diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4636 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9994304157964686,
+  "eval_steps": 500,
+  "global_step": 658,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0015188912094171254,
+      "grad_norm": 494.0,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 21.3162,
+      "step": 1
+    },
+    {
+      "epoch": 0.003037782418834251,
+      "grad_norm": 440.0,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 19.5214,
+      "step": 2
+    },
+    {
+      "epoch": 0.004556673628251376,
+      "grad_norm": 462.0,
+      "learning_rate": 1.5e-06,
+      "loss": 20.3955,
+      "step": 3
+    },
+    {
+      "epoch": 0.006075564837668502,
+      "grad_norm": 446.0,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 19.8595,
+      "step": 4
+    },
+    {
+      "epoch": 0.007594456047085627,
+      "grad_norm": 440.0,
+      "learning_rate": 2.5e-06,
+      "loss": 19.3269,
+      "step": 5
+    },
+    {
+      "epoch": 0.009113347256502752,
+      "grad_norm": 432.0,
+      "learning_rate": 3e-06,
+      "loss": 20.3605,
+      "step": 6
+    },
+    {
+      "epoch": 0.010632238465919878,
+      "grad_norm": 408.0,
+      "learning_rate": 3.5e-06,
+      "loss": 18.9548,
+      "step": 7
+    },
+    {
+      "epoch": 0.012151129675337003,
+      "grad_norm": 366.0,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 17.6873,
+      "step": 8
+    },
+    {
+      "epoch": 0.013670020884754129,
+      "grad_norm": 346.0,
+      "learning_rate": 4.5e-06,
+      "loss": 17.031,
+      "step": 9
+    },
+    {
+      "epoch": 0.015188912094171255,
+      "grad_norm": 284.0,
+      "learning_rate": 5e-06,
+      "loss": 15.499,
+      "step": 10
+    },
+    {
+      "epoch": 0.01670780330358838,
+      "grad_norm": 238.0,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 13.135,
+      "step": 11
+    },
+    {
+      "epoch": 0.018226694513005504,
+      "grad_norm": 248.0,
+      "learning_rate": 6e-06,
+      "loss": 12.58,
+      "step": 12
+    },
+    {
+      "epoch": 0.01974558572242263,
+      "grad_norm": 251.0,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 10.6039,
+      "step": 13
+    },
+    {
+      "epoch": 0.021264476931839756,
+      "grad_norm": 180.0,
+      "learning_rate": 7e-06,
+      "loss": 9.4173,
+      "step": 14
+    },
+    {
+      "epoch": 0.022783368141256883,
+      "grad_norm": 108.5,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 8.5445,
+      "step": 15
+    },
+    {
+      "epoch": 0.024302259350674007,
+      "grad_norm": 81.5,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 8.3828,
+      "step": 16
+    },
+    {
+      "epoch": 0.025821150560091134,
+      "grad_norm": 67.0,
+      "learning_rate": 8.5e-06,
+      "loss": 7.3555,
+      "step": 17
+    },
+    {
+      "epoch": 0.027340041769508258,
+      "grad_norm": 46.25,
+      "learning_rate": 9e-06,
+      "loss": 6.5261,
+      "step": 18
+    },
+    {
+      "epoch": 0.028858932978925386,
+      "grad_norm": 62.5,
+      "learning_rate": 9.5e-06,
+      "loss": 6.4529,
+      "step": 19
+    },
+    {
+      "epoch": 0.03037782418834251,
+      "grad_norm": 57.25,
+      "learning_rate": 1e-05,
+      "loss": 6.1533,
+      "step": 20
+    },
+    {
+      "epoch": 0.03189671539775964,
+      "grad_norm": 36.75,
+      "learning_rate": 9.999939382570075e-06,
+      "loss": 5.6474,
+      "step": 21
+    },
+    {
+      "epoch": 0.03341560660717676,
+      "grad_norm": 20.375,
+      "learning_rate": 9.999757531750086e-06,
+      "loss": 5.379,
+      "step": 22
+    },
+    {
+      "epoch": 0.034934497816593885,
+      "grad_norm": 17.125,
+      "learning_rate": 9.999454451949364e-06,
+      "loss": 5.1134,
+      "step": 23
+    },
+    {
+      "epoch": 0.03645338902601101,
+      "grad_norm": 11.6875,
+      "learning_rate": 9.999030150516681e-06,
+      "loss": 4.8157,
+      "step": 24
+    },
+    {
+      "epoch": 0.03797228023542814,
+      "grad_norm": 10.1875,
+      "learning_rate": 9.998484637740058e-06,
+      "loss": 4.7658,
+      "step": 25
+    },
+    {
+      "epoch": 0.03949117144484526,
+      "grad_norm": 8.625,
+      "learning_rate": 9.997817926846528e-06,
+      "loss": 4.6094,
+      "step": 26
+    },
+    {
+      "epoch": 0.04101006265426239,
+      "grad_norm": 7.25,
+      "learning_rate": 9.997030034001815e-06,
+      "loss": 4.389,
+      "step": 27
+    },
+    {
+      "epoch": 0.04252895386367951,
+      "grad_norm": 6.78125,
+      "learning_rate": 9.99612097830993e-06,
+      "loss": 4.3941,
+      "step": 28
+    },
+    {
+      "epoch": 0.04404784507309664,
+      "grad_norm": 6.9375,
+      "learning_rate": 9.995090781812724e-06,
+      "loss": 4.3693,
+      "step": 29
+    },
+    {
+      "epoch": 0.045566736282513766,
+      "grad_norm": 6.84375,
+      "learning_rate": 9.993939469489342e-06,
+      "loss": 4.2591,
+      "step": 30
+    },
+    {
+      "epoch": 0.04708562749193089,
+      "grad_norm": 5.53125,
+      "learning_rate": 9.99266706925562e-06,
+      "loss": 4.2106,
+      "step": 31
+    },
+    {
+      "epoch": 0.048604518701348014,
+      "grad_norm": 5.09375,
+      "learning_rate": 9.991273611963413e-06,
+      "loss": 4.131,
+      "step": 32
+    },
+    {
+      "epoch": 0.050123409910765145,
+      "grad_norm": 6.09375,
+      "learning_rate": 9.98975913139984e-06,
+      "loss": 4.0936,
+      "step": 33
+    },
+    {
+      "epoch": 0.05164230112018227,
+      "grad_norm": 7.96875,
+      "learning_rate": 9.98812366428647e-06,
+      "loss": 4.0474,
+      "step": 34
+    },
+    {
+      "epoch": 0.05316119232959939,
+      "grad_norm": 6.375,
+      "learning_rate": 9.986367250278423e-06,
+      "loss": 4.0911,
+      "step": 35
+    },
+    {
+      "epoch": 0.054680083539016516,
+      "grad_norm": 8.0,
+      "learning_rate": 9.984489931963429e-06,
+      "loss": 3.8623,
+      "step": 36
+    },
+    {
+      "epoch": 0.05619897474843364,
+      "grad_norm": 4.46875,
+      "learning_rate": 9.982491754860763e-06,
+      "loss": 3.7118,
+      "step": 37
+    },
+    {
+      "epoch": 0.05771786595785077,
+      "grad_norm": 5.8125,
+      "learning_rate": 9.980372767420179e-06,
+      "loss": 3.9689,
+      "step": 38
+    },
+    {
+      "epoch": 0.059236757167267895,
+      "grad_norm": 4.3125,
+      "learning_rate": 9.978133021020697e-06,
+      "loss": 3.782,
+      "step": 39
+    },
+    {
+      "epoch": 0.06075564837668502,
+      "grad_norm": 5.0,
+      "learning_rate": 9.97577256996939e-06,
+      "loss": 3.7709,
+      "step": 40
+    },
+    {
+      "epoch": 0.06227453958610214,
+      "grad_norm": 5.34375,
+      "learning_rate": 9.97329147150005e-06,
+      "loss": 3.6426,
+      "step": 41
+    },
+    {
+      "epoch": 0.06379343079551927,
+      "grad_norm": 5.09375,
+      "learning_rate": 9.970689785771798e-06,
+      "loss": 3.5875,
+      "step": 42
+    },
+    {
+      "epoch": 0.06531232200493639,
+      "grad_norm": 5.90625,
+      "learning_rate": 9.96796757586764e-06,
+      "loss": 3.7528,
+      "step": 43
+    },
+    {
+      "epoch": 0.06683121321435352,
+      "grad_norm": 4.0,
+      "learning_rate": 9.965124907792916e-06,
+      "loss": 3.5526,
+      "step": 44
+    },
+    {
+      "epoch": 0.06835010442377065,
+      "grad_norm": 5.15625,
+      "learning_rate": 9.962161850473723e-06,
+      "loss": 3.4634,
+      "step": 45
+    },
+    {
+      "epoch": 0.06986899563318777,
+      "grad_norm": 4.25,
+      "learning_rate": 9.95907847575523e-06,
+      "loss": 3.534,
+      "step": 46
+    },
+    {
+      "epoch": 0.0713878868426049,
+      "grad_norm": 4.09375,
+      "learning_rate": 9.955874858399936e-06,
+      "loss": 3.4982,
+      "step": 47
+    },
+    {
+      "epoch": 0.07290677805202202,
+      "grad_norm": 4.40625,
+      "learning_rate": 9.952551076085864e-06,
+      "loss": 3.5063,
+      "step": 48
+    },
+    {
+      "epoch": 0.07442566926143915,
+      "grad_norm": 4.4375,
+      "learning_rate": 9.949107209404664e-06,
+      "loss": 3.4372,
+      "step": 49
+    },
+    {
+      "epoch": 0.07594456047085628,
+      "grad_norm": 4.46875,
+      "learning_rate": 9.945543341859681e-06,
+      "loss": 3.4007,
+      "step": 50
+    },
+    {
+      "epoch": 0.0774634516802734,
+      "grad_norm": 4.5,
+      "learning_rate": 9.94185955986391e-06,
+      "loss": 3.2577,
+      "step": 51
+    },
+    {
+      "epoch": 0.07898234288969053,
+      "grad_norm": 4.0625,
+      "learning_rate": 9.938055952737908e-06,
+      "loss": 3.2997,
+      "step": 52
+    },
+    {
+      "epoch": 0.08050123409910766,
+      "grad_norm": 3.984375,
+      "learning_rate": 9.934132612707631e-06,
+      "loss": 3.3514,
+      "step": 53
+    },
+    {
+      "epoch": 0.08202012530852477,
+      "grad_norm": 3.390625,
+      "learning_rate": 9.930089634902197e-06,
+      "loss": 3.1966,
+      "step": 54
+    },
+    {
+      "epoch": 0.0835390165179419,
+      "grad_norm": 5.78125,
+      "learning_rate": 9.925927117351573e-06,
+      "loss": 3.1777,
+      "step": 55
+    },
+    {
+      "epoch": 0.08505790772735902,
+      "grad_norm": 4.125,
+      "learning_rate": 9.921645160984205e-06,
+      "loss": 3.0858,
+      "step": 56
+    },
+    {
+      "epoch": 0.08657679893677615,
+      "grad_norm": 4.90625,
+      "learning_rate": 9.917243869624573e-06,
+      "loss": 3.1884,
+      "step": 57
+    },
+    {
+      "epoch": 0.08809569014619328,
+      "grad_norm": 4.6875,
+      "learning_rate": 9.91272334999066e-06,
+      "loss": 3.1061,
+      "step": 58
+    },
+    {
+      "epoch": 0.0896145813556104,
+      "grad_norm": 4.15625,
+      "learning_rate": 9.908083711691383e-06,
+      "loss": 3.1701,
+      "step": 59
+    },
+    {
+      "epoch": 0.09113347256502753,
+      "grad_norm": 5.125,
+      "learning_rate": 9.903325067223918e-06,
+      "loss": 3.1281,
+      "step": 60
+    },
+    {
+      "epoch": 0.09265236377444465,
+      "grad_norm": 4.59375,
+      "learning_rate": 9.898447531970989e-06,
+      "loss": 3.0717,
+      "step": 61
+    },
+    {
+      "epoch": 0.09417125498386178,
+      "grad_norm": 5.9375,
+      "learning_rate": 9.893451224198051e-06,
+      "loss": 3.0151,
+      "step": 62
+    },
+    {
+      "epoch": 0.09569014619327891,
+      "grad_norm": 4.25,
+      "learning_rate": 9.888336265050443e-06,
+      "loss": 3.0585,
+      "step": 63
+    },
+    {
+      "epoch": 0.09720903740269603,
+      "grad_norm": 5.28125,
+      "learning_rate": 9.883102778550434e-06,
+      "loss": 3.0649,
+      "step": 64
+    },
+    {
+      "epoch": 0.09872792861211316,
+      "grad_norm": 3.53125,
+      "learning_rate": 9.877750891594224e-06,
+      "loss": 3.0054,
+      "step": 65
+    },
+    {
+      "epoch": 0.10024681982153029,
+      "grad_norm": 4.3125,
+      "learning_rate": 9.872280733948867e-06,
+      "loss": 2.911,
+      "step": 66
+    },
+    {
+      "epoch": 0.1017657110309474,
+      "grad_norm": 4.375,
+      "learning_rate": 9.866692438249124e-06,
+      "loss": 3.0651,
+      "step": 67
+    },
+    {
+      "epoch": 0.10328460224036454,
+      "grad_norm": 4.34375,
+      "learning_rate": 9.86098613999424e-06,
+      "loss": 2.9609,
+      "step": 68
+    },
+    {
+      "epoch": 0.10480349344978165,
+      "grad_norm": 3.75,
+      "learning_rate": 9.855161977544672e-06,
+      "loss": 3.0308,
+      "step": 69
+    },
+    {
+      "epoch": 0.10632238465919878,
+      "grad_norm": 3.078125,
+      "learning_rate": 9.849220092118721e-06,
+      "loss": 2.97,
+      "step": 70
+    },
+    {
+      "epoch": 0.10784127586861592,
+      "grad_norm": 3.390625,
+      "learning_rate": 9.84316062778912e-06,
+      "loss": 2.9543,
+      "step": 71
+    },
+    {
+      "epoch": 0.10936016707803303,
+      "grad_norm": 2.6875,
+      "learning_rate": 9.836983731479526e-06,
+      "loss": 2.8797,
+      "step": 72
+    },
+    {
+      "epoch": 0.11087905828745016,
+      "grad_norm": 2.84375,
+      "learning_rate": 9.830689552960974e-06,
+      "loss": 2.8188,
+      "step": 73
+    },
+    {
+      "epoch": 0.11239794949686728,
+      "grad_norm": 3.171875,
+      "learning_rate": 9.824278244848236e-06,
+      "loss": 2.9654,
+      "step": 74
+    },
+    {
+      "epoch": 0.11391684070628441,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.817749962596115e-06,
+      "loss": 2.8417,
+      "step": 75
+    },
+    {
+      "epoch": 0.11543573191570154,
+      "grad_norm": 2.671875,
+      "learning_rate": 9.811104864495691e-06,
+      "loss": 2.9101,
+      "step": 76
+    },
+    {
+      "epoch": 0.11695462312511866,
+      "grad_norm": 2.6875,
+      "learning_rate": 9.804343111670472e-06,
+      "loss": 2.8579,
+      "step": 77
+    },
+    {
+      "epoch": 0.11847351433453579,
+      "grad_norm": 2.53125,
+      "learning_rate": 9.797464868072489e-06,
+      "loss": 2.8388,
+      "step": 78
+    },
+    {
+      "epoch": 0.11999240554395292,
+      "grad_norm": 3.078125,
+      "learning_rate": 9.790470300478318e-06,
+      "loss": 2.8716,
+      "step": 79
+    },
+    {
+      "epoch": 0.12151129675337004,
+      "grad_norm": 2.046875,
+      "learning_rate": 9.783359578485047e-06,
+      "loss": 2.8199,
+      "step": 80
+    },
+    {
+      "epoch": 0.12303018796278717,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.776132874506153e-06,
+      "loss": 2.8302,
+      "step": 81
+    },
+    {
+      "epoch": 0.12454907917220429,
+      "grad_norm": 2.140625,
+      "learning_rate": 9.768790363767321e-06,
+      "loss": 2.795,
+      "step": 82
+    },
+    {
+      "epoch": 0.1260679703816214,
+      "grad_norm": 2.125,
+      "learning_rate": 9.761332224302209e-06,
+      "loss": 2.7479,
+      "step": 83
+    },
+    {
+      "epoch": 0.12758686159103855,
+      "grad_norm": 2.765625,
