diff --git "a/checkpoint-6500/trainer_state.json" "b/checkpoint-6500/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/checkpoint-6500/trainer_state.json"
@@ -0,0 +1,4584 @@
+{
+  "best_global_step": null,
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0051031256644696,
+  "eval_steps": 500,
+  "global_step": 6500,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0015464017165059054,
+      "grad_norm": 4352.0,
+      "learning_rate": 1.9972307692307693e-05,
+      "loss": 10.9174,
+      "step": 10
+    },
+    {
+      "epoch": 0.0030928034330118107,
+      "grad_norm": 71168.0,
+      "learning_rate": 1.9941538461538464e-05,
+      "loss": 11.9649,
+      "step": 20
+    },
+    {
+      "epoch": 0.004639205149517716,
+      "grad_norm": 190.0,
+      "learning_rate": 1.9910769230769232e-05,
+      "loss": 5.27,
+      "step": 30
+    },
+    {
+      "epoch": 0.0061856068660236215,
+      "grad_norm": 16.125,
+      "learning_rate": 1.9880000000000003e-05,
+      "loss": 0.3647,
+      "step": 40
+    },
+    {
+      "epoch": 0.007732008582529527,
+      "grad_norm": 3.890625,
+      "learning_rate": 1.984923076923077e-05,
+      "loss": 0.3099,
+      "step": 50
+    },
+    {
+      "epoch": 0.009278410299035433,
+      "grad_norm": 1.8828125,
+      "learning_rate": 1.9818461538461538e-05,
+      "loss": 0.2842,
+      "step": 60
+    },
+    {
+      "epoch": 0.010824812015541337,
+      "grad_norm": 1.6953125,
+      "learning_rate": 1.978769230769231e-05,
+      "loss": 0.2943,
+      "step": 70
+    },
+    {
+      "epoch": 0.012371213732047243,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.9756923076923077e-05,
+      "loss": 0.3539,
+      "step": 80
+    },
+    {
+      "epoch": 0.013917615448553147,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.9726153846153848e-05,
+      "loss": 0.259,
+      "step": 90
+    },
+    {
+      "epoch": 0.015464017165059053,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.9695384615384616e-05,
+      "loss": 0.2741,
+      "step": 100
+    },
+    {
+      "epoch": 0.01701041888156496,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.9664615384615387e-05,
+      "loss": 0.281,
+      "step": 110
+    },
+    {
+      "epoch": 0.018556820598070865,
+      "grad_norm": 0.921875,
+      "learning_rate": 1.9633846153846155e-05,
+      "loss": 0.2586,
+      "step": 120
+    },
+    {
+      "epoch": 0.020103222314576768,
+      "grad_norm": 0.921875,
+      "learning_rate": 1.9603076923076926e-05,
+      "loss": 0.2776,
+      "step": 130
+    },
+    {
+      "epoch": 0.021649624031082674,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.9572307692307693e-05,
+      "loss": 0.3186,
+      "step": 140
+    },
+    {
+      "epoch": 0.02319602574758858,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.9541538461538464e-05,
+      "loss": 0.3315,
+      "step": 150
+    },
+    {
+      "epoch": 0.024742427464094486,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.9510769230769232e-05,
+      "loss": 0.257,
+      "step": 160
+    },
+    {
+      "epoch": 0.026288829180600392,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.948e-05,
+      "loss": 0.2592,
+      "step": 170
+    },
+    {
+      "epoch": 0.027835230897106295,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.944923076923077e-05,
+      "loss": 0.2703,
+      "step": 180
+    },
+    {
+      "epoch": 0.0293816326136122,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.941846153846154e-05,
+      "loss": 0.2547,
+      "step": 190
+    },
+    {
+      "epoch": 0.030928034330118107,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.938769230769231e-05,
+      "loss": 0.3182,
+      "step": 200
+    },
+    {
+      "epoch": 0.03247443604662401,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.9356923076923077e-05,
+      "loss": 0.3005,
+      "step": 210
+    },
+    {
+      "epoch": 0.03402083776312992,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.932615384615385e-05,
+      "loss": 0.2693,
+      "step": 220
+    },
+    {
+      "epoch": 0.035567239479635825,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.929538461538462e-05,
+      "loss": 0.2925,
+      "step": 230
+    },
+    {
+      "epoch": 0.03711364119614173,
+      "grad_norm": 1.078125,
+      "learning_rate": 1.9264615384615387e-05,
+      "loss": 0.3165,
+      "step": 240
+    },
+    {
+      "epoch": 0.03866004291264763,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.9233846153846155e-05,
+      "loss": 0.2606,
+      "step": 250
+    },
+    {
+      "epoch": 0.040206444629153536,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.9203076923076923e-05,
+      "loss": 0.324,
+      "step": 260
+    },
+    {
+      "epoch": 0.04175284634565944,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.9172307692307694e-05,
+      "loss": 0.2787,
+      "step": 270
+    },
+    {
+      "epoch": 0.04329924806216535,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.914153846153846e-05,
+      "loss": 0.3092,
+      "step": 280
+    },
+    {
+      "epoch": 0.044845649778671254,
+      "grad_norm": 0.875,
+      "learning_rate": 1.9110769230769233e-05,
+      "loss": 0.2831,
+      "step": 290
+    },
+    {
+      "epoch": 0.04639205149517716,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.908e-05,
+      "loss": 0.282,
+      "step": 300
+    },
+    {
+      "epoch": 0.047938453211683066,
+      "grad_norm": 1.8125,
+      "learning_rate": 1.904923076923077e-05,
+      "loss": 0.3863,
+      "step": 310
+    },
+    {
+      "epoch": 0.04948485492818897,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.901846153846154e-05,
+      "loss": 0.246,
+      "step": 320
+    },
+    {
+      "epoch": 0.05103125664469488,
+      "grad_norm": 1.3984375,
+      "learning_rate": 1.898769230769231e-05,
+      "loss": 0.3483,
+      "step": 330
+    },
+    {
+      "epoch": 0.052577658361200784,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.8956923076923078e-05,
+      "loss": 0.4107,
+      "step": 340
+    },
+    {
+      "epoch": 0.05412406007770668,
+      "grad_norm": 1.0,
+      "learning_rate": 1.892615384615385e-05,
+      "loss": 0.2813,
+      "step": 350
+    },
+    {
+      "epoch": 0.05567046179421259,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.8895384615384617e-05,
+      "loss": 0.283,
+      "step": 360
+    },
+    {
+      "epoch": 0.057216863510718495,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.8864615384615384e-05,
+      "loss": 0.268,
+      "step": 370
+    },
+    {
+      "epoch": 0.0587632652272244,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.8833846153846155e-05,
+      "loss": 0.2852,
+      "step": 380
+    },
+    {
+      "epoch": 0.06030966694373031,
+      "grad_norm": 0.921875,
+      "learning_rate": 1.8803076923076923e-05,
+      "loss": 0.2477,
+      "step": 390
+    },
+    {
+      "epoch": 0.06185606866023621,
+      "grad_norm": 1.0,
+      "learning_rate": 1.8772307692307694e-05,
+      "loss": 0.2418,
+      "step": 400
+    },
+    {
+      "epoch": 0.06340247037674211,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.8741538461538462e-05,
+      "loss": 0.2218,
+      "step": 410
+    },
+    {
+      "epoch": 0.06494887209324803,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.8710769230769233e-05,
+      "loss": 0.2616,
+      "step": 420
+    },
+    {
+      "epoch": 0.06649527380975392,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.8680000000000004e-05,
+      "loss": 0.3475,
+      "step": 430
+    },
+    {
+      "epoch": 0.06804167552625984,
+      "grad_norm": 1.125,
+      "learning_rate": 1.8649230769230772e-05,
+      "loss": 0.3025,
+      "step": 440
+    },
+    {
+      "epoch": 0.06958807724276574,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.861846153846154e-05,
+      "loss": 0.3119,
+      "step": 450
+    },
+    {
+      "epoch": 0.07113447895927165,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.8587692307692307e-05,
+      "loss": 0.3004,
+      "step": 460
+    },
+    {
+      "epoch": 0.07268088067577755,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.8556923076923078e-05,
+      "loss": 0.2957,
+      "step": 470
+    },
+    {
+      "epoch": 0.07422728239228346,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.8526153846153846e-05,
+      "loss": 0.3162,
+      "step": 480
+    },
+    {
+      "epoch": 0.07577368410878936,
+      "grad_norm": 1.75,
+      "learning_rate": 1.8495384615384617e-05,
+      "loss": 0.3637,
+      "step": 490
+    },
+    {
+      "epoch": 0.07732008582529526,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.8464615384615385e-05,
+      "loss": 0.2379,
+      "step": 500
+    },
+    {
+      "epoch": 0.07886648754180117,
+      "grad_norm": 0.984375,
+      "learning_rate": 1.8433846153846156e-05,
+      "loss": 0.3098,
+      "step": 510
+    },
+    {
+      "epoch": 0.08041288925830707,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.8403076923076924e-05,
+      "loss": 0.3977,
+      "step": 520
+    },
+    {
+      "epoch": 0.08195929097481298,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.8372307692307695e-05,
+      "loss": 0.3034,
+      "step": 530
+    },
+    {
+      "epoch": 0.08350569269131888,
+      "grad_norm": 0.7421875,
+      "learning_rate": 1.8341538461538462e-05,
+      "loss": 0.2327,
+      "step": 540
+    },
+    {
+      "epoch": 0.0850520944078248,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.8310769230769233e-05,
+      "loss": 0.2561,
+      "step": 550
+    },
+    {
+      "epoch": 0.0865984961243307,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.828e-05,
+      "loss": 0.3739,
+      "step": 560
+    },
+    {
+      "epoch": 0.08814489784083661,
+      "grad_norm": 0.87890625,
+      "learning_rate": 1.824923076923077e-05,
+      "loss": 0.3605,
+      "step": 570
+    },
+    {
+      "epoch": 0.08969129955734251,
+      "grad_norm": 1.0,
+      "learning_rate": 1.821846153846154e-05,
+      "loss": 0.2557,
+      "step": 580
+    },
+    {
+      "epoch": 0.09123770127384842,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.8187692307692308e-05,
+      "loss": 0.2806,
+      "step": 590
+    },
+    {
+      "epoch": 0.09278410299035432,
+      "grad_norm": 0.875,
+      "learning_rate": 1.815692307692308e-05,
+      "loss": 0.2977,
+      "step": 600
+    },
+    {
+      "epoch": 0.09433050470686022,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.8126153846153846e-05,
+      "loss": 0.2845,
+      "step": 610
+    },
+    {
+      "epoch": 0.09587690642336613,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.8095384615384618e-05,
+      "loss": 0.3309,
+      "step": 620
+    },
+    {
+      "epoch": 0.09742330813987203,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.806461538461539e-05,
+      "loss": 0.3197,
+      "step": 630
+    },
+    {
+      "epoch": 0.09896970985637794,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.8033846153846156e-05,
+      "loss": 0.2654,
+      "step": 640
+    },
+    {
+      "epoch": 0.10051611157288384,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.8003076923076924e-05,
+      "loss": 0.2954,
+      "step": 650
+    },
+    {
+      "epoch": 0.10206251328938976,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.7972307692307692e-05,
+      "loss": 0.3237,
+      "step": 660
+    },
+    {
+      "epoch": 0.10360891500589565,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.7941538461538463e-05,
+      "loss": 0.2887,
+      "step": 670
+    },
+    {
+      "epoch": 0.10515531672240157,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.791076923076923e-05,
+      "loss": 0.3018,
+      "step": 680
+    },
+    {
+      "epoch": 0.10670171843890747,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.788e-05,
+      "loss": 0.261,
+      "step": 690
+    },
+    {
+      "epoch": 0.10824812015541337,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.784923076923077e-05,
+      "loss": 0.254,
+      "step": 700
+    },
+    {
+      "epoch": 0.10979452187191928,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.781846153846154e-05,
+      "loss": 0.2944,
+      "step": 710
+    },
+    {
+      "epoch": 0.11134092358842518,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.778769230769231e-05,
+      "loss": 0.3163,
+      "step": 720
+    },
+    {
+      "epoch": 0.11288732530493109,
+      "grad_norm": 0.640625,
+      "learning_rate": 1.775692307692308e-05,
+      "loss": 0.2838,
+      "step": 730
+    },
+    {
+      "epoch": 0.11443372702143699,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.7726153846153847e-05,
+      "loss": 0.236,
+      "step": 740
+    },
+    {
+      "epoch": 0.1159801287379429,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.7695384615384618e-05,
+      "loss": 0.2164,
+      "step": 750
+    },
+    {
+      "epoch": 0.1175265304544488,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.7664615384615386e-05,
+      "loss": 0.3331,
+      "step": 760
+    },
+    {
+      "epoch": 0.11907293217095472,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.7633846153846153e-05,
+      "loss": 0.303,
+      "step": 770
+    },
+    {
+      "epoch": 0.12061933388746061,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.7603076923076924e-05,
+      "loss": 0.3264,
+      "step": 780
+    },
+    {
+      "epoch": 0.12216573560396653,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.7572307692307692e-05,
+      "loss": 0.2097,
+      "step": 790
+    },
+    {
+      "epoch": 0.12371213732047243,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.7541538461538463e-05,
+      "loss": 0.2456,
+      "step": 800
+    },
+    {
+      "epoch": 0.12525853903697834,
