diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,35033 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 5.859141341051616,
+  "eval_steps": 500,
+  "global_step": 25000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00120598166907863,
+      "grad_norm": 17.625,
+      "learning_rate": 3e-06,
+      "loss": 10.4116,
+      "step": 5
+    },
+    {
+      "epoch": 0.00241196333815726,
+      "grad_norm": 15.625,
+      "learning_rate": 6e-06,
+      "loss": 10.3625,
+      "step": 10
+    },
+    {
+      "epoch": 0.00361794500723589,
+      "grad_norm": 16.875,
+      "learning_rate": 9e-06,
+      "loss": 10.2274,
+      "step": 15
+    },
+    {
+      "epoch": 0.00482392667631452,
+      "grad_norm": 14.625,
+      "learning_rate": 1.2e-05,
+      "loss": 9.9835,
+      "step": 20
+    },
+    {
+      "epoch": 0.00602990834539315,
+      "grad_norm": 14.0,
+      "learning_rate": 1.5e-05,
+      "loss": 9.5794,
+      "step": 25
+    },
+    {
+      "epoch": 0.00723589001447178,
+      "grad_norm": 10.6875,
+      "learning_rate": 1.8e-05,
+      "loss": 9.0401,
+      "step": 30
+    },
+    {
+      "epoch": 0.00844187168355041,
+      "grad_norm": 4.0625,
+      "learning_rate": 2.1e-05,
+      "loss": 8.2806,
+      "step": 35
+    },
+    {
+      "epoch": 0.00964785335262904,
+      "grad_norm": 7.46875,
+      "learning_rate": 2.4e-05,
+      "loss": 8.2083,
+      "step": 40
+    },
+    {
+      "epoch": 0.01085383502170767,
+      "grad_norm": 7.09375,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 8.0693,
+      "step": 45
+    },
+    {
+      "epoch": 0.0120598166907863,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 7.8499,
+      "step": 50
+    },
+    {
+      "epoch": 0.01326579835986493,
+      "grad_norm": 4.3125,
+      "learning_rate": 3e-05,
+      "loss": 7.771,
+      "step": 55
+    },
+    {
+      "epoch": 0.01447178002894356,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 7.7109,
+      "step": 60
+    },
+    {
+      "epoch": 0.01567776169802219,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 7.6555,
+      "step": 65
+    },
+    {
+      "epoch": 0.01688374336710082,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 7.6062,
+      "step": 70
+    },
+    {
+      "epoch": 0.01808972503617945,
+      "grad_norm": 1.828125,
+      "learning_rate": 3e-05,
+      "loss": 7.5032,
+      "step": 75
+    },
+    {
+      "epoch": 0.01929570670525808,
+      "grad_norm": 1.78125,
+      "learning_rate": 3e-05,
+      "loss": 7.3975,
+      "step": 80
+    },
+    {
+      "epoch": 0.02050168837433671,
+      "grad_norm": 1.90625,
+      "learning_rate": 3e-05,
+      "loss": 7.3448,
+      "step": 85
+    },
+    {
+      "epoch": 0.02170767004341534,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 7.3475,
+      "step": 90
+    },
+    {
+      "epoch": 0.02291365171249397,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3e-05,
+      "loss": 7.1952,
+      "step": 95
+    },
+    {
+      "epoch": 0.0241196333815726,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3e-05,
+      "loss": 7.2894,
+      "step": 100
+    },
+    {
+      "epoch": 0.02532561505065123,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3e-05,
+      "loss": 7.3062,
+      "step": 105
+    },
+    {
+      "epoch": 0.02653159671972986,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3e-05,
+      "loss": 7.1772,
+      "step": 110
+    },
+    {
+      "epoch": 0.02773757838880849,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3e-05,
+      "loss": 7.1534,
+      "step": 115
+    },
+    {
+      "epoch": 0.02894356005788712,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3e-05,
+      "loss": 7.2145,
+      "step": 120
+    },
+    {
+      "epoch": 0.03014954172696575,
+      "grad_norm": 1.859375,
+      "learning_rate": 3e-05,
+      "loss": 7.0982,
+      "step": 125
+    },
+    {
+      "epoch": 0.03135552339604438,
+      "grad_norm": 1.78125,
+      "learning_rate": 3e-05,
+      "loss": 7.1474,
+      "step": 130
+    },
+    {
+      "epoch": 0.03256150506512301,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 7.061,
+      "step": 135
+    },
+    {
+      "epoch": 0.03376748673420164,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 7.0864,
+      "step": 140
+    },
+    {
+      "epoch": 0.03497346840328027,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 7.0517,
+      "step": 145
+    },
+    {
+      "epoch": 0.0361794500723589,
+      "grad_norm": 1.890625,
+      "learning_rate": 3e-05,
+      "loss": 6.9559,
+      "step": 150
+    },
+    {
+      "epoch": 0.03738543174143753,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 6.9734,
+      "step": 155
+    },
+    {
+      "epoch": 0.03859141341051616,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 6.9503,
+      "step": 160
+    },
+    {
+      "epoch": 0.03979739507959479,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 6.9415,
+      "step": 165
+    },
+    {
+      "epoch": 0.04100337674867342,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 7.001,
+      "step": 170
+    },
+    {
+      "epoch": 0.04220935841775205,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3e-05,
+      "loss": 6.8792,
+      "step": 175
+    },
+    {
+      "epoch": 0.04341534008683068,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3e-05,
+      "loss": 6.854,
+      "step": 180
+    },
+    {
+      "epoch": 0.04462132175590931,
+      "grad_norm": 1.625,
+      "learning_rate": 3e-05,
+      "loss": 6.8928,
+      "step": 185
+    },
+    {
+      "epoch": 0.04582730342498794,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 6.7267,
+      "step": 190
+    },
+    {
+      "epoch": 0.04703328509406657,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 6.8235,
+      "step": 195
+    },
+    {
+      "epoch": 0.0482392667631452,
+      "grad_norm": 1.4921875,
+      "learning_rate": 3e-05,
+      "loss": 6.7739,
+      "step": 200
+    },
+    {
+      "epoch": 0.04944524843222383,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 6.7484,
+      "step": 205
+    },
+    {
+      "epoch": 0.05065123010130246,
+      "grad_norm": 1.984375,
+      "learning_rate": 3e-05,
+      "loss": 6.7177,
+      "step": 210
+    },
+    {
+      "epoch": 0.05185721177038109,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 6.7426,
+      "step": 215
+    },
+    {
+      "epoch": 0.05306319343945972,
+      "grad_norm": 1.7421875,
+      "learning_rate": 3e-05,
+      "loss": 6.6993,
+      "step": 220
+    },
+    {
+      "epoch": 0.05426917510853835,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 6.7761,
+      "step": 225
+    },
+    {
+      "epoch": 0.05547515677761698,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 6.5162,
+      "step": 230
+    },
+    {
+      "epoch": 0.05668113844669561,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 6.6471,
+      "step": 235
+    },
+    {
+      "epoch": 0.05788712011577424,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 6.5741,
+      "step": 240
+    },
+    {
+      "epoch": 0.05909310178485287,
+      "grad_norm": 4.1875,
+      "learning_rate": 3e-05,
+      "loss": 6.4797,
+      "step": 245
+    },
+    {
+      "epoch": 0.0602990834539315,
+      "grad_norm": 1.921875,
+      "learning_rate": 3e-05,
+      "loss": 6.5313,
+      "step": 250
+    },
+    {
+      "epoch": 0.06150506512301013,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 6.4377,
+      "step": 255
+    },
+    {
+      "epoch": 0.06271104679208876,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 6.4948,
+      "step": 260
+    },
+    {
+      "epoch": 0.0639170284611674,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 6.4971,
+      "step": 265
+    },
+    {
+      "epoch": 0.06512301013024602,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 6.3971,
+      "step": 270
+    },
+    {
+      "epoch": 0.06632899179932465,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 6.4961,
+      "step": 275
+    },
+    {
+      "epoch": 0.06753497346840329,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 6.4457,
+      "step": 280
+    },
+    {
+      "epoch": 0.06874095513748191,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 6.4788,
+      "step": 285
+    },
+    {
+      "epoch": 0.06994693680656054,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 6.4388,
+      "step": 290
+    },
+    {
+      "epoch": 0.07115291847563918,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 6.2615,
+      "step": 295
+    },
+    {
+      "epoch": 0.0723589001447178,
+      "grad_norm": 1.859375,
+      "learning_rate": 3e-05,
+      "loss": 6.369,
+      "step": 300
+    },
+    {
+      "epoch": 0.07356488181379643,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 6.4014,
+      "step": 305
+    },
+    {
+      "epoch": 0.07477086348287507,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 6.3217,
+      "step": 310
+    },
+    {
+      "epoch": 0.07597684515195369,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 6.424,
+      "step": 315
+    },
+    {
+      "epoch": 0.07718282682103232,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 6.2718,
+      "step": 320
+    },
+    {
+      "epoch": 0.07838880849011096,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 6.2224,
+      "step": 325
+    },
+    {
+      "epoch": 0.07959479015918958,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 6.2538,
+      "step": 330
+    },
+    {
+      "epoch": 0.08080077182826821,
+      "grad_norm": 1.78125,
+      "learning_rate": 3e-05,
+      "loss": 6.2736,
+      "step": 335
+    },
+    {
+      "epoch": 0.08200675349734685,
+      "grad_norm": 1.7421875,
+      "learning_rate": 3e-05,
+      "loss": 6.248,
+      "step": 340
+    },
+    {
+      "epoch": 0.08321273516642547,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 6.3282,
+      "step": 345
+    },
+    {
+      "epoch": 0.0844187168355041,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 6.3235,
+      "step": 350
+    },
+    {
+      "epoch": 0.08562469850458274,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 6.37,
+      "step": 355
+    },
+    {
+      "epoch": 0.08683068017366136,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 6.1852,
+      "step": 360
+    },
+    {
+      "epoch": 0.08803666184273999,
+      "grad_norm": 3.5,
+      "learning_rate": 3e-05,
+      "loss": 6.2639,
+      "step": 365
+    },
+    {
+      "epoch": 0.08924264351181863,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 6.2367,
+      "step": 370
+    },
+    {
+      "epoch": 0.09044862518089725,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 6.0995,
+      "step": 375
+    },
+    {
+      "epoch": 0.09165460684997588,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 6.1735,
+      "step": 380
+    },
+    {
+      "epoch": 0.09286058851905452,
+      "grad_norm": 1.890625,
+      "learning_rate": 3e-05,
+      "loss": 6.1528,
+      "step": 385
+    },
+    {
+      "epoch": 0.09406657018813314,
+      "grad_norm": 1.8359375,
+      "learning_rate": 3e-05,
+      "loss": 6.1329,
+      "step": 390
+    },
+    {
+      "epoch": 0.09527255185721177,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 6.1236,
+      "step": 395
+    },
+    {
+      "epoch": 0.0964785335262904,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 6.198,
+      "step": 400
+    },
+    {
+      "epoch": 0.09768451519536903,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 6.1956,
+      "step": 405
+    },
+    {
+      "epoch": 0.09889049686444766,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 6.1735,
+      "step": 410
+    },
+    {
+      "epoch": 0.1000964785335263,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 6.1953,
+      "step": 415
+    },
+    {
+      "epoch": 0.10130246020260492,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 6.1069,
+      "step": 420
+    },
+    {
+      "epoch": 0.10250844187168355,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 6.0713,
+      "step": 425
+    },
+    {
+      "epoch": 0.10371442354076219,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 6.3303,
+      "step": 430
+    },
+    {
+      "epoch": 0.1049204052098408,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 6.012,
+      "step": 435
+    },
+    {
+      "epoch": 0.10612638687891944,
+      "grad_norm": 1.78125,
+      "learning_rate": 3e-05,
+      "loss": 6.05,
+      "step": 440
+    },
+    {
+      "epoch": 0.10733236854799807,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 6.0952,
+      "step": 445
+    },
+    {
+      "epoch": 0.1085383502170767,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3e-05,
+      "loss": 6.1188,
+      "step": 450
+    },
+    {
+      "epoch": 0.10974433188615533,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 6.1039,
+      "step": 455
+    },
+    {
+      "epoch": 0.11095031355523396,
+      "grad_norm": 1.984375,
+      "learning_rate": 3e-05,
+      "loss": 6.0926,
+      "step": 460
+    },
+    {
+      "epoch": 0.11215629522431259,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 6.0924,
+      "step": 465
+    },
+    {
+      "epoch": 0.11336227689339122,
+      "grad_norm": 1.859375,
+      "learning_rate": 3e-05,
+      "loss": 6.012,
+      "step": 470
+    },
+    {
+      "epoch": 0.11456825856246985,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 5.9395,
+      "step": 475
+    },
+    {
+      "epoch": 0.11577424023154848,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 6.071,
+      "step": 480
+    },
+    {
+      "epoch": 0.11698022190062711,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 6.0479,
+      "step": 485
+    },
+    {
+      "epoch": 0.11818620356970574,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 5.9637,
+      "step": 490
+    },
+    {
+      "epoch": 0.11939218523878437,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3e-05,
+      "loss": 5.9824,
+      "step": 495
+    },
+    {
+      "epoch": 0.120598166907863,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 5.993,
+      "step": 500
+    },
+    {
+      "epoch": 0.12180414857694163,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 5.8498,
+      "step": 505
+    },
+    {
+      "epoch": 0.12301013024602026,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 5.8376,
+      "step": 510
+    },
+    {
+      "epoch": 0.12421611191509889,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 5.9112,
+      "step": 515
+    },
+    {
+      "epoch": 0.12542209358417752,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 5.9365,
+      "step": 520
+    },
+    {
+      "epoch": 0.12662807525325614,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3e-05,
+      "loss": 6.0778,
+      "step": 525
+    },
+    {
+      "epoch": 0.1278340569223348,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 5.9356,
+      "step": 530
+    },
+    {
+      "epoch": 0.12904003859141341,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3e-05,
+      "loss": 5.8361,
+      "step": 535
+    },
+    {
+      "epoch": 0.13024602026049203,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 5.89,
+      "step": 540
+    },
+    {
+      "epoch": 0.13145200192957068,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 5.9576,
+      "step": 545
+    },
+    {
+      "epoch": 0.1326579835986493,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 5.9431,
+      "step": 550
+    },
+    {
+      "epoch": 0.13386396526772792,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 5.8577,
+      "step": 555
+    },
+    {
+      "epoch": 0.13506994693680657,
+      "grad_norm": 3.671875,
+      "learning_rate": 3e-05,
+      "loss": 5.9617,
+      "step": 560
+    },
+    {
+      "epoch": 0.1362759286058852,
+      "grad_norm": 3.3125,
+      "learning_rate": 3e-05,
+      "loss": 5.8371,
+      "step": 565
+    },
+    {
+      "epoch": 0.13748191027496381,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 5.8252,
+      "step": 570
+    },
+    {
+      "epoch": 0.13868789194404246,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 5.808,
+      "step": 575
+    },
+    {
+      "epoch": 0.13989387361312108,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 5.9049,
+      "step": 580
+    },
+    {
+      "epoch": 0.1410998552821997,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 5.836,
+      "step": 585
+    },
+    {
+      "epoch": 0.14230583695127835,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 5.7272,
+      "step": 590
+    },
+    {
+      "epoch": 0.14351181862035697,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 5.8204,
+      "step": 595
+    },
+    {
+      "epoch": 0.1447178002894356,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 5.9753,
+      "step": 600
+    },
+    {
+      "epoch": 0.14592378195851424,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 5.7331,
+      "step": 605
+    },
+    {
+      "epoch": 0.14712976362759286,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 5.8421,
+      "step": 610
+    },
+    {
+      "epoch": 0.14833574529667148,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 5.7543,
+      "step": 615
+    },
+    {
+      "epoch": 0.14954172696575013,
+      "grad_norm": 1.890625,
+      "learning_rate": 3e-05,
+      "loss": 5.7815,
+      "step": 620
+    },
+    {
+      "epoch": 0.15074770863482875,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.6632,
+      "step": 625
+    },
+    {
+      "epoch": 0.15195369030390737,
+      "grad_norm": 1.7265625,
+      "learning_rate": 3e-05,
+      "loss": 5.7123,
+      "step": 630
+    },
+    {
+      "epoch": 0.15315967197298602,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 5.8582,
+      "step": 635
+    },
+    {
+      "epoch": 0.15436565364206464,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3e-05,
+      "loss": 5.7628,
+      "step": 640
+    },
+    {
+      "epoch": 0.15557163531114326,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 5.9128,
+      "step": 645
+    },
+    {
+      "epoch": 0.1567776169802219,
+      "grad_norm": 1.984375,
+      "learning_rate": 3e-05,
+      "loss": 5.8285,
+      "step": 650
+    },
+    {
+      "epoch": 0.15798359864930053,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 5.7964,
+      "step": 655
+    },
+    {
+      "epoch": 0.15918958031837915,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 5.8595,
+      "step": 660
+    },
+    {
+      "epoch": 0.1603955619874578,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 5.7467,
+      "step": 665
+    },
+    {
+      "epoch": 0.16160154365653642,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 5.5928,
+      "step": 670
+    },
+    {
+      "epoch": 0.16280752532561504,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 5.738,
+      "step": 675
+    },
+    {
+      "epoch": 0.1640135069946937,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 5.7686,
+      "step": 680
+    },
+    {
+      "epoch": 0.1652194886637723,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 5.7783,
+      "step": 685
+    },
+    {
+      "epoch": 0.16642547033285093,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 5.7515,
+      "step": 690
+    },
+    {
+      "epoch": 0.16763145200192958,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 5.6172,
+      "step": 695
+    },
+    {
+      "epoch": 0.1688374336710082,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 5.6604,
+      "step": 700
+    },
+    {
+      "epoch": 0.17004341534008682,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 5.8269,
+      "step": 705
+    },
+    {
+      "epoch": 0.17124939700916547,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 5.7508,
+      "step": 710
+    },
+    {
+      "epoch": 0.1724553786782441,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3e-05,
+      "loss": 5.6119,
+      "step": 715
+    },
+    {
+      "epoch": 0.1736613603473227,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 5.8154,
+      "step": 720
+    },
+    {
+      "epoch": 0.17486734201640136,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.6788,
+      "step": 725
+    },
+    {
+      "epoch": 0.17607332368547998,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 5.6748,
+      "step": 730
+    },
+    {
+      "epoch": 0.1772793053545586,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 5.6841,
+      "step": 735
+    },
+    {
+      "epoch": 0.17848528702363725,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 5.5893,
+      "step": 740
+    },
+    {
+      "epoch": 0.17969126869271587,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 5.6872,
+      "step": 745
+    },
+    {
+      "epoch": 0.1808972503617945,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 5.721,
+      "step": 750
+    },
+    {
+      "epoch": 0.18210323203087314,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 5.7009,
+      "step": 755
+    },
+    {
+      "epoch": 0.18330921369995176,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 5.6044,
+      "step": 760
+    },
+    {
+      "epoch": 0.18451519536903038,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.5983,
+      "step": 765
+    },
+    {
+      "epoch": 0.18572117703810903,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 5.6824,
+      "step": 770
+    },
+    {
+      "epoch": 0.18692715870718765,
+      "grad_norm": 6.625,
+      "learning_rate": 3e-05,
+      "loss": 5.5232,
+      "step": 775
+    },
+    {
+      "epoch": 0.18813314037626627,
+      "grad_norm": 3.96875,
+      "learning_rate": 3e-05,
+      "loss": 5.6313,
+      "step": 780
+    },
+    {
+      "epoch": 0.18933912204534492,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 5.7965,
+      "step": 785
+    },
+    {
+      "epoch": 0.19054510371442354,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 5.5971,
+      "step": 790
+    },
+    {
+      "epoch": 0.19175108538350216,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 5.6727,
+      "step": 795
+    },
+    {
+      "epoch": 0.1929570670525808,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 5.7194,
+      "step": 800
+    },
+    {
+      "epoch": 0.19416304872165943,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 5.6573,
+      "step": 805
+    },
+    {
+      "epoch": 0.19536903039073805,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 5.5429,
+      "step": 810
+    },
+    {
+      "epoch": 0.1965750120598167,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 5.541,
+      "step": 815
+    },
+    {
+      "epoch": 0.19778099372889532,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 5.5469,
+      "step": 820
+    },
+    {
+      "epoch": 0.19898697539797394,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.4902,
+      "step": 825
+    },
+    {
+      "epoch": 0.2001929570670526,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 5.5931,
+      "step": 830
+    },
+    {
+      "epoch": 0.2013989387361312,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 5.4769,
+      "step": 835
+    },
+    {
+      "epoch": 0.20260492040520983,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.6509,
+      "step": 840
+    },
+    {
+      "epoch": 0.20381090207428848,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 5.6149,
+      "step": 845
+    },
+    {
+      "epoch": 0.2050168837433671,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 5.661,
+      "step": 850
+    },
+    {
+      "epoch": 0.20622286541244572,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 5.4612,
+      "step": 855
+    },
+    {
+      "epoch": 0.20742884708152437,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 5.5261,
+      "step": 860
+    },
+    {
+      "epoch": 0.208634828750603,
+      "grad_norm": 1.859375,
+      "learning_rate": 3e-05,
+      "loss": 5.7081,
+      "step": 865
+    },
+    {
+      "epoch": 0.2098408104196816,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 5.6594,
+      "step": 870
+    },
+    {
+      "epoch": 0.21104679208876026,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 5.3843,
+      "step": 875
+    },
+    {
+      "epoch": 0.21225277375783888,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.6101,
+      "step": 880
+    },
+    {
+      "epoch": 0.2134587554269175,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 5.4925,
+      "step": 885
+    },
+    {
+      "epoch": 0.21466473709599615,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 5.4863,
+      "step": 890
+    },
+    {
+      "epoch": 0.21587071876507477,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3e-05,
+      "loss": 5.5784,
+      "step": 895
+    },
+    {
+      "epoch": 0.2170767004341534,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 5.5706,
+      "step": 900
+    },
+    {
+      "epoch": 0.21828268210323204,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 5.4767,
+      "step": 905
+    },
+    {
+      "epoch": 0.21948866377231066,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 5.5478,
+      "step": 910
+    },
+    {
+      "epoch": 0.22069464544138928,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 5.4582,
+      "step": 915
+    },
+    {
+      "epoch": 0.22190062711046793,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 5.4833,
+      "step": 920
+    },
+    {
+      "epoch": 0.22310660877954655,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 5.3878,
+      "step": 925
+    },
+    {
+      "epoch": 0.22431259044862517,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 5.6182,
+      "step": 930
+    },
+    {
+      "epoch": 0.22551857211770382,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 5.3843,
+      "step": 935
+    },
+    {
+      "epoch": 0.22672455378678244,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 5.398,
+      "step": 940
+    },
+    {
+      "epoch": 0.22793053545586106,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 5.4073,
+      "step": 945
+    },
+    {
+      "epoch": 0.2291365171249397,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 5.4073,
+      "step": 950
+    },
+    {
+      "epoch": 0.23034249879401833,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 5.5549,
+      "step": 955
+    },
+    {
+      "epoch": 0.23154848046309695,
+      "grad_norm": 1.8046875,
+      "learning_rate": 3e-05,
+      "loss": 5.3167,
+      "step": 960
+    },
+    {
+      "epoch": 0.2327544621321756,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3e-05,
+      "loss": 5.4991,
+      "step": 965
+    },
+    {
+      "epoch": 0.23396044380125422,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 5.617,
+      "step": 970
+    },
+    {
+      "epoch": 0.23516642547033284,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 5.5679,
+      "step": 975
+    },
+    {
+      "epoch": 0.2363724071394115,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.394,
+      "step": 980
+    },
+    {
+      "epoch": 0.2375783888084901,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 5.4975,
+      "step": 985
+    },
+    {
+      "epoch": 0.23878437047756873,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 5.3327,
+      "step": 990
+    },
+    {
+      "epoch": 0.23999035214664738,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 5.5471,
+      "step": 995
+    },
+    {
+      "epoch": 0.241196333815726,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 5.4136,
+      "step": 1000
+    },
+    {
+      "epoch": 0.24240231548480462,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.494,
+      "step": 1005
+    },
+    {
+      "epoch": 0.24360829715388327,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 5.5097,
+      "step": 1010
+    },
+    {
+      "epoch": 0.2448142788229619,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 5.4636,
+      "step": 1015
+    },
+    {
+      "epoch": 0.2460202604920405,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 5.3754,
+      "step": 1020
+    },
+    {
+      "epoch": 0.24722624216111916,
+      "grad_norm": 1.8046875,
+      "learning_rate": 3e-05,
+      "loss": 5.5269,
+      "step": 1025
+    },
+    {
+      "epoch": 0.24843222383019778,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 5.3828,
+      "step": 1030
+    },
+    {
+      "epoch": 0.2496382054992764,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 5.3867,
+      "step": 1035
+    },
+    {
+      "epoch": 0.25084418716835505,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 5.4392,
+      "step": 1040
+    },
+    {
+      "epoch": 0.2520501688374337,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 5.4154,
+      "step": 1045
+    },
+    {
+      "epoch": 0.2532561505065123,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 5.382,
+      "step": 1050
+    },
+    {
+      "epoch": 0.25446213217559094,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 5.3914,
+      "step": 1055
+    },
+    {
+      "epoch": 0.2556681138446696,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 5.2091,
+      "step": 1060
+    },
+    {
+      "epoch": 0.2568740955137482,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 5.402,
+      "step": 1065
+    },
+    {
+      "epoch": 0.25808007718282683,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 5.3478,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2592860588519055,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 5.278,
+      "step": 1075
+    },
+    {
+      "epoch": 0.26049204052098407,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 5.402,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2616980221900627,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 5.3795,
+      "step": 1085
+    },
+    {
+      "epoch": 0.26290400385914137,
+      "grad_norm": 1.9375,
+      "learning_rate": 3e-05,
+      "loss": 5.2764,
+      "step": 1090
+    },
+    {
+      "epoch": 0.26410998552821996,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 5.3733,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2653159671972986,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 5.281,
+      "step": 1100
+    },
+    {
+      "epoch": 0.26652194886637726,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 5.2757,
+      "step": 1105
+    },
+    {
+      "epoch": 0.26772793053545585,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 5.2248,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2689339122045345,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 5.4059,
+      "step": 1115
+    },
+    {
+      "epoch": 0.27013989387361315,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 5.2239,
+      "step": 1120
+    },
+    {
+      "epoch": 0.27134587554269174,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 5.1511,
+      "step": 1125
+    },
+    {
+      "epoch": 0.2725518572117704,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 5.2924,
+      "step": 1130
+    },
+    {
+      "epoch": 0.27375783888084904,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 5.2489,
+      "step": 1135
+    },
+    {
+      "epoch": 0.27496382054992763,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 5.329,
+      "step": 1140
+    },
+    {
+      "epoch": 0.2761698022190063,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3e-05,
+      "loss": 5.1569,
+      "step": 1145
+    },
+    {
+      "epoch": 0.2773757838880849,
+      "grad_norm": 1.6796875,
+      "learning_rate": 3e-05,
+      "loss": 5.3047,
+      "step": 1150
+    },
+    {
+      "epoch": 0.2785817655571635,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.4468,
+      "step": 1155
+    },
+    {
+      "epoch": 0.27978774722624217,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 5.409,
+      "step": 1160
+    },
+    {
+      "epoch": 0.2809937288953208,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.2111,
+      "step": 1165
+    },
+    {
+      "epoch": 0.2821997105643994,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 5.3326,
+      "step": 1170
+    },
+    {
+      "epoch": 0.28340569223347806,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 5.1767,
+      "step": 1175
+    },
+    {
+      "epoch": 0.2846116739025567,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 5.3533,
+      "step": 1180
+    },
+    {
+      "epoch": 0.2858176555716353,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 5.2169,
+      "step": 1185
+    },
+    {
+      "epoch": 0.28702363724071395,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.351,
+      "step": 1190
+    },
+    {
+      "epoch": 0.2882296189097926,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 5.3343,
+      "step": 1195
+    },
+    {
+      "epoch": 0.2894356005788712,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 5.31,
+      "step": 1200
+    },
+    {
+      "epoch": 0.29064158224794984,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 5.2081,
+      "step": 1205
+    },
+    {
+      "epoch": 0.2918475639170285,
+      "grad_norm": 1.796875,
+      "learning_rate": 3e-05,
+      "loss": 5.3053,
+      "step": 1210
+    },
+    {
+      "epoch": 0.2930535455861071,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 5.2586,
+      "step": 1215
+    },
+    {
+      "epoch": 0.2942595272551857,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 5.1507,
+      "step": 1220
+    },
+    {
+      "epoch": 0.2954655089242644,
+      "grad_norm": 1.7578125,
+      "learning_rate": 3e-05,
+      "loss": 5.3245,
+      "step": 1225
+    },
+    {
+      "epoch": 0.29667149059334297,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 5.2317,
+      "step": 1230
+    },
+    {
+      "epoch": 0.2978774722624216,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.2238,
+      "step": 1235
+    },
+    {
+      "epoch": 0.29908345393150026,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 5.2438,
+      "step": 1240
+    },
+    {
+      "epoch": 0.30028943560057886,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 5.2629,
+      "step": 1245
+    },
+    {
+      "epoch": 0.3014954172696575,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 5.1487,
+      "step": 1250
+    },
+    {
+      "epoch": 0.30270139893873615,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 5.2241,
+      "step": 1255
+    },
+    {
+      "epoch": 0.30390738060781475,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 5.2636,
+      "step": 1260
+    },
+    {
+      "epoch": 0.3051133622768934,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 5.1245,
+      "step": 1265
+    },
+    {
+      "epoch": 0.30631934394597204,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 5.1451,
+      "step": 1270
+    },
+    {
+      "epoch": 0.30752532561505064,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 5.2969,
+      "step": 1275
+    },
+    {
+      "epoch": 0.3087313072841293,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 5.1689,
+      "step": 1280
+    },
+    {
+      "epoch": 0.30993728895320793,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 5.2066,
+      "step": 1285
+    },
+    {
+      "epoch": 0.3111432706222865,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.9629,
+      "step": 1290
+    },
+    {
+      "epoch": 0.3123492522913652,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 5.1767,
+      "step": 1295
+    },
+    {
+      "epoch": 0.3135552339604438,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 5.2642,
+      "step": 1300
+    },
+    {
+      "epoch": 0.3147612156295224,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 5.2788,
+      "step": 1305
+    },
+    {
+      "epoch": 0.31596719729860107,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 5.1955,
+      "step": 1310
+    },
+    {
+      "epoch": 0.3171731789676797,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 5.2811,
+      "step": 1315
+    },
+    {
+      "epoch": 0.3183791606367583,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 5.2512,
+      "step": 1320
+    },
+    {
+      "epoch": 0.31958514230583696,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 5.126,
+      "step": 1325
+    },
+    {
+      "epoch": 0.3207911239749156,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 5.0701,
+      "step": 1330
+    },
+    {
+      "epoch": 0.3219971056439942,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.1081,
+      "step": 1335
+    },
+    {
+      "epoch": 0.32320308731307285,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 5.2009,
+      "step": 1340
+    },
+    {
+      "epoch": 0.3244090689821515,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.25,
+      "step": 1345
+    },
+    {
+      "epoch": 0.3256150506512301,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 5.1474,
+      "step": 1350
+    },
+    {
+      "epoch": 0.32682103232030874,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 5.4087,
+      "step": 1355
+    },
+    {
+      "epoch": 0.3280270139893874,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 5.2222,
+      "step": 1360
+    },
+    {
+      "epoch": 0.329232995658466,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 5.1554,
+      "step": 1365
+    },
+    {
+      "epoch": 0.3304389773275446,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 5.19,
+      "step": 1370
+    },
+    {
+      "epoch": 0.3316449589966233,
+      "grad_norm": 3.78125,
+      "learning_rate": 3e-05,
+      "loss": 5.1978,
+      "step": 1375
+    },
+    {
+      "epoch": 0.33285094066570187,
+      "grad_norm": 1.875,
+      "learning_rate": 3e-05,
+      "loss": 5.1126,
+      "step": 1380
+    },
+    {
+      "epoch": 0.3340569223347805,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 5.1463,
+      "step": 1385
+    },
+    {
+      "epoch": 0.33526290400385916,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 5.0063,
+      "step": 1390
+    },
+    {
+      "epoch": 0.33646888567293776,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 5.199,
+      "step": 1395
+    },
+    {
+      "epoch": 0.3376748673420164,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 5.1133,
+      "step": 1400
+    },
+    {
+      "epoch": 0.33888084901109505,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 5.3344,
+      "step": 1405
+    },
+    {
+      "epoch": 0.34008683068017365,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 5.2786,
+      "step": 1410
+    },
+    {
+      "epoch": 0.3412928123492523,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 5.1066,
+      "step": 1415
+    },
+    {
+      "epoch": 0.34249879401833094,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 5.2149,
+      "step": 1420
+    },
+    {
+      "epoch": 0.34370477568740954,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 5.0942,
+      "step": 1425
+    },
+    {
+      "epoch": 0.3449107573564882,
+      "grad_norm": 1.9375,
+      "learning_rate": 3e-05,
+      "loss": 5.1109,
+      "step": 1430
+    },
+    {
+      "epoch": 0.34611673902556683,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 5.2741,
+      "step": 1435
+    },
+    {
+      "epoch": 0.3473227206946454,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 5.1884,
+      "step": 1440
+    },
+    {
+      "epoch": 0.3485287023637241,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3e-05,
+      "loss": 5.184,
+      "step": 1445
+    },
+    {
+      "epoch": 0.3497346840328027,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 5.113,
+      "step": 1450
+    },
+    {
+      "epoch": 0.3509406657018813,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 5.0311,
+      "step": 1455
+    },
+    {
+      "epoch": 0.35214664737095996,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 5.0447,
+      "step": 1460
+    },
+    {
+      "epoch": 0.3533526290400386,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 5.263,
+      "step": 1465
+    },
+    {
+      "epoch": 0.3545586107091172,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 5.0733,
+      "step": 1470
+    },
+    {
+      "epoch": 0.35576459237819585,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 5.176,
+      "step": 1475
+    },
+    {
+      "epoch": 0.3569705740472745,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.9958,
+      "step": 1480
+    },
+    {
+      "epoch": 0.3581765557163531,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 5.0778,
+      "step": 1485
+    },
+    {
+      "epoch": 0.35938253738543174,
+      "grad_norm": 1.890625,
+      "learning_rate": 3e-05,
+      "loss": 5.1228,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3605885190545104,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 5.053,
+      "step": 1495
+    },
+    {
+      "epoch": 0.361794500723589,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 5.0317,
+      "step": 1500
+    },
+    {
+      "epoch": 0.36300048239266763,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 5.0683,
+      "step": 1505
+    },
+    {
+      "epoch": 0.3642064640617463,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 5.0274,
+      "step": 1510
+    },
+    {
+      "epoch": 0.3654124457308249,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 5.0054,
+      "step": 1515
+    },
+    {
+      "epoch": 0.3666184273999035,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 5.0405,
+      "step": 1520
+    },
+    {
+      "epoch": 0.36782440906898217,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 5.1625,
+      "step": 1525
+    },
+    {
+      "epoch": 0.36903039073806077,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.281,
+      "step": 1530
+    },
+    {
+      "epoch": 0.3702363724071394,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 5.1182,
+      "step": 1535
+    },
+    {
+      "epoch": 0.37144235407621806,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 5.088,
+      "step": 1540
+    },
+    {
+      "epoch": 0.37264833574529665,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.9542,
+      "step": 1545
+    },
+    {
+      "epoch": 0.3738543174143753,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 5.0302,
+      "step": 1550
+    },
+    {
+      "epoch": 0.37506029908345395,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 5.0856,
+      "step": 1555
+    },
+    {
+      "epoch": 0.37626628075253254,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 5.018,
+      "step": 1560
+    },
+    {
+      "epoch": 0.3774722624216112,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.9291,
+      "step": 1565
+    },
+    {
+      "epoch": 0.37867824409068984,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.909,
+      "step": 1570
+    },
+    {
+      "epoch": 0.37988422575976843,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 5.111,
+      "step": 1575
+    },
+    {
+      "epoch": 0.3810902074288471,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 5.1551,
+      "step": 1580
+    },
+    {
+      "epoch": 0.38229618909792573,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.9441,
+      "step": 1585
+    },
+    {
+      "epoch": 0.3835021707670043,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3e-05,
+      "loss": 4.9529,
+      "step": 1590
+    },
+    {
+      "epoch": 0.384708152436083,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.9265,
+      "step": 1595
+    },
+    {
+      "epoch": 0.3859141341051616,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.9057,
+      "step": 1600
+    },
+    {
+      "epoch": 0.3871201157742402,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 5.2044,
+      "step": 1605
+    },
+    {
+      "epoch": 0.38832609744331886,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 5.0659,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3895320791123975,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 5.0833,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3907380607814761,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.9692,
+      "step": 1620
+    },
+    {
+      "epoch": 0.39194404245055475,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 5.0722,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3931500241196334,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.9036,
+      "step": 1630
+    },
+    {
+      "epoch": 0.394356005788712,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.9615,
+      "step": 1635
+    },
+    {
+      "epoch": 0.39556198745779064,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3e-05,
+      "loss": 5.0087,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3967679691268693,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3e-05,
+      "loss": 5.0289,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3979739507959479,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 5.0801,
+      "step": 1650
+    },
+    {
+      "epoch": 0.39917993246502653,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.8894,
+      "step": 1655
+    },
+    {
+      "epoch": 0.4003859141341052,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 5.0647,
+      "step": 1660
+    },
+    {
+      "epoch": 0.4015918958031838,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 5.0765,
+      "step": 1665
+    },
+    {
+      "epoch": 0.4027978774722624,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.9994,
+      "step": 1670
+    },
+    {
+      "epoch": 0.40400385914134107,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 5.1047,
+      "step": 1675
+    },
+    {
+      "epoch": 0.40520984081041966,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 5.0652,
+      "step": 1680
+    },
+    {
+      "epoch": 0.4064158224794983,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 5.1016,
+      "step": 1685
+    },
+    {
+      "epoch": 0.40762180414857696,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.0802,
+      "step": 1690
+    },
+    {
+      "epoch": 0.40882778581765555,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.938,
+      "step": 1695
+    },
+    {
+      "epoch": 0.4100337674867342,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 5.0301,
+      "step": 1700
+    },
+    {
+      "epoch": 0.41123974915581285,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.0248,
+      "step": 1705
+    },
+    {
+      "epoch": 0.41244573082489144,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.7854,
+      "step": 1710
+    },
+    {
+      "epoch": 0.4136517124939701,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 5.0096,
+      "step": 1715
+    },
+    {
+      "epoch": 0.41485769416304874,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.9715,
+      "step": 1720
+    },
+    {
+      "epoch": 0.41606367583212733,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.9592,
+      "step": 1725
+    },
+    {
+      "epoch": 0.417269657501206,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.8384,
+      "step": 1730
+    },
+    {
+      "epoch": 0.41847563917028463,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.8799,
+      "step": 1735
+    },
+    {
+      "epoch": 0.4196816208393632,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 5.0099,
+      "step": 1740
+    },
+    {
+      "epoch": 0.42088760250844187,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.7933,
+      "step": 1745
+    },
+    {
+      "epoch": 0.4220935841775205,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 5.1248,
+      "step": 1750
+    },
+    {
+      "epoch": 0.4232995658465991,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.9463,
+      "step": 1755
+    },
+    {
+      "epoch": 0.42450554751567776,
+      "grad_norm": 1.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.9497,
+      "step": 1760
+    },
+    {
+      "epoch": 0.4257115291847564,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.8903,
+      "step": 1765
+    },
+    {
+      "epoch": 0.426917510853835,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.8761,
+      "step": 1770
+    },
+    {
+      "epoch": 0.42812349252291365,
+      "grad_norm": 1.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.9089,
+      "step": 1775
+    },
+    {
+      "epoch": 0.4293294741919923,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.9017,
+      "step": 1780
+    },
+    {
+      "epoch": 0.4305354558610709,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.8968,
+      "step": 1785
+    },
+    {
+      "epoch": 0.43174143753014954,
+      "grad_norm": 1.875,
+      "learning_rate": 3e-05,
+      "loss": 4.9408,
+      "step": 1790
+    },
+    {
+      "epoch": 0.4329474191992282,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.9692,
+      "step": 1795
+    },
+    {
+      "epoch": 0.4341534008683068,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 5.0574,
+      "step": 1800
+    },
+    {
+      "epoch": 0.43535938253738543,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.9252,
+      "step": 1805
+    },
+    {
+      "epoch": 0.4365653642064641,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.91,
+      "step": 1810
+    },
+    {
+      "epoch": 0.4377713458755427,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 4.8551,
+      "step": 1815
+    },
+    {
+      "epoch": 0.4389773275446213,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.8312,
+      "step": 1820
+    },
+    {
+      "epoch": 0.44018330921369997,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.9764,
+      "step": 1825
+    },
+    {
+      "epoch": 0.44138929088277856,
+      "grad_norm": 1.8984375,
+      "learning_rate": 3e-05,
+      "loss": 4.8777,
+      "step": 1830
+    },
+    {
+      "epoch": 0.4425952725518572,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.9967,
+      "step": 1835
+    },
+    {
+      "epoch": 0.44380125422093586,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.9884,
+      "step": 1840
+    },
+    {
+      "epoch": 0.44500723589001445,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.9541,
+      "step": 1845
+    },
+    {
+      "epoch": 0.4462132175590931,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.7961,
+      "step": 1850
+    },
+    {
+      "epoch": 0.44741919922817175,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 5.0094,
+      "step": 1855
+    },
+    {
+      "epoch": 0.44862518089725034,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.9893,
+      "step": 1860
+    },
+    {
+      "epoch": 0.449831162566329,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.8225,
+      "step": 1865
+    },
+    {
+      "epoch": 0.45103714423540764,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 5.0,
+      "step": 1870
+    },
+    {
+      "epoch": 0.45224312590448623,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.8553,
+      "step": 1875
+    },
+    {
+      "epoch": 0.4534491075735649,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.9396,
+      "step": 1880
+    },
+    {
+      "epoch": 0.45465508924264353,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.8937,
+      "step": 1885
+    },
+    {
+      "epoch": 0.4558610709117221,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.6979,
+      "step": 1890
+    },
+    {
+      "epoch": 0.45706705258080077,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.914,
+      "step": 1895
+    },
+    {
+      "epoch": 0.4582730342498794,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.9191,
+      "step": 1900
+    },
+    {
+      "epoch": 0.459479015918958,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.8465,
+      "step": 1905
+    },
+    {
+      "epoch": 0.46068499758803666,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.853,
+      "step": 1910
+    },
+    {
+      "epoch": 0.4618909792571153,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.8957,
+      "step": 1915
+    },
+    {
+      "epoch": 0.4630969609261939,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.9466,
+      "step": 1920
+    },
+    {
+      "epoch": 0.46430294259527255,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.9664,
+      "step": 1925
+    },
+    {
+      "epoch": 0.4655089242643512,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.8807,
+      "step": 1930
+    },
+    {
+      "epoch": 0.4667149059334298,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.9079,
+      "step": 1935
+    },
+    {
+      "epoch": 0.46792088760250844,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.99,
+      "step": 1940
+    },
+    {
+      "epoch": 0.4691268692715871,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.9467,
+      "step": 1945
+    },
+    {
+      "epoch": 0.4703328509406657,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.8917,
+      "step": 1950
+    },
+    {
+      "epoch": 0.47153883260974433,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.9579,
+      "step": 1955
+    },
+    {
+      "epoch": 0.472744814278823,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.9121,
+      "step": 1960
+    },
+    {
+      "epoch": 0.47395079594790157,
+      "grad_norm": 1.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.9916,
+      "step": 1965
+    },
+    {
+      "epoch": 0.4751567776169802,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.9472,
+      "step": 1970
+    },
+    {
+      "epoch": 0.47636275928605887,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 5.0118,
+      "step": 1975
+    },
+    {
+      "epoch": 0.47756874095513746,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.8575,
+      "step": 1980
+    },
+    {
+      "epoch": 0.4787747226242161,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.8229,
+      "step": 1985
+    },
+    {
+      "epoch": 0.47998070429329476,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.9579,
+      "step": 1990
+    },
+    {
+      "epoch": 0.48118668596237335,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.7751,
+      "step": 1995
+    },
+    {
+      "epoch": 0.482392667631452,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.9796,
+      "step": 2000
+    },
+    {
+      "epoch": 0.48359864930053065,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.7714,
+      "step": 2005
+    },
+    {
+      "epoch": 0.48480463096960924,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.7068,
+      "step": 2010
+    },
+    {
+      "epoch": 0.4860106126386879,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.8003,
+      "step": 2015
+    },
+    {
+      "epoch": 0.48721659430776654,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.8965,
+      "step": 2020
+    },
+    {
+      "epoch": 0.48842257597684513,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.8556,
+      "step": 2025
+    },
+    {
+      "epoch": 0.4896285576459238,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3e-05,
+      "loss": 4.7593,
+      "step": 2030
+    },
+    {
+      "epoch": 0.4908345393150024,
+      "grad_norm": 1.875,
+      "learning_rate": 3e-05,
+      "loss": 4.8081,
+      "step": 2035
+    },
+    {
+      "epoch": 0.492040520984081,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3e-05,
+      "loss": 4.9076,
+      "step": 2040
+    },
+    {
+      "epoch": 0.49324650265315967,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.7247,
+      "step": 2045
+    },
+    {
+      "epoch": 0.4944524843222383,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.8586,
+      "step": 2050
+    },
+    {
+      "epoch": 0.4956584659913169,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.7292,
+      "step": 2055
+    },
+    {
+      "epoch": 0.49686444766039556,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.7707,
+      "step": 2060
+    },
+    {
+      "epoch": 0.4980704293294742,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.8343,
+      "step": 2065
+    },
+    {
+      "epoch": 0.4992764109985528,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 4.8711,
+      "step": 2070
+    },
+    {
+      "epoch": 0.5004823926676315,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.7347,
+      "step": 2075
+    },
+    {
+      "epoch": 0.5016883743367101,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.7419,
+      "step": 2080
+    },
+    {
+      "epoch": 0.5028943560057887,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.7517,
+      "step": 2085
+    },
+    {
+      "epoch": 0.5041003376748674,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.755,
+      "step": 2090
+    },
+    {
+      "epoch": 0.505306319343946,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.8667,
+      "step": 2095
+    },
+    {
+      "epoch": 0.5065123010130246,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.7972,
+      "step": 2100
+    },
+    {
+      "epoch": 0.5077182826821033,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.881,
+      "step": 2105
+    },
+    {
+      "epoch": 0.5089242643511819,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.83,
+      "step": 2110
+    },
+    {
+      "epoch": 0.5101302460202605,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.8669,
+      "step": 2115
+    },
+    {
+      "epoch": 0.5113362276893392,
+      "grad_norm": 3.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.8719,
+      "step": 2120
+    },
+    {
+      "epoch": 0.5125422093584178,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 4.8299,
+      "step": 2125
+    },
+    {
+      "epoch": 0.5137481910274964,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.8699,
+      "step": 2130
+    },
+    {
+      "epoch": 0.5149541726965751,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.8022,
+      "step": 2135
+    },
+    {
+      "epoch": 0.5161601543656537,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.815,
+      "step": 2140
+    },
+    {
+      "epoch": 0.5173661360347322,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.7154,
+      "step": 2145
+    },
+    {
+      "epoch": 0.518572117703811,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.7906,
+      "step": 2150
+    },
+    {
+      "epoch": 0.5197780993728895,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.7535,
+      "step": 2155
+    },
+    {
+      "epoch": 0.5209840810419681,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.6967,
+      "step": 2160
+    },
+    {
+      "epoch": 0.5221900627110468,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.6081,
+      "step": 2165
+    },
+    {
+      "epoch": 0.5233960443801254,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.7869,
+      "step": 2170
+    },
+    {
+      "epoch": 0.524602026049204,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.7334,
+      "step": 2175
+    },
+    {
+      "epoch": 0.5258080077182827,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.7554,
+      "step": 2180
+    },
+    {
+      "epoch": 0.5270139893873613,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.6895,
+      "step": 2185
+    },
+    {
+      "epoch": 0.5282199710564399,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.7401,
+      "step": 2190
+    },
+    {
+      "epoch": 0.5294259527255186,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.7774,
+      "step": 2195
+    },
+    {
+      "epoch": 0.5306319343945972,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.665,
+      "step": 2200
+    },
+    {
+      "epoch": 0.5318379160636758,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.7083,
+      "step": 2205
+    },
+    {
+      "epoch": 0.5330438977327545,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.7318,
+      "step": 2210
+    },
+    {
+      "epoch": 0.5342498794018331,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.7261,
+      "step": 2215
+    },
+    {
+      "epoch": 0.5354558610709117,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.6802,
+      "step": 2220
+    },
+    {
+      "epoch": 0.5366618427399904,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.7567,
+      "step": 2225
+    },
+    {
+      "epoch": 0.537867824409069,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.6111,
+      "step": 2230
+    },
+    {
+      "epoch": 0.5390738060781476,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.7502,
+      "step": 2235
+    },
+    {
+      "epoch": 0.5402797877472263,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.5811,
+      "step": 2240
+    },
+    {
+      "epoch": 0.5414857694163049,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.7408,
+      "step": 2245
+    },
+    {
+      "epoch": 0.5426917510853835,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.7826,
+      "step": 2250
+    },
+    {
+      "epoch": 0.5438977327544622,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.6391,
+      "step": 2255
+    },
+    {
+      "epoch": 0.5451037144235408,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.7174,
+      "step": 2260
+    },
+    {
+      "epoch": 0.5463096960926194,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3e-05,
+      "loss": 4.6833,
+      "step": 2265
+    },
+    {
+      "epoch": 0.5475156777616981,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.7697,
+      "step": 2270
+    },
+    {
+      "epoch": 0.5487216594307767,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.7611,
+      "step": 2275
+    },
+    {
+      "epoch": 0.5499276410998553,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.5988,
+      "step": 2280
+    },
+    {
+      "epoch": 0.551133622768934,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.7913,
+      "step": 2285
+    },
+    {
+      "epoch": 0.5523396044380126,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.7792,
+      "step": 2290
+    },
+    {
+      "epoch": 0.5535455861070911,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.6986,
+      "step": 2295
+    },
+    {
+      "epoch": 0.5547515677761699,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.8458,
+      "step": 2300
+    },
+    {
+      "epoch": 0.5559575494452484,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.7553,
+      "step": 2305
+    },
+    {
+      "epoch": 0.557163531114327,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.5985,
+      "step": 2310
+    },
+    {
+      "epoch": 0.5583695127834057,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.6457,
+      "step": 2315
+    },
+    {
+      "epoch": 0.5595754944524843,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.8647,
+      "step": 2320
+    },
+    {
+      "epoch": 0.5607814761215629,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.7258,
+      "step": 2325
+    },
+    {
+      "epoch": 0.5619874577906416,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.7187,
+      "step": 2330
+    },
+    {
+      "epoch": 0.5631934394597202,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.5816,
+      "step": 2335
+    },
+    {
+      "epoch": 0.5643994211287988,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.6699,
+      "step": 2340
+    },
+    {
+      "epoch": 0.5656054027978775,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.5736,
+      "step": 2345
+    },
+    {
+      "epoch": 0.5668113844669561,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.7388,
+      "step": 2350
+    },
+    {
+      "epoch": 0.5680173661360347,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.9559,
+      "step": 2355
+    },
+    {
+      "epoch": 0.5692233478051134,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3e-05,
+      "loss": 4.6631,
+      "step": 2360
+    },
+    {
+      "epoch": 0.570429329474192,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.4889,
+      "step": 2365
+    },
+    {
+      "epoch": 0.5716353111432706,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.7725,
+      "step": 2370
+    },
+    {
+      "epoch": 0.5728412928123493,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.8148,
+      "step": 2375
+    },
+    {
+      "epoch": 0.5740472744814279,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.5893,
+      "step": 2380
+    },
+    {
+      "epoch": 0.5752532561505065,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.7027,
+      "step": 2385
+    },
+    {
+      "epoch": 0.5764592378195852,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.6606,
+      "step": 2390
+    },
+    {
+      "epoch": 0.5776652194886638,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.7504,
+      "step": 2395
+    },
+    {
+      "epoch": 0.5788712011577424,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.6643,
+      "step": 2400
+    },
+    {
+      "epoch": 0.5800771828268211,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.7539,
+      "step": 2405
+    },
+    {
+      "epoch": 0.5812831644958997,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.6503,
+      "step": 2410
+    },
+    {
+      "epoch": 0.5824891461649783,
+      "grad_norm": 1.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.7897,
+      "step": 2415
+    },
+    {
+      "epoch": 0.583695127834057,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.6341,
+      "step": 2420
+    },
+    {
+      "epoch": 0.5849011095031356,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 4.5938,
+      "step": 2425
+    },
+    {
+      "epoch": 0.5861070911722142,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.8009,
+      "step": 2430
+    },
+    {
+      "epoch": 0.5873130728412929,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.6057,
+      "step": 2435
+    },
+    {
+      "epoch": 0.5885190545103715,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.7151,
+      "step": 2440
+    },
+    {
+      "epoch": 0.58972503617945,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.72,
+      "step": 2445
+    },
+    {
+      "epoch": 0.5909310178485288,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.6675,
+      "step": 2450
+    },
+    {
+      "epoch": 0.5921369995176073,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.6394,
+      "step": 2455
+    },
+    {
+      "epoch": 0.5933429811866859,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.7655,
+      "step": 2460
+    },
+    {
+      "epoch": 0.5945489628557646,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.5585,
+      "step": 2465
+    },
+    {
+      "epoch": 0.5957549445248432,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.6956,
+      "step": 2470
+    },
+    {
+      "epoch": 0.5969609261939218,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3e-05,
+      "loss": 4.5783,
+      "step": 2475
+    },
+    {
+      "epoch": 0.5981669078630005,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.5324,
+      "step": 2480
+    },
+    {
+      "epoch": 0.5993728895320791,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.6965,
+      "step": 2485
+    },
+    {
+      "epoch": 0.6005788712011577,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.5978,
+      "step": 2490
+    },
+    {
+      "epoch": 0.6017848528702364,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.7335,
+      "step": 2495
+    },
+    {
+      "epoch": 0.602990834539315,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.6361,
+      "step": 2500
+    },
+    {
+      "epoch": 0.6041968162083936,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.7529,
+      "step": 2505
+    },
+    {
+      "epoch": 0.6054027978774723,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.7204,
+      "step": 2510
+    },
+    {
+      "epoch": 0.6066087795465509,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.6322,
+      "step": 2515
+    },
+    {
+      "epoch": 0.6078147612156295,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.568,
+      "step": 2520
+    },
+    {
+      "epoch": 0.6090207428847082,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.704,
+      "step": 2525
+    },
+    {
+      "epoch": 0.6102267245537868,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.6954,
+      "step": 2530
+    },
+    {
+      "epoch": 0.6114327062228654,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.675,
+      "step": 2535
+    },
+    {
+      "epoch": 0.6126386878919441,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.7965,
+      "step": 2540
+    },
+    {
+      "epoch": 0.6138446695610227,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.7336,
+      "step": 2545
+    },
+    {
+      "epoch": 0.6150506512301013,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.7419,
+      "step": 2550
+    },
+    {
+      "epoch": 0.61625663289918,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.5528,
+      "step": 2555
+    },
+    {
+      "epoch": 0.6174626145682586,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 4.5106,
+      "step": 2560
+    },
+    {
+      "epoch": 0.6186685962373372,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.5695,
+      "step": 2565
+    },
+    {
+      "epoch": 0.6198745779064159,
+      "grad_norm": 1.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.6957,
+      "step": 2570
+    },
+    {
+      "epoch": 0.6210805595754945,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.5537,
+      "step": 2575
+    },
+    {
+      "epoch": 0.622286541244573,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.6757,
+      "step": 2580
+    },
+    {
+      "epoch": 0.6234925229136518,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.5825,
+      "step": 2585
+    },
+    {
+      "epoch": 0.6246985045827304,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.5221,
+      "step": 2590
+    },
+    {
+      "epoch": 0.625904486251809,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.6702,
+      "step": 2595
+    },
+    {
+      "epoch": 0.6271104679208876,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.4973,
+      "step": 2600
+    },
+    {
+      "epoch": 0.6283164495899662,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.7989,
+      "step": 2605
+    },
+    {
+      "epoch": 0.6295224312590448,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.6821,
+      "step": 2610
+    },
+    {
+      "epoch": 0.6307284129281235,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.6938,
+      "step": 2615
+    },
+    {
+      "epoch": 0.6319343945972021,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.5866,
+      "step": 2620
+    },
+    {
+      "epoch": 0.6331403762662807,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.6921,
+      "step": 2625
+    },
+    {
+      "epoch": 0.6343463579353594,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.5261,
+      "step": 2630
+    },
+    {
+      "epoch": 0.635552339604438,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 4.616,
+      "step": 2635
+    },
+    {
+      "epoch": 0.6367583212735166,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.5105,
+      "step": 2640
+    },
+    {
+      "epoch": 0.6379643029425953,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.5342,
+      "step": 2645
+    },
+    {
+      "epoch": 0.6391702846116739,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.5575,
+      "step": 2650
+    },
+    {
+      "epoch": 0.6403762662807525,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.6319,
+      "step": 2655
+    },
+    {
+      "epoch": 0.6415822479498312,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.7018,
+      "step": 2660
+    },
+    {
+      "epoch": 0.6427882296189098,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.774,
+      "step": 2665
+    },
+    {
+      "epoch": 0.6439942112879884,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.6651,
+      "step": 2670
+    },
+    {
+      "epoch": 0.6452001929570671,
+      "grad_norm": 3.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.6475,
+      "step": 2675
+    },
+    {
+      "epoch": 0.6464061746261457,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.4535,
+      "step": 2680
+    },
+    {
+      "epoch": 0.6476121562952243,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.5088,
+      "step": 2685
+    },
+    {
+      "epoch": 0.648818137964303,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.4722,
+      "step": 2690
+    },
+    {
+      "epoch": 0.6500241196333816,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 4.5021,
+      "step": 2695
+    },
+    {
+      "epoch": 0.6512301013024602,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.5359,
+      "step": 2700
+    },
+    {
+      "epoch": 0.6524360829715389,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.6285,
+      "step": 2705
+    },
+    {
+      "epoch": 0.6536420646406175,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.6462,
+      "step": 2710
+    },
+    {
+      "epoch": 0.6548480463096961,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.578,
+      "step": 2715
+    },
+    {
+      "epoch": 0.6560540279787748,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.5873,
+      "step": 2720
+    },
+    {
+      "epoch": 0.6572600096478534,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.5252,
+      "step": 2725
+    },
+    {
+      "epoch": 0.658465991316932,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.6603,
+      "step": 2730
+    },
+    {
+      "epoch": 0.6596719729860107,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.707,
+      "step": 2735
+    },
+    {
+      "epoch": 0.6608779546550893,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.7296,
+      "step": 2740
+    },
+    {
+      "epoch": 0.6620839363241678,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.554,
+      "step": 2745
+    },
+    {
+      "epoch": 0.6632899179932465,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 4.6896,
+      "step": 2750
+    },
+    {
+      "epoch": 0.6644958996623251,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.5461,
+      "step": 2755
+    },
+    {
+      "epoch": 0.6657018813314037,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.5021,
+      "step": 2760
+    },
+    {
+      "epoch": 0.6669078630004824,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.7011,
+      "step": 2765
+    },
+    {
+      "epoch": 0.668113844669561,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.5652,
+      "step": 2770
+    },
+    {
+      "epoch": 0.6693198263386396,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.6102,
+      "step": 2775
+    },
+    {
+      "epoch": 0.6705258080077183,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.6903,
+      "step": 2780
+    },
+    {
+      "epoch": 0.6717317896767969,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.7055,
+      "step": 2785
+    },
+    {
+      "epoch": 0.6729377713458755,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.5797,
+      "step": 2790
+    },
+    {
+      "epoch": 0.6741437530149542,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.4644,
+      "step": 2795
+    },
+    {
+      "epoch": 0.6753497346840328,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.5142,
+      "step": 2800
+    },
+    {
+      "epoch": 0.6765557163531114,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.6188,
+      "step": 2805
+    },
+    {
+      "epoch": 0.6777616980221901,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.2891,
+      "step": 2810
+    },
+    {
+      "epoch": 0.6789676796912687,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.6081,
+      "step": 2815
+    },
+    {
+      "epoch": 0.6801736613603473,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.6215,
+      "step": 2820
+    },
+    {
+      "epoch": 0.681379643029426,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.4295,
+      "step": 2825
+    },
+    {
+      "epoch": 0.6825856246985046,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.5824,
+      "step": 2830
+    },
+    {
+      "epoch": 0.6837916063675832,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.4736,
+      "step": 2835
+    },
+    {
+      "epoch": 0.6849975880366619,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.5933,
+      "step": 2840
+    },
+    {
+      "epoch": 0.6862035697057405,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.5604,
+      "step": 2845
+    },
+    {
+      "epoch": 0.6874095513748191,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.4067,
+      "step": 2850
+    },
+    {
+      "epoch": 0.6886155330438978,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.4835,
+      "step": 2855
+    },
+    {
+      "epoch": 0.6898215147129764,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.4981,
+      "step": 2860
+    },
+    {
+      "epoch": 0.691027496382055,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.5609,
+      "step": 2865
+    },
+    {
+      "epoch": 0.6922334780511337,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.4212,
+      "step": 2870
+    },
+    {
+      "epoch": 0.6934394597202123,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.533,
+      "step": 2875
+    },
+    {
+      "epoch": 0.6946454413892909,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.5442,
+      "step": 2880
+    },
+    {
+      "epoch": 0.6958514230583696,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.5254,
+      "step": 2885
+    },
+    {
+      "epoch": 0.6970574047274481,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.352,
+      "step": 2890
+    },
+    {
+      "epoch": 0.6982633863965267,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.5573,
+      "step": 2895
+    },
+    {
+      "epoch": 0.6994693680656054,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.5445,
+      "step": 2900
+    },
+    {
+      "epoch": 0.700675349734684,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.5631,
+      "step": 2905
+    },
+    {
+      "epoch": 0.7018813314037626,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.4409,
+      "step": 2910
+    },
+    {
+      "epoch": 0.7030873130728413,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.5282,
+      "step": 2915
+    },
+    {
+      "epoch": 0.7042932947419199,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.3651,
+      "step": 2920
+    },
+    {
+      "epoch": 0.7054992764109985,
+      "grad_norm": 1.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.5946,
+      "step": 2925
+    },
+    {
+      "epoch": 0.7067052580800772,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.5357,
+      "step": 2930
+    },
+    {
+      "epoch": 0.7079112397491558,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.5109,
+      "step": 2935
+    },
+    {
+      "epoch": 0.7091172214182344,
+      "grad_norm": 1.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.4849,
+      "step": 2940
+    },
+    {
+      "epoch": 0.7103232030873131,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.4213,
+      "step": 2945
+    },
+    {
+      "epoch": 0.7115291847563917,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.4284,
+      "step": 2950
+    },
+    {
+      "epoch": 0.7127351664254703,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.4445,
+      "step": 2955
+    },
+    {
+      "epoch": 0.713941148094549,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.5524,
+      "step": 2960
+    },
+    {
+      "epoch": 0.7151471297636276,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.5115,
+      "step": 2965
+    },
+    {
+      "epoch": 0.7163531114327062,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.5595,
+      "step": 2970
+    },
+    {
+      "epoch": 0.7175590931017849,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.4096,
+      "step": 2975
+    },
+    {
+      "epoch": 0.7187650747708635,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.6125,
+      "step": 2980
+    },
+    {
+      "epoch": 0.7199710564399421,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.5777,
+      "step": 2985
+    },
+    {
+      "epoch": 0.7211770381090208,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.411,
+      "step": 2990
+    },
+    {
+      "epoch": 0.7223830197780994,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.6638,
+      "step": 2995
+    },
+    {
+      "epoch": 0.723589001447178,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.4474,
+      "step": 3000
+    },
+    {
+      "epoch": 0.7247949831162567,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.4597,
+      "step": 3005
+    },
+    {
+      "epoch": 0.7260009647853353,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.4419,
+      "step": 3010
+    },
+    {
+      "epoch": 0.7272069464544139,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.3696,
+      "step": 3015
+    },
+    {
+      "epoch": 0.7284129281234926,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.442,
+      "step": 3020
+    },
+    {
+      "epoch": 0.7296189097925712,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.3994,
+      "step": 3025
+    },
+    {
+      "epoch": 0.7308248914616498,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.5059,
+      "step": 3030
+    },
+    {
+      "epoch": 0.7320308731307285,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.5434,
+      "step": 3035
+    },
+    {
+      "epoch": 0.733236854799807,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.4819,
+      "step": 3040
+    },
+    {
+      "epoch": 0.7344428364688856,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.4458,
+      "step": 3045
+    },
+    {
+      "epoch": 0.7356488181379643,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.4622,
+      "step": 3050
+    },
+    {
+      "epoch": 0.7368547998070429,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.521,
+      "step": 3055
+    },
