diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5523 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9997421351211965,
+  "eval_steps": 500,
+  "global_step": 3877,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0012893243940175349,
+      "grad_norm": 5.59375,
+      "learning_rate": 6.04062581035517e-06,
+      "loss": 1.2115,
+      "step": 5
+    },
+    {
+      "epoch": 0.0025786487880350697,
+      "grad_norm": 3.953125,
+      "learning_rate": 1.3591408073299131e-05,
+      "loss": 1.0466,
+      "step": 10
+    },
+    {
+      "epoch": 0.0038679731820526046,
+      "grad_norm": 3.828125,
+      "learning_rate": 2.1142190336243096e-05,
+      "loss": 1.1259,
+      "step": 15
+    },
+    {
+      "epoch": 0.005157297576070139,
+      "grad_norm": 3.546875,
+      "learning_rate": 2.8692972599187053e-05,
+      "loss": 0.9623,
+      "step": 20
+    },
+    {
+      "epoch": 0.006446621970087674,
+      "grad_norm": 3.296875,
+      "learning_rate": 3.624375486213102e-05,
+      "loss": 1.0801,
+      "step": 25
+    },
+    {
+      "epoch": 0.007735946364105209,
+      "grad_norm": 2.921875,
+      "learning_rate": 4.379453712507498e-05,
+      "loss": 1.0694,
+      "step": 30
+    },
+    {
+      "epoch": 0.009025270758122744,
+      "grad_norm": 3.0625,
+      "learning_rate": 5.1345319388018946e-05,
+      "loss": 1.0959,
+      "step": 35
+    },
+    {
+      "epoch": 0.010314595152140279,
+      "grad_norm": 3.046875,
+      "learning_rate": 5.2855355744897876e-05,
+      "loss": 1.1499,
+      "step": 40
+    },
+    {
+      "epoch": 0.011603919546157814,
+      "grad_norm": 2.734375,
+      "learning_rate": 5.285486785827738e-05,
+      "loss": 1.1016,
+      "step": 45
+    },
+    {
+      "epoch": 0.012893243940175348,
+      "grad_norm": 2.78125,
+      "learning_rate": 5.285400468290924e-05,
+      "loss": 1.1458,
+      "step": 50
+    },
+    {
+      "epoch": 0.014182568334192883,
+      "grad_norm": 2.703125,
+      "learning_rate": 5.285276623321451e-05,
+      "loss": 1.1322,
+      "step": 55
+    },
+    {
+      "epoch": 0.015471892728210418,
+      "grad_norm": 2.765625,
+      "learning_rate": 5.285115252988398e-05,
+      "loss": 1.2314,
+      "step": 60
+    },
+    {
+      "epoch": 0.016761217122227952,
+      "grad_norm": 2.859375,
+      "learning_rate": 5.284916359987777e-05,
+      "loss": 1.1943,
+      "step": 65
+    },
+    {
+      "epoch": 0.018050541516245487,
+      "grad_norm": 2.546875,
+      "learning_rate": 5.284679947642494e-05,
+      "loss": 1.1607,
+      "step": 70
+    },
+    {
+      "epoch": 0.019339865910263022,
+      "grad_norm": 2.703125,
+      "learning_rate": 5.284406019902289e-05,
+      "loss": 1.2431,
+      "step": 75
+    },
+    {
+      "epoch": 0.020629190304280558,
+      "grad_norm": 2.640625,
+      "learning_rate": 5.284094581343671e-05,
+      "loss": 1.1656,
+      "step": 80
+    },
+    {
+      "epoch": 0.021918514698298093,
+      "grad_norm": 2.8125,
+      "learning_rate": 5.283745637169844e-05,
+      "loss": 1.246,
+      "step": 85
+    },
+    {
+      "epoch": 0.02320783909231563,
+      "grad_norm": 2.546875,
+      "learning_rate": 5.2833591932106155e-05,
+      "loss": 1.2611,
+      "step": 90
+    },
+    {
+      "epoch": 0.02449716348633316,
+      "grad_norm": 2.5625,
+      "learning_rate": 5.2829352559223046e-05,
+      "loss": 1.2741,
+      "step": 95
+    },
+    {
+      "epoch": 0.025786487880350695,
+      "grad_norm": 2.359375,
+      "learning_rate": 5.28247383238763e-05,
+      "loss": 1.1718,
+      "step": 100
+    },
+    {
+      "epoch": 0.02707581227436823,
+      "grad_norm": 2.453125,
+      "learning_rate": 5.281974930315593e-05,
+      "loss": 1.2055,
+      "step": 105
+    },
+    {
+      "epoch": 0.028365136668385766,
+      "grad_norm": 2.609375,
+      "learning_rate": 5.2814385580413484e-05,
+      "loss": 1.2243,
+      "step": 110
+    },
+    {
+      "epoch": 0.0296544610624033,
+      "grad_norm": 2.640625,
+      "learning_rate": 5.2808647245260656e-05,
+      "loss": 1.0913,
+      "step": 115
+    },
+    {
+      "epoch": 0.030943785456420837,
+      "grad_norm": 2.375,
+      "learning_rate": 5.28025343935678e-05,
+      "loss": 1.1677,
+      "step": 120
+    },
+    {
+      "epoch": 0.03223310985043837,
+      "grad_norm": 2.578125,
+      "learning_rate": 5.2796047127462307e-05,
+      "loss": 1.2382,
+      "step": 125
+    },
+    {
+      "epoch": 0.033522434244455904,
+      "grad_norm": 2.375,
+      "learning_rate": 5.2789185555326894e-05,
+      "loss": 1.2948,
+      "step": 130
+    },
+    {
+      "epoch": 0.03481175863847344,
+      "grad_norm": 2.53125,
+      "learning_rate": 5.2781949791797845e-05,
+      "loss": 1.2487,
+      "step": 135
+    },
+    {
+      "epoch": 0.036101083032490974,
+      "grad_norm": 2.265625,
+      "learning_rate": 5.277433995776302e-05,
+      "loss": 1.1657,
+      "step": 140
+    },
+    {
+      "epoch": 0.03739040742650851,
+      "grad_norm": 2.390625,
+      "learning_rate": 5.276635618035989e-05,
+      "loss": 1.1609,
+      "step": 145
+    },
+    {
+      "epoch": 0.038679731820526045,
+      "grad_norm": 2.59375,
+      "learning_rate": 5.2757998592973404e-05,
+      "loss": 1.1451,
+      "step": 150
+    },
+    {
+      "epoch": 0.03996905621454358,
+      "grad_norm": 2.34375,
+      "learning_rate": 5.274926733523374e-05,
+      "loss": 1.2336,
+      "step": 155
+    },
+    {
+      "epoch": 0.041258380608561115,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.274016255301399e-05,
+      "loss": 1.2118,
+      "step": 160
+    },
+    {
+      "epoch": 0.04254770500257865,
+      "grad_norm": 2.609375,
+      "learning_rate": 5.273068439842771e-05,
+      "loss": 1.2228,
+      "step": 165
+    },
+    {
+      "epoch": 0.043837029396596186,
+      "grad_norm": 2.546875,
+      "learning_rate": 5.272083302982642e-05,
+      "loss": 1.2513,
+      "step": 170
+    },
+    {
+      "epoch": 0.04512635379061372,
+      "grad_norm": 2.828125,
+      "learning_rate": 5.271060861179687e-05,
+      "loss": 1.293,
+      "step": 175
+    },
+    {
+      "epoch": 0.04641567818463126,
+      "grad_norm": 2.546875,
+      "learning_rate": 5.27000113151584e-05,
+      "loss": 1.2207,
+      "step": 180
+    },
+    {
+      "epoch": 0.047705002578648785,
+      "grad_norm": 2.28125,
+      "learning_rate": 5.268904131695998e-05,
+      "loss": 1.1663,
+      "step": 185
+    },
+    {
+      "epoch": 0.04899432697266632,
+      "grad_norm": 2.328125,
+      "learning_rate": 5.267769880047737e-05,
+      "loss": 1.0902,
+      "step": 190
+    },
+    {
+      "epoch": 0.050283651366683856,
+      "grad_norm": 2.609375,
+      "learning_rate": 5.266598395520992e-05,
+      "loss": 1.17,
+      "step": 195
+    },
+    {
+      "epoch": 0.05157297576070139,
+      "grad_norm": 2.71875,
+      "learning_rate": 5.265389697687752e-05,
+      "loss": 1.2415,
+      "step": 200
+    },
+    {
+      "epoch": 0.052862300154718926,
+      "grad_norm": 2.46875,
+      "learning_rate": 5.264143806741727e-05,
+      "loss": 1.1643,
+      "step": 205
+    },
+    {
+      "epoch": 0.05415162454873646,
+      "grad_norm": 2.359375,
+      "learning_rate": 5.262860743498013e-05,
+      "loss": 1.1885,
+      "step": 210
+    },
+    {
+      "epoch": 0.055440948942754,
+      "grad_norm": 2.5,
+      "learning_rate": 5.261540529392744e-05,
+      "loss": 1.1791,
+      "step": 215
+    },
+    {
+      "epoch": 0.05673027333677153,
+      "grad_norm": 2.0,
+      "learning_rate": 5.260183186482731e-05,
+      "loss": 1.176,
+      "step": 220
+    },
+    {
+      "epoch": 0.05801959773078907,
+      "grad_norm": 2.359375,
+      "learning_rate": 5.258788737445096e-05,
+      "loss": 1.2173,
+      "step": 225
+    },
+    {
+      "epoch": 0.0593089221248066,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.257357205576898e-05,
+      "loss": 1.1201,
+      "step": 230
+    },
+    {
+      "epoch": 0.06059824651882414,
+      "grad_norm": 2.203125,
+      "learning_rate": 5.2558886147947305e-05,
+      "loss": 1.2142,
+      "step": 235
+    },
+    {
+      "epoch": 0.06188757091284167,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.254382989634337e-05,
+      "loss": 1.1628,
+      "step": 240
+    },
+    {
+      "epoch": 0.0631768953068592,
+      "grad_norm": 2.859375,
+      "learning_rate": 5.2528403552501924e-05,
+      "loss": 1.2297,
+      "step": 245
+    },
+    {
+      "epoch": 0.06446621970087674,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.251260737415082e-05,
+      "loss": 1.1902,
+      "step": 250
+    },
+    {
+      "epoch": 0.06575554409489427,
+      "grad_norm": 2.296875,
+      "learning_rate": 5.249644162519676e-05,
+      "loss": 1.2065,
+      "step": 255
+    },
+    {
+      "epoch": 0.06704486848891181,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.2479906575720856e-05,
+      "loss": 1.1921,
+      "step": 260
+    },
+    {
+      "epoch": 0.06833419288292934,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.246300250197413e-05,
+      "loss": 1.2166,
+      "step": 265
+    },
+    {
+      "epoch": 0.06962351727694688,
+      "grad_norm": 2.359375,
+      "learning_rate": 5.244572968637286e-05,
+      "loss": 1.2223,
+      "step": 270
+    },
+    {
+      "epoch": 0.07091284167096441,
+      "grad_norm": 2.390625,
+      "learning_rate": 5.242808841749393e-05,
+      "loss": 1.2558,
+      "step": 275
+    },
+    {
+      "epoch": 0.07220216606498195,
+      "grad_norm": 2.015625,
+      "learning_rate": 5.241007899006996e-05,
+      "loss": 1.2317,
+      "step": 280
+    },
+    {
+      "epoch": 0.07349149045899948,
+      "grad_norm": 2.765625,
+      "learning_rate": 5.239170170498438e-05,
+      "loss": 1.17,
+      "step": 285
+    },
+    {
+      "epoch": 0.07478081485301702,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.237295686926644e-05,
+      "loss": 1.1697,
+      "step": 290
+    },
+    {
+      "epoch": 0.07607013924703455,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.235384479608603e-05,
+      "loss": 1.2806,
+      "step": 295
+    },
+    {
+      "epoch": 0.07735946364105209,
+      "grad_norm": 2.375,
+      "learning_rate": 5.233436580474849e-05,
+      "loss": 1.1799,
+      "step": 300
+    },
+    {
+      "epoch": 0.07864878803506963,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.231452022068927e-05,
+      "loss": 1.1573,
+      "step": 305
+    },
+    {
+      "epoch": 0.07993811242908716,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.229430837546846e-05,
+      "loss": 1.2525,
+      "step": 310
+    },
+    {
+      "epoch": 0.0812274368231047,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.2273730606765305e-05,
+      "loss": 1.1929,
+      "step": 315
+    },
+    {
+      "epoch": 0.08251676121712223,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.2252787258372486e-05,
+      "loss": 1.194,
+      "step": 320
+    },
+    {
+      "epoch": 0.08380608561113977,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.223147868019047e-05,
+      "loss": 1.1309,
+      "step": 325
+    },
+    {
+      "epoch": 0.0850954100051573,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.22098052282216e-05,
+      "loss": 1.2773,
+      "step": 330
+    },
+    {
+      "epoch": 0.08638473439917484,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.218776726456415e-05,
+      "loss": 1.1824,
+      "step": 335
+    },
+    {
+      "epoch": 0.08767405879319237,
+      "grad_norm": 2.375,
+      "learning_rate": 5.21653651574063e-05,
+      "loss": 1.1324,
+      "step": 340
+    },
+    {
+      "epoch": 0.08896338318720991,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.214259928101999e-05,
+      "loss": 1.2117,
+      "step": 345
+    },
+    {
+      "epoch": 0.09025270758122744,
+      "grad_norm": 2.09375,
+      "learning_rate": 5.2119470015754616e-05,
+      "loss": 1.2373,
+      "step": 350
+    },
+    {
+      "epoch": 0.09154203197524498,
+      "grad_norm": 2.109375,
+      "learning_rate": 5.209597774803075e-05,
+      "loss": 1.0479,
+      "step": 355
+    },
+    {
+      "epoch": 0.09283135636926251,
+      "grad_norm": 2.234375,
+      "learning_rate": 5.20721228703336e-05,
+      "loss": 1.2103,
+      "step": 360
+    },
+    {
+      "epoch": 0.09412068076328003,
+      "grad_norm": 2.34375,
+      "learning_rate": 5.2047905781206555e-05,
+      "loss": 1.1763,
+      "step": 365
+    },
+    {
+      "epoch": 0.09541000515729757,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.202332688524442e-05,
+      "loss": 1.242,
+      "step": 370
+    },
+    {
+      "epoch": 0.0966993295513151,
+      "grad_norm": 2.265625,
+      "learning_rate": 5.1998386593086725e-05,
+      "loss": 1.1736,
+      "step": 375
+    },
+    {
+      "epoch": 0.09798865394533264,
+      "grad_norm": 2.328125,
+      "learning_rate": 5.197308532141083e-05,
+      "loss": 1.1454,
+      "step": 380
+    },
+    {
+      "epoch": 0.09927797833935018,
+      "grad_norm": 2.296875,
+      "learning_rate": 5.1947423492925016e-05,
+      "loss": 1.1139,
+      "step": 385
+    },
+    {
+      "epoch": 0.10056730273336771,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.1921401536361344e-05,
+      "loss": 1.1734,
+      "step": 390
+    },
+    {
+      "epoch": 0.10185662712738525,
+      "grad_norm": 2.125,
+      "learning_rate": 5.1895019886468554e-05,
+      "loss": 1.1896,
+      "step": 395
+    },
+    {
+      "epoch": 0.10314595152140278,
+      "grad_norm": 2.8125,
+      "learning_rate": 5.186827898400478e-05,
+      "loss": 1.199,
+      "step": 400
+    },
+    {
+      "epoch": 0.10443527591542032,
+      "grad_norm": 2.546875,
+      "learning_rate": 5.184117927573018e-05,
+      "loss": 1.1354,
+      "step": 405
+    },
+    {
+      "epoch": 0.10572460030943785,
+      "grad_norm": 2.03125,
+      "learning_rate": 5.181372121439948e-05,
+      "loss": 1.0959,
+      "step": 410
+    },
+    {
+      "epoch": 0.10701392470345539,
+      "grad_norm": 2.125,
+      "learning_rate": 5.1785905258754425e-05,
+      "loss": 1.2448,
+      "step": 415
+    },
+    {
+      "epoch": 0.10830324909747292,
+      "grad_norm": 2.09375,
+      "learning_rate": 5.175773187351609e-05,
+      "loss": 1.1474,
+      "step": 420
+    },
+    {
+      "epoch": 0.10959257349149046,
+      "grad_norm": 2.25,
+      "learning_rate": 5.1729201529377095e-05,
+      "loss": 1.1742,
+      "step": 425
+    },
+    {
+      "epoch": 0.110881897885508,
+      "grad_norm": 2.0,
+      "learning_rate": 5.170031470299384e-05,
+      "loss": 1.0778,
+      "step": 430
+    },
+    {
+      "epoch": 0.11217122227952553,
+      "grad_norm": 2.03125,
+      "learning_rate": 5.167107187697839e-05,
+      "loss": 1.2032,
+      "step": 435
+    },
+    {
+      "epoch": 0.11346054667354306,
+      "grad_norm": 2.234375,
+      "learning_rate": 5.164147353989056e-05,
+      "loss": 1.0743,
+      "step": 440
+    },
+    {
+      "epoch": 0.1147498710675606,
+      "grad_norm": 2.234375,
+      "learning_rate": 5.161152018622965e-05,
+      "loss": 1.1793,
+      "step": 445
+    },
+    {
+      "epoch": 0.11603919546157813,
+      "grad_norm": 2.171875,
+      "learning_rate": 5.1581212316426254e-05,
+      "loss": 1.1221,
+      "step": 450
+    },
+    {
+      "epoch": 0.11732851985559567,
+      "grad_norm": 2.109375,
+      "learning_rate": 5.155055043683382e-05,
+      "loss": 1.193,
+      "step": 455
+    },
+    {
+      "epoch": 0.1186178442496132,
+      "grad_norm": 2.34375,
+      "learning_rate": 5.151953505972029e-05,
+      "loss": 1.2236,
+      "step": 460
+    },
+    {
+      "epoch": 0.11990716864363074,
+      "grad_norm": 2.234375,
+      "learning_rate": 5.1488166703259424e-05,
+      "loss": 1.2349,
+      "step": 465
+    },
+    {
+      "epoch": 0.12119649303764828,
+      "grad_norm": 2.109375,
+      "learning_rate": 5.1456445891522276e-05,
+      "loss": 1.229,
+      "step": 470
+    },
+    {
+      "epoch": 0.12248581743166581,
+      "grad_norm": 2.28125,
+      "learning_rate": 5.142437315446831e-05,
+      "loss": 1.1681,
+      "step": 475
+    },
+    {
+      "epoch": 0.12377514182568335,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.139194902793664e-05,
+      "loss": 1.1831,
+      "step": 480
+    },
+    {
+      "epoch": 0.12506446621970088,
+      "grad_norm": 2.40625,
+      "learning_rate": 5.1359174053637043e-05,
+      "loss": 1.2434,
+      "step": 485
+    },
+    {
