diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7034 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.9237341954852492,
+  "eval_steps": 2000,
+  "global_step": 10000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0009237341954852491,
+      "grad_norm": 10.377836227416992,
+      "learning_rate": 0.0001,
+      "loss": 0.7753,
+      "step": 10
+    },
+    {
+      "epoch": 0.0018474683909704983,
+      "grad_norm": 3.7491939067840576,
+      "learning_rate": 0.0001,
+      "loss": 0.4055,
+      "step": 20
+    },
+    {
+      "epoch": 0.002771202586455747,
+      "grad_norm": 4.031465530395508,
+      "learning_rate": 0.0001,
+      "loss": 0.3717,
+      "step": 30
+    },
+    {
+      "epoch": 0.0036949367819409966,
+      "grad_norm": 2.0838451385498047,
+      "learning_rate": 0.0001,
+      "loss": 0.3579,
+      "step": 40
+    },
+    {
+      "epoch": 0.004618670977426246,
+      "grad_norm": 1.9277153015136719,
+      "learning_rate": 0.0001,
+      "loss": 0.3472,
+      "step": 50
+    },
+    {
+      "epoch": 0.005542405172911494,
+      "grad_norm": 2.6250083446502686,
+      "learning_rate": 0.0001,
+      "loss": 0.3329,
+      "step": 60
+    },
+    {
+      "epoch": 0.006466139368396744,
+      "grad_norm": 2.675837278366089,
+      "learning_rate": 0.0001,
+      "loss": 0.3285,
+      "step": 70
+    },
+    {
+      "epoch": 0.007389873563881993,
+      "grad_norm": 1.3599060773849487,
+      "learning_rate": 0.0001,
+      "loss": 0.323,
+      "step": 80
+    },
+    {
+      "epoch": 0.008313607759367242,
+      "grad_norm": 2.4787392616271973,
+      "learning_rate": 0.0001,
+      "loss": 0.3184,
+      "step": 90
+    },
+    {
+      "epoch": 0.009237341954852492,
+      "grad_norm": 1.6417860984802246,
+      "learning_rate": 0.0001,
+      "loss": 0.3176,
+      "step": 100
+    },
+    {
+      "epoch": 0.01016107615033774,
+      "grad_norm": 1.5067483186721802,
+      "learning_rate": 0.0001,
+      "loss": 0.3076,
+      "step": 110
+    },
+    {
+      "epoch": 0.011084810345822989,
+      "grad_norm": 1.5528444051742554,
+      "learning_rate": 0.0001,
+      "loss": 0.3126,
+      "step": 120
+    },
+    {
+      "epoch": 0.012008544541308239,
+      "grad_norm": 1.6968024969100952,
+      "learning_rate": 0.0001,
+      "loss": 0.3105,
+      "step": 130
+    },
+    {
+      "epoch": 0.012932278736793488,
+      "grad_norm": 2.3010127544403076,
+      "learning_rate": 0.0001,
+      "loss": 0.3115,
+      "step": 140
+    },
+    {
+      "epoch": 0.013856012932278736,
+      "grad_norm": 1.9934582710266113,
+      "learning_rate": 0.0001,
+      "loss": 0.3053,
+      "step": 150
+    },
+    {
+      "epoch": 0.014779747127763986,
+      "grad_norm": 1.4010335206985474,
+      "learning_rate": 0.0001,
+      "loss": 0.3021,
+      "step": 160
+    },
+    {
+      "epoch": 0.015703481323249235,
+      "grad_norm": 1.210365891456604,
+      "learning_rate": 0.0001,
+      "loss": 0.3021,
+      "step": 170
+    },
+    {
+      "epoch": 0.016627215518734483,
+      "grad_norm": 0.8452786803245544,
+      "learning_rate": 0.0001,
+      "loss": 0.3021,
+      "step": 180
+    },
+    {
+      "epoch": 0.01755094971421973,
+      "grad_norm": 0.8893113136291504,
+      "learning_rate": 0.0001,
+      "loss": 0.2993,
+      "step": 190
+    },
+    {
+      "epoch": 0.018474683909704984,
+      "grad_norm": 1.0572125911712646,
+      "learning_rate": 0.0001,
+      "loss": 0.2995,
+      "step": 200
+    },
+    {
+      "epoch": 0.019398418105190232,
+      "grad_norm": 1.1769524812698364,
+      "learning_rate": 0.0001,
+      "loss": 0.2991,
+      "step": 210
+    },
+    {
+      "epoch": 0.02032215230067548,
+      "grad_norm": 1.1253703832626343,
+      "learning_rate": 0.0001,
+      "loss": 0.297,
+      "step": 220
+    },
+    {
+      "epoch": 0.02124588649616073,
+      "grad_norm": 0.9401121735572815,
+      "learning_rate": 0.0001,
+      "loss": 0.2967,
+      "step": 230
+    },
+    {
+      "epoch": 0.022169620691645978,
+      "grad_norm": 0.8937715291976929,
+      "learning_rate": 0.0001,
+      "loss": 0.2914,
+      "step": 240
+    },
+    {
+      "epoch": 0.02309335488713123,
+      "grad_norm": 1.2040687799453735,
+      "learning_rate": 0.0001,
+      "loss": 0.2956,
+      "step": 250
+    },
+    {
+      "epoch": 0.024017089082616478,
+      "grad_norm": 0.8526865243911743,
+      "learning_rate": 0.0001,
+      "loss": 0.2893,
+      "step": 260
+    },
+    {
+      "epoch": 0.024940823278101727,
+      "grad_norm": 0.9076613783836365,
+      "learning_rate": 0.0001,
+      "loss": 0.2879,
+      "step": 270
+    },
+    {
+      "epoch": 0.025864557473586975,
+      "grad_norm": 0.8890297412872314,
+      "learning_rate": 0.0001,
+      "loss": 0.2837,
+      "step": 280
+    },
+    {
+      "epoch": 0.026788291669072224,
+      "grad_norm": 0.8901604413986206,
+      "learning_rate": 0.0001,
+      "loss": 0.2783,
+      "step": 290
+    },
+    {
+      "epoch": 0.027712025864557472,
+      "grad_norm": 1.1589536666870117,
+      "learning_rate": 0.0001,
+      "loss": 1.1256,
+      "step": 300
+    },
+    {
+      "epoch": 0.028635760060042724,
+      "grad_norm": 0.6688721776008606,
+      "learning_rate": 0.0001,
+      "loss": 0.2779,
+      "step": 310
+    },
+    {
+      "epoch": 0.029559494255527972,
+      "grad_norm": 0.7300515174865723,
+      "learning_rate": 0.0001,
+      "loss": 0.2769,
+      "step": 320
+    },
+    {
+      "epoch": 0.03048322845101322,
+      "grad_norm": 0.8623325228691101,
+      "learning_rate": 0.0001,
+      "loss": 0.2862,
+      "step": 330
+    },
+    {
+      "epoch": 0.03140696264649847,
+      "grad_norm": 0.8834196925163269,
+      "learning_rate": 0.0001,
+      "loss": 0.2815,
+      "step": 340
+    },
+    {
+      "epoch": 0.03233069684198372,
+      "grad_norm": 0.622388482093811,
+      "learning_rate": 0.0001,
+      "loss": 0.2776,
+      "step": 350
+    },
+    {
+      "epoch": 0.033254431037468966,
+      "grad_norm": 0.7375735640525818,
+      "learning_rate": 0.0001,
+      "loss": 0.2805,
+      "step": 360
+    },
+    {
+      "epoch": 0.034178165232954215,
+      "grad_norm": 0.7450881600379944,
+      "learning_rate": 0.0001,
+      "loss": 0.2821,
+      "step": 370
+    },
+    {
+      "epoch": 0.03510189942843946,
+      "grad_norm": 0.9458677768707275,
+      "learning_rate": 0.0001,
+      "loss": 0.2756,
+      "step": 380
+    },
+    {
+      "epoch": 0.03602563362392472,
+      "grad_norm": 0.9151942133903503,
+      "learning_rate": 0.0001,
+      "loss": 0.2795,
+      "step": 390
+    },
+    {
+      "epoch": 0.03694936781940997,
+      "grad_norm": 0.7280314564704895,
+      "learning_rate": 0.0001,
+      "loss": 0.6648,
+      "step": 400
+    },
+    {
+      "epoch": 0.037873102014895216,
+      "grad_norm": 0.7620455026626587,
+      "learning_rate": 0.0001,
+      "loss": 0.2716,
+      "step": 410
+    },
+    {
+      "epoch": 0.038796836210380464,
+      "grad_norm": 0.5944289565086365,
+      "learning_rate": 0.0001,
+      "loss": 0.2704,
+      "step": 420
+    },
+    {
+      "epoch": 0.03972057040586571,
+      "grad_norm": 1.0736445188522339,
+      "learning_rate": 0.0001,
+      "loss": 0.2728,
+      "step": 430
+    },
+    {
+      "epoch": 0.04064430460135096,
+      "grad_norm": 0.5833008885383606,
+      "learning_rate": 0.0001,
+      "loss": 0.2778,
+      "step": 440
+    },
+    {
+      "epoch": 0.04156803879683621,
+      "grad_norm": 0.6026699542999268,
+      "learning_rate": 0.0001,
+      "loss": 0.2726,
+      "step": 450
+    },
+    {
+      "epoch": 0.04249177299232146,
+      "grad_norm": 0.6159306764602661,
+      "learning_rate": 0.0001,
+      "loss": 0.2698,
+      "step": 460
+    },
+    {
+      "epoch": 0.04341550718780671,
+      "grad_norm": 0.5370445251464844,
+      "learning_rate": 0.0001,
+      "loss": 0.273,
+      "step": 470
+    },
+    {
+      "epoch": 0.044339241383291955,
+      "grad_norm": 0.6999521255493164,
+      "learning_rate": 0.0001,
+      "loss": 0.2738,
+      "step": 480
+    },
+    {
+      "epoch": 0.045262975578777204,
+      "grad_norm": 0.7118927836418152,
+      "learning_rate": 0.0001,
+      "loss": 0.2695,
+      "step": 490
+    },
+    {
+      "epoch": 0.04618670977426246,
+      "grad_norm": 0.5837290287017822,
+      "learning_rate": 0.0001,
+      "loss": 0.6541,
+      "step": 500
+    },
+    {
+      "epoch": 0.04711044396974771,
+      "grad_norm": 0.5503808259963989,
+      "learning_rate": 0.0001,
+      "loss": 0.2675,
+      "step": 510
+    },
+    {
+      "epoch": 0.048034178165232956,
+      "grad_norm": 0.7749144434928894,
+      "learning_rate": 0.0001,
+      "loss": 0.2651,
+      "step": 520
+    },
+    {
+      "epoch": 0.048957912360718205,
+      "grad_norm": 0.657133162021637,
+      "learning_rate": 0.0001,
+      "loss": 0.2663,
+      "step": 530
+    },
+    {
+      "epoch": 0.04988164655620345,
+      "grad_norm": 0.6993033289909363,
+      "learning_rate": 0.0001,
+      "loss": 0.2667,
+      "step": 540
+    },
+    {
+      "epoch": 0.0508053807516887,
+      "grad_norm": 0.6134759187698364,
+      "learning_rate": 0.0001,
+      "loss": 0.6345,
+      "step": 550
+    },
+    {
+      "epoch": 0.05172911494717395,
+      "grad_norm": 0.8342564702033997,
+      "learning_rate": 0.0001,
+      "loss": 0.2667,
+      "step": 560
+    },
+    {
+      "epoch": 0.0526528491426592,
+      "grad_norm": 0.6970536112785339,
+      "learning_rate": 0.0001,
+      "loss": 0.2743,
+      "step": 570
+    },
+    {
+      "epoch": 0.05357658333814445,
+      "grad_norm": 0.5340526103973389,
+      "learning_rate": 0.0001,
+      "loss": 0.266,
+      "step": 580
+    },
+    {
+      "epoch": 0.054500317533629695,
+      "grad_norm": 0.5798579454421997,
+      "learning_rate": 0.0001,
+      "loss": 0.2645,
+      "step": 590
+    },
+    {
+      "epoch": 0.055424051729114944,
+      "grad_norm": 0.5978800654411316,
+      "learning_rate": 0.0001,
+      "loss": 0.2658,
+      "step": 600
+    },
+    {
+      "epoch": 0.0563477859246002,
+      "grad_norm": 0.6169262528419495,
+      "learning_rate": 0.0001,
+      "loss": 0.2648,
+      "step": 610
+    },
+    {
+      "epoch": 0.05727152012008545,
+      "grad_norm": 0.49252912402153015,
+      "learning_rate": 0.0001,
+      "loss": 0.2687,
+      "step": 620
+    },
+    {
+      "epoch": 0.058195254315570696,
+      "grad_norm": 0.5621038675308228,
+      "learning_rate": 0.0001,
+      "loss": 0.2646,
+      "step": 630
+    },
+    {
+      "epoch": 0.059118988511055945,
+      "grad_norm": 0.5030767321586609,
+      "learning_rate": 0.0001,
+      "loss": 0.2677,
+      "step": 640
+    },
+    {
+      "epoch": 0.06004272270654119,
+      "grad_norm": 0.44811710715293884,
+      "learning_rate": 0.0001,
+      "loss": 0.2636,
+      "step": 650
+    },
+    {
+      "epoch": 0.06096645690202644,
+      "grad_norm": 0.645104706287384,
+      "learning_rate": 0.0001,
+      "loss": 0.2698,
+      "step": 660
+    },
+    {
+      "epoch": 0.06189019109751169,
+      "grad_norm": 0.5724523663520813,
+      "learning_rate": 0.0001,
+      "loss": 0.264,
+      "step": 670
+    },
+    {
+      "epoch": 0.06281392529299694,
+      "grad_norm": 0.49830982089042664,
+      "learning_rate": 0.0001,
+      "loss": 0.261,
+      "step": 680
+    },
+    {
+      "epoch": 0.06373765948848219,
+      "grad_norm": 0.4987528622150421,
+      "learning_rate": 0.0001,
+      "loss": 0.6213,
+      "step": 690
+    },
+    {
+      "epoch": 0.06466139368396744,
+      "grad_norm": 0.4653536081314087,
+      "learning_rate": 0.0001,
+      "loss": 0.2657,
+      "step": 700
+    },
+    {
+      "epoch": 0.06558512787945268,
+      "grad_norm": 0.6882783770561218,
+      "learning_rate": 0.0001,
+      "loss": 0.2585,
+      "step": 710
+    },
+    {
+      "epoch": 0.06650886207493793,
+      "grad_norm": 0.6810200214385986,
+      "learning_rate": 0.0001,
+      "loss": 0.2606,
+      "step": 720
+    },
+    {
+      "epoch": 0.06743259627042318,
+      "grad_norm": 0.7022980451583862,
+      "learning_rate": 0.0001,
+      "loss": 0.2623,
+      "step": 730
+    },
+    {
+      "epoch": 0.06835633046590843,
+      "grad_norm": 0.6302197575569153,
+      "learning_rate": 0.0001,
+      "loss": 0.262,
+      "step": 740
+    },
+    {
+      "epoch": 0.06928006466139368,
+      "grad_norm": 0.5974870324134827,
+      "learning_rate": 0.0001,
+      "loss": 0.2622,
+      "step": 750
+    },
+    {
+      "epoch": 0.07020379885687893,
+      "grad_norm": 0.7654616236686707,
+      "learning_rate": 0.0001,
+      "loss": 0.2613,
+      "step": 760
+    },
+    {
+      "epoch": 0.07112753305236419,
+      "grad_norm": 0.6381949186325073,
+      "learning_rate": 0.0001,
+      "loss": 0.2618,
+      "step": 770
+    },
+    {
+      "epoch": 0.07205126724784944,
+      "grad_norm": 0.5837942957878113,
+      "learning_rate": 0.0001,
+      "loss": 0.256,
+      "step": 780
+    },
+    {
+      "epoch": 0.07297500144333469,
+      "grad_norm": 0.5880801677703857,
+      "learning_rate": 0.0001,
+      "loss": 0.2551,
+      "step": 790
+    },
+    {
+      "epoch": 0.07389873563881993,
+      "grad_norm": 0.5989728569984436,
+      "learning_rate": 0.0001,
+      "loss": 0.2584,
+      "step": 800
+    },
+    {
+      "epoch": 0.07482246983430518,
+      "grad_norm": 0.5575709342956543,
+      "learning_rate": 0.0001,
+      "loss": 0.2556,
+      "step": 810
+    },
+    {
+      "epoch": 0.07574620402979043,
+      "grad_norm": 0.46623289585113525,
+      "learning_rate": 0.0001,
+      "loss": 0.2517,
+      "step": 820
+    },
+    {
+      "epoch": 0.07666993822527568,
+      "grad_norm": 0.668117105960846,
+      "learning_rate": 0.0001,
+      "loss": 0.2539,
+      "step": 830
+    },
+    {
+      "epoch": 0.07759367242076093,
+      "grad_norm": 0.5178112387657166,
+      "learning_rate": 0.0001,
+      "loss": 0.254,
+      "step": 840
+    },
+    {
+      "epoch": 0.07851740661624618,
+      "grad_norm": 0.48676905035972595,
+      "learning_rate": 0.0001,
+      "loss": 0.2554,
+      "step": 850
+    },
+    {
+      "epoch": 0.07944114081173143,
+      "grad_norm": 0.6930618286132812,
+      "learning_rate": 0.0001,
+      "loss": 0.2659,
+      "step": 860
+    },
+    {
+      "epoch": 0.08036487500721667,
+      "grad_norm": 0.6086645722389221,
+      "learning_rate": 0.0001,
+      "loss": 0.2544,
+      "step": 870
+    },
+    {
+      "epoch": 0.08128860920270192,
+      "grad_norm": 0.5384548306465149,
+      "learning_rate": 0.0001,
+      "loss": 0.2517,
+      "step": 880
+    },
+    {
+      "epoch": 0.08221234339818717,
+      "grad_norm": 0.43174371123313904,
+      "learning_rate": 0.0001,
+      "loss": 0.254,
+      "step": 890
+    },
+    {
+      "epoch": 0.08313607759367242,
+      "grad_norm": 0.6086358428001404,
+      "learning_rate": 0.0001,
+      "loss": 0.2541,
+      "step": 900
+    },
+    {
+      "epoch": 0.08405981178915767,
+      "grad_norm": 0.7949267029762268,
+      "learning_rate": 0.0001,
+      "loss": 0.6148,
+      "step": 910
+    },
+    {
+      "epoch": 0.08498354598464292,
+      "grad_norm": 0.5551185607910156,
+      "learning_rate": 0.0001,
+      "loss": 0.2555,
+      "step": 920
+    },
+    {
+      "epoch": 0.08590728018012816,
+      "grad_norm": 0.5146967172622681,
+      "learning_rate": 0.0001,
+      "loss": 0.261,
+      "step": 930
+    },
+    {
+      "epoch": 0.08683101437561341,
+      "grad_norm": 0.653384268283844,
+      "learning_rate": 0.0001,
+      "loss": 0.2576,
+      "step": 940
+    },
+    {
+      "epoch": 0.08775474857109866,
+      "grad_norm": 0.444113552570343,
+      "learning_rate": 0.0001,
+      "loss": 0.2538,
+      "step": 950
+    },
+    {
+      "epoch": 0.08867848276658391,
+      "grad_norm": 0.4909529983997345,
+      "learning_rate": 0.0001,
+      "loss": 0.2518,
+      "step": 960
+    },
+    {
+      "epoch": 0.08960221696206916,
+      "grad_norm": 0.4676177203655243,
+      "learning_rate": 0.0001,
+      "loss": 0.2495,
+      "step": 970
+    },
+    {
+      "epoch": 0.09052595115755441,
+      "grad_norm": 0.576374888420105,
+      "learning_rate": 0.0001,
+      "loss": 0.2491,
+      "step": 980
+    },
+    {
+      "epoch": 0.09144968535303967,
+      "grad_norm": 0.6059390306472778,
+      "learning_rate": 0.0001,
+      "loss": 0.2482,
+      "step": 990
+    },
+    {
+      "epoch": 0.09237341954852492,
+      "grad_norm": 0.6289136409759521,
+      "learning_rate": 0.0001,
+      "loss": 0.2523,
+      "step": 1000
+    },
+    {
+      "epoch": 0.09329715374401017,
+      "grad_norm": 0.6940639019012451,
+      "learning_rate": 0.0001,
+      "loss": 0.2508,
+      "step": 1010
+    },
+    {
+      "epoch": 0.09422088793949542,
+      "grad_norm": 0.49572697281837463,
+      "learning_rate": 0.0001,
+      "loss": 0.2526,
+      "step": 1020
+    },
+    {
+      "epoch": 0.09514462213498066,
+      "grad_norm": 0.9866797924041748,
+      "learning_rate": 0.0001,
+      "loss": 0.6079,
+      "step": 1030
+    },
+    {
+      "epoch": 0.09606835633046591,
+      "grad_norm": 0.3926860988140106,
+      "learning_rate": 0.0001,
+      "loss": 0.2502,
+      "step": 1040
+    },
+    {
+      "epoch": 0.09699209052595116,
+      "grad_norm": 0.5195950865745544,
+      "learning_rate": 0.0001,
+      "loss": 0.2505,
+      "step": 1050
+    },
+    {
+      "epoch": 0.09791582472143641,
+      "grad_norm": 0.6011677980422974,
+      "learning_rate": 0.0001,
+      "loss": 0.2531,
+      "step": 1060
+    },
+    {
+      "epoch": 0.09883955891692166,
+      "grad_norm": 0.45717713236808777,
+      "learning_rate": 0.0001,
+      "loss": 0.2539,
+      "step": 1070
+    },
+    {
+      "epoch": 0.0997632931124069,
+      "grad_norm": 0.45463839173316956,
+      "learning_rate": 0.0001,
+      "loss": 0.2573,
+      "step": 1080
+    },
+    {
+      "epoch": 0.10068702730789215,
+      "grad_norm": 0.5337753295898438,
+      "learning_rate": 0.0001,
+      "loss": 0.2466,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1016107615033774,
+      "grad_norm": 0.6251022815704346,
+      "learning_rate": 0.0001,
+      "loss": 0.2446,
+      "step": 1100
+    },
+    {
+      "epoch": 0.10253449569886265,
+      "grad_norm": 0.5094243884086609,
+      "learning_rate": 0.0001,
+      "loss": 0.2478,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1034582298943479,
+      "grad_norm": 0.4294573664665222,
+      "learning_rate": 0.0001,
+      "loss": 0.2447,
+      "step": 1120
+    },
+    {
+      "epoch": 0.10438196408983315,
+      "grad_norm": 0.619016706943512,
+      "learning_rate": 0.0001,
+      "loss": 0.6017,
+      "step": 1130
+    },
+    {
+      "epoch": 0.1053056982853184,
+      "grad_norm": 0.7880054116249084,
+      "learning_rate": 0.0001,
+      "loss": 0.2521,
+      "step": 1140
+    },
+    {
+      "epoch": 0.10622943248080365,
+      "grad_norm": 0.49156785011291504,
+      "learning_rate": 0.0001,
+      "loss": 0.2527,
+      "step": 1150
+    },
+    {
+      "epoch": 0.1071531666762889,
+      "grad_norm": 0.5413324236869812,
+      "learning_rate": 0.0001,
+      "loss": 0.2558,
+      "step": 1160
+    },
+    {
+      "epoch": 0.10807690087177414,
+      "grad_norm": 0.4010981619358063,
+      "learning_rate": 0.0001,
+      "loss": 0.245,
+      "step": 1170
+    },
+    {
+      "epoch": 0.10900063506725939,
+      "grad_norm": 0.41714775562286377,
+      "learning_rate": 0.0001,
+      "loss": 0.2462,
+      "step": 1180
+    },
+    {
+      "epoch": 0.10992436926274464,
+      "grad_norm": 0.5162696838378906,
+      "learning_rate": 0.0001,
+      "loss": 0.2494,
+      "step": 1190
+    },
+    {
+      "epoch": 0.11084810345822989,
+      "grad_norm": 0.546750545501709,
+      "learning_rate": 0.0001,
+      "loss": 0.2504,
+      "step": 1200
+    },
+    {
+      "epoch": 0.11177183765371515,
+      "grad_norm": 0.4263443648815155,
+      "learning_rate": 0.0001,
+      "loss": 0.5993,
+      "step": 1210
+    },
+    {
+      "epoch": 0.1126955718492004,
+      "grad_norm": 0.42017892003059387,
+      "learning_rate": 0.0001,
+      "loss": 0.5952,
+      "step": 1220
+    },
+    {
+      "epoch": 0.11361930604468565,
+      "grad_norm": 0.49449294805526733,