+    {
+      "epoch": 0.7380607814761215,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.4673,
+      "step": 3060
+    },
+    {
+      "epoch": 0.7392667631452002,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.4084,
+      "step": 3065
+    },
+    {
+      "epoch": 0.7404727448142788,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.416,
+      "step": 3070
+    },
+    {
+      "epoch": 0.7416787264833574,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.4623,
+      "step": 3075
+    },
+    {
+      "epoch": 0.7428847081524361,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.6505,
+      "step": 3080
+    },
+    {
+      "epoch": 0.7440906898215147,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.4949,
+      "step": 3085
+    },
+    {
+      "epoch": 0.7452966714905933,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.5213,
+      "step": 3090
+    },
+    {
+      "epoch": 0.746502653159672,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3e-05,
+      "loss": 4.3938,
+      "step": 3095
+    },
+    {
+      "epoch": 0.7477086348287506,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.3599,
+      "step": 3100
+    },
+    {
+      "epoch": 0.7489146164978292,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.2375,
+      "step": 3105
+    },
+    {
+      "epoch": 0.7501205981669079,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.566,
+      "step": 3110
+    },
+    {
+      "epoch": 0.7513265798359865,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.6481,
+      "step": 3115
+    },
+    {
+      "epoch": 0.7525325615050651,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.6129,
+      "step": 3120
+    },
+    {
+      "epoch": 0.7537385431741438,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.6412,
+      "step": 3125
+    },
+    {
+      "epoch": 0.7549445248432224,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.3374,
+      "step": 3130
+    },
+    {
+      "epoch": 0.756150506512301,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.5581,
+      "step": 3135
+    },
+    {
+      "epoch": 0.7573564881813797,
+      "grad_norm": 1.8046875,
+      "learning_rate": 3e-05,
+      "loss": 4.4636,
+      "step": 3140
+    },
+    {
+      "epoch": 0.7585624698504583,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.7031,
+      "step": 3145
+    },
+    {
+      "epoch": 0.7597684515195369,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.4857,
+      "step": 3150
+    },
+    {
+      "epoch": 0.7609744331886156,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.3713,
+      "step": 3155
+    },
+    {
+      "epoch": 0.7621804148576942,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.5158,
+      "step": 3160
+    },
+    {
+      "epoch": 0.7633863965267728,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.3696,
+      "step": 3165
+    },
+    {
+      "epoch": 0.7645923781958515,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.3256,
+      "step": 3170
+    },
+    {
+      "epoch": 0.7657983598649301,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3e-05,
+      "loss": 4.3505,
+      "step": 3175
+    },
+    {
+      "epoch": 0.7670043415340086,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.4221,
+      "step": 3180
+    },
+    {
+      "epoch": 0.7682103232030874,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.609,
+      "step": 3185
+    },
+    {
+      "epoch": 0.769416304872166,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.4097,
+      "step": 3190
+    },
+    {
+      "epoch": 0.7706222865412445,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.3881,
+      "step": 3195
+    },
+    {
+      "epoch": 0.7718282682103232,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.3572,
+      "step": 3200
+    },
+    {
+      "epoch": 0.7730342498794018,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.3264,
+      "step": 3205
+    },
+    {
+      "epoch": 0.7742402315484804,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1551,
+      "step": 3210
+    },
+    {
+      "epoch": 0.7754462132175591,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.6972,
+      "step": 3215
+    },
+    {
+      "epoch": 0.7766521948866377,
+      "grad_norm": 3.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.6838,
+      "step": 3220
+    },
+    {
+      "epoch": 0.7778581765557163,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.4757,
+      "step": 3225
+    },
+    {
+      "epoch": 0.779064158224795,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.4435,
+      "step": 3230
+    },
+    {
+      "epoch": 0.7802701398938736,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.3932,
+      "step": 3235
+    },
+    {
+      "epoch": 0.7814761215629522,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.4354,
+      "step": 3240
+    },
+    {
+      "epoch": 0.7826821032320309,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.4226,
+      "step": 3245
+    },
+    {
+      "epoch": 0.7838880849011095,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.5854,
+      "step": 3250
+    },
+    {
+      "epoch": 0.7850940665701881,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.417,
+      "step": 3255
+    },
+    {
+      "epoch": 0.7863000482392668,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.4898,
+      "step": 3260
+    },
+    {
+      "epoch": 0.7875060299083454,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.5403,
+      "step": 3265
+    },
+    {
+      "epoch": 0.788712011577424,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.6457,
+      "step": 3270
+    },
+    {
+      "epoch": 0.7899179932465027,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.3593,
+      "step": 3275
+    },
+    {
+      "epoch": 0.7911239749155813,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.3323,
+      "step": 3280
+    },
+    {
+      "epoch": 0.7923299565846599,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.5542,
+      "step": 3285
+    },
+    {
+      "epoch": 0.7935359382537386,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.521,
+      "step": 3290
+    },
+    {
+      "epoch": 0.7947419199228172,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.4324,
+      "step": 3295
+    },
+    {
+      "epoch": 0.7959479015918958,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.4935,
+      "step": 3300
+    },
+    {
+      "epoch": 0.7971538832609745,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.4216,
+      "step": 3305
+    },
+    {
+      "epoch": 0.7983598649300531,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.4761,
+      "step": 3310
+    },
+    {
+      "epoch": 0.7995658465991317,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.4752,
+      "step": 3315
+    },
+    {
+      "epoch": 0.8007718282682104,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.409,
+      "step": 3320
+    },
+    {
+      "epoch": 0.801977809937289,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.3812,
+      "step": 3325
+    },
+    {
+      "epoch": 0.8031837916063675,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.4513,
+      "step": 3330
+    },
+    {
+      "epoch": 0.8043897732754463,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.502,
+      "step": 3335
+    },
+    {
+      "epoch": 0.8055957549445248,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.3314,
+      "step": 3340
+    },
+    {
+      "epoch": 0.8068017366136034,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.513,
+      "step": 3345
+    },
+    {
+      "epoch": 0.8080077182826821,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.4607,
+      "step": 3350
+    },
+    {
+      "epoch": 0.8092136999517607,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.382,
+      "step": 3355
+    },
+    {
+      "epoch": 0.8104196816208393,
+      "grad_norm": 1.8671875,
+      "learning_rate": 3e-05,
+      "loss": 4.3694,
+      "step": 3360
+    },
+    {
+      "epoch": 0.811625663289918,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.36,
+      "step": 3365
+    },
+    {
+      "epoch": 0.8128316449589966,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.4718,
+      "step": 3370
+    },
+    {
+      "epoch": 0.8140376266280752,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.302,
+      "step": 3375
+    },
+    {
+      "epoch": 0.8152436082971539,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.4969,
+      "step": 3380
+    },
+    {
+      "epoch": 0.8164495899662325,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.3885,
+      "step": 3385
+    },
+    {
+      "epoch": 0.8176555716353111,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.4318,
+      "step": 3390
+    },
+    {
+      "epoch": 0.8188615533043898,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.4583,
+      "step": 3395
+    },
+    {
+      "epoch": 0.8200675349734684,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.2654,
+      "step": 3400
+    },
+    {
+      "epoch": 0.821273516642547,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.3499,
+      "step": 3405
+    },
+    {
+      "epoch": 0.8224794983116257,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.3371,
+      "step": 3410
+    },
+    {
+      "epoch": 0.8236854799807043,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.3177,
+      "step": 3415
+    },
+    {
+      "epoch": 0.8248914616497829,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.3892,
+      "step": 3420
+    },
+    {
+      "epoch": 0.8260974433188616,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.3503,
+      "step": 3425
+    },
+    {
+      "epoch": 0.8273034249879402,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.4576,
+      "step": 3430
+    },
+    {
+      "epoch": 0.8285094066570188,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.4192,
+      "step": 3435
+    },
+    {
+      "epoch": 0.8297153883260975,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.4293,
+      "step": 3440
+    },
+    {
+      "epoch": 0.8309213699951761,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.3069,
+      "step": 3445
+    },
+    {
+      "epoch": 0.8321273516642547,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.432,
+      "step": 3450
+    },
+    {
+      "epoch": 0.8333333333333334,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.3897,
+      "step": 3455
+    },
+    {
+      "epoch": 0.834539315002412,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.4484,
+      "step": 3460
+    },
+    {
+      "epoch": 0.8357452966714906,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.2252,
+      "step": 3465
+    },
+    {
+      "epoch": 0.8369512783405693,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.3953,
+      "step": 3470
+    },
+    {
+      "epoch": 0.8381572600096479,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.4112,
+      "step": 3475
+    },
+    {
+      "epoch": 0.8393632416787264,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.4335,
+      "step": 3480
+    },
+    {
+      "epoch": 0.8405692233478052,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.4711,
+      "step": 3485
+    },
+    {
+      "epoch": 0.8417752050168837,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.2862,
+      "step": 3490
+    },
+    {
+      "epoch": 0.8429811866859623,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.2646,
+      "step": 3495
+    },
+    {
+      "epoch": 0.844187168355041,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.3446,
+      "step": 3500
+    },
+    {
+      "epoch": 0.8453931500241196,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.2989,
+      "step": 3505
+    },
+    {
+      "epoch": 0.8465991316931982,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.3109,
+      "step": 3510
+    },
+    {
+      "epoch": 0.8478051133622769,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.5348,
+      "step": 3515
+    },
+    {
+      "epoch": 0.8490110950313555,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.3558,
+      "step": 3520
+    },
+    {
+      "epoch": 0.8502170767004341,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.5995,
+      "step": 3525
+    },
+    {
+      "epoch": 0.8514230583695128,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.4,
+      "step": 3530
+    },
+    {
+      "epoch": 0.8526290400385914,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.3557,
+      "step": 3535
+    },
+    {
+      "epoch": 0.85383502170767,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.3707,
+      "step": 3540
+    },
+    {
+      "epoch": 0.8550410033767487,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.2456,
+      "step": 3545
+    },
+    {
+      "epoch": 0.8562469850458273,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.2662,
+      "step": 3550
+    },
+    {
+      "epoch": 0.8574529667149059,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.3337,
+      "step": 3555
+    },
+    {
+      "epoch": 0.8586589483839846,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.4318,
+      "step": 3560
+    },
+    {
+      "epoch": 0.8598649300530632,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.2292,
+      "step": 3565
+    },
+    {
+      "epoch": 0.8610709117221418,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.201,
+      "step": 3570
+    },
+    {
+      "epoch": 0.8622768933912205,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.4233,
+      "step": 3575
+    },
+    {
+      "epoch": 0.8634828750602991,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.5371,
+      "step": 3580
+    },
+    {
+      "epoch": 0.8646888567293777,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.2995,
+      "step": 3585
+    },
+    {
+      "epoch": 0.8658948383984564,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.4251,
+      "step": 3590
+    },
+    {
+      "epoch": 0.867100820067535,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.3139,
+      "step": 3595
+    },
+    {
+      "epoch": 0.8683068017366136,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.2808,
+      "step": 3600
+    },
+    {
+      "epoch": 0.8695127834056923,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.3172,
+      "step": 3605
+    },
+    {
+      "epoch": 0.8707187650747709,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.3034,
+      "step": 3610
+    },
+    {
+      "epoch": 0.8719247467438495,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2873,
+      "step": 3615
+    },
+    {
+      "epoch": 0.8731307284129282,
+      "grad_norm": 3.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.3963,
+      "step": 3620
+    },
+    {
+      "epoch": 0.8743367100820068,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.3578,
+      "step": 3625
+    },
+    {
+      "epoch": 0.8755426917510853,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.2587,
+      "step": 3630
+    },
+    {
+      "epoch": 0.876748673420164,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.4614,
+      "step": 3635
+    },
+    {
+      "epoch": 0.8779546550892426,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1304,
+      "step": 3640
+    },
+    {
+      "epoch": 0.8791606367583212,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.2083,
+      "step": 3645
+    },
+    {
+      "epoch": 0.8803666184273999,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.2815,
+      "step": 3650
+    },
+    {
+      "epoch": 0.8815726000964785,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.4023,
+      "step": 3655
+    },
+    {
+      "epoch": 0.8827785817655571,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0872,
+      "step": 3660
+    },
+    {
+      "epoch": 0.8839845634346358,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 4.4202,
+      "step": 3665
+    },
+    {
+      "epoch": 0.8851905451037144,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.4167,
+      "step": 3670
+    },
+    {
+      "epoch": 0.886396526772793,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 4.2922,
+      "step": 3675
+    },
+    {
+      "epoch": 0.8876025084418717,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.5491,
+      "step": 3680
+    },
+    {
+      "epoch": 0.8888084901109503,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0671,
+      "step": 3685
+    },
+    {
+      "epoch": 0.8900144717800289,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.4984,
+      "step": 3690
+    },
+    {
+      "epoch": 0.8912204534491076,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.3098,
+      "step": 3695
+    },
+    {
+      "epoch": 0.8924264351181862,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.2841,
+      "step": 3700
+    },
+    {
+      "epoch": 0.8936324167872648,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.4347,
+      "step": 3705
+    },
+    {
+      "epoch": 0.8948383984563435,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.3184,
+      "step": 3710
+    },
+    {
+      "epoch": 0.8960443801254221,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2526,
+      "step": 3715
+    },
+    {
+      "epoch": 0.8972503617945007,
+      "grad_norm": 3.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.3653,
+      "step": 3720
+    },
+    {
+      "epoch": 0.8984563434635794,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.1791,
+      "step": 3725
+    },
+    {
+      "epoch": 0.899662325132658,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.3713,
+      "step": 3730
+    },
+    {
+      "epoch": 0.9008683068017366,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.278,
+      "step": 3735
+    },
+    {
+      "epoch": 0.9020742884708153,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.3953,
+      "step": 3740
+    },
+    {
+      "epoch": 0.9032802701398939,
+      "grad_norm": 1.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.248,
+      "step": 3745
+    },
+    {
+      "epoch": 0.9044862518089725,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.4365,
+      "step": 3750
+    },
+    {
+      "epoch": 0.9056922334780512,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.3308,
+      "step": 3755
+    },
+    {
+      "epoch": 0.9068982151471298,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.404,
+      "step": 3760
+    },
+    {
+      "epoch": 0.9081041968162084,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.2406,
+      "step": 3765
+    },
+    {
+      "epoch": 0.9093101784852871,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.4286,
+      "step": 3770
+    },
+    {
+      "epoch": 0.9105161601543657,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 4.2325,
+      "step": 3775
+    },
+    {
+      "epoch": 0.9117221418234442,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.1555,
+      "step": 3780
+    },
+    {
+      "epoch": 0.912928123492523,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.261,
+      "step": 3785
+    },
+    {
+      "epoch": 0.9141341051616015,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.234,
+      "step": 3790
+    },
+    {
+      "epoch": 0.9153400868306801,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.2657,
+      "step": 3795
+    },
+    {
+      "epoch": 0.9165460684997588,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.183,
+      "step": 3800
+    },
+    {
+      "epoch": 0.9177520501688374,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.4021,
+      "step": 3805
+    },
+    {
+      "epoch": 0.918958031837916,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.3369,
+      "step": 3810
+    },
+    {
+      "epoch": 0.9201640135069947,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.2254,
+      "step": 3815
+    },
+    {
+      "epoch": 0.9213699951760733,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.364,
+      "step": 3820
+    },
+    {
+      "epoch": 0.9225759768451519,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.2053,
+      "step": 3825
+    },
+    {
+      "epoch": 0.9237819585142306,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.2441,
+      "step": 3830
+    },
+    {
+      "epoch": 0.9249879401833092,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.2054,
+      "step": 3835
+    },
+    {
+      "epoch": 0.9261939218523878,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.3761,
+      "step": 3840
+    },
+    {
+      "epoch": 0.9273999035214665,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 4.2734,
+      "step": 3845
+    },
+    {
+      "epoch": 0.9286058851905451,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1227,
+      "step": 3850
+    },
+    {
+      "epoch": 0.9298118668596237,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.4329,
+      "step": 3855
+    },
+    {
+      "epoch": 0.9310178485287024,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.3229,
+      "step": 3860
+    },
+    {
+      "epoch": 0.932223830197781,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.2651,
+      "step": 3865
+    },
+    {
+      "epoch": 0.9334298118668596,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.3458,
+      "step": 3870
+    },
+    {
+      "epoch": 0.9346357935359383,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.2634,
+      "step": 3875
+    },
+    {
+      "epoch": 0.9358417752050169,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.3467,
+      "step": 3880
+    },
+    {
+      "epoch": 0.9370477568740955,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1485,
+      "step": 3885
+    },
+    {
+      "epoch": 0.9382537385431742,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.368,
+      "step": 3890
+    },
+    {
+      "epoch": 0.9394597202122528,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.2311,
+      "step": 3895
+    },
+    {
+      "epoch": 0.9406657018813314,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.3066,
+      "step": 3900
+    },
+    {
+      "epoch": 0.9418716835504101,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.2629,
+      "step": 3905
+    },
+    {
+      "epoch": 0.9430776652194887,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.2645,
+      "step": 3910
+    },
+    {
+      "epoch": 0.9442836468885673,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.3612,
+      "step": 3915
+    },
+    {
+      "epoch": 0.945489628557646,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.245,
+      "step": 3920
+    },
+    {
+      "epoch": 0.9466956102267245,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.3571,
+      "step": 3925
+    },
+    {
+      "epoch": 0.9479015918958031,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1609,
+      "step": 3930
+    },
+    {
+      "epoch": 0.9491075735648818,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.2638,
+      "step": 3935
+    },
+    {
+      "epoch": 0.9503135552339604,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.3612,
+      "step": 3940
+    },
+    {
+      "epoch": 0.951519536903039,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.2925,
+      "step": 3945
+    },
+    {
+      "epoch": 0.9527255185721177,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1713,
+      "step": 3950
+    },
+    {
+      "epoch": 0.9539315002411963,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1754,
+      "step": 3955
+    },
+    {
+      "epoch": 0.9551374819102749,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.3065,
+      "step": 3960
+    },
+    {
+      "epoch": 0.9563434635793536,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.1348,
+      "step": 3965
+    },
+    {
+      "epoch": 0.9575494452484322,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.2436,
+      "step": 3970
+    },
+    {
+      "epoch": 0.9587554269175108,
+      "grad_norm": 3.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.3451,
+      "step": 3975
+    },
+    {
+      "epoch": 0.9599614085865895,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.3137,
+      "step": 3980
+    },
+    {
+      "epoch": 0.9611673902556681,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1208,
+      "step": 3985
+    },
+    {
+      "epoch": 0.9623733719247467,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0812,
+      "step": 3990
+    },
+    {
+      "epoch": 0.9635793535938254,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.3247,
+      "step": 3995
+    },
+    {
+      "epoch": 0.964785335262904,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 4.304,
+      "step": 4000
+    },
+    {
+      "epoch": 0.9659913169319826,
+      "grad_norm": 1.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.3427,
+      "step": 4005
+    },
+    {
+      "epoch": 0.9671972986010613,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.1484,
+      "step": 4010
+    },
+    {
+      "epoch": 0.9684032802701399,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.1438,
+      "step": 4015
+    },
+    {
+      "epoch": 0.9696092619392185,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.1749,
+      "step": 4020
+    },
+    {
+      "epoch": 0.9708152436082972,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.2491,
+      "step": 4025
+    },
+    {
+      "epoch": 0.9720212252773758,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.2584,
+      "step": 4030
+    },
+    {
+      "epoch": 0.9732272069464544,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.296,
+      "step": 4035
+    },
+    {
+      "epoch": 0.9744331886155331,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.3104,
+      "step": 4040
+    },
+    {
+      "epoch": 0.9756391702846117,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.1227,
+      "step": 4045
+    },
+    {
+      "epoch": 0.9768451519536903,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0633,
+      "step": 4050
+    },
+    {
+      "epoch": 0.978051133622769,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.2781,
+      "step": 4055
+    },
+    {
+      "epoch": 0.9792571152918476,
+      "grad_norm": 3.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.3384,
+      "step": 4060
+    },
+    {
+      "epoch": 0.9804630969609262,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.3135,
+      "step": 4065
+    },
+    {
+      "epoch": 0.9816690786300049,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.364,
+      "step": 4070
+    },
+    {
+      "epoch": 0.9828750602990834,
+      "grad_norm": 3.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1734,
+      "step": 4075
+    },
+    {
+      "epoch": 0.984081041968162,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1808,
+      "step": 4080
+    },
+    {
+      "epoch": 0.9852870236372407,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1457,
+      "step": 4085
+    },
+    {
+      "epoch": 0.9864930053063193,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1074,
+      "step": 4090
+    },
+    {
+      "epoch": 0.9876989869753979,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1239,
+      "step": 4095
+    },
+    {
+      "epoch": 0.9889049686444766,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.4142,
+      "step": 4100
+    },
+    {
+      "epoch": 0.9901109503135552,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.2597,
+      "step": 4105
+    },
+    {
+      "epoch": 0.9913169319826338,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.2161,
+      "step": 4110
+    },
+    {
+      "epoch": 0.9925229136517125,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1805,
+      "step": 4115
+    },
+    {
+      "epoch": 0.9937288953207911,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.342,
+      "step": 4120
+    },
+    {
+      "epoch": 0.9949348769898697,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0627,
+      "step": 4125
+    },
+    {
+      "epoch": 0.9961408586589484,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.1565,
+      "step": 4130
+    },
+    {
+      "epoch": 0.997346840328027,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.1849,
+      "step": 4135
+    },
+    {
+      "epoch": 0.9985528219971056,
+      "grad_norm": 1.8671875,
+      "learning_rate": 3e-05,
+      "loss": 4.2711,
+      "step": 4140
+    },
+    {
+      "epoch": 0.9997588036661843,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.3792,
+      "step": 4145
+    },
+    {
+      "epoch": 1.000964785335263,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.1339,
+      "step": 4150
+    },
+    {
+      "epoch": 1.0021707670043416,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0774,
+      "step": 4155
+    },
+    {
+      "epoch": 1.0033767486734202,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.3682,
+      "step": 4160
+    },
+    {
+      "epoch": 1.0045827303424988,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.2177,
+      "step": 4165
+    },
+    {
+      "epoch": 1.0057887120115774,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.2855,
+      "step": 4170
+    },
+    {
+      "epoch": 1.006994693680656,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1398,
+      "step": 4175
+    },
+    {
+      "epoch": 1.0082006753497348,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.1209,
+      "step": 4180
+    },
+    {
+      "epoch": 1.0094066570188134,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.3527,
+      "step": 4185
+    },
+    {
+      "epoch": 1.010612638687892,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.3545,
+      "step": 4190
+    },
+    {
+      "epoch": 1.0118186203569706,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0483,
+      "step": 4195
+    },
+    {
+      "epoch": 1.0130246020260492,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.2368,
+      "step": 4200
+    },
+    {
+      "epoch": 1.0142305836951278,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0668,
+      "step": 4205
+    },
+    {
+      "epoch": 1.0154365653642066,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.2598,
+      "step": 4210
+    },
+    {
+      "epoch": 1.0166425470332852,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.2961,
+      "step": 4215
+    },
+    {
+      "epoch": 1.0178485287023638,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.2173,
+      "step": 4220
+    },
+    {
+      "epoch": 1.0190545103714423,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1629,
+      "step": 4225
+    },
+    {
+      "epoch": 1.020260492040521,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.102,
+      "step": 4230
+    },
+    {
+      "epoch": 1.0214664737095995,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.2525,
+      "step": 4235
+    },
+    {
+      "epoch": 1.0226724553786783,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0938,
+      "step": 4240
+    },
+    {
+      "epoch": 1.023878437047757,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.171,
+      "step": 4245
+    },
+    {
+      "epoch": 1.0250844187168355,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.2585,
+      "step": 4250
+    },
+    {
+      "epoch": 1.0262904003859141,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.1278,
+      "step": 4255
+    },
+    {
+      "epoch": 1.0274963820549927,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.2064,
+      "step": 4260
+    },
+    {
+      "epoch": 1.0287023637240713,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.2195,
+      "step": 4265
+    },
+    {
+      "epoch": 1.0299083453931501,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.2334,
+      "step": 4270
+    },
+    {
+      "epoch": 1.0311143270622287,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.128,
+      "step": 4275
+    },
+    {
+      "epoch": 1.0323203087313073,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.232,
+      "step": 4280
+    },
+    {
+      "epoch": 1.033526290400386,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1578,
+      "step": 4285
+    },
+    {
+      "epoch": 1.0347322720694645,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.2872,
+      "step": 4290
+    },
+    {
+      "epoch": 1.035938253738543,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9836,
+      "step": 4295
+    },
+    {
+      "epoch": 1.037144235407622,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.306,
+      "step": 4300
+    },
+    {
+      "epoch": 1.0383502170767005,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2063,
+      "step": 4305
+    },
+    {
+      "epoch": 1.039556198745779,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.1627,
+      "step": 4310
+    },
+    {
+      "epoch": 1.0407621804148577,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.3906,
+      "step": 4315
+    },
+    {
+      "epoch": 1.0419681620839363,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.16,
+      "step": 4320
+    },
+    {
+      "epoch": 1.0431741437530149,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.2564,
+      "step": 4325
+    },
+    {
+      "epoch": 1.0443801254220937,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.1919,
+      "step": 4330
+    },
+    {
+      "epoch": 1.0455861070911723,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1206,
+      "step": 4335
+    },
+    {
+      "epoch": 1.0467920887602509,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.2585,
+      "step": 4340
+    },
+    {
+      "epoch": 1.0479980704293295,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.2129,
+      "step": 4345
+    },
+    {
+      "epoch": 1.049204052098408,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.1281,
+      "step": 4350
+    },
+    {
+      "epoch": 1.0504100337674867,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3e-05,
+      "loss": 4.1529,
+      "step": 4355
+    },
+    {
+      "epoch": 1.0516160154365655,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.289,
+      "step": 4360
+    },
+    {
+      "epoch": 1.052821997105644,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.191,
+      "step": 4365
+    },
+    {
+      "epoch": 1.0540279787747227,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.3676,
+      "step": 4370
+    },
+    {
+      "epoch": 1.0552339604438012,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0414,
+      "step": 4375
+    },
+    {
+      "epoch": 1.0564399421128798,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1719,
+      "step": 4380
+    },
+    {
+      "epoch": 1.0576459237819584,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.2354,
+      "step": 4385
+    },
+    {
+      "epoch": 1.0588519054510372,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0022,
+      "step": 4390
+    },
+    {
+      "epoch": 1.0600578871201158,
+      "grad_norm": 1.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0969,
+      "step": 4395
+    },
+    {
+      "epoch": 1.0612638687891944,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9865,
+      "step": 4400
+    },
+    {
+      "epoch": 1.062469850458273,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1781,
+      "step": 4405
+    },
+    {
+      "epoch": 1.0636758321273516,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.2007,
+      "step": 4410
+    },
+    {
+      "epoch": 1.0648818137964302,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1212,
+      "step": 4415
+    },
+    {
+      "epoch": 1.066087795465509,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.114,
+      "step": 4420
+    },
+    {
+      "epoch": 1.0672937771345876,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.1021,
+      "step": 4425
+    },
+    {
+      "epoch": 1.0684997588036662,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.2303,
+      "step": 4430
+    },
+    {
+      "epoch": 1.0697057404727448,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1789,
+      "step": 4435
+    },
+    {
+      "epoch": 1.0709117221418234,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0521,
+      "step": 4440
+    },
+    {
+      "epoch": 1.072117703810902,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0793,
+      "step": 4445
+    },
+    {
+      "epoch": 1.0733236854799808,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.2611,
+      "step": 4450
+    },
+    {
+      "epoch": 1.0745296671490594,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.1296,
+      "step": 4455
+    },
+    {
+      "epoch": 1.075735648818138,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.1777,
+      "step": 4460
+    },
+    {
+      "epoch": 1.0769416304872166,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0622,
+      "step": 4465
+    },
+    {
+      "epoch": 1.0781476121562952,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1491,
+      "step": 4470
+    },
+    {
+      "epoch": 1.0793535938253738,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0251,
+      "step": 4475
+    },
+    {
+      "epoch": 1.0805595754944526,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1231,
+      "step": 4480
+    },
+    {
+      "epoch": 1.0817655571635312,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1179,
+      "step": 4485
+    },
+    {
+      "epoch": 1.0829715388326098,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.2324,
+      "step": 4490
+    },
+    {
+      "epoch": 1.0841775205016884,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.2711,
+      "step": 4495
+    },
+    {
+      "epoch": 1.085383502170767,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.2723,
+      "step": 4500
+    },
+    {
+      "epoch": 1.0865894838398455,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.1012,
+      "step": 4505
+    },
+    {
+      "epoch": 1.0877954655089244,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.2708,
+      "step": 4510
+    },
+    {
+      "epoch": 1.089001447178003,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1696,
+      "step": 4515
+    },
+    {
+      "epoch": 1.0902074288470815,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9826,
+      "step": 4520
+    },
+    {
+      "epoch": 1.0914134105161601,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.168,
+      "step": 4525
+    },
+    {
+      "epoch": 1.0926193921852387,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.087,
+      "step": 4530
+    },
+    {
+      "epoch": 1.0938253738543173,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1323,
+      "step": 4535
+    },
+    {
+      "epoch": 1.0950313555233961,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1108,
+      "step": 4540
+    },
+    {
+      "epoch": 1.0962373371924747,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.2787,
+      "step": 4545
+    },
+    {
+      "epoch": 1.0974433188615533,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.1382,
+      "step": 4550
+    },
+    {
+      "epoch": 1.098649300530632,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.2285,
+      "step": 4555
+    },
+    {
+      "epoch": 1.0998552821997105,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.2904,
+      "step": 4560
+    },
+    {
+      "epoch": 1.101061263868789,
+      "grad_norm": 1.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.0092,
+      "step": 4565
+    },
+    {
+      "epoch": 1.102267245537868,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0829,
+      "step": 4570
+    },
+    {
+      "epoch": 1.1034732272069465,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.3022,
+      "step": 4575
+    },
+    {
+      "epoch": 1.104679208876025,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.2022,
+      "step": 4580
+    },
+    {
+      "epoch": 1.1058851905451037,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0575,
+      "step": 4585
+    },
+    {
+      "epoch": 1.1070911722141823,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1738,
+      "step": 4590
+    },
+    {
+      "epoch": 1.1082971538832609,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.1463,
+      "step": 4595
+    },
+    {
+      "epoch": 1.1095031355523397,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.1492,
+      "step": 4600
+    },
+    {
+      "epoch": 1.1107091172214183,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.173,
+      "step": 4605
+    },
+    {
+      "epoch": 1.111915098890497,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.115,
+      "step": 4610
+    },
+    {
+      "epoch": 1.1131210805595755,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.2306,
+      "step": 4615
+    },
+    {
+      "epoch": 1.114327062228654,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0668,
+      "step": 4620
+    },
+    {
+      "epoch": 1.1155330438977327,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1718,
+      "step": 4625
+    },
+    {
+      "epoch": 1.1167390255668115,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.3329,
+      "step": 4630
+    },
+    {
+      "epoch": 1.11794500723589,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1353,
+      "step": 4635
+    },
+    {
+      "epoch": 1.1191509889049687,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1285,
+      "step": 4640
+    },
+    {
+      "epoch": 1.1203569705740473,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.143,
+      "step": 4645
+    },
+    {
+      "epoch": 1.1215629522431259,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.086,
+      "step": 4650
+    },
+    {
+      "epoch": 1.1227689339122044,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0245,
+      "step": 4655
+    },
+    {
+      "epoch": 1.1239749155812833,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0672,
+      "step": 4660
+    },
+    {
+      "epoch": 1.1251808972503619,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0632,
+      "step": 4665
+    },
+    {
+      "epoch": 1.1263868789194404,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.2852,
+      "step": 4670
+    },
+    {
+      "epoch": 1.127592860588519,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0793,
+      "step": 4675
+    },
+    {
+      "epoch": 1.1287988422575976,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1748,
+      "step": 4680
+    },
+    {
+      "epoch": 1.1300048239266762,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.111,
+      "step": 4685
+    },
+    {
+      "epoch": 1.1312108055957548,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1002,
+      "step": 4690
+    },
+    {
+      "epoch": 1.1324167872648336,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1802,
+      "step": 4695
+    },
+    {
+      "epoch": 1.1336227689339122,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.2181,
+      "step": 4700
+    },
+    {
+      "epoch": 1.1348287506029908,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1413,
+      "step": 4705
+    },
+    {
+      "epoch": 1.1360347322720694,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.1944,
+      "step": 4710
+    },
+    {
+      "epoch": 1.137240713941148,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.0352,
+      "step": 4715
+    },
+    {
+      "epoch": 1.1384466956102268,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0962,
+      "step": 4720
+    },
+    {
+      "epoch": 1.1396526772793054,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1969,
+      "step": 4725
+    },
+    {
+      "epoch": 1.140858658948384,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1846,
+      "step": 4730
+    },
+    {
+      "epoch": 1.1420646406174626,
+      "grad_norm": 1.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9844,
+      "step": 4735
+    },
+    {
+      "epoch": 1.1432706222865412,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1057,
+      "step": 4740
+    },
+    {
+      "epoch": 1.1444766039556198,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1298,
+      "step": 4745
+    },
+    {
+      "epoch": 1.1456825856246984,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.1007,
+      "step": 4750
+    },
+    {
+      "epoch": 1.1468885672937772,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.1214,
+      "step": 4755
+    },
+    {
+      "epoch": 1.1480945489628558,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.13,
+      "step": 4760
+    },
+    {
+      "epoch": 1.1493005306319344,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0098,
+      "step": 4765
+    },
+    {
+      "epoch": 1.150506512301013,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0388,
+      "step": 4770
+    },
+    {
+      "epoch": 1.1517124939700916,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9719,
+      "step": 4775
+    },
+    {
+      "epoch": 1.1529184756391704,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0633,
+      "step": 4780
+    },
+    {
+      "epoch": 1.154124457308249,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.152,
+      "step": 4785
+    },
+    {
+      "epoch": 1.1553304389773276,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.2158,
+      "step": 4790
+    },
+    {
+      "epoch": 1.1565364206464062,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.126,
+      "step": 4795
+    },
+    {
+      "epoch": 1.1577424023154848,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.2521,
+      "step": 4800
+    },
+    {
+      "epoch": 1.1589483839845633,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.2653,
+      "step": 4805
+    },
+    {
+      "epoch": 1.1601543656536422,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0527,
+      "step": 4810
+    },
+    {
+      "epoch": 1.1613603473227208,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9317,
+      "step": 4815
+    },
+    {
+      "epoch": 1.1625663289917993,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.9856,
+      "step": 4820
+    },
+    {
+      "epoch": 1.163772310660878,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1945,
+      "step": 4825
+    },
+    {
+      "epoch": 1.1649782923299565,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.1806,
+      "step": 4830
+    },
+    {
+      "epoch": 1.1661842739990351,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.1163,
+      "step": 4835
+    },
+    {
+      "epoch": 1.167390255668114,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0936,
+      "step": 4840
+    },
+    {
+      "epoch": 1.1685962373371925,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9,
+      "step": 4845
+    },
+    {
+      "epoch": 1.1698022190062711,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.161,
+      "step": 4850
+    },
+    {
+      "epoch": 1.1710082006753497,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.1315,
+      "step": 4855
+    },
+    {
+      "epoch": 1.1722141823444283,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.991,
+      "step": 4860
+    },
+    {
+      "epoch": 1.173420164013507,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2845,
+      "step": 4865
+    },
+    {
+      "epoch": 1.1746261456825857,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0925,
+      "step": 4870
+    },
+    {
+      "epoch": 1.1758321273516643,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0806,
+      "step": 4875
+    },
+    {
+      "epoch": 1.177038109020743,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0887,
+      "step": 4880
+    },
+    {
+      "epoch": 1.1782440906898215,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0005,
+      "step": 4885
+    },
+    {
+      "epoch": 1.1794500723589,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0836,
+      "step": 4890
+    },
+    {
+      "epoch": 1.1806560540279787,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1957,
+      "step": 4895
+    },
+    {
+      "epoch": 1.1818620356970575,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0634,
+      "step": 4900
+    },
+    {
+      "epoch": 1.183068017366136,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1855,
+      "step": 4905
+    },
+    {
+      "epoch": 1.1842739990352147,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.0014,
+      "step": 4910
+    },
+    {
+      "epoch": 1.1854799807042933,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0946,
+      "step": 4915
+    },
+    {
+      "epoch": 1.1866859623733719,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9955,
+      "step": 4920
+    },
+    {
+      "epoch": 1.1878919440424505,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1028,
+      "step": 4925
+    },
+    {
+      "epoch": 1.1890979257115293,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1742,
+      "step": 4930
+    },
+    {
+      "epoch": 1.1903039073806079,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.1621,
+      "step": 4935
+    },
+    {
+      "epoch": 1.1915098890496865,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0406,
+      "step": 4940
+    },
+    {
+      "epoch": 1.192715870718765,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.2553,
+      "step": 4945
+    },
+    {
+      "epoch": 1.1939218523878437,
+      "grad_norm": 1.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1071,
+      "step": 4950
+    },
+    {
+      "epoch": 1.1951278340569222,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0448,
+      "step": 4955
+    },
+    {
+      "epoch": 1.196333815726001,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0193,
+      "step": 4960
+    },
+    {
+      "epoch": 1.1975397973950797,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0086,
+      "step": 4965
+    },
+    {
+      "epoch": 1.1987457790641582,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9806,
+      "step": 4970
+    },
+    {
+      "epoch": 1.1999517607332368,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1022,
+      "step": 4975
+    },
+    {
+      "epoch": 1.2011577424023154,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.075,
+      "step": 4980
+    },
+    {
+      "epoch": 1.202363724071394,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0788,
+      "step": 4985
+    },
+    {
+      "epoch": 1.2035697057404728,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.1779,
+      "step": 4990
+    },
+    {
+      "epoch": 1.2047756874095514,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2796,
+      "step": 4995
+    },
+    {
+      "epoch": 1.20598166907863,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0545,
+      "step": 5000
+    },
+    {
+      "epoch": 1.2071876507477086,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9616,
+      "step": 5005
+    },
+    {
+      "epoch": 1.2083936324167872,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.2295,
+      "step": 5010
+    },
+    {
+      "epoch": 1.2095996140858658,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0503,
+      "step": 5015
+    },
+    {
+      "epoch": 1.2108055957549446,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.1597,
+      "step": 5020
+    },
+    {
+      "epoch": 1.2120115774240232,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.0367,
+      "step": 5025
+    },
+    {
+      "epoch": 1.2132175590931018,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0732,
+      "step": 5030
+    },
+    {
+      "epoch": 1.2144235407621804,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0282,
+      "step": 5035
+    },
+    {
+      "epoch": 1.215629522431259,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.132,
+      "step": 5040
+    },
+    {
+      "epoch": 1.2168355041003376,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0851,
+      "step": 5045
+    },
+    {
+      "epoch": 1.2180414857694164,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0377,
+      "step": 5050
+    },
+    {
+      "epoch": 1.219247467438495,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0493,
+      "step": 5055
+    },
+    {
+      "epoch": 1.2204534491075736,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1075,
+      "step": 5060
+    },
+    {
+      "epoch": 1.2216594307766522,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0531,
+      "step": 5065
+    },
+    {
+      "epoch": 1.2228654124457308,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8709,
+      "step": 5070
+    },
+    {
+      "epoch": 1.2240713941148094,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1531,
+      "step": 5075
+    },
+    {
+      "epoch": 1.2252773757838882,
+      "grad_norm": 1.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9813,
+      "step": 5080
+    },
+    {
+      "epoch": 1.2264833574529668,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9545,
+      "step": 5085
+    },
+    {
+      "epoch": 1.2276893391220454,
+      "grad_norm": 1.8359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0415,
+      "step": 5090
+    },
+    {
+      "epoch": 1.228895320791124,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9904,
+      "step": 5095
+    },
+    {
+      "epoch": 1.2301013024602026,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9211,
+      "step": 5100
+    },
+    {
+      "epoch": 1.2313072841292811,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0547,
+      "step": 5105
+    },
+    {
+      "epoch": 1.23251326579836,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.0466,
+      "step": 5110
+    },
+    {
+      "epoch": 1.2337192474674386,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1817,
+      "step": 5115
+    },
+    {
+      "epoch": 1.2349252291365171,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.3072,
+      "step": 5120
+    },
+    {
+      "epoch": 1.2361312108055957,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0433,
+      "step": 5125
+    },
+    {
+      "epoch": 1.2373371924746743,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0285,
+      "step": 5130
+    },
+    {
+      "epoch": 1.238543174143753,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0347,
+      "step": 5135
+    },
+    {
+      "epoch": 1.2397491558128317,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1013,
+      "step": 5140
+    },
+    {
+      "epoch": 1.2409551374819103,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9632,
+      "step": 5145
+    },
+    {
+      "epoch": 1.242161119150989,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1824,
+      "step": 5150
+    },
+    {
+      "epoch": 1.2433671008200675,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1593,
+      "step": 5155
+    },
+    {
+      "epoch": 1.244573082489146,
+      "grad_norm": 1.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.01,
+      "step": 5160
+    },
+    {
+      "epoch": 1.2457790641582247,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.075,
+      "step": 5165
+    },
+    {
+      "epoch": 1.2469850458273035,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.1642,
+      "step": 5170
+    },
+    {
+      "epoch": 1.248191027496382,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.087,
+      "step": 5175
+    },
+    {
+      "epoch": 1.2493970091654607,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9084,
+      "step": 5180
+    },
+    {
+      "epoch": 1.2506029908345393,
+      "grad_norm": 1.8515625,
+      "learning_rate": 3e-05,
+      "loss": 4.1324,
+      "step": 5185
+    },
+    {
+      "epoch": 1.251808972503618,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9775,
+      "step": 5190
+    },
+    {
+      "epoch": 1.2530149541726967,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0716,
+      "step": 5195
+    },
+    {
+      "epoch": 1.2542209358417753,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8476,
+      "step": 5200
+    },
+    {
+      "epoch": 1.255426917510854,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1585,
+      "step": 5205
+    },
+    {
+      "epoch": 1.2566328991799325,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9763,
+      "step": 5210
+    },
+    {
+      "epoch": 1.257838880849011,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.0321,
+      "step": 5215
+    },
+    {
+      "epoch": 1.2590448625180897,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9563,
+      "step": 5220
+    },
+    {
+      "epoch": 1.2602508441871683,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0625,
+      "step": 5225
+    },
+    {
+      "epoch": 1.2614568258562469,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0721,
+      "step": 5230
+    },
+    {
+      "epoch": 1.2626628075253257,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9653,
+      "step": 5235
+    },
+    {
+      "epoch": 1.2638687891944043,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9962,
+      "step": 5240
+    },
+    {
+      "epoch": 1.2650747708634829,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0589,
+      "step": 5245
+    },
+    {
+      "epoch": 1.2662807525325614,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0012,
+      "step": 5250
+    },
+    {
+      "epoch": 1.2674867342016403,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.805,
+      "step": 5255
+    },
+    {
+      "epoch": 1.2686927158707189,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.1071,
+      "step": 5260
+    },
+    {
+      "epoch": 1.2698986975397974,
+      "grad_norm": 1.7734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9521,
+      "step": 5265
+    },
+    {
+      "epoch": 1.271104679208876,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.8906,
+      "step": 5270
+    },
+    {
+      "epoch": 1.2723106608779546,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.012,
+      "step": 5275
+    },
+    {
+      "epoch": 1.2735166425470332,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9555,
+      "step": 5280
+    },
+    {
+      "epoch": 1.2747226242161118,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8404,
+      "step": 5285
+    },
+    {
+      "epoch": 1.2759286058851904,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0432,
+      "step": 5290
+    },
+    {
+      "epoch": 1.2771345875542692,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9693,
+      "step": 5295
+    },
+    {
+      "epoch": 1.2783405692233478,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1444,
+      "step": 5300
+    },
+    {
+      "epoch": 1.2795465508924264,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.2652,
+      "step": 5305
+    },
+    {
+      "epoch": 1.280752532561505,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9923,
+      "step": 5310
+    },
+    {
+      "epoch": 1.2819585142305838,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9645,
+      "step": 5315
+    },
+    {
+      "epoch": 1.2831644958996624,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9688,
+      "step": 5320
+    },
+    {
+      "epoch": 1.284370477568741,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0604,
+      "step": 5325
+    },
+    {
+      "epoch": 1.2855764592378196,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9091,
+      "step": 5330
+    },
+    {
+      "epoch": 1.2867824409068982,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.1342,
+      "step": 5335
+    },
+    {
+      "epoch": 1.2879884225759768,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0989,
+      "step": 5340
+    },
+    {
+      "epoch": 1.2891944042450554,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 4.002,
+      "step": 5345
+    },
+    {
+      "epoch": 1.290400385914134,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0439,
+      "step": 5350
+    },
+    {
+      "epoch": 1.2916063675832128,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0985,
+      "step": 5355
+    },
+    {
+      "epoch": 1.2928123492522914,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0396,
+      "step": 5360
+    },
+    {
+      "epoch": 1.29401833092137,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8272,
+      "step": 5365
+    },
+    {
+      "epoch": 1.2952243125904486,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0941,
+      "step": 5370
+    },
+    {
+      "epoch": 1.2964302942595274,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9093,
+      "step": 5375
+    },
+    {
+      "epoch": 1.297636275928606,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0413,
+      "step": 5380
+    },
+    {
+      "epoch": 1.2988422575976846,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9375,
+      "step": 5385
+    },
+    {
+      "epoch": 1.3000482392667632,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0401,
+      "step": 5390
+    },
+    {
+      "epoch": 1.3012542209358418,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9941,
+      "step": 5395
+    },
+    {
+      "epoch": 1.3024602026049203,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1502,
+      "step": 5400
+    },
+    {
+      "epoch": 1.303666184273999,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9638,
+      "step": 5405
+    },
+    {
+      "epoch": 1.3048721659430775,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.8541,
+      "step": 5410
+    },
+    {
+      "epoch": 1.3060781476121563,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9946,
+      "step": 5415
+    },
+    {
+      "epoch": 1.307284129281235,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0078,
+      "step": 5420
+    },
+    {
+      "epoch": 1.3084901109503135,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9895,
+      "step": 5425
+    },
+    {
+      "epoch": 1.3096960926193921,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9607,
+      "step": 5430
+    },
+    {
+      "epoch": 1.310902074288471,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 3.8822,
+      "step": 5435
+    },
+    {
+      "epoch": 1.3121080559575495,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.8596,
+      "step": 5440
+    },
+    {
+      "epoch": 1.3133140376266281,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.1279,
+      "step": 5445
+    },
+    {
+      "epoch": 1.3145200192957067,
+      "grad_norm": 1.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.0723,
+      "step": 5450
+    },
+    {
+      "epoch": 1.3157260009647853,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0458,
+      "step": 5455
+    },
+    {
+      "epoch": 1.316931982633864,
+      "grad_norm": 1.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0843,
+      "step": 5460
+    },
+    {
+      "epoch": 1.3181379643029425,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.0233,
+      "step": 5465
+    },
+    {
+      "epoch": 1.319343945972021,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9558,
+      "step": 5470
+    },
+    {
+      "epoch": 1.3205499276411,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9242,
+      "step": 5475
+    },
+    {
+      "epoch": 1.3217559093101785,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8995,
+      "step": 5480
+    },
+    {
+      "epoch": 1.322961890979257,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9838,
+      "step": 5485
+    },
+    {
+      "epoch": 1.3241678726483357,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1444,
+      "step": 5490
+    },
+    {
+      "epoch": 1.3253738543174145,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9579,
+      "step": 5495
+    },
+    {
+      "epoch": 1.326579835986493,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1027,
+      "step": 5500
+    },
+    {
+      "epoch": 1.3277858176555717,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1139,
+      "step": 5505
+    },
+    {
+      "epoch": 1.3289917993246503,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0427,
+      "step": 5510
+    },
+    {
+      "epoch": 1.3301977809937289,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9941,
+      "step": 5515
+    },
+    {
+      "epoch": 1.3314037626628075,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0884,
+      "step": 5520
+    },
+    {
+      "epoch": 1.332609744331886,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9796,
+      "step": 5525
+    },
+    {
+      "epoch": 1.3338157260009647,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9593,
+      "step": 5530
+    },
+    {
+      "epoch": 1.3350217076700435,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9173,
+      "step": 5535
+    },
+    {
+      "epoch": 1.336227689339122,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.0316,
+      "step": 5540
+    },
+    {
+      "epoch": 1.3374336710082007,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0766,
+      "step": 5545
+    },
+    {
+      "epoch": 1.3386396526772792,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1248,
+      "step": 5550
+    },
+    {
+      "epoch": 1.339845634346358,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0593,
+      "step": 5555
+    },
+    {
+      "epoch": 1.3410516160154367,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0802,
+      "step": 5560
+    },
+    {
+      "epoch": 1.3422575976845152,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0298,
+      "step": 5565
+    },
+    {
+      "epoch": 1.3434635793535938,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9824,
+      "step": 5570
+    },
+    {
+      "epoch": 1.3446695610226724,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.993,
+      "step": 5575
+    },
+    {
+      "epoch": 1.345875542691751,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.1292,
+      "step": 5580
+    },
+    {
+      "epoch": 1.3470815243608296,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.1268,
+      "step": 5585
+    },
+    {
+      "epoch": 1.3482875060299082,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0361,
+      "step": 5590
+    },
+    {
+      "epoch": 1.349493487698987,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 4.057,
+      "step": 5595
+    },
+    {
+      "epoch": 1.3506994693680656,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.89,
+      "step": 5600
+    },
+    {
+      "epoch": 1.3519054510371442,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.929,
+      "step": 5605
+    },
+    {
+      "epoch": 1.3531114327062228,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1667,
+      "step": 5610
+    },
+    {
+      "epoch": 1.3543174143753016,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.9858,
+      "step": 5615
+    },
+    {
+      "epoch": 1.3555233960443802,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.1026,
+      "step": 5620
+    },
+    {
+      "epoch": 1.3567293777134588,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9049,
+      "step": 5625
+    },
+    {
+      "epoch": 1.3579353593825374,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9312,
+      "step": 5630
+    },
+    {
+      "epoch": 1.359141341051616,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0745,
+      "step": 5635
+    },
+    {
+      "epoch": 1.3603473227206946,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0108,
+      "step": 5640
+    },
+    {
+      "epoch": 1.3615533043897732,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9133,
+      "step": 5645
+    },
+    {
+      "epoch": 1.3627592860588518,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9778,
+      "step": 5650
+    },
+    {
+      "epoch": 1.3639652677279306,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0117,
+      "step": 5655
+    },
+    {
+      "epoch": 1.3651712493970092,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8992,
+      "step": 5660
+    },
+    {
+      "epoch": 1.3663772310660878,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8815,
+      "step": 5665
+    },
+    {
+      "epoch": 1.3675832127351664,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0667,
+      "step": 5670
+    },
+    {
+      "epoch": 1.3687891944042452,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1417,
+      "step": 5675
+    },
+    {
+      "epoch": 1.3699951760733238,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1762,
+      "step": 5680
+    },
+    {
+      "epoch": 1.3712011577424024,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9421,
+      "step": 5685
+    },
+    {
+      "epoch": 1.372407139411481,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9013,
+      "step": 5690
+    },
+    {
+      "epoch": 1.3736131210805596,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9713,
+      "step": 5695
+    },
+    {
+      "epoch": 1.3748191027496381,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9916,
+      "step": 5700
+    },
+    {
+      "epoch": 1.3760250844187167,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9628,
+      "step": 5705
+    },
+    {
+      "epoch": 1.3772310660877953,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8958,
+      "step": 5710
+    },
+    {
+      "epoch": 1.3784370477568741,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.8121,
+      "step": 5715
+    },
+    {
+      "epoch": 1.3796430294259527,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9822,
+      "step": 5720
+    },
+    {
+      "epoch": 1.3808490110950313,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1571,
+      "step": 5725
+    },
+    {
+      "epoch": 1.38205499276411,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.8781,
+      "step": 5730
+    },
+    {
+      "epoch": 1.3832609744331887,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8821,
+      "step": 5735
+    },
+    {
+      "epoch": 1.3844669561022673,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9592,
+      "step": 5740
+    },
+    {
+      "epoch": 1.385672937771346,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8345,
+      "step": 5745
+    },
+    {
+      "epoch": 1.3868789194404245,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0398,
+      "step": 5750
+    },
+    {
+      "epoch": 1.388084901109503,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1608,
+      "step": 5755
+    },
+    {
+      "epoch": 1.3892908827785817,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9649,
+      "step": 5760
+    },
+    {
+      "epoch": 1.3904968644476603,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9321,
+      "step": 5765
+    },
+    {
+      "epoch": 1.391702846116739,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9942,
+      "step": 5770
+    },
+    {
+      "epoch": 1.3929088277858177,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9593,
+      "step": 5775
+    },
+    {
+      "epoch": 1.3941148094548963,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9289,
+      "step": 5780
+    },
+    {
+      "epoch": 1.395320791123975,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9472,
+      "step": 5785
+    },
+    {
+      "epoch": 1.3965267727930535,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0542,
+      "step": 5790
+    },
+    {
+      "epoch": 1.3977327544621323,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9841,
+      "step": 5795
+    },
+    {
+      "epoch": 1.398938736131211,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.8916,
+      "step": 5800
+    },
+    {
+      "epoch": 1.4001447178002895,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0977,
+      "step": 5805
+    },
+    {
+      "epoch": 1.401350699469368,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0226,
+      "step": 5810
+    },
+    {
+      "epoch": 1.4025566811384467,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0059,
+      "step": 5815
+    },
+    {
+      "epoch": 1.4037626628075253,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1549,
+      "step": 5820
+    },
+    {
+      "epoch": 1.4049686444766039,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0359,
+      "step": 5825
+    },
+    {
+      "epoch": 1.4061746261456824,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1193,
+      "step": 5830
+    },
+    {
+      "epoch": 1.4073806078147613,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1605,
+      "step": 5835
+    },
+    {
+      "epoch": 1.4085865894838399,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8778,
+      "step": 5840
+    },
+    {
+      "epoch": 1.4097925711529185,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9953,
+      "step": 5845
+    },
+    {
+      "epoch": 1.410998552821997,
+      "grad_norm": 3.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0284,
+      "step": 5850
+    },
+    {
+      "epoch": 1.4122045344910759,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.8557,
+      "step": 5855
+    },
+    {
+      "epoch": 1.4134105161601545,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9324,
+      "step": 5860
+    },
+    {
+      "epoch": 1.414616497829233,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9125,
+      "step": 5865
+    },
+    {
+      "epoch": 1.4158224794983116,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.1176,
+      "step": 5870
+    },
+    {
+      "epoch": 1.4170284611673902,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.7876,
+      "step": 5875
+    },
+    {
+      "epoch": 1.4182344428364688,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.8916,
+      "step": 5880
+    },
+    {
+      "epoch": 1.4194404245055474,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0576,
+      "step": 5885
+    },
+    {
+      "epoch": 1.420646406174626,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.847,
+      "step": 5890
+    },
+    {
+      "epoch": 1.4218523878437048,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.9892,
+      "step": 5895
+    },
+    {
+      "epoch": 1.4230583695127834,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0812,
+      "step": 5900
+    },
+    {
+      "epoch": 1.424264351181862,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8944,
+      "step": 5905
+    },
+    {
+      "epoch": 1.4254703328509406,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9525,
+      "step": 5910
+    },
+    {
+      "epoch": 1.4266763145200194,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9512,
+      "step": 5915
+    },
+    {
+      "epoch": 1.427882296189098,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8585,
+      "step": 5920
+    },
+    {
+      "epoch": 1.4290882778581766,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0108,
+      "step": 5925
+    },
+    {
+      "epoch": 1.4302942595272552,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8406,
+      "step": 5930
+    },
+    {
+      "epoch": 1.4315002411963338,
+      "grad_norm": 1.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9009,
+      "step": 5935
+    },
+    {
+      "epoch": 1.4327062228654124,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0334,
+      "step": 5940
+    },
+    {
+      "epoch": 1.433912204534491,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0876,
+      "step": 5945
+    },
+    {
+      "epoch": 1.4351181862035696,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9512,
+      "step": 5950
+    },
+    {
+      "epoch": 1.4363241678726484,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3e-05,
+      "loss": 3.8528,
+      "step": 5955
+    },
+    {
+      "epoch": 1.437530149541727,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9492,
+      "step": 5960
+    },
+    {
+      "epoch": 1.4387361312108056,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9196,
+      "step": 5965
+    },
+    {
+      "epoch": 1.4399421128798842,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0004,
+      "step": 5970
+    },
+    {
+      "epoch": 1.441148094548963,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9415,
+      "step": 5975
+    },
+    {
+      "epoch": 1.4423540762180416,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9929,
+      "step": 5980
+    },
+    {
+      "epoch": 1.4435600578871202,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.039,
+      "step": 5985
+    },
+    {
+      "epoch": 1.4447660395561988,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9964,
+      "step": 5990
+    },
+    {
+      "epoch": 1.4459720212252773,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.8487,
+      "step": 5995
+    },
+    {
+      "epoch": 1.447178002894356,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0825,
+      "step": 6000
+    },
+    {
+      "epoch": 1.4483839845634345,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0041,
+      "step": 6005
+    },
+    {
+      "epoch": 1.4495899662325131,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8463,
+      "step": 6010
+    },
+    {
+      "epoch": 1.450795947901592,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0803,
+      "step": 6015
+    },
+    {
+      "epoch": 1.4520019295706705,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9207,
+      "step": 6020
+    },
+    {
+      "epoch": 1.4532079112397491,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9713,
+      "step": 6025
+    },
+    {
+      "epoch": 1.4544138929088277,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.945,
+      "step": 6030
+    },
+    {
+      "epoch": 1.4556198745779065,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8737,
+      "step": 6035
+    },
+    {
+      "epoch": 1.4568258562469851,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.964,
+      "step": 6040
+    },
+    {
+      "epoch": 1.4580318379160637,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8763,
+      "step": 6045
+    },
+    {
+      "epoch": 1.4592378195851423,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9784,
+      "step": 6050
+    },
+    {
+      "epoch": 1.460443801254221,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8817,
+      "step": 6055
+    },
+    {
+      "epoch": 1.4616497829232995,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1275,
+      "step": 6060
+    },
+    {
+      "epoch": 1.462855764592378,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0263,
+      "step": 6065
+    },
+    {
+      "epoch": 1.464061746261457,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.949,
+      "step": 6070
+    },
+    {
+      "epoch": 1.4652677279305355,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9922,
+      "step": 6075
+    },
+    {
+      "epoch": 1.466473709599614,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0292,
+      "step": 6080
+    },
+    {
+      "epoch": 1.4676796912686927,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0366,
+      "step": 6085
+    },
+    {
+      "epoch": 1.4688856729377713,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1351,
+      "step": 6090
+    },
+    {
+      "epoch": 1.47009165460685,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9434,
+      "step": 6095
+    },
+    {
+      "epoch": 1.4712976362759287,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0487,
+      "step": 6100
+    },
+    {
+      "epoch": 1.4725036179450073,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9642,
+      "step": 6105
+    },
+    {
+      "epoch": 1.4737095996140859,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.1484,
+      "step": 6110
+    },
+    {
+      "epoch": 1.4749155812831645,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0544,
+      "step": 6115
+    },
+    {
+      "epoch": 1.476121562952243,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0747,
+      "step": 6120
+    },
+    {
+      "epoch": 1.4773275446213217,
+      "grad_norm": 1.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9309,
+      "step": 6125
+    },
+    {
+      "epoch": 1.4785335262904005,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0152,
+      "step": 6130
+    },
+    {
+      "epoch": 1.479739507959479,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0743,
+      "step": 6135
+    },
+    {
+      "epoch": 1.4809454896285577,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9126,
+      "step": 6140
+    },
+    {
+      "epoch": 1.4821514712976362,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9772,
+      "step": 6145
+    },
+    {
+      "epoch": 1.4833574529667148,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8644,
+      "step": 6150
+    },
+    {
+      "epoch": 1.4845634346357937,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8759,
+      "step": 6155
+    },
+    {
+      "epoch": 1.4857694163048722,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8819,
+      "step": 6160
+    },
+    {
+      "epoch": 1.4869753979739508,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9676,
+      "step": 6165
+    },
+    {
+      "epoch": 1.4881813796430294,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9589,
+      "step": 6170
+    },
+    {
+      "epoch": 1.489387361312108,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9092,
+      "step": 6175
+    },
+    {
+      "epoch": 1.4905933429811866,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9514,
+      "step": 6180
+    },
+    {
+      "epoch": 1.4917993246502652,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0319,
+      "step": 6185
+    },
+    {
+      "epoch": 1.493005306319344,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8165,
+      "step": 6190
+    },
+    {
+      "epoch": 1.4942112879884226,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0019,
+      "step": 6195
+    },
+    {
+      "epoch": 1.4954172696575012,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8178,
+      "step": 6200
+    },
+    {
+      "epoch": 1.4966232513265798,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.878,
+      "step": 6205
+    },
+    {
