diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,9089 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 9.992193598750976,
+  "eval_steps": 500,
+  "global_step": 6400,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00156128024980484,
+      "grad_norm": 15.554603576660156,
+      "learning_rate": 3.125e-07,
+      "loss": 1.2028,
+      "step": 1
+    },
+    {
+      "epoch": 0.0078064012490242,
+      "grad_norm": 14.548426628112793,
+      "learning_rate": 1.5625e-06,
+      "loss": 1.23,
+      "step": 5
+    },
+    {
+      "epoch": 0.0156128024980484,
+      "grad_norm": 10.735854148864746,
+      "learning_rate": 3.125e-06,
+      "loss": 1.1711,
+      "step": 10
+    },
+    {
+      "epoch": 0.0234192037470726,
+      "grad_norm": 9.330694198608398,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 1.137,
+      "step": 15
+    },
+    {
+      "epoch": 0.0312256049960968,
+      "grad_norm": 5.181936740875244,
+      "learning_rate": 6.25e-06,
+      "loss": 1.0235,
+      "step": 20
+    },
+    {
+      "epoch": 0.039032006245121,
+      "grad_norm": 3.862912654876709,
+      "learning_rate": 7.8125e-06,
+      "loss": 0.9359,
+      "step": 25
+    },
+    {
+      "epoch": 0.0468384074941452,
+      "grad_norm": 3.142946720123291,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 0.8746,
+      "step": 30
+    },
+    {
+      "epoch": 0.0546448087431694,
+      "grad_norm": 1.8955553770065308,
+      "learning_rate": 1.09375e-05,
+      "loss": 0.8135,
+      "step": 35
+    },
+    {
+      "epoch": 0.0624512099921936,
+      "grad_norm": 1.3491796255111694,
+      "learning_rate": 1.25e-05,
+      "loss": 0.7632,
+      "step": 40
+    },
+    {
+      "epoch": 0.0702576112412178,
+      "grad_norm": 1.4237151145935059,
+      "learning_rate": 1.4062500000000001e-05,
+      "loss": 0.7152,
+      "step": 45
+    },
+    {
+      "epoch": 0.078064012490242,
+      "grad_norm": 1.992169737815857,
+      "learning_rate": 1.5625e-05,
+      "loss": 0.6802,
+      "step": 50
+    },
+    {
+      "epoch": 0.0858704137392662,
+      "grad_norm": 1.0017743110656738,
+      "learning_rate": 1.71875e-05,
+      "loss": 0.6617,
+      "step": 55
+    },
+    {
+      "epoch": 0.0936768149882904,
+      "grad_norm": 0.9323798418045044,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 0.6236,
+      "step": 60
+    },
+    {
+      "epoch": 0.1014832162373146,
+      "grad_norm": 0.7461961507797241,
+      "learning_rate": 2.0312500000000002e-05,
+      "loss": 0.6,
+      "step": 65
+    },
+    {
+      "epoch": 0.1092896174863388,
+      "grad_norm": 0.7259904742240906,
+      "learning_rate": 2.1875e-05,
+      "loss": 0.5814,
+      "step": 70
+    },
+    {
+      "epoch": 0.117096018735363,
+      "grad_norm": 0.5570379495620728,
+      "learning_rate": 2.34375e-05,
+      "loss": 0.5752,
+      "step": 75
+    },
+    {
+      "epoch": 0.1249024199843872,
+      "grad_norm": 0.5848241448402405,
+      "learning_rate": 2.5e-05,
+      "loss": 0.5455,
+      "step": 80
+    },
+    {
+      "epoch": 0.1327088212334114,
+      "grad_norm": 0.9087640643119812,
+      "learning_rate": 2.6562500000000002e-05,
+      "loss": 0.549,
+      "step": 85
+    },
+    {
+      "epoch": 0.1405152224824356,
+      "grad_norm": 0.6453709602355957,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 0.5382,
+      "step": 90
+    },
+    {
+      "epoch": 0.1483216237314598,
+      "grad_norm": 0.6244491934776306,
+      "learning_rate": 2.96875e-05,
+      "loss": 0.5416,
+      "step": 95
+    },
+    {
+      "epoch": 0.156128024980484,
+      "grad_norm": 0.6923064589500427,
+      "learning_rate": 3.125e-05,
+      "loss": 0.5209,
+      "step": 100
+    },
+    {
+      "epoch": 0.16393442622950818,
+      "grad_norm": 0.8899492025375366,
+      "learning_rate": 3.2812500000000005e-05,
+      "loss": 0.5193,
+      "step": 105
+    },
+    {
+      "epoch": 0.1717408274785324,
+      "grad_norm": 0.6282894611358643,
+      "learning_rate": 3.4375e-05,
+      "loss": 0.5283,
+      "step": 110
+    },
+    {
+      "epoch": 0.1795472287275566,
+      "grad_norm": 0.637545645236969,
+      "learning_rate": 3.59375e-05,
+      "loss": 0.5161,
+      "step": 115
+    },
+    {
+      "epoch": 0.1873536299765808,
+      "grad_norm": 0.6604511141777039,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.5024,
+      "step": 120
+    },
+    {
+      "epoch": 0.195160031225605,
+      "grad_norm": 0.5799457430839539,
+      "learning_rate": 3.90625e-05,
+      "loss": 0.5113,
+      "step": 125
+    },
+    {
+      "epoch": 0.2029664324746292,
+      "grad_norm": 0.745388388633728,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 0.509,
+      "step": 130
+    },
+    {
+      "epoch": 0.2107728337236534,
+      "grad_norm": 0.8749244809150696,
+      "learning_rate": 4.21875e-05,
+      "loss": 0.5085,
+      "step": 135
+    },
+    {
+      "epoch": 0.2185792349726776,
+      "grad_norm": 0.832904040813446,
+      "learning_rate": 4.375e-05,
+      "loss": 0.4885,
+      "step": 140
+    },
+    {
+      "epoch": 0.2263856362217018,
+      "grad_norm": 0.7716369032859802,
+      "learning_rate": 4.5312500000000004e-05,
+      "loss": 0.4886,
+      "step": 145
+    },
+    {
+      "epoch": 0.234192037470726,
+      "grad_norm": 0.6549221277236938,
+      "learning_rate": 4.6875e-05,
+      "loss": 0.4879,
+      "step": 150
+    },
+    {
+      "epoch": 0.2419984387197502,
+      "grad_norm": 0.5921856164932251,
+      "learning_rate": 4.8437500000000005e-05,
+      "loss": 0.4902,
+      "step": 155
+    },
+    {
+      "epoch": 0.2498048399687744,
+      "grad_norm": 0.5582632422447205,
+      "learning_rate": 5e-05,
+      "loss": 0.4954,
+      "step": 160
+    },
+    {
+      "epoch": 0.2576112412177986,
+      "grad_norm": 0.5605296492576599,
+      "learning_rate": 5.15625e-05,
+      "loss": 0.4882,
+      "step": 165
+    },
+    {
+      "epoch": 0.2654176424668228,
+      "grad_norm": 0.7804284691810608,
+      "learning_rate": 5.3125000000000004e-05,
+      "loss": 0.4751,
+      "step": 170
+    },
+    {
+      "epoch": 0.273224043715847,
+      "grad_norm": 0.9444372653961182,
+      "learning_rate": 5.46875e-05,
+      "loss": 0.4777,
+      "step": 175
+    },
+    {
+      "epoch": 0.2810304449648712,
+      "grad_norm": 0.5533527135848999,
+      "learning_rate": 5.6250000000000005e-05,
+      "loss": 0.4889,
+      "step": 180
+    },
+    {
+      "epoch": 0.2888368462138954,
+      "grad_norm": 0.5322621464729309,
+      "learning_rate": 5.78125e-05,
+      "loss": 0.4837,
+      "step": 185
+    },
+    {
+      "epoch": 0.2966432474629196,
+      "grad_norm": 0.5449932217597961,
+      "learning_rate": 5.9375e-05,
+      "loss": 0.4695,
+      "step": 190
+    },
+    {
+      "epoch": 0.3044496487119438,
+      "grad_norm": 0.6229299306869507,
+      "learning_rate": 6.0937500000000004e-05,
+      "loss": 0.4891,
+      "step": 195
+    },
+    {
+      "epoch": 0.312256049960968,
+      "grad_norm": 0.5832580327987671,
+      "learning_rate": 6.25e-05,
+      "loss": 0.4697,
+      "step": 200
+    },
+    {
+      "epoch": 0.3200624512099922,
+      "grad_norm": 0.6701719164848328,
+      "learning_rate": 6.40625e-05,
+      "loss": 0.466,
+      "step": 205
+    },
+    {
+      "epoch": 0.32786885245901637,
+      "grad_norm": 1.088781476020813,
+      "learning_rate": 6.562500000000001e-05,
+      "loss": 0.4684,
+      "step": 210
+    },
+    {
+      "epoch": 0.3356752537080406,
+      "grad_norm": 0.5555649399757385,
+      "learning_rate": 6.71875e-05,
+      "loss": 0.471,
+      "step": 215
+    },
+    {
+      "epoch": 0.3434816549570648,
+      "grad_norm": 0.7747379541397095,
+      "learning_rate": 6.875e-05,
+      "loss": 0.4749,
+      "step": 220
+    },
+    {
+      "epoch": 0.351288056206089,
+      "grad_norm": 0.5692268013954163,
+      "learning_rate": 7.031250000000001e-05,
+      "loss": 0.4716,
+      "step": 225
+    },
+    {
+      "epoch": 0.3590944574551132,
+      "grad_norm": 0.6532680988311768,
+      "learning_rate": 7.1875e-05,
+      "loss": 0.4658,
+      "step": 230
+    },
+    {
+      "epoch": 0.3669008587041374,
+      "grad_norm": 0.5181140303611755,
+      "learning_rate": 7.34375e-05,
+      "loss": 0.4631,
+      "step": 235
+    },
+    {
+      "epoch": 0.3747072599531616,
+      "grad_norm": 0.5234228372573853,
+      "learning_rate": 7.500000000000001e-05,
+      "loss": 0.4632,
+      "step": 240
+    },
+    {
+      "epoch": 0.3825136612021858,
+      "grad_norm": 0.7775989174842834,
+      "learning_rate": 7.65625e-05,
+      "loss": 0.4767,
+      "step": 245
+    },
+    {
+      "epoch": 0.39032006245121,
+      "grad_norm": 0.736663281917572,
+      "learning_rate": 7.8125e-05,
+      "loss": 0.4552,
+      "step": 250
+    },
+    {
+      "epoch": 0.3981264637002342,
+      "grad_norm": 0.49683699011802673,
+      "learning_rate": 7.96875e-05,
+      "loss": 0.458,
+      "step": 255
+    },
+    {
+      "epoch": 0.4059328649492584,
+      "grad_norm": 0.5580967664718628,
+      "learning_rate": 8.125000000000001e-05,
+      "loss": 0.4553,
+      "step": 260
+    },
+    {
+      "epoch": 0.4137392661982826,
+      "grad_norm": 0.5070011019706726,
+      "learning_rate": 8.28125e-05,
+      "loss": 0.4667,
+      "step": 265
+    },
+    {
+      "epoch": 0.4215456674473068,
+      "grad_norm": 0.5351033806800842,
+      "learning_rate": 8.4375e-05,
+      "loss": 0.4633,
+      "step": 270
+    },
+    {
+      "epoch": 0.42935206869633097,
+      "grad_norm": 0.5223821997642517,
+      "learning_rate": 8.593750000000001e-05,
+      "loss": 0.4614,
+      "step": 275
+    },
+    {
+      "epoch": 0.4371584699453552,
+      "grad_norm": 0.5035797953605652,
+      "learning_rate": 8.75e-05,
+      "loss": 0.4577,
+      "step": 280
+    },
+    {
+      "epoch": 0.4449648711943794,
+      "grad_norm": 0.5388506054878235,
+      "learning_rate": 8.90625e-05,
+      "loss": 0.4647,
+      "step": 285
+    },
+    {
+      "epoch": 0.4527712724434036,
+      "grad_norm": 0.5311102867126465,
+      "learning_rate": 9.062500000000001e-05,
+      "loss": 0.4537,
+      "step": 290
+    },
+    {
+      "epoch": 0.4605776736924278,
+      "grad_norm": 0.5614621043205261,
+      "learning_rate": 9.21875e-05,
+      "loss": 0.4635,
+      "step": 295
+    },
+    {
+      "epoch": 0.468384074941452,
+      "grad_norm": 0.7148458957672119,
+      "learning_rate": 9.375e-05,
+      "loss": 0.457,
+      "step": 300
+    },
+    {
+      "epoch": 0.47619047619047616,
+      "grad_norm": 0.7659621834754944,
+      "learning_rate": 9.53125e-05,
+      "loss": 0.4522,
+      "step": 305
+    },
+    {
+      "epoch": 0.4839968774395004,
+      "grad_norm": 0.6256824731826782,
+      "learning_rate": 9.687500000000001e-05,
+      "loss": 0.4547,
+      "step": 310
+    },
+    {
+      "epoch": 0.4918032786885246,
+      "grad_norm": 0.5664246082305908,
+      "learning_rate": 9.84375e-05,
+      "loss": 0.456,
+      "step": 315
+    },
+    {
+      "epoch": 0.4996096799375488,
+      "grad_norm": 0.6836294531822205,
+      "learning_rate": 0.0001,
+      "loss": 0.4525,
+      "step": 320
+    },
+    {
+      "epoch": 0.507416081186573,
+      "grad_norm": 0.5858592391014099,
+      "learning_rate": 0.00010156250000000001,
+      "loss": 0.4439,
+      "step": 325
+    },
+    {
+      "epoch": 0.5152224824355972,
+      "grad_norm": 0.5709215402603149,
+      "learning_rate": 0.000103125,
+      "loss": 0.4591,
+      "step": 330
+    },
+    {
+      "epoch": 0.5230288836846214,
+      "grad_norm": 0.5698142647743225,
+      "learning_rate": 0.0001046875,
+      "loss": 0.4412,
+      "step": 335
+    },
+    {
+      "epoch": 0.5308352849336456,
+      "grad_norm": 0.5309858918190002,
+      "learning_rate": 0.00010625000000000001,
+      "loss": 0.4473,
+      "step": 340
+    },
+    {
+      "epoch": 0.5386416861826698,
+      "grad_norm": 0.49240779876708984,
+      "learning_rate": 0.00010781250000000001,
+      "loss": 0.4415,
+      "step": 345
+    },
+    {
+      "epoch": 0.546448087431694,
+      "grad_norm": 0.4742411971092224,
+      "learning_rate": 0.000109375,
+      "loss": 0.441,
+      "step": 350
+    },
+    {
+      "epoch": 0.5542544886807181,
+      "grad_norm": 0.4116191864013672,
+      "learning_rate": 0.0001109375,
+      "loss": 0.4389,
+      "step": 355
+    },
+    {
+      "epoch": 0.5620608899297423,
+      "grad_norm": 0.5430789589881897,
+      "learning_rate": 0.00011250000000000001,
+      "loss": 0.4636,
+      "step": 360
+    },
+    {
+      "epoch": 0.5698672911787666,
+      "grad_norm": 0.6789748668670654,
+      "learning_rate": 0.0001140625,
+      "loss": 0.4391,
+      "step": 365
+    },
+    {
+      "epoch": 0.5776736924277908,
+      "grad_norm": 0.4417556822299957,
+      "learning_rate": 0.000115625,
+      "loss": 0.4383,
+      "step": 370
+    },
+    {
+      "epoch": 0.585480093676815,
+      "grad_norm": 0.4192645847797394,
+      "learning_rate": 0.00011718750000000001,
+      "loss": 0.4392,
+      "step": 375
+    },
+    {
+      "epoch": 0.5932864949258392,
+      "grad_norm": 0.42620405554771423,
+      "learning_rate": 0.00011875,
+      "loss": 0.4509,
+      "step": 380
+    },
+    {
+      "epoch": 0.6010928961748634,
+      "grad_norm": 0.44085854291915894,
+      "learning_rate": 0.0001203125,
+      "loss": 0.4458,
+      "step": 385
+    },
+    {
+      "epoch": 0.6088992974238876,
+      "grad_norm": 0.45056504011154175,
+      "learning_rate": 0.00012187500000000001,
+      "loss": 0.4461,
+      "step": 390
+    },
+    {
+      "epoch": 0.6167056986729118,
+      "grad_norm": 0.5633181929588318,
+      "learning_rate": 0.0001234375,
+      "loss": 0.4358,
+      "step": 395
+    },
+    {
+      "epoch": 0.624512099921936,
+      "grad_norm": 0.5012897253036499,
+      "learning_rate": 0.000125,
+      "loss": 0.4498,
+      "step": 400
+    },
+    {
+      "epoch": 0.6323185011709602,
+      "grad_norm": 0.4812583923339844,
+      "learning_rate": 0.0001265625,
+      "loss": 0.4239,
+      "step": 405
+    },
+    {
+      "epoch": 0.6401249024199844,
+      "grad_norm": 0.42709508538246155,
+      "learning_rate": 0.000128125,
+      "loss": 0.452,
+      "step": 410
+    },
+    {
+      "epoch": 0.6479313036690086,
+      "grad_norm": 0.5168142914772034,
+      "learning_rate": 0.0001296875,
+      "loss": 0.4365,
+      "step": 415
+    },
+    {
+      "epoch": 0.6557377049180327,
+      "grad_norm": 0.4785514771938324,
+      "learning_rate": 0.00013125000000000002,
+      "loss": 0.4356,
+      "step": 420
+    },
+    {
+      "epoch": 0.663544106167057,
+      "grad_norm": 0.44413048028945923,
+      "learning_rate": 0.0001328125,
+      "loss": 0.4398,
+      "step": 425
+    },
+    {
+      "epoch": 0.6713505074160812,
+      "grad_norm": 0.4412420094013214,
+      "learning_rate": 0.000134375,
+      "loss": 0.4363,
+      "step": 430
+    },
+    {
+      "epoch": 0.6791569086651054,
+      "grad_norm": 0.4392535090446472,
+      "learning_rate": 0.00013593750000000002,
+      "loss": 0.4339,
+      "step": 435
+    },
+    {
+      "epoch": 0.6869633099141296,
+      "grad_norm": 0.42931196093559265,
+      "learning_rate": 0.0001375,
+      "loss": 0.4279,
+      "step": 440
+    },
+    {
+      "epoch": 0.6947697111631538,
+      "grad_norm": 0.41424766182899475,
+      "learning_rate": 0.0001390625,
+      "loss": 0.4313,
+      "step": 445
+    },
+    {
+      "epoch": 0.702576112412178,
+      "grad_norm": 0.4159262776374817,
+      "learning_rate": 0.00014062500000000002,
+      "loss": 0.4273,
+      "step": 450
+    },
+    {
+      "epoch": 0.7103825136612022,
+      "grad_norm": 0.397058367729187,
+      "learning_rate": 0.0001421875,
+      "loss": 0.4416,
+      "step": 455
+    },
+    {
+      "epoch": 0.7181889149102264,
+      "grad_norm": 0.42139485478401184,
+      "learning_rate": 0.00014375,
+      "loss": 0.4348,
+      "step": 460
+    },
+    {
+      "epoch": 0.7259953161592506,
+      "grad_norm": 0.47808170318603516,
+      "learning_rate": 0.00014531250000000002,
+      "loss": 0.4244,
+      "step": 465
+    },
+    {
+      "epoch": 0.7338017174082748,
+      "grad_norm": 0.4168015122413635,
+      "learning_rate": 0.000146875,
+      "loss": 0.4351,
+      "step": 470
+    },
+    {
+      "epoch": 0.741608118657299,
+      "grad_norm": 0.3950536251068115,
+      "learning_rate": 0.0001484375,
+      "loss": 0.4338,
+      "step": 475
+    },
+    {
+      "epoch": 0.7494145199063232,
+      "grad_norm": 0.5723301768302917,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.4348,
+      "step": 480
+    },
+    {
+      "epoch": 0.7572209211553473,
+      "grad_norm": 0.3689775764942169,
+      "learning_rate": 0.0001515625,
+      "loss": 0.4355,
+      "step": 485
+    },
+    {
+      "epoch": 0.7650273224043715,
+      "grad_norm": 0.4008382260799408,
+      "learning_rate": 0.000153125,
+      "loss": 0.422,
+      "step": 490
+    },
+    {
+      "epoch": 0.7728337236533958,
+      "grad_norm": 0.49685850739479065,
+      "learning_rate": 0.0001546875,
+      "loss": 0.4258,
+      "step": 495
+    },
+    {
+      "epoch": 0.78064012490242,
+      "grad_norm": 0.39265674352645874,
+      "learning_rate": 0.00015625,
+      "loss": 0.4334,
+      "step": 500
+    },
+    {
+      "epoch": 0.7884465261514442,
+      "grad_norm": 0.6721764206886292,
+      "learning_rate": 0.00015781250000000002,
+      "loss": 0.4283,
+      "step": 505
+    },
+    {
+      "epoch": 0.7962529274004684,
+      "grad_norm": 0.43760669231414795,
+      "learning_rate": 0.000159375,
+      "loss": 0.4315,
+      "step": 510
+    },
+    {
+      "epoch": 0.8040593286494926,
+      "grad_norm": 0.3985048532485962,
+      "learning_rate": 0.0001609375,
+      "loss": 0.4318,
+      "step": 515
+    },
+    {
+      "epoch": 0.8118657298985168,
+      "grad_norm": 0.3855305016040802,
+      "learning_rate": 0.00016250000000000002,
+      "loss": 0.4236,
+      "step": 520
+    },
+    {
+      "epoch": 0.819672131147541,
+      "grad_norm": 0.4576222002506256,
+      "learning_rate": 0.0001640625,
+      "loss": 0.4246,
+      "step": 525
+    },
+    {
+      "epoch": 0.8274785323965652,
+      "grad_norm": 0.3545086979866028,
+      "learning_rate": 0.000165625,
+      "loss": 0.4248,
+      "step": 530
+    },
+    {
+      "epoch": 0.8352849336455894,
+      "grad_norm": 0.360779345035553,
+      "learning_rate": 0.00016718750000000002,
+      "loss": 0.4225,
+      "step": 535
+    },
+    {
+      "epoch": 0.8430913348946136,
+      "grad_norm": 0.344586044549942,
+      "learning_rate": 0.00016875,
+      "loss": 0.4231,
+      "step": 540
+    },
+    {
+      "epoch": 0.8508977361436377,
+      "grad_norm": 0.4501068890094757,
+      "learning_rate": 0.0001703125,
+      "loss": 0.423,
+      "step": 545
+    },
+    {
+      "epoch": 0.8587041373926619,
+      "grad_norm": 0.5521011352539062,
+      "learning_rate": 0.00017187500000000002,
+      "loss": 0.424,
+      "step": 550
+    },
+    {
+      "epoch": 0.8665105386416861,
+      "grad_norm": 0.4081909954547882,
+      "learning_rate": 0.0001734375,
+      "loss": 0.4126,
+      "step": 555
+    },
+    {
+      "epoch": 0.8743169398907104,
+      "grad_norm": 0.36447080969810486,
+      "learning_rate": 0.000175,
+      "loss": 0.4197,
+      "step": 560
+    },
+    {
+      "epoch": 0.8821233411397346,
+      "grad_norm": 0.43303537368774414,
+      "learning_rate": 0.00017656250000000002,
+      "loss": 0.4261,
+      "step": 565
+    },
+    {
+      "epoch": 0.8899297423887588,
+      "grad_norm": 0.35191860795021057,
+      "learning_rate": 0.000178125,
+      "loss": 0.4233,
+      "step": 570
+    },
+    {
+      "epoch": 0.897736143637783,
+      "grad_norm": 0.33722788095474243,
+      "learning_rate": 0.0001796875,
+      "loss": 0.4149,
+      "step": 575
+    },
+    {
+      "epoch": 0.9055425448868072,
+      "grad_norm": 0.36060288548469543,
+      "learning_rate": 0.00018125000000000001,
+      "loss": 0.4244,
+      "step": 580
+    },
+    {
+      "epoch": 0.9133489461358314,
+      "grad_norm": 0.353076696395874,
+      "learning_rate": 0.0001828125,
+      "loss": 0.4202,
+      "step": 585
+    },
+    {
+      "epoch": 0.9211553473848556,
+      "grad_norm": 0.39187219738960266,
+      "learning_rate": 0.000184375,
+      "loss": 0.4133,
+      "step": 590
+    },
+    {
+      "epoch": 0.9289617486338798,
+      "grad_norm": 0.5925287008285522,
+      "learning_rate": 0.0001859375,
+      "loss": 0.4231,
+      "step": 595
+    },
+    {
+      "epoch": 0.936768149882904,
+      "grad_norm": 0.5176442861557007,
+      "learning_rate": 0.0001875,
+      "loss": 0.4245,
+      "step": 600
+    },
+    {
+      "epoch": 0.9445745511319282,
+      "grad_norm": 0.33046722412109375,
+      "learning_rate": 0.00018906250000000002,
+      "loss": 0.4147,
+      "step": 605
+    },
+    {
+      "epoch": 0.9523809523809523,
+      "grad_norm": 0.3665611147880554,
+      "learning_rate": 0.000190625,
+      "loss": 0.4275,
+      "step": 610
+    },
+    {
+      "epoch": 0.9601873536299765,
+      "grad_norm": 0.3403601050376892,
+      "learning_rate": 0.0001921875,
+      "loss": 0.4158,
+      "step": 615
+    },
+    {
+      "epoch": 0.9679937548790007,
+      "grad_norm": 0.3726797103881836,
+      "learning_rate": 0.00019375000000000002,
+      "loss": 0.4197,
+      "step": 620
+    },
+    {
+      "epoch": 0.975800156128025,
+      "grad_norm": 0.38149702548980713,
+      "learning_rate": 0.0001953125,
+      "loss": 0.4102,
+      "step": 625
+    },
+    {
+      "epoch": 0.9836065573770492,
+      "grad_norm": 0.3473910987377167,
+      "learning_rate": 0.000196875,
+      "loss": 0.4294,
+      "step": 630
+    },
+    {
+      "epoch": 0.9914129586260734,
+      "grad_norm": 0.32832297682762146,
+      "learning_rate": 0.00019843750000000002,
+      "loss": 0.4162,
+      "step": 635
+    },
+    {
+      "epoch": 0.9992193598750976,
+      "grad_norm": 0.33780425786972046,
+      "learning_rate": 0.0002,
+      "loss": 0.419,
+      "step": 640
+    },
+    {
+      "epoch": 0.9992193598750976,
+      "eval_loss": 1.0502797365188599,
+      "eval_runtime": 0.3387,
+      "eval_samples_per_second": 14.76,
+      "eval_steps_per_second": 2.952,
+      "step": 640
+    },
+    {
+      "epoch": 1.0070257611241218,
+      "grad_norm": 0.34341296553611755,
+      "learning_rate": 0.00019999962815279418,
+      "loss": 0.4058,
+      "step": 645
+    },
+    {
+      "epoch": 1.014832162373146,
+      "grad_norm": 0.3360665440559387,
+      "learning_rate": 0.00019999851261394218,
+      "loss": 0.4051,
+      "step": 650
+    },
+    {
+      "epoch": 1.0226385636221702,
+      "grad_norm": 0.33649712800979614,
+      "learning_rate": 0.00019999665339174013,
+      "loss": 0.4055,
+      "step": 655
+    },
+    {
+      "epoch": 1.0304449648711944,
+      "grad_norm": 0.33783483505249023,
+      "learning_rate": 0.00019999405050001497,
+      "loss": 0.4124,
+      "step": 660
+    },
+    {
+      "epoch": 1.0382513661202186,
+      "grad_norm": 0.35244283080101013,
+      "learning_rate": 0.0001999907039581243,
+      "loss": 0.4148,
+      "step": 665
+    },
+    {
+      "epoch": 1.0460577673692428,
+      "grad_norm": 0.3347136974334717,
+      "learning_rate": 0.0001999866137909562,
+      "loss": 0.4032,
+      "step": 670
+    },
+    {
+      "epoch": 1.053864168618267,
+      "grad_norm": 0.34604641795158386,
+      "learning_rate": 0.0001999817800289289,
+      "loss": 0.4093,
+      "step": 675
+    },
+    {
+      "epoch": 1.0616705698672912,
+      "grad_norm": 0.36726704239845276,
+      "learning_rate": 0.00019997620270799092,
+      "loss": 0.4022,
+      "step": 680
+    },
+    {
+      "epoch": 1.0694769711163155,
+      "grad_norm": 0.410674512386322,
+      "learning_rate": 0.00019996988186962041,
+      "loss": 0.4095,
+      "step": 685
+    },
+    {
+      "epoch": 1.0772833723653397,
+      "grad_norm": 0.3476228713989258,
+      "learning_rate": 0.00019996281756082517,
+      "loss": 0.4141,
+      "step": 690
+    },
+    {
+      "epoch": 1.0850897736143639,
+      "grad_norm": 0.3787543475627899,
+      "learning_rate": 0.00019995500983414202,
+      "loss": 0.3973,
+      "step": 695
+    },
+    {
+      "epoch": 1.092896174863388,
+      "grad_norm": 0.35693004727363586,
+      "learning_rate": 0.00019994645874763658,
+      "loss": 0.4155,
+      "step": 700
+    },
+    {
+      "epoch": 1.100702576112412,
+      "grad_norm": 0.3264383375644684,
+      "learning_rate": 0.0001999371643649028,
+      "loss": 0.3979,
+      "step": 705
+    },
+    {
+      "epoch": 1.1085089773614363,
+      "grad_norm": 0.32450568675994873,
+      "learning_rate": 0.00019992712675506253,
+      "loss": 0.41,
+      "step": 710
+    },
+    {
+      "epoch": 1.1163153786104605,
+      "grad_norm": 0.3869469165802002,
+      "learning_rate": 0.00019991634599276486,
+      "loss": 0.4127,
+      "step": 715
+    },
+    {
+      "epoch": 1.1241217798594847,
+      "grad_norm": 0.46077024936676025,
+      "learning_rate": 0.0001999048221581858,
+      "loss": 0.4067,
+      "step": 720
+    },
+    {
+      "epoch": 1.131928181108509,
+      "grad_norm": 0.3199751675128937,
+      "learning_rate": 0.00019989255533702736,
+      "loss": 0.4092,
+      "step": 725
+    },
+    {
+      "epoch": 1.139734582357533,
+      "grad_norm": 0.341310054063797,
+      "learning_rate": 0.00019987954562051725,
+      "loss": 0.4192,
+      "step": 730
+    },
+    {
+      "epoch": 1.1475409836065573,
+      "grad_norm": 0.3507387340068817,
+      "learning_rate": 0.000199865793105408,
+      "loss": 0.4063,
+      "step": 735
+    },
+    {
+      "epoch": 1.1553473848555815,
+      "grad_norm": 0.33445993065834045,
+      "learning_rate": 0.00019985129789397633,
+      "loss": 0.4178,
+      "step": 740
+    },
+    {
+      "epoch": 1.1631537861046057,
+      "grad_norm": 0.3056560158729553,
+      "learning_rate": 0.00019983606009402224,
+      "loss": 0.395,
+      "step": 745
+    },
+    {
+      "epoch": 1.17096018735363,
+      "grad_norm": 0.31608232855796814,
+      "learning_rate": 0.00019982007981886847,
+      "loss": 0.4089,
+      "step": 750
+    },
+    {
+      "epoch": 1.1787665886026542,
+      "grad_norm": 0.44849371910095215,
+      "learning_rate": 0.00019980335718735944,
+      "loss": 0.4042,
+      "step": 755
+    },
+    {
+      "epoch": 1.1865729898516784,
+      "grad_norm": 0.34841999411582947,
+      "learning_rate": 0.00019978589232386035,
+      "loss": 0.4017,
+      "step": 760
+    },
+    {
+      "epoch": 1.1943793911007026,
+      "grad_norm": 0.30834051966667175,
+      "learning_rate": 0.0001997676853582565,
+      "loss": 0.4059,
+      "step": 765
+    },
+    {
+      "epoch": 1.2021857923497268,
+      "grad_norm": 0.319698303937912,
+      "learning_rate": 0.000199748736425952,
+      "loss": 0.4133,
+      "step": 770
+    },
+    {
+      "epoch": 1.209992193598751,
+      "grad_norm": 0.3290104269981384,
+      "learning_rate": 0.00019972904566786903,
+      "loss": 0.4025,
+      "step": 775
+    },
+    {
+      "epoch": 1.2177985948477752,
+      "grad_norm": 0.3340936601161957,
+      "learning_rate": 0.00019970861323044667,
+      "loss": 0.4056,
+      "step": 780
+    },
+    {
+      "epoch": 1.2256049960967994,
+      "grad_norm": 0.32487761974334717,
+      "learning_rate": 0.00019968743926563983,
+      "loss": 0.4043,
+      "step": 785
+    },
+    {
+      "epoch": 1.2334113973458236,
+      "grad_norm": 0.3178825378417969,
+      "learning_rate": 0.00019966552393091804,
+      "loss": 0.4063,
+      "step": 790
+    },
+    {
+      "epoch": 1.2412177985948478,
+      "grad_norm": 0.3003855347633362,
+      "learning_rate": 0.00019964286738926448,
+      "loss": 0.3961,
+      "step": 795
+    },
+    {
+      "epoch": 1.249024199843872,
+      "grad_norm": 0.32938238978385925,
+      "learning_rate": 0.00019961946980917456,
+      "loss": 0.3991,
+      "step": 800
+    },
+    {
+      "epoch": 1.2568306010928962,
+      "grad_norm": 0.3072170615196228,
+      "learning_rate": 0.0001995953313646548,
+      "loss": 0.3943,
+      "step": 805
+    },
+    {
+      "epoch": 1.2646370023419204,
+      "grad_norm": 0.36918073892593384,
+      "learning_rate": 0.0001995704522352214,
+      "loss": 0.3998,
+      "step": 810
+    },
+    {
+      "epoch": 1.2724434035909447,
+      "grad_norm": 0.3280356228351593,
+      "learning_rate": 0.00019954483260589911,
+      "loss": 0.4035,
+      "step": 815
+    },
+    {
+      "epoch": 1.2802498048399689,
+      "grad_norm": 0.29702678322792053,
+      "learning_rate": 0.0001995184726672197,
+      "loss": 0.3901,
+      "step": 820
+    },
+    {
+      "epoch": 1.288056206088993,
