diff --git "a/checkpoint-3000/trainer_state.json" "b/checkpoint-3000/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-3000/trainer_state.json"
@@ -0,0 +1,10533 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.7536168347216132,
+  "eval_steps": 500,
+  "global_step": 3000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.001169077889814409,
+      "grad_norm": 6.78125,
+      "learning_rate": 3.898635477582846e-07,
+      "loss": 5.0851,
+      "step": 2
+    },
+    {
+      "epoch": 0.002338155779628818,
+      "grad_norm": 7.0625,
+      "learning_rate": 7.797270955165692e-07,
+      "loss": 5.1921,
+      "step": 4
+    },
+    {
+      "epoch": 0.0035072336694432268,
+      "grad_norm": 7.34375,
+      "learning_rate": 1.1695906432748538e-06,
+      "loss": 5.0756,
+      "step": 6
+    },
+    {
+      "epoch": 0.004676311559257636,
+      "grad_norm": 7.03125,
+      "learning_rate": 1.5594541910331385e-06,
+      "loss": 5.1097,
+      "step": 8
+    },
+    {
+      "epoch": 0.005845389449072045,
+      "grad_norm": 6.40625,
+      "learning_rate": 1.949317738791423e-06,
+      "loss": 5.0591,
+      "step": 10
+    },
+    {
+      "epoch": 0.0070144673388864535,
+      "grad_norm": 5.8125,
+      "learning_rate": 2.3391812865497075e-06,
+      "loss": 5.096,
+      "step": 12
+    },
+    {
+      "epoch": 0.008183545228700862,
+      "grad_norm": 5.1875,
+      "learning_rate": 2.729044834307992e-06,
+      "loss": 5.0234,
+      "step": 14
+    },
+    {
+      "epoch": 0.009352623118515271,
+      "grad_norm": 5.03125,
+      "learning_rate": 3.118908382066277e-06,
+      "loss": 5.0266,
+      "step": 16
+    },
+    {
+      "epoch": 0.01052170100832968,
+      "grad_norm": 4.90625,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 5.031,
+      "step": 18
+    },
+    {
+      "epoch": 0.01169077889814409,
+      "grad_norm": 4.90625,
+      "learning_rate": 3.898635477582846e-06,
+      "loss": 4.9845,
+      "step": 20
+    },
+    {
+      "epoch": 0.012859856787958498,
+      "grad_norm": 4.6875,
+      "learning_rate": 4.2884990253411305e-06,
+      "loss": 4.9933,
+      "step": 22
+    },
+    {
+      "epoch": 0.014028934677772907,
+      "grad_norm": 4.625,
+      "learning_rate": 4.678362573099415e-06,
+      "loss": 5.0822,
+      "step": 24
+    },
+    {
+      "epoch": 0.015198012567587316,
+      "grad_norm": 4.71875,
+      "learning_rate": 5.0682261208576995e-06,
+      "loss": 4.9392,
+      "step": 26
+    },
+    {
+      "epoch": 0.016367090457401725,
+      "grad_norm": 4.65625,
+      "learning_rate": 5.458089668615984e-06,
+      "loss": 5.0441,
+      "step": 28
+    },
+    {
+      "epoch": 0.017536168347216132,
+      "grad_norm": 4.6875,
+      "learning_rate": 5.8479532163742686e-06,
+      "loss": 4.8865,
+      "step": 30
+    },
+    {
+      "epoch": 0.018705246237030543,
+      "grad_norm": 4.59375,
+      "learning_rate": 6.237816764132554e-06,
+      "loss": 4.9227,
+      "step": 32
+    },
+    {
+      "epoch": 0.01987432412684495,
+      "grad_norm": 4.65625,
+      "learning_rate": 6.6276803118908384e-06,
+      "loss": 4.9035,
+      "step": 34
+    },
+    {
+      "epoch": 0.02104340201665936,
+      "grad_norm": 4.59375,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 4.863,
+      "step": 36
+    },
+    {
+      "epoch": 0.022212479906473768,
+      "grad_norm": 4.71875,
+      "learning_rate": 7.4074074074074075e-06,
+      "loss": 4.7735,
+      "step": 38
+    },
+    {
+      "epoch": 0.02338155779628818,
+      "grad_norm": 4.84375,
+      "learning_rate": 7.797270955165692e-06,
+      "loss": 4.7932,
+      "step": 40
+    },
+    {
+      "epoch": 0.024550635686102586,
+      "grad_norm": 5.3125,
+      "learning_rate": 8.187134502923977e-06,
+      "loss": 4.6933,
+      "step": 42
+    },
+    {
+      "epoch": 0.025719713575916996,
+      "grad_norm": 7.15625,
+      "learning_rate": 8.576998050682261e-06,
+      "loss": 4.5906,
+      "step": 44
+    },
+    {
+      "epoch": 0.026888791465731404,
+      "grad_norm": 8.8125,
+      "learning_rate": 8.966861598440546e-06,
+      "loss": 4.4291,
+      "step": 46
+    },
+    {
+      "epoch": 0.028057869355545814,
+      "grad_norm": 9.6875,
+      "learning_rate": 9.35672514619883e-06,
+      "loss": 4.221,
+      "step": 48
+    },
+    {
+      "epoch": 0.02922694724536022,
+      "grad_norm": 8.8125,
+      "learning_rate": 9.746588693957115e-06,
+      "loss": 4.032,
+      "step": 50
+    },
+    {
+      "epoch": 0.030396025135174632,
+      "grad_norm": 9.0,
+      "learning_rate": 1.0136452241715399e-05,
+      "loss": 3.8603,
+      "step": 52
+    },
+    {
+      "epoch": 0.03156510302498904,
+      "grad_norm": 7.46875,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 3.6522,
+      "step": 54
+    },
+    {
+      "epoch": 0.03273418091480345,
+      "grad_norm": 8.375,
+      "learning_rate": 1.0916179337231968e-05,
+      "loss": 3.4394,
+      "step": 56
+    },
+    {
+      "epoch": 0.03390325880461786,
+      "grad_norm": 6.4375,
+      "learning_rate": 1.1306042884990253e-05,
+      "loss": 3.2265,
+      "step": 58
+    },
+    {
+      "epoch": 0.035072336694432264,
+      "grad_norm": 5.75,
+      "learning_rate": 1.1695906432748537e-05,
+      "loss": 3.091,
+      "step": 60
+    },
+    {
+      "epoch": 0.03624141458424668,
+      "grad_norm": 4.8125,
+      "learning_rate": 1.2085769980506823e-05,
+      "loss": 2.8856,
+      "step": 62
+    },
+    {
+      "epoch": 0.037410492474061086,
+      "grad_norm": 3.8125,
+      "learning_rate": 1.2475633528265108e-05,
+      "loss": 2.6299,
+      "step": 64
+    },
+    {
+      "epoch": 0.03857957036387549,
+      "grad_norm": 3.234375,
+      "learning_rate": 1.2865497076023392e-05,
+      "loss": 2.5917,
+      "step": 66
+    },
+    {
+      "epoch": 0.0397486482536899,
+      "grad_norm": 4.0625,
+      "learning_rate": 1.3255360623781677e-05,
+      "loss": 2.3757,
+      "step": 68
+    },
+    {
+      "epoch": 0.040917726143504314,
+      "grad_norm": 2.453125,
+      "learning_rate": 1.364522417153996e-05,
+      "loss": 2.308,
+      "step": 70
+    },
+    {
+      "epoch": 0.04208680403331872,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 2.3042,
+      "step": 72
+    },
+    {
+      "epoch": 0.04325588192313313,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.442495126705653e-05,
+      "loss": 2.2149,
+      "step": 74
+    },
+    {
+      "epoch": 0.044424959812947536,
+      "grad_norm": 2.71875,
+      "learning_rate": 1.4814814814814815e-05,
+      "loss": 2.0738,
+      "step": 76
+    },
+    {
+      "epoch": 0.04559403770276195,
+      "grad_norm": 2.46875,
+      "learning_rate": 1.5204678362573099e-05,
+      "loss": 1.9663,
+      "step": 78
+    },
+    {
+      "epoch": 0.04676311559257636,
+      "grad_norm": 2.640625,
+      "learning_rate": 1.5594541910331384e-05,
+      "loss": 1.9384,
+      "step": 80
+    },
+    {
+      "epoch": 0.047932193482390764,
+      "grad_norm": 2.0,
+      "learning_rate": 1.5984405458089668e-05,
+      "loss": 1.894,
+      "step": 82
+    },
+    {
+      "epoch": 0.04910127137220517,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.6374269005847955e-05,
+      "loss": 1.8204,
+      "step": 84
+    },
+    {
+      "epoch": 0.050270349262019585,
+      "grad_norm": 2.953125,
+      "learning_rate": 1.676413255360624e-05,
+      "loss": 1.8051,
+      "step": 86
+    },
+    {
+      "epoch": 0.05143942715183399,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7153996101364522e-05,
+      "loss": 1.5832,
+      "step": 88
+    },
+    {
+      "epoch": 0.0526085050416484,
+      "grad_norm": 1.75,
+      "learning_rate": 1.7543859649122806e-05,
+      "loss": 1.5298,
+      "step": 90
+    },
+    {
+      "epoch": 0.05377758293146281,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.7933723196881093e-05,
+      "loss": 1.4964,
+      "step": 92
+    },
+    {
+      "epoch": 0.054946660821277214,
+      "grad_norm": 1.3515625,
+      "learning_rate": 1.8323586744639376e-05,
+      "loss": 1.3987,
+      "step": 94
+    },
+    {
+      "epoch": 0.05611573871109163,
+      "grad_norm": 1.9140625,
+      "learning_rate": 1.871345029239766e-05,
+      "loss": 1.3959,
+      "step": 96
+    },
+    {
+      "epoch": 0.057284816600906036,
+      "grad_norm": 1.34375,
+      "learning_rate": 1.9103313840155944e-05,
+      "loss": 1.3673,
+      "step": 98
+    },
+    {
+      "epoch": 0.05845389449072044,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.949317738791423e-05,
+      "loss": 1.2864,
+      "step": 100
+    },
+    {
+      "epoch": 0.05962297238053485,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.9883040935672515e-05,
+      "loss": 1.2535,
+      "step": 102
+    },
+    {
+      "epoch": 0.060792050270349264,
+      "grad_norm": 1.2109375,
+      "learning_rate": 2.0272904483430798e-05,
+      "loss": 1.2242,
+      "step": 104
+    },
+    {
+      "epoch": 0.06196112816016367,
+      "grad_norm": 2.59375,
+      "learning_rate": 2.0662768031189085e-05,
+      "loss": 1.1936,
+      "step": 106
+    },
+    {
+      "epoch": 0.06313020604997809,
+      "grad_norm": 1.953125,
+      "learning_rate": 2.105263157894737e-05,
+      "loss": 1.1858,
+      "step": 108
+    },
+    {
+      "epoch": 0.06429928393979249,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.1442495126705653e-05,
+      "loss": 1.1438,
+      "step": 110
+    },
+    {
+      "epoch": 0.0654683618296069,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.1832358674463936e-05,
+      "loss": 1.1217,
+      "step": 112
+    },
+    {
+      "epoch": 0.06663743971942131,
+      "grad_norm": 1.8671875,
+      "learning_rate": 2.2222222222222223e-05,
+      "loss": 1.1508,
+      "step": 114
+    },
+    {
+      "epoch": 0.06780651760923571,
+      "grad_norm": 1.046875,
+      "learning_rate": 2.2612085769980507e-05,
+      "loss": 1.0976,
+      "step": 116
+    },
+    {
+      "epoch": 0.06897559549905012,
+      "grad_norm": 1.234375,
+      "learning_rate": 2.300194931773879e-05,
+      "loss": 1.0716,
+      "step": 118
+    },
+    {
+      "epoch": 0.07014467338886453,
+      "grad_norm": 1.6171875,
+      "learning_rate": 2.3391812865497074e-05,
+      "loss": 1.0152,
+      "step": 120
+    },
+    {
+      "epoch": 0.07131375127867894,
+      "grad_norm": 3.125,
+      "learning_rate": 2.378167641325536e-05,
+      "loss": 1.0265,
+      "step": 122
+    },
+    {
+      "epoch": 0.07248282916849336,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.4171539961013645e-05,
+      "loss": 1.2355,
+      "step": 124
+    },
+    {
+      "epoch": 0.07365190705830776,
+      "grad_norm": 1.1796875,
+      "learning_rate": 2.456140350877193e-05,
+      "loss": 1.0111,
+      "step": 126
+    },
+    {
+      "epoch": 0.07482098494812217,
+      "grad_norm": 2.453125,
+      "learning_rate": 2.4951267056530216e-05,
+      "loss": 1.2317,
+      "step": 128
+    },
+    {
+      "epoch": 0.07599006283793658,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.53411306042885e-05,
+      "loss": 0.9988,
+      "step": 130
+    },
+    {
+      "epoch": 0.07715914072775099,
+      "grad_norm": 1.09375,
+      "learning_rate": 2.5730994152046783e-05,
+      "loss": 1.0076,
+      "step": 132
+    },
+    {
+      "epoch": 0.07832821861756539,
+      "grad_norm": 1.4375,
+      "learning_rate": 2.6120857699805067e-05,
+      "loss": 1.0155,
+      "step": 134
+    },
+    {
+      "epoch": 0.0794972965073798,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.6510721247563354e-05,
+      "loss": 1.1216,
+      "step": 136
+    },
+    {
+      "epoch": 0.08066637439719421,
+      "grad_norm": 0.9921875,
+      "learning_rate": 2.6900584795321637e-05,
+      "loss": 0.9788,
+      "step": 138
+    },
+    {
+      "epoch": 0.08183545228700863,
+      "grad_norm": 1.328125,
+      "learning_rate": 2.729044834307992e-05,
+      "loss": 0.9372,
+      "step": 140
+    },
+    {
+      "epoch": 0.08300453017682304,
+      "grad_norm": 1.296875,
+      "learning_rate": 2.7680311890838205e-05,
+      "loss": 0.897,
+      "step": 142
+    },
+    {
+      "epoch": 0.08417360806663744,
+      "grad_norm": 1.0703125,
+      "learning_rate": 2.8070175438596492e-05,
+      "loss": 0.9766,
+      "step": 144
+    },
+    {
+      "epoch": 0.08534268595645185,
+      "grad_norm": 0.84765625,
+      "learning_rate": 2.8460038986354776e-05,
+      "loss": 0.9315,
+      "step": 146
+    },
+    {
+      "epoch": 0.08651176384626626,
+      "grad_norm": 0.9140625,
+      "learning_rate": 2.884990253411306e-05,
+      "loss": 0.9425,
+      "step": 148
+    },
+    {
+      "epoch": 0.08768084173608066,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.9239766081871346e-05,
+      "loss": 0.9242,
+      "step": 150
+    },
+    {
+      "epoch": 0.08884991962589507,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.962962962962963e-05,
+      "loss": 0.9396,
+      "step": 152
+    },
+    {
+      "epoch": 0.09001899751570948,
+      "grad_norm": 0.92578125,
+      "learning_rate": 3.0019493177387914e-05,
+      "loss": 0.8932,
+      "step": 154
+    },
+    {
+      "epoch": 0.0911880754055239,
+      "grad_norm": 72.5,
+      "learning_rate": 3.0409356725146197e-05,
+      "loss": 0.9062,
+      "step": 156
+    },
+    {
+      "epoch": 0.0923571532953383,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.0799220272904484e-05,
+      "loss": 0.9534,
+      "step": 158
+    },
+    {
+      "epoch": 0.09352623118515271,
+      "grad_norm": 1.453125,
+      "learning_rate": 3.118908382066277e-05,
+      "loss": 0.8715,
+      "step": 160
+    },
+    {
+      "epoch": 0.09469530907496712,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.157894736842105e-05,
+      "loss": 0.9029,
+      "step": 162
+    },
+    {
+      "epoch": 0.09586438696478153,
+      "grad_norm": 1.15625,
+      "learning_rate": 3.1968810916179335e-05,
+      "loss": 0.8968,
+      "step": 164
+    },
+    {
+      "epoch": 0.09703346485459594,
+      "grad_norm": 1.4296875,
+      "learning_rate": 3.235867446393762e-05,
+      "loss": 0.9784,
+      "step": 166
+    },
+    {
+      "epoch": 0.09820254274441034,
+      "grad_norm": 1.0078125,
+      "learning_rate": 3.274853801169591e-05,
+      "loss": 0.9012,
+      "step": 168
+    },
+    {
+      "epoch": 0.09937162063422475,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.313840155945419e-05,
+      "loss": 0.8895,
+      "step": 170
+    },
+    {
+      "epoch": 0.10054069852403917,
+      "grad_norm": 0.8046875,
+      "learning_rate": 3.352826510721248e-05,
+      "loss": 0.8858,
+      "step": 172
+    },
+    {
+      "epoch": 0.10170977641385358,
+      "grad_norm": 0.83203125,
+      "learning_rate": 3.391812865497076e-05,
+      "loss": 0.8408,
+      "step": 174
+    },
+    {
+      "epoch": 0.10287885430366799,
+      "grad_norm": 0.8671875,
+      "learning_rate": 3.4307992202729044e-05,
+      "loss": 0.8896,
+      "step": 176
+    },
+    {
+      "epoch": 0.10404793219348239,
+      "grad_norm": 1.78125,
+      "learning_rate": 3.469785575048733e-05,
+      "loss": 0.8941,
+      "step": 178
+    },
+    {
+      "epoch": 0.1052170100832968,
+      "grad_norm": 0.8984375,
+      "learning_rate": 3.508771929824561e-05,
+      "loss": 0.9013,
+      "step": 180
+    },
+    {
+      "epoch": 0.1063860879731112,
+      "grad_norm": 1.2734375,
+      "learning_rate": 3.5477582846003895e-05,
+      "loss": 0.8585,
+      "step": 182
+    },
+    {
+      "epoch": 0.10755516586292561,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.5867446393762186e-05,
+      "loss": 1.0231,
+      "step": 184
+    },
+    {
+      "epoch": 0.10872424375274002,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3.625730994152047e-05,
+      "loss": 0.8835,
+      "step": 186
+    },
+    {
+      "epoch": 0.10989332164255443,
+      "grad_norm": 1.078125,
+      "learning_rate": 3.664717348927875e-05,
+      "loss": 0.8465,
+      "step": 188
+    },
+    {
+      "epoch": 0.11106239953236885,
+      "grad_norm": 2.75,
+      "learning_rate": 3.7037037037037037e-05,
+      "loss": 1.1014,
+      "step": 190
+    },
+    {
+      "epoch": 0.11223147742218326,
+      "grad_norm": 2.09375,
+      "learning_rate": 3.742690058479532e-05,
+      "loss": 0.8691,
+      "step": 192
+    },
+    {
+      "epoch": 0.11340055531199766,
+      "grad_norm": 0.96875,
+      "learning_rate": 3.7816764132553604e-05,
+      "loss": 0.8547,
+      "step": 194
+    },
+    {
+      "epoch": 0.11456963320181207,
+      "grad_norm": 1.5859375,
+      "learning_rate": 3.820662768031189e-05,
+      "loss": 0.8725,
+      "step": 196
+    },
+    {
+      "epoch": 0.11573871109162648,
+      "grad_norm": 0.88671875,
+      "learning_rate": 3.859649122807018e-05,
+      "loss": 0.8393,
+      "step": 198
+    },
+    {
+      "epoch": 0.11690778898144089,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.898635477582846e-05,
+      "loss": 0.8345,
+      "step": 200
+    },
+    {
+      "epoch": 0.11807686687125529,
+      "grad_norm": 0.82421875,
+      "learning_rate": 3.9376218323586745e-05,
+      "loss": 0.8966,
+      "step": 202
+    },
+    {
+      "epoch": 0.1192459447610697,
+      "grad_norm": 0.79296875,
+      "learning_rate": 3.976608187134503e-05,
+      "loss": 0.8291,
+      "step": 204
+    },
+    {
+      "epoch": 0.12041502265088412,
+      "grad_norm": 0.83203125,
+      "learning_rate": 4.015594541910331e-05,
+      "loss": 0.8175,
+      "step": 206
+    },
+    {
+      "epoch": 0.12158410054069853,
+      "grad_norm": 0.74609375,
+      "learning_rate": 4.0545808966861596e-05,
+      "loss": 0.8353,
+      "step": 208
+    },
+    {
+      "epoch": 0.12275317843051294,
+      "grad_norm": 1.796875,
+      "learning_rate": 4.093567251461988e-05,
+      "loss": 1.0084,
+      "step": 210
+    },
+    {
+      "epoch": 0.12392225632032734,
+      "grad_norm": 59.75,
+      "learning_rate": 4.132553606237817e-05,
+      "loss": 0.8478,
+      "step": 212
+    },
+    {
+      "epoch": 0.12509133421014176,
+      "grad_norm": 1.265625,
+      "learning_rate": 4.1715399610136454e-05,
+      "loss": 0.8267,
+      "step": 214
+    },
+    {
+      "epoch": 0.12626041209995617,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.210526315789474e-05,
+      "loss": 0.836,
+      "step": 216
+    },
+    {
+      "epoch": 0.12742948998977058,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.249512670565302e-05,
+      "loss": 0.857,
+      "step": 218
+    },
+    {
+      "epoch": 0.12859856787958499,
+      "grad_norm": 0.92578125,
+      "learning_rate": 4.2884990253411305e-05,
+      "loss": 0.8478,
+      "step": 220
+    },
+    {
+      "epoch": 0.1297676457693994,
+      "grad_norm": 0.953125,
+      "learning_rate": 4.327485380116959e-05,
+      "loss": 0.844,
+      "step": 222
+    },
+    {
+      "epoch": 0.1309367236592138,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.366471734892787e-05,
+      "loss": 0.8128,
+      "step": 224
+    },
+    {
+      "epoch": 0.1321058015490282,
+      "grad_norm": 1.28125,
+      "learning_rate": 4.4054580896686156e-05,
+      "loss": 0.8276,
+      "step": 226
+    },
+    {
+      "epoch": 0.13327487943884261,
+      "grad_norm": 0.94921875,
+      "learning_rate": 4.4444444444444447e-05,
+      "loss": 0.8478,
+      "step": 228
+    },
+    {
+      "epoch": 0.13444395732865702,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.483430799220273e-05,
+      "loss": 0.814,
+      "step": 230
+    },
+    {
+      "epoch": 0.13561303521847143,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.5224171539961014e-05,
+      "loss": 0.765,
+      "step": 232
+    },
+    {
+      "epoch": 0.13678211310828584,
+      "grad_norm": 0.87109375,
+      "learning_rate": 4.56140350877193e-05,
+      "loss": 0.8277,
+      "step": 234
+    },
+    {
+      "epoch": 0.13795119099810024,
+      "grad_norm": 0.921875,
+      "learning_rate": 4.600389863547758e-05,
+      "loss": 0.8103,
+      "step": 236
+    },
+    {
+      "epoch": 0.13912026888791465,
+      "grad_norm": 1.4921875,
+      "learning_rate": 4.6393762183235865e-05,
+      "loss": 0.7897,
+      "step": 238
+    },
+    {
+      "epoch": 0.14028934677772906,
+      "grad_norm": 0.7734375,
+      "learning_rate": 4.678362573099415e-05,
+      "loss": 0.7798,
+      "step": 240
+    },
+    {
+      "epoch": 0.14145842466754346,
+      "grad_norm": 0.98828125,
+      "learning_rate": 4.717348927875244e-05,
+      "loss": 0.8442,
+      "step": 242
+    },
+    {
+      "epoch": 0.14262750255735787,
+      "grad_norm": 0.91796875,
+      "learning_rate": 4.756335282651072e-05,
+      "loss": 0.7936,
+      "step": 244
+    },
+    {
+      "epoch": 0.1437965804471723,
+      "grad_norm": 0.859375,
+      "learning_rate": 4.7953216374269006e-05,
+      "loss": 0.7726,
+      "step": 246
+    },
+    {
+      "epoch": 0.1449656583369867,
+      "grad_norm": 1.3671875,
+      "learning_rate": 4.834307992202729e-05,
+      "loss": 0.9626,
+      "step": 248
+    },
+    {
+      "epoch": 0.14613473622680112,
+      "grad_norm": 0.921875,
+      "learning_rate": 4.8732943469785574e-05,
+      "loss": 0.7929,
+      "step": 250
+    },
+    {
+      "epoch": 0.14730381411661553,
+      "grad_norm": 0.92578125,
+      "learning_rate": 4.912280701754386e-05,
+      "loss": 0.8589,
+      "step": 252
+    },
+    {
+      "epoch": 0.14847289200642994,
+      "grad_norm": 0.83203125,
+      "learning_rate": 4.951267056530214e-05,
+      "loss": 0.7383,
+      "step": 254
+    },
+    {
+      "epoch": 0.14964196989624434,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.990253411306043e-05,
+      "loss": 0.8011,
+      "step": 256
+    },
+    {
+      "epoch": 0.15081104778605875,
+      "grad_norm": 1.3359375,
+      "learning_rate": 5.0292397660818715e-05,
+      "loss": 0.8404,
+      "step": 258
+    },
+    {
+      "epoch": 0.15198012567587316,
+      "grad_norm": 1.21875,
+      "learning_rate": 5.0682261208577e-05,
+      "loss": 0.8301,
+      "step": 260
+    },
+    {
+      "epoch": 0.15314920356568756,
+      "grad_norm": 0.93359375,
+      "learning_rate": 5.107212475633528e-05,
+      "loss": 0.8157,
+      "step": 262
+    },
+    {
+      "epoch": 0.15431828145550197,
+      "grad_norm": 1.3203125,
+      "learning_rate": 5.1461988304093566e-05,
+      "loss": 0.7724,
+      "step": 264
+    },
+    {
+      "epoch": 0.15548735934531638,
+      "grad_norm": 1.09375,
+      "learning_rate": 5.185185185185185e-05,
+      "loss": 0.7586,
+      "step": 266
+    },
+    {
+      "epoch": 0.15665643723513079,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.2241715399610133e-05,
+      "loss": 0.7812,
+      "step": 268
+    },
+    {
+      "epoch": 0.1578255151249452,
+      "grad_norm": 0.74609375,
+      "learning_rate": 5.2631578947368424e-05,
+      "loss": 0.7486,
+      "step": 270
+    },
+    {
+      "epoch": 0.1589945930147596,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.302144249512671e-05,
+      "loss": 0.7339,
+      "step": 272
+    },
+    {
+      "epoch": 0.160163670904574,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.341130604288499e-05,
+      "loss": 0.8381,
+      "step": 274
+    },
+    {
+      "epoch": 0.16133274879438841,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.3801169590643275e-05,
+      "loss": 0.774,
+      "step": 276
+    },
+    {
+      "epoch": 0.16250182668420282,
+      "grad_norm": 0.796875,
+      "learning_rate": 5.419103313840156e-05,
+      "loss": 0.8114,
+      "step": 278
+    },
+    {
+      "epoch": 0.16367090457401726,
+      "grad_norm": 0.9140625,
+      "learning_rate": 5.458089668615984e-05,
+      "loss": 0.7549,
+      "step": 280
+    },
+    {
+      "epoch": 0.16483998246383166,
+      "grad_norm": 1.0625,
+      "learning_rate": 5.4970760233918126e-05,
+      "loss": 0.7393,
+      "step": 282
+    },
+    {
+      "epoch": 0.16600906035364607,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.536062378167641e-05,
+      "loss": 0.7607,
+      "step": 284
+    },
+    {
+      "epoch": 0.16717813824346048,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.57504873294347e-05,
+      "loss": 0.8158,
+      "step": 286
+    },
+    {
+      "epoch": 0.16834721613327489,
+      "grad_norm": 1.734375,
+      "learning_rate": 5.6140350877192984e-05,
+      "loss": 0.7403,
+      "step": 288
+    },
+    {
+      "epoch": 0.1695162940230893,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.653021442495127e-05,
+      "loss": 0.7278,
+      "step": 290
+    },
+    {
+      "epoch": 0.1706853719129037,
+      "grad_norm": 1.7578125,
+      "learning_rate": 5.692007797270955e-05,
+      "loss": 0.821,
+      "step": 292
+    },
+    {
+      "epoch": 0.1718544498027181,
+      "grad_norm": 3.1875,
+      "learning_rate": 5.7309941520467835e-05,
+      "loss": 0.8276,
+      "step": 294
+    },
+    {
+      "epoch": 0.1730235276925325,
+      "grad_norm": 1.1484375,
+      "learning_rate": 5.769980506822612e-05,
+      "loss": 0.8505,
+      "step": 296
+    },
+    {
+      "epoch": 0.17419260558234692,
+      "grad_norm": 1.2265625,
+      "learning_rate": 5.80896686159844e-05,
+      "loss": 0.7941,
+      "step": 298
+    },
+    {
+      "epoch": 0.17536168347216133,
+      "grad_norm": 1.4765625,
+      "learning_rate": 5.847953216374269e-05,
+      "loss": 0.8621,
+      "step": 300
+    },
+    {
+      "epoch": 0.17653076136197574,
+      "grad_norm": 2.25,
+      "learning_rate": 5.8869395711500976e-05,
+      "loss": 0.8071,
+      "step": 302
+    },
+    {
+      "epoch": 0.17769983925179014,
+      "grad_norm": 0.92578125,
+      "learning_rate": 5.925925925925926e-05,
+      "loss": 0.803,
+      "step": 304
+    },
+    {
+      "epoch": 0.17886891714160455,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.9649122807017544e-05,
+      "loss": 0.7569,
+      "step": 306
+    },
+    {
+      "epoch": 0.18003799503141896,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.003898635477583e-05,
+      "loss": 0.7722,
+      "step": 308
+    },
+    {
+      "epoch": 0.18120707292123336,
+      "grad_norm": 0.984375,
+      "learning_rate": 6.042884990253411e-05,
+      "loss": 0.7746,
+      "step": 310
+    },
+    {
+      "epoch": 0.1823761508110478,
+      "grad_norm": 2.296875,
+      "learning_rate": 6.0818713450292395e-05,
+      "loss": 0.7292,
+      "step": 312
+    },
+    {
+      "epoch": 0.1835452287008622,
+      "grad_norm": 1.140625,
+      "learning_rate": 6.120857699805068e-05,
+      "loss": 0.7751,
+      "step": 314
+    },
+    {
+      "epoch": 0.1847143065906766,
+      "grad_norm": 0.8203125,
+      "learning_rate": 6.159844054580897e-05,
+      "loss": 0.7473,
+      "step": 316
+    },
+    {
+      "epoch": 0.18588338448049102,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.198830409356725e-05,
+      "loss": 0.7718,
+      "step": 318
+    },
+    {
+      "epoch": 0.18705246237030543,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.237816764132554e-05,
+      "loss": 0.7812,
+      "step": 320
+    },
+    {
+      "epoch": 0.18822154026011984,
+      "grad_norm": 0.859375,
+      "learning_rate": 6.276803118908382e-05,
+      "loss": 0.7552,
+      "step": 322
+    },
+    {
+      "epoch": 0.18939061814993424,
+      "grad_norm": 0.765625,
+      "learning_rate": 6.31578947368421e-05,
+      "loss": 0.7244,
+      "step": 324
+    },
+    {
+      "epoch": 0.19055969603974865,
+      "grad_norm": 1.2734375,
+      "learning_rate": 6.354775828460039e-05,
+      "loss": 0.8182,
+      "step": 326
+    },
+    {
+      "epoch": 0.19172877392956306,
+      "grad_norm": 0.93359375,
+      "learning_rate": 6.393762183235867e-05,
+      "loss": 0.7578,
+      "step": 328
+    },
+    {
+      "epoch": 0.19289785181937746,
+      "grad_norm": 0.875,
+      "learning_rate": 6.432748538011695e-05,
+      "loss": 0.8755,
+      "step": 330
+    },
+    {
+      "epoch": 0.19406692970919187,
+      "grad_norm": 1.4296875,
+      "learning_rate": 6.471734892787524e-05,
+      "loss": 0.7436,
+      "step": 332
+    },
+    {
+      "epoch": 0.19523600759900628,
+      "grad_norm": 1.859375,
+      "learning_rate": 6.510721247563352e-05,
+      "loss": 0.7415,
+      "step": 334
+    },
+    {
+      "epoch": 0.19640508548882069,
+      "grad_norm": 1.265625,
+      "learning_rate": 6.549707602339182e-05,
+      "loss": 0.8082,
+      "step": 336
+    },
+    {
+      "epoch": 0.1975741633786351,
+      "grad_norm": 1.4453125,
+      "learning_rate": 6.58869395711501e-05,
+      "loss": 0.7226,
+      "step": 338
+    },
+    {
+      "epoch": 0.1987432412684495,
+      "grad_norm": 0.8203125,
+      "learning_rate": 6.627680311890839e-05,
+      "loss": 0.7408,
+      "step": 340
+    },
+    {
+      "epoch": 0.1999123191582639,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.666666666666667e-05,
+      "loss": 0.7752,
+      "step": 342
+    },
+    {
+      "epoch": 0.20108139704807834,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.705653021442495e-05,
+      "loss": 0.7313,
+      "step": 344
+    },
+    {
+      "epoch": 0.20225047493789275,
+      "grad_norm": 1.1015625,
+      "learning_rate": 6.744639376218324e-05,
+      "loss": 0.7164,
+      "step": 346
+    },
+    {
+      "epoch": 0.20341955282770716,
+      "grad_norm": 0.9296875,
+      "learning_rate": 6.783625730994152e-05,
+      "loss": 0.7836,
+      "step": 348
+    },
+    {
+      "epoch": 0.20458863071752156,
+      "grad_norm": 1.9140625,
+      "learning_rate": 6.82261208576998e-05,
+      "loss": 0.8936,
+      "step": 350
+    },
+    {
+      "epoch": 0.20575770860733597,
+      "grad_norm": 0.8984375,
+      "learning_rate": 6.861598440545809e-05,
+      "loss": 0.7386,
+      "step": 352
+    },
+    {
+      "epoch": 0.20692678649715038,
+      "grad_norm": 0.9609375,
+      "learning_rate": 6.900584795321637e-05,
+      "loss": 0.8893,
+      "step": 354
+    },
+    {
+      "epoch": 0.20809586438696479,
+      "grad_norm": 0.9140625,
+      "learning_rate": 6.939571150097466e-05,
+      "loss": 0.715,
+      "step": 356
+    },
+    {
+      "epoch": 0.2092649422767792,
+      "grad_norm": 0.94921875,
+      "learning_rate": 6.978557504873294e-05,
+      "loss": 0.7087,
+      "step": 358
+    },
+    {
+      "epoch": 0.2104340201665936,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.017543859649122e-05,
+      "loss": 0.7067,
+      "step": 360
+    },
+    {
+      "epoch": 0.211603098056408,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.05653021442495e-05,
+      "loss": 0.6979,
+      "step": 362
+    },
+    {
+      "epoch": 0.2127721759462224,
+      "grad_norm": 1.8046875,
+      "learning_rate": 7.095516569200779e-05,
+      "loss": 0.8533,
+      "step": 364
+    },
+    {
+      "epoch": 0.21394125383603682,
+      "grad_norm": 0.95703125,
+      "learning_rate": 7.134502923976609e-05,
+      "loss": 0.6971,
+      "step": 366
+    },
+    {
+      "epoch": 0.21511033172585123,
+      "grad_norm": 1.1640625,
+      "learning_rate": 7.173489278752437e-05,
+      "loss": 0.7152,
+      "step": 368
+    },
+    {
+      "epoch": 0.21627940961566564,
+      "grad_norm": 1.3984375,
+      "learning_rate": 7.212475633528265e-05,