+      "learning_rate": 0.0001,
+      "loss": 0.2493,
+      "step": 1230
+    },
+    {
+      "epoch": 0.1145430402401709,
+      "grad_norm": 0.5061096549034119,
+      "learning_rate": 0.0001,
+      "loss": 0.2441,
+      "step": 1240
+    },
+    {
+      "epoch": 0.11546677443565614,
+      "grad_norm": 0.7036425471305847,
+      "learning_rate": 0.0001,
+      "loss": 0.241,
+      "step": 1250
+    },
+    {
+      "epoch": 0.11639050863114139,
+      "grad_norm": 0.46319207549095154,
+      "learning_rate": 0.0001,
+      "loss": 0.2506,
+      "step": 1260
+    },
+    {
+      "epoch": 0.11731424282662664,
+      "grad_norm": 0.5542079210281372,
+      "learning_rate": 0.0001,
+      "loss": 0.2405,
+      "step": 1270
+    },
+    {
+      "epoch": 0.11823797702211189,
+      "grad_norm": 0.5105220675468445,
+      "learning_rate": 0.0001,
+      "loss": 0.2445,
+      "step": 1280
+    },
+    {
+      "epoch": 0.11916171121759714,
+      "grad_norm": 0.5735233426094055,
+      "learning_rate": 0.0001,
+      "loss": 0.2478,
+      "step": 1290
+    },
+    {
+      "epoch": 0.12008544541308239,
+      "grad_norm": 0.4917200207710266,
+      "learning_rate": 0.0001,
+      "loss": 0.2469,
+      "step": 1300
+    },
+    {
+      "epoch": 0.12100917960856764,
+      "grad_norm": 0.4899098873138428,
+      "learning_rate": 0.0001,
+      "loss": 0.2405,
+      "step": 1310
+    },
+    {
+      "epoch": 0.12193291380405288,
+      "grad_norm": 0.42172908782958984,
+      "learning_rate": 0.0001,
+      "loss": 0.2442,
+      "step": 1320
+    },
+    {
+      "epoch": 0.12285664799953813,
+      "grad_norm": 0.5698174834251404,
+      "learning_rate": 0.0001,
+      "loss": 0.2448,
+      "step": 1330
+    },
+    {
+      "epoch": 0.12378038219502338,
+      "grad_norm": 0.43075141310691833,
+      "learning_rate": 0.0001,
+      "loss": 0.5928,
+      "step": 1340
+    },
+    {
+      "epoch": 0.12470411639050863,
+      "grad_norm": 0.4610329270362854,
+      "learning_rate": 0.0001,
+      "loss": 0.5987,
+      "step": 1350
+    },
+    {
+      "epoch": 0.12562785058599388,
+      "grad_norm": 0.4741334021091461,
+      "learning_rate": 0.0001,
+      "loss": 0.246,
+      "step": 1360
+    },
+    {
+      "epoch": 0.12655158478147913,
+      "grad_norm": 0.8104658126831055,
+      "learning_rate": 0.0001,
+      "loss": 0.2426,
+      "step": 1370
+    },
+    {
+      "epoch": 0.12747531897696437,
+      "grad_norm": 0.46889594197273254,
+      "learning_rate": 0.0001,
+      "loss": 0.2478,
+      "step": 1380
+    },
+    {
+      "epoch": 0.12839905317244962,
+      "grad_norm": 0.5051830410957336,
+      "learning_rate": 0.0001,
+      "loss": 0.2453,
+      "step": 1390
+    },
+    {
+      "epoch": 0.12932278736793487,
+      "grad_norm": 0.6313543319702148,
+      "learning_rate": 0.0001,
+      "loss": 0.247,
+      "step": 1400
+    },
+    {
+      "epoch": 0.13024652156342012,
+      "grad_norm": 0.4182421863079071,
+      "learning_rate": 0.0001,
+      "loss": 0.2441,
+      "step": 1410
+    },
+    {
+      "epoch": 0.13117025575890537,
+      "grad_norm": 0.5988187789916992,
+      "learning_rate": 0.0001,
+      "loss": 0.2449,
+      "step": 1420
+    },
+    {
+      "epoch": 0.13209398995439062,
+      "grad_norm": 0.5029876828193665,
+      "learning_rate": 0.0001,
+      "loss": 0.2391,
+      "step": 1430
+    },
+    {
+      "epoch": 0.13301772414987587,
+      "grad_norm": 0.3747965395450592,
+      "learning_rate": 0.0001,
+      "loss": 0.2396,
+      "step": 1440
+    },
+    {
+      "epoch": 0.13394145834536111,
+      "grad_norm": 0.4476030170917511,
+      "learning_rate": 0.0001,
+      "loss": 0.2484,
+      "step": 1450
+    },
+    {
+      "epoch": 0.13486519254084636,
+      "grad_norm": 0.4146966338157654,
+      "learning_rate": 0.0001,
+      "loss": 0.2439,
+      "step": 1460
+    },
+    {
+      "epoch": 0.1357889267363316,
+      "grad_norm": 0.439266562461853,
+      "learning_rate": 0.0001,
+      "loss": 0.2358,
+      "step": 1470
+    },
+    {
+      "epoch": 0.13671266093181686,
+      "grad_norm": 0.45384058356285095,
+      "learning_rate": 0.0001,
+      "loss": 0.2465,
+      "step": 1480
+    },
+    {
+      "epoch": 0.1376363951273021,
+      "grad_norm": 0.41790732741355896,
+      "learning_rate": 0.0001,
+      "loss": 0.2443,
+      "step": 1490
+    },
+    {
+      "epoch": 0.13856012932278736,
+      "grad_norm": 0.4739314317703247,
+      "learning_rate": 0.0001,
+      "loss": 0.2412,
+      "step": 1500
+    },
+    {
+      "epoch": 0.1394838635182726,
+      "grad_norm": 0.40792784094810486,
+      "learning_rate": 0.0001,
+      "loss": 0.239,
+      "step": 1510
+    },
+    {
+      "epoch": 0.14040759771375785,
+      "grad_norm": 0.38325121998786926,
+      "learning_rate": 0.0001,
+      "loss": 0.2397,
+      "step": 1520
+    },
+    {
+      "epoch": 0.1413313319092431,
+      "grad_norm": 0.5073789358139038,
+      "learning_rate": 0.0001,
+      "loss": 0.2407,
+      "step": 1530
+    },
+    {
+      "epoch": 0.14225506610472838,
+      "grad_norm": 0.42200055718421936,
+      "learning_rate": 0.0001,
+      "loss": 0.2452,
+      "step": 1540
+    },
+    {
+      "epoch": 0.14317880030021363,
+      "grad_norm": 0.42271173000335693,
+      "learning_rate": 0.0001,
+      "loss": 0.2435,
+      "step": 1550
+    },
+    {
+      "epoch": 0.14410253449569888,
+      "grad_norm": 0.4510470926761627,
+      "learning_rate": 0.0001,
+      "loss": 0.586,
+      "step": 1560
+    },
+    {
+      "epoch": 0.14502626869118412,
+      "grad_norm": 0.47332921624183655,
+      "learning_rate": 0.0001,
+      "loss": 0.2404,
+      "step": 1570
+    },
+    {
+      "epoch": 0.14595000288666937,
+      "grad_norm": 0.44405099749565125,
+      "learning_rate": 0.0001,
+      "loss": 0.2339,
+      "step": 1580
+    },
+    {
+      "epoch": 0.14687373708215462,
+      "grad_norm": 0.4454667270183563,
+      "learning_rate": 0.0001,
+      "loss": 0.2431,
+      "step": 1590
+    },
+    {
+      "epoch": 0.14779747127763987,
+      "grad_norm": 0.5730986595153809,
+      "learning_rate": 0.0001,
+      "loss": 0.2383,
+      "step": 1600
+    },
+    {
+      "epoch": 0.14872120547312512,
+      "grad_norm": 0.42024850845336914,
+      "learning_rate": 0.0001,
+      "loss": 0.2409,
+      "step": 1610
+    },
+    {
+      "epoch": 0.14964493966861037,
+      "grad_norm": 0.47302043437957764,
+      "learning_rate": 0.0001,
+      "loss": 0.5866,
+      "step": 1620
+    },
+    {
+      "epoch": 0.15056867386409561,
+      "grad_norm": 0.5812174677848816,
+      "learning_rate": 0.0001,
+      "loss": 0.2406,
+      "step": 1630
+    },
+    {
+      "epoch": 0.15149240805958086,
+      "grad_norm": 0.4626239240169525,
+      "learning_rate": 0.0001,
+      "loss": 0.2399,
+      "step": 1640
+    },
+    {
+      "epoch": 0.1524161422550661,
+      "grad_norm": 0.43332046270370483,
+      "learning_rate": 0.0001,
+      "loss": 0.2372,
+      "step": 1650
+    },
+    {
+      "epoch": 0.15333987645055136,
+      "grad_norm": 0.503600537776947,
+      "learning_rate": 0.0001,
+      "loss": 0.2353,
+      "step": 1660
+    },
+    {
+      "epoch": 0.1542636106460366,
+      "grad_norm": 0.47614091634750366,
+      "learning_rate": 0.0001,
+      "loss": 0.2374,
+      "step": 1670
+    },
+    {
+      "epoch": 0.15518734484152186,
+      "grad_norm": 0.584269106388092,
+      "learning_rate": 0.0001,
+      "loss": 0.2409,
+      "step": 1680
+    },
+    {
+      "epoch": 0.1561110790370071,
+      "grad_norm": 0.4429151117801666,
+      "learning_rate": 0.0001,
+      "loss": 0.2434,
+      "step": 1690
+    },
+    {
+      "epoch": 0.15703481323249235,
+      "grad_norm": 0.38515257835388184,
+      "learning_rate": 0.0001,
+      "loss": 0.2378,
+      "step": 1700
+    },
+    {
+      "epoch": 0.1579585474279776,
+      "grad_norm": 0.4715421795845032,
+      "learning_rate": 0.0001,
+      "loss": 0.2375,
+      "step": 1710
+    },
+    {
+      "epoch": 0.15888228162346285,
+      "grad_norm": 0.3542056679725647,
+      "learning_rate": 0.0001,
+      "loss": 0.2326,
+      "step": 1720
+    },
+    {
+      "epoch": 0.1598060158189481,
+      "grad_norm": 0.5648353695869446,
+      "learning_rate": 0.0001,
+      "loss": 0.2418,
+      "step": 1730
+    },
+    {
+      "epoch": 0.16072975001443335,
+      "grad_norm": 0.3755480647087097,
+      "learning_rate": 0.0001,
+      "loss": 0.2386,
+      "step": 1740
+    },
+    {
+      "epoch": 0.1616534842099186,
+      "grad_norm": 0.718487024307251,
+      "learning_rate": 0.0001,
+      "loss": 0.2384,
+      "step": 1750
+    },
+    {
+      "epoch": 0.16257721840540384,
+      "grad_norm": 0.48689281940460205,
+      "learning_rate": 0.0001,
+      "loss": 0.2366,
+      "step": 1760
+    },
+    {
+      "epoch": 0.1635009526008891,
+      "grad_norm": 0.44187742471694946,
+      "learning_rate": 0.0001,
+      "loss": 0.2397,
+      "step": 1770
+    },
+    {
+      "epoch": 0.16442468679637434,
+      "grad_norm": 0.5262730121612549,
+      "learning_rate": 0.0001,
+      "loss": 0.241,
+      "step": 1780
+    },
+    {
+      "epoch": 0.1653484209918596,
+      "grad_norm": 0.402015745639801,
+      "learning_rate": 0.0001,
+      "loss": 0.2357,
+      "step": 1790
+    },
+    {
+      "epoch": 0.16627215518734484,
+      "grad_norm": 0.3373963236808777,
+      "learning_rate": 0.0001,
+      "loss": 0.2305,
+      "step": 1800
+    },
+    {
+      "epoch": 0.1671958893828301,
+      "grad_norm": 0.4744202792644501,
+      "learning_rate": 0.0001,
+      "loss": 0.2385,
+      "step": 1810
+    },
+    {
+      "epoch": 0.16811962357831534,
+      "grad_norm": 0.5223276019096375,
+      "learning_rate": 0.0001,
+      "loss": 0.2349,
+      "step": 1820
+    },
+    {
+      "epoch": 0.16904335777380058,
+      "grad_norm": 0.4173218607902527,
+      "learning_rate": 0.0001,
+      "loss": 0.2352,
+      "step": 1830
+    },
+    {
+      "epoch": 0.16996709196928583,
+      "grad_norm": 0.4088299870491028,
+      "learning_rate": 0.0001,
+      "loss": 0.2391,
+      "step": 1840
+    },
+    {
+      "epoch": 0.17089082616477108,
+      "grad_norm": 0.6284809708595276,
+      "learning_rate": 0.0001,
+      "loss": 0.2299,
+      "step": 1850
+    },
+    {
+      "epoch": 0.17181456036025633,
+      "grad_norm": 0.30865636467933655,
+      "learning_rate": 0.0001,
+      "loss": 0.2354,
+      "step": 1860
+    },
+    {
+      "epoch": 0.17273829455574158,
+      "grad_norm": 0.44444739818573,
+      "learning_rate": 0.0001,
+      "loss": 0.2317,
+      "step": 1870
+    },
+    {
+      "epoch": 0.17366202875122683,
+      "grad_norm": 0.3890416622161865,
+      "learning_rate": 0.0001,
+      "loss": 0.2302,
+      "step": 1880
+    },
+    {
+      "epoch": 0.17458576294671208,
+      "grad_norm": 0.38967961072921753,
+      "learning_rate": 0.0001,
+      "loss": 0.2407,
+      "step": 1890
+    },
+    {
+      "epoch": 0.17550949714219732,
+      "grad_norm": 0.4953692555427551,
+      "learning_rate": 0.0001,
+      "loss": 0.2347,
+      "step": 1900
+    },
+    {
+      "epoch": 0.17643323133768257,
+      "grad_norm": 0.38823607563972473,
+      "learning_rate": 0.0001,
+      "loss": 0.5787,
+      "step": 1910
+    },
+    {
+      "epoch": 0.17735696553316782,
+      "grad_norm": 0.4435032606124878,
+      "learning_rate": 0.0001,
+      "loss": 0.2344,
+      "step": 1920
+    },
+    {
+      "epoch": 0.17828069972865307,
+      "grad_norm": 0.3668000400066376,
+      "learning_rate": 0.0001,
+      "loss": 0.24,
+      "step": 1930
+    },
+    {
+      "epoch": 0.17920443392413832,
+      "grad_norm": 0.3940606713294983,
+      "learning_rate": 0.0001,
+      "loss": 0.2363,
+      "step": 1940
+    },
+    {
+      "epoch": 0.18012816811962357,
+      "grad_norm": 0.44965189695358276,
+      "learning_rate": 0.0001,
+      "loss": 0.2379,
+      "step": 1950
+    },
+    {
+      "epoch": 0.18105190231510881,
+      "grad_norm": 0.4830586612224579,
+      "learning_rate": 0.0001,
+      "loss": 0.2368,
+      "step": 1960
+    },
+    {
+      "epoch": 0.18197563651059406,
+      "grad_norm": 0.4827195107936859,
+      "learning_rate": 0.0001,
+      "loss": 0.2347,
+      "step": 1970
+    },
+    {
+      "epoch": 0.18289937070607934,
+      "grad_norm": 0.4296880066394806,
+      "learning_rate": 0.0001,
+      "loss": 0.2333,
+      "step": 1980
+    },
+    {
+      "epoch": 0.1838231049015646,
+      "grad_norm": 0.35778701305389404,
+      "learning_rate": 0.0001,
+      "loss": 0.2373,
+      "step": 1990
+    },
+    {
+      "epoch": 0.18474683909704984,
+      "grad_norm": 0.3922187089920044,
+      "learning_rate": 0.0001,
+      "loss": 0.2369,
+      "step": 2000
+    },
+    {
+      "epoch": 0.18567057329253508,
+      "grad_norm": 0.4088379740715027,
+      "learning_rate": 0.0001,
+      "loss": 0.2357,
+      "step": 2010
+    },
+    {
+      "epoch": 0.18659430748802033,
+      "grad_norm": 0.5719146728515625,
+      "learning_rate": 0.0001,
+      "loss": 0.2296,
+      "step": 2020
+    },
+    {
+      "epoch": 0.18751804168350558,
+      "grad_norm": 0.6169770956039429,
+      "learning_rate": 0.0001,
+      "loss": 0.2384,
+      "step": 2030
+    },
+    {
+      "epoch": 0.18844177587899083,
+      "grad_norm": 0.36770936846733093,
+      "learning_rate": 0.0001,
+      "loss": 0.5772,
+      "step": 2040
+    },
+    {
+      "epoch": 0.18936551007447608,
+      "grad_norm": 0.43697449564933777,
+      "learning_rate": 0.0001,
+      "loss": 0.2385,
+      "step": 2050
+    },
+    {
+      "epoch": 0.19028924426996133,
+      "grad_norm": 0.4228357672691345,
+      "learning_rate": 0.0001,
+      "loss": 0.2353,
+      "step": 2060
+    },
+    {
+      "epoch": 0.19121297846544658,
+      "grad_norm": 0.42827463150024414,
+      "learning_rate": 0.0001,
+      "loss": 0.232,
+      "step": 2070
+    },
+    {
+      "epoch": 0.19213671266093182,
+      "grad_norm": 0.4047262966632843,
+      "learning_rate": 0.0001,
+      "loss": 0.2294,
+      "step": 2080
+    },
+    {
+      "epoch": 0.19306044685641707,
+      "grad_norm": 0.5330992937088013,
+      "learning_rate": 0.0001,
+      "loss": 0.2332,
+      "step": 2090
+    },
+    {
+      "epoch": 0.19398418105190232,
+      "grad_norm": 0.47420746088027954,
+      "learning_rate": 0.0001,
+      "loss": 0.2331,
+      "step": 2100
+    },
+    {
+      "epoch": 0.19490791524738757,
+      "grad_norm": 0.3812981843948364,
+      "learning_rate": 0.0001,
+      "loss": 0.2362,
+      "step": 2110
+    },
+    {
+      "epoch": 0.19583164944287282,
+      "grad_norm": 0.3819383978843689,
+      "learning_rate": 0.0001,
+      "loss": 0.231,
+      "step": 2120
+    },
+    {
+      "epoch": 0.19675538363835807,
+      "grad_norm": 0.30570483207702637,
+      "learning_rate": 0.0001,
+      "loss": 0.2268,
+      "step": 2130
+    },
+    {
+      "epoch": 0.19767911783384332,
+      "grad_norm": 0.44649067521095276,
+      "learning_rate": 0.0001,
+      "loss": 0.5768,
+      "step": 2140
+    },
+    {
+      "epoch": 0.19860285202932856,
+      "grad_norm": 0.3471992015838623,
+      "learning_rate": 0.0001,
+      "loss": 0.2314,
+      "step": 2150
+    },
+    {
+      "epoch": 0.1995265862248138,
+      "grad_norm": 0.38094842433929443,
+      "learning_rate": 0.0001,
+      "loss": 0.2293,
+      "step": 2160
+    },
+    {
+      "epoch": 0.20045032042029906,
+      "grad_norm": 0.3608592450618744,
+      "learning_rate": 0.0001,
+      "loss": 0.2284,
+      "step": 2170
+    },
+    {
+      "epoch": 0.2013740546157843,
+      "grad_norm": 0.37288790941238403,
+      "learning_rate": 0.0001,
+      "loss": 0.2745,
+      "step": 2180
+    },
+    {
+      "epoch": 0.20229778881126956,
+      "grad_norm": 0.3077997863292694,
+      "learning_rate": 0.0001,
+      "loss": 0.233,
+      "step": 2190
+    },
+    {
+      "epoch": 0.2032215230067548,
+      "grad_norm": 0.4390782415866852,
+      "learning_rate": 0.0001,
+      "loss": 0.233,
+      "step": 2200
+    },
+    {
+      "epoch": 0.20414525720224005,
+      "grad_norm": 0.450605183839798,
+      "learning_rate": 0.0001,
+      "loss": 0.2307,
+      "step": 2210
+    },
+    {
+      "epoch": 0.2050689913977253,
+      "grad_norm": 0.43789491057395935,
+      "learning_rate": 0.0001,
+      "loss": 0.2288,
+      "step": 2220
+    },
+    {
+      "epoch": 0.20599272559321055,
+      "grad_norm": 0.4309540092945099,
+      "learning_rate": 0.0001,
+      "loss": 0.2289,
+      "step": 2230
+    },
+    {
+      "epoch": 0.2069164597886958,
+      "grad_norm": 0.973432183265686,
+      "learning_rate": 0.0001,
+      "loss": 0.2342,
+      "step": 2240
+    },
+    {
+      "epoch": 0.20784019398418105,
+      "grad_norm": 0.6154965162277222,
+      "learning_rate": 0.0001,
+      "loss": 0.2324,
+      "step": 2250
+    },
+    {
+      "epoch": 0.2087639281796663,
+      "grad_norm": 0.5133722424507141,
+      "learning_rate": 0.0001,
+      "loss": 0.2317,
+      "step": 2260
+    },
+    {
+      "epoch": 0.20968766237515155,
+      "grad_norm": 0.5500303506851196,
+      "learning_rate": 0.0001,
+      "loss": 0.2325,
+      "step": 2270
+    },
+    {
+      "epoch": 0.2106113965706368,
+      "grad_norm": 0.4004790484905243,
+      "learning_rate": 0.0001,
+      "loss": 0.2348,
+      "step": 2280
+    },
+    {
+      "epoch": 0.21153513076612204,
+      "grad_norm": 0.4432072341442108,
+      "learning_rate": 0.0001,
+      "loss": 0.2298,
+      "step": 2290
+    },
+    {
+      "epoch": 0.2124588649616073,
+      "grad_norm": 0.32310307025909424,
+      "learning_rate": 0.0001,
+      "loss": 0.227,
+      "step": 2300
+    },
+    {
+      "epoch": 0.21338259915709254,
+      "grad_norm": 0.4177485406398773,
+      "learning_rate": 0.0001,
+      "loss": 0.2262,
+      "step": 2310
+    },
+    {
+      "epoch": 0.2143063333525778,
+      "grad_norm": 0.46155500411987305,
+      "learning_rate": 0.0001,
+      "loss": 0.2293,
+      "step": 2320
+    },
+    {
+      "epoch": 0.21523006754806304,
+      "grad_norm": 0.5199536085128784,
+      "learning_rate": 0.0001,
+      "loss": 0.2318,
+      "step": 2330
+    },
+    {
+      "epoch": 0.21615380174354829,
+      "grad_norm": 0.45795243978500366,
+      "learning_rate": 0.0001,
+      "loss": 0.2322,
+      "step": 2340
+    },
+    {
+      "epoch": 0.21707753593903353,
+      "grad_norm": 0.36111390590667725,
+      "learning_rate": 0.0001,
+      "loss": 0.2308,
+      "step": 2350
+    },
+    {
+      "epoch": 0.21800127013451878,
+      "grad_norm": 0.3496291935443878,
+      "learning_rate": 0.0001,
+      "loss": 0.2304,
+      "step": 2360
+    },
+    {
+      "epoch": 0.21892500433000403,
+      "grad_norm": 0.38132452964782715,
+      "learning_rate": 0.0001,
+      "loss": 0.2353,
+      "step": 2370
+    },
+    {
+      "epoch": 0.21984873852548928,
+      "grad_norm": 0.3942013084888458,
+      "learning_rate": 0.0001,
+      "loss": 0.231,
+      "step": 2380
+    },
+    {
+      "epoch": 0.22077247272097453,
+      "grad_norm": 0.45268431305885315,
+      "learning_rate": 0.0001,
+      "loss": 0.2311,
+      "step": 2390
+    },
+    {
+      "epoch": 0.22169620691645978,
+      "grad_norm": 0.5048559308052063,
+      "learning_rate": 0.0001,
+      "loss": 0.2339,
+      "step": 2400
+    },
+    {
+      "epoch": 0.22261994111194502,
+      "grad_norm": 0.4176543354988098,
+      "learning_rate": 0.0001,
+      "loss": 0.2324,
+      "step": 2410
+    },
+    {
+      "epoch": 0.2235436753074303,
+      "grad_norm": 0.49830180406570435,
+      "learning_rate": 0.0001,
+      "loss": 0.2255,
+      "step": 2420
+    },
+    {
+      "epoch": 0.22446740950291555,
+      "grad_norm": 0.45104148983955383,
+      "learning_rate": 0.0001,
+      "loss": 0.2305,
+      "step": 2430
+    },
+    {
+      "epoch": 0.2253911436984008,
+      "grad_norm": 0.5595024824142456,
+      "learning_rate": 0.0001,
+      "loss": 0.2303,
+      "step": 2440
+    },
+    {
+      "epoch": 0.22631487789388605,
+      "grad_norm": 0.3488145172595978,
+      "learning_rate": 0.0001,
+      "loss": 0.2298,
+      "step": 2450
+    },
+    {
+      "epoch": 0.2272386120893713,
+      "grad_norm": 0.4085540473461151,
+      "learning_rate": 0.0001,
+      "loss": 0.2274,
