diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4641 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.9814814814814814,
+  "eval_steps": 81,
+  "global_step": 648,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0030864197530864196,
+      "grad_norm": 0.11897344887256622,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 0.6253,
+      "step": 1
+    },
+    {
+      "epoch": 0.0030864197530864196,
+      "eval_loss": 0.6252603530883789,
+      "eval_runtime": 44.2936,
+      "eval_samples_per_second": 8.308,
+      "eval_steps_per_second": 1.039,
+      "step": 1
+    },
+    {
+      "epoch": 0.006172839506172839,
+      "grad_norm": 0.11417510360479355,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 0.6376,
+      "step": 2
+    },
+    {
+      "epoch": 0.009259259259259259,
+      "grad_norm": 0.0693814605474472,
+      "learning_rate": 3e-06,
+      "loss": 0.2684,
+      "step": 3
+    },
+    {
+      "epoch": 0.012345679012345678,
+      "grad_norm": 0.1110842302441597,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 0.5096,
+      "step": 4
+    },
+    {
+      "epoch": 0.015432098765432098,
+      "grad_norm": 0.09205043315887451,
+      "learning_rate": 5e-06,
+      "loss": 0.5674,
+      "step": 5
+    },
+    {
+      "epoch": 0.018518518518518517,
+      "grad_norm": 0.1063380092382431,
+      "learning_rate": 6e-06,
+      "loss": 0.6219,
+      "step": 6
+    },
+    {
+      "epoch": 0.021604938271604937,
+      "grad_norm": 0.0740552470088005,
+      "learning_rate": 7e-06,
+      "loss": 0.5478,
+      "step": 7
+    },
+    {
+      "epoch": 0.024691358024691357,
+      "grad_norm": 0.10674550384283066,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 0.6168,
+      "step": 8
+    },
+    {
+      "epoch": 0.027777777777777776,
+      "grad_norm": 0.1061239168047905,
+      "learning_rate": 9e-06,
+      "loss": 0.7106,
+      "step": 9
+    },
+    {
+      "epoch": 0.030864197530864196,
+      "grad_norm": 0.10123332589864731,
+      "learning_rate": 1e-05,
+      "loss": 0.5221,
+      "step": 10
+    },
+    {
+      "epoch": 0.033950617283950615,
+      "grad_norm": 0.06680818647146225,
+      "learning_rate": 9.999939382570075e-06,
+      "loss": 0.2592,
+      "step": 11
+    },
+    {
+      "epoch": 0.037037037037037035,
+      "grad_norm": 0.09670277684926987,
+      "learning_rate": 9.999757531750086e-06,
+      "loss": 0.5183,
+      "step": 12
+    },
+    {
+      "epoch": 0.040123456790123455,
+      "grad_norm": 0.07567557692527771,
+      "learning_rate": 9.999454451949364e-06,
+      "loss": 0.3257,
+      "step": 13
+    },
+    {
+      "epoch": 0.043209876543209874,
+      "grad_norm": 0.10101059824228287,
+      "learning_rate": 9.999030150516681e-06,
+      "loss": 0.4788,
+      "step": 14
+    },
+    {
+      "epoch": 0.046296296296296294,
+      "grad_norm": 0.1238669604063034,
+      "learning_rate": 9.998484637740058e-06,
+      "loss": 0.6218,
+      "step": 15
+    },
+    {
+      "epoch": 0.04938271604938271,
+      "grad_norm": 0.10699903219938278,
+      "learning_rate": 9.997817926846528e-06,
+      "loss": 0.6429,
+      "step": 16
+    },
+    {
+      "epoch": 0.05246913580246913,
+      "grad_norm": 0.08470468968153,
+      "learning_rate": 9.997030034001815e-06,
+      "loss": 0.3134,
+      "step": 17
+    },
+    {
+      "epoch": 0.05555555555555555,
+      "grad_norm": 0.1229688748717308,
+      "learning_rate": 9.99612097830993e-06,
+      "loss": 0.712,
+      "step": 18
+    },
+    {
+      "epoch": 0.05864197530864197,
+      "grad_norm": 0.10526233166456223,
+      "learning_rate": 9.995090781812724e-06,
+      "loss": 0.504,
+      "step": 19
+    },
+    {
+      "epoch": 0.06172839506172839,
+      "grad_norm": 0.11165868490934372,
+      "learning_rate": 9.993939469489342e-06,
+      "loss": 0.5122,
+      "step": 20
+    },
+    {
+      "epoch": 0.06481481481481481,
+      "grad_norm": 0.09065920859575272,
+      "learning_rate": 9.99266706925562e-06,
+      "loss": 0.4664,
+      "step": 21
+    },
+    {
+      "epoch": 0.06790123456790123,
+      "grad_norm": 0.10060250014066696,
+      "learning_rate": 9.991273611963413e-06,
+      "loss": 0.4732,
+      "step": 22
+    },
+    {
+      "epoch": 0.07098765432098765,
+      "grad_norm": 0.10402392596006393,
+      "learning_rate": 9.98975913139984e-06,
+      "loss": 0.4899,
+      "step": 23
+    },
+    {
+      "epoch": 0.07407407407407407,
+      "grad_norm": 0.11345162242650986,
+      "learning_rate": 9.98812366428647e-06,
+      "loss": 0.5365,
+      "step": 24
+    },
+    {
+      "epoch": 0.07716049382716049,
+      "grad_norm": 0.1189904510974884,
+      "learning_rate": 9.986367250278423e-06,
+      "loss": 0.6293,
+      "step": 25
+    },
+    {
+      "epoch": 0.08024691358024691,
+      "grad_norm": 0.11722761392593384,
+      "learning_rate": 9.984489931963429e-06,
+      "loss": 0.4991,
+      "step": 26
+    },
+    {
+      "epoch": 0.08333333333333333,
+      "grad_norm": 0.08803360909223557,
+      "learning_rate": 9.982491754860763e-06,
+      "loss": 0.381,
+      "step": 27
+    },
+    {
+      "epoch": 0.08641975308641975,
+      "grad_norm": 0.11037921905517578,
+      "learning_rate": 9.980372767420179e-06,
+      "loss": 0.5814,
+      "step": 28
+    },
+    {
+      "epoch": 0.08950617283950617,
+      "grad_norm": 0.0851665586233139,
+      "learning_rate": 9.978133021020697e-06,
+      "loss": 0.3629,
+      "step": 29
+    },
+    {
+      "epoch": 0.09259259259259259,
+      "grad_norm": 0.10195960849523544,
+      "learning_rate": 9.97577256996939e-06,
+      "loss": 0.5672,
+      "step": 30
+    },
+    {
+      "epoch": 0.09567901234567901,
+      "grad_norm": 0.12112904340028763,
+      "learning_rate": 9.97329147150005e-06,
+      "loss": 0.6165,
+      "step": 31
+    },
+    {
+      "epoch": 0.09876543209876543,
+      "grad_norm": 0.07611838728189468,
+      "learning_rate": 9.970689785771798e-06,
+      "loss": 0.3902,
+      "step": 32
+    },
+    {
+      "epoch": 0.10185185185185185,
+      "grad_norm": 0.1013374775648117,
+      "learning_rate": 9.96796757586764e-06,
+      "loss": 0.5096,
+      "step": 33
+    },
+    {
+      "epoch": 0.10493827160493827,
+      "grad_norm": 0.08809865266084671,
+      "learning_rate": 9.965124907792916e-06,
+      "loss": 0.3333,
+      "step": 34
+    },
+    {
+      "epoch": 0.10802469135802469,
+      "grad_norm": 0.0764087364077568,
+      "learning_rate": 9.962161850473723e-06,
+      "loss": 0.3461,
+      "step": 35
+    },
+    {
+      "epoch": 0.1111111111111111,
+      "grad_norm": 0.0995788499712944,
+      "learning_rate": 9.95907847575523e-06,
+      "loss": 0.4225,
+      "step": 36
+    },
+    {
+      "epoch": 0.11419753086419752,
+      "grad_norm": 0.11751396954059601,
+      "learning_rate": 9.955874858399936e-06,
+      "loss": 0.4991,
+      "step": 37
+    },
+    {
+      "epoch": 0.11728395061728394,
+      "grad_norm": 0.10502217710018158,
+      "learning_rate": 9.952551076085864e-06,
+      "loss": 0.5847,
+      "step": 38
+    },
+    {
+      "epoch": 0.12037037037037036,
+      "grad_norm": 0.1077880784869194,
+      "learning_rate": 9.949107209404664e-06,
+      "loss": 0.4901,
+      "step": 39
+    },
+    {
+      "epoch": 0.12345679012345678,
+      "grad_norm": 0.08844556659460068,
+      "learning_rate": 9.945543341859681e-06,
+      "loss": 0.5752,
+      "step": 40
+    },
+    {
+      "epoch": 0.12654320987654322,
+      "grad_norm": 0.10771756619215012,
+      "learning_rate": 9.94185955986391e-06,
+      "loss": 0.5393,
+      "step": 41
+    },
+    {
+      "epoch": 0.12962962962962962,
+      "grad_norm": 0.07496192306280136,
+      "learning_rate": 9.938055952737908e-06,
+      "loss": 0.3334,
+      "step": 42
+    },
+    {
+      "epoch": 0.13271604938271606,
+      "grad_norm": 0.106163389980793,
+      "learning_rate": 9.934132612707631e-06,
+      "loss": 0.5319,
+      "step": 43
+    },
+    {
+      "epoch": 0.13580246913580246,
+      "grad_norm": 0.09276831895112991,
+      "learning_rate": 9.930089634902197e-06,
+      "loss": 0.486,
+      "step": 44
+    },
+    {
+      "epoch": 0.1388888888888889,
+      "grad_norm": 0.09449384361505508,
+      "learning_rate": 9.925927117351573e-06,
+      "loss": 0.3858,
+      "step": 45
+    },
+    {
+      "epoch": 0.1419753086419753,
+      "grad_norm": 0.07955848425626755,
+      "learning_rate": 9.921645160984205e-06,
+      "loss": 0.4648,
+      "step": 46
+    },
+    {
+      "epoch": 0.14506172839506173,
+      "grad_norm": 0.10575301945209503,
+      "learning_rate": 9.917243869624573e-06,
+      "loss": 0.4704,
+      "step": 47
+    },
+    {
+      "epoch": 0.14814814814814814,
+      "grad_norm": 0.0714716911315918,
+      "learning_rate": 9.91272334999066e-06,
+      "loss": 0.372,
+      "step": 48
+    },
+    {
+      "epoch": 0.15123456790123457,
+      "grad_norm": 0.08894475549459457,
+      "learning_rate": 9.908083711691383e-06,
+      "loss": 0.5005,
+      "step": 49
+    },
+    {
+      "epoch": 0.15432098765432098,
+      "grad_norm": 0.0800170972943306,
+      "learning_rate": 9.903325067223918e-06,
+      "loss": 0.3688,
+      "step": 50
+    },
+    {
+      "epoch": 0.1574074074074074,
+      "grad_norm": 0.09310433268547058,
+      "learning_rate": 9.898447531970989e-06,
+      "loss": 0.5127,
+      "step": 51
+    },
+    {
+      "epoch": 0.16049382716049382,
+      "grad_norm": 0.07690192013978958,
+      "learning_rate": 9.893451224198051e-06,
+      "loss": 0.2993,
+      "step": 52
+    },
+    {
+      "epoch": 0.16358024691358025,
+      "grad_norm": 0.08025282621383667,
+      "learning_rate": 9.888336265050443e-06,
+      "loss": 0.4004,
+      "step": 53
+    },
+    {
+      "epoch": 0.16666666666666666,
+      "grad_norm": 0.06500386446714401,
+      "learning_rate": 9.883102778550434e-06,
+      "loss": 0.3317,
+      "step": 54
+    },
+    {
+      "epoch": 0.1697530864197531,
+      "grad_norm": 0.07926575839519501,
+      "learning_rate": 9.877750891594224e-06,
+      "loss": 0.3606,
+      "step": 55
+    },
+    {
+      "epoch": 0.1728395061728395,
+      "grad_norm": 0.07245253026485443,
+      "learning_rate": 9.872280733948867e-06,
+      "loss": 0.4437,
+      "step": 56
+    },
+    {
+      "epoch": 0.17592592592592593,
+      "grad_norm": 0.07353054732084274,
+      "learning_rate": 9.866692438249124e-06,
+      "loss": 0.36,
+      "step": 57
+    },
+    {
+      "epoch": 0.17901234567901234,
+      "grad_norm": 0.09307980537414551,
+      "learning_rate": 9.86098613999424e-06,
+      "loss": 0.5175,
+      "step": 58
+    },
+    {
+      "epoch": 0.18209876543209877,
+      "grad_norm": 0.07782690227031708,
+      "learning_rate": 9.855161977544672e-06,
+      "loss": 0.4332,
+      "step": 59
+    },
+    {
+      "epoch": 0.18518518518518517,
+      "grad_norm": 0.06865860521793365,
+      "learning_rate": 9.849220092118721e-06,
+      "loss": 0.3464,
+      "step": 60
+    },
+    {
+      "epoch": 0.1882716049382716,
+      "grad_norm": 0.0760008841753006,
+      "learning_rate": 9.84316062778912e-06,
+      "loss": 0.3808,
+      "step": 61
+    },
+    {
+      "epoch": 0.19135802469135801,
+      "grad_norm": 0.07834326475858688,
+      "learning_rate": 9.836983731479526e-06,
+      "loss": 0.499,
+      "step": 62
+    },
+    {
+      "epoch": 0.19444444444444445,
+      "grad_norm": 0.08240173012018204,
+      "learning_rate": 9.830689552960974e-06,
+      "loss": 0.4432,
+      "step": 63
+    },
+    {
+      "epoch": 0.19753086419753085,
+      "grad_norm": 0.06976404786109924,
+      "learning_rate": 9.824278244848236e-06,
+      "loss": 0.3482,
+      "step": 64
+    },
+    {
+      "epoch": 0.2006172839506173,
+      "grad_norm": 0.09335274249315262,
+      "learning_rate": 9.817749962596115e-06,
+      "loss": 0.4533,
+      "step": 65
+    },
+    {
+      "epoch": 0.2037037037037037,
+      "grad_norm": 0.10973995178937912,
+      "learning_rate": 9.811104864495691e-06,
+      "loss": 0.6042,
+      "step": 66
+    },
+    {
+      "epoch": 0.20679012345679013,
+      "grad_norm": 0.08284437656402588,
+      "learning_rate": 9.804343111670472e-06,
+      "loss": 0.4818,
+      "step": 67
+    },
+    {
+      "epoch": 0.20987654320987653,
+      "grad_norm": 0.08448096364736557,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 0.518,
+      "step": 68
+    },
+    {
+      "epoch": 0.21296296296296297,
+      "grad_norm": 0.07667321711778641,
+      "learning_rate": 9.790470300478318e-06,
+      "loss": 0.3757,
+      "step": 69
+    },
+    {
+      "epoch": 0.21604938271604937,
+      "grad_norm": 0.0944654569029808,
+      "learning_rate": 9.783359578485047e-06,
+      "loss": 0.4863,
+      "step": 70
+    },
+    {
+      "epoch": 0.2191358024691358,
+      "grad_norm": 0.07617281377315521,
+      "learning_rate": 9.776132874506153e-06,
+      "loss": 0.3484,
+      "step": 71
+    },
+    {
+      "epoch": 0.2222222222222222,
+      "grad_norm": 0.09038567543029785,
+      "learning_rate": 9.768790363767321e-06,
+      "loss": 0.596,
+      "step": 72
+    },
+    {
+      "epoch": 0.22530864197530864,
+      "grad_norm": 0.0843636766076088,
+      "learning_rate": 9.761332224302209e-06,
+      "loss": 0.4042,
+      "step": 73
+    },
+    {
+      "epoch": 0.22839506172839505,
+      "grad_norm": 0.09003959596157074,
+      "learning_rate": 9.753758636948112e-06,
+      "loss": 0.5011,
+      "step": 74
+    },
+    {
+      "epoch": 0.23148148148148148,
+      "grad_norm": 0.079057976603508,
+      "learning_rate": 9.74606978534159e-06,
+      "loss": 0.4703,
+      "step": 75
+    },
+    {
+      "epoch": 0.2345679012345679,
+      "grad_norm": 0.07765232026576996,
+      "learning_rate": 9.738265855914014e-06,
+      "loss": 0.3294,
+      "step": 76
+    },
+    {
+      "epoch": 0.23765432098765432,
+      "grad_norm": 0.07654544711112976,
+      "learning_rate": 9.730347037887041e-06,
+      "loss": 0.4039,
+      "step": 77
+    },
+    {
+      "epoch": 0.24074074074074073,
+      "grad_norm": 0.05925621837377548,
+      "learning_rate": 9.722313523268028e-06,
+      "loss": 0.2295,
+      "step": 78
+    },
+    {
+      "epoch": 0.24382716049382716,
+      "grad_norm": 0.07830403745174408,
+      "learning_rate": 9.714165506845381e-06,
+      "loss": 0.3721,
+      "step": 79
+    },
+    {
+      "epoch": 0.24691358024691357,
+      "grad_norm": 0.09928114712238312,
+      "learning_rate": 9.705903186183828e-06,
+      "loss": 0.5154,
+      "step": 80
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 0.06352175772190094,
+      "learning_rate": 9.697526761619621e-06,
+      "loss": 0.2613,
+      "step": 81
+    },