+      "grad_norm": 0.875,
+      "learning_rate": 1.751076923076923e-05,
+      "loss": 0.2877,
+      "step": 810
+    },
+    {
+      "epoch": 0.12680494075348422,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.7480000000000002e-05,
+      "loss": 0.2902,
+      "step": 820
+    },
+    {
+      "epoch": 0.12835134246999014,
+      "grad_norm": 0.875,
+      "learning_rate": 1.7449230769230773e-05,
+      "loss": 0.2357,
+      "step": 830
+    },
+    {
+      "epoch": 0.12989774418649605,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.741846153846154e-05,
+      "loss": 0.2926,
+      "step": 840
+    },
+    {
+      "epoch": 0.13144414590300196,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.738769230769231e-05,
+      "loss": 0.2301,
+      "step": 850
+    },
+    {
+      "epoch": 0.13299054761950785,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.7356923076923076e-05,
+      "loss": 0.2501,
+      "step": 860
+    },
+    {
+      "epoch": 0.13453694933601376,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.7326153846153847e-05,
+      "loss": 0.2393,
+      "step": 870
+    },
+    {
+      "epoch": 0.13608335105251967,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.7295384615384615e-05,
+      "loss": 0.2337,
+      "step": 880
+    },
+    {
+      "epoch": 0.1376297527690256,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.7264615384615386e-05,
+      "loss": 0.3147,
+      "step": 890
+    },
+    {
+      "epoch": 0.13917615448553147,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.7233846153846154e-05,
+      "loss": 0.2949,
+      "step": 900
+    },
+    {
+      "epoch": 0.14072255620203739,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.7203076923076925e-05,
+      "loss": 0.3394,
+      "step": 910
+    },
+    {
+      "epoch": 0.1422689579185433,
+      "grad_norm": 0.72265625,
+      "learning_rate": 1.7172307692307696e-05,
+      "loss": 0.3119,
+      "step": 920
+    },
+    {
+      "epoch": 0.14381535963504918,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.7141538461538464e-05,
+      "loss": 0.2959,
+      "step": 930
+    },
+    {
+      "epoch": 0.1453617613515551,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.711076923076923e-05,
+      "loss": 0.2677,
+      "step": 940
+    },
+    {
+      "epoch": 0.146908163068061,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.7080000000000002e-05,
+      "loss": 0.2575,
+      "step": 950
+    },
+    {
+      "epoch": 0.14845456478456692,
+      "grad_norm": 1.1640625,
+      "learning_rate": 1.704923076923077e-05,
+      "loss": 0.2419,
+      "step": 960
+    },
+    {
+      "epoch": 0.1500009665010728,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.7018461538461538e-05,
+      "loss": 0.2631,
+      "step": 970
+    },
+    {
+      "epoch": 0.15154736821757872,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.698769230769231e-05,
+      "loss": 0.2415,
+      "step": 980
+    },
+    {
+      "epoch": 0.15309376993408463,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.6956923076923077e-05,
+      "loss": 0.2498,
+      "step": 990
+    },
+    {
+      "epoch": 0.15464017165059052,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.6926153846153848e-05,
+      "loss": 0.2845,
+      "step": 1000
+    },
+    {
+      "epoch": 0.15618657336709643,
+      "grad_norm": 1.0,
+      "learning_rate": 1.6895384615384615e-05,
+      "loss": 0.3159,
+      "step": 1010
+    },
+    {
+      "epoch": 0.15773297508360234,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.6864615384615387e-05,
+      "loss": 0.2969,
+      "step": 1020
+    },
+    {
+      "epoch": 0.15927937680010826,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.6833846153846158e-05,
+      "loss": 0.3195,
+      "step": 1030
+    },
+    {
+      "epoch": 0.16082577851661414,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.6803076923076925e-05,
+      "loss": 0.3086,
+      "step": 1040
+    },
+    {
+      "epoch": 0.16237218023312006,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.6772307692307693e-05,
+      "loss": 0.297,
+      "step": 1050
+    },
+    {
+      "epoch": 0.16391858194962597,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.674153846153846e-05,
+      "loss": 0.2677,
+      "step": 1060
+    },
+    {
+      "epoch": 0.16546498366613188,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.6710769230769232e-05,
+      "loss": 0.294,
+      "step": 1070
+    },
+    {
+      "epoch": 0.16701138538263777,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.668e-05,
+      "loss": 0.2483,
+      "step": 1080
+    },
+    {
+      "epoch": 0.16855778709914368,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.664923076923077e-05,
+      "loss": 0.2564,
+      "step": 1090
+    },
+    {
+      "epoch": 0.1701041888156496,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.661846153846154e-05,
+      "loss": 0.2363,
+      "step": 1100
+    },
+    {
+      "epoch": 0.17165059053215548,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.658769230769231e-05,
+      "loss": 0.2486,
+      "step": 1110
+    },
+    {
+      "epoch": 0.1731969922486614,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.655692307692308e-05,
+      "loss": 0.3142,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1747433939651673,
+      "grad_norm": 1.609375,
+      "learning_rate": 1.6526153846153848e-05,
+      "loss": 0.4319,
+      "step": 1130
+    },
+    {
+      "epoch": 0.17628979568167322,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.6495384615384616e-05,
+      "loss": 0.2727,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1778361973981791,
+      "grad_norm": 0.92578125,
+      "learning_rate": 1.6464615384615387e-05,
+      "loss": 0.2472,
+      "step": 1150
+    },
+    {
+      "epoch": 0.17938259911468502,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.6433846153846155e-05,
+      "loss": 0.3036,
+      "step": 1160
+    },
+    {
+      "epoch": 0.18092900083119093,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.6403076923076922e-05,
+      "loss": 0.2199,
+      "step": 1170
+    },
+    {
+      "epoch": 0.18247540254769684,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.6372307692307693e-05,
+      "loss": 0.2474,
+      "step": 1180
+    },
+    {
+      "epoch": 0.18402180426420273,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.634153846153846e-05,
+      "loss": 0.2892,
+      "step": 1190
+    },
+    {
+      "epoch": 0.18556820598070864,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.6310769230769232e-05,
+      "loss": 0.3317,
+      "step": 1200
+    },
+    {
+      "epoch": 0.18711460769721455,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.628e-05,
+      "loss": 0.3066,
+      "step": 1210
+    },
+    {
+      "epoch": 0.18866100941372044,
+      "grad_norm": 0.88671875,
+      "learning_rate": 1.624923076923077e-05,
+      "loss": 0.2811,
+      "step": 1220
+    },
+    {
+      "epoch": 0.19020741113022635,
+      "grad_norm": 0.87890625,
+      "learning_rate": 1.6218461538461542e-05,
+      "loss": 0.2503,
+      "step": 1230
+    },
+    {
+      "epoch": 0.19175381284673226,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.618769230769231e-05,
+      "loss": 0.3128,
+      "step": 1240
+    },
+    {
+      "epoch": 0.19330021456323818,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.6156923076923078e-05,
+      "loss": 0.3067,
+      "step": 1250
+    },
+    {
+      "epoch": 0.19484661627974406,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.6126153846153845e-05,
+      "loss": 0.2975,
+      "step": 1260
+    },
+    {
+      "epoch": 0.19639301799624997,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.6095384615384616e-05,
+      "loss": 0.3083,
+      "step": 1270
+    },
+    {
+      "epoch": 0.1979394197127559,
+      "grad_norm": 0.66796875,
+      "learning_rate": 1.6064615384615384e-05,
+      "loss": 0.2786,
+      "step": 1280
+    },
+    {
+      "epoch": 0.19948582142926177,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.6033846153846155e-05,
+      "loss": 0.404,
+      "step": 1290
+    },
+    {
+      "epoch": 0.20103222314576769,
+      "grad_norm": 0.8984375,
+      "learning_rate": 1.6003076923076923e-05,
+      "loss": 0.3213,
+      "step": 1300
+    },
+    {
+      "epoch": 0.2025786248622736,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.5972307692307694e-05,
+      "loss": 0.24,
+      "step": 1310
+    },
+    {
+      "epoch": 0.2041250265787795,
+      "grad_norm": 1.2890625,
+      "learning_rate": 1.5941538461538465e-05,
+      "loss": 0.2711,
+      "step": 1320
+    },
+    {
+      "epoch": 0.2056714282952854,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.5910769230769233e-05,
+      "loss": 0.2493,
+      "step": 1330
+    },
+    {
+      "epoch": 0.2072178300117913,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.588e-05,
+      "loss": 0.353,
+      "step": 1340
+    },
+    {
+      "epoch": 0.20876423172829722,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.584923076923077e-05,
+      "loss": 0.2534,
+      "step": 1350
+    },
+    {
+      "epoch": 0.21031063344480314,
+      "grad_norm": 0.64453125,
+      "learning_rate": 1.581846153846154e-05,
+      "loss": 0.2088,
+      "step": 1360
+    },
+    {
+      "epoch": 0.21185703516130902,
+      "grad_norm": 0.94921875,
+      "learning_rate": 1.5787692307692307e-05,
+      "loss": 0.3146,
+      "step": 1370
+    },
+    {
+      "epoch": 0.21340343687781493,
+      "grad_norm": 1.125,
+      "learning_rate": 1.5756923076923078e-05,
+      "loss": 0.2947,
+      "step": 1380
+    },
+    {
+      "epoch": 0.21494983859432085,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.5726153846153846e-05,
+      "loss": 0.2039,
+      "step": 1390
+    },
+    {
+      "epoch": 0.21649624031082673,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.5695384615384617e-05,
+      "loss": 0.252,
+      "step": 1400
+    },
+    {
+      "epoch": 0.21804264202733264,
+      "grad_norm": 0.875,
+      "learning_rate": 1.5664615384615388e-05,
+      "loss": 0.2689,
+      "step": 1410
+    },
+    {
+      "epoch": 0.21958904374383856,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.5633846153846156e-05,
+      "loss": 0.3239,
+      "step": 1420
+    },
+    {
+      "epoch": 0.22113544546034447,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.5603076923076927e-05,
+      "loss": 0.2891,
+      "step": 1430
+    },
+    {
+      "epoch": 0.22268184717685036,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.5572307692307694e-05,
+      "loss": 0.3306,
+      "step": 1440
+    },
+    {
+      "epoch": 0.22422824889335627,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.5541538461538462e-05,
+      "loss": 0.2971,
+      "step": 1450
+    },
+    {
+      "epoch": 0.22577465060986218,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.551076923076923e-05,
+      "loss": 0.2892,
+      "step": 1460
+    },
+    {
+      "epoch": 0.2273210523263681,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.548e-05,
+      "loss": 0.2773,
+      "step": 1470
+    },
+    {
+      "epoch": 0.22886745404287398,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.544923076923077e-05,
+      "loss": 0.2767,
+      "step": 1480
+    },
+    {
+      "epoch": 0.2304138557593799,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.541846153846154e-05,
+      "loss": 0.2899,
+      "step": 1490
+    },
+    {
+      "epoch": 0.2319602574758858,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.5387692307692307e-05,
+      "loss": 0.2521,
+      "step": 1500
+    },
+    {
+      "epoch": 0.2335066591923917,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.535692307692308e-05,
+      "loss": 0.2479,
+      "step": 1510
+    },
+    {
+      "epoch": 0.2350530609088976,
+      "grad_norm": 1.0,
+      "learning_rate": 1.532615384615385e-05,
+      "loss": 0.3154,
+      "step": 1520
+    },
+    {
+      "epoch": 0.23659946262540352,
+      "grad_norm": 0.83984375,
+      "learning_rate": 1.5295384615384617e-05,
+      "loss": 0.3391,
+      "step": 1530
+    },
+    {
+      "epoch": 0.23814586434190943,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.5264615384615385e-05,
+      "loss": 0.265,
+      "step": 1540
+    },
+    {
+      "epoch": 0.23969226605841532,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.5233846153846154e-05,
+      "loss": 0.2949,
+      "step": 1550
+    },
+    {
+      "epoch": 0.24123866777492123,
+      "grad_norm": 1.265625,
+      "learning_rate": 1.5203076923076925e-05,
+      "loss": 0.3136,
+      "step": 1560
+    },
+    {
+      "epoch": 0.24278506949142714,
+      "grad_norm": 1.4609375,
+      "learning_rate": 1.5172307692307693e-05,
+      "loss": 0.3073,
+      "step": 1570
+    },
+    {
+      "epoch": 0.24433147120793305,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.5141538461538463e-05,
+      "loss": 0.3271,
+      "step": 1580
+    },
+    {
+      "epoch": 0.24587787292443894,
+      "grad_norm": 1.03125,
+      "learning_rate": 1.5110769230769232e-05,
+      "loss": 0.2722,
+      "step": 1590
+    },
+    {
+      "epoch": 0.24742427464094485,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.5080000000000001e-05,
+      "loss": 0.3513,
+      "step": 1600
+    },
+    {
+      "epoch": 0.24897067635745077,
+      "grad_norm": 1.1796875,
+      "learning_rate": 1.504923076923077e-05,
+      "loss": 0.2212,
+      "step": 1610
+    },
+    {
+      "epoch": 0.2505170780739567,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.501846153846154e-05,
+      "loss": 0.2914,
+      "step": 1620
+    },
+    {
+      "epoch": 0.25206347979046256,
+      "grad_norm": 0.83984375,