+      "epoch": 0.1263537906137184,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.1326048779140886e-05,
+      "loss": 1.1939,
+      "step": 490
+    },
+    {
+      "epoch": 0.12764311500773595,
+      "grad_norm": 2.203125,
+      "learning_rate": 5.129257375787202e-05,
+      "loss": 1.1533,
+      "step": 495
+    },
+    {
+      "epoch": 0.12893243940175347,
+      "grad_norm": 1.9296875,
+      "learning_rate": 5.1258749549097495e-05,
+      "loss": 1.1803,
+      "step": 500
+    },
+    {
+      "epoch": 0.12893243940175347,
+      "eval_loss": 1.1997082233428955,
+      "eval_runtime": 5.9796,
+      "eval_samples_per_second": 10.201,
+      "eval_steps_per_second": 10.201,
+      "step": 500
+    },
+    {
+      "epoch": 0.13022176379577102,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.122457671791826e-05,
+      "loss": 1.2035,
+      "step": 505
+    },
+    {
+      "epoch": 0.13151108818978854,
+      "grad_norm": 2.125,
+      "learning_rate": 5.119005583525967e-05,
+      "loss": 1.1817,
+      "step": 510
+    },
+    {
+      "epoch": 0.1328004125838061,
+      "grad_norm": 2.25,
+      "learning_rate": 5.115518747786201e-05,
+      "loss": 1.2583,
+      "step": 515
+    },
+    {
+      "epoch": 0.13408973697782361,
+      "grad_norm": 2.015625,
+      "learning_rate": 5.111997222827079e-05,
+      "loss": 1.2218,
+      "step": 520
+    },
+    {
+      "epoch": 0.13537906137184116,
+      "grad_norm": 2.078125,
+      "learning_rate": 5.108441067482706e-05,
+      "loss": 1.192,
+      "step": 525
+    },
+    {
+      "epoch": 0.13666838576585869,
+      "grad_norm": 2.0625,
+      "learning_rate": 5.1048503411657564e-05,
+      "loss": 1.2815,
+      "step": 530
+    },
+    {
+      "epoch": 0.13795771015987623,
+      "grad_norm": 2.0625,
+      "learning_rate": 5.1012251038664823e-05,
+      "loss": 1.1606,
+      "step": 535
+    },
+    {
+      "epoch": 0.13924703455389376,
+      "grad_norm": 2.25,
+      "learning_rate": 5.0975654161517116e-05,
+      "loss": 1.1756,
+      "step": 540
+    },
+    {
+      "epoch": 0.1405363589479113,
+      "grad_norm": 2.09375,
+      "learning_rate": 5.093871339163833e-05,
+      "loss": 1.2096,
+      "step": 545
+    },
+    {
+      "epoch": 0.14182568334192883,
+      "grad_norm": 2.015625,
+      "learning_rate": 5.090142934619778e-05,
+      "loss": 1.0802,
+      "step": 550
+    },
+    {
+      "epoch": 0.14311500773594638,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.086380264809987e-05,
+      "loss": 1.1596,
+      "step": 555
+    },
+    {
+      "epoch": 0.1444043321299639,
+      "grad_norm": 2.453125,
+      "learning_rate": 5.082583392597372e-05,
+      "loss": 1.1686,
+      "step": 560
+    },
+    {
+      "epoch": 0.14569365652398145,
+      "grad_norm": 1.9609375,
+      "learning_rate": 5.0787523814162645e-05,
+      "loss": 1.1517,
+      "step": 565
+    },
+    {
+      "epoch": 0.14698298091799897,
+      "grad_norm": 2.03125,
+      "learning_rate": 5.0748872952713536e-05,
+      "loss": 1.2205,
+      "step": 570
+    },
+    {
+      "epoch": 0.14827230531201652,
+      "grad_norm": 2.15625,
+      "learning_rate": 5.07098819873662e-05,
+      "loss": 1.1786,
+      "step": 575
+    },
+    {
+      "epoch": 0.14956162970603404,
+      "grad_norm": 2.171875,
+      "learning_rate": 5.0670551569542557e-05,
+      "loss": 1.1836,
+      "step": 580
+    },
+    {
+      "epoch": 0.15085095410005156,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.063088235633576e-05,
+      "loss": 1.1965,
+      "step": 585
+    },
+    {
+      "epoch": 0.1521402784940691,
+      "grad_norm": 2.03125,
+      "learning_rate": 5.0590875010499206e-05,
+      "loss": 1.1585,
+      "step": 590
+    },
+    {
+      "epoch": 0.15342960288808663,
+      "grad_norm": 2.203125,
+      "learning_rate": 5.055053020043549e-05,
+      "loss": 1.2677,
+      "step": 595
+    },
+    {
+      "epoch": 0.15471892728210418,
+      "grad_norm": 2.0625,
+      "learning_rate": 5.050984860018521e-05,
+      "loss": 1.1154,
+      "step": 600
+    },
+    {
+      "epoch": 0.1560082516761217,
+      "grad_norm": 2.21875,
+      "learning_rate": 5.046883088941571e-05,
+      "loss": 1.1757,
+      "step": 605
+    },
+    {
+      "epoch": 0.15729757607013925,
+      "grad_norm": 2.171875,
+      "learning_rate": 5.042747775340974e-05,
+      "loss": 1.2166,
+      "step": 610
+    },
+    {
+      "epoch": 0.15858690046415677,
+      "grad_norm": 1.8671875,
+      "learning_rate": 5.0385789883054e-05,
+      "loss": 1.1866,
+      "step": 615
+    },
+    {
+      "epoch": 0.15987622485817432,
+      "grad_norm": 2.078125,
+      "learning_rate": 5.03437679748276e-05,
+      "loss": 1.2244,
+      "step": 620
+    },
+    {
+      "epoch": 0.16116554925219184,
+      "grad_norm": 1.90625,
+      "learning_rate": 5.030141273079039e-05,
+      "loss": 1.1816,
+      "step": 625
+    },
+    {
+      "epoch": 0.1624548736462094,
+      "grad_norm": 2.1875,
+      "learning_rate": 5.0258724858571304e-05,
+      "loss": 1.1844,
+      "step": 630
+    },
+    {
+      "epoch": 0.1637441980402269,
+      "grad_norm": 1.9921875,
+      "learning_rate": 5.021570507135647e-05,
+      "loss": 1.1354,
+      "step": 635
+    },
+    {
+      "epoch": 0.16503352243424446,
+      "grad_norm": 2.140625,
+      "learning_rate": 5.01723540878773e-05,
+      "loss": 1.1669,
+      "step": 640
+    },
+    {
+      "epoch": 0.16632284682826198,
+      "grad_norm": 1.9921875,
+      "learning_rate": 5.012867263239854e-05,
+      "loss": 1.1339,
+      "step": 645
+    },
+    {
+      "epoch": 0.16761217122227953,
+      "grad_norm": 2.234375,
+      "learning_rate": 5.0084661434706116e-05,
+      "loss": 1.1506,
+      "step": 650
+    },
+    {
+      "epoch": 0.16890149561629705,
+      "grad_norm": 2.234375,
+      "learning_rate": 5.0040321230094935e-05,
+      "loss": 1.1925,
+      "step": 655
+    },
+    {
+      "epoch": 0.1701908200103146,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.9995652759356636e-05,
+      "loss": 1.2023,
+      "step": 660
+    },
+    {
+      "epoch": 0.17148014440433212,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.995065676876723e-05,
+      "loss": 1.1732,
+      "step": 665
+    },
+    {
+      "epoch": 0.17276946879834967,
+      "grad_norm": 2.765625,
+      "learning_rate": 4.9905334010074547e-05,
+      "loss": 1.2815,
+      "step": 670
+    },
+    {
+      "epoch": 0.1740587931923672,
+      "grad_norm": 2.0625,
+      "learning_rate": 4.985968524048576e-05,
+      "loss": 1.1651,
+      "step": 675
+    },
+    {
+      "epoch": 0.17534811758638474,
+      "grad_norm": 2.03125,
+      "learning_rate": 4.98137112226547e-05,
+      "loss": 1.0955,
+      "step": 680
+    },
+    {
+      "epoch": 0.17663744198040227,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.976741272466913e-05,
+      "loss": 1.1798,
+      "step": 685
+    },
+    {
+      "epoch": 0.17792676637441981,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.972079052003789e-05,
+      "loss": 1.2084,
+      "step": 690
+    },
+    {
+      "epoch": 0.17921609076843734,
+      "grad_norm": 1.921875,
+      "learning_rate": 4.9673845387677976e-05,
+      "loss": 1.1601,
+      "step": 695
+    },
+    {
+      "epoch": 0.18050541516245489,
+      "grad_norm": 2.28125,
+      "learning_rate": 4.962657811190156e-05,
+      "loss": 1.1366,
+      "step": 700
+    },
+    {
+      "epoch": 0.1817947395564724,
+      "grad_norm": 2.0,
+      "learning_rate": 4.957898948240284e-05,
+      "loss": 1.1559,
+      "step": 705
+    },
+    {
+      "epoch": 0.18308406395048996,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.953108029424488e-05,
+      "loss": 1.2221,
+      "step": 710
+    },
+    {
+      "epoch": 0.18437338834450748,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.948285134784633e-05,
+      "loss": 1.1051,
+      "step": 715
+    },
+    {
+      "epoch": 0.18566271273852503,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.943430344896802e-05,
+      "loss": 1.1687,
+      "step": 720
+    },
+    {
+      "epoch": 0.18695203713254255,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.938543740869953e-05,
+      "loss": 1.2572,
+      "step": 725
+    },
+    {
+      "epoch": 0.18824136152656007,
+      "grad_norm": 1.9765625,
+      "learning_rate": 4.9336254043445605e-05,
+      "loss": 1.1043,
+      "step": 730
+    },
+    {
+      "epoch": 0.18953068592057762,
+      "grad_norm": 2.15625,
+      "learning_rate": 4.9286754174912566e-05,
+      "loss": 1.1089,
+      "step": 735
+    },
+    {
+      "epoch": 0.19082001031459514,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.923693863009455e-05,
+      "loss": 1.1846,
+      "step": 740
+    },
+    {
+      "epoch": 0.1921093347086127,
+      "grad_norm": 2.0625,
+      "learning_rate": 4.918680824125966e-05,
+      "loss": 1.1286,
+      "step": 745
+    },
+    {
+      "epoch": 0.1933986591026302,
+      "grad_norm": 2.203125,
+      "learning_rate": 4.913636384593615e-05,
+      "loss": 1.0553,
+      "step": 750
+    },
+    {
+      "epoch": 0.19468798349664776,
+      "grad_norm": 1.7734375,
+      "learning_rate": 4.908560628689835e-05,
+      "loss": 1.0351,
+      "step": 755
+    },
+    {
+      "epoch": 0.19597730789066528,
+      "grad_norm": 1.890625,
+      "learning_rate": 4.903453641215261e-05,
+      "loss": 1.1585,
+      "step": 760
+    },
+    {
+      "epoch": 0.19726663228468283,
+      "grad_norm": 2.03125,
+      "learning_rate": 4.898315507492315e-05,
+      "loss": 1.0893,
+      "step": 765
+    },
+    {
+      "epoch": 0.19855595667870035,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.8931463133637794e-05,
+      "loss": 1.2285,
+      "step": 770
+    },
+    {
+      "epoch": 0.1998452810727179,
+      "grad_norm": 2.328125,
+      "learning_rate": 4.8879461451913624e-05,
+      "loss": 1.2796,
+      "step": 775
+    },
+    {
+      "epoch": 0.20113460546673542,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.8827150898542534e-05,
+      "loss": 1.1078,
+      "step": 780
+    },
+    {
+      "epoch": 0.20242392986075297,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.877453234747677e-05,
+      "loss": 1.1233,
+      "step": 785
+    },
+    {
+      "epoch": 0.2037132542547705,
+      "grad_norm": 1.9140625,
+      "learning_rate": 4.8721606677814265e-05,
+      "loss": 1.1082,
+      "step": 790
+    },
+    {
+      "epoch": 0.20500257864878804,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.8668374773784023e-05,
+      "loss": 1.088,
+      "step": 795
+    },
+    {
+      "epoch": 0.20629190304280556,
+      "grad_norm": 1.9140625,
+      "learning_rate": 4.861483752473125e-05,
+      "loss": 1.1169,
+      "step": 800
+    },
+    {
+      "epoch": 0.2075812274368231,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.856099582510257e-05,
+      "loss": 1.1559,
+      "step": 805
+    },
+    {
+      "epoch": 0.20887055183084063,
+      "grad_norm": 1.9765625,
+      "learning_rate": 4.85068505744311e-05,
+      "loss": 1.2187,
+      "step": 810
+    },
+    {
+      "epoch": 0.21015987622485818,
+      "grad_norm": 1.9296875,
+      "learning_rate": 4.845240267732132e-05,
+      "loss": 1.1704,
+      "step": 815
+    },
+    {
+      "epoch": 0.2114492006188757,
+      "grad_norm": 1.875,
+      "learning_rate": 4.8397653043434065e-05,
+      "loss": 1.1339,
+      "step": 820
+    },
+    {
+      "epoch": 0.21273852501289325,
+      "grad_norm": 1.890625,
+      "learning_rate": 4.834260258747129e-05,
+      "loss": 1.1171,
+      "step": 825
+    },
+    {
+      "epoch": 0.21402784940691078,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.828725222916073e-05,
+      "loss": 1.0852,
+      "step": 830
+    },
+    {
+      "epoch": 0.21531717380092832,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.823160289324066e-05,
+      "loss": 1.1294,
+      "step": 835
+    },
+    {
+      "epoch": 0.21660649819494585,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.817565550944433e-05,
+      "loss": 1.0255,
+      "step": 840
+    },
+    {
+      "epoch": 0.2178958225889634,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.8119411012484475e-05,
+      "loss": 1.075,
+      "step": 845
+    },
+    {
+      "epoch": 0.21918514698298092,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.8062870342037706e-05,
+      "loss": 1.2015,
+      "step": 850
+    },
+    {
+      "epoch": 0.22047447137699847,
+      "grad_norm": 1.9609375,
+      "learning_rate": 4.8006034442728814e-05,
+      "loss": 1.0623,
+      "step": 855
+    },
+    {
+      "epoch": 0.221763795771016,
+      "grad_norm": 2.25,
+      "learning_rate": 4.794890426411496e-05,
+      "loss": 1.1991,
+      "step": 860
+    },
+    {
+      "epoch": 0.22305312016503354,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.789148076066983e-05,
+      "loss": 1.1276,
+      "step": 865
+    },
+    {
+      "epoch": 0.22434244455905106,
+      "grad_norm": 1.8671875,
+      "learning_rate": 4.783376489176771e-05,
+      "loss": 1.1259,
+      "step": 870
+    },
+    {
+      "epoch": 0.22563176895306858,
+      "grad_norm": 1.7890625,
+      "learning_rate": 4.77757576216674e-05,
+      "loss": 1.1575,
+      "step": 875
+    },
+    {
+      "epoch": 0.22692109334708613,
+      "grad_norm": 1.9296875,
+      "learning_rate": 4.771745991949617e-05,
+      "loss": 1.045,
+      "step": 880
+    },
+    {
+      "epoch": 0.22821041774110365,
+      "grad_norm": 2.25,
+      "learning_rate": 4.7658872759233505e-05,
+      "loss": 1.1183,
+      "step": 885
+    },
+    {
+      "epoch": 0.2294997421351212,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.75999971196949e-05,
+      "loss": 1.1308,
+      "step": 890
+    },
+    {
+      "epoch": 0.23078906652913872,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.7540833984515455e-05,
+      "loss": 1.0878,
+      "step": 895
+    },
+    {
+      "epoch": 0.23207839092315627,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.748138434213345e-05,
+      "loss": 1.0997,
+      "step": 900
+    },
+    {
+      "epoch": 0.2333677153171738,
+      "grad_norm": 1.9921875,
+      "learning_rate": 4.742164918577386e-05,
+      "loss": 1.1204,
+      "step": 905
+    },
+    {
+      "epoch": 0.23465703971119134,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.736162951343173e-05,
+      "loss": 1.1391,
+      "step": 910
+    },
+    {
+      "epoch": 0.23594636410520886,
+      "grad_norm": 2.25,
+      "learning_rate": 4.7301326327855505e-05,
+      "loss": 1.1672,
+      "step": 915
+    },
+    {
+      "epoch": 0.2372356884992264,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.72407406365303e-05,
+      "loss": 1.1424,
+      "step": 920
+    },
+    {
+      "epoch": 0.23852501289324393,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.717987345166103e-05,
+      "loss": 1.1088,
+      "step": 925
+    },
+    {
+      "epoch": 0.23981433728726148,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.711872579015556e-05,
+      "loss": 1.0471,
+      "step": 930
+    },
+    {
+      "epoch": 0.241103661681279,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.705729867360764e-05,
+      "loss": 1.1416,
+      "step": 935
+    },
+    {
+      "epoch": 0.24239298607529655,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.699559312827989e-05,
+      "loss": 1.0923,
+      "step": 940
+    },
+    {
+      "epoch": 0.24368231046931407,
+      "grad_norm": 2.09375,
+      "learning_rate": 4.693361018508664e-05,
+      "loss": 1.1861,
+      "step": 945
+    },
+    {
+      "epoch": 0.24497163486333162,
+      "grad_norm": 2.125,
+      "learning_rate": 4.687135087957671e-05,
+      "loss": 1.103,
+      "step": 950
+    },
+    {
+      "epoch": 0.24626095925734914,
+      "grad_norm": 1.828125,
+      "learning_rate": 4.680881625191608e-05,
+      "loss": 1.1496,
+      "step": 955
+    },
+    {
+      "epoch": 0.2475502836513667,
+      "grad_norm": 2.03125,
+      "learning_rate": 4.674600734687056e-05,
+      "loss": 1.1968,
+      "step": 960
+    },
+    {
+      "epoch": 0.24883960804538421,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.66829252137883e-05,
+      "loss": 1.1518,
+      "step": 965
+    },
+    {
+      "epoch": 0.25012893243940176,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.661957090658228e-05,
+      "loss": 1.0874,
+      "step": 970
+    },
+    {
+      "epoch": 0.2514182568334193,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.655594548371269e-05,
