diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,5944 @@
+{
+  "best_metric": 2.4195876121520996,
+  "best_model_checkpoint": "./output/training_results/C016_random_sample_llama3-8b-base_pretrain_20240504_181744/checkpoint-1000",
+  "epoch": 4.0,
+  "eval_steps": 200,
+  "global_step": 4108,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0009737098344693282,
+      "grad_norm": 4.622597308389442,
+      "learning_rate": 7.5e-07,
+      "loss": 2.7686,
+      "step": 1
+    },
+    {
+      "epoch": 0.004868549172346641,
+      "grad_norm": 2.9010810501364586,
+      "learning_rate": 3e-06,
+      "loss": 2.7274,
+      "step": 5
+    },
+    {
+      "epoch": 0.009737098344693282,
+      "grad_norm": 2.6032598047562,
+      "learning_rate": 6e-06,
+      "loss": 2.667,
+      "step": 10
+    },
+    {
+      "epoch": 0.014605647517039922,
+      "grad_norm": 2.570724795196223,
+      "learning_rate": 9.75e-06,
+      "loss": 2.6551,
+      "step": 15
+    },
+    {
+      "epoch": 0.019474196689386564,
+      "grad_norm": 2.968267308985469,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 2.643,
+      "step": 20
+    },
+    {
+      "epoch": 0.024342745861733205,
+      "grad_norm": 2.8739885340338285,
+      "learning_rate": 1.487975935855361e-05,
+      "loss": 2.6294,
+      "step": 25
+    },
+    {
+      "epoch": 0.029211295034079845,
+      "grad_norm": 3.023512383227986,
+      "learning_rate": 1.4681310547763171e-05,
+      "loss": 2.627,
+      "step": 30
+    },
+    {
+      "epoch": 0.034079844206426485,
+      "grad_norm": 2.528499970351483,
+      "learning_rate": 1.4485278864193794e-05,
+      "loss": 2.5955,
+      "step": 35
+    },
+    {
+      "epoch": 0.03894839337877313,
+      "grad_norm": 2.547095976749082,
+      "learning_rate": 1.4291637778717775e-05,
+      "loss": 2.6183,
+      "step": 40
+    },
+    {
+      "epoch": 0.043816942551119765,
+      "grad_norm": 1.9843941555908906,
+      "learning_rate": 1.4100361021341427e-05,
+      "loss": 2.5925,
+      "step": 45
+    },
+    {
+      "epoch": 0.04868549172346641,
+      "grad_norm": 2.0711055751945815,
+      "learning_rate": 1.3911422578987613e-05,
+      "loss": 2.5762,
+      "step": 50
+    },
+    {
+      "epoch": 0.053554040895813046,
+      "grad_norm": 2.1149009622981736,
+      "learning_rate": 1.3724796693294575e-05,
+      "loss": 2.6115,
+      "step": 55
+    },
+    {
+      "epoch": 0.05842259006815969,
+      "grad_norm": 2.125830374210055,
+      "learning_rate": 1.3540457858430796e-05,
+      "loss": 2.6139,
+      "step": 60
+    },
+    {
+      "epoch": 0.06329113924050633,
+      "grad_norm": 1.9410826947795146,
+      "learning_rate": 1.3394616487103794e-05,
+      "loss": 2.5748,
+      "step": 65
+    },
+    {
+      "epoch": 0.06815968841285297,
+      "grad_norm": 2.1512046862388146,
+      "learning_rate": 1.3214330866815837e-05,
+      "loss": 2.6165,
+      "step": 70
+    },
+    {
+      "epoch": 0.0730282375851996,
+      "grad_norm": 2.0456636080847157,
+      "learning_rate": 1.3036262206089142e-05,
+      "loss": 2.529,
+      "step": 75
+    },
+    {
+      "epoch": 0.07789678675754626,
+      "grad_norm": 2.232330041514811,
+      "learning_rate": 1.2860385939062166e-05,
+      "loss": 2.6501,
+      "step": 80
+    },
+    {
+      "epoch": 0.0827653359298929,
+      "grad_norm": 2.415240950467998,
+      "learning_rate": 1.2686677742137694e-05,
+      "loss": 2.5821,
+      "step": 85
+    },
+    {
+      "epoch": 0.08763388510223953,
+      "grad_norm": 2.4124279122793375,
+      "learning_rate": 1.2515113531889656e-05,
+      "loss": 2.5964,
+      "step": 90
+    },
+    {
+      "epoch": 0.09250243427458617,
+      "grad_norm": 2.2985857393283364,
+      "learning_rate": 1.2345669462985584e-05,
+      "loss": 2.6227,
+      "step": 95
+    },
+    {
+      "epoch": 0.09737098344693282,
+      "grad_norm": 1.984056163436905,
+      "learning_rate": 1.2178321926124443e-05,
+      "loss": 2.5646,
+      "step": 100
+    },
+    {
+      "epoch": 0.10223953261927946,
+      "grad_norm": 1.9469124194664196,
+      "learning_rate": 1.201304754598983e-05,
+      "loss": 2.5677,
+      "step": 105
+    },
+    {
+      "epoch": 0.10710808179162609,
+      "grad_norm": 2.000422025073212,
+      "learning_rate": 1.1849823179218302e-05,
+      "loss": 2.6047,
+      "step": 110
+    },
+    {
+      "epoch": 0.11197663096397274,
+      "grad_norm": 1.8457596949821262,
+      "learning_rate": 1.168862591238295e-05,
+      "loss": 2.6201,
+      "step": 115
+    },
+    {
+      "epoch": 0.11684518013631938,
+      "grad_norm": 1.8876145847065753,
+      "learning_rate": 1.1529433059992029e-05,
+      "loss": 2.5327,
+      "step": 120
+    },
+    {
+      "epoch": 0.12171372930866602,
+      "grad_norm": 1.8657408134698463,
+      "learning_rate": 1.1372222162502398e-05,
+      "loss": 2.5759,
+      "step": 125
+    },
+    {
+      "epoch": 0.12658227848101267,
+      "grad_norm": 1.9689520876834024,
+      "learning_rate": 1.1216970984347951e-05,
+      "loss": 2.5822,
+      "step": 130
+    },
+    {
+      "epoch": 0.1314508276533593,
+      "grad_norm": 1.8787257890744393,
+      "learning_rate": 1.1063657511982824e-05,
+      "loss": 2.5703,
+      "step": 135
+    },
+    {
+      "epoch": 0.13631937682570594,
+      "grad_norm": 2.3946720962934673,
+      "learning_rate": 1.0912259951939131e-05,
+      "loss": 2.5048,
+      "step": 140
+    },
+    {
+      "epoch": 0.14118792599805258,
+      "grad_norm": 2.069484826525082,
+      "learning_rate": 1.0762756728899413e-05,
+      "loss": 2.5433,
+      "step": 145
+    },
+    {
+      "epoch": 0.1460564751703992,
+      "grad_norm": 1.9728419903336738,
+      "learning_rate": 1.0615126483783578e-05,
+      "loss": 2.5451,
+      "step": 150
+    },
+    {
+      "epoch": 0.15092502434274585,
+      "grad_norm": 1.8965209810159926,
+      "learning_rate": 1.0469348071850116e-05,
+      "loss": 2.5761,
+      "step": 155
+    },
+    {
+      "epoch": 0.15579357351509251,
+      "grad_norm": 1.9728913609239898,
+      "learning_rate": 1.0325400560811745e-05,
+      "loss": 2.5509,
+      "step": 160
+    },
+    {
+      "epoch": 0.16066212268743915,
+      "grad_norm": 1.7695814333016529,
+      "learning_rate": 1.0183263228965269e-05,
+      "loss": 2.5695,
+      "step": 165
+    },
+    {
+      "epoch": 0.1655306718597858,
+      "grad_norm": 1.9109787480384295,
+      "learning_rate": 1.0042915563335448e-05,
+      "loss": 2.5843,
+      "step": 170
+    },
+    {
+      "epoch": 0.17039922103213243,
+      "grad_norm": 1.8950613853479759,
+      "learning_rate": 9.904337257833037e-06,
+      "loss": 2.5447,
+      "step": 175
+    },
+    {
+      "epoch": 0.17526777020447906,
+      "grad_norm": 1.9462634859013892,
+      "learning_rate": 9.767508211426758e-06,
+      "loss": 2.5236,
+      "step": 180
+    },
+    {
+      "epoch": 0.1801363193768257,
+      "grad_norm": 1.9007929646893995,
+      "learning_rate": 9.632408526329031e-06,
+      "loss": 2.5511,
+      "step": 185
+    },
+    {
+      "epoch": 0.18500486854917234,
+      "grad_norm": 1.9203529678658229,
+      "learning_rate": 9.4990185061956e-06,
+      "loss": 2.5294,
+      "step": 190
+    },
+    {
+      "epoch": 0.189873417721519,
+      "grad_norm": 1.8492777850666573,
+      "learning_rate": 9.367318654338783e-06,
+      "loss": 2.5313,
+      "step": 195
+    },
+    {
+      "epoch": 0.19474196689386564,
+      "grad_norm": 2.053277986812429,
+      "learning_rate": 9.23728967195438e-06,
+      "loss": 2.5472,
+      "step": 200
+    },
+    {
+      "epoch": 0.19474196689386564,
+      "eval_loss": 2.526153564453125,
+      "eval_runtime": 86.8653,
+      "eval_samples_per_second": 84.027,
+      "eval_steps_per_second": 0.668,
+      "step": 200
+    },
+    {
+      "epoch": 0.19961051606621227,
+      "grad_norm": 1.8721064281522946,
+      "learning_rate": 9.108912456361987e-06,
+      "loss": 2.5221,
+      "step": 205
+    },
+    {
+      "epoch": 0.2044790652385589,
+      "grad_norm": 2.0199844942072707,
+      "learning_rate": 8.982168099258849e-06,
+      "loss": 2.5093,
+      "step": 210
+    },
+    {
+      "epoch": 0.20934761441090555,
+      "grad_norm": 1.7890996151627292,
+      "learning_rate": 8.857037884987036e-06,
+      "loss": 2.5197,
+      "step": 215
+    },
+    {
+      "epoch": 0.21421616358325218,
+      "grad_norm": 1.8519229946590852,
+      "learning_rate": 8.733503288813807e-06,
+      "loss": 2.4945,
+      "step": 220
+    },
+    {
+      "epoch": 0.21908471275559882,
+      "grad_norm": 1.7635414105359053,
+      "learning_rate": 8.611545975225206e-06,
+      "loss": 2.508,
+      "step": 225
+    },
+    {
+      "epoch": 0.22395326192794549,
+      "grad_norm": 1.8628950690860595,
+      "learning_rate": 8.491147796232788e-06,
+      "loss": 2.5093,
+      "step": 230
+    },
+    {
+      "epoch": 0.22882181110029212,
+      "grad_norm": 1.8308684769690282,
+      "learning_rate": 8.372290789693232e-06,
+      "loss": 2.4912,
+      "step": 235
+    },
+    {
+      "epoch": 0.23369036027263876,
+      "grad_norm": 1.8771495332116028,
+      "learning_rate": 8.254957177640966e-06,
+      "loss": 2.5112,
+      "step": 240
+    },
+    {
+      "epoch": 0.2385589094449854,
+      "grad_norm": 1.848114881393312,
+      "learning_rate": 8.13912936463366e-06,
+      "loss": 2.5004,
+      "step": 245
+    },
+    {
+      "epoch": 0.24342745861733203,
+      "grad_norm": 1.8532877282661944,
+      "learning_rate": 8.024789936110332e-06,
+      "loss": 2.5063,
+      "step": 250
+    },
+    {
+      "epoch": 0.24829600778967867,
+      "grad_norm": 1.8072290437694474,
+      "learning_rate": 7.911921656762299e-06,
+      "loss": 2.4812,
+      "step": 255
+    },
+    {
+      "epoch": 0.25316455696202533,
+      "grad_norm": 1.7086014592127396,
+      "learning_rate": 7.800507468916596e-06,
+      "loss": 2.5107,
+      "step": 260
+    },
+    {
+      "epoch": 0.25803310613437197,
+      "grad_norm": 1.713632600119656,
+      "learning_rate": 7.690530490932004e-06,
+      "loss": 2.4908,
+      "step": 265
+    },
+    {
+      "epoch": 0.2629016553067186,
+      "grad_norm": 1.7625004775631927,
+      "learning_rate": 7.581974015607404e-06,
+      "loss": 2.5364,
+      "step": 270
+    },
+    {
+      "epoch": 0.26777020447906524,
+      "grad_norm": 1.73225200198116,
+      "learning_rate": 7.474821508602552e-06,
+      "loss": 2.4639,
+      "step": 275
+    },
+    {
+      "epoch": 0.2726387536514119,
+      "grad_norm": 1.7849989992249207,
+      "learning_rate": 7.369056606871157e-06,
+      "loss": 2.5102,
+      "step": 280
+    },
+    {
+      "epoch": 0.2775073028237585,
+      "grad_norm": 1.8370989545866505,
+      "learning_rate": 7.2646631171060325e-06,
+      "loss": 2.4668,
+      "step": 285
+    },
+    {
+      "epoch": 0.28237585199610515,
+      "grad_norm": 1.7638888058771318,
+      "learning_rate": 7.161625014196459e-06,
+      "loss": 2.4778,
+      "step": 290
+    },
+    {
+      "epoch": 0.2872444011684518,
+      "grad_norm": 1.840487665783078,
+      "learning_rate": 7.059926439697582e-06,
+      "loss": 2.4949,
+      "step": 295
+    },
+    {
+      "epoch": 0.2921129503407984,
+      "grad_norm": 1.7654501649580268,
+      "learning_rate": 6.959551700311679e-06,
+      "loss": 2.5254,
+      "step": 300
+    },
+    {
+      "epoch": 0.29698149951314506,
+      "grad_norm": 1.9353359489811188,
+      "learning_rate": 6.860485266381383e-06,
+      "loss": 2.4986,
+      "step": 305
+    },
+    {
+      "epoch": 0.3018500486854917,
+      "grad_norm": 1.9617969898380334,
+      "learning_rate": 6.7627117703947036e-06,
+      "loss": 2.5186,
+      "step": 310
+    },
+    {
+      "epoch": 0.30671859785783834,
+      "grad_norm": 1.702196140955499,
+      "learning_rate": 6.6662160055017e-06,
+      "loss": 2.4988,
+      "step": 315
+    },
+    {
+      "epoch": 0.31158714703018503,
+      "grad_norm": 1.6844416023733326,
+      "learning_rate": 6.570982924042871e-06,
+      "loss": 2.5118,
+      "step": 320
+    },
+    {
+      "epoch": 0.31645569620253167,
+      "grad_norm": 1.7294022438189514,
+      "learning_rate": 6.476997636089112e-06,
+      "loss": 2.5248,
+      "step": 325
+    },
+    {
+      "epoch": 0.3213242453748783,
+      "grad_norm": 2.0648141831567983,
+      "learning_rate": 6.384245407993103e-06,
+      "loss": 2.5276,
+      "step": 330
+    },
+    {
+      "epoch": 0.32619279454722494,
+      "grad_norm": 1.728780446047325,
+      "learning_rate": 6.292711660952165e-06,
+      "loss": 2.4661,
+      "step": 335
+    },
+    {
+      "epoch": 0.3310613437195716,
+      "grad_norm": 1.7061810945106262,
+      "learning_rate": 6.202381969582487e-06,
+      "loss": 2.4577,
+      "step": 340
+    },
+    {
+      "epoch": 0.3359298928919182,
+      "grad_norm": 1.6870860598605661,
+      "learning_rate": 6.11324206050453e-06,
+      "loss": 2.4968,
+      "step": 345
+    },
+    {
+      "epoch": 0.34079844206426485,
+      "grad_norm": 1.7294693647713457,
+      "learning_rate": 6.025277810939709e-06,
+      "loss": 2.4665,
+      "step": 350
+    },
+    {
+      "epoch": 0.3456669912366115,
+      "grad_norm": 1.771451497626332,
+      "learning_rate": 5.9384752473181495e-06,
+      "loss": 2.4601,
+      "step": 355
+    },
+    {
+      "epoch": 0.3505355404089581,
+      "grad_norm": 1.7336392065891828,
+      "learning_rate": 5.8528205438975404e-06,
+      "loss": 2.4733,
+      "step": 360
+    },
+    {
+      "epoch": 0.35540408958130476,
+      "grad_norm": 1.751587409627844,
+      "learning_rate": 5.768300021392871e-06,
+      "loss": 2.4184,
+      "step": 365
+    },
+    {
+      "epoch": 0.3602726387536514,
+      "grad_norm": 1.735366924052735,
+      "learning_rate": 5.684900145617154e-06,
+      "loss": 2.4582,
+      "step": 370
+    },
+    {
+      "epoch": 0.36514118792599803,
+      "grad_norm": 1.8313267360344203,
+      "learning_rate": 5.6026075261329595e-06,
+      "loss": 2.4746,
+      "step": 375
+    },
+    {
+      "epoch": 0.37000973709834467,
+      "grad_norm": 1.8199109237326785,
+      "learning_rate": 5.521408914914653e-06,
+      "loss": 2.4395,
+      "step": 380
+    },
+    {
+      "epoch": 0.3748782862706913,
+      "grad_norm": 1.655319976322653,
+      "learning_rate": 5.441291205021376e-06,
+      "loss": 2.4805,
+      "step": 385
+    },
+    {
+      "epoch": 0.379746835443038,
+      "grad_norm": 1.6994219665911234,
+      "learning_rate": 5.3622414292806385e-06,
+      "loss": 2.4505,
+      "step": 390
+    },
+    {
+      "epoch": 0.38461538461538464,
+      "grad_norm": 1.7131089370429131,
+      "learning_rate": 5.2842467589823945e-06,
+      "loss": 2.5166,
+      "step": 395
+    },
+    {
+      "epoch": 0.3894839337877313,
+      "grad_norm": 1.7028942553739241,
+      "learning_rate": 5.207294502583657e-06,
+      "loss": 2.4431,
+      "step": 400
+    },
+    {
+      "epoch": 0.3894839337877313,
+      "eval_loss": 2.4732775688171387,
+      "eval_runtime": 85.4562,
+      "eval_samples_per_second": 85.412,
+      "eval_steps_per_second": 0.679,
+      "step": 400
+    },
+    {
+      "epoch": 0.3943524829600779,
+      "grad_norm": 1.772274911090255,
+      "learning_rate": 5.131372104423501e-06,
+      "loss": 2.5029,
+      "step": 405
+    },
+    {
+      "epoch": 0.39922103213242455,
+      "grad_norm": 1.6531175372154208,
+      "learning_rate": 5.0564671434483495e-06,
+      "loss": 2.5081,
+      "step": 410
+    },
+    {
+      "epoch": 0.4040895813047712,
+      "grad_norm": 1.649280502586864,
+      "learning_rate": 4.9825673319475865e-06,
+      "loss": 2.4378,
+      "step": 415
+    },
+    {
+      "epoch": 0.4089581304771178,
+      "grad_norm": 1.8019468764704318,
+      "learning_rate": 4.909660514299323e-06,
+      "loss": 2.5013,
+      "step": 420
+    },
+    {
+      "epoch": 0.41382667964946446,
+      "grad_norm": 2.001260057759529,
+      "learning_rate": 4.837734665726331e-06,
+      "loss": 2.4586,
+      "step": 425
+    },
+    {
+      "epoch": 0.4186952288218111,
+      "grad_norm": 1.7081822720906132,
+      "learning_rate": 4.766777891061954e-06,
+      "loss": 2.513,
+      "step": 430
+    },
+    {
+      "epoch": 0.42356377799415773,
+      "grad_norm": 1.8238918791581111,
+      "learning_rate": 4.696778423526121e-06,
+      "loss": 2.4336,
+      "step": 435
+    },
+    {
+      "epoch": 0.42843232716650437,
+      "grad_norm": 1.7162241591421625,
+      "learning_rate": 4.627724623511167e-06,
+      "loss": 2.4923,
+      "step": 440
+    },
+    {
+      "epoch": 0.433300876338851,
+      "grad_norm": 1.791273570594169,
+      "learning_rate": 4.559604977377591e-06,
+      "loss": 2.4524,
+      "step": 445
+    },
+    {
+      "epoch": 0.43816942551119764,
+      "grad_norm": 1.585355030794431,
+      "learning_rate": 4.4924080962595615e-06,
+      "loss": 2.4547,
+      "step": 450
+    },
+    {
+      "epoch": 0.4430379746835443,
+      "grad_norm": 1.627715490257672,
+      "learning_rate": 4.426122714880177e-06,
+      "loss": 2.4271,
+      "step": 455
+    },
+    {
+      "epoch": 0.44790652385589097,
+      "grad_norm": 1.7001533187863673,
+      "learning_rate": 4.360737690376327e-06,
+      "loss": 2.5004,
+      "step": 460
+    },
+    {
+      "epoch": 0.4527750730282376,
+      "grad_norm": 1.645312039241492,
+      "learning_rate": 4.29624200113319e-06,
+      "loss": 2.4227,
+      "step": 465
+    },
+    {
+      "epoch": 0.45764362220058424,
+      "grad_norm": 1.6935805821575964,
+      "learning_rate": 4.232624745628264e-06,
+      "loss": 2.4804,
+      "step": 470
+    },
+    {
+      "epoch": 0.4625121713729309,
+      "grad_norm": 1.6013179989515345,
+      "learning_rate": 4.169875141284801e-06,
+      "loss": 2.4438,
+      "step": 475
+    },
+    {
+      "epoch": 0.4673807205452775,
+      "grad_norm": 1.6369493229279692,
+      "learning_rate": 4.107982523334686e-06,
+      "loss": 2.468,
+      "step": 480
+    },
+    {
+      "epoch": 0.47224926971762415,
+      "grad_norm": 1.5838983294803852,
+      "learning_rate": 4.046936343690647e-06,
+      "loss": 2.4468,
+      "step": 485
+    },
+    {
+      "epoch": 0.4771178188899708,
+      "grad_norm": 1.652021178997152,
+      "learning_rate": 3.986726169827688e-06,
+      "loss": 2.4484,
+      "step": 490
+    },
+    {
+      "epoch": 0.4819863680623174,
+      "grad_norm": 1.6100438156106431,
+      "learning_rate": 3.9391530173378065e-06,
+      "loss": 2.4699,
+      "step": 495
+    },
+    {
+      "epoch": 0.48685491723466406,
+      "grad_norm": 1.6357278449106312,
+      "learning_rate": 3.880421728549051e-06,
+      "loss": 2.4972,
+      "step": 500
+    },
+    {
+      "epoch": 0.4917234664070107,
+      "grad_norm": 1.665827366459226,
+      "learning_rate": 3.822497839855098e-06,
+      "loss": 2.4498,
+      "step": 505
+    },
+    {
+      "epoch": 0.49659201557935734,
