171120192811485,
      "grad_norm": 1.5046316385269165,
      "learning_rate": 3.6783904605563246e-07,
      "loss": 0.8554,
      "step": 5470
    },
    {
      "epoch": 0.9172796814418945,
      "grad_norm": 0.4091032147407532,
      "learning_rate": 3.663811164029152e-07,
      "loss": 0.8805,
      "step": 5471
    },
    {
      "epoch": 0.9174473436026407,
      "grad_norm": 0.35924676060676575,
      "learning_rate": 3.6492602782369657e-07,
      "loss": 0.8437,
      "step": 5472
    },
    {
      "epoch": 0.9176150057633867,
      "grad_norm": 0.5918196439743042,
      "learning_rate": 3.634737807471023e-07,
      "loss": 0.9303,
      "step": 5473
    },
    {
      "epoch": 0.9177826679241329,
      "grad_norm": 0.3623254895210266,
      "learning_rate": 3.620243756014175e-07,
      "loss": 0.8151,
      "step": 5474
    },
    {
      "epoch": 0.9179503300848789,
      "grad_norm": 0.5924551486968994,
      "learning_rate": 3.605778128140891e-07,
      "loss": 0.8822,
      "step": 5475
    },
    {
      "epoch": 0.9181179922456251,
      "grad_norm": 0.3630390465259552,
      "learning_rate": 3.5913409281172706e-07,
      "loss": 0.8507,
      "step": 5476
    },
    {
      "epoch": 0.9182856544063711,
      "grad_norm": 1.2100481986999512,
      "learning_rate": 3.57693216020103e-07,
      "loss": 0.7789,
      "step": 5477
    },
    {
      "epoch": 0.9184533165671173,
      "grad_norm": 0.4164574444293976,
      "learning_rate": 3.5625518286414696e-07,
      "loss": 0.83,
      "step": 5478
    },
    {
      "epoch": 0.9186209787278633,
      "grad_norm": 0.4077569842338562,
      "learning_rate": 3.548199937679553e-07,
      "loss": 0.9101,
      "step": 5479
    },
    {
      "epoch": 0.9187886408886095,
      "grad_norm": 0.6429985761642456,
      "learning_rate": 3.533876491547805e-07,
      "loss": 0.868,
      "step": 5480
    },
    {
      "epoch": 0.9189563030493555,
      "grad_norm": 0.4084544777870178,
      "learning_rate": 3.5195814944704143e-07,
      "loss": 0.9758,
      "step": 5481
    },
    {
      "epoch": 0.9191239652101016,
      "grad_norm": 0.43578848242759705,
      "learning_rate": 3.5053149506631413e-07,
      "loss": 0.8348,
      "step": 5482
    },
    {
      "epoch": 0.9192916273708478,
      "grad_norm": 0.4982907176017761,
      "learning_rate": 3.491076864333354e-07,
      "loss": 0.844,
      "step": 5483
    },
    {
      "epoch": 0.9194592895315938,
      "grad_norm": 0.4257267415523529,
      "learning_rate": 3.476867239680071e-07,
      "loss": 0.7825,
      "step": 5484
    },
    {
      "epoch": 0.91962695169234,
      "grad_norm": 0.4639023244380951,
      "learning_rate": 3.46268608089384e-07,
      "loss": 0.8848,
      "step": 5485
    },
    {
      "epoch": 0.919794613853086,
      "grad_norm": 0.5799000263214111,
      "learning_rate": 3.4485333921569166e-07,
      "loss": 0.8532,
      "step": 5486
    },
    {
      "epoch": 0.9199622760138322,
      "grad_norm": 0.532958447933197,
      "learning_rate": 3.434409177643061e-07,
      "loss": 0.8224,
      "step": 5487
    },
    {
      "epoch": 0.9201299381745782,
      "grad_norm": 0.31039291620254517,
      "learning_rate": 3.420313441517686e-07,
      "loss": 0.8132,
      "step": 5488
    },
    {
      "epoch": 0.9202976003353244,
      "grad_norm": 0.41978487372398376,
      "learning_rate": 3.4062461879378005e-07,
      "loss": 0.8231,
      "step": 5489
    },
    {
      "epoch": 0.9204652624960704,
      "grad_norm": 0.3548107147216797,
      "learning_rate": 3.3922074210520407e-07,
      "loss": 0.9061,
      "step": 5490
    },
    {
      "epoch": 0.9206329246568166,
      "grad_norm": 0.4682113826274872,
      "learning_rate": 3.3781971450005724e-07,
      "loss": 0.8242,
      "step": 5491
    },
    {
      "epoch": 0.9208005868175626,
      "grad_norm": 0.46550726890563965,
      "learning_rate": 3.364215363915213e-07,
      "loss": 0.9678,
      "step": 5492
    },
    {
      "epoch": 0.9209682489783088,
      "grad_norm": 0.46769365668296814,
      "learning_rate": 3.350262081919375e-07,
      "loss": 0.8332,
      "step": 5493
    },
    {
      "epoch": 0.9211359111390548,
      "grad_norm": 0.4007333219051361,
      "learning_rate": 3.336337303128034e-07,
      "loss": 0.8389,
      "step": 5494
    },
    {
      "epoch": 0.921303573299801,
      "grad_norm": 0.4930897057056427,
      "learning_rate": 3.322441031647816e-07,
      "loss": 0.7938,
      "step": 5495
    },
    {
      "epoch": 0.921471235460547,
      "grad_norm": 0.4777367115020752,
      "learning_rate": 3.308573271576854e-07,
      "loss": 0.7612,
      "step": 5496
    },
    {
      "epoch": 0.9216388976212931,
      "grad_norm": 1.0620161294937134,
      "learning_rate": 3.2947340270049424e-07,
      "loss": 0.8541,
      "step": 5497
    },
    {
      "epoch": 0.9218065597820392,
      "grad_norm": 0.462362140417099,
      "learning_rate": 3.2809233020134614e-07,
      "loss": 0.8281,
      "step": 5498
    },
    {
      "epoch": 0.9219742219427853,
      "grad_norm": 0.651817262172699,
      "learning_rate": 3.267141100675353e-07,
      "loss": 0.8666,
      "step": 5499
    },
    {
      "epoch": 0.9221418841035314,
      "grad_norm": 0.3979651927947998,
      "learning_rate": 3.2533874270551546e-07,
      "loss": 0.8032,
      "step": 5500
    },
    {
      "epoch": 0.9223095462642775,
      "grad_norm": 0.4143291413784027,
      "learning_rate": 3.239662285209e-07,
      "loss": 0.8331,
      "step": 5501
    },
    {
      "epoch": 0.9224772084250236,
      "grad_norm": 0.3905138373374939,
      "learning_rate": 3.225965679184595e-07,
      "loss": 0.8526,
      "step": 5502
    },
    {
      "epoch": 0.9226448705857697,
      "grad_norm": 0.4227253198623657,
      "learning_rate": 3.2122976130212644e-07,
      "loss": 0.7858,
      "step": 5503
    },
    {
      "epoch": 0.9228125327465158,
      "grad_norm": 0.37597405910491943,
      "learning_rate": 3.198658090749862e-07,
      "loss": 0.8548,
      "step": 5504
    },
    {
      "epoch": 0.9229801949072619,
      "grad_norm": 0.611361026763916,
      "learning_rate": 3.185047116392859e-07,
      "loss": 0.8773,
      "step": 5505
    },
    {
      "epoch": 0.923147857068008,
      "grad_norm": 0.38209253549575806,
      "learning_rate": 3.17146469396431e-07,
      "loss": 0.7404,
      "step": 5506
    },
    {
      "epoch": 0.9233155192287541,
      "grad_norm": 0.8735145330429077,
      "learning_rate": 3.157910827469823e-07,
      "loss": 0.8744,
      "step": 5507
    },
    {
      "epoch": 0.9234831813895001,
      "grad_norm": 0.35140228271484375,
      "learning_rate": 3.144385520906623e-07,
      "loss": 0.8837,
      "step": 5508
    },
    {
      "epoch": 0.9236508435502463,
      "grad_norm": 0.38805145025253296,
      "learning_rate": 3.1308887782634746e-07,
      "loss": 0.7759,
      "step": 5509
    },
    {
      "epoch": 0.9238185057109923,
      "grad_norm": 0.4310365617275238,
      "learning_rate": 3.1174206035207265e-07,
      "loss": 0.7953,
      "step": 5510
    },
    {
      "epoch": 0.9239861678717385,
      "grad_norm": 0.367244154214859,
      "learning_rate": 3.1039810006503357e-07,
      "loss": 0.811,
      "step": 5511
    },
    {
      "epoch": 0.9241538300324845,
      "grad_norm": 1.0172940492630005,
      "learning_rate": 3.0905699736157756e-07,
      "loss": 0.9186,
      "step": 5512
    },
    {
      "epoch": 0.9243214921932307,
      "grad_norm": 0.47557225823402405,
      "learning_rate": 3.077187526372161e-07,
      "loss": 0.8462,
      "step": 5513
    },
    {
      "epoch": 0.9244891543539767,
      "grad_norm": 0.42626944184303284,
      "learning_rate": 3.063833662866111e-07,
      "loss": 0.8948,
      "step": 5514
    },
    {
      "epoch": 0.9246568165147229,
      "grad_norm": 0.6726216077804565,
      "learning_rate": 3.0505083870358555e-07,
      "loss": 0.8881,
      "step": 5515
    },
    {
      "epoch": 0.9248244786754689,
      "grad_norm": 1.152274250984192,
      "learning_rate": 3.0372117028111825e-07,
      "loss": 0.7966,
      "step": 5516
    },
    {
      "epoch": 0.9249921408362151,
      "grad_norm": 0.4575171172618866,
      "learning_rate": 3.023943614113456e-07,
      "loss": 0.7819,
      "step": 5517
    },
    {
      "epoch": 0.9251598029969611,
      "grad_norm": 0.4207251965999603,
      "learning_rate": 3.0107041248556015e-07,
      "loss": 0.7604,
      "step": 5518
    },
    {
      "epoch": 0.9253274651577073,
      "grad_norm": 0.4027195870876312,
      "learning_rate": 2.997493238942084e-07,
      "loss": 0.8249,
      "step": 5519
    },
    {
      "epoch": 0.9254951273184533,
      "grad_norm": 0.32407307624816895,
      "learning_rate": 2.984310960268988e-07,
      "loss": 0.7537,
      "step": 5520
    },
    {
      "epoch": 0.9256627894791994,
      "grad_norm": 0.5370124578475952,
      "learning_rate": 2.971157292723903e-07,
      "loss": 0.8758,
      "step": 5521
    },
    {
      "epoch": 0.9258304516399455,
      "grad_norm": 0.4623180329799652,
      "learning_rate": 2.9580322401860374e-07,
      "loss": 0.8286,
      "step": 5522
    },
    {
      "epoch": 0.9259981138006916,
      "grad_norm": 0.6158974766731262,
      "learning_rate": 2.944935806526117e-07,
      "loss": 0.732,
      "step": 5523
    },
    {
      "epoch": 0.9261657759614377,
      "grad_norm": 0.4130534529685974,
      "learning_rate": 2.93186799560643e-07,
      "loss": 0.8576,
      "step": 5524
    },
    {
      "epoch": 0.9263334381221838,
      "grad_norm": 0.979594349861145,
      "learning_rate": 2.9188288112808606e-07,
      "loss": 0.8856,
      "step": 5525
    },
    {
      "epoch": 0.9265011002829299,
      "grad_norm": 0.4869593679904938,
      "learning_rate": 2.905818257394799e-07,
      "loss": 0.8958,
      "step": 5526
    },
    {
      "epoch": 0.926668762443676,
      "grad_norm": 0.43282878398895264,
      "learning_rate": 2.892836337785243e-07,
      "loss": 0.816,
      "step": 5527
    },
    {
      "epoch": 0.9268364246044221,
      "grad_norm": 0.5063544511795044,
      "learning_rate": 2.8798830562807077e-07,
      "loss": 0.8357,
      "step": 5528
    },
    {
      "epoch": 0.9270040867651682,
      "grad_norm": 0.3956616520881653,
      "learning_rate": 2.866958416701271e-07,
      "loss": 0.8457,
      "step": 5529
    },
    {
      "epoch": 0.9271717489259143,
      "grad_norm": 0.46155110001564026,
      "learning_rate": 2.8540624228585947e-07,
      "loss": 0.9228,
      "step": 5530
    },
    {
      "epoch": 0.9273394110866604,
      "grad_norm": 0.3728257417678833,
      "learning_rate": 2.8411950785558474e-07,
      "loss": 0.8098,
      "step": 5531
    },
    {
      "epoch": 0.9275070732474064,
      "grad_norm": 0.4718465805053711,
      "learning_rate": 2.8283563875877494e-07,