+      "learning_rate": 9.753758636948112e-06,
+      "loss": 2.7841,
+      "step": 84
+    },
+    {
+      "epoch": 0.12910575280045566,
+      "grad_norm": 2.125,
+      "learning_rate": 9.74606978534159e-06,
+      "loss": 2.8815,
+      "step": 85
+    },
+    {
+      "epoch": 0.13062464400987278,
+      "grad_norm": 3.328125,
+      "learning_rate": 9.738265855914014e-06,
+      "loss": 2.7458,
+      "step": 86
+    },
+    {
+      "epoch": 0.13214353521928993,
+      "grad_norm": 2.3125,
+      "learning_rate": 9.730347037887041e-06,
+      "loss": 2.8145,
+      "step": 87
+    },
+    {
+      "epoch": 0.13366242642870704,
+      "grad_norm": 2.9375,
+      "learning_rate": 9.722313523268028e-06,
+      "loss": 2.7804,
+      "step": 88
+    },
+    {
+      "epoch": 0.13518131763812416,
+      "grad_norm": 2.578125,
+      "learning_rate": 9.714165506845381e-06,
+      "loss": 2.7899,
+      "step": 89
+    },
+    {
+      "epoch": 0.1367002088475413,
+      "grad_norm": 3.359375,
+      "learning_rate": 9.705903186183828e-06,
+      "loss": 2.7296,
+      "step": 90
+    },
+    {
+      "epoch": 0.13821910005695842,
+      "grad_norm": 2.96875,
+      "learning_rate": 9.697526761619621e-06,
+      "loss": 2.7233,
+      "step": 91
+    },
+    {
+      "epoch": 0.13973799126637554,
+      "grad_norm": 2.09375,
+      "learning_rate": 9.689036436255698e-06,
+      "loss": 2.7421,
+      "step": 92
+    },
+    {
+      "epoch": 0.14125688247579268,
+      "grad_norm": 2.828125,
+      "learning_rate": 9.680432415956736e-06,
+      "loss": 2.7627,
+      "step": 93
+    },
+    {
+      "epoch": 0.1427757736852098,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.671714909344175e-06,
+      "loss": 2.7611,
+      "step": 94
+    },
+    {
+      "epoch": 0.14429466489462692,
+      "grad_norm": 1.9140625,
+      "learning_rate": 9.66288412779115e-06,
+      "loss": 2.753,
+      "step": 95
+    },
+    {
+      "epoch": 0.14581355610404403,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.653940285417381e-06,
+      "loss": 2.7191,
+      "step": 96
+    },
+    {
+      "epoch": 0.14733244731346118,
+      "grad_norm": 1.4765625,
+      "learning_rate": 9.644883599083959e-06,
+      "loss": 2.7194,
+      "step": 97
+    },
+    {
+      "epoch": 0.1488513385228783,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.635714288388103e-06,
+      "loss": 2.6886,
+      "step": 98
+    },
+    {
+      "epoch": 0.1503702297322954,
+      "grad_norm": 1.7265625,
+      "learning_rate": 9.626432575657834e-06,
+      "loss": 2.7631,
+      "step": 99
+    },
+    {
+      "epoch": 0.15188912094171256,
+      "grad_norm": 1.6796875,
+      "learning_rate": 9.617038685946578e-06,
+      "loss": 2.6749,
+      "step": 100
+    },
+    {
+      "epoch": 0.15340801215112967,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.60753284702772e-06,
+      "loss": 2.6299,
+      "step": 101
+    },
+    {
+      "epoch": 0.1549269033605468,
+      "grad_norm": 1.421875,
+      "learning_rate": 9.597915289389067e-06,
+      "loss": 2.6723,
+      "step": 102
+    },
+    {
+      "epoch": 0.15644579456996394,
+      "grad_norm": 1.5625,
+      "learning_rate": 9.58818624622727e-06,
+      "loss": 2.7104,
+      "step": 103
+    },
+    {
+      "epoch": 0.15796468577938105,
+      "grad_norm": 1.640625,
+      "learning_rate": 9.578345953442163e-06,
+      "loss": 2.6862,
+      "step": 104
+    },
+    {
+      "epoch": 0.15948357698879817,
+      "grad_norm": 1.8515625,
+      "learning_rate": 9.568394649631055e-06,
+      "loss": 2.7183,
+      "step": 105
+    },
+    {
+      "epoch": 0.16100246819821531,
+      "grad_norm": 1.9453125,
+      "learning_rate": 9.558332576082925e-06,
+      "loss": 2.6991,
+      "step": 106
+    },
+    {
+      "epoch": 0.16252135940763243,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.548159976772593e-06,
+      "loss": 2.591,
+      "step": 107
+    },
+    {
+      "epoch": 0.16404025061704955,
+      "grad_norm": 1.703125,
+      "learning_rate": 9.537877098354787e-06,
+      "loss": 2.6819,
+      "step": 108
+    },
+    {
+      "epoch": 0.16555914182646667,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.527484190158171e-06,
+      "loss": 2.6957,
+      "step": 109
+    },
+    {
+      "epoch": 0.1670780330358838,
+      "grad_norm": 2.265625,
+      "learning_rate": 9.5169815041793e-06,
+      "loss": 2.7064,
+      "step": 110
+    },
+    {
+      "epoch": 0.16859692424530093,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.506369295076505e-06,
+      "loss": 2.6742,
+      "step": 111
+    },
+    {
+      "epoch": 0.17011581545471804,
+      "grad_norm": 1.640625,
+      "learning_rate": 9.495647820163725e-06,
+      "loss": 2.6757,
+      "step": 112
+    },
+    {
+      "epoch": 0.1716347066641352,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.484817339404261e-06,
+      "loss": 2.6503,
+      "step": 113
+    },
+    {
+      "epoch": 0.1731535978735523,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.473878115404477e-06,
+      "loss": 2.6608,
+      "step": 114
+    },
+    {
+      "epoch": 0.17467248908296942,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.462830413407427e-06,
+      "loss": 2.6217,
+      "step": 115
+    },
+    {
+      "epoch": 0.17619138029238657,
+      "grad_norm": 1.578125,
+      "learning_rate": 9.451674501286436e-06,
+      "loss": 2.6885,
+      "step": 116
+    },
+    {
+      "epoch": 0.17771027150180368,
+      "grad_norm": 1.71875,
+      "learning_rate": 9.440410649538592e-06,
+      "loss": 2.6041,
+      "step": 117
+    },
+    {
+      "epoch": 0.1792291627112208,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.42903913127819e-06,
+      "loss": 2.6156,
+      "step": 118
+    },
+    {
+      "epoch": 0.18074805392063795,
+      "grad_norm": 1.796875,
+      "learning_rate": 9.417560222230115e-06,
+      "loss": 2.5769,
+      "step": 119
+    },
+    {
+      "epoch": 0.18226694513005506,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.405974200723156e-06,
+      "loss": 2.5831,
+      "step": 120
+    },
+    {
+      "epoch": 0.18378583633947218,
+      "grad_norm": 1.59375,
+      "learning_rate": 9.394281347683247e-06,
+      "loss": 2.7265,
+      "step": 121
+    },
+    {
+      "epoch": 0.1853047275488893,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.382481946626673e-06,
+      "loss": 2.6134,
+      "step": 122
+    },
+    {
+      "epoch": 0.18682361875830644,
+      "grad_norm": 1.6015625,
+      "learning_rate": 9.370576283653178e-06,
+      "loss": 2.5806,
+      "step": 123
+    },
+    {
+      "epoch": 0.18834250996772356,
+      "grad_norm": 1.5078125,
+      "learning_rate": 9.358564647439037e-06,
+      "loss": 2.5152,
+      "step": 124
+    },
+    {
+      "epoch": 0.18986140117714068,
+      "grad_norm": 1.546875,
+      "learning_rate": 9.34644732923006e-06,
+      "loss": 2.6123,
+      "step": 125
+    },
+    {
+      "epoch": 0.19138029238655782,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.33422462283452e-06,
+      "loss": 2.6245,
+      "step": 126
+    },
+    {
+      "epoch": 0.19289918359597494,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.321896824616036e-06,
+      "loss": 2.5982,
+      "step": 127
+    },
+    {
+      "epoch": 0.19441807480539205,
+      "grad_norm": 2.1875,
+      "learning_rate": 9.309464233486386e-06,
+      "loss": 2.61,
+      "step": 128
+    },
+    {
+      "epoch": 0.1959369660148092,
+      "grad_norm": 1.3984375,
+      "learning_rate": 9.29692715089826e-06,
+      "loss": 2.5773,
+      "step": 129
+    },
+    {
+      "epoch": 0.19745585722422632,
+      "grad_norm": 1.4296875,
+      "learning_rate": 9.284285880837947e-06,
+      "loss": 2.5853,
+      "step": 130
+    },
+    {
+      "epoch": 0.19897474843364343,
+      "grad_norm": 1.46875,
+      "learning_rate": 9.271540729817969e-06,
+      "loss": 2.5594,
+      "step": 131
+    },
+    {
+      "epoch": 0.20049363964306058,
+      "grad_norm": 1.765625,
+      "learning_rate": 9.258692006869644e-06,
+      "loss": 2.5041,
+      "step": 132
+    },
+    {
+      "epoch": 0.2020125308524777,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.245740023535596e-06,
+      "loss": 2.5043,
+      "step": 133
+    },
+    {
+      "epoch": 0.2035314220618948,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.232685093862206e-06,
+      "loss": 2.5629,
+      "step": 134
+    },
+    {
+      "epoch": 0.20505031327131193,
+      "grad_norm": 1.34375,
+      "learning_rate": 9.219527534391983e-06,
+      "loss": 2.5121,
+      "step": 135
+    },
+    {
+      "epoch": 0.20656920448072907,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.206267664155906e-06,
+      "loss": 2.5578,
+      "step": 136
+    },
+    {
+      "epoch": 0.2080880956901462,
+      "grad_norm": 1.203125,
+      "learning_rate": 9.192905804665677e-06,
+      "loss": 2.5341,
+      "step": 137
+    },
+    {
+      "epoch": 0.2096069868995633,
+      "grad_norm": 2.203125,
+      "learning_rate": 9.179442279905927e-06,
+      "loss": 2.5867,
+      "step": 138
+    },
+    {
+      "epoch": 0.21112587810898045,
+      "grad_norm": 1.2890625,
+      "learning_rate": 9.165877416326365e-06,
+      "loss": 2.6066,
+      "step": 139
+    },
+    {
+      "epoch": 0.21264476931839757,
+      "grad_norm": 1.8828125,
+      "learning_rate": 9.152211542833856e-06,
+      "loss": 2.6332,
+      "step": 140
+    },
+    {
+      "epoch": 0.2141636605278147,
+      "grad_norm": 1.6640625,
+      "learning_rate": 9.138444990784455e-06,
+      "loss": 2.5835,
+      "step": 141
+    },
+    {
+      "epoch": 0.21568255173723183,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.124578093975358e-06,
+      "loss": 2.622,
+      "step": 142
+    },
+    {
+      "epoch": 0.21720144294664895,
+      "grad_norm": 1.4453125,
+      "learning_rate": 9.110611188636828e-06,
+      "loss": 2.5668,
+      "step": 143
+    },
+    {
+      "epoch": 0.21872033415606607,
+      "grad_norm": 1.1640625,
+      "learning_rate": 9.096544613424026e-06,
+      "loss": 2.6229,
+      "step": 144
+    },
+    {
+      "epoch": 0.2202392253654832,
+      "grad_norm": 1.515625,
+      "learning_rate": 9.082378709408805e-06,
+      "loss": 2.5456,
+      "step": 145
+    },
+    {
+      "epoch": 0.22175811657490033,
+      "grad_norm": 1.296875,
+      "learning_rate": 9.068113820071447e-06,
+      "loss": 2.4778,
+      "step": 146
+    },
+    {
+      "epoch": 0.22327700778431744,
+      "grad_norm": 1.53125,
+      "learning_rate": 9.053750291292321e-06,
+      "loss": 2.4628,
+      "step": 147
+    },
+    {
+      "epoch": 0.22479589899373456,
+      "grad_norm": 1.265625,
+      "learning_rate": 9.039288471343505e-06,
+      "loss": 2.5147,
+      "step": 148
+    },
+    {
+      "epoch": 0.2263147902031517,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.024728710880345e-06,
+      "loss": 2.5673,
+      "step": 149
+    },
+    {
+      "epoch": 0.22783368141256882,
+      "grad_norm": 1.2265625,
+      "learning_rate": 9.010071362932945e-06,
+      "loss": 2.5198,
+      "step": 150
+    },
+    {
+      "epoch": 0.22935257262198594,
+      "grad_norm": 1.40625,
+      "learning_rate": 8.995316782897605e-06,
+      "loss": 2.5247,
+      "step": 151
+    },
+    {
+      "epoch": 0.23087146383140308,
+      "grad_norm": 1.3046875,
+      "learning_rate": 8.98046532852822e-06,
+      "loss": 2.5068,
+      "step": 152
+    },
+    {
+      "epoch": 0.2323903550408202,
+      "grad_norm": 1.2421875,
+      "learning_rate": 8.965517359927583e-06,
+      "loss": 2.5362,
+      "step": 153
+    },
+    {
+      "epoch": 0.23390924625023732,
+      "grad_norm": 1.2265625,
+      "learning_rate": 8.950473239538672e-06,
+      "loss": 2.5166,
+      "step": 154
+    },
+    {
+      "epoch": 0.23542813745965446,
+      "grad_norm": 1.421875,
+      "learning_rate": 8.935333332135853e-06,
+      "loss": 2.502,
+      "step": 155
+    },
+    {
+      "epoch": 0.23694702866907158,
+      "grad_norm": 1.6796875,
+      "learning_rate": 8.920098004816035e-06,
+      "loss": 2.5186,
+      "step": 156
+    },
+    {
+      "epoch": 0.2384659198784887,
+      "grad_norm": 1.0859375,
+      "learning_rate": 8.904767626989774e-06,
+      "loss": 2.5439,
+      "step": 157
+    },
+    {
+      "epoch": 0.23998481108790584,
+      "grad_norm": 1.1640625,
+      "learning_rate": 8.88934257037231e-06,
+      "loss": 2.5026,
+      "step": 158
+    },
+    {
+      "epoch": 0.24150370229732296,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.873823208974557e-06,
+      "loss": 2.5948,
+      "step": 159
+    },
+    {
+      "epoch": 0.24302259350674008,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.85820991909404e-06,
+      "loss": 2.5613,
+      "step": 160
+    },
+    {
+      "epoch": 0.2445414847161572,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.842503079305757e-06,
+      "loss": 2.5053,
+      "step": 161
+    },
+    {
+      "epoch": 0.24606037592557434,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.826703070453014e-06,
+      "loss": 2.4929,
+      "step": 162
+    },
+    {
+      "epoch": 0.24757926713499145,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.810810275638183e-06,
+      "loss": 2.4863,
+      "step": 163
+    },
+    {
+      "epoch": 0.24909815834440857,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.794825080213415e-06,
+      "loss": 2.5587,
+      "step": 164
+    },
+    {
+      "epoch": 0.2506170495538257,
+      "grad_norm": 1.15625,
+      "learning_rate": 8.778747871771293e-06,
+      "loss": 2.5292,
+      "step": 165
+    },
+    {
+      "epoch": 0.2521359407632428,
+      "grad_norm": 1.375,
+      "learning_rate": 8.76257904013544e-06,
+      "loss": 2.5414,