+      "loss": 0.7176,
+      "step": 370
+    },
+    {
+      "epoch": 0.21744848750548004,
+      "grad_norm": 0.84375,
+      "learning_rate": 7.251461988304094e-05,
+      "loss": 0.7054,
+      "step": 372
+    },
+    {
+      "epoch": 0.21861756539529445,
+      "grad_norm": 0.859375,
+      "learning_rate": 7.290448343079922e-05,
+      "loss": 0.7157,
+      "step": 374
+    },
+    {
+      "epoch": 0.21978664328510886,
+      "grad_norm": 0.99609375,
+      "learning_rate": 7.32943469785575e-05,
+      "loss": 0.742,
+      "step": 376
+    },
+    {
+      "epoch": 0.2209557211749233,
+      "grad_norm": 1.1171875,
+      "learning_rate": 7.368421052631579e-05,
+      "loss": 0.7016,
+      "step": 378
+    },
+    {
+      "epoch": 0.2221247990647377,
+      "grad_norm": 1.3359375,
+      "learning_rate": 7.407407407407407e-05,
+      "loss": 0.7263,
+      "step": 380
+    },
+    {
+      "epoch": 0.2232938769545521,
+      "grad_norm": 0.8515625,
+      "learning_rate": 7.446393762183236e-05,
+      "loss": 0.7386,
+      "step": 382
+    },
+    {
+      "epoch": 0.2244629548443665,
+      "grad_norm": 1.3359375,
+      "learning_rate": 7.485380116959064e-05,
+      "loss": 0.7488,
+      "step": 384
+    },
+    {
+      "epoch": 0.22563203273418092,
+      "grad_norm": 0.98828125,
+      "learning_rate": 7.524366471734892e-05,
+      "loss": 0.7304,
+      "step": 386
+    },
+    {
+      "epoch": 0.22680111062399533,
+      "grad_norm": 0.90625,
+      "learning_rate": 7.563352826510721e-05,
+      "loss": 0.6926,
+      "step": 388
+    },
+    {
+      "epoch": 0.22797018851380974,
+      "grad_norm": 1.578125,
+      "learning_rate": 7.602339181286549e-05,
+      "loss": 0.7258,
+      "step": 390
+    },
+    {
+      "epoch": 0.22913926640362414,
+      "grad_norm": 0.82421875,
+      "learning_rate": 7.641325536062378e-05,
+      "loss": 0.7169,
+      "step": 392
+    },
+    {
+      "epoch": 0.23030834429343855,
+      "grad_norm": 0.78125,
+      "learning_rate": 7.680311890838207e-05,
+      "loss": 0.6962,
+      "step": 394
+    },
+    {
+      "epoch": 0.23147742218325296,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.719298245614036e-05,
+      "loss": 0.718,
+      "step": 396
+    },
+    {
+      "epoch": 0.23264650007306736,
+      "grad_norm": 0.84375,
+      "learning_rate": 7.758284600389864e-05,
+      "loss": 0.7106,
+      "step": 398
+    },
+    {
+      "epoch": 0.23381557796288177,
+      "grad_norm": 1.109375,
+      "learning_rate": 7.797270955165692e-05,
+      "loss": 0.739,
+      "step": 400
+    },
+    {
+      "epoch": 0.23498465585269618,
+      "grad_norm": 0.9140625,
+      "learning_rate": 7.836257309941521e-05,
+      "loss": 0.7037,
+      "step": 402
+    },
+    {
+      "epoch": 0.23615373374251059,
+      "grad_norm": 0.76953125,
+      "learning_rate": 7.875243664717349e-05,
+      "loss": 0.717,
+      "step": 404
+    },
+    {
+      "epoch": 0.237322811632325,
+      "grad_norm": 1.0546875,
+      "learning_rate": 7.914230019493177e-05,
+      "loss": 0.7041,
+      "step": 406
+    },
+    {
+      "epoch": 0.2384918895221394,
+      "grad_norm": 0.921875,
+      "learning_rate": 7.953216374269006e-05,
+      "loss": 0.7016,
+      "step": 408
+    },
+    {
+      "epoch": 0.23966096741195383,
+      "grad_norm": 0.8515625,
+      "learning_rate": 7.992202729044834e-05,
+      "loss": 0.7423,
+      "step": 410
+    },
+    {
+      "epoch": 0.24083004530176824,
+      "grad_norm": 0.88671875,
+      "learning_rate": 8.031189083820663e-05,
+      "loss": 0.7128,
+      "step": 412
+    },
+    {
+      "epoch": 0.24199912319158265,
+      "grad_norm": 0.9453125,
+      "learning_rate": 8.070175438596491e-05,
+      "loss": 0.7144,
+      "step": 414
+    },
+    {
+      "epoch": 0.24316820108139706,
+      "grad_norm": 1.1328125,
+      "learning_rate": 8.109161793372319e-05,
+      "loss": 0.6987,
+      "step": 416
+    },
+    {
+      "epoch": 0.24433727897121146,
+      "grad_norm": 0.9296875,
+      "learning_rate": 8.148148148148148e-05,
+      "loss": 0.6946,
+      "step": 418
+    },
+    {
+      "epoch": 0.24550635686102587,
+      "grad_norm": 0.9296875,
+      "learning_rate": 8.187134502923976e-05,
+      "loss": 0.6954,
+      "step": 420
+    },
+    {
+      "epoch": 0.24667543475084028,
+      "grad_norm": 0.84765625,
+      "learning_rate": 8.226120857699804e-05,
+      "loss": 0.7507,
+      "step": 422
+    },
+    {
+      "epoch": 0.24784451264065469,
+      "grad_norm": 2.171875,
+      "learning_rate": 8.265107212475634e-05,
+      "loss": 0.9267,
+      "step": 424
+    },
+    {
+      "epoch": 0.2490135905304691,
+      "grad_norm": 0.81640625,
+      "learning_rate": 8.304093567251462e-05,
+      "loss": 0.7222,
+      "step": 426
+    },
+    {
+      "epoch": 0.2501826684202835,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.343079922027291e-05,
+      "loss": 0.8919,
+      "step": 428
+    },
+    {
+      "epoch": 0.25135174631009793,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.382066276803119e-05,
+      "loss": 0.7451,
+      "step": 430
+    },
+    {
+      "epoch": 0.25252082419991234,
+      "grad_norm": 0.81640625,
+      "learning_rate": 8.421052631578948e-05,
+      "loss": 0.7027,
+      "step": 432
+    },
+    {
+      "epoch": 0.25368990208972675,
+      "grad_norm": 0.79296875,
+      "learning_rate": 8.460038986354776e-05,
+      "loss": 0.712,
+      "step": 434
+    },
+    {
+      "epoch": 0.25485897997954116,
+      "grad_norm": 0.8984375,
+      "learning_rate": 8.499025341130604e-05,
+      "loss": 0.7122,
+      "step": 436
+    },
+    {
+      "epoch": 0.25602805786935556,
+      "grad_norm": 0.72265625,
+      "learning_rate": 8.538011695906433e-05,
+      "loss": 0.6871,
+      "step": 438
+    },
+    {
+      "epoch": 0.25719713575916997,
+      "grad_norm": 0.77734375,
+      "learning_rate": 8.576998050682261e-05,
+      "loss": 0.7278,
+      "step": 440
+    },
+    {
+      "epoch": 0.2583662136489844,
+      "grad_norm": 1.1171875,
+      "learning_rate": 8.61598440545809e-05,
+      "loss": 0.7023,
+      "step": 442
+    },
+    {
+      "epoch": 0.2595352915387988,
+      "grad_norm": 0.8359375,
+      "learning_rate": 8.654970760233918e-05,
+      "loss": 0.7167,
+      "step": 444
+    },
+    {
+      "epoch": 0.2607043694286132,
+      "grad_norm": 0.8203125,
+      "learning_rate": 8.693957115009746e-05,
+      "loss": 0.7346,
+      "step": 446
+    },
+    {
+      "epoch": 0.2618734473184276,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.732943469785574e-05,
+      "loss": 0.7223,
+      "step": 448
+    },
+    {
+      "epoch": 0.263042525208242,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.771929824561403e-05,
+      "loss": 0.7032,
+      "step": 450
+    },
+    {
+      "epoch": 0.2642116030980564,
+      "grad_norm": 0.9609375,
+      "learning_rate": 8.810916179337231e-05,
+      "loss": 0.6804,
+      "step": 452
+    },
+    {
+      "epoch": 0.2653806809878708,
+      "grad_norm": 0.875,
+      "learning_rate": 8.849902534113061e-05,
+      "loss": 0.7224,
+      "step": 454
+    },
+    {
+      "epoch": 0.26654975887768523,
+      "grad_norm": 0.8359375,
+      "learning_rate": 8.888888888888889e-05,
+      "loss": 0.7138,
+      "step": 456
+    },
+    {
+      "epoch": 0.26771883676749963,
+      "grad_norm": 0.859375,
+      "learning_rate": 8.927875243664718e-05,
+      "loss": 0.7065,
+      "step": 458
+    },
+    {
+      "epoch": 0.26888791465731404,
+      "grad_norm": 0.96484375,
+      "learning_rate": 8.966861598440546e-05,
+      "loss": 0.7289,
+      "step": 460
+    },
+    {
+      "epoch": 0.27005699254712845,
+      "grad_norm": 1.0703125,
+      "learning_rate": 9.005847953216374e-05,
+      "loss": 0.6917,
+      "step": 462
+    },
+    {
+      "epoch": 0.27122607043694286,
+      "grad_norm": 1.0078125,
+      "learning_rate": 9.044834307992203e-05,
+      "loss": 0.7493,
+      "step": 464
+    },
+    {
+      "epoch": 0.27239514832675726,
+      "grad_norm": 1.0546875,
+      "learning_rate": 9.083820662768031e-05,
+      "loss": 0.7075,
+      "step": 466
+    },
+    {
+      "epoch": 0.27356422621657167,
+      "grad_norm": 0.88671875,
+      "learning_rate": 9.12280701754386e-05,
+      "loss": 0.6926,
+      "step": 468
+    },
+    {
+      "epoch": 0.2747333041063861,
+      "grad_norm": 0.734375,
+      "learning_rate": 9.161793372319688e-05,
+      "loss": 0.6714,
+      "step": 470
+    },
+    {
+      "epoch": 0.2759023819962005,
+      "grad_norm": 0.73046875,
+      "learning_rate": 9.200779727095516e-05,
+      "loss": 0.6973,
+      "step": 472
+    },
+    {
+      "epoch": 0.2770714598860149,
+      "grad_norm": 0.7578125,
+      "learning_rate": 9.239766081871345e-05,
+      "loss": 0.7289,
+      "step": 474
+    },
+    {
+      "epoch": 0.2782405377758293,
+      "grad_norm": 0.84765625,
+      "learning_rate": 9.278752436647173e-05,
+      "loss": 0.7296,
+      "step": 476
+    },
+    {
+      "epoch": 0.2794096156656437,
+      "grad_norm": 0.99609375,
+      "learning_rate": 9.317738791423001e-05,
+      "loss": 0.7229,
+      "step": 478
+    },
+    {
+      "epoch": 0.2805786935554581,
+      "grad_norm": 1.0859375,
+      "learning_rate": 9.35672514619883e-05,
+      "loss": 0.7158,
+      "step": 480
+    },
+    {
+      "epoch": 0.2817477714452725,
+      "grad_norm": 0.8671875,
+      "learning_rate": 9.39571150097466e-05,
+      "loss": 0.6862,
+      "step": 482
+    },
+    {
+      "epoch": 0.28291684933508693,
+      "grad_norm": 0.77734375,
+      "learning_rate": 9.434697855750488e-05,
+      "loss": 0.6535,
+      "step": 484
+    },
+    {
+      "epoch": 0.28408592722490134,
+      "grad_norm": 0.90234375,
+      "learning_rate": 9.473684210526316e-05,
+      "loss": 0.703,
+      "step": 486
+    },
+    {
+      "epoch": 0.28525500511471574,
+      "grad_norm": 0.80078125,
+      "learning_rate": 9.512670565302145e-05,
+      "loss": 0.7296,
+      "step": 488
+    },
+    {
+      "epoch": 0.28642408300453015,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.551656920077973e-05,
+      "loss": 0.7061,
+      "step": 490
+    },
+    {
+      "epoch": 0.2875931608943446,
+      "grad_norm": 0.76953125,
+      "learning_rate": 9.590643274853801e-05,
+      "loss": 0.6731,
+      "step": 492
+    },
+    {
+      "epoch": 0.288762238784159,
+      "grad_norm": 1.0234375,
+      "learning_rate": 9.62962962962963e-05,
+      "loss": 0.7141,
+      "step": 494
+    },
+    {
+      "epoch": 0.2899313166739734,
+      "grad_norm": 1.8359375,
+      "learning_rate": 9.668615984405458e-05,
+      "loss": 0.7874,
+      "step": 496
+    },
+    {
+      "epoch": 0.29110039456378783,
+      "grad_norm": 0.97265625,
+      "learning_rate": 9.707602339181286e-05,
+      "loss": 0.7179,
+      "step": 498
+    },
+    {
+      "epoch": 0.29226947245360224,
+      "grad_norm": 0.8984375,
+      "learning_rate": 9.746588693957115e-05,
+      "loss": 0.6819,
+      "step": 500
+    },
+    {
+      "epoch": 0.29343855034341665,
+      "grad_norm": 0.8359375,
+      "learning_rate": 9.785575048732943e-05,
+      "loss": 0.688,
+      "step": 502
+    },
+    {
+      "epoch": 0.29460762823323106,
+      "grad_norm": 2.21875,
+      "learning_rate": 9.824561403508771e-05,
+      "loss": 0.8356,
+      "step": 504
+    },
+    {
+      "epoch": 0.29577670612304546,
+      "grad_norm": 2.171875,
+      "learning_rate": 9.8635477582846e-05,
+      "loss": 0.7733,
+      "step": 506
+    },
+    {
+      "epoch": 0.29694578401285987,
+      "grad_norm": 0.8203125,
+      "learning_rate": 9.902534113060428e-05,
+      "loss": 0.7087,
+      "step": 508
+    },
+    {
+      "epoch": 0.2981148619026743,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.941520467836257e-05,
+      "loss": 0.7023,
+      "step": 510
+    },
+    {
+      "epoch": 0.2992839397924887,
+      "grad_norm": 1.28125,
+      "learning_rate": 9.980506822612086e-05,
+      "loss": 0.7087,
+      "step": 512
+    },
+    {
+      "epoch": 0.3004530176823031,
+      "grad_norm": 1.328125,
+      "learning_rate": 0.00010019493177387915,
+      "loss": 0.7002,
+      "step": 514
+    },
+    {
+      "epoch": 0.3016220955721175,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00010058479532163743,
+      "loss": 0.7355,
+      "step": 516
+    },
+    {
+      "epoch": 0.3027911734619319,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00010097465886939573,
+      "loss": 0.6983,
+      "step": 518
+    },
+    {
+      "epoch": 0.3039602513517463,
+      "grad_norm": 1.65625,
+      "learning_rate": 0.000101364522417154,
+      "loss": 0.7016,
+      "step": 520
+    },
+    {
+      "epoch": 0.3051293292415607,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.0001017543859649123,
+      "loss": 0.7096,
+      "step": 522
+    },
+    {
+      "epoch": 0.3062984071313751,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00010214424951267056,
+      "loss": 0.7099,
+      "step": 524
+    },
+    {
+      "epoch": 0.30746748502118953,
+      "grad_norm": 1.4375,
+      "learning_rate": 0.00010253411306042886,
+      "loss": 0.6931,
+      "step": 526
+    },
+    {
+      "epoch": 0.30863656291100394,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00010292397660818713,
+      "loss": 0.7427,
+      "step": 528
+    },
+    {
+      "epoch": 0.30980564080081835,
+      "grad_norm": 1.234375,
+      "learning_rate": 0.00010331384015594543,
+      "loss": 0.6609,
+      "step": 530
+    },
+    {
+      "epoch": 0.31097471869063276,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001037037037037037,
+      "loss": 0.7195,
+      "step": 532
+    },
+    {
+      "epoch": 0.31214379658044716,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.000104093567251462,
+      "loss": 0.6735,
+      "step": 534
+    },
+    {
+      "epoch": 0.31331287447026157,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00010448343079922027,
+      "loss": 0.7455,
+      "step": 536
+    },
+    {
+      "epoch": 0.314481952360076,
+      "grad_norm": 0.91015625,
+      "learning_rate": 0.00010487329434697856,
+      "loss": 0.6683,
+      "step": 538
+    },
+    {
+      "epoch": 0.3156510302498904,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00010526315789473685,
+      "loss": 0.6981,
+      "step": 540
+    },
+    {
+      "epoch": 0.3168201081397048,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00010565302144249513,
+      "loss": 0.65,
+      "step": 542
+    },
+    {
+      "epoch": 0.3179891860295192,
+      "grad_norm": 2.203125,
+      "learning_rate": 0.00010604288499025342,
+      "loss": 0.7997,
+      "step": 544
+    },
+    {
+      "epoch": 0.3191582639193336,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00010643274853801171,
+      "loss": 0.7143,
+      "step": 546
+    },
+    {
+      "epoch": 0.320327341809148,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00010682261208576998,
+      "loss": 0.6619,
+      "step": 548
+    },
+    {
+      "epoch": 0.3214964196989624,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.00010721247563352828,
+      "loss": 0.6632,
+      "step": 550
+    },
+    {
+      "epoch": 0.32266549758877683,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00010760233918128655,
+      "loss": 0.6844,
+      "step": 552
+    },
+    {
+      "epoch": 0.32383457547859124,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00010799220272904485,
+      "loss": 0.698,
+      "step": 554
+    },
+    {
+      "epoch": 0.32500365336840564,
+      "grad_norm": 0.94140625,
+      "learning_rate": 0.00010838206627680312,
+      "loss": 0.6132,
+      "step": 556
+    },
+    {
+      "epoch": 0.3261727312582201,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00010877192982456141,
+      "loss": 0.6507,
+      "step": 558
+    },
+    {
+      "epoch": 0.3273418091480345,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00010916179337231968,
+      "loss": 0.696,
+      "step": 560
+    },
+    {
+      "epoch": 0.3285108870378489,
+      "grad_norm": 3.0,
+      "learning_rate": 0.00010955165692007798,
+      "loss": 0.926,
+      "step": 562
+    },
+    {
+      "epoch": 0.3296799649276633,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00010994152046783625,
+      "loss": 0.7797,
+      "step": 564
+    },
+    {
+      "epoch": 0.33084904281747773,
+      "grad_norm": 1.71875,
+      "learning_rate": 0.00011033138401559455,
+      "loss": 0.7379,
+      "step": 566
+    },
+    {
+      "epoch": 0.33201812070729214,
+      "grad_norm": 0.9921875,
+      "learning_rate": 0.00011072124756335282,
+      "loss": 0.679,
+      "step": 568
+    },
+    {
+      "epoch": 0.33318719859710655,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00011111111111111112,
+      "loss": 0.6668,
+      "step": 570
+    },
+    {
+      "epoch": 0.33435627648692096,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.0001115009746588694,
+      "loss": 0.7378,
+      "step": 572
+    },
+    {
+      "epoch": 0.33552535437673536,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.0001118908382066277,
+      "loss": 0.7341,
+      "step": 574
+    },
+    {
+      "epoch": 0.33669443226654977,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.00011228070175438597,
+      "loss": 0.6793,
+      "step": 576
+    },
+    {
+      "epoch": 0.3378635101563642,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00011267056530214426,
+      "loss": 0.6926,
+      "step": 578
+    },
+    {
+      "epoch": 0.3390325880461786,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00011306042884990253,
+      "loss": 0.6393,
+      "step": 580
+    },
+    {
+      "epoch": 0.340201665935993,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00011345029239766083,
+      "loss": 0.7207,
+      "step": 582
+    },
+    {
+      "epoch": 0.3413707438258074,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.0001138401559454191,
+      "loss": 0.6168,
+      "step": 584
+    },
+    {
+      "epoch": 0.3425398217156218,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0001142300194931774,
+      "loss": 0.712,
+      "step": 586
+    },
+    {
+      "epoch": 0.3437088996054362,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00011461988304093567,
+      "loss": 0.6532,
+      "step": 588
+    },
+    {
+      "epoch": 0.3448779774952506,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00011500974658869397,
+      "loss": 0.6766,
+      "step": 590
+    },
+    {
+      "epoch": 0.346047055385065,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00011539961013645224,
+      "loss": 0.6986,
+      "step": 592
+    },
+    {
+      "epoch": 0.34721613327487943,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00011578947368421053,
+      "loss": 0.6867,
+      "step": 594
+    },
+    {
+      "epoch": 0.34838521116469384,
+      "grad_norm": 1.484375,
+      "learning_rate": 0.0001161793372319688,
+      "loss": 0.8175,
+      "step": 596
+    },
+    {
+      "epoch": 0.34955428905450825,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001165692007797271,
+      "loss": 0.8122,
+      "step": 598
+    },
+    {
+      "epoch": 0.35072336694432266,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00011695906432748539,
+      "loss": 0.7642,
+      "step": 600
+    },
+    {
+      "epoch": 0.35189244483413706,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00011734892787524367,
+      "loss": 0.703,
+      "step": 602
+    },
+    {
+      "epoch": 0.35306152272395147,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00011773879142300195,
+      "loss": 0.7031,
+      "step": 604
+    },
+    {
+      "epoch": 0.3542306006137659,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00011812865497076025,
+      "loss": 0.686,
+      "step": 606
+    },
+    {
+      "epoch": 0.3553996785035803,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00011851851851851852,
+      "loss": 0.6708,
+      "step": 608
+    },
+    {
+      "epoch": 0.3565687563933947,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.00011890838206627682,
+      "loss": 0.8084,
+      "step": 610
+    },
+    {
+      "epoch": 0.3577378342832091,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00011929824561403509,
+      "loss": 0.6705,
+      "step": 612
+    },
+    {
+      "epoch": 0.3589069121730235,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00011968810916179338,
+      "loss": 0.6409,
+      "step": 614
+    },
+    {
+      "epoch": 0.3600759900628379,
+      "grad_norm": 0.9375,
+      "learning_rate": 0.00012007797270955165,
+      "loss": 0.6805,
+      "step": 616
+    },
+    {
+      "epoch": 0.3612450679526523,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00012046783625730995,
+      "loss": 0.7129,
+      "step": 618
+    },
+    {
+      "epoch": 0.36241414584246673,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00012085769980506822,
+      "loss": 0.6819,
+      "step": 620
+    },
+    {
+      "epoch": 0.36358322373228114,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.00012124756335282652,
+      "loss": 0.6424,
+      "step": 622
+    },
+    {
+      "epoch": 0.3647523016220956,
+      "grad_norm": 0.7890625,
+      "learning_rate": 0.00012163742690058479,
+      "loss": 0.704,
+      "step": 624
+    },
+    {
+      "epoch": 0.36592137951191,
+      "grad_norm": 1.3359375,
+      "learning_rate": 0.00012202729044834309,
+      "loss": 0.8172,
+      "step": 626
+    },
+    {
+      "epoch": 0.3670904574017244,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00012241715399610137,
+      "loss": 0.6486,
+      "step": 628
+    },
+    {
+      "epoch": 0.3682595352915388,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00012280701754385965,
+      "loss": 0.6772,
+      "step": 630
+    },
+    {
+      "epoch": 0.3694286131813532,
+      "grad_norm": 1.1171875,
+      "learning_rate": 0.00012319688109161794,
+      "loss": 0.6997,
+      "step": 632
+    },
+    {
+      "epoch": 0.37059769107116763,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00012358674463937622,
+      "loss": 0.6984,
+      "step": 634
+    },
+    {
+      "epoch": 0.37176676896098204,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0001239766081871345,
+      "loss": 0.6796,
+      "step": 636
+    },
+    {
+      "epoch": 0.37293584685079645,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.0001243664717348928,
+      "loss": 0.8194,
+      "step": 638
+    },
+    {
+      "epoch": 0.37410492474061086,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00012475633528265107,
+      "loss": 0.7189,
+      "step": 640
+    },
+    {
+      "epoch": 0.37527400263042526,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00012514619883040936,
+      "loss": 0.6934,
+      "step": 642
+    },
+    {
+      "epoch": 0.37644308052023967,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00012553606237816764,
+      "loss": 0.6885,
+      "step": 644
+    },
+    {
+      "epoch": 0.3776121584100541,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00012592592592592592,
+      "loss": 0.671,
+      "step": 646
+    },
+    {
+      "epoch": 0.3787812362998685,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0001263157894736842,
+      "loss": 0.6515,
+      "step": 648
+    },
+    {
+      "epoch": 0.3799503141896829,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00012670565302144252,
+      "loss": 0.6833,
+      "step": 650
+    },
+    {
+      "epoch": 0.3811193920794973,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00012709551656920077,
+      "loss": 0.6363,
+      "step": 652
+    },
+    {
+      "epoch": 0.3822884699693117,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00012748538011695908,
+      "loss": 0.6848,
+      "step": 654
+    },
+    {
+      "epoch": 0.3834575478591261,
+      "grad_norm": 0.93359375,
+      "learning_rate": 0.00012787524366471734,
+      "loss": 0.7383,
+      "step": 656
+    },
+    {
+      "epoch": 0.3846266257489405,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00012826510721247565,
+      "loss": 0.7057,
+      "step": 658
+    },
+    {
+      "epoch": 0.3857957036387549,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0001286549707602339,
+      "loss": 0.7039,
+      "step": 660
+    },
+    {
+      "epoch": 0.38696478152856933,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00012904483430799222,
+      "loss": 0.727,
+      "step": 662
+    },
+    {
+      "epoch": 0.38813385941838374,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00012943469785575048,
+      "loss": 0.7128,
+      "step": 664
+    },
+    {
+      "epoch": 0.38930293730819815,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.0001298245614035088,
+      "loss": 0.6446,
+      "step": 666
+    },
+    {
+      "epoch": 0.39047201519801256,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00013021442495126704,
+      "loss": 0.6434,
+      "step": 668
+    },
+    {
+      "epoch": 0.39164109308782696,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00013060428849902535,
+      "loss": 0.642,
+      "step": 670
+    },
+    {
+      "epoch": 0.39281017097764137,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00013099415204678364,
+      "loss": 0.6805,
+      "step": 672
+    },
+    {
+      "epoch": 0.3939792488674558,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00013138401559454192,
+      "loss": 0.6923,
+      "step": 674
+    },
+    {
+      "epoch": 0.3951483267572702,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.0001317738791423002,
+      "loss": 0.7944,
+      "step": 676
+    },
+    {
+      "epoch": 0.3963174046470846,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0001321637426900585,
+      "loss": 0.6876,
+      "step": 678
+    },
+    {
+      "epoch": 0.397486482536899,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00013255360623781677,
+      "loss": 0.6857,
+      "step": 680
+    },
+    {
+      "epoch": 0.3986555604267134,
+      "grad_norm": 0.83984375,
+      "learning_rate": 0.00013294346978557506,
+      "loss": 0.6796,
+      "step": 682
+    },
+    {
+      "epoch": 0.3998246383165278,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.6604,
+      "step": 684
+    },
+    {
+      "epoch": 0.4009937162063422,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00013372319688109162,
+      "loss": 0.6999,
+      "step": 686
+    },
+    {
+      "epoch": 0.4021627940961567,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.0001341130604288499,
+      "loss": 0.663,
+      "step": 688
+    },
+    {
+      "epoch": 0.4033318719859711,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.0001345029239766082,
+      "loss": 0.6553,
+      "step": 690
+    },
+    {
+      "epoch": 0.4045009498757855,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00013489278752436647,
+      "loss": 0.6892,
+      "step": 692
+    },
+    {
+      "epoch": 0.4056700277655999,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00013528265107212476,
+      "loss": 0.6529,
+      "step": 694
+    },
+    {
+      "epoch": 0.4068391056554143,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00013567251461988304,
+      "loss": 0.6728,
+      "step": 696
+    },
+    {
+      "epoch": 0.4080081835452287,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00013606237816764133,
+      "loss": 0.6847,
+      "step": 698
+    },
+    {
+      "epoch": 0.4091772614350431,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001364522417153996,
+      "loss": 0.7475,
+      "step": 700
+    },
+    {
+      "epoch": 0.41034633932485753,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.0001368421052631579,
+      "loss": 0.6805,
+      "step": 702
+    },
+    {
+      "epoch": 0.41151541721467194,
+      "grad_norm": 0.875,
+      "learning_rate": 0.00013723196881091618,
+      "loss": 0.8327,
+      "step": 704
+    },
+    {
+      "epoch": 0.41268449510448635,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.0001376218323586745,
+      "loss": 0.663,
+      "step": 706
+    },
+    {
+      "epoch": 0.41385357299430076,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00013801169590643274,
+      "loss": 0.6675,
+      "step": 708
+    },
+    {
+      "epoch": 0.41502265088411516,
+      "grad_norm": 0.95703125,
+      "learning_rate": 0.00013840155945419105,
+      "loss": 0.7458,
+      "step": 710
+    },
+    {
+      "epoch": 0.41619172877392957,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.0001387914230019493,
+      "loss": 0.6385,
+      "step": 712
+    },
+    {
+      "epoch": 0.417360806663744,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00013918128654970762,
+      "loss": 0.7979,
+      "step": 714
+    },
+    {
+      "epoch": 0.4185298845535584,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00013957115009746588,
+      "loss": 0.6949,
+      "step": 716
+    },
+    {
+      "epoch": 0.4196989624433728,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001399610136452242,
+      "loss": 0.6713,
+      "step": 718
+    },
+    {
+      "epoch": 0.4208680403331872,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00014035087719298245,
+      "loss": 0.6503,
+      "step": 720
+    },
+    {
+      "epoch": 0.4220371182230016,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00014074074074074076,
+      "loss": 0.6641,
+      "step": 722
+    },
+    {
+      "epoch": 0.423206196112816,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.000141130604288499,
+      "loss": 0.7159,
+      "step": 724
+    },
+    {
+      "epoch": 0.4243752740026304,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00014152046783625732,
+      "loss": 0.786,
+      "step": 726
+    },
+    {
+      "epoch": 0.4255443518924448,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00014191033138401558,
+      "loss": 0.6511,
+      "step": 728
+    },
+    {
+      "epoch": 0.42671342978225923,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.0001423001949317739,
+      "loss": 0.8375,
+      "step": 730
+    },
+    {
+      "epoch": 0.42788250767207364,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.00014269005847953217,
+      "loss": 0.7475,
+      "step": 732
+    },
+    {
+      "epoch": 0.42905158556188805,
+      "grad_norm": 1.3046875,
+      "learning_rate": 0.00014307992202729046,
+      "loss": 0.6989,
+      "step": 734
+    },
+    {
+      "epoch": 0.43022066345170246,
+      "grad_norm": 0.9453125,
+      "learning_rate": 0.00014346978557504874,
+      "loss": 0.6792,
+      "step": 736
+    },
+    {
+      "epoch": 0.43138974134151686,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00014385964912280703,
+      "loss": 0.7084,
+      "step": 738
+    },
+    {
+      "epoch": 0.43255881923133127,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.0001442495126705653,
+      "loss": 0.7263,
+      "step": 740
+    },
+    {
+      "epoch": 0.4337278971211457,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.0001446393762183236,
+      "loss": 0.706,
+      "step": 742
+    },
+    {
+      "epoch": 0.4348969750109601,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00014502923976608188,
+      "loss": 0.6989,
+      "step": 744
+    },
+    {