+      "learning_rate": 1.498769230769231e-05,
+      "loss": 0.281,
+      "step": 1630
+    },
+    {
+      "epoch": 0.25360988150696845,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.4956923076923077e-05,
+      "loss": 0.2509,
+      "step": 1640
+    },
+    {
+      "epoch": 0.2551562832234744,
+      "grad_norm": 1.2578125,
+      "learning_rate": 1.4926153846153848e-05,
+      "loss": 0.2994,
+      "step": 1650
+    },
+    {
+      "epoch": 0.2567026849399803,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.4895384615384616e-05,
+      "loss": 0.2839,
+      "step": 1660
+    },
+    {
+      "epoch": 0.2582490866564862,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.4864615384615385e-05,
+      "loss": 0.229,
+      "step": 1670
+    },
+    {
+      "epoch": 0.2597954883729921,
+      "grad_norm": 0.9453125,
+      "learning_rate": 1.4833846153846155e-05,
+      "loss": 0.2381,
+      "step": 1680
+    },
+    {
+      "epoch": 0.261341890089498,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.4803076923076924e-05,
+      "loss": 0.3495,
+      "step": 1690
+    },
+    {
+      "epoch": 0.2628882918060039,
+      "grad_norm": 0.7421875,
+      "learning_rate": 1.4772307692307692e-05,
+      "loss": 0.2756,
+      "step": 1700
+    },
+    {
+      "epoch": 0.2644346935225098,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.4741538461538463e-05,
+      "loss": 0.3189,
+      "step": 1710
+    },
+    {
+      "epoch": 0.2659810952390157,
+      "grad_norm": 0.765625,
+      "learning_rate": 1.4710769230769232e-05,
+      "loss": 0.289,
+      "step": 1720
+    },
+    {
+      "epoch": 0.26752749695552164,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.4680000000000002e-05,
+      "loss": 0.2848,
+      "step": 1730
+    },
+    {
+      "epoch": 0.2690738986720275,
+      "grad_norm": 2.96875,
+      "learning_rate": 1.4649230769230771e-05,
+      "loss": 0.3115,
+      "step": 1740
+    },
+    {
+      "epoch": 0.2706203003885334,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.4618461538461539e-05,
+      "loss": 0.3075,
+      "step": 1750
+    },
+    {
+      "epoch": 0.27216670210503935,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.458769230769231e-05,
+      "loss": 0.2705,
+      "step": 1760
+    },
+    {
+      "epoch": 0.27371310382154523,
+      "grad_norm": 0.98046875,
+      "learning_rate": 1.4556923076923078e-05,
+      "loss": 0.266,
+      "step": 1770
+    },
+    {
+      "epoch": 0.2752595055380512,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.4526153846153847e-05,
+      "loss": 0.3179,
+      "step": 1780
+    },
+    {
+      "epoch": 0.27680590725455706,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.4495384615384616e-05,
+      "loss": 0.2775,
+      "step": 1790
+    },
+    {
+      "epoch": 0.27835230897106295,
+      "grad_norm": 0.87890625,
+      "learning_rate": 1.4464615384615386e-05,
+      "loss": 0.3279,
+      "step": 1800
+    },
+    {
+      "epoch": 0.2798987106875689,
+      "grad_norm": 0.8515625,
+      "learning_rate": 1.4433846153846155e-05,
+      "loss": 0.2373,
+      "step": 1810
+    },
+    {
+      "epoch": 0.28144511240407477,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.4403076923076925e-05,
+      "loss": 0.2216,
+      "step": 1820
+    },
+    {
+      "epoch": 0.28299151412058066,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.4372307692307694e-05,
+      "loss": 0.3206,
+      "step": 1830
+    },
+    {
+      "epoch": 0.2845379158370866,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.4341538461538462e-05,
+      "loss": 0.2467,
+      "step": 1840
+    },
+    {
+      "epoch": 0.2860843175535925,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.4310769230769233e-05,
+      "loss": 0.2818,
+      "step": 1850
+    },
+    {
+      "epoch": 0.28763071927009837,
+      "grad_norm": 0.8984375,
+      "learning_rate": 1.428e-05,
+      "loss": 0.2477,
+      "step": 1860
+    },
+    {
+      "epoch": 0.2891771209866043,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.4249230769230772e-05,
+      "loss": 0.2576,
+      "step": 1870
+    },
+    {
+      "epoch": 0.2907235227031102,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.421846153846154e-05,
+      "loss": 0.2841,
+      "step": 1880
+    },
+    {
+      "epoch": 0.2922699244196161,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.4187692307692309e-05,
+      "loss": 0.3371,
+      "step": 1890
+    },
+    {
+      "epoch": 0.293816326136122,
+      "grad_norm": 1.0,
+      "learning_rate": 1.4156923076923076e-05,
+      "loss": 0.3037,
+      "step": 1900
+    },
+    {
+      "epoch": 0.2953627278526279,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.4126153846153847e-05,
+      "loss": 0.2526,
+      "step": 1910
+    },
+    {
+      "epoch": 0.29690912956913385,
+      "grad_norm": 0.98828125,
+      "learning_rate": 1.4095384615384617e-05,
+      "loss": 0.2125,
+      "step": 1920
+    },
+    {
+      "epoch": 0.29845553128563973,
+      "grad_norm": 0.8125,
+      "learning_rate": 1.4064615384615386e-05,
+      "loss": 0.2783,
+      "step": 1930
+    },
+    {
+      "epoch": 0.3000019330021456,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.4033846153846156e-05,
+      "loss": 0.3131,
+      "step": 1940
+    },
+    {
+      "epoch": 0.30154833471865156,
+      "grad_norm": 0.78125,
+      "learning_rate": 1.4003076923076923e-05,
+      "loss": 0.3226,
+      "step": 1950
+    },
+    {
+      "epoch": 0.30309473643515744,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.3972307692307694e-05,
+      "loss": 0.2819,
+      "step": 1960
+    },
+    {
+      "epoch": 0.3046411381516633,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.3941538461538462e-05,
+      "loss": 0.2868,
+      "step": 1970
+    },
+    {
+      "epoch": 0.30618753986816927,
+      "grad_norm": 0.92578125,
+      "learning_rate": 1.3910769230769232e-05,
+      "loss": 0.2615,
+      "step": 1980
+    },
+    {
+      "epoch": 0.30773394158467515,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.3880000000000001e-05,
+      "loss": 0.255,
+      "step": 1990
+    },
+    {
+      "epoch": 0.30928034330118104,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.384923076923077e-05,
+      "loss": 0.251,
+      "step": 2000
+    },
+    {
+      "epoch": 0.310826745017687,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.3818461538461541e-05,
+      "loss": 0.2983,
+      "step": 2010
+    },
+    {
+      "epoch": 0.31237314673419286,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.3787692307692309e-05,
+      "loss": 0.2705,
+      "step": 2020
+    },
+    {
+      "epoch": 0.3139195484506988,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.3756923076923079e-05,
+      "loss": 0.2937,
+      "step": 2030
+    },
+    {
+      "epoch": 0.3154659501672047,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.3726153846153846e-05,
+      "loss": 0.3296,
+      "step": 2040
+    },
+    {
+      "epoch": 0.3170123518837106,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.3695384615384617e-05,
+      "loss": 0.2666,
+      "step": 2050
+    },
+    {
+      "epoch": 0.3185587536002165,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.3664615384615385e-05,
+      "loss": 0.3124,
+      "step": 2060
+    },
+    {
+      "epoch": 0.3201051553167224,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.3633846153846156e-05,
+      "loss": 0.3752,
+      "step": 2070
+    },
+    {
+      "epoch": 0.3216515570332283,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.3603076923076924e-05,
+      "loss": 0.2622,
+      "step": 2080
+    },
+    {
+      "epoch": 0.3231979587497342,
+      "grad_norm": 0.734375,
+      "learning_rate": 1.3572307692307693e-05,
+      "loss": 0.2526,
+      "step": 2090
+    },
+    {
+      "epoch": 0.3247443604662401,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.3541538461538464e-05,
+      "loss": 0.2775,
+      "step": 2100
+    },
+    {
+      "epoch": 0.326290762182746,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.3510769230769232e-05,
+      "loss": 0.3322,
+      "step": 2110
+    },
+    {
+      "epoch": 0.32783716389925194,
+      "grad_norm": 1.0,
+      "learning_rate": 1.3480000000000001e-05,
+      "loss": 0.2897,
+      "step": 2120
+    },
+    {
+      "epoch": 0.3293835656157578,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.344923076923077e-05,
+      "loss": 0.2815,
+      "step": 2130
+    },
+    {
+      "epoch": 0.33092996733226376,
+      "grad_norm": 0.7109375,
+      "learning_rate": 1.341846153846154e-05,
+      "loss": 0.2959,
+      "step": 2140
+    },
+    {
+      "epoch": 0.33247636904876965,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.3387692307692308e-05,
+      "loss": 0.2571,
+      "step": 2150
+    },
+    {
+      "epoch": 0.33402277076527553,
+      "grad_norm": 0.91015625,
+      "learning_rate": 1.3356923076923079e-05,
+      "loss": 0.247,
+      "step": 2160
+    },
+    {
+      "epoch": 0.3355691724817815,
+      "grad_norm": 0.828125,
+      "learning_rate": 1.3326153846153847e-05,
+      "loss": 0.248,
+      "step": 2170
+    },
+    {
+      "epoch": 0.33711557419828736,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.3295384615384616e-05,
+      "loss": 0.2438,
+      "step": 2180
+    },
+    {
+      "epoch": 0.33866197591479325,
+      "grad_norm": 1.2734375,
+      "learning_rate": 1.3264615384615385e-05,
+      "loss": 0.3612,
+      "step": 2190
+    },
+    {
+      "epoch": 0.3402083776312992,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.3233846153846155e-05,
+      "loss": 0.3287,
+      "step": 2200
+    },
+    {
+      "epoch": 0.34175477934780507,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.3203076923076926e-05,
+      "loss": 0.2756,
+      "step": 2210
+    },
+    {
+      "epoch": 0.34330118106431096,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.3172307692307694e-05,
+      "loss": 0.2886,
+      "step": 2220
+    },
+    {
+      "epoch": 0.3448475827808169,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.3141538461538463e-05,
+      "loss": 0.2446,
+      "step": 2230
+    },
+    {
+      "epoch": 0.3463939844973228,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.311076923076923e-05,
+      "loss": 0.3064,
+      "step": 2240
+    },
+    {
+      "epoch": 0.3479403862138287,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.3080000000000002e-05,
+      "loss": 0.2376,
+      "step": 2250
+    },
+    {
+      "epoch": 0.3494867879303346,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.304923076923077e-05,
+      "loss": 0.2932,
+      "step": 2260
+    },
+    {
+      "epoch": 0.3510331896468405,
+      "grad_norm": 1.4296875,
+      "learning_rate": 1.301846153846154e-05,
+      "loss": 0.2979,
+      "step": 2270
+    },
+    {
+      "epoch": 0.35257959136334643,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.2987692307692308e-05,
+      "loss": 0.2897,
+      "step": 2280
+    },
+    {
+      "epoch": 0.3541259930798523,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.2956923076923078e-05,
+      "loss": 0.2744,
+      "step": 2290
+    },
+    {
+      "epoch": 0.3556723947963582,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.2926153846153849e-05,
+      "loss": 0.2708,
+      "step": 2300
+    },
+    {
+      "epoch": 0.35721879651286415,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.2895384615384616e-05,
+      "loss": 0.2224,
+      "step": 2310
+    },
+    {
+      "epoch": 0.35876519822937003,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.2864615384615386e-05,
+      "loss": 0.2728,
+      "step": 2320
+    },
+    {
+      "epoch": 0.3603115999458759,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.2833846153846155e-05,
+      "loss": 0.2661,
+      "step": 2330
+    },
+    {
+      "epoch": 0.36185800166238186,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.2803076923076925e-05,
+      "loss": 0.2892,
+      "step": 2340
+    },
+    {
+      "epoch": 0.36340440337888774,
+      "grad_norm": 0.6640625,
+      "learning_rate": 1.2772307692307692e-05,
+      "loss": 0.3092,
+      "step": 2350
+    },
+    {
+      "epoch": 0.3649508050953937,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.2741538461538463e-05,
+      "loss": 0.2542,
+      "step": 2360
+    },
+    {
+      "epoch": 0.36649720681189957,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.2710769230769231e-05,
+      "loss": 0.3589,
+      "step": 2370
+    },
+    {
+      "epoch": 0.36804360852840545,
+      "grad_norm": 0.71484375,
+      "learning_rate": 1.268e-05,
+      "loss": 0.2237,
+      "step": 2380
+    },
+    {
+      "epoch": 0.3695900102449114,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.264923076923077e-05,
+      "loss": 0.3413,
+      "step": 2390
+    },
+    {
+      "epoch": 0.3711364119614173,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.261846153846154e-05,
+      "loss": 0.2556,
+      "step": 2400
+    },
+    {
+      "epoch": 0.37268281367792316,
+      "grad_norm": 0.9296875,
+      "learning_rate": 1.258769230769231e-05,
+      "loss": 0.3087,
+      "step": 2410
+    },
+    {
+      "epoch": 0.3742292153944291,
+      "grad_norm": 0.95703125,
+      "learning_rate": 1.2556923076923078e-05,
+      "loss": 0.2609,
+      "step": 2420
+    },
+    {
+      "epoch": 0.375775617110935,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.2526153846153848e-05,
+      "loss": 0.2572,
+      "step": 2430
+    },
+    {
+      "epoch": 0.3773220188274409,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.2495384615384615e-05,
+      "loss": 0.3003,
+      "step": 2440
+    },
+    {
+      "epoch": 0.3788684205439468,