+      "loss": 1.1538,
+      "step": 975
+    },
+    {
+      "epoch": 0.2527075812274368,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.649205000816923e-05,
+      "loss": 1.0443,
+      "step": 980
+    },
+    {
+      "epoch": 0.2539969056214544,
+      "grad_norm": 1.7890625,
+      "learning_rate": 4.64278855474534e-05,
+      "loss": 1.1422,
+      "step": 985
+    },
+    {
+      "epoch": 0.2552862300154719,
+      "grad_norm": 2.125,
+      "learning_rate": 4.6363453173560604e-05,
+      "loss": 1.1429,
+      "step": 990
+    },
+    {
+      "epoch": 0.2565755544094894,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.62987539629623e-05,
+      "loss": 1.0597,
+      "step": 995
+    },
+    {
+      "epoch": 0.25786487880350695,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.623378899658795e-05,
+      "loss": 1.0648,
+      "step": 1000
+    },
+    {
+      "epoch": 0.25786487880350695,
+      "eval_loss": 1.1551861763000488,
+      "eval_runtime": 5.9584,
+      "eval_samples_per_second": 10.238,
+      "eval_steps_per_second": 10.238,
+      "step": 1000
+    },
+    {
+      "epoch": 0.2591542031975245,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.616855935980703e-05,
+      "loss": 1.1178,
+      "step": 1005
+    },
+    {
+      "epoch": 0.26044352759154205,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.610306614241085e-05,
+      "loss": 1.1416,
+      "step": 1010
+    },
+    {
+      "epoch": 0.26173285198555957,
+      "grad_norm": 2.03125,
+      "learning_rate": 4.603731043859435e-05,
+      "loss": 1.1197,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2630221763795771,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.5971293346937864e-05,
+      "loss": 1.1287,
+      "step": 1020
+    },
+    {
+      "epoch": 0.2643115007735946,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.5905015970388676e-05,
+      "loss": 1.1263,
+      "step": 1025
+    },
+    {
+      "epoch": 0.2656008251676122,
+      "grad_norm": 2.0,
+      "learning_rate": 4.583847941624272e-05,
+      "loss": 1.1601,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2668901495616297,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.5771684796125925e-05,
+      "loss": 1.1069,
+      "step": 1035
+    },
+    {
+      "epoch": 0.26817947395564723,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.570463322597582e-05,
+      "loss": 1.1125,
+      "step": 1040
+    },
+    {
+      "epoch": 0.26946879834966475,
+      "grad_norm": 2.03125,
+      "learning_rate": 4.563732582602271e-05,
+      "loss": 1.0749,
+      "step": 1045
+    },
+    {
+      "epoch": 0.27075812274368233,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.556976372077111e-05,
+      "loss": 1.1864,
+      "step": 1050
+    },
+    {
+      "epoch": 0.27204744713769985,
+      "grad_norm": 1.9921875,
+      "learning_rate": 4.5501948038980866e-05,
+      "loss": 1.0876,
+      "step": 1055
+    },
+    {
+      "epoch": 0.27333677153171737,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.543387991364833e-05,
+      "loss": 1.0913,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2746260959257349,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.5365560481987445e-05,
+      "loss": 1.0716,
+      "step": 1065
+    },
+    {
+      "epoch": 0.27591542031975247,
+      "grad_norm": 1.9375,
+      "learning_rate": 4.52969908854107e-05,
+      "loss": 1.1896,
+      "step": 1070
+    },
+    {
+      "epoch": 0.27720474471377,
+      "grad_norm": 1.8671875,
+      "learning_rate": 4.5228172269510094e-05,
+      "loss": 1.1034,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2784940691077875,
+      "grad_norm": 1.9140625,
+      "learning_rate": 4.515910578403803e-05,
+      "loss": 1.1209,
+      "step": 1080
+    },
+    {
+      "epoch": 0.27978339350180503,
+      "grad_norm": 2.171875,
+      "learning_rate": 4.5089792582888e-05,
+      "loss": 1.0667,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2810727178958226,
+      "grad_norm": 2.078125,
+      "learning_rate": 4.502023382407543e-05,
+      "loss": 1.1368,
+      "step": 1090
+    },
+    {
+      "epoch": 0.28236204228984013,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.4950430669718266e-05,
+      "loss": 1.1341,
+      "step": 1095
+    },
+    {
+      "epoch": 0.28365136668385765,
+      "grad_norm": 1.9765625,
+      "learning_rate": 4.4880384286017535e-05,
+      "loss": 1.1036,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2849406910778752,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.481009584323793e-05,
+      "loss": 1.1085,
+      "step": 1105
+    },
+    {
+      "epoch": 0.28623001547189275,
+      "grad_norm": 1.9296875,
+      "learning_rate": 4.4739566515688235e-05,
+      "loss": 1.0887,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2875193398659103,
+      "grad_norm": 2.0,
+      "learning_rate": 4.4668797481701654e-05,
+      "loss": 1.115,
+      "step": 1115
+    },
+    {
+      "epoch": 0.2888086642599278,
+      "grad_norm": 1.9765625,
+      "learning_rate": 4.459778992361618e-05,
+      "loss": 1.1317,
+      "step": 1120
+    },
+    {
+      "epoch": 0.2900979886539453,
+      "grad_norm": 1.890625,
+      "learning_rate": 4.452654502775486e-05,
+      "loss": 1.0472,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2913873130479629,
+      "grad_norm": 1.84375,
+      "learning_rate": 4.44550639844059e-05,
+      "loss": 1.0344,
+      "step": 1130
+    },
+    {
+      "epoch": 0.2926766374419804,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.438334798780285e-05,
+      "loss": 1.1265,
+      "step": 1135
+    },
+    {
+      "epoch": 0.29396596183599794,
+      "grad_norm": 2.015625,
+      "learning_rate": 4.43113982361046e-05,
+      "loss": 1.143,
+      "step": 1140
+    },
+    {
+      "epoch": 0.29525528623001546,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.4239215931375395e-05,
+      "loss": 1.2072,
+      "step": 1145
+    },
+    {
+      "epoch": 0.29654461062403303,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.416680227956475e-05,
+      "loss": 1.181,
+      "step": 1150
+    },
+    {
+      "epoch": 0.29783393501805056,
+      "grad_norm": 1.9140625,
+      "learning_rate": 4.409415849048729e-05,
+      "loss": 1.1152,
+      "step": 1155
+    },
+    {
+      "epoch": 0.2991232594120681,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.4021285777802536e-05,
+      "loss": 1.076,
+      "step": 1160
+    },
+    {
+      "epoch": 0.3004125838060856,
+      "grad_norm": 1.7734375,
+      "learning_rate": 4.394818535899465e-05,
+      "loss": 1.0386,
+      "step": 1165
+    },
+    {
+      "epoch": 0.3017019082001031,
+      "grad_norm": 1.8984375,
+      "learning_rate": 4.3874858455352065e-05,
+      "loss": 1.1222,
+      "step": 1170
+    },
+    {
+      "epoch": 0.3029912325941207,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.380130629194709e-05,
+      "loss": 1.1347,
+      "step": 1175
+    },
+    {
+      "epoch": 0.3042805569881382,
+      "grad_norm": 1.8359375,
+      "learning_rate": 4.372753009761548e-05,
+      "loss": 1.0434,
+      "step": 1180
+    },
+    {
+      "epoch": 0.30556988138215574,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.365353110493583e-05,
+      "loss": 1.1653,
+      "step": 1185
+    },
+    {
+      "epoch": 0.30685920577617326,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.357931055020906e-05,
+      "loss": 1.0697,
+      "step": 1190
+    },
+    {
+      "epoch": 0.30814853017019084,
+      "grad_norm": 1.7578125,
+      "learning_rate": 4.3504869673437725e-05,
+      "loss": 1.1026,
+      "step": 1195
+    },
+    {
+      "epoch": 0.30943785456420836,
+      "grad_norm": 1.75,
+      "learning_rate": 4.343020971830526e-05,
+      "loss": 1.0369,
+      "step": 1200
+    },
+    {
+      "epoch": 0.3107271789582259,
+      "grad_norm": 2.0,
+      "learning_rate": 4.33553319321553e-05,
+      "loss": 1.1722,
+      "step": 1205
+    },
+    {
+      "epoch": 0.3120165033522434,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.328023756597074e-05,
+      "loss": 1.0676,
+      "step": 1210
+    },
+    {
+      "epoch": 0.313305827746261,
+      "grad_norm": 2.046875,
+      "learning_rate": 4.320492787435289e-05,
+      "loss": 1.1024,
+      "step": 1215
+    },
+    {
+      "epoch": 0.3145951521402785,
+      "grad_norm": 1.7421875,
+      "learning_rate": 4.3129404115500496e-05,
+      "loss": 1.0666,
+      "step": 1220
+    },
+    {
+      "epoch": 0.315884476534296,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.305366755118875e-05,
+      "loss": 1.0873,
+      "step": 1225
+    },
+    {
+      "epoch": 0.31717380092831354,
+      "grad_norm": 1.8828125,
+      "learning_rate": 4.297771944674813e-05,
+      "loss": 1.1047,
+      "step": 1230
+    },
+    {
+      "epoch": 0.3184631253223311,
+      "grad_norm": 2.0625,
+      "learning_rate": 4.290156107104338e-05,
+      "loss": 1.143,
+      "step": 1235
+    },
+    {
+      "epoch": 0.31975244971634864,
+      "grad_norm": 1.921875,
+      "learning_rate": 4.2825193696452206e-05,
+      "loss": 1.1102,
+      "step": 1240
+    },
+    {
+      "epoch": 0.32104177411036616,
+      "grad_norm": 1.7265625,
+      "learning_rate": 4.274861859884404e-05,
+      "loss": 1.0131,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3223310985043837,
+      "grad_norm": 1.8671875,
+      "learning_rate": 4.2671837057558804e-05,
+      "loss": 1.0149,
+      "step": 1250
+    },
+    {
+      "epoch": 0.32362042289840126,
+      "grad_norm": 1.78125,
+      "learning_rate": 4.25948503553854e-05,
+      "loss": 1.0634,
+      "step": 1255
+    },
+    {
+      "epoch": 0.3249097472924188,
+      "grad_norm": 2.140625,
+      "learning_rate": 4.2517659778540415e-05,
+      "loss": 1.1028,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3261990716864363,
+      "grad_norm": 1.7421875,
+      "learning_rate": 4.244026661664653e-05,
+      "loss": 1.0244,
+      "step": 1265
+    },
+    {
+      "epoch": 0.3274883960804538,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.2362672162711035e-05,
+      "loss": 1.0678,
+      "step": 1270
+    },
+    {
+      "epoch": 0.3287777204744714,
+      "grad_norm": 1.8515625,
+      "learning_rate": 4.228487771310418e-05,
+      "loss": 1.0802,
+      "step": 1275
+    },
+    {
+      "epoch": 0.3300670448684889,
+      "grad_norm": 1.9765625,
+      "learning_rate": 4.220688456753759e-05,
+      "loss": 1.1543,
+      "step": 1280
+    },
+    {
+      "epoch": 0.33135636926250644,
+      "grad_norm": 1.875,
+      "learning_rate": 4.2128694029042434e-05,
+      "loss": 1.1074,
+      "step": 1285
+    },
+    {
+      "epoch": 0.33264569365652397,
+      "grad_norm": 1.7890625,
+      "learning_rate": 4.205030740394778e-05,
+      "loss": 1.0334,
+      "step": 1290
+    },
+    {
+      "epoch": 0.33393501805054154,
+      "grad_norm": 1.8203125,
+      "learning_rate": 4.197172600185871e-05,
+      "loss": 1.0696,
+      "step": 1295
+    },
+    {
+      "epoch": 0.33522434244455906,
+      "grad_norm": 2.109375,
+      "learning_rate": 4.18929511356344e-05,
+      "loss": 1.1449,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3365136668385766,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.181398412136629e-05,
+      "loss": 1.0391,
+      "step": 1305
+    },
+    {
+      "epoch": 0.3378029912325941,
+      "grad_norm": 1.90625,
+      "learning_rate": 4.173482627835598e-05,
+      "loss": 1.1672,
+      "step": 1310
+    },
+    {
+      "epoch": 0.33909231562661163,
+      "grad_norm": 1.703125,
+      "learning_rate": 4.16554789290933e-05,
+      "loss": 1.0117,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3403816400206292,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.1575943399234106e-05,
+      "loss": 1.1189,
+      "step": 1320
+    },
+    {
+      "epoch": 0.3416709644146467,
+      "grad_norm": 1.9140625,
+      "learning_rate": 4.149622101757823e-05,
+      "loss": 1.1284,
+      "step": 1325
+    },
+    {
+      "epoch": 0.34296028880866425,
+      "grad_norm": 1.71875,
+      "learning_rate": 4.1416313116047214e-05,
+      "loss": 1.1232,
+      "step": 1330
+    },
+    {
+      "epoch": 0.34424961320268177,
+      "grad_norm": 2.125,
+      "learning_rate": 4.13362210296621e-05,
+      "loss": 1.0743,
+      "step": 1335
+    },
+    {
+      "epoch": 0.34553893759669935,
+      "grad_norm": 1.765625,
+      "learning_rate": 4.1255946096521076e-05,
+      "loss": 1.1301,
+      "step": 1340
+    },
+    {
+      "epoch": 0.34682826199071687,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.117548965777716e-05,
+      "loss": 1.01,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3481175863847344,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.109485305761581e-05,
+      "loss": 1.0666,
+      "step": 1350
+    },
+    {
+      "epoch": 0.3494069107787519,
+      "grad_norm": 1.9453125,
+      "learning_rate": 4.1014037643232394e-05,
+      "loss": 1.1641,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3506962351727695,
+      "grad_norm": 1.625,
+      "learning_rate": 4.093304476480978e-05,
+      "loss": 1.0863,
+      "step": 1360
+    },
+    {
+      "epoch": 0.351985559566787,
+      "grad_norm": 1.8671875,
+      "learning_rate": 4.085187577549568e-05,
+      "loss": 1.0774,
+      "step": 1365
+    },
+    {
+      "epoch": 0.35327488396080453,
+      "grad_norm": 1.984375,
+      "learning_rate": 4.077053203138012e-05,
+      "loss": 1.0408,
+      "step": 1370
+    },
+    {
+      "epoch": 0.35456420835482205,
+      "grad_norm": 1.7109375,
+      "learning_rate": 4.068901489147275e-05,
+      "loss": 1.0915,
+      "step": 1375
+    },
+    {
+      "epoch": 0.35585353274883963,
+      "grad_norm": 2.0625,
+      "learning_rate": 4.060732571768012e-05,
+      "loss": 1.0972,
+      "step": 1380
+    },
+    {
+      "epoch": 0.35714285714285715,
+      "grad_norm": 1.921875,
+      "learning_rate": 4.052546587478298e-05,
+      "loss": 1.0948,
+      "step": 1385
+    },
+    {
+      "epoch": 0.35843218153687467,
+      "grad_norm": 1.671875,
+      "learning_rate": 4.044343673041344e-05,
+      "loss": 1.0768,
+      "step": 1390
+    },
+    {
+      "epoch": 0.3597215059308922,
+      "grad_norm": 1.65625,
+      "learning_rate": 4.036123965503211e-05,
+      "loss": 1.0934,
+      "step": 1395
+    },
+    {
+      "epoch": 0.36101083032490977,
+      "grad_norm": 1.96875,
+      "learning_rate": 4.027887602190524e-05,
+      "loss": 1.0385,
+      "step": 1400
+    },
+    {
+      "epoch": 0.3623001547189273,
+      "grad_norm": 1.890625,
+      "learning_rate": 4.019634720708178e-05,
+      "loss": 1.1174,
+      "step": 1405
+    },
+    {
+      "epoch": 0.3635894791129448,
+      "grad_norm": 1.6875,
+      "learning_rate": 4.0113654589370305e-05,
+      "loss": 1.0295,
+      "step": 1410
+    },
+    {
+      "epoch": 0.36487880350696233,
+      "grad_norm": 1.8125,
+      "learning_rate": 4.003079955031613e-05,
+      "loss": 1.1139,
+      "step": 1415
+    },
+    {
+      "epoch": 0.3661681279009799,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.994778347417806e-05,
+      "loss": 1.0385,
+      "step": 1420
+    },
+    {
+      "epoch": 0.36745745229499743,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3.9864607747905403e-05,
+      "loss": 1.1079,
+      "step": 1425
+    },
+    {
+      "epoch": 0.36874677668901495,
+      "grad_norm": 1.9375,
+      "learning_rate": 3.978127376111474e-05,
+      "loss": 1.0183,
+      "step": 1430
+    },
+    {
+      "epoch": 0.3700361010830325,
+      "grad_norm": 1.796875,
+      "learning_rate": 3.969778290606666e-05,
+      "loss": 1.0554,
+      "step": 1435
+    },
+    {
+      "epoch": 0.37132542547705005,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.96141365776426e-05,
+      "loss": 1.0603,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3726147498710676,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.953033617332148e-05,
+      "loss": 1.1558,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3739040742650851,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.944638309315634e-05,
+      "loss": 1.0739,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3751933986591026,
+      "grad_norm": 2.0,
+      "learning_rate": 3.9362278739751e-05,
+      "loss": 1.0382,
+      "step": 1455
+    },
+    {
+      "epoch": 0.37648272305312014,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.92780245182366e-05,