+      "epoch": 0.4360660529007745,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00014541910331384016,
+      "loss": 0.7097,
+      "step": 746
+    },
+    {
+      "epoch": 0.4372351307905889,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00014580896686159844,
+      "loss": 0.6767,
+      "step": 748
+    },
+    {
+      "epoch": 0.4384042086804033,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00014619883040935673,
+      "loss": 0.7045,
+      "step": 750
+    },
+    {
+      "epoch": 0.4395732865702177,
+      "grad_norm": 0.828125,
+      "learning_rate": 0.000146588693957115,
+      "loss": 0.6492,
+      "step": 752
+    },
+    {
+      "epoch": 0.4407423644600322,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001469785575048733,
+      "loss": 0.6776,
+      "step": 754
+    },
+    {
+      "epoch": 0.4419114423498466,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00014736842105263158,
+      "loss": 0.8008,
+      "step": 756
+    },
+    {
+      "epoch": 0.443080520239661,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00014775828460038986,
+      "loss": 0.6677,
+      "step": 758
+    },
+    {
+      "epoch": 0.4442495981294754,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00014814814814814815,
+      "loss": 0.6799,
+      "step": 760
+    },
+    {
+      "epoch": 0.4454186760192898,
+      "grad_norm": 0.796875,
+      "learning_rate": 0.00014853801169590643,
+      "loss": 0.6626,
+      "step": 762
+    },
+    {
+      "epoch": 0.4465877539091042,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001489278752436647,
+      "loss": 0.6814,
+      "step": 764
+    },
+    {
+      "epoch": 0.4477568317989186,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00014931773879142302,
+      "loss": 0.7181,
+      "step": 766
+    },
+    {
+      "epoch": 0.448925909688733,
+      "grad_norm": 0.8046875,
+      "learning_rate": 0.00014970760233918128,
+      "loss": 0.681,
+      "step": 768
+    },
+    {
+      "epoch": 0.45009498757854743,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.0001500974658869396,
+      "loss": 0.6625,
+      "step": 770
+    },
+    {
+      "epoch": 0.45126406546836184,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00015048732943469785,
+      "loss": 0.6558,
+      "step": 772
+    },
+    {
+      "epoch": 0.45243314335817625,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00015087719298245616,
+      "loss": 0.6968,
+      "step": 774
+    },
+    {
+      "epoch": 0.45360222124799066,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00015126705653021442,
+      "loss": 0.7073,
+      "step": 776
+    },
+    {
+      "epoch": 0.45477129913780506,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00015165692007797273,
+      "loss": 0.7151,
+      "step": 778
+    },
+    {
+      "epoch": 0.45594037702761947,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00015204678362573098,
+      "loss": 0.7192,
+      "step": 780
+    },
+    {
+      "epoch": 0.4571094549174339,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0001524366471734893,
+      "loss": 0.6937,
+      "step": 782
+    },
+    {
+      "epoch": 0.4582785328072483,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00015282651072124755,
+      "loss": 0.6859,
+      "step": 784
+    },
+    {
+      "epoch": 0.4594476106970627,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00015321637426900586,
+      "loss": 0.6491,
+      "step": 786
+    },
+    {
+      "epoch": 0.4606166885868771,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00015360623781676414,
+      "loss": 0.719,
+      "step": 788
+    },
+    {
+      "epoch": 0.4617857664766915,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00015399610136452243,
+      "loss": 0.6906,
+      "step": 790
+    },
+    {
+      "epoch": 0.4629548443665059,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0001543859649122807,
+      "loss": 0.6957,
+      "step": 792
+    },
+    {
+      "epoch": 0.4641239222563203,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.000154775828460039,
+      "loss": 0.6863,
+      "step": 794
+    },
+    {
+      "epoch": 0.4652930001461347,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00015516569200779728,
+      "loss": 0.6845,
+      "step": 796
+    },
+    {
+      "epoch": 0.46646207803594913,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00015555555555555556,
+      "loss": 0.6393,
+      "step": 798
+    },
+    {
+      "epoch": 0.46763115592576354,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00015594541910331385,
+      "loss": 0.6739,
+      "step": 800
+    },
+    {
+      "epoch": 0.46880023381557795,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00015633528265107213,
+      "loss": 0.6919,
+      "step": 802
+    },
+    {
+      "epoch": 0.46996931170539236,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00015672514619883041,
+      "loss": 0.6885,
+      "step": 804
+    },
+    {
+      "epoch": 0.47113838959520676,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0001571150097465887,
+      "loss": 0.7075,
+      "step": 806
+    },
+    {
+      "epoch": 0.47230746748502117,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00015750487329434698,
+      "loss": 0.6716,
+      "step": 808
+    },
+    {
+      "epoch": 0.4734765453748356,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00015789473684210527,
+      "loss": 0.6595,
+      "step": 810
+    },
+    {
+      "epoch": 0.47464562326465,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00015828460038986355,
+      "loss": 0.6272,
+      "step": 812
+    },
+    {
+      "epoch": 0.4758147011544644,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00015867446393762183,
+      "loss": 0.6849,
+      "step": 814
+    },
+    {
+      "epoch": 0.4769837790442788,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00015906432748538012,
+      "loss": 0.6523,
+      "step": 816
+    },
+    {
+      "epoch": 0.4781528569340932,
+      "grad_norm": 1.40625,
+      "learning_rate": 0.0001594541910331384,
+      "loss": 0.7112,
+      "step": 818
+    },
+    {
+      "epoch": 0.47932193482390767,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00015984405458089668,
+      "loss": 0.7073,
+      "step": 820
+    },
+    {
+      "epoch": 0.4804910127137221,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.000160233918128655,
+      "loss": 0.6436,
+      "step": 822
+    },
+    {
+      "epoch": 0.4816600906035365,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00016062378167641325,
+      "loss": 0.669,
+      "step": 824
+    },
+    {
+      "epoch": 0.4828291684933509,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00016101364522417156,
+      "loss": 0.8224,
+      "step": 826
+    },
+    {
+      "epoch": 0.4839982463831653,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00016140350877192982,
+      "loss": 0.7526,
+      "step": 828
+    },
+    {
+      "epoch": 0.4851673242729797,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00016179337231968813,
+      "loss": 0.6986,
+      "step": 830
+    },
+    {
+      "epoch": 0.4863364021627941,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00016218323586744639,
+      "loss": 0.6496,
+      "step": 832
+    },
+    {
+      "epoch": 0.4875054800526085,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001625730994152047,
+      "loss": 0.7165,
+      "step": 834
+    },
+    {
+      "epoch": 0.4886745579424229,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00016296296296296295,
+      "loss": 0.6453,
+      "step": 836
+    },
+    {
+      "epoch": 0.48984363583223733,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00016335282651072126,
+      "loss": 0.617,
+      "step": 838
+    },
+    {
+      "epoch": 0.49101271372205174,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00016374269005847952,
+      "loss": 0.6841,
+      "step": 840
+    },
+    {
+      "epoch": 0.49218179161186615,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00016413255360623783,
+      "loss": 0.6668,
+      "step": 842
+    },
+    {
+      "epoch": 0.49335086950168056,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0001645224171539961,
+      "loss": 0.7092,
+      "step": 844
+    },
+    {
+      "epoch": 0.49451994739149496,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.0001649122807017544,
+      "loss": 0.6479,
+      "step": 846
+    },
+    {
+      "epoch": 0.49568902528130937,
+      "grad_norm": 0.81640625,
+      "learning_rate": 0.00016530214424951268,
+      "loss": 0.6687,
+      "step": 848
+    },
+    {
+      "epoch": 0.4968581031711238,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00016569200779727097,
+      "loss": 0.6704,
+      "step": 850
+    },
+    {
+      "epoch": 0.4980271810609382,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00016608187134502925,
+      "loss": 0.6678,
+      "step": 852
+    },
+    {
+      "epoch": 0.4991962589507526,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00016647173489278753,
+      "loss": 0.6744,
+      "step": 854
+    },
+    {
+      "epoch": 0.500365336840567,
+      "grad_norm": 0.79296875,
+      "learning_rate": 0.00016686159844054582,
+      "loss": 0.6565,
+      "step": 856
+    },
+    {
+      "epoch": 0.5015344147303814,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0001672514619883041,
+      "loss": 0.6499,
+      "step": 858
+    },
+    {
+      "epoch": 0.5027034926201959,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00016764132553606238,
+      "loss": 0.6619,
+      "step": 860
+    },
+    {
+      "epoch": 0.5038725705100102,
+      "grad_norm": 1.703125,
+      "learning_rate": 0.00016803118908382067,
+      "loss": 0.8945,
+      "step": 862
+    },
+    {
+      "epoch": 0.5050416483998247,
+      "grad_norm": 0.9296875,
+      "learning_rate": 0.00016842105263157895,
+      "loss": 0.7597,
+      "step": 864
+    },
+    {
+      "epoch": 0.506210726289639,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00016881091617933723,
+      "loss": 0.6698,
+      "step": 866
+    },
+    {
+      "epoch": 0.5073798041794535,
+      "grad_norm": 1.1953125,
+      "learning_rate": 0.00016920077972709552,
+      "loss": 0.7135,
+      "step": 868
+    },
+    {
+      "epoch": 0.5085488820692678,
+      "grad_norm": 1.0859375,
+      "learning_rate": 0.0001695906432748538,
+      "loss": 0.8298,
+      "step": 870
+    },
+    {
+      "epoch": 0.5097179599590823,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00016998050682261209,
+      "loss": 0.6749,
+      "step": 872
+    },
+    {
+      "epoch": 0.5108870378488967,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00017037037037037037,
+      "loss": 0.8318,
+      "step": 874
+    },
+    {
+      "epoch": 0.5120561157387111,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.00017076023391812865,
+      "loss": 0.7329,
+      "step": 876
+    },
+    {
+      "epoch": 0.5132251936285255,
+      "grad_norm": 0.87109375,
+      "learning_rate": 0.00017115009746588694,
+      "loss": 0.7127,
+      "step": 878
+    },
+    {
+      "epoch": 0.5143942715183399,
+      "grad_norm": 9.375,
+      "learning_rate": 0.00017153996101364522,
+      "loss": 0.7969,
+      "step": 880
+    },
+    {
+      "epoch": 0.5155633494081543,
+      "grad_norm": 2.015625,
+      "learning_rate": 0.00017192982456140353,
+      "loss": 0.8313,
+      "step": 882
+    },
+    {
+      "epoch": 0.5167324272979688,
+      "grad_norm": 0.71875,
+      "learning_rate": 0.0001723196881091618,
+      "loss": 0.7693,
+      "step": 884
+    },
+    {
+      "epoch": 0.5179015051877831,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0001727095516569201,
+      "loss": 0.7127,
+      "step": 886
+    },
+    {
+      "epoch": 0.5190705830775976,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00017309941520467836,
+      "loss": 0.7283,
+      "step": 888
+    },
+    {
+      "epoch": 0.5202396609674119,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00017348927875243667,
+      "loss": 0.6591,
+      "step": 890
+    },
+    {
+      "epoch": 0.5214087388572264,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00017387914230019492,
+      "loss": 0.6794,
+      "step": 892
+    },
+    {
+      "epoch": 0.5225778167470407,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00017426900584795323,
+      "loss": 0.6299,
+      "step": 894
+    },
+    {
+      "epoch": 0.5237468946368552,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.0001746588693957115,
+      "loss": 0.73,
+      "step": 896
+    },
+    {
+      "epoch": 0.5249159725266695,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0001750487329434698,
+      "loss": 0.6538,
+      "step": 898
+    },
+    {
+      "epoch": 0.526085050416484,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00017543859649122806,
+      "loss": 0.7013,
+      "step": 900
+    },
+    {
+      "epoch": 0.5272541283062984,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00017582846003898637,
+      "loss": 0.6633,
+      "step": 902
+    },
+    {
+      "epoch": 0.5284232061961128,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00017621832358674462,
+      "loss": 0.6638,
+      "step": 904
+    },
+    {
+      "epoch": 0.5295922840859272,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00017660818713450294,
+      "loss": 0.6616,
+      "step": 906
+    },
+    {
+      "epoch": 0.5307613619757416,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00017699805068226122,
+      "loss": 0.6935,
+      "step": 908
+    },
+    {
+      "epoch": 0.531930439865556,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0001773879142300195,
+      "loss": 0.6795,
+      "step": 910
+    },
+    {
+      "epoch": 0.5330995177553705,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00017777777777777779,
+      "loss": 0.6797,
+      "step": 912
+    },
+    {
+      "epoch": 0.5342685956451848,
+      "grad_norm": 0.75,
+      "learning_rate": 0.00017816764132553607,
+      "loss": 0.6583,
+      "step": 914
+    },
+    {
+      "epoch": 0.5354376735349993,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00017855750487329435,
+      "loss": 0.6592,
+      "step": 916
+    },
+    {
+      "epoch": 0.5366067514248137,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00017894736842105264,
+      "loss": 0.7534,
+      "step": 918
+    },
+    {
+      "epoch": 0.5377758293146281,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00017933723196881092,
+      "loss": 0.699,
+      "step": 920
+    },
+    {
+      "epoch": 0.5389449072044425,
+      "grad_norm": 0.7109375,
+      "learning_rate": 0.0001797270955165692,
+      "loss": 0.6759,
+      "step": 922
+    },
+    {
+      "epoch": 0.5401139850942569,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001801169590643275,
+      "loss": 0.6676,
+      "step": 924
+    },
+    {
+      "epoch": 0.5412830629840714,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00018050682261208577,
+      "loss": 0.6734,
+      "step": 926
+    },
+    {
+      "epoch": 0.5424521408738857,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00018089668615984406,
+      "loss": 0.6545,
+      "step": 928
+    },
+    {
+      "epoch": 0.5436212187637002,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00018128654970760234,
+      "loss": 0.6863,
+      "step": 930
+    },
+    {
+      "epoch": 0.5447902966535145,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00018167641325536062,
+      "loss": 0.6536,
+      "step": 932
+    },
+    {
+      "epoch": 0.545959374543329,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0001820662768031189,
+      "loss": 0.6825,
+      "step": 934
+    },
+    {
+      "epoch": 0.5471284524331433,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.0001824561403508772,
+      "loss": 0.7075,
+      "step": 936
+    },
+    {
+      "epoch": 0.5482975303229578,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00018284600389863547,
+      "loss": 0.8103,
+      "step": 938
+    },
+    {
+      "epoch": 0.5494666082127722,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00018323586744639376,
+      "loss": 0.7081,
+      "step": 940
+    },
+    {
+      "epoch": 0.5506356861025866,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00018362573099415207,
+      "loss": 0.6647,
+      "step": 942
+    },
+    {
+      "epoch": 0.551804763992401,
+      "grad_norm": 0.98828125,
+      "learning_rate": 0.00018401559454191032,
+      "loss": 0.7202,
+      "step": 944
+    },
+    {
+      "epoch": 0.5529738418822154,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00018440545808966864,
+      "loss": 0.7734,
+      "step": 946
+    },
+    {
+      "epoch": 0.5541429197720298,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0001847953216374269,
+      "loss": 0.6381,
+      "step": 948
+    },
+    {
+      "epoch": 0.5553119976618442,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0001851851851851852,
+      "loss": 0.7235,
+      "step": 950
+    },
+    {
+      "epoch": 0.5564810755516586,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00018557504873294346,
+      "loss": 0.8296,
+      "step": 952
+    },
+    {
+      "epoch": 0.5576501534414731,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00018596491228070177,
+      "loss": 0.6901,
+      "step": 954
+    },
+    {
+      "epoch": 0.5588192313312874,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00018635477582846003,
+      "loss": 0.6628,
+      "step": 956
+    },
+    {
+      "epoch": 0.5599883092211019,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00018674463937621834,
+      "loss": 0.6689,
+      "step": 958
+    },
+    {
+      "epoch": 0.5611573871109162,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.0001871345029239766,
+      "loss": 0.7377,
+      "step": 960
+    },
+    {
+      "epoch": 0.5623264650007307,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001875243664717349,
+      "loss": 0.7217,
+      "step": 962
+    },
+    {
+      "epoch": 0.563495542890545,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0001879142300194932,
+      "loss": 0.6776,
+      "step": 964
+    },
+    {
+      "epoch": 0.5646646207803595,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00018830409356725147,
+      "loss": 0.694,
+      "step": 966
+    },
+    {
+      "epoch": 0.5658336986701739,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00018869395711500976,
+      "loss": 0.7305,
+      "step": 968
+    },
+    {
+      "epoch": 0.5670027765599883,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00018908382066276804,
+      "loss": 0.6659,
+      "step": 970
+    },
+    {
+      "epoch": 0.5681718544498027,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00018947368421052632,
+      "loss": 0.705,
+      "step": 972
+    },
+    {
+      "epoch": 0.5693409323396171,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.0001898635477582846,
+      "loss": 0.8348,
+      "step": 974
+    },
+    {
+      "epoch": 0.5705100102294315,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.0001902534113060429,
+      "loss": 0.7383,
+      "step": 976
+    },
+    {
+      "epoch": 0.571679088119246,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00019064327485380117,
+      "loss": 0.7039,
+      "step": 978
+    },
+    {
+      "epoch": 0.5728481660090603,
+      "grad_norm": 0.69921875,
+      "learning_rate": 0.00019103313840155946,
+      "loss": 0.7163,
+      "step": 980
+    },
+    {
+      "epoch": 0.5740172438988748,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019142300194931774,
+      "loss": 0.6648,
+      "step": 982
+    },
+    {
+      "epoch": 0.5751863217886892,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00019181286549707603,
+      "loss": 0.6896,
+      "step": 984
+    },
+    {
+      "epoch": 0.5763553996785036,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0001922027290448343,
+      "loss": 0.7127,
+      "step": 986
+    },
+    {
+      "epoch": 0.577524477568318,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001925925925925926,
+      "loss": 0.6549,
+      "step": 988
+    },
+    {
+      "epoch": 0.5786935554581324,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00019298245614035088,
+      "loss": 0.6581,
+      "step": 990
+    },
+    {
+      "epoch": 0.5798626333479469,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00019337231968810916,
+      "loss": 0.6802,
+      "step": 992
+    },
+    {
+      "epoch": 0.5810317112377612,
+      "grad_norm": 1.03125,
+      "learning_rate": 0.00019376218323586744,
+      "loss": 0.6906,
+      "step": 994
+    },
+    {
+      "epoch": 0.5822007891275757,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019415204678362573,
+      "loss": 0.7383,
+      "step": 996
+    },
+    {
+      "epoch": 0.58336986701739,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00019454191033138404,
+      "loss": 0.7141,
+      "step": 998
+    },
+    {
+      "epoch": 0.5845389449072045,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001949317738791423,
+      "loss": 0.6729,
+      "step": 1000
+    },
+    {
+      "epoch": 0.5857080227970188,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.0001953216374269006,
+      "loss": 0.6561,
+      "step": 1002
+    },
+    {
+      "epoch": 0.5868771006868333,
+      "grad_norm": 1.109375,
+      "learning_rate": 0.00019571150097465886,
+      "loss": 0.9202,
+      "step": 1004
+    },
+    {
+      "epoch": 0.5880461785766476,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019610136452241717,
+      "loss": 0.7142,
+      "step": 1006
+    },
+    {
+      "epoch": 0.5892152564664621,
+      "grad_norm": 105.0,
+      "learning_rate": 0.00019649122807017543,
+      "loss": 1.1271,
+      "step": 1008
+    },
+    {
+      "epoch": 0.5903843343562765,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00019688109161793374,
+      "loss": 1.1422,
+      "step": 1010
+    },
+    {
+      "epoch": 0.5915534122460909,
+      "grad_norm": 1.0390625,
+      "learning_rate": 0.000197270955165692,
+      "loss": 0.7422,
+      "step": 1012
+    },
+    {
+      "epoch": 0.5927224901359053,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.0001976608187134503,
+      "loss": 0.7443,
+      "step": 1014
+    },
+    {
+      "epoch": 0.5938915680257197,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00019805068226120856,
+      "loss": 0.7227,
+      "step": 1016
+    },
+    {
+      "epoch": 0.5950606459155341,
+      "grad_norm": 1.28125,
+      "learning_rate": 0.00019844054580896687,
+      "loss": 0.7005,
+      "step": 1018
+    },
+    {
+      "epoch": 0.5962297238053486,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.00019883040935672513,
+      "loss": 0.6514,
+      "step": 1020
+    },
+    {
+      "epoch": 0.5973988016951629,
+      "grad_norm": 0.96875,
+      "learning_rate": 0.00019922027290448344,
+      "loss": 0.7211,
+      "step": 1022
+    },
+    {
+      "epoch": 0.5985678795849774,
+      "grad_norm": 0.87890625,
+      "learning_rate": 0.00019961013645224173,
+      "loss": 0.6627,
+      "step": 1024
+    },
+    {
+      "epoch": 0.5997369574747917,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0002,
+      "loss": 0.6844,
+      "step": 1026
+    },
+    {
+      "epoch": 0.6009060353646062,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00019990253411306043,
+      "loss": 0.7113,
+      "step": 1028
+    },
+    {
+      "epoch": 0.6020751132544205,
+      "grad_norm": 12.5,
+      "learning_rate": 0.00019980506822612085,
+      "loss": 0.751,
+      "step": 1030
+    },
+    {
+      "epoch": 0.603244191144235,
+      "grad_norm": 1.9296875,
+      "learning_rate": 0.00019970760233918128,
+      "loss": 0.7035,
+      "step": 1032
+    },
+    {
+      "epoch": 0.6044132690340493,
+      "grad_norm": 3.96875,
+      "learning_rate": 0.00019961013645224173,
+      "loss": 0.8057,
+      "step": 1034
+    },
+    {
+      "epoch": 0.6055823469238638,
+      "grad_norm": 2.125,
+      "learning_rate": 0.00019951267056530218,
+      "loss": 1.0457,
+      "step": 1036
+    },
+    {
+      "epoch": 0.6067514248136782,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0001994152046783626,
+      "loss": 0.7618,
+      "step": 1038
+    },
+    {
+      "epoch": 0.6079205027034926,
+      "grad_norm": 1.3515625,
+      "learning_rate": 0.00019931773879142302,
+      "loss": 0.8979,
+      "step": 1040
+    },
+    {
+      "epoch": 0.609089580593307,
+      "grad_norm": 1.1484375,
+      "learning_rate": 0.00019922027290448344,
+      "loss": 0.7494,
+      "step": 1042
+    },
+    {
+      "epoch": 0.6102586584831214,
+      "grad_norm": 1.7109375,
+      "learning_rate": 0.00019912280701754386,
+      "loss": 0.7505,
+      "step": 1044
+    },
+    {
+      "epoch": 0.6114277363729358,
+      "grad_norm": 1.2578125,
+      "learning_rate": 0.0001990253411306043,
+      "loss": 0.7356,
+      "step": 1046
+    },
+    {
+      "epoch": 0.6125968142627503,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0001989278752436647,
+      "loss": 0.7106,
+      "step": 1048
+    },
+    {
+      "epoch": 0.6137658921525647,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019883040935672513,
+      "loss": 0.7151,
+      "step": 1050
+    },
+    {
+      "epoch": 0.6149349700423791,
+      "grad_norm": 2.609375,
+      "learning_rate": 0.00019873294346978558,
+      "loss": 0.7294,
+      "step": 1052
+    },
+    {
+      "epoch": 0.6161040479321935,
+      "grad_norm": 0.91796875,
+      "learning_rate": 0.00019863547758284603,
+      "loss": 0.7628,
+      "step": 1054
+    },
+    {
+      "epoch": 0.6172731258220079,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00019853801169590645,
+      "loss": 0.6701,
+      "step": 1056
+    },
+    {
+      "epoch": 0.6184422037118223,
+      "grad_norm": 11.5,
+      "learning_rate": 0.00019844054580896687,
+      "loss": 0.8591,
+      "step": 1058
+    },
+    {
+      "epoch": 0.6196112816016367,
+      "grad_norm": 6.125,
+      "learning_rate": 0.0001983430799220273,
+      "loss": 0.7596,
+      "step": 1060
+    },
+    {
+      "epoch": 0.6207803594914512,
+      "grad_norm": 10.625,
+      "learning_rate": 0.00019824561403508772,
+      "loss": 0.7751,
+      "step": 1062
+    },
+    {
+      "epoch": 0.6219494373812655,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00019814814814814814,
+      "loss": 0.8455,
+      "step": 1064
+    },
+    {
+      "epoch": 0.62311851527108,
+      "grad_norm": 0.88671875,
+      "learning_rate": 0.00019805068226120856,
+      "loss": 0.7299,
+      "step": 1066
+    },
+    {
+      "epoch": 0.6242875931608943,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.000197953216374269,
+      "loss": 0.7077,
+      "step": 1068
+    },
+    {
+      "epoch": 0.6254566710507088,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00019785575048732946,
+      "loss": 0.6911,
+      "step": 1070
+    },
+    {
+      "epoch": 0.6266257489405231,
+      "grad_norm": 10.375,
+      "learning_rate": 0.00019775828460038989,
+      "loss": 0.6806,
+      "step": 1072
+    },
+    {
+      "epoch": 0.6277948268303376,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.0001976608187134503,
+      "loss": 0.7528,
+      "step": 1074
+    },
+    {
+      "epoch": 0.628963904720152,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00019756335282651073,
+      "loss": 0.6915,
+      "step": 1076
+    },
+    {
+      "epoch": 0.6301329826099664,
+      "grad_norm": 4.125,
+      "learning_rate": 0.00019746588693957115,
+      "loss": 0.7393,
+      "step": 1078
+    },
+    {
+      "epoch": 0.6313020604997808,
+      "grad_norm": 1.8515625,
+      "learning_rate": 0.00019736842105263157,
+      "loss": 0.698,
+      "step": 1080
+    },
+    {
+      "epoch": 0.6324711383895952,
+      "grad_norm": 1.0546875,
+      "learning_rate": 0.000197270955165692,
+      "loss": 0.701,
+      "step": 1082
+    },
+    {
+      "epoch": 0.6336402162794096,
+      "grad_norm": 2.15625,
+      "learning_rate": 0.00019717348927875245,
+      "loss": 0.6641,
+      "step": 1084
+    },
+    {
+      "epoch": 0.634809294169224,
+      "grad_norm": 2.4375,
+      "learning_rate": 0.00019707602339181287,
+      "loss": 0.7128,
+      "step": 1086
+    },
+    {
+      "epoch": 0.6359783720590384,
+      "grad_norm": 1.265625,
+      "learning_rate": 0.00019697855750487332,
+      "loss": 0.8843,
+      "step": 1088
+    },
+    {
+      "epoch": 0.6371474499488529,
+      "grad_norm": 5.5625,
+      "learning_rate": 0.00019688109161793374,
+      "loss": 0.7482,
+      "step": 1090
+    },
+    {
+      "epoch": 0.6383165278386672,
+      "grad_norm": 1.4765625,
+      "learning_rate": 0.00019678362573099416,
+      "loss": 0.8855,
+      "step": 1092
+    },
+    {
+      "epoch": 0.6394856057284817,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00019668615984405458,
+      "loss": 0.7744,
+      "step": 1094
+    },
+    {
+      "epoch": 0.640654683618296,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.000196588693957115,
+      "loss": 0.72,
+      "step": 1096
+    },
+    {
+      "epoch": 0.6418237615081105,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00019649122807017543,
+      "loss": 0.6577,
+      "step": 1098
+    },
+    {
+      "epoch": 0.6429928393979248,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00019639376218323588,
+      "loss": 0.736,
+      "step": 1100
+    },
+    {
+      "epoch": 0.6441619172877393,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0001962962962962963,
+      "loss": 0.7025,
+      "step": 1102
+    },
+    {
+      "epoch": 0.6453309951775537,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019619883040935675,
+      "loss": 0.7076,
+      "step": 1104
+    },
+    {
+      "epoch": 0.6465000730673681,
+      "grad_norm": 1.609375,
+      "learning_rate": 0.00019610136452241717,
+      "loss": 0.7121,
+      "step": 1106
+    },
+    {
+      "epoch": 0.6476691509571825,
+      "grad_norm": 1.1015625,
+      "learning_rate": 0.0001960038986354776,
+      "loss": 0.705,
+      "step": 1108
+    },
+    {
+      "epoch": 0.6488382288469969,
+      "grad_norm": 1.734375,
+      "learning_rate": 0.00019590643274853802,
+      "loss": 0.8037,
+      "step": 1110
+    },
+    {
+      "epoch": 0.6500073067368113,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00019580896686159844,
+      "loss": 0.7236,
+      "step": 1112
+    },
+    {
+      "epoch": 0.6511763846266257,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019571150097465886,
+      "loss": 0.764,
+      "step": 1114
+    },
+    {
+      "epoch": 0.6523454625164402,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001956140350877193,
+      "loss": 0.6921,
+      "step": 1116
+    },
+    {
+      "epoch": 0.6535145404062546,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019551656920077973,
+      "loss": 0.7026,
+      "step": 1118
+    },
+    {
+      "epoch": 0.654683618296069,
+      "grad_norm": 0.921875,
+      "learning_rate": 0.00019541910331384016,
+      "loss": 0.6641,
+      "step": 1120
+    },
+    {
+      "epoch": 0.6558526961858834,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001953216374269006,