+    {
+      "epoch": 0.25,
+      "eval_loss": 0.5444870591163635,
+      "eval_runtime": 44.3715,
+      "eval_samples_per_second": 8.294,
+      "eval_steps_per_second": 1.037,
+      "step": 81
+    },
+    {
+      "epoch": 0.25308641975308643,
+      "grad_norm": 0.07308296114206314,
+      "learning_rate": 9.689036436255698e-06,
+      "loss": 0.3455,
+      "step": 82
+    },
+    {
+      "epoch": 0.25617283950617287,
+      "grad_norm": 0.07788842916488647,
+      "learning_rate": 9.680432415956736e-06,
+      "loss": 0.4675,
+      "step": 83
+    },
+    {
+      "epoch": 0.25925925925925924,
+      "grad_norm": 0.09506388008594513,
+      "learning_rate": 9.671714909344175e-06,
+      "loss": 0.5544,
+      "step": 84
+    },
+    {
+      "epoch": 0.2623456790123457,
+      "grad_norm": 0.08810863643884659,
+      "learning_rate": 9.66288412779115e-06,
+      "loss": 0.497,
+      "step": 85
+    },
+    {
+      "epoch": 0.2654320987654321,
+      "grad_norm": 0.06235141307115555,
+      "learning_rate": 9.653940285417381e-06,
+      "loss": 0.2775,
+      "step": 86
+    },
+    {
+      "epoch": 0.26851851851851855,
+      "grad_norm": 0.07534658908843994,
+      "learning_rate": 9.644883599083959e-06,
+      "loss": 0.3706,
+      "step": 87
+    },
+    {
+      "epoch": 0.2716049382716049,
+      "grad_norm": 0.11235971748828888,
+      "learning_rate": 9.635714288388103e-06,
+      "loss": 0.6166,
+      "step": 88
+    },
+    {
+      "epoch": 0.27469135802469136,
+      "grad_norm": 0.07352706789970398,
+      "learning_rate": 9.626432575657834e-06,
+      "loss": 0.4254,
+      "step": 89
+    },
+    {
+      "epoch": 0.2777777777777778,
+      "grad_norm": 0.10939712822437286,
+      "learning_rate": 9.617038685946578e-06,
+      "loss": 0.3768,
+      "step": 90
+    },
+    {
+      "epoch": 0.2808641975308642,
+      "grad_norm": 0.0766228511929512,
+      "learning_rate": 9.60753284702772e-06,
+      "loss": 0.3562,
+      "step": 91
+    },
+    {
+      "epoch": 0.2839506172839506,
+      "grad_norm": 0.08354140818119049,
+      "learning_rate": 9.597915289389067e-06,
+      "loss": 0.4783,
+      "step": 92
+    },
+    {
+      "epoch": 0.28703703703703703,
+      "grad_norm": 0.08200543373823166,
+      "learning_rate": 9.58818624622727e-06,
+      "loss": 0.3947,
+      "step": 93
+    },
+    {
+      "epoch": 0.29012345679012347,
+      "grad_norm": 0.08410683274269104,
+      "learning_rate": 9.578345953442163e-06,
+      "loss": 0.5048,
+      "step": 94
+    },
+    {
+      "epoch": 0.2932098765432099,
+      "grad_norm": 0.1019473522901535,
+      "learning_rate": 9.568394649631055e-06,
+      "loss": 0.5842,
+      "step": 95
+    },
+    {
+      "epoch": 0.2962962962962963,
+      "grad_norm": 0.08855041116476059,
+      "learning_rate": 9.558332576082925e-06,
+      "loss": 0.4176,
+      "step": 96
+    },
+    {
+      "epoch": 0.2993827160493827,
+      "grad_norm": 0.08165948837995529,
+      "learning_rate": 9.548159976772593e-06,
+      "loss": 0.4098,
+      "step": 97
+    },
+    {
+      "epoch": 0.30246913580246915,
+      "grad_norm": 0.07580746710300446,
+      "learning_rate": 9.537877098354787e-06,
+      "loss": 0.3886,
+      "step": 98
+    },
+    {
+      "epoch": 0.3055555555555556,
+      "grad_norm": 0.0938824713230133,
+      "learning_rate": 9.527484190158171e-06,
+      "loss": 0.4551,
+      "step": 99
+    },
+    {
+      "epoch": 0.30864197530864196,
+      "grad_norm": 0.07878723740577698,
+      "learning_rate": 9.5169815041793e-06,
+      "loss": 0.4042,
+      "step": 100
+    },
+    {
+      "epoch": 0.3117283950617284,
+      "grad_norm": 0.07207982987165451,
+      "learning_rate": 9.506369295076505e-06,
+      "loss": 0.3541,
+      "step": 101
+    },
+    {
+      "epoch": 0.3148148148148148,
+      "grad_norm": 0.06538520753383636,
+      "learning_rate": 9.495647820163725e-06,
+      "loss": 0.2972,
+      "step": 102
+    },
+    {
+      "epoch": 0.31790123456790126,
+      "grad_norm": 0.08196717500686646,
+      "learning_rate": 9.484817339404261e-06,
+      "loss": 0.401,
+      "step": 103
+    },
+    {
+      "epoch": 0.32098765432098764,
+      "grad_norm": 0.07677263766527176,
+      "learning_rate": 9.473878115404477e-06,
+      "loss": 0.4073,
+      "step": 104
+    },
+    {
+      "epoch": 0.32407407407407407,
+      "grad_norm": 0.11730651557445526,
+      "learning_rate": 9.462830413407427e-06,
+      "loss": 0.4501,
+      "step": 105
+    },
+    {
+      "epoch": 0.3271604938271605,
+      "grad_norm": 0.06849709898233414,
+      "learning_rate": 9.451674501286436e-06,
+      "loss": 0.2538,
+      "step": 106
+    },
+    {
+      "epoch": 0.33024691358024694,
+      "grad_norm": 0.09413019567728043,
+      "learning_rate": 9.440410649538592e-06,
+      "loss": 0.4646,
+      "step": 107
+    },
+    {
+      "epoch": 0.3333333333333333,
+      "grad_norm": 0.15361227095127106,
+      "learning_rate": 9.42903913127819e-06,
+      "loss": 0.5303,
+      "step": 108
+    },
+    {
+      "epoch": 0.33641975308641975,
+      "grad_norm": 0.08900155127048492,
+      "learning_rate": 9.417560222230115e-06,
+      "loss": 0.383,
+      "step": 109
+    },
+    {
+      "epoch": 0.3395061728395062,
+      "grad_norm": 0.07807417958974838,
+      "learning_rate": 9.405974200723156e-06,
+      "loss": 0.3673,
+      "step": 110
+    },
+    {
+      "epoch": 0.3425925925925926,
+      "grad_norm": 0.1323561668395996,
+      "learning_rate": 9.394281347683247e-06,
+      "loss": 0.597,
+      "step": 111
+    },
+    {
+      "epoch": 0.345679012345679,
+      "grad_norm": 0.11236107349395752,
+      "learning_rate": 9.382481946626673e-06,
+      "loss": 0.5051,
+      "step": 112
+    },
+    {
+      "epoch": 0.3487654320987654,
+      "grad_norm": 0.09908317029476166,
+      "learning_rate": 9.370576283653178e-06,
+      "loss": 0.3208,
+      "step": 113
+    },
+    {
+      "epoch": 0.35185185185185186,
+      "grad_norm": 0.08509659022092819,
+      "learning_rate": 9.358564647439037e-06,
+      "loss": 0.3801,
+      "step": 114
+    },
+    {
+      "epoch": 0.3549382716049383,
+      "grad_norm": 0.05896300822496414,
+      "learning_rate": 9.34644732923006e-06,
+      "loss": 0.2217,
+      "step": 115
+    },
+    {
+      "epoch": 0.35802469135802467,
+      "grad_norm": 0.06763949990272522,
+      "learning_rate": 9.33422462283452e-06,
+      "loss": 0.3583,
+      "step": 116
+    },
+    {
+      "epoch": 0.3611111111111111,
+      "grad_norm": 0.0857081338763237,
+      "learning_rate": 9.321896824616036e-06,
+      "loss": 0.4122,
+      "step": 117
+    },
+    {
+      "epoch": 0.36419753086419754,
+      "grad_norm": 0.07149571180343628,
+      "learning_rate": 9.309464233486386e-06,
+      "loss": 0.2959,
+      "step": 118
+    },
+    {
+      "epoch": 0.36728395061728397,
+      "grad_norm": 0.09094710648059845,
+      "learning_rate": 9.29692715089826e-06,
+      "loss": 0.3633,
+      "step": 119
+    },
+    {
+      "epoch": 0.37037037037037035,
+      "grad_norm": 0.07034748792648315,
+      "learning_rate": 9.284285880837947e-06,
+      "loss": 0.2826,
+      "step": 120
+    },
+    {
+      "epoch": 0.3734567901234568,
+      "grad_norm": 0.0919278934597969,
+      "learning_rate": 9.271540729817969e-06,
+      "loss": 0.389,
+      "step": 121
+    },
+    {
+      "epoch": 0.3765432098765432,
+      "grad_norm": 0.07186863571405411,
+      "learning_rate": 9.258692006869644e-06,
+      "loss": 0.296,
+      "step": 122
+    },
+    {
+      "epoch": 0.37962962962962965,
+      "grad_norm": 0.09665773808956146,
+      "learning_rate": 9.245740023535596e-06,
+      "loss": 0.4324,
+      "step": 123
+    },
+    {
+      "epoch": 0.38271604938271603,
+      "grad_norm": 0.08115452527999878,
+      "learning_rate": 9.232685093862206e-06,
+      "loss": 0.3555,
+      "step": 124
+    },
+    {
+      "epoch": 0.38580246913580246,
+      "grad_norm": 0.07702954113483429,
+      "learning_rate": 9.219527534391983e-06,
+      "loss": 0.3385,
+      "step": 125
+    },
+    {
+      "epoch": 0.3888888888888889,
+      "grad_norm": 0.10876493901014328,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 0.4446,
+      "step": 126
+    },
+    {
+      "epoch": 0.39197530864197533,
+      "grad_norm": 0.07764764875173569,
+      "learning_rate": 9.192905804665677e-06,
+      "loss": 0.369,
+      "step": 127
+    },
+    {
+      "epoch": 0.3950617283950617,
+      "grad_norm": 0.10887006670236588,
+      "learning_rate": 9.179442279905927e-06,
+      "loss": 0.4297,
+      "step": 128
+    },
+    {
+      "epoch": 0.39814814814814814,
+      "grad_norm": 0.10183979570865631,
+      "learning_rate": 9.165877416326365e-06,
+      "loss": 0.5906,
+      "step": 129
+    },
+    {
+      "epoch": 0.4012345679012346,
+      "grad_norm": 0.07278673350811005,
+      "learning_rate": 9.152211542833856e-06,
+      "loss": 0.3017,
+      "step": 130
+    },
+    {
+      "epoch": 0.404320987654321,
+      "grad_norm": 0.08892305195331573,
+      "learning_rate": 9.138444990784455e-06,
+      "loss": 0.3919,
+      "step": 131
+    },
+    {
+      "epoch": 0.4074074074074074,
+      "grad_norm": 0.0926053375005722,
+      "learning_rate": 9.124578093975358e-06,
+      "loss": 0.4833,
+      "step": 132
+    },
+    {
+      "epoch": 0.4104938271604938,
+      "grad_norm": 0.1312541514635086,
+      "learning_rate": 9.110611188636828e-06,
+      "loss": 0.4139,
+      "step": 133
+    },
+    {
+      "epoch": 0.41358024691358025,
+      "grad_norm": 0.07399484515190125,
+      "learning_rate": 9.096544613424026e-06,
+      "loss": 0.3156,
+      "step": 134
+    },
+    {
+      "epoch": 0.4166666666666667,
+      "grad_norm": 0.0757204219698906,
+      "learning_rate": 9.082378709408805e-06,
+      "loss": 0.3355,
+      "step": 135
+    },
+    {
+      "epoch": 0.41975308641975306,
+      "grad_norm": 0.08242496103048325,
+      "learning_rate": 9.068113820071447e-06,
+      "loss": 0.3647,
+      "step": 136
+    },
+    {
+      "epoch": 0.4228395061728395,
+      "grad_norm": 0.08191465586423874,
+      "learning_rate": 9.053750291292321e-06,
+      "loss": 0.3801,
+      "step": 137
+    },
+    {
+      "epoch": 0.42592592592592593,
+      "grad_norm": 0.08579788357019424,
+      "learning_rate": 9.039288471343505e-06,
+      "loss": 0.4375,
+      "step": 138
+    },
+    {
+      "epoch": 0.42901234567901236,
+      "grad_norm": 0.09289571642875671,
+      "learning_rate": 9.024728710880345e-06,
+      "loss": 0.3733,
+      "step": 139
+    },
+    {
+      "epoch": 0.43209876543209874,
+      "grad_norm": 0.09474348276853561,
+      "learning_rate": 9.010071362932945e-06,
+      "loss": 0.5004,
+      "step": 140
+    },
+    {
+      "epoch": 0.4351851851851852,
+      "grad_norm": 0.09607541561126709,
+      "learning_rate": 8.995316782897605e-06,
+      "loss": 0.3496,
+      "step": 141
+    },
+    {
+      "epoch": 0.4382716049382716,
+      "grad_norm": 0.08354438096284866,
+      "learning_rate": 8.98046532852822e-06,
+      "loss": 0.3528,
+      "step": 142
+    },
+    {
+      "epoch": 0.44135802469135804,
+      "grad_norm": 0.08367566019296646,
+      "learning_rate": 8.965517359927583e-06,
+      "loss": 0.3365,
+      "step": 143
+    },
+    {
+      "epoch": 0.4444444444444444,
+      "grad_norm": 0.08424922823905945,
+      "learning_rate": 8.950473239538672e-06,
+      "loss": 0.3636,
+      "step": 144
+    },
+    {
+      "epoch": 0.44753086419753085,
+      "grad_norm": 0.07770823687314987,
+      "learning_rate": 8.935333332135853e-06,
+      "loss": 0.2757,
+      "step": 145
+    },
+    {
+      "epoch": 0.4506172839506173,
+      "grad_norm": 0.08803431689739227,
+      "learning_rate": 8.920098004816035e-06,
+      "loss": 0.3397,
+      "step": 146
+    },
+    {
+      "epoch": 0.4537037037037037,
+      "grad_norm": 0.11619243025779724,
+      "learning_rate": 8.904767626989774e-06,
+      "loss": 0.4058,
+      "step": 147
+    },
+    {
+      "epoch": 0.4567901234567901,
+      "grad_norm": 0.08595902472734451,
+      "learning_rate": 8.88934257037231e-06,
+      "loss": 0.3447,
+      "step": 148
+    },
+    {
+      "epoch": 0.45987654320987653,
+      "grad_norm": 0.08116041868925095,
+      "learning_rate": 8.873823208974557e-06,
+      "loss": 0.3578,
+      "step": 149
+    },
+    {
+      "epoch": 0.46296296296296297,
+      "grad_norm": 0.13053898513317108,
+      "learning_rate": 8.85820991909404e-06,
+      "loss": 0.5429,
+      "step": 150
+    },
+    {
+      "epoch": 0.4660493827160494,
+      "grad_norm": 0.08137528598308563,
+      "learning_rate": 8.842503079305757e-06,
+      "loss": 0.3078,
+      "step": 151
+    },
+    {
+      "epoch": 0.4691358024691358,
+      "grad_norm": 0.0843534767627716,
+      "learning_rate": 8.826703070453014e-06,
+      "loss": 0.3807,
+      "step": 152
+    },
+    {
+      "epoch": 0.4722222222222222,
+      "grad_norm": 0.13925758004188538,
+      "learning_rate": 8.810810275638183e-06,
+      "loss": 0.4771,
+      "step": 153
+    },
+    {
+      "epoch": 0.47530864197530864,
+      "grad_norm": 0.08117470145225525,
+      "learning_rate": 8.794825080213415e-06,
+      "loss": 0.3197,
+      "step": 154
+    },
+    {
+      "epoch": 0.4783950617283951,
+      "grad_norm": 0.07650022953748703,
+      "learning_rate": 8.778747871771293e-06,
+      "loss": 0.2993,
+      "step": 155
+    },
+    {
+      "epoch": 0.48148148148148145,
+      "grad_norm": 0.09445349872112274,
+      "learning_rate": 8.76257904013544e-06,
+      "loss": 0.3641,
+      "step": 156
+    },
+    {
+      "epoch": 0.4845679012345679,
+      "grad_norm": 0.097043976187706,
+      "learning_rate": 8.746318977351066e-06,
+      "loss": 0.4181,
+      "step": 157
+    },
+    {
+      "epoch": 0.4876543209876543,
+      "grad_norm": 0.1167394146323204,
+      "learning_rate": 8.729968077675454e-06,
+      "loss": 0.5277,
+      "step": 158
+    },
+    {
+      "epoch": 0.49074074074074076,
+      "grad_norm": 0.08402277529239655,
+      "learning_rate": 8.713526737568415e-06,
+      "loss": 0.2867,
+      "step": 159
+    },
+    {
+      "epoch": 0.49382716049382713,
+      "grad_norm": 0.09060430526733398,
+      "learning_rate": 8.696995355682656e-06,
+      "loss": 0.3219,
+      "step": 160
+    },
+    {
+      "epoch": 0.49691358024691357,
+      "grad_norm": 0.1259710192680359,
+      "learning_rate": 8.680374332854134e-06,
+      "loss": 0.5394,
+      "step": 161
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.09654678404331207,
+      "learning_rate": 8.663664072092324e-06,
+      "loss": 0.3679,
+      "step": 162
+    },
+    {
+      "epoch": 0.5,
+      "eval_loss": 0.5044411420822144,