+      "step": 2460
+    },
+    {
+      "epoch": 0.22816234628485654,
+      "grad_norm": 0.3607054054737091,
+      "learning_rate": 0.0001,
+      "loss": 0.2298,
+      "step": 2470
+    },
+    {
+      "epoch": 0.2290860804803418,
+      "grad_norm": 0.46505624055862427,
+      "learning_rate": 0.0001,
+      "loss": 0.2287,
+      "step": 2480
+    },
+    {
+      "epoch": 0.23000981467582704,
+      "grad_norm": 0.3342452645301819,
+      "learning_rate": 0.0001,
+      "loss": 0.2284,
+      "step": 2490
+    },
+    {
+      "epoch": 0.2309335488713123,
+      "grad_norm": 0.383774071931839,
+      "learning_rate": 0.0001,
+      "loss": 0.2323,
+      "step": 2500
+    },
+    {
+      "epoch": 0.23185728306679754,
+      "grad_norm": 0.48557889461517334,
+      "learning_rate": 0.0001,
+      "loss": 0.2374,
+      "step": 2510
+    },
+    {
+      "epoch": 0.23278101726228279,
+      "grad_norm": 0.478712260723114,
+      "learning_rate": 0.0001,
+      "loss": 0.2357,
+      "step": 2520
+    },
+    {
+      "epoch": 0.23370475145776803,
+      "grad_norm": 0.45071640610694885,
+      "learning_rate": 0.0001,
+      "loss": 0.2325,
+      "step": 2530
+    },
+    {
+      "epoch": 0.23462848565325328,
+      "grad_norm": 0.3682229816913605,
+      "learning_rate": 0.0001,
+      "loss": 0.2281,
+      "step": 2540
+    },
+    {
+      "epoch": 0.23555221984873853,
+      "grad_norm": 0.4806053042411804,
+      "learning_rate": 0.0001,
+      "loss": 0.2289,
+      "step": 2550
+    },
+    {
+      "epoch": 0.23647595404422378,
+      "grad_norm": 0.44345441460609436,
+      "learning_rate": 0.0001,
+      "loss": 0.2339,
+      "step": 2560
+    },
+    {
+      "epoch": 0.23739968823970903,
+      "grad_norm": 0.38094815611839294,
+      "learning_rate": 0.0001,
+      "loss": 0.2303,
+      "step": 2570
+    },
+    {
+      "epoch": 0.23832342243519428,
+      "grad_norm": 0.4847588539123535,
+      "learning_rate": 0.0001,
+      "loss": 0.2267,
+      "step": 2580
+    },
+    {
+      "epoch": 0.23924715663067952,
+      "grad_norm": 0.4299144148826599,
+      "learning_rate": 0.0001,
+      "loss": 0.2285,
+      "step": 2590
+    },
+    {
+      "epoch": 0.24017089082616477,
+      "grad_norm": 0.3692493736743927,
+      "learning_rate": 0.0001,
+      "loss": 0.2292,
+      "step": 2600
+    },
+    {
+      "epoch": 0.24109462502165002,
+      "grad_norm": 0.4436064660549164,
+      "learning_rate": 0.0001,
+      "loss": 0.2259,
+      "step": 2610
+    },
+    {
+      "epoch": 0.24201835921713527,
+      "grad_norm": 0.5403839349746704,
+      "learning_rate": 0.0001,
+      "loss": 0.5703,
+      "step": 2620
+    },
+    {
+      "epoch": 0.24294209341262052,
+      "grad_norm": 0.39412447810173035,
+      "learning_rate": 0.0001,
+      "loss": 0.2295,
+      "step": 2630
+    },
+    {
+      "epoch": 0.24386582760810577,
+      "grad_norm": 0.45101356506347656,
+      "learning_rate": 0.0001,
+      "loss": 0.2322,
+      "step": 2640
+    },
+    {
+      "epoch": 0.24478956180359102,
+      "grad_norm": 0.32571062445640564,
+      "learning_rate": 0.0001,
+      "loss": 0.228,
+      "step": 2650
+    },
+    {
+      "epoch": 0.24571329599907626,
+      "grad_norm": 0.39001643657684326,
+      "learning_rate": 0.0001,
+      "loss": 0.2283,
+      "step": 2660
+    },
+    {
+      "epoch": 0.2466370301945615,
+      "grad_norm": 0.396857351064682,
+      "learning_rate": 0.0001,
+      "loss": 0.2305,
+      "step": 2670
+    },
+    {
+      "epoch": 0.24756076439004676,
+      "grad_norm": 0.40750470757484436,
+      "learning_rate": 0.0001,
+      "loss": 0.2287,
+      "step": 2680
+    },
+    {
+      "epoch": 0.248484498585532,
+      "grad_norm": 0.330258846282959,
+      "learning_rate": 0.0001,
+      "loss": 0.2237,
+      "step": 2690
+    },
+    {
+      "epoch": 0.24940823278101726,
+      "grad_norm": 0.4315931499004364,
+      "learning_rate": 0.0001,
+      "loss": 0.2278,
+      "step": 2700
+    },
+    {
+      "epoch": 0.2503319669765025,
+      "grad_norm": 0.32770416140556335,
+      "learning_rate": 0.0001,
+      "loss": 0.2297,
+      "step": 2710
+    },
+    {
+      "epoch": 0.25125570117198776,
+      "grad_norm": 0.6947807669639587,
+      "learning_rate": 0.0001,
+      "loss": 0.2345,
+      "step": 2720
+    },
+    {
+      "epoch": 0.252179435367473,
+      "grad_norm": 1.181774377822876,
+      "learning_rate": 0.0001,
+      "loss": 0.5722,
+      "step": 2730
+    },
+    {
+      "epoch": 0.25310316956295825,
+      "grad_norm": 0.4993976056575775,
+      "learning_rate": 0.0001,
+      "loss": 0.2307,
+      "step": 2740
+    },
+    {
+      "epoch": 0.2540269037584435,
+      "grad_norm": 0.3862954080104828,
+      "learning_rate": 0.0001,
+      "loss": 0.2303,
+      "step": 2750
+    },
+    {
+      "epoch": 0.25495063795392875,
+      "grad_norm": 0.4661349356174469,
+      "learning_rate": 0.0001,
+      "loss": 0.5788,
+      "step": 2760
+    },
+    {
+      "epoch": 0.255874372149414,
+      "grad_norm": 0.4150824248790741,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 2770
+    },
+    {
+      "epoch": 0.25679810634489925,
+      "grad_norm": 0.33502984046936035,
+      "learning_rate": 0.0001,
+      "loss": 0.2256,
+      "step": 2780
+    },
+    {
+      "epoch": 0.2577218405403845,
+      "grad_norm": 0.5108284950256348,
+      "learning_rate": 0.0001,
+      "loss": 0.2303,
+      "step": 2790
+    },
+    {
+      "epoch": 0.25864557473586974,
+      "grad_norm": 0.3284357488155365,
+      "learning_rate": 0.0001,
+      "loss": 0.2291,
+      "step": 2800
+    },
+    {
+      "epoch": 0.259569308931355,
+      "grad_norm": 0.40002018213272095,
+      "learning_rate": 0.0001,
+      "loss": 0.2252,
+      "step": 2810
+    },
+    {
+      "epoch": 0.26049304312684024,
+      "grad_norm": 0.4224354922771454,
+      "learning_rate": 0.0001,
+      "loss": 0.2278,
+      "step": 2820
+    },
+    {
+      "epoch": 0.2614167773223255,
+      "grad_norm": 0.4732987582683563,
+      "learning_rate": 0.0001,
+      "loss": 0.2276,
+      "step": 2830
+    },
+    {
+      "epoch": 0.26234051151781074,
+      "grad_norm": 0.3268199563026428,
+      "learning_rate": 0.0001,
+      "loss": 0.2296,
+      "step": 2840
+    },
+    {
+      "epoch": 0.263264245713296,
+      "grad_norm": 0.4282330274581909,
+      "learning_rate": 0.0001,
+      "loss": 0.2299,
+      "step": 2850
+    },
+    {
+      "epoch": 0.26418797990878123,
+      "grad_norm": 0.3902899920940399,
+      "learning_rate": 0.0001,
+      "loss": 0.228,
+      "step": 2860
+    },
+    {
+      "epoch": 0.2651117141042665,
+      "grad_norm": 0.2985314428806305,
+      "learning_rate": 0.0001,
+      "loss": 0.2234,
+      "step": 2870
+    },
+    {
+      "epoch": 0.26603544829975173,
+      "grad_norm": 0.6039664149284363,
+      "learning_rate": 0.0001,
+      "loss": 0.2293,
+      "step": 2880
+    },
+    {
+      "epoch": 0.266959182495237,
+      "grad_norm": 0.5956476926803589,
+      "learning_rate": 0.0001,
+      "loss": 0.2246,
+      "step": 2890
+    },
+    {
+      "epoch": 0.26788291669072223,
+      "grad_norm": 0.48536789417266846,
+      "learning_rate": 0.0001,
+      "loss": 0.2287,
+      "step": 2900
+    },
+    {
+      "epoch": 0.2688066508862075,
+      "grad_norm": 0.5488937497138977,
+      "learning_rate": 0.0001,
+      "loss": 0.228,
+      "step": 2910
+    },
+    {
+      "epoch": 0.2697303850816927,
+      "grad_norm": 0.34997400641441345,
+      "learning_rate": 0.0001,
+      "loss": 0.2263,
+      "step": 2920
+    },
+    {
+      "epoch": 0.270654119277178,
+      "grad_norm": 0.44494467973709106,
+      "learning_rate": 0.0001,
+      "loss": 0.2248,
+      "step": 2930
+    },
+    {
+      "epoch": 0.2715778534726632,
+      "grad_norm": 0.4488919973373413,
+      "learning_rate": 0.0001,
+      "loss": 0.2205,
+      "step": 2940
+    },
+    {
+      "epoch": 0.27250158766814847,
+      "grad_norm": 0.4541712701320648,
+      "learning_rate": 0.0001,
+      "loss": 0.2259,
+      "step": 2950
+    },
+    {
+      "epoch": 0.2734253218636337,
+      "grad_norm": 0.5734704732894897,
+      "learning_rate": 0.0001,
+      "loss": 0.5651,
+      "step": 2960
+    },
+    {
+      "epoch": 0.27434905605911897,
+      "grad_norm": 0.5139559507369995,
+      "learning_rate": 0.0001,
+      "loss": 0.2234,
+      "step": 2970
+    },
+    {
+      "epoch": 0.2752727902546042,
+      "grad_norm": 0.3888668119907379,
+      "learning_rate": 0.0001,
+      "loss": 0.2268,
+      "step": 2980
+    },
+    {
+      "epoch": 0.27619652445008946,
+      "grad_norm": 0.4781576991081238,
+      "learning_rate": 0.0001,
+      "loss": 0.2252,
+      "step": 2990
+    },
+    {
+      "epoch": 0.2771202586455747,
+      "grad_norm": 0.507675051689148,
+      "learning_rate": 0.0001,
+      "loss": 0.228,
+      "step": 3000
+    },
+    {
+      "epoch": 0.27804399284105996,
+      "grad_norm": 0.5105670690536499,
+      "learning_rate": 0.0001,
+      "loss": 0.2342,
+      "step": 3010
+    },
+    {
+      "epoch": 0.2789677270365452,
+      "grad_norm": 0.39887920022010803,
+      "learning_rate": 0.0001,
+      "loss": 0.2296,
+      "step": 3020
+    },
+    {
+      "epoch": 0.27989146123203046,
+      "grad_norm": 0.33971166610717773,
+      "learning_rate": 0.0001,
+      "loss": 0.2302,
+      "step": 3030
+    },
+    {
+      "epoch": 0.2808151954275157,
+      "grad_norm": 0.34017518162727356,
+      "learning_rate": 0.0001,
+      "loss": 0.2245,
+      "step": 3040
+    },
+    {
+      "epoch": 0.28173892962300096,
+      "grad_norm": 0.3974183201789856,
+      "learning_rate": 0.0001,
+      "loss": 0.2254,
+      "step": 3050
+    },
+    {
+      "epoch": 0.2826626638184862,
+      "grad_norm": 0.4833984673023224,
+      "learning_rate": 0.0001,
+      "loss": 0.2276,
+      "step": 3060
+    },
+    {
+      "epoch": 0.28358639801397145,
+      "grad_norm": 0.5275608897209167,
+      "learning_rate": 0.0001,
+      "loss": 0.2261,
+      "step": 3070
+    },
+    {
+      "epoch": 0.28451013220945676,
+      "grad_norm": 0.384132981300354,
+      "learning_rate": 0.0001,
+      "loss": 0.2273,
+      "step": 3080
+    },
+    {
+      "epoch": 0.285433866404942,
+      "grad_norm": 0.506726086139679,
+      "learning_rate": 0.0001,
+      "loss": 0.2257,
+      "step": 3090
+    },
+    {
+      "epoch": 0.28635760060042725,
+      "grad_norm": 0.3069717288017273,
+      "learning_rate": 0.0001,
+      "loss": 0.2245,
+      "step": 3100
+    },
+    {
+      "epoch": 0.2872813347959125,
+      "grad_norm": 0.3742898404598236,
+      "learning_rate": 0.0001,
+      "loss": 0.2239,
+      "step": 3110
+    },
+    {
+      "epoch": 0.28820506899139775,
+      "grad_norm": 0.4463520348072052,
+      "learning_rate": 0.0001,
+      "loss": 0.2247,
+      "step": 3120
+    },
+    {
+      "epoch": 0.289128803186883,
+      "grad_norm": 0.4115908443927765,
+      "learning_rate": 0.0001,
+      "loss": 0.2235,
+      "step": 3130
+    },
+    {
+      "epoch": 0.29005253738236825,
+      "grad_norm": 0.3671577274799347,
+      "learning_rate": 0.0001,
+      "loss": 0.222,
+      "step": 3140
+    },
+    {
+      "epoch": 0.2909762715778535,
+      "grad_norm": 0.39772334694862366,
+      "learning_rate": 0.0001,
+      "loss": 0.2268,
+      "step": 3150
+    },
+    {
+      "epoch": 0.29190000577333874,
+      "grad_norm": 0.33106935024261475,
+      "learning_rate": 0.0001,
+      "loss": 0.2234,
+      "step": 3160
+    },
+    {
+      "epoch": 0.292823739968824,
+      "grad_norm": 0.45472344756126404,
+      "learning_rate": 0.0001,
+      "loss": 0.5588,
+      "step": 3170
+    },
+    {
+      "epoch": 0.29374747416430924,
+      "grad_norm": 0.4509011507034302,
+      "learning_rate": 0.0001,
+      "loss": 0.2254,
+      "step": 3180
+    },
+    {
+      "epoch": 0.2946712083597945,
+      "grad_norm": 0.35373446345329285,
+      "learning_rate": 0.0001,
+      "loss": 0.2257,
+      "step": 3190
+    },
+    {
+      "epoch": 0.29559494255527974,
+      "grad_norm": 0.3876577615737915,
+      "learning_rate": 0.0001,
+      "loss": 0.2211,
+      "step": 3200
+    },
+    {
+      "epoch": 0.296518676750765,
+      "grad_norm": 0.37119758129119873,
+      "learning_rate": 0.0001,
+      "loss": 0.2192,
+      "step": 3210
+    },
+    {
+      "epoch": 0.29744241094625024,
+      "grad_norm": 0.3517870008945465,
+      "learning_rate": 0.0001,
+      "loss": 0.2245,
+      "step": 3220
+    },
+    {
+      "epoch": 0.2983661451417355,
+      "grad_norm": 0.33061233162879944,
+      "learning_rate": 0.0001,
+      "loss": 0.219,
+      "step": 3230
+    },
+    {
+      "epoch": 0.29928987933722073,
+      "grad_norm": 0.4985819458961487,
+      "learning_rate": 0.0001,
+      "loss": 0.2272,
+      "step": 3240
+    },
+    {
+      "epoch": 0.300213613532706,
+      "grad_norm": 0.33760136365890503,
+      "learning_rate": 0.0001,
+      "loss": 0.2226,
+      "step": 3250
+    },
+    {
+      "epoch": 0.30113734772819123,
+      "grad_norm": 0.30659353733062744,
+      "learning_rate": 0.0001,
+      "loss": 0.2203,
+      "step": 3260
+    },
+    {
+      "epoch": 0.3020610819236765,
+      "grad_norm": 0.37383535504341125,
+      "learning_rate": 0.0001,
+      "loss": 0.2216,
+      "step": 3270
+    },
+    {
+      "epoch": 0.3029848161191617,
+      "grad_norm": 0.3123575747013092,
+      "learning_rate": 0.0001,
+      "loss": 0.2212,
+      "step": 3280
+    },
+    {
+      "epoch": 0.303908550314647,
+      "grad_norm": 0.41256213188171387,
+      "learning_rate": 0.0001,
+      "loss": 0.218,
+      "step": 3290
+    },
+    {
+      "epoch": 0.3048322845101322,
+      "grad_norm": 0.41308340430259705,
+      "learning_rate": 0.0001,
+      "loss": 0.2234,
+      "step": 3300
+    },
+    {
+      "epoch": 0.30575601870561747,
+      "grad_norm": 0.42367058992385864,
+      "learning_rate": 0.0001,
+      "loss": 0.2268,
+      "step": 3310
+    },
+    {
+      "epoch": 0.3066797529011027,
+      "grad_norm": 0.4047527611255646,
+      "learning_rate": 0.0001,
+      "loss": 0.2224,
+      "step": 3320
+    },
+    {
+      "epoch": 0.30760348709658797,
+      "grad_norm": 0.4354739487171173,
+      "learning_rate": 0.0001,
+      "loss": 0.2221,
+      "step": 3330
+    },
+    {
+      "epoch": 0.3085272212920732,
+      "grad_norm": 0.4134977161884308,
+      "learning_rate": 0.0001,
+      "loss": 0.2238,
+      "step": 3340
+    },
+    {
+      "epoch": 0.30945095548755847,
+      "grad_norm": 0.3472785949707031,
+      "learning_rate": 0.0001,
+      "loss": 0.221,
+      "step": 3350
+    },
+    {
+      "epoch": 0.3103746896830437,
+      "grad_norm": 0.5074013471603394,
+      "learning_rate": 0.0001,
+      "loss": 0.2246,
+      "step": 3360
+    },
+    {
+      "epoch": 0.31129842387852896,
+      "grad_norm": 0.4216979146003723,
+      "learning_rate": 0.0001,
+      "loss": 0.2172,
+      "step": 3370
+    },
+    {
+      "epoch": 0.3122221580740142,
+      "grad_norm": 0.5406137108802795,
+      "learning_rate": 0.0001,
+      "loss": 0.222,
+      "step": 3380
+    },
+    {
+      "epoch": 0.31314589226949946,
+      "grad_norm": 0.39189383387565613,
+      "learning_rate": 0.0001,
+      "loss": 0.2202,
+      "step": 3390
+    },
+    {
+      "epoch": 0.3140696264649847,
+      "grad_norm": 0.32048338651657104,
+      "learning_rate": 0.0001,
+      "loss": 0.2193,
+      "step": 3400
+    },
+    {
+      "epoch": 0.31499336066046996,
+      "grad_norm": 0.39686036109924316,
+      "learning_rate": 0.0001,
+      "loss": 0.2233,
+      "step": 3410
+    },
+    {
+      "epoch": 0.3159170948559552,
+      "grad_norm": 0.3520762622356415,
+      "learning_rate": 0.0001,
+      "loss": 0.2158,
+      "step": 3420
+    },
+    {
+      "epoch": 0.31684082905144045,
+      "grad_norm": 0.3739902675151825,
+      "learning_rate": 0.0001,
+      "loss": 0.2247,
+      "step": 3430
+    },
+    {
+      "epoch": 0.3177645632469257,
+      "grad_norm": 0.42624348402023315,
+      "learning_rate": 0.0001,
+      "loss": 0.5581,
+      "step": 3440
+    },
+    {
+      "epoch": 0.31868829744241095,
+      "grad_norm": 0.4178403615951538,
+      "learning_rate": 0.0001,
+      "loss": 0.2222,
+      "step": 3450
+    },
+    {
+      "epoch": 0.3196120316378962,
+      "grad_norm": 0.4947364330291748,
+      "learning_rate": 0.0001,
+      "loss": 0.2223,
+      "step": 3460
+    },
+    {
+      "epoch": 0.32053576583338145,
+      "grad_norm": 0.3784151077270508,
+      "learning_rate": 0.0001,
+      "loss": 0.2164,
+      "step": 3470
+    },
+    {
+      "epoch": 0.3214595000288667,
+      "grad_norm": 0.46106189489364624,
+      "learning_rate": 0.0001,
+      "loss": 0.2169,
+      "step": 3480
+    },
+    {
+      "epoch": 0.32238323422435194,
+      "grad_norm": 0.29418033361434937,
+      "learning_rate": 0.0001,
+      "loss": 0.2252,
+      "step": 3490
+    },
+    {
+      "epoch": 0.3233069684198372,
+      "grad_norm": 0.3958749771118164,
+      "learning_rate": 0.0001,
+      "loss": 0.2205,
+      "step": 3500
+    },
+    {
+      "epoch": 0.32423070261532244,
+      "grad_norm": 0.3951239287853241,
+      "learning_rate": 0.0001,
+      "loss": 0.2218,
+      "step": 3510
+    },
+    {
+      "epoch": 0.3251544368108077,
+      "grad_norm": 0.32840287685394287,
+      "learning_rate": 0.0001,
+      "loss": 0.2204,
+      "step": 3520
+    },
+    {
+      "epoch": 0.32607817100629294,
+      "grad_norm": 0.3020164370536804,
+      "learning_rate": 0.0001,
+      "loss": 0.2235,
+      "step": 3530
+    },
+    {
+      "epoch": 0.3270019052017782,
+      "grad_norm": 0.4775121510028839,
+      "learning_rate": 0.0001,
+      "loss": 0.2182,
+      "step": 3540
+    },
+    {
+      "epoch": 0.32792563939726344,
+      "grad_norm": 0.3426738381385803,
+      "learning_rate": 0.0001,
+      "loss": 0.2239,
+      "step": 3550
+    },
+    {
+      "epoch": 0.3288493735927487,
+      "grad_norm": 0.37116217613220215,
+      "learning_rate": 0.0001,
+      "loss": 0.2226,
+      "step": 3560
+    },
+    {
+      "epoch": 0.32977310778823393,
+      "grad_norm": 0.5717690587043762,
+      "learning_rate": 0.0001,
+      "loss": 0.219,
+      "step": 3570
+    },
+    {
+      "epoch": 0.3306968419837192,
+      "grad_norm": 0.4740220010280609,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 3580
+    },
+    {
+      "epoch": 0.33162057617920443,
+      "grad_norm": 0.3198816776275635,
+      "learning_rate": 0.0001,
+      "loss": 0.2214,
+      "step": 3590
+    },
+    {
+      "epoch": 0.3325443103746897,
+      "grad_norm": 0.4375385046005249,
+      "learning_rate": 0.0001,
+      "loss": 0.2253,
+      "step": 3600
+    },
+    {
+      "epoch": 0.3334680445701749,
+      "grad_norm": 0.3628276586532593,
+      "learning_rate": 0.0001,
+      "loss": 0.2181,
+      "step": 3610
+    },
+    {
+      "epoch": 0.3343917787656602,
+      "grad_norm": 0.6822785139083862,
+      "learning_rate": 0.0001,
+      "loss": 0.2222,
+      "step": 3620
+    },
+    {
+      "epoch": 0.3353155129611454,
+      "grad_norm": 0.45200440287590027,
+      "learning_rate": 0.0001,
+      "loss": 0.2233,
+      "step": 3630
+    },
+    {
+      "epoch": 0.33623924715663067,
+      "grad_norm": 0.42117175459861755,
+      "learning_rate": 0.0001,
+      "loss": 0.2223,
+      "step": 3640
+    },
+    {
+      "epoch": 0.3371629813521159,
+      "grad_norm": 0.42544686794281006,
+      "learning_rate": 0.0001,
+      "loss": 0.2242,
+      "step": 3650
+    },
+    {
+      "epoch": 0.33808671554760117,
+      "grad_norm": 0.44248509407043457,
+      "learning_rate": 0.0001,
+      "loss": 0.219,
+      "step": 3660
+    },
+    {
+      "epoch": 0.3390104497430864,
+      "grad_norm": 0.32826924324035645,
+      "learning_rate": 0.0001,
+      "loss": 0.2185,
+      "step": 3670
+    },
+    {
+      "epoch": 0.33993418393857167,
+      "grad_norm": 0.4132959842681885,
+      "learning_rate": 0.0001,
+      "loss": 0.2235,
+      "step": 3680
+    },
+    {
+      "epoch": 0.3408579181340569,
+      "grad_norm": 0.4707147181034088,
+      "learning_rate": 0.0001,
+      "loss": 0.2227,
+      "step": 3690
+    },
+    {
+      "epoch": 0.34178165232954216,
+      "grad_norm": 0.55455482006073,