+      "grad_norm": 1.6865163895540334,
+      "learning_rate": 3.7653713478280894e-06,
+      "loss": 2.4344,
+      "step": 510
+    },
+    {
+      "epoch": 0.501460564751704,
+      "grad_norm": 1.6041809903831323,
+      "learning_rate": 3.70903235936047e-06,
+      "loss": 2.4389,
+      "step": 515
+    },
+    {
+      "epoch": 0.5063291139240507,
+      "grad_norm": 1.5953703380549231,
+      "learning_rate": 3.653471090598976e-06,
+      "loss": 2.456,
+      "step": 520
+    },
+    {
+      "epoch": 0.5111976630963972,
+      "grad_norm": 1.651511590173226,
+      "learning_rate": 3.59867786588742e-06,
+      "loss": 2.4241,
+      "step": 525
+    },
+    {
+      "epoch": 0.5160662122687439,
+      "grad_norm": 1.7103129608293084,
+      "learning_rate": 3.5446431167182903e-06,
+      "loss": 2.4655,
+      "step": 530
+    },
+    {
+      "epoch": 0.5209347614410905,
+      "grad_norm": 1.6921307716416762,
+      "learning_rate": 3.49135738069308e-06,
+      "loss": 2.4682,
+      "step": 535
+    },
+    {
+      "epoch": 0.5258033106134372,
+      "grad_norm": 2.4654732646603064,
+      "learning_rate": 3.438811300491226e-06,
+      "loss": 2.4509,
+      "step": 540
+    },
+    {
+      "epoch": 0.5306718597857838,
+      "grad_norm": 1.8390744205983383,
+      "learning_rate": 3.386995622847693e-06,
+      "loss": 2.4649,
+      "step": 545
+    },
+    {
+      "epoch": 0.5355404089581305,
+      "grad_norm": 1.644182617363712,
+      "learning_rate": 3.3359011975390947e-06,
+      "loss": 2.4421,
+      "step": 550
+    },
+    {
+      "epoch": 0.5404089581304771,
+      "grad_norm": 1.6574734258059267,
+      "learning_rate": 3.2855189763782522e-06,
+      "loss": 2.4339,
+      "step": 555
+    },
+    {
+      "epoch": 0.5452775073028238,
+      "grad_norm": 1.645146888773675,
+      "learning_rate": 3.2358400122172194e-06,
+      "loss": 2.4573,
+      "step": 560
+    },
+    {
+      "epoch": 0.5501460564751705,
+      "grad_norm": 1.6192183546217223,
+      "learning_rate": 3.1868554579586605e-06,
+      "loss": 2.4574,
+      "step": 565
+    },
+    {
+      "epoch": 0.555014605647517,
+      "grad_norm": 1.671822469835026,
+      "learning_rate": 3.1385565655755006e-06,
+      "loss": 2.4685,
+      "step": 570
+    },
+    {
+      "epoch": 0.5598831548198637,
+      "grad_norm": 1.6395167213969633,
+      "learning_rate": 3.0909346851388586e-06,
+      "loss": 2.4845,
+      "step": 575
+    },
+    {
+      "epoch": 0.5647517039922103,
+      "grad_norm": 1.6945968363900514,
+      "learning_rate": 3.0439812638541433e-06,
+      "loss": 2.4732,
+      "step": 580
+    },
+    {
+      "epoch": 0.569620253164557,
+      "grad_norm": 1.6428936915822159,
+      "learning_rate": 2.9976878451053044e-06,
+      "loss": 2.4622,
+      "step": 585
+    },
+    {
+      "epoch": 0.5744888023369036,
+      "grad_norm": 1.6443646175732205,
+      "learning_rate": 2.9520460675071095e-06,
+      "loss": 2.4466,
+      "step": 590
+    },
+    {
+      "epoch": 0.5793573515092503,
+      "grad_norm": 1.6462331908034773,
+      "learning_rate": 2.907047663965498e-06,
+      "loss": 2.466,
+      "step": 595
+    },
+    {
+      "epoch": 0.5842259006815969,
+      "grad_norm": 1.6579152402369843,
+      "learning_rate": 2.862684460745835e-06,
+      "loss": 2.4163,
+      "step": 600
+    },
+    {
+      "epoch": 0.5842259006815969,
+      "eval_loss": 2.4442696571350098,
+      "eval_runtime": 85.0459,
+      "eval_samples_per_second": 85.824,
+      "eval_steps_per_second": 0.682,
+      "step": 600
+    },
+    {
+      "epoch": 0.5890944498539435,
+      "grad_norm": 1.6546825117019377,
+      "learning_rate": 2.8189483765491077e-06,
+      "loss": 2.4857,
+      "step": 605
+    },
+    {
+      "epoch": 0.5939629990262901,
+      "grad_norm": 1.6868467390816666,
+      "learning_rate": 2.775831421595948e-06,
+      "loss": 2.4588,
+      "step": 610
+    },
+    {
+      "epoch": 0.5988315481986368,
+      "grad_norm": 1.631864642870465,
+      "learning_rate": 2.7333256967184713e-06,
+      "loss": 2.4542,
+      "step": 615
+    },
+    {
+      "epoch": 0.6037000973709834,
+      "grad_norm": 1.5765796237275709,
+      "learning_rate": 2.6914233924598045e-06,
+      "loss": 2.4277,
+      "step": 620
+    },
+    {
+      "epoch": 0.6085686465433301,
+      "grad_norm": 1.6329988363272994,
+      "learning_rate": 2.6501167881813393e-06,
+      "loss": 2.3847,
+      "step": 625
+    },
+    {
+      "epoch": 0.6134371957156767,
+      "grad_norm": 1.6799124493504514,
+      "learning_rate": 2.6093982511776034e-06,
+      "loss": 2.4285,
+      "step": 630
+    },
+    {
+      "epoch": 0.6183057448880234,
+      "grad_norm": 1.6239504885103953,
+      "learning_rate": 2.569260235798673e-06,
+      "loss": 2.4432,
+      "step": 635
+    },
+    {
+      "epoch": 0.6231742940603701,
+      "grad_norm": 1.6009220872039354,
+      "learning_rate": 2.5296952825801428e-06,
+      "loss": 2.4118,
+      "step": 640
+    },
+    {
+      "epoch": 0.6280428432327166,
+      "grad_norm": 1.668850128404625,
+      "learning_rate": 2.49069601738053e-06,
+      "loss": 2.4968,
+      "step": 645
+    },
+    {
+      "epoch": 0.6329113924050633,
+      "grad_norm": 1.6629353311695643,
+      "learning_rate": 2.45225515052612e-06,
+      "loss": 2.4499,
+      "step": 650
+    },
+    {
+      "epoch": 0.6377799415774099,
+      "grad_norm": 1.5929095731919567,
+      "learning_rate": 2.414365475963123e-06,
+      "loss": 2.4344,
+      "step": 655
+    },
+    {
+      "epoch": 0.6426484907497566,
+      "grad_norm": 1.6171386676286523,
+      "learning_rate": 2.3770198704171897e-06,
+      "loss": 2.4291,
+      "step": 660
+    },
+    {
+      "epoch": 0.6475170399221032,
+      "grad_norm": 1.6420238027175638,
+      "learning_rate": 2.3402112925601284e-06,
+      "loss": 2.4089,
+      "step": 665
+    },
+    {
+      "epoch": 0.6523855890944499,
+      "grad_norm": 1.6788147362161117,
+      "learning_rate": 2.30393278218386e-06,
+      "loss": 2.4573,
+      "step": 670
+    },
+    {
+      "epoch": 0.6572541382667965,
+      "grad_norm": 1.6870203522696865,
+      "learning_rate": 2.2681774593814984e-06,
+      "loss": 2.4088,
+      "step": 675
+    },
+    {
+      "epoch": 0.6621226874391432,
+      "grad_norm": 1.635603955024262,
+      "learning_rate": 2.232938523735545e-06,
+      "loss": 2.4588,
+      "step": 680
+    },
+    {
+      "epoch": 0.6669912366114897,
+      "grad_norm": 1.6513072405624771,
+      "learning_rate": 2.1982092535131004e-06,
+      "loss": 2.4498,
+      "step": 685
+    },
+    {
+      "epoch": 0.6718597857838364,
+      "grad_norm": 1.6924700394115,
+      "learning_rate": 2.1639830048681073e-06,
+      "loss": 2.4365,
+      "step": 690
+    },
+    {
+      "epoch": 0.676728334956183,
+      "grad_norm": 1.6617480593936398,
+      "learning_rate": 2.130253211050499e-06,
+      "loss": 2.4644,
+      "step": 695
+    },
+    {
+      "epoch": 0.6815968841285297,
+      "grad_norm": 1.6227857828645276,
+      "learning_rate": 2.097013381622268e-06,
+      "loss": 2.4065,
+      "step": 700
+    },
+    {
+      "epoch": 0.6864654333008764,
+      "grad_norm": 1.6566150548613952,
+      "learning_rate": 2.0642571016803623e-06,
+      "loss": 2.4368,
+      "step": 705
+    },
+    {
+      "epoch": 0.691333982473223,
+      "grad_norm": 2.0603714204760255,
+      "learning_rate": 2.031978031086392e-06,
+      "loss": 2.4595,
+      "step": 710
+    },
+    {
+      "epoch": 0.6962025316455697,
+      "grad_norm": 1.6243505097480613,
+      "learning_rate": 2.0001699037030468e-06,
+      "loss": 2.4347,
+      "step": 715
+    },
+    {
+      "epoch": 0.7010710808179162,
+      "grad_norm": 1.7249749503578298,
+      "learning_rate": 1.968826526637239e-06,
+      "loss": 2.4648,
+      "step": 720
+    },
+    {
+      "epoch": 0.7059396299902629,
+      "grad_norm": 1.7262462882996954,
+      "learning_rate": 1.9379417794898855e-06,
+      "loss": 2.439,
+      "step": 725
+    },
+    {
+      "epoch": 0.7108081791626095,
+      "grad_norm": 1.6204017543437201,
+      "learning_rate": 1.907509613612265e-06,
+      "loss": 2.4204,
+      "step": 730
+    },
+    {
+      "epoch": 0.7156767283349562,
+      "grad_norm": 1.6204489322103508,
+      "learning_rate": 1.8775240513689499e-06,
+      "loss": 2.4594,
+      "step": 735
+    },
+    {
+      "epoch": 0.7205452775073028,
+      "grad_norm": 1.6106742671779046,
+      "learning_rate": 1.8479791854072195e-06,
+      "loss": 2.4215,
+      "step": 740
+    },
+    {
+      "epoch": 0.7254138266796495,
+      "grad_norm": 1.6278702344437153,
+      "learning_rate": 1.818869177932943e-06,
+      "loss": 2.4177,
+      "step": 745
+    },
+    {
+      "epoch": 0.7302823758519961,
+      "grad_norm": 1.702303345942966,
+      "learning_rate": 1.7901882599928386e-06,
+      "loss": 2.4412,
+      "step": 750
+    },
+    {
+      "epoch": 0.7351509250243428,
+      "grad_norm": 1.616327757952299,
+      "learning_rate": 1.7619307307631266e-06,
+      "loss": 2.4,
+      "step": 755
+    },
+    {
+      "epoch": 0.7400194741966893,
+      "grad_norm": 1.581113737981139,
+      "learning_rate": 1.734090956844456e-06,
+      "loss": 2.4072,
+      "step": 760
+    },
+    {
+      "epoch": 0.744888023369036,
+      "grad_norm": 1.556080290425103,
+      "learning_rate": 1.7066633715631195e-06,
+      "loss": 2.4384,
+      "step": 765
+    },
+    {
+      "epoch": 0.7497565725413826,
+      "grad_norm": 1.6038706235589455,
+      "learning_rate": 1.6796424742784764e-06,
+      "loss": 2.4738,
+      "step": 770
+    },
+    {
+      "epoch": 0.7546251217137293,
+      "grad_norm": 1.662078041711206,
+      "learning_rate": 1.6530228296965529e-06,
+      "loss": 2.4957,
+      "step": 775
+    },
+    {
+      "epoch": 0.759493670886076,
+      "grad_norm": 1.6328340091897662,
+      "learning_rate": 1.6267990671897495e-06,
+      "loss": 2.4816,
+      "step": 780
+    },
+    {
+      "epoch": 0.7643622200584226,
+      "grad_norm": 1.6438111553611983,
+      "learning_rate": 1.6009658801226467e-06,
+      "loss": 2.4171,
+      "step": 785
+    },
+    {
+      "epoch": 0.7692307692307693,
+      "grad_norm": 1.7211490194262304,
+      "learning_rate": 1.575518025183845e-06,
+      "loss": 2.4268,
+      "step": 790
+    },
+    {
+      "epoch": 0.7740993184031159,
+      "grad_norm": 1.7200203924352186,
+      "learning_rate": 1.5504503217237718e-06,
+      "loss": 2.4661,
+      "step": 795
+    },
+    {
+      "epoch": 0.7789678675754625,
+      "grad_norm": 1.6451751546473583,
+      "learning_rate": 1.5257576510984581e-06,
+      "loss": 2.4462,
+      "step": 800
+    },
+    {
+      "epoch": 0.7789678675754625,
+      "eval_loss": 2.4281296730041504,
+      "eval_runtime": 85.85,
+      "eval_samples_per_second": 85.02,
+      "eval_steps_per_second": 0.676,
+      "step": 800
+    },
+    {
+      "epoch": 0.7838364167478091,
+      "grad_norm": 1.7186162803254572,
+      "learning_rate": 1.5014349560191975e-06,
+      "loss": 2.4447,
+      "step": 805
+    },
+    {
+      "epoch": 0.7887049659201558,
+      "grad_norm": 1.6390532449357682,
+      "learning_rate": 1.4774772399080773e-06,
+      "loss": 2.3768,
+      "step": 810
+    },
+    {
+      "epoch": 0.7935735150925024,
+      "grad_norm": 1.6479758648696174,
+      "learning_rate": 1.4538795662592943e-06,
+      "loss": 2.4559,
+      "step": 815
+    },
+    {
+      "epoch": 0.7984420642648491,
+      "grad_norm": 1.6376458723290501,
+      "learning_rate": 1.4306370580062715e-06,
+      "loss": 2.4113,
+      "step": 820
+    },
+    {
+      "epoch": 0.8033106134371957,
+      "grad_norm": 1.6642605682645168,
+      "learning_rate": 1.4077448968944671e-06,
+      "loss": 2.4438,
+      "step": 825
+    },
+    {
+      "epoch": 0.8081791626095424,
+      "grad_norm": 1.5597665460776493,
+      "learning_rate": 1.3851983228598879e-06,
+      "loss": 2.4615,
+      "step": 830
+    },
+    {
+      "epoch": 0.813047711781889,
+      "grad_norm": 1.5998155659125195,
+      "learning_rate": 1.3629926334132263e-06,
+      "loss": 2.4245,
+      "step": 835
+    },
+    {
+      "epoch": 0.8179162609542356,
+      "grad_norm": 1.6231320102269173,
+      "learning_rate": 1.341123183029607e-06,
+      "loss": 2.4677,
+      "step": 840
+    },
+    {
+      "epoch": 0.8227848101265823,
+      "grad_norm": 1.587206597114467,
+      "learning_rate": 1.3195853825438637e-06,
+      "loss": 2.4268,
+      "step": 845
+    },
+    {
+      "epoch": 0.8276533592989289,
+      "grad_norm": 1.69258641098763,
+      "learning_rate": 1.2983746985513552e-06,
+      "loss": 2.4468,
+      "step": 850
+    },
+    {
+      "epoch": 0.8325219084712756,
+      "grad_norm": 1.6329982694488412,
+      "learning_rate": 1.2774866528142222e-06,
+      "loss": 2.4483,
+      "step": 855
+    },
+    {
+      "epoch": 0.8373904576436222,
+      "grad_norm": 1.6510229207341078,
+      "learning_rate": 1.2569168216730917e-06,
+      "loss": 2.4192,
+      "step": 860
+    },
+    {
+      "epoch": 0.8422590068159689,
+      "grad_norm": 1.5817582463558826,
+      "learning_rate": 1.2366608354641566e-06,
+      "loss": 2.4214,
+      "step": 865
+    },
+    {
+      "epoch": 0.8471275559883155,
+      "grad_norm": 2.4364871892272117,
+      "learning_rate": 1.216714377941609e-06,
+      "loss": 2.3713,
+      "step": 870
+    },
+    {
+      "epoch": 0.8519961051606622,
+      "grad_norm": 2.5372621502402186,
+      "learning_rate": 1.197073185705361e-06,
+      "loss": 2.415,
+      "step": 875
+    },
+    {
+      "epoch": 0.8568646543330087,
+      "grad_norm": 1.7440355879318754,
+      "learning_rate": 1.1777330476340439e-06,
+      "loss": 2.4468,
+      "step": 880
+    },
+    {
+      "epoch": 0.8617332035053554,
+      "grad_norm": 1.5959852776726504,
+      "learning_rate": 1.1624748993325157e-06,
+      "loss": 2.4416,
+      "step": 885
+    },
+    {
+      "epoch": 0.866601752677702,
+      "grad_norm": 1.5792041005336184,
+      "learning_rate": 1.1436662114572268e-06,
+      "loss": 2.4699,
+      "step": 890
+    },
+    {
+      "epoch": 0.8714703018500487,
+      "grad_norm": 1.6749035075354037,
+      "learning_rate": 1.1251470599677525e-06,
+      "loss": 2.3906,
+      "step": 895
+    },
+    {
+      "epoch": 0.8763388510223953,
+      "grad_norm": 1.6974831202238516,
+      "learning_rate": 1.106913427322351e-06,
+      "loss": 2.4473,
+      "step": 900
+    },
+    {
+      "epoch": 0.881207400194742,
+      "grad_norm": 1.629005550261729,
+      "learning_rate": 1.0889613456083674e-06,
+      "loss": 2.3865,
+      "step": 905
+    },
+    {
+      "epoch": 0.8860759493670886,
+      "grad_norm": 1.541318076309618,
+      "learning_rate": 1.0712868960049544e-06,
+      "loss": 2.4382,
+      "step": 910
+    },
+    {
+      "epoch": 0.8909444985394352,
+      "grad_norm": 1.5752275103135045,
+      "learning_rate": 1.0538862082508074e-06,
+      "loss": 2.452,
+      "step": 915
+    },
+    {
+      "epoch": 0.8958130477117819,
+      "grad_norm": 1.699492645238764,
+      "learning_rate": 1.036755460116832e-06,
+      "loss": 2.4333,
+      "step": 920
+    },
+    {
+      "epoch": 0.9006815968841285,
+      "grad_norm": 1.6186519953923517,
+      "learning_rate": 1.019890876883744e-06,
+      "loss": 2.4319,
+      "step": 925
+    },
+    {
+      "epoch": 0.9055501460564752,
+      "grad_norm": 1.586657649597322,
+      "learning_rate": 1.0032887308245372e-06,
+      "loss": 2.3888,
+      "step": 930
+    },
+    {
+      "epoch": 0.9104186952288218,
+      "grad_norm": 1.6044497498274506,
+      "learning_rate": 9.869453406918023e-07,
+      "loss": 2.4039,
+      "step": 935
+    },
+    {
+      "epoch": 0.9152872444011685,
+      "grad_norm": 1.6723989958086949,
+      "learning_rate": 9.708570712098284e-07,
+      "loss": 2.4005,
+      "step": 940
+    },
+    {
+      "epoch": 0.9201557935735151,
+      "grad_norm": 1.6902859760264846,
+      "learning_rate": 9.550203325714876e-07,
+      "loss": 2.4261,
+      "step": 945
+    },
+    {
+      "epoch": 0.9250243427458618,
+      "grad_norm": 1.6544740657432804,
+      "learning_rate": 9.394315799398425e-07,
+      "loss": 2.4451,
+      "step": 950
+    },
+    {
+      "epoch": 0.9298928919182083,
+      "grad_norm": 1.54882787053935,
+      "learning_rate": 9.240873129544315e-07,
+      "loss": 2.444,
+      "step": 955
+    },
+    {
+      "epoch": 0.934761441090555,
+      "grad_norm": 1.5909955762857377,
+      "learning_rate": 9.089840752422235e-07,
+      "loss": 2.3899,
+      "step": 960
+    },
+    {
+      "epoch": 0.9396299902629016,
+      "grad_norm": 1.6145099358774662,
+      "learning_rate": 8.941184539331783e-07,
+      "loss": 2.4215,
+      "step": 965
+    },
+    {
+      "epoch": 0.9444985394352483,
+      "grad_norm": 1.6027897997067373,
+      "learning_rate": 8.794870791803967e-07,
+      "loss": 2.4725,
+      "step": 970
+    },
+    {
+      "epoch": 0.9493670886075949,
+      "grad_norm": 1.5457405929725805,
+      "learning_rate": 8.650866236848001e-07,
+      "loss": 2.4511,
+      "step": 975
+    },
+    {
+      "epoch": 0.9542356377799416,
+      "grad_norm": 1.5741657202863206,
+      "learning_rate": 8.509138022243367e-07,
+      "loss": 2.4241,
+      "step": 980
+    },
+    {
+      "epoch": 0.9591041869522883,
+      "grad_norm": 1.6377152121905334,
+      "learning_rate": 8.369653711876414e-07,
+      "loss": 2.4034,
+      "step": 985
+    },
+    {
+      "epoch": 0.9639727361246349,
+      "grad_norm": 1.5672375026688792,
+      "learning_rate": 8.232381281121472e-07,
+      "loss": 2.4561,
+      "step": 990
+    },
+    {
+      "epoch": 0.9688412852969815,
+      "grad_norm": 1.5788396321776008,
+      "learning_rate": 8.097289112265904e-07,
+      "loss": 2.4023,
+      "step": 995
+    },
+    {
+      "epoch": 0.9737098344693281,
+      "grad_norm": 1.6492145516163694,
+      "learning_rate": 7.964345989978902e-07,
+      "loss": 2.4353,
+      "step": 1000
+    },
+    {
+      "epoch": 0.9737098344693281,
+      "eval_loss": 2.4195876121520996,
+      "eval_runtime": 85.322,
+      "eval_samples_per_second": 85.547,
+      "eval_steps_per_second": 0.68,
+      "step": 1000
+    },
+    {
+      "epoch": 0.9785783836416748,
+      "grad_norm": 1.5879573832642313,
+      "learning_rate": 7.83352109682346e-07,
+      "loss": 2.4211,
+      "step": 1005
+    },
+    {
+      "epoch": 0.9834469328140214,
+      "grad_norm": 1.6569952022837342,
+      "learning_rate": 7.704784008811466e-07,
+      "loss": 2.4405,
+      "step": 1010
+    },
+    {
+      "epoch": 0.9883154819863681,
+      "grad_norm": 1.59377337317319,
+      "learning_rate": 7.5781046910014e-07,
+      "loss": 2.4239,
+      "step": 1015
+    },
+    {
+      "epoch": 0.9931840311587147,
+      "grad_norm": 1.6314113718182768,
+      "learning_rate": 7.453453493138208e-07,
+      "loss": 2.3844,
+      "step": 1020