+      "eval_runtime": 44.4479,
+      "eval_samples_per_second": 8.279,
+      "eval_steps_per_second": 1.035,
+      "step": 162
+    },
+    {
+      "epoch": 0.5030864197530864,
+      "grad_norm": 0.13062100112438202,
+      "learning_rate": 8.646864978570445e-06,
+      "loss": 0.38,
+      "step": 163
+    },
+    {
+      "epoch": 0.5061728395061729,
+      "grad_norm": 0.11305861920118332,
+      "learning_rate": 8.629977459615655e-06,
+      "loss": 0.3435,
+      "step": 164
+    },
+    {
+      "epoch": 0.5092592592592593,
+      "grad_norm": 0.07454624772071838,
+      "learning_rate": 8.613001924699146e-06,
+      "loss": 0.2768,
+      "step": 165
+    },
+    {
+      "epoch": 0.5123456790123457,
+      "grad_norm": 0.08615926653146744,
+      "learning_rate": 8.595938785426241e-06,
+      "loss": 0.3404,
+      "step": 166
+    },
+    {
+      "epoch": 0.5154320987654321,
+      "grad_norm": 0.09183604270219803,
+      "learning_rate": 8.578788455526398e-06,
+      "loss": 0.3493,
+      "step": 167
+    },
+    {
+      "epoch": 0.5185185185185185,
+      "grad_norm": 0.08047281205654144,
+      "learning_rate": 8.561551350843185e-06,
+      "loss": 0.3271,
+      "step": 168
+    },
+    {
+      "epoch": 0.5216049382716049,
+      "grad_norm": 0.08007708936929703,
+      "learning_rate": 8.544227889324199e-06,
+      "loss": 0.2844,
+      "step": 169
+    },
+    {
+      "epoch": 0.5246913580246914,
+      "grad_norm": 0.08152032643556595,
+      "learning_rate": 8.526818491010922e-06,
+      "loss": 0.3033,
+      "step": 170
+    },
+    {
+      "epoch": 0.5277777777777778,
+      "grad_norm": 0.10703514516353607,
+      "learning_rate": 8.509323578028547e-06,
+      "loss": 0.4296,
+      "step": 171
+    },
+    {
+      "epoch": 0.5308641975308642,
+      "grad_norm": 0.07901628315448761,
+      "learning_rate": 8.491743574575743e-06,
+      "loss": 0.29,
+      "step": 172
+    },
+    {
+      "epoch": 0.5339506172839507,
+      "grad_norm": 0.09099699556827545,
+      "learning_rate": 8.474078906914359e-06,
+      "loss": 0.3021,
+      "step": 173
+    },
+    {
+      "epoch": 0.5370370370370371,
+      "grad_norm": 0.0866774320602417,
+      "learning_rate": 8.456330003359093e-06,
+      "loss": 0.2633,
+      "step": 174
+    },
+    {
+      "epoch": 0.5401234567901234,
+      "grad_norm": 0.10114055871963501,
+      "learning_rate": 8.438497294267117e-06,
+      "loss": 0.3735,
+      "step": 175
+    },
+    {
+      "epoch": 0.5432098765432098,
+      "grad_norm": 0.1260298639535904,
+      "learning_rate": 8.420581212027625e-06,
+      "loss": 0.4687,
+      "step": 176
+    },
+    {
+      "epoch": 0.5462962962962963,
+      "grad_norm": 0.1004004031419754,
+      "learning_rate": 8.402582191051365e-06,
+      "loss": 0.29,
+      "step": 177
+    },
+    {
+      "epoch": 0.5493827160493827,
+      "grad_norm": 0.08794572949409485,
+      "learning_rate": 8.38450066776009e-06,
+      "loss": 0.3589,
+      "step": 178
+    },
+    {
+      "epoch": 0.5524691358024691,
+      "grad_norm": 0.10174311697483063,
+      "learning_rate": 8.36633708057599e-06,
+      "loss": 0.3832,
+      "step": 179
+    },
+    {
+      "epoch": 0.5555555555555556,
+      "grad_norm": 0.11463697254657745,
+      "learning_rate": 8.348091869911054e-06,
+      "loss": 0.4172,
+      "step": 180
+    },
+    {
+      "epoch": 0.558641975308642,
+      "grad_norm": 0.11808864772319794,
+      "learning_rate": 8.329765478156394e-06,
+      "loss": 0.494,
+      "step": 181
+    },
+    {
+      "epoch": 0.5617283950617284,
+      "grad_norm": 0.11152324080467224,
+      "learning_rate": 8.311358349671516e-06,
+      "loss": 0.3973,
+      "step": 182
+    },
+    {
+      "epoch": 0.5648148148148148,
+      "grad_norm": 0.09295979887247086,
+      "learning_rate": 8.292870930773551e-06,
+      "loss": 0.3696,
+      "step": 183
+    },
+    {
+      "epoch": 0.5679012345679012,
+      "grad_norm": 0.10292661935091019,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 0.3408,
+      "step": 184
+    },
+    {
+      "epoch": 0.5709876543209876,
+      "grad_norm": 0.10190277546644211,
+      "learning_rate": 8.255657016729997e-06,
+      "loss": 0.3513,
+      "step": 185
+    },
+    {
+      "epoch": 0.5740740740740741,
+      "grad_norm": 0.08307984471321106,
+      "learning_rate": 8.23693142390914e-06,
+      "loss": 0.2577,
+      "step": 186
+    },
+    {
+      "epoch": 0.5771604938271605,
+      "grad_norm": 0.11023180931806564,
+      "learning_rate": 8.218127345302775e-06,
+      "loss": 0.4168,
+      "step": 187
+    },
+    {
+      "epoch": 0.5802469135802469,
+      "grad_norm": 0.10529080033302307,
+      "learning_rate": 8.199245236852871e-06,
+      "loss": 0.4223,
+      "step": 188
+    },
+    {
+      "epoch": 0.5833333333333334,
+      "grad_norm": 0.14696502685546875,
+      "learning_rate": 8.180285556393384e-06,
+      "loss": 0.5283,
+      "step": 189
+    },
+    {
+      "epoch": 0.5864197530864198,
+      "grad_norm": 0.15351015329360962,
+      "learning_rate": 8.161248763639154e-06,
+      "loss": 0.5173,
+      "step": 190
+    },
+    {
+      "epoch": 0.5895061728395061,
+      "grad_norm": 0.10003789514303207,
+      "learning_rate": 8.142135320174758e-06,
+      "loss": 0.3617,
+      "step": 191
+    },
+    {
+      "epoch": 0.5925925925925926,
+      "grad_norm": 0.09017117321491241,
+      "learning_rate": 8.122945689443328e-06,
+      "loss": 0.2601,
+      "step": 192
+    },
+    {
+      "epoch": 0.595679012345679,
+      "grad_norm": 0.11840925365686417,
+      "learning_rate": 8.1036803367353e-06,
+      "loss": 0.4291,
+      "step": 193
+    },
+    {
+      "epoch": 0.5987654320987654,
+      "grad_norm": 0.09116993844509125,
+      "learning_rate": 8.084339729177142e-06,
+      "loss": 0.3061,
+      "step": 194
+    },
+    {
+      "epoch": 0.6018518518518519,
+      "grad_norm": 0.11056546866893768,
+      "learning_rate": 8.064924335720023e-06,
+      "loss": 0.3712,
+      "step": 195
+    },
+    {
+      "epoch": 0.6049382716049383,
+      "grad_norm": 0.10576466470956802,
+      "learning_rate": 8.045434627128446e-06,
+      "loss": 0.3591,
+      "step": 196
+    },
+    {
+      "epoch": 0.6080246913580247,
+      "grad_norm": 0.09751347452402115,
+      "learning_rate": 8.025871075968828e-06,
+      "loss": 0.3268,
+      "step": 197
+    },
+    {
+      "epoch": 0.6111111111111112,
+      "grad_norm": 0.11890437453985214,
+      "learning_rate": 8.006234156598043e-06,
+      "loss": 0.3256,
+      "step": 198
+    },
+    {
+      "epoch": 0.6141975308641975,
+      "grad_norm": 0.12418389320373535,
+      "learning_rate": 7.986524345151924e-06,
+      "loss": 0.5357,
+      "step": 199
+    },
+    {
+      "epoch": 0.6172839506172839,
+      "grad_norm": 0.11261377483606339,
+      "learning_rate": 7.966742119533724e-06,
+      "loss": 0.4537,
+      "step": 200
+    },
+    {
+      "epoch": 0.6203703703703703,
+      "grad_norm": 0.12626801431179047,
+      "learning_rate": 7.946887959402504e-06,
+      "loss": 0.3786,
+      "step": 201
+    },
+    {
+      "epoch": 0.6234567901234568,
+      "grad_norm": 0.12130914628505707,
+      "learning_rate": 7.926962346161535e-06,
+      "loss": 0.4564,
+      "step": 202
+    },
+    {
+      "epoch": 0.6265432098765432,
+      "grad_norm": 0.10559491068124771,
+      "learning_rate": 7.9069657629466e-06,
+      "loss": 0.3984,
+      "step": 203
+    },
+    {
+      "epoch": 0.6296296296296297,
+      "grad_norm": 0.11549825966358185,
+      "learning_rate": 7.886898694614292e-06,
+      "loss": 0.4251,
+      "step": 204
+    },
+    {
+      "epoch": 0.6327160493827161,
+      "grad_norm": 0.10902281850576401,
+      "learning_rate": 7.866761627730253e-06,
+      "loss": 0.4012,
+      "step": 205
+    },
+    {
+      "epoch": 0.6358024691358025,
+      "grad_norm": 0.11586394906044006,
+      "learning_rate": 7.846555050557381e-06,
+      "loss": 0.3586,
+      "step": 206
+    },
+    {
+      "epoch": 0.6388888888888888,
+      "grad_norm": 0.10988422483205795,
+      "learning_rate": 7.826279453043985e-06,
+      "loss": 0.4294,
+      "step": 207
+    },
+    {
+      "epoch": 0.6419753086419753,
+      "grad_norm": 0.1205698624253273,
+      "learning_rate": 7.805935326811913e-06,
+      "loss": 0.4782,
+      "step": 208
+    },
+    {
+      "epoch": 0.6450617283950617,
+      "grad_norm": 0.08950233459472656,
+      "learning_rate": 7.78552316514462e-06,
+      "loss": 0.2901,
+      "step": 209
+    },
+    {
+      "epoch": 0.6481481481481481,
+      "grad_norm": 0.13640360534191132,
+      "learning_rate": 7.765043462975217e-06,
+      "loss": 0.4403,
+      "step": 210
+    },
+    {
+      "epoch": 0.6512345679012346,
+      "grad_norm": 0.13739749789237976,
+      "learning_rate": 7.744496716874472e-06,
+      "loss": 0.472,
+      "step": 211
+    },
+    {
+      "epoch": 0.654320987654321,
+      "grad_norm": 0.10840674489736557,
+      "learning_rate": 7.723883425038759e-06,
+      "loss": 0.3961,
+      "step": 212
+    },
+    {
+      "epoch": 0.6574074074074074,
+      "grad_norm": 0.11287008225917816,
+      "learning_rate": 7.703204087277989e-06,
+      "loss": 0.4169,
+      "step": 213
+    },
+    {
+      "epoch": 0.6604938271604939,
+      "grad_norm": 0.1013006791472435,
+      "learning_rate": 7.682459205003484e-06,
+      "loss": 0.3537,
+      "step": 214
+    },
+    {
+      "epoch": 0.6635802469135802,
+      "grad_norm": 0.12204479426145554,
+      "learning_rate": 7.661649281215823e-06,
+      "loss": 0.3444,
+      "step": 215
+    },
+    {
+      "epoch": 0.6666666666666666,
+      "grad_norm": 0.1041225790977478,
+      "learning_rate": 7.640774820492647e-06,
+      "loss": 0.3432,
+      "step": 216
+    },
+    {
+      "epoch": 0.6697530864197531,
+      "grad_norm": 0.12317519634962082,
+      "learning_rate": 7.619836328976416e-06,
+      "loss": 0.4119,
+      "step": 217
+    },
+    {
+      "epoch": 0.6728395061728395,
+      "grad_norm": 0.15862716734409332,
+      "learning_rate": 7.598834314362151e-06,
+      "loss": 0.3585,
+      "step": 218
+    },
+    {
+      "epoch": 0.6759259259259259,
+      "grad_norm": 0.10013571381568909,
+      "learning_rate": 7.57776928588511e-06,
+      "loss": 0.3589,
+      "step": 219
+    },
+    {
+      "epoch": 0.6790123456790124,
+      "grad_norm": 0.11820396035909653,
+      "learning_rate": 7.556641754308447e-06,
+      "loss": 0.2838,
+      "step": 220
+    },
+    {
+      "epoch": 0.6820987654320988,
+      "grad_norm": 0.08206115663051605,
+      "learning_rate": 7.535452231910829e-06,
+      "loss": 0.1639,
+      "step": 221
+    },
+    {
+      "epoch": 0.6851851851851852,
+      "grad_norm": 0.13305512070655823,
+      "learning_rate": 7.514201232474012e-06,
+      "loss": 0.3923,
+      "step": 222
+    },
+    {
+      "epoch": 0.6882716049382716,
+      "grad_norm": 0.1208796426653862,
+      "learning_rate": 7.492889271270382e-06,
+      "loss": 0.3698,
+      "step": 223
+    },
+    {
+      "epoch": 0.691358024691358,
+      "grad_norm": 0.11946754902601242,
+      "learning_rate": 7.471516865050468e-06,
+      "loss": 0.3797,
+      "step": 224
+    },
+    {
+      "epoch": 0.6944444444444444,
+      "grad_norm": 0.08816403150558472,
+      "learning_rate": 7.450084532030402e-06,
+      "loss": 0.2238,
+      "step": 225
+    },
+    {
+      "epoch": 0.6975308641975309,
+      "grad_norm": 0.12045780569314957,
+      "learning_rate": 7.428592791879361e-06,
+      "loss": 0.3699,
+      "step": 226
+    },
+    {
+      "epoch": 0.7006172839506173,
+      "grad_norm": 0.11096329241991043,
+      "learning_rate": 7.407042165706969e-06,
+      "loss": 0.362,
+      "step": 227
+    },
+    {
+      "epoch": 0.7037037037037037,
+      "grad_norm": 0.14540982246398926,
+      "learning_rate": 7.385433176050654e-06,
+      "loss": 0.4543,
+      "step": 228
+    },
+    {
+      "epoch": 0.7067901234567902,
+      "grad_norm": 0.11663732677698135,
+      "learning_rate": 7.36376634686298e-06,
+      "loss": 0.4606,
+      "step": 229
+    },
+    {
+      "epoch": 0.7098765432098766,
+      "grad_norm": 0.11102988570928574,
+      "learning_rate": 7.342042203498952e-06,
+      "loss": 0.3526,
+      "step": 230
+    },
+    {
+      "epoch": 0.7129629629629629,
+      "grad_norm": 0.11012902110815048,
+      "learning_rate": 7.320261272703259e-06,
+      "loss": 0.4337,
+      "step": 231
+    },
+    {
+      "epoch": 0.7160493827160493,
+      "grad_norm": 0.09911687672138214,
+      "learning_rate": 7.298424082597526e-06,
+      "loss": 0.2504,
+      "step": 232
+    },
+    {
+      "epoch": 0.7191358024691358,
+      "grad_norm": 0.13727596402168274,
+      "learning_rate": 7.276531162667484e-06,
+      "loss": 0.4725,
+      "step": 233
+    },
+    {
+      "epoch": 0.7222222222222222,
+      "grad_norm": 0.10461889952421188,
+      "learning_rate": 7.254583043750152e-06,
+      "loss": 0.3202,
+      "step": 234
+    },
+    {
+      "epoch": 0.7253086419753086,
+      "grad_norm": 0.18260876834392548,
+      "learning_rate": 7.232580258020952e-06,
+      "loss": 0.4248,
+      "step": 235
+    },
+    {
+      "epoch": 0.7283950617283951,
+      "grad_norm": 0.13938364386558533,
+      "learning_rate": 7.210523338980814e-06,
+      "loss": 0.2602,
+      "step": 236
+    },
+    {
+      "epoch": 0.7314814814814815,
+      "grad_norm": 0.11910004913806915,
+      "learning_rate": 7.1884128214432366e-06,
+      "loss": 0.4185,
+      "step": 237
+    },
+    {
+      "epoch": 0.7345679012345679,
+      "grad_norm": 0.10073763877153397,
+      "learning_rate": 7.1662492415213194e-06,
+      "loss": 0.2697,
+      "step": 238
+    },
+    {
+      "epoch": 0.7376543209876543,
+      "grad_norm": 0.11307626962661743,
+      "learning_rate": 7.14403313661476e-06,
+      "loss": 0.4232,
+      "step": 239
+    },
+    {
+      "epoch": 0.7407407407407407,
+      "grad_norm": 0.10806172341108322,
+      "learning_rate": 7.1217650453968335e-06,
+      "loss": 0.2928,
+      "step": 240
+    },
+    {
+      "epoch": 0.7438271604938271,
+      "grad_norm": 0.14010940492153168,
+      "learning_rate": 7.099445507801324e-06,
+      "loss": 0.3915,
+      "step": 241
+    },
+    {
+      "epoch": 0.7469135802469136,
+      "grad_norm": 0.09002690017223358,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 0.2801,
+      "step": 242
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.11942241340875626,
+      "learning_rate": 7.0546542594366605e-06,
+      "loss": 0.4149,
+      "step": 243
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 0.4767835736274719,
+      "eval_runtime": 44.3688,
+      "eval_samples_per_second": 8.294,