+      "learning_rate": 0.0001,
+      "loss": 0.2207,
+      "step": 3700
+    },
+    {
+      "epoch": 0.3427053865250274,
+      "grad_norm": 0.41418227553367615,
+      "learning_rate": 0.0001,
+      "loss": 0.5481,
+      "step": 3710
+    },
+    {
+      "epoch": 0.34362912072051266,
+      "grad_norm": 0.40281161665916443,
+      "learning_rate": 0.0001,
+      "loss": 0.2239,
+      "step": 3720
+    },
+    {
+      "epoch": 0.3445528549159979,
+      "grad_norm": 0.41627874970436096,
+      "learning_rate": 0.0001,
+      "loss": 0.3852,
+      "step": 3730
+    },
+    {
+      "epoch": 0.34547658911148316,
+      "grad_norm": 0.5185238122940063,
+      "learning_rate": 0.0001,
+      "loss": 0.2265,
+      "step": 3740
+    },
+    {
+      "epoch": 0.3464003233069684,
+      "grad_norm": 0.3757447302341461,
+      "learning_rate": 0.0001,
+      "loss": 0.2225,
+      "step": 3750
+    },
+    {
+      "epoch": 0.34732405750245365,
+      "grad_norm": 0.4384530484676361,
+      "learning_rate": 0.0001,
+      "loss": 0.2177,
+      "step": 3760
+    },
+    {
+      "epoch": 0.3482477916979389,
+      "grad_norm": 0.48497986793518066,
+      "learning_rate": 0.0001,
+      "loss": 0.2138,
+      "step": 3770
+    },
+    {
+      "epoch": 0.34917152589342415,
+      "grad_norm": 0.3871825933456421,
+      "learning_rate": 0.0001,
+      "loss": 0.5491,
+      "step": 3780
+    },
+    {
+      "epoch": 0.3500952600889094,
+      "grad_norm": 0.38259807229042053,
+      "learning_rate": 0.0001,
+      "loss": 0.2219,
+      "step": 3790
+    },
+    {
+      "epoch": 0.35101899428439465,
+      "grad_norm": 0.39300450682640076,
+      "learning_rate": 0.0001,
+      "loss": 0.2206,
+      "step": 3800
+    },
+    {
+      "epoch": 0.3519427284798799,
+      "grad_norm": 0.40899527072906494,
+      "learning_rate": 0.0001,
+      "loss": 0.2176,
+      "step": 3810
+    },
+    {
+      "epoch": 0.35286646267536514,
+      "grad_norm": 0.3768922686576843,
+      "learning_rate": 0.0001,
+      "loss": 0.2241,
+      "step": 3820
+    },
+    {
+      "epoch": 0.3537901968708504,
+      "grad_norm": 0.3415662944316864,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 3830
+    },
+    {
+      "epoch": 0.35471393106633564,
+      "grad_norm": 0.37897399067878723,
+      "learning_rate": 0.0001,
+      "loss": 0.2166,
+      "step": 3840
+    },
+    {
+      "epoch": 0.3556376652618209,
+      "grad_norm": 0.30442744493484497,
+      "learning_rate": 0.0001,
+      "loss": 0.2182,
+      "step": 3850
+    },
+    {
+      "epoch": 0.35656139945730614,
+      "grad_norm": 0.31671950221061707,
+      "learning_rate": 0.0001,
+      "loss": 0.2149,
+      "step": 3860
+    },
+    {
+      "epoch": 0.3574851336527914,
+      "grad_norm": 0.35188165307044983,
+      "learning_rate": 0.0001,
+      "loss": 0.2172,
+      "step": 3870
+    },
+    {
+      "epoch": 0.35840886784827664,
+      "grad_norm": 0.3252941071987152,
+      "learning_rate": 0.0001,
+      "loss": 0.2139,
+      "step": 3880
+    },
+    {
+      "epoch": 0.3593326020437619,
+      "grad_norm": 0.3823052644729614,
+      "learning_rate": 0.0001,
+      "loss": 0.2195,
+      "step": 3890
+    },
+    {
+      "epoch": 0.36025633623924713,
+      "grad_norm": 0.3905535340309143,
+      "learning_rate": 0.0001,
+      "loss": 0.5448,
+      "step": 3900
+    },
+    {
+      "epoch": 0.3611800704347324,
+      "grad_norm": 0.3888482451438904,
+      "learning_rate": 0.0001,
+      "loss": 0.5429,
+      "step": 3910
+    },
+    {
+      "epoch": 0.36210380463021763,
+      "grad_norm": 0.41537418961524963,
+      "learning_rate": 0.0001,
+      "loss": 0.2155,
+      "step": 3920
+    },
+    {
+      "epoch": 0.3630275388257029,
+      "grad_norm": 0.43847179412841797,
+      "learning_rate": 0.0001,
+      "loss": 0.2152,
+      "step": 3930
+    },
+    {
+      "epoch": 0.3639512730211881,
+      "grad_norm": 0.3908667266368866,
+      "learning_rate": 0.0001,
+      "loss": 0.2209,
+      "step": 3940
+    },
+    {
+      "epoch": 0.3648750072166734,
+      "grad_norm": 0.3972265124320984,
+      "learning_rate": 0.0001,
+      "loss": 0.2201,
+      "step": 3950
+    },
+    {
+      "epoch": 0.3657987414121587,
+      "grad_norm": 0.3925420343875885,
+      "learning_rate": 0.0001,
+      "loss": 0.2146,
+      "step": 3960
+    },
+    {
+      "epoch": 0.3667224756076439,
+      "grad_norm": 0.36612439155578613,
+      "learning_rate": 0.0001,
+      "loss": 0.2159,
+      "step": 3970
+    },
+    {
+      "epoch": 0.3676462098031292,
+      "grad_norm": 0.3378424346446991,
+      "learning_rate": 0.0001,
+      "loss": 0.214,
+      "step": 3980
+    },
+    {
+      "epoch": 0.3685699439986144,
+      "grad_norm": 0.38814422488212585,
+      "learning_rate": 0.0001,
+      "loss": 0.2161,
+      "step": 3990
+    },
+    {
+      "epoch": 0.3694936781940997,
+      "grad_norm": 0.4463529884815216,
+      "learning_rate": 0.0001,
+      "loss": 0.2145,
+      "step": 4000
+    },
+    {
+      "epoch": 0.3704174123895849,
+      "grad_norm": 0.359531432390213,
+      "learning_rate": 0.0001,
+      "loss": 0.2148,
+      "step": 4010
+    },
+    {
+      "epoch": 0.37134114658507017,
+      "grad_norm": 0.43108657002449036,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 4020
+    },
+    {
+      "epoch": 0.3722648807805554,
+      "grad_norm": 0.355692982673645,
+      "learning_rate": 0.0001,
+      "loss": 0.2189,
+      "step": 4030
+    },
+    {
+      "epoch": 0.37318861497604067,
+      "grad_norm": 0.4168572425842285,
+      "learning_rate": 0.0001,
+      "loss": 0.2219,
+      "step": 4040
+    },
+    {
+      "epoch": 0.3741123491715259,
+      "grad_norm": 0.4103034734725952,
+      "learning_rate": 0.0001,
+      "loss": 0.2246,
+      "step": 4050
+    },
+    {
+      "epoch": 0.37503608336701116,
+      "grad_norm": 0.42007574439048767,
+      "learning_rate": 0.0001,
+      "loss": 0.2151,
+      "step": 4060
+    },
+    {
+      "epoch": 0.3759598175624964,
+      "grad_norm": 0.563727617263794,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 4070
+    },
+    {
+      "epoch": 0.37688355175798166,
+      "grad_norm": 0.3448803424835205,
+      "learning_rate": 0.0001,
+      "loss": 0.215,
+      "step": 4080
+    },
+    {
+      "epoch": 0.3778072859534669,
+      "grad_norm": 0.35881146788597107,
+      "learning_rate": 0.0001,
+      "loss": 0.2153,
+      "step": 4090
+    },
+    {
+      "epoch": 0.37873102014895216,
+      "grad_norm": 0.42269548773765564,
+      "learning_rate": 0.0001,
+      "loss": 0.2166,
+      "step": 4100
+    },
+    {
+      "epoch": 0.3796547543444374,
+      "grad_norm": 0.3665271997451782,
+      "learning_rate": 0.0001,
+      "loss": 0.2129,
+      "step": 4110
+    },
+    {
+      "epoch": 0.38057848853992265,
+      "grad_norm": 0.3933054208755493,
+      "learning_rate": 0.0001,
+      "loss": 0.2154,
+      "step": 4120
+    },
+    {
+      "epoch": 0.3815022227354079,
+      "grad_norm": 0.348173588514328,
+      "learning_rate": 0.0001,
+      "loss": 0.2143,
+      "step": 4130
+    },
+    {
+      "epoch": 0.38242595693089315,
+      "grad_norm": 0.321692556142807,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 4140
+    },
+    {
+      "epoch": 0.3833496911263784,
+      "grad_norm": 0.37629473209381104,
+      "learning_rate": 0.0001,
+      "loss": 0.546,
+      "step": 4150
+    },
+    {
+      "epoch": 0.38427342532186365,
+      "grad_norm": 0.5757617950439453,
+      "learning_rate": 0.0001,
+      "loss": 0.221,
+      "step": 4160
+    },
+    {
+      "epoch": 0.3851971595173489,
+      "grad_norm": 0.43266335129737854,
+      "learning_rate": 0.0001,
+      "loss": 0.2153,
+      "step": 4170
+    },
+    {
+      "epoch": 0.38612089371283415,
+      "grad_norm": 0.4368360936641693,
+      "learning_rate": 0.0001,
+      "loss": 0.2228,
+      "step": 4180
+    },
+    {
+      "epoch": 0.3870446279083194,
+      "grad_norm": 0.29642513394355774,
+      "learning_rate": 0.0001,
+      "loss": 0.2138,
+      "step": 4190
+    },
+    {
+      "epoch": 0.38796836210380464,
+      "grad_norm": 0.3685188889503479,
+      "learning_rate": 0.0001,
+      "loss": 0.2155,
+      "step": 4200
+    },
+    {
+      "epoch": 0.3888920962992899,
+      "grad_norm": 0.41517990827560425,
+      "learning_rate": 0.0001,
+      "loss": 0.5366,
+      "step": 4210
+    },
+    {
+      "epoch": 0.38981583049477514,
+      "grad_norm": 0.5697425603866577,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 4220
+    },
+    {
+      "epoch": 0.3907395646902604,
+      "grad_norm": 0.29118719696998596,
+      "learning_rate": 0.0001,
+      "loss": 0.2112,
+      "step": 4230
+    },
+    {
+      "epoch": 0.39166329888574564,
+      "grad_norm": 0.39508700370788574,
+      "learning_rate": 0.0001,
+      "loss": 0.2182,
+      "step": 4240
+    },
+    {
+      "epoch": 0.3925870330812309,
+      "grad_norm": 0.374363511800766,
+      "learning_rate": 0.0001,
+      "loss": 0.2132,
+      "step": 4250
+    },
+    {
+      "epoch": 0.39351076727671613,
+      "grad_norm": 0.42270106077194214,
+      "learning_rate": 0.0001,
+      "loss": 0.2158,
+      "step": 4260
+    },
+    {
+      "epoch": 0.3944345014722014,
+      "grad_norm": 0.44743838906288147,
+      "learning_rate": 0.0001,
+      "loss": 0.2196,
+      "step": 4270
+    },
+    {
+      "epoch": 0.39535823566768663,
+      "grad_norm": 0.39016926288604736,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 4280
+    },
+    {
+      "epoch": 0.3962819698631719,
+      "grad_norm": 0.4054979681968689,
+      "learning_rate": 0.0001,
+      "loss": 0.2097,
+      "step": 4290
+    },
+    {
+      "epoch": 0.3972057040586571,
+      "grad_norm": 0.39118272066116333,
+      "learning_rate": 0.0001,
+      "loss": 0.2151,
+      "step": 4300
+    },
+    {
+      "epoch": 0.3981294382541424,
+      "grad_norm": 0.47753405570983887,
+      "learning_rate": 0.0001,
+      "loss": 0.2162,
+      "step": 4310
+    },
+    {
+      "epoch": 0.3990531724496276,
+      "grad_norm": 0.4376755356788635,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 4320
+    },
+    {
+      "epoch": 0.3999769066451129,
+      "grad_norm": 0.35959693789482117,
+      "learning_rate": 0.0001,
+      "loss": 0.213,
+      "step": 4330
+    },
+    {
+      "epoch": 0.4009006408405981,
+      "grad_norm": 0.36669230461120605,
+      "learning_rate": 0.0001,
+      "loss": 0.5367,
+      "step": 4340
+    },
+    {
+      "epoch": 0.40182437503608337,
+      "grad_norm": 0.37655845284461975,
+      "learning_rate": 0.0001,
+      "loss": 0.5361,
+      "step": 4350
+    },
+    {
+      "epoch": 0.4027481092315686,
+      "grad_norm": 0.33536455035209656,
+      "learning_rate": 0.0001,
+      "loss": 0.2155,
+      "step": 4360
+    },
+    {
+      "epoch": 0.40367184342705387,
+      "grad_norm": 0.49370628595352173,
+      "learning_rate": 0.0001,
+      "loss": 0.2119,
+      "step": 4370
+    },
+    {
+      "epoch": 0.4045955776225391,
+      "grad_norm": 0.38602909445762634,
+      "learning_rate": 0.0001,
+      "loss": 0.2163,
+      "step": 4380
+    },
+    {
+      "epoch": 0.40551931181802436,
+      "grad_norm": 0.3964555859565735,
+      "learning_rate": 0.0001,
+      "loss": 0.2142,
+      "step": 4390
+    },
+    {
+      "epoch": 0.4064430460135096,
+      "grad_norm": 0.5216159820556641,
+      "learning_rate": 0.0001,
+      "loss": 0.215,
+      "step": 4400
+    },
+    {
+      "epoch": 0.40736678020899486,
+      "grad_norm": 0.5131031274795532,
+      "learning_rate": 0.0001,
+      "loss": 0.215,
+      "step": 4410
+    },
+    {
+      "epoch": 0.4082905144044801,
+      "grad_norm": 0.3819705545902252,
+      "learning_rate": 0.0001,
+      "loss": 0.2218,
+      "step": 4420
+    },
+    {
+      "epoch": 0.40921424859996536,
+      "grad_norm": 0.41002157330513,
+      "learning_rate": 0.0001,
+      "loss": 0.2134,
+      "step": 4430
+    },
+    {
+      "epoch": 0.4101379827954506,
+      "grad_norm": 0.3606174886226654,
+      "learning_rate": 0.0001,
+      "loss": 0.2112,
+      "step": 4440
+    },
+    {
+      "epoch": 0.41106171699093585,
+      "grad_norm": 0.4673618674278259,
+      "learning_rate": 0.0001,
+      "loss": 0.2102,
+      "step": 4450
+    },
+    {
+      "epoch": 0.4119854511864211,
+      "grad_norm": 0.3812722861766815,
+      "learning_rate": 0.0001,
+      "loss": 0.2119,
+      "step": 4460
+    },
+    {
+      "epoch": 0.41290918538190635,
+      "grad_norm": 0.3641897439956665,
+      "learning_rate": 0.0001,
+      "loss": 0.2136,
+      "step": 4470
+    },
+    {
+      "epoch": 0.4138329195773916,
+      "grad_norm": 0.5161442756652832,
+      "learning_rate": 0.0001,
+      "loss": 0.217,
+      "step": 4480
+    },
+    {
+      "epoch": 0.41475665377287685,
+      "grad_norm": 0.4585149884223938,
+      "learning_rate": 0.0001,
+      "loss": 0.2099,
+      "step": 4490
+    },
+    {
+      "epoch": 0.4156803879683621,
+      "grad_norm": 0.468485563993454,
+      "learning_rate": 0.0001,
+      "loss": 0.2145,
+      "step": 4500
+    },
+    {
+      "epoch": 0.41660412216384735,
+      "grad_norm": 0.3875979781150818,
+      "learning_rate": 0.0001,
+      "loss": 0.2147,
+      "step": 4510
+    },
+    {
+      "epoch": 0.4175278563593326,
+      "grad_norm": 0.5004295110702515,
+      "learning_rate": 0.0001,
+      "loss": 0.2159,
+      "step": 4520
+    },
+    {
+      "epoch": 0.41845159055481784,
+      "grad_norm": 0.45981934666633606,
+      "learning_rate": 0.0001,
+      "loss": 0.216,
+      "step": 4530
+    },
+    {
+      "epoch": 0.4193753247503031,
+      "grad_norm": 0.40167751908302307,
+      "learning_rate": 0.0001,
+      "loss": 0.2152,
+      "step": 4540
+    },
+    {
+      "epoch": 0.42029905894578834,
+      "grad_norm": 0.41078290343284607,
+      "learning_rate": 0.0001,
+      "loss": 0.2169,
+      "step": 4550
+    },
+    {
+      "epoch": 0.4212227931412736,
+      "grad_norm": 0.4501461386680603,
+      "learning_rate": 0.0001,
+      "loss": 0.5322,
+      "step": 4560
+    },
+    {
+      "epoch": 0.42214652733675884,
+      "grad_norm": 0.3375072777271271,
+      "learning_rate": 0.0001,
+      "loss": 0.2088,
+      "step": 4570
+    },
+    {
+      "epoch": 0.4230702615322441,
+      "grad_norm": 0.43444013595581055,
+      "learning_rate": 0.0001,
+      "loss": 0.2103,
+      "step": 4580
+    },
+    {
+      "epoch": 0.42399399572772933,
+      "grad_norm": 0.6476168632507324,
+      "learning_rate": 0.0001,
+      "loss": 0.2138,
+      "step": 4590
+    },
+    {
+      "epoch": 0.4249177299232146,
+      "grad_norm": 0.4376696050167084,
+      "learning_rate": 0.0001,
+      "loss": 0.2094,
+      "step": 4600
+    },
+    {
+      "epoch": 0.42584146411869983,
+      "grad_norm": 0.37766605615615845,
+      "learning_rate": 0.0001,
+      "loss": 0.21,
+      "step": 4610
+    },
+    {
+      "epoch": 0.4267651983141851,
+      "grad_norm": 0.551823079586029,
+      "learning_rate": 0.0001,
+      "loss": 0.211,
+      "step": 4620
+    },
+    {
+      "epoch": 0.4276889325096703,
+      "grad_norm": 0.37839171290397644,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 4630
+    },
+    {
+      "epoch": 0.4286126667051556,
+      "grad_norm": 0.4079412817955017,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 4640
+    },
+    {
+      "epoch": 0.4295364009006408,
+      "grad_norm": 0.4214341342449188,
+      "learning_rate": 0.0001,
+      "loss": 0.2123,
+      "step": 4650
+    },
+    {
+      "epoch": 0.4304601350961261,
+      "grad_norm": 0.37837374210357666,
+      "learning_rate": 0.0001,
+      "loss": 0.213,
+      "step": 4660
+    },
+    {
+      "epoch": 0.4313838692916113,
+      "grad_norm": 0.4620187282562256,
+      "learning_rate": 0.0001,
+      "loss": 0.2123,
+      "step": 4670
+    },
+    {
+      "epoch": 0.43230760348709657,
+      "grad_norm": 0.3782137334346771,
+      "learning_rate": 0.0001,
+      "loss": 0.2108,
+      "step": 4680
+    },
+    {
+      "epoch": 0.4332313376825818,
+      "grad_norm": 0.43671566247940063,
+      "learning_rate": 0.0001,
+      "loss": 0.2137,
+      "step": 4690
+    },
+    {
+      "epoch": 0.43415507187806707,
+      "grad_norm": 0.4650036096572876,
+      "learning_rate": 0.0001,
+      "loss": 0.2123,
+      "step": 4700
+    },
+    {
+      "epoch": 0.4350788060735523,
+      "grad_norm": 0.3813154697418213,
+      "learning_rate": 0.0001,
+      "loss": 0.2112,
+      "step": 4710
+    },
+    {
+      "epoch": 0.43600254026903756,
+      "grad_norm": 0.40941861271858215,
+      "learning_rate": 0.0001,
+      "loss": 0.5279,
+      "step": 4720
+    },
+    {
+      "epoch": 0.4369262744645228,
+      "grad_norm": 0.5895905494689941,
+      "learning_rate": 0.0001,
+      "loss": 0.2112,
+      "step": 4730
+    },
+    {
+      "epoch": 0.43785000866000806,
+      "grad_norm": 0.4762829542160034,
+      "learning_rate": 0.0001,
+      "loss": 0.2084,
+      "step": 4740
+    },
+    {
+      "epoch": 0.4387737428554933,
+      "grad_norm": 0.4353892207145691,
+      "learning_rate": 0.0001,
+      "loss": 0.2056,
+      "step": 4750
+    },
+    {
+      "epoch": 0.43969747705097856,
+      "grad_norm": 0.3636105954647064,
+      "learning_rate": 0.0001,
+      "loss": 0.2092,
+      "step": 4760
+    },
+    {
+      "epoch": 0.4406212112464638,
+      "grad_norm": 0.3632349371910095,
+      "learning_rate": 0.0001,
+      "loss": 0.2128,
+      "step": 4770
+    },
+    {
+      "epoch": 0.44154494544194905,
+      "grad_norm": 0.3997797966003418,
+      "learning_rate": 0.0001,
+      "loss": 0.2109,
+      "step": 4780
+    },
+    {
+      "epoch": 0.4424686796374343,
+      "grad_norm": 0.3572961688041687,
+      "learning_rate": 0.0001,
+      "loss": 0.2095,
+      "step": 4790
+    },
+    {
+      "epoch": 0.44339241383291955,
+      "grad_norm": 0.5888026356697083,
+      "learning_rate": 0.0001,
+      "loss": 0.2071,
+      "step": 4800
+    },
+    {
+      "epoch": 0.4443161480284048,
+      "grad_norm": 0.5110175013542175,
+      "learning_rate": 0.0001,
+      "loss": 0.2123,
+      "step": 4810
+    },
+    {
+      "epoch": 0.44523988222389005,
+      "grad_norm": 0.5052554607391357,
+      "learning_rate": 0.0001,
+      "loss": 0.2074,
+      "step": 4820
+    },
+    {
+      "epoch": 0.4461636164193753,
+      "grad_norm": 0.3634883463382721,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 4830
+    },
+    {
+      "epoch": 0.4470873506148606,
+      "grad_norm": 0.5420510172843933,
+      "learning_rate": 0.0001,
+      "loss": 0.2067,
+      "step": 4840
+    },
+    {
+      "epoch": 0.44801108481034585,
+      "grad_norm": 0.41634082794189453,
+      "learning_rate": 0.0001,
+      "loss": 0.2066,
+      "step": 4850
+    },
+    {
+      "epoch": 0.4489348190058311,
+      "grad_norm": 0.42388445138931274,
+      "learning_rate": 0.0001,
+      "loss": 0.2105,
+      "step": 4860
+    },
+    {
+      "epoch": 0.44985855320131635,
+      "grad_norm": 0.3504672944545746,
+      "learning_rate": 0.0001,
+      "loss": 0.2138,
+      "step": 4870
+    },
+    {
+      "epoch": 0.4507822873968016,
+      "grad_norm": 0.4212481677532196,
+      "learning_rate": 0.0001,
+      "loss": 0.2121,
+      "step": 4880
+    },
+    {
+      "epoch": 0.45170602159228684,
+      "grad_norm": 0.404214471578598,
+      "learning_rate": 0.0001,
+      "loss": 0.211,
+      "step": 4890
+    },
+    {
+      "epoch": 0.4526297557877721,
+      "grad_norm": 0.33001455664634705,
+      "learning_rate": 0.0001,
+      "loss": 0.2117,
+      "step": 4900
+    },
+    {
+      "epoch": 0.45355348998325734,
+      "grad_norm": 0.44681602716445923,
+      "learning_rate": 0.0001,
+      "loss": 0.2065,
+      "step": 4910
+    },
+    {
+      "epoch": 0.4544772241787426,
+      "grad_norm": 0.38129082322120667,
+      "learning_rate": 0.0001,
+      "loss": 0.2127,
+      "step": 4920
+    },
+    {
+      "epoch": 0.45540095837422784,
+      "grad_norm": 0.6335559487342834,
+      "learning_rate": 0.0001,
+      "loss": 0.2096,
+      "step": 4930
+    },
+    {
+      "epoch": 0.4563246925697131,
+      "grad_norm": 0.34582656621932983,
+      "learning_rate": 0.0001,
+      "loss": 0.2081,
+      "step": 4940
+    },