+      "step": 166
+    },
+    {
+      "epoch": 0.25365483197266,
+      "grad_norm": 1.234375,
+      "learning_rate": 8.746318977351066e-06,
+      "loss": 2.5376,
+      "step": 167
+    },
+    {
+      "epoch": 0.2551737231820771,
+      "grad_norm": 1.390625,
+      "learning_rate": 8.729968077675454e-06,
+      "loss": 2.5962,
+      "step": 168
+    },
+    {
+      "epoch": 0.2566926143914942,
+      "grad_norm": 1.265625,
+      "learning_rate": 8.713526737568415e-06,
+      "loss": 2.5141,
+      "step": 169
+    },
+    {
+      "epoch": 0.25821150560091133,
+      "grad_norm": 1.375,
+      "learning_rate": 8.696995355682656e-06,
+      "loss": 2.6006,
+      "step": 170
+    },
+    {
+      "epoch": 0.25973039681032845,
+      "grad_norm": 1.5390625,
+      "learning_rate": 8.680374332854134e-06,
+      "loss": 2.4407,
+      "step": 171
+    },
+    {
+      "epoch": 0.26124928801974556,
+      "grad_norm": 1.359375,
+      "learning_rate": 8.663664072092324e-06,
+      "loss": 2.4783,
+      "step": 172
+    },
+    {
+      "epoch": 0.26276817922916273,
+      "grad_norm": 1.421875,
+      "learning_rate": 8.646864978570445e-06,
+      "loss": 2.513,
+      "step": 173
+    },
+    {
+      "epoch": 0.26428707043857985,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.629977459615655e-06,
+      "loss": 2.4805,
+      "step": 174
+    },
+    {
+      "epoch": 0.26580596164799697,
+      "grad_norm": 1.390625,
+      "learning_rate": 8.613001924699146e-06,
+      "loss": 2.5176,
+      "step": 175
+    },
+    {
+      "epoch": 0.2673248528574141,
+      "grad_norm": 1.3125,
+      "learning_rate": 8.595938785426241e-06,
+      "loss": 2.5586,
+      "step": 176
+    },
+    {
+      "epoch": 0.2688437440668312,
+      "grad_norm": 1.3671875,
+      "learning_rate": 8.578788455526398e-06,
+      "loss": 2.4918,
+      "step": 177
+    },
+    {
+      "epoch": 0.2703626352762483,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.561551350843185e-06,
+      "loss": 2.5574,
+      "step": 178
+    },
+    {
+      "epoch": 0.27188152648566544,
+      "grad_norm": 1.3828125,
+      "learning_rate": 8.544227889324199e-06,
+      "loss": 2.4831,
+      "step": 179
+    },
+    {
+      "epoch": 0.2734004176950826,
+      "grad_norm": 1.4296875,
+      "learning_rate": 8.526818491010922e-06,
+      "loss": 2.5496,
+      "step": 180
+    },
+    {
+      "epoch": 0.2749193089044997,
+      "grad_norm": 1.5078125,
+      "learning_rate": 8.509323578028547e-06,
+      "loss": 2.5047,
+      "step": 181
+    },
+    {
+      "epoch": 0.27643820011391684,
+      "grad_norm": 1.5234375,
+      "learning_rate": 8.491743574575743e-06,
+      "loss": 2.4502,
+      "step": 182
+    },
+    {
+      "epoch": 0.27795709132333396,
+      "grad_norm": 1.46875,
+      "learning_rate": 8.474078906914359e-06,
+      "loss": 2.5617,
+      "step": 183
+    },
+    {
+      "epoch": 0.2794759825327511,
+      "grad_norm": 1.234375,
+      "learning_rate": 8.456330003359093e-06,
+      "loss": 2.4827,
+      "step": 184
+    },
+    {
+      "epoch": 0.2809948737421682,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.438497294267117e-06,
+      "loss": 2.5726,
+      "step": 185
+    },
+    {
+      "epoch": 0.28251376495158537,
+      "grad_norm": 1.453125,
+      "learning_rate": 8.420581212027625e-06,
+      "loss": 2.4644,
+      "step": 186
+    },
+    {
+      "epoch": 0.2840326561610025,
+      "grad_norm": 1.25,
+      "learning_rate": 8.402582191051365e-06,
+      "loss": 2.4287,
+      "step": 187
+    },
+    {
+      "epoch": 0.2855515473704196,
+      "grad_norm": 1.640625,
+      "learning_rate": 8.38450066776009e-06,
+      "loss": 2.5467,
+      "step": 188
+    },
+    {
+      "epoch": 0.2870704385798367,
+      "grad_norm": 1.359375,
+      "learning_rate": 8.36633708057599e-06,
+      "loss": 2.4555,
+      "step": 189
+    },
+    {
+      "epoch": 0.28858932978925383,
+      "grad_norm": 1.3203125,
+      "learning_rate": 8.348091869911054e-06,
+      "loss": 2.5255,
+      "step": 190
+    },
+    {
+      "epoch": 0.29010822099867095,
+      "grad_norm": 1.34375,
+      "learning_rate": 8.329765478156394e-06,
+      "loss": 2.5443,
+      "step": 191
+    },
+    {
+      "epoch": 0.29162711220808807,
+      "grad_norm": 1.2890625,
+      "learning_rate": 8.311358349671516e-06,
+      "loss": 2.5705,
+      "step": 192
+    },
+    {
+      "epoch": 0.29314600341750524,
+      "grad_norm": 1.203125,
+      "learning_rate": 8.292870930773551e-06,
+      "loss": 2.4817,
+      "step": 193
+    },
+    {
+      "epoch": 0.29466489462692236,
+      "grad_norm": 1.3671875,
+      "learning_rate": 8.274303669726427e-06,
+      "loss": 2.5036,
+      "step": 194
+    },
+    {
+      "epoch": 0.2961837858363395,
+      "grad_norm": 1.1484375,
+      "learning_rate": 8.255657016729997e-06,
+      "loss": 2.5086,
+      "step": 195
+    },
+    {
+      "epoch": 0.2977026770457566,
+      "grad_norm": 1.078125,
+      "learning_rate": 8.23693142390914e-06,
+      "loss": 2.4332,
+      "step": 196
+    },
+    {
+      "epoch": 0.2992215682551737,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.218127345302775e-06,
+      "loss": 2.4879,
+      "step": 197
+    },
+    {
+      "epoch": 0.3007404594645908,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.199245236852871e-06,
+      "loss": 2.5324,
+      "step": 198
+    },
+    {
+      "epoch": 0.302259350674008,
+      "grad_norm": 1.328125,
+      "learning_rate": 8.180285556393384e-06,
+      "loss": 2.4826,
+      "step": 199
+    },
+    {
+      "epoch": 0.3037782418834251,
+      "grad_norm": 1.109375,
+      "learning_rate": 8.161248763639154e-06,
+      "loss": 2.4943,
+      "step": 200
+    },
+    {
+      "epoch": 0.30529713309284223,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.142135320174758e-06,
+      "loss": 2.4246,
+      "step": 201
+    },
+    {
+      "epoch": 0.30681602430225935,
+      "grad_norm": 1.296875,
+      "learning_rate": 8.122945689443328e-06,
+      "loss": 2.4313,
+      "step": 202
+    },
+    {
+      "epoch": 0.30833491551167647,
+      "grad_norm": 1.09375,
+      "learning_rate": 8.1036803367353e-06,
+      "loss": 2.559,
+      "step": 203
+    },
+    {
+      "epoch": 0.3098538067210936,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.084339729177142e-06,
+      "loss": 2.514,
+      "step": 204
+    },
+    {
+      "epoch": 0.3113726979305107,
+      "grad_norm": 1.2578125,
+      "learning_rate": 8.064924335720023e-06,
+      "loss": 2.4889,
+      "step": 205
+    },
+    {
+      "epoch": 0.3128915891399279,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.045434627128446e-06,
+      "loss": 2.4603,
+      "step": 206
+    },
+    {
+      "epoch": 0.314410480349345,
+      "grad_norm": 1.2734375,
+      "learning_rate": 8.025871075968828e-06,
+      "loss": 2.4971,
+      "step": 207
+    },
+    {
+      "epoch": 0.3159293715587621,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.006234156598043e-06,
+      "loss": 2.5571,
+      "step": 208
+    },
+    {
+      "epoch": 0.3174482627681792,
+      "grad_norm": 1.0625,
+      "learning_rate": 7.986524345151924e-06,
+      "loss": 2.5384,
+      "step": 209
+    },
+    {
+      "epoch": 0.31896715397759634,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.966742119533724e-06,
+      "loss": 2.4885,
+      "step": 210
+    },
+    {
+      "epoch": 0.32048604518701346,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.946887959402504e-06,
+      "loss": 2.4381,
+      "step": 211
+    },
+    {
+      "epoch": 0.32200493639643063,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.926962346161535e-06,
+      "loss": 2.5199,
+      "step": 212
+    },
+    {
+      "epoch": 0.32352382760584775,
+      "grad_norm": 1.078125,
+      "learning_rate": 7.9069657629466e-06,
+      "loss": 2.5338,
+      "step": 213
+    },
+    {
+      "epoch": 0.32504271881526486,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.886898694614292e-06,
+      "loss": 2.4292,
+      "step": 214
+    },
+    {
+      "epoch": 0.326561610024682,
+      "grad_norm": 1.2890625,
+      "learning_rate": 7.866761627730253e-06,
+      "loss": 2.4925,
+      "step": 215
+    },
+    {
+      "epoch": 0.3280805012340991,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.846555050557381e-06,
+      "loss": 2.4482,
+      "step": 216
+    },
+    {
+      "epoch": 0.3295993924435162,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.826279453043985e-06,
+      "loss": 2.5279,
+      "step": 217
+    },
+    {
+      "epoch": 0.33111828365293333,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.805935326811913e-06,
+      "loss": 2.4002,
+      "step": 218
+    },
+    {
+      "epoch": 0.3326371748623505,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.78552316514462e-06,
+      "loss": 2.3698,
+      "step": 219
+    },
+    {
+      "epoch": 0.3341560660717676,
+      "grad_norm": 1.0859375,
+      "learning_rate": 7.765043462975217e-06,
+      "loss": 2.4626,
+      "step": 220
+    },
+    {
+      "epoch": 0.33567495728118474,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.744496716874472e-06,
+      "loss": 2.5538,
+      "step": 221
+    },
+    {
+      "epoch": 0.33719384849060186,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.723883425038759e-06,
+      "loss": 2.4648,
+      "step": 222
+    },
+    {
+      "epoch": 0.33871273970001897,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.703204087277989e-06,
+      "loss": 2.3859,
+      "step": 223
+    },
+    {
+      "epoch": 0.3402316309094361,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.682459205003484e-06,
+      "loss": 2.4553,
+      "step": 224
+    },
+    {
+      "epoch": 0.34175052211885326,
+      "grad_norm": 1.2578125,
+      "learning_rate": 7.661649281215823e-06,
+      "loss": 2.5044,
+      "step": 225
+    },
+    {
+      "epoch": 0.3432694133282704,
+      "grad_norm": 1.3828125,
+      "learning_rate": 7.640774820492647e-06,
+      "loss": 2.4061,
+      "step": 226
+    },
+    {
+      "epoch": 0.3447883045376875,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.619836328976416e-06,
+      "loss": 2.4957,
+      "step": 227
+    },
+    {
+      "epoch": 0.3463071957471046,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.598834314362151e-06,
+      "loss": 2.383,
+      "step": 228
+    },
+    {
+      "epoch": 0.34782608695652173,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.57776928588511e-06,
+      "loss": 2.4998,
+      "step": 229
+    },
+    {
+      "epoch": 0.34934497816593885,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.556641754308447e-06,
+      "loss": 2.4103,
+      "step": 230
+    },
+    {
+      "epoch": 0.35086386937535596,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.535452231910829e-06,
+      "loss": 2.4448,
+      "step": 231
+    },
+    {
+      "epoch": 0.35238276058477314,
+      "grad_norm": 1.2421875,
+      "learning_rate": 7.514201232474012e-06,
+      "loss": 2.4017,
+      "step": 232
+    },
+    {
+      "epoch": 0.35390165179419025,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.492889271270382e-06,
+      "loss": 2.4629,
+      "step": 233
+    },
+    {
+      "epoch": 0.35542054300360737,
+      "grad_norm": 1.296875,
+      "learning_rate": 7.471516865050468e-06,
+      "loss": 2.4281,
+      "step": 234
+    },
+    {
+      "epoch": 0.3569394342130245,
+      "grad_norm": 1.125,
+      "learning_rate": 7.450084532030402e-06,
+      "loss": 2.4231,
+      "step": 235
+    },
+    {
+      "epoch": 0.3584583254224416,
+      "grad_norm": 1.0,
+      "learning_rate": 7.428592791879361e-06,
+      "loss": 2.4429,
+      "step": 236
+    },
+    {
+      "epoch": 0.3599772166318587,
+      "grad_norm": 1.015625,
+      "learning_rate": 7.407042165706969e-06,
+      "loss": 2.469,
+      "step": 237
+    },
+    {
+      "epoch": 0.3614961078412759,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.385433176050654e-06,
+      "loss": 2.5211,
+      "step": 238
+    },
+    {
+      "epoch": 0.363014999050693,
+      "grad_norm": 1.2109375,
+      "learning_rate": 7.36376634686298e-06,
+      "loss": 2.4338,
+      "step": 239
+    },
+    {
+      "epoch": 0.3645338902601101,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.342042203498952e-06,
+      "loss": 2.5161,
+      "step": 240
+    },
+    {
+      "epoch": 0.36605278146952724,
+      "grad_norm": 1.171875,
+      "learning_rate": 7.320261272703259e-06,
+      "loss": 2.4728,
+      "step": 241
+    },
+    {
+      "epoch": 0.36757167267894436,
+      "grad_norm": 1.203125,
+      "learning_rate": 7.298424082597526e-06,
+      "loss": 2.4609,
+      "step": 242
+    },
+    {
+      "epoch": 0.3690905638883615,
+      "grad_norm": 1.21875,
+      "learning_rate": 7.276531162667484e-06,
+      "loss": 2.4977,
+      "step": 243
+    },
+    {
+      "epoch": 0.3706094550977786,
+      "grad_norm": 1.203125,
+      "learning_rate": 7.254583043750152e-06,
+      "loss": 2.4324,
+      "step": 244
+    },
+    {
+      "epoch": 0.37212834630719577,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.232580258020952e-06,
+      "loss": 2.4407,
+      "step": 245
+    },
+    {
+      "epoch": 0.3736472375166129,
+      "grad_norm": 0.91796875,
+      "learning_rate": 7.210523338980814e-06,
+      "loss": 2.5021,
+      "step": 246
+    },
+    {
+      "epoch": 0.37516612872603,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.1884128214432366e-06,
+      "loss": 2.4268,
+      "step": 247
+    },
+    {
+      "epoch": 0.3766850199354471,
+      "grad_norm": 1.0390625,
+      "learning_rate": 7.1662492415213194e-06,
+      "loss": 2.4063,
+      "step": 248
+    },
+    {
+      "epoch": 0.37820391114486424,
+      "grad_norm": 1.0078125,