+    },
+    {
+      "epoch": 0.9980525803310614,
+      "grad_norm": 1.6620011915571502,
+      "learning_rate": 7.330801145335265e-07,
+      "loss": 2.4181,
+      "step": 1025
+    },
+    {
+      "epoch": 1.002921129503408,
+      "grad_norm": 1.9251273144439607,
+      "learning_rate": 7.210118753797866e-07,
+      "loss": 2.3064,
+      "step": 1030
+    },
+    {
+      "epoch": 1.0077896786757545,
+      "grad_norm": 1.6577940198631569,
+      "learning_rate": 7.091377796588075e-07,
+      "loss": 2.1998,
+      "step": 1035
+    },
+    {
+      "epoch": 1.0126582278481013,
+      "grad_norm": 1.7530256550139145,
+      "learning_rate": 6.974550119430372e-07,
+      "loss": 2.2708,
+      "step": 1040
+    },
+    {
+      "epoch": 1.017526777020448,
+      "grad_norm": 1.6969458887579825,
+      "learning_rate": 6.859607931558122e-07,
+      "loss": 2.2173,
+      "step": 1045
+    },
+    {
+      "epoch": 1.0223953261927945,
+      "grad_norm": 1.7016018631062932,
+      "learning_rate": 6.746523801600123e-07,
+      "loss": 2.2381,
+      "step": 1050
+    },
+    {
+      "epoch": 1.0272638753651413,
+      "grad_norm": 1.7593644025263198,
+      "learning_rate": 6.635270653507262e-07,
+      "loss": 2.2734,
+      "step": 1055
+    },
+    {
+      "epoch": 1.0321324245374879,
+      "grad_norm": 1.654744797257189,
+      "learning_rate": 6.525821762518746e-07,
+      "loss": 2.2064,
+      "step": 1060
+    },
+    {
+      "epoch": 1.0370009737098345,
+      "grad_norm": 1.7421937503061258,
+      "learning_rate": 6.41815075116768e-07,
+      "loss": 2.2413,
+      "step": 1065
+    },
+    {
+      "epoch": 1.041869522882181,
+      "grad_norm": 1.6658497077947585,
+      "learning_rate": 6.312231585325566e-07,
+      "loss": 2.2099,
+      "step": 1070
+    },
+    {
+      "epoch": 1.0467380720545278,
+      "grad_norm": 1.6993685133537,
+      "learning_rate": 6.208038570285602e-07,
+      "loss": 2.2881,
+      "step": 1075
+    },
+    {
+      "epoch": 1.0516066212268744,
+      "grad_norm": 1.7131084770608689,
+      "learning_rate": 6.105546346884208e-07,
+      "loss": 2.2381,
+      "step": 1080
+    },
+    {
+      "epoch": 1.056475170399221,
+      "grad_norm": 1.7081524078028645,
+      "learning_rate": 6.004729887660704e-07,
+      "loss": 2.236,
+      "step": 1085
+    },
+    {
+      "epoch": 1.0613437195715676,
+      "grad_norm": 1.8169379524340188,
+      "learning_rate": 5.905564493054672e-07,
+      "loss": 2.2342,
+      "step": 1090
+    },
+    {
+      "epoch": 1.0662122687439144,
+      "grad_norm": 1.6685262696722296,
+      "learning_rate": 5.808025787640756e-07,
+      "loss": 2.222,
+      "step": 1095
+    },
+    {
+      "epoch": 1.071080817916261,
+      "grad_norm": 1.6952684753454728,
+      "learning_rate": 5.712089716400602e-07,
+      "loss": 2.212,
+      "step": 1100
+    },
+    {
+      "epoch": 1.0759493670886076,
+      "grad_norm": 1.7168463474300166,
+      "learning_rate": 5.617732541031583e-07,
+      "loss": 2.2186,
+      "step": 1105
+    },
+    {
+      "epoch": 1.0808179162609541,
+      "grad_norm": 1.750097139848988,
+      "learning_rate": 5.524930836292023e-07,
+      "loss": 2.2494,
+      "step": 1110
+    },
+    {
+      "epoch": 1.085686465433301,
+      "grad_norm": 1.7330864953187293,
+      "learning_rate": 5.433661486382679e-07,
+      "loss": 2.2102,
+      "step": 1115
+    },
+    {
+      "epoch": 1.0905550146056475,
+      "grad_norm": 1.6803544506498977,
+      "learning_rate": 5.343901681364102e-07,
+      "loss": 2.2419,
+      "step": 1120
+    },
+    {
+      "epoch": 1.095423563777994,
+      "grad_norm": 1.7013937838659918,
+      "learning_rate": 5.255628913609601e-07,
+      "loss": 2.2194,
+      "step": 1125
+    },
+    {
+      "epoch": 1.1002921129503407,
+      "grad_norm": 1.7495270131848708,
+      "learning_rate": 5.168820974293567e-07,
+      "loss": 2.2213,
+      "step": 1130
+    },
+    {
+      "epoch": 1.1051606621226875,
+      "grad_norm": 1.7709656106164202,
+      "learning_rate": 5.08345594991474e-07,
+      "loss": 2.2428,
+      "step": 1135
+    },
+    {
+      "epoch": 1.110029211295034,
+      "grad_norm": 1.7338624116926167,
+      "learning_rate": 4.999512218854316e-07,
+      "loss": 2.1828,
+      "step": 1140
+    },
+    {
+      "epoch": 1.1148977604673806,
+      "grad_norm": 1.6915851593817088,
+      "learning_rate": 4.916968447968384e-07,
+      "loss": 2.2286,
+      "step": 1145
+    },
+    {
+      "epoch": 1.1197663096397275,
+      "grad_norm": 1.7318300441723782,
+      "learning_rate": 4.835803589214607e-07,
+      "loss": 2.2443,
+      "step": 1150
+    },
+    {
+      "epoch": 1.124634858812074,
+      "grad_norm": 1.6773573534938775,
+      "learning_rate": 4.7559968763127113e-07,
+      "loss": 2.2427,
+      "step": 1155
+    },
+    {
+      "epoch": 1.1295034079844206,
+      "grad_norm": 1.7187378671025355,
+      "learning_rate": 4.67752782143862e-07,
+      "loss": 2.2726,
+      "step": 1160
+    },
+    {
+      "epoch": 1.1343719571567672,
+      "grad_norm": 1.740542786504217,
+      "learning_rate": 4.6003762119518e-07,
+      "loss": 2.1869,
+      "step": 1165
+    },
+    {
+      "epoch": 1.139240506329114,
+      "grad_norm": 1.6945482053223642,
+      "learning_rate": 4.524522107155746e-07,
+      "loss": 2.2248,
+      "step": 1170
+    },
+    {
+      "epoch": 1.1441090555014606,
+      "grad_norm": 1.6615466442429803,
+      "learning_rate": 4.4499458350910965e-07,
+      "loss": 2.2248,
+      "step": 1175
+    },
+    {
+      "epoch": 1.1489776046738072,
+      "grad_norm": 1.7112548018169453,
+      "learning_rate": 4.3766279893612986e-07,
+      "loss": 2.2471,
+      "step": 1180
+    },
+    {
+      "epoch": 1.1538461538461537,
+      "grad_norm": 1.860899105148827,
+      "learning_rate": 4.304549425990434e-07,
+      "loss": 2.2272,
+      "step": 1185
+    },
+    {
+      "epoch": 1.1587147030185005,
+      "grad_norm": 1.691810639062161,
+      "learning_rate": 4.23369126031296e-07,
+      "loss": 2.263,
+      "step": 1190
+    },
+    {
+      "epoch": 1.1635832521908471,
+      "grad_norm": 1.775044069455589,
+      "learning_rate": 4.1640348638951344e-07,
+      "loss": 2.2447,
+      "step": 1195
+    },
+    {
+      "epoch": 1.1684518013631937,
+      "grad_norm": 1.9319359346716891,
+      "learning_rate": 4.0955618614877875e-07,
+      "loss": 2.2111,
+      "step": 1200
+    },
+    {
+      "epoch": 1.1684518013631937,
+      "eval_loss": 2.429011106491089,
+      "eval_runtime": 85.6795,
+      "eval_samples_per_second": 85.19,
+      "eval_steps_per_second": 0.677,
+      "step": 1200
+    },
+    {
+      "epoch": 1.1733203505355405,
+      "grad_norm": 1.7574340980020409,
+      "learning_rate": 4.0282541280102037e-07,
+      "loss": 2.2332,
+      "step": 1205
+    },
+    {
+      "epoch": 1.178188899707887,
+      "grad_norm": 1.6980825882291957,
+      "learning_rate": 3.9620937855648926e-07,
+      "loss": 2.2235,
+      "step": 1210
+    },
+    {
+      "epoch": 1.1830574488802337,
+      "grad_norm": 1.7998015292888088,
+      "learning_rate": 3.897063200482918e-07,
+      "loss": 2.21,
+      "step": 1215
+    },
+    {
+      "epoch": 1.1879259980525803,
+      "grad_norm": 1.631324517663901,
+      "learning_rate": 3.833144980399548e-07,
+      "loss": 2.219,
+      "step": 1220
+    },
+    {
+      "epoch": 1.192794547224927,
+      "grad_norm": 1.6407460745986788,
+      "learning_rate": 3.770321971360005e-07,
+      "loss": 2.1846,
+      "step": 1225
+    },
+    {
+      "epoch": 1.1976630963972736,
+      "grad_norm": 1.700809400503958,
+      "learning_rate": 3.7085772549549546e-07,
+      "loss": 2.2605,
+      "step": 1230
+    },
+    {
+      "epoch": 1.2025316455696202,
+      "grad_norm": 1.6715825199948962,
+      "learning_rate": 3.647894145485631e-07,
+      "loss": 2.2213,
+      "step": 1235
+    },
+    {
+      "epoch": 1.2074001947419668,
+      "grad_norm": 1.6992864948362203,
+      "learning_rate": 3.588256187158179e-07,
+      "loss": 2.2256,
+      "step": 1240
+    },
+    {
+      "epoch": 1.2122687439143136,
+      "grad_norm": 1.7332695276540628,
+      "learning_rate": 3.5296471513070895e-07,
+      "loss": 2.2602,
+      "step": 1245
+    },
+    {
+      "epoch": 1.2171372930866602,
+      "grad_norm": 1.7489006914371872,
+      "learning_rate": 3.472051033647415e-07,
+      "loss": 2.2514,
+      "step": 1250
+    },
+    {
+      "epoch": 1.2220058422590068,
+      "grad_norm": 1.7282302399042118,
+      "learning_rate": 3.4154520515555147e-07,
+      "loss": 2.2403,
+      "step": 1255
+    },
+    {
+      "epoch": 1.2268743914313536,
+      "grad_norm": 1.702612549564712,
+      "learning_rate": 3.3598346413781214e-07,
+      "loss": 2.2181,
+      "step": 1260
+    },
+    {
+      "epoch": 1.2317429406037002,
+      "grad_norm": 1.7033226596241964,
+      "learning_rate": 3.305183455769435e-07,
+      "loss": 2.2761,
+      "step": 1265
+    },
+    {
+      "epoch": 1.2366114897760467,
+      "grad_norm": 1.750385373058268,
+      "learning_rate": 3.251483361056018e-07,
+      "loss": 2.2387,
+      "step": 1270
+    },
+    {
+      "epoch": 1.2414800389483933,
+      "grad_norm": 1.7510710962758804,
+      "learning_rate": 3.1987194346292783e-07,
+      "loss": 2.2642,
+      "step": 1275
+    },
+    {
+      "epoch": 1.24634858812074,
+      "grad_norm": 1.699193378636474,
+      "learning_rate": 3.1468769623652614e-07,
+      "loss": 2.2675,
+      "step": 1280
+    },
+    {
+      "epoch": 1.2512171372930867,
+      "grad_norm": 1.6670573040794279,
+      "learning_rate": 3.0959414360715174e-07,
+      "loss": 2.2596,
+      "step": 1285
+    },
+    {
+      "epoch": 1.2560856864654333,
+      "grad_norm": 1.7667142100191355,
+      "learning_rate": 3.045898550960847e-07,
+      "loss": 2.2037,
+      "step": 1290
+    },
+    {
+      "epoch": 1.2609542356377799,
+      "grad_norm": 1.7069461836357607,
+      "learning_rate": 2.9967342031515984e-07,
+      "loss": 2.1602,
+      "step": 1295
+    },
+    {
+      "epoch": 1.2658227848101267,
+      "grad_norm": 1.6514498602889256,
+      "learning_rate": 2.948434487194415e-07,
+      "loss": 2.2278,
+      "step": 1300
+    },
+    {
+      "epoch": 1.2706913339824732,
+      "grad_norm": 1.661661974641321,
+      "learning_rate": 2.900985693625059e-07,
+      "loss": 2.2009,
+      "step": 1305
+    },
+    {
+      "epoch": 1.2755598831548198,
+      "grad_norm": 1.774692203861925,
+      "learning_rate": 2.854374306543197e-07,
+      "loss": 2.2549,
+      "step": 1310
+    },
+    {
+      "epoch": 1.2804284323271666,
+      "grad_norm": 1.678761386598376,
+      "learning_rate": 2.8085870012168447e-07,
+      "loss": 2.2362,
+      "step": 1315
+    },
+    {
+      "epoch": 1.2852969814995132,
+      "grad_norm": 1.717206031840906,
+      "learning_rate": 2.763610641712288e-07,
+      "loss": 2.2259,
+      "step": 1320
+    },
+    {
+      "epoch": 1.2901655306718598,
+      "grad_norm": 1.725788802654923,
+      "learning_rate": 2.7194322785492024e-07,
+      "loss": 2.2743,
+      "step": 1325
+    },
+    {
+      "epoch": 1.2950340798442064,
+      "grad_norm": 1.7272128729108773,
+      "learning_rate": 2.6760391463808273e-07,
+      "loss": 2.2238,
+      "step": 1330
+    },
+    {
+      "epoch": 1.299902629016553,
+      "grad_norm": 1.673730997692294,
+      "learning_rate": 2.633418661698872e-07,
+      "loss": 2.213,
+      "step": 1335
+    },
+    {
+      "epoch": 1.3047711781888998,
+      "grad_norm": 1.67421773725042,
+      "learning_rate": 2.591558420563027e-07,
+      "loss": 2.2159,
+      "step": 1340
+    },
+    {
+      "epoch": 1.3096397273612463,
+      "grad_norm": 1.7181866282412452,
+      "learning_rate": 2.550446196354801e-07,
+      "loss": 2.1931,
+      "step": 1345
+    },
+    {
+      "epoch": 1.314508276533593,
+      "grad_norm": 1.9995904186048974,
+      "learning_rate": 2.5100699375554764e-07,
+      "loss": 2.2394,
+      "step": 1350
+    },
+    {
+      "epoch": 1.3193768257059397,
+      "grad_norm": 1.6972697466343885,
+      "learning_rate": 2.4704177655480044e-07,
+      "loss": 2.2046,
+      "step": 1355
+    },
+    {
+      "epoch": 1.3242453748782863,
+      "grad_norm": 1.7352888329956384,
+      "learning_rate": 2.431477972442557e-07,
+      "loss": 2.2582,
+      "step": 1360
+    },
+    {
+      "epoch": 1.3291139240506329,
+      "grad_norm": 1.7536755936521693,
+      "learning_rate": 2.3932390189255786e-07,
+      "loss": 2.2573,
+      "step": 1365
+    },
+    {
+      "epoch": 1.3339824732229795,
+      "grad_norm": 1.7063896078961662,
+      "learning_rate": 2.3556895321321132e-07,
+      "loss": 2.2665,
+      "step": 1370
+    },
+    {
+      "epoch": 1.3388510223953263,
+      "grad_norm": 1.6740223494854998,
+      "learning_rate": 2.3188183035411849e-07,
+      "loss": 2.2163,
+      "step": 1375
+    },
+    {
+      "epoch": 1.3437195715676729,
+      "grad_norm": 1.7492684308128166,
+      "learning_rate": 2.2826142868940236e-07,
+      "loss": 2.1842,
+      "step": 1380
+    },
+    {
+      "epoch": 1.3485881207400194,
+      "grad_norm": 1.7405438484291114,
+      "learning_rate": 2.2470665961349557e-07,
+      "loss": 2.1983,
+      "step": 1385
+    },
+    {
+      "epoch": 1.353456669912366,
+      "grad_norm": 1.764967629336457,
+      "learning_rate": 2.2121645033746942e-07,
+      "loss": 2.2207,
+      "step": 1390
+    },
+    {
+      "epoch": 1.3583252190847128,
+      "grad_norm": 1.7077892049631243,
+      "learning_rate": 2.1778974368759137e-07,
+      "loss": 2.2475,
+      "step": 1395
+    },
+    {
+      "epoch": 1.3631937682570594,
+      "grad_norm": 1.7009940340048213,
+      "learning_rate": 2.1442549790608077e-07,
+      "loss": 2.2503,
+      "step": 1400
+    },
+    {
+      "epoch": 1.3631937682570594,
+      "eval_loss": 2.428063154220581,
+      "eval_runtime": 85.6482,
+      "eval_samples_per_second": 85.221,
+      "eval_steps_per_second": 0.677,
+      "step": 1400
+    },
+    {
+      "epoch": 1.368062317429406,
+      "grad_norm": 1.6852036518237539,
+      "learning_rate": 2.1112268645405111e-07,
+      "loss": 2.2092,
+      "step": 1405
+    },
+    {
+      "epoch": 1.3729308666017528,
+      "grad_norm": 1.7762881669503894,
+      "learning_rate": 2.0788029781661377e-07,
+      "loss": 2.275,
+      "step": 1410
+    },
+    {
+      "epoch": 1.3777994157740994,
+      "grad_norm": 1.7187728321020497,
+      "learning_rate": 2.0469733531012517e-07,
+      "loss": 2.205,
+      "step": 1415
+    },
+    {
+      "epoch": 1.382667964946446,
+      "grad_norm": 1.75561614859316,
+      "learning_rate": 2.015728168915585e-07,
+      "loss": 2.2513,
+      "step": 1420
+    },
+    {
+      "epoch": 1.3875365141187925,
+      "grad_norm": 1.7108273702116936,
+      "learning_rate": 1.9850577496997802e-07,
+      "loss": 2.3129,
+      "step": 1425
+    },
+    {
+      "epoch": 1.3924050632911391,
+      "grad_norm": 1.7373174761439103,
+      "learning_rate": 1.9549525622009822e-07,
+      "loss": 2.3009,
+      "step": 1430
+    },
+    {
+      "epoch": 1.397273612463486,
+      "grad_norm": 1.6944981878481962,
+      "learning_rate": 1.9254032139790948e-07,
+      "loss": 2.2276,
+      "step": 1435
+    },
+    {
+      "epoch": 1.4021421616358325,
+      "grad_norm": 1.7020044804533232,
+      "learning_rate": 1.8964004515834934e-07,
+      "loss": 2.2515,
+      "step": 1440
+    },
+    {
+      "epoch": 1.407010710808179,
+      "grad_norm": 1.7104910024357038,
+      "learning_rate": 1.8679351587500074e-07,
+      "loss": 2.2176,
+      "step": 1445
+    },
+    {
+      "epoch": 1.4118792599805259,
+      "grad_norm": 1.706154278461277,
+      "learning_rate": 1.839998354618001e-07,
+      "loss": 2.2426,
+      "step": 1450
+    },
+    {
+      "epoch": 1.4167478091528725,
+      "grad_norm": 1.6546136448975106,
+      "learning_rate": 1.8125811919673288e-07,
+      "loss": 2.2729,
+      "step": 1455
+    },
+    {
+      "epoch": 1.421616358325219,
+      "grad_norm": 1.7298929182929772,
+      "learning_rate": 1.7856749554750352e-07,
+      "loss": 2.2213,
+      "step": 1460
+    },
+    {
+      "epoch": 1.4264849074975658,
+      "grad_norm": 1.6799052546726045,
+      "learning_rate": 1.7592710599915514e-07,
+      "loss": 2.2494,
+      "step": 1465
+    },
+    {
+      "epoch": 1.4313534566699124,
+      "grad_norm": 1.7192873393376786,
+      "learning_rate": 1.7333610488362632e-07,
+      "loss": 2.2219,
+      "step": 1470
+    },
+    {
+      "epoch": 1.436222005842259,
+      "grad_norm": 1.6974268468535265,
+      "learning_rate": 1.7079365921122288e-07,
+      "loss": 2.2159,
+      "step": 1475
+    },
+    {
+      "epoch": 1.4410905550146056,
+      "grad_norm": 1.736932687979258,
+      "learning_rate": 1.682989485039898e-07,
+      "loss": 2.2001,
+      "step": 1480
+    },
+    {
+      "epoch": 1.4459591041869522,
+      "grad_norm": 1.7389015197229705,
+      "learning_rate": 1.65851164630961e-07,
+      "loss": 2.1815,
+      "step": 1485
+    },
+    {
+      "epoch": 1.450827653359299,
+      "grad_norm": 1.7402452727637348,
+      "learning_rate": 1.634495116452756e-07,
+      "loss": 2.274,
+      "step": 1490
+    },
+    {
+      "epoch": 1.4556962025316456,
+      "grad_norm": 1.689117128248476,
+      "learning_rate": 1.6109320562313633e-07,
+      "loss": 2.2296,
+      "step": 1495
+    },
+    {
+      "epoch": 1.4605647517039921,
+      "grad_norm": 1.7021757020476915,
+      "learning_rate": 1.587814745045978e-07,
+      "loss": 2.2488,
+      "step": 1500
+    },
+    {
+      "epoch": 1.465433300876339,
+      "grad_norm": 1.726585875668609,
+      "learning_rate": 1.565135579361648e-07,
+      "loss": 2.248,
+      "step": 1505
+    },
+    {
+      "epoch": 1.4703018500486855,
+      "grad_norm": 1.7896520417513544,
+      "learning_rate": 1.542887071151836e-07,
+      "loss": 2.2768,
+      "step": 1510
+    },
+    {
+      "epoch": 1.475170399221032,
+      "grad_norm": 1.8424772477578564,
+      "learning_rate": 1.5210618463601055e-07,
+      "loss": 2.2177,
+      "step": 1515
+    },
+    {
+      "epoch": 1.480038948393379,
+      "grad_norm": 1.7094835426748392,
+      "learning_rate": 1.4996526433793895e-07,
+      "loss": 2.1814,
+      "step": 1520
+    },
+    {
+      "epoch": 1.4849074975657255,
+      "grad_norm": 1.774002445036824,
+      "learning_rate": 1.478652311548687e-07,
+      "loss": 2.241,
+      "step": 1525
+    },
+    {
+      "epoch": 1.489776046738072,
+      "grad_norm": 1.7528924437568645,
+      "learning_rate": 1.4580538096670264e-07,
+      "loss": 2.2446,
+      "step": 1530
+    },
+    {
+      "epoch": 1.4946445959104186,
+      "grad_norm": 1.7450272228199162,
+      "learning_rate": 1.4378502045245165e-07,
+      "loss": 2.2098,
+      "step": 1535
+    },
+    {
+      "epoch": 1.4995131450827652,
+      "grad_norm": 1.71350388367754,