+      "grad_norm": 1.1484375,
+      "learning_rate": 1.2464615384615386e-05,
+      "loss": 0.259,
+      "step": 2450
+    },
+    {
+      "epoch": 0.3804148222604527,
+      "grad_norm": 1.203125,
+      "learning_rate": 1.2433846153846154e-05,
+      "loss": 0.2606,
+      "step": 2460
+    },
+    {
+      "epoch": 0.38196122397695864,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.2403076923076925e-05,
+      "loss": 0.2351,
+      "step": 2470
+    },
+    {
+      "epoch": 0.3835076256934645,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.2372307692307693e-05,
+      "loss": 0.2664,
+      "step": 2480
+    },
+    {
+      "epoch": 0.3850540274099704,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.2341538461538462e-05,
+      "loss": 0.245,
+      "step": 2490
+    },
+    {
+      "epoch": 0.38660042912647635,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.2310769230769233e-05,
+      "loss": 0.2781,
+      "step": 2500
+    },
+    {
+      "epoch": 0.38814683084298224,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.2280000000000001e-05,
+      "loss": 0.2847,
+      "step": 2510
+    },
+    {
+      "epoch": 0.3896932325594881,
+      "grad_norm": 0.8203125,
+      "learning_rate": 1.224923076923077e-05,
+      "loss": 0.3223,
+      "step": 2520
+    },
+    {
+      "epoch": 0.39123963427599406,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.221846153846154e-05,
+      "loss": 0.3068,
+      "step": 2530
+    },
+    {
+      "epoch": 0.39278603599249995,
+      "grad_norm": 0.9609375,
+      "learning_rate": 1.218769230769231e-05,
+      "loss": 0.2393,
+      "step": 2540
+    },
+    {
+      "epoch": 0.39433243770900583,
+      "grad_norm": 1.09375,
+      "learning_rate": 1.2156923076923077e-05,
+      "loss": 0.2918,
+      "step": 2550
+    },
+    {
+      "epoch": 0.3958788394255118,
+      "grad_norm": 0.84375,
+      "learning_rate": 1.2126153846153848e-05,
+      "loss": 0.2146,
+      "step": 2560
+    },
+    {
+      "epoch": 0.39742524114201766,
+      "grad_norm": 0.875,
+      "learning_rate": 1.2095384615384616e-05,
+      "loss": 0.3178,
+      "step": 2570
+    },
+    {
+      "epoch": 0.39897164285852355,
+      "grad_norm": 0.78515625,
+      "learning_rate": 1.2064615384615385e-05,
+      "loss": 0.2247,
+      "step": 2580
+    },
+    {
+      "epoch": 0.4005180445750295,
+      "grad_norm": 0.93359375,
+      "learning_rate": 1.2033846153846154e-05,
+      "loss": 0.2684,
+      "step": 2590
+    },
+    {
+      "epoch": 0.40206444629153537,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.2003076923076924e-05,
+      "loss": 0.2332,
+      "step": 2600
+    },
+    {
+      "epoch": 0.4036108480080413,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.1972307692307695e-05,
+      "loss": 0.4153,
+      "step": 2610
+    },
+    {
+      "epoch": 0.4051572497245472,
+      "grad_norm": 0.96484375,
+      "learning_rate": 1.1941538461538463e-05,
+      "loss": 0.2559,
+      "step": 2620
+    },
+    {
+      "epoch": 0.4067036514410531,
+      "grad_norm": 1.53125,
+      "learning_rate": 1.1910769230769232e-05,
+      "loss": 0.2974,
+      "step": 2630
+    },
+    {
+      "epoch": 0.408250053157559,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.188e-05,
+      "loss": 0.3523,
+      "step": 2640
+    },
+    {
+      "epoch": 0.4097964548740649,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.1849230769230771e-05,
+      "loss": 0.2619,
+      "step": 2650
+    },
+    {
+      "epoch": 0.4113428565905708,
+      "grad_norm": 1.109375,
+      "learning_rate": 1.1818461538461539e-05,
+      "loss": 0.2833,
+      "step": 2660
+    },
+    {
+      "epoch": 0.41288925830707673,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.178769230769231e-05,
+      "loss": 0.244,
+      "step": 2670
+    },
+    {
+      "epoch": 0.4144356600235826,
+      "grad_norm": 0.8046875,
+      "learning_rate": 1.1756923076923077e-05,
+      "loss": 0.2238,
+      "step": 2680
+    },
+    {
+      "epoch": 0.4159820617400885,
+      "grad_norm": 0.87890625,
+      "learning_rate": 1.1726153846153847e-05,
+      "loss": 0.2839,
+      "step": 2690
+    },
+    {
+      "epoch": 0.41752846345659445,
+      "grad_norm": 1.46875,
+      "learning_rate": 1.1695384615384618e-05,
+      "loss": 0.3264,
+      "step": 2700
+    },
+    {
+      "epoch": 0.41907486517310033,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.1664615384615386e-05,
+      "loss": 0.2501,
+      "step": 2710
+    },
+    {
+      "epoch": 0.42062126688960627,
+      "grad_norm": 0.94140625,
+      "learning_rate": 1.1633846153846155e-05,
+      "loss": 0.3618,
+      "step": 2720
+    },
+    {
+      "epoch": 0.42216766860611216,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.1603076923076924e-05,
+      "loss": 0.2353,
+      "step": 2730
+    },
+    {
+      "epoch": 0.42371407032261804,
+      "grad_norm": 0.79296875,
+      "learning_rate": 1.1572307692307694e-05,
+      "loss": 0.2745,
+      "step": 2740
+    },
+    {
+      "epoch": 0.425260472039124,
+      "grad_norm": 1.1171875,
+      "learning_rate": 1.1541538461538461e-05,
+      "loss": 0.2673,
+      "step": 2750
+    },
+    {
+      "epoch": 0.42680687375562987,
+      "grad_norm": 1.1796875,
+      "learning_rate": 1.1510769230769232e-05,
+      "loss": 0.2448,
+      "step": 2760
+    },
+    {
+      "epoch": 0.42835327547213575,
+      "grad_norm": 0.82421875,
+      "learning_rate": 1.148e-05,
+      "loss": 0.2428,
+      "step": 2770
+    },
+    {
+      "epoch": 0.4298996771886417,
+      "grad_norm": 0.97265625,
+      "learning_rate": 1.144923076923077e-05,
+      "loss": 0.2259,
+      "step": 2780
+    },
+    {
+      "epoch": 0.4314460789051476,
+      "grad_norm": 1.0625,
+      "learning_rate": 1.141846153846154e-05,
+      "loss": 0.3214,
+      "step": 2790
+    },
+    {
+      "epoch": 0.43299248062165346,
+      "grad_norm": 0.90625,
+      "learning_rate": 1.1387692307692308e-05,
+      "loss": 0.272,
+      "step": 2800
+    },
+    {
+      "epoch": 0.4345388823381594,
+      "grad_norm": 0.7890625,
+      "learning_rate": 1.135692307692308e-05,
+      "loss": 0.3055,
+      "step": 2810
+    },
+    {
+      "epoch": 0.4360852840546653,
+      "grad_norm": 1.3359375,
+      "learning_rate": 1.1326153846153847e-05,
+      "loss": 0.3783,
+      "step": 2820
+    },
+    {
+      "epoch": 0.43763168577117123,
+      "grad_norm": 0.83203125,
+      "learning_rate": 1.1295384615384617e-05,
+      "loss": 0.2318,
+      "step": 2830
+    },
+    {
+      "epoch": 0.4391780874876771,
+      "grad_norm": 1.0,
+      "learning_rate": 1.1264615384615384e-05,
+      "loss": 0.3072,
+      "step": 2840
+    },
+    {
+      "epoch": 0.440724489204183,
+      "grad_norm": 1.453125,
+      "learning_rate": 1.1233846153846155e-05,
+      "loss": 0.3272,
+      "step": 2850
+    },
+    {
+      "epoch": 0.44227089092068894,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.1203076923076923e-05,
+      "loss": 0.2422,
+      "step": 2860
+    },
+    {
+      "epoch": 0.4438172926371948,
+      "grad_norm": 0.83984375,
+      "learning_rate": 1.1172307692307694e-05,
+      "loss": 0.2424,
+      "step": 2870
+    },
+    {
+      "epoch": 0.4453636943537007,
+      "grad_norm": 1.0703125,
+      "learning_rate": 1.1141538461538462e-05,
+      "loss": 0.288,
+      "step": 2880
+    },
+    {
+      "epoch": 0.44691009607020665,
+      "grad_norm": 0.90234375,
+      "learning_rate": 1.1110769230769231e-05,
+      "loss": 0.2376,
+      "step": 2890
+    },
+    {
+      "epoch": 0.44845649778671254,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.1080000000000002e-05,
+      "loss": 0.2541,
+      "step": 2900
+    },
+    {
+      "epoch": 0.4500028995032184,
+      "grad_norm": 0.80078125,
+      "learning_rate": 1.104923076923077e-05,
+      "loss": 0.2928,
+      "step": 2910
+    },
+    {
+      "epoch": 0.45154930121972436,
+      "grad_norm": 0.83984375,
+      "learning_rate": 1.101846153846154e-05,
+      "loss": 0.2582,
+      "step": 2920
+    },
+    {
+      "epoch": 0.45309570293623025,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.0987692307692309e-05,
+      "loss": 0.2548,
+      "step": 2930
+    },
+    {
+      "epoch": 0.4546421046527362,
+      "grad_norm": 1.421875,
+      "learning_rate": 1.0956923076923078e-05,
+      "loss": 0.3462,
+      "step": 2940
+    },
+    {
+      "epoch": 0.4561885063692421,
+      "grad_norm": 1.0,
+      "learning_rate": 1.0926153846153846e-05,
+      "loss": 0.3076,
+      "step": 2950
+    },
+    {
+      "epoch": 0.45773490808574796,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.0895384615384617e-05,
+      "loss": 0.2761,
+      "step": 2960
+    },
+    {
+      "epoch": 0.4592813098022539,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.0864615384615385e-05,
+      "loss": 0.3359,
+      "step": 2970
+    },
+    {
+      "epoch": 0.4608277115187598,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.0833846153846154e-05,
+      "loss": 0.3213,
+      "step": 2980
+    },
+    {
+      "epoch": 0.46237411323526567,
+      "grad_norm": 0.85546875,
+      "learning_rate": 1.0803076923076925e-05,
+      "loss": 0.2917,
+      "step": 2990
+    },
+    {
+      "epoch": 0.4639205149517716,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.0772307692307693e-05,
+      "loss": 0.2774,
+      "step": 3000
+    },
+    {
+      "epoch": 0.4654669166682775,
+      "grad_norm": 0.7734375,
+      "learning_rate": 1.0741538461538464e-05,
+      "loss": 0.3373,
+      "step": 3010
+    },
+    {
+      "epoch": 0.4670133183847834,
+      "grad_norm": 0.57421875,
+      "learning_rate": 1.0710769230769232e-05,
+      "loss": 0.248,
+      "step": 3020
+    },
+    {
+      "epoch": 0.4685597201012893,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.0680000000000001e-05,
+      "loss": 0.2782,
+      "step": 3030
+    },
+    {
+      "epoch": 0.4701061218177952,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.0649230769230769e-05,
+      "loss": 0.3041,
+      "step": 3040
+    },
+    {
+      "epoch": 0.47165252353430115,
+      "grad_norm": 0.8359375,
+      "learning_rate": 1.061846153846154e-05,
+      "loss": 0.2109,
+      "step": 3050
+    },
+    {
+      "epoch": 0.47319892525080703,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.0587692307692308e-05,
+      "loss": 0.2815,
+      "step": 3060
+    },
+    {
+      "epoch": 0.4747453269673129,
+      "grad_norm": 1.1875,
+      "learning_rate": 1.0556923076923079e-05,
+      "loss": 0.2775,
+      "step": 3070
+    },
+    {
+      "epoch": 0.47629172868381886,
+      "grad_norm": 1.15625,
+      "learning_rate": 1.0526153846153846e-05,
+      "loss": 0.2645,
+      "step": 3080
+    },
+    {
+      "epoch": 0.47783813040032475,
+      "grad_norm": 0.91796875,
+      "learning_rate": 1.0495384615384616e-05,
+      "loss": 0.2738,
+      "step": 3090
+    },
+    {
+      "epoch": 0.47938453211683063,
+      "grad_norm": 0.73828125,
+      "learning_rate": 1.0464615384615387e-05,
+      "loss": 0.2912,
+      "step": 3100
+    },
+    {
+      "epoch": 0.48093093383333657,
+      "grad_norm": 1.046875,
+      "learning_rate": 1.0433846153846155e-05,
+      "loss": 0.217,
+      "step": 3110
+    },
+    {
+      "epoch": 0.48247733554984246,
+      "grad_norm": 0.96875,
+      "learning_rate": 1.0403076923076924e-05,
+      "loss": 0.3397,
+      "step": 3120
+    },
+    {
+      "epoch": 0.48402373726634834,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.0372307692307693e-05,
+      "loss": 0.215,
+      "step": 3130
+    },
+    {
+      "epoch": 0.4855701389828543,
+      "grad_norm": 1.125,
+      "learning_rate": 1.0341538461538463e-05,
+      "loss": 0.2587,
+      "step": 3140
+    },
+    {
+      "epoch": 0.48711654069936017,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.031076923076923e-05,
+      "loss": 0.2183,
+      "step": 3150
+    },
+    {
+      "epoch": 0.4886629424158661,
+      "grad_norm": 1.140625,
+      "learning_rate": 1.0280000000000002e-05,
+      "loss": 0.2551,
+      "step": 3160
+    },
+    {
+      "epoch": 0.490209344132372,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.024923076923077e-05,
+      "loss": 0.2389,
+      "step": 3170
+    },
+    {
+      "epoch": 0.4917557458488779,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.0218461538461539e-05,
+      "loss": 0.2774,
+      "step": 3180
+    },
+    {
+      "epoch": 0.4933021475653838,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.018769230769231e-05,
+      "loss": 0.2608,
+      "step": 3190
+    },
+    {
+      "epoch": 0.4948485492818897,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.0156923076923077e-05,
+      "loss": 0.3287,
+      "step": 3200
+    },
+    {
+      "epoch": 0.4963949509983956,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.0126153846153849e-05,
+      "loss": 0.236,
+      "step": 3210
+    },
+    {
+      "epoch": 0.49794135271490153,
+      "grad_norm": 1.0,
+      "learning_rate": 1.0095384615384616e-05,
+      "loss": 0.259,
+      "step": 3220
+    },
+    {
+      "epoch": 0.4994877544314074,
+      "grad_norm": 0.6796875,
+      "learning_rate": 1.0064615384615386e-05,
+      "loss": 0.2668,
+      "step": 3230
+    },
+    {
+      "epoch": 0.5010341561479134,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.0033846153846153e-05,
+      "loss": 0.3078,
+      "step": 3240
+    },
+    {
+      "epoch": 0.5025805578644192,
+      "grad_norm": 1.1328125,
+      "learning_rate": 1.0003076923076924e-05,
+      "loss": 0.2674,
+      "step": 3250
+    },
+    {
+      "epoch": 0.5041269595809251,
+      "grad_norm": 1.03125,