+    {
+      "epoch": 0.45724842676519833,
+      "grad_norm": 0.5229650139808655,
+      "learning_rate": 0.0001,
+      "loss": 0.2115,
+      "step": 4950
+    },
+    {
+      "epoch": 0.4581721609606836,
+      "grad_norm": 0.34041497111320496,
+      "learning_rate": 0.0001,
+      "loss": 0.525,
+      "step": 4960
+    },
+    {
+      "epoch": 0.45909589515616883,
+      "grad_norm": 0.4637944996356964,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 4970
+    },
+    {
+      "epoch": 0.4600196293516541,
+      "grad_norm": 0.42019012570381165,
+      "learning_rate": 0.0001,
+      "loss": 0.2041,
+      "step": 4980
+    },
+    {
+      "epoch": 0.46094336354713933,
+      "grad_norm": 0.49644148349761963,
+      "learning_rate": 0.0001,
+      "loss": 0.2105,
+      "step": 4990
+    },
+    {
+      "epoch": 0.4618670977426246,
+      "grad_norm": 0.44564756751060486,
+      "learning_rate": 0.0001,
+      "loss": 0.2069,
+      "step": 5000
+    },
+    {
+      "epoch": 0.4627908319381098,
+      "grad_norm": 0.48275625705718994,
+      "learning_rate": 0.0001,
+      "loss": 0.2101,
+      "step": 5010
+    },
+    {
+      "epoch": 0.4637145661335951,
+      "grad_norm": 0.37899744510650635,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 5020
+    },
+    {
+      "epoch": 0.4646383003290803,
+      "grad_norm": 0.3303026556968689,
+      "learning_rate": 0.0001,
+      "loss": 0.2036,
+      "step": 5030
+    },
+    {
+      "epoch": 0.46556203452456557,
+      "grad_norm": 0.369429349899292,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 5040
+    },
+    {
+      "epoch": 0.4664857687200508,
+      "grad_norm": 0.5387573838233948,
+      "learning_rate": 0.0001,
+      "loss": 0.207,
+      "step": 5050
+    },
+    {
+      "epoch": 0.46740950291553607,
+      "grad_norm": 0.4687996804714203,
+      "learning_rate": 0.0001,
+      "loss": 0.2093,
+      "step": 5060
+    },
+    {
+      "epoch": 0.4683332371110213,
+      "grad_norm": 0.4017786383628845,
+      "learning_rate": 0.0001,
+      "loss": 0.208,
+      "step": 5070
+    },
+    {
+      "epoch": 0.46925697130650657,
+      "grad_norm": 0.41899704933166504,
+      "learning_rate": 0.0001,
+      "loss": 0.2074,
+      "step": 5080
+    },
+    {
+      "epoch": 0.4701807055019918,
+      "grad_norm": 0.3573046326637268,
+      "learning_rate": 0.0001,
+      "loss": 0.2097,
+      "step": 5090
+    },
+    {
+      "epoch": 0.47110443969747706,
+      "grad_norm": 0.7022538185119629,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 5100
+    },
+    {
+      "epoch": 0.4720281738929623,
+      "grad_norm": 0.403980016708374,
+      "learning_rate": 0.0001,
+      "loss": 0.2084,
+      "step": 5110
+    },
+    {
+      "epoch": 0.47295190808844756,
+      "grad_norm": 0.3694177567958832,
+      "learning_rate": 0.0001,
+      "loss": 0.2061,
+      "step": 5120
+    },
+    {
+      "epoch": 0.4738756422839328,
+      "grad_norm": 0.32916390895843506,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 5130
+    },
+    {
+      "epoch": 0.47479937647941806,
+      "grad_norm": 0.36460039019584656,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 5140
+    },
+    {
+      "epoch": 0.4757231106749033,
+      "grad_norm": 0.33318039774894714,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 5150
+    },
+    {
+      "epoch": 0.47664684487038855,
+      "grad_norm": 0.33933624625205994,
+      "learning_rate": 0.0001,
+      "loss": 0.2043,
+      "step": 5160
+    },
+    {
+      "epoch": 0.4775705790658738,
+      "grad_norm": 0.4552598297595978,
+      "learning_rate": 0.0001,
+      "loss": 0.2079,
+      "step": 5170
+    },
+    {
+      "epoch": 0.47849431326135905,
+      "grad_norm": 0.3418557047843933,
+      "learning_rate": 0.0001,
+      "loss": 0.2051,
+      "step": 5180
+    },
+    {
+      "epoch": 0.4794180474568443,
+      "grad_norm": 0.4285419285297394,
+      "learning_rate": 0.0001,
+      "loss": 0.2095,
+      "step": 5190
+    },
+    {
+      "epoch": 0.48034178165232955,
+      "grad_norm": 0.48558488488197327,
+      "learning_rate": 0.0001,
+      "loss": 0.2038,
+      "step": 5200
+    },
+    {
+      "epoch": 0.4812655158478148,
+      "grad_norm": 0.3706708550453186,
+      "learning_rate": 0.0001,
+      "loss": 0.209,
+      "step": 5210
+    },
+    {
+      "epoch": 0.48218925004330004,
+      "grad_norm": 0.39956068992614746,
+      "learning_rate": 0.0001,
+      "loss": 0.2074,
+      "step": 5220
+    },
+    {
+      "epoch": 0.4831129842387853,
+      "grad_norm": 0.5333150625228882,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 5230
+    },
+    {
+      "epoch": 0.48403671843427054,
+      "grad_norm": 0.6540710926055908,
+      "learning_rate": 0.0001,
+      "loss": 0.207,
+      "step": 5240
+    },
+    {
+      "epoch": 0.4849604526297558,
+      "grad_norm": 0.47686120867729187,
+      "learning_rate": 0.0001,
+      "loss": 0.2116,
+      "step": 5250
+    },
+    {
+      "epoch": 0.48588418682524104,
+      "grad_norm": 0.3650057911872864,
+      "learning_rate": 0.0001,
+      "loss": 0.2099,
+      "step": 5260
+    },
+    {
+      "epoch": 0.4868079210207263,
+      "grad_norm": 0.37506401538848877,
+      "learning_rate": 0.0001,
+      "loss": 0.2096,
+      "step": 5270
+    },
+    {
+      "epoch": 0.48773165521621153,
+      "grad_norm": 0.4392995536327362,
+      "learning_rate": 0.0001,
+      "loss": 0.2047,
+      "step": 5280
+    },
+    {
+      "epoch": 0.4886553894116968,
+      "grad_norm": 0.3806658089160919,
+      "learning_rate": 0.0001,
+      "loss": 0.2098,
+      "step": 5290
+    },
+    {
+      "epoch": 0.48957912360718203,
+      "grad_norm": 0.3960641324520111,
+      "learning_rate": 0.0001,
+      "loss": 0.2061,
+      "step": 5300
+    },
+    {
+      "epoch": 0.4905028578026673,
+      "grad_norm": 0.3831325173377991,
+      "learning_rate": 0.0001,
+      "loss": 0.203,
+      "step": 5310
+    },
+    {
+      "epoch": 0.49142659199815253,
+      "grad_norm": 0.46465015411376953,
+      "learning_rate": 0.0001,
+      "loss": 0.2048,
+      "step": 5320
+    },
+    {
+      "epoch": 0.4923503261936378,
+      "grad_norm": 0.426073282957077,
+      "learning_rate": 0.0001,
+      "loss": 0.1992,
+      "step": 5330
+    },
+    {
+      "epoch": 0.493274060389123,
+      "grad_norm": 0.3883862793445587,
+      "learning_rate": 0.0001,
+      "loss": 0.2063,
+      "step": 5340
+    },
+    {
+      "epoch": 0.4941977945846083,
+      "grad_norm": 0.5800668597221375,
+      "learning_rate": 0.0001,
+      "loss": 0.2054,
+      "step": 5350
+    },
+    {
+      "epoch": 0.4951215287800935,
+      "grad_norm": 0.4094981551170349,
+      "learning_rate": 0.0001,
+      "loss": 0.2052,
+      "step": 5360
+    },
+    {
+      "epoch": 0.49604526297557877,
+      "grad_norm": 0.4575849771499634,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 5370
+    },
+    {
+      "epoch": 0.496968997171064,
+      "grad_norm": 0.44456157088279724,
+      "learning_rate": 0.0001,
+      "loss": 0.208,
+      "step": 5380
+    },
+    {
+      "epoch": 0.49789273136654927,
+      "grad_norm": 0.47655409574508667,
+      "learning_rate": 0.0001,
+      "loss": 0.2085,
+      "step": 5390
+    },
+    {
+      "epoch": 0.4988164655620345,
+      "grad_norm": 0.5304370522499084,
+      "learning_rate": 0.0001,
+      "loss": 0.2029,
+      "step": 5400
+    },
+    {
+      "epoch": 0.49974019975751977,
+      "grad_norm": 0.44756844639778137,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 5410
+    },
+    {
+      "epoch": 0.500663933953005,
+      "grad_norm": 0.6823918223381042,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 5420
+    },
+    {
+      "epoch": 0.5015876681484903,
+      "grad_norm": 0.3913891613483429,
+      "learning_rate": 0.0001,
+      "loss": 0.2057,
+      "step": 5430
+    },
+    {
+      "epoch": 0.5025114023439755,
+      "grad_norm": 0.40689653158187866,
+      "learning_rate": 0.0001,
+      "loss": 0.2062,
+      "step": 5440
+    },
+    {
+      "epoch": 0.5034351365394608,
+      "grad_norm": 0.6231473684310913,
+      "learning_rate": 0.0001,
+      "loss": 0.2039,
+      "step": 5450
+    },
+    {
+      "epoch": 0.504358870734946,
+      "grad_norm": 0.3745615780353546,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 5460
+    },
+    {
+      "epoch": 0.5052826049304313,
+      "grad_norm": 0.42019689083099365,
+      "learning_rate": 0.0001,
+      "loss": 0.2066,
+      "step": 5470
+    },
+    {
+      "epoch": 0.5062063391259165,
+      "grad_norm": 0.37668848037719727,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 5480
+    },
+    {
+      "epoch": 0.5071300733214018,
+      "grad_norm": 0.4221283197402954,
+      "learning_rate": 0.0001,
+      "loss": 0.2068,
+      "step": 5490
+    },
+    {
+      "epoch": 0.508053807516887,
+      "grad_norm": 0.3790681064128876,
+      "learning_rate": 0.0001,
+      "loss": 0.2015,
+      "step": 5500
+    },
+    {
+      "epoch": 0.5089775417123723,
+      "grad_norm": 0.3551619052886963,
+      "learning_rate": 0.0001,
+      "loss": 0.203,
+      "step": 5510
+    },
+    {
+      "epoch": 0.5099012759078575,
+      "grad_norm": 0.45014235377311707,
+      "learning_rate": 0.0001,
+      "loss": 0.5225,
+      "step": 5520
+    },
+    {
+      "epoch": 0.5108250101033428,
+      "grad_norm": 0.6005759835243225,
+      "learning_rate": 0.0001,
+      "loss": 0.2062,
+      "step": 5530
+    },
+    {
+      "epoch": 0.511748744298828,
+      "grad_norm": 0.43494895100593567,
+      "learning_rate": 0.0001,
+      "loss": 0.2062,
+      "step": 5540
+    },
+    {
+      "epoch": 0.5126724784943133,
+      "grad_norm": 0.4632158577442169,
+      "learning_rate": 0.0001,
+      "loss": 0.2018,
+      "step": 5550
+    },
+    {
+      "epoch": 0.5135962126897985,
+      "grad_norm": 0.4338931739330292,
+      "learning_rate": 0.0001,
+      "loss": 0.2045,
+      "step": 5560
+    },
+    {
+      "epoch": 0.5145199468852838,
+      "grad_norm": 0.39191317558288574,
+      "learning_rate": 0.0001,
+      "loss": 0.5165,
+      "step": 5570
+    },
+    {
+      "epoch": 0.515443681080769,
+      "grad_norm": 0.37209969758987427,
+      "learning_rate": 0.0001,
+      "loss": 0.2035,
+      "step": 5580
+    },
+    {
+      "epoch": 0.5163674152762543,
+      "grad_norm": 0.4409072697162628,
+      "learning_rate": 0.0001,
+      "loss": 0.2026,
+      "step": 5590
+    },
+    {
+      "epoch": 0.5172911494717395,
+      "grad_norm": 0.4884057641029358,
+      "learning_rate": 0.0001,
+      "loss": 0.208,
+      "step": 5600
+    },
+    {
+      "epoch": 0.5182148836672248,
+      "grad_norm": 0.6604700684547424,
+      "learning_rate": 0.0001,
+      "loss": 0.2083,
+      "step": 5610
+    },
+    {
+      "epoch": 0.51913861786271,
+      "grad_norm": 0.45961496233940125,
+      "learning_rate": 0.0001,
+      "loss": 0.2062,
+      "step": 5620
+    },
+    {
+      "epoch": 0.5200623520581953,
+      "grad_norm": 0.5905351638793945,
+      "learning_rate": 0.0001,
+      "loss": 0.2033,
+      "step": 5630
+    },
+    {
+      "epoch": 0.5209860862536805,
+      "grad_norm": 0.4417837858200073,
+      "learning_rate": 0.0001,
+      "loss": 0.2014,
+      "step": 5640
+    },
+    {
+      "epoch": 0.5219098204491658,
+      "grad_norm": 0.4360920488834381,
+      "learning_rate": 0.0001,
+      "loss": 0.2066,
+      "step": 5650
+    },
+    {
+      "epoch": 0.522833554644651,
+      "grad_norm": 0.39159780740737915,
+      "learning_rate": 0.0001,
+      "loss": 0.2047,
+      "step": 5660
+    },
+    {
+      "epoch": 0.5237572888401363,
+      "grad_norm": 0.49577730894088745,
+      "learning_rate": 0.0001,
+      "loss": 0.2026,
+      "step": 5670
+    },
+    {
+      "epoch": 0.5246810230356215,
+      "grad_norm": 0.4545651376247406,
+      "learning_rate": 0.0001,
+      "loss": 0.1994,
+      "step": 5680
+    },
+    {
+      "epoch": 0.5256047572311068,
+      "grad_norm": 0.44074729084968567,
+      "learning_rate": 0.0001,
+      "loss": 0.2049,
+      "step": 5690
+    },
+    {
+      "epoch": 0.526528491426592,
+      "grad_norm": 0.3732854425907135,
+      "learning_rate": 0.0001,
+      "loss": 0.207,
+      "step": 5700
+    },
+    {
+      "epoch": 0.5274522256220773,
+      "grad_norm": 0.3655250072479248,
+      "learning_rate": 0.0001,
+      "loss": 0.2026,
+      "step": 5710
+    },
+    {
+      "epoch": 0.5283759598175625,
+      "grad_norm": 1.0543566942214966,
+      "learning_rate": 0.0001,
+      "loss": 0.2007,
+      "step": 5720
+    },
+    {
+      "epoch": 0.5292996940130478,
+      "grad_norm": 0.63393634557724,
+      "learning_rate": 0.0001,
+      "loss": 0.2004,
+      "step": 5730
+    },
+    {
+      "epoch": 0.530223428208533,
+      "grad_norm": 0.5240088701248169,
+      "learning_rate": 0.0001,
+      "loss": 0.206,
+      "step": 5740
+    },
+    {
+      "epoch": 0.5311471624040183,
+      "grad_norm": 0.3218691349029541,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 5750
+    },
+    {
+      "epoch": 0.5320708965995035,
+      "grad_norm": 0.4333510398864746,
+      "learning_rate": 0.0001,
+      "loss": 0.2009,
+      "step": 5760
+    },
+    {
+      "epoch": 0.5329946307949888,
+      "grad_norm": 0.3151180148124695,
+      "learning_rate": 0.0001,
+      "loss": 0.2003,
+      "step": 5770
+    },
+    {
+      "epoch": 0.533918364990474,
+      "grad_norm": 0.5382466912269592,
+      "learning_rate": 0.0001,
+      "loss": 0.2021,
+      "step": 5780
+    },
+    {
+      "epoch": 0.5348420991859593,
+      "grad_norm": 0.44026586413383484,
+      "learning_rate": 0.0001,
+      "loss": 0.204,
+      "step": 5790
+    },
+    {
+      "epoch": 0.5357658333814445,
+      "grad_norm": 0.3506523072719574,
+      "learning_rate": 0.0001,
+      "loss": 0.2034,
+      "step": 5800
+    },
+    {
+      "epoch": 0.5366895675769298,
+      "grad_norm": 0.5124208331108093,
+      "learning_rate": 0.0001,
+      "loss": 0.2043,
+      "step": 5810
+    },
+    {
+      "epoch": 0.537613301772415,
+      "grad_norm": 0.428034245967865,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 5820
+    },
+    {
+      "epoch": 0.5385370359679003,
+      "grad_norm": 0.38046014308929443,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 5830
+    },
+    {
+      "epoch": 0.5394607701633855,
+      "grad_norm": 0.37654662132263184,
+      "learning_rate": 0.0001,
+      "loss": 0.2018,
+      "step": 5840
+    },
+    {
+      "epoch": 0.5403845043588708,
+      "grad_norm": 0.38294240832328796,
+      "learning_rate": 0.0001,
+      "loss": 0.205,
+      "step": 5850
+    },
+    {
+      "epoch": 0.541308238554356,
+      "grad_norm": 0.5114345550537109,
+      "learning_rate": 0.0001,
+      "loss": 0.2013,
+      "step": 5860
+    },
+    {
+      "epoch": 0.5422319727498413,
+      "grad_norm": 0.39832645654678345,
+      "learning_rate": 0.0001,
+      "loss": 0.199,
+      "step": 5870
+    },
+    {
+      "epoch": 0.5431557069453264,
+      "grad_norm": 0.4142138957977295,
+      "learning_rate": 0.0001,
+      "loss": 0.1982,
+      "step": 5880
+    },
+    {
+      "epoch": 0.5440794411408117,
+      "grad_norm": 0.5544411540031433,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 5890
+    },
+    {
+      "epoch": 0.5450031753362969,
+      "grad_norm": 0.4589371383190155,
+      "learning_rate": 0.0001,
+      "loss": 0.1977,
+      "step": 5900
+    },
+    {
+      "epoch": 0.5459269095317822,
+      "grad_norm": 0.5622234344482422,
+      "learning_rate": 0.0001,
+      "loss": 0.1984,
+      "step": 5910
+    },
+    {
+      "epoch": 0.5468506437272674,
+      "grad_norm": 0.4233631193637848,
+      "learning_rate": 0.0001,
+      "loss": 0.1997,
+      "step": 5920
+    },
+    {
+      "epoch": 0.5477743779227527,
+      "grad_norm": 0.7320846915245056,
+      "learning_rate": 0.0001,
+      "loss": 0.1991,
+      "step": 5930
+    },
+    {
+      "epoch": 0.5486981121182379,
+      "grad_norm": 0.38207074999809265,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 5940
+    },
+    {
+      "epoch": 0.5496218463137232,
+      "grad_norm": 0.35408392548561096,
+      "learning_rate": 0.0001,
+      "loss": 0.1984,
+      "step": 5950
+    },
+    {
+      "epoch": 0.5505455805092084,
+      "grad_norm": 0.5252432227134705,
+      "learning_rate": 0.0001,
+      "loss": 0.2022,
+      "step": 5960
+    },
+    {
+      "epoch": 0.5514693147046937,
+      "grad_norm": 0.536348819732666,
+      "learning_rate": 0.0001,
+      "loss": 0.1986,
+      "step": 5970
+    },
+    {
+      "epoch": 0.5523930489001789,
+      "grad_norm": 0.45002058148384094,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 5980
+    },
+    {
+      "epoch": 0.5533167830956642,
+      "grad_norm": 0.8043459057807922,
+      "learning_rate": 0.0001,
+      "loss": 0.2021,
+      "step": 5990
+    },
+    {
+      "epoch": 0.5542405172911494,
+      "grad_norm": 0.3983522355556488,
+      "learning_rate": 0.0001,
+      "loss": 0.1972,
+      "step": 6000
+    },
+    {
+      "epoch": 0.5551642514866347,
+      "grad_norm": 0.7973201274871826,
+      "learning_rate": 0.0001,
+      "loss": 0.1969,
+      "step": 6010
+    },
+    {
+      "epoch": 0.5560879856821199,
+      "grad_norm": 0.4090383052825928,
+      "learning_rate": 0.0001,
+      "loss": 0.1988,
+      "step": 6020
+    },
+    {
+      "epoch": 0.5570117198776052,
+      "grad_norm": 0.4011801779270172,
+      "learning_rate": 0.0001,
+      "loss": 0.1974,
+      "step": 6030
+    },
+    {
+      "epoch": 0.5579354540730904,
+      "grad_norm": 0.34157446026802063,
+      "learning_rate": 0.0001,
+      "loss": 0.1958,
+      "step": 6040
+    },
+    {
+      "epoch": 0.5588591882685757,
+      "grad_norm": 0.4531498849391937,
+      "learning_rate": 0.0001,
+      "loss": 0.194,
+      "step": 6050
+    },
+    {
+      "epoch": 0.5597829224640609,
+      "grad_norm": 0.3886425197124481,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 6060
+    },
+    {
+      "epoch": 0.5607066566595462,
+      "grad_norm": 0.6903582811355591,
+      "learning_rate": 0.0001,
+      "loss": 0.1998,
+      "step": 6070
+    },
+    {
+      "epoch": 0.5616303908550314,
+      "grad_norm": 0.4828319251537323,
+      "learning_rate": 0.0001,
+      "loss": 0.1989,
+      "step": 6080
+    },
+    {
+      "epoch": 0.5625541250505167,
+      "grad_norm": 0.3676108717918396,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 6090
+    },
+    {
+      "epoch": 0.5634778592460019,
+      "grad_norm": 0.36137014627456665,
+      "learning_rate": 0.0001,
+      "loss": 0.198,
+      "step": 6100
+    },
+    {
+      "epoch": 0.5644015934414872,
+      "grad_norm": 0.523659884929657,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 6110
+    },
+    {
+      "epoch": 0.5653253276369724,
+      "grad_norm": 0.3308408260345459,
+      "learning_rate": 0.0001,
+      "loss": 0.1973,
+      "step": 6120
+    },
+    {
+      "epoch": 0.5662490618324577,
+      "grad_norm": 0.39561885595321655,
+      "learning_rate": 0.0001,
+      "loss": 0.2012,
+      "step": 6130
+    },
+    {
+      "epoch": 0.5671727960279429,
+      "grad_norm": 0.4854455590248108,
+      "learning_rate": 0.0001,
+      "loss": 0.2025,
+      "step": 6140
+    },
+    {
+      "epoch": 0.5680965302234282,
+      "grad_norm": 0.4626583456993103,
+      "learning_rate": 0.0001,
+      "loss": 0.203,
+      "step": 6150
+    },
+    {
+      "epoch": 0.5690202644189135,
+      "grad_norm": 0.36548173427581787,
+      "learning_rate": 0.0001,
+      "loss": 0.2001,
+      "step": 6160
+    },
+    {
+      "epoch": 0.5699439986143987,
+      "grad_norm": 0.4229353070259094,
+      "learning_rate": 0.0001,
+      "loss": 0.1982,
+      "step": 6170
+    },
+    {
+      "epoch": 0.570867732809884,
+      "grad_norm": 0.3363434076309204,
+      "learning_rate": 0.0001,
+      "loss": 0.1956,
+      "step": 6180
+    },
+    {
+      "epoch": 0.5717914670053692,
+      "grad_norm": 0.6675479412078857,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 6190
+    },
+    {
+      "epoch": 0.5727152012008545,
+      "grad_norm": 0.4045729339122772,
+      "learning_rate": 0.0001,
+      "loss": 0.1976,
+      "step": 6200
+    },
+    {
+      "epoch": 0.5736389353963397,