+      "loss": 0.9809,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3777720474471377,
+      "grad_norm": 1.96875,
+      "learning_rate": 3.919362183624811e-05,
+      "loss": 1.0984,
+      "step": 1465
+    },
+    {
+      "epoch": 0.37906137184115524,
+      "grad_norm": 1.8359375,
+      "learning_rate": 3.9109072103900854e-05,
+      "loss": 0.9802,
+      "step": 1470
+    },
+    {
+      "epoch": 0.38035069623517276,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.9024376733766895e-05,
+      "loss": 1.0936,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3816400206291903,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.8939537140851486e-05,
+      "loss": 1.0586,
+      "step": 1480
+    },
+    {
+      "epoch": 0.38292934502320786,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3.885455474256942e-05,
+      "loss": 1.0392,
+      "step": 1485
+    },
+    {
+      "epoch": 0.3842186694172254,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.87694309587213e-05,
+      "loss": 1.0565,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3855079938112429,
+      "grad_norm": 1.8671875,
+      "learning_rate": 3.868416721146989e-05,
+      "loss": 1.1257,
+      "step": 1495
+    },
+    {
+      "epoch": 0.3867973182052604,
+      "grad_norm": 1.8203125,
+      "learning_rate": 3.859876492531633e-05,
+      "loss": 1.0121,
+      "step": 1500
+    },
+    {
+      "epoch": 0.3867973182052604,
+      "eval_loss": 1.1112358570098877,
+      "eval_runtime": 5.9297,
+      "eval_samples_per_second": 10.287,
+      "eval_steps_per_second": 10.287,
+      "step": 1500
+    },
+    {
+      "epoch": 0.388086642599278,
+      "grad_norm": 2.046875,
+      "learning_rate": 3.851322552707628e-05,
+      "loss": 1.0925,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3893759669932955,
+      "grad_norm": 1.78125,
+      "learning_rate": 3.842755044585618e-05,
+      "loss": 1.1934,
+      "step": 1510
+    },
+    {
+      "epoch": 0.39066529138731304,
+      "grad_norm": 1.78125,
+      "learning_rate": 3.834174111302932e-05,
+      "loss": 1.0437,
+      "step": 1515
+    },
+    {
+      "epoch": 0.39195461578133056,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.8255798962211906e-05,
+      "loss": 1.0796,
+      "step": 1520
+    },
+    {
+      "epoch": 0.39324394017534814,
+      "grad_norm": 1.84375,
+      "learning_rate": 3.816972542923915e-05,
+      "loss": 1.1371,
+      "step": 1525
+    },
+    {
+      "epoch": 0.39453326456936566,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.808352195214126e-05,
+      "loss": 1.1143,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3958225889633832,
+      "grad_norm": 1.828125,
+      "learning_rate": 3.7997189971119424e-05,
+      "loss": 1.013,
+      "step": 1535
+    },
+    {
+      "epoch": 0.3971119133574007,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.791073092852174e-05,
+      "loss": 0.9783,
+      "step": 1540
+    },
+    {
+      "epoch": 0.3984012377514183,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.782414626881914e-05,
+      "loss": 1.0597,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3996905621454358,
+      "grad_norm": 1.984375,
+      "learning_rate": 3.773743743858121e-05,
+      "loss": 1.0715,
+      "step": 1550
+    },
+    {
+      "epoch": 0.4009798865394533,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.7650605886452074e-05,
+      "loss": 1.0855,
+      "step": 1555
+    },
+    {
+      "epoch": 0.40226921093347084,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.756365306312618e-05,
+      "loss": 1.1035,
+      "step": 1560
+    },
+    {
+      "epoch": 0.4035585353274884,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.7476580421324e-05,
+      "loss": 1.0607,
+      "step": 1565
+    },
+    {
+      "epoch": 0.40484785972150594,
+      "grad_norm": 2.03125,
+      "learning_rate": 3.73893894157679e-05,
+      "loss": 1.0923,
+      "step": 1570
+    },
+    {
+      "epoch": 0.40613718411552346,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.7302081503157674e-05,
+      "loss": 1.0617,
+      "step": 1575
+    },
+    {
+      "epoch": 0.407426508509541,
+      "grad_norm": 1.96875,
+      "learning_rate": 3.721465814214632e-05,
+      "loss": 1.12,
+      "step": 1580
+    },
+    {
+      "epoch": 0.40871583290355856,
+      "grad_norm": 1.875,
+      "learning_rate": 3.712712079331561e-05,
+      "loss": 1.1134,
+      "step": 1585
+    },
+    {
+      "epoch": 0.4100051572975761,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.703947091915171e-05,
+      "loss": 1.042,
+      "step": 1590
+    },
+    {
+      "epoch": 0.4112944816915936,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.695170998402079e-05,
+      "loss": 1.1391,
+      "step": 1595
+    },
+    {
+      "epoch": 0.4125838060856111,
+      "grad_norm": 2.03125,
+      "learning_rate": 3.6863839454144464e-05,
+      "loss": 1.1358,
+      "step": 1600
+    },
+    {
+      "epoch": 0.41387313047962865,
+      "grad_norm": 1.7421875,
+      "learning_rate": 3.6775860797575356e-05,
+      "loss": 1.071,
+      "step": 1605
+    },
+    {
+      "epoch": 0.4151624548736462,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3.668777548417259e-05,
+      "loss": 1.0495,
+      "step": 1610
+    },
+    {
+      "epoch": 0.41645177926766375,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.6599584985577163e-05,
+      "loss": 1.0827,
+      "step": 1615
+    },
+    {
+      "epoch": 0.41774110366168127,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.651129077518743e-05,
+      "loss": 1.1104,
+      "step": 1620
+    },
+    {
+      "epoch": 0.4190304280556988,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3.642289432813446e-05,
+      "loss": 1.0891,
+      "step": 1625
+    },
+    {
+      "epoch": 0.42031975244971637,
+      "grad_norm": 1.703125,
+      "learning_rate": 3.633439712125737e-05,
+      "loss": 1.1566,
+      "step": 1630
+    },
+    {
+      "epoch": 0.4216090768437339,
+      "grad_norm": 1.6875,
+      "learning_rate": 3.624580063307868e-05,
+      "loss": 1.072,
+      "step": 1635
+    },
+    {
+      "epoch": 0.4228984012377514,
+      "grad_norm": 2.125,
+      "learning_rate": 3.6157106343779614e-05,
+      "loss": 1.114,
+      "step": 1640
+    },
+    {
+      "epoch": 0.42418772563176893,
+      "grad_norm": 1.6171875,
+      "learning_rate": 3.6068315735175344e-05,
+      "loss": 1.0469,
+      "step": 1645
+    },
+    {
+      "epoch": 0.4254770500257865,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.597943029069025e-05,
+      "loss": 1.1008,
+      "step": 1650
+    },
+    {
+      "epoch": 0.42676637441980403,
+      "grad_norm": 1.578125,
+      "learning_rate": 3.589045149533315e-05,
+      "loss": 1.0672,
+      "step": 1655
+    },
+    {
+      "epoch": 0.42805569881382155,
+      "grad_norm": 1.8359375,
+      "learning_rate": 3.5801380835672444e-05,
+      "loss": 1.0739,
+      "step": 1660
+    },
+    {
+      "epoch": 0.42934502320783907,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.571221979981133e-05,
+      "loss": 1.0629,
+      "step": 1665
+    },
+    {
+      "epoch": 0.43063434760185665,
+      "grad_norm": 1.8359375,
+      "learning_rate": 3.562296987736295e-05,
+      "loss": 1.0751,
+      "step": 1670
+    },
+    {
+      "epoch": 0.43192367199587417,
+      "grad_norm": 1.875,
+      "learning_rate": 3.553363255942539e-05,
+      "loss": 1.0313,
+      "step": 1675
+    },
+    {
+      "epoch": 0.4332129963898917,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3.5444209338556955e-05,
+      "loss": 1.0101,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4345023207839092,
+      "grad_norm": 2.046875,
+      "learning_rate": 3.535470170875104e-05,
+      "loss": 1.0779,
+      "step": 1685
+    },
+    {
+      "epoch": 0.4357916451779268,
+      "grad_norm": 2.140625,
+      "learning_rate": 3.526511116541132e-05,
+      "loss": 1.0378,
+      "step": 1690
+    },
+    {
+      "epoch": 0.4370809695719443,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.5175439205326684e-05,
+      "loss": 1.1487,
+      "step": 1695
+    },
+    {
+      "epoch": 0.43837029396596183,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3.508568732664624e-05,
+      "loss": 1.0822,
+      "step": 1700
+    },
+    {
+      "epoch": 0.43965961835997935,
+      "grad_norm": 1.8046875,
+      "learning_rate": 3.499585702885432e-05,
+      "loss": 1.0489,
+      "step": 1705
+    },
+    {
+      "epoch": 0.44094894275399693,
+      "grad_norm": 2.078125,
+      "learning_rate": 3.490594981274538e-05,
+      "loss": 1.1533,
+      "step": 1710
+    },
+    {
+      "epoch": 0.44223826714801445,
+      "grad_norm": 1.859375,
+      "learning_rate": 3.4815967180398966e-05,
+      "loss": 1.0537,
+      "step": 1715
+    },
+    {
+      "epoch": 0.443527591542032,
+      "grad_norm": 1.84375,
+      "learning_rate": 3.47259106351546e-05,
+      "loss": 0.9471,
+      "step": 1720
+    },
+    {
+      "epoch": 0.4448169159360495,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.463578168158666e-05,
+      "loss": 1.0691,
+      "step": 1725
+    },
+    {
+      "epoch": 0.44610624033006707,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.454558182547926e-05,
+      "loss": 1.0992,
+      "step": 1730
+    },
+    {
+      "epoch": 0.4473955647240846,
+      "grad_norm": 1.828125,
+      "learning_rate": 3.445531257380107e-05,
+      "loss": 1.0514,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4486848891181021,
+      "grad_norm": 1.7109375,
+      "learning_rate": 3.436497543468016e-05,
+      "loss": 1.0751,
+      "step": 1740
+    },
+    {
+      "epoch": 0.44997421351211964,
+      "grad_norm": 1.921875,
+      "learning_rate": 3.427457191737879e-05,
+      "loss": 1.0634,
+      "step": 1745
+    },
+    {
+      "epoch": 0.45126353790613716,
+      "grad_norm": 1.6015625,
+      "learning_rate": 3.418410353226821e-05,
+      "loss": 0.9649,
+      "step": 1750
+    },
+    {
+      "epoch": 0.45255286230015473,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3.4093571790803406e-05,
+      "loss": 1.045,
+      "step": 1755
+    },
+    {
+      "epoch": 0.45384218669417226,
+      "grad_norm": 1.8671875,
+      "learning_rate": 3.400297820549785e-05,
+      "loss": 1.0857,
+      "step": 1760
+    },
+    {
+      "epoch": 0.4551315110881898,
+      "grad_norm": 1.953125,
+      "learning_rate": 3.3912324289898276e-05,
+      "loss": 1.0083,
+      "step": 1765
+    },
+    {
+      "epoch": 0.4564208354822073,
+      "grad_norm": 1.640625,
+      "learning_rate": 3.3821611558559314e-05,
+      "loss": 1.0745,
+      "step": 1770
+    },
+    {
+      "epoch": 0.4577101598762249,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3.3730841527018246e-05,
+      "loss": 1.0229,
+      "step": 1775
+    },
+    {
+      "epoch": 0.4589994842702424,
+      "grad_norm": 1.75,
+      "learning_rate": 3.364001571176967e-05,
+      "loss": 1.0087,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4602888086642599,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.354913563024015e-05,
+      "loss": 1.0314,
+      "step": 1785
+    },
+    {
+      "epoch": 0.46157813305827744,
+      "grad_norm": 2.078125,
+      "learning_rate": 3.345820280076291e-05,
+      "loss": 1.0225,
+      "step": 1790
+    },
+    {
+      "epoch": 0.462867457452295,
+      "grad_norm": 2.078125,
+      "learning_rate": 3.3367218742552386e-05,
+      "loss": 1.0863,
+      "step": 1795
+    },
+    {
+      "epoch": 0.46415678184631254,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.3276184975678923e-05,
+      "loss": 0.9991,
+      "step": 1800
+    },
+    {
+      "epoch": 0.46544610624033006,
+      "grad_norm": 2.0,
+      "learning_rate": 3.318510302104335e-05,
+      "loss": 1.1004,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4667354306343476,
+      "grad_norm": 1.6328125,
+      "learning_rate": 3.3093974400351546e-05,
+      "loss": 1.0244,
+      "step": 1810
+    },
+    {
+      "epoch": 0.46802475502836516,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.3002800636089057e-05,
+      "loss": 1.0605,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4693140794223827,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.2911583251495654e-05,
+      "loss": 1.0927,
+      "step": 1820
+    },
+    {
+      "epoch": 0.4706034038164002,
+      "grad_norm": 1.9375,
+      "learning_rate": 3.282032377053984e-05,
+      "loss": 1.0402,
+      "step": 1825
+    },
+    {
+      "epoch": 0.4718927282104177,
+      "grad_norm": 2.0,
+      "learning_rate": 3.272902371789345e-05,
+      "loss": 1.1045,
+      "step": 1830
+    },
+    {
+      "epoch": 0.4731820526044353,
+      "grad_norm": 1.734375,
+      "learning_rate": 3.263768461890614e-05,
+      "loss": 1.0115,
+      "step": 1835
+    },
+    {
+      "epoch": 0.4744713769984528,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.2546307999579904e-05,
+      "loss": 0.9979,
+      "step": 1840
+    },
+    {
+      "epoch": 0.47576070139247034,
+      "grad_norm": 1.8125,
+      "learning_rate": 3.2454895386543614e-05,
+      "loss": 1.1165,
+      "step": 1845
+    },
+    {
+      "epoch": 0.47705002578648786,
+      "grad_norm": 2.234375,
+      "learning_rate": 3.2363448307027455e-05,
+      "loss": 1.1466,
+      "step": 1850
+    },
+    {
+      "epoch": 0.47833935018050544,
+      "grad_norm": 2.03125,
+      "learning_rate": 3.227196828883747e-05,
+      "loss": 1.0755,
+      "step": 1855
+    },
+    {
+      "epoch": 0.47962867457452296,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3.218045686033e-05,
+      "loss": 1.0108,
+      "step": 1860
+    },
+    {
+      "epoch": 0.4809179989685405,
+      "grad_norm": 1.9375,
+      "learning_rate": 3.2088915550386144e-05,
+      "loss": 1.038,
+      "step": 1865
+    },
+    {
+      "epoch": 0.482207323362558,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3.199734588838625e-05,
+      "loss": 1.015,
+      "step": 1870
+    },
+    {
+      "epoch": 0.4834966477565756,
+      "grad_norm": 2.125,
+      "learning_rate": 3.190574940418435e-05,
+      "loss": 1.0388,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4847859721505931,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3.181412762808256e-05,
+      "loss": 1.0731,
+      "step": 1880
+    },
+    {
+      "epoch": 0.4860752965446106,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3.172248209080558e-05,
+      "loss": 1.0242,
+      "step": 1885
+    },
+    {
+      "epoch": 0.48736462093862815,
+      "grad_norm": 1.8046875,
+      "learning_rate": 3.163081432347508e-05,
+      "loss": 1.0533,
+      "step": 1890
+    },
+    {
+      "epoch": 0.48865394533264567,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.1539125857584104e-05,
+      "loss": 1.0957,
+      "step": 1895
+    },
+    {
+      "epoch": 0.48994326972666324,
+      "grad_norm": 1.90625,
+      "learning_rate": 3.1447418224971553e-05,
+      "loss": 0.9937,
+      "step": 1900
+    },
+    {
+      "epoch": 0.49123259412068077,
+      "grad_norm": 1.8046875,
+      "learning_rate": 3.1355692957796505e-05,
+      "loss": 1.045,
+      "step": 1905
+    },
+    {
+      "epoch": 0.4925219185146983,
+      "grad_norm": 1.7890625,
+      "learning_rate": 3.1263951588512674e-05,
+      "loss": 0.9935,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4938112429087158,
+      "grad_norm": 1.84375,
+      "learning_rate": 3.1172195649842795e-05,
+      "loss": 1.1245,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4951005673027334,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.108042667475302e-05,
+      "loss": 1.0937,
+      "step": 1920
+    },
+    {
+      "epoch": 0.4963898916967509,
+      "grad_norm": 2.03125,
+      "learning_rate": 3.098864619642726e-05,
+      "loss": 1.0607,
+      "step": 1925
+    },
+    {
+      "epoch": 0.49767921609076843,
+      "grad_norm": 1.6953125,
+      "learning_rate": 3.0896855748241674e-05,
+      "loss": 1.074,
+      "step": 1930
+    },
+    {
+      "epoch": 0.49896854048478595,
+      "grad_norm": 2.03125,
+      "learning_rate": 3.080505686373894e-05,
+      "loss": 0.9997,
+      "step": 1935
+    },
+    {
+      "epoch": 0.5002578648788035,
+      "grad_norm": 1.84375,
+      "learning_rate": 3.0713251076602694e-05,
+      "loss": 1.0496,
+      "step": 1940
+    },
+    {
+      "epoch": 0.501547189272821,
+      "grad_norm": 1.96875,
