diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,4978 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 3.0,
+  "eval_steps": 500,
+  "global_step": 1410,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.004259850905218318,
+      "grad_norm": 0.80859375,
+      "learning_rate": 4.225352112676057e-07,
+      "loss": 1.8638979196548462,
+      "step": 2
+    },
+    {
+      "epoch": 0.008519701810436636,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.267605633802817e-06,
+      "loss": 1.9382712841033936,
+      "step": 4
+    },
+    {
+      "epoch": 0.012779552715654952,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.1126760563380285e-06,
+      "loss": 1.8919719457626343,
+      "step": 6
+    },
+    {
+      "epoch": 0.01703940362087327,
+      "grad_norm": 0.6875,
+      "learning_rate": 2.957746478873239e-06,
+      "loss": 1.9754539728164673,
+      "step": 8
+    },
+    {
+      "epoch": 0.021299254526091587,
+      "grad_norm": 0.8046875,
+      "learning_rate": 3.8028169014084508e-06,
+      "loss": 1.9735431671142578,
+      "step": 10
+    },
+    {
+      "epoch": 0.025559105431309903,
+      "grad_norm": 0.490234375,
+      "learning_rate": 4.6478873239436615e-06,
+      "loss": 1.962188959121704,
+      "step": 12
+    },
+    {
+      "epoch": 0.029818956336528223,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.492957746478874e-06,
+      "loss": 1.8216444253921509,
+      "step": 14
+    },
+    {
+      "epoch": 0.03407880724174654,
+      "grad_norm": 0.65234375,
+      "learning_rate": 6.338028169014085e-06,
+      "loss": 1.879197359085083,
+      "step": 16
+    },
+    {
+      "epoch": 0.038338658146964855,
+      "grad_norm": 0.58203125,
+      "learning_rate": 7.183098591549295e-06,
+      "loss": 1.9074592590332031,
+      "step": 18
+    },
+    {
+      "epoch": 0.042598509052183174,
+      "grad_norm": 0.5625,
+      "learning_rate": 8.028169014084507e-06,
+      "loss": 1.8535538911819458,
+      "step": 20
+    },
+    {
+      "epoch": 0.046858359957401494,
+      "grad_norm": 0.70703125,
+      "learning_rate": 8.873239436619718e-06,
+      "loss": 1.7652872800827026,
+      "step": 22
+    },
+    {
+      "epoch": 0.051118210862619806,
+      "grad_norm": 0.470703125,
+      "learning_rate": 9.71830985915493e-06,
+      "loss": 1.7537739276885986,
+      "step": 24
+    },
+    {
+      "epoch": 0.055378061767838126,
+      "grad_norm": 0.4375,
+      "learning_rate": 1.056338028169014e-05,
+      "loss": 1.6868540048599243,
+      "step": 26
+    },
+    {
+      "epoch": 0.059637912673056445,
+      "grad_norm": 0.419921875,
+      "learning_rate": 1.1408450704225351e-05,
+      "loss": 1.694838523864746,
+      "step": 28
+    },
+    {
+      "epoch": 0.06389776357827476,
+      "grad_norm": 0.6953125,
+      "learning_rate": 1.2253521126760564e-05,
+      "loss": 1.7118496894836426,
+      "step": 30
+    },
+    {
+      "epoch": 0.06815761448349308,
+      "grad_norm": 0.5,
+      "learning_rate": 1.3098591549295775e-05,
+      "loss": 1.7367674112319946,
+      "step": 32
+    },
+    {
+      "epoch": 0.0724174653887114,
+      "grad_norm": 0.625,
+      "learning_rate": 1.3943661971830985e-05,
+      "loss": 1.6931723356246948,
+      "step": 34
+    },
+    {
+      "epoch": 0.07667731629392971,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.4788732394366198e-05,
+      "loss": 1.8322988748550415,
+      "step": 36
+    },
+    {
+      "epoch": 0.08093716719914804,
+      "grad_norm": 0.71875,
+      "learning_rate": 1.563380281690141e-05,
+      "loss": 1.723473072052002,
+      "step": 38
+    },
+    {
+      "epoch": 0.08519701810436635,
+      "grad_norm": 0.765625,
+      "learning_rate": 1.6478873239436623e-05,
+      "loss": 1.584614872932434,
+      "step": 40
+    },
+    {
+      "epoch": 0.08945686900958466,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.7323943661971833e-05,
+      "loss": 1.757703423500061,
+      "step": 42
+    },
+    {
+      "epoch": 0.09371671991480299,
+      "grad_norm": 0.384765625,
+      "learning_rate": 1.816901408450704e-05,
+      "loss": 1.5781840085983276,
+      "step": 44
+    },
+    {
+      "epoch": 0.0979765708200213,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.9014084507042255e-05,
+      "loss": 1.6014955043792725,
+      "step": 46
+    },
+    {
+      "epoch": 0.10223642172523961,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.9859154929577465e-05,
+      "loss": 1.674490213394165,
+      "step": 48
+    },
+    {
+      "epoch": 0.10649627263045794,
+      "grad_norm": 0.61328125,
+      "learning_rate": 2.0704225352112676e-05,
+      "loss": 1.393676519393921,
+      "step": 50
+    },
+    {
+      "epoch": 0.11075612353567625,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.154929577464789e-05,
+      "loss": 1.495132327079773,
+      "step": 52
+    },
+    {
+      "epoch": 0.11501597444089456,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.23943661971831e-05,
+      "loss": 1.5442848205566406,
+      "step": 54
+    },
+    {
+      "epoch": 0.11927582534611289,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.3239436619718308e-05,
+      "loss": 1.459320068359375,
+      "step": 56
+    },
+    {
+      "epoch": 0.1235356762513312,
+      "grad_norm": 0.7578125,
+      "learning_rate": 2.4084507042253522e-05,
+      "loss": 1.4848005771636963,
+      "step": 58
+    },
+    {
+      "epoch": 0.12779552715654952,
+      "grad_norm": 0.310546875,
+      "learning_rate": 2.4929577464788733e-05,
+      "loss": 1.3594304323196411,
+      "step": 60
+    },
+    {
+      "epoch": 0.13205537806176784,
+      "grad_norm": 0.6640625,
+      "learning_rate": 2.5774647887323944e-05,
+      "loss": 1.4234025478363037,
+      "step": 62
+    },
+    {
+      "epoch": 0.13631522896698617,
+      "grad_norm": 0.27734375,
+      "learning_rate": 2.6619718309859158e-05,
+      "loss": 1.3102433681488037,
+      "step": 64
+    },
+    {
+      "epoch": 0.14057507987220447,
+      "grad_norm": 0.3046875,
+      "learning_rate": 2.746478873239437e-05,
+      "loss": 1.3784610033035278,
+      "step": 66
+    },
+    {
+      "epoch": 0.1448349307774228,
+      "grad_norm": 0.5625,
+      "learning_rate": 2.8309859154929576e-05,
+      "loss": 1.325601577758789,
+      "step": 68
+    },
+    {
+      "epoch": 0.14909478168264112,
+      "grad_norm": 0.36328125,
+      "learning_rate": 2.915492957746479e-05,
+      "loss": 1.3749815225601196,
+      "step": 70
+    },
+    {
+      "epoch": 0.15335463258785942,
+      "grad_norm": 0.2890625,
+      "learning_rate": 3e-05,
+      "loss": 1.4330412149429321,
+      "step": 72
+    },
+    {
+      "epoch": 0.15761448349307774,
+      "grad_norm": 1.609375,
+      "learning_rate": 2.9999867885940888e-05,
+      "loss": 1.4420013427734375,
+      "step": 74
+    },
+    {
+      "epoch": 0.16187433439829607,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.999947154667255e-05,
+      "loss": 1.3299309015274048,
+      "step": 76
+    },
+    {
+      "epoch": 0.16613418530351437,
+      "grad_norm": 0.3125,
+      "learning_rate": 2.9998810990921997e-05,
+      "loss": 1.2926124334335327,
+      "step": 78
+    },
+    {
+      "epoch": 0.1703940362087327,
+      "grad_norm": 0.34765625,
+      "learning_rate": 2.999788623323402e-05,
+      "loss": 1.4053008556365967,
+      "step": 80
+    },
+    {
+      "epoch": 0.17465388711395102,
+      "grad_norm": 0.341796875,
+      "learning_rate": 2.999669729397085e-05,
+      "loss": 1.3341435194015503,
+      "step": 82
+    },
+    {
+      "epoch": 0.17891373801916932,
+      "grad_norm": 1.1171875,
+      "learning_rate": 2.999524419931176e-05,
+      "loss": 1.3351154327392578,
+      "step": 84
+    },
+    {
+      "epoch": 0.18317358892438765,
+      "grad_norm": 0.275390625,
+      "learning_rate": 2.9993526981252465e-05,
+      "loss": 1.279821515083313,
+      "step": 86
+    },
+    {
+      "epoch": 0.18743343982960597,
+      "grad_norm": 0.29296875,
+      "learning_rate": 2.999154567760439e-05,
+      "loss": 1.2992417812347412,
+      "step": 88
+    },
+    {
+      "epoch": 0.19169329073482427,
+      "grad_norm": 0.43359375,
+      "learning_rate": 2.998930033199389e-05,
+      "loss": 1.2671334743499756,
+      "step": 90
+    },
+    {
+      "epoch": 0.1959531416400426,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.9986790993861245e-05,
+      "loss": 1.4086840152740479,
+      "step": 92
+    },
+    {
+      "epoch": 0.20021299254526093,
+      "grad_norm": 0.4765625,
+      "learning_rate": 2.9984017718459603e-05,
+      "loss": 1.260171890258789,
+      "step": 94
+    },
+    {
+      "epoch": 0.20447284345047922,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.998098056685374e-05,
+      "loss": 1.337839126586914,
+      "step": 96
+    },
+    {
+      "epoch": 0.20873269435569755,
+      "grad_norm": 0.44921875,
+      "learning_rate": 2.9977679605918732e-05,
+      "loss": 1.3128660917282104,
+      "step": 98
+    },
+    {
+      "epoch": 0.21299254526091588,
+      "grad_norm": 0.3515625,
+      "learning_rate": 2.9974114908338454e-05,
+      "loss": 1.3010621070861816,
+      "step": 100
+    },
+    {
+      "epoch": 0.21725239616613418,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 2.9970286552604036e-05,
+      "loss": 1.3007886409759521,
+      "step": 102
+    },
+    {
+      "epoch": 0.2215122470713525,
+      "grad_norm": 0.251953125,
+      "learning_rate": 2.996619462301207e-05,
+      "loss": 1.3102291822433472,
+      "step": 104
+    },
+    {
+      "epoch": 0.22577209797657083,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.9961839209662808e-05,
+      "loss": 1.4082542657852173,
+      "step": 106
+    },
+    {
+      "epoch": 0.23003194888178913,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.9957220408458118e-05,
+      "loss": 1.220694899559021,
+      "step": 108
+    },
+    {
+      "epoch": 0.23429179978700745,
+      "grad_norm": 0.494140625,
+      "learning_rate": 2.9952338321099435e-05,
+      "loss": 1.3067396879196167,
+      "step": 110
+    },
+    {
+      "epoch": 0.23855165069222578,
+      "grad_norm": 0.34765625,
+      "learning_rate": 2.9947193055085505e-05,
+      "loss": 1.283180594444275,
+      "step": 112
+    },
+    {
+      "epoch": 0.24281150159744408,
+      "grad_norm": 0.255859375,
+      "learning_rate": 2.9941784723709973e-05,
+      "loss": 1.273285150527954,
+      "step": 114
+    },
+    {
+      "epoch": 0.2470713525026624,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.993611344605895e-05,
+      "loss": 1.2464628219604492,
+      "step": 116
+    },
+    {
+      "epoch": 0.25133120340788073,
+      "grad_norm": 0.255859375,
+      "learning_rate": 2.9930179347008347e-05,
+      "loss": 1.2899105548858643,
+      "step": 118
+    },
+    {
+      "epoch": 0.25559105431309903,
+      "grad_norm": 0.71484375,
+      "learning_rate": 2.9923982557221154e-05,
+      "loss": 1.166140079498291,
+      "step": 120
+    },
+    {
+      "epoch": 0.2598509052183174,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.9917523213144554e-05,
+      "loss": 1.3073238134384155,
+      "step": 122
+    },
+    {
+      "epoch": 0.2641107561235357,
+      "grad_norm": 0.251953125,
+      "learning_rate": 2.9910801457006897e-05,
+      "loss": 1.2734519243240356,
+      "step": 124
+    },
+    {
+      "epoch": 0.268370607028754,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 2.9903817436814603e-05,
+      "loss": 1.190434455871582,
+      "step": 126
+    },
+    {
+      "epoch": 0.27263045793397234,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.9896571306348874e-05,
+      "loss": 1.2509433031082153,
+      "step": 128
+    },
+    {
+      "epoch": 0.27689030883919064,
+      "grad_norm": 0.48828125,
+      "learning_rate": 2.9889063225162337e-05,
+      "loss": 1.3253034353256226,
+      "step": 130
+    },
+    {
+      "epoch": 0.28115015974440893,
+      "grad_norm": 0.318359375,
+      "learning_rate": 2.98812933585755e-05,
+      "loss": 1.2004448175430298,
+      "step": 132
+    },
+    {
+      "epoch": 0.2854100106496273,
+      "grad_norm": 0.2890625,
+      "learning_rate": 2.9873261877673142e-05,
+      "loss": 1.1677073240280151,
+      "step": 134
+    },
+    {
+      "epoch": 0.2896698615548456,
+      "grad_norm": 0.48046875,
+      "learning_rate": 2.9864968959300505e-05,
+      "loss": 1.2530547380447388,
+      "step": 136
+    },
+    {
+      "epoch": 0.2939297124600639,
+      "grad_norm": 0.482421875,
+      "learning_rate": 2.985641478605945e-05,
+      "loss": 1.2573705911636353,
+      "step": 138
+    },
+    {
+      "epoch": 0.29818956336528224,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.9847599546304395e-05,
+      "loss": 1.3057535886764526,
+      "step": 140
+    },
+    {
+      "epoch": 0.30244941427050054,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.9838523434138204e-05,
+      "loss": 1.2737255096435547,
+      "step": 142
+    },
+    {
+      "epoch": 0.30670926517571884,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.982918664940787e-05,
+      "loss": 1.2411197423934937,
+      "step": 144
+    },
+    {
+      "epoch": 0.3109691160809372,
+      "grad_norm": 0.3203125,
+      "learning_rate": 2.9819589397700148e-05,
+      "loss": 1.2790652513504028,
+      "step": 146
+    },
+    {
+      "epoch": 0.3152289669861555,
+      "grad_norm": 0.408203125,
+      "learning_rate": 2.9809731890337017e-05,
+      "loss": 1.2759779691696167,
+      "step": 148
+    },
+    {
+      "epoch": 0.3194888178913738,
+      "grad_norm": 0.337890625,
+      "learning_rate": 2.979961434437103e-05,
+      "loss": 1.3018522262573242,
+      "step": 150
+    },
+    {
+      "epoch": 0.32374866879659214,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.9789236982580538e-05,
+      "loss": 1.3175352811813354,
+      "step": 152
+    },
+    {
+      "epoch": 0.32800851970181044,
+      "grad_norm": 0.32421875,
+      "learning_rate": 2.9778600033464767e-05,
+      "loss": 1.291448712348938,
+      "step": 154
+    },
+    {
+      "epoch": 0.33226837060702874,
+      "grad_norm": 0.275390625,
+      "learning_rate": 2.97677037312388e-05,
+      "loss": 1.2841699123382568,
+      "step": 156
+    },
+    {
+      "epoch": 0.3365282215122471,
+      "grad_norm": 0.33203125,
+      "learning_rate": 2.975654831582843e-05,
+      "loss": 1.3081012964248657,
+      "step": 158
+    },
+    {
+      "epoch": 0.3407880724174654,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.9745134032864862e-05,
+      "loss": 1.2524945735931396,
+      "step": 160
+    },
+    {
+      "epoch": 0.3450479233226837,
+      "grad_norm": 0.228515625,
+      "learning_rate": 2.973346113367929e-05,
+      "loss": 1.1932268142700195,
+      "step": 162
+    },
+    {
+      "epoch": 0.34930777422790205,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.972152987529741e-05,
+      "loss": 1.2276166677474976,
+      "step": 164
+    },
+    {
+      "epoch": 0.35356762513312034,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.9709340520433722e-05,
+      "loss": 1.2343382835388184,
+      "step": 166
+    },
+    {
+      "epoch": 0.35782747603833864,
+      "grad_norm": 0.96875,
+      "learning_rate": 2.9696893337485734e-05,
+      "loss": 1.3475210666656494,
+      "step": 168
+    },
+    {
+      "epoch": 0.362087326943557,
+      "grad_norm": 0.69921875,
+      "learning_rate": 2.9684188600528098e-05,
+      "loss": 1.2921943664550781,
+      "step": 170
+    },
+    {
+      "epoch": 0.3663471778487753,
+      "grad_norm": 0.515625,
+      "learning_rate": 2.967122658930654e-05,
+      "loss": 1.1600617170333862,
+      "step": 172
+    },
+    {
+      "epoch": 0.3706070287539936,
+      "grad_norm": 0.341796875,
+      "learning_rate": 2.9658007589231723e-05,
+      "loss": 1.3178966045379639,
+      "step": 174
+    },
+    {
+      "epoch": 0.37486687965921195,
+      "grad_norm": 0.33203125,
+      "learning_rate": 2.9644531891372925e-05,
+      "loss": 1.3098689317703247,
+      "step": 176