+      "grad_norm": 0.30307725071907043,
+      "learning_rate": 0.00019949137261522052,
+      "loss": 0.4053,
+      "step": 825
+    },
+    {
+      "epoch": 1.2958626073380173,
+      "grad_norm": 0.31273552775382996,
+      "learning_rate": 0.00019946353265144312,
+      "loss": 0.4136,
+      "step": 830
+    },
+    {
+      "epoch": 1.3036690085870415,
+      "grad_norm": 0.30774056911468506,
+      "learning_rate": 0.00019943495298293181,
+      "loss": 0.3995,
+      "step": 835
+    },
+    {
+      "epoch": 1.3114754098360657,
+      "grad_norm": 0.3068085312843323,
+      "learning_rate": 0.00019940563382223197,
+      "loss": 0.4136,
+      "step": 840
+    },
+    {
+      "epoch": 1.31928181108509,
+      "grad_norm": 0.33068013191223145,
+      "learning_rate": 0.00019937557538738854,
+      "loss": 0.3957,
+      "step": 845
+    },
+    {
+      "epoch": 1.327088212334114,
+      "grad_norm": 0.36017951369285583,
+      "learning_rate": 0.00019934477790194445,
+      "loss": 0.4094,
+      "step": 850
+    },
+    {
+      "epoch": 1.334894613583138,
+      "grad_norm": 0.3393058776855469,
+      "learning_rate": 0.00019931324159493886,
+      "loss": 0.4072,
+      "step": 855
+    },
+    {
+      "epoch": 1.3427010148321623,
+      "grad_norm": 0.29926055669784546,
+      "learning_rate": 0.0001992809667009055,
+      "loss": 0.3991,
+      "step": 860
+    },
+    {
+      "epoch": 1.3505074160811865,
+      "grad_norm": 0.3126608431339264,
+      "learning_rate": 0.000199247953459871,
+      "loss": 0.3997,
+      "step": 865
+    },
+    {
+      "epoch": 1.3583138173302107,
+      "grad_norm": 0.31976112723350525,
+      "learning_rate": 0.00019921420211735299,
+      "loss": 0.4043,
+      "step": 870
+    },
+    {
+      "epoch": 1.366120218579235,
+      "grad_norm": 0.37549304962158203,
+      "learning_rate": 0.00019917971292435826,
+      "loss": 0.4112,
+      "step": 875
+    },
+    {
+      "epoch": 1.3739266198282591,
+      "grad_norm": 0.3190479278564453,
+      "learning_rate": 0.00019914448613738106,
+      "loss": 0.402,
+      "step": 880
+    },
+    {
+      "epoch": 1.3817330210772834,
+      "grad_norm": 0.30075913667678833,
+      "learning_rate": 0.00019910852201840103,
+      "loss": 0.4159,
+      "step": 885
+    },
+    {
+      "epoch": 1.3895394223263076,
+      "grad_norm": 0.3074491024017334,
+      "learning_rate": 0.00019907182083488129,
+      "loss": 0.4045,
+      "step": 890
+    },
+    {
+      "epoch": 1.3973458235753318,
+      "grad_norm": 0.3855604827404022,
+      "learning_rate": 0.0001990343828597665,
+      "loss": 0.4045,
+      "step": 895
+    },
+    {
+      "epoch": 1.405152224824356,
+      "grad_norm": 0.3558739721775055,
+      "learning_rate": 0.00019899620837148077,
+      "loss": 0.3969,
+      "step": 900
+    },
+    {
+      "epoch": 1.4129586260733802,
+      "grad_norm": 0.3128485083580017,
+      "learning_rate": 0.00019895729765392572,
+      "loss": 0.403,
+      "step": 905
+    },
+    {
+      "epoch": 1.4207650273224044,
+      "grad_norm": 0.30848538875579834,
+      "learning_rate": 0.0001989176509964781,
+      "loss": 0.398,
+      "step": 910
+    },
+    {
+      "epoch": 1.4285714285714286,
+      "grad_norm": 0.2908243238925934,
+      "learning_rate": 0.00019887726869398793,
+      "loss": 0.394,
+      "step": 915
+    },
+    {
+      "epoch": 1.4363778298204528,
+      "grad_norm": 0.29241564869880676,
+      "learning_rate": 0.00019883615104677608,
+      "loss": 0.4007,
+      "step": 920
+    },
+    {
+      "epoch": 1.444184231069477,
+      "grad_norm": 0.29839983582496643,
+      "learning_rate": 0.00019879429836063226,
+      "loss": 0.3961,
+      "step": 925
+    },
+    {
+      "epoch": 1.4519906323185012,
+      "grad_norm": 0.298441618680954,
+      "learning_rate": 0.00019875171094681248,
+      "loss": 0.4003,
+      "step": 930
+    },
+    {
+      "epoch": 1.4597970335675254,
+      "grad_norm": 0.30495795607566833,
+      "learning_rate": 0.00019870838912203705,
+      "loss": 0.4036,
+      "step": 935
+    },
+    {
+      "epoch": 1.4676034348165496,
+      "grad_norm": 0.2891780436038971,
+      "learning_rate": 0.0001986643332084879,
+      "loss": 0.3901,
+      "step": 940
+    },
+    {
+      "epoch": 1.4754098360655736,
+      "grad_norm": 0.2930201292037964,
+      "learning_rate": 0.00019861954353380643,
+      "loss": 0.3943,
+      "step": 945
+    },
+    {
+      "epoch": 1.4832162373145978,
+      "grad_norm": 0.289411336183548,
+      "learning_rate": 0.0001985740204310909,
+      "loss": 0.4003,
+      "step": 950
+    },
+    {
+      "epoch": 1.491022638563622,
+      "grad_norm": 0.3018743693828583,
+      "learning_rate": 0.0001985277642388941,
+      "loss": 0.3949,
+      "step": 955
+    },
+    {
+      "epoch": 1.4988290398126463,
+      "grad_norm": 0.3346049189567566,
+      "learning_rate": 0.00019848077530122083,
+      "loss": 0.3949,
+      "step": 960
+    },
+    {
+      "epoch": 1.5066354410616705,
+      "grad_norm": 0.35057857632637024,
+      "learning_rate": 0.00019843305396752509,
+      "loss": 0.3916,
+      "step": 965
+    },
+    {
+      "epoch": 1.5144418423106947,
+      "grad_norm": 0.33775442838668823,
+      "learning_rate": 0.00019838460059270775,
+      "loss": 0.4003,
+      "step": 970
+    },
+    {
+      "epoch": 1.5222482435597189,
+      "grad_norm": 0.3016526997089386,
+      "learning_rate": 0.00019833541553711395,
+      "loss": 0.4032,
+      "step": 975
+    },
+    {
+      "epoch": 1.530054644808743,
+      "grad_norm": 0.2809526026248932,
+      "learning_rate": 0.0001982854991665301,
+      "loss": 0.3983,
+      "step": 980
+    },
+    {
+      "epoch": 1.5378610460577673,
+      "grad_norm": 0.3081335723400116,
+      "learning_rate": 0.00019823485185218158,
+      "loss": 0.3997,
+      "step": 985
+    },
+    {
+      "epoch": 1.5456674473067915,
+      "grad_norm": 0.28975608944892883,
+      "learning_rate": 0.00019818347397072955,
+      "loss": 0.3943,
+      "step": 990
+    },
+    {
+      "epoch": 1.5534738485558157,
+      "grad_norm": 0.32257601618766785,
+      "learning_rate": 0.00019813136590426844,
+      "loss": 0.3948,
+      "step": 995
+    },
+    {
+      "epoch": 1.56128024980484,
+      "grad_norm": 0.3275687098503113,
+      "learning_rate": 0.00019807852804032305,
+      "loss": 0.4001,
+      "step": 1000
+    },
+    {
+      "epoch": 1.5690866510538641,
+      "grad_norm": 0.3463219404220581,
+      "learning_rate": 0.0001980249607718456,
+      "loss": 0.3992,
+      "step": 1005
+    },
+    {
+      "epoch": 1.5768930523028883,
+      "grad_norm": 0.3080371022224426,
+      "learning_rate": 0.00019797066449721295,
+      "loss": 0.4023,
+      "step": 1010
+    },
+    {
+      "epoch": 1.5846994535519126,
+      "grad_norm": 0.29140976071357727,
+      "learning_rate": 0.00019791563962022338,
+      "loss": 0.4036,
+      "step": 1015
+    },
+    {
+      "epoch": 1.5925058548009368,
+      "grad_norm": 0.3033461570739746,
+      "learning_rate": 0.00019785988655009385,
+      "loss": 0.4051,
+      "step": 1020
+    },
+    {
+      "epoch": 1.600312256049961,
+      "grad_norm": 0.27476766705513,
+      "learning_rate": 0.0001978034057014568,
+      "loss": 0.3818,
+      "step": 1025
+    },
+    {
+      "epoch": 1.6081186572989852,
+      "grad_norm": 0.3026203513145447,
+      "learning_rate": 0.0001977461974943572,
+      "loss": 0.3959,
+      "step": 1030
+    },
+    {
+      "epoch": 1.6159250585480094,
+      "grad_norm": 0.28781238198280334,
+      "learning_rate": 0.00019768826235424926,
+      "loss": 0.4067,
+      "step": 1035
+    },
+    {
+      "epoch": 1.6237314597970336,
+      "grad_norm": 0.28916335105895996,
+      "learning_rate": 0.00019762960071199333,
+      "loss": 0.3999,
+      "step": 1040
+    },
+    {
+      "epoch": 1.6315378610460578,
+      "grad_norm": 0.2844083607196808,
+      "learning_rate": 0.00019757021300385286,
+      "loss": 0.3958,
+      "step": 1045
+    },
+    {
+      "epoch": 1.639344262295082,
+      "grad_norm": 0.30581745505332947,
+      "learning_rate": 0.00019751009967149087,
+      "loss": 0.4016,
+      "step": 1050
+    },
+    {
+      "epoch": 1.6471506635441062,
+      "grad_norm": 0.27910909056663513,
+      "learning_rate": 0.00019744926116196685,
+      "loss": 0.3903,
+      "step": 1055
+    },
+    {
+      "epoch": 1.6549570647931304,
+      "grad_norm": 0.2814294695854187,
+      "learning_rate": 0.00019738769792773336,
+      "loss": 0.3896,
+      "step": 1060
+    },
+    {
+      "epoch": 1.6627634660421546,
+      "grad_norm": 0.29259413480758667,
+      "learning_rate": 0.00019732541042663282,
+      "loss": 0.4039,
+      "step": 1065
+    },
+    {
+      "epoch": 1.6705698672911788,
+      "grad_norm": 0.2868483364582062,
+      "learning_rate": 0.00019726239912189382,
+      "loss": 0.3893,
+      "step": 1070
+    },
+    {
+      "epoch": 1.678376268540203,
+      "grad_norm": 0.30284589529037476,
+      "learning_rate": 0.00019719866448212795,
+      "loss": 0.3973,
+      "step": 1075
+    },
+    {
+      "epoch": 1.6861826697892273,
+      "grad_norm": 0.29238229990005493,
+      "learning_rate": 0.00019713420698132614,
+      "loss": 0.404,
+      "step": 1080
+    },
+    {
+      "epoch": 1.6939890710382515,
+      "grad_norm": 0.2957955300807953,
+      "learning_rate": 0.00019706902709885524,
+      "loss": 0.3929,
+      "step": 1085
+    },
+    {
+      "epoch": 1.7017954722872757,
+      "grad_norm": 0.29362279176712036,
+      "learning_rate": 0.00019700312531945442,
+      "loss": 0.393,
+      "step": 1090
+    },
+    {
+      "epoch": 1.7096018735362999,
+      "grad_norm": 0.3154332637786865,
+      "learning_rate": 0.00019693650213323144,
+      "loss": 0.3971,
+      "step": 1095
+    },
+    {
+      "epoch": 1.717408274785324,
+      "grad_norm": 0.31909048557281494,
+      "learning_rate": 0.00019686915803565934,
+      "loss": 0.3991,
+      "step": 1100
+    },
+    {
+      "epoch": 1.7252146760343483,
+      "grad_norm": 0.31985750794410706,
+      "learning_rate": 0.00019680109352757227,
+      "loss": 0.4076,
+      "step": 1105
+    },
+    {
+      "epoch": 1.7330210772833725,
+      "grad_norm": 0.2867513597011566,
+      "learning_rate": 0.00019673230911516226,
+      "loss": 0.3957,
+      "step": 1110
+    },
+    {
+      "epoch": 1.7408274785323967,
+      "grad_norm": 0.30128851532936096,
+      "learning_rate": 0.00019666280530997518,
+      "loss": 0.4041,
+      "step": 1115
+    },
+    {
+      "epoch": 1.748633879781421,
+      "grad_norm": 0.2900910973548889,
+      "learning_rate": 0.00019659258262890683,
+      "loss": 0.3898,
+      "step": 1120
+    },
+    {
+      "epoch": 1.756440281030445,
+      "grad_norm": 0.3005325496196747,
+      "learning_rate": 0.00019652164159419946,
+      "loss": 0.4086,
+      "step": 1125
+    },
+    {
+      "epoch": 1.7642466822794691,
+      "grad_norm": 0.2881026268005371,
+      "learning_rate": 0.00019644998273343753,
+      "loss": 0.4005,
+      "step": 1130
+    },
+    {
+      "epoch": 1.7720530835284933,
+      "grad_norm": 0.2854783535003662,
+      "learning_rate": 0.000196377606579544,
+      "loss": 0.3942,
+      "step": 1135
+    },
+    {
+      "epoch": 1.7798594847775175,
+      "grad_norm": 0.2979474365711212,
+      "learning_rate": 0.00019630451367077628,
+      "loss": 0.3935,
+      "step": 1140
+    },
+    {
+      "epoch": 1.7876658860265418,
+      "grad_norm": 0.28160732984542847,
+      "learning_rate": 0.00019623070455072224,
+      "loss": 0.4153,
+      "step": 1145
+    },
+    {
+      "epoch": 1.795472287275566,
+      "grad_norm": 0.27550724148750305,
+      "learning_rate": 0.0001961561797682962,
+      "loss": 0.3921,
+      "step": 1150
+    },
+    {
+      "epoch": 1.8032786885245902,
+      "grad_norm": 0.27477726340293884,
+      "learning_rate": 0.00019608093987773478,
+      "loss": 0.4017,
+      "step": 1155
+    },
+    {
+      "epoch": 1.8110850897736144,
+      "grad_norm": 0.2824558913707733,
+      "learning_rate": 0.0001960049854385929,
+      "loss": 0.3897,
+      "step": 1160
+    },
+    {
+      "epoch": 1.8188914910226386,
+      "grad_norm": 0.2761404514312744,
+      "learning_rate": 0.00019592831701573937,
+      "loss": 0.4095,
+      "step": 1165
+    },
+    {
+      "epoch": 1.8266978922716628,
+      "grad_norm": 0.28211623430252075,
+      "learning_rate": 0.00019585093517935305,
+      "loss": 0.3912,
+      "step": 1170
+    },
+    {
+      "epoch": 1.834504293520687,
+      "grad_norm": 0.27295398712158203,
+      "learning_rate": 0.0001957728405049183,
+      "loss": 0.4125,
+      "step": 1175
+    },
+    {
+      "epoch": 1.8423106947697112,
+      "grad_norm": 0.2896745800971985,
+      "learning_rate": 0.0001956940335732209,
+      "loss": 0.3927,
+      "step": 1180
+    },
+    {
+      "epoch": 1.8501170960187352,
+      "grad_norm": 0.3007790446281433,
+      "learning_rate": 0.00019561451497034354,
+      "loss": 0.3932,
+      "step": 1185
+    },
+    {
+      "epoch": 1.8579234972677594,
+      "grad_norm": 0.29394078254699707,
+      "learning_rate": 0.00019553428528766163,
+      "loss": 0.3905,
+      "step": 1190
+    },
+    {
+      "epoch": 1.8657298985167836,
+      "grad_norm": 0.2767265737056732,
+      "learning_rate": 0.00019545334512183885,
+      "loss": 0.395,
+      "step": 1195
+    },
+    {
+      "epoch": 1.8735362997658078,
+      "grad_norm": 0.26828575134277344,
+      "learning_rate": 0.0001953716950748227,
+      "loss": 0.3996,
+      "step": 1200
+    },
+    {
+      "epoch": 1.881342701014832,
+      "grad_norm": 0.2718822658061981,
+      "learning_rate": 0.00019528933575384,
+      "loss": 0.3803,
+      "step": 1205
+    },
+    {
+      "epoch": 1.8891491022638562,
+      "grad_norm": 0.2866761386394501,
+      "learning_rate": 0.00019520626777139245,
+      "loss": 0.4056,
+      "step": 1210
+    },
+    {
+      "epoch": 1.8969555035128804,
+      "grad_norm": 0.3028269410133362,
+      "learning_rate": 0.00019512249174525193,
+      "loss": 0.4,
+      "step": 1215
+    },
+    {
+      "epoch": 1.9047619047619047,
+      "grad_norm": 0.2755603492259979,
+      "learning_rate": 0.0001950380082984561,
+      "loss": 0.3829,
+      "step": 1220
+    },
+    {
+      "epoch": 1.9125683060109289,
+      "grad_norm": 0.28492555022239685,
+      "learning_rate": 0.00019495281805930367,
+      "loss": 0.3839,
+      "step": 1225
+    },
+    {
+      "epoch": 1.920374707259953,
+      "grad_norm": 0.2751103639602661,
+      "learning_rate": 0.00019486692166134964,
+      "loss": 0.3907,
+      "step": 1230
+    },
+    {
+      "epoch": 1.9281811085089773,
+      "grad_norm": 0.2826032042503357,
+      "learning_rate": 0.00019478031974340073,
+      "loss": 0.3884,
+      "step": 1235
+    },
+    {
+      "epoch": 1.9359875097580015,
+      "grad_norm": 0.28445425629615784,
+      "learning_rate": 0.0001946930129495106,
+      "loss": 0.3956,
+      "step": 1240
+    },
+    {
+      "epoch": 1.9437939110070257,
+      "grad_norm": 0.2712079584598541,
+      "learning_rate": 0.0001946050019289749,
+      "loss": 0.3893,
+      "step": 1245
+    },
+    {
+      "epoch": 1.95160031225605,
+      "grad_norm": 0.27717921137809753,
+      "learning_rate": 0.0001945162873363268,
+      "loss": 0.3912,
+      "step": 1250
+    },
+    {
+      "epoch": 1.9594067135050741,
+      "grad_norm": 0.2907007336616516,
+      "learning_rate": 0.00019442686983133168,
+      "loss": 0.4004,
+      "step": 1255
+    },
+    {
+      "epoch": 1.9672131147540983,
+      "grad_norm": 0.2781454026699066,
+      "learning_rate": 0.00019433675007898255,
+      "loss": 0.4037,
+      "step": 1260
+    },
+    {
+      "epoch": 1.9750195160031225,
+      "grad_norm": 0.27712446451187134,
+      "learning_rate": 0.00019424592874949496,
+      "loss": 0.3902,
+      "step": 1265
+    },
+    {
+      "epoch": 1.9828259172521467,
+      "grad_norm": 0.2906958758831024,
+      "learning_rate": 0.00019415440651830208,
+      "loss": 0.3955,
+      "step": 1270
+    },
+    {
+      "epoch": 1.990632318501171,
+      "grad_norm": 0.28120070695877075,
+      "learning_rate": 0.00019406218406604965,
+      "loss": 0.3878,
+      "step": 1275
+    },
+    {
+      "epoch": 1.9984387197501952,
+      "grad_norm": 0.2804274260997772,
+      "learning_rate": 0.00019396926207859084,
+      "loss": 0.3829,
+      "step": 1280
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 1.0417580604553223,
+      "eval_runtime": 0.3233,
+      "eval_samples_per_second": 15.465,
+      "eval_steps_per_second": 3.093,
+      "step": 1281
+    },
+    {
+      "epoch": 2.0062451209992194,
+      "grad_norm": 0.2731517255306244,
+      "learning_rate": 0.00019387564124698133,
+      "loss": 0.3747,
+      "step": 1285
+    },
+    {
+      "epoch": 2.0140515222482436,
+      "grad_norm": 0.2837215065956116,
+      "learning_rate": 0.00019378132226747398,
+      "loss": 0.3651,
+      "step": 1290
+    },
+    {
+      "epoch": 2.021857923497268,
+      "grad_norm": 0.29142889380455017,
+      "learning_rate": 0.0001936863058415138,
+      "loss": 0.3736,
+      "step": 1295
+    },
+    {
+      "epoch": 2.029664324746292,
+      "grad_norm": 0.29420003294944763,
+      "learning_rate": 0.0001935905926757326,
+      "loss": 0.3764,
+      "step": 1300
+    },
+    {
+      "epoch": 2.037470725995316,
+      "grad_norm": 0.2812630832195282,
+      "learning_rate": 0.00019349418348194383,
+      "loss": 0.3674,
+      "step": 1305
+    },
+    {
+      "epoch": 2.0452771272443404,
+      "grad_norm": 0.3169541656970978,
+      "learning_rate": 0.00019339707897713738,
+      "loss": 0.3625,
+      "step": 1310
+    },
+    {
+      "epoch": 2.0530835284933646,
+      "grad_norm": 0.31742027401924133,
+      "learning_rate": 0.0001932992798834739,
+      "loss": 0.3718,
+      "step": 1315
+    },
+    {
+      "epoch": 2.060889929742389,
+      "grad_norm": 0.2889857888221741,
+      "learning_rate": 0.00019320078692827987,
+      "loss": 0.3798,
+      "step": 1320
+    },
+    {
+      "epoch": 2.068696330991413,
+      "grad_norm": 0.27867743372917175,
+      "learning_rate": 0.00019310160084404186,
+      "loss": 0.3647,
+      "step": 1325
+    },
+    {
+      "epoch": 2.0765027322404372,
+      "grad_norm": 0.29241612553596497,
+      "learning_rate": 0.00019300172236840122,
+      "loss": 0.3761,
+      "step": 1330
+    },
+    {
+      "epoch": 2.0843091334894615,
+      "grad_norm": 0.28482168912887573,
+      "learning_rate": 0.00019290115224414862,
+      "loss": 0.3743,
+      "step": 1335
+    },
+    {
+      "epoch": 2.0921155347384857,
+      "grad_norm": 0.3034825325012207,
+      "learning_rate": 0.00019279989121921847,
+      "loss": 0.3757,
+      "step": 1340
+    },
+    {
+      "epoch": 2.09992193598751,
+      "grad_norm": 0.3016110956668854,
+      "learning_rate": 0.0001926979400466833,
+      "loss": 0.3774,
+      "step": 1345
+    },
+    {
+      "epoch": 2.107728337236534,
+      "grad_norm": 0.3048888146877289,
+      "learning_rate": 0.00019259529948474833,
+      "loss": 0.369,
+      "step": 1350
+    },
+    {
+      "epoch": 2.1155347384855583,
+      "grad_norm": 0.2792314291000366,
+      "learning_rate": 0.00019249197029674566,
+      "loss": 0.3727,
+      "step": 1355
+    },
+    {
+      "epoch": 2.1233411397345825,
+      "grad_norm": 0.30155235528945923,
+      "learning_rate": 0.0001923879532511287,
+      "loss": 0.3691,
+      "step": 1360
+    },
+    {
+      "epoch": 2.1311475409836067,
+      "grad_norm": 0.2807096838951111,
+      "learning_rate": 0.0001922832491214664,
+      "loss": 0.3618,
+      "step": 1365
+    },
+    {
+      "epoch": 2.138953942232631,
+      "grad_norm": 0.2897432744503021,
+      "learning_rate": 0.00019217785868643747,
+      "loss": 0.3644,
+      "step": 1370
+    },
+    {
+      "epoch": 2.146760343481655,
+      "grad_norm": 0.2933688759803772,
+      "learning_rate": 0.0001920717827298248,
+      "loss": 0.3753,
+      "step": 1375
+    },
+    {
+      "epoch": 2.1545667447306793,
+      "grad_norm": 0.2933692932128906,
+      "learning_rate": 0.00019196502204050922,
+      "loss": 0.3725,
+      "step": 1380
+    },
+    {
+      "epoch": 2.1623731459797035,
+      "grad_norm": 0.28731822967529297,
+      "learning_rate": 0.00019185757741246412,
+      "loss": 0.3733,
+      "step": 1385
+    },
+    {
+      "epoch": 2.1701795472287277,
+      "grad_norm": 0.3193456530570984,
+      "learning_rate": 0.00019174944964474912,
+      "loss": 0.3741,
+      "step": 1390
+    },
+    {
+      "epoch": 2.177985948477752,
+      "grad_norm": 0.2853163480758667,
+      "learning_rate": 0.00019164063954150447,
+      "loss": 0.3774,
+      "step": 1395
+    },
+    {
+      "epoch": 2.185792349726776,
+      "grad_norm": 0.3006320893764496,
+      "learning_rate": 0.00019153114791194473,
+      "loss": 0.3677,
+      "step": 1400
+    },
+    {
+      "epoch": 2.1935987509758004,
+      "grad_norm": 0.2752683758735657,
+      "learning_rate": 0.00019142097557035308,
+      "loss": 0.3675,
+      "step": 1405
+    },
+    {
+      "epoch": 2.201405152224824,
+      "grad_norm": 0.2900081276893616,
+      "learning_rate": 0.00019131012333607506,
+      "loss": 0.375,
+      "step": 1410
+    },
+    {
+      "epoch": 2.209211553473849,
+      "grad_norm": 0.3021766245365143,
+      "learning_rate": 0.00019119859203351255,
+      "loss": 0.3778,
+      "step": 1415
+    },
+    {
+      "epoch": 2.2170179547228726,
+      "grad_norm": 0.2842157185077667,
+      "learning_rate": 0.00019108638249211758,
+      "loss": 0.3668,
+      "step": 1420
+    },
+    {
+      "epoch": 2.2248243559718968,
+      "grad_norm": 0.27962204813957214,
+      "learning_rate": 0.0001909734955463863,
+      "loss": 0.3731,
+      "step": 1425
+    },
+    {
+      "epoch": 2.232630757220921,
+      "grad_norm": 0.29117563366889954,
+      "learning_rate": 0.0001908599320358526,
+      "loss": 0.3726,
+      "step": 1430
+    },
+    {
+      "epoch": 2.240437158469945,
+      "grad_norm": 0.307558536529541,
+      "learning_rate": 0.0001907456928050819,
+      "loss": 0.37,
+      "step": 1435
+    },
+    {
+      "epoch": 2.2482435597189694,
+      "grad_norm": 0.3160630762577057,
+      "learning_rate": 0.000190630778703665,
+      "loss": 0.3769,
+      "step": 1440
+    },
+    {
+      "epoch": 2.2560499609679936,
+      "grad_norm": 0.2901064157485962,
+      "learning_rate": 0.0001905151905862117,
+      "loss": 0.3703,
+      "step": 1445
+    },
+    {
+      "epoch": 2.263856362217018,
+      "grad_norm": 0.2909891605377197,
+      "learning_rate": 0.00019039892931234435,
+      "loss": 0.374,
+      "step": 1450
+    },
+    {
+      "epoch": 2.271662763466042,
+      "grad_norm": 0.3095717132091522,
+      "learning_rate": 0.00019028199574669153,
+      "loss": 0.3848,
+      "step": 1455
+    },
+    {
+      "epoch": 2.279469164715066,
+      "grad_norm": 0.2892928421497345,
+      "learning_rate": 0.00019016439075888162,
+      "loss": 0.364,
+      "step": 1460
+    },
+    {
+      "epoch": 2.2872755659640904,
+      "grad_norm": 0.3012056350708008,
+      "learning_rate": 0.00019004611522353636,
+      "loss": 0.3704,
+      "step": 1465
+    },
+    {
+      "epoch": 2.2950819672131146,
+      "grad_norm": 0.28810742497444153,
+      "learning_rate": 0.00018992717002026433,
+      "loss": 0.3757,
+      "step": 1470
+    },
+    {
+      "epoch": 2.302888368462139,
+      "grad_norm": 0.3010272979736328,
+      "learning_rate": 0.0001898075560336543,
+      "loss": 0.3695,
+      "step": 1475
+    },
+    {
+      "epoch": 2.310694769711163,
+      "grad_norm": 0.2966187000274658,
+      "learning_rate": 0.00018968727415326884,
+      "loss": 0.3753,
+      "step": 1480
+    },
+    {
+      "epoch": 2.3185011709601873,
+      "grad_norm": 0.303035169839859,
+      "learning_rate": 0.00018956632527363757,
+      "loss": 0.3844,
+      "step": 1485
+    },
+    {
+      "epoch": 2.3263075722092115,
+      "grad_norm": 0.2904973030090332,
+      "learning_rate": 0.00018944471029425054,
+      "loss": 0.3733,
+      "step": 1490
+    },
+    {
+      "epoch": 2.3341139734582357,
+      "grad_norm": 0.2944584786891937,
+      "learning_rate": 0.00018932243011955154,
+      "loss": 0.3705,
+      "step": 1495
+    },
+    {
+      "epoch": 2.34192037470726,
+      "grad_norm": 0.3047090470790863,
+      "learning_rate": 0.00018919948565893142,
+      "loss": 0.3776,
+      "step": 1500
+    },
+    {
+      "epoch": 2.349726775956284,
+      "grad_norm": 0.28889063000679016,
+      "learning_rate": 0.00018907587782672128,
+      "loss": 0.3748,
+      "step": 1505
+    },
+    {
+      "epoch": 2.3575331772053083,
+      "grad_norm": 0.29488006234169006,
+      "learning_rate": 0.00018895160754218561,
+      "loss": 0.3719,
+      "step": 1510
+    },
+    {
+      "epoch": 2.3653395784543325,
+      "grad_norm": 0.28277385234832764,
+      "learning_rate": 0.00018882667572951562,
+      "loss": 0.3737,
+      "step": 1515
+    },
+    {
+      "epoch": 2.3731459797033567,
+      "grad_norm": 0.2788912057876587,
+      "learning_rate": 0.00018870108331782217,
+      "loss": 0.37,
+      "step": 1520
+    },
+    {
+      "epoch": 2.380952380952381,
+      "grad_norm": 0.2984495759010315,
+      "learning_rate": 0.00018857483124112907,
+      "loss": 0.3816,
+      "step": 1525
+    },
+    {
+      "epoch": 2.388758782201405,
+      "grad_norm": 0.28799429535865784,
+      "learning_rate": 0.00018844792043836589,
+      "loss": 0.3849,
+      "step": 1530
+    },
+    {
+      "epoch": 2.3965651834504293,
+      "grad_norm": 0.2892215847969055,
+      "learning_rate": 0.00018832035185336124,
+      "loss": 0.372,
+      "step": 1535
+    },
+    {
+      "epoch": 2.4043715846994536,
+      "grad_norm": 0.30197983980178833,
+      "learning_rate": 0.0001881921264348355,
+      "loss": 0.3764,
+      "step": 1540
+    },
+    {
+      "epoch": 2.4121779859484778,
+      "grad_norm": 0.3232642412185669,
+      "learning_rate": 0.000188063245136394,
+      "loss": 0.3788,
+      "step": 1545
+    },
+    {
+      "epoch": 2.419984387197502,
+      "grad_norm": 0.29107898473739624,
+      "learning_rate": 0.00018793370891651972,
+      "loss": 0.3786,
+      "step": 1550
+    },
+    {
+      "epoch": 2.427790788446526,
+      "grad_norm": 0.29140210151672363,
+      "learning_rate": 0.00018780351873856627,
+      "loss": 0.3763,
+      "step": 1555
+    },
+    {
+      "epoch": 2.4355971896955504,
+      "grad_norm": 0.30477312207221985,
+      "learning_rate": 0.0001876726755707508,
+      "loss": 0.3866,
+      "step": 1560
+    },
+    {
+      "epoch": 2.4434035909445746,
+      "grad_norm": 0.3051372170448303,
+      "learning_rate": 0.00018754118038614653,
+      "loss": 0.378,
+      "step": 1565
+    },
+    {
+      "epoch": 2.451209992193599,
+      "grad_norm": 0.2856997847557068,
+      "learning_rate": 0.0001874090341626759,
+      "loss": 0.3744,
+      "step": 1570
+    },
+    {
+      "epoch": 2.459016393442623,
+      "grad_norm": 0.26825210452079773,
+      "learning_rate": 0.00018727623788310292,
+      "loss": 0.369,
+      "step": 1575
+    },
+    {
+      "epoch": 2.4668227946916472,
+      "grad_norm": 0.2855939269065857,
+      "learning_rate": 0.00018714279253502616,
+      "loss": 0.3829,
+      "step": 1580
+    },
+    {
+      "epoch": 2.4746291959406714,
+      "grad_norm": 0.28602635860443115,
+      "learning_rate": 0.00018700869911087115,
+      "loss": 0.3657,
+      "step": 1585
+    },
+    {
+      "epoch": 2.4824355971896956,
+      "grad_norm": 0.30138617753982544,
+      "learning_rate": 0.00018687395860788323,
+      "loss": 0.3694,
+      "step": 1590
+    },
+    {
+      "epoch": 2.49024199843872,
+      "grad_norm": 0.2945271134376526,
+      "learning_rate": 0.00018673857202812,
+      "loss": 0.3663,
+      "step": 1595
+    },
+    {
+      "epoch": 2.498048399687744,
+      "grad_norm": 0.28233060240745544,
+      "learning_rate": 0.00018660254037844388,
+      "loss": 0.3839,
+      "step": 1600
+    },
+    {
+      "epoch": 2.5058548009367683,
+      "grad_norm": 0.30950790643692017,
+      "learning_rate": 0.00018646586467051463,
+      "loss": 0.3864,
+      "step": 1605
+    },
+    {
+      "epoch": 2.5136612021857925,
+      "grad_norm": 0.30938637256622314,
+      "learning_rate": 0.00018632854592078184,
+      "loss": 0.3703,
+      "step": 1610
+    },
+    {
+      "epoch": 2.5214676034348167,
+      "grad_norm": 0.2714425325393677,
+      "learning_rate": 0.00018619058515047745,
+      "loss": 0.3748,
+      "step": 1615
+    },
+    {
+      "epoch": 2.529274004683841,
+      "grad_norm": 0.30516138672828674,
+      "learning_rate": 0.0001860519833856079,
+      "loss": 0.3868,