+      "loss": 0.7031,
+      "step": 1122
+    },
+    {
+      "epoch": 0.6570217740756978,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00019522417153996103,
+      "loss": 0.6727,
+      "step": 1124
+    },
+    {
+      "epoch": 0.6581908519655122,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00019512670565302145,
+      "loss": 0.6963,
+      "step": 1126
+    },
+    {
+      "epoch": 0.6593599298553267,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00019502923976608187,
+      "loss": 0.665,
+      "step": 1128
+    },
+    {
+      "epoch": 0.660529007745141,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001949317738791423,
+      "loss": 0.6752,
+      "step": 1130
+    },
+    {
+      "epoch": 0.6616980856349555,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00019483430799220274,
+      "loss": 0.6616,
+      "step": 1132
+    },
+    {
+      "epoch": 0.6628671635247698,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00019473684210526317,
+      "loss": 0.6426,
+      "step": 1134
+    },
+    {
+      "epoch": 0.6640362414145843,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001946393762183236,
+      "loss": 0.675,
+      "step": 1136
+    },
+    {
+      "epoch": 0.6652053193043986,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00019454191033138404,
+      "loss": 0.732,
+      "step": 1138
+    },
+    {
+      "epoch": 0.6663743971942131,
+      "grad_norm": 0.703125,
+      "learning_rate": 0.00019444444444444446,
+      "loss": 0.6701,
+      "step": 1140
+    },
+    {
+      "epoch": 0.6675434750840274,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00019434697855750488,
+      "loss": 0.6637,
+      "step": 1142
+    },
+    {
+      "epoch": 0.6687125529738419,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0001942495126705653,
+      "loss": 0.8085,
+      "step": 1144
+    },
+    {
+      "epoch": 0.6698816308636563,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00019415204678362573,
+      "loss": 0.6828,
+      "step": 1146
+    },
+    {
+      "epoch": 0.6710507087534707,
+      "grad_norm": 0.92578125,
+      "learning_rate": 0.00019405458089668618,
+      "loss": 0.7319,
+      "step": 1148
+    },
+    {
+      "epoch": 0.6722197866432851,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0001939571150097466,
+      "loss": 0.6919,
+      "step": 1150
+    },
+    {
+      "epoch": 0.6733888645330995,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019385964912280702,
+      "loss": 0.7407,
+      "step": 1152
+    },
+    {
+      "epoch": 0.6745579424229139,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00019376218323586744,
+      "loss": 0.7096,
+      "step": 1154
+    },
+    {
+      "epoch": 0.6757270203127284,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.0001936647173489279,
+      "loss": 0.6948,
+      "step": 1156
+    },
+    {
+      "epoch": 0.6768960982025427,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00019356725146198832,
+      "loss": 0.6902,
+      "step": 1158
+    },
+    {
+      "epoch": 0.6780651760923572,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019346978557504874,
+      "loss": 0.6991,
+      "step": 1160
+    },
+    {
+      "epoch": 0.6792342539821715,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.00019337231968810916,
+      "loss": 0.7059,
+      "step": 1162
+    },
+    {
+      "epoch": 0.680403331871986,
+      "grad_norm": 0.953125,
+      "learning_rate": 0.0001932748538011696,
+      "loss": 0.8268,
+      "step": 1164
+    },
+    {
+      "epoch": 0.6815724097618003,
+      "grad_norm": 0.69140625,
+      "learning_rate": 0.00019317738791423003,
+      "loss": 0.6697,
+      "step": 1166
+    },
+    {
+      "epoch": 0.6827414876516148,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00019307992202729045,
+      "loss": 0.6828,
+      "step": 1168
+    },
+    {
+      "epoch": 0.6839105655414291,
+      "grad_norm": 0.76953125,
+      "learning_rate": 0.00019298245614035088,
+      "loss": 0.8377,
+      "step": 1170
+    },
+    {
+      "epoch": 0.6850796434312436,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.00019288499025341133,
+      "loss": 0.7405,
+      "step": 1172
+    },
+    {
+      "epoch": 0.686248721321058,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00019278752436647175,
+      "loss": 0.7039,
+      "step": 1174
+    },
+    {
+      "epoch": 0.6874177992108724,
+      "grad_norm": 0.71484375,
+      "learning_rate": 0.00019269005847953217,
+      "loss": 0.7217,
+      "step": 1176
+    },
+    {
+      "epoch": 0.6885868771006868,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.0001925925925925926,
+      "loss": 0.6667,
+      "step": 1178
+    },
+    {
+      "epoch": 0.6897559549905012,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00019249512670565304,
+      "loss": 0.7186,
+      "step": 1180
+    },
+    {
+      "epoch": 0.6909250328803157,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00019239766081871346,
+      "loss": 0.7037,
+      "step": 1182
+    },
+    {
+      "epoch": 0.69209411077013,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001923001949317739,
+      "loss": 0.6953,
+      "step": 1184
+    },
+    {
+      "epoch": 0.6932631886599445,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001922027290448343,
+      "loss": 0.6952,
+      "step": 1186
+    },
+    {
+      "epoch": 0.6944322665497589,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019210526315789473,
+      "loss": 0.7018,
+      "step": 1188
+    },
+    {
+      "epoch": 0.6956013444395733,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00019200779727095518,
+      "loss": 0.675,
+      "step": 1190
+    },
+    {
+      "epoch": 0.6967704223293877,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001919103313840156,
+      "loss": 0.6897,
+      "step": 1192
+    },
+    {
+      "epoch": 0.6979395002192021,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019181286549707603,
+      "loss": 0.6644,
+      "step": 1194
+    },
+    {
+      "epoch": 0.6991085781090165,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00019171539961013647,
+      "loss": 0.7869,
+      "step": 1196
+    },
+    {
+      "epoch": 0.700277655998831,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.0001916179337231969,
+      "loss": 0.6748,
+      "step": 1198
+    },
+    {
+      "epoch": 0.7014467338886453,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00019152046783625732,
+      "loss": 0.6426,
+      "step": 1200
+    },
+    {
+      "epoch": 0.7026158117784598,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00019142300194931774,
+      "loss": 0.6642,
+      "step": 1202
+    },
+    {
+      "epoch": 0.7037848896682741,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00019132553606237816,
+      "loss": 0.7248,
+      "step": 1204
+    },
+    {
+      "epoch": 0.7049539675580886,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001912280701754386,
+      "loss": 0.6599,
+      "step": 1206
+    },
+    {
+      "epoch": 0.7061230454479029,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00019113060428849904,
+      "loss": 0.6483,
+      "step": 1208
+    },
+    {
+      "epoch": 0.7072921233377174,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00019103313840155946,
+      "loss": 0.7163,
+      "step": 1210
+    },
+    {
+      "epoch": 0.7084612012275318,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0001909356725146199,
+      "loss": 0.6607,
+      "step": 1212
+    },
+    {
+      "epoch": 0.7096302791173462,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00019083820662768033,
+      "loss": 0.6647,
+      "step": 1214
+    },
+    {
+      "epoch": 0.7107993570071606,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00019074074074074075,
+      "loss": 0.7187,
+      "step": 1216
+    },
+    {
+      "epoch": 0.711968434896975,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019064327485380117,
+      "loss": 0.6364,
+      "step": 1218
+    },
+    {
+      "epoch": 0.7131375127867894,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0001905458089668616,
+      "loss": 0.6514,
+      "step": 1220
+    },
+    {
+      "epoch": 0.7143065906766038,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00019044834307992202,
+      "loss": 0.6375,
+      "step": 1222
+    },
+    {
+      "epoch": 0.7154756685664182,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00019035087719298247,
+      "loss": 0.653,
+      "step": 1224
+    },
+    {
+      "epoch": 0.7166447464562327,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001902534113060429,
+      "loss": 0.711,
+      "step": 1226
+    },
+    {
+      "epoch": 0.717813824346047,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00019015594541910334,
+      "loss": 0.6696,
+      "step": 1228
+    },
+    {
+      "epoch": 0.7189829022358615,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00019005847953216376,
+      "loss": 0.6747,
+      "step": 1230
+    },
+    {
+      "epoch": 0.7201519801256758,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.00018996101364522418,
+      "loss": 0.6863,
+      "step": 1232
+    },
+    {
+      "epoch": 0.7213210580154903,
+      "grad_norm": 2.953125,
+      "learning_rate": 0.0001898635477582846,
+      "loss": 0.7781,
+      "step": 1234
+    },
+    {
+      "epoch": 0.7224901359053046,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00018976608187134503,
+      "loss": 0.7072,
+      "step": 1236
+    },
+    {
+      "epoch": 0.7236592137951191,
+      "grad_norm": 2.359375,
+      "learning_rate": 0.00018966861598440545,
+      "loss": 0.8231,
+      "step": 1238
+    },
+    {
+      "epoch": 0.7248282916849335,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0001895711500974659,
+      "loss": 0.6973,
+      "step": 1240
+    },
+    {
+      "epoch": 0.7259973695747479,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00018947368421052632,
+      "loss": 0.6941,
+      "step": 1242
+    },
+    {
+      "epoch": 0.7271664474645623,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00018937621832358677,
+      "loss": 0.6481,
+      "step": 1244
+    },
+    {
+      "epoch": 0.7283355253543767,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001892787524366472,
+      "loss": 0.6984,
+      "step": 1246
+    },
+    {
+      "epoch": 0.7295046032441912,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00018918128654970762,
+      "loss": 0.7148,
+      "step": 1248
+    },
+    {
+      "epoch": 0.7306736811340055,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00018908382066276804,
+      "loss": 0.6359,
+      "step": 1250
+    },
+    {
+      "epoch": 0.73184275902382,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00018898635477582846,
+      "loss": 0.6469,
+      "step": 1252
+    },
+    {
+      "epoch": 0.7330118369136344,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00018888888888888888,
+      "loss": 0.6634,
+      "step": 1254
+    },
+    {
+      "epoch": 0.7341809148034488,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001887914230019493,
+      "loss": 0.788,
+      "step": 1256
+    },
+    {
+      "epoch": 0.7353499926932632,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00018869395711500976,
+      "loss": 0.6582,
+      "step": 1258
+    },
+    {
+      "epoch": 0.7365190705830776,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.0001885964912280702,
+      "loss": 0.6793,
+      "step": 1260
+    },
+    {
+      "epoch": 0.737688148472892,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00018849902534113063,
+      "loss": 0.6758,
+      "step": 1262
+    },
+    {
+      "epoch": 0.7388572263627065,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00018840155945419105,
+      "loss": 0.6362,
+      "step": 1264
+    },
+    {
+      "epoch": 0.7400263042525208,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00018830409356725147,
+      "loss": 0.692,
+      "step": 1266
+    },
+    {
+      "epoch": 0.7411953821423353,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.0001882066276803119,
+      "loss": 0.6731,
+      "step": 1268
+    },
+    {
+      "epoch": 0.7423644600321496,
+      "grad_norm": 0.75390625,
+      "learning_rate": 0.00018810916179337232,
+      "loss": 0.7731,
+      "step": 1270
+    },
+    {
+      "epoch": 0.7435335379219641,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00018801169590643274,
+      "loss": 0.6723,
+      "step": 1272
+    },
+    {
+      "epoch": 0.7447026158117784,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001879142300194932,
+      "loss": 0.6414,
+      "step": 1274
+    },
+    {
+      "epoch": 0.7458716937015929,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001878167641325536,
+      "loss": 0.6392,
+      "step": 1276
+    },
+    {
+      "epoch": 0.7470407715914072,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00018771929824561406,
+      "loss": 0.6615,
+      "step": 1278
+    },
+    {
+      "epoch": 0.7482098494812217,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00018762183235867448,
+      "loss": 0.6489,
+      "step": 1280
+    },
+    {
+      "epoch": 0.7493789273710361,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0001875243664717349,
+      "loss": 0.6432,
+      "step": 1282
+    },
+    {
+      "epoch": 0.7505480052608505,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00018742690058479533,
+      "loss": 0.7061,
+      "step": 1284
+    },
+    {
+      "epoch": 0.7517170831506649,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00018732943469785575,
+      "loss": 0.692,
+      "step": 1286
+    },
+    {
+      "epoch": 0.7528861610404793,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00018723196881091617,
+      "loss": 0.641,
+      "step": 1288
+    },
+    {
+      "epoch": 0.7540552389302937,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0001871345029239766,
+      "loss": 0.6842,
+      "step": 1290
+    },
+    {
+      "epoch": 0.7552243168201082,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00018703703703703704,
+      "loss": 0.6471,
+      "step": 1292
+    },
+    {
+      "epoch": 0.7563933947099225,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001869395711500975,
+      "loss": 0.6824,
+      "step": 1294
+    },
+    {
+      "epoch": 0.757562472599737,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00018684210526315792,
+      "loss": 0.6166,
+      "step": 1296
+    },
+    {
+      "epoch": 0.7587315504895513,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.00018674463937621834,
+      "loss": 0.749,
+      "step": 1298
+    },
+    {
+      "epoch": 0.7599006283793658,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.00018664717348927876,
+      "loss": 0.7687,
+      "step": 1300
+    },
+    {
+      "epoch": 0.7610697062691801,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.00018654970760233918,
+      "loss": 0.7456,
+      "step": 1302
+    },
+    {
+      "epoch": 0.7622387841589946,
+      "grad_norm": 0.8828125,
+      "learning_rate": 0.0001864522417153996,
+      "loss": 0.706,
+      "step": 1304
+    },
+    {
+      "epoch": 0.763407862048809,
+      "grad_norm": 0.734375,
+      "learning_rate": 0.00018635477582846003,
+      "loss": 0.9545,
+      "step": 1306
+    },
+    {
+      "epoch": 0.7645769399386234,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00018625730994152048,
+      "loss": 0.6858,
+      "step": 1308
+    },
+    {
+      "epoch": 0.7657460178284379,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001861598440545809,
+      "loss": 0.6233,
+      "step": 1310
+    },
+    {
+      "epoch": 0.7669150957182522,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00018606237816764135,
+      "loss": 0.6781,
+      "step": 1312
+    },
+    {
+      "epoch": 0.7680841736080667,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00018596491228070177,
+      "loss": 0.6687,
+      "step": 1314
+    },
+    {
+      "epoch": 0.769253251497881,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001858674463937622,
+      "loss": 0.7021,
+      "step": 1316
+    },
+    {
+      "epoch": 0.7704223293876955,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00018576998050682262,
+      "loss": 0.6391,
+      "step": 1318
+    },
+    {
+      "epoch": 0.7715914072775099,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00018567251461988304,
+      "loss": 0.6663,
+      "step": 1320
+    },
+    {
+      "epoch": 0.7727604851673243,
+      "grad_norm": 0.98046875,
+      "learning_rate": 0.00018557504873294346,
+      "loss": 0.6668,
+      "step": 1322
+    },
+    {
+      "epoch": 0.7739295630571387,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001854775828460039,
+      "loss": 0.6051,
+      "step": 1324
+    },
+    {
+      "epoch": 0.7750986409469531,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00018538011695906433,
+      "loss": 0.6486,
+      "step": 1326
+    },
+    {
+      "epoch": 0.7762677188367675,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00018528265107212478,
+      "loss": 0.6645,
+      "step": 1328
+    },
+    {
+      "epoch": 0.777436796726582,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001851851851851852,
+      "loss": 0.6235,
+      "step": 1330
+    },
+    {
+      "epoch": 0.7786058746163963,
+      "grad_norm": 1.46875,
+      "learning_rate": 0.00018508771929824563,
+      "loss": 0.7003,
+      "step": 1332
+    },
+    {
+      "epoch": 0.7797749525062108,
+      "grad_norm": 0.890625,
+      "learning_rate": 0.00018499025341130605,
+      "loss": 0.7938,
+      "step": 1334
+    },
+    {
+      "epoch": 0.7809440303960251,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00018489278752436647,
+      "loss": 0.6972,
+      "step": 1336
+    },
+    {
+      "epoch": 0.7821131082858396,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001847953216374269,
+      "loss": 0.6443,
+      "step": 1338
+    },
+    {
+      "epoch": 0.7832821861756539,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00018469785575048734,
+      "loss": 0.5978,
+      "step": 1340
+    },
+    {
+      "epoch": 0.7844512640654684,
+      "grad_norm": 1.2109375,
+      "learning_rate": 0.00018460038986354776,
+      "loss": 0.7214,
+      "step": 1342
+    },
+    {
+      "epoch": 0.7856203419552827,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00018450292397660819,
+      "loss": 0.6563,
+      "step": 1344
+    },
+    {
+      "epoch": 0.7867894198450972,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00018440545808966864,
+      "loss": 0.6051,
+      "step": 1346
+    },
+    {
+      "epoch": 0.7879584977349116,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00018430799220272906,
+      "loss": 0.6615,
+      "step": 1348
+    },
+    {
+      "epoch": 0.789127575624726,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00018421052631578948,
+      "loss": 0.727,
+      "step": 1350
+    },
+    {
+      "epoch": 0.7902966535145404,
+      "grad_norm": 0.7734375,
+      "learning_rate": 0.0001841130604288499,
+      "loss": 0.6712,
+      "step": 1352
+    },
+    {
+      "epoch": 0.7914657314043548,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00018401559454191032,
+      "loss": 0.7094,
+      "step": 1354
+    },
+    {
+      "epoch": 0.7926348092941692,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00018391812865497077,
+      "loss": 0.6863,
+      "step": 1356
+    },
+    {
+      "epoch": 0.7938038871839836,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001838206627680312,
+      "loss": 0.6378,
+      "step": 1358
+    },
+    {
+      "epoch": 0.794972965073798,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.00018372319688109162,
+      "loss": 0.8658,
+      "step": 1360
+    },
+    {
+      "epoch": 0.7961420429636125,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00018362573099415207,
+      "loss": 0.6507,
+      "step": 1362
+    },
+    {
+      "epoch": 0.7973111208534268,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001835282651072125,
+      "loss": 0.7661,
+      "step": 1364
+    },
+    {
+      "epoch": 0.7984801987432413,
+      "grad_norm": 0.625,
+      "learning_rate": 0.0001834307992202729,
+      "loss": 0.6932,
+      "step": 1366
+    },
+    {
+      "epoch": 0.7996492766330556,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00018333333333333334,
+      "loss": 0.681,
+      "step": 1368
+    },
+    {
+      "epoch": 0.8008183545228701,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00018323586744639376,
+      "loss": 0.6374,
+      "step": 1370
+    },
+    {
+      "epoch": 0.8019874324126844,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001831384015594542,
+      "loss": 0.657,
+      "step": 1372
+    },
+    {
+      "epoch": 0.8031565103024989,
+      "grad_norm": 0.83203125,
+      "learning_rate": 0.00018304093567251463,
+      "loss": 0.6511,
+      "step": 1374
+    },
+    {
+      "epoch": 0.8043255881923134,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00018294346978557505,
+      "loss": 0.6286,
+      "step": 1376
+    },
+    {
+      "epoch": 0.8054946660821277,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00018284600389863547,
+      "loss": 0.6964,
+      "step": 1378
+    },
+    {
+      "epoch": 0.8066637439719422,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00018274853801169592,
+      "loss": 0.6383,
+      "step": 1380
+    },
+    {
+      "epoch": 0.8078328218617565,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00018265107212475635,
+      "loss": 0.6459,
+      "step": 1382
+    },
+    {
+      "epoch": 0.809001899751571,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00018255360623781677,
+      "loss": 0.7644,
+      "step": 1384
+    },
+    {
+      "epoch": 0.8101709776413853,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.0001824561403508772,
+      "loss": 0.6872,
+      "step": 1386
+    },
+    {
+      "epoch": 0.8113400555311998,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00018235867446393764,
+      "loss": 0.7238,
+      "step": 1388
+    },
+    {
+      "epoch": 0.8125091334210142,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00018226120857699806,
+      "loss": 0.6579,
+      "step": 1390
+    },
+    {
+      "epoch": 0.8136782113108286,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00018216374269005848,
+      "loss": 0.6741,
+      "step": 1392
+    },
+    {
+      "epoch": 0.814847289200643,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001820662768031189,
+      "loss": 0.6218,
+      "step": 1394
+    },
+    {
+      "epoch": 0.8160163670904574,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00018196881091617936,
+      "loss": 0.6077,
+      "step": 1396
+    },
+    {
+      "epoch": 0.8171854449802718,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00018187134502923978,
+      "loss": 0.6797,
+      "step": 1398
+    },
+    {
+      "epoch": 0.8183545228700863,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0001817738791423002,
+      "loss": 0.6416,
+      "step": 1400
+    },
+    {
+      "epoch": 0.8195236007599006,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00018167641325536062,
+      "loss": 0.6664,
+      "step": 1402
+    },
+    {
+      "epoch": 0.8206926786497151,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00018157894736842107,
+      "loss": 0.6795,
+      "step": 1404
+    },
+    {
+      "epoch": 0.8218617565395294,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.0001814814814814815,
+      "loss": 0.7478,
+      "step": 1406
+    },
+    {
+      "epoch": 0.8230308344293439,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00018138401559454192,
+      "loss": 0.6926,
+      "step": 1408
+    },
+    {
+      "epoch": 0.8241999123191582,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00018128654970760234,
+      "loss": 0.6235,
+      "step": 1410
+    },
+    {
+      "epoch": 0.8253689902089727,
+      "grad_norm": 1.203125,
+      "learning_rate": 0.00018118908382066276,
+      "loss": 0.7571,
+      "step": 1412
+    },
+    {
+      "epoch": 0.826538068098787,
+      "grad_norm": 1.046875,
+      "learning_rate": 0.0001810916179337232,
+      "loss": 0.6671,
+      "step": 1414
+    },
+    {
+      "epoch": 0.8277071459886015,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00018099415204678363,
+      "loss": 0.715,
+      "step": 1416
+    },
+    {
+      "epoch": 0.8288762238784159,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00018089668615984406,
+      "loss": 0.643,
+      "step": 1418
+    },
+    {
+      "epoch": 0.8300453017682303,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0001807992202729045,
+      "loss": 0.6701,
+      "step": 1420
+    },
+    {
+      "epoch": 0.8312143796580447,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00018070175438596493,
+      "loss": 0.6674,
+      "step": 1422
+    },
+    {
+      "epoch": 0.8323834575478591,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00018060428849902535,
+      "loss": 0.6442,
+      "step": 1424
+    },
+    {
+      "epoch": 0.8335525354376735,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00018050682261208577,
+      "loss": 0.6345,
+      "step": 1426
+    },
+    {
+      "epoch": 0.834721613327488,
+      "grad_norm": 0.640625,
+      "learning_rate": 0.0001804093567251462,
+      "loss": 0.7375,
+      "step": 1428
+    },
+    {
+      "epoch": 0.8358906912173023,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00018031189083820664,
+      "loss": 0.6404,
+      "step": 1430
+    },
+    {
+      "epoch": 0.8370597691071168,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00018021442495126707,
+      "loss": 0.6356,
+      "step": 1432
+    },
+    {
+      "epoch": 0.8382288469969311,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001801169590643275,
+      "loss": 0.6207,
+      "step": 1434
+    },
+    {
+      "epoch": 0.8393979248867456,
+      "grad_norm": 0.6171875,
+      "learning_rate": 0.00018001949317738794,
+      "loss": 0.6375,
+      "step": 1436
+    },
+    {
+      "epoch": 0.8405670027765599,
+      "grad_norm": 1.453125,
+      "learning_rate": 0.00017992202729044836,
+      "loss": 0.8081,
+      "step": 1438
+    },
+    {
+      "epoch": 0.8417360806663744,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00017982456140350878,
+      "loss": 0.7778,
+      "step": 1440
+    },
+    {
+      "epoch": 0.8429051585561889,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.0001797270955165692,
+      "loss": 0.7142,
+      "step": 1442
+    },
+    {
+      "epoch": 0.8440742364460032,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00017962962962962963,
+      "loss": 0.6502,
+      "step": 1444
+    },
+    {
+      "epoch": 0.8452433143358177,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00017953216374269005,
+      "loss": 0.6668,
+      "step": 1446
+    },
+    {
+      "epoch": 0.846412392225632,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001794346978557505,
+      "loss": 0.6663,
+      "step": 1448
+    },
+    {
+      "epoch": 0.8475814701154465,
+      "grad_norm": 3.65625,
+      "learning_rate": 0.00017933723196881092,
+      "loss": 0.7231,
+      "step": 1450
+    },
+    {
+      "epoch": 0.8487505480052608,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00017923976608187137,
+      "loss": 0.6152,
+      "step": 1452
+    },
+    {
+      "epoch": 0.8499196258950753,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001791423001949318,
+      "loss": 0.7494,
+      "step": 1454
+    },
+    {
+      "epoch": 0.8510887037848897,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00017904483430799221,
+      "loss": 0.6777,
+      "step": 1456
+    },
+    {
+      "epoch": 0.8522577816747041,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00017894736842105264,
+      "loss": 0.6667,
+      "step": 1458
+    },
+    {
+      "epoch": 0.8534268595645185,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00017884990253411306,
+      "loss": 0.6389,
+      "step": 1460
+    },
+    {
+      "epoch": 0.8545959374543329,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00017875243664717348,
+      "loss": 0.6435,
+      "step": 1462
+    },
+    {
+      "epoch": 0.8557650153441473,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00017865497076023393,
+      "loss": 0.654,
+      "step": 1464
+    },
+    {
+      "epoch": 0.8569340932339617,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00017855750487329435,
+      "loss": 0.6692,
+      "step": 1466
+    },
+    {
+      "epoch": 0.8581031711237761,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.0001784600389863548,
+      "loss": 0.6807,
+      "step": 1468
+    },
+    {
+      "epoch": 0.8592722490135906,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00017836257309941523,
+      "loss": 0.6314,
+      "step": 1470
+    },
+    {
+      "epoch": 0.8604413269034049,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00017826510721247565,
+      "loss": 0.6068,
+      "step": 1472
+    },
+    {
+      "epoch": 0.8616104047932194,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00017816764132553607,
+      "loss": 0.6545,
+      "step": 1474
+    },
+    {
+      "epoch": 0.8627794826830337,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0001780701754385965,
+      "loss": 0.6898,
+      "step": 1476
+    },
+    {
+      "epoch": 0.8639485605728482,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00017797270955165691,
+      "loss": 0.6213,
+      "step": 1478
+    },
+    {
+      "epoch": 0.8651176384626625,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00017787524366471734,
+      "loss": 0.6487,
+      "step": 1480
+    },
+    {
+      "epoch": 0.866286716352477,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00017777777777777779,
+      "loss": 0.6895,
+      "step": 1482
+    },
+    {
+      "epoch": 0.8674557942422914,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00017768031189083824,
+      "loss": 0.6522,
+      "step": 1484
+    },
+    {
+      "epoch": 0.8686248721321058,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00017758284600389866,
+      "loss": 0.6484,
+      "step": 1486
+    },
+    {
+      "epoch": 0.8697939500219202,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00017748538011695908,
+      "loss": 0.6517,
+      "step": 1488
+    },
+    {
+      "epoch": 0.8709630279117346,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0001773879142300195,
+      "loss": 0.6644,
+      "step": 1490
+    },
+    {
+      "epoch": 0.872132105801549,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00017729044834307992,
+      "loss": 0.6383,
+      "step": 1492
+    },
+    {
+      "epoch": 0.8733011836913634,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00017719298245614035,
+      "loss": 0.7941,
+      "step": 1494
+    },
+    {
+      "epoch": 0.8744702615811778,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00017709551656920077,
+      "loss": 0.6882,
+      "step": 1496
+    },
+    {
+      "epoch": 0.8756393394709923,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00017699805068226122,
+      "loss": 0.627,
+      "step": 1498
+    },
+    {