+      "learning_rate": 1.4180346694503253e-07,
+      "loss": 2.2407,
+      "step": 1540
+    },
+    {
+      "epoch": 1.504381694255112,
+      "grad_norm": 1.6655187619092102,
+      "learning_rate": 1.398600482877437e-07,
+      "loss": 2.2516,
+      "step": 1545
+    },
+    {
+      "epoch": 1.5092502434274586,
+      "grad_norm": 1.730038860188208,
+      "learning_rate": 1.379541026923992e-07,
+      "loss": 2.2244,
+      "step": 1550
+    },
+    {
+      "epoch": 1.5141187925998052,
+      "grad_norm": 1.7106880841934986,
+      "learning_rate": 1.360849785991099e-07,
+      "loss": 2.2359,
+      "step": 1555
+    },
+    {
+      "epoch": 1.518987341772152,
+      "grad_norm": 1.7507950020460012,
+      "learning_rate": 1.3425203453769053e-07,
+      "loss": 2.2067,
+      "step": 1560
+    },
+    {
+      "epoch": 1.5238558909444986,
+      "grad_norm": 1.6877714873976932,
+      "learning_rate": 1.3245463899068173e-07,
+      "loss": 2.2203,
+      "step": 1565
+    },
+    {
+      "epoch": 1.5287244401168452,
+      "grad_norm": 1.7990198522805025,
+      "learning_rate": 1.306921702579676e-07,
+      "loss": 2.2176,
+      "step": 1570
+    },
+    {
+      "epoch": 1.533592989289192,
+      "grad_norm": 1.7072472713071833,
+      "learning_rate": 1.2896401632297622e-07,
+      "loss": 2.2199,
+      "step": 1575
+    },
+    {
+      "epoch": 1.5384615384615383,
+      "grad_norm": 1.7137109372098451,
+      "learning_rate": 1.2726957472044431e-07,
+      "loss": 2.2593,
+      "step": 1580
+    },
+    {
+      "epoch": 1.5433300876338851,
+      "grad_norm": 1.7487536889121376,
+      "learning_rate": 1.2560825240573496e-07,
+      "loss": 2.2512,
+      "step": 1585
+    },
+    {
+      "epoch": 1.5481986368062317,
+      "grad_norm": 1.689517701653058,
+      "learning_rate": 1.23979465625689e-07,
+      "loss": 2.2594,
+      "step": 1590
+    },
+    {
+      "epoch": 1.5530671859785783,
+      "grad_norm": 1.7198840121589114,
+      "learning_rate": 1.223826397909982e-07,
+      "loss": 2.1724,
+      "step": 1595
+    },
+    {
+      "epoch": 1.557935735150925,
+      "grad_norm": 1.7280648830031222,
+      "learning_rate": 1.2081720935008362e-07,
+      "loss": 2.258,
+      "step": 1600
+    },
+    {
+      "epoch": 1.557935735150925,
+      "eval_loss": 2.4271326065063477,
+      "eval_runtime": 85.3342,
+      "eval_samples_per_second": 85.534,
+      "eval_steps_per_second": 0.68,
+      "step": 1600
+    },
+    {
+      "epoch": 1.5628042843232717,
+      "grad_norm": 1.7166373586948793,
+      "learning_rate": 1.1928261766446461e-07,
+      "loss": 2.2507,
+      "step": 1605
+    },
+    {
+      "epoch": 1.5676728334956183,
+      "grad_norm": 1.6962264139269825,
+      "learning_rate": 1.1777831688560428e-07,
+      "loss": 2.214,
+      "step": 1610
+    },
+    {
+      "epoch": 1.572541382667965,
+      "grad_norm": 1.667890993114025,
+      "learning_rate": 1.1630376783321605e-07,
+      "loss": 2.2107,
+      "step": 1615
+    },
+    {
+      "epoch": 1.5774099318403116,
+      "grad_norm": 1.7705130739867874,
+      "learning_rate": 1.1485843987501698e-07,
+      "loss": 2.2298,
+      "step": 1620
+    },
+    {
+      "epoch": 1.5822784810126582,
+      "grad_norm": 1.7067117715143372,
+      "learning_rate": 1.134418108079144e-07,
+      "loss": 2.2568,
+      "step": 1625
+    },
+    {
+      "epoch": 1.587147030185005,
+      "grad_norm": 1.698686461270445,
+      "learning_rate": 1.1205336674061039e-07,
+      "loss": 2.256,
+      "step": 1630
+    },
+    {
+      "epoch": 1.5920155793573514,
+      "grad_norm": 1.6914580859058206,
+      "learning_rate": 1.1069260197761055e-07,
+      "loss": 2.2255,
+      "step": 1635
+    },
+    {
+      "epoch": 1.5968841285296982,
+      "grad_norm": 1.695162008456652,
+      "learning_rate": 1.0935901890462346e-07,
+      "loss": 2.1925,
+      "step": 1640
+    },
+    {
+      "epoch": 1.6017526777020448,
+      "grad_norm": 1.8353600777066885,
+      "learning_rate": 1.0805212787533523e-07,
+      "loss": 2.259,
+      "step": 1645
+    },
+    {
+      "epoch": 1.6066212268743914,
+      "grad_norm": 1.7146893423140706,
+      "learning_rate": 1.0677144709954816e-07,
+      "loss": 2.251,
+      "step": 1650
+    },
+    {
+      "epoch": 1.6114897760467382,
+      "grad_norm": 1.7479655666276084,
+      "learning_rate": 1.0551650253266659e-07,
+      "loss": 2.2297,
+      "step": 1655
+    },
+    {
+      "epoch": 1.6163583252190847,
+      "grad_norm": 1.7879147626083056,
+      "learning_rate": 1.0428682776651918e-07,
+      "loss": 2.1994,
+      "step": 1660
+    },
+    {
+      "epoch": 1.6212268743914313,
+      "grad_norm": 1.7970614657397905,
+      "learning_rate": 1.030819639215023e-07,
+      "loss": 2.2049,
+      "step": 1665
+    },
+    {
+      "epoch": 1.6260954235637781,
+      "grad_norm": 1.7518141901380913,
+      "learning_rate": 1.0190145954003192e-07,
+      "loss": 2.2203,
+      "step": 1670
+    },
+    {
+      "epoch": 1.6309639727361245,
+      "grad_norm": 1.7501918042535358,
+      "learning_rate": 1.0074487048129093e-07,
+      "loss": 2.2585,
+      "step": 1675
+    },
+    {
+      "epoch": 1.6358325219084713,
+      "grad_norm": 1.756138340727838,
+      "learning_rate": 9.961175981725802e-08,
+      "loss": 2.2441,
+      "step": 1680
+    },
+    {
+      "epoch": 1.6407010710808179,
+      "grad_norm": 1.6916112013631275,
+      "learning_rate": 9.850169773000545e-08,
+      "loss": 2.2229,
+      "step": 1685
+    },
+    {
+      "epoch": 1.6455696202531644,
+      "grad_norm": 1.7520093451689733,
+      "learning_rate": 9.741426141025332e-08,
+      "loss": 2.2418,
+      "step": 1690
+    },
+    {
+      "epoch": 1.6504381694255112,
+      "grad_norm": 1.721678894466571,
+      "learning_rate": 9.634903495716675e-08,
+      "loss": 2.215,
+      "step": 1695
+    },
+    {
+      "epoch": 1.6553067185978578,
+      "grad_norm": 1.7726004550340564,
+      "learning_rate": 9.530560927938334e-08,
+      "loss": 2.2609,
+      "step": 1700
+    },
+    {
+      "epoch": 1.6601752677702044,
+      "grad_norm": 1.7520211987580594,
+      "learning_rate": 9.428358199725911e-08,
+      "loss": 2.2388,
+      "step": 1705
+    },
+    {
+      "epoch": 1.6650438169425512,
+      "grad_norm": 1.8152868603337584,
+      "learning_rate": 9.328255734631862e-08,
+      "loss": 2.2346,
+      "step": 1710
+    },
+    {
+      "epoch": 1.6699123661148978,
+      "grad_norm": 1.7051369362258122,
+      "learning_rate": 9.230214608189946e-08,
+      "loss": 2.201,
+      "step": 1715
+    },
+    {
+      "epoch": 1.6747809152872444,
+      "grad_norm": 1.7253608618380802,
+      "learning_rate": 9.1341965384976e-08,
+      "loss": 2.2579,
+      "step": 1720
+    },
+    {
+      "epoch": 1.6796494644595912,
+      "grad_norm": 1.6955731274986756,
+      "learning_rate": 9.040163876915256e-08,
+      "loss": 2.2456,
+      "step": 1725
+    },
+    {
+      "epoch": 1.6845180136319375,
+      "grad_norm": 1.8315504305607004,
+      "learning_rate": 8.948079598881221e-08,
+      "loss": 2.2085,
+      "step": 1730
+    },
+    {
+      "epoch": 1.6893865628042843,
+      "grad_norm": 1.680264316266455,
+      "learning_rate": 8.857907294841052e-08,
+      "loss": 2.2427,
+      "step": 1735
+    },
+    {
+      "epoch": 1.694255111976631,
+      "grad_norm": 1.7344138245065615,
+      "learning_rate": 8.76961116129008e-08,
+      "loss": 2.2503,
+      "step": 1740
+    },
+    {
+      "epoch": 1.6991236611489775,
+      "grad_norm": 1.7400368127622114,
+      "learning_rate": 8.683155991928096e-08,
+      "loss": 2.2245,
+      "step": 1745
+    },
+    {
+      "epoch": 1.7039922103213243,
+      "grad_norm": 1.7436841437009907,
+      "learning_rate": 8.598507168924832e-08,
+      "loss": 2.229,
+      "step": 1750
+    },
+    {
+      "epoch": 1.7088607594936709,
+      "grad_norm": 1.718735182879372,
+      "learning_rate": 8.515630654295236e-08,
+      "loss": 2.214,
+      "step": 1755
+    },
+    {
+      "epoch": 1.7137293086660175,
+      "grad_norm": 2.4325656334712416,
+      "learning_rate": 8.43449298138329e-08,
+      "loss": 2.2273,
+      "step": 1760
+    },
+    {
+      "epoch": 1.7185978578383643,
+      "grad_norm": 1.898796758188917,
+      "learning_rate": 8.355061246453264e-08,
+      "loss": 2.2182,
+      "step": 1765
+    },
+    {
+      "epoch": 1.7234664070107109,
+      "grad_norm": 1.7705269671220705,
+      "learning_rate": 8.277303100387306e-08,
+      "loss": 2.2375,
+      "step": 1770
+    },
+    {
+      "epoch": 1.7283349561830574,
+      "grad_norm": 1.7467032834457301,
+      "learning_rate": 8.201186740488162e-08,
+      "loss": 2.2547,
+      "step": 1775
+    },
+    {
+      "epoch": 1.7332035053554042,
+      "grad_norm": 1.7519508473607348,
+      "learning_rate": 8.126680902385984e-08,
+      "loss": 2.2533,
+      "step": 1780
+    },
+    {
+      "epoch": 1.7380720545277506,
+      "grad_norm": 1.6602010890382983,
+      "learning_rate": 8.053754852048118e-08,
+      "loss": 2.2409,
+      "step": 1785
+    },
+    {
+      "epoch": 1.7429406037000974,
+      "grad_norm": 1.7182534815184765,
+      "learning_rate": 7.982378377890754e-08,
+      "loss": 2.2169,
+      "step": 1790
+    },
+    {
+      "epoch": 1.747809152872444,
+      "grad_norm": 1.7132799050190683,
+      "learning_rate": 7.912521782991345e-08,
+      "loss": 2.21,
+      "step": 1795
+    },
+    {
+      "epoch": 1.7526777020447906,
+      "grad_norm": 1.7492486479656935,
+      "learning_rate": 7.844155877400776e-08,
+      "loss": 2.254,
+      "step": 1800
+    },
+    {
+      "epoch": 1.7526777020447906,
+      "eval_loss": 2.4266231060028076,
+      "eval_runtime": 84.8728,
+      "eval_samples_per_second": 85.999,
+      "eval_steps_per_second": 0.683,
+      "step": 1800
+    },
+    {
+      "epoch": 1.7575462512171374,
+      "grad_norm": 1.7982513495954076,
+      "learning_rate": 7.777251970554109e-08,
+      "loss": 2.2446,
+      "step": 1805
+    },
+    {
+      "epoch": 1.762414800389484,
+      "grad_norm": 1.7235517571422887,
+      "learning_rate": 7.711781863779007e-08,
+      "loss": 2.1939,
+      "step": 1810
+    },
+    {
+      "epoch": 1.7672833495618305,
+      "grad_norm": 1.7506770945570296,
+      "learning_rate": 7.64771784290061e-08,
+      "loss": 2.2339,
+      "step": 1815
+    },
+    {
+      "epoch": 1.7721518987341773,
+      "grad_norm": 1.961410951418005,
+      "learning_rate": 7.58503267094196e-08,
+      "loss": 2.2686,
+      "step": 1820
+    },
+    {
+      "epoch": 1.7770204479065237,
+      "grad_norm": 1.8140605591823493,
+      "learning_rate": 7.523699580918899e-08,
+      "loss": 2.2136,
+      "step": 1825
+    },
+    {
+      "epoch": 1.7818889970788705,
+      "grad_norm": 1.733018609908853,
+      "learning_rate": 7.463692268728381e-08,
+      "loss": 2.165,
+      "step": 1830
+    },
+    {
+      "epoch": 1.7867575462512173,
+      "grad_norm": 1.6824331979801876,
+      "learning_rate": 7.404984886129268e-08,
+      "loss": 2.2369,
+      "step": 1835
+    },
+    {
+      "epoch": 1.7916260954235637,
+      "grad_norm": 1.8520431760674112,
+      "learning_rate": 7.347552033814525e-08,
+      "loss": 2.2612,
+      "step": 1840
+    },
+    {
+      "epoch": 1.7964946445959105,
+      "grad_norm": 1.7117045631055423,
+      "learning_rate": 7.291368754573846e-08,
+      "loss": 2.2425,
+      "step": 1845
+    },
+    {
+      "epoch": 1.801363193768257,
+      "grad_norm": 1.7052819169533668,
+      "learning_rate": 7.236410526545777e-08,
+      "loss": 2.196,
+      "step": 1850
+    },
+    {
+      "epoch": 1.8062317429406036,
+      "grad_norm": 1.6384344799557868,
+      "learning_rate": 7.182653256558277e-08,
+      "loss": 2.2563,
+      "step": 1855
+    },
+    {
+      "epoch": 1.8111002921129504,
+      "grad_norm": 1.726079611316107,
+      "learning_rate": 7.130073273556794e-08,
+      "loss": 2.2455,
+      "step": 1860
+    },
+    {
+      "epoch": 1.815968841285297,
+      "grad_norm": 1.8072162191497745,
+      "learning_rate": 7.07864732211891e-08,
+      "loss": 2.2181,
+      "step": 1865
+    },
+    {
+      "epoch": 1.8208373904576436,
+      "grad_norm": 1.720087653439122,
+      "learning_rate": 7.028352556054533e-08,
+      "loss": 2.2295,
+      "step": 1870
+    },
+    {
+      "epoch": 1.8257059396299904,
+      "grad_norm": 1.7654960116615692,
+      "learning_rate": 6.979166532090796e-08,
+      "loss": 2.2391,
+      "step": 1875
+    },
+    {
+      "epoch": 1.8305744888023368,
+      "grad_norm": 1.735053092580652,
+      "learning_rate": 6.931067203640622e-08,
+      "loss": 2.2372,
+      "step": 1880
+    },
+    {
+      "epoch": 1.8354430379746836,
+      "grad_norm": 0.8624508534667337,
+      "learning_rate": 6.884032914654115e-08,
+      "loss": 2.2893,
+      "step": 1885
+    },
+    {
+      "epoch": 1.8403115871470301,
+      "grad_norm": 1.7141597133248032,
+      "learning_rate": 6.838042393551797e-08,
+      "loss": 2.1609,
+      "step": 1890
+    },
+    {
+      "epoch": 1.8451801363193767,
+      "grad_norm": 1.7126007991228827,
+      "learning_rate": 6.793074747238838e-08,
+      "loss": 2.1911,
+      "step": 1895
+    },
+    {
+      "epoch": 1.8500486854917235,
+      "grad_norm": 1.7133412238684549,
+      "learning_rate": 6.749109455199282e-08,
+      "loss": 2.2345,
+      "step": 1900
+    },
+    {
+      "epoch": 1.85491723466407,
+      "grad_norm": 1.7274466496979972,
+      "learning_rate": 6.706126363669515e-08,
+      "loss": 2.2514,
+      "step": 1905
+    },
+    {
+      "epoch": 1.8597857838364167,
+      "grad_norm": 1.7189469301901343,
+      "learning_rate": 6.664105679889947e-08,
+      "loss": 2.2458,
+      "step": 1910
+    },
+    {
+      "epoch": 1.8646543330087635,
+      "grad_norm": 1.690182741334165,
+      "learning_rate": 6.623027966434113e-08,
+      "loss": 2.2038,
+      "step": 1915
+    },
+    {
+      "epoch": 1.86952288218111,
+      "grad_norm": 1.7617528535634877,
+      "learning_rate": 6.582874135614325e-08,
+      "loss": 2.1569,
+      "step": 1920
+    },
+    {
+      "epoch": 1.8743914313534566,
+      "grad_norm": 1.7229096608807917,
+      "learning_rate": 6.54362544396295e-08,
+      "loss": 2.2416,
+      "step": 1925
+    },
+    {
+      "epoch": 1.8792599805258035,
+      "grad_norm": 1.796873830830146,
+      "learning_rate": 6.505263486788538e-08,
+      "loss": 2.2388,
+      "step": 1930
+    },
+    {
+      "epoch": 1.8841285296981498,
+      "grad_norm": 1.6550586704575283,
+      "learning_rate": 6.46777019280588e-08,
+      "loss": 2.2445,
+      "step": 1935
+    },
+    {
+      "epoch": 1.8889970788704966,
+      "grad_norm": 1.7133733906940865,
+      "learning_rate": 6.431127818839212e-08,
+      "loss": 2.1907,
+      "step": 1940
+    },
+    {
+      "epoch": 1.8938656280428432,
+      "grad_norm": 1.7131092326344233,
+      "learning_rate": 6.395318944597702e-08,
+      "loss": 2.2508,
+      "step": 1945
+    },
+    {
+      "epoch": 1.8987341772151898,
+      "grad_norm": 1.7975356797064426,
+      "learning_rate": 6.360326467522418e-08,
+      "loss": 2.2619,
+      "step": 1950
+    },
+    {
+      "epoch": 1.9036027263875366,
+      "grad_norm": 1.7927617071731754,
+      "learning_rate": 6.326133597703928e-08,
+      "loss": 2.2306,
+      "step": 1955
+    },
+    {
+      "epoch": 1.9084712755598832,
+      "grad_norm": 1.7327613795288057,
+      "learning_rate": 6.292723852869772e-08,
+      "loss": 2.2258,
+      "step": 1960
+    },
+    {
+      "epoch": 1.9133398247322297,
+      "grad_norm": 1.7114888287588472,
+      "learning_rate": 6.260081053440938e-08,
+      "loss": 2.2258,
+      "step": 1965
+    },
+    {
+      "epoch": 1.9182083739045765,
+      "grad_norm": 1.7617910174679117,
+      "learning_rate": 6.228189317656625e-08,
+      "loss": 2.2214,
+      "step": 1970
+    },
+    {
+      "epoch": 1.9230769230769231,
+      "grad_norm": 1.69813136105815,
+      "learning_rate": 6.197033056766434e-08,
+      "loss": 2.2189,
+      "step": 1975
+    },
+    {
+      "epoch": 1.9279454722492697,
+      "grad_norm": 1.668741197040639,
+      "learning_rate": 6.166596970289254e-08,
+      "loss": 2.2168,
+      "step": 1980
+    },
+    {
+      "epoch": 1.9328140214216165,
+      "grad_norm": 1.7177940007984969,
+      "learning_rate": 6.136866041338061e-08,
+      "loss": 2.2131,
+      "step": 1985
+    },
+    {
+      "epoch": 1.9376825705939629,
+      "grad_norm": 2.097413251733257,
+      "learning_rate": 6.107825532009846e-08,
+      "loss": 2.2431,
+      "step": 1990
+    },
+    {
+      "epoch": 1.9425511197663097,
+      "grad_norm": 1.6798259076762905,
+      "learning_rate": 6.079460978839945e-08,
+      "loss": 2.273,
+      "step": 1995
+    },
+    {
+      "epoch": 1.9474196689386563,
+      "grad_norm": 1.7260224966683506,
+      "learning_rate": 6.051758188319985e-08,
+      "loss": 2.2508,
+      "step": 2000
+    },
+    {
+      "epoch": 1.9474196689386563,
+      "eval_loss": 2.4265518188476562,
+      "eval_runtime": 85.2652,
+      "eval_samples_per_second": 85.604,
+      "eval_steps_per_second": 0.68,
+      "step": 2000
+    },
+    {
+      "epoch": 1.9522882181110028,
+      "grad_norm": 1.6953303266550364,
+      "learning_rate": 6.024703232478723e-08,
+      "loss": 2.2536,
+      "step": 2005
+    },
+    {
+      "epoch": 1.9571567672833496,
+      "grad_norm": 1.7537091815900079,
+      "learning_rate": 5.998282444525052e-08,
+      "loss": 2.1932,
+      "step": 2010
+    },
+    {
+      "epoch": 1.9620253164556962,
+      "grad_norm": 1.6913312559085918,
+      "learning_rate": 5.97248241455241e-08,
+      "loss": 2.2097,
+      "step": 2015
+    },
+    {
+      "epoch": 1.9668938656280428,
+      "grad_norm": 1.709640143824205,
+      "learning_rate": 5.947289985303898e-08,
+      "loss": 2.2739,
+      "step": 2020
+    },
+    {
+      "epoch": 1.9717624148003896,
+      "grad_norm": 1.769591500268775,
+      "learning_rate": 5.922692247997387e-08,
+      "loss": 2.2345,
+      "step": 2025
+    },
+    {
+      "epoch": 1.976630963972736,
+      "grad_norm": 1.7089545500913113,
+      "learning_rate": 5.898676538209882e-08,
+      "loss": 2.1989,
+      "step": 2030
+    },
+    {
+      "epoch": 1.9814995131450828,
+      "grad_norm": 1.7324755039666533,
+      "learning_rate": 5.8752304318204655e-08,
+      "loss": 2.2296,
+      "step": 2035
+    },
+    {
+      "epoch": 1.9863680623174296,
+      "grad_norm": 1.7254800842506772,
+      "learning_rate": 5.852341741011111e-08,
+      "loss": 2.193,
+      "step": 2040
+    },
+    {
+      "epoch": 1.991236611489776,
+      "grad_norm": 1.7451878538885703,
+      "learning_rate": 5.829998510324686e-08,
+      "loss": 2.2829,
+      "step": 2045
+    },
+    {
+      "epoch": 1.9961051606621227,
+      "grad_norm": 1.7607671547384287,
+      "learning_rate": 5.8081890127794496e-08,
+      "loss": 2.2133,