+      "learning_rate": 9.972307692307694e-06,
+      "loss": 0.276,
+      "step": 3260
+    },
+    {
+      "epoch": 0.505673361297431,
+      "grad_norm": 0.58203125,
+      "learning_rate": 9.941538461538463e-06,
+      "loss": 0.2331,
+      "step": 3270
+    },
+    {
+      "epoch": 0.5072197630139369,
+      "grad_norm": 0.7734375,
+      "learning_rate": 9.910769230769231e-06,
+      "loss": 0.2518,
+      "step": 3280
+    },
+    {
+      "epoch": 0.5087661647304429,
+      "grad_norm": 0.79296875,
+      "learning_rate": 9.88e-06,
+      "loss": 0.3217,
+      "step": 3290
+    },
+    {
+      "epoch": 0.5103125664469488,
+      "grad_norm": 0.78125,
+      "learning_rate": 9.84923076923077e-06,
+      "loss": 0.2582,
+      "step": 3300
+    },
+    {
+      "epoch": 0.5118589681634547,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.818461538461539e-06,
+      "loss": 0.2967,
+      "step": 3310
+    },
+    {
+      "epoch": 0.5134053698799605,
+      "grad_norm": 0.921875,
+      "learning_rate": 9.787692307692308e-06,
+      "loss": 0.2508,
+      "step": 3320
+    },
+    {
+      "epoch": 0.5149517715964664,
+      "grad_norm": 1.3203125,
+      "learning_rate": 9.756923076923078e-06,
+      "loss": 0.247,
+      "step": 3330
+    },
+    {
+      "epoch": 0.5164981733129724,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.726153846153847e-06,
+      "loss": 0.2664,
+      "step": 3340
+    },
+    {
+      "epoch": 0.5180445750294783,
+      "grad_norm": 0.96875,
+      "learning_rate": 9.695384615384617e-06,
+      "loss": 0.2963,
+      "step": 3350
+    },
+    {
+      "epoch": 0.5195909767459842,
+      "grad_norm": 1.2109375,
+      "learning_rate": 9.664615384615386e-06,
+      "loss": 0.2869,
+      "step": 3360
+    },
+    {
+      "epoch": 0.5211373784624901,
+      "grad_norm": 0.796875,
+      "learning_rate": 9.633846153846155e-06,
+      "loss": 0.2125,
+      "step": 3370
+    },
+    {
+      "epoch": 0.522683780178996,
+      "grad_norm": 0.83203125,
+      "learning_rate": 9.603076923076923e-06,
+      "loss": 0.2379,
+      "step": 3380
+    },
+    {
+      "epoch": 0.5242301818955019,
+      "grad_norm": 1.2578125,
+      "learning_rate": 9.572307692307693e-06,
+      "loss": 0.2897,
+      "step": 3390
+    },
+    {
+      "epoch": 0.5257765836120079,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.541538461538462e-06,
+      "loss": 0.2954,
+      "step": 3400
+    },
+    {
+      "epoch": 0.5273229853285137,
+      "grad_norm": 0.75390625,
+      "learning_rate": 9.510769230769231e-06,
+      "loss": 0.257,
+      "step": 3410
+    },
+    {
+      "epoch": 0.5288693870450196,
+      "grad_norm": 0.77734375,
+      "learning_rate": 9.48e-06,
+      "loss": 0.3107,
+      "step": 3420
+    },
+    {
+      "epoch": 0.5304157887615255,
+      "grad_norm": 0.7890625,
+      "learning_rate": 9.44923076923077e-06,
+      "loss": 0.2178,
+      "step": 3430
+    },
+    {
+      "epoch": 0.5319621904780314,
+      "grad_norm": 1.0625,
+      "learning_rate": 9.41846153846154e-06,
+      "loss": 0.2407,
+      "step": 3440
+    },
+    {
+      "epoch": 0.5335085921945374,
+      "grad_norm": 1.2421875,
+      "learning_rate": 9.387692307692309e-06,
+      "loss": 0.284,
+      "step": 3450
+    },
+    {
+      "epoch": 0.5350549939110433,
+      "grad_norm": 0.8984375,
+      "learning_rate": 9.356923076923078e-06,
+      "loss": 0.2358,
+      "step": 3460
+    },
+    {
+      "epoch": 0.5366013956275492,
+      "grad_norm": 1.1328125,
+      "learning_rate": 9.326153846153848e-06,
+      "loss": 0.2455,
+      "step": 3470
+    },
+    {
+      "epoch": 0.538147797344055,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.295384615384615e-06,
+      "loss": 0.3416,
+      "step": 3480
+    },
+    {
+      "epoch": 0.5396941990605609,
+      "grad_norm": 1.171875,
+      "learning_rate": 9.264615384615385e-06,
+      "loss": 0.2908,
+      "step": 3490
+    },
+    {
+      "epoch": 0.5412406007770668,
+      "grad_norm": 0.83984375,
+      "learning_rate": 9.233846153846154e-06,
+      "loss": 0.2648,
+      "step": 3500
+    },
+    {
+      "epoch": 0.5427870024935728,
+      "grad_norm": 0.984375,
+      "learning_rate": 9.203076923076924e-06,
+      "loss": 0.2159,
+      "step": 3510
+    },
+    {
+      "epoch": 0.5443334042100787,
+      "grad_norm": 1.078125,
+      "learning_rate": 9.172307692307693e-06,
+      "loss": 0.3019,
+      "step": 3520
+    },
+    {
+      "epoch": 0.5458798059265846,
+      "grad_norm": 1.4375,
+      "learning_rate": 9.141538461538462e-06,
+      "loss": 0.2886,
+      "step": 3530
+    },
+    {
+      "epoch": 0.5474262076430905,
+      "grad_norm": 1.1484375,
+      "learning_rate": 9.110769230769232e-06,
+      "loss": 0.2674,
+      "step": 3540
+    },
+    {
+      "epoch": 0.5489726093595964,
+      "grad_norm": 1.015625,
+      "learning_rate": 9.080000000000001e-06,
+      "loss": 0.2809,
+      "step": 3550
+    },
+    {
+      "epoch": 0.5505190110761023,
+      "grad_norm": 0.94140625,
+      "learning_rate": 9.04923076923077e-06,
+      "loss": 0.3105,
+      "step": 3560
+    },
+    {
+      "epoch": 0.5520654127926082,
+      "grad_norm": 0.859375,
+      "learning_rate": 9.01846153846154e-06,
+      "loss": 0.3115,
+      "step": 3570
+    },
+    {
+      "epoch": 0.5536118145091141,
+      "grad_norm": 0.90234375,
+      "learning_rate": 8.987692307692308e-06,
+      "loss": 0.2605,
+      "step": 3580
+    },
+    {
+      "epoch": 0.55515821622562,
+      "grad_norm": 0.703125,
+      "learning_rate": 8.956923076923077e-06,
+      "loss": 0.2281,
+      "step": 3590
+    },
+    {
+      "epoch": 0.5567046179421259,
+      "grad_norm": 0.953125,
+      "learning_rate": 8.926153846153846e-06,
+      "loss": 0.2732,
+      "step": 3600
+    },
+    {
+      "epoch": 0.5582510196586318,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.895384615384616e-06,
+      "loss": 0.2134,
+      "step": 3610
+    },
+    {
+      "epoch": 0.5597974213751378,
+      "grad_norm": 1.140625,
+      "learning_rate": 8.864615384615385e-06,
+      "loss": 0.2788,
+      "step": 3620
+    },
+    {
+      "epoch": 0.5613438230916437,
+      "grad_norm": 0.8984375,
+      "learning_rate": 8.833846153846155e-06,
+      "loss": 0.2558,
+      "step": 3630
+    },
+    {
+      "epoch": 0.5628902248081495,
+      "grad_norm": 0.7265625,
+      "learning_rate": 8.803076923076924e-06,
+      "loss": 0.2719,
+      "step": 3640
+    },
+    {
+      "epoch": 0.5644366265246554,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.772307692307693e-06,
+      "loss": 0.2596,
+      "step": 3650
+    },
+    {
+      "epoch": 0.5659830282411613,
+      "grad_norm": 0.8828125,
+      "learning_rate": 8.741538461538463e-06,
+      "loss": 0.2484,
+      "step": 3660
+    },
+    {
+      "epoch": 0.5675294299576673,
+      "grad_norm": 0.734375,
+      "learning_rate": 8.710769230769232e-06,
+      "loss": 0.2734,
+      "step": 3670
+    },
+    {
+      "epoch": 0.5690758316741732,
+      "grad_norm": 0.91796875,
+      "learning_rate": 8.68e-06,
+      "loss": 0.2728,
+      "step": 3680
+    },
+    {
+      "epoch": 0.5706222333906791,
+      "grad_norm": 0.9765625,
+      "learning_rate": 8.64923076923077e-06,
+      "loss": 0.2746,
+      "step": 3690
+    },
+    {
+      "epoch": 0.572168635107185,
+      "grad_norm": 0.94140625,
+      "learning_rate": 8.618461538461539e-06,
+      "loss": 0.2767,
+      "step": 3700
+    },
+    {
+      "epoch": 0.5737150368236908,
+      "grad_norm": 0.80859375,
+      "learning_rate": 8.587692307692308e-06,
+      "loss": 0.2798,
+      "step": 3710
+    },
+    {
+      "epoch": 0.5752614385401967,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.556923076923077e-06,
+      "loss": 0.2573,
+      "step": 3720
+    },
+    {
+      "epoch": 0.5768078402567027,
+      "grad_norm": 1.1015625,
+      "learning_rate": 8.526153846153847e-06,
+      "loss": 0.2756,
+      "step": 3730
+    },
+    {
+      "epoch": 0.5783542419732086,
+      "grad_norm": 1.03125,
+      "learning_rate": 8.495384615384616e-06,
+      "loss": 0.2819,
+      "step": 3740
+    },
+    {
+      "epoch": 0.5799006436897145,
+      "grad_norm": 0.8046875,
+      "learning_rate": 8.464615384615386e-06,
+      "loss": 0.22,
+      "step": 3750
+    },
+    {
+      "epoch": 0.5814470454062204,
+      "grad_norm": 1.0078125,
+      "learning_rate": 8.433846153846155e-06,
+      "loss": 0.2857,
+      "step": 3760
+    },
+    {
+      "epoch": 0.5829934471227263,
+      "grad_norm": 1.0703125,
+      "learning_rate": 8.403076923076924e-06,
+      "loss": 0.2803,
+      "step": 3770
+    },
+    {
+      "epoch": 0.5845398488392322,
+      "grad_norm": 1.0625,
+      "learning_rate": 8.372307692307692e-06,
+      "loss": 0.2207,
+      "step": 3780
+    },
+    {
+      "epoch": 0.5860862505557382,
+      "grad_norm": 1.375,
+      "learning_rate": 8.341538461538462e-06,
+      "loss": 0.2684,
+      "step": 3790
+    },
+    {
+      "epoch": 0.587632652272244,
+      "grad_norm": 0.7578125,
+      "learning_rate": 8.310769230769231e-06,
+      "loss": 0.2353,
+      "step": 3800
+    },
+    {
+      "epoch": 0.5891790539887499,
+      "grad_norm": 0.6484375,
+      "learning_rate": 8.28e-06,
+      "loss": 0.3244,
+      "step": 3810
+    },
+    {
+      "epoch": 0.5907254557052558,
+      "grad_norm": 1.15625,
+      "learning_rate": 8.24923076923077e-06,
+      "loss": 0.288,
+      "step": 3820
+    },
+    {
+      "epoch": 0.5922718574217617,
+      "grad_norm": 0.91015625,
+      "learning_rate": 8.218461538461539e-06,
+      "loss": 0.261,
+      "step": 3830
+    },
+    {
+      "epoch": 0.5938182591382677,
+      "grad_norm": 0.78125,
+      "learning_rate": 8.187692307692309e-06,
+      "loss": 0.3277,
+      "step": 3840
+    },
+    {
+      "epoch": 0.5953646608547736,
+      "grad_norm": 0.87890625,
+      "learning_rate": 8.156923076923078e-06,
+      "loss": 0.2727,
+      "step": 3850
+    },
+    {
+      "epoch": 0.5969110625712795,
+      "grad_norm": 0.80078125,
+      "learning_rate": 8.126153846153847e-06,
+      "loss": 0.2319,
+      "step": 3860
+    },
+    {
+      "epoch": 0.5984574642877853,
+      "grad_norm": 0.97265625,
+      "learning_rate": 8.095384615384617e-06,
+      "loss": 0.252,
+      "step": 3870
+    },
+    {
+      "epoch": 0.6000038660042912,
+      "grad_norm": 0.96875,
+      "learning_rate": 8.064615384615384e-06,
+      "loss": 0.2683,
+      "step": 3880
+    },
+    {
+      "epoch": 0.6015502677207971,
+      "grad_norm": 0.89453125,
+      "learning_rate": 8.033846153846154e-06,
+      "loss": 0.3251,
+      "step": 3890
+    },
+    {
+      "epoch": 0.6030966694373031,
+      "grad_norm": 1.015625,
+      "learning_rate": 8.003076923076923e-06,
+      "loss": 0.3153,
+      "step": 3900
+    },
+    {
+      "epoch": 0.604643071153809,
+      "grad_norm": 0.9296875,
+      "learning_rate": 7.972307692307693e-06,
+      "loss": 0.3234,
+      "step": 3910
+    },
+    {
+      "epoch": 0.6061894728703149,
+      "grad_norm": 0.88671875,
+      "learning_rate": 7.941538461538462e-06,
+      "loss": 0.2812,
+      "step": 3920
+    },
+    {
+      "epoch": 0.6077358745868208,
+      "grad_norm": 0.890625,
+      "learning_rate": 7.910769230769231e-06,
+      "loss": 0.2959,
+      "step": 3930
+    },
+    {
+      "epoch": 0.6092822763033267,
+      "grad_norm": 0.98046875,
+      "learning_rate": 7.88e-06,
+      "loss": 0.2818,
+      "step": 3940
+    },
+    {
+      "epoch": 0.6108286780198326,
+      "grad_norm": 1.15625,
+      "learning_rate": 7.84923076923077e-06,
+      "loss": 0.3152,
+      "step": 3950
+    },
+    {
+      "epoch": 0.6123750797363385,
+      "grad_norm": 0.9375,
+      "learning_rate": 7.81846153846154e-06,
+      "loss": 0.2988,
+      "step": 3960
+    },
+    {
+      "epoch": 0.6139214814528444,
+      "grad_norm": 0.984375,
+      "learning_rate": 7.787692307692309e-06,
+      "loss": 0.2835,
+      "step": 3970
+    },
+    {
+      "epoch": 0.6154678831693503,
+      "grad_norm": 1.359375,
+      "learning_rate": 7.756923076923077e-06,
+      "loss": 0.3486,
+      "step": 3980
+    },
+    {
+      "epoch": 0.6170142848858562,
+      "grad_norm": 1.3828125,
+      "learning_rate": 7.726153846153846e-06,
+      "loss": 0.2934,
+      "step": 3990
+    },
+    {
+      "epoch": 0.6185606866023621,
+      "grad_norm": 0.92578125,
+      "learning_rate": 7.695384615384615e-06,
+      "loss": 0.2678,
+      "step": 4000
+    },
+    {
+      "epoch": 0.6201070883188681,
+      "grad_norm": 0.90625,
+      "learning_rate": 7.664615384615385e-06,
+      "loss": 0.2608,
+      "step": 4010
+    },
+    {
+      "epoch": 0.621653490035374,
+      "grad_norm": 1.0703125,
+      "learning_rate": 7.633846153846154e-06,
+      "loss": 0.289,
+      "step": 4020
+    },
+    {
+      "epoch": 0.6231998917518798,
+      "grad_norm": 1.421875,
+      "learning_rate": 7.6030769230769245e-06,
+      "loss": 0.2856,
+      "step": 4030
+    },
+    {
+      "epoch": 0.6247462934683857,
+      "grad_norm": 0.67578125,
+      "learning_rate": 7.572307692307693e-06,
+      "loss": 0.2569,
+      "step": 4040
+    },
+    {
+      "epoch": 0.6262926951848916,
+      "grad_norm": 0.96484375,
+      "learning_rate": 7.5415384615384624e-06,
+      "loss": 0.2727,
+      "step": 4050
+    },
+    {
+      "epoch": 0.6278390969013976,
+      "grad_norm": 0.75390625,
+      "learning_rate": 7.510769230769232e-06,
+      "loss": 0.279,
+      "step": 4060
+    },
+    {
+      "epoch": 0.6293854986179035,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.48e-06,
+      "loss": 0.3606,
+      "step": 4070
+    },
+    {
+      "epoch": 0.6309319003344094,