+      "epoch": 0.8768084173608066,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00017690058479532167,
+      "loss": 0.6141,
+      "step": 1500
+    },
+    {
+      "epoch": 0.8779774952506211,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001768031189083821,
+      "loss": 0.6226,
+      "step": 1502
+    },
+    {
+      "epoch": 0.8791465731404354,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001767056530214425,
+      "loss": 0.6834,
+      "step": 1504
+    },
+    {
+      "epoch": 0.8803156510302499,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00017660818713450294,
+      "loss": 0.6269,
+      "step": 1506
+    },
+    {
+      "epoch": 0.8814847289200644,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00017651072124756336,
+      "loss": 0.6494,
+      "step": 1508
+    },
+    {
+      "epoch": 0.8826538068098787,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00017641325536062378,
+      "loss": 0.6028,
+      "step": 1510
+    },
+    {
+      "epoch": 0.8838228846996932,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001763157894736842,
+      "loss": 0.6428,
+      "step": 1512
+    },
+    {
+      "epoch": 0.8849919625895075,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00017621832358674462,
+      "loss": 0.6303,
+      "step": 1514
+    },
+    {
+      "epoch": 0.886161040479322,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00017612085769980507,
+      "loss": 0.6522,
+      "step": 1516
+    },
+    {
+      "epoch": 0.8873301183691363,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00017602339181286552,
+      "loss": 0.6279,
+      "step": 1518
+    },
+    {
+      "epoch": 0.8884991962589508,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00017592592592592595,
+      "loss": 0.6507,
+      "step": 1520
+    },
+    {
+      "epoch": 0.8896682741487651,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00017582846003898637,
+      "loss": 0.6137,
+      "step": 1522
+    },
+    {
+      "epoch": 0.8908373520385796,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001757309941520468,
+      "loss": 0.649,
+      "step": 1524
+    },
+    {
+      "epoch": 0.892006429928394,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001756335282651072,
+      "loss": 0.6498,
+      "step": 1526
+    },
+    {
+      "epoch": 0.8931755078182084,
+      "grad_norm": 1.59375,
+      "learning_rate": 0.00017553606237816763,
+      "loss": 0.6804,
+      "step": 1528
+    },
+    {
+      "epoch": 0.8943445857080228,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00017543859649122806,
+      "loss": 0.6186,
+      "step": 1530
+    },
+    {
+      "epoch": 0.8955136635978372,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0001753411306042885,
+      "loss": 0.6518,
+      "step": 1532
+    },
+    {
+      "epoch": 0.8966827414876516,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00017524366471734896,
+      "loss": 0.6458,
+      "step": 1534
+    },
+    {
+      "epoch": 0.897851819377466,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00017514619883040938,
+      "loss": 0.6642,
+      "step": 1536
+    },
+    {
+      "epoch": 0.8990208972672804,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001750487329434698,
+      "loss": 0.6254,
+      "step": 1538
+    },
+    {
+      "epoch": 0.9001899751570949,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00017495126705653022,
+      "loss": 0.6158,
+      "step": 1540
+    },
+    {
+      "epoch": 0.9013590530469092,
+      "grad_norm": 0.65234375,
+      "learning_rate": 0.00017485380116959065,
+      "loss": 0.6635,
+      "step": 1542
+    },
+    {
+      "epoch": 0.9025281309367237,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00017475633528265107,
+      "loss": 0.6242,
+      "step": 1544
+    },
+    {
+      "epoch": 0.903697208826538,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.0001746588693957115,
+      "loss": 0.6503,
+      "step": 1546
+    },
+    {
+      "epoch": 0.9048662867163525,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00017456140350877194,
+      "loss": 0.5875,
+      "step": 1548
+    },
+    {
+      "epoch": 0.9060353646061668,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00017446393762183236,
+      "loss": 0.6683,
+      "step": 1550
+    },
+    {
+      "epoch": 0.9072044424959813,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.0001743664717348928,
+      "loss": 0.6176,
+      "step": 1552
+    },
+    {
+      "epoch": 0.9083735203857957,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00017426900584795323,
+      "loss": 0.6351,
+      "step": 1554
+    },
+    {
+      "epoch": 0.9095425982756101,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00017417153996101366,
+      "loss": 0.668,
+      "step": 1556
+    },
+    {
+      "epoch": 0.9107116761654245,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00017407407407407408,
+      "loss": 0.6425,
+      "step": 1558
+    },
+    {
+      "epoch": 0.9118807540552389,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001739766081871345,
+      "loss": 0.6127,
+      "step": 1560
+    },
+    {
+      "epoch": 0.9130498319450533,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00017387914230019492,
+      "loss": 0.6055,
+      "step": 1562
+    },
+    {
+      "epoch": 0.9142189098348678,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00017378167641325537,
+      "loss": 0.6276,
+      "step": 1564
+    },
+    {
+      "epoch": 0.9153879877246821,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001736842105263158,
+      "loss": 0.6818,
+      "step": 1566
+    },
+    {
+      "epoch": 0.9165570656144966,
+      "grad_norm": 0.66015625,
+      "learning_rate": 0.00017358674463937624,
+      "loss": 0.76,
+      "step": 1568
+    },
+    {
+      "epoch": 0.9177261435043109,
+      "grad_norm": 4.78125,
+      "learning_rate": 0.00017348927875243667,
+      "loss": 0.6739,
+      "step": 1570
+    },
+    {
+      "epoch": 0.9188952213941254,
+      "grad_norm": 2.078125,
+      "learning_rate": 0.0001733918128654971,
+      "loss": 0.6578,
+      "step": 1572
+    },
+    {
+      "epoch": 0.9200642992839398,
+      "grad_norm": 0.8515625,
+      "learning_rate": 0.0001732943469785575,
+      "loss": 0.6705,
+      "step": 1574
+    },
+    {
+      "epoch": 0.9212333771737542,
+      "grad_norm": 0.80078125,
+      "learning_rate": 0.00017319688109161793,
+      "loss": 0.6605,
+      "step": 1576
+    },
+    {
+      "epoch": 0.9224024550635687,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00017309941520467836,
+      "loss": 0.6346,
+      "step": 1578
+    },
+    {
+      "epoch": 0.923571532953383,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.0001730019493177388,
+      "loss": 0.6291,
+      "step": 1580
+    },
+    {
+      "epoch": 0.9247406108431975,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00017290448343079923,
+      "loss": 0.6774,
+      "step": 1582
+    },
+    {
+      "epoch": 0.9259096887330118,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00017280701754385965,
+      "loss": 0.6603,
+      "step": 1584
+    },
+    {
+      "epoch": 0.9270787666228263,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.0001727095516569201,
+      "loss": 0.6435,
+      "step": 1586
+    },
+    {
+      "epoch": 0.9282478445126406,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.00017261208576998052,
+      "loss": 0.7842,
+      "step": 1588
+    },
+    {
+      "epoch": 0.9294169224024551,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00017251461988304094,
+      "loss": 0.6378,
+      "step": 1590
+    },
+    {
+      "epoch": 0.9305860002922695,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00017241715399610137,
+      "loss": 0.6809,
+      "step": 1592
+    },
+    {
+      "epoch": 0.9317550781820839,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.0001723196881091618,
+      "loss": 0.6226,
+      "step": 1594
+    },
+    {
+      "epoch": 0.9329241560718983,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00017222222222222224,
+      "loss": 0.6763,
+      "step": 1596
+    },
+    {
+      "epoch": 0.9340932339617127,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00017212475633528266,
+      "loss": 0.7279,
+      "step": 1598
+    },
+    {
+      "epoch": 0.9352623118515271,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00017202729044834308,
+      "loss": 0.6655,
+      "step": 1600
+    },
+    {
+      "epoch": 0.9364313897413415,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00017192982456140353,
+      "loss": 0.6658,
+      "step": 1602
+    },
+    {
+      "epoch": 0.9376004676311559,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00017183235867446395,
+      "loss": 0.7698,
+      "step": 1604
+    },
+    {
+      "epoch": 0.9387695455209704,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00017173489278752438,
+      "loss": 0.6555,
+      "step": 1606
+    },
+    {
+      "epoch": 0.9399386234107847,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001716374269005848,
+      "loss": 0.6465,
+      "step": 1608
+    },
+    {
+      "epoch": 0.9411077013005992,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00017153996101364522,
+      "loss": 0.6642,
+      "step": 1610
+    },
+    {
+      "epoch": 0.9422767791904135,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00017144249512670567,
+      "loss": 0.6579,
+      "step": 1612
+    },
+    {
+      "epoch": 0.943445857080228,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0001713450292397661,
+      "loss": 0.6427,
+      "step": 1614
+    },
+    {
+      "epoch": 0.9446149349700423,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00017124756335282651,
+      "loss": 0.6114,
+      "step": 1616
+    },
+    {
+      "epoch": 0.9457840128598568,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00017115009746588694,
+      "loss": 0.5887,
+      "step": 1618
+    },
+    {
+      "epoch": 0.9469530907496712,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00017105263157894739,
+      "loss": 0.6266,
+      "step": 1620
+    },
+    {
+      "epoch": 0.9481221686394856,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0001709551656920078,
+      "loss": 0.642,
+      "step": 1622
+    },
+    {
+      "epoch": 0.9492912465293,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00017085769980506823,
+      "loss": 0.605,
+      "step": 1624
+    },
+    {
+      "epoch": 0.9504603244191144,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00017076023391812865,
+      "loss": 0.5981,
+      "step": 1626
+    },
+    {
+      "epoch": 0.9516294023089288,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001706627680311891,
+      "loss": 0.7126,
+      "step": 1628
+    },
+    {
+      "epoch": 0.9527984801987432,
+      "grad_norm": 0.78125,
+      "learning_rate": 0.00017056530214424952,
+      "loss": 0.7645,
+      "step": 1630
+    },
+    {
+      "epoch": 0.9539675580885576,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00017046783625730995,
+      "loss": 0.7024,
+      "step": 1632
+    },
+    {
+      "epoch": 0.9551366359783721,
+      "grad_norm": 0.99609375,
+      "learning_rate": 0.00017037037037037037,
+      "loss": 0.7639,
+      "step": 1634
+    },
+    {
+      "epoch": 0.9563057138681864,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00017027290448343082,
+      "loss": 0.624,
+      "step": 1636
+    },
+    {
+      "epoch": 0.9574747917580009,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00017017543859649124,
+      "loss": 0.6589,
+      "step": 1638
+    },
+    {
+      "epoch": 0.9586438696478153,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00017007797270955166,
+      "loss": 0.6068,
+      "step": 1640
+    },
+    {
+      "epoch": 0.9598129475376297,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00016998050682261209,
+      "loss": 0.6223,
+      "step": 1642
+    },
+    {
+      "epoch": 0.9609820254274442,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00016988304093567254,
+      "loss": 0.6217,
+      "step": 1644
+    },
+    {
+      "epoch": 0.9621511033172585,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00016978557504873296,
+      "loss": 0.6249,
+      "step": 1646
+    },
+    {
+      "epoch": 0.963320181207073,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00016968810916179338,
+      "loss": 0.6676,
+      "step": 1648
+    },
+    {
+      "epoch": 0.9644892590968873,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001695906432748538,
+      "loss": 0.6327,
+      "step": 1650
+    },
+    {
+      "epoch": 0.9656583369867018,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00016949317738791422,
+      "loss": 0.6233,
+      "step": 1652
+    },
+    {
+      "epoch": 0.9668274148765161,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00016939571150097467,
+      "loss": 0.6466,
+      "step": 1654
+    },
+    {
+      "epoch": 0.9679964927663306,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0001692982456140351,
+      "loss": 0.6173,
+      "step": 1656
+    },
+    {
+      "epoch": 0.969165570656145,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00016920077972709552,
+      "loss": 0.6279,
+      "step": 1658
+    },
+    {
+      "epoch": 0.9703346485459594,
+      "grad_norm": 4.78125,
+      "learning_rate": 0.00016910331384015597,
+      "loss": 0.8359,
+      "step": 1660
+    },
+    {
+      "epoch": 0.9715037264357738,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.0001690058479532164,
+      "loss": 0.6702,
+      "step": 1662
+    },
+    {
+      "epoch": 0.9726728043255882,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0001689083820662768,
+      "loss": 0.6553,
+      "step": 1664
+    },
+    {
+      "epoch": 0.9738418822154026,
+      "grad_norm": 1.9453125,
+      "learning_rate": 0.00016881091617933723,
+      "loss": 0.7719,
+      "step": 1666
+    },
+    {
+      "epoch": 0.975010960105217,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00016871345029239766,
+      "loss": 0.6581,
+      "step": 1668
+    },
+    {
+      "epoch": 0.9761800379950314,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001686159844054581,
+      "loss": 0.6169,
+      "step": 1670
+    },
+    {
+      "epoch": 0.9773491158848459,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00016851851851851853,
+      "loss": 0.6449,
+      "step": 1672
+    },
+    {
+      "epoch": 0.9785181937746602,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00016842105263157895,
+      "loss": 0.6011,
+      "step": 1674
+    },
+    {
+      "epoch": 0.9796872716644747,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0001683235867446394,
+      "loss": 0.6331,
+      "step": 1676
+    },
+    {
+      "epoch": 0.980856349554289,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00016822612085769982,
+      "loss": 0.635,
+      "step": 1678
+    },
+    {
+      "epoch": 0.9820254274441035,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00016812865497076025,
+      "loss": 0.6567,
+      "step": 1680
+    },
+    {
+      "epoch": 0.9831945053339178,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00016803118908382067,
+      "loss": 0.592,
+      "step": 1682
+    },
+    {
+      "epoch": 0.9843635832237323,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.0001679337231968811,
+      "loss": 0.6291,
+      "step": 1684
+    },
+    {
+      "epoch": 0.9855326611135466,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001678362573099415,
+      "loss": 0.6125,
+      "step": 1686
+    },
+    {
+      "epoch": 0.9867017390033611,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00016773879142300196,
+      "loss": 0.5989,
+      "step": 1688
+    },
+    {
+      "epoch": 0.9878708168931755,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00016764132553606238,
+      "loss": 0.6214,
+      "step": 1690
+    },
+    {
+      "epoch": 0.9890398947829899,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00016754385964912283,
+      "loss": 0.5794,
+      "step": 1692
+    },
+    {
+      "epoch": 0.9902089726728043,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00016744639376218326,
+      "loss": 0.6138,
+      "step": 1694
+    },
+    {
+      "epoch": 0.9913780505626187,
+      "grad_norm": 0.765625,
+      "learning_rate": 0.00016734892787524368,
+      "loss": 0.5763,
+      "step": 1696
+    },
+    {
+      "epoch": 0.9925471284524331,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001672514619883041,
+      "loss": 0.5501,
+      "step": 1698
+    },
+    {
+      "epoch": 0.9937162063422476,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00016715399610136452,
+      "loss": 0.605,
+      "step": 1700
+    },
+    {
+      "epoch": 0.9948852842320619,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00016705653021442494,
+      "loss": 0.5973,
+      "step": 1702
+    },
+    {
+      "epoch": 0.9960543621218764,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.0001669590643274854,
+      "loss": 0.5792,
+      "step": 1704
+    },
+    {
+      "epoch": 0.9972234400116908,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00016686159844054582,
+      "loss": 0.5833,
+      "step": 1706
+    },
+    {
+      "epoch": 0.9983925179015052,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00016676413255360627,
+      "loss": 0.6092,
+      "step": 1708
+    },
+    {
+      "epoch": 0.9995615957913196,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001666666666666667,
+      "loss": 0.6591,
+      "step": 1710
+    },
+    {
+      "epoch": 1.000730673681134,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001665692007797271,
+      "loss": 0.5507,
+      "step": 1712
+    },
+    {
+      "epoch": 1.0018997515709485,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00016647173489278753,
+      "loss": 0.5538,
+      "step": 1714
+    },
+    {
+      "epoch": 1.0030688294607628,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00016637426900584796,
+      "loss": 0.511,
+      "step": 1716
+    },
+    {
+      "epoch": 1.0042379073505772,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00016627680311890838,
+      "loss": 0.5208,
+      "step": 1718
+    },
+    {
+      "epoch": 1.0054069852403917,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0001661793372319688,
+      "loss": 0.5428,
+      "step": 1720
+    },
+    {
+      "epoch": 1.006576063130206,
+      "grad_norm": 2.71875,
+      "learning_rate": 0.00016608187134502925,
+      "loss": 0.5979,
+      "step": 1722
+    },
+    {
+      "epoch": 1.0077451410200204,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0001659844054580897,
+      "loss": 0.5444,
+      "step": 1724
+    },
+    {
+      "epoch": 1.0089142189098348,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00016588693957115012,
+      "loss": 0.574,
+      "step": 1726
+    },
+    {
+      "epoch": 1.0100832967996494,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00016578947368421054,
+      "loss": 0.4927,
+      "step": 1728
+    },
+    {
+      "epoch": 1.0112523746894637,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00016569200779727097,
+      "loss": 0.5525,
+      "step": 1730
+    },
+    {
+      "epoch": 1.012421452579278,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0001655945419103314,
+      "loss": 0.5884,
+      "step": 1732
+    },
+    {
+      "epoch": 1.0135905304690924,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0001654970760233918,
+      "loss": 0.4956,
+      "step": 1734
+    },
+    {
+      "epoch": 1.014759608358907,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00016539961013645223,
+      "loss": 0.5446,
+      "step": 1736
+    },
+    {
+      "epoch": 1.0159286862487213,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00016530214424951268,
+      "loss": 0.5602,
+      "step": 1738
+    },
+    {
+      "epoch": 1.0170977641385357,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.0001652046783625731,
+      "loss": 0.5462,
+      "step": 1740
+    },
+    {
+      "epoch": 1.01826684202835,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00016510721247563355,
+      "loss": 0.53,
+      "step": 1742
+    },
+    {
+      "epoch": 1.0194359199181646,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00016500974658869398,
+      "loss": 0.5374,
+      "step": 1744
+    },
+    {
+      "epoch": 1.020604997807979,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0001649122807017544,
+      "loss": 0.5258,
+      "step": 1746
+    },
+    {
+      "epoch": 1.0217740756977933,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00016481481481481482,
+      "loss": 0.5118,
+      "step": 1748
+    },
+    {
+      "epoch": 1.0229431535876077,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00016471734892787524,
+      "loss": 0.4822,
+      "step": 1750
+    },
+    {
+      "epoch": 1.0241122314774223,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00016461988304093567,
+      "loss": 0.5435,
+      "step": 1752
+    },
+    {
+      "epoch": 1.0252813093672366,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001645224171539961,
+      "loss": 0.5703,
+      "step": 1754
+    },
+    {
+      "epoch": 1.026450387257051,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00016442495126705654,
+      "loss": 0.587,
+      "step": 1756
+    },
+    {
+      "epoch": 1.0276194651468653,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00016432748538011699,
+      "loss": 0.5146,
+      "step": 1758
+    },
+    {
+      "epoch": 1.0287885430366799,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001642300194931774,
+      "loss": 0.5102,
+      "step": 1760
+    },
+    {
+      "epoch": 1.0299576209264942,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00016413255360623783,
+      "loss": 0.5178,
+      "step": 1762
+    },
+    {
+      "epoch": 1.0311266988163086,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00016403508771929825,
+      "loss": 0.4984,
+      "step": 1764
+    },
+    {
+      "epoch": 1.032295776706123,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00016393762183235868,
+      "loss": 0.4915,
+      "step": 1766
+    },
+    {
+      "epoch": 1.0334648545959375,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0001638401559454191,
+      "loss": 0.5493,
+      "step": 1768
+    },
+    {
+      "epoch": 1.0346339324857519,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00016374269005847952,
+      "loss": 0.4906,
+      "step": 1770
+    },
+    {
+      "epoch": 1.0358030103755662,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00016364522417153997,
+      "loss": 0.5478,
+      "step": 1772
+    },
+    {
+      "epoch": 1.0369720882653808,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001635477582846004,
+      "loss": 0.5187,
+      "step": 1774
+    },
+    {
+      "epoch": 1.0381411661551951,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00016345029239766084,
+      "loss": 0.5421,
+      "step": 1776
+    },
+    {
+      "epoch": 1.0393102440450095,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00016335282651072126,
+      "loss": 0.5583,
+      "step": 1778
+    },
+    {
+      "epoch": 1.0404793219348238,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00016325536062378169,
+      "loss": 0.5283,
+      "step": 1780
+    },
+    {
+      "epoch": 1.0416483998246384,
+      "grad_norm": 0.7578125,
+      "learning_rate": 0.0001631578947368421,
+      "loss": 0.5294,
+      "step": 1782
+    },
+    {
+      "epoch": 1.0428174777144528,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00016306042884990253,
+      "loss": 0.5682,
+      "step": 1784
+    },
+    {
+      "epoch": 1.0439865556042671,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00016296296296296295,
+      "loss": 0.5088,
+      "step": 1786
+    },
+    {
+      "epoch": 1.0451556334940815,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001628654970760234,
+      "loss": 0.4887,
+      "step": 1788
+    },
+    {
+      "epoch": 1.046324711383896,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00016276803118908382,
+      "loss": 0.5478,
+      "step": 1790
+    },
+    {
+      "epoch": 1.0474937892737104,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00016267056530214427,
+      "loss": 0.5156,
+      "step": 1792
+    },
+    {
+      "epoch": 1.0486628671635247,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001625730994152047,
+      "loss": 0.528,
+      "step": 1794
+    },
+    {
+      "epoch": 1.049831945053339,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00016247563352826512,
+      "loss": 0.5,
+      "step": 1796
+    },
+    {
+      "epoch": 1.0510010229431537,
+      "grad_norm": 0.984375,
+      "learning_rate": 0.00016237816764132554,
+      "loss": 0.6486,
+      "step": 1798
+    },
+    {
+      "epoch": 1.052170100832968,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00016228070175438596,
+      "loss": 0.5215,
+      "step": 1800
+    },
+    {
+      "epoch": 1.0533391787227824,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00016218323586744639,
+      "loss": 0.5179,
+      "step": 1802
+    },
+    {
+      "epoch": 1.0545082566125967,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00016208576998050683,
+      "loss": 0.5772,
+      "step": 1804
+    },
+    {
+      "epoch": 1.0556773345024113,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00016198830409356726,
+      "loss": 0.504,
+      "step": 1806
+    },
+    {
+      "epoch": 1.0568464123922257,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001618908382066277,
+      "loss": 0.5463,
+      "step": 1808
+    },
+    {
+      "epoch": 1.05801549028204,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00016179337231968813,
+      "loss": 0.5146,
+      "step": 1810
+    },
+    {
+      "epoch": 1.0591845681718544,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00016169590643274855,
+      "loss": 0.5626,
+      "step": 1812
+    },
+    {
+      "epoch": 1.060353646061669,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00016159844054580897,
+      "loss": 0.5224,
+      "step": 1814
+    },
+    {
+      "epoch": 1.0615227239514833,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001615009746588694,
+      "loss": 0.5456,
+      "step": 1816
+    },
+    {
+      "epoch": 1.0626918018412976,
+      "grad_norm": 0.8125,
+      "learning_rate": 0.00016140350877192982,
+      "loss": 0.5831,
+      "step": 1818
+    },
+    {
+      "epoch": 1.063860879731112,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00016130604288499027,
+      "loss": 0.5131,
+      "step": 1820
+    },
+    {
+      "epoch": 1.0650299576209266,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001612085769980507,
+      "loss": 0.6097,
+      "step": 1822
+    },
+    {
+      "epoch": 1.066199035510741,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001611111111111111,
+      "loss": 0.5895,
+      "step": 1824
+    },
+    {
+      "epoch": 1.0673681134005553,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00016101364522417156,
+      "loss": 0.5048,
+      "step": 1826
+    },
+    {
+      "epoch": 1.0685371912903698,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00016091617933723198,
+      "loss": 0.5041,
+      "step": 1828
+    },
+    {
+      "epoch": 1.0697062691801842,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001608187134502924,
+      "loss": 0.5293,
+      "step": 1830
+    },
+    {
+      "epoch": 1.0708753470699985,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00016072124756335283,
+      "loss": 0.5054,
+      "step": 1832
+    },
+    {
+      "epoch": 1.072044424959813,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00016062378167641325,
+      "loss": 0.5345,
+      "step": 1834
+    },
+    {
+      "epoch": 1.0732135028496272,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.0001605263157894737,
+      "loss": 0.5057,
+      "step": 1836
+    },
+    {
+      "epoch": 1.0743825807394418,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00016042884990253412,
+      "loss": 0.5478,
+      "step": 1838
+    },
+    {
+      "epoch": 1.0755516586292562,
+      "grad_norm": 0.6640625,
+      "learning_rate": 0.00016033138401559454,
+      "loss": 0.6639,
+      "step": 1840
+    },
+    {
+      "epoch": 1.0767207365190705,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.000160233918128655,
+      "loss": 0.529,
+      "step": 1842
+    },
+    {
+      "epoch": 1.077889814408885,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00016013645224171542,
+      "loss": 0.5433,
+      "step": 1844
+    },
+    {
+      "epoch": 1.0790588922986994,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00016003898635477584,
+      "loss": 0.6416,
+      "step": 1846
+    },
+    {
+      "epoch": 1.0802279701885138,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00015994152046783626,
+      "loss": 0.5125,
+      "step": 1848
+    },
+    {
+      "epoch": 1.0813970480783281,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00015984405458089668,
+      "loss": 0.5199,
+      "step": 1850
+    },
+    {
+      "epoch": 1.0825661259681427,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00015974658869395713,
+      "loss": 0.5225,
+      "step": 1852
+    },
+    {
+      "epoch": 1.083735203857957,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00015964912280701756,
+      "loss": 0.4977,
+      "step": 1854
+    },
+    {
+      "epoch": 1.0849042817477714,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00015955165692007798,
+      "loss": 0.5717,
+      "step": 1856
+    },
+    {
+      "epoch": 1.0860733596375858,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001594541910331384,
+      "loss": 0.5352,
+      "step": 1858
+    },
+    {
+      "epoch": 1.0872424375274004,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00015935672514619885,
+      "loss": 0.5502,
+      "step": 1860
+    },
+    {
+      "epoch": 1.0884115154172147,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00015925925925925927,
+      "loss": 0.5091,
+      "step": 1862
+    },
+    {
+      "epoch": 1.089580593307029,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001591617933723197,
+      "loss": 0.5104,
+      "step": 1864
+    },
+    {
+      "epoch": 1.0907496711968434,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00015906432748538012,
+      "loss": 0.5178,
+      "step": 1866
+    },
+    {
+      "epoch": 1.091918749086658,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00015896686159844057,
+      "loss": 0.5414,
+      "step": 1868
+    },
+    {
+      "epoch": 1.0930878269764723,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.000158869395711501,
+      "loss": 0.4606,
+      "step": 1870
+    },
+    {
+      "epoch": 1.0942569048662867,
+      "grad_norm": 0.5,
+      "learning_rate": 0.0001587719298245614,
+      "loss": 0.5489,
+      "step": 1872
+    },
+    {
+      "epoch": 1.095425982756101,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00015867446393762183,
+      "loss": 0.4745,
+      "step": 1874
+    },
+    {
+      "epoch": 1.0965950606459156,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00015857699805068228,
+      "loss": 0.5481,
+      "step": 1876