+      "step": 2050
+    },
+    {
+      "epoch": 2.0009737098344695,
+      "grad_norm": 1.7100416555620679,
+      "learning_rate": 5.7869017460393855e-08,
+      "loss": 2.233,
+      "step": 2055
+    },
+    {
+      "epoch": 2.005842259006816,
+      "grad_norm": 1.746899480839477,
+      "learning_rate": 5.7661254286396653e-08,
+      "loss": 2.2424,
+      "step": 2060
+    },
+    {
+      "epoch": 2.0107108081791627,
+      "grad_norm": 1.7305460820757212,
+      "learning_rate": 5.74584899626664e-08,
+      "loss": 2.1762,
+      "step": 2065
+    },
+    {
+      "epoch": 2.015579357351509,
+      "grad_norm": 1.7063747134288176,
+      "learning_rate": 5.726061598091625e-08,
+      "loss": 2.2345,
+      "step": 2070
+    },
+    {
+      "epoch": 2.020447906523856,
+      "grad_norm": 1.7115321830413128,
+      "learning_rate": 5.706752593157899e-08,
+      "loss": 2.2415,
+      "step": 2075
+    },
+    {
+      "epoch": 2.0253164556962027,
+      "grad_norm": 1.7315759852916324,
+      "learning_rate": 5.687911546820225e-08,
+      "loss": 2.2013,
+      "step": 2080
+    },
+    {
+      "epoch": 2.030185004868549,
+      "grad_norm": 1.7053153190560344,
+      "learning_rate": 5.669528227236286e-08,
+      "loss": 2.1833,
+      "step": 2085
+    },
+    {
+      "epoch": 2.035053554040896,
+      "grad_norm": 1.7480977685627526,
+      "learning_rate": 5.6515926019093655e-08,
+      "loss": 2.211,
+      "step": 2090
+    },
+    {
+      "epoch": 2.0399221032132426,
+      "grad_norm": 1.7325131107565894,
+      "learning_rate": 5.6340948342816956e-08,
+      "loss": 2.1746,
+      "step": 2095
+    },
+    {
+      "epoch": 2.044790652385589,
+      "grad_norm": 1.7861322062259097,
+      "learning_rate": 5.617025280377817e-08,
+      "loss": 2.2137,
+      "step": 2100
+    },
+    {
+      "epoch": 2.049659201557936,
+      "grad_norm": 1.7407845229020498,
+      "learning_rate": 5.6003744854973425e-08,
+      "loss": 2.2327,
+      "step": 2105
+    },
+    {
+      "epoch": 2.0545277507302826,
+      "grad_norm": 1.7703593488238105,
+      "learning_rate": 5.584133180956534e-08,
+      "loss": 2.2133,
+      "step": 2110
+    },
+    {
+      "epoch": 2.059396299902629,
+      "grad_norm": 1.7217536220585616,
+      "learning_rate": 5.568292280878073e-08,
+      "loss": 2.1892,
+      "step": 2115
+    },
+    {
+      "epoch": 2.0642648490749758,
+      "grad_norm": 1.7276020192361135,
+      "learning_rate": 5.552842879028437e-08,
+      "loss": 2.2204,
+      "step": 2120
+    },
+    {
+      "epoch": 2.069133398247322,
+      "grad_norm": 1.727960780515485,
+      "learning_rate": 5.537776245702285e-08,
+      "loss": 2.1996,
+      "step": 2125
+    },
+    {
+      "epoch": 2.074001947419669,
+      "grad_norm": 1.7514773544281632,
+      "learning_rate": 5.523083824653292e-08,
+      "loss": 2.1812,
+      "step": 2130
+    },
+    {
+      "epoch": 2.0788704965920157,
+      "grad_norm": 1.7204243933149874,
+      "learning_rate": 5.5087572300708064e-08,
+      "loss": 2.2412,
+      "step": 2135
+    },
+    {
+      "epoch": 2.083739045764362,
+      "grad_norm": 1.7770689913358608,
+      "learning_rate": 5.49478824360182e-08,
+      "loss": 2.2175,
+      "step": 2140
+    },
+    {
+      "epoch": 2.088607594936709,
+      "grad_norm": 1.8118845495638691,
+      "learning_rate": 5.4811688114176284e-08,
+      "loss": 2.2016,
+      "step": 2145
+    },
+    {
+      "epoch": 2.0934761441090557,
+      "grad_norm": 1.746154499517824,
+      "learning_rate": 5.46789104132466e-08,
+      "loss": 2.2338,
+      "step": 2150
+    },
+    {
+      "epoch": 2.098344693281402,
+      "grad_norm": 1.7040405445808213,
+      "learning_rate": 5.454947199918886e-08,
+      "loss": 2.2298,
+      "step": 2155
+    },
+    {
+      "epoch": 2.103213242453749,
+      "grad_norm": 1.6905682803922102,
+      "learning_rate": 5.4423297097832996e-08,
+      "loss": 2.2539,
+      "step": 2160
+    },
+    {
+      "epoch": 2.108081791626095,
+      "grad_norm": 1.7302575758498189,
+      "learning_rate": 5.430031146727882e-08,
+      "loss": 2.2279,
+      "step": 2165
+    },
+    {
+      "epoch": 2.112950340798442,
+      "grad_norm": 1.771661817904592,
+      "learning_rate": 5.418044237071543e-08,
+      "loss": 2.1828,
+      "step": 2170
+    },
+    {
+      "epoch": 2.117818889970789,
+      "grad_norm": 1.726387016486399,
+      "learning_rate": 5.406361854965489e-08,
+      "loss": 2.2167,
+      "step": 2175
+    },
+    {
+      "epoch": 2.122687439143135,
+      "grad_norm": 1.72736148684106,
+      "learning_rate": 5.394977019757503e-08,
+      "loss": 2.1627,
+      "step": 2180
+    },
+    {
+      "epoch": 2.127555988315482,
+      "grad_norm": 1.732688655222483,
+      "learning_rate": 5.3838828933965965e-08,
+      "loss": 2.2482,
+      "step": 2185
+    },
+    {
+      "epoch": 2.132424537487829,
+      "grad_norm": 1.765418024490754,
+      "learning_rate": 5.373072777877539e-08,
+      "loss": 2.2416,
+      "step": 2190
+    },
+    {
+      "epoch": 2.137293086660175,
+      "grad_norm": 1.7905409309741185,
+      "learning_rate": 5.3625401127247335e-08,
+      "loss": 2.2049,
+      "step": 2195
+    },
+    {
+      "epoch": 2.142161635832522,
+      "grad_norm": 1.7285450482838325,
+      "learning_rate": 5.3522784725149425e-08,
+      "loss": 2.2112,
+      "step": 2200
+    },
+    {
+      "epoch": 2.142161635832522,
+      "eval_loss": 2.4287049770355225,
+      "eval_runtime": 85.2698,
+      "eval_samples_per_second": 85.599,
+      "eval_steps_per_second": 0.68,
+      "step": 2200
+    },
+    {
+      "epoch": 2.1470301850048688,
+      "grad_norm": 1.7298233620729722,
+      "learning_rate": 5.3422815644383576e-08,
+      "loss": 2.2268,
+      "step": 2205
+    },
+    {
+      "epoch": 2.151898734177215,
+      "grad_norm": 1.7106243642725834,
+      "learning_rate": 5.332543225897528e-08,
+      "loss": 2.1823,
+      "step": 2210
+    },
+    {
+      "epoch": 2.156767283349562,
+      "grad_norm": 1.7093421731817005,
+      "learning_rate": 5.3230574221436374e-08,
+      "loss": 2.2282,
+      "step": 2215
+    },
+    {
+      "epoch": 2.1616358325219083,
+      "grad_norm": 1.7311568429821071,
+      "learning_rate": 5.313818243949664e-08,
+      "loss": 2.2213,
+      "step": 2220
+    },
+    {
+      "epoch": 2.166504381694255,
+      "grad_norm": 1.7674044638597066,
+      "learning_rate": 5.3048199053199294e-08,
+      "loss": 2.2254,
+      "step": 2225
+    },
+    {
+      "epoch": 2.171372930866602,
+      "grad_norm": 1.7115381111757135,
+      "learning_rate": 5.296056741235573e-08,
+      "loss": 2.2418,
+      "step": 2230
+    },
+    {
+      "epoch": 2.1762414800389482,
+      "grad_norm": 1.6790013492814684,
+      "learning_rate": 5.2875232054354564e-08,
+      "loss": 2.2243,
+      "step": 2235
+    },
+    {
+      "epoch": 2.181110029211295,
+      "grad_norm": 1.7513052402271196,
+      "learning_rate": 5.279213868232073e-08,
+      "loss": 2.247,
+      "step": 2240
+    },
+    {
+      "epoch": 2.185978578383642,
+      "grad_norm": 1.8041144340469755,
+      "learning_rate": 5.271123414361961e-08,
+      "loss": 2.2318,
+      "step": 2245
+    },
+    {
+      "epoch": 2.190847127555988,
+      "grad_norm": 1.711266621732132,
+      "learning_rate": 5.263246640870184e-08,
+      "loss": 2.2393,
+      "step": 2250
+    },
+    {
+      "epoch": 2.195715676728335,
+      "grad_norm": 1.6969264596911122,
+      "learning_rate": 5.2555784550284366e-08,
+      "loss": 2.2087,
+      "step": 2255
+    },
+    {
+      "epoch": 2.2005842259006814,
+      "grad_norm": 1.7651857140540494,
+      "learning_rate": 5.2481138722863035e-08,
+      "loss": 2.2069,
+      "step": 2260
+    },
+    {
+      "epoch": 2.205452775073028,
+      "grad_norm": 1.7817072494501394,
+      "learning_rate": 5.240848014255253e-08,
+      "loss": 2.2051,
+      "step": 2265
+    },
+    {
+      "epoch": 2.210321324245375,
+      "grad_norm": 1.7343663502264346,
+      "learning_rate": 5.233776106724918e-08,
+      "loss": 2.2389,
+      "step": 2270
+    },
+    {
+      "epoch": 2.2151898734177213,
+      "grad_norm": 1.6921255572364247,
+      "learning_rate": 5.2268934777112365e-08,
+      "loss": 2.2362,
+      "step": 2275
+    },
+    {
+      "epoch": 2.220058422590068,
+      "grad_norm": 1.7013539832560813,
+      "learning_rate": 5.220195555536015e-08,
+      "loss": 2.2519,
+      "step": 2280
+    },
+    {
+      "epoch": 2.224926971762415,
+      "grad_norm": 1.785593606024491,
+      "learning_rate": 5.213677866937508e-08,
+      "loss": 2.2231,
+      "step": 2285
+    },
+    {
+      "epoch": 2.2297955209347613,
+      "grad_norm": 1.7400661751479558,
+      "learning_rate": 5.207336035211581e-08,
+      "loss": 2.2306,
+      "step": 2290
+    },
+    {
+      "epoch": 2.234664070107108,
+      "grad_norm": 1.7513785103045856,
+      "learning_rate": 5.201165778383047e-08,
+      "loss": 2.184,
+      "step": 2295
+    },
+    {
+      "epoch": 2.239532619279455,
+      "grad_norm": 1.8372816676992572,
+      "learning_rate": 5.19516290740677e-08,
+      "loss": 2.1913,
+      "step": 2300
+    },
+    {
+      "epoch": 2.2444011684518013,
+      "grad_norm": 1.7411317592151172,
+      "learning_rate": 5.189323324398133e-08,
+      "loss": 2.1749,
+      "step": 2305
+    },
+    {
+      "epoch": 2.249269717624148,
+      "grad_norm": 1.7262426588954412,
+      "learning_rate": 5.1836430208924614e-08,
+      "loss": 2.1902,
+      "step": 2310
+    },
+    {
+      "epoch": 2.2541382667964944,
+      "grad_norm": 1.7496058178900977,
+      "learning_rate": 5.1781180761330104e-08,
+      "loss": 2.2354,
+      "step": 2315
+    },
+    {
+      "epoch": 2.2590068159688412,
+      "grad_norm": 1.7515829381648897,
+      "learning_rate": 5.1727446553871265e-08,
+      "loss": 2.1717,
+      "step": 2320
+    },
+    {
+      "epoch": 2.263875365141188,
+      "grad_norm": 1.7101981117837521,
+      "learning_rate": 5.1675190082901985e-08,
+      "loss": 2.1983,
+      "step": 2325
+    },
+    {
+      "epoch": 2.2687439143135344,
+      "grad_norm": 1.674731414303109,
+      "learning_rate": 5.1624374672169996e-08,
+      "loss": 2.1964,
+      "step": 2330
+    },
+    {
+      "epoch": 2.273612463485881,
+      "grad_norm": 1.8194729895132122,
+      "learning_rate": 5.157496445680062e-08,
+      "loss": 2.208,
+      "step": 2335
+    },
+    {
+      "epoch": 2.278481012658228,
+      "grad_norm": 1.7198751813449369,
+      "learning_rate": 5.152692436754697e-08,
+      "loss": 2.2088,
+      "step": 2340
+    },
+    {
+      "epoch": 2.2833495618305744,
+      "grad_norm": 1.7052083287832007,
+      "learning_rate": 5.1480220115302935e-08,
+      "loss": 2.2412,
+      "step": 2345
+    },
+    {
+      "epoch": 2.288218111002921,
+      "grad_norm": 1.6994269312553734,
+      "learning_rate": 5.143481817587523e-08,
+      "loss": 2.2523,
+      "step": 2350
+    },
+    {
+      "epoch": 2.293086660175268,
+      "grad_norm": 1.7821059623881625,
+      "learning_rate": 5.139068577501104e-08,
+      "loss": 2.2306,
+      "step": 2355
+    },
+    {
+      "epoch": 2.2979552093476143,
+      "grad_norm": 1.7655828205712845,
+      "learning_rate": 5.134779087367746e-08,
+      "loss": 2.2192,
+      "step": 2360
+    },
+    {
+      "epoch": 2.302823758519961,
+      "grad_norm": 1.764331702411005,
+      "learning_rate": 5.130610215358936e-08,
+      "loss": 2.2528,
+      "step": 2365
+    },
+    {
+      "epoch": 2.3076923076923075,
+      "grad_norm": 1.814774977446024,
+      "learning_rate": 5.126558900298217e-08,
+      "loss": 2.2311,
+      "step": 2370
+    },
+    {
+      "epoch": 2.3125608568646543,
+      "grad_norm": 1.7749398004090056,
+      "learning_rate": 5.122622150262591e-08,
+      "loss": 2.1965,
+      "step": 2375
+    },
+    {
+      "epoch": 2.317429406037001,
+      "grad_norm": 1.7732754323290614,
+      "learning_rate": 5.118797041207741e-08,
+      "loss": 2.2307,
+      "step": 2380
+    },
+    {
+      "epoch": 2.3222979552093475,
+      "grad_norm": 1.7204661711193003,
+      "learning_rate": 5.1150807156166916e-08,
+      "loss": 2.1813,
+      "step": 2385
+    },
+    {
+      "epoch": 2.3271665043816943,
+      "grad_norm": 1.7830208677372077,
+      "learning_rate": 5.111470381171611e-08,
+      "loss": 2.1976,
+      "step": 2390
+    },
+    {
+      "epoch": 2.332035053554041,
+      "grad_norm": 1.7100457234343944,
+      "learning_rate": 5.1079633094484e-08,
+      "loss": 2.2421,
+      "step": 2395
+    },
+    {
+      "epoch": 2.3369036027263874,
+      "grad_norm": 1.7579822935018907,
+      "learning_rate": 5.104556834633745e-08,
+      "loss": 2.2063,
+      "step": 2400
+    },
+    {
+      "epoch": 2.3369036027263874,
+      "eval_loss": 2.4292638301849365,
+      "eval_runtime": 85.8612,
+      "eval_samples_per_second": 85.009,
+      "eval_steps_per_second": 0.676,
+      "step": 2400
+    },
+    {
+      "epoch": 2.3417721518987342,
+      "grad_norm": 1.7345007813133775,
+      "learning_rate": 5.101248352264326e-08,
+      "loss": 2.2243,
+      "step": 2405
+    },
+    {
+      "epoch": 2.346640701071081,
+      "grad_norm": 1.7248955210954013,
+      "learning_rate": 5.098035317987838e-08,
+      "loss": 2.2096,
+      "step": 2410
+    },
+    {
+      "epoch": 2.3515092502434274,
+      "grad_norm": 1.7460382186199548,
+      "learning_rate": 5.0949152463455285e-08,
+      "loss": 2.2034,
+      "step": 2415
+    },
+    {
+      "epoch": 2.356377799415774,
+      "grad_norm": 1.7441544396124007,
+      "learning_rate": 5.09188570957593e-08,
+      "loss": 2.1584,
+      "step": 2420
+    },
+    {
+      "epoch": 2.3612463485881205,
+      "grad_norm": 1.7037153131860179,
+      "learning_rate": 5.0889443364394804e-08,
+      "loss": 2.222,
+      "step": 2425
+    },
+    {
+      "epoch": 2.3661148977604674,
+      "grad_norm": 1.7280060531644463,
+      "learning_rate": 5.0860888110637265e-08,
+      "loss": 2.2324,
+      "step": 2430
+    },
+    {
+      "epoch": 2.370983446932814,
+      "grad_norm": 1.7906585557070347,
+      "learning_rate": 5.083316871808814e-08,
+      "loss": 2.2047,
+      "step": 2435
+    },
+    {
+      "epoch": 2.3758519961051605,
+      "grad_norm": 1.7845861555659082,
+      "learning_rate": 5.080626310152955e-08,
+      "loss": 2.1902,
+      "step": 2440
+    },
+    {
+      "epoch": 2.3807205452775073,
+      "grad_norm": 1.6806913458915649,
+      "learning_rate": 5.078014969597595e-08,
+      "loss": 2.2,
+      "step": 2445
+    },
+    {
+      "epoch": 2.385589094449854,
+      "grad_norm": 1.7905906194923593,
+      "learning_rate": 5.075480744591971e-08,
+      "loss": 2.1989,
+      "step": 2450
+    },
+    {
+      "epoch": 2.3904576436222005,
+      "grad_norm": 1.7807103616717044,
+      "learning_rate": 5.073021579476786e-08,
+      "loss": 2.2194,
+      "step": 2455
+    },
+    {
+      "epoch": 2.3953261927945473,
+      "grad_norm": 1.6957268999562412,
+      "learning_rate": 5.070635467446715e-08,
+      "loss": 2.1982,
+      "step": 2460
+    },
+    {
+      "epoch": 2.400194741966894,
+      "grad_norm": 1.775675901114813,
+      "learning_rate": 5.0683204495314504e-08,
+      "loss": 2.213,
+      "step": 2465
+    },
+    {
+      "epoch": 2.4050632911392404,
+      "grad_norm": 1.7829403999278695,
+      "learning_rate": 5.06607461359503e-08,
+      "loss": 2.2226,
+      "step": 2470
+    },
+    {
+      "epoch": 2.4099318403115872,
+      "grad_norm": 1.766705522512629,
+      "learning_rate": 5.06389609335315e-08,
+      "loss": 2.1896,
+      "step": 2475
+    },
+    {
+      "epoch": 2.4148003894839336,
+      "grad_norm": 1.7827496357552255,
+      "learning_rate": 5.0617830674082116e-08,
+      "loss": 2.1947,
+      "step": 2480
+    },
+    {
+      "epoch": 2.4196689386562804,
+      "grad_norm": 1.7451999252246684,
+      "learning_rate": 5.059733758301827e-08,
+      "loss": 2.2393,
+      "step": 2485
+    },
+    {
+      "epoch": 2.424537487828627,
+      "grad_norm": 1.721359686507254,
+      "learning_rate": 5.057746431584517e-08,
+      "loss": 2.2388,
+      "step": 2490
+    },
+    {
+      "epoch": 2.4294060370009736,
+      "grad_norm": 1.7903488742098908,
+      "learning_rate": 5.055819394902345e-08,
+      "loss": 2.1765,
+      "step": 2495
+    },
+    {
+      "epoch": 2.4342745861733204,
+      "grad_norm": 1.7278059859253259,
+      "learning_rate": 5.053950997100227e-08,
+      "loss": 2.188,
+      "step": 2500
+    },
+    {
+      "epoch": 2.439143135345667,
+      "grad_norm": 1.723905932162126,
+      "learning_rate": 5.052139627341665e-08,
+      "loss": 2.1994,
+      "step": 2505
+    },
+    {
+      "epoch": 2.4440116845180135,
+      "grad_norm": 1.8796866769396487,
+      "learning_rate": 5.050383714244649e-08,
+      "loss": 2.2311,
+      "step": 2510
+    },
+    {
+      "epoch": 2.4488802336903603,
+      "grad_norm": 1.7459716379486228,
+      "learning_rate": 5.0486817250334816e-08,
+      "loss": 2.2485,
+      "step": 2515
+    },
+    {
+      "epoch": 2.453748782862707,
+      "grad_norm": 1.7365026105109531,
+      "learning_rate": 5.047032164706284e-08,
+      "loss": 2.1676,
+      "step": 2520
+    },
+    {
+      "epoch": 2.4586173320350535,
+      "grad_norm": 1.7081299835491885,
+      "learning_rate": 5.045433575217931e-08,
+      "loss": 2.1875,
+      "step": 2525
+    },
+    {
+      "epoch": 2.4634858812074003,
+      "grad_norm": 1.744755862782243,
+      "learning_rate": 5.043884534678184e-08,
+      "loss": 2.245,
+      "step": 2530
+    },
+    {
+      "epoch": 2.4683544303797467,
+      "grad_norm": 1.7422537825667799,
+      "learning_rate": 5.042383656564784e-08,
+      "loss": 2.217,
+      "step": 2535
+    },
+    {
+      "epoch": 2.4732229795520935,
+      "grad_norm": 1.734841704470881,
+      "learning_rate": 5.040929588951272e-08,
+      "loss": 2.2173,
+      "step": 2540
+    },
+    {
+      "epoch": 2.4780915287244403,
+      "grad_norm": 1.779810459600651,
+      "learning_rate": 5.039521013749303e-08,
+      "loss": 2.247,
+      "step": 2545
+    },
+    {
+      "epoch": 2.4829600778967866,
+      "grad_norm": 1.8117164085811932,
+      "learning_rate": 5.0381566459652284e-08,
+      "loss": 2.237,
+      "step": 2550
+    },
+    {
+      "epoch": 2.4878286270691334,
+      "grad_norm": 1.7287658099278131,
+      "learning_rate": 5.0368352329707235e-08,
+      "loss": 2.1902,
+      "step": 2555
+    },
+    {
+      "epoch": 2.49269717624148,
+      "grad_norm": 1.7649214315832387,
+      "learning_rate": 5.0355555537872345e-08,
+      "loss": 2.2109,
+      "step": 2560
+    },
+    {
+      "epoch": 2.4975657254138266,
+      "grad_norm": 1.662642992634986,
+      "learning_rate": 5.0343164183840344e-08,
+      "loss": 2.197,
+      "step": 2565
+    },
+    {
+      "epoch": 2.5024342745861734,
+      "grad_norm": 1.7337104838406339,