+      "epoch": 1.4978292329956584,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9808,
+      "step": 6210
+    },
+    {
+      "epoch": 1.4990352146647372,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9069,
+      "step": 6215
+    },
+    {
+      "epoch": 1.5002411963338158,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9508,
+      "step": 6220
+    },
+    {
+      "epoch": 1.5014471780028944,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.7668,
+      "step": 6225
+    },
+    {
+      "epoch": 1.502653159671973,
+      "grad_norm": 1.7890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9336,
+      "step": 6230
+    },
+    {
+      "epoch": 1.5038591413410516,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8866,
+      "step": 6235
+    },
+    {
+      "epoch": 1.5050651230101302,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0279,
+      "step": 6240
+    },
+    {
+      "epoch": 1.5062711046792088,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0068,
+      "step": 6245
+    },
+    {
+      "epoch": 1.5074770863482874,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9693,
+      "step": 6250
+    },
+    {
+      "epoch": 1.5086830680173662,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9907,
+      "step": 6255
+    },
+    {
+      "epoch": 1.5098890496864448,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0405,
+      "step": 6260
+    },
+    {
+      "epoch": 1.5110950313555234,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9773,
+      "step": 6265
+    },
+    {
+      "epoch": 1.5123010130246022,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9862,
+      "step": 6270
+    },
+    {
+      "epoch": 1.5135069946936808,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.1404,
+      "step": 6275
+    },
+    {
+      "epoch": 1.5147129763627594,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8186,
+      "step": 6280
+    },
+    {
+      "epoch": 1.515918958031838,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0414,
+      "step": 6285
+    },
+    {
+      "epoch": 1.5171249397009166,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9972,
+      "step": 6290
+    },
+    {
+      "epoch": 1.5183309213699951,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8544,
+      "step": 6295
+    },
+    {
+      "epoch": 1.5195369030390737,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.9612,
+      "step": 6300
+    },
+    {
+      "epoch": 1.5207428847081523,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8324,
+      "step": 6305
+    },
+    {
+      "epoch": 1.521948866377231,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.7584,
+      "step": 6310
+    },
+    {
+      "epoch": 1.5231548480463097,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9257,
+      "step": 6315
+    },
+    {
+      "epoch": 1.5243608297153883,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9332,
+      "step": 6320
+    },
+    {
+      "epoch": 1.525566811384467,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8832,
+      "step": 6325
+    },
+    {
+      "epoch": 1.5267727930535457,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8925,
+      "step": 6330
+    },
+    {
+      "epoch": 1.5279787747226243,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8892,
+      "step": 6335
+    },
+    {
+      "epoch": 1.529184756391703,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9855,
+      "step": 6340
+    },
+    {
+      "epoch": 1.5303907380607815,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8522,
+      "step": 6345
+    },
+    {
+      "epoch": 1.5315967197298601,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9123,
+      "step": 6350
+    },
+    {
+      "epoch": 1.5328027013989387,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.7973,
+      "step": 6355
+    },
+    {
+      "epoch": 1.5340086830680173,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9352,
+      "step": 6360
+    },
+    {
+      "epoch": 1.535214664737096,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8727,
+      "step": 6365
+    },
+    {
+      "epoch": 1.5364206464061745,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.917,
+      "step": 6370
+    },
+    {
+      "epoch": 1.5376266280752533,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.8183,
+      "step": 6375
+    },
+    {
+      "epoch": 1.538832609744332,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.8579,
+      "step": 6380
+    },
+    {
+      "epoch": 1.5400385914134105,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.8495,
+      "step": 6385
+    },
+    {
+      "epoch": 1.5412445730824893,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9046,
+      "step": 6390
+    },
+    {
+      "epoch": 1.542450554751568,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.969,
+      "step": 6395
+    },
+    {
+      "epoch": 1.5436565364206465,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.8441,
+      "step": 6400
+    },
+    {
+      "epoch": 1.544862518089725,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9219,
+      "step": 6405
+    },
+    {
+      "epoch": 1.5460684997588037,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.904,
+      "step": 6410
+    },
+    {
+      "epoch": 1.5472744814278823,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8468,
+      "step": 6415
+    },
+    {
+      "epoch": 1.5484804630969609,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0883,
+      "step": 6420
+    },
+    {
+      "epoch": 1.5496864447660395,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.759,
+      "step": 6425
+    },
+    {
+      "epoch": 1.550892426435118,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9091,
+      "step": 6430
+    },
+    {
+      "epoch": 1.5520984081041969,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0657,
+      "step": 6435
+    },
+    {
+      "epoch": 1.5533043897732755,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.8907,
+      "step": 6440
+    },
+    {
+      "epoch": 1.554510371442354,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0508,
+      "step": 6445
+    },
+    {
+      "epoch": 1.5557163531114329,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.976,
+      "step": 6450
+    },
+    {
+      "epoch": 1.5569223347805115,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.831,
+      "step": 6455
+    },
+    {
+      "epoch": 1.55812831644959,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.807,
+      "step": 6460
+    },
+    {
+      "epoch": 1.5593342981186686,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.012,
+      "step": 6465
+    },
+    {
+      "epoch": 1.5605402797877472,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9841,
+      "step": 6470
+    },
+    {
+      "epoch": 1.5617462614568258,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.8583,
+      "step": 6475
+    },
+    {
+      "epoch": 1.5629522431259044,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.802,
+      "step": 6480
+    },
+    {
+      "epoch": 1.564158224794983,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8552,
+      "step": 6485
+    },
+    {
+      "epoch": 1.5653642064640616,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7725,
+      "step": 6490
+    },
+    {
+      "epoch": 1.5665701881331404,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.944,
+      "step": 6495
+    },
+    {
+      "epoch": 1.567776169802219,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9519,
+      "step": 6500
+    },
+    {
+      "epoch": 1.5689821514712976,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.1496,
+      "step": 6505
+    },
+    {
+      "epoch": 1.5701881331403764,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.6896,
+      "step": 6510
+    },
+    {
+      "epoch": 1.571394114809455,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.054,
+      "step": 6515
+    },
+    {
+      "epoch": 1.5726000964785336,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0407,
+      "step": 6520
+    },
+    {
+      "epoch": 1.5738060781476122,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.7924,
+      "step": 6525
+    },
+    {
+      "epoch": 1.5750120598166908,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8803,
+      "step": 6530
+    },
+    {
+      "epoch": 1.5762180414857694,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9293,
+      "step": 6535
+    },
+    {
+      "epoch": 1.577424023154848,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9784,
+      "step": 6540
+    },
+    {
+      "epoch": 1.5786300048239266,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.957,
+      "step": 6545
+    },
+    {
+      "epoch": 1.5798359864930052,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9482,
+      "step": 6550
+    },
+    {
+      "epoch": 1.581041968162084,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8502,
+      "step": 6555
+    },
+    {
+      "epoch": 1.5822479498311626,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9748,
+      "step": 6560
+    },
+    {
+      "epoch": 1.5834539315002412,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9425,
+      "step": 6565
+    },
+    {
+      "epoch": 1.58465991316932,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8511,
+      "step": 6570
+    },
+    {
+      "epoch": 1.5858658948383986,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9477,
+      "step": 6575
+    },
+    {
+      "epoch": 1.5870718765074772,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8881,
+      "step": 6580
+    },
+    {
+      "epoch": 1.5882778581765558,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9599,
+      "step": 6585
+    },
+    {
+      "epoch": 1.5894838398456343,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9477,
+      "step": 6590
+    },
+    {
+      "epoch": 1.590689821514713,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9076,
+      "step": 6595
+    },
+    {
+      "epoch": 1.5918958031837915,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9423,
+      "step": 6600
+    },
+    {
+      "epoch": 1.5931017848528701,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0652,
+      "step": 6605
+    },
+    {
+      "epoch": 1.5943077665219487,
+      "grad_norm": 3.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.7895,
+      "step": 6610
+    },
+    {
+      "epoch": 1.5955137481910275,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.8779,
+      "step": 6615
+    },
+    {
+      "epoch": 1.5967197298601061,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8829,
+      "step": 6620
+    },
+    {
+      "epoch": 1.5979257115291847,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.7635,
+      "step": 6625
+    },
+    {
+      "epoch": 1.5991316931982635,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9663,
+      "step": 6630
+    },
+    {
+      "epoch": 1.6003376748673421,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8111,
+      "step": 6635
+    },
+    {
+      "epoch": 1.6015436565364207,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9601,
+      "step": 6640
+    },
+    {
+      "epoch": 1.6027496382054993,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9679,
+      "step": 6645
+    },
+    {
+      "epoch": 1.603955619874578,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9529,
+      "step": 6650
+    },
+    {
+      "epoch": 1.6051616015436565,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0332,
+      "step": 6655
+    },
+    {
+      "epoch": 1.606367583212735,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.8873,
+      "step": 6660
+    },
+    {
+      "epoch": 1.6075735648818137,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8165,
+      "step": 6665
+    },
+    {
+      "epoch": 1.6087795465508923,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9351,
+      "step": 6670
+    },
+    {
+      "epoch": 1.609985528219971,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0391,
+      "step": 6675
+    },
+    {
+      "epoch": 1.6111915098890497,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9085,
+      "step": 6680
+    },
+    {
+      "epoch": 1.6123974915581283,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0265,
+      "step": 6685
+    },
+    {
+      "epoch": 1.613603473227207,
+      "grad_norm": 3.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0292,
+      "step": 6690
+    },
+    {
+      "epoch": 1.6148094548962857,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0291,
+      "step": 6695
+    },
+    {
+      "epoch": 1.6160154365653643,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9243,
+      "step": 6700
+    },
+    {
+      "epoch": 1.6172214182344429,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.7057,
+      "step": 6705
+    },
+    {
+      "epoch": 1.6184273999035215,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.8733,
+      "step": 6710
+    },
+    {
+      "epoch": 1.6196333815726,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0025,
+      "step": 6715
+    },
+    {
+      "epoch": 1.6208393632416787,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8148,
+      "step": 6720
+    },
+    {
+      "epoch": 1.6220453449107572,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9253,
+      "step": 6725
+    },
+    {
+      "epoch": 1.6232513265798358,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9053,
+      "step": 6730
+    },
+    {
+      "epoch": 1.6244573082489147,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.7442,
+      "step": 6735
+    },
+    {
+      "epoch": 1.6256632899179932,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9988,
+      "step": 6740
+    },
+    {
+      "epoch": 1.6268692715870718,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8321,
+      "step": 6745
+    },
+    {
+      "epoch": 1.6280752532561507,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8806,
+      "step": 6750
+    },
+    {
+      "epoch": 1.6292812349252292,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0618,
+      "step": 6755
+    },
+    {
+      "epoch": 1.6304872165943078,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0236,
+      "step": 6760
+    },
+    {
+      "epoch": 1.6316931982633864,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.894,
+      "step": 6765
+    },
+    {
+      "epoch": 1.632899179932465,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9581,
+      "step": 6770
+    },
+    {
+      "epoch": 1.6341051616015436,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.7907,
+      "step": 6775
+    },
+    {
+      "epoch": 1.6353111432706222,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.861,
+      "step": 6780
+    },
+    {
+      "epoch": 1.6365171249397008,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.8668,
+      "step": 6785
+    },
+    {
+      "epoch": 1.6377231066087794,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8156,
+      "step": 6790
+    },
+    {
+      "epoch": 1.6389290882778582,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8422,
+      "step": 6795
+    },
+    {
+      "epoch": 1.6401350699469368,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0078,
+      "step": 6800
+    },
+    {
+      "epoch": 1.6413410516160154,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9526,
+      "step": 6805
+    },
+    {
+      "epoch": 1.6425470332850942,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0902,
+      "step": 6810
+    },
+    {
+      "epoch": 1.6437530149541728,
+      "grad_norm": 1.9765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9587,
+      "step": 6815
+    },
+    {
+      "epoch": 1.6449589966232514,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.916,
+      "step": 6820
+    },
+    {
+      "epoch": 1.64616497829233,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.7942,
+      "step": 6825
+    },
+    {
+      "epoch": 1.6473709599614086,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.7968,
+      "step": 6830
+    },
+    {
+      "epoch": 1.6485769416304872,
+      "grad_norm": 1.9140625,
+      "learning_rate": 3e-05,
+      "loss": 3.8271,
+      "step": 6835
+    },
+    {
+      "epoch": 1.6497829232995658,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.7452,
+      "step": 6840
+    },
+    {
+      "epoch": 1.6509889049686444,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.788,
+      "step": 6845
+    },
+    {
+      "epoch": 1.652194886637723,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9527,
+      "step": 6850
+    },
+    {
+      "epoch": 1.6534008683068018,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9491,
+      "step": 6855
+    },
+    {
+      "epoch": 1.6546068499758804,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.7877,
+      "step": 6860
+    },
+    {
+      "epoch": 1.655812831644959,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0136,
+      "step": 6865
+    },
+    {
+      "epoch": 1.6570188133140378,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.8773,
+      "step": 6870
+    },
+    {
+      "epoch": 1.6582247949831164,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9202,
+      "step": 6875
+    },
+    {
+      "epoch": 1.659430776652195,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.061,
+      "step": 6880
+    },
+    {
+      "epoch": 1.6606367583212736,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0245,
+      "step": 6885
+    },
+    {
+      "epoch": 1.6618427399903521,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8493,
+      "step": 6890
+    },
+    {
+      "epoch": 1.6630487216594307,
+      "grad_norm": 1.8828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0109,
+      "step": 6895
+    },
+    {
+      "epoch": 1.6642547033285093,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9281,
+      "step": 6900
+    },
+    {
+      "epoch": 1.665460684997588,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8341,
+      "step": 6905
+    },
+    {
+      "epoch": 1.6666666666666665,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9277,
+      "step": 6910
+    },
+    {
+      "epoch": 1.6678726483357453,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7938,
+      "step": 6915
+    },
+    {
+      "epoch": 1.669078630004824,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0384,
+      "step": 6920
+    },
+    {
+      "epoch": 1.6702846116739025,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0376,
+      "step": 6925
+    },
+    {
+      "epoch": 1.6714905933429813,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9776,
+      "step": 6930
+    },
+    {
+      "epoch": 1.67269657501206,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.834,
+      "step": 6935
+    },
+    {
+      "epoch": 1.6739025566811385,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8761,
+      "step": 6940
+    },
+    {
+      "epoch": 1.6751085383502171,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8635,
+      "step": 6945
+    },
+    {
+      "epoch": 1.6763145200192957,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.8935,
+      "step": 6950
+    },
+    {
+      "epoch": 1.6775205016883743,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7317,
+      "step": 6955
+    },
+    {
+      "epoch": 1.678726483357453,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8441,
+      "step": 6960
+    },
+    {
+      "epoch": 1.6799324650265315,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9865,
+      "step": 6965
+    },
+    {
+      "epoch": 1.68113844669561,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.818,
+      "step": 6970
+    },
+    {
+      "epoch": 1.682344428364689,
+      "grad_norm": 3.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9333,
+      "step": 6975
+    },
+    {
+      "epoch": 1.6835504100337675,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.7855,
+      "step": 6980
+    },
+    {
+      "epoch": 1.684756391702846,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.8617,
+      "step": 6985
+    },
+    {
+      "epoch": 1.685962373371925,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9757,
+      "step": 6990
+    },
+    {
+      "epoch": 1.6871683550410035,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.7579,
+      "step": 6995
+    },
+    {
+      "epoch": 1.688374336710082,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8122,
+      "step": 7000
+    },
+    {
+      "epoch": 1.6895803183791607,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8582,
+      "step": 7005
+    },
+    {
+      "epoch": 1.6907863000482393,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.8424,
+      "step": 7010
+    },
+    {
+      "epoch": 1.6919922817173179,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.7794,
+      "step": 7015
+    },
+    {
+      "epoch": 1.6931982633863965,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.883,
+      "step": 7020
+    },
+    {
+      "epoch": 1.694404245055475,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.7467,
+      "step": 7025
+    },
+    {
+      "epoch": 1.6956102267245536,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9066,
+      "step": 7030
+    },
+    {
+      "epoch": 1.6968162083936325,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.7149,
+      "step": 7035
+    },
+    {
+      "epoch": 1.698022190062711,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9981,
+      "step": 7040
+    },
+    {
+      "epoch": 1.6992281717317896,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.885,
+      "step": 7045
+    },
+    {
+      "epoch": 1.7004341534008685,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8583,
+      "step": 7050
+    },
+    {
+      "epoch": 1.701640135069947,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.798,
+      "step": 7055
+    },
+    {
+      "epoch": 1.7028461167390256,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8205,
+      "step": 7060
+    },
+    {
+      "epoch": 1.7040520984081042,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7779,
+      "step": 7065
+    },
+    {
+      "epoch": 1.7052580800771828,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9621,
+      "step": 7070
+    },
+    {
+      "epoch": 1.7064640617462614,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.7453,
+      "step": 7075
+    },
+    {
+      "epoch": 1.70767004341534,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9981,
+      "step": 7080
+    },
+    {
+      "epoch": 1.7088760250844186,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8412,
+      "step": 7085
+    },
+    {
+      "epoch": 1.7100820067534972,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.7884,
+      "step": 7090
+    },
+    {
+      "epoch": 1.711287988422576,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.787,
+      "step": 7095
+    },
+    {
+      "epoch": 1.7124939700916546,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.7944,
+      "step": 7100
+    },
+    {
+      "epoch": 1.7136999517607332,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.8066,
+      "step": 7105
+    },
+    {
+      "epoch": 1.714905933429812,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9211,
+      "step": 7110
+    },
+    {
+      "epoch": 1.7161119150988906,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8854,
+      "step": 7115
+    },
+    {
+      "epoch": 1.7173178967679692,
+      "grad_norm": 1.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.81,
+      "step": 7120
+    },
+    {
+      "epoch": 1.7185238784370478,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9024,
+      "step": 7125
+    },
+    {
+      "epoch": 1.7197298601061264,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.0509,
+      "step": 7130
+    },
+    {
+      "epoch": 1.720935841775205,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.7665,
+      "step": 7135
+    },
+    {
+      "epoch": 1.7221418234442836,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0687,
+      "step": 7140
+    },
+    {
+      "epoch": 1.7233478051133622,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.7784,
+      "step": 7145
+    },
+    {
+      "epoch": 1.7245537867824408,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.8031,
+      "step": 7150
+    },
+    {
+      "epoch": 1.7257597684515196,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8097,
+      "step": 7155
+    },
+    {
+      "epoch": 1.7269657501205982,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8283,
+      "step": 7160
+    },
+    {
+      "epoch": 1.7281717317896768,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7616,
+      "step": 7165
+    },
+    {
+      "epoch": 1.7293777134587556,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8086,
+      "step": 7170
+    },
+    {
+      "epoch": 1.7305836951278342,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8023,
+      "step": 7175
+    },
+    {
+      "epoch": 1.7317896767969128,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.939,
+      "step": 7180
+    },
+    {
+      "epoch": 1.7329956584659914,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.86,
+      "step": 7185
+    },
+    {
+      "epoch": 1.73420164013507,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7831,
+      "step": 7190
+    },
+    {
+      "epoch": 1.7354076218041485,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8882,
+      "step": 7195
+    },
+    {
+      "epoch": 1.7366136034732271,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8688,
+      "step": 7200
+    },
+    {
+      "epoch": 1.7378195851423057,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8119,
+      "step": 7205
+    },
+    {
+      "epoch": 1.7390255668113843,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.8169,
+      "step": 7210
+    },
+    {
+      "epoch": 1.7402315484804631,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8267,
+      "step": 7215
+    },
+    {
+      "epoch": 1.7414375301495417,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8566,
+      "step": 7220
+    },
+    {
+      "epoch": 1.7426435118186203,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.947,
+      "step": 7225
+    },
+    {
+      "epoch": 1.7438494934876991,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9169,
+      "step": 7230
+    },
+    {
+      "epoch": 1.7450554751567777,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.944,
+      "step": 7235
+    },
+    {
+      "epoch": 1.7462614568258563,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7584,
+      "step": 7240
+    },
+    {
+      "epoch": 1.747467438494935,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.7844,
+      "step": 7245
+    },
+    {
+      "epoch": 1.7486734201640135,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.5996,
+      "step": 7250
+    },
+    {
+      "epoch": 1.749879401833092,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.7866,
+      "step": 7255
+    },
+    {
+      "epoch": 1.7510853835021707,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1632,
+      "step": 7260
+    },
+    {
+      "epoch": 1.7522913651712493,
+      "grad_norm": 3.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9928,
+      "step": 7265
+    },
+    {
+      "epoch": 1.7534973468403279,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9223,
+      "step": 7270
+    },
+    {
+      "epoch": 1.7547033285094067,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8657,
+      "step": 7275
+    },
+    {
+      "epoch": 1.7559093101784853,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9718,
+      "step": 7280
+    },
+    {
+      "epoch": 1.7571152918475639,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.7933,
+      "step": 7285
+    },
+    {
+      "epoch": 1.7583212735166427,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0573,
+      "step": 7290
+    },
+    {
+      "epoch": 1.7595272551857213,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.8796,
+      "step": 7295
+    },
+    {
+      "epoch": 1.7607332368547999,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.7767,
+      "step": 7300
+    },
+    {
+      "epoch": 1.7619392185238785,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8606,
+      "step": 7305
+    },
+    {
+      "epoch": 1.763145200192957,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8195,
+      "step": 7310
+    },
+    {
+      "epoch": 1.7643511818620357,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7947,
+      "step": 7315
+    },
+    {
+      "epoch": 1.7655571635311142,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.7401,
+      "step": 7320
+    },
+    {
+      "epoch": 1.7667631452001928,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.7819,
+      "step": 7325
+    },
+    {
+      "epoch": 1.7679691268692714,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0019,
+      "step": 7330
+    },
+    {
+      "epoch": 1.7691751085383502,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8072,
+      "step": 7335
+    },
+    {
+      "epoch": 1.7703810902074288,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8622,
+      "step": 7340
+    },
+    {
+      "epoch": 1.7715870718765074,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 3.7629,
+      "step": 7345
+    },
+    {
+      "epoch": 1.7727930535455863,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 3.7102,
+      "step": 7350
+    },
+    {
+      "epoch": 1.7739990352146648,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.541,
+      "step": 7355
+    },
+    {
+      "epoch": 1.7752050168837434,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0336,
+      "step": 7360
+    },
+    {
+      "epoch": 1.776410998552822,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0071,
+      "step": 7365
+    },
+    {
+      "epoch": 1.7776169802219006,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9898,
+      "step": 7370
+    },
+    {
+      "epoch": 1.7788229618909792,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9848,
+      "step": 7375
+    },
+    {
+      "epoch": 1.7800289435600578,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.6977,
+      "step": 7380
+    },
+    {
+      "epoch": 1.7812349252291364,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9025,
+      "step": 7385
+    },
+    {
+      "epoch": 1.782440906898215,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 3.7842,
+      "step": 7390
+    },
+    {
+      "epoch": 1.7836468885672938,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9929,
+      "step": 7395
+    },
+    {
+      "epoch": 1.7848528702363724,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8596,
+      "step": 7400
+    },
+    {
+      "epoch": 1.786058851905451,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8727,
+      "step": 7405
+    },
+    {
+      "epoch": 1.7872648335745298,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9179,
+      "step": 7410
+    },
+    {
+      "epoch": 1.7884708152436084,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1607,
+      "step": 7415
+    },
+    {
+      "epoch": 1.789676796912687,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.7633,
+      "step": 7420
+    },
+    {
+      "epoch": 1.7908827785817656,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.7805,
+      "step": 7425
+    },
+    {
+      "epoch": 1.7920887602508442,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9107,
+      "step": 7430
+    },
+    {
+      "epoch": 1.7932947419199228,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9827,
+      "step": 7435
+    },
+    {
+      "epoch": 1.7945007235890014,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8112,
+      "step": 7440
+    },
+    {
+      "epoch": 1.79570670525808,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9435,
+      "step": 7445
+    },
+    {
+      "epoch": 1.7969126869271586,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.821,
+      "step": 7450
+    },
+    {
+      "epoch": 1.7981186685962374,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.7391,
+      "step": 7455
+    },
+    {
+      "epoch": 1.799324650265316,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0559,
+      "step": 7460
+    },
+    {
+      "epoch": 1.8005306319343946,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8256,
+      "step": 7465
+    },
+    {
+      "epoch": 1.8017366136034734,
+      "grad_norm": 1.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.7895,
+      "step": 7470
+    },
+    {
+      "epoch": 1.802942595272552,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.8965,
+      "step": 7475
+    },
+    {
+      "epoch": 1.8041485769416306,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9159,
+      "step": 7480
+    },
+    {
+      "epoch": 1.8053545586107091,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8632,
+      "step": 7485
+    },
+    {
+      "epoch": 1.8065605402797877,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9129,
+      "step": 7490
+    },
+    {
+      "epoch": 1.8077665219488663,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9412,
+      "step": 7495
+    },
+    {
+      "epoch": 1.808972503617945,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8699,
+      "step": 7500
+    },
+    {
+      "epoch": 1.8101784852870235,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.779,
+      "step": 7505
+    },
+    {
+      "epoch": 1.8113844669561021,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8178,
+      "step": 7510
+    },
+    {
+      "epoch": 1.812590448625181,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.8654,
+      "step": 7515
+    },
+    {
+      "epoch": 1.8137964302942595,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.7555,
+      "step": 7520
+    },
+    {
+      "epoch": 1.8150024119633381,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8698,
+      "step": 7525
+    },
+    {
+      "epoch": 1.816208393632417,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9038,
+      "step": 7530
+    },
+    {
+      "epoch": 1.8174143753014955,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8027,
+      "step": 7535
+    },
+    {
+      "epoch": 1.8186203569705741,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9413,
+      "step": 7540
+    },
+    {
+      "epoch": 1.8198263386396527,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.767,
+      "step": 7545
+    },
+    {
+      "epoch": 1.8210323203087313,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.7247,
+      "step": 7550
+    },
+    {
+      "epoch": 1.82223830197781,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.7961,
+      "step": 7555
+    },
+    {
+      "epoch": 1.8234442836468885,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.7912,
+      "step": 7560
+    },
+    {
+      "epoch": 1.824650265315967,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.7878,
+      "step": 7565
+    },
+    {
+      "epoch": 1.8258562469850457,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.7986,
+      "step": 7570
+    },
+    {
+      "epoch": 1.8270622286541245,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9248,
+      "step": 7575
+    },
+    {
+      "epoch": 1.828268210323203,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8577,
+      "step": 7580
+    },
+    {
+      "epoch": 1.8294741919922817,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9105,
+      "step": 7585
+    },
+    {
+      "epoch": 1.8306801736613605,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.7991,
+      "step": 7590
+    },
+    {
+      "epoch": 1.831886155330439,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8738,
+      "step": 7595
+    },
+    {
+      "epoch": 1.8330921369995177,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.7792,
+      "step": 7600
+    },
+    {
+      "epoch": 1.8342981186685963,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.9745,
+      "step": 7605
+    },
+    {
+      "epoch": 1.8355041003376749,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.5487,
+      "step": 7610
+    },
+    {
+      "epoch": 1.8367100820067535,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8766,
+      "step": 7615
+    },
+    {
+      "epoch": 1.837916063675832,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9019,
+      "step": 7620
+    },
+    {
+      "epoch": 1.8391220453449106,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8466,
+      "step": 7625
+    },
+    {
+      "epoch": 1.8403280270139892,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9464,
+      "step": 7630
+    },
+    {
+      "epoch": 1.841534008683068,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.7597,
+      "step": 7635
+    },
+    {
+      "epoch": 1.8427399903521466,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.7028,
+      "step": 7640
+    },
+    {
+      "epoch": 1.8439459720212252,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8406,
+      "step": 7645
+    },
+    {
+      "epoch": 1.845151953690304,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.6806,
+      "step": 7650
+    },
+    {
+      "epoch": 1.8463579353593826,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.739,
+      "step": 7655
+    },
+    {
+      "epoch": 1.8475639170284612,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9497,
+      "step": 7660
+    },
+    {
+      "epoch": 1.8487698986975398,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8381,
+      "step": 7665
+    },
+    {
+      "epoch": 1.8499758803666184,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9928,
+      "step": 7670
+    },
+    {
+      "epoch": 1.851181862035697,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9728,
+      "step": 7675
+    },
+    {
+      "epoch": 1.8523878437047756,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8359,
+      "step": 7680
+    },
+    {
+      "epoch": 1.8535938253738542,
+      "grad_norm": 4.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.7726,
+      "step": 7685
+    },
+    {
+      "epoch": 1.8547998070429328,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.7752,
+      "step": 7690
+    },
+    {
+      "epoch": 1.8560057887120116,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.6874,
+      "step": 7695
+    },
+    {
+      "epoch": 1.8572117703810902,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8096,
+      "step": 7700
+    },
+    {
+      "epoch": 1.8584177520501688,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8184,
+      "step": 7705
+    },
+    {
+      "epoch": 1.8596237337192476,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.7646,
+      "step": 7710
+    },
+    {
+      "epoch": 1.8608297153883262,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.7085,
+      "step": 7715
+    },
+    {
+      "epoch": 1.8620356970574048,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.7508,
+      "step": 7720
+    },
+    {
+      "epoch": 1.8632416787264834,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0698,
+      "step": 7725
+    },
+    {
+      "epoch": 1.864447660395562,
+      "grad_norm": 1.9453125,
+      "learning_rate": 3e-05,
+      "loss": 3.7501,
+      "step": 7730
+    },
+    {
+      "epoch": 1.8656536420646406,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9893,
+      "step": 7735
+    },
+    {
+      "epoch": 1.8668596237337192,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.7567,
+      "step": 7740
+    },
+    {
+      "epoch": 1.8680656054027978,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.7308,
+      "step": 7745
+    },
+    {
+      "epoch": 1.8692715870718764,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.7238,
+      "step": 7750
+    },
+    {
+      "epoch": 1.8704775687409552,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8657,
+      "step": 7755
+    },
+    {
+      "epoch": 1.8716835504100338,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.7781,
+      "step": 7760
+    },
+    {
+      "epoch": 1.8728895320791124,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.7808,
+      "step": 7765
+    },
+    {
+      "epoch": 1.8740955137481912,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9164,
+      "step": 7770
+    },
+    {
+      "epoch": 1.8753014954172698,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3e-05,
+      "loss": 3.7068,
+      "step": 7775
+    },
+    {
+      "epoch": 1.8765074770863484,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8565,
+      "step": 7780
+    },
+    {
+      "epoch": 1.877713458755427,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7286,
+      "step": 7785
+    },
+    {
+      "epoch": 1.8789194404245055,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.6623,
+      "step": 7790
+    },
+    {
+      "epoch": 1.8801254220935841,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.7901,
+      "step": 7795
+    },
+    {
+      "epoch": 1.8813314037626627,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.883,
+      "step": 7800
+    },
+    {
+      "epoch": 1.8825373854317413,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.6748,
+      "step": 7805
+    },
+    {
+      "epoch": 1.88374336710082,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.7408,
+      "step": 7810
+    },
+    {
+      "epoch": 1.8849493487698987,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9042,
+      "step": 7815
+    },
+    {
+      "epoch": 1.8861553304389773,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8211,
+      "step": 7820
+    },
+    {
+      "epoch": 1.887361312108056,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0569,
+      "step": 7825
+    },
+    {
+      "epoch": 1.8885672937771347,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.5554,
+      "step": 7830
+    },
+    {
+      "epoch": 1.8897732754462133,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9346,
+      "step": 7835
+    },
+    {
+      "epoch": 1.890979257115292,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8821,
+      "step": 7840
+    },
+    {
+      "epoch": 1.8921852387843705,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.683,
+      "step": 7845
+    },
+    {
+      "epoch": 1.893391220453449,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9123,
+      "step": 7850
+    },
+    {
+      "epoch": 1.8945972021225277,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.821,
+      "step": 7855
+    },
+    {
+      "epoch": 1.8958031837916063,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.69,
+      "step": 7860
+    },
+    {
+      "epoch": 1.8970091654606849,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9012,
+      "step": 7865
+    },
+    {
+      "epoch": 1.8982151471297635,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.6692,
+      "step": 7870
+    },
+    {
+      "epoch": 1.8994211287988423,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8644,
+      "step": 7875
+    },
+    {
+      "epoch": 1.9006271104679209,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.819,
+      "step": 7880
+    },
+    {
+      "epoch": 1.9018330921369995,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7935,
+      "step": 7885
+    },
+    {
+      "epoch": 1.9030390738060783,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8848,
+      "step": 7890
+    },
+    {
+      "epoch": 1.9042450554751569,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8953,
+      "step": 7895
+    },
+    {
+      "epoch": 1.9054510371442355,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8023,
+      "step": 7900
+    },
+    {
+      "epoch": 1.906657018813314,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9101,
+      "step": 7905
+    },
+    {
+      "epoch": 1.9078630004823927,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7766,
+      "step": 7910
+    },
+    {
+      "epoch": 1.9090689821514712,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.7748,
+      "step": 7915
+    },
+    {
+      "epoch": 1.9102749638205498,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8324,
+      "step": 7920
+    },
+    {
+      "epoch": 1.9114809454896284,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7237,
+      "step": 7925
+    },
+    {
+      "epoch": 1.912686927158707,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.7041,
+      "step": 7930
+    },
+    {
+      "epoch": 1.9138929088277858,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.7389,
+      "step": 7935
+    },
+    {
+      "epoch": 1.9150988904968644,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.7476,
+      "step": 7940
+    },
+    {
+      "epoch": 1.916304872165943,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.7843,
+      "step": 7945
+    },
+    {
+      "epoch": 1.9175108538350218,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.8001,
+      "step": 7950
+    },
+    {
+      "epoch": 1.9187168355041004,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8482,
+      "step": 7955
+    },
+    {
+      "epoch": 1.919922817173179,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.7813,
+      "step": 7960
+    },
+    {
+      "epoch": 1.9211287988422576,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8595,
+      "step": 7965
+    },
+    {
+      "epoch": 1.9223347805113362,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7564,
+      "step": 7970
+    },
+    {
+      "epoch": 1.9235407621804148,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.672,
+      "step": 7975
+    },
+    {
+      "epoch": 1.9247467438494934,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.761,
+      "step": 7980
+    },
+    {
+      "epoch": 1.925952725518572,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8834,
+      "step": 7985
+    },
+    {
+      "epoch": 1.9271587071876506,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.7795,
+      "step": 7990
+    },
+    {
+      "epoch": 1.9283646888567294,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.6395,
+      "step": 7995
+    },
+    {
+      "epoch": 1.929570670525808,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8351,
+      "step": 8000
+    },
+    {
+      "epoch": 1.9307766521948866,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9706,
+      "step": 8005
+    },
+    {
+      "epoch": 1.9319826338639654,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.7406,
+      "step": 8010
+    },
+    {
+      "epoch": 1.933188615533044,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.7666,
+      "step": 8015
+    },
+    {
+      "epoch": 1.9343945972021226,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8246,
+      "step": 8020
+    },
+    {
+      "epoch": 1.9356005788712012,
+      "grad_norm": 6.125,
+      "learning_rate": 3e-05,
+      "loss": 3.761,
+      "step": 8025
+    },
+    {
+      "epoch": 1.9368065605402798,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8339,
+      "step": 8030
+    },
+    {
+      "epoch": 1.9380125422093584,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.7689,
+      "step": 8035
+    },
+    {
+      "epoch": 1.939218523878437,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.7775,
+      "step": 8040
+    },
+    {
+      "epoch": 1.9404245055475156,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8596,
+      "step": 8045
+    },
+    {
+      "epoch": 1.9416304872165941,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7681,
+      "step": 8050
+    },
+    {
+      "epoch": 1.942836468885673,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.7345,
+      "step": 8055
+    },
+    {
+      "epoch": 1.9440424505547516,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8306,
+      "step": 8060
+    },
+    {
+      "epoch": 1.9452484322238301,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.8441,
+      "step": 8065
+    },
+    {
+      "epoch": 1.946454413892909,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.8137,
+      "step": 8070
+    },
+    {
+      "epoch": 1.9476603955619876,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.7487,
+      "step": 8075
+    },
+    {
+      "epoch": 1.9488663772310661,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.7596,
+      "step": 8080
+    },
+    {
+      "epoch": 1.9500723589001447,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7839,
+      "step": 8085
+    },
+    {
+      "epoch": 1.9512783405692233,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9499,
+      "step": 8090
+    },
+    {
+      "epoch": 1.952484322238302,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.653,
+      "step": 8095
+    },
+    {
+      "epoch": 1.9536903039073805,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.6592,
+      "step": 8100
+    },
+    {
+      "epoch": 1.9548962855764591,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8865,
+      "step": 8105
+    },
+    {
+      "epoch": 1.9561022672455377,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.6638,
+      "step": 8110
+    },
+    {
+      "epoch": 1.9573082489146165,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.6911,
+      "step": 8115
+    },
+    {
+      "epoch": 1.9585142305836951,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.7635,
+      "step": 8120
+    },
+    {
+      "epoch": 1.9597202122527737,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.952,
+      "step": 8125
+    },
+    {
+      "epoch": 1.9609261939218525,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.6066,
+      "step": 8130
+    },
+    {
+      "epoch": 1.9621321755909311,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.633,
+      "step": 8135
+    },
+    {
+      "epoch": 1.9633381572600097,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7623,
+      "step": 8140
+    },
+    {
+      "epoch": 1.9645441389290883,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.8725,
+      "step": 8145
+    },
+    {
+      "epoch": 1.965750120598167,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9518,
+      "step": 8150
+    },
+    {
+      "epoch": 1.9669561022672455,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.6851,
+      "step": 8155
+    },
+    {
+      "epoch": 1.968162083936324,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.6479,
+      "step": 8160
+    },
+    {
+      "epoch": 1.9693680656054027,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7243,
+      "step": 8165
+    },
+    {
+      "epoch": 1.9705740472744813,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.6661,
+      "step": 8170
+    },
+    {
+      "epoch": 1.97178002894356,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8267,
+      "step": 8175
+    },
+    {
+      "epoch": 1.9729860106126387,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.8583,
+      "step": 8180
+    },
+    {
+      "epoch": 1.9741919922817173,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.8219,
+      "step": 8185
+    },
+    {
+      "epoch": 1.975397973950796,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.6852,
+      "step": 8190
+    },
+    {
+      "epoch": 1.9766039556198747,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.666,
+      "step": 8195
+    },
+    {
+      "epoch": 1.9778099372889533,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.7167,
+      "step": 8200
+    },
+    {
+      "epoch": 1.9790159189580319,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.906,
+      "step": 8205
+    },
+    {
+      "epoch": 1.9802219006271105,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8129,
+      "step": 8210
+    },
+    {
+      "epoch": 1.981427882296189,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9011,
+      "step": 8215
+    },
+    {
+      "epoch": 1.9826338639652676,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.7331,
+      "step": 8220
+    },
+    {
+      "epoch": 1.9838398456343462,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.7144,
+      "step": 8225
+    },
+    {
+      "epoch": 1.9850458273034248,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.6859,
+      "step": 8230
+    },
+    {
+      "epoch": 1.9862518089725036,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.62,
+      "step": 8235
+    },
+    {
+      "epoch": 1.9874577906415822,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.6762,
+      "step": 8240
+    },
+    {
+      "epoch": 1.9886637723106608,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0014,
+      "step": 8245
+    },
+    {
+      "epoch": 1.9898697539797396,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.7749,
+      "step": 8250
+    },
+    {
+      "epoch": 1.9910757356488182,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.6529,
+      "step": 8255
+    },
+    {
+      "epoch": 1.9922817173178968,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.7956,
+      "step": 8260
+    },
+    {
+      "epoch": 1.9934876989869754,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8536,
+      "step": 8265
+    },
+    {
+      "epoch": 1.994693680656054,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.6658,
+      "step": 8270
+    },
+    {
+      "epoch": 1.9958996623251326,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.7088,
+      "step": 8275
+    },
+    {
+      "epoch": 1.9971056439942112,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.7479,
+      "step": 8280
+    },
+    {
+      "epoch": 1.9983116256632898,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.7478,
+      "step": 8285
+    },
+    {
+      "epoch": 1.9995176073323684,
+      "grad_norm": 3.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9395,
+      "step": 8290
+    },
+    {
+      "epoch": 2.000723589001447,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.648,
+      "step": 8295
+    },
+    {
+      "epoch": 2.001929570670526,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.6676,
+      "step": 8300
+    },
+    {
+      "epoch": 2.0031355523396046,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8753,
+      "step": 8305
+    },
+    {
+      "epoch": 2.004341534008683,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8172,
+      "step": 8310
+    },
+    {
+      "epoch": 2.005547515677762,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.836,
+      "step": 8315
+    },
+    {
+      "epoch": 2.0067534973468404,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.7457,
+      "step": 8320
+    },
+    {
+      "epoch": 2.007959479015919,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.6816,
+      "step": 8325
+    },
+    {
+      "epoch": 2.0091654606849976,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.7902,
+      "step": 8330
+    },
+    {
+      "epoch": 2.010371442354076,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9442,
+      "step": 8335
+    },
+    {
+      "epoch": 2.0115774240231548,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.674,
+      "step": 8340
+    },
+    {
+      "epoch": 2.0127834056922334,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7012,
+      "step": 8345
+    },
+    {
+      "epoch": 2.013989387361312,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.6978,
+      "step": 8350
+    },
+    {
+      "epoch": 2.0151953690303905,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.7883,
+      "step": 8355
+    },
+    {
+      "epoch": 2.0164013506994696,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8538,
+      "step": 8360
+    },
+    {
+      "epoch": 2.017607332368548,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.7824,
+      "step": 8365
+    },
+    {
+      "epoch": 2.0188133140376268,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.713,
+      "step": 8370
+    },
+    {
+      "epoch": 2.0200192957067054,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.6853,
+      "step": 8375
+    },
+    {
+      "epoch": 2.021225277375784,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.752,
+      "step": 8380
+    },
+    {
+      "epoch": 2.0224312590448625,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.7355,
+      "step": 8385
+    },
+    {
+      "epoch": 2.023637240713941,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.6408,
+      "step": 8390
+    },
+    {
+      "epoch": 2.0248432223830197,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8283,
+      "step": 8395
+    },
+    {
+      "epoch": 2.0260492040520983,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.7683,
+      "step": 8400
+    },
+    {
+      "epoch": 2.027255185721177,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.7757,
+      "step": 8405
+    },
+    {
+      "epoch": 2.0284611673902555,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.7638,
+      "step": 8410
+    },
+    {
+      "epoch": 2.029667149059334,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.7866,
+      "step": 8415
+    },
+    {
+      "epoch": 2.030873130728413,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.6612,
+      "step": 8420
+    },
+    {
+      "epoch": 2.0320791123974917,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.7902,
+      "step": 8425
+    },
+    {
+      "epoch": 2.0332850940665703,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.7404,
+      "step": 8430
+    },
+    {
+      "epoch": 2.034491075735649,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.7562,
+      "step": 8435
+    },
+    {
+      "epoch": 2.0356970574047275,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.6321,
+      "step": 8440
+    },
+    {
+      "epoch": 2.036903039073806,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8358,
+      "step": 8445
+    },
+    {
+      "epoch": 2.0381090207428847,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.7849,
+      "step": 8450
+    },
+    {
+      "epoch": 2.0393150024119633,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.7036,
+      "step": 8455
+    },
+    {
+      "epoch": 2.040520984081042,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8943,
+      "step": 8460
+    },
+    {
+      "epoch": 2.0417269657501205,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.755,
+      "step": 8465
+    },
+    {
+      "epoch": 2.042932947419199,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8691,
+      "step": 8470
+    },
+    {
+      "epoch": 2.0441389290882777,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.8007,
+      "step": 8475
+    },
+    {
+      "epoch": 2.0453449107573567,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.6498,
+      "step": 8480
+    },
+    {
+      "epoch": 2.0465508924264353,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.8584,
+      "step": 8485
+    },
+    {
+      "epoch": 2.047756874095514,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8621,
+      "step": 8490
+    },
+    {
+      "epoch": 2.0489628557645925,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.6308,
+      "step": 8495
+    },
+    {
+      "epoch": 2.050168837433671,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.6876,
+      "step": 8500
+    },
+    {
+      "epoch": 2.0513748191027497,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.8568,
+      "step": 8505
+    },
+    {
+      "epoch": 2.0525808007718283,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.7444,
+      "step": 8510
+    },
+    {
+      "epoch": 2.053786782440907,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9624,
+      "step": 8515
+    },
+    {
+      "epoch": 2.0549927641099854,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.6974,
+      "step": 8520
+    },
+    {
+      "epoch": 2.056198745779064,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.7123,
+      "step": 8525
+    },
+    {
+      "epoch": 2.0574047274481426,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8225,
+      "step": 8530
+    },
+    {
+      "epoch": 2.058610709117221,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.65,
+      "step": 8535
+    },
+    {
+      "epoch": 2.0598166907863003,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.5742,
+      "step": 8540
+    },
+    {
+      "epoch": 2.061022672455379,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.5805,
+      "step": 8545
+    },
+    {
+      "epoch": 2.0622286541244574,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.7155,
+      "step": 8550
+    },
+    {
+      "epoch": 2.063434635793536,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8111,
+      "step": 8555
+    },
+    {
+      "epoch": 2.0646406174626146,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.7434,
+      "step": 8560
+    },
+    {
+      "epoch": 2.065846599131693,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.6068,
+      "step": 8565
+    },
+    {
+      "epoch": 2.067052580800772,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.7453,
+      "step": 8570
+    },
+    {
+      "epoch": 2.0682585624698504,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.7809,
+      "step": 8575
+    },
+    {
+      "epoch": 2.069464544138929,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.7651,
+      "step": 8580
+    },
+    {
+      "epoch": 2.0706705258080076,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.6758,
+      "step": 8585
+    },
+    {
+      "epoch": 2.071876507477086,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.5966,
+      "step": 8590
+    },
+    {
+      "epoch": 2.0730824891461648,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8621,
+      "step": 8595
+    },
+    {
+      "epoch": 2.074288470815244,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.7196,
+      "step": 8600
+    },
+    {
+      "epoch": 2.0754944524843224,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.7583,
+      "step": 8605
+    },
+    {
+      "epoch": 2.076700434153401,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7316,
+      "step": 8610
+    },
+    {
+      "epoch": 2.0779064158224796,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.7126,
+      "step": 8615
+    },
+    {
+      "epoch": 2.079112397491558,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.5228,
+      "step": 8620
+    },
+    {
+      "epoch": 2.0803183791606368,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.6893,
+      "step": 8625
+    },
+    {
+      "epoch": 2.0815243608297154,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.7325,
+      "step": 8630
+    },
+    {
+      "epoch": 2.082730342498794,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.7844,
+      "step": 8635
+    },
+    {
+      "epoch": 2.0839363241678726,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3e-05,
+      "loss": 3.8857,
+      "step": 8640
+    },
+    {
+      "epoch": 2.085142305836951,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.8558,
+      "step": 8645
+    },
+    {
+      "epoch": 2.0863482875060297,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.6917,
+      "step": 8650
+    },
+    {
+      "epoch": 2.0875542691751083,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.836,
+      "step": 8655
+    },
+    {
+      "epoch": 2.0887602508441874,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.7832,
+      "step": 8660
+    },
+    {
+      "epoch": 2.089966232513266,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.6065,
+      "step": 8665
+    },
+    {
+      "epoch": 2.0911722141823446,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.6953,
+      "step": 8670
+    },
+    {
+      "epoch": 2.092378195851423,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.6726,
+      "step": 8675
+    },
+    {
+      "epoch": 2.0935841775205017,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.7379,
+      "step": 8680
+    },
+    {
+      "epoch": 2.0947901591895803,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.7176,
+      "step": 8685
+    },
+    {
+      "epoch": 2.095996140858659,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8576,
+      "step": 8690
+    },
+    {
+      "epoch": 2.0972021225277375,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.6537,
+      "step": 8695
+    },
+    {
+      "epoch": 2.098408104196816,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8747,
+      "step": 8700
+    },
+    {
+      "epoch": 2.0996140858658947,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9617,
+      "step": 8705
+    },
+    {
+      "epoch": 2.1008200675349733,
+      "grad_norm": 3.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.6493,
+      "step": 8710
+    },
+    {
+      "epoch": 2.102026049204052,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.5946,
+      "step": 8715
+    },
+    {
+      "epoch": 2.103232030873131,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8339,
+      "step": 8720
+    },
+    {
+      "epoch": 2.1044380125422095,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8405,
+      "step": 8725
+    },
+    {
+      "epoch": 2.105643994211288,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.6722,
+      "step": 8730
+    },
+    {
+      "epoch": 2.1068499758803667,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.7286,
+      "step": 8735
+    },
+    {
+      "epoch": 2.1080559575494453,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.6787,
+      "step": 8740
+    },
+    {
+      "epoch": 2.109261939218524,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.7956,
+      "step": 8745
+    },
+    {
+      "epoch": 2.1104679208876025,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.7473,
+      "step": 8750
+    },
+    {
+      "epoch": 2.111673902556681,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.7366,
+      "step": 8755
+    },
+    {
+      "epoch": 2.1128798842257597,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.8498,
+      "step": 8760
+    },
+    {
+      "epoch": 2.1140858658948383,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.6918,
+      "step": 8765
+    },
+    {
+      "epoch": 2.115291847563917,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.777,
+      "step": 8770
+    },
+    {
+      "epoch": 2.1164978292329955,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.8905,
+      "step": 8775
+    },
+    {
+      "epoch": 2.1177038109020745,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.8154,
+      "step": 8780
+    },
+    {
+      "epoch": 2.118909792571153,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.6955,
+      "step": 8785
+    },
+    {
+      "epoch": 2.1201157742402317,
+      "grad_norm": 3.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.6849,
+      "step": 8790
+    },
+    {
+      "epoch": 2.1213217559093103,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.7385,
+      "step": 8795
+    },
+    {
+      "epoch": 2.122527737578389,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.6102,
+      "step": 8800
+    },
+    {
+      "epoch": 2.1237337192474675,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.6665,
+      "step": 8805
+    },
+    {
+      "epoch": 2.124939700916546,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.6648,
+      "step": 8810
+    },
+    {
+      "epoch": 2.1261456825856246,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8518,
+      "step": 8815
+    },
+    {
+      "epoch": 2.1273516642547032,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.739,
+      "step": 8820
+    },
+    {
+      "epoch": 2.128557645923782,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.7587,
+      "step": 8825
+    },
+    {
+      "epoch": 2.1297636275928604,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.7438,
+      "step": 8830
+    },
+    {
+      "epoch": 2.130969609261939,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.6728,
+      "step": 8835
+    },
+    {
+      "epoch": 2.132175590931018,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.7619,
+      "step": 8840
+    },
+    {
+      "epoch": 2.1333815726000966,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.864,
+      "step": 8845
+    },
+    {
+      "epoch": 2.1345875542691752,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 3.6923,
+      "step": 8850
+    },
+    {
+      "epoch": 2.135793535938254,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8623,
+      "step": 8855
+    },
+    {
+      "epoch": 2.1369995176073324,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 3.6683,
+      "step": 8860
+    },
+    {
+      "epoch": 2.138205499276411,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.649,
+      "step": 8865
+    },
+    {
+      "epoch": 2.1394114809454896,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.7971,
+      "step": 8870
+    },
+    {
+      "epoch": 2.140617462614568,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.78,
+      "step": 8875
+    },
+    {
+      "epoch": 2.141823444283647,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7552,
+      "step": 8880
+    },
+    {
+      "epoch": 2.1430294259527254,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.6223,
+      "step": 8885
+    },
+    {
+      "epoch": 2.144235407621804,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7211,
+      "step": 8890
+    },
+    {
+      "epoch": 2.1454413892908826,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.6437,
+      "step": 8895
+    },
+    {
+      "epoch": 2.1466473709599616,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8143,
+      "step": 8900
+    },
+    {
+      "epoch": 2.14785335262904,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.7189,
+      "step": 8905
+    },
+    {
+      "epoch": 2.149059334298119,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.6055,
+      "step": 8910
+    },
+    {
+      "epoch": 2.1502653159671974,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.7114,
+      "step": 8915
+    },
+    {
+      "epoch": 2.151471297636276,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.5794,
+      "step": 8920
+    },
+    {
+      "epoch": 2.1526772793053546,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.6117,
+      "step": 8925
+    },
+    {
+      "epoch": 2.153883260974433,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.7645,
+      "step": 8930
+    },
+    {
+      "epoch": 2.1550892426435118,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8283,
+      "step": 8935
+    },
+    {
+      "epoch": 2.1562952243125904,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.7704,
+      "step": 8940
+    },
+    {
+      "epoch": 2.157501205981669,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.8142,
+      "step": 8945
+    },
+    {
+      "epoch": 2.1587071876507475,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9524,
+      "step": 8950
+    },
+    {
+      "epoch": 2.159913169319826,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.6887,
+      "step": 8955
+    },
+    {
+      "epoch": 2.161119150988905,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.573,
+      "step": 8960
+    },
+    {
+      "epoch": 2.1623251326579838,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.5456,
+      "step": 8965
+    },
+    {
+      "epoch": 2.1635311143270624,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7748,
+      "step": 8970
+    },
+    {
+      "epoch": 2.164737095996141,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.7593,
+      "step": 8975
+    },
+    {
+      "epoch": 2.1659430776652195,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.7741,
+      "step": 8980
+    },
+    {
+      "epoch": 2.167149059334298,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.7436,
+      "step": 8985
+    },
+    {
+      "epoch": 2.1683550410033767,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.5113,
+      "step": 8990
+    },
+    {
+      "epoch": 2.1695610226724553,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.7764,
+      "step": 8995
+    },
+    {
+      "epoch": 2.170767004341534,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.719,
+      "step": 9000
+    },
+    {
+      "epoch": 2.0012059816690786,
+      "grad_norm": 3.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.2476,
+      "step": 9005
+    },
+    {
+      "epoch": 2.002411963338157,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.3297,
+      "step": 9010
+    },
+    {
+      "epoch": 2.003617945007236,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.3612,
+      "step": 9015
+    },
+    {
+      "epoch": 2.0048239266763144,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.3168,
+      "step": 9020
+    },
+    {
+      "epoch": 2.006029908345393,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.2866,
+      "step": 9025
+    },
+    {
+      "epoch": 2.0072358900144716,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.2497,
+      "step": 9030
+    },
+    {
+      "epoch": 2.0084418716835506,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1764,
+      "step": 9035
+    },
+    {
+      "epoch": 2.009647853352629,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.3846,
+      "step": 9040
+    },
+    {
+      "epoch": 2.010853835021708,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.3504,
+      "step": 9045
+    },
+    {
+      "epoch": 2.0120598166907864,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1678,
+      "step": 9050
+    },
+    {
+      "epoch": 2.013265798359865,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.2396,
+      "step": 9055
+    },
+    {
+      "epoch": 2.0144717800289436,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1989,
+      "step": 9060
+    },
+    {
+      "epoch": 2.015677761698022,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.2574,
+      "step": 9065
+    },
+    {
+      "epoch": 2.0168837433671007,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.3173,
+      "step": 9070
+    },
+    {
+      "epoch": 2.0180897250361793,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.2554,
+      "step": 9075
+    },
+    {
+      "epoch": 2.019295706705258,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.1888,
+      "step": 9080
+    },
+    {
+      "epoch": 2.0205016883743365,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.2018,
+      "step": 9085
+    },
+    {
+      "epoch": 2.0217076700434156,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.2681,
+      "step": 9090
+    },
+    {
+      "epoch": 2.022913651712494,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1149,
+      "step": 9095
+    },
+    {
+      "epoch": 2.0241196333815727,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.2563,
+      "step": 9100
+    },
+    {
+      "epoch": 2.0253256150506513,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.3224,
+      "step": 9105
+    },
+    {
+      "epoch": 2.02653159671973,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1351,
+      "step": 9110
+    },
+    {
+      "epoch": 2.0277375783888085,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.2382,
+      "step": 9115
+    },
+    {
+      "epoch": 2.028943560057887,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.2655,
+      "step": 9120
+    },
+    {
+      "epoch": 2.0301495417269657,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.215,
+      "step": 9125
+    },
+    {
+      "epoch": 2.0313555233960443,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.217,
+      "step": 9130
+    },
+    {
+      "epoch": 2.032561505065123,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.2117,
+      "step": 9135
+    },
+    {
+      "epoch": 2.0337674867342015,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.2194,
+      "step": 9140
+    },
+    {
+      "epoch": 2.03497346840328,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.3021,
+      "step": 9145
+    },
+    {
+      "epoch": 2.0361794500723587,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0593,
+      "step": 9150
+    },
+    {
+      "epoch": 2.0373854317414377,
+      "grad_norm": 3.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.3386,
+      "step": 9155
+    },
+    {
+      "epoch": 2.0385914134105163,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.2206,
+      "step": 9160
+    },
+    {
+      "epoch": 2.039797395079595,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.2414,
+      "step": 9165
+    },
+    {
+      "epoch": 2.0410033767486735,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.3838,
+      "step": 9170
+    },
+    {
+      "epoch": 2.042209358417752,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 4.2496,
+      "step": 9175
+    },
+    {
+      "epoch": 2.0434153400868307,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.2251,
+      "step": 9180
+    },
+    {
+      "epoch": 2.0446213217559093,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.3328,
+      "step": 9185
+    },
+    {
+      "epoch": 2.045827303424988,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0846,
+      "step": 9190
+    },
+    {
+      "epoch": 2.0470332850940665,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.2989,
+      "step": 9195
+    },
+    {
+      "epoch": 2.048239266763145,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 4.2227,
+      "step": 9200
+    },
+    {
+      "epoch": 2.0494452484322236,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.2321,
+      "step": 9205
+    },
+    {
+      "epoch": 2.0506512301013027,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1743,
+      "step": 9210
+    },
+    {
+      "epoch": 2.0518572117703813,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.31,
+      "step": 9215
+    },
+    {
+      "epoch": 2.05306319343946,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.2312,
+      "step": 9220
+    },
+    {
+      "epoch": 2.0542691751085385,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.361,
+      "step": 9225
+    },
+    {
+      "epoch": 2.055475156777617,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0888,
+      "step": 9230
+    },
+    {
+      "epoch": 2.0566811384466956,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.3034,
+      "step": 9235
+    },
+    {
+      "epoch": 2.0578871201157742,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.1833,
+      "step": 9240
+    },
+    {
+      "epoch": 2.059093101784853,
+      "grad_norm": 3.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0506,
+      "step": 9245
+    },
+    {
+      "epoch": 2.0602990834539314,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1551,
+      "step": 9250
+    },
+    {
+      "epoch": 2.06150506512301,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.165,
+      "step": 9255
+    },
+    {
+      "epoch": 2.0627110467920886,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1569,
+      "step": 9260
+    },
+    {
+      "epoch": 2.063917028461167,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.2441,
+      "step": 9265
+    },
+    {
+      "epoch": 2.065123010130246,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1088,
+      "step": 9270
+    },
+    {
+      "epoch": 2.066328991799325,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.176,
+      "step": 9275
+    },
+    {
+      "epoch": 2.0675349734684034,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.1904,
+      "step": 9280
+    },
+    {
+      "epoch": 2.068740955137482,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.257,
+      "step": 9285
+    },
+    {
+      "epoch": 2.0699469368065606,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.236,
+      "step": 9290
+    },
+    {
+      "epoch": 2.071152918475639,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0485,
+      "step": 9295
+    },
+    {
+      "epoch": 2.072358900144718,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1858,
+      "step": 9300
+    },
+    {
+      "epoch": 2.0735648818137964,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.2637,
+      "step": 9305
+    },
+    {
+      "epoch": 2.074770863482875,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.1634,
+      "step": 9310
+    },
+    {
+      "epoch": 2.0759768451519536,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.2879,
+      "step": 9315
+    },
+    {
+      "epoch": 2.077182826821032,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0801,
+      "step": 9320
+    },
+    {
+      "epoch": 2.0783888084901108,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.1329,
+      "step": 9325
+    },
+    {
+      "epoch": 2.07959479015919,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.1094,
+      "step": 9330
+    },
+    {
+      "epoch": 2.0808007718282684,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.1697,
+      "step": 9335
+    },
+    {
+      "epoch": 2.082006753497347,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.156,
+      "step": 9340
+    },
+    {
+      "epoch": 2.0832127351664256,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.3122,
+      "step": 9345
+    },
+    {
+      "epoch": 2.084418716835504,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.2945,
+      "step": 9350
+    },
+    {
+      "epoch": 2.0856246985045828,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.3271,
+      "step": 9355
+    },
+    {
+      "epoch": 2.0868306801736614,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.1325,
+      "step": 9360
+    },
+    {
+      "epoch": 2.08803666184274,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.2733,
+      "step": 9365
+    },
+    {
+      "epoch": 2.0892426435118185,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.2,