+      "eval_steps_per_second": 1.037,
+      "step": 243
+    },
+    {
+      "epoch": 0.7530864197530864,
+      "grad_norm": 0.16698460280895233,
+      "learning_rate": 7.03218363471965e-06,
+      "loss": 0.4605,
+      "step": 244
+    },
+    {
+      "epoch": 0.7561728395061729,
+      "grad_norm": 0.12310118973255157,
+      "learning_rate": 7.0096637357030105e-06,
+      "loss": 0.4328,
+      "step": 245
+    },
+    {
+      "epoch": 0.7592592592592593,
+      "grad_norm": 0.11915367841720581,
+      "learning_rate": 6.987095108426102e-06,
+      "loss": 0.3907,
+      "step": 246
+    },
+    {
+      "epoch": 0.7623456790123457,
+      "grad_norm": 0.1066504493355751,
+      "learning_rate": 6.964478300109796e-06,
+      "loss": 0.3148,
+      "step": 247
+    },
+    {
+      "epoch": 0.7654320987654321,
+      "grad_norm": 0.09711527079343796,
+      "learning_rate": 6.94181385914321e-06,
+      "loss": 0.2736,
+      "step": 248
+    },
+    {
+      "epoch": 0.7685185185185185,
+      "grad_norm": 0.08204776048660278,
+      "learning_rate": 6.91910233507041e-06,
+      "loss": 0.1607,
+      "step": 249
+    },
+    {
+      "epoch": 0.7716049382716049,
+      "grad_norm": 0.13877205550670624,
+      "learning_rate": 6.896344278577083e-06,
+      "loss": 0.3763,
+      "step": 250
+    },
+    {
+      "epoch": 0.7746913580246914,
+      "grad_norm": 0.11828643828630447,
+      "learning_rate": 6.873540241477189e-06,
+      "loss": 0.4063,
+      "step": 251
+    },
+    {
+      "epoch": 0.7777777777777778,
+      "grad_norm": 0.13950656354427338,
+      "learning_rate": 6.850690776699574e-06,
+      "loss": 0.4348,
+      "step": 252
+    },
+    {
+      "epoch": 0.7808641975308642,
+      "grad_norm": 0.13861550390720367,
+      "learning_rate": 6.8277964382745675e-06,
+      "loss": 0.4007,
+      "step": 253
+    },
+    {
+      "epoch": 0.7839506172839507,
+      "grad_norm": 0.12502089142799377,
+      "learning_rate": 6.804857781320558e-06,
+      "loss": 0.4157,
+      "step": 254
+    },
+    {
+      "epoch": 0.7870370370370371,
+      "grad_norm": 0.1129172146320343,
+      "learning_rate": 6.781875362030512e-06,
+      "loss": 0.3087,
+      "step": 255
+    },
+    {
+      "epoch": 0.7901234567901234,
+      "grad_norm": 0.18749450147151947,
+      "learning_rate": 6.758849737658508e-06,
+      "loss": 0.381,
+      "step": 256
+    },
+    {
+      "epoch": 0.7932098765432098,
+      "grad_norm": 0.11505936086177826,
+      "learning_rate": 6.735781466506216e-06,
+      "loss": 0.3639,
+      "step": 257
+    },
+    {
+      "epoch": 0.7962962962962963,
+      "grad_norm": 0.13606995344161987,
+      "learning_rate": 6.712671107909359e-06,
+      "loss": 0.4504,
+      "step": 258
+    },
+    {
+      "epoch": 0.7993827160493827,
+      "grad_norm": 0.13360187411308289,
+      "learning_rate": 6.6895192222241534e-06,
+      "loss": 0.4113,
+      "step": 259
+    },
+    {
+      "epoch": 0.8024691358024691,
+      "grad_norm": 0.1227497085928917,
+      "learning_rate": 6.666326370813722e-06,
+      "loss": 0.3156,
+      "step": 260
+    },
+    {
+      "epoch": 0.8055555555555556,
+      "grad_norm": 0.1294088065624237,
+      "learning_rate": 6.643093116034486e-06,
+      "loss": 0.2544,
+      "step": 261
+    },
+    {
+      "epoch": 0.808641975308642,
+      "grad_norm": 0.11842790246009827,
+      "learning_rate": 6.619820021222518e-06,
+      "loss": 0.2796,
+      "step": 262
+    },
+    {
+      "epoch": 0.8117283950617284,
+      "grad_norm": 0.11302869021892548,
+      "learning_rate": 6.5965076506799e-06,
+      "loss": 0.3225,
+      "step": 263
+    },
+    {
+      "epoch": 0.8148148148148148,
+      "grad_norm": 0.1153462752699852,
+      "learning_rate": 6.573156569661026e-06,
+      "loss": 0.3168,
+      "step": 264
+    },
+    {
+      "epoch": 0.8179012345679012,
+      "grad_norm": 0.14865292608737946,
+      "learning_rate": 6.549767344358903e-06,
+      "loss": 0.3793,
+      "step": 265
+    },
+    {
+      "epoch": 0.8209876543209876,
+      "grad_norm": 0.18601423501968384,
+      "learning_rate": 6.526340541891418e-06,
+      "loss": 0.383,
+      "step": 266
+    },
+    {
+      "epoch": 0.8240740740740741,
+      "grad_norm": 0.11983994394540787,
+      "learning_rate": 6.5028767302875974e-06,
+      "loss": 0.3366,
+      "step": 267
+    },
+    {
+      "epoch": 0.8271604938271605,
+      "grad_norm": 0.11204046756029129,
+      "learning_rate": 6.479376478473822e-06,
+      "loss": 0.2842,
+      "step": 268
+    },
+    {
+      "epoch": 0.8302469135802469,
+      "grad_norm": 0.12731367349624634,
+      "learning_rate": 6.455840356260041e-06,
+      "loss": 0.3664,
+      "step": 269
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 0.12762831151485443,
+      "learning_rate": 6.432268934325947e-06,
+      "loss": 0.4333,
+      "step": 270
+    },
+    {
+      "epoch": 0.8364197530864198,
+      "grad_norm": 0.1425330489873886,
+      "learning_rate": 6.408662784207149e-06,
+      "loss": 0.283,
+      "step": 271
+    },
+    {
+      "epoch": 0.8395061728395061,
+      "grad_norm": 0.1323920488357544,
+      "learning_rate": 6.385022478281307e-06,
+      "loss": 0.4108,
+      "step": 272
+    },
+    {
+      "epoch": 0.8425925925925926,
+      "grad_norm": 0.1550484001636505,
+      "learning_rate": 6.361348589754255e-06,
+      "loss": 0.3396,
+      "step": 273
+    },
+    {
+      "epoch": 0.845679012345679,
+      "grad_norm": 0.09628990292549133,
+      "learning_rate": 6.337641692646106e-06,
+      "loss": 0.246,
+      "step": 274
+    },
+    {
+      "epoch": 0.8487654320987654,
+      "grad_norm": 0.1477012187242508,
+      "learning_rate": 6.313902361777327e-06,
+      "loss": 0.4705,
+      "step": 275
+    },
+    {
+      "epoch": 0.8518518518518519,
+      "grad_norm": 0.14865955710411072,
+      "learning_rate": 6.290131172754811e-06,
+      "loss": 0.417,
+      "step": 276
+    },
+    {
+      "epoch": 0.8549382716049383,
+      "grad_norm": 0.11468877643346786,
+      "learning_rate": 6.266328701957911e-06,
+      "loss": 0.3683,
+      "step": 277
+    },
+    {
+      "epoch": 0.8580246913580247,
+      "grad_norm": 0.1273777186870575,
+      "learning_rate": 6.24249552652447e-06,
+      "loss": 0.2808,
+      "step": 278
+    },
+    {
+      "epoch": 0.8611111111111112,
+      "grad_norm": 0.10113878548145294,
+      "learning_rate": 6.2186322243368236e-06,
+      "loss": 0.3368,
+      "step": 279
+    },
+    {
+      "epoch": 0.8641975308641975,
+      "grad_norm": 0.1183820515871048,
+      "learning_rate": 6.194739374007792e-06,
+      "loss": 0.3095,
+      "step": 280
+    },
+    {
+      "epoch": 0.8672839506172839,
+      "grad_norm": 0.12614701688289642,
+      "learning_rate": 6.170817554866646e-06,
+      "loss": 0.3772,
+      "step": 281
+    },
+    {
+      "epoch": 0.8703703703703703,
+      "grad_norm": 0.19127966463565826,
+      "learning_rate": 6.1468673469450655e-06,
+      "loss": 0.3179,
+      "step": 282
+    },
+    {
+      "epoch": 0.8734567901234568,
+      "grad_norm": 0.14781445264816284,
+      "learning_rate": 6.122889330963069e-06,
+      "loss": 0.3659,
+      "step": 283
+    },
+    {
+      "epoch": 0.8765432098765432,
+      "grad_norm": 0.1360250860452652,
+      "learning_rate": 6.098884088314938e-06,
+      "loss": 0.4211,
+      "step": 284
+    },
+    {
+      "epoch": 0.8796296296296297,
+      "grad_norm": 0.1149686872959137,
+      "learning_rate": 6.074852201055121e-06,
+      "loss": 0.2571,
+      "step": 285
+    },
+    {
+      "epoch": 0.8827160493827161,
+      "grad_norm": 0.14958076179027557,
+      "learning_rate": 6.050794251884112e-06,
+      "loss": 0.4164,
+      "step": 286
+    },
+    {
+      "epoch": 0.8858024691358025,
+      "grad_norm": 0.12140931189060211,
+      "learning_rate": 6.026710824134331e-06,
+      "loss": 0.2203,
+      "step": 287
+    },
+    {
+      "epoch": 0.8888888888888888,
+      "grad_norm": 0.12924239039421082,
+      "learning_rate": 6.002602501755974e-06,
+      "loss": 0.4255,
+      "step": 288
+    },
+    {
+      "epoch": 0.8919753086419753,
+      "grad_norm": 0.1369277834892273,
+      "learning_rate": 5.978469869302861e-06,
+      "loss": 0.4083,
+      "step": 289
+    },
+    {
+      "epoch": 0.8950617283950617,
+      "grad_norm": 0.13165542483329773,
+      "learning_rate": 5.954313511918252e-06,
+      "loss": 0.3317,
+      "step": 290
+    },
+    {
+      "epoch": 0.8981481481481481,
+      "grad_norm": 0.16248537600040436,
+      "learning_rate": 5.9301340153206685e-06,
+      "loss": 0.4079,
+      "step": 291
+    },
+    {
+      "epoch": 0.9012345679012346,
+      "grad_norm": 0.14584743976593018,
+      "learning_rate": 5.905931965789688e-06,
+      "loss": 0.3508,
+      "step": 292
+    },
+    {
+      "epoch": 0.904320987654321,
+      "grad_norm": 0.15875974297523499,
+      "learning_rate": 5.881707950151725e-06,
+      "loss": 0.3597,
+      "step": 293
+    },
+    {
+      "epoch": 0.9074074074074074,
+      "grad_norm": 0.11724277585744858,
+      "learning_rate": 5.857462555765809e-06,
+      "loss": 0.3152,
+      "step": 294
+    },
+    {
+      "epoch": 0.9104938271604939,
+      "grad_norm": 0.12342196702957153,
+      "learning_rate": 5.8331963705093375e-06,
+      "loss": 0.318,
+      "step": 295
+    },
+    {
+      "epoch": 0.9135802469135802,
+      "grad_norm": 0.12013120949268341,
+      "learning_rate": 5.808909982763825e-06,
+      "loss": 0.3951,
+      "step": 296
+    },
+    {
+      "epoch": 0.9166666666666666,
+      "grad_norm": 0.10280231386423111,
+      "learning_rate": 5.784603981400632e-06,
+      "loss": 0.2725,
+      "step": 297
+    },
+    {
+      "epoch": 0.9197530864197531,
+      "grad_norm": 0.12491166591644287,
+      "learning_rate": 5.760278955766695e-06,
+      "loss": 0.3837,
+      "step": 298
+    },
+    {
+      "epoch": 0.9228395061728395,
+      "grad_norm": 0.11760140210390091,
+      "learning_rate": 5.735935495670229e-06,
+      "loss": 0.2464,
+      "step": 299
+    },
+    {
+      "epoch": 0.9259259259259259,
+      "grad_norm": 0.13774855434894562,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 0.3504,
+      "step": 300
+    },
+    {
+      "epoch": 0.9290123456790124,
+      "grad_norm": 0.09982441365718842,
+      "learning_rate": 5.687195633543151e-06,
+      "loss": 0.2457,
+      "step": 301
+    },
+    {
+      "epoch": 0.9320987654320988,
+      "grad_norm": 0.11534377187490463,
+      "learning_rate": 5.662800413306611e-06,
+      "loss": 0.2951,
+      "step": 302
+    },
+    {
+      "epoch": 0.9351851851851852,
+      "grad_norm": 0.100958451628685,
+      "learning_rate": 5.6383891221670275e-06,
+      "loss": 0.19,
+      "step": 303
+    },
+    {
+      "epoch": 0.9382716049382716,
+      "grad_norm": 0.17198745906352997,
+      "learning_rate": 5.613962352024293e-06,
+      "loss": 0.3832,
+      "step": 304
+    },
+    {
+      "epoch": 0.941358024691358,
+      "grad_norm": 0.16045625507831573,
+      "learning_rate": 5.589520695153618e-06,
+      "loss": 0.4173,
+      "step": 305
+    },
+    {
+      "epoch": 0.9444444444444444,
+      "grad_norm": 0.12690144777297974,
+      "learning_rate": 5.5650647441911706e-06,
+      "loss": 0.3318,
+      "step": 306
+    },
+    {
+      "epoch": 0.9475308641975309,
+      "grad_norm": 0.12933467328548431,
+      "learning_rate": 5.540595092119709e-06,
+      "loss": 0.3169,
+      "step": 307
+    },
+    {
+      "epoch": 0.9506172839506173,
+      "grad_norm": 0.1863582581281662,
+      "learning_rate": 5.516112332254203e-06,
+      "loss": 0.3925,
+      "step": 308
+    },
+    {
+      "epoch": 0.9537037037037037,
+      "grad_norm": 0.15057547390460968,
+      "learning_rate": 5.491617058227443e-06,
+      "loss": 0.4953,
+      "step": 309
+    },
+    {
+      "epoch": 0.9567901234567902,
+      "grad_norm": 0.159704327583313,
+      "learning_rate": 5.46710986397565e-06,
+      "loss": 0.3831,
+      "step": 310
+    },
+    {
+      "epoch": 0.9598765432098766,
+      "grad_norm": 0.0988263189792633,
+      "learning_rate": 5.442591343724081e-06,
+      "loss": 0.1455,
+      "step": 311
+    },
+    {
+      "epoch": 0.9629629629629629,
+      "grad_norm": 0.13106189668178558,
+      "learning_rate": 5.418062091972604e-06,
+      "loss": 0.227,
+      "step": 312
+    },
+    {
+      "epoch": 0.9660493827160493,
+      "grad_norm": 0.17571298778057098,
+      "learning_rate": 5.393522703481303e-06,
+      "loss": 0.4638,
+      "step": 313
+    },
+    {
+      "epoch": 0.9691358024691358,
+      "grad_norm": 0.12073665857315063,
+      "learning_rate": 5.36897377325604e-06,
+      "loss": 0.2587,
+      "step": 314
+    },
+    {
+      "epoch": 0.9722222222222222,
+      "grad_norm": 0.08656695485115051,
+      "learning_rate": 5.344415896534039e-06,
+      "loss": 0.2088,
+      "step": 315
+    },
+    {
+      "epoch": 0.9753086419753086,
+      "grad_norm": 0.1401841789484024,
+      "learning_rate": 5.319849668769449e-06,
+      "loss": 0.3667,
+      "step": 316
+    },
+    {
+      "epoch": 0.9783950617283951,
+      "grad_norm": 0.1650845855474472,
+      "learning_rate": 5.295275685618905e-06,
+      "loss": 0.3667,
+      "step": 317
+    },
+    {
+      "epoch": 0.9814814814814815,
+      "grad_norm": 0.13909409940242767,
+      "learning_rate": 5.270694542927089e-06,
+      "loss": 0.3811,
+      "step": 318
+    },
+    {
+      "epoch": 0.9845679012345679,
+      "grad_norm": 0.11377997696399689,
+      "learning_rate": 5.246106836712277e-06,
+      "loss": 0.2349,
+      "step": 319
+    },
+    {
+      "epoch": 0.9876543209876543,
+      "grad_norm": 0.12037783116102219,
+      "learning_rate": 5.2215131631518945e-06,
+      "loss": 0.2901,
+      "step": 320
+    },
+    {
+      "epoch": 0.9907407407407407,
+      "grad_norm": 0.13020600378513336,
+      "learning_rate": 5.196914118568054e-06,
+      "loss": 0.3427,
+      "step": 321
+    },
+    {
+      "epoch": 0.9938271604938271,
+      "grad_norm": 0.15103194117546082,
+      "learning_rate": 5.1723102994130994e-06,
+      "loss": 0.4012,
+      "step": 322
+    },
+    {
+      "epoch": 0.9969135802469136,
+      "grad_norm": 0.105732262134552,
+      "learning_rate": 5.147702302255143e-06,
+      "loss": 0.175,
+      "step": 323
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.17236697673797607,
+      "learning_rate": 5.123090723763607e-06,
+      "loss": 0.3751,
+      "step": 324
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.4522034823894501,
+      "eval_runtime": 44.5334,
+      "eval_samples_per_second": 8.263,
+      "eval_steps_per_second": 1.033,
+      "step": 324
+    },
+    {