+      "step": 1620
+    },
+    {
+      "epoch": 2.537080405932865,
+      "grad_norm": 0.30349424481391907,
+      "learning_rate": 0.00018591274165694687,
+      "loss": 0.3682,
+      "step": 1625
+    },
+    {
+      "epoch": 2.5448868071818893,
+      "grad_norm": 0.29412826895713806,
+      "learning_rate": 0.00018577286100002723,
+      "loss": 0.3683,
+      "step": 1630
+    },
+    {
+      "epoch": 2.552693208430913,
+      "grad_norm": 0.2976747155189514,
+      "learning_rate": 0.00018563234245513364,
+      "loss": 0.3697,
+      "step": 1635
+    },
+    {
+      "epoch": 2.5604996096799377,
+      "grad_norm": 0.28876587748527527,
+      "learning_rate": 0.00018549118706729468,
+      "loss": 0.3751,
+      "step": 1640
+    },
+    {
+      "epoch": 2.5683060109289615,
+      "grad_norm": 0.2986242175102234,
+      "learning_rate": 0.00018534939588627503,
+      "loss": 0.3772,
+      "step": 1645
+    },
+    {
+      "epoch": 2.576112412177986,
+      "grad_norm": 0.29054194688796997,
+      "learning_rate": 0.00018520696996656788,
+      "loss": 0.388,
+      "step": 1650
+    },
+    {
+      "epoch": 2.58391881342701,
+      "grad_norm": 0.2982032001018524,
+      "learning_rate": 0.00018506391036738674,
+      "loss": 0.3705,
+      "step": 1655
+    },
+    {
+      "epoch": 2.5917252146760346,
+      "grad_norm": 0.2838602364063263,
+      "learning_rate": 0.0001849202181526579,
+      "loss": 0.3705,
+      "step": 1660
+    },
+    {
+      "epoch": 2.5995316159250583,
+      "grad_norm": 0.28826209902763367,
+      "learning_rate": 0.00018477589439101232,
+      "loss": 0.3698,
+      "step": 1665
+    },
+    {
+      "epoch": 2.607338017174083,
+      "grad_norm": 0.2916529178619385,
+      "learning_rate": 0.00018463094015577775,
+      "loss": 0.3711,
+      "step": 1670
+    },
+    {
+      "epoch": 2.6151444184231067,
+      "grad_norm": 0.2822958528995514,
+      "learning_rate": 0.00018448535652497073,
+      "loss": 0.3773,
+      "step": 1675
+    },
+    {
+      "epoch": 2.6229508196721314,
+      "grad_norm": 0.2998236119747162,
+      "learning_rate": 0.0001843391445812886,
+      "loss": 0.3769,
+      "step": 1680
+    },
+    {
+      "epoch": 2.630757220921155,
+      "grad_norm": 0.2773909866809845,
+      "learning_rate": 0.00018419230541210139,
+      "loss": 0.3656,
+      "step": 1685
+    },
+    {
+      "epoch": 2.63856362217018,
+      "grad_norm": 0.2902582585811615,
+      "learning_rate": 0.0001840448401094438,
+      "loss": 0.3811,
+      "step": 1690
+    },
+    {
+      "epoch": 2.6463700234192036,
+      "grad_norm": 0.28828758001327515,
+      "learning_rate": 0.00018389674977000708,
+      "loss": 0.3718,
+      "step": 1695
+    },
+    {
+      "epoch": 2.654176424668228,
+      "grad_norm": 0.31182461977005005,
+      "learning_rate": 0.0001837480354951308,
+      "loss": 0.3719,
+      "step": 1700
+    },
+    {
+      "epoch": 2.661982825917252,
+      "grad_norm": 0.2795233130455017,
+      "learning_rate": 0.0001835986983907947,
+      "loss": 0.3635,
+      "step": 1705
+    },
+    {
+      "epoch": 2.669789227166276,
+      "grad_norm": 0.28305554389953613,
+      "learning_rate": 0.00018344873956761045,
+      "loss": 0.3659,
+      "step": 1710
+    },
+    {
+      "epoch": 2.6775956284153004,
+      "grad_norm": 0.27487748861312866,
+      "learning_rate": 0.00018329816014081346,
+      "loss": 0.3738,
+      "step": 1715
+    },
+    {
+      "epoch": 2.6854020296643246,
+      "grad_norm": 0.2843542695045471,
+      "learning_rate": 0.00018314696123025454,
+      "loss": 0.3729,
+      "step": 1720
+    },
+    {
+      "epoch": 2.693208430913349,
+      "grad_norm": 0.28278475999832153,
+      "learning_rate": 0.0001829951439603915,
+      "loss": 0.3675,
+      "step": 1725
+    },
+    {
+      "epoch": 2.701014832162373,
+      "grad_norm": 0.28049561381340027,
+      "learning_rate": 0.00018284270946028092,
+      "loss": 0.3722,
+      "step": 1730
+    },
+    {
+      "epoch": 2.7088212334113972,
+      "grad_norm": 0.2751220464706421,
+      "learning_rate": 0.00018268965886356964,
+      "loss": 0.3708,
+      "step": 1735
+    },
+    {
+      "epoch": 2.7166276346604215,
+      "grad_norm": 0.2924399971961975,
+      "learning_rate": 0.00018253599330848638,
+      "loss": 0.3728,
+      "step": 1740
+    },
+    {
+      "epoch": 2.7244340359094457,
+      "grad_norm": 0.27712133526802063,
+      "learning_rate": 0.00018238171393783335,
+      "loss": 0.3716,
+      "step": 1745
+    },
+    {
+      "epoch": 2.73224043715847,
+      "grad_norm": 0.28451547026634216,
+      "learning_rate": 0.00018222682189897752,
+      "loss": 0.3826,
+      "step": 1750
+    },
+    {
+      "epoch": 2.740046838407494,
+      "grad_norm": 0.275689035654068,
+      "learning_rate": 0.0001820713183438424,
+      "loss": 0.3809,
+      "step": 1755
+    },
+    {
+      "epoch": 2.7478532396565183,
+      "grad_norm": 0.2970561683177948,
+      "learning_rate": 0.0001819152044288992,
+      "loss": 0.3807,
+      "step": 1760
+    },
+    {
+      "epoch": 2.7556596409055425,
+      "grad_norm": 0.27083873748779297,
+      "learning_rate": 0.00018175848131515837,
+      "loss": 0.3777,
+      "step": 1765
+    },
+    {
+      "epoch": 2.7634660421545667,
+      "grad_norm": 0.2864970266819,
+      "learning_rate": 0.000181601150168161,
+      "loss": 0.3697,
+      "step": 1770
+    },
+    {
+      "epoch": 2.771272443403591,
+      "grad_norm": 0.2779759466648102,
+      "learning_rate": 0.00018144321215797,
+      "loss": 0.371,
+      "step": 1775
+    },
+    {
+      "epoch": 2.779078844652615,
+      "grad_norm": 0.2953694462776184,
+      "learning_rate": 0.00018128466845916154,
+      "loss": 0.3782,
+      "step": 1780
+    },
+    {
+      "epoch": 2.7868852459016393,
+      "grad_norm": 0.27182242274284363,
+      "learning_rate": 0.00018112552025081625,
+      "loss": 0.3688,
+      "step": 1785
+    },
+    {
+      "epoch": 2.7946916471506635,
+      "grad_norm": 0.28873521089553833,
+      "learning_rate": 0.00018096576871651042,
+      "loss": 0.3703,
+      "step": 1790
+    },
+    {
+      "epoch": 2.8024980483996877,
+      "grad_norm": 0.2740427851676941,
+      "learning_rate": 0.00018080541504430733,
+      "loss": 0.3751,
+      "step": 1795
+    },
+    {
+      "epoch": 2.810304449648712,
+      "grad_norm": 0.2902701199054718,
+      "learning_rate": 0.00018064446042674828,
+      "loss": 0.3691,
+      "step": 1800
+    },
+    {
+      "epoch": 2.818110850897736,
+      "grad_norm": 0.28125742077827454,
+      "learning_rate": 0.00018048290606084371,
+      "loss": 0.3767,
+      "step": 1805
+    },
+    {
+      "epoch": 2.8259172521467604,
+      "grad_norm": 0.28101617097854614,
+      "learning_rate": 0.00018032075314806448,
+      "loss": 0.3767,
+      "step": 1810
+    },
+    {
+      "epoch": 2.8337236533957846,
+      "grad_norm": 0.31095340847969055,
+      "learning_rate": 0.00018015800289433273,
+      "loss": 0.3841,
+      "step": 1815
+    },
+    {
+      "epoch": 2.841530054644809,
+      "grad_norm": 0.26109251379966736,
+      "learning_rate": 0.00017999465651001296,
+      "loss": 0.3821,
+      "step": 1820
+    },
+    {
+      "epoch": 2.849336455893833,
+      "grad_norm": 0.293575257062912,
+      "learning_rate": 0.00017983071520990315,
+      "loss": 0.3809,
+      "step": 1825
+    },
+    {
+      "epoch": 2.857142857142857,
+      "grad_norm": 0.26493799686431885,
+      "learning_rate": 0.00017966618021322557,
+      "loss": 0.3752,
+      "step": 1830
+    },
+    {
+      "epoch": 2.8649492583918814,
+      "grad_norm": 0.2778892517089844,
+      "learning_rate": 0.00017950105274361776,
+      "loss": 0.369,
+      "step": 1835
+    },
+    {
+      "epoch": 2.8727556596409056,
+      "grad_norm": 0.28492116928100586,
+      "learning_rate": 0.00017933533402912354,
+      "loss": 0.3714,
+      "step": 1840
+    },
+    {
+      "epoch": 2.88056206088993,
+      "grad_norm": 0.281095415353775,
+      "learning_rate": 0.00017916902530218368,
+      "loss": 0.3727,
+      "step": 1845
+    },
+    {
+      "epoch": 2.888368462138954,
+      "grad_norm": 0.277915358543396,
+      "learning_rate": 0.0001790021277996269,
+      "loss": 0.3728,
+      "step": 1850
+    },
+    {
+      "epoch": 2.8961748633879782,
+      "grad_norm": 0.27478280663490295,
+      "learning_rate": 0.00017883464276266064,
+      "loss": 0.3799,
+      "step": 1855
+    },
+    {
+      "epoch": 2.9039812646370025,
+      "grad_norm": 0.28556427359580994,
+      "learning_rate": 0.00017866657143686168,
+      "loss": 0.3762,
+      "step": 1860
+    },
+    {
+      "epoch": 2.9117876658860267,
+      "grad_norm": 0.2784564197063446,
+      "learning_rate": 0.0001784979150721672,
+      "loss": 0.3772,
+      "step": 1865
+    },
+    {
+      "epoch": 2.919594067135051,
+      "grad_norm": 0.3048378825187683,
+      "learning_rate": 0.00017832867492286505,
+      "loss": 0.3691,
+      "step": 1870
+    },
+    {
+      "epoch": 2.927400468384075,
+      "grad_norm": 0.2872634828090668,
+      "learning_rate": 0.0001781588522475848,
+      "loss": 0.3686,
+      "step": 1875
+    },
+    {
+      "epoch": 2.9352068696330993,
+      "grad_norm": 0.28422531485557556,
+      "learning_rate": 0.00017798844830928817,
+      "loss": 0.3814,
+      "step": 1880
+    },
+    {
+      "epoch": 2.9430132708821235,
+      "grad_norm": 0.30014801025390625,
+      "learning_rate": 0.0001778174643752598,
+      "loss": 0.383,
+      "step": 1885
+    },
+    {
+      "epoch": 2.9508196721311473,
+      "grad_norm": 0.3135896921157837,
+      "learning_rate": 0.0001776459017170976,
+      "loss": 0.3665,
+      "step": 1890
+    },
+    {
+      "epoch": 2.958626073380172,
+      "grad_norm": 0.28260594606399536,
+      "learning_rate": 0.00017747376161070344,
+      "loss": 0.3856,
+      "step": 1895
+    },
+    {
+      "epoch": 2.9664324746291957,
+      "grad_norm": 0.26851585507392883,
+      "learning_rate": 0.0001773010453362737,
+      "loss": 0.3838,
+      "step": 1900
+    },
+    {
+      "epoch": 2.9742388758782203,
+      "grad_norm": 0.2853715121746063,
+      "learning_rate": 0.00017712775417828968,
+      "loss": 0.3776,
+      "step": 1905
+    },
+    {
+      "epoch": 2.982045277127244,
+      "grad_norm": 0.28013601899147034,
+      "learning_rate": 0.00017695388942550805,
+      "loss": 0.3741,
+      "step": 1910
+    },
+    {
+      "epoch": 2.9898516783762688,
+      "grad_norm": 0.270375519990921,
+      "learning_rate": 0.00017677945237095123,
+      "loss": 0.3707,
+      "step": 1915
+    },
+    {
+      "epoch": 2.9976580796252925,
+      "grad_norm": 0.29110994935035706,
+      "learning_rate": 0.0001766044443118978,
+      "loss": 0.3774,
+      "step": 1920
+    },
+    {
+      "epoch": 2.999219359875098,
+      "eval_loss": 1.0361436605453491,
+      "eval_runtime": 0.337,
+      "eval_samples_per_second": 14.838,
+      "eval_steps_per_second": 2.968,
+      "step": 1921
+    },
+    {
+      "epoch": 3.0054644808743167,
+      "grad_norm": 0.27365708351135254,
+      "learning_rate": 0.000176428866549873,
+      "loss": 0.357,
+      "step": 1925
+    },
+    {
+      "epoch": 3.013270882123341,
+      "grad_norm": 0.29497936367988586,
+      "learning_rate": 0.00017625272039063883,
+      "loss": 0.3487,
+      "step": 1930
+    },
+    {
+      "epoch": 3.021077283372365,
+      "grad_norm": 0.28680676221847534,
+      "learning_rate": 0.00017607600714418436,
+      "loss": 0.3427,
+      "step": 1935
+    },
+    {
+      "epoch": 3.0288836846213893,
+      "grad_norm": 0.29143229126930237,
+      "learning_rate": 0.0001758987281247162,
+      "loss": 0.3472,
+      "step": 1940
+    },
+    {
+      "epoch": 3.0366900858704136,
+      "grad_norm": 0.28377851843833923,
+      "learning_rate": 0.00017572088465064848,
+      "loss": 0.3468,
+      "step": 1945
+    },
+    {
+      "epoch": 3.0444964871194378,
+      "grad_norm": 0.27560412883758545,
+      "learning_rate": 0.00017554247804459316,
+      "loss": 0.3444,
+      "step": 1950
+    },
+    {
+      "epoch": 3.052302888368462,
+      "grad_norm": 0.2871435880661011,
+      "learning_rate": 0.00017536350963335022,
+      "loss": 0.355,
+      "step": 1955
+    },
+    {
+      "epoch": 3.060109289617486,
+      "grad_norm": 0.29234611988067627,
+      "learning_rate": 0.00017518398074789775,
+      "loss": 0.3405,
+      "step": 1960
+    },
+    {
+      "epoch": 3.0679156908665104,
+      "grad_norm": 0.2977675199508667,
+      "learning_rate": 0.000175003892723382,
+      "loss": 0.3522,
+      "step": 1965
+    },
+    {
+      "epoch": 3.0757220921155346,
+      "grad_norm": 0.2990856468677521,
+      "learning_rate": 0.00017482324689910758,
+      "loss": 0.352,
+      "step": 1970
+    },
+    {
+      "epoch": 3.083528493364559,
+      "grad_norm": 0.28264856338500977,
+      "learning_rate": 0.00017464204461852738,
+      "loss": 0.3497,
+      "step": 1975
+    },
+    {
+      "epoch": 3.091334894613583,
+      "grad_norm": 0.3084184229373932,
+      "learning_rate": 0.00017446028722923265,
+      "loss": 0.3586,
+      "step": 1980
+    },
+    {
+      "epoch": 3.0991412958626072,
+      "grad_norm": 0.29917633533477783,
+      "learning_rate": 0.00017427797608294288,
+      "loss": 0.3542,
+      "step": 1985
+    },
+    {
+      "epoch": 3.1069476971116314,
+      "grad_norm": 0.2877940237522125,
+      "learning_rate": 0.00017409511253549593,
+      "loss": 0.3527,
+      "step": 1990
+    },
+    {
+      "epoch": 3.1147540983606556,
+      "grad_norm": 0.3050166070461273,
+      "learning_rate": 0.00017391169794683776,
+      "loss": 0.3481,
+      "step": 1995
+    },
+    {
+      "epoch": 3.12256049960968,
+      "grad_norm": 0.29156407713890076,
+      "learning_rate": 0.0001737277336810124,
+      "loss": 0.3614,
+      "step": 2000
+    },
+    {
+      "epoch": 3.130366900858704,
+      "grad_norm": 0.3042657971382141,
+      "learning_rate": 0.00017354322110615188,
+      "loss": 0.3608,
+      "step": 2005
+    },
+    {
+      "epoch": 3.1381733021077283,
+      "grad_norm": 0.29401668906211853,
+      "learning_rate": 0.00017335816159446584,
+      "loss": 0.3618,
+      "step": 2010
+    },
+    {
+      "epoch": 3.1459797033567525,
+      "grad_norm": 0.3223430812358856,
+      "learning_rate": 0.00017317255652223157,
+      "loss": 0.3637,
+      "step": 2015
+    },
+    {
+      "epoch": 3.1537861046057767,
+      "grad_norm": 0.3130653500556946,
+      "learning_rate": 0.00017298640726978357,
+      "loss": 0.3561,
+      "step": 2020
+    },
+    {
+      "epoch": 3.161592505854801,
+      "grad_norm": 0.3067520260810852,
+      "learning_rate": 0.00017279971522150348,
+      "loss": 0.3536,
+      "step": 2025
+    },
+    {
+      "epoch": 3.169398907103825,
+      "grad_norm": 0.30372655391693115,
+      "learning_rate": 0.00017261248176580958,
+      "loss": 0.3511,
+      "step": 2030
+    },
+    {
+      "epoch": 3.1772053083528493,
+      "grad_norm": 0.34251293540000916,
+      "learning_rate": 0.00017242470829514672,
+      "loss": 0.3591,
+      "step": 2035
+    },
+    {
+      "epoch": 3.1850117096018735,
+      "grad_norm": 0.34350651502609253,
+      "learning_rate": 0.00017223639620597556,
+      "loss": 0.345,
+      "step": 2040
+    },
+    {
+      "epoch": 3.1928181108508977,
+      "grad_norm": 0.29779571294784546,
+      "learning_rate": 0.0001720475468987627,
+      "loss": 0.3496,
+      "step": 2045
+    },
+    {
+      "epoch": 3.200624512099922,
+      "grad_norm": 0.29767748713493347,
+      "learning_rate": 0.0001718581617779698,
+      "loss": 0.35,
+      "step": 2050
+    },
+    {
+      "epoch": 3.208430913348946,
+      "grad_norm": 0.30058038234710693,
+      "learning_rate": 0.00017166824225204352,
+      "loss": 0.3481,
+      "step": 2055
+    },
+    {
+      "epoch": 3.2162373145979704,
+      "grad_norm": 0.2958409786224365,
+      "learning_rate": 0.00017147778973340466,
+      "loss": 0.3554,
+      "step": 2060
+    },
+    {
+      "epoch": 3.2240437158469946,
+      "grad_norm": 0.29515761137008667,
+      "learning_rate": 0.000171286805638438,
+      "loss": 0.3489,
+      "step": 2065
+    },
+    {
+      "epoch": 3.2318501170960188,
+      "grad_norm": 0.30113205313682556,
+      "learning_rate": 0.00017109529138748157,
+      "loss": 0.3437,
+      "step": 2070
+    },
+    {
+      "epoch": 3.239656518345043,
+      "grad_norm": 0.2853497862815857,
+      "learning_rate": 0.0001709032484048162,
+      "loss": 0.3515,
+      "step": 2075
+    },
+    {
+      "epoch": 3.247462919594067,
+      "grad_norm": 0.3135252892971039,
+      "learning_rate": 0.00017071067811865476,
+      "loss": 0.3502,
+      "step": 2080
+    },
+    {
+      "epoch": 3.2552693208430914,
+      "grad_norm": 0.30913615226745605,
+      "learning_rate": 0.00017051758196113173,
+      "loss": 0.3468,
+      "step": 2085
+    },
+    {
+      "epoch": 3.2630757220921156,
+      "grad_norm": 0.2965530753135681,
+      "learning_rate": 0.00017032396136829246,
+      "loss": 0.3565,
+      "step": 2090
+    },
+    {
+      "epoch": 3.27088212334114,
+      "grad_norm": 0.2974913418292999,
+      "learning_rate": 0.00017012981778008243,
+      "loss": 0.3442,
+      "step": 2095
+    },
+    {
+      "epoch": 3.278688524590164,
+      "grad_norm": 0.2917117476463318,
+      "learning_rate": 0.00016993515264033672,
+      "loss": 0.3479,
+      "step": 2100
+    },
+    {
+      "epoch": 3.2864949258391882,
+      "grad_norm": 0.29723039269447327,
+      "learning_rate": 0.00016973996739676905,
+      "loss": 0.3523,
+      "step": 2105
+    },
+    {
+      "epoch": 3.2943013270882124,
+      "grad_norm": 0.3127197027206421,
+      "learning_rate": 0.00016954426350096116,
+      "loss": 0.3574,
+      "step": 2110
+    },
+    {
+      "epoch": 3.3021077283372366,
+      "grad_norm": 0.28733986616134644,
+      "learning_rate": 0.00016934804240835205,
+      "loss": 0.3495,
+      "step": 2115
+    },
+    {
+      "epoch": 3.309914129586261,
+      "grad_norm": 0.30774375796318054,
+      "learning_rate": 0.00016915130557822695,
+      "loss": 0.3515,
+      "step": 2120
+    },
+    {
+      "epoch": 3.317720530835285,
+      "grad_norm": 0.3072982132434845,
+      "learning_rate": 0.0001689540544737067,
+      "loss": 0.3473,
+      "step": 2125
+    },
+    {
+      "epoch": 3.3255269320843093,
+      "grad_norm": 0.30740848183631897,
+      "learning_rate": 0.00016875629056173675,
+      "loss": 0.3631,
+      "step": 2130
+    },
+    {
+      "epoch": 3.3333333333333335,
+      "grad_norm": 0.2990294396877289,
+      "learning_rate": 0.00016855801531307624,
+      "loss": 0.3508,
+      "step": 2135
+    },
+    {
+      "epoch": 3.3411397345823577,
+      "grad_norm": 0.28540557622909546,
+      "learning_rate": 0.00016835923020228712,
+      "loss": 0.353,
+      "step": 2140
+    },
+    {
+      "epoch": 3.348946135831382,
+      "grad_norm": 0.29948052763938904,
+      "learning_rate": 0.0001681599367077232,
+      "loss": 0.3627,
+      "step": 2145
+    },
+    {
+      "epoch": 3.356752537080406,
+      "grad_norm": 0.30102795362472534,
+      "learning_rate": 0.00016796013631151897,
+      "loss": 0.3526,
+      "step": 2150
+    },
+    {
+      "epoch": 3.3645589383294303,
+      "grad_norm": 0.2947191298007965,
+      "learning_rate": 0.00016775983049957887,
+      "loss": 0.3645,
+      "step": 2155
+    },
+    {
+      "epoch": 3.3723653395784545,
+      "grad_norm": 0.3223245143890381,
+      "learning_rate": 0.00016755902076156604,
+      "loss": 0.3623,
+      "step": 2160
+    },
+    {
+      "epoch": 3.3801717408274783,
+      "grad_norm": 0.28947457671165466,
+      "learning_rate": 0.00016735770859089123,
+      "loss": 0.3638,
+      "step": 2165
+    },
+    {
+      "epoch": 3.387978142076503,
+      "grad_norm": 0.28693681955337524,
+      "learning_rate": 0.00016715589548470185,
+      "loss": 0.3614,
+      "step": 2170
+    },
+    {
+      "epoch": 3.3957845433255267,
+      "grad_norm": 0.29449644684791565,
+      "learning_rate": 0.00016695358294387065,
+      "loss": 0.3529,
+      "step": 2175
+    },
+    {
+      "epoch": 3.4035909445745514,
+      "grad_norm": 0.289899080991745,
+      "learning_rate": 0.00016675077247298474,
+      "loss": 0.3556,
+      "step": 2180
+    },
+    {
+      "epoch": 3.411397345823575,
+      "grad_norm": 0.2969209849834442,
+      "learning_rate": 0.00016654746558033423,
+      "loss": 0.3598,
+      "step": 2185
+    },
+    {
+      "epoch": 3.4192037470725998,
+      "grad_norm": 0.3075496554374695,
+      "learning_rate": 0.00016634366377790114,
+      "loss": 0.3509,
+      "step": 2190
+    },
+    {
+      "epoch": 3.4270101483216235,
+      "grad_norm": 0.28629180788993835,
+      "learning_rate": 0.00016613936858134802,
+      "loss": 0.3481,
+      "step": 2195
+    },
+    {
+      "epoch": 3.4348165495706477,
+      "grad_norm": 0.3028337061405182,
+      "learning_rate": 0.00016593458151000688,
+      "loss": 0.358,
+      "step": 2200
+    },
+    {
+      "epoch": 3.442622950819672,
+      "grad_norm": 0.2874985933303833,
+      "learning_rate": 0.00016572930408686776,
+      "loss": 0.3593,
+      "step": 2205
+    },
+    {
+      "epoch": 3.450429352068696,
+      "grad_norm": 0.29345887899398804,
+      "learning_rate": 0.00016552353783856732,
+      "loss": 0.3485,
+      "step": 2210
+    },
+    {
+      "epoch": 3.4582357533177204,
+      "grad_norm": 0.28079068660736084,
+      "learning_rate": 0.00016531728429537766,
+      "loss": 0.3606,
+      "step": 2215
+    },
+    {
+      "epoch": 3.4660421545667446,
+      "grad_norm": 0.2924216389656067,
+      "learning_rate": 0.0001651105449911949,
+      "loss": 0.351,
+      "step": 2220
+    },
+    {
+      "epoch": 3.473848555815769,
+      "grad_norm": 0.29718244075775146,
+      "learning_rate": 0.0001649033214635277,
+      "loss": 0.3521,
+      "step": 2225
+    },
+    {
+      "epoch": 3.481654957064793,
+      "grad_norm": 0.2965914309024811,
+      "learning_rate": 0.00016469561525348577,
+      "loss": 0.3564,
+      "step": 2230
+    },
+    {
+      "epoch": 3.489461358313817,
+      "grad_norm": 0.2874649167060852,
+      "learning_rate": 0.0001644874279057686,
+      "loss": 0.3554,
+      "step": 2235
+    },
+    {
+      "epoch": 3.4972677595628414,
+      "grad_norm": 0.2855139970779419,
+      "learning_rate": 0.00016427876096865394,
+      "loss": 0.3619,
+      "step": 2240
+    },
+    {
+      "epoch": 3.5050741608118656,
+      "grad_norm": 0.2949666976928711,
+      "learning_rate": 0.0001640696159939861,
+      "loss": 0.3558,
+      "step": 2245
+    },
+    {
+      "epoch": 3.51288056206089,
+      "grad_norm": 0.29184606671333313,
+      "learning_rate": 0.00016385999453716454,
+      "loss": 0.3625,
+      "step": 2250
+    },
+    {
+      "epoch": 3.520686963309914,
+      "grad_norm": 0.29414233565330505,
+      "learning_rate": 0.00016364989815713233,
+      "loss": 0.3559,
+      "step": 2255
+    },
+    {
+      "epoch": 3.5284933645589383,
+      "grad_norm": 0.2970221936702728,
+      "learning_rate": 0.00016343932841636456,
+      "loss": 0.3488,
+      "step": 2260
+    },
+    {
+      "epoch": 3.5362997658079625,
+      "grad_norm": 0.305912047624588,
+      "learning_rate": 0.00016322828688085657,
+      "loss": 0.3553,
+      "step": 2265
+    },
+    {
+      "epoch": 3.5441061670569867,
+      "grad_norm": 0.3016732931137085,
+      "learning_rate": 0.00016301677512011248,
+      "loss": 0.359,
+      "step": 2270
+    },
+    {
+      "epoch": 3.551912568306011,
+      "grad_norm": 0.2802044749259949,
+      "learning_rate": 0.00016280479470713344,
+      "loss": 0.347,
+      "step": 2275
+    },
+    {
+      "epoch": 3.559718969555035,
+      "grad_norm": 0.3013835549354553,
+      "learning_rate": 0.00016259234721840591,
+      "loss": 0.3664,
+      "step": 2280
+    },
+    {
+      "epoch": 3.5675253708040593,
+      "grad_norm": 0.2899465262889862,
+      "learning_rate": 0.00016237943423389001,
+      "loss": 0.3638,
+      "step": 2285
+    },
+    {
+      "epoch": 3.5753317720530835,
+      "grad_norm": 0.2953374981880188,
+      "learning_rate": 0.00016216605733700775,
+      "loss": 0.3621,
+      "step": 2290
+    },
+    {
+      "epoch": 3.5831381733021077,
+      "grad_norm": 0.28580641746520996,
+      "learning_rate": 0.00016195221811463113,
+      "loss": 0.3558,
+      "step": 2295
+    },
+    {
+      "epoch": 3.590944574551132,
+      "grad_norm": 0.2812565565109253,
+      "learning_rate": 0.00016173791815707051,
+      "loss": 0.357,
+      "step": 2300
+    },
+    {
+      "epoch": 3.598750975800156,
+      "grad_norm": 0.3074684739112854,
+      "learning_rate": 0.00016152315905806268,
+      "loss": 0.3593,
+      "step": 2305
+    },
+    {
+      "epoch": 3.6065573770491803,
+      "grad_norm": 0.29887428879737854,
+      "learning_rate": 0.0001613079424147591,
+      "loss": 0.3607,
+      "step": 2310
+    },
+    {
+      "epoch": 3.6143637782982045,
+      "grad_norm": 0.2820027470588684,
+      "learning_rate": 0.00016109226982771393,
+      "loss": 0.3535,
+      "step": 2315
+    },
+    {
+      "epoch": 3.6221701795472288,
+      "grad_norm": 0.28952109813690186,
+      "learning_rate": 0.00016087614290087208,
+      "loss": 0.3563,
+      "step": 2320
+    },
+    {
+      "epoch": 3.629976580796253,
+      "grad_norm": 0.29456931352615356,
+      "learning_rate": 0.00016065956324155746,
+      "loss": 0.3601,
+      "step": 2325
+    },
+    {
+      "epoch": 3.637782982045277,
+      "grad_norm": 0.29095590114593506,
+      "learning_rate": 0.0001604425324604609,
+      "loss": 0.3633,
+      "step": 2330
+    },
+    {
+      "epoch": 3.6455893832943014,
+      "grad_norm": 0.27606719732284546,
+      "learning_rate": 0.00016022505217162822,
+      "loss": 0.3621,
+      "step": 2335
+    },
+    {
+      "epoch": 3.6533957845433256,
+      "grad_norm": 0.2846406400203705,
+      "learning_rate": 0.0001600071239924481,
+      "loss": 0.349,
+      "step": 2340
+    },
+    {
+      "epoch": 3.66120218579235,
+      "grad_norm": 0.28319764137268066,
+      "learning_rate": 0.00015978874954364033,
+      "loss": 0.3514,
+      "step": 2345
+    },
+    {
+      "epoch": 3.669008587041374,
+      "grad_norm": 0.3063547909259796,
+      "learning_rate": 0.00015956993044924334,
+      "loss": 0.3504,
+      "step": 2350
+    },
+    {
+      "epoch": 3.676814988290398,
+      "grad_norm": 0.27655765414237976,
+      "learning_rate": 0.0001593506683366026,
+      "loss": 0.3466,
+      "step": 2355
+    },
+    {
+      "epoch": 3.6846213895394224,
+      "grad_norm": 0.2881523072719574,
+      "learning_rate": 0.00015913096483635824,
+      "loss": 0.3466,
+      "step": 2360
+    },
+    {
+      "epoch": 3.6924277907884466,
+      "grad_norm": 0.2950555682182312,
+      "learning_rate": 0.00015891082158243282,
+      "loss": 0.3512,
+      "step": 2365
+    },
+    {
+      "epoch": 3.700234192037471,
+      "grad_norm": 0.2711721360683441,
+      "learning_rate": 0.00015869024021201948,
+      "loss": 0.3545,
+      "step": 2370
+    },
+    {
+      "epoch": 3.708040593286495,
+      "grad_norm": 0.31127429008483887,
+      "learning_rate": 0.00015846922236556946,
+      "loss": 0.3616,
+      "step": 2375
+    },
+    {
+      "epoch": 3.7158469945355193,
+      "grad_norm": 0.28716492652893066,
+      "learning_rate": 0.00015824776968678024,
+      "loss": 0.3548,
+      "step": 2380
+    },
+    {
+      "epoch": 3.7236533957845435,
+      "grad_norm": 0.28540754318237305,
+      "learning_rate": 0.00015802588382258292,
+      "loss": 0.3571,
+      "step": 2385
+    },
+    {
+      "epoch": 3.7314597970335677,
+      "grad_norm": 0.30483660101890564,
+      "learning_rate": 0.00015780356642313036,
+      "loss": 0.3475,
+      "step": 2390
+    },
+    {
+      "epoch": 3.739266198282592,
+      "grad_norm": 0.28474995493888855,
+      "learning_rate": 0.00015758081914178456,
+      "loss": 0.3475,
+      "step": 2395
+    },
+    {
+      "epoch": 3.747072599531616,
+      "grad_norm": 0.28476497530937195,
+      "learning_rate": 0.0001573576436351046,
+      "loss": 0.3525,
+      "step": 2400
+    },
+    {
+      "epoch": 3.7548790007806403,
+      "grad_norm": 0.29259222745895386,
+      "learning_rate": 0.00015713404156283435,
+      "loss": 0.3683,
+      "step": 2405
+    },
+    {
+      "epoch": 3.762685402029664,
+      "grad_norm": 0.2946162223815918,
+      "learning_rate": 0.00015691001458788983,
+      "loss": 0.3587,
+      "step": 2410
+    },
+    {
+      "epoch": 3.7704918032786887,
+      "grad_norm": 0.2999446988105774,
+      "learning_rate": 0.0001566855643763472,
+      "loss": 0.355,
+      "step": 2415
+    },
+    {
+      "epoch": 3.7782982045277125,
