diff --git "a/Flan-T5-Typosquat-detect/trainer_state.json" "b/Flan-T5-Typosquat-detect/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/Flan-T5-Typosquat-detect/trainer_state.json"
@@ -0,0 +1,14065 @@
+{
+  "best_metric": 0.012842352502048016,
+  "best_model_checkpoint": "typosquat_flan_model_peft/checkpoint-20000",
+  "epoch": 4.0,
+  "eval_steps": 500,
+  "global_step": 20000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.002,
+      "grad_norm": 2.9712789058685303,
+      "learning_rate": 4.9980000000000006e-05,
+      "loss": 4.4769,
+      "step": 10
+    },
+    {
+      "epoch": 0.004,
+      "grad_norm": 2.7549290657043457,
+      "learning_rate": 4.996e-05,
+      "loss": 4.1624,
+      "step": 20
+    },
+    {
+      "epoch": 0.006,
+      "grad_norm": 3.280527114868164,
+      "learning_rate": 4.9940000000000006e-05,
+      "loss": 3.9022,
+      "step": 30
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 3.6827425956726074,
+      "learning_rate": 4.992e-05,
+      "loss": 3.289,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "grad_norm": 4.800412178039551,
+      "learning_rate": 4.99e-05,
+      "loss": 2.6414,
+      "step": 50
+    },
+    {
+      "epoch": 0.012,
+      "grad_norm": 3.7369704246520996,
+      "learning_rate": 4.9880000000000004e-05,
+      "loss": 1.8745,
+      "step": 60
+    },
+    {
+      "epoch": 0.014,
+      "grad_norm": 4.058594226837158,
+      "learning_rate": 4.986e-05,
+      "loss": 1.1633,
+      "step": 70
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 4.471256256103516,
+      "learning_rate": 4.9840000000000004e-05,
+      "loss": 0.5817,
+      "step": 80
+    },
+    {
+      "epoch": 0.018,
+      "grad_norm": 2.974433660507202,
+      "learning_rate": 4.982e-05,
+      "loss": 0.2911,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "grad_norm": 0.5866895914077759,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 0.3218,
+      "step": 100
+    },
+    {
+      "epoch": 0.022,
+      "grad_norm": 0.8520813584327698,
+      "learning_rate": 4.978e-05,
+      "loss": 0.2104,
+      "step": 110
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 0.2147858440876007,
+      "learning_rate": 4.976e-05,
+      "loss": 0.1706,
+      "step": 120
+    },
+    {
+      "epoch": 0.026,
+      "grad_norm": 3.007725238800049,
+      "learning_rate": 4.974e-05,
+      "loss": 0.3214,
+      "step": 130
+    },
+    {
+      "epoch": 0.028,
+      "grad_norm": 0.693766713142395,
+      "learning_rate": 4.972e-05,
+      "loss": 0.137,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "grad_norm": 1.4026163816452026,
+      "learning_rate": 4.97e-05,
+      "loss": 0.1514,
+      "step": 150
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 1.3514808416366577,
+      "learning_rate": 4.9680000000000005e-05,
+      "loss": 0.2081,
+      "step": 160
+    },
+    {
+      "epoch": 0.034,
+      "grad_norm": 1.0751250982284546,
+      "learning_rate": 4.966e-05,
+      "loss": 0.1561,
+      "step": 170
+    },
+    {
+      "epoch": 0.036,
+      "grad_norm": 1.2864580154418945,
+      "learning_rate": 4.9640000000000006e-05,
+      "loss": 0.0576,
+      "step": 180
+    },
+    {
+      "epoch": 0.038,
+      "grad_norm": 3.6556406021118164,
+      "learning_rate": 4.962e-05,
+      "loss": 0.2457,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 2.740448236465454,
+      "learning_rate": 4.96e-05,
+      "loss": 0.2094,
+      "step": 200
+    },
+    {
+      "epoch": 0.042,
+      "grad_norm": 1.7056159973144531,
+      "learning_rate": 4.958e-05,
+      "loss": 0.1938,
+      "step": 210
+    },
+    {
+      "epoch": 0.044,
+      "grad_norm": 0.500120997428894,
+      "learning_rate": 4.956e-05,
+      "loss": 0.2285,
+      "step": 220
+    },
+    {
+      "epoch": 0.046,
+      "grad_norm": 1.6456410884857178,
+      "learning_rate": 4.9540000000000003e-05,
+      "loss": 0.2164,
+      "step": 230
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 0.29337406158447266,
+      "learning_rate": 4.952e-05,
+      "loss": 0.1568,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "grad_norm": 1.0422818660736084,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 0.1407,
+      "step": 250
+    },
+    {
+      "epoch": 0.052,
+      "grad_norm": 3.711244821548462,
+      "learning_rate": 4.948000000000001e-05,
+      "loss": 0.2156,
+      "step": 260
+    },
+    {
+      "epoch": 0.054,
+      "grad_norm": 1.4521507024765015,
+      "learning_rate": 4.946e-05,
+      "loss": 0.1499,
+      "step": 270
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 1.1018396615982056,
+      "learning_rate": 4.944e-05,
+      "loss": 0.1733,
+      "step": 280
+    },
+    {
+      "epoch": 0.058,
+      "grad_norm": 3.3761978149414062,
+      "learning_rate": 4.942e-05,
+      "loss": 0.2515,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "grad_norm": 1.1806458234786987,
+      "learning_rate": 4.94e-05,
+      "loss": 0.1081,
+      "step": 300
+    },
+    {
+      "epoch": 0.062,
+      "grad_norm": 1.5186078548431396,
+      "learning_rate": 4.9380000000000005e-05,
+      "loss": 0.1387,
+      "step": 310
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 0.5397683382034302,
+      "learning_rate": 4.936e-05,
+      "loss": 0.087,
+      "step": 320
+    },
+    {
+      "epoch": 0.066,
+      "grad_norm": 0.28479281067848206,
+      "learning_rate": 4.9340000000000005e-05,
+      "loss": 0.0819,
+      "step": 330
+    },
+    {
+      "epoch": 0.068,
+      "grad_norm": 0.3176600933074951,
+      "learning_rate": 4.932e-05,
+      "loss": 0.1401,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "grad_norm": 0.14727644622325897,
+      "learning_rate": 4.93e-05,
+      "loss": 0.1139,
+      "step": 350
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 0.6405311226844788,
+      "learning_rate": 4.928e-05,
+      "loss": 0.1413,
+      "step": 360
+    },
+    {
+      "epoch": 0.074,
+      "grad_norm": 2.0455615520477295,
+      "learning_rate": 4.926e-05,
+      "loss": 0.102,
+      "step": 370
+    },
+    {
+      "epoch": 0.076,
+      "grad_norm": 1.2913883924484253,
+      "learning_rate": 4.924e-05,
+      "loss": 0.0346,
+      "step": 380
+    },
+    {
+      "epoch": 0.078,
+      "grad_norm": 0.3396105170249939,
+      "learning_rate": 4.9220000000000006e-05,
+      "loss": 0.1247,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": 0.8023151159286499,
+      "learning_rate": 4.92e-05,
+      "loss": 0.0915,
+      "step": 400
+    },
+    {
+      "epoch": 0.082,
+      "grad_norm": 2.867755651473999,
+      "learning_rate": 4.918000000000001e-05,
+      "loss": 0.0638,
+      "step": 410
+    },
+    {
+      "epoch": 0.084,
+      "grad_norm": 4.464131832122803,
+      "learning_rate": 4.9160000000000004e-05,
+      "loss": 0.1038,
+      "step": 420
+    },
+    {
+      "epoch": 0.086,
+      "grad_norm": 0.06573939323425293,
+      "learning_rate": 4.914e-05,
+      "loss": 0.0835,
+      "step": 430
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 3.679654121398926,
+      "learning_rate": 4.9120000000000004e-05,
+      "loss": 0.0967,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "grad_norm": 0.6470697522163391,
+      "learning_rate": 4.91e-05,
+      "loss": 0.1244,
+      "step": 450
+    },
+    {
+      "epoch": 0.092,
+      "grad_norm": 0.11164885014295578,
+      "learning_rate": 4.9080000000000004e-05,
+      "loss": 0.2084,
+      "step": 460
+    },
+    {
+      "epoch": 0.094,
+      "grad_norm": 0.10385742038488388,
+      "learning_rate": 4.906e-05,
+      "loss": 0.0626,
+      "step": 470
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 2.7525277137756348,
+      "learning_rate": 4.9040000000000005e-05,
+      "loss": 0.07,
+      "step": 480
+    },
+    {
+      "epoch": 0.098,
+      "grad_norm": 0.11531264334917068,
+      "learning_rate": 4.902e-05,
+      "loss": 0.0261,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "grad_norm": 3.795445203781128,
+      "learning_rate": 4.9e-05,
+      "loss": 0.1517,
+      "step": 500
+    },
+    {
+      "epoch": 0.102,
+      "grad_norm": 0.14794982969760895,
+      "learning_rate": 4.898e-05,
+      "loss": 0.1346,
+      "step": 510
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 3.277275800704956,
+      "learning_rate": 4.896e-05,
+      "loss": 0.0195,
+      "step": 520
+    },
+    {
+      "epoch": 0.106,
+      "grad_norm": 5.350284099578857,
+      "learning_rate": 4.894e-05,
+      "loss": 0.1441,
+      "step": 530
+    },
+    {
+      "epoch": 0.108,
+      "grad_norm": 0.12020950764417648,
+      "learning_rate": 4.8920000000000006e-05,
+      "loss": 0.0465,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "grad_norm": 0.7666958570480347,
+      "learning_rate": 4.89e-05,
+      "loss": 0.0169,
+      "step": 550
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 2.0386059284210205,
+      "learning_rate": 4.8880000000000006e-05,
+      "loss": 0.1544,
+      "step": 560
+    },
+    {
+      "epoch": 0.114,
+      "grad_norm": 0.1720697283744812,
+      "learning_rate": 4.886e-05,
+      "loss": 0.0535,
+      "step": 570
+    },
+    {
+      "epoch": 0.116,
+      "grad_norm": 0.04188171774148941,
+      "learning_rate": 4.884e-05,
+      "loss": 0.0547,
+      "step": 580
+    },
+    {
+      "epoch": 0.118,
+      "grad_norm": 0.36014607548713684,
+      "learning_rate": 4.8820000000000004e-05,
+      "loss": 0.0707,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 0.06317629665136337,
+      "learning_rate": 4.88e-05,
+      "loss": 0.0984,
+      "step": 600
+    },
+    {
+      "epoch": 0.122,
+      "grad_norm": 0.04827451333403587,
+      "learning_rate": 4.8780000000000004e-05,
+      "loss": 0.0322,
+      "step": 610
+    },
+    {
+      "epoch": 0.124,
+      "grad_norm": 0.09840811043977737,
+      "learning_rate": 4.876e-05,
+      "loss": 0.0579,
+      "step": 620
+    },
+    {
+      "epoch": 0.126,
+      "grad_norm": 1.4731272459030151,
+      "learning_rate": 4.8740000000000004e-05,
+      "loss": 0.0203,
+      "step": 630
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 0.04075102135539055,
+      "learning_rate": 4.872000000000001e-05,
+      "loss": 0.0479,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "grad_norm": 2.8817827701568604,
+      "learning_rate": 4.87e-05,
+      "loss": 0.0964,
+      "step": 650
+    },
+    {
+      "epoch": 0.132,
+      "grad_norm": 3.602795124053955,
+      "learning_rate": 4.868e-05,
+      "loss": 0.0772,
+      "step": 660
+    },
+    {
+      "epoch": 0.134,
+      "grad_norm": 7.180420875549316,
+      "learning_rate": 4.866e-05,
+      "loss": 0.0225,
+      "step": 670
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 0.40698742866516113,
+      "learning_rate": 4.864e-05,
+      "loss": 0.0883,
+      "step": 680
+    },
+    {
+      "epoch": 0.138,
+      "grad_norm": 4.945473670959473,
+      "learning_rate": 4.8620000000000005e-05,
+      "loss": 0.106,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "grad_norm": 0.16914227604866028,
+      "learning_rate": 4.86e-05,
+      "loss": 0.1672,
+      "step": 700
+    },
+    {
+      "epoch": 0.142,
+      "grad_norm": 0.5895906090736389,
+      "learning_rate": 4.8580000000000006e-05,
+      "loss": 0.0754,
+      "step": 710
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 0.2772788107395172,
+      "learning_rate": 4.856e-05,
+      "loss": 0.0339,
+      "step": 720
+    },
+    {
+      "epoch": 0.146,
+      "grad_norm": 0.1928812563419342,
+      "learning_rate": 4.854e-05,
+      "loss": 0.0592,
+      "step": 730
+    },
+    {
+      "epoch": 0.148,
+      "grad_norm": 0.046394895762205124,
+      "learning_rate": 4.852e-05,
+      "loss": 0.0454,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "grad_norm": 4.180880069732666,
+      "learning_rate": 4.85e-05,
+      "loss": 0.1045,
+      "step": 750
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 3.7465639114379883,
+      "learning_rate": 4.8480000000000003e-05,
+      "loss": 0.0181,
+      "step": 760
+    },
+    {
+      "epoch": 0.154,
+      "grad_norm": 0.293394535779953,
+      "learning_rate": 4.846e-05,
+      "loss": 0.0094,
+      "step": 770
+    },
+    {
+      "epoch": 0.156,
+      "grad_norm": 0.0170027744024992,
+      "learning_rate": 4.8440000000000004e-05,
+      "loss": 0.0174,
+      "step": 780
+    },
+    {
+      "epoch": 0.158,
+      "grad_norm": 0.021193701773881912,
+      "learning_rate": 4.842000000000001e-05,
+      "loss": 0.1072,
+      "step": 790
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": 0.5707680583000183,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 0.0646,
+      "step": 800
+    },
+    {
+      "epoch": 0.162,
+      "grad_norm": 0.08426664769649506,
+      "learning_rate": 4.838e-05,
+      "loss": 0.1026,
+      "step": 810
+    },
+    {
+      "epoch": 0.164,
+      "grad_norm": 1.9383238554000854,
+      "learning_rate": 4.836e-05,
+      "loss": 0.0185,
+      "step": 820
+    },
+    {
+      "epoch": 0.166,
+      "grad_norm": 0.04476593807339668,
+      "learning_rate": 4.834e-05,
+      "loss": 0.0397,
+      "step": 830
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 7.816348552703857,
+      "learning_rate": 4.8320000000000005e-05,
+      "loss": 0.1227,
+      "step": 840
+    },
+    {
+      "epoch": 0.17,
+      "grad_norm": 0.10927680134773254,
+      "learning_rate": 4.83e-05,
+      "loss": 0.0911,
+      "step": 850
+    },
+    {
+      "epoch": 0.172,
+      "grad_norm": 0.10985364019870758,
+      "learning_rate": 4.8280000000000005e-05,
+      "loss": 0.0128,
+      "step": 860
+    },
+    {
+      "epoch": 0.174,
+      "grad_norm": 1.405458927154541,
+      "learning_rate": 4.826e-05,
+      "loss": 0.0182,
+      "step": 870
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 3.5338621139526367,
+      "learning_rate": 4.824e-05,
+      "loss": 0.0624,
+      "step": 880
+    },
+    {
+      "epoch": 0.178,
+      "grad_norm": 0.1506468653678894,
+      "learning_rate": 4.822e-05,
+      "loss": 0.0269,
+      "step": 890
+    },
+    {
+      "epoch": 0.18,
+      "grad_norm": 0.00646029831841588,
+      "learning_rate": 4.82e-05,
+      "loss": 0.0407,
+      "step": 900
+    },
+    {
+      "epoch": 0.182,
+      "grad_norm": 0.12899932265281677,
+      "learning_rate": 4.818e-05,
+      "loss": 0.0659,
+      "step": 910
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 2.2137913703918457,
+      "learning_rate": 4.816e-05,
+      "loss": 0.0145,
+      "step": 920
+    },
+    {
+      "epoch": 0.186,
+      "grad_norm": 0.20166102051734924,
+      "learning_rate": 4.814e-05,
+      "loss": 0.0912,
+      "step": 930
+    },
+    {
+      "epoch": 0.188,
+      "grad_norm": 0.6006794571876526,
+      "learning_rate": 4.812000000000001e-05,
+      "loss": 0.018,
+      "step": 940
+    },
+    {
+      "epoch": 0.19,
+      "grad_norm": 0.8807332515716553,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 0.0028,
+      "step": 950
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 0.03953103348612785,
+      "learning_rate": 4.808e-05,
+      "loss": 0.0028,
+      "step": 960
+    },
+    {
+      "epoch": 0.194,
+      "grad_norm": 0.0035690057557076216,
+      "learning_rate": 4.8060000000000004e-05,
+      "loss": 0.018,
+      "step": 970
+    },
+    {
+      "epoch": 0.196,
+      "grad_norm": 0.009828136302530766,
+      "learning_rate": 4.804e-05,
+      "loss": 0.027,
+      "step": 980
+    },
+    {
+      "epoch": 0.198,
+      "grad_norm": 0.19281725585460663,
+      "learning_rate": 4.8020000000000004e-05,
+      "loss": 0.0752,
+      "step": 990
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 0.008602121844887733,
+      "learning_rate": 4.8e-05,
+      "loss": 0.2189,
+      "step": 1000
+    },
+    {
+      "epoch": 0.202,
+      "grad_norm": 3.5480191707611084,
+      "learning_rate": 4.7980000000000005e-05,
+      "loss": 0.197,
+      "step": 1010
+    },
+    {
+      "epoch": 0.204,
+      "grad_norm": 0.06215432286262512,
+      "learning_rate": 4.796e-05,
+      "loss": 0.0947,
+      "step": 1020
+    },
+    {
+      "epoch": 0.206,
+      "grad_norm": 0.5615971684455872,
+      "learning_rate": 4.794e-05,
+      "loss": 0.037,
+      "step": 1030
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 0.30555933713912964,
+      "learning_rate": 4.792e-05,
+      "loss": 0.0095,
+      "step": 1040
+    },
+    {
+      "epoch": 0.21,
+      "grad_norm": 0.0339435338973999,
+      "learning_rate": 4.79e-05,
+      "loss": 0.0199,
+      "step": 1050
+    },
+    {
+      "epoch": 0.212,
+      "grad_norm": 0.011141384951770306,
+      "learning_rate": 4.788e-05,
+      "loss": 0.0678,
+      "step": 1060
+    },
+    {
+      "epoch": 0.214,
+      "grad_norm": 0.003661577822640538,
+      "learning_rate": 4.7860000000000006e-05,
+      "loss": 0.163,
+      "step": 1070
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 0.01145030278712511,
+      "learning_rate": 4.784e-05,
+      "loss": 0.0227,
+      "step": 1080
+    },
+    {
+      "epoch": 0.218,
+      "grad_norm": 0.08786927908658981,
+      "learning_rate": 4.7820000000000006e-05,
+      "loss": 0.0769,
+      "step": 1090
+    },
+    {
+      "epoch": 0.22,
+      "grad_norm": 0.008924941532313824,
+      "learning_rate": 4.78e-05,
+      "loss": 0.0196,
+      "step": 1100
+    },
+    {
+      "epoch": 0.222,
+      "grad_norm": 4.367251873016357,
+      "learning_rate": 4.778e-05,
+      "loss": 0.0152,
+      "step": 1110
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 0.030621623620390892,
+      "learning_rate": 4.7760000000000004e-05,
+      "loss": 0.0948,
+      "step": 1120
+    },
+    {
+      "epoch": 0.226,
+      "grad_norm": 0.0778178870677948,
+      "learning_rate": 4.774e-05,
+      "loss": 0.0352,
+      "step": 1130
+    },
+    {
+      "epoch": 0.228,
+      "grad_norm": 0.10039296001195908,
+      "learning_rate": 4.7720000000000004e-05,
+      "loss": 0.0405,
+      "step": 1140
+    },
+    {
+      "epoch": 0.23,
+      "grad_norm": 1.098934531211853,
+      "learning_rate": 4.77e-05,
+      "loss": 0.0287,
+      "step": 1150
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 0.3654647767543793,
+      "learning_rate": 4.7680000000000004e-05,
+      "loss": 0.0643,
+      "step": 1160
+    },
+    {
+      "epoch": 0.234,
+      "grad_norm": 0.3039003312587738,
+      "learning_rate": 4.766000000000001e-05,
+      "loss": 0.0244,
+      "step": 1170
+    },
+    {
+      "epoch": 0.236,
+      "grad_norm": 3.5462236404418945,
+      "learning_rate": 4.7640000000000005e-05,
+      "loss": 0.0468,
+      "step": 1180
+    },
+    {
+      "epoch": 0.238,
+      "grad_norm": 0.01479172334074974,
+      "learning_rate": 4.762e-05,
+      "loss": 0.0078,
+      "step": 1190
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 4.272269248962402,
+      "learning_rate": 4.76e-05,
+      "loss": 0.0113,
+      "step": 1200
+    },
+    {
+      "epoch": 0.242,
+      "grad_norm": 4.091390609741211,
+      "learning_rate": 4.758e-05,
+      "loss": 0.0121,
+      "step": 1210
+    },
+    {
+      "epoch": 0.244,
+      "grad_norm": 0.03659886121749878,
+      "learning_rate": 4.7560000000000005e-05,
+      "loss": 0.1266,
+      "step": 1220
+    },
+    {
+      "epoch": 0.246,
+      "grad_norm": 0.157633975148201,
+      "learning_rate": 4.754e-05,
+      "loss": 0.0193,
+      "step": 1230
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 0.0051111564971506596,
+      "learning_rate": 4.7520000000000006e-05,
+      "loss": 0.0081,
+      "step": 1240
+    },
+    {
+      "epoch": 0.25,
+      "grad_norm": 1.567456603050232,
+      "learning_rate": 4.75e-05,
+      "loss": 0.0048,
+      "step": 1250
+    },
+    {
+      "epoch": 0.252,
+      "grad_norm": 3.683023452758789,
+      "learning_rate": 4.748e-05,
+      "loss": 0.0195,
+      "step": 1260
+    },
+    {
+      "epoch": 0.254,
+      "grad_norm": 0.06732743978500366,
+      "learning_rate": 4.746e-05,
+      "loss": 0.0374,
+      "step": 1270
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 0.10905110090970993,
+      "learning_rate": 4.744e-05,
+      "loss": 0.0392,
+      "step": 1280
+    },
+    {
+      "epoch": 0.258,
+      "grad_norm": 0.47610998153686523,
+      "learning_rate": 4.742e-05,
+      "loss": 0.0029,
+      "step": 1290
+    },
+    {
+      "epoch": 0.26,
+      "grad_norm": 0.13898958265781403,
+      "learning_rate": 4.74e-05,
+      "loss": 0.0551,
+      "step": 1300
+    },
+    {
+      "epoch": 0.262,
+      "grad_norm": 0.012497484683990479,
+      "learning_rate": 4.7380000000000004e-05,
+      "loss": 0.1259,
+      "step": 1310
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 0.013291290029883385,
+      "learning_rate": 4.736000000000001e-05,
+      "loss": 0.0031,
+      "step": 1320
+    },
+    {
+      "epoch": 0.266,
+      "grad_norm": 0.020174721255898476,
+      "learning_rate": 4.7340000000000004e-05,
+      "loss": 0.0552,
+      "step": 1330
+    },
+    {
+      "epoch": 0.268,
+      "grad_norm": 0.36099356412887573,
+      "learning_rate": 4.732e-05,
+      "loss": 0.0522,
+      "step": 1340
+    },
+    {
+      "epoch": 0.27,
+      "grad_norm": 0.03428328409790993,
+      "learning_rate": 4.73e-05,
+      "loss": 0.0121,
+      "step": 1350
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 0.10378572344779968,
+      "learning_rate": 4.728e-05,
+      "loss": 0.1406,
+      "step": 1360
+    },
+    {
+      "epoch": 0.274,
+      "grad_norm": 0.06635264307260513,
+      "learning_rate": 4.7260000000000005e-05,
+      "loss": 0.0073,
+      "step": 1370
+    },
+    {
+      "epoch": 0.276,
+      "grad_norm": 0.03210114687681198,
+      "learning_rate": 4.724e-05,
+      "loss": 0.0383,
+      "step": 1380
+    },
+    {
+      "epoch": 0.278,
+      "grad_norm": 0.016254939138889313,
+      "learning_rate": 4.7220000000000005e-05,
+      "loss": 0.0192,
+      "step": 1390
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 0.9825054407119751,
+      "learning_rate": 4.72e-05,
+      "loss": 0.0063,
+      "step": 1400
+    },
+    {
+      "epoch": 0.282,
+      "grad_norm": 0.39390048384666443,
+      "learning_rate": 4.718e-05,
+      "loss": 0.0018,
+      "step": 1410
+    },
+    {
+      "epoch": 0.284,
+      "grad_norm": 0.0035170663613826036,
+      "learning_rate": 4.716e-05,
+      "loss": 0.0057,
+      "step": 1420
+    },
+    {
+      "epoch": 0.286,
+      "grad_norm": 0.38211411237716675,
+      "learning_rate": 4.714e-05,
+      "loss": 0.0115,
+      "step": 1430
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 0.008729588240385056,
+      "learning_rate": 4.712e-05,
+      "loss": 0.0408,
+      "step": 1440
+    },
+    {
+      "epoch": 0.29,
+      "grad_norm": 0.8136197328567505,
+      "learning_rate": 4.71e-05,
+      "loss": 0.0021,
+      "step": 1450
+    },
+    {
+      "epoch": 0.292,
+      "grad_norm": 2.9295754432678223,
+      "learning_rate": 4.708e-05,
+      "loss": 0.1572,
+      "step": 1460
+    },
+    {
+      "epoch": 0.294,
+      "grad_norm": 0.9183974862098694,
+      "learning_rate": 4.706000000000001e-05,
+      "loss": 0.088,
+      "step": 1470
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 0.022023877128958702,
+      "learning_rate": 4.7040000000000004e-05,
+      "loss": 0.0209,
+      "step": 1480
+    },
+    {
+      "epoch": 0.298,
+      "grad_norm": 3.611262559890747,
+      "learning_rate": 4.702e-05,
+      "loss": 0.0424,
+      "step": 1490
+    },
+    {
+      "epoch": 0.3,
+      "grad_norm": 3.3381121158599854,
+      "learning_rate": 4.7e-05,
+      "loss": 0.0403,
+      "step": 1500
+    },
+    {
+      "epoch": 0.302,
+      "grad_norm": 0.0031520898919552565,
+      "learning_rate": 4.698e-05,
+      "loss": 0.0016,
+      "step": 1510
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 0.007907604798674583,
+      "learning_rate": 4.6960000000000004e-05,
+      "loss": 0.0437,
+      "step": 1520
+    },
+    {
+      "epoch": 0.306,
+      "grad_norm": 0.18256665766239166,
+      "learning_rate": 4.694e-05,
+      "loss": 0.0468,
+      "step": 1530
+    },
+    {
+      "epoch": 0.308,
+      "grad_norm": 0.08469197154045105,
+      "learning_rate": 4.6920000000000005e-05,
+      "loss": 0.094,
+      "step": 1540
+    },
+    {
+      "epoch": 0.31,
+      "grad_norm": 0.023916004225611687,
+      "learning_rate": 4.69e-05,
+      "loss": 0.0338,
+      "step": 1550
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 0.047887589782476425,
+      "learning_rate": 4.688e-05,
+      "loss": 0.0154,
+      "step": 1560
+    },
+    {
+      "epoch": 0.314,
+      "grad_norm": 11.07988452911377,
+      "learning_rate": 4.686e-05,
+      "loss": 0.0529,
+      "step": 1570
+    },
+    {
+      "epoch": 0.316,
+      "grad_norm": 0.3059365749359131,
+      "learning_rate": 4.684e-05,
+      "loss": 0.0585,
+      "step": 1580
+    },
+    {
+      "epoch": 0.318,
+      "grad_norm": 0.30327826738357544,
+      "learning_rate": 4.682e-05,
+      "loss": 0.0207,
+      "step": 1590
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": 0.0919349268078804,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 0.0873,
+      "step": 1600
+    },
+    {
+      "epoch": 0.322,
+      "grad_norm": 1.0741651058197021,
+      "learning_rate": 4.678e-05,
+      "loss": 0.0699,
+      "step": 1610
+    },
+    {
+      "epoch": 0.324,
+      "grad_norm": 0.00857605691999197,
+      "learning_rate": 4.6760000000000006e-05,
+      "loss": 0.0869,
+      "step": 1620
+    },
+    {
+      "epoch": 0.326,
+      "grad_norm": 0.035690005868673325,
+      "learning_rate": 4.674e-05,
+      "loss": 0.006,
+      "step": 1630
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 0.08569003641605377,
+      "learning_rate": 4.672e-05,
+      "loss": 0.0024,
+      "step": 1640
+    },
+    {
+      "epoch": 0.33,
+      "grad_norm": 0.11262888461351395,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 0.0475,
+      "step": 1650
+    },
+    {
+      "epoch": 0.332,
+      "grad_norm": 0.0029759930912405252,
+      "learning_rate": 4.668e-05,
+      "loss": 0.1158,
+      "step": 1660
+    },
+    {
+      "epoch": 0.334,
+      "grad_norm": 0.14019815623760223,
+      "learning_rate": 4.6660000000000004e-05,
+      "loss": 0.0015,
+      "step": 1670
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 0.06434296816587448,
+      "learning_rate": 4.664e-05,
+      "loss": 0.072,
+      "step": 1680
+    },
+    {
+      "epoch": 0.338,
+      "grad_norm": 0.02371792495250702,
+      "learning_rate": 4.6620000000000004e-05,
+      "loss": 0.0292,
+      "step": 1690
+    },
+    {
+      "epoch": 0.34,
+      "grad_norm": 6.26638126373291,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 0.0654,
+      "step": 1700
+    },
+    {
+      "epoch": 0.342,
+      "grad_norm": 0.013329977169632912,
+      "learning_rate": 4.6580000000000005e-05,
+      "loss": 0.0507,
+      "step": 1710
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 0.020742950960993767,
+      "learning_rate": 4.656e-05,
+      "loss": 0.0008,
+      "step": 1720
+    },
+    {
+      "epoch": 0.346,
+      "grad_norm": 0.10419709235429764,
+      "learning_rate": 4.654e-05,
+      "loss": 0.0156,
+      "step": 1730
+    },
+    {
+      "epoch": 0.348,
+      "grad_norm": 0.001128773088566959,
+      "learning_rate": 4.652e-05,
+      "loss": 0.0046,
+      "step": 1740
+    },
+    {
+      "epoch": 0.35,
+      "grad_norm": 0.02164996601641178,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 0.0009,
+      "step": 1750
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.002971704350784421,
+      "learning_rate": 4.648e-05,
+      "loss": 0.0299,
+      "step": 1760
+    },
+    {
+      "epoch": 0.354,
+      "grad_norm": 0.06722798943519592,
+      "learning_rate": 4.6460000000000006e-05,
+      "loss": 0.0352,
+      "step": 1770
+    },
+    {
+      "epoch": 0.356,
+      "grad_norm": 0.006361286621540785,
+      "learning_rate": 4.644e-05,
+      "loss": 0.0042,
+      "step": 1780
+    },
+    {
+      "epoch": 0.358,
+      "grad_norm": 5.5403571128845215,
+      "learning_rate": 4.642e-05,
+      "loss": 0.1371,
+      "step": 1790
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 0.2011372298002243,
+      "learning_rate": 4.64e-05,
+      "loss": 0.0963,
+      "step": 1800
+    },
+    {
+      "epoch": 0.362,
+      "grad_norm": 0.008815037086606026,
+      "learning_rate": 4.638e-05,
+      "loss": 0.0445,
+      "step": 1810
+    },
+    {
+      "epoch": 0.364,
+      "grad_norm": 0.010142517276108265,
+      "learning_rate": 4.636e-05,
+      "loss": 0.0581,
+      "step": 1820
+    },
+    {
+      "epoch": 0.366,
+      "grad_norm": 0.002696698298677802,
+      "learning_rate": 4.634e-05,
+      "loss": 0.0435,
+      "step": 1830
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 0.05711021274328232,
+      "learning_rate": 4.6320000000000004e-05,
+      "loss": 0.0221,
+      "step": 1840
+    },
+    {
+      "epoch": 0.37,
+      "grad_norm": 0.1295560747385025,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 0.0039,
+      "step": 1850
+    },
+    {
+      "epoch": 0.372,
+      "grad_norm": 8.641880989074707,
+      "learning_rate": 4.6280000000000004e-05,
+      "loss": 0.05,
+      "step": 1860
+    },
+    {
+      "epoch": 0.374,
+      "grad_norm": 0.008340752683579922,
+      "learning_rate": 4.626e-05,
+      "loss": 0.1107,
+      "step": 1870
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 0.041587088257074356,
+      "learning_rate": 4.624e-05,
+      "loss": 0.0902,
+      "step": 1880
+    },
+    {
+      "epoch": 0.378,
+      "grad_norm": 1.1638970375061035,
+      "learning_rate": 4.622e-05,
+      "loss": 0.0297,
+      "step": 1890
+    },
+    {
+      "epoch": 0.38,
+      "grad_norm": 0.016888264566659927,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 0.0078,
+      "step": 1900
+    },
+    {
+      "epoch": 0.382,
+      "grad_norm": 0.07481393963098526,
+      "learning_rate": 4.618e-05,
+      "loss": 0.0348,
+      "step": 1910
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 4.364596843719482,
+      "learning_rate": 4.6160000000000005e-05,
+      "loss": 0.1122,
+      "step": 1920
+    },
+    {
+      "epoch": 0.386,
+      "grad_norm": 0.0195784792304039,
+      "learning_rate": 4.614e-05,
+      "loss": 0.0434,
+      "step": 1930
+    },
+    {
+      "epoch": 0.388,
+      "grad_norm": 6.045379161834717,
+      "learning_rate": 4.612e-05,
+      "loss": 0.0194,
+      "step": 1940
+    },
+    {
+      "epoch": 0.39,
+      "grad_norm": 0.0069319577887654305,
+      "learning_rate": 4.61e-05,
+      "loss": 0.0809,
+      "step": 1950
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 3.495983839035034,
+      "learning_rate": 4.608e-05,
+      "loss": 0.0738,
+      "step": 1960
+    },
+    {
+      "epoch": 0.394,
+      "grad_norm": 0.19603276252746582,
+      "learning_rate": 4.606e-05,
+      "loss": 0.0115,
+      "step": 1970
+    },
+    {
+      "epoch": 0.396,
+      "grad_norm": 3.1189138889312744,
+      "learning_rate": 4.604e-05,
+      "loss": 0.0076,
+      "step": 1980
+    },
+    {
+      "epoch": 0.398,
+      "grad_norm": 0.010036235675215721,
+      "learning_rate": 4.602e-05,
+      "loss": 0.0124,
+      "step": 1990
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 0.02806951105594635,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 0.0127,
+      "step": 2000
+    },
+    {
+      "epoch": 0.402,
+      "grad_norm": 0.19674068689346313,
+      "learning_rate": 4.5980000000000004e-05,
+      "loss": 0.0184,
+      "step": 2010
+    },
+    {
+      "epoch": 0.404,
+      "grad_norm": 0.021522896364331245,
+      "learning_rate": 4.596e-05,
+      "loss": 0.0021,
+      "step": 2020
+    },
+    {
+      "epoch": 0.406,
+      "grad_norm": 0.044582124799489975,
+      "learning_rate": 4.594e-05,
+      "loss": 0.0632,
+      "step": 2030
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 0.004063369706273079,
+      "learning_rate": 4.592e-05,
+      "loss": 0.001,
+      "step": 2040
+    },
+    {
+      "epoch": 0.41,
+      "grad_norm": 0.032624807208776474,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 0.0493,
+      "step": 2050
+    },
+    {
+      "epoch": 0.412,
+      "grad_norm": 0.018941815942525864,
+      "learning_rate": 4.588e-05,
+      "loss": 0.0024,
+      "step": 2060
+    },
+    {
+      "epoch": 0.414,
+      "grad_norm": 0.4768604636192322,
+      "learning_rate": 4.5860000000000005e-05,
+      "loss": 0.0317,
+      "step": 2070
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 0.007809026166796684,
+      "learning_rate": 4.584e-05,
+      "loss": 0.0849,
+      "step": 2080
+    },
+    {
+      "epoch": 0.418,
+      "grad_norm": 0.09230257570743561,
+      "learning_rate": 4.5820000000000005e-05,
+      "loss": 0.0032,
+      "step": 2090
+    },
+    {
+      "epoch": 0.42,
+      "grad_norm": 0.0211187656968832,
+      "learning_rate": 4.58e-05,
+      "loss": 0.0148,
+      "step": 2100
+    },
+    {
+      "epoch": 0.422,
+      "grad_norm": 0.027965866029262543,
+      "learning_rate": 4.578e-05,
+      "loss": 0.0776,
+      "step": 2110
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 0.006458558142185211,
+      "learning_rate": 4.576e-05,
+      "loss": 0.0127,
+      "step": 2120
+    },
+    {
+      "epoch": 0.426,
+      "grad_norm": 2.797787666320801,
+      "learning_rate": 4.574e-05,
+      "loss": 0.0425,
+      "step": 2130
+    },
+    {
+      "epoch": 0.428,
+      "grad_norm": 0.008578737266361713,
+      "learning_rate": 4.572e-05,
+      "loss": 0.0697,
+      "step": 2140
+    },
+    {
+      "epoch": 0.43,
+      "grad_norm": 0.8794230818748474,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 0.049,
+      "step": 2150
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 0.09468420594930649,
+      "learning_rate": 4.568e-05,
+      "loss": 0.0515,
+      "step": 2160
+    },
+    {
+      "epoch": 0.434,
+      "grad_norm": 0.010913840495049953,
+      "learning_rate": 4.566e-05,
+      "loss": 0.0129,
+      "step": 2170
+    },
+    {
+      "epoch": 0.436,
+      "grad_norm": 0.024601969867944717,
+      "learning_rate": 4.564e-05,
+      "loss": 0.007,
+      "step": 2180
+    },
+    {
+      "epoch": 0.438,
+      "grad_norm": 0.08446158468723297,
+      "learning_rate": 4.562e-05,
+      "loss": 0.0543,
+      "step": 2190
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.03246144950389862,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 0.0081,
+      "step": 2200
+    },
+    {
+      "epoch": 0.442,
+      "grad_norm": 5.424055576324463,
+      "learning_rate": 4.558e-05,
+      "loss": 0.033,
+      "step": 2210
+    },
+    {
+      "epoch": 0.444,
+      "grad_norm": 0.060880791395902634,
+      "learning_rate": 4.5560000000000004e-05,
+      "loss": 0.0291,
+      "step": 2220
+    },
+    {
+      "epoch": 0.446,
+      "grad_norm": 0.01959436945617199,
+      "learning_rate": 4.554000000000001e-05,
+      "loss": 0.002,
+      "step": 2230
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 0.02229149639606476,
+      "learning_rate": 4.5520000000000005e-05,
+      "loss": 0.0222,
+      "step": 2240
+    },
+    {
+      "epoch": 0.45,
+      "grad_norm": 0.8051292300224304,
+      "learning_rate": 4.55e-05,
+      "loss": 0.0921,
+      "step": 2250
+    },
+    {
+      "epoch": 0.452,
+      "grad_norm": 4.567664623260498,
+      "learning_rate": 4.548e-05,
+      "loss": 0.0898,
+      "step": 2260
+    },
+    {
+      "epoch": 0.454,
+      "grad_norm": 1.3188562393188477,
+      "learning_rate": 4.546e-05,
+      "loss": 0.059,
+      "step": 2270
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 0.07389594614505768,
+      "learning_rate": 4.5440000000000005e-05,
+      "loss": 0.0776,
+      "step": 2280
+    },
+    {
+      "epoch": 0.458,
+      "grad_norm": 0.07509294152259827,
+      "learning_rate": 4.542e-05,
+      "loss": 0.0774,
+      "step": 2290
+    },
+    {
+      "epoch": 0.46,
+      "grad_norm": 0.020364753901958466,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 0.0062,
+      "step": 2300
+    },
+    {
+      "epoch": 0.462,
+      "grad_norm": 0.08534001559019089,
+      "learning_rate": 4.538e-05,
+      "loss": 0.0177,
+      "step": 2310
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 0.019280171021819115,
+      "learning_rate": 4.536e-05,
+      "loss": 0.002,
+      "step": 2320
+    },
+    {
+      "epoch": 0.466,
+      "grad_norm": 0.1653839647769928,
+      "learning_rate": 4.534e-05,
+      "loss": 0.0472,
+      "step": 2330
+    },
+    {
+      "epoch": 0.468,
+      "grad_norm": 0.01105203852057457,
+      "learning_rate": 4.532e-05,
+      "loss": 0.0281,
+      "step": 2340
+    },
+    {
+      "epoch": 0.47,
+      "grad_norm": 0.6273080110549927,
+      "learning_rate": 4.53e-05,
+      "loss": 0.0851,
+      "step": 2350
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 0.04479006305336952,
+      "learning_rate": 4.528e-05,
+      "loss": 0.0011,
+      "step": 2360
+    },
+    {
+      "epoch": 0.474,
+      "grad_norm": 0.013826013542711735,
+      "learning_rate": 4.5260000000000004e-05,
+      "loss": 0.0055,
+      "step": 2370
+    },
+    {
+      "epoch": 0.476,
+      "grad_norm": 0.008165967650711536,
+      "learning_rate": 4.524000000000001e-05,
+      "loss": 0.0009,
+      "step": 2380
+    },
+    {
+      "epoch": 0.478,
+      "grad_norm": 0.09005433320999146,
+      "learning_rate": 4.5220000000000004e-05,
+      "loss": 0.0367,
+      "step": 2390
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 0.059469930827617645,
+      "learning_rate": 4.52e-05,
+      "loss": 0.0509,
+      "step": 2400
+    },
+    {
+      "epoch": 0.482,
+      "grad_norm": 0.0066351089626550674,
+      "learning_rate": 4.518e-05,
+      "loss": 0.071,
+      "step": 2410
+    },
+    {
+      "epoch": 0.484,
+      "grad_norm": 6.272415637969971,
+      "learning_rate": 4.516e-05,
+      "loss": 0.0809,
+      "step": 2420
+    },
+    {
+      "epoch": 0.486,
+      "grad_norm": 0.011232190765440464,
+      "learning_rate": 4.5140000000000005e-05,
+      "loss": 0.049,
+      "step": 2430
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 0.8765757083892822,
+      "learning_rate": 4.512e-05,
+      "loss": 0.0486,
+      "step": 2440
+    },
+    {
+      "epoch": 0.49,
+      "grad_norm": 0.014707114547491074,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 0.0346,
+      "step": 2450
+    },
+    {
+      "epoch": 0.492,
+      "grad_norm": 0.02815588191151619,
+      "learning_rate": 4.508e-05,
+      "loss": 0.0285,
+      "step": 2460
+    },
+    {
+      "epoch": 0.494,
+      "grad_norm": 0.013969714753329754,
+      "learning_rate": 4.506e-05,
+      "loss": 0.0052,
+      "step": 2470
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 0.030105378478765488,
+      "learning_rate": 4.504e-05,
+      "loss": 0.0329,
+      "step": 2480
+    },
+    {
+      "epoch": 0.498,
+      "grad_norm": 0.028038593009114265,
+      "learning_rate": 4.502e-05,
+      "loss": 0.0279,
+      "step": 2490
+    },
+    {
+      "epoch": 0.5,
+      "grad_norm": 0.09044764190912247,
+      "learning_rate": 4.5e-05,
+      "loss": 0.0314,
+      "step": 2500
+    },
+    {
+      "epoch": 0.502,
+      "grad_norm": 0.02895580418407917,
+      "learning_rate": 4.498e-05,
+      "loss": 0.0823,
+      "step": 2510
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 0.004999326542019844,
+      "learning_rate": 4.496e-05,
+      "loss": 0.0015,
+      "step": 2520
+    },
+    {
+      "epoch": 0.506,
+      "grad_norm": 0.10353109240531921,
+      "learning_rate": 4.494000000000001e-05,
+      "loss": 0.0315,
+      "step": 2530
+    },
+    {
+      "epoch": 0.508,
+      "grad_norm": 1.6411314010620117,
+      "learning_rate": 4.4920000000000004e-05,
+      "loss": 0.0109,
+      "step": 2540
+    },
+    {
+      "epoch": 0.51,
+      "grad_norm": 0.11047064512968063,
+      "learning_rate": 4.49e-05,
+      "loss": 0.0227,
+      "step": 2550
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 0.027552004903554916,