+      "grad_norm": 0.6667457222938538,
+      "learning_rate": 0.0001,
+      "loss": 0.1973,
+      "step": 6210
+    },
+    {
+      "epoch": 0.574562669591825,
+      "grad_norm": 0.3881010413169861,
+      "learning_rate": 0.0001,
+      "loss": 0.5135,
+      "step": 6220
+    },
+    {
+      "epoch": 0.5754864037873102,
+      "grad_norm": 0.4003308415412903,
+      "learning_rate": 0.0001,
+      "loss": 0.1966,
+      "step": 6230
+    },
+    {
+      "epoch": 0.5764101379827955,
+      "grad_norm": 0.3514181971549988,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 6240
+    },
+    {
+      "epoch": 0.5773338721782807,
+      "grad_norm": 0.46444565057754517,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 6250
+    },
+    {
+      "epoch": 0.578257606373766,
+      "grad_norm": 0.4301440715789795,
+      "learning_rate": 0.0001,
+      "loss": 0.1975,
+      "step": 6260
+    },
+    {
+      "epoch": 0.5791813405692512,
+      "grad_norm": 0.4304162561893463,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 6270
+    },
+    {
+      "epoch": 0.5801050747647365,
+      "grad_norm": 0.3856479525566101,
+      "learning_rate": 0.0001,
+      "loss": 0.1944,
+      "step": 6280
+    },
+    {
+      "epoch": 0.5810288089602217,
+      "grad_norm": 0.4279865324497223,
+      "learning_rate": 0.0001,
+      "loss": 0.1926,
+      "step": 6290
+    },
+    {
+      "epoch": 0.581952543155707,
+      "grad_norm": 0.44592881202697754,
+      "learning_rate": 0.0001,
+      "loss": 0.1936,
+      "step": 6300
+    },
+    {
+      "epoch": 0.5828762773511922,
+      "grad_norm": 0.43160223960876465,
+      "learning_rate": 0.0001,
+      "loss": 0.193,
+      "step": 6310
+    },
+    {
+      "epoch": 0.5838000115466775,
+      "grad_norm": 0.4333481192588806,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 6320
+    },
+    {
+      "epoch": 0.5847237457421627,
+      "grad_norm": 0.45316243171691895,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 6330
+    },
+    {
+      "epoch": 0.585647479937648,
+      "grad_norm": 0.2942434251308441,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 6340
+    },
+    {
+      "epoch": 0.5865712141331332,
+      "grad_norm": 0.49979010224342346,
+      "learning_rate": 0.0001,
+      "loss": 0.1925,
+      "step": 6350
+    },
+    {
+      "epoch": 0.5874949483286185,
+      "grad_norm": 0.4264412522315979,
+      "learning_rate": 0.0001,
+      "loss": 0.196,
+      "step": 6360
+    },
+    {
+      "epoch": 0.5884186825241037,
+      "grad_norm": 0.38610175251960754,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 6370
+    },
+    {
+      "epoch": 0.589342416719589,
+      "grad_norm": 0.6601687669754028,
+      "learning_rate": 0.0001,
+      "loss": 0.1967,
+      "step": 6380
+    },
+    {
+      "epoch": 0.5902661509150742,
+      "grad_norm": 0.34420180320739746,
+      "learning_rate": 0.0001,
+      "loss": 0.1993,
+      "step": 6390
+    },
+    {
+      "epoch": 0.5911898851105595,
+      "grad_norm": 0.4261429011821747,
+      "learning_rate": 0.0001,
+      "loss": 0.1932,
+      "step": 6400
+    },
+    {
+      "epoch": 0.5921136193060447,
+      "grad_norm": 0.3757234513759613,
+      "learning_rate": 0.0001,
+      "loss": 0.1971,
+      "step": 6410
+    },
+    {
+      "epoch": 0.59303735350153,
+      "grad_norm": 0.3245389461517334,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 6420
+    },
+    {
+      "epoch": 0.5939610876970152,
+      "grad_norm": 0.4480787515640259,
+      "learning_rate": 0.0001,
+      "loss": 0.1968,
+      "step": 6430
+    },
+    {
+      "epoch": 0.5948848218925005,
+      "grad_norm": 0.3903449475765228,
+      "learning_rate": 0.0001,
+      "loss": 0.197,
+      "step": 6440
+    },
+    {
+      "epoch": 0.5958085560879857,
+      "grad_norm": 0.38698723912239075,
+      "learning_rate": 0.0001,
+      "loss": 0.4993,
+      "step": 6450
+    },
+    {
+      "epoch": 0.596732290283471,
+      "grad_norm": 0.37320080399513245,
+      "learning_rate": 0.0001,
+      "loss": 0.1903,
+      "step": 6460
+    },
+    {
+      "epoch": 0.5976560244789562,
+      "grad_norm": 0.40918853878974915,
+      "learning_rate": 0.0001,
+      "loss": 0.1931,
+      "step": 6470
+    },
+    {
+      "epoch": 0.5985797586744415,
+      "grad_norm": 0.5983522534370422,
+      "learning_rate": 0.0001,
+      "loss": 0.1925,
+      "step": 6480
+    },
+    {
+      "epoch": 0.5995034928699267,
+      "grad_norm": 0.422124981880188,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 6490
+    },
+    {
+      "epoch": 0.600427227065412,
+      "grad_norm": 0.549231767654419,
+      "learning_rate": 0.0001,
+      "loss": 0.1961,
+      "step": 6500
+    },
+    {
+      "epoch": 0.6013509612608972,
+      "grad_norm": 0.47191357612609863,
+      "learning_rate": 0.0001,
+      "loss": 0.1945,
+      "step": 6510
+    },
+    {
+      "epoch": 0.6022746954563825,
+      "grad_norm": 0.40886831283569336,
+      "learning_rate": 0.0001,
+      "loss": 0.1931,
+      "step": 6520
+    },
+    {
+      "epoch": 0.6031984296518677,
+      "grad_norm": 0.5115428566932678,
+      "learning_rate": 0.0001,
+      "loss": 0.1972,
+      "step": 6530
+    },
+    {
+      "epoch": 0.604122163847353,
+      "grad_norm": 0.44815415143966675,
+      "learning_rate": 0.0001,
+      "loss": 0.1937,
+      "step": 6540
+    },
+    {
+      "epoch": 0.6050458980428381,
+      "grad_norm": 0.46659350395202637,
+      "learning_rate": 0.0001,
+      "loss": 0.195,
+      "step": 6550
+    },
+    {
+      "epoch": 0.6059696322383235,
+      "grad_norm": 0.38357239961624146,
+      "learning_rate": 0.0001,
+      "loss": 0.1963,
+      "step": 6560
+    },
+    {
+      "epoch": 0.6068933664338086,
+      "grad_norm": 0.5536012649536133,
+      "learning_rate": 0.0001,
+      "loss": 0.1927,
+      "step": 6570
+    },
+    {
+      "epoch": 0.607817100629294,
+      "grad_norm": 0.4727837145328522,
+      "learning_rate": 0.0001,
+      "loss": 0.193,
+      "step": 6580
+    },
+    {
+      "epoch": 0.6087408348247791,
+      "grad_norm": 0.4362563490867615,
+      "learning_rate": 0.0001,
+      "loss": 0.194,
+      "step": 6590
+    },
+    {
+      "epoch": 0.6096645690202644,
+      "grad_norm": 0.4345818758010864,
+      "learning_rate": 0.0001,
+      "loss": 0.1954,
+      "step": 6600
+    },
+    {
+      "epoch": 0.6105883032157496,
+      "grad_norm": 0.38879844546318054,
+      "learning_rate": 0.0001,
+      "loss": 0.1923,
+      "step": 6610
+    },
+    {
+      "epoch": 0.6115120374112349,
+      "grad_norm": 0.401516318321228,
+      "learning_rate": 0.0001,
+      "loss": 0.1909,
+      "step": 6620
+    },
+    {
+      "epoch": 0.6124357716067201,
+      "grad_norm": 0.36760014295578003,
+      "learning_rate": 0.0001,
+      "loss": 0.1922,
+      "step": 6630
+    },
+    {
+      "epoch": 0.6133595058022054,
+      "grad_norm": 0.37221571803092957,
+      "learning_rate": 0.0001,
+      "loss": 0.1837,
+      "step": 6640
+    },
+    {
+      "epoch": 0.6142832399976906,
+      "grad_norm": 0.3736136555671692,
+      "learning_rate": 0.0001,
+      "loss": 0.1886,
+      "step": 6650
+    },
+    {
+      "epoch": 0.6152069741931759,
+      "grad_norm": 0.4183785617351532,
+      "learning_rate": 0.0001,
+      "loss": 0.1943,
+      "step": 6660
+    },
+    {
+      "epoch": 0.6161307083886611,
+      "grad_norm": 0.5676806569099426,
+      "learning_rate": 0.0001,
+      "loss": 0.1957,
+      "step": 6670
+    },
+    {
+      "epoch": 0.6170544425841464,
+      "grad_norm": 0.4059545695781708,
+      "learning_rate": 0.0001,
+      "loss": 0.1929,
+      "step": 6680
+    },
+    {
+      "epoch": 0.6179781767796316,
+      "grad_norm": 0.3749110698699951,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 6690
+    },
+    {
+      "epoch": 0.6189019109751169,
+      "grad_norm": 0.34088799357414246,
+      "learning_rate": 0.0001,
+      "loss": 0.191,
+      "step": 6700
+    },
+    {
+      "epoch": 0.6198256451706021,
+      "grad_norm": 0.3927398920059204,
+      "learning_rate": 0.0001,
+      "loss": 0.1946,
+      "step": 6710
+    },
+    {
+      "epoch": 0.6207493793660874,
+      "grad_norm": 0.4718213975429535,
+      "learning_rate": 0.0001,
+      "loss": 0.1948,
+      "step": 6720
+    },
+    {
+      "epoch": 0.6216731135615726,
+      "grad_norm": 0.4251835346221924,
+      "learning_rate": 0.0001,
+      "loss": 0.5033,
+      "step": 6730
+    },
+    {
+      "epoch": 0.6225968477570579,
+      "grad_norm": 0.39704465866088867,
+      "learning_rate": 0.0001,
+      "loss": 0.1924,
+      "step": 6740
+    },
+    {
+      "epoch": 0.6235205819525431,
+      "grad_norm": 0.4783962368965149,
+      "learning_rate": 0.0001,
+      "loss": 0.1922,
+      "step": 6750
+    },
+    {
+      "epoch": 0.6244443161480284,
+      "grad_norm": 0.4048413038253784,
+      "learning_rate": 0.0001,
+      "loss": 0.1962,
+      "step": 6760
+    },
+    {
+      "epoch": 0.6253680503435136,
+      "grad_norm": 0.3915632665157318,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 6770
+    },
+    {
+      "epoch": 0.6262917845389989,
+      "grad_norm": 0.39853039383888245,
+      "learning_rate": 0.0001,
+      "loss": 0.5015,
+      "step": 6780
+    },
+    {
+      "epoch": 0.6272155187344841,
+      "grad_norm": 0.40785282850265503,
+      "learning_rate": 0.0001,
+      "loss": 0.1896,
+      "step": 6790
+    },
+    {
+      "epoch": 0.6281392529299694,
+      "grad_norm": 0.6974129676818848,
+      "learning_rate": 0.0001,
+      "loss": 0.4933,
+      "step": 6800
+    },
+    {
+      "epoch": 0.6290629871254546,
+      "grad_norm": 0.3204994797706604,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 6810
+    },
+    {
+      "epoch": 0.6299867213209399,
+      "grad_norm": 0.8155962228775024,
+      "learning_rate": 0.0001,
+      "loss": 0.1903,
+      "step": 6820
+    },
+    {
+      "epoch": 0.6309104555164251,
+      "grad_norm": 0.8928115963935852,
+      "learning_rate": 0.0001,
+      "loss": 0.1893,
+      "step": 6830
+    },
+    {
+      "epoch": 0.6318341897119104,
+      "grad_norm": 0.5847592353820801,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 6840
+    },
+    {
+      "epoch": 0.6327579239073956,
+      "grad_norm": 0.4507416784763336,
+      "learning_rate": 0.0001,
+      "loss": 0.1902,
+      "step": 6850
+    },
+    {
+      "epoch": 0.6336816581028809,
+      "grad_norm": 0.6219226121902466,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 6860
+    },
+    {
+      "epoch": 0.6346053922983661,
+      "grad_norm": 0.4082391560077667,
+      "learning_rate": 0.0001,
+      "loss": 0.19,
+      "step": 6870
+    },
+    {
+      "epoch": 0.6355291264938514,
+      "grad_norm": 0.48560988903045654,
+      "learning_rate": 0.0001,
+      "loss": 0.2649,
+      "step": 6880
+    },
+    {
+      "epoch": 0.6364528606893366,
+      "grad_norm": 0.8206206560134888,
+      "learning_rate": 0.0001,
+      "loss": 0.1883,
+      "step": 6890
+    },
+    {
+      "epoch": 0.6373765948848219,
+      "grad_norm": 0.4298798143863678,
+      "learning_rate": 0.0001,
+      "loss": 0.3829,
+      "step": 6900
+    },
+    {
+      "epoch": 0.6383003290803071,
+      "grad_norm": 0.4346946179866791,
+      "learning_rate": 0.0001,
+      "loss": 0.1901,
+      "step": 6910
+    },
+    {
+      "epoch": 0.6392240632757924,
+      "grad_norm": 0.4600197672843933,
+      "learning_rate": 0.0001,
+      "loss": 0.1916,
+      "step": 6920
+    },
+    {
+      "epoch": 0.6401477974712776,
+      "grad_norm": 0.3899524211883545,
+      "learning_rate": 0.0001,
+      "loss": 0.1898,
+      "step": 6930
+    },
+    {
+      "epoch": 0.6410715316667629,
+      "grad_norm": 0.4084889888763428,
+      "learning_rate": 0.0001,
+      "loss": 0.4899,
+      "step": 6940
+    },
+    {
+      "epoch": 0.6419952658622481,
+      "grad_norm": 0.4361218512058258,
+      "learning_rate": 0.0001,
+      "loss": 0.188,
+      "step": 6950
+    },
+    {
+      "epoch": 0.6429190000577334,
+      "grad_norm": 0.504452109336853,
+      "learning_rate": 0.0001,
+      "loss": 0.1876,
+      "step": 6960
+    },
+    {
+      "epoch": 0.6438427342532186,
+      "grad_norm": 0.4924841821193695,
+      "learning_rate": 0.0001,
+      "loss": 0.1928,
+      "step": 6970
+    },
+    {
+      "epoch": 0.6447664684487039,
+      "grad_norm": 0.47029873728752136,
+      "learning_rate": 0.0001,
+      "loss": 0.1906,
+      "step": 6980
+    },
+    {
+      "epoch": 0.6456902026441891,
+      "grad_norm": 0.3729427456855774,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 6990
+    },
+    {
+      "epoch": 0.6466139368396744,
+      "grad_norm": 0.3899003267288208,
+      "learning_rate": 0.0001,
+      "loss": 0.1912,
+      "step": 7000
+    },
+    {
+      "epoch": 0.6475376710351596,
+      "grad_norm": 0.41368311643600464,
+      "learning_rate": 0.0001,
+      "loss": 0.1891,
+      "step": 7010
+    },
+    {
+      "epoch": 0.6484614052306449,
+      "grad_norm": 0.5094907283782959,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 7020
+    },
+    {
+      "epoch": 0.6493851394261301,
+      "grad_norm": 0.442466676235199,
+      "learning_rate": 0.0001,
+      "loss": 0.1896,
+      "step": 7030
+    },
+    {
+      "epoch": 0.6503088736216154,
+      "grad_norm": 0.37364810705184937,
+      "learning_rate": 0.0001,
+      "loss": 0.1873,
+      "step": 7040
+    },
+    {
+      "epoch": 0.6512326078171007,
+      "grad_norm": 0.5380529761314392,
+      "learning_rate": 0.0001,
+      "loss": 0.1879,
+      "step": 7050
+    },
+    {
+      "epoch": 0.6521563420125859,
+      "grad_norm": 0.6053389310836792,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 7060
+    },
+    {
+      "epoch": 0.6530800762080712,
+      "grad_norm": 0.5834764838218689,
+      "learning_rate": 0.0001,
+      "loss": 0.1878,
+      "step": 7070
+    },
+    {
+      "epoch": 0.6540038104035564,
+      "grad_norm": 0.4722538888454437,
+      "learning_rate": 0.0001,
+      "loss": 0.4849,
+      "step": 7080
+    },
+    {
+      "epoch": 0.6549275445990417,
+      "grad_norm": 0.5411726832389832,
+      "learning_rate": 0.0001,
+      "loss": 0.1826,
+      "step": 7090
+    },
+    {
+      "epoch": 0.6558512787945269,
+      "grad_norm": 0.7647090554237366,
+      "learning_rate": 0.0001,
+      "loss": 0.1883,
+      "step": 7100
+    },
+    {
+      "epoch": 0.6567750129900122,
+      "grad_norm": 0.5438753962516785,
+      "learning_rate": 0.0001,
+      "loss": 0.4819,
+      "step": 7110
+    },
+    {
+      "epoch": 0.6576987471854974,
+      "grad_norm": 0.3922753632068634,
+      "learning_rate": 0.0001,
+      "loss": 0.1842,
+      "step": 7120
+    },
+    {
+      "epoch": 0.6586224813809827,
+      "grad_norm": 0.6545330882072449,
+      "learning_rate": 0.0001,
+      "loss": 0.1864,
+      "step": 7130
+    },
+    {
+      "epoch": 0.6595462155764679,
+      "grad_norm": 0.4712873697280884,
+      "learning_rate": 0.0001,
+      "loss": 0.1871,
+      "step": 7140
+    },
+    {
+      "epoch": 0.6604699497719532,
+      "grad_norm": 0.4064500033855438,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 7150
+    },
+    {
+      "epoch": 0.6613936839674384,
+      "grad_norm": 0.4518815577030182,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 7160
+    },
+    {
+      "epoch": 0.6623174181629237,
+      "grad_norm": 0.7918335795402527,
+      "learning_rate": 0.0001,
+      "loss": 0.187,
+      "step": 7170
+    },
+    {
+      "epoch": 0.6632411523584089,
+      "grad_norm": 0.48166733980178833,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 7180
+    },
+    {
+      "epoch": 0.6641648865538942,
+      "grad_norm": 0.6200546026229858,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 7190
+    },
+    {
+      "epoch": 0.6650886207493794,
+      "grad_norm": 0.6837434768676758,
+      "learning_rate": 0.0001,
+      "loss": 0.4852,
+      "step": 7200
+    },
+    {
+      "epoch": 0.6660123549448647,
+      "grad_norm": 0.6038565039634705,
+      "learning_rate": 0.0001,
+      "loss": 0.1889,
+      "step": 7210
+    },
+    {
+      "epoch": 0.6669360891403499,
+      "grad_norm": 0.7944003939628601,
+      "learning_rate": 0.0001,
+      "loss": 0.186,
+      "step": 7220
+    },
+    {
+      "epoch": 0.6678598233358352,
+      "grad_norm": 0.47415363788604736,
+      "learning_rate": 0.0001,
+      "loss": 0.1862,
+      "step": 7230
+    },
+    {
+      "epoch": 0.6687835575313203,
+      "grad_norm": 0.5343001484870911,
+      "learning_rate": 0.0001,
+      "loss": 0.1867,
+      "step": 7240
+    },
+    {
+      "epoch": 0.6697072917268057,
+      "grad_norm": 0.4093886613845825,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 7250
+    },
+    {
+      "epoch": 0.6706310259222908,
+      "grad_norm": 0.47353625297546387,
+      "learning_rate": 0.0001,
+      "loss": 0.1875,
+      "step": 7260
+    },
+    {
+      "epoch": 0.6715547601177762,
+      "grad_norm": 0.6467652916908264,
+      "learning_rate": 0.0001,
+      "loss": 0.1834,
+      "step": 7270
+    },
+    {
+      "epoch": 0.6724784943132613,
+      "grad_norm": 0.561173677444458,
+      "learning_rate": 0.0001,
+      "loss": 0.1872,
+      "step": 7280
+    },
+    {
+      "epoch": 0.6734022285087466,
+      "grad_norm": 0.44280222058296204,
+      "learning_rate": 0.0001,
+      "loss": 0.1855,
+      "step": 7290
+    },
+    {
+      "epoch": 0.6743259627042318,
+      "grad_norm": 0.45622822642326355,
+      "learning_rate": 0.0001,
+      "loss": 0.1817,
+      "step": 7300
+    },
+    {
+      "epoch": 0.6752496968997171,
+      "grad_norm": 0.43211063742637634,
+      "learning_rate": 0.0001,
+      "loss": 0.4778,
+      "step": 7310
+    },
+    {
+      "epoch": 0.6761734310952023,
+      "grad_norm": 0.4531904458999634,
+      "learning_rate": 0.0001,
+      "loss": 0.1853,
+      "step": 7320
+    },
+    {
+      "epoch": 0.6770971652906876,
+      "grad_norm": 0.4295077621936798,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 7330
+    },
+    {
+      "epoch": 0.6780208994861728,
+      "grad_norm": 0.4116787612438202,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 7340
+    },
+    {
+      "epoch": 0.6789446336816581,
+      "grad_norm": 0.4715608060359955,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 7350
+    },
+    {
+      "epoch": 0.6798683678771433,
+      "grad_norm": 0.49722760915756226,
+      "learning_rate": 0.0001,
+      "loss": 0.1852,
+      "step": 7360
+    },
+    {
+      "epoch": 0.6807921020726286,
+      "grad_norm": 0.6285380721092224,
+      "learning_rate": 0.0001,
+      "loss": 0.183,
+      "step": 7370
+    },
+    {
+      "epoch": 0.6817158362681138,
+      "grad_norm": 0.6626774668693542,
+      "learning_rate": 0.0001,
+      "loss": 0.1848,
+      "step": 7380
+    },
+    {
+      "epoch": 0.6826395704635991,
+      "grad_norm": 0.3493516445159912,
+      "learning_rate": 0.0001,
+      "loss": 0.1816,
+      "step": 7390
+    },
+    {
+      "epoch": 0.6835633046590843,
+      "grad_norm": 0.39721986651420593,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 7400
+    },
+    {
+      "epoch": 0.6844870388545696,
+      "grad_norm": 0.36072245240211487,
+      "learning_rate": 0.0001,
+      "loss": 0.1834,
+      "step": 7410
+    },
+    {
+      "epoch": 0.6854107730500548,
+      "grad_norm": 0.3699972629547119,
+      "learning_rate": 0.0001,
+      "loss": 0.1838,
+      "step": 7420
+    },
+    {
+      "epoch": 0.6863345072455401,
+      "grad_norm": 0.353380024433136,
+      "learning_rate": 0.0001,
+      "loss": 0.1839,
+      "step": 7430
+    },
+    {
+      "epoch": 0.6872582414410253,
+      "grad_norm": 0.43993157148361206,
+      "learning_rate": 0.0001,
+      "loss": 0.1808,
+      "step": 7440
+    },
+    {
+      "epoch": 0.6881819756365106,
+      "grad_norm": 0.3863504230976105,
+      "learning_rate": 0.0001,
+      "loss": 0.1841,
+      "step": 7450
+    },
+    {
+      "epoch": 0.6891057098319958,
+      "grad_norm": 0.6462234854698181,
+      "learning_rate": 0.0001,
+      "loss": 0.1823,
+      "step": 7460
+    },
+    {
+      "epoch": 0.6900294440274811,
+      "grad_norm": 0.45021629333496094,
+      "learning_rate": 0.0001,
+      "loss": 0.1878,