+      "epoch": 1.0030864197530864,
+      "grad_norm": 0.15303292870521545,
+      "learning_rate": 5.098476160694741e-06,
+      "loss": 0.4663,
+      "step": 325
+    },
+    {
+      "epoch": 1.0061728395061729,
+      "grad_norm": 0.10959513485431671,
+      "learning_rate": 5.073859209877167e-06,
+      "loss": 0.2389,
+      "step": 326
+    },
+    {
+      "epoch": 1.0092592592592593,
+      "grad_norm": 0.14050254225730896,
+      "learning_rate": 5.049240468197401e-06,
+      "loss": 0.3591,
+      "step": 327
+    },
+    {
+      "epoch": 1.0123456790123457,
+      "grad_norm": 0.12712690234184265,
+      "learning_rate": 5.0246205325853824e-06,
+      "loss": 0.3452,
+      "step": 328
+    },
+    {
+      "epoch": 1.0154320987654322,
+      "grad_norm": 0.1756986677646637,
+      "learning_rate": 5e-06,
+      "loss": 0.4289,
+      "step": 329
+    },
+    {
+      "epoch": 1.0185185185185186,
+      "grad_norm": 0.14214292168617249,
+      "learning_rate": 4.975379467414621e-06,
+      "loss": 0.3695,
+      "step": 330
+    },
+    {
+      "epoch": 1.0030864197530864,
+      "grad_norm": 0.1542719155550003,
+      "learning_rate": 4.950759531802602e-06,
+      "loss": 0.3824,
+      "step": 331
+    },
+    {
+      "epoch": 1.0061728395061729,
+      "grad_norm": 0.12223492562770844,
+      "learning_rate": 4.926140790122835e-06,
+      "loss": 0.2753,
+      "step": 332
+    },
+    {
+      "epoch": 1.0092592592592593,
+      "grad_norm": 0.12852071225643158,
+      "learning_rate": 4.90152383930526e-06,
+      "loss": 0.2418,
+      "step": 333
+    },
+    {
+      "epoch": 1.0123456790123457,
+      "grad_norm": 0.1099737137556076,
+      "learning_rate": 4.876909276236395e-06,
+      "loss": 0.2964,
+      "step": 334
+    },
+    {
+      "epoch": 1.0154320987654322,
+      "grad_norm": 0.1437702178955078,
+      "learning_rate": 4.852297697744857e-06,
+      "loss": 0.355,
+      "step": 335
+    },
+    {
+      "epoch": 1.0185185185185186,
+      "grad_norm": 0.12063878774642944,
+      "learning_rate": 4.827689700586902e-06,
+      "loss": 0.2879,
+      "step": 336
+    },
+    {
+      "epoch": 1.021604938271605,
+      "grad_norm": 0.19743777811527252,
+      "learning_rate": 4.803085881431949e-06,
+      "loss": 0.3412,
+      "step": 337
+    },
+    {
+      "epoch": 1.0246913580246915,
+      "grad_norm": 0.22067442536354065,
+      "learning_rate": 4.778486836848107e-06,
+      "loss": 0.3051,
+      "step": 338
+    },
+    {
+      "epoch": 1.0277777777777777,
+      "grad_norm": 0.1556781828403473,
+      "learning_rate": 4.7538931632877254e-06,
+      "loss": 0.3369,
+      "step": 339
+    },
+    {
+      "epoch": 1.0308641975308641,
+      "grad_norm": 0.132530078291893,
+      "learning_rate": 4.729305457072913e-06,
+      "loss": 0.3452,
+      "step": 340
+    },
+    {
+      "epoch": 1.0339506172839505,
+      "grad_norm": 0.16023634374141693,
+      "learning_rate": 4.704724314381097e-06,
+      "loss": 0.3887,
+      "step": 341
+    },
+    {
+      "epoch": 1.037037037037037,
+      "grad_norm": 0.14671647548675537,
+      "learning_rate": 4.680150331230552e-06,
+      "loss": 0.3082,
+      "step": 342
+    },
+    {
+      "epoch": 1.0401234567901234,
+      "grad_norm": 0.20157098770141602,
+      "learning_rate": 4.6555841034659625e-06,
+      "loss": 0.5004,
+      "step": 343
+    },
+    {
+      "epoch": 1.0432098765432098,
+      "grad_norm": 0.14635726809501648,
+      "learning_rate": 4.631026226743962e-06,
+      "loss": 0.4104,
+      "step": 344
+    },
+    {
+      "epoch": 1.0462962962962963,
+      "grad_norm": 0.14289334416389465,
+      "learning_rate": 4.606477296518698e-06,
+      "loss": 0.3206,
+      "step": 345
+    },
+    {
+      "epoch": 1.0493827160493827,
+      "grad_norm": 0.14635069668293,
+      "learning_rate": 4.581937908027397e-06,
+      "loss": 0.2957,
+      "step": 346
+    },
+    {
+      "epoch": 1.0524691358024691,
+      "grad_norm": 0.1479678899049759,
+      "learning_rate": 4.55740865627592e-06,
+      "loss": 0.3168,
+      "step": 347
+    },
+    {
+      "epoch": 1.0555555555555556,
+      "grad_norm": 0.12210693210363388,
+      "learning_rate": 4.532890136024351e-06,
+      "loss": 0.2854,
+      "step": 348
+    },
+    {
+      "epoch": 1.058641975308642,
+      "grad_norm": 0.16018199920654297,
+      "learning_rate": 4.508382941772558e-06,
+      "loss": 0.2937,
+      "step": 349
+    },
+    {
+      "epoch": 1.0617283950617284,
+      "grad_norm": 0.14056287705898285,
+      "learning_rate": 4.483887667745798e-06,
+      "loss": 0.3246,
+      "step": 350
+    },
+    {
+      "epoch": 1.0648148148148149,
+      "grad_norm": 0.14486226439476013,
+      "learning_rate": 4.459404907880293e-06,
+      "loss": 0.3133,
+      "step": 351
+    },
+    {
+      "epoch": 1.0679012345679013,
+      "grad_norm": 0.1279231458902359,
+      "learning_rate": 4.434935255808831e-06,
+      "loss": 0.2219,
+      "step": 352
+    },
+    {
+      "epoch": 1.0709876543209877,
+      "grad_norm": 0.16269516944885254,
+      "learning_rate": 4.410479304846385e-06,
+      "loss": 0.3531,
+      "step": 353
+    },
+    {
+      "epoch": 1.074074074074074,
+      "grad_norm": 0.15139630436897278,
+      "learning_rate": 4.386037647975708e-06,
+      "loss": 0.2508,
+      "step": 354
+    },
+    {
+      "epoch": 1.0771604938271604,
+      "grad_norm": 0.15115757286548615,
+      "learning_rate": 4.361610877832974e-06,
+      "loss": 0.3908,
+      "step": 355
+    },
+    {
+      "epoch": 1.0802469135802468,
+      "grad_norm": 0.17080338299274445,
+      "learning_rate": 4.337199586693389e-06,
+      "loss": 0.4233,
+      "step": 356
+    },
+    {
+      "epoch": 1.0833333333333333,
+      "grad_norm": 0.149905264377594,
+      "learning_rate": 4.312804366456851e-06,
+      "loss": 0.3354,
+      "step": 357
+    },
+    {
+      "epoch": 1.0864197530864197,
+      "grad_norm": 0.2038925588130951,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 0.422,
+      "step": 358
+    },
+    {
+      "epoch": 1.0895061728395061,
+      "grad_norm": 0.1319386065006256,
+      "learning_rate": 4.2640645043297715e-06,
+      "loss": 0.2812,
+      "step": 359
+    },
+    {
+      "epoch": 1.0925925925925926,
+      "grad_norm": 0.210116446018219,
+      "learning_rate": 4.239721044233306e-06,
+      "loss": 0.3266,
+      "step": 360
+    },
+    {
+      "epoch": 1.095679012345679,
+      "grad_norm": 0.15533123910427094,
+      "learning_rate": 4.215396018599369e-06,
+      "loss": 0.3106,
+      "step": 361
+    },
+    {
+      "epoch": 1.0987654320987654,
+      "grad_norm": 0.15208472311496735,
+      "learning_rate": 4.191090017236177e-06,
+      "loss": 0.3423,
+      "step": 362
+    },
+    {
+      "epoch": 1.1018518518518519,
+      "grad_norm": 0.12684912979602814,
+      "learning_rate": 4.166803629490664e-06,
+      "loss": 0.2755,
+      "step": 363
+    },
+    {
+      "epoch": 1.1049382716049383,
+      "grad_norm": 0.18555931746959686,
+      "learning_rate": 4.142537444234192e-06,
+      "loss": 0.4007,
+      "step": 364
+    },
+    {
+      "epoch": 1.1080246913580247,
+      "grad_norm": 0.20792073011398315,
+      "learning_rate": 4.118292049848277e-06,
+      "loss": 0.2467,
+      "step": 365
+    },
+    {
+      "epoch": 1.1111111111111112,
+      "grad_norm": 0.13857008516788483,
+      "learning_rate": 4.094068034210313e-06,
+      "loss": 0.3666,
+      "step": 366
+    },
+    {
+      "epoch": 1.1141975308641976,
+      "grad_norm": 0.10900649428367615,
+      "learning_rate": 4.069865984679332e-06,
+      "loss": 0.1954,
+      "step": 367
+    },
+    {
+      "epoch": 1.117283950617284,
+      "grad_norm": 0.13190750777721405,
+      "learning_rate": 4.045686488081748e-06,
+      "loss": 0.309,
+      "step": 368
+    },
+    {
+      "epoch": 1.1203703703703705,
+      "grad_norm": 0.16032575070858002,
+      "learning_rate": 4.021530130697141e-06,
+      "loss": 0.3524,
+      "step": 369
+    },
+    {
+      "epoch": 1.123456790123457,
+      "grad_norm": 0.14147287607192993,
+      "learning_rate": 3.997397498244028e-06,
+      "loss": 0.3088,
+      "step": 370
+    },
+    {
+      "epoch": 1.126543209876543,
+      "grad_norm": 0.1288299709558487,
+      "learning_rate": 3.97328917586567e-06,
+      "loss": 0.3216,
+      "step": 371
+    },
+    {
+      "epoch": 1.1296296296296295,
+      "grad_norm": 0.17235535383224487,
+      "learning_rate": 3.9492057481158905e-06,
+      "loss": 0.3339,
+      "step": 372
+    },
+    {
+      "epoch": 1.132716049382716,
+      "grad_norm": 0.21856486797332764,
+      "learning_rate": 3.92514779894488e-06,
+      "loss": 0.3691,
+      "step": 373
+    },
+    {
+      "epoch": 1.1358024691358024,
+      "grad_norm": 0.188248872756958,
+      "learning_rate": 3.901115911685063e-06,
+      "loss": 0.3879,
+      "step": 374
+    },
+    {
+      "epoch": 1.1388888888888888,
+      "grad_norm": 0.17136438190937042,
+      "learning_rate": 3.877110669036932e-06,
+      "loss": 0.4754,
+      "step": 375
+    },
+    {
+      "epoch": 1.1419753086419753,
+      "grad_norm": 0.14845937490463257,
+      "learning_rate": 3.853132653054936e-06,
+      "loss": 0.4178,
+      "step": 376
+    },
+    {
+      "epoch": 1.1450617283950617,
+      "grad_norm": 0.14598865807056427,
+      "learning_rate": 3.829182445133356e-06,
+      "loss": 0.2653,
+      "step": 377
+    },
+    {
+      "epoch": 1.1481481481481481,
+      "grad_norm": 0.12898695468902588,
+      "learning_rate": 3.8052606259922097e-06,
+      "loss": 0.2613,
+      "step": 378
+    },
+    {
+      "epoch": 1.1512345679012346,
+      "grad_norm": 0.12332043796777725,
+      "learning_rate": 3.7813677756631773e-06,
+      "loss": 0.2803,
+      "step": 379
+    },
+    {
+      "epoch": 1.154320987654321,
+      "grad_norm": 0.1356392502784729,
+      "learning_rate": 3.75750447347553e-06,
+      "loss": 0.4038,
+      "step": 380
+    },
+    {
+      "epoch": 1.1574074074074074,
+      "grad_norm": 0.25393664836883545,
+      "learning_rate": 3.7336712980420897e-06,
+      "loss": 0.5067,
+      "step": 381
+    },
+    {
+      "epoch": 1.1604938271604939,
+      "grad_norm": 0.12110210955142975,
+      "learning_rate": 3.7098688272451893e-06,
+      "loss": 0.2413,
+      "step": 382
+    },
+    {
+      "epoch": 1.1635802469135803,
+      "grad_norm": 0.12632521986961365,
+      "learning_rate": 3.6860976382226747e-06,
+      "loss": 0.2583,
+      "step": 383
+    },
+    {
+      "epoch": 1.1666666666666667,
+      "grad_norm": 0.15142959356307983,
+      "learning_rate": 3.662358307353897e-06,
+      "loss": 0.4542,
+      "step": 384
+    },
+    {
+      "epoch": 1.1697530864197532,
+      "grad_norm": 0.11639465391635895,
+      "learning_rate": 3.638651410245746e-06,
+      "loss": 0.1849,
+      "step": 385
+    },
+    {
+      "epoch": 1.1728395061728394,
+      "grad_norm": 0.14406833052635193,
+      "learning_rate": 3.6149775217186954e-06,
+      "loss": 0.3171,
+      "step": 386
+    },
+    {
+      "epoch": 1.175925925925926,
+      "grad_norm": 0.1374572366476059,
+      "learning_rate": 3.5913372157928515e-06,
+      "loss": 0.2849,
+      "step": 387
+    },
+    {
+      "epoch": 1.1790123456790123,
+      "grad_norm": 0.16935373842716217,
+      "learning_rate": 3.5677310656740537e-06,
+      "loss": 0.3982,
+      "step": 388
+    },
+    {
+      "epoch": 1.1820987654320987,
+      "grad_norm": 0.1098417416214943,
+      "learning_rate": 3.5441596437399596e-06,
+      "loss": 0.2149,
+      "step": 389
+    },
+    {
+      "epoch": 1.1851851851851851,
+      "grad_norm": 0.14076852798461914,
+      "learning_rate": 3.5206235215261785e-06,
+      "loss": 0.2685,
+      "step": 390
+    },
+    {
+      "epoch": 1.1882716049382716,
+      "grad_norm": 0.12600207328796387,
+      "learning_rate": 3.4971232697124046e-06,
+      "loss": 0.2009,
+      "step": 391
+    },
+    {
+      "epoch": 1.191358024691358,
+      "grad_norm": 0.13086476922035217,
+      "learning_rate": 3.4736594581085837e-06,
+      "loss": 0.3062,
+      "step": 392
+    },
+    {
+      "epoch": 1.1944444444444444,
+      "grad_norm": 0.16587767004966736,
+      "learning_rate": 3.4502326556411e-06,
+      "loss": 0.2432,
+      "step": 393
+    },
+    {
+      "epoch": 1.1975308641975309,
+      "grad_norm": 0.13524991273880005,
+      "learning_rate": 3.4268434303389747e-06,
+      "loss": 0.3204,
+      "step": 394
+    },
+    {
+      "epoch": 1.2006172839506173,
+      "grad_norm": 0.15923044085502625,
+      "learning_rate": 3.403492349320101e-06,
+      "loss": 0.36,
+      "step": 395
+    },
+    {
+      "epoch": 1.2037037037037037,
+      "grad_norm": 0.19655781984329224,
+      "learning_rate": 3.380179978777482e-06,
+      "loss": 0.4863,
+      "step": 396
+    },
+    {
+      "epoch": 1.2067901234567902,
+      "grad_norm": 0.13031858205795288,
+      "learning_rate": 3.356906883965516e-06,
+      "loss": 0.2884,
+      "step": 397
+    },
+    {
+      "epoch": 1.2098765432098766,
+      "grad_norm": 0.12421680986881256,
+      "learning_rate": 3.33367362918628e-06,
+      "loss": 0.1891,
+      "step": 398
+    },
+    {
+      "epoch": 1.212962962962963,
+      "grad_norm": 0.15903340280056,
+      "learning_rate": 3.3104807777758487e-06,
+      "loss": 0.4381,
+      "step": 399
+    },
+    {
+      "epoch": 1.2160493827160495,
+      "grad_norm": 0.11143235117197037,
+      "learning_rate": 3.2873288920906436e-06,
+      "loss": 0.2269,
+      "step": 400
+    },
+    {
+      "epoch": 1.2191358024691359,
+      "grad_norm": 0.1427583545446396,
+      "learning_rate": 3.2642185334937853e-06,
+      "loss": 0.3874,
+      "step": 401
+    },
+    {
+      "epoch": 1.2222222222222223,
+      "grad_norm": 0.21431690454483032,
+      "learning_rate": 3.2411502623414925e-06,
+      "loss": 0.4815,
+      "step": 402
+    },
+    {
+      "epoch": 1.2253086419753085,
+      "grad_norm": 0.20369336009025574,
+      "learning_rate": 3.2181246379694886e-06,
+      "loss": 0.429,
+      "step": 403
+    },
+    {
+      "epoch": 1.228395061728395,
+      "grad_norm": 0.21474803984165192,
+      "learning_rate": 3.1951422186794447e-06,
+      "loss": 0.4217,
+      "step": 404
+    },
+    {
+      "epoch": 1.2314814814814814,
+      "grad_norm": 0.1690702587366104,
+      "learning_rate": 3.1722035617254333e-06,
+      "loss": 0.3388,
+      "step": 405
+    },
+    {
+      "epoch": 1.2314814814814814,
+      "eval_loss": 0.4383295774459839,
+      "eval_runtime": 44.45,
+      "eval_samples_per_second": 8.279,
+      "eval_steps_per_second": 1.035,
+      "step": 405
+    },
+    {
+      "epoch": 1.2345679012345678,
+      "grad_norm": 0.13106146454811096,