      "loss": 0.7529,
      "step": 5532
    },
    {
      "epoch": 0.9276747354081526,
      "grad_norm": 0.4334772825241089,
      "learning_rate": 2.815546353740639e-07,
      "loss": 0.83,
      "step": 5533
    },
    {
      "epoch": 0.9278423975688986,
      "grad_norm": 0.7013729810714722,
      "learning_rate": 2.8027649807923053e-07,
      "loss": 0.8759,
      "step": 5534
    },
    {
      "epoch": 0.9280100597296448,
      "grad_norm": 0.4334397614002228,
      "learning_rate": 2.7900122725121547e-07,
      "loss": 0.8663,
      "step": 5535
    },
    {
      "epoch": 0.9281777218903908,
      "grad_norm": 0.4532167613506317,
      "learning_rate": 2.777288232661135e-07,
      "loss": 0.7878,
      "step": 5536
    },
    {
      "epoch": 0.928345384051137,
      "grad_norm": 0.37148404121398926,
      "learning_rate": 2.764592864991689e-07,
      "loss": 0.8114,
      "step": 5537
    },
    {
      "epoch": 0.928513046211883,
      "grad_norm": 0.43077895045280457,
      "learning_rate": 2.751926173247865e-07,
      "loss": 0.7621,
      "step": 5538
    },
    {
      "epoch": 0.9286807083726292,
      "grad_norm": 0.5970816016197205,
      "learning_rate": 2.73928816116521e-07,
      "loss": 0.8411,
      "step": 5539
    },
    {
      "epoch": 0.9288483705333752,
      "grad_norm": 0.4724122881889343,
      "learning_rate": 2.7266788324708524e-07,
      "loss": 0.8413,
      "step": 5540
    },
    {
      "epoch": 0.9290160326941214,
      "grad_norm": 0.4347897171974182,
      "learning_rate": 2.714098190883407e-07,
      "loss": 0.9437,
      "step": 5541
    },
    {
      "epoch": 0.9291836948548674,
      "grad_norm": 0.42247769236564636,
      "learning_rate": 2.7015462401130843e-07,
      "loss": 0.7411,
      "step": 5542
    },
    {
      "epoch": 0.9293513570156136,
      "grad_norm": 0.5921533107757568,
      "learning_rate": 2.6890229838616e-07,
      "loss": 0.8147,
      "step": 5543
    },
    {
      "epoch": 0.9295190191763596,
      "grad_norm": 0.28696200251579285,
      "learning_rate": 2.6765284258222337e-07,
      "loss": 0.8451,
      "step": 5544
    },
    {
      "epoch": 0.9296866813371057,
      "grad_norm": 0.8700627684593201,
      "learning_rate": 2.6640625696797706e-07,
      "loss": 0.87,
      "step": 5545
    },
    {
      "epoch": 0.9298543434978518,
      "grad_norm": 0.308586448431015,
      "learning_rate": 2.6516254191105374e-07,
      "loss": 0.7779,
      "step": 5546
    },
    {
      "epoch": 0.9300220056585979,
      "grad_norm": 0.4023432433605194,
      "learning_rate": 2.639216977782433e-07,
      "loss": 0.8421,
      "step": 5547
    },
    {
      "epoch": 0.930189667819344,
      "grad_norm": 0.502148449420929,
      "learning_rate": 2.626837249354819e-07,
      "loss": 0.8936,
      "step": 5548
    },
    {
      "epoch": 0.9303573299800901,
      "grad_norm": 0.9387050271034241,
      "learning_rate": 2.614486237478675e-07,
      "loss": 0.8268,
      "step": 5549
    },
    {
      "epoch": 0.9305249921408362,
      "grad_norm": 0.6043473482131958,
      "learning_rate": 2.602163945796432e-07,
      "loss": 0.8885,
      "step": 5550
    },
    {
      "epoch": 0.9306926543015823,
      "grad_norm": 0.4056151509284973,
      "learning_rate": 2.589870377942105e-07,
      "loss": 0.7908,
      "step": 5551
    },
    {
      "epoch": 0.9308603164623284,
      "grad_norm": 0.41021332144737244,
      "learning_rate": 2.577605537541217e-07,
      "loss": 0.8381,
      "step": 5552
    },
    {
      "epoch": 0.9310279786230745,
      "grad_norm": 0.447323739528656,
      "learning_rate": 2.565369428210818e-07,
      "loss": 0.8611,
      "step": 5553
    },
    {
      "epoch": 0.9311956407838206,
      "grad_norm": 0.4121238887310028,
      "learning_rate": 2.553162053559488e-07,
      "loss": 0.7491,
      "step": 5554
    },
    {
      "epoch": 0.9313633029445667,
      "grad_norm": 0.35634592175483704,
      "learning_rate": 2.540983417187348e-07,
      "loss": 0.8236,
      "step": 5555
    },
    {
      "epoch": 0.9315309651053127,
      "grad_norm": 0.3375703990459442,
      "learning_rate": 2.528833522685992e-07,
      "loss": 0.8379,
      "step": 5556
    },
    {
      "epoch": 0.9316986272660589,
      "grad_norm": 0.524369478225708,
      "learning_rate": 2.516712373638608e-07,
      "loss": 0.8166,
      "step": 5557
    },
    {
      "epoch": 0.9318662894268049,
      "grad_norm": 0.43301674723625183,
      "learning_rate": 2.5046199736198596e-07,
      "loss": 0.8631,
      "step": 5558
    },
    {
      "epoch": 0.9320339515875511,
      "grad_norm": 0.47489795088768005,
      "learning_rate": 2.4925563261959606e-07,
      "loss": 0.7435,
      "step": 5559
    },
    {
      "epoch": 0.9322016137482971,
      "grad_norm": 0.41612544655799866,
      "learning_rate": 2.48052143492461e-07,
      "loss": 0.8816,
      "step": 5560
    },
    {
      "epoch": 0.9323692759090433,
      "grad_norm": 0.43906569480895996,
      "learning_rate": 2.4685153033550457e-07,
      "loss": 0.7912,
      "step": 5561
    },
    {
      "epoch": 0.9325369380697893,
      "grad_norm": 0.4497454762458801,
      "learning_rate": 2.456537935028036e-07,
      "loss": 0.8735,
      "step": 5562
    },
    {
      "epoch": 0.9327046002305355,
      "grad_norm": 0.3805798590183258,
      "learning_rate": 2.4445893334758664e-07,
      "loss": 0.7605,
      "step": 5563
    },
    {
      "epoch": 0.9328722623912815,
      "grad_norm": 0.4280644357204437,
      "learning_rate": 2.4326695022222956e-07,
      "loss": 0.7823,
      "step": 5564
    },
    {
      "epoch": 0.9330399245520277,
      "grad_norm": 0.47225022315979004,
      "learning_rate": 2.420778444782645e-07,
      "loss": 0.9112,
      "step": 5565
    },
    {
      "epoch": 0.9332075867127737,
      "grad_norm": 0.3943757116794586,
      "learning_rate": 2.408916164663755e-07,
      "loss": 0.8492,
      "step": 5566
    },
    {
      "epoch": 0.9333752488735199,
      "grad_norm": 0.3952292799949646,
      "learning_rate": 2.397082665363915e-07,
      "loss": 0.8742,
      "step": 5567
    },
    {
      "epoch": 0.9335429110342659,
      "grad_norm": 0.3373555541038513,
      "learning_rate": 2.3852779503730217e-07,
      "loss": 0.7162,
      "step": 5568
    },
    {
      "epoch": 0.933710573195012,
      "grad_norm": 0.46765002608299255,
      "learning_rate": 2.3735020231723915e-07,
      "loss": 0.8617,
      "step": 5569
    },
    {
      "epoch": 0.9338782353557582,
      "grad_norm": 0.44101208448410034,
      "learning_rate": 2.3617548872349015e-07,
      "loss": 0.8869,
      "step": 5570
    },
    {
      "epoch": 0.9340458975165042,
      "grad_norm": 0.3950546383857727,
      "learning_rate": 2.3500365460249474e-07,
      "loss": 0.7568,
      "step": 5571
    },
    {
      "epoch": 0.9342135596772504,
      "grad_norm": 0.38100630044937134,
      "learning_rate": 2.3383470029983868e-07,
      "loss": 0.8651,
      "step": 5572
    },
    {
      "epoch": 0.9343812218379964,
      "grad_norm": 0.33927208185195923,
      "learning_rate": 2.3266862616026175e-07,
      "loss": 0.7564,
      "step": 5573
    },
    {
      "epoch": 0.9345488839987426,
      "grad_norm": 0.31735849380493164,
      "learning_rate": 2.3150543252765556e-07,
      "loss": 0.7747,
      "step": 5574
    },
    {
      "epoch": 0.9347165461594886,
      "grad_norm": 0.5408655405044556,
      "learning_rate": 2.303451197450568e-07,
      "loss": 0.767,
      "step": 5575
    },
    {
      "epoch": 0.9348842083202348,
      "grad_norm": 0.45583081245422363,
      "learning_rate": 2.2918768815465951e-07,
      "loss": 0.7727,
      "step": 5576
    },
    {
      "epoch": 0.9350518704809808,
      "grad_norm": 0.8801291584968567,
      "learning_rate": 2.280331380978029e-07,
      "loss": 0.7964,
      "step": 5577
    },
    {
      "epoch": 0.935219532641727,
      "grad_norm": 0.2814652621746063,
      "learning_rate": 2.2688146991497683e-07,
      "loss": 0.8358,
      "step": 5578
    },
    {
      "epoch": 0.935387194802473,
      "grad_norm": 0.532590925693512,
      "learning_rate": 2.2573268394582625e-07,
      "loss": 0.8855,
      "step": 5579
    },
    {
      "epoch": 0.9355548569632192,
      "grad_norm": 0.34196722507476807,
      "learning_rate": 2.2458678052913796e-07,
      "loss": 0.8477,
      "step": 5580
    },
    {
      "epoch": 0.9357225191239652,
      "grad_norm": 0.39009302854537964,
      "learning_rate": 2.2344376000285606e-07,
      "loss": 0.8902,
      "step": 5581
    },
    {
      "epoch": 0.9358901812847114,
      "grad_norm": 0.4790273904800415,
      "learning_rate": 2.2230362270407203e-07,
      "loss": 0.8542,
      "step": 5582
    },
    {
      "epoch": 0.9360578434454574,
      "grad_norm": 0.45219048857688904,
      "learning_rate": 2.2116636896902244e-07,
      "loss": 0.8752,
      "step": 5583
    },
    {
      "epoch": 0.9362255056062035,
      "grad_norm": 0.35386043787002563,
      "learning_rate": 2.2003199913310124e-07,
      "loss": 0.9029,
      "step": 5584
    },
    {
      "epoch": 0.9363931677669496,
      "grad_norm": 0.43155017495155334,
      "learning_rate": 2.189005135308475e-07,
      "loss": 0.8639,
      "step": 5585
    },
    {
      "epoch": 0.9365608299276957,
      "grad_norm": 0.40214449167251587,
      "learning_rate": 2.1777191249594986e-07,
      "loss": 0.8996,
      "step": 5586
    },
    {
      "epoch": 0.9367284920884418,
      "grad_norm": 0.33067846298217773,
      "learning_rate": 2.1664619636124652e-07,
      "loss": 0.781,
      "step": 5587
    },
    {
      "epoch": 0.9368961542491879,
      "grad_norm": 0.6097775101661682,
      "learning_rate": 2.1552336545872522e-07,
      "loss": 0.8176,
      "step": 5588
    },
    {
      "epoch": 0.937063816409934,
      "grad_norm": 0.36495113372802734,
      "learning_rate": 2.144034201195222e-07,
      "loss": 0.9063,
      "step": 5589
    },
    {
      "epoch": 0.9372314785706801,
      "grad_norm": 0.3748246133327484,
      "learning_rate": 2.132863606739255e-07,
      "loss": 0.8615,
      "step": 5590
    },
    {
      "epoch": 0.9373991407314262,
      "grad_norm": 0.3747648596763611,
      "learning_rate": 2.1217218745136715e-07,
      "loss": 0.8227,
      "step": 5591
    },
    {
      "epoch": 0.9375668028921723,
      "grad_norm": 0.4228968322277069,
      "learning_rate": 2.1106090078043095e-07,
      "loss": 0.7919,
      "step": 5592
    },
    {
      "epoch": 0.9377344650529184,
      "grad_norm": 0.43143996596336365,
      "learning_rate": 2.0995250098885034e-07,
      "loss": 0.7377,
      "step": 5593
    },
    {
      "epoch": 0.9379021272136645,
      "grad_norm": 0.611870527267456,
      "learning_rate": 2.0884698840350492e-07,
      "loss": 0.8197,
      "step": 5594
    },