+    },
+    {
+      "epoch": 0.37912673056443025,
+      "grad_norm": 0.234375,
+      "learning_rate": 2.9630799792451687e-05,
+      "loss": 1.1713343858718872,
+      "step": 178
+    },
+    {
+      "epoch": 0.38338658146964855,
+      "grad_norm": 0.2109375,
+      "learning_rate": 2.9616811594835214e-05,
+      "loss": 1.2428940534591675,
+      "step": 180
+    },
+    {
+      "epoch": 0.3876464323748669,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.9602567606529776e-05,
+      "loss": 1.2774041891098022,
+      "step": 182
+    },
+    {
+      "epoch": 0.3919062832800852,
+      "grad_norm": 0.33984375,
+      "learning_rate": 2.9588068141173888e-05,
+      "loss": 1.2816710472106934,
+      "step": 184
+    },
+    {
+      "epoch": 0.3961661341853035,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.9573313518031424e-05,
+      "loss": 1.1415907144546509,
+      "step": 186
+    },
+    {
+      "epoch": 0.40042598509052185,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.955830406198458e-05,
+      "loss": 1.232388973236084,
+      "step": 188
+    },
+    {
+      "epoch": 0.40468583599574015,
+      "grad_norm": 0.3125,
+      "learning_rate": 2.95430401035267e-05,
+      "loss": 1.250954031944275,
+      "step": 190
+    },
+    {
+      "epoch": 0.40894568690095845,
+      "grad_norm": 0.310546875,
+      "learning_rate": 2.9527521978755053e-05,
+      "loss": 1.205154299736023,
+      "step": 192
+    },
+    {
+      "epoch": 0.4132055378061768,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.9511750029363377e-05,
+      "loss": 1.2991074323654175,
+      "step": 194
+    },
+    {
+      "epoch": 0.4174653887113951,
+      "grad_norm": 0.296875,
+      "learning_rate": 2.949572460263438e-05,
+      "loss": 1.2244809865951538,
+      "step": 196
+    },
+    {
+      "epoch": 0.4217252396166134,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.947944605143208e-05,
+      "loss": 1.2750012874603271,
+      "step": 198
+    },
+    {
+      "epoch": 0.42598509052183176,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.9462914734194078e-05,
+      "loss": 1.3423129320144653,
+      "step": 200
+    },
+    {
+      "epoch": 0.43024494142705005,
+      "grad_norm": 0.35546875,
+      "learning_rate": 2.9446131014923593e-05,
+      "loss": 1.280989646911621,
+      "step": 202
+    },
+    {
+      "epoch": 0.43450479233226835,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.9429095263181514e-05,
+      "loss": 1.2904020547866821,
+      "step": 204
+    },
+    {
+      "epoch": 0.4387646432374867,
+      "grad_norm": 0.427734375,
+      "learning_rate": 2.9411807854078226e-05,
+      "loss": 1.2392964363098145,
+      "step": 206
+    },
+    {
+      "epoch": 0.443024494142705,
+      "grad_norm": 0.3046875,
+      "learning_rate": 2.9394269168265358e-05,
+      "loss": 1.2662076950073242,
+      "step": 208
+    },
+    {
+      "epoch": 0.4472843450479233,
+      "grad_norm": 0.3046875,
+      "learning_rate": 2.9376479591927408e-05,
+      "loss": 1.2238541841506958,
+      "step": 210
+    },
+    {
+      "epoch": 0.45154419595314166,
+      "grad_norm": 0.263671875,
+      "learning_rate": 2.935843951677323e-05,
+      "loss": 1.2052886486053467,
+      "step": 212
+    },
+    {
+      "epoch": 0.45580404685835996,
+      "grad_norm": 0.3515625,
+      "learning_rate": 2.9340149340027412e-05,
+      "loss": 1.2680332660675049,
+      "step": 214
+    },
+    {
+      "epoch": 0.46006389776357826,
+      "grad_norm": 0.318359375,
+      "learning_rate": 2.9321609464421546e-05,
+      "loss": 1.233550786972046,
+      "step": 216
+    },
+    {
+      "epoch": 0.4643237486687966,
+      "grad_norm": 0.578125,
+      "learning_rate": 2.930282029818533e-05,
+      "loss": 1.2138795852661133,
+      "step": 218
+    },
+    {
+      "epoch": 0.4685835995740149,
+      "grad_norm": 0.2578125,
+      "learning_rate": 2.92837822550376e-05,
+      "loss": 1.1213371753692627,
+      "step": 220
+    },
+    {
+      "epoch": 0.4728434504792332,
+      "grad_norm": 0.404296875,
+      "learning_rate": 2.9264495754177225e-05,
+      "loss": 1.2127740383148193,
+      "step": 222
+    },
+    {
+      "epoch": 0.47710330138445156,
+      "grad_norm": 1.5078125,
+      "learning_rate": 2.924496122027384e-05,
+      "loss": 1.3384878635406494,
+      "step": 224
+    },
+    {
+      "epoch": 0.48136315228966986,
+      "grad_norm": 0.345703125,
+      "learning_rate": 2.9225179083458555e-05,
+      "loss": 1.1937229633331299,
+      "step": 226
+    },
+    {
+      "epoch": 0.48562300319488816,
+      "grad_norm": 0.52734375,
+      "learning_rate": 2.9205149779314425e-05,
+      "loss": 1.2608391046524048,
+      "step": 228
+    },
+    {
+      "epoch": 0.4898828541001065,
+      "grad_norm": 0.31640625,
+      "learning_rate": 2.918487374886691e-05,
+      "loss": 1.2325993776321411,
+      "step": 230
+    },
+    {
+      "epoch": 0.4941427050053248,
+      "grad_norm": 0.287109375,
+      "learning_rate": 2.91643514385741e-05,
+      "loss": 1.2050917148590088,
+      "step": 232
+    },
+    {
+      "epoch": 0.4984025559105431,
+      "grad_norm": 0.296875,
+      "learning_rate": 2.9143583300316975e-05,
+      "loss": 1.2299753427505493,
+      "step": 234
+    },
+    {
+      "epoch": 0.5026624068157615,
+      "grad_norm": 0.322265625,
+      "learning_rate": 2.9122569791389354e-05,
+      "loss": 1.2500553131103516,
+      "step": 236
+    },
+    {
+      "epoch": 0.5069222577209798,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.9101311374487908e-05,
+      "loss": 1.3044551610946655,
+      "step": 238
+    },
+    {
+      "epoch": 0.5111821086261981,
+      "grad_norm": 0.28125,
+      "learning_rate": 2.907980851770193e-05,
+      "loss": 1.1923537254333496,
+      "step": 240
+    },
+    {
+      "epoch": 0.5154419595314164,
+      "grad_norm": 0.419921875,
+      "learning_rate": 2.905806169450303e-05,
+      "loss": 1.2567352056503296,
+      "step": 242
+    },
+    {
+      "epoch": 0.5197018104366348,
+      "grad_norm": 0.369140625,
+      "learning_rate": 2.9036071383734716e-05,
+      "loss": 1.2812081575393677,
+      "step": 244
+    },
+    {
+      "epoch": 0.5239616613418531,
+      "grad_norm": 0.4609375,
+      "learning_rate": 2.9013838069601874e-05,
+      "loss": 1.2612706422805786,
+      "step": 246
+    },
+    {
+      "epoch": 0.5282215122470714,
+      "grad_norm": 0.27734375,
+      "learning_rate": 2.8991362241660053e-05,
+      "loss": 1.2162076234817505,
+      "step": 248
+    },
+    {
+      "epoch": 0.5324813631522897,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.8968644394804736e-05,
+      "loss": 1.2357534170150757,
+      "step": 250
+    },
+    {
+      "epoch": 0.536741214057508,
+      "grad_norm": 0.255859375,
+      "learning_rate": 2.894568502926042e-05,
+      "loss": 1.144363284111023,
+      "step": 252
+    },
+    {
+      "epoch": 0.5410010649627263,
+      "grad_norm": 0.68359375,
+      "learning_rate": 2.8922484650569597e-05,
+      "loss": 1.1998339891433716,
+      "step": 254
+    },
+    {
+      "epoch": 0.5452609158679447,
+      "grad_norm": 0.435546875,
+      "learning_rate": 2.8899043769581627e-05,
+      "loss": 1.1842751502990723,
+      "step": 256
+    },
+    {
+      "epoch": 0.549520766773163,
+      "grad_norm": 0.78515625,
+      "learning_rate": 2.8875362902441517e-05,
+      "loss": 1.1901715993881226,
+      "step": 258
+    },
+    {
+      "epoch": 0.5537806176783813,
+      "grad_norm": 0.447265625,
+      "learning_rate": 2.885144257057849e-05,
+      "loss": 1.3038347959518433,
+      "step": 260
+    },
+    {
+      "epoch": 0.5580404685835996,
+      "grad_norm": 0.46484375,
+      "learning_rate": 2.8827283300694593e-05,
+      "loss": 1.2350062131881714,
+      "step": 262
+    },
+    {
+      "epoch": 0.5623003194888179,
+      "grad_norm": 0.443359375,
+      "learning_rate": 2.8802885624753013e-05,
+      "loss": 1.2469710111618042,
+      "step": 264
+    },
+    {
+      "epoch": 0.5665601703940362,
+      "grad_norm": 0.38671875,
+      "learning_rate": 2.8778250079966417e-05,
+      "loss": 1.2484819889068604,
+      "step": 266
+    },
+    {
+      "epoch": 0.5708200212992546,
+      "grad_norm": 0.48046875,
+      "learning_rate": 2.875337720878512e-05,
+      "loss": 1.213232159614563,
+      "step": 268
+    },
+    {
+      "epoch": 0.5750798722044729,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.8728267558885102e-05,
+      "loss": 1.1985093355178833,
+      "step": 270
+    },
+    {
+      "epoch": 0.5793397231096912,
+      "grad_norm": 0.28125,
+      "learning_rate": 2.8702921683156e-05,
+      "loss": 1.2459266185760498,
+      "step": 272
+    },
+    {
+      "epoch": 0.5835995740149095,
+      "grad_norm": 0.251953125,
+      "learning_rate": 2.867734013968891e-05,
+      "loss": 1.3075346946716309,
+      "step": 274
+    },
+    {
+      "epoch": 0.5878594249201278,
+      "grad_norm": 0.63671875,
+      "learning_rate": 2.8651523491764074e-05,
+      "loss": 1.254473090171814,
+      "step": 276
+    },
+    {
+      "epoch": 0.5921192758253461,
+      "grad_norm": 0.5703125,
+      "learning_rate": 2.8625472307838518e-05,
+      "loss": 1.2639200687408447,
+      "step": 278
+    },
+    {
+      "epoch": 0.5963791267305645,
+      "grad_norm": 0.703125,
+      "learning_rate": 2.8599187161533533e-05,
+      "loss": 1.23056161403656,
+      "step": 280
+    },
+    {
+      "epoch": 0.6006389776357828,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.8572668631622e-05,
+      "loss": 1.2778501510620117,
+      "step": 282
+    },
+    {
+      "epoch": 0.6048988285410011,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.8545917302015693e-05,
+      "loss": 1.240308403968811,
+      "step": 284
+    },
+    {
+      "epoch": 0.6091586794462194,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.851893376175241e-05,
+      "loss": 1.3061432838439941,
+      "step": 286
+    },
+    {
+      "epoch": 0.6134185303514377,
+      "grad_norm": 0.326171875,
+      "learning_rate": 2.849171860498298e-05,
+      "loss": 1.1693536043167114,
+      "step": 288
+    },
+    {
+      "epoch": 0.617678381256656,
+      "grad_norm": 0.396484375,
+      "learning_rate": 2.8464272430958208e-05,
+      "loss": 1.3255276679992676,
+      "step": 290
+    },
+    {
+      "epoch": 0.6219382321618744,
+      "grad_norm": 0.4140625,
+      "learning_rate": 2.843659584401568e-05,
+      "loss": 1.1839312314987183,
+      "step": 292
+    },
+    {
+      "epoch": 0.6261980830670927,
+      "grad_norm": 0.400390625,
+      "learning_rate": 2.840868945356643e-05,
+      "loss": 1.2237545251846313,
+      "step": 294
+    },
+    {
+      "epoch": 0.630457933972311,
+      "grad_norm": 0.279296875,
+      "learning_rate": 2.8380553874081544e-05,
+      "loss": 1.219810962677002,
+      "step": 296
+    },
+    {
+      "epoch": 0.6347177848775293,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 2.8352189725078623e-05,
+      "loss": 1.148103952407837,
+      "step": 298
+    },
+    {
+      "epoch": 0.6389776357827476,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.8323597631108148e-05,
+      "loss": 1.266182780265808,
+      "step": 300
+    },
+    {
+      "epoch": 0.6432374866879659,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.829477822173972e-05,
+      "loss": 1.1832197904586792,
+      "step": 302
+    },
+    {
+      "epoch": 0.6474973375931843,
+      "grad_norm": 0.322265625,
+      "learning_rate": 2.8265732131548185e-05,
+      "loss": 1.2743726968765259,
+      "step": 304
+    },
+    {
+      "epoch": 0.6517571884984026,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 2.82364600000997e-05,
+      "loss": 1.2408907413482666,
+      "step": 306
+    },
+    {
+      "epoch": 0.6560170394036209,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.8206962471937612e-05,
+      "loss": 1.2314817905426025,
+      "step": 308
+    },
+    {
+      "epoch": 0.6602768903088392,
+      "grad_norm": 0.279296875,
+      "learning_rate": 2.817724019656829e-05,
+      "loss": 1.0730669498443604,
+      "step": 310
+    },
+    {
+      "epoch": 0.6645367412140575,
+      "grad_norm": 0.3203125,
+      "learning_rate": 2.81472938284468e-05,
+      "loss": 1.250943660736084,
+      "step": 312
+    },
+    {
+      "epoch": 0.6687965921192758,
+      "grad_norm": 0.98046875,
+      "learning_rate": 2.811712402696252e-05,
+      "loss": 1.1586111783981323,
+      "step": 314
+    },
+    {
+      "epoch": 0.6730564430244942,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.808673145642461e-05,
+      "loss": 1.2091357707977295,
+      "step": 316
+    },
+    {
+      "epoch": 0.6773162939297125,
+      "grad_norm": 0.7890625,
+      "learning_rate": 2.805611678604737e-05,
+      "loss": 1.219393253326416,
+      "step": 318
+    },
+    {
+      "epoch": 0.6815761448349308,
+      "grad_norm": 0.302734375,
+      "learning_rate": 2.8025280689935538e-05,
+      "loss": 1.2416179180145264,
+      "step": 320
+    },
+    {
+      "epoch": 0.6858359957401491,
+      "grad_norm": 0.46875,
+      "learning_rate": 2.7994223847069417e-05,
+      "loss": 1.2236298322677612,
+      "step": 322
+    },
+    {
+      "epoch": 0.6900958466453674,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.7962946941289932e-05,
+      "loss": 1.1898835897445679,
+      "step": 324
+    },
+    {
+      "epoch": 0.6943556975505857,
+      "grad_norm": 0.357421875,
+      "learning_rate": 2.7931450661283587e-05,
+      "loss": 1.1595722436904907,
+      "step": 326
+    },
+    {
+      "epoch": 0.6986155484558041,
+      "grad_norm": 0.53125,
+      "learning_rate": 2.7899735700567272e-05,
+      "loss": 1.221711277961731,
+      "step": 328
+    },
+    {
+      "epoch": 0.7028753993610224,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 2.7867802757473023e-05,
+      "loss": 1.2105400562286377,
+      "step": 330
+    },
+    {
+      "epoch": 0.7071352502662407,
+      "grad_norm": 0.30078125,
+      "learning_rate": 2.7835652535132635e-05,
+      "loss": 1.2640867233276367,
+      "step": 332
+    },
+    {
+      "epoch": 0.711395101171459,
+      "grad_norm": 0.330078125,
+      "learning_rate": 2.780328574146216e-05,
+      "loss": 1.259413480758667,
+      "step": 334
+    },
+    {
+      "epoch": 0.7156549520766773,
+      "grad_norm": 0.4453125,
+      "learning_rate": 2.7770703089146355e-05,
+      "loss": 1.3237056732177734,
+      "step": 336
+    },
+    {
+      "epoch": 0.7199148029818956,
+      "grad_norm": 0.400390625,
+      "learning_rate": 2.7737905295622957e-05,
+      "loss": 1.2199316024780273,
+      "step": 338
+    },
+    {
+      "epoch": 0.724174653887114,
+      "grad_norm": 0.51953125,
+      "learning_rate": 2.7704893083066906e-05,
+      "loss": 1.1969261169433594,
+      "step": 340
+    },
+    {
+      "epoch": 0.7284345047923323,
+      "grad_norm": 0.466796875,
+      "learning_rate": 2.7671667178374443e-05,
+      "loss": 1.2693402767181396,
+      "step": 342
+    },
+    {
+      "epoch": 0.7326943556975506,
+      "grad_norm": 0.265625,
+      "learning_rate": 2.7638228313147083e-05,
+      "loss": 1.230875015258789,
+      "step": 344
+    },
+    {
+      "epoch": 0.7369542066027689,
+      "grad_norm": 0.375,
+      "learning_rate": 2.760457722367553e-05,
+      "loss": 1.1558018922805786,
+      "step": 346
+    },
+    {
+      "epoch": 0.7412140575079872,
+      "grad_norm": 2.40625,
+      "learning_rate": 2.7570714650923446e-05,
+      "loss": 1.3312543630599976,
+      "step": 348
+    },
+    {
+      "epoch": 0.7454739084132055,
+      "grad_norm": 0.283203125,
+      "learning_rate": 2.7536641340511177e-05,
+      "loss": 1.1423282623291016,
+      "step": 350
+    },
+    {
+      "epoch": 0.7497337593184239,
+      "grad_norm": 0.41796875,
+      "learning_rate": 2.7502358042699257e-05,
+      "loss": 1.1751903295516968,
+      "step": 352
+    },
+    {
+      "epoch": 0.7539936102236422,
+      "grad_norm": 0.384765625,
+      "learning_rate": 2.7467865512371974e-05,