+      "learning_rate": 3.149309223300428e-06,
+      "loss": 0.2537,
+      "step": 406
+    },
+    {
+      "epoch": 1.2376543209876543,
+      "grad_norm": 0.18745112419128418,
+      "learning_rate": 3.126459758522813e-06,
+      "loss": 0.3825,
+      "step": 407
+    },
+    {
+      "epoch": 1.2407407407407407,
+      "grad_norm": 0.1358872950077057,
+      "learning_rate": 3.103655721422917e-06,
+      "loss": 0.3057,
+      "step": 408
+    },
+    {
+      "epoch": 1.2438271604938271,
+      "grad_norm": 0.15695077180862427,
+      "learning_rate": 3.080897664929592e-06,
+      "loss": 0.412,
+      "step": 409
+    },
+    {
+      "epoch": 1.2469135802469136,
+      "grad_norm": 0.15740308165550232,
+      "learning_rate": 3.0581861408567907e-06,
+      "loss": 0.371,
+      "step": 410
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.17210154235363007,
+      "learning_rate": 3.035521699890206e-06,
+      "loss": 0.4671,
+      "step": 411
+    },
+    {
+      "epoch": 1.2530864197530864,
+      "grad_norm": 0.1564391851425171,
+      "learning_rate": 3.0129048915739013e-06,
+      "loss": 0.397,
+      "step": 412
+    },
+    {
+      "epoch": 1.2561728395061729,
+      "grad_norm": 0.15035340189933777,
+      "learning_rate": 2.9903362642969903e-06,
+      "loss": 0.3696,
+      "step": 413
+    },
+    {
+      "epoch": 1.2592592592592593,
+      "grad_norm": 0.12334346026182175,
+      "learning_rate": 2.967816365280351e-06,
+      "loss": 0.2595,
+      "step": 414
+    },
+    {
+      "epoch": 1.2623456790123457,
+      "grad_norm": 0.159285768866539,
+      "learning_rate": 2.94534574056334e-06,
+      "loss": 0.3444,
+      "step": 415
+    },
+    {
+      "epoch": 1.2654320987654322,
+      "grad_norm": 0.14071713387966156,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 0.264,
+      "step": 416
+    },
+    {
+      "epoch": 1.2685185185185186,
+      "grad_norm": 0.17824961245059967,
+      "learning_rate": 2.9005544921986774e-06,
+      "loss": 0.3823,
+      "step": 417
+    },
+    {
+      "epoch": 1.2716049382716048,
+      "grad_norm": 0.14212675392627716,
+      "learning_rate": 2.8782349546031673e-06,
+      "loss": 0.253,
+      "step": 418
+    },
+    {
+      "epoch": 1.2746913580246915,
+      "grad_norm": 0.21493245661258698,
+      "learning_rate": 2.8559668633852433e-06,
+      "loss": 0.3181,
+      "step": 419
+    },
+    {
+      "epoch": 1.2777777777777777,
+      "grad_norm": 0.14115536212921143,
+      "learning_rate": 2.8337507584786826e-06,
+      "loss": 0.3007,
+      "step": 420
+    },
+    {
+      "epoch": 1.2808641975308643,
+      "grad_norm": 0.16807730495929718,
+      "learning_rate": 2.811587178556764e-06,
+      "loss": 0.271,
+      "step": 421
+    },
+    {
+      "epoch": 1.2839506172839505,
+      "grad_norm": 0.19324727356433868,
+      "learning_rate": 2.789476661019186e-06,
+      "loss": 0.3613,
+      "step": 422
+    },
+    {
+      "epoch": 1.287037037037037,
+      "grad_norm": 0.22242026031017303,
+      "learning_rate": 2.7674197419790493e-06,
+      "loss": 0.3391,
+      "step": 423
+    },
+    {
+      "epoch": 1.2901234567901234,
+      "grad_norm": 0.1270921379327774,
+      "learning_rate": 2.7454169562498503e-06,
+      "loss": 0.2094,
+      "step": 424
+    },
+    {
+      "epoch": 1.2932098765432098,
+      "grad_norm": 0.12505224347114563,
+      "learning_rate": 2.723468837332517e-06,
+      "loss": 0.2807,
+      "step": 425
+    },
+    {
+      "epoch": 1.2962962962962963,
+      "grad_norm": 0.16030734777450562,
+      "learning_rate": 2.7015759174024756e-06,
+      "loss": 0.3266,
+      "step": 426
+    },
+    {
+      "epoch": 1.2993827160493827,
+      "grad_norm": 0.1334860622882843,
+      "learning_rate": 2.6797387272967414e-06,
+      "loss": 0.2262,
+      "step": 427
+    },
+    {
+      "epoch": 1.3024691358024691,
+      "grad_norm": 0.16829054057598114,
+      "learning_rate": 2.65795779650105e-06,
+      "loss": 0.3483,
+      "step": 428
+    },
+    {
+      "epoch": 1.3055555555555556,
+      "grad_norm": 0.16048014163970947,
+      "learning_rate": 2.63623365313702e-06,
+      "loss": 0.3673,
+      "step": 429
+    },
+    {
+      "epoch": 1.308641975308642,
+      "grad_norm": 0.22250574827194214,
+      "learning_rate": 2.614566823949348e-06,
+      "loss": 0.3418,
+      "step": 430
+    },
+    {
+      "epoch": 1.3117283950617284,
+      "grad_norm": 0.13716565072536469,
+      "learning_rate": 2.592957834293033e-06,
+      "loss": 0.2986,
+      "step": 431
+    },
+    {
+      "epoch": 1.3148148148148149,
+      "grad_norm": 0.15584644675254822,
+      "learning_rate": 2.5714072081206407e-06,
+      "loss": 0.3419,
+      "step": 432
+    },
+    {
+      "epoch": 1.3179012345679013,
+      "grad_norm": 0.17043578624725342,
+      "learning_rate": 2.5499154679696014e-06,
+      "loss": 0.3133,
+      "step": 433
+    },
+    {
+      "epoch": 1.3209876543209877,
+      "grad_norm": 0.1307077258825302,
+      "learning_rate": 2.528483134949535e-06,
+      "loss": 0.2484,
+      "step": 434
+    },
+    {
+      "epoch": 1.324074074074074,
+      "grad_norm": 0.19332851469516754,
+      "learning_rate": 2.50711072872962e-06,
+      "loss": 0.338,
+      "step": 435
+    },
+    {
+      "epoch": 1.3271604938271606,
+      "grad_norm": 0.18752485513687134,
+      "learning_rate": 2.4857987675259887e-06,
+      "loss": 0.3693,
+      "step": 436
+    },
+    {
+      "epoch": 1.3302469135802468,
+      "grad_norm": 0.171221524477005,
+      "learning_rate": 2.4645477680891734e-06,
+      "loss": 0.3222,
+      "step": 437
+    },
+    {
+      "epoch": 1.3333333333333333,
+      "grad_norm": 0.2540048062801361,
+      "learning_rate": 2.4433582456915556e-06,
+      "loss": 0.4404,
+      "step": 438
+    },
+    {
+      "epoch": 1.3364197530864197,
+      "grad_norm": 0.13886091113090515,
+      "learning_rate": 2.422230714114891e-06,
+      "loss": 0.3246,
+      "step": 439
+    },
+    {
+      "epoch": 1.3395061728395061,
+      "grad_norm": 0.11673127859830856,
+      "learning_rate": 2.4011656856378513e-06,
+      "loss": 0.1878,
+      "step": 440
+    },
+    {
+      "epoch": 1.3425925925925926,
+      "grad_norm": 0.20191854238510132,
+      "learning_rate": 2.3801636710235836e-06,
+      "loss": 0.2979,
+      "step": 441
+    },
+    {
+      "epoch": 1.345679012345679,
+      "grad_norm": 0.16786165535449982,
+      "learning_rate": 2.3592251795073564e-06,
+      "loss": 0.2931,
+      "step": 442
+    },
+    {
+      "epoch": 1.3487654320987654,
+      "grad_norm": 0.1304280310869217,
+      "learning_rate": 2.338350718784177e-06,
+      "loss": 0.2368,
+      "step": 443
+    },
+    {
+      "epoch": 1.3518518518518519,
+      "grad_norm": 0.14287714660167694,
+      "learning_rate": 2.3175407949965167e-06,
+      "loss": 0.286,
+      "step": 444
+    },
+    {
+      "epoch": 1.3549382716049383,
+      "grad_norm": 0.13601404428482056,
+      "learning_rate": 2.296795912722014e-06,
+      "loss": 0.268,
+      "step": 445
+    },
+    {
+      "epoch": 1.3580246913580247,
+      "grad_norm": 0.1764301061630249,
+      "learning_rate": 2.2761165749612417e-06,
+      "loss": 0.355,
+      "step": 446
+    },
+    {
+      "epoch": 1.3611111111111112,
+      "grad_norm": 0.1622696816921234,
+      "learning_rate": 2.25550328312553e-06,
+      "loss": 0.3438,
+      "step": 447
+    },
+    {
+      "epoch": 1.3641975308641976,
+      "grad_norm": 0.15518330037593842,
+      "learning_rate": 2.2349565370247837e-06,
+      "loss": 0.2844,
+      "step": 448
+    },
+    {
+      "epoch": 1.367283950617284,
+      "grad_norm": 0.13542047142982483,
+      "learning_rate": 2.214476834855382e-06,
+      "loss": 0.324,
+      "step": 449
+    },
+    {
+      "epoch": 1.3703703703703702,
+      "grad_norm": 0.20794177055358887,
+      "learning_rate": 2.1940646731880887e-06,
+      "loss": 0.5443,
+      "step": 450
+    },
+    {
+      "epoch": 1.373456790123457,
+      "grad_norm": 0.1371917873620987,
+      "learning_rate": 2.173720546956015e-06,
+      "loss": 0.3663,
+      "step": 451
+    },
+    {
+      "epoch": 1.376543209876543,
+      "grad_norm": 0.17952483892440796,
+      "learning_rate": 2.1534449494426203e-06,
+      "loss": 0.3209,
+      "step": 452
+    },
+    {
+      "epoch": 1.3796296296296298,
+      "grad_norm": 0.1383998692035675,
+      "learning_rate": 2.1332383722697483e-06,
+      "loss": 0.2407,
+      "step": 453
+    },
+    {
+      "epoch": 1.382716049382716,
+      "grad_norm": 0.17842058837413788,
+      "learning_rate": 2.1131013053857097e-06,
+      "loss": 0.5964,
+      "step": 454
+    },
+    {
+      "epoch": 1.3858024691358024,
+      "grad_norm": 0.13012441992759705,
+      "learning_rate": 2.0930342370534013e-06,
+      "loss": 0.2686,
+      "step": 455
+    },
+    {
+      "epoch": 1.3888888888888888,
+      "grad_norm": 0.1683279275894165,
+      "learning_rate": 2.073037653838466e-06,
+      "loss": 0.4134,
+      "step": 456
+    },
+    {
+      "epoch": 1.3919753086419753,
+      "grad_norm": 0.18860593438148499,
+      "learning_rate": 2.053112040597495e-06,
+      "loss": 0.2766,
+      "step": 457
+    },
+    {
+      "epoch": 1.3950617283950617,
+      "grad_norm": 0.15948981046676636,
+      "learning_rate": 2.0332578804662783e-06,
+      "loss": 0.452,
+      "step": 458
+    },
+    {
+      "epoch": 1.3981481481481481,
+      "grad_norm": 0.13614550232887268,
+      "learning_rate": 2.013475654848076e-06,
+      "loss": 0.3028,
+      "step": 459
+    },
+    {
+      "epoch": 1.4012345679012346,
+      "grad_norm": 0.1575852334499359,
+      "learning_rate": 1.99376584340196e-06,
+      "loss": 0.3772,
+      "step": 460
+    },
+    {
+      "epoch": 1.404320987654321,
+      "grad_norm": 0.1815677434206009,
+      "learning_rate": 1.9741289240311757e-06,
+      "loss": 0.4218,
+      "step": 461
+    },
+    {
+      "epoch": 1.4074074074074074,
+      "grad_norm": 0.16409048438072205,
+      "learning_rate": 1.954565372871554e-06,
+      "loss": 0.4449,
+      "step": 462
+    },
+    {
+      "epoch": 1.4104938271604939,
+      "grad_norm": 0.17997804284095764,
+      "learning_rate": 1.935075664279978e-06,
+      "loss": 0.3908,
+      "step": 463
+    },
+    {
+      "epoch": 1.4135802469135803,
+      "grad_norm": 0.17692823708057404,
+      "learning_rate": 1.9156602708228584e-06,
+      "loss": 0.3506,
+      "step": 464
+    },
+    {
+      "epoch": 1.4166666666666667,
+      "grad_norm": 0.17066018283367157,
+      "learning_rate": 1.8963196632647008e-06,
+      "loss": 0.4187,
+      "step": 465
+    },
+    {
+      "epoch": 1.4197530864197532,
+      "grad_norm": 0.17325402796268463,
+      "learning_rate": 1.8770543105566752e-06,
+      "loss": 0.3865,
+      "step": 466
+    },
+    {
+      "epoch": 1.4228395061728394,
+      "grad_norm": 0.1373230516910553,
+      "learning_rate": 1.8578646798252432e-06,
+      "loss": 0.2194,
+      "step": 467
+    },
+    {
+      "epoch": 1.425925925925926,
+      "grad_norm": 0.14924941956996918,
+      "learning_rate": 1.8387512363608496e-06,
+      "loss": 0.3415,
+      "step": 468
+    },
+    {
+      "epoch": 1.4290123456790123,
+      "grad_norm": 0.15401771664619446,
+      "learning_rate": 1.8197144436066167e-06,
+      "loss": 0.3132,
+      "step": 469
+    },
+    {
+      "epoch": 1.4320987654320987,
+      "grad_norm": 0.24441462755203247,
+      "learning_rate": 1.8007547631471289e-06,
+      "loss": 0.365,
+      "step": 470
+    },
+    {
+      "epoch": 1.4351851851851851,
+      "grad_norm": 0.2641655206680298,
+      "learning_rate": 1.781872654697226e-06,
+      "loss": 0.4653,
+      "step": 471
+    },
+    {
+      "epoch": 1.4382716049382716,
+      "grad_norm": 0.18639406561851501,
+      "learning_rate": 1.7630685760908623e-06,
+      "loss": 0.3422,
+      "step": 472
+    },
+    {
+      "epoch": 1.441358024691358,
+      "grad_norm": 0.14547406136989594,
+      "learning_rate": 1.7443429832700038e-06,
+      "loss": 0.3541,
+      "step": 473
+    },
+    {
+      "epoch": 1.4444444444444444,
+      "grad_norm": 0.179130420088768,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": 0.3341,
+      "step": 474
+    },
+    {
+      "epoch": 1.4475308641975309,
+      "grad_norm": 0.1942981481552124,
+      "learning_rate": 1.7071290692264492e-06,
+      "loss": 0.392,
+      "step": 475
+    },
+    {
+      "epoch": 1.4506172839506173,
+      "grad_norm": 0.10643615573644638,
+      "learning_rate": 1.6886416503284835e-06,
+      "loss": 0.2317,
+      "step": 476
+    },
+    {
+      "epoch": 1.4537037037037037,
+      "grad_norm": 0.14966462552547455,
+      "learning_rate": 1.6702345218436066e-06,
+      "loss": 0.2882,
+      "step": 477
+    },
+    {
+      "epoch": 1.4567901234567902,
+      "grad_norm": 0.1604948490858078,
+      "learning_rate": 1.6519081300889472e-06,
+      "loss": 0.3337,
+      "step": 478
+    },
+    {
+      "epoch": 1.4598765432098766,
+      "grad_norm": 0.23344826698303223,
+      "learning_rate": 1.6336629194240118e-06,
+      "loss": 0.3655,
+      "step": 479
+    },
+    {
+      "epoch": 1.462962962962963,
+      "grad_norm": 0.1553526222705841,
+      "learning_rate": 1.6154993322399114e-06,
+      "loss": 0.316,
+      "step": 480
+    },
+    {
+      "epoch": 1.4660493827160495,
+      "grad_norm": 0.1312614530324936,
+      "learning_rate": 1.5974178089486364e-06,
+      "loss": 0.301,
+      "step": 481
+    },
+    {
+      "epoch": 1.4691358024691357,
+      "grad_norm": 0.13480979204177856,
+      "learning_rate": 1.5794187879723755e-06,
+      "loss": 0.356,
+      "step": 482
+    },
+    {
+      "epoch": 1.4722222222222223,
+      "grad_norm": 0.14350688457489014,
+      "learning_rate": 1.561502705732883e-06,
+      "loss": 0.3021,
+      "step": 483
+    },
+    {
+      "epoch": 1.4753086419753085,
+      "grad_norm": 0.13871291279792786,
+      "learning_rate": 1.543669996640908e-06,
+      "loss": 0.4188,
+      "step": 484
+    },
+    {
+      "epoch": 1.4783950617283952,
+      "grad_norm": 0.16152562201023102,
+      "learning_rate": 1.5259210930856423e-06,
+      "loss": 0.3632,
+      "step": 485
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "grad_norm": 0.17420196533203125,
+      "learning_rate": 1.5082564254242583e-06,
+      "loss": 0.3735,
+      "step": 486
+    },
+    {
+      "epoch": 1.4814814814814814,
+      "eval_loss": 0.430364727973938,
+      "eval_runtime": 44.4346,
+      "eval_samples_per_second": 8.282,
+      "eval_steps_per_second": 1.035,
+      "step": 486
+    },
+    {
+      "epoch": 1.4845679012345678,
+      "grad_norm": 0.15298381447792053,
+      "learning_rate": 1.4906764219714537e-06,