+      "learning_rate": 3.06214399206319e-05,
+      "loss": 1.116,
+      "step": 1945
+    },
+    {
+      "epoch": 0.5028365136668386,
+      "grad_norm": 1.75,
+      "learning_rate": 3.052962492971522e-05,
+      "loss": 0.9968,
+      "step": 1950
+    },
+    {
+      "epoch": 0.5041258380608561,
+      "grad_norm": 1.890625,
+      "learning_rate": 3.0437807637805365e-05,
+      "loss": 1.0227,
+      "step": 1955
+    },
+    {
+      "epoch": 0.5054151624548736,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3.0345989578893525e-05,
+      "loss": 1.1049,
+      "step": 1960
+    },
+    {
+      "epoch": 0.5067044868488911,
+      "grad_norm": 1.875,
+      "learning_rate": 3.025417228698368e-05,
+      "loss": 1.0208,
+      "step": 1965
+    },
+    {
+      "epoch": 0.5079938112429088,
+      "grad_norm": 1.65625,
+      "learning_rate": 3.0162357296067e-05,
+      "loss": 1.0379,
+      "step": 1970
+    },
+    {
+      "epoch": 0.5092831356369263,
+      "grad_norm": 2.015625,
+      "learning_rate": 3.00705461400962e-05,
+      "loss": 1.1005,
+      "step": 1975
+    },
+    {
+      "epoch": 0.5105724600309438,
+      "grad_norm": 1.90625,
+      "learning_rate": 2.9978740352959958e-05,
+      "loss": 1.0733,
+      "step": 1980
+    },
+    {
+      "epoch": 0.5118617844249613,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.988694146845723e-05,
+      "loss": 0.9714,
+      "step": 1985
+    },
+    {
+      "epoch": 0.5131511088189789,
+      "grad_norm": 2.171875,
+      "learning_rate": 2.979515102027163e-05,
+      "loss": 1.0943,
+      "step": 1990
+    },
+    {
+      "epoch": 0.5144404332129964,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.970337054194588e-05,
+      "loss": 1.0451,
+      "step": 1995
+    },
+    {
+      "epoch": 0.5157297576070139,
+      "grad_norm": 1.890625,
+      "learning_rate": 2.96116015668561e-05,
+      "loss": 1.0278,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5157297576070139,
+      "eval_loss": 1.0737704038619995,
+      "eval_runtime": 5.9321,
+      "eval_samples_per_second": 10.283,
+      "eval_steps_per_second": 10.283,
+      "step": 2000
+    },
+    {
+      "epoch": 0.5170190820010314,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.9519845628186215e-05,
+      "loss": 0.9885,
+      "step": 2005
+    },
+    {
+      "epoch": 0.518308406395049,
+      "grad_norm": 1.828125,
+      "learning_rate": 2.942810425890239e-05,
+      "loss": 1.0262,
+      "step": 2010
+    },
+    {
+      "epoch": 0.5195977307890666,
+      "grad_norm": 1.765625,
+      "learning_rate": 2.9336378991727346e-05,
+      "loss": 1.1212,
+      "step": 2015
+    },
+    {
+      "epoch": 0.5208870551830841,
+      "grad_norm": 1.9140625,
+      "learning_rate": 2.9244671359114785e-05,
+      "loss": 1.1245,
+      "step": 2020
+    },
+    {
+      "epoch": 0.5221763795771016,
+      "grad_norm": 1.6953125,
+      "learning_rate": 2.9152982893223816e-05,
+      "loss": 1.094,
+      "step": 2025
+    },
+    {
+      "epoch": 0.5234657039711191,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.9061315125893318e-05,
+      "loss": 1.1426,
+      "step": 2030
+    },
+    {
+      "epoch": 0.5247550283651367,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.896966958861634e-05,
+      "loss": 1.0176,
+      "step": 2035
+    },
+    {
+      "epoch": 0.5260443527591542,
+      "grad_norm": 1.8671875,
+      "learning_rate": 2.8878047812514548e-05,
+      "loss": 1.0857,
+      "step": 2040
+    },
+    {
+      "epoch": 0.5273336771531717,
+      "grad_norm": 1.921875,
+      "learning_rate": 2.8786451328312642e-05,
+      "loss": 0.9217,
+      "step": 2045
+    },
+    {
+      "epoch": 0.5286230015471892,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.869488166631276e-05,
+      "loss": 1.0058,
+      "step": 2050
+    },
+    {
+      "epoch": 0.5299123259412069,
+      "grad_norm": 2.109375,
+      "learning_rate": 2.8603340356368896e-05,
+      "loss": 1.0062,
+      "step": 2055
+    },
+    {
+      "epoch": 0.5312016503352244,
+      "grad_norm": 1.96875,
+      "learning_rate": 2.851182892786143e-05,
+      "loss": 0.9938,
+      "step": 2060
+    },
+    {
+      "epoch": 0.5324909747292419,
+      "grad_norm": 1.890625,
+      "learning_rate": 2.8420348909671448e-05,
+      "loss": 1.0247,
+      "step": 2065
+    },
+    {
+      "epoch": 0.5337802991232594,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.832890183015528e-05,
+      "loss": 1.0371,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5350696235172769,
+      "grad_norm": 1.953125,
+      "learning_rate": 2.8237489217118988e-05,
+      "loss": 1.0502,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5363589479112945,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.814611259779276e-05,
+      "loss": 1.134,
+      "step": 2080
+    },
+    {
+      "epoch": 0.537648272305312,
+      "grad_norm": 1.9609375,
+      "learning_rate": 2.805477349880544e-05,
+      "loss": 1.0545,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5389375966993295,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.796347344615905e-05,
+      "loss": 1.04,
+      "step": 2090
+    },
+    {
+      "epoch": 0.5402269210933471,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.7872213965203245e-05,
+      "loss": 1.068,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5415162454873647,
+      "grad_norm": 1.890625,
+      "learning_rate": 2.778099658060983e-05,
+      "loss": 1.0346,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5428055698813822,
+      "grad_norm": 1.9140625,
+      "learning_rate": 2.7689822816347353e-05,
+      "loss": 1.0194,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5440948942753997,
+      "grad_norm": 1.6640625,
+      "learning_rate": 2.7598694195655555e-05,
+      "loss": 1.0642,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5453842186694172,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.750761224101997e-05,
+      "loss": 1.0599,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5466735430634347,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.7416578474146507e-05,
+      "loss": 1.0245,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5479628674574523,
+      "grad_norm": 2.15625,
+      "learning_rate": 2.732559441593599e-05,
+      "loss": 1.0541,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5492521918514698,
+      "grad_norm": 1.8671875,
+      "learning_rate": 2.7234661586458737e-05,
+      "loss": 1.0549,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5505415162454874,
+      "grad_norm": 1.6484375,
+      "learning_rate": 2.7143781504929226e-05,
+      "loss": 1.0084,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5518308406395049,
+      "grad_norm": 1.640625,
+      "learning_rate": 2.7052955689680653e-05,
+      "loss": 1.0363,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5531201650335225,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.6962185658139582e-05,
+      "loss": 0.9836,
+      "step": 2145
+    },
+    {
+      "epoch": 0.55440948942754,
+      "grad_norm": 1.8515625,
+      "learning_rate": 2.6871472926800613e-05,
+      "loss": 1.0215,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5556988138215575,
+      "grad_norm": 1.9765625,
+      "learning_rate": 2.678081901120104e-05,
+      "loss": 1.0312,
+      "step": 2155
+    },
+    {
+      "epoch": 0.556988138215575,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.6690225425895504e-05,
+      "loss": 0.9963,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5582774626095925,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.6599693684430685e-05,
+      "loss": 1.0511,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5595667870036101,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.6509225299320107e-05,
+      "loss": 1.0052,
+      "step": 2170
+    },
+    {
+      "epoch": 0.5608561113976276,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.641882178201874e-05,
+      "loss": 0.9411,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5621454357916452,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.6328484642897824e-05,
+      "loss": 1.0797,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5634347601856627,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.6238215391219643e-05,
+      "loss": 1.0059,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5647240845796803,
+      "grad_norm": 1.6796875,
+      "learning_rate": 2.614801553511224e-05,
+      "loss": 0.9014,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5660134089736978,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.6057886581544292e-05,
+      "loss": 1.1011,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5673027333677153,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.5967830036299933e-05,
+      "loss": 1.0821,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5685920577617328,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.5877847403953526e-05,
+      "loss": 1.0629,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5698813821557503,
+      "grad_norm": 2.03125,
+      "learning_rate": 2.578794018784458e-05,
+      "loss": 1.0789,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5711707065497679,
+      "grad_norm": 2.0,
+      "learning_rate": 2.5698109890052658e-05,
+      "loss": 1.0434,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5724600309437855,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.560835801137222e-05,
+      "loss": 1.0453,
+      "step": 2220
+    },
+    {
+      "epoch": 0.573749355337803,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.5518686051287572e-05,
+      "loss": 1.0478,
+      "step": 2225
+    },
+    {
+      "epoch": 0.5750386797318205,
+      "grad_norm": 1.859375,
+      "learning_rate": 2.5429095507947854e-05,
+      "loss": 1.0431,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5763280041258381,
+      "grad_norm": 1.875,
+      "learning_rate": 2.5339587878141955e-05,
+      "loss": 1.0132,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5776173285198556,
+      "grad_norm": 2.03125,
+      "learning_rate": 2.5250164657273497e-05,
+      "loss": 1.0215,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5789066529138731,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.5160827339335956e-05,
+      "loss": 1.0457,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5801959773078906,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.5071577416887566e-05,
+      "loss": 0.9901,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5814853017019082,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.4982416381026455e-05,
+      "loss": 1.1305,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5827746260959258,
+      "grad_norm": 2.0,
+      "learning_rate": 2.489334572136575e-05,
+      "loss": 1.015,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5840639504899433,
+      "grad_norm": 1.703125,
+      "learning_rate": 2.4804366926008652e-05,
+      "loss": 0.9877,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5853532748839608,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.4715481481523555e-05,
+      "loss": 1.038,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5866425992779783,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.4626690872919282e-05,
+      "loss": 0.9913,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5879319236719959,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.4537996583620226e-05,
+      "loss": 1.0046,
+      "step": 2280
+    },
+    {
+      "epoch": 0.5892212480660134,
+      "grad_norm": 1.75,
+      "learning_rate": 2.4449400095441535e-05,
+      "loss": 1.0308,
+      "step": 2285
+    },
+    {
+      "epoch": 0.5905105724600309,
+      "grad_norm": 1.828125,
+      "learning_rate": 2.4360902888564438e-05,
+      "loss": 1.0234,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5917998968540484,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.4272506441511473e-05,
+      "loss": 0.9995,
+      "step": 2295
+    },
+    {
+      "epoch": 0.5930892212480661,
+      "grad_norm": 2.03125,
+      "learning_rate": 2.4184212231121743e-05,
+      "loss": 1.0432,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5943785456420836,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.409602173252631e-05,
+      "loss": 1.0965,
+      "step": 2305
+    },
+    {
+      "epoch": 0.5956678700361011,
+      "grad_norm": 1.90625,
+      "learning_rate": 2.400793641912354e-05,
+      "loss": 0.9769,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5969571944301186,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.3919957762554435e-05,
+      "loss": 1.0718,
+      "step": 2315
+    },
+    {
+      "epoch": 0.5982465188241362,
+      "grad_norm": 1.828125,
+      "learning_rate": 2.3832087232678104e-05,
+      "loss": 1.02,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5995358432181537,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.374432629754718e-05,
+      "loss": 0.9902,
+      "step": 2325
+    },
+    {
+      "epoch": 0.6008251676121712,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.3656676423383297e-05,
+      "loss": 1.0366,
+      "step": 2330
+    },
+    {
+      "epoch": 0.6021144920061887,
+      "grad_norm": 2.09375,
+      "learning_rate": 2.3569139074552582e-05,
+      "loss": 1.0425,
+      "step": 2335
+    },
+    {
+      "epoch": 0.6034038164002062,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.3481715713541222e-05,
+      "loss": 1.0575,
+      "step": 2340
+    },
+    {
+      "epoch": 0.6046931407942239,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.3394407800930997e-05,
+      "loss": 0.9437,
+      "step": 2345
+    },
+    {
+      "epoch": 0.6059824651882414,
+      "grad_norm": 1.7734375,
+      "learning_rate": 2.330721679537489e-05,
+      "loss": 1.0057,
+      "step": 2350
+    },
+    {
+      "epoch": 0.6072717895822589,
+      "grad_norm": 1.71875,
+      "learning_rate": 2.3220144153572727e-05,
+      "loss": 0.9554,
+      "step": 2355
+    },
+    {
+      "epoch": 0.6085611139762764,
+      "grad_norm": 1.8203125,
+      "learning_rate": 2.3133191330246826e-05,
+      "loss": 1.0164,
+      "step": 2360
+    },
+    {
+      "epoch": 0.609850438370294,
+      "grad_norm": 1.8984375,
+      "learning_rate": 2.3046359778117684e-05,
+      "loss": 1.0752,
+      "step": 2365
+    },
+    {
+      "epoch": 0.6111397627643115,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.2959650947879754e-05,
+      "loss": 0.9909,
+      "step": 2370
+    },
+    {
+      "epoch": 0.612429087158329,
+      "grad_norm": 2.125,
+      "learning_rate": 2.2873066288177156e-05,
+      "loss": 1.0401,
+      "step": 2375
+    },
+    {
+      "epoch": 0.6137184115523465,
+      "grad_norm": 1.7890625,
+      "learning_rate": 2.278660724557947e-05,
+      "loss": 1.061,
+      "step": 2380
+    },
+    {
+      "epoch": 0.6150077359463642,
+      "grad_norm": 2.1875,
+      "learning_rate": 2.2700275264557638e-05,
+      "loss": 1.1276,
+      "step": 2385
+    },
+    {
+      "epoch": 0.6162970603403817,
+      "grad_norm": 1.734375,
+      "learning_rate": 2.2614071787459753e-05,
+      "loss": 1.0116,
+      "step": 2390
+    },
+    {
+      "epoch": 0.6175863847343992,
+      "grad_norm": 2.125,
+      "learning_rate": 2.252799825448699e-05,
+      "loss": 1.0583,
+      "step": 2395
+    },
+    {
+      "epoch": 0.6188757091284167,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.244205610366957e-05,
+      "loss": 1.0676,
+      "step": 2400
+    },
+    {
+      "epoch": 0.6201650335224342,
+      "grad_norm": 1.9453125,
+      "learning_rate": 2.2356246770842717e-05,
+      "loss": 1.0411,
+      "step": 2405
+    },
+    {
+      "epoch": 0.6214543579164518,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.2270571689622625e-05,
+      "loss": 1.0155,
+      "step": 2410
+    },
+    {
+      "epoch": 0.6227436823104693,
+      "grad_norm": 2.03125,
+      "learning_rate": 2.2185032291382576e-05,
+      "loss": 1.0121,
+      "step": 2415
+    },
+    {
+      "epoch": 0.6240330067044868,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.2099630005229015e-05,
+      "loss": 1.1067,
+      "step": 2420
+    },
+    {
+      "epoch": 0.6253223310985044,
+      "grad_norm": 1.9140625,
+      "learning_rate": 2.2014366257977607e-05,
+      "loss": 0.9744,
+      "step": 2425
+    },
+    {
+      "epoch": 0.626611655492522,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.192924247412948e-05,
+      "loss": 1.0325,