+      "grad_norm": 0.2700735032558441,
+      "learning_rate": 0.00015646069259743007,
+      "loss": 0.3531,
+      "step": 2420
+    },
+    {
+      "epoch": 3.786104605776737,
+      "grad_norm": 0.28989067673683167,
+      "learning_rate": 0.00015623540092349732,
+      "loss": 0.3568,
+      "step": 2425
+    },
+    {
+      "epoch": 3.793911007025761,
+      "grad_norm": 0.2979315221309662,
+      "learning_rate": 0.00015600969103003056,
+      "loss": 0.3514,
+      "step": 2430
+    },
+    {
+      "epoch": 3.8017174082747855,
+      "grad_norm": 0.27748894691467285,
+      "learning_rate": 0.00015578356459562163,
+      "loss": 0.3474,
+      "step": 2435
+    },
+    {
+      "epoch": 3.8095238095238093,
+      "grad_norm": 0.28995972871780396,
+      "learning_rate": 0.00015555702330196023,
+      "loss": 0.352,
+      "step": 2440
+    },
+    {
+      "epoch": 3.817330210772834,
+      "grad_norm": 0.2819332480430603,
+      "learning_rate": 0.00015533006883382123,
+      "loss": 0.3482,
+      "step": 2445
+    },
+    {
+      "epoch": 3.8251366120218577,
+      "grad_norm": 0.2890748083591461,
+      "learning_rate": 0.0001551027028790524,
+      "loss": 0.3608,
+      "step": 2450
+    },
+    {
+      "epoch": 3.8329430132708824,
+      "grad_norm": 0.295684278011322,
+      "learning_rate": 0.0001548749271285616,
+      "loss": 0.3541,
+      "step": 2455
+    },
+    {
+      "epoch": 3.840749414519906,
+      "grad_norm": 0.2782337963581085,
+      "learning_rate": 0.00015464674327630434,
+      "loss": 0.3467,
+      "step": 2460
+    },
+    {
+      "epoch": 3.848555815768931,
+      "grad_norm": 0.2889052629470825,
+      "learning_rate": 0.00015441815301927122,
+      "loss": 0.3597,
+      "step": 2465
+    },
+    {
+      "epoch": 3.8563622170179546,
+      "grad_norm": 0.27629801630973816,
+      "learning_rate": 0.00015418915805747517,
+      "loss": 0.3543,
+      "step": 2470
+    },
+    {
+      "epoch": 3.8641686182669788,
+      "grad_norm": 0.2963428497314453,
+      "learning_rate": 0.00015395976009393894,
+      "loss": 0.3637,
+      "step": 2475
+    },
+    {
+      "epoch": 3.871975019516003,
+      "grad_norm": 0.29081323742866516,
+      "learning_rate": 0.0001537299608346824,
+      "loss": 0.3563,
+      "step": 2480
+    },
+    {
+      "epoch": 3.879781420765027,
+      "grad_norm": 0.29357287287712097,
+      "learning_rate": 0.00015349976198870973,
+      "loss": 0.3633,
+      "step": 2485
+    },
+    {
+      "epoch": 3.8875878220140514,
+      "grad_norm": 0.29132893681526184,
+      "learning_rate": 0.00015326916526799692,
+      "loss": 0.3563,
+      "step": 2490
+    },
+    {
+      "epoch": 3.8953942232630756,
+      "grad_norm": 0.29773959517478943,
+      "learning_rate": 0.00015303817238747892,
+      "loss": 0.3511,
+      "step": 2495
+    },
+    {
+      "epoch": 3.9032006245121,
+      "grad_norm": 0.2842431366443634,
+      "learning_rate": 0.0001528067850650368,
+      "loss": 0.3533,
+      "step": 2500
+    },
+    {
+      "epoch": 3.911007025761124,
+      "grad_norm": 0.277873694896698,
+      "learning_rate": 0.00015257500502148522,
+      "loss": 0.3549,
+      "step": 2505
+    },
+    {
+      "epoch": 3.9188134270101482,
+      "grad_norm": 0.2919045388698578,
+      "learning_rate": 0.0001523428339805594,
+      "loss": 0.3521,
+      "step": 2510
+    },
+    {
+      "epoch": 3.9266198282591724,
+      "grad_norm": 0.289386123418808,
+      "learning_rate": 0.00015211027366890235,
+      "loss": 0.3603,
+      "step": 2515
+    },
+    {
+      "epoch": 3.9344262295081966,
+      "grad_norm": 0.27099329233169556,
+      "learning_rate": 0.00015187732581605217,
+      "loss": 0.3553,
+      "step": 2520
+    },
+    {
+      "epoch": 3.942232630757221,
+      "grad_norm": 0.2989470660686493,
+      "learning_rate": 0.00015164399215442898,
+      "loss": 0.3562,
+      "step": 2525
+    },
+    {
+      "epoch": 3.950039032006245,
+      "grad_norm": 0.2968309223651886,
+      "learning_rate": 0.00015141027441932216,
+      "loss": 0.3581,
+      "step": 2530
+    },
+    {
+      "epoch": 3.9578454332552693,
+      "grad_norm": 0.29724830389022827,
+      "learning_rate": 0.00015117617434887753,
+      "loss": 0.3514,
+      "step": 2535
+    },
+    {
+      "epoch": 3.9656518345042935,
+      "grad_norm": 0.30498626828193665,
+      "learning_rate": 0.0001509416936840842,
+      "loss": 0.3691,
+      "step": 2540
+    },
+    {
+      "epoch": 3.9734582357533177,
+      "grad_norm": 0.28580141067504883,
+      "learning_rate": 0.00015070683416876172,
+      "loss": 0.3509,
+      "step": 2545
+    },
+    {
+      "epoch": 3.981264637002342,
+      "grad_norm": 0.28979378938674927,
+      "learning_rate": 0.0001504715975495472,
+      "loss": 0.3598,
+      "step": 2550
+    },
+    {
+      "epoch": 3.989071038251366,
+      "grad_norm": 0.30698639154434204,
+      "learning_rate": 0.0001502359855758823,
+      "loss": 0.357,
+      "step": 2555
+    },
+    {
+      "epoch": 3.9968774395003903,
+      "grad_norm": 0.2776874303817749,
+      "learning_rate": 0.00015000000000000001,
+      "loss": 0.355,
+      "step": 2560
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 1.041857361793518,
+      "eval_runtime": 0.3229,
+      "eval_samples_per_second": 15.485,
+      "eval_steps_per_second": 3.097,
+      "step": 2562
+    },
+    {
+      "epoch": 4.004683840749414,
+      "grad_norm": 0.2781620919704437,
+      "learning_rate": 0.00014976364257691192,
+      "loss": 0.3413,
+      "step": 2565
+    },
+    {
+      "epoch": 4.012490241998439,
+      "grad_norm": 0.3088448643684387,
+      "learning_rate": 0.00014952691506439495,
+      "loss": 0.3303,
+      "step": 2570
+    },
+    {
+      "epoch": 4.0202966432474625,
+      "grad_norm": 0.28063511848449707,
+      "learning_rate": 0.00014928981922297842,
+      "loss": 0.3268,
+      "step": 2575
+    },
+    {
+      "epoch": 4.028103044496487,
+      "grad_norm": 0.32527846097946167,
+      "learning_rate": 0.0001490523568159308,
+      "loss": 0.3184,
+      "step": 2580
+    },
+    {
+      "epoch": 4.035909445745511,
+      "grad_norm": 0.29716962575912476,
+      "learning_rate": 0.00014881452960924677,
+      "loss": 0.3287,
+      "step": 2585
+    },
+    {
+      "epoch": 4.043715846994536,
+      "grad_norm": 0.30141621828079224,
+      "learning_rate": 0.00014857633937163402,
+      "loss": 0.3331,
+      "step": 2590
+    },
+    {
+      "epoch": 4.051522248243559,
+      "grad_norm": 0.2995307743549347,
+      "learning_rate": 0.00014833778787449998,
+      "loss": 0.3313,
+      "step": 2595
+    },
+    {
+      "epoch": 4.059328649492584,
+      "grad_norm": 0.30431655049324036,
+      "learning_rate": 0.00014809887689193877,
+      "loss": 0.334,
+      "step": 2600
+    },
+    {
+      "epoch": 4.067135050741608,
+      "grad_norm": 0.3243234157562256,
+      "learning_rate": 0.0001478596082007181,
+      "loss": 0.3339,
+      "step": 2605
+    },
+    {
+      "epoch": 4.074941451990632,
+      "grad_norm": 0.3121318817138672,
+      "learning_rate": 0.0001476199835802658,
+      "loss": 0.3306,
+      "step": 2610
+    },
+    {
+      "epoch": 4.082747853239656,
+      "grad_norm": 0.31301409006118774,
+      "learning_rate": 0.0001473800048126568,
+      "loss": 0.3336,
+      "step": 2615
+    },
+    {
+      "epoch": 4.090554254488681,
+      "grad_norm": 0.29308003187179565,
+      "learning_rate": 0.0001471396736825998,
+      "loss": 0.3239,
+      "step": 2620
+    },
+    {
+      "epoch": 4.098360655737705,
+      "grad_norm": 0.30689167976379395,
+      "learning_rate": 0.0001468989919774239,
+      "loss": 0.3342,
+      "step": 2625
+    },
+    {
+      "epoch": 4.106167056986729,
+      "grad_norm": 0.32097750902175903,
+      "learning_rate": 0.0001466579614870656,
+      "loss": 0.3386,
+      "step": 2630
+    },
+    {
+      "epoch": 4.113973458235753,
+      "grad_norm": 0.3277100622653961,
+      "learning_rate": 0.00014641658400405515,
+      "loss": 0.3434,
+      "step": 2635
+    },
+    {
+      "epoch": 4.121779859484778,
+      "grad_norm": 0.30586951971054077,
+      "learning_rate": 0.00014617486132350343,
+      "loss": 0.3328,
+      "step": 2640
+    },
+    {
+      "epoch": 4.129586260733801,
+      "grad_norm": 0.30447080731391907,
+      "learning_rate": 0.0001459327952430884,
+      "loss": 0.3343,
+      "step": 2645
+    },
+    {
+      "epoch": 4.137392661982826,
+      "grad_norm": 0.31396037340164185,
+      "learning_rate": 0.00014569038756304207,
+      "loss": 0.3331,
+      "step": 2650
+    },
+    {
+      "epoch": 4.14519906323185,
+      "grad_norm": 0.30158740282058716,
+      "learning_rate": 0.0001454476400861368,
+      "loss": 0.33,
+      "step": 2655
+    },
+    {
+      "epoch": 4.1530054644808745,
+      "grad_norm": 0.30618518590927124,
+      "learning_rate": 0.000145204554617672,
+      "loss": 0.3297,
+      "step": 2660
+    },
+    {
+      "epoch": 4.160811865729898,
+      "grad_norm": 0.3204987049102783,
+      "learning_rate": 0.00014496113296546067,
+      "loss": 0.3356,
+      "step": 2665
+    },
+    {
+      "epoch": 4.168618266978923,
+      "grad_norm": 0.31640496850013733,
+      "learning_rate": 0.0001447173769398161,
+      "loss": 0.3397,
+      "step": 2670
+    },
+    {
+      "epoch": 4.176424668227947,
+      "grad_norm": 0.3094062805175781,
+      "learning_rate": 0.0001444732883535382,
+      "loss": 0.3359,
+      "step": 2675
+    },
+    {
+      "epoch": 4.184231069476971,
+      "grad_norm": 0.3020561933517456,
+      "learning_rate": 0.00014422886902190014,
+      "loss": 0.3313,
+      "step": 2680
+    },
+    {
+      "epoch": 4.192037470725995,
+      "grad_norm": 0.3145982027053833,
+      "learning_rate": 0.00014398412076263486,
+      "loss": 0.3441,
+      "step": 2685
+    },
+    {
+      "epoch": 4.19984387197502,
+      "grad_norm": 0.3120006322860718,
+      "learning_rate": 0.00014373904539592144,
+      "loss": 0.3334,
+      "step": 2690
+    },
+    {
+      "epoch": 4.2076502732240435,
+      "grad_norm": 0.319619357585907,
+      "learning_rate": 0.00014349364474437173,
+      "loss": 0.3287,
+      "step": 2695
+    },
+    {
+      "epoch": 4.215456674473068,
+      "grad_norm": 0.3287660479545593,
+      "learning_rate": 0.00014324792063301662,
+      "loss": 0.3296,
+      "step": 2700
+    },
+    {
+      "epoch": 4.223263075722092,
+      "grad_norm": 0.3166675567626953,
+      "learning_rate": 0.00014300187488929265,
+      "loss": 0.3433,
+      "step": 2705
+    },
+    {
+      "epoch": 4.231069476971117,
+      "grad_norm": 0.3454567492008209,
+      "learning_rate": 0.00014275550934302823,
+      "loss": 0.3384,
+      "step": 2710
+    },
+    {
+      "epoch": 4.23887587822014,
+      "grad_norm": 0.30334779620170593,
+      "learning_rate": 0.00014250882582643012,
+      "loss": 0.334,
+      "step": 2715
+    },
+    {
+      "epoch": 4.246682279469165,
+      "grad_norm": 0.31963109970092773,
+      "learning_rate": 0.00014226182617406996,
+      "loss": 0.3428,
+      "step": 2720
+    },
+    {
+      "epoch": 4.254488680718189,
+      "grad_norm": 0.30890125036239624,
+      "learning_rate": 0.00014201451222287025,
+      "loss": 0.3326,
+      "step": 2725
+    },
+    {
+      "epoch": 4.262295081967213,
+      "grad_norm": 0.32028844952583313,
+      "learning_rate": 0.0001417668858120911,
+      "loss": 0.333,
+      "step": 2730
+    },
+    {
+      "epoch": 4.270101483216237,
+      "grad_norm": 0.3037426471710205,
+      "learning_rate": 0.00014151894878331622,
+      "loss": 0.3302,
+      "step": 2735
+    },
+    {
+      "epoch": 4.277907884465262,
+      "grad_norm": 0.305708110332489,
+      "learning_rate": 0.00014127070298043947,
+      "loss": 0.3387,
+      "step": 2740
+    },
+    {
+      "epoch": 4.285714285714286,
+      "grad_norm": 0.3226141333580017,
+      "learning_rate": 0.00014102215024965105,
+      "loss": 0.3352,
+      "step": 2745
+    },
+    {
+      "epoch": 4.29352068696331,
+      "grad_norm": 0.3047237992286682,
+      "learning_rate": 0.00014077329243942369,
+      "loss": 0.3377,
+      "step": 2750
+    },
+    {
+      "epoch": 4.301327088212334,
+      "grad_norm": 0.3010615408420563,
+      "learning_rate": 0.000140524131400499,
+      "loss": 0.3325,
+      "step": 2755
+    },
+    {
+      "epoch": 4.309133489461359,
+      "grad_norm": 0.3127407431602478,
+      "learning_rate": 0.00014027466898587374,
+      "loss": 0.3357,
+      "step": 2760
+    },
+    {
+      "epoch": 4.316939890710382,
+      "grad_norm": 0.31216832995414734,
+      "learning_rate": 0.00014002490705078592,
+      "loss": 0.346,
+      "step": 2765
+    },
+    {
+      "epoch": 4.324746291959407,
+      "grad_norm": 0.31948304176330566,
+      "learning_rate": 0.00013977484745270112,
+      "loss": 0.3292,
+      "step": 2770
+    },
+    {
+      "epoch": 4.332552693208431,
+      "grad_norm": 0.3235209584236145,
+      "learning_rate": 0.00013952449205129855,
+      "loss": 0.3415,
+      "step": 2775
+    },
+    {
+      "epoch": 4.3403590944574555,
+      "grad_norm": 0.3177550137042999,
+      "learning_rate": 0.0001392738427084574,
+      "loss": 0.3341,
+      "step": 2780
+    },
+    {
+      "epoch": 4.348165495706479,
+      "grad_norm": 0.31263765692710876,
+      "learning_rate": 0.00013902290128824283,
+      "loss": 0.3348,
+      "step": 2785
+    },
+    {
+      "epoch": 4.355971896955504,
+      "grad_norm": 0.31187278032302856,
+      "learning_rate": 0.00013877166965689205,
+      "loss": 0.3405,
+      "step": 2790
+    },
+    {
+      "epoch": 4.363778298204528,
+      "grad_norm": 0.3060557246208191,
+      "learning_rate": 0.00013852014968280082,
+      "loss": 0.3328,
+      "step": 2795
+    },
+    {
+      "epoch": 4.371584699453552,
+      "grad_norm": 0.30810457468032837,
+      "learning_rate": 0.000138268343236509,
+      "loss": 0.3303,
+      "step": 2800
+    },
+    {
+      "epoch": 4.379391100702576,
+      "grad_norm": 0.309929758310318,
+      "learning_rate": 0.0001380162521906871,
+      "loss": 0.3371,
+      "step": 2805
+    },
+    {
+      "epoch": 4.387197501951601,
+      "grad_norm": 0.31695547699928284,
+      "learning_rate": 0.00013776387842012218,
+      "loss": 0.3393,
+      "step": 2810
+    },
+    {
+      "epoch": 4.3950039032006245,
+      "grad_norm": 0.31112900376319885,
+      "learning_rate": 0.0001375112238017038,
+      "loss": 0.338,
+      "step": 2815
+    },
+    {
+      "epoch": 4.402810304449648,
+      "grad_norm": 0.29906710982322693,
+      "learning_rate": 0.0001372582902144103,
+      "loss": 0.3318,
+      "step": 2820
+    },
+    {
+      "epoch": 4.410616705698673,
+      "grad_norm": 0.30970847606658936,
+      "learning_rate": 0.00013700507953929463,
+      "loss": 0.3387,
+      "step": 2825
+    },
+    {
+      "epoch": 4.418423106947698,
+      "grad_norm": 0.3163338005542755,
+      "learning_rate": 0.00013675159365947036,
+      "loss": 0.3371,
+      "step": 2830
+    },
+    {
+      "epoch": 4.426229508196721,
+      "grad_norm": 0.31127607822418213,
+      "learning_rate": 0.0001364978344600979,
+      "loss": 0.3372,
+      "step": 2835
+    },
+    {
+      "epoch": 4.434035909445745,
+      "grad_norm": 0.2967492640018463,
+      "learning_rate": 0.00013624380382837016,
+      "loss": 0.339,
+      "step": 2840
+    },
+    {
+      "epoch": 4.44184231069477,
+      "grad_norm": 0.31639572978019714,
+      "learning_rate": 0.00013598950365349883,
+      "loss": 0.3411,
+      "step": 2845
+    },
+    {
+      "epoch": 4.4496487119437935,
+      "grad_norm": 0.3143470883369446,
+      "learning_rate": 0.00013573493582670003,
+      "loss": 0.3386,
+      "step": 2850
+    },
+    {
+      "epoch": 4.457455113192818,
+      "grad_norm": 0.3157510757446289,
+      "learning_rate": 0.0001354801022411805,
+      "loss": 0.3339,
+      "step": 2855
+    },
+    {
+      "epoch": 4.465261514441842,
+      "grad_norm": 0.3069648742675781,
+      "learning_rate": 0.00013522500479212337,
+      "loss": 0.3385,
+      "step": 2860
+    },
+    {
+      "epoch": 4.473067915690867,
+      "grad_norm": 0.31706541776657104,
+      "learning_rate": 0.0001349696453766741,
+      "loss": 0.3353,
+      "step": 2865
+    },
+    {
+      "epoch": 4.48087431693989,
+      "grad_norm": 0.3200201988220215,
+      "learning_rate": 0.00013471402589392638,
+      "loss": 0.3419,
+      "step": 2870
+    },
+    {
+      "epoch": 4.488680718188915,
+      "grad_norm": 0.32029232382774353,
+      "learning_rate": 0.00013445814824490805,
+      "loss": 0.3417,
+      "step": 2875
+    },
+    {
+      "epoch": 4.496487119437939,
+      "grad_norm": 0.3023779094219208,
+      "learning_rate": 0.00013420201433256689,
+      "loss": 0.3388,
+      "step": 2880
+    },
+    {
+      "epoch": 4.504293520686963,
+      "grad_norm": 0.3167456090450287,
+      "learning_rate": 0.00013394562606175648,
+      "loss": 0.335,
+      "step": 2885
+    },
+    {
+      "epoch": 4.512099921935987,
+      "grad_norm": 0.31978872418403625,
+      "learning_rate": 0.000133688985339222,
+      "loss": 0.3425,
+      "step": 2890
+    },
+    {
+      "epoch": 4.519906323185012,
+      "grad_norm": 0.32426413893699646,
+      "learning_rate": 0.00013343209407358627,
+      "loss": 0.3325,
+      "step": 2895
+    },
+    {
+      "epoch": 4.527712724434036,
+      "grad_norm": 0.28932276368141174,
+      "learning_rate": 0.00013317495417533524,
+      "loss": 0.3365,
+      "step": 2900
+    },
+    {
+      "epoch": 4.53551912568306,
+      "grad_norm": 0.31238290667533875,
+      "learning_rate": 0.00013291756755680388,
+      "loss": 0.3353,
+      "step": 2905
+    },
+    {
+      "epoch": 4.543325526932084,
+      "grad_norm": 0.3126748204231262,
+      "learning_rate": 0.00013265993613216224,
+      "loss": 0.3363,
+      "step": 2910
+    },
+    {
+      "epoch": 4.551131928181109,
+      "grad_norm": 0.331447571516037,
+      "learning_rate": 0.0001324020618174007,
+      "loss": 0.3434,
+      "step": 2915
+    },
+    {
+      "epoch": 4.558938329430132,
+      "grad_norm": 0.31981754302978516,
+      "learning_rate": 0.00013214394653031616,
+      "loss": 0.3321,
+      "step": 2920
+    },
+    {
+      "epoch": 4.566744730679157,
+      "grad_norm": 0.30969661474227905,
+      "learning_rate": 0.0001318855921904976,
+      "loss": 0.3432,
+      "step": 2925
+    },
+    {
+      "epoch": 4.574551131928181,
+      "grad_norm": 0.3188318610191345,
+      "learning_rate": 0.00013162700071931184,
+      "loss": 0.3397,
+      "step": 2930
+    },
+    {
+      "epoch": 4.5823575331772055,
+      "grad_norm": 0.3379952907562256,
+      "learning_rate": 0.00013136817403988917,
+      "loss": 0.3389,
+      "step": 2935
+    },
+    {
+      "epoch": 4.590163934426229,
+      "grad_norm": 0.2957393229007721,
+      "learning_rate": 0.00013110911407710908,
+      "loss": 0.3294,
+      "step": 2940
+    },
+    {
+      "epoch": 4.597970335675254,
+      "grad_norm": 0.30884405970573425,
+      "learning_rate": 0.00013084982275758615,
+      "loss": 0.3367,
+      "step": 2945
+    },
+    {
+      "epoch": 4.605776736924278,
+      "grad_norm": 0.31198689341545105,
+      "learning_rate": 0.00013059030200965536,
+      "loss": 0.3301,
+      "step": 2950
+    },
+    {
+      "epoch": 4.613583138173302,
+      "grad_norm": 0.3176429271697998,
+      "learning_rate": 0.00013033055376335804,
+      "loss": 0.3356,
+      "step": 2955
+    },
+    {
+      "epoch": 4.621389539422326,
+      "grad_norm": 0.3074142038822174,
+      "learning_rate": 0.00013007057995042732,
+      "loss": 0.3315,
+      "step": 2960
+    },
+    {
+      "epoch": 4.629195940671351,
+      "grad_norm": 0.30663779377937317,
+      "learning_rate": 0.00012981038250427403,
+      "loss": 0.3376,
+      "step": 2965
+    },
+    {
+      "epoch": 4.6370023419203745,
+      "grad_norm": 0.30579113960266113,
+      "learning_rate": 0.0001295499633599719,
+      "loss": 0.3351,
+      "step": 2970
+    },
+    {
+      "epoch": 4.644808743169399,
+      "grad_norm": 0.3152572214603424,
+      "learning_rate": 0.00012928932445424365,
+      "loss": 0.3369,
+      "step": 2975
+    },
+    {
+      "epoch": 4.652615144418423,
+      "grad_norm": 0.3045886754989624,
+      "learning_rate": 0.00012902846772544624,
+      "loss": 0.3317,
+      "step": 2980
+    },
+    {
+      "epoch": 4.660421545667448,
+      "grad_norm": 0.3164273500442505,
+      "learning_rate": 0.00012876739511355657,
+      "loss": 0.3408,
+      "step": 2985
+    },
+    {
+      "epoch": 4.668227946916471,
+      "grad_norm": 0.3016981780529022,
+      "learning_rate": 0.0001285061085601571,
+      "loss": 0.3417,
+      "step": 2990
+    },
+    {
+      "epoch": 4.676034348165496,
+      "grad_norm": 0.3070276081562042,
+      "learning_rate": 0.00012824461000842126,
+      "loss": 0.3283,
+      "step": 2995
+    },
+    {
+      "epoch": 4.68384074941452,
+      "grad_norm": 0.30681294202804565,
+      "learning_rate": 0.00012798290140309923,
+      "loss": 0.3395,
+      "step": 3000
+    },
+    {
+      "epoch": 4.691647150663544,
+      "grad_norm": 0.3165208697319031,
+      "learning_rate": 0.0001277209846905032,
+      "loss": 0.3358,
+      "step": 3005
+    },
+    {
+      "epoch": 4.699453551912568,
+      "grad_norm": 0.30810585618019104,
+      "learning_rate": 0.00012745886181849325,
+      "loss": 0.3409,
+      "step": 3010
+    },
+    {
+      "epoch": 4.707259953161593,
+      "grad_norm": 0.3146737515926361,
+      "learning_rate": 0.00012719653473646243,
+      "loss": 0.3449,
+      "step": 3015
+    },
+    {
+      "epoch": 4.715066354410617,
+      "grad_norm": 0.29630905389785767,
+      "learning_rate": 0.00012693400539532263,
+      "loss": 0.3384,
+      "step": 3020
+    },
+    {
+      "epoch": 4.722872755659641,
+      "grad_norm": 0.3214413523674011,
+      "learning_rate": 0.00012667127574748986,
+      "loss": 0.3347,
+      "step": 3025
+    },
+    {
+      "epoch": 4.730679156908665,
+      "grad_norm": 0.3026057183742523,
+      "learning_rate": 0.00012640834774686985,
+      "loss": 0.3342,
+      "step": 3030
+    },
+    {
+      "epoch": 4.73848555815769,
+      "grad_norm": 0.31248968839645386,
+      "learning_rate": 0.00012614522334884344,
+      "loss": 0.3381,
+      "step": 3035
+    },
+    {
+      "epoch": 4.7462919594067134,
+      "grad_norm": 0.3201672434806824,
+      "learning_rate": 0.00012588190451025207,
+      "loss": 0.3282,
+      "step": 3040
+    },
+    {
+      "epoch": 4.754098360655737,
+      "grad_norm": 0.32542794942855835,
+      "learning_rate": 0.00012561839318938326,
+      "loss": 0.3464,
+      "step": 3045
+    },
+    {
+      "epoch": 4.761904761904762,
+      "grad_norm": 0.2962396740913391,
+      "learning_rate": 0.00012535469134595595,
+      "loss": 0.3451,
+      "step": 3050
+    },
+    {
+      "epoch": 4.7697111631537865,
+      "grad_norm": 0.3082274794578552,
+      "learning_rate": 0.00012509080094110604,
+      "loss": 0.3325,
+      "step": 3055
+    },
+    {
+      "epoch": 4.77751756440281,
+      "grad_norm": 0.31192833185195923,
+      "learning_rate": 0.00012482672393737164,
+      "loss": 0.3348,
+      "step": 3060
+    },
+    {
+      "epoch": 4.785323965651834,
+      "grad_norm": 0.30983906984329224,
+      "learning_rate": 0.00012456246229867877,
+      "loss": 0.3418,
+      "step": 3065
+    },
+    {
+      "epoch": 4.793130366900859,
+      "grad_norm": 0.34676092863082886,
+      "learning_rate": 0.0001242980179903264,
+      "loss": 0.3435,
+      "step": 3070
+    },
+    {
+      "epoch": 4.800936768149883,
+      "grad_norm": 0.3103564977645874,
+      "learning_rate": 0.0001240333929789721,
+      "loss": 0.3392,
+      "step": 3075
+    },
+    {
+      "epoch": 4.808743169398907,
+      "grad_norm": 0.32498899102211,
+      "learning_rate": 0.00012376858923261733,
+      "loss": 0.3387,
+      "step": 3080
+    },
+    {
+      "epoch": 4.816549570647931,
+      "grad_norm": 0.3146662414073944,
+      "learning_rate": 0.0001235036087205927,
+      "loss": 0.3406,
+      "step": 3085
+    },
+    {
+      "epoch": 4.8243559718969555,
+      "grad_norm": 0.32241055369377136,
+      "learning_rate": 0.0001232384534135435,
+      "loss": 0.3421,
+      "step": 3090
+    },
+    {
+      "epoch": 4.83216237314598,
+      "grad_norm": 0.29581108689308167,
+      "learning_rate": 0.00012297312528341484,
+      "loss": 0.3309,
+      "step": 3095
+    },
+    {
+      "epoch": 4.839968774395004,
+      "grad_norm": 0.3068925440311432,
+      "learning_rate": 0.00012270762630343734,
+      "loss": 0.3419,
+      "step": 3100
+    },
+    {
+      "epoch": 4.847775175644028,
+      "grad_norm": 0.3263617753982544,
+      "learning_rate": 0.000122441958448112,
+      "loss": 0.3359,
+      "step": 3105
+    },
+    {
+      "epoch": 4.855581576893052,
+      "grad_norm": 0.3101602792739868,
+      "learning_rate": 0.00012217612369319579,
+      "loss": 0.342,
+      "step": 3110
+    },
+    {
+      "epoch": 4.863387978142076,
+      "grad_norm": 0.3126956522464752,
+      "learning_rate": 0.00012191012401568698,
+      "loss": 0.3386,
+      "step": 3115
+    },
+    {
+      "epoch": 4.871194379391101,
+      "grad_norm": 0.3099190890789032,
+      "learning_rate": 0.00012164396139381029,
+      "loss": 0.3288,
+      "step": 3120
+    },
+    {
+      "epoch": 4.8790007806401245,
+      "grad_norm": 0.3113436996936798,
+      "learning_rate": 0.00012137763780700227,
+      "loss": 0.3412,
+      "step": 3125
+    },
+    {
+      "epoch": 4.886807181889149,
+      "grad_norm": 0.29880639910697937,
+      "learning_rate": 0.00012111115523589651,
+      "loss": 0.3292,
+      "step": 3130
+    },
+    {
+      "epoch": 4.894613583138173,
+      "grad_norm": 0.31602269411087036,
+      "learning_rate": 0.00012084451566230906,
+      "loss": 0.3372,
+      "step": 3135
+    },
+    {
+      "epoch": 4.902419984387198,
+      "grad_norm": 0.3172509968280792,
+      "learning_rate": 0.00012057772106922349,
+      "loss": 0.3395,
+      "step": 3140
+    },
+    {
+      "epoch": 4.910226385636221,
+      "grad_norm": 0.31268373131752014,
+      "learning_rate": 0.0001203107734407763,
+      "loss": 0.339,
+      "step": 3145
+    },
+    {
+      "epoch": 4.918032786885246,
+      "grad_norm": 0.3162590265274048,
+      "learning_rate": 0.00012004367476224206,
+      "loss": 0.3484,
+      "step": 3150
+    },
+    {
+      "epoch": 4.92583918813427,
+      "grad_norm": 0.34318986535072327,
+      "learning_rate": 0.00011977642702001875,
+      "loss": 0.3473,
+      "step": 3155
+    },
+    {
+      "epoch": 4.9336455893832944,
+      "grad_norm": 0.2913156747817993,
+      "learning_rate": 0.00011950903220161285,
+      "loss": 0.3415,
+      "step": 3160
+    },
+    {
+      "epoch": 4.941451990632318,
+      "grad_norm": 0.29538801312446594,
+      "learning_rate": 0.0001192414922956247,
+      "loss": 0.3361,
+      "step": 3165
+    },
+    {
+      "epoch": 4.949258391881343,
+      "grad_norm": 0.3098941743373871,
+      "learning_rate": 0.00011897380929173364,
+      "loss": 0.3384,
+      "step": 3170
+    },
+    {
+      "epoch": 4.957064793130367,
+      "grad_norm": 0.3089110851287842,
+      "learning_rate": 0.0001187059851806832,
+      "loss": 0.335,
+      "step": 3175
+    },
+    {
+      "epoch": 4.964871194379391,
+      "grad_norm": 0.29929882287979126,
+      "learning_rate": 0.00011843802195426635,
+      "loss": 0.3364,
+      "step": 3180
+    },
+    {
+      "epoch": 4.972677595628415,
+      "grad_norm": 0.31270211935043335,
+      "learning_rate": 0.00011816992160531062,
+      "loss": 0.3482,
+      "step": 3185
+    },
+    {
+      "epoch": 4.98048399687744,
+      "grad_norm": 0.3091076910495758,
+      "learning_rate": 0.0001179016861276633,
+      "loss": 0.3387,
+      "step": 3190
+    },
+    {
+      "epoch": 4.9882903981264635,
+      "grad_norm": 0.3052199184894562,
+      "learning_rate": 0.00011763331751617663,
+      "loss": 0.3454,
+      "step": 3195
+    },
+    {
+      "epoch": 4.996096799375488,
+      "grad_norm": 0.2918078303337097,
+      "learning_rate": 0.00011736481776669306,
+      "loss": 0.3413,
+      "step": 3200
+    },
+    {
+      "epoch": 4.999219359875098,
+      "eval_loss": 1.0547945499420166,
+      "eval_runtime": 0.336,
+      "eval_samples_per_second": 14.88,
+      "eval_steps_per_second": 2.976,
+      "step": 3202
+    },
+    {
+      "epoch": 5.003903200624512,
+      "grad_norm": 0.2900161147117615,
+      "learning_rate": 0.00011709618887603014,
+      "loss": 0.3236,
+      "step": 3205
+    },
+    {
+      "epoch": 5.0117096018735365,
+      "grad_norm": 0.3155598044395447,
+      "learning_rate": 0.00011682743284196593,
+      "loss": 0.3129,
+      "step": 3210
+    },
+    {