    {
      "epoch": 0.9380697893744105,
      "grad_norm": 0.44937756657600403,
      "learning_rate": 2.0774436335042614e-07,
      "loss": 0.9146,
      "step": 5595
    },
    {
      "epoch": 0.9382374515351567,
      "grad_norm": 0.40494683384895325,
      "learning_rate": 2.0664462615478942e-07,
      "loss": 0.8464,
      "step": 5596
    },
    {
      "epoch": 0.9384051136959027,
      "grad_norm": 0.3957933485507965,
      "learning_rate": 2.055477771409209e-07,
      "loss": 0.7419,
      "step": 5597
    },
    {
      "epoch": 0.9385727758566489,
      "grad_norm": 0.6733267903327942,
      "learning_rate": 2.0445381663229734e-07,
      "loss": 0.9049,
      "step": 5598
    },
    {
      "epoch": 0.9387404380173949,
      "grad_norm": 0.42263564467430115,
      "learning_rate": 2.033627449515374e-07,
      "loss": 0.8662,
      "step": 5599
    },
    {
      "epoch": 0.9389081001781411,
      "grad_norm": 0.5451824069023132,
      "learning_rate": 2.0227456242041589e-07,
      "loss": 0.8229,
      "step": 5600
    },
    {
      "epoch": 0.9390757623388871,
      "grad_norm": 0.36930420994758606,
      "learning_rate": 2.0118926935984939e-07,
      "loss": 0.8783,
      "step": 5601
    },
    {
      "epoch": 0.9392434244996333,
      "grad_norm": 0.36576372385025024,
      "learning_rate": 2.0010686608990304e-07,
      "loss": 0.8387,
      "step": 5602
    },
    {
      "epoch": 0.9394110866603793,
      "grad_norm": 0.32187724113464355,
      "learning_rate": 1.9902735292979257e-07,
      "loss": 0.8154,
      "step": 5603
    },
    {
      "epoch": 0.9395787488211255,
      "grad_norm": 0.4188230037689209,
      "learning_rate": 1.9795073019788002e-07,
      "loss": 0.8459,
      "step": 5604
    },
    {
      "epoch": 0.9397464109818715,
      "grad_norm": 0.33690860867500305,
      "learning_rate": 1.968769982116736e-07,
      "loss": 0.8161,
      "step": 5605
    },
    {
      "epoch": 0.9399140731426177,
      "grad_norm": 0.3758140504360199,
      "learning_rate": 1.9580615728783004e-07,
      "loss": 0.7501,
      "step": 5606
    },
    {
      "epoch": 0.9400817353033637,
      "grad_norm": 0.9181646108627319,
      "learning_rate": 1.9473820774215557e-07,
      "loss": 0.8525,
      "step": 5607
    },
    {
      "epoch": 0.9402493974641098,
      "grad_norm": 0.44023045897483826,
      "learning_rate": 1.9367314988959829e-07,
      "loss": 0.7978,
      "step": 5608
    },
    {
      "epoch": 0.9404170596248559,
      "grad_norm": 0.5621471405029297,
      "learning_rate": 1.9261098404426136e-07,
      "loss": 0.8633,
      "step": 5609
    },
    {
      "epoch": 0.940584721785602,
      "grad_norm": 0.7647391557693481,
      "learning_rate": 1.9155171051938758e-07,
      "loss": 0.7623,
      "step": 5610
    },
    {
      "epoch": 0.9407523839463481,
      "grad_norm": 0.34152573347091675,
      "learning_rate": 1.9049532962736926e-07,
      "loss": 0.7409,
      "step": 5611
    },
    {
      "epoch": 0.9409200461070942,
      "grad_norm": 0.4390283226966858,
      "learning_rate": 1.8944184167974832e-07,
      "loss": 0.8357,
      "step": 5612
    },
    {
      "epoch": 0.9410877082678403,
      "grad_norm": 0.3948175609111786,
      "learning_rate": 1.883912469872118e-07,
      "loss": 0.8497,
      "step": 5613
    },
    {
      "epoch": 0.9412553704285864,
      "grad_norm": 0.4613439440727234,
      "learning_rate": 1.8734354585958958e-07,
      "loss": 0.8871,
      "step": 5614
    },
    {
      "epoch": 0.9414230325893325,
      "grad_norm": 0.4463028609752655,
      "learning_rate": 1.8629873860586567e-07,
      "loss": 0.8103,
      "step": 5615
    },
    {
      "epoch": 0.9415906947500786,
      "grad_norm": 0.3706364929676056,
      "learning_rate": 1.8525682553416468e-07,
      "loss": 0.8776,
      "step": 5616
    },
    {
      "epoch": 0.9417583569108247,
      "grad_norm": 0.5275152325630188,
      "learning_rate": 1.8421780695175973e-07,
      "loss": 0.8077,
      "step": 5617
    },
    {
      "epoch": 0.9419260190715708,
      "grad_norm": 0.5429855585098267,
      "learning_rate": 1.8318168316507123e-07,
      "loss": 0.8583,
      "step": 5618
    },
    {
      "epoch": 0.9420936812323168,
      "grad_norm": 0.3907036781311035,
      "learning_rate": 1.8214845447966256e-07,
      "loss": 0.8475,
      "step": 5619
    },
    {
      "epoch": 0.942261343393063,
      "grad_norm": 0.3771839141845703,
      "learning_rate": 1.8111812120024884e-07,
      "loss": 0.8338,
      "step": 5620
    },
    {
      "epoch": 0.942429005553809,
      "grad_norm": 0.38360580801963806,
      "learning_rate": 1.8009068363068595e-07,
      "loss": 0.8439,
      "step": 5621
    },
    {
      "epoch": 0.9425966677145552,
      "grad_norm": 0.6593756675720215,
      "learning_rate": 1.790661420739781e-07,
      "loss": 0.8713,
      "step": 5622
    },
    {
      "epoch": 0.9427643298753012,
      "grad_norm": 0.35056811571121216,
      "learning_rate": 1.7804449683227697e-07,
      "loss": 0.8065,
      "step": 5623
    },
    {
      "epoch": 0.9429319920360474,
      "grad_norm": 0.5193197131156921,
      "learning_rate": 1.7702574820687602e-07,
      "loss": 0.9337,
      "step": 5624
    },
    {
      "epoch": 0.9430996541967934,
      "grad_norm": 0.4338991045951843,
      "learning_rate": 1.7600989649821932e-07,
      "loss": 0.7891,
      "step": 5625
    },
    {
      "epoch": 0.9432673163575396,
      "grad_norm": 0.4451599717140198,
      "learning_rate": 1.7499694200589057e-07,
      "loss": 0.9064,
      "step": 5626
    },
    {
      "epoch": 0.9434349785182856,
      "grad_norm": 0.3576316833496094,
      "learning_rate": 1.7398688502862748e-07,
      "loss": 0.8033,
      "step": 5627
    },
    {
      "epoch": 0.9436026406790318,
      "grad_norm": 0.3730756938457489,
      "learning_rate": 1.7297972586430401e-07,
      "loss": 0.882,
      "step": 5628
    },
    {
      "epoch": 0.9437703028397778,
      "grad_norm": 0.4072226583957672,
      "learning_rate": 1.7197546480994587e-07,
      "loss": 0.8271,
      "step": 5629
    },
    {
      "epoch": 0.943937965000524,
      "grad_norm": 0.3262406289577484,
      "learning_rate": 1.7097410216172062e-07,
      "loss": 0.7549,
      "step": 5630
    },
    {
      "epoch": 0.94410562716127,
      "grad_norm": 0.47185268998146057,
      "learning_rate": 1.6997563821494533e-07,
      "loss": 0.9325,
      "step": 5631
    },
    {
      "epoch": 0.9442732893220162,
      "grad_norm": 0.4424075186252594,
      "learning_rate": 1.6898007326407784e-07,
      "loss": 0.7956,
      "step": 5632
    },
    {
      "epoch": 0.9444409514827622,
      "grad_norm": 0.5058478713035583,
      "learning_rate": 1.6798740760272104e-07,
      "loss": 0.9044,
      "step": 5633
    },
    {
      "epoch": 0.9446086136435083,
      "grad_norm": 0.6497793197631836,
      "learning_rate": 1.6699764152362628e-07,
      "loss": 0.7954,
      "step": 5634
    },
    {
      "epoch": 0.9447762758042544,
      "grad_norm": 0.38459107279777527,
      "learning_rate": 1.6601077531868794e-07,
      "loss": 0.8354,
      "step": 5635
    },
    {
      "epoch": 0.9449439379650005,
      "grad_norm": 0.40192103385925293,
      "learning_rate": 1.650268092789442e-07,
      "loss": 0.8224,
      "step": 5636
    },
    {
      "epoch": 0.9451116001257466,
      "grad_norm": 0.40187108516693115,
      "learning_rate": 1.640457436945786e-07,
      "loss": 0.9093,
      "step": 5637
    },
    {
      "epoch": 0.9452792622864927,
      "grad_norm": 0.42827141284942627,
      "learning_rate": 1.6306757885491964e-07,
      "loss": 0.817,
      "step": 5638
    },
    {
      "epoch": 0.9454469244472388,
      "grad_norm": 0.37759557366371155,
      "learning_rate": 1.6209231504844102e-07,
      "loss": 0.7827,
      "step": 5639
    },
    {
      "epoch": 0.9456145866079849,
      "grad_norm": 0.3767712712287903,
      "learning_rate": 1.6111995256275936e-07,
      "loss": 0.8499,
      "step": 5640
    },
    {
      "epoch": 0.945782248768731,
      "grad_norm": 0.3564743995666504,
      "learning_rate": 1.6015049168463748e-07,
      "loss": 0.758,
      "step": 5641
    },
    {
      "epoch": 0.9459499109294771,
      "grad_norm": 0.6888882517814636,
      "learning_rate": 1.5918393269998112e-07,
      "loss": 0.8285,
      "step": 5642
    },
    {
      "epoch": 0.9461175730902232,
      "grad_norm": 0.46881723403930664,
      "learning_rate": 1.582202758938378e-07,
      "loss": 0.9183,
      "step": 5643
    },
    {
      "epoch": 0.9462852352509693,
      "grad_norm": 0.8559983372688293,
      "learning_rate": 1.572595215504058e-07,
      "loss": 0.7672,
      "step": 5644
    },
    {
      "epoch": 0.9464528974117153,
      "grad_norm": 0.9411035180091858,
      "learning_rate": 1.5630166995302177e-07,
      "loss": 0.8157,
      "step": 5645
    },
    {
      "epoch": 0.9466205595724615,
      "grad_norm": 0.4257885217666626,
      "learning_rate": 1.553467213841664e-07,
      "loss": 0.8358,
      "step": 5646
    },
    {
      "epoch": 0.9467882217332075,
      "grad_norm": 0.697925865650177,
      "learning_rate": 1.543946761254689e-07,
      "loss": 0.8611,
      "step": 5647
    },
    {
      "epoch": 0.9469558838939537,
      "grad_norm": 0.5628939867019653,
      "learning_rate": 1.5344553445769683e-07,
      "loss": 0.8576,
      "step": 5648
    },
    {
      "epoch": 0.9471235460546997,
      "grad_norm": 0.6172940731048584,
      "learning_rate": 1.5249929666076525e-07,
      "loss": 0.8426,
      "step": 5649
    },
    {
      "epoch": 0.9472912082154459,
      "grad_norm": 0.4763745963573456,
      "learning_rate": 1.51555963013732e-07,
      "loss": 0.8808,
      "step": 5650
    },
    {
      "epoch": 0.9474588703761919,
      "grad_norm": 0.3848874866962433,
      "learning_rate": 1.5061553379479455e-07,
      "loss": 0.8111,
      "step": 5651
    },
    {
      "epoch": 0.9476265325369381,
      "grad_norm": 0.39039984345436096,
      "learning_rate": 1.4967800928129994e-07,
      "loss": 0.8439,
      "step": 5652
    },
    {
      "epoch": 0.9477941946976841,
      "grad_norm": 0.3452758193016052,
      "learning_rate": 1.4874338974973478e-07,
      "loss": 0.8758,
      "step": 5653
    },
    {
      "epoch": 0.9479618568584303,
      "grad_norm": 0.4906795024871826,
      "learning_rate": 1.4781167547573194e-07,
      "loss": 0.8095,
      "step": 5654
    },
    {
      "epoch": 0.9481295190191764,
      "grad_norm": 0.37396731972694397,
      "learning_rate": 1.468828667340616e-07,
      "loss": 0.7964,
      "step": 5655
    },
    {
      "epoch": 0.9482971811799225,
      "grad_norm": 1.0198975801467896,
      "learning_rate": 1.4595696379864467e-07,
      "loss": 0.9225,
      "step": 5656
    },
    {
      "epoch": 0.9484648433406686,