+      "step": 2430
+    },
+    {
+      "epoch": 0.6279009798865395,
+      "grad_norm": 1.953125,
+      "learning_rate": 2.184426007584741e-05,
+      "loss": 1.0486,
+      "step": 2435
+    },
+    {
+      "epoch": 0.629190304280557,
+      "grad_norm": 2.140625,
+      "learning_rate": 2.1759420482932008e-05,
+      "loss": 1.0464,
+      "step": 2440
+    },
+    {
+      "epoch": 0.6304796286745745,
+      "grad_norm": 1.7109375,
+      "learning_rate": 2.1674725112798042e-05,
+      "loss": 1.0279,
+      "step": 2445
+    },
+    {
+      "epoch": 0.631768953068592,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.1590175380450785e-05,
+      "loss": 1.03,
+      "step": 2450
+    },
+    {
+      "epoch": 0.6330582774626096,
+      "grad_norm": 1.6875,
+      "learning_rate": 2.15057726984623e-05,
+      "loss": 1.145,
+      "step": 2455
+    },
+    {
+      "epoch": 0.6343476018566271,
+      "grad_norm": 1.78125,
+      "learning_rate": 2.1421518476947895e-05,
+      "loss": 1.0365,
+      "step": 2460
+    },
+    {
+      "epoch": 0.6356369262506446,
+      "grad_norm": 2.03125,
+      "learning_rate": 2.1337414123542552e-05,
+      "loss": 1.0355,
+      "step": 2465
+    },
+    {
+      "epoch": 0.6369262506446622,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.1253461043377427e-05,
+      "loss": 1.041,
+      "step": 2470
+    },
+    {
+      "epoch": 0.6382155750386798,
+      "grad_norm": 1.796875,
+      "learning_rate": 2.116966063905629e-05,
+      "loss": 0.9655,
+      "step": 2475
+    },
+    {
+      "epoch": 0.6395048994326973,
+      "grad_norm": 1.84375,
+      "learning_rate": 2.1086014310632236e-05,
+      "loss": 1.0204,
+      "step": 2480
+    },
+    {
+      "epoch": 0.6407942238267148,
+      "grad_norm": 1.8671875,
+      "learning_rate": 2.1002523455584162e-05,
+      "loss": 1.0523,
+      "step": 2485
+    },
+    {
+      "epoch": 0.6420835482207323,
+      "grad_norm": 1.953125,
+      "learning_rate": 2.0919189468793493e-05,
+      "loss": 1.0599,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6433728726147498,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.083601374252084e-05,
+      "loss": 1.0395,
+      "step": 2495
+    },
+    {
+      "epoch": 0.6446621970087674,
+      "grad_norm": 1.7421875,
+      "learning_rate": 2.075299766638278e-05,
+      "loss": 0.9891,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6446621970087674,
+      "eval_loss": 1.046289324760437,
+      "eval_runtime": 5.9309,
+      "eval_samples_per_second": 10.285,
+      "eval_steps_per_second": 10.285,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6459515214027849,
+      "grad_norm": 1.8125,
+      "learning_rate": 2.0670142627328585e-05,
+      "loss": 1.0521,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6472408457968025,
+      "grad_norm": 2.0625,
+      "learning_rate": 2.0587450009617123e-05,
+      "loss": 1.0291,
+      "step": 2510
+    },
+    {
+      "epoch": 0.64853017019082,
+      "grad_norm": 1.9921875,
+      "learning_rate": 2.050492119479365e-05,
+      "loss": 1.059,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6498194945848376,
+      "grad_norm": 1.8515625,
+      "learning_rate": 2.042255756166679e-05,
+      "loss": 1.0072,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6511088189788551,
+      "grad_norm": 1.859375,
+      "learning_rate": 2.0340360486285464e-05,
+      "loss": 1.0762,
+      "step": 2525
+    },
+    {
+      "epoch": 0.6523981433728726,
+      "grad_norm": 1.7265625,
+      "learning_rate": 2.025833134191592e-05,
+      "loss": 1.107,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6536874677668901,
+      "grad_norm": 1.9296875,
+      "learning_rate": 2.0176471499018788e-05,
+      "loss": 1.0689,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6549767921609077,
+      "grad_norm": 1.8046875,
+      "learning_rate": 2.009478232522615e-05,
+      "loss": 1.0118,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6562661165549252,
+      "grad_norm": 2.03125,
+      "learning_rate": 2.0013265185318775e-05,
+      "loss": 1.1118,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6575554409489428,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.993192144120322e-05,
+      "loss": 1.0804,
+      "step": 2550
+    },
+    {
+      "epoch": 0.6588447653429603,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.985075245188912e-05,
+      "loss": 1.0594,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6601340897369778,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.9769759573466502e-05,
+      "loss": 1.0435,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6614234141309954,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.9688944159083095e-05,
+      "loss": 0.9609,
+      "step": 2565
+    },
+    {
+      "epoch": 0.6627127385250129,
+      "grad_norm": 1.828125,
+      "learning_rate": 1.9608307558921733e-05,
+      "loss": 1.017,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6640020629190304,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.9527851120177823e-05,
+      "loss": 0.9764,
+      "step": 2575
+    },
+    {
+      "epoch": 0.6652913873130479,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.9447576187036802e-05,
+      "loss": 1.0188,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6665807117070655,
+      "grad_norm": 1.9921875,
+      "learning_rate": 1.9367484100651682e-05,
+      "loss": 0.9877,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6678700361010831,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.928757619912067e-05,
+      "loss": 0.9892,
+      "step": 2590
+    },
+    {
+      "epoch": 0.6691593604951006,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.9207853817464797e-05,
+      "loss": 0.9466,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6704486848891181,
+      "grad_norm": 2.09375,
+      "learning_rate": 1.9128318287605603e-05,
+      "loss": 1.009,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6717380092831357,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.9048970938342917e-05,
+      "loss": 0.9773,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6730273336771532,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.8969813095332614e-05,
+      "loss": 1.0355,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6743166580711707,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.8890846081064497e-05,
+      "loss": 1.0589,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6756059824651882,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.8812071214840193e-05,
+      "loss": 1.0397,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6768953068592057,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.8733489812751114e-05,
+      "loss": 1.0045,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6781846312532233,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.865510318765646e-05,
+      "loss": 0.9839,
+      "step": 2630
+    },
+    {
+      "epoch": 0.6794739556472409,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.8576912649161312e-05,
+      "loss": 0.969,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6807632800412584,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.8498919503594713e-05,
+      "loss": 1.0719,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6820526044352759,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.8421125053987865e-05,
+      "loss": 1.0143,
+      "step": 2645
+    },
+    {
+      "epoch": 0.6833419288292935,
+      "grad_norm": 1.625,
+      "learning_rate": 1.8343530600052367e-05,
+      "loss": 0.9688,
+      "step": 2650
+    },
+    {
+      "epoch": 0.684631253223311,
+      "grad_norm": 2.296875,
+      "learning_rate": 1.8266137438158484e-05,
+      "loss": 1.0113,
+      "step": 2655
+    },
+    {
+      "epoch": 0.6859205776173285,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.818894686131349e-05,
+      "loss": 1.0184,
+      "step": 2660
+    },
+    {
+      "epoch": 0.687209902011346,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.81119601591401e-05,
+      "loss": 1.0351,
+      "step": 2665
+    },
+    {
+      "epoch": 0.6884992264053635,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.803517861785485e-05,
+      "loss": 1.0266,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6897885507993812,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.7958603520246697e-05,
+      "loss": 0.9812,
+      "step": 2675
+    },
+    {
+      "epoch": 0.6910778751933987,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.7882236145655516e-05,
+      "loss": 0.9585,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6923671995874162,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.7806077769950766e-05,
+      "loss": 1.0526,
+      "step": 2685
+    },
+    {
+      "epoch": 0.6936565239814337,
+      "grad_norm": 1.828125,
+      "learning_rate": 1.7730129665510155e-05,
+      "loss": 1.0162,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6949458483754513,
+      "grad_norm": 2.390625,
+      "learning_rate": 1.7654393101198393e-05,
+      "loss": 0.9874,
+      "step": 2695
+    },
+    {
+      "epoch": 0.6962351727694688,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.7578869342346008e-05,
+      "loss": 0.9908,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6975244971634863,
+      "grad_norm": 1.875,
+      "learning_rate": 1.750355965072816e-05,
+      "loss": 0.9867,
+      "step": 2705
+    },
+    {
+      "epoch": 0.6988138215575038,
+      "grad_norm": 2.0,
+      "learning_rate": 1.74284652845436e-05,
+      "loss": 1.0304,
+      "step": 2710
+    },
+    {
+      "epoch": 0.7001031459515215,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.7353587498393638e-05,
+      "loss": 0.9274,
+      "step": 2715
+    },
+    {
+      "epoch": 0.701392470345539,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.727892754326118e-05,
+      "loss": 0.937,
+      "step": 2720
+    },
+    {
+      "epoch": 0.7026817947395565,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.7204486666489832e-05,
+      "loss": 1.0404,
+      "step": 2725
+    },
+    {
+      "epoch": 0.703971119133574,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.7130266111763068e-05,
+      "loss": 1.0085,
+      "step": 2730
+    },
+    {
+      "epoch": 0.7052604435275915,
+      "grad_norm": 1.9921875,
+      "learning_rate": 1.7056267119083418e-05,
+      "loss": 1.0239,
+      "step": 2735
+    },
+    {
+      "epoch": 0.7065497679216091,
+      "grad_norm": 1.75,
+      "learning_rate": 1.698249092475181e-05,
+      "loss": 1.0884,
+      "step": 2740
+    },
+    {
+      "epoch": 0.7078390923156266,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.6908938761346837e-05,
+      "loss": 1.0603,
+      "step": 2745
+    },
+    {
+      "epoch": 0.7091284167096441,
+      "grad_norm": 3.078125,
+      "learning_rate": 1.683561185770425e-05,
+      "loss": 1.0534,
+      "step": 2750
+    },
+    {
+      "epoch": 0.7104177411036616,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.6762511438896357e-05,
+      "loss": 1.1067,
+      "step": 2755
+    },
+    {
+      "epoch": 0.7117070654976793,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.6689638726211603e-05,
+      "loss": 1.0621,
+      "step": 2760
+    },
+    {
+      "epoch": 0.7129963898916968,
+      "grad_norm": 1.640625,
+      "learning_rate": 1.6616994937134146e-05,
+      "loss": 0.9698,
+      "step": 2765
+    },
+    {
+      "epoch": 0.7142857142857143,
+      "grad_norm": 1.828125,
+      "learning_rate": 1.6544581285323504e-05,
+      "loss": 0.9792,
+      "step": 2770
+    },
+    {
+      "epoch": 0.7155750386797318,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.6472398980594305e-05,
+      "loss": 0.9448,
+      "step": 2775
+    },
+    {
+      "epoch": 0.7168643630737493,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.6400449228896048e-05,
+      "loss": 0.9549,
+      "step": 2780
+    },
+    {
+      "epoch": 0.7181536874677669,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.6328733232292987e-05,
+      "loss": 1.0743,
+      "step": 2785
+    },
+    {
+      "epoch": 0.7194430118617844,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.6257252188944034e-05,
+      "loss": 0.9276,
+      "step": 2790
+    },
+    {
+      "epoch": 0.7207323362558019,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.618600729308271e-05,
+      "loss": 0.9445,
+      "step": 2795
+    },
+    {
+      "epoch": 0.7220216606498195,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.6114999734997253e-05,
+      "loss": 1.0704,
+      "step": 2800
+    },
+    {
+      "epoch": 0.7233109850438371,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.604423070101067e-05,
+      "loss": 1.0213,
+      "step": 2805
+    },
+    {
+      "epoch": 0.7246003094378546,
+      "grad_norm": 1.7109375,
+      "learning_rate": 1.597370137346097e-05,
+      "loss": 0.9388,
+      "step": 2810
+    },
+    {
+      "epoch": 0.7258896338318721,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.5903412930681368e-05,
+      "loss": 0.9158,
+      "step": 2815
+    },
+    {
+      "epoch": 0.7271789582258896,
+      "grad_norm": 2.078125,
+      "learning_rate": 1.5833366546980637e-05,
+      "loss": 1.1032,
+      "step": 2820
+    },
+    {
+      "epoch": 0.7284682826199071,
+      "grad_norm": 1.875,
+      "learning_rate": 1.5763563392623466e-05,
+      "loss": 0.9779,
+      "step": 2825
+    },
+    {
+      "epoch": 0.7297576070139247,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.56940046338109e-05,
+      "loss": 0.9669,
+      "step": 2830
+    },
+    {
+      "epoch": 0.7310469314079422,
+      "grad_norm": 2.125,
+      "learning_rate": 1.5624691432660877e-05,
+      "loss": 1.0766,
+      "step": 2835
+    },
+    {
+      "epoch": 0.7323362558019598,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.5555624947188803e-05,
+      "loss": 0.969,
+      "step": 2840
+    },
+    {
+      "epoch": 0.7336255801959773,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.5486806331288205e-05,
+      "loss": 1.012,
+      "step": 2845
+    },
+    {
+      "epoch": 0.7349149045899949,
+      "grad_norm": 1.96875,
+      "learning_rate": 1.541823673471145e-05,
+      "loss": 1.0372,
+      "step": 2850
+    },
+    {
+      "epoch": 0.7362042289840124,
+      "grad_norm": 1.75,
+      "learning_rate": 1.534991730305056e-05,
+      "loss": 1.0502,
+      "step": 2855
+    },
+    {
+      "epoch": 0.7374935533780299,
+      "grad_norm": 1.625,
+      "learning_rate": 1.5281849177718033e-05,
+      "loss": 0.9381,
+      "step": 2860
+    },
+    {
+      "epoch": 0.7387828777720474,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.521403349592779e-05,
+      "loss": 1.0346,
+      "step": 2865
+    },
+    {
+      "epoch": 0.740072202166065,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.514647139067619e-05,
+      "loss": 0.9806,
+      "step": 2870
+    },
+    {
+      "epoch": 0.7413615265600825,
+      "grad_norm": 1.9453125,
+      "learning_rate": 1.5079163990723088e-05,
+      "loss": 1.0132,
+      "step": 2875
+    },
+    {
+      "epoch": 0.7426508509541001,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.5012112420572964e-05,
+      "loss": 1.046,
+      "step": 2880
+    },
+    {
+      "epoch": 0.7439401753481176,
+      "grad_norm": 1.734375,
+      "learning_rate": 1.494531780045618e-05,
+      "loss": 0.9981,
+      "step": 2885
+    },
+    {
+      "epoch": 0.7452294997421351,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.4878781246310215e-05,
+      "loss": 1.0359,
+      "step": 2890
+    },
+    {
+      "epoch": 0.7465188241361527,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.4812503869761036e-05,
+      "loss": 0.9655,
+      "step": 2895
+    },
+    {
+      "epoch": 0.7478081485301702,
+      "grad_norm": 1.7109375,
+      "learning_rate": 1.4746486778104545e-05,
+      "loss": 0.9679,
+      "step": 2900
+    },
+    {
+      "epoch": 0.7490974729241877,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.4680731074288052e-05,
+      "loss": 1.0166,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7503867973182052,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.4615237856891862e-05,
+      "loss": 0.9635,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7516761217122228,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.4550008220110944e-05,
+      "loss": 1.0444,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7529654461062403,