+      "step": 9370
+    },
+    {
+      "epoch": 2.090448625180897,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0549,
+      "step": 9375
+    },
+    {
+      "epoch": 2.0916546068499757,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.2768,
+      "step": 9380
+    },
+    {
+      "epoch": 2.0928605885190543,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1369,
+      "step": 9385
+    },
+    {
+      "epoch": 2.094066570188133,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.1692,
+      "step": 9390
+    },
+    {
+      "epoch": 2.095272551857212,
+      "grad_norm": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1711,
+      "step": 9395
+    },
+    {
+      "epoch": 2.0964785335262905,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.2876,
+      "step": 9400
+    },
+    {
+      "epoch": 2.097684515195369,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.2371,
+      "step": 9405
+    },
+    {
+      "epoch": 2.0988904968644477,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.2242,
+      "step": 9410
+    },
+    {
+      "epoch": 2.1000964785335263,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.2682,
+      "step": 9415
+    },
+    {
+      "epoch": 2.101302460202605,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1245,
+      "step": 9420
+    },
+    {
+      "epoch": 2.1025084418716835,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1288,
+      "step": 9425
+    },
+    {
+      "epoch": 2.103714423540762,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.4769,
+      "step": 9430
+    },
+    {
+      "epoch": 2.1049204052098407,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.1007,
+      "step": 9435
+    },
+    {
+      "epoch": 2.1061263868789193,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.09,
+      "step": 9440
+    },
+    {
+      "epoch": 2.107332368547998,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1662,
+      "step": 9445
+    },
+    {
+      "epoch": 2.108538350217077,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.2663,
+      "step": 9450
+    },
+    {
+      "epoch": 2.1097443318861555,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.2089,
+      "step": 9455
+    },
+    {
+      "epoch": 2.110950313555234,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.2068,
+      "step": 9460
+    },
+    {
+      "epoch": 2.1121562952243127,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.2668,
+      "step": 9465
+    },
+    {
+      "epoch": 2.1133622768933913,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.2088,
+      "step": 9470
+    },
+    {
+      "epoch": 2.11456825856247,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.1264,
+      "step": 9475
+    },
+    {
+      "epoch": 2.1157742402315485,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.3307,
+      "step": 9480
+    },
+    {
+      "epoch": 2.116980221900627,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.3448,
+      "step": 9485
+    },
+    {
+      "epoch": 2.1181862035697057,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1575,
+      "step": 9490
+    },
+    {
+      "epoch": 2.1193921852387843,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1604,
+      "step": 9495
+    },
+    {
+      "epoch": 2.120598166907863,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.2618,
+      "step": 9500
+    },
+    {
+      "epoch": 2.1218041485769414,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0784,
+      "step": 9505
+    },
+    {
+      "epoch": 2.12301013024602,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0385,
+      "step": 9510
+    },
+    {
+      "epoch": 2.124216111915099,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1256,
+      "step": 9515
+    },
+    {
+      "epoch": 2.1254220935841777,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1635,
+      "step": 9520
+    },
+    {
+      "epoch": 2.1266280752532563,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.3009,
+      "step": 9525
+    },
+    {
+      "epoch": 2.127834056922335,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.2126,
+      "step": 9530
+    },
+    {
+      "epoch": 2.1290400385914134,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.1022,
+      "step": 9535
+    },
+    {
+      "epoch": 2.130246020260492,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1408,
+      "step": 9540
+    },
+    {
+      "epoch": 2.1314520019295706,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.2241,
+      "step": 9545
+    },
+    {
+      "epoch": 2.132657983598649,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.2544,
+      "step": 9550
+    },
+    {
+      "epoch": 2.133863965267728,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.203,
+      "step": 9555
+    },
+    {
+      "epoch": 2.1350699469368064,
+      "grad_norm": 4.25,
+      "learning_rate": 3e-05,
+      "loss": 4.2632,
+      "step": 9560
+    },
+    {
+      "epoch": 2.136275928605885,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.1642,
+      "step": 9565
+    },
+    {
+      "epoch": 2.137481910274964,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1256,
+      "step": 9570
+    },
+    {
+      "epoch": 2.1386878919440426,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1642,
+      "step": 9575
+    },
+    {
+      "epoch": 2.1398938736131212,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.2296,
+      "step": 9580
+    },
+    {
+      "epoch": 2.1410998552822,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.2119,
+      "step": 9585
+    },
+    {
+      "epoch": 2.1423058369512784,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0698,
+      "step": 9590
+    },
+    {
+      "epoch": 2.143511818620357,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1828,
+      "step": 9595
+    },
+    {
+      "epoch": 2.1447178002894356,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.2621,
+      "step": 9600
+    },
+    {
+      "epoch": 2.145923781958514,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0704,
+      "step": 9605
+    },
+    {
+      "epoch": 2.147129763627593,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.2167,
+      "step": 9610
+    },
+    {
+      "epoch": 2.1483357452966714,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1329,
+      "step": 9615
+    },
+    {
+      "epoch": 2.14954172696575,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1253,
+      "step": 9620
+    },
+    {
+      "epoch": 2.1507477086348286,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.0756,
+      "step": 9625
+    },
+    {
+      "epoch": 2.151953690303907,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0534,
+      "step": 9630
+    },
+    {
+      "epoch": 2.153159671972986,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.193,
+      "step": 9635
+    },
+    {
+      "epoch": 2.154365653642065,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.2204,
+      "step": 9640
+    },
+    {
+      "epoch": 2.1555716353111434,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.2818,
+      "step": 9645
+    },
+    {
+      "epoch": 2.156777616980222,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1848,
+      "step": 9650
+    },
+    {
+      "epoch": 2.1579835986493006,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.2621,
+      "step": 9655
+    },
+    {
+      "epoch": 2.159189580318379,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.2945,
+      "step": 9660
+    },
+    {
+      "epoch": 2.1603955619874577,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1566,
+      "step": 9665
+    },
+    {
+      "epoch": 2.1616015436565363,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0023,
+      "step": 9670
+    },
+    {
+      "epoch": 2.162807525325615,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1066,
+      "step": 9675
+    },
+    {
+      "epoch": 2.1640135069946935,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.2053,
+      "step": 9680
+    },
+    {
+      "epoch": 2.165219488663772,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.2369,
+      "step": 9685
+    },
+    {
+      "epoch": 2.166425470332851,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.2122,
+      "step": 9690
+    },
+    {
+      "epoch": 2.1676314520019297,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0694,
+      "step": 9695
+    },
+    {
+      "epoch": 2.1688374336710083,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0493,
+      "step": 9700
+    },
+    {
+      "epoch": 2.170043415340087,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.2667,
+      "step": 9705
+    },
+    {
+      "epoch": 2.1712493970091655,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.1695,
+      "step": 9710
+    },
+    {
+      "epoch": 2.172455378678244,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1245,
+      "step": 9715
+    },
+    {
+      "epoch": 2.1736613603473227,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.3299,
+      "step": 9720
+    },
+    {
+      "epoch": 2.1748673420164013,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.1907,
+      "step": 9725
+    },
+    {
+      "epoch": 2.17607332368548,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.1464,
+      "step": 9730
+    },
+    {
+      "epoch": 2.1772793053545585,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1861,
+      "step": 9735
+    },
+    {
+      "epoch": 2.178485287023637,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0978,
+      "step": 9740
+    },
+    {
+      "epoch": 2.1796912686927157,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1823,
+      "step": 9745
+    },
+    {
+      "epoch": 2.1808972503617943,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.2823,
+      "step": 9750
+    },
+    {
+      "epoch": 2.1821032320308733,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.2414,
+      "step": 9755
+    },
+    {
+      "epoch": 2.183309213699952,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.1525,
+      "step": 9760
+    },
+    {
+      "epoch": 2.1845151953690305,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.1164,
+      "step": 9765
+    },
+    {
+      "epoch": 2.185721177038109,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.1465,
+      "step": 9770
+    },
+    {
+      "epoch": 2.1869271587071877,
+      "grad_norm": 4.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1222,
+      "step": 9775
+    },
+    {
+      "epoch": 2.1881331403762663,
+      "grad_norm": 3.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.1895,
+      "step": 9780
+    },
+    {
+      "epoch": 2.189339122045345,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.3288,
+      "step": 9785
+    },
+    {
+      "epoch": 2.1905451037144235,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.1525,
+      "step": 9790
+    },
+    {
+      "epoch": 2.191751085383502,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.2228,
+      "step": 9795
+    },
+    {
+      "epoch": 2.1929570670525806,
+      "grad_norm": 3.5,
+      "learning_rate": 3e-05,
+      "loss": 4.3341,
+      "step": 9800
+    },
+    {
+      "epoch": 2.1941630487216592,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.2245,
+      "step": 9805
+    },
+    {
+      "epoch": 2.1953690303907383,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0826,
+      "step": 9810
+    },
+    {
+      "epoch": 2.196575012059817,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.1116,
+      "step": 9815
+    },
+    {
+      "epoch": 2.1977809937288955,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.1183,
+      "step": 9820
+    },
+    {
+      "epoch": 2.198986975397974,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0587,
+      "step": 9825
+    },
+    {
+      "epoch": 2.2001929570670526,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1993,
+      "step": 9830
+    },
+    {
+      "epoch": 2.2013989387361312,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.1627,
+      "step": 9835
+    },
+    {
+      "epoch": 2.20260492040521,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.2873,
+      "step": 9840
+    },
+    {
+      "epoch": 2.2038109020742884,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.2517,
+      "step": 9845
+    },
+    {
+      "epoch": 2.205016883743367,
+      "grad_norm": 2.0,
+      "learning_rate": 3e-05,
+      "loss": 4.2902,
+      "step": 9850
+    },
+    {
+      "epoch": 2.2062228654124456,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.1335,
+      "step": 9855
+    },
+    {
+      "epoch": 2.207428847081524,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1005,
+      "step": 9860
+    },
+    {
+      "epoch": 2.208634828750603,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.3688,
+      "step": 9865
+    },
+    {
+      "epoch": 2.2098408104196814,
+      "grad_norm": 3.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.2761,
+      "step": 9870
+    },
+    {
+      "epoch": 2.2110467920887604,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9508,
+      "step": 9875
+    },
+    {
+      "epoch": 2.212252773757839,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.2251,
+      "step": 9880
+    },
+    {
+      "epoch": 2.2134587554269176,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1957,
+      "step": 9885
+    },
+    {
+      "epoch": 2.214664737095996,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.1688,
+      "step": 9890
+    },
+    {
+      "epoch": 2.215870718765075,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.2139,
+      "step": 9895
+    },
+    {
+      "epoch": 2.2170767004341534,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.2213,
+      "step": 9900
+    },
+    {
+      "epoch": 2.218282682103232,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.13,
+      "step": 9905
+    },
+    {
+      "epoch": 2.2194886637723106,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1776,
+      "step": 9910
+    },
+    {
+      "epoch": 2.220694645441389,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1426,
+      "step": 9915
+    },
+    {
+      "epoch": 2.2219006271104678,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1578,
+      "step": 9920
+    },
+    {
+      "epoch": 2.2231066087795464,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0207,
+      "step": 9925
+    },
+    {
+      "epoch": 2.2243125904486254,
+      "grad_norm": 3.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.2973,
+      "step": 9930
+    },
+    {
+      "epoch": 2.225518572117704,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0343,
+      "step": 9935
+    },
+    {
+      "epoch": 2.2267245537867826,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0924,
+      "step": 9940
+    },
+    {
+      "epoch": 2.227930535455861,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1201,
+      "step": 9945
+    },
+    {
+      "epoch": 2.2291365171249398,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0651,
+      "step": 9950
+    },
+    {
+      "epoch": 2.2303424987940184,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.2184,
+      "step": 9955
+    },
+    {
+      "epoch": 2.231548480463097,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.0086,
+      "step": 9960
+    },
+    {
+      "epoch": 2.2327544621321755,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.2001,
+      "step": 9965
+    },
+    {
+      "epoch": 2.233960443801254,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.3328,
+      "step": 9970
+    },
+    {
+      "epoch": 2.2351664254703327,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.2686,
+      "step": 9975
+    },
+    {
+      "epoch": 2.2363724071394113,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0926,
+      "step": 9980
+    },
+    {
+      "epoch": 2.23757838880849,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.2253,
+      "step": 9985
+    },
+    {
+      "epoch": 2.2387843704775685,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0817,
+      "step": 9990
+    },
+    {
+      "epoch": 2.2399903521466475,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.2543,
+      "step": 9995
+    },
+    {
+      "epoch": 2.241196333815726,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0528,
+      "step": 10000
+    },
+    {
+      "epoch": 2.2424023154848047,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.2123,
+      "step": 10005
+    },
+    {
+      "epoch": 2.2436082971538833,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1992,
+      "step": 10010
+    },
+    {
+      "epoch": 2.244814278822962,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.212,
+      "step": 10015
+    },
+    {
+      "epoch": 2.2460202604920405,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.1248,
+      "step": 10020
+    },
+    {
+      "epoch": 2.247226242161119,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.2627,
+      "step": 10025
+    },
+    {
+      "epoch": 2.2484322238301977,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.103,
+      "step": 10030
+    },
+    {
+      "epoch": 2.2496382054992763,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1098,
+      "step": 10035
+    },
+    {
+      "epoch": 2.250844187168355,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.1861,
+      "step": 10040
+    },
+    {
+      "epoch": 2.2520501688374335,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1603,
+      "step": 10045
+    },
+    {
+      "epoch": 2.2532561505065125,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1436,
+      "step": 10050
+    },
+    {
+      "epoch": 2.254462132175591,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.1656,
+      "step": 10055
+    },
+    {
+      "epoch": 2.2556681138446697,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9487,
+      "step": 10060
+    },
+    {
+      "epoch": 2.2568740955137483,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1735,
+      "step": 10065
+    },
+    {
+      "epoch": 2.258080077182827,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1959,
+      "step": 10070
+    },
+    {
+      "epoch": 2.2592860588519055,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.1209,
+      "step": 10075
+    },
+    {
+      "epoch": 2.260492040520984,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1955,
+      "step": 10080
+    },
+    {
+      "epoch": 2.2616980221900627,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1891,
+      "step": 10085
+    },
+    {
+      "epoch": 2.2629040038591413,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0758,
+      "step": 10090
+    },
+    {
+      "epoch": 2.26410998552822,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1674,
+      "step": 10095
+    },
+    {
+      "epoch": 2.2653159671972984,
+      "grad_norm": 3.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1084,
+      "step": 10100
+    },
+    {
+      "epoch": 2.2665219488663775,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1182,
+      "step": 10105
+    },
+    {
+      "epoch": 2.2677279305354556,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0088,
+      "step": 10110
+    },
+    {
+      "epoch": 2.2689339122045347,
+      "grad_norm": 3.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.2343,
+      "step": 10115
+    },
+    {
+      "epoch": 2.2701398938736133,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0165,
+      "step": 10120
+    },
+    {
+      "epoch": 2.271345875542692,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0021,
+      "step": 10125
+    },
+    {
+      "epoch": 2.2725518572117704,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1563,
+      "step": 10130
+    },
+    {
+      "epoch": 2.273757838880849,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0757,
+      "step": 10135
+    },
+    {
+      "epoch": 2.2749638205499276,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.1532,
+      "step": 10140
+    },
+    {
+      "epoch": 2.276169802219006,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9921,
+      "step": 10145
+    },
+    {
+      "epoch": 2.277375783888085,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1515,
+      "step": 10150
+    },
+    {
+      "epoch": 2.2785817655571634,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.2955,
+      "step": 10155
+    },
+    {
+      "epoch": 2.279787747226242,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.266,
+      "step": 10160
+    },
+    {
+      "epoch": 2.2809937288953206,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.061,
+      "step": 10165
+    },
+    {
+      "epoch": 2.2821997105643996,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.19,
+      "step": 10170
+    },
+    {
+      "epoch": 2.2834056922334782,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0745,
+      "step": 10175
+    },
+    {
+      "epoch": 2.284611673902557,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.1898,
+      "step": 10180
+    },
+    {
+      "epoch": 2.2858176555716354,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0209,
+      "step": 10185
+    },
+    {
+      "epoch": 2.287023637240714,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.2058,
+      "step": 10190
+    },
+    {
+      "epoch": 2.2882296189097926,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.183,
+      "step": 10195
+    },
+    {
+      "epoch": 2.289435600578871,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1823,
+      "step": 10200
+    },
+    {
+      "epoch": 2.29064158224795,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.1024,
+      "step": 10205
+    },
+    {
+      "epoch": 2.2918475639170284,
+      "grad_norm": 1.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.1754,
+      "step": 10210
+    },
+    {
+      "epoch": 2.293053545586107,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.117,
+      "step": 10215
+    },
+    {
+      "epoch": 2.2942595272551856,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.022,
+      "step": 10220
+    },
+    {
+      "epoch": 2.2954655089242646,
+      "grad_norm": 1.9921875,
+      "learning_rate": 3e-05,
+      "loss": 4.2116,
+      "step": 10225
+    },
+    {
+      "epoch": 2.2966714905933427,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1104,
+      "step": 10230
+    },
+    {
+      "epoch": 2.297877472262422,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1166,
+      "step": 10235
+    },
+    {
+      "epoch": 2.2990834539315004,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.1814,
+      "step": 10240
+    },
+    {
+      "epoch": 2.300289435600579,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.1516,
+      "step": 10245
+    },
+    {
+      "epoch": 2.3014954172696576,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0829,
+      "step": 10250
+    },
+    {
+      "epoch": 2.302701398938736,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1533,
+      "step": 10255
+    },
+    {
+      "epoch": 2.3039073806078147,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1477,
+      "step": 10260
+    },
+    {
+      "epoch": 2.3051133622768933,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0464,
+      "step": 10265
+    },
+    {
+      "epoch": 2.306319343945972,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.086,
+      "step": 10270
+    },
+    {
+      "epoch": 2.3075253256150505,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.2252,
+      "step": 10275
+    },
+    {
+      "epoch": 2.308731307284129,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0657,
+      "step": 10280
+    },
+    {
+      "epoch": 2.3099372889532077,
+      "grad_norm": 3.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1504,
+      "step": 10285
+    },
+    {
+      "epoch": 2.3111432706222867,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.8939,
+      "step": 10290
+    },
+    {
+      "epoch": 2.3123492522913653,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.1186,
+      "step": 10295
+    },
+    {
+      "epoch": 2.313555233960444,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.1986,
+      "step": 10300
+    },
+    {
+      "epoch": 2.3147612156295225,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.2401,
+      "step": 10305
+    },
+    {
+      "epoch": 2.315967197298601,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.1075,
+      "step": 10310
+    },
+    {
+      "epoch": 2.3171731789676797,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.2698,
+      "step": 10315
+    },
+    {
+      "epoch": 2.3183791606367583,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.2006,
+      "step": 10320
+    },
+    {
+      "epoch": 2.319585142305837,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0476,
+      "step": 10325
+    },
+    {
+      "epoch": 2.3207911239749155,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0846,
+      "step": 10330
+    },
+    {
+      "epoch": 2.321997105643994,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0502,
+      "step": 10335
+    },
+    {
+      "epoch": 2.3232030873130727,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.1355,
+      "step": 10340
+    },
+    {
+      "epoch": 2.3244090689821517,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.2342,
+      "step": 10345
+    },
+    {
+      "epoch": 2.32561505065123,
+      "grad_norm": 3.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1098,
+      "step": 10350
+    },
+    {
+      "epoch": 2.326821032320309,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.3324,
+      "step": 10355
+    },
+    {
+      "epoch": 2.3280270139893875,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1441,
+      "step": 10360
+    },
+    {
+      "epoch": 2.329232995658466,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1193,
+      "step": 10365
+    },
+    {
+      "epoch": 2.3304389773275447,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1809,
+      "step": 10370
+    },
+    {
+      "epoch": 2.3316449589966233,
+      "grad_norm": 3.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.2074,
+      "step": 10375
+    },
+    {
+      "epoch": 2.332850940665702,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.1207,
+      "step": 10380
+    },
+    {
+      "epoch": 2.3340569223347805,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1611,
+      "step": 10385
+    },
+    {
+      "epoch": 2.335262904003859,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9704,
+      "step": 10390
+    },
+    {
+      "epoch": 2.3364688856729376,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.2003,
+      "step": 10395
+    },
+    {
+      "epoch": 2.3376748673420162,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1499,
+      "step": 10400
+    },
+    {
+      "epoch": 2.338880849011095,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.333,
+      "step": 10405
+    },
+    {
+      "epoch": 2.340086830680174,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.2561,
+      "step": 10410
+    },
+    {
+      "epoch": 2.3412928123492525,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0917,
+      "step": 10415
+    },
+    {
+      "epoch": 2.342498794018331,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1949,
+      "step": 10420
+    },
+    {
+      "epoch": 2.3437047756874096,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0743,
+      "step": 10425
+    },
+    {
+      "epoch": 2.3449107573564882,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1132,
+      "step": 10430
+    },
+    {
+      "epoch": 2.346116739025567,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.3364,
+      "step": 10435
+    },
+    {
+      "epoch": 2.3473227206946454,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.2251,
+      "step": 10440
+    },
+    {
+      "epoch": 2.348528702363724,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.1887,
+      "step": 10445
+    },
+    {
+      "epoch": 2.3497346840328026,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.2043,
+      "step": 10450
+    },
+    {
+      "epoch": 2.350940665701881,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.077,
+      "step": 10455
+    },
+    {
+      "epoch": 2.35214664737096,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0716,
+      "step": 10460
+    },
+    {
+      "epoch": 2.353352629040039,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.3089,
+      "step": 10465
+    },
+    {
+      "epoch": 2.354558610709117,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1143,
+      "step": 10470
+    },
+    {
+      "epoch": 2.355764592378196,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.2583,
+      "step": 10475
+    },
+    {
+      "epoch": 2.3569705740472746,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.029,
+      "step": 10480
+    },
+    {
+      "epoch": 2.358176555716353,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.1102,
+      "step": 10485
+    },
+    {
+      "epoch": 2.359382537385432,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1991,
+      "step": 10490
+    },
+    {
+      "epoch": 2.3605885190545104,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.1438,
+      "step": 10495
+    },
+    {
+      "epoch": 2.361794500723589,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0709,
+      "step": 10500
+    },
+    {
+      "epoch": 2.3630004823926676,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1308,
+      "step": 10505
+    },
+    {
+      "epoch": 2.364206464061746,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0969,
+      "step": 10510
+    },
+    {
+      "epoch": 2.3654124457308248,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0492,
+      "step": 10515
+    },
+    {
+      "epoch": 2.3666184273999034,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.079,
+      "step": 10520
+    },
+    {
+      "epoch": 2.367824409068982,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.2073,
+      "step": 10525
+    },
+    {
+      "epoch": 2.369030390738061,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.3452,
+      "step": 10530
+    },
+    {
+      "epoch": 2.3702363724071396,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.1712,
+      "step": 10535
+    },
+    {
+      "epoch": 2.371442354076218,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1475,
+      "step": 10540
+    },
+    {
+      "epoch": 2.3726483357452968,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0714,
+      "step": 10545
+    },
+    {
+      "epoch": 2.3738543174143754,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0849,
+      "step": 10550
+    },
+    {
+      "epoch": 2.375060299083454,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1826,
+      "step": 10555
+    },
+    {
+      "epoch": 2.3762662807525325,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1175,
+      "step": 10560
+    },
+    {
+      "epoch": 2.377472262421611,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9737,
+      "step": 10565
+    },
+    {
+      "epoch": 2.3786782440906897,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0004,
+      "step": 10570
+    },
+    {
+      "epoch": 2.3798842257597683,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.2172,
+      "step": 10575
+    },
+    {
+      "epoch": 2.381090207428847,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.2544,
+      "step": 10580
+    },
+    {
+      "epoch": 2.382296189097926,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0196,
+      "step": 10585
+    },
+    {
+      "epoch": 2.383502170767004,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0762,
+      "step": 10590
+    },
+    {
+      "epoch": 2.384708152436083,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0776,
+      "step": 10595
+    },
+    {
+      "epoch": 2.3859141341051617,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9896,
+      "step": 10600
+    },
+    {
+      "epoch": 2.3871201157742403,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.3171,
+      "step": 10605
+    },
+    {
+      "epoch": 2.388326097443319,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1992,
+      "step": 10610
+    },
+    {
+      "epoch": 2.3895320791123975,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.2065,
+      "step": 10615
+    },
+    {
+      "epoch": 2.390738060781476,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0532,
+      "step": 10620
+    },
+    {
+      "epoch": 2.3919440424505547,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1709,
+      "step": 10625
+    },
+    {
+      "epoch": 2.3931500241196333,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0705,
+      "step": 10630
+    },
+    {
+      "epoch": 2.394356005788712,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.1072,
+      "step": 10635
+    },
+    {
+      "epoch": 2.3955619874577905,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1484,
+      "step": 10640
+    },
+    {
+      "epoch": 2.396767969126869,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.1344,
+      "step": 10645
+    },
+    {
+      "epoch": 2.397973950795948,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.2026,
+      "step": 10650
+    },
+    {
+      "epoch": 2.3991799324650267,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0286,
+      "step": 10655
+    },
+    {
+      "epoch": 2.4003859141341053,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1975,
+      "step": 10660
+    },
+    {
+      "epoch": 2.401591895803184,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.2446,
+      "step": 10665
+    },
+    {
+      "epoch": 2.4027978774722625,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1889,
+      "step": 10670
+    },
+    {
+      "epoch": 2.404003859141341,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.2531,
+      "step": 10675
+    },
+    {
+      "epoch": 2.4052098408104197,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.2263,
+      "step": 10680
+    },
+    {
+      "epoch": 2.4064158224794983,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.2489,
+      "step": 10685
+    },
+    {
+      "epoch": 2.407621804148577,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.2175,
+      "step": 10690
+    },
+    {
+      "epoch": 2.4088277858176554,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0807,
+      "step": 10695
+    },
+    {
+      "epoch": 2.410033767486734,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1817,
+      "step": 10700
+    },
+    {
+      "epoch": 2.411239749155813,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.1689,
+      "step": 10705
+    },
+    {
+      "epoch": 2.412445730824891,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9629,
+      "step": 10710
+    },
+    {
+      "epoch": 2.4136517124939703,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1763,
+      "step": 10715
+    },
+    {
+      "epoch": 2.414857694163049,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1753,
+      "step": 10720
+    },
+    {
+      "epoch": 2.4160636758321274,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.1507,
+      "step": 10725
+    },
+    {
+      "epoch": 2.417269657501206,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0001,
+      "step": 10730
+    },
+    {
+      "epoch": 2.4184756391702846,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.108,
+      "step": 10735
+    },
+    {
+      "epoch": 2.4196816208393632,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.1973,
+      "step": 10740
+    },
+    {
+      "epoch": 2.420887602508442,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.973,
+      "step": 10745
+    },
+    {
+      "epoch": 2.4220935841775204,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.2821,
+      "step": 10750
+    },
+    {
+      "epoch": 2.423299565846599,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.143,
+      "step": 10755
+    },
+    {
+      "epoch": 2.4245055475156776,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0657,
+      "step": 10760
+    },
+    {
+      "epoch": 2.425711529184756,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1022,
+      "step": 10765
+    },
+    {
+      "epoch": 2.4269175108538352,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0939,
+      "step": 10770
+    },
+    {
+      "epoch": 2.428123492522914,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.077,
+      "step": 10775
+    },
+    {
+      "epoch": 2.4293294741919924,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0794,
+      "step": 10780
+    },
+    {
+      "epoch": 2.430535455861071,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.104,
+      "step": 10785
+    },
+    {
+      "epoch": 2.4317414375301496,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1079,
+      "step": 10790
+    },
+    {
+      "epoch": 2.432947419199228,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.1413,
+      "step": 10795
+    },
+    {
+      "epoch": 2.434153400868307,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.2506,
+      "step": 10800
+    },
+    {
+      "epoch": 2.4353593825373854,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1035,
+      "step": 10805
+    },
+    {
+      "epoch": 2.436565364206464,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1023,
+      "step": 10810
+    },
+    {
+      "epoch": 2.4377713458755426,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0791,
+      "step": 10815
+    },
+    {
+      "epoch": 2.438977327544621,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0464,
+      "step": 10820
+    },
+    {
+      "epoch": 2.4401833092137,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1936,
+      "step": 10825
+    },
+    {
+      "epoch": 2.4413892908827783,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.1159,
+      "step": 10830
+    },
+    {
+      "epoch": 2.4425952725518574,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.2167,
+      "step": 10835
+    },
+    {
+      "epoch": 2.443801254220936,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1819,
+      "step": 10840
+    },
+    {
+      "epoch": 2.4450072358900146,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1397,
+      "step": 10845
+    },
+    {
+      "epoch": 2.446213217559093,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.0418,
+      "step": 10850
+    },
+    {
+      "epoch": 2.4474191992281717,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.2023,
+      "step": 10855
+    },
+    {
+      "epoch": 2.4486251808972503,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1893,
+      "step": 10860
+    },
+    {
+      "epoch": 2.449831162566329,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0535,
+      "step": 10865
+    },
+    {
+      "epoch": 2.4510371442354075,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.231,
+      "step": 10870
+    },
+    {
+      "epoch": 2.452243125904486,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0816,
+      "step": 10875
+    },
+    {
+      "epoch": 2.4534491075735647,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1265,
+      "step": 10880
+    },
+    {
+      "epoch": 2.4546550892426433,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.1443,
+      "step": 10885
+    },
+    {
+      "epoch": 2.4558610709117223,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9236,
+      "step": 10890
+    },
+    {
+      "epoch": 2.457067052580801,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1654,
+      "step": 10895
+    },
+    {
+      "epoch": 2.4582730342498795,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1724,
+      "step": 10900
+    },
+    {
+      "epoch": 2.459479015918958,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0959,
+      "step": 10905
+    },
+    {
+      "epoch": 2.4606849975880367,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.1286,
+      "step": 10910
+    },
+    {
+      "epoch": 2.4618909792571153,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1728,
+      "step": 10915
+    },
+    {
+      "epoch": 2.463096960926194,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.2028,
+      "step": 10920
+    },
+    {
+      "epoch": 2.4643029425952725,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1971,
+      "step": 10925
+    },
+    {
+      "epoch": 2.465508924264351,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1073,
+      "step": 10930
+    },
+    {
+      "epoch": 2.4667149059334297,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.1725,
+      "step": 10935
+    },
+    {
+      "epoch": 2.4679208876025083,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.2382,
+      "step": 10940
+    },
+    {
+      "epoch": 2.4691268692715873,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.2438,
+      "step": 10945
+    },
+    {
+      "epoch": 2.4703328509406655,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.144,
+      "step": 10950
+    },
+    {
+      "epoch": 2.4715388326097445,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.2194,
+      "step": 10955
+    },
+    {
+      "epoch": 2.472744814278823,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.1659,
+      "step": 10960
+    },
+    {
+      "epoch": 2.4739507959479017,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2302,
+      "step": 10965
+    },
+    {
+      "epoch": 2.4751567776169803,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1926,
+      "step": 10970
+    },
+    {
+      "epoch": 2.476362759286059,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.2911,
+      "step": 10975
+    },
+    {
+      "epoch": 2.4775687409551375,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1058,
+      "step": 10980
+    },
+    {
+      "epoch": 2.478774722624216,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0897,
+      "step": 10985
+    },
+    {
+      "epoch": 2.4799807042932946,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.2864,
+      "step": 10990
+    },
+    {
+      "epoch": 2.4811866859623732,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.0392,
+      "step": 10995
+    },
+    {
+      "epoch": 2.482392667631452,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.2263,
+      "step": 11000
+    },
+    {
+      "epoch": 2.4835986493005304,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0793,
+      "step": 11005
+    },
+    {
+      "epoch": 2.4848046309696095,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9963,
+      "step": 11010
+    },
+    {
+      "epoch": 2.486010612638688,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.1093,
+      "step": 11015
+    },
+    {
+      "epoch": 2.4872165943077666,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.1803,
+      "step": 11020
+    },
+    {
+      "epoch": 2.4884225759768452,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1233,
+      "step": 11025
+    },
+    {
+      "epoch": 2.489628557645924,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0788,
+      "step": 11030
+    },
+    {
+      "epoch": 2.4908345393150024,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1316,
+      "step": 11035
+    },
+    {
+      "epoch": 2.492040520984081,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.1877,
+      "step": 11040
+    },
+    {
+      "epoch": 2.4932465026531596,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0385,
+      "step": 11045
+    },
+    {
+      "epoch": 2.494452484322238,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.1771,
+      "step": 11050
+    },
+    {
+      "epoch": 2.495658465991317,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9814,
+      "step": 11055
+    },
+    {
+      "epoch": 2.4968644476603954,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.0832,
+      "step": 11060
+    },
+    {
+      "epoch": 2.4980704293294744,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.1175,
+      "step": 11065
+    },
+    {
+      "epoch": 2.4992764109985526,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1414,
+      "step": 11070
+    },
+    {
+      "epoch": 2.5004823926676316,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0533,
+      "step": 11075
+    },
+    {
+      "epoch": 2.50168837433671,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0339,
+      "step": 11080
+    },
+    {
+      "epoch": 2.502894356005789,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0856,
+      "step": 11085
+    },
+    {
+      "epoch": 2.5041003376748674,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0876,
+      "step": 11090
+    },
+    {
+      "epoch": 2.505306319343946,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.2181,
+      "step": 11095
+    },
+    {
+      "epoch": 2.5065123010130246,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.153,
+      "step": 11100
+    },
+    {
+      "epoch": 2.507718282682103,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.2022,
+      "step": 11105
+    },
+    {
+      "epoch": 2.5089242643511818,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1784,
+      "step": 11110
+    },
+    {
+      "epoch": 2.5101302460202604,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1909,
+      "step": 11115
+    },
+    {
+      "epoch": 2.5113362276893394,
+      "grad_norm": 3.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.19,
+      "step": 11120
+    },
+    {
+      "epoch": 2.5125422093584175,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1537,
+      "step": 11125
+    },
+    {
+      "epoch": 2.5137481910274966,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.2259,
+      "step": 11130
+    },
+    {
+      "epoch": 2.514954172696575,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1133,
+      "step": 11135
+    },
+    {
+      "epoch": 2.5161601543656538,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1808,
+      "step": 11140
+    },
+    {
+      "epoch": 2.5173661360347324,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.1005,
+      "step": 11145
+    },
+    {
+      "epoch": 2.518572117703811,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0976,
+      "step": 11150
+    },
+    {
+      "epoch": 2.5197780993728895,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1336,
+      "step": 11155
+    },
+    {
+      "epoch": 2.520984081041968,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0542,
+      "step": 11160
+    },
+    {
+      "epoch": 2.5221900627110467,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9754,
+      "step": 11165
+    },
+    {
+      "epoch": 2.5233960443801253,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1391,
+      "step": 11170
+    },
+    {
+      "epoch": 2.524602026049204,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0785,
+      "step": 11175
+    },
+    {
+      "epoch": 2.5258080077182825,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1229,
+      "step": 11180
+    },
+    {
+      "epoch": 2.5270139893873615,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0562,
+      "step": 11185
+    },
+    {
+      "epoch": 2.5282199710564397,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1304,
+      "step": 11190
+    },
+    {
+      "epoch": 2.5294259527255187,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1364,
+      "step": 11195
+    },
+    {
+      "epoch": 2.5306319343945973,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.061,
+      "step": 11200
+    },
+    {
+      "epoch": 2.531837916063676,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0671,
+      "step": 11205
+    },
+    {
+      "epoch": 2.5330438977327545,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0776,
+      "step": 11210
+    },
+    {
+      "epoch": 2.534249879401833,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1167,
+      "step": 11215
+    },
+    {
+      "epoch": 2.5354558610709117,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0123,
+      "step": 11220
+    },
+    {
+      "epoch": 2.5366618427399903,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.1062,
+      "step": 11225
+    },
+    {
+      "epoch": 2.537867824409069,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9938,
+      "step": 11230
+    },
+    {
+      "epoch": 2.5390738060781475,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1367,
+      "step": 11235
+    },
+    {
+      "epoch": 2.5402797877472265,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9744,
+      "step": 11240
+    },
+    {
+      "epoch": 2.5414857694163047,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.1372,
+      "step": 11245
+    },
+    {
+      "epoch": 2.5426917510853837,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.1641,
+      "step": 11250
+    },
+    {
+      "epoch": 2.5438977327544623,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0557,
+      "step": 11255
+    },
+    {
+      "epoch": 2.545103714423541,
+      "grad_norm": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1024,
+      "step": 11260
+    },
+    {
+      "epoch": 2.5463096960926195,
+      "grad_norm": 1.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.0954,
+      "step": 11265
+    },
+    {
+      "epoch": 2.547515677761698,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1508,
+      "step": 11270
+    },
+    {
+      "epoch": 2.5487216594307767,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1624,
+      "step": 11275
+    },
+    {
+      "epoch": 2.5499276410998553,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.989,
+      "step": 11280
+    },
+    {
+      "epoch": 2.551133622768934,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1942,
+      "step": 11285
+    },
+    {
+      "epoch": 2.5523396044380124,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1694,
+      "step": 11290
+    },
+    {
+      "epoch": 2.553545586107091,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0985,
+      "step": 11295
+    },
+    {
+      "epoch": 2.5547515677761696,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.2778,
+      "step": 11300
+    },
+    {
+      "epoch": 2.5559575494452487,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1524,
+      "step": 11305
+    },
+    {
+      "epoch": 2.557163531114327,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0022,
+      "step": 11310
+    },
+    {
+      "epoch": 2.558369512783406,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.044,
+      "step": 11315
+    },
+    {
+      "epoch": 2.5595754944524844,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2498,
+      "step": 11320
+    },
+    {
+      "epoch": 2.560781476121563,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1297,
+      "step": 11325
+    },
+    {
+      "epoch": 2.5619874577906416,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1091,
+      "step": 11330
+    },
+    {
+      "epoch": 2.5631934394597202,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9753,
+      "step": 11335
+    },
+    {
+      "epoch": 2.564399421128799,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0582,
+      "step": 11340
+    },
+    {
+      "epoch": 2.5656054027978774,
+      "grad_norm": 3.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9721,
+      "step": 11345
+    },
+    {
+      "epoch": 2.566811384466956,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.142,
+      "step": 11350
+    },
+    {
+      "epoch": 2.5680173661360346,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.3733,
+      "step": 11355
+    },
+    {
+      "epoch": 2.5692233478051136,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1027,
+      "step": 11360
+    },
+    {
+      "epoch": 2.570429329474192,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9416,
+      "step": 11365
+    },
+    {
+      "epoch": 2.571635311143271,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1964,
+      "step": 11370
+    },
+    {
+      "epoch": 2.5728412928123494,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.2444,
+      "step": 11375
+    },
+    {
+      "epoch": 2.574047274481428,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9891,
+      "step": 11380
+    },
+    {
+      "epoch": 2.5752532561505066,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1265,
+      "step": 11385
+    },
+    {
+      "epoch": 2.576459237819585,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1045,
+      "step": 11390
+    },
+    {
+      "epoch": 2.577665219488664,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.181,
+      "step": 11395
+    },
+    {
+      "epoch": 2.5788712011577424,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1055,
+      "step": 11400
+    },
+    {
+      "epoch": 2.580077182826821,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1873,
+      "step": 11405
+    },
+    {
+      "epoch": 2.5812831644958996,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0753,
+      "step": 11410
+    },
+    {
+      "epoch": 2.582489146164978,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2336,
+      "step": 11415
+    },
+    {
+      "epoch": 2.5836951278340567,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0895,
+      "step": 11420
+    },
+    {
+      "epoch": 2.584901109503136,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0433,
+      "step": 11425
+    },
+    {
+      "epoch": 2.586107091172214,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.2776,
+      "step": 11430
+    },
+    {
+      "epoch": 2.587313072841293,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0463,
+      "step": 11435
+    },
+    {
+      "epoch": 2.5885190545103716,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1704,
+      "step": 11440
+    },
+    {
+      "epoch": 2.58972503617945,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.1732,
+      "step": 11445
+    },
+    {
+      "epoch": 2.5909310178485288,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1178,
+      "step": 11450
+    },
+    {
+      "epoch": 2.5921369995176073,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1073,
+      "step": 11455
+    },
+    {
+      "epoch": 2.593342981186686,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.2243,
+      "step": 11460
+    },
+    {
+      "epoch": 2.5945489628557645,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.023,
+      "step": 11465
+    },
+    {
+      "epoch": 2.595754944524843,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1639,
+      "step": 11470
+    },
+    {
+      "epoch": 2.5969609261939217,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0331,
+      "step": 11475
+    },
+    {
+      "epoch": 2.5981669078630008,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9988,
+      "step": 11480
+    },
+    {
+      "epoch": 2.599372889532079,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.176,
+      "step": 11485
+    },
+    {
+      "epoch": 2.600578871201158,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.067,
+      "step": 11490
+    },
+    {
+      "epoch": 2.6017848528702365,
+      "grad_norm": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1904,
+      "step": 11495
+    },
+    {
+      "epoch": 2.602990834539315,
+      "grad_norm": 3.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.098,
+      "step": 11500
+    },
+    {
+      "epoch": 2.6041968162083937,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1901,
+      "step": 11505
+    },
+    {
+      "epoch": 2.6054027978774723,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1816,
+      "step": 11510
+    },
+    {
+      "epoch": 2.606608779546551,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.137,
+      "step": 11515
+    },
+    {
+      "epoch": 2.6078147612156295,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0361,
+      "step": 11520
+    },
+    {
+      "epoch": 2.609020742884708,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.2011,
+      "step": 11525
+    },
+    {
+      "epoch": 2.6102267245537867,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.137,
+      "step": 11530
+    },
+    {
+      "epoch": 2.6114327062228653,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1566,
+      "step": 11535
+    },
+    {
+      "epoch": 2.612638687891944,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.2693,
+      "step": 11540
+    },
+    {
+      "epoch": 2.613844669561023,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.2253,
+      "step": 11545
+    },
+    {
+      "epoch": 2.615050651230101,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.2158,
+      "step": 11550
+    },
+    {
+      "epoch": 2.61625663289918,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0551,
+      "step": 11555
+    },
+    {
+      "epoch": 2.6174626145682587,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0063,
+      "step": 11560
+    },
+    {
+      "epoch": 2.6186685962373373,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.081,
+      "step": 11565
+    },
+    {
+      "epoch": 2.619874577906416,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.1738,
+      "step": 11570
+    },
+    {
+      "epoch": 2.6210805595754945,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0546,
+      "step": 11575
+    },
+    {
+      "epoch": 2.622286541244573,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1841,
+      "step": 11580
+    },
+    {
+      "epoch": 2.6234925229136516,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0773,
+      "step": 11585
+    },
+    {
+      "epoch": 2.6246985045827302,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0238,
+      "step": 11590
+    },
+    {
+      "epoch": 2.625904486251809,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1272,
+      "step": 11595
+    },
+    {
+      "epoch": 2.627110467920888,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0367,
+      "step": 11600
+    },
+    {
+      "epoch": 2.628316449589966,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.2925,
+      "step": 11605
+    },
+    {
+      "epoch": 2.629522431259045,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.1793,
+      "step": 11610
+    },
+    {
+      "epoch": 2.6307284129281236,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.203,
+      "step": 11615
+    },
+    {
+      "epoch": 2.6319343945972022,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0658,
+      "step": 11620
+    },
+    {
+      "epoch": 2.633140376266281,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.1987,
+      "step": 11625
+    },
+    {
+      "epoch": 2.6343463579353594,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.048,
+      "step": 11630
+    },
+    {
+      "epoch": 2.635552339604438,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.124,
+      "step": 11635
+    },
+    {
+      "epoch": 2.6367583212735166,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0372,
+      "step": 11640
+    },
+    {
+      "epoch": 2.637964302942595,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0557,
+      "step": 11645
+    },
+    {
+      "epoch": 2.639170284611674,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0853,
+      "step": 11650
+    },
+    {
+      "epoch": 2.6403762662807524,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1164,
+      "step": 11655
+    },
+    {
+      "epoch": 2.641582247949831,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.2068,
+      "step": 11660
+    },
+    {
+      "epoch": 2.64278822961891,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.2862,
+      "step": 11665
+    },
+    {
+      "epoch": 2.643994211287988,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.1415,
+      "step": 11670
+    },
+    {
+      "epoch": 2.645200192957067,
+      "grad_norm": 4.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.1748,
+      "step": 11675
+    },
+    {
+      "epoch": 2.646406174626146,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9935,
+      "step": 11680
+    },
+    {
+      "epoch": 2.6476121562952244,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0582,
+      "step": 11685
+    },
+    {
+      "epoch": 2.648818137964303,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0076,
+      "step": 11690
+    },
+    {
+      "epoch": 2.6500241196333816,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9872,
+      "step": 11695
+    },
+    {
+      "epoch": 2.65123010130246,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0618,
+      "step": 11700
+    },
+    {
+      "epoch": 2.6524360829715388,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.1482,
+      "step": 11705
+    },
+    {
+      "epoch": 2.6536420646406174,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1659,
+      "step": 11710
+    },
+    {
+      "epoch": 2.654848046309696,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0781,
+      "step": 11715
+    },
+    {
+      "epoch": 2.656054027978775,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1076,
+      "step": 11720
+    },
+    {
+      "epoch": 2.657260009647853,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0904,
+      "step": 11725
+    },
+    {
+      "epoch": 2.658465991316932,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.2026,
+      "step": 11730
+    },
+    {
+      "epoch": 2.6596719729860108,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.2472,
+      "step": 11735
+    },
+    {
+      "epoch": 2.6608779546550894,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.2508,
+      "step": 11740
+    },
+    {
+      "epoch": 2.662083936324168,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0978,
+      "step": 11745
+    },
+    {
+      "epoch": 2.6632899179932465,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.2511,
+      "step": 11750
+    },
+    {
+      "epoch": 2.664495899662325,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0748,
+      "step": 11755
+    },
+    {
+      "epoch": 2.6657018813314037,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0397,
+      "step": 11760
+    },
+    {
+      "epoch": 2.6669078630004823,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.2276,
+      "step": 11765
+    },
+    {
+      "epoch": 2.668113844669561,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1341,
+      "step": 11770
+    },
+    {
+      "epoch": 2.6693198263386395,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.1468,
+      "step": 11775
+    },
+    {
+      "epoch": 2.670525808007718,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.2332,
+      "step": 11780
+    },
+    {
+      "epoch": 2.671731789676797,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1936,
+      "step": 11785
+    },
+    {
+      "epoch": 2.6729377713458753,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.128,
+      "step": 11790
+    },
+    {
+      "epoch": 2.6741437530149543,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.989,
+      "step": 11795
+    },
+    {
+      "epoch": 2.675349734684033,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1033,
+      "step": 11800
+    },
+    {
+      "epoch": 2.6765557163531115,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1828,
+      "step": 11805
+    },
+    {
+      "epoch": 2.67776169802219,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8892,
+      "step": 11810
+    },
+    {
+      "epoch": 2.6789676796912687,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1484,
+      "step": 11815
+    },
+    {
+      "epoch": 2.6801736613603473,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1791,
+      "step": 11820
+    },
+    {
+      "epoch": 2.681379643029426,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0164,
+      "step": 11825
+    },
+    {
+      "epoch": 2.6825856246985045,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.1231,
+      "step": 11830
+    },
+    {
+      "epoch": 2.683791606367583,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0432,
+      "step": 11835
+    },
+    {
+      "epoch": 2.684997588036662,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.1452,
+      "step": 11840
+    },
+    {
+      "epoch": 2.6862035697057403,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.136,
+      "step": 11845
+    },
+    {
+      "epoch": 2.6874095513748193,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0017,
+      "step": 11850
+    },
+    {
+      "epoch": 2.688615533043898,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0565,
+      "step": 11855
+    },
+    {
+      "epoch": 2.6898215147129765,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0813,
+      "step": 11860
+    },
+    {
+      "epoch": 2.691027496382055,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.1229,
+      "step": 11865
+    },
+    {
+      "epoch": 2.6922334780511337,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0003,
+      "step": 11870
+    },
+    {
+      "epoch": 2.6934394597202123,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0572,
+      "step": 11875
+    },
+    {
+      "epoch": 2.694645441389291,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0974,
+      "step": 11880
+    },
+    {
+      "epoch": 2.6958514230583694,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.1187,
+      "step": 11885
+    },
+    {
+      "epoch": 2.697057404727448,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9405,
+      "step": 11890
+    },
+    {
+      "epoch": 2.6982633863965266,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1633,
+      "step": 11895
+    },
+    {
+      "epoch": 2.6994693680656052,
+      "grad_norm": 3.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0612,
+      "step": 11900
+    },
+    {
+      "epoch": 2.7006753497346843,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1518,
+      "step": 11905
+    },
+    {
+      "epoch": 2.7018813314037624,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0476,
+      "step": 11910
+    },
+    {
+      "epoch": 2.7030873130728414,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1044,
+      "step": 11915
+    },
+    {
+      "epoch": 2.70429329474192,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9564,
+      "step": 11920
+    },
+    {
+      "epoch": 2.7054992764109986,
+      "grad_norm": 1.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.1638,
+      "step": 11925
+    },
+    {
+      "epoch": 2.7067052580800772,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1089,
+      "step": 11930
+    },
+    {
+      "epoch": 2.707911239749156,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.1196,
+      "step": 11935
+    },
+    {
+      "epoch": 2.7091172214182344,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0696,
+      "step": 11940
+    },
+    {
+      "epoch": 2.710323203087313,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0352,
+      "step": 11945
+    },
+    {
+      "epoch": 2.7115291847563916,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0309,
+      "step": 11950
+    },
+    {
+      "epoch": 2.71273516642547,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0522,
+      "step": 11955
+    },
+    {
+      "epoch": 2.7139411480945492,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.1092,
+      "step": 11960
+    },
+    {
+      "epoch": 2.7151471297636274,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0894,
+      "step": 11965
+    },
+    {
+      "epoch": 2.7163531114327064,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.1613,
+      "step": 11970
+    },
+    {
+      "epoch": 2.717559093101785,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9797,
+      "step": 11975
+    },
+    {
+      "epoch": 2.7187650747708636,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.2203,
+      "step": 11980
+    },
+    {
+      "epoch": 2.719971056439942,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.1743,
+      "step": 11985
+    },
+    {
+      "epoch": 2.721177038109021,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0163,
+      "step": 11990
+    },
+    {
+      "epoch": 2.7223830197780994,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.2629,
+      "step": 11995
+    },
+    {
+      "epoch": 2.723589001447178,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0506,
+      "step": 12000
+    },
+    {
+      "epoch": 2.7247949831162566,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0748,
+      "step": 12005
+    },
+    {
+      "epoch": 2.726000964785335,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0374,
+      "step": 12010
+    },
+    {
+      "epoch": 2.7272069464544137,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9765,
+      "step": 12015
+    },
+    {
+      "epoch": 2.7284129281234923,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0611,
+      "step": 12020
+    },
+    {
+      "epoch": 2.7296189097925714,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0109,
+      "step": 12025
+    },
+    {
+      "epoch": 2.7308248914616495,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1325,
+      "step": 12030
+    },
+    {
+      "epoch": 2.7320308731307286,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1419,
+      "step": 12035
+    },
+    {
+      "epoch": 2.733236854799807,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0651,
+      "step": 12040
+    },
+    {
+      "epoch": 2.7344428364688858,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0466,
+      "step": 12045
+    },
+    {
+      "epoch": 2.7356488181379643,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0669,
+      "step": 12050
+    },
+    {
+      "epoch": 2.736854799807043,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1362,
+      "step": 12055
+    },
+    {
+      "epoch": 2.7380607814761215,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0632,
+      "step": 12060
+    },
+    {
+      "epoch": 2.7392667631452,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0498,
+      "step": 12065
+    },
+    {
+      "epoch": 2.7404727448142787,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0151,
+      "step": 12070
+    },
+    {
+      "epoch": 2.7416787264833573,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0584,
+      "step": 12075
+    },
+    {
+      "epoch": 2.7428847081524363,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.2482,
+      "step": 12080
+    },
+    {
+      "epoch": 2.7440906898215145,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1087,
+      "step": 12085
+    },
+    {
+      "epoch": 2.7452966714905935,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.1298,
+      "step": 12090
+    },
+    {
+      "epoch": 2.746502653159672,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0261,
+      "step": 12095
+    },
+    {
+      "epoch": 2.7477086348287507,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9859,
+      "step": 12100
+    },
+    {
+      "epoch": 2.7489146164978293,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8792,
+      "step": 12105
+    },
+    {
+      "epoch": 2.750120598166908,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.1922,
+      "step": 12110
+    },
+    {
+      "epoch": 2.7513265798359865,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.2514,
+      "step": 12115
+    },
+    {
+      "epoch": 2.752532561505065,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.2351,
+      "step": 12120
+    },
+    {
+      "epoch": 2.7537385431741437,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.2309,
+      "step": 12125
+    },
+    {
+      "epoch": 2.7549445248432223,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9893,
+      "step": 12130
+    },
+    {
+      "epoch": 2.756150506512301,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1744,
+      "step": 12135
+    },
+    {
+      "epoch": 2.7573564881813795,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0635,
+      "step": 12140
+    },
+    {
+      "epoch": 2.7585624698504585,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.2654,
+      "step": 12145
+    },
+    {
+      "epoch": 2.7597684515195366,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1203,
+      "step": 12150
+    },
+    {
+      "epoch": 2.7609744331886157,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0367,
+      "step": 12155
+    },
+    {
+      "epoch": 2.7621804148576943,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.1486,
+      "step": 12160
+    },
+    {
+      "epoch": 2.763386396526773,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0081,
+      "step": 12165
+    },
+    {
+      "epoch": 2.7645923781958515,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9865,
+      "step": 12170
+    },
+    {
+      "epoch": 2.76579835986493,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0075,
+      "step": 12175
+    },
+    {
+      "epoch": 2.7670043415340086,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0424,
+      "step": 12180
+    },
+    {
+      "epoch": 2.7682103232030872,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.2542,
+      "step": 12185
+    },
+    {
+      "epoch": 2.769416304872166,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.055,
+      "step": 12190
+    },
+    {
+      "epoch": 2.7706222865412444,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.045,
+      "step": 12195
+    },
+    {
+      "epoch": 2.7718282682103235,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0115,
+      "step": 12200
+    },
+    {
+      "epoch": 2.7730342498794016,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.989,
+      "step": 12205
+    },
+    {
+      "epoch": 2.7742402315484807,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8249,
+      "step": 12210
+    },
+    {
+      "epoch": 2.7754462132175592,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.3465,
+      "step": 12215
+    },
+    {
+      "epoch": 2.776652194886638,
+      "grad_norm": 3.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.3276,
+      "step": 12220
+    },
+    {
+      "epoch": 2.7778581765557164,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.1293,
+      "step": 12225
+    },
+    {
+      "epoch": 2.779064158224795,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0856,
+      "step": 12230
+    },
+    {
+      "epoch": 2.7802701398938736,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0849,
+      "step": 12235
+    },
+    {
+      "epoch": 2.781476121562952,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0482,
+      "step": 12240
+    },
+    {
+      "epoch": 2.782682103232031,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0615,
+      "step": 12245
+    },
+    {
+      "epoch": 2.7838880849011094,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.2041,
+      "step": 12250
+    },
+    {
+      "epoch": 2.785094066570188,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0746,
+      "step": 12255
+    },
+    {
+      "epoch": 2.7863000482392666,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.1682,
+      "step": 12260
+    },
+    {
+      "epoch": 2.7875060299083456,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.1719,
+      "step": 12265
+    },
+    {
+      "epoch": 2.7887120115774238,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.333,
+      "step": 12270
+    },
+    {
+      "epoch": 2.789917993246503,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0211,
+      "step": 12275
+    },
+    {
+      "epoch": 2.7911239749155814,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0016,
+      "step": 12280
+    },
+    {
+      "epoch": 2.79232995658466,
+      "grad_norm": 3.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.236,
+      "step": 12285
+    },
+    {
+      "epoch": 2.7935359382537386,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1864,
+      "step": 12290
+    },
+    {
+      "epoch": 2.794741919922817,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0912,
+      "step": 12295
+    },
+    {
+      "epoch": 2.7959479015918958,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1669,
+      "step": 12300
+    },
+    {
+      "epoch": 2.7971538832609744,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0738,
+      "step": 12305
+    },
+    {
+      "epoch": 2.798359864930053,
+      "grad_norm": 3.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1521,
+      "step": 12310
+    },
+    {
+      "epoch": 2.7995658465991315,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.1464,
+      "step": 12315
+    },
+    {
+      "epoch": 2.8007718282682106,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.1016,
+      "step": 12320
+    },
+    {
+      "epoch": 2.8019778099372887,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0554,
+      "step": 12325
+    },
+    {
+      "epoch": 2.8031837916063678,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1224,
+      "step": 12330
+    },
+    {
+      "epoch": 2.8043897732754464,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.1699,
+      "step": 12335
+    },
+    {
+      "epoch": 2.805595754944525,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0422,
+      "step": 12340
+    },
+    {
+      "epoch": 2.8068017366136035,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1896,
+      "step": 12345
+    },
+    {
+      "epoch": 2.808007718282682,
+      "grad_norm": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1451,
+      "step": 12350
+    },
+    {
+      "epoch": 2.8092136999517607,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0473,
+      "step": 12355
+    },
+    {
+      "epoch": 2.8104196816208393,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0793,
+      "step": 12360
+    },
+    {
+      "epoch": 2.811625663289918,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0617,
+      "step": 12365
+    },
+    {
+      "epoch": 2.8128316449589965,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.169,
+      "step": 12370
+    },
+    {
+      "epoch": 2.814037626628075,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9993,
+      "step": 12375
+    },
+    {
+      "epoch": 2.8152436082971537,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.1945,
+      "step": 12380
+    },
+    {
+      "epoch": 2.8164495899662327,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0516,
+      "step": 12385
+    },
+    {
+      "epoch": 2.817655571635311,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.1206,
+      "step": 12390
+    },
+    {
+      "epoch": 2.81886155330439,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1641,
+      "step": 12395
+    },
+    {
+      "epoch": 2.8200675349734685,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9757,
+      "step": 12400
+    },
+    {
+      "epoch": 2.821273516642547,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0516,
+      "step": 12405
+    },
+    {
+      "epoch": 2.8224794983116257,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.0335,
+      "step": 12410
+    },
+    {
+      "epoch": 2.8236854799807043,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0036,
+      "step": 12415
+    },
+    {
+      "epoch": 2.824891461649783,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.1023,
+      "step": 12420
+    },
+    {
+      "epoch": 2.8260974433188615,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0767,
+      "step": 12425
+    },
+    {
+      "epoch": 2.82730342498794,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.1463,
+      "step": 12430
+    },
+    {
+      "epoch": 2.8285094066570187,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1257,
+      "step": 12435
+    },
+    {
+      "epoch": 2.8297153883260977,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.1518,
+      "step": 12440
+    },
+    {
+      "epoch": 2.830921369995176,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0254,
+      "step": 12445
+    },
+    {
+      "epoch": 2.832127351664255,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.1223,
+      "step": 12450
+    },
+    {
+      "epoch": 2.8333333333333335,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0784,
+      "step": 12455
+    },
+    {
+      "epoch": 2.834539315002412,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1354,
+      "step": 12460
+    },
+    {
+      "epoch": 2.8357452966714907,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9273,
+      "step": 12465
+    },
+    {
+      "epoch": 2.8369512783405693,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.127,
+      "step": 12470
+    },
+    {
+      "epoch": 2.838157260009648,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.1316,
+      "step": 12475
+    },
+    {
+      "epoch": 2.8393632416787264,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.1644,
+      "step": 12480
+    },
+    {
+      "epoch": 2.840569223347805,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1627,
+      "step": 12485
+    },
+    {
+      "epoch": 2.8417752050168836,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9926,
+      "step": 12490
+    },
+    {
+      "epoch": 2.8429811866859622,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9624,
+      "step": 12495
+    },
+    {
+      "epoch": 2.844187168355041,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0651,
+      "step": 12500
+    },
+    {
+      "epoch": 2.84539315002412,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0333,
+      "step": 12505
+    },
+    {
+      "epoch": 2.846599131693198,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0222,
+      "step": 12510
+    },