      "grad_norm": 0.47543761134147644,
      "learning_rate": 1.4503396694253936e-07,
      "loss": 0.8874,
      "step": 5657
    },
    {
      "epoch": 0.9486325055014146,
      "grad_norm": 0.3990827202796936,
      "learning_rate": 1.4411387643795017e-07,
      "loss": 0.815,
      "step": 5658
    },
    {
      "epoch": 0.9488001676621608,
      "grad_norm": 0.3973783850669861,
      "learning_rate": 1.4319669255622115e-07,
      "loss": 0.7715,
      "step": 5659
    },
    {
      "epoch": 0.9489678298229068,
      "grad_norm": 0.35283103585243225,
      "learning_rate": 1.422824155678426e-07,
      "loss": 0.8014,
      "step": 5660
    },
    {
      "epoch": 0.949135491983653,
      "grad_norm": 0.5298373103141785,
      "learning_rate": 1.4137104574244665e-07,
      "loss": 0.9061,
      "step": 5661
    },
    {
      "epoch": 0.949303154144399,
      "grad_norm": 0.3985663056373596,
      "learning_rate": 1.4046258334880714e-07,
      "loss": 0.8046,
      "step": 5662
    },
    {
      "epoch": 0.9494708163051452,
      "grad_norm": 0.5122137069702148,
      "learning_rate": 1.3955702865483977e-07,
      "loss": 0.9043,
      "step": 5663
    },
    {
      "epoch": 0.9496384784658912,
      "grad_norm": 0.5534654855728149,
      "learning_rate": 1.386543819276054e-07,
      "loss": 0.8845,
      "step": 5664
    },
    {
      "epoch": 0.9498061406266374,
      "grad_norm": 0.38909614086151123,
      "learning_rate": 1.3775464343330437e-07,
      "loss": 0.8299,
      "step": 5665
    },
    {
      "epoch": 0.9499738027873834,
      "grad_norm": 0.5959657430648804,
      "learning_rate": 1.3685781343728222e-07,
      "loss": 0.8008,
      "step": 5666
    },
    {
      "epoch": 0.9501414649481296,
      "grad_norm": 0.4258873760700226,
      "learning_rate": 1.3596389220402296e-07,
      "loss": 0.794,
      "step": 5667
    },
    {
      "epoch": 0.9503091271088756,
      "grad_norm": 0.46945369243621826,
      "learning_rate": 1.350728799971579e-07,
      "loss": 0.7653,
      "step": 5668
    },
    {
      "epoch": 0.9504767892696218,
      "grad_norm": 0.3949854373931885,
      "learning_rate": 1.3418477707945576e-07,
      "loss": 0.8105,
      "step": 5669
    },
    {
      "epoch": 0.9506444514303678,
      "grad_norm": 0.4090568721294403,
      "learning_rate": 1.3329958371282815e-07,
      "loss": 0.9001,
      "step": 5670
    },
    {
      "epoch": 0.950812113591114,
      "grad_norm": 0.4272141456604004,
      "learning_rate": 1.3241730015833176e-07,
      "loss": 0.8375,
      "step": 5671
    },
    {
      "epoch": 0.95097977575186,
      "grad_norm": 0.46921345591545105,
      "learning_rate": 1.3153792667616183e-07,
      "loss": 0.8256,
      "step": 5672
    },
    {
      "epoch": 0.9511474379126061,
      "grad_norm": 0.4632411599159241,
      "learning_rate": 1.3066146352565755e-07,
      "loss": 0.8708,
      "step": 5673
    },
    {
      "epoch": 0.9513151000733522,
      "grad_norm": 0.8034499883651733,
      "learning_rate": 1.2978791096529663e-07,
      "loss": 0.8364,
      "step": 5674
    },
    {
      "epoch": 0.9514827622340983,
      "grad_norm": 0.4491528570652008,
      "learning_rate": 1.2891726925270186e-07,
      "loss": 0.8402,
      "step": 5675
    },
    {
      "epoch": 0.9516504243948444,
      "grad_norm": 0.4233703911304474,
      "learning_rate": 1.2804953864463676e-07,
      "loss": 0.7798,
      "step": 5676
    },
    {
      "epoch": 0.9518180865555905,
      "grad_norm": 0.37281307578086853,
      "learning_rate": 1.271847193970055e-07,
      "loss": 0.8267,
      "step": 5677
    },
    {
      "epoch": 0.9519857487163366,
      "grad_norm": 0.43550345301628113,
      "learning_rate": 1.2632281176485295e-07,
      "loss": 0.813,
      "step": 5678
    },
    {
      "epoch": 0.9521534108770827,
      "grad_norm": 0.6797903180122375,
      "learning_rate": 1.254638160023669e-07,
      "loss": 0.8983,
      "step": 5679
    },
    {
      "epoch": 0.9523210730378288,
      "grad_norm": 0.43068262934684753,
      "learning_rate": 1.2460773236287694e-07,
      "loss": 0.8743,
      "step": 5680
    },
    {
      "epoch": 0.9524887351985749,
      "grad_norm": 0.36471620202064514,
      "learning_rate": 1.2375456109885108e-07,
      "loss": 0.7999,
      "step": 5681
    },
    {
      "epoch": 0.952656397359321,
      "grad_norm": 0.3796907067298889,
      "learning_rate": 1.2290430246190145e-07,
      "loss": 0.8054,
      "step": 5682
    },
    {
      "epoch": 0.9528240595200671,
      "grad_norm": 0.6740773320198059,
      "learning_rate": 1.2205695670277963e-07,
      "loss": 0.8228,
      "step": 5683
    },
    {
      "epoch": 0.9529917216808131,
      "grad_norm": 0.4136049747467041,
      "learning_rate": 1.2121252407137685e-07,
      "loss": 0.7726,
      "step": 5684
    },
    {
      "epoch": 0.9531593838415593,
      "grad_norm": 0.45437702536582947,
      "learning_rate": 1.2037100481672836e-07,
      "loss": 0.8748,
      "step": 5685
    },
    {
      "epoch": 0.9533270460023053,
      "grad_norm": 0.4190893769264221,
      "learning_rate": 1.1953239918700897e-07,
      "loss": 0.8073,
      "step": 5686
    },
    {
      "epoch": 0.9534947081630515,
      "grad_norm": 1.0747568607330322,
      "learning_rate": 1.1869670742953199e-07,
      "loss": 0.865,
      "step": 5687
    },
    {
      "epoch": 0.9536623703237975,
      "grad_norm": 0.5825490951538086,
      "learning_rate": 1.1786392979075467e-07,
      "loss": 0.8079,
      "step": 5688
    },
    {
      "epoch": 0.9538300324845437,
      "grad_norm": 0.44110995531082153,
      "learning_rate": 1.170340665162728e-07,
      "loss": 0.8474,
      "step": 5689
    },
    {
      "epoch": 0.9539976946452897,
      "grad_norm": 0.34802770614624023,
      "learning_rate": 1.1620711785082395e-07,
      "loss": 0.761,
      "step": 5690
    },
    {
      "epoch": 0.9541653568060359,
      "grad_norm": 0.4393647313117981,
      "learning_rate": 1.1538308403828524e-07,
      "loss": 0.801,
      "step": 5691
    },
    {
      "epoch": 0.9543330189667819,
      "grad_norm": 0.3807205259799957,
      "learning_rate": 1.1456196532167452e-07,
      "loss": 0.8628,
      "step": 5692
    },
    {
      "epoch": 0.9545006811275281,
      "grad_norm": 0.4649239778518677,
      "learning_rate": 1.1374376194315028e-07,
      "loss": 0.834,
      "step": 5693
    },
    {
      "epoch": 0.9546683432882741,
      "grad_norm": 0.47582659125328064,
      "learning_rate": 1.1292847414400843e-07,
      "loss": 0.9082,
      "step": 5694
    },
    {
      "epoch": 0.9548360054490203,
      "grad_norm": 0.4253336489200592,
      "learning_rate": 1.1211610216469105e-07,
      "loss": 0.7836,
      "step": 5695
    },
    {
      "epoch": 0.9550036676097663,
      "grad_norm": 0.4425368010997772,
      "learning_rate": 1.1130664624477539e-07,
      "loss": 0.8245,
      "step": 5696
    },
    {
      "epoch": 0.9551713297705124,
      "grad_norm": 0.4651656746864319,
      "learning_rate": 1.1050010662297827e-07,
      "loss": 0.7881,
      "step": 5697
    },
    {
      "epoch": 0.9553389919312585,
      "grad_norm": 0.6690202355384827,
      "learning_rate": 1.0969648353715945e-07,
      "loss": 0.846,
      "step": 5698
    },
    {
      "epoch": 0.9555066540920046,
      "grad_norm": 0.4050973653793335,
      "learning_rate": 1.0889577722431821e-07,
      "loss": 0.8111,
      "step": 5699
    },
    {
      "epoch": 0.9556743162527507,
      "grad_norm": 0.4939645528793335,
      "learning_rate": 1.0809798792059234e-07,
      "loss": 0.7909,
      "step": 5700
    },
    {
      "epoch": 0.9558419784134968,
      "grad_norm": 0.36605989933013916,
      "learning_rate": 1.0730311586125808e-07,
      "loss": 0.8195,
      "step": 5701
    },
    {
      "epoch": 0.9560096405742429,
      "grad_norm": 0.4095785319805145,
      "learning_rate": 1.0651116128073568e-07,
      "loss": 0.8232,
      "step": 5702
    },
    {
      "epoch": 0.956177302734989,
      "grad_norm": 0.40485769510269165,
      "learning_rate": 1.0572212441257945e-07,
      "loss": 0.757,
      "step": 5703
    },
    {
      "epoch": 0.9563449648957351,
      "grad_norm": 0.5039658546447754,
      "learning_rate": 1.0493600548948879e-07,
      "loss": 0.7698,
      "step": 5704
    },
    {
      "epoch": 0.9565126270564812,
      "grad_norm": 0.42550402879714966,
      "learning_rate": 1.0415280474329936e-07,
      "loss": 0.825,
      "step": 5705
    },
    {
      "epoch": 0.9566802892172273,
      "grad_norm": 0.425540566444397,
      "learning_rate": 1.033725224049853e-07,
      "loss": 0.868,
      "step": 5706
    },
    {
      "epoch": 0.9568479513779734,
      "grad_norm": 0.5083727836608887,
      "learning_rate": 1.0259515870466363e-07,
      "loss": 0.7654,
      "step": 5707
    },
    {
      "epoch": 0.9570156135387194,
      "grad_norm": 0.3627629578113556,
      "learning_rate": 1.0182071387158543e-07,
      "loss": 0.88,
      "step": 5708
    },
    {
      "epoch": 0.9571832756994656,
      "grad_norm": 0.7076917886734009,
      "learning_rate": 1.01049188134148e-07,
      "loss": 0.831,
      "step": 5709
    },
    {
      "epoch": 0.9573509378602116,
      "grad_norm": 0.502890408039093,
      "learning_rate": 1.0028058171988153e-07,
      "loss": 0.8811,
      "step": 5710
    },
    {
      "epoch": 0.9575186000209578,
      "grad_norm": 0.7879444360733032,
      "learning_rate": 9.951489485545696e-08,
      "loss": 0.836,
      "step": 5711
    },
    {
      "epoch": 0.9576862621817038,
      "grad_norm": 0.5187103152275085,
      "learning_rate": 9.875212776668586e-08,
      "loss": 0.8376,
      "step": 5712
    },
    {
      "epoch": 0.95785392434245,
      "grad_norm": 0.4570627808570862,
      "learning_rate": 9.799228067851829e-08,
      "loss": 0.8397,
      "step": 5713
    },
    {
      "epoch": 0.958021586503196,
      "grad_norm": 0.41080382466316223,
      "learning_rate": 9.723535381504167e-08,
      "loss": 0.8595,
      "step": 5714
    },
    {
      "epoch": 0.9581892486639422,
      "grad_norm": 0.7080820202827454,
      "learning_rate": 9.6481347399483e-08,
      "loss": 0.8357,
      "step": 5715
    },
    {
      "epoch": 0.9583569108246882,
      "grad_norm": 0.41043514013290405,
      "learning_rate": 9.573026165420773e-08,
      "loss": 0.8083,
      "step": 5716
    },
    {
      "epoch": 0.9585245729854344,
      "grad_norm": 0.48574718832969666,
      "learning_rate": 9.498209680072202e-08,
      "loss": 0.8493,
      "step": 5717
    },
    {
      "epoch": 0.9586922351461804,
      "grad_norm": 0.8123177886009216,
      "learning_rate": 9.423685305966824e-08,
      "loss": 0.8967,
      "step": 5718
    },
    {
      "epoch": 0.9588598973069266,
      "grad_norm": 0.4755695164203644,