+      "learning_rate": 4.488e-05,
+      "loss": 0.0481,
+      "step": 2560
+    },
+    {
+      "epoch": 0.514,
+      "grad_norm": 0.2809949815273285,
+      "learning_rate": 4.486e-05,
+      "loss": 0.0016,
+      "step": 2570
+    },
+    {
+      "epoch": 0.516,
+      "grad_norm": 0.012851985171437263,
+      "learning_rate": 4.4840000000000004e-05,
+      "loss": 0.0014,
+      "step": 2580
+    },
+    {
+      "epoch": 0.518,
+      "grad_norm": 5.742981910705566,
+      "learning_rate": 4.482e-05,
+      "loss": 0.0187,
+      "step": 2590
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 0.013996957801282406,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 0.0005,
+      "step": 2600
+    },
+    {
+      "epoch": 0.522,
+      "grad_norm": 0.1463635265827179,
+      "learning_rate": 4.478e-05,
+      "loss": 0.0517,
+      "step": 2610
+    },
+    {
+      "epoch": 0.524,
+      "grad_norm": 0.06374786049127579,
+      "learning_rate": 4.4760000000000005e-05,
+      "loss": 0.0455,
+      "step": 2620
+    },
+    {
+      "epoch": 0.526,
+      "grad_norm": 1.9763230085372925,
+      "learning_rate": 4.474e-05,
+      "loss": 0.0811,
+      "step": 2630
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 0.0060223303735256195,
+      "learning_rate": 4.472e-05,
+      "loss": 0.0037,
+      "step": 2640
+    },
+    {
+      "epoch": 0.53,
+      "grad_norm": 3.1282496452331543,
+      "learning_rate": 4.47e-05,
+      "loss": 0.028,
+      "step": 2650
+    },
+    {
+      "epoch": 0.532,
+      "grad_norm": 0.04501995071768761,
+      "learning_rate": 4.468e-05,
+      "loss": 0.0634,
+      "step": 2660
+    },
+    {
+      "epoch": 0.534,
+      "grad_norm": 0.9025501012802124,
+      "learning_rate": 4.466e-05,
+      "loss": 0.0253,
+      "step": 2670
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 0.013647008687257767,
+      "learning_rate": 4.4640000000000006e-05,
+      "loss": 0.0048,
+      "step": 2680
+    },
+    {
+      "epoch": 0.538,
+      "grad_norm": 0.016055762767791748,
+      "learning_rate": 4.462e-05,
+      "loss": 0.0005,
+      "step": 2690
+    },
+    {
+      "epoch": 0.54,
+      "grad_norm": 0.009782964363694191,
+      "learning_rate": 4.46e-05,
+      "loss": 0.0082,
+      "step": 2700
+    },
+    {
+      "epoch": 0.542,
+      "grad_norm": 0.008022104389965534,
+      "learning_rate": 4.458e-05,
+      "loss": 0.0188,
+      "step": 2710
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 0.0033532867673784494,
+      "learning_rate": 4.456e-05,
+      "loss": 0.0039,
+      "step": 2720
+    },
+    {
+      "epoch": 0.546,
+      "grad_norm": 0.08850495517253876,
+      "learning_rate": 4.4540000000000004e-05,
+      "loss": 0.0042,
+      "step": 2730
+    },
+    {
+      "epoch": 0.548,
+      "grad_norm": 3.59587025642395,
+      "learning_rate": 4.452e-05,
+      "loss": 0.0495,
+      "step": 2740
+    },
+    {
+      "epoch": 0.55,
+      "grad_norm": 5.187063217163086,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 0.0632,
+      "step": 2750
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 0.005760591477155685,
+      "learning_rate": 4.448e-05,
+      "loss": 0.0732,
+      "step": 2760
+    },
+    {
+      "epoch": 0.554,
+      "grad_norm": 0.11296861618757248,
+      "learning_rate": 4.4460000000000005e-05,
+      "loss": 0.0331,
+      "step": 2770
+    },
+    {
+      "epoch": 0.556,
+      "grad_norm": 0.02798079513013363,
+      "learning_rate": 4.444e-05,
+      "loss": 0.0484,
+      "step": 2780
+    },
+    {
+      "epoch": 0.558,
+      "grad_norm": 0.11013288795948029,
+      "learning_rate": 4.442e-05,
+      "loss": 0.0033,
+      "step": 2790
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 4.912937164306641,
+      "learning_rate": 4.44e-05,
+      "loss": 0.0248,
+      "step": 2800
+    },
+    {
+      "epoch": 0.562,
+      "grad_norm": 8.702404022216797,
+      "learning_rate": 4.438e-05,
+      "loss": 0.0565,
+      "step": 2810
+    },
+    {
+      "epoch": 0.564,
+      "grad_norm": 0.01780632883310318,
+      "learning_rate": 4.436e-05,
+      "loss": 0.0085,
+      "step": 2820
+    },
+    {
+      "epoch": 0.566,
+      "grad_norm": 0.008804120123386383,
+      "learning_rate": 4.4340000000000006e-05,
+      "loss": 0.0498,
+      "step": 2830
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 0.02988666482269764,
+      "learning_rate": 4.432e-05,
+      "loss": 0.0482,
+      "step": 2840
+    },
+    {
+      "epoch": 0.57,
+      "grad_norm": 0.0162635650485754,
+      "learning_rate": 4.43e-05,
+      "loss": 0.0564,
+      "step": 2850
+    },
+    {
+      "epoch": 0.572,
+      "grad_norm": 0.15490345656871796,
+      "learning_rate": 4.428e-05,
+      "loss": 0.0328,
+      "step": 2860
+    },
+    {
+      "epoch": 0.574,
+      "grad_norm": 0.017897745594382286,
+      "learning_rate": 4.426e-05,
+      "loss": 0.0025,
+      "step": 2870
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 0.2046775221824646,
+      "learning_rate": 4.424e-05,
+      "loss": 0.0018,
+      "step": 2880
+    },
+    {
+      "epoch": 0.578,
+      "grad_norm": 0.097975954413414,
+      "learning_rate": 4.422e-05,
+      "loss": 0.0231,
+      "step": 2890
+    },
+    {
+      "epoch": 0.58,
+      "grad_norm": 0.050555840134620667,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 0.0005,
+      "step": 2900
+    },
+    {
+      "epoch": 0.582,
+      "grad_norm": 0.006744217127561569,
+      "learning_rate": 4.418000000000001e-05,
+      "loss": 0.0333,
+      "step": 2910
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 0.11287890374660492,
+      "learning_rate": 4.4160000000000004e-05,
+      "loss": 0.0361,
+      "step": 2920
+    },
+    {
+      "epoch": 0.586,
+      "grad_norm": 0.03264043852686882,
+      "learning_rate": 4.414e-05,
+      "loss": 0.0316,
+      "step": 2930
+    },
+    {
+      "epoch": 0.588,
+      "grad_norm": 0.004322608467191458,
+      "learning_rate": 4.412e-05,
+      "loss": 0.0549,
+      "step": 2940
+    },
+    {
+      "epoch": 0.59,
+      "grad_norm": 0.09659099578857422,
+      "learning_rate": 4.41e-05,
+      "loss": 0.094,
+      "step": 2950
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 0.06353981792926788,
+      "learning_rate": 4.4080000000000005e-05,
+      "loss": 0.0231,
+      "step": 2960
+    },
+    {
+      "epoch": 0.594,
+      "grad_norm": 0.0059261200949549675,
+      "learning_rate": 4.406e-05,
+      "loss": 0.0008,
+      "step": 2970
+    },
+    {
+      "epoch": 0.596,
+      "grad_norm": 0.09767893701791763,
+      "learning_rate": 4.4040000000000005e-05,
+      "loss": 0.0234,
+      "step": 2980
+    },
+    {
+      "epoch": 0.598,
+      "grad_norm": 0.005928624887019396,
+      "learning_rate": 4.402e-05,
+      "loss": 0.0241,
+      "step": 2990
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 0.0034908633679151535,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 0.004,
+      "step": 3000
+    },
+    {
+      "epoch": 0.602,
+      "grad_norm": 0.18380242586135864,
+      "learning_rate": 4.398e-05,
+      "loss": 0.0699,
+      "step": 3010
+    },
+    {
+      "epoch": 0.604,
+      "grad_norm": 0.013765583746135235,
+      "learning_rate": 4.396e-05,
+      "loss": 0.0647,
+      "step": 3020
+    },
+    {
+      "epoch": 0.606,
+      "grad_norm": 0.2933019995689392,
+      "learning_rate": 4.394e-05,
+      "loss": 0.1148,
+      "step": 3030
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 0.025059092789888382,
+      "learning_rate": 4.392e-05,
+      "loss": 0.0013,
+      "step": 3040
+    },
+    {
+      "epoch": 0.61,
+      "grad_norm": 0.025903893634676933,
+      "learning_rate": 4.39e-05,
+      "loss": 0.0101,
+      "step": 3050
+    },
+    {
+      "epoch": 0.612,
+      "grad_norm": 0.02233230695128441,
+      "learning_rate": 4.388000000000001e-05,
+      "loss": 0.0929,
+      "step": 3060
+    },
+    {
+      "epoch": 0.614,
+      "grad_norm": 0.041354354470968246,
+      "learning_rate": 4.3860000000000004e-05,
+      "loss": 0.0291,
+      "step": 3070
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 0.004411335568875074,
+      "learning_rate": 4.384e-05,
+      "loss": 0.0314,
+      "step": 3080
+    },
+    {
+      "epoch": 0.618,
+      "grad_norm": 0.05570805072784424,
+      "learning_rate": 4.382e-05,
+      "loss": 0.001,
+      "step": 3090
+    },
+    {
+      "epoch": 0.62,
+      "grad_norm": 2.777430772781372,
+      "learning_rate": 4.38e-05,
+      "loss": 0.1514,
+      "step": 3100
+    },
+    {
+      "epoch": 0.622,
+      "grad_norm": 4.803562164306641,
+      "learning_rate": 4.3780000000000004e-05,
+      "loss": 0.0401,
+      "step": 3110
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 0.07038727402687073,
+      "learning_rate": 4.376e-05,
+      "loss": 0.0186,
+      "step": 3120
+    },
+    {
+      "epoch": 0.626,
+      "grad_norm": 0.024878088384866714,
+      "learning_rate": 4.3740000000000005e-05,
+      "loss": 0.0644,
+      "step": 3130
+    },
+    {
+      "epoch": 0.628,
+      "grad_norm": 0.010944769717752934,
+      "learning_rate": 4.372e-05,
+      "loss": 0.0191,
+      "step": 3140
+    },
+    {
+      "epoch": 0.63,
+      "grad_norm": 0.01699189841747284,
+      "learning_rate": 4.3700000000000005e-05,
+      "loss": 0.0154,
+      "step": 3150
+    },
+    {
+      "epoch": 0.632,
+      "grad_norm": 4.613604545593262,
+      "learning_rate": 4.368e-05,
+      "loss": 0.0147,
+      "step": 3160
+    },
+    {
+      "epoch": 0.634,
+      "grad_norm": 0.010693288408219814,
+      "learning_rate": 4.366e-05,
+      "loss": 0.0008,
+      "step": 3170
+    },
+    {
+      "epoch": 0.636,
+      "grad_norm": 0.005737672559916973,
+      "learning_rate": 4.364e-05,
+      "loss": 0.002,
+      "step": 3180
+    },
+    {
+      "epoch": 0.638,
+      "grad_norm": 0.004944519139826298,
+      "learning_rate": 4.362e-05,
+      "loss": 0.0013,
+      "step": 3190
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": 0.0188067015260458,
+      "learning_rate": 4.36e-05,
+      "loss": 0.0012,
+      "step": 3200
+    },
+    {
+      "epoch": 0.642,
+      "grad_norm": 0.0032054812181741,
+      "learning_rate": 4.3580000000000006e-05,
+      "loss": 0.0145,
+      "step": 3210
+    },
+    {
+      "epoch": 0.644,
+      "grad_norm": 0.13175809383392334,
+      "learning_rate": 4.356e-05,
+      "loss": 0.0392,
+      "step": 3220
+    },
+    {
+      "epoch": 0.646,
+      "grad_norm": 4.678634166717529,
+      "learning_rate": 4.354e-05,
+      "loss": 0.0444,
+      "step": 3230
+    },
+    {
+      "epoch": 0.648,
+      "grad_norm": 9.381738662719727,
+      "learning_rate": 4.352e-05,
+      "loss": 0.0287,
+      "step": 3240
+    },
+    {
+      "epoch": 0.65,
+      "grad_norm": 0.00795825943350792,
+      "learning_rate": 4.35e-05,
+      "loss": 0.0803,
+      "step": 3250
+    },
+    {
+      "epoch": 0.652,
+      "grad_norm": 0.04332546144723892,
+      "learning_rate": 4.3480000000000004e-05,
+      "loss": 0.0154,
+      "step": 3260
+    },
+    {
+      "epoch": 0.654,
+      "grad_norm": 0.2564730644226074,
+      "learning_rate": 4.346e-05,
+      "loss": 0.0476,
+      "step": 3270
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 3.7965617179870605,
+      "learning_rate": 4.3440000000000004e-05,
+      "loss": 0.1515,
+      "step": 3280
+    },
+    {
+      "epoch": 0.658,
+      "grad_norm": 0.0158467348664999,
+      "learning_rate": 4.342e-05,
+      "loss": 0.0529,
+      "step": 3290
+    },
+    {
+      "epoch": 0.66,
+      "grad_norm": 0.6378382444381714,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 0.0225,
+      "step": 3300
+    },
+    {
+      "epoch": 0.662,
+      "grad_norm": 3.4232382774353027,
+      "learning_rate": 4.338e-05,
+      "loss": 0.0941,
+      "step": 3310
+    },
+    {
+      "epoch": 0.664,
+      "grad_norm": 0.0431983545422554,
+      "learning_rate": 4.336e-05,
+      "loss": 0.0196,
+      "step": 3320
+    },
+    {
+      "epoch": 0.666,
+      "grad_norm": 0.15850763022899628,
+      "learning_rate": 4.334e-05,
+      "loss": 0.0229,
+      "step": 3330
+    },
+    {
+      "epoch": 0.668,
+      "grad_norm": 0.6902371644973755,
+      "learning_rate": 4.332e-05,
+      "loss": 0.0387,
+      "step": 3340
+    },
+    {
+      "epoch": 0.67,
+      "grad_norm": 0.028398212045431137,
+      "learning_rate": 4.33e-05,
+      "loss": 0.0053,
+      "step": 3350
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 0.048552609980106354,
+      "learning_rate": 4.3280000000000006e-05,
+      "loss": 0.0017,
+      "step": 3360
+    },
+    {
+      "epoch": 0.674,
+      "grad_norm": 0.011205868795514107,
+      "learning_rate": 4.326e-05,
+      "loss": 0.002,
+      "step": 3370
+    },
+    {
+      "epoch": 0.676,
+      "grad_norm": 6.537700653076172,
+      "learning_rate": 4.324e-05,
+      "loss": 0.0238,
+      "step": 3380
+    },
+    {
+      "epoch": 0.678,
+      "grad_norm": 0.03194591403007507,
+      "learning_rate": 4.3219999999999996e-05,
+      "loss": 0.0004,
+      "step": 3390
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 0.04188118129968643,
+      "learning_rate": 4.32e-05,
+      "loss": 0.0028,
+      "step": 3400
+    },
+    {
+      "epoch": 0.682,
+      "grad_norm": 1.9461021423339844,
+      "learning_rate": 4.318e-05,
+      "loss": 0.0039,
+      "step": 3410
+    },
+    {
+      "epoch": 0.684,
+      "grad_norm": 0.028302855789661407,
+      "learning_rate": 4.316e-05,
+      "loss": 0.0921,
+      "step": 3420
+    },
+    {
+      "epoch": 0.686,
+      "grad_norm": 0.06954790651798248,
+      "learning_rate": 4.3140000000000004e-05,
+      "loss": 0.0018,
+      "step": 3430
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 0.9867027401924133,
+      "learning_rate": 4.312000000000001e-05,
+      "loss": 0.0037,
+      "step": 3440
+    },
+    {
+      "epoch": 0.69,
+      "grad_norm": 0.005265855696052313,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 0.026,
+      "step": 3450
+    },
+    {
+      "epoch": 0.692,
+      "grad_norm": 0.6946307420730591,
+      "learning_rate": 4.308e-05,
+      "loss": 0.061,
+      "step": 3460
+    },
+    {
+      "epoch": 0.694,
+      "grad_norm": 0.06465889513492584,
+      "learning_rate": 4.306e-05,
+      "loss": 0.0023,
+      "step": 3470
+    },
+    {
+      "epoch": 0.696,
+      "grad_norm": 0.7860224843025208,
+      "learning_rate": 4.304e-05,
+      "loss": 0.0055,
+      "step": 3480
+    },
+    {
+      "epoch": 0.698,
+      "grad_norm": 0.04062548652291298,
+      "learning_rate": 4.3020000000000005e-05,
+      "loss": 0.0454,
+      "step": 3490
+    },
+    {
+      "epoch": 0.7,
+      "grad_norm": 0.011811433359980583,
+      "learning_rate": 4.3e-05,
+      "loss": 0.0151,
+      "step": 3500
+    },
+    {
+      "epoch": 0.702,
+      "grad_norm": 2.8018949031829834,
+      "learning_rate": 4.2980000000000005e-05,
+      "loss": 0.0043,
+      "step": 3510
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 0.6071522831916809,
+      "learning_rate": 4.296e-05,
+      "loss": 0.0011,
+      "step": 3520
+    },
+    {
+      "epoch": 0.706,
+      "grad_norm": 0.032536111772060394,
+      "learning_rate": 4.2940000000000006e-05,
+      "loss": 0.0034,
+      "step": 3530
+    },
+    {
+      "epoch": 0.708,
+      "grad_norm": 0.006422733422368765,
+      "learning_rate": 4.292e-05,
+      "loss": 0.0019,
+      "step": 3540
+    },
+    {
+      "epoch": 0.71,
+      "grad_norm": 4.8893818855285645,
+      "learning_rate": 4.29e-05,
+      "loss": 0.0786,
+      "step": 3550
+    },
+    {
+      "epoch": 0.712,
+      "grad_norm": 0.009687143377959728,
+      "learning_rate": 4.288e-05,
+      "loss": 0.0002,
+      "step": 3560
+    },
+    {
+      "epoch": 0.714,
+      "grad_norm": 0.02869969606399536,
+      "learning_rate": 4.286e-05,
+      "loss": 0.0818,
+      "step": 3570
+    },
+    {
+      "epoch": 0.716,
+      "grad_norm": 0.02002692222595215,
+      "learning_rate": 4.284e-05,
+      "loss": 0.0002,
+      "step": 3580
+    },
+    {
+      "epoch": 0.718,
+      "grad_norm": 0.12229043245315552,
+      "learning_rate": 4.282000000000001e-05,
+      "loss": 0.0034,
+      "step": 3590
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": 0.9271475076675415,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 0.0228,
+      "step": 3600
+    },
+    {
+      "epoch": 0.722,
+      "grad_norm": 0.015138584189116955,
+      "learning_rate": 4.278e-05,
+      "loss": 0.1131,
+      "step": 3610
+    },
+    {
+      "epoch": 0.724,
+      "grad_norm": 0.028255993500351906,
+      "learning_rate": 4.276e-05,
+      "loss": 0.0303,
+      "step": 3620
+    },
+    {
+      "epoch": 0.726,
+      "grad_norm": 0.03335114195942879,
+      "learning_rate": 4.274e-05,
+      "loss": 0.0049,
+      "step": 3630
+    },
+    {
+      "epoch": 0.728,
+      "grad_norm": 0.005619376432150602,
+      "learning_rate": 4.2720000000000004e-05,
+      "loss": 0.0018,
+      "step": 3640
+    },
+    {
+      "epoch": 0.73,
+      "grad_norm": 7.768834114074707,
+      "learning_rate": 4.27e-05,
+      "loss": 0.0237,
+      "step": 3650
+    },
+    {
+      "epoch": 0.732,
+      "grad_norm": 0.013384620659053326,
+      "learning_rate": 4.2680000000000005e-05,
+      "loss": 0.0041,
+      "step": 3660
+    },
+    {
+      "epoch": 0.734,
+      "grad_norm": 0.02822156995534897,
+      "learning_rate": 4.266e-05,
+      "loss": 0.0932,
+      "step": 3670
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 0.006292347330600023,
+      "learning_rate": 4.2640000000000005e-05,
+      "loss": 0.0019,
+      "step": 3680
+    },
+    {
+      "epoch": 0.738,
+      "grad_norm": 0.04632676765322685,
+      "learning_rate": 4.262e-05,
+      "loss": 0.0387,
+      "step": 3690
+    },
+    {
+      "epoch": 0.74,
+      "grad_norm": 0.011406010948121548,
+      "learning_rate": 4.26e-05,
+      "loss": 0.0383,
+      "step": 3700
+    },
+    {
+      "epoch": 0.742,
+      "grad_norm": 0.015299669466912746,
+      "learning_rate": 4.258e-05,
+      "loss": 0.0507,
+      "step": 3710
+    },
+    {
+      "epoch": 0.744,
+      "grad_norm": 0.002912426833063364,
+      "learning_rate": 4.256e-05,
+      "loss": 0.0136,
+      "step": 3720
+    },
+    {
+      "epoch": 0.746,
+      "grad_norm": 7.015591144561768,
+      "learning_rate": 4.254e-05,
+      "loss": 0.0677,
+      "step": 3730
+    },
+    {
+      "epoch": 0.748,
+      "grad_norm": 0.029005423188209534,
+      "learning_rate": 4.2520000000000006e-05,
+      "loss": 0.048,
+      "step": 3740
+    },
+    {
+      "epoch": 0.75,
+      "grad_norm": 0.006496680434793234,
+      "learning_rate": 4.25e-05,
+      "loss": 0.0174,
+      "step": 3750
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 0.005983166862279177,
+      "learning_rate": 4.248e-05,
+      "loss": 0.0023,
+      "step": 3760
+    },
+    {
+      "epoch": 0.754,
+      "grad_norm": 5.252658843994141,
+      "learning_rate": 4.246e-05,
+      "loss": 0.0902,
+      "step": 3770
+    },
+    {
+      "epoch": 0.756,
+      "grad_norm": 0.008069785311818123,
+      "learning_rate": 4.244e-05,
+      "loss": 0.004,
+      "step": 3780
+    },
+    {
+      "epoch": 0.758,
+      "grad_norm": 2.6544532775878906,
+      "learning_rate": 4.2420000000000004e-05,
+      "loss": 0.057,
+      "step": 3790
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 0.013155023567378521,
+      "learning_rate": 4.24e-05,
+      "loss": 0.0066,
+      "step": 3800
+    },
+    {
+      "epoch": 0.762,
+      "grad_norm": 4.488767623901367,
+      "learning_rate": 4.2380000000000004e-05,
+      "loss": 0.0157,
+      "step": 3810
+    },
+    {
+      "epoch": 0.764,
+      "grad_norm": 0.006404740270227194,
+      "learning_rate": 4.236e-05,
+      "loss": 0.0496,
+      "step": 3820
+    },
+    {
+      "epoch": 0.766,
+      "grad_norm": 0.027893809601664543,
+      "learning_rate": 4.2340000000000005e-05,
+      "loss": 0.0393,
+      "step": 3830
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 0.2745389938354492,
+      "learning_rate": 4.232e-05,
+      "loss": 0.0013,
+      "step": 3840
+    },
+    {
+      "epoch": 0.77,
+      "grad_norm": 5.9310197830200195,
+      "learning_rate": 4.23e-05,
+      "loss": 0.0727,
+      "step": 3850
+    },
+    {
+      "epoch": 0.772,
+      "grad_norm": 0.014957119710743427,
+      "learning_rate": 4.228e-05,
+      "loss": 0.0184,
+      "step": 3860
+    },
+    {
+      "epoch": 0.774,
+      "grad_norm": 0.047319989651441574,
+      "learning_rate": 4.226e-05,
+      "loss": 0.038,
+      "step": 3870
+    },
+    {
+      "epoch": 0.776,
+      "grad_norm": 0.0042962199077010155,
+      "learning_rate": 4.224e-05,
+      "loss": 0.0041,
+      "step": 3880
+    },
+    {
+      "epoch": 0.778,
+      "grad_norm": 0.0794585719704628,
+      "learning_rate": 4.2220000000000006e-05,
+      "loss": 0.0008,
+      "step": 3890
+    },
+    {
+      "epoch": 0.78,
+      "grad_norm": 0.015754858031868935,
+      "learning_rate": 4.22e-05,
+      "loss": 0.0148,
+      "step": 3900
+    },
+    {
+      "epoch": 0.782,
+      "grad_norm": 0.020532747730612755,
+      "learning_rate": 4.2180000000000006e-05,
+      "loss": 0.0345,
+      "step": 3910
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 0.0342375785112381,
+      "learning_rate": 4.2159999999999996e-05,
+      "loss": 0.0027,
+      "step": 3920
+    },
+    {
+      "epoch": 0.786,
+      "grad_norm": 0.4745217561721802,
+      "learning_rate": 4.214e-05,
+      "loss": 0.0126,
+      "step": 3930
+    },
+    {
+      "epoch": 0.788,
+      "grad_norm": 0.0014779713237658143,
+      "learning_rate": 4.212e-05,
+      "loss": 0.1061,
+      "step": 3940
+    },
+    {
+      "epoch": 0.79,
+      "grad_norm": 0.043706584721803665,
+      "learning_rate": 4.21e-05,
+      "loss": 0.0195,
+      "step": 3950
+    },
+    {
+      "epoch": 0.792,
+      "grad_norm": 0.04930654540657997,
+      "learning_rate": 4.2080000000000004e-05,
+      "loss": 0.0029,
+      "step": 3960
+    },
+    {
+      "epoch": 0.794,
+      "grad_norm": 0.025867952033877373,
+      "learning_rate": 4.206e-05,
+      "loss": 0.0709,
+      "step": 3970
+    },
+    {
+      "epoch": 0.796,
+      "grad_norm": 0.6700100898742676,
+      "learning_rate": 4.2040000000000004e-05,
+      "loss": 0.0149,
+      "step": 3980
+    },
+    {
+      "epoch": 0.798,
+      "grad_norm": 0.0404072068631649,
+      "learning_rate": 4.202e-05,
+      "loss": 0.0251,
+      "step": 3990
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 0.008657311089336872,
+      "learning_rate": 4.2e-05,
+      "loss": 0.0163,
+      "step": 4000
+    },
+    {
+      "epoch": 0.802,
+      "grad_norm": 0.02054884470999241,
+      "learning_rate": 4.198e-05,
+      "loss": 0.0374,
+      "step": 4010
+    },
+    {
+      "epoch": 0.804,
+      "grad_norm": 0.01718149520456791,
+      "learning_rate": 4.196e-05,
+      "loss": 0.0288,
+      "step": 4020
+    },
+    {
+      "epoch": 0.806,
+      "grad_norm": 0.05612902343273163,
+      "learning_rate": 4.194e-05,
+      "loss": 0.0184,
+      "step": 4030
+    },
+    {
+      "epoch": 0.808,
+      "grad_norm": 0.009326531551778316,
+      "learning_rate": 4.1920000000000005e-05,
+      "loss": 0.0023,
+      "step": 4040
+    },
+    {
+      "epoch": 0.81,
+      "grad_norm": 0.09644528478384018,
+      "learning_rate": 4.19e-05,
+      "loss": 0.0184,
+      "step": 4050
+    },
+    {
+      "epoch": 0.812,
+      "grad_norm": 0.011878632009029388,
+      "learning_rate": 4.1880000000000006e-05,
+      "loss": 0.0311,
+      "step": 4060
+    },
+    {
+      "epoch": 0.814,
+      "grad_norm": 0.014384951442480087,
+      "learning_rate": 4.186e-05,
+      "loss": 0.0041,
+      "step": 4070
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 5.182435035705566,
+      "learning_rate": 4.184e-05,
+      "loss": 0.0216,
+      "step": 4080
+    },
+    {
+      "epoch": 0.818,
+      "grad_norm": 5.221963882446289,
+      "learning_rate": 4.182e-05,
+      "loss": 0.051,
+      "step": 4090
+    },
+    {
+      "epoch": 0.82,
+      "grad_norm": 0.01110916119068861,
+      "learning_rate": 4.18e-05,
+      "loss": 0.0223,
+      "step": 4100
+    },
+    {
+      "epoch": 0.822,
+      "grad_norm": 0.062016647309064865,
+      "learning_rate": 4.178e-05,
+      "loss": 0.0102,
+      "step": 4110
+    },
+    {
+      "epoch": 0.824,
+      "grad_norm": 0.6081287860870361,
+      "learning_rate": 4.176000000000001e-05,
+      "loss": 0.0168,
+      "step": 4120
+    },
+    {
+      "epoch": 0.826,
+      "grad_norm": 0.010736220516264439,
+      "learning_rate": 4.1740000000000004e-05,
+      "loss": 0.0153,
+      "step": 4130
+    },
+    {
+      "epoch": 0.828,
+      "grad_norm": 0.03323579207062721,
+      "learning_rate": 4.172e-05,
+      "loss": 0.0224,
+      "step": 4140
+    },
+    {
+      "epoch": 0.83,
+      "grad_norm": 0.7267991304397583,
+      "learning_rate": 4.17e-05,
+      "loss": 0.0033,
+      "step": 4150
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 0.029643530026078224,
+      "learning_rate": 4.168e-05,
+      "loss": 0.0077,
+      "step": 4160
+    },
+    {
+      "epoch": 0.834,
+      "grad_norm": 0.014343446120619774,
+      "learning_rate": 4.1660000000000004e-05,
+      "loss": 0.0025,
+      "step": 4170
+    },
+    {
+      "epoch": 0.836,
+      "grad_norm": 0.007818669080734253,
+      "learning_rate": 4.164e-05,
+      "loss": 0.0254,
+      "step": 4180
+    },
+    {
+      "epoch": 0.838,
+      "grad_norm": 0.004449554719030857,
+      "learning_rate": 4.1620000000000005e-05,
+      "loss": 0.0005,
+      "step": 4190
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 0.005323763936758041,
+      "learning_rate": 4.16e-05,
+      "loss": 0.0761,
+      "step": 4200
+    },
+    {
+      "epoch": 0.842,
+      "grad_norm": 0.17715303599834442,
+      "learning_rate": 4.1580000000000005e-05,
+      "loss": 0.0098,
+      "step": 4210
+    },
+    {
+      "epoch": 0.844,
+      "grad_norm": 0.013948998413980007,
+      "learning_rate": 4.156e-05,
+      "loss": 0.0105,
+      "step": 4220
+    },
+    {
+      "epoch": 0.846,
+      "grad_norm": 0.029932260513305664,
+      "learning_rate": 4.154e-05,
+      "loss": 0.0022,
+      "step": 4230
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 5.534884929656982,
+      "learning_rate": 4.152e-05,
+      "loss": 0.0178,
+      "step": 4240
+    },
+    {
+      "epoch": 0.85,
+      "grad_norm": 2.4610676765441895,
+      "learning_rate": 4.15e-05,
+      "loss": 0.0397,
+      "step": 4250
+    },
+    {
+      "epoch": 0.852,
+      "grad_norm": 0.00479121832177043,
+      "learning_rate": 4.148e-05,
+      "loss": 0.0056,
+      "step": 4260
+    },
+    {
+      "epoch": 0.854,
+      "grad_norm": 0.4627273976802826,
+      "learning_rate": 4.1460000000000006e-05,
+      "loss": 0.0665,
+      "step": 4270
+    },
+    {
+      "epoch": 0.856,
+      "grad_norm": 0.022942064329981804,
+      "learning_rate": 4.144e-05,
+      "loss": 0.0586,
+      "step": 4280
+    },
+    {
+      "epoch": 0.858,
+      "grad_norm": 0.0016935011371970177,
+      "learning_rate": 4.142000000000001e-05,
+      "loss": 0.0358,
+      "step": 4290
+    },
+    {
+      "epoch": 0.86,
+      "grad_norm": 4.48838472366333,
+      "learning_rate": 4.14e-05,
+      "loss": 0.0626,
+      "step": 4300
+    },
+    {
+      "epoch": 0.862,
+      "grad_norm": 0.0164650809019804,
+      "learning_rate": 4.138e-05,
+      "loss": 0.0551,
+      "step": 4310
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 0.012818397022783756,
+      "learning_rate": 4.1360000000000004e-05,
+      "loss": 0.0031,
+      "step": 4320
+    },
+    {
+      "epoch": 0.866,
+      "grad_norm": 0.02281295880675316,
+      "learning_rate": 4.134e-05,
+      "loss": 0.0713,
+      "step": 4330
+    },
+    {
+      "epoch": 0.868,
+      "grad_norm": 0.37325066328048706,
+      "learning_rate": 4.1320000000000004e-05,
+      "loss": 0.0345,
+      "step": 4340
+    },
+    {
+      "epoch": 0.87,
+      "grad_norm": 0.022145600989460945,
+      "learning_rate": 4.13e-05,
+      "loss": 0.0285,
+      "step": 4350
+    },
+    {
+      "epoch": 0.872,
+      "grad_norm": 0.006575500126928091,
+      "learning_rate": 4.1280000000000005e-05,
+      "loss": 0.0099,
+      "step": 4360
+    },
+    {
+      "epoch": 0.874,
+      "grad_norm": 0.03781846538186073,
+      "learning_rate": 4.126e-05,
+      "loss": 0.0673,
+      "step": 4370
+    },
+    {
+      "epoch": 0.876,
+      "grad_norm": 0.004571052268147469,
+      "learning_rate": 4.124e-05,
+      "loss": 0.015,
+      "step": 4380
+    },
+    {
+      "epoch": 0.878,
+      "grad_norm": 0.01689932681620121,
+      "learning_rate": 4.122e-05,
+      "loss": 0.0311,
+      "step": 4390
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 0.06380756199359894,
+      "learning_rate": 4.12e-05,
+      "loss": 0.0041,
+      "step": 4400
+    },
+    {
+      "epoch": 0.882,
+      "grad_norm": 9.5936279296875,
+      "learning_rate": 4.118e-05,
+      "loss": 0.0814,
+      "step": 4410
+    },
+    {
+      "epoch": 0.884,
+      "grad_norm": 0.013598248362541199,
+      "learning_rate": 4.1160000000000006e-05,
+      "loss": 0.0174,
+      "step": 4420
+    },
+    {
+      "epoch": 0.886,
+      "grad_norm": 0.0035734116099774837,
+      "learning_rate": 4.114e-05,
+      "loss": 0.141,
+      "step": 4430
+    },
+    {
+      "epoch": 0.888,
+      "grad_norm": 0.025886977091431618,
+      "learning_rate": 4.1120000000000006e-05,
+      "loss": 0.0285,
+      "step": 4440
+    },
+    {
+      "epoch": 0.89,
+      "grad_norm": 4.5155487060546875,
+      "learning_rate": 4.11e-05,
+      "loss": 0.0211,
+      "step": 4450
+    },
+    {
+      "epoch": 0.892,
+      "grad_norm": 0.05487843602895737,
+      "learning_rate": 4.108e-05,
+      "loss": 0.0241,
+      "step": 4460
+    },
+    {
+      "epoch": 0.894,
+      "grad_norm": 0.013455720618367195,
+      "learning_rate": 4.106e-05,
+      "loss": 0.036,
+      "step": 4470
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 0.012445429340004921,
+      "learning_rate": 4.104e-05,
+      "loss": 0.0015,
+      "step": 4480
+    },
+    {
+      "epoch": 0.898,
+      "grad_norm": 0.07584936916828156,
+      "learning_rate": 4.1020000000000004e-05,
+      "loss": 0.064,
+      "step": 4490
+    },
+    {
+      "epoch": 0.9,
+      "grad_norm": 0.004590052645653486,
+      "learning_rate": 4.1e-05,
+      "loss": 0.0007,
+      "step": 4500
+    },
+    {
+      "epoch": 0.902,
+      "grad_norm": 0.03764956444501877,
+      "learning_rate": 4.0980000000000004e-05,
+      "loss": 0.0008,
+      "step": 4510
+    },
+    {
+      "epoch": 0.904,
+      "grad_norm": 0.019499754533171654,
+      "learning_rate": 4.096e-05,
+      "loss": 0.0052,
+      "step": 4520
+    },
+    {
+      "epoch": 0.906,
+      "grad_norm": 0.004517500754445791,
+      "learning_rate": 4.094e-05,
+      "loss": 0.0004,
+      "step": 4530
+    },
+    {
+      "epoch": 0.908,
+      "grad_norm": 0.007550814189016819,
+      "learning_rate": 4.092e-05,
+      "loss": 0.0202,
+      "step": 4540
+    },
+    {
+      "epoch": 0.91,
+      "grad_norm": 0.00612619798630476,
+      "learning_rate": 4.09e-05,
+      "loss": 0.035,
+      "step": 4550
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 0.004684742074459791,
+      "learning_rate": 4.088e-05,
+      "loss": 0.0004,
+      "step": 4560
+    },
+    {
+      "epoch": 0.914,
+      "grad_norm": 0.0021801223047077656,
+      "learning_rate": 4.0860000000000005e-05,
+      "loss": 0.0505,
+      "step": 4570
+    },
+    {
+      "epoch": 0.916,
+      "grad_norm": 0.06538788974285126,
+      "learning_rate": 4.084e-05,
+      "loss": 0.0037,
+      "step": 4580
+    },
+    {
+      "epoch": 0.918,
+      "grad_norm": 0.007073140237480402,
+      "learning_rate": 4.0820000000000006e-05,
+      "loss": 0.1086,
+      "step": 4590
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 4.945827484130859,
+      "learning_rate": 4.08e-05,
+      "loss": 0.021,
+      "step": 4600
+    },
+    {
+      "epoch": 0.922,
+      "grad_norm": 0.04866447672247887,
+      "learning_rate": 4.078e-05,
+      "loss": 0.0444,
+      "step": 4610
+    },
+    {
+      "epoch": 0.924,
+      "grad_norm": 0.004719861783087254,
+      "learning_rate": 4.076e-05,
+      "loss": 0.0752,
+      "step": 4620
+    },
+    {
+      "epoch": 0.926,
+      "grad_norm": 0.8237816095352173,
+      "learning_rate": 4.074e-05,
+      "loss": 0.0286,
+      "step": 4630
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 0.12901705503463745,
+      "learning_rate": 4.072e-05,
+      "loss": 0.0099,
+      "step": 4640
+    },
+    {
+      "epoch": 0.93,
+      "grad_norm": 0.03178408369421959,
+      "learning_rate": 4.07e-05,
+      "loss": 0.032,
+      "step": 4650
+    },
+    {
+      "epoch": 0.932,
+      "grad_norm": 0.04729364439845085,
+      "learning_rate": 4.0680000000000004e-05,
+      "loss": 0.0386,
+      "step": 4660
+    },
+    {
+      "epoch": 0.934,
+      "grad_norm": 0.022903740406036377,
+      "learning_rate": 4.066e-05,
+      "loss": 0.0175,
+      "step": 4670
+    },
+    {
+      "epoch": 0.936,
+      "grad_norm": 0.006164549384266138,
+      "learning_rate": 4.064e-05,
+      "loss": 0.0022,
+      "step": 4680
+    },
+    {
+      "epoch": 0.938,
+      "grad_norm": 0.004589389078319073,
+      "learning_rate": 4.062e-05,
+      "loss": 0.066,
+      "step": 4690
+    },
+    {
+      "epoch": 0.94,
+      "grad_norm": 0.012154348194599152,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 0.0731,
+      "step": 4700
+    },
+    {
+      "epoch": 0.942,
+      "grad_norm": 6.234540939331055,
+      "learning_rate": 4.058e-05,
+      "loss": 0.0222,
+      "step": 4710
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 0.17935433983802795,
+      "learning_rate": 4.0560000000000005e-05,
+      "loss": 0.0019,
+      "step": 4720
+    },
+    {
+      "epoch": 0.946,
+      "grad_norm": 0.21610985696315765,
+      "learning_rate": 4.054e-05,
+      "loss": 0.0588,
+      "step": 4730
+    },
+    {
+      "epoch": 0.948,
+      "grad_norm": 0.010893936268985271,
+      "learning_rate": 4.0520000000000005e-05,
+      "loss": 0.0124,
+      "step": 4740
+    },
+    {
+      "epoch": 0.95,
+      "grad_norm": 0.027378089725971222,
+      "learning_rate": 4.05e-05,
+      "loss": 0.0012,
+      "step": 4750
+    },
+    {
+      "epoch": 0.952,
+      "grad_norm": 0.19339287281036377,
+      "learning_rate": 4.048e-05,
+      "loss": 0.0696,
+      "step": 4760
+    },
+    {
+      "epoch": 0.954,
+      "grad_norm": 3.2441658973693848,
+      "learning_rate": 4.046e-05,
+      "loss": 0.041,
+      "step": 4770
+    },
+    {
+      "epoch": 0.956,
+      "grad_norm": 0.012117883190512657,
+      "learning_rate": 4.044e-05,
+      "loss": 0.0396,
+      "step": 4780
+    },
+    {
+      "epoch": 0.958,
+      "grad_norm": 0.018160009756684303,
+      "learning_rate": 4.042e-05,
+      "loss": 0.0009,
+      "step": 4790
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 0.005018259398639202,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 0.0006,
+      "step": 4800
+    },
+    {
+      "epoch": 0.962,
+      "grad_norm": 1.8187236785888672,
+      "learning_rate": 4.038e-05,
+      "loss": 0.0034,
+      "step": 4810
+    },
+    {
+      "epoch": 0.964,
+      "grad_norm": 0.01619546115398407,
+      "learning_rate": 4.0360000000000007e-05,
+      "loss": 0.0138,
+      "step": 4820
+    },
+    {
+      "epoch": 0.966,
+      "grad_norm": 0.0023929474409669638,
+      "learning_rate": 4.034e-05,
+      "loss": 0.0542,
+      "step": 4830
+    },
+    {
+      "epoch": 0.968,
+      "grad_norm": 0.36933594942092896,
+      "learning_rate": 4.032e-05,
+      "loss": 0.0055,
+      "step": 4840
+    },
+    {
+      "epoch": 0.97,
+      "grad_norm": 0.0013171678874641657,
+      "learning_rate": 4.0300000000000004e-05,
+      "loss": 0.0121,
+      "step": 4850
+    },
+    {
+      "epoch": 0.972,
+      "grad_norm": 0.0078110783360898495,
+      "learning_rate": 4.028e-05,
+      "loss": 0.0356,
+      "step": 4860
+    },
+    {
+      "epoch": 0.974,
+      "grad_norm": 0.0026302209589630365,
+      "learning_rate": 4.0260000000000004e-05,
+      "loss": 0.0472,
+      "step": 4870
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 0.008590531535446644,
+      "learning_rate": 4.024e-05,
+      "loss": 0.0344,
+      "step": 4880
+    },
+    {
+      "epoch": 0.978,
+      "grad_norm": 0.0045235357247292995,
+      "learning_rate": 4.0220000000000005e-05,
+      "loss": 0.0314,
+      "step": 4890
+    },
+    {
+      "epoch": 0.98,
+      "grad_norm": 0.012373453937470913,
+      "learning_rate": 4.02e-05,
+      "loss": 0.0508,
+      "step": 4900
+    },
+    {
+      "epoch": 0.982,
+      "grad_norm": 4.436814785003662,
+      "learning_rate": 4.018e-05,
+      "loss": 0.0161,
+      "step": 4910
+    },
+    {
+      "epoch": 0.984,
+      "grad_norm": 0.010414298623800278,
+      "learning_rate": 4.016e-05,
+      "loss": 0.058,
+      "step": 4920
+    },
+    {
+      "epoch": 0.986,
+      "grad_norm": 0.17478038370609283,
+      "learning_rate": 4.014e-05,
+      "loss": 0.002,
+      "step": 4930
+    },
+    {
+      "epoch": 0.988,
+      "grad_norm": 0.11370497941970825,
+      "learning_rate": 4.012e-05,
+      "loss": 0.018,
+      "step": 4940
+    },
+    {
+      "epoch": 0.99,
+      "grad_norm": 0.00878177396953106,
+      "learning_rate": 4.0100000000000006e-05,
+      "loss": 0.0273,
+      "step": 4950
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 0.004751706030219793,
+      "learning_rate": 4.008e-05,
+      "loss": 0.0365,
+      "step": 4960
+    },
+    {
+      "epoch": 0.994,
+      "grad_norm": 0.6816898584365845,
+      "learning_rate": 4.0060000000000006e-05,
+      "loss": 0.0021,
+      "step": 4970
+    },
+    {
+      "epoch": 0.996,
+      "grad_norm": 0.023028748109936714,
+      "learning_rate": 4.004e-05,
+      "loss": 0.0046,
+      "step": 4980
+    },
+    {
+      "epoch": 0.998,
+      "grad_norm": 0.008332543075084686,
+      "learning_rate": 4.002e-05,
+      "loss": 0.0289,
+      "step": 4990
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": 0.7706586718559265,
+      "learning_rate": 4e-05,
+      "loss": 0.0807,
+      "step": 5000
+    },
+    {
+      "epoch": 1.0,
+      "eval_loss": 0.016496026888489723,
+      "eval_runtime": 91.0716,
+      "eval_samples_per_second": 109.804,
+      "eval_steps_per_second": 13.725,
+      "step": 5000
+    },
+    {
+      "epoch": 1.002,
+      "grad_norm": 0.3708021640777588,
+      "learning_rate": 3.998e-05,
+      "loss": 0.072,
+      "step": 5010
+    },
+    {
+      "epoch": 1.004,
+      "grad_norm": 0.47253674268722534,
+      "learning_rate": 3.9960000000000004e-05,
+      "loss": 0.0386,
+      "step": 5020
+    },
+    {
+      "epoch": 1.006,
+      "grad_norm": 0.04058856889605522,
+      "learning_rate": 3.994e-05,
+      "loss": 0.022,
+      "step": 5030
+    },
+    {
+      "epoch": 1.008,
+      "grad_norm": 0.0035456023178994656,
+      "learning_rate": 3.9920000000000004e-05,
+      "loss": 0.008,
+      "step": 5040
+    },
+    {
+      "epoch": 1.01,
+      "grad_norm": 0.00531780207529664,
+      "learning_rate": 3.99e-05,
+      "loss": 0.0622,
+      "step": 5050
+    },
+    {
+      "epoch": 1.012,
+      "grad_norm": 3.454925298690796,
+      "learning_rate": 3.988e-05,
+      "loss": 0.0087,
+      "step": 5060
+    },
+    {
+      "epoch": 1.014,
+      "grad_norm": 0.017686668783426285,
+      "learning_rate": 3.986e-05,
+      "loss": 0.0291,
+      "step": 5070
+    },
+    {
+      "epoch": 1.016,
+      "grad_norm": 0.007442487869411707,
+      "learning_rate": 3.984e-05,
+      "loss": 0.0094,
+      "step": 5080
+    },
+    {
+      "epoch": 1.018,
+      "grad_norm": 0.12783612310886383,
+      "learning_rate": 3.982e-05,
+      "loss": 0.004,
+      "step": 5090
+    },
+    {
+      "epoch": 1.02,