+      "epoch": 5.01951600312256,
+      "grad_norm": 0.3263396918773651,
+      "learning_rate": 0.00011655855166322409,
+      "loss": 0.3154,
+      "step": 3215
+    },
+    {
+      "epoch": 5.027322404371585,
+      "grad_norm": 0.3178098797798157,
+      "learning_rate": 0.0001162895473394589,
+      "loss": 0.3133,
+      "step": 3220
+    },
+    {
+      "epoch": 5.035128805620609,
+      "grad_norm": 0.3300285339355469,
+      "learning_rate": 0.00011602042187124045,
+      "loss": 0.3103,
+      "step": 3225
+    },
+    {
+      "epoch": 5.042935206869633,
+      "grad_norm": 0.3271562159061432,
+      "learning_rate": 0.00011575117726003979,
+      "loss": 0.3092,
+      "step": 3230
+    },
+    {
+      "epoch": 5.050741608118657,
+      "grad_norm": 0.3253972828388214,
+      "learning_rate": 0.00011548181550821411,
+      "loss": 0.3076,
+      "step": 3235
+    },
+    {
+      "epoch": 5.058548009367682,
+      "grad_norm": 0.33159464597702026,
+      "learning_rate": 0.00011521233861899167,
+      "loss": 0.3223,
+      "step": 3240
+    },
+    {
+      "epoch": 5.0663544106167056,
+      "grad_norm": 0.3398463726043701,
+      "learning_rate": 0.00011494274859645707,
+      "loss": 0.3163,
+      "step": 3245
+    },
+    {
+      "epoch": 5.07416081186573,
+      "grad_norm": 0.332904577255249,
+      "learning_rate": 0.00011467304744553618,
+      "loss": 0.308,
+      "step": 3250
+    },
+    {
+      "epoch": 5.081967213114754,
+      "grad_norm": 0.32349440455436707,
+      "learning_rate": 0.00011440323717198144,
+      "loss": 0.3045,
+      "step": 3255
+    },
+    {
+      "epoch": 5.089773614363779,
+      "grad_norm": 0.3356363773345947,
+      "learning_rate": 0.00011413331978235678,
+      "loss": 0.3188,
+      "step": 3260
+    },
+    {
+      "epoch": 5.097580015612802,
+      "grad_norm": 0.3371128737926483,
+      "learning_rate": 0.00011386329728402268,
+      "loss": 0.3103,
+      "step": 3265
+    },
+    {
+      "epoch": 5.105386416861827,
+      "grad_norm": 0.3300149142742157,
+      "learning_rate": 0.00011359317168512142,
+      "loss": 0.3055,
+      "step": 3270
+    },
+    {
+      "epoch": 5.113192818110851,
+      "grad_norm": 0.3270377814769745,
+      "learning_rate": 0.000113322944994562,
+      "loss": 0.3115,
+      "step": 3275
+    },
+    {
+      "epoch": 5.1209992193598755,
+      "grad_norm": 0.3231879472732544,
+      "learning_rate": 0.00011305261922200519,
+      "loss": 0.3075,
+      "step": 3280
+    },
+    {
+      "epoch": 5.128805620608899,
+      "grad_norm": 0.33695080876350403,
+      "learning_rate": 0.00011278219637784861,
+      "loss": 0.316,
+      "step": 3285
+    },
+    {
+      "epoch": 5.136612021857924,
+      "grad_norm": 0.34212931990623474,
+      "learning_rate": 0.00011251167847321193,
+      "loss": 0.3158,
+      "step": 3290
+    },
+    {
+      "epoch": 5.144418423106948,
+      "grad_norm": 0.3377404808998108,
+      "learning_rate": 0.00011224106751992163,
+      "loss": 0.317,
+      "step": 3295
+    },
+    {
+      "epoch": 5.152224824355972,
+      "grad_norm": 0.34452709555625916,
+      "learning_rate": 0.00011197036553049625,
+      "loss": 0.3228,
+      "step": 3300
+    },
+    {
+      "epoch": 5.160031225604996,
+      "grad_norm": 0.3356159031391144,
+      "learning_rate": 0.00011169957451813138,
+      "loss": 0.3124,
+      "step": 3305
+    },
+    {
+      "epoch": 5.167837626854021,
+      "grad_norm": 0.3585182726383209,
+      "learning_rate": 0.00011142869649668466,
+      "loss": 0.3168,
+      "step": 3310
+    },
+    {
+      "epoch": 5.1756440281030445,
+      "grad_norm": 0.3430497348308563,
+      "learning_rate": 0.00011115773348066074,
+      "loss": 0.3178,
+      "step": 3315
+    },
+    {
+      "epoch": 5.183450429352069,
+      "grad_norm": 0.3532644808292389,
+      "learning_rate": 0.00011088668748519647,
+      "loss": 0.3192,
+      "step": 3320
+    },
+    {
+      "epoch": 5.191256830601093,
+      "grad_norm": 0.34457042813301086,
+      "learning_rate": 0.00011061556052604578,
+      "loss": 0.3181,
+      "step": 3325
+    },
+    {
+      "epoch": 5.199063231850117,
+      "grad_norm": 0.33349376916885376,
+      "learning_rate": 0.00011034435461956466,
+      "loss": 0.3121,
+      "step": 3330
+    },
+    {
+      "epoch": 5.206869633099141,
+      "grad_norm": 0.3395507335662842,
+      "learning_rate": 0.00011007307178269632,
+      "loss": 0.3183,
+      "step": 3335
+    },
+    {
+      "epoch": 5.214676034348165,
+      "grad_norm": 0.3544360399246216,
+      "learning_rate": 0.0001098017140329561,
+      "loss": 0.3243,
+      "step": 3340
+    },
+    {
+      "epoch": 5.22248243559719,
+      "grad_norm": 0.3494921326637268,
+      "learning_rate": 0.0001095302833884163,
+      "loss": 0.3146,
+      "step": 3345
+    },
+    {
+      "epoch": 5.2302888368462135,
+      "grad_norm": 0.33652257919311523,
+      "learning_rate": 0.00010925878186769158,
+      "loss": 0.3135,
+      "step": 3350
+    },
+    {
+      "epoch": 5.238095238095238,
+      "grad_norm": 0.3407340943813324,
+      "learning_rate": 0.00010898721148992351,
+      "loss": 0.3205,
+      "step": 3355
+    },
+    {
+      "epoch": 5.245901639344262,
+      "grad_norm": 0.3510075509548187,
+      "learning_rate": 0.00010871557427476583,
+      "loss": 0.3127,
+      "step": 3360
+    },
+    {
+      "epoch": 5.2537080405932866,
+      "grad_norm": 0.34021666646003723,
+      "learning_rate": 0.00010844387224236932,
+      "loss": 0.3192,
+      "step": 3365
+    },
+    {
+      "epoch": 5.26151444184231,
+      "grad_norm": 0.32510459423065186,
+      "learning_rate": 0.00010817210741336684,
+      "loss": 0.3108,
+      "step": 3370
+    },
+    {
+      "epoch": 5.269320843091335,
+      "grad_norm": 0.36883097887039185,
+      "learning_rate": 0.00010790028180885821,
+      "loss": 0.3212,
+      "step": 3375
+    },
+    {
+      "epoch": 5.277127244340359,
+      "grad_norm": 0.3306715190410614,
+      "learning_rate": 0.00010762839745039526,
+      "loss": 0.325,
+      "step": 3380
+    },
+    {
+      "epoch": 5.284933645589383,
+      "grad_norm": 0.3390478491783142,
+      "learning_rate": 0.00010735645635996676,
+      "loss": 0.3159,
+      "step": 3385
+    },
+    {
+      "epoch": 5.292740046838407,
+      "grad_norm": 0.32592055201530457,
+      "learning_rate": 0.00010708446055998343,
+      "loss": 0.3127,
+      "step": 3390
+    },
+    {
+      "epoch": 5.300546448087432,
+      "grad_norm": 0.3241756856441498,
+      "learning_rate": 0.00010681241207326282,
+      "loss": 0.3087,
+      "step": 3395
+    },
+    {
+      "epoch": 5.308352849336456,
+      "grad_norm": 0.3321205675601959,
+      "learning_rate": 0.00010654031292301432,
+      "loss": 0.3191,
+      "step": 3400
+    },
+    {
+      "epoch": 5.31615925058548,
+      "grad_norm": 0.3427116572856903,
+      "learning_rate": 0.00010626816513282412,
+      "loss": 0.3167,
+      "step": 3405
+    },
+    {
+      "epoch": 5.323965651834504,
+      "grad_norm": 0.33668723702430725,
+      "learning_rate": 0.00010599597072664012,
+      "loss": 0.3173,
+      "step": 3410
+    },
+    {
+      "epoch": 5.331772053083529,
+      "grad_norm": 0.3475385010242462,
+      "learning_rate": 0.0001057237317287569,
+      "loss": 0.3211,
+      "step": 3415
+    },
+    {
+      "epoch": 5.339578454332552,
+      "grad_norm": 0.3414543867111206,
+      "learning_rate": 0.00010545145016380065,
+      "loss": 0.3171,
+      "step": 3420
+    },
+    {
+      "epoch": 5.347384855581577,
+      "grad_norm": 0.37167057394981384,
+      "learning_rate": 0.00010517912805671419,
+      "loss": 0.3199,
+      "step": 3425
+    },
+    {
+      "epoch": 5.355191256830601,
+      "grad_norm": 0.3445757031440735,
+      "learning_rate": 0.00010490676743274181,
+      "loss": 0.3176,
+      "step": 3430
+    },
+    {
+      "epoch": 5.3629976580796255,
+      "grad_norm": 0.33130553364753723,
+      "learning_rate": 0.00010463437031741423,
+      "loss": 0.3171,
+      "step": 3435
+    },
+    {
+      "epoch": 5.370804059328649,
+      "grad_norm": 0.34514832496643066,
+      "learning_rate": 0.00010436193873653361,
+      "loss": 0.3174,
+      "step": 3440
+    },
+    {
+      "epoch": 5.378610460577674,
+      "grad_norm": 0.3419683575630188,
+      "learning_rate": 0.00010408947471615835,
+      "loss": 0.3229,
+      "step": 3445
+    },
+    {
+      "epoch": 5.386416861826698,
+      "grad_norm": 0.3344433903694153,
+      "learning_rate": 0.00010381698028258817,
+      "loss": 0.3238,
+      "step": 3450
+    },
+    {
+      "epoch": 5.394223263075722,
+      "grad_norm": 0.3199335038661957,
+      "learning_rate": 0.00010354445746234894,
+      "loss": 0.3218,
+      "step": 3455
+    },
+    {
+      "epoch": 5.402029664324746,
+      "grad_norm": 0.3573337197303772,
+      "learning_rate": 0.00010327190828217763,
+      "loss": 0.3274,
+      "step": 3460
+    },
+    {
+      "epoch": 5.409836065573771,
+      "grad_norm": 0.3186776638031006,
+      "learning_rate": 0.00010299933476900727,
+      "loss": 0.3278,
+      "step": 3465
+    },
+    {
+      "epoch": 5.4176424668227945,
+      "grad_norm": 0.3291524648666382,
+      "learning_rate": 0.00010272673894995187,
+      "loss": 0.3207,
+      "step": 3470
+    },
+    {
+      "epoch": 5.425448868071819,
+      "grad_norm": 0.3383654057979584,
+      "learning_rate": 0.00010245412285229124,
+      "loss": 0.309,
+      "step": 3475
+    },
+    {
+      "epoch": 5.433255269320843,
+      "grad_norm": 0.33380231261253357,
+      "learning_rate": 0.00010218148850345613,
+      "loss": 0.3139,
+      "step": 3480
+    },
+    {
+      "epoch": 5.441061670569868,
+      "grad_norm": 0.3124459385871887,
+      "learning_rate": 0.0001019088379310129,
+      "loss": 0.3167,
+      "step": 3485
+    },
+    {
+      "epoch": 5.448868071818891,
+      "grad_norm": 0.34139421582221985,
+      "learning_rate": 0.00010163617316264869,
+      "loss": 0.3177,
+      "step": 3490
+    },
+    {
+      "epoch": 5.456674473067916,
+      "grad_norm": 0.33428099751472473,
+      "learning_rate": 0.00010136349622615608,
+      "loss": 0.3185,
+      "step": 3495
+    },
+    {
+      "epoch": 5.46448087431694,
+      "grad_norm": 0.3460736870765686,
+      "learning_rate": 0.00010109080914941824,
+      "loss": 0.3187,
+      "step": 3500
+    },
+    {
+      "epoch": 5.472287275565964,
+      "grad_norm": 0.35471439361572266,
+      "learning_rate": 0.00010081811396039373,
+      "loss": 0.3213,
+      "step": 3505
+    },
+    {
+      "epoch": 5.480093676814988,
+      "grad_norm": 0.3312179446220398,
+      "learning_rate": 0.00010054541268710138,
+      "loss": 0.3148,
+      "step": 3510
+    },
+    {
+      "epoch": 5.487900078064013,
+      "grad_norm": 0.34200319647789,
+      "learning_rate": 0.00010027270735760541,
+      "loss": 0.3268,
+      "step": 3515
+    },
+    {
+      "epoch": 5.495706479313037,
+      "grad_norm": 0.339599072933197,
+      "learning_rate": 0.0001,
+      "loss": 0.3148,
+      "step": 3520
+    },
+    {
+      "epoch": 5.503512880562061,
+      "grad_norm": 0.3320743143558502,
+      "learning_rate": 9.972729264239461e-05,
+      "loss": 0.3176,
+      "step": 3525
+    },
+    {
+      "epoch": 5.511319281811085,
+      "grad_norm": 0.3287266790866852,
+      "learning_rate": 9.945458731289862e-05,
+      "loss": 0.3235,
+      "step": 3530
+    },
+    {
+      "epoch": 5.51912568306011,
+      "grad_norm": 0.3429989814758301,
+      "learning_rate": 9.918188603960632e-05,
+      "loss": 0.3257,
+      "step": 3535
+    },
+    {
+      "epoch": 5.526932084309133,
+      "grad_norm": 0.3366888165473938,
+      "learning_rate": 9.890919085058178e-05,
+      "loss": 0.3194,
+      "step": 3540
+    },
+    {
+      "epoch": 5.534738485558158,
+      "grad_norm": 0.3251045346260071,
+      "learning_rate": 9.863650377384395e-05,
+      "loss": 0.315,
+      "step": 3545
+    },
+    {
+      "epoch": 5.542544886807182,
+      "grad_norm": 0.3391668200492859,
+      "learning_rate": 9.836382683735132e-05,
+      "loss": 0.3225,
+      "step": 3550
+    },
+    {
+      "epoch": 5.550351288056206,
+      "grad_norm": 0.3379986584186554,
+      "learning_rate": 9.809116206898712e-05,
+      "loss": 0.3251,
+      "step": 3555
+    },
+    {
+      "epoch": 5.55815768930523,
+      "grad_norm": 0.336603581905365,
+      "learning_rate": 9.78185114965439e-05,
+      "loss": 0.3211,
+      "step": 3560
+    },
+    {
+      "epoch": 5.565964090554255,
+      "grad_norm": 0.3426755368709564,
+      "learning_rate": 9.754587714770878e-05,
+      "loss": 0.323,
+      "step": 3565
+    },
+    {
+      "epoch": 5.573770491803279,
+      "grad_norm": 0.3463711738586426,
+      "learning_rate": 9.727326105004817e-05,
+      "loss": 0.3202,
+      "step": 3570
+    },
+    {
+      "epoch": 5.581576893052302,
+      "grad_norm": 0.36325374245643616,
+      "learning_rate": 9.700066523099273e-05,
+      "loss": 0.3143,
+      "step": 3575
+    },
+    {
+      "epoch": 5.589383294301327,
+      "grad_norm": 0.32558673620224,
+      "learning_rate": 9.67280917178224e-05,
+      "loss": 0.3286,
+      "step": 3580
+    },
+    {
+      "epoch": 5.597189695550352,
+      "grad_norm": 0.33436307311058044,
+      "learning_rate": 9.64555425376511e-05,
+      "loss": 0.3168,
+      "step": 3585
+    },
+    {
+      "epoch": 5.6049960967993755,
+      "grad_norm": 0.33413755893707275,
+      "learning_rate": 9.618301971741184e-05,
+      "loss": 0.3135,
+      "step": 3590
+    },
+    {
+      "epoch": 5.612802498048399,
+      "grad_norm": 0.3465372920036316,
+      "learning_rate": 9.591052528384167e-05,
+      "loss": 0.3175,
+      "step": 3595
+    },
+    {
+      "epoch": 5.620608899297424,
+      "grad_norm": 0.3320391774177551,
+      "learning_rate": 9.563806126346642e-05,
+      "loss": 0.3201,
+      "step": 3600
+    },
+    {
+      "epoch": 5.628415300546449,
+      "grad_norm": 0.3287176191806793,
+      "learning_rate": 9.536562968258578e-05,
+      "loss": 0.3224,
+      "step": 3605
+    },
+    {
+      "epoch": 5.636221701795472,
+      "grad_norm": 0.3486092686653137,
+      "learning_rate": 9.509323256725821e-05,
+      "loss": 0.3254,
+      "step": 3610
+    },
+    {
+      "epoch": 5.644028103044496,
+      "grad_norm": 0.3497357666492462,
+      "learning_rate": 9.482087194328583e-05,
+      "loss": 0.3178,
+      "step": 3615
+    },
+    {
+      "epoch": 5.651834504293521,
+      "grad_norm": 0.338351309299469,
+      "learning_rate": 9.454854983619936e-05,
+      "loss": 0.3312,
+      "step": 3620
+    },
+    {
+      "epoch": 5.6596409055425445,
+      "grad_norm": 0.3393017649650574,
+      "learning_rate": 9.427626827124317e-05,
+      "loss": 0.3171,
+      "step": 3625
+    },
+    {
+      "epoch": 5.667447306791569,
+      "grad_norm": 0.34753820300102234,
+      "learning_rate": 9.400402927335992e-05,
+      "loss": 0.3231,
+      "step": 3630
+    },
+    {
+      "epoch": 5.675253708040593,
+      "grad_norm": 0.3426099121570587,
+      "learning_rate": 9.373183486717592e-05,
+      "loss": 0.3195,
+      "step": 3635
+    },
+    {
+      "epoch": 5.683060109289618,
+      "grad_norm": 0.3154798150062561,
+      "learning_rate": 9.345968707698569e-05,
+      "loss": 0.3234,
+      "step": 3640
+    },
+    {
+      "epoch": 5.690866510538641,
+      "grad_norm": 0.3466772735118866,
+      "learning_rate": 9.318758792673722e-05,
+      "loss": 0.3117,
+      "step": 3645
+    },
+    {
+      "epoch": 5.698672911787666,
+      "grad_norm": 0.3402402997016907,
+      "learning_rate": 9.29155394400166e-05,
+      "loss": 0.3197,
+      "step": 3650
+    },
+    {
+      "epoch": 5.70647931303669,
+      "grad_norm": 0.34711048007011414,
+      "learning_rate": 9.264354364003327e-05,
+      "loss": 0.3238,
+      "step": 3655
+    },
+    {
+      "epoch": 5.714285714285714,
+      "grad_norm": 0.33199217915534973,
+      "learning_rate": 9.237160254960476e-05,
+      "loss": 0.3136,
+      "step": 3660
+    },
+    {
+      "epoch": 5.722092115534738,
+      "grad_norm": 0.3648219108581543,
+      "learning_rate": 9.209971819114181e-05,
+      "loss": 0.3257,
+      "step": 3665
+    },
+    {
+      "epoch": 5.729898516783763,
+      "grad_norm": 0.3323144316673279,
+      "learning_rate": 9.182789258663321e-05,
+      "loss": 0.3162,
+      "step": 3670
+    },
+    {
+      "epoch": 5.737704918032787,
+      "grad_norm": 0.32556962966918945,
+      "learning_rate": 9.155612775763069e-05,
+      "loss": 0.3191,
+      "step": 3675
+    },
+    {
+      "epoch": 5.745511319281811,
+      "grad_norm": 0.33691149950027466,
+      "learning_rate": 9.128442572523417e-05,
+      "loss": 0.3221,
+      "step": 3680
+    },
+    {
+      "epoch": 5.753317720530835,
+      "grad_norm": 0.33508819341659546,
+      "learning_rate": 9.10127885100765e-05,
+      "loss": 0.3215,
+      "step": 3685
+    },
+    {
+      "epoch": 5.76112412177986,
+      "grad_norm": 0.31548982858657837,
+      "learning_rate": 9.074121813230847e-05,
+      "loss": 0.3134,
+      "step": 3690
+    },
+    {
+      "epoch": 5.768930523028883,
+      "grad_norm": 0.33069273829460144,
+      "learning_rate": 9.046971661158372e-05,
+      "loss": 0.3138,
+      "step": 3695
+    },
+    {
+      "epoch": 5.776736924277908,
+      "grad_norm": 0.3511803150177002,
+      "learning_rate": 9.019828596704394e-05,
+      "loss": 0.322,
+      "step": 3700
+    },
+    {
+      "epoch": 5.784543325526932,
+      "grad_norm": 0.34525299072265625,
+      "learning_rate": 8.992692821730368e-05,
+      "loss": 0.3268,
+      "step": 3705
+    },
+    {
+      "epoch": 5.7923497267759565,
+      "grad_norm": 0.32578587532043457,
+      "learning_rate": 8.965564538043535e-05,
+      "loss": 0.3254,
+      "step": 3710
+    },
+    {
+      "epoch": 5.80015612802498,
+      "grad_norm": 0.33581212162971497,
+      "learning_rate": 8.938443947395427e-05,
+      "loss": 0.3249,
+      "step": 3715
+    },
+    {
+      "epoch": 5.807962529274005,
+      "grad_norm": 0.34412479400634766,
+      "learning_rate": 8.911331251480357e-05,
+      "loss": 0.3263,
+      "step": 3720
+    },
+    {
+      "epoch": 5.815768930523029,
+      "grad_norm": 0.3513209819793701,
+      "learning_rate": 8.884226651933927e-05,
+      "loss": 0.3199,
+      "step": 3725
+    },
+    {
+      "epoch": 5.823575331772053,
+      "grad_norm": 0.3510242700576782,
+      "learning_rate": 8.857130350331535e-05,
+      "loss": 0.3156,
+      "step": 3730
+    },
+    {
+      "epoch": 5.831381733021077,
+      "grad_norm": 0.3359852135181427,
+      "learning_rate": 8.830042548186863e-05,
+      "loss": 0.3225,
+      "step": 3735
+    },
+    {
+      "epoch": 5.839188134270102,
+      "grad_norm": 0.3355961740016937,
+      "learning_rate": 8.802963446950377e-05,
+      "loss": 0.3229,
+      "step": 3740
+    },
+    {
+      "epoch": 5.8469945355191255,
+      "grad_norm": 0.3400302827358246,
+      "learning_rate": 8.775893248007839e-05,
+      "loss": 0.331,
+      "step": 3745
+    },
+    {
+      "epoch": 5.85480093676815,
+      "grad_norm": 0.3256356120109558,
+      "learning_rate": 8.74883215267881e-05,
+      "loss": 0.327,
+      "step": 3750
+    },
+    {
+      "epoch": 5.862607338017174,
+      "grad_norm": 0.35503053665161133,
+      "learning_rate": 8.721780362215138e-05,
+      "loss": 0.3273,
+      "step": 3755
+    },
+    {
+      "epoch": 5.870413739266199,
+      "grad_norm": 0.34520792961120605,
+      "learning_rate": 8.694738077799488e-05,
+      "loss": 0.3198,
+      "step": 3760
+    },
+    {
+      "epoch": 5.878220140515222,
+      "grad_norm": 0.32543712854385376,
+      "learning_rate": 8.667705500543803e-05,
+      "loss": 0.3171,
+      "step": 3765
+    },
+    {
+      "epoch": 5.886026541764247,
+      "grad_norm": 0.33950576186180115,
+      "learning_rate": 8.64068283148786e-05,
+      "loss": 0.3207,
+      "step": 3770
+    },
+    {
+      "epoch": 5.893832943013271,
+      "grad_norm": 0.3403869569301605,
+      "learning_rate": 8.613670271597733e-05,
+      "loss": 0.3243,
+      "step": 3775
+    },
+    {
+      "epoch": 5.901639344262295,
+      "grad_norm": 0.34109991788864136,
+      "learning_rate": 8.586668021764329e-05,
+      "loss": 0.3221,
+      "step": 3780
+    },
+    {
+      "epoch": 5.909445745511319,
+      "grad_norm": 0.3320345878601074,
+      "learning_rate": 8.559676282801858e-05,
+      "loss": 0.3171,
+      "step": 3785
+    },
+    {
+      "epoch": 5.917252146760344,
+      "grad_norm": 0.34376391768455505,
+      "learning_rate": 8.532695255446383e-05,
+      "loss": 0.3191,
+      "step": 3790
+    },
+    {
+      "epoch": 5.925058548009368,
+      "grad_norm": 0.33876165747642517,
+      "learning_rate": 8.505725140354294e-05,
+      "loss": 0.3198,
+      "step": 3795
+    },
+    {
+      "epoch": 5.932864949258392,
+      "grad_norm": 0.34862759709358215,
+      "learning_rate": 8.478766138100834e-05,
+      "loss": 0.3195,
+      "step": 3800
+    },
+    {
+      "epoch": 5.940671350507416,
+      "grad_norm": 0.34864258766174316,
+      "learning_rate": 8.451818449178591e-05,
+      "loss": 0.3271,
+      "step": 3805
+    },
+    {
+      "epoch": 5.948477751756441,
+      "grad_norm": 0.35281243920326233,
+      "learning_rate": 8.424882273996024e-05,
+      "loss": 0.3246,
+      "step": 3810
+    },
+    {
+      "epoch": 5.956284153005464,
+      "grad_norm": 0.33554619550704956,
+      "learning_rate": 8.397957812875959e-05,
+      "loss": 0.3239,
+      "step": 3815
+    },
+    {
+      "epoch": 5.964090554254488,
+      "grad_norm": 0.3569699227809906,
+      "learning_rate": 8.371045266054114e-05,
+      "loss": 0.3182,
+      "step": 3820
+    },
+    {
+      "epoch": 5.971896955503513,
+      "grad_norm": 0.3249468505382538,
+      "learning_rate": 8.344144833677594e-05,
+      "loss": 0.3218,
+      "step": 3825
+    },
+    {
+      "epoch": 5.9797033567525375,
+      "grad_norm": 0.3645569682121277,
+      "learning_rate": 8.317256715803407e-05,
+      "loss": 0.3158,
+      "step": 3830
+    },
+    {
+      "epoch": 5.987509758001561,
+      "grad_norm": 0.32271501421928406,
+      "learning_rate": 8.290381112396987e-05,
+      "loss": 0.3194,
+      "step": 3835
+    },
+    {
+      "epoch": 5.995316159250585,
+      "grad_norm": 0.3314792215824127,
+      "learning_rate": 8.263518223330697e-05,
+      "loss": 0.318,
+      "step": 3840
+    },
+    {
+      "epoch": 6.0,
+      "eval_loss": 1.0760587453842163,
+      "eval_runtime": 0.3225,
+      "eval_samples_per_second": 15.506,
+      "eval_steps_per_second": 3.101,
+      "step": 3843
+    },
+    {
+      "epoch": 6.00312256049961,
+      "grad_norm": 0.31583714485168457,
+      "learning_rate": 8.236668248382337e-05,
+      "loss": 0.3,
+      "step": 3845
+    },
+    {
+      "epoch": 6.0109289617486334,
+      "grad_norm": 0.36563432216644287,
+      "learning_rate": 8.209831387233676e-05,
+      "loss": 0.2928,
+      "step": 3850
+    },
+    {
+      "epoch": 6.018735362997658,
+      "grad_norm": 0.3586893081665039,
+      "learning_rate": 8.183007839468942e-05,
+      "loss": 0.2947,
+      "step": 3855
+    },
+    {
+      "epoch": 6.026541764246682,
+      "grad_norm": 0.351180762052536,
+      "learning_rate": 8.156197804573366e-05,
+      "loss": 0.2974,
+      "step": 3860
+    },
+    {
+      "epoch": 6.0343481654957065,
+      "grad_norm": 0.3694261610507965,
+      "learning_rate": 8.12940148193168e-05,
+      "loss": 0.2943,
+      "step": 3865
+    },
+    {
+      "epoch": 6.04215456674473,
+      "grad_norm": 0.3579952120780945,
+      "learning_rate": 8.102619070826639e-05,
+      "loss": 0.2993,
+      "step": 3870
+    },
+    {
+      "epoch": 6.049960967993755,
+      "grad_norm": 0.3623141348361969,
+      "learning_rate": 8.075850770437534e-05,
+      "loss": 0.3012,
+      "step": 3875
+    },
+    {
+      "epoch": 6.057767369242779,
+      "grad_norm": 0.35049453377723694,
+      "learning_rate": 8.049096779838719e-05,
+      "loss": 0.2954,
+      "step": 3880
+    },
+    {
+      "epoch": 6.065573770491803,
+      "grad_norm": 0.36742958426475525,
+      "learning_rate": 8.022357297998127e-05,
+      "loss": 0.292,
+      "step": 3885
+    },
+    {
+      "epoch": 6.073380171740827,
+      "grad_norm": 0.36501386761665344,
+      "learning_rate": 7.995632523775795e-05,
+      "loss": 0.2938,
+      "step": 3890
+    },
+    {
+      "epoch": 6.081186572989852,
+      "grad_norm": 0.35817086696624756,
+      "learning_rate": 7.968922655922374e-05,
+      "loss": 0.302,
+      "step": 3895
+    },
+    {
+      "epoch": 6.0889929742388755,
+      "grad_norm": 0.3561493158340454,
+      "learning_rate": 7.942227893077652e-05,
+      "loss": 0.3042,
+      "step": 3900
+    },
+    {
+      "epoch": 6.0967993754879,
+      "grad_norm": 0.3571511507034302,
+      "learning_rate": 7.915548433769095e-05,
+      "loss": 0.2971,
+      "step": 3905
+    },
+    {
+      "epoch": 6.104605776736924,
+      "grad_norm": 0.34343209862709045,
+      "learning_rate": 7.888884476410348e-05,
+      "loss": 0.2956,
+      "step": 3910
+    },
+    {
+      "epoch": 6.112412177985949,
+      "grad_norm": 0.3509383201599121,
+      "learning_rate": 7.862236219299777e-05,
+      "loss": 0.2991,
+      "step": 3915
+    },
+    {
+      "epoch": 6.120218579234972,
+      "grad_norm": 0.35262179374694824,
+      "learning_rate": 7.835603860618972e-05,
+      "loss": 0.2984,
+      "step": 3920
+    },
+    {
+      "epoch": 6.128024980483997,
+      "grad_norm": 0.35837092995643616,
+      "learning_rate": 7.808987598431303e-05,
+      "loss": 0.2949,
+      "step": 3925
+    },
+    {
+      "epoch": 6.135831381733021,
+      "grad_norm": 0.3698333501815796,
+      "learning_rate": 7.782387630680421e-05,
+      "loss": 0.3056,
+      "step": 3930
+    },
+    {
+      "epoch": 6.143637782982045,
+      "grad_norm": 0.3528449535369873,
+      "learning_rate": 7.755804155188802e-05,
+      "loss": 0.2861,
+      "step": 3935
+    },
+    {
+      "epoch": 6.151444184231069,
+      "grad_norm": 0.3811301290988922,
+      "learning_rate": 7.729237369656269e-05,
+      "loss": 0.3052,
+      "step": 3940
+    },
+    {
+      "epoch": 6.159250585480094,
+      "grad_norm": 0.37010228633880615,
+      "learning_rate": 7.702687471658518e-05,
+      "loss": 0.2911,
+      "step": 3945
+    },
+    {
+      "epoch": 6.167056986729118,
+      "grad_norm": 0.37986573576927185,
+      "learning_rate": 7.676154658645656e-05,
+      "loss": 0.2997,
+      "step": 3950
+    },
+    {
+      "epoch": 6.174863387978142,
+      "grad_norm": 0.34996771812438965,
+      "learning_rate": 7.649639127940735e-05,
+      "loss": 0.3029,
+      "step": 3955
+    },
+    {
+      "epoch": 6.182669789227166,
+      "grad_norm": 0.3746223449707031,
+      "learning_rate": 7.623141076738271e-05,
+      "loss": 0.3064,
+      "step": 3960
+    },
+    {
+      "epoch": 6.190476190476191,
+      "grad_norm": 0.3640713393688202,
+      "learning_rate": 7.596660702102791e-05,
+      "loss": 0.3013,
+      "step": 3965
+    },
+    {
+      "epoch": 6.1982825917252145,
+      "grad_norm": 0.3776625096797943,
+      "learning_rate": 7.570198200967362e-05,
+      "loss": 0.3063,
+      "step": 3970
+    },
+    {
+      "epoch": 6.206088992974239,
+      "grad_norm": 0.3669188916683197,
+      "learning_rate": 7.543753770132127e-05,
+      "loss": 0.2982,
+      "step": 3975
+    },
+    {
+      "epoch": 6.213895394223263,
+      "grad_norm": 0.37241482734680176,
+      "learning_rate": 7.517327606262836e-05,
+      "loss": 0.3,
+      "step": 3980
+    },
+    {
+      "epoch": 6.2217017954722875,
+      "grad_norm": 0.36023199558258057,
+      "learning_rate": 7.490919905889403e-05,
+      "loss": 0.2997,
+      "step": 3985
+    },
+    {
+      "epoch": 6.229508196721311,
+      "grad_norm": 0.36012396216392517,
+      "learning_rate": 7.464530865404407e-05,
+      "loss": 0.2926,
+      "step": 3990
+    },
+    {
+      "epoch": 6.237314597970336,
+      "grad_norm": 0.3803520202636719,
+      "learning_rate": 7.438160681061676e-05,
+      "loss": 0.3045,
+      "step": 3995
+    },
+    {
+      "epoch": 6.24512099921936,
+      "grad_norm": 0.34980347752571106,
+      "learning_rate": 7.411809548974792e-05,
+      "loss": 0.2991,
+      "step": 4000
+    },
+    {
+      "epoch": 6.252927400468384,
+      "grad_norm": 0.3561069369316101,
+      "learning_rate": 7.385477665115658e-05,
+      "loss": 0.3094,
+      "step": 4005
+    },
+    {
+      "epoch": 6.260733801717408,
+      "grad_norm": 0.36508041620254517,
+      "learning_rate": 7.359165225313019e-05,
+      "loss": 0.2969,
+      "step": 4010
+    },
+    {
+      "epoch": 6.268540202966433,
+      "grad_norm": 0.36842647194862366,
+      "learning_rate": 7.332872425251018e-05,