      "learning_rate": 9.349453065082725e-08,
      "loss": 0.7297,
      "step": 5719
    },
    {
      "epoch": 0.9590275594676726,
      "grad_norm": 0.3649205267429352,
      "learning_rate": 9.275512979311952e-08,
      "loss": 0.8195,
      "step": 5720
    },
    {
      "epoch": 0.9591952216284187,
      "grad_norm": 0.7999193668365479,
      "learning_rate": 9.201865070460392e-08,
      "loss": 0.8144,
      "step": 5721
    },
    {
      "epoch": 0.9593628837891648,
      "grad_norm": 0.3566318452358246,
      "learning_rate": 9.128509360247562e-08,
      "loss": 0.7668,
      "step": 5722
    },
    {
      "epoch": 0.9595305459499109,
      "grad_norm": 0.5108819007873535,
      "learning_rate": 9.055445870307267e-08,
      "loss": 0.7538,
      "step": 5723
    },
    {
      "epoch": 0.959698208110657,
      "grad_norm": 0.4764617681503296,
      "learning_rate": 8.982674622186605e-08,
      "loss": 0.843,
      "step": 5724
    },
    {
      "epoch": 0.9598658702714031,
      "grad_norm": 0.35026487708091736,
      "learning_rate": 8.910195637346741e-08,
      "loss": 0.8462,
      "step": 5725
    },
    {
      "epoch": 0.9600335324321492,
      "grad_norm": 0.3958798050880432,
      "learning_rate": 8.83800893716269e-08,
      "loss": 0.8281,
      "step": 5726
    },
    {
      "epoch": 0.9602011945928953,
      "grad_norm": 0.6569202542304993,
      "learning_rate": 8.76611454292331e-08,
      "loss": 0.7438,
      "step": 5727
    },
    {
      "epoch": 0.9603688567536414,
      "grad_norm": 1.2969248294830322,
      "learning_rate": 8.694512475830863e-08,
      "loss": 0.76,
      "step": 5728
    },
    {
      "epoch": 0.9605365189143875,
      "grad_norm": 0.44533786177635193,
      "learning_rate": 8.623202757001792e-08,
      "loss": 0.8143,
      "step": 5729
    },
    {
      "epoch": 0.9607041810751336,
      "grad_norm": 0.5025610327720642,
      "learning_rate": 8.552185407466384e-08,
      "loss": 0.8676,
      "step": 5730
    },
    {
      "epoch": 0.9608718432358797,
      "grad_norm": 0.46003156900405884,
      "learning_rate": 8.481460448168333e-08,
      "loss": 0.8146,
      "step": 5731
    },
    {
      "epoch": 0.9610395053966257,
      "grad_norm": 0.3955024480819702,
      "learning_rate": 8.411027899965507e-08,
      "loss": 0.8441,
      "step": 5732
    },
    {
      "epoch": 0.9612071675573719,
      "grad_norm": 0.4125998020172119,
      "learning_rate": 8.34088778362907e-08,
      "loss": 0.7981,
      "step": 5733
    },
    {
      "epoch": 0.9613748297181179,
      "grad_norm": 0.509367823600769,
      "learning_rate": 8.271040119844476e-08,
      "loss": 0.8146,
      "step": 5734
    },
    {
      "epoch": 0.9615424918788641,
      "grad_norm": 0.5742055773735046,
      "learning_rate": 8.201484929210357e-08,
      "loss": 0.8677,
      "step": 5735
    },
    {
      "epoch": 0.9617101540396101,
      "grad_norm": 0.5999758243560791,
      "learning_rate": 8.13222223223975e-08,
      "loss": 0.8455,
      "step": 5736
    },
    {
      "epoch": 0.9618778162003563,
      "grad_norm": 0.3258301615715027,
      "learning_rate": 8.063252049358983e-08,
      "loss": 0.8064,
      "step": 5737
    },
    {
      "epoch": 0.9620454783611023,
      "grad_norm": 0.5386455059051514,
      "learning_rate": 7.994574400908007e-08,
      "loss": 0.8215,
      "step": 5738
    },
    {
      "epoch": 0.9622131405218485,
      "grad_norm": 0.452926903963089,
      "learning_rate": 7.926189307140952e-08,
      "loss": 0.8158,
      "step": 5739
    },
    {
      "epoch": 0.9623808026825946,
      "grad_norm": 0.45257359743118286,
      "learning_rate": 7.858096788225245e-08,
      "loss": 0.9341,
      "step": 5740
    },
    {
      "epoch": 0.9625484648433407,
      "grad_norm": 0.7201059460639954,
      "learning_rate": 7.790296864242374e-08,
      "loss": 0.8401,
      "step": 5741
    },
    {
      "epoch": 0.9627161270040868,
      "grad_norm": 0.4358721077442169,
      "learning_rate": 7.722789555187238e-08,
      "loss": 0.7329,
      "step": 5742
    },
    {
      "epoch": 0.9628837891648329,
      "grad_norm": 0.42482608556747437,
      "learning_rate": 7.655574880968685e-08,
      "loss": 0.8294,
      "step": 5743
    },
    {
      "epoch": 0.963051451325579,
      "grad_norm": 0.38124197721481323,
      "learning_rate": 7.588652861409084e-08,
      "loss": 0.7149,
      "step": 5744
    },
    {
      "epoch": 0.963219113486325,
      "grad_norm": 0.4425771236419678,
      "learning_rate": 7.522023516244648e-08,
      "loss": 0.759,
      "step": 5745
    },
    {
      "epoch": 0.9633867756470712,
      "grad_norm": 0.567328929901123,
      "learning_rate": 7.455686865125212e-08,
      "loss": 0.8754,
      "step": 5746
    },
    {
      "epoch": 0.9635544378078172,
      "grad_norm": 0.35585835576057434,
      "learning_rate": 7.389642927614127e-08,
      "loss": 0.8482,
      "step": 5747
    },
    {
      "epoch": 0.9637220999685634,
      "grad_norm": 0.3198508024215698,
      "learning_rate": 7.323891723188592e-08,
      "loss": 0.829,
      "step": 5748
    },
    {
      "epoch": 0.9638897621293094,
      "grad_norm": 0.4167691171169281,
      "learning_rate": 7.258433271239651e-08,
      "loss": 0.8377,
      "step": 5749
    },
    {
      "epoch": 0.9640574242900556,
      "grad_norm": 0.5370296835899353,
      "learning_rate": 7.193267591071529e-08,
      "loss": 0.8478,
      "step": 5750
    },
    {
      "epoch": 0.9642250864508016,
      "grad_norm": 0.8849428296089172,
      "learning_rate": 7.12839470190263e-08,
      "loss": 0.952,
      "step": 5751
    },
    {
      "epoch": 0.9643927486115478,
      "grad_norm": 0.40987876057624817,
      "learning_rate": 7.063814622864651e-08,
      "loss": 0.8097,
      "step": 5752
    },
    {
      "epoch": 0.9645604107722938,
      "grad_norm": 0.4275193512439728,
      "learning_rate": 6.999527373003134e-08,
      "loss": 0.7957,
      "step": 5753
    },
    {
      "epoch": 0.96472807293304,
      "grad_norm": 0.49265700578689575,
      "learning_rate": 6.935532971277026e-08,
      "loss": 0.8611,
      "step": 5754
    },
    {
      "epoch": 0.964895735093786,
      "grad_norm": 0.4670751094818115,
      "learning_rate": 6.871831436559339e-08,
      "loss": 0.8502,
      "step": 5755
    },
    {
      "epoch": 0.9650633972545322,
      "grad_norm": 0.419053852558136,
      "learning_rate": 6.80842278763616e-08,
      "loss": 0.8133,
      "step": 5756
    },
    {
      "epoch": 0.9652310594152782,
      "grad_norm": 0.3705775737762451,
      "learning_rate": 6.745307043207638e-08,
      "loss": 0.7635,
      "step": 5757
    },
    {
      "epoch": 0.9653987215760244,
      "grad_norm": 0.40138015151023865,
      "learning_rate": 6.682484221887331e-08,
      "loss": 0.8773,
      "step": 5758
    },
    {
      "epoch": 0.9655663837367704,
      "grad_norm": 0.4074966013431549,
      "learning_rate": 6.619954342202528e-08,
      "loss": 0.8126,
      "step": 5759
    },
    {
      "epoch": 0.9657340458975165,
      "grad_norm": 0.5536928772926331,
      "learning_rate": 6.557717422594035e-08,
      "loss": 0.8109,
      "step": 5760
    },
    {
      "epoch": 0.9659017080582626,
      "grad_norm": 0.4149467647075653,
      "learning_rate": 6.495773481416278e-08,
      "loss": 0.7873,
      "step": 5761
    },
    {
      "epoch": 0.9660693702190087,
      "grad_norm": 0.4304526448249817,
      "learning_rate": 6.434122536937204e-08,
      "loss": 0.8521,
      "step": 5762
    },
    {
      "epoch": 0.9662370323797548,
      "grad_norm": 0.6651942729949951,
      "learning_rate": 6.3727646073386e-08,
      "loss": 0.8533,
      "step": 5763
    },
    {
      "epoch": 0.9664046945405009,
      "grad_norm": 0.415158212184906,
      "learning_rate": 6.311699710715657e-08,
      "loss": 0.8647,
      "step": 5764
    },
    {
      "epoch": 0.966572356701247,
      "grad_norm": 0.38120999932289124,
      "learning_rate": 6.250927865077083e-08,
      "loss": 0.7783,
      "step": 5765
    },
    {
      "epoch": 0.9667400188619931,
      "grad_norm": 1.4348599910736084,
      "learning_rate": 6.190449088345318e-08,
      "loss": 0.836,
      "step": 5766
    },
    {
      "epoch": 0.9669076810227392,
      "grad_norm": 0.3521517217159271,
      "learning_rate": 6.130263398356318e-08,
      "loss": 0.8043,
      "step": 5767
    },
    {
      "epoch": 0.9670753431834853,
      "grad_norm": 0.3433744013309479,
      "learning_rate": 6.070370812859438e-08,
      "loss": 0.8868,
      "step": 5768
    },
    {
      "epoch": 0.9672430053442314,
      "grad_norm": 0.46505022048950195,
      "learning_rate": 6.010771349517885e-08,
      "loss": 1.0081,
      "step": 5769
    },
    {
      "epoch": 0.9674106675049775,
      "grad_norm": 0.29260390996932983,
      "learning_rate": 5.951465025908265e-08,
      "loss": 0.7565,
      "step": 5770
    },
    {
      "epoch": 0.9675783296657235,
      "grad_norm": 0.5390307903289795,
      "learning_rate": 5.892451859520698e-08,
      "loss": 0.8892,
      "step": 5771
    },
    {
      "epoch": 0.9677459918264697,
      "grad_norm": 0.5621128678321838,
      "learning_rate": 5.83373186775904e-08,
      "loss": 0.8231,
      "step": 5772
    },
    {
      "epoch": 0.9679136539872157,
      "grad_norm": 0.41262245178222656,
      "learning_rate": 5.775305067940329e-08,
      "loss": 0.8835,
      "step": 5773
    },
    {
      "epoch": 0.9680813161479619,
      "grad_norm": 0.5214173197746277,
      "learning_rate": 5.717171477295447e-08,
      "loss": 0.9254,
      "step": 5774
    },
    {
      "epoch": 0.9682489783087079,
      "grad_norm": 0.5759170651435852,
      "learning_rate": 5.659331112968791e-08,
      "loss": 0.9011,
      "step": 5775
    },
    {
      "epoch": 0.9684166404694541,
      "grad_norm": 0.5135636925697327,
      "learning_rate": 5.6017839920180506e-08,
      "loss": 0.8083,
      "step": 5776
    },
    {
      "epoch": 0.9685843026302001,
      "grad_norm": 0.3557618260383606,
      "learning_rate": 5.544530131414649e-08,
      "loss": 0.9134,
      "step": 5777
    },
    {
      "epoch": 0.9687519647909463,
      "grad_norm": 0.5884382724761963,
      "learning_rate": 5.4875695480435255e-08,
      "loss": 0.8587,
      "step": 5778
    },
    {
      "epoch": 0.9689196269516923,
      "grad_norm": 0.4821484088897705,
      "learning_rate": 5.4309022587029084e-08,
      "loss": 0.8129,
      "step": 5779
    },
    {
      "epoch": 0.9690872891124385,
      "grad_norm": 0.6065845489501953,
      "learning_rate": 5.374528280104874e-08,
      "loss": 0.8444,
      "step": 5780
    },
    {
      "epoch": 0.9692549512731845,
      "grad_norm": 0.4409104287624359,
      "learning_rate": 5.318447628874679e-08,
      "loss": 0.903,
      "step": 5781
    },
    {
      "epoch": 0.9694226134339307,
