diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,7052 @@
+{
+  "best_metric": 0.5115005185204882,
+  "best_model_checkpoint": "/m/triton/scratch/elec/puhe/p/palp3/MUCS/indicwav2vec_outputs/pd_warmup_2000/s300_shuff500/checkpoint-1000",
+  "epoch": 1.6,
+  "eval_steps": 1000,
+  "global_step": 1000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0016,
+      "grad_norm": NaN,
+      "learning_rate": 0.0,
+      "loss": 76.924,
+      "step": 1
+    },
+    {
+      "epoch": 0.0032,
+      "grad_norm": Infinity,
+      "learning_rate": 0.0,
+      "loss": 40.9666,
+      "step": 2
+    },
+    {
+      "epoch": 0.0048,
+      "grad_norm": 22.267667770385742,
+      "learning_rate": 3e-07,
+      "loss": 40.3514,
+      "step": 3
+    },
+    {
+      "epoch": 0.0064,
+      "grad_norm": 11.958697319030762,
+      "learning_rate": 6e-07,
+      "loss": 28.7886,
+      "step": 4
+    },
+    {
+      "epoch": 0.008,
+      "grad_norm": 14.436713218688965,
+      "learning_rate": 9e-07,
+      "loss": 33.1337,
+      "step": 5
+    },
+    {
+      "epoch": 0.0096,
+      "grad_norm": 15.921396255493164,
+      "learning_rate": 1.2e-06,
+      "loss": 29.2715,
+      "step": 6
+    },
+    {
+      "epoch": 0.0112,
+      "grad_norm": 11.616898536682129,
+      "learning_rate": 1.4999999999999998e-06,
+      "loss": 28.6694,
+      "step": 7
+    },
+    {
+      "epoch": 0.0128,
+      "grad_norm": 12.15279483795166,
+      "learning_rate": 1.8e-06,
+      "loss": 26.7664,
+      "step": 8
+    },
+    {
+      "epoch": 0.0144,
+      "grad_norm": 15.99345874786377,
+      "learning_rate": 2.1e-06,
+      "loss": 27.2963,
+      "step": 9
+    },
+    {
+      "epoch": 0.016,
+      "grad_norm": 10.025712966918945,
+      "learning_rate": 2.4e-06,
+      "loss": 22.7932,
+      "step": 10
+    },
+    {
+      "epoch": 0.0176,
+      "grad_norm": 9.379335403442383,
+      "learning_rate": 2.6999999999999996e-06,
+      "loss": 20.7226,
+      "step": 11
+    },
+    {
+      "epoch": 0.0192,
+      "grad_norm": 12.390824317932129,
+      "learning_rate": 2.9999999999999997e-06,
+      "loss": 27.5995,
+      "step": 12
+    },
+    {
+      "epoch": 0.0208,
+      "grad_norm": 10.201970100402832,
+      "learning_rate": 3.2999999999999993e-06,
+      "loss": 23.3013,
+      "step": 13
+    },
+    {
+      "epoch": 0.0224,
+      "grad_norm": 11.483911514282227,
+      "learning_rate": 3.6e-06,
+      "loss": 24.6987,
+      "step": 14
+    },
+    {
+      "epoch": 0.024,
+      "grad_norm": 9.661028861999512,
+      "learning_rate": 3.899999999999999e-06,
+      "loss": 22.0543,
+      "step": 15
+    },
+    {
+      "epoch": 0.0256,
+      "grad_norm": 9.904827117919922,
+      "learning_rate": 4.2e-06,
+      "loss": 20.3867,
+      "step": 16
+    },
+    {
+      "epoch": 0.0272,
+      "grad_norm": 10.597962379455566,
+      "learning_rate": 4.499999999999999e-06,
+      "loss": 24.6232,
+      "step": 17
+    },
+    {
+      "epoch": 0.0288,
+      "grad_norm": 12.7444486618042,
+      "learning_rate": 4.8e-06,
+      "loss": 25.3891,
+      "step": 18
+    },
+    {
+      "epoch": 0.0304,
+      "grad_norm": 10.090996742248535,
+      "learning_rate": 5.1e-06,
+      "loss": 21.5661,
+      "step": 19
+    },
+    {
+      "epoch": 0.032,
+      "grad_norm": 10.649155616760254,
+      "learning_rate": 5.399999999999999e-06,
+      "loss": 23.0623,
+      "step": 20
+    },
+    {
+      "epoch": 0.0336,
+      "grad_norm": 10.286359786987305,
+      "learning_rate": 5.7e-06,
+      "loss": 20.7193,
+      "step": 21
+    },
+    {