+      "learning_rate": 5.033116666989654e-08,
+      "loss": 2.2101,
+      "step": 2570
+    },
+    {
+      "epoch": 2.50730282375852,
+      "grad_norm": 1.8134178787153215,
+      "learning_rate": 5.031955169416503e-08,
+      "loss": 2.1644,
+      "step": 2575
+    },
+    {
+      "epoch": 2.5121713729308666,
+      "grad_norm": 1.780434641418463,
+      "learning_rate": 5.0308308243984355e-08,
+      "loss": 2.2137,
+      "step": 2580
+    },
+    {
+      "epoch": 2.5170399221032134,
+      "grad_norm": 1.744873235024722,
+      "learning_rate": 5.0297425589410844e-08,
+      "loss": 2.1802,
+      "step": 2585
+    },
+    {
+      "epoch": 2.5219084712755597,
+      "grad_norm": 1.7946463871437135,
+      "learning_rate": 5.0286893276847386e-08,
+      "loss": 2.2194,
+      "step": 2590
+    },
+    {
+      "epoch": 2.5267770204479065,
+      "grad_norm": 1.7550387629748228,
+      "learning_rate": 5.0276701122795665e-08,
+      "loss": 2.1656,
+      "step": 2595
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "grad_norm": 1.7656891772283707,
+      "learning_rate": 5.02668392077299e-08,
+      "loss": 2.2544,
+      "step": 2600
+    },
+    {
+      "epoch": 2.5316455696202533,
+      "eval_loss": 2.429072141647339,
+      "eval_runtime": 85.9551,
+      "eval_samples_per_second": 84.916,
+      "eval_steps_per_second": 0.675,
+      "step": 2600
+    },
+    {
+      "epoch": 2.5365141187925997,
+      "grad_norm": 1.8122468666175253,
+      "learning_rate": 5.025729787009003e-08,
+      "loss": 2.2305,
+      "step": 2605
+    },
+    {
+      "epoch": 2.5413826679649465,
+      "grad_norm": 1.6973041191639477,
+      "learning_rate": 5.024806770039247e-08,
+      "loss": 2.2292,
+      "step": 2610
+    },
+    {
+      "epoch": 2.546251217137293,
+      "grad_norm": 1.729259299213179,
+      "learning_rate": 5.023913953545651e-08,
+      "loss": 2.2614,
+      "step": 2615
+    },
+    {
+      "epoch": 2.5511197663096397,
+      "grad_norm": 1.7611375878198783,
+      "learning_rate": 5.023050445274437e-08,
+      "loss": 2.2225,
+      "step": 2620
+    },
+    {
+      "epoch": 2.5559883154819865,
+      "grad_norm": 1.7936750412570266,
+      "learning_rate": 5.022215376481317e-08,
+      "loss": 2.2203,
+      "step": 2625
+    },
+    {
+      "epoch": 2.5608568646543333,
+      "grad_norm": 1.7697623039425605,
+      "learning_rate": 5.021407901387688e-08,
+      "loss": 2.1863,
+      "step": 2630
+    },
+    {
+      "epoch": 2.5657254138266796,
+      "grad_norm": 1.7467899336813517,
+      "learning_rate": 5.02062719664764e-08,
+      "loss": 2.2103,
+      "step": 2635
+    },
+    {
+      "epoch": 2.5705939629990264,
+      "grad_norm": 1.7889652968951377,
+      "learning_rate": 5.019872460825613e-08,
+      "loss": 2.2229,
+      "step": 2640
+    },
+    {
+      "epoch": 2.575462512171373,
+      "grad_norm": 1.722371905671811,
+      "learning_rate": 5.019142913884503e-08,
+      "loss": 2.1491,
+      "step": 2645
+    },
+    {
+      "epoch": 2.5803310613437196,
+      "grad_norm": 1.7502842489092898,
+      "learning_rate": 5.018437796684058e-08,
+      "loss": 2.2225,
+      "step": 2650
+    },
+    {
+      "epoch": 2.5851996105160664,
+      "grad_norm": 1.733643511652611,
+      "learning_rate": 5.0177563704893894e-08,
+      "loss": 2.2308,
+      "step": 2655
+    },
+    {
+      "epoch": 2.5900681596884128,
+      "grad_norm": 1.7507169934067448,
+      "learning_rate": 5.0170979164894177e-08,
+      "loss": 2.2319,
+      "step": 2660
+    },
+    {
+      "epoch": 2.5949367088607596,
+      "grad_norm": 1.7597206213080367,
+      "learning_rate": 5.016461735325101e-08,
+      "loss": 2.2221,
+      "step": 2665
+    },
+    {
+      "epoch": 2.599805258033106,
+      "grad_norm": 1.7484941859889187,
+      "learning_rate": 5.0158471466272625e-08,
+      "loss": 2.2408,
+      "step": 2670
+    },
+    {
+      "epoch": 2.6046738072054527,
+      "grad_norm": 1.7575142192005702,
+      "learning_rate": 5.0152534885638713e-08,
+      "loss": 2.2199,
+      "step": 2675
+    },
+    {
+      "epoch": 2.6095423563777995,
+      "grad_norm": 1.7220313810152217,
+      "learning_rate": 5.014680117396598e-08,
+      "loss": 2.172,
+      "step": 2680
+    },
+    {
+      "epoch": 2.6144109055501463,
+      "grad_norm": 1.7246730790979932,
+      "learning_rate": 5.0141264070464985e-08,
+      "loss": 2.2053,
+      "step": 2685
+    },
+    {
+      "epoch": 2.6192794547224927,
+      "grad_norm": 1.733539469198201,
+      "learning_rate": 5.013591748668665e-08,
+      "loss": 2.1883,
+      "step": 2690
+    },
+    {
+      "epoch": 2.6241480038948395,
+      "grad_norm": 1.708722183137744,
+      "learning_rate": 5.0130755502356856e-08,
+      "loss": 2.2084,
+      "step": 2695
+    },
+    {
+      "epoch": 2.629016553067186,
+      "grad_norm": 1.743046661098172,
+      "learning_rate": 5.0125772361297664e-08,
+      "loss": 2.1743,
+      "step": 2700
+    },
+    {
+      "epoch": 2.6338851022395326,
+      "grad_norm": 1.8128256440501649,
+      "learning_rate": 5.0120962467433614e-08,
+      "loss": 2.2052,
+      "step": 2705
+    },
+    {
+      "epoch": 2.6387536514118795,
+      "grad_norm": 1.7757973503695443,
+      "learning_rate": 5.0116320380881606e-08,
+      "loss": 2.2504,
+      "step": 2710
+    },
+    {
+      "epoch": 2.643622200584226,
+      "grad_norm": 1.7774946391866604,
+      "learning_rate": 5.0111840814122985e-08,
+      "loss": 2.1926,
+      "step": 2715
+    },
+    {
+      "epoch": 2.6484907497565726,
+      "grad_norm": 1.7302618686374809,
+      "learning_rate": 5.010751862825623e-08,
+      "loss": 2.1849,
+      "step": 2720
+    },
+    {
+      "epoch": 2.653359298928919,
+      "grad_norm": 1.7071516593785698,
+      "learning_rate": 5.0103348829328986e-08,
+      "loss": 2.1692,
+      "step": 2725
+    },
+    {
+      "epoch": 2.6582278481012658,
+      "grad_norm": 1.721147660276721,
+      "learning_rate": 5.009932656474795e-08,
+      "loss": 2.2232,
+      "step": 2730
+    },
+    {
+      "epoch": 2.6630963972736126,
+      "grad_norm": 1.8757024025402924,
+      "learning_rate": 5.0095447119765244e-08,
+      "loss": 2.193,
+      "step": 2735
+    },
+    {
+      "epoch": 2.667964946445959,
+      "grad_norm": 1.7734949594595664,
+      "learning_rate": 5.009170591403991e-08,
+      "loss": 2.2176,
+      "step": 2740
+    },
+    {
+      "epoch": 2.6728334956183057,
+      "grad_norm": 1.7300923376938375,
+      "learning_rate": 5.0088098498273176e-08,
+      "loss": 2.2405,
+      "step": 2745
+    },
+    {
+      "epoch": 2.6777020447906525,
+      "grad_norm": 1.715381643484051,
+      "learning_rate": 5.008462055091624e-08,
+      "loss": 2.1716,
+      "step": 2750
+    },
+    {
+      "epoch": 2.682570593962999,
+      "grad_norm": 1.7201490901820933,
+      "learning_rate": 5.00812678749491e-08,
+      "loss": 2.2163,
+      "step": 2755
+    },
+    {
+      "epoch": 2.6874391431353457,
+      "grad_norm": 1.776422114245558,
+      "learning_rate": 5.0078036394729376e-08,
+      "loss": 2.2385,
+      "step": 2760
+    },
+    {
+      "epoch": 2.6923076923076925,
+      "grad_norm": 1.7434342909806673,
+      "learning_rate": 5.007492215290964e-08,
+      "loss": 2.2148,
+      "step": 2765
+    },
+    {
+      "epoch": 2.697176241480039,
+      "grad_norm": 1.7303585866040545,
+      "learning_rate": 5.0071921307422136e-08,
+      "loss": 2.2202,
+      "step": 2770
+    },
+    {
+      "epoch": 2.7020447906523857,
+      "grad_norm": 1.7653235238408616,
+      "learning_rate": 5.00690301285296e-08,
+      "loss": 2.2438,
+      "step": 2775
+    },
+    {
+      "epoch": 2.706913339824732,
+      "grad_norm": 1.732574656691103,
+      "learning_rate": 5.006624499594101e-08,
+      "loss": 2.1988,
+      "step": 2780
+    },
+    {
+      "epoch": 2.711781888997079,
+      "grad_norm": 1.7030324021473955,
+      "learning_rate": 5.0063562395991076e-08,
+      "loss": 2.2623,
+      "step": 2785
+    },
+    {
+      "epoch": 2.7166504381694256,
+      "grad_norm": 1.9331342916421579,
+      "learning_rate": 5.006097891888214e-08,
+      "loss": 2.1995,
+      "step": 2790
+    },
+    {
+      "epoch": 2.721518987341772,
+      "grad_norm": 1.7201710098478449,
+      "learning_rate": 5.005849125598762e-08,
+      "loss": 2.2411,
+      "step": 2795
+    },
+    {
+      "epoch": 2.726387536514119,
+      "grad_norm": 1.6978407566709903,
+      "learning_rate": 5.0056096197215594e-08,
+      "loss": 2.2024,
+      "step": 2800
+    },
+    {
+      "epoch": 2.726387536514119,
+      "eval_loss": 2.4288954734802246,
+      "eval_runtime": 85.7694,
+      "eval_samples_per_second": 85.1,
+      "eval_steps_per_second": 0.676,
+      "step": 2800
+    },
+    {
+      "epoch": 2.731256085686465,
+      "grad_norm": 1.7625821445707386,
+      "learning_rate": 5.005379062843146e-08,
+      "loss": 2.245,
+      "step": 2805
+    },
+    {
+      "epoch": 2.736124634858812,
+      "grad_norm": 1.7304216492048587,
+      "learning_rate": 5.005157152893869e-08,
+      "loss": 2.223,
+      "step": 2810
+    },
+    {
+      "epoch": 2.7409931840311588,
+      "grad_norm": 1.755183680113624,
+      "learning_rate": 5.0049435969016435e-08,
+      "loss": 2.2239,
+      "step": 2815
+    },
+    {
+      "epoch": 2.7458617332035056,
+      "grad_norm": 1.728272875903093,
+      "learning_rate": 5.004738110751295e-08,
+      "loss": 2.237,
+      "step": 2820
+    },
+    {
+      "epoch": 2.750730282375852,
+      "grad_norm": 1.764407536913778,
+      "learning_rate": 5.0045404189493816e-08,
+      "loss": 2.2193,
+      "step": 2825
+    },
+    {
+      "epoch": 2.7555988315481987,
+      "grad_norm": 1.7145100854618731,
+      "learning_rate": 5.0043502543943914e-08,
+      "loss": 2.1959,
+      "step": 2830
+    },
+    {
+      "epoch": 2.760467380720545,
+      "grad_norm": 1.7322275336580597,
+      "learning_rate": 5.0041673581522057e-08,
+      "loss": 2.1898,
+      "step": 2835
+    },
+    {
+      "epoch": 2.765335929892892,
+      "grad_norm": 1.7454769702317845,
+      "learning_rate": 5.003991479236733e-08,
+      "loss": 2.2852,
+      "step": 2840
+    },
+    {
+      "epoch": 2.7702044790652387,
+      "grad_norm": 1.8214614189910026,
+      "learning_rate": 5.003822374395615e-08,
+      "loss": 2.2305,
+      "step": 2845
+    },
+    {
+      "epoch": 2.775073028237585,
+      "grad_norm": 1.8310587465226418,
+      "learning_rate": 5.0036598079009014e-08,
+      "loss": 2.1533,
+      "step": 2850
+    },
+    {
+      "epoch": 2.779941577409932,
+      "grad_norm": 1.773580257235929,
+      "learning_rate": 5.003503551344602e-08,
+      "loss": 2.2247,
+      "step": 2855
+    },
+    {
+      "epoch": 2.7848101265822782,
+      "grad_norm": 1.6972419215418109,
+      "learning_rate": 5.003353383439017e-08,
+      "loss": 2.2464,
+      "step": 2860
+    },
+    {
+      "epoch": 2.789678675754625,
+      "grad_norm": 1.7541525971250715,
+      "learning_rate": 5.003209089821755e-08,
+      "loss": 2.2041,
+      "step": 2865
+    },
+    {
+      "epoch": 2.794547224926972,
+      "grad_norm": 1.7726611140742992,
+      "learning_rate": 5.003070462865347e-08,
+      "loss": 2.2351,
+      "step": 2870
+    },
+    {
+      "epoch": 2.7994157740993186,
+      "grad_norm": 1.7257606879168454,
+      "learning_rate": 5.002937301491361e-08,
+      "loss": 2.1964,
+      "step": 2875
+    },
+    {
+      "epoch": 2.804284323271665,
+      "grad_norm": 1.7665569916717356,
+      "learning_rate": 5.002809410988936e-08,
+      "loss": 2.2198,
+      "step": 2880
+    },
+    {
+      "epoch": 2.809152872444012,
+      "grad_norm": 0.8460817495380386,
+      "learning_rate": 5.00268660283764e-08,
+      "loss": 2.2017,
+      "step": 2885
+    },
+    {
+      "epoch": 2.814021421616358,
+      "grad_norm": 1.7781068744183743,
+      "learning_rate": 5.002568694534567e-08,
+      "loss": 2.2113,
+      "step": 2890
+    },
+    {
+      "epoch": 2.818889970788705,
+      "grad_norm": 1.70436430018188,
+      "learning_rate": 5.0024555094256e-08,
+      "loss": 2.2296,
+      "step": 2895
+    },
+    {
+      "epoch": 2.8237585199610518,
+      "grad_norm": 1.801068375798501,
+      "learning_rate": 5.002346876540729e-08,
+      "loss": 2.2755,
+      "step": 2900
+    },
+    {
+      "epoch": 2.828627069133398,
+      "grad_norm": 1.7984300376552464,
+      "learning_rate": 5.0022426304333776e-08,
+      "loss": 2.2464,
+      "step": 2905
+    },
+    {
+      "epoch": 2.833495618305745,
+      "grad_norm": 1.7278702590224992,
+      "learning_rate": 5.002142611023626e-08,
+      "loss": 2.2223,
+      "step": 2910
+    },
+    {
+      "epoch": 2.8383641674780913,
+      "grad_norm": 1.7544765756817247,
+      "learning_rate": 5.0020466634452695e-08,
+      "loss": 2.2096,
+      "step": 2915
+    },
+    {
+      "epoch": 2.843232716650438,
+      "grad_norm": 1.7574601723708598,
+      "learning_rate": 5.001954637896626e-08,
+      "loss": 2.2034,
+      "step": 2920
+    },
+    {
+      "epoch": 2.848101265822785,
+      "grad_norm": 1.6850076585078704,
+      "learning_rate": 5.0018663894950185e-08,
+      "loss": 2.1788,
+      "step": 2925
+    },
+    {
+      "epoch": 2.8529698149951317,
+      "grad_norm": 1.734136831684721,
+      "learning_rate": 5.001781778134857e-08,
+      "loss": 2.171,
+      "step": 2930
+    },
+    {
+      "epoch": 2.857838364167478,
+      "grad_norm": 1.8530752594311186,
+      "learning_rate": 5.0017006683492444e-08,
+      "loss": 2.2323,
+      "step": 2935
+    },
+    {
+      "epoch": 2.862706913339825,
+      "grad_norm": 1.7452631968653791,
+      "learning_rate": 5.001622929175032e-08,
+      "loss": 2.2002,
+      "step": 2940
+    },
+    {
+      "epoch": 2.867575462512171,
+      "grad_norm": 1.765852324361209,
+      "learning_rate": 5.0015484340212554e-08,
+      "loss": 2.173,
+      "step": 2945
+    },
+    {
+      "epoch": 2.872444011684518,
+      "grad_norm": 1.7504024622386836,
+      "learning_rate": 5.0014770605408766e-08,
+      "loss": 2.1818,
+      "step": 2950
+    },
+    {
+      "epoch": 2.877312560856865,
+      "grad_norm": 1.709084582717418,
+      "learning_rate": 5.001408690505769e-08,
+      "loss": 2.2248,
+      "step": 2955
+    },
+    {
+      "epoch": 2.882181110029211,
+      "grad_norm": 1.726469521262587,
+      "learning_rate": 5.0013432096848647e-08,
+      "loss": 2.1754,
+      "step": 2960
+    },
+    {
+      "epoch": 2.887049659201558,
+      "grad_norm": 1.735487847062397,
+      "learning_rate": 5.001280507725414e-08,
+      "loss": 2.1754,
+      "step": 2965
+    },
+    {
+      "epoch": 2.8919182083739043,
+      "grad_norm": 1.7901732327186204,
+      "learning_rate": 5.0012204780372755e-08,
+      "loss": 2.2104,
+      "step": 2970
+    },
+    {
+      "epoch": 2.896786757546251,
+      "grad_norm": 1.7497305238142942,
+      "learning_rate": 5.001163017680177e-08,
+      "loss": 2.2131,
+      "step": 2975
+    },
+    {
+      "epoch": 2.901655306718598,
+      "grad_norm": 1.6912311894555403,
+      "learning_rate": 5.0011080272538895e-08,
+      "loss": 2.1984,
+      "step": 2980
+    },
+    {
+      "epoch": 2.9065238558909448,
+      "grad_norm": 1.7550540502025136,
+      "learning_rate": 5.0010657486604285e-08,
+      "loss": 2.2325,
+      "step": 2985
+    },
+    {
+      "epoch": 2.911392405063291,
+      "grad_norm": 1.7830928527818988,
+      "learning_rate": 5.001014964512753e-08,
+      "loss": 2.2295,
+      "step": 2990
+    },
+    {
+      "epoch": 2.916260954235638,
+      "grad_norm": 1.7967938470353484,
+      "learning_rate": 5.000966389908247e-08,
+      "loss": 2.2238,
+      "step": 2995
+    },
+    {
+      "epoch": 2.9211295034079843,
+      "grad_norm": 1.73275794534534,
+      "learning_rate": 5.0009199379466084e-08,
+      "loss": 2.2074,
+      "step": 3000
+    },
+    {
+      "epoch": 2.9211295034079843,
+      "eval_loss": 2.4288480281829834,
+      "eval_runtime": 85.4779,
+      "eval_samples_per_second": 85.391,
+      "eval_steps_per_second": 0.679,
+      "step": 3000
+    },
+    {
+      "epoch": 2.925998052580331,
+      "grad_norm": 1.798848636728742,
+      "learning_rate": 5.000875524778925e-08,
+      "loss": 2.1889,
+      "step": 3005
+    },
+    {
+      "epoch": 2.930866601752678,
+      "grad_norm": 1.7669718423903236,
+      "learning_rate": 5.0008330695135104e-08,
+      "loss": 2.2008,
+      "step": 3010
+    },
+    {
+      "epoch": 2.9357351509250242,
+      "grad_norm": 1.7532733346720555,
+      "learning_rate": 5.000792494124235e-08,
+      "loss": 2.193,
+      "step": 3015
+    },
+    {
+      "epoch": 2.940603700097371,
+      "grad_norm": 1.7095158480632124,
+      "learning_rate": 5.00075372336131e-08,
+      "loss": 2.2389,
+      "step": 3020
+    },
+    {
+      "epoch": 2.9454722492697174,
+      "grad_norm": 1.7236728786130506,
+      "learning_rate": 5.0007166846644555e-08,
+      "loss": 2.2444,
+      "step": 3025
+    },
+    {
+      "epoch": 2.950340798442064,
+      "grad_norm": 1.7363740627339725,
+      "learning_rate": 5.000681308078417e-08,
+      "loss": 2.22,
+      "step": 3030
+    },
+    {
+      "epoch": 2.955209347614411,
+      "grad_norm": 1.728808934644446,
+      "learning_rate": 5.000647526170754e-08,
+      "loss": 2.2163,
+      "step": 3035
+    },
+    {
+      "epoch": 2.960077896786758,
+      "grad_norm": 1.7449413432648693,
+      "learning_rate": 5.000615273951875e-08,
+      "loss": 2.2779,
+      "step": 3040
+    },
+    {
+      "epoch": 2.964946445959104,
+      "grad_norm": 1.753647141186398,
+      "learning_rate": 5.000584488797249e-08,
+      "loss": 2.1997,
+      "step": 3045
+    },
+    {
+      "epoch": 2.969814995131451,
+      "grad_norm": 1.729320335247063,
+      "learning_rate": 5.000555110371747e-08,
+      "loss": 2.2329,
+      "step": 3050
+    },
+    {
+      "epoch": 2.9746835443037973,
+      "grad_norm": 1.7435846324509203,
+      "learning_rate": 5.000527080556074e-08,
+      "loss": 2.2455,
+      "step": 3055
+    },
+    {
+      "epoch": 2.979552093476144,
+      "grad_norm": 1.7544585151474008,
+      "learning_rate": 5.000500343375234e-08,
+      "loss": 2.2232,
+      "step": 3060
+    },
+    {
+      "epoch": 2.984420642648491,
+      "grad_norm": 1.8297856934072276,
+      "learning_rate": 5.00047484492898e-08,
+      "loss": 2.2025,
+      "step": 3065
+    },
+    {
+      "epoch": 2.9892891918208373,
+      "grad_norm": 1.8331124463008972,
+      "learning_rate": 5.000450533324217e-08,
+      "loss": 2.2346,
+      "step": 3070
+    },
+    {
+      "epoch": 2.994157740993184,
+      "grad_norm": 1.77523332941347,
+      "learning_rate": 5.0004273586092874e-08,
+      "loss": 2.2476,
+      "step": 3075
+    },
+    {
+      "epoch": 2.9990262901655305,
+      "grad_norm": 1.719560891287608,
+      "learning_rate": 5.0004052727101194e-08,
+      "loss": 2.2219,
+      "step": 3080
+    },
+    {