+      "loss": 0.3162,
+      "step": 487
+    },
+    {
+      "epoch": 1.4876543209876543,
+      "grad_norm": 0.17767275869846344,
+      "learning_rate": 1.4731815089890795e-06,
+      "loss": 0.451,
+      "step": 488
+    },
+    {
+      "epoch": 1.4907407407407407,
+      "grad_norm": 0.2112477868795395,
+      "learning_rate": 1.455772110675804e-06,
+      "loss": 0.3914,
+      "step": 489
+    },
+    {
+      "epoch": 1.4938271604938271,
+      "grad_norm": 0.18488173186779022,
+      "learning_rate": 1.438448649156815e-06,
+      "loss": 0.3242,
+      "step": 490
+    },
+    {
+      "epoch": 1.4969135802469136,
+      "grad_norm": 0.19138255715370178,
+      "learning_rate": 1.4212115444736024e-06,
+      "loss": 0.3273,
+      "step": 491
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.17519411444664001,
+      "learning_rate": 1.4040612145737608e-06,
+      "loss": 0.314,
+      "step": 492
+    },
+    {
+      "epoch": 1.5030864197530864,
+      "grad_norm": 0.11331440508365631,
+      "learning_rate": 1.3869980753008537e-06,
+      "loss": 0.2184,
+      "step": 493
+    },
+    {
+      "epoch": 1.5061728395061729,
+      "grad_norm": 0.1674378216266632,
+      "learning_rate": 1.370022540384347e-06,
+      "loss": 0.3075,
+      "step": 494
+    },
+    {
+      "epoch": 1.5092592592592593,
+      "grad_norm": 0.14736564457416534,
+      "learning_rate": 1.353135021429554e-06,
+      "loss": 0.3719,
+      "step": 495
+    },
+    {
+      "epoch": 1.5123456790123457,
+      "grad_norm": 0.14618776738643646,
+      "learning_rate": 1.3363359279076776e-06,
+      "loss": 0.3625,
+      "step": 496
+    },
+    {
+      "epoch": 1.515432098765432,
+      "grad_norm": 0.15497514605522156,
+      "learning_rate": 1.3196256671458663e-06,
+      "loss": 0.3522,
+      "step": 497
+    },
+    {
+      "epoch": 1.5185185185185186,
+      "grad_norm": 0.1439277082681656,
+      "learning_rate": 1.3030046443173445e-06,
+      "loss": 0.2904,
+      "step": 498
+    },
+    {
+      "epoch": 1.5216049382716048,
+      "grad_norm": 0.14361339807510376,
+      "learning_rate": 1.2864732624315867e-06,
+      "loss": 0.3338,
+      "step": 499
+    },
+    {
+      "epoch": 1.5246913580246915,
+      "grad_norm": 0.1480712592601776,
+      "learning_rate": 1.270031922324546e-06,
+      "loss": 0.4092,
+      "step": 500
+    },
+    {
+      "epoch": 1.5277777777777777,
+      "grad_norm": 0.156494140625,
+      "learning_rate": 1.2536810226489354e-06,
+      "loss": 0.3855,
+      "step": 501
+    },
+    {
+      "epoch": 1.5308641975308643,
+      "grad_norm": 0.2111222743988037,
+      "learning_rate": 1.237420959864561e-06,
+      "loss": 0.4681,
+      "step": 502
+    },
+    {
+      "epoch": 1.5339506172839505,
+      "grad_norm": 0.20178188383579254,
+      "learning_rate": 1.2212521282287093e-06,
+      "loss": 0.3472,
+      "step": 503
+    },
+    {
+      "epoch": 1.5370370370370372,
+      "grad_norm": 0.14656566083431244,
+      "learning_rate": 1.2051749197865875e-06,
+      "loss": 0.2829,
+      "step": 504
+    },
+    {
+      "epoch": 1.5401234567901234,
+      "grad_norm": 0.17030468583106995,
+      "learning_rate": 1.1891897243618184e-06,
+      "loss": 0.457,
+      "step": 505
+    },
+    {
+      "epoch": 1.5432098765432098,
+      "grad_norm": 0.16490556299686432,
+      "learning_rate": 1.173296929546987e-06,
+      "loss": 0.4265,
+      "step": 506
+    },
+    {
+      "epoch": 1.5462962962962963,
+      "grad_norm": 0.15814335644245148,
+      "learning_rate": 1.1574969206942443e-06,
+      "loss": 0.3079,
+      "step": 507
+    },
+    {
+      "epoch": 1.5493827160493827,
+      "grad_norm": 0.15672267973423004,
+      "learning_rate": 1.1417900809059623e-06,
+      "loss": 0.2618,
+      "step": 508
+    },
+    {
+      "epoch": 1.5524691358024691,
+      "grad_norm": 0.26926475763320923,
+      "learning_rate": 1.1261767910254422e-06,
+      "loss": 0.4501,
+      "step": 509
+    },
+    {
+      "epoch": 1.5555555555555556,
+      "grad_norm": 0.22438615560531616,
+      "learning_rate": 1.1106574296276923e-06,
+      "loss": 0.5102,
+      "step": 510
+    },
+    {
+      "epoch": 1.558641975308642,
+      "grad_norm": 0.16849224269390106,
+      "learning_rate": 1.095232373010226e-06,
+      "loss": 0.4356,
+      "step": 511
+    },
+    {
+      "epoch": 1.5617283950617284,
+      "grad_norm": 0.15593089163303375,
+      "learning_rate": 1.0799019951839656e-06,
+      "loss": 0.2973,
+      "step": 512
+    },
+    {
+      "epoch": 1.5648148148148149,
+      "grad_norm": 0.14039039611816406,
+      "learning_rate": 1.0646666678641477e-06,
+      "loss": 0.4104,
+      "step": 513
+    },
+    {
+      "epoch": 1.567901234567901,
+      "grad_norm": 0.11041123420000076,
+      "learning_rate": 1.0495267604613273e-06,
+      "loss": 0.2541,
+      "step": 514
+    },
+    {
+      "epoch": 1.5709876543209877,
+      "grad_norm": 0.1312185525894165,
+      "learning_rate": 1.0344826400724185e-06,
+      "loss": 0.2818,
+      "step": 515
+    },
+    {
+      "epoch": 1.574074074074074,
+      "grad_norm": 0.20511452853679657,
+      "learning_rate": 1.0195346714717813e-06,
+      "loss": 0.3218,
+      "step": 516
+    },
+    {
+      "epoch": 1.5771604938271606,
+      "grad_norm": 0.2118871957063675,
+      "learning_rate": 1.0046832171023952e-06,
+      "loss": 0.2921,
+      "step": 517
+    },
+    {
+      "epoch": 1.5802469135802468,
+      "grad_norm": 0.18419800698757172,
+      "learning_rate": 9.899286370670575e-07,
+      "loss": 0.4502,
+      "step": 518
+    },
+    {
+      "epoch": 1.5833333333333335,
+      "grad_norm": 0.1755116879940033,
+      "learning_rate": 9.752712891196558e-07,
+      "loss": 0.3514,
+      "step": 519
+    },
+    {
+      "epoch": 1.5864197530864197,
+      "grad_norm": 0.16331788897514343,
+      "learning_rate": 9.607115286564972e-07,
+      "loss": 0.318,
+      "step": 520
+    },
+    {
+      "epoch": 1.5895061728395061,
+      "grad_norm": 0.18510426580905914,
+      "learning_rate": 9.46249708707681e-07,
+      "loss": 0.3207,
+      "step": 521
+    },
+    {
+      "epoch": 1.5925925925925926,
+      "grad_norm": 0.1467633843421936,
+      "learning_rate": 9.318861799285539e-07,
+      "loss": 0.32,
+      "step": 522
+    },
+    {
+      "epoch": 1.595679012345679,
+      "grad_norm": 0.21128030121326447,
+      "learning_rate": 9.176212905911946e-07,
+      "loss": 0.4566,
+      "step": 523
+    },
+    {
+      "epoch": 1.5987654320987654,
+      "grad_norm": 0.14944253861904144,
+      "learning_rate": 9.034553865759754e-07,
+      "loss": 0.4221,
+      "step": 524
+    },
+    {
+      "epoch": 1.6018518518518519,
+      "grad_norm": 0.1913837343454361,
+      "learning_rate": 8.893888113631732e-07,
+      "loss": 0.3236,
+      "step": 525
+    },
+    {
+      "epoch": 1.6049382716049383,
+      "grad_norm": 0.14830860495567322,
+      "learning_rate": 8.754219060246432e-07,
+      "loss": 0.3504,
+      "step": 526
+    },
+    {
+      "epoch": 1.6080246913580247,
+      "grad_norm": 0.1303461194038391,
+      "learning_rate": 8.615550092155478e-07,
+      "loss": 0.2281,
+      "step": 527
+    },
+    {
+      "epoch": 1.6111111111111112,
+      "grad_norm": 0.11773131787776947,
+      "learning_rate": 8.477884571661449e-07,
+      "loss": 0.2038,
+      "step": 528
+    },
+    {
+      "epoch": 1.6141975308641974,
+      "grad_norm": 0.16557615995407104,
+      "learning_rate": 8.341225836736367e-07,
+      "loss": 0.2965,
+      "step": 529
+    },
+    {
+      "epoch": 1.617283950617284,
+      "grad_norm": 0.15140382945537567,
+      "learning_rate": 8.20557720094074e-07,
+      "loss": 0.2804,
+      "step": 530
+    },
+    {
+      "epoch": 1.6203703703703702,
+      "grad_norm": 0.15120923519134521,
+      "learning_rate": 8.070941953343242e-07,
+      "loss": 0.3037,
+      "step": 531
+    },
+    {
+      "epoch": 1.623456790123457,
+      "grad_norm": 0.28693991899490356,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 0.4625,
+      "step": 532
+    },
+    {
+      "epoch": 1.626543209876543,
+      "grad_norm": 0.226279154419899,
+      "learning_rate": 7.804724656080182e-07,
+      "loss": 0.3529,
+      "step": 533
+    },
+    {
+      "epoch": 1.6296296296296298,
+      "grad_norm": 0.14384153485298157,
+      "learning_rate": 7.673149061377966e-07,
+      "loss": 0.4064,
+      "step": 534
+    },
+    {
+      "epoch": 1.632716049382716,
+      "grad_norm": 0.153773695230484,
+      "learning_rate": 7.542599764644049e-07,
+      "loss": 0.2779,
+      "step": 535
+    },
+    {
+      "epoch": 1.6358024691358026,
+      "grad_norm": 0.2235001176595688,
+      "learning_rate": 7.413079931303591e-07,
+      "loss": 0.4181,
+      "step": 536
+    },
+    {
+      "epoch": 1.6388888888888888,
+      "grad_norm": 0.1906222552061081,
+      "learning_rate": 7.284592701820325e-07,
+      "loss": 0.2867,
+      "step": 537
+    },
+    {
+      "epoch": 1.6419753086419753,
+      "grad_norm": 0.189738929271698,
+      "learning_rate": 7.157141191620548e-07,
+      "loss": 0.3274,
+      "step": 538
+    },
+    {
+      "epoch": 1.6450617283950617,
+      "grad_norm": 0.15748707950115204,
+      "learning_rate": 7.030728491017408e-07,
+      "loss": 0.2892,
+      "step": 539
+    },
+    {
+      "epoch": 1.6481481481481481,
+      "grad_norm": 0.2472158521413803,
+      "learning_rate": 6.905357665136142e-07,
+      "loss": 0.3892,
+      "step": 540
+    },
+    {
+      "epoch": 1.6512345679012346,
+      "grad_norm": 0.18736745417118073,
+      "learning_rate": 6.781031753839662e-07,
+      "loss": 0.3192,
+      "step": 541
+    },
+    {
+      "epoch": 1.654320987654321,
+      "grad_norm": 0.15377798676490784,
+      "learning_rate": 6.657753771654812e-07,
+      "loss": 0.2991,
+      "step": 542
+    },
+    {
+      "epoch": 1.6574074074074074,
+      "grad_norm": 0.16992682218551636,
+      "learning_rate": 6.535526707699408e-07,
+      "loss": 0.3628,
+      "step": 543
+    },
+    {
+      "epoch": 1.6604938271604939,
+      "grad_norm": 0.201069176197052,
+      "learning_rate": 6.414353525609628e-07,
+      "loss": 0.3127,
+      "step": 544
+    },
+    {
+      "epoch": 1.6635802469135803,
+      "grad_norm": 0.14373762905597687,
+      "learning_rate": 6.294237163468231e-07,
+      "loss": 0.2488,
+      "step": 545
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 0.16759946942329407,
+      "learning_rate": 6.175180533733277e-07,
+      "loss": 0.3833,
+      "step": 546
+    },
+    {
+      "epoch": 1.6697530864197532,
+      "grad_norm": 0.2061176598072052,
+      "learning_rate": 6.057186523167529e-07,
+      "loss": 0.252,
+      "step": 547
+    },
+    {
+      "epoch": 1.6728395061728394,
+      "grad_norm": 0.18383823335170746,
+      "learning_rate": 5.940257992768456e-07,
+      "loss": 0.3677,
+      "step": 548
+    },
+    {
+      "epoch": 1.675925925925926,
+      "grad_norm": 0.2329624891281128,
+      "learning_rate": 5.824397777698859e-07,
+      "loss": 0.3821,
+      "step": 549
+    },
+    {
+      "epoch": 1.6790123456790123,
+      "grad_norm": 0.16050845384597778,
+      "learning_rate": 5.709608687218116e-07,
+      "loss": 0.3203,
+      "step": 550
+    },
+    {
+      "epoch": 1.682098765432099,
+      "grad_norm": 0.1575547456741333,
+      "learning_rate": 5.595893504614097e-07,
+      "loss": 0.4154,
+      "step": 551
+    },
+    {
+      "epoch": 1.6851851851851851,
+      "grad_norm": 0.14166632294654846,
+      "learning_rate": 5.483254987135644e-07,
+      "loss": 0.2528,
+      "step": 552
+    },
+    {
+      "epoch": 1.6882716049382716,
+      "grad_norm": 0.1413419544696808,
+      "learning_rate": 5.371695865925736e-07,
+      "loss": 0.2011,
+      "step": 553
+    },
+    {
+      "epoch": 1.691358024691358,
+      "grad_norm": 0.14001396298408508,
+      "learning_rate": 5.261218845955246e-07,
+      "loss": 0.2521,
+      "step": 554
+    },
+    {
+      "epoch": 1.6944444444444444,
+      "grad_norm": 0.2379157692193985,
+      "learning_rate": 5.151826605957394e-07,
+      "loss": 0.3396,
+      "step": 555
+    },
+    {
+      "epoch": 1.6975308641975309,
+      "grad_norm": 0.1787138283252716,
+      "learning_rate": 5.043521798362755e-07,
+      "loss": 0.2596,
+      "step": 556
+    },
+    {
+      "epoch": 1.7006172839506173,
+      "grad_norm": 0.41910964250564575,
+      "learning_rate": 4.936307049234956e-07,
+      "loss": 0.3327,
+      "step": 557
+    },
+    {
+      "epoch": 1.7037037037037037,
+      "grad_norm": 0.1860780268907547,
+      "learning_rate": 4.830184958207007e-07,
+      "loss": 0.399,
+      "step": 558
+    },
+    {
+      "epoch": 1.7067901234567902,
+      "grad_norm": 0.16398878395557404,
+      "learning_rate": 4.725158098418309e-07,
+      "loss": 0.3953,
+      "step": 559
+    },
+    {
+      "epoch": 1.7098765432098766,
+      "grad_norm": 0.1744304746389389,
+      "learning_rate": 4.6212290164521554e-07,
+      "loss": 0.2567,
+      "step": 560
+    },
+    {
+      "epoch": 1.7129629629629628,
+      "grad_norm": 0.19683323800563812,
+      "learning_rate": 4.5184002322740784e-07,
+      "loss": 0.4327,
+      "step": 561
+    },
+    {
+      "epoch": 1.7160493827160495,
+      "grad_norm": 0.17663246393203735,
+      "learning_rate": 4.4166742391707593e-07,
+      "loss": 0.2145,
+      "step": 562
+    },
+    {
+      "epoch": 1.7191358024691357,
+      "grad_norm": 0.16606709361076355,
+      "learning_rate": 4.316053503689466e-07,
+      "loss": 0.3419,
+      "step": 563
+    },
+    {
+      "epoch": 1.7222222222222223,
+      "grad_norm": 0.21532438695430756,
+      "learning_rate": 4.2165404655783836e-07,
+      "loss": 0.379,
+      "step": 564
+    },
+    {
+      "epoch": 1.7253086419753085,
+      "grad_norm": 0.1450224667787552,
+      "learning_rate": 4.1181375377273237e-07,
+      "loss": 0.19,
+      "step": 565
+    },
+    {
+      "epoch": 1.7283950617283952,
+      "grad_norm": 0.18900087475776672,
+      "learning_rate": 4.020847106109349e-07,
+      "loss": 0.3304,
+      "step": 566
+    },
+    {
+      "epoch": 1.7314814814814814,
+      "grad_norm": 0.1328793317079544,
+      "learning_rate": 3.9246715297228176e-07,
+      "loss": 0.283,
+      "step": 567
+    },
+    {
+      "epoch": 1.7314814814814814,
+      "eval_loss": 0.42760223150253296,
+      "eval_runtime": 44.2033,
+      "eval_samples_per_second": 8.325,
+      "eval_steps_per_second": 1.041,
+      "step": 567
+    },
+    {
+      "epoch": 1.734567901234568,
+      "grad_norm": 0.14145122468471527,
+      "learning_rate": 3.829613140534222e-07,
+      "loss": 0.3045,
+      "step": 568
+    },