+      "step": 7470
+    },
+    {
+      "epoch": 0.6909531782229663,
+      "grad_norm": 0.5298671722412109,
+      "learning_rate": 0.0001,
+      "loss": 0.1836,
+      "step": 7480
+    },
+    {
+      "epoch": 0.6918769124184516,
+      "grad_norm": 0.544666051864624,
+      "learning_rate": 0.0001,
+      "loss": 0.1843,
+      "step": 7490
+    },
+    {
+      "epoch": 0.6928006466139368,
+      "grad_norm": 0.4883028566837311,
+      "learning_rate": 0.0001,
+      "loss": 0.1831,
+      "step": 7500
+    },
+    {
+      "epoch": 0.6937243808094221,
+      "grad_norm": 0.5272738933563232,
+      "learning_rate": 0.0001,
+      "loss": 0.1887,
+      "step": 7510
+    },
+    {
+      "epoch": 0.6946481150049073,
+      "grad_norm": 0.4605458378791809,
+      "learning_rate": 0.0001,
+      "loss": 0.1834,
+      "step": 7520
+    },
+    {
+      "epoch": 0.6955718492003926,
+      "grad_norm": 0.4206326901912689,
+      "learning_rate": 0.0001,
+      "loss": 0.1803,
+      "step": 7530
+    },
+    {
+      "epoch": 0.6964955833958778,
+      "grad_norm": 0.4763469398021698,
+      "learning_rate": 0.0001,
+      "loss": 0.1839,
+      "step": 7540
+    },
+    {
+      "epoch": 0.6974193175913631,
+      "grad_norm": 0.9533921480178833,
+      "learning_rate": 0.0001,
+      "loss": 0.184,
+      "step": 7550
+    },
+    {
+      "epoch": 0.6983430517868483,
+      "grad_norm": 0.4474931061267853,
+      "learning_rate": 0.0001,
+      "loss": 0.1845,
+      "step": 7560
+    },
+    {
+      "epoch": 0.6992667859823336,
+      "grad_norm": 0.7657787799835205,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 7570
+    },
+    {
+      "epoch": 0.7001905201778188,
+      "grad_norm": 0.4875025451183319,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 7580
+    },
+    {
+      "epoch": 0.7011142543733041,
+      "grad_norm": 0.42080581188201904,
+      "learning_rate": 0.0001,
+      "loss": 0.178,
+      "step": 7590
+    },
+    {
+      "epoch": 0.7020379885687893,
+      "grad_norm": 0.36648714542388916,
+      "learning_rate": 0.0001,
+      "loss": 0.1849,
+      "step": 7600
+    },
+    {
+      "epoch": 0.7029617227642746,
+      "grad_norm": 0.4213377833366394,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 7610
+    },
+    {
+      "epoch": 0.7038854569597598,
+      "grad_norm": 0.35371777415275574,
+      "learning_rate": 0.0001,
+      "loss": 0.185,
+      "step": 7620
+    },
+    {
+      "epoch": 0.7048091911552451,
+      "grad_norm": 0.4143765866756439,
+      "learning_rate": 0.0001,
+      "loss": 0.1833,
+      "step": 7630
+    },
+    {
+      "epoch": 0.7057329253507303,
+      "grad_norm": 0.34513357281684875,
+      "learning_rate": 0.0001,
+      "loss": 0.1816,
+      "step": 7640
+    },
+    {
+      "epoch": 0.7066566595462156,
+      "grad_norm": 0.56122225522995,
+      "learning_rate": 0.0001,
+      "loss": 0.4756,
+      "step": 7650
+    },
+    {
+      "epoch": 0.7075803937417008,
+      "grad_norm": 0.3687942624092102,
+      "learning_rate": 0.0001,
+      "loss": 0.1794,
+      "step": 7660
+    },
+    {
+      "epoch": 0.7085041279371861,
+      "grad_norm": 0.6322677731513977,
+      "learning_rate": 0.0001,
+      "loss": 0.4732,
+      "step": 7670
+    },
+    {
+      "epoch": 0.7094278621326713,
+      "grad_norm": 0.781364381313324,
+      "learning_rate": 0.0001,
+      "loss": 0.1824,
+      "step": 7680
+    },
+    {
+      "epoch": 0.7103515963281566,
+      "grad_norm": 0.7070577144622803,
+      "learning_rate": 0.0001,
+      "loss": 0.1819,
+      "step": 7690
+    },
+    {
+      "epoch": 0.7112753305236418,
+      "grad_norm": 0.6206178069114685,
+      "learning_rate": 0.0001,
+      "loss": 0.1807,
+      "step": 7700
+    },
+    {
+      "epoch": 0.7121990647191271,
+      "grad_norm": 0.6880868077278137,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 7710
+    },
+    {
+      "epoch": 0.7131227989146123,
+      "grad_norm": 0.5743620991706848,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 7720
+    },
+    {
+      "epoch": 0.7140465331100976,
+      "grad_norm": 0.6810530424118042,
+      "learning_rate": 0.0001,
+      "loss": 0.1859,
+      "step": 7730
+    },
+    {
+      "epoch": 0.7149702673055828,
+      "grad_norm": 0.6227778792381287,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 7740
+    },
+    {
+      "epoch": 0.7158940015010681,
+      "grad_norm": 0.84186851978302,
+      "learning_rate": 0.0001,
+      "loss": 0.1827,
+      "step": 7750
+    },
+    {
+      "epoch": 0.7168177356965533,
+      "grad_norm": 0.5849947929382324,
+      "learning_rate": 0.0001,
+      "loss": 0.1816,
+      "step": 7760
+    },
+    {
+      "epoch": 0.7177414698920386,
+      "grad_norm": 0.42177727818489075,
+      "learning_rate": 0.0001,
+      "loss": 0.1792,
+      "step": 7770
+    },
+    {
+      "epoch": 0.7186652040875238,
+      "grad_norm": 0.7109232544898987,
+      "learning_rate": 0.0001,
+      "loss": 0.1795,
+      "step": 7780
+    },
+    {
+      "epoch": 0.7195889382830091,
+      "grad_norm": 0.6489834785461426,
+      "learning_rate": 0.0001,
+      "loss": 0.1768,
+      "step": 7790
+    },
+    {
+      "epoch": 0.7205126724784943,
+      "grad_norm": 0.5706067085266113,
+      "learning_rate": 0.0001,
+      "loss": 0.1795,
+      "step": 7800
+    },
+    {
+      "epoch": 0.7214364066739796,
+      "grad_norm": 0.6116433143615723,
+      "learning_rate": 0.0001,
+      "loss": 0.1796,
+      "step": 7810
+    },
+    {
+      "epoch": 0.7223601408694648,
+      "grad_norm": 0.40196338295936584,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 7820
+    },
+    {
+      "epoch": 0.7232838750649501,
+      "grad_norm": 0.48718318343162537,
+      "learning_rate": 0.0001,
+      "loss": 0.1775,
+      "step": 7830
+    },
+    {
+      "epoch": 0.7242076092604353,
+      "grad_norm": 0.4317290186882019,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 7840
+    },
+    {
+      "epoch": 0.7251313434559206,
+      "grad_norm": 0.4202496111392975,
+      "learning_rate": 0.0001,
+      "loss": 0.1764,
+      "step": 7850
+    },
+    {
+      "epoch": 0.7260550776514058,
+      "grad_norm": 0.6695144772529602,
+      "learning_rate": 0.0001,
+      "loss": 0.4703,
+      "step": 7860
+    },
+    {
+      "epoch": 0.7269788118468911,
+      "grad_norm": 0.596879780292511,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 7870
+    },
+    {
+      "epoch": 0.7279025460423763,
+      "grad_norm": 0.4169802665710449,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 7880
+    },
+    {
+      "epoch": 0.7288262802378616,
+      "grad_norm": 0.3892427980899811,
+      "learning_rate": 0.0001,
+      "loss": 0.1793,
+      "step": 7890
+    },
+    {
+      "epoch": 0.7297500144333467,
+      "grad_norm": 0.4820833206176758,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 7900
+    },
+    {
+      "epoch": 0.730673748628832,
+      "grad_norm": 0.6452758312225342,
+      "learning_rate": 0.0001,
+      "loss": 0.182,
+      "step": 7910
+    },
+    {
+      "epoch": 0.7315974828243174,
+      "grad_norm": 0.4922819137573242,
+      "learning_rate": 0.0001,
+      "loss": 0.179,
+      "step": 7920
+    },
+    {
+      "epoch": 0.7325212170198026,
+      "grad_norm": 0.5007845759391785,
+      "learning_rate": 0.0001,
+      "loss": 0.172,
+      "step": 7930
+    },
+    {
+      "epoch": 0.7334449512152879,
+      "grad_norm": 0.44310009479522705,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 7940
+    },
+    {
+      "epoch": 0.734368685410773,
+      "grad_norm": 0.46110036969184875,
+      "learning_rate": 0.0001,
+      "loss": 0.3969,
+      "step": 7950
+    },
+    {
+      "epoch": 0.7352924196062584,
+      "grad_norm": 0.7349547743797302,
+      "learning_rate": 0.0001,
+      "loss": 0.1795,
+      "step": 7960
+    },
+    {
+      "epoch": 0.7362161538017435,
+      "grad_norm": 0.5728829503059387,
+      "learning_rate": 0.0001,
+      "loss": 0.1774,
+      "step": 7970
+    },
+    {
+      "epoch": 0.7371398879972288,
+      "grad_norm": 0.4830996096134186,
+      "learning_rate": 0.0001,
+      "loss": 0.1798,
+      "step": 7980
+    },
+    {
+      "epoch": 0.738063622192714,
+      "grad_norm": 0.8080053925514221,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 7990
+    },
+    {
+      "epoch": 0.7389873563881993,
+      "grad_norm": 0.9416603446006775,
+      "learning_rate": 0.0001,
+      "loss": 0.1791,
+      "step": 8000
+    },
+    {
+      "epoch": 0.7399110905836845,
+      "grad_norm": 0.602627694606781,
+      "learning_rate": 0.0001,
+      "loss": 0.175,
+      "step": 8010
+    },
+    {
+      "epoch": 0.7408348247791698,
+      "grad_norm": 0.8102542757987976,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 8020
+    },
+    {
+      "epoch": 0.741758558974655,
+      "grad_norm": 0.5637885928153992,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 8030
+    },
+    {
+      "epoch": 0.7426822931701403,
+      "grad_norm": 0.34650176763534546,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 8040
+    },
+    {
+      "epoch": 0.7436060273656255,
+      "grad_norm": 0.4475821554660797,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 8050
+    },
+    {
+      "epoch": 0.7445297615611108,
+      "grad_norm": 0.4095354378223419,
+      "learning_rate": 0.0001,
+      "loss": 0.1785,
+      "step": 8060
+    },
+    {
+      "epoch": 0.745453495756596,
+      "grad_norm": 0.4277932941913605,
+      "learning_rate": 0.0001,
+      "loss": 0.1769,
+      "step": 8070
+    },
+    {
+      "epoch": 0.7463772299520813,
+      "grad_norm": 0.3410409986972809,
+      "learning_rate": 0.0001,
+      "loss": 0.4647,
+      "step": 8080
+    },
+    {
+      "epoch": 0.7473009641475665,
+      "grad_norm": 0.3772905468940735,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 8090
+    },
+    {
+      "epoch": 0.7482246983430518,
+      "grad_norm": 0.43052271008491516,
+      "learning_rate": 0.0001,
+      "loss": 0.1762,
+      "step": 8100
+    },
+    {
+      "epoch": 0.749148432538537,
+      "grad_norm": 0.44517043232917786,
+      "learning_rate": 0.0001,
+      "loss": 0.1768,
+      "step": 8110
+    },
+    {
+      "epoch": 0.7500721667340223,
+      "grad_norm": 0.3778810501098633,
+      "learning_rate": 0.0001,
+      "loss": 0.1768,
+      "step": 8120
+    },
+    {
+      "epoch": 0.7509959009295075,
+      "grad_norm": 0.7292577624320984,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 8130
+    },
+    {
+      "epoch": 0.7519196351249928,
+      "grad_norm": 0.5761885046958923,
+      "learning_rate": 0.0001,
+      "loss": 0.1768,
+      "step": 8140
+    },
+    {
+      "epoch": 0.752843369320478,
+      "grad_norm": 0.5460628271102905,
+      "learning_rate": 0.0001,
+      "loss": 0.1789,
+      "step": 8150
+    },
+    {
+      "epoch": 0.7537671035159633,
+      "grad_norm": 0.49501052498817444,
+      "learning_rate": 0.0001,
+      "loss": 0.1783,
+      "step": 8160
+    },
+    {
+      "epoch": 0.7546908377114485,
+      "grad_norm": 0.4425870180130005,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 8170
+    },
+    {
+      "epoch": 0.7556145719069338,
+      "grad_norm": 0.39895811676979065,
+      "learning_rate": 0.0001,
+      "loss": 0.1714,
+      "step": 8180
+    },
+    {
+      "epoch": 0.756538306102419,
+      "grad_norm": 0.4607703685760498,
+      "learning_rate": 0.0001,
+      "loss": 0.1806,
+      "step": 8190
+    },
+    {
+      "epoch": 0.7574620402979043,
+      "grad_norm": 0.38244667649269104,
+      "learning_rate": 0.0001,
+      "loss": 0.1768,
+      "step": 8200
+    },
+    {
+      "epoch": 0.7583857744933895,
+      "grad_norm": 0.5269780158996582,
+      "learning_rate": 0.0001,
+      "loss": 0.1755,
+      "step": 8210
+    },
+    {
+      "epoch": 0.7593095086888748,
+      "grad_norm": 0.42601966857910156,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 8220
+    },
+    {
+      "epoch": 0.76023324288436,
+      "grad_norm": 0.5692285895347595,
+      "learning_rate": 0.0001,
+      "loss": 0.1751,
+      "step": 8230
+    },
+    {
+      "epoch": 0.7611569770798453,
+      "grad_norm": 0.862949550151825,
+      "learning_rate": 0.0001,
+      "loss": 0.1752,
+      "step": 8240
+    },
+    {
+      "epoch": 0.7620807112753305,
+      "grad_norm": 0.5871086716651917,
+      "learning_rate": 0.0001,
+      "loss": 0.1736,
+      "step": 8250
+    },
+    {
+      "epoch": 0.7630044454708158,
+      "grad_norm": 0.36870133876800537,
+      "learning_rate": 0.0001,
+      "loss": 0.1758,
+      "step": 8260
+    },
+    {
+      "epoch": 0.763928179666301,
+      "grad_norm": 0.5182648301124573,
+      "learning_rate": 0.0001,
+      "loss": 0.1712,
+      "step": 8270
+    },
+    {
+      "epoch": 0.7648519138617863,
+      "grad_norm": 0.44941583275794983,
+      "learning_rate": 0.0001,
+      "loss": 0.1761,
+      "step": 8280
+    },
+    {
+      "epoch": 0.7657756480572715,
+      "grad_norm": 0.4947168529033661,
+      "learning_rate": 0.0001,
+      "loss": 0.1687,
+      "step": 8290
+    },
+    {
+      "epoch": 0.7666993822527568,
+      "grad_norm": 0.4370608627796173,
+      "learning_rate": 0.0001,
+      "loss": 0.1757,
+      "step": 8300
+    },
+    {
+      "epoch": 0.767623116448242,
+      "grad_norm": 0.4258075952529907,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 8310
+    },
+    {
+      "epoch": 0.7685468506437273,
+      "grad_norm": 0.4772101640701294,
+      "learning_rate": 0.0001,
+      "loss": 0.177,
+      "step": 8320
+    },
+    {
+      "epoch": 0.7694705848392125,
+      "grad_norm": 0.45485544204711914,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 8330
+    },
+    {
+      "epoch": 0.7703943190346978,
+      "grad_norm": 0.6943185925483704,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 8340
+    },
+    {
+      "epoch": 0.771318053230183,
+      "grad_norm": 0.8926262855529785,
+      "learning_rate": 0.0001,
+      "loss": 0.1749,
+      "step": 8350
+    },
+    {
+      "epoch": 0.7722417874256683,
+      "grad_norm": 0.7720366716384888,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 8360
+    },
+    {
+      "epoch": 0.7731655216211535,
+      "grad_norm": 0.6796042919158936,
+      "learning_rate": 0.0001,
+      "loss": 0.1743,
+      "step": 8370
+    },
+    {
+      "epoch": 0.7740892558166388,
+      "grad_norm": 0.40225762128829956,
+      "learning_rate": 0.0001,
+      "loss": 0.1728,
+      "step": 8380
+    },
+    {
+      "epoch": 0.775012990012124,
+      "grad_norm": 0.5170344710350037,
+      "learning_rate": 0.0001,
+      "loss": 0.1744,
+      "step": 8390
+    },
+    {
+      "epoch": 0.7759367242076093,
+      "grad_norm": 0.43774446845054626,
+      "learning_rate": 0.0001,
+      "loss": 0.171,
+      "step": 8400
+    },
+    {
+      "epoch": 0.7768604584030945,
+      "grad_norm": 0.35116827487945557,
+      "learning_rate": 0.0001,
+      "loss": 0.1722,
+      "step": 8410
+    },
+    {
+      "epoch": 0.7777841925985798,
+      "grad_norm": 0.42142215371131897,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 8420
+    },
+    {
+      "epoch": 0.778707926794065,
+      "grad_norm": 0.45540010929107666,
+      "learning_rate": 0.0001,
+      "loss": 0.1708,
+      "step": 8430
+    },
+    {
+      "epoch": 0.7796316609895503,
+      "grad_norm": 0.5589233040809631,
+      "learning_rate": 0.0001,
+      "loss": 0.1746,
+      "step": 8440
+    },
+    {
+      "epoch": 0.7805553951850355,
+      "grad_norm": 0.3629968464374542,
+      "learning_rate": 0.0001,
+      "loss": 0.1748,
+      "step": 8450
+    },
+    {
+      "epoch": 0.7814791293805208,
+      "grad_norm": 0.806050717830658,
+      "learning_rate": 0.0001,
+      "loss": 0.173,
+      "step": 8460
+    },
+    {
+      "epoch": 0.782402863576006,
+      "grad_norm": 0.7409831285476685,
+      "learning_rate": 0.0001,
+      "loss": 0.1771,
+      "step": 8470
+    },
+    {
+      "epoch": 0.7833265977714913,
+      "grad_norm": 0.4119178354740143,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 8480
+    },
+    {
+      "epoch": 0.7842503319669765,
+      "grad_norm": 0.4585535526275635,
+      "learning_rate": 0.0001,
+      "loss": 0.1732,
+      "step": 8490
+    },
+    {
+      "epoch": 0.7851740661624618,
+      "grad_norm": 0.4914413094520569,
+      "learning_rate": 0.0001,
+      "loss": 0.1742,
+      "step": 8500
+    },
+    {
+      "epoch": 0.786097800357947,
+      "grad_norm": 0.451362282037735,
+      "learning_rate": 0.0001,
+      "loss": 0.1713,
+      "step": 8510
+    },
+    {
+      "epoch": 0.7870215345534323,
+      "grad_norm": 0.39437827467918396,
+      "learning_rate": 0.0001,
+      "loss": 0.1733,
+      "step": 8520
+    },
+    {
+      "epoch": 0.7879452687489175,
+      "grad_norm": 0.6410636305809021,
+      "learning_rate": 0.0001,
+      "loss": 0.1737,
+      "step": 8530
+    },
+    {
+      "epoch": 0.7888690029444028,
+      "grad_norm": 0.8889188170433044,
+      "learning_rate": 0.0001,
+      "loss": 0.1751,
+      "step": 8540
+    },
+    {
+      "epoch": 0.789792737139888,
+      "grad_norm": 0.5608518123626709,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 8550
+    },
+    {
+      "epoch": 0.7907164713353733,
+      "grad_norm": 0.4477889835834503,
+      "learning_rate": 0.0001,
+      "loss": 0.1718,
+      "step": 8560
+    },
+    {
+      "epoch": 0.7916402055308585,
+      "grad_norm": 0.543216347694397,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 8570
+    },
+    {
+      "epoch": 0.7925639397263438,
+      "grad_norm": 0.47616589069366455,
+      "learning_rate": 0.0001,
+      "loss": 0.1747,
+      "step": 8580
+    },
+    {
+      "epoch": 0.793487673921829,
+      "grad_norm": 0.5494805574417114,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 8590
+    },
+    {
+      "epoch": 0.7944114081173143,
+      "grad_norm": 0.3553082346916199,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 8600
+    },
+    {
+      "epoch": 0.7953351423127994,
+      "grad_norm": 0.9502201676368713,
+      "learning_rate": 0.0001,
+      "loss": 0.1763,
+      "step": 8610
+    },
+    {
+      "epoch": 0.7962588765082848,
+      "grad_norm": 0.47275465726852417,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 8620
+    },
+    {
+      "epoch": 0.79718261070377,
+      "grad_norm": 0.8205789923667908,
+      "learning_rate": 0.0001,
+      "loss": 0.1742,
+      "step": 8630
+    },
+    {
+      "epoch": 0.7981063448992552,
+      "grad_norm": 0.5286776423454285,
+      "learning_rate": 0.0001,
+      "loss": 0.1734,
+      "step": 8640
+    },
+    {
+      "epoch": 0.7990300790947404,
+      "grad_norm": 0.4189944565296173,
+      "learning_rate": 0.0001,
+      "loss": 0.1689,
+      "step": 8650
+    },
+    {
+      "epoch": 0.7999538132902257,
+      "grad_norm": 0.4479851722717285,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 8660
+    },
+    {
+      "epoch": 0.8008775474857109,
+      "grad_norm": 0.739837646484375,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 8670
+    },
+    {
+      "epoch": 0.8018012816811962,
+      "grad_norm": 0.34144723415374756,
+      "learning_rate": 0.0001,
+      "loss": 0.1717,
+      "step": 8680
+    },
+    {
+      "epoch": 0.8027250158766814,
+      "grad_norm": 0.556404173374176,
+      "learning_rate": 0.0001,
+      "loss": 0.1721,
+      "step": 8690
+    },
+    {
+      "epoch": 0.8036487500721667,
+      "grad_norm": 0.6465960144996643,
+      "learning_rate": 0.0001,
+      "loss": 0.1698,
+      "step": 8700
+    },
+    {
+      "epoch": 0.8045724842676519,
+      "grad_norm": 0.4784127473831177,
+      "learning_rate": 0.0001,
+      "loss": 0.1726,
+      "step": 8710
+    },
+    {
+      "epoch": 0.8054962184631372,
+      "grad_norm": 0.446210652589798,
+      "learning_rate": 0.0001,
+      "loss": 0.1709,
+      "step": 8720
+    },
+    {
+      "epoch": 0.8064199526586224,
+      "grad_norm": 0.6317527294158936,
+      "learning_rate": 0.0001,
+      "loss": 0.17,
+      "step": 8730
+    },
+    {
+      "epoch": 0.8073436868541077,
+      "grad_norm": 0.9491158127784729,
+      "learning_rate": 0.0001,
+      "loss": 0.1738,
+      "step": 8740
+    },
+    {
+      "epoch": 0.8082674210495929,