+    },
+    {
+      "epoch": 1.09776413853573,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001584795321637427,
+      "loss": 0.5632,
+      "step": 1878
+    },
+    {
+      "epoch": 1.0989332164255443,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00015838206627680313,
+      "loss": 0.5136,
+      "step": 1880
+    },
+    {
+      "epoch": 1.1001022943153587,
+      "grad_norm": 0.61328125,
+      "learning_rate": 0.00015828460038986355,
+      "loss": 0.51,
+      "step": 1882
+    },
+    {
+      "epoch": 1.1012713722051732,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.000158187134502924,
+      "loss": 0.5066,
+      "step": 1884
+    },
+    {
+      "epoch": 1.1024404500949876,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00015808966861598442,
+      "loss": 0.6068,
+      "step": 1886
+    },
+    {
+      "epoch": 1.103609527984802,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00015799220272904484,
+      "loss": 0.5732,
+      "step": 1888
+    },
+    {
+      "epoch": 1.1047786058746163,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00015789473684210527,
+      "loss": 0.4733,
+      "step": 1890
+    },
+    {
+      "epoch": 1.1059476837644309,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001577972709551657,
+      "loss": 0.5106,
+      "step": 1892
+    },
+    {
+      "epoch": 1.1071167616542452,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00015769980506822614,
+      "loss": 0.545,
+      "step": 1894
+    },
+    {
+      "epoch": 1.1082858395440596,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00015760233918128656,
+      "loss": 0.5188,
+      "step": 1896
+    },
+    {
+      "epoch": 1.1094549174338741,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00015750487329434698,
+      "loss": 0.5393,
+      "step": 1898
+    },
+    {
+      "epoch": 1.1106239953236885,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00015740740740740743,
+      "loss": 0.5315,
+      "step": 1900
+    },
+    {
+      "epoch": 1.1117930732135028,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00015730994152046785,
+      "loss": 0.5042,
+      "step": 1902
+    },
+    {
+      "epoch": 1.1129621511033172,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00015721247563352828,
+      "loss": 0.5491,
+      "step": 1904
+    },
+    {
+      "epoch": 1.1141312289931316,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001571150097465887,
+      "loss": 0.5612,
+      "step": 1906
+    },
+    {
+      "epoch": 1.1153003068829461,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00015701754385964912,
+      "loss": 0.5288,
+      "step": 1908
+    },
+    {
+      "epoch": 1.1164693847727605,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00015692007797270957,
+      "loss": 0.5249,
+      "step": 1910
+    },
+    {
+      "epoch": 1.1176384626625748,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00015682261208577,
+      "loss": 0.5217,
+      "step": 1912
+    },
+    {
+      "epoch": 1.1188075405523894,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00015672514619883041,
+      "loss": 0.5379,
+      "step": 1914
+    },
+    {
+      "epoch": 1.1199766184422038,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00015662768031189086,
+      "loss": 0.5196,
+      "step": 1916
+    },
+    {
+      "epoch": 1.121145696332018,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00015653021442495129,
+      "loss": 0.5316,
+      "step": 1918
+    },
+    {
+      "epoch": 1.1223147742218325,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0001564327485380117,
+      "loss": 0.5255,
+      "step": 1920
+    },
+    {
+      "epoch": 1.123483852111647,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00015633528265107213,
+      "loss": 0.5311,
+      "step": 1922
+    },
+    {
+      "epoch": 1.1246529300014614,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00015623781676413255,
+      "loss": 0.5729,
+      "step": 1924
+    },
+    {
+      "epoch": 1.1258220078912757,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00015614035087719297,
+      "loss": 0.495,
+      "step": 1926
+    },
+    {
+      "epoch": 1.12699108578109,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00015604288499025342,
+      "loss": 0.5984,
+      "step": 1928
+    },
+    {
+      "epoch": 1.1281601636709047,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00015594541910331385,
+      "loss": 0.5055,
+      "step": 1930
+    },
+    {
+      "epoch": 1.129329241560719,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001558479532163743,
+      "loss": 0.5671,
+      "step": 1932
+    },
+    {
+      "epoch": 1.1304983194505334,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00015575048732943472,
+      "loss": 0.5577,
+      "step": 1934
+    },
+    {
+      "epoch": 1.1316673973403477,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00015565302144249514,
+      "loss": 0.5637,
+      "step": 1936
+    },
+    {
+      "epoch": 1.1328364752301623,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00015555555555555556,
+      "loss": 0.5337,
+      "step": 1938
+    },
+    {
+      "epoch": 1.1340055531199766,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00015545808966861599,
+      "loss": 0.5213,
+      "step": 1940
+    },
+    {
+      "epoch": 1.135174631009791,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001553606237816764,
+      "loss": 0.493,
+      "step": 1942
+    },
+    {
+      "epoch": 1.1363437088996053,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00015526315789473686,
+      "loss": 0.5246,
+      "step": 1944
+    },
+    {
+      "epoch": 1.13751278678942,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00015516569200779728,
+      "loss": 0.4862,
+      "step": 1946
+    },
+    {
+      "epoch": 1.1386818646792343,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00015506822612085773,
+      "loss": 0.6714,
+      "step": 1948
+    },
+    {
+      "epoch": 1.1398509425690486,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00015497076023391815,
+      "loss": 0.57,
+      "step": 1950
+    },
+    {
+      "epoch": 1.1410200204588632,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00015487329434697857,
+      "loss": 0.488,
+      "step": 1952
+    },
+    {
+      "epoch": 1.1421890983486775,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.000154775828460039,
+      "loss": 0.5453,
+      "step": 1954
+    },
+    {
+      "epoch": 1.143358176238492,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00015467836257309942,
+      "loss": 0.5405,
+      "step": 1956
+    },
+    {
+      "epoch": 1.1445272541283062,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00015458089668615984,
+      "loss": 0.4999,
+      "step": 1958
+    },
+    {
+      "epoch": 1.1456963320181206,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.00015448343079922026,
+      "loss": 0.5213,
+      "step": 1960
+    },
+    {
+      "epoch": 1.1468654099079352,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001543859649122807,
+      "loss": 0.5285,
+      "step": 1962
+    },
+    {
+      "epoch": 1.1480344877977495,
+      "grad_norm": 3.71875,
+      "learning_rate": 0.00015428849902534116,
+      "loss": 0.6403,
+      "step": 1964
+    },
+    {
+      "epoch": 1.1492035656875639,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00015419103313840158,
+      "loss": 0.5461,
+      "step": 1966
+    },
+    {
+      "epoch": 1.1503726435773785,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.000154093567251462,
+      "loss": 0.581,
+      "step": 1968
+    },
+    {
+      "epoch": 1.1515417214671928,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00015399610136452243,
+      "loss": 0.5225,
+      "step": 1970
+    },
+    {
+      "epoch": 1.1527107993570072,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00015389863547758285,
+      "loss": 0.5275,
+      "step": 1972
+    },
+    {
+      "epoch": 1.1538798772468215,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00015380116959064327,
+      "loss": 0.5758,
+      "step": 1974
+    },
+    {
+      "epoch": 1.1550489551366359,
+      "grad_norm": 1.3984375,
+      "learning_rate": 0.0001537037037037037,
+      "loss": 0.7684,
+      "step": 1976
+    },
+    {
+      "epoch": 1.1562180330264504,
+      "grad_norm": 1.0234375,
+      "learning_rate": 0.00015360623781676414,
+      "loss": 0.6212,
+      "step": 1978
+    },
+    {
+      "epoch": 1.1573871109162648,
+      "grad_norm": 1.09375,
+      "learning_rate": 0.00015350877192982457,
+      "loss": 0.5519,
+      "step": 1980
+    },
+    {
+      "epoch": 1.1585561888060791,
+      "grad_norm": 2.21875,
+      "learning_rate": 0.00015341130604288502,
+      "loss": 0.693,
+      "step": 1982
+    },
+    {
+      "epoch": 1.1597252666958937,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00015331384015594544,
+      "loss": 0.5857,
+      "step": 1984
+    },
+    {
+      "epoch": 1.160894344585708,
+      "grad_norm": 0.65625,
+      "learning_rate": 0.00015321637426900586,
+      "loss": 0.5599,
+      "step": 1986
+    },
+    {
+      "epoch": 1.1620634224755224,
+      "grad_norm": 0.84375,
+      "learning_rate": 0.00015311890838206628,
+      "loss": 0.5946,
+      "step": 1988
+    },
+    {
+      "epoch": 1.1632325003653368,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.0001530214424951267,
+      "loss": 0.5712,
+      "step": 1990
+    },
+    {
+      "epoch": 1.1644015782551513,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00015292397660818713,
+      "loss": 0.5825,
+      "step": 1992
+    },
+    {
+      "epoch": 1.1655706561449657,
+      "grad_norm": 0.89453125,
+      "learning_rate": 0.00015282651072124755,
+      "loss": 0.5646,
+      "step": 1994
+    },
+    {
+      "epoch": 1.16673973403478,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.000152729044834308,
+      "loss": 0.5377,
+      "step": 1996
+    },
+    {
+      "epoch": 1.1679088119245944,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00015263157894736845,
+      "loss": 0.5101,
+      "step": 1998
+    },
+    {
+      "epoch": 1.169077889814409,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00015253411306042887,
+      "loss": 0.4811,
+      "step": 2000
+    },
+    {
+      "epoch": 1.1702469677042233,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0001524366471734893,
+      "loss": 0.5085,
+      "step": 2002
+    },
+    {
+      "epoch": 1.1714160455940377,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00015233918128654972,
+      "loss": 0.5708,
+      "step": 2004
+    },
+    {
+      "epoch": 1.172585123483852,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00015224171539961014,
+      "loss": 0.5335,
+      "step": 2006
+    },
+    {
+      "epoch": 1.1737542013736666,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00015214424951267056,
+      "loss": 0.6306,
+      "step": 2008
+    },
+    {
+      "epoch": 1.174923279263481,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00015204678362573098,
+      "loss": 0.5656,
+      "step": 2010
+    },
+    {
+      "epoch": 1.1760923571532953,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00015194931773879143,
+      "loss": 0.5031,
+      "step": 2012
+    },
+    {
+      "epoch": 1.1772614350431096,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00015185185185185185,
+      "loss": 0.5342,
+      "step": 2014
+    },
+    {
+      "epoch": 1.1784305129329242,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001517543859649123,
+      "loss": 0.5409,
+      "step": 2016
+    },
+    {
+      "epoch": 1.1795995908227386,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00015165692007797273,
+      "loss": 0.5312,
+      "step": 2018
+    },
+    {
+      "epoch": 1.180768668712553,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00015155945419103315,
+      "loss": 0.5754,
+      "step": 2020
+    },
+    {
+      "epoch": 1.1819377466023675,
+      "grad_norm": 1.015625,
+      "learning_rate": 0.00015146198830409357,
+      "loss": 0.648,
+      "step": 2022
+    },
+    {
+      "epoch": 1.1831068244921819,
+      "grad_norm": 0.82421875,
+      "learning_rate": 0.000151364522417154,
+      "loss": 0.6954,
+      "step": 2024
+    },
+    {
+      "epoch": 1.1842759023819962,
+      "grad_norm": 0.78515625,
+      "learning_rate": 0.00015126705653021442,
+      "loss": 0.5343,
+      "step": 2026
+    },
+    {
+      "epoch": 1.1854449802718106,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00015116959064327486,
+      "loss": 0.4973,
+      "step": 2028
+    },
+    {
+      "epoch": 1.186614058161625,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.0001510721247563353,
+      "loss": 0.5654,
+      "step": 2030
+    },
+    {
+      "epoch": 1.1877831360514395,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00015097465886939574,
+      "loss": 0.557,
+      "step": 2032
+    },
+    {
+      "epoch": 1.1889522139412538,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00015087719298245616,
+      "loss": 0.5554,
+      "step": 2034
+    },
+    {
+      "epoch": 1.1901212918310682,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00015077972709551658,
+      "loss": 0.5727,
+      "step": 2036
+    },
+    {
+      "epoch": 1.1912903697208828,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.000150682261208577,
+      "loss": 0.5447,
+      "step": 2038
+    },
+    {
+      "epoch": 1.192459447610697,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00015058479532163743,
+      "loss": 0.5653,
+      "step": 2040
+    },
+    {
+      "epoch": 1.1936285255005115,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00015048732943469785,
+      "loss": 0.5057,
+      "step": 2042
+    },
+    {
+      "epoch": 1.1947976033903258,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0001503898635477583,
+      "loss": 0.5383,
+      "step": 2044
+    },
+    {
+      "epoch": 1.1959666812801402,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00015029239766081872,
+      "loss": 0.5763,
+      "step": 2046
+    },
+    {
+      "epoch": 1.1971357591699547,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00015019493177387914,
+      "loss": 0.5312,
+      "step": 2048
+    },
+    {
+      "epoch": 1.198304837059769,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001500974658869396,
+      "loss": 0.5504,
+      "step": 2050
+    },
+    {
+      "epoch": 1.1994739149495834,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.5234,
+      "step": 2052
+    },
+    {
+      "epoch": 1.200642992839398,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00014990253411306044,
+      "loss": 0.5458,
+      "step": 2054
+    },
+    {
+      "epoch": 1.2018120707292124,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00014980506822612086,
+      "loss": 0.5399,
+      "step": 2056
+    },
+    {
+      "epoch": 1.2029811486190267,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00014970760233918128,
+      "loss": 0.5323,
+      "step": 2058
+    },
+    {
+      "epoch": 1.204150226508841,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001496101364522417,
+      "loss": 0.5314,
+      "step": 2060
+    },
+    {
+      "epoch": 1.2053193043986556,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00014951267056530215,
+      "loss": 0.4901,
+      "step": 2062
+    },
+    {
+      "epoch": 1.20648838228847,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00014941520467836257,
+      "loss": 0.5305,
+      "step": 2064
+    },
+    {
+      "epoch": 1.2076574601782843,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00014931773879142302,
+      "loss": 0.4642,
+      "step": 2066
+    },
+    {
+      "epoch": 1.2088265380680987,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00014922027290448345,
+      "loss": 0.508,
+      "step": 2068
+    },
+    {
+      "epoch": 1.2099956159579133,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00014912280701754387,
+      "loss": 0.5329,
+      "step": 2070
+    },
+    {
+      "epoch": 1.2111646938477276,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001490253411306043,
+      "loss": 0.5502,
+      "step": 2072
+    },
+    {
+      "epoch": 1.212333771737542,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001489278752436647,
+      "loss": 0.5335,
+      "step": 2074
+    },
+    {
+      "epoch": 1.2135028496273563,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00014883040935672514,
+      "loss": 0.5298,
+      "step": 2076
+    },
+    {
+      "epoch": 1.214671927517171,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00014873294346978559,
+      "loss": 0.5271,
+      "step": 2078
+    },
+    {
+      "epoch": 1.2158410054069853,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.000148635477582846,
+      "loss": 0.5326,
+      "step": 2080
+    },
+    {
+      "epoch": 1.2170100832967996,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00014853801169590643,
+      "loss": 0.5556,
+      "step": 2082
+    },
+    {
+      "epoch": 1.218179161186614,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00014844054580896688,
+      "loss": 0.4727,
+      "step": 2084
+    },
+    {
+      "epoch": 1.2193482390764285,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001483430799220273,
+      "loss": 0.5172,
+      "step": 2086
+    },
+    {
+      "epoch": 1.2205173169662429,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00014824561403508772,
+      "loss": 0.5698,
+      "step": 2088
+    },
+    {
+      "epoch": 1.2216863948560572,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00014814814814814815,
+      "loss": 0.6561,
+      "step": 2090
+    },
+    {
+      "epoch": 1.2228554727458718,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00014805068226120857,
+      "loss": 0.5546,
+      "step": 2092
+    },
+    {
+      "epoch": 1.2240245506356862,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00014795321637426902,
+      "loss": 0.5504,
+      "step": 2094
+    },
+    {
+      "epoch": 1.2251936285255005,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00014785575048732944,
+      "loss": 0.5061,
+      "step": 2096
+    },
+    {
+      "epoch": 1.2263627064153149,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00014775828460038986,
+      "loss": 0.5257,
+      "step": 2098
+    },
+    {
+      "epoch": 1.2275317843051292,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0001476608187134503,
+      "loss": 0.5323,
+      "step": 2100
+    },
+    {
+      "epoch": 1.2287008621949438,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00014756335282651073,
+      "loss": 0.5322,
+      "step": 2102
+    },
+    {
+      "epoch": 1.2298699400847581,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00014746588693957116,
+      "loss": 0.5758,
+      "step": 2104
+    },
+    {
+      "epoch": 1.2310390179745725,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00014736842105263158,
+      "loss": 0.5396,
+      "step": 2106
+    },
+    {
+      "epoch": 1.232208095864387,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.000147270955165692,
+      "loss": 0.5596,
+      "step": 2108
+    },
+    {
+      "epoch": 1.2333771737542014,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00014717348927875245,
+      "loss": 0.5493,
+      "step": 2110
+    },
+    {
+      "epoch": 1.2345462516440158,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00014707602339181287,
+      "loss": 0.5155,
+      "step": 2112
+    },
+    {
+      "epoch": 1.2357153295338301,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.0001469785575048733,
+      "loss": 0.5305,
+      "step": 2114
+    },
+    {
+      "epoch": 1.2368844074236447,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00014688109161793372,
+      "loss": 0.6194,
+      "step": 2116
+    },
+    {
+      "epoch": 1.238053485313459,
+      "grad_norm": 0.77734375,
+      "learning_rate": 0.00014678362573099417,
+      "loss": 0.5856,
+      "step": 2118
+    },
+    {
+      "epoch": 1.2392225632032734,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.0001466861598440546,
+      "loss": 0.5392,
+      "step": 2120
+    },
+    {
+      "epoch": 1.2403916410930877,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.000146588693957115,
+      "loss": 0.5132,
+      "step": 2122
+    },
+    {
+      "epoch": 1.2415607189829023,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00014649122807017543,
+      "loss": 0.5524,
+      "step": 2124
+    },
+    {
+      "epoch": 1.2427297968727167,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00014639376218323588,
+      "loss": 0.5383,
+      "step": 2126
+    },
+    {
+      "epoch": 1.243898874762531,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.0001462962962962963,
+      "loss": 0.548,
+      "step": 2128
+    },
+    {
+      "epoch": 1.2450679526523454,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00014619883040935673,
+      "loss": 0.5194,
+      "step": 2130
+    },
+    {
+      "epoch": 1.24623703054216,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00014610136452241715,
+      "loss": 0.5007,
+      "step": 2132
+    },
+    {
+      "epoch": 1.2474061084319743,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001460038986354776,
+      "loss": 0.5092,
+      "step": 2134
+    },
+    {
+      "epoch": 1.2485751863217887,
+      "grad_norm": 0.94921875,
+      "learning_rate": 0.00014590643274853802,
+      "loss": 0.5948,
+      "step": 2136
+    },
+    {
+      "epoch": 1.249744264211603,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00014580896686159844,
+      "loss": 0.5545,
+      "step": 2138
+    },
+    {
+      "epoch": 1.2509133421014176,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00014571150097465887,
+      "loss": 0.5342,
+      "step": 2140
+    },
+    {
+      "epoch": 1.252082419991232,
+      "grad_norm": 1.2890625,
+      "learning_rate": 0.00014561403508771932,
+      "loss": 0.616,
+      "step": 2142
+    },
+    {
+      "epoch": 1.2532514978810463,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00014551656920077974,
+      "loss": 0.5408,
+      "step": 2144
+    },
+    {
+      "epoch": 1.2544205757708609,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00014541910331384016,
+      "loss": 0.5604,
+      "step": 2146
+    },
+    {
+      "epoch": 1.2555896536606752,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00014532163742690058,
+      "loss": 0.5168,
+      "step": 2148
+    },
+    {
+      "epoch": 1.2567587315504896,
+      "grad_norm": 0.59375,
+      "learning_rate": 0.000145224171539961,
+      "loss": 0.5451,
+      "step": 2150
+    },
+    {
+      "epoch": 1.257927809440304,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00014512670565302145,
+      "loss": 0.5365,
+      "step": 2152
+    },
+    {
+      "epoch": 1.2590968873301183,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00014502923976608188,
+      "loss": 0.5516,
+      "step": 2154
+    },
+    {
+      "epoch": 1.2602659652199328,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001449317738791423,
+      "loss": 0.6553,
+      "step": 2156
+    },
+    {
+      "epoch": 1.2614350431097472,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00014483430799220275,
+      "loss": 0.5779,
+      "step": 2158
+    },
+    {
+      "epoch": 1.2626041209995615,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00014473684210526317,
+      "loss": 0.5469,
+      "step": 2160
+    },
+    {
+      "epoch": 1.2637731988893761,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001446393762183236,
+      "loss": 0.5338,
+      "step": 2162
+    },
+    {
+      "epoch": 1.2649422767791905,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00014454191033138402,
+      "loss": 0.4948,
+      "step": 2164
+    },
+    {
+      "epoch": 1.2661113546690048,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00014444444444444444,
+      "loss": 0.5273,
+      "step": 2166
+    },
+    {
+      "epoch": 1.2672804325588192,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0001443469785575049,
+      "loss": 0.5314,
+      "step": 2168
+    },
+    {
+      "epoch": 1.2684495104486335,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.0001442495126705653,
+      "loss": 0.5253,
+      "step": 2170
+    },
+    {
+      "epoch": 1.269618588338448,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00014415204678362573,
+      "loss": 0.6898,
+      "step": 2172
+    },
+    {
+      "epoch": 1.2707876662282624,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00014405458089668618,
+      "loss": 0.4904,
+      "step": 2174
+    },
+    {
+      "epoch": 1.2719567441180768,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.0001439571150097466,
+      "loss": 0.5021,
+      "step": 2176
+    },
+    {
+      "epoch": 1.2731258220078914,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00014385964912280703,
+      "loss": 0.5574,
+      "step": 2178
+    },
+    {
+      "epoch": 1.2742948998977057,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00014376218323586745,
+      "loss": 0.563,
+      "step": 2180
+    },
+    {
+      "epoch": 1.27546397778752,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00014366471734892787,
+      "loss": 0.5243,
+      "step": 2182
+    },
+    {
+      "epoch": 1.2766330556773344,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001435672514619883,
+      "loss": 0.596,
+      "step": 2184
+    },
+    {
+      "epoch": 1.2778021335671488,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.00014346978557504874,
+      "loss": 0.5895,
+      "step": 2186
+    },
+    {
+      "epoch": 1.2789712114569634,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00014337231968810916,
+      "loss": 0.4975,
+      "step": 2188
+    },
+    {
+      "epoch": 1.2801402893467777,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00014327485380116961,
+      "loss": 0.5065,
+      "step": 2190
+    },
+    {
+      "epoch": 1.281309367236592,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00014317738791423004,
+      "loss": 0.4677,
+      "step": 2192
+    },
+    {
+      "epoch": 1.2824784451264066,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00014307992202729046,
+      "loss": 0.5402,
+      "step": 2194
+    },
+    {
+      "epoch": 1.283647523016221,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00014298245614035088,
+      "loss": 0.5598,
+      "step": 2196
+    },
+    {
+      "epoch": 1.2848166009060353,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0001428849902534113,
+      "loss": 0.5707,
+      "step": 2198
+    },
+    {
+      "epoch": 1.28598567879585,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00014278752436647173,
+      "loss": 0.5603,
+      "step": 2200
+    },
+    {
+      "epoch": 1.2871547566856643,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00014269005847953217,
+      "loss": 0.494,
+      "step": 2202
+    },
+    {
+      "epoch": 1.2883238345754786,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001425925925925926,
+      "loss": 0.5525,
+      "step": 2204
+    },
+    {
+      "epoch": 1.289492912465293,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00014249512670565305,
+      "loss": 0.5143,
+      "step": 2206
+    },
+    {
+      "epoch": 1.2906619903551073,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00014239766081871347,
+      "loss": 0.5176,
+      "step": 2208
+    },
+    {
+      "epoch": 1.2918310682449219,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001423001949317739,
+      "loss": 0.5043,
+      "step": 2210
+    },
+    {
+      "epoch": 1.2930001461347362,
+      "grad_norm": 0.6953125,
+      "learning_rate": 0.0001422027290448343,
+      "loss": 0.571,
+      "step": 2212
+    },
+    {
+      "epoch": 1.2941692240245506,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00014210526315789474,
+      "loss": 0.5369,
+      "step": 2214
+    },
+    {
+      "epoch": 1.2953383019143652,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00014200779727095516,
+      "loss": 0.5103,
+      "step": 2216
+    },
+    {
+      "epoch": 1.2965073798041795,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00014191033138401558,
+      "loss": 0.4751,
+      "step": 2218
+    },
+    {
+      "epoch": 1.2976764576939939,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00014181286549707603,
+      "loss": 0.5009,
+      "step": 2220
+    },
+    {
+      "epoch": 1.2988455355838082,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00014171539961013648,
+      "loss": 0.5003,
+      "step": 2222
+    },
+    {
+      "epoch": 1.3000146134736226,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0001416179337231969,
+      "loss": 0.4975,
+      "step": 2224
+    },
+    {
+      "epoch": 1.3011836913634371,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00014152046783625732,
+      "loss": 0.4553,
+      "step": 2226
+    },
+    {
+      "epoch": 1.3023527692532515,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00014142300194931775,
+      "loss": 0.5384,
+      "step": 2228
+    },
+    {
+      "epoch": 1.3035218471430658,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00014132553606237817,
+      "loss": 0.4887,
+      "step": 2230
+    },
+    {
+      "epoch": 1.3046909250328804,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001412280701754386,
+      "loss": 0.4916,
+      "step": 2232
+    },
+    {
+      "epoch": 1.3058600029226948,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.000141130604288499,
+      "loss": 0.5218,
+      "step": 2234
+    },
+    {
+      "epoch": 1.3070290808125091,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00014103313840155946,
+      "loss": 0.5185,
+      "step": 2236
+    },
+    {
+      "epoch": 1.3081981587023235,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.0001409356725146199,
+      "loss": 0.5197,
+      "step": 2238
+    },
+    {
+      "epoch": 1.3093672365921378,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00014083820662768033,
+      "loss": 0.4974,
+      "step": 2240
+    },
+    {
+      "epoch": 1.3105363144819524,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00014074074074074076,
+      "loss": 0.5262,
+      "step": 2242
+    },
+    {
+      "epoch": 1.3117053923717668,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00014064327485380118,
+      "loss": 0.5682,
+      "step": 2244
+    },
+    {
+      "epoch": 1.312874470261581,
+      "grad_norm": 1.21875,
+      "learning_rate": 0.0001405458089668616,
+      "loss": 0.5945,
+      "step": 2246
+    },
+    {
+      "epoch": 1.3140435481513957,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00014044834307992202,
+      "loss": 0.5376,
+      "step": 2248
+    },
+    {
+      "epoch": 1.31521262604121,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00014035087719298245,
+      "loss": 0.5386,
+      "step": 2250
+    },
+    {
+      "epoch": 1.3163817039310244,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00014025341130604287,
+      "loss": 0.5174,
+      "step": 2252
+    },
+    {
+      "epoch": 1.3175507818208387,
+      "grad_norm": 0.671875,
+      "learning_rate": 0.00014015594541910332,
+      "loss": 0.5459,