+      "grad_norm": 0.4321322441101074,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 0.001,
+      "step": 5100
+    },
+    {
+      "epoch": 1.022,
+      "grad_norm": 0.008714113384485245,
+      "learning_rate": 3.978e-05,
+      "loss": 0.0153,
+      "step": 5110
+    },
+    {
+      "epoch": 1.024,
+      "grad_norm": 0.8053523302078247,
+      "learning_rate": 3.9760000000000006e-05,
+      "loss": 0.0015,
+      "step": 5120
+    },
+    {
+      "epoch": 1.026,
+      "grad_norm": 0.007202483713626862,
+      "learning_rate": 3.974e-05,
+      "loss": 0.0638,
+      "step": 5130
+    },
+    {
+      "epoch": 1.028,
+      "grad_norm": 0.005011177621781826,
+      "learning_rate": 3.972e-05,
+      "loss": 0.015,
+      "step": 5140
+    },
+    {
+      "epoch": 1.03,
+      "grad_norm": 0.014276087284088135,
+      "learning_rate": 3.97e-05,
+      "loss": 0.0396,
+      "step": 5150
+    },
+    {
+      "epoch": 1.032,
+      "grad_norm": 0.02816883847117424,
+      "learning_rate": 3.968e-05,
+      "loss": 0.0009,
+      "step": 5160
+    },
+    {
+      "epoch": 1.034,
+      "grad_norm": 0.012690484523773193,
+      "learning_rate": 3.966e-05,
+      "loss": 0.0004,
+      "step": 5170
+    },
+    {
+      "epoch": 1.036,
+      "grad_norm": 0.01476860698312521,
+      "learning_rate": 3.964e-05,
+      "loss": 0.0265,
+      "step": 5180
+    },
+    {
+      "epoch": 1.038,
+      "grad_norm": 0.04537089914083481,
+      "learning_rate": 3.9620000000000004e-05,
+      "loss": 0.0569,
+      "step": 5190
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 0.004969718400388956,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 0.0908,
+      "step": 5200
+    },
+    {
+      "epoch": 1.042,
+      "grad_norm": 0.03211864456534386,
+      "learning_rate": 3.958e-05,
+      "loss": 0.0028,
+      "step": 5210
+    },
+    {
+      "epoch": 1.044,
+      "grad_norm": 0.058294814079999924,
+      "learning_rate": 3.956e-05,
+      "loss": 0.001,
+      "step": 5220
+    },
+    {
+      "epoch": 1.046,
+      "grad_norm": 0.015622718259692192,
+      "learning_rate": 3.954e-05,
+      "loss": 0.017,
+      "step": 5230
+    },
+    {
+      "epoch": 1.048,
+      "grad_norm": 0.014199018478393555,
+      "learning_rate": 3.952e-05,
+      "loss": 0.0462,
+      "step": 5240
+    },
+    {
+      "epoch": 1.05,
+      "grad_norm": 5.004390716552734,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 0.0108,
+      "step": 5250
+    },
+    {
+      "epoch": 1.052,
+      "grad_norm": 0.0037176634650677443,
+      "learning_rate": 3.948e-05,
+      "loss": 0.0018,
+      "step": 5260
+    },
+    {
+      "epoch": 1.054,
+      "grad_norm": 0.19786491990089417,
+      "learning_rate": 3.9460000000000005e-05,
+      "loss": 0.0017,
+      "step": 5270
+    },
+    {
+      "epoch": 1.056,
+      "grad_norm": 0.002558846725150943,
+      "learning_rate": 3.944e-05,
+      "loss": 0.0425,
+      "step": 5280
+    },
+    {
+      "epoch": 1.058,
+      "grad_norm": 0.044299229979515076,
+      "learning_rate": 3.942e-05,
+      "loss": 0.0135,
+      "step": 5290
+    },
+    {
+      "epoch": 1.06,
+      "grad_norm": 0.027034727856516838,
+      "learning_rate": 3.94e-05,
+      "loss": 0.0005,
+      "step": 5300
+    },
+    {
+      "epoch": 1.062,
+      "grad_norm": 0.009834056720137596,
+      "learning_rate": 3.938e-05,
+      "loss": 0.019,
+      "step": 5310
+    },
+    {
+      "epoch": 1.064,
+      "grad_norm": 0.11118980497121811,
+      "learning_rate": 3.936e-05,
+      "loss": 0.0018,
+      "step": 5320
+    },
+    {
+      "epoch": 1.066,
+      "grad_norm": 0.05060764402151108,
+      "learning_rate": 3.9340000000000006e-05,
+      "loss": 0.0017,
+      "step": 5330
+    },
+    {
+      "epoch": 1.068,
+      "grad_norm": 0.011010345071554184,
+      "learning_rate": 3.932e-05,
+      "loss": 0.002,
+      "step": 5340
+    },
+    {
+      "epoch": 1.07,
+      "grad_norm": 0.005932790692895651,
+      "learning_rate": 3.9300000000000007e-05,
+      "loss": 0.0043,
+      "step": 5350
+    },
+    {
+      "epoch": 1.072,
+      "grad_norm": 0.10913924872875214,
+      "learning_rate": 3.9280000000000003e-05,
+      "loss": 0.0006,
+      "step": 5360
+    },
+    {
+      "epoch": 1.074,
+      "grad_norm": 0.030183983966708183,
+      "learning_rate": 3.926e-05,
+      "loss": 0.0758,
+      "step": 5370
+    },
+    {
+      "epoch": 1.076,
+      "grad_norm": 0.00351043906994164,
+      "learning_rate": 3.9240000000000004e-05,
+      "loss": 0.0004,
+      "step": 5380
+    },
+    {
+      "epoch": 1.078,
+      "grad_norm": 0.018562721088528633,
+      "learning_rate": 3.922e-05,
+      "loss": 0.0348,
+      "step": 5390
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 0.015087517909705639,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 0.002,
+      "step": 5400
+    },
+    {
+      "epoch": 1.082,
+      "grad_norm": 0.017430707812309265,
+      "learning_rate": 3.918e-05,
+      "loss": 0.0093,
+      "step": 5410
+    },
+    {
+      "epoch": 1.084,
+      "grad_norm": 3.7334983348846436,
+      "learning_rate": 3.9160000000000005e-05,
+      "loss": 0.0808,
+      "step": 5420
+    },
+    {
+      "epoch": 1.086,
+      "grad_norm": 4.454714298248291,
+      "learning_rate": 3.914e-05,
+      "loss": 0.109,
+      "step": 5430
+    },
+    {
+      "epoch": 1.088,
+      "grad_norm": 0.027748622000217438,
+      "learning_rate": 3.912e-05,
+      "loss": 0.0336,
+      "step": 5440
+    },
+    {
+      "epoch": 1.09,
+      "grad_norm": 0.01548692025244236,
+      "learning_rate": 3.91e-05,
+      "loss": 0.002,
+      "step": 5450
+    },
+    {
+      "epoch": 1.092,
+      "grad_norm": 0.008610161021351814,
+      "learning_rate": 3.908e-05,
+      "loss": 0.0358,
+      "step": 5460
+    },
+    {
+      "epoch": 1.094,
+      "grad_norm": 0.00959813967347145,
+      "learning_rate": 3.906e-05,
+      "loss": 0.0432,
+      "step": 5470
+    },
+    {
+      "epoch": 1.096,
+      "grad_norm": 0.006911910604685545,
+      "learning_rate": 3.9040000000000006e-05,
+      "loss": 0.0312,
+      "step": 5480
+    },
+    {
+      "epoch": 1.098,
+      "grad_norm": 0.19846265017986298,
+      "learning_rate": 3.902e-05,
+      "loss": 0.0019,
+      "step": 5490
+    },
+    {
+      "epoch": 1.1,
+      "grad_norm": 0.028356514871120453,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 0.0291,
+      "step": 5500
+    },
+    {
+      "epoch": 1.102,
+      "grad_norm": 0.007134539540857077,
+      "learning_rate": 3.898e-05,
+      "loss": 0.0305,
+      "step": 5510
+    },
+    {
+      "epoch": 1.104,
+      "grad_norm": 0.004375297110527754,
+      "learning_rate": 3.896e-05,
+      "loss": 0.0089,
+      "step": 5520
+    },
+    {
+      "epoch": 1.106,
+      "grad_norm": 0.010319947265088558,
+      "learning_rate": 3.894e-05,
+      "loss": 0.0333,
+      "step": 5530
+    },
+    {
+      "epoch": 1.108,
+      "grad_norm": 0.0030403188429772854,
+      "learning_rate": 3.892e-05,
+      "loss": 0.0247,
+      "step": 5540
+    },
+    {
+      "epoch": 1.11,
+      "grad_norm": 0.024173550307750702,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 0.0335,
+      "step": 5550
+    },
+    {
+      "epoch": 1.112,
+      "grad_norm": 0.004791913088411093,
+      "learning_rate": 3.888e-05,
+      "loss": 0.0004,
+      "step": 5560
+    },
+    {
+      "epoch": 1.114,
+      "grad_norm": 0.0013288178015500307,
+      "learning_rate": 3.8860000000000004e-05,
+      "loss": 0.0142,
+      "step": 5570
+    },
+    {
+      "epoch": 1.116,
+      "grad_norm": 0.0035326327197253704,
+      "learning_rate": 3.884e-05,
+      "loss": 0.0226,
+      "step": 5580
+    },
+    {
+      "epoch": 1.1179999999999999,
+      "grad_norm": 0.004453238099813461,
+      "learning_rate": 3.882e-05,
+      "loss": 0.0104,
+      "step": 5590
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.02598814107477665,
+      "learning_rate": 3.88e-05,
+      "loss": 0.0007,
+      "step": 5600
+    },
+    {
+      "epoch": 1.1219999999999999,
+      "grad_norm": 0.005187725182622671,
+      "learning_rate": 3.878e-05,
+      "loss": 0.0316,
+      "step": 5610
+    },
+    {
+      "epoch": 1.124,
+      "grad_norm": 0.002645947737619281,
+      "learning_rate": 3.876e-05,
+      "loss": 0.0394,
+      "step": 5620
+    },
+    {
+      "epoch": 1.126,
+      "grad_norm": 0.039730992168188095,
+      "learning_rate": 3.8740000000000005e-05,
+      "loss": 0.0451,
+      "step": 5630
+    },
+    {
+      "epoch": 1.1280000000000001,
+      "grad_norm": 0.04360891878604889,
+      "learning_rate": 3.872e-05,
+      "loss": 0.0011,
+      "step": 5640
+    },
+    {
+      "epoch": 1.13,
+      "grad_norm": 4.470123291015625,
+      "learning_rate": 3.8700000000000006e-05,
+      "loss": 0.0098,
+      "step": 5650
+    },
+    {
+      "epoch": 1.1320000000000001,
+      "grad_norm": 0.056175097823143005,
+      "learning_rate": 3.868e-05,
+      "loss": 0.0343,
+      "step": 5660
+    },
+    {
+      "epoch": 1.134,
+      "grad_norm": 6.098271369934082,
+      "learning_rate": 3.866e-05,
+      "loss": 0.061,
+      "step": 5670
+    },
+    {
+      "epoch": 1.1360000000000001,
+      "grad_norm": 0.5581873655319214,
+      "learning_rate": 3.864e-05,
+      "loss": 0.0014,
+      "step": 5680
+    },
+    {
+      "epoch": 1.138,
+      "grad_norm": 2.2527987957000732,
+      "learning_rate": 3.862e-05,
+      "loss": 0.0088,
+      "step": 5690
+    },
+    {
+      "epoch": 1.1400000000000001,
+      "grad_norm": 0.008088558912277222,
+      "learning_rate": 3.86e-05,
+      "loss": 0.0767,
+      "step": 5700
+    },
+    {
+      "epoch": 1.142,
+      "grad_norm": 0.036031920462846756,
+      "learning_rate": 3.858e-05,
+      "loss": 0.0007,
+      "step": 5710
+    },
+    {
+      "epoch": 1.144,
+      "grad_norm": 0.030742233619093895,
+      "learning_rate": 3.8560000000000004e-05,
+      "loss": 0.0008,
+      "step": 5720
+    },
+    {
+      "epoch": 1.146,
+      "grad_norm": 0.0673534944653511,
+      "learning_rate": 3.854000000000001e-05,
+      "loss": 0.0637,
+      "step": 5730
+    },
+    {
+      "epoch": 1.148,
+      "grad_norm": 0.016540292650461197,
+      "learning_rate": 3.8520000000000004e-05,
+      "loss": 0.041,
+      "step": 5740
+    },
+    {
+      "epoch": 1.15,
+      "grad_norm": 0.05118989571928978,
+      "learning_rate": 3.85e-05,
+      "loss": 0.0011,
+      "step": 5750
+    },
+    {
+      "epoch": 1.152,
+      "grad_norm": 0.0019592768512666225,
+      "learning_rate": 3.848e-05,
+      "loss": 0.0168,
+      "step": 5760
+    },
+    {
+      "epoch": 1.154,
+      "grad_norm": 0.2414097934961319,
+      "learning_rate": 3.846e-05,
+      "loss": 0.0008,
+      "step": 5770
+    },
+    {
+      "epoch": 1.156,
+      "grad_norm": 1.9626208543777466,
+      "learning_rate": 3.8440000000000005e-05,
+      "loss": 0.0078,
+      "step": 5780
+    },
+    {
+      "epoch": 1.158,
+      "grad_norm": 0.006050242111086845,
+      "learning_rate": 3.842e-05,
+      "loss": 0.034,
+      "step": 5790
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 0.0023852037265896797,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 0.0318,
+      "step": 5800
+    },
+    {
+      "epoch": 1.162,
+      "grad_norm": 0.014977013692259789,
+      "learning_rate": 3.838e-05,
+      "loss": 0.0339,
+      "step": 5810
+    },
+    {
+      "epoch": 1.164,
+      "grad_norm": 0.00630151154473424,
+      "learning_rate": 3.836e-05,
+      "loss": 0.017,
+      "step": 5820
+    },
+    {
+      "epoch": 1.166,
+      "grad_norm": 2.7326033115386963,
+      "learning_rate": 3.834e-05,
+      "loss": 0.0497,
+      "step": 5830
+    },
+    {
+      "epoch": 1.168,
+      "grad_norm": 0.01748991757631302,
+      "learning_rate": 3.832e-05,
+      "loss": 0.0259,
+      "step": 5840
+    },
+    {
+      "epoch": 1.17,
+      "grad_norm": 0.04222341999411583,
+      "learning_rate": 3.83e-05,
+      "loss": 0.0725,
+      "step": 5850
+    },
+    {
+      "epoch": 1.172,
+      "grad_norm": 0.013558843173086643,
+      "learning_rate": 3.828e-05,
+      "loss": 0.0557,
+      "step": 5860
+    },
+    {
+      "epoch": 1.174,
+      "grad_norm": 0.0790916159749031,
+      "learning_rate": 3.826e-05,
+      "loss": 0.0354,
+      "step": 5870
+    },
+    {
+      "epoch": 1.176,
+      "grad_norm": 0.038206301629543304,
+      "learning_rate": 3.8240000000000007e-05,
+      "loss": 0.0337,
+      "step": 5880
+    },
+    {
+      "epoch": 1.178,
+      "grad_norm": 0.033259790390729904,
+      "learning_rate": 3.822e-05,
+      "loss": 0.0588,
+      "step": 5890
+    },
+    {
+      "epoch": 1.18,
+      "grad_norm": 0.9643262028694153,
+      "learning_rate": 3.82e-05,
+      "loss": 0.014,
+      "step": 5900
+    },
+    {
+      "epoch": 1.182,
+      "grad_norm": 0.007181926164776087,
+      "learning_rate": 3.818e-05,
+      "loss": 0.0247,
+      "step": 5910
+    },
+    {
+      "epoch": 1.184,
+      "grad_norm": 0.09112616628408432,
+      "learning_rate": 3.816e-05,
+      "loss": 0.0211,
+      "step": 5920
+    },
+    {
+      "epoch": 1.186,
+      "grad_norm": 0.006056137848645449,
+      "learning_rate": 3.8140000000000004e-05,
+      "loss": 0.0186,
+      "step": 5930
+    },
+    {
+      "epoch": 1.188,
+      "grad_norm": 0.029104258865118027,
+      "learning_rate": 3.812e-05,
+      "loss": 0.0327,
+      "step": 5940
+    },
+    {
+      "epoch": 1.19,
+      "grad_norm": 0.02636542171239853,
+      "learning_rate": 3.8100000000000005e-05,
+      "loss": 0.0018,
+      "step": 5950
+    },
+    {
+      "epoch": 1.192,
+      "grad_norm": 0.058980781584978104,
+      "learning_rate": 3.808e-05,
+      "loss": 0.0152,
+      "step": 5960
+    },
+    {
+      "epoch": 1.194,
+      "grad_norm": 0.09248950332403183,
+      "learning_rate": 3.806e-05,
+      "loss": 0.0325,
+      "step": 5970
+    },
+    {
+      "epoch": 1.196,
+      "grad_norm": 0.0237438902258873,
+      "learning_rate": 3.804e-05,
+      "loss": 0.0018,
+      "step": 5980
+    },
+    {
+      "epoch": 1.198,
+      "grad_norm": 0.006177122704684734,
+      "learning_rate": 3.802e-05,
+      "loss": 0.0365,
+      "step": 5990
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 0.6815275549888611,
+      "learning_rate": 3.8e-05,
+      "loss": 0.026,
+      "step": 6000
+    },
+    {
+      "epoch": 1.202,
+      "grad_norm": 0.010962256230413914,
+      "learning_rate": 3.7980000000000006e-05,
+      "loss": 0.0847,
+      "step": 6010
+    },
+    {
+      "epoch": 1.204,
+      "grad_norm": 0.024839259684085846,
+      "learning_rate": 3.796e-05,
+      "loss": 0.0283,
+      "step": 6020
+    },
+    {
+      "epoch": 1.206,
+      "grad_norm": 0.003172196913510561,
+      "learning_rate": 3.7940000000000006e-05,
+      "loss": 0.0146,
+      "step": 6030
+    },
+    {
+      "epoch": 1.208,
+      "grad_norm": 0.015222206711769104,
+      "learning_rate": 3.792e-05,
+      "loss": 0.0188,
+      "step": 6040
+    },
+    {
+      "epoch": 1.21,
+      "grad_norm": 10.11510181427002,
+      "learning_rate": 3.79e-05,
+      "loss": 0.0616,
+      "step": 6050
+    },
+    {
+      "epoch": 1.212,
+      "grad_norm": 0.20069774985313416,
+      "learning_rate": 3.788e-05,
+      "loss": 0.0099,
+      "step": 6060
+    },
+    {
+      "epoch": 1.214,
+      "grad_norm": 0.021253420040011406,
+      "learning_rate": 3.786e-05,
+      "loss": 0.0869,
+      "step": 6070
+    },
+    {
+      "epoch": 1.216,
+      "grad_norm": 0.016821395605802536,
+      "learning_rate": 3.7840000000000004e-05,
+      "loss": 0.0076,
+      "step": 6080
+    },
+    {
+      "epoch": 1.218,
+      "grad_norm": 0.032200634479522705,
+      "learning_rate": 3.782e-05,
+      "loss": 0.0549,
+      "step": 6090
+    },
+    {
+      "epoch": 1.22,
+      "grad_norm": 0.015066594816744328,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 0.0348,
+      "step": 6100
+    },
+    {
+      "epoch": 1.222,
+      "grad_norm": 0.027172857895493507,
+      "learning_rate": 3.778000000000001e-05,
+      "loss": 0.0022,
+      "step": 6110
+    },
+    {
+      "epoch": 1.224,
+      "grad_norm": 5.351569175720215,
+      "learning_rate": 3.776e-05,
+      "loss": 0.007,
+      "step": 6120
+    },
+    {
+      "epoch": 1.226,
+      "grad_norm": 0.017085228115320206,
+      "learning_rate": 3.774e-05,
+      "loss": 0.0018,
+      "step": 6130
+    },
+    {
+      "epoch": 1.228,
+      "grad_norm": 2.938812017440796,
+      "learning_rate": 3.772e-05,
+      "loss": 0.0356,
+      "step": 6140
+    },
+    {
+      "epoch": 1.23,
+      "grad_norm": 0.009132993407547474,
+      "learning_rate": 3.77e-05,
+      "loss": 0.0085,
+      "step": 6150
+    },
+    {
+      "epoch": 1.232,
+      "grad_norm": 0.007850081659853458,
+      "learning_rate": 3.7680000000000005e-05,
+      "loss": 0.0086,
+      "step": 6160
+    },
+    {
+      "epoch": 1.234,
+      "grad_norm": 0.007628438528627157,
+      "learning_rate": 3.766e-05,
+      "loss": 0.004,
+      "step": 6170
+    },
+    {
+      "epoch": 1.236,
+      "grad_norm": 0.05944610759615898,
+      "learning_rate": 3.7640000000000006e-05,
+      "loss": 0.0347,
+      "step": 6180
+    },
+    {
+      "epoch": 1.238,
+      "grad_norm": 0.06916775554418564,
+      "learning_rate": 3.762e-05,
+      "loss": 0.0031,
+      "step": 6190
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": 0.0030985125340521336,
+      "learning_rate": 3.76e-05,
+      "loss": 0.057,
+      "step": 6200
+    },
+    {
+      "epoch": 1.242,
+      "grad_norm": 5.392228126525879,
+      "learning_rate": 3.758e-05,
+      "loss": 0.0098,
+      "step": 6210
+    },
+    {
+      "epoch": 1.244,
+      "grad_norm": 0.014599272049963474,
+      "learning_rate": 3.756e-05,
+      "loss": 0.096,
+      "step": 6220
+    },
+    {
+      "epoch": 1.246,
+      "grad_norm": 0.005305968225002289,
+      "learning_rate": 3.754e-05,
+      "loss": 0.0026,
+      "step": 6230
+    },
+    {
+      "epoch": 1.248,
+      "grad_norm": 0.016365058720111847,
+      "learning_rate": 3.752e-05,
+      "loss": 0.0005,
+      "step": 6240
+    },
+    {
+      "epoch": 1.25,
+      "grad_norm": 0.02052203379571438,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 0.0026,
+      "step": 6250
+    },
+    {
+      "epoch": 1.252,
+      "grad_norm": 0.027127837762236595,
+      "learning_rate": 3.748000000000001e-05,
+      "loss": 0.0031,
+      "step": 6260
+    },
+    {
+      "epoch": 1.254,
+      "grad_norm": 0.011660023592412472,
+      "learning_rate": 3.7460000000000004e-05,
+      "loss": 0.0007,
+      "step": 6270
+    },
+    {
+      "epoch": 1.256,
+      "grad_norm": 0.0043875472620129585,
+      "learning_rate": 3.744e-05,
+      "loss": 0.0381,
+      "step": 6280
+    },
+    {
+      "epoch": 1.258,
+      "grad_norm": 0.004166820086538792,
+      "learning_rate": 3.742e-05,
+      "loss": 0.0005,
+      "step": 6290
+    },
+    {
+      "epoch": 1.26,
+      "grad_norm": 0.010409657843410969,
+      "learning_rate": 3.74e-05,
+      "loss": 0.0004,
+      "step": 6300
+    },
+    {
+      "epoch": 1.262,
+      "grad_norm": 0.005970802158117294,
+      "learning_rate": 3.7380000000000005e-05,
+      "loss": 0.0003,
+      "step": 6310
+    },
+    {
+      "epoch": 1.264,
+      "grad_norm": 0.001832369016483426,
+      "learning_rate": 3.736e-05,
+      "loss": 0.0008,
+      "step": 6320
+    },
+    {
+      "epoch": 1.266,
+      "grad_norm": 4.269883632659912,
+      "learning_rate": 3.7340000000000005e-05,
+      "loss": 0.0912,
+      "step": 6330
+    },
+    {
+      "epoch": 1.268,
+      "grad_norm": 0.018679887056350708,
+      "learning_rate": 3.732e-05,
+      "loss": 0.0231,
+      "step": 6340
+    },
+    {
+      "epoch": 1.27,
+      "grad_norm": 0.0038905141409486532,
+      "learning_rate": 3.73e-05,
+      "loss": 0.0963,
+      "step": 6350
+    },
+    {
+      "epoch": 1.272,
+      "grad_norm": 0.0063622696325182915,
+      "learning_rate": 3.728e-05,
+      "loss": 0.0363,
+      "step": 6360
+    },
+    {
+      "epoch": 1.274,
+      "grad_norm": 7.225616931915283,
+      "learning_rate": 3.726e-05,
+      "loss": 0.0092,
+      "step": 6370
+    },
+    {
+      "epoch": 1.276,
+      "grad_norm": 0.19642934203147888,
+      "learning_rate": 3.724e-05,
+      "loss": 0.0198,
+      "step": 6380
+    },
+    {
+      "epoch": 1.278,
+      "grad_norm": 0.4732407033443451,
+      "learning_rate": 3.722e-05,
+      "loss": 0.0282,
+      "step": 6390
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 0.008111849427223206,
+      "learning_rate": 3.72e-05,
+      "loss": 0.0011,
+      "step": 6400
+    },
+    {
+      "epoch": 1.282,
+      "grad_norm": 0.01679563894867897,
+      "learning_rate": 3.7180000000000007e-05,
+      "loss": 0.0026,
+      "step": 6410
+    },
+    {
+      "epoch": 1.284,
+      "grad_norm": 0.007721078582108021,
+      "learning_rate": 3.716e-05,
+      "loss": 0.0026,
+      "step": 6420
+    },
+    {
+      "epoch": 1.286,
+      "grad_norm": 0.0026185016613453627,
+      "learning_rate": 3.714e-05,
+      "loss": 0.036,
+      "step": 6430
+    },
+    {
+      "epoch": 1.288,
+      "grad_norm": 0.020146319642663002,
+      "learning_rate": 3.712e-05,
+      "loss": 0.0123,
+      "step": 6440
+    },
+    {
+      "epoch": 1.29,
+      "grad_norm": 0.13414332270622253,
+      "learning_rate": 3.71e-05,
+      "loss": 0.0607,
+      "step": 6450
+    },
+    {
+      "epoch": 1.292,
+      "grad_norm": 0.006200847215950489,
+      "learning_rate": 3.7080000000000004e-05,
+      "loss": 0.022,
+      "step": 6460
+    },
+    {
+      "epoch": 1.294,
+      "grad_norm": 5.099764823913574,
+      "learning_rate": 3.706e-05,
+      "loss": 0.0139,
+      "step": 6470
+    },
+    {
+      "epoch": 1.296,
+      "grad_norm": 0.02171763963997364,
+      "learning_rate": 3.7040000000000005e-05,
+      "loss": 0.0077,
+      "step": 6480
+    },
+    {
+      "epoch": 1.298,
+      "grad_norm": 0.008273259736597538,
+      "learning_rate": 3.702e-05,
+      "loss": 0.001,
+      "step": 6490
+    },
+    {
+      "epoch": 1.3,
+      "grad_norm": 0.004964989610016346,
+      "learning_rate": 3.7e-05,
+      "loss": 0.051,
+      "step": 6500
+    },
+    {
+      "epoch": 1.302,
+      "grad_norm": 0.008970340713858604,
+      "learning_rate": 3.698e-05,
+      "loss": 0.0004,
+      "step": 6510
+    },
+    {
+      "epoch": 1.304,
+      "grad_norm": 0.04173256456851959,
+      "learning_rate": 3.696e-05,
+      "loss": 0.0006,
+      "step": 6520
+    },
+    {
+      "epoch": 1.306,
+      "grad_norm": 0.005524017848074436,
+      "learning_rate": 3.694e-05,
+      "loss": 0.0003,
+      "step": 6530
+    },
+    {
+      "epoch": 1.308,
+      "grad_norm": 5.520667552947998,
+      "learning_rate": 3.692e-05,
+      "loss": 0.0523,
+      "step": 6540
+    },
+    {
+      "epoch": 1.31,
+      "grad_norm": 0.0042858850210905075,
+      "learning_rate": 3.69e-05,
+      "loss": 0.0007,
+      "step": 6550
+    },
+    {
+      "epoch": 1.312,
+      "grad_norm": 7.040893077850342,
+      "learning_rate": 3.6880000000000006e-05,
+      "loss": 0.1054,
+      "step": 6560
+    },
+    {
+      "epoch": 1.314,
+      "grad_norm": 0.0050101811066269875,
+      "learning_rate": 3.686e-05,
+      "loss": 0.0344,
+      "step": 6570
+    },
+    {
+      "epoch": 1.316,
+      "grad_norm": 0.39104142785072327,
+      "learning_rate": 3.684e-05,
+      "loss": 0.0063,
+      "step": 6580
+    },
+    {
+      "epoch": 1.318,
+      "grad_norm": 0.005544619634747505,
+      "learning_rate": 3.682e-05,
+      "loss": 0.0082,
+      "step": 6590
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 5.666801452636719,
+      "learning_rate": 3.68e-05,
+      "loss": 0.058,
+      "step": 6600
+    },
+    {
+      "epoch": 1.322,
+      "grad_norm": 0.04448606073856354,
+      "learning_rate": 3.6780000000000004e-05,
+      "loss": 0.0059,
+      "step": 6610
+    },
+    {
+      "epoch": 1.324,
+      "grad_norm": 0.007718035951256752,
+      "learning_rate": 3.676e-05,
+      "loss": 0.0012,
+      "step": 6620
+    },
+    {
+      "epoch": 1.326,
+      "grad_norm": 0.0031523483339697123,
+      "learning_rate": 3.6740000000000004e-05,
+      "loss": 0.0535,
+      "step": 6630
+    },
+    {
+      "epoch": 1.328,
+      "grad_norm": 8.072606086730957,
+      "learning_rate": 3.672000000000001e-05,
+      "loss": 0.0175,
+      "step": 6640
+    },
+    {
+      "epoch": 1.33,
+      "grad_norm": 0.01579626277089119,
+      "learning_rate": 3.6700000000000004e-05,
+      "loss": 0.0361,
+      "step": 6650
+    },
+    {
+      "epoch": 1.332,
+      "grad_norm": 0.006640156265348196,
+      "learning_rate": 3.668e-05,
+      "loss": 0.0742,
+      "step": 6660
+    },
+    {
+      "epoch": 1.334,
+      "grad_norm": 0.027232594788074493,
+      "learning_rate": 3.666e-05,
+      "loss": 0.0408,
+      "step": 6670
+    },
+    {
+      "epoch": 1.336,
+      "grad_norm": 0.01830277405679226,
+      "learning_rate": 3.664e-05,
+      "loss": 0.0106,
+      "step": 6680
+    },
+    {
+      "epoch": 1.338,
+      "grad_norm": 0.0011746941599994898,
+      "learning_rate": 3.6620000000000005e-05,
+      "loss": 0.0152,
+      "step": 6690
+    },
+    {
+      "epoch": 1.34,
+      "grad_norm": 0.10199079662561417,
+      "learning_rate": 3.66e-05,
+      "loss": 0.0375,
+      "step": 6700
+    },
+    {
+      "epoch": 1.342,
+      "grad_norm": 0.0049275667406618595,
+      "learning_rate": 3.6580000000000006e-05,
+      "loss": 0.0301,
+      "step": 6710
+    },
+    {
+      "epoch": 1.3439999999999999,
+      "grad_norm": 0.1531529277563095,
+      "learning_rate": 3.656e-05,
+      "loss": 0.0135,
+      "step": 6720
+    },
+    {
+      "epoch": 1.346,
+      "grad_norm": 3.4101831912994385,
+      "learning_rate": 3.654e-05,
+      "loss": 0.0459,
+      "step": 6730
+    },
+    {
+      "epoch": 1.3479999999999999,
+      "grad_norm": 0.02141091600060463,
+      "learning_rate": 3.652e-05,
+      "loss": 0.0061,
+      "step": 6740
+    },
+    {
+      "epoch": 1.35,
+      "grad_norm": 0.01527430210262537,
+      "learning_rate": 3.65e-05,
+      "loss": 0.09,
+      "step": 6750
+    },
+    {
+      "epoch": 1.3519999999999999,
+      "grad_norm": 0.007771384436637163,
+      "learning_rate": 3.648e-05,
+      "loss": 0.0069,
+      "step": 6760
+    },
+    {
+      "epoch": 1.354,
+      "grad_norm": 0.012513011693954468,
+      "learning_rate": 3.646e-05,
+      "loss": 0.0025,
+      "step": 6770
+    },
+    {
+      "epoch": 1.3559999999999999,
+      "grad_norm": 0.009514408186078072,
+      "learning_rate": 3.6440000000000003e-05,
+      "loss": 0.0008,
+      "step": 6780
+    },
+    {
+      "epoch": 1.358,
+      "grad_norm": 0.01860465668141842,
+      "learning_rate": 3.642000000000001e-05,
+      "loss": 0.0726,
+      "step": 6790
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 0.2263597846031189,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 0.0188,
+      "step": 6800
+    },
+    {
+      "epoch": 1.362,
+      "grad_norm": 5.2578654289245605,
+      "learning_rate": 3.638e-05,
+      "loss": 0.0486,
+      "step": 6810
+    },
+    {
+      "epoch": 1.3639999999999999,
+      "grad_norm": 0.008457668125629425,
+      "learning_rate": 3.636e-05,
+      "loss": 0.0071,
+      "step": 6820
+    },
+    {
+      "epoch": 1.366,
+      "grad_norm": 0.012101087719202042,
+      "learning_rate": 3.634e-05,
+      "loss": 0.0286,
+      "step": 6830
+    },
+    {
+      "epoch": 1.3679999999999999,
+      "grad_norm": 0.04453659430146217,
+      "learning_rate": 3.6320000000000005e-05,
+      "loss": 0.0014,
+      "step": 6840
+    },
+    {
+      "epoch": 1.37,
+      "grad_norm": 0.4392252564430237,
+      "learning_rate": 3.63e-05,
+      "loss": 0.0015,
+      "step": 6850
+    },
+    {
+      "epoch": 1.3719999999999999,
+      "grad_norm": 0.0513470284640789,
+      "learning_rate": 3.6280000000000005e-05,
+      "loss": 0.0477,
+      "step": 6860
+    },
+    {
+      "epoch": 1.374,
+      "grad_norm": 0.006927871610969305,
+      "learning_rate": 3.626e-05,
+      "loss": 0.0525,
+      "step": 6870
+    },
+    {
+      "epoch": 1.376,
+      "grad_norm": 0.04651854932308197,
+      "learning_rate": 3.624e-05,
+      "loss": 0.0282,
+      "step": 6880
+    },
+    {
+      "epoch": 1.3780000000000001,
+      "grad_norm": 7.312324523925781,
+      "learning_rate": 3.622e-05,
+      "loss": 0.0129,
+      "step": 6890
+    },
+    {
+      "epoch": 1.38,
+      "grad_norm": 0.0063580735586583614,
+      "learning_rate": 3.62e-05,
+      "loss": 0.0104,
+      "step": 6900
+    },
+    {
+      "epoch": 1.3820000000000001,
+      "grad_norm": 0.006133576389402151,
+      "learning_rate": 3.618e-05,
+      "loss": 0.0012,
+      "step": 6910
+    },
+    {
+      "epoch": 1.384,
+      "grad_norm": 0.005199335515499115,
+      "learning_rate": 3.616e-05,
+      "loss": 0.0239,
+      "step": 6920
+    },
+    {
+      "epoch": 1.3860000000000001,
+      "grad_norm": 0.10993275791406631,
+      "learning_rate": 3.614e-05,
+      "loss": 0.0036,
+      "step": 6930
+    },
+    {
+      "epoch": 1.388,
+      "grad_norm": 0.006421110592782497,
+      "learning_rate": 3.6120000000000007e-05,
+      "loss": 0.0265,
+      "step": 6940
+    },
+    {
+      "epoch": 1.3900000000000001,
+      "grad_norm": 0.008753607049584389,
+      "learning_rate": 3.61e-05,
+      "loss": 0.0847,
+      "step": 6950
+    },
+    {
+      "epoch": 1.392,
+      "grad_norm": 0.004009403754025698,
+      "learning_rate": 3.608e-05,
+      "loss": 0.001,
+      "step": 6960
+    },
+    {
+      "epoch": 1.3940000000000001,
+      "grad_norm": 0.27608707547187805,
+      "learning_rate": 3.606e-05,
+      "loss": 0.0025,
+      "step": 6970
+    },
+    {
+      "epoch": 1.396,
+      "grad_norm": 0.002582039451226592,
+      "learning_rate": 3.604e-05,
+      "loss": 0.0256,
+      "step": 6980
+    },
+    {
+      "epoch": 1.3980000000000001,
+      "grad_norm": 0.004374573472887278,
+      "learning_rate": 3.6020000000000004e-05,
+      "loss": 0.0006,
+      "step": 6990
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 0.06569258868694305,
+      "learning_rate": 3.6e-05,
+      "loss": 0.0446,
+      "step": 7000
+    },
+    {
+      "epoch": 1.4020000000000001,
+      "grad_norm": 0.007347286678850651,
+      "learning_rate": 3.5980000000000004e-05,
+      "loss": 0.0015,
+      "step": 7010
+    },
+    {
+      "epoch": 1.404,
+      "grad_norm": 0.00714182248339057,
+      "learning_rate": 3.596e-05,
+      "loss": 0.0138,
+      "step": 7020
+    },
+    {
+      "epoch": 1.4060000000000001,
+      "grad_norm": 0.005495099350810051,
+      "learning_rate": 3.594e-05,
+      "loss": 0.0006,
+      "step": 7030
+    },
+    {
+      "epoch": 1.408,
+      "grad_norm": 0.2169579118490219,
+      "learning_rate": 3.592e-05,
+      "loss": 0.0022,
+      "step": 7040
+    },
+    {
+      "epoch": 1.41,
+      "grad_norm": 0.006320688873529434,
+      "learning_rate": 3.59e-05,
+      "loss": 0.0009,
+      "step": 7050
+    },
+    {
+      "epoch": 1.412,
+      "grad_norm": 0.004149855114519596,
+      "learning_rate": 3.588e-05,
+      "loss": 0.0007,
+      "step": 7060
+    },
+    {
+      "epoch": 1.414,
+      "grad_norm": 0.163759246468544,
+      "learning_rate": 3.586e-05,
+      "loss": 0.0396,
+      "step": 7070
+    },
+    {
+      "epoch": 1.416,
+      "grad_norm": 3.08158802986145,
+      "learning_rate": 3.584e-05,
+      "loss": 0.0337,
+      "step": 7080
+    },
+    {
+      "epoch": 1.418,
+      "grad_norm": 0.06443344056606293,
+      "learning_rate": 3.5820000000000006e-05,
+      "loss": 0.0085,
+      "step": 7090
+    },
+    {
+      "epoch": 1.42,
+      "grad_norm": 0.13582125306129456,
+      "learning_rate": 3.58e-05,
+      "loss": 0.0372,
+      "step": 7100
+    },
+    {
+      "epoch": 1.422,
+      "grad_norm": 0.04201153665781021,
+      "learning_rate": 3.578e-05,
+      "loss": 0.0481,
+      "step": 7110
+    },
+    {
+      "epoch": 1.424,
+      "grad_norm": 0.7387461066246033,
+      "learning_rate": 3.5759999999999996e-05,
+      "loss": 0.094,
+      "step": 7120
+    },
+    {
+      "epoch": 1.426,
+      "grad_norm": 0.0052914912812411785,
+      "learning_rate": 3.574e-05,
+      "loss": 0.0677,
+      "step": 7130
+    },
+    {
+      "epoch": 1.428,
+      "grad_norm": 0.03011983446776867,
+      "learning_rate": 3.5720000000000004e-05,
+      "loss": 0.0548,
+      "step": 7140
+    },
+    {
+      "epoch": 1.43,
+      "grad_norm": 0.024330249056220055,
+      "learning_rate": 3.57e-05,
+      "loss": 0.0254,
+      "step": 7150
+    },
+    {
+      "epoch": 1.432,
+      "grad_norm": 0.02639131061732769,
+      "learning_rate": 3.5680000000000004e-05,
+      "loss": 0.0037,
+      "step": 7160
+    },
+    {
+      "epoch": 1.434,
+      "grad_norm": 0.008115893229842186,
+      "learning_rate": 3.566e-05,
+      "loss": 0.0022,
+      "step": 7170
+    },
+    {
+      "epoch": 1.436,
+      "grad_norm": 0.031278759241104126,
+      "learning_rate": 3.5640000000000004e-05,
+      "loss": 0.0404,
+      "step": 7180
+    },
+    {
+      "epoch": 1.438,
+      "grad_norm": 0.01463502086699009,
+      "learning_rate": 3.562e-05,
+      "loss": 0.009,
+      "step": 7190
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 0.009483623318374157,
+      "learning_rate": 3.56e-05,
+      "loss": 0.0004,
+      "step": 7200
+    },
+    {
+      "epoch": 1.442,
+      "grad_norm": 0.003925948403775692,
+      "learning_rate": 3.558e-05,
+      "loss": 0.0235,
+      "step": 7210
+    },
+    {
+      "epoch": 1.444,
+      "grad_norm": 0.021284854039549828,
+      "learning_rate": 3.5560000000000005e-05,
+      "loss": 0.0628,
+      "step": 7220
+    },
+    {
+      "epoch": 1.446,
+      "grad_norm": 0.00911774393171072,
+      "learning_rate": 3.554e-05,
+      "loss": 0.064,
+      "step": 7230
+    },
+    {
+      "epoch": 1.448,
+      "grad_norm": 7.336594581604004,
+      "learning_rate": 3.5520000000000006e-05,
+      "loss": 0.0262,
+      "step": 7240
+    },
+    {
+      "epoch": 1.45,
+      "grad_norm": 0.018722884356975555,
+      "learning_rate": 3.55e-05,
+      "loss": 0.0041,
+      "step": 7250
+    },
+    {
+      "epoch": 1.452,
+      "grad_norm": 7.197745323181152,
+      "learning_rate": 3.548e-05,
+      "loss": 0.0204,
+      "step": 7260
+    },
+    {
+      "epoch": 1.454,
+      "grad_norm": 0.029174281284213066,
+      "learning_rate": 3.546e-05,
+      "loss": 0.0115,
+      "step": 7270
+    },
+    {
+      "epoch": 1.456,
+      "grad_norm": 0.022253194823861122,
+      "learning_rate": 3.544e-05,
+      "loss": 0.005,
+      "step": 7280
+    },
+    {
+      "epoch": 1.458,
+      "grad_norm": 0.004992309492081404,
+      "learning_rate": 3.542e-05,
+      "loss": 0.0009,
+      "step": 7290
+    },
+    {
+      "epoch": 1.46,
+      "grad_norm": 0.013181679882109165,
+      "learning_rate": 3.54e-05,
+      "loss": 0.0316,
+      "step": 7300
+    },
+    {
+      "epoch": 1.462,
+      "grad_norm": 3.856898069381714,
+      "learning_rate": 3.5380000000000003e-05,
+      "loss": 0.0599,
+      "step": 7310
+    },
+    {
+      "epoch": 1.464,
+      "grad_norm": 0.015816127881407738,
+      "learning_rate": 3.536000000000001e-05,
+      "loss": 0.0051,
+      "step": 7320
+    },
+    {
+      "epoch": 1.466,
+      "grad_norm": 0.012996883131563663,
+      "learning_rate": 3.5340000000000004e-05,
+      "loss": 0.0018,
+      "step": 7330
+    },
+    {
+      "epoch": 1.468,
+      "grad_norm": 0.08451968431472778,
+      "learning_rate": 3.532e-05,
+      "loss": 0.0504,
+      "step": 7340
+    },
+    {
+      "epoch": 1.47,
+      "grad_norm": 0.01942392811179161,
+      "learning_rate": 3.53e-05,
+      "loss": 0.0338,
+      "step": 7350
+    },
+    {
+      "epoch": 1.472,
+      "grad_norm": 0.007807042449712753,
+      "learning_rate": 3.528e-05,
+      "loss": 0.0014,
+      "step": 7360
+    },
+    {
+      "epoch": 1.474,
+      "grad_norm": 0.35328736901283264,
+      "learning_rate": 3.5260000000000005e-05,
+      "loss": 0.0331,
+      "step": 7370
+    },
+    {
+      "epoch": 1.476,
+      "grad_norm": 0.007190255913883448,
+      "learning_rate": 3.524e-05,
+      "loss": 0.0023,
+      "step": 7380
+    },
+    {
+      "epoch": 1.478,
+      "grad_norm": 0.005118261557072401,
+      "learning_rate": 3.5220000000000005e-05,
+      "loss": 0.0117,
+      "step": 7390
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 0.01209124457091093,
+      "learning_rate": 3.52e-05,
+      "loss": 0.0293,
+      "step": 7400
+    },
+    {
+      "epoch": 1.482,
+      "grad_norm": 0.0048452820628881454,
+      "learning_rate": 3.518e-05,
+      "loss": 0.0268,
+      "step": 7410
+    },
+    {
+      "epoch": 1.484,
+      "grad_norm": 0.004197114147245884,
+      "learning_rate": 3.516e-05,
+      "loss": 0.0012,
+      "step": 7420
+    },
+    {
+      "epoch": 1.486,
+      "grad_norm": 0.02889455109834671,
+      "learning_rate": 3.514e-05,
+      "loss": 0.0059,
+      "step": 7430
+    },
+    {
+      "epoch": 1.488,
+      "grad_norm": 0.0030213482677936554,
+      "learning_rate": 3.512e-05,
+      "loss": 0.0032,
+      "step": 7440
+    },
+    {
+      "epoch": 1.49,
+      "grad_norm": 0.008681685663759708,
+      "learning_rate": 3.51e-05,
+      "loss": 0.01,
+      "step": 7450
+    },
+    {
+      "epoch": 1.492,
+      "grad_norm": 0.0030685935635119677,
+      "learning_rate": 3.508e-05,
+      "loss": 0.0174,
+      "step": 7460
+    },
+    {
+      "epoch": 1.494,
+      "grad_norm": 0.0032798685133457184,
+      "learning_rate": 3.5060000000000007e-05,
+      "loss": 0.0443,
+      "step": 7470
+    },
+    {
+      "epoch": 1.496,
+      "grad_norm": 0.009949653409421444,
+      "learning_rate": 3.504e-05,
+      "loss": 0.0352,
+      "step": 7480
+    },
+    {
+      "epoch": 1.498,
+      "grad_norm": 0.7159678339958191,
+      "learning_rate": 3.502e-05,
+      "loss": 0.0952,
+      "step": 7490
+    },
+    {
+      "epoch": 1.5,
+      "grad_norm": 0.00631822319701314,
+      "learning_rate": 3.5e-05,
+      "loss": 0.0775,
+      "step": 7500
+    },
+    {
+      "epoch": 1.502,
+      "grad_norm": 0.09870759397745132,
+      "learning_rate": 3.498e-05,
+      "loss": 0.002,
+      "step": 7510
+    },
+    {
+      "epoch": 1.504,
+      "grad_norm": 0.018792299553751945,
+      "learning_rate": 3.4960000000000004e-05,
+      "loss": 0.0036,
+      "step": 7520
+    },
+    {
+      "epoch": 1.506,
+      "grad_norm": 0.8837906718254089,
+      "learning_rate": 3.494e-05,
+      "loss": 0.0052,
+      "step": 7530
+    },
+    {
+      "epoch": 1.508,
+      "grad_norm": 0.009946630336344242,
+      "learning_rate": 3.4920000000000004e-05,
+      "loss": 0.0012,
+      "step": 7540
+    },
+    {
+      "epoch": 1.51,
+      "grad_norm": 1.3218886852264404,
+      "learning_rate": 3.49e-05,
+      "loss": 0.0327,
+      "step": 7550
+    },
+    {
+      "epoch": 1.512,
+      "grad_norm": 0.44792118668556213,
+      "learning_rate": 3.4880000000000005e-05,
+      "loss": 0.0033,
+      "step": 7560
+    },
+    {
+      "epoch": 1.514,
+      "grad_norm": 0.05720994248986244,
+      "learning_rate": 3.486e-05,
+      "loss": 0.0006,
+      "step": 7570
+    },
+    {
+      "epoch": 1.516,
+      "grad_norm": 0.04546366259455681,
+      "learning_rate": 3.484e-05,
+      "loss": 0.0004,
+      "step": 7580
+    },
+    {
+      "epoch": 1.518,
+      "grad_norm": 0.003919788636267185,
+      "learning_rate": 3.482e-05,
+      "loss": 0.0001,
+      "step": 7590
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 0.008556053973734379,
+      "learning_rate": 3.48e-05,
+      "loss": 0.0007,
+      "step": 7600
+    },
+    {
+      "epoch": 1.522,
+      "grad_norm": 0.012111378833651543,
+      "learning_rate": 3.478e-05,
+      "loss": 0.0052,
+      "step": 7610
+    },
+    {
+      "epoch": 1.524,
+      "grad_norm": 0.003584041493013501,
+      "learning_rate": 3.4760000000000006e-05,
+      "loss": 0.0003,
+      "step": 7620
+    },
+    {