+      "grad_norm": 0.7109375,
+      "learning_rate": 7.44923076923077e-06,
+      "loss": 0.2959,
+      "step": 4080
+    },
+    {
+      "epoch": 0.6324783020509153,
+      "grad_norm": 1.03125,
+      "learning_rate": 7.418461538461539e-06,
+      "loss": 0.2622,
+      "step": 4090
+    },
+    {
+      "epoch": 0.6340247037674211,
+      "grad_norm": 0.8046875,
+      "learning_rate": 7.387692307692308e-06,
+      "loss": 0.2207,
+      "step": 4100
+    },
+    {
+      "epoch": 0.635571105483927,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.356923076923077e-06,
+      "loss": 0.3007,
+      "step": 4110
+    },
+    {
+      "epoch": 0.637117507200433,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.326153846153847e-06,
+      "loss": 0.2815,
+      "step": 4120
+    },
+    {
+      "epoch": 0.6386639089169389,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.295384615384617e-06,
+      "loss": 0.2587,
+      "step": 4130
+    },
+    {
+      "epoch": 0.6402103106334448,
+      "grad_norm": 0.9453125,
+      "learning_rate": 7.264615384615385e-06,
+      "loss": 0.2999,
+      "step": 4140
+    },
+    {
+      "epoch": 0.6417567123499507,
+      "grad_norm": 0.90234375,
+      "learning_rate": 7.233846153846155e-06,
+      "loss": 0.2398,
+      "step": 4150
+    },
+    {
+      "epoch": 0.6433031140664566,
+      "grad_norm": 0.98046875,
+      "learning_rate": 7.203076923076924e-06,
+      "loss": 0.2716,
+      "step": 4160
+    },
+    {
+      "epoch": 0.6448495157829626,
+      "grad_norm": 1.0234375,
+      "learning_rate": 7.172307692307693e-06,
+      "loss": 0.2222,
+      "step": 4170
+    },
+    {
+      "epoch": 0.6463959174994685,
+      "grad_norm": 0.73828125,
+      "learning_rate": 7.141538461538462e-06,
+      "loss": 0.285,
+      "step": 4180
+    },
+    {
+      "epoch": 0.6479423192159743,
+      "grad_norm": 0.74609375,
+      "learning_rate": 7.1107692307692314e-06,
+      "loss": 0.3562,
+      "step": 4190
+    },
+    {
+      "epoch": 0.6494887209324802,
+      "grad_norm": 0.9921875,
+      "learning_rate": 7.08e-06,
+      "loss": 0.283,
+      "step": 4200
+    },
+    {
+      "epoch": 0.6510351226489861,
+      "grad_norm": 0.85546875,
+      "learning_rate": 7.049230769230769e-06,
+      "loss": 0.2915,
+      "step": 4210
+    },
+    {
+      "epoch": 0.652581524365492,
+      "grad_norm": 0.765625,
+      "learning_rate": 7.01846153846154e-06,
+      "loss": 0.2424,
+      "step": 4220
+    },
+    {
+      "epoch": 0.654127926081998,
+      "grad_norm": 0.81640625,
+      "learning_rate": 6.987692307692309e-06,
+      "loss": 0.2456,
+      "step": 4230
+    },
+    {
+      "epoch": 0.6556743277985039,
+      "grad_norm": 0.97265625,
+      "learning_rate": 6.9569230769230776e-06,
+      "loss": 0.2946,
+      "step": 4240
+    },
+    {
+      "epoch": 0.6572207295150098,
+      "grad_norm": 0.77734375,
+      "learning_rate": 6.926153846153847e-06,
+      "loss": 0.3338,
+      "step": 4250
+    },
+    {
+      "epoch": 0.6587671312315156,
+      "grad_norm": 1.2265625,
+      "learning_rate": 6.895384615384616e-06,
+      "loss": 0.2645,
+      "step": 4260
+    },
+    {
+      "epoch": 0.6603135329480215,
+      "grad_norm": 0.96484375,
+      "learning_rate": 6.864615384615385e-06,
+      "loss": 0.2671,
+      "step": 4270
+    },
+    {
+      "epoch": 0.6618599346645275,
+      "grad_norm": 1.09375,
+      "learning_rate": 6.833846153846154e-06,
+      "loss": 0.2627,
+      "step": 4280
+    },
+    {
+      "epoch": 0.6634063363810334,
+      "grad_norm": 1.2890625,
+      "learning_rate": 6.803076923076924e-06,
+      "loss": 0.2972,
+      "step": 4290
+    },
+    {
+      "epoch": 0.6649527380975393,
+      "grad_norm": 0.8359375,
+      "learning_rate": 6.772307692307692e-06,
+      "loss": 0.2637,
+      "step": 4300
+    },
+    {
+      "epoch": 0.6664991398140452,
+      "grad_norm": 0.87109375,
+      "learning_rate": 6.741538461538462e-06,
+      "loss": 0.2459,
+      "step": 4310
+    },
+    {
+      "epoch": 0.6680455415305511,
+      "grad_norm": 0.93359375,
+      "learning_rate": 6.710769230769232e-06,
+      "loss": 0.3008,
+      "step": 4320
+    },
+    {
+      "epoch": 0.669591943247057,
+      "grad_norm": 1.109375,
+      "learning_rate": 6.680000000000001e-06,
+      "loss": 0.2881,
+      "step": 4330
+    },
+    {
+      "epoch": 0.671138344963563,
+      "grad_norm": 0.67578125,
+      "learning_rate": 6.64923076923077e-06,
+      "loss": 0.2887,
+      "step": 4340
+    },
+    {
+      "epoch": 0.6726847466800688,
+      "grad_norm": 0.875,
+      "learning_rate": 6.618461538461539e-06,
+      "loss": 0.3097,
+      "step": 4350
+    },
+    {
+      "epoch": 0.6742311483965747,
+      "grad_norm": 0.8671875,
+      "learning_rate": 6.587692307692309e-06,
+      "loss": 0.2623,
+      "step": 4360
+    },
+    {
+      "epoch": 0.6757775501130806,
+      "grad_norm": 0.8984375,
+      "learning_rate": 6.556923076923077e-06,
+      "loss": 0.2589,
+      "step": 4370
+    },
+    {
+      "epoch": 0.6773239518295865,
+      "grad_norm": 0.77734375,
+      "learning_rate": 6.5261538461538465e-06,
+      "loss": 0.2149,
+      "step": 4380
+    },
+    {
+      "epoch": 0.6788703535460925,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.495384615384616e-06,
+      "loss": 0.22,
+      "step": 4390
+    },
+    {
+      "epoch": 0.6804167552625984,
+      "grad_norm": 0.98828125,
+      "learning_rate": 6.4646153846153845e-06,
+      "loss": 0.2636,
+      "step": 4400
+    },
+    {
+      "epoch": 0.6819631569791043,
+      "grad_norm": 1.0234375,
+      "learning_rate": 6.433846153846154e-06,
+      "loss": 0.3088,
+      "step": 4410
+    },
+    {
+      "epoch": 0.6835095586956101,
+      "grad_norm": 0.92578125,
+      "learning_rate": 6.403076923076924e-06,
+      "loss": 0.2378,
+      "step": 4420
+    },
+    {
+      "epoch": 0.685055960412116,
+      "grad_norm": 0.91796875,
+      "learning_rate": 6.3723076923076935e-06,
+      "loss": 0.2405,
+      "step": 4430
+    },
+    {
+      "epoch": 0.6866023621286219,
+      "grad_norm": 0.83203125,
+      "learning_rate": 6.341538461538462e-06,
+      "loss": 0.2825,
+      "step": 4440
+    },
+    {
+      "epoch": 0.6881487638451279,
+      "grad_norm": 0.9296875,
+      "learning_rate": 6.3107692307692315e-06,
+      "loss": 0.3513,
+      "step": 4450
+    },
+    {
+      "epoch": 0.6896951655616338,
+      "grad_norm": 1.078125,
+      "learning_rate": 6.280000000000001e-06,
+      "loss": 0.3271,
+      "step": 4460
+    },
+    {
+      "epoch": 0.6912415672781397,
+      "grad_norm": 0.82421875,
+      "learning_rate": 6.249230769230769e-06,
+      "loss": 0.2742,
+      "step": 4470
+    },
+    {
+      "epoch": 0.6927879689946456,
+      "grad_norm": 0.9921875,
+      "learning_rate": 6.218461538461539e-06,
+      "loss": 0.2642,
+      "step": 4480
+    },
+    {
+      "epoch": 0.6943343707111514,
+      "grad_norm": 0.96875,
+      "learning_rate": 6.187692307692308e-06,
+      "loss": 0.2364,
+      "step": 4490
+    },
+    {
+      "epoch": 0.6958807724276574,
+      "grad_norm": 1.1875,
+      "learning_rate": 6.156923076923077e-06,
+      "loss": 0.2883,
+      "step": 4500
+    },
+    {
+      "epoch": 0.6974271741441633,
+      "grad_norm": 1.03125,
+      "learning_rate": 6.126153846153846e-06,
+      "loss": 0.2219,
+      "step": 4510
+    },
+    {
+      "epoch": 0.6989735758606692,
+      "grad_norm": 0.9453125,
+      "learning_rate": 6.095384615384616e-06,
+      "loss": 0.2639,
+      "step": 4520
+    },
+    {
+      "epoch": 0.7005199775771751,
+      "grad_norm": 0.80859375,
+      "learning_rate": 6.064615384615386e-06,
+      "loss": 0.2764,
+      "step": 4530
+    },
+    {
+      "epoch": 0.702066379293681,
+      "grad_norm": 1.0390625,
+      "learning_rate": 6.033846153846154e-06,
+      "loss": 0.3097,
+      "step": 4540
+    },
+    {
+      "epoch": 0.7036127810101869,
+      "grad_norm": 0.890625,
+      "learning_rate": 6.003076923076924e-06,
+      "loss": 0.2262,
+      "step": 4550
+    },
+    {
+      "epoch": 0.7051591827266929,
+      "grad_norm": 0.82421875,
+      "learning_rate": 5.972307692307693e-06,
+      "loss": 0.2537,
+      "step": 4560
+    },
+    {
+      "epoch": 0.7067055844431988,
+      "grad_norm": 0.7734375,
+      "learning_rate": 5.941538461538462e-06,
+      "loss": 0.347,
+      "step": 4570
+    },
+    {
+      "epoch": 0.7082519861597046,
+      "grad_norm": 1.4140625,
+      "learning_rate": 5.910769230769231e-06,
+      "loss": 0.3193,
+      "step": 4580
+    },
+    {
+      "epoch": 0.7097983878762105,
+      "grad_norm": 0.75,
+      "learning_rate": 5.8800000000000005e-06,
+      "loss": 0.273,
+      "step": 4590
+    },
+    {
+      "epoch": 0.7113447895927164,
+      "grad_norm": 0.921875,
+      "learning_rate": 5.849230769230769e-06,
+      "loss": 0.2902,
+      "step": 4600
+    },
+    {
+      "epoch": 0.7128911913092224,
+      "grad_norm": 0.8046875,
+      "learning_rate": 5.818461538461538e-06,
+      "loss": 0.3653,
+      "step": 4610
+    },
+    {
+      "epoch": 0.7144375930257283,
+      "grad_norm": 0.76171875,
+      "learning_rate": 5.787692307692309e-06,
+      "loss": 0.3106,
+      "step": 4620
+    },
+    {
+      "epoch": 0.7159839947422342,
+      "grad_norm": 0.953125,
+      "learning_rate": 5.756923076923078e-06,
+      "loss": 0.2368,
+      "step": 4630
+    },
+    {
+      "epoch": 0.7175303964587401,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.726153846153847e-06,
+      "loss": 0.249,
+      "step": 4640
+    },
+    {
+      "epoch": 0.719076798175246,
+      "grad_norm": 1.1328125,
+      "learning_rate": 5.695384615384616e-06,
+      "loss": 0.3709,
+      "step": 4650
+    },
+    {
+      "epoch": 0.7206231998917518,
+      "grad_norm": 0.88671875,
+      "learning_rate": 5.664615384615385e-06,
+      "loss": 0.2921,
+      "step": 4660
+    },
+    {
+      "epoch": 0.7221696016082578,
+      "grad_norm": 1.015625,
+      "learning_rate": 5.633846153846154e-06,
+      "loss": 0.3115,
+      "step": 4670
+    },
+    {
+      "epoch": 0.7237160033247637,
+      "grad_norm": 0.87890625,
+      "learning_rate": 5.603076923076923e-06,
+      "loss": 0.2479,
+      "step": 4680
+    },
+    {
+      "epoch": 0.7252624050412696,
+      "grad_norm": 1.046875,
+      "learning_rate": 5.572307692307693e-06,
+      "loss": 0.2297,
+      "step": 4690
+    },
+    {
+      "epoch": 0.7268088067577755,
+      "grad_norm": 0.89453125,
+      "learning_rate": 5.541538461538461e-06,
+      "loss": 0.2454,
+      "step": 4700
+    },
+    {
+      "epoch": 0.7283552084742814,
+      "grad_norm": 0.79296875,
+      "learning_rate": 5.5107692307692315e-06,
+      "loss": 0.2849,
+      "step": 4710
+    },
+    {
+      "epoch": 0.7299016101907874,
+      "grad_norm": 0.74609375,
+      "learning_rate": 5.480000000000001e-06,
+      "loss": 0.2797,
+      "step": 4720
+    },
+    {
+      "epoch": 0.7314480119072932,
+      "grad_norm": 1.0234375,
+      "learning_rate": 5.44923076923077e-06,
+      "loss": 0.3882,
+      "step": 4730
+    },
+    {
+      "epoch": 0.7329944136237991,
+      "grad_norm": 0.90234375,
+      "learning_rate": 5.418461538461539e-06,
+      "loss": 0.2509,
+      "step": 4740
+    },
+    {
+      "epoch": 0.734540815340305,
+      "grad_norm": 0.81640625,
+      "learning_rate": 5.387692307692308e-06,
+      "loss": 0.2408,
+      "step": 4750
+    },
+    {
+      "epoch": 0.7360872170568109,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5.356923076923078e-06,
+      "loss": 0.2413,
+      "step": 4760
+    },
+    {
+      "epoch": 0.7376336187733168,
+      "grad_norm": 0.7890625,
+      "learning_rate": 5.326153846153846e-06,
+      "loss": 0.2432,
+      "step": 4770
+    },
+    {
+      "epoch": 0.7391800204898228,
+      "grad_norm": 0.9921875,
+      "learning_rate": 5.2953846153846156e-06,
+      "loss": 0.277,
+      "step": 4780
+    },
+    {
+      "epoch": 0.7407264222063287,
+      "grad_norm": 1.15625,
+      "learning_rate": 5.264615384615385e-06,
+      "loss": 0.2486,
+      "step": 4790
+    },
+    {
+      "epoch": 0.7422728239228346,
+      "grad_norm": 0.8203125,
+      "learning_rate": 5.2338461538461535e-06,
+      "loss": 0.3,
+      "step": 4800
+    },
+    {
+      "epoch": 0.7438192256393404,
+      "grad_norm": 0.87109375,
+      "learning_rate": 5.203076923076924e-06,
+      "loss": 0.26,
+      "step": 4810
+    },
+    {
+      "epoch": 0.7453656273558463,
+      "grad_norm": 1.328125,
+      "learning_rate": 5.172307692307693e-06,
+      "loss": 0.2937,
+      "step": 4820
+    },
+    {
+      "epoch": 0.7469120290723523,
+      "grad_norm": 0.9453125,
+      "learning_rate": 5.1415384615384625e-06,
+      "loss": 0.3057,
+      "step": 4830
+    },
+    {
+      "epoch": 0.7484584307888582,
+      "grad_norm": 0.8828125,
+      "learning_rate": 5.110769230769231e-06,
+      "loss": 0.3284,
+      "step": 4840
+    },
+    {
+      "epoch": 0.7500048325053641,
+      "grad_norm": 0.98046875,
+      "learning_rate": 5.0800000000000005e-06,
+      "loss": 0.2434,
+      "step": 4850
+    },
+    {
+      "epoch": 0.75155123422187,
+      "grad_norm": 0.97265625,
+      "learning_rate": 5.04923076923077e-06,
+      "loss": 0.257,
+      "step": 4860
+    },
+    {
+      "epoch": 0.7530976359383759,
+      "grad_norm": 0.91015625,
+      "learning_rate": 5.0184615384615384e-06,
+      "loss": 0.2714,
+      "step": 4870
+    },
+    {
+      "epoch": 0.7546440376548817,
+      "grad_norm": 0.609375,
+      "learning_rate": 4.987692307692308e-06,
+      "loss": 0.2182,
+      "step": 4880
+    },
+    {
+      "epoch": 0.7561904393713877,
+      "grad_norm": 0.85546875,