+      "step": 2254
+    },
+    {
+      "epoch": 1.318719859710653,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00014005847953216377,
+      "loss": 0.5424,
+      "step": 2256
+    },
+    {
+      "epoch": 1.3198889376004677,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001399610136452242,
+      "loss": 0.5088,
+      "step": 2258
+    },
+    {
+      "epoch": 1.321058015490282,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0001398635477582846,
+      "loss": 0.5042,
+      "step": 2260
+    },
+    {
+      "epoch": 1.3222270933800964,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00013976608187134503,
+      "loss": 0.5704,
+      "step": 2262
+    },
+    {
+      "epoch": 1.323396171269911,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00013966861598440546,
+      "loss": 0.5154,
+      "step": 2264
+    },
+    {
+      "epoch": 1.3245652491597253,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00013957115009746588,
+      "loss": 0.5377,
+      "step": 2266
+    },
+    {
+      "epoch": 1.3257343270495396,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.0001394736842105263,
+      "loss": 0.5435,
+      "step": 2268
+    },
+    {
+      "epoch": 1.3269034049393542,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00013937621832358675,
+      "loss": 0.4831,
+      "step": 2270
+    },
+    {
+      "epoch": 1.3280724828291686,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.0001392787524366472,
+      "loss": 0.4896,
+      "step": 2272
+    },
+    {
+      "epoch": 1.329241560718983,
+      "grad_norm": 2.96875,
+      "learning_rate": 0.00013918128654970762,
+      "loss": 0.6227,
+      "step": 2274
+    },
+    {
+      "epoch": 1.3304106386087973,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00013908382066276804,
+      "loss": 0.5034,
+      "step": 2276
+    },
+    {
+      "epoch": 1.3315797164986116,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00013898635477582847,
+      "loss": 0.5445,
+      "step": 2278
+    },
+    {
+      "epoch": 1.3327487943884262,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.0001388888888888889,
+      "loss": 0.5254,
+      "step": 2280
+    },
+    {
+      "epoch": 1.3339178722782405,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001387914230019493,
+      "loss": 0.5563,
+      "step": 2282
+    },
+    {
+      "epoch": 1.335086950168055,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00013869395711500973,
+      "loss": 0.5262,
+      "step": 2284
+    },
+    {
+      "epoch": 1.3362560280578695,
+      "grad_norm": 0.431640625,
+      "learning_rate": 0.00013859649122807018,
+      "loss": 0.4994,
+      "step": 2286
+    },
+    {
+      "epoch": 1.3374251059476838,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0001384990253411306,
+      "loss": 0.5218,
+      "step": 2288
+    },
+    {
+      "epoch": 1.3385941838374982,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00013840155945419105,
+      "loss": 0.5622,
+      "step": 2290
+    },
+    {
+      "epoch": 1.3397632617273125,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00013830409356725148,
+      "loss": 0.5173,
+      "step": 2292
+    },
+    {
+      "epoch": 1.3409323396171269,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001382066276803119,
+      "loss": 0.496,
+      "step": 2294
+    },
+    {
+      "epoch": 1.3421014175069415,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00013810916179337232,
+      "loss": 0.5524,
+      "step": 2296
+    },
+    {
+      "epoch": 1.3432704953967558,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00013801169590643274,
+      "loss": 0.4955,
+      "step": 2298
+    },
+    {
+      "epoch": 1.3444395732865702,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00013791423001949317,
+      "loss": 0.5226,
+      "step": 2300
+    },
+    {
+      "epoch": 1.3456086511763847,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.00013781676413255362,
+      "loss": 0.5202,
+      "step": 2302
+    },
+    {
+      "epoch": 1.346777729066199,
+      "grad_norm": 0.96484375,
+      "learning_rate": 0.00013771929824561404,
+      "loss": 0.801,
+      "step": 2304
+    },
+    {
+      "epoch": 1.3479468069560134,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.0001376218323586745,
+      "loss": 0.5402,
+      "step": 2306
+    },
+    {
+      "epoch": 1.3491158848458278,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001375243664717349,
+      "loss": 0.5445,
+      "step": 2308
+    },
+    {
+      "epoch": 1.3502849627356421,
+      "grad_norm": 0.6484375,
+      "learning_rate": 0.00013742690058479533,
+      "loss": 0.5235,
+      "step": 2310
+    },
+    {
+      "epoch": 1.3514540406254567,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00013732943469785575,
+      "loss": 0.5099,
+      "step": 2312
+    },
+    {
+      "epoch": 1.352623118515271,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00013723196881091618,
+      "loss": 0.4943,
+      "step": 2314
+    },
+    {
+      "epoch": 1.3537921964050854,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001371345029239766,
+      "loss": 0.5101,
+      "step": 2316
+    },
+    {
+      "epoch": 1.3549612742949,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00013703703703703705,
+      "loss": 0.5452,
+      "step": 2318
+    },
+    {
+      "epoch": 1.3561303521847143,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00013693957115009747,
+      "loss": 0.532,
+      "step": 2320
+    },
+    {
+      "epoch": 1.3572994300745287,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0001368421052631579,
+      "loss": 0.5339,
+      "step": 2322
+    },
+    {
+      "epoch": 1.358468507964343,
+      "grad_norm": 0.86328125,
+      "learning_rate": 0.00013674463937621834,
+      "loss": 0.6001,
+      "step": 2324
+    },
+    {
+      "epoch": 1.3596375858541576,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.00013664717348927876,
+      "loss": 0.5809,
+      "step": 2326
+    },
+    {
+      "epoch": 1.360806663743972,
+      "grad_norm": 0.59765625,
+      "learning_rate": 0.0001365497076023392,
+      "loss": 0.4819,
+      "step": 2328
+    },
+    {
+      "epoch": 1.3619757416337863,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.0001364522417153996,
+      "loss": 0.5462,
+      "step": 2330
+    },
+    {
+      "epoch": 1.3631448195236007,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00013635477582846003,
+      "loss": 0.5619,
+      "step": 2332
+    },
+    {
+      "epoch": 1.3643138974134152,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00013625730994152048,
+      "loss": 0.514,
+      "step": 2334
+    },
+    {
+      "epoch": 1.3654829753032296,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.0001361598440545809,
+      "loss": 0.5458,
+      "step": 2336
+    },
+    {
+      "epoch": 1.366652053193044,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00013606237816764133,
+      "loss": 0.5545,
+      "step": 2338
+    },
+    {
+      "epoch": 1.3678211310828585,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00013596491228070177,
+      "loss": 0.5931,
+      "step": 2340
+    },
+    {
+      "epoch": 1.3689902089726729,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.0001358674463937622,
+      "loss": 0.553,
+      "step": 2342
+    },
+    {
+      "epoch": 1.3701592868624872,
+      "grad_norm": 0.72265625,
+      "learning_rate": 0.00013576998050682262,
+      "loss": 0.5348,
+      "step": 2344
+    },
+    {
+      "epoch": 1.3713283647523016,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00013567251461988304,
+      "loss": 0.584,
+      "step": 2346
+    },
+    {
+      "epoch": 1.372497442642116,
+      "grad_norm": 0.423828125,
+      "learning_rate": 0.00013557504873294346,
+      "loss": 0.523,
+      "step": 2348
+    },
+    {
+      "epoch": 1.3736665205319305,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.0001354775828460039,
+      "loss": 0.5684,
+      "step": 2350
+    },
+    {
+      "epoch": 1.3748355984217449,
+      "grad_norm": 0.9609375,
+      "learning_rate": 0.00013538011695906434,
+      "loss": 0.5319,
+      "step": 2352
+    },
+    {
+      "epoch": 1.3760046763115592,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00013528265107212476,
+      "loss": 0.5758,
+      "step": 2354
+    },
+    {
+      "epoch": 1.3771737542013738,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00013518518518518518,
+      "loss": 0.4881,
+      "step": 2356
+    },
+    {
+      "epoch": 1.3783428320911881,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00013508771929824563,
+      "loss": 0.5436,
+      "step": 2358
+    },
+    {
+      "epoch": 1.3795119099810025,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00013499025341130605,
+      "loss": 0.6471,
+      "step": 2360
+    },
+    {
+      "epoch": 1.3806809878708168,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00013489278752436647,
+      "loss": 0.5255,
+      "step": 2362
+    },
+    {
+      "epoch": 1.3818500657606312,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001347953216374269,
+      "loss": 0.5235,
+      "step": 2364
+    },
+    {
+      "epoch": 1.3830191436504458,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00013469785575048735,
+      "loss": 0.4981,
+      "step": 2366
+    },
+    {
+      "epoch": 1.38418822154026,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00013460038986354777,
+      "loss": 0.4869,
+      "step": 2368
+    },
+    {
+      "epoch": 1.3853572994300745,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001345029239766082,
+      "loss": 0.4993,
+      "step": 2370
+    },
+    {
+      "epoch": 1.386526377319889,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.0001344054580896686,
+      "loss": 0.4922,
+      "step": 2372
+    },
+    {
+      "epoch": 1.3876954552097034,
+      "grad_norm": 1.171875,
+      "learning_rate": 0.00013430799220272906,
+      "loss": 0.653,
+      "step": 2374
+    },
+    {
+      "epoch": 1.3888645330995177,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00013421052631578948,
+      "loss": 0.5303,
+      "step": 2376
+    },
+    {
+      "epoch": 1.390033610989332,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001341130604288499,
+      "loss": 0.5234,
+      "step": 2378
+    },
+    {
+      "epoch": 1.3912026888791464,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00013401559454191033,
+      "loss": 0.5408,
+      "step": 2380
+    },
+    {
+      "epoch": 1.392371766768961,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00013391812865497078,
+      "loss": 0.5112,
+      "step": 2382
+    },
+    {
+      "epoch": 1.3935408446587754,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0001338206627680312,
+      "loss": 0.541,
+      "step": 2384
+    },
+    {
+      "epoch": 1.3947099225485897,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00013372319688109162,
+      "loss": 0.5173,
+      "step": 2386
+    },
+    {
+      "epoch": 1.3958790004384043,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00013362573099415205,
+      "loss": 0.4972,
+      "step": 2388
+    },
+    {
+      "epoch": 1.3970480783282186,
+      "grad_norm": 0.60546875,
+      "learning_rate": 0.00013352826510721247,
+      "loss": 0.5849,
+      "step": 2390
+    },
+    {
+      "epoch": 1.398217156218033,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00013343079922027292,
+      "loss": 0.4755,
+      "step": 2392
+    },
+    {
+      "epoch": 1.3993862341078476,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00013333333333333334,
+      "loss": 0.4992,
+      "step": 2394
+    },
+    {
+      "epoch": 1.400555311997662,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00013323586744639376,
+      "loss": 0.4972,
+      "step": 2396
+    },
+    {
+      "epoch": 1.4017243898874763,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001331384015594542,
+      "loss": 0.517,
+      "step": 2398
+    },
+    {
+      "epoch": 1.4028934677772906,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00013304093567251463,
+      "loss": 0.5337,
+      "step": 2400
+    },
+    {
+      "epoch": 1.404062545667105,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00013294346978557506,
+      "loss": 0.6145,
+      "step": 2402
+    },
+    {
+      "epoch": 1.4052316235569196,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00013284600389863548,
+      "loss": 0.529,
+      "step": 2404
+    },
+    {
+      "epoch": 1.406400701446734,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.0001327485380116959,
+      "loss": 0.5566,
+      "step": 2406
+    },
+    {
+      "epoch": 1.4075697793365483,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00013265107212475635,
+      "loss": 0.5198,
+      "step": 2408
+    },
+    {
+      "epoch": 1.4087388572263628,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00013255360623781677,
+      "loss": 0.4804,
+      "step": 2410
+    },
+    {
+      "epoch": 1.4099079351161772,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001324561403508772,
+      "loss": 0.4942,
+      "step": 2412
+    },
+    {
+      "epoch": 1.4110770130059915,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00013235867446393764,
+      "loss": 0.5272,
+      "step": 2414
+    },
+    {
+      "epoch": 1.4122460908958059,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00013226120857699807,
+      "loss": 0.5051,
+      "step": 2416
+    },
+    {
+      "epoch": 1.4134151687856202,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.0001321637426900585,
+      "loss": 0.5318,
+      "step": 2418
+    },
+    {
+      "epoch": 1.4145842466754348,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001320662768031189,
+      "loss": 0.5298,
+      "step": 2420
+    },
+    {
+      "epoch": 1.4157533245652492,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00013196881091617933,
+      "loss": 0.5054,
+      "step": 2422
+    },
+    {
+      "epoch": 1.4169224024550635,
+      "grad_norm": 0.8359375,
+      "learning_rate": 0.00013187134502923976,
+      "loss": 0.5184,
+      "step": 2424
+    },
+    {
+      "epoch": 1.418091480344878,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001317738791423002,
+      "loss": 0.5057,
+      "step": 2426
+    },
+    {
+      "epoch": 1.4192605582346924,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00013167641325536063,
+      "loss": 0.5926,
+      "step": 2428
+    },
+    {
+      "epoch": 1.4204296361245068,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00013157894736842108,
+      "loss": 0.6182,
+      "step": 2430
+    },
+    {
+      "epoch": 1.4215987140143211,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.0001314814814814815,
+      "loss": 0.5316,
+      "step": 2432
+    },
+    {
+      "epoch": 1.4227677919041355,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00013138401559454192,
+      "loss": 0.5063,
+      "step": 2434
+    },
+    {
+      "epoch": 1.42393686979395,
+      "grad_norm": 0.7265625,
+      "learning_rate": 0.00013128654970760234,
+      "loss": 0.4917,
+      "step": 2436
+    },
+    {
+      "epoch": 1.4251059476837644,
+      "grad_norm": 0.5546875,
+      "learning_rate": 0.00013118908382066277,
+      "loss": 0.5486,
+      "step": 2438
+    },
+    {
+      "epoch": 1.4262750255735788,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.0001310916179337232,
+      "loss": 0.4851,
+      "step": 2440
+    },
+    {
+      "epoch": 1.4274441034633933,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00013099415204678364,
+      "loss": 0.5135,
+      "step": 2442
+    },
+    {
+      "epoch": 1.4286131813532077,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00013089668615984406,
+      "loss": 0.483,
+      "step": 2444
+    },
+    {
+      "epoch": 1.429782259243022,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.0001307992202729045,
+      "loss": 0.5652,
+      "step": 2446
+    },
+    {
+      "epoch": 1.4309513371328364,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00013070175438596493,
+      "loss": 0.5252,
+      "step": 2448
+    },
+    {
+      "epoch": 1.4321204150226508,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00013060428849902535,
+      "loss": 0.5489,
+      "step": 2450
+    },
+    {
+      "epoch": 1.4332894929124653,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00013050682261208578,
+      "loss": 0.5033,
+      "step": 2452
+    },
+    {
+      "epoch": 1.4344585708022797,
+      "grad_norm": 0.490234375,
+      "learning_rate": 0.0001304093567251462,
+      "loss": 0.5016,
+      "step": 2454
+    },
+    {
+      "epoch": 1.435627648692094,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00013031189083820662,
+      "loss": 0.4923,
+      "step": 2456
+    },
+    {
+      "epoch": 1.4367967265819086,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00013021442495126704,
+      "loss": 0.52,
+      "step": 2458
+    },
+    {
+      "epoch": 1.437965804471723,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001301169590643275,
+      "loss": 0.513,
+      "step": 2460
+    },
+    {
+      "epoch": 1.4391348823615373,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00013001949317738794,
+      "loss": 0.4825,
+      "step": 2462
+    },
+    {
+      "epoch": 1.4403039602513519,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00012992202729044836,
+      "loss": 0.54,
+      "step": 2464
+    },
+    {
+      "epoch": 1.4414730381411662,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001298245614035088,
+      "loss": 0.5213,
+      "step": 2466
+    },
+    {
+      "epoch": 1.4426421160309806,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.0001297270955165692,
+      "loss": 0.4893,
+      "step": 2468
+    },
+    {
+      "epoch": 1.443811193920795,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00012962962962962963,
+      "loss": 0.501,
+      "step": 2470
+    },
+    {
+      "epoch": 1.4449802718106093,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00012953216374269005,
+      "loss": 0.5218,
+      "step": 2472
+    },
+    {
+      "epoch": 1.4461493497004239,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00012943469785575048,
+      "loss": 0.5996,
+      "step": 2474
+    },
+    {
+      "epoch": 1.4473184275902382,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00012933723196881093,
+      "loss": 0.5197,
+      "step": 2476
+    },
+    {
+      "epoch": 1.4484875054800526,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00012923976608187135,
+      "loss": 0.5161,
+      "step": 2478
+    },
+    {
+      "epoch": 1.4496565833698671,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001291423001949318,
+      "loss": 0.5125,
+      "step": 2480
+    },
+    {
+      "epoch": 1.4508256612596815,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00012904483430799222,
+      "loss": 0.4591,
+      "step": 2482
+    },
+    {
+      "epoch": 1.4519947391494958,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00012894736842105264,
+      "loss": 0.5067,
+      "step": 2484
+    },
+    {
+      "epoch": 1.4531638170393102,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00012884990253411306,
+      "loss": 0.5034,
+      "step": 2486
+    },
+    {
+      "epoch": 1.4543328949291245,
+      "grad_norm": 0.9765625,
+      "learning_rate": 0.00012875243664717349,
+      "loss": 0.6573,
+      "step": 2488
+    },
+    {
+      "epoch": 1.4555019728189391,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.0001286549707602339,
+      "loss": 0.5001,
+      "step": 2490
+    },
+    {
+      "epoch": 1.4566710507087535,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00012855750487329433,
+      "loss": 0.4992,
+      "step": 2492
+    },
+    {
+      "epoch": 1.4578401285985678,
+      "grad_norm": 0.6875,
+      "learning_rate": 0.00012846003898635478,
+      "loss": 0.5203,
+      "step": 2494
+    },
+    {
+      "epoch": 1.4590092064883824,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00012836257309941523,
+      "loss": 0.5263,
+      "step": 2496
+    },
+    {
+      "epoch": 1.4601782843781967,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00012826510721247565,
+      "loss": 0.5118,
+      "step": 2498
+    },
+    {
+      "epoch": 1.461347362268011,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.00012816764132553607,
+      "loss": 0.4759,
+      "step": 2500
+    },
+    {
+      "epoch": 1.4625164401578254,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001280701754385965,
+      "loss": 0.5321,
+      "step": 2502
+    },
+    {
+      "epoch": 1.4636855180476398,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00012797270955165692,
+      "loss": 0.4693,
+      "step": 2504
+    },
+    {
+      "epoch": 1.4648545959374544,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00012787524366471734,
+      "loss": 0.4409,
+      "step": 2506
+    },
+    {
+      "epoch": 1.4660236738272687,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00012777777777777776,
+      "loss": 0.4841,
+      "step": 2508
+    },
+    {
+      "epoch": 1.467192751717083,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001276803118908382,
+      "loss": 0.482,
+      "step": 2510
+    },
+    {
+      "epoch": 1.4683618296068977,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00012758284600389864,
+      "loss": 0.5053,
+      "step": 2512
+    },
+    {
+      "epoch": 1.469530907496712,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00012748538011695908,
+      "loss": 0.4906,
+      "step": 2514
+    },
+    {
+      "epoch": 1.4706999853865264,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0001273879142300195,
+      "loss": 0.5435,
+      "step": 2516
+    },
+    {
+      "epoch": 1.4718690632763407,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00012729044834307993,
+      "loss": 0.5278,
+      "step": 2518
+    },
+    {
+      "epoch": 1.473038141166155,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00012719298245614035,
+      "loss": 0.4676,
+      "step": 2520
+    },
+    {
+      "epoch": 1.4742072190559696,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00012709551656920077,
+      "loss": 0.4983,
+      "step": 2522
+    },
+    {
+      "epoch": 1.475376296945784,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.0001269980506822612,
+      "loss": 0.5437,
+      "step": 2524
+    },
+    {
+      "epoch": 1.4765453748355983,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00012690058479532165,
+      "loss": 0.4985,
+      "step": 2526
+    },
+    {
+      "epoch": 1.477714452725413,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00012680311890838207,
+      "loss": 0.5014,
+      "step": 2528
+    },
+    {
+      "epoch": 1.4788835306152273,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00012670565302144252,
+      "loss": 0.4793,
+      "step": 2530
+    },
+    {
+      "epoch": 1.4800526085050416,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00012660818713450294,
+      "loss": 0.5015,
+      "step": 2532
+    },
+    {
+      "epoch": 1.4812216863948562,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.00012651072124756336,
+      "loss": 0.5777,
+      "step": 2534
+    },
+    {
+      "epoch": 1.4823907642846705,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00012641325536062378,
+      "loss": 0.5194,
+      "step": 2536
+    },
+    {
+      "epoch": 1.4835598421744849,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001263157894736842,
+      "loss": 0.5067,
+      "step": 2538
+    },
+    {
+      "epoch": 1.4847289200642992,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00012621832358674463,
+      "loss": 0.4664,
+      "step": 2540
+    },
+    {
+      "epoch": 1.4858979979541136,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00012612085769980508,
+      "loss": 0.5486,
+      "step": 2542
+    },
+    {
+      "epoch": 1.4870670758439282,
+      "grad_norm": 0.97265625,
+      "learning_rate": 0.0001260233918128655,
+      "loss": 0.4842,
+      "step": 2544
+    },
+    {
+      "epoch": 1.4882361537337425,
+      "grad_norm": 0.486328125,
+      "learning_rate": 0.00012592592592592592,
+      "loss": 0.4884,
+      "step": 2546
+    },
+    {
+      "epoch": 1.4894052316235569,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00012582846003898637,
+      "loss": 0.5559,
+      "step": 2548
+    },
+    {
+      "epoch": 1.4905743095133714,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.0001257309941520468,
+      "loss": 0.5301,
+      "step": 2550
+    },
+    {
+      "epoch": 1.4917433874031858,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00012563352826510722,
+      "loss": 0.5174,
+      "step": 2552
+    },
+    {
+      "epoch": 1.4929124652930001,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00012553606237816764,
+      "loss": 0.4971,
+      "step": 2554
+    },
+    {
+      "epoch": 1.4940815431828145,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00012543859649122806,
+      "loss": 0.4853,
+      "step": 2556
+    },
+    {
+      "epoch": 1.4952506210726288,
+      "grad_norm": 0.578125,
+      "learning_rate": 0.0001253411306042885,
+      "loss": 0.4976,
+      "step": 2558
+    },
+    {
+      "epoch": 1.4964196989624434,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00012524366471734893,
+      "loss": 0.4784,
+      "step": 2560
+    },
+    {
+      "epoch": 1.4975887768522578,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00012514619883040936,
+      "loss": 0.4954,
+      "step": 2562
+    },
+    {
+      "epoch": 1.4987578547420721,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.0001250487329434698,
+      "loss": 0.4958,
+      "step": 2564
+    },
+    {
+      "epoch": 1.4999269326318867,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00012495126705653023,
+      "loss": 0.4747,
+      "step": 2566
+    },
+    {
+      "epoch": 1.501096010521701,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00012485380116959065,
+      "loss": 0.4837,
+      "step": 2568
+    },
+    {
+      "epoch": 1.5022650884115154,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00012475633528265107,
+      "loss": 0.5141,
+      "step": 2570
+    },
+    {
+      "epoch": 1.50343416630133,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001246588693957115,
+      "loss": 0.497,
+      "step": 2572
+    },
+    {
+      "epoch": 1.504603244191144,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00012456140350877194,
+      "loss": 0.5424,
+      "step": 2574
+    },
+    {
+      "epoch": 1.5057723220809587,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00012446393762183237,
+      "loss": 0.4895,
+      "step": 2576
+    },
+    {
+      "epoch": 1.506941399970773,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.0001243664717348928,
+      "loss": 0.4274,
+      "step": 2578
+    },
+    {
+      "epoch": 1.5081104778605874,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.0001242690058479532,
+      "loss": 0.474,
+      "step": 2580
+    },
+    {
+      "epoch": 1.509279555750402,
+      "grad_norm": 0.5,
+      "learning_rate": 0.00012417153996101366,
+      "loss": 0.4974,
+      "step": 2582
+    },
+    {
+      "epoch": 1.5104486336402163,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00012407407407407408,
+      "loss": 0.487,
+      "step": 2584
+    },
+    {
+      "epoch": 1.5116177115300307,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.0001239766081871345,
+      "loss": 0.5199,
+      "step": 2586
+    },
+    {
+      "epoch": 1.5127867894198452,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00012387914230019493,
+      "loss": 0.4695,
+      "step": 2588
+    },
+    {
+      "epoch": 1.5139558673096594,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00012378167641325538,
+      "loss": 0.5281,
+      "step": 2590
+    },
+    {
+      "epoch": 1.515124945199474,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.0001236842105263158,
+      "loss": 0.5104,
+      "step": 2592
+    },
+    {
+      "epoch": 1.5162940230892883,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00012358674463937622,
+      "loss": 0.4797,
+      "step": 2594
+    },
+    {
+      "epoch": 1.5174631009791026,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00012348927875243664,
+      "loss": 0.5521,
+      "step": 2596
+    },
+    {
+      "epoch": 1.5186321788689172,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.0001233918128654971,
+      "loss": 0.5292,
+      "step": 2598
+    },
+    {
+      "epoch": 1.5198012567587316,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00012329434697855751,
+      "loss": 0.5144,
+      "step": 2600
+    },
+    {
+      "epoch": 1.520970334648546,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.00012319688109161794,
+      "loss": 0.5575,
+      "step": 2602
+    },
+    {
+      "epoch": 1.5221394125383605,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00012309941520467836,
+      "loss": 0.5626,
+      "step": 2604
+    },
+    {
+      "epoch": 1.5233084904281746,
+      "grad_norm": 0.58984375,
+      "learning_rate": 0.0001230019493177388,
+      "loss": 0.5235,
+      "step": 2606
+    },
+    {
+      "epoch": 1.5244775683179892,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00012290448343079923,
+      "loss": 0.508,
+      "step": 2608
+    },
+    {
+      "epoch": 1.5256466462078035,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00012280701754385965,
+      "loss": 0.4921,
+      "step": 2610
+    },
+    {
+      "epoch": 1.526815724097618,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00012270955165692008,
+      "loss": 0.4954,
+      "step": 2612
+    },
+    {
+      "epoch": 1.5279848019874325,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.0001226120857699805,
+      "loss": 0.5225,
+      "step": 2614
+    },
+    {
+      "epoch": 1.5291538798772468,
+      "grad_norm": 1.6875,
+      "learning_rate": 0.00012251461988304095,
+      "loss": 0.4836,
+      "step": 2616
+    },
+    {
+      "epoch": 1.5303229577670612,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00012241715399610137,
+      "loss": 0.4991,
+      "step": 2618
+    },
+    {
+      "epoch": 1.5314920356568757,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001223196881091618,
+      "loss": 0.4973,
+      "step": 2620
+    },
+    {
+      "epoch": 1.53266111354669,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00012222222222222224,
+      "loss": 0.5018,
+      "step": 2622
+    },
+    {
+      "epoch": 1.5338301914365045,
+      "grad_norm": 0.609375,
+      "learning_rate": 0.00012212475633528266,
+      "loss": 0.5728,
+      "step": 2624
+    },
+    {
+      "epoch": 1.534999269326319,
+      "grad_norm": 0.474609375,
+      "learning_rate": 0.00012202729044834309,
+      "loss": 0.5437,
+      "step": 2626
+    },
+    {
+      "epoch": 1.5361683472161332,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00012192982456140352,
+      "loss": 0.533,
+      "step": 2628
+    },
+    {
+      "epoch": 1.5373374251059477,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00012183235867446394,
+      "loss": 0.4743,
+      "step": 2630
+    },
+    {
+      "epoch": 1.538506502995762,