      "grad_norm": 0.3960440456867218,
      "learning_rate": 5.262660321551316e-08,
      "loss": 0.7283,
      "step": 5782
    },
    {
      "epoch": 0.9695902755946767,
      "grad_norm": 0.40227869153022766,
      "learning_rate": 5.2071663745869585e-08,
      "loss": 0.7678,
      "step": 5783
    },
    {
      "epoch": 0.9697579377554228,
      "grad_norm": 0.4411965012550354,
      "learning_rate": 5.151965804347625e-08,
      "loss": 0.8097,
      "step": 5784
    },
    {
      "epoch": 0.9699255999161689,
      "grad_norm": 0.5686452984809875,
      "learning_rate": 5.0970586271126274e-08,
      "loss": 0.8054,
      "step": 5785
    },
    {
      "epoch": 0.970093262076915,
      "grad_norm": 0.39256587624549866,
      "learning_rate": 5.0424448590746796e-08,
      "loss": 0.8827,
      "step": 5786
    },
    {
      "epoch": 0.9702609242376611,
      "grad_norm": 0.3850398659706116,
      "learning_rate": 4.98812451634012e-08,
      "loss": 0.8567,
      "step": 5787
    },
    {
      "epoch": 0.9704285863984072,
      "grad_norm": 0.35744708776474,
      "learning_rate": 4.93409761492869e-08,
      "loss": 0.8392,
      "step": 5788
    },
    {
      "epoch": 0.9705962485591533,
      "grad_norm": 0.4973679780960083,
      "learning_rate": 4.880364170773533e-08,
      "loss": 0.8695,
      "step": 5789
    },
    {
      "epoch": 0.9707639107198994,
      "grad_norm": 0.3606968820095062,
      "learning_rate": 4.8269241997214166e-08,
      "loss": 0.8918,
      "step": 5790
    },
    {
      "epoch": 0.9709315728806455,
      "grad_norm": 0.4637780487537384,
      "learning_rate": 4.773777717532513e-08,
      "loss": 0.8227,
      "step": 5791
    },
    {
      "epoch": 0.9710992350413916,
      "grad_norm": 0.9876329898834229,
      "learning_rate": 4.720924739880062e-08,
      "loss": 0.7865,
      "step": 5792
    },
    {
      "epoch": 0.9712668972021377,
      "grad_norm": 0.5332596302032471,
      "learning_rate": 4.6683652823513725e-08,
      "loss": 0.8898,
      "step": 5793
    },
    {
      "epoch": 0.9714345593628838,
      "grad_norm": 0.47035887837409973,
      "learning_rate": 4.616099360446713e-08,
      "loss": 0.8151,
      "step": 5794
    },
    {
      "epoch": 0.9716022215236298,
      "grad_norm": 0.40923890471458435,
      "learning_rate": 4.5641269895800864e-08,
      "loss": 0.9865,
      "step": 5795
    },
    {
      "epoch": 0.971769883684376,
      "grad_norm": 0.6284613013267517,
      "learning_rate": 4.512448185078677e-08,
      "loss": 0.9175,
      "step": 5796
    },
    {
      "epoch": 0.971937545845122,
      "grad_norm": 0.4579513967037201,
      "learning_rate": 4.461062962183183e-08,
      "loss": 0.8939,
      "step": 5797
    },
    {
      "epoch": 0.9721052080058682,
      "grad_norm": 0.396099328994751,
      "learning_rate": 4.4099713360478133e-08,
      "loss": 0.8508,
      "step": 5798
    },
    {
      "epoch": 0.9722728701666142,
      "grad_norm": 0.42241689562797546,
      "learning_rate": 4.359173321740184e-08,
      "loss": 0.7861,
      "step": 5799
    },
    {
      "epoch": 0.9724405323273604,
      "grad_norm": 0.35261476039886475,
      "learning_rate": 4.308668934241089e-08,
      "loss": 0.7617,
      "step": 5800
    },
    {
      "epoch": 0.9726081944881064,
      "grad_norm": 0.387504518032074,
      "learning_rate": 4.2584581884450584e-08,
      "loss": 0.8291,
      "step": 5801
    },
    {
      "epoch": 0.9727758566488526,
      "grad_norm": 0.4456618130207062,
      "learning_rate": 4.208541099159691e-08,
      "loss": 0.8707,
      "step": 5802
    },
    {
      "epoch": 0.9729435188095986,
      "grad_norm": 0.5461599826812744,
      "learning_rate": 4.158917681106322e-08,
      "loss": 0.8829,
      "step": 5803
    },
    {
      "epoch": 0.9731111809703448,
      "grad_norm": 0.4105444848537445,
      "learning_rate": 4.109587948919469e-08,
      "loss": 0.858,
      "step": 5804
    },
    {
      "epoch": 0.9732788431310908,
      "grad_norm": 0.937830924987793,
      "learning_rate": 4.0605519171469375e-08,
      "loss": 0.7827,
      "step": 5805
    },
    {
      "epoch": 0.973446505291837,
      "grad_norm": 0.5319908857345581,
      "learning_rate": 4.011809600250272e-08,
      "loss": 0.7846,
      "step": 5806
    },
    {
      "epoch": 0.973614167452583,
      "grad_norm": 0.3267498314380646,
      "learning_rate": 3.9633610126040836e-08,
      "loss": 0.8895,
      "step": 5807
    },
    {
      "epoch": 0.9737818296133292,
      "grad_norm": 0.8419097065925598,
      "learning_rate": 3.9152061684964993e-08,
      "loss": 0.8483,
      "step": 5808
    },
    {
      "epoch": 0.9739494917740752,
      "grad_norm": 0.5133106112480164,
      "learning_rate": 3.8673450821288264e-08,
      "loss": 0.8842,
      "step": 5809
    },
    {
      "epoch": 0.9741171539348213,
      "grad_norm": 0.8418963551521301,
      "learning_rate": 3.8197777676161064e-08,
      "loss": 0.8217,
      "step": 5810
    },
    {
      "epoch": 0.9742848160955674,
      "grad_norm": 0.37620463967323303,
      "learning_rate": 3.772504238986452e-08,
      "loss": 0.7553,
      "step": 5811
    },
    {
      "epoch": 0.9744524782563135,
      "grad_norm": 0.4085863530635834,
      "learning_rate": 3.725524510181378e-08,
      "loss": 0.8598,
      "step": 5812
    },
    {
      "epoch": 0.9746201404170596,
      "grad_norm": 0.3842092752456665,
      "learning_rate": 3.6788385950558e-08,
      "loss": 0.9393,
      "step": 5813
    },
    {
      "epoch": 0.9747878025778057,
      "grad_norm": 0.4535366892814636,
      "learning_rate": 3.63244650737804e-08,
      "loss": 0.8294,
      "step": 5814
    },
    {
      "epoch": 0.9749554647385518,
      "grad_norm": 0.5589142441749573,
      "learning_rate": 3.586348260829486e-08,
      "loss": 0.8977,
      "step": 5815
    },
    {
      "epoch": 0.9751231268992979,
      "grad_norm": 0.4944474399089813,
      "learning_rate": 3.540543869005375e-08,
      "loss": 0.8625,
      "step": 5816
    },
    {
      "epoch": 0.975290789060044,
      "grad_norm": 0.3996366858482361,
      "learning_rate": 3.4950333454137895e-08,
      "loss": 0.8303,
      "step": 5817
    },
    {
      "epoch": 0.9754584512207901,
      "grad_norm": 0.44617748260498047,
      "learning_rate": 3.449816703476438e-08,
      "loss": 0.892,
      "step": 5818
    },
    {
      "epoch": 0.9756261133815362,
      "grad_norm": 0.45459893345832825,
      "learning_rate": 3.40489395652821e-08,
      "loss": 0.8965,
      "step": 5819
    },
    {
      "epoch": 0.9757937755422823,
      "grad_norm": 0.3187973201274872,
      "learning_rate": 3.360265117817396e-08,
      "loss": 0.7541,
      "step": 5820
    },
    {
      "epoch": 0.9759614377030283,
      "grad_norm": 0.38300496339797974,
      "learning_rate": 3.315930200505579e-08,
      "loss": 0.8877,
      "step": 5821
    },
    {
      "epoch": 0.9761290998637745,
      "grad_norm": 0.5122457146644592,
      "learning_rate": 3.271889217667745e-08,
      "loss": 0.8554,
      "step": 5822
    },
    {
      "epoch": 0.9762967620245205,
      "grad_norm": 0.3872210383415222,
      "learning_rate": 3.22814218229206e-08,
      "loss": 0.7912,
      "step": 5823
    },
    {
      "epoch": 0.9764644241852667,
      "grad_norm": 0.48558467626571655,
      "learning_rate": 3.1846891072802034e-08,
      "loss": 0.8185,
      "step": 5824
    },
    {
      "epoch": 0.9766320863460127,
      "grad_norm": 0.4107375144958496,
      "learning_rate": 3.141530005446702e-08,
      "loss": 0.8521,
      "step": 5825
    },
    {
      "epoch": 0.9767997485067589,
      "grad_norm": 0.5492359399795532,
      "learning_rate": 3.098664889520042e-08,
      "loss": 0.8031,
      "step": 5826
    },
    {
      "epoch": 0.976967410667505,
      "grad_norm": 0.3595879077911377,
      "learning_rate": 3.0560937721415554e-08,
      "loss": 0.9382,
      "step": 5827
    },
    {
      "epoch": 0.9771350728282511,
      "grad_norm": 0.6469846367835999,
      "learning_rate": 3.013816665865976e-08,
      "loss": 0.8775,
      "step": 5828
    },
    {
      "epoch": 0.9773027349889972,
      "grad_norm": 0.45741432905197144,
      "learning_rate": 2.971833583161332e-08,
      "loss": 0.8431,
      "step": 5829
    },
    {
      "epoch": 0.9774703971497433,
      "grad_norm": 0.37668368220329285,
      "learning_rate": 2.9301445364089408e-08,
      "loss": 0.8204,
      "step": 5830
    },
    {
      "epoch": 0.9776380593104894,
      "grad_norm": 0.3521329164505005,
      "learning_rate": 2.8887495379035235e-08,
      "loss": 0.8736,
      "step": 5831
    },
    {
      "epoch": 0.9778057214712355,
      "grad_norm": 0.42450135946273804,
      "learning_rate": 2.8476485998528703e-08,
      "loss": 0.8061,
      "step": 5832
    },
    {
      "epoch": 0.9779733836319816,
      "grad_norm": 0.43218979239463806,
      "learning_rate": 2.806841734378174e-08,
      "loss": 0.8153,
      "step": 5833
    },
    {
      "epoch": 0.9781410457927276,
      "grad_norm": 0.4481380581855774,
      "learning_rate": 2.7663289535140302e-08,
      "loss": 0.7531,
      "step": 5834
    },
    {
      "epoch": 0.9783087079534738,
      "grad_norm": 0.41322800517082214,
      "learning_rate": 2.7261102692078822e-08,
      "loss": 0.8489,
      "step": 5835
    },
    {
      "epoch": 0.9784763701142198,
      "grad_norm": 0.45851191878318787,
      "learning_rate": 2.6861856933209086e-08,
      "loss": 0.7952,
      "step": 5836
    },
    {
      "epoch": 0.978644032274966,
      "grad_norm": 0.5079235434532166,
      "learning_rate": 2.6465552376274683e-08,
      "loss": 0.8555,
      "step": 5837
    },
    {
      "epoch": 0.978811694435712,
      "grad_norm": 0.38337695598602295,
      "learning_rate": 2.6072189138147686e-08,
      "loss": 0.7607,
      "step": 5838
    },
    {
      "epoch": 0.9789793565964582,
      "grad_norm": 0.40837207436561584,
      "learning_rate": 2.5681767334837514e-08,
      "loss": 0.8458,
      "step": 5839
    },
    {
      "epoch": 0.9791470187572042,
      "grad_norm": 0.35640430450439453,
      "learning_rate": 2.5294287081485404e-08,
      "loss": 0.8652,
      "step": 5840
    },
    {
      "epoch": 0.9793146809179504,
      "grad_norm": 0.8105623722076416,
      "learning_rate": 2.4909748492362162e-08,
      "loss": 0.873,
      "step": 5841
    },
    {
      "epoch": 0.9794823430786964,
      "grad_norm": 0.4976637661457062,
      "learning_rate": 2.4528151680873748e-08,
      "loss": 0.9144,
      "step": 5842
    },
    {
      "epoch": 0.9796500052394426,
      "grad_norm": 0.5497066378593445,
      "learning_rate": 2.4149496759557912e-08,
      "loss": 0.8361,
      "step": 5843
    },
    {
      "epoch": 0.9798176674001886,
      "grad_norm": 0.5602542757987976,
      "learning_rate": 2.377378384008422e-08,