+      "epoch": 1.526,
+      "grad_norm": 0.0034352331422269344,
+      "learning_rate": 3.474e-05,
+      "loss": 0.1027,
+      "step": 7630
+    },
+    {
+      "epoch": 1.528,
+      "grad_norm": 0.005325534846633673,
+      "learning_rate": 3.472e-05,
+      "loss": 0.0002,
+      "step": 7640
+    },
+    {
+      "epoch": 1.53,
+      "grad_norm": 2.399014949798584,
+      "learning_rate": 3.4699999999999996e-05,
+      "loss": 0.0229,
+      "step": 7650
+    },
+    {
+      "epoch": 1.532,
+      "grad_norm": 0.03801944851875305,
+      "learning_rate": 3.468e-05,
+      "loss": 0.0583,
+      "step": 7660
+    },
+    {
+      "epoch": 1.534,
+      "grad_norm": 4.686843395233154,
+      "learning_rate": 3.4660000000000004e-05,
+      "loss": 0.0621,
+      "step": 7670
+    },
+    {
+      "epoch": 1.536,
+      "grad_norm": 0.02090873382985592,
+      "learning_rate": 3.464e-05,
+      "loss": 0.0435,
+      "step": 7680
+    },
+    {
+      "epoch": 1.538,
+      "grad_norm": 0.003576928051188588,
+      "learning_rate": 3.4620000000000004e-05,
+      "loss": 0.0189,
+      "step": 7690
+    },
+    {
+      "epoch": 1.54,
+      "grad_norm": 0.00641475897282362,
+      "learning_rate": 3.46e-05,
+      "loss": 0.0065,
+      "step": 7700
+    },
+    {
+      "epoch": 1.542,
+      "grad_norm": 0.010015168227255344,
+      "learning_rate": 3.4580000000000004e-05,
+      "loss": 0.005,
+      "step": 7710
+    },
+    {
+      "epoch": 1.544,
+      "grad_norm": 0.002555366838350892,
+      "learning_rate": 3.456e-05,
+      "loss": 0.0736,
+      "step": 7720
+    },
+    {
+      "epoch": 1.546,
+      "grad_norm": 0.007367596495896578,
+      "learning_rate": 3.454e-05,
+      "loss": 0.0006,
+      "step": 7730
+    },
+    {
+      "epoch": 1.548,
+      "grad_norm": 4.383460521697998,
+      "learning_rate": 3.452e-05,
+      "loss": 0.0353,
+      "step": 7740
+    },
+    {
+      "epoch": 1.55,
+      "grad_norm": 8.912064552307129,
+      "learning_rate": 3.45e-05,
+      "loss": 0.0293,
+      "step": 7750
+    },
+    {
+      "epoch": 1.552,
+      "grad_norm": 0.026212628930807114,
+      "learning_rate": 3.448e-05,
+      "loss": 0.0319,
+      "step": 7760
+    },
+    {
+      "epoch": 1.554,
+      "grad_norm": 0.0010424124775454402,
+      "learning_rate": 3.4460000000000005e-05,
+      "loss": 0.0112,
+      "step": 7770
+    },
+    {
+      "epoch": 1.556,
+      "grad_norm": 0.23557564616203308,
+      "learning_rate": 3.444e-05,
+      "loss": 0.02,
+      "step": 7780
+    },
+    {
+      "epoch": 1.558,
+      "grad_norm": 0.016170961782336235,
+      "learning_rate": 3.442e-05,
+      "loss": 0.0408,
+      "step": 7790
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 4.391359329223633,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 0.0577,
+      "step": 7800
+    },
+    {
+      "epoch": 1.562,
+      "grad_norm": 0.023899197578430176,
+      "learning_rate": 3.438e-05,
+      "loss": 0.007,
+      "step": 7810
+    },
+    {
+      "epoch": 1.564,
+      "grad_norm": 1.6332088708877563,
+      "learning_rate": 3.436e-05,
+      "loss": 0.0548,
+      "step": 7820
+    },
+    {
+      "epoch": 1.5659999999999998,
+      "grad_norm": 0.06435178220272064,
+      "learning_rate": 3.434e-05,
+      "loss": 0.0006,
+      "step": 7830
+    },
+    {
+      "epoch": 1.568,
+      "grad_norm": 0.019996432587504387,
+      "learning_rate": 3.4320000000000003e-05,
+      "loss": 0.002,
+      "step": 7840
+    },
+    {
+      "epoch": 1.5699999999999998,
+      "grad_norm": 0.010787452571094036,
+      "learning_rate": 3.430000000000001e-05,
+      "loss": 0.0024,
+      "step": 7850
+    },
+    {
+      "epoch": 1.572,
+      "grad_norm": 0.012626455165445805,
+      "learning_rate": 3.4280000000000004e-05,
+      "loss": 0.0193,
+      "step": 7860
+    },
+    {
+      "epoch": 1.5739999999999998,
+      "grad_norm": 0.029891546815633774,
+      "learning_rate": 3.426e-05,
+      "loss": 0.0032,
+      "step": 7870
+    },
+    {
+      "epoch": 1.576,
+      "grad_norm": 0.006990303285419941,
+      "learning_rate": 3.424e-05,
+      "loss": 0.0304,
+      "step": 7880
+    },
+    {
+      "epoch": 1.5779999999999998,
+      "grad_norm": 0.005956538952887058,
+      "learning_rate": 3.422e-05,
+      "loss": 0.0004,
+      "step": 7890
+    },
+    {
+      "epoch": 1.58,
+      "grad_norm": 0.07458526641130447,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 0.0878,
+      "step": 7900
+    },
+    {
+      "epoch": 1.5819999999999999,
+      "grad_norm": 0.010691751725971699,
+      "learning_rate": 3.418e-05,
+      "loss": 0.0007,
+      "step": 7910
+    },
+    {
+      "epoch": 1.584,
+      "grad_norm": 0.015178173780441284,
+      "learning_rate": 3.4160000000000005e-05,
+      "loss": 0.003,
+      "step": 7920
+    },
+    {
+      "epoch": 1.5859999999999999,
+      "grad_norm": 0.025410959497094154,
+      "learning_rate": 3.414e-05,
+      "loss": 0.0011,
+      "step": 7930
+    },
+    {
+      "epoch": 1.588,
+      "grad_norm": 0.014401872642338276,
+      "learning_rate": 3.412e-05,
+      "loss": 0.0046,
+      "step": 7940
+    },
+    {
+      "epoch": 1.5899999999999999,
+      "grad_norm": 0.006295850966125727,
+      "learning_rate": 3.41e-05,
+      "loss": 0.0879,
+      "step": 7950
+    },
+    {
+      "epoch": 1.592,
+      "grad_norm": 0.09975335747003555,
+      "learning_rate": 3.408e-05,
+      "loss": 0.0503,
+      "step": 7960
+    },
+    {
+      "epoch": 1.5939999999999999,
+      "grad_norm": 0.017794815823435783,
+      "learning_rate": 3.406e-05,
+      "loss": 0.0421,
+      "step": 7970
+    },
+    {
+      "epoch": 1.596,
+      "grad_norm": 0.9266834259033203,
+      "learning_rate": 3.404e-05,
+      "loss": 0.0038,
+      "step": 7980
+    },
+    {
+      "epoch": 1.5979999999999999,
+      "grad_norm": 0.013778113760054111,
+      "learning_rate": 3.402e-05,
+      "loss": 0.0005,
+      "step": 7990
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 0.0268776323646307,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 0.0183,
+      "step": 8000
+    },
+    {
+      "epoch": 1.6019999999999999,
+      "grad_norm": 0.006763918790966272,
+      "learning_rate": 3.398e-05,
+      "loss": 0.0108,
+      "step": 8010
+    },
+    {
+      "epoch": 1.604,
+      "grad_norm": 6.1889872550964355,
+      "learning_rate": 3.396e-05,
+      "loss": 0.0292,
+      "step": 8020
+    },
+    {
+      "epoch": 1.6059999999999999,
+      "grad_norm": 0.023212481290102005,
+      "learning_rate": 3.394e-05,
+      "loss": 0.0009,
+      "step": 8030
+    },
+    {
+      "epoch": 1.608,
+      "grad_norm": 0.33053529262542725,
+      "learning_rate": 3.392e-05,
+      "loss": 0.0011,
+      "step": 8040
+    },
+    {
+      "epoch": 1.6099999999999999,
+      "grad_norm": 0.005435945466160774,
+      "learning_rate": 3.3900000000000004e-05,
+      "loss": 0.042,
+      "step": 8050
+    },
+    {
+      "epoch": 1.612,
+      "grad_norm": 2.227583408355713,
+      "learning_rate": 3.388e-05,
+      "loss": 0.0297,
+      "step": 8060
+    },
+    {
+      "epoch": 1.6139999999999999,
+      "grad_norm": 0.06277505308389664,
+      "learning_rate": 3.3860000000000004e-05,
+      "loss": 0.0313,
+      "step": 8070
+    },
+    {
+      "epoch": 1.616,
+      "grad_norm": 0.04577331617474556,
+      "learning_rate": 3.384e-05,
+      "loss": 0.0169,
+      "step": 8080
+    },
+    {
+      "epoch": 1.6179999999999999,
+      "grad_norm": 0.0067207179963588715,
+      "learning_rate": 3.3820000000000005e-05,
+      "loss": 0.0401,
+      "step": 8090
+    },
+    {
+      "epoch": 1.62,
+      "grad_norm": 0.0028235744684934616,
+      "learning_rate": 3.38e-05,
+      "loss": 0.0007,
+      "step": 8100
+    },
+    {
+      "epoch": 1.6219999999999999,
+      "grad_norm": 0.02251609042286873,
+      "learning_rate": 3.378e-05,
+      "loss": 0.0388,
+      "step": 8110
+    },
+    {
+      "epoch": 1.624,
+      "grad_norm": 0.008003746159374714,
+      "learning_rate": 3.376e-05,
+      "loss": 0.0017,
+      "step": 8120
+    },
+    {
+      "epoch": 1.626,
+      "grad_norm": 0.02333793044090271,
+      "learning_rate": 3.374e-05,
+      "loss": 0.0006,
+      "step": 8130
+    },
+    {
+      "epoch": 1.6280000000000001,
+      "grad_norm": 0.01334973331540823,
+      "learning_rate": 3.372e-05,
+      "loss": 0.0095,
+      "step": 8140
+    },
+    {
+      "epoch": 1.63,
+      "grad_norm": 0.04795200005173683,
+      "learning_rate": 3.3700000000000006e-05,
+      "loss": 0.0131,
+      "step": 8150
+    },
+    {
+      "epoch": 1.6320000000000001,
+      "grad_norm": 0.014241354539990425,
+      "learning_rate": 3.368e-05,
+      "loss": 0.0164,
+      "step": 8160
+    },
+    {
+      "epoch": 1.634,
+      "grad_norm": 0.0266971867531538,
+      "learning_rate": 3.366e-05,
+      "loss": 0.0011,
+      "step": 8170
+    },
+    {
+      "epoch": 1.6360000000000001,
+      "grad_norm": 0.0070693595334887505,
+      "learning_rate": 3.3639999999999996e-05,
+      "loss": 0.0207,
+      "step": 8180
+    },
+    {
+      "epoch": 1.638,
+      "grad_norm": 0.001665965304709971,
+      "learning_rate": 3.362e-05,
+      "loss": 0.0005,
+      "step": 8190
+    },
+    {
+      "epoch": 1.6400000000000001,
+      "grad_norm": 0.0016495983581990004,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 0.001,
+      "step": 8200
+    },
+    {
+      "epoch": 1.642,
+      "grad_norm": 0.007819821126759052,
+      "learning_rate": 3.358e-05,
+      "loss": 0.0066,
+      "step": 8210
+    },
+    {
+      "epoch": 1.6440000000000001,
+      "grad_norm": 0.002417607232928276,
+      "learning_rate": 3.3560000000000004e-05,
+      "loss": 0.0028,
+      "step": 8220
+    },
+    {
+      "epoch": 1.646,
+      "grad_norm": 0.016434961929917336,
+      "learning_rate": 3.354e-05,
+      "loss": 0.0005,
+      "step": 8230
+    },
+    {
+      "epoch": 1.6480000000000001,
+      "grad_norm": 7.141128063201904,
+      "learning_rate": 3.3520000000000004e-05,
+      "loss": 0.1391,
+      "step": 8240
+    },
+    {
+      "epoch": 1.65,
+      "grad_norm": 0.012079566717147827,
+      "learning_rate": 3.35e-05,
+      "loss": 0.0823,
+      "step": 8250
+    },
+    {
+      "epoch": 1.6520000000000001,
+      "grad_norm": 0.051017794758081436,
+      "learning_rate": 3.348e-05,
+      "loss": 0.0054,
+      "step": 8260
+    },
+    {
+      "epoch": 1.654,
+      "grad_norm": 0.005140508990734816,
+      "learning_rate": 3.346e-05,
+      "loss": 0.0271,
+      "step": 8270
+    },
+    {
+      "epoch": 1.6560000000000001,
+      "grad_norm": 0.017056677490472794,
+      "learning_rate": 3.344e-05,
+      "loss": 0.0892,
+      "step": 8280
+    },
+    {
+      "epoch": 1.658,
+      "grad_norm": 0.024035515263676643,
+      "learning_rate": 3.342e-05,
+      "loss": 0.0075,
+      "step": 8290
+    },
+    {
+      "epoch": 1.6600000000000001,
+      "grad_norm": 4.052472114562988,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 0.0297,
+      "step": 8300
+    },
+    {
+      "epoch": 1.662,
+      "grad_norm": 0.006922661792486906,
+      "learning_rate": 3.338e-05,
+      "loss": 0.0558,
+      "step": 8310
+    },
+    {
+      "epoch": 1.6640000000000001,
+      "grad_norm": 0.018334083259105682,
+      "learning_rate": 3.336e-05,
+      "loss": 0.0004,
+      "step": 8320
+    },
+    {
+      "epoch": 1.666,
+      "grad_norm": 0.011962796561419964,
+      "learning_rate": 3.3339999999999996e-05,
+      "loss": 0.0421,
+      "step": 8330
+    },
+    {
+      "epoch": 1.6680000000000001,
+      "grad_norm": 0.00863931979984045,
+      "learning_rate": 3.332e-05,
+      "loss": 0.0386,
+      "step": 8340
+    },
+    {
+      "epoch": 1.67,
+      "grad_norm": 0.016877811402082443,
+      "learning_rate": 3.33e-05,
+      "loss": 0.0013,
+      "step": 8350
+    },
+    {
+      "epoch": 1.6720000000000002,
+      "grad_norm": 3.0169286727905273,
+      "learning_rate": 3.328e-05,
+      "loss": 0.0532,
+      "step": 8360
+    },
+    {
+      "epoch": 1.674,
+      "grad_norm": 0.00712529057636857,
+      "learning_rate": 3.3260000000000003e-05,
+      "loss": 0.0355,
+      "step": 8370
+    },
+    {
+      "epoch": 1.6760000000000002,
+      "grad_norm": 0.008485515601933002,
+      "learning_rate": 3.324e-05,
+      "loss": 0.0014,
+      "step": 8380
+    },
+    {
+      "epoch": 1.678,
+      "grad_norm": 0.034667760133743286,
+      "learning_rate": 3.3220000000000004e-05,
+      "loss": 0.0248,
+      "step": 8390
+    },
+    {
+      "epoch": 1.6800000000000002,
+      "grad_norm": 0.007112329825758934,
+      "learning_rate": 3.32e-05,
+      "loss": 0.0042,
+      "step": 8400
+    },
+    {
+      "epoch": 1.682,
+      "grad_norm": 0.0033175209537148476,
+      "learning_rate": 3.318e-05,
+      "loss": 0.0395,
+      "step": 8410
+    },
+    {
+      "epoch": 1.6840000000000002,
+      "grad_norm": 4.283847808837891,
+      "learning_rate": 3.316e-05,
+      "loss": 0.0193,
+      "step": 8420
+    },
+    {
+      "epoch": 1.686,
+      "grad_norm": 0.006708692759275436,
+      "learning_rate": 3.314e-05,
+      "loss": 0.0008,
+      "step": 8430
+    },
+    {
+      "epoch": 1.688,
+      "grad_norm": 0.6214526891708374,
+      "learning_rate": 3.312e-05,
+      "loss": 0.0018,
+      "step": 8440
+    },
+    {
+      "epoch": 1.69,
+      "grad_norm": 0.01082578394562006,
+      "learning_rate": 3.3100000000000005e-05,
+      "loss": 0.0012,
+      "step": 8450
+    },
+    {
+      "epoch": 1.692,
+      "grad_norm": 0.016304902732372284,
+      "learning_rate": 3.308e-05,
+      "loss": 0.0036,
+      "step": 8460
+    },
+    {
+      "epoch": 1.694,
+      "grad_norm": 0.12819497287273407,
+      "learning_rate": 3.3060000000000005e-05,
+      "loss": 0.0012,
+      "step": 8470
+    },
+    {
+      "epoch": 1.696,
+      "grad_norm": 0.0035316101275384426,
+      "learning_rate": 3.304e-05,
+      "loss": 0.0157,
+      "step": 8480
+    },
+    {
+      "epoch": 1.698,
+      "grad_norm": 0.0035381668712943792,
+      "learning_rate": 3.302e-05,
+      "loss": 0.0006,
+      "step": 8490
+    },
+    {
+      "epoch": 1.7,
+      "grad_norm": 0.0053281825967133045,
+      "learning_rate": 3.3e-05,
+      "loss": 0.0067,
+      "step": 8500
+    },
+    {
+      "epoch": 1.702,
+      "grad_norm": 0.001686160103417933,
+      "learning_rate": 3.298e-05,
+      "loss": 0.0011,
+      "step": 8510
+    },
+    {
+      "epoch": 1.704,
+      "grad_norm": 0.039347004145383835,
+      "learning_rate": 3.296e-05,
+      "loss": 0.0004,
+      "step": 8520
+    },
+    {
+      "epoch": 1.706,
+      "grad_norm": 0.015970556065440178,
+      "learning_rate": 3.2940000000000006e-05,
+      "loss": 0.0412,
+      "step": 8530
+    },
+    {
+      "epoch": 1.708,
+      "grad_norm": 0.2755222022533417,
+      "learning_rate": 3.292e-05,
+      "loss": 0.001,
+      "step": 8540
+    },
+    {
+      "epoch": 1.71,
+      "grad_norm": 0.03691761940717697,
+      "learning_rate": 3.29e-05,
+      "loss": 0.0005,
+      "step": 8550
+    },
+    {
+      "epoch": 1.712,
+      "grad_norm": 0.002575872465968132,
+      "learning_rate": 3.288e-05,
+      "loss": 0.0687,
+      "step": 8560
+    },
+    {
+      "epoch": 1.714,
+      "grad_norm": 0.17692694067955017,
+      "learning_rate": 3.286e-05,
+      "loss": 0.0306,
+      "step": 8570
+    },
+    {
+      "epoch": 1.716,
+      "grad_norm": 0.0052223242819309235,
+      "learning_rate": 3.2840000000000004e-05,
+      "loss": 0.0002,
+      "step": 8580
+    },
+    {
+      "epoch": 1.718,
+      "grad_norm": 0.014517701230943203,
+      "learning_rate": 3.282e-05,
+      "loss": 0.0261,
+      "step": 8590
+    },
+    {
+      "epoch": 1.72,
+      "grad_norm": 0.002597847953438759,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 0.0033,
+      "step": 8600
+    },
+    {
+      "epoch": 1.722,
+      "grad_norm": 0.002658716170117259,
+      "learning_rate": 3.278e-05,
+      "loss": 0.0017,
+      "step": 8610
+    },
+    {
+      "epoch": 1.724,
+      "grad_norm": 0.13059301674365997,
+      "learning_rate": 3.2760000000000005e-05,
+      "loss": 0.0492,
+      "step": 8620
+    },
+    {
+      "epoch": 1.726,
+      "grad_norm": 0.012524668127298355,
+      "learning_rate": 3.274e-05,
+      "loss": 0.0004,
+      "step": 8630
+    },
+    {
+      "epoch": 1.728,
+      "grad_norm": 8.583258628845215,
+      "learning_rate": 3.272e-05,
+      "loss": 0.0534,
+      "step": 8640
+    },
+    {
+      "epoch": 1.73,
+      "grad_norm": 0.0053812297992408276,
+      "learning_rate": 3.27e-05,
+      "loss": 0.0078,
+      "step": 8650
+    },
+    {
+      "epoch": 1.732,
+      "grad_norm": 0.005391386337578297,
+      "learning_rate": 3.268e-05,
+      "loss": 0.0112,
+      "step": 8660
+    },
+    {
+      "epoch": 1.734,
+      "grad_norm": 0.0013112288434058428,
+      "learning_rate": 3.266e-05,
+      "loss": 0.0574,
+      "step": 8670
+    },
+    {
+      "epoch": 1.736,
+      "grad_norm": 0.014318425208330154,
+      "learning_rate": 3.2640000000000006e-05,
+      "loss": 0.0364,
+      "step": 8680
+    },
+    {
+      "epoch": 1.738,
+      "grad_norm": 0.03327896445989609,
+      "learning_rate": 3.262e-05,
+      "loss": 0.0168,
+      "step": 8690
+    },
+    {
+      "epoch": 1.74,
+      "grad_norm": 0.4554899036884308,
+      "learning_rate": 3.26e-05,
+      "loss": 0.009,
+      "step": 8700
+    },
+    {
+      "epoch": 1.742,
+      "grad_norm": 5.39903450012207,
+      "learning_rate": 3.2579999999999996e-05,
+      "loss": 0.009,
+      "step": 8710
+    },
+    {
+      "epoch": 1.744,
+      "grad_norm": 0.006803723517805338,
+      "learning_rate": 3.256e-05,
+      "loss": 0.0148,
+      "step": 8720
+    },
+    {
+      "epoch": 1.746,
+      "grad_norm": 0.005707309115678072,
+      "learning_rate": 3.2540000000000004e-05,
+      "loss": 0.0532,
+      "step": 8730
+    },
+    {
+      "epoch": 1.748,
+      "grad_norm": 0.005006300285458565,
+      "learning_rate": 3.252e-05,
+      "loss": 0.0272,
+      "step": 8740
+    },
+    {
+      "epoch": 1.75,
+      "grad_norm": 0.012628933414816856,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 0.0625,
+      "step": 8750
+    },
+    {
+      "epoch": 1.752,
+      "grad_norm": 0.2593526542186737,
+      "learning_rate": 3.248e-05,
+      "loss": 0.004,
+      "step": 8760
+    },
+    {
+      "epoch": 1.754,
+      "grad_norm": 0.012107155285775661,
+      "learning_rate": 3.2460000000000004e-05,
+      "loss": 0.0116,
+      "step": 8770
+    },
+    {
+      "epoch": 1.756,
+      "grad_norm": 0.012258902192115784,
+      "learning_rate": 3.244e-05,
+      "loss": 0.0075,
+      "step": 8780
+    },
+    {
+      "epoch": 1.758,
+      "grad_norm": 0.007064484991133213,
+      "learning_rate": 3.242e-05,
+      "loss": 0.0338,
+      "step": 8790
+    },
+    {
+      "epoch": 1.76,
+      "grad_norm": 0.28371575474739075,
+      "learning_rate": 3.24e-05,
+      "loss": 0.0526,
+      "step": 8800
+    },
+    {
+      "epoch": 1.762,
+      "grad_norm": 3.631308078765869,
+      "learning_rate": 3.238e-05,
+      "loss": 0.0622,
+      "step": 8810
+    },
+    {
+      "epoch": 1.764,
+      "grad_norm": 0.022568710148334503,
+      "learning_rate": 3.236e-05,
+      "loss": 0.0618,
+      "step": 8820
+    },
+    {
+      "epoch": 1.766,
+      "grad_norm": 0.0031279297545552254,
+      "learning_rate": 3.2340000000000005e-05,
+      "loss": 0.02,
+      "step": 8830
+    },
+    {
+      "epoch": 1.768,
+      "grad_norm": 0.0031439117155969143,
+      "learning_rate": 3.232e-05,
+      "loss": 0.0007,
+      "step": 8840
+    },
+    {
+      "epoch": 1.77,
+      "grad_norm": 0.013959056697785854,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 0.003,
+      "step": 8850
+    },
+    {
+      "epoch": 1.772,
+      "grad_norm": 0.004907474387437105,
+      "learning_rate": 3.2279999999999996e-05,
+      "loss": 0.047,
+      "step": 8860
+    },
+    {
+      "epoch": 1.774,
+      "grad_norm": 4.480982780456543,
+      "learning_rate": 3.226e-05,
+      "loss": 0.0606,
+      "step": 8870
+    },
+    {
+      "epoch": 1.776,
+      "grad_norm": 0.17286856472492218,
+      "learning_rate": 3.224e-05,
+      "loss": 0.0044,
+      "step": 8880
+    },
+    {
+      "epoch": 1.778,
+      "grad_norm": 0.008956176228821278,
+      "learning_rate": 3.222e-05,
+      "loss": 0.0461,
+      "step": 8890
+    },
+    {
+      "epoch": 1.78,
+      "grad_norm": 0.01405914593487978,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 0.027,
+      "step": 8900
+    },
+    {
+      "epoch": 1.782,
+      "grad_norm": 0.02167750708758831,
+      "learning_rate": 3.218e-05,
+      "loss": 0.0037,
+      "step": 8910
+    },
+    {
+      "epoch": 1.784,
+      "grad_norm": 0.009961246512830257,
+      "learning_rate": 3.2160000000000004e-05,
+      "loss": 0.0256,
+      "step": 8920
+    },
+    {
+      "epoch": 1.786,
+      "grad_norm": 0.012369144707918167,
+      "learning_rate": 3.214e-05,
+      "loss": 0.0025,
+      "step": 8930
+    },
+    {
+      "epoch": 1.788,
+      "grad_norm": 0.9500744938850403,
+      "learning_rate": 3.212e-05,
+      "loss": 0.0311,
+      "step": 8940
+    },
+    {
+      "epoch": 1.79,
+      "grad_norm": 0.0016900122864171863,
+      "learning_rate": 3.21e-05,
+      "loss": 0.0007,
+      "step": 8950
+    },
+    {
+      "epoch": 1.792,
+      "grad_norm": 0.002476219553500414,
+      "learning_rate": 3.208e-05,
+      "loss": 0.0279,
+      "step": 8960
+    },
+    {
+      "epoch": 1.794,
+      "grad_norm": 0.004267267417162657,
+      "learning_rate": 3.206e-05,
+      "loss": 0.0004,
+      "step": 8970
+    },
+    {
+      "epoch": 1.796,
+      "grad_norm": 0.04748089611530304,
+      "learning_rate": 3.2040000000000005e-05,
+      "loss": 0.0236,
+      "step": 8980
+    },
+    {
+      "epoch": 1.798,
+      "grad_norm": 0.03387998789548874,
+      "learning_rate": 3.202e-05,
+      "loss": 0.0004,
+      "step": 8990
+    },
+    {
+      "epoch": 1.8,
+      "grad_norm": 0.0028094626031816006,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 0.0123,
+      "step": 9000
+    },
+    {
+      "epoch": 1.802,
+      "grad_norm": 0.001941610942594707,
+      "learning_rate": 3.198e-05,
+      "loss": 0.0037,
+      "step": 9010
+    },
+    {
+      "epoch": 1.804,
+      "grad_norm": 5.164578914642334,
+      "learning_rate": 3.196e-05,
+      "loss": 0.0457,
+      "step": 9020
+    },
+    {
+      "epoch": 1.806,
+      "grad_norm": 0.8224331140518188,
+      "learning_rate": 3.194e-05,
+      "loss": 0.0011,
+      "step": 9030
+    },
+    {
+      "epoch": 1.808,
+      "grad_norm": 0.00412099901586771,
+      "learning_rate": 3.192e-05,
+      "loss": 0.0061,
+      "step": 9040
+    },
+    {
+      "epoch": 1.81,
+      "grad_norm": 3.7033016681671143,
+      "learning_rate": 3.19e-05,
+      "loss": 0.0528,
+      "step": 9050
+    },
+    {
+      "epoch": 1.812,
+      "grad_norm": 0.004599456675350666,
+      "learning_rate": 3.188e-05,
+      "loss": 0.0008,
+      "step": 9060
+    },
+    {
+      "epoch": 1.814,
+      "grad_norm": 0.011987713165581226,
+      "learning_rate": 3.186e-05,
+      "loss": 0.0007,
+      "step": 9070
+    },
+    {
+      "epoch": 1.8159999999999998,
+      "grad_norm": 0.006984145380556583,
+      "learning_rate": 3.184e-05,
+      "loss": 0.0008,
+      "step": 9080
+    },
+    {
+      "epoch": 1.818,
+      "grad_norm": 0.0012510290835052729,
+      "learning_rate": 3.182e-05,
+      "loss": 0.0006,
+      "step": 9090
+    },
+    {
+      "epoch": 1.8199999999999998,
+      "grad_norm": 0.008051293902099133,
+      "learning_rate": 3.18e-05,
+      "loss": 0.0002,
+      "step": 9100
+    },
+    {
+      "epoch": 1.822,
+      "grad_norm": 0.001987141091376543,
+      "learning_rate": 3.1780000000000004e-05,
+      "loss": 0.0009,
+      "step": 9110
+    },
+    {
+      "epoch": 1.8239999999999998,
+      "grad_norm": 0.003431364195421338,
+      "learning_rate": 3.176e-05,
+      "loss": 0.0038,
+      "step": 9120
+    },
+    {
+      "epoch": 1.826,
+      "grad_norm": 0.0020495145581662655,
+      "learning_rate": 3.1740000000000004e-05,
+      "loss": 0.0013,
+      "step": 9130
+    },
+    {
+      "epoch": 1.8279999999999998,
+      "grad_norm": 0.005789860151708126,
+      "learning_rate": 3.172e-05,
+      "loss": 0.0002,
+      "step": 9140
+    },
+    {
+      "epoch": 1.83,
+      "grad_norm": 0.012445029802620411,
+      "learning_rate": 3.1700000000000005e-05,
+      "loss": 0.0029,
+      "step": 9150
+    },
+    {
+      "epoch": 1.8319999999999999,
+      "grad_norm": 1.2400474548339844,
+      "learning_rate": 3.168e-05,
+      "loss": 0.0008,
+      "step": 9160
+    },
+    {
+      "epoch": 1.834,
+      "grad_norm": 0.008145655505359173,
+      "learning_rate": 3.166e-05,
+      "loss": 0.0047,
+      "step": 9170
+    },
+    {
+      "epoch": 1.8359999999999999,
+      "grad_norm": 0.001066625933162868,
+      "learning_rate": 3.164e-05,
+      "loss": 0.0006,
+      "step": 9180
+    },
+    {
+      "epoch": 1.838,
+      "grad_norm": 0.00029511668253690004,
+      "learning_rate": 3.162e-05,
+      "loss": 0.0787,
+      "step": 9190
+    },
+    {
+      "epoch": 1.8399999999999999,
+      "grad_norm": 0.003314885776489973,
+      "learning_rate": 3.16e-05,
+      "loss": 0.0002,
+      "step": 9200
+    },
+    {
+      "epoch": 1.842,
+      "grad_norm": 0.00821120385080576,
+      "learning_rate": 3.1580000000000006e-05,
+      "loss": 0.0036,
+      "step": 9210
+    },
+    {
+      "epoch": 1.8439999999999999,
+      "grad_norm": 0.0037178942002356052,
+      "learning_rate": 3.156e-05,
+      "loss": 0.0464,
+      "step": 9220
+    },
+    {
+      "epoch": 1.846,
+      "grad_norm": 0.0036436987575143576,
+      "learning_rate": 3.154e-05,
+      "loss": 0.0121,
+      "step": 9230
+    },
+    {
+      "epoch": 1.8479999999999999,
+      "grad_norm": 0.009318316355347633,
+      "learning_rate": 3.1519999999999996e-05,
+      "loss": 0.0007,
+      "step": 9240
+    },
+    {
+      "epoch": 1.85,
+      "grad_norm": 0.001607164042070508,
+      "learning_rate": 3.15e-05,
+      "loss": 0.0448,
+      "step": 9250
+    },
+    {
+      "epoch": 1.8519999999999999,
+      "grad_norm": 0.011071178130805492,
+      "learning_rate": 3.1480000000000004e-05,
+      "loss": 0.0022,
+      "step": 9260
+    },
+    {
+      "epoch": 1.854,
+      "grad_norm": 0.034108154475688934,
+      "learning_rate": 3.146e-05,
+      "loss": 0.0045,
+      "step": 9270
+    },
+    {
+      "epoch": 1.8559999999999999,
+      "grad_norm": 0.002131303306668997,
+      "learning_rate": 3.1440000000000004e-05,
+      "loss": 0.001,
+      "step": 9280
+    },
+    {
+      "epoch": 1.858,
+      "grad_norm": 0.017153270542621613,
+      "learning_rate": 3.142e-05,
+      "loss": 0.001,
+      "step": 9290
+    },
+    {
+      "epoch": 1.8599999999999999,
+      "grad_norm": 0.036032069474458694,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 0.0003,
+      "step": 9300
+    },
+    {
+      "epoch": 1.862,
+      "grad_norm": 0.0025088468100875616,
+      "learning_rate": 3.138e-05,
+      "loss": 0.0665,
+      "step": 9310
+    },
+    {
+      "epoch": 1.8639999999999999,
+      "grad_norm": 5.318429946899414,
+      "learning_rate": 3.136e-05,
+      "loss": 0.0054,
+      "step": 9320
+    },
+    {
+      "epoch": 1.866,
+      "grad_norm": 0.008601468987762928,
+      "learning_rate": 3.134e-05,
+      "loss": 0.0413,
+      "step": 9330
+    },
+    {
+      "epoch": 1.8679999999999999,
+      "grad_norm": 8.6300687789917,
+      "learning_rate": 3.132e-05,
+      "loss": 0.0326,
+      "step": 9340
+    },
+    {
+      "epoch": 1.87,
+      "grad_norm": 0.004777857102453709,
+      "learning_rate": 3.13e-05,
+      "loss": 0.0006,
+      "step": 9350
+    },
+    {
+      "epoch": 1.8719999999999999,
+      "grad_norm": 3.332656145095825,
+      "learning_rate": 3.1280000000000005e-05,
+      "loss": 0.0028,
+      "step": 9360
+    },
+    {
+      "epoch": 1.874,
+      "grad_norm": 0.08430604636669159,
+      "learning_rate": 3.126e-05,
+      "loss": 0.0128,
+      "step": 9370
+    },
+    {
+      "epoch": 1.876,
+      "grad_norm": 0.0012110250536352396,
+      "learning_rate": 3.1240000000000006e-05,
+      "loss": 0.0733,
+      "step": 9380
+    },
+    {
+      "epoch": 1.8780000000000001,
+      "grad_norm": 0.0015350721077993512,
+      "learning_rate": 3.122e-05,
+      "loss": 0.0224,
+      "step": 9390
+    },
+    {
+      "epoch": 1.88,
+      "grad_norm": 0.0013420511968433857,
+      "learning_rate": 3.12e-05,
+      "loss": 0.0003,
+      "step": 9400
+    },
+    {
+      "epoch": 1.8820000000000001,
+      "grad_norm": 0.008449506014585495,
+      "learning_rate": 3.118e-05,
+      "loss": 0.0013,
+      "step": 9410
+    },
+    {
+      "epoch": 1.884,
+      "grad_norm": 0.09631182253360748,
+      "learning_rate": 3.116e-05,
+      "loss": 0.013,
+      "step": 9420
+    },
+    {
+      "epoch": 1.8860000000000001,
+      "grad_norm": 0.004350234754383564,
+      "learning_rate": 3.1140000000000003e-05,
+      "loss": 0.0004,
+      "step": 9430
+    },
+    {
+      "epoch": 1.888,
+      "grad_norm": 0.0016877787420526147,
+      "learning_rate": 3.112e-05,
+      "loss": 0.0768,
+      "step": 9440
+    },
+    {
+      "epoch": 1.8900000000000001,
+      "grad_norm": 6.81647253036499,
+      "learning_rate": 3.1100000000000004e-05,
+      "loss": 0.0102,
+      "step": 9450
+    },
+    {
+      "epoch": 1.892,
+      "grad_norm": 0.028952328488230705,
+      "learning_rate": 3.108e-05,
+      "loss": 0.0286,
+      "step": 9460
+    },
+    {
+      "epoch": 1.8940000000000001,
+      "grad_norm": 0.002540194895118475,
+      "learning_rate": 3.106e-05,
+      "loss": 0.0284,
+      "step": 9470
+    },
+    {
+      "epoch": 1.896,
+      "grad_norm": 0.0016325548058375716,
+      "learning_rate": 3.104e-05,
+      "loss": 0.1019,
+      "step": 9480
+    },
+    {
+      "epoch": 1.8980000000000001,
+      "grad_norm": 0.08182340115308762,
+      "learning_rate": 3.102e-05,
+      "loss": 0.0005,
+      "step": 9490
+    },
+    {
+      "epoch": 1.9,
+      "grad_norm": 0.04047470539808273,
+      "learning_rate": 3.1e-05,
+      "loss": 0.0003,
+      "step": 9500
+    },
+    {
+      "epoch": 1.9020000000000001,
+      "grad_norm": 0.0023045740090310574,
+      "learning_rate": 3.0980000000000005e-05,
+      "loss": 0.0216,
+      "step": 9510
+    },
+    {
+      "epoch": 1.904,
+      "grad_norm": 0.0018172425916418433,
+      "learning_rate": 3.096e-05,
+      "loss": 0.0315,
+      "step": 9520
+    },
+    {
+      "epoch": 1.9060000000000001,
+      "grad_norm": 0.3413315415382385,
+      "learning_rate": 3.0940000000000005e-05,
+      "loss": 0.0008,
+      "step": 9530
+    },
+    {
+      "epoch": 1.908,
+      "grad_norm": 0.002091864123940468,
+      "learning_rate": 3.092e-05,
+      "loss": 0.0094,
+      "step": 9540
+    },
+    {
+      "epoch": 1.9100000000000001,
+      "grad_norm": 0.016177967190742493,
+      "learning_rate": 3.09e-05,
+      "loss": 0.0004,
+      "step": 9550
+    },
+    {
+      "epoch": 1.912,
+      "grad_norm": 3.8514859676361084,
+      "learning_rate": 3.088e-05,
+      "loss": 0.1336,
+      "step": 9560
+    },
+    {
+      "epoch": 1.9140000000000001,
+      "grad_norm": 0.24707883596420288,
+      "learning_rate": 3.086e-05,
+      "loss": 0.0268,
+      "step": 9570
+    },
+    {
+      "epoch": 1.916,
+      "grad_norm": 2.6880104541778564,
+      "learning_rate": 3.084e-05,
+      "loss": 0.0048,
+      "step": 9580
+    },
+    {
+      "epoch": 1.9180000000000001,
+      "grad_norm": 0.06657981127500534,
+      "learning_rate": 3.082e-05,
+      "loss": 0.0442,
+      "step": 9590
+    },
+    {
+      "epoch": 1.92,
+      "grad_norm": 0.39712944626808167,
+      "learning_rate": 3.08e-05,
+      "loss": 0.0031,
+      "step": 9600
+    },
+    {
+      "epoch": 1.9220000000000002,
+      "grad_norm": 0.006004078313708305,
+      "learning_rate": 3.078e-05,
+      "loss": 0.0005,
+      "step": 9610
+    },
+    {
+      "epoch": 1.924,
+      "grad_norm": 7.00140905380249,
+      "learning_rate": 3.076e-05,
+      "loss": 0.0547,
+      "step": 9620
+    },
+    {
+      "epoch": 1.9260000000000002,
+      "grad_norm": 0.016468839719891548,
+      "learning_rate": 3.074e-05,
+      "loss": 0.0005,
+      "step": 9630
+    },
+    {
+      "epoch": 1.928,
+      "grad_norm": 0.06567908078432083,
+      "learning_rate": 3.072e-05,
+      "loss": 0.0009,
+      "step": 9640
+    },
+    {
+      "epoch": 1.9300000000000002,
+      "grad_norm": 11.211073875427246,
+      "learning_rate": 3.07e-05,
+      "loss": 0.0558,
+      "step": 9650
+    },
+    {
+      "epoch": 1.932,
+      "grad_norm": 0.008082142099738121,
+      "learning_rate": 3.0680000000000004e-05,
+      "loss": 0.0013,
+      "step": 9660
+    },
+    {
+      "epoch": 1.9340000000000002,
+      "grad_norm": 10.757091522216797,
+      "learning_rate": 3.066e-05,
+      "loss": 0.0194,
+      "step": 9670
+    },
+    {
+      "epoch": 1.936,
+      "grad_norm": 0.05785878747701645,
+      "learning_rate": 3.0640000000000005e-05,
+      "loss": 0.0006,
+      "step": 9680
+    },
+    {
+      "epoch": 1.938,
+      "grad_norm": 0.002469051396474242,
+      "learning_rate": 3.062e-05,
+      "loss": 0.0296,
+      "step": 9690
+    },
+    {
+      "epoch": 1.94,
+      "grad_norm": 0.006106134038418531,
+      "learning_rate": 3.06e-05,
+      "loss": 0.0029,
+      "step": 9700
+    },
+    {
+      "epoch": 1.942,
+      "grad_norm": 0.0031414697878062725,
+      "learning_rate": 3.058e-05,
+      "loss": 0.046,
+      "step": 9710
+    },
+    {
+      "epoch": 1.944,
+      "grad_norm": 0.011137120425701141,
+      "learning_rate": 3.056e-05,
+      "loss": 0.0308,
+      "step": 9720
+    },
+    {
+      "epoch": 1.946,
+      "grad_norm": 0.03261066600680351,
+      "learning_rate": 3.054e-05,
+      "loss": 0.0022,
+      "step": 9730
+    },
+    {
+      "epoch": 1.948,
+      "grad_norm": 0.05034400150179863,
+      "learning_rate": 3.0520000000000006e-05,
+      "loss": 0.0451,
+      "step": 9740
+    },
+    {
+      "epoch": 1.95,
+      "grad_norm": 2.934372663497925,
+      "learning_rate": 3.05e-05,
+      "loss": 0.0029,
+      "step": 9750
+    },
+    {
+      "epoch": 1.952,
+      "grad_norm": 0.0024814677890390158,
+      "learning_rate": 3.0480000000000003e-05,
+      "loss": 0.008,
+      "step": 9760
+    },
+    {
+      "epoch": 1.954,
+      "grad_norm": 0.012240572832524776,
+      "learning_rate": 3.046e-05,
+      "loss": 0.0142,
+      "step": 9770
+    },
+    {
+      "epoch": 1.956,
+      "grad_norm": 0.012755164876580238,
+      "learning_rate": 3.0440000000000003e-05,
+      "loss": 0.0056,
+      "step": 9780
+    },
+    {
+      "epoch": 1.958,
+      "grad_norm": 0.20030856132507324,
+      "learning_rate": 3.0420000000000004e-05,
+      "loss": 0.0093,
+      "step": 9790
+    },
+    {
+      "epoch": 1.96,
+      "grad_norm": 5.824985980987549,
+      "learning_rate": 3.04e-05,
+      "loss": 0.0199,
+      "step": 9800
+    },
+    {
+      "epoch": 1.962,
+      "grad_norm": 4.64138126373291,
+      "learning_rate": 3.0380000000000004e-05,
+      "loss": 0.0224,
+      "step": 9810
+    },
+    {
+      "epoch": 1.964,
+      "grad_norm": 0.002365650376304984,
+      "learning_rate": 3.036e-05,
+      "loss": 0.101,
+      "step": 9820
+    },
+    {
+      "epoch": 1.966,
+      "grad_norm": 0.0037491919938474894,
+      "learning_rate": 3.034e-05,
+      "loss": 0.0226,
+      "step": 9830
+    },
+    {
+      "epoch": 1.968,
+      "grad_norm": 0.018977370113134384,
+      "learning_rate": 3.0320000000000004e-05,
+      "loss": 0.0252,
+      "step": 9840
+    },
+    {
+      "epoch": 1.97,
+      "grad_norm": 0.006231269333511591,
+      "learning_rate": 3.03e-05,
+      "loss": 0.0017,
+      "step": 9850
+    },
+    {
+      "epoch": 1.972,
+      "grad_norm": 0.0033881240524351597,
+      "learning_rate": 3.028e-05,
+      "loss": 0.001,
+      "step": 9860
+    },
+    {
+      "epoch": 1.974,
+      "grad_norm": 0.0041534435003995895,
+      "learning_rate": 3.0259999999999998e-05,
+      "loss": 0.004,
+      "step": 9870
+    },
+    {
+      "epoch": 1.976,
+      "grad_norm": 7.33493185043335,
+      "learning_rate": 3.0240000000000002e-05,
+      "loss": 0.0649,
+      "step": 9880
+    },
+    {
+      "epoch": 1.978,
+      "grad_norm": 0.004932209383696318,
+      "learning_rate": 3.0220000000000005e-05,
+      "loss": 0.0178,
+      "step": 9890
+    },
+    {
+      "epoch": 1.98,
+      "grad_norm": 0.0016821910394355655,
+      "learning_rate": 3.02e-05,
+      "loss": 0.0047,
+      "step": 9900
+    },
+    {
+      "epoch": 1.982,
+      "grad_norm": 0.004557713866233826,
+      "learning_rate": 3.0180000000000002e-05,
+      "loss": 0.0435,
+      "step": 9910
+    },
+    {
+      "epoch": 1.984,
+      "grad_norm": 5.0917863845825195,
+      "learning_rate": 3.016e-05,
+      "loss": 0.0597,
+      "step": 9920
+    },
+    {
+      "epoch": 1.986,
+      "grad_norm": 0.0021672865841537714,
+      "learning_rate": 3.0140000000000003e-05,
+      "loss": 0.0003,
+      "step": 9930
+    },
+    {
+      "epoch": 1.988,
+      "grad_norm": 0.0008831778541207314,
+      "learning_rate": 3.0120000000000003e-05,
+      "loss": 0.0009,
+      "step": 9940
+    },
+    {
+      "epoch": 1.99,
+      "grad_norm": 0.006184358615428209,
+      "learning_rate": 3.01e-05,
+      "loss": 0.0038,
+      "step": 9950
+    },
+    {
+      "epoch": 1.992,
+      "grad_norm": 0.001267423969693482,
+      "learning_rate": 3.0080000000000003e-05,
+      "loss": 0.037,
+      "step": 9960
+    },
+    {
+      "epoch": 1.994,
+      "grad_norm": 0.005642888136208057,
+      "learning_rate": 3.006e-05,
+      "loss": 0.0051,
+      "step": 9970
+    },
+    {
+      "epoch": 1.996,
+      "grad_norm": 0.0026595494709908962,
+      "learning_rate": 3.004e-05,
+      "loss": 0.0655,
+      "step": 9980
+    },
+    {
+      "epoch": 1.998,
+      "grad_norm": 0.013930058106780052,
+      "learning_rate": 3.0020000000000004e-05,
+      "loss": 0.0271,
+      "step": 9990
+    },
+    {
+      "epoch": 2.0,
+      "grad_norm": 0.004837405402213335,
+      "learning_rate": 3e-05,
+      "loss": 0.027,
+      "step": 10000
+    },
+    {
+      "epoch": 2.0,
+      "eval_loss": 0.018645154312253,
+      "eval_runtime": 93.8446,
+      "eval_samples_per_second": 106.559,
+      "eval_steps_per_second": 13.32,
+      "step": 10000
+    },
+    {
+      "epoch": 2.002,
+      "grad_norm": 0.022907476872205734,
+      "learning_rate": 2.998e-05,
+      "loss": 0.0216,
+      "step": 10010
+    },
+    {
+      "epoch": 2.004,
+      "grad_norm": 0.19703127443790436,
+      "learning_rate": 2.9959999999999998e-05,
+      "loss": 0.002,
+      "step": 10020
+    },
+    {
+      "epoch": 2.006,
+      "grad_norm": 0.05761381983757019,
+      "learning_rate": 2.994e-05,
+      "loss": 0.0056,
+      "step": 10030
+    },
+    {
+      "epoch": 2.008,
+      "grad_norm": 0.027498705312609673,
+      "learning_rate": 2.9920000000000005e-05,
+      "loss": 0.0175,
+      "step": 10040
+    },
+    {
+      "epoch": 2.01,
+      "grad_norm": 3.6588926315307617,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 0.0119,
+      "step": 10050
+    },
+    {
+      "epoch": 2.012,
+      "grad_norm": 0.002984186401590705,
+      "learning_rate": 2.9880000000000002e-05,
+      "loss": 0.0121,
+      "step": 10060
+    },
+    {
+      "epoch": 2.014,
+      "grad_norm": 0.022666804492473602,
+      "learning_rate": 2.986e-05,
+      "loss": 0.0006,
+      "step": 10070
+    },
+    {
+      "epoch": 2.016,
+      "grad_norm": 0.015680214390158653,
+      "learning_rate": 2.9840000000000002e-05,
+      "loss": 0.0249,
+      "step": 10080
+    },
+    {
+      "epoch": 2.018,
+      "grad_norm": 0.010176047682762146,
+      "learning_rate": 2.9820000000000002e-05,
+      "loss": 0.0069,
+      "step": 10090
+    },
+    {
+      "epoch": 2.02,
+      "grad_norm": 0.1402248740196228,