+      "learning_rate": 7.14403313661476e-06,
+      "loss": 2.4253,
+      "step": 249
+    },
+    {
+      "epoch": 0.37972280235428135,
+      "grad_norm": 1.265625,
+      "learning_rate": 7.1217650453968335e-06,
+      "loss": 2.461,
+      "step": 250
+    },
+    {
+      "epoch": 0.3812416935636985,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.099445507801324e-06,
+      "loss": 2.5152,
+      "step": 251
+    },
+    {
+      "epoch": 0.38276058477311564,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.0770750650094335e-06,
+      "loss": 2.4101,
+      "step": 252
+    },
+    {
+      "epoch": 0.38427947598253276,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.0546542594366605e-06,
+      "loss": 2.5015,
+      "step": 253
+    },
+    {
+      "epoch": 0.3857983671919499,
+      "grad_norm": 0.96875,
+      "learning_rate": 7.03218363471965e-06,
+      "loss": 2.4964,
+      "step": 254
+    },
+    {
+      "epoch": 0.387317258401367,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.0096637357030105e-06,
+      "loss": 2.4427,
+      "step": 255
+    },
+    {
+      "epoch": 0.3888361496107841,
+      "grad_norm": 1.390625,
+      "learning_rate": 6.987095108426102e-06,
+      "loss": 2.3514,
+      "step": 256
+    },
+    {
+      "epoch": 0.3903550408202012,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.964478300109796e-06,
+      "loss": 2.4921,
+      "step": 257
+    },
+    {
+      "epoch": 0.3918739320296184,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.94181385914321e-06,
+      "loss": 2.4503,
+      "step": 258
+    },
+    {
+      "epoch": 0.3933928232390355,
+      "grad_norm": 0.9375,
+      "learning_rate": 6.91910233507041e-06,
+      "loss": 2.3622,
+      "step": 259
+    },
+    {
+      "epoch": 0.39491171444845263,
+      "grad_norm": 1.0703125,
+      "learning_rate": 6.896344278577083e-06,
+      "loss": 2.4076,
+      "step": 260
+    },
+    {
+      "epoch": 0.39643060565786975,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.873540241477189e-06,
+      "loss": 2.4053,
+      "step": 261
+    },
+    {
+      "epoch": 0.39794949686728687,
+      "grad_norm": 0.90625,
+      "learning_rate": 6.850690776699574e-06,
+      "loss": 2.4828,
+      "step": 262
+    },
+    {
+      "epoch": 0.399468388076704,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.8277964382745675e-06,
+      "loss": 2.4118,
+      "step": 263
+    },
+    {
+      "epoch": 0.40098727928612116,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.804857781320558e-06,
+      "loss": 2.4467,
+      "step": 264
+    },
+    {
+      "epoch": 0.4025061704955383,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.781875362030512e-06,
+      "loss": 2.5718,
+      "step": 265
+    },
+    {
+      "epoch": 0.4040250617049554,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.758849737658508e-06,
+      "loss": 2.5041,
+      "step": 266
+    },
+    {
+      "epoch": 0.4055439529143725,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.735781466506216e-06,
+      "loss": 2.4383,
+      "step": 267
+    },
+    {
+      "epoch": 0.4070628441237896,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.712671107909359e-06,
+      "loss": 2.478,
+      "step": 268
+    },
+    {
+      "epoch": 0.40858173533320674,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.6895192222241534e-06,
+      "loss": 2.4207,
+      "step": 269
+    },
+    {
+      "epoch": 0.41010062654262386,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.666326370813722e-06,
+      "loss": 2.4511,
+      "step": 270
+    },
+    {
+      "epoch": 0.41161951775204103,
+      "grad_norm": 1.0078125,
+      "learning_rate": 6.643093116034486e-06,
+      "loss": 2.4288,
+      "step": 271
+    },
+    {
+      "epoch": 0.41313840896145815,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.619820021222518e-06,
+      "loss": 2.437,
+      "step": 272
+    },
+    {
+      "epoch": 0.41465730017087526,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.5965076506799e-06,
+      "loss": 2.4671,
+      "step": 273
+    },
+    {
+      "epoch": 0.4161761913802924,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.573156569661026e-06,
+      "loss": 2.5018,
+      "step": 274
+    },
+    {
+      "epoch": 0.4176950825897095,
+      "grad_norm": 0.890625,
+      "learning_rate": 6.549767344358903e-06,
+      "loss": 2.3747,
+      "step": 275
+    },
+    {
+      "epoch": 0.4192139737991266,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.526340541891418e-06,
+      "loss": 2.4567,
+      "step": 276
+    },
+    {
+      "epoch": 0.4207328650085438,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.5028767302875974e-06,
+      "loss": 2.4371,
+      "step": 277
+    },
+    {
+      "epoch": 0.4222517562179609,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.479376478473822e-06,
+      "loss": 2.3193,
+      "step": 278
+    },
+    {
+      "epoch": 0.423770647427378,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.455840356260041e-06,
+      "loss": 2.4587,
+      "step": 279
+    },
+    {
+      "epoch": 0.42528953863679514,
+      "grad_norm": 1.1484375,
+      "learning_rate": 6.432268934325947e-06,
+      "loss": 2.4221,
+      "step": 280
+    },
+    {
+      "epoch": 0.42680842984621226,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.408662784207149e-06,
+      "loss": 2.4534,
+      "step": 281
+    },
+    {
+      "epoch": 0.4283273210556294,
+      "grad_norm": 1.1640625,
+      "learning_rate": 6.385022478281307e-06,
+      "loss": 2.4384,
+      "step": 282
+    },
+    {
+      "epoch": 0.4298462122650465,
+      "grad_norm": 1.15625,
+      "learning_rate": 6.361348589754255e-06,
+      "loss": 2.4464,
+      "step": 283
+    },
+    {
+      "epoch": 0.43136510347446366,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.337641692646106e-06,
+      "loss": 2.4387,
+      "step": 284
+    },
+    {
+      "epoch": 0.4328839946838808,
+      "grad_norm": 1.2734375,
+      "learning_rate": 6.313902361777327e-06,
+      "loss": 2.5006,
+      "step": 285
+    },
+    {
+      "epoch": 0.4344028858932979,
+      "grad_norm": 1.359375,
+      "learning_rate": 6.290131172754811e-06,
+      "loss": 2.5047,
+      "step": 286
+    },
+    {
+      "epoch": 0.435921777102715,
+      "grad_norm": 0.94140625,
+      "learning_rate": 6.266328701957911e-06,
+      "loss": 2.3543,
+      "step": 287
+    },
+    {
+      "epoch": 0.43744066831213213,
+      "grad_norm": 1.171875,
+      "learning_rate": 6.24249552652447e-06,
+      "loss": 2.5089,
+      "step": 288
+    },
+    {
+      "epoch": 0.43895955952154925,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.2186322243368236e-06,
+      "loss": 2.4545,
+      "step": 289
+    },
+    {
+      "epoch": 0.4404784507309664,
+      "grad_norm": 0.96484375,
+      "learning_rate": 6.194739374007792e-06,
+      "loss": 2.4691,
+      "step": 290
+    },
+    {
+      "epoch": 0.44199734194038354,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.170817554866646e-06,
+      "loss": 2.5019,
+      "step": 291
+    },
+    {
+      "epoch": 0.44351623314980065,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.1468673469450655e-06,
+      "loss": 2.4309,
+      "step": 292
+    },
+    {
+      "epoch": 0.44503512435921777,
+      "grad_norm": 0.91796875,
+      "learning_rate": 6.122889330963069e-06,
+      "loss": 2.4174,
+      "step": 293
+    },
+    {
+      "epoch": 0.4465540155686349,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.098884088314938e-06,
+      "loss": 2.4323,
+      "step": 294
+    },
+    {
+      "epoch": 0.448072906778052,
+      "grad_norm": 1.1171875,
+      "learning_rate": 6.074852201055121e-06,
+      "loss": 2.4597,
+      "step": 295
+    },
+    {
+      "epoch": 0.4495917979874691,
+      "grad_norm": 1.046875,
+      "learning_rate": 6.050794251884112e-06,
+      "loss": 2.4288,
+      "step": 296
+    },
+    {
+      "epoch": 0.4511106891968863,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.026710824134331e-06,
+      "loss": 2.4338,
+      "step": 297
+    },
+    {
+      "epoch": 0.4526295804063034,
+      "grad_norm": 0.94140625,
+      "learning_rate": 6.002602501755974e-06,
+      "loss": 2.4154,
+      "step": 298
+    },
+    {
+      "epoch": 0.45414847161572053,
+      "grad_norm": 0.9765625,
+      "learning_rate": 5.978469869302861e-06,
+      "loss": 2.4921,
+      "step": 299
+    },
+    {
+      "epoch": 0.45566736282513765,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.954313511918252e-06,
+      "loss": 2.4616,
+      "step": 300
+    },
+    {
+      "epoch": 0.45718625403455476,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5.9301340153206685e-06,
+      "loss": 2.4872,
+      "step": 301
+    },
+    {
+      "epoch": 0.4587051452439719,
+      "grad_norm": 1.0390625,
+      "learning_rate": 5.905931965789688e-06,
+      "loss": 2.4421,
+      "step": 302
+    },
+    {
+      "epoch": 0.46022403645338905,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.881707950151725e-06,
+      "loss": 2.4822,
+      "step": 303
+    },
+    {
+      "epoch": 0.46174292766280617,
+      "grad_norm": 1.140625,
+      "learning_rate": 5.857462555765809e-06,
+      "loss": 2.4562,
+      "step": 304
+    },
+    {
+      "epoch": 0.4632618188722233,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.8331963705093375e-06,
+      "loss": 2.5125,
+      "step": 305
+    },
+    {
+      "epoch": 0.4647807100816404,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.808909982763825e-06,
+      "loss": 2.4263,
+      "step": 306
+    },
+    {
+      "epoch": 0.4662996012910575,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.784603981400632e-06,
+      "loss": 2.4522,
+      "step": 307
+    },
+    {
+      "epoch": 0.46781849250047464,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.760278955766695e-06,
+      "loss": 2.4475,
+      "step": 308
+    },
+    {
+      "epoch": 0.46933738370989175,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5.735935495670229e-06,
+      "loss": 2.4194,
+      "step": 309
+    },
+    {
+      "epoch": 0.4708562749193089,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.711574191366427e-06,
+      "loss": 2.3649,
+      "step": 310
+    },
+    {
+      "epoch": 0.47237516612872604,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.687195633543151e-06,
+      "loss": 2.4821,
+      "step": 311
+    },
+    {
+      "epoch": 0.47389405733814316,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5.662800413306611e-06,
+      "loss": 2.475,
+      "step": 312
+    },
+    {
+      "epoch": 0.4754129485475603,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.6383891221670275e-06,
+      "loss": 2.3078,
+      "step": 313
+    },
+    {
+      "epoch": 0.4769318397569774,
+      "grad_norm": 1.1171875,
+      "learning_rate": 5.613962352024293e-06,
+      "loss": 2.3925,
+      "step": 314
+    },
+    {
+      "epoch": 0.4784507309663945,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.589520695153618e-06,
+      "loss": 2.4178,
+      "step": 315
+    },
+    {
+      "epoch": 0.4799696221758117,
+      "grad_norm": 0.890625,
+      "learning_rate": 5.5650647441911706e-06,
+      "loss": 2.5059,
+      "step": 316
+    },
+    {
+      "epoch": 0.4814885133852288,
+      "grad_norm": 1.078125,
+      "learning_rate": 5.540595092119709e-06,
+      "loss": 2.5159,
+      "step": 317
+    },
+    {
+      "epoch": 0.4830074045946459,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.516112332254203e-06,
+      "loss": 2.334,
+      "step": 318
+    },
+    {
+      "epoch": 0.48452629580406303,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.491617058227443e-06,
+      "loss": 2.4316,
+      "step": 319
+    },
+    {
+      "epoch": 0.48604518701348015,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.46710986397565e-06,
+      "loss": 2.5104,
+      "step": 320
+    },
+    {
+      "epoch": 0.48756407822289727,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.442591343724081e-06,
+      "loss": 2.4709,
+      "step": 321
+    },
+    {
+      "epoch": 0.4890829694323144,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5.418062091972604e-06,
+      "loss": 2.4407,
+      "step": 322
+    },
+    {
+      "epoch": 0.49060186064173156,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5.393522703481303e-06,
+      "loss": 2.3959,
+      "step": 323
+    },
+    {
+      "epoch": 0.4921207518511487,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.36897377325604e-06,
+      "loss": 2.4172,
+      "step": 324
+    },
+    {
+      "epoch": 0.4936396430605658,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.344415896534039e-06,
+      "loss": 2.3965,
+      "step": 325
+    },
+    {
+      "epoch": 0.4951585342699829,
+      "grad_norm": 0.90625,
+      "learning_rate": 5.319849668769449e-06,
+      "loss": 2.4589,
+      "step": 326
+    },
+    {
+      "epoch": 0.4966774254794,
+      "grad_norm": 0.9375,
+      "learning_rate": 5.295275685618905e-06,
+      "loss": 2.3558,
+      "step": 327
+    },
+    {
+      "epoch": 0.49819631668881714,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.270694542927089e-06,
+      "loss": 2.5089,
+      "step": 328
+    },
+    {
+      "epoch": 0.4997152078982343,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.246106836712277e-06,
+      "loss": 2.4541,
+      "step": 329
+    },
+    {
+      "epoch": 0.5012340991076514,
+      "grad_norm": 0.91796875,
+      "learning_rate": 5.2215131631518945e-06,
+      "loss": 2.3858,
+      "step": 330
+    },
+    {
+      "epoch": 0.5027529903170685,
+      "grad_norm": 1.0078125,
+      "learning_rate": 5.196914118568054e-06,
+      "loss": 2.4189,
+      "step": 331
+    },
+    {
+      "epoch": 0.5042718815264856,