+      "loss": 0.3009,
+      "step": 4015
+    },
+    {
+      "epoch": 6.2763466042154565,
+      "grad_norm": 0.3771759569644928,
+      "learning_rate": 7.30659946046774e-05,
+      "loss": 0.3027,
+      "step": 4020
+    },
+    {
+      "epoch": 6.284153005464481,
+      "grad_norm": 0.36950060725212097,
+      "learning_rate": 7.280346526353759e-05,
+      "loss": 0.3052,
+      "step": 4025
+    },
+    {
+      "epoch": 6.291959406713505,
+      "grad_norm": 0.36776700615882874,
+      "learning_rate": 7.25411381815068e-05,
+      "loss": 0.298,
+      "step": 4030
+    },
+    {
+      "epoch": 6.29976580796253,
+      "grad_norm": 0.3699568510055542,
+      "learning_rate": 7.22790153094968e-05,
+      "loss": 0.2961,
+      "step": 4035
+    },
+    {
+      "epoch": 6.307572209211553,
+      "grad_norm": 0.36055341362953186,
+      "learning_rate": 7.20170985969008e-05,
+      "loss": 0.297,
+      "step": 4040
+    },
+    {
+      "epoch": 6.315378610460578,
+      "grad_norm": 0.37468183040618896,
+      "learning_rate": 7.175538999157876e-05,
+      "loss": 0.2989,
+      "step": 4045
+    },
+    {
+      "epoch": 6.323185011709602,
+      "grad_norm": 0.3778867721557617,
+      "learning_rate": 7.149389143984295e-05,
+      "loss": 0.2994,
+      "step": 4050
+    },
+    {
+      "epoch": 6.330991412958626,
+      "grad_norm": 0.37846189737319946,
+      "learning_rate": 7.123260488644345e-05,
+      "loss": 0.2949,
+      "step": 4055
+    },
+    {
+      "epoch": 6.33879781420765,
+      "grad_norm": 0.3595486581325531,
+      "learning_rate": 7.097153227455379e-05,
+      "loss": 0.3063,
+      "step": 4060
+    },
+    {
+      "epoch": 6.346604215456675,
+      "grad_norm": 0.38495534658432007,
+      "learning_rate": 7.071067554575637e-05,
+      "loss": 0.3037,
+      "step": 4065
+    },
+    {
+      "epoch": 6.354410616705699,
+      "grad_norm": 0.37678053975105286,
+      "learning_rate": 7.045003664002809e-05,
+      "loss": 0.3111,
+      "step": 4070
+    },
+    {
+      "epoch": 6.362217017954723,
+      "grad_norm": 0.353000283241272,
+      "learning_rate": 7.018961749572604e-05,
+      "loss": 0.3003,
+      "step": 4075
+    },
+    {
+      "epoch": 6.370023419203747,
+      "grad_norm": 0.381798654794693,
+      "learning_rate": 6.992942004957271e-05,
+      "loss": 0.3037,
+      "step": 4080
+    },
+    {
+      "epoch": 6.377829820452771,
+      "grad_norm": 0.34018296003341675,
+      "learning_rate": 6.9669446236642e-05,
+      "loss": 0.2956,
+      "step": 4085
+    },
+    {
+      "epoch": 6.3856362217017955,
+      "grad_norm": 0.3533703684806824,
+      "learning_rate": 6.940969799034465e-05,
+      "loss": 0.298,
+      "step": 4090
+    },
+    {
+      "epoch": 6.39344262295082,
+      "grad_norm": 0.36671626567840576,
+      "learning_rate": 6.915017724241389e-05,
+      "loss": 0.3004,
+      "step": 4095
+    },
+    {
+      "epoch": 6.401249024199844,
+      "grad_norm": 0.3814579248428345,
+      "learning_rate": 6.889088592289093e-05,
+      "loss": 0.2992,
+      "step": 4100
+    },
+    {
+      "epoch": 6.409055425448868,
+      "grad_norm": 0.38459497690200806,
+      "learning_rate": 6.863182596011087e-05,
+      "loss": 0.3036,
+      "step": 4105
+    },
+    {
+      "epoch": 6.416861826697892,
+      "grad_norm": 0.3617273271083832,
+      "learning_rate": 6.837299928068817e-05,
+      "loss": 0.3037,
+      "step": 4110
+    },
+    {
+      "epoch": 6.424668227946917,
+      "grad_norm": 0.3558304011821747,
+      "learning_rate": 6.811440780950237e-05,
+      "loss": 0.2934,
+      "step": 4115
+    },
+    {
+      "epoch": 6.432474629195941,
+      "grad_norm": 0.36529627442359924,
+      "learning_rate": 6.785605346968386e-05,
+      "loss": 0.3034,
+      "step": 4120
+    },
+    {
+      "epoch": 6.4402810304449645,
+      "grad_norm": 0.3953593373298645,
+      "learning_rate": 6.759793818259933e-05,
+      "loss": 0.3005,
+      "step": 4125
+    },
+    {
+      "epoch": 6.448087431693989,
+      "grad_norm": 0.38395586609840393,
+      "learning_rate": 6.73400638678378e-05,
+      "loss": 0.3009,
+      "step": 4130
+    },
+    {
+      "epoch": 6.455893832943013,
+      "grad_norm": 0.35242578387260437,
+      "learning_rate": 6.708243244319611e-05,
+      "loss": 0.3053,
+      "step": 4135
+    },
+    {
+      "epoch": 6.4637002341920375,
+      "grad_norm": 0.3590448200702667,
+      "learning_rate": 6.682504582466482e-05,
+      "loss": 0.2994,
+      "step": 4140
+    },
+    {
+      "epoch": 6.471506635441061,
+      "grad_norm": 0.37192437052726746,
+      "learning_rate": 6.656790592641375e-05,
+      "loss": 0.3056,
+      "step": 4145
+    },
+    {
+      "epoch": 6.479313036690086,
+      "grad_norm": 0.35877934098243713,
+      "learning_rate": 6.6311014660778e-05,
+      "loss": 0.2958,
+      "step": 4150
+    },
+    {
+      "epoch": 6.48711943793911,
+      "grad_norm": 0.3730364143848419,
+      "learning_rate": 6.605437393824356e-05,
+      "loss": 0.3047,
+      "step": 4155
+    },
+    {
+      "epoch": 6.494925839188134,
+      "grad_norm": 0.3790695071220398,
+      "learning_rate": 6.579798566743314e-05,
+      "loss": 0.3027,
+      "step": 4160
+    },
+    {
+      "epoch": 6.502732240437158,
+      "grad_norm": 0.36754992604255676,
+      "learning_rate": 6.554185175509197e-05,
+      "loss": 0.2916,
+      "step": 4165
+    },
+    {
+      "epoch": 6.510538641686183,
+      "grad_norm": 0.37978771328926086,
+      "learning_rate": 6.528597410607364e-05,
+      "loss": 0.3023,
+      "step": 4170
+    },
+    {
+      "epoch": 6.518345042935207,
+      "grad_norm": 0.35783836245536804,
+      "learning_rate": 6.503035462332592e-05,
+      "loss": 0.3031,
+      "step": 4175
+    },
+    {
+      "epoch": 6.526151444184231,
+      "grad_norm": 0.3850097060203552,
+      "learning_rate": 6.477499520787665e-05,
+      "loss": 0.3025,
+      "step": 4180
+    },
+    {
+      "epoch": 6.533957845433255,
+      "grad_norm": 0.38699930906295776,
+      "learning_rate": 6.451989775881955e-05,
+      "loss": 0.3032,
+      "step": 4185
+    },
+    {
+      "epoch": 6.54176424668228,
+      "grad_norm": 0.3620811998844147,
+      "learning_rate": 6.42650641733e-05,
+      "loss": 0.3029,
+      "step": 4190
+    },
+    {
+      "epoch": 6.549570647931303,
+      "grad_norm": 0.3833983838558197,
+      "learning_rate": 6.401049634650118e-05,
+      "loss": 0.3018,
+      "step": 4195
+    },
+    {
+      "epoch": 6.557377049180328,
+      "grad_norm": 0.3642504811286926,
+      "learning_rate": 6.375619617162985e-05,
+      "loss": 0.3052,
+      "step": 4200
+    },
+    {
+      "epoch": 6.565183450429352,
+      "grad_norm": 0.34987977147102356,
+      "learning_rate": 6.350216553990212e-05,
+      "loss": 0.3027,
+      "step": 4205
+    },
+    {
+      "epoch": 6.5729898516783765,
+      "grad_norm": 0.3789624571800232,
+      "learning_rate": 6.324840634052967e-05,
+      "loss": 0.2983,
+      "step": 4210
+    },
+    {
+      "epoch": 6.5807962529274,
+      "grad_norm": 0.39051589369773865,
+      "learning_rate": 6.29949204607054e-05,
+      "loss": 0.3099,
+      "step": 4215
+    },
+    {
+      "epoch": 6.588602654176425,
+      "grad_norm": 0.36654597520828247,
+      "learning_rate": 6.27417097855897e-05,
+      "loss": 0.3067,
+      "step": 4220
+    },
+    {
+      "epoch": 6.596409055425449,
+      "grad_norm": 0.3750831186771393,
+      "learning_rate": 6.248877619829619e-05,
+      "loss": 0.304,
+      "step": 4225
+    },
+    {
+      "epoch": 6.604215456674473,
+      "grad_norm": 0.3867753744125366,
+      "learning_rate": 6.223612157987786e-05,
+      "loss": 0.3082,
+      "step": 4230
+    },
+    {
+      "epoch": 6.612021857923497,
+      "grad_norm": 0.3690248429775238,
+      "learning_rate": 6.198374780931293e-05,
+      "loss": 0.3044,
+      "step": 4235
+    },
+    {
+      "epoch": 6.619828259172522,
+      "grad_norm": 0.37476563453674316,
+      "learning_rate": 6.173165676349103e-05,
+      "loss": 0.3027,
+      "step": 4240
+    },
+    {
+      "epoch": 6.6276346604215455,
+      "grad_norm": 0.36592650413513184,
+      "learning_rate": 6.14798503171992e-05,
+      "loss": 0.3015,
+      "step": 4245
+    },
+    {
+      "epoch": 6.63544106167057,
+      "grad_norm": 0.3869721293449402,
+      "learning_rate": 6.122833034310793e-05,
+      "loss": 0.2986,
+      "step": 4250
+    },
+    {
+      "epoch": 6.643247462919594,
+      "grad_norm": 0.3800086975097656,
+      "learning_rate": 6.097709871175723e-05,
+      "loss": 0.3026,
+      "step": 4255
+    },
+    {
+      "epoch": 6.6510538641686185,
+      "grad_norm": 0.3717212677001953,
+      "learning_rate": 6.0726157291542605e-05,
+      "loss": 0.303,
+      "step": 4260
+    },
+    {
+      "epoch": 6.658860265417642,
+      "grad_norm": 0.35861897468566895,
+      "learning_rate": 6.047550794870145e-05,
+      "loss": 0.3008,
+      "step": 4265
+    },
+    {
+      "epoch": 6.666666666666667,
+      "grad_norm": 0.3755732476711273,
+      "learning_rate": 6.02251525472989e-05,
+      "loss": 0.3006,
+      "step": 4270
+    },
+    {
+      "epoch": 6.674473067915691,
+      "grad_norm": 0.3563028872013092,
+      "learning_rate": 5.9975092949214116e-05,
+      "loss": 0.3041,
+      "step": 4275
+    },
+    {
+      "epoch": 6.682279469164715,
+      "grad_norm": 0.3711301386356354,
+      "learning_rate": 5.9725331014126294e-05,
+      "loss": 0.3047,
+      "step": 4280
+    },
+    {
+      "epoch": 6.690085870413739,
+      "grad_norm": 0.3619491457939148,
+      "learning_rate": 5.947586859950103e-05,
+      "loss": 0.3034,
+      "step": 4285
+    },
+    {
+      "epoch": 6.697892271662764,
+      "grad_norm": 0.3621855080127716,
+      "learning_rate": 5.922670756057633e-05,
+      "loss": 0.2993,
+      "step": 4290
+    },
+    {
+      "epoch": 6.705698672911788,
+      "grad_norm": 0.3833586573600769,
+      "learning_rate": 5.8977849750348944e-05,
+      "loss": 0.2969,
+      "step": 4295
+    },
+    {
+      "epoch": 6.713505074160812,
+      "grad_norm": 0.36535149812698364,
+      "learning_rate": 5.872929701956054e-05,
+      "loss": 0.3109,
+      "step": 4300
+    },
+    {
+      "epoch": 6.721311475409836,
+      "grad_norm": 0.34852883219718933,
+      "learning_rate": 5.848105121668381e-05,
+      "loss": 0.2996,
+      "step": 4305
+    },
+    {
+      "epoch": 6.729117876658861,
+      "grad_norm": 0.35054901242256165,
+      "learning_rate": 5.8233114187908935e-05,
+      "loss": 0.3038,
+      "step": 4310
+    },
+    {
+      "epoch": 6.736924277907884,
+      "grad_norm": 0.38017308712005615,
+      "learning_rate": 5.7985487777129765e-05,
+      "loss": 0.3023,
+      "step": 4315
+    },
+    {
+      "epoch": 6.744730679156909,
+      "grad_norm": 0.3646107017993927,
+      "learning_rate": 5.773817382593008e-05,
+      "loss": 0.3097,
+      "step": 4320
+    },
+    {
+      "epoch": 6.752537080405933,
+      "grad_norm": 0.3594595491886139,
+      "learning_rate": 5.749117417356988e-05,
+      "loss": 0.3088,
+      "step": 4325
+    },
+    {
+      "epoch": 6.760343481654957,
+      "grad_norm": 0.39560645818710327,
+      "learning_rate": 5.7244490656971815e-05,
+      "loss": 0.3058,
+      "step": 4330
+    },
+    {
+      "epoch": 6.768149882903981,
+      "grad_norm": 0.3735295236110687,
+      "learning_rate": 5.699812511070734e-05,
+      "loss": 0.3109,
+      "step": 4335
+    },
+    {
+      "epoch": 6.775956284153006,
+      "grad_norm": 0.3793189823627472,
+      "learning_rate": 5.675207936698337e-05,
+      "loss": 0.2962,
+      "step": 4340
+    },
+    {
+      "epoch": 6.78376268540203,
+      "grad_norm": 0.36545222997665405,
+      "learning_rate": 5.6506355255628284e-05,
+      "loss": 0.3034,
+      "step": 4345
+    },
+    {
+      "epoch": 6.791569086651053,
+      "grad_norm": 0.3819272816181183,
+      "learning_rate": 5.6260954604078585e-05,
+      "loss": 0.3055,
+      "step": 4350
+    },
+    {
+      "epoch": 6.799375487900078,
+      "grad_norm": 0.40963688492774963,
+      "learning_rate": 5.601587923736518e-05,
+      "loss": 0.297,
+      "step": 4355
+    },
+    {
+      "epoch": 6.807181889149103,
+      "grad_norm": 0.3582998812198639,
+      "learning_rate": 5.577113097809989e-05,
+      "loss": 0.3024,
+      "step": 4360
+    },
+    {
+      "epoch": 6.8149882903981265,
+      "grad_norm": 0.3594261407852173,
+      "learning_rate": 5.5526711646461835e-05,
+      "loss": 0.2976,
+      "step": 4365
+    },
+    {
+      "epoch": 6.82279469164715,
+      "grad_norm": 0.36158230900764465,
+      "learning_rate": 5.528262306018395e-05,
+      "loss": 0.3015,
+      "step": 4370
+    },
+    {
+      "epoch": 6.830601092896175,
+      "grad_norm": 0.37753212451934814,
+      "learning_rate": 5.503886703453933e-05,
+      "loss": 0.2942,
+      "step": 4375
+    },
+    {
+      "epoch": 6.8384074941451995,
+      "grad_norm": 0.37955158948898315,
+      "learning_rate": 5.4795445382328037e-05,
+      "loss": 0.3017,
+      "step": 4380
+    },
+    {
+      "epoch": 6.846213895394223,
+      "grad_norm": 0.3600054979324341,
+      "learning_rate": 5.4552359913863214e-05,
+      "loss": 0.3061,
+      "step": 4385
+    },
+    {
+      "epoch": 6.854020296643247,
+      "grad_norm": 0.3673628270626068,
+      "learning_rate": 5.4309612436957937e-05,
+      "loss": 0.3049,
+      "step": 4390
+    },
+    {
+      "epoch": 6.861826697892272,
+      "grad_norm": 0.36428987979888916,
+      "learning_rate": 5.4067204756911624e-05,
+      "loss": 0.3096,
+      "step": 4395
+    },
+    {
+      "epoch": 6.8696330991412955,
+      "grad_norm": 0.3699190616607666,
+      "learning_rate": 5.382513867649663e-05,
+      "loss": 0.3008,
+      "step": 4400
+    },
+    {
+      "epoch": 6.87743950039032,
+      "grad_norm": 0.36936306953430176,
+      "learning_rate": 5.358341599594483e-05,
+      "loss": 0.2979,
+      "step": 4405
+    },
+    {
+      "epoch": 6.885245901639344,
+      "grad_norm": 0.3822793960571289,
+      "learning_rate": 5.3342038512934424e-05,
+      "loss": 0.3009,
+      "step": 4410
+    },
+    {
+      "epoch": 6.893052302888369,
+      "grad_norm": 0.37353184819221497,
+      "learning_rate": 5.3101008022576126e-05,
+      "loss": 0.3061,
+      "step": 4415
+    },
+    {
+      "epoch": 6.900858704137392,
+      "grad_norm": 0.37411707639694214,
+      "learning_rate": 5.286032631740023e-05,
+      "loss": 0.3022,
+      "step": 4420
+    },
+    {
+      "epoch": 6.908665105386417,
+      "grad_norm": 0.3575087785720825,
+      "learning_rate": 5.261999518734322e-05,
+      "loss": 0.302,
+      "step": 4425
+    },
+    {
+      "epoch": 6.916471506635441,
+      "grad_norm": 0.3561764359474182,
+      "learning_rate": 5.238001641973422e-05,
+      "loss": 0.2998,
+      "step": 4430
+    },
+    {
+      "epoch": 6.924277907884465,
+      "grad_norm": 0.3756863474845886,
+      "learning_rate": 5.214039179928194e-05,
+      "loss": 0.3009,
+      "step": 4435
+    },
+    {
+      "epoch": 6.932084309133489,
+      "grad_norm": 0.36556264758110046,
+      "learning_rate": 5.190112310806126e-05,
+      "loss": 0.296,
+      "step": 4440
+    },
+    {
+      "epoch": 6.939890710382514,
+      "grad_norm": 0.37201789021492004,
+      "learning_rate": 5.1662212125500075e-05,
+      "loss": 0.2993,
+      "step": 4445
+    },
+    {
+      "epoch": 6.947697111631538,
+      "grad_norm": 0.34447523951530457,
+      "learning_rate": 5.142366062836599e-05,
+      "loss": 0.3002,
+      "step": 4450
+    },
+    {
+      "epoch": 6.955503512880562,
+      "grad_norm": 0.3652222156524658,
+      "learning_rate": 5.118547039075326e-05,
+      "loss": 0.3015,
+      "step": 4455
+    },
+    {
+      "epoch": 6.963309914129586,
+      "grad_norm": 0.3540942072868347,
+      "learning_rate": 5.09476431840692e-05,
+      "loss": 0.3031,
+      "step": 4460
+    },
+    {
+      "epoch": 6.971116315378611,
+      "grad_norm": 0.36961525678634644,
+      "learning_rate": 5.071018077702161e-05,
+      "loss": 0.3134,
+      "step": 4465
+    },
+    {
+      "epoch": 6.978922716627634,
+      "grad_norm": 0.3560226559638977,
+      "learning_rate": 5.047308493560506e-05,
+      "loss": 0.2953,
+      "step": 4470
+    },
+    {
+      "epoch": 6.986729117876659,
+      "grad_norm": 0.3567667007446289,
+      "learning_rate": 5.023635742308807e-05,
+      "loss": 0.3081,
+      "step": 4475
+    },
+    {
+      "epoch": 6.994535519125683,
+      "grad_norm": 0.3771991431713104,
+      "learning_rate": 5.000000000000002e-05,
+      "loss": 0.3056,
+      "step": 4480
+    },
+    {
+      "epoch": 6.999219359875098,
+      "eval_loss": 1.1023039817810059,
+      "eval_runtime": 0.3388,
+      "eval_samples_per_second": 14.759,
+      "eval_steps_per_second": 2.952,
+      "step": 4483
+    },
+    {
+      "epoch": 7.0023419203747075,
+      "grad_norm": 0.34990745782852173,
+      "learning_rate": 4.976401442411774e-05,
+      "loss": 0.2964,
+      "step": 4485
+    },
+    {
+      "epoch": 7.010148321623731,
+      "grad_norm": 0.3766123354434967,
+      "learning_rate": 4.952840245045278e-05,
+      "loss": 0.2868,
+      "step": 4490
+    },
+    {
+      "epoch": 7.017954722872756,
+      "grad_norm": 0.3829561769962311,
+      "learning_rate": 4.92931658312383e-05,
+      "loss": 0.2865,
+      "step": 4495
+    },
+    {
+      "epoch": 7.02576112412178,
+      "grad_norm": 0.3801458179950714,
+      "learning_rate": 4.9058306315915826e-05,
+      "loss": 0.277,
+      "step": 4500
+    },
+    {
+      "epoch": 7.033567525370804,
+      "grad_norm": 0.3700219690799713,
+      "learning_rate": 4.882382565112248e-05,
+      "loss": 0.2803,
+      "step": 4505
+    },
+    {
+      "epoch": 7.041373926619828,
+      "grad_norm": 0.3695278763771057,
+      "learning_rate": 4.8589725580677835e-05,
+      "loss": 0.2817,
+      "step": 4510
+    },
+    {
+      "epoch": 7.049180327868853,
+      "grad_norm": 0.3825577199459076,
+      "learning_rate": 4.835600784557106e-05,
+      "loss": 0.2791,
+      "step": 4515
+    },
+    {
+      "epoch": 7.0569867291178765,
+      "grad_norm": 0.36848512291908264,
+      "learning_rate": 4.8122674183947836e-05,
+      "loss": 0.2808,
+      "step": 4520
+    },
+    {
+      "epoch": 7.064793130366901,
+      "grad_norm": 0.3669551610946655,
+      "learning_rate": 4.7889726331097686e-05,
+      "loss": 0.2813,
+      "step": 4525
+    },
+    {
+      "epoch": 7.072599531615925,
+      "grad_norm": 0.3677949905395508,
+      "learning_rate": 4.7657166019440614e-05,
+      "loss": 0.2761,
+      "step": 4530
+    },
+    {
+      "epoch": 7.08040593286495,
+      "grad_norm": 0.39766794443130493,
+      "learning_rate": 4.742499497851478e-05,
+      "loss": 0.2888,
+      "step": 4535
+    },
+    {
+      "epoch": 7.088212334113973,
+      "grad_norm": 0.39841169118881226,
+      "learning_rate": 4.7193214934963206e-05,
+      "loss": 0.2874,
+      "step": 4540
+    },
+    {
+      "epoch": 7.096018735362998,
+      "grad_norm": 0.37418073415756226,
+      "learning_rate": 4.696182761252112e-05,
+      "loss": 0.2794,
+      "step": 4545
+    },
+    {
+      "epoch": 7.103825136612022,
+      "grad_norm": 0.3763929307460785,
+      "learning_rate": 4.6730834732003104e-05,
+      "loss": 0.2851,
+      "step": 4550
+    },
+    {
+      "epoch": 7.111631537861046,
+      "grad_norm": 0.39079606533050537,
+      "learning_rate": 4.6500238011290295e-05,
+      "loss": 0.2804,
+      "step": 4555
+    },
+    {
+      "epoch": 7.11943793911007,
+      "grad_norm": 0.3997804522514343,
+      "learning_rate": 4.6270039165317605e-05,
+      "loss": 0.2827,
+      "step": 4560
+    },
+    {
+      "epoch": 7.127244340359095,
+      "grad_norm": 0.3801449239253998,
+      "learning_rate": 4.604023990606105e-05,
+      "loss": 0.2845,
+      "step": 4565
+    },
+    {
+      "epoch": 7.135050741608119,
+      "grad_norm": 0.3795543611049652,
+      "learning_rate": 4.5810841942524864e-05,
+      "loss": 0.2793,
+      "step": 4570
+    },
+    {
+      "epoch": 7.142857142857143,
+      "grad_norm": 0.4058801829814911,
+      "learning_rate": 4.5581846980728794e-05,
+      "loss": 0.2844,
+      "step": 4575
+    },
+    {
+      "epoch": 7.150663544106167,
+      "grad_norm": 0.38217005133628845,
+      "learning_rate": 4.535325672369567e-05,
+      "loss": 0.2879,
+      "step": 4580
+    },
+    {
+      "epoch": 7.158469945355192,
+      "grad_norm": 0.4147985279560089,
+      "learning_rate": 4.512507287143842e-05,
+      "loss": 0.2829,
+      "step": 4585
+    },
+    {
+      "epoch": 7.166276346604215,
+      "grad_norm": 0.3910820484161377,
+      "learning_rate": 4.4897297120947624e-05,
+      "loss": 0.2898,
+      "step": 4590
+    },
+    {
+      "epoch": 7.17408274785324,
+      "grad_norm": 0.3985479176044464,
+      "learning_rate": 4.466993116617878e-05,
+      "loss": 0.2851,
+      "step": 4595
+    },
+    {
+      "epoch": 7.181889149102264,
+      "grad_norm": 0.3941868245601654,
+      "learning_rate": 4.444297669803981e-05,
+      "loss": 0.2828,
+      "step": 4600
+    },
+    {
+      "epoch": 7.1896955503512885,
+      "grad_norm": 0.3968215882778168,
+      "learning_rate": 4.4216435404378355e-05,
+      "loss": 0.2917,
+      "step": 4605
+    },
+    {
+      "epoch": 7.197501951600312,
+      "grad_norm": 0.3764592707157135,
+      "learning_rate": 4.399030896996945e-05,
+      "loss": 0.2811,
+      "step": 4610
+    },
+    {
+      "epoch": 7.205308352849336,
+      "grad_norm": 0.3845047056674957,
+      "learning_rate": 4.3764599076502696e-05,
+      "loss": 0.2765,
+      "step": 4615
+    },
+    {
+      "epoch": 7.213114754098361,
+      "grad_norm": 0.39302965998649597,
+      "learning_rate": 4.353930740256996e-05,
+      "loss": 0.2839,
+      "step": 4620
+    },
+    {
+      "epoch": 7.220921155347384,
+      "grad_norm": 0.38965940475463867,
+      "learning_rate": 4.331443562365285e-05,
+      "loss": 0.2834,
+      "step": 4625
+    },
+    {
+      "epoch": 7.228727556596409,
+      "grad_norm": 0.36734485626220703,
+      "learning_rate": 4.308998541211015e-05,
+      "loss": 0.2846,
+      "step": 4630
+    },
+    {
+      "epoch": 7.236533957845433,
+      "grad_norm": 0.41275784373283386,
+      "learning_rate": 4.286595843716569e-05,
+      "loss": 0.2903,
+      "step": 4635
+    },
+    {
+      "epoch": 7.2443403590944575,
+      "grad_norm": 0.38159552216529846,
+      "learning_rate": 4.264235636489542e-05,
+      "loss": 0.288,
+      "step": 4640
+    },
+    {
+      "epoch": 7.252146760343481,
+      "grad_norm": 0.38839638233184814,
+      "learning_rate": 4.241918085821547e-05,
+      "loss": 0.2831,
+      "step": 4645
+    },
+    {
+      "epoch": 7.259953161592506,
+      "grad_norm": 0.42755383253097534,
+      "learning_rate": 4.219643357686967e-05,
+      "loss": 0.2908,
+      "step": 4650
+    },
+    {
+      "epoch": 7.26775956284153,
+      "grad_norm": 0.43094560503959656,
+      "learning_rate": 4.19741161774171e-05,
+      "loss": 0.2841,
+      "step": 4655
+    },
+    {
+      "epoch": 7.275565964090554,
+      "grad_norm": 0.3952085077762604,
+      "learning_rate": 4.17522303132198e-05,
+      "loss": 0.2858,
+      "step": 4660
+    },
+    {
+      "epoch": 7.283372365339578,
+      "grad_norm": 0.401659220457077,
+      "learning_rate": 4.153077763443055e-05,
+      "loss": 0.2801,
+      "step": 4665
+    },
+    {
+      "epoch": 7.291178766588603,
+      "grad_norm": 0.3862559199333191,
+      "learning_rate": 4.1309759787980565e-05,
+      "loss": 0.2818,
+      "step": 4670
+    },
+    {
+      "epoch": 7.2989851678376265,
+      "grad_norm": 0.39351990818977356,
+      "learning_rate": 4.1089178417567164e-05,
+      "loss": 0.2916,
+      "step": 4675
+    },
+    {
+      "epoch": 7.306791569086651,
+      "grad_norm": 0.39333629608154297,
+      "learning_rate": 4.086903516364179e-05,
+      "loss": 0.2885,
+      "step": 4680
+    },
+    {
+      "epoch": 7.314597970335675,
+      "grad_norm": 0.40127068758010864,
+      "learning_rate": 4.064933166339737e-05,
+      "loss": 0.2892,
+      "step": 4685
+    },
+    {
+      "epoch": 7.3224043715847,
+      "grad_norm": 0.4141682982444763,
+      "learning_rate": 4.0430069550756665e-05,
+      "loss": 0.2823,
+      "step": 4690
+    },
+    {
+      "epoch": 7.330210772833723,
+      "grad_norm": 0.39413705468177795,
+      "learning_rate": 4.021125045635973e-05,
+      "loss": 0.2775,
+      "step": 4695
+    },
+    {
+      "epoch": 7.338017174082748,
+      "grad_norm": 0.3831061124801636,
+      "learning_rate": 3.999287600755192e-05,
+      "loss": 0.2867,
+      "step": 4700
+    },
+    {
+      "epoch": 7.345823575331772,
+      "grad_norm": 0.3949141204357147,
+      "learning_rate": 3.977494782837182e-05,
+      "loss": 0.2895,
+      "step": 4705
+    },
+    {
+      "epoch": 7.353629976580796,
+      "grad_norm": 0.41716331243515015,
+      "learning_rate": 3.9557467539539115e-05,
+      "loss": 0.2857,
+      "step": 4710
+    },
+    {
+      "epoch": 7.36143637782982,
+      "grad_norm": 0.385775625705719,
+      "learning_rate": 3.9340436758442536e-05,
+      "loss": 0.2819,
+      "step": 4715
+    },
+    {
+      "epoch": 7.369242779078845,
+      "grad_norm": 0.40228816866874695,
+      "learning_rate": 3.9123857099127936e-05,
+      "loss": 0.2891,
+      "step": 4720
+    },
+    {
+      "epoch": 7.377049180327869,
+      "grad_norm": 0.38272202014923096,
+      "learning_rate": 3.8907730172286124e-05,
+      "loss": 0.284,
+      "step": 4725
+    },
+    {
+      "epoch": 7.384855581576893,
+      "grad_norm": 0.39481300115585327,
+      "learning_rate": 3.8692057585240905e-05,
+      "loss": 0.2822,
+      "step": 4730
+    },
+    {
+      "epoch": 7.392661982825917,
+      "grad_norm": 0.3933485150337219,
+      "learning_rate": 3.847684094193733e-05,
+      "loss": 0.294,
+      "step": 4735
+    },
+    {
+      "epoch": 7.400468384074942,
+      "grad_norm": 0.542770504951477,
+      "learning_rate": 3.826208184292952e-05,
+      "loss": 0.2887,
+      "step": 4740
+    },
+    {
+      "epoch": 7.408274785323965,
+      "grad_norm": 0.4088205397129059,
+      "learning_rate": 3.804778188536887e-05,
+      "loss": 0.2891,
+      "step": 4745
+    },
+    {
+      "epoch": 7.41608118657299,
+      "grad_norm": 0.3968678116798401,
+      "learning_rate": 3.783394266299228e-05,
+      "loss": 0.2853,
+      "step": 4750
+    },
+    {
+      "epoch": 7.423887587822014,
+      "grad_norm": 0.41700446605682373,
+      "learning_rate": 3.7620565766109975e-05,
+      "loss": 0.2903,
+      "step": 4755
+    },
+    {
+      "epoch": 7.4316939890710385,
+      "grad_norm": 0.399271160364151,
+      "learning_rate": 3.7407652781594095e-05,
+      "loss": 0.289,
+      "step": 4760
+    },
+    {
+      "epoch": 7.439500390320062,
+      "grad_norm": 0.3900463879108429,
+      "learning_rate": 3.719520529286659e-05,
+      "loss": 0.2846,
+      "step": 4765
+    },
+    {
+      "epoch": 7.447306791569087,
+      "grad_norm": 0.43415582180023193,
+      "learning_rate": 3.698322487988755e-05,
+      "loss": 0.2898,
+      "step": 4770
+    },
+    {
+      "epoch": 7.455113192818111,
+      "grad_norm": 0.38591185212135315,
+      "learning_rate": 3.677171311914346e-05,
+      "loss": 0.2832,
+      "step": 4775
+    },
+    {
+      "epoch": 7.462919594067135,
+      "grad_norm": 0.3995843827724457,
+      "learning_rate": 3.6560671583635467e-05,
+      "loss": 0.2904,
+      "step": 4780
+    },
+    {
+      "epoch": 7.470725995316159,
+      "grad_norm": 0.4198484718799591,
+      "learning_rate": 3.6350101842867645e-05,
+      "loss": 0.2769,
+      "step": 4785
+    },
+    {
+      "epoch": 7.478532396565184,
+      "grad_norm": 0.39973580837249756,
+      "learning_rate": 3.614000546283547e-05,
+      "loss": 0.2867,
+      "step": 4790
+    },
+    {
+      "epoch": 7.4863387978142075,
+      "grad_norm": 0.37141865491867065,
+      "learning_rate": 3.593038400601395e-05,
+      "loss": 0.2834,
+      "step": 4795
+    },
+    {
+      "epoch": 7.494145199063232,
+      "grad_norm": 0.4066362679004669,
+      "learning_rate": 3.5721239031346066e-05,
+      "loss": 0.2847,
+      "step": 4800
+    },
+    {
+      "epoch": 7.501951600312256,
+      "grad_norm": 0.40516340732574463,
+      "learning_rate": 3.55125720942314e-05,
+      "loss": 0.2854,
+      "step": 4805
+    },
+    {
+      "epoch": 7.509758001561281,
+      "grad_norm": 0.428204208612442,
+      "learning_rate": 3.530438474651428e-05,
+      "loss": 0.2816,
+      "step": 4810
+    },
+    {
+      "epoch": 7.517564402810304,
+      "grad_norm": 0.4114013612270355,
+      "learning_rate": 3.509667853647235e-05,
+      "loss": 0.2783,
+      "step": 4815
+    },
+    {