+      "grad_norm": 0.41015625,
+      "learning_rate": 0.00012173489278752437,
+      "loss": 0.4799,
+      "step": 2632
+    },
+    {
+      "epoch": 1.5396755808855764,
+      "grad_norm": 0.5859375,
+      "learning_rate": 0.00012163742690058479,
+      "loss": 0.5407,
+      "step": 2634
+    },
+    {
+      "epoch": 1.540844658775391,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00012153996101364522,
+      "loss": 0.4884,
+      "step": 2636
+    },
+    {
+      "epoch": 1.5420137366652054,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00012144249512670567,
+      "loss": 0.5327,
+      "step": 2638
+    },
+    {
+      "epoch": 1.5431828145550197,
+      "grad_norm": 1.0703125,
+      "learning_rate": 0.0001213450292397661,
+      "loss": 0.6054,
+      "step": 2640
+    },
+    {
+      "epoch": 1.5443518924448343,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00012124756335282652,
+      "loss": 0.5225,
+      "step": 2642
+    },
+    {
+      "epoch": 1.5455209703346484,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00012115009746588694,
+      "loss": 0.4628,
+      "step": 2644
+    },
+    {
+      "epoch": 1.546690048224463,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00012105263157894738,
+      "loss": 0.5348,
+      "step": 2646
+    },
+    {
+      "epoch": 1.5478591261142773,
+      "grad_norm": 0.5078125,
+      "learning_rate": 0.0001209551656920078,
+      "loss": 0.5142,
+      "step": 2648
+    },
+    {
+      "epoch": 1.5490282040040917,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00012085769980506822,
+      "loss": 0.4994,
+      "step": 2650
+    },
+    {
+      "epoch": 1.5501972818939063,
+      "grad_norm": 0.515625,
+      "learning_rate": 0.00012076023391812866,
+      "loss": 0.5443,
+      "step": 2652
+    },
+    {
+      "epoch": 1.5513663597837206,
+      "grad_norm": 0.458984375,
+      "learning_rate": 0.0001206627680311891,
+      "loss": 0.4998,
+      "step": 2654
+    },
+    {
+      "epoch": 1.552535437673535,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00012056530214424953,
+      "loss": 0.5259,
+      "step": 2656
+    },
+    {
+      "epoch": 1.5537045155633495,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00012046783625730995,
+      "loss": 0.5091,
+      "step": 2658
+    },
+    {
+      "epoch": 1.5548735934531637,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00012037037037037037,
+      "loss": 0.6077,
+      "step": 2660
+    },
+    {
+      "epoch": 1.5560426713429782,
+      "grad_norm": 0.859375,
+      "learning_rate": 0.00012027290448343081,
+      "loss": 0.4897,
+      "step": 2662
+    },
+    {
+      "epoch": 1.5572117492327926,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00012017543859649123,
+      "loss": 0.53,
+      "step": 2664
+    },
+    {
+      "epoch": 1.558380827122607,
+      "grad_norm": 0.67578125,
+      "learning_rate": 0.00012007797270955165,
+      "loss": 0.5213,
+      "step": 2666
+    },
+    {
+      "epoch": 1.5595499050124215,
+      "grad_norm": 0.73046875,
+      "learning_rate": 0.00011998050682261208,
+      "loss": 0.4866,
+      "step": 2668
+    },
+    {
+      "epoch": 1.5607189829022359,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00011988304093567253,
+      "loss": 0.5091,
+      "step": 2670
+    },
+    {
+      "epoch": 1.5618880607920502,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00011978557504873296,
+      "loss": 0.4714,
+      "step": 2672
+    },
+    {
+      "epoch": 1.5630571386818648,
+      "grad_norm": 0.74609375,
+      "learning_rate": 0.00011968810916179338,
+      "loss": 0.56,
+      "step": 2674
+    },
+    {
+      "epoch": 1.564226216571679,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.0001195906432748538,
+      "loss": 0.5047,
+      "step": 2676
+    },
+    {
+      "epoch": 1.5653952944614935,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00011949317738791423,
+      "loss": 0.5009,
+      "step": 2678
+    },
+    {
+      "epoch": 1.5665643723513079,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00011939571150097466,
+      "loss": 0.4999,
+      "step": 2680
+    },
+    {
+      "epoch": 1.5677334502411222,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00011929824561403509,
+      "loss": 0.5216,
+      "step": 2682
+    },
+    {
+      "epoch": 1.5689025281309368,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00011920077972709551,
+      "loss": 0.4842,
+      "step": 2684
+    },
+    {
+      "epoch": 1.5700716060207511,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00011910331384015596,
+      "loss": 0.5055,
+      "step": 2686
+    },
+    {
+      "epoch": 1.5712406839105655,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00011900584795321638,
+      "loss": 0.4664,
+      "step": 2688
+    },
+    {
+      "epoch": 1.57240976180038,
+      "grad_norm": 0.56640625,
+      "learning_rate": 0.00011890838206627682,
+      "loss": 0.5221,
+      "step": 2690
+    },
+    {
+      "epoch": 1.5735788396901944,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00011881091617933724,
+      "loss": 0.481,
+      "step": 2692
+    },
+    {
+      "epoch": 1.5747479175800088,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00011871345029239766,
+      "loss": 0.6087,
+      "step": 2694
+    },
+    {
+      "epoch": 1.5759169954698233,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.0001186159844054581,
+      "loss": 0.5085,
+      "step": 2696
+    },
+    {
+      "epoch": 1.5770860733596375,
+      "grad_norm": 0.40625,
+      "learning_rate": 0.00011851851851851852,
+      "loss": 0.4165,
+      "step": 2698
+    },
+    {
+      "epoch": 1.578255151249452,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00011842105263157894,
+      "loss": 0.5322,
+      "step": 2700
+    },
+    {
+      "epoch": 1.5794242291392664,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00011832358674463939,
+      "loss": 0.5114,
+      "step": 2702
+    },
+    {
+      "epoch": 1.5805933070290807,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00011822612085769981,
+      "loss": 0.4954,
+      "step": 2704
+    },
+    {
+      "epoch": 1.5817623849188953,
+      "grad_norm": 0.5703125,
+      "learning_rate": 0.00011812865497076025,
+      "loss": 0.5638,
+      "step": 2706
+    },
+    {
+      "epoch": 1.5829314628087097,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00011803118908382067,
+      "loss": 0.5842,
+      "step": 2708
+    },
+    {
+      "epoch": 1.584100540698524,
+      "grad_norm": 0.84765625,
+      "learning_rate": 0.0001179337231968811,
+      "loss": 0.5528,
+      "step": 2710
+    },
+    {
+      "epoch": 1.5852696185883386,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00011783625730994152,
+      "loss": 0.5358,
+      "step": 2712
+    },
+    {
+      "epoch": 1.5864386964781527,
+      "grad_norm": 0.6328125,
+      "learning_rate": 0.00011773879142300195,
+      "loss": 0.6037,
+      "step": 2714
+    },
+    {
+      "epoch": 1.5876077743679673,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00011764132553606237,
+      "loss": 0.4884,
+      "step": 2716
+    },
+    {
+      "epoch": 1.5887768522577816,
+      "grad_norm": 0.546875,
+      "learning_rate": 0.00011754385964912282,
+      "loss": 0.5316,
+      "step": 2718
+    },
+    {
+      "epoch": 1.589945930147596,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00011744639376218325,
+      "loss": 0.5015,
+      "step": 2720
+    },
+    {
+      "epoch": 1.5911150080374106,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00011734892787524367,
+      "loss": 0.4958,
+      "step": 2722
+    },
+    {
+      "epoch": 1.592284085927225,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.0001172514619883041,
+      "loss": 0.4921,
+      "step": 2724
+    },
+    {
+      "epoch": 1.5934531638170393,
+      "grad_norm": 0.421875,
+      "learning_rate": 0.00011715399610136453,
+      "loss": 0.4645,
+      "step": 2726
+    },
+    {
+      "epoch": 1.5946222417068538,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00011705653021442495,
+      "loss": 0.4816,
+      "step": 2728
+    },
+    {
+      "epoch": 1.595791319596668,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00011695906432748539,
+      "loss": 0.5128,
+      "step": 2730
+    },
+    {
+      "epoch": 1.5969603974864826,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00011686159844054581,
+      "loss": 0.4852,
+      "step": 2732
+    },
+    {
+      "epoch": 1.598129475376297,
+      "grad_norm": 0.55078125,
+      "learning_rate": 0.00011676413255360626,
+      "loss": 0.5084,
+      "step": 2734
+    },
+    {
+      "epoch": 1.5992985532661113,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00011666666666666668,
+      "loss": 0.4967,
+      "step": 2736
+    },
+    {
+      "epoch": 1.6004676311559258,
+      "grad_norm": 0.408203125,
+      "learning_rate": 0.0001165692007797271,
+      "loss": 0.4603,
+      "step": 2738
+    },
+    {
+      "epoch": 1.6016367090457402,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00011647173489278754,
+      "loss": 0.4917,
+      "step": 2740
+    },
+    {
+      "epoch": 1.6028057869355545,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00011637426900584796,
+      "loss": 0.5217,
+      "step": 2742
+    },
+    {
+      "epoch": 1.603974864825369,
+      "grad_norm": 0.4140625,
+      "learning_rate": 0.00011627680311890838,
+      "loss": 0.4649,
+      "step": 2744
+    },
+    {
+      "epoch": 1.6051439427151832,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001161793372319688,
+      "loss": 0.505,
+      "step": 2746
+    },
+    {
+      "epoch": 1.6063130206049978,
+      "grad_norm": 0.419921875,
+      "learning_rate": 0.00011608187134502924,
+      "loss": 0.4309,
+      "step": 2748
+    },
+    {
+      "epoch": 1.6074820984948122,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00011598440545808969,
+      "loss": 0.5105,
+      "step": 2750
+    },
+    {
+      "epoch": 1.6086511763846265,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00011588693957115011,
+      "loss": 0.5273,
+      "step": 2752
+    },
+    {
+      "epoch": 1.609820254274441,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00011578947368421053,
+      "loss": 0.5164,
+      "step": 2754
+    },
+    {
+      "epoch": 1.6109893321642554,
+      "grad_norm": 0.435546875,
+      "learning_rate": 0.00011569200779727096,
+      "loss": 0.4782,
+      "step": 2756
+    },
+    {
+      "epoch": 1.6121584100540698,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00011559454191033139,
+      "loss": 0.4889,
+      "step": 2758
+    },
+    {
+      "epoch": 1.6133274879438844,
+      "grad_norm": 0.63671875,
+      "learning_rate": 0.00011549707602339181,
+      "loss": 0.5186,
+      "step": 2760
+    },
+    {
+      "epoch": 1.6144965658336987,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00011539961013645224,
+      "loss": 0.4711,
+      "step": 2762
+    },
+    {
+      "epoch": 1.615665643723513,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00011530214424951267,
+      "loss": 0.4893,
+      "step": 2764
+    },
+    {
+      "epoch": 1.6168347216133276,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00011520467836257311,
+      "loss": 0.5227,
+      "step": 2766
+    },
+    {
+      "epoch": 1.6180037995031418,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00011510721247563354,
+      "loss": 0.4831,
+      "step": 2768
+    },
+    {
+      "epoch": 1.6191728773929563,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00011500974658869397,
+      "loss": 0.5047,
+      "step": 2770
+    },
+    {
+      "epoch": 1.6203419552827707,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00011491228070175439,
+      "loss": 0.512,
+      "step": 2772
+    },
+    {
+      "epoch": 1.621511033172585,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00011481481481481482,
+      "loss": 0.4838,
+      "step": 2774
+    },
+    {
+      "epoch": 1.6226801110623996,
+      "grad_norm": 1.2265625,
+      "learning_rate": 0.00011471734892787525,
+      "loss": 0.5778,
+      "step": 2776
+    },
+    {
+      "epoch": 1.623849188952214,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00011461988304093567,
+      "loss": 0.4849,
+      "step": 2778
+    },
+    {
+      "epoch": 1.6250182668420283,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00011452241715399609,
+      "loss": 0.5256,
+      "step": 2780
+    },
+    {
+      "epoch": 1.626187344731843,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00011442495126705654,
+      "loss": 0.5054,
+      "step": 2782
+    },
+    {
+      "epoch": 1.627356422621657,
+      "grad_norm": 0.57421875,
+      "learning_rate": 0.00011432748538011698,
+      "loss": 0.4739,
+      "step": 2784
+    },
+    {
+      "epoch": 1.6285255005114716,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.0001142300194931774,
+      "loss": 0.6099,
+      "step": 2786
+    },
+    {
+      "epoch": 1.629694578401286,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00011413255360623782,
+      "loss": 0.5187,
+      "step": 2788
+    },
+    {
+      "epoch": 1.6308636562911003,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00011403508771929824,
+      "loss": 0.46,
+      "step": 2790
+    },
+    {
+      "epoch": 1.6320327341809149,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00011393762183235868,
+      "loss": 0.5229,
+      "step": 2792
+    },
+    {
+      "epoch": 1.6332018120707292,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.0001138401559454191,
+      "loss": 0.4822,
+      "step": 2794
+    },
+    {
+      "epoch": 1.6343708899605436,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.00011374269005847952,
+      "loss": 0.5022,
+      "step": 2796
+    },
+    {
+      "epoch": 1.6355399678503582,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00011364522417153997,
+      "loss": 0.4819,
+      "step": 2798
+    },
+    {
+      "epoch": 1.6367090457401723,
+      "grad_norm": 0.70703125,
+      "learning_rate": 0.0001135477582846004,
+      "loss": 0.4742,
+      "step": 2800
+    },
+    {
+      "epoch": 1.6378781236299869,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00011345029239766083,
+      "loss": 0.4819,
+      "step": 2802
+    },
+    {
+      "epoch": 1.6390472015198012,
+      "grad_norm": 0.62890625,
+      "learning_rate": 0.00011335282651072125,
+      "loss": 0.4933,
+      "step": 2804
+    },
+    {
+      "epoch": 1.6402162794096156,
+      "grad_norm": 0.54296875,
+      "learning_rate": 0.00011325536062378168,
+      "loss": 0.4929,
+      "step": 2806
+    },
+    {
+      "epoch": 1.6413853572994301,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00011315789473684211,
+      "loss": 0.4854,
+      "step": 2808
+    },
+    {
+      "epoch": 1.6425544351892445,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.00011306042884990253,
+      "loss": 0.5641,
+      "step": 2810
+    },
+    {
+      "epoch": 1.6437235130790588,
+      "grad_norm": 0.52734375,
+      "learning_rate": 0.00011296296296296296,
+      "loss": 0.5033,
+      "step": 2812
+    },
+    {
+      "epoch": 1.6448925909688734,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.0001128654970760234,
+      "loss": 0.487,
+      "step": 2814
+    },
+    {
+      "epoch": 1.6460616688586878,
+      "grad_norm": 0.478515625,
+      "learning_rate": 0.00011276803118908383,
+      "loss": 0.4896,
+      "step": 2816
+    },
+    {
+      "epoch": 1.6472307467485021,
+      "grad_norm": 0.5390625,
+      "learning_rate": 0.00011267056530214426,
+      "loss": 0.4961,
+      "step": 2818
+    },
+    {
+      "epoch": 1.6483998246383167,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00011257309941520469,
+      "loss": 0.4786,
+      "step": 2820
+    },
+    {
+      "epoch": 1.6495689025281308,
+      "grad_norm": 0.8984375,
+      "learning_rate": 0.00011247563352826511,
+      "loss": 0.55,
+      "step": 2822
+    },
+    {
+      "epoch": 1.6507379804179454,
+      "grad_norm": 0.5625,
+      "learning_rate": 0.00011237816764132553,
+      "loss": 0.5174,
+      "step": 2824
+    },
+    {
+      "epoch": 1.6519070583077597,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00011228070175438597,
+      "loss": 0.4896,
+      "step": 2826
+    },
+    {
+      "epoch": 1.653076136197574,
+      "grad_norm": 0.5234375,
+      "learning_rate": 0.00011218323586744639,
+      "loss": 0.5106,
+      "step": 2828
+    },
+    {
+      "epoch": 1.6542452140873887,
+      "grad_norm": 0.68359375,
+      "learning_rate": 0.00011208576998050684,
+      "loss": 0.6019,
+      "step": 2830
+    },
+    {
+      "epoch": 1.655414291977203,
+      "grad_norm": 0.76171875,
+      "learning_rate": 0.00011198830409356726,
+      "loss": 0.5032,
+      "step": 2832
+    },
+    {
+      "epoch": 1.6565833698670174,
+      "grad_norm": 0.6015625,
+      "learning_rate": 0.0001118908382066277,
+      "loss": 0.56,
+      "step": 2834
+    },
+    {
+      "epoch": 1.657752447756832,
+      "grad_norm": 0.58203125,
+      "learning_rate": 0.00011179337231968812,
+      "loss": 0.4863,
+      "step": 2836
+    },
+    {
+      "epoch": 1.658921525646646,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00011169590643274854,
+      "loss": 0.499,
+      "step": 2838
+    },
+    {
+      "epoch": 1.6600906035364607,
+      "grad_norm": 1.125,
+      "learning_rate": 0.00011159844054580896,
+      "loss": 0.5816,
+      "step": 2840
+    },
+    {
+      "epoch": 1.661259681426275,
+      "grad_norm": 0.6796875,
+      "learning_rate": 0.0001115009746588694,
+      "loss": 0.4982,
+      "step": 2842
+    },
+    {
+      "epoch": 1.6624287593160894,
+      "grad_norm": 0.48828125,
+      "learning_rate": 0.00011140350877192982,
+      "loss": 0.5131,
+      "step": 2844
+    },
+    {
+      "epoch": 1.663597837205904,
+      "grad_norm": 0.625,
+      "learning_rate": 0.00011130604288499027,
+      "loss": 0.4955,
+      "step": 2846
+    },
+    {
+      "epoch": 1.6647669150957183,
+      "grad_norm": 0.498046875,
+      "learning_rate": 0.0001112085769980507,
+      "loss": 0.5469,
+      "step": 2848
+    },
+    {
+      "epoch": 1.6659359929855326,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00011111111111111112,
+      "loss": 0.5019,
+      "step": 2850
+    },
+    {
+      "epoch": 1.6671050708753472,
+      "grad_norm": 0.53515625,
+      "learning_rate": 0.00011101364522417155,
+      "loss": 0.4795,
+      "step": 2852
+    },
+    {
+      "epoch": 1.6682741487651613,
+      "grad_norm": 0.64453125,
+      "learning_rate": 0.00011091617933723197,
+      "loss": 0.5395,
+      "step": 2854
+    },
+    {
+      "epoch": 1.669443226654976,
+      "grad_norm": 0.51953125,
+      "learning_rate": 0.0001108187134502924,
+      "loss": 0.4709,
+      "step": 2856
+    },
+    {
+      "epoch": 1.6706123045447903,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00011072124756335282,
+      "loss": 0.4927,
+      "step": 2858
+    },
+    {
+      "epoch": 1.6717813824346046,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00011062378167641326,
+      "loss": 0.4645,
+      "step": 2860
+    },
+    {
+      "epoch": 1.6729504603244192,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.0001105263157894737,
+      "loss": 0.5206,
+      "step": 2862
+    },
+    {
+      "epoch": 1.6741195382142335,
+      "grad_norm": 0.4453125,
+      "learning_rate": 0.00011042884990253413,
+      "loss": 0.4863,
+      "step": 2864
+    },
+    {
+      "epoch": 1.675288616104048,
+      "grad_norm": 0.4921875,
+      "learning_rate": 0.00011033138401559455,
+      "loss": 0.4934,
+      "step": 2866
+    },
+    {
+      "epoch": 1.6764576939938625,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00011023391812865498,
+      "loss": 0.494,
+      "step": 2868
+    },
+    {
+      "epoch": 1.6776267718836766,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00011013645224171541,
+      "loss": 0.4683,
+      "step": 2870
+    },
+    {
+      "epoch": 1.6787958497734912,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00011003898635477583,
+      "loss": 0.4944,
+      "step": 2872
+    },
+    {
+      "epoch": 1.6799649276633055,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00010994152046783625,
+      "loss": 0.4921,
+      "step": 2874
+    },
+    {
+      "epoch": 1.6811340055531199,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00010984405458089669,
+      "loss": 0.4924,
+      "step": 2876
+    },
+    {
+      "epoch": 1.6823030834429344,
+      "grad_norm": 0.470703125,
+      "learning_rate": 0.00010974658869395714,
+      "loss": 0.4903,
+      "step": 2878
+    },
+    {
+      "epoch": 1.6834721613327488,
+      "grad_norm": 0.47265625,
+      "learning_rate": 0.00010964912280701756,
+      "loss": 0.4925,
+      "step": 2880
+    },
+    {
+      "epoch": 1.6846412392225631,
+      "grad_norm": 0.427734375,
+      "learning_rate": 0.00010955165692007798,
+      "loss": 0.509,
+      "step": 2882
+    },
+    {
+      "epoch": 1.6858103171123777,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001094541910331384,
+      "loss": 0.5393,
+      "step": 2884
+    },
+    {
+      "epoch": 1.686979395002192,
+      "grad_norm": 0.42578125,
+      "learning_rate": 0.00010935672514619884,
+      "loss": 0.4775,
+      "step": 2886
+    },
+    {
+      "epoch": 1.6881484728920064,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00010925925925925926,
+      "loss": 0.5132,
+      "step": 2888
+    },
+    {
+      "epoch": 1.689317550781821,
+      "grad_norm": 0.439453125,
+      "learning_rate": 0.00010916179337231968,
+      "loss": 0.4972,
+      "step": 2890
+    },
+    {
+      "epoch": 1.6904866286716351,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00010906432748538011,
+      "loss": 0.482,
+      "step": 2892
+    },
+    {
+      "epoch": 1.6916557065614497,
+      "grad_norm": 0.4296875,
+      "learning_rate": 0.00010896686159844056,
+      "loss": 0.4649,
+      "step": 2894
+    },
+    {
+      "epoch": 1.692824784451264,
+      "grad_norm": 1.0625,
+      "learning_rate": 0.00010886939571150099,
+      "loss": 0.5683,
+      "step": 2896
+    },
+    {
+      "epoch": 1.6939938623410784,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00010877192982456141,
+      "loss": 0.4917,
+      "step": 2898
+    },
+    {
+      "epoch": 1.695162940230893,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00010867446393762184,
+      "loss": 0.5137,
+      "step": 2900
+    },
+    {
+      "epoch": 1.6963320181207073,
+      "grad_norm": 0.46484375,
+      "learning_rate": 0.00010857699805068227,
+      "loss": 0.5205,
+      "step": 2902
+    },
+    {
+      "epoch": 1.6975010960105217,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.0001084795321637427,
+      "loss": 0.4467,
+      "step": 2904
+    },
+    {
+      "epoch": 1.6986701739003363,
+      "grad_norm": 0.453125,
+      "learning_rate": 0.00010838206627680312,
+      "loss": 0.5017,
+      "step": 2906
+    },
+    {
+      "epoch": 1.6998392517901504,
+      "grad_norm": 0.455078125,
+      "learning_rate": 0.00010828460038986354,
+      "loss": 0.4427,
+      "step": 2908
+    },
+    {
+      "epoch": 1.701008329679965,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00010818713450292399,
+      "loss": 0.486,
+      "step": 2910
+    },
+    {
+      "epoch": 1.7021774075697793,
+      "grad_norm": 0.62109375,
+      "learning_rate": 0.00010808966861598442,
+      "loss": 0.5212,
+      "step": 2912
+    },
+    {
+      "epoch": 1.7033464854595937,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00010799220272904485,
+      "loss": 0.449,
+      "step": 2914
+    },
+    {
+      "epoch": 1.7045155633494082,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00010789473684210527,
+      "loss": 0.4504,
+      "step": 2916
+    },
+    {
+      "epoch": 1.7056846412392226,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00010779727095516569,
+      "loss": 0.4477,
+      "step": 2918
+    },
+    {
+      "epoch": 1.706853719129037,
+      "grad_norm": 0.66796875,
+      "learning_rate": 0.00010769980506822613,
+      "loss": 0.5277,
+      "step": 2920
+    },
+    {
+      "epoch": 1.7080227970188515,
+      "grad_norm": 0.48046875,
+      "learning_rate": 0.00010760233918128655,
+      "loss": 0.5377,
+      "step": 2922
+    },
+    {
+      "epoch": 1.7091918749086656,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00010750487329434697,
+      "loss": 0.4533,
+      "step": 2924
+    },
+    {
+      "epoch": 1.7103609527984802,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00010740740740740742,
+      "loss": 0.514,
+      "step": 2926
+    },
+    {
+      "epoch": 1.7115300306882946,
+      "grad_norm": 0.50390625,
+      "learning_rate": 0.00010730994152046784,
+      "loss": 0.4668,
+      "step": 2928
+    },
+    {
+      "epoch": 1.712699108578109,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00010721247563352828,
+      "loss": 0.4949,
+      "step": 2930
+    },
+    {
+      "epoch": 1.7138681864679235,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.0001071150097465887,
+      "loss": 0.4852,
+      "step": 2932
+    },
+    {
+      "epoch": 1.7150372643577378,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00010701754385964912,
+      "loss": 0.4805,
+      "step": 2934
+    },
+    {
+      "epoch": 1.7162063422475522,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.00010692007797270956,
+      "loss": 0.4607,
+      "step": 2936
+    },
+    {
+      "epoch": 1.7173754201373668,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00010682261208576998,
+      "loss": 0.4716,
+      "step": 2938
+    },
+    {
+      "epoch": 1.718544498027181,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.0001067251461988304,
+      "loss": 0.4457,
+      "step": 2940
+    },
+    {
+      "epoch": 1.7197135759169955,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00010662768031189085,
+      "loss": 0.481,
+      "step": 2942
+    },
+    {
+      "epoch": 1.7208826538068098,
+      "grad_norm": 0.4375,
+      "learning_rate": 0.00010653021442495128,
+      "loss": 0.4695,
+      "step": 2944
+    },
+    {
+      "epoch": 1.7220517316966242,
+      "grad_norm": 0.416015625,
+      "learning_rate": 0.00010643274853801171,
+      "loss": 0.4423,
+      "step": 2946
+    },
+    {
+      "epoch": 1.7232208095864388,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00010633528265107213,
+      "loss": 0.4777,
+      "step": 2948
+    },
+    {
+      "epoch": 1.724389887476253,
+      "grad_norm": 0.55859375,
+      "learning_rate": 0.00010623781676413256,
+      "loss": 0.5063,
+      "step": 2950
+    },
+    {
+      "epoch": 1.7255589653660675,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00010614035087719298,
+      "loss": 0.487,
+      "step": 2952
+    },
+    {
+      "epoch": 1.726728043255882,
+      "grad_norm": 0.484375,
+      "learning_rate": 0.00010604288499025342,
+      "loss": 0.4766,
+      "step": 2954
+    },
+    {
+      "epoch": 1.7278971211456964,
+      "grad_norm": 0.482421875,
+      "learning_rate": 0.00010594541910331384,
+      "loss": 0.4855,
+      "step": 2956
+    },
+    {
+      "epoch": 1.7290661990355107,
+      "grad_norm": 0.51171875,
+      "learning_rate": 0.00010584795321637429,
+      "loss": 0.5181,
+      "step": 2958
+    },
+    {
+      "epoch": 1.7302352769253253,
+      "grad_norm": 0.49609375,
+      "learning_rate": 0.00010575048732943471,
+      "loss": 0.513,
+      "step": 2960
+    },
+    {
+      "epoch": 1.7314043548151394,
+      "grad_norm": 0.451171875,
+      "learning_rate": 0.00010565302144249513,
+      "loss": 0.4913,
+      "step": 2962
+    },
+    {
+      "epoch": 1.732573432704954,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.00010555555555555557,
+      "loss": 0.4949,
+      "step": 2964
+    },
+    {
+      "epoch": 1.7337425105947684,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00010545808966861599,
+      "loss": 0.4734,
+      "step": 2966
+    },
+    {
+      "epoch": 1.7349115884845827,
+      "grad_norm": 0.53125,
+      "learning_rate": 0.00010536062378167641,
+      "loss": 0.4935,
+      "step": 2968
+    },
+    {
+      "epoch": 1.7360806663743973,
+      "grad_norm": 0.447265625,
+      "learning_rate": 0.00010526315789473685,
+      "loss": 0.4807,
+      "step": 2970
+    },
+    {
+      "epoch": 1.7372497442642116,
+      "grad_norm": 0.4609375,
+      "learning_rate": 0.00010516569200779727,
+      "loss": 0.4963,
+      "step": 2972
+    },
+    {
+      "epoch": 1.738418822154026,
+      "grad_norm": 0.4765625,
+      "learning_rate": 0.00010506822612085772,
+      "loss": 0.4904,
+      "step": 2974
+    },
+    {
+      "epoch": 1.7395879000438406,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00010497076023391814,
+      "loss": 0.4773,
+      "step": 2976
+    },
+    {
+      "epoch": 1.7407569779336547,
+      "grad_norm": 0.443359375,
+      "learning_rate": 0.00010487329434697856,
+      "loss": 0.5176,
+      "step": 2978
+    },
+    {
+      "epoch": 1.7419260558234693,
+      "grad_norm": 0.466796875,
+      "learning_rate": 0.000104775828460039,
+      "loss": 0.4819,
+      "step": 2980
+    },
+    {
+      "epoch": 1.7430951337132836,
+      "grad_norm": 0.46875,
+      "learning_rate": 0.00010467836257309942,
+      "loss": 0.4813,
+      "step": 2982
+    },
+    {
+      "epoch": 1.744264211603098,
+      "grad_norm": 0.44921875,
+      "learning_rate": 0.00010458089668615984,
+      "loss": 0.4732,
+      "step": 2984
+    },
+    {
+      "epoch": 1.7454332894929125,
+      "grad_norm": 0.41796875,
+      "learning_rate": 0.00010448343079922027,
+      "loss": 0.4551,
+      "step": 2986
+    },
+    {
+      "epoch": 1.746602367382727,
+      "grad_norm": 0.44140625,
+      "learning_rate": 0.0001043859649122807,
+      "loss": 0.4832,
+      "step": 2988
+    },
+    {
+      "epoch": 1.7477714452725412,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00010428849902534115,
+      "loss": 0.5009,
+      "step": 2990
+    },
+    {
+      "epoch": 1.7489405231623558,
+      "grad_norm": 0.43359375,
+      "learning_rate": 0.00010419103313840157,
+      "loss": 0.4791,
+      "step": 2992
+    },
+    {
+      "epoch": 1.75010960105217,
+      "grad_norm": 0.5,
+      "learning_rate": 0.000104093567251462,
+      "loss": 0.4453,
+      "step": 2994
+    },
+    {
+      "epoch": 1.7512786789419845,
+      "grad_norm": 0.462890625,
+      "learning_rate": 0.00010399610136452242,
+      "loss": 0.4768,
+      "step": 2996
+    },
+    {
+      "epoch": 1.7524477568317989,
+      "grad_norm": 0.494140625,
+      "learning_rate": 0.00010389863547758286,
+      "loss": 0.5192,
+      "step": 2998
+    },
+    {
+      "epoch": 1.7536168347216132,
+      "grad_norm": 0.45703125,
+      "learning_rate": 0.00010380116959064328,
+      "loss": 0.5391,
+      "step": 3000
+    }
+  ],
+  "logging_steps": 2,
+  "max_steps": 5130,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 3,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.72870338772992e+17,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}