+      "epoch": 3.0038948393378773,
+      "grad_norm": 1.7451899184767314,
+      "learning_rate": 5.000384229368183e-08,
+      "loss": 2.267,
+      "step": 3085
+    },
+    {
+      "epoch": 3.008763388510224,
+      "grad_norm": 1.7132088078853227,
+      "learning_rate": 5.0003641840802084e-08,
+      "loss": 2.232,
+      "step": 3090
+    },
+    {
+      "epoch": 3.0136319376825704,
+      "grad_norm": 1.7074710408698504,
+      "learning_rate": 5.000345094039628e-08,
+      "loss": 2.2226,
+      "step": 3095
+    },
+    {
+      "epoch": 3.0185004868549172,
+      "grad_norm": 1.7116453045970612,
+      "learning_rate": 5.000326918079705e-08,
+      "loss": 2.2132,
+      "step": 3100
+    },
+    {
+      "epoch": 3.023369036027264,
+      "grad_norm": 1.7478082412483065,
+      "learning_rate": 5.000309616618301e-08,
+      "loss": 2.2595,
+      "step": 3105
+    },
+    {
+      "epoch": 3.0282375851996104,
+      "grad_norm": 1.7390659451407764,
+      "learning_rate": 5.0002931516042464e-08,
+      "loss": 2.2532,
+      "step": 3110
+    },
+    {
+      "epoch": 3.033106134371957,
+      "grad_norm": 1.6629867778767704,
+      "learning_rate": 5.0002774864652723e-08,
+      "loss": 2.2157,
+      "step": 3115
+    },
+    {
+      "epoch": 3.037974683544304,
+      "grad_norm": 1.7430106410373127,
+      "learning_rate": 5.000262586057472e-08,
+      "loss": 2.2654,
+      "step": 3120
+    },
+    {
+      "epoch": 3.0428432327166504,
+      "grad_norm": 1.7699263854675031,
+      "learning_rate": 5.0002484166162455e-08,
+      "loss": 2.1838,
+      "step": 3125
+    },
+    {
+      "epoch": 3.047711781888997,
+      "grad_norm": 1.7970379405255692,
+      "learning_rate": 5.0002349457087e-08,
+      "loss": 2.1844,
+      "step": 3130
+    },
+    {
+      "epoch": 3.0525803310613435,
+      "grad_norm": 1.740063554373285,
+      "learning_rate": 5.00022214218746e-08,
+      "loss": 2.1975,
+      "step": 3135
+    },
+    {
+      "epoch": 3.0574488802336903,
+      "grad_norm": 1.7669764533755763,
+      "learning_rate": 5.000209976145863e-08,
+      "loss": 2.173,
+      "step": 3140
+    },
+    {
+      "epoch": 3.062317429406037,
+      "grad_norm": 1.7570288798515004,
+      "learning_rate": 5.000198418874495e-08,
+      "loss": 2.202,
+      "step": 3145
+    },
+    {
+      "epoch": 3.0671859785783835,
+      "grad_norm": 1.7640313096994278,
+      "learning_rate": 5.000187442819041e-08,
+      "loss": 2.2743,
+      "step": 3150
+    },
+    {
+      "epoch": 3.0720545277507303,
+      "grad_norm": 1.776697414652845,
+      "learning_rate": 5.000177021539416e-08,
+      "loss": 2.1953,
+      "step": 3155
+    },
+    {
+      "epoch": 3.076923076923077,
+      "grad_norm": 1.7469177371601763,
+      "learning_rate": 5.000167129670136e-08,
+      "loss": 2.2643,
+      "step": 3160
+    },
+    {
+      "epoch": 3.0817916260954235,
+      "grad_norm": 1.7981065248029078,
+      "learning_rate": 5.000157742881906e-08,
+      "loss": 2.2192,
+      "step": 3165
+    },
+    {
+      "epoch": 3.0866601752677703,
+      "grad_norm": 1.8048538458330963,
+      "learning_rate": 5.000148837844394e-08,
+      "loss": 2.2188,
+      "step": 3170
+    },
+    {
+      "epoch": 3.091528724440117,
+      "grad_norm": 1.77587620793656,
+      "learning_rate": 5.000140392190154e-08,
+      "loss": 2.205,
+      "step": 3175
+    },
+    {
+      "epoch": 3.0963972736124634,
+      "grad_norm": 1.7277030937813718,
+      "learning_rate": 5.000132384479669e-08,
+      "loss": 2.2418,
+      "step": 3180
+    },
+    {
+      "epoch": 3.1012658227848102,
+      "grad_norm": 1.7715455250981111,
+      "learning_rate": 5.0001247941674944e-08,
+      "loss": 2.195,
+      "step": 3185
+    },
+    {
+      "epoch": 3.1061343719571566,
+      "grad_norm": 1.7418270265834659,
+      "learning_rate": 5.00011760156946e-08,
+      "loss": 2.2188,
+      "step": 3190
+    },
+    {
+      "epoch": 3.1110029211295034,
+      "grad_norm": 1.7626475968666995,
+      "learning_rate": 5.000110787830913e-08,
+      "loss": 2.1969,
+      "step": 3195
+    },
+    {
+      "epoch": 3.11587147030185,
+      "grad_norm": 1.7554268739931094,
+      "learning_rate": 5.000104334895965e-08,
+      "loss": 2.2268,
+      "step": 3200
+    },
+    {
+      "epoch": 3.11587147030185,
+      "eval_loss": 2.4296844005584717,
+      "eval_runtime": 85.8007,
+      "eval_samples_per_second": 85.069,
+      "eval_steps_per_second": 0.676,
+      "step": 3200
+    },
+    {
+      "epoch": 3.1207400194741965,
+      "grad_norm": 1.768111175754245,
+      "learning_rate": 5.000098225477726e-08,
+      "loss": 2.1925,
+      "step": 3205
+    },
+    {
+      "epoch": 3.1256085686465434,
+      "grad_norm": 1.731673904293537,
+      "learning_rate": 5.000092443029491e-08,
+      "loss": 2.2291,
+      "step": 3210
+    },
+    {
+      "epoch": 3.13047711781889,
+      "grad_norm": 1.7474269730832317,
+      "learning_rate": 5.0000869717168615e-08,
+      "loss": 2.2431,
+      "step": 3215
+    },
+    {
+      "epoch": 3.1353456669912365,
+      "grad_norm": 1.7068082256216377,
+      "learning_rate": 5.000081796390766e-08,
+      "loss": 2.2197,
+      "step": 3220
+    },
+    {
+      "epoch": 3.1402142161635833,
+      "grad_norm": 1.791689784795618,
+      "learning_rate": 5.000076902561367e-08,
+      "loss": 2.1983,
+      "step": 3225
+    },
+    {
+      "epoch": 3.14508276533593,
+      "grad_norm": 1.7624390988374115,
+      "learning_rate": 5.000072276372817e-08,
+      "loss": 2.2693,
+      "step": 3230
+    },
+    {
+      "epoch": 3.1499513145082765,
+      "grad_norm": 1.8338687668629183,
+      "learning_rate": 5.0000679045788575e-08,
+      "loss": 2.1917,
+      "step": 3235
+    },
+    {
+      "epoch": 3.1548198636806233,
+      "grad_norm": 1.709244589398017,
+      "learning_rate": 5.000063774519218e-08,
+      "loss": 2.1911,
+      "step": 3240
+    },
+    {
+      "epoch": 3.1596884128529696,
+      "grad_norm": 1.7192540273943981,
+      "learning_rate": 5.0000598740968074e-08,
+      "loss": 2.1583,
+      "step": 3245
+    },
+    {
+      "epoch": 3.1645569620253164,
+      "grad_norm": 1.728010839757635,
+      "learning_rate": 5.000056191755672e-08,
+      "loss": 2.1972,
+      "step": 3250
+    },
+    {
+      "epoch": 3.1694255111976632,
+      "grad_norm": 1.788796272197145,
+      "learning_rate": 5.0000527164596915e-08,
+      "loss": 2.2011,
+      "step": 3255
+    },
+    {
+      "epoch": 3.1742940603700096,
+      "grad_norm": 1.7635432660617776,
+      "learning_rate": 5.000049437672004e-08,
+      "loss": 2.2152,
+      "step": 3260
+    },
+    {
+      "epoch": 3.1791626095423564,
+      "grad_norm": 1.7957521394770797,
+      "learning_rate": 5.000046345335129e-08,
+      "loss": 2.2182,
+      "step": 3265
+    },
+    {
+      "epoch": 3.184031158714703,
+      "grad_norm": 1.8009009243137868,
+      "learning_rate": 5.000043429851777e-08,
+      "loss": 2.2061,
+      "step": 3270
+    },
+    {
+      "epoch": 3.1888997078870496,
+      "grad_norm": 1.7948019721772699,
+      "learning_rate": 5.0000406820663126e-08,
+      "loss": 2.1857,
+      "step": 3275
+    },
+    {
+      "epoch": 3.1937682570593964,
+      "grad_norm": 1.8100815667189885,
+      "learning_rate": 5.0000380932468733e-08,
+      "loss": 2.1777,
+      "step": 3280
+    },
+    {
+      "epoch": 3.1986368062317427,
+      "grad_norm": 1.7827984808365251,
+      "learning_rate": 5.000035655068104e-08,
+      "loss": 2.2366,
+      "step": 3285
+    },
+    {
+      "epoch": 3.2035053554040895,
+      "grad_norm": 1.748286864117926,
+      "learning_rate": 5.0000333595944974e-08,
+      "loss": 2.207,
+      "step": 3290
+    },
+    {
+      "epoch": 3.2083739045764363,
+      "grad_norm": 1.703650849390858,
+      "learning_rate": 5.0000311992643325e-08,
+      "loss": 2.1896,
+      "step": 3295
+    },
+    {
+      "epoch": 3.2132424537487827,
+      "grad_norm": 1.7698280644275959,
+      "learning_rate": 5.0000291668741705e-08,
+      "loss": 2.1639,
+      "step": 3300
+    },
+    {
+      "epoch": 3.2181110029211295,
+      "grad_norm": 1.756366599744074,
+      "learning_rate": 5.000027255563917e-08,
+      "loss": 2.2012,
+      "step": 3305
+    },
+    {
+      "epoch": 3.2229795520934763,
+      "grad_norm": 1.759461002068701,
+      "learning_rate": 5.000025458802413e-08,
+      "loss": 2.2341,
+      "step": 3310
+    },
+    {
+      "epoch": 3.2278481012658227,
+      "grad_norm": 1.7494660148125987,
+      "learning_rate": 5.000023770373551e-08,
+      "loss": 2.207,
+      "step": 3315
+    },
+    {
+      "epoch": 3.2327166504381695,
+      "grad_norm": 1.7785100579340334,
+      "learning_rate": 5.000022184362899e-08,
+      "loss": 2.1999,
+      "step": 3320
+    },
+    {
+      "epoch": 3.2375851996105163,
+      "grad_norm": 1.7645954769969932,
+      "learning_rate": 5.0000206951448066e-08,
+      "loss": 2.2683,
+      "step": 3325
+    },
+    {
+      "epoch": 3.2424537487828626,
+      "grad_norm": 1.7745088927616617,
+      "learning_rate": 5.000019297369995e-08,
+      "loss": 2.2251,
+      "step": 3330
+    },
+    {
+      "epoch": 3.2473222979552094,
+      "grad_norm": 1.8131533062966378,
+      "learning_rate": 5.0000179859535986e-08,
+      "loss": 2.2074,
+      "step": 3335
+    },
+    {
+      "epoch": 3.252190847127556,
+      "grad_norm": 1.744211372153281,
+      "learning_rate": 5.000016756063664e-08,
+      "loss": 2.172,
+      "step": 3340
+    },
+    {
+      "epoch": 3.2570593962999026,
+      "grad_norm": 1.7741244031891719,
+      "learning_rate": 5.000015603110066e-08,
+      "loss": 2.2425,
+      "step": 3345
+    },
+    {
+      "epoch": 3.2619279454722494,
+      "grad_norm": 1.7486152934920032,
+      "learning_rate": 5.0000145227338575e-08,
+      "loss": 2.2143,
+      "step": 3350
+    },
+    {
+      "epoch": 3.2667964946445958,
+      "grad_norm": 1.757112828413238,
+      "learning_rate": 5.000013510797011e-08,
+      "loss": 2.1976,
+      "step": 3355
+    },
+    {
+      "epoch": 3.2716650438169426,
+      "grad_norm": 1.7240716262707467,
+      "learning_rate": 5.000012563372563e-08,
+      "loss": 2.2618,
+      "step": 3360
+    },
+    {
+      "epoch": 3.2765335929892894,
+      "grad_norm": 1.7213155493984982,
+      "learning_rate": 5.0000116767351296e-08,
+      "loss": 2.2115,
+      "step": 3365
+    },
+    {
+      "epoch": 3.2814021421616357,
+      "grad_norm": 1.8311141582937929,
+      "learning_rate": 5.000010847351797e-08,
+      "loss": 2.1974,
+      "step": 3370
+    },
+    {
+      "epoch": 3.2862706913339825,
+      "grad_norm": 1.7864630099921617,
+      "learning_rate": 5.000010071873363e-08,
+      "loss": 2.238,
+      "step": 3375
+    },
+    {
+      "epoch": 3.291139240506329,
+      "grad_norm": 1.7883459145585372,
+      "learning_rate": 5.00000934712592e-08,
+      "loss": 2.2031,
+      "step": 3380
+    },
+    {
+      "epoch": 3.2960077896786757,
+      "grad_norm": 1.699599322514979,
+      "learning_rate": 5.000008670102778e-08,
+      "loss": 2.2327,
+      "step": 3385
+    },
+    {
+      "epoch": 3.3008763388510225,
+      "grad_norm": 1.7788870047828989,
+      "learning_rate": 5.0000080379567014e-08,
+      "loss": 2.1472,
+      "step": 3390
+    },
+    {
+      "epoch": 3.305744888023369,
+      "grad_norm": 1.7365094304977247,
+      "learning_rate": 5.000007447992463e-08,
+      "loss": 2.2068,
+      "step": 3395
+    },
+    {
+      "epoch": 3.3106134371957157,
+      "grad_norm": 1.8492451880826386,
+      "learning_rate": 5.0000068976596906e-08,
+      "loss": 2.1556,
+      "step": 3400
+    },
+    {
+      "epoch": 3.3106134371957157,
+      "eval_loss": 2.429415464401245,
+      "eval_runtime": 85.0239,
+      "eval_samples_per_second": 85.846,
+      "eval_steps_per_second": 0.682,
+      "step": 3400
+    },
+    {
+      "epoch": 3.3154819863680625,
+      "grad_norm": 1.769012775563444,
+      "learning_rate": 5.0000063845460134e-08,
+      "loss": 2.1791,
+      "step": 3405
+    },
+    {
+      "epoch": 3.320350535540409,
+      "grad_norm": 1.7426924775202128,
+      "learning_rate": 5.000005999315767e-08,
+      "loss": 2.2323,
+      "step": 3410
+    },
+    {
+      "epoch": 3.3252190847127556,
+      "grad_norm": 1.764543573084179,
+      "learning_rate": 5.000005547532745e-08,
+      "loss": 2.1965,
+      "step": 3415
+    },
+    {
+      "epoch": 3.3300876338851024,
+      "grad_norm": 1.8373323305401401,
+      "learning_rate": 5.0000051268933666e-08,
+      "loss": 2.2295,
+      "step": 3420
+    },
+    {
+      "epoch": 3.334956183057449,
+      "grad_norm": 1.713591544576181,
+      "learning_rate": 5.00000473545193e-08,
+      "loss": 2.1847,
+      "step": 3425
+    },
+    {
+      "epoch": 3.3398247322297956,
+      "grad_norm": 1.7559454568881032,
+      "learning_rate": 5.000004371371542e-08,
+      "loss": 2.2148,
+      "step": 3430
+    },
+    {
+      "epoch": 3.344693281402142,
+      "grad_norm": 1.7485006911314465,
+      "learning_rate": 5.000004032918764e-08,
+      "loss": 2.2065,
+      "step": 3435
+    },
+    {
+      "epoch": 3.3495618305744888,
+      "grad_norm": 1.7533601433403279,
+      "learning_rate": 5.00000371845847e-08,
+      "loss": 2.1955,
+      "step": 3440
+    },
+    {
+      "epoch": 3.3544303797468356,
+      "grad_norm": 1.7543658707594614,
+      "learning_rate": 5.000003426448936e-08,
+      "loss": 2.2087,
+      "step": 3445
+    },
+    {
+      "epoch": 3.359298928919182,
+      "grad_norm": 1.7904175039614314,
+      "learning_rate": 5.0000031554371375e-08,
+      "loss": 2.272,
+      "step": 3450
+    },
+    {
+      "epoch": 3.3641674780915287,
+      "grad_norm": 1.7502721668385577,
+      "learning_rate": 5.000002904054251e-08,
+      "loss": 2.2101,
+      "step": 3455
+    },
+    {
+      "epoch": 3.3690360272638755,
+      "grad_norm": 1.706493138085354,
+      "learning_rate": 5.000002671011354e-08,
+      "loss": 2.1955,
+      "step": 3460
+    },
+    {
+      "epoch": 3.373904576436222,
+      "grad_norm": 1.7655034010253434,
+      "learning_rate": 5.0000024550953135e-08,
+      "loss": 2.2103,
+      "step": 3465
+    },
+    {
+      "epoch": 3.3787731256085687,
+      "grad_norm": 1.7580566741004937,
+      "learning_rate": 5.0000022551648575e-08,
+      "loss": 2.2243,
+      "step": 3470
+    },
+    {
+      "epoch": 3.3836416747809155,
+      "grad_norm": 1.7951986660353298,
+      "learning_rate": 5.000002070146821e-08,
+      "loss": 2.233,
+      "step": 3475
+    },
+    {
+      "epoch": 3.388510223953262,
+      "grad_norm": 1.7464673450399157,
+      "learning_rate": 5.000001899032566e-08,
+      "loss": 2.1938,
+      "step": 3480
+    },
+    {
+      "epoch": 3.3933787731256086,
+      "grad_norm": 1.7506731049190618,
+      "learning_rate": 5.0000017408745575e-08,
+      "loss": 2.2132,
+      "step": 3485
+    },
+    {
+      "epoch": 3.398247322297955,
+      "grad_norm": 1.7158313234275915,
+      "learning_rate": 5.0000015947831005e-08,
+      "loss": 2.2255,
+      "step": 3490
+    },
+    {
+      "epoch": 3.403115871470302,
+      "grad_norm": 1.7741448257648114,
+      "learning_rate": 5.000001459923229e-08,
+      "loss": 2.2561,
+      "step": 3495
+    },
+    {
+      "epoch": 3.4079844206426486,
+      "grad_norm": 1.7680229196146826,
+      "learning_rate": 5.0000013355117324e-08,
+      "loss": 2.2174,
+      "step": 3500
+    },
+    {
+      "epoch": 3.412852969814995,
+      "grad_norm": 1.7566868775329945,
+      "learning_rate": 5.000001220814333e-08,
+      "loss": 2.2049,
+      "step": 3505
+    },
+    {
+      "epoch": 3.4177215189873418,
+      "grad_norm": 1.7231310116084255,
+      "learning_rate": 5.0000011151429815e-08,
+      "loss": 2.243,
+      "step": 3510
+    },
+    {
+      "epoch": 3.4225900681596886,
+      "grad_norm": 1.7525419468509682,
+      "learning_rate": 5.0000010178532987e-08,
+      "loss": 2.2271,
+      "step": 3515
+    },
+    {
+      "epoch": 3.427458617332035,
+      "grad_norm": 1.7743431604185957,
+      "learning_rate": 5.000000928342124e-08,
+      "loss": 2.1975,
+      "step": 3520
+    },
+    {
+      "epoch": 3.4323271665043817,
+      "grad_norm": 1.738567333307858,
+      "learning_rate": 5.000000846045193e-08,
+      "loss": 2.1717,
+      "step": 3525
+    },
+    {
+      "epoch": 3.4371957156767285,
+      "grad_norm": 1.7930958206448022,
+      "learning_rate": 5.000000770434924e-08,
+      "loss": 2.1939,
+      "step": 3530
+    },
+    {
+      "epoch": 3.442064264849075,
+      "grad_norm": 1.7498436482409632,
+      "learning_rate": 5.0000007010183126e-08,
+      "loss": 2.213,
+      "step": 3535
+    },
+    {
+      "epoch": 3.4469328140214217,
+      "grad_norm": 1.7055196904583134,
+      "learning_rate": 5.0000006373349365e-08,
+      "loss": 2.2058,
+      "step": 3540
+    },
+    {
+      "epoch": 3.451801363193768,
+      "grad_norm": 1.7510250220237107,
+      "learning_rate": 5.00000057895505e-08,
+      "loss": 2.2337,
+      "step": 3545
+    },
+    {
+      "epoch": 3.456669912366115,
+      "grad_norm": 1.713344136042387,
+      "learning_rate": 5.0000005254777845e-08,
+      "loss": 2.1989,
+      "step": 3550
+    },
+    {
+      "epoch": 3.4615384615384617,
+      "grad_norm": 1.7371767245978642,
+      "learning_rate": 5.000000476529434e-08,
+      "loss": 2.233,
+      "step": 3555
+    },
+    {
+      "epoch": 3.466407010710808,
+      "grad_norm": 1.7636410740956914,
+      "learning_rate": 5.000000431761833e-08,
+      "loss": 2.1615,
+      "step": 3560
+    },
+    {
+      "epoch": 3.471275559883155,
+      "grad_norm": 1.7144161091063708,
+      "learning_rate": 5.000000390850812e-08,
+      "loss": 2.2205,
+      "step": 3565
+    },
+    {
+      "epoch": 3.4761441090555016,
+      "grad_norm": 1.8042049963234217,
+      "learning_rate": 5.0000003534947425e-08,
+      "loss": 2.178,
+      "step": 3570
+    },
+    {
+      "epoch": 3.481012658227848,
+      "grad_norm": 1.7853951456456785,
+      "learning_rate": 5.00000031941315e-08,
+      "loss": 2.2185,
+      "step": 3575
+    },
+    {
+      "epoch": 3.485881207400195,
+      "grad_norm": 1.6707790882813032,
+      "learning_rate": 5.000000288345406e-08,
+      "loss": 2.1868,
+      "step": 3580
+    },
+    {
+      "epoch": 3.4907497565725416,
+      "grad_norm": 1.7236640638911755,
+      "learning_rate": 5.000000260049486e-08,
+      "loss": 2.2153,
+      "step": 3585
+    },
+    {
+      "epoch": 3.495618305744888,
+      "grad_norm": 1.7103907676750174,
+      "learning_rate": 5.0000002343008e-08,
+      "loss": 2.2806,
+      "step": 3590
+    },
+    {
+      "epoch": 3.5004868549172348,
+      "grad_norm": 1.7458900493895921,
+      "learning_rate": 5.000000210891081e-08,
+      "loss": 2.1832,
+      "step": 3595
+    },
+    {
+      "epoch": 3.505355404089581,
+      "grad_norm": 1.8597626382974297,
+      "learning_rate": 5.00000018962734e-08,