+      "epoch": 7.525370804059329,
+      "grad_norm": 0.4049510955810547,
+      "learning_rate": 3.4889455008805106e-05,
+      "loss": 0.2904,
+      "step": 4820
+    },
+    {
+      "epoch": 7.533177205308353,
+      "grad_norm": 0.4051942825317383,
+      "learning_rate": 3.468271570462235e-05,
+      "loss": 0.2893,
+      "step": 4825
+    },
+    {
+      "epoch": 7.540983606557377,
+      "grad_norm": 0.4027831554412842,
+      "learning_rate": 3.447646216143268e-05,
+      "loss": 0.2903,
+      "step": 4830
+    },
+    {
+      "epoch": 7.548790007806401,
+      "grad_norm": 0.3996686339378357,
+      "learning_rate": 3.427069591313226e-05,
+      "loss": 0.2823,
+      "step": 4835
+    },
+    {
+      "epoch": 7.556596409055425,
+      "grad_norm": 0.40199506282806396,
+      "learning_rate": 3.406541848999312e-05,
+      "loss": 0.2817,
+      "step": 4840
+    },
+    {
+      "epoch": 7.56440281030445,
+      "grad_norm": 0.4444729685783386,
+      "learning_rate": 3.3860631418652e-05,
+      "loss": 0.2859,
+      "step": 4845
+    },
+    {
+      "epoch": 7.572209211553474,
+      "grad_norm": 0.40995243191719055,
+      "learning_rate": 3.365633622209891e-05,
+      "loss": 0.2869,
+      "step": 4850
+    },
+    {
+      "epoch": 7.580015612802498,
+      "grad_norm": 0.4094167649745941,
+      "learning_rate": 3.345253441966579e-05,
+      "loss": 0.2862,
+      "step": 4855
+    },
+    {
+      "epoch": 7.587822014051522,
+      "grad_norm": 0.38202327489852905,
+      "learning_rate": 3.324922752701528e-05,
+      "loss": 0.2842,
+      "step": 4860
+    },
+    {
+      "epoch": 7.595628415300546,
+      "grad_norm": 0.4048561155796051,
+      "learning_rate": 3.3046417056129366e-05,
+      "loss": 0.2829,
+      "step": 4865
+    },
+    {
+      "epoch": 7.603434816549571,
+      "grad_norm": 0.3873648941516876,
+      "learning_rate": 3.2844104515298155e-05,
+      "loss": 0.2837,
+      "step": 4870
+    },
+    {
+      "epoch": 7.611241217798595,
+      "grad_norm": 0.39025747776031494,
+      "learning_rate": 3.2642291409108775e-05,
+      "loss": 0.2777,
+      "step": 4875
+    },
+    {
+      "epoch": 7.619047619047619,
+      "grad_norm": 0.39962899684906006,
+      "learning_rate": 3.244097923843398e-05,
+      "loss": 0.2897,
+      "step": 4880
+    },
+    {
+      "epoch": 7.626854020296643,
+      "grad_norm": 0.39887768030166626,
+      "learning_rate": 3.2240169500421135e-05,
+      "loss": 0.2777,
+      "step": 4885
+    },
+    {
+      "epoch": 7.634660421545668,
+      "grad_norm": 0.42155277729034424,
+      "learning_rate": 3.2039863688481055e-05,
+      "loss": 0.2882,
+      "step": 4890
+    },
+    {
+      "epoch": 7.642466822794692,
+      "grad_norm": 0.40622079372406006,
+      "learning_rate": 3.184006329227684e-05,
+      "loss": 0.292,
+      "step": 4895
+    },
+    {
+      "epoch": 7.6502732240437155,
+      "grad_norm": 0.41190096735954285,
+      "learning_rate": 3.164076979771287e-05,
+      "loss": 0.283,
+      "step": 4900
+    },
+    {
+      "epoch": 7.65807962529274,
+      "grad_norm": 0.41508612036705017,
+      "learning_rate": 3.144198468692379e-05,
+      "loss": 0.279,
+      "step": 4905
+    },
+    {
+      "epoch": 7.665886026541764,
+      "grad_norm": 0.39837446808815,
+      "learning_rate": 3.1243709438263255e-05,
+      "loss": 0.2843,
+      "step": 4910
+    },
+    {
+      "epoch": 7.6736924277907885,
+      "grad_norm": 0.4004950523376465,
+      "learning_rate": 3.104594552629331e-05,
+      "loss": 0.2866,
+      "step": 4915
+    },
+    {
+      "epoch": 7.681498829039812,
+      "grad_norm": 0.3863111436367035,
+      "learning_rate": 3.0848694421773075e-05,
+      "loss": 0.2803,
+      "step": 4920
+    },
+    {
+      "epoch": 7.689305230288837,
+      "grad_norm": 0.39881014823913574,
+      "learning_rate": 3.065195759164797e-05,
+      "loss": 0.2832,
+      "step": 4925
+    },
+    {
+      "epoch": 7.697111631537861,
+      "grad_norm": 0.3900108337402344,
+      "learning_rate": 3.0455736499038845e-05,
+      "loss": 0.2955,
+      "step": 4930
+    },
+    {
+      "epoch": 7.704918032786885,
+      "grad_norm": 0.4136905372142792,
+      "learning_rate": 3.0260032603230982e-05,
+      "loss": 0.2938,
+      "step": 4935
+    },
+    {
+      "epoch": 7.712724434035909,
+      "grad_norm": 0.4050344228744507,
+      "learning_rate": 3.0064847359663284e-05,
+      "loss": 0.2903,
+      "step": 4940
+    },
+    {
+      "epoch": 7.720530835284934,
+      "grad_norm": 0.4190680980682373,
+      "learning_rate": 2.9870182219917564e-05,
+      "loss": 0.2944,
+      "step": 4945
+    },
+    {
+      "epoch": 7.7283372365339575,
+      "grad_norm": 0.39599907398223877,
+      "learning_rate": 2.9676038631707593e-05,
+      "loss": 0.2911,
+      "step": 4950
+    },
+    {
+      "epoch": 7.736143637782982,
+      "grad_norm": 0.39728233218193054,
+      "learning_rate": 2.9482418038868275e-05,
+      "loss": 0.2861,
+      "step": 4955
+    },
+    {
+      "epoch": 7.743950039032006,
+      "grad_norm": 0.39015698432922363,
+      "learning_rate": 2.9289321881345254e-05,
+      "loss": 0.2846,
+      "step": 4960
+    },
+    {
+      "epoch": 7.751756440281031,
+      "grad_norm": 0.394378662109375,
+      "learning_rate": 2.9096751595183823e-05,
+      "loss": 0.2879,
+      "step": 4965
+    },
+    {
+      "epoch": 7.759562841530054,
+      "grad_norm": 0.39528360962867737,
+      "learning_rate": 2.8904708612518404e-05,
+      "loss": 0.2899,
+      "step": 4970
+    },
+    {
+      "epoch": 7.767369242779079,
+      "grad_norm": 0.41304799914360046,
+      "learning_rate": 2.8713194361562036e-05,
+      "loss": 0.2898,
+      "step": 4975
+    },
+    {
+      "epoch": 7.775175644028103,
+      "grad_norm": 0.4061037003993988,
+      "learning_rate": 2.8522210266595384e-05,
+      "loss": 0.2901,
+      "step": 4980
+    },
+    {
+      "epoch": 7.7829820452771274,
+      "grad_norm": 0.42109227180480957,
+      "learning_rate": 2.833175774795651e-05,
+      "loss": 0.2868,
+      "step": 4985
+    },
+    {
+      "epoch": 7.790788446526151,
+      "grad_norm": 0.38924074172973633,
+      "learning_rate": 2.814183822203019e-05,
+      "loss": 0.2887,
+      "step": 4990
+    },
+    {
+      "epoch": 7.798594847775176,
+      "grad_norm": 0.4080544114112854,
+      "learning_rate": 2.795245310123732e-05,
+      "loss": 0.2884,
+      "step": 4995
+    },
+    {
+      "epoch": 7.8064012490242,
+      "grad_norm": 0.39689868688583374,
+      "learning_rate": 2.776360379402445e-05,
+      "loss": 0.2859,
+      "step": 5000
+    },
+    {
+      "epoch": 7.814207650273224,
+      "grad_norm": 0.40437766909599304,
+      "learning_rate": 2.7575291704853323e-05,
+      "loss": 0.287,
+      "step": 5005
+    },
+    {
+      "epoch": 7.822014051522248,
+      "grad_norm": 0.4085007905960083,
+      "learning_rate": 2.738751823419041e-05,
+      "loss": 0.2875,
+      "step": 5010
+    },
+    {
+      "epoch": 7.829820452771273,
+      "grad_norm": 0.3995596468448639,
+      "learning_rate": 2.720028477849652e-05,
+      "loss": 0.2818,
+      "step": 5015
+    },
+    {
+      "epoch": 7.8376268540202965,
+      "grad_norm": 0.3908860683441162,
+      "learning_rate": 2.7013592730216465e-05,
+      "loss": 0.283,
+      "step": 5020
+    },
+    {
+      "epoch": 7.845433255269321,
+      "grad_norm": 0.39162319898605347,
+      "learning_rate": 2.6827443477768454e-05,
+      "loss": 0.2833,
+      "step": 5025
+    },
+    {
+      "epoch": 7.853239656518345,
+      "grad_norm": 0.3898274600505829,
+      "learning_rate": 2.664183840553417e-05,
+      "loss": 0.2903,
+      "step": 5030
+    },
+    {
+      "epoch": 7.8610460577673695,
+      "grad_norm": 0.4035649597644806,
+      "learning_rate": 2.6456778893848144e-05,
+      "loss": 0.2826,
+      "step": 5035
+    },
+    {
+      "epoch": 7.868852459016393,
+      "grad_norm": 0.4092506170272827,
+      "learning_rate": 2.6272266318987603e-05,
+      "loss": 0.284,
+      "step": 5040
+    },
+    {
+      "epoch": 7.876658860265418,
+      "grad_norm": 0.40052658319473267,
+      "learning_rate": 2.6088302053162272e-05,
+      "loss": 0.2894,
+      "step": 5045
+    },
+    {
+      "epoch": 7.884465261514442,
+      "grad_norm": 0.3952447175979614,
+      "learning_rate": 2.5904887464504114e-05,
+      "loss": 0.281,
+      "step": 5050
+    },
+    {
+      "epoch": 7.892271662763466,
+      "grad_norm": 0.3892611563205719,
+      "learning_rate": 2.5722023917057125e-05,
+      "loss": 0.2824,
+      "step": 5055
+    },
+    {
+      "epoch": 7.90007806401249,
+      "grad_norm": 0.3946845531463623,
+      "learning_rate": 2.5539712770767376e-05,
+      "loss": 0.2865,
+      "step": 5060
+    },
+    {
+      "epoch": 7.907884465261515,
+      "grad_norm": 0.39248526096343994,
+      "learning_rate": 2.535795538147262e-05,
+      "loss": 0.2863,
+      "step": 5065
+    },
+    {
+      "epoch": 7.9156908665105385,
+      "grad_norm": 0.4364306330680847,
+      "learning_rate": 2.5176753100892426e-05,
+      "loss": 0.2916,
+      "step": 5070
+    },
+    {
+      "epoch": 7.923497267759563,
+      "grad_norm": 0.3956867754459381,
+      "learning_rate": 2.4996107276618008e-05,
+      "loss": 0.2819,
+      "step": 5075
+    },
+    {
+      "epoch": 7.931303669008587,
+      "grad_norm": 0.3908245265483856,
+      "learning_rate": 2.4816019252102273e-05,
+      "loss": 0.2864,
+      "step": 5080
+    },
+    {
+      "epoch": 7.939110070257612,
+      "grad_norm": 0.3847108483314514,
+      "learning_rate": 2.46364903666498e-05,
+      "loss": 0.2937,
+      "step": 5085
+    },
+    {
+      "epoch": 7.946916471506635,
+      "grad_norm": 0.412410169839859,
+      "learning_rate": 2.445752195540687e-05,
+      "loss": 0.2795,
+      "step": 5090
+    },
+    {
+      "epoch": 7.95472287275566,
+      "grad_norm": 0.39699193835258484,
+      "learning_rate": 2.4279115349351543e-05,
+      "loss": 0.2909,
+      "step": 5095
+    },
+    {
+      "epoch": 7.962529274004684,
+      "grad_norm": 0.4221264719963074,
+      "learning_rate": 2.4101271875283817e-05,
+      "loss": 0.2862,
+      "step": 5100
+    },
+    {
+      "epoch": 7.970335675253708,
+      "grad_norm": 0.38855311274528503,
+      "learning_rate": 2.3923992855815647e-05,
+      "loss": 0.2847,
+      "step": 5105
+    },
+    {
+      "epoch": 7.978142076502732,
+      "grad_norm": 0.4506378173828125,
+      "learning_rate": 2.3747279609361196e-05,
+      "loss": 0.2904,
+      "step": 5110
+    },
+    {
+      "epoch": 7.985948477751757,
+      "grad_norm": 0.4097312390804291,
+      "learning_rate": 2.3571133450127002e-05,
+      "loss": 0.2903,
+      "step": 5115
+    },
+    {
+      "epoch": 7.993754879000781,
+      "grad_norm": 0.4036017060279846,
+      "learning_rate": 2.339555568810221e-05,
+      "loss": 0.2857,
+      "step": 5120
+    },
+    {
+      "epoch": 8.0,
+      "eval_loss": 1.1256442070007324,
+      "eval_runtime": 0.3225,
+      "eval_samples_per_second": 15.502,
+      "eval_steps_per_second": 3.1,
+      "step": 5124
+    },
+    {
+      "epoch": 8.001561280249804,
+      "grad_norm": 0.3814958333969116,
+      "learning_rate": 2.3220547629048796e-05,
+      "loss": 0.2894,
+      "step": 5125
+    },
+    {
+      "epoch": 8.009367681498828,
+      "grad_norm": 0.3972679376602173,
+      "learning_rate": 2.3046110574491985e-05,
+      "loss": 0.2791,
+      "step": 5130
+    },
+    {
+      "epoch": 8.017174082747854,
+      "grad_norm": 0.3934348225593567,
+      "learning_rate": 2.2872245821710346e-05,
+      "loss": 0.2719,
+      "step": 5135
+    },
+    {
+      "epoch": 8.024980483996877,
+      "grad_norm": 0.4403572380542755,
+      "learning_rate": 2.26989546637263e-05,
+      "loss": 0.2688,
+      "step": 5140
+    },
+    {
+      "epoch": 8.032786885245901,
+      "grad_norm": 0.43504083156585693,
+      "learning_rate": 2.2526238389296585e-05,
+      "loss": 0.2727,
+      "step": 5145
+    },
+    {
+      "epoch": 8.040593286494925,
+      "grad_norm": 0.40251481533050537,
+      "learning_rate": 2.2354098282902446e-05,
+      "loss": 0.2706,
+      "step": 5150
+    },
+    {
+      "epoch": 8.04839968774395,
+      "grad_norm": 0.3875437080860138,
+      "learning_rate": 2.218253562474023e-05,
+      "loss": 0.2684,
+      "step": 5155
+    },
+    {
+      "epoch": 8.056206088992974,
+      "grad_norm": 0.39775198698043823,
+      "learning_rate": 2.201155169071184e-05,
+      "loss": 0.2685,
+      "step": 5160
+    },
+    {
+      "epoch": 8.064012490241998,
+      "grad_norm": 0.4047357439994812,
+      "learning_rate": 2.1841147752415235e-05,
+      "loss": 0.2686,
+      "step": 5165
+    },
+    {
+      "epoch": 8.071818891491022,
+      "grad_norm": 0.4089979827404022,
+      "learning_rate": 2.1671325077134963e-05,
+      "loss": 0.2695,
+      "step": 5170
+    },
+    {
+      "epoch": 8.079625292740047,
+      "grad_norm": 0.40754222869873047,
+      "learning_rate": 2.1502084927832845e-05,
+      "loss": 0.2801,
+      "step": 5175
+    },
+    {
+      "epoch": 8.087431693989071,
+      "grad_norm": 0.4052791893482208,
+      "learning_rate": 2.1333428563138303e-05,
+      "loss": 0.2708,
+      "step": 5180
+    },
+    {
+      "epoch": 8.095238095238095,
+      "grad_norm": 0.4029444754123688,
+      "learning_rate": 2.116535723733938e-05,
+      "loss": 0.2688,
+      "step": 5185
+    },
+    {
+      "epoch": 8.103044496487119,
+      "grad_norm": 0.43849509954452515,
+      "learning_rate": 2.0997872200373116e-05,
+      "loss": 0.2749,
+      "step": 5190
+    },
+    {
+      "epoch": 8.110850897736144,
+      "grad_norm": 0.4012914001941681,
+      "learning_rate": 2.083097469781632e-05,
+      "loss": 0.2726,
+      "step": 5195
+    },
+    {
+      "epoch": 8.118657298985168,
+      "grad_norm": 0.42504939436912537,
+      "learning_rate": 2.0664665970876496e-05,
+      "loss": 0.2791,
+      "step": 5200
+    },
+    {
+      "epoch": 8.126463700234192,
+      "grad_norm": 0.41797903180122375,
+      "learning_rate": 2.0498947256382273e-05,
+      "loss": 0.2728,
+      "step": 5205
+    },
+    {
+      "epoch": 8.134270101483215,
+      "grad_norm": 0.3988747000694275,
+      "learning_rate": 2.0333819786774444e-05,
+      "loss": 0.2757,
+      "step": 5210
+    },
+    {
+      "epoch": 8.142076502732241,
+      "grad_norm": 0.4232763350009918,
+      "learning_rate": 2.0169284790096853e-05,
+      "loss": 0.2749,
+      "step": 5215
+    },
+    {
+      "epoch": 8.149882903981265,
+      "grad_norm": 0.41461604833602905,
+      "learning_rate": 2.000534348998704e-05,
+      "loss": 0.2747,
+      "step": 5220
+    },
+    {
+      "epoch": 8.157689305230289,
+      "grad_norm": 0.4166080355644226,
+      "learning_rate": 1.9841997105667275e-05,
+      "loss": 0.2721,
+      "step": 5225
+    },
+    {
+      "epoch": 8.165495706479312,
+      "grad_norm": 0.41639989614486694,
+      "learning_rate": 1.967924685193552e-05,
+      "loss": 0.2645,
+      "step": 5230
+    },
+    {
+      "epoch": 8.173302107728338,
+      "grad_norm": 0.4174750745296478,
+      "learning_rate": 1.9517093939156295e-05,
+      "loss": 0.2732,
+      "step": 5235
+    },
+    {
+      "epoch": 8.181108508977362,
+      "grad_norm": 0.4450904130935669,
+      "learning_rate": 1.9355539573251734e-05,
+      "loss": 0.2777,
+      "step": 5240
+    },
+    {
+      "epoch": 8.188914910226385,
+      "grad_norm": 0.4246358573436737,
+      "learning_rate": 1.9194584955692706e-05,
+      "loss": 0.2791,
+      "step": 5245
+    },
+    {
+      "epoch": 8.19672131147541,
+      "grad_norm": 0.4177229106426239,
+      "learning_rate": 1.903423128348959e-05,
+      "loss": 0.2743,
+      "step": 5250
+    },
+    {
+      "epoch": 8.204527712724435,
+      "grad_norm": 0.42174917459487915,
+      "learning_rate": 1.8874479749183782e-05,
+      "loss": 0.2754,
+      "step": 5255
+    },
+    {
+      "epoch": 8.212334113973458,
+      "grad_norm": 0.4163845181465149,
+      "learning_rate": 1.8715331540838487e-05,
+      "loss": 0.2718,
+      "step": 5260
+    },
+    {
+      "epoch": 8.220140515222482,
+      "grad_norm": 0.4283103346824646,
+      "learning_rate": 1.855678784203002e-05,
+      "loss": 0.2704,
+      "step": 5265
+    },
+    {
+      "epoch": 8.227946916471506,
+      "grad_norm": 0.4160711467266083,
+      "learning_rate": 1.8398849831839014e-05,
+      "loss": 0.2679,
+      "step": 5270
+    },
+    {
+      "epoch": 8.235753317720532,
+      "grad_norm": 0.4049237370491028,
+      "learning_rate": 1.824151868484164e-05,
+      "loss": 0.269,
+      "step": 5275
+    },
+    {
+      "epoch": 8.243559718969555,
+      "grad_norm": 0.4092932641506195,
+      "learning_rate": 1.808479557110081e-05,
+      "loss": 0.2725,
+      "step": 5280
+    },
+    {
+      "epoch": 8.251366120218579,
+      "grad_norm": 0.42792901396751404,
+      "learning_rate": 1.792868165615762e-05,
+      "loss": 0.2778,
+      "step": 5285
+    },
+    {
+      "epoch": 8.259172521467603,
+      "grad_norm": 0.4112985134124756,
+      "learning_rate": 1.7773178101022514e-05,
+      "loss": 0.2708,
+      "step": 5290
+    },
+    {
+      "epoch": 8.266978922716628,
+      "grad_norm": 0.4147091209888458,
+      "learning_rate": 1.7618286062166677e-05,
+      "loss": 0.2772,
+      "step": 5295
+    },
+    {
+      "epoch": 8.274785323965652,
+      "grad_norm": 0.4274720847606659,
+      "learning_rate": 1.7464006691513623e-05,
+      "loss": 0.273,
+      "step": 5300
+    },
+    {
+      "epoch": 8.282591725214676,
+      "grad_norm": 0.40364208817481995,
+      "learning_rate": 1.7310341136430385e-05,
+      "loss": 0.2778,
+      "step": 5305
+    },
+    {
+      "epoch": 8.2903981264637,
+      "grad_norm": 0.42018428444862366,
+      "learning_rate": 1.7157290539719106e-05,
+      "loss": 0.2743,
+      "step": 5310
+    },
+    {
+      "epoch": 8.298204527712725,
+      "grad_norm": 0.4169312119483948,
+      "learning_rate": 1.700485603960853e-05,
+      "loss": 0.275,
+      "step": 5315
+    },
+    {
+      "epoch": 8.306010928961749,
+      "grad_norm": 0.40692824125289917,
+      "learning_rate": 1.6853038769745467e-05,
+      "loss": 0.2683,
+      "step": 5320
+    },
+    {
+      "epoch": 8.313817330210773,
+      "grad_norm": 0.41851934790611267,
+      "learning_rate": 1.6701839859186542e-05,
+      "loss": 0.2677,
+      "step": 5325
+    },
+    {
+      "epoch": 8.321623731459797,
+      "grad_norm": 0.4233751595020294,
+      "learning_rate": 1.655126043238957e-05,
+      "loss": 0.27,
+      "step": 5330
+    },
+    {
+      "epoch": 8.329430132708822,
+      "grad_norm": 0.4326801002025604,
+      "learning_rate": 1.6401301609205335e-05,
+      "loss": 0.2774,
+      "step": 5335
+    },
+    {
+      "epoch": 8.337236533957846,
+      "grad_norm": 0.42542752623558044,
+      "learning_rate": 1.6251964504869222e-05,
+      "loss": 0.2692,
+      "step": 5340
+    },
+    {
+      "epoch": 8.34504293520687,
+      "grad_norm": 0.4217033386230469,
+      "learning_rate": 1.6103250229992927e-05,
+      "loss": 0.2804,
+      "step": 5345
+    },
+    {
+      "epoch": 8.352849336455893,
+      "grad_norm": 0.4149416387081146,
+      "learning_rate": 1.595515989055618e-05,
+      "loss": 0.2732,
+      "step": 5350
+    },
+    {
+      "epoch": 8.360655737704919,
+      "grad_norm": 0.42739465832710266,
+      "learning_rate": 1.5807694587898648e-05,
+      "loss": 0.2782,
+      "step": 5355
+    },
+    {
+      "epoch": 8.368462138953943,
+      "grad_norm": 0.42222627997398376,
+      "learning_rate": 1.566085541871145e-05,
+      "loss": 0.2699,
+      "step": 5360
+    },
+    {
+      "epoch": 8.376268540202966,
+      "grad_norm": 0.410786509513855,
+      "learning_rate": 1.551464347502929e-05,
+      "loss": 0.2743,
+      "step": 5365
+    },
+    {
+      "epoch": 8.38407494145199,
+      "grad_norm": 0.42929258942604065,
+      "learning_rate": 1.5369059844222278e-05,
+      "loss": 0.2785,
+      "step": 5370
+    },
+    {
+      "epoch": 8.391881342701016,
+      "grad_norm": 0.4153624475002289,
+      "learning_rate": 1.5224105608987704e-05,
+      "loss": 0.2687,
+      "step": 5375
+    },
+    {
+      "epoch": 8.39968774395004,
+      "grad_norm": 0.3873469829559326,
+      "learning_rate": 1.5079781847342123e-05,
+      "loss": 0.2678,
+      "step": 5380
+    },
+    {
+      "epoch": 8.407494145199063,
+      "grad_norm": 0.43077465891838074,
+      "learning_rate": 1.4936089632613281e-05,
+      "loss": 0.2765,
+      "step": 5385
+    },
+    {
+      "epoch": 8.415300546448087,
+      "grad_norm": 0.4113737940788269,
+      "learning_rate": 1.4793030033432142e-05,
+      "loss": 0.2688,
+      "step": 5390
+    },
+    {
+      "epoch": 8.42310694769711,
+      "grad_norm": 0.40002962946891785,
+      "learning_rate": 1.4650604113724953e-05,
+      "loss": 0.2693,
+      "step": 5395
+    },
+    {
+      "epoch": 8.430913348946136,
+      "grad_norm": 0.4060341417789459,
+      "learning_rate": 1.4508812932705363e-05,
+      "loss": 0.2754,
+      "step": 5400
+    },
+    {
+      "epoch": 8.43871975019516,
+      "grad_norm": 0.43747276067733765,
+      "learning_rate": 1.4367657544866375e-05,
+      "loss": 0.273,
+      "step": 5405
+    },
+    {
+      "epoch": 8.446526151444184,
+      "grad_norm": 0.41913843154907227,
+      "learning_rate": 1.42271389999728e-05,
+      "loss": 0.2756,
+      "step": 5410
+    },
+    {
+      "epoch": 8.454332552693208,
+      "grad_norm": 0.39348214864730835,
+      "learning_rate": 1.4087258343053167e-05,
+      "loss": 0.2701,
+      "step": 5415
+    },
+    {
+      "epoch": 8.462138953942233,
+      "grad_norm": 0.4029614329338074,
+      "learning_rate": 1.3948016614392112e-05,
+      "loss": 0.269,
+      "step": 5420
+    },
+    {
+      "epoch": 8.469945355191257,
+      "grad_norm": 0.45508867502212524,
+      "learning_rate": 1.3809414849522584e-05,
+      "loss": 0.2738,
+      "step": 5425
+    },
+    {
+      "epoch": 8.47775175644028,
+      "grad_norm": 0.4153214395046234,
+      "learning_rate": 1.367145407921817e-05,
+      "loss": 0.2713,
+      "step": 5430
+    },
+    {
+      "epoch": 8.485558157689304,
+      "grad_norm": 0.4336087703704834,
+      "learning_rate": 1.3534135329485376e-05,
+      "loss": 0.2786,
+      "step": 5435
+    },
+    {
+      "epoch": 8.49336455893833,
+      "grad_norm": 0.4236217141151428,
+      "learning_rate": 1.339745962155613e-05,
+      "loss": 0.277,
+      "step": 5440
+    },
+    {
+      "epoch": 8.501170960187354,
+      "grad_norm": 0.43844056129455566,
+      "learning_rate": 1.3261427971880025e-05,
+      "loss": 0.2709,
+      "step": 5445
+    },
+    {
+      "epoch": 8.508977361436378,
+      "grad_norm": 0.43588918447494507,
+      "learning_rate": 1.3126041392116772e-05,
+      "loss": 0.2759,
+      "step": 5450
+    },
+    {
+      "epoch": 8.516783762685401,
+      "grad_norm": 0.4094368517398834,
+      "learning_rate": 1.2991300889128866e-05,
+      "loss": 0.2776,
+      "step": 5455
+    },
+    {
+      "epoch": 8.524590163934427,
+      "grad_norm": 0.41443923115730286,
+      "learning_rate": 1.2857207464973875e-05,
+      "loss": 0.2746,
+      "step": 5460
+    },
+    {
+      "epoch": 8.53239656518345,
+      "grad_norm": 0.44371187686920166,
+      "learning_rate": 1.2723762116897075e-05,
+      "loss": 0.2821,
+      "step": 5465
+    },
+    {
+      "epoch": 8.540202966432474,
+      "grad_norm": 0.3933749198913574,
+      "learning_rate": 1.2590965837324131e-05,
+      "loss": 0.2713,
+      "step": 5470
+    },
+    {
+      "epoch": 8.548009367681498,
+      "grad_norm": 0.4023985266685486,
+      "learning_rate": 1.2458819613853468e-05,
+      "loss": 0.2714,
+      "step": 5475
+    },
+    {
+      "epoch": 8.555815768930524,
+      "grad_norm": 0.4138771891593933,
+      "learning_rate": 1.2327324429249232e-05,
+      "loss": 0.2749,
+      "step": 5480
+    },
+    {
+      "epoch": 8.563622170179547,
+      "grad_norm": 0.4046929180622101,
+      "learning_rate": 1.2196481261433735e-05,
+      "loss": 0.2732,
+      "step": 5485
+    },
+    {
+      "epoch": 8.571428571428571,
+      "grad_norm": 0.4175087511539459,
+      "learning_rate": 1.2066291083480296e-05,
+      "loss": 0.2691,
+      "step": 5490
+    },
+    {
+      "epoch": 8.579234972677595,
+      "grad_norm": 0.4213341176509857,
+      "learning_rate": 1.1936754863606014e-05,
+      "loss": 0.2769,
+      "step": 5495
+    },
+    {
+      "epoch": 8.58704137392662,
+      "grad_norm": 0.43556031584739685,
+      "learning_rate": 1.1807873565164506e-05,
+      "loss": 0.2712,
+      "step": 5500
+    },
+    {
+      "epoch": 8.594847775175644,
+      "grad_norm": 0.39848992228507996,
+      "learning_rate": 1.1679648146638766e-05,
+      "loss": 0.2718,
+      "step": 5505
+    },
+    {
+      "epoch": 8.602654176424668,
+      "grad_norm": 0.4068874418735504,
+      "learning_rate": 1.155207956163411e-05,
+      "loss": 0.2714,
+      "step": 5510
+    },
+    {
+      "epoch": 8.610460577673692,
+      "grad_norm": 0.41433241963386536,
+      "learning_rate": 1.1425168758870964e-05,
+      "loss": 0.275,
+      "step": 5515
+    },
+    {
+      "epoch": 8.618266978922717,
+      "grad_norm": 0.42830371856689453,
+      "learning_rate": 1.129891668217783e-05,
+      "loss": 0.2743,
+      "step": 5520
+    },
+    {
+      "epoch": 8.626073380171741,
+      "grad_norm": 0.4217815697193146,
+      "learning_rate": 1.1173324270484397e-05,
+      "loss": 0.2756,
+      "step": 5525
+    },
+    {
+      "epoch": 8.633879781420765,
+      "grad_norm": 0.3956458270549774,
+      "learning_rate": 1.1048392457814405e-05,
+      "loss": 0.2743,
+      "step": 5530
+    },
+    {
+      "epoch": 8.641686182669789,
+      "grad_norm": 0.42598819732666016,
+      "learning_rate": 1.0924122173278751e-05,
+      "loss": 0.2741,
+      "step": 5535
+    },
+    {
+      "epoch": 8.649492583918814,
+      "grad_norm": 0.44088292121887207,
+      "learning_rate": 1.080051434106859e-05,
+      "loss": 0.2784,
+      "step": 5540
+    },
+    {
+      "epoch": 8.657298985167838,
+      "grad_norm": 0.40059834718704224,
+      "learning_rate": 1.067756988044848e-05,
+      "loss": 0.2736,
+      "step": 5545
+    },
+    {
+      "epoch": 8.665105386416862,
+      "grad_norm": 0.41283807158470154,
+      "learning_rate": 1.0555289705749483e-05,
+      "loss": 0.2781,
+      "step": 5550
+    },
+    {
+      "epoch": 8.672911787665885,
+      "grad_norm": 0.40279173851013184,
+      "learning_rate": 1.0433674726362442e-05,
+      "loss": 0.2689,
+      "step": 5555
+    },
+    {
+      "epoch": 8.680718188914911,
+      "grad_norm": 0.4150051176548004,
+      "learning_rate": 1.0312725846731175e-05,
+      "loss": 0.2733,
+      "step": 5560
+    },
+    {
+      "epoch": 8.688524590163935,
+      "grad_norm": 0.4160909950733185,
+      "learning_rate": 1.019244396634571e-05,
+      "loss": 0.272,
+      "step": 5565
+    },
+    {
+      "epoch": 8.696330991412959,
+      "grad_norm": 0.4090689718723297,
+      "learning_rate": 1.0072829979735699e-05,
+      "loss": 0.2779,
+      "step": 5570
+    },
+    {
+      "epoch": 8.704137392661982,
+      "grad_norm": 0.43035534024238586,
+      "learning_rate": 9.953884776463652e-06,
+      "loss": 0.2799,
+      "step": 5575
+    },
+    {
+      "epoch": 8.711943793911008,
+      "grad_norm": 0.4087963402271271,
+      "learning_rate": 9.835609241118404e-06,
+      "loss": 0.2724,
+      "step": 5580
+    },
+    {
+      "epoch": 8.719750195160032,
+      "grad_norm": 0.4282284677028656,
+      "learning_rate": 9.71800425330851e-06,
+      "loss": 0.2766,
+      "step": 5585
+    },
+    {
+      "epoch": 8.727556596409055,
+      "grad_norm": 0.39750391244888306,
+      "learning_rate": 9.601070687655667e-06,
+      "loss": 0.2751,
+      "step": 5590
+    },
+    {
+      "epoch": 8.735362997658079,
+      "grad_norm": 0.43247750401496887,
+      "learning_rate": 9.48480941378831e-06,
+      "loss": 0.2701,
+      "step": 5595
+    },
+    {
+      "epoch": 8.743169398907105,
+      "grad_norm": 0.42617279291152954,
+      "learning_rate": 9.369221296335006e-06,
+      "loss": 0.2774,
+      "step": 5600
+    },
+    {
+      "epoch": 8.750975800156128,
+      "grad_norm": 0.4098498225212097,
+      "learning_rate": 9.254307194918144e-06,
+      "loss": 0.2779,
+      "step": 5605
+    },
+    {
+      "epoch": 8.758782201405152,
+      "grad_norm": 0.4429432153701782,
+      "learning_rate": 9.140067964147447e-06,
+      "loss": 0.2809,
+      "step": 5610
+    },
+    {
+      "epoch": 8.766588602654176,
+      "grad_norm": 0.42042461037635803,
+      "learning_rate": 9.026504453613726e-06,
+      "loss": 0.2755,
+      "step": 5615
+    },
+    {
+      "epoch": 8.774395003903201,