+      "grad_norm": 1.03125,
+      "learning_rate": 5.1723102994130994e-06,
+      "loss": 2.4426,
+      "step": 332
+    },
+    {
+      "epoch": 0.5057907727359028,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.147702302255143e-06,
+      "loss": 2.4418,
+      "step": 333
+    },
+    {
+      "epoch": 0.50730966394532,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.123090723763607e-06,
+      "loss": 2.3768,
+      "step": 334
+    },
+    {
+      "epoch": 0.508828555154737,
+      "grad_norm": 0.9296875,
+      "learning_rate": 5.098476160694741e-06,
+      "loss": 2.4267,
+      "step": 335
+    },
+    {
+      "epoch": 0.5103474463641542,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.073859209877167e-06,
+      "loss": 2.4311,
+      "step": 336
+    },
+    {
+      "epoch": 0.5118663375735713,
+      "grad_norm": 0.99609375,
+      "learning_rate": 5.049240468197401e-06,
+      "loss": 2.5122,
+      "step": 337
+    },
+    {
+      "epoch": 0.5133852287829884,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5.0246205325853824e-06,
+      "loss": 2.384,
+      "step": 338
+    },
+    {
+      "epoch": 0.5149041199924056,
+      "grad_norm": 0.9375,
+      "learning_rate": 5e-06,
+      "loss": 2.4921,
+      "step": 339
+    },
+    {
+      "epoch": 0.5164230112018227,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.975379467414621e-06,
+      "loss": 2.4624,
+      "step": 340
+    },
+    {
+      "epoch": 0.5179419024112398,
+      "grad_norm": 0.92578125,
+      "learning_rate": 4.950759531802602e-06,
+      "loss": 2.402,
+      "step": 341
+    },
+    {
+      "epoch": 0.5194607936206569,
+      "grad_norm": 0.90625,
+      "learning_rate": 4.926140790122835e-06,
+      "loss": 2.4924,
+      "step": 342
+    },
+    {
+      "epoch": 0.5209796848300741,
+      "grad_norm": 0.89453125,
+      "learning_rate": 4.90152383930526e-06,
+      "loss": 2.4247,
+      "step": 343
+    },
+    {
+      "epoch": 0.5224985760394911,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.876909276236395e-06,
+      "loss": 2.3653,
+      "step": 344
+    },
+    {
+      "epoch": 0.5240174672489083,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.852297697744857e-06,
+      "loss": 2.4583,
+      "step": 345
+    },
+    {
+      "epoch": 0.5255363584583255,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.827689700586902e-06,
+      "loss": 2.4109,
+      "step": 346
+    },
+    {
+      "epoch": 0.5270552496677425,
+      "grad_norm": 0.86328125,
+      "learning_rate": 4.803085881431949e-06,
+      "loss": 2.4321,
+      "step": 347
+    },
+    {
+      "epoch": 0.5285741408771597,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.778486836848107e-06,
+      "loss": 2.5462,
+      "step": 348
+    },
+    {
+      "epoch": 0.5300930320865768,
+      "grad_norm": 0.9921875,
+      "learning_rate": 4.7538931632877254e-06,
+      "loss": 2.3965,
+      "step": 349
+    },
+    {
+      "epoch": 0.5316119232959939,
+      "grad_norm": 1.03125,
+      "learning_rate": 4.729305457072913e-06,
+      "loss": 2.4898,
+      "step": 350
+    },
+    {
+      "epoch": 0.533130814505411,
+      "grad_norm": 0.9140625,
+      "learning_rate": 4.704724314381097e-06,
+      "loss": 2.4746,
+      "step": 351
+    },
+    {
+      "epoch": 0.5346497057148282,
+      "grad_norm": 0.875,
+      "learning_rate": 4.680150331230552e-06,
+      "loss": 2.3629,
+      "step": 352
+    },
+    {
+      "epoch": 0.5361685969242453,
+      "grad_norm": 1.0234375,
+      "learning_rate": 4.6555841034659625e-06,
+      "loss": 2.3738,
+      "step": 353
+    },
+    {
+      "epoch": 0.5376874881336624,
+      "grad_norm": 0.9140625,
+      "learning_rate": 4.631026226743962e-06,
+      "loss": 2.4069,
+      "step": 354
+    },
+    {
+      "epoch": 0.5392063793430796,
+      "grad_norm": 0.921875,
+      "learning_rate": 4.606477296518698e-06,
+      "loss": 2.344,
+      "step": 355
+    },
+    {
+      "epoch": 0.5407252705524966,
+      "grad_norm": 0.91015625,
+      "learning_rate": 4.581937908027397e-06,
+      "loss": 2.4081,
+      "step": 356
+    },
+    {
+      "epoch": 0.5422441617619138,
+      "grad_norm": 0.921875,
+      "learning_rate": 4.55740865627592e-06,
+      "loss": 2.4161,
+      "step": 357
+    },
+    {
+      "epoch": 0.5437630529713309,
+      "grad_norm": 0.91796875,
+      "learning_rate": 4.532890136024351e-06,
+      "loss": 2.4184,
+      "step": 358
+    },
+    {
+      "epoch": 0.545281944180748,
+      "grad_norm": 0.93359375,
+      "learning_rate": 4.508382941772558e-06,
+      "loss": 2.4924,
+      "step": 359
+    },
+    {
+      "epoch": 0.5468008353901652,
+      "grad_norm": 0.90234375,
+      "learning_rate": 4.483887667745798e-06,
+      "loss": 2.3855,
+      "step": 360
+    },
+    {
+      "epoch": 0.5483197265995823,
+      "grad_norm": 0.96484375,
+      "learning_rate": 4.459404907880293e-06,
+      "loss": 2.4468,
+      "step": 361
+    },
+    {
+      "epoch": 0.5498386178089995,
+      "grad_norm": 0.953125,
+      "learning_rate": 4.434935255808831e-06,
+      "loss": 2.3802,
+      "step": 362
+    },
+    {
+      "epoch": 0.5513575090184165,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.410479304846385e-06,
+      "loss": 2.35,
+      "step": 363
+    },
+    {
+      "epoch": 0.5528764002278337,
+      "grad_norm": 0.91796875,
+      "learning_rate": 4.386037647975708e-06,
+      "loss": 2.3869,
+      "step": 364
+    },
+    {
+      "epoch": 0.5543952914372509,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.361610877832974e-06,
+      "loss": 2.41,
+      "step": 365
+    },
+    {
+      "epoch": 0.5559141826466679,
+      "grad_norm": 0.9296875,
+      "learning_rate": 4.337199586693389e-06,
+      "loss": 2.3689,
+      "step": 366
+    },
+    {
+      "epoch": 0.5574330738560851,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.312804366456851e-06,
+      "loss": 2.4504,
+      "step": 367
+    },
+    {
+      "epoch": 0.5589519650655022,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.2884258086335755e-06,
+      "loss": 2.4652,
+      "step": 368
+    },
+    {
+      "epoch": 0.5604708562749193,
+      "grad_norm": 0.9140625,
+      "learning_rate": 4.2640645043297715e-06,
+      "loss": 2.4019,
+      "step": 369
+    },
+    {
+      "epoch": 0.5619897474843364,
+      "grad_norm": 0.8828125,
+      "learning_rate": 4.239721044233306e-06,
+      "loss": 2.306,
+      "step": 370
+    },
+    {
+      "epoch": 0.5635086386937536,
+      "grad_norm": 0.85546875,
+      "learning_rate": 4.215396018599369e-06,
+      "loss": 2.3859,
+      "step": 371
+    },
+    {
+      "epoch": 0.5650275299031707,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.191090017236177e-06,
+      "loss": 2.3734,
+      "step": 372
+    },
+    {
+      "epoch": 0.5665464211125878,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.166803629490664e-06,
+      "loss": 2.4194,
+      "step": 373
+    },
+    {
+      "epoch": 0.568065312322005,
+      "grad_norm": 0.93359375,
+      "learning_rate": 4.142537444234192e-06,
+      "loss": 2.4815,
+      "step": 374
+    },
+    {
+      "epoch": 0.569584203531422,
+      "grad_norm": 0.8828125,
+      "learning_rate": 4.118292049848277e-06,
+      "loss": 2.4493,
+      "step": 375
+    },
+    {
+      "epoch": 0.5711030947408392,
+      "grad_norm": 0.8828125,
+      "learning_rate": 4.094068034210313e-06,
+      "loss": 2.4458,
+      "step": 376
+    },
+    {
+      "epoch": 0.5726219859502563,
+      "grad_norm": 0.91796875,
+      "learning_rate": 4.069865984679332e-06,
+      "loss": 2.4004,
+      "step": 377
+    },
+    {
+      "epoch": 0.5741408771596734,
+      "grad_norm": 0.87109375,
+      "learning_rate": 4.045686488081748e-06,
+      "loss": 2.4643,
+      "step": 378
+    },
+    {
+      "epoch": 0.5756597683690906,
+      "grad_norm": 0.91796875,
+      "learning_rate": 4.021530130697141e-06,
+      "loss": 2.4709,
+      "step": 379
+    },
+    {
+      "epoch": 0.5771786595785077,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.997397498244028e-06,
+      "loss": 2.4853,
+      "step": 380
+    },
+    {
+      "epoch": 0.5786975507879248,
+      "grad_norm": 0.9453125,
+      "learning_rate": 3.97328917586567e-06,
+      "loss": 2.5029,
+      "step": 381
+    },
+    {
+      "epoch": 0.5802164419973419,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.9492057481158905e-06,
+      "loss": 2.4853,
+      "step": 382
+    },
+    {
+      "epoch": 0.5817353332067591,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.92514779894488e-06,
+      "loss": 2.4058,
+      "step": 383
+    },
+    {
+      "epoch": 0.5832542244161761,
+      "grad_norm": 1.2578125,
+      "learning_rate": 3.901115911685063e-06,
+      "loss": 2.4426,
+      "step": 384
+    },
+    {
+      "epoch": 0.5847731156255933,
+      "grad_norm": 1.25,
+      "learning_rate": 3.877110669036932e-06,
+      "loss": 2.4151,
+      "step": 385
+    },
+    {
+      "epoch": 0.5862920068350105,
+      "grad_norm": 0.89453125,
+      "learning_rate": 3.853132653054936e-06,
+      "loss": 2.4061,
+      "step": 386
+    },
+    {
+      "epoch": 0.5878108980444275,
+      "grad_norm": 0.89453125,
+      "learning_rate": 3.829182445133356e-06,
+      "loss": 2.4192,
+      "step": 387
+    },
+    {
+      "epoch": 0.5893297892538447,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.8052606259922097e-06,
+      "loss": 2.3988,
+      "step": 388
+    },
+    {
+      "epoch": 0.5908486804632618,
+      "grad_norm": 0.9140625,
+      "learning_rate": 3.7813677756631773e-06,
+      "loss": 2.4392,
+      "step": 389
+    },
+    {
+      "epoch": 0.592367571672679,
+      "grad_norm": 0.94140625,
+      "learning_rate": 3.75750447347553e-06,
+      "loss": 2.3739,
+      "step": 390
+    },
+    {
+      "epoch": 0.5938864628820961,
+      "grad_norm": 0.91796875,
+      "learning_rate": 3.7336712980420897e-06,
+      "loss": 2.4588,
+      "step": 391
+    },
+    {
+      "epoch": 0.5954053540915132,
+      "grad_norm": 0.91015625,
+      "learning_rate": 3.7098688272451893e-06,
+      "loss": 2.4266,
+      "step": 392
+    },
+    {
+      "epoch": 0.5969242453009304,
+      "grad_norm": 0.97265625,
+      "learning_rate": 3.6860976382226747e-06,
+      "loss": 2.3617,
+      "step": 393
+    },
+    {
+      "epoch": 0.5984431365103474,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.662358307353897e-06,
+      "loss": 2.4095,
+      "step": 394
+    },
+    {
+      "epoch": 0.5999620277197646,
+      "grad_norm": 0.921875,
+      "learning_rate": 3.638651410245746e-06,
+      "loss": 2.5,
+      "step": 395
+    },
+    {
+      "epoch": 0.6014809189291817,
+      "grad_norm": 0.96484375,
+      "learning_rate": 3.6149775217186954e-06,
+      "loss": 2.4601,
+      "step": 396
+    },
+    {
+      "epoch": 0.6029998101385988,
+      "grad_norm": 0.89453125,
+      "learning_rate": 3.5913372157928515e-06,
+      "loss": 2.4496,
+      "step": 397
+    },
+    {
+      "epoch": 0.604518701348016,
+      "grad_norm": 0.90234375,
+      "learning_rate": 3.5677310656740537e-06,
+      "loss": 2.367,
+      "step": 398
+    },
+    {
+      "epoch": 0.6060375925574331,
+      "grad_norm": 0.85546875,
+      "learning_rate": 3.5441596437399596e-06,
+      "loss": 2.4182,
+      "step": 399
+    },
+    {
+      "epoch": 0.6075564837668502,
+      "grad_norm": 0.91015625,
+      "learning_rate": 3.5206235215261785e-06,
+      "loss": 2.4485,
+      "step": 400
+    },
+    {
+      "epoch": 0.6090753749762673,
+      "grad_norm": 0.859375,
+      "learning_rate": 3.4971232697124046e-06,
+      "loss": 2.3709,
+      "step": 401
+    },
+    {
+      "epoch": 0.6105942661856845,
+      "grad_norm": 0.9140625,
+      "learning_rate": 3.4736594581085837e-06,
+      "loss": 2.4012,
+      "step": 402
+    },
+    {
+      "epoch": 0.6121131573951015,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.4502326556411e-06,
+      "loss": 2.484,
+      "step": 403
+    },
+    {
+      "epoch": 0.6136320486045187,
+      "grad_norm": 0.9140625,
+      "learning_rate": 3.4268434303389747e-06,
+      "loss": 2.3686,
+      "step": 404
+    },
+    {
+      "epoch": 0.6151509398139359,
+      "grad_norm": 0.9140625,
+      "learning_rate": 3.403492349320101e-06,
+      "loss": 2.4893,
+      "step": 405
+    },
+    {
+      "epoch": 0.6166698310233529,
+      "grad_norm": 0.8828125,
+      "learning_rate": 3.380179978777482e-06,
+      "loss": 2.4911,
+      "step": 406
+    },
+    {
+      "epoch": 0.6181887222327701,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.356906883965516e-06,
+      "loss": 2.4203,
+      "step": 407
+    },
+    {
+      "epoch": 0.6197076134421872,
+      "grad_norm": 0.8671875,
+      "learning_rate": 3.33367362918628e-06,
+      "loss": 2.4613,
+      "step": 408
+    },
+    {
+      "epoch": 0.6212265046516043,
+      "grad_norm": 0.91796875,
+      "learning_rate": 3.3104807777758487e-06,
+      "loss": 2.4544,
+      "step": 409
+    },
+    {
+      "epoch": 0.6227453958610214,
+      "grad_norm": 0.86328125,
+      "learning_rate": 3.2873288920906436e-06,
+      "loss": 2.4338,
+      "step": 410
+    },
+    {
+      "epoch": 0.6242642870704386,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.2642185334937853e-06,
+      "loss": 2.5019,
+      "step": 411
+    },
+    {
+      "epoch": 0.6257831782798557,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.2411502623414925e-06,
+      "loss": 2.4375,
+      "step": 412
+    },
+    {
+      "epoch": 0.6273020694892728,
+      "grad_norm": 0.90625,
+      "learning_rate": 3.2181246379694886e-06,
+      "loss": 2.4859,
+      "step": 413
+    },
+    {
+      "epoch": 0.62882096069869,
+      "grad_norm": 0.94921875,