+      "learning_rate": 4.956923076923077e-06,
+      "loss": 0.2855,
+      "step": 4890
+    },
+    {
+      "epoch": 0.7577368410878936,
+      "grad_norm": 1.0703125,
+      "learning_rate": 4.926153846153847e-06,
+      "loss": 0.2774,
+      "step": 4900
+    },
+    {
+      "epoch": 0.7592832428043995,
+      "grad_norm": 0.86328125,
+      "learning_rate": 4.895384615384616e-06,
+      "loss": 0.2489,
+      "step": 4910
+    },
+    {
+      "epoch": 0.7608296445209054,
+      "grad_norm": 1.1328125,
+      "learning_rate": 4.8646153846153846e-06,
+      "loss": 0.3157,
+      "step": 4920
+    },
+    {
+      "epoch": 0.7623760462374113,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.833846153846154e-06,
+      "loss": 0.2704,
+      "step": 4930
+    },
+    {
+      "epoch": 0.7639224479539173,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4.803076923076923e-06,
+      "loss": 0.2995,
+      "step": 4940
+    },
+    {
+      "epoch": 0.7654688496704232,
+      "grad_norm": 0.98046875,
+      "learning_rate": 4.772307692307693e-06,
+      "loss": 0.2422,
+      "step": 4950
+    },
+    {
+      "epoch": 0.767015251386929,
+      "grad_norm": 1.296875,
+      "learning_rate": 4.741538461538462e-06,
+      "loss": 0.2692,
+      "step": 4960
+    },
+    {
+      "epoch": 0.7685616531034349,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.710769230769231e-06,
+      "loss": 0.2704,
+      "step": 4970
+    },
+    {
+      "epoch": 0.7701080548199408,
+      "grad_norm": 0.85546875,
+      "learning_rate": 4.680000000000001e-06,
+      "loss": 0.3147,
+      "step": 4980
+    },
+    {
+      "epoch": 0.7716544565364467,
+      "grad_norm": 0.9609375,
+      "learning_rate": 4.6492307692307695e-06,
+      "loss": 0.2867,
+      "step": 4990
+    },
+    {
+      "epoch": 0.7732008582529527,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.618461538461539e-06,
+      "loss": 0.2896,
+      "step": 5000
+    },
+    {
+      "epoch": 0.7747472599694586,
+      "grad_norm": 1.0546875,
+      "learning_rate": 4.587692307692308e-06,
+      "loss": 0.2335,
+      "step": 5010
+    },
+    {
+      "epoch": 0.7762936616859645,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.556923076923077e-06,
+      "loss": 0.2441,
+      "step": 5020
+    },
+    {
+      "epoch": 0.7778400634024704,
+      "grad_norm": 1.4453125,
+      "learning_rate": 4.526153846153847e-06,
+      "loss": 0.3049,
+      "step": 5030
+    },
+    {
+      "epoch": 0.7793864651189762,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.495384615384616e-06,
+      "loss": 0.2605,
+      "step": 5040
+    },
+    {
+      "epoch": 0.7809328668354822,
+      "grad_norm": 1.2265625,
+      "learning_rate": 4.464615384615385e-06,
+      "loss": 0.2876,
+      "step": 5050
+    },
+    {
+      "epoch": 0.7824792685519881,
+      "grad_norm": 1.09375,
+      "learning_rate": 4.433846153846154e-06,
+      "loss": 0.3434,
+      "step": 5060
+    },
+    {
+      "epoch": 0.784025670268494,
+      "grad_norm": 1.046875,
+      "learning_rate": 4.403076923076923e-06,
+      "loss": 0.2956,
+      "step": 5070
+    },
+    {
+      "epoch": 0.7855720719849999,
+      "grad_norm": 1.0078125,
+      "learning_rate": 4.372307692307693e-06,
+      "loss": 0.3175,
+      "step": 5080
+    },
+    {
+      "epoch": 0.7871184737015058,
+      "grad_norm": 1.1015625,
+      "learning_rate": 4.341538461538462e-06,
+      "loss": 0.2914,
+      "step": 5090
+    },
+    {
+      "epoch": 0.7886648754180117,
+      "grad_norm": 1.125,
+      "learning_rate": 4.310769230769231e-06,
+      "loss": 0.2657,
+      "step": 5100
+    },
+    {
+      "epoch": 0.7902112771345177,
+      "grad_norm": 0.9453125,
+      "learning_rate": 4.2800000000000005e-06,
+      "loss": 0.3168,
+      "step": 5110
+    },
+    {
+      "epoch": 0.7917576788510235,
+      "grad_norm": 1.0390625,
+      "learning_rate": 4.249230769230769e-06,
+      "loss": 0.2422,
+      "step": 5120
+    },
+    {
+      "epoch": 0.7933040805675294,
+      "grad_norm": 0.8125,
+      "learning_rate": 4.218461538461539e-06,
+      "loss": 0.2651,
+      "step": 5130
+    },
+    {
+      "epoch": 0.7948504822840353,
+      "grad_norm": 0.984375,
+      "learning_rate": 4.187692307692308e-06,
+      "loss": 0.245,
+      "step": 5140
+    },
+    {
+      "epoch": 0.7963968840005412,
+      "grad_norm": 0.82421875,
+      "learning_rate": 4.156923076923077e-06,
+      "loss": 0.3055,
+      "step": 5150
+    },
+    {
+      "epoch": 0.7979432857170471,
+      "grad_norm": 1.015625,
+      "learning_rate": 4.126153846153847e-06,
+      "loss": 0.2992,
+      "step": 5160
+    },
+    {
+      "epoch": 0.7994896874335531,
+      "grad_norm": 0.796875,
+      "learning_rate": 4.095384615384615e-06,
+      "loss": 0.3123,
+      "step": 5170
+    },
+    {
+      "epoch": 0.801036089150059,
+      "grad_norm": 1.1796875,
+      "learning_rate": 4.0646153846153854e-06,
+      "loss": 0.2849,
+      "step": 5180
+    },
+    {
+      "epoch": 0.8025824908665649,
+      "grad_norm": 0.84765625,
+      "learning_rate": 4.033846153846154e-06,
+      "loss": 0.317,
+      "step": 5190
+    },
+    {
+      "epoch": 0.8041288925830707,
+      "grad_norm": 0.88671875,
+      "learning_rate": 4.003076923076923e-06,
+      "loss": 0.2567,
+      "step": 5200
+    },
+    {
+      "epoch": 0.8056752942995766,
+      "grad_norm": 1.109375,
+      "learning_rate": 3.972307692307693e-06,
+      "loss": 0.2918,
+      "step": 5210
+    },
+    {
+      "epoch": 0.8072216960160826,
+      "grad_norm": 0.9765625,
+      "learning_rate": 3.941538461538461e-06,
+      "loss": 0.3973,
+      "step": 5220
+    },
+    {
+      "epoch": 0.8087680977325885,
+      "grad_norm": 1.1015625,
+      "learning_rate": 3.9107692307692316e-06,
+      "loss": 0.3034,
+      "step": 5230
+    },
+    {
+      "epoch": 0.8103144994490944,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.88e-06,
+      "loss": 0.2369,
+      "step": 5240
+    },
+    {
+      "epoch": 0.8118609011656003,
+      "grad_norm": 0.8125,
+      "learning_rate": 3.8492307692307695e-06,
+      "loss": 0.261,
+      "step": 5250
+    },
+    {
+      "epoch": 0.8134073028821062,
+      "grad_norm": 1.015625,
+      "learning_rate": 3.818461538461539e-06,
+      "loss": 0.2657,
+      "step": 5260
+    },
+    {
+      "epoch": 0.814953704598612,
+      "grad_norm": 0.91796875,
+      "learning_rate": 3.787692307692308e-06,
+      "loss": 0.2336,
+      "step": 5270
+    },
+    {
+      "epoch": 0.816500106315118,
+      "grad_norm": 0.83984375,
+      "learning_rate": 3.7569230769230773e-06,
+      "loss": 0.2683,
+      "step": 5280
+    },
+    {
+      "epoch": 0.8180465080316239,
+      "grad_norm": 1.09375,
+      "learning_rate": 3.7261538461538467e-06,
+      "loss": 0.2703,
+      "step": 5290
+    },
+    {
+      "epoch": 0.8195929097481298,
+      "grad_norm": 1.28125,
+      "learning_rate": 3.6953846153846156e-06,
+      "loss": 0.2907,
+      "step": 5300
+    },
+    {
+      "epoch": 0.8211393114646357,
+      "grad_norm": 0.95703125,
+      "learning_rate": 3.6646153846153846e-06,
+      "loss": 0.3177,
+      "step": 5310
+    },
+    {
+      "epoch": 0.8226857131811416,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.633846153846154e-06,
+      "loss": 0.3023,
+      "step": 5320
+    },
+    {
+      "epoch": 0.8242321148976476,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.6030769230769234e-06,
+      "loss": 0.2169,
+      "step": 5330
+    },
+    {
+      "epoch": 0.8257785166141535,
+      "grad_norm": 0.7890625,
+      "learning_rate": 3.572307692307693e-06,
+      "loss": 0.24,
+      "step": 5340
+    },
+    {
+      "epoch": 0.8273249183306594,
+      "grad_norm": 1.0546875,
+      "learning_rate": 3.5415384615384618e-06,
+      "loss": 0.3394,
+      "step": 5350
+    },
+    {
+      "epoch": 0.8288713200471652,
+      "grad_norm": 1.1875,
+      "learning_rate": 3.5107692307692307e-06,
+      "loss": 0.2527,
+      "step": 5360
+    },
+    {
+      "epoch": 0.8304177217636711,
+      "grad_norm": 0.69921875,
+      "learning_rate": 3.48e-06,
+      "loss": 0.2447,
+      "step": 5370
+    },
+    {
+      "epoch": 0.831964123480177,
+      "grad_norm": 1.0703125,
+      "learning_rate": 3.4492307692307695e-06,
+      "loss": 0.2509,
+      "step": 5380
+    },
+    {
+      "epoch": 0.833510525196683,
+      "grad_norm": 0.80859375,
+      "learning_rate": 3.418461538461539e-06,
+      "loss": 0.3633,
+      "step": 5390
+    },
+    {
+      "epoch": 0.8350569269131889,
+      "grad_norm": 0.69921875,
+      "learning_rate": 3.387692307692308e-06,
+      "loss": 0.3206,
+      "step": 5400
+    },
+    {
+      "epoch": 0.8366033286296948,
+      "grad_norm": 1.5078125,
+      "learning_rate": 3.356923076923077e-06,
+      "loss": 0.3542,
+      "step": 5410
+    },
+    {
+      "epoch": 0.8381497303462007,
+      "grad_norm": 1.0234375,
+      "learning_rate": 3.3261538461538463e-06,
+      "loss": 0.2731,
+      "step": 5420
+    },
+    {
+      "epoch": 0.8396961320627065,
+      "grad_norm": 0.92578125,
+      "learning_rate": 3.2953846153846157e-06,
+      "loss": 0.3256,
+      "step": 5430
+    },
+    {
+      "epoch": 0.8412425337792125,
+      "grad_norm": 0.8359375,
+      "learning_rate": 3.264615384615385e-06,
+      "loss": 0.2471,
+      "step": 5440
+    },
+    {
+      "epoch": 0.8427889354957184,
+      "grad_norm": 0.9453125,
+      "learning_rate": 3.233846153846154e-06,
+      "loss": 0.2755,
+      "step": 5450
+    },
+    {
+      "epoch": 0.8443353372122243,
+      "grad_norm": 1.046875,
+      "learning_rate": 3.203076923076923e-06,
+      "loss": 0.3139,
+      "step": 5460
+    },
+    {
+      "epoch": 0.8458817389287302,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.1723076923076924e-06,
+      "loss": 0.2722,
+      "step": 5470
+    },
+    {
+      "epoch": 0.8474281406452361,
+      "grad_norm": 0.828125,
+      "learning_rate": 3.141538461538462e-06,
+      "loss": 0.3058,
+      "step": 5480
+    },
+    {
+      "epoch": 0.848974542361742,
+      "grad_norm": 1.03125,
+      "learning_rate": 3.110769230769231e-06,
+      "loss": 0.2424,
+      "step": 5490
+    },
+    {
+      "epoch": 0.850520944078248,
+      "grad_norm": 0.98828125,
+      "learning_rate": 3.08e-06,
+      "loss": 0.2752,
+      "step": 5500
+    },
+    {
+      "epoch": 0.8520673457947538,
+      "grad_norm": 0.6640625,
+      "learning_rate": 3.049230769230769e-06,
+      "loss": 0.2309,
+      "step": 5510
+    },
+    {
+      "epoch": 0.8536137475112597,
+      "grad_norm": 0.85546875,
+      "learning_rate": 3.0184615384615385e-06,
+      "loss": 0.33,
+      "step": 5520
+    },
+    {
+      "epoch": 0.8551601492277656,
+      "grad_norm": 0.765625,
+      "learning_rate": 2.987692307692308e-06,
+      "loss": 0.2942,
+      "step": 5530
+    },
+    {
+      "epoch": 0.8567065509442715,
+      "grad_norm": 0.76171875,
+      "learning_rate": 2.9569230769230773e-06,
+      "loss": 0.3103,
+      "step": 5540
+    },
+    {
+      "epoch": 0.8582529526607775,
+      "grad_norm": 0.91796875,
+      "learning_rate": 2.9261538461538463e-06,
+      "loss": 0.2775,
+      "step": 5550
+    },
+    {
+      "epoch": 0.8597993543772834,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.8953846153846153e-06,
+      "loss": 0.2941,
+      "step": 5560
+    },
+    {
+      "epoch": 0.8613457560937893,
+      "grad_norm": 0.72265625,
+      "learning_rate": 2.8646153846153847e-06,
+      "loss": 0.2591,
+      "step": 5570
+    },
+    {
+      "epoch": 0.8628921578102952,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.833846153846154e-06,
+      "loss": 0.2801,
+      "step": 5580
+    },
+    {
+      "epoch": 0.864438559526801,
+      "grad_norm": 0.859375,
+      "learning_rate": 2.8030769230769234e-06,
+      "loss": 0.3041,
+      "step": 5590
+    },
+    {
+      "epoch": 0.8659849612433069,
+      "grad_norm": 1.2265625,
+      "learning_rate": 2.7723076923076924e-06,
+      "loss": 0.2866,
+      "step": 5600
+    },
+    {
+      "epoch": 0.8675313629598129,
+      "grad_norm": 0.99609375,
+      "learning_rate": 2.7415384615384614e-06,
+      "loss": 0.3128,
+      "step": 5610
+    },
+    {
+      "epoch": 0.8690777646763188,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.710769230769231e-06,
+      "loss": 0.3121,
+      "step": 5620
+    },
+    {
+      "epoch": 0.8706241663928247,
+      "grad_norm": 0.7578125,
+      "learning_rate": 2.68e-06,
+      "loss": 0.2264,
+      "step": 5630
+    },
+    {
+      "epoch": 0.8721705681093306,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.6492307692307696e-06,
+      "loss": 0.2619,
+      "step": 5640
+    },
+    {
+      "epoch": 0.8737169698258365,
+      "grad_norm": 0.70703125,
+      "learning_rate": 2.6184615384615385e-06,
+      "loss": 0.2631,
+      "step": 5650
+    },
+    {
+      "epoch": 0.8752633715423425,
+      "grad_norm": 0.9765625,
+      "learning_rate": 2.587692307692308e-06,
+      "loss": 0.2636,
+      "step": 5660
+    },
+    {
+      "epoch": 0.8768097732588483,
+      "grad_norm": 1.03125,
+      "learning_rate": 2.5569230769230773e-06,
+      "loss": 0.3569,
+      "step": 5670
+    },
+    {
+      "epoch": 0.8783561749753542,
+      "grad_norm": 0.76953125,
+      "learning_rate": 2.5261538461538463e-06,
+      "loss": 0.2297,
+      "step": 5680
+    },
+    {
+      "epoch": 0.8799025766918601,
+      "grad_norm": 0.89453125,
+      "learning_rate": 2.4953846153846157e-06,
+      "loss": 0.2181,
+      "step": 5690
+    },
+    {
+      "epoch": 0.881448978408366,
+      "grad_norm": 1.3359375,
+      "learning_rate": 2.4646153846153847e-06,
+      "loss": 0.3117,