+      "loss": 1.2713823318481445,
+      "step": 354
+    },
+    {
+      "epoch": 0.7582534611288605,
+      "grad_norm": 0.330078125,
+      "learning_rate": 2.7433164509020684e-05,
+      "loss": 1.2887362241744995,
+      "step": 356
+    },
+    {
+      "epoch": 0.7625133120340788,
+      "grad_norm": 0.341796875,
+      "learning_rate": 2.7398255796727127e-05,
+      "loss": 1.2369112968444824,
+      "step": 358
+    },
+    {
+      "epoch": 0.7667731629392971,
+      "grad_norm": 0.34375,
+      "learning_rate": 2.7363140144146578e-05,
+      "loss": 1.150454044342041,
+      "step": 360
+    },
+    {
+      "epoch": 0.7710330138445154,
+      "grad_norm": 0.275390625,
+      "learning_rate": 2.7327818324490938e-05,
+      "loss": 1.2185767889022827,
+      "step": 362
+    },
+    {
+      "epoch": 0.7752928647497338,
+      "grad_norm": 0.6953125,
+      "learning_rate": 2.729229111551171e-05,
+      "loss": 1.2292591333389282,
+      "step": 364
+    },
+    {
+      "epoch": 0.7795527156549521,
+      "grad_norm": 0.236328125,
+      "learning_rate": 2.725655929948285e-05,
+      "loss": 1.2185684442520142,
+      "step": 366
+    },
+    {
+      "epoch": 0.7838125665601704,
+      "grad_norm": 0.30078125,
+      "learning_rate": 2.722062366318357e-05,
+      "loss": 1.1981046199798584,
+      "step": 368
+    },
+    {
+      "epoch": 0.7880724174653887,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 2.7184484997881e-05,
+      "loss": 1.1411432027816772,
+      "step": 370
+    },
+    {
+      "epoch": 0.792332268370607,
+      "grad_norm": 0.85546875,
+      "learning_rate": 2.7148144099312765e-05,
+      "loss": 1.2738561630249023,
+      "step": 372
+    },
+    {
+      "epoch": 0.7965921192758253,
+      "grad_norm": 0.275390625,
+      "learning_rate": 2.7111601767669473e-05,
+      "loss": 1.1942780017852783,
+      "step": 374
+    },
+    {
+      "epoch": 0.8008519701810437,
+      "grad_norm": 0.212890625,
+      "learning_rate": 2.7074858807577084e-05,
+      "loss": 1.1684967279434204,
+      "step": 376
+    },
+    {
+      "epoch": 0.805111821086262,
+      "grad_norm": 0.171875,
+      "learning_rate": 2.7037916028079198e-05,
+      "loss": 1.1836313009262085,
+      "step": 378
+    },
+    {
+      "epoch": 0.8093716719914803,
+      "grad_norm": 0.365234375,
+      "learning_rate": 2.7000774242619235e-05,
+      "loss": 1.2047457695007324,
+      "step": 380
+    },
+    {
+      "epoch": 0.8136315228966986,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 2.696343426902254e-05,
+      "loss": 1.186992883682251,
+      "step": 382
+    },
+    {
+      "epoch": 0.8178913738019169,
+      "grad_norm": 0.376953125,
+      "learning_rate": 2.6925896929478355e-05,
+      "loss": 1.1887181997299194,
+      "step": 384
+    },
+    {
+      "epoch": 0.8221512247071352,
+      "grad_norm": 0.328125,
+      "learning_rate": 2.6888163050521734e-05,
+      "loss": 1.2181212902069092,
+      "step": 386
+    },
+    {
+      "epoch": 0.8264110756123536,
+      "grad_norm": 0.55078125,
+      "learning_rate": 2.6850233463015334e-05,
+      "loss": 1.1820951700210571,
+      "step": 388
+    },
+    {
+      "epoch": 0.8306709265175719,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.6812109002131106e-05,
+      "loss": 1.1575113534927368,
+      "step": 390
+    },
+    {
+      "epoch": 0.8349307774227902,
+      "grad_norm": 0.40625,
+      "learning_rate": 2.6773790507331936e-05,
+      "loss": 1.1017088890075684,
+      "step": 392
+    },
+    {
+      "epoch": 0.8391906283280085,
+      "grad_norm": 0.310546875,
+      "learning_rate": 2.673527882235314e-05,
+      "loss": 1.1889958381652832,
+      "step": 394
+    },
+    {
+      "epoch": 0.8434504792332268,
+      "grad_norm": 0.373046875,
+      "learning_rate": 2.6696574795183882e-05,
+      "loss": 1.1406269073486328,
+      "step": 396
+    },
+    {
+      "epoch": 0.8477103301384451,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.665767927804852e-05,
+      "loss": 1.172967791557312,
+      "step": 398
+    },
+    {
+      "epoch": 0.8519701810436635,
+      "grad_norm": 0.23046875,
+      "learning_rate": 2.661859312738783e-05,
+      "loss": 1.2290892601013184,
+      "step": 400
+    },
+    {
+      "epoch": 0.8562300319488818,
+      "grad_norm": 0.255859375,
+      "learning_rate": 2.6579317203840154e-05,
+      "loss": 1.0655782222747803,
+      "step": 402
+    },
+    {
+      "epoch": 0.8604898828541001,
+      "grad_norm": 0.208984375,
+      "learning_rate": 2.6539852372222434e-05,
+      "loss": 1.1730587482452393,
+      "step": 404
+    },
+    {
+      "epoch": 0.8647497337593184,
+      "grad_norm": 0.283203125,
+      "learning_rate": 2.6500199501511184e-05,
+      "loss": 1.2667183876037598,
+      "step": 406
+    },
+    {
+      "epoch": 0.8690095846645367,
+      "grad_norm": 0.3125,
+      "learning_rate": 2.646035946482336e-05,
+      "loss": 1.2611602544784546,
+      "step": 408
+    },
+    {
+      "epoch": 0.873269435569755,
+      "grad_norm": 0.234375,
+      "learning_rate": 2.6420333139397122e-05,
+      "loss": 1.2684861421585083,
+      "step": 410
+    },
+    {
+      "epoch": 0.8775292864749734,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 2.638012140657252e-05,
+      "loss": 1.2144488096237183,
+      "step": 412
+    },
+    {
+      "epoch": 0.8817891373801917,
+      "grad_norm": 0.44921875,
+      "learning_rate": 2.6339725151772095e-05,
+      "loss": 1.2024558782577515,
+      "step": 414
+    },
+    {
+      "epoch": 0.88604898828541,
+      "grad_norm": 0.478515625,
+      "learning_rate": 2.6299145264481386e-05,
+      "loss": 1.2472572326660156,
+      "step": 416
+    },
+    {
+      "epoch": 0.8903088391906283,
+      "grad_norm": 0.310546875,
+      "learning_rate": 2.625838263822932e-05,
+      "loss": 1.15989351272583,
+      "step": 418
+    },
+    {
+      "epoch": 0.8945686900958466,
+      "grad_norm": 0.236328125,
+      "learning_rate": 2.621743817056858e-05,
+      "loss": 1.2214092016220093,
+      "step": 420
+    },
+    {
+      "epoch": 0.898828541001065,
+      "grad_norm": 0.37109375,
+      "learning_rate": 2.6176312763055795e-05,
+      "loss": 1.1031744480133057,
+      "step": 422
+    },
+    {
+      "epoch": 0.9030883919062833,
+      "grad_norm": 0.251953125,
+      "learning_rate": 2.6135007321231715e-05,
+      "loss": 1.0990759134292603,
+      "step": 424
+    },
+    {
+      "epoch": 0.9073482428115016,
+      "grad_norm": 0.41015625,
+      "learning_rate": 2.6093522754601284e-05,
+      "loss": 1.180249810218811,
+      "step": 426
+    },
+    {
+      "epoch": 0.9116080937167199,
+      "grad_norm": 0.185546875,
+      "learning_rate": 2.6051859976613564e-05,
+      "loss": 1.1679967641830444,
+      "step": 428
+    },
+    {
+      "epoch": 0.9158679446219382,
+      "grad_norm": 0.294921875,
+      "learning_rate": 2.601001990464169e-05,
+      "loss": 1.1675636768341064,
+      "step": 430
+    },
+    {
+      "epoch": 0.9201277955271565,
+      "grad_norm": 0.93359375,
+      "learning_rate": 2.5968003459962608e-05,
+      "loss": 1.187214732170105,
+      "step": 432
+    },
+    {
+      "epoch": 0.9243876464323749,
+      "grad_norm": 0.23828125,
+      "learning_rate": 2.592581156773684e-05,
+      "loss": 1.1574485301971436,
+      "step": 434
+    },
+    {
+      "epoch": 0.9286474973375932,
+      "grad_norm": 0.33203125,
+      "learning_rate": 2.588344515698806e-05,
+      "loss": 1.207824468612671,
+      "step": 436
+    },
+    {
+      "epoch": 0.9329073482428115,
+      "grad_norm": 0.275390625,
+      "learning_rate": 2.58409051605827e-05,
+      "loss": 1.160508155822754,
+      "step": 438
+    },
+    {
+      "epoch": 0.9371671991480298,
+      "grad_norm": 0.3125,
+      "learning_rate": 2.5798192515209343e-05,
+      "loss": 1.1380846500396729,
+      "step": 440
+    },
+    {
+      "epoch": 0.9414270500532481,
+      "grad_norm": 0.1884765625,
+      "learning_rate": 2.5755308161358166e-05,
+      "loss": 1.1430374383926392,
+      "step": 442
+    },
+    {
+      "epoch": 0.9456869009584664,
+      "grad_norm": 1.0625,
+      "learning_rate": 2.5712253043300174e-05,
+      "loss": 1.1965644359588623,
+      "step": 444
+    },
+    {
+      "epoch": 0.9499467518636848,
+      "grad_norm": 0.302734375,
+      "learning_rate": 2.5669028109066426e-05,
+      "loss": 1.2050869464874268,
+      "step": 446
+    },
+    {
+      "epoch": 0.9542066027689031,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.5625634310427188e-05,
+      "loss": 1.1945817470550537,
+      "step": 448
+    },
+    {
+      "epoch": 0.9584664536741214,
+      "grad_norm": 0.275390625,
+      "learning_rate": 2.558207260287093e-05,
+      "loss": 1.1947966814041138,
+      "step": 450
+    },
+    {
+      "epoch": 0.9627263045793397,
+      "grad_norm": 0.279296875,
+      "learning_rate": 2.553834394558332e-05,
+      "loss": 1.134352445602417,
+      "step": 452
+    },
+    {
+      "epoch": 0.966986155484558,
+      "grad_norm": 1.0859375,
+      "learning_rate": 2.5494449301426102e-05,
+      "loss": 1.2251217365264893,
+      "step": 454
+    },
+    {
+      "epoch": 0.9712460063897763,
+      "grad_norm": 0.2177734375,
+      "learning_rate": 2.5450389636915867e-05,
+      "loss": 1.081860899925232,
+      "step": 456
+    },
+    {
+      "epoch": 0.9755058572949947,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 2.540616592220281e-05,
+      "loss": 1.182367205619812,
+      "step": 458
+    },
+    {
+      "epoch": 0.979765708200213,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 2.5361779131049344e-05,
+      "loss": 1.158174991607666,
+      "step": 460
+    },
+    {
+      "epoch": 0.9840255591054313,
+      "grad_norm": 0.2001953125,
+      "learning_rate": 2.5317230240808656e-05,
+      "loss": 1.1436811685562134,
+      "step": 462
+    },
+    {
+      "epoch": 0.9882854100106496,
+      "grad_norm": 0.31640625,
+      "learning_rate": 2.527252023240319e-05,
+      "loss": 1.1009982824325562,
+      "step": 464
+    },
+    {
+      "epoch": 0.9925452609158679,
+      "grad_norm": 0.39453125,
+      "learning_rate": 2.5227650090303083e-05,
+      "loss": 1.2242732048034668,
+      "step": 466
+    },
+    {
+      "epoch": 0.9968051118210862,
+      "grad_norm": 0.255859375,
+      "learning_rate": 2.5182620802504415e-05,
+      "loss": 1.1412031650543213,
+      "step": 468
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.453125,
+      "learning_rate": 2.513743336050753e-05,
+      "loss": 1.3368866443634033,
+      "step": 470
+    },
+    {
+      "epoch": 1.0042598509052183,
+      "grad_norm": 0.23828125,
+      "learning_rate": 2.5092088759295147e-05,
+      "loss": 0.9358726739883423,
+      "step": 472
+    },
+    {
+      "epoch": 1.0085197018104366,
+      "grad_norm": 0.189453125,
+      "learning_rate": 2.5046587997310503e-05,
+      "loss": 0.9842238426208496,
+      "step": 474
+    },
+    {
+      "epoch": 1.012779552715655,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 2.500093207643532e-05,
+      "loss": 0.909864068031311,
+      "step": 476
+    },
+    {
+      "epoch": 1.0170394036208732,
+      "grad_norm": 0.248046875,
+      "learning_rate": 2.4955122001967757e-05,
+      "loss": 0.8217376470565796,
+      "step": 478
+    },
+    {
+      "epoch": 1.0212992545260915,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 2.4909158782600303e-05,
+      "loss": 0.9412868618965149,
+      "step": 480
+    },
+    {
+      "epoch": 1.0255591054313098,
+      "grad_norm": 0.423828125,
+      "learning_rate": 2.4863043430397546e-05,
+      "loss": 0.9232436418533325,
+      "step": 482
+    },
+    {
+      "epoch": 1.0298189563365283,
+      "grad_norm": 0.1708984375,
+      "learning_rate": 2.481677696077387e-05,
+      "loss": 0.9075867533683777,
+      "step": 484
+    },
+    {
+      "epoch": 1.0340788072417466,
+      "grad_norm": 0.3203125,
+      "learning_rate": 2.477036039247113e-05,
+      "loss": 0.9229554533958435,
+      "step": 486
+    },
+    {
+      "epoch": 1.038338658146965,
+      "grad_norm": 0.28515625,
+      "learning_rate": 2.4723794747536204e-05,
+      "loss": 0.8909753561019897,
+      "step": 488
+    },
+    {
+      "epoch": 1.0425985090521832,
+      "grad_norm": 0.2578125,
+      "learning_rate": 2.4677081051298473e-05,
+      "loss": 0.8516156077384949,
+      "step": 490
+    },
+    {
+      "epoch": 1.0468583599574015,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.4630220332347293e-05,
+      "loss": 0.910189151763916,
+      "step": 492
+    },
+    {
+      "epoch": 1.0511182108626198,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 2.458321362250928e-05,
+      "loss": 0.8809674978256226,
+      "step": 494
+    },
+    {
+      "epoch": 1.055378061767838,
+      "grad_norm": 0.337890625,
+      "learning_rate": 2.4536061956825653e-05,
+      "loss": 0.9545248746871948,
+      "step": 496
+    },
+    {
+      "epoch": 1.0596379126730564,
+      "grad_norm": 0.359375,
+      "learning_rate": 2.44887663735294e-05,
+      "loss": 0.8128166794776917,
+      "step": 498
+    },
+    {
+      "epoch": 1.0638977635782747,
+      "grad_norm": 0.271484375,
+      "learning_rate": 2.4441327914022435e-05,
+      "loss": 0.7933678030967712,
+      "step": 500
+    },
+    {
+      "epoch": 1.068157614483493,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.4393747622852666e-05,
+      "loss": 0.845329761505127,
+      "step": 502
+    },
+    {
+      "epoch": 1.0724174653887113,
+      "grad_norm": 0.291015625,
+      "learning_rate": 2.4346026547690983e-05,
+      "loss": 0.8825768232345581,
+      "step": 504
+    },
+    {
+      "epoch": 1.0766773162939298,
+      "grad_norm": 0.322265625,
+      "learning_rate": 2.4298165739308227e-05,
+      "loss": 0.9173828959465027,
+      "step": 506
+    },
+    {
+      "epoch": 1.0809371671991481,
+      "grad_norm": 0.625,
+      "learning_rate": 2.4250166251551998e-05,
+      "loss": 0.9571421146392822,
+      "step": 508
+    },
+    {
+      "epoch": 1.0851970181043664,
+      "grad_norm": 0.29296875,
+      "learning_rate": 2.4202029141323492e-05,
+      "loss": 0.8474833369255066,
+      "step": 510
+    },
+    {
+      "epoch": 1.0894568690095847,
+      "grad_norm": 0.34765625,
+      "learning_rate": 2.415375546855422e-05,
+      "loss": 0.8801344633102417,
+      "step": 512
+    },
+    {
+      "epoch": 1.093716719914803,
+      "grad_norm": 0.33984375,
+      "learning_rate": 2.4105346296182648e-05,
+      "loss": 0.8761341571807861,
+      "step": 514
+    },
+    {
+      "epoch": 1.0979765708200213,
+      "grad_norm": 0.921875,
+      "learning_rate": 2.4056802690130826e-05,
+      "loss": 0.8511140942573547,
+      "step": 516
+    },
+    {
+      "epoch": 1.1022364217252396,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 2.4008125719280893e-05,
+      "loss": 0.8243319392204285,
+      "step": 518
+    },
+    {
+      "epoch": 1.106496272630458,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 2.395931645545155e-05,
+      "loss": 0.9023821949958801,
+      "step": 520
+    },
+    {
+      "epoch": 1.1107561235356762,
+      "grad_norm": 0.5859375,
+      "learning_rate": 2.391037597337446e-05,
+      "loss": 0.8977804183959961,
+      "step": 522
+    },
+    {
+      "epoch": 1.1150159744408945,
+      "grad_norm": 0.271484375,
+      "learning_rate": 2.3861305350670564e-05,
+      "loss": 0.8644490242004395,
+      "step": 524
+    },
+    {
+      "epoch": 1.1192758253461128,
+      "grad_norm": 0.220703125,
+      "learning_rate": 2.381210566782642e-05,
+      "loss": 0.8652825951576233,
+      "step": 526
+    },
+    {
+      "epoch": 1.123535676251331,
+      "grad_norm": 0.48828125,
+      "learning_rate": 2.3762778008170296e-05,
+      "loss": 0.9315000176429749,
+      "step": 528
+    },
+    {
+      "epoch": 1.1277955271565494,
+      "grad_norm": 0.298828125,
+      "learning_rate": 2.3713323457848425e-05,
+      "loss": 0.8627546429634094,
+      "step": 530
+    },
+    {
+      "epoch": 1.132055378061768,