+      "learning_rate": 2.98e-05,
+      "loss": 0.001,
+      "step": 10100
+    },
+    {
+      "epoch": 2.022,
+      "grad_norm": 0.039656609296798706,
+      "learning_rate": 2.9780000000000003e-05,
+      "loss": 0.0007,
+      "step": 10110
+    },
+    {
+      "epoch": 2.024,
+      "grad_norm": 0.0013894840376451612,
+      "learning_rate": 2.976e-05,
+      "loss": 0.0026,
+      "step": 10120
+    },
+    {
+      "epoch": 2.026,
+      "grad_norm": 0.05445275083184242,
+      "learning_rate": 2.974e-05,
+      "loss": 0.0387,
+      "step": 10130
+    },
+    {
+      "epoch": 2.028,
+      "grad_norm": 0.0013749465579167008,
+      "learning_rate": 2.9720000000000003e-05,
+      "loss": 0.0383,
+      "step": 10140
+    },
+    {
+      "epoch": 2.03,
+      "grad_norm": 0.005243366118520498,
+      "learning_rate": 2.97e-05,
+      "loss": 0.0094,
+      "step": 10150
+    },
+    {
+      "epoch": 2.032,
+      "grad_norm": 0.006576089188456535,
+      "learning_rate": 2.9680000000000004e-05,
+      "loss": 0.0216,
+      "step": 10160
+    },
+    {
+      "epoch": 2.034,
+      "grad_norm": 0.0026048733852803707,
+      "learning_rate": 2.9659999999999997e-05,
+      "loss": 0.0732,
+      "step": 10170
+    },
+    {
+      "epoch": 2.036,
+      "grad_norm": 0.006093551870435476,
+      "learning_rate": 2.964e-05,
+      "loss": 0.0416,
+      "step": 10180
+    },
+    {
+      "epoch": 2.038,
+      "grad_norm": 0.014537391252815723,
+      "learning_rate": 2.9620000000000004e-05,
+      "loss": 0.0017,
+      "step": 10190
+    },
+    {
+      "epoch": 2.04,
+      "grad_norm": 0.010529935359954834,
+      "learning_rate": 2.96e-05,
+      "loss": 0.0136,
+      "step": 10200
+    },
+    {
+      "epoch": 2.042,
+      "grad_norm": 2.0856871604919434,
+      "learning_rate": 2.958e-05,
+      "loss": 0.0047,
+      "step": 10210
+    },
+    {
+      "epoch": 2.044,
+      "grad_norm": 0.014749602414667606,
+      "learning_rate": 2.9559999999999998e-05,
+      "loss": 0.1169,
+      "step": 10220
+    },
+    {
+      "epoch": 2.046,
+      "grad_norm": 0.0697651207447052,
+      "learning_rate": 2.9540000000000002e-05,
+      "loss": 0.0203,
+      "step": 10230
+    },
+    {
+      "epoch": 2.048,
+      "grad_norm": 0.016631267964839935,
+      "learning_rate": 2.9520000000000002e-05,
+      "loss": 0.0015,
+      "step": 10240
+    },
+    {
+      "epoch": 2.05,
+      "grad_norm": 0.005667282734066248,
+      "learning_rate": 2.95e-05,
+      "loss": 0.0023,
+      "step": 10250
+    },
+    {
+      "epoch": 2.052,
+      "grad_norm": 0.0238018911331892,
+      "learning_rate": 2.9480000000000002e-05,
+      "loss": 0.0029,
+      "step": 10260
+    },
+    {
+      "epoch": 2.054,
+      "grad_norm": 0.007609200663864613,
+      "learning_rate": 2.946e-05,
+      "loss": 0.0003,
+      "step": 10270
+    },
+    {
+      "epoch": 2.056,
+      "grad_norm": 0.012714263051748276,
+      "learning_rate": 2.944e-05,
+      "loss": 0.0015,
+      "step": 10280
+    },
+    {
+      "epoch": 2.058,
+      "grad_norm": 0.001624607597477734,
+      "learning_rate": 2.9420000000000003e-05,
+      "loss": 0.0089,
+      "step": 10290
+    },
+    {
+      "epoch": 2.06,
+      "grad_norm": 0.057245634496212006,
+      "learning_rate": 2.94e-05,
+      "loss": 0.0062,
+      "step": 10300
+    },
+    {
+      "epoch": 2.062,
+      "grad_norm": 0.1072976291179657,
+      "learning_rate": 2.9380000000000003e-05,
+      "loss": 0.001,
+      "step": 10310
+    },
+    {
+      "epoch": 2.064,
+      "grad_norm": 0.0009422163129784167,
+      "learning_rate": 2.9360000000000003e-05,
+      "loss": 0.0004,
+      "step": 10320
+    },
+    {
+      "epoch": 2.066,
+      "grad_norm": 8.735067367553711,
+      "learning_rate": 2.934e-05,
+      "loss": 0.0731,
+      "step": 10330
+    },
+    {
+      "epoch": 2.068,
+      "grad_norm": 0.0013592693721875548,
+      "learning_rate": 2.9320000000000004e-05,
+      "loss": 0.0004,
+      "step": 10340
+    },
+    {
+      "epoch": 2.07,
+      "grad_norm": 0.03364754840731621,
+      "learning_rate": 2.93e-05,
+      "loss": 0.0191,
+      "step": 10350
+    },
+    {
+      "epoch": 2.072,
+      "grad_norm": 0.004056991543620825,
+      "learning_rate": 2.928e-05,
+      "loss": 0.001,
+      "step": 10360
+    },
+    {
+      "epoch": 2.074,
+      "grad_norm": 0.008462933823466301,
+      "learning_rate": 2.9260000000000004e-05,
+      "loss": 0.0009,
+      "step": 10370
+    },
+    {
+      "epoch": 2.076,
+      "grad_norm": 0.01646353490650654,
+      "learning_rate": 2.924e-05,
+      "loss": 0.0046,
+      "step": 10380
+    },
+    {
+      "epoch": 2.078,
+      "grad_norm": 0.010266261175274849,
+      "learning_rate": 2.922e-05,
+      "loss": 0.0382,
+      "step": 10390
+    },
+    {
+      "epoch": 2.08,
+      "grad_norm": 0.01267939805984497,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 0.0133,
+      "step": 10400
+    },
+    {
+      "epoch": 2.082,
+      "grad_norm": 0.0029067820869386196,
+      "learning_rate": 2.9180000000000002e-05,
+      "loss": 0.0407,
+      "step": 10410
+    },
+    {
+      "epoch": 2.084,
+      "grad_norm": 0.004823862109333277,
+      "learning_rate": 2.9160000000000005e-05,
+      "loss": 0.0199,
+      "step": 10420
+    },
+    {
+      "epoch": 2.086,
+      "grad_norm": 9.204266548156738,
+      "learning_rate": 2.9140000000000002e-05,
+      "loss": 0.0357,
+      "step": 10430
+    },
+    {
+      "epoch": 2.088,
+      "grad_norm": 0.0023752956185489893,
+      "learning_rate": 2.9120000000000002e-05,
+      "loss": 0.0003,
+      "step": 10440
+    },
+    {
+      "epoch": 2.09,
+      "grad_norm": 0.1312880516052246,
+      "learning_rate": 2.91e-05,
+      "loss": 0.0004,
+      "step": 10450
+    },
+    {
+      "epoch": 2.092,
+      "grad_norm": 0.0029505027923732996,
+      "learning_rate": 2.9080000000000003e-05,
+      "loss": 0.0065,
+      "step": 10460
+    },
+    {
+      "epoch": 2.094,
+      "grad_norm": 0.009261439554393291,
+      "learning_rate": 2.9060000000000003e-05,
+      "loss": 0.0132,
+      "step": 10470
+    },
+    {
+      "epoch": 2.096,
+      "grad_norm": 5.55691385269165,
+      "learning_rate": 2.904e-05,
+      "loss": 0.0151,
+      "step": 10480
+    },
+    {
+      "epoch": 2.098,
+      "grad_norm": 0.002771316794678569,
+      "learning_rate": 2.9020000000000003e-05,
+      "loss": 0.0631,
+      "step": 10490
+    },
+    {
+      "epoch": 2.1,
+      "grad_norm": 0.006200423464179039,
+      "learning_rate": 2.9e-05,
+      "loss": 0.0005,
+      "step": 10500
+    },
+    {
+      "epoch": 2.102,
+      "grad_norm": 0.9081658720970154,
+      "learning_rate": 2.898e-05,
+      "loss": 0.0021,
+      "step": 10510
+    },
+    {
+      "epoch": 2.104,
+      "grad_norm": 0.008039538748562336,
+      "learning_rate": 2.8960000000000004e-05,
+      "loss": 0.0005,
+      "step": 10520
+    },
+    {
+      "epoch": 2.106,
+      "grad_norm": 0.0029554793145507574,
+      "learning_rate": 2.894e-05,
+      "loss": 0.0003,
+      "step": 10530
+    },
+    {
+      "epoch": 2.108,
+      "grad_norm": 0.046315453946590424,
+      "learning_rate": 2.8920000000000004e-05,
+      "loss": 0.001,
+      "step": 10540
+    },
+    {
+      "epoch": 2.11,
+      "grad_norm": 0.003512721508741379,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 0.0005,
+      "step": 10550
+    },
+    {
+      "epoch": 2.112,
+      "grad_norm": 1.5078462362289429,
+      "learning_rate": 2.888e-05,
+      "loss": 0.0346,
+      "step": 10560
+    },
+    {
+      "epoch": 2.114,
+      "grad_norm": 0.017720790579915047,
+      "learning_rate": 2.8860000000000005e-05,
+      "loss": 0.0422,
+      "step": 10570
+    },
+    {
+      "epoch": 2.116,
+      "grad_norm": 0.9635635018348694,
+      "learning_rate": 2.8840000000000002e-05,
+      "loss": 0.0434,
+      "step": 10580
+    },
+    {
+      "epoch": 2.118,
+      "grad_norm": 0.0331871435046196,
+      "learning_rate": 2.8820000000000002e-05,
+      "loss": 0.0002,
+      "step": 10590
+    },
+    {
+      "epoch": 2.12,
+      "grad_norm": 0.0016624679556116462,
+      "learning_rate": 2.88e-05,
+      "loss": 0.0006,
+      "step": 10600
+    },
+    {
+      "epoch": 2.122,
+      "grad_norm": 0.0020574277732521296,
+      "learning_rate": 2.8780000000000002e-05,
+      "loss": 0.0219,
+      "step": 10610
+    },
+    {
+      "epoch": 2.124,
+      "grad_norm": 0.010397090576589108,
+      "learning_rate": 2.8760000000000002e-05,
+      "loss": 0.0763,
+      "step": 10620
+    },
+    {
+      "epoch": 2.126,
+      "grad_norm": 0.0018888169433921576,
+      "learning_rate": 2.874e-05,
+      "loss": 0.025,
+      "step": 10630
+    },
+    {
+      "epoch": 2.128,
+      "grad_norm": 0.0008277022279798985,
+      "learning_rate": 2.8720000000000003e-05,
+      "loss": 0.0002,
+      "step": 10640
+    },
+    {
+      "epoch": 2.13,
+      "grad_norm": 0.005590827204287052,
+      "learning_rate": 2.87e-05,
+      "loss": 0.0037,
+      "step": 10650
+    },
+    {
+      "epoch": 2.132,
+      "grad_norm": 2.0354769229888916,
+      "learning_rate": 2.868e-05,
+      "loss": 0.0015,
+      "step": 10660
+    },
+    {
+      "epoch": 2.134,
+      "grad_norm": 3.090385913848877,
+      "learning_rate": 2.8660000000000003e-05,
+      "loss": 0.0193,
+      "step": 10670
+    },
+    {
+      "epoch": 2.136,
+      "grad_norm": 0.0038130644243210554,
+      "learning_rate": 2.864e-05,
+      "loss": 0.0011,
+      "step": 10680
+    },
+    {
+      "epoch": 2.138,
+      "grad_norm": 0.0742473378777504,
+      "learning_rate": 2.8620000000000004e-05,
+      "loss": 0.0508,
+      "step": 10690
+    },
+    {
+      "epoch": 2.14,
+      "grad_norm": 0.008754884824156761,
+      "learning_rate": 2.86e-05,
+      "loss": 0.0269,
+      "step": 10700
+    },
+    {
+      "epoch": 2.142,
+      "grad_norm": 0.05013411119580269,
+      "learning_rate": 2.858e-05,
+      "loss": 0.0007,
+      "step": 10710
+    },
+    {
+      "epoch": 2.144,
+      "grad_norm": 0.006567743141204119,
+      "learning_rate": 2.8560000000000004e-05,
+      "loss": 0.0011,
+      "step": 10720
+    },
+    {
+      "epoch": 2.146,
+      "grad_norm": 0.0007309935172088444,
+      "learning_rate": 2.854e-05,
+      "loss": 0.0311,
+      "step": 10730
+    },
+    {
+      "epoch": 2.148,
+      "grad_norm": 0.1684633046388626,
+      "learning_rate": 2.852e-05,
+      "loss": 0.0496,
+      "step": 10740
+    },
+    {
+      "epoch": 2.15,
+      "grad_norm": 7.855876922607422,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 0.0145,
+      "step": 10750
+    },
+    {
+      "epoch": 2.152,
+      "grad_norm": 0.02232036367058754,
+      "learning_rate": 2.8480000000000002e-05,
+      "loss": 0.0366,
+      "step": 10760
+    },
+    {
+      "epoch": 2.154,
+      "grad_norm": 0.007120494265109301,
+      "learning_rate": 2.8460000000000002e-05,
+      "loss": 0.0795,
+      "step": 10770
+    },
+    {
+      "epoch": 2.156,
+      "grad_norm": 0.06669113039970398,
+      "learning_rate": 2.844e-05,
+      "loss": 0.0342,
+      "step": 10780
+    },
+    {
+      "epoch": 2.158,
+      "grad_norm": 0.004165279679000378,
+      "learning_rate": 2.8420000000000002e-05,
+      "loss": 0.0007,
+      "step": 10790
+    },
+    {
+      "epoch": 2.16,
+      "grad_norm": 0.0076158307492733,
+      "learning_rate": 2.84e-05,
+      "loss": 0.0011,
+      "step": 10800
+    },
+    {
+      "epoch": 2.162,
+      "grad_norm": 0.043945301324129105,
+      "learning_rate": 2.8380000000000003e-05,
+      "loss": 0.0146,
+      "step": 10810
+    },
+    {
+      "epoch": 2.164,
+      "grad_norm": 0.02764349803328514,
+      "learning_rate": 2.8360000000000003e-05,
+      "loss": 0.0007,
+      "step": 10820
+    },
+    {
+      "epoch": 2.166,
+      "grad_norm": 0.006873734761029482,
+      "learning_rate": 2.834e-05,
+      "loss": 0.0439,
+      "step": 10830
+    },
+    {
+      "epoch": 2.168,
+      "grad_norm": 0.1033577173948288,
+      "learning_rate": 2.8320000000000003e-05,
+      "loss": 0.0008,
+      "step": 10840
+    },
+    {
+      "epoch": 2.17,
+      "grad_norm": 0.00217875256203115,
+      "learning_rate": 2.83e-05,
+      "loss": 0.0024,
+      "step": 10850
+    },
+    {
+      "epoch": 2.172,
+      "grad_norm": 0.004244733136147261,
+      "learning_rate": 2.828e-05,
+      "loss": 0.0034,
+      "step": 10860
+    },
+    {
+      "epoch": 2.174,
+      "grad_norm": 2.9411137104034424,
+      "learning_rate": 2.8260000000000004e-05,
+      "loss": 0.0159,
+      "step": 10870
+    },
+    {
+      "epoch": 2.176,
+      "grad_norm": 0.00518818199634552,
+      "learning_rate": 2.824e-05,
+      "loss": 0.1114,
+      "step": 10880
+    },
+    {
+      "epoch": 2.178,
+      "grad_norm": 0.042840536683797836,
+      "learning_rate": 2.822e-05,
+      "loss": 0.0241,
+      "step": 10890
+    },
+    {
+      "epoch": 2.18,
+      "grad_norm": 0.006402214057743549,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 0.0014,
+      "step": 10900
+    },
+    {
+      "epoch": 2.182,
+      "grad_norm": 0.053806014358997345,
+      "learning_rate": 2.818e-05,
+      "loss": 0.0045,
+      "step": 10910
+    },
+    {
+      "epoch": 2.184,
+      "grad_norm": 0.019120024517178535,
+      "learning_rate": 2.816e-05,
+      "loss": 0.0017,
+      "step": 10920
+    },
+    {
+      "epoch": 2.186,
+      "grad_norm": 0.050662651658058167,
+      "learning_rate": 2.8139999999999998e-05,
+      "loss": 0.0121,
+      "step": 10930
+    },
+    {
+      "epoch": 2.188,
+      "grad_norm": 0.004558287560939789,
+      "learning_rate": 2.8120000000000002e-05,
+      "loss": 0.0123,
+      "step": 10940
+    },
+    {
+      "epoch": 2.19,
+      "grad_norm": 0.07387060672044754,
+      "learning_rate": 2.8100000000000005e-05,
+      "loss": 0.0156,
+      "step": 10950
+    },
+    {
+      "epoch": 2.192,
+      "grad_norm": 0.013174938969314098,
+      "learning_rate": 2.8080000000000002e-05,
+      "loss": 0.0004,
+      "step": 10960
+    },
+    {
+      "epoch": 2.194,
+      "grad_norm": 0.08706201612949371,
+      "learning_rate": 2.8060000000000002e-05,
+      "loss": 0.0035,
+      "step": 10970
+    },
+    {
+      "epoch": 2.196,
+      "grad_norm": 0.05324581637978554,
+      "learning_rate": 2.804e-05,
+      "loss": 0.0738,
+      "step": 10980
+    },
+    {
+      "epoch": 2.198,
+      "grad_norm": 2.949378252029419,
+      "learning_rate": 2.8020000000000003e-05,
+      "loss": 0.0015,
+      "step": 10990
+    },
+    {
+      "epoch": 2.2,
+      "grad_norm": 0.03104395419359207,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 0.0143,
+      "step": 11000
+    },
+    {
+      "epoch": 2.202,
+      "grad_norm": 0.0049886321648955345,
+      "learning_rate": 2.798e-05,
+      "loss": 0.0633,
+      "step": 11010
+    },
+    {
+      "epoch": 2.204,
+      "grad_norm": 0.019628267735242844,
+      "learning_rate": 2.7960000000000003e-05,
+      "loss": 0.0004,
+      "step": 11020
+    },
+    {
+      "epoch": 2.206,
+      "grad_norm": 0.03688063099980354,
+      "learning_rate": 2.794e-05,
+      "loss": 0.0004,
+      "step": 11030
+    },
+    {
+      "epoch": 2.208,
+      "grad_norm": 0.08032286167144775,
+      "learning_rate": 2.792e-05,
+      "loss": 0.0799,
+      "step": 11040
+    },
+    {
+      "epoch": 2.21,
+      "grad_norm": 0.0025247232988476753,
+      "learning_rate": 2.7900000000000004e-05,
+      "loss": 0.0017,
+      "step": 11050
+    },
+    {
+      "epoch": 2.212,
+      "grad_norm": 8.440220832824707,
+      "learning_rate": 2.788e-05,
+      "loss": 0.0193,
+      "step": 11060
+    },
+    {
+      "epoch": 2.214,
+      "grad_norm": 0.019714640453457832,
+      "learning_rate": 2.7860000000000004e-05,
+      "loss": 0.022,
+      "step": 11070
+    },
+    {
+      "epoch": 2.216,
+      "grad_norm": 0.0017304858192801476,
+      "learning_rate": 2.7839999999999998e-05,
+      "loss": 0.0128,
+      "step": 11080
+    },
+    {
+      "epoch": 2.218,
+      "grad_norm": 0.002389088971540332,
+      "learning_rate": 2.782e-05,
+      "loss": 0.0003,
+      "step": 11090
+    },
+    {
+      "epoch": 2.22,
+      "grad_norm": 0.002172302920371294,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 0.001,
+      "step": 11100
+    },
+    {
+      "epoch": 2.222,
+      "grad_norm": 0.009733350947499275,
+      "learning_rate": 2.778e-05,
+      "loss": 0.0004,
+      "step": 11110
+    },
+    {
+      "epoch": 2.224,
+      "grad_norm": 0.0028315680101513863,
+      "learning_rate": 2.7760000000000002e-05,
+      "loss": 0.0003,
+      "step": 11120
+    },
+    {
+      "epoch": 2.226,
+      "grad_norm": 0.6095342040061951,
+      "learning_rate": 2.774e-05,
+      "loss": 0.0045,
+      "step": 11130
+    },
+    {
+      "epoch": 2.228,
+      "grad_norm": 0.0029441171791404486,
+      "learning_rate": 2.7720000000000002e-05,
+      "loss": 0.0003,
+      "step": 11140
+    },
+    {
+      "epoch": 2.23,
+      "grad_norm": 0.0030777750071138144,
+      "learning_rate": 2.7700000000000002e-05,
+      "loss": 0.0044,
+      "step": 11150
+    },
+    {
+      "epoch": 2.232,
+      "grad_norm": 0.0014329582918435335,
+      "learning_rate": 2.768e-05,
+      "loss": 0.0071,
+      "step": 11160
+    },
+    {
+      "epoch": 2.234,
+      "grad_norm": 0.0009634523303247988,
+      "learning_rate": 2.7660000000000003e-05,
+      "loss": 0.0248,
+      "step": 11170
+    },
+    {
+      "epoch": 2.2359999999999998,
+      "grad_norm": 0.008729507215321064,
+      "learning_rate": 2.764e-05,
+      "loss": 0.0004,
+      "step": 11180
+    },
+    {
+      "epoch": 2.238,
+      "grad_norm": 0.004630973096936941,
+      "learning_rate": 2.762e-05,
+      "loss": 0.0105,
+      "step": 11190
+    },
+    {
+      "epoch": 2.24,
+      "grad_norm": 0.003606861224398017,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 0.0251,
+      "step": 11200
+    },
+    {
+      "epoch": 2.242,
+      "grad_norm": 0.004198799841105938,
+      "learning_rate": 2.758e-05,
+      "loss": 0.003,
+      "step": 11210
+    },
+    {
+      "epoch": 2.2439999999999998,
+      "grad_norm": 0.23881489038467407,
+      "learning_rate": 2.7560000000000004e-05,
+      "loss": 0.0153,
+      "step": 11220
+    },
+    {
+      "epoch": 2.246,
+      "grad_norm": 0.009163503535091877,
+      "learning_rate": 2.754e-05,
+      "loss": 0.001,
+      "step": 11230
+    },
+    {
+      "epoch": 2.248,
+      "grad_norm": 2.6871395111083984,
+      "learning_rate": 2.752e-05,
+      "loss": 0.0262,
+      "step": 11240
+    },
+    {
+      "epoch": 2.25,
+      "grad_norm": 0.02001805603504181,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 0.0015,
+      "step": 11250
+    },
+    {
+      "epoch": 2.252,
+      "grad_norm": 0.014580647461116314,
+      "learning_rate": 2.748e-05,
+      "loss": 0.0033,
+      "step": 11260
+    },
+    {
+      "epoch": 2.254,
+      "grad_norm": 0.010014360770583153,
+      "learning_rate": 2.746e-05,
+      "loss": 0.0067,
+      "step": 11270
+    },
+    {
+      "epoch": 2.2560000000000002,
+      "grad_norm": 0.02044554613530636,
+      "learning_rate": 2.7439999999999998e-05,
+      "loss": 0.001,
+      "step": 11280
+    },
+    {
+      "epoch": 2.258,
+      "grad_norm": 0.0019547122064977884,
+      "learning_rate": 2.7420000000000002e-05,
+      "loss": 0.0009,
+      "step": 11290
+    },
+    {
+      "epoch": 2.26,
+      "grad_norm": 0.02893057093024254,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 0.0019,
+      "step": 11300
+    },
+    {
+      "epoch": 2.262,
+      "grad_norm": 0.003301553428173065,
+      "learning_rate": 2.738e-05,
+      "loss": 0.096,
+      "step": 11310
+    },
+    {
+      "epoch": 2.2640000000000002,
+      "grad_norm": 0.002293473342433572,
+      "learning_rate": 2.7360000000000002e-05,
+      "loss": 0.001,
+      "step": 11320
+    },
+    {
+      "epoch": 2.266,
+      "grad_norm": 0.0028984423261135817,
+      "learning_rate": 2.734e-05,
+      "loss": 0.0014,
+      "step": 11330
+    },
+    {
+      "epoch": 2.268,
+      "grad_norm": 0.0026602321304380894,
+      "learning_rate": 2.7320000000000003e-05,
+      "loss": 0.0655,
+      "step": 11340
+    },
+    {
+      "epoch": 2.27,
+      "grad_norm": 9.377867698669434,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 0.0291,
+      "step": 11350
+    },
+    {
+      "epoch": 2.2720000000000002,
+      "grad_norm": 0.9569777846336365,
+      "learning_rate": 2.728e-05,
+      "loss": 0.0017,
+      "step": 11360
+    },
+    {
+      "epoch": 2.274,
+      "grad_norm": 0.004788199905306101,
+      "learning_rate": 2.7260000000000003e-05,
+      "loss": 0.0067,
+      "step": 11370
+    },
+    {
+      "epoch": 2.276,
+      "grad_norm": 3.850484609603882,
+      "learning_rate": 2.724e-05,
+      "loss": 0.0367,
+      "step": 11380
+    },
+    {
+      "epoch": 2.278,
+      "grad_norm": 0.0053862943314015865,
+      "learning_rate": 2.722e-05,
+      "loss": 0.0367,
+      "step": 11390
+    },
+    {
+      "epoch": 2.2800000000000002,
+      "grad_norm": 0.32168224453926086,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 0.0047,
+      "step": 11400
+    },
+    {
+      "epoch": 2.282,
+      "grad_norm": 0.007178562227636576,
+      "learning_rate": 2.718e-05,
+      "loss": 0.0009,
+      "step": 11410
+    },
+    {
+      "epoch": 2.284,
+      "grad_norm": 0.03468381240963936,
+      "learning_rate": 2.716e-05,
+      "loss": 0.0003,
+      "step": 11420
+    },
+    {
+      "epoch": 2.286,
+      "grad_norm": 0.06162317097187042,
+      "learning_rate": 2.7139999999999998e-05,
+      "loss": 0.0003,
+      "step": 11430
+    },
+    {
+      "epoch": 2.288,
+      "grad_norm": 0.0026774811558425426,
+      "learning_rate": 2.712e-05,
+      "loss": 0.0012,
+      "step": 11440
+    },
+    {
+      "epoch": 2.29,
+      "grad_norm": 0.004388803616166115,
+      "learning_rate": 2.7100000000000005e-05,
+      "loss": 0.0011,
+      "step": 11450
+    },
+    {
+      "epoch": 2.292,
+      "grad_norm": 0.007513342425227165,
+      "learning_rate": 2.7079999999999998e-05,
+      "loss": 0.0228,
+      "step": 11460
+    },
+    {
+      "epoch": 2.294,
+      "grad_norm": 0.0048149460926651955,
+      "learning_rate": 2.7060000000000002e-05,
+      "loss": 0.1007,
+      "step": 11470
+    },
+    {
+      "epoch": 2.296,
+      "grad_norm": 0.0016075849998742342,
+      "learning_rate": 2.704e-05,
+      "loss": 0.0416,
+      "step": 11480
+    },
+    {
+      "epoch": 2.298,
+      "grad_norm": 2.9490482807159424,
+      "learning_rate": 2.7020000000000002e-05,
+      "loss": 0.0033,
+      "step": 11490
+    },
+    {
+      "epoch": 2.3,
+      "grad_norm": 0.0028781858272850513,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 0.0007,
+      "step": 11500
+    },
+    {
+      "epoch": 2.302,
+      "grad_norm": 7.440994739532471,
+      "learning_rate": 2.698e-05,
+      "loss": 0.0077,
+      "step": 11510
+    },
+    {
+      "epoch": 2.304,
+      "grad_norm": 2.7259159088134766,
+      "learning_rate": 2.6960000000000003e-05,
+      "loss": 0.0058,
+      "step": 11520
+    },
+    {
+      "epoch": 2.306,
+      "grad_norm": 0.0006665542023256421,
+      "learning_rate": 2.694e-05,
+      "loss": 0.03,
+      "step": 11530
+    },
+    {
+      "epoch": 2.308,
+      "grad_norm": 3.5304527282714844,
+      "learning_rate": 2.692e-05,
+      "loss": 0.0573,
+      "step": 11540
+    },
+    {
+      "epoch": 2.31,
+      "grad_norm": 0.026721453294157982,
+      "learning_rate": 2.6900000000000003e-05,
+      "loss": 0.0142,
+      "step": 11550
+    },
+    {
+      "epoch": 2.312,
+      "grad_norm": 0.0015324908308684826,
+      "learning_rate": 2.688e-05,
+      "loss": 0.0249,
+      "step": 11560
+    },
+    {
+      "epoch": 2.314,
+      "grad_norm": 0.01611916720867157,
+      "learning_rate": 2.686e-05,
+      "loss": 0.0638,
+      "step": 11570
+    },
+    {
+      "epoch": 2.316,
+      "grad_norm": 0.010218789801001549,
+      "learning_rate": 2.6840000000000004e-05,
+      "loss": 0.0465,
+      "step": 11580
+    },
+    {
+      "epoch": 2.318,
+      "grad_norm": 0.01205415278673172,
+      "learning_rate": 2.682e-05,
+      "loss": 0.0564,
+      "step": 11590
+    },
+    {
+      "epoch": 2.32,
+      "grad_norm": 0.4571046829223633,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 0.0273,
+      "step": 11600
+    },
+    {
+      "epoch": 2.322,
+      "grad_norm": 0.06596440821886063,
+      "learning_rate": 2.678e-05,
+      "loss": 0.0008,
+      "step": 11610
+    },
+    {
+      "epoch": 2.324,
+      "grad_norm": 0.010993361473083496,
+      "learning_rate": 2.676e-05,
+      "loss": 0.0364,
+      "step": 11620
+    },
+    {
+      "epoch": 2.326,
+      "grad_norm": 0.012889869511127472,
+      "learning_rate": 2.6740000000000005e-05,
+      "loss": 0.0009,
+      "step": 11630
+    },
+    {
+      "epoch": 2.328,
+      "grad_norm": 0.04803529009222984,
+      "learning_rate": 2.672e-05,
+      "loss": 0.0017,
+      "step": 11640
+    },
+    {
+      "epoch": 2.33,
+      "grad_norm": 0.005940242204815149,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 0.0009,
+      "step": 11650
+    },
+    {
+      "epoch": 2.332,
+      "grad_norm": 0.05164859816431999,
+      "learning_rate": 2.668e-05,
+      "loss": 0.004,
+      "step": 11660
+    },
+    {
+      "epoch": 2.334,
+      "grad_norm": 0.004865141119807959,
+      "learning_rate": 2.6660000000000002e-05,
+      "loss": 0.0016,
+      "step": 11670
+    },
+    {
+      "epoch": 2.336,
+      "grad_norm": 0.0013962871162220836,
+      "learning_rate": 2.6640000000000002e-05,
+      "loss": 0.0012,
+      "step": 11680
+    },
+    {
+      "epoch": 2.338,
+      "grad_norm": 0.0495498850941658,
+      "learning_rate": 2.662e-05,
+      "loss": 0.0012,
+      "step": 11690
+    },
+    {
+      "epoch": 2.34,
+      "grad_norm": 0.002483365125954151,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 0.001,
+      "step": 11700
+    },
+    {
+      "epoch": 2.342,
+      "grad_norm": 0.007892883382737637,
+      "learning_rate": 2.658e-05,
+      "loss": 0.0019,
+      "step": 11710
+    },
+    {
+      "epoch": 2.344,
+      "grad_norm": 0.6501008868217468,
+      "learning_rate": 2.6560000000000003e-05,
+      "loss": 0.001,
+      "step": 11720
+    },
+    {
+      "epoch": 2.346,
+      "grad_norm": 0.004912431817501783,
+      "learning_rate": 2.6540000000000003e-05,
+      "loss": 0.0446,
+      "step": 11730
+    },
+    {
+      "epoch": 2.348,
+      "grad_norm": 0.0070532746613025665,
+      "learning_rate": 2.652e-05,
+      "loss": 0.0337,
+      "step": 11740
+    },
+    {
+      "epoch": 2.35,
+      "grad_norm": 0.0037770241033285856,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 0.0197,
+      "step": 11750
+    },
+    {
+      "epoch": 2.352,
+      "grad_norm": 0.09849817305803299,
+      "learning_rate": 2.648e-05,
+      "loss": 0.0279,
+      "step": 11760
+    },
+    {
+      "epoch": 2.354,
+      "grad_norm": 0.0006417520926333964,
+      "learning_rate": 2.646e-05,
+      "loss": 0.001,
+      "step": 11770
+    },
+    {
+      "epoch": 2.356,
+      "grad_norm": 0.00028482265770435333,
+      "learning_rate": 2.6440000000000004e-05,
+      "loss": 0.0012,
+      "step": 11780
+    },
+    {
+      "epoch": 2.358,
+      "grad_norm": 0.03328564390540123,
+      "learning_rate": 2.642e-05,
+      "loss": 0.1101,
+      "step": 11790
+    },
+    {
+      "epoch": 2.36,
+      "grad_norm": 0.007043204270303249,
+      "learning_rate": 2.64e-05,
+      "loss": 0.0003,
+      "step": 11800
+    },
+    {
+      "epoch": 2.362,
+      "grad_norm": 0.016940118744969368,
+      "learning_rate": 2.6379999999999998e-05,
+      "loss": 0.0054,
+      "step": 11810
+    },
+    {
+      "epoch": 2.364,
+      "grad_norm": 0.008778809569776058,
+      "learning_rate": 2.6360000000000002e-05,
+      "loss": 0.0008,
+      "step": 11820
+    },
+    {
+      "epoch": 2.366,
+      "grad_norm": 0.005758294835686684,
+      "learning_rate": 2.6340000000000002e-05,
+      "loss": 0.0003,
+      "step": 11830
+    },
+    {
+      "epoch": 2.368,
+      "grad_norm": 0.026752840727567673,
+      "learning_rate": 2.632e-05,
+      "loss": 0.0101,
+      "step": 11840
+    },
+    {
+      "epoch": 2.37,
+      "grad_norm": 0.009118887595832348,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 0.0005,
+      "step": 11850
+    },
+    {
+      "epoch": 2.372,
+      "grad_norm": 0.0074037835001945496,
+      "learning_rate": 2.628e-05,
+      "loss": 0.0007,
+      "step": 11860
+    },
+    {
+      "epoch": 2.374,
+      "grad_norm": 0.005524361506104469,
+      "learning_rate": 2.6260000000000003e-05,
+      "loss": 0.0003,
+      "step": 11870
+    },
+    {
+      "epoch": 2.376,
+      "grad_norm": 0.0064158011227846146,
+      "learning_rate": 2.6240000000000003e-05,
+      "loss": 0.0239,
+      "step": 11880
+    },
+    {
+      "epoch": 2.378,
+      "grad_norm": 0.009333696216344833,
+      "learning_rate": 2.622e-05,
+      "loss": 0.0009,
+      "step": 11890
+    },
+    {
+      "epoch": 2.38,
+      "grad_norm": 0.0013852640986442566,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 0.0047,
+      "step": 11900
+    },
+    {
+      "epoch": 2.382,
+      "grad_norm": 0.002884730463847518,
+      "learning_rate": 2.618e-05,
+      "loss": 0.0008,
+      "step": 11910
+    },
+    {
+      "epoch": 2.384,
+      "grad_norm": 0.02573120780289173,
+      "learning_rate": 2.616e-05,
+      "loss": 0.0798,
+      "step": 11920
+    },
+    {
+      "epoch": 2.386,
+      "grad_norm": 0.004389532376080751,
+      "learning_rate": 2.6140000000000004e-05,
+      "loss": 0.0109,
+      "step": 11930
+    },
+    {
+      "epoch": 2.388,
+      "grad_norm": 0.017901601269841194,
+      "learning_rate": 2.612e-05,
+      "loss": 0.0004,
+      "step": 11940
+    },
+    {
+      "epoch": 2.39,
+      "grad_norm": 2.5958914756774902,
+      "learning_rate": 2.61e-05,
+      "loss": 0.048,
+      "step": 11950
+    },
+    {
+      "epoch": 2.392,
+      "grad_norm": 0.0030245562084019184,
+      "learning_rate": 2.6079999999999998e-05,
+      "loss": 0.0384,
+      "step": 11960
+    },
+    {
+      "epoch": 2.394,
+      "grad_norm": 0.001944447518326342,
+      "learning_rate": 2.606e-05,
+      "loss": 0.014,
+      "step": 11970
+    },
+    {
+      "epoch": 2.396,
+      "grad_norm": 0.013363710604608059,
+      "learning_rate": 2.6040000000000005e-05,
+      "loss": 0.0268,
+      "step": 11980
+    },
+    {
+      "epoch": 2.398,
+      "grad_norm": 0.016894085332751274,
+      "learning_rate": 2.602e-05,
+      "loss": 0.0021,
+      "step": 11990
+    },
+    {
+      "epoch": 2.4,
+      "grad_norm": 0.008409597910940647,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 0.0023,
+      "step": 12000
+    },
+    {
+      "epoch": 2.402,
+      "grad_norm": 0.030583485960960388,
+      "learning_rate": 2.598e-05,
+      "loss": 0.0953,
+      "step": 12010
+    },
+    {
+      "epoch": 2.404,
+      "grad_norm": 0.01641802117228508,
+      "learning_rate": 2.5960000000000002e-05,
+      "loss": 0.0008,
+      "step": 12020
+    },
+    {
+      "epoch": 2.406,
+      "grad_norm": 0.0023062105756253004,
+      "learning_rate": 2.5940000000000002e-05,
+      "loss": 0.0087,
+      "step": 12030
+    },
+    {
+      "epoch": 2.408,
+      "grad_norm": 0.006509549915790558,
+      "learning_rate": 2.592e-05,
+      "loss": 0.0081,
+      "step": 12040
+    },
+    {
+      "epoch": 2.41,
+      "grad_norm": 0.0068439035676419735,
+      "learning_rate": 2.5900000000000003e-05,
+      "loss": 0.0253,
+      "step": 12050
+    },
+    {
+      "epoch": 2.412,
+      "grad_norm": 0.005552592687308788,
+      "learning_rate": 2.588e-05,
+      "loss": 0.0326,
+      "step": 12060
+    },
+    {
+      "epoch": 2.414,
+      "grad_norm": 0.011574423871934414,
+      "learning_rate": 2.586e-05,
+      "loss": 0.0318,
+      "step": 12070
+    },
+    {
+      "epoch": 2.416,
+      "grad_norm": 0.0022241093683987856,
+      "learning_rate": 2.5840000000000003e-05,
+      "loss": 0.0006,
+      "step": 12080
+    },
+    {
+      "epoch": 2.418,
+      "grad_norm": 0.00799054279923439,
+      "learning_rate": 2.582e-05,
+      "loss": 0.0007,
+      "step": 12090
+    },
+    {
+      "epoch": 2.42,
+      "grad_norm": 0.09423800557851791,
+      "learning_rate": 2.58e-05,
+      "loss": 0.0013,
+      "step": 12100
+    },
+    {
+      "epoch": 2.422,
+      "grad_norm": 0.005380244459956884,
+      "learning_rate": 2.5779999999999997e-05,
+      "loss": 0.0338,
+      "step": 12110
+    },
+    {
+      "epoch": 2.424,
+      "grad_norm": 0.025284456089138985,
+      "learning_rate": 2.576e-05,
+      "loss": 0.0023,
+      "step": 12120
+    },
+    {
+      "epoch": 2.426,
+      "grad_norm": 0.005192068871110678,
+      "learning_rate": 2.5740000000000004e-05,
+      "loss": 0.0063,
+      "step": 12130
+    },
+    {
+      "epoch": 2.428,
+      "grad_norm": 0.006014976650476456,
+      "learning_rate": 2.572e-05,
+      "loss": 0.0006,
+      "step": 12140
+    },
+    {
+      "epoch": 2.43,
+      "grad_norm": 0.0024557739961892366,
+      "learning_rate": 2.57e-05,
+      "loss": 0.0164,
+      "step": 12150
+    },
+    {
+      "epoch": 2.432,
+      "grad_norm": 0.03051261603832245,
+      "learning_rate": 2.5679999999999998e-05,
+      "loss": 0.0009,
+      "step": 12160
+    },
+    {
+      "epoch": 2.434,
+      "grad_norm": 0.00140279158949852,
+      "learning_rate": 2.566e-05,
+      "loss": 0.0005,
+      "step": 12170
+    },
+    {
+      "epoch": 2.436,
+      "grad_norm": 5.22982931137085,
+      "learning_rate": 2.5640000000000002e-05,
+      "loss": 0.0207,
+      "step": 12180
+    },
+    {
+      "epoch": 2.438,
+      "grad_norm": 0.003986484836786985,
+      "learning_rate": 2.562e-05,
+      "loss": 0.0008,
+      "step": 12190
+    },
+    {
+      "epoch": 2.44,
+      "grad_norm": 0.0018658132757991552,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 0.0142,
+      "step": 12200
+    },
+    {
+      "epoch": 2.442,
+      "grad_norm": 0.905129611492157,
+      "learning_rate": 2.5580000000000002e-05,
+      "loss": 0.0026,
+      "step": 12210
+    },
+    {
+      "epoch": 2.444,
+      "grad_norm": 0.0045546796172857285,
+      "learning_rate": 2.556e-05,
+      "loss": 0.0922,
+      "step": 12220
+    },
+    {
+      "epoch": 2.446,
+      "grad_norm": 0.02099277451634407,
+      "learning_rate": 2.5540000000000003e-05,
+      "loss": 0.0006,
+      "step": 12230
+    },
+    {
+      "epoch": 2.448,
+      "grad_norm": 0.01280425488948822,
+      "learning_rate": 2.552e-05,
+      "loss": 0.0101,
+      "step": 12240
+    },
+    {
+      "epoch": 2.45,
+      "grad_norm": 5.127621650695801,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 0.0715,
+      "step": 12250
+    },
+    {
+      "epoch": 2.452,
+      "grad_norm": 0.006172411143779755,
+      "learning_rate": 2.5480000000000003e-05,
+      "loss": 0.0008,
+      "step": 12260
+    },
+    {
+      "epoch": 2.454,
+      "grad_norm": 0.0678882971405983,
+      "learning_rate": 2.546e-05,
+      "loss": 0.0009,
+      "step": 12270
+    },
+    {
+      "epoch": 2.456,
+      "grad_norm": 0.003381960093975067,
+      "learning_rate": 2.5440000000000004e-05,
+      "loss": 0.0173,
+      "step": 12280
+    },
+    {
+      "epoch": 2.458,
+      "grad_norm": 0.005731347016990185,
+      "learning_rate": 2.542e-05,
+      "loss": 0.0114,
+      "step": 12290
+    },
+    {
+      "epoch": 2.46,
+      "grad_norm": 0.058216921985149384,
+      "learning_rate": 2.54e-05,
+      "loss": 0.0034,
+      "step": 12300
+    },
+    {
+      "epoch": 2.462,
+      "grad_norm": 0.003912481013685465,
+      "learning_rate": 2.5380000000000004e-05,
+      "loss": 0.0001,
+      "step": 12310
+    },
+    {
+      "epoch": 2.464,
+      "grad_norm": 1.2991969585418701,
+      "learning_rate": 2.536e-05,
+      "loss": 0.0021,
+      "step": 12320
+    },
+    {
+      "epoch": 2.466,
+      "grad_norm": 0.008072658441960812,
+      "learning_rate": 2.534e-05,
+      "loss": 0.0082,
+      "step": 12330
+    },
+    {
+      "epoch": 2.468,
+      "grad_norm": 0.1591532677412033,
+      "learning_rate": 2.5319999999999998e-05,
+      "loss": 0.0006,
+      "step": 12340
+    },
+    {
+      "epoch": 2.4699999999999998,
+      "grad_norm": 0.002749226987361908,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 0.0003,
+      "step": 12350
+    },
+    {
+      "epoch": 2.472,
+      "grad_norm": 0.013632843270897865,
+      "learning_rate": 2.5280000000000005e-05,
+      "loss": 0.0491,
+      "step": 12360
+    },
+    {
+      "epoch": 2.474,
+      "grad_norm": 0.002205546712502837,
+      "learning_rate": 2.526e-05,
+      "loss": 0.0003,
+      "step": 12370
+    },
+    {
+      "epoch": 2.476,
+      "grad_norm": 0.0033999320585280657,
+      "learning_rate": 2.5240000000000002e-05,
+      "loss": 0.0119,
+      "step": 12380
+    },
+    {
+      "epoch": 2.4779999999999998,
+      "grad_norm": 0.004555273801088333,
+      "learning_rate": 2.522e-05,
+      "loss": 0.0013,
+      "step": 12390
+    },
+    {
+      "epoch": 2.48,
+      "grad_norm": 0.02334682084619999,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 0.0181,
+      "step": 12400
+    },
+    {
+      "epoch": 2.482,
+      "grad_norm": 0.0025895722210407257,
+      "learning_rate": 2.5180000000000003e-05,
+      "loss": 0.0042,
+      "step": 12410
+    },
+    {
+      "epoch": 2.484,
+      "grad_norm": 0.05390416085720062,
+      "learning_rate": 2.516e-05,
+      "loss": 0.0011,
+      "step": 12420
+    },
+    {
+      "epoch": 2.4859999999999998,
+      "grad_norm": 0.0009085657657124102,
+      "learning_rate": 2.5140000000000003e-05,
+      "loss": 0.0284,
+      "step": 12430
+    },
+    {
+      "epoch": 2.488,
+      "grad_norm": 0.26577678322792053,
+      "learning_rate": 2.512e-05,
+      "loss": 0.001,
+      "step": 12440
+    },
+    {
+      "epoch": 2.49,
+      "grad_norm": 0.0048172916285693645,
+      "learning_rate": 2.51e-05,
+      "loss": 0.0004,
+      "step": 12450
+    },
+    {
+      "epoch": 2.492,
+      "grad_norm": 0.009927072562277317,
+      "learning_rate": 2.5080000000000004e-05,
+      "loss": 0.0482,
+      "step": 12460
+    },
+    {
+      "epoch": 2.4939999999999998,
+      "grad_norm": 0.0019616412464529276,
+      "learning_rate": 2.506e-05,
+      "loss": 0.0765,
+      "step": 12470
+    },
+    {
+      "epoch": 2.496,
+      "grad_norm": 0.0017744253855198622,
+      "learning_rate": 2.504e-05,
+      "loss": 0.0003,
+      "step": 12480
+    },
+    {
+      "epoch": 2.498,
+      "grad_norm": 0.012268969789147377,
+      "learning_rate": 2.5019999999999998e-05,
+      "loss": 0.0175,
+      "step": 12490
+    },
+    {
+      "epoch": 2.5,
+      "grad_norm": 5.593498229980469,
+      "learning_rate": 2.5e-05,
+      "loss": 0.0125,
+      "step": 12500
+    },
+    {
+      "epoch": 2.502,
+      "grad_norm": 0.039957594126462936,
+      "learning_rate": 2.498e-05,
+      "loss": 0.0034,
+      "step": 12510
+    },
+    {
+      "epoch": 2.504,
+      "grad_norm": 0.00404326943680644,
+      "learning_rate": 2.496e-05,
+      "loss": 0.0004,
+      "step": 12520
+    },
+    {
+      "epoch": 2.5060000000000002,
+      "grad_norm": 0.0023503380361944437,
+      "learning_rate": 2.4940000000000002e-05,
+      "loss": 0.0075,
+      "step": 12530
+    },
+    {
+      "epoch": 2.508,
+      "grad_norm": 0.005890808999538422,
+      "learning_rate": 2.4920000000000002e-05,
+      "loss": 0.0006,
+      "step": 12540
+    },
+    {
+      "epoch": 2.51,
+      "grad_norm": 0.0037465377245098352,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 0.0015,
+      "step": 12550
+    },
+    {
+      "epoch": 2.512,
+      "grad_norm": 0.001902339281514287,
+      "learning_rate": 2.488e-05,
+      "loss": 0.0008,
+      "step": 12560
+    },
+    {
+      "epoch": 2.5140000000000002,
+      "grad_norm": 0.035102877765893936,
+      "learning_rate": 2.486e-05,
+      "loss": 0.0013,