+      "grad_norm": 1.890625,
+      "learning_rate": 1.4485043253736602e-05,
+      "loss": 1.0087,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7542547705002579,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.442034404313829e-05,
+      "loss": 0.9099,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7555440948942754,
+      "grad_norm": 1.8359375,
+      "learning_rate": 1.4355911669245498e-05,
+      "loss": 1.0327,
+      "step": 2930
+    },
+    {
+      "epoch": 0.756833419288293,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.4291747208529667e-05,
+      "loss": 0.9717,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7581227436823105,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.4227851732986211e-05,
+      "loss": 0.9807,
+      "step": 2940
+    },
+    {
+      "epoch": 0.759412068076328,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.416422631011661e-05,
+      "loss": 1.0287,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7607013924703455,
+      "grad_norm": 1.828125,
+      "learning_rate": 1.4100872002910596e-05,
+      "loss": 0.9726,
+      "step": 2950
+    },
+    {
+      "epoch": 0.761990716864363,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.4037789869828342e-05,
+      "loss": 1.0324,
+      "step": 2955
+    },
+    {
+      "epoch": 0.7632800412583806,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.3974980964782822e-05,
+      "loss": 1.0568,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7645693656523982,
+      "grad_norm": 2.109375,
+      "learning_rate": 1.3912446337122194e-05,
+      "loss": 1.0554,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7658586900464157,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.3850187031612258e-05,
+      "loss": 0.9739,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7671480144404332,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.3788204088419005e-05,
+      "loss": 1.0046,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7684373388344508,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.3726498543091257e-05,
+      "loss": 1.0389,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7697266632284683,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.3665071426543339e-05,
+      "loss": 1.0316,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7710159876224858,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.360392376503787e-05,
+      "loss": 1.011,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7723053120165033,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.3543056580168609e-05,
+      "loss": 0.8865,
+      "step": 2995
+    },
+    {
+      "epoch": 0.7735946364105208,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.3482470888843401e-05,
+      "loss": 1.0311,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7735946364105208,
+      "eval_loss": 1.0309275388717651,
+      "eval_runtime": 5.9351,
+      "eval_samples_per_second": 10.278,
+      "eval_steps_per_second": 10.278,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7748839608045385,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.342216770326717e-05,
+      "loss": 0.9621,
+      "step": 3005
+    },
+    {
+      "epoch": 0.776173285198556,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.3362148030925034e-05,
+      "loss": 0.9984,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7774626095925735,
+      "grad_norm": 1.84375,
+      "learning_rate": 1.3302412874565446e-05,
+      "loss": 1.0217,
+      "step": 3015
+    },
+    {
+      "epoch": 0.778751933986591,
+      "grad_norm": 1.5703125,
+      "learning_rate": 1.3242963232183446e-05,
+      "loss": 0.9911,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7800412583806086,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.3183800097003996e-05,
+      "loss": 0.9811,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7813305827746261,
+      "grad_norm": 1.65625,
+      "learning_rate": 1.3124924457465393e-05,
+      "loss": 0.9311,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7826199071686436,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.3066337297202728e-05,
+      "loss": 0.9823,
+      "step": 3035
+    },
+    {
+      "epoch": 0.7839092315626611,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.3008039595031494e-05,
+      "loss": 0.9727,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7851985559566786,
+      "grad_norm": 1.953125,
+      "learning_rate": 1.2950032324931187e-05,
+      "loss": 0.9876,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7864878803506963,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.2892316456029063e-05,
+      "loss": 0.9482,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7877772047447138,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.283489295258394e-05,
+      "loss": 1.0071,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7890665291387313,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.2777762773970085e-05,
+      "loss": 1.0459,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7903558535327488,
+      "grad_norm": 2.0,
+      "learning_rate": 1.2720926874661193e-05,
+      "loss": 1.0633,
+      "step": 3065
+    },
+    {
+      "epoch": 0.7916451779267664,
+      "grad_norm": 1.984375,
+      "learning_rate": 1.2664386204214421e-05,
+      "loss": 1.0249,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7929345023207839,
+      "grad_norm": 1.8046875,
+      "learning_rate": 1.2608141707254567e-05,
+      "loss": 1.0072,
+      "step": 3075
+    },
+    {
+      "epoch": 0.7942238267148014,
+      "grad_norm": 1.71875,
+      "learning_rate": 1.2552194323458235e-05,
+      "loss": 0.9606,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7955131511088189,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.2496544987538164e-05,
+      "loss": 1.0381,
+      "step": 3085
+    },
+    {
+      "epoch": 0.7968024755028366,
+      "grad_norm": 1.703125,
+      "learning_rate": 1.2441194629227616e-05,
+      "loss": 1.0054,
+      "step": 3090
+    },
+    {
+      "epoch": 0.7980917998968541,
+      "grad_norm": 1.7578125,
+      "learning_rate": 1.2386144173264833e-05,
+      "loss": 0.9685,
+      "step": 3095
+    },
+    {
+      "epoch": 0.7993811242908716,
+      "grad_norm": 1.7109375,
+      "learning_rate": 1.2331394539377574e-05,
+      "loss": 0.9869,
+      "step": 3100
+    },
+    {
+      "epoch": 0.8006704486848891,
+      "grad_norm": 1.7109375,
+      "learning_rate": 1.2276946642267801e-05,
+      "loss": 0.9892,
+      "step": 3105
+    },
+    {
+      "epoch": 0.8019597730789066,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.2222801391596322e-05,
+      "loss": 0.9621,
+      "step": 3110
+    },
+    {
+      "epoch": 0.8032490974729242,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.2168959691967653e-05,
+      "loss": 1.0096,
+      "step": 3115
+    },
+    {
+      "epoch": 0.8045384218669417,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.2115422442914878e-05,
+      "loss": 0.954,
+      "step": 3120
+    },
+    {
+      "epoch": 0.8058277462609592,
+      "grad_norm": 1.7890625,
+      "learning_rate": 1.2062190538884626e-05,
+      "loss": 1.052,
+      "step": 3125
+    },
+    {
+      "epoch": 0.8071170706549768,
+      "grad_norm": 2.328125,
+      "learning_rate": 1.2009264869222126e-05,
+      "loss": 1.0579,
+      "step": 3130
+    },
+    {
+      "epoch": 0.8084063950489944,
+      "grad_norm": 1.765625,
+      "learning_rate": 1.1956646318156364e-05,
+      "loss": 0.9956,
+      "step": 3135
+    },
+    {
+      "epoch": 0.8096957194430119,
+      "grad_norm": 1.8515625,
+      "learning_rate": 1.190433576478528e-05,
+      "loss": 0.8616,
+      "step": 3140
+    },
+    {
+      "epoch": 0.8109850438370294,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.1852334083061102e-05,
+      "loss": 1.0347,
+      "step": 3145
+    },
+    {
+      "epoch": 0.8122743682310469,
+      "grad_norm": 1.625,
+      "learning_rate": 1.1800642141775746e-05,
+      "loss": 0.9059,
+      "step": 3150
+    },
+    {
+      "epoch": 0.8135636926250644,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.1749260804546288e-05,
+      "loss": 0.9367,
+      "step": 3155
+    },
+    {
+      "epoch": 0.814853017019082,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.1698190929800547e-05,
+      "loss": 1.0462,
+      "step": 3160
+    },
+    {
+      "epoch": 0.8161423414130995,
+      "grad_norm": 1.9140625,
+      "learning_rate": 1.1647433370762743e-05,
+      "loss": 1.0531,
+      "step": 3165
+    },
+    {
+      "epoch": 0.8174316658071171,
+      "grad_norm": 1.828125,
+      "learning_rate": 1.1596988975439234e-05,
+      "loss": 0.8727,
+      "step": 3170
+    },
+    {
+      "epoch": 0.8187209902011346,
+      "grad_norm": 1.9140625,
+      "learning_rate": 1.1546858586604354e-05,
+      "loss": 1.0065,
+      "step": 3175
+    },
+    {
+      "epoch": 0.8200103145951522,
+      "grad_norm": 1.90625,
+      "learning_rate": 1.1497043041786332e-05,
+      "loss": 1.0106,
+      "step": 3180
+    },
+    {
+      "epoch": 0.8212996389891697,
+      "grad_norm": 1.9765625,
+      "learning_rate": 1.1447543173253292e-05,
+      "loss": 1.052,
+      "step": 3185
+    },
+    {
+      "epoch": 0.8225889633831872,
+      "grad_norm": 2.0,
+      "learning_rate": 1.1398359807999372e-05,
+      "loss": 0.9672,
+      "step": 3190
+    },
+    {
+      "epoch": 0.8238782877772047,
+      "grad_norm": 1.9375,
+      "learning_rate": 1.1349493767730873e-05,
+      "loss": 0.9907,
+      "step": 3195
+    },
+    {
+      "epoch": 0.8251676121712223,
+      "grad_norm": 1.671875,
+      "learning_rate": 1.1300945868852562e-05,
+      "loss": 1.0021,
+      "step": 3200
+    },
+    {
+      "epoch": 0.8264569365652398,
+      "grad_norm": 1.921875,
+      "learning_rate": 1.1252716922454011e-05,
+      "loss": 1.0254,
+      "step": 3205
+    },
+    {
+      "epoch": 0.8277462609592573,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.1204807734296061e-05,
+      "loss": 0.9832,
+      "step": 3210
+    },
+    {
+      "epoch": 0.8290355853532749,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.1157219104797343e-05,
+      "loss": 0.9811,
+      "step": 3215
+    },
+    {
+      "epoch": 0.8303249097472925,
+      "grad_norm": 1.9296875,
+      "learning_rate": 1.1109951829020923e-05,
+      "loss": 1.0738,
+      "step": 3220
+    },
+    {
+      "epoch": 0.83161423414131,
+      "grad_norm": 1.8984375,
+      "learning_rate": 1.1063006696661005e-05,
+      "loss": 1.0695,
+      "step": 3225
+    },
+    {
+      "epoch": 0.8329035585353275,
+      "grad_norm": 1.9140625,
+      "learning_rate": 1.101638449202976e-05,
+      "loss": 0.9994,
+      "step": 3230
+    },
+    {
+      "epoch": 0.834192882929345,
+      "grad_norm": 1.7734375,
+      "learning_rate": 1.0970085994044191e-05,
+      "loss": 1.0161,
+      "step": 3235
+    },
+    {
+      "epoch": 0.8354822073233625,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.0924111976213137e-05,
+      "loss": 1.0409,
+      "step": 3240
+    },
+    {
+      "epoch": 0.8367715317173801,
+      "grad_norm": 1.8671875,
+      "learning_rate": 1.0878463206624355e-05,
+      "loss": 0.9981,
+      "step": 3245
+    },
+    {
+      "epoch": 0.8380608561113976,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.083314044793167e-05,
+      "loss": 0.9965,
+      "step": 3250
+    },
+    {
+      "epoch": 0.8393501805054152,
+      "grad_norm": 2.265625,
+      "learning_rate": 1.0788144457342251e-05,
+      "loss": 0.9594,
+      "step": 3255
+    },
+    {
+      "epoch": 0.8406395048994327,
+      "grad_norm": 2.046875,
+      "learning_rate": 1.0743475986603965e-05,
+      "loss": 1.0307,
+      "step": 3260
+    },
+    {
+      "epoch": 0.8419288292934503,
+      "grad_norm": 1.9140625,
+      "learning_rate": 1.0699135781992785e-05,
+      "loss": 0.9831,
+      "step": 3265
+    },
+    {
+      "epoch": 0.8432181536874678,
+      "grad_norm": 2.0625,
+      "learning_rate": 1.0655124584300354e-05,
+      "loss": 1.0567,
+      "step": 3270
+    },
+    {
+      "epoch": 0.8445074780814853,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.0611443128821596e-05,
+      "loss": 1.0023,
+      "step": 3275
+    },
+    {
+      "epoch": 0.8457968024755028,
+      "grad_norm": 1.8203125,
+      "learning_rate": 1.0568092145342435e-05,
+      "loss": 0.9446,
+      "step": 3280
+    },
+    {
+      "epoch": 0.8470861268695203,
+      "grad_norm": 1.796875,
+      "learning_rate": 1.0525072358127589e-05,
+      "loss": 0.9916,
+      "step": 3285
+    },
+    {
+      "epoch": 0.8483754512635379,
+      "grad_norm": 1.859375,
+      "learning_rate": 1.0482384485908503e-05,
+      "loss": 1.0467,
+      "step": 3290
+    },
+    {
+      "epoch": 0.8496647756575555,
+      "grad_norm": 1.828125,
+      "learning_rate": 1.0440029241871301e-05,
+      "loss": 1.0734,
+      "step": 3295
+    },
+    {
+      "epoch": 0.850954100051573,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.0398007333644897e-05,
+      "loss": 1.0442,
+      "step": 3300
+    },
+    {
+      "epoch": 0.8522434244455905,
+      "grad_norm": 1.7265625,
+      "learning_rate": 1.035631946328916e-05,
+      "loss": 1.0284,
+      "step": 3305
+    },
+    {
+      "epoch": 0.8535327488396081,
+      "grad_norm": 1.875,
+      "learning_rate": 1.0314966327283193e-05,
+      "loss": 0.9905,
+      "step": 3310
+    },
+    {
+      "epoch": 0.8548220732336256,
+      "grad_norm": 1.7421875,
+      "learning_rate": 1.0273948616513696e-05,
+      "loss": 1.0478,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8561113976276431,
+      "grad_norm": 2.171875,
+      "learning_rate": 1.0233267016263407e-05,
+      "loss": 1.0102,
+      "step": 3320
+    },
+    {
+      "epoch": 0.8574007220216606,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.019292220619969e-05,
+      "loss": 0.9438,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8586900464156781,
+      "grad_norm": 2.140625,
+      "learning_rate": 1.0152914860363142e-05,
+      "loss": 0.9739,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8599793708096957,
+      "grad_norm": 1.78125,
+      "learning_rate": 1.0113245647156345e-05,
+      "loss": 1.0104,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8612686952037133,
+      "grad_norm": 2.015625,
+      "learning_rate": 1.0073915229332698e-05,
+      "loss": 1.0052,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8625580195977308,
+      "grad_norm": 2.03125,
+      "learning_rate": 1.0034924263985362e-05,
+      "loss": 1.0722,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8638473439917483,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.996273402536247e-06,
+      "loss": 0.9685,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8651366683857659,
+      "grad_norm": 1.8046875,
+      "learning_rate": 9.95796329072517e-06,
+      "loss": 0.9873,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8664259927797834,
+      "grad_norm": 1.96875,
+      "learning_rate": 9.919994568599023e-06,
+      "loss": 1.0418,
+      "step": 3360
+    },
+    {
+      "epoch": 0.8677153171738009,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.882367870501118e-06,
+      "loss": 1.0223,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8690046415678184,
+      "grad_norm": 1.921875,
+      "learning_rate": 9.845083825060565e-06,
+      "loss": 1.0222,
+      "step": 3370
+    },
+    {
+      "epoch": 0.870293965961836,
+      "grad_norm": 2.0,
+      "learning_rate": 9.808143055181779e-06,
+      "loss": 0.9932,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8715832903558536,
+      "grad_norm": 2.03125,
+      "learning_rate": 9.771546178034068e-06,
+      "loss": 0.9962,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8728726147498711,
+      "grad_norm": 1.6953125,
+      "learning_rate": 9.735293805041332e-06,
+      "loss": 0.9477,
+      "step": 3385
+    },
+    {
+      "epoch": 0.8741619391438886,
+      "grad_norm": 1.8125,
+      "learning_rate": 9.699386541871843e-06,
+      "loss": 0.9873,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8754512635379061,
+      "grad_norm": 2.171875,
+      "learning_rate": 9.663824988428113e-06,
+      "loss": 1.0182,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8767405879319237,
+      "grad_norm": 1.90625,
+      "learning_rate": 9.628609738836891e-06,
+      "loss": 0.9724,
+      "step": 3400
+    },
+    {