+      "grad_norm": 0.3515625,
+      "learning_rate": 2.366374310580106e-05,
+      "loss": 0.8466436266899109,
+      "step": 532
+    },
+    {
+      "epoch": 1.1363152289669862,
+      "grad_norm": 0.234375,
+      "learning_rate": 2.3614038043738432e-05,
+      "loss": 0.8433495163917542,
+      "step": 534
+    },
+    {
+      "epoch": 1.1405750798722045,
+      "grad_norm": 0.671875,
+      "learning_rate": 2.35642093661168e-05,
+      "loss": 0.9653686285018921,
+      "step": 536
+    },
+    {
+      "epoch": 1.1448349307774228,
+      "grad_norm": 0.46484375,
+      "learning_rate": 2.351425817011432e-05,
+      "loss": 0.9155454039573669,
+      "step": 538
+    },
+    {
+      "epoch": 1.1490947816826411,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 2.3464185555606854e-05,
+      "loss": 0.8044310212135315,
+      "step": 540
+    },
+    {
+      "epoch": 1.1533546325878594,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 2.3413992625143808e-05,
+      "loss": 0.8448784947395325,
+      "step": 542
+    },
+    {
+      "epoch": 1.1576144834930777,
+      "grad_norm": 0.51171875,
+      "learning_rate": 2.3363680483923794e-05,
+      "loss": 0.9145954251289368,
+      "step": 544
+    },
+    {
+      "epoch": 1.161874334398296,
+      "grad_norm": 0.251953125,
+      "learning_rate": 2.3313250239770364e-05,
+      "loss": 0.8059402108192444,
+      "step": 546
+    },
+    {
+      "epoch": 1.1661341853035143,
+      "grad_norm": 0.32421875,
+      "learning_rate": 2.326270300310756e-05,
+      "loss": 0.910370945930481,
+      "step": 548
+    },
+    {
+      "epoch": 1.1703940362087326,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 2.3212039886935464e-05,
+      "loss": 0.8459041118621826,
+      "step": 550
+    },
+    {
+      "epoch": 1.174653887113951,
+      "grad_norm": 0.23046875,
+      "learning_rate": 2.3161262006805744e-05,
+      "loss": 0.8679651618003845,
+      "step": 552
+    },
+    {
+      "epoch": 1.1789137380191694,
+      "grad_norm": 0.353515625,
+      "learning_rate": 2.3110370480797046e-05,
+      "loss": 0.8998923897743225,
+      "step": 554
+    },
+    {
+      "epoch": 1.1831735889243877,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.3059366429490382e-05,
+      "loss": 0.9410486817359924,
+      "step": 556
+    },
+    {
+      "epoch": 1.187433439829606,
+      "grad_norm": 0.294921875,
+      "learning_rate": 2.3008250975944458e-05,
+      "loss": 0.8485605120658875,
+      "step": 558
+    },
+    {
+      "epoch": 1.1916932907348243,
+      "grad_norm": 0.240234375,
+      "learning_rate": 2.2957025245670945e-05,
+      "loss": 0.8777744770050049,
+      "step": 560
+    },
+    {
+      "epoch": 1.1959531416400426,
+      "grad_norm": 0.40234375,
+      "learning_rate": 2.2905690366609703e-05,
+      "loss": 0.9006752967834473,
+      "step": 562
+    },
+    {
+      "epoch": 1.200212992545261,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.2854247469103943e-05,
+      "loss": 0.8309807181358337,
+      "step": 564
+    },
+    {
+      "epoch": 1.2044728434504792,
+      "grad_norm": 0.26953125,
+      "learning_rate": 2.280269768587534e-05,
+      "loss": 0.9057250618934631,
+      "step": 566
+    },
+    {
+      "epoch": 1.2087326943556975,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 2.2751042151999064e-05,
+      "loss": 0.829549252986908,
+      "step": 568
+    },
+    {
+      "epoch": 1.2129925452609158,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.2699282004878834e-05,
+      "loss": 0.9091805219650269,
+      "step": 570
+    },
+    {
+      "epoch": 1.2172523961661341,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 2.264741838422183e-05,
+      "loss": 0.8178958296775818,
+      "step": 572
+    },
+    {
+      "epoch": 1.2215122470713524,
+      "grad_norm": 0.62109375,
+      "learning_rate": 2.2595452432013637e-05,
+      "loss": 0.9319694638252258,
+      "step": 574
+    },
+    {
+      "epoch": 1.225772097976571,
+      "grad_norm": 0.255859375,
+      "learning_rate": 2.2543385292493068e-05,
+      "loss": 0.8054318428039551,
+      "step": 576
+    },
+    {
+      "epoch": 1.230031948881789,
+      "grad_norm": 0.37890625,
+      "learning_rate": 2.2491218112126974e-05,
+      "loss": 0.8717759847640991,
+      "step": 578
+    },
+    {
+      "epoch": 1.2342917997870075,
+      "grad_norm": 0.28515625,
+      "learning_rate": 2.2438952039585023e-05,
+      "loss": 0.9084351062774658,
+      "step": 580
+    },
+    {
+      "epoch": 1.2385516506922258,
+      "grad_norm": 0.25390625,
+      "learning_rate": 2.238658822571437e-05,
+      "loss": 0.877246618270874,
+      "step": 582
+    },
+    {
+      "epoch": 1.2428115015974441,
+      "grad_norm": 0.380859375,
+      "learning_rate": 2.2334127823514353e-05,
+      "loss": 0.8917878866195679,
+      "step": 584
+    },
+    {
+      "epoch": 1.2470713525026624,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 2.2281571988111087e-05,
+      "loss": 0.9018102884292603,
+      "step": 586
+    },
+    {
+      "epoch": 1.2513312034078807,
+      "grad_norm": 0.181640625,
+      "learning_rate": 2.222892187673203e-05,
+      "loss": 0.8929234147071838,
+      "step": 588
+    },
+    {
+      "epoch": 1.255591054313099,
+      "grad_norm": 0.291015625,
+      "learning_rate": 2.2176178648680504e-05,
+      "loss": 0.9248031973838806,
+      "step": 590
+    },
+    {
+      "epoch": 1.2598509052183173,
+      "grad_norm": 0.275390625,
+      "learning_rate": 2.2123343465310163e-05,
+      "loss": 0.9204663038253784,
+      "step": 592
+    },
+    {
+      "epoch": 1.2641107561235356,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 2.2070417489999427e-05,
+      "loss": 0.8040061593055725,
+      "step": 594
+    },
+    {
+      "epoch": 1.268370607028754,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.201740188812588e-05,
+      "loss": 0.9146944880485535,
+      "step": 596
+    },
+    {
+      "epoch": 1.2726304579339724,
+      "grad_norm": 0.2578125,
+      "learning_rate": 2.196429782704057e-05,
+      "loss": 0.8526248931884766,
+      "step": 598
+    },
+    {
+      "epoch": 1.2768903088391905,
+      "grad_norm": 0.328125,
+      "learning_rate": 2.191110647604235e-05,
+      "loss": 0.8366101384162903,
+      "step": 600
+    },
+    {
+      "epoch": 1.281150159744409,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 2.1857829006352092e-05,
+      "loss": 0.8716267347335815,
+      "step": 602
+    },
+    {
+      "epoch": 1.2854100106496273,
+      "grad_norm": 0.333984375,
+      "learning_rate": 2.180446659108693e-05,
+      "loss": 0.9040926694869995,
+      "step": 604
+    },
+    {
+      "epoch": 1.2896698615548456,
+      "grad_norm": 0.251953125,
+      "learning_rate": 2.1751020405234427e-05,
+      "loss": 0.8583382368087769,
+      "step": 606
+    },
+    {
+      "epoch": 1.293929712460064,
+      "grad_norm": 0.314453125,
+      "learning_rate": 2.1697491625626652e-05,
+      "loss": 0.8685941696166992,
+      "step": 608
+    },
+    {
+      "epoch": 1.2981895633652822,
+      "grad_norm": 0.21484375,
+      "learning_rate": 2.1643881430914343e-05,
+      "loss": 0.8654310703277588,
+      "step": 610
+    },
+    {
+      "epoch": 1.3024494142705005,
+      "grad_norm": 0.2265625,
+      "learning_rate": 2.1590191001540903e-05,
+      "loss": 0.8943390846252441,
+      "step": 612
+    },
+    {
+      "epoch": 1.3067092651757188,
+      "grad_norm": 0.26171875,
+      "learning_rate": 2.153642151971643e-05,
+      "loss": 0.8576252460479736,
+      "step": 614
+    },
+    {
+      "epoch": 1.3109691160809371,
+      "grad_norm": 0.4375,
+      "learning_rate": 2.1482574169391664e-05,
+      "loss": 0.8761968612670898,
+      "step": 616
+    },
+    {
+      "epoch": 1.3152289669861554,
+      "grad_norm": 0.232421875,
+      "learning_rate": 2.1428650136231948e-05,
+      "loss": 0.8207455277442932,
+      "step": 618
+    },
+    {
+      "epoch": 1.3194888178913737,
+      "grad_norm": 0.22265625,
+      "learning_rate": 2.1374650607591106e-05,
+      "loss": 0.8694437742233276,
+      "step": 620
+    },
+    {
+      "epoch": 1.323748668796592,
+      "grad_norm": 0.408203125,
+      "learning_rate": 2.1320576772485284e-05,
+      "loss": 0.872995138168335,
+      "step": 622
+    },
+    {
+      "epoch": 1.3280085197018106,
+      "grad_norm": 0.30859375,
+      "learning_rate": 2.126642982156679e-05,
+      "loss": 0.9666632413864136,
+      "step": 624
+    },
+    {
+      "epoch": 1.3322683706070286,
+      "grad_norm": 0.3828125,
+      "learning_rate": 2.1212210947097873e-05,
+      "loss": 0.8025370836257935,
+      "step": 626
+    },
+    {
+      "epoch": 1.3365282215122471,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.1157921342924457e-05,
+      "loss": 0.8531129956245422,
+      "step": 628
+    },
+    {
+      "epoch": 1.3407880724174654,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.1103562204449876e-05,
+      "loss": 0.8310921788215637,
+      "step": 630
+    },
+    {
+      "epoch": 1.3450479233226837,
+      "grad_norm": 0.291015625,
+      "learning_rate": 2.1049134728608537e-05,
+      "loss": 0.903289794921875,
+      "step": 632
+    },
+    {
+      "epoch": 1.349307774227902,
+      "grad_norm": 0.28125,
+      "learning_rate": 2.0994640113839568e-05,
+      "loss": 0.8707770705223083,
+      "step": 634
+    },
+    {
+      "epoch": 1.3535676251331203,
+      "grad_norm": 0.267578125,
+      "learning_rate": 2.0940079560060427e-05,
+      "loss": 0.8999609351158142,
+      "step": 636
+    },
+    {
+      "epoch": 1.3578274760383386,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 2.088545426864048e-05,
+      "loss": 0.8670209646224976,
+      "step": 638
+    },
+    {
+      "epoch": 1.362087326943557,
+      "grad_norm": 0.349609375,
+      "learning_rate": 2.0830765442374563e-05,
+      "loss": 0.8102102279663086,
+      "step": 640
+    },
+    {
+      "epoch": 1.3663471778487752,
+      "grad_norm": 0.2041015625,
+      "learning_rate": 2.077601428545648e-05,
+      "loss": 0.8202542662620544,
+      "step": 642
+    },
+    {
+      "epoch": 1.3706070287539935,
+      "grad_norm": 0.234375,
+      "learning_rate": 2.0721202003452496e-05,
+      "loss": 0.8944796323776245,
+      "step": 644
+    },
+    {
+      "epoch": 1.374866879659212,
+      "grad_norm": 0.208984375,
+      "learning_rate": 2.066632980327478e-05,
+      "loss": 0.9467480778694153,
+      "step": 646
+    },
+    {
+      "epoch": 1.3791267305644301,
+      "grad_norm": 0.482421875,
+      "learning_rate": 2.061139889315486e-05,
+      "loss": 0.8729652762413025,
+      "step": 648
+    },
+    {
+      "epoch": 1.3833865814696487,
+      "grad_norm": 0.275390625,
+      "learning_rate": 2.0556410482616977e-05,
+      "loss": 0.8954660892486572,
+      "step": 650
+    },
+    {
+      "epoch": 1.387646432374867,
+      "grad_norm": 0.2734375,
+      "learning_rate": 2.050136578245149e-05,
+      "loss": 0.870725691318512,
+      "step": 652
+    },
+    {
+      "epoch": 1.3919062832800853,
+      "grad_norm": 0.251953125,
+      "learning_rate": 2.0446266004688197e-05,
+      "loss": 0.8651110529899597,
+      "step": 654
+    },
+    {
+      "epoch": 1.3961661341853036,
+      "grad_norm": 0.240234375,
+      "learning_rate": 2.039111236256964e-05,
+      "loss": 0.8937119841575623,
+      "step": 656
+    },
+    {
+      "epoch": 1.4004259850905219,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 2.0335906070524416e-05,
+      "loss": 0.8803120851516724,
+      "step": 658
+    },
+    {
+      "epoch": 1.4046858359957402,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 2.02806483441404e-05,
+      "loss": 0.8514755368232727,
+      "step": 660
+    },
+    {
+      "epoch": 1.4089456869009584,
+      "grad_norm": 0.1630859375,
+      "learning_rate": 2.0225340400138033e-05,
+      "loss": 0.8654860258102417,
+      "step": 662
+    },
+    {
+      "epoch": 1.4132055378061767,
+      "grad_norm": 0.1904296875,
+      "learning_rate": 2.0169983456343464e-05,
+      "loss": 0.861249566078186,
+      "step": 664
+    },
+    {
+      "epoch": 1.417465388711395,
+      "grad_norm": 0.48046875,
+      "learning_rate": 2.011457873166179e-05,
+      "loss": 0.8996407389640808,
+      "step": 666
+    },
+    {
+      "epoch": 1.4217252396166133,
+      "grad_norm": 0.390625,
+      "learning_rate": 2.005912744605019e-05,
+      "loss": 0.822201132774353,
+      "step": 668
+    },
+    {
+      "epoch": 1.4259850905218316,
+      "grad_norm": 0.294921875,
+      "learning_rate": 2.0003630820491066e-05,
+      "loss": 0.8432199358940125,
+      "step": 670
+    },
+    {
+      "epoch": 1.4302449414270502,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.9948090076965163e-05,
+      "loss": 0.8672274351119995,
+      "step": 672
+    },
+    {
+      "epoch": 1.4345047923322682,
+      "grad_norm": 0.25,
+      "learning_rate": 1.9892506438424666e-05,
+      "loss": 0.8486787676811218,
+      "step": 674
+    },
+    {
+      "epoch": 1.4387646432374868,
+      "grad_norm": 0.279296875,
+      "learning_rate": 1.9836881128766248e-05,
+      "loss": 0.8892148733139038,
+      "step": 676
+    },
+    {
+      "epoch": 1.443024494142705,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.9781215372804158e-05,
+      "loss": 0.8915472030639648,
+      "step": 678
+    },
+    {
+      "epoch": 1.4472843450479234,
+      "grad_norm": 0.3203125,
+      "learning_rate": 1.9725510396243226e-05,
+      "loss": 0.8767306804656982,
+      "step": 680
+    },
+    {
+      "epoch": 1.4515441959531417,
+      "grad_norm": 0.421875,
+      "learning_rate": 1.9669767425651873e-05,
+      "loss": 1.0251777172088623,
+      "step": 682
+    },
+    {
+      "epoch": 1.45580404685836,
+      "grad_norm": 0.32421875,
+      "learning_rate": 1.9613987688435132e-05,
+      "loss": 0.8821164965629578,
+      "step": 684
+    },
+    {
+      "epoch": 1.4600638977635783,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 1.955817241280757e-05,
+      "loss": 0.8836470246315002,
+      "step": 686
+    },
+    {
+      "epoch": 1.4643237486687966,
+      "grad_norm": 0.173828125,
+      "learning_rate": 1.9502322827766297e-05,
+      "loss": 0.9067674279212952,
+      "step": 688
+    },
+    {
+      "epoch": 1.4685835995740149,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 1.9446440163063875e-05,
+      "loss": 0.9052207469940186,
+      "step": 690
+    },
+    {
+      "epoch": 1.4728434504792332,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.939052564918126e-05,
+      "loss": 0.8458245396614075,
+      "step": 692
+    },
+    {
+      "epoch": 1.4771033013844517,
+      "grad_norm": 0.328125,
+      "learning_rate": 1.9334580517300668e-05,
+      "loss": 0.9541709423065186,
+      "step": 694
+    },
+    {
+      "epoch": 1.4813631522896697,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 1.9278605999278513e-05,
+      "loss": 0.9391557574272156,
+      "step": 696
+    },
+    {
+      "epoch": 1.4856230031948883,
+      "grad_norm": 0.18359375,
+      "learning_rate": 1.922260332761827e-05,
+      "loss": 0.9119634628295898,
+      "step": 698
+    },
+    {
+      "epoch": 1.4898828541001066,
+      "grad_norm": 0.3671875,
+      "learning_rate": 1.9166573735443302e-05,
+      "loss": 0.872115433216095,
+      "step": 700
+    },
+    {
+      "epoch": 1.4941427050053249,
+      "grad_norm": 2.609375,
+      "learning_rate": 1.9110518456469764e-05,
+      "loss": 0.9491547346115112,
+      "step": 702
+    },
+    {
+      "epoch": 1.4984025559105432,
+      "grad_norm": 0.1826171875,
+      "learning_rate": 1.905443872497939e-05,
+      "loss": 0.8039662837982178,
+      "step": 704
+    },
+    {
+      "epoch": 1.5026624068157615,
+      "grad_norm": 0.26953125,
+      "learning_rate": 1.8998335775792343e-05,
+      "loss": 0.8376708030700684,
+      "step": 706
+    },
+    {
+      "epoch": 1.5069222577209798,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 1.894221084424001e-05,
+      "loss": 0.8669439554214478,