+      "step": 12570
+    },
+    {
+      "epoch": 2.516,
+      "grad_norm": 0.00217231223359704,
+      "learning_rate": 2.4840000000000003e-05,
+      "loss": 0.0913,
+      "step": 12580
+    },
+    {
+      "epoch": 2.518,
+      "grad_norm": 0.00795662309974432,
+      "learning_rate": 2.4820000000000003e-05,
+      "loss": 0.0363,
+      "step": 12590
+    },
+    {
+      "epoch": 2.52,
+      "grad_norm": 0.007800019811838865,
+      "learning_rate": 2.48e-05,
+      "loss": 0.0185,
+      "step": 12600
+    },
+    {
+      "epoch": 2.5220000000000002,
+      "grad_norm": 0.005222728475928307,
+      "learning_rate": 2.478e-05,
+      "loss": 0.0047,
+      "step": 12610
+    },
+    {
+      "epoch": 2.524,
+      "grad_norm": 0.004378939513117075,
+      "learning_rate": 2.476e-05,
+      "loss": 0.0004,
+      "step": 12620
+    },
+    {
+      "epoch": 2.526,
+      "grad_norm": 0.015952318906784058,
+      "learning_rate": 2.4740000000000004e-05,
+      "loss": 0.0019,
+      "step": 12630
+    },
+    {
+      "epoch": 2.528,
+      "grad_norm": 0.001999328611418605,
+      "learning_rate": 2.472e-05,
+      "loss": 0.0324,
+      "step": 12640
+    },
+    {
+      "epoch": 2.5300000000000002,
+      "grad_norm": 0.03502897545695305,
+      "learning_rate": 2.47e-05,
+      "loss": 0.039,
+      "step": 12650
+    },
+    {
+      "epoch": 2.532,
+      "grad_norm": 2.233900785446167,
+      "learning_rate": 2.468e-05,
+      "loss": 0.0422,
+      "step": 12660
+    },
+    {
+      "epoch": 2.534,
+      "grad_norm": 0.02098439261317253,
+      "learning_rate": 2.466e-05,
+      "loss": 0.0364,
+      "step": 12670
+    },
+    {
+      "epoch": 2.536,
+      "grad_norm": 0.005417154636234045,
+      "learning_rate": 2.464e-05,
+      "loss": 0.0863,
+      "step": 12680
+    },
+    {
+      "epoch": 2.5380000000000003,
+      "grad_norm": 0.0077430736273527145,
+      "learning_rate": 2.462e-05,
+      "loss": 0.008,
+      "step": 12690
+    },
+    {
+      "epoch": 2.54,
+      "grad_norm": 0.00703767966479063,
+      "learning_rate": 2.46e-05,
+      "loss": 0.0017,
+      "step": 12700
+    },
+    {
+      "epoch": 2.542,
+      "grad_norm": 6.646230220794678,
+      "learning_rate": 2.4580000000000002e-05,
+      "loss": 0.0383,
+      "step": 12710
+    },
+    {
+      "epoch": 2.544,
+      "grad_norm": 0.011931142769753933,
+      "learning_rate": 2.4560000000000002e-05,
+      "loss": 0.0024,
+      "step": 12720
+    },
+    {
+      "epoch": 2.5460000000000003,
+      "grad_norm": 0.006545823533087969,
+      "learning_rate": 2.4540000000000002e-05,
+      "loss": 0.051,
+      "step": 12730
+    },
+    {
+      "epoch": 2.548,
+      "grad_norm": 0.004485706333070993,
+      "learning_rate": 2.4520000000000002e-05,
+      "loss": 0.0014,
+      "step": 12740
+    },
+    {
+      "epoch": 2.55,
+      "grad_norm": 6.2546892166137695,
+      "learning_rate": 2.45e-05,
+      "loss": 0.0658,
+      "step": 12750
+    },
+    {
+      "epoch": 2.552,
+      "grad_norm": 0.024582767859101295,
+      "learning_rate": 2.448e-05,
+      "loss": 0.0048,
+      "step": 12760
+    },
+    {
+      "epoch": 2.5540000000000003,
+      "grad_norm": 0.009772568009793758,
+      "learning_rate": 2.4460000000000003e-05,
+      "loss": 0.0011,
+      "step": 12770
+    },
+    {
+      "epoch": 2.556,
+      "grad_norm": 0.3461325168609619,
+      "learning_rate": 2.4440000000000003e-05,
+      "loss": 0.0012,
+      "step": 12780
+    },
+    {
+      "epoch": 2.558,
+      "grad_norm": 0.004639799706637859,
+      "learning_rate": 2.442e-05,
+      "loss": 0.0037,
+      "step": 12790
+    },
+    {
+      "epoch": 2.56,
+      "grad_norm": 0.018950438126921654,
+      "learning_rate": 2.44e-05,
+      "loss": 0.0105,
+      "step": 12800
+    },
+    {
+      "epoch": 2.5620000000000003,
+      "grad_norm": 0.010657204315066338,
+      "learning_rate": 2.438e-05,
+      "loss": 0.0208,
+      "step": 12810
+    },
+    {
+      "epoch": 2.564,
+      "grad_norm": 0.0076634702272713184,
+      "learning_rate": 2.4360000000000004e-05,
+      "loss": 0.0008,
+      "step": 12820
+    },
+    {
+      "epoch": 2.566,
+      "grad_norm": 0.004654941149055958,
+      "learning_rate": 2.434e-05,
+      "loss": 0.0023,
+      "step": 12830
+    },
+    {
+      "epoch": 2.568,
+      "grad_norm": 0.010933748446404934,
+      "learning_rate": 2.432e-05,
+      "loss": 0.0039,
+      "step": 12840
+    },
+    {
+      "epoch": 2.57,
+      "grad_norm": 9.478780746459961,
+      "learning_rate": 2.43e-05,
+      "loss": 0.0268,
+      "step": 12850
+    },
+    {
+      "epoch": 2.572,
+      "grad_norm": 4.93724250793457,
+      "learning_rate": 2.428e-05,
+      "loss": 0.0525,
+      "step": 12860
+    },
+    {
+      "epoch": 2.574,
+      "grad_norm": 0.006643955130130053,
+      "learning_rate": 2.426e-05,
+      "loss": 0.0006,
+      "step": 12870
+    },
+    {
+      "epoch": 2.576,
+      "grad_norm": 0.06824769824743271,
+      "learning_rate": 2.4240000000000002e-05,
+      "loss": 0.058,
+      "step": 12880
+    },
+    {
+      "epoch": 2.578,
+      "grad_norm": 0.003721135901287198,
+      "learning_rate": 2.4220000000000002e-05,
+      "loss": 0.0022,
+      "step": 12890
+    },
+    {
+      "epoch": 2.58,
+      "grad_norm": 0.076459601521492,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 0.0114,
+      "step": 12900
+    },
+    {
+      "epoch": 2.582,
+      "grad_norm": 0.0034186700358986855,
+      "learning_rate": 2.418e-05,
+      "loss": 0.0252,
+      "step": 12910
+    },
+    {
+      "epoch": 2.584,
+      "grad_norm": 0.019875500351190567,
+      "learning_rate": 2.4160000000000002e-05,
+      "loss": 0.0146,
+      "step": 12920
+    },
+    {
+      "epoch": 2.586,
+      "grad_norm": 0.004298020154237747,
+      "learning_rate": 2.4140000000000003e-05,
+      "loss": 0.0044,
+      "step": 12930
+    },
+    {
+      "epoch": 2.588,
+      "grad_norm": 0.005100682843476534,
+      "learning_rate": 2.412e-05,
+      "loss": 0.0016,
+      "step": 12940
+    },
+    {
+      "epoch": 2.59,
+      "grad_norm": 14.05281925201416,
+      "learning_rate": 2.41e-05,
+      "loss": 0.0197,
+      "step": 12950
+    },
+    {
+      "epoch": 2.592,
+      "grad_norm": 0.5248048305511475,
+      "learning_rate": 2.408e-05,
+      "loss": 0.001,
+      "step": 12960
+    },
+    {
+      "epoch": 2.594,
+      "grad_norm": 0.0029409376438707113,
+      "learning_rate": 2.4060000000000003e-05,
+      "loss": 0.0398,
+      "step": 12970
+    },
+    {
+      "epoch": 2.596,
+      "grad_norm": 0.005220264196395874,
+      "learning_rate": 2.404e-05,
+      "loss": 0.0123,
+      "step": 12980
+    },
+    {
+      "epoch": 2.598,
+      "grad_norm": 0.004500200040638447,
+      "learning_rate": 2.402e-05,
+      "loss": 0.0084,
+      "step": 12990
+    },
+    {
+      "epoch": 2.6,
+      "grad_norm": 0.016876282170414925,
+      "learning_rate": 2.4e-05,
+      "loss": 0.0004,
+      "step": 13000
+    },
+    {
+      "epoch": 2.602,
+      "grad_norm": 0.00897781178355217,
+      "learning_rate": 2.398e-05,
+      "loss": 0.0004,
+      "step": 13010
+    },
+    {
+      "epoch": 2.604,
+      "grad_norm": 0.8232995867729187,
+      "learning_rate": 2.396e-05,
+      "loss": 0.0175,
+      "step": 13020
+    },
+    {
+      "epoch": 2.606,
+      "grad_norm": 2.4105169773101807,
+      "learning_rate": 2.394e-05,
+      "loss": 0.0417,
+      "step": 13030
+    },
+    {
+      "epoch": 2.608,
+      "grad_norm": 0.003810339141637087,
+      "learning_rate": 2.392e-05,
+      "loss": 0.0004,
+      "step": 13040
+    },
+    {
+      "epoch": 2.61,
+      "grad_norm": 0.040436211973428726,
+      "learning_rate": 2.39e-05,
+      "loss": 0.0008,
+      "step": 13050
+    },
+    {
+      "epoch": 2.612,
+      "grad_norm": 0.01878192462027073,
+      "learning_rate": 2.3880000000000002e-05,
+      "loss": 0.075,
+      "step": 13060
+    },
+    {
+      "epoch": 2.614,
+      "grad_norm": 0.005005359649658203,
+      "learning_rate": 2.3860000000000002e-05,
+      "loss": 0.0587,
+      "step": 13070
+    },
+    {
+      "epoch": 2.616,
+      "grad_norm": 0.004173360764980316,
+      "learning_rate": 2.3840000000000002e-05,
+      "loss": 0.0152,
+      "step": 13080
+    },
+    {
+      "epoch": 2.618,
+      "grad_norm": 0.003978619817644358,
+      "learning_rate": 2.3820000000000002e-05,
+      "loss": 0.0693,
+      "step": 13090
+    },
+    {
+      "epoch": 2.62,
+      "grad_norm": 0.014284418895840645,
+      "learning_rate": 2.38e-05,
+      "loss": 0.0008,
+      "step": 13100
+    },
+    {
+      "epoch": 2.622,
+      "grad_norm": 0.012868880294263363,
+      "learning_rate": 2.3780000000000003e-05,
+      "loss": 0.0027,
+      "step": 13110
+    },
+    {
+      "epoch": 2.624,
+      "grad_norm": 0.021820498630404472,
+      "learning_rate": 2.3760000000000003e-05,
+      "loss": 0.0013,
+      "step": 13120
+    },
+    {
+      "epoch": 2.626,
+      "grad_norm": 0.020173553377389908,
+      "learning_rate": 2.374e-05,
+      "loss": 0.0088,
+      "step": 13130
+    },
+    {
+      "epoch": 2.628,
+      "grad_norm": 0.023448575288057327,
+      "learning_rate": 2.372e-05,
+      "loss": 0.0008,
+      "step": 13140
+    },
+    {
+      "epoch": 2.63,
+      "grad_norm": 0.012240973301231861,
+      "learning_rate": 2.37e-05,
+      "loss": 0.0108,
+      "step": 13150
+    },
+    {
+      "epoch": 2.632,
+      "grad_norm": 0.36354461312294006,
+      "learning_rate": 2.3680000000000004e-05,
+      "loss": 0.0447,
+      "step": 13160
+    },
+    {
+      "epoch": 2.634,
+      "grad_norm": 0.6124907732009888,
+      "learning_rate": 2.366e-05,
+      "loss": 0.0096,
+      "step": 13170
+    },
+    {
+      "epoch": 2.636,
+      "grad_norm": 0.007227941881865263,
+      "learning_rate": 2.364e-05,
+      "loss": 0.0014,
+      "step": 13180
+    },
+    {
+      "epoch": 2.638,
+      "grad_norm": 0.0026091760955750942,
+      "learning_rate": 2.362e-05,
+      "loss": 0.0097,
+      "step": 13190
+    },
+    {
+      "epoch": 2.64,
+      "grad_norm": 0.004290744196623564,
+      "learning_rate": 2.36e-05,
+      "loss": 0.0008,
+      "step": 13200
+    },
+    {
+      "epoch": 2.642,
+      "grad_norm": 0.003058998379856348,
+      "learning_rate": 2.358e-05,
+      "loss": 0.0005,
+      "step": 13210
+    },
+    {
+      "epoch": 2.644,
+      "grad_norm": 0.00327027915045619,
+      "learning_rate": 2.356e-05,
+      "loss": 0.0437,
+      "step": 13220
+    },
+    {
+      "epoch": 2.646,
+      "grad_norm": 0.02532695233821869,
+      "learning_rate": 2.354e-05,
+      "loss": 0.0613,
+      "step": 13230
+    },
+    {
+      "epoch": 2.648,
+      "grad_norm": 2.3604378700256348,
+      "learning_rate": 2.3520000000000002e-05,
+      "loss": 0.0556,
+      "step": 13240
+    },
+    {
+      "epoch": 2.65,
+      "grad_norm": 8.695077896118164,
+      "learning_rate": 2.35e-05,
+      "loss": 0.0323,
+      "step": 13250
+    },
+    {
+      "epoch": 2.652,
+      "grad_norm": 0.009923191741108894,
+      "learning_rate": 2.3480000000000002e-05,
+      "loss": 0.001,
+      "step": 13260
+    },
+    {
+      "epoch": 2.654,
+      "grad_norm": 0.029513264074921608,
+      "learning_rate": 2.3460000000000002e-05,
+      "loss": 0.0006,
+      "step": 13270
+    },
+    {
+      "epoch": 2.656,
+      "grad_norm": 3.3521392345428467,
+      "learning_rate": 2.344e-05,
+      "loss": 0.0318,
+      "step": 13280
+    },
+    {
+      "epoch": 2.658,
+      "grad_norm": 0.2049526572227478,
+      "learning_rate": 2.342e-05,
+      "loss": 0.0025,
+      "step": 13290
+    },
+    {
+      "epoch": 2.66,
+      "grad_norm": 0.007929221726953983,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 0.0016,
+      "step": 13300
+    },
+    {
+      "epoch": 2.662,
+      "grad_norm": 0.002996771829202771,
+      "learning_rate": 2.3380000000000003e-05,
+      "loss": 0.0028,
+      "step": 13310
+    },
+    {
+      "epoch": 2.664,
+      "grad_norm": 0.007703165523707867,
+      "learning_rate": 2.336e-05,
+      "loss": 0.0006,
+      "step": 13320
+    },
+    {
+      "epoch": 2.666,
+      "grad_norm": 0.003229381749406457,
+      "learning_rate": 2.334e-05,
+      "loss": 0.002,
+      "step": 13330
+    },
+    {
+      "epoch": 2.668,
+      "grad_norm": 0.014454679563641548,
+      "learning_rate": 2.332e-05,
+      "loss": 0.0713,
+      "step": 13340
+    },
+    {
+      "epoch": 2.67,
+      "grad_norm": 0.008571525104343891,
+      "learning_rate": 2.3300000000000004e-05,
+      "loss": 0.0019,
+      "step": 13350
+    },
+    {
+      "epoch": 2.672,
+      "grad_norm": 0.006012748461216688,
+      "learning_rate": 2.328e-05,
+      "loss": 0.0296,
+      "step": 13360
+    },
+    {
+      "epoch": 2.674,
+      "grad_norm": 0.004160825163125992,
+      "learning_rate": 2.326e-05,
+      "loss": 0.0336,
+      "step": 13370
+    },
+    {
+      "epoch": 2.676,
+      "grad_norm": 0.031649768352508545,
+      "learning_rate": 2.324e-05,
+      "loss": 0.019,
+      "step": 13380
+    },
+    {
+      "epoch": 2.678,
+      "grad_norm": 0.019651921465992928,
+      "learning_rate": 2.322e-05,
+      "loss": 0.0407,
+      "step": 13390
+    },
+    {
+      "epoch": 2.68,
+      "grad_norm": 0.028730370104312897,
+      "learning_rate": 2.32e-05,
+      "loss": 0.0735,
+      "step": 13400
+    },
+    {
+      "epoch": 2.682,
+      "grad_norm": 0.06920024007558823,
+      "learning_rate": 2.318e-05,
+      "loss": 0.0044,
+      "step": 13410
+    },
+    {
+      "epoch": 2.684,
+      "grad_norm": 0.03945730999112129,
+      "learning_rate": 2.3160000000000002e-05,
+      "loss": 0.0004,
+      "step": 13420
+    },
+    {
+      "epoch": 2.686,
+      "grad_norm": 0.01288557332009077,
+      "learning_rate": 2.3140000000000002e-05,
+      "loss": 0.0026,
+      "step": 13430
+    },
+    {
+      "epoch": 2.6879999999999997,
+      "grad_norm": 0.5390536189079285,
+      "learning_rate": 2.312e-05,
+      "loss": 0.0332,
+      "step": 13440
+    },
+    {
+      "epoch": 2.69,
+      "grad_norm": 0.006479345262050629,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 0.0005,
+      "step": 13450
+    },
+    {
+      "epoch": 2.692,
+      "grad_norm": 0.008790848776698112,
+      "learning_rate": 2.3080000000000003e-05,
+      "loss": 0.0004,
+      "step": 13460
+    },
+    {
+      "epoch": 2.694,
+      "grad_norm": 0.004521473310887814,
+      "learning_rate": 2.306e-05,
+      "loss": 0.004,
+      "step": 13470
+    },
+    {
+      "epoch": 2.6959999999999997,
+      "grad_norm": 6.016846179962158,
+      "learning_rate": 2.304e-05,
+      "loss": 0.0207,
+      "step": 13480
+    },
+    {
+      "epoch": 2.698,
+      "grad_norm": 0.016385849565267563,
+      "learning_rate": 2.302e-05,
+      "loss": 0.0006,
+      "step": 13490
+    },
+    {
+      "epoch": 2.7,
+      "grad_norm": 0.0029000595677644014,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 0.0246,
+      "step": 13500
+    },
+    {
+      "epoch": 2.702,
+      "grad_norm": 0.017534567043185234,
+      "learning_rate": 2.298e-05,
+      "loss": 0.0008,
+      "step": 13510
+    },
+    {
+      "epoch": 2.7039999999999997,
+      "grad_norm": 0.008091018535196781,
+      "learning_rate": 2.296e-05,
+      "loss": 0.0006,
+      "step": 13520
+    },
+    {
+      "epoch": 2.706,
+      "grad_norm": 0.008187901228666306,
+      "learning_rate": 2.294e-05,
+      "loss": 0.0005,
+      "step": 13530
+    },
+    {
+      "epoch": 2.708,
+      "grad_norm": 0.004407752305269241,
+      "learning_rate": 2.292e-05,
+      "loss": 0.0228,
+      "step": 13540
+    },
+    {
+      "epoch": 2.71,
+      "grad_norm": 0.03731589391827583,
+      "learning_rate": 2.29e-05,
+      "loss": 0.0063,
+      "step": 13550
+    },
+    {
+      "epoch": 2.7119999999999997,
+      "grad_norm": 0.005584000609815121,
+      "learning_rate": 2.288e-05,
+      "loss": 0.0008,
+      "step": 13560
+    },
+    {
+      "epoch": 2.714,
+      "grad_norm": 0.03260991722345352,
+      "learning_rate": 2.286e-05,
+      "loss": 0.0006,
+      "step": 13570
+    },
+    {
+      "epoch": 2.716,
+      "grad_norm": 0.006006445270031691,
+      "learning_rate": 2.284e-05,
+      "loss": 0.0337,
+      "step": 13580
+    },
+    {
+      "epoch": 2.718,
+      "grad_norm": 0.003870969172567129,
+      "learning_rate": 2.282e-05,
+      "loss": 0.0005,
+      "step": 13590
+    },
+    {
+      "epoch": 2.7199999999999998,
+      "grad_norm": 0.013978478498756886,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 0.001,
+      "step": 13600
+    },
+    {
+      "epoch": 2.722,
+      "grad_norm": 0.003443440655246377,
+      "learning_rate": 2.2780000000000002e-05,
+      "loss": 0.0763,
+      "step": 13610
+    },
+    {
+      "epoch": 2.724,
+      "grad_norm": 0.0038477852940559387,
+      "learning_rate": 2.2760000000000002e-05,
+      "loss": 0.0007,
+      "step": 13620
+    },
+    {
+      "epoch": 2.726,
+      "grad_norm": 0.002752868691459298,
+      "learning_rate": 2.274e-05,
+      "loss": 0.0466,
+      "step": 13630
+    },
+    {
+      "epoch": 2.7279999999999998,
+      "grad_norm": 0.010040170513093472,
+      "learning_rate": 2.2720000000000003e-05,
+      "loss": 0.0022,
+      "step": 13640
+    },
+    {
+      "epoch": 2.73,
+      "grad_norm": 2.712369203567505,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 0.0411,
+      "step": 13650
+    },
+    {
+      "epoch": 2.732,
+      "grad_norm": 0.018802713602781296,
+      "learning_rate": 2.268e-05,
+      "loss": 0.0206,
+      "step": 13660
+    },
+    {
+      "epoch": 2.734,
+      "grad_norm": 0.04902869462966919,
+      "learning_rate": 2.266e-05,
+      "loss": 0.0009,
+      "step": 13670
+    },
+    {
+      "epoch": 2.7359999999999998,
+      "grad_norm": 0.0029866821132600307,
+      "learning_rate": 2.264e-05,
+      "loss": 0.0255,
+      "step": 13680
+    },
+    {
+      "epoch": 2.738,
+      "grad_norm": 0.004088583867996931,
+      "learning_rate": 2.2620000000000004e-05,
+      "loss": 0.0288,
+      "step": 13690
+    },
+    {
+      "epoch": 2.74,
+      "grad_norm": 0.04366866126656532,
+      "learning_rate": 2.26e-05,
+      "loss": 0.0027,
+      "step": 13700
+    },
+    {
+      "epoch": 2.742,
+      "grad_norm": 0.6947624683380127,
+      "learning_rate": 2.258e-05,
+      "loss": 0.0009,
+      "step": 13710
+    },
+    {
+      "epoch": 2.7439999999999998,
+      "grad_norm": 0.011672383174300194,
+      "learning_rate": 2.256e-05,
+      "loss": 0.06,
+      "step": 13720
+    },
+    {
+      "epoch": 2.746,
+      "grad_norm": 0.0022892083507031202,
+      "learning_rate": 2.254e-05,
+      "loss": 0.0385,
+      "step": 13730
+    },
+    {
+      "epoch": 2.748,
+      "grad_norm": 0.014559238217771053,
+      "learning_rate": 2.252e-05,
+      "loss": 0.001,
+      "step": 13740
+    },
+    {
+      "epoch": 2.75,
+      "grad_norm": 0.0139195891097188,
+      "learning_rate": 2.25e-05,
+      "loss": 0.0528,
+      "step": 13750
+    },
+    {
+      "epoch": 2.752,
+      "grad_norm": 0.011245626956224442,
+      "learning_rate": 2.248e-05,
+      "loss": 0.0125,
+      "step": 13760
+    },
+    {
+      "epoch": 2.754,
+      "grad_norm": 0.009724706411361694,
+      "learning_rate": 2.2460000000000002e-05,
+      "loss": 0.0006,
+      "step": 13770
+    },
+    {
+      "epoch": 2.7560000000000002,
+      "grad_norm": 0.017409011721611023,
+      "learning_rate": 2.244e-05,
+      "loss": 0.018,
+      "step": 13780
+    },
+    {
+      "epoch": 2.758,
+      "grad_norm": 0.03546595573425293,
+      "learning_rate": 2.2420000000000002e-05,
+      "loss": 0.0517,
+      "step": 13790
+    },
+    {
+      "epoch": 2.76,
+      "grad_norm": 0.004322000779211521,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 0.0185,
+      "step": 13800
+    },
+    {
+      "epoch": 2.762,
+      "grad_norm": 0.013103820383548737,
+      "learning_rate": 2.2380000000000003e-05,
+      "loss": 0.0113,
+      "step": 13810
+    },
+    {
+      "epoch": 2.7640000000000002,
+      "grad_norm": 0.008694900199770927,
+      "learning_rate": 2.236e-05,
+      "loss": 0.0322,
+      "step": 13820
+    },
+    {
+      "epoch": 2.766,
+      "grad_norm": 0.026327064260840416,
+      "learning_rate": 2.234e-05,
+      "loss": 0.0014,
+      "step": 13830
+    },
+    {
+      "epoch": 2.768,
+      "grad_norm": 0.025663388893008232,
+      "learning_rate": 2.2320000000000003e-05,
+      "loss": 0.0107,
+      "step": 13840
+    },
+    {
+      "epoch": 2.77,
+      "grad_norm": 0.0036214159335941076,
+      "learning_rate": 2.23e-05,
+      "loss": 0.0395,
+      "step": 13850
+    },
+    {
+      "epoch": 2.7720000000000002,
+      "grad_norm": 0.002212496241554618,
+      "learning_rate": 2.228e-05,
+      "loss": 0.0008,
+      "step": 13860
+    },
+    {
+      "epoch": 2.774,
+      "grad_norm": 0.007965583354234695,
+      "learning_rate": 2.226e-05,
+      "loss": 0.0433,
+      "step": 13870
+    },
+    {
+      "epoch": 2.776,
+      "grad_norm": 0.006145271472632885,
+      "learning_rate": 2.224e-05,
+      "loss": 0.0205,
+      "step": 13880
+    },
+    {
+      "epoch": 2.778,
+      "grad_norm": 0.004019021987915039,
+      "learning_rate": 2.222e-05,
+      "loss": 0.0005,
+      "step": 13890
+    },
+    {
+      "epoch": 2.7800000000000002,
+      "grad_norm": 0.013389227911829948,
+      "learning_rate": 2.22e-05,
+      "loss": 0.0006,
+      "step": 13900
+    },
+    {
+      "epoch": 2.782,
+      "grad_norm": 0.009832042269408703,
+      "learning_rate": 2.218e-05,
+      "loss": 0.0021,
+      "step": 13910
+    },
+    {
+      "epoch": 2.784,
+      "grad_norm": 0.0037811503279954195,
+      "learning_rate": 2.216e-05,
+      "loss": 0.0152,
+      "step": 13920
+    },
+    {
+      "epoch": 2.786,
+      "grad_norm": 0.00249070068821311,
+      "learning_rate": 2.214e-05,
+      "loss": 0.0013,
+      "step": 13930
+    },
+    {
+      "epoch": 2.7880000000000003,
+      "grad_norm": 0.008035962469875813,
+      "learning_rate": 2.212e-05,
+      "loss": 0.0216,
+      "step": 13940
+    },
+    {
+      "epoch": 2.79,
+      "grad_norm": 0.003760936204344034,
+      "learning_rate": 2.2100000000000002e-05,
+      "loss": 0.0454,
+      "step": 13950
+    },
+    {
+      "epoch": 2.792,
+      "grad_norm": 0.15059198439121246,
+      "learning_rate": 2.2080000000000002e-05,
+      "loss": 0.0389,
+      "step": 13960
+    },
+    {
+      "epoch": 2.794,
+      "grad_norm": 0.00915328599512577,
+      "learning_rate": 2.206e-05,
+      "loss": 0.0007,
+      "step": 13970
+    },
+    {
+      "epoch": 2.7960000000000003,
+      "grad_norm": 0.019227130338549614,
+      "learning_rate": 2.2040000000000002e-05,
+      "loss": 0.0225,
+      "step": 13980
+    },
+    {
+      "epoch": 2.798,
+      "grad_norm": 0.0020889227744191885,
+      "learning_rate": 2.2020000000000003e-05,
+      "loss": 0.0029,
+      "step": 13990
+    },
+    {
+      "epoch": 2.8,
+      "grad_norm": 0.022324854508042336,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 0.0008,
+      "step": 14000
+    },
+    {
+      "epoch": 2.802,
+      "grad_norm": 0.010531187057495117,
+      "learning_rate": 2.198e-05,
+      "loss": 0.0039,
+      "step": 14010
+    },
+    {
+      "epoch": 2.8040000000000003,
+      "grad_norm": 0.10176002234220505,
+      "learning_rate": 2.196e-05,
+      "loss": 0.0275,
+      "step": 14020
+    },
+    {
+      "epoch": 2.806,
+      "grad_norm": 0.013241732493042946,
+      "learning_rate": 2.1940000000000003e-05,
+      "loss": 0.0005,
+      "step": 14030
+    },
+    {
+      "epoch": 2.808,
+      "grad_norm": 0.009389587678015232,
+      "learning_rate": 2.192e-05,
+      "loss": 0.0387,
+      "step": 14040
+    },
+    {
+      "epoch": 2.81,
+      "grad_norm": 0.003362020943313837,
+      "learning_rate": 2.19e-05,
+      "loss": 0.0007,
+      "step": 14050
+    },
+    {
+      "epoch": 2.8120000000000003,
+      "grad_norm": 0.0029993215575814247,
+      "learning_rate": 2.188e-05,
+      "loss": 0.0324,
+      "step": 14060
+    },
+    {
+      "epoch": 2.814,
+      "grad_norm": 0.04549940302968025,
+      "learning_rate": 2.186e-05,
+      "loss": 0.0656,
+      "step": 14070
+    },
+    {
+      "epoch": 2.816,
+      "grad_norm": 9.865263938903809,
+      "learning_rate": 2.184e-05,
+      "loss": 0.0314,
+      "step": 14080
+    },
+    {
+      "epoch": 2.818,
+      "grad_norm": 0.007688828743994236,
+      "learning_rate": 2.182e-05,
+      "loss": 0.0312,
+      "step": 14090
+    },
+    {
+      "epoch": 2.82,
+      "grad_norm": 0.01804133504629135,
+      "learning_rate": 2.18e-05,
+      "loss": 0.0003,
+      "step": 14100
+    },
+    {
+      "epoch": 2.822,
+      "grad_norm": 0.004246327094733715,
+      "learning_rate": 2.178e-05,
+      "loss": 0.021,
+      "step": 14110
+    },
+    {
+      "epoch": 2.824,
+      "grad_norm": 0.01003707479685545,
+      "learning_rate": 2.176e-05,
+      "loss": 0.0014,
+      "step": 14120
+    },
+    {
+      "epoch": 2.826,
+      "grad_norm": 0.015356244519352913,
+      "learning_rate": 2.1740000000000002e-05,
+      "loss": 0.0007,
+      "step": 14130
+    },
+    {
+      "epoch": 2.828,
+      "grad_norm": 0.0139754144474864,
+      "learning_rate": 2.1720000000000002e-05,
+      "loss": 0.0004,
+      "step": 14140
+    },
+    {
+      "epoch": 2.83,
+      "grad_norm": 0.1567094773054123,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 0.001,
+      "step": 14150
+    },
+    {
+      "epoch": 2.832,
+      "grad_norm": 0.012544950470328331,
+      "learning_rate": 2.168e-05,
+      "loss": 0.0147,
+      "step": 14160
+    },
+    {
+      "epoch": 2.834,
+      "grad_norm": 0.005918503273278475,
+      "learning_rate": 2.166e-05,
+      "loss": 0.0014,
+      "step": 14170
+    },
+    {
+      "epoch": 2.836,
+      "grad_norm": 4.1460065841674805,
+      "learning_rate": 2.1640000000000003e-05,
+      "loss": 0.0416,
+      "step": 14180
+    },
+    {
+      "epoch": 2.838,
+      "grad_norm": 0.0017820524517446756,
+      "learning_rate": 2.162e-05,
+      "loss": 0.0352,
+      "step": 14190
+    },
+    {
+      "epoch": 2.84,
+      "grad_norm": 0.01237359270453453,
+      "learning_rate": 2.16e-05,
+      "loss": 0.0016,
+      "step": 14200
+    },
+    {
+      "epoch": 2.842,
+      "grad_norm": 0.002615107223391533,
+      "learning_rate": 2.158e-05,
+      "loss": 0.0083,
+      "step": 14210
+    },
+    {
+      "epoch": 2.844,
+      "grad_norm": 0.006978666875511408,
+      "learning_rate": 2.1560000000000004e-05,
+      "loss": 0.0426,
+      "step": 14220
+    },
+    {
+      "epoch": 2.846,
+      "grad_norm": 0.0061668511480093,
+      "learning_rate": 2.154e-05,
+      "loss": 0.0171,
+      "step": 14230
+    },
+    {
+      "epoch": 2.848,
+      "grad_norm": 0.004591886419802904,
+      "learning_rate": 2.152e-05,
+      "loss": 0.0529,
+      "step": 14240
+    },
+    {
+      "epoch": 2.85,
+      "grad_norm": 0.0053439270704984665,
+      "learning_rate": 2.15e-05,
+      "loss": 0.0039,
+      "step": 14250
+    },
+    {
+      "epoch": 2.852,
+      "grad_norm": 0.005400092341005802,
+      "learning_rate": 2.148e-05,
+      "loss": 0.0006,
+      "step": 14260
+    },
+    {
+      "epoch": 2.854,
+      "grad_norm": 0.011323225684463978,
+      "learning_rate": 2.146e-05,
+      "loss": 0.0053,
+      "step": 14270
+    },
+    {
+      "epoch": 2.856,
+      "grad_norm": 9.84250545501709,
+      "learning_rate": 2.144e-05,
+      "loss": 0.0966,
+      "step": 14280
+    },
+    {
+      "epoch": 2.858,
+      "grad_norm": 0.013491962105035782,
+      "learning_rate": 2.142e-05,
+      "loss": 0.0007,
+      "step": 14290
+    },
+    {
+      "epoch": 2.86,
+      "grad_norm": 0.003985918127000332,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 0.015,
+      "step": 14300
+    },
+    {
+      "epoch": 2.862,
+      "grad_norm": 3.3448140621185303,
+      "learning_rate": 2.138e-05,
+      "loss": 0.0021,
+      "step": 14310
+    },
+    {
+      "epoch": 2.864,
+      "grad_norm": 0.02373521216213703,
+      "learning_rate": 2.1360000000000002e-05,
+      "loss": 0.0245,
+      "step": 14320
+    },
+    {
+      "epoch": 2.866,
+      "grad_norm": 0.029405981302261353,
+      "learning_rate": 2.1340000000000002e-05,
+      "loss": 0.0061,
+      "step": 14330
+    },
+    {
+      "epoch": 2.868,
+      "grad_norm": 0.007544131018221378,
+      "learning_rate": 2.1320000000000003e-05,
+      "loss": 0.0445,
+      "step": 14340
+    },
+    {
+      "epoch": 2.87,
+      "grad_norm": 0.035514336079359055,
+      "learning_rate": 2.13e-05,
+      "loss": 0.0183,
+      "step": 14350
+    },
+    {
+      "epoch": 2.872,
+      "grad_norm": 0.0044698393903672695,
+      "learning_rate": 2.128e-05,
+      "loss": 0.0425,
+      "step": 14360
+    },
+    {
+      "epoch": 2.874,
+      "grad_norm": 0.14416873455047607,
+      "learning_rate": 2.1260000000000003e-05,
+      "loss": 0.0067,
+      "step": 14370
+    },
+    {
+      "epoch": 2.876,
+      "grad_norm": 4.21664571762085,
+      "learning_rate": 2.124e-05,
+      "loss": 0.0773,
+      "step": 14380
+    },
+    {
+      "epoch": 2.878,
+      "grad_norm": 5.436165809631348,
+      "learning_rate": 2.122e-05,
+      "loss": 0.041,
+      "step": 14390
+    },
+    {
+      "epoch": 2.88,
+      "grad_norm": 0.02734544686973095,
+      "learning_rate": 2.12e-05,
+      "loss": 0.0032,
+      "step": 14400
+    },
+    {
+      "epoch": 2.882,
+      "grad_norm": 0.0023346785455942154,
+      "learning_rate": 2.118e-05,
+      "loss": 0.0655,
+      "step": 14410
+    },
+    {
+      "epoch": 2.884,
+      "grad_norm": 0.006281679030507803,
+      "learning_rate": 2.116e-05,
+      "loss": 0.0142,
+      "step": 14420
+    },
+    {
+      "epoch": 2.886,
+      "grad_norm": 0.024367235600948334,
+      "learning_rate": 2.114e-05,
+      "loss": 0.0011,
+      "step": 14430
+    },
+    {
+      "epoch": 2.888,
+      "grad_norm": 0.00849449634552002,
+      "learning_rate": 2.112e-05,
+      "loss": 0.0981,
+      "step": 14440
+    },
+    {
+      "epoch": 2.89,
+      "grad_norm": 0.021741211414337158,
+      "learning_rate": 2.11e-05,
+      "loss": 0.0032,
+      "step": 14450
+    },
+    {
+      "epoch": 2.892,
+      "grad_norm": 0.0908471941947937,
+      "learning_rate": 2.1079999999999998e-05,
+      "loss": 0.006,
+      "step": 14460
+    },
+    {
+      "epoch": 2.894,
+      "grad_norm": 0.009153512306511402,
+      "learning_rate": 2.106e-05,
+      "loss": 0.0031,
+      "step": 14470
+    },
+    {
+      "epoch": 2.896,
+      "grad_norm": 0.0878956988453865,
+      "learning_rate": 2.1040000000000002e-05,
+      "loss": 0.0023,
+      "step": 14480
+    },
+    {
+      "epoch": 2.898,
+      "grad_norm": 0.22682340443134308,
+      "learning_rate": 2.1020000000000002e-05,
+      "loss": 0.0026,
+      "step": 14490
+    },
+    {
+      "epoch": 2.9,
+      "grad_norm": 0.010414546355605125,
+      "learning_rate": 2.1e-05,
+      "loss": 0.0239,
+      "step": 14500
+    },
+    {
+      "epoch": 2.902,
+      "grad_norm": 1.4128164052963257,
+      "learning_rate": 2.098e-05,
+      "loss": 0.0017,
+      "step": 14510
+    },
+    {
+      "epoch": 2.904,
+      "grad_norm": 0.1310981959104538,
+      "learning_rate": 2.0960000000000003e-05,
+      "loss": 0.0339,
+      "step": 14520
+    },
+    {
+      "epoch": 2.906,
+      "grad_norm": 0.0020759692415595055,
+      "learning_rate": 2.0940000000000003e-05,
+      "loss": 0.001,
+      "step": 14530
+    },
+    {
+      "epoch": 2.908,
+      "grad_norm": 6.847393989562988,
+      "learning_rate": 2.092e-05,
+      "loss": 0.0242,
+      "step": 14540
+    },
+    {
+      "epoch": 2.91,
+      "grad_norm": 0.002632703399285674,
+      "learning_rate": 2.09e-05,
+      "loss": 0.0043,
+      "step": 14550
+    },
+    {
+      "epoch": 2.912,
+      "grad_norm": 0.12055040150880814,
+      "learning_rate": 2.0880000000000003e-05,
+      "loss": 0.0106,
+      "step": 14560
+    },
+    {
+      "epoch": 2.914,
+      "grad_norm": 0.008990177884697914,
+      "learning_rate": 2.086e-05,
+      "loss": 0.0003,
+      "step": 14570
+    },
+    {
+      "epoch": 2.916,
+      "grad_norm": 0.005368634592741728,
+      "learning_rate": 2.084e-05,
+      "loss": 0.032,
+      "step": 14580
+    },
+    {
+      "epoch": 2.918,
+      "grad_norm": 0.003290928667411208,
+      "learning_rate": 2.082e-05,
+      "loss": 0.0009,
+      "step": 14590
+    },
+    {
+      "epoch": 2.92,
+      "grad_norm": 0.0031266543082892895,
+      "learning_rate": 2.08e-05,
+      "loss": 0.0061,
+      "step": 14600
+    },
+    {
+      "epoch": 2.922,
+      "grad_norm": 0.020411498844623566,
+      "learning_rate": 2.078e-05,
+      "loss": 0.0033,
+      "step": 14610
+    },
+    {
+      "epoch": 2.924,
+      "grad_norm": 0.01999702863395214,
+      "learning_rate": 2.076e-05,
+      "loss": 0.0025,
+      "step": 14620
+    },
+    {
+      "epoch": 2.926,
+      "grad_norm": 2.3530824184417725,
+      "learning_rate": 2.074e-05,
+      "loss": 0.0325,
+      "step": 14630
+    },
+    {
+      "epoch": 2.928,
+      "grad_norm": 0.0432647280395031,
+      "learning_rate": 2.072e-05,
+      "loss": 0.0173,
+      "step": 14640
+    },
+    {
+      "epoch": 2.93,
+      "grad_norm": 0.03774017095565796,
+      "learning_rate": 2.07e-05,
+      "loss": 0.0638,
+      "step": 14650
+    },
+    {
+      "epoch": 2.932,
+      "grad_norm": 0.023275645449757576,
+      "learning_rate": 2.0680000000000002e-05,
+      "loss": 0.0008,
+      "step": 14660
+    },
+    {
+      "epoch": 2.934,
+      "grad_norm": 0.0035587181337177753,
+      "learning_rate": 2.0660000000000002e-05,
+      "loss": 0.0615,
+      "step": 14670
+    },
+    {
+      "epoch": 2.936,
+      "grad_norm": 0.012016230262815952,
+      "learning_rate": 2.0640000000000002e-05,
+      "loss": 0.0012,
+      "step": 14680
+    },
+    {
+      "epoch": 2.9379999999999997,
+      "grad_norm": 0.0014051154721528292,
+      "learning_rate": 2.062e-05,
+      "loss": 0.0021,
+      "step": 14690
+    },
+    {
+      "epoch": 2.94,
+      "grad_norm": 0.002413893351331353,
+      "learning_rate": 2.06e-05,
+      "loss": 0.0089,
+      "step": 14700
+    },
+    {
+      "epoch": 2.942,
+      "grad_norm": 0.002096255775541067,
+      "learning_rate": 2.0580000000000003e-05,
+      "loss": 0.0213,
+      "step": 14710
+    },
+    {
+      "epoch": 2.944,
+      "grad_norm": 0.1285342127084732,
+      "learning_rate": 2.0560000000000003e-05,
+      "loss": 0.0377,
+      "step": 14720
+    },
+    {
+      "epoch": 2.9459999999999997,
+      "grad_norm": 0.01958276331424713,
+      "learning_rate": 2.054e-05,
+      "loss": 0.001,
+      "step": 14730
+    },
+    {
+      "epoch": 2.948,
+      "grad_norm": 0.03653479367494583,
+      "learning_rate": 2.052e-05,
+      "loss": 0.0005,
+      "step": 14740
+    },
+    {
+      "epoch": 2.95,
+      "grad_norm": 0.009381549432873726,
+      "learning_rate": 2.05e-05,
+      "loss": 0.0014,
+      "step": 14750
+    },
+    {
+      "epoch": 2.952,
+      "grad_norm": 0.03101208433508873,
+      "learning_rate": 2.048e-05,
+      "loss": 0.0331,
+      "step": 14760
+    },
+    {
+      "epoch": 2.9539999999999997,
+      "grad_norm": 0.01349340658634901,
+      "learning_rate": 2.046e-05,
+      "loss": 0.0417,
+      "step": 14770
+    },
+    {
+      "epoch": 2.956,
+      "grad_norm": 0.0013646255247294903,
+      "learning_rate": 2.044e-05,
+      "loss": 0.0406,
+      "step": 14780
+    },
+    {
+      "epoch": 2.958,
+      "grad_norm": 0.0011694016866385937,
+      "learning_rate": 2.042e-05,
+      "loss": 0.0306,
+      "step": 14790
+    },
+    {
+      "epoch": 2.96,
+      "grad_norm": 0.5294227004051208,
+      "learning_rate": 2.04e-05,
+      "loss": 0.0279,
+      "step": 14800
+    },
+    {
+      "epoch": 2.9619999999999997,
+      "grad_norm": 0.006853798869997263,
+      "learning_rate": 2.038e-05,
+      "loss": 0.0045,
+      "step": 14810
+    },
+    {
+      "epoch": 2.964,
+      "grad_norm": 0.0035395545419305563,
+      "learning_rate": 2.036e-05,
+      "loss": 0.0244,
+      "step": 14820
+    },
+    {
+      "epoch": 2.966,
+      "grad_norm": 0.0022441355977207422,
+      "learning_rate": 2.0340000000000002e-05,
+      "loss": 0.0012,
+      "step": 14830
+    },
+    {
+      "epoch": 2.968,
+      "grad_norm": 0.016829049214720726,
+      "learning_rate": 2.032e-05,
+      "loss": 0.0013,
+      "step": 14840
+    },
+    {
+      "epoch": 2.9699999999999998,
+      "grad_norm": 0.11161936819553375,
+      "learning_rate": 2.0300000000000002e-05,
+      "loss": 0.0435,
+      "step": 14850
+    },
+    {
+      "epoch": 2.972,
+      "grad_norm": 0.006620103493332863,
+      "learning_rate": 2.0280000000000002e-05,
+      "loss": 0.0388,
+      "step": 14860
+    },
+    {
+      "epoch": 2.974,
+      "grad_norm": 0.27429601550102234,
+      "learning_rate": 2.0260000000000003e-05,
+      "loss": 0.001,
+      "step": 14870
+    },
+    {
+      "epoch": 2.976,
+      "grad_norm": 0.01480827946215868,
+      "learning_rate": 2.024e-05,
+      "loss": 0.0008,
+      "step": 14880
+    },
+    {
+      "epoch": 2.9779999999999998,
+      "grad_norm": 0.012657509185373783,
+      "learning_rate": 2.022e-05,
+      "loss": 0.0022,
+      "step": 14890
+    },
+    {
+      "epoch": 2.98,
+      "grad_norm": 0.00810237880796194,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 0.0025,
+      "step": 14900
+    },
+    {
+      "epoch": 2.982,
+      "grad_norm": 0.0014344545779749751,
+      "learning_rate": 2.0180000000000003e-05,
+      "loss": 0.0162,
+      "step": 14910
+    },
+    {
+      "epoch": 2.984,
+      "grad_norm": 0.004025655798614025,
+      "learning_rate": 2.016e-05,
+      "loss": 0.0004,
+      "step": 14920
+    },
+    {
+      "epoch": 2.9859999999999998,
+      "grad_norm": 0.004194976761937141,
+      "learning_rate": 2.014e-05,
+      "loss": 0.0006,
+      "step": 14930
+    },
+    {
+      "epoch": 2.988,
+      "grad_norm": 0.009011647664010525,
+      "learning_rate": 2.012e-05,
+      "loss": 0.0558,
+      "step": 14940
+    },
+    {
+      "epoch": 2.99,
+      "grad_norm": 0.003613577224314213,
+      "learning_rate": 2.01e-05,
+      "loss": 0.0033,
+      "step": 14950
+    },
+    {
+      "epoch": 2.992,
+      "grad_norm": 0.016397807747125626,
+      "learning_rate": 2.008e-05,
+      "loss": 0.0066,
+      "step": 14960
+    },
+    {
+      "epoch": 2.9939999999999998,
+      "grad_norm": 0.006938213016837835,
+      "learning_rate": 2.006e-05,
+      "loss": 0.003,
+      "step": 14970
+    },
+    {
+      "epoch": 2.996,
+      "grad_norm": 0.0022056391462683678,
+      "learning_rate": 2.004e-05,
+      "loss": 0.0008,
+      "step": 14980
+    },
+    {
+      "epoch": 2.998,
+      "grad_norm": 0.1262286901473999,
+      "learning_rate": 2.002e-05,
+      "loss": 0.0048,
+      "step": 14990
+    },
+    {
+      "epoch": 3.0,
+      "grad_norm": 1.3995038270950317,
+      "learning_rate": 2e-05,
+      "loss": 0.0034,
+      "step": 15000
+    },
+    {
+      "epoch": 3.0,
+      "eval_loss": 0.01657683774828911,
+      "eval_runtime": 95.2432,
+      "eval_samples_per_second": 104.994,
+      "eval_steps_per_second": 13.124,
+      "step": 15000
+    },
+    {
+      "epoch": 3.002,
+      "grad_norm": 3.7540829181671143,
+      "learning_rate": 1.9980000000000002e-05,
+      "loss": 0.0197,
+      "step": 15010
+    },
+    {
+      "epoch": 3.004,
+      "grad_norm": 0.0008035671780817211,
+      "learning_rate": 1.9960000000000002e-05,
+      "loss": 0.0389,
+      "step": 15020
+    },
+    {
+      "epoch": 3.006,
+      "grad_norm": 0.006811047438532114,
+      "learning_rate": 1.994e-05,
+      "loss": 0.0407,
+      "step": 15030
+    },
+    {
+      "epoch": 3.008,
+      "grad_norm": 0.000698752177413553,
+      "learning_rate": 1.992e-05,
+      "loss": 0.0006,
+      "step": 15040
+    },
+    {
+      "epoch": 3.01,
+      "grad_norm": 0.012239865958690643,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 0.0213,