+    {
+      "epoch": 2.847805113362277,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.2556,
+      "step": 12515
+    },
+    {
+      "epoch": 2.8490110950313556,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.046,
+      "step": 12520
+    },
+    {
+      "epoch": 2.8502170767004342,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.2979,
+      "step": 12525
+    },
+    {
+      "epoch": 2.851423058369513,
+      "grad_norm": 3.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1178,
+      "step": 12530
+    },
+    {
+      "epoch": 2.8526290400385914,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0731,
+      "step": 12535
+    },
+    {
+      "epoch": 2.85383502170767,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.0941,
+      "step": 12540
+    },
+    {
+      "epoch": 2.8550410033767486,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9681,
+      "step": 12545
+    },
+    {
+      "epoch": 2.856246985045827,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9995,
+      "step": 12550
+    },
+    {
+      "epoch": 2.857452966714906,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0625,
+      "step": 12555
+    },
+    {
+      "epoch": 2.858658948383985,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.1462,
+      "step": 12560
+    },
+    {
+      "epoch": 2.859864930053063,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.967,
+      "step": 12565
+    },
+    {
+      "epoch": 2.861070911722142,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9447,
+      "step": 12570
+    },
+    {
+      "epoch": 2.8622768933912206,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.1469,
+      "step": 12575
+    },
+    {
+      "epoch": 2.863482875060299,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.2723,
+      "step": 12580
+    },
+    {
+      "epoch": 2.864688856729378,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0287,
+      "step": 12585
+    },
+    {
+      "epoch": 2.8658948383984564,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1408,
+      "step": 12590
+    },
+    {
+      "epoch": 2.867100820067535,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0394,
+      "step": 12595
+    },
+    {
+      "epoch": 2.8683068017366136,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 4.0287,
+      "step": 12600
+    },
+    {
+      "epoch": 2.869512783405692,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0445,
+      "step": 12605
+    },
+    {
+      "epoch": 2.8707187650747708,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0422,
+      "step": 12610
+    },
+    {
+      "epoch": 2.8719247467438493,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0228,
+      "step": 12615
+    },
+    {
+      "epoch": 2.873130728412928,
+      "grad_norm": 3.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.1007,
+      "step": 12620
+    },
+    {
+      "epoch": 2.874336710082007,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0885,
+      "step": 12625
+    },
+    {
+      "epoch": 2.875542691751085,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9827,
+      "step": 12630
+    },
+    {
+      "epoch": 2.876748673420164,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.1984,
+      "step": 12635
+    },
+    {
+      "epoch": 2.8779546550892428,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9131,
+      "step": 12640
+    },
+    {
+      "epoch": 2.8791606367583213,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9683,
+      "step": 12645
+    },
+    {
+      "epoch": 2.8803666184274,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0378,
+      "step": 12650
+    },
+    {
+      "epoch": 2.8815726000964785,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.148,
+      "step": 12655
+    },
+    {
+      "epoch": 2.882778581765557,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8546,
+      "step": 12660
+    },
+    {
+      "epoch": 2.8839845634346357,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.193,
+      "step": 12665
+    },
+    {
+      "epoch": 2.8851905451037143,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1806,
+      "step": 12670
+    },
+    {
+      "epoch": 2.886396526772793,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0654,
+      "step": 12675
+    },
+    {
+      "epoch": 2.887602508441872,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.2687,
+      "step": 12680
+    },
+    {
+      "epoch": 2.88880849011095,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8384,
+      "step": 12685
+    },
+    {
+      "epoch": 2.890014471780029,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.2077,
+      "step": 12690
+    },
+    {
+      "epoch": 2.8912204534491077,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0338,
+      "step": 12695
+    },
+    {
+      "epoch": 2.8924264351181863,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0343,
+      "step": 12700
+    },
+    {
+      "epoch": 2.893632416787265,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1802,
+      "step": 12705
+    },
+    {
+      "epoch": 2.8948383984563435,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0843,
+      "step": 12710
+    },
+    {
+      "epoch": 2.896044380125422,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.014,
+      "step": 12715
+    },
+    {
+      "epoch": 2.8972503617945007,
+      "grad_norm": 4.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1161,
+      "step": 12720
+    },
+    {
+      "epoch": 2.8984563434635793,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9769,
+      "step": 12725
+    },
+    {
+      "epoch": 2.899662325132658,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.1166,
+      "step": 12730
+    },
+    {
+      "epoch": 2.9008683068017365,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0279,
+      "step": 12735
+    },
+    {
+      "epoch": 2.902074288470815,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1594,
+      "step": 12740
+    },
+    {
+      "epoch": 2.903280270139894,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.0105,
+      "step": 12745
+    },
+    {
+      "epoch": 2.9044862518089722,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.2134,
+      "step": 12750
+    },
+    {
+      "epoch": 2.9056922334780513,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0715,
+      "step": 12755
+    },
+    {
+      "epoch": 2.90689821514713,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1808,
+      "step": 12760
+    },
+    {
+      "epoch": 2.9081041968162085,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0269,
+      "step": 12765
+    },
+    {
+      "epoch": 2.909310178485287,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.167,
+      "step": 12770
+    },
+    {
+      "epoch": 2.9105161601543657,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0173,
+      "step": 12775
+    },
+    {
+      "epoch": 2.9117221418234442,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9264,
+      "step": 12780
+    },
+    {
+      "epoch": 2.912928123492523,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.042,
+      "step": 12785
+    },
+    {
+      "epoch": 2.9141341051616014,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0058,
+      "step": 12790
+    },
+    {
+      "epoch": 2.91534008683068,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0471,
+      "step": 12795
+    },
+    {
+      "epoch": 2.916546068499759,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9633,
+      "step": 12800
+    },
+    {
+      "epoch": 2.917752050168837,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1622,
+      "step": 12805
+    },
+    {
+      "epoch": 2.9189580318379162,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.131,
+      "step": 12810
+    },
+    {
+      "epoch": 2.920164013506995,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0153,
+      "step": 12815
+    },
+    {
+      "epoch": 2.9213699951760734,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1133,
+      "step": 12820
+    },
+    {
+      "epoch": 2.922575976845152,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.0067,
+      "step": 12825
+    },
+    {
+      "epoch": 2.9237819585142306,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.013,
+      "step": 12830
+    },
+    {
+      "epoch": 2.924987940183309,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9936,
+      "step": 12835
+    },
+    {
+      "epoch": 2.926193921852388,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.1483,
+      "step": 12840
+    },
+    {
+      "epoch": 2.9273999035214664,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0721,
+      "step": 12845
+    },
+    {
+      "epoch": 2.928605885190545,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.9175,
+      "step": 12850
+    },
+    {
+      "epoch": 2.9298118668596236,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.2047,
+      "step": 12855
+    },
+    {
+      "epoch": 2.931017848528702,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.1067,
+      "step": 12860
+    },
+    {
+      "epoch": 2.932223830197781,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0596,
+      "step": 12865
+    },
+    {
+      "epoch": 2.9334298118668594,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.1225,
+      "step": 12870
+    },
+    {
+      "epoch": 2.9346357935359384,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0444,
+      "step": 12875
+    },
+    {
+      "epoch": 2.935841775205017,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1243,
+      "step": 12880
+    },
+    {
+      "epoch": 2.9370477568740956,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9408,
+      "step": 12885
+    },
+    {
+      "epoch": 2.938253738543174,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.1349,
+      "step": 12890
+    },
+    {
+      "epoch": 2.9394597202122528,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0184,
+      "step": 12895
+    },
+    {
+      "epoch": 2.9406657018813314,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0827,
+      "step": 12900
+    },
+    {
+      "epoch": 2.94187168355041,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0646,
+      "step": 12905
+    },
+    {
+      "epoch": 2.9430776652194885,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.043,
+      "step": 12910
+    },
+    {
+      "epoch": 2.944283646888567,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1694,
+      "step": 12915
+    },
+    {
+      "epoch": 2.945489628557646,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0645,
+      "step": 12920
+    },
+    {
+      "epoch": 2.9466956102267243,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1264,
+      "step": 12925
+    },
+    {
+      "epoch": 2.9479015918958034,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9371,
+      "step": 12930
+    },
+    {
+      "epoch": 2.949107573564882,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0511,
+      "step": 12935
+    },
+    {
+      "epoch": 2.9503135552339605,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.171,
+      "step": 12940
+    },
+    {
+      "epoch": 2.951519536903039,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0845,
+      "step": 12945
+    },
+    {
+      "epoch": 2.9527255185721177,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9512,
+      "step": 12950
+    },
+    {
+      "epoch": 2.9539315002411963,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9474,
+      "step": 12955
+    },
+    {
+      "epoch": 2.955137481910275,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1066,
+      "step": 12960
+    },
+    {
+      "epoch": 2.9563434635793535,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9189,
+      "step": 12965
+    },
+    {
+      "epoch": 2.957549445248432,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0173,
+      "step": 12970
+    },
+    {
+      "epoch": 2.9587554269175107,
+      "grad_norm": 3.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.139,
+      "step": 12975
+    },
+    {
+      "epoch": 2.9599614085865893,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0827,
+      "step": 12980
+    },
+    {
+      "epoch": 2.9611673902556683,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9359,
+      "step": 12985
+    },
+    {
+      "epoch": 2.9623733719247465,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8856,
+      "step": 12990
+    },
+    {
+      "epoch": 2.9635793535938255,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1431,
+      "step": 12995
+    },
+    {
+      "epoch": 2.964785335262904,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1186,
+      "step": 13000
+    },
+    {
+      "epoch": 2.9659913169319827,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.1581,
+      "step": 13005
+    },
+    {
+      "epoch": 2.9671972986010613,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9509,
+      "step": 13010
+    },
+    {
+      "epoch": 2.96840328027014,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9609,
+      "step": 13015
+    },
+    {
+      "epoch": 2.9696092619392185,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9756,
+      "step": 13020
+    },
+    {
+      "epoch": 2.970815243608297,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0841,
+      "step": 13025
+    },
+    {
+      "epoch": 2.9720212252773757,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0665,
+      "step": 13030
+    },
+    {
+      "epoch": 2.9732272069464543,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1547,
+      "step": 13035
+    },
+    {
+      "epoch": 2.9744331886155333,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1022,
+      "step": 13040
+    },
+    {
+      "epoch": 2.9756391702846114,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9579,
+      "step": 13045
+    },
+    {
+      "epoch": 2.9768451519536905,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8809,
+      "step": 13050
+    },
+    {
+      "epoch": 2.978051133622769,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0751,
+      "step": 13055
+    },
+    {
+      "epoch": 2.9792571152918477,
+      "grad_norm": 3.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0962,
+      "step": 13060
+    },
+    {
+      "epoch": 2.9804630969609263,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.1125,
+      "step": 13065
+    },
+    {
+      "epoch": 2.981669078630005,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1935,
+      "step": 13070
+    },
+    {
+      "epoch": 2.9828750602990834,
+      "grad_norm": 3.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0026,
+      "step": 13075
+    },
+    {
+      "epoch": 2.984081041968162,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.015,
+      "step": 13080
+    },
+    {
+      "epoch": 2.9852870236372406,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9765,
+      "step": 13085
+    },
+    {
+      "epoch": 2.9864930053063192,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.921,
+      "step": 13090
+    },
+    {
+      "epoch": 2.987698986975398,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9594,
+      "step": 13095
+    },
+    {
+      "epoch": 2.9889049686444764,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.213,
+      "step": 13100
+    },
+    {
+      "epoch": 2.9901109503135554,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.053,
+      "step": 13105
+    },
+    {
+      "epoch": 2.9913169319826336,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0336,
+      "step": 13110
+    },
+    {
+      "epoch": 2.9925229136517126,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0058,
+      "step": 13115
+    },
+    {
+      "epoch": 2.9937288953207912,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1504,
+      "step": 13120
+    },
+    {
+      "epoch": 2.99493487698987,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9082,
+      "step": 13125
+    },
+    {
+      "epoch": 2.9961408586589484,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0186,
+      "step": 13130
+    },
+    {
+      "epoch": 2.997346840328027,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0363,
+      "step": 13135
+    },
+    {
+      "epoch": 2.9985528219971056,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.099,
+      "step": 13140
+    },
+    {
+      "epoch": 2.999758803666184,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.1806,
+      "step": 13145
+    },
+    {
+      "epoch": 3.000964785335263,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9605,
+      "step": 13150
+    },
+    {
+      "epoch": 3.0021707670043414,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9325,
+      "step": 13155
+    },
+    {
+      "epoch": 3.00337674867342,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1737,
+      "step": 13160
+    },
+    {
+      "epoch": 3.004582730342499,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0322,
+      "step": 13165
+    },
+    {
+      "epoch": 3.0057887120115776,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1062,
+      "step": 13170
+    },
+    {
+      "epoch": 3.006994693680656,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9885,
+      "step": 13175
+    },
+    {
+      "epoch": 3.008200675349735,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9667,
+      "step": 13180
+    },
+    {
+      "epoch": 3.0094066570188134,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.1605,
+      "step": 13185
+    },
+    {
+      "epoch": 3.010612638687892,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1717,
+      "step": 13190
+    },
+    {
+      "epoch": 3.0118186203569706,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.8992,
+      "step": 13195
+    },
+    {
+      "epoch": 3.013024602026049,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0766,
+      "step": 13200
+    },
+    {
+      "epoch": 3.0142305836951278,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9022,
+      "step": 13205
+    },
+    {
+      "epoch": 3.0154365653642063,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0784,
+      "step": 13210
+    },
+    {
+      "epoch": 3.016642547033285,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.1089,
+      "step": 13215
+    },
+    {
+      "epoch": 3.0178485287023635,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0501,
+      "step": 13220
+    },
+    {
+      "epoch": 3.0190545103714426,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9829,
+      "step": 13225
+    },
+    {
+      "epoch": 3.020260492040521,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9425,
+      "step": 13230
+    },
+    {
+      "epoch": 3.0214664737095998,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.101,
+      "step": 13235
+    },
+    {
+      "epoch": 3.0226724553786783,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9544,
+      "step": 13240
+    },
+    {
+      "epoch": 3.023878437047757,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0155,
+      "step": 13245
+    },
+    {
+      "epoch": 3.0250844187168355,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0865,
+      "step": 13250
+    },
+    {
+      "epoch": 3.026290400385914,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9694,
+      "step": 13255
+    },
+    {
+      "epoch": 3.0274963820549927,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.027,
+      "step": 13260
+    },
+    {
+      "epoch": 3.0287023637240713,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0497,
+      "step": 13265
+    },
+    {
+      "epoch": 3.02990834539315,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0664,
+      "step": 13270
+    },
+    {
+      "epoch": 3.0311143270622285,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9588,
+      "step": 13275
+    },
+    {
+      "epoch": 3.032320308731307,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.068,
+      "step": 13280
+    },
+    {
+      "epoch": 3.033526290400386,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9816,
+      "step": 13285
+    },
+    {
+      "epoch": 3.0347322720694647,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1299,
+      "step": 13290
+    },
+    {
+      "epoch": 3.0359382537385433,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8364,
+      "step": 13295
+    },
+    {
+      "epoch": 3.037144235407622,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.1571,
+      "step": 13300
+    },
+    {
+      "epoch": 3.0383502170767005,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0643,
+      "step": 13305
+    },
+    {
+      "epoch": 3.039556198745779,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0234,
+      "step": 13310
+    },
+    {
+      "epoch": 3.0407621804148577,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.1982,
+      "step": 13315
+    },
+    {
+      "epoch": 3.0419681620839363,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0241,
+      "step": 13320
+    },
+    {
+      "epoch": 3.043174143753015,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1117,
+      "step": 13325
+    },
+    {
+      "epoch": 3.0443801254220935,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0489,
+      "step": 13330
+    },
+    {
+      "epoch": 3.045586107091172,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.974,
+      "step": 13335
+    },
+    {
+      "epoch": 3.0467920887602506,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1092,
+      "step": 13340
+    },
+    {
+      "epoch": 3.0479980704293297,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0894,
+      "step": 13345
+    },
+    {
+      "epoch": 3.0492040520984083,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9961,
+      "step": 13350
+    },
+    {
+      "epoch": 3.050410033767487,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9915,
+      "step": 13355
+    },
+    {
+      "epoch": 3.0516160154365655,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.1449,
+      "step": 13360
+    },
+    {
+      "epoch": 3.052821997105644,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0539,
+      "step": 13365
+    },
+    {
+      "epoch": 3.0540279787747227,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.2002,
+      "step": 13370
+    },
+    {
+      "epoch": 3.0552339604438012,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.934,
+      "step": 13375
+    },
+    {
+      "epoch": 3.05643994211288,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0279,
+      "step": 13380
+    },
+    {
+      "epoch": 3.0576459237819584,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0714,
+      "step": 13385
+    },
+    {
+      "epoch": 3.058851905451037,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8936,
+      "step": 13390
+    },
+    {
+      "epoch": 3.0600578871201156,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9457,
+      "step": 13395
+    },
+    {
+      "epoch": 3.061263868789194,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8864,
+      "step": 13400
+    },
+    {
+      "epoch": 3.0624698504582732,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0401,
+      "step": 13405
+    },
+    {
+      "epoch": 3.063675832127352,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0836,
+      "step": 13410
+    },
+    {
+      "epoch": 3.0648818137964304,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9798,
+      "step": 13415
+    },
+    {
+      "epoch": 3.066087795465509,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.966,
+      "step": 13420
+    },
+    {
+      "epoch": 3.0672937771345876,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9688,
+      "step": 13425
+    },
+    {
+      "epoch": 3.068499758803666,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0855,
+      "step": 13430
+    },
+    {
+      "epoch": 3.069705740472745,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0465,
+      "step": 13435
+    },
+    {
+      "epoch": 3.0709117221418234,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9355,
+      "step": 13440
+    },
+    {
+      "epoch": 3.072117703810902,
+      "grad_norm": 3.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9578,
+      "step": 13445
+    },
+    {
+      "epoch": 3.0733236854799806,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1251,
+      "step": 13450
+    },
+    {
+      "epoch": 3.074529667149059,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0133,
+      "step": 13455
+    },
+    {
+      "epoch": 3.0757356488181378,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0401,
+      "step": 13460
+    },
+    {
+      "epoch": 3.076941630487217,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9156,
+      "step": 13465
+    },
+    {
+      "epoch": 3.0781476121562954,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0138,
+      "step": 13470
+    },
+    {
+      "epoch": 3.079353593825374,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8903,
+      "step": 13475
+    },
+    {
+      "epoch": 3.0805595754944526,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9927,
+      "step": 13480
+    },
+    {
+      "epoch": 3.081765557163531,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0072,
+      "step": 13485
+    },
+    {
+      "epoch": 3.0829715388326098,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0956,
+      "step": 13490
+    },
+    {
+      "epoch": 3.0841775205016884,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1261,
+      "step": 13495
+    },
+    {
+      "epoch": 3.085383502170767,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.157,
+      "step": 13500
+    },
+    {
+      "epoch": 3.0865894838398455,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9861,
+      "step": 13505
+    },
+    {
+      "epoch": 3.087795465508924,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.124,
+      "step": 13510
+    },
+    {
+      "epoch": 3.0890014471780027,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0408,
+      "step": 13515
+    },
+    {
+      "epoch": 3.0902074288470813,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8824,
+      "step": 13520
+    },
+    {
+      "epoch": 3.0914134105161604,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.057,
+      "step": 13525
+    },
+    {
+      "epoch": 3.092619392185239,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9654,
+      "step": 13530
+    },
+    {
+      "epoch": 3.0938253738543176,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0035,
+      "step": 13535
+    },
+    {
+      "epoch": 3.095031355523396,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9868,
+      "step": 13540
+    },
+    {
+      "epoch": 3.0962373371924747,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.1533,
+      "step": 13545
+    },
+    {
+      "epoch": 3.0974433188615533,
+      "grad_norm": 3.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0024,
+      "step": 13550
+    },
+    {
+      "epoch": 3.098649300530632,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.1052,
+      "step": 13555
+    },
+    {
+      "epoch": 3.0998552821997105,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1562,
+      "step": 13560
+    },
+    {
+      "epoch": 3.101061263868789,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9155,
+      "step": 13565
+    },
+    {
+      "epoch": 3.1022672455378677,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9797,
+      "step": 13570
+    },
+    {
+      "epoch": 3.1034732272069463,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1746,
+      "step": 13575
+    },
+    {
+      "epoch": 3.104679208876025,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0701,
+      "step": 13580
+    },
+    {
+      "epoch": 3.105885190545104,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9489,
+      "step": 13585
+    },
+    {
+      "epoch": 3.1070911722141825,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0601,
+      "step": 13590
+    },
+    {
+      "epoch": 3.108297153883261,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0415,
+      "step": 13595
+    },
+    {
+      "epoch": 3.1095031355523397,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0321,
+      "step": 13600
+    },
+    {
+      "epoch": 3.1107091172214183,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.048,
+      "step": 13605
+    },
+    {
+      "epoch": 3.111915098890497,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0143,
+      "step": 13610
+    },
+    {
+      "epoch": 3.1131210805595755,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.118,
+      "step": 13615
+    },
+    {
+      "epoch": 3.114327062228654,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9826,
+      "step": 13620
+    },
+    {
+      "epoch": 3.1155330438977327,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0644,
+      "step": 13625
+    },
+    {
+      "epoch": 3.1167390255668113,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.2254,
+      "step": 13630
+    },
+    {
+      "epoch": 3.11794500723589,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.028,
+      "step": 13635
+    },
+    {
+      "epoch": 3.1191509889049684,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0133,
+      "step": 13640
+    },
+    {
+      "epoch": 3.1203569705740475,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0251,
+      "step": 13645
+    },
+    {
+      "epoch": 3.121562952243126,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9699,
+      "step": 13650
+    },
+    {
+      "epoch": 3.1227689339122047,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9452,
+      "step": 13655
+    },
+    {
+      "epoch": 3.1239749155812833,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9581,
+      "step": 13660
+    },
+    {
+      "epoch": 3.125180897250362,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9676,
+      "step": 13665
+    },
+    {
+      "epoch": 3.1263868789194404,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1417,
+      "step": 13670
+    },
+    {
+      "epoch": 3.127592860588519,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9902,
+      "step": 13675
+    },
+    {
+      "epoch": 3.1287988422575976,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0673,
+      "step": 13680
+    },
+    {
+      "epoch": 3.1300048239266762,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.998,
+      "step": 13685
+    },
+    {
+      "epoch": 3.131210805595755,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9742,
+      "step": 13690
+    },
+    {
+      "epoch": 3.1324167872648334,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0822,
+      "step": 13695
+    },
+    {
+      "epoch": 3.133622768933912,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.1118,
+      "step": 13700
+    },
+    {
+      "epoch": 3.134828750602991,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0555,
+      "step": 13705
+    },
+    {
+      "epoch": 3.1360347322720696,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0973,
+      "step": 13710
+    },
+    {
+      "epoch": 3.1372407139411482,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9268,
+      "step": 13715
+    },
+    {
+      "epoch": 3.138446695610227,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.015,
+      "step": 13720
+    },
+    {
+      "epoch": 3.1396526772793054,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0813,
+      "step": 13725
+    },
+    {
+      "epoch": 3.140858658948384,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0813,
+      "step": 13730
+    },
+    {
+      "epoch": 3.1420646406174626,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.8976,
+      "step": 13735
+    },
+    {
+      "epoch": 3.143270622286541,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0243,
+      "step": 13740
+    },
+    {
+      "epoch": 3.14447660395562,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0255,
+      "step": 13745
+    },
+    {
+      "epoch": 3.1456825856246984,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0229,
+      "step": 13750
+    },
+    {
+      "epoch": 3.146888567293777,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0125,
+      "step": 13755
+    },
+    {
+      "epoch": 3.1480945489628556,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0281,
+      "step": 13760
+    },
+    {
+      "epoch": 3.1493005306319346,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9351,
+      "step": 13765
+    },
+    {
+      "epoch": 3.150506512301013,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9611,
+      "step": 13770
+    },
+    {
+      "epoch": 3.151712493970092,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8776,
+      "step": 13775
+    },
+    {
+      "epoch": 3.1529184756391704,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9645,
+      "step": 13780
+    },
+    {
+      "epoch": 3.154124457308249,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0643,
+      "step": 13785
+    },
+    {
+      "epoch": 3.1553304389773276,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1082,
+      "step": 13790
+    },
+    {
+      "epoch": 3.156536420646406,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0164,
+      "step": 13795
+    },
+    {
+      "epoch": 3.1577424023154848,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.1538,
+      "step": 13800
+    },
+    {
+      "epoch": 3.1589483839845633,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1769,
+      "step": 13805
+    },
+    {
+      "epoch": 3.160154365653642,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9665,
+      "step": 13810
+    },
+    {
+      "epoch": 3.1613603473227205,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8724,
+      "step": 13815
+    },
+    {
+      "epoch": 3.162566328991799,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8985,
+      "step": 13820
+    },
+    {
+      "epoch": 3.163772310660878,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1009,
+      "step": 13825
+    },
+    {
+      "epoch": 3.1649782923299568,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1008,
+      "step": 13830
+    },
+    {
+      "epoch": 3.1661842739990353,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0372,
+      "step": 13835
+    },
+    {
+      "epoch": 3.167390255668114,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.014,
+      "step": 13840
+    },
+    {
+      "epoch": 3.1685962373371925,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.8106,
+      "step": 13845
+    },
+    {
+      "epoch": 3.169802219006271,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0814,
+      "step": 13850
+    },
+    {
+      "epoch": 3.1710082006753497,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0285,
+      "step": 13855
+    },
+    {
+      "epoch": 3.1722141823444283,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9147,
+      "step": 13860
+    },
+    {
+      "epoch": 3.173420164013507,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.2018,
+      "step": 13865
+    },
+    {
+      "epoch": 3.1746261456825855,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0175,
+      "step": 13870
+    },
+    {
+      "epoch": 3.175832127351664,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0163,
+      "step": 13875
+    },
+    {
+      "epoch": 3.1770381090207427,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0386,
+      "step": 13880
+    },
+    {
+      "epoch": 3.1782440906898217,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9444,
+      "step": 13885
+    },
+    {
+      "epoch": 3.1794500723589003,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9977,
+      "step": 13890
+    },
+    {
+      "epoch": 3.180656054027979,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.1229,
+      "step": 13895
+    },
+    {
+      "epoch": 3.1818620356970575,
+      "grad_norm": 3.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0196,
+      "step": 13900
+    },
+    {
+      "epoch": 3.183068017366136,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.1247,
+      "step": 13905
+    },
+    {
+      "epoch": 3.1842739990352147,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9441,
+      "step": 13910
+    },
+    {
+      "epoch": 3.1854799807042933,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0142,
+      "step": 13915
+    },
+    {
+      "epoch": 3.186685962373372,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9364,
+      "step": 13920
+    },
+    {
+      "epoch": 3.1878919440424505,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0615,
+      "step": 13925
+    },
+    {
+      "epoch": 3.189097925711529,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0958,
+      "step": 13930
+    },
+    {
+      "epoch": 3.1903039073806077,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0952,
+      "step": 13935
+    },
+    {
+      "epoch": 3.1915098890496862,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.993,
+      "step": 13940
+    },
+    {
+      "epoch": 3.1927158707187653,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1895,
+      "step": 13945
+    },
+    {
+      "epoch": 3.193921852387844,
+      "grad_norm": 1.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0379,
+      "step": 13950
+    },
+    {
+      "epoch": 3.1951278340569225,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9688,
+      "step": 13955
+    },
+    {
+      "epoch": 3.196333815726001,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9644,
+      "step": 13960
+    },
+    {
+      "epoch": 3.1975397973950797,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9363,
+      "step": 13965
+    },
+    {
+      "epoch": 3.1987457790641582,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9434,
+      "step": 13970
+    },
+    {
+      "epoch": 3.199951760733237,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0571,
+      "step": 13975
+    },
+    {
+      "epoch": 3.2011577424023154,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0266,
+      "step": 13980
+    },
+    {
+      "epoch": 3.202363724071394,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.034,
+      "step": 13985
+    },
+    {
+      "epoch": 3.2035697057404726,
+      "grad_norm": 3.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1168,
+      "step": 13990
+    },
+    {
+      "epoch": 3.204775687409551,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.2014,
+      "step": 13995
+    },
+    {
+      "epoch": 3.20598166907863,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.016,
+      "step": 14000
+    },
+    {
+      "epoch": 3.207187650747709,
+      "grad_norm": 2.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.899,
+      "step": 14005
+    },
+    {
+      "epoch": 3.2083936324167874,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1745,
+      "step": 14010
+    },
+    {
+      "epoch": 3.209599614085866,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9857,
+      "step": 14015
+    },
+    {
+      "epoch": 3.2108055957549446,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.0834,
+      "step": 14020
+    },
+    {
+      "epoch": 3.212011577424023,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9924,
+      "step": 14025
+    },
+    {
+      "epoch": 3.213217559093102,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0381,
+      "step": 14030
+    },
+    {
+      "epoch": 3.2144235407621804,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9991,
+      "step": 14035
+    },
+    {
+      "epoch": 3.215629522431259,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0751,
+      "step": 14040
+    },
+    {
+      "epoch": 3.2168355041003376,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0358,
+      "step": 14045
+    },
+    {
+      "epoch": 3.218041485769416,
+      "grad_norm": 3.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0009,
+      "step": 14050
+    },
+    {
+      "epoch": 3.2192474674384948,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.994,
+      "step": 14055
+    },
+    {
+      "epoch": 3.2204534491075734,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0452,
+      "step": 14060
+    },
+    {
+      "epoch": 3.2216594307766524,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0047,
+      "step": 14065
+    },
+    {
+      "epoch": 3.222865412445731,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8072,
+      "step": 14070
+    },
+    {
+      "epoch": 3.2240713941148096,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.1151,
+      "step": 14075
+    },
+    {
+      "epoch": 3.225277375783888,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.9328,
+      "step": 14080
+    },
+    {
+      "epoch": 3.2264833574529668,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9196,
+      "step": 14085
+    },
+    {
+      "epoch": 3.2276893391220454,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0239,
+      "step": 14090
+    },
+    {
+      "epoch": 3.228895320791124,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9504,
+      "step": 14095
+    },
+    {
+      "epoch": 3.2301013024602026,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8764,
+      "step": 14100
+    },
+    {
+      "epoch": 3.231307284129281,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9942,
+      "step": 14105
+    },
+    {
+      "epoch": 3.2325132657983597,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9936,
+      "step": 14110
+    },
+    {
+      "epoch": 3.2337192474674383,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.1299,
+      "step": 14115
+    },
+    {
+      "epoch": 3.234925229136517,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.2354,
+      "step": 14120
+    },
+    {
+      "epoch": 3.236131210805596,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0096,
+      "step": 14125
+    },
+    {
+      "epoch": 3.2373371924746746,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9859,
+      "step": 14130
+    },
+    {
+      "epoch": 3.238543174143753,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9901,
+      "step": 14135
+    },
+    {
+      "epoch": 3.2397491558128317,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.048,
+      "step": 14140
+    },
+    {
+      "epoch": 3.2409551374819103,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8809,
+      "step": 14145
+    },
+    {
+      "epoch": 3.242161119150989,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.1033,
+      "step": 14150
+    },
+    {
+      "epoch": 3.2433671008200675,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0967,
+      "step": 14155
+    },
+    {
+      "epoch": 3.244573082489146,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9584,
+      "step": 14160
+    },
+    {
+      "epoch": 3.2457790641582247,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.0257,
+      "step": 14165
+    },
+    {
+      "epoch": 3.2469850458273033,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0908,
+      "step": 14170
+    },
+    {
+      "epoch": 3.248191027496382,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0343,
+      "step": 14175
+    },
+    {
+      "epoch": 3.2493970091654605,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.882,
+      "step": 14180
+    },
+    {
+      "epoch": 3.2506029908345395,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0731,
+      "step": 14185
+    },
+    {
+      "epoch": 3.251808972503618,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9578,
+      "step": 14190
+    },
+    {
+      "epoch": 3.2530149541726967,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0419,
+      "step": 14195
+    },
+    {
+      "epoch": 3.2542209358417753,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8306,
+      "step": 14200
+    },
+    {
+      "epoch": 3.255426917510854,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1088,
+      "step": 14205
+    },
+    {
+      "epoch": 3.2566328991799325,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9569,
+      "step": 14210
+    },
+    {
+      "epoch": 3.257838880849011,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0281,
+      "step": 14215
+    },
+    {
+      "epoch": 3.2590448625180897,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9602,
+      "step": 14220
+    },
+    {
+      "epoch": 3.2602508441871683,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0232,
+      "step": 14225
+    },
+    {
+      "epoch": 3.261456825856247,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.052,
+      "step": 14230
+    },
+    {
+      "epoch": 3.2626628075253254,
+      "grad_norm": 3.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9574,
+      "step": 14235
+    },
+    {
+      "epoch": 3.2638687891944045,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9721,
+      "step": 14240
+    },
+    {
+      "epoch": 3.2650747708634826,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0276,
+      "step": 14245
+    },
+    {
+      "epoch": 3.2662807525325617,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9899,
+      "step": 14250
+    },
+    {
+      "epoch": 3.2674867342016403,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.7672,
+      "step": 14255
+    },
+    {
+      "epoch": 3.268692715870719,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.1001,
+      "step": 14260
+    },
+    {
+      "epoch": 3.2698986975397974,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9353,
+      "step": 14265
+    },
+    {
+      "epoch": 3.271104679208876,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.8915,
+      "step": 14270
+    },
+    {
+      "epoch": 3.2723106608779546,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0175,
+      "step": 14275
+    },
+    {
+      "epoch": 3.2735166425470332,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9433,
+      "step": 14280
+    },
+    {
+      "epoch": 3.274722624216112,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8234,
+      "step": 14285
+    },
+    {
+      "epoch": 3.2759286058851904,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.011,
+      "step": 14290
+    },
+    {
+      "epoch": 3.277134587554269,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9454,
+      "step": 14295
+    },
+    {
+      "epoch": 3.2783405692233476,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1123,
+      "step": 14300
+    },
+    {
+      "epoch": 3.2795465508924266,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.2219,
+      "step": 14305
+    },
+    {
+      "epoch": 3.2807525325615052,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9955,
+      "step": 14310
+    },
+    {
+      "epoch": 3.281958514230584,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9424,
+      "step": 14315
+    },
+    {
+      "epoch": 3.2831644958996624,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9537,
+      "step": 14320
+    },
+    {
+      "epoch": 3.284370477568741,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0333,
+      "step": 14325
+    },
+    {
+      "epoch": 3.2855764592378196,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8545,
+      "step": 14330
+    },
+    {
+      "epoch": 3.286782440906898,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.083,
+      "step": 14335
+    },
+    {
+      "epoch": 3.287988422575977,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0703,
+      "step": 14340
+    },
+    {
+      "epoch": 3.2891944042450554,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9802,
+      "step": 14345
+    },
+    {
+      "epoch": 3.290400385914134,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0247,
+      "step": 14350
+    },
+    {
+      "epoch": 3.2916063675832126,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0461,
+      "step": 14355
+    },
+    {
+      "epoch": 3.2928123492522916,
+      "grad_norm": 3.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0102,
+      "step": 14360
+    },
+    {
+      "epoch": 3.2940183309213698,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8292,
+      "step": 14365
+    },
+    {
+      "epoch": 3.295224312590449,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.0706,
+      "step": 14370
+    },
+    {
+      "epoch": 3.2964302942595274,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.914,
+      "step": 14375
+    },
+    {
+      "epoch": 3.297636275928606,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0359,
+      "step": 14380
+    },
+    {
+      "epoch": 3.2988422575976846,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9523,
+      "step": 14385
+    },
+    {
+      "epoch": 3.300048239266763,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0287,
+      "step": 14390
+    },
+    {
+      "epoch": 3.3012542209358418,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9867,
+      "step": 14395
+    },
+    {
+      "epoch": 3.3024602026049203,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1293,
+      "step": 14400
+    },
+    {
+      "epoch": 3.303666184273999,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9522,
+      "step": 14405
+    },
+    {
+      "epoch": 3.3048721659430775,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8667,
+      "step": 14410
+    },
+    {
+      "epoch": 3.306078147612156,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9863,
+      "step": 14415
+    },
+    {
+      "epoch": 3.3072841292812347,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.013,
+      "step": 14420
+    },
+    {
+      "epoch": 3.3084901109503138,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9983,
+      "step": 14425
+    },
+    {
+      "epoch": 3.3096960926193923,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.972,
+      "step": 14430
+    },
+    {
+      "epoch": 3.310902074288471,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.8923,
+      "step": 14435
+    },
+    {
+      "epoch": 3.3121080559575495,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8711,
+      "step": 14440
+    },
+    {
+      "epoch": 3.313314037626628,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1087,
+      "step": 14445
+    },
+    {
+      "epoch": 3.3145200192957067,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0565,
+      "step": 14450
+    },
+    {
+      "epoch": 3.3157260009647853,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0258,
+      "step": 14455
+    },
+    {
+      "epoch": 3.316931982633864,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0745,
+      "step": 14460
+    },
+    {
+      "epoch": 3.3181379643029425,
+      "grad_norm": 3.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0424,
+      "step": 14465
+    },
+    {
+      "epoch": 3.319343945972021,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9648,
+      "step": 14470
+    },
+    {
+      "epoch": 3.3205499276410997,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9329,
+      "step": 14475
+    },
+    {
+      "epoch": 3.3217559093101787,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9133,
+      "step": 14480
+    },
+    {
+      "epoch": 3.322961890979257,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9903,
+      "step": 14485
+    },
+    {
+      "epoch": 3.324167872648336,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.1389,
+      "step": 14490
+    },
+    {
+      "epoch": 3.3253738543174145,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.96,
+      "step": 14495
+    },
+    {
+      "epoch": 3.326579835986493,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0746,
+      "step": 14500
+    },
+    {
+      "epoch": 3.3277858176555717,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0971,
+      "step": 14505
+    },
+    {
+      "epoch": 3.3289917993246503,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0437,
+      "step": 14510
+    },
+    {
+      "epoch": 3.330197780993729,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9961,
+      "step": 14515
+    },
+    {
+      "epoch": 3.3314037626628075,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0781,
+      "step": 14520
+    },
+    {
+      "epoch": 3.332609744331886,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.9975,
+      "step": 14525
+    },
+    {
+      "epoch": 3.3338157260009647,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9685,
+      "step": 14530
+    },
+    {
+      "epoch": 3.3350217076700432,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9129,
+      "step": 14535
+    },
+    {
+      "epoch": 3.336227689339122,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0438,
+      "step": 14540
+    },
+    {
+      "epoch": 3.337433671008201,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.067,
+      "step": 14545
+    },
+    {
+      "epoch": 3.3386396526772795,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0993,
+      "step": 14550
+    },
+    {
+      "epoch": 3.339845634346358,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0744,
+      "step": 14555
+    },
+    {
+      "epoch": 3.3410516160154367,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0717,
+      "step": 14560
+    },
+    {
+      "epoch": 3.3422575976845152,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0287,
+      "step": 14565
+    },
+    {
+      "epoch": 3.343463579353594,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9798,
+      "step": 14570
+    },
+    {
+      "epoch": 3.3446695610226724,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0063,
+      "step": 14575
+    },
+    {
+      "epoch": 3.345875542691751,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.1417,
+      "step": 14580
+    },
+    {
+      "epoch": 3.3470815243608296,
+      "grad_norm": 3.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1565,
+      "step": 14585
+    },
+    {
+      "epoch": 3.348287506029908,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0453,
+      "step": 14590
+    },
+    {
+      "epoch": 3.349493487698987,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0766,
+      "step": 14595
+    },
+    {
+      "epoch": 3.350699469368066,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9373,
+      "step": 14600
+    },
+    {
+      "epoch": 3.351905451037144,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9683,
+      "step": 14605
+    },
+    {
+      "epoch": 3.353111432706223,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1546,
+      "step": 14610
+    },
+    {
+      "epoch": 3.3543174143753016,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9953,
+      "step": 14615
+    },
+    {
+      "epoch": 3.35552339604438,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1147,
+      "step": 14620
+    },
+    {
+      "epoch": 3.356729377713459,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9063,
+      "step": 14625
+    },
+    {
+      "epoch": 3.3579353593825374,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9578,
+      "step": 14630
+    },
+    {
+      "epoch": 3.359141341051616,
+      "grad_norm": 3.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0816,
+      "step": 14635
+    },
+    {
+      "epoch": 3.3603473227206946,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0369,
+      "step": 14640
+    },
+    {
+      "epoch": 3.361553304389773,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9274,
+      "step": 14645
+    },
+    {
+      "epoch": 3.3627592860588518,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9866,
+      "step": 14650
+    },
+    {
+      "epoch": 3.3639652677279304,
+      "grad_norm": 3.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0215,
+      "step": 14655
+    },
+    {
+      "epoch": 3.365171249397009,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9415,
+      "step": 14660
+    },
+    {
+      "epoch": 3.366377231066088,
+      "grad_norm": 3.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8939,
+      "step": 14665
+    },
+    {
+      "epoch": 3.3675832127351666,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0955,
+      "step": 14670
+    },
+    {
+      "epoch": 3.368789194404245,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1219,
+      "step": 14675
+    },
+    {
+      "epoch": 3.3699951760733238,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1545,
+      "step": 14680
+    },
+    {
+      "epoch": 3.3712011577424024,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9588,
+      "step": 14685
+    },
+    {
+      "epoch": 3.372407139411481,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9333,
+      "step": 14690
+    },
+    {
+      "epoch": 3.3736131210805596,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9928,
+      "step": 14695
+    },
+    {
+      "epoch": 3.374819102749638,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0348,
+      "step": 14700
+    },
+    {
+      "epoch": 3.3760250844187167,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0069,
+      "step": 14705
+    },
+    {
+      "epoch": 3.3772310660877953,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9182,
+      "step": 14710
+    },
+    {
+      "epoch": 3.378437047756874,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8373,
+      "step": 14715
+    },
+    {
+      "epoch": 3.379643029425953,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0303,
+      "step": 14720
+    },
+    {
+      "epoch": 3.380849011095031,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.1722,
+      "step": 14725
+    },
+    {
+      "epoch": 3.38205499276411,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9155,
+      "step": 14730
+    },
+    {
+      "epoch": 3.3832609744331887,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9361,
+      "step": 14735
+    },
+    {
+      "epoch": 3.3844669561022673,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0036,
+      "step": 14740
+    },
+    {
+      "epoch": 3.385672937771346,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8562,
+      "step": 14745
+    },
+    {
+      "epoch": 3.3868789194404245,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0685,
+      "step": 14750
+    },
+    {
+      "epoch": 3.388084901109503,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.1758,
+      "step": 14755
+    },
+    {
+      "epoch": 3.3892908827785817,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0137,
+      "step": 14760
+    },
+    {
+      "epoch": 3.3904968644476603,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.978,
+      "step": 14765
+    },
+    {
+      "epoch": 3.391702846116739,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0286,
+      "step": 14770
+    },
+    {
+      "epoch": 3.3929088277858175,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0036,
+      "step": 14775
+    },
+    {
+      "epoch": 3.394114809454896,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9598,
+      "step": 14780
+    },
+    {
+      "epoch": 3.395320791123975,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9941,
+      "step": 14785
+    },
+    {
+      "epoch": 3.3965267727930537,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0702,
+      "step": 14790
+    },
+    {
+      "epoch": 3.3977327544621323,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.008,
+      "step": 14795
+    },
+    {
+      "epoch": 3.398938736131211,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9393,
+      "step": 14800
+    },
+    {
+      "epoch": 3.4001447178002895,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.1036,
+      "step": 14805
+    },
+    {
+      "epoch": 3.401350699469368,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.054,
+      "step": 14810
+    },
+    {
+      "epoch": 3.4025566811384467,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0621,
+      "step": 14815
+    },
+    {
+      "epoch": 3.4037626628075253,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.1669,
+      "step": 14820
+    },
+    {
+      "epoch": 3.404968644476604,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0668,
+      "step": 14825
+    },
+    {
+      "epoch": 3.4061746261456824,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.1343,
+      "step": 14830
+    },
+    {
+      "epoch": 3.407380607814761,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.1739,
+      "step": 14835
+    },
+    {
+      "epoch": 3.40858658948384,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9357,
+      "step": 14840
+    },
+    {
+      "epoch": 3.4097925711529182,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0191,
+      "step": 14845
+    },
+    {
+      "epoch": 3.4109985528219973,
+      "grad_norm": 3.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0613,
+      "step": 14850
+    },
+    {
+      "epoch": 3.412204534491076,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9014,
+      "step": 14855
+    },
+    {
+      "epoch": 3.4134105161601545,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9721,
+      "step": 14860
+    },
+    {
+      "epoch": 3.414616497829233,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9494,
+      "step": 14865
+    },
+    {
+      "epoch": 3.4158224794983116,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1432,
+      "step": 14870
+    },
+    {
+      "epoch": 3.4170284611673902,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8406,
+      "step": 14875
+    },
+    {
+      "epoch": 3.418234442836469,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9471,
+      "step": 14880
+    },
+    {
+      "epoch": 3.4194404245055474,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1186,
+      "step": 14885
+    },
+    {
+      "epoch": 3.420646406174626,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.9049,
+      "step": 14890
+    },
+    {
+      "epoch": 3.4218523878437046,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0118,
+      "step": 14895
+    },
+    {
+      "epoch": 3.423058369512783,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.1187,
+      "step": 14900
+    },
+    {
+      "epoch": 3.4242643511818622,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9578,
+      "step": 14905
+    },
+    {
+      "epoch": 3.425470332850941,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.998,
+      "step": 14910
+    },
+    {
+      "epoch": 3.4266763145200194,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9749,
+      "step": 14915
+    },
+    {
+      "epoch": 3.427882296189098,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9134,
+      "step": 14920
+    },
+    {
+      "epoch": 3.4290882778581766,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0398,
+      "step": 14925
+    },
+    {
+      "epoch": 3.430294259527255,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9159,
+      "step": 14930
+    },
+    {
+      "epoch": 3.431500241196334,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9447,
+      "step": 14935
+    },
+    {
+      "epoch": 3.4327062228654124,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0588,
+      "step": 14940
+    },
+    {
+      "epoch": 3.433912204534491,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.1365,
+      "step": 14945
+    },
+    {
+      "epoch": 3.4351181862035696,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.988,
+      "step": 14950
+    },
+    {
+      "epoch": 3.436324167872648,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9131,
+      "step": 14955
+    },
+    {
+      "epoch": 3.437530149541727,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0076,
+      "step": 14960
+    },
+    {
+      "epoch": 3.4387361312108053,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9779,
+      "step": 14965
+    },
+    {
+      "epoch": 3.4399421128798844,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0355,
+      "step": 14970
+    },
+    {
+      "epoch": 3.441148094548963,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0125,
+      "step": 14975
+    },
+    {
+      "epoch": 3.4423540762180416,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.029,
+      "step": 14980
+    },
+    {
+      "epoch": 3.44356005788712,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.0693,
+      "step": 14985
+    },
+    {
+      "epoch": 3.4447660395561988,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.0339,
+      "step": 14990
+    },
+    {
+      "epoch": 3.4459720212252773,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9148,
+      "step": 14995
+    },
+    {
+      "epoch": 3.447178002894356,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.1223,
+      "step": 15000
+    },
+    {
+      "epoch": 3.4483839845634345,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0484,
+      "step": 15005
+    },
+    {
+      "epoch": 3.449589966232513,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.905,
+      "step": 15010
+    },
+    {
+      "epoch": 3.4507959479015917,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1181,
+      "step": 15015
+    },
+    {
+      "epoch": 3.4520019295706703,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9647,
+      "step": 15020
+    },
+    {
+      "epoch": 3.4532079112397494,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0063,
+      "step": 15025
+    },
+    {
+      "epoch": 3.454413892908828,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9917,
+      "step": 15030
+    },
+    {
+      "epoch": 3.4556198745779065,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9416,
+      "step": 15035
+    },
+    {
+      "epoch": 3.456825856246985,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.003,
+      "step": 15040
+    },
+    {
+      "epoch": 3.4580318379160637,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9431,
+      "step": 15045
+    },
+    {
+      "epoch": 3.4592378195851423,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0297,
+      "step": 15050
+    },
+    {
+      "epoch": 3.460443801254221,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9578,
+      "step": 15055
+    },
+    {
+      "epoch": 3.4616497829232995,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.1545,
+      "step": 15060
+    },
+    {
+      "epoch": 3.462855764592378,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0698,
+      "step": 15065
+    },
+    {
+      "epoch": 3.4640617462614567,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0099,
+      "step": 15070
+    },
+    {
+      "epoch": 3.4652677279305353,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0312,
+      "step": 15075
+    },
+    {
+      "epoch": 3.4664737095996143,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.068,
+      "step": 15080
+    },
+    {
+      "epoch": 3.467679691268693,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.0546,
+      "step": 15085
+    },
+    {
+      "epoch": 3.4688856729377715,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1771,
+      "step": 15090
+    },
+    {
+      "epoch": 3.47009165460685,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0114,
+      "step": 15095
+    },
+    {
+      "epoch": 3.4712976362759287,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0987,
+      "step": 15100
+    },
+    {
+      "epoch": 3.4725036179450073,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0167,
+      "step": 15105
+    },
+    {
+      "epoch": 3.473709599614086,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1671,
+      "step": 15110
+    },
+    {
+      "epoch": 3.4749155812831645,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0783,
+      "step": 15115
+    },
+    {
+      "epoch": 3.476121562952243,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 4.112,
+      "step": 15120
+    },
+    {
+      "epoch": 3.4773275446213217,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9935,
+      "step": 15125
+    },
+    {
+      "epoch": 3.4785335262904002,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0511,
+      "step": 15130
+    },
+    {
+      "epoch": 3.479739507959479,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1281,
+      "step": 15135
+    },
+    {
+      "epoch": 3.4809454896285574,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9762,
+      "step": 15140
+    },
+    {
+      "epoch": 3.4821514712976365,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0346,
+      "step": 15145
+    },
+    {
+      "epoch": 3.483357452966715,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9322,
+      "step": 15150
+    },
+    {
+      "epoch": 3.4845634346357937,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9412,
+      "step": 15155
+    },
+    {
+      "epoch": 3.4857694163048722,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.953,
+      "step": 15160
+    },
+    {
+      "epoch": 3.486975397973951,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0236,
+      "step": 15165
+    },
+    {
+      "epoch": 3.4881813796430294,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0193,
+      "step": 15170
+    },
+    {
+      "epoch": 3.489387361312108,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9847,
+      "step": 15175
+    },
+    {
+      "epoch": 3.4905933429811866,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0215,
+      "step": 15180
+    },
+    {
+      "epoch": 3.491799324650265,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0822,
+      "step": 15185
+    },
+    {
+      "epoch": 3.493005306319344,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8903,
+      "step": 15190
+    },
+    {
+      "epoch": 3.4942112879884224,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0781,
+      "step": 15195
+    },
+    {
+      "epoch": 3.4954172696575014,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8788,
+      "step": 15200
+    },
+    {
+      "epoch": 3.49662325132658,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9371,
+      "step": 15205
+    },
+    {
+      "epoch": 3.4978292329956586,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0242,
+      "step": 15210
+    },
+    {
+      "epoch": 3.499035214664737,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9758,
+      "step": 15215
+    },
+    {
+      "epoch": 3.500241196333816,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9946,
+      "step": 15220
+    },
+    {
+      "epoch": 3.5014471780028944,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8467,
+      "step": 15225
+    },
+    {
+      "epoch": 3.502653159671973,
+      "grad_norm": 1.9609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9953,
+      "step": 15230
+    },
+    {
+      "epoch": 3.5038591413410516,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9511,
+      "step": 15235
+    },
+    {
+      "epoch": 3.50506512301013,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.082,
+      "step": 15240
+    },
+    {
+      "epoch": 3.5062711046792088,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0789,
+      "step": 15245
+    },
+    {
+      "epoch": 3.5074770863482874,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0445,
+      "step": 15250
+    },
+    {
+      "epoch": 3.5086830680173664,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.0717,
+      "step": 15255
+    },
+    {
+      "epoch": 3.5098890496864446,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.1007,
+      "step": 15260
+    },
+    {
+      "epoch": 3.5110950313555236,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0185,
+      "step": 15265
+    },
+    {
+      "epoch": 3.512301013024602,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0267,
+      "step": 15270
+    },
+    {
+      "epoch": 3.5135069946936808,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.2083,
+      "step": 15275
+    },
+    {
+      "epoch": 3.5147129763627594,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8944,
+      "step": 15280
+    },
+    {
+      "epoch": 3.515918958031838,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0898,
+      "step": 15285
+    },
+    {
+      "epoch": 3.5171249397009166,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0848,
+      "step": 15290
+    },
+    {
+      "epoch": 3.518330921369995,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9434,
+      "step": 15295
+    },
+    {
+      "epoch": 3.5195369030390737,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0369,
+      "step": 15300
+    },
+    {
+      "epoch": 3.5207428847081523,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.918,
+      "step": 15305
+    },
+    {
+      "epoch": 3.521948866377231,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.854,
+      "step": 15310
+    },
+    {
+      "epoch": 3.5231548480463095,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0014,
+      "step": 15315
+    },
+    {
+      "epoch": 3.5243608297153886,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0185,
+      "step": 15320
+    },
+    {
+      "epoch": 3.5255668113844667,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9638,
+      "step": 15325
+    },
+    {
+      "epoch": 3.5267727930535457,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.993,
+      "step": 15330
+    },
+    {
+      "epoch": 3.5279787747226243,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9725,
+      "step": 15335
+    },
+    {
+      "epoch": 3.529184756391703,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.0577,
+      "step": 15340
+    },
+    {
+      "epoch": 3.5303907380607815,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9415,
+      "step": 15345
+    },
+    {
+      "epoch": 3.53159671972986,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0072,
+      "step": 15350
+    },
+    {
+      "epoch": 3.5328027013989387,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8658,
+      "step": 15355
+    },
+    {
+      "epoch": 3.5340086830680173,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0158,
+      "step": 15360
+    },
+    {
+      "epoch": 3.535214664737096,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.933,
+      "step": 15365
+    },
+    {
+      "epoch": 3.5364206464061745,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9914,
+      "step": 15370
+    },
+    {
+      "epoch": 3.5376266280752535,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9102,
+      "step": 15375
+    },
+    {
+      "epoch": 3.5388326097443317,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9443,
+      "step": 15380
+    },
+    {
+      "epoch": 3.5400385914134107,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9349,
+      "step": 15385
+    },
+    {
+      "epoch": 3.5412445730824893,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9924,
+      "step": 15390
+    },
+    {
+      "epoch": 3.542450554751568,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.0462,
+      "step": 15395
+    },
+    {
+      "epoch": 3.5436565364206465,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9478,
+      "step": 15400
+    },
+    {
+      "epoch": 3.544862518089725,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9993,
+      "step": 15405
+    },
+    {
+      "epoch": 3.5460684997588037,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0022,
+      "step": 15410
+    },
+    {
+      "epoch": 3.5472744814278823,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9451,
+      "step": 15415
+    },
+    {
+      "epoch": 3.548480463096961,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1631,
+      "step": 15420
+    },
+    {
+      "epoch": 3.5496864447660395,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8508,
+      "step": 15425
+    },
+    {
+      "epoch": 3.550892426435118,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.007,
+      "step": 15430
+    },
+    {
+      "epoch": 3.5520984081041966,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1412,
+      "step": 15435
+    },
+    {
+      "epoch": 3.5533043897732757,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9812,
+      "step": 15440
+    },
+    {
+      "epoch": 3.554510371442354,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.1371,
+      "step": 15445
+    },
+    {
+      "epoch": 3.555716353111433,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.0788,
+      "step": 15450
+    },
+    {
+      "epoch": 3.5569223347805115,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9216,
+      "step": 15455
+    },
+    {
+      "epoch": 3.55812831644959,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8934,
+      "step": 15460
+    },
+    {
+      "epoch": 3.5593342981186686,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0998,
+      "step": 15465
+    },
+    {
+      "epoch": 3.5605402797877472,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0685,
+      "step": 15470
+    },
+    {
+      "epoch": 3.561746261456826,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9627,
+      "step": 15475
+    },
+    {
+      "epoch": 3.5629522431259044,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9114,
+      "step": 15480
+    },
+    {
+      "epoch": 3.564158224794983,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9346,
+      "step": 15485
+    },
+    {
+      "epoch": 3.5653642064640616,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8659,
+      "step": 15490
+    },
+    {
+      "epoch": 3.5665701881331406,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0312,
+      "step": 15495
+    },
+    {
+      "epoch": 3.567776169802219,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0441,
+      "step": 15500
+    },
+    {
+      "epoch": 3.568982151471298,
+      "grad_norm": 3.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.2124,
+      "step": 15505
+    },
+    {
+      "epoch": 3.5701881331403764,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8043,
+      "step": 15510
+    },
+    {
+      "epoch": 3.571394114809455,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1218,
+      "step": 15515
+    },
+    {
+      "epoch": 3.5726000964785336,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.1294,
+      "step": 15520
+    },
+    {
+      "epoch": 3.573806078147612,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8942,
+      "step": 15525
+    },
+    {
+      "epoch": 3.575012059816691,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9531,
+      "step": 15530
+    },
+    {
+      "epoch": 3.5762180414857694,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0333,
+      "step": 15535
+    },
+    {
+      "epoch": 3.577424023154848,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0552,
+      "step": 15540
+    },
+    {
+      "epoch": 3.5786300048239266,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0429,
+      "step": 15545
+    },
+    {
+      "epoch": 3.579835986493005,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0418,
+      "step": 15550
+    },
+    {
+      "epoch": 3.5810419681620838,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9592,
+      "step": 15555
+    },
+    {
+      "epoch": 3.582247949831163,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0796,
+      "step": 15560
+    },
+    {
+      "epoch": 3.583453931500241,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0507,
+      "step": 15565
+    },
+    {
+      "epoch": 3.58465991316932,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9588,
+      "step": 15570
+    },
+    {
+      "epoch": 3.5858658948383986,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.052,
+      "step": 15575
+    },
+    {
+      "epoch": 3.587071876507477,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9889,
+      "step": 15580
+    },
+    {
+      "epoch": 3.5882778581765558,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0662,
+      "step": 15585
+    },
+    {
+      "epoch": 3.5894838398456343,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0463,
+      "step": 15590
+    },
+    {
+      "epoch": 3.590689821514713,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.998,
+      "step": 15595
+    },
+    {
+      "epoch": 3.5918958031837915,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0381,
+      "step": 15600
+    },
+    {
+      "epoch": 3.59310178485287,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.155,
+      "step": 15605
+    },
+    {
+      "epoch": 3.5943077665219487,
+      "grad_norm": 3.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8984,
+      "step": 15610
+    },
+    {
+      "epoch": 3.5955137481910278,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9862,
+      "step": 15615
+    },
+    {
+      "epoch": 3.596719729860106,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.988,
+      "step": 15620
+    },
+    {
+      "epoch": 3.597925711529185,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8587,
+      "step": 15625
+    },
+    {
+      "epoch": 3.5991316931982635,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.094,
+      "step": 15630
+    },
+    {
+      "epoch": 3.600337674867342,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9377,
+      "step": 15635
+    },
+    {
+      "epoch": 3.6015436565364207,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0605,
+      "step": 15640
+    },
+    {
+      "epoch": 3.6027496382054993,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0728,
+      "step": 15645
+    },
+    {
+      "epoch": 3.603955619874578,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0383,
+      "step": 15650
+    },
+    {
+      "epoch": 3.6051616015436565,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.1243,
+      "step": 15655
+    },
+    {
+      "epoch": 3.606367583212735,