+      "epoch": 0.8780299123259412,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.593741381439226e-06,
+      "loss": 0.9572,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8793192367199587,
+      "grad_norm": 1.7734375,
+      "learning_rate": 9.559220498780638e-06,
+      "loss": 0.9457,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8806085611139762,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.5250476676014e-06,
+      "loss": 1.0024,
+      "step": 3415
+    },
+    {
+      "epoch": 0.8818978855079939,
+      "grad_norm": 1.9765625,
+      "learning_rate": 9.491223458826881e-06,
+      "loss": 1.0743,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8831872099020114,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.45774843755801e-06,
+      "loss": 1.0251,
+      "step": 3425
+    },
+    {
+      "epoch": 0.8844765342960289,
+      "grad_norm": 2.0,
+      "learning_rate": 9.424623163061851e-06,
+      "loss": 1.0152,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8857658586900464,
+      "grad_norm": 1.8984375,
+      "learning_rate": 9.391848188762251e-06,
+      "loss": 1.0406,
+      "step": 3435
+    },
+    {
+      "epoch": 0.887055183084064,
+      "grad_norm": 2.0,
+      "learning_rate": 9.359424062230587e-06,
+      "loss": 0.9555,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8883445074780815,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.32735132517662e-06,
+      "loss": 1.0079,
+      "step": 3445
+    },
+    {
+      "epoch": 0.889633831872099,
+      "grad_norm": 2.078125,
+      "learning_rate": 9.29563051343947e-06,
+      "loss": 1.0293,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8909231562661165,
+      "grad_norm": 1.6484375,
+      "learning_rate": 9.26426215697861e-06,
+      "loss": 0.9844,
+      "step": 3455
+    },
+    {
+      "epoch": 0.8922124806601341,
+      "grad_norm": 2.0625,
+      "learning_rate": 9.23324677986507e-06,
+      "loss": 1.0013,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8935018050541517,
+      "grad_norm": 1.5859375,
+      "learning_rate": 9.202584900272644e-06,
+      "loss": 0.9861,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8947911294481692,
+      "grad_norm": 1.7734375,
+      "learning_rate": 9.172277030469241e-06,
+      "loss": 1.0327,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8960804538421867,
+      "grad_norm": 1.9296875,
+      "learning_rate": 9.142323676808336e-06,
+      "loss": 1.0442,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8973697782362042,
+      "grad_norm": 1.875,
+      "learning_rate": 9.112725339720507e-06,
+      "loss": 0.9581,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8986591026302218,
+      "grad_norm": 1.953125,
+      "learning_rate": 9.083482513705065e-06,
+      "loss": 1.0484,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8999484270242393,
+      "grad_norm": 1.7890625,
+      "learning_rate": 9.0545956873218e-06,
+      "loss": 1.0416,
+      "step": 3490
+    },
+    {
+      "epoch": 0.9012377514182568,
+      "grad_norm": 1.8203125,
+      "learning_rate": 9.026065343182814e-06,
+      "loss": 1.022,
+      "step": 3495
+    },
+    {
+      "epoch": 0.9025270758122743,
+      "grad_norm": 1.8515625,
+      "learning_rate": 8.997891957944469e-06,
+      "loss": 1.0437,
+      "step": 3500
+    },
+    {
+      "epoch": 0.9025270758122743,
+      "eval_loss": 1.0246747732162476,
+      "eval_runtime": 5.9432,
+      "eval_samples_per_second": 10.264,
+      "eval_steps_per_second": 10.264,
+      "step": 3500
+    },
+    {
+      "epoch": 0.903816400206292,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.970076002299412e-06,
+      "loss": 0.9608,
+      "step": 3505
+    },
+    {
+      "epoch": 0.9051057246003095,
+      "grad_norm": 1.8125,
+      "learning_rate": 8.942617940968718e-06,
+      "loss": 1.0789,
+      "step": 3510
+    },
+    {
+      "epoch": 0.906395048994327,
+      "grad_norm": 2.015625,
+      "learning_rate": 8.915518232694117e-06,
+      "loss": 1.0299,
+      "step": 3515
+    },
+    {
+      "epoch": 0.9076843733883445,
+      "grad_norm": 1.96875,
+      "learning_rate": 8.88877733023034e-06,
+      "loss": 1.0237,
+      "step": 3520
+    },
+    {
+      "epoch": 0.908973697782362,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.862395680337552e-06,
+      "loss": 0.9328,
+      "step": 3525
+    },
+    {
+      "epoch": 0.9102630221763796,
+      "grad_norm": 1.7421875,
+      "learning_rate": 8.836373723773878e-06,
+      "loss": 1.0397,
+      "step": 3530
+    },
+    {
+      "epoch": 0.9115523465703971,
+      "grad_norm": 1.8203125,
+      "learning_rate": 8.810711895288062e-06,
+      "loss": 0.9778,
+      "step": 3535
+    },
+    {
+      "epoch": 0.9128416709644146,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.785410623612173e-06,
+      "loss": 1.0516,
+      "step": 3540
+    },
+    {
+      "epoch": 0.9141309953584322,
+      "grad_norm": 1.890625,
+      "learning_rate": 8.760470331454475e-06,
+      "loss": 1.0376,
+      "step": 3545
+    },
+    {
+      "epoch": 0.9154203197524498,
+      "grad_norm": 2.09375,
+      "learning_rate": 8.735891435492338e-06,
+      "loss": 1.1082,
+      "step": 3550
+    },
+    {
+      "epoch": 0.9167096441464673,
+      "grad_norm": 1.71875,
+      "learning_rate": 8.71167434636529e-06,
+      "loss": 0.9414,
+      "step": 3555
+    },
+    {
+      "epoch": 0.9179989685404848,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.687819468668153e-06,
+      "loss": 1.0192,
+      "step": 3560
+    },
+    {
+      "epoch": 0.9192882929345023,
+      "grad_norm": 1.8125,
+      "learning_rate": 8.66432720094428e-06,
+      "loss": 0.9368,
+      "step": 3565
+    },
+    {
+      "epoch": 0.9205776173285198,
+      "grad_norm": 1.9296875,
+      "learning_rate": 8.641197935678909e-06,
+      "loss": 1.0606,
+      "step": 3570
+    },
+    {
+      "epoch": 0.9218669417225374,
+      "grad_norm": 1.8984375,
+      "learning_rate": 8.618432059292594e-06,
+      "loss": 1.0411,
+      "step": 3575
+    },
+    {
+      "epoch": 0.9231562661165549,
+      "grad_norm": 1.828125,
+      "learning_rate": 8.596029952134748e-06,
+      "loss": 1.073,
+      "step": 3580
+    },
+    {
+      "epoch": 0.9244455905105725,
+      "grad_norm": 1.796875,
+      "learning_rate": 8.5739919884773e-06,
+      "loss": 0.9726,
+      "step": 3585
+    },
+    {
+      "epoch": 0.92573491490459,
+      "grad_norm": 1.890625,
+      "learning_rate": 8.552318536508424e-06,
+      "loss": 1.0113,
+      "step": 3590
+    },
+    {
+      "epoch": 0.9270242392986076,
+      "grad_norm": 1.953125,
+      "learning_rate": 8.531009958326412e-06,
+      "loss": 0.9652,
+      "step": 3595
+    },
+    {
+      "epoch": 0.9283135636926251,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.510066609933596e-06,
+      "loss": 1.0442,
+      "step": 3600
+    },
+    {
+      "epoch": 0.9296028880866426,
+      "grad_norm": 1.9921875,
+      "learning_rate": 8.489488841230432e-06,
+      "loss": 0.9804,
+      "step": 3605
+    },
+    {
+      "epoch": 0.9308922124806601,
+      "grad_norm": 1.9609375,
+      "learning_rate": 8.469276996009629e-06,
+      "loss": 1.0182,
+      "step": 3610
+    },
+    {
+      "epoch": 0.9321815368746776,
+      "grad_norm": 1.8125,
+      "learning_rate": 8.449431411950405e-06,
+      "loss": 0.967,
+      "step": 3615
+    },
+    {
+      "epoch": 0.9334708612686952,
+      "grad_norm": 1.9921875,
+      "learning_rate": 8.429952420612871e-06,
+      "loss": 0.9747,
+      "step": 3620
+    },
+    {
+      "epoch": 0.9347601856627127,
+      "grad_norm": 2.078125,
+      "learning_rate": 8.41084034743246e-06,
+      "loss": 1.052,
+      "step": 3625
+    },
+    {
+      "epoch": 0.9360495100567303,
+      "grad_norm": 2.046875,
+      "learning_rate": 8.392095511714513e-06,
+      "loss": 1.0107,
+      "step": 3630
+    },
+    {
+      "epoch": 0.9373388344507478,
+      "grad_norm": 1.78125,
+      "learning_rate": 8.373718226628935e-06,
+      "loss": 0.9588,
+      "step": 3635
+    },
+    {
+      "epoch": 0.9386281588447654,
+      "grad_norm": 1.7578125,
+      "learning_rate": 8.355708799204965e-06,
+      "loss": 1.0021,
+      "step": 3640
+    },
+    {
+      "epoch": 0.9399174832387829,
+      "grad_norm": 1.9140625,
+      "learning_rate": 8.338067530326038e-06,
+      "loss": 0.934,
+      "step": 3645
+    },
+    {
+      "epoch": 0.9412068076328004,
+      "grad_norm": 1.890625,
+      "learning_rate": 8.320794714724771e-06,
+      "loss": 1.0268,
+      "step": 3650
+    },
+    {
+      "epoch": 0.9424961320268179,
+      "grad_norm": 2.03125,
+      "learning_rate": 8.303890640978039e-06,
+      "loss": 0.9972,
+      "step": 3655
+    },
+    {
+      "epoch": 0.9437854564208354,
+      "grad_norm": 1.96875,
+      "learning_rate": 8.287355591502135e-06,
+      "loss": 0.9224,
+      "step": 3660
+    },
+    {
+      "epoch": 0.945074780814853,
+      "grad_norm": 2.046875,
+      "learning_rate": 8.271189842548078e-06,
+      "loss": 0.9846,
+      "step": 3665
+    },
+    {
+      "epoch": 0.9463641052088706,
+      "grad_norm": 1.75,
+      "learning_rate": 8.255393664196975e-06,
+      "loss": 1.0072,
+      "step": 3670
+    },
+    {
+      "epoch": 0.9476534296028881,
+      "grad_norm": 1.9140625,
+      "learning_rate": 8.239967320355523e-06,
+      "loss": 1.0633,
+      "step": 3675
+    },
+    {
+      "epoch": 0.9489427539969056,
+      "grad_norm": 1.859375,
+      "learning_rate": 8.224911068751592e-06,
+      "loss": 0.923,
+      "step": 3680
+    },
+    {
+      "epoch": 0.9502320783909232,
+      "grad_norm": 1.734375,
+      "learning_rate": 8.210225160929927e-06,
+      "loss": 0.9009,
+      "step": 3685
+    },
+    {
+      "epoch": 0.9515214027849407,
+      "grad_norm": 1.953125,
+      "learning_rate": 8.195909842247936e-06,
+      "loss": 1.0314,
+      "step": 3690
+    },
+    {
+      "epoch": 0.9528107271789582,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.181965351871594e-06,
+      "loss": 1.024,
+      "step": 3695
+    },
+    {
+      "epoch": 0.9541000515729757,
+      "grad_norm": 1.7890625,
+      "learning_rate": 8.168391922771464e-06,
+      "loss": 1.0027,
+      "step": 3700
+    },
+    {
+      "epoch": 0.9553893759669932,
+      "grad_norm": 2.140625,
+      "learning_rate": 8.155189781718763e-06,
+      "loss": 1.0896,
+      "step": 3705
+    },
+    {
+      "epoch": 0.9566787003610109,
+      "grad_norm": 1.9453125,
+      "learning_rate": 8.142359149281627e-06,
+      "loss": 1.0547,
+      "step": 3710
+    },
+    {
+      "epoch": 0.9579680247550284,
+      "grad_norm": 1.84375,
+      "learning_rate": 8.129900239821378e-06,
+      "loss": 0.8805,
+      "step": 3715
+    },
+    {
+      "epoch": 0.9592573491490459,
+      "grad_norm": 1.671875,
+      "learning_rate": 8.117813261488977e-06,
+      "loss": 1.0267,
+      "step": 3720
+    },
+    {
+      "epoch": 0.9605466735430634,
+      "grad_norm": 1.984375,
+      "learning_rate": 8.106098416221526e-06,
+      "loss": 1.0638,
+      "step": 3725
+    },
+    {
+      "epoch": 0.961835997937081,
+      "grad_norm": 1.921875,
+      "learning_rate": 8.094755899738908e-06,
+      "loss": 0.9863,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9631253223310985,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.0837859015405e-06,
+      "loss": 1.0291,
+      "step": 3735
+    },
+    {
+      "epoch": 0.964414646725116,
+      "grad_norm": 2.09375,
+      "learning_rate": 8.073188604902021e-06,
+      "loss": 1.0061,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9657039711191335,
+      "grad_norm": 1.984375,
+      "learning_rate": 8.062964186872479e-06,
+      "loss": 0.999,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9669932955131512,
+      "grad_norm": 2.03125,
+      "learning_rate": 8.053112818271178e-06,
+      "loss": 0.993,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9682826199071687,
+      "grad_norm": 1.8984375,
+      "learning_rate": 8.043634663684905e-06,
+      "loss": 0.9366,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9695719443011862,
+      "grad_norm": 1.75,
+      "learning_rate": 8.034529881465155e-06,
+      "loss": 1.0075,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9708612686952037,
+      "grad_norm": 1.8671875,
+      "learning_rate": 8.02579862372549e-06,
+      "loss": 1.0466,
+      "step": 3765
+    },
+    {
+      "epoch": 0.9721505930892212,
+      "grad_norm": 1.765625,
+      "learning_rate": 8.017441036339002e-06,
+      "loss": 0.9364,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9734399174832388,
+      "grad_norm": 1.703125,
+      "learning_rate": 8.009457258935876e-06,
+      "loss": 0.9968,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9747292418772563,
+      "grad_norm": 1.8046875,
+      "learning_rate": 8.001847424901053e-06,
+      "loss": 0.9744,
+      "step": 3780
+    },
+    {
+      "epoch": 0.9760185662712738,
+      "grad_norm": 1.9375,
+      "learning_rate": 7.994611661371999e-06,
+      "loss": 1.0311,
+      "step": 3785
+    },
+    {
+      "epoch": 0.9773078906652913,
+      "grad_norm": 1.890625,
+      "learning_rate": 7.987750089236593e-06,
+      "loss": 0.9923,
+      "step": 3790
+    },
+    {
+      "epoch": 0.978597215059309,
+      "grad_norm": 1.9453125,
+      "learning_rate": 7.981262823131096e-06,
+      "loss": 0.9846,
+      "step": 3795
+    },
+    {
+      "epoch": 0.9798865394533265,
+      "grad_norm": 1.8984375,
+      "learning_rate": 7.97514997143824e-06,
+      "loss": 1.037,
+      "step": 3800
+    },
+    {
+      "epoch": 0.981175863847344,
+      "grad_norm": 1.8125,
+      "learning_rate": 7.969411636285418e-06,
+      "loss": 0.9694,
+      "step": 3805
+    },
+    {
+      "epoch": 0.9824651882413615,
+      "grad_norm": 1.8671875,
+      "learning_rate": 7.96404791354297e-06,
+      "loss": 1.095,
+      "step": 3810
+    },
+    {
+      "epoch": 0.983754512635379,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.959058892822598e-06,
+      "loss": 0.8972,
+      "step": 3815
+    },
+    {
+      "epoch": 0.9850438370293966,
+      "grad_norm": 1.7890625,
+      "learning_rate": 7.95444465747585e-06,
+      "loss": 1.0229,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9863331614234141,
+      "grad_norm": 1.609375,
+      "learning_rate": 7.950205284592742e-06,
+      "loss": 0.8876,
+      "step": 3825
+    },
+    {
+      "epoch": 0.9876224858174316,
+      "grad_norm": 1.890625,
+      "learning_rate": 7.946340845000463e-06,
+      "loss": 1.0351,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9889118102114492,
+      "grad_norm": 2.015625,
+      "learning_rate": 7.942851403262187e-06,
+      "loss": 1.0178,
+      "step": 3835
+    },
+    {
+      "epoch": 0.9902011346054668,
+      "grad_norm": 1.7578125,
+      "learning_rate": 7.939737017676008e-06,
+      "loss": 1.0092,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9914904589994843,
+      "grad_norm": 1.8359375,
+      "learning_rate": 7.936997740273952e-06,
+      "loss": 0.9306,
+      "step": 3845
+    },
+    {
+      "epoch": 0.9927797833935018,
+      "grad_norm": 1.875,
+      "learning_rate": 7.934633616821124e-06,
+      "loss": 1.0749,
+      "step": 3850
+    },
+    {
+      "epoch": 0.9940691077875193,
+      "grad_norm": 1.7734375,
+      "learning_rate": 7.932644686814917e-06,
+      "loss": 1.0432,
+      "step": 3855
+    },
+    {
+      "epoch": 0.9953584321815369,
+      "grad_norm": 2.109375,
+      "learning_rate": 7.931030983484384e-06,
+      "loss": 1.0019,
+      "step": 3860
+    },
+    {
+      "epoch": 0.9966477565755544,
+      "grad_norm": 1.8984375,
+      "learning_rate": 7.92979253378966e-06,
+      "loss": 1.0068,
+      "step": 3865
+    },
+    {
+      "epoch": 0.9979370809695719,
+      "grad_norm": 1.6484375,
+      "learning_rate": 7.928929358421515e-06,
+      "loss": 0.964,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9992264053635895,
+      "grad_norm": 1.90625,
+      "learning_rate": 7.92844147180102e-06,
+      "loss": 1.0317,
+      "step": 3875
+    },
+    {
+      "epoch": 0.9997421351211965,
+      "eval_loss": 1.0225045680999756,
+      "eval_runtime": 5.9883,
+      "eval_samples_per_second": 10.187,
+      "eval_steps_per_second": 10.187,
+      "step": 3877
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 3878,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 7.715935864529879e+17,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}