+      "learning_rate": 3.1951422186794447e-06,
+      "loss": 2.4439,
+      "step": 414
+    },
+    {
+      "epoch": 0.630339851908107,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.1722035617254333e-06,
+      "loss": 2.411,
+      "step": 415
+    },
+    {
+      "epoch": 0.6318587431175242,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.149309223300428e-06,
+      "loss": 2.3643,
+      "step": 416
+    },
+    {
+      "epoch": 0.6333776343269414,
+      "grad_norm": 1.078125,
+      "learning_rate": 3.126459758522813e-06,
+      "loss": 2.4063,
+      "step": 417
+    },
+    {
+      "epoch": 0.6348965255363584,
+      "grad_norm": 1.0390625,
+      "learning_rate": 3.103655721422917e-06,
+      "loss": 2.4542,
+      "step": 418
+    },
+    {
+      "epoch": 0.6364154167457756,
+      "grad_norm": 0.8671875,
+      "learning_rate": 3.080897664929592e-06,
+      "loss": 2.392,
+      "step": 419
+    },
+    {
+      "epoch": 0.6379343079551927,
+      "grad_norm": 0.92578125,
+      "learning_rate": 3.0581861408567907e-06,
+      "loss": 2.4415,
+      "step": 420
+    },
+    {
+      "epoch": 0.6394531991646099,
+      "grad_norm": 0.86328125,
+      "learning_rate": 3.035521699890206e-06,
+      "loss": 2.414,
+      "step": 421
+    },
+    {
+      "epoch": 0.6409720903740269,
+      "grad_norm": 0.8828125,
+      "learning_rate": 3.0129048915739013e-06,
+      "loss": 2.48,
+      "step": 422
+    },
+    {
+      "epoch": 0.6424909815834441,
+      "grad_norm": 1.0546875,
+      "learning_rate": 2.9903362642969903e-06,
+      "loss": 2.4107,
+      "step": 423
+    },
+    {
+      "epoch": 0.6440098727928613,
+      "grad_norm": 4.15625,
+      "learning_rate": 2.967816365280351e-06,
+      "loss": 2.4132,
+      "step": 424
+    },
+    {
+      "epoch": 0.6455287640022783,
+      "grad_norm": 0.9140625,
+      "learning_rate": 2.94534574056334e-06,
+      "loss": 2.4418,
+      "step": 425
+    },
+    {
+      "epoch": 0.6470476552116955,
+      "grad_norm": 0.87890625,
+      "learning_rate": 2.9229249349905686e-06,
+      "loss": 2.4867,
+      "step": 426
+    },
+    {
+      "epoch": 0.6485665464211126,
+      "grad_norm": 0.84375,
+      "learning_rate": 2.9005544921986774e-06,
+      "loss": 2.4427,
+      "step": 427
+    },
+    {
+      "epoch": 0.6500854376305297,
+      "grad_norm": 1.140625,
+      "learning_rate": 2.8782349546031673e-06,
+      "loss": 2.4603,
+      "step": 428
+    },
+    {
+      "epoch": 0.6516043288399468,
+      "grad_norm": 0.84375,
+      "learning_rate": 2.8559668633852433e-06,
+      "loss": 2.382,
+      "step": 429
+    },
+    {
+      "epoch": 0.653123220049364,
+      "grad_norm": 0.8359375,
+      "learning_rate": 2.8337507584786826e-06,
+      "loss": 2.4133,
+      "step": 430
+    },
+    {
+      "epoch": 0.6546421112587811,
+      "grad_norm": 0.8671875,
+      "learning_rate": 2.811587178556764e-06,
+      "loss": 2.4165,
+      "step": 431
+    },
+    {
+      "epoch": 0.6561610024681982,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.789476661019186e-06,
+      "loss": 2.4604,
+      "step": 432
+    },
+    {
+      "epoch": 0.6576798936776154,
+      "grad_norm": 0.93359375,
+      "learning_rate": 2.7674197419790493e-06,
+      "loss": 2.4372,
+      "step": 433
+    },
+    {
+      "epoch": 0.6591987848870324,
+      "grad_norm": 0.87109375,
+      "learning_rate": 2.7454169562498503e-06,
+      "loss": 2.4066,
+      "step": 434
+    },
+    {
+      "epoch": 0.6607176760964496,
+      "grad_norm": 0.91796875,
+      "learning_rate": 2.723468837332517e-06,
+      "loss": 2.42,
+      "step": 435
+    },
+    {
+      "epoch": 0.6622365673058667,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.7015759174024756e-06,
+      "loss": 2.3347,
+      "step": 436
+    },
+    {
+      "epoch": 0.6637554585152838,
+      "grad_norm": 0.87109375,
+      "learning_rate": 2.6797387272967414e-06,
+      "loss": 2.4629,
+      "step": 437
+    },
+    {
+      "epoch": 0.665274349724701,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.65795779650105e-06,
+      "loss": 2.4078,
+      "step": 438
+    },
+    {
+      "epoch": 0.6667932409341181,
+      "grad_norm": 0.8984375,
+      "learning_rate": 2.63623365313702e-06,
+      "loss": 2.4538,
+      "step": 439
+    },
+    {
+      "epoch": 0.6683121321435352,
+      "grad_norm": 0.88671875,
+      "learning_rate": 2.614566823949348e-06,
+      "loss": 2.4317,
+      "step": 440
+    },
+    {
+      "epoch": 0.6698310233529523,
+      "grad_norm": 0.89453125,
+      "learning_rate": 2.592957834293033e-06,
+      "loss": 2.3749,
+      "step": 441
+    },
+    {
+      "epoch": 0.6713499145623695,
+      "grad_norm": 0.88671875,
+      "learning_rate": 2.5714072081206407e-06,
+      "loss": 2.4384,
+      "step": 442
+    },
+    {
+      "epoch": 0.6728688057717866,
+      "grad_norm": 0.8515625,
+      "learning_rate": 2.5499154679696014e-06,
+      "loss": 2.4108,
+      "step": 443
+    },
+    {
+      "epoch": 0.6743876969812037,
+      "grad_norm": 0.8828125,
+      "learning_rate": 2.528483134949535e-06,
+      "loss": 2.4876,
+      "step": 444
+    },
+    {
+      "epoch": 0.6759065881906209,
+      "grad_norm": 0.84375,
+      "learning_rate": 2.50711072872962e-06,
+      "loss": 2.4413,
+      "step": 445
+    },
+    {
+      "epoch": 0.6774254794000379,
+      "grad_norm": 0.87890625,
+      "learning_rate": 2.4857987675259887e-06,
+      "loss": 2.4378,
+      "step": 446
+    },
+    {
+      "epoch": 0.6789443706094551,
+      "grad_norm": 0.90625,
+      "learning_rate": 2.4645477680891734e-06,
+      "loss": 2.4039,
+      "step": 447
+    },
+    {
+      "epoch": 0.6804632618188722,
+      "grad_norm": 0.890625,
+      "learning_rate": 2.4433582456915556e-06,
+      "loss": 2.5087,
+      "step": 448
+    },
+    {
+      "epoch": 0.6819821530282894,
+      "grad_norm": 0.93359375,
+      "learning_rate": 2.422230714114891e-06,
+      "loss": 2.3984,
+      "step": 449
+    },
+    {
+      "epoch": 0.6835010442377065,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.4011656856378513e-06,
+      "loss": 2.3983,
+      "step": 450
+    },
+    {
+      "epoch": 0.6850199354471236,
+      "grad_norm": 0.9296875,
+      "learning_rate": 2.3801636710235836e-06,
+      "loss": 2.4803,
+      "step": 451
+    },
+    {
+      "epoch": 0.6865388266565408,
+      "grad_norm": 0.85546875,
+      "learning_rate": 2.3592251795073564e-06,
+      "loss": 2.4262,
+      "step": 452
+    },
+    {
+      "epoch": 0.6880577178659578,
+      "grad_norm": 0.8828125,
+      "learning_rate": 2.338350718784177e-06,
+      "loss": 2.4503,
+      "step": 453
+    },
+    {
+      "epoch": 0.689576609075375,
+      "grad_norm": 0.87109375,
+      "learning_rate": 2.3175407949965167e-06,
+      "loss": 2.4356,
+      "step": 454
+    },
+    {
+      "epoch": 0.691095500284792,
+      "grad_norm": 0.8671875,
+      "learning_rate": 2.296795912722014e-06,
+      "loss": 2.4435,
+      "step": 455
+    },
+    {
+      "epoch": 0.6926143914942092,
+      "grad_norm": 1.0,
+      "learning_rate": 2.2761165749612417e-06,
+      "loss": 2.3985,
+      "step": 456
+    },
+    {
+      "epoch": 0.6941332827036264,
+      "grad_norm": 0.94921875,
+      "learning_rate": 2.25550328312553e-06,
+      "loss": 2.4073,
+      "step": 457
+    },
+    {
+      "epoch": 0.6956521739130435,
+      "grad_norm": 0.88671875,
+      "learning_rate": 2.2349565370247837e-06,
+      "loss": 2.3838,
+      "step": 458
+    },
+    {
+      "epoch": 0.6971710651224606,
+      "grad_norm": 0.91015625,
+      "learning_rate": 2.214476834855382e-06,
+      "loss": 2.4342,
+      "step": 459
+    },
+    {
+      "epoch": 0.6986899563318777,
+      "grad_norm": 0.84375,
+      "learning_rate": 2.1940646731880887e-06,
+      "loss": 2.4411,
+      "step": 460
+    },
+    {
+      "epoch": 0.7002088475412949,
+      "grad_norm": 0.859375,
+      "learning_rate": 2.173720546956015e-06,
+      "loss": 2.4685,
+      "step": 461
+    },
+    {
+      "epoch": 0.7017277387507119,
+      "grad_norm": 0.9375,
+      "learning_rate": 2.1534449494426203e-06,
+      "loss": 2.4395,
+      "step": 462
+    },
+    {
+      "epoch": 0.7032466299601291,
+      "grad_norm": 0.921875,
+      "learning_rate": 2.1332383722697483e-06,
+      "loss": 2.3932,
+      "step": 463
+    },
+    {
+      "epoch": 0.7047655211695463,
+      "grad_norm": 0.9375,
+      "learning_rate": 2.1131013053857097e-06,
+      "loss": 2.4367,
+      "step": 464
+    },
+    {
+      "epoch": 0.7062844123789633,
+      "grad_norm": 1.15625,
+      "learning_rate": 2.0930342370534013e-06,
+      "loss": 2.4825,
+      "step": 465
+    },
+    {
+      "epoch": 0.7078033035883805,
+      "grad_norm": 0.87890625,
+      "learning_rate": 2.073037653838466e-06,
+      "loss": 2.4702,
+      "step": 466
+    },
+    {
+      "epoch": 0.7093221947977976,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.053112040597495e-06,
+      "loss": 2.3709,
+      "step": 467
+    },
+    {
+      "epoch": 0.7108410860072147,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.0332578804662783e-06,
+      "loss": 2.3477,
+      "step": 468
+    },
+    {
+      "epoch": 0.7123599772166319,
+      "grad_norm": 0.859375,
+      "learning_rate": 2.013475654848076e-06,
+      "loss": 2.3872,
+      "step": 469
+    },
+    {
+      "epoch": 0.713878868426049,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.99376584340196e-06,
+      "loss": 2.4375,
+      "step": 470
+    },
+    {
+      "epoch": 0.7153977596354661,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.9741289240311757e-06,
+      "loss": 2.4044,
+      "step": 471
+    },
+    {
+      "epoch": 0.7169166508448832,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.954565372871554e-06,
+      "loss": 2.3834,
+      "step": 472
+    },
+    {
+      "epoch": 0.7184355420543004,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.935075664279978e-06,
+      "loss": 2.3561,
+      "step": 473
+    },
+    {
+      "epoch": 0.7199544332637174,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.9156602708228584e-06,
+      "loss": 2.4273,
+      "step": 474
+    },
+    {
+      "epoch": 0.7214733244731346,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.8963196632647008e-06,
+      "loss": 2.4413,
+      "step": 475
+    },
+    {
+      "epoch": 0.7229922156825518,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.8770543105566752e-06,
+      "loss": 2.4814,
+      "step": 476
+    },
+    {
+      "epoch": 0.7245111068919688,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.8578646798252432e-06,
+      "loss": 2.3857,
+      "step": 477
+    },
+    {
+      "epoch": 0.726029998101386,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.8387512363608496e-06,
+      "loss": 2.4597,
+      "step": 478
+    },
+    {
+      "epoch": 0.7275488893108031,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.8197144436066167e-06,
+      "loss": 2.4199,
+      "step": 479
+    },
+    {
+      "epoch": 0.7290677805202203,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.8007547631471289e-06,
+      "loss": 2.4812,
+      "step": 480
+    },
+    {
+      "epoch": 0.7305866717296373,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.781872654697226e-06,
+      "loss": 2.3901,
+      "step": 481
+    },
+    {
+      "epoch": 0.7321055629390545,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.7630685760908623e-06,
+      "loss": 2.4603,
+      "step": 482
+    },
+    {
+      "epoch": 0.7336244541484717,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.7443429832700038e-06,
+      "loss": 2.4276,
+      "step": 483
+    },
+    {
+      "epoch": 0.7351433453578887,
+      "grad_norm": 0.8984375,
+      "learning_rate": 1.7256963302735752e-06,
+      "loss": 2.4272,
+      "step": 484
+    },
+    {
+      "epoch": 0.7366622365673059,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.7071290692264492e-06,
+      "loss": 2.3496,
+      "step": 485
+    },
+    {
+      "epoch": 0.738181127776723,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.6886416503284835e-06,
+      "loss": 2.4147,
+      "step": 486
+    },
+    {
+      "epoch": 0.7397000189861401,
+      "grad_norm": 0.8671875,
+      "learning_rate": 1.6702345218436066e-06,
+      "loss": 2.4918,
+      "step": 487
+    },
+    {
+      "epoch": 0.7412189101955572,
+      "grad_norm": 0.8984375,
+      "learning_rate": 1.6519081300889472e-06,
+      "loss": 2.4301,
+      "step": 488
+    },
+    {
+      "epoch": 0.7427378014049744,
+      "grad_norm": 0.8671875,
+      "learning_rate": 1.6336629194240118e-06,
+      "loss": 2.4587,
+      "step": 489
+    },
+    {
+      "epoch": 0.7442566926143915,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.6154993322399114e-06,
+      "loss": 2.386,
+      "step": 490
+    },
+    {
+      "epoch": 0.7457755838238086,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.5974178089486364e-06,
+      "loss": 2.4635,
+      "step": 491
+    },
+    {
+      "epoch": 0.7472944750332258,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.5794187879723755e-06,
+      "loss": 2.3878,
+      "step": 492
+    },
+    {
+      "epoch": 0.7488133662426428,
+      "grad_norm": 0.87890625,
+      "learning_rate": 1.561502705732883e-06,
+      "loss": 2.4565,
+      "step": 493
+    },
+    {
+      "epoch": 0.75033225745206,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.543669996640908e-06,
+      "loss": 2.3761,
+      "step": 494
+    },
+    {
+      "epoch": 0.7518511486614772,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.5259210930856423e-06,
+      "loss": 2.4103,
+      "step": 495
+    },
+    {
+      "epoch": 0.7533700398708942,
+      "grad_norm": 0.875,
+      "learning_rate": 1.5082564254242583e-06,
+      "loss": 2.4664,