+    {
+      "epoch": 1.7376543209876543,
+      "grad_norm": 0.1800602227449417,
+      "learning_rate": 3.7356742434216775e-07,
+      "loss": 0.2553,
+      "step": 569
+    },
+    {
+      "epoch": 1.7407407407407407,
+      "grad_norm": 0.18250073492527008,
+      "learning_rate": 3.642857116118986e-07,
+      "loss": 0.23,
+      "step": 570
+    },
+    {
+      "epoch": 1.7438271604938271,
+      "grad_norm": 0.14363303780555725,
+      "learning_rate": 3.5511640091604293e-07,
+      "loss": 0.2744,
+      "step": 571
+    },
+    {
+      "epoch": 1.7469135802469136,
+      "grad_norm": 0.16794289648532867,
+      "learning_rate": 3.4605971458262e-07,
+      "loss": 0.3806,
+      "step": 572
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.15108714997768402,
+      "learning_rate": 3.371158722088497e-07,
+      "loss": 0.2868,
+      "step": 573
+    },
+    {
+      "epoch": 1.7530864197530864,
+      "grad_norm": 0.2250644415616989,
+      "learning_rate": 3.2828509065582713e-07,
+      "loss": 0.4173,
+      "step": 574
+    },
+    {
+      "epoch": 1.7561728395061729,
+      "grad_norm": 0.16634950041770935,
+      "learning_rate": 3.195675840432655e-07,
+      "loss": 0.3429,
+      "step": 575
+    },
+    {
+      "epoch": 1.7592592592592593,
+      "grad_norm": 0.3840501010417938,
+      "learning_rate": 3.109635637443026e-07,
+      "loss": 0.3564,
+      "step": 576
+    },
+    {
+      "epoch": 1.7623456790123457,
+      "grad_norm": 0.1317005604505539,
+      "learning_rate": 3.02473238380378e-07,
+      "loss": 0.2571,
+      "step": 577
+    },
+    {
+      "epoch": 1.765432098765432,
+      "grad_norm": 0.16465657949447632,
+      "learning_rate": 2.9409681381617315e-07,
+      "loss": 0.3739,
+      "step": 578
+    },
+    {
+      "epoch": 1.7685185185185186,
+      "grad_norm": 0.14124394953250885,
+      "learning_rate": 2.858344931546181e-07,
+      "loss": 0.2025,
+      "step": 579
+    },
+    {
+      "epoch": 1.7716049382716048,
+      "grad_norm": 0.19090065360069275,
+      "learning_rate": 2.776864767319731e-07,
+      "loss": 0.3652,
+      "step": 580
+    },
+    {
+      "epoch": 1.7746913580246915,
+      "grad_norm": 0.16761578619480133,
+      "learning_rate": 2.696529621129618e-07,
+      "loss": 0.3257,
+      "step": 581
+    },
+    {
+      "epoch": 1.7777777777777777,
+      "grad_norm": 0.17358000576496124,
+      "learning_rate": 2.617341440859883e-07,
+      "loss": 0.3162,
+      "step": 582
+    },
+    {
+      "epoch": 1.7808641975308643,
+      "grad_norm": 0.13688547909259796,
+      "learning_rate": 2.539302146584116e-07,
+      "loss": 0.2838,
+      "step": 583
+    },
+    {
+      "epoch": 1.7839506172839505,
+      "grad_norm": 0.12233246117830276,
+      "learning_rate": 2.4624136305188895e-07,
+      "loss": 0.2656,
+      "step": 584
+    },
+    {
+      "epoch": 1.7870370370370372,
+      "grad_norm": 0.14487585425376892,
+      "learning_rate": 2.3866777569779234e-07,
+      "loss": 0.2808,
+      "step": 585
+    },
+    {
+      "epoch": 1.7901234567901234,
+      "grad_norm": 0.1593523919582367,
+      "learning_rate": 2.3120963623267822e-07,
+      "loss": 0.3441,
+      "step": 586
+    },
+    {
+      "epoch": 1.7932098765432098,
+      "grad_norm": 0.1122526079416275,
+      "learning_rate": 2.2386712549384848e-07,
+      "loss": 0.1452,
+      "step": 587
+    },
+    {
+      "epoch": 1.7962962962962963,
+      "grad_norm": 0.1848554015159607,
+      "learning_rate": 2.1664042151495424e-07,
+      "loss": 0.407,
+      "step": 588
+    },
+    {
+      "epoch": 1.7993827160493827,
+      "grad_norm": 0.17059315741062164,
+      "learning_rate": 2.095296995216828e-07,
+      "loss": 0.3516,
+      "step": 589
+    },
+    {
+      "epoch": 1.8024691358024691,
+      "grad_norm": 0.18412597477436066,
+      "learning_rate": 2.0253513192751374e-07,
+      "loss": 0.2922,
+      "step": 590
+    },
+    {
+      "epoch": 1.8055555555555556,
+      "grad_norm": 0.17134982347488403,
+      "learning_rate": 1.9565688832952846e-07,
+      "loss": 0.2951,
+      "step": 591
+    },
+    {
+      "epoch": 1.808641975308642,
+      "grad_norm": 0.11777715384960175,
+      "learning_rate": 1.8889513550430892e-07,
+      "loss": 0.24,
+      "step": 592
+    },
+    {
+      "epoch": 1.8117283950617284,
+      "grad_norm": 0.18584772944450378,
+      "learning_rate": 1.8225003740388546e-07,
+      "loss": 0.3498,
+      "step": 593
+    },
+    {
+      "epoch": 1.8148148148148149,
+      "grad_norm": 0.15893200039863586,
+      "learning_rate": 1.7572175515176538e-07,
+      "loss": 0.3392,
+      "step": 594
+    },
+    {
+      "epoch": 1.817901234567901,
+      "grad_norm": 0.152305468916893,
+      "learning_rate": 1.693104470390261e-07,
+      "loss": 0.2333,
+      "step": 595
+    },
+    {
+      "epoch": 1.8209876543209877,
+      "grad_norm": 0.15064826607704163,
+      "learning_rate": 1.6301626852047504e-07,
+      "loss": 0.2935,
+      "step": 596
+    },
+    {
+      "epoch": 1.824074074074074,
+      "grad_norm": 0.18689890205860138,
+      "learning_rate": 1.5683937221088242e-07,
+      "loss": 0.4082,
+      "step": 597
+    },
+    {
+      "epoch": 1.8271604938271606,
+      "grad_norm": 0.16067026555538177,
+      "learning_rate": 1.5077990788127993e-07,
+      "loss": 0.2624,
+      "step": 598
+    },
+    {
+      "epoch": 1.8302469135802468,
+      "grad_norm": 0.15756982564926147,
+      "learning_rate": 1.448380224553303e-07,
+      "loss": 0.3681,
+      "step": 599
+    },
+    {
+      "epoch": 1.8333333333333335,
+      "grad_norm": 0.16193000972270966,
+      "learning_rate": 1.3901386000576112e-07,
+      "loss": 0.5148,
+      "step": 600
+    },
+    {
+      "epoch": 1.8364197530864197,
+      "grad_norm": 0.1545064002275467,
+      "learning_rate": 1.3330756175087778e-07,
+      "loss": 0.2837,
+      "step": 601
+    },
+    {
+      "epoch": 1.8395061728395061,
+      "grad_norm": 0.1584656536579132,
+      "learning_rate": 1.2771926605113283e-07,
+      "loss": 0.267,
+      "step": 602
+    },
+    {
+      "epoch": 1.8425925925925926,
+      "grad_norm": 0.23085588216781616,
+      "learning_rate": 1.2224910840577642e-07,
+      "loss": 0.3637,
+      "step": 603
+    },
+    {
+      "epoch": 1.845679012345679,
+      "grad_norm": 0.15698540210723877,
+      "learning_rate": 1.1689722144956672e-07,
+      "loss": 0.2152,
+      "step": 604
+    },
+    {
+      "epoch": 1.8487654320987654,
+      "grad_norm": 0.1545877605676651,
+      "learning_rate": 1.1166373494955696e-07,
+      "loss": 0.3073,
+      "step": 605
+    },
+    {
+      "epoch": 1.8518518518518519,
+      "grad_norm": 0.16467563807964325,
+      "learning_rate": 1.06548775801949e-07,
+      "loss": 0.3654,
+      "step": 606
+    },
+    {
+      "epoch": 1.8549382716049383,
+      "grad_norm": 0.20076429843902588,
+      "learning_rate": 1.0155246802901198e-07,
+      "loss": 0.3131,
+      "step": 607
+    },
+    {
+      "epoch": 1.8580246913580247,
+      "grad_norm": 0.14146511256694794,
+      "learning_rate": 9.667493277608187e-08,
+      "loss": 0.3651,
+      "step": 608
+    },
+    {
+      "epoch": 1.8611111111111112,
+      "grad_norm": 0.15111708641052246,
+      "learning_rate": 9.191628830861832e-08,
+      "loss": 0.267,
+      "step": 609
+    },
+    {
+      "epoch": 1.8641975308641974,
+      "grad_norm": 0.13036541640758514,
+      "learning_rate": 8.727665000934027e-08,
+      "loss": 0.2568,
+      "step": 610
+    },
+    {
+      "epoch": 1.867283950617284,
+      "grad_norm": 0.16827543079853058,
+      "learning_rate": 8.275613037542873e-08,
+      "loss": 0.4188,
+      "step": 611
+    },
+    {
+      "epoch": 1.8703703703703702,
+      "grad_norm": 0.18110865354537964,
+      "learning_rate": 7.835483901579454e-08,
+      "loss": 0.3361,
+      "step": 612
+    },
+    {
+      "epoch": 1.873456790123457,
+      "grad_norm": 0.1515679508447647,
+      "learning_rate": 7.407288264842772e-08,
+      "loss": 0.3421,
+      "step": 613
+    },
+    {
+      "epoch": 1.876543209876543,
+      "grad_norm": 0.1735447645187378,
+      "learning_rate": 6.991036509780391e-08,
+      "loss": 0.3908,
+      "step": 614
+    },
+    {
+      "epoch": 1.8796296296296298,
+      "grad_norm": 0.15131166577339172,
+      "learning_rate": 6.58673872923693e-08,
+      "loss": 0.2439,
+      "step": 615
+    },
+    {
+      "epoch": 1.882716049382716,
+      "grad_norm": 0.12076130509376526,
+      "learning_rate": 6.194404726209358e-08,
+      "loss": 0.2178,
+      "step": 616
+    },
+    {
+      "epoch": 1.8858024691358026,
+      "grad_norm": 0.1315135806798935,
+      "learning_rate": 5.8140440136091326e-08,
+      "loss": 0.2291,
+      "step": 617
+    },
+    {
+      "epoch": 1.8888888888888888,
+      "grad_norm": 0.17915165424346924,
+      "learning_rate": 5.445665814031942e-08,
+      "loss": 0.2377,
+      "step": 618
+    },
+    {
+      "epoch": 1.8919753086419753,
+      "grad_norm": 0.14008641242980957,
+      "learning_rate": 5.089279059533658e-08,
+      "loss": 0.2266,
+      "step": 619
+    },
+    {
+      "epoch": 1.8950617283950617,
+      "grad_norm": 0.18772335350513458,
+      "learning_rate": 4.744892391413791e-08,
+      "loss": 0.4006,
+      "step": 620
+    },
+    {
+      "epoch": 1.8981481481481481,
+      "grad_norm": 0.14937154948711395,
+      "learning_rate": 4.412514160006376e-08,
+      "loss": 0.3891,
+      "step": 621
+    },
+    {
+      "epoch": 1.9012345679012346,
+      "grad_norm": 0.12767252326011658,
+      "learning_rate": 4.092152424477025e-08,
+      "loss": 0.2397,
+      "step": 622
+    },
+    {
+      "epoch": 1.904320987654321,
+      "grad_norm": 0.16874873638153076,
+      "learning_rate": 3.7838149526277514e-08,
+      "loss": 0.3338,
+      "step": 623
+    },
+    {
+      "epoch": 1.9074074074074074,
+      "grad_norm": 0.1845911145210266,
+      "learning_rate": 3.487509220708563e-08,
+      "loss": 0.4378,
+      "step": 624
+    },
+    {
+      "epoch": 1.9104938271604939,
+      "grad_norm": 0.14064140617847443,
+      "learning_rate": 3.2032424132362736e-08,
+      "loss": 0.2801,
+      "step": 625
+    },
+    {
+      "epoch": 1.9135802469135803,
+      "grad_norm": 0.14805810153484344,
+      "learning_rate": 2.9310214228202016e-08,
+      "loss": 0.3122,
+      "step": 626
+    },
+    {
+      "epoch": 1.9166666666666665,
+      "grad_norm": 0.1921551674604416,
+      "learning_rate": 2.6708528499950758e-08,
+      "loss": 0.2982,
+      "step": 627
+    },
+    {
+      "epoch": 1.9197530864197532,
+      "grad_norm": 0.14775682985782623,
+      "learning_rate": 2.4227430030609455e-08,
+      "loss": 0.3503,
+      "step": 628
+    },
+    {
+      "epoch": 1.9228395061728394,
+      "grad_norm": 0.17906314134597778,
+      "learning_rate": 2.1866978979303567e-08,
+      "loss": 0.3863,
+      "step": 629
+    },
+    {
+      "epoch": 1.925925925925926,
+      "grad_norm": 0.1467551589012146,
+      "learning_rate": 1.962723257982302e-08,
+      "loss": 0.2993,
+      "step": 630
+    },
+    {
+      "epoch": 1.9290123456790123,
+      "grad_norm": 0.2205621749162674,
+      "learning_rate": 1.7508245139236658e-08,
+      "loss": 0.3168,
+      "step": 631
+    },
+    {
+      "epoch": 1.932098765432099,
+      "grad_norm": 0.1704474836587906,
+      "learning_rate": 1.5510068036573288e-08,
+      "loss": 0.3177,
+      "step": 632
+    },
+    {
+      "epoch": 1.9351851851851851,
+      "grad_norm": 0.15591393411159515,
+      "learning_rate": 1.3632749721577132e-08,
+      "loss": 0.2671,
+      "step": 633
+    },
+    {
+      "epoch": 1.9382716049382716,
+      "grad_norm": 0.1339595913887024,
+      "learning_rate": 1.1876335713532638e-08,
+      "loss": 0.196,
+      "step": 634
+    },
+    {
+      "epoch": 1.941358024691358,
+      "grad_norm": 0.15144091844558716,
+      "learning_rate": 1.024086860016149e-08,
+      "loss": 0.306,
+      "step": 635
+    },
+    {
+      "epoch": 1.9444444444444444,
+      "grad_norm": 0.14868693053722382,
+      "learning_rate": 8.726388036587874e-09,
+      "loss": 0.271,
+      "step": 636
+    },
+    {
+      "epoch": 1.9475308641975309,
+      "grad_norm": 0.14298443496227264,
+      "learning_rate": 7.332930744380906e-09,
+      "loss": 0.225,
+      "step": 637
+    },
+    {
+      "epoch": 1.9506172839506173,
+      "grad_norm": 0.14053991436958313,
+      "learning_rate": 6.060530510659246e-09,
+      "loss": 0.32,
+      "step": 638
+    },
+    {
+      "epoch": 1.9537037037037037,
+      "grad_norm": 0.2039446085691452,
+      "learning_rate": 4.909218187276743e-09,
+      "loss": 0.4306,
+      "step": 639
+    },
+    {
+      "epoch": 1.9567901234567902,
+      "grad_norm": 0.20658931136131287,
+      "learning_rate": 3.8790216900702615e-09,
+      "loss": 0.4053,
+      "step": 640
+    },
+    {
+      "epoch": 1.9598765432098766,
+      "grad_norm": 0.30260926485061646,
+      "learning_rate": 2.9699659981863306e-09,
+      "loss": 0.3979,
+      "step": 641
+    },
+    {
+      "epoch": 1.9629629629629628,
+      "grad_norm": 0.1412692815065384,
+      "learning_rate": 2.182073153471631e-09,
+      "loss": 0.1879,
+      "step": 642
+    },
+    {
+      "epoch": 1.9660493827160495,
+      "grad_norm": 0.11770602315664291,
+      "learning_rate": 1.5153622599428652e-09,
+      "loss": 0.2462,
+      "step": 643
+    },
+    {
+      "epoch": 1.9691358024691357,
+      "grad_norm": 0.156539648771286,
+      "learning_rate": 9.698494833199068e-10,
+      "loss": 0.3218,
+      "step": 644
+    },
+    {
+      "epoch": 1.9722222222222223,
+      "grad_norm": 0.19168072938919067,
+      "learning_rate": 5.455480506355582e-10,
+      "loss": 0.4821,
+      "step": 645
+    },
+    {
+      "epoch": 1.9753086419753085,
+      "grad_norm": 0.13230177760124207,
+      "learning_rate": 2.4246824991525085e-10,
+      "loss": 0.3134,
+      "step": 646
+    },
+    {
+      "epoch": 1.9783950617283952,
+      "grad_norm": 0.1942073255777359,
+      "learning_rate": 6.061742992613529e-11,
+      "loss": 0.3413,
+      "step": 647
+    },
+    {
+      "epoch": 1.9814814814814814,
+      "grad_norm": 0.15652911365032196,
+      "learning_rate": 0.0,
+      "loss": 0.2942,
+      "step": 648
+    },
+    {
+      "epoch": 1.9814814814814814,
+      "eval_loss": 0.42709851264953613,
+      "eval_runtime": 44.317,
+      "eval_samples_per_second": 8.304,
+      "eval_steps_per_second": 1.038,
+      "step": 648
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 648,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 162,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.584525189221712e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}