+      "grad_norm": 3.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9975,
+      "step": 15660
+    },
+    {
+      "epoch": 3.6075735648818137,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.918,
+      "step": 15665
+    },
+    {
+      "epoch": 3.6087795465508923,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0236,
+      "step": 15670
+    },
+    {
+      "epoch": 3.609985528219971,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1219,
+      "step": 15675
+    },
+    {
+      "epoch": 3.61119150988905,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0131,
+      "step": 15680
+    },
+    {
+      "epoch": 3.612397491558128,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1084,
+      "step": 15685
+    },
+    {
+      "epoch": 3.613603473227207,
+      "grad_norm": 3.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.1355,
+      "step": 15690
+    },
+    {
+      "epoch": 3.6148094548962857,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1247,
+      "step": 15695
+    },
+    {
+      "epoch": 3.6160154365653643,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0449,
+      "step": 15700
+    },
+    {
+      "epoch": 3.617221418234443,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.8251,
+      "step": 15705
+    },
+    {
+      "epoch": 3.6184273999035215,
+      "grad_norm": 3.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0023,
+      "step": 15710
+    },
+    {
+      "epoch": 3.6196333815726,
+      "grad_norm": 3.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.1111,
+      "step": 15715
+    },
+    {
+      "epoch": 3.6208393632416787,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.929,
+      "step": 15720
+    },
+    {
+      "epoch": 3.6220453449107572,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0471,
+      "step": 15725
+    },
+    {
+      "epoch": 3.623251326579836,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0213,
+      "step": 15730
+    },
+    {
+      "epoch": 3.624457308248915,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.8792,
+      "step": 15735
+    },
+    {
+      "epoch": 3.625663289917993,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0717,
+      "step": 15740
+    },
+    {
+      "epoch": 3.626869271587072,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9672,
+      "step": 15745
+    },
+    {
+      "epoch": 3.6280752532561507,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9882,
+      "step": 15750
+    },
+    {
+      "epoch": 3.6292812349252292,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.1598,
+      "step": 15755
+    },
+    {
+      "epoch": 3.630487216594308,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.1315,
+      "step": 15760
+    },
+    {
+      "epoch": 3.6316931982633864,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0041,
+      "step": 15765
+    },
+    {
+      "epoch": 3.632899179932465,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0705,
+      "step": 15770
+    },
+    {
+      "epoch": 3.6341051616015436,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9164,
+      "step": 15775
+    },
+    {
+      "epoch": 3.635311143270622,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9844,
+      "step": 15780
+    },
+    {
+      "epoch": 3.636517124939701,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9853,
+      "step": 15785
+    },
+    {
+      "epoch": 3.6377231066087794,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9188,
+      "step": 15790
+    },
+    {
+      "epoch": 3.638929088277858,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9599,
+      "step": 15795
+    },
+    {
+      "epoch": 3.640135069946937,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1056,
+      "step": 15800
+    },
+    {
+      "epoch": 3.641341051616015,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0439,
+      "step": 15805
+    },
+    {
+      "epoch": 3.642547033285094,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1838,
+      "step": 15810
+    },
+    {
+      "epoch": 3.643753014954173,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0717,
+      "step": 15815
+    },
+    {
+      "epoch": 3.6449589966232514,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0409,
+      "step": 15820
+    },
+    {
+      "epoch": 3.64616497829233,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.921,
+      "step": 15825
+    },
+    {
+      "epoch": 3.6473709599614086,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9315,
+      "step": 15830
+    },
+    {
+      "epoch": 3.648576941630487,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9507,
+      "step": 15835
+    },
+    {
+      "epoch": 3.6497829232995658,
+      "grad_norm": 3.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8639,
+      "step": 15840
+    },
+    {
+      "epoch": 3.6509889049686444,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9103,
+      "step": 15845
+    },
+    {
+      "epoch": 3.652194886637723,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0831,
+      "step": 15850
+    },
+    {
+      "epoch": 3.653400868306802,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0593,
+      "step": 15855
+    },
+    {
+      "epoch": 3.65460684997588,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.896,
+      "step": 15860
+    },
+    {
+      "epoch": 3.655812831644959,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1176,
+      "step": 15865
+    },
+    {
+      "epoch": 3.6570188133140378,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9909,
+      "step": 15870
+    },
+    {
+      "epoch": 3.6582247949831164,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0337,
+      "step": 15875
+    },
+    {
+      "epoch": 3.659430776652195,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.1702,
+      "step": 15880
+    },
+    {
+      "epoch": 3.6606367583212736,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1194,
+      "step": 15885
+    },
+    {
+      "epoch": 3.661842739990352,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9726,
+      "step": 15890
+    },
+    {
+      "epoch": 3.6630487216594307,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.1218,
+      "step": 15895
+    },
+    {
+      "epoch": 3.6642547033285093,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.037,
+      "step": 15900
+    },
+    {
+      "epoch": 3.665460684997588,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9541,
+      "step": 15905
+    },
+    {
+      "epoch": 3.6666666666666665,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.038,
+      "step": 15910
+    },
+    {
+      "epoch": 3.667872648335745,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9506,
+      "step": 15915
+    },
+    {
+      "epoch": 3.669078630004824,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1574,
+      "step": 15920
+    },
+    {
+      "epoch": 3.6702846116739023,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1581,
+      "step": 15925
+    },
+    {
+      "epoch": 3.6714905933429813,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0647,
+      "step": 15930
+    },
+    {
+      "epoch": 3.67269657501206,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9469,
+      "step": 15935
+    },
+    {
+      "epoch": 3.6739025566811385,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9963,
+      "step": 15940
+    },
+    {
+      "epoch": 3.675108538350217,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.996,
+      "step": 15945
+    },
+    {
+      "epoch": 3.6763145200192957,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0304,
+      "step": 15950
+    },
+    {
+      "epoch": 3.6775205016883743,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8884,
+      "step": 15955
+    },
+    {
+      "epoch": 3.678726483357453,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9658,
+      "step": 15960
+    },
+    {
+      "epoch": 3.6799324650265315,
+      "grad_norm": 3.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0913,
+      "step": 15965
+    },
+    {
+      "epoch": 3.68113844669561,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9585,
+      "step": 15970
+    },
+    {
+      "epoch": 3.682344428364689,
+      "grad_norm": 3.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0505,
+      "step": 15975
+    },
+    {
+      "epoch": 3.6835504100337673,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9165,
+      "step": 15980
+    },
+    {
+      "epoch": 3.6847563917028463,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.976,
+      "step": 15985
+    },
+    {
+      "epoch": 3.685962373371925,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.1032,
+      "step": 15990
+    },
+    {
+      "epoch": 3.6871683550410035,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9082,
+      "step": 15995
+    },
+    {
+      "epoch": 3.688374336710082,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9664,
+      "step": 16000
+    },
+    {
+      "epoch": 3.6895803183791607,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9856,
+      "step": 16005
+    },
+    {
+      "epoch": 3.6907863000482393,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9754,
+      "step": 16010
+    },
+    {
+      "epoch": 3.691992281717318,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9251,
+      "step": 16015
+    },
+    {
+      "epoch": 3.6931982633863965,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9983,
+      "step": 16020
+    },
+    {
+      "epoch": 3.694404245055475,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.878,
+      "step": 16025
+    },
+    {
+      "epoch": 3.6956102267245536,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0377,
+      "step": 16030
+    },
+    {
+      "epoch": 3.6968162083936322,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8615,
+      "step": 16035
+    },
+    {
+      "epoch": 3.6980221900627113,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.1154,
+      "step": 16040
+    },
+    {
+      "epoch": 3.6992281717317894,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9776,
+      "step": 16045
+    },
+    {
+      "epoch": 3.7004341534008685,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9946,
+      "step": 16050
+    },
+    {
+      "epoch": 3.701640135069947,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9427,
+      "step": 16055
+    },
+    {
+      "epoch": 3.7028461167390256,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9502,
+      "step": 16060
+    },
+    {
+      "epoch": 3.7040520984081042,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9252,
+      "step": 16065
+    },
+    {
+      "epoch": 3.705258080077183,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0952,
+      "step": 16070
+    },
+    {
+      "epoch": 3.7064640617462614,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.887,
+      "step": 16075
+    },
+    {
+      "epoch": 3.70767004341534,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.122,
+      "step": 16080
+    },
+    {
+      "epoch": 3.7088760250844186,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9845,
+      "step": 16085
+    },
+    {
+      "epoch": 3.710082006753497,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9335,
+      "step": 16090
+    },
+    {
+      "epoch": 3.7112879884225762,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9416,
+      "step": 16095
+    },
+    {
+      "epoch": 3.7124939700916544,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9335,
+      "step": 16100
+    },
+    {
+      "epoch": 3.7136999517607334,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9316,
+      "step": 16105
+    },
+    {
+      "epoch": 3.714905933429812,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0515,
+      "step": 16110
+    },
+    {
+      "epoch": 3.7161119150988906,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.0138,
+      "step": 16115
+    },
+    {
+      "epoch": 3.717317896767969,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9327,
+      "step": 16120
+    },
+    {
+      "epoch": 3.718523878437048,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.0349,
+      "step": 16125
+    },
+    {
+      "epoch": 3.7197298601061264,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.1848,
+      "step": 16130
+    },
+    {
+      "epoch": 3.720935841775205,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.898,
+      "step": 16135
+    },
+    {
+      "epoch": 3.7221418234442836,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1881,
+      "step": 16140
+    },
+    {
+      "epoch": 3.723347805113362,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.9333,
+      "step": 16145
+    },
+    {
+      "epoch": 3.7245537867824408,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9446,
+      "step": 16150
+    },
+    {
+      "epoch": 3.7257597684515193,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9377,
+      "step": 16155
+    },
+    {
+      "epoch": 3.7269657501205984,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9663,
+      "step": 16160
+    },
+    {
+      "epoch": 3.7281717317896765,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8904,
+      "step": 16165
+    },
+    {
+      "epoch": 3.7293777134587556,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9516,
+      "step": 16170
+    },
+    {
+      "epoch": 3.730583695127834,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9578,
+      "step": 16175
+    },
+    {
+      "epoch": 3.7317896767969128,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0739,
+      "step": 16180
+    },
+    {
+      "epoch": 3.7329956584659914,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9925,
+      "step": 16185
+    },
+    {
+      "epoch": 3.73420164013507,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9177,
+      "step": 16190
+    },
+    {
+      "epoch": 3.7354076218041485,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0116,
+      "step": 16195
+    },
+    {
+      "epoch": 3.736613603473227,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0218,
+      "step": 16200
+    },
+    {
+      "epoch": 3.7378195851423057,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.9417,
+      "step": 16205
+    },
+    {
+      "epoch": 3.7390255668113843,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9539,
+      "step": 16210
+    },
+    {
+      "epoch": 3.7402315484804634,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9554,
+      "step": 16215
+    },
+    {
+      "epoch": 3.7414375301495415,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9798,
+      "step": 16220
+    },
+    {
+      "epoch": 3.7426435118186205,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.0595,
+      "step": 16225
+    },
+    {
+      "epoch": 3.743849493487699,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0568,
+      "step": 16230
+    },
+    {
+      "epoch": 3.7450554751567777,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.0669,
+      "step": 16235
+    },
+    {
+      "epoch": 3.7462614568258563,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9098,
+      "step": 16240
+    },
+    {
+      "epoch": 3.747467438494935,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9163,
+      "step": 16245
+    },
+    {
+      "epoch": 3.7486734201640135,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.7712,
+      "step": 16250
+    },
+    {
+      "epoch": 3.749879401833092,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9362,
+      "step": 16255
+    },
+    {
+      "epoch": 3.7510853835021707,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.2692,
+      "step": 16260
+    },
+    {
+      "epoch": 3.7522913651712493,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.1035,
+      "step": 16265
+    },
+    {
+      "epoch": 3.753497346840328,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0493,
+      "step": 16270
+    },
+    {
+      "epoch": 3.7547033285094065,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0008,
+      "step": 16275
+    },
+    {
+      "epoch": 3.7559093101784855,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1084,
+      "step": 16280
+    },
+    {
+      "epoch": 3.7571152918475637,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9278,
+      "step": 16285
+    },
+    {
+      "epoch": 3.7583212735166427,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.1495,
+      "step": 16290
+    },
+    {
+      "epoch": 3.7595272551857213,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0115,
+      "step": 16295
+    },
+    {
+      "epoch": 3.7607332368548,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9386,
+      "step": 16300
+    },
+    {
+      "epoch": 3.7619392185238785,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0107,
+      "step": 16305
+    },
+    {
+      "epoch": 3.763145200192957,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9651,
+      "step": 16310
+    },
+    {
+      "epoch": 3.7643511818620357,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9544,
+      "step": 16315
+    },
+    {
+      "epoch": 3.7655571635311142,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8844,
+      "step": 16320
+    },
+    {
+      "epoch": 3.766763145200193,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9236,
+      "step": 16325
+    },
+    {
+      "epoch": 3.7679691268692714,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.1234,
+      "step": 16330
+    },
+    {
+      "epoch": 3.7691751085383505,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9655,
+      "step": 16335
+    },
+    {
+      "epoch": 3.7703810902074286,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0073,
+      "step": 16340
+    },
+    {
+      "epoch": 3.7715870718765077,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9181,
+      "step": 16345
+    },
+    {
+      "epoch": 3.7727930535455863,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.8662,
+      "step": 16350
+    },
+    {
+      "epoch": 3.773999035214665,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7226,
+      "step": 16355
+    },
+    {
+      "epoch": 3.7752050168837434,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.1805,
+      "step": 16360
+    },
+    {
+      "epoch": 3.776410998552822,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.1371,
+      "step": 16365
+    },
+    {
+      "epoch": 3.7776169802219006,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.129,
+      "step": 16370
+    },
+    {
+      "epoch": 3.778822961890979,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.1049,
+      "step": 16375
+    },
+    {
+      "epoch": 3.780028943560058,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8706,
+      "step": 16380
+    },
+    {
+      "epoch": 3.7812349252291364,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0336,
+      "step": 16385
+    },
+    {
+      "epoch": 3.782440906898215,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9422,
+      "step": 16390
+    },
+    {
+      "epoch": 3.7836468885672936,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0953,
+      "step": 16395
+    },
+    {
+      "epoch": 3.7848528702363726,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0049,
+      "step": 16400
+    },
+    {
+      "epoch": 3.7860588519054508,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0195,
+      "step": 16405
+    },
+    {
+      "epoch": 3.78726483357453,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0637,
+      "step": 16410
+    },
+    {
+      "epoch": 3.7884708152436084,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.3091,
+      "step": 16415
+    },
+    {
+      "epoch": 3.789676796912687,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.933,
+      "step": 16420
+    },
+    {
+      "epoch": 3.7908827785817656,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9491,
+      "step": 16425
+    },
+    {
+      "epoch": 3.792088760250844,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0612,
+      "step": 16430
+    },
+    {
+      "epoch": 3.7932947419199228,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.128,
+      "step": 16435
+    },
+    {
+      "epoch": 3.7945007235890014,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9772,
+      "step": 16440
+    },
+    {
+      "epoch": 3.79570670525808,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0806,
+      "step": 16445
+    },
+    {
+      "epoch": 3.7969126869271586,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9721,
+      "step": 16450
+    },
+    {
+      "epoch": 3.7981186685962376,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9053,
+      "step": 16455
+    },
+    {
+      "epoch": 3.7993246502653157,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.2064,
+      "step": 16460
+    },
+    {
+      "epoch": 3.8005306319343948,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9998,
+      "step": 16465
+    },
+    {
+      "epoch": 3.8017366136034734,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.93,
+      "step": 16470
+    },
+    {
+      "epoch": 3.802942595272552,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0474,
+      "step": 16475
+    },
+    {
+      "epoch": 3.8041485769416306,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0622,
+      "step": 16480
+    },
+    {
+      "epoch": 3.805354558610709,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0207,
+      "step": 16485
+    },
+    {
+      "epoch": 3.8065605402797877,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0581,
+      "step": 16490
+    },
+    {
+      "epoch": 3.8077665219488663,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0917,
+      "step": 16495
+    },
+    {
+      "epoch": 3.808972503617945,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0036,
+      "step": 16500
+    },
+    {
+      "epoch": 3.8101784852870235,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9413,
+      "step": 16505
+    },
+    {
+      "epoch": 3.811384466956102,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9802,
+      "step": 16510
+    },
+    {
+      "epoch": 3.8125904486251807,
+      "grad_norm": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0398,
+      "step": 16515
+    },
+    {
+      "epoch": 3.8137964302942597,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9177,
+      "step": 16520
+    },
+    {
+      "epoch": 3.815002411963338,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0421,
+      "step": 16525
+    },
+    {
+      "epoch": 3.816208393632417,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0367,
+      "step": 16530
+    },
+    {
+      "epoch": 3.8174143753014955,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9691,
+      "step": 16535
+    },
+    {
+      "epoch": 3.818620356970574,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0991,
+      "step": 16540
+    },
+    {
+      "epoch": 3.8198263386396527,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9303,
+      "step": 16545
+    },
+    {
+      "epoch": 3.8210323203087313,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8994,
+      "step": 16550
+    },
+    {
+      "epoch": 3.82223830197781,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9727,
+      "step": 16555
+    },
+    {
+      "epoch": 3.8234442836468885,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9319,
+      "step": 16560
+    },
+    {
+      "epoch": 3.824650265315967,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9565,
+      "step": 16565
+    },
+    {
+      "epoch": 3.8258562469850457,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9581,
+      "step": 16570
+    },
+    {
+      "epoch": 3.8270622286541247,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0812,
+      "step": 16575
+    },
+    {
+      "epoch": 3.828268210323203,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0111,
+      "step": 16580
+    },
+    {
+      "epoch": 3.829474191992282,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0788,
+      "step": 16585
+    },
+    {
+      "epoch": 3.8306801736613605,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9713,
+      "step": 16590
+    },
+    {
+      "epoch": 3.831886155330439,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0475,
+      "step": 16595
+    },
+    {
+      "epoch": 3.8330921369995177,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9336,
+      "step": 16600
+    },
+    {
+      "epoch": 3.8342981186685963,
+      "grad_norm": 3.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1285,
+      "step": 16605
+    },
+    {
+      "epoch": 3.835504100337675,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.7279,
+      "step": 16610
+    },
+    {
+      "epoch": 3.8367100820067535,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0651,
+      "step": 16615
+    },
+    {
+      "epoch": 3.837916063675832,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.066,
+      "step": 16620
+    },
+    {
+      "epoch": 3.8391220453449106,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0235,
+      "step": 16625
+    },
+    {
+      "epoch": 3.8403280270139892,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.1096,
+      "step": 16630
+    },
+    {
+      "epoch": 3.841534008683068,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9193,
+      "step": 16635
+    },
+    {
+      "epoch": 3.842739990352147,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8714,
+      "step": 16640
+    },
+    {
+      "epoch": 3.843945972021225,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0109,
+      "step": 16645
+    },
+    {
+      "epoch": 3.845151953690304,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8672,
+      "step": 16650
+    },
+    {
+      "epoch": 3.8463579353593826,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.923,
+      "step": 16655
+    },
+    {
+      "epoch": 3.8475639170284612,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 4.1278,
+      "step": 16660
+    },
+    {
+      "epoch": 3.84876989869754,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9864,
+      "step": 16665
+    },
+    {
+      "epoch": 3.8499758803666184,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.1395,
+      "step": 16670
+    },
+    {
+      "epoch": 3.851181862035697,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1347,
+      "step": 16675
+    },
+    {
+      "epoch": 3.8523878437047756,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0089,
+      "step": 16680
+    },
+    {
+      "epoch": 3.853593825373854,
+      "grad_norm": 4.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9326,
+      "step": 16685
+    },
+    {
+      "epoch": 3.854799807042933,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9616,
+      "step": 16690
+    },
+    {
+      "epoch": 3.856005788712012,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8703,
+      "step": 16695
+    },
+    {
+      "epoch": 3.85721177038109,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9703,
+      "step": 16700
+    },
+    {
+      "epoch": 3.858417752050169,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9828,
+      "step": 16705
+    },
+    {
+      "epoch": 3.8596237337192476,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9348,
+      "step": 16710
+    },
+    {
+      "epoch": 3.860829715388326,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8907,
+      "step": 16715
+    },
+    {
+      "epoch": 3.862035697057405,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9368,
+      "step": 16720
+    },
+    {
+      "epoch": 3.8632416787264834,
+      "grad_norm": 3.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.2211,
+      "step": 16725
+    },
+    {
+      "epoch": 3.864447660395562,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9402,
+      "step": 16730
+    },
+    {
+      "epoch": 3.8656536420646406,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.1418,
+      "step": 16735
+    },
+    {
+      "epoch": 3.866859623733719,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9347,
+      "step": 16740
+    },
+    {
+      "epoch": 3.8680656054027978,
+      "grad_norm": 3.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.917,
+      "step": 16745
+    },
+    {
+      "epoch": 3.8692715870718764,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.8988,
+      "step": 16750
+    },
+    {
+      "epoch": 3.870477568740955,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0261,
+      "step": 16755
+    },
+    {
+      "epoch": 3.871683550410034,
+      "grad_norm": 3.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9484,
+      "step": 16760
+    },
+    {
+      "epoch": 3.872889532079112,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9292,
+      "step": 16765
+    },
+    {
+      "epoch": 3.874095513748191,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0743,
+      "step": 16770
+    },
+    {
+      "epoch": 3.8753014954172698,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.8979,
+      "step": 16775
+    },
+    {
+      "epoch": 3.8765074770863484,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0227,
+      "step": 16780
+    },
+    {
+      "epoch": 3.877713458755427,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9069,
+      "step": 16785
+    },
+    {
+      "epoch": 3.8789194404245055,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.8504,
+      "step": 16790
+    },
+    {
+      "epoch": 3.880125422093584,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9777,
+      "step": 16795
+    },
+    {
+      "epoch": 3.8813314037626627,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0615,
+      "step": 16800
+    },
+    {
+      "epoch": 3.8825373854317413,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8637,
+      "step": 16805
+    },
+    {
+      "epoch": 3.88374336710082,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9371,
+      "step": 16810
+    },
+    {
+      "epoch": 3.884949348769899,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0929,
+      "step": 16815
+    },
+    {
+      "epoch": 3.886155330438977,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0039,
+      "step": 16820
+    },
+    {
+      "epoch": 3.887361312108056,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.2203,
+      "step": 16825
+    },
+    {
+      "epoch": 3.8885672937771347,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.7446,
+      "step": 16830
+    },
+    {
+      "epoch": 3.8897732754462133,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0843,
+      "step": 16835
+    },
+    {
+      "epoch": 3.890979257115292,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0489,
+      "step": 16840
+    },
+    {
+      "epoch": 3.8921852387843705,
+      "grad_norm": 3.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8601,
+      "step": 16845
+    },
+    {
+      "epoch": 3.893391220453449,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0869,
+      "step": 16850
+    },
+    {
+      "epoch": 3.8945972021225277,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.015,
+      "step": 16855
+    },
+    {
+      "epoch": 3.8958031837916063,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8737,
+      "step": 16860
+    },
+    {
+      "epoch": 3.897009165460685,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0627,
+      "step": 16865
+    },
+    {
+      "epoch": 3.8982151471297635,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8719,
+      "step": 16870
+    },
+    {
+      "epoch": 3.899421128798842,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0355,
+      "step": 16875
+    },
+    {
+      "epoch": 3.900627110467921,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9983,
+      "step": 16880
+    },
+    {
+      "epoch": 3.9018330921369992,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.968,
+      "step": 16885
+    },
+    {
+      "epoch": 3.9030390738060783,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.055,
+      "step": 16890
+    },
+    {
+      "epoch": 3.904245055475157,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0793,
+      "step": 16895
+    },
+    {
+      "epoch": 3.9054510371442355,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9698,
+      "step": 16900
+    },
+    {
+      "epoch": 3.906657018813314,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0943,
+      "step": 16905
+    },
+    {
+      "epoch": 3.9078630004823927,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9656,
+      "step": 16910
+    },
+    {
+      "epoch": 3.9090689821514712,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9591,
+      "step": 16915
+    },
+    {
+      "epoch": 3.91027496382055,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0096,
+      "step": 16920
+    },
+    {
+      "epoch": 3.9114809454896284,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9084,
+      "step": 16925
+    },
+    {
+      "epoch": 3.912686927158707,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8909,
+      "step": 16930
+    },
+    {
+      "epoch": 3.913892908827786,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9246,
+      "step": 16935
+    },
+    {
+      "epoch": 3.915098890496864,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9368,
+      "step": 16940
+    },
+    {
+      "epoch": 3.9163048721659433,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9595,
+      "step": 16945
+    },
+    {
+      "epoch": 3.917510853835022,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9792,
+      "step": 16950
+    },
+    {
+      "epoch": 3.9187168355041004,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0542,
+      "step": 16955
+    },
+    {
+      "epoch": 3.919922817173179,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9786,
+      "step": 16960
+    },
+    {
+      "epoch": 3.9211287988422576,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0183,
+      "step": 16965
+    },
+    {
+      "epoch": 3.922334780511336,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9496,
+      "step": 16970
+    },
+    {
+      "epoch": 3.923540762180415,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8776,
+      "step": 16975
+    },
+    {
+      "epoch": 3.9247467438494934,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9536,
+      "step": 16980
+    },
+    {
+      "epoch": 3.925952725518572,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0673,
+      "step": 16985
+    },
+    {
+      "epoch": 3.9271587071876506,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9777,
+      "step": 16990
+    },
+    {
+      "epoch": 3.928364688856729,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8375,
+      "step": 16995
+    },
+    {
+      "epoch": 3.929570670525808,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0203,
+      "step": 17000
+    },
+    {
+      "epoch": 3.9307766521948864,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1498,
+      "step": 17005
+    },
+    {
+      "epoch": 3.9319826338639654,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9497,
+      "step": 17010
+    },
+    {
+      "epoch": 3.933188615533044,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9592,
+      "step": 17015
+    },
+    {
+      "epoch": 3.9343945972021226,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0012,
+      "step": 17020
+    },
+    {
+      "epoch": 3.935600578871201,
+      "grad_norm": 5.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9361,
+      "step": 17025
+    },
+    {
+      "epoch": 3.9368065605402798,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0225,
+      "step": 17030
+    },
+    {
+      "epoch": 3.9380125422093584,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9514,
+      "step": 17035
+    },
+    {
+      "epoch": 3.939218523878437,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9751,
+      "step": 17040
+    },
+    {
+      "epoch": 3.9404245055475156,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0302,
+      "step": 17045
+    },
+    {
+      "epoch": 3.941630487216594,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9687,
+      "step": 17050
+    },
+    {
+      "epoch": 3.942836468885673,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9369,
+      "step": 17055
+    },
+    {
+      "epoch": 3.9440424505547513,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0217,
+      "step": 17060
+    },
+    {
+      "epoch": 3.9452484322238304,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0357,
+      "step": 17065
+    },
+    {
+      "epoch": 3.946454413892909,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9942,
+      "step": 17070
+    },
+    {
+      "epoch": 3.9476603955619876,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9284,
+      "step": 17075
+    },
+    {
+      "epoch": 3.948866377231066,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.9487,
+      "step": 17080
+    },
+    {
+      "epoch": 3.9500723589001447,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9801,
+      "step": 17085
+    },
+    {
+      "epoch": 3.9512783405692233,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1226,
+      "step": 17090
+    },
+    {
+      "epoch": 3.952484322238302,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.844,
+      "step": 17095
+    },
+    {
+      "epoch": 3.9536903039073805,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8442,
+      "step": 17100
+    },
+    {
+      "epoch": 3.954896285576459,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.073,
+      "step": 17105
+    },
+    {
+      "epoch": 3.9561022672455377,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.8494,
+      "step": 17110
+    },
+    {
+      "epoch": 3.9573082489146163,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8586,
+      "step": 17115
+    },
+    {
+      "epoch": 3.9585142305836953,
+      "grad_norm": 3.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9582,
+      "step": 17120
+    },
+    {
+      "epoch": 3.9597202122527735,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.1137,
+      "step": 17125
+    },
+    {
+      "epoch": 3.9609261939218525,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.7992,
+      "step": 17130
+    },
+    {
+      "epoch": 3.962132175590931,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8212,
+      "step": 17135
+    },
+    {
+      "epoch": 3.9633381572600097,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.961,
+      "step": 17140
+    },
+    {
+      "epoch": 3.9645441389290883,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.0676,
+      "step": 17145
+    },
+    {
+      "epoch": 3.965750120598167,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 4.1459,
+      "step": 17150
+    },
+    {
+      "epoch": 3.9669561022672455,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8767,
+      "step": 17155
+    },
+    {
+      "epoch": 3.968162083936324,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8575,
+      "step": 17160
+    },
+    {
+      "epoch": 3.9693680656054027,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9358,
+      "step": 17165
+    },
+    {
+      "epoch": 3.9705740472744813,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8776,
+      "step": 17170
+    },
+    {
+      "epoch": 3.9717800289435603,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0366,
+      "step": 17175
+    },
+    {
+      "epoch": 3.9729860106126385,
+      "grad_norm": 3.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0652,
+      "step": 17180
+    },
+    {
+      "epoch": 3.9741919922817175,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.0154,
+      "step": 17185
+    },
+    {
+      "epoch": 3.975397973950796,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8953,
+      "step": 17190
+    },
+    {
+      "epoch": 3.9766039556198747,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8697,
+      "step": 17195
+    },
+    {
+      "epoch": 3.9778099372889533,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9085,
+      "step": 17200
+    },
+    {
+      "epoch": 3.979015918958032,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0601,
+      "step": 17205
+    },
+    {
+      "epoch": 3.9802219006271105,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0051,
+      "step": 17210
+    },
+    {
+      "epoch": 3.981427882296189,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.096,
+      "step": 17215
+    },
+    {
+      "epoch": 3.9826338639652676,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.952,
+      "step": 17220
+    },
+    {
+      "epoch": 3.9838398456343462,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9228,
+      "step": 17225
+    },
+    {
+      "epoch": 3.985045827303425,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.893,
+      "step": 17230
+    },
+    {
+      "epoch": 3.9862518089725034,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.814,
+      "step": 17235
+    },
+    {
+      "epoch": 3.9874577906415825,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8744,
+      "step": 17240
+    },
+    {
+      "epoch": 3.9886637723106606,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.1822,
+      "step": 17245
+    },
+    {
+      "epoch": 3.9898697539797396,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.9633,
+      "step": 17250
+    },
+    {
+      "epoch": 3.9910757356488182,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8521,
+      "step": 17255
+    },
+    {
+      "epoch": 3.992281717317897,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9787,
+      "step": 17260
+    },
+    {
+      "epoch": 3.9934876989869754,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.048,
+      "step": 17265
+    },
+    {
+      "epoch": 3.994693680656054,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8688,
+      "step": 17270
+    },
+    {
+      "epoch": 3.9958996623251326,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9372,
+      "step": 17275
+    },
+    {
+      "epoch": 3.997105643994211,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9689,
+      "step": 17280
+    },
+    {
+      "epoch": 3.99831162566329,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9447,
+      "step": 17285
+    },
+    {
+      "epoch": 3.9995176073323684,
+      "grad_norm": 4.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.1225,
+      "step": 17290
+    },
+    {
+      "epoch": 4.000723589001447,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.8388,
+      "step": 17295
+    },
+    {
+      "epoch": 4.001929570670526,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.8877,
+      "step": 17300
+    },
+    {
+      "epoch": 4.003135552339605,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0602,
+      "step": 17305
+    },
+    {
+      "epoch": 4.004341534008683,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0037,
+      "step": 17310
+    },
+    {
+      "epoch": 4.005547515677762,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0028,
+      "step": 17315
+    },
+    {
+      "epoch": 4.00675349734684,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9476,
+      "step": 17320
+    },
+    {
+      "epoch": 4.007959479015919,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.8961,
+      "step": 17325
+    },
+    {
+      "epoch": 4.009165460684998,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9938,
+      "step": 17330
+    },
+    {
+      "epoch": 4.010371442354076,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1277,
+      "step": 17335
+    },
+    {
+      "epoch": 4.011577424023155,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.885,
+      "step": 17340
+    },
+    {
+      "epoch": 4.012783405692233,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9108,
+      "step": 17345
+    },
+    {
+      "epoch": 4.013989387361312,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8904,
+      "step": 17350
+    },
+    {
+      "epoch": 4.0151953690303905,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9804,
+      "step": 17355
+    },
+    {
+      "epoch": 4.01640135069947,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.017,
+      "step": 17360
+    },
+    {
+      "epoch": 4.017607332368548,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9892,
+      "step": 17365
+    },
+    {
+      "epoch": 4.018813314037627,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8946,
+      "step": 17370
+    },
+    {
+      "epoch": 4.020019295706705,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8947,
+      "step": 17375
+    },
+    {
+      "epoch": 4.021225277375784,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.957,
+      "step": 17380
+    },
+    {
+      "epoch": 4.022431259044862,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9355,
+      "step": 17385
+    },
+    {
+      "epoch": 4.023637240713941,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8422,
+      "step": 17390
+    },
+    {
+      "epoch": 4.02484322238302,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0214,
+      "step": 17395
+    },
+    {
+      "epoch": 4.026049204052098,
+      "grad_norm": 3.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9652,
+      "step": 17400
+    },
+    {
+      "epoch": 4.027255185721177,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9663,
+      "step": 17405
+    },
+    {
+      "epoch": 4.0284611673902555,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.96,
+      "step": 17410
+    },
+    {
+      "epoch": 4.0296671490593345,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9893,
+      "step": 17415
+    },
+    {
+      "epoch": 4.030873130728413,
+      "grad_norm": 3.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8778,
+      "step": 17420
+    },
+    {
+      "epoch": 4.032079112397492,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9751,
+      "step": 17425
+    },
+    {
+      "epoch": 4.03328509406657,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.934,
+      "step": 17430
+    },
+    {
+      "epoch": 4.034491075735649,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9501,
+      "step": 17435
+    },
+    {
+      "epoch": 4.035697057404727,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8527,
+      "step": 17440
+    },
+    {
+      "epoch": 4.036903039073806,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0261,
+      "step": 17445
+    },
+    {
+      "epoch": 4.038109020742885,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0066,
+      "step": 17450
+    },
+    {
+      "epoch": 4.039315002411963,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9134,
+      "step": 17455
+    },
+    {
+      "epoch": 4.040520984081042,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0793,
+      "step": 17460
+    },
+    {
+      "epoch": 4.0417269657501205,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9565,
+      "step": 17465
+    },
+    {
+      "epoch": 4.0429329474191995,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0828,
+      "step": 17470
+    },
+    {
+      "epoch": 4.044138929088278,
+      "grad_norm": 3.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9935,
+      "step": 17475
+    },
+    {
+      "epoch": 4.045344910757357,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8752,
+      "step": 17480
+    },
+    {
+      "epoch": 4.046550892426435,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0542,
+      "step": 17485
+    },
+    {
+      "epoch": 4.047756874095514,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0809,
+      "step": 17490
+    },
+    {
+      "epoch": 4.048962855764592,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8445,
+      "step": 17495
+    },
+    {
+      "epoch": 4.050168837433671,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8905,
+      "step": 17500
+    },
+    {
+      "epoch": 4.051374819102749,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0455,
+      "step": 17505
+    },
+    {
+      "epoch": 4.052580800771828,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9642,
+      "step": 17510
+    },
+    {
+      "epoch": 4.053786782440907,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1512,
+      "step": 17515
+    },
+    {
+      "epoch": 4.054992764109985,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9247,
+      "step": 17520
+    },
+    {
+      "epoch": 4.0561987457790645,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9222,
+      "step": 17525
+    },
+    {
+      "epoch": 4.057404727448143,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0195,
+      "step": 17530
+    },
+    {
+      "epoch": 4.058610709117222,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8717,
+      "step": 17535
+    },
+    {
+      "epoch": 4.0598166907863,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.7809,
+      "step": 17540
+    },
+    {
+      "epoch": 4.061022672455379,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.807,
+      "step": 17545
+    },
+    {
+      "epoch": 4.062228654124457,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9341,
+      "step": 17550
+    },
+    {
+      "epoch": 4.063434635793536,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.016,
+      "step": 17555
+    },
+    {
+      "epoch": 4.064640617462614,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.952,
+      "step": 17560
+    },
+    {
+      "epoch": 4.065846599131693,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8231,
+      "step": 17565
+    },
+    {
+      "epoch": 4.067052580800772,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9479,
+      "step": 17570
+    },
+    {
+      "epoch": 4.06825856246985,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9743,
+      "step": 17575
+    },
+    {
+      "epoch": 4.069464544138929,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9869,
+      "step": 17580
+    },
+    {
+      "epoch": 4.070670525808008,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8977,
+      "step": 17585
+    },
+    {
+      "epoch": 4.071876507477087,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8267,
+      "step": 17590
+    },
+    {
+      "epoch": 4.073082489146165,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0622,
+      "step": 17595
+    },
+    {
+      "epoch": 4.074288470815244,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9417,
+      "step": 17600
+    },
+    {
+      "epoch": 4.075494452484322,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9521,
+      "step": 17605
+    },
+    {
+      "epoch": 4.076700434153401,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.934,
+      "step": 17610
+    },
+    {
+      "epoch": 4.077906415822479,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9232,
+      "step": 17615
+    },
+    {
+      "epoch": 4.079112397491558,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.7429,
+      "step": 17620
+    },
+    {
+      "epoch": 4.080318379160636,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.8876,
+      "step": 17625
+    },
+    {
+      "epoch": 4.081524360829715,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9486,
+      "step": 17630
+    },
+    {
+      "epoch": 4.082730342498794,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0074,
+      "step": 17635
+    },
+    {
+      "epoch": 4.0839363241678726,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0851,
+      "step": 17640
+    },
+    {
+      "epoch": 4.085142305836952,
+      "grad_norm": 3.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0495,
+      "step": 17645
+    },
+    {
+      "epoch": 4.08634828750603,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9278,
+      "step": 17650
+    },
+    {
+      "epoch": 4.087554269175109,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0431,
+      "step": 17655
+    },
+    {
+      "epoch": 4.088760250844187,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9829,
+      "step": 17660
+    },
+    {
+      "epoch": 4.089966232513266,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8358,
+      "step": 17665
+    },
+    {
+      "epoch": 4.091172214182344,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9264,
+      "step": 17670
+    },
+    {
+      "epoch": 4.092378195851423,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8923,
+      "step": 17675
+    },
+    {
+      "epoch": 4.093584177520501,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9424,
+      "step": 17680
+    },
+    {
+      "epoch": 4.09479015918958,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.927,
+      "step": 17685
+    },
+    {
+      "epoch": 4.095996140858659,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0665,
+      "step": 17690
+    },
+    {
+      "epoch": 4.0972021225277375,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8743,
+      "step": 17695
+    },
+    {
+      "epoch": 4.098408104196817,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0551,
+      "step": 17700
+    },
+    {
+      "epoch": 4.099614085865895,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1643,
+      "step": 17705
+    },
+    {
+      "epoch": 4.100820067534974,
+      "grad_norm": 3.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8589,
+      "step": 17710
+    },
+    {
+      "epoch": 4.102026049204052,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8367,
+      "step": 17715
+    },
+    {
+      "epoch": 4.103232030873131,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0369,
+      "step": 17720
+    },
+    {
+      "epoch": 4.104438012542209,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0402,
+      "step": 17725
+    },
+    {
+      "epoch": 4.105643994211288,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9079,
+      "step": 17730
+    },
+    {
+      "epoch": 4.106849975880366,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9417,
+      "step": 17735
+    },
+    {
+      "epoch": 4.108055957549445,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9048,
+      "step": 17740
+    },
+    {
+      "epoch": 4.1092619392185235,
+      "grad_norm": 3.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0145,
+      "step": 17745
+    },
+    {
+      "epoch": 4.1104679208876025,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9552,
+      "step": 17750
+    },
+    {
+      "epoch": 4.1116739025566815,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9672,
+      "step": 17755
+    },
+    {
+      "epoch": 4.11287988422576,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0467,
+      "step": 17760
+    },
+    {
+      "epoch": 4.114085865894839,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9248,
+      "step": 17765
+    },
+    {
+      "epoch": 4.115291847563917,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0002,
+      "step": 17770
+    },
+    {
+      "epoch": 4.116497829232996,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 4.0882,
+      "step": 17775
+    },
+    {
+      "epoch": 4.117703810902074,
+      "grad_norm": 3.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0125,
+      "step": 17780
+    },
+    {
+      "epoch": 4.118909792571153,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9268,
+      "step": 17785
+    },
+    {
+      "epoch": 4.120115774240231,
+      "grad_norm": 3.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8774,
+      "step": 17790
+    },
+    {
+      "epoch": 4.12132175590931,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9526,
+      "step": 17795
+    },
+    {
+      "epoch": 4.122527737578388,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8421,
+      "step": 17800
+    },
+    {
+      "epoch": 4.1237337192474675,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9007,
+      "step": 17805
+    },
+    {
+      "epoch": 4.1249397009165465,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8851,
+      "step": 17810
+    },
+    {
+      "epoch": 4.126145682585625,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0274,
+      "step": 17815
+    },
+    {
+      "epoch": 4.127351664254704,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9692,
+      "step": 17820
+    },
+    {
+      "epoch": 4.128557645923782,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9722,
+      "step": 17825
+    },
+    {
+      "epoch": 4.129763627592861,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9472,
+      "step": 17830
+    },
+    {
+      "epoch": 4.130969609261939,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.8693,
+      "step": 17835
+    },
+    {
+      "epoch": 4.132175590931018,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.992,
+      "step": 17840
+    },
+    {
+      "epoch": 4.133381572600096,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0836,
+      "step": 17845
+    },
+    {
+      "epoch": 4.134587554269175,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9054,
+      "step": 17850
+    },
+    {
+      "epoch": 4.135793535938253,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0747,
+      "step": 17855
+    },
+    {
+      "epoch": 4.136999517607332,
+      "grad_norm": 2.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.877,
+      "step": 17860
+    },
+    {
+      "epoch": 4.138205499276411,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8684,
+      "step": 17865
+    },
+    {
+      "epoch": 4.13941148094549,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0057,
+      "step": 17870
+    },
+    {
+      "epoch": 4.140617462614569,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9937,
+      "step": 17875
+    },
+    {
+      "epoch": 4.141823444283647,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9655,
+      "step": 17880
+    },
+    {
+      "epoch": 4.143029425952726,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8509,
+      "step": 17885
+    },
+    {
+      "epoch": 4.144235407621804,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.945,
+      "step": 17890
+    },
+    {
+      "epoch": 4.145441389290883,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.882,
+      "step": 17895
+    },
+    {
+      "epoch": 4.146647370959961,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0138,
+      "step": 17900
+    },
+    {
+      "epoch": 4.14785335262904,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9427,
+      "step": 17905
+    },
+    {
+      "epoch": 4.149059334298118,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8479,
+      "step": 17910
+    },
+    {
+      "epoch": 4.150265315967197,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9487,
+      "step": 17915
+    },
+    {
+      "epoch": 4.1514712976362755,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.7906,
+      "step": 17920
+    },
+    {
+      "epoch": 4.152677279305355,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8365,
+      "step": 17925
+    },
+    {
+      "epoch": 4.153883260974434,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9746,
+      "step": 17930
+    },
+    {
+      "epoch": 4.155089242643512,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0496,
+      "step": 17935
+    },
+    {
+      "epoch": 4.156295224312591,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9872,
+      "step": 17940
+    },
+    {
+      "epoch": 4.157501205981669,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0278,
+      "step": 17945
+    },
+    {
+      "epoch": 4.158707187650748,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.1542,
+      "step": 17950
+    },
+    {
+      "epoch": 4.159913169319826,
+      "grad_norm": 3.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9273,
+      "step": 17955
+    },
+    {
+      "epoch": 4.161119150988905,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.8292,
+      "step": 17960
+    },
+    {
+      "epoch": 4.162325132657983,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.7716,
+      "step": 17965
+    },
+    {
+      "epoch": 4.163531114327062,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0024,
+      "step": 17970
+    },
+    {
+      "epoch": 4.1647370959961405,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9923,
+      "step": 17975
+    },
+    {
+      "epoch": 4.1659430776652195,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 4.0109,
+      "step": 17980
+    },
+    {
+      "epoch": 4.167149059334298,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9655,
+      "step": 17985
+    },
+    {
+      "epoch": 4.168355041003377,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.7403,
+      "step": 17990
+    },
+    {
+      "epoch": 4.169561022672456,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9884,
+      "step": 17995
+    },
+    {
+      "epoch": 4.170767004341534,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9317,
+      "step": 18000
+    },
+    {
+      "epoch": 4.171972986010613,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8992,
+      "step": 18005
+    },
+    {
+      "epoch": 4.173178967679691,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9805,
+      "step": 18010
+    },
+    {
+      "epoch": 4.17438494934877,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0573,
+      "step": 18015
+    },
+    {
+      "epoch": 4.175590931017848,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8974,
+      "step": 18020
+    },
+    {
+      "epoch": 4.176796912686927,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0002,
+      "step": 18025
+    },
+    {
+      "epoch": 4.1780028943560055,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8932,
+      "step": 18030
+    },
+    {
+      "epoch": 4.1792088760250845,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8835,
+      "step": 18035
+    },
+    {
+      "epoch": 4.180414857694163,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0431,
+      "step": 18040
+    },
+    {
+      "epoch": 4.181620839363242,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9927,
+      "step": 18045
+    },
+    {
+      "epoch": 4.182826821032321,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9982,
+      "step": 18050
+    },
+    {
+      "epoch": 4.184032802701399,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9321,
+      "step": 18055
+    },
+    {
+      "epoch": 4.185238784370478,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.8708,
+      "step": 18060
+    },
+    {
+      "epoch": 4.186444766039556,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9317,
+      "step": 18065
+    },
+    {
+      "epoch": 4.187650747708635,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9679,
+      "step": 18070
+    },
+    {
+      "epoch": 4.188856729377713,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9209,
+      "step": 18075
+    },
+    {
+      "epoch": 4.190062711046792,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.1022,
+      "step": 18080
+    },
+    {
+      "epoch": 4.19126869271587,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9368,
+      "step": 18085
+    },
+    {
+      "epoch": 4.1924746743849495,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.1196,
+      "step": 18090
+    },
+    {
+      "epoch": 4.193680656054028,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0144,
+      "step": 18095
+    },
+    {
+      "epoch": 4.194886637723107,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8412,
+      "step": 18100
+    },
+    {
+      "epoch": 4.196092619392186,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9223,
+      "step": 18105
+    },
+    {
+      "epoch": 4.197298601061264,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8457,
+      "step": 18110
+    },
+    {
+      "epoch": 4.198504582730343,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8579,
+      "step": 18115
+    },
+    {
+      "epoch": 4.199710564399421,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0089,
+      "step": 18120
+    },
+    {
+      "epoch": 4.2009165460685,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9215,
+      "step": 18125
+    },
+    {
+      "epoch": 4.202122527737578,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.8947,
+      "step": 18130
+    },
+    {
+      "epoch": 4.203328509406657,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.132,
+      "step": 18135
+    },
+    {
+      "epoch": 4.204534491075735,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0845,
+      "step": 18140
+    },
+    {
+      "epoch": 4.205740472744814,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0138,
+      "step": 18145
+    },
+    {
+      "epoch": 4.206946454413893,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.7772,
+      "step": 18150
+    },
+    {
+      "epoch": 4.208152436082972,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0942,
+      "step": 18155
+    },
+    {
+      "epoch": 4.20935841775205,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9259,
+      "step": 18160
+    },
+    {
+      "epoch": 4.210564399421129,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.0173,
+      "step": 18165
+    },
+    {
+      "epoch": 4.211770381090208,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9079,
+      "step": 18170
+    },
+    {
+      "epoch": 4.212976362759286,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9434,
+      "step": 18175
+    },
+    {
+      "epoch": 4.214182344428365,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.8833,
+      "step": 18180
+    },
+    {
+      "epoch": 4.215388326097443,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.997,
+      "step": 18185
+    },
+    {
+      "epoch": 4.216594307766522,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0224,
+      "step": 18190
+    },
+    {
+      "epoch": 4.2178002894356,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9269,
+      "step": 18195
+    },
+    {
+      "epoch": 4.219006271104679,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8506,
+      "step": 18200
+    },
+    {
+      "epoch": 4.2202122527737576,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9768,
+      "step": 18205
+    },
+    {
+      "epoch": 4.221418234442837,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9676,
+      "step": 18210
+    },
+    {
+      "epoch": 4.222624216111915,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.7846,
+      "step": 18215
+    },
+    {
+      "epoch": 4.223830197780994,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.979,
+      "step": 18220
+    },
+    {
+      "epoch": 4.225036179450072,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9122,
+      "step": 18225
+    },
+    {
+      "epoch": 4.226242161119151,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8589,
+      "step": 18230
+    },
+    {
+      "epoch": 4.22744814278823,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9471,
+      "step": 18235
+    },
+    {
+      "epoch": 4.228654124457308,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8577,
+      "step": 18240
+    },
+    {
+      "epoch": 4.229860106126387,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.794,
+      "step": 18245
+    },
+    {
+      "epoch": 4.231066087795465,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9315,
+      "step": 18250
+    },
+    {
+      "epoch": 4.232272069464544,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9148,
+      "step": 18255
+    },
+    {
+      "epoch": 4.2334780511336225,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0313,
+      "step": 18260
+    },
+    {
+      "epoch": 4.234684032802702,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.2064,
+      "step": 18265
+    },
+    {
+      "epoch": 4.23589001447178,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8789,
+      "step": 18270
+    },
+    {
+      "epoch": 4.237095996140859,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9149,
+      "step": 18275
+    },
+    {
+      "epoch": 4.238301977809937,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9243,
+      "step": 18280
+    },
+    {
+      "epoch": 4.239507959479016,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9675,
+      "step": 18285
+    },
+    {
+      "epoch": 4.240713941148095,
+      "grad_norm": 5.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8315,
+      "step": 18290
+    },
+    {
+      "epoch": 4.241919922817173,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.919,
+      "step": 18295
+    },
+    {
+      "epoch": 4.243125904486252,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1006,
+      "step": 18300
+    },
+    {
+      "epoch": 4.24433188615533,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.896,
+      "step": 18305
+    },
+    {
+      "epoch": 4.245537867824409,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0299,
+      "step": 18310
+    },
+    {
+      "epoch": 4.2467438494934875,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9486,
+      "step": 18315
+    },
+    {
+      "epoch": 4.2479498311625665,
+      "grad_norm": 3.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9588,
+      "step": 18320
+    },
+    {
+      "epoch": 4.249155812831645,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.7608,
+      "step": 18325
+    },
+    {
+      "epoch": 4.250361794500724,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0113,
+      "step": 18330
+    },
+    {
+      "epoch": 4.251567776169802,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8923,
+      "step": 18335
+    },
+    {
+      "epoch": 4.252773757838881,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9763,
+      "step": 18340
+    },
+    {
+      "epoch": 4.25397973950796,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.809,
+      "step": 18345
+    },
+    {
+      "epoch": 4.255185721177038,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0002,
+      "step": 18350
+    },
+    {
+      "epoch": 4.256391702846117,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8559,
+      "step": 18355
+    },
+    {
+      "epoch": 4.257597684515195,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9349,
+      "step": 18360
+    },
+    {
+      "epoch": 4.258803666184274,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9087,
+      "step": 18365
+    },
+    {
+      "epoch": 4.2600096478533525,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9506,
+      "step": 18370
+    },
+    {
+      "epoch": 4.2612156295224315,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9924,
+      "step": 18375
+    },
+    {
+      "epoch": 4.26242161119151,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9285,
+      "step": 18380
+    },
+    {
+      "epoch": 4.263627592860589,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8252,
+      "step": 18385
+    },
+    {
+      "epoch": 4.264833574529667,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9538,
+      "step": 18390
+    },
+    {
+      "epoch": 4.266039556198746,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9166,
+      "step": 18395
+    },
+    {
+      "epoch": 4.267245537867824,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.7875,
+      "step": 18400
+    },
+    {
+      "epoch": 4.268451519536903,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.947,
+      "step": 18405
+    },
+    {
+      "epoch": 4.269657501205982,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9321,
+      "step": 18410
+    },
+    {
+      "epoch": 4.27086348287506,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.8441,
+      "step": 18415
+    },
+    {
+      "epoch": 4.272069464544139,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9337,
+      "step": 18420
+    },
+    {
+      "epoch": 4.273275446213217,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.802,
+      "step": 18425
+    },
+    {
+      "epoch": 4.2744814278822965,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8029,
+      "step": 18430
+    },
+    {
+      "epoch": 4.275687409551375,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9194,
+      "step": 18435
+    },
+    {
+      "epoch": 4.276893391220454,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8878,
+      "step": 18440
+    },
+    {
+      "epoch": 4.278099372889532,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0155,
+      "step": 18445
+    },
+    {
+      "epoch": 4.279305354558611,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.1717,
+      "step": 18450
+    },
+    {
+      "epoch": 4.280511336227689,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.8489,
+      "step": 18455
+    },
+    {
+      "epoch": 4.281717317896768,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9181,
+      "step": 18460
+    },
+    {
+      "epoch": 4.282923299565846,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9272,
+      "step": 18465
+    },
+    {
+      "epoch": 4.284129281234925,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8896,
+      "step": 18470
+    },
+    {
+      "epoch": 4.285335262904004,
+      "grad_norm": 5.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.8618,
+      "step": 18475
+    },
+    {
+      "epoch": 4.286541244573082,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8792,
+      "step": 18480
+    },
+    {
+      "epoch": 4.287747226242161,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0755,
+      "step": 18485
+    },
+    {
+      "epoch": 4.28895320791124,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.8929,
+      "step": 18490
+    },
+    {
+      "epoch": 4.290159189580319,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9933,
+      "step": 18495
+    },
+    {
+      "epoch": 4.291365171249397,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9579,
+      "step": 18500
+    },
+    {
+      "epoch": 4.292571152918476,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.937,
+      "step": 18505
+    },
+    {
+      "epoch": 4.293777134587554,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.7516,
+      "step": 18510
+    },
+    {
+      "epoch": 4.294983116256633,
+      "grad_norm": 3.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9543,
+      "step": 18515
+    },
+    {
+      "epoch": 4.296189097925711,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.931,
+      "step": 18520
+    },
+    {
+      "epoch": 4.29739507959479,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.941,
+      "step": 18525
+    },
+    {
+      "epoch": 4.298601061263869,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.848,
+      "step": 18530
+    },
+    {
+      "epoch": 4.299807042932947,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.9776,
+      "step": 18535
+    },
+    {
+      "epoch": 4.301013024602026,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9061,
+      "step": 18540
+    },
+    {
+      "epoch": 4.3022190062711045,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.062,
+      "step": 18545
+    },
+    {
+      "epoch": 4.303424987940184,
+      "grad_norm": 3.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.907,
+      "step": 18550
+    },
+    {
+      "epoch": 4.304630969609262,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8085,
+      "step": 18555
+    },
+    {
+      "epoch": 4.305836951278341,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.83,
+      "step": 18560
+    },
+    {
+      "epoch": 4.307042932947419,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9847,
+      "step": 18565
+    },
+    {
+      "epoch": 4.308248914616498,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9728,
+      "step": 18570
+    },
+    {
+      "epoch": 4.309454896285576,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.886,
+      "step": 18575
+    },
+    {
+      "epoch": 4.310660877954655,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.8431,
+      "step": 18580
+    },
+    {
+      "epoch": 4.311866859623734,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.7705,
+      "step": 18585
+    },
+    {
+      "epoch": 4.313072841292812,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0417,
+      "step": 18590
+    },
+    {
+      "epoch": 4.314278822961891,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9488,
+      "step": 18595
+    },
+    {
+      "epoch": 4.3154848046309695,
+      "grad_norm": 3.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9963,
+      "step": 18600
+    },
+    {
+      "epoch": 4.3166907863000485,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0354,
+      "step": 18605
+    },
+    {
+      "epoch": 4.317896767969127,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9202,
+      "step": 18610
+    },
+    {
+      "epoch": 4.319102749638206,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9808,
+      "step": 18615
+    },
+    {
+      "epoch": 4.320308731307284,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8379,
+      "step": 18620
+    },
+    {
+      "epoch": 4.321514712976363,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8134,
+      "step": 18625
+    },
+    {
+      "epoch": 4.322720694645441,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9399,
+      "step": 18630
+    },
+    {
+      "epoch": 4.32392667631452,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9967,
+      "step": 18635
+    },
+    {
+      "epoch": 4.325132657983598,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9444,
+      "step": 18640
+    },
+    {
+      "epoch": 4.326338639652677,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9807,
+      "step": 18645
+    },
+    {
+      "epoch": 4.327544621321756,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0261,
+      "step": 18650
+    },
+    {
+      "epoch": 4.3287506029908345,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9628,
+      "step": 18655
+    },
+    {
+      "epoch": 4.3299565846599135,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.936,
+      "step": 18660
+    },
+    {
+      "epoch": 4.331162566328992,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0522,
+      "step": 18665
+    },
+    {
+      "epoch": 4.332368547998071,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9081,
+      "step": 18670
+    },
+    {
+      "epoch": 4.333574529667149,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8924,
+      "step": 18675
+    },
+    {
+      "epoch": 4.334780511336228,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9114,
+      "step": 18680
+    },
+    {
+      "epoch": 4.335986493005306,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9564,
+      "step": 18685
+    },
+    {
+      "epoch": 4.337192474674385,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9293,
+      "step": 18690
+    },
+    {
+      "epoch": 4.338398456343463,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0191,
+      "step": 18695
+    },
+    {
+      "epoch": 4.339604438012542,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0384,
+      "step": 18700
+    },
+    {
+      "epoch": 4.34081041968162,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.0502,
+      "step": 18705
+    },
+    {
+      "epoch": 4.342016401350699,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.926,
+      "step": 18710
+    },
+    {
+      "epoch": 4.3432223830197785,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9212,
+      "step": 18715
+    },
+    {
+      "epoch": 4.344428364688857,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9521,
+      "step": 18720
+    },
+    {
+      "epoch": 4.345634346357936,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.0382,
+      "step": 18725
+    },
+    {
+      "epoch": 4.346840328027014,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9874,
+      "step": 18730
+    },
+    {
+      "epoch": 4.348046309696093,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 4.0693,
+      "step": 18735
+    },
+    {
+      "epoch": 4.349252291365171,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0806,
+      "step": 18740
+    },
+    {
+      "epoch": 4.35045827303425,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8479,
+      "step": 18745
+    },
+    {
+      "epoch": 4.351664254703328,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8612,
+      "step": 18750
+    },
+    {
+      "epoch": 4.352870236372407,
+      "grad_norm": 3.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.1342,
+      "step": 18755
+    },
+    {
+      "epoch": 4.354076218041485,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9161,
+      "step": 18760
+    },
+    {
+      "epoch": 4.355282199710564,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0085,
+      "step": 18765
+    },
+    {
+      "epoch": 4.356488181379643,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9504,
+      "step": 18770
+    },
+    {
+      "epoch": 4.357694163048722,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8287,
+      "step": 18775
+    },
+    {
+      "epoch": 4.358900144717801,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9961,
+      "step": 18780
+    },
+    {
+      "epoch": 4.360106126386879,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9507,
+      "step": 18785
+    },
+    {
+      "epoch": 4.361312108055958,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.8648,
+      "step": 18790
+    },
+    {
+      "epoch": 4.362518089725036,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9403,
+      "step": 18795
+    },
+    {
+      "epoch": 4.363724071394115,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8749,