      "loss": 0.8222,
      "step": 5844
    },
    {
      "epoch": 0.9799853295609348,
      "grad_norm": 0.3840111494064331,
      "learning_rate": 2.340101303325626e-08,
      "loss": 0.8269,
      "step": 5845
    },
    {
      "epoch": 0.9801529917216808,
      "grad_norm": 0.5249937176704407,
      "learning_rate": 2.3031184449007204e-08,
      "loss": 0.8154,
      "step": 5846
    },
    {
      "epoch": 0.980320653882427,
      "grad_norm": 0.8526900410652161,
      "learning_rate": 2.2664298196405367e-08,
      "loss": 0.8309,
      "step": 5847
    },
    {
      "epoch": 0.980488316043173,
      "grad_norm": 1.0609358549118042,
      "learning_rate": 2.2300354383648636e-08,
      "loss": 0.771,
      "step": 5848
    },
    {
      "epoch": 0.9806559782039191,
      "grad_norm": 0.36970868706703186,
      "learning_rate": 2.193935311807005e-08,
      "loss": 0.8063,
      "step": 5849
    },
    {
      "epoch": 0.9808236403646652,
      "grad_norm": 0.4812318682670593,
      "learning_rate": 2.1581294506132223e-08,
      "loss": 0.8783,
      "step": 5850
    },
    {
      "epoch": 0.9809913025254113,
      "grad_norm": 0.39542368054389954,
      "learning_rate": 2.1226178653432905e-08,
      "loss": 0.8123,
      "step": 5851
    },
    {
      "epoch": 0.9811589646861574,
      "grad_norm": 0.4277333617210388,
      "learning_rate": 2.0874005664697218e-08,
      "loss": 0.8418,
      "step": 5852
    },
    {
      "epoch": 0.9813266268469035,
      "grad_norm": 0.5052667856216431,
      "learning_rate": 2.052477564378763e-08,
      "loss": 0.7711,
      "step": 5853
    },
    {
      "epoch": 0.9814942890076496,
      "grad_norm": 0.5133655071258545,
      "learning_rate": 2.0178488693695096e-08,
      "loss": 0.9144,
      "step": 5854
    },
    {
      "epoch": 0.9816619511683957,
      "grad_norm": 0.4648545980453491,
      "learning_rate": 1.98351449165457e-08,
      "loss": 0.8852,
      "step": 5855
    },
    {
      "epoch": 0.9818296133291418,
      "grad_norm": 1.2075058221817017,
      "learning_rate": 1.9494744413594003e-08,
      "loss": 0.7866,
      "step": 5856
    },
    {
      "epoch": 0.9819972754898879,
      "grad_norm": 0.4560854434967041,
      "learning_rate": 1.9157287285228587e-08,
      "loss": 0.8429,
      "step": 5857
    },
    {
      "epoch": 0.982164937650634,
      "grad_norm": 0.3814972937107086,
      "learning_rate": 1.8822773630969847e-08,
      "loss": 0.8514,
      "step": 5858
    },
    {
      "epoch": 0.9823325998113801,
      "grad_norm": 0.38170596957206726,
      "learning_rate": 1.8491203549471092e-08,
      "loss": 0.7808,
      "step": 5859
    },
    {
      "epoch": 0.9825002619721261,
      "grad_norm": 0.40041080117225647,
      "learning_rate": 1.8162577138515215e-08,
      "loss": 0.8326,
      "step": 5860
    },
    {
      "epoch": 0.9826679241328723,
      "grad_norm": 0.37729009985923767,
      "learning_rate": 1.783689449501802e-08,
      "loss": 0.8323,
      "step": 5861
    },
    {
      "epoch": 0.9828355862936183,
      "grad_norm": 0.4086211919784546,
      "learning_rate": 1.7514155715028236e-08,
      "loss": 0.8604,
      "step": 5862
    },
    {
      "epoch": 0.9830032484543645,
      "grad_norm": 0.3598542809486389,
      "learning_rate": 1.7194360893725282e-08,
      "loss": 0.8003,
      "step": 5863
    },
    {
      "epoch": 0.9831709106151105,
      "grad_norm": 0.5794903039932251,
      "learning_rate": 1.6877510125420384e-08,
      "loss": 0.9411,
      "step": 5864
    },
    {
      "epoch": 0.9833385727758567,
      "grad_norm": 0.390186071395874,
      "learning_rate": 1.656360350355768e-08,
      "loss": 0.7728,
      "step": 5865
    },
    {
      "epoch": 0.9835062349366027,
      "grad_norm": 0.4856814444065094,
      "learning_rate": 1.6252641120710898e-08,
      "loss": 0.9004,
      "step": 5866
    },
    {
      "epoch": 0.9836738970973489,
      "grad_norm": 0.5970719456672668,
      "learning_rate": 1.5944623068586683e-08,
      "loss": 0.872,
      "step": 5867
    },
    {
      "epoch": 0.9838415592580949,
      "grad_norm": 0.3844272494316101,
      "learning_rate": 1.5639549438023483e-08,
      "loss": 0.6954,
      "step": 5868
    },
    {
      "epoch": 0.9840092214188411,
      "grad_norm": 0.3902965188026428,
      "learning_rate": 1.5337420318992656e-08,
      "loss": 0.8182,
      "step": 5869
    },
    {
      "epoch": 0.9841768835795871,
      "grad_norm": 0.42631059885025024,
      "learning_rate": 1.5038235800595158e-08,
      "loss": 0.8431,
      "step": 5870
    },
    {
      "epoch": 0.9843445457403333,
      "grad_norm": 0.4387223720550537,
      "learning_rate": 1.4741995971062627e-08,
      "loss": 0.7704,
      "step": 5871
    },
    {
      "epoch": 0.9845122079010793,
      "grad_norm": 0.4834582805633545,
      "learning_rate": 1.4448700917761849e-08,
      "loss": 0.9173,
      "step": 5872
    },
    {
      "epoch": 0.9846798700618254,
      "grad_norm": 0.4982442557811737,
      "learning_rate": 1.4158350727188075e-08,
      "loss": 0.7943,
      "step": 5873
    },
    {
      "epoch": 0.9848475322225715,
      "grad_norm": 0.37011417746543884,
      "learning_rate": 1.3870945484969478e-08,
      "loss": 0.9477,
      "step": 5874
    },
    {
      "epoch": 0.9850151943833176,
      "grad_norm": 0.49638018012046814,
      "learning_rate": 1.3586485275864925e-08,
      "loss": 0.7223,
      "step": 5875
    },
    {
      "epoch": 0.9851828565440637,
      "grad_norm": 0.33733877539634705,
      "learning_rate": 1.3304970183766198e-08,
      "loss": 0.8092,
      "step": 5876
    },
    {
      "epoch": 0.9853505187048098,
      "grad_norm": 0.3381383717060089,
      "learning_rate": 1.3026400291695774e-08,
      "loss": 0.8701,
      "step": 5877
    },
    {
      "epoch": 0.9855181808655559,
      "grad_norm": 0.35699698328971863,
      "learning_rate": 1.2750775681805717e-08,
      "loss": 0.8487,
      "step": 5878
    },
    {
      "epoch": 0.985685843026302,
      "grad_norm": 0.4587336480617523,
      "learning_rate": 1.2478096435383225e-08,
      "loss": 0.9753,
      "step": 5879
    },
    {
      "epoch": 0.9858535051870481,
      "grad_norm": 0.5707885026931763,
      "learning_rate": 1.2208362632842863e-08,
      "loss": 0.8661,
      "step": 5880
    },
    {
      "epoch": 0.9860211673477942,
      "grad_norm": 0.40471991896629333,
      "learning_rate": 1.1941574353733221e-08,
      "loss": 0.9193,
      "step": 5881
    },
    {
      "epoch": 0.9861888295085403,
      "grad_norm": 0.4383912980556488,
      "learning_rate": 1.1677731676733584e-08,
      "loss": 0.8396,
      "step": 5882
    },
    {
      "epoch": 0.9863564916692864,
      "grad_norm": 0.4767834544181824,
      "learning_rate": 1.1416834679655042e-08,
      "loss": 0.8994,
      "step": 5883
    },
    {
      "epoch": 0.9865241538300324,
      "grad_norm": 0.593654215335846,
      "learning_rate": 1.1158883439438273e-08,
      "loss": 0.7644,
      "step": 5884
    },
    {
      "epoch": 0.9866918159907786,
      "grad_norm": 0.38353127241134644,
      "learning_rate": 1.0903878032156867e-08,
      "loss": 0.8433,
      "step": 5885
    },
    {
      "epoch": 0.9868594781515246,
      "grad_norm": 0.43884798884391785,
      "learning_rate": 1.0651818533015113e-08,
      "loss": 0.7387,
      "step": 5886
    },
    {
      "epoch": 0.9870271403122708,
      "grad_norm": 0.3942774534225464,
      "learning_rate": 1.0402705016349102e-08,
      "loss": 0.8686,
      "step": 5887
    },
    {
      "epoch": 0.9871948024730168,
      "grad_norm": 0.36644846200942993,
      "learning_rate": 1.0156537555623402e-08,
      "loss": 0.7794,
      "step": 5888
    },
    {
      "epoch": 0.987362464633763,
      "grad_norm": 0.4375738203525543,
      "learning_rate": 9.913316223437718e-09,
      "loss": 0.831,
      "step": 5889
    },
    {
      "epoch": 0.987530126794509,
      "grad_norm": 0.49118849635124207,
      "learning_rate": 9.67304109152023e-09,
      "loss": 0.8345,
      "step": 5890
    },
    {
      "epoch": 0.9876977889552552,
      "grad_norm": 0.4413753151893616,
      "learning_rate": 9.43571223073092e-09,
      "loss": 0.8346,
      "step": 5891
    },
    {
      "epoch": 0.9878654511160012,
      "grad_norm": 0.4078384339809418,
      "learning_rate": 9.20132971106269e-09,
      "loss": 0.805,
      "step": 5892
    },
    {
      "epoch": 0.9880331132767474,
      "grad_norm": 0.42576390504837036,
      "learning_rate": 8.969893601634694e-09,
      "loss": 0.8319,
      "step": 5893
    },
    {
      "epoch": 0.9882007754374934,
      "grad_norm": 0.46708157658576965,
      "learning_rate": 8.741403970703443e-09,
      "loss": 0.8839,
      "step": 5894
    },
    {
      "epoch": 0.9883684375982396,
      "grad_norm": 0.34723997116088867,
      "learning_rate": 8.515860885651706e-09,
      "loss": 0.798,
      "step": 5895
    },
    {
      "epoch": 0.9885360997589856,
      "grad_norm": 0.37282437086105347,
      "learning_rate": 8.293264412995161e-09,
      "loss": 0.8861,
      "step": 5896
    },
    {
      "epoch": 0.9887037619197317,
      "grad_norm": 0.42206084728240967,
      "learning_rate": 8.073614618380188e-09,
      "loss": 0.8341,
      "step": 5897
    },
    {
      "epoch": 0.9888714240804778,
      "grad_norm": 0.4986555278301239,
      "learning_rate": 7.856911566584968e-09,
      "loss": 0.7804,
      "step": 5898
    },
    {
      "epoch": 0.9890390862412239,
      "grad_norm": 1.1532371044158936,
      "learning_rate": 7.64315532151727e-09,
      "loss": 0.7834,
      "step": 5899
    },
    {
      "epoch": 0.98920674840197,
      "grad_norm": 1.007047414779663,
      "learning_rate": 7.432345946216668e-09,
      "loss": 0.837,
      "step": 5900
    },
    {
      "epoch": 0.9893744105627161,
      "grad_norm": 0.46277493238449097,
      "learning_rate": 7.22448350285343e-09,
      "loss": 0.8525,
      "step": 5901
    },
    {
      "epoch": 0.9895420727234622,
      "grad_norm": 0.38527214527130127,
      "learning_rate": 7.01956805272852e-09,
      "loss": 0.7731,
      "step": 5902
    },
    {
      "epoch": 0.9897097348842083,
      "grad_norm": 0.45211055874824524,
      "learning_rate": 6.817599656273599e-09,
      "loss": 0.9063,
      "step": 5903
    },
    {
      "epoch": 0.9898773970449544,
      "grad_norm": 0.42291173338890076,
      "learning_rate": 6.6185783730521315e-09,
      "loss": 0.7942,
      "step": 5904
    },
    {
      "epoch": 0.9900450592057005,
      "grad_norm": 0.35738232731819153,
      "learning_rate": 6.422504261759388e-09,
      "loss": 0.7929,
      "step": 5905
    },
    {
      "epoch": 0.9902127213664466,
      "grad_norm": 0.5838137865066528,
      "learning_rate": 6.229377380218005e-09,
      "loss": 0.845,
      "step": 5906
    },