+      "step": 15050
+    },
+    {
+      "epoch": 3.012,
+      "grad_norm": 0.0010888595134019852,
+      "learning_rate": 1.9880000000000003e-05,
+      "loss": 0.0509,
+      "step": 15060
+    },
+    {
+      "epoch": 3.014,
+      "grad_norm": 5.646819114685059,
+      "learning_rate": 1.986e-05,
+      "loss": 0.0442,
+      "step": 15070
+    },
+    {
+      "epoch": 3.016,
+      "grad_norm": 0.3257085978984833,
+      "learning_rate": 1.984e-05,
+      "loss": 0.0031,
+      "step": 15080
+    },
+    {
+      "epoch": 3.018,
+      "grad_norm": 0.166681170463562,
+      "learning_rate": 1.982e-05,
+      "loss": 0.0015,
+      "step": 15090
+    },
+    {
+      "epoch": 3.02,
+      "grad_norm": 0.010647795163094997,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 0.0003,
+      "step": 15100
+    },
+    {
+      "epoch": 3.022,
+      "grad_norm": 10.128013610839844,
+      "learning_rate": 1.978e-05,
+      "loss": 0.0077,
+      "step": 15110
+    },
+    {
+      "epoch": 3.024,
+      "grad_norm": 0.005560045130550861,
+      "learning_rate": 1.976e-05,
+      "loss": 0.0423,
+      "step": 15120
+    },
+    {
+      "epoch": 3.026,
+      "grad_norm": 0.01081987377256155,
+      "learning_rate": 1.974e-05,
+      "loss": 0.0735,
+      "step": 15130
+    },
+    {
+      "epoch": 3.028,
+      "grad_norm": 0.0014237303985282779,
+      "learning_rate": 1.972e-05,
+      "loss": 0.0004,
+      "step": 15140
+    },
+    {
+      "epoch": 3.03,
+      "grad_norm": 4.161525726318359,
+      "learning_rate": 1.97e-05,
+      "loss": 0.0254,
+      "step": 15150
+    },
+    {
+      "epoch": 3.032,
+      "grad_norm": 4.0388641357421875,
+      "learning_rate": 1.968e-05,
+      "loss": 0.006,
+      "step": 15160
+    },
+    {
+      "epoch": 3.034,
+      "grad_norm": 0.008463465608656406,
+      "learning_rate": 1.966e-05,
+      "loss": 0.0144,
+      "step": 15170
+    },
+    {
+      "epoch": 3.036,
+      "grad_norm": 0.006266901269555092,
+      "learning_rate": 1.9640000000000002e-05,
+      "loss": 0.0045,
+      "step": 15180
+    },
+    {
+      "epoch": 3.038,
+      "grad_norm": 0.020866410806775093,
+      "learning_rate": 1.9620000000000002e-05,
+      "loss": 0.051,
+      "step": 15190
+    },
+    {
+      "epoch": 3.04,
+      "grad_norm": 0.06101961433887482,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 0.0082,
+      "step": 15200
+    },
+    {
+      "epoch": 3.042,
+      "grad_norm": 0.0015061348676681519,
+      "learning_rate": 1.9580000000000002e-05,
+      "loss": 0.0237,
+      "step": 15210
+    },
+    {
+      "epoch": 3.044,
+      "grad_norm": 0.004363874439150095,
+      "learning_rate": 1.956e-05,
+      "loss": 0.0161,
+      "step": 15220
+    },
+    {
+      "epoch": 3.046,
+      "grad_norm": 0.003860832192003727,
+      "learning_rate": 1.954e-05,
+      "loss": 0.0452,
+      "step": 15230
+    },
+    {
+      "epoch": 3.048,
+      "grad_norm": 0.0023255248088389635,
+      "learning_rate": 1.9520000000000003e-05,
+      "loss": 0.002,
+      "step": 15240
+    },
+    {
+      "epoch": 3.05,
+      "grad_norm": 2.2219414710998535,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 0.003,
+      "step": 15250
+    },
+    {
+      "epoch": 3.052,
+      "grad_norm": 0.13828614354133606,
+      "learning_rate": 1.948e-05,
+      "loss": 0.0455,
+      "step": 15260
+    },
+    {
+      "epoch": 3.054,
+      "grad_norm": 0.003867669263854623,
+      "learning_rate": 1.946e-05,
+      "loss": 0.0004,
+      "step": 15270
+    },
+    {
+      "epoch": 3.056,
+      "grad_norm": 0.007381816394627094,
+      "learning_rate": 1.944e-05,
+      "loss": 0.0005,
+      "step": 15280
+    },
+    {
+      "epoch": 3.058,
+      "grad_norm": 0.008525327779352665,
+      "learning_rate": 1.942e-05,
+      "loss": 0.0427,
+      "step": 15290
+    },
+    {
+      "epoch": 3.06,
+      "grad_norm": 0.03834341838955879,
+      "learning_rate": 1.94e-05,
+      "loss": 0.0023,
+      "step": 15300
+    },
+    {
+      "epoch": 3.062,
+      "grad_norm": 0.10945117473602295,
+      "learning_rate": 1.938e-05,
+      "loss": 0.0174,
+      "step": 15310
+    },
+    {
+      "epoch": 3.064,
+      "grad_norm": 0.011007886379957199,
+      "learning_rate": 1.936e-05,
+      "loss": 0.0007,
+      "step": 15320
+    },
+    {
+      "epoch": 3.066,
+      "grad_norm": 0.006719260010868311,
+      "learning_rate": 1.934e-05,
+      "loss": 0.0003,
+      "step": 15330
+    },
+    {
+      "epoch": 3.068,
+      "grad_norm": 0.0017096955562010407,
+      "learning_rate": 1.932e-05,
+      "loss": 0.0163,
+      "step": 15340
+    },
+    {
+      "epoch": 3.07,
+      "grad_norm": 0.00444265129044652,
+      "learning_rate": 1.93e-05,
+      "loss": 0.0008,
+      "step": 15350
+    },
+    {
+      "epoch": 3.072,
+      "grad_norm": 0.0027511902153491974,
+      "learning_rate": 1.9280000000000002e-05,
+      "loss": 0.0048,
+      "step": 15360
+    },
+    {
+      "epoch": 3.074,
+      "grad_norm": 0.08696744590997696,
+      "learning_rate": 1.9260000000000002e-05,
+      "loss": 0.0007,
+      "step": 15370
+    },
+    {
+      "epoch": 3.076,
+      "grad_norm": 0.010581353679299355,
+      "learning_rate": 1.924e-05,
+      "loss": 0.0006,
+      "step": 15380
+    },
+    {
+      "epoch": 3.078,
+      "grad_norm": 6.367326736450195,
+      "learning_rate": 1.9220000000000002e-05,
+      "loss": 0.0151,
+      "step": 15390
+    },
+    {
+      "epoch": 3.08,
+      "grad_norm": 0.038151416927576065,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 0.0007,
+      "step": 15400
+    },
+    {
+      "epoch": 3.082,
+      "grad_norm": 0.0035086292773485184,
+      "learning_rate": 1.918e-05,
+      "loss": 0.0039,
+      "step": 15410
+    },
+    {
+      "epoch": 3.084,
+      "grad_norm": 0.006965567357838154,
+      "learning_rate": 1.916e-05,
+      "loss": 0.0107,
+      "step": 15420
+    },
+    {
+      "epoch": 3.086,
+      "grad_norm": 0.006669293623417616,
+      "learning_rate": 1.914e-05,
+      "loss": 0.0393,
+      "step": 15430
+    },
+    {
+      "epoch": 3.088,
+      "grad_norm": 0.015491176396608353,
+      "learning_rate": 1.9120000000000003e-05,
+      "loss": 0.0008,
+      "step": 15440
+    },
+    {
+      "epoch": 3.09,
+      "grad_norm": 0.0049215275794267654,
+      "learning_rate": 1.91e-05,
+      "loss": 0.0156,
+      "step": 15450
+    },
+    {
+      "epoch": 3.092,
+      "grad_norm": 0.006124439183622599,
+      "learning_rate": 1.908e-05,
+      "loss": 0.0012,
+      "step": 15460
+    },
+    {
+      "epoch": 3.094,
+      "grad_norm": 0.00404756236821413,
+      "learning_rate": 1.906e-05,
+      "loss": 0.0004,
+      "step": 15470
+    },
+    {
+      "epoch": 3.096,
+      "grad_norm": 0.0698126032948494,
+      "learning_rate": 1.904e-05,
+      "loss": 0.0005,
+      "step": 15480
+    },
+    {
+      "epoch": 3.098,
+      "grad_norm": 0.2247430384159088,
+      "learning_rate": 1.902e-05,
+      "loss": 0.0006,
+      "step": 15490
+    },
+    {
+      "epoch": 3.1,
+      "grad_norm": 0.028489235788583755,
+      "learning_rate": 1.9e-05,
+      "loss": 0.0127,
+      "step": 15500
+    },
+    {
+      "epoch": 3.102,
+      "grad_norm": 0.0063696675933897495,
+      "learning_rate": 1.898e-05,
+      "loss": 0.0007,
+      "step": 15510
+    },
+    {
+      "epoch": 3.104,
+      "grad_norm": 0.0047574895434081554,
+      "learning_rate": 1.896e-05,
+      "loss": 0.0002,
+      "step": 15520
+    },
+    {
+      "epoch": 3.106,
+      "grad_norm": 0.05734870955348015,
+      "learning_rate": 1.894e-05,
+      "loss": 0.0011,
+      "step": 15530
+    },
+    {
+      "epoch": 3.108,
+      "grad_norm": 0.0044009024277329445,
+      "learning_rate": 1.8920000000000002e-05,
+      "loss": 0.0823,
+      "step": 15540
+    },
+    {
+      "epoch": 3.11,
+      "grad_norm": 0.002054518321529031,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 0.0005,
+      "step": 15550
+    },
+    {
+      "epoch": 3.112,
+      "grad_norm": 0.017458245158195496,
+      "learning_rate": 1.888e-05,
+      "loss": 0.001,
+      "step": 15560
+    },
+    {
+      "epoch": 3.114,
+      "grad_norm": 0.008351721800863743,
+      "learning_rate": 1.886e-05,
+      "loss": 0.0024,
+      "step": 15570
+    },
+    {
+      "epoch": 3.116,
+      "grad_norm": 0.002828374970704317,
+      "learning_rate": 1.8840000000000003e-05,
+      "loss": 0.0004,
+      "step": 15580
+    },
+    {
+      "epoch": 3.118,
+      "grad_norm": 0.0023917988874018192,
+      "learning_rate": 1.8820000000000003e-05,
+      "loss": 0.0003,
+      "step": 15590
+    },
+    {
+      "epoch": 3.12,
+      "grad_norm": 0.02163778804242611,
+      "learning_rate": 1.88e-05,
+      "loss": 0.0456,
+      "step": 15600
+    },
+    {
+      "epoch": 3.122,
+      "grad_norm": 0.0021023666486144066,
+      "learning_rate": 1.878e-05,
+      "loss": 0.0014,
+      "step": 15610
+    },
+    {
+      "epoch": 3.124,
+      "grad_norm": 0.006280581932514906,
+      "learning_rate": 1.876e-05,
+      "loss": 0.0151,
+      "step": 15620
+    },
+    {
+      "epoch": 3.126,
+      "grad_norm": 0.23775038123130798,
+      "learning_rate": 1.8740000000000004e-05,
+      "loss": 0.0005,
+      "step": 15630
+    },
+    {
+      "epoch": 3.128,
+      "grad_norm": 0.7629896402359009,
+      "learning_rate": 1.872e-05,
+      "loss": 0.0009,
+      "step": 15640
+    },
+    {
+      "epoch": 3.13,
+      "grad_norm": 0.07593518495559692,
+      "learning_rate": 1.87e-05,
+      "loss": 0.0013,
+      "step": 15650
+    },
+    {
+      "epoch": 3.132,
+      "grad_norm": 0.0027242007199674845,
+      "learning_rate": 1.868e-05,
+      "loss": 0.0936,
+      "step": 15660
+    },
+    {
+      "epoch": 3.134,
+      "grad_norm": 0.00594953540712595,
+      "learning_rate": 1.866e-05,
+      "loss": 0.0412,
+      "step": 15670
+    },
+    {
+      "epoch": 3.136,
+      "grad_norm": 0.007625892758369446,
+      "learning_rate": 1.864e-05,
+      "loss": 0.0003,
+      "step": 15680
+    },
+    {
+      "epoch": 3.138,
+      "grad_norm": 0.003998244646936655,
+      "learning_rate": 1.862e-05,
+      "loss": 0.0086,
+      "step": 15690
+    },
+    {
+      "epoch": 3.14,
+      "grad_norm": 0.02065129391849041,
+      "learning_rate": 1.86e-05,
+      "loss": 0.0052,
+      "step": 15700
+    },
+    {
+      "epoch": 3.142,
+      "grad_norm": 0.0024038213305175304,
+      "learning_rate": 1.858e-05,
+      "loss": 0.0003,
+      "step": 15710
+    },
+    {
+      "epoch": 3.144,
+      "grad_norm": 0.0023801634088158607,
+      "learning_rate": 1.856e-05,
+      "loss": 0.0519,
+      "step": 15720
+    },
+    {
+      "epoch": 3.146,
+      "grad_norm": 0.050345417112112045,
+      "learning_rate": 1.8540000000000002e-05,
+      "loss": 0.0507,
+      "step": 15730
+    },
+    {
+      "epoch": 3.148,
+      "grad_norm": 3.0590100288391113,
+      "learning_rate": 1.8520000000000002e-05,
+      "loss": 0.0224,
+      "step": 15740
+    },
+    {
+      "epoch": 3.15,
+      "grad_norm": 0.051652975380420685,
+      "learning_rate": 1.85e-05,
+      "loss": 0.0006,
+      "step": 15750
+    },
+    {
+      "epoch": 3.152,
+      "grad_norm": 0.00765568669885397,
+      "learning_rate": 1.848e-05,
+      "loss": 0.0007,
+      "step": 15760
+    },
+    {
+      "epoch": 3.154,
+      "grad_norm": 0.08669667690992355,
+      "learning_rate": 1.846e-05,
+      "loss": 0.0471,
+      "step": 15770
+    },
+    {
+      "epoch": 3.156,
+      "grad_norm": 0.0020077722147107124,
+      "learning_rate": 1.8440000000000003e-05,
+      "loss": 0.0436,
+      "step": 15780
+    },
+    {
+      "epoch": 3.158,
+      "grad_norm": 0.008714812807738781,
+      "learning_rate": 1.842e-05,
+      "loss": 0.0017,
+      "step": 15790
+    },
+    {
+      "epoch": 3.16,
+      "grad_norm": 0.008934718556702137,
+      "learning_rate": 1.84e-05,
+      "loss": 0.0003,
+      "step": 15800
+    },
+    {
+      "epoch": 3.162,
+      "grad_norm": 0.0029634966049343348,
+      "learning_rate": 1.838e-05,
+      "loss": 0.0048,
+      "step": 15810
+    },
+    {
+      "epoch": 3.164,
+      "grad_norm": 0.26886558532714844,
+      "learning_rate": 1.8360000000000004e-05,
+      "loss": 0.0015,
+      "step": 15820
+    },
+    {
+      "epoch": 3.166,
+      "grad_norm": 4.406497955322266,
+      "learning_rate": 1.834e-05,
+      "loss": 0.1182,
+      "step": 15830
+    },
+    {
+      "epoch": 3.168,
+      "grad_norm": 0.006616260390728712,
+      "learning_rate": 1.832e-05,
+      "loss": 0.0223,
+      "step": 15840
+    },
+    {
+      "epoch": 3.17,
+      "grad_norm": 0.007440556772053242,
+      "learning_rate": 1.83e-05,
+      "loss": 0.0015,
+      "step": 15850
+    },
+    {
+      "epoch": 3.172,
+      "grad_norm": 0.08389567583799362,
+      "learning_rate": 1.828e-05,
+      "loss": 0.0223,
+      "step": 15860
+    },
+    {
+      "epoch": 3.174,
+      "grad_norm": 0.08651598542928696,
+      "learning_rate": 1.826e-05,
+      "loss": 0.0119,
+      "step": 15870
+    },
+    {
+      "epoch": 3.176,
+      "grad_norm": 0.0037528069224208593,
+      "learning_rate": 1.824e-05,
+      "loss": 0.001,
+      "step": 15880
+    },
+    {
+      "epoch": 3.178,
+      "grad_norm": 0.000578329199925065,
+      "learning_rate": 1.8220000000000002e-05,
+      "loss": 0.0381,
+      "step": 15890
+    },
+    {
+      "epoch": 3.18,
+      "grad_norm": 0.0035956057254225016,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 0.0006,
+      "step": 15900
+    },
+    {
+      "epoch": 3.182,
+      "grad_norm": 0.013380602933466434,
+      "learning_rate": 1.818e-05,
+      "loss": 0.0178,
+      "step": 15910
+    },
+    {
+      "epoch": 3.184,
+      "grad_norm": 0.023049576207995415,
+      "learning_rate": 1.8160000000000002e-05,
+      "loss": 0.001,
+      "step": 15920
+    },
+    {
+      "epoch": 3.186,
+      "grad_norm": 0.0015177453169599175,
+      "learning_rate": 1.8140000000000003e-05,
+      "loss": 0.0016,
+      "step": 15930
+    },
+    {
+      "epoch": 3.188,
+      "grad_norm": 0.009679178707301617,
+      "learning_rate": 1.812e-05,
+      "loss": 0.0062,
+      "step": 15940
+    },
+    {
+      "epoch": 3.19,
+      "grad_norm": 0.0049837734550237656,
+      "learning_rate": 1.81e-05,
+      "loss": 0.0086,
+      "step": 15950
+    },
+    {
+      "epoch": 3.192,
+      "grad_norm": 3.9954798221588135,
+      "learning_rate": 1.808e-05,
+      "loss": 0.0424,
+      "step": 15960
+    },
+    {
+      "epoch": 3.194,
+      "grad_norm": 0.029047487303614616,
+      "learning_rate": 1.8060000000000003e-05,
+      "loss": 0.0254,
+      "step": 15970
+    },
+    {
+      "epoch": 3.196,
+      "grad_norm": 0.007384998258203268,
+      "learning_rate": 1.804e-05,
+      "loss": 0.0487,
+      "step": 15980
+    },
+    {
+      "epoch": 3.198,
+      "grad_norm": 2.2781753540039062,
+      "learning_rate": 1.802e-05,
+      "loss": 0.0428,
+      "step": 15990
+    },
+    {
+      "epoch": 3.2,
+      "grad_norm": 2.6462597846984863,
+      "learning_rate": 1.8e-05,
+      "loss": 0.0718,
+      "step": 16000
+    },
+    {
+      "epoch": 3.202,
+      "grad_norm": 0.005736429709941149,
+      "learning_rate": 1.798e-05,
+      "loss": 0.0014,
+      "step": 16010
+    },
+    {
+      "epoch": 3.204,
+      "grad_norm": 0.006327548995614052,
+      "learning_rate": 1.796e-05,
+      "loss": 0.0066,
+      "step": 16020
+    },
+    {
+      "epoch": 3.206,
+      "grad_norm": 0.006256358232349157,
+      "learning_rate": 1.794e-05,
+      "loss": 0.0191,
+      "step": 16030
+    },
+    {
+      "epoch": 3.208,
+      "grad_norm": 0.008648771792650223,
+      "learning_rate": 1.792e-05,
+      "loss": 0.0005,
+      "step": 16040
+    },
+    {
+      "epoch": 3.21,
+      "grad_norm": 1.948608160018921,
+      "learning_rate": 1.79e-05,
+      "loss": 0.0298,
+      "step": 16050
+    },
+    {
+      "epoch": 3.212,
+      "grad_norm": 0.21498942375183105,
+      "learning_rate": 1.7879999999999998e-05,
+      "loss": 0.0007,
+      "step": 16060
+    },
+    {
+      "epoch": 3.214,
+      "grad_norm": 0.004237090237438679,
+      "learning_rate": 1.7860000000000002e-05,
+      "loss": 0.0303,
+      "step": 16070
+    },
+    {
+      "epoch": 3.216,
+      "grad_norm": 0.0034377332776784897,
+      "learning_rate": 1.7840000000000002e-05,
+      "loss": 0.0252,
+      "step": 16080
+    },
+    {
+      "epoch": 3.218,
+      "grad_norm": 0.36846575140953064,
+      "learning_rate": 1.7820000000000002e-05,
+      "loss": 0.0419,
+      "step": 16090
+    },
+    {
+      "epoch": 3.22,
+      "grad_norm": 0.13659149408340454,
+      "learning_rate": 1.78e-05,
+      "loss": 0.0007,
+      "step": 16100
+    },
+    {
+      "epoch": 3.222,
+      "grad_norm": 0.015541269443929195,
+      "learning_rate": 1.7780000000000003e-05,
+      "loss": 0.0005,
+      "step": 16110
+    },
+    {
+      "epoch": 3.224,
+      "grad_norm": 0.03992128372192383,
+      "learning_rate": 1.7760000000000003e-05,
+      "loss": 0.0103,
+      "step": 16120
+    },
+    {
+      "epoch": 3.226,
+      "grad_norm": 0.005680619738996029,
+      "learning_rate": 1.774e-05,
+      "loss": 0.0011,
+      "step": 16130
+    },
+    {
+      "epoch": 3.228,
+      "grad_norm": 0.009108452126383781,
+      "learning_rate": 1.772e-05,
+      "loss": 0.0011,
+      "step": 16140
+    },
+    {
+      "epoch": 3.23,
+      "grad_norm": 0.0023171931970864534,
+      "learning_rate": 1.77e-05,
+      "loss": 0.0325,
+      "step": 16150
+    },
+    {
+      "epoch": 3.232,
+      "grad_norm": 0.01577291637659073,
+      "learning_rate": 1.7680000000000004e-05,
+      "loss": 0.0006,
+      "step": 16160
+    },
+    {
+      "epoch": 3.234,
+      "grad_norm": 0.03504559025168419,
+      "learning_rate": 1.766e-05,
+      "loss": 0.0491,
+      "step": 16170
+    },
+    {
+      "epoch": 3.2359999999999998,
+      "grad_norm": 0.033147890120744705,
+      "learning_rate": 1.764e-05,
+      "loss": 0.0039,
+      "step": 16180
+    },
+    {
+      "epoch": 3.238,
+      "grad_norm": 0.027644271031022072,
+      "learning_rate": 1.762e-05,
+      "loss": 0.0007,
+      "step": 16190
+    },
+    {
+      "epoch": 3.24,
+      "grad_norm": 0.2593711316585541,
+      "learning_rate": 1.76e-05,
+      "loss": 0.0306,
+      "step": 16200
+    },
+    {
+      "epoch": 3.242,
+      "grad_norm": 0.02173049934208393,
+      "learning_rate": 1.758e-05,
+      "loss": 0.0008,
+      "step": 16210
+    },
+    {
+      "epoch": 3.2439999999999998,
+      "grad_norm": 0.3893928825855255,
+      "learning_rate": 1.756e-05,
+      "loss": 0.0015,
+      "step": 16220
+    },
+    {
+      "epoch": 3.246,
+      "grad_norm": 1.5050745010375977,
+      "learning_rate": 1.754e-05,
+      "loss": 0.0016,
+      "step": 16230
+    },
+    {
+      "epoch": 3.248,
+      "grad_norm": 0.014095143415033817,
+      "learning_rate": 1.752e-05,
+      "loss": 0.0291,
+      "step": 16240
+    },
+    {
+      "epoch": 3.25,
+      "grad_norm": 0.020162342116236687,
+      "learning_rate": 1.75e-05,
+      "loss": 0.0011,
+      "step": 16250
+    },
+    {
+      "epoch": 3.252,
+      "grad_norm": 0.009506700560450554,
+      "learning_rate": 1.7480000000000002e-05,
+      "loss": 0.0011,
+      "step": 16260
+    },
+    {
+      "epoch": 3.254,
+      "grad_norm": 0.02688973769545555,
+      "learning_rate": 1.7460000000000002e-05,
+      "loss": 0.0235,
+      "step": 16270
+    },
+    {
+      "epoch": 3.2560000000000002,
+      "grad_norm": 0.0014943861169740558,
+      "learning_rate": 1.7440000000000002e-05,
+      "loss": 0.0301,
+      "step": 16280
+    },
+    {
+      "epoch": 3.258,
+      "grad_norm": 0.008569997735321522,
+      "learning_rate": 1.742e-05,
+      "loss": 0.0105,
+      "step": 16290
+    },
+    {
+      "epoch": 3.26,
+      "grad_norm": 0.7792108058929443,
+      "learning_rate": 1.74e-05,
+      "loss": 0.0012,
+      "step": 16300
+    },
+    {
+      "epoch": 3.262,
+      "grad_norm": 0.034945737570524216,
+      "learning_rate": 1.7380000000000003e-05,
+      "loss": 0.0003,
+      "step": 16310
+    },
+    {
+      "epoch": 3.2640000000000002,
+      "grad_norm": 0.2685021162033081,
+      "learning_rate": 1.736e-05,
+      "loss": 0.0013,
+      "step": 16320
+    },
+    {
+      "epoch": 3.266,
+      "grad_norm": 0.0574602410197258,
+      "learning_rate": 1.734e-05,
+      "loss": 0.0005,
+      "step": 16330
+    },
+    {
+      "epoch": 3.268,
+      "grad_norm": 0.003966180607676506,
+      "learning_rate": 1.732e-05,
+      "loss": 0.0007,
+      "step": 16340
+    },
+    {
+      "epoch": 3.27,
+      "grad_norm": 4.431776523590088,
+      "learning_rate": 1.73e-05,
+      "loss": 0.004,
+      "step": 16350
+    },
+    {
+      "epoch": 3.2720000000000002,
+      "grad_norm": 0.0023266009520739317,
+      "learning_rate": 1.728e-05,
+      "loss": 0.0004,
+      "step": 16360
+    },
+    {
+      "epoch": 3.274,
+      "grad_norm": 0.006505758501589298,
+      "learning_rate": 1.726e-05,
+      "loss": 0.0014,
+      "step": 16370
+    },
+    {
+      "epoch": 3.276,
+      "grad_norm": 0.09441948682069778,
+      "learning_rate": 1.724e-05,
+      "loss": 0.0004,
+      "step": 16380
+    },
+    {
+      "epoch": 3.278,
+      "grad_norm": 0.06182579696178436,
+      "learning_rate": 1.722e-05,
+      "loss": 0.0117,
+      "step": 16390
+    },
+    {
+      "epoch": 3.2800000000000002,
+      "grad_norm": 7.060178756713867,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 0.0468,
+      "step": 16400
+    },
+    {
+      "epoch": 3.282,
+      "grad_norm": 0.00426541967317462,
+      "learning_rate": 1.718e-05,
+      "loss": 0.0009,
+      "step": 16410
+    },
+    {
+      "epoch": 3.284,
+      "grad_norm": 0.008011597208678722,
+      "learning_rate": 1.7160000000000002e-05,
+      "loss": 0.0005,
+      "step": 16420
+    },
+    {
+      "epoch": 3.286,
+      "grad_norm": 0.00453008059412241,
+      "learning_rate": 1.7140000000000002e-05,
+      "loss": 0.0006,
+      "step": 16430
+    },
+    {
+      "epoch": 3.288,
+      "grad_norm": 0.005792681127786636,
+      "learning_rate": 1.712e-05,
+      "loss": 0.0012,
+      "step": 16440
+    },
+    {
+      "epoch": 3.29,
+      "grad_norm": 8.664570808410645,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 0.0734,
+      "step": 16450
+    },
+    {
+      "epoch": 3.292,
+      "grad_norm": 0.02087589167058468,
+      "learning_rate": 1.7080000000000002e-05,
+      "loss": 0.0007,
+      "step": 16460
+    },
+    {
+      "epoch": 3.294,
+      "grad_norm": 0.01756032183766365,
+      "learning_rate": 1.706e-05,
+      "loss": 0.0337,
+      "step": 16470
+    },
+    {
+      "epoch": 3.296,
+      "grad_norm": 0.0026294682174921036,
+      "learning_rate": 1.704e-05,
+      "loss": 0.0353,
+      "step": 16480
+    },
+    {
+      "epoch": 3.298,
+      "grad_norm": 0.018317170441150665,
+      "learning_rate": 1.702e-05,
+      "loss": 0.0004,
+      "step": 16490
+    },
+    {
+      "epoch": 3.3,
+      "grad_norm": 0.022388644516468048,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 0.0175,
+      "step": 16500
+    },
+    {
+      "epoch": 3.302,
+      "grad_norm": 0.003864126745611429,
+      "learning_rate": 1.698e-05,
+      "loss": 0.0621,
+      "step": 16510
+    },
+    {
+      "epoch": 3.304,
+      "grad_norm": 0.0023028342984616756,
+      "learning_rate": 1.696e-05,
+      "loss": 0.0025,
+      "step": 16520
+    },
+    {
+      "epoch": 3.306,
+      "grad_norm": 0.021012943238019943,
+      "learning_rate": 1.694e-05,
+      "loss": 0.0332,
+      "step": 16530
+    },
+    {
+      "epoch": 3.308,
+      "grad_norm": 0.003342842683196068,
+      "learning_rate": 1.692e-05,
+      "loss": 0.0206,
+      "step": 16540
+    },
+    {
+      "epoch": 3.31,
+      "grad_norm": 0.0015880778664723039,
+      "learning_rate": 1.69e-05,
+      "loss": 0.0007,
+      "step": 16550
+    },
+    {
+      "epoch": 3.312,
+      "grad_norm": 0.002224528230726719,
+      "learning_rate": 1.688e-05,
+      "loss": 0.0057,
+      "step": 16560
+    },
+    {
+      "epoch": 3.314,
+      "grad_norm": 0.0382666289806366,
+      "learning_rate": 1.686e-05,
+      "loss": 0.0006,
+      "step": 16570
+    },
+    {
+      "epoch": 3.316,
+      "grad_norm": 3.435798168182373,
+      "learning_rate": 1.684e-05,
+      "loss": 0.0117,
+      "step": 16580
+    },
+    {
+      "epoch": 3.318,
+      "grad_norm": 0.009352126158773899,
+      "learning_rate": 1.6819999999999998e-05,
+      "loss": 0.0035,
+      "step": 16590
+    },
+    {
+      "epoch": 3.32,
+      "grad_norm": 0.00058965221978724,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 0.0511,
+      "step": 16600
+    },
+    {
+      "epoch": 3.322,
+      "grad_norm": 0.0026318279560655355,
+      "learning_rate": 1.6780000000000002e-05,
+      "loss": 0.0196,
+      "step": 16610
+    },
+    {
+      "epoch": 3.324,
+      "grad_norm": 0.011510156095027924,
+      "learning_rate": 1.6760000000000002e-05,
+      "loss": 0.0004,
+      "step": 16620
+    },
+    {
+      "epoch": 3.326,
+      "grad_norm": 0.008059993386268616,
+      "learning_rate": 1.674e-05,
+      "loss": 0.0006,
+      "step": 16630
+    },
+    {
+      "epoch": 3.328,
+      "grad_norm": 0.0023515725042670965,
+      "learning_rate": 1.672e-05,
+      "loss": 0.0015,
+      "step": 16640
+    },
+    {
+      "epoch": 3.33,
+      "grad_norm": 0.006723460275679827,
+      "learning_rate": 1.6700000000000003e-05,
+      "loss": 0.0005,
+      "step": 16650
+    },
+    {
+      "epoch": 3.332,
+      "grad_norm": 0.02267891727387905,
+      "learning_rate": 1.668e-05,
+      "loss": 0.0004,
+      "step": 16660
+    },
+    {
+      "epoch": 3.334,
+      "grad_norm": 0.04994427040219307,
+      "learning_rate": 1.666e-05,
+      "loss": 0.0664,
+      "step": 16670
+    },
+    {
+      "epoch": 3.336,
+      "grad_norm": 0.023461012169718742,
+      "learning_rate": 1.664e-05,
+      "loss": 0.0242,
+      "step": 16680
+    },
+    {
+      "epoch": 3.338,
+      "grad_norm": 0.006013381294906139,
+      "learning_rate": 1.662e-05,
+      "loss": 0.0421,
+      "step": 16690
+    },
+    {
+      "epoch": 3.34,
+      "grad_norm": 1.3734526634216309,
+      "learning_rate": 1.66e-05,
+      "loss": 0.0533,
+      "step": 16700
+    },
+    {
+      "epoch": 3.342,
+      "grad_norm": 0.006099446676671505,
+      "learning_rate": 1.658e-05,
+      "loss": 0.0682,
+      "step": 16710
+    },
+    {
+      "epoch": 3.344,
+      "grad_norm": 0.0032595868688076735,
+      "learning_rate": 1.656e-05,
+      "loss": 0.0126,
+      "step": 16720
+    },
+    {
+      "epoch": 3.346,
+      "grad_norm": 2.1808266639709473,
+      "learning_rate": 1.654e-05,
+      "loss": 0.0465,
+      "step": 16730
+    },
+    {
+      "epoch": 3.348,
+      "grad_norm": 0.019032666459679604,
+      "learning_rate": 1.652e-05,
+      "loss": 0.0009,
+      "step": 16740
+    },
+    {
+      "epoch": 3.35,
+      "grad_norm": 0.01130975503474474,
+      "learning_rate": 1.65e-05,
+      "loss": 0.013,
+      "step": 16750
+    },
+    {
+      "epoch": 3.352,
+      "grad_norm": 0.00591658428311348,
+      "learning_rate": 1.648e-05,
+      "loss": 0.0008,
+      "step": 16760
+    },
+    {
+      "epoch": 3.354,
+      "grad_norm": 0.0044304500333964825,
+      "learning_rate": 1.646e-05,
+      "loss": 0.0043,
+      "step": 16770
+    },
+    {
+      "epoch": 3.356,
+      "grad_norm": 0.15547245740890503,
+      "learning_rate": 1.644e-05,
+      "loss": 0.058,
+      "step": 16780
+    },
+    {
+      "epoch": 3.358,
+      "grad_norm": 0.021581124514341354,
+      "learning_rate": 1.6420000000000002e-05,
+      "loss": 0.0017,
+      "step": 16790
+    },
+    {
+      "epoch": 3.36,
+      "grad_norm": 0.06020371988415718,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 0.0276,
+      "step": 16800
+    },
+    {
+      "epoch": 3.362,
+      "grad_norm": 12.31493091583252,
+      "learning_rate": 1.6380000000000002e-05,
+      "loss": 0.0159,
+      "step": 16810
+    },
+    {
+      "epoch": 3.364,
+      "grad_norm": 0.016272541135549545,
+      "learning_rate": 1.636e-05,
+      "loss": 0.0587,
+      "step": 16820
+    },
+    {
+      "epoch": 3.366,
+      "grad_norm": 0.060270678251981735,
+      "learning_rate": 1.634e-05,
+      "loss": 0.0007,
+      "step": 16830
+    },
+    {
+      "epoch": 3.368,
+      "grad_norm": 0.03014090657234192,
+      "learning_rate": 1.6320000000000003e-05,
+      "loss": 0.0007,
+      "step": 16840
+    },
+    {
+      "epoch": 3.37,
+      "grad_norm": 0.012263710610568523,
+      "learning_rate": 1.63e-05,
+      "loss": 0.0442,
+      "step": 16850
+    },
+    {
+      "epoch": 3.372,
+      "grad_norm": 0.1727428138256073,
+      "learning_rate": 1.628e-05,
+      "loss": 0.0189,
+      "step": 16860
+    },
+    {
+      "epoch": 3.374,
+      "grad_norm": 0.016265198588371277,
+      "learning_rate": 1.626e-05,
+      "loss": 0.022,
+      "step": 16870
+    },
+    {
+      "epoch": 3.376,
+      "grad_norm": 0.0468362420797348,
+      "learning_rate": 1.624e-05,
+      "loss": 0.0008,
+      "step": 16880
+    },
+    {
+      "epoch": 3.378,
+      "grad_norm": 0.015331187285482883,
+      "learning_rate": 1.622e-05,
+      "loss": 0.0006,
+      "step": 16890
+    },
+    {
+      "epoch": 3.38,
+      "grad_norm": 0.05118807777762413,
+      "learning_rate": 1.62e-05,
+      "loss": 0.0023,
+      "step": 16900
+    },
+    {
+      "epoch": 3.382,
+      "grad_norm": 0.01582205668091774,
+      "learning_rate": 1.618e-05,
+      "loss": 0.0292,
+      "step": 16910
+    },
+    {
+      "epoch": 3.384,
+      "grad_norm": 2.682358503341675,
+      "learning_rate": 1.616e-05,
+      "loss": 0.0025,
+      "step": 16920
+    },
+    {
+      "epoch": 3.386,
+      "grad_norm": 0.013448044657707214,
+      "learning_rate": 1.6139999999999998e-05,
+      "loss": 0.0019,
+      "step": 16930
+    },
+    {
+      "epoch": 3.388,
+      "grad_norm": 0.005577258300036192,
+      "learning_rate": 1.612e-05,
+      "loss": 0.0007,
+      "step": 16940
+    },
+    {
+      "epoch": 3.39,
+      "grad_norm": 0.01986427791416645,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 0.0007,
+      "step": 16950
+    },
+    {
+      "epoch": 3.392,
+      "grad_norm": 0.1588340550661087,
+      "learning_rate": 1.6080000000000002e-05,
+      "loss": 0.0384,
+      "step": 16960
+    },
+    {
+      "epoch": 3.394,
+      "grad_norm": 0.0015054995892569423,
+      "learning_rate": 1.606e-05,
+      "loss": 0.0006,
+      "step": 16970
+    },
+    {
+      "epoch": 3.396,
+      "grad_norm": 0.012828778475522995,
+      "learning_rate": 1.604e-05,
+      "loss": 0.0016,
+      "step": 16980
+    },
+    {
+      "epoch": 3.398,
+      "grad_norm": 0.019603174179792404,
+      "learning_rate": 1.6020000000000002e-05,
+      "loss": 0.0003,
+      "step": 16990
+    },
+    {
+      "epoch": 3.4,
+      "grad_norm": 0.02867083251476288,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 0.0007,
+      "step": 17000
+    },
+    {
+      "epoch": 3.402,
+      "grad_norm": 0.026190217584371567,
+      "learning_rate": 1.598e-05,
+      "loss": 0.0013,
+      "step": 17010
+    },
+    {
+      "epoch": 3.404,
+      "grad_norm": 0.001176636666059494,
+      "learning_rate": 1.596e-05,
+      "loss": 0.002,
+      "step": 17020
+    },
+    {
+      "epoch": 3.406,
+      "grad_norm": 0.0011765063973143697,
+      "learning_rate": 1.594e-05,
+      "loss": 0.0004,
+      "step": 17030
+    },
+    {
+      "epoch": 3.408,
+      "grad_norm": 0.001738901250064373,
+      "learning_rate": 1.592e-05,
+      "loss": 0.0004,
+      "step": 17040
+    },
+    {
+      "epoch": 3.41,
+      "grad_norm": 0.007277410943061113,
+      "learning_rate": 1.59e-05,
+      "loss": 0.0003,
+      "step": 17050
+    },
+    {
+      "epoch": 3.412,
+      "grad_norm": 0.0007897934992797673,
+      "learning_rate": 1.588e-05,
+      "loss": 0.0099,
+      "step": 17060
+    },
+    {
+      "epoch": 3.414,
+      "grad_norm": 0.22555957734584808,
+      "learning_rate": 1.586e-05,
+      "loss": 0.1061,
+      "step": 17070
+    },
+    {
+      "epoch": 3.416,
+      "grad_norm": 0.005760278087109327,
+      "learning_rate": 1.584e-05,
+      "loss": 0.0169,
+      "step": 17080
+    },
+    {
+      "epoch": 3.418,
+      "grad_norm": 0.00394419115036726,
+      "learning_rate": 1.582e-05,
+      "loss": 0.0021,
+      "step": 17090
+    },
+    {
+      "epoch": 3.42,
+      "grad_norm": 0.008633914403617382,
+      "learning_rate": 1.58e-05,
+      "loss": 0.0012,
+      "step": 17100
+    },
+    {
+      "epoch": 3.422,
+      "grad_norm": 0.014838412404060364,
+      "learning_rate": 1.578e-05,
+      "loss": 0.027,
+      "step": 17110
+    },
+    {
+      "epoch": 3.424,
+      "grad_norm": 0.005310640204697847,
+      "learning_rate": 1.5759999999999998e-05,
+      "loss": 0.0007,
+      "step": 17120
+    },
+    {
+      "epoch": 3.426,
+      "grad_norm": 0.003743154229596257,
+      "learning_rate": 1.5740000000000002e-05,
+      "loss": 0.0004,
+      "step": 17130
+    },
+    {
+      "epoch": 3.428,
+      "grad_norm": 6.841172218322754,
+      "learning_rate": 1.5720000000000002e-05,
+      "loss": 0.0212,
+      "step": 17140
+    },
+    {
+      "epoch": 3.43,
+      "grad_norm": 0.0082596680149436,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 0.0029,
+      "step": 17150
+    },
+    {
+      "epoch": 3.432,
+      "grad_norm": 0.004711842630058527,
+      "learning_rate": 1.568e-05,
+      "loss": 0.0079,
+      "step": 17160
+    },
+    {
+      "epoch": 3.434,
+      "grad_norm": 0.0029079318046569824,
+      "learning_rate": 1.566e-05,
+      "loss": 0.0002,
+      "step": 17170
+    },
+    {
+      "epoch": 3.436,
+      "grad_norm": 0.0035103964619338512,
+      "learning_rate": 1.5640000000000003e-05,
+      "loss": 0.0006,
+      "step": 17180
+    },
+    {
+      "epoch": 3.438,
+      "grad_norm": 0.0038129487074911594,
+      "learning_rate": 1.5620000000000003e-05,
+      "loss": 0.0006,
+      "step": 17190
+    },
+    {
+      "epoch": 3.44,
+      "grad_norm": 0.025164583697915077,
+      "learning_rate": 1.56e-05,
+      "loss": 0.0003,
+      "step": 17200
+    },
+    {
+      "epoch": 3.442,
+      "grad_norm": 0.0032029796857386827,
+      "learning_rate": 1.558e-05,
+      "loss": 0.002,
+      "step": 17210
+    },
+    {
+      "epoch": 3.444,
+      "grad_norm": 0.008071298711001873,
+      "learning_rate": 1.556e-05,
+      "loss": 0.0092,
+      "step": 17220
+    },
+    {
+      "epoch": 3.446,
+      "grad_norm": 0.025147704407572746,
+      "learning_rate": 1.554e-05,
+      "loss": 0.0006,
+      "step": 17230
+    },
+    {
+      "epoch": 3.448,
+      "grad_norm": 0.0043593524023890495,
+      "learning_rate": 1.552e-05,
+      "loss": 0.0002,
+      "step": 17240
+    },
+    {
+      "epoch": 3.45,
+      "grad_norm": 0.0024096034467220306,
+      "learning_rate": 1.55e-05,
+      "loss": 0.024,
+      "step": 17250
+    },
+    {
+      "epoch": 3.452,
+      "grad_norm": 2.802278518676758,
+      "learning_rate": 1.548e-05,
+      "loss": 0.0362,
+      "step": 17260
+    },
+    {
+      "epoch": 3.454,
+      "grad_norm": 0.009577326476573944,
+      "learning_rate": 1.546e-05,
+      "loss": 0.0123,
+      "step": 17270
+    },
+    {
+      "epoch": 3.456,
+      "grad_norm": 0.0024973542895168066,
+      "learning_rate": 1.544e-05,
+      "loss": 0.0022,
+      "step": 17280
+    },
+    {
+      "epoch": 3.458,
+      "grad_norm": 0.003990876022726297,
+      "learning_rate": 1.542e-05,
+      "loss": 0.0678,
+      "step": 17290
+    },
+    {
+      "epoch": 3.46,
+      "grad_norm": 0.02418040670454502,
+      "learning_rate": 1.54e-05,
+      "loss": 0.0028,
+      "step": 17300
+    },
+    {
+      "epoch": 3.462,
+      "grad_norm": 0.002280684420838952,
+      "learning_rate": 1.538e-05,
+      "loss": 0.0012,
+      "step": 17310
+    },
+    {
+      "epoch": 3.464,
+      "grad_norm": 0.6858948469161987,
+      "learning_rate": 1.536e-05,
+      "loss": 0.0012,
+      "step": 17320
+    },
+    {
+      "epoch": 3.466,
+      "grad_norm": 0.00416453555226326,
+      "learning_rate": 1.5340000000000002e-05,
+      "loss": 0.0099,
+      "step": 17330
+    },
+    {
+      "epoch": 3.468,
+      "grad_norm": 8.234037399291992,
+      "learning_rate": 1.5320000000000002e-05,
+      "loss": 0.0406,
+      "step": 17340
+    },
+    {
+      "epoch": 3.4699999999999998,
+      "grad_norm": 0.011349360458552837,
+      "learning_rate": 1.53e-05,
+      "loss": 0.0003,
+      "step": 17350
+    },
+    {
+      "epoch": 3.472,
+      "grad_norm": 0.011367256753146648,
+      "learning_rate": 1.528e-05,
+      "loss": 0.0439,
+      "step": 17360
+    },
+    {
+      "epoch": 3.474,
+      "grad_norm": 0.006158503703773022,
+      "learning_rate": 1.5260000000000003e-05,
+      "loss": 0.0162,
+      "step": 17370
+    },
+    {
+      "epoch": 3.476,
+      "grad_norm": 0.003388832788914442,
+      "learning_rate": 1.5240000000000001e-05,
+      "loss": 0.0015,
+      "step": 17380
+    },
+    {
+      "epoch": 3.4779999999999998,
+      "grad_norm": 0.09154608845710754,
+      "learning_rate": 1.5220000000000002e-05,
+      "loss": 0.0595,
+      "step": 17390
+    },
+    {
+      "epoch": 3.48,
+      "grad_norm": 0.0007251560455188155,
+      "learning_rate": 1.52e-05,
+      "loss": 0.0015,
+      "step": 17400
+    },
+    {
+      "epoch": 3.482,
+      "grad_norm": 0.0014719640603289008,
+      "learning_rate": 1.518e-05,
+      "loss": 0.0272,
+      "step": 17410
+    },
+    {
+      "epoch": 3.484,
+      "grad_norm": 0.004138593561947346,
+      "learning_rate": 1.5160000000000002e-05,
+      "loss": 0.0039,
+      "step": 17420
+    },
+    {
+      "epoch": 3.4859999999999998,
+      "grad_norm": 0.002763948170468211,
+      "learning_rate": 1.514e-05,
+      "loss": 0.0447,
+      "step": 17430
+    },
+    {
+      "epoch": 3.488,
+      "grad_norm": 0.002828595694154501,
+      "learning_rate": 1.5120000000000001e-05,
+      "loss": 0.0004,
+      "step": 17440
+    },
+    {
+      "epoch": 3.49,
+      "grad_norm": 0.007722462061792612,
+      "learning_rate": 1.51e-05,
+      "loss": 0.0088,
+      "step": 17450
+    },
+    {
+      "epoch": 3.492,
+      "grad_norm": 0.024451201781630516,
+      "learning_rate": 1.508e-05,
+      "loss": 0.0034,
+      "step": 17460
+    },
+    {
+      "epoch": 3.4939999999999998,
+      "grad_norm": 0.01881391368806362,
+      "learning_rate": 1.5060000000000001e-05,
+      "loss": 0.0377,
+      "step": 17470
+    },
+    {
+      "epoch": 3.496,
+      "grad_norm": 0.004509866703301668,
+      "learning_rate": 1.5040000000000002e-05,
+      "loss": 0.0004,
+      "step": 17480
+    },
+    {
+      "epoch": 3.498,
+      "grad_norm": 0.09744896739721298,
+      "learning_rate": 1.502e-05,
+      "loss": 0.0116,
+      "step": 17490
+    },
+    {
+      "epoch": 3.5,
+      "grad_norm": 0.9149200320243835,
+      "learning_rate": 1.5e-05,
+      "loss": 0.0302,
+      "step": 17500
+    },
+    {
+      "epoch": 3.502,
+      "grad_norm": 0.014574715867638588,
+      "learning_rate": 1.4979999999999999e-05,
+      "loss": 0.0005,
+      "step": 17510
+    },
+    {
+      "epoch": 3.504,
+      "grad_norm": 0.0013671967899426818,
+      "learning_rate": 1.4960000000000002e-05,
+      "loss": 0.0004,
+      "step": 17520
+    },
+    {
+      "epoch": 3.5060000000000002,
+      "grad_norm": 0.044884104281663895,
+      "learning_rate": 1.4940000000000001e-05,
+      "loss": 0.0034,
+      "step": 17530
+    },
+    {
+      "epoch": 3.508,
+      "grad_norm": 0.002155133057385683,
+      "learning_rate": 1.4920000000000001e-05,
+      "loss": 0.0251,
+      "step": 17540
+    },
+    {
+      "epoch": 3.51,
+      "grad_norm": 0.005239092279225588,
+      "learning_rate": 1.49e-05,
+      "loss": 0.0031,
+      "step": 17550
+    },
+    {
+      "epoch": 3.512,
+      "grad_norm": 0.004131457768380642,
+      "learning_rate": 1.488e-05,
+      "loss": 0.0315,
+      "step": 17560
+    },