+      "step": 496
+    },
+    {
+      "epoch": 0.7548889310803114,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.4906764219714537e-06,
+      "loss": 2.3815,
+      "step": 497
+    },
+    {
+      "epoch": 0.7564078222897285,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.4731815089890795e-06,
+      "loss": 2.3936,
+      "step": 498
+    },
+    {
+      "epoch": 0.7579267134991456,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.455772110675804e-06,
+      "loss": 2.4351,
+      "step": 499
+    },
+    {
+      "epoch": 0.7594456047085627,
+      "grad_norm": 0.92578125,
+      "learning_rate": 1.438448649156815e-06,
+      "loss": 2.4531,
+      "step": 500
+    },
+    {
+      "epoch": 0.7609644959179799,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.4212115444736024e-06,
+      "loss": 2.3204,
+      "step": 501
+    },
+    {
+      "epoch": 0.762483387127397,
+      "grad_norm": 0.8671875,
+      "learning_rate": 1.4040612145737608e-06,
+      "loss": 2.4785,
+      "step": 502
+    },
+    {
+      "epoch": 0.7640022783368141,
+      "grad_norm": 0.8984375,
+      "learning_rate": 1.3869980753008537e-06,
+      "loss": 2.4757,
+      "step": 503
+    },
+    {
+      "epoch": 0.7655211695462313,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.370022540384347e-06,
+      "loss": 2.3726,
+      "step": 504
+    },
+    {
+      "epoch": 0.7670400607556483,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.353135021429554e-06,
+      "loss": 2.4188,
+      "step": 505
+    },
+    {
+      "epoch": 0.7685589519650655,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.3363359279076776e-06,
+      "loss": 2.3275,
+      "step": 506
+    },
+    {
+      "epoch": 0.7700778431744826,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.3196256671458663e-06,
+      "loss": 2.4475,
+      "step": 507
+    },
+    {
+      "epoch": 0.7715967343838998,
+      "grad_norm": 0.875,
+      "learning_rate": 1.3030046443173445e-06,
+      "loss": 2.3754,
+      "step": 508
+    },
+    {
+      "epoch": 0.7731156255933169,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.2864732624315867e-06,
+      "loss": 2.4502,
+      "step": 509
+    },
+    {
+      "epoch": 0.774634516802734,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.270031922324546e-06,
+      "loss": 2.4717,
+      "step": 510
+    },
+    {
+      "epoch": 0.7761534080121512,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.2536810226489354e-06,
+      "loss": 2.4426,
+      "step": 511
+    },
+    {
+      "epoch": 0.7776722992215682,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.237420959864561e-06,
+      "loss": 2.3612,
+      "step": 512
+    },
+    {
+      "epoch": 0.7791911904309854,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.2212521282287093e-06,
+      "loss": 2.4509,
+      "step": 513
+    },
+    {
+      "epoch": 0.7807100816404025,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.2051749197865875e-06,
+      "loss": 2.4396,
+      "step": 514
+    },
+    {
+      "epoch": 0.7822289728498196,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1891897243618184e-06,
+      "loss": 2.4232,
+      "step": 515
+    },
+    {
+      "epoch": 0.7837478640592368,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.173296929546987e-06,
+      "loss": 2.4108,
+      "step": 516
+    },
+    {
+      "epoch": 0.7852667552686539,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.1574969206942443e-06,
+      "loss": 2.4057,
+      "step": 517
+    },
+    {
+      "epoch": 0.786785646478071,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.1417900809059623e-06,
+      "loss": 2.4187,
+      "step": 518
+    },
+    {
+      "epoch": 0.7883045376874881,
+      "grad_norm": 0.87890625,
+      "learning_rate": 1.1261767910254422e-06,
+      "loss": 2.4451,
+      "step": 519
+    },
+    {
+      "epoch": 0.7898234288969053,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.1106574296276923e-06,
+      "loss": 2.3739,
+      "step": 520
+    },
+    {
+      "epoch": 0.7913423201063224,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.095232373010226e-06,
+      "loss": 2.4306,
+      "step": 521
+    },
+    {
+      "epoch": 0.7928612113157395,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.0799019951839656e-06,
+      "loss": 2.3165,
+      "step": 522
+    },
+    {
+      "epoch": 0.7943801025251567,
+      "grad_norm": 0.89453125,
+      "learning_rate": 1.0646666678641477e-06,
+      "loss": 2.4474,
+      "step": 523
+    },
+    {
+      "epoch": 0.7958989937345737,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.0495267604613273e-06,
+      "loss": 2.4753,
+      "step": 524
+    },
+    {
+      "epoch": 0.7974178849439909,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.0344826400724185e-06,
+      "loss": 2.3867,
+      "step": 525
+    },
+    {
+      "epoch": 0.798936776153408,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.0195346714717813e-06,
+      "loss": 2.4024,
+      "step": 526
+    },
+    {
+      "epoch": 0.8004556673628251,
+      "grad_norm": 0.83984375,
+      "learning_rate": 1.0046832171023952e-06,
+      "loss": 2.4095,
+      "step": 527
+    },
+    {
+      "epoch": 0.8019745585722423,
+      "grad_norm": 0.8828125,
+      "learning_rate": 9.899286370670575e-07,
+      "loss": 2.4188,
+      "step": 528
+    },
+    {
+      "epoch": 0.8034934497816594,
+      "grad_norm": 0.9921875,
+      "learning_rate": 9.752712891196558e-07,
+      "loss": 2.3859,
+      "step": 529
+    },
+    {
+      "epoch": 0.8050123409910765,
+      "grad_norm": 0.8515625,
+      "learning_rate": 9.607115286564972e-07,
+      "loss": 2.4168,
+      "step": 530
+    },
+    {
+      "epoch": 0.8065312322004936,
+      "grad_norm": 0.83984375,
+      "learning_rate": 9.46249708707681e-07,
+      "loss": 2.4961,
+      "step": 531
+    },
+    {
+      "epoch": 0.8080501234099108,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.318861799285539e-07,
+      "loss": 2.4079,
+      "step": 532
+    },
+    {
+      "epoch": 0.8095690146193278,
+      "grad_norm": 0.8203125,
+      "learning_rate": 9.176212905911946e-07,
+      "loss": 2.4301,
+      "step": 533
+    },
+    {
+      "epoch": 0.811087905828745,
+      "grad_norm": 0.9375,
+      "learning_rate": 9.034553865759754e-07,
+      "loss": 2.3373,
+      "step": 534
+    },
+    {
+      "epoch": 0.8126067970381622,
+      "grad_norm": 0.8515625,
+      "learning_rate": 8.893888113631732e-07,
+      "loss": 2.4129,
+      "step": 535
+    },
+    {
+      "epoch": 0.8141256882475792,
+      "grad_norm": 0.8359375,
+      "learning_rate": 8.754219060246432e-07,
+      "loss": 2.4006,
+      "step": 536
+    },
+    {
+      "epoch": 0.8156445794569964,
+      "grad_norm": 0.875,
+      "learning_rate": 8.615550092155478e-07,
+      "loss": 2.4266,
+      "step": 537
+    },
+    {
+      "epoch": 0.8171634706664135,
+      "grad_norm": 0.859375,
+      "learning_rate": 8.477884571661449e-07,
+      "loss": 2.338,
+      "step": 538
+    },
+    {
+      "epoch": 0.8186823618758307,
+      "grad_norm": 1.0234375,
+      "learning_rate": 8.341225836736367e-07,
+      "loss": 2.3338,
+      "step": 539
+    },
+    {
+      "epoch": 0.8202012530852477,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.20557720094074e-07,
+      "loss": 2.4662,
+      "step": 540
+    },
+    {
+      "epoch": 0.8217201442946649,
+      "grad_norm": 0.84765625,
+      "learning_rate": 8.070941953343242e-07,
+      "loss": 2.4553,
+      "step": 541
+    },
+    {
+      "epoch": 0.8232390355040821,
+      "grad_norm": 0.828125,
+      "learning_rate": 7.937323358440935e-07,
+      "loss": 2.385,
+      "step": 542
+    },
+    {
+      "epoch": 0.8247579267134991,
+      "grad_norm": 0.82421875,
+      "learning_rate": 7.804724656080182e-07,
+      "loss": 2.3504,
+      "step": 543
+    },
+    {
+      "epoch": 0.8262768179229163,
+      "grad_norm": 0.86328125,
+      "learning_rate": 7.673149061377966e-07,
+      "loss": 2.4097,
+      "step": 544
+    },
+    {
+      "epoch": 0.8277957091323334,
+      "grad_norm": 0.84765625,
+      "learning_rate": 7.542599764644049e-07,
+      "loss": 2.3469,
+      "step": 545
+    },
+    {
+      "epoch": 0.8293146003417505,
+      "grad_norm": 0.8671875,
+      "learning_rate": 7.413079931303591e-07,
+      "loss": 2.3823,
+      "step": 546
+    },
+    {
+      "epoch": 0.8308334915511677,
+      "grad_norm": 0.8671875,
+      "learning_rate": 7.284592701820325e-07,
+      "loss": 2.4564,
+      "step": 547
+    },
+    {
+      "epoch": 0.8323523827605848,
+      "grad_norm": 0.93359375,
+      "learning_rate": 7.157141191620548e-07,
+      "loss": 2.4415,
+      "step": 548
+    },
+    {
+      "epoch": 0.8338712739700019,
+      "grad_norm": 0.87890625,
+      "learning_rate": 7.030728491017408e-07,
+      "loss": 2.4262,
+      "step": 549
+    },
+    {
+      "epoch": 0.835390165179419,
+      "grad_norm": 0.87890625,
+      "learning_rate": 6.905357665136142e-07,
+      "loss": 2.3055,
+      "step": 550
+    },
+    {
+      "epoch": 0.8369090563888362,
+      "grad_norm": 0.88671875,
+      "learning_rate": 6.781031753839662e-07,
+      "loss": 2.5036,
+      "step": 551
+    },
+    {
+      "epoch": 0.8384279475982532,
+      "grad_norm": 0.8515625,
+      "learning_rate": 6.657753771654812e-07,
+      "loss": 2.3845,
+      "step": 552
+    },
+    {
+      "epoch": 0.8399468388076704,
+      "grad_norm": 0.828125,
+      "learning_rate": 6.535526707699408e-07,
+      "loss": 2.3961,
+      "step": 553
+    },
+    {
+      "epoch": 0.8414657300170876,
+      "grad_norm": 0.83203125,
+      "learning_rate": 6.414353525609628e-07,
+      "loss": 2.4355,
+      "step": 554
+    },
+    {
+      "epoch": 0.8429846212265046,
+      "grad_norm": 0.87890625,
+      "learning_rate": 6.294237163468231e-07,
+      "loss": 2.4291,
+      "step": 555
+    },
+    {
+      "epoch": 0.8445035124359218,
+      "grad_norm": 0.875,
+      "learning_rate": 6.175180533733277e-07,
+      "loss": 2.4261,
+      "step": 556
+    },
+    {
+      "epoch": 0.8460224036453389,
+      "grad_norm": 0.890625,
+      "learning_rate": 6.057186523167529e-07,
+      "loss": 2.4062,
+      "step": 557
+    },
+    {
+      "epoch": 0.847541294854756,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.940257992768456e-07,
+      "loss": 2.4546,
+      "step": 558
+    },
+    {
+      "epoch": 0.8490601860641731,
+      "grad_norm": 0.8671875,
+      "learning_rate": 5.824397777698859e-07,
+      "loss": 2.4496,
+      "step": 559
+    },
+    {
+      "epoch": 0.8505790772735903,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.709608687218116e-07,
+      "loss": 2.3831,
+      "step": 560
+    },
+    {
+      "epoch": 0.8520979684830075,
+      "grad_norm": 0.875,
+      "learning_rate": 5.595893504614097e-07,
+      "loss": 2.4172,
+      "step": 561
+    },
+    {
+      "epoch": 0.8536168596924245,
+      "grad_norm": 0.84375,
+      "learning_rate": 5.483254987135644e-07,
+      "loss": 2.4194,
+      "step": 562
+    },
+    {
+      "epoch": 0.8551357509018417,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5.371695865925736e-07,
+      "loss": 2.3656,
+      "step": 563
+    },
+    {
+      "epoch": 0.8566546421112587,
+      "grad_norm": 0.84765625,
+      "learning_rate": 5.261218845955246e-07,
+      "loss": 2.3516,
+      "step": 564
+    },
+    {
+      "epoch": 0.8581735333206759,
+      "grad_norm": 0.859375,
+      "learning_rate": 5.151826605957394e-07,
+      "loss": 2.4581,
+      "step": 565
+    },
+    {
+      "epoch": 0.859692424530093,
+      "grad_norm": 0.90625,
+      "learning_rate": 5.043521798362755e-07,
+      "loss": 2.4911,
+      "step": 566
+    },
+    {
+      "epoch": 0.8612113157395102,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.936307049234956e-07,
+      "loss": 2.4658,
+      "step": 567
+    },
+    {
+      "epoch": 0.8627302069489273,
+      "grad_norm": 0.859375,
+      "learning_rate": 4.830184958207007e-07,
+      "loss": 2.4005,
+      "step": 568
+    },
+    {
+      "epoch": 0.8642490981583444,
+      "grad_norm": 0.875,
+      "learning_rate": 4.725158098418309e-07,
+      "loss": 2.4644,
+      "step": 569
+    },
+    {
+      "epoch": 0.8657679893677616,
+      "grad_norm": 0.8125,
+      "learning_rate": 4.6212290164521554e-07,
+      "loss": 2.2967,
+      "step": 570
+    },
+    {
+      "epoch": 0.8672868805771786,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.5184002322740784e-07,
+      "loss": 2.3816,
+      "step": 571
+    },
+    {
+      "epoch": 0.8688057717865958,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.4166742391707593e-07,
+      "loss": 2.4745,
+      "step": 572
+    },
+    {
+      "epoch": 0.870324662996013,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.316053503689466e-07,
+      "loss": 2.4481,
+      "step": 573
+    },
+    {
+      "epoch": 0.87184355420543,
+      "grad_norm": 0.89453125,
+      "learning_rate": 4.2165404655783836e-07,
+      "loss": 2.4813,
+      "step": 574
+    },
+    {
+      "epoch": 0.8733624454148472,
+      "grad_norm": 0.8671875,
+      "learning_rate": 4.1181375377273237e-07,
+      "loss": 2.4648,
+      "step": 575
+    },
+    {
+      "epoch": 0.8748813366242643,
+      "grad_norm": 0.8515625,
+      "learning_rate": 4.020847106109349e-07,
+      "loss": 2.5166,
+      "step": 576
+    },
+    {
+      "epoch": 0.8764002278336814,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.9246715297228176e-07,
+      "loss": 2.3706,
+      "step": 577
+    },
+    {
+      "epoch": 0.8779191190430985,
+      "grad_norm": 0.84375,
+      "learning_rate": 3.829613140534222e-07,
+      "loss": 2.4946,
+      "step": 578
+    },
+    {
+      "epoch": 0.8794380102525157,