+      "step": 708
+    },
+    {
+      "epoch": 1.511182108626198,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 1.888606516613781e-05,
+      "loss": 0.8526804447174072,
+      "step": 710
+    },
+    {
+      "epoch": 1.5154419595314164,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 1.8829899977757996e-05,
+      "loss": 0.838132381439209,
+      "step": 712
+    },
+    {
+      "epoch": 1.5197018104366347,
+      "grad_norm": 0.294921875,
+      "learning_rate": 1.8773716515802387e-05,
+      "loss": 0.9030261635780334,
+      "step": 714
+    },
+    {
+      "epoch": 1.5239616613418532,
+      "grad_norm": 0.298828125,
+      "learning_rate": 1.8717516017375192e-05,
+      "loss": 0.8684689998626709,
+      "step": 716
+    },
+    {
+      "epoch": 1.5282215122470713,
+      "grad_norm": 0.208984375,
+      "learning_rate": 1.866129971995575e-05,
+      "loss": 0.950151264667511,
+      "step": 718
+    },
+    {
+      "epoch": 1.5324813631522898,
+      "grad_norm": 0.4765625,
+      "learning_rate": 1.8605068861371255e-05,
+      "loss": 0.9864886403083801,
+      "step": 720
+    },
+    {
+      "epoch": 1.5367412140575079,
+      "grad_norm": 0.2421875,
+      "learning_rate": 1.8548824679769538e-05,
+      "loss": 0.9203893542289734,
+      "step": 722
+    },
+    {
+      "epoch": 1.5410010649627264,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 1.8492568413591787e-05,
+      "loss": 0.8589147329330444,
+      "step": 724
+    },
+    {
+      "epoch": 1.5452609158679447,
+      "grad_norm": 0.2265625,
+      "learning_rate": 1.8436301301545282e-05,
+      "loss": 0.7150123119354248,
+      "step": 726
+    },
+    {
+      "epoch": 1.549520766773163,
+      "grad_norm": 0.1796875,
+      "learning_rate": 1.8380024582576128e-05,
+      "loss": 0.843291163444519,
+      "step": 728
+    },
+    {
+      "epoch": 1.5537806176783813,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.8323739495841943e-05,
+      "loss": 0.8748659491539001,
+      "step": 730
+    },
+    {
+      "epoch": 1.5580404685835996,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 1.8267447280684607e-05,
+      "loss": 0.8816359043121338,
+      "step": 732
+    },
+    {
+      "epoch": 1.5623003194888179,
+      "grad_norm": 0.1611328125,
+      "learning_rate": 1.8211149176602964e-05,
+      "loss": 0.9086512923240662,
+      "step": 734
+    },
+    {
+      "epoch": 1.5665601703940362,
+      "grad_norm": 0.19140625,
+      "learning_rate": 1.8154846423225515e-05,
+      "loss": 0.9282605648040771,
+      "step": 736
+    },
+    {
+      "epoch": 1.5708200212992547,
+      "grad_norm": 0.486328125,
+      "learning_rate": 1.8098540260283158e-05,
+      "loss": 0.8508008122444153,
+      "step": 738
+    },
+    {
+      "epoch": 1.5750798722044728,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 1.8042231927581833e-05,
+      "loss": 0.7999932169914246,
+      "step": 740
+    },
+    {
+      "epoch": 1.5793397231096913,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.7985922664975274e-05,
+      "loss": 0.9391716718673706,
+      "step": 742
+    },
+    {
+      "epoch": 1.5835995740149094,
+      "grad_norm": 0.201171875,
+      "learning_rate": 1.79296137123377e-05,
+      "loss": 0.8545106649398804,
+      "step": 744
+    },
+    {
+      "epoch": 1.5878594249201279,
+      "grad_norm": 0.197265625,
+      "learning_rate": 1.7873306309536485e-05,
+      "loss": 0.8491992950439453,
+      "step": 746
+    },
+    {
+      "epoch": 1.592119275825346,
+      "grad_norm": 0.306640625,
+      "learning_rate": 1.7817001696404894e-05,
+      "loss": 0.8515585064888,
+      "step": 748
+    },
+    {
+      "epoch": 1.5963791267305645,
+      "grad_norm": 0.212890625,
+      "learning_rate": 1.7760701112714742e-05,
+      "loss": 0.8558241128921509,
+      "step": 750
+    },
+    {
+      "epoch": 1.6006389776357828,
+      "grad_norm": 0.28125,
+      "learning_rate": 1.7704405798149154e-05,
+      "loss": 0.8748922944068909,
+      "step": 752
+    },
+    {
+      "epoch": 1.604898828541001,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.764811699227521e-05,
+      "loss": 0.881086528301239,
+      "step": 754
+    },
+    {
+      "epoch": 1.6091586794462194,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 1.7591835934516677e-05,
+      "loss": 0.8601434230804443,
+      "step": 756
+    },
+    {
+      "epoch": 1.6134185303514377,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.7535563864126723e-05,
+      "loss": 0.925481915473938,
+      "step": 758
+    },
+    {
+      "epoch": 1.617678381256656,
+      "grad_norm": 0.224609375,
+      "learning_rate": 1.7479302020160627e-05,
+      "loss": 0.8856874108314514,
+      "step": 760
+    },
+    {
+      "epoch": 1.6219382321618743,
+      "grad_norm": 0.6875,
+      "learning_rate": 1.7423051641448478e-05,
+      "loss": 0.9088162779808044,
+      "step": 762
+    },
+    {
+      "epoch": 1.6261980830670928,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.7366813966567914e-05,
+      "loss": 0.7893877029418945,
+      "step": 764
+    },
+    {
+      "epoch": 1.6304579339723109,
+      "grad_norm": 0.3046875,
+      "learning_rate": 1.7310590233816868e-05,
+      "loss": 0.8651562929153442,
+      "step": 766
+    },
+    {
+      "epoch": 1.6347177848775294,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 1.7254381681186248e-05,
+      "loss": 0.8518175482749939,
+      "step": 768
+    },
+    {
+      "epoch": 1.6389776357827475,
+      "grad_norm": 0.306640625,
+      "learning_rate": 1.7198189546332738e-05,
+      "loss": 0.8798878192901611,
+      "step": 770
+    },
+    {
+      "epoch": 1.643237486687966,
+      "grad_norm": 0.248046875,
+      "learning_rate": 1.7142015066551515e-05,
+      "loss": 0.815255343914032,
+      "step": 772
+    },
+    {
+      "epoch": 1.6474973375931843,
+      "grad_norm": 0.2373046875,
+      "learning_rate": 1.7085859478748988e-05,
+      "loss": 0.936029314994812,
+      "step": 774
+    },
+    {
+      "epoch": 1.6517571884984026,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 1.7029724019415604e-05,
+      "loss": 0.9097844362258911,
+      "step": 776
+    },
+    {
+      "epoch": 1.6560170394036209,
+      "grad_norm": 0.29296875,
+      "learning_rate": 1.6973609924598605e-05,
+      "loss": 0.8360726833343506,
+      "step": 778
+    },
+    {
+      "epoch": 1.6602768903088392,
+      "grad_norm": 0.31640625,
+      "learning_rate": 1.691751842987478e-05,
+      "loss": 0.7691276669502258,
+      "step": 780
+    },
+    {
+      "epoch": 1.6645367412140575,
+      "grad_norm": 0.412109375,
+      "learning_rate": 1.6861450770323317e-05,
+      "loss": 0.9032488465309143,
+      "step": 782
+    },
+    {
+      "epoch": 1.6687965921192758,
+      "grad_norm": 0.30859375,
+      "learning_rate": 1.680540818049856e-05,
+      "loss": 0.8317678570747375,
+      "step": 784
+    },
+    {
+      "epoch": 1.6730564430244943,
+      "grad_norm": 0.455078125,
+      "learning_rate": 1.674939189440285e-05,
+      "loss": 0.8583813905715942,
+      "step": 786
+    },
+    {
+      "epoch": 1.6773162939297124,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 1.6693403145459335e-05,
+      "loss": 0.8612514138221741,
+      "step": 788
+    },
+    {
+      "epoch": 1.681576144834931,
+      "grad_norm": 0.19921875,
+      "learning_rate": 1.6637443166484836e-05,
+      "loss": 0.8975757360458374,
+      "step": 790
+    },
+    {
+      "epoch": 1.685835995740149,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.6581513189662684e-05,
+      "loss": 0.8868735432624817,
+      "step": 792
+    },
+    {
+      "epoch": 1.6900958466453675,
+      "grad_norm": 0.251953125,
+      "learning_rate": 1.652561444651558e-05,
+      "loss": 0.887550950050354,
+      "step": 794
+    },
+    {
+      "epoch": 1.6943556975505856,
+      "grad_norm": 0.25,
+      "learning_rate": 1.6469748167878502e-05,
+      "loss": 0.8832526803016663,
+      "step": 796
+    },
+    {
+      "epoch": 1.698615548455804,
+      "grad_norm": 0.203125,
+      "learning_rate": 1.64139155838716e-05,
+      "loss": 0.8911911845207214,
+      "step": 798
+    },
+    {
+      "epoch": 1.7028753993610224,
+      "grad_norm": 0.181640625,
+      "learning_rate": 1.635811792387308e-05,
+      "loss": 0.8105019927024841,
+      "step": 800
+    },
+    {
+      "epoch": 1.7071352502662407,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.630235641649217e-05,
+      "loss": 0.8116901516914368,
+      "step": 802
+    },
+    {
+      "epoch": 1.711395101171459,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 1.6246632289542054e-05,
+      "loss": 0.936326801776886,
+      "step": 804
+    },
+    {
+      "epoch": 1.7156549520766773,
+      "grad_norm": 0.251953125,
+      "learning_rate": 1.6190946770012838e-05,
+      "loss": 0.7342237234115601,
+      "step": 806
+    },
+    {
+      "epoch": 1.7199148029818956,
+      "grad_norm": 0.314453125,
+      "learning_rate": 1.613530108404451e-05,
+      "loss": 0.8804312944412231,
+      "step": 808
+    },
+    {
+      "epoch": 1.7241746538871139,
+      "grad_norm": 0.255859375,
+      "learning_rate": 1.6079696456899987e-05,
+      "loss": 0.900128960609436,
+      "step": 810
+    },
+    {
+      "epoch": 1.7284345047923324,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 1.6024134112938102e-05,
+      "loss": 0.9259054660797119,
+      "step": 812
+    },
+    {
+      "epoch": 1.7326943556975505,
+      "grad_norm": 0.353515625,
+      "learning_rate": 1.5968615275586648e-05,
+      "loss": 0.7679681777954102,
+      "step": 814
+    },
+    {
+      "epoch": 1.736954206602769,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.5913141167315455e-05,
+      "loss": 0.8207501173019409,
+      "step": 816
+    },
+    {
+      "epoch": 1.741214057507987,
+      "grad_norm": 0.69140625,
+      "learning_rate": 1.5857713009609468e-05,
+      "loss": 0.8840711116790771,
+      "step": 818
+    },
+    {
+      "epoch": 1.7454739084132056,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.5802332022941827e-05,
+      "loss": 0.87161785364151,
+      "step": 820
+    },
+    {
+      "epoch": 1.749733759318424,
+      "grad_norm": 0.1552734375,
+      "learning_rate": 1.5746999426747028e-05,
+      "loss": 0.8653435111045837,
+      "step": 822
+    },
+    {
+      "epoch": 1.7539936102236422,
+      "grad_norm": 0.39453125,
+      "learning_rate": 1.5691716439394043e-05,
+      "loss": 0.8810278177261353,
+      "step": 824
+    },
+    {
+      "epoch": 1.7582534611288605,
+      "grad_norm": 0.29296875,
+      "learning_rate": 1.563648427815953e-05,
+      "loss": 0.8902249336242676,
+      "step": 826
+    },
+    {
+      "epoch": 1.7625133120340788,
+      "grad_norm": 0.32421875,
+      "learning_rate": 1.558130415920098e-05,
+      "loss": 0.8972048163414001,
+      "step": 828
+    },
+    {
+      "epoch": 1.766773162939297,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 1.552617729752998e-05,
+      "loss": 0.8320347666740417,
+      "step": 830
+    },
+    {
+      "epoch": 1.7710330138445154,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 1.5471104906985447e-05,
+      "loss": 0.8805668354034424,
+      "step": 832
+    },
+    {
+      "epoch": 1.775292864749734,
+      "grad_norm": 0.146484375,
+      "learning_rate": 1.5416088200206873e-05,
+      "loss": 0.8669639229774475,
+      "step": 834
+    },
+    {
+      "epoch": 1.779552715654952,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 1.5361128388607685e-05,
+      "loss": 0.8641019463539124,
+      "step": 836
+    },
+    {
+      "epoch": 1.7838125665601705,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 1.5306226682348513e-05,
+      "loss": 0.8257539868354797,
+      "step": 838
+    },
+    {
+      "epoch": 1.7880724174653886,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 1.525138429031056e-05,
+      "loss": 0.8225594758987427,
+      "step": 840
+    },
+    {
+      "epoch": 1.792332268370607,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.5196602420068995e-05,
+      "loss": 0.8701678514480591,
+      "step": 842
+    },
+    {
+      "epoch": 1.7965921192758252,
+      "grad_norm": 0.28515625,
+      "learning_rate": 1.514188227786637e-05,
+      "loss": 0.8979432582855225,
+      "step": 844
+    },
+    {
+      "epoch": 1.8008519701810437,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.5087225068586032e-05,
+      "loss": 0.8577451109886169,
+      "step": 846
+    },
+    {
+      "epoch": 1.805111821086262,
+      "grad_norm": 0.375,
+      "learning_rate": 1.5032631995725602e-05,
+      "loss": 0.7677904367446899,
+      "step": 848
+    },
+    {
+      "epoch": 1.8093716719914803,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 1.4978104261370499e-05,
+      "loss": 0.8740429878234863,
+      "step": 850
+    },
+    {
+      "epoch": 1.8136315228966986,
+      "grad_norm": 0.2080078125,
+      "learning_rate": 1.4923643066167442e-05,
+      "loss": 0.8772373795509338,
+      "step": 852
+    },
+    {
+      "epoch": 1.817891373801917,
+      "grad_norm": 0.1923828125,
+      "learning_rate": 1.4869249609298016e-05,
+      "loss": 0.8475224375724792,
+      "step": 854
+    },
+    {
+      "epoch": 1.8221512247071352,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 1.4814925088452294e-05,
+      "loss": 0.8336386680603027,
+      "step": 856
+    },
+    {
+      "epoch": 1.8264110756123535,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.4760670699802433e-05,
+      "loss": 0.8594604730606079,
+      "step": 858
+    },
+    {
+      "epoch": 1.830670926517572,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.4706487637976349e-05,
+      "loss": 0.8947794437408447,
+      "step": 860
+    },
+    {
+      "epoch": 1.83493077742279,
+      "grad_norm": 0.181640625,
+      "learning_rate": 1.4652377096031413e-05,
+      "loss": 0.8090410828590393,
+      "step": 862
+    },
+    {
+      "epoch": 1.8391906283280086,
+      "grad_norm": 0.255859375,
+      "learning_rate": 1.4598340265428186e-05,
+      "loss": 0.8447999954223633,
+      "step": 864
+    },
+    {
+      "epoch": 1.8434504792332267,
+      "grad_norm": 0.1728515625,
+      "learning_rate": 1.4544378336004174e-05,
+      "loss": 0.8753990530967712,
+      "step": 866
+    },
+    {
+      "epoch": 1.8477103301384452,
+      "grad_norm": 0.291015625,
+      "learning_rate": 1.4490492495947626e-05,
+      "loss": 0.8337631225585938,
+      "step": 868
+    },
+    {
+      "epoch": 1.8519701810436635,
+      "grad_norm": 0.515625,
+      "learning_rate": 1.4436683931771386e-05,
+      "loss": 0.8855006098747253,
+      "step": 870
+    },
+    {
+      "epoch": 1.8562300319488818,
+      "grad_norm": 0.25,
+      "learning_rate": 1.4382953828286769e-05,
+      "loss": 0.8446431756019592,
+      "step": 872
+    },
+    {
+      "epoch": 1.8604898828541,
+      "grad_norm": 0.1845703125,
+      "learning_rate": 1.4329303368577442e-05,
+      "loss": 0.9195294976234436,
+      "step": 874
+    },
+    {
+      "epoch": 1.8647497337593184,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 1.4275733733973408e-05,
+      "loss": 0.8846089243888855,
+      "step": 876
+    },
+    {
+      "epoch": 1.8690095846645367,
+      "grad_norm": 0.310546875,
+      "learning_rate": 1.4222246104024985e-05,
+      "loss": 0.8711283802986145,
+      "step": 878
+    },
+    {
+      "epoch": 1.873269435569755,
+      "grad_norm": 0.2421875,
+      "learning_rate": 1.4168841656476817e-05,
+      "loss": 0.8777478337287903,
+      "step": 880
+    },
+    {
+      "epoch": 1.8775292864749735,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 1.411552156724196e-05,
+      "loss": 0.9211516976356506,
+      "step": 882
+    },
+    {
+      "epoch": 1.8817891373801916,
+      "grad_norm": 0.2060546875,
+      "learning_rate": 1.4062287010375991e-05,
+      "loss": 0.7991109490394592,
+      "step": 884
+    },
+    {
+      "epoch": 1.8860489882854101,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.4009139158051142e-05,