    {
      "epoch": 0.9903803835271927,
      "grad_norm": 0.4058472812175751,
      "learning_rate": 6.039197785383533e-09,
      "loss": 0.8274,
      "step": 5907
    },
    {
      "epoch": 0.9905480456879387,
      "grad_norm": 0.44134339690208435,
      "learning_rate": 5.8519655333444394e-09,
      "loss": 0.8507,
      "step": 5908
    },
    {
      "epoch": 0.9907157078486849,
      "grad_norm": 0.34918412566185,
      "learning_rate": 5.667680679315446e-09,
      "loss": 0.7411,
      "step": 5909
    },
    {
      "epoch": 0.9908833700094309,
      "grad_norm": 0.3565990924835205,
      "learning_rate": 5.4863432776453006e-09,
      "loss": 0.8605,
      "step": 5910
    },
    {
      "epoch": 0.9910510321701771,
      "grad_norm": 0.7799057960510254,
      "learning_rate": 5.307953381813447e-09,
      "loss": 0.793,
      "step": 5911
    },
    {
      "epoch": 0.9912186943309232,
      "grad_norm": 0.412158340215683,
      "learning_rate": 5.132511044427801e-09,
      "loss": 0.8211,
      "step": 5912
    },
    {
      "epoch": 0.9913863564916693,
      "grad_norm": 0.4002961218357086,
      "learning_rate": 4.960016317229199e-09,
      "loss": 0.8245,
      "step": 5913
    },
    {
      "epoch": 0.9915540186524154,
      "grad_norm": 0.4357675611972809,
      "learning_rate": 4.7904692510891695e-09,
      "loss": 0.7889,
      "step": 5914
    },
    {
      "epoch": 0.9917216808131615,
      "grad_norm": 0.4863226115703583,
      "learning_rate": 4.623869896008826e-09,
      "loss": 0.8934,
      "step": 5915
    },
    {
      "epoch": 0.9918893429739076,
      "grad_norm": 0.41823381185531616,
      "learning_rate": 4.4602183011188685e-09,
      "loss": 0.9074,
      "step": 5916
    },
    {
      "epoch": 0.9920570051346537,
      "grad_norm": 0.3756341338157654,
      "learning_rate": 4.299514514684022e-09,
      "loss": 0.7179,
      "step": 5917
    },
    {
      "epoch": 0.9922246672953998,
      "grad_norm": 0.34542134404182434,
      "learning_rate": 4.141758584097488e-09,
      "loss": 0.7625,
      "step": 5918
    },
    {
      "epoch": 0.9923923294561459,
      "grad_norm": 0.43031448125839233,
      "learning_rate": 3.986950555883162e-09,
      "loss": 0.7759,
      "step": 5919
    },
    {
      "epoch": 0.992559991616892,
      "grad_norm": 0.38881802558898926,
      "learning_rate": 3.835090475695635e-09,
      "loss": 0.8077,
      "step": 5920
    },
    {
      "epoch": 0.992727653777638,
      "grad_norm": 0.3611317574977875,
      "learning_rate": 3.686178388321304e-09,
      "loss": 0.7856,
      "step": 5921
    },
    {
      "epoch": 0.9928953159383842,
      "grad_norm": 0.47091519832611084,
      "learning_rate": 3.540214337675041e-09,
      "loss": 0.7925,
      "step": 5922
    },
    {
      "epoch": 0.9930629780991302,
      "grad_norm": 0.46980783343315125,
      "learning_rate": 3.3971983668035224e-09,
      "loss": 0.8761,
      "step": 5923
    },
    {
      "epoch": 0.9932306402598764,
      "grad_norm": 0.3938907980918884,
      "learning_rate": 3.2571305178841218e-09,
      "loss": 0.775,
      "step": 5924
    },
    {
      "epoch": 0.9933983024206224,
      "grad_norm": 0.42583513259887695,
      "learning_rate": 3.1200108322249066e-09,
      "loss": 0.8806,
      "step": 5925
    },
    {
      "epoch": 0.9935659645813686,
      "grad_norm": 0.4112643897533417,
      "learning_rate": 2.985839350264641e-09,
      "loss": 0.8161,
      "step": 5926
    },
    {
      "epoch": 0.9937336267421146,
      "grad_norm": 0.4502149820327759,
      "learning_rate": 2.854616111570563e-09,
      "loss": 0.8438,
      "step": 5927
    },
    {
      "epoch": 0.9939012889028608,
      "grad_norm": 0.5014587044715881,
      "learning_rate": 2.726341154843937e-09,
      "loss": 0.8073,
      "step": 5928
    },
    {
      "epoch": 0.9940689510636068,
      "grad_norm": 0.42569249868392944,
      "learning_rate": 2.6010145179133916e-09,
      "loss": 0.8726,
      "step": 5929
    },
    {
      "epoch": 0.994236613224353,
      "grad_norm": 0.5752376914024353,
      "learning_rate": 2.4786362377393624e-09,
      "loss": 0.7958,
      "step": 5930
    },
    {
      "epoch": 0.994404275385099,
      "grad_norm": 0.439833402633667,
      "learning_rate": 2.3592063504129793e-09,
      "loss": 0.8239,
      "step": 5931
    },
    {
      "epoch": 0.9945719375458452,
      "grad_norm": 0.3875184655189514,
      "learning_rate": 2.242724891156067e-09,
      "loss": 0.792,
      "step": 5932
    },
    {
      "epoch": 0.9947395997065912,
      "grad_norm": 0.42492565512657166,
      "learning_rate": 2.1291918943189272e-09,
      "loss": 0.8352,
      "step": 5933
    },
    {
      "epoch": 0.9949072618673374,
      "grad_norm": 0.5234774947166443,
      "learning_rate": 2.0186073933858852e-09,
      "loss": 0.7744,
      "step": 5934
    },
    {
      "epoch": 0.9950749240280834,
      "grad_norm": 0.40094274282455444,
      "learning_rate": 1.9109714209675223e-09,
      "loss": 0.8669,
      "step": 5935
    },
    {
      "epoch": 0.9952425861888295,
      "grad_norm": 0.5302832126617432,
      "learning_rate": 1.8062840088095558e-09,
      "loss": 0.8151,
      "step": 5936
    },
    {
      "epoch": 0.9954102483495756,
      "grad_norm": 0.438462495803833,
      "learning_rate": 1.7045451877828467e-09,
      "loss": 0.843,
      "step": 5937
    },
    {
      "epoch": 0.9955779105103217,
      "grad_norm": 0.489299476146698,
      "learning_rate": 1.6057549878933931e-09,
      "loss": 0.9438,
      "step": 5938
    },
    {
      "epoch": 0.9957455726710678,
      "grad_norm": 0.4687711298465729,
      "learning_rate": 1.5099134382756676e-09,
      "loss": 0.8566,
      "step": 5939
    },
    {
      "epoch": 0.9959132348318139,
      "grad_norm": 0.43907466530799866,
      "learning_rate": 1.4170205671937275e-09,
      "loss": 0.7816,
      "step": 5940
    },
    {
      "epoch": 0.99608089699256,
      "grad_norm": 0.3608318865299225,
      "learning_rate": 1.3270764020423265e-09,
      "loss": 0.8377,
      "step": 5941
    },
    {
      "epoch": 0.9962485591533061,
      "grad_norm": 0.4274871349334717,
      "learning_rate": 1.2400809693480232e-09,
      "loss": 0.8222,
      "step": 5942
    },
    {
      "epoch": 0.9964162213140522,
      "grad_norm": 0.4970013201236725,
      "learning_rate": 1.156034294766961e-09,
      "loss": 0.8703,
      "step": 5943
    },
    {
      "epoch": 0.9965838834747983,
      "grad_norm": 0.4293815791606903,
      "learning_rate": 1.0749364030848697e-09,
      "loss": 0.8397,
      "step": 5944
    },
    {
      "epoch": 0.9967515456355444,
      "grad_norm": 0.451774537563324,
      "learning_rate": 9.96787318218173e-10,
      "loss": 0.8459,
      "step": 5945
    },
    {
      "epoch": 0.9969192077962905,
      "grad_norm": 0.5851801037788391,
      "learning_rate": 9.215870632162117e-10,
      "loss": 0.873,
      "step": 5946
    },
    {
      "epoch": 0.9970868699570365,
      "grad_norm": 0.6409519910812378,
      "learning_rate": 8.493356602545799e-10,
      "loss": 0.8277,
      "step": 5947
    },
    {
      "epoch": 0.9972545321177827,
      "grad_norm": 0.4242671728134155,
      "learning_rate": 7.800331306406783e-10,
      "loss": 0.8626,
      "step": 5948
    },
    {
      "epoch": 0.9974221942785287,
      "grad_norm": 0.5252429246902466,
      "learning_rate": 7.136794948137127e-10,
      "loss": 0.8759,
      "step": 5949
    },
    {
      "epoch": 0.9975898564392749,
      "grad_norm": 0.4827357232570648,
      "learning_rate": 6.502747723424741e-10,
      "loss": 0.9101,
      "step": 5950
    },
    {
      "epoch": 0.9977575186000209,
      "grad_norm": 0.45835551619529724,
      "learning_rate": 5.898189819253386e-10,
      "loss": 0.7571,
      "step": 5951
    },
    {
      "epoch": 0.9979251807607671,
      "grad_norm": 0.4265470802783966,
      "learning_rate": 5.323121413902677e-10,
      "loss": 0.7681,
      "step": 5952
    },
    {
      "epoch": 0.9980928429215131,
      "grad_norm": 0.3536740839481354,
      "learning_rate": 4.777542676992486e-10,
      "loss": 0.794,
      "step": 5953
    },
    {
      "epoch": 0.9982605050822593,
      "grad_norm": 1.0968900918960571,
      "learning_rate": 4.2614537693941304e-10,
      "loss": 0.7514,
      "step": 5954
    },
    {
      "epoch": 0.9984281672430053,
      "grad_norm": 0.3914303183555603,
      "learning_rate": 3.7748548433302887e-10,
      "loss": 0.7851,
      "step": 5955
    },
    {
      "epoch": 0.9985958294037515,
      "grad_norm": 0.39772719144821167,
      "learning_rate": 3.317746042297287e-10,
      "loss": 0.8082,
      "step": 5956
    },
    {
      "epoch": 0.9987634915644975,
      "grad_norm": 0.5874171853065491,
      "learning_rate": 2.8901275010984055e-10,
      "loss": 0.8401,
      "step": 5957
    },
    {
      "epoch": 0.9989311537252437,
      "grad_norm": 0.42458993196487427,
      "learning_rate": 2.4919993458549783e-10,
      "loss": 0.8856,
      "step": 5958
    },
    {
      "epoch": 0.9990988158859897,
      "grad_norm": 0.622805655002594,
      "learning_rate": 2.1233616939619896e-10,
      "loss": 0.775,
      "step": 5959
    },
    {
      "epoch": 0.9992664780467359,
      "grad_norm": 0.4151398837566376,
      "learning_rate": 1.784214654143579e-10,
      "loss": 0.8683,
      "step": 5960
    },
    {
      "epoch": 0.9994341402074819,
      "grad_norm": 0.42758768796920776,
      "learning_rate": 1.4745583264197394e-10,
      "loss": 0.8124,
      "step": 5961
    },
    {
      "epoch": 0.999601802368228,
      "grad_norm": 0.5456928014755249,
      "learning_rate": 1.1943928021174167e-10,
      "loss": 0.8261,
      "step": 5962
    },
    {
      "epoch": 0.9997694645289741,
      "grad_norm": 0.9677426218986511,
      "learning_rate": 9.437181638483061e-11,
      "loss": 0.8238,
      "step": 5963
    },
    {
      "epoch": 0.9999371266897202,
      "grad_norm": 0.5349898934364319,
      "learning_rate": 7.225344855532612e-11,
      "loss": 0.8254,
      "step": 5964
    },
    {
      "epoch": 0.9999371266897202,
      "step": 5964,
      "total_flos": 5.467049278255923e+18,
      "train_loss": 0.8923786733971595,
      "train_runtime": 124734.865,
      "train_samples_per_second": 6.12,
      "train_steps_per_second": 0.048
    }
  ],
  "logging_steps": 1.0,
  "max_steps": 5964,
  "num_input_tokens_seen": 0,
  "num_train_epochs": 1,
  "save_steps": 500,
  "stateful_callbacks": {
    "TrainerControl": {
      "args": {
        "should_epoch_stop": false,
        "should_evaluate": false,
        "should_log": false,
        "should_save": true,
        "should_training_stop": true
      },
      "attributes": {}
    }
  },
  "total_flos": 5.467049278255923e+18,
  "train_batch_size": 1,
  "trial_name": null,
  "trial_params": null
}