+      "grad_norm": 0.890625,
+      "learning_rate": 3.7356742434216775e-07,
+      "loss": 2.4344,
+      "step": 579
+    },
+    {
+      "epoch": 0.8809569014619328,
+      "grad_norm": 0.84765625,
+      "learning_rate": 3.642857116118986e-07,
+      "loss": 2.4431,
+      "step": 580
+    },
+    {
+      "epoch": 0.8824757926713499,
+      "grad_norm": 0.875,
+      "learning_rate": 3.5511640091604293e-07,
+      "loss": 2.3728,
+      "step": 581
+    },
+    {
+      "epoch": 0.8839946838807671,
+      "grad_norm": 0.81640625,
+      "learning_rate": 3.4605971458262e-07,
+      "loss": 2.358,
+      "step": 582
+    },
+    {
+      "epoch": 0.8855135750901841,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.371158722088497e-07,
+      "loss": 2.4764,
+      "step": 583
+    },
+    {
+      "epoch": 0.8870324662996013,
+      "grad_norm": 0.8984375,
+      "learning_rate": 3.2828509065582713e-07,
+      "loss": 2.4414,
+      "step": 584
+    },
+    {
+      "epoch": 0.8885513575090184,
+      "grad_norm": 0.8515625,
+      "learning_rate": 3.195675840432655e-07,
+      "loss": 2.4016,
+      "step": 585
+    },
+    {
+      "epoch": 0.8900702487184355,
+      "grad_norm": 0.85546875,
+      "learning_rate": 3.109635637443026e-07,
+      "loss": 2.4212,
+      "step": 586
+    },
+    {
+      "epoch": 0.8915891399278527,
+      "grad_norm": 0.890625,
+      "learning_rate": 3.02473238380378e-07,
+      "loss": 2.2654,
+      "step": 587
+    },
+    {
+      "epoch": 0.8931080311372698,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.9409681381617315e-07,
+      "loss": 2.2745,
+      "step": 588
+    },
+    {
+      "epoch": 0.894626922346687,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.858344931546181e-07,
+      "loss": 2.3712,
+      "step": 589
+    },
+    {
+      "epoch": 0.896145813556104,
+      "grad_norm": 0.86328125,
+      "learning_rate": 2.776864767319731e-07,
+      "loss": 2.4357,
+      "step": 590
+    },
+    {
+      "epoch": 0.8976647047655212,
+      "grad_norm": 0.890625,
+      "learning_rate": 2.696529621129618e-07,
+      "loss": 2.4577,
+      "step": 591
+    },
+    {
+      "epoch": 0.8991835959749382,
+      "grad_norm": 0.828125,
+      "learning_rate": 2.617341440859883e-07,
+      "loss": 2.3876,
+      "step": 592
+    },
+    {
+      "epoch": 0.9007024871843554,
+      "grad_norm": 0.9609375,
+      "learning_rate": 2.539302146584116e-07,
+      "loss": 2.3993,
+      "step": 593
+    },
+    {
+      "epoch": 0.9022213783937726,
+      "grad_norm": 0.828125,
+      "learning_rate": 2.4624136305188895e-07,
+      "loss": 2.3749,
+      "step": 594
+    },
+    {
+      "epoch": 0.9037402696031896,
+      "grad_norm": 0.87109375,
+      "learning_rate": 2.3866777569779234e-07,
+      "loss": 2.4471,
+      "step": 595
+    },
+    {
+      "epoch": 0.9052591608126068,
+      "grad_norm": 0.86328125,
+      "learning_rate": 2.3120963623267822e-07,
+      "loss": 2.3111,
+      "step": 596
+    },
+    {
+      "epoch": 0.9067780520220239,
+      "grad_norm": 0.86328125,
+      "learning_rate": 2.2386712549384848e-07,
+      "loss": 2.3943,
+      "step": 597
+    },
+    {
+      "epoch": 0.9082969432314411,
+      "grad_norm": 0.875,
+      "learning_rate": 2.1664042151495424e-07,
+      "loss": 2.4127,
+      "step": 598
+    },
+    {
+      "epoch": 0.9098158344408582,
+      "grad_norm": 0.859375,
+      "learning_rate": 2.095296995216828e-07,
+      "loss": 2.5428,
+      "step": 599
+    },
+    {
+      "epoch": 0.9113347256502753,
+      "grad_norm": 0.88671875,
+      "learning_rate": 2.0253513192751374e-07,
+      "loss": 2.4433,
+      "step": 600
+    },
+    {
+      "epoch": 0.9128536168596925,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.9565688832952846e-07,
+      "loss": 2.4961,
+      "step": 601
+    },
+    {
+      "epoch": 0.9143725080691095,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.8889513550430892e-07,
+      "loss": 2.4045,
+      "step": 602
+    },
+    {
+      "epoch": 0.9158913992785267,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.8225003740388546e-07,
+      "loss": 2.4311,
+      "step": 603
+    },
+    {
+      "epoch": 0.9174102904879438,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.7572175515176538e-07,
+      "loss": 2.4039,
+      "step": 604
+    },
+    {
+      "epoch": 0.9189291816973609,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.693104470390261e-07,
+      "loss": 2.4362,
+      "step": 605
+    },
+    {
+      "epoch": 0.9204480729067781,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.6301626852047504e-07,
+      "loss": 2.3737,
+      "step": 606
+    },
+    {
+      "epoch": 0.9219669641161952,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.5683937221088242e-07,
+      "loss": 2.3526,
+      "step": 607
+    },
+    {
+      "epoch": 0.9234858553256123,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.5077990788127993e-07,
+      "loss": 2.4345,
+      "step": 608
+    },
+    {
+      "epoch": 0.9250047465350294,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.448380224553303e-07,
+      "loss": 2.359,
+      "step": 609
+    },
+    {
+      "epoch": 0.9265236377444466,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.3901386000576112e-07,
+      "loss": 2.4759,
+      "step": 610
+    },
+    {
+      "epoch": 0.9280425289538636,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.3330756175087778e-07,
+      "loss": 2.4237,
+      "step": 611
+    },
+    {
+      "epoch": 0.9295614201632808,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.2771926605113283e-07,
+      "loss": 2.411,
+      "step": 612
+    },
+    {
+      "epoch": 0.931080311372698,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.2224910840577642e-07,
+      "loss": 2.495,
+      "step": 613
+    },
+    {
+      "epoch": 0.932599202582115,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.1689722144956672e-07,
+      "loss": 2.4696,
+      "step": 614
+    },
+    {
+      "epoch": 0.9341180937915322,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.1166373494955696e-07,
+      "loss": 2.4935,
+      "step": 615
+    },
+    {
+      "epoch": 0.9356369850009493,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.06548775801949e-07,
+      "loss": 2.4556,
+      "step": 616
+    },
+    {
+      "epoch": 0.9371558762103664,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.0155246802901198e-07,
+      "loss": 2.4549,
+      "step": 617
+    },
+    {
+      "epoch": 0.9386747674197835,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.667493277608187e-08,
+      "loss": 2.3727,
+      "step": 618
+    },
+    {
+      "epoch": 0.9401936586292007,
+      "grad_norm": 0.84375,
+      "learning_rate": 9.191628830861832e-08,
+      "loss": 2.4479,
+      "step": 619
+    },
+    {
+      "epoch": 0.9417125498386179,
+      "grad_norm": 0.83984375,
+      "learning_rate": 8.727665000934027e-08,
+      "loss": 2.4592,
+      "step": 620
+    },
+    {
+      "epoch": 0.9432314410480349,
+      "grad_norm": 0.875,
+      "learning_rate": 8.275613037542873e-08,
+      "loss": 2.3427,
+      "step": 621
+    },
+    {
+      "epoch": 0.9447503322574521,
+      "grad_norm": 0.85546875,
+      "learning_rate": 7.835483901579454e-08,
+      "loss": 2.3793,
+      "step": 622
+    },
+    {
+      "epoch": 0.9462692234668691,
+      "grad_norm": 0.8515625,
+      "learning_rate": 7.407288264842772e-08,
+      "loss": 2.3967,
+      "step": 623
+    },
+    {
+      "epoch": 0.9477881146762863,
+      "grad_norm": 1.3203125,
+      "learning_rate": 6.991036509780391e-08,
+      "loss": 2.3673,
+      "step": 624
+    },
+    {
+      "epoch": 0.9493070058857034,
+      "grad_norm": 0.83984375,
+      "learning_rate": 6.58673872923693e-08,
+      "loss": 2.3546,
+      "step": 625
+    },
+    {
+      "epoch": 0.9508258970951206,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.194404726209358e-08,
+      "loss": 2.4605,
+      "step": 626
+    },
+    {
+      "epoch": 0.9523447883045377,
+      "grad_norm": 0.828125,
+      "learning_rate": 5.8140440136091326e-08,
+      "loss": 2.4121,
+      "step": 627
+    },
+    {
+      "epoch": 0.9538636795139548,
+      "grad_norm": 0.8984375,
+      "learning_rate": 5.445665814031942e-08,
+      "loss": 2.3989,
+      "step": 628
+    },
+    {
+      "epoch": 0.955382570723372,
+      "grad_norm": 0.84375,
+      "learning_rate": 5.089279059533658e-08,
+      "loss": 2.4139,
+      "step": 629
+    },
+    {
+      "epoch": 0.956901461932789,
+      "grad_norm": 0.92578125,
+      "learning_rate": 4.744892391413791e-08,
+      "loss": 2.3641,
+      "step": 630
+    },
+    {
+      "epoch": 0.9584203531422062,
+      "grad_norm": 0.875,
+      "learning_rate": 4.412514160006376e-08,
+      "loss": 2.445,
+      "step": 631
+    },
+    {
+      "epoch": 0.9599392443516234,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.092152424477025e-08,
+      "loss": 2.4109,
+      "step": 632
+    },
+    {
+      "epoch": 0.9614581355610404,
+      "grad_norm": 0.875,
+      "learning_rate": 3.7838149526277514e-08,
+      "loss": 2.4756,
+      "step": 633
+    },
+    {
+      "epoch": 0.9629770267704576,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.487509220708563e-08,
+      "loss": 2.3943,
+      "step": 634
+    },
+    {
+      "epoch": 0.9644959179798747,
+      "grad_norm": 0.859375,
+      "learning_rate": 3.2032424132362736e-08,
+      "loss": 2.4921,
+      "step": 635
+    },
+    {
+      "epoch": 0.9660148091892918,
+      "grad_norm": 0.83203125,
+      "learning_rate": 2.9310214228202016e-08,
+      "loss": 2.4052,
+      "step": 636
+    },
+    {
+      "epoch": 0.9675337003987089,
+      "grad_norm": 0.82421875,
+      "learning_rate": 2.6708528499950758e-08,
+      "loss": 2.4681,
+      "step": 637
+    },
+    {
+      "epoch": 0.9690525916081261,
+      "grad_norm": 0.87109375,
+      "learning_rate": 2.4227430030609455e-08,
+      "loss": 2.378,
+      "step": 638
+    },
+    {
+      "epoch": 0.9705714828175432,
+      "grad_norm": 1.25,
+      "learning_rate": 2.1866978979303567e-08,
+      "loss": 2.4043,
+      "step": 639
+    },
+    {
+      "epoch": 0.9720903740269603,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.962723257982302e-08,
+      "loss": 2.4691,
+      "step": 640
+    },
+    {
+      "epoch": 0.9736092652363775,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.7508245139236658e-08,
+      "loss": 2.4154,
+      "step": 641
+    },
+    {
+      "epoch": 0.9751281564457945,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.5510068036573288e-08,
+      "loss": 2.4022,
+      "step": 642
+    },
+    {
+      "epoch": 0.9766470476552117,
+      "grad_norm": 0.8671875,
+      "learning_rate": 1.3632749721577132e-08,
+      "loss": 2.4114,
+      "step": 643
+    },
+    {
+      "epoch": 0.9781659388646288,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.1876335713532638e-08,
+      "loss": 2.4238,
+      "step": 644
+    },
+    {
+      "epoch": 0.9796848300740459,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.024086860016149e-08,
+      "loss": 2.3759,
+      "step": 645
+    },
+    {
+      "epoch": 0.9812037212834631,
+      "grad_norm": 0.85546875,
+      "learning_rate": 8.726388036587874e-09,
+      "loss": 2.4658,
+      "step": 646
+    },
+    {
+      "epoch": 0.9827226124928802,
+      "grad_norm": 0.90625,
+      "learning_rate": 7.332930744380906e-09,
+      "loss": 2.424,
+      "step": 647
+    },
+    {
+      "epoch": 0.9842415037022973,
+      "grad_norm": 0.8671875,
+      "learning_rate": 6.060530510659246e-09,
+      "loss": 2.4868,
+      "step": 648
+    },
+    {
+      "epoch": 0.9857603949117144,
+      "grad_norm": 0.890625,
+      "learning_rate": 4.909218187276743e-09,
+      "loss": 2.5088,
+      "step": 649
+    },
+    {
+      "epoch": 0.9872792861211316,
+      "grad_norm": 0.83203125,
+      "learning_rate": 3.8790216900702615e-09,
+      "loss": 2.4717,
+      "step": 650
+    },
+    {
+      "epoch": 0.9887981773305486,
+      "grad_norm": 0.890625,
+      "learning_rate": 2.9699659981863306e-09,
+      "loss": 2.4196,
+      "step": 651
+    },
+    {
+      "epoch": 0.9903170685399658,
+      "grad_norm": 0.82421875,
+      "learning_rate": 2.182073153471631e-09,
+      "loss": 2.4242,
+      "step": 652
+    },
+    {
+      "epoch": 0.991835959749383,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.5153622599428652e-09,
+      "loss": 2.3848,
+      "step": 653
+    },
+    {
+      "epoch": 0.9933548509588,
+      "grad_norm": 0.85546875,
+      "learning_rate": 9.698494833199068e-10,
+      "loss": 2.4641,
+      "step": 654
+    },
+    {
+      "epoch": 0.9948737421682172,
+      "grad_norm": 0.83984375,
+      "learning_rate": 5.455480506355582e-10,
+      "loss": 2.4073,
+      "step": 655
+    },
+    {
+      "epoch": 0.9963926333776343,
+      "grad_norm": 0.84375,
+      "learning_rate": 2.4246824991525085e-10,
+      "loss": 2.4084,
+      "step": 656
+    },
+    {
+      "epoch": 0.9979115245870515,
+      "grad_norm": 0.90625,
+      "learning_rate": 6.061742992613529e-11,
+      "loss": 2.4904,
+      "step": 657
+    },
+    {
+      "epoch": 0.9994304157964686,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.0,
+      "loss": 2.4394,
+      "step": 658
+    },
+    {
+      "epoch": 0.9994304157964686,
+      "step": 658,
+      "total_flos": 1.2030734967018357e+19,
+      "train_loss": 0.9496028307361081,
+      "train_runtime": 40588.5091,
+      "train_samples_per_second": 2.076,
+      "train_steps_per_second": 0.016
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 658,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 200,
+  "total_flos": 1.2030734967018357e+19,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}