+      "loss": 0.7523772120475769,
+      "step": 886
+    },
+    {
+      "epoch": 1.8903088391906282,
+      "grad_norm": 0.306640625,
+      "learning_rate": 1.3956079180530488e-05,
+      "loss": 0.8029102087020874,
+      "step": 888
+    },
+    {
+      "epoch": 1.8945686900958467,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.3903108246142204e-05,
+      "loss": 0.9185020923614502,
+      "step": 890
+    },
+    {
+      "epoch": 1.898828541001065,
+      "grad_norm": 0.185546875,
+      "learning_rate": 1.3850227521253819e-05,
+      "loss": 0.8490954041481018,
+      "step": 892
+    },
+    {
+      "epoch": 1.9030883919062833,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 1.379743817024653e-05,
+      "loss": 0.9293335676193237,
+      "step": 894
+    },
+    {
+      "epoch": 1.9073482428115016,
+      "grad_norm": 0.26953125,
+      "learning_rate": 1.3744741355489573e-05,
+      "loss": 0.83982253074646,
+      "step": 896
+    },
+    {
+      "epoch": 1.91160809371672,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.3692138237314642e-05,
+      "loss": 0.8462101817131042,
+      "step": 898
+    },
+    {
+      "epoch": 1.9158679446219382,
+      "grad_norm": 0.296875,
+      "learning_rate": 1.3639629973990308e-05,
+      "loss": 0.8812525272369385,
+      "step": 900
+    },
+    {
+      "epoch": 1.9201277955271565,
+      "grad_norm": 0.2021484375,
+      "learning_rate": 1.3587217721696534e-05,
+      "loss": 0.8216854929924011,
+      "step": 902
+    },
+    {
+      "epoch": 1.924387646432375,
+      "grad_norm": 0.267578125,
+      "learning_rate": 1.3534902634499233e-05,
+      "loss": 0.8462478518486023,
+      "step": 904
+    },
+    {
+      "epoch": 1.928647497337593,
+      "grad_norm": 0.2099609375,
+      "learning_rate": 1.3482685864324816e-05,
+      "loss": 0.8769442439079285,
+      "step": 906
+    },
+    {
+      "epoch": 1.9329073482428116,
+      "grad_norm": 0.2890625,
+      "learning_rate": 1.3430568560934854e-05,
+      "loss": 0.8453910946846008,
+      "step": 908
+    },
+    {
+      "epoch": 1.9371671991480297,
+      "grad_norm": 0.236328125,
+      "learning_rate": 1.3378551871900778e-05,
+      "loss": 0.7549237012863159,
+      "step": 910
+    },
+    {
+      "epoch": 1.9414270500532482,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.332663694257857e-05,
+      "loss": 0.8484979867935181,
+      "step": 912
+    },
+    {
+      "epoch": 1.9456869009584663,
+      "grad_norm": 0.33203125,
+      "learning_rate": 1.3274824916083569e-05,
+      "loss": 0.8290879130363464,
+      "step": 914
+    },
+    {
+      "epoch": 1.9499467518636848,
+      "grad_norm": 0.314453125,
+      "learning_rate": 1.3223116933265295e-05,
+      "loss": 0.880619466304779,
+      "step": 916
+    },
+    {
+      "epoch": 1.9542066027689031,
+      "grad_norm": 0.205078125,
+      "learning_rate": 1.3171514132682338e-05,
+      "loss": 0.8705392479896545,
+      "step": 918
+    },
+    {
+      "epoch": 1.9584664536741214,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 1.3120017650577267e-05,
+      "loss": 0.849368691444397,
+      "step": 920
+    },
+    {
+      "epoch": 1.9627263045793397,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.3068628620851627e-05,
+      "loss": 0.8190315961837769,
+      "step": 922
+    },
+    {
+      "epoch": 1.966986155484558,
+      "grad_norm": 0.390625,
+      "learning_rate": 1.3017348175040983e-05,
+      "loss": 0.8338907361030579,
+      "step": 924
+    },
+    {
+      "epoch": 1.9712460063897763,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 1.2966177442289958e-05,
+      "loss": 0.783728837966919,
+      "step": 926
+    },
+    {
+      "epoch": 1.9755058572949946,
+      "grad_norm": 0.16796875,
+      "learning_rate": 1.2915117549327428e-05,
+      "loss": 0.8934606313705444,
+      "step": 928
+    },
+    {
+      "epoch": 1.9797657082002131,
+      "grad_norm": 0.458984375,
+      "learning_rate": 1.2864169620441688e-05,
+      "loss": 0.8038821220397949,
+      "step": 930
+    },
+    {
+      "epoch": 1.9840255591054312,
+      "grad_norm": 0.349609375,
+      "learning_rate": 1.2813334777455677e-05,
+      "loss": 0.9299109578132629,
+      "step": 932
+    },
+    {
+      "epoch": 1.9882854100106497,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 1.27626141397023e-05,
+      "loss": 0.7765668034553528,
+      "step": 934
+    },
+    {
+      "epoch": 1.9925452609158678,
+      "grad_norm": 0.224609375,
+      "learning_rate": 1.2712008823999787e-05,
+      "loss": 0.8893784284591675,
+      "step": 936
+    },
+    {
+      "epoch": 1.9968051118210863,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 1.2661519944627085e-05,
+      "loss": 0.8529191017150879,
+      "step": 938
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.365234375,
+      "learning_rate": 1.2611148613299316e-05,
+      "loss": 0.8112186789512634,
+      "step": 940
+    },
+    {
+      "epoch": 2.0042598509052185,
+      "grad_norm": 0.1748046875,
+      "learning_rate": 1.2560895939143335e-05,
+      "loss": 0.6377139687538147,
+      "step": 942
+    },
+    {
+      "epoch": 2.0085197018104366,
+      "grad_norm": 0.15625,
+      "learning_rate": 1.2510763028673259e-05,
+      "loss": 0.5881322026252747,
+      "step": 944
+    },
+    {
+      "epoch": 2.012779552715655,
+      "grad_norm": 0.16015625,
+      "learning_rate": 1.2460750985766133e-05,
+      "loss": 0.5497787594795227,
+      "step": 946
+    },
+    {
+      "epoch": 2.017039403620873,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.2410860911637633e-05,
+      "loss": 0.6513974070549011,
+      "step": 948
+    },
+    {
+      "epoch": 2.0212992545260917,
+      "grad_norm": 0.16796875,
+      "learning_rate": 1.2361093904817794e-05,
+      "loss": 0.6880634427070618,
+      "step": 950
+    },
+    {
+      "epoch": 2.02555910543131,
+      "grad_norm": 0.2470703125,
+      "learning_rate": 1.2311451061126825e-05,
+      "loss": 0.669802188873291,
+      "step": 952
+    },
+    {
+      "epoch": 2.0298189563365283,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 1.2261933473650986e-05,
+      "loss": 0.6532925963401794,
+      "step": 954
+    },
+    {
+      "epoch": 2.0340788072417464,
+      "grad_norm": 0.216796875,
+      "learning_rate": 1.2212542232718526e-05,
+      "loss": 0.6424761414527893,
+      "step": 956
+    },
+    {
+      "epoch": 2.038338658146965,
+      "grad_norm": 0.3046875,
+      "learning_rate": 1.2163278425875673e-05,
+      "loss": 0.599922776222229,
+      "step": 958
+    },
+    {
+      "epoch": 2.042598509052183,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.211414313786267e-05,
+      "loss": 0.5999573469161987,
+      "step": 960
+    },
+    {
+      "epoch": 2.0468583599574015,
+      "grad_norm": 0.341796875,
+      "learning_rate": 1.2065137450589902e-05,
+      "loss": 0.5664547681808472,
+      "step": 962
+    },
+    {
+      "epoch": 2.0511182108626196,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.2016262443114092e-05,
+      "loss": 0.6771121025085449,
+      "step": 964
+    },
+    {
+      "epoch": 2.055378061767838,
+      "grad_norm": 0.251953125,
+      "learning_rate": 1.19675191916145e-05,
+      "loss": 0.6011976003646851,
+      "step": 966
+    },
+    {
+      "epoch": 2.0596379126730566,
+      "grad_norm": 0.45703125,
+      "learning_rate": 1.1918908769369263e-05,
+      "loss": 0.624125599861145,
+      "step": 968
+    },
+    {
+      "epoch": 2.0638977635782747,
+      "grad_norm": 0.52734375,
+      "learning_rate": 1.187043224673176e-05,
+      "loss": 0.5838209390640259,
+      "step": 970
+    },
+    {
+      "epoch": 2.0681576144834932,
+      "grad_norm": 0.4765625,
+      "learning_rate": 1.1822090691107007e-05,
+      "loss": 0.6163349151611328,
+      "step": 972
+    },
+    {
+      "epoch": 2.0724174653887113,
+      "grad_norm": 0.2578125,
+      "learning_rate": 1.1773885166928193e-05,
+      "loss": 0.6664748787879944,
+      "step": 974
+    },
+    {
+      "epoch": 2.07667731629393,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.1725816735633235e-05,
+      "loss": 0.6090631484985352,
+      "step": 976
+    },
+    {
+      "epoch": 2.080937167199148,
+      "grad_norm": 0.2734375,
+      "learning_rate": 1.1677886455641398e-05,
+      "loss": 0.6150251030921936,
+      "step": 978
+    },
+    {
+      "epoch": 2.0851970181043664,
+      "grad_norm": 0.59375,
+      "learning_rate": 1.1630095382329988e-05,
+      "loss": 0.6834192872047424,
+      "step": 980
+    },
+    {
+      "epoch": 2.0894568690095845,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.158244456801111e-05,
+      "loss": 0.5855680108070374,
+      "step": 982
+    },
+    {
+      "epoch": 2.093716719914803,
+      "grad_norm": 0.26171875,
+      "learning_rate": 1.1534935061908528e-05,
+      "loss": 0.6290924549102783,
+      "step": 984
+    },
+    {
+      "epoch": 2.097976570820021,
+      "grad_norm": 0.27734375,
+      "learning_rate": 1.1487567910134513e-05,
+      "loss": 0.5710505247116089,
+      "step": 986
+    },
+    {
+      "epoch": 2.1022364217252396,
+      "grad_norm": 0.3203125,
+      "learning_rate": 1.1440344155666851e-05,
+      "loss": 0.6610984802246094,
+      "step": 988
+    },
+    {
+      "epoch": 2.106496272630458,
+      "grad_norm": 0.2138671875,
+      "learning_rate": 1.1393264838325865e-05,
+      "loss": 0.6294957995414734,
+      "step": 990
+    },
+    {
+      "epoch": 2.110756123535676,
+      "grad_norm": 0.46484375,
+      "learning_rate": 1.1346330994751497e-05,
+      "loss": 0.6489307880401611,
+      "step": 992
+    },
+    {
+      "epoch": 2.1150159744408947,
+      "grad_norm": 0.287109375,
+      "learning_rate": 1.1299543658380509e-05,
+      "loss": 0.5717250108718872,
+      "step": 994
+    },
+    {
+      "epoch": 2.119275825346113,
+      "grad_norm": 0.236328125,
+      "learning_rate": 1.1252903859423728e-05,
+      "loss": 0.5853033065795898,
+      "step": 996
+    },
+    {
+      "epoch": 2.1235356762513313,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 1.120641262484335e-05,
+      "loss": 0.608925461769104,
+      "step": 998
+    },
+    {
+      "epoch": 2.1277955271565494,
+      "grad_norm": 0.216796875,
+      "learning_rate": 1.1160070978330323e-05,
+      "loss": 0.6262862086296082,
+      "step": 1000
+    },
+    {
+      "epoch": 2.132055378061768,
+      "grad_norm": 0.30078125,
+      "learning_rate": 1.1113879940281813e-05,
+      "loss": 0.5531333088874817,
+      "step": 1002
+    },
+    {
+      "epoch": 2.136315228966986,
+      "grad_norm": 0.302734375,
+      "learning_rate": 1.1067840527778752e-05,
+      "loss": 0.6142609119415283,
+      "step": 1004
+    },
+    {
+      "epoch": 2.1405750798722045,
+      "grad_norm": 0.271484375,
+      "learning_rate": 1.1021953754563406e-05,
+      "loss": 0.6254585981369019,
+      "step": 1006
+    },
+    {
+      "epoch": 2.1448349307774226,
+      "grad_norm": 0.224609375,
+      "learning_rate": 1.0976220631017094e-05,
+      "loss": 0.648613691329956,
+      "step": 1008
+    },
+    {
+      "epoch": 2.149094781682641,
+      "grad_norm": 0.400390625,
+      "learning_rate": 1.0930642164137922e-05,
+      "loss": 0.4957270324230194,
+      "step": 1010
+    },
+    {
+      "epoch": 2.1533546325878596,
+      "grad_norm": 0.2353515625,
+      "learning_rate": 1.0885219357518583e-05,
+      "loss": 0.6625660061836243,
+      "step": 1012
+    },
+    {
+      "epoch": 2.1576144834930777,
+      "grad_norm": 0.21875,
+      "learning_rate": 1.0839953211324313e-05,
+      "loss": 0.6448312401771545,
+      "step": 1014
+    },
+    {
+      "epoch": 2.1618743343982962,
+      "grad_norm": 0.310546875,
+      "learning_rate": 1.0794844722270831e-05,
+      "loss": 0.6265139579772949,
+      "step": 1016
+    },
+    {
+      "epoch": 2.1661341853035143,
+      "grad_norm": 0.3515625,
+      "learning_rate": 1.0749894883602406e-05,
+      "loss": 0.58893221616745,
+      "step": 1018
+    },
+    {
+      "epoch": 2.170394036208733,
+      "grad_norm": 0.5234375,
+      "learning_rate": 1.0705104685069973e-05,
+      "loss": 0.524358332157135,
+      "step": 1020
+    },
+    {
+      "epoch": 2.174653887113951,
+      "grad_norm": 0.392578125,
+      "learning_rate": 1.0660475112909354e-05,
+      "loss": 0.6041074395179749,
+      "step": 1022
+    },
+    {
+      "epoch": 2.1789137380191694,
+      "grad_norm": 0.25,
+      "learning_rate": 1.0616007149819543e-05,
+      "loss": 0.6296215653419495,
+      "step": 1024
+    },
+    {
+      "epoch": 2.1831735889243875,
+      "grad_norm": 0.37890625,
+      "learning_rate": 1.057170177494105e-05,
+      "loss": 0.6504489779472351,
+      "step": 1026
+    },
+    {
+      "epoch": 2.187433439829606,
+      "grad_norm": 0.361328125,
+      "learning_rate": 1.052755996383437e-05,
+      "loss": 0.6803485155105591,
+      "step": 1028
+    },
+    {
+      "epoch": 2.191693290734824,
+      "grad_norm": 0.2333984375,
+      "learning_rate": 1.0483582688458472e-05,
+      "loss": 0.6579641699790955,
+      "step": 1030
+    },
+    {
+      "epoch": 2.1959531416400426,
+      "grad_norm": 0.5234375,
+      "learning_rate": 1.0439770917149414e-05,
+      "loss": 0.6605786085128784,
+      "step": 1032
+    },
+    {
+      "epoch": 2.2002129925452607,
+      "grad_norm": 0.189453125,
+      "learning_rate": 1.0396125614599018e-05,
+      "loss": 0.6570585370063782,
+      "step": 1034
+    },
+    {
+      "epoch": 2.2044728434504792,
+      "grad_norm": 0.337890625,
+      "learning_rate": 1.0352647741833637e-05,
+      "loss": 0.6363896131515503,
+      "step": 1036
+    },
+    {
+      "epoch": 2.2087326943556977,
+      "grad_norm": 0.296875,
+      "learning_rate": 1.0309338256192982e-05,
+      "loss": 0.6393426656723022,
+      "step": 1038
+    },
+    {
+      "epoch": 2.212992545260916,
+      "grad_norm": 0.345703125,
+      "learning_rate": 1.0266198111309041e-05,
+      "loss": 0.7091052532196045,
+      "step": 1040
+    },
+    {
+      "epoch": 2.2172523961661343,
+      "grad_norm": 0.75,
+      "learning_rate": 1.0223228257085083e-05,
+      "loss": 0.6515456438064575,
+      "step": 1042
+    },
+    {
+      "epoch": 2.2215122470713524,
+      "grad_norm": 0.244140625,
+      "learning_rate": 1.0180429639674761e-05,
+      "loss": 0.6235453486442566,
+      "step": 1044
+    },
+    {
+      "epoch": 2.225772097976571,
+      "grad_norm": 0.263671875,
+      "learning_rate": 1.0137803201461248e-05,
+      "loss": 0.5850796699523926,
+      "step": 1046
+    },
+    {
+      "epoch": 2.230031948881789,
+      "grad_norm": 0.443359375,
+      "learning_rate": 1.0095349881036508e-05,
+      "loss": 0.5203170776367188,
+      "step": 1048
+    },
+    {
+      "epoch": 2.2342917997870075,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.0053070613180625e-05,
+      "loss": 0.6159985065460205,
+      "step": 1050
+    },
+    {
+      "epoch": 2.2385516506922256,
+      "grad_norm": 0.203125,
+      "learning_rate": 1.0010966328841206e-05,
+      "loss": 0.6239602565765381,
+      "step": 1052
+    },
+    {
+      "epoch": 2.242811501597444,
+      "grad_norm": 0.265625,
+      "learning_rate": 9.969037955112908e-06,
+      "loss": 0.6027981042861938,
+      "step": 1054
+    },
+    {
+      "epoch": 2.247071352502662,
+      "grad_norm": 0.30078125,
+      "learning_rate": 9.927286415217005e-06,
+      "loss": 0.591469407081604,
+      "step": 1056
+    },
+    {
+      "epoch": 2.2513312034078807,
+      "grad_norm": 0.1982421875,
+      "learning_rate": 9.88571262848107e-06,
+      "loss": 0.5683766007423401,
+      "step": 1058
+    },
+    {
+      "epoch": 2.255591054313099,
+      "grad_norm": 0.35546875,
+      "learning_rate": 9.844317510318719e-06,
+      "loss": 0.6158217191696167,
+      "step": 1060
+    },
+    {
+      "epoch": 2.2598509052183173,
+      "grad_norm": 0.60546875,
+      "learning_rate": 9.803101972209462e-06,
+      "loss": 0.5769312381744385,
+      "step": 1062
+    },
+    {