+      "step": 5700
+    },
+    {
+      "epoch": 0.8829953801248719,
+      "grad_norm": 0.9296875,
+      "learning_rate": 2.433846153846154e-06,
+      "loss": 0.3071,
+      "step": 5710
+    },
+    {
+      "epoch": 0.8845417818413779,
+      "grad_norm": 0.828125,
+      "learning_rate": 2.4030769230769235e-06,
+      "loss": 0.2599,
+      "step": 5720
+    },
+    {
+      "epoch": 0.8860881835578838,
+      "grad_norm": 1.0234375,
+      "learning_rate": 2.3723076923076924e-06,
+      "loss": 0.265,
+      "step": 5730
+    },
+    {
+      "epoch": 0.8876345852743897,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.341538461538462e-06,
+      "loss": 0.2922,
+      "step": 5740
+    },
+    {
+      "epoch": 0.8891809869908955,
+      "grad_norm": 0.984375,
+      "learning_rate": 2.310769230769231e-06,
+      "loss": 0.3616,
+      "step": 5750
+    },
+    {
+      "epoch": 0.8907273887074014,
+      "grad_norm": 1.2578125,
+      "learning_rate": 2.28e-06,
+      "loss": 0.2587,
+      "step": 5760
+    },
+    {
+      "epoch": 0.8922737904239074,
+      "grad_norm": 0.7890625,
+      "learning_rate": 2.2492307692307696e-06,
+      "loss": 0.335,
+      "step": 5770
+    },
+    {
+      "epoch": 0.8938201921404133,
+      "grad_norm": 0.9375,
+      "learning_rate": 2.218461538461539e-06,
+      "loss": 0.288,
+      "step": 5780
+    },
+    {
+      "epoch": 0.8953665938569192,
+      "grad_norm": 0.921875,
+      "learning_rate": 2.187692307692308e-06,
+      "loss": 0.2932,
+      "step": 5790
+    },
+    {
+      "epoch": 0.8969129955734251,
+      "grad_norm": 1.109375,
+      "learning_rate": 2.156923076923077e-06,
+      "loss": 0.282,
+      "step": 5800
+    },
+    {
+      "epoch": 0.898459397289931,
+      "grad_norm": 0.8671875,
+      "learning_rate": 2.1261538461538463e-06,
+      "loss": 0.2073,
+      "step": 5810
+    },
+    {
+      "epoch": 0.9000057990064368,
+      "grad_norm": 0.87890625,
+      "learning_rate": 2.0953846153846157e-06,
+      "loss": 0.2583,
+      "step": 5820
+    },
+    {
+      "epoch": 0.9015522007229428,
+      "grad_norm": 1.0390625,
+      "learning_rate": 2.064615384615385e-06,
+      "loss": 0.2805,
+      "step": 5830
+    },
+    {
+      "epoch": 0.9030986024394487,
+      "grad_norm": 0.828125,
+      "learning_rate": 2.033846153846154e-06,
+      "loss": 0.2416,
+      "step": 5840
+    },
+    {
+      "epoch": 0.9046450041559546,
+      "grad_norm": 0.7890625,
+      "learning_rate": 2.003076923076923e-06,
+      "loss": 0.2826,
+      "step": 5850
+    },
+    {
+      "epoch": 0.9061914058724605,
+      "grad_norm": 0.9375,
+      "learning_rate": 1.9723076923076924e-06,
+      "loss": 0.3072,
+      "step": 5860
+    },
+    {
+      "epoch": 0.9077378075889664,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.941538461538462e-06,
+      "loss": 0.3551,
+      "step": 5870
+    },
+    {
+      "epoch": 0.9092842093054724,
+      "grad_norm": 0.84765625,
+      "learning_rate": 1.9107692307692312e-06,
+      "loss": 0.3224,
+      "step": 5880
+    },
+    {
+      "epoch": 0.9108306110219783,
+      "grad_norm": 0.890625,
+      "learning_rate": 1.8800000000000002e-06,
+      "loss": 0.2501,
+      "step": 5890
+    },
+    {
+      "epoch": 0.9123770127384841,
+      "grad_norm": 0.671875,
+      "learning_rate": 1.8492307692307692e-06,
+      "loss": 0.2555,
+      "step": 5900
+    },
+    {
+      "epoch": 0.91392341445499,
+      "grad_norm": 0.8671875,
+      "learning_rate": 1.8184615384615386e-06,
+      "loss": 0.329,
+      "step": 5910
+    },
+    {
+      "epoch": 0.9154698161714959,
+      "grad_norm": 0.953125,
+      "learning_rate": 1.7876923076923078e-06,
+      "loss": 0.3193,
+      "step": 5920
+    },
+    {
+      "epoch": 0.9170162178880018,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.7569230769230772e-06,
+      "loss": 0.3162,
+      "step": 5930
+    },
+    {
+      "epoch": 0.9185626196045078,
+      "grad_norm": 0.87109375,
+      "learning_rate": 1.7261538461538463e-06,
+      "loss": 0.29,
+      "step": 5940
+    },
+    {
+      "epoch": 0.9201090213210137,
+      "grad_norm": 0.859375,
+      "learning_rate": 1.6953846153846153e-06,
+      "loss": 0.3122,
+      "step": 5950
+    },
+    {
+      "epoch": 0.9216554230375196,
+      "grad_norm": 0.75,
+      "learning_rate": 1.6646153846153847e-06,
+      "loss": 0.2374,
+      "step": 5960
+    },
+    {
+      "epoch": 0.9232018247540255,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.6338461538461539e-06,
+      "loss": 0.2562,
+      "step": 5970
+    },
+    {
+      "epoch": 0.9247482264705313,
+      "grad_norm": 1.0859375,
+      "learning_rate": 1.6030769230769233e-06,
+      "loss": 0.2854,
+      "step": 5980
+    },
+    {
+      "epoch": 0.9262946281870373,
+      "grad_norm": 1.1015625,
+      "learning_rate": 1.5723076923076925e-06,
+      "loss": 0.3549,
+      "step": 5990
+    },
+    {
+      "epoch": 0.9278410299035432,
+      "grad_norm": 1.1953125,
+      "learning_rate": 1.5415384615384614e-06,
+      "loss": 0.4152,
+      "step": 6000
+    },
+    {
+      "epoch": 0.9293874316200491,
+      "grad_norm": 1.0546875,
+      "learning_rate": 1.5107692307692308e-06,
+      "loss": 0.2626,
+      "step": 6010
+    },
+    {
+      "epoch": 0.930933833336555,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.48e-06,
+      "loss": 0.3302,
+      "step": 6020
+    },
+    {
+      "epoch": 0.9324802350530609,
+      "grad_norm": 1.0390625,
+      "learning_rate": 1.4492307692307694e-06,
+      "loss": 0.3063,
+      "step": 6030
+    },
+    {
+      "epoch": 0.9340266367695668,
+      "grad_norm": 1.2265625,
+      "learning_rate": 1.4184615384615386e-06,
+      "loss": 0.2758,
+      "step": 6040
+    },
+    {
+      "epoch": 0.9355730384860728,
+      "grad_norm": 1.015625,
+      "learning_rate": 1.3876923076923076e-06,
+      "loss": 0.2999,
+      "step": 6050
+    },
+    {
+      "epoch": 0.9371194402025786,
+      "grad_norm": 1.0234375,
+      "learning_rate": 1.356923076923077e-06,
+      "loss": 0.3438,
+      "step": 6060
+    },
+    {
+      "epoch": 0.9386658419190845,
+      "grad_norm": 0.70703125,
+      "learning_rate": 1.3261538461538461e-06,
+      "loss": 0.2368,
+      "step": 6070
+    },
+    {
+      "epoch": 0.9402122436355904,
+      "grad_norm": 0.80859375,
+      "learning_rate": 1.2953846153846155e-06,
+      "loss": 0.2524,
+      "step": 6080
+    },
+    {
+      "epoch": 0.9417586453520963,
+      "grad_norm": 0.74609375,
+      "learning_rate": 1.2646153846153847e-06,
+      "loss": 0.2264,
+      "step": 6090
+    },
+    {
+      "epoch": 0.9433050470686023,
+      "grad_norm": 0.75390625,
+      "learning_rate": 1.233846153846154e-06,
+      "loss": 0.2067,
+      "step": 6100
+    },
+    {
+      "epoch": 0.9448514487851082,
+      "grad_norm": 0.81640625,
+      "learning_rate": 1.2030769230769233e-06,
+      "loss": 0.231,
+      "step": 6110
+    },
+    {
+      "epoch": 0.9463978505016141,
+      "grad_norm": 0.9765625,
+      "learning_rate": 1.1723076923076925e-06,
+      "loss": 0.2731,
+      "step": 6120
+    },
+    {
+      "epoch": 0.94794425221812,
+      "grad_norm": 0.9140625,
+      "learning_rate": 1.1415384615384617e-06,
+      "loss": 0.2837,
+      "step": 6130
+    },
+    {
+      "epoch": 0.9494906539346258,
+      "grad_norm": 1.0078125,
+      "learning_rate": 1.1107692307692309e-06,
+      "loss": 0.2272,
+      "step": 6140
+    },
+    {
+      "epoch": 0.9510370556511317,
+      "grad_norm": 0.9921875,
+      "learning_rate": 1.08e-06,
+      "loss": 0.3152,
+      "step": 6150
+    },
+    {
+      "epoch": 0.9525834573676377,
+      "grad_norm": 0.86328125,
+      "learning_rate": 1.0492307692307694e-06,
+      "loss": 0.3017,
+      "step": 6160
+    },
+    {
+      "epoch": 0.9541298590841436,
+      "grad_norm": 0.8828125,
+      "learning_rate": 1.0184615384615386e-06,
+      "loss": 0.3378,
+      "step": 6170
+    },
+    {
+      "epoch": 0.9556762608006495,
+      "grad_norm": 0.92578125,
+      "learning_rate": 9.876923076923078e-07,
+      "loss": 0.2503,
+      "step": 6180
+    },
+    {
+      "epoch": 0.9572226625171554,
+      "grad_norm": 0.87109375,
+      "learning_rate": 9.56923076923077e-07,
+      "loss": 0.3438,
+      "step": 6190
+    },
+    {
+      "epoch": 0.9587690642336613,
+      "grad_norm": 0.875,
+      "learning_rate": 9.261538461538462e-07,
+      "loss": 0.2667,
+      "step": 6200
+    },
+    {
+      "epoch": 0.9603154659501673,
+      "grad_norm": 0.64453125,
+      "learning_rate": 8.953846153846155e-07,
+      "loss": 0.2745,
+      "step": 6210
+    },
+    {
+      "epoch": 0.9618618676666731,
+      "grad_norm": 0.84765625,
+      "learning_rate": 8.646153846153847e-07,
+      "loss": 0.2683,
+      "step": 6220
+    },
+    {
+      "epoch": 0.963408269383179,
+      "grad_norm": 0.91015625,
+      "learning_rate": 8.338461538461539e-07,
+      "loss": 0.2422,
+      "step": 6230
+    },
+    {
+      "epoch": 0.9649546710996849,
+      "grad_norm": 0.89453125,
+      "learning_rate": 8.030769230769231e-07,
+      "loss": 0.2588,
+      "step": 6240
+    },
+    {
+      "epoch": 0.9665010728161908,
+      "grad_norm": 0.87890625,
+      "learning_rate": 7.723076923076923e-07,
+      "loss": 0.2812,
+      "step": 6250
+    },
+    {
+      "epoch": 0.9680474745326967,
+      "grad_norm": 1.09375,
+      "learning_rate": 7.415384615384616e-07,
+      "loss": 0.3232,
+      "step": 6260
+    },
+    {
+      "epoch": 0.9695938762492027,
+      "grad_norm": 0.92578125,
+      "learning_rate": 7.107692307692309e-07,
+      "loss": 0.2422,
+      "step": 6270
+    },
+    {
+      "epoch": 0.9711402779657086,
+      "grad_norm": 0.9765625,
+      "learning_rate": 6.800000000000001e-07,
+      "loss": 0.2833,
+      "step": 6280
+    },
+    {
+      "epoch": 0.9726866796822145,
+      "grad_norm": 0.890625,
+      "learning_rate": 6.492307692307692e-07,
+      "loss": 0.2517,
+      "step": 6290
+    },
+    {
+      "epoch": 0.9742330813987203,
+      "grad_norm": 0.99609375,
+      "learning_rate": 6.184615384615385e-07,
+      "loss": 0.2534,
+      "step": 6300
+    },
+    {
+      "epoch": 0.9757794831152262,
+      "grad_norm": 0.8125,
+      "learning_rate": 5.876923076923077e-07,
+      "loss": 0.2911,
+      "step": 6310
+    },
+    {
+      "epoch": 0.9773258848317322,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.56923076923077e-07,
+      "loss": 0.2628,
+      "step": 6320
+    },
+    {
+      "epoch": 0.9788722865482381,
+      "grad_norm": 1.1015625,
+      "learning_rate": 5.261538461538462e-07,
+      "loss": 0.3191,
+      "step": 6330
+    },
+    {
+      "epoch": 0.980418688264744,
+      "grad_norm": 0.9765625,
+      "learning_rate": 4.953846153846155e-07,
+      "loss": 0.3225,
+      "step": 6340
+    },
+    {
+      "epoch": 0.9819650899812499,
+      "grad_norm": 0.92578125,
+      "learning_rate": 4.6461538461538465e-07,
+      "loss": 0.2836,
+      "step": 6350
+    },
+    {
+      "epoch": 0.9835114916977558,
+      "grad_norm": 1.0625,
+      "learning_rate": 4.3384615384615384e-07,
+      "loss": 0.303,
+      "step": 6360
+    },
+    {
+      "epoch": 0.9850578934142616,
+      "grad_norm": 1.171875,
+      "learning_rate": 4.0307692307692313e-07,
+      "loss": 0.2381,
+      "step": 6370
+    },
+    {
+      "epoch": 0.9866042951307676,
+      "grad_norm": 0.77734375,
+      "learning_rate": 3.7230769230769236e-07,
+      "loss": 0.2821,
+      "step": 6380
+    },
+    {
+      "epoch": 0.9881506968472735,
+      "grad_norm": 0.796875,
+      "learning_rate": 3.4153846153846155e-07,
+      "loss": 0.3187,
+      "step": 6390
+    },
+    {
+      "epoch": 0.9896970985637794,
+      "grad_norm": 0.80859375,
+      "learning_rate": 3.107692307692308e-07,
+      "loss": 0.2993,
+      "step": 6400
+    },
+    {
+      "epoch": 0.9912435002802853,
+      "grad_norm": 0.7421875,
+      "learning_rate": 2.8e-07,
+      "loss": 0.2923,
+      "step": 6410
+    },
+    {
+      "epoch": 0.9927899019967912,
+      "grad_norm": 0.953125,
+      "learning_rate": 2.4923076923076926e-07,
+      "loss": 0.2407,
+      "step": 6420
+    },
+    {
+      "epoch": 0.9943363037132972,
+      "grad_norm": 1.0078125,
+      "learning_rate": 2.1846153846153847e-07,
+      "loss": 0.257,
+      "step": 6430
+    },
+    {
+      "epoch": 0.9958827054298031,
+      "grad_norm": 1.171875,
+      "learning_rate": 1.8769230769230773e-07,
+      "loss": 0.2531,
+      "step": 6440
+    },
+    {
+      "epoch": 0.997429107146309,
+      "grad_norm": 0.69921875,
+      "learning_rate": 1.5692307692307694e-07,
+      "loss": 0.273,
+      "step": 6450
+    },
+    {
+      "epoch": 0.9989755088628148,
+      "grad_norm": 0.99609375,
+      "learning_rate": 1.2615384615384617e-07,
+      "loss": 0.313,
+      "step": 6460
+    },
+    {
+      "epoch": 1.0004639205149517,
+      "grad_norm": 0.75,
+      "learning_rate": 9.53846153846154e-08,
+      "loss": 0.2218,
+      "step": 6470
+    },
+    {
+      "epoch": 1.0020103222314576,
+      "grad_norm": 1.0625,
+      "learning_rate": 6.461538461538462e-08,
+      "loss": 0.3402,
+      "step": 6480
+    },
+    {
+      "epoch": 1.0035567239479637,
+      "grad_norm": 0.80078125,
+      "learning_rate": 3.384615384615385e-08,
+      "loss": 0.2327,
+      "step": 6490
+    },
+    {
+      "epoch": 1.0051031256644696,
+      "grad_norm": 0.98046875,
+      "learning_rate": 3.0769230769230774e-09,
+      "loss": 0.2185,
+      "step": 6500
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 6500,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.4935783864782275e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}