+      "step": 18800
+    },
+    {
+      "epoch": 4.364930053063193,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9702,
+      "step": 18805
+    },
+    {
+      "epoch": 4.366136034732272,
+      "grad_norm": 3.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8223,
+      "step": 18810
+    },
+    {
+      "epoch": 4.36734201640135,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.966,
+      "step": 18815
+    },
+    {
+      "epoch": 4.368547998070429,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.1193,
+      "step": 18820
+    },
+    {
+      "epoch": 4.369753979739508,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0578,
+      "step": 18825
+    },
+    {
+      "epoch": 4.370959961408587,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9657,
+      "step": 18830
+    },
+    {
+      "epoch": 4.372165943077666,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8123,
+      "step": 18835
+    },
+    {
+      "epoch": 4.373371924746744,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9681,
+      "step": 18840
+    },
+    {
+      "epoch": 4.374577906415823,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8817,
+      "step": 18845
+    },
+    {
+      "epoch": 4.375783888084901,
+      "grad_norm": 3.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9669,
+      "step": 18850
+    },
+    {
+      "epoch": 4.37698986975398,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.86,
+      "step": 18855
+    },
+    {
+      "epoch": 4.378195851423058,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.7974,
+      "step": 18860
+    },
+    {
+      "epoch": 4.379401833092137,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9139,
+      "step": 18865
+    },
+    {
+      "epoch": 4.380607814761215,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 4.1553,
+      "step": 18870
+    },
+    {
+      "epoch": 4.381813796430294,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8111,
+      "step": 18875
+    },
+    {
+      "epoch": 4.3830197780993725,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8528,
+      "step": 18880
+    },
+    {
+      "epoch": 4.3842257597684515,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9521,
+      "step": 18885
+    },
+    {
+      "epoch": 4.385431741437531,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8524,
+      "step": 18890
+    },
+    {
+      "epoch": 4.386637723106609,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9402,
+      "step": 18895
+    },
+    {
+      "epoch": 4.387843704775688,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0998,
+      "step": 18900
+    },
+    {
+      "epoch": 4.389049686444766,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9386,
+      "step": 18905
+    },
+    {
+      "epoch": 4.390255668113845,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9202,
+      "step": 18910
+    },
+    {
+      "epoch": 4.391461649782923,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9179,
+      "step": 18915
+    },
+    {
+      "epoch": 4.392667631452002,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9382,
+      "step": 18920
+    },
+    {
+      "epoch": 4.39387361312108,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9612,
+      "step": 18925
+    },
+    {
+      "epoch": 4.395079594790159,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.891,
+      "step": 18930
+    },
+    {
+      "epoch": 4.3962855764592375,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9813,
+      "step": 18935
+    },
+    {
+      "epoch": 4.3974915581283165,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9354,
+      "step": 18940
+    },
+    {
+      "epoch": 4.398697539797395,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9109,
+      "step": 18945
+    },
+    {
+      "epoch": 4.399903521466474,
+      "grad_norm": 4.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0227,
+      "step": 18950
+    },
+    {
+      "epoch": 4.401109503135553,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.972,
+      "step": 18955
+    },
+    {
+      "epoch": 4.402315484804631,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0383,
+      "step": 18960
+    },
+    {
+      "epoch": 4.40352146647371,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 4.0109,
+      "step": 18965
+    },
+    {
+      "epoch": 4.404727448142788,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0041,
+      "step": 18970
+    },
+    {
+      "epoch": 4.405933429811867,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0996,
+      "step": 18975
+    },
+    {
+      "epoch": 4.407139411480945,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1244,
+      "step": 18980
+    },
+    {
+      "epoch": 4.408345393150024,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.8899,
+      "step": 18985
+    },
+    {
+      "epoch": 4.409551374819102,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9266,
+      "step": 18990
+    },
+    {
+      "epoch": 4.4107573564881815,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0406,
+      "step": 18995
+    },
+    {
+      "epoch": 4.41196333815726,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.8808,
+      "step": 19000
+    },
+    {
+      "epoch": 4.413169319826339,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.8306,
+      "step": 19005
+    },
+    {
+      "epoch": 4.414375301495418,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.8822,
+      "step": 19010
+    },
+    {
+      "epoch": 4.415581283164496,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.102,
+      "step": 19015
+    },
+    {
+      "epoch": 4.416787264833575,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8508,
+      "step": 19020
+    },
+    {
+      "epoch": 4.417993246502653,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8282,
+      "step": 19025
+    },
+    {
+      "epoch": 4.419199228171732,
+      "grad_norm": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 3.98,
+      "step": 19030
+    },
+    {
+      "epoch": 4.42040520984081,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9701,
+      "step": 19035
+    },
+    {
+      "epoch": 4.421611191509889,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.89,
+      "step": 19040
+    },
+    {
+      "epoch": 4.422817173178967,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0001,
+      "step": 19045
+    },
+    {
+      "epoch": 4.424023154848046,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9654,
+      "step": 19050
+    },
+    {
+      "epoch": 4.425229136517125,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8927,
+      "step": 19055
+    },
+    {
+      "epoch": 4.426435118186204,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8706,
+      "step": 19060
+    },
+    {
+      "epoch": 4.427641099855283,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.8929,
+      "step": 19065
+    },
+    {
+      "epoch": 4.428847081524361,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0062,
+      "step": 19070
+    },
+    {
+      "epoch": 4.43005306319344,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7936,
+      "step": 19075
+    },
+    {
+      "epoch": 4.431259044862518,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9664,
+      "step": 19080
+    },
+    {
+      "epoch": 4.432465026531597,
+      "grad_norm": 4.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9573,
+      "step": 19085
+    },
+    {
+      "epoch": 4.433671008200675,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0506,
+      "step": 19090
+    },
+    {
+      "epoch": 4.434876989869754,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9315,
+      "step": 19095
+    },
+    {
+      "epoch": 4.436082971538832,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9096,
+      "step": 19100
+    },
+    {
+      "epoch": 4.437288953207911,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8916,
+      "step": 19105
+    },
+    {
+      "epoch": 4.4384949348769895,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9185,
+      "step": 19110
+    },
+    {
+      "epoch": 4.439700916546069,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9442,
+      "step": 19115
+    },
+    {
+      "epoch": 4.440906898215147,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.976,
+      "step": 19120
+    },
+    {
+      "epoch": 4.442112879884226,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9757,
+      "step": 19125
+    },
+    {
+      "epoch": 4.443318861553305,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9799,
+      "step": 19130
+    },
+    {
+      "epoch": 4.444524843222383,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.947,
+      "step": 19135
+    },
+    {
+      "epoch": 4.445730824891462,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9425,
+      "step": 19140
+    },
+    {
+      "epoch": 4.44693680656054,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9554,
+      "step": 19145
+    },
+    {
+      "epoch": 4.448142788229619,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9967,
+      "step": 19150
+    },
+    {
+      "epoch": 4.449348769898697,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8435,
+      "step": 19155
+    },
+    {
+      "epoch": 4.450554751567776,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0367,
+      "step": 19160
+    },
+    {
+      "epoch": 4.4517607332368545,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9623,
+      "step": 19165
+    },
+    {
+      "epoch": 4.4529667149059335,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9176,
+      "step": 19170
+    },
+    {
+      "epoch": 4.454172696575012,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9261,
+      "step": 19175
+    },
+    {
+      "epoch": 4.455378678244091,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9105,
+      "step": 19180
+    },
+    {
+      "epoch": 4.456584659913169,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8613,
+      "step": 19185
+    },
+    {
+      "epoch": 4.457790641582248,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8935,
+      "step": 19190
+    },
+    {
+      "epoch": 4.458996623251327,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0224,
+      "step": 19195
+    },
+    {
+      "epoch": 4.460202604920405,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9101,
+      "step": 19200
+    },
+    {
+      "epoch": 4.461408586589484,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0449,
+      "step": 19205
+    },
+    {
+      "epoch": 4.462614568258562,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9605,
+      "step": 19210
+    },
+    {
+      "epoch": 4.463820549927641,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.9333,
+      "step": 19215
+    },
+    {
+      "epoch": 4.4650265315967195,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0092,
+      "step": 19220
+    },
+    {
+      "epoch": 4.4662325132657985,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0187,
+      "step": 19225
+    },
+    {
+      "epoch": 4.467438494934877,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9404,
+      "step": 19230
+    },
+    {
+      "epoch": 4.468644476603956,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0867,
+      "step": 19235
+    },
+    {
+      "epoch": 4.469850458273034,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9912,
+      "step": 19240
+    },
+    {
+      "epoch": 4.471056439942113,
+      "grad_norm": 3.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0355,
+      "step": 19245
+    },
+    {
+      "epoch": 4.472262421611192,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9804,
+      "step": 19250
+    },
+    {
+      "epoch": 4.47346840328027,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0691,
+      "step": 19255
+    },
+    {
+      "epoch": 4.474674384949349,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9969,
+      "step": 19260
+    },
+    {
+      "epoch": 4.475880366618427,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.0161,
+      "step": 19265
+    },
+    {
+      "epoch": 4.477086348287506,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0143,
+      "step": 19270
+    },
+    {
+      "epoch": 4.478292329956584,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9209,
+      "step": 19275
+    },
+    {
+      "epoch": 4.4794983116256635,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9923,
+      "step": 19280
+    },
+    {
+      "epoch": 4.480704293294742,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.037,
+      "step": 19285
+    },
+    {
+      "epoch": 4.481910274963821,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9443,
+      "step": 19290
+    },
+    {
+      "epoch": 4.483116256632899,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9038,
+      "step": 19295
+    },
+    {
+      "epoch": 4.484322238301978,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.912,
+      "step": 19300
+    },
+    {
+      "epoch": 4.485528219971057,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.8101,
+      "step": 19305
+    },
+    {
+      "epoch": 4.486734201640135,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0292,
+      "step": 19310
+    },
+    {
+      "epoch": 4.487940183309214,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8724,
+      "step": 19315
+    },
+    {
+      "epoch": 4.489146164978292,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9854,
+      "step": 19320
+    },
+    {
+      "epoch": 4.490352146647371,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9615,
+      "step": 19325
+    },
+    {
+      "epoch": 4.491558128316449,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.9114,
+      "step": 19330
+    },
+    {
+      "epoch": 4.492764109985528,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9932,
+      "step": 19335
+    },
+    {
+      "epoch": 4.493970091654607,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8948,
+      "step": 19340
+    },
+    {
+      "epoch": 4.495176073323686,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8645,
+      "step": 19345
+    },
+    {
+      "epoch": 4.496382054992764,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9229,
+      "step": 19350
+    },
+    {
+      "epoch": 4.497588036661843,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9432,
+      "step": 19355
+    },
+    {
+      "epoch": 4.498794018330921,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8388,
+      "step": 19360
+    },
+    {
+      "epoch": 4.5,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.948,
+      "step": 19365
+    },
+    {
+      "epoch": 4.501205981669079,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.7818,
+      "step": 19370
+    },
+    {
+      "epoch": 4.502411963338157,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9572,
+      "step": 19375
+    },
+    {
+      "epoch": 4.503617945007236,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8603,
+      "step": 19380
+    },
+    {
+      "epoch": 4.504823926676314,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9934,
+      "step": 19385
+    },
+    {
+      "epoch": 4.506029908345393,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0172,
+      "step": 19390
+    },
+    {
+      "epoch": 4.507235890014472,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9433,
+      "step": 19395
+    },
+    {
+      "epoch": 4.508441871683551,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0175,
+      "step": 19400
+    },
+    {
+      "epoch": 4.509647853352629,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9929,
+      "step": 19405
+    },
+    {
+      "epoch": 4.510853835021708,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0433,
+      "step": 19410
+    },
+    {
+      "epoch": 4.512059816690786,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9826,
+      "step": 19415
+    },
+    {
+      "epoch": 4.513265798359865,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0318,
+      "step": 19420
+    },
+    {
+      "epoch": 4.514471780028943,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9771,
+      "step": 19425
+    },
+    {
+      "epoch": 4.515677761698022,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9192,
+      "step": 19430
+    },
+    {
+      "epoch": 4.516883743367101,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0718,
+      "step": 19435
+    },
+    {
+      "epoch": 4.518089725036179,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.886,
+      "step": 19440
+    },
+    {
+      "epoch": 4.519295706705258,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9867,
+      "step": 19445
+    },
+    {
+      "epoch": 4.5205016883743365,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8466,
+      "step": 19450
+    },
+    {
+      "epoch": 4.521707670043416,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8188,
+      "step": 19455
+    },
+    {
+      "epoch": 4.522913651712494,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8884,
+      "step": 19460
+    },
+    {
+      "epoch": 4.524119633381573,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9647,
+      "step": 19465
+    },
+    {
+      "epoch": 4.525325615050651,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9551,
+      "step": 19470
+    },
+    {
+      "epoch": 4.52653159671973,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9088,
+      "step": 19475
+    },
+    {
+      "epoch": 4.527737578388809,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8979,
+      "step": 19480
+    },
+    {
+      "epoch": 4.528943560057887,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9685,
+      "step": 19485
+    },
+    {
+      "epoch": 4.530149541726965,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8511,
+      "step": 19490
+    },
+    {
+      "epoch": 4.531355523396044,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0053,
+      "step": 19495
+    },
+    {
+      "epoch": 4.532561505065123,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.76,
+      "step": 19500
+    },
+    {
+      "epoch": 4.5337674867342015,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8996,
+      "step": 19505
+    },
+    {
+      "epoch": 4.5349734684032805,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9456,
+      "step": 19510
+    },
+    {
+      "epoch": 4.536179450072359,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8806,
+      "step": 19515
+    },
+    {
+      "epoch": 4.537385431741438,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9195,
+      "step": 19520
+    },
+    {
+      "epoch": 4.538591413410516,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.8215,
+      "step": 19525
+    },
+    {
+      "epoch": 4.539797395079595,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9012,
+      "step": 19530
+    },
+    {
+      "epoch": 4.541003376748673,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8991,
+      "step": 19535
+    },
+    {
+      "epoch": 4.542209358417752,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9642,
+      "step": 19540
+    },
+    {
+      "epoch": 4.543415340086831,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9493,
+      "step": 19545
+    },
+    {
+      "epoch": 4.544621321755909,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.839,
+      "step": 19550
+    },
+    {
+      "epoch": 4.545827303424988,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9932,
+      "step": 19555
+    },
+    {
+      "epoch": 4.5470332850940665,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.932,
+      "step": 19560
+    },
+    {
+      "epoch": 4.5482392667631455,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.1326,
+      "step": 19565
+    },
+    {
+      "epoch": 4.549445248432224,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.7192,
+      "step": 19570
+    },
+    {
+      "epoch": 4.550651230101303,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9293,
+      "step": 19575
+    },
+    {
+      "epoch": 4.551857211770381,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1116,
+      "step": 19580
+    },
+    {
+      "epoch": 4.55306319343946,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.917,
+      "step": 19585
+    },
+    {
+      "epoch": 4.554269175108538,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.1351,
+      "step": 19590
+    },
+    {
+      "epoch": 4.555475156777617,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9761,
+      "step": 19595
+    },
+    {
+      "epoch": 4.556681138446695,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.849,
+      "step": 19600
+    },
+    {
+      "epoch": 4.557887120115774,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8289,
+      "step": 19605
+    },
+    {
+      "epoch": 4.559093101784853,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0511,
+      "step": 19610
+    },
+    {
+      "epoch": 4.560299083453931,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0382,
+      "step": 19615
+    },
+    {
+      "epoch": 4.5615050651230105,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9575,
+      "step": 19620
+    },
+    {
+      "epoch": 4.562711046792089,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8191,
+      "step": 19625
+    },
+    {
+      "epoch": 4.563917028461168,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.901,
+      "step": 19630
+    },
+    {
+      "epoch": 4.565123010130246,
+      "grad_norm": 3.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.766,
+      "step": 19635
+    },
+    {
+      "epoch": 4.566328991799325,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9071,
+      "step": 19640
+    },
+    {
+      "epoch": 4.567534973468403,
+      "grad_norm": 3.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0219,
+      "step": 19645
+    },
+    {
+      "epoch": 4.568740955137482,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.1069,
+      "step": 19650
+    },
+    {
+      "epoch": 4.56994693680656,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.8216,
+      "step": 19655
+    },
+    {
+      "epoch": 4.571152918475639,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9227,
+      "step": 19660
+    },
+    {
+      "epoch": 4.572358900144717,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1128,
+      "step": 19665
+    },
+    {
+      "epoch": 4.573564881813796,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.8642,
+      "step": 19670
+    },
+    {
+      "epoch": 4.574770863482875,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.8425,
+      "step": 19675
+    },
+    {
+      "epoch": 4.575976845151954,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0141,
+      "step": 19680
+    },
+    {
+      "epoch": 4.577182826821033,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9448,
+      "step": 19685
+    },
+    {
+      "epoch": 4.578388808490111,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0479,
+      "step": 19690
+    },
+    {
+      "epoch": 4.57959479015919,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.9816,
+      "step": 19695
+    },
+    {
+      "epoch": 4.580800771828268,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9442,
+      "step": 19700
+    },
+    {
+      "epoch": 4.582006753497347,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9765,
+      "step": 19705
+    },
+    {
+      "epoch": 4.583212735166425,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0233,
+      "step": 19710
+    },
+    {
+      "epoch": 4.584418716835504,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8763,
+      "step": 19715
+    },
+    {
+      "epoch": 4.585624698504583,
+      "grad_norm": 4.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9905,
+      "step": 19720
+    },
+    {
+      "epoch": 4.586830680173661,
+      "grad_norm": 3.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9975,
+      "step": 19725
+    },
+    {
+      "epoch": 4.5880366618427395,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9168,
+      "step": 19730
+    },
+    {
+      "epoch": 4.5892426435118185,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9945,
+      "step": 19735
+    },
+    {
+      "epoch": 4.590448625180898,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0105,
+      "step": 19740
+    },
+    {
+      "epoch": 4.591654606849976,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8938,
+      "step": 19745
+    },
+    {
+      "epoch": 4.592860588519055,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0372,
+      "step": 19750
+    },
+    {
+      "epoch": 4.594066570188133,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9091,
+      "step": 19755
+    },
+    {
+      "epoch": 4.595272551857212,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9619,
+      "step": 19760
+    },
+    {
+      "epoch": 4.59647853352629,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9201,
+      "step": 19765
+    },
+    {
+      "epoch": 4.597684515195369,
+      "grad_norm": 3.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8557,
+      "step": 19770
+    },
+    {
+      "epoch": 4.598890496864447,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9338,
+      "step": 19775
+    },
+    {
+      "epoch": 4.600096478533526,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9232,
+      "step": 19780
+    },
+    {
+      "epoch": 4.601302460202605,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9607,
+      "step": 19785
+    },
+    {
+      "epoch": 4.6025084418716835,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9891,
+      "step": 19790
+    },
+    {
+      "epoch": 4.6037144235407625,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9689,
+      "step": 19795
+    },
+    {
+      "epoch": 4.604920405209841,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0237,
+      "step": 19800
+    },
+    {
+      "epoch": 4.60612638687892,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9795,
+      "step": 19805
+    },
+    {
+      "epoch": 4.607332368547998,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9382,
+      "step": 19810
+    },
+    {
+      "epoch": 4.608538350217077,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.924,
+      "step": 19815
+    },
+    {
+      "epoch": 4.609744331886155,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0634,
+      "step": 19820
+    },
+    {
+      "epoch": 4.610950313555234,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9434,
+      "step": 19825
+    },
+    {
+      "epoch": 4.612156295224312,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0461,
+      "step": 19830
+    },
+    {
+      "epoch": 4.613362276893391,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0901,
+      "step": 19835
+    },
+    {
+      "epoch": 4.614568258562469,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.075,
+      "step": 19840
+    },
+    {
+      "epoch": 4.6157742402315485,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9903,
+      "step": 19845
+    },
+    {
+      "epoch": 4.6169802219006275,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.7537,
+      "step": 19850
+    },
+    {
+      "epoch": 4.618186203569706,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.8785,
+      "step": 19855
+    },
+    {
+      "epoch": 4.619392185238785,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 4.0932,
+      "step": 19860
+    },
+    {
+      "epoch": 4.620598166907863,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9081,
+      "step": 19865
+    },
+    {
+      "epoch": 4.621804148576942,
+      "grad_norm": 4.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9645,
+      "step": 19870
+    },
+    {
+      "epoch": 4.62301013024602,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.002,
+      "step": 19875
+    },
+    {
+      "epoch": 4.624216111915099,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.7916,
+      "step": 19880
+    },
+    {
+      "epoch": 4.625422093584177,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0679,
+      "step": 19885
+    },
+    {
+      "epoch": 4.626628075253256,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8554,
+      "step": 19890
+    },
+    {
+      "epoch": 4.627834056922334,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.8759,
+      "step": 19895
+    },
+    {
+      "epoch": 4.629040038591413,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1692,
+      "step": 19900
+    },
+    {
+      "epoch": 4.630246020260492,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0769,
+      "step": 19905
+    },
+    {
+      "epoch": 4.631452001929571,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9471,
+      "step": 19910
+    },
+    {
+      "epoch": 4.63265798359865,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9627,
+      "step": 19915
+    },
+    {
+      "epoch": 4.633863965267728,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.9123,
+      "step": 19920
+    },
+    {
+      "epoch": 4.635069946936807,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9039,
+      "step": 19925
+    },
+    {
+      "epoch": 4.636275928605885,
+      "grad_norm": 3.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9443,
+      "step": 19930
+    },
+    {
+      "epoch": 4.637481910274964,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8838,
+      "step": 19935
+    },
+    {
+      "epoch": 4.638687891944042,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8633,
+      "step": 19940
+    },
+    {
+      "epoch": 4.639893873613121,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0481,
+      "step": 19945
+    },
+    {
+      "epoch": 4.641099855282199,
+      "grad_norm": 3.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9399,
+      "step": 19950
+    },
+    {
+      "epoch": 4.642305836951278,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.077,
+      "step": 19955
+    },
+    {
+      "epoch": 4.6435118186203574,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.1191,
+      "step": 19960
+    },
+    {
+      "epoch": 4.644717800289436,
+      "grad_norm": 3.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9741,
+      "step": 19965
+    },
+    {
+      "epoch": 4.645923781958514,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8167,
+      "step": 19970
+    },
+    {
+      "epoch": 4.647129763627593,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8412,
+      "step": 19975
+    },
+    {
+      "epoch": 4.648335745296672,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9654,
+      "step": 19980
+    },
+    {
+      "epoch": 4.64954172696575,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.7964,
+      "step": 19985
+    },
+    {
+      "epoch": 4.650747708634829,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.8671,
+      "step": 19990
+    },
+    {
+      "epoch": 4.651953690303907,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.0256,
+      "step": 19995
+    },
+    {
+      "epoch": 4.653159671972986,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0187,
+      "step": 20000
+    },
+    {
+      "epoch": 4.654365653642064,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8417,
+      "step": 20005
+    },
+    {
+      "epoch": 4.655571635311143,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0317,
+      "step": 20010
+    },
+    {
+      "epoch": 4.6567776169802215,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8415,
+      "step": 20015
+    },
+    {
+      "epoch": 4.657983598649301,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0008,
+      "step": 20020
+    },
+    {
+      "epoch": 4.65918958031838,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 4.171,
+      "step": 20025
+    },
+    {
+      "epoch": 4.660395561987458,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0814,
+      "step": 20030
+    },
+    {
+      "epoch": 4.661601543656537,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9164,
+      "step": 20035
+    },
+    {
+      "epoch": 4.662807525325615,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.043,
+      "step": 20040
+    },
+    {
+      "epoch": 4.664013506994694,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0084,
+      "step": 20045
+    },
+    {
+      "epoch": 4.665219488663772,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8783,
+      "step": 20050
+    },
+    {
+      "epoch": 4.666425470332851,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8936,
+      "step": 20055
+    },
+    {
+      "epoch": 4.667631452001929,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.009,
+      "step": 20060
+    },
+    {
+      "epoch": 4.668837433671008,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0824,
+      "step": 20065
+    },
+    {
+      "epoch": 4.6700434153400865,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0993,
+      "step": 20070
+    },
+    {
+      "epoch": 4.6712493970091655,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.0027,
+      "step": 20075
+    },
+    {
+      "epoch": 4.672455378678244,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8808,
+      "step": 20080
+    },
+    {
+      "epoch": 4.673661360347323,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9794,
+      "step": 20085
+    },
+    {
+      "epoch": 4.674867342016402,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9214,
+      "step": 20090
+    },
+    {
+      "epoch": 4.67607332368548,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9029,
+      "step": 20095
+    },
+    {
+      "epoch": 4.677279305354559,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9026,
+      "step": 20100
+    },
+    {
+      "epoch": 4.678485287023637,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8744,
+      "step": 20105
+    },
+    {
+      "epoch": 4.679691268692716,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0173,
+      "step": 20110
+    },
+    {
+      "epoch": 4.680897250361794,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9444,
+      "step": 20115
+    },
+    {
+      "epoch": 4.682103232030873,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9932,
+      "step": 20120
+    },
+    {
+      "epoch": 4.6833092136999515,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8665,
+      "step": 20125
+    },
+    {
+      "epoch": 4.6845151953690305,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.877,
+      "step": 20130
+    },
+    {
+      "epoch": 4.685721177038109,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 4.0382,
+      "step": 20135
+    },
+    {
+      "epoch": 4.686927158707188,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9295,
+      "step": 20140
+    },
+    {
+      "epoch": 4.688133140376266,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8698,
+      "step": 20145
+    },
+    {
+      "epoch": 4.689339122045345,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9039,
+      "step": 20150
+    },
+    {
+      "epoch": 4.690545103714424,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9007,
+      "step": 20155
+    },
+    {
+      "epoch": 4.691751085383502,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.896,
+      "step": 20160
+    },
+    {
+      "epoch": 4.692957067052581,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8963,
+      "step": 20165
+    },
+    {
+      "epoch": 4.694163048721659,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8924,
+      "step": 20170
+    },
+    {
+      "epoch": 4.695369030390738,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9467,
+      "step": 20175
+    },
+    {
+      "epoch": 4.696575012059816,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.8768,
+      "step": 20180
+    },
+    {
+      "epoch": 4.6977809937288955,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9075,
+      "step": 20185
+    },
+    {
+      "epoch": 4.698986975397974,
+      "grad_norm": 3.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0124,
+      "step": 20190
+    },
+    {
+      "epoch": 4.700192957067053,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8964,
+      "step": 20195
+    },
+    {
+      "epoch": 4.701398938736132,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9017,
+      "step": 20200
+    },
+    {
+      "epoch": 4.70260492040521,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9123,
+      "step": 20205
+    },
+    {
+      "epoch": 4.703810902074288,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8724,
+      "step": 20210
+    },
+    {
+      "epoch": 4.705016883743367,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9993,
+      "step": 20215
+    },
+    {
+      "epoch": 4.706222865412446,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.8621,
+      "step": 20220
+    },
+    {
+      "epoch": 4.707428847081524,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0181,
+      "step": 20225
+    },
+    {
+      "epoch": 4.708634828750603,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9327,
+      "step": 20230
+    },
+    {
+      "epoch": 4.709840810419681,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8933,
+      "step": 20235
+    },
+    {
+      "epoch": 4.71104679208876,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9689,
+      "step": 20240
+    },
+    {
+      "epoch": 4.712252773757839,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.7931,
+      "step": 20245
+    },
+    {
+      "epoch": 4.713458755426918,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.8618,
+      "step": 20250
+    },
+    {
+      "epoch": 4.714664737095996,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0381,
+      "step": 20255
+    },
+    {
+      "epoch": 4.715870718765075,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9483,
+      "step": 20260
+    },
+    {
+      "epoch": 4.717076700434154,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.933,
+      "step": 20265
+    },
+    {
+      "epoch": 4.718282682103232,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8767,
+      "step": 20270
+    },
+    {
+      "epoch": 4.719488663772311,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.1403,
+      "step": 20275
+    },
+    {
+      "epoch": 4.720694645441389,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9109,
+      "step": 20280
+    },
+    {
+      "epoch": 4.721900627110468,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0063,
+      "step": 20285
+    },
+    {
+      "epoch": 4.723106608779546,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9716,
+      "step": 20290
+    },
+    {
+      "epoch": 4.724312590448625,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.868,
+      "step": 20295
+    },
+    {
+      "epoch": 4.7255185721177035,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9417,
+      "step": 20300
+    },
+    {
+      "epoch": 4.726724553786783,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8789,
+      "step": 20305
+    },
+    {
+      "epoch": 4.727930535455861,
+      "grad_norm": 2.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.7718,
+      "step": 20310
+    },
+    {
+      "epoch": 4.72913651712494,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.924,
+      "step": 20315
+    },
+    {
+      "epoch": 4.730342498794018,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9154,
+      "step": 20320
+    },
+    {
+      "epoch": 4.731548480463097,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0157,
+      "step": 20325
+    },
+    {
+      "epoch": 4.732754462132176,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9031,
+      "step": 20330
+    },
+    {
+      "epoch": 4.733960443801254,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9127,
+      "step": 20335
+    },
+    {
+      "epoch": 4.735166425470333,
+      "grad_norm": 3.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.935,
+      "step": 20340
+    },
+    {
+      "epoch": 4.736372407139411,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9078,
+      "step": 20345
+    },
+    {
+      "epoch": 4.73757838880849,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9356,
+      "step": 20350
+    },
+    {
+      "epoch": 4.7387843704775685,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9006,
+      "step": 20355
+    },
+    {
+      "epoch": 4.7399903521466475,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8435,
+      "step": 20360
+    },
+    {
+      "epoch": 4.741196333815726,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9632,
+      "step": 20365
+    },
+    {
+      "epoch": 4.742402315484805,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.931,
+      "step": 20370
+    },
+    {
+      "epoch": 4.743608297153883,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0474,
+      "step": 20375
+    },
+    {
+      "epoch": 4.744814278822962,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9415,
+      "step": 20380
+    },
+    {
+      "epoch": 4.74602026049204,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.961,
+      "step": 20385
+    },
+    {
+      "epoch": 4.747226242161119,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8224,
+      "step": 20390
+    },
+    {
+      "epoch": 4.748432223830198,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.7832,
+      "step": 20395
+    },
+    {
+      "epoch": 4.749638205499276,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8409,
+      "step": 20400
+    },
+    {
+      "epoch": 4.750844187168355,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1794,
+      "step": 20405
+    },
+    {
+      "epoch": 4.7520501688374335,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0168,
+      "step": 20410
+    },
+    {
+      "epoch": 4.7532561505065125,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0308,
+      "step": 20415
+    },
+    {
+      "epoch": 4.754462132175591,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9314,
+      "step": 20420
+    },
+    {
+      "epoch": 4.75566811384467,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.038,
+      "step": 20425
+    },
+    {
+      "epoch": 4.756874095513748,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9364,
+      "step": 20430
+    },
+    {
+      "epoch": 4.758080077182827,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0435,
+      "step": 20435
+    },
+    {
+      "epoch": 4.759286058851906,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9514,
+      "step": 20440
+    },
+    {
+      "epoch": 4.760492040520984,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9252,
+      "step": 20445
+    },
+    {
+      "epoch": 4.761698022190062,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.9576,
+      "step": 20450
+    },
+    {
+      "epoch": 4.762904003859141,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9271,
+      "step": 20455
+    },
+    {
+      "epoch": 4.76410998552822,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.8534,
+      "step": 20460
+    },
+    {
+      "epoch": 4.765315967197298,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.858,
+      "step": 20465
+    },
+    {
+      "epoch": 4.7665219488663775,
+      "grad_norm": 3.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8748,
+      "step": 20470
+    },
+    {
+      "epoch": 4.767727930535456,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9712,
+      "step": 20475
+    },
+    {
+      "epoch": 4.768933912204535,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9745,
+      "step": 20480
+    },
+    {
+      "epoch": 4.770139893873613,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.98,
+      "step": 20485
+    },
+    {
+      "epoch": 4.771345875542692,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8438,
+      "step": 20490
+    },
+    {
+      "epoch": 4.77255185721177,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8551,
+      "step": 20495
+    },
+    {
+      "epoch": 4.773757838880849,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.654,
+      "step": 20500
+    },
+    {
+      "epoch": 4.774963820549928,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 4.0144,
+      "step": 20505
+    },
+    {
+      "epoch": 4.776169802219006,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.1523,
+      "step": 20510
+    },
+    {
+      "epoch": 4.777375783888085,
+      "grad_norm": 2.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.1056,
+      "step": 20515
+    },
+    {
+      "epoch": 4.778581765557163,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9953,
+      "step": 20520
+    },
+    {
+      "epoch": 4.7797877472262424,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.8346,
+      "step": 20525
+    },
+    {
+      "epoch": 4.780993728895321,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0517,
+      "step": 20530
+    },
+    {
+      "epoch": 4.7821997105644,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8489,
+      "step": 20535
+    },
+    {
+      "epoch": 4.783405692233478,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.9816,
+      "step": 20540
+    },
+    {
+      "epoch": 4.784611673902557,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9426,
+      "step": 20545
+    },
+    {
+      "epoch": 4.785817655571635,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9752,
+      "step": 20550
+    },
+    {
+      "epoch": 4.787023637240714,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0684,
+      "step": 20555
+    },
+    {
+      "epoch": 4.788229618909792,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 4.1085,
+      "step": 20560
+    },
+    {
+      "epoch": 4.789435600578871,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9727,
+      "step": 20565
+    },
+    {
+      "epoch": 4.79064158224795,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9538,
+      "step": 20570
+    },
+    {
+      "epoch": 4.791847563917028,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9423,
+      "step": 20575
+    },
+    {
+      "epoch": 4.793053545586107,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0441,
+      "step": 20580
+    },
+    {
+      "epoch": 4.794259527255186,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9836,
+      "step": 20585
+    },
+    {
+      "epoch": 4.795465508924265,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9774,
+      "step": 20590
+    },
+    {
+      "epoch": 4.796671490593343,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.912,
+      "step": 20595
+    },
+    {
+      "epoch": 4.797877472262422,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9208,
+      "step": 20600
+    },
+    {
+      "epoch": 4.7990834539315,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0449,
+      "step": 20605
+    },
+    {
+      "epoch": 4.800289435600579,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.031,
+      "step": 20610
+    },
+    {
+      "epoch": 4.801495417269657,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8891,
+      "step": 20615
+    },
+    {
+      "epoch": 4.802701398938736,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9214,
+      "step": 20620
+    },
+    {
+      "epoch": 4.803907380607814,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0434,
+      "step": 20625
+    },
+    {
+      "epoch": 4.805113362276893,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9426,
+      "step": 20630
+    },
+    {
+      "epoch": 4.806319343945972,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0018,
+      "step": 20635
+    },
+    {
+      "epoch": 4.8075253256150505,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0681,
+      "step": 20640
+    },
+    {
+      "epoch": 4.80873130728413,
+      "grad_norm": 4.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9522,
+      "step": 20645
+    },
+    {
+      "epoch": 4.809937288953208,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8802,
+      "step": 20650
+    },
+    {
+      "epoch": 4.811143270622287,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8852,
+      "step": 20655
+    },
+    {
+      "epoch": 4.812349252291365,
+      "grad_norm": 3.375,
+      "learning_rate": 3e-05,
+      "loss": 3.9538,
+      "step": 20660
+    },
+    {
+      "epoch": 4.813555233960444,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9321,
+      "step": 20665
+    },
+    {
+      "epoch": 4.814761215629522,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9748,
+      "step": 20670
+    },
+    {
+      "epoch": 4.815967197298601,
+      "grad_norm": 6.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0384,
+      "step": 20675
+    },
+    {
+      "epoch": 4.81717317896768,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.8488,
+      "step": 20680
+    },
+    {
+      "epoch": 4.818379160636758,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0321,
+      "step": 20685
+    },
+    {
+      "epoch": 4.8195851423058365,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.968,
+      "step": 20690
+    },
+    {
+      "epoch": 4.8207911239749155,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.7767,
+      "step": 20695
+    },
+    {
+      "epoch": 4.8219971056439945,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9967,
+      "step": 20700
+    },
+    {
+      "epoch": 4.823203087313073,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.7879,
+      "step": 20705
+    },
+    {
+      "epoch": 4.824409068982152,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.8706,
+      "step": 20710
+    },
+    {
+      "epoch": 4.82561505065123,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0078,
+      "step": 20715
+    },
+    {
+      "epoch": 4.826821032320309,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0377,
+      "step": 20720
+    },
+    {
+      "epoch": 4.828027013989387,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.913,
+      "step": 20725
+    },
+    {
+      "epoch": 4.829232995658466,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9725,
+      "step": 20730
+    },
+    {
+      "epoch": 4.830438977327544,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9714,
+      "step": 20735
+    },
+    {
+      "epoch": 4.831644958996623,
+      "grad_norm": 3.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9658,
+      "step": 20740
+    },
+    {
+      "epoch": 4.832850940665702,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8669,
+      "step": 20745
+    },
+    {
+      "epoch": 4.8340569223347805,
+      "grad_norm": 3.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0164,
+      "step": 20750
+    },
+    {
+      "epoch": 4.8352629040038595,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.7978,
+      "step": 20755
+    },
+    {
+      "epoch": 4.836468885672938,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9663,
+      "step": 20760
+    },
+    {
+      "epoch": 4.837674867342017,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0615,
+      "step": 20765
+    },
+    {
+      "epoch": 4.838880849011095,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8136,
+      "step": 20770
+    },
+    {
+      "epoch": 4.840086830680174,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.1187,
+      "step": 20775
+    },
+    {
+      "epoch": 4.841292812349252,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.9137,
+      "step": 20780
+    },
+    {
+      "epoch": 4.842498794018331,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8554,
+      "step": 20785
+    },
+    {
+      "epoch": 4.843704775687409,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9178,
+      "step": 20790
+    },
+    {
+      "epoch": 4.844910757356488,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.758,
+      "step": 20795
+    },
+    {
+      "epoch": 4.846116739025566,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9255,
+      "step": 20800
+    },
+    {
+      "epoch": 4.847322720694645,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0064,
+      "step": 20805
+    },
+    {
+      "epoch": 4.8485287023637245,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0108,
+      "step": 20810
+    },
+    {
+      "epoch": 4.849734684032803,
+      "grad_norm": 3.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.969,
+      "step": 20815
+    },
+    {
+      "epoch": 4.850940665701882,
+      "grad_norm": 1.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.1743,
+      "step": 20820
+    },
+    {
+      "epoch": 4.85214664737096,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9218,
+      "step": 20825
+    },
+    {
+      "epoch": 4.853352629040039,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9121,
+      "step": 20830
+    },
+    {
+      "epoch": 4.854558610709117,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9341,
+      "step": 20835
+    },
+    {
+      "epoch": 4.855764592378196,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8066,
+      "step": 20840
+    },
+    {
+      "epoch": 4.856970574047274,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9598,
+      "step": 20845
+    },
+    {
+      "epoch": 4.858176555716353,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8488,
+      "step": 20850
+    },
+    {
+      "epoch": 4.859382537385431,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9551,
+      "step": 20855
+    },
+    {
+      "epoch": 4.86058851905451,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.8587,
+      "step": 20860
+    },
+    {
+      "epoch": 4.8617945007235885,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8243,
+      "step": 20865
+    },
+    {
+      "epoch": 4.863000482392668,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0493,
+      "step": 20870
+    },
+    {
+      "epoch": 4.864206464061747,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0363,
+      "step": 20875
+    },
+    {
+      "epoch": 4.865412445730825,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0081,
+      "step": 20880
+    },
+    {
+      "epoch": 4.866618427399904,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.958,
+      "step": 20885
+    },
+    {
+      "epoch": 4.867824409068982,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.7951,
+      "step": 20890
+    },
+    {
+      "epoch": 4.869030390738061,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8614,
+      "step": 20895
+    },
+    {
+      "epoch": 4.870236372407139,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 4.03,
+      "step": 20900
+    },
+    {
+      "epoch": 4.871442354076218,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8715,
+      "step": 20905
+    },
+    {
+      "epoch": 4.872648335745296,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9043,
+      "step": 20910
+    },
+    {
+      "epoch": 4.873854317414375,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9583,
+      "step": 20915
+    },
+    {
+      "epoch": 4.875060299083454,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9146,
+      "step": 20920
+    },
+    {
+      "epoch": 4.8762662807525325,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.8752,
+      "step": 20925
+    },
+    {
+      "epoch": 4.877472262421611,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9535,
+      "step": 20930
+    },
+    {
+      "epoch": 4.87867824409069,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.7113,
+      "step": 20935
+    },
+    {
+      "epoch": 4.879884225759769,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.932,
+      "step": 20940
+    },
+    {
+      "epoch": 4.881090207428847,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.021,
+      "step": 20945
+    },
+    {
+      "epoch": 4.882296189097926,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9005,
+      "step": 20950
+    },
+    {
+      "epoch": 4.883502170767004,
+      "grad_norm": 1.9296875,
+      "learning_rate": 3e-05,
+      "loss": 3.7932,
+      "step": 20955
+    },
+    {
+      "epoch": 4.884708152436083,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.1057,
+      "step": 20960
+    },
+    {
+      "epoch": 4.885914134105161,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9113,
+      "step": 20965
+    },
+    {
+      "epoch": 4.88712011577424,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0753,
+      "step": 20970
+    },
+    {
+      "epoch": 4.8883260974433185,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8068,
+      "step": 20975
+    },
+    {
+      "epoch": 4.8895320791123975,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 4.0359,
+      "step": 20980
+    },
+    {
+      "epoch": 4.8907380607814765,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9581,
+      "step": 20985
+    },
+    {
+      "epoch": 4.891944042450555,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8416,
+      "step": 20990
+    },
+    {
+      "epoch": 4.893150024119634,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.996,
+      "step": 20995
+    },
+    {
+      "epoch": 4.894356005788712,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0332,
+      "step": 21000
+    },
+    {
+      "epoch": 4.895561987457791,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8406,
+      "step": 21005
+    },
+    {
+      "epoch": 4.896767969126869,
+      "grad_norm": 3.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.005,
+      "step": 21010
+    },
+    {
+      "epoch": 4.897973950795948,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.7687,
+      "step": 21015
+    },
+    {
+      "epoch": 4.899179932465026,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0207,
+      "step": 21020
+    },
+    {
+      "epoch": 4.900385914134105,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9626,
+      "step": 21025
+    },
+    {
+      "epoch": 4.901591895803183,
+      "grad_norm": 2.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.8243,
+      "step": 21030
+    },
+    {
+      "epoch": 4.9027978774722625,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0623,
+      "step": 21035
+    },
+    {
+      "epoch": 4.904003859141341,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 4.0195,
+      "step": 21040
+    },
+    {
+      "epoch": 4.90520984081042,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.8954,
+      "step": 21045
+    },
+    {
+      "epoch": 4.906415822479499,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0211,
+      "step": 21050
+    },
+    {
+      "epoch": 4.907621804148577,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9705,
+      "step": 21055
+    },
+    {
+      "epoch": 4.908827785817656,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9381,
+      "step": 21060
+    },
+    {
+      "epoch": 4.910033767486734,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9312,
+      "step": 21065
+    },
+    {
+      "epoch": 4.911239749155813,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8781,
+      "step": 21070
+    },
+    {
+      "epoch": 4.912445730824891,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.8277,
+      "step": 21075
+    },
+    {
+      "epoch": 4.91365171249397,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8868,
+      "step": 21080
+    },
+    {
+      "epoch": 4.914857694163048,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8644,
+      "step": 21085
+    },
+    {
+      "epoch": 4.9160636758321274,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.92,
+      "step": 21090
+    },
+    {
+      "epoch": 4.9172696575012065,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8915,
+      "step": 21095
+    },
+    {
+      "epoch": 4.918475639170285,
+      "grad_norm": 4.1875,
+      "learning_rate": 3e-05,
+      "loss": 4.0556,
+      "step": 21100
+    },
+    {
+      "epoch": 4.919681620839363,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9364,
+      "step": 21105
+    },
+    {
+      "epoch": 4.920887602508442,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9488,
+      "step": 21110
+    },
+    {
+      "epoch": 4.922093584177521,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8182,
+      "step": 21115
+    },
+    {
+      "epoch": 4.923299565846599,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8677,
+      "step": 21120
+    },
+    {
+      "epoch": 4.924505547515678,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9054,
+      "step": 21125
+    },
+    {
+      "epoch": 4.925711529184756,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9512,
+      "step": 21130
+    },
+    {
+      "epoch": 4.926917510853835,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.0463,
+      "step": 21135
+    },
+    {
+      "epoch": 4.928123492522913,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.782,
+      "step": 21140
+    },
+    {
+      "epoch": 4.929329474191992,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.8505,
+      "step": 21145
+    },
+    {
+      "epoch": 4.930535455861071,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.1929,
+      "step": 21150
+    },
+    {
+      "epoch": 4.93174143753015,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9182,
+      "step": 21155
+    },
+    {
+      "epoch": 4.932947419199229,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.9292,
+      "step": 21160
+    },
+    {
+      "epoch": 4.934153400868307,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.877,
+      "step": 21165
+    },
+    {
+      "epoch": 4.935359382537386,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9097,
+      "step": 21170
+    },
+    {
+      "epoch": 4.936565364206464,
+      "grad_norm": 3.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9824,
+      "step": 21175
+    },
+    {
+      "epoch": 4.937771345875543,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.8611,
+      "step": 21180
+    },
+    {
+      "epoch": 4.938977327544621,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.889,
+      "step": 21185
+    },
+    {
+      "epoch": 4.9401833092137,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9685,
+      "step": 21190
+    },
+    {
+      "epoch": 4.941389290882778,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9564,
+      "step": 21195
+    },
+    {
+      "epoch": 4.942595272551857,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.8476,
+      "step": 21200
+    },
+    {
+      "epoch": 4.9438012542209355,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.994,
+      "step": 21205
+    },
+    {
+      "epoch": 4.945007235890015,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9816,
+      "step": 21210
+    },
+    {
+      "epoch": 4.946213217559093,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8729,
+      "step": 21215
+    },
+    {
+      "epoch": 4.947419199228172,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0304,
+      "step": 21220
+    },
+    {
+      "epoch": 4.948625180897251,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8374,
+      "step": 21225
+    },
+    {
+      "epoch": 4.949831162566329,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9099,
+      "step": 21230
+    },
+    {
+      "epoch": 4.951037144235408,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 4.0382,
+      "step": 21235
+    },
+    {
+      "epoch": 4.952243125904486,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8176,
+      "step": 21240
+    },
+    {
+      "epoch": 4.953449107573565,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8897,
+      "step": 21245
+    },
+    {
+      "epoch": 4.954655089242643,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9768,
+      "step": 21250
+    },
+    {
+      "epoch": 4.955861070911722,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8429,
+      "step": 21255
+    },
+    {
+      "epoch": 4.9570670525808005,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.7557,
+      "step": 21260
+    },
+    {
+      "epoch": 4.9582730342498795,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.8978,
+      "step": 21265
+    },
+    {
+      "epoch": 4.959479015918958,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 4.064,
+      "step": 21270
+    },
+    {
+      "epoch": 4.960684997588037,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.7338,
+      "step": 21275
+    },
+    {
+      "epoch": 4.961890979257115,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8576,
+      "step": 21280
+    },
+    {
+      "epoch": 4.963096960926194,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.8683,
+      "step": 21285
+    },
+    {
+      "epoch": 4.964302942595273,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 4.0564,
+      "step": 21290
+    },
+    {
+      "epoch": 4.965508924264351,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.1415,
+      "step": 21295
+    },
+    {
+      "epoch": 4.96671490593343,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.7619,
+      "step": 21300
+    },
+    {
+      "epoch": 4.967920887602508,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.8409,
+      "step": 21305
+    },
+    {
+      "epoch": 4.969126869271587,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8675,
+      "step": 21310
+    },
+    {
+      "epoch": 4.9703328509406655,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.805,
+      "step": 21315
+    },
+    {
+      "epoch": 4.9715388326097445,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9616,
+      "step": 21320
+    },
+    {
+      "epoch": 4.972744814278823,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9983,
+      "step": 21325
+    },
+    {
+      "epoch": 4.973950795947902,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.0076,
+      "step": 21330
+    },
+    {
+      "epoch": 4.975156777616981,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8523,
+      "step": 21335
+    },
+    {
+      "epoch": 4.976362759286059,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8214,
+      "step": 21340
+    },
+    {
+      "epoch": 4.977568740955137,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8259,
+      "step": 21345
+    },
+    {
+      "epoch": 4.978774722624216,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9151,
+      "step": 21350
+    },
+    {
+      "epoch": 4.979980704293295,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0295,
+      "step": 21355
+    },
+    {
+      "epoch": 4.981186685962373,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0624,
+      "step": 21360
+    },
+    {
+      "epoch": 4.982392667631452,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8917,
+      "step": 21365
+    },
+    {
+      "epoch": 4.98359864930053,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.89,
+      "step": 21370
+    },
+    {
+      "epoch": 4.9848046309696095,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.857,
+      "step": 21375
+    },
+    {
+      "epoch": 4.986010612638688,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8069,
+      "step": 21380
+    },
+    {
+      "epoch": 4.987216594307767,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.7648,
+      "step": 21385
+    },
+    {
+      "epoch": 4.988422575976845,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0919,
+      "step": 21390
+    },
+    {
+      "epoch": 4.989628557645924,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.956,
+      "step": 21395
+    },
+    {
+      "epoch": 4.990834539315003,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8665,
+      "step": 21400
+    },
+    {
+      "epoch": 4.992040520984081,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8915,
+      "step": 21405
+    },
+    {
+      "epoch": 4.99324650265316,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.0173,
+      "step": 21410
+    },
+    {
+      "epoch": 4.994452484322238,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.7982,
+      "step": 21415
+    },
+    {
+      "epoch": 4.995658465991317,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8948,
+      "step": 21420
+    },
+    {
+      "epoch": 4.996864447660395,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.8085,
+      "step": 21425
+    },
+    {
+      "epoch": 4.998070429329474,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.9634,
+      "step": 21430
+    },
+    {
+      "epoch": 4.999276410998553,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 4.0828,
+      "step": 21435
+    },
+    {
+      "epoch": 5.000482392667632,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7937,
+      "step": 21440
+    },
+    {
+      "epoch": 5.00168837433671,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9184,
+      "step": 21445
+    },
+    {
+      "epoch": 5.002894356005789,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9379,
+      "step": 21450
+    },
+    {
+      "epoch": 5.004100337674867,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 4.0123,
+      "step": 21455
+    },
+    {
+      "epoch": 5.005306319343946,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.8683,
+      "step": 21460
+    },
+    {
+      "epoch": 5.006512301013025,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8978,
+      "step": 21465
+    },
+    {
+      "epoch": 5.007718282682103,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8722,
+      "step": 21470
+    },
+    {
+      "epoch": 5.008924264351182,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9295,
+      "step": 21475
+    },
+    {
+      "epoch": 5.01013024602026,
+      "grad_norm": 3.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1149,
+      "step": 21480
+    },
+    {
+      "epoch": 5.011336227689339,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8613,
+      "step": 21485
+    },
+    {
+      "epoch": 5.0125422093584175,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8367,
+      "step": 21490
+    },
+    {
+      "epoch": 5.013748191027497,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8382,
+      "step": 21495
+    },
+    {
+      "epoch": 5.014954172696575,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9636,
+      "step": 21500
+    },
+    {
+      "epoch": 5.016160154365654,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9675,
+      "step": 21505
+    },
+    {
+      "epoch": 5.017366136034732,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8949,
+      "step": 21510
+    },
+    {
+      "epoch": 5.018572117703811,
+      "grad_norm": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8907,
+      "step": 21515
+    },
+    {
+      "epoch": 5.019778099372889,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8416,
+      "step": 21520
+    },
+    {
+      "epoch": 5.020984081041968,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9238,
+      "step": 21525
+    },
+    {
+      "epoch": 5.022190062711047,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.959,
+      "step": 21530
+    },
+    {
+      "epoch": 5.023396044380125,
+      "grad_norm": 3.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.7752,
+      "step": 21535
+    },
+    {
+      "epoch": 5.024602026049204,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8786,
+      "step": 21540
+    },
+    {
+      "epoch": 5.0258080077182825,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9937,
+      "step": 21545
+    },
+    {
+      "epoch": 5.0270139893873615,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8521,
+      "step": 21550
+    },
+    {
+      "epoch": 5.02821997105644,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9118,
+      "step": 21555
+    },
+    {
+      "epoch": 5.029425952725519,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9575,
+      "step": 21560
+    },
+    {
+      "epoch": 5.030631934394597,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.8176,
+      "step": 21565
+    },
+    {
+      "epoch": 5.031837916063676,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9273,
+      "step": 21570
+    },
+    {
+      "epoch": 5.033043897732754,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9293,
+      "step": 21575
+    },
+    {
+      "epoch": 5.034249879401833,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.836,
+      "step": 21580
+    },
+    {
+      "epoch": 5.035455861070912,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8771,
+      "step": 21585
+    },
+    {
+      "epoch": 5.03666184273999,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9086,
+      "step": 21590
+    },
+    {
+      "epoch": 5.037867824409069,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0582,
+      "step": 21595
+    },
+    {
+      "epoch": 5.0390738060781475,
+      "grad_norm": 2.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.7782,
+      "step": 21600
+    },
+    {
+      "epoch": 5.0402797877472265,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0771,
+      "step": 21605
+    },
+    {
+      "epoch": 5.041485769416305,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9776,
+      "step": 21610
+    },
+    {
+      "epoch": 5.042691751085384,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9155,
+      "step": 21615
+    },
+    {
+      "epoch": 5.043897732754462,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8897,
+      "step": 21620
+    },
+    {
+      "epoch": 5.045103714423541,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9852,
+      "step": 21625
+    },
+    {
+      "epoch": 5.046309696092619,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9083,
+      "step": 21630
+    },
+    {
+      "epoch": 5.047515677761698,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9973,
+      "step": 21635
+    },
+    {
+      "epoch": 5.048721659430776,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.885,
+      "step": 21640
+    },
+    {
+      "epoch": 5.049927641099855,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8372,
+      "step": 21645
+    },
+    {
+      "epoch": 5.051133622768934,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9772,
+      "step": 21650
+    },
+    {
+      "epoch": 5.052339604438012,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.947,
+      "step": 21655
+    },
+    {
+      "epoch": 5.0535455861070915,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0524,
+      "step": 21660
+    },
+    {
+      "epoch": 5.05475156777617,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9114,
+      "step": 21665
+    },
+    {
+      "epoch": 5.055957549445249,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.845,
+      "step": 21670
+    },
+    {
+      "epoch": 5.057163531114327,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9679,
+      "step": 21675
+    },
+    {
+      "epoch": 5.058369512783406,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.894,
+      "step": 21680
+    },
+    {
+      "epoch": 5.059575494452484,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.7347,
+      "step": 21685
+    },
+    {
+      "epoch": 5.060781476121563,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.7781,
+      "step": 21690
+    },
+    {
+      "epoch": 5.061987457790641,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.8644,
+      "step": 21695
+    },
+    {
+      "epoch": 5.06319343945972,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.898,
+      "step": 21700
+    },
+    {
+      "epoch": 5.064399421128799,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9921,
+      "step": 21705
+    },
+    {
+      "epoch": 5.065605402797877,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.7472,
+      "step": 21710
+    },
+    {
+      "epoch": 5.0668113844669564,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9397,
+      "step": 21715
+    },
+    {
+      "epoch": 5.068017366136035,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8533,
+      "step": 21720
+    },
+    {
+      "epoch": 5.069223347805114,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9585,
+      "step": 21725
+    },
+    {
+      "epoch": 5.070429329474192,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.934,
+      "step": 21730
+    },
+    {
+      "epoch": 5.071635311143271,
+      "grad_norm": 3.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.7228,
+      "step": 21735
+    },
+    {
+      "epoch": 5.072841292812349,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9274,
+      "step": 21740
+    },
+    {
+      "epoch": 5.074047274481428,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9427,
+      "step": 21745
+    },
+    {
+      "epoch": 5.075253256150506,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.865,
+      "step": 21750
+    },
+    {
+      "epoch": 5.076459237819585,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9054,
+      "step": 21755
+    },
+    {
+      "epoch": 5.077665219488663,
+      "grad_norm": 3.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9283,
+      "step": 21760
+    },
+    {
+      "epoch": 5.078871201157742,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.6981,
+      "step": 21765
+    },
+    {
+      "epoch": 5.080077182826821,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8309,
+      "step": 21770
+    },
+    {
+      "epoch": 5.0812831644959,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9549,
+      "step": 21775
+    },
+    {
+      "epoch": 5.082489146164979,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9689,
+      "step": 21780
+    },
+    {
+      "epoch": 5.083695127834057,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.957,
+      "step": 21785
+    },
+    {
+      "epoch": 5.084901109503136,
+      "grad_norm": 3.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.922,
+      "step": 21790
+    },
+    {
+      "epoch": 5.086107091172214,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9898,
+      "step": 21795
+    },
+    {
+      "epoch": 5.087313072841293,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9402,
+      "step": 21800
+    },
+    {
+      "epoch": 5.088519054510371,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 4.0113,
+      "step": 21805
+    },
+    {
+      "epoch": 5.08972503617945,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.7446,
+      "step": 21810
+    },
+    {
+      "epoch": 5.090931017848528,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8729,
+      "step": 21815
+    },
+    {
+      "epoch": 5.092136999517607,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9414,
+      "step": 21820
+    },
+    {
+      "epoch": 5.093342981186686,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8305,
+      "step": 21825
+    },
+    {
+      "epoch": 5.0945489628557645,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9127,
+      "step": 21830
+    },
+    {
+      "epoch": 5.095754944524844,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9617,
+      "step": 21835
+    },
+    {
+      "epoch": 5.096960926193922,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9096,
+      "step": 21840
+    },
+    {
+      "epoch": 5.098166907863001,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9059,
+      "step": 21845
+    },
+    {
+      "epoch": 5.099372889532079,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.1668,
+      "step": 21850
+    },
+    {
+      "epoch": 5.100578871201158,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.8048,
+      "step": 21855
+    },
+    {
+      "epoch": 5.101784852870236,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8474,
+      "step": 21860
+    },
+    {
+      "epoch": 5.102990834539315,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.8594,
+      "step": 21865
+    },
+    {
+      "epoch": 5.104196816208393,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.093,
+      "step": 21870
+    },
+    {
+      "epoch": 5.105402797877472,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9099,
+      "step": 21875
+    },
+    {
+      "epoch": 5.1066087795465505,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.8213,
+      "step": 21880
+    },
+    {
+      "epoch": 5.1078147612156295,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8325,
+      "step": 21885
+    },
+    {
+      "epoch": 5.1090207428847085,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9916,
+      "step": 21890
+    },
+    {
+      "epoch": 5.110226724553787,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9432,
+      "step": 21895
+    },
+    {
+      "epoch": 5.111432706222866,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.8655,
+      "step": 21900
+    },
+    {
+      "epoch": 5.112638687891944,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.0019,
+      "step": 21905
+    },
+    {
+      "epoch": 5.113844669561023,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8892,
+      "step": 21910
+    },
+    {
+      "epoch": 5.115050651230101,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.94,
+      "step": 21915
+    },
+    {
+      "epoch": 5.11625663289918,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9665,
+      "step": 21920
+    },
+    {
+      "epoch": 5.117462614568258,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.096,
+      "step": 21925
+    },
+    {
+      "epoch": 5.118668596237337,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8654,
+      "step": 21930
+    },
+    {
+      "epoch": 5.119874577906415,
+      "grad_norm": 3.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8154,
+      "step": 21935
+    },
+    {
+      "epoch": 5.1210805595754945,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9213,
+      "step": 21940
+    },
+    {