+      "epoch": 2.264110756123536,
+      "grad_norm": 0.24609375,
+      "learning_rate": 9.762066921678647e-06,
+      "loss": 0.5810741186141968,
+      "step": 1064
+    },
+    {
+      "epoch": 2.268370607028754,
+      "grad_norm": 0.28125,
+      "learning_rate": 9.721213262277447e-06,
+      "loss": 0.5853366255760193,
+      "step": 1066
+    },
+    {
+      "epoch": 2.2726304579339724,
+      "grad_norm": 0.2314453125,
+      "learning_rate": 9.680541893563e-06,
+      "loss": 0.5754764676094055,
+      "step": 1068
+    },
+    {
+      "epoch": 2.2768903088391905,
+      "grad_norm": 0.376953125,
+      "learning_rate": 9.640053711078571e-06,
+      "loss": 0.6414265632629395,
+      "step": 1070
+    },
+    {
+      "epoch": 2.281150159744409,
+      "grad_norm": 0.703125,
+      "learning_rate": 9.599749606333844e-06,
+      "loss": 0.5730122327804565,
+      "step": 1072
+    },
+    {
+      "epoch": 2.285410010649627,
+      "grad_norm": 0.224609375,
+      "learning_rate": 9.559630466785301e-06,
+      "loss": 0.6548243761062622,
+      "step": 1074
+    },
+    {
+      "epoch": 2.2896698615548456,
+      "grad_norm": 0.359375,
+      "learning_rate": 9.519697175816675e-06,
+      "loss": 0.6757615804672241,
+      "step": 1076
+    },
+    {
+      "epoch": 2.2939297124600637,
+      "grad_norm": 0.271484375,
+      "learning_rate": 9.4799506127195e-06,
+      "loss": 0.6540831923484802,
+      "step": 1078
+    },
+    {
+      "epoch": 2.2981895633652822,
+      "grad_norm": 0.5390625,
+      "learning_rate": 9.44039165267372e-06,
+      "loss": 0.5985897779464722,
+      "step": 1080
+    },
+    {
+      "epoch": 2.3024494142705008,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.40102116672848e-06,
+      "loss": 0.6373129487037659,
+      "step": 1082
+    },
+    {
+      "epoch": 2.306709265175719,
+      "grad_norm": 0.32421875,
+      "learning_rate": 9.361840021782899e-06,
+      "loss": 0.5798696279525757,
+      "step": 1084
+    },
+    {
+      "epoch": 2.3109691160809374,
+      "grad_norm": 0.53125,
+      "learning_rate": 9.322849080566986e-06,
+      "loss": 0.6472339034080505,
+      "step": 1086
+    },
+    {
+      "epoch": 2.3152289669861554,
+      "grad_norm": 0.2236328125,
+      "learning_rate": 9.284049201622668e-06,
+      "loss": 0.5931280851364136,
+      "step": 1088
+    },
+    {
+      "epoch": 2.319488817891374,
+      "grad_norm": 0.26171875,
+      "learning_rate": 9.245441239284858e-06,
+      "loss": 0.6150895953178406,
+      "step": 1090
+    },
+    {
+      "epoch": 2.323748668796592,
+      "grad_norm": 0.37890625,
+      "learning_rate": 9.207026043662654e-06,
+      "loss": 0.5743486285209656,
+      "step": 1092
+    },
+    {
+      "epoch": 2.3280085197018106,
+      "grad_norm": 0.302734375,
+      "learning_rate": 9.168804460620634e-06,
+      "loss": 0.6586934328079224,
+      "step": 1094
+    },
+    {
+      "epoch": 2.3322683706070286,
+      "grad_norm": 0.333984375,
+      "learning_rate": 9.130777331760208e-06,
+      "loss": 0.581457793712616,
+      "step": 1096
+    },
+    {
+      "epoch": 2.336528221512247,
+      "grad_norm": 0.236328125,
+      "learning_rate": 9.092945494401107e-06,
+      "loss": 0.602104663848877,
+      "step": 1098
+    },
+    {
+      "epoch": 2.3407880724174652,
+      "grad_norm": 0.318359375,
+      "learning_rate": 9.055309781562922e-06,
+      "loss": 0.5987313985824585,
+      "step": 1100
+    },
+    {
+      "epoch": 2.3450479233226837,
+      "grad_norm": 0.212890625,
+      "learning_rate": 9.017871021946787e-06,
+      "loss": 0.5123194456100464,
+      "step": 1102
+    },
+    {
+      "epoch": 2.349307774227902,
+      "grad_norm": 0.494140625,
+      "learning_rate": 8.980630039917124e-06,
+      "loss": 0.5810441374778748,
+      "step": 1104
+    },
+    {
+      "epoch": 2.3535676251331203,
+      "grad_norm": 0.5390625,
+      "learning_rate": 8.943587655483478e-06,
+      "loss": 0.5871768593788147,
+      "step": 1106
+    },
+    {
+      "epoch": 2.357827476038339,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 8.906744684282483e-06,
+      "loss": 0.6104775667190552,
+      "step": 1108
+    },
+    {
+      "epoch": 2.362087326943557,
+      "grad_norm": 0.296875,
+      "learning_rate": 8.870101937559877e-06,
+      "loss": 0.6351394653320312,
+      "step": 1110
+    },
+    {
+      "epoch": 2.3663471778487755,
+      "grad_norm": 0.52734375,
+      "learning_rate": 8.833660222152663e-06,
+      "loss": 0.6355900168418884,
+      "step": 1112
+    },
+    {
+      "epoch": 2.3706070287539935,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 8.797420340471334e-06,
+      "loss": 0.5833765268325806,
+      "step": 1114
+    },
+    {
+      "epoch": 2.374866879659212,
+      "grad_norm": 0.2216796875,
+      "learning_rate": 8.761383090482205e-06,
+      "loss": 0.6019313931465149,
+      "step": 1116
+    },
+    {
+      "epoch": 2.37912673056443,
+      "grad_norm": 0.2265625,
+      "learning_rate": 8.725549265689833e-06,
+      "loss": 0.5999468564987183,
+      "step": 1118
+    },
+    {
+      "epoch": 2.3833865814696487,
+      "grad_norm": 0.484375,
+      "learning_rate": 8.689919655119559e-06,
+      "loss": 0.6521666646003723,
+      "step": 1120
+    },
+    {
+      "epoch": 2.3876464323748667,
+      "grad_norm": 0.224609375,
+      "learning_rate": 8.654495043300129e-06,
+      "loss": 0.612395703792572,
+      "step": 1122
+    },
+    {
+      "epoch": 2.3919062832800853,
+      "grad_norm": 0.466796875,
+      "learning_rate": 8.619276210246427e-06,
+      "loss": 0.5964239239692688,
+      "step": 1124
+    },
+    {
+      "epoch": 2.3961661341853033,
+      "grad_norm": 0.26171875,
+      "learning_rate": 8.584263931442275e-06,
+      "loss": 0.6384221911430359,
+      "step": 1126
+    },
+    {
+      "epoch": 2.400425985090522,
+      "grad_norm": 0.28125,
+      "learning_rate": 8.549458977823395e-06,
+      "loss": 0.6933798789978027,
+      "step": 1128
+    },
+    {
+      "epoch": 2.40468583599574,
+      "grad_norm": 0.1962890625,
+      "learning_rate": 8.514862115760396e-06,
+      "loss": 0.5889874696731567,
+      "step": 1130
+    },
+    {
+      "epoch": 2.4089456869009584,
+      "grad_norm": 0.30859375,
+      "learning_rate": 8.480474107041925e-06,
+      "loss": 0.6254542469978333,
+      "step": 1132
+    },
+    {
+      "epoch": 2.413205537806177,
+      "grad_norm": 0.314453125,
+      "learning_rate": 8.446295708857888e-06,
+      "loss": 0.6616327166557312,
+      "step": 1134
+    },
+    {
+      "epoch": 2.417465388711395,
+      "grad_norm": 0.375,
+      "learning_rate": 8.412327673782774e-06,
+      "loss": 0.6202198266983032,
+      "step": 1136
+    },
+    {
+      "epoch": 2.4217252396166136,
+      "grad_norm": 0.267578125,
+      "learning_rate": 8.378570749759076e-06,
+      "loss": 0.6176246404647827,
+      "step": 1138
+    },
+    {
+      "epoch": 2.4259850905218316,
+      "grad_norm": 0.369140625,
+      "learning_rate": 8.345025680080836e-06,
+      "loss": 0.5884604454040527,
+      "step": 1140
+    },
+    {
+      "epoch": 2.43024494142705,
+      "grad_norm": 0.4453125,
+      "learning_rate": 8.311693203377277e-06,
+      "loss": 0.5704495906829834,
+      "step": 1142
+    },
+    {
+      "epoch": 2.4345047923322682,
+      "grad_norm": 0.357421875,
+      "learning_rate": 8.278574053596534e-06,
+      "loss": 0.5104537606239319,
+      "step": 1144
+    },
+    {
+      "epoch": 2.4387646432374868,
+      "grad_norm": 0.375,
+      "learning_rate": 8.245668959989489e-06,
+      "loss": 0.6920484900474548,
+      "step": 1146
+    },
+    {
+      "epoch": 2.443024494142705,
+      "grad_norm": 0.3359375,
+      "learning_rate": 8.212978647093724e-06,
+      "loss": 0.605790376663208,
+      "step": 1148
+    },
+    {
+      "epoch": 2.4472843450479234,
+      "grad_norm": 0.431640625,
+      "learning_rate": 8.180503834717563e-06,
+      "loss": 0.6005589962005615,
+      "step": 1150
+    },
+    {
+      "epoch": 2.451544195953142,
+      "grad_norm": 0.30078125,
+      "learning_rate": 8.148245237924212e-06,
+      "loss": 0.6908122301101685,
+      "step": 1152
+    },
+    {
+      "epoch": 2.45580404685836,
+      "grad_norm": 0.392578125,
+      "learning_rate": 8.116203567016035e-06,
+      "loss": 0.5939027667045593,
+      "step": 1154
+    },
+    {
+      "epoch": 2.460063897763578,
+      "grad_norm": 0.357421875,
+      "learning_rate": 8.084379527518908e-06,
+      "loss": 0.6245042681694031,
+      "step": 1156
+    },
+    {
+      "epoch": 2.4643237486687966,
+      "grad_norm": 0.25390625,
+      "learning_rate": 8.05277382016666e-06,
+      "loss": 0.5638337731361389,
+      "step": 1158
+    },
+    {
+      "epoch": 2.468583599574015,
+      "grad_norm": 0.2890625,
+      "learning_rate": 8.021387140885672e-06,
+      "loss": 0.665945291519165,
+      "step": 1160
+    },
+    {
+      "epoch": 2.472843450479233,
+      "grad_norm": 0.2255859375,
+      "learning_rate": 7.99022018077955e-06,
+      "loss": 0.5603002309799194,
+      "step": 1162
+    },
+    {
+      "epoch": 2.4771033013844517,
+      "grad_norm": 0.287109375,
+      "learning_rate": 7.959273626113896e-06,
+      "loss": 0.5992410182952881,
+      "step": 1164
+    },
+    {
+      "epoch": 2.4813631522896697,
+      "grad_norm": 0.216796875,
+      "learning_rate": 7.9285481583012e-06,
+      "loss": 0.6628497242927551,
+      "step": 1166
+    },
+    {
+      "epoch": 2.4856230031948883,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.898044453885837e-06,
+      "loss": 0.5260273218154907,
+      "step": 1168
+    },
+    {
+      "epoch": 2.4898828541001063,
+      "grad_norm": 0.265625,
+      "learning_rate": 7.867763184529182e-06,
+      "loss": 0.6244964599609375,
+      "step": 1170
+    },
+    {
+      "epoch": 2.494142705005325,
+      "grad_norm": 0.369140625,
+      "learning_rate": 7.837705016994796e-06,
+      "loss": 0.6657370328903198,
+      "step": 1172
+    },
+    {
+      "epoch": 2.498402555910543,
+      "grad_norm": 0.333984375,
+      "learning_rate": 7.80787061313377e-06,
+      "loss": 0.6410002708435059,
+      "step": 1174
+    },
+    {
+      "epoch": 2.5026624068157615,
+      "grad_norm": 0.1953125,
+      "learning_rate": 7.77826062987014e-06,
+      "loss": 0.5408449769020081,
+      "step": 1176
+    },
+    {
+      "epoch": 2.50692225772098,
+      "grad_norm": 0.35546875,
+      "learning_rate": 7.748875719186413e-06,
+      "loss": 0.5735031962394714,
+      "step": 1178
+    },
+    {
+      "epoch": 2.511182108626198,
+      "grad_norm": 0.322265625,
+      "learning_rate": 7.71971652810923e-06,
+      "loss": 0.6153873801231384,
+      "step": 1180
+    },
+    {
+      "epoch": 2.515441959531416,
+      "grad_norm": 0.33984375,
+      "learning_rate": 7.690783698695106e-06,
+      "loss": 0.5873544216156006,
+      "step": 1182
+    },
+    {
+      "epoch": 2.5197018104366347,
+      "grad_norm": 0.259765625,
+      "learning_rate": 7.662077868016297e-06,
+      "loss": 0.6717422604560852,
+      "step": 1184
+    },
+    {
+      "epoch": 2.523961661341853,
+      "grad_norm": 0.94921875,
+      "learning_rate": 7.633599668146775e-06,
+      "loss": 0.6083505153656006,
+      "step": 1186
+    },
+    {
+      "epoch": 2.5282215122470713,
+      "grad_norm": 0.30078125,
+      "learning_rate": 7.605349726148296e-06,
+      "loss": 0.6134154200553894,
+      "step": 1188
+    },
+    {
+      "epoch": 2.5324813631522898,
+      "grad_norm": 0.275390625,
+      "learning_rate": 7.577328664056617e-06,
+      "loss": 0.589963972568512,
+      "step": 1190
+    },
+    {
+      "epoch": 2.536741214057508,
+      "grad_norm": 0.51171875,
+      "learning_rate": 7.549537098867776e-06,
+      "loss": 0.5288025140762329,
+      "step": 1192
+    },
+    {
+      "epoch": 2.5410010649627264,
+      "grad_norm": 0.2451171875,
+      "learning_rate": 7.521975642524525e-06,
+      "loss": 0.616111159324646,
+      "step": 1194
+    },
+    {
+      "epoch": 2.545260915867945,
+      "grad_norm": 0.3046875,
+      "learning_rate": 7.494644901902843e-06,
+      "loss": 0.6015118360519409,
+      "step": 1196
+    },
+    {
+      "epoch": 2.549520766773163,
+      "grad_norm": 0.30078125,
+      "learning_rate": 7.467545478798574e-06,
+      "loss": 0.5770639777183533,
+      "step": 1198
+    },
+    {
+      "epoch": 2.553780617678381,
+      "grad_norm": 0.365234375,
+      "learning_rate": 7.440677969914182e-06,
+      "loss": 0.6590741872787476,
+      "step": 1200
+    },
+    {
+      "epoch": 2.5580404685835996,
+      "grad_norm": 0.2392578125,
+      "learning_rate": 7.4140429668456115e-06,
+      "loss": 0.47983720898628235,
+      "step": 1202
+    },
+    {
+      "epoch": 2.562300319488818,
+      "grad_norm": 0.31640625,
+      "learning_rate": 7.38764105606926e-06,
+      "loss": 0.549656093120575,
+      "step": 1204
+    },
+    {
+      "epoch": 2.566560170394036,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.361472818929058e-06,
+      "loss": 0.5793447494506836,
+      "step": 1206
+    },
+    {
+      "epoch": 2.5708200212992547,
+      "grad_norm": 0.251953125,
+      "learning_rate": 7.335538831623676e-06,
+      "loss": 0.637956976890564,
+      "step": 1208
+    },
+    {
+      "epoch": 2.5750798722044728,
+      "grad_norm": 0.5625,
+      "learning_rate": 7.309839665193839e-06,
+      "loss": 0.5784144401550293,
+      "step": 1210
+    },
+    {
+      "epoch": 2.5793397231096913,
+      "grad_norm": 0.2275390625,
+      "learning_rate": 7.284375885509741e-06,
+      "loss": 0.6299670338630676,
+      "step": 1212
+    },
+    {
+      "epoch": 2.5835995740149094,
+      "grad_norm": 0.322265625,
+      "learning_rate": 7.259148053258603e-06,
+      "loss": 0.674586296081543,
+      "step": 1214
+    },
+    {
+      "epoch": 2.587859424920128,
+      "grad_norm": 0.3671875,
+      "learning_rate": 7.234156723932312e-06,
+      "loss": 0.6188330054283142,
+      "step": 1216
+    },
+    {
+      "epoch": 2.592119275825346,
+      "grad_norm": 0.271484375,
+      "learning_rate": 7.20940244781519e-06,
+      "loss": 0.6208375096321106,
+      "step": 1218
+    },
+    {
+      "epoch": 2.5963791267305645,
+      "grad_norm": 0.388671875,
+      "learning_rate": 7.184885769971888e-06,
+      "loss": 0.6017476916313171,
+      "step": 1220
+    },
+    {
+      "epoch": 2.600638977635783,
+      "grad_norm": 0.333984375,
+      "learning_rate": 7.160607230235378e-06,
+      "loss": 0.6354559659957886,
+      "step": 1222
+    },
+    {
+      "epoch": 2.604898828541001,
+      "grad_norm": 0.31640625,
+      "learning_rate": 7.136567363195069e-06,
+      "loss": 0.6745753884315491,
+      "step": 1224
+    },
+    {
+      "epoch": 2.609158679446219,
+      "grad_norm": 0.427734375,
+      "learning_rate": 7.112766698185027e-06,
+      "loss": 0.5988171100616455,
+      "step": 1226
+    },
+    {
+      "epoch": 2.6134185303514377,
+      "grad_norm": 0.25390625,
+      "learning_rate": 7.089205759272327e-06,
+      "loss": 0.6004793643951416,
+      "step": 1228
+    },
+    {
+      "epoch": 2.617678381256656,
+      "grad_norm": 0.31640625,
+      "learning_rate": 7.06588506524552e-06,
+      "loss": 0.5850980877876282,
+      "step": 1230
+    },
+    {
+      "epoch": 2.6219382321618743,
+      "grad_norm": 0.33203125,
+      "learning_rate": 7.042805129603193e-06,
+      "loss": 0.5615159869194031,
+      "step": 1232
+    },
+    {
+      "epoch": 2.626198083067093,
+      "grad_norm": 0.333984375,
+      "learning_rate": 7.019966460542681e-06,
+      "loss": 0.6120025515556335,
+      "step": 1234
+    },
+    {
+      "epoch": 2.630457933972311,
+      "grad_norm": 0.35546875,
+      "learning_rate": 6.997369560948859e-06,
+      "loss": 0.6796953082084656,
+      "step": 1236
+    },
+    {
+      "epoch": 2.6347177848775294,
+      "grad_norm": 0.306640625,
+      "learning_rate": 6.975014928383083e-06,
+      "loss": 0.5794081091880798,
+      "step": 1238
+    },
+    {
+      "epoch": 2.6389776357827475,
+      "grad_norm": 0.28515625,