+      "grad_norm": 0.485317587852478,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 8750
+    },
+    {
+      "epoch": 0.8091911552450782,
+      "grad_norm": 0.4314129054546356,
+      "learning_rate": 0.0001,
+      "loss": 0.1694,
+      "step": 8760
+    },
+    {
+      "epoch": 0.8101148894405634,
+      "grad_norm": 0.43169793486595154,
+      "learning_rate": 0.0001,
+      "loss": 0.4557,
+      "step": 8770
+    },
+    {
+      "epoch": 0.8110386236360487,
+      "grad_norm": 0.7022022008895874,
+      "learning_rate": 0.0001,
+      "loss": 0.1692,
+      "step": 8780
+    },
+    {
+      "epoch": 0.8119623578315339,
+      "grad_norm": 0.8071302771568298,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 8790
+    },
+    {
+      "epoch": 0.8128860920270192,
+      "grad_norm": 0.4125478267669678,
+      "learning_rate": 0.0001,
+      "loss": 0.1695,
+      "step": 8800
+    },
+    {
+      "epoch": 0.8138098262225045,
+      "grad_norm": 0.669414222240448,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 8810
+    },
+    {
+      "epoch": 0.8147335604179897,
+      "grad_norm": 0.5178114175796509,
+      "learning_rate": 0.0001,
+      "loss": 0.1707,
+      "step": 8820
+    },
+    {
+      "epoch": 0.815657294613475,
+      "grad_norm": 0.6118435263633728,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 8830
+    },
+    {
+      "epoch": 0.8165810288089602,
+      "grad_norm": 0.5061274766921997,
+      "learning_rate": 0.0001,
+      "loss": 0.168,
+      "step": 8840
+    },
+    {
+      "epoch": 0.8175047630044455,
+      "grad_norm": 0.44611838459968567,
+      "learning_rate": 0.0001,
+      "loss": 0.4568,
+      "step": 8850
+    },
+    {
+      "epoch": 0.8184284971999307,
+      "grad_norm": 0.5419828295707703,
+      "learning_rate": 0.0001,
+      "loss": 0.1687,
+      "step": 8860
+    },
+    {
+      "epoch": 0.819352231395416,
+      "grad_norm": 0.47415947914123535,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 8870
+    },
+    {
+      "epoch": 0.8202759655909012,
+      "grad_norm": 0.3742285370826721,
+      "learning_rate": 0.0001,
+      "loss": 0.4504,
+      "step": 8880
+    },
+    {
+      "epoch": 0.8211996997863865,
+      "grad_norm": 0.4165765941143036,
+      "learning_rate": 0.0001,
+      "loss": 0.1691,
+      "step": 8890
+    },
+    {
+      "epoch": 0.8221234339818717,
+      "grad_norm": 0.4976545572280884,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 8900
+    },
+    {
+      "epoch": 0.823047168177357,
+      "grad_norm": 0.3930950164794922,
+      "learning_rate": 0.0001,
+      "loss": 0.1669,
+      "step": 8910
+    },
+    {
+      "epoch": 0.8239709023728422,
+      "grad_norm": 0.7222493290901184,
+      "learning_rate": 0.0001,
+      "loss": 0.1678,
+      "step": 8920
+    },
+    {
+      "epoch": 0.8248946365683275,
+      "grad_norm": 0.6513727903366089,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 8930
+    },
+    {
+      "epoch": 0.8258183707638127,
+      "grad_norm": 0.36553195118904114,
+      "learning_rate": 0.0001,
+      "loss": 0.1685,
+      "step": 8940
+    },
+    {
+      "epoch": 0.826742104959298,
+      "grad_norm": 0.4141073226928711,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 8950
+    },
+    {
+      "epoch": 0.8276658391547832,
+      "grad_norm": 0.9026088714599609,
+      "learning_rate": 0.0001,
+      "loss": 0.1655,
+      "step": 8960
+    },
+    {
+      "epoch": 0.8285895733502685,
+      "grad_norm": 0.4109916687011719,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 8970
+    },
+    {
+      "epoch": 0.8295133075457537,
+      "grad_norm": 0.4617246389389038,
+      "learning_rate": 0.0001,
+      "loss": 0.164,
+      "step": 8980
+    },
+    {
+      "epoch": 0.830437041741239,
+      "grad_norm": 0.6527224779129028,
+      "learning_rate": 0.0001,
+      "loss": 0.1638,
+      "step": 8990
+    },
+    {
+      "epoch": 0.8313607759367242,
+      "grad_norm": 1.0628283023834229,
+      "learning_rate": 0.0001,
+      "loss": 0.1663,
+      "step": 9000
+    },
+    {
+      "epoch": 0.8322845101322095,
+      "grad_norm": 0.4224163293838501,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 9010
+    },
+    {
+      "epoch": 0.8332082443276947,
+      "grad_norm": 0.5163058638572693,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 9020
+    },
+    {
+      "epoch": 0.83413197852318,
+      "grad_norm": 0.5277922749519348,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 9030
+    },
+    {
+      "epoch": 0.8350557127186652,
+      "grad_norm": 1.1311513185501099,
+      "learning_rate": 0.0001,
+      "loss": 0.1668,
+      "step": 9040
+    },
+    {
+      "epoch": 0.8359794469141505,
+      "grad_norm": 0.7585214376449585,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 9050
+    },
+    {
+      "epoch": 0.8369031811096357,
+      "grad_norm": 0.8250924944877625,
+      "learning_rate": 0.0001,
+      "loss": 0.1631,
+      "step": 9060
+    },
+    {
+      "epoch": 0.837826915305121,
+      "grad_norm": 0.664091944694519,
+      "learning_rate": 0.0001,
+      "loss": 0.1643,
+      "step": 9070
+    },
+    {
+      "epoch": 0.8387506495006062,
+      "grad_norm": 0.5413016676902771,
+      "learning_rate": 0.0001,
+      "loss": 0.4144,
+      "step": 9080
+    },
+    {
+      "epoch": 0.8396743836960915,
+      "grad_norm": 0.533757746219635,
+      "learning_rate": 0.0001,
+      "loss": 0.1679,
+      "step": 9090
+    },
+    {
+      "epoch": 0.8405981178915767,
+      "grad_norm": 0.5132860541343689,
+      "learning_rate": 0.0001,
+      "loss": 0.1662,
+      "step": 9100
+    },
+    {
+      "epoch": 0.841521852087062,
+      "grad_norm": 1.0943279266357422,
+      "learning_rate": 0.0001,
+      "loss": 0.1674,
+      "step": 9110
+    },
+    {
+      "epoch": 0.8424455862825472,
+      "grad_norm": 0.9697147011756897,
+      "learning_rate": 0.0001,
+      "loss": 0.1701,
+      "step": 9120
+    },
+    {
+      "epoch": 0.8433693204780325,
+      "grad_norm": 0.4370110332965851,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 9130
+    },
+    {
+      "epoch": 0.8442930546735177,
+      "grad_norm": 0.49583470821380615,
+      "learning_rate": 0.0001,
+      "loss": 0.1675,
+      "step": 9140
+    },
+    {
+      "epoch": 0.845216788869003,
+      "grad_norm": 0.3518935441970825,
+      "learning_rate": 0.0001,
+      "loss": 0.1649,
+      "step": 9150
+    },
+    {
+      "epoch": 0.8461405230644882,
+      "grad_norm": 0.4974721670150757,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 9160
+    },
+    {
+      "epoch": 0.8470642572599735,
+      "grad_norm": 0.5408349633216858,
+      "learning_rate": 0.0001,
+      "loss": 0.1644,
+      "step": 9170
+    },
+    {
+      "epoch": 0.8479879914554587,
+      "grad_norm": 0.4251846373081207,
+      "learning_rate": 0.0001,
+      "loss": 0.1627,
+      "step": 9180
+    },
+    {
+      "epoch": 0.848911725650944,
+      "grad_norm": 0.47709429264068604,
+      "learning_rate": 0.0001,
+      "loss": 0.1672,
+      "step": 9190
+    },
+    {
+      "epoch": 0.8498354598464292,
+      "grad_norm": 0.45655733346939087,
+      "learning_rate": 0.0001,
+      "loss": 0.1664,
+      "step": 9200
+    },
+    {
+      "epoch": 0.8507591940419145,
+      "grad_norm": 0.5739580988883972,
+      "learning_rate": 0.0001,
+      "loss": 0.1628,
+      "step": 9210
+    },
+    {
+      "epoch": 0.8516829282373997,
+      "grad_norm": 0.41699105501174927,
+      "learning_rate": 0.0001,
+      "loss": 0.1659,
+      "step": 9220
+    },
+    {
+      "epoch": 0.852606662432885,
+      "grad_norm": 0.5231941938400269,
+      "learning_rate": 0.0001,
+      "loss": 0.1667,
+      "step": 9230
+    },
+    {
+      "epoch": 0.8535303966283702,
+      "grad_norm": 0.3867189884185791,
+      "learning_rate": 0.0001,
+      "loss": 0.1656,
+      "step": 9240
+    },
+    {
+      "epoch": 0.8544541308238555,
+      "grad_norm": 0.42265090346336365,
+      "learning_rate": 0.0001,
+      "loss": 0.1621,
+      "step": 9250
+    },
+    {
+      "epoch": 0.8553778650193407,
+      "grad_norm": 0.7266954779624939,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 9260
+    },
+    {
+      "epoch": 0.856301599214826,
+      "grad_norm": 0.4829242527484894,
+      "learning_rate": 0.0001,
+      "loss": 0.1705,
+      "step": 9270
+    },
+    {
+      "epoch": 0.8572253334103112,
+      "grad_norm": 0.4225987493991852,
+      "learning_rate": 0.0001,
+      "loss": 0.1716,
+      "step": 9280
+    },
+    {
+      "epoch": 0.8581490676057965,
+      "grad_norm": 0.7197133898735046,
+      "learning_rate": 0.0001,
+      "loss": 0.1674,
+      "step": 9290
+    },
+    {
+      "epoch": 0.8590728018012816,
+      "grad_norm": 0.5460970997810364,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 9300
+    },
+    {
+      "epoch": 0.859996535996767,
+      "grad_norm": 0.3880206048488617,
+      "learning_rate": 0.0001,
+      "loss": 0.1651,
+      "step": 9310
+    },
+    {
+      "epoch": 0.8609202701922521,
+      "grad_norm": 0.5950866937637329,
+      "learning_rate": 0.0001,
+      "loss": 0.163,
+      "step": 9320
+    },
+    {
+      "epoch": 0.8618440043877375,
+      "grad_norm": 0.4253402352333069,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 9330
+    },
+    {
+      "epoch": 0.8627677385832226,
+      "grad_norm": 0.3970343768596649,
+      "learning_rate": 0.0001,
+      "loss": 0.1647,
+      "step": 9340
+    },
+    {
+      "epoch": 0.863691472778708,
+      "grad_norm": 0.43371790647506714,
+      "learning_rate": 0.0001,
+      "loss": 0.1671,
+      "step": 9350
+    },
+    {
+      "epoch": 0.8646152069741931,
+      "grad_norm": 0.634937047958374,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 9360
+    },
+    {
+      "epoch": 0.8655389411696784,
+      "grad_norm": 0.7563658356666565,
+      "learning_rate": 0.0001,
+      "loss": 0.1625,
+      "step": 9370
+    },
+    {
+      "epoch": 0.8664626753651636,
+      "grad_norm": 0.498696506023407,
+      "learning_rate": 0.0001,
+      "loss": 0.4461,
+      "step": 9380
+    },
+    {
+      "epoch": 0.8673864095606489,
+      "grad_norm": 0.39479953050613403,
+      "learning_rate": 0.0001,
+      "loss": 0.1595,
+      "step": 9390
+    },
+    {
+      "epoch": 0.8683101437561341,
+      "grad_norm": 0.5282758474349976,
+      "learning_rate": 0.0001,
+      "loss": 0.165,
+      "step": 9400
+    },
+    {
+      "epoch": 0.8692338779516194,
+      "grad_norm": 0.502001941204071,
+      "learning_rate": 0.0001,
+      "loss": 0.161,
+      "step": 9410
+    },
+    {
+      "epoch": 0.8701576121471046,
+      "grad_norm": 0.44498735666275024,
+      "learning_rate": 0.0001,
+      "loss": 0.166,
+      "step": 9420
+    },
+    {
+      "epoch": 0.8710813463425899,
+      "grad_norm": 0.4128197729587555,
+      "learning_rate": 0.0001,
+      "loss": 0.4407,
+      "step": 9430
+    },
+    {
+      "epoch": 0.8720050805380751,
+      "grad_norm": 0.4395778775215149,
+      "learning_rate": 0.0001,
+      "loss": 0.1637,
+      "step": 9440
+    },
+    {
+      "epoch": 0.8729288147335604,
+      "grad_norm": 0.42124509811401367,
+      "learning_rate": 0.0001,
+      "loss": 0.1619,
+      "step": 9450
+    },
+    {
+      "epoch": 0.8738525489290456,
+      "grad_norm": 0.6406058073043823,
+      "learning_rate": 0.0001,
+      "loss": 0.1609,
+      "step": 9460
+    },
+    {
+      "epoch": 0.8747762831245309,
+      "grad_norm": 0.48441171646118164,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 9470
+    },
+    {
+      "epoch": 0.8757000173200161,
+      "grad_norm": 0.41274169087409973,
+      "learning_rate": 0.0001,
+      "loss": 0.1599,
+      "step": 9480
+    },
+    {
+      "epoch": 0.8766237515155014,
+      "grad_norm": 0.7884708046913147,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 9490
+    },
+    {
+      "epoch": 0.8775474857109866,
+      "grad_norm": 0.3887585997581482,
+      "learning_rate": 0.0001,
+      "loss": 0.1597,
+      "step": 9500
+    },
+    {
+      "epoch": 0.8784712199064719,
+      "grad_norm": 0.5127732157707214,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 9510
+    },
+    {
+      "epoch": 0.8793949541019571,
+      "grad_norm": 0.6500501036643982,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 9520
+    },
+    {
+      "epoch": 0.8803186882974424,
+      "grad_norm": 0.9996128082275391,
+      "learning_rate": 0.0001,
+      "loss": 0.1616,
+      "step": 9530
+    },
+    {
+      "epoch": 0.8812424224929276,
+      "grad_norm": 0.6707596182823181,
+      "learning_rate": 0.0001,
+      "loss": 0.1639,
+      "step": 9540
+    },
+    {
+      "epoch": 0.8821661566884129,
+      "grad_norm": 0.47952958941459656,
+      "learning_rate": 0.0001,
+      "loss": 0.1551,
+      "step": 9550
+    },
+    {
+      "epoch": 0.8830898908838981,
+      "grad_norm": 0.5806353688240051,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 9560
+    },
+    {
+      "epoch": 0.8840136250793834,
+      "grad_norm": 0.5127092003822327,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 9570
+    },
+    {
+      "epoch": 0.8849373592748686,
+      "grad_norm": 0.7458198070526123,
+      "learning_rate": 0.0001,
+      "loss": 0.1598,
+      "step": 9580
+    },
+    {
+      "epoch": 0.8858610934703539,
+      "grad_norm": 0.4138084650039673,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 9590
+    },
+    {
+      "epoch": 0.8867848276658391,
+      "grad_norm": 0.6283295154571533,
+      "learning_rate": 0.0001,
+      "loss": 0.1607,
+      "step": 9600
+    },
+    {
+      "epoch": 0.8877085618613244,
+      "grad_norm": 0.8556647300720215,
+      "learning_rate": 0.0001,
+      "loss": 0.1603,
+      "step": 9610
+    },
+    {
+      "epoch": 0.8886322960568096,
+      "grad_norm": 0.962799608707428,
+      "learning_rate": 0.0001,
+      "loss": 0.1596,
+      "step": 9620
+    },
+    {
+      "epoch": 0.8895560302522949,
+      "grad_norm": 0.4691963493824005,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 9630
+    },
+    {
+      "epoch": 0.8904797644477801,
+      "grad_norm": 0.5351691246032715,
+      "learning_rate": 0.0001,
+      "loss": 0.1594,
+      "step": 9640
+    },
+    {
+      "epoch": 0.8914034986432654,
+      "grad_norm": 0.43255048990249634,
+      "learning_rate": 0.0001,
+      "loss": 0.1617,
+      "step": 9650
+    },
+    {
+      "epoch": 0.8923272328387506,
+      "grad_norm": 0.4275433123111725,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 9660
+    },
+    {
+      "epoch": 0.8932509670342359,
+      "grad_norm": 0.41553664207458496,
+      "learning_rate": 0.0001,
+      "loss": 0.1548,
+      "step": 9670
+    },
+    {
+      "epoch": 0.8941747012297212,
+      "grad_norm": 0.5009292960166931,
+      "learning_rate": 0.0001,
+      "loss": 0.1587,
+      "step": 9680
+    },
+    {
+      "epoch": 0.8950984354252064,
+      "grad_norm": 0.47097280621528625,
+      "learning_rate": 0.0001,
+      "loss": 0.1593,
+      "step": 9690
+    },
+    {
+      "epoch": 0.8960221696206917,
+      "grad_norm": 0.4848572313785553,
+      "learning_rate": 0.0001,
+      "loss": 0.16,
+      "step": 9700
+    },
+    {
+      "epoch": 0.8969459038161769,
+      "grad_norm": 0.41388949751853943,
+      "learning_rate": 0.0001,
+      "loss": 0.1569,
+      "step": 9710
+    },
+    {
+      "epoch": 0.8978696380116622,
+      "grad_norm": 1.0132074356079102,
+      "learning_rate": 0.0001,
+      "loss": 0.1601,
+      "step": 9720
+    },
+    {
+      "epoch": 0.8987933722071474,
+      "grad_norm": 0.5202128887176514,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 9730
+    },
+    {
+      "epoch": 0.8997171064026327,
+      "grad_norm": 0.5012717843055725,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 9740
+    },
+    {
+      "epoch": 0.9006408405981179,
+      "grad_norm": 0.6582213640213013,
+      "learning_rate": 0.0001,
+      "loss": 0.162,
+      "step": 9750
+    },
+    {
+      "epoch": 0.9015645747936032,
+      "grad_norm": 0.5900190472602844,
+      "learning_rate": 0.0001,
+      "loss": 0.1582,
+      "step": 9760
+    },
+    {
+      "epoch": 0.9024883089890884,
+      "grad_norm": 0.8734009861946106,
+      "learning_rate": 0.0001,
+      "loss": 0.1577,
+      "step": 9770
+    },
+    {
+      "epoch": 0.9034120431845737,
+      "grad_norm": 0.5617889165878296,
+      "learning_rate": 0.0001,
+      "loss": 0.1566,
+      "step": 9780
+    },
+    {
+      "epoch": 0.9043357773800589,
+      "grad_norm": 0.4171944260597229,
+      "learning_rate": 0.0001,
+      "loss": 0.1611,
+      "step": 9790
+    },
+    {
+      "epoch": 0.9052595115755442,
+      "grad_norm": 0.6544126272201538,
+      "learning_rate": 0.0001,
+      "loss": 0.1562,
+      "step": 9800
+    },
+    {
+      "epoch": 0.9061832457710294,
+      "grad_norm": 0.48378440737724304,
+      "learning_rate": 0.0001,
+      "loss": 0.1586,
+      "step": 9810
+    },
+    {
+      "epoch": 0.9071069799665147,
+      "grad_norm": 0.7782066464424133,
+      "learning_rate": 0.0001,
+      "loss": 0.4362,
+      "step": 9820
+    },
+    {
+      "epoch": 0.9080307141619999,
+      "grad_norm": 0.4387872517108917,
+      "learning_rate": 0.0001,
+      "loss": 0.1584,
+      "step": 9830
+    },
+    {
+      "epoch": 0.9089544483574852,
+      "grad_norm": 0.5127896070480347,
+      "learning_rate": 0.0001,
+      "loss": 0.1554,
+      "step": 9840
+    },
+    {
+      "epoch": 0.9098781825529704,
+      "grad_norm": 1.0956830978393555,
+      "learning_rate": 0.0001,
+      "loss": 0.1641,
+      "step": 9850
+    },
+    {
+      "epoch": 0.9108019167484557,
+      "grad_norm": 0.4374365210533142,
+      "learning_rate": 0.0001,
+      "loss": 0.1602,
+      "step": 9860
+    },
+    {
+      "epoch": 0.9117256509439409,
+      "grad_norm": 0.3823205828666687,
+      "learning_rate": 0.0001,
+      "loss": 0.1583,
+      "step": 9870
+    },
+    {
+      "epoch": 0.9126493851394262,
+      "grad_norm": 0.47566449642181396,
+      "learning_rate": 0.0001,
+      "loss": 0.1606,
+      "step": 9880
+    },
+    {
+      "epoch": 0.9135731193349114,
+      "grad_norm": 0.5634418725967407,
+      "learning_rate": 0.0001,
+      "loss": 0.1555,
+      "step": 9890
+    },
+    {
+      "epoch": 0.9144968535303967,
+      "grad_norm": 0.543655276298523,
+      "learning_rate": 0.0001,
+      "loss": 0.1574,
+      "step": 9900
+    },
+    {
+      "epoch": 0.9154205877258819,
+      "grad_norm": 0.7198746204376221,
+      "learning_rate": 0.0001,
+      "loss": 0.1589,
+      "step": 9910
+    },
+    {
+      "epoch": 0.9163443219213672,
+      "grad_norm": 0.4656532108783722,
+      "learning_rate": 0.0001,
+      "loss": 0.1588,
+      "step": 9920
+    },
+    {
+      "epoch": 0.9172680561168524,
+      "grad_norm": 0.4030615985393524,
+      "learning_rate": 0.0001,
+      "loss": 0.1599,
+      "step": 9930
+    },
+    {
+      "epoch": 0.9181917903123377,
+      "grad_norm": 0.5512037873268127,
+      "learning_rate": 0.0001,
+      "loss": 0.1579,
+      "step": 9940
+    },
+    {
+      "epoch": 0.9191155245078229,
+      "grad_norm": 1.127402901649475,
+      "learning_rate": 0.0001,
+      "loss": 0.1591,
+      "step": 9950
+    },
+    {
+      "epoch": 0.9200392587033082,
+      "grad_norm": 1.4228541851043701,
+      "learning_rate": 0.0001,
+      "loss": 0.1572,
+      "step": 9960
+    },
+    {
+      "epoch": 0.9209629928987934,
+      "grad_norm": 1.1439874172210693,
+      "learning_rate": 0.0001,
+      "loss": 0.5687,
+      "step": 9970
+    },
+    {
+      "epoch": 0.9218867270942787,
+      "grad_norm": 1.3781626224517822,
+      "learning_rate": 0.0001,
+      "loss": 0.1619,
+      "step": 9980
+    },
+    {
+      "epoch": 0.9228104612897639,
+      "grad_norm": 0.9378222823143005,
+      "learning_rate": 0.0001,
+      "loss": 0.1613,
+      "step": 9990
+    },
+    {
+      "epoch": 0.9237341954852492,
+      "grad_norm": 0.42424276471138,
+      "learning_rate": 0.0001,
+      "loss": 0.1576,
+      "step": 10000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 100000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 10,
+  "save_steps": 10000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 0.0,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}