+      "grad_norm": 0.41905051469802856,
+      "learning_rate": 8.91361750788241e-06,
+      "loss": 0.2679,
+      "step": 5620
+    },
+    {
+      "epoch": 8.782201405152225,
+      "grad_norm": 0.4195885956287384,
+      "learning_rate": 8.801407966487486e-06,
+      "loss": 0.2734,
+      "step": 5625
+    },
+    {
+      "epoch": 8.790007806401249,
+      "grad_norm": 0.4253963232040405,
+      "learning_rate": 8.689876663924957e-06,
+      "loss": 0.2674,
+      "step": 5630
+    },
+    {
+      "epoch": 8.797814207650273,
+      "grad_norm": 0.43018728494644165,
+      "learning_rate": 8.579024429646932e-06,
+      "loss": 0.2762,
+      "step": 5635
+    },
+    {
+      "epoch": 8.805620608899297,
+      "grad_norm": 0.41742637753486633,
+      "learning_rate": 8.46885208805529e-06,
+      "loss": 0.2768,
+      "step": 5640
+    },
+    {
+      "epoch": 8.813427010148322,
+      "grad_norm": 0.41150280833244324,
+      "learning_rate": 8.359360458495557e-06,
+      "loss": 0.2777,
+      "step": 5645
+    },
+    {
+      "epoch": 8.821233411397346,
+      "grad_norm": 0.4265673756599426,
+      "learning_rate": 8.250550355250875e-06,
+      "loss": 0.2743,
+      "step": 5650
+    },
+    {
+      "epoch": 8.82903981264637,
+      "grad_norm": 0.4104432463645935,
+      "learning_rate": 8.142422587535903e-06,
+      "loss": 0.279,
+      "step": 5655
+    },
+    {
+      "epoch": 8.836846213895395,
+      "grad_norm": 0.4088299572467804,
+      "learning_rate": 8.034977959490775e-06,
+      "loss": 0.2723,
+      "step": 5660
+    },
+    {
+      "epoch": 8.844652615144419,
+      "grad_norm": 0.4326065182685852,
+      "learning_rate": 7.92821727017523e-06,
+      "loss": 0.2748,
+      "step": 5665
+    },
+    {
+      "epoch": 8.852459016393443,
+      "grad_norm": 0.4133962392807007,
+      "learning_rate": 7.822141313562547e-06,
+      "loss": 0.2699,
+      "step": 5670
+    },
+    {
+      "epoch": 8.860265417642466,
+      "grad_norm": 0.42703020572662354,
+      "learning_rate": 7.71675087853364e-06,
+      "loss": 0.2735,
+      "step": 5675
+    },
+    {
+      "epoch": 8.86807181889149,
+      "grad_norm": 0.43023622035980225,
+      "learning_rate": 7.612046748871327e-06,
+      "loss": 0.2764,
+      "step": 5680
+    },
+    {
+      "epoch": 8.875878220140516,
+      "grad_norm": 0.4223518371582031,
+      "learning_rate": 7.50802970325436e-06,
+      "loss": 0.276,
+      "step": 5685
+    },
+    {
+      "epoch": 8.88368462138954,
+      "grad_norm": 0.43024882674217224,
+      "learning_rate": 7.404700515251672e-06,
+      "loss": 0.2745,
+      "step": 5690
+    },
+    {
+      "epoch": 8.891491022638563,
+      "grad_norm": 0.4196966886520386,
+      "learning_rate": 7.30205995331672e-06,
+      "loss": 0.2791,
+      "step": 5695
+    },
+    {
+      "epoch": 8.899297423887587,
+      "grad_norm": 0.39749372005462646,
+      "learning_rate": 7.200108780781556e-06,
+      "loss": 0.2753,
+      "step": 5700
+    },
+    {
+      "epoch": 8.907103825136613,
+      "grad_norm": 0.4266413748264313,
+      "learning_rate": 7.0988477558513785e-06,
+      "loss": 0.2769,
+      "step": 5705
+    },
+    {
+      "epoch": 8.914910226385636,
+      "grad_norm": 0.4070897102355957,
+      "learning_rate": 6.998277631598793e-06,
+      "loss": 0.2757,
+      "step": 5710
+    },
+    {
+      "epoch": 8.92271662763466,
+      "grad_norm": 0.39892005920410156,
+      "learning_rate": 6.898399155958168e-06,
+      "loss": 0.2741,
+      "step": 5715
+    },
+    {
+      "epoch": 8.930523028883684,
+      "grad_norm": 0.42477500438690186,
+      "learning_rate": 6.7992130717201564e-06,
+      "loss": 0.2791,
+      "step": 5720
+    },
+    {
+      "epoch": 8.93832943013271,
+      "grad_norm": 0.44194042682647705,
+      "learning_rate": 6.700720116526116e-06,
+      "loss": 0.2786,
+      "step": 5725
+    },
+    {
+      "epoch": 8.946135831381733,
+      "grad_norm": 0.4111786484718323,
+      "learning_rate": 6.6029210228626626e-06,
+      "loss": 0.2728,
+      "step": 5730
+    },
+    {
+      "epoch": 8.953942232630757,
+      "grad_norm": 0.43993067741394043,
+      "learning_rate": 6.505816518056162e-06,
+      "loss": 0.2799,
+      "step": 5735
+    },
+    {
+      "epoch": 8.96174863387978,
+      "grad_norm": 0.413835346698761,
+      "learning_rate": 6.409407324267447e-06,
+      "loss": 0.2761,
+      "step": 5740
+    },
+    {
+      "epoch": 8.969555035128806,
+      "grad_norm": 0.4155209958553314,
+      "learning_rate": 6.313694158486228e-06,
+      "loss": 0.2726,
+      "step": 5745
+    },
+    {
+      "epoch": 8.97736143637783,
+      "grad_norm": 0.43269336223602295,
+      "learning_rate": 6.218677732526035e-06,
+      "loss": 0.2697,
+      "step": 5750
+    },
+    {
+      "epoch": 8.985167837626854,
+      "grad_norm": 0.4364638924598694,
+      "learning_rate": 6.124358753018689e-06,
+      "loss": 0.2788,
+      "step": 5755
+    },
+    {
+      "epoch": 8.992974238875878,
+      "grad_norm": 0.40509918332099915,
+      "learning_rate": 6.030737921409169e-06,
+      "loss": 0.2732,
+      "step": 5760
+    },
+    {
+      "epoch": 8.999219359875097,
+      "eval_loss": 1.1471812725067139,
+      "eval_runtime": 0.3388,
+      "eval_samples_per_second": 14.758,
+      "eval_steps_per_second": 2.952,
+      "step": 5764
+    },
+    {
+      "epoch": 9.000780640124903,
+      "grad_norm": 0.40420717000961304,
+      "learning_rate": 5.937815933950375e-06,
+      "loss": 0.2633,
+      "step": 5765
+    },
+    {
+      "epoch": 9.008587041373927,
+      "grad_norm": 0.41395288705825806,
+      "learning_rate": 5.8455934816979305e-06,
+      "loss": 0.2729,
+      "step": 5770
+    },
+    {
+      "epoch": 9.01639344262295,
+      "grad_norm": 0.4095096290111542,
+      "learning_rate": 5.7540712505050444e-06,
+      "loss": 0.265,
+      "step": 5775
+    },
+    {
+      "epoch": 9.024199843871974,
+      "grad_norm": 0.41725656390190125,
+      "learning_rate": 5.663249921017477e-06,
+      "loss": 0.2632,
+      "step": 5780
+    },
+    {
+      "epoch": 9.032006245121,
+      "grad_norm": 0.41932108998298645,
+      "learning_rate": 5.57313016866835e-06,
+      "loss": 0.2709,
+      "step": 5785
+    },
+    {
+      "epoch": 9.039812646370024,
+      "grad_norm": 0.41799718141555786,
+      "learning_rate": 5.483712663673224e-06,
+      "loss": 0.2696,
+      "step": 5790
+    },
+    {
+      "epoch": 9.047619047619047,
+      "grad_norm": 0.41482260823249817,
+      "learning_rate": 5.394998071025104e-06,
+      "loss": 0.2623,
+      "step": 5795
+    },
+    {
+      "epoch": 9.055425448868071,
+      "grad_norm": 0.43754109740257263,
+      "learning_rate": 5.306987050489442e-06,
+      "loss": 0.2673,
+      "step": 5800
+    },
+    {
+      "epoch": 9.063231850117097,
+      "grad_norm": 0.41212213039398193,
+      "learning_rate": 5.21968025659928e-06,
+      "loss": 0.2643,
+      "step": 5805
+    },
+    {
+      "epoch": 9.07103825136612,
+      "grad_norm": 0.43795979022979736,
+      "learning_rate": 5.1330783386503765e-06,
+      "loss": 0.2672,
+      "step": 5810
+    },
+    {
+      "epoch": 9.078844652615144,
+      "grad_norm": 0.416830837726593,
+      "learning_rate": 5.047181940696333e-06,
+      "loss": 0.2637,
+      "step": 5815
+    },
+    {
+      "epoch": 9.086651053864168,
+      "grad_norm": 0.4341103136539459,
+      "learning_rate": 4.961991701543889e-06,
+      "loss": 0.2712,
+      "step": 5820
+    },
+    {
+      "epoch": 9.094457455113194,
+      "grad_norm": 0.4144740402698517,
+      "learning_rate": 4.877508254748076e-06,
+      "loss": 0.2665,
+      "step": 5825
+    },
+    {
+      "epoch": 9.102263856362217,
+      "grad_norm": 0.42530107498168945,
+      "learning_rate": 4.7937322286075725e-06,
+      "loss": 0.2678,
+      "step": 5830
+    },
+    {
+      "epoch": 9.110070257611241,
+      "grad_norm": 0.4266839623451233,
+      "learning_rate": 4.710664246160013e-06,
+      "loss": 0.2709,
+      "step": 5835
+    },
+    {
+      "epoch": 9.117876658860265,
+      "grad_norm": 0.42478978633880615,
+      "learning_rate": 4.628304925177318e-06,
+      "loss": 0.2692,
+      "step": 5840
+    },
+    {
+      "epoch": 9.12568306010929,
+      "grad_norm": 0.42860227823257446,
+      "learning_rate": 4.54665487816115e-06,
+      "loss": 0.2694,
+      "step": 5845
+    },
+    {
+      "epoch": 9.133489461358314,
+      "grad_norm": 0.43037670850753784,
+      "learning_rate": 4.465714712338398e-06,
+      "loss": 0.2686,
+      "step": 5850
+    },
+    {
+      "epoch": 9.141295862607338,
+      "grad_norm": 0.4481579661369324,
+      "learning_rate": 4.385485029656489e-06,
+      "loss": 0.2672,
+      "step": 5855
+    },
+    {
+      "epoch": 9.149102263856362,
+      "grad_norm": 0.406190425157547,
+      "learning_rate": 4.305966426779118e-06,
+      "loss": 0.2697,
+      "step": 5860
+    },
+    {
+      "epoch": 9.156908665105387,
+      "grad_norm": 0.41195255517959595,
+      "learning_rate": 4.22715949508169e-06,
+      "loss": 0.268,
+      "step": 5865
+    },
+    {
+      "epoch": 9.164715066354411,
+      "grad_norm": 0.41223201155662537,
+      "learning_rate": 4.149064820646953e-06,
+      "loss": 0.2625,
+      "step": 5870
+    },
+    {
+      "epoch": 9.172521467603435,
+      "grad_norm": 0.40198594331741333,
+      "learning_rate": 4.071682984260638e-06,
+      "loss": 0.2654,
+      "step": 5875
+    },
+    {
+      "epoch": 9.180327868852459,
+      "grad_norm": 0.4083414673805237,
+      "learning_rate": 3.99501456140714e-06,
+      "loss": 0.2666,
+      "step": 5880
+    },
+    {
+      "epoch": 9.188134270101482,
+      "grad_norm": 0.43248122930526733,
+      "learning_rate": 3.919060122265228e-06,
+      "loss": 0.2712,
+      "step": 5885
+    },
+    {
+      "epoch": 9.195940671350508,
+      "grad_norm": 0.4074991047382355,
+      "learning_rate": 3.8438202317037986e-06,
+      "loss": 0.2636,
+      "step": 5890
+    },
+    {
+      "epoch": 9.203747072599532,
+      "grad_norm": 0.410256028175354,
+      "learning_rate": 3.7692954492777764e-06,
+      "loss": 0.2663,
+      "step": 5895
+    },
+    {
+      "epoch": 9.211553473848555,
+      "grad_norm": 0.41726231575012207,
+      "learning_rate": 3.6954863292237297e-06,
+      "loss": 0.2649,
+      "step": 5900
+    },
+    {
+      "epoch": 9.21935987509758,
+      "grad_norm": 0.40415236353874207,
+      "learning_rate": 3.622393420456016e-06,
+      "loss": 0.2691,
+      "step": 5905
+    },
+    {
+      "epoch": 9.227166276346605,
+      "grad_norm": 0.40107595920562744,
+      "learning_rate": 3.550017266562489e-06,
+      "loss": 0.2636,
+      "step": 5910
+    },
+    {
+      "epoch": 9.234972677595628,
+      "grad_norm": 0.4478661119937897,
+      "learning_rate": 3.4783584058005527e-06,
+      "loss": 0.2711,
+      "step": 5915
+    },
+    {
+      "epoch": 9.242779078844652,
+      "grad_norm": 0.40624120831489563,
+      "learning_rate": 3.40741737109318e-06,
+      "loss": 0.2652,
+      "step": 5920
+    },
+    {
+      "epoch": 9.250585480093676,
+      "grad_norm": 0.42705732583999634,
+      "learning_rate": 3.3371946900248473e-06,
+      "loss": 0.2675,
+      "step": 5925
+    },
+    {
+      "epoch": 9.258391881342702,
+      "grad_norm": 0.4034976661205292,
+      "learning_rate": 3.267690884837726e-06,
+      "loss": 0.2631,
+      "step": 5930
+    },
+    {
+      "epoch": 9.266198282591725,
+      "grad_norm": 0.4358353912830353,
+      "learning_rate": 3.198906472427732e-06,
+      "loss": 0.2702,
+      "step": 5935
+    },
+    {
+      "epoch": 9.274004683840749,
+      "grad_norm": 0.42044761776924133,
+      "learning_rate": 3.130841964340692e-06,
+      "loss": 0.269,
+      "step": 5940
+    },
+    {
+      "epoch": 9.281811085089773,
+      "grad_norm": 0.4694254994392395,
+      "learning_rate": 3.06349786676855e-06,
+      "loss": 0.27,
+      "step": 5945
+    },
+    {
+      "epoch": 9.289617486338798,
+      "grad_norm": 0.4230342507362366,
+      "learning_rate": 2.996874680545603e-06,
+      "loss": 0.2645,
+      "step": 5950
+    },
+    {
+      "epoch": 9.297423887587822,
+      "grad_norm": 0.436117947101593,
+      "learning_rate": 2.9309729011447573e-06,
+      "loss": 0.2713,
+      "step": 5955
+    },
+    {
+      "epoch": 9.305230288836846,
+      "grad_norm": 0.4295603632926941,
+      "learning_rate": 2.8657930186738567e-06,
+      "loss": 0.2666,
+      "step": 5960
+    },
+    {
+      "epoch": 9.31303669008587,
+      "grad_norm": 0.42392897605895996,
+      "learning_rate": 2.8013355178720612e-06,
+      "loss": 0.2682,
+      "step": 5965
+    },
+    {
+      "epoch": 9.320843091334895,
+      "grad_norm": 0.40090325474739075,
+      "learning_rate": 2.7376008781061835e-06,
+      "loss": 0.2624,
+      "step": 5970
+    },
+    {
+      "epoch": 9.328649492583919,
+      "grad_norm": 0.4344378113746643,
+      "learning_rate": 2.674589573367192e-06,
+      "loss": 0.274,
+      "step": 5975
+    },
+    {
+      "epoch": 9.336455893832943,
+      "grad_norm": 0.4555279016494751,
+      "learning_rate": 2.612302072266637e-06,
+      "loss": 0.2696,
+      "step": 5980
+    },
+    {
+      "epoch": 9.344262295081966,
+      "grad_norm": 0.4352642297744751,
+      "learning_rate": 2.5507388380331843e-06,
+      "loss": 0.2677,
+      "step": 5985
+    },
+    {
+      "epoch": 9.352068696330992,
+      "grad_norm": 0.4107036590576172,
+      "learning_rate": 2.4899003285091536e-06,
+      "loss": 0.2692,
+      "step": 5990
+    },
+    {
+      "epoch": 9.359875097580016,
+      "grad_norm": 0.4284498691558838,
+      "learning_rate": 2.429786996147154e-06,
+      "loss": 0.2623,
+      "step": 5995
+    },
+    {
+      "epoch": 9.36768149882904,
+      "grad_norm": 0.42979347705841064,
+      "learning_rate": 2.3703992880066638e-06,
+      "loss": 0.2705,
+      "step": 6000
+    },
+    {
+      "epoch": 9.375487900078063,
+      "grad_norm": 0.4255581796169281,
+      "learning_rate": 2.3117376457507667e-06,
+      "loss": 0.2678,
+      "step": 6005
+    },
+    {
+      "epoch": 9.383294301327089,
+      "grad_norm": 0.4213410019874573,
+      "learning_rate": 2.2538025056428214e-06,
+      "loss": 0.2679,
+      "step": 6010
+    },
+    {
+      "epoch": 9.391100702576113,
+      "grad_norm": 0.42378342151641846,
+      "learning_rate": 2.1965942985431977e-06,
+      "loss": 0.2603,
+      "step": 6015
+    },
+    {
+      "epoch": 9.398907103825136,
+      "grad_norm": 0.43251511454582214,
+      "learning_rate": 2.140113449906167e-06,
+      "loss": 0.2678,
+      "step": 6020
+    },
+    {
+      "epoch": 9.40671350507416,
+      "grad_norm": 0.428505539894104,
+      "learning_rate": 2.0843603797766287e-06,
+      "loss": 0.2716,
+      "step": 6025
+    },
+    {
+      "epoch": 9.414519906323186,
+      "grad_norm": 0.40141308307647705,
+      "learning_rate": 2.0293355027870554e-06,
+      "loss": 0.264,
+      "step": 6030
+    },
+    {
+      "epoch": 9.42232630757221,
+      "grad_norm": 0.42421501874923706,
+      "learning_rate": 1.975039228154385e-06,
+      "loss": 0.2675,
+      "step": 6035
+    },
+    {
+      "epoch": 9.430132708821233,
+      "grad_norm": 0.41700848937034607,
+      "learning_rate": 1.921471959676957e-06,
+      "loss": 0.2663,
+      "step": 6040
+    },
+    {
+      "epoch": 9.437939110070257,
+      "grad_norm": 0.4247893989086151,
+      "learning_rate": 1.8686340957315695e-06,
+      "loss": 0.2671,
+      "step": 6045
+    },
+    {
+      "epoch": 9.445745511319283,
+      "grad_norm": 0.4123760163784027,
+      "learning_rate": 1.8165260292704711e-06,
+      "loss": 0.264,
+      "step": 6050
+    },
+    {
+      "epoch": 9.453551912568306,
+      "grad_norm": 0.4423754811286926,
+      "learning_rate": 1.7651481478184296e-06,
+      "loss": 0.2617,
+      "step": 6055
+    },
+    {
+      "epoch": 9.46135831381733,
+      "grad_norm": 0.4133644998073578,
+      "learning_rate": 1.7145008334698898e-06,
+      "loss": 0.2624,
+      "step": 6060
+    },
+    {
+      "epoch": 9.469164715066354,
+      "grad_norm": 0.4270353317260742,
+      "learning_rate": 1.6645844628860764e-06,
+      "loss": 0.2674,
+      "step": 6065
+    },
+    {
+      "epoch": 9.47697111631538,
+      "grad_norm": 0.41615259647369385,
+      "learning_rate": 1.6153994072922506e-06,
+      "loss": 0.2708,
+      "step": 6070
+    },
+    {
+      "epoch": 9.484777517564403,
+      "grad_norm": 0.4244523346424103,
+      "learning_rate": 1.5669460324749586e-06,
+      "loss": 0.2698,
+      "step": 6075
+    },
+    {
+      "epoch": 9.492583918813427,
+      "grad_norm": 0.4362230598926544,
+      "learning_rate": 1.5192246987791981e-06,
+      "loss": 0.2617,
+      "step": 6080
+    },
+    {
+      "epoch": 9.50039032006245,
+      "grad_norm": 0.4365394413471222,
+      "learning_rate": 1.472235761105878e-06,
+      "loss": 0.2669,
+      "step": 6085
+    },
+    {
+      "epoch": 9.508196721311476,
+      "grad_norm": 0.4158194363117218,
+      "learning_rate": 1.4259795689090972e-06,
+      "loss": 0.2694,
+      "step": 6090
+    },
+    {
+      "epoch": 9.5160031225605,
+      "grad_norm": 0.41145652532577515,
+      "learning_rate": 1.3804564661935915e-06,
+      "loss": 0.2629,
+      "step": 6095
+    },
+    {
+      "epoch": 9.523809523809524,
+      "grad_norm": 0.39693784713745117,
+      "learning_rate": 1.3356667915121025e-06,
+      "loss": 0.2683,
+      "step": 6100
+    },
+    {
+      "epoch": 9.531615925058547,
+      "grad_norm": 0.4337891638278961,
+      "learning_rate": 1.2916108779629566e-06,
+      "loss": 0.2691,
+      "step": 6105
+    },
+    {
+      "epoch": 9.539422326307573,
+      "grad_norm": 0.431761234998703,
+      "learning_rate": 1.2482890531875125e-06,
+      "loss": 0.2704,
+      "step": 6110
+    },
+    {
+      "epoch": 9.547228727556597,
+      "grad_norm": 0.42788565158843994,
+      "learning_rate": 1.2057016393677623e-06,
+      "loss": 0.2721,
+      "step": 6115
+    },
+    {
+      "epoch": 9.55503512880562,
+      "grad_norm": 0.4381605088710785,
+      "learning_rate": 1.163848953223934e-06,
+      "loss": 0.2731,
+      "step": 6120
+    },
+    {
+      "epoch": 9.562841530054644,
+      "grad_norm": 0.42890533804893494,
+      "learning_rate": 1.1227313060120926e-06,
+      "loss": 0.2683,
+      "step": 6125
+    },
+    {
+      "epoch": 9.570647931303668,
+      "grad_norm": 0.41614386439323425,
+      "learning_rate": 1.0823490035218987e-06,
+      "loss": 0.2634,
+      "step": 6130
+    },
+    {
+      "epoch": 9.578454332552694,
+      "grad_norm": 0.41682758927345276,
+      "learning_rate": 1.042702346074287e-06,
+      "loss": 0.2621,
+      "step": 6135
+    },
+    {
+      "epoch": 9.586260733801717,
+      "grad_norm": 0.4195772707462311,
+      "learning_rate": 1.003791628519213e-06,
+      "loss": 0.2677,
+      "step": 6140
+    },
+    {
+      "epoch": 9.594067135050741,
+      "grad_norm": 0.40635907649993896,
+      "learning_rate": 9.656171402335213e-07,
+      "loss": 0.2668,
+      "step": 6145
+    },
+    {
+      "epoch": 9.601873536299767,
+      "grad_norm": 0.40993356704711914,
+      "learning_rate": 9.281791651187366e-07,
+      "loss": 0.2661,
+      "step": 6150
+    },
+    {
+      "epoch": 9.60967993754879,
+      "grad_norm": 0.4103589355945587,
+      "learning_rate": 8.914779815989982e-07,
+      "loss": 0.2691,
+      "step": 6155
+    },
+    {
+      "epoch": 9.617486338797814,
+      "grad_norm": 0.44015854597091675,
+      "learning_rate": 8.555138626189618e-07,
+      "loss": 0.2649,
+      "step": 6160
+    },
+    {
+      "epoch": 9.625292740046838,
+      "grad_norm": 0.414385586977005,
+      "learning_rate": 8.202870756417569e-07,
+      "loss": 0.2616,
+      "step": 6165
+    },
+    {
+      "epoch": 9.633099141295862,
+      "grad_norm": 0.4170667827129364,
+      "learning_rate": 7.857978826470325e-07,
+      "loss": 0.2658,
+      "step": 6170
+    },
+    {
+      "epoch": 9.640905542544887,
+      "grad_norm": 0.42618632316589355,
+      "learning_rate": 7.520465401290033e-07,
+      "loss": 0.2691,
+      "step": 6175
+    },
+    {
+      "epoch": 9.648711943793911,
+      "grad_norm": 0.41920483112335205,
+      "learning_rate": 7.19033299094496e-07,
+      "loss": 0.2639,
+      "step": 6180
+    },
+    {
+      "epoch": 9.656518345042935,
+      "grad_norm": 0.4324471056461334,
+      "learning_rate": 6.867584050611498e-07,
+      "loss": 0.2715,
+      "step": 6185
+    },
+    {
+      "epoch": 9.664324746291959,
+      "grad_norm": 0.4518769681453705,
+      "learning_rate": 6.552220980555635e-07,
+      "loss": 0.2749,
+      "step": 6190
+    },
+    {
+      "epoch": 9.672131147540984,
+      "grad_norm": 0.4381062984466553,
+      "learning_rate": 6.244246126114627e-07,
+      "loss": 0.2734,
+      "step": 6195
+    },
+    {
+      "epoch": 9.679937548790008,
+      "grad_norm": 0.4453407824039459,
+      "learning_rate": 5.943661777680354e-07,
+      "loss": 0.2692,
+      "step": 6200
+    },
+    {
+      "epoch": 9.687743950039032,
+      "grad_norm": 0.41984906792640686,
+      "learning_rate": 5.650470170681876e-07,
+      "loss": 0.2644,
+      "step": 6205
+    },
+    {
+      "epoch": 9.695550351288055,
+      "grad_norm": 0.4198761284351349,
+      "learning_rate": 5.364673485568794e-07,
+      "loss": 0.2676,
+      "step": 6210
+    },
+    {
+      "epoch": 9.703356752537081,
+      "grad_norm": 0.42462319135665894,
+      "learning_rate": 5.086273847795031e-07,
+      "loss": 0.2671,
+      "step": 6215
+    },
+    {
+      "epoch": 9.711163153786105,
+      "grad_norm": 0.4393913447856903,
+      "learning_rate": 4.815273327803182e-07,
+      "loss": 0.2689,
+      "step": 6220
+    },
+    {
+      "epoch": 9.718969555035128,
+      "grad_norm": 0.4282847046852112,
+      "learning_rate": 4.5516739410087494e-07,
+      "loss": 0.2696,
+      "step": 6225
+    },
+    {
+      "epoch": 9.726775956284152,
+      "grad_norm": 0.41957032680511475,
+      "learning_rate": 4.2954776477860393e-07,
+      "loss": 0.2748,
+      "step": 6230
+    },
+    {
+      "epoch": 9.734582357533178,
+      "grad_norm": 0.4238908886909485,
+      "learning_rate": 4.0466863534522893e-07,
+      "loss": 0.2741,
+      "step": 6235
+    },
+    {
+      "epoch": 9.742388758782202,
+      "grad_norm": 0.4146591126918793,
+      "learning_rate": 3.805301908254455e-07,
+      "loss": 0.2653,
+      "step": 6240
+    },
+    {
+      "epoch": 9.750195160031225,
+      "grad_norm": 0.4316459000110626,
+      "learning_rate": 3.571326107355333e-07,
+      "loss": 0.2687,
+      "step": 6245
+    },
+    {
+      "epoch": 9.758001561280249,
+      "grad_norm": 0.4224886894226074,
+      "learning_rate": 3.3447606908196817e-07,
+      "loss": 0.2646,
+      "step": 6250
+    },
+    {
+      "epoch": 9.765807962529275,
+      "grad_norm": 0.439411461353302,
+      "learning_rate": 3.125607343602011e-07,
+      "loss": 0.2657,
+      "step": 6255
+    },
+    {
+      "epoch": 9.773614363778298,
+      "grad_norm": 0.4261423647403717,
+      "learning_rate": 2.9138676955333676e-07,
+      "loss": 0.2606,
+      "step": 6260
+    },
+    {
+      "epoch": 9.781420765027322,
+      "grad_norm": 0.41440731287002563,
+      "learning_rate": 2.709543321309793e-07,
+      "loss": 0.2673,
+      "step": 6265
+    },
+    {
+      "epoch": 9.789227166276346,
+      "grad_norm": 0.42373111844062805,
+      "learning_rate": 2.5126357404802183e-07,
+      "loss": 0.2751,
+      "step": 6270
+    },
+    {
+      "epoch": 9.797033567525371,
+      "grad_norm": 0.4395394027233124,
+      "learning_rate": 2.3231464174352512e-07,
+      "loss": 0.2747,
+      "step": 6275
+    },
+    {
+      "epoch": 9.804839968774395,
+      "grad_norm": 0.4383135735988617,
+      "learning_rate": 2.141076761396521e-07,
+      "loss": 0.2708,
+      "step": 6280
+    },
+    {
+      "epoch": 9.812646370023419,
+      "grad_norm": 0.4313544034957886,
+      "learning_rate": 1.966428126405795e-07,
+      "loss": 0.2686,
+      "step": 6285
+    },
+    {
+      "epoch": 9.820452771272443,
+      "grad_norm": 0.39951956272125244,
+      "learning_rate": 1.7992018113152098e-07,
+      "loss": 0.2591,
+      "step": 6290
+    },
+    {
+      "epoch": 9.828259172521468,
+      "grad_norm": 0.4347188174724579,
+      "learning_rate": 1.6393990597775022e-07,
+      "loss": 0.2709,
+      "step": 6295
+    },
+    {
+      "epoch": 9.836065573770492,
+      "grad_norm": 0.4294663965702057,
+      "learning_rate": 1.487021060236904e-07,
+      "loss": 0.2644,
+      "step": 6300
+    },
+    {
+      "epoch": 9.843871975019516,
+      "grad_norm": 0.437920480966568,
+      "learning_rate": 1.3420689459200386e-07,
+      "loss": 0.2693,
+      "step": 6305
+    },
+    {
+      "epoch": 9.85167837626854,
+      "grad_norm": 0.4170249104499817,
+      "learning_rate": 1.204543794827595e-07,
+      "loss": 0.2748,
+      "step": 6310
+    },
+    {
+      "epoch": 9.859484777517565,
+      "grad_norm": 0.40519675612449646,
+      "learning_rate": 1.0744466297265554e-07,
+      "loss": 0.267,
+      "step": 6315
+    },
+    {
+      "epoch": 9.867291178766589,
+      "grad_norm": 0.40046894550323486,
+      "learning_rate": 9.517784181422019e-08,
+      "loss": 0.2614,
+      "step": 6320
+    },
+    {
+      "epoch": 9.875097580015613,
+      "grad_norm": 0.4310961067676544,
+      "learning_rate": 8.365400723512328e-08,
+      "loss": 0.2617,
+      "step": 6325
+    },
+    {
+      "epoch": 9.882903981264636,
+      "grad_norm": 0.43310293555259705,
+      "learning_rate": 7.287324493747693e-08,
+      "loss": 0.2685,
+      "step": 6330
+    },
+    {
+      "epoch": 9.890710382513662,
+      "grad_norm": 0.43305081129074097,
+      "learning_rate": 6.283563509719148e-08,
+      "loss": 0.2607,
+      "step": 6335
+    },
+    {
+      "epoch": 9.898516783762686,
+      "grad_norm": 0.42313021421432495,
+      "learning_rate": 5.354125236343155e-08,
+      "loss": 0.2714,
+      "step": 6340
+    },
+    {
+      "epoch": 9.90632318501171,
+      "grad_norm": 0.43003058433532715,
+      "learning_rate": 4.499016585799431e-08,
+      "loss": 0.2685,
+      "step": 6345
+    },
+    {
+      "epoch": 9.914129586260733,
+      "grad_norm": 0.4060039818286896,
+      "learning_rate": 3.7182439174832106e-08,
+      "loss": 0.2728,
+      "step": 6350
+    },
+    {
+      "epoch": 9.921935987509759,
+      "grad_norm": 0.4126955270767212,
+      "learning_rate": 3.0118130379575005e-08,
+      "loss": 0.2675,
+      "step": 6355
+    },
+    {
+      "epoch": 9.929742388758783,
+      "grad_norm": 0.43146437406539917,
+      "learning_rate": 2.379729200908676e-08,
+      "loss": 0.2695,
+      "step": 6360
+    },
+    {
+      "epoch": 9.937548790007806,
+      "grad_norm": 0.4640655219554901,
+      "learning_rate": 1.8219971071098408e-08,
+      "loss": 0.2713,
+      "step": 6365
+    },
+    {
+      "epoch": 9.94535519125683,
+      "grad_norm": 0.41949042677879333,
+      "learning_rate": 1.3386209043819708e-08,
+      "loss": 0.2707,
+      "step": 6370
+    },
+    {
+      "epoch": 9.953161592505854,
+      "grad_norm": 0.42296043038368225,
+      "learning_rate": 9.296041875683781e-09,
+      "loss": 0.2636,
+      "step": 6375
+    },
+    {
+      "epoch": 9.96096799375488,
+      "grad_norm": 0.45912352204322815,
+      "learning_rate": 5.949499985025142e-09,
+      "loss": 0.2702,
+      "step": 6380
+    },
+    {
+      "epoch": 9.968774395003903,
+      "grad_norm": 0.4141922891139984,
+      "learning_rate": 3.346608259890971e-09,
+      "loss": 0.2663,
+      "step": 6385
+    },
+    {
+      "epoch": 9.976580796252927,
+      "grad_norm": 0.42737066745758057,
+      "learning_rate": 1.4873860578412668e-09,
+      "loss": 0.2692,
+      "step": 6390
+    },
+    {
+      "epoch": 9.984387197501952,
+      "grad_norm": 0.441078782081604,
+      "learning_rate": 3.7184720581562306e-10,
+      "loss": 0.2707,
+      "step": 6395
+    },
+    {
+      "epoch": 9.992193598750976,
+      "grad_norm": 0.45411616563796997,
+      "learning_rate": 0.0,
+      "loss": 0.2714,
+      "step": 6400
+    },
+    {
+      "epoch": 9.992193598750976,
+      "eval_loss": 1.1572602987289429,
+      "eval_runtime": 0.3457,
+      "eval_samples_per_second": 14.464,
+      "eval_steps_per_second": 2.893,
+      "step": 6400
+    },
+    {
+      "epoch": 9.992193598750976,
+      "step": 6400,
+      "total_flos": 8.968401612833817e+18,
+      "train_loss": 0.34100163986906407,
+      "train_runtime": 19497.7947,
+      "train_samples_per_second": 10.509,
+      "train_steps_per_second": 0.328
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 6400,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 8.968401612833817e+18,
+  "train_batch_size": 4,
+  "trial_name": null,
+  "trial_params": null
+}