+      "learning_rate": 6.952903055072226e-06,
+      "loss": 0.5920906066894531,
+      "step": 1240
+    },
+    {
+      "epoch": 2.643237486687966,
+      "grad_norm": 0.2412109375,
+      "learning_rate": 6.9310344278978505e-06,
+      "loss": 0.5745714902877808,
+      "step": 1242
+    },
+    {
+      "epoch": 2.647497337593184,
+      "grad_norm": 0.2578125,
+      "learning_rate": 6.909409528385466e-06,
+      "loss": 0.5876143574714661,
+      "step": 1244
+    },
+    {
+      "epoch": 2.6517571884984026,
+      "grad_norm": 0.2734375,
+      "learning_rate": 6.888028832693953e-06,
+      "loss": 0.586786150932312,
+      "step": 1246
+    },
+    {
+      "epoch": 2.656017039403621,
+      "grad_norm": 0.380859375,
+      "learning_rate": 6.86689281160506e-06,
+      "loss": 0.5594542622566223,
+      "step": 1248
+    },
+    {
+      "epoch": 2.660276890308839,
+      "grad_norm": 0.23046875,
+      "learning_rate": 6.846001930513041e-06,
+      "loss": 0.6434107422828674,
+      "step": 1250
+    },
+    {
+      "epoch": 2.6645367412140573,
+      "grad_norm": 0.361328125,
+      "learning_rate": 6.825356649414415e-06,
+      "loss": 0.6385661959648132,
+      "step": 1252
+    },
+    {
+      "epoch": 2.668796592119276,
+      "grad_norm": 0.291015625,
+      "learning_rate": 6.80495742289783e-06,
+      "loss": 0.6039466261863708,
+      "step": 1254
+    },
+    {
+      "epoch": 2.6730564430244943,
+      "grad_norm": 0.298828125,
+      "learning_rate": 6.784804700134056e-06,
+      "loss": 0.6025973558425903,
+      "step": 1256
+    },
+    {
+      "epoch": 2.6773162939297124,
+      "grad_norm": 0.296875,
+      "learning_rate": 6.764898924866091e-06,
+      "loss": 0.6119323372840881,
+      "step": 1258
+    },
+    {
+      "epoch": 2.681576144834931,
+      "grad_norm": 0.32421875,
+      "learning_rate": 6.7452405353993985e-06,
+      "loss": 0.617369532585144,
+      "step": 1260
+    },
+    {
+      "epoch": 2.685835995740149,
+      "grad_norm": 0.330078125,
+      "learning_rate": 6.72582996459225e-06,
+      "loss": 0.6640692949295044,
+      "step": 1262
+    },
+    {
+      "epoch": 2.6900958466453675,
+      "grad_norm": 0.3359375,
+      "learning_rate": 6.706667639846196e-06,
+      "loss": 0.6609706282615662,
+      "step": 1264
+    },
+    {
+      "epoch": 2.6943556975505856,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 6.687753983096654e-06,
+      "loss": 0.53211909532547,
+      "step": 1266
+    },
+    {
+      "epoch": 2.698615548455804,
+      "grad_norm": 0.455078125,
+      "learning_rate": 6.669089410803617e-06,
+      "loss": 0.667971134185791,
+      "step": 1268
+    },
+    {
+      "epoch": 2.702875399361022,
+      "grad_norm": 0.265625,
+      "learning_rate": 6.650674333942487e-06,
+      "loss": 0.5798393487930298,
+      "step": 1270
+    },
+    {
+      "epoch": 2.7071352502662407,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.632509157995023e-06,
+      "loss": 0.6258153915405273,
+      "step": 1272
+    },
+    {
+      "epoch": 2.711395101171459,
+      "grad_norm": 0.392578125,
+      "learning_rate": 6.614594282940414e-06,
+      "loss": 0.624832272529602,
+      "step": 1274
+    },
+    {
+      "epoch": 2.7156549520766773,
+      "grad_norm": 0.349609375,
+      "learning_rate": 6.596930103246468e-06,
+      "loss": 0.5772223472595215,
+      "step": 1276
+    },
+    {
+      "epoch": 2.7199148029818954,
+      "grad_norm": 0.61328125,
+      "learning_rate": 6.579517007860933e-06,
+      "loss": 0.5936267971992493,
+      "step": 1278
+    },
+    {
+      "epoch": 2.724174653887114,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 6.562355380202927e-06,
+      "loss": 0.668041467666626,
+      "step": 1280
+    },
+    {
+      "epoch": 2.7284345047923324,
+      "grad_norm": 0.25,
+      "learning_rate": 6.5454455981545e-06,
+      "loss": 0.5487772226333618,
+      "step": 1282
+    },
+    {
+      "epoch": 2.7326943556975505,
+      "grad_norm": 0.205078125,
+      "learning_rate": 6.528788034052311e-06,
+      "loss": 0.6349499225616455,
+      "step": 1284
+    },
+    {
+      "epoch": 2.736954206602769,
+      "grad_norm": 0.23828125,
+      "learning_rate": 6.512383054679422e-06,
+      "loss": 0.5938593149185181,
+      "step": 1286
+    },
+    {
+      "epoch": 2.741214057507987,
+      "grad_norm": 0.263671875,
+      "learning_rate": 6.496231021257242e-06,
+      "loss": 0.6245843172073364,
+      "step": 1288
+    },
+    {
+      "epoch": 2.7454739084132056,
+      "grad_norm": 0.357421875,
+      "learning_rate": 6.480332289437552e-06,
+      "loss": 0.5823163390159607,
+      "step": 1290
+    },
+    {
+      "epoch": 2.749733759318424,
+      "grad_norm": 0.3359375,
+      "learning_rate": 6.464687209294682e-06,
+      "loss": 0.5846402049064636,
+      "step": 1292
+    },
+    {
+      "epoch": 2.753993610223642,
+      "grad_norm": 0.412109375,
+      "learning_rate": 6.44929612531781e-06,
+      "loss": 0.6277037262916565,
+      "step": 1294
+    },
+    {
+      "epoch": 2.7582534611288603,
+      "grad_norm": 0.20703125,
+      "learning_rate": 6.434159376403363e-06,
+      "loss": 0.6208704113960266,
+      "step": 1296
+    },
+    {
+      "epoch": 2.762513312034079,
+      "grad_norm": 0.32421875,
+      "learning_rate": 6.419277295847563e-06,
+      "loss": 0.5632691979408264,
+      "step": 1298
+    },
+    {
+      "epoch": 2.7667731629392973,
+      "grad_norm": 0.296875,
+      "learning_rate": 6.404650211339093e-06,
+      "loss": 0.6156328320503235,
+      "step": 1300
+    },
+    {
+      "epoch": 2.7710330138445154,
+      "grad_norm": 0.259765625,
+      "learning_rate": 6.390278444951868e-06,
+      "loss": 0.6689990758895874,
+      "step": 1302
+    },
+    {
+      "epoch": 2.775292864749734,
+      "grad_norm": 0.25,
+      "learning_rate": 6.376162313137955e-06,
+      "loss": 0.6374217867851257,
+      "step": 1304
+    },
+    {
+      "epoch": 2.779552715654952,
+      "grad_norm": 0.439453125,
+      "learning_rate": 6.3623021267205975e-06,
+      "loss": 0.6087695360183716,
+      "step": 1306
+    },
+    {
+      "epoch": 2.7838125665601705,
+      "grad_norm": 0.2734375,
+      "learning_rate": 6.348698190887377e-06,
+      "loss": 0.5766043066978455,
+      "step": 1308
+    },
+    {
+      "epoch": 2.7880724174653886,
+      "grad_norm": 0.359375,
+      "learning_rate": 6.3353508051834924e-06,
+      "loss": 0.6857935786247253,
+      "step": 1310
+    },
+    {
+      "epoch": 2.792332268370607,
+      "grad_norm": 0.259765625,
+      "learning_rate": 6.322260263505159e-06,
+      "loss": 0.6080771684646606,
+      "step": 1312
+    },
+    {
+      "epoch": 2.796592119275825,
+      "grad_norm": 0.232421875,
+      "learning_rate": 6.309426854093147e-06,
+      "loss": 0.5428948402404785,
+      "step": 1314
+    },
+    {
+      "epoch": 2.8008519701810437,
+      "grad_norm": 0.609375,
+      "learning_rate": 6.2968508595264195e-06,
+      "loss": 0.6500948667526245,
+      "step": 1316
+    },
+    {
+      "epoch": 2.8051118210862622,
+      "grad_norm": 0.306640625,
+      "learning_rate": 6.284532556715927e-06,
+      "loss": 0.6038864850997925,
+      "step": 1318
+    },
+    {
+      "epoch": 2.8093716719914803,
+      "grad_norm": 0.404296875,
+      "learning_rate": 6.272472216898501e-06,
+      "loss": 0.6369448304176331,
+      "step": 1320
+    },
+    {
+      "epoch": 2.8136315228966984,
+      "grad_norm": 0.32421875,
+      "learning_rate": 6.260670105630885e-06,
+      "loss": 0.6288717985153198,
+      "step": 1322
+    },
+    {
+      "epoch": 2.817891373801917,
+      "grad_norm": 0.3984375,
+      "learning_rate": 6.2491264827838775e-06,
+      "loss": 0.6535931825637817,
+      "step": 1324
+    },
+    {
+      "epoch": 2.8221512247071354,
+      "grad_norm": 0.28515625,
+      "learning_rate": 6.237841602536627e-06,
+      "loss": 0.6414341330528259,
+      "step": 1326
+    },
+    {
+      "epoch": 2.8264110756123535,
+      "grad_norm": 0.2578125,
+      "learning_rate": 6.226815713371023e-06,
+      "loss": 0.5740489959716797,
+      "step": 1328
+    },
+    {
+      "epoch": 2.830670926517572,
+      "grad_norm": 0.2890625,
+      "learning_rate": 6.216049058066229e-06,
+      "loss": 0.5453130602836609,
+      "step": 1330
+    },
+    {
+      "epoch": 2.83493077742279,
+      "grad_norm": 0.228515625,
+      "learning_rate": 6.205541873693331e-06,
+      "loss": 0.531428873538971,
+      "step": 1332
+    },
+    {
+      "epoch": 2.8391906283280086,
+      "grad_norm": 0.3046875,
+      "learning_rate": 6.195294391610128e-06,
+      "loss": 0.6185562014579773,
+      "step": 1334
+    },
+    {
+      "epoch": 2.8434504792332267,
+      "grad_norm": 0.3671875,
+      "learning_rate": 6.185306837456027e-06,
+      "loss": 0.6069992184638977,
+      "step": 1336
+    },
+    {
+      "epoch": 2.847710330138445,
+      "grad_norm": 0.26953125,
+      "learning_rate": 6.1755794311470824e-06,
+      "loss": 0.5699736475944519,
+      "step": 1338
+    },
+    {
+      "epoch": 2.8519701810436633,
+      "grad_norm": 0.875,
+      "learning_rate": 6.166112386871149e-06,
+      "loss": 0.5937331318855286,
+      "step": 1340
+    },
+    {
+      "epoch": 2.856230031948882,
+      "grad_norm": 0.30078125,
+      "learning_rate": 6.15690591308317e-06,
+      "loss": 0.5383535623550415,
+      "step": 1342
+    },
+    {
+      "epoch": 2.8604898828541003,
+      "grad_norm": 0.2119140625,
+      "learning_rate": 6.14796021250058e-06,
+      "loss": 0.5439456701278687,
+      "step": 1344
+    },
+    {
+      "epoch": 2.8647497337593184,
+      "grad_norm": 0.2578125,
+      "learning_rate": 6.139275482098847e-06,
+      "loss": 0.5950272083282471,
+      "step": 1346
+    },
+    {
+      "epoch": 2.8690095846645365,
+      "grad_norm": 0.255859375,
+      "learning_rate": 6.130851913107137e-06,
+      "loss": 0.5372447967529297,
+      "step": 1348
+    },
+    {
+      "epoch": 2.873269435569755,
+      "grad_norm": 0.2294921875,
+      "learning_rate": 6.122689691004103e-06,
+      "loss": 0.5755343437194824,
+      "step": 1350
+    },
+    {
+      "epoch": 2.8775292864749735,
+      "grad_norm": 0.28515625,
+      "learning_rate": 6.114788995513787e-06,
+      "loss": 0.6370142102241516,
+      "step": 1352
+    },
+    {
+      "epoch": 2.8817891373801916,
+      "grad_norm": 0.28125,
+      "learning_rate": 6.107150000601684e-06,
+      "loss": 0.5815765261650085,
+      "step": 1354
+    },
+    {
+      "epoch": 2.88604898828541,
+      "grad_norm": 0.2490234375,
+      "learning_rate": 6.099772874470899e-06,
+      "loss": 0.6185727715492249,
+      "step": 1356
+    },
+    {
+      "epoch": 2.890308839190628,
+      "grad_norm": 0.36328125,
+      "learning_rate": 6.092657779558442e-06,
+      "loss": 0.5713162422180176,
+      "step": 1358
+    },
+    {
+      "epoch": 2.8945686900958467,
+      "grad_norm": 0.283203125,
+      "learning_rate": 6.08580487253166e-06,
+      "loss": 0.6169173121452332,
+      "step": 1360
+    },
+    {
+      "epoch": 2.8988285410010652,
+      "grad_norm": 0.34375,
+      "learning_rate": 6.079214304284781e-06,
+      "loss": 0.5929686427116394,
+      "step": 1362
+    },
+    {
+      "epoch": 2.9030883919062833,
+      "grad_norm": 0.294921875,
+      "learning_rate": 6.072886219935593e-06,
+      "loss": 0.5761704444885254,
+      "step": 1364
+    },
+    {
+      "epoch": 2.9073482428115014,
+      "grad_norm": 0.25390625,
+      "learning_rate": 6.066820758822244e-06,
+      "loss": 0.5787940621376038,
+      "step": 1366
+    },
+    {
+      "epoch": 2.91160809371672,
+      "grad_norm": 0.26171875,
+      "learning_rate": 6.0610180545001845e-06,
+      "loss": 0.5501613020896912,
+      "step": 1368
+    },
+    {
+      "epoch": 2.9158679446219384,
+      "grad_norm": 0.1943359375,
+      "learning_rate": 6.055478234739217e-06,
+      "loss": 0.5612152218818665,
+      "step": 1370
+    },
+    {
+      "epoch": 2.9201277955271565,
+      "grad_norm": 0.255859375,
+      "learning_rate": 6.050201421520689e-06,
+      "loss": 0.6078463792800903,
+      "step": 1372
+    },
+    {
+      "epoch": 2.924387646432375,
+      "grad_norm": 0.1953125,
+      "learning_rate": 6.045187731034801e-06,
+      "loss": 0.5890936255455017,
+      "step": 1374
+    },
+    {
+      "epoch": 2.928647497337593,
+      "grad_norm": 0.302734375,
+      "learning_rate": 6.040437273678055e-06,
+      "loss": 0.6533024311065674,
+      "step": 1376
+    },
+    {
+      "epoch": 2.9329073482428116,
+      "grad_norm": 4.3125,
+      "learning_rate": 6.0359501540508174e-06,
+      "loss": 0.6827770471572876,
+      "step": 1378
+    },
+    {
+      "epoch": 2.9371671991480297,
+      "grad_norm": 0.3359375,
+      "learning_rate": 6.0317264709550185e-06,
+      "loss": 0.6418617963790894,
+      "step": 1380
+    },
+    {
+      "epoch": 2.9414270500532482,
+      "grad_norm": 0.2431640625,
+      "learning_rate": 6.02776631739198e-06,
+      "loss": 0.5774567127227783,
+      "step": 1382
+    },
+    {
+      "epoch": 2.9456869009584663,
+      "grad_norm": 0.38671875,
+      "learning_rate": 6.0240697805603594e-06,
+      "loss": 0.6014460921287537,
+      "step": 1384
+    },
+    {
+      "epoch": 2.949946751863685,
+      "grad_norm": 0.51171875,
+      "learning_rate": 6.020636941854242e-06,
+      "loss": 0.5642235279083252,
+      "step": 1386
+    },
+    {
+      "epoch": 2.9542066027689033,
+      "grad_norm": 0.333984375,
+      "learning_rate": 6.017467876861333e-06,
+      "loss": 0.5891353487968445,
+      "step": 1388
+    },
+    {
+      "epoch": 2.9584664536741214,
+      "grad_norm": 0.244140625,
+      "learning_rate": 6.014562655361307e-06,
+      "loss": 0.5744375586509705,
+      "step": 1390
+    },
+    {
+      "epoch": 2.9627263045793395,
+      "grad_norm": 0.255859375,
+      "learning_rate": 6.011921341324265e-06,
+      "loss": 0.5458447933197021,
+      "step": 1392
+    },
+    {
+      "epoch": 2.966986155484558,
+      "grad_norm": 0.23828125,
+      "learning_rate": 6.009543992909327e-06,
+      "loss": 0.6621728539466858,
+      "step": 1394
+    },
+    {
+      "epoch": 2.9712460063897765,
+      "grad_norm": 0.236328125,
+      "learning_rate": 6.007430662463352e-06,
+      "loss": 0.5778822898864746,
+      "step": 1396
+    },
+    {
+      "epoch": 2.9755058572949946,
+      "grad_norm": 0.5390625,
+      "learning_rate": 6.005581396519782e-06,
+      "loss": 0.5913535952568054,
+      "step": 1398
+    },
+    {
+      "epoch": 2.979765708200213,
+      "grad_norm": 1.1328125,
+      "learning_rate": 6.0039962357976234e-06,
+      "loss": 0.5911454558372498,
+      "step": 1400
+    },
+    {
+      "epoch": 2.984025559105431,
+      "grad_norm": 0.2197265625,
+      "learning_rate": 6.002675215200546e-06,
+      "loss": 0.5291861295700073,
+      "step": 1402
+    },
+    {
+      "epoch": 2.9882854100106497,
+      "grad_norm": 0.29296875,
+      "learning_rate": 6.001618363816112e-06,
+      "loss": 0.577559232711792,
+      "step": 1404
+    },
+    {
+      "epoch": 2.992545260915868,
+      "grad_norm": 0.28125,
+      "learning_rate": 6.000825704915147e-06,
+      "loss": 0.5995616912841797,
+      "step": 1406
+    },
+    {
+      "epoch": 2.9968051118210863,
+      "grad_norm": 0.251953125,
+      "learning_rate": 6.000297255951213e-06,
+      "loss": 0.5123644471168518,
+      "step": 1408
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 0.2890625,
+      "learning_rate": 6.000033028560234e-06,
+      "loss": 0.584560215473175,
+      "step": 1410
+    },
+    {
+      "epoch": 3.0,
+      "step": 1410,
+      "total_flos": 3.6491913262740275e+18,
+      "train_loss": 0.9255026633857836,
+      "train_runtime": 18814.6678,
+      "train_samples_per_second": 2.396,
+      "train_steps_per_second": 0.075
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 1410,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 99999,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 3.6491913262740275e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}