+      "loss": 2.1953,
+      "step": 3600
+    },
+    {
+      "epoch": 3.505355404089581,
+      "eval_loss": 2.4295620918273926,
+      "eval_runtime": 85.4888,
+      "eval_samples_per_second": 85.38,
+      "eval_steps_per_second": 0.678,
+      "step": 3600
+    },
+    {
+      "epoch": 3.510223953261928,
+      "grad_norm": 1.7712796854715114,
+      "learning_rate": 5.000000170330872e-08,
+      "loss": 2.1841,
+      "step": 3605
+    },
+    {
+      "epoch": 3.5150925024342747,
+      "grad_norm": 1.7646034875776253,
+      "learning_rate": 5.000000152836327e-08,
+      "loss": 2.2182,
+      "step": 3610
+    },
+    {
+      "epoch": 3.519961051606621,
+      "grad_norm": 1.7404348200627402,
+      "learning_rate": 5.000000136990825e-08,
+      "loss": 2.2245,
+      "step": 3615
+    },
+    {
+      "epoch": 3.524829600778968,
+      "grad_norm": 1.7098638272240443,
+      "learning_rate": 5.000000122653126e-08,
+      "loss": 2.2094,
+      "step": 3620
+    },
+    {
+      "epoch": 3.5296981499513143,
+      "grad_norm": 1.7920393050904846,
+      "learning_rate": 5.000000109692848e-08,
+      "loss": 2.2366,
+      "step": 3625
+    },
+    {
+      "epoch": 3.534566699123661,
+      "grad_norm": 1.746241513422538,
+      "learning_rate": 5.0000000979897296e-08,
+      "loss": 2.1925,
+      "step": 3630
+    },
+    {
+      "epoch": 3.539435248296008,
+      "grad_norm": 1.8183591426270245,
+      "learning_rate": 5.000000087432932e-08,
+      "loss": 2.2159,
+      "step": 3635
+    },
+    {
+      "epoch": 3.5443037974683547,
+      "grad_norm": 1.821208130477108,
+      "learning_rate": 5.0000000779203936e-08,
+      "loss": 2.2038,
+      "step": 3640
+    },
+    {
+      "epoch": 3.549172346640701,
+      "grad_norm": 1.7255936567126264,
+      "learning_rate": 5.000000069358209e-08,
+      "loss": 2.1901,
+      "step": 3645
+    },
+    {
+      "epoch": 3.554040895813048,
+      "grad_norm": 1.855947848851622,
+      "learning_rate": 5.0000000616600557e-08,
+      "loss": 2.2198,
+      "step": 3650
+    },
+    {
+      "epoch": 3.558909444985394,
+      "grad_norm": 1.8303786214402666,
+      "learning_rate": 5.000000054746654e-08,
+      "loss": 2.1946,
+      "step": 3655
+    },
+    {
+      "epoch": 3.563777994157741,
+      "grad_norm": 1.7392508415046033,
+      "learning_rate": 5.000000048545256e-08,
+      "loss": 2.1778,
+      "step": 3660
+    },
+    {
+      "epoch": 3.568646543330088,
+      "grad_norm": 1.8213274923689418,
+      "learning_rate": 5.0000000429891724e-08,
+      "loss": 2.1742,
+      "step": 3665
+    },
+    {
+      "epoch": 3.573515092502434,
+      "grad_norm": 1.8017867977855504,
+      "learning_rate": 5.0000000380173246e-08,
+      "loss": 2.1915,
+      "step": 3670
+    },
+    {
+      "epoch": 3.578383641674781,
+      "grad_norm": 1.7382969587337458,
+      "learning_rate": 5.000000033573829e-08,
+      "loss": 2.1693,
+      "step": 3675
+    },
+    {
+      "epoch": 3.5832521908471273,
+      "grad_norm": 1.6944625431597526,
+      "learning_rate": 5.0000000296076037e-08,
+      "loss": 2.1787,
+      "step": 3680
+    },
+    {
+      "epoch": 3.588120740019474,
+      "grad_norm": 1.726132620813893,
+      "learning_rate": 5.0000000260720054e-08,
+      "loss": 2.2046,
+      "step": 3685
+    },
+    {
+      "epoch": 3.592989289191821,
+      "grad_norm": 1.715704002261475,
+      "learning_rate": 5.0000000229244886e-08,
+      "loss": 2.2242,
+      "step": 3690
+    },
+    {
+      "epoch": 3.5978578383641677,
+      "grad_norm": 1.7355446090598186,
+      "learning_rate": 5.0000000201262854e-08,
+      "loss": 2.2162,
+      "step": 3695
+    },
+    {
+      "epoch": 3.602726387536514,
+      "grad_norm": 1.744140516288637,
+      "learning_rate": 5.0000000176421085e-08,
+      "loss": 2.1987,
+      "step": 3700
+    },
+    {
+      "epoch": 3.607594936708861,
+      "grad_norm": 1.801416253511192,
+      "learning_rate": 5.000000015439878e-08,
+      "loss": 2.2142,
+      "step": 3705
+    },
+    {
+      "epoch": 3.6124634858812072,
+      "grad_norm": 1.72380914141294,
+      "learning_rate": 5.000000013490458e-08,
+      "loss": 2.1938,
+      "step": 3710
+    },
+    {
+      "epoch": 3.617332035053554,
+      "grad_norm": 1.6829634226373547,
+      "learning_rate": 5.000000011767422e-08,
+      "loss": 2.2291,
+      "step": 3715
+    },
+    {
+      "epoch": 3.622200584225901,
+      "grad_norm": 1.71665602485241,
+      "learning_rate": 5.0000000102468287e-08,
+      "loss": 2.2024,
+      "step": 3720
+    },
+    {
+      "epoch": 3.627069133398247,
+      "grad_norm": 1.7001891231981483,
+      "learning_rate": 5.000000008907013e-08,
+      "loss": 2.2397,
+      "step": 3725
+    },
+    {
+      "epoch": 3.631937682570594,
+      "grad_norm": 1.7560852361384687,
+      "learning_rate": 5.000000007728397e-08,
+      "loss": 2.2149,
+      "step": 3730
+    },
+    {
+      "epoch": 3.6368062317429404,
+      "grad_norm": 1.8239331341308023,
+      "learning_rate": 5.0000000066933104e-08,
+      "loss": 2.2073,
+      "step": 3735
+    },
+    {
+      "epoch": 3.641674780915287,
+      "grad_norm": 1.7435406529125186,
+      "learning_rate": 5.000000005785826e-08,
+      "loss": 2.202,
+      "step": 3740
+    },
+    {
+      "epoch": 3.646543330087634,
+      "grad_norm": 1.769830204490899,
+      "learning_rate": 5.0000000049916066e-08,
+      "loss": 2.1874,
+      "step": 3745
+    },
+    {
+      "epoch": 3.651411879259981,
+      "grad_norm": 1.7365544971501097,
+      "learning_rate": 5.0000000042977675e-08,
+      "loss": 2.2312,
+      "step": 3750
+    },
+    {
+      "epoch": 3.656280428432327,
+      "grad_norm": 1.754918261553824,
+      "learning_rate": 5.000000003692742e-08,
+      "loss": 2.233,
+      "step": 3755
+    },
+    {
+      "epoch": 3.661148977604674,
+      "grad_norm": 1.7251476744734402,
+      "learning_rate": 5.000000003166163e-08,
+      "loss": 2.1785,
+      "step": 3760
+    },
+    {
+      "epoch": 3.6660175267770203,
+      "grad_norm": 1.6885151956951188,
+      "learning_rate": 5.000000002708756e-08,
+      "loss": 2.213,
+      "step": 3765
+    },
+    {
+      "epoch": 3.670886075949367,
+      "grad_norm": 1.8104164593434913,
+      "learning_rate": 5.000000002312232e-08,
+      "loss": 2.2031,
+      "step": 3770
+    },
+    {
+      "epoch": 3.675754625121714,
+      "grad_norm": 1.6964800041666812,
+      "learning_rate": 5.0000000019691985e-08,
+      "loss": 2.2106,
+      "step": 3775
+    },
+    {
+      "epoch": 3.6806231742940603,
+      "grad_norm": 1.7770021912734775,
+      "learning_rate": 5.0000000016730705e-08,
+      "loss": 2.1958,
+      "step": 3780
+    },
+    {
+      "epoch": 3.685491723466407,
+      "grad_norm": 1.7854463027276486,
+      "learning_rate": 5.000000001417995e-08,
+      "loss": 2.1748,
+      "step": 3785
+    },
+    {
+      "epoch": 3.6903602726387534,
+      "grad_norm": 1.7314209770363447,
+      "learning_rate": 5.000000001198777e-08,
+      "loss": 2.2285,
+      "step": 3790
+    },
+    {
+      "epoch": 3.6952288218111002,
+      "grad_norm": 1.7315368458494336,
+      "learning_rate": 5.000000001010814e-08,
+      "loss": 2.2168,
+      "step": 3795
+    },
+    {
+      "epoch": 3.700097370983447,
+      "grad_norm": 1.7953711064417144,
+      "learning_rate": 5.0000000008500364e-08,
+      "loss": 2.2002,
+      "step": 3800
+    },
+    {
+      "epoch": 3.700097370983447,
+      "eval_loss": 2.4294352531433105,
+      "eval_runtime": 85.6309,
+      "eval_samples_per_second": 85.238,
+      "eval_steps_per_second": 0.677,
+      "step": 3800
+    },
+    {
+      "epoch": 3.704965920155794,
+      "grad_norm": 1.792810318597494,
+      "learning_rate": 5.000000000712853e-08,
+      "loss": 2.2033,
+      "step": 3805
+    },
+    {
+      "epoch": 3.70983446932814,
+      "grad_norm": 1.7519531928443899,
+      "learning_rate": 5.0000000005961005e-08,
+      "loss": 2.2504,
+      "step": 3810
+    },
+    {
+      "epoch": 3.714703018500487,
+      "grad_norm": 1.8276341931990958,
+      "learning_rate": 5.0000000004969983e-08,
+      "loss": 2.1868,
+      "step": 3815
+    },
+    {
+      "epoch": 3.7195715676728334,
+      "grad_norm": 1.7578731153653255,
+      "learning_rate": 5.0000000004131077e-08,
+      "loss": 2.2114,
+      "step": 3820
+    },
+    {
+      "epoch": 3.72444011684518,
+      "grad_norm": 1.7261756289309933,
+      "learning_rate": 5.000000000342294e-08,
+      "loss": 2.1714,
+      "step": 3825
+    },
+    {
+      "epoch": 3.729308666017527,
+      "grad_norm": 1.7847142724316984,
+      "learning_rate": 5.0000000002826924e-08,
+      "loss": 2.1744,
+      "step": 3830
+    },
+    {
+      "epoch": 3.7341772151898733,
+      "grad_norm": 1.7506323765963263,
+      "learning_rate": 5.0000000002326804e-08,
+      "loss": 2.2474,
+      "step": 3835
+    },
+    {
+      "epoch": 3.73904576436222,
+      "grad_norm": 1.7726668919984896,
+      "learning_rate": 5.000000000190845e-08,
+      "loss": 2.1832,
+      "step": 3840
+    },
+    {
+      "epoch": 3.7439143135345665,
+      "grad_norm": 1.791217388707096,
+      "learning_rate": 5.000000000155963e-08,
+      "loss": 2.2552,
+      "step": 3845
+    },
+    {
+      "epoch": 3.7487828627069133,
+      "grad_norm": 1.7260937822564961,
+      "learning_rate": 5.000000000126977e-08,
+      "loss": 2.2135,
+      "step": 3850
+    },
+    {
+      "epoch": 3.75365141187926,
+      "grad_norm": 1.8761923429228229,
+      "learning_rate": 5.000000000102974e-08,
+      "loss": 2.2143,
+      "step": 3855
+    },
+    {
+      "epoch": 3.7585199610516065,
+      "grad_norm": 1.8000916225190866,
+      "learning_rate": 5.000000000083169e-08,
+      "loss": 2.1963,
+      "step": 3860
+    },
+    {
+      "epoch": 3.7633885102239533,
+      "grad_norm": 1.8139182234780402,
+      "learning_rate": 5.0000000000668896e-08,
+      "loss": 2.2333,
+      "step": 3865
+    },
+    {
+      "epoch": 3.7682570593963,
+      "grad_norm": 1.779253821129295,
+      "learning_rate": 5.000000000053561e-08,
+      "loss": 2.2217,
+      "step": 3870
+    },
+    {
+      "epoch": 3.7731256085686464,
+      "grad_norm": 1.714027215371266,
+      "learning_rate": 5.0000000000426916e-08,
+      "loss": 2.2047,
+      "step": 3875
+    },
+    {
+      "epoch": 3.7779941577409932,
+      "grad_norm": 1.803155267019016,
+      "learning_rate": 5.000000000033866e-08,
+      "loss": 2.2063,
+      "step": 3880
+    },
+    {
+      "epoch": 3.78286270691334,
+      "grad_norm": 1.7914082013975081,
+      "learning_rate": 5.0000000000267316e-08,
+      "loss": 2.2199,
+      "step": 3885
+    },
+    {
+      "epoch": 3.7877312560856864,
+      "grad_norm": 1.7354663482603165,
+      "learning_rate": 5.00000000002099e-08,
+      "loss": 2.2118,
+      "step": 3890
+    },
+    {
+      "epoch": 3.792599805258033,
+      "grad_norm": 1.9003792042576655,
+      "learning_rate": 5.000000000016393e-08,
+      "loss": 2.1497,
+      "step": 3895
+    },
+    {
+      "epoch": 3.7974683544303796,
+      "grad_norm": 1.76200601661607,
+      "learning_rate": 5.0000000000127304e-08,
+      "loss": 2.1832,
+      "step": 3900
+    },
+    {
+      "epoch": 3.8023369036027264,
+      "grad_norm": 1.831527213740064,
+      "learning_rate": 5.0000000000098273e-08,
+      "loss": 2.2266,
+      "step": 3905
+    },
+    {
+      "epoch": 3.807205452775073,
+      "grad_norm": 1.796422966992554,
+      "learning_rate": 5.000000000007539e-08,
+      "loss": 2.2171,
+      "step": 3910
+    },
+    {
+      "epoch": 3.8120740019474195,
+      "grad_norm": 1.8025926494413154,
+      "learning_rate": 5.000000000005746e-08,
+      "loss": 2.2325,
+      "step": 3915
+    },
+    {
+      "epoch": 3.8169425511197663,
+      "grad_norm": 1.793793977270209,
+      "learning_rate": 5.000000000004349e-08,
+      "loss": 2.2545,
+      "step": 3920
+    },
+    {
+      "epoch": 3.8218111002921127,
+      "grad_norm": 1.7928832433425952,
+      "learning_rate": 5.000000000003268e-08,
+      "loss": 2.182,
+      "step": 3925
+    },
+    {
+      "epoch": 3.8266796494644595,
+      "grad_norm": 1.7947342233214896,
+      "learning_rate": 5.000000000002437e-08,
+      "loss": 2.1897,
+      "step": 3930
+    },
+    {
+      "epoch": 3.8315481986368063,
+      "grad_norm": 1.7630875066863383,
+      "learning_rate": 5.000000000001803e-08,
+      "loss": 2.1764,
+      "step": 3935
+    },
+    {
+      "epoch": 3.836416747809153,
+      "grad_norm": 1.7899032248006117,
+      "learning_rate": 5.0000000000013226e-08,
+      "loss": 2.1868,
+      "step": 3940
+    },
+    {
+      "epoch": 3.8412852969814995,
+      "grad_norm": 1.7615217853308245,
+      "learning_rate": 5.000000000000961e-08,
+      "loss": 2.1874,
+      "step": 3945
+    },
+    {
+      "epoch": 3.8461538461538463,
+      "grad_norm": 1.8191785253367547,
+      "learning_rate": 5.000000000000692e-08,
+      "loss": 2.2096,
+      "step": 3950
+    },
+    {
+      "epoch": 3.8510223953261926,
+      "grad_norm": 1.7142116315629525,
+      "learning_rate": 5.0000000000004934e-08,
+      "loss": 2.1825,
+      "step": 3955
+    },
+    {
+      "epoch": 3.8558909444985394,
+      "grad_norm": 1.8193971497080461,
+      "learning_rate": 5.000000000000348e-08,
+      "loss": 2.1726,
+      "step": 3960
+    },
+    {
+      "epoch": 3.8607594936708862,
+      "grad_norm": 1.7556627121596253,
+      "learning_rate": 5.0000000000002426e-08,
+      "loss": 2.2139,
+      "step": 3965
+    },
+    {
+      "epoch": 3.8656280428432326,
+      "grad_norm": 1.744589516587126,
+      "learning_rate": 5.000000000000167e-08,
+      "loss": 2.2201,
+      "step": 3970
+    },
+    {
+      "epoch": 3.8704965920155794,
+      "grad_norm": 1.714747302449748,
+      "learning_rate": 5.0000000000001136e-08,
+      "loss": 2.2009,
+      "step": 3975
+    },
+    {
+      "epoch": 3.8753651411879257,
+      "grad_norm": 1.768710509140083,
+      "learning_rate": 5.000000000000076e-08,
+      "loss": 2.2031,
+      "step": 3980
+    },
+    {
+      "epoch": 3.8802336903602725,
+      "grad_norm": 1.7322811102830442,
+      "learning_rate": 5.00000000000005e-08,
+      "loss": 2.2165,
+      "step": 3985
+    },
+    {
+      "epoch": 3.8851022395326194,
+      "grad_norm": 1.8064547108979563,
+      "learning_rate": 5.000000000000032e-08,
+      "loss": 2.251,
+      "step": 3990
+    },
+    {
+      "epoch": 3.889970788704966,
+      "grad_norm": 1.7490086693064326,
+      "learning_rate": 5.000000000000021e-08,
+      "loss": 2.2078,
+      "step": 3995
+    },
+    {
+      "epoch": 3.8948393378773125,
+      "grad_norm": 1.8272955450221657,
+      "learning_rate": 5.000000000000013e-08,
+      "loss": 2.2437,
+      "step": 4000
+    },
+    {
+      "epoch": 3.8948393378773125,
+      "eval_loss": 2.4291422367095947,
+      "eval_runtime": 85.7591,
+      "eval_samples_per_second": 85.111,
+      "eval_steps_per_second": 0.676,
+      "step": 4000
+    },
+    {
+      "epoch": 3.8997078870496593,
+      "grad_norm": 1.7590084186341215,
+      "learning_rate": 5.000000000000008e-08,
+      "loss": 2.2069,
+      "step": 4005
+    },
+    {
+      "epoch": 3.9045764362220057,
+      "grad_norm": 1.7702717336824394,
+      "learning_rate": 5.0000000000000044e-08,
+      "loss": 2.1958,
+      "step": 4010
+    },
+    {
+      "epoch": 3.9094449853943525,
+      "grad_norm": 1.7719070457541741,
+      "learning_rate": 5.0000000000000024e-08,
+      "loss": 2.1862,
+      "step": 4015
+    },
+    {
+      "epoch": 3.9143135345666993,
+      "grad_norm": 1.729790978044211,
+      "learning_rate": 5.000000000000001e-08,
+      "loss": 2.201,
+      "step": 4020
+    },
+    {
+      "epoch": 3.9191820837390456,
+      "grad_norm": 1.751806658583263,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 2.2071,
+      "step": 4025
+    },
+    {
+      "epoch": 3.9240506329113924,
+      "grad_norm": 1.7425275946141223,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 2.2775,
+      "step": 4030
+    },
+    {
+      "epoch": 3.928919182083739,
+      "grad_norm": 1.792394595138285,
+      "learning_rate": 5e-08,
+      "loss": 2.1683,
+      "step": 4035
+    },
+    {
+      "epoch": 3.9337877312560856,
+      "grad_norm": 1.7949487182841233,
+      "learning_rate": 5e-08,
+      "loss": 2.1962,
+      "step": 4040
+    },
+    {
+      "epoch": 3.9386562804284324,
+      "grad_norm": 1.8476999521081328,
+      "learning_rate": 5e-08,
+      "loss": 2.2004,
+      "step": 4045
+    },
+    {
+      "epoch": 3.943524829600779,
+      "grad_norm": 1.8225182124751913,
+      "learning_rate": 5e-08,
+      "loss": 2.2092,
+      "step": 4050
+    },
+    {
+      "epoch": 3.9483933787731256,
+      "grad_norm": 1.819449204564278,
+      "learning_rate": 5e-08,
+      "loss": 2.2543,
+      "step": 4055
+    },
+    {
+      "epoch": 3.9532619279454724,
+      "grad_norm": 1.8290111060640115,
+      "learning_rate": 5e-08,
+      "loss": 2.1963,
+      "step": 4060
+    },
+    {
+      "epoch": 3.9581304771178187,
+      "grad_norm": 1.7296724483111061,
+      "learning_rate": 5e-08,
+      "loss": 2.2167,
+      "step": 4065
+    },
+    {
+      "epoch": 3.9629990262901655,
+      "grad_norm": 1.768342281901305,
+      "learning_rate": 5e-08,
+      "loss": 2.2145,
+      "step": 4070
+    },
+    {
+      "epoch": 3.9678675754625123,
+      "grad_norm": 1.7810841025261306,
+      "learning_rate": 5e-08,
+      "loss": 2.2205,
+      "step": 4075
+    },
+    {
+      "epoch": 3.9727361246348587,
+      "grad_norm": 1.7724437878330235,
+      "learning_rate": 5e-08,
+      "loss": 2.2227,
+      "step": 4080
+    },
+    {
+      "epoch": 3.9776046738072055,
+      "grad_norm": 1.7693706025925438,
+      "learning_rate": 5e-08,
+      "loss": 2.26,
+      "step": 4085
+    },
+    {
+      "epoch": 3.982473222979552,
+      "grad_norm": 1.8547893721966033,
+      "learning_rate": 5e-08,
+      "loss": 2.1705,
+      "step": 4090
+    },
+    {
+      "epoch": 3.9873417721518987,
+      "grad_norm": 1.7856750671515411,
+      "learning_rate": 5e-08,
+      "loss": 2.2304,
+      "step": 4095
+    },
+    {
+      "epoch": 3.9922103213242455,
+      "grad_norm": 1.789882341208583,
+      "learning_rate": 5e-08,
+      "loss": 2.2391,
+      "step": 4100
+    },
+    {
+      "epoch": 3.9970788704965923,
+      "grad_norm": 1.8424154832009545,
+      "learning_rate": 5e-08,
+      "loss": 2.2157,
+      "step": 4105
+    },
+    {
+      "epoch": 4.0,
+      "step": 4108,
+      "total_flos": 429332983971840.0,
+      "train_loss": 2.284057284167586,
+      "train_runtime": 15107.0718,
+      "train_samples_per_second": 17.393,
+      "train_steps_per_second": 0.272
+    }
+  ],
+  "logging_steps": 5,
+  "max_steps": 4108,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 4,
+  "save_steps": 200,
+  "total_flos": 429332983971840.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}