+      "epoch": 5.1222865412445735,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8733,
+      "step": 21945
+    },
+    {
+      "epoch": 5.123492522913652,
+      "grad_norm": 3.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.7978,
+      "step": 21950
+    },
+    {
+      "epoch": 5.124698504582731,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.7867,
+      "step": 21955
+    },
+    {
+      "epoch": 5.125904486251809,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0244,
+      "step": 21960
+    },
+    {
+      "epoch": 5.127110467920888,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9427,
+      "step": 21965
+    },
+    {
+      "epoch": 5.128316449589966,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9321,
+      "step": 21970
+    },
+    {
+      "epoch": 5.129522431259045,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.8548,
+      "step": 21975
+    },
+    {
+      "epoch": 5.130728412928123,
+      "grad_norm": 3.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8351,
+      "step": 21980
+    },
+    {
+      "epoch": 5.131934394597202,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9124,
+      "step": 21985
+    },
+    {
+      "epoch": 5.13314037626628,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 4.0818,
+      "step": 21990
+    },
+    {
+      "epoch": 5.134346357935359,
+      "grad_norm": 2.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.7985,
+      "step": 21995
+    },
+    {
+      "epoch": 5.135552339604438,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.0759,
+      "step": 22000
+    },
+    {
+      "epoch": 5.136758321273517,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.8537,
+      "step": 22005
+    },
+    {
+      "epoch": 5.137964302942596,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.7829,
+      "step": 22010
+    },
+    {
+      "epoch": 5.139170284611674,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.923,
+      "step": 22015
+    },
+    {
+      "epoch": 5.140376266280753,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9953,
+      "step": 22020
+    },
+    {
+      "epoch": 5.141582247949831,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9761,
+      "step": 22025
+    },
+    {
+      "epoch": 5.14278822961891,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.7898,
+      "step": 22030
+    },
+    {
+      "epoch": 5.143994211287988,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8231,
+      "step": 22035
+    },
+    {
+      "epoch": 5.145200192957067,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.8964,
+      "step": 22040
+    },
+    {
+      "epoch": 5.146406174626145,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9405,
+      "step": 22045
+    },
+    {
+      "epoch": 5.147612156295224,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9334,
+      "step": 22050
+    },
+    {
+      "epoch": 5.1488181379643025,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8232,
+      "step": 22055
+    },
+    {
+      "epoch": 5.150024119633382,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8534,
+      "step": 22060
+    },
+    {
+      "epoch": 5.151230101302461,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.7698,
+      "step": 22065
+    },
+    {
+      "epoch": 5.152436082971539,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.797,
+      "step": 22070
+    },
+    {
+      "epoch": 5.153642064640618,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.94,
+      "step": 22075
+    },
+    {
+      "epoch": 5.154848046309696,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9255,
+      "step": 22080
+    },
+    {
+      "epoch": 5.156054027978775,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.0454,
+      "step": 22085
+    },
+    {
+      "epoch": 5.157260009647853,
+      "grad_norm": 3.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9132,
+      "step": 22090
+    },
+    {
+      "epoch": 5.158465991316932,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0732,
+      "step": 22095
+    },
+    {
+      "epoch": 5.15967197298601,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9151,
+      "step": 22100
+    },
+    {
+      "epoch": 5.160877954655089,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.8621,
+      "step": 22105
+    },
+    {
+      "epoch": 5.1620839363241675,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.701,
+      "step": 22110
+    },
+    {
+      "epoch": 5.1632899179932465,
+      "grad_norm": 2.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9158,
+      "step": 22115
+    },
+    {
+      "epoch": 5.164495899662326,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9542,
+      "step": 22120
+    },
+    {
+      "epoch": 5.165701881331404,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9618,
+      "step": 22125
+    },
+    {
+      "epoch": 5.166907863000483,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8922,
+      "step": 22130
+    },
+    {
+      "epoch": 5.168113844669561,
+      "grad_norm": 3.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8107,
+      "step": 22135
+    },
+    {
+      "epoch": 5.16931982633864,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8175,
+      "step": 22140
+    },
+    {
+      "epoch": 5.170525808007718,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9216,
+      "step": 22145
+    },
+    {
+      "epoch": 5.171731789676797,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8877,
+      "step": 22150
+    },
+    {
+      "epoch": 5.172937771345875,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.8928,
+      "step": 22155
+    },
+    {
+      "epoch": 5.174143753014954,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 4.0834,
+      "step": 22160
+    },
+    {
+      "epoch": 5.1753497346840325,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.8334,
+      "step": 22165
+    },
+    {
+      "epoch": 5.1765557163531115,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9299,
+      "step": 22170
+    },
+    {
+      "epoch": 5.17776169802219,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8922,
+      "step": 22175
+    },
+    {
+      "epoch": 5.178967679691269,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8433,
+      "step": 22180
+    },
+    {
+      "epoch": 5.180173661360348,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9261,
+      "step": 22185
+    },
+    {
+      "epoch": 5.181379643029426,
+      "grad_norm": 3.421875,
+      "learning_rate": 3e-05,
+      "loss": 4.0145,
+      "step": 22190
+    },
+    {
+      "epoch": 5.182585624698505,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.978,
+      "step": 22195
+    },
+    {
+      "epoch": 5.183791606367583,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8131,
+      "step": 22200
+    },
+    {
+      "epoch": 5.184997588036662,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.8114,
+      "step": 22205
+    },
+    {
+      "epoch": 5.18620356970574,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9253,
+      "step": 22210
+    },
+    {
+      "epoch": 5.187409551374819,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8948,
+      "step": 22215
+    },
+    {
+      "epoch": 5.188615533043897,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8952,
+      "step": 22220
+    },
+    {
+      "epoch": 5.1898215147129765,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 4.0042,
+      "step": 22225
+    },
+    {
+      "epoch": 5.191027496382055,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8473,
+      "step": 22230
+    },
+    {
+      "epoch": 5.192233478051134,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.1594,
+      "step": 22235
+    },
+    {
+      "epoch": 5.193439459720212,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.927,
+      "step": 22240
+    },
+    {
+      "epoch": 5.194645441389291,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.859,
+      "step": 22245
+    },
+    {
+      "epoch": 5.19585142305837,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8284,
+      "step": 22250
+    },
+    {
+      "epoch": 5.197057404727448,
+      "grad_norm": 3.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.8632,
+      "step": 22255
+    },
+    {
+      "epoch": 5.198263386396527,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.7852,
+      "step": 22260
+    },
+    {
+      "epoch": 5.199469368065605,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9096,
+      "step": 22265
+    },
+    {
+      "epoch": 5.200675349734684,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9676,
+      "step": 22270
+    },
+    {
+      "epoch": 5.201881331403762,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.7712,
+      "step": 22275
+    },
+    {
+      "epoch": 5.2030873130728414,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0899,
+      "step": 22280
+    },
+    {
+      "epoch": 5.20429329474192,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.0446,
+      "step": 22285
+    },
+    {
+      "epoch": 5.205499276410999,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9552,
+      "step": 22290
+    },
+    {
+      "epoch": 5.206705258080077,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8262,
+      "step": 22295
+    },
+    {
+      "epoch": 5.207911239749156,
+      "grad_norm": 3.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9497,
+      "step": 22300
+    },
+    {
+      "epoch": 5.209117221418235,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.953,
+      "step": 22305
+    },
+    {
+      "epoch": 5.210323203087313,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9856,
+      "step": 22310
+    },
+    {
+      "epoch": 5.211529184756392,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.7899,
+      "step": 22315
+    },
+    {
+      "epoch": 5.21273516642547,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8281,
+      "step": 22320
+    },
+    {
+      "epoch": 5.213941148094549,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9231,
+      "step": 22325
+    },
+    {
+      "epoch": 5.215147129763627,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9871,
+      "step": 22330
+    },
+    {
+      "epoch": 5.216353111432706,
+      "grad_norm": 3.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9694,
+      "step": 22335
+    },
+    {
+      "epoch": 5.217559093101785,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8839,
+      "step": 22340
+    },
+    {
+      "epoch": 5.218765074770864,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.8151,
+      "step": 22345
+    },
+    {
+      "epoch": 5.219971056439942,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8759,
+      "step": 22350
+    },
+    {
+      "epoch": 5.221177038109021,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9434,
+      "step": 22355
+    },
+    {
+      "epoch": 5.2223830197781,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.7774,
+      "step": 22360
+    },
+    {
+      "epoch": 5.223589001447178,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.8739,
+      "step": 22365
+    },
+    {
+      "epoch": 5.224794983116257,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9417,
+      "step": 22370
+    },
+    {
+      "epoch": 5.226000964785335,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8387,
+      "step": 22375
+    },
+    {
+      "epoch": 5.227206946454414,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8646,
+      "step": 22380
+    },
+    {
+      "epoch": 5.228412928123492,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.7746,
+      "step": 22385
+    },
+    {
+      "epoch": 5.229618909792571,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.8178,
+      "step": 22390
+    },
+    {
+      "epoch": 5.2308248914616495,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8928,
+      "step": 22395
+    },
+    {
+      "epoch": 5.232030873130729,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8171,
+      "step": 22400
+    },
+    {
+      "epoch": 5.233236854799807,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 4.0171,
+      "step": 22405
+    },
+    {
+      "epoch": 5.234442836468886,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 4.0651,
+      "step": 22410
+    },
+    {
+      "epoch": 5.235648818137964,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8623,
+      "step": 22415
+    },
+    {
+      "epoch": 5.236854799807043,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8643,
+      "step": 22420
+    },
+    {
+      "epoch": 5.238060781476122,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8966,
+      "step": 22425
+    },
+    {
+      "epoch": 5.2392667631452,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8332,
+      "step": 22430
+    },
+    {
+      "epoch": 5.240472744814279,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8822,
+      "step": 22435
+    },
+    {
+      "epoch": 5.241678726483357,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8462,
+      "step": 22440
+    },
+    {
+      "epoch": 5.242884708152436,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0783,
+      "step": 22445
+    },
+    {
+      "epoch": 5.2440906898215145,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8279,
+      "step": 22450
+    },
+    {
+      "epoch": 5.2452966714905935,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9815,
+      "step": 22455
+    },
+    {
+      "epoch": 5.246502653159672,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8487,
+      "step": 22460
+    },
+    {
+      "epoch": 5.247708634828751,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9799,
+      "step": 22465
+    },
+    {
+      "epoch": 5.248914616497829,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.7613,
+      "step": 22470
+    },
+    {
+      "epoch": 5.250120598166908,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8872,
+      "step": 22475
+    },
+    {
+      "epoch": 5.251326579835986,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8842,
+      "step": 22480
+    },
+    {
+      "epoch": 5.252532561505065,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9607,
+      "step": 22485
+    },
+    {
+      "epoch": 5.253738543174144,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.7689,
+      "step": 22490
+    },
+    {
+      "epoch": 5.254944524843222,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9501,
+      "step": 22495
+    },
+    {
+      "epoch": 5.256150506512301,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.7814,
+      "step": 22500
+    },
+    {
+      "epoch": 5.2573564881813795,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.833,
+      "step": 22505
+    },
+    {
+      "epoch": 5.2585624698504585,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.8897,
+      "step": 22510
+    },
+    {
+      "epoch": 5.259768451519537,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.0058,
+      "step": 22515
+    },
+    {
+      "epoch": 5.260974433188616,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8924,
+      "step": 22520
+    },
+    {
+      "epoch": 5.262180414857694,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9175,
+      "step": 22525
+    },
+    {
+      "epoch": 5.263386396526773,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.7812,
+      "step": 22530
+    },
+    {
+      "epoch": 5.264592378195851,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.8766,
+      "step": 22535
+    },
+    {
+      "epoch": 5.26579835986493,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9242,
+      "step": 22540
+    },
+    {
+      "epoch": 5.267004341534009,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.7476,
+      "step": 22545
+    },
+    {
+      "epoch": 5.268210323203087,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.8466,
+      "step": 22550
+    },
+    {
+      "epoch": 5.269416304872166,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9543,
+      "step": 22555
+    },
+    {
+      "epoch": 5.270622286541244,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.7943,
+      "step": 22560
+    },
+    {
+      "epoch": 5.2718282682103235,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.8315,
+      "step": 22565
+    },
+    {
+      "epoch": 5.273034249879402,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.7683,
+      "step": 22570
+    },
+    {
+      "epoch": 5.274240231548481,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8152,
+      "step": 22575
+    },
+    {
+      "epoch": 5.275446213217559,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8659,
+      "step": 22580
+    },
+    {
+      "epoch": 5.276652194886638,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.7819,
+      "step": 22585
+    },
+    {
+      "epoch": 5.277858176555716,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 4.0173,
+      "step": 22590
+    },
+    {
+      "epoch": 5.279064158224795,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0548,
+      "step": 22595
+    },
+    {
+      "epoch": 5.280270139893874,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8177,
+      "step": 22600
+    },
+    {
+      "epoch": 5.281476121562952,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.875,
+      "step": 22605
+    },
+    {
+      "epoch": 5.282682103232031,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8638,
+      "step": 22610
+    },
+    {
+      "epoch": 5.283888084901109,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8335,
+      "step": 22615
+    },
+    {
+      "epoch": 5.285094066570188,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8747,
+      "step": 22620
+    },
+    {
+      "epoch": 5.286300048239267,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8161,
+      "step": 22625
+    },
+    {
+      "epoch": 5.287506029908346,
+      "grad_norm": 3.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.041,
+      "step": 22630
+    },
+    {
+      "epoch": 5.288712011577424,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8093,
+      "step": 22635
+    },
+    {
+      "epoch": 5.289917993246503,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9673,
+      "step": 22640
+    },
+    {
+      "epoch": 5.291123974915581,
+      "grad_norm": 2.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.8403,
+      "step": 22645
+    },
+    {
+      "epoch": 5.29232995658466,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9492,
+      "step": 22650
+    },
+    {
+      "epoch": 5.293535938253738,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.7767,
+      "step": 22655
+    },
+    {
+      "epoch": 5.294741919922817,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8635,
+      "step": 22660
+    },
+    {
+      "epoch": 5.295947901591896,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8812,
+      "step": 22665
+    },
+    {
+      "epoch": 5.297153883260974,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9282,
+      "step": 22670
+    },
+    {
+      "epoch": 5.298359864930053,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.7618,
+      "step": 22675
+    },
+    {
+      "epoch": 5.2995658465991315,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.9495,
+      "step": 22680
+    },
+    {
+      "epoch": 5.300771828268211,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.915,
+      "step": 22685
+    },
+    {
+      "epoch": 5.301977809937289,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9178,
+      "step": 22690
+    },
+    {
+      "epoch": 5.303183791606368,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9037,
+      "step": 22695
+    },
+    {
+      "epoch": 5.304389773275446,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.7155,
+      "step": 22700
+    },
+    {
+      "epoch": 5.305595754944525,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8979,
+      "step": 22705
+    },
+    {
+      "epoch": 5.306801736613603,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.8654,
+      "step": 22710
+    },
+    {
+      "epoch": 5.308007718282682,
+      "grad_norm": 3.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9831,
+      "step": 22715
+    },
+    {
+      "epoch": 5.30921369995176,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8501,
+      "step": 22720
+    },
+    {
+      "epoch": 5.310419681620839,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8332,
+      "step": 22725
+    },
+    {
+      "epoch": 5.311625663289918,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.6915,
+      "step": 22730
+    },
+    {
+      "epoch": 5.3128316449589965,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9509,
+      "step": 22735
+    },
+    {
+      "epoch": 5.3140376266280756,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.9541,
+      "step": 22740
+    },
+    {
+      "epoch": 5.315243608297154,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9391,
+      "step": 22745
+    },
+    {
+      "epoch": 5.316449589966233,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8767,
+      "step": 22750
+    },
+    {
+      "epoch": 5.317655571635311,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9737,
+      "step": 22755
+    },
+    {
+      "epoch": 5.31886155330439,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9249,
+      "step": 22760
+    },
+    {
+      "epoch": 5.320067534973468,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.8289,
+      "step": 22765
+    },
+    {
+      "epoch": 5.321273516642547,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.7654,
+      "step": 22770
+    },
+    {
+      "epoch": 5.322479498311625,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8648,
+      "step": 22775
+    },
+    {
+      "epoch": 5.323685479980704,
+      "grad_norm": 3.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.951,
+      "step": 22780
+    },
+    {
+      "epoch": 5.324891461649783,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.9552,
+      "step": 22785
+    },
+    {
+      "epoch": 5.3260974433188615,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.9069,
+      "step": 22790
+    },
+    {
+      "epoch": 5.3273034249879405,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0112,
+      "step": 22795
+    },
+    {
+      "epoch": 5.328509406657019,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8401,
+      "step": 22800
+    },
+    {
+      "epoch": 5.329715388326098,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9011,
+      "step": 22805
+    },
+    {
+      "epoch": 5.330921369995176,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0513,
+      "step": 22810
+    },
+    {
+      "epoch": 5.332127351664255,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8436,
+      "step": 22815
+    },
+    {
+      "epoch": 5.333333333333333,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8568,
+      "step": 22820
+    },
+    {
+      "epoch": 5.334539315002412,
+      "grad_norm": 3.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8491,
+      "step": 22825
+    },
+    {
+      "epoch": 5.33574529667149,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8814,
+      "step": 22830
+    },
+    {
+      "epoch": 5.336951278340569,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9669,
+      "step": 22835
+    },
+    {
+      "epoch": 5.338157260009648,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9322,
+      "step": 22840
+    },
+    {
+      "epoch": 5.3393632416787264,
+      "grad_norm": 3.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0186,
+      "step": 22845
+    },
+    {
+      "epoch": 5.3405692233478055,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9937,
+      "step": 22850
+    },
+    {
+      "epoch": 5.341775205016884,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9043,
+      "step": 22855
+    },
+    {
+      "epoch": 5.342981186685963,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.8663,
+      "step": 22860
+    },
+    {
+      "epoch": 5.344187168355041,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8442,
+      "step": 22865
+    },
+    {
+      "epoch": 5.34539315002412,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9621,
+      "step": 22870
+    },
+    {
+      "epoch": 5.346599131693198,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 4.0337,
+      "step": 22875
+    },
+    {
+      "epoch": 5.347805113362277,
+      "grad_norm": 3.328125,
+      "learning_rate": 3e-05,
+      "loss": 4.0618,
+      "step": 22880
+    },
+    {
+      "epoch": 5.349011095031355,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.965,
+      "step": 22885
+    },
+    {
+      "epoch": 5.350217076700434,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.819,
+      "step": 22890
+    },
+    {
+      "epoch": 5.351423058369512,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.8295,
+      "step": 22895
+    },
+    {
+      "epoch": 5.352629040038591,
+      "grad_norm": 3.34375,
+      "learning_rate": 3e-05,
+      "loss": 4.0718,
+      "step": 22900
+    },
+    {
+      "epoch": 5.3538350217076704,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.889,
+      "step": 22905
+    },
+    {
+      "epoch": 5.355041003376749,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.963,
+      "step": 22910
+    },
+    {
+      "epoch": 5.356246985045828,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9911,
+      "step": 22915
+    },
+    {
+      "epoch": 5.357452966714906,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.7436,
+      "step": 22920
+    },
+    {
+      "epoch": 5.358658948383985,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8823,
+      "step": 22925
+    },
+    {
+      "epoch": 5.359864930053063,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9227,
+      "step": 22930
+    },
+    {
+      "epoch": 5.361070911722142,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.8426,
+      "step": 22935
+    },
+    {
+      "epoch": 5.36227689339122,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9322,
+      "step": 22940
+    },
+    {
+      "epoch": 5.363482875060299,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8069,
+      "step": 22945
+    },
+    {
+      "epoch": 5.364688856729377,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8967,
+      "step": 22950
+    },
+    {
+      "epoch": 5.365894838398456,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.7528,
+      "step": 22955
+    },
+    {
+      "epoch": 5.3671008200675345,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9744,
+      "step": 22960
+    },
+    {
+      "epoch": 5.368306801736614,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 4.1065,
+      "step": 22965
+    },
+    {
+      "epoch": 5.369512783405693,
+      "grad_norm": 4.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9464,
+      "step": 22970
+    },
+    {
+      "epoch": 5.370718765074771,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9455,
+      "step": 22975
+    },
+    {
+      "epoch": 5.37192474674385,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.7549,
+      "step": 22980
+    },
+    {
+      "epoch": 5.373130728412928,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.931,
+      "step": 22985
+    },
+    {
+      "epoch": 5.374336710082007,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8414,
+      "step": 22990
+    },
+    {
+      "epoch": 5.375542691751085,
+      "grad_norm": 3.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9864,
+      "step": 22995
+    },
+    {
+      "epoch": 5.376748673420164,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.7807,
+      "step": 23000
+    },
+    {
+      "epoch": 5.377954655089242,
+      "grad_norm": 3.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.7735,
+      "step": 23005
+    },
+    {
+      "epoch": 5.379160636758321,
+      "grad_norm": 3.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.8382,
+      "step": 23010
+    },
+    {
+      "epoch": 5.3803666184273995,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 4.0343,
+      "step": 23015
+    },
+    {
+      "epoch": 5.3815726000964785,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8926,
+      "step": 23020
+    },
+    {
+      "epoch": 5.382778581765558,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8172,
+      "step": 23025
+    },
+    {
+      "epoch": 5.383984563434636,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8929,
+      "step": 23030
+    },
+    {
+      "epoch": 5.385190545103715,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8203,
+      "step": 23035
+    },
+    {
+      "epoch": 5.386396526772793,
+      "grad_norm": 3.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8249,
+      "step": 23040
+    },
+    {
+      "epoch": 5.387602508441872,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.0976,
+      "step": 23045
+    },
+    {
+      "epoch": 5.38880849011095,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9134,
+      "step": 23050
+    },
+    {
+      "epoch": 5.390014471780029,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8834,
+      "step": 23055
+    },
+    {
+      "epoch": 5.391220453449107,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.883,
+      "step": 23060
+    },
+    {
+      "epoch": 5.392426435118186,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.905,
+      "step": 23065
+    },
+    {
+      "epoch": 5.3936324167872645,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8558,
+      "step": 23070
+    },
+    {
+      "epoch": 5.3948383984563435,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9013,
+      "step": 23075
+    },
+    {
+      "epoch": 5.3960443801254225,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9328,
+      "step": 23080
+    },
+    {
+      "epoch": 5.397250361794501,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8126,
+      "step": 23085
+    },
+    {
+      "epoch": 5.39845634346358,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0001,
+      "step": 23090
+    },
+    {
+      "epoch": 5.399662325132658,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.8751,
+      "step": 23095
+    },
+    {
+      "epoch": 5.400868306801737,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.972,
+      "step": 23100
+    },
+    {
+      "epoch": 5.402074288470815,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 4.0142,
+      "step": 23105
+    },
+    {
+      "epoch": 5.403280270139894,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0198,
+      "step": 23110
+    },
+    {
+      "epoch": 5.404486251808972,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9366,
+      "step": 23115
+    },
+    {
+      "epoch": 5.405692233478051,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0044,
+      "step": 23120
+    },
+    {
+      "epoch": 5.406898215147129,
+      "grad_norm": 3.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1258,
+      "step": 23125
+    },
+    {
+      "epoch": 5.4081041968162085,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9112,
+      "step": 23130
+    },
+    {
+      "epoch": 5.409310178485287,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.8063,
+      "step": 23135
+    },
+    {
+      "epoch": 5.410516160154366,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 4.0434,
+      "step": 23140
+    },
+    {
+      "epoch": 5.411722141823445,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.878,
+      "step": 23145
+    },
+    {
+      "epoch": 5.412928123492523,
+      "grad_norm": 3.171875,
+      "learning_rate": 3e-05,
+      "loss": 3.6575,
+      "step": 23150
+    },
+    {
+      "epoch": 5.414134105161602,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.882,
+      "step": 23155
+    },
+    {
+      "epoch": 5.41534008683068,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.1331,
+      "step": 23160
+    },
+    {
+      "epoch": 5.416546068499759,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.7655,
+      "step": 23165
+    },
+    {
+      "epoch": 5.417752050168837,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7729,
+      "step": 23170
+    },
+    {
+      "epoch": 5.418958031837916,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.957,
+      "step": 23175
+    },
+    {
+      "epoch": 5.420164013506994,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9156,
+      "step": 23180
+    },
+    {
+      "epoch": 5.421369995176073,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.856,
+      "step": 23185
+    },
+    {
+      "epoch": 5.422575976845152,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9478,
+      "step": 23190
+    },
+    {
+      "epoch": 5.423781958514231,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9211,
+      "step": 23195
+    },
+    {
+      "epoch": 5.424987940183309,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8786,
+      "step": 23200
+    },
+    {
+      "epoch": 5.426193921852388,
+      "grad_norm": 3.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8665,
+      "step": 23205
+    },
+    {
+      "epoch": 5.427399903521467,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.8091,
+      "step": 23210
+    },
+    {
+      "epoch": 5.428605885190545,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9709,
+      "step": 23215
+    },
+    {
+      "epoch": 5.429811866859624,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.7431,
+      "step": 23220
+    },
+    {
+      "epoch": 5.431017848528702,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.857,
+      "step": 23225
+    },
+    {
+      "epoch": 5.432223830197781,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9715,
+      "step": 23230
+    },
+    {
+      "epoch": 5.433429811866859,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9681,
+      "step": 23235
+    },
+    {
+      "epoch": 5.434635793535938,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.938,
+      "step": 23240
+    },
+    {
+      "epoch": 5.4358417752050165,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8178,
+      "step": 23245
+    },
+    {
+      "epoch": 5.437047756874096,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8495,
+      "step": 23250
+    },
+    {
+      "epoch": 5.438253738543174,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8898,
+      "step": 23255
+    },
+    {
+      "epoch": 5.439459720212253,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8435,
+      "step": 23260
+    },
+    {
+      "epoch": 5.440665701881332,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0859,
+      "step": 23265
+    },
+    {
+      "epoch": 5.44187168355041,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8726,
+      "step": 23270
+    },
+    {
+      "epoch": 5.443077665219489,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0161,
+      "step": 23275
+    },
+    {
+      "epoch": 5.444283646888567,
+      "grad_norm": 3.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8961,
+      "step": 23280
+    },
+    {
+      "epoch": 5.445489628557646,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.7977,
+      "step": 23285
+    },
+    {
+      "epoch": 5.446695610226724,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9329,
+      "step": 23290
+    },
+    {
+      "epoch": 5.447901591895803,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.9534,
+      "step": 23295
+    },
+    {
+      "epoch": 5.4491075735648815,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8252,
+      "step": 23300
+    },
+    {
+      "epoch": 5.4503135552339605,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9868,
+      "step": 23305
+    },
+    {
+      "epoch": 5.451519536903039,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9234,
+      "step": 23310
+    },
+    {
+      "epoch": 5.452725518572118,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9167,
+      "step": 23315
+    },
+    {
+      "epoch": 5.453931500241197,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.8283,
+      "step": 23320
+    },
+    {
+      "epoch": 5.455137481910275,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.895,
+      "step": 23325
+    },
+    {
+      "epoch": 5.456343463579354,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.7375,
+      "step": 23330
+    },
+    {
+      "epoch": 5.457549445248432,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9086,
+      "step": 23335
+    },
+    {
+      "epoch": 5.458755426917511,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9871,
+      "step": 23340
+    },
+    {
+      "epoch": 5.459961408586589,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8431,
+      "step": 23345
+    },
+    {
+      "epoch": 5.461167390255668,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9762,
+      "step": 23350
+    },
+    {
+      "epoch": 5.4623733719247465,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9696,
+      "step": 23355
+    },
+    {
+      "epoch": 5.4635793535938255,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.913,
+      "step": 23360
+    },
+    {
+      "epoch": 5.464785335262904,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9877,
+      "step": 23365
+    },
+    {
+      "epoch": 5.465991316931983,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.9073,
+      "step": 23370
+    },
+    {
+      "epoch": 5.467197298601061,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9377,
+      "step": 23375
+    },
+    {
+      "epoch": 5.46840328027014,
+      "grad_norm": 3.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0128,
+      "step": 23380
+    },
+    {
+      "epoch": 5.469609261939219,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 4.0208,
+      "step": 23385
+    },
+    {
+      "epoch": 5.470815243608297,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.939,
+      "step": 23390
+    },
+    {
+      "epoch": 5.472021225277376,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9935,
+      "step": 23395
+    },
+    {
+      "epoch": 5.473227206946454,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9437,
+      "step": 23400
+    },
+    {
+      "epoch": 5.474433188615533,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0159,
+      "step": 23405
+    },
+    {
+      "epoch": 5.4756391702846114,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.9312,
+      "step": 23410
+    },
+    {
+      "epoch": 5.4768451519536905,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0047,
+      "step": 23415
+    },
+    {
+      "epoch": 5.478051133622769,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8874,
+      "step": 23420
+    },
+    {
+      "epoch": 5.479257115291848,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9742,
+      "step": 23425
+    },
+    {
+      "epoch": 5.480463096960926,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9367,
+      "step": 23430
+    },
+    {
+      "epoch": 5.481669078630005,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8547,
+      "step": 23435
+    },
+    {
+      "epoch": 5.482875060299083,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9652,
+      "step": 23440
+    },
+    {
+      "epoch": 5.484081041968162,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.838,
+      "step": 23445
+    },
+    {
+      "epoch": 5.485287023637241,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.7544,
+      "step": 23450
+    },
+    {
+      "epoch": 5.486493005306319,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9559,
+      "step": 23455
+    },
+    {
+      "epoch": 5.487698986975398,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9173,
+      "step": 23460
+    },
+    {
+      "epoch": 5.488904968644476,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8572,
+      "step": 23465
+    },
+    {
+      "epoch": 5.4901109503135554,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9472,
+      "step": 23470
+    },
+    {
+      "epoch": 5.491316931982634,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8662,
+      "step": 23475
+    },
+    {
+      "epoch": 5.492522913651713,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.9692,
+      "step": 23480
+    },
+    {
+      "epoch": 5.493728895320791,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.8551,
+      "step": 23485
+    },
+    {
+      "epoch": 5.49493487698987,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.791,
+      "step": 23490
+    },
+    {
+      "epoch": 5.496140858658948,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.847,
+      "step": 23495
+    },
+    {
+      "epoch": 5.497346840328027,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.956,
+      "step": 23500
+    },
+    {
+      "epoch": 5.498552821997106,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.7946,
+      "step": 23505
+    },
+    {
+      "epoch": 5.499758803666184,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8917,
+      "step": 23510
+    },
+    {
+      "epoch": 5.500964785335263,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.7733,
+      "step": 23515
+    },
+    {
+      "epoch": 5.502170767004341,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.8953,
+      "step": 23520
+    },
+    {
+      "epoch": 5.50337674867342,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 3.8253,
+      "step": 23525
+    },
+    {
+      "epoch": 5.504582730342499,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9044,
+      "step": 23530
+    },
+    {
+      "epoch": 5.505788712011578,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 4.0008,
+      "step": 23535
+    },
+    {
+      "epoch": 5.506994693680656,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9245,
+      "step": 23540
+    },
+    {
+      "epoch": 5.508200675349735,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0074,
+      "step": 23545
+    },
+    {
+      "epoch": 5.509406657018813,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8905,
+      "step": 23550
+    },
+    {
+      "epoch": 5.510612638687892,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0509,
+      "step": 23555
+    },
+    {
+      "epoch": 5.511818620356971,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9083,
+      "step": 23560
+    },
+    {
+      "epoch": 5.513024602026049,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9471,
+      "step": 23565
+    },
+    {
+      "epoch": 5.514230583695128,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9761,
+      "step": 23570
+    },
+    {
+      "epoch": 5.515436565364206,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8881,
+      "step": 23575
+    },
+    {
+      "epoch": 5.516642547033285,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.009,
+      "step": 23580
+    },
+    {
+      "epoch": 5.5178485287023635,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.9264,
+      "step": 23585
+    },
+    {
+      "epoch": 5.519054510371443,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8979,
+      "step": 23590
+    },
+    {
+      "epoch": 5.520260492040521,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.8037,
+      "step": 23595
+    },
+    {
+      "epoch": 5.5214664737096,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.8167,
+      "step": 23600
+    },
+    {
+      "epoch": 5.522672455378678,
+      "grad_norm": 3.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.8055,
+      "step": 23605
+    },
+    {
+      "epoch": 5.523878437047757,
+      "grad_norm": 2.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.9042,
+      "step": 23610
+    },
+    {
+      "epoch": 5.525084418716835,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.8375,
+      "step": 23615
+    },
+    {
+      "epoch": 5.526290400385914,
+      "grad_norm": 2.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9302,
+      "step": 23620
+    },
+    {
+      "epoch": 5.527496382054993,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.883,
+      "step": 23625
+    },
+    {
+      "epoch": 5.528702363724071,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.9157,
+      "step": 23630
+    },
+    {
+      "epoch": 5.52990834539315,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.826,
+      "step": 23635
+    },
+    {
+      "epoch": 5.5311143270622285,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9703,
+      "step": 23640
+    },
+    {
+      "epoch": 5.5323203087313075,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.7943,
+      "step": 23645
+    },
+    {
+      "epoch": 5.533526290400386,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.8437,
+      "step": 23650
+    },
+    {
+      "epoch": 5.534732272069465,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.8803,
+      "step": 23655
+    },
+    {
+      "epoch": 5.535938253738543,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.7778,
+      "step": 23660
+    },
+    {
+      "epoch": 5.537144235407622,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9392,
+      "step": 23665
+    },
+    {
+      "epoch": 5.5383502170767,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.7376,
+      "step": 23670
+    },
+    {
+      "epoch": 5.539556198745779,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9211,
+      "step": 23675
+    },
+    {
+      "epoch": 5.540762180414857,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8134,
+      "step": 23680
+    },
+    {
+      "epoch": 5.541968162083936,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9135,
+      "step": 23685
+    },
+    {
+      "epoch": 5.543174143753015,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9313,
+      "step": 23690
+    },
+    {
+      "epoch": 5.5443801254220935,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.7804,
+      "step": 23695
+    },
+    {
+      "epoch": 5.5455861070911725,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9341,
+      "step": 23700
+    },
+    {
+      "epoch": 5.546792088760251,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.9208,
+      "step": 23705
+    },
+    {
+      "epoch": 5.54799807042933,
+      "grad_norm": 3.9375,
+      "learning_rate": 3e-05,
+      "loss": 4.0624,
+      "step": 23710
+    },
+    {
+      "epoch": 5.549204052098408,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.7657,
+      "step": 23715
+    },
+    {
+      "epoch": 5.550410033767487,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.8927,
+      "step": 23720
+    },
+    {
+      "epoch": 5.551616015436565,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 4.0104,
+      "step": 23725
+    },
+    {
+      "epoch": 5.552821997105644,
+      "grad_norm": 3.046875,
+      "learning_rate": 3e-05,
+      "loss": 3.8402,
+      "step": 23730
+    },
+    {
+      "epoch": 5.554027978774723,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.0567,
+      "step": 23735
+    },
+    {
+      "epoch": 5.555233960443801,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9959,
+      "step": 23740
+    },
+    {
+      "epoch": 5.556439942112879,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.8047,
+      "step": 23745
+    },
+    {
+      "epoch": 5.557645923781958,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8724,
+      "step": 23750
+    },
+    {
+      "epoch": 5.5588519054510375,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9087,
+      "step": 23755
+    },
+    {
+      "epoch": 5.560057887120116,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 4.0107,
+      "step": 23760
+    },
+    {
+      "epoch": 5.561263868789195,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9476,
+      "step": 23765
+    },
+    {
+      "epoch": 5.562469850458273,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 3.7981,
+      "step": 23770
+    },
+    {
+      "epoch": 5.563675832127352,
+      "grad_norm": 3.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8564,
+      "step": 23775
+    },
+    {
+      "epoch": 5.56488181379643,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.7502,
+      "step": 23780
+    },
+    {
+      "epoch": 5.566087795465509,
+      "grad_norm": 2.125,
+      "learning_rate": 3e-05,
+      "loss": 3.832,
+      "step": 23785
+    },
+    {
+      "epoch": 5.567293777134587,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9683,
+      "step": 23790
+    },
+    {
+      "epoch": 5.568499758803666,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.1233,
+      "step": 23795
+    },
+    {
+      "epoch": 5.569705740472745,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.7684,
+      "step": 23800
+    },
+    {
+      "epoch": 5.570911722141823,
+      "grad_norm": 2.703125,
+      "learning_rate": 3e-05,
+      "loss": 3.8199,
+      "step": 23805
+    },
+    {
+      "epoch": 5.572117703810902,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0923,
+      "step": 23810
+    },
+    {
+      "epoch": 5.573323685479981,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9091,
+      "step": 23815
+    },
+    {
+      "epoch": 5.57452966714906,
+      "grad_norm": 2.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.7912,
+      "step": 23820
+    },
+    {
+      "epoch": 5.575735648818138,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9635,
+      "step": 23825
+    },
+    {
+      "epoch": 5.576941630487217,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.8467,
+      "step": 23830
+    },
+    {
+      "epoch": 5.578147612156295,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 4.0456,
+      "step": 23835
+    },
+    {
+      "epoch": 5.579353593825374,
+      "grad_norm": 2.25,
+      "learning_rate": 3e-05,
+      "loss": 3.8758,
+      "step": 23840
+    },
+    {
+      "epoch": 5.580559575494452,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9819,
+      "step": 23845
+    },
+    {
+      "epoch": 5.581765557163531,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.899,
+      "step": 23850
+    },
+    {
+      "epoch": 5.582971538832609,
+      "grad_norm": 3.25,
+      "learning_rate": 3e-05,
+      "loss": 3.9845,
+      "step": 23855
+    },
+    {
+      "epoch": 5.584177520501688,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.8407,
+      "step": 23860
+    },
+    {
+      "epoch": 5.585383502170767,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 3.9287,
+      "step": 23865
+    },
+    {
+      "epoch": 5.5865894838398455,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0203,
+      "step": 23870
+    },
+    {
+      "epoch": 5.587795465508925,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.7997,
+      "step": 23875
+    },
+    {
+      "epoch": 5.589001447178003,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 4.0247,
+      "step": 23880
+    },
+    {
+      "epoch": 5.590207428847082,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.9555,
+      "step": 23885
+    },
+    {
+      "epoch": 5.59141341051616,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.8785,
+      "step": 23890
+    },
+    {
+      "epoch": 5.592619392185239,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9272,
+      "step": 23895
+    },
+    {
+      "epoch": 5.593825373854317,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 3.9239,
+      "step": 23900
+    },
+    {
+      "epoch": 5.595031355523396,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.853,
+      "step": 23905
+    },
+    {
+      "epoch": 5.596237337192474,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9492,
+      "step": 23910
+    },
+    {
+      "epoch": 5.597443318861553,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.7937,
+      "step": 23915
+    },
+    {
+      "epoch": 5.5986493005306315,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8552,
+      "step": 23920
+    },
+    {
+      "epoch": 5.5998552821997105,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9371,
+      "step": 23925
+    },
+    {
+      "epoch": 5.6010612638687896,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9063,
+      "step": 23930
+    },
+    {
+      "epoch": 5.602267245537868,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9762,
+      "step": 23935
+    },
+    {
+      "epoch": 5.603473227206947,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.8711,
+      "step": 23940
+    },
+    {
+      "epoch": 5.604679208876025,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0296,
+      "step": 23945
+    },
+    {
+      "epoch": 5.605885190545104,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9397,
+      "step": 23950
+    },
+    {
+      "epoch": 5.607091172214182,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.914,
+      "step": 23955
+    },
+    {
+      "epoch": 5.608297153883261,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.8729,
+      "step": 23960
+    },
+    {
+      "epoch": 5.609503135552339,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.9831,
+      "step": 23965
+    },
+    {
+      "epoch": 5.610709117221418,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9341,
+      "step": 23970
+    },
+    {
+      "epoch": 5.611915098890497,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9959,
+      "step": 23975
+    },
+    {
+      "epoch": 5.6131210805595755,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 4.0269,
+      "step": 23980
+    },
+    {
+      "epoch": 5.614327062228654,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.077,
+      "step": 23985
+    },
+    {
+      "epoch": 5.615533043897733,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.9276,
+      "step": 23990
+    },
+    {
+      "epoch": 5.616739025566812,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.7895,
+      "step": 23995
+    },
+    {
+      "epoch": 5.61794500723589,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.8373,
+      "step": 24000
+    },
+    {
+      "epoch": 5.619150988904969,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9947,
+      "step": 24005
+    },
+    {
+      "epoch": 5.620356970574047,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.8984,
+      "step": 24010
+    },
+    {
+      "epoch": 5.621562952243126,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9236,
+      "step": 24015
+    },
+    {
+      "epoch": 5.622768933912204,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9875,
+      "step": 24020
+    },
+    {
+      "epoch": 5.623974915581283,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.7336,
+      "step": 24025
+    },
+    {
+      "epoch": 5.625180897250361,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 4.0664,
+      "step": 24030
+    },
+    {
+      "epoch": 5.6263868789194404,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.7686,
+      "step": 24035
+    },
+    {
+      "epoch": 5.6275928605885195,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.8031,
+      "step": 24040
+    },
+    {
+      "epoch": 5.628798842257598,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 4.1759,
+      "step": 24045
+    },
+    {
+      "epoch": 5.630004823926677,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9652,
+      "step": 24050
+    },
+    {
+      "epoch": 5.631210805595755,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9655,
+      "step": 24055
+    },
+    {
+      "epoch": 5.632416787264834,
+      "grad_norm": 2.953125,
+      "learning_rate": 3e-05,
+      "loss": 3.8859,
+      "step": 24060
+    },
+    {
+      "epoch": 5.633622768933912,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.8852,
+      "step": 24065
+    },
+    {
+      "epoch": 5.634828750602991,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9097,
+      "step": 24070
+    },
+    {
+      "epoch": 5.636034732272069,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.921,
+      "step": 24075
+    },
+    {
+      "epoch": 5.637240713941148,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8161,
+      "step": 24080
+    },
+    {
+      "epoch": 5.638446695610226,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.8408,
+      "step": 24085
+    },
+    {
+      "epoch": 5.639652677279305,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.9192,
+      "step": 24090
+    },
+    {
+      "epoch": 5.640858658948384,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8856,
+      "step": 24095
+    },
+    {
+      "epoch": 5.642064640617463,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 4.1615,
+      "step": 24100
+    },
+    {
+      "epoch": 5.643270622286542,
+      "grad_norm": 3.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9952,
+      "step": 24105
+    },
+    {
+      "epoch": 5.64447660395562,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.9389,
+      "step": 24110
+    },
+    {
+      "epoch": 5.645682585624699,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.8854,
+      "step": 24115
+    },
+    {
+      "epoch": 5.646888567293777,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.7662,
+      "step": 24120
+    },
+    {
+      "epoch": 5.648094548962856,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9437,
+      "step": 24125
+    },
+    {
+      "epoch": 5.649300530631934,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.7892,
+      "step": 24130
+    },
+    {
+      "epoch": 5.650506512301013,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.7409,
+      "step": 24135
+    },
+    {
+      "epoch": 5.651712493970091,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9944,
+      "step": 24140
+    },
+    {
+      "epoch": 5.65291847563917,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.9909,
+      "step": 24145
+    },
+    {
+      "epoch": 5.6541244573082485,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 3.8207,
+      "step": 24150
+    },
+    {
+      "epoch": 5.655330438977328,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 4.0181,
+      "step": 24155
+    },
+    {
+      "epoch": 5.656536420646406,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8528,
+      "step": 24160
+    },
+    {
+      "epoch": 5.657742402315485,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8931,
+      "step": 24165
+    },
+    {
+      "epoch": 5.658948383984564,
+      "grad_norm": 2.921875,
+      "learning_rate": 3e-05,
+      "loss": 4.0681,
+      "step": 24170
+    },
+    {
+      "epoch": 5.660154365653642,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0327,
+      "step": 24175
+    },
+    {
+      "epoch": 5.661360347322721,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8842,
+      "step": 24180
+    },
+    {
+      "epoch": 5.662566328991799,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 4.0374,
+      "step": 24185
+    },
+    {
+      "epoch": 5.663772310660878,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9792,
+      "step": 24190
+    },
+    {
+      "epoch": 5.664978292329956,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.9157,
+      "step": 24195
+    },
+    {
+      "epoch": 5.666184273999035,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8082,
+      "step": 24200
+    },
+    {
+      "epoch": 5.6673902556681135,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9661,
+      "step": 24205
+    },
+    {
+      "epoch": 5.6685962373371925,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 4.0323,
+      "step": 24210
+    },
+    {
+      "epoch": 5.669802219006272,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.0348,
+      "step": 24215
+    },
+    {
+      "epoch": 5.67100820067535,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.9035,
+      "step": 24220
+    },
+    {
+      "epoch": 5.672214182344428,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.9819,
+      "step": 24225
+    },
+    {
+      "epoch": 5.673420164013507,
+      "grad_norm": 2.296875,
+      "learning_rate": 3e-05,
+      "loss": 3.9195,
+      "step": 24230
+    },
+    {
+      "epoch": 5.674626145682586,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.8516,
+      "step": 24235
+    },
+    {
+      "epoch": 5.675832127351664,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.9088,
+      "step": 24240
+    },
+    {
+      "epoch": 5.677038109020743,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.8743,
+      "step": 24245
+    },
+    {
+      "epoch": 5.678244090689821,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.792,
+      "step": 24250
+    },
+    {
+      "epoch": 5.6794500723589,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9193,
+      "step": 24255
+    },
+    {
+      "epoch": 5.6806560540279785,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9117,
+      "step": 24260
+    },
+    {
+      "epoch": 5.6818620356970575,
+      "grad_norm": 2.8125,
+      "learning_rate": 3e-05,
+      "loss": 4.032,
+      "step": 24265
+    },
+    {
+      "epoch": 5.683068017366136,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.793,
+      "step": 24270
+    },
+    {
+      "epoch": 5.684273999035215,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.8266,
+      "step": 24275
+    },
+    {
+      "epoch": 5.685479980704294,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 4.0098,
+      "step": 24280
+    },
+    {
+      "epoch": 5.686685962373372,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9192,
+      "step": 24285
+    },
+    {
+      "epoch": 5.687891944042451,
+      "grad_norm": 3.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.8154,
+      "step": 24290
+    },
+    {
+      "epoch": 5.689097925711529,
+      "grad_norm": 2.75,
+      "learning_rate": 3e-05,
+      "loss": 3.8686,
+      "step": 24295
+    },
+    {
+      "epoch": 5.690303907380608,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.91,
+      "step": 24300
+    },
+    {
+      "epoch": 5.691509889049686,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8052,
+      "step": 24305
+    },
+    {
+      "epoch": 5.692715870718765,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.8589,
+      "step": 24310
+    },
+    {
+      "epoch": 5.693921852387843,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9194,
+      "step": 24315
+    },
+    {
+      "epoch": 5.6951278340569225,
+      "grad_norm": 2.203125,
+      "learning_rate": 3e-05,
+      "loss": 3.86,
+      "step": 24320
+    },
+    {
+      "epoch": 5.696333815726001,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8757,
+      "step": 24325
+    },
+    {
+      "epoch": 5.69753979739508,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.8218,
+      "step": 24330
+    },
+    {
+      "epoch": 5.698745779064158,
+      "grad_norm": 3.125,
+      "learning_rate": 3e-05,
+      "loss": 3.9438,
+      "step": 24335
+    },
+    {
+      "epoch": 5.699951760733237,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.8521,
+      "step": 24340
+    },
+    {
+      "epoch": 5.701157742402316,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 3.9065,
+      "step": 24345
+    },
+    {
+      "epoch": 5.702363724071394,
+      "grad_norm": 3.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.8656,
+      "step": 24350
+    },
+    {
+      "epoch": 5.703569705740473,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.879,
+      "step": 24355
+    },
+    {
+      "epoch": 5.704775687409551,
+      "grad_norm": 3.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9726,
+      "step": 24360
+    },
+    {
+      "epoch": 5.70598166907863,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.7641,
+      "step": 24365
+    },
+    {
+      "epoch": 5.707187650747708,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9876,
+      "step": 24370
+    },
+    {
+      "epoch": 5.708393632416787,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9063,
+      "step": 24375
+    },
+    {
+      "epoch": 5.709599614085866,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8726,
+      "step": 24380
+    },
+    {
+      "epoch": 5.710805595754945,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.9036,
+      "step": 24385
+    },
+    {
+      "epoch": 5.712011577424023,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.833,
+      "step": 24390
+    },
+    {
+      "epoch": 5.713217559093102,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.767,
+      "step": 24395
+    },
+    {
+      "epoch": 5.71442354076218,
+      "grad_norm": 2.421875,
+      "learning_rate": 3e-05,
+      "loss": 3.9863,
+      "step": 24400
+    },
+    {
+      "epoch": 5.715629522431259,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.8597,
+      "step": 24405
+    },
+    {
+      "epoch": 5.716835504100338,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9471,
+      "step": 24410
+    },
+    {
+      "epoch": 5.718041485769416,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8303,
+      "step": 24415
+    },
+    {
+      "epoch": 5.719247467438495,
+      "grad_norm": 3.0,
+      "learning_rate": 3e-05,
+      "loss": 4.0843,
+      "step": 24420
+    },
+    {
+      "epoch": 5.720453449107573,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.9113,
+      "step": 24425
+    },
+    {
+      "epoch": 5.721659430776652,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9092,
+      "step": 24430
+    },
+    {
+      "epoch": 5.7228654124457305,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.9785,
+      "step": 24435
+    },
+    {
+      "epoch": 5.72407139411481,
+      "grad_norm": 2.546875,
+      "learning_rate": 3e-05,
+      "loss": 3.8626,
+      "step": 24440
+    },
+    {
+      "epoch": 5.725277375783888,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.8585,
+      "step": 24445
+    },
+    {
+      "epoch": 5.726483357452967,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8717,
+      "step": 24450
+    },
+    {
+      "epoch": 5.727689339122046,
+      "grad_norm": 3.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.7441,
+      "step": 24455
+    },
+    {
+      "epoch": 5.728895320791124,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9387,
+      "step": 24460
+    },
+    {
+      "epoch": 5.730101302460202,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.7744,
+      "step": 24465
+    },
+    {
+      "epoch": 5.731307284129281,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9583,
+      "step": 24470
+    },
+    {
+      "epoch": 5.73251326579836,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.9269,
+      "step": 24475
+    },
+    {
+      "epoch": 5.733719247467438,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9111,
+      "step": 24480
+    },
+    {
+      "epoch": 5.734925229136517,
+      "grad_norm": 2.578125,
+      "learning_rate": 3e-05,
+      "loss": 3.8222,
+      "step": 24485
+    },
+    {
+      "epoch": 5.7361312108055955,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.8979,
+      "step": 24490
+    },
+    {
+      "epoch": 5.7373371924746746,
+      "grad_norm": 2.859375,
+      "learning_rate": 3e-05,
+      "loss": 3.9525,
+      "step": 24495
+    },
+    {
+      "epoch": 5.738543174143753,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.8084,
+      "step": 24500
+    },
+    {
+      "epoch": 5.739749155812832,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8157,
+      "step": 24505
+    },
+    {
+      "epoch": 5.74095513748191,
+      "grad_norm": 2.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8661,
+      "step": 24510
+    },
+    {
+      "epoch": 5.742161119150989,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.853,
+      "step": 24515
+    },
+    {
+      "epoch": 5.743367100820068,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 4.1209,
+      "step": 24520
+    },
+    {
+      "epoch": 5.744573082489146,
+      "grad_norm": 2.609375,
+      "learning_rate": 3e-05,
+      "loss": 3.896,
+      "step": 24525
+    },
+    {
+      "epoch": 5.745779064158225,
+      "grad_norm": 2.625,
+      "learning_rate": 3e-05,
+      "loss": 3.8586,
+      "step": 24530
+    },
+    {
+      "epoch": 5.746985045827303,
+      "grad_norm": 3.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8605,
+      "step": 24535
+    },
+    {
+      "epoch": 5.748191027496382,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.7692,
+      "step": 24540
+    },
+    {
+      "epoch": 5.7493970091654605,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.6939,
+      "step": 24545
+    },
+    {
+      "epoch": 5.7506029908345395,
+      "grad_norm": 3.234375,
+      "learning_rate": 3e-05,
+      "loss": 4.1683,
+      "step": 24550
+    },
+    {
+      "epoch": 5.751808972503618,
+      "grad_norm": 3.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0264,
+      "step": 24555
+    },
+    {
+      "epoch": 5.753014954172697,
+      "grad_norm": 2.5,
+      "learning_rate": 3e-05,
+      "loss": 3.9376,
+      "step": 24560
+    },
+    {
+      "epoch": 5.754220935841775,
+      "grad_norm": 3.140625,
+      "learning_rate": 3e-05,
+      "loss": 3.9573,
+      "step": 24565
+    },
+    {
+      "epoch": 5.755426917510854,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9609,
+      "step": 24570
+    },
+    {
+      "epoch": 5.756632899179932,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.8621,
+      "step": 24575
+    },
+    {
+      "epoch": 5.757838880849011,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.9926,
+      "step": 24580
+    },
+    {
+      "epoch": 5.75904486251809,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.979,
+      "step": 24585
+    },
+    {
+      "epoch": 5.760250844187168,
+      "grad_norm": 3.21875,
+      "learning_rate": 3e-05,
+      "loss": 3.9041,
+      "step": 24590
+    },
+    {
+      "epoch": 5.761456825856247,
+      "grad_norm": 2.984375,
+      "learning_rate": 3e-05,
+      "loss": 3.8563,
+      "step": 24595
+    },
+    {
+      "epoch": 5.7626628075253254,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9112,
+      "step": 24600
+    },
+    {
+      "epoch": 5.7638687891944045,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8316,
+      "step": 24605
+    },
+    {
+      "epoch": 5.765074770863483,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.8034,
+      "step": 24610
+    },
+    {
+      "epoch": 5.766280752532562,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.8455,
+      "step": 24615
+    },
+    {
+      "epoch": 5.76748673420164,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.8887,
+      "step": 24620
+    },
+    {
+      "epoch": 5.768692715870719,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.978,
+      "step": 24625
+    },
+    {
+      "epoch": 5.769898697539798,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.8693,
+      "step": 24630
+    },
+    {
+      "epoch": 5.771104679208876,
+      "grad_norm": 3.015625,
+      "learning_rate": 3e-05,
+      "loss": 3.93,
+      "step": 24635
+    },
+    {
+      "epoch": 5.772310660877954,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.7529,
+      "step": 24640
+    },
+    {
+      "epoch": 5.773516642547033,
+      "grad_norm": 2.3125,
+      "learning_rate": 3e-05,
+      "loss": 3.7131,
+      "step": 24645
+    },
+    {
+      "epoch": 5.774722624216112,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.797,
+      "step": 24650
+    },
+    {
+      "epoch": 5.77592860588519,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 4.1485,
+      "step": 24655
+    },
+    {
+      "epoch": 5.7771345875542695,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 4.1069,
+      "step": 24660
+    },
+    {
+      "epoch": 5.778340569223348,
+      "grad_norm": 2.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.9295,
+      "step": 24665
+    },
+    {
+      "epoch": 5.779546550892427,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.8998,
+      "step": 24670
+    },
+    {
+      "epoch": 5.780752532561505,
+      "grad_norm": 2.765625,
+      "learning_rate": 3e-05,
+      "loss": 3.9761,
+      "step": 24675
+    },
+    {
+      "epoch": 5.781958514230584,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.845,
+      "step": 24680
+    },
+    {
+      "epoch": 5.783164495899662,
+      "grad_norm": 2.5625,
+      "learning_rate": 3e-05,
+      "loss": 3.9596,
+      "step": 24685
+    },
+    {
+      "epoch": 5.784370477568741,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.8565,
+      "step": 24690
+    },
+    {
+      "epoch": 5.78557645923782,
+      "grad_norm": 3.09375,
+      "learning_rate": 3e-05,
+      "loss": 3.9124,
+      "step": 24695
+    },
+    {
+      "epoch": 5.786782440906898,
+      "grad_norm": 4.59375,
+      "learning_rate": 3e-05,
+      "loss": 4.0294,
+      "step": 24700
+    },
+    {
+      "epoch": 5.787988422575977,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 4.0272,
+      "step": 24705
+    },
+    {
+      "epoch": 5.789194404245055,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9979,
+      "step": 24710
+    },
+    {
+      "epoch": 5.790400385914134,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9102,
+      "step": 24715
+    },
+    {
+      "epoch": 5.791606367583213,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.8692,
+      "step": 24720
+    },
+    {
+      "epoch": 5.792812349252292,
+      "grad_norm": 2.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0655,
+      "step": 24725
+    },
+    {
+      "epoch": 5.79401833092137,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.94,
+      "step": 24730
+    },
+    {
+      "epoch": 5.795224312590449,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.9427,
+      "step": 24735
+    },
+    {
+      "epoch": 5.796430294259527,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.8959,
+      "step": 24740
+    },
+    {
+      "epoch": 5.797636275928606,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8888,
+      "step": 24745
+    },
+    {
+      "epoch": 5.798842257597684,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.9183,
+      "step": 24750
+    },
+    {
+      "epoch": 5.800048239266763,
+      "grad_norm": 3.03125,
+      "learning_rate": 3e-05,
+      "loss": 4.0635,
+      "step": 24755
+    },
+    {
+      "epoch": 5.801254220935842,
+      "grad_norm": 2.65625,
+      "learning_rate": 3e-05,
+      "loss": 3.8348,
+      "step": 24760
+    },
+    {
+      "epoch": 5.80246020260492,
+      "grad_norm": 2.59375,
+      "learning_rate": 3e-05,
+      "loss": 3.8865,
+      "step": 24765
+    },
+    {
+      "epoch": 5.803666184273999,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.9758,
+      "step": 24770
+    },
+    {
+      "epoch": 5.8048721659430775,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0289,
+      "step": 24775
+    },
+    {
+      "epoch": 5.806078147612157,
+      "grad_norm": 2.375,
+      "learning_rate": 3e-05,
+      "loss": 3.8561,
+      "step": 24780
+    },
+    {
+      "epoch": 5.807284129281235,
+      "grad_norm": 3.4375,
+      "learning_rate": 3e-05,
+      "loss": 4.0404,
+      "step": 24785
+    },
+    {
+      "epoch": 5.808490110950314,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9954,
+      "step": 24790
+    },
+    {
+      "epoch": 5.809696092619392,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.7289,
+      "step": 24795
+    },
+    {
+      "epoch": 5.810902074288471,
+      "grad_norm": 3.1875,
+      "learning_rate": 3e-05,
+      "loss": 3.9266,
+      "step": 24800
+    },
+    {
+      "epoch": 5.812108055957549,
+      "grad_norm": 2.84375,
+      "learning_rate": 3e-05,
+      "loss": 3.8585,
+      "step": 24805
+    },
+    {
+      "epoch": 5.813314037626628,
+      "grad_norm": 2.359375,
+      "learning_rate": 3e-05,
+      "loss": 3.9895,
+      "step": 24810
+    },
+    {
+      "epoch": 5.814520019295706,
+      "grad_norm": 2.328125,
+      "learning_rate": 3e-05,
+      "loss": 3.901,
+      "step": 24815
+    },
+    {
+      "epoch": 5.815726000964785,
+      "grad_norm": 3.109375,
+      "learning_rate": 3e-05,
+      "loss": 3.9541,
+      "step": 24820
+    },
+    {
+      "epoch": 5.816931982633864,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8988,
+      "step": 24825
+    },
+    {
+      "epoch": 5.8181379643029425,
+      "grad_norm": 2.96875,
+      "learning_rate": 3e-05,
+      "loss": 3.9047,
+      "step": 24830
+    },
+    {
+      "epoch": 5.8193439459720215,
+      "grad_norm": 2.46875,
+      "learning_rate": 3e-05,
+      "loss": 3.9864,
+      "step": 24835
+    },
+    {
+      "epoch": 5.8205499276411,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 3.7528,
+      "step": 24840
+    },
+    {
+      "epoch": 5.821755909310179,
+      "grad_norm": 2.390625,
+      "learning_rate": 3e-05,
+      "loss": 3.9339,
+      "step": 24845
+    },
+    {
+      "epoch": 5.822961890979257,
+      "grad_norm": 3.265625,
+      "learning_rate": 3e-05,
+      "loss": 3.8203,
+      "step": 24850
+    },
+    {
+      "epoch": 5.824167872648336,
+      "grad_norm": 2.90625,
+      "learning_rate": 3e-05,
+      "loss": 3.8625,
+      "step": 24855
+    },
+    {
+      "epoch": 5.825373854317414,
+      "grad_norm": 2.796875,
+      "learning_rate": 3e-05,
+      "loss": 3.9287,
+      "step": 24860
+    },
+    {
+      "epoch": 5.826579835986493,
+      "grad_norm": 2.34375,
+      "learning_rate": 3e-05,
+      "loss": 3.9621,
+      "step": 24865
+    },
+    {
+      "epoch": 5.827785817655572,
+      "grad_norm": 3.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9187,
+      "step": 24870
+    },
+    {
+      "epoch": 5.82899179932465,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.8991,
+      "step": 24875
+    },
+    {
+      "epoch": 5.830197780993728,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.9435,
+      "step": 24880
+    },
+    {
+      "epoch": 5.8314037626628075,
+      "grad_norm": 2.640625,
+      "learning_rate": 3e-05,
+      "loss": 3.9192,
+      "step": 24885
+    },
+    {
+      "epoch": 5.8326097443318865,
+      "grad_norm": 2.78125,
+      "learning_rate": 3e-05,
+      "loss": 3.9059,
+      "step": 24890
+    },
+    {
+      "epoch": 5.833815726000965,
+      "grad_norm": 2.671875,
+      "learning_rate": 3e-05,
+      "loss": 3.8766,
+      "step": 24895
+    },
+    {
+      "epoch": 5.835021707670044,
+      "grad_norm": 3.0625,
+      "learning_rate": 3e-05,
+      "loss": 3.8632,
+      "step": 24900
+    },
+    {
+      "epoch": 5.836227689339122,
+      "grad_norm": 3.078125,
+      "learning_rate": 3e-05,
+      "loss": 3.8248,
+      "step": 24905
+    },
+    {
+      "epoch": 5.837433671008201,
+      "grad_norm": 2.828125,
+      "learning_rate": 3e-05,
+      "loss": 4.0409,
+      "step": 24910
+    },
+    {
+      "epoch": 5.838639652677279,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8307,
+      "step": 24915
+    },
+    {
+      "epoch": 5.839845634346358,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 4.0343,
+      "step": 24920
+    },
+    {
+      "epoch": 5.841051616015436,
+      "grad_norm": 2.53125,
+      "learning_rate": 3e-05,
+      "loss": 3.9369,
+      "step": 24925
+    },
+    {
+      "epoch": 5.842257597684515,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.7994,
+      "step": 24930
+    },
+    {
+      "epoch": 5.843463579353594,
+      "grad_norm": 3.15625,
+      "learning_rate": 3e-05,
+      "loss": 3.8435,
+      "step": 24935
+    },
+    {
+      "epoch": 5.844669561022672,
+      "grad_norm": 2.71875,
+      "learning_rate": 3e-05,
+      "loss": 3.7457,
+      "step": 24940
+    },
+    {
+      "epoch": 5.8458755426917515,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8773,
+      "step": 24945
+    },
+    {
+      "epoch": 5.84708152436083,
+      "grad_norm": 2.28125,
+      "learning_rate": 3e-05,
+      "loss": 3.9961,
+      "step": 24950
+    },
+    {
+      "epoch": 5.848287506029909,
+      "grad_norm": 2.9375,
+      "learning_rate": 3e-05,
+      "loss": 3.9634,
+      "step": 24955
+    },
+    {
+      "epoch": 5.849493487698987,
+      "grad_norm": 2.484375,
+      "learning_rate": 3e-05,
+      "loss": 3.8916,
+      "step": 24960
+    },
+    {
+      "epoch": 5.850699469368066,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 4.1893,
+      "step": 24965
+    },
+    {
+      "epoch": 5.851905451037144,
+      "grad_norm": 2.40625,
+      "learning_rate": 3e-05,
+      "loss": 3.9133,
+      "step": 24970
+    },
+    {
+      "epoch": 5.853111432706223,
+      "grad_norm": 2.453125,
+      "learning_rate": 3e-05,
+      "loss": 3.8251,
+      "step": 24975
+    },
+    {
+      "epoch": 5.854317414375301,
+      "grad_norm": 2.734375,
+      "learning_rate": 3e-05,
+      "loss": 3.9746,
+      "step": 24980
+    },
+    {
+      "epoch": 5.85552339604438,
+      "grad_norm": 2.515625,
+      "learning_rate": 3e-05,
+      "loss": 3.6877,
+      "step": 24985
+    },
+    {
+      "epoch": 5.856729377713458,
+      "grad_norm": 2.875,
+      "learning_rate": 3e-05,
+      "loss": 3.9787,
+      "step": 24990
+    },
+    {
+      "epoch": 5.857935359382537,
+      "grad_norm": 2.6875,
+      "learning_rate": 3e-05,
+      "loss": 3.7577,
+      "step": 24995
+    },
+    {
+      "epoch": 5.859141341051616,
+      "grad_norm": 2.890625,
+      "learning_rate": 3e-05,
+      "loss": 3.9542,
+      "step": 25000
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 41460,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "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": 1.7338551769487114e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}