+    {
+      "epoch": 3.5140000000000002,
+      "grad_norm": 0.0030689926352351904,
+      "learning_rate": 1.4860000000000002e-05,
+      "loss": 0.0001,
+      "step": 17570
+    },
+    {
+      "epoch": 3.516,
+      "grad_norm": 0.02878253161907196,
+      "learning_rate": 1.4840000000000002e-05,
+      "loss": 0.0379,
+      "step": 17580
+    },
+    {
+      "epoch": 3.518,
+      "grad_norm": 0.011847962625324726,
+      "learning_rate": 1.482e-05,
+      "loss": 0.0004,
+      "step": 17590
+    },
+    {
+      "epoch": 3.52,
+      "grad_norm": 0.00936802290380001,
+      "learning_rate": 1.48e-05,
+      "loss": 0.0144,
+      "step": 17600
+    },
+    {
+      "epoch": 3.5220000000000002,
+      "grad_norm": 0.0026822967920452356,
+      "learning_rate": 1.4779999999999999e-05,
+      "loss": 0.0002,
+      "step": 17610
+    },
+    {
+      "epoch": 3.524,
+      "grad_norm": 0.03320813924074173,
+      "learning_rate": 1.4760000000000001e-05,
+      "loss": 0.0166,
+      "step": 17620
+    },
+    {
+      "epoch": 3.526,
+      "grad_norm": 0.0031809851061552763,
+      "learning_rate": 1.4740000000000001e-05,
+      "loss": 0.0004,
+      "step": 17630
+    },
+    {
+      "epoch": 3.528,
+      "grad_norm": 0.0020093093626201153,
+      "learning_rate": 1.472e-05,
+      "loss": 0.0374,
+      "step": 17640
+    },
+    {
+      "epoch": 3.5300000000000002,
+      "grad_norm": 0.013656358234584332,
+      "learning_rate": 1.47e-05,
+      "loss": 0.0003,
+      "step": 17650
+    },
+    {
+      "epoch": 3.532,
+      "grad_norm": 0.0178862102329731,
+      "learning_rate": 1.4680000000000002e-05,
+      "loss": 0.0002,
+      "step": 17660
+    },
+    {
+      "epoch": 3.534,
+      "grad_norm": 0.010191562585532665,
+      "learning_rate": 1.4660000000000002e-05,
+      "loss": 0.0513,
+      "step": 17670
+    },
+    {
+      "epoch": 3.536,
+      "grad_norm": 0.0026134855579584837,
+      "learning_rate": 1.464e-05,
+      "loss": 0.0464,
+      "step": 17680
+    },
+    {
+      "epoch": 3.5380000000000003,
+      "grad_norm": 0.009755929931998253,
+      "learning_rate": 1.462e-05,
+      "loss": 0.0054,
+      "step": 17690
+    },
+    {
+      "epoch": 3.54,
+      "grad_norm": 0.015786219388246536,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 0.0214,
+      "step": 17700
+    },
+    {
+      "epoch": 3.542,
+      "grad_norm": 0.005827093962579966,
+      "learning_rate": 1.4580000000000003e-05,
+      "loss": 0.0005,
+      "step": 17710
+    },
+    {
+      "epoch": 3.544,
+      "grad_norm": 0.004923023749142885,
+      "learning_rate": 1.4560000000000001e-05,
+      "loss": 0.0279,
+      "step": 17720
+    },
+    {
+      "epoch": 3.5460000000000003,
+      "grad_norm": 0.002227694494649768,
+      "learning_rate": 1.4540000000000001e-05,
+      "loss": 0.0039,
+      "step": 17730
+    },
+    {
+      "epoch": 3.548,
+      "grad_norm": 0.0029734703712165356,
+      "learning_rate": 1.452e-05,
+      "loss": 0.0004,
+      "step": 17740
+    },
+    {
+      "epoch": 3.55,
+      "grad_norm": 4.266994476318359,
+      "learning_rate": 1.45e-05,
+      "loss": 0.087,
+      "step": 17750
+    },
+    {
+      "epoch": 3.552,
+      "grad_norm": 0.007924766279757023,
+      "learning_rate": 1.4480000000000002e-05,
+      "loss": 0.0007,
+      "step": 17760
+    },
+    {
+      "epoch": 3.5540000000000003,
+      "grad_norm": 0.00312436418607831,
+      "learning_rate": 1.4460000000000002e-05,
+      "loss": 0.0108,
+      "step": 17770
+    },
+    {
+      "epoch": 3.556,
+      "grad_norm": 0.004613260738551617,
+      "learning_rate": 1.444e-05,
+      "loss": 0.0265,
+      "step": 17780
+    },
+    {
+      "epoch": 3.558,
+      "grad_norm": 0.014302493073046207,
+      "learning_rate": 1.4420000000000001e-05,
+      "loss": 0.0004,
+      "step": 17790
+    },
+    {
+      "epoch": 3.56,
+      "grad_norm": 0.024875864386558533,
+      "learning_rate": 1.44e-05,
+      "loss": 0.0303,
+      "step": 17800
+    },
+    {
+      "epoch": 3.5620000000000003,
+      "grad_norm": 0.04414631798863411,
+      "learning_rate": 1.4380000000000001e-05,
+      "loss": 0.0074,
+      "step": 17810
+    },
+    {
+      "epoch": 3.564,
+      "grad_norm": 0.03683282062411308,
+      "learning_rate": 1.4360000000000001e-05,
+      "loss": 0.0457,
+      "step": 17820
+    },
+    {
+      "epoch": 3.566,
+      "grad_norm": 0.021348169073462486,
+      "learning_rate": 1.434e-05,
+      "loss": 0.0018,
+      "step": 17830
+    },
+    {
+      "epoch": 3.568,
+      "grad_norm": 0.002948606386780739,
+      "learning_rate": 1.432e-05,
+      "loss": 0.0255,
+      "step": 17840
+    },
+    {
+      "epoch": 3.57,
+      "grad_norm": 0.7394766211509705,
+      "learning_rate": 1.43e-05,
+      "loss": 0.001,
+      "step": 17850
+    },
+    {
+      "epoch": 3.572,
+      "grad_norm": 0.14344385266304016,
+      "learning_rate": 1.4280000000000002e-05,
+      "loss": 0.0008,
+      "step": 17860
+    },
+    {
+      "epoch": 3.574,
+      "grad_norm": 0.007859619334340096,
+      "learning_rate": 1.426e-05,
+      "loss": 0.0007,
+      "step": 17870
+    },
+    {
+      "epoch": 3.576,
+      "grad_norm": 0.003310305532068014,
+      "learning_rate": 1.4240000000000001e-05,
+      "loss": 0.0005,
+      "step": 17880
+    },
+    {
+      "epoch": 3.578,
+      "grad_norm": 0.011208192445337772,
+      "learning_rate": 1.422e-05,
+      "loss": 0.0007,
+      "step": 17890
+    },
+    {
+      "epoch": 3.58,
+      "grad_norm": 0.0034019567538052797,
+      "learning_rate": 1.42e-05,
+      "loss": 0.0006,
+      "step": 17900
+    },
+    {
+      "epoch": 3.582,
+      "grad_norm": 1.4686945676803589,
+      "learning_rate": 1.4180000000000001e-05,
+      "loss": 0.002,
+      "step": 17910
+    },
+    {
+      "epoch": 3.584,
+      "grad_norm": 0.0022530239075422287,
+      "learning_rate": 1.4160000000000002e-05,
+      "loss": 0.0004,
+      "step": 17920
+    },
+    {
+      "epoch": 3.586,
+      "grad_norm": 0.003713824087753892,
+      "learning_rate": 1.414e-05,
+      "loss": 0.0262,
+      "step": 17930
+    },
+    {
+      "epoch": 3.588,
+      "grad_norm": 0.003505872329697013,
+      "learning_rate": 1.412e-05,
+      "loss": 0.0012,
+      "step": 17940
+    },
+    {
+      "epoch": 3.59,
+      "grad_norm": 0.012412111274898052,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 0.0004,
+      "step": 17950
+    },
+    {
+      "epoch": 3.592,
+      "grad_norm": 0.0049460954032838345,
+      "learning_rate": 1.408e-05,
+      "loss": 0.0231,
+      "step": 17960
+    },
+    {
+      "epoch": 3.594,
+      "grad_norm": 0.007156440056860447,
+      "learning_rate": 1.4060000000000001e-05,
+      "loss": 0.0004,
+      "step": 17970
+    },
+    {
+      "epoch": 3.596,
+      "grad_norm": 0.27144309878349304,
+      "learning_rate": 1.4040000000000001e-05,
+      "loss": 0.0196,
+      "step": 17980
+    },
+    {
+      "epoch": 3.598,
+      "grad_norm": 0.025920584797859192,
+      "learning_rate": 1.402e-05,
+      "loss": 0.0036,
+      "step": 17990
+    },
+    {
+      "epoch": 3.6,
+      "grad_norm": 0.005095527973026037,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 0.012,
+      "step": 18000
+    },
+    {
+      "epoch": 3.602,
+      "grad_norm": 0.0032979496754705906,
+      "learning_rate": 1.3980000000000002e-05,
+      "loss": 0.0073,
+      "step": 18010
+    },
+    {
+      "epoch": 3.604,
+      "grad_norm": 0.0015533601399511099,
+      "learning_rate": 1.396e-05,
+      "loss": 0.0817,
+      "step": 18020
+    },
+    {
+      "epoch": 3.606,
+      "grad_norm": 0.005558553151786327,
+      "learning_rate": 1.394e-05,
+      "loss": 0.0022,
+      "step": 18030
+    },
+    {
+      "epoch": 3.608,
+      "grad_norm": 0.01786387525498867,
+      "learning_rate": 1.3919999999999999e-05,
+      "loss": 0.0003,
+      "step": 18040
+    },
+    {
+      "epoch": 3.61,
+      "grad_norm": 0.006026718765497208,
+      "learning_rate": 1.3900000000000002e-05,
+      "loss": 0.0007,
+      "step": 18050
+    },
+    {
+      "epoch": 3.612,
+      "grad_norm": 0.008753432892262936,
+      "learning_rate": 1.3880000000000001e-05,
+      "loss": 0.0007,
+      "step": 18060
+    },
+    {
+      "epoch": 3.614,
+      "grad_norm": 0.012564820237457752,
+      "learning_rate": 1.3860000000000001e-05,
+      "loss": 0.0007,
+      "step": 18070
+    },
+    {
+      "epoch": 3.616,
+      "grad_norm": 0.03756599500775337,
+      "learning_rate": 1.384e-05,
+      "loss": 0.0011,
+      "step": 18080
+    },
+    {
+      "epoch": 3.618,
+      "grad_norm": 0.002842110814526677,
+      "learning_rate": 1.382e-05,
+      "loss": 0.0256,
+      "step": 18090
+    },
+    {
+      "epoch": 3.62,
+      "grad_norm": 3.56378173828125,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 0.0338,
+      "step": 18100
+    },
+    {
+      "epoch": 3.622,
+      "grad_norm": 0.005345380399376154,
+      "learning_rate": 1.3780000000000002e-05,
+      "loss": 0.0002,
+      "step": 18110
+    },
+    {
+      "epoch": 3.624,
+      "grad_norm": 0.004506293218582869,
+      "learning_rate": 1.376e-05,
+      "loss": 0.0003,
+      "step": 18120
+    },
+    {
+      "epoch": 3.626,
+      "grad_norm": 3.007185697555542,
+      "learning_rate": 1.374e-05,
+      "loss": 0.0373,
+      "step": 18130
+    },
+    {
+      "epoch": 3.628,
+      "grad_norm": 0.027455611154437065,
+      "learning_rate": 1.3719999999999999e-05,
+      "loss": 0.0003,
+      "step": 18140
+    },
+    {
+      "epoch": 3.63,
+      "grad_norm": 0.004361871164292097,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 0.0761,
+      "step": 18150
+    },
+    {
+      "epoch": 3.632,
+      "grad_norm": 0.49487006664276123,
+      "learning_rate": 1.3680000000000001e-05,
+      "loss": 0.0009,
+      "step": 18160
+    },
+    {
+      "epoch": 3.634,
+      "grad_norm": 0.0041132434271276,
+      "learning_rate": 1.3660000000000001e-05,
+      "loss": 0.0012,
+      "step": 18170
+    },
+    {
+      "epoch": 3.636,
+      "grad_norm": 0.007894878275692463,
+      "learning_rate": 1.364e-05,
+      "loss": 0.0035,
+      "step": 18180
+    },
+    {
+      "epoch": 3.638,
+      "grad_norm": 0.003962747287005186,
+      "learning_rate": 1.362e-05,
+      "loss": 0.0014,
+      "step": 18190
+    },
+    {
+      "epoch": 3.64,
+      "grad_norm": 0.007062731776386499,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 0.0007,
+      "step": 18200
+    },
+    {
+      "epoch": 3.642,
+      "grad_norm": 9.764742851257324,
+      "learning_rate": 1.358e-05,
+      "loss": 0.0567,
+      "step": 18210
+    },
+    {
+      "epoch": 3.644,
+      "grad_norm": 0.01080237701535225,
+      "learning_rate": 1.356e-05,
+      "loss": 0.0033,
+      "step": 18220
+    },
+    {
+      "epoch": 3.646,
+      "grad_norm": 0.0007170343305915594,
+      "learning_rate": 1.3539999999999999e-05,
+      "loss": 0.0006,
+      "step": 18230
+    },
+    {
+      "epoch": 3.648,
+      "grad_norm": 6.265512466430664,
+      "learning_rate": 1.352e-05,
+      "loss": 0.0549,
+      "step": 18240
+    },
+    {
+      "epoch": 3.65,
+      "grad_norm": 0.11348658055067062,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 0.0355,
+      "step": 18250
+    },
+    {
+      "epoch": 3.652,
+      "grad_norm": 3.7011401653289795,
+      "learning_rate": 1.3480000000000001e-05,
+      "loss": 0.052,
+      "step": 18260
+    },
+    {
+      "epoch": 3.654,
+      "grad_norm": 0.004853812512010336,
+      "learning_rate": 1.346e-05,
+      "loss": 0.0002,
+      "step": 18270
+    },
+    {
+      "epoch": 3.656,
+      "grad_norm": 0.0019084862433373928,
+      "learning_rate": 1.344e-05,
+      "loss": 0.0063,
+      "step": 18280
+    },
+    {
+      "epoch": 3.658,
+      "grad_norm": 0.002195891458541155,
+      "learning_rate": 1.3420000000000002e-05,
+      "loss": 0.0062,
+      "step": 18290
+    },
+    {
+      "epoch": 3.66,
+      "grad_norm": 0.028412865474820137,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 0.0125,
+      "step": 18300
+    },
+    {
+      "epoch": 3.662,
+      "grad_norm": 0.02995235100388527,
+      "learning_rate": 1.338e-05,
+      "loss": 0.0308,
+      "step": 18310
+    },
+    {
+      "epoch": 3.664,
+      "grad_norm": 0.05932564288377762,
+      "learning_rate": 1.336e-05,
+      "loss": 0.0178,
+      "step": 18320
+    },
+    {
+      "epoch": 3.666,
+      "grad_norm": 4.091817855834961,
+      "learning_rate": 1.334e-05,
+      "loss": 0.0817,
+      "step": 18330
+    },
+    {
+      "epoch": 3.668,
+      "grad_norm": 0.007290849462151527,
+      "learning_rate": 1.3320000000000001e-05,
+      "loss": 0.0351,
+      "step": 18340
+    },
+    {
+      "epoch": 3.67,
+      "grad_norm": 0.0027278633788228035,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 0.0011,
+      "step": 18350
+    },
+    {
+      "epoch": 3.672,
+      "grad_norm": 0.010340756736695766,
+      "learning_rate": 1.3280000000000002e-05,
+      "loss": 0.0014,
+      "step": 18360
+    },
+    {
+      "epoch": 3.674,
+      "grad_norm": 0.001625576289370656,
+      "learning_rate": 1.326e-05,
+      "loss": 0.0005,
+      "step": 18370
+    },
+    {
+      "epoch": 3.676,
+      "grad_norm": 0.013287074863910675,
+      "learning_rate": 1.324e-05,
+      "loss": 0.0008,
+      "step": 18380
+    },
+    {
+      "epoch": 3.678,
+      "grad_norm": 0.05007312819361687,
+      "learning_rate": 1.3220000000000002e-05,
+      "loss": 0.0008,
+      "step": 18390
+    },
+    {
+      "epoch": 3.68,
+      "grad_norm": 0.050255417823791504,
+      "learning_rate": 1.32e-05,
+      "loss": 0.0091,
+      "step": 18400
+    },
+    {
+      "epoch": 3.682,
+      "grad_norm": 0.004707262851297855,
+      "learning_rate": 1.3180000000000001e-05,
+      "loss": 0.0009,
+      "step": 18410
+    },
+    {
+      "epoch": 3.684,
+      "grad_norm": 0.0015249394346028566,
+      "learning_rate": 1.316e-05,
+      "loss": 0.0296,
+      "step": 18420
+    },
+    {
+      "epoch": 3.686,
+      "grad_norm": 0.06252807378768921,
+      "learning_rate": 1.314e-05,
+      "loss": 0.0059,
+      "step": 18430
+    },
+    {
+      "epoch": 3.6879999999999997,
+      "grad_norm": 0.02390551008284092,
+      "learning_rate": 1.3120000000000001e-05,
+      "loss": 0.0157,
+      "step": 18440
+    },
+    {
+      "epoch": 3.69,
+      "grad_norm": 0.009191851131618023,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 0.033,
+      "step": 18450
+    },
+    {
+      "epoch": 3.692,
+      "grad_norm": 0.020053191110491753,
+      "learning_rate": 1.308e-05,
+      "loss": 0.002,
+      "step": 18460
+    },
+    {
+      "epoch": 3.694,
+      "grad_norm": 0.0027913597878068686,
+      "learning_rate": 1.306e-05,
+      "loss": 0.0034,
+      "step": 18470
+    },
+    {
+      "epoch": 3.6959999999999997,
+      "grad_norm": 5.262605667114258,
+      "learning_rate": 1.3039999999999999e-05,
+      "loss": 0.0049,
+      "step": 18480
+    },
+    {
+      "epoch": 3.698,
+      "grad_norm": 0.0016802336322143674,
+      "learning_rate": 1.3020000000000002e-05,
+      "loss": 0.0023,
+      "step": 18490
+    },
+    {
+      "epoch": 3.7,
+      "grad_norm": 0.002672248985618353,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 0.0013,
+      "step": 18500
+    },
+    {
+      "epoch": 3.702,
+      "grad_norm": 0.0664878785610199,
+      "learning_rate": 1.2980000000000001e-05,
+      "loss": 0.0016,
+      "step": 18510
+    },
+    {
+      "epoch": 3.7039999999999997,
+      "grad_norm": 3.239924192428589,
+      "learning_rate": 1.296e-05,
+      "loss": 0.0425,
+      "step": 18520
+    },
+    {
+      "epoch": 3.706,
+      "grad_norm": 0.0018718920182436705,
+      "learning_rate": 1.294e-05,
+      "loss": 0.0198,
+      "step": 18530
+    },
+    {
+      "epoch": 3.708,
+      "grad_norm": 0.01981085166335106,
+      "learning_rate": 1.2920000000000002e-05,
+      "loss": 0.0013,
+      "step": 18540
+    },
+    {
+      "epoch": 3.71,
+      "grad_norm": 0.0019104615785181522,
+      "learning_rate": 1.29e-05,
+      "loss": 0.0006,
+      "step": 18550
+    },
+    {
+      "epoch": 3.7119999999999997,
+      "grad_norm": 0.05587019398808479,
+      "learning_rate": 1.288e-05,
+      "loss": 0.1196,
+      "step": 18560
+    },
+    {
+      "epoch": 3.714,
+      "grad_norm": 0.14568588137626648,
+      "learning_rate": 1.286e-05,
+      "loss": 0.0008,
+      "step": 18570
+    },
+    {
+      "epoch": 3.716,
+      "grad_norm": 0.008522302843630314,
+      "learning_rate": 1.2839999999999999e-05,
+      "loss": 0.0346,
+      "step": 18580
+    },
+    {
+      "epoch": 3.718,
+      "grad_norm": 0.0016074853483587503,
+      "learning_rate": 1.2820000000000001e-05,
+      "loss": 0.01,
+      "step": 18590
+    },
+    {
+      "epoch": 3.7199999999999998,
+      "grad_norm": 0.021763231605291367,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 0.0312,
+      "step": 18600
+    },
+    {
+      "epoch": 3.722,
+      "grad_norm": 0.1613340526819229,
+      "learning_rate": 1.278e-05,
+      "loss": 0.0035,
+      "step": 18610
+    },
+    {
+      "epoch": 3.724,
+      "grad_norm": 0.03853601589798927,
+      "learning_rate": 1.276e-05,
+      "loss": 0.0007,
+      "step": 18620
+    },
+    {
+      "epoch": 3.726,
+      "grad_norm": 0.012178819626569748,
+      "learning_rate": 1.2740000000000002e-05,
+      "loss": 0.0027,
+      "step": 18630
+    },
+    {
+      "epoch": 3.7279999999999998,
+      "grad_norm": 0.009596653282642365,
+      "learning_rate": 1.2720000000000002e-05,
+      "loss": 0.0153,
+      "step": 18640
+    },
+    {
+      "epoch": 3.73,
+      "grad_norm": 0.030010340735316277,
+      "learning_rate": 1.27e-05,
+      "loss": 0.0039,
+      "step": 18650
+    },
+    {
+      "epoch": 3.732,
+      "grad_norm": 0.005089831538498402,
+      "learning_rate": 1.268e-05,
+      "loss": 0.0002,
+      "step": 18660
+    },
+    {
+      "epoch": 3.734,
+      "grad_norm": 0.018710069358348846,
+      "learning_rate": 1.2659999999999999e-05,
+      "loss": 0.0272,
+      "step": 18670
+    },
+    {
+      "epoch": 3.7359999999999998,
+      "grad_norm": 0.0020607642363756895,
+      "learning_rate": 1.2640000000000003e-05,
+      "loss": 0.0003,
+      "step": 18680
+    },
+    {
+      "epoch": 3.738,
+      "grad_norm": 0.0018226850079372525,
+      "learning_rate": 1.2620000000000001e-05,
+      "loss": 0.0004,
+      "step": 18690
+    },
+    {
+      "epoch": 3.74,
+      "grad_norm": 0.0032087010331451893,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 0.0067,
+      "step": 18700
+    },
+    {
+      "epoch": 3.742,
+      "grad_norm": 0.002074528718367219,
+      "learning_rate": 1.258e-05,
+      "loss": 0.0002,
+      "step": 18710
+    },
+    {
+      "epoch": 3.7439999999999998,
+      "grad_norm": 0.009091685526072979,
+      "learning_rate": 1.256e-05,
+      "loss": 0.0001,
+      "step": 18720
+    },
+    {
+      "epoch": 3.746,
+      "grad_norm": 0.0004810831160284579,
+      "learning_rate": 1.2540000000000002e-05,
+      "loss": 0.0007,
+      "step": 18730
+    },
+    {
+      "epoch": 3.748,
+      "grad_norm": 0.010521749965846539,
+      "learning_rate": 1.252e-05,
+      "loss": 0.0009,
+      "step": 18740
+    },
+    {
+      "epoch": 3.75,
+      "grad_norm": 5.149264335632324,
+      "learning_rate": 1.25e-05,
+      "loss": 0.1288,
+      "step": 18750
+    },
+    {
+      "epoch": 3.752,
+      "grad_norm": 0.005692175589501858,
+      "learning_rate": 1.248e-05,
+      "loss": 0.0241,
+      "step": 18760
+    },
+    {
+      "epoch": 3.754,
+      "grad_norm": 0.04272817447781563,
+      "learning_rate": 1.2460000000000001e-05,
+      "loss": 0.0011,
+      "step": 18770
+    },
+    {
+      "epoch": 3.7560000000000002,
+      "grad_norm": 4.0740203857421875,
+      "learning_rate": 1.244e-05,
+      "loss": 0.0046,
+      "step": 18780
+    },
+    {
+      "epoch": 3.758,
+      "grad_norm": 0.020532237365841866,
+      "learning_rate": 1.2420000000000001e-05,
+      "loss": 0.0007,
+      "step": 18790
+    },
+    {
+      "epoch": 3.76,
+      "grad_norm": 0.003703763009980321,
+      "learning_rate": 1.24e-05,
+      "loss": 0.0294,
+      "step": 18800
+    },
+    {
+      "epoch": 3.762,
+      "grad_norm": 0.008622046560049057,
+      "learning_rate": 1.238e-05,
+      "loss": 0.0198,
+      "step": 18810
+    },
+    {
+      "epoch": 3.7640000000000002,
+      "grad_norm": 0.010497739538550377,
+      "learning_rate": 1.236e-05,
+      "loss": 0.0173,
+      "step": 18820
+    },
+    {
+      "epoch": 3.766,
+      "grad_norm": 0.0006842405418865383,
+      "learning_rate": 1.234e-05,
+      "loss": 0.0003,
+      "step": 18830
+    },
+    {
+      "epoch": 3.768,
+      "grad_norm": 0.0067459153942763805,
+      "learning_rate": 1.232e-05,
+      "loss": 0.001,
+      "step": 18840
+    },
+    {
+      "epoch": 3.77,
+      "grad_norm": 0.00805757287889719,
+      "learning_rate": 1.23e-05,
+      "loss": 0.0009,
+      "step": 18850
+    },
+    {
+      "epoch": 3.7720000000000002,
+      "grad_norm": 0.0038732087705284357,
+      "learning_rate": 1.2280000000000001e-05,
+      "loss": 0.0073,
+      "step": 18860
+    },
+    {
+      "epoch": 3.774,
+      "grad_norm": 0.005831129848957062,
+      "learning_rate": 1.2260000000000001e-05,
+      "loss": 0.0002,
+      "step": 18870
+    },
+    {
+      "epoch": 3.776,
+      "grad_norm": 0.004823405761271715,
+      "learning_rate": 1.224e-05,
+      "loss": 0.0017,
+      "step": 18880
+    },
+    {
+      "epoch": 3.778,
+      "grad_norm": 0.0019176382338628173,
+      "learning_rate": 1.2220000000000002e-05,
+      "loss": 0.0088,
+      "step": 18890
+    },
+    {
+      "epoch": 3.7800000000000002,
+      "grad_norm": 0.018935251981019974,
+      "learning_rate": 1.22e-05,
+      "loss": 0.0021,
+      "step": 18900
+    },
+    {
+      "epoch": 3.782,
+      "grad_norm": 0.010264934971928596,
+      "learning_rate": 1.2180000000000002e-05,
+      "loss": 0.067,
+      "step": 18910
+    },
+    {
+      "epoch": 3.784,
+      "grad_norm": 0.024122945964336395,
+      "learning_rate": 1.216e-05,
+      "loss": 0.0151,
+      "step": 18920
+    },
+    {
+      "epoch": 3.786,
+      "grad_norm": 0.46055740118026733,
+      "learning_rate": 1.214e-05,
+      "loss": 0.0005,
+      "step": 18930
+    },
+    {
+      "epoch": 3.7880000000000003,
+      "grad_norm": 0.004659720230847597,
+      "learning_rate": 1.2120000000000001e-05,
+      "loss": 0.0003,
+      "step": 18940
+    },
+    {
+      "epoch": 3.79,
+      "grad_norm": 0.11232586205005646,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 0.0011,
+      "step": 18950
+    },
+    {
+      "epoch": 3.792,
+      "grad_norm": 0.0011392510496079922,
+      "learning_rate": 1.2080000000000001e-05,
+      "loss": 0.0003,
+      "step": 18960
+    },
+    {
+      "epoch": 3.794,
+      "grad_norm": 0.0054550268687307835,
+      "learning_rate": 1.206e-05,
+      "loss": 0.0203,
+      "step": 18970
+    },
+    {
+      "epoch": 3.7960000000000003,
+      "grad_norm": 0.035335637629032135,
+      "learning_rate": 1.204e-05,
+      "loss": 0.0005,
+      "step": 18980
+    },
+    {
+      "epoch": 3.798,
+      "grad_norm": 0.0006725691491737962,
+      "learning_rate": 1.202e-05,
+      "loss": 0.014,
+      "step": 18990
+    },
+    {
+      "epoch": 3.8,
+      "grad_norm": 0.006045354995876551,
+      "learning_rate": 1.2e-05,
+      "loss": 0.0102,
+      "step": 19000
+    },
+    {
+      "epoch": 3.802,
+      "grad_norm": 0.0034727011807262897,
+      "learning_rate": 1.198e-05,
+      "loss": 0.0035,
+      "step": 19010
+    },
+    {
+      "epoch": 3.8040000000000003,
+      "grad_norm": 0.006682442035526037,
+      "learning_rate": 1.196e-05,
+      "loss": 0.0004,
+      "step": 19020
+    },
+    {
+      "epoch": 3.806,
+      "grad_norm": 3.7649495601654053,
+      "learning_rate": 1.1940000000000001e-05,
+      "loss": 0.0034,
+      "step": 19030
+    },
+    {
+      "epoch": 3.808,
+      "grad_norm": 0.0046861134469509125,
+      "learning_rate": 1.1920000000000001e-05,
+      "loss": 0.0209,
+      "step": 19040
+    },
+    {
+      "epoch": 3.81,
+      "grad_norm": 0.0016382812755182385,
+      "learning_rate": 1.19e-05,
+      "loss": 0.0277,
+      "step": 19050
+    },
+    {
+      "epoch": 3.8120000000000003,
+      "grad_norm": 0.104917511343956,
+      "learning_rate": 1.1880000000000001e-05,
+      "loss": 0.0132,
+      "step": 19060
+    },
+    {
+      "epoch": 3.814,
+      "grad_norm": 6.915656089782715,
+      "learning_rate": 1.186e-05,
+      "loss": 0.0171,
+      "step": 19070
+    },
+    {
+      "epoch": 3.816,
+      "grad_norm": 0.00960025005042553,
+      "learning_rate": 1.1840000000000002e-05,
+      "loss": 0.0004,
+      "step": 19080
+    },
+    {
+      "epoch": 3.818,
+      "grad_norm": 0.004476846661418676,
+      "learning_rate": 1.182e-05,
+      "loss": 0.0487,
+      "step": 19090
+    },
+    {
+      "epoch": 3.82,
+      "grad_norm": 0.006105990614742041,
+      "learning_rate": 1.18e-05,
+      "loss": 0.0003,
+      "step": 19100
+    },
+    {
+      "epoch": 3.822,
+      "grad_norm": 1.390354871749878,
+      "learning_rate": 1.178e-05,
+      "loss": 0.0155,
+      "step": 19110
+    },
+    {
+      "epoch": 3.824,
+      "grad_norm": 5.109686851501465,
+      "learning_rate": 1.1760000000000001e-05,
+      "loss": 0.0233,
+      "step": 19120
+    },
+    {
+      "epoch": 3.826,
+      "grad_norm": 0.16702832281589508,
+      "learning_rate": 1.1740000000000001e-05,
+      "loss": 0.0324,
+      "step": 19130
+    },
+    {
+      "epoch": 3.828,
+      "grad_norm": 0.02535170502960682,
+      "learning_rate": 1.172e-05,
+      "loss": 0.0244,
+      "step": 19140
+    },
+    {
+      "epoch": 3.83,
+      "grad_norm": 0.0075645833276212215,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 0.0002,
+      "step": 19150
+    },
+    {
+      "epoch": 3.832,
+      "grad_norm": 0.029271816834807396,
+      "learning_rate": 1.168e-05,
+      "loss": 0.0003,
+      "step": 19160
+    },
+    {
+      "epoch": 3.834,
+      "grad_norm": 0.04714861512184143,
+      "learning_rate": 1.166e-05,
+      "loss": 0.0015,
+      "step": 19170
+    },
+    {
+      "epoch": 3.836,
+      "grad_norm": 0.005103900097310543,
+      "learning_rate": 1.164e-05,
+      "loss": 0.0008,
+      "step": 19180
+    },
+    {
+      "epoch": 3.838,
+      "grad_norm": 0.004146083723753691,
+      "learning_rate": 1.162e-05,
+      "loss": 0.0007,
+      "step": 19190
+    },
+    {
+      "epoch": 3.84,
+      "grad_norm": 0.06597728282213211,
+      "learning_rate": 1.16e-05,
+      "loss": 0.0005,
+      "step": 19200
+    },
+    {
+      "epoch": 3.842,
+      "grad_norm": 0.038979671895504,
+      "learning_rate": 1.1580000000000001e-05,
+      "loss": 0.0385,
+      "step": 19210
+    },
+    {
+      "epoch": 3.844,
+      "grad_norm": 0.222853884100914,
+      "learning_rate": 1.156e-05,
+      "loss": 0.0238,
+      "step": 19220
+    },
+    {
+      "epoch": 3.846,
+      "grad_norm": 0.0034583851229399443,
+      "learning_rate": 1.1540000000000001e-05,
+      "loss": 0.0003,
+      "step": 19230
+    },
+    {
+      "epoch": 3.848,
+      "grad_norm": 0.0020076294895261526,
+      "learning_rate": 1.152e-05,
+      "loss": 0.0002,
+      "step": 19240
+    },
+    {
+      "epoch": 3.85,
+      "grad_norm": 0.0918976441025734,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 0.0006,
+      "step": 19250
+    },
+    {
+      "epoch": 3.852,
+      "grad_norm": 0.0026816155295819044,
+      "learning_rate": 1.148e-05,
+      "loss": 0.0003,
+      "step": 19260
+    },
+    {
+      "epoch": 3.854,
+      "grad_norm": 0.003534205723553896,
+      "learning_rate": 1.146e-05,
+      "loss": 0.0532,
+      "step": 19270
+    },
+    {
+      "epoch": 3.856,
+      "grad_norm": 0.0011774309678003192,
+      "learning_rate": 1.144e-05,
+      "loss": 0.0114,
+      "step": 19280
+    },
+    {
+      "epoch": 3.858,
+      "grad_norm": 10.538480758666992,
+      "learning_rate": 1.142e-05,
+      "loss": 0.0395,
+      "step": 19290
+    },
+    {
+      "epoch": 3.86,
+      "grad_norm": 2.4867305755615234,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 0.0599,
+      "step": 19300
+    },
+    {
+      "epoch": 3.862,
+      "grad_norm": 0.009613540023565292,
+      "learning_rate": 1.1380000000000001e-05,
+      "loss": 0.0571,
+      "step": 19310
+    },
+    {
+      "epoch": 3.864,
+      "grad_norm": 0.08618064969778061,
+      "learning_rate": 1.1360000000000001e-05,
+      "loss": 0.003,
+      "step": 19320
+    },
+    {
+      "epoch": 3.866,
+      "grad_norm": 0.013218753039836884,
+      "learning_rate": 1.134e-05,
+      "loss": 0.0019,
+      "step": 19330
+    },
+    {
+      "epoch": 3.868,
+      "grad_norm": 0.019840750843286514,
+      "learning_rate": 1.132e-05,
+      "loss": 0.0182,
+      "step": 19340
+    },
+    {
+      "epoch": 3.87,
+      "grad_norm": 0.029982879757881165,
+      "learning_rate": 1.13e-05,
+      "loss": 0.0007,
+      "step": 19350
+    },
+    {
+      "epoch": 3.872,
+      "grad_norm": 0.004936496261507273,
+      "learning_rate": 1.128e-05,
+      "loss": 0.0003,
+      "step": 19360
+    },
+    {
+      "epoch": 3.874,
+      "grad_norm": 0.006760491523891687,
+      "learning_rate": 1.126e-05,
+      "loss": 0.0227,
+      "step": 19370
+    },
+    {
+      "epoch": 3.876,
+      "grad_norm": 4.958995342254639,
+      "learning_rate": 1.124e-05,
+      "loss": 0.0459,
+      "step": 19380
+    },
+    {
+      "epoch": 3.878,
+      "grad_norm": 0.013879384845495224,
+      "learning_rate": 1.122e-05,
+      "loss": 0.0105,
+      "step": 19390
+    },
+    {
+      "epoch": 3.88,
+      "grad_norm": 0.0024123447947204113,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 0.0348,
+      "step": 19400
+    },
+    {
+      "epoch": 3.882,
+      "grad_norm": 0.006186407525092363,
+      "learning_rate": 1.118e-05,
+      "loss": 0.0026,
+      "step": 19410
+    },
+    {
+      "epoch": 3.884,
+      "grad_norm": 0.005400556139647961,
+      "learning_rate": 1.1160000000000002e-05,
+      "loss": 0.0003,
+      "step": 19420
+    },
+    {
+      "epoch": 3.886,
+      "grad_norm": 0.010396105237305164,
+      "learning_rate": 1.114e-05,
+      "loss": 0.0006,
+      "step": 19430
+    },
+    {
+      "epoch": 3.888,
+      "grad_norm": 0.05669211596250534,
+      "learning_rate": 1.112e-05,
+      "loss": 0.0094,
+      "step": 19440
+    },
+    {
+      "epoch": 3.89,
+      "grad_norm": 0.011665274389088154,
+      "learning_rate": 1.11e-05,
+      "loss": 0.0003,
+      "step": 19450
+    },
+    {
+      "epoch": 3.892,
+      "grad_norm": 0.0033667744137346745,
+      "learning_rate": 1.108e-05,
+      "loss": 0.0003,
+      "step": 19460
+    },
+    {
+      "epoch": 3.894,
+      "grad_norm": 0.0014184300089254975,
+      "learning_rate": 1.106e-05,
+      "loss": 0.0487,
+      "step": 19470
+    },
+    {
+      "epoch": 3.896,
+      "grad_norm": 0.010493861511349678,
+      "learning_rate": 1.1040000000000001e-05,
+      "loss": 0.0004,
+      "step": 19480
+    },
+    {
+      "epoch": 3.898,
+      "grad_norm": 5.450310230255127,
+      "learning_rate": 1.1020000000000001e-05,
+      "loss": 0.0328,
+      "step": 19490
+    },
+    {
+      "epoch": 3.9,
+      "grad_norm": 0.008521239273250103,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 0.0007,
+      "step": 19500
+    },
+    {
+      "epoch": 3.902,
+      "grad_norm": 0.002282432746142149,
+      "learning_rate": 1.098e-05,
+      "loss": 0.003,
+      "step": 19510
+    },
+    {
+      "epoch": 3.904,
+      "grad_norm": 0.008011306636035442,
+      "learning_rate": 1.096e-05,
+      "loss": 0.0017,
+      "step": 19520
+    },
+    {
+      "epoch": 3.906,
+      "grad_norm": 0.006831697188317776,
+      "learning_rate": 1.094e-05,
+      "loss": 0.0005,
+      "step": 19530
+    },
+    {
+      "epoch": 3.908,
+      "grad_norm": 0.03718950226902962,
+      "learning_rate": 1.092e-05,
+      "loss": 0.0007,
+      "step": 19540
+    },
+    {
+      "epoch": 3.91,
+      "grad_norm": 0.0013135026674717665,
+      "learning_rate": 1.09e-05,
+      "loss": 0.0308,
+      "step": 19550
+    },
+    {
+      "epoch": 3.912,
+      "grad_norm": 0.005439362488687038,
+      "learning_rate": 1.088e-05,
+      "loss": 0.0373,
+      "step": 19560
+    },
+    {
+      "epoch": 3.914,
+      "grad_norm": 0.0057830181904137135,
+      "learning_rate": 1.0860000000000001e-05,
+      "loss": 0.0488,
+      "step": 19570
+    },
+    {
+      "epoch": 3.916,
+      "grad_norm": 0.0024540277663618326,
+      "learning_rate": 1.084e-05,
+      "loss": 0.0004,
+      "step": 19580
+    },
+    {
+      "epoch": 3.918,
+      "grad_norm": 0.03300853818655014,
+      "learning_rate": 1.0820000000000001e-05,
+      "loss": 0.0005,
+      "step": 19590
+    },
+    {
+      "epoch": 3.92,
+      "grad_norm": 0.016842743381857872,
+      "learning_rate": 1.08e-05,
+      "loss": 0.0007,
+      "step": 19600
+    },
+    {
+      "epoch": 3.922,
+      "grad_norm": 0.03553108125925064,
+      "learning_rate": 1.0780000000000002e-05,
+      "loss": 0.0527,
+      "step": 19610
+    },
+    {
+      "epoch": 3.924,
+      "grad_norm": 0.007941951975226402,
+      "learning_rate": 1.076e-05,
+      "loss": 0.0003,
+      "step": 19620
+    },
+    {
+      "epoch": 3.926,
+      "grad_norm": 0.0028262296691536903,
+      "learning_rate": 1.074e-05,
+      "loss": 0.0214,
+      "step": 19630
+    },
+    {
+      "epoch": 3.928,
+      "grad_norm": 0.2088228166103363,
+      "learning_rate": 1.072e-05,
+      "loss": 0.0007,
+      "step": 19640
+    },
+    {
+      "epoch": 3.93,
+      "grad_norm": 0.012677551247179508,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 0.0414,
+      "step": 19650
+    },
+    {
+      "epoch": 3.932,
+      "grad_norm": 0.1734095960855484,
+      "learning_rate": 1.0680000000000001e-05,
+      "loss": 0.0065,
+      "step": 19660
+    },
+    {
+      "epoch": 3.934,
+      "grad_norm": 0.0029586998280137777,
+      "learning_rate": 1.0660000000000001e-05,
+      "loss": 0.0002,
+      "step": 19670
+    },
+    {
+      "epoch": 3.936,
+      "grad_norm": 0.018373634666204453,
+      "learning_rate": 1.064e-05,
+      "loss": 0.0011,
+      "step": 19680
+    },
+    {
+      "epoch": 3.9379999999999997,
+      "grad_norm": 0.010094398632645607,
+      "learning_rate": 1.062e-05,
+      "loss": 0.0497,
+      "step": 19690
+    },
+    {
+      "epoch": 3.94,
+      "grad_norm": 0.06180315464735031,
+      "learning_rate": 1.06e-05,
+      "loss": 0.0927,
+      "step": 19700
+    },
+    {
+      "epoch": 3.942,
+      "grad_norm": 0.0060223545879125595,
+      "learning_rate": 1.058e-05,
+      "loss": 0.001,
+      "step": 19710
+    },
+    {
+      "epoch": 3.944,
+      "grad_norm": 0.006959962192922831,
+      "learning_rate": 1.056e-05,
+      "loss": 0.0007,
+      "step": 19720
+    },
+    {
+      "epoch": 3.9459999999999997,
+      "grad_norm": 0.17345768213272095,
+      "learning_rate": 1.0539999999999999e-05,
+      "loss": 0.0333,
+      "step": 19730
+    },
+    {
+      "epoch": 3.948,
+      "grad_norm": 0.003703049151226878,
+      "learning_rate": 1.0520000000000001e-05,
+      "loss": 0.0433,
+      "step": 19740
+    },
+    {
+      "epoch": 3.95,
+      "grad_norm": 0.0042807250283658504,
+      "learning_rate": 1.05e-05,
+      "loss": 0.0009,
+      "step": 19750
+    },
+    {
+      "epoch": 3.952,
+      "grad_norm": 0.012641056440770626,
+      "learning_rate": 1.0480000000000001e-05,
+      "loss": 0.0014,
+      "step": 19760
+    },
+    {
+      "epoch": 3.9539999999999997,
+      "grad_norm": 0.012951981276273727,
+      "learning_rate": 1.046e-05,
+      "loss": 0.0241,
+      "step": 19770
+    },
+    {
+      "epoch": 3.956,
+      "grad_norm": 0.068780317902565,
+      "learning_rate": 1.0440000000000002e-05,
+      "loss": 0.0309,
+      "step": 19780
+    },
+    {
+      "epoch": 3.958,
+      "grad_norm": 0.005025830119848251,
+      "learning_rate": 1.042e-05,
+      "loss": 0.0008,
+      "step": 19790
+    },
+    {
+      "epoch": 3.96,
+      "grad_norm": 0.39109471440315247,
+      "learning_rate": 1.04e-05,
+      "loss": 0.0009,
+      "step": 19800
+    },
+    {
+      "epoch": 3.9619999999999997,
+      "grad_norm": 0.0031092409044504166,
+      "learning_rate": 1.038e-05,
+      "loss": 0.0017,
+      "step": 19810
+    },
+    {
+      "epoch": 3.964,
+      "grad_norm": 0.013690159656107426,
+      "learning_rate": 1.036e-05,
+      "loss": 0.0066,
+      "step": 19820
+    },
+    {
+      "epoch": 3.966,
+      "grad_norm": 0.050386037677526474,
+      "learning_rate": 1.0340000000000001e-05,
+      "loss": 0.0004,
+      "step": 19830
+    },
+    {
+      "epoch": 3.968,
+      "grad_norm": 0.009826398454606533,
+      "learning_rate": 1.0320000000000001e-05,
+      "loss": 0.0006,
+      "step": 19840
+    },
+    {
+      "epoch": 3.9699999999999998,
+      "grad_norm": 0.005541136488318443,
+      "learning_rate": 1.03e-05,
+      "loss": 0.001,
+      "step": 19850
+    },
+    {
+      "epoch": 3.972,
+      "grad_norm": 0.03238997980952263,
+      "learning_rate": 1.0280000000000002e-05,
+      "loss": 0.0013,
+      "step": 19860
+    },
+    {
+      "epoch": 3.974,
+      "grad_norm": 0.01623641885817051,
+      "learning_rate": 1.026e-05,
+      "loss": 0.03,
+      "step": 19870
+    },
+    {
+      "epoch": 3.976,
+      "grad_norm": 2.500455856323242,
+      "learning_rate": 1.024e-05,
+      "loss": 0.0758,
+      "step": 19880
+    },
+    {
+      "epoch": 3.9779999999999998,
+      "grad_norm": 0.0059792581014335155,
+      "learning_rate": 1.022e-05,
+      "loss": 0.0005,
+      "step": 19890
+    },
+    {
+      "epoch": 3.98,
+      "grad_norm": 0.0036019564140588045,
+      "learning_rate": 1.02e-05,
+      "loss": 0.0019,
+      "step": 19900
+    },
+    {
+      "epoch": 3.982,
+      "grad_norm": 0.006227910052984953,
+      "learning_rate": 1.018e-05,
+      "loss": 0.0174,
+      "step": 19910
+    },
+    {
+      "epoch": 3.984,
+      "grad_norm": 0.05088542029261589,
+      "learning_rate": 1.016e-05,
+      "loss": 0.0022,
+      "step": 19920
+    },
+    {
+      "epoch": 3.9859999999999998,
+      "grad_norm": 0.21118783950805664,
+      "learning_rate": 1.0140000000000001e-05,
+      "loss": 0.0053,
+      "step": 19930
+    },
+    {
+      "epoch": 3.988,
+      "grad_norm": 0.003003940684720874,
+      "learning_rate": 1.012e-05,
+      "loss": 0.0005,
+      "step": 19940
+    },
+    {
+      "epoch": 3.99,
+      "grad_norm": 0.012095367535948753,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 0.0414,
+      "step": 19950
+    },
+    {
+      "epoch": 3.992,
+      "grad_norm": 0.012738720513880253,
+      "learning_rate": 1.008e-05,
+      "loss": 0.0462,
+      "step": 19960
+    },
+    {
+      "epoch": 3.9939999999999998,
+      "grad_norm": 0.04133102670311928,
+      "learning_rate": 1.006e-05,
+      "loss": 0.0019,
+      "step": 19970
+    },
+    {
+      "epoch": 3.996,
+      "grad_norm": 0.0019321141298860312,
+      "learning_rate": 1.004e-05,
+      "loss": 0.0026,
+      "step": 19980
+    },
+    {
+      "epoch": 3.998,
+      "grad_norm": 0.006852047983556986,
+      "learning_rate": 1.002e-05,
+      "loss": 0.0002,
+      "step": 19990
+    },
+    {
+      "epoch": 4.0,
+      "grad_norm": 0.0022855447605252266,
+      "learning_rate": 1e-05,
+      "loss": 0.0002,
+      "step": 20000
+    },
+    {
+      "epoch": 4.0,
+      "eval_loss": 0.012842352502048016,
+      "eval_runtime": 94.2048,
+      "eval_samples_per_second": 106.152,
+      "eval_steps_per_second": 13.269,
+      "step": 20000
+    }
+  ],
+  "logging_steps": 10,
+  "max_steps": 25000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 5,
+  "save_steps": 500,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": false
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 4.62403141632e+16,
+  "train_batch_size": 8,
+  "trial_name": null,
+  "trial_params": null
+}