+      "epoch": 0.0352,
+      "grad_norm": 10.968955993652344,
+      "learning_rate": 5.999999999999999e-06,
+      "loss": 24.0174,
+      "step": 22
+    },
+    {
+      "epoch": 0.0368,
+      "grad_norm": 9.74566650390625,
+      "learning_rate": 6.3e-06,
+      "loss": 21.454,
+      "step": 23
+    },
+    {
+      "epoch": 0.0384,
+      "grad_norm": 11.162517547607422,
+      "learning_rate": 6.599999999999999e-06,
+      "loss": 23.5266,
+      "step": 24
+    },
+    {
+      "epoch": 0.04,
+      "grad_norm": 9.547463417053223,
+      "learning_rate": 6.899999999999999e-06,
+      "loss": 20.214,
+      "step": 25
+    },
+    {
+      "epoch": 0.0416,
+      "grad_norm": 10.754602432250977,
+      "learning_rate": 7.2e-06,
+      "loss": 22.6765,
+      "step": 26
+    },
+    {
+      "epoch": 0.0432,
+      "grad_norm": 9.74982738494873,
+      "learning_rate": 7.499999999999999e-06,
+      "loss": 20.4099,
+      "step": 27
+    },
+    {
+      "epoch": 0.0448,
+      "grad_norm": 10.397897720336914,
+      "learning_rate": 7.799999999999998e-06,
+      "loss": 21.5807,
+      "step": 28
+    },
+    {
+      "epoch": 0.0464,
+      "grad_norm": 10.956497192382812,
+      "learning_rate": 8.099999999999999e-06,
+      "loss": 23.4123,
+      "step": 29
+    },
+    {
+      "epoch": 0.048,
+      "grad_norm": 10.610095977783203,
+      "learning_rate": 8.4e-06,
+      "loss": 21.6038,
+      "step": 30
+    },
+    {
+      "epoch": 0.0496,
+      "grad_norm": 10.559882164001465,
+      "learning_rate": 8.7e-06,
+      "loss": 21.7059,
+      "step": 31
+    },
+    {
+      "epoch": 0.0512,
+      "grad_norm": 10.241806983947754,
+      "learning_rate": 8.999999999999999e-06,
+      "loss": 21.5684,
+      "step": 32
+    },
+    {
+      "epoch": 0.0528,
+      "grad_norm": 9.802021980285645,
+      "learning_rate": 9.299999999999999e-06,
+      "loss": 19.0525,
+      "step": 33
+    },
+    {
+      "epoch": 0.0544,
+      "grad_norm": 10.859997749328613,
+      "learning_rate": 9.6e-06,
+      "loss": 20.7641,
+      "step": 34
+    },
+    {
+      "epoch": 0.056,
+      "grad_norm": 11.377524375915527,
+      "learning_rate": 9.9e-06,
+      "loss": 21.9332,
+      "step": 35
+    },
+    {
+      "epoch": 0.0576,
+      "grad_norm": 10.088302612304688,
+      "learning_rate": 1.02e-05,
+      "loss": 19.4306,
+      "step": 36
+    },
+    {
+      "epoch": 0.0592,
+      "grad_norm": 12.069904327392578,
+      "learning_rate": 1.05e-05,
+      "loss": 23.6146,
+      "step": 37
+    },
+    {
+      "epoch": 0.0608,
+      "grad_norm": 12.05452823638916,
+      "learning_rate": 1.0799999999999998e-05,
+      "loss": 22.8087,
+      "step": 38
+    },
+    {
+      "epoch": 0.0624,
+      "grad_norm": 12.891792297363281,
+      "learning_rate": 1.1099999999999999e-05,
+      "loss": 21.6699,
+      "step": 39
+    },
+    {
+      "epoch": 0.064,
+      "grad_norm": 10.262922286987305,
+      "learning_rate": 1.14e-05,
+      "loss": 19.0892,
+      "step": 40
+    },
+    {
+      "epoch": 0.0656,
+      "grad_norm": 11.921724319458008,
+      "learning_rate": 1.17e-05,
+      "loss": 21.5905,
+      "step": 41
+    },
+    {
+      "epoch": 0.0672,
+      "grad_norm": 11.676680564880371,
+      "learning_rate": 1.1999999999999999e-05,
+      "loss": 20.87,
+      "step": 42
+    },
+    {
+      "epoch": 0.0688,
+      "grad_norm": 10.26872730255127,
+      "learning_rate": 1.2299999999999999e-05,
+      "loss": 18.0263,
+      "step": 43
+    },
+    {
+      "epoch": 0.0704,
+      "grad_norm": 10.563227653503418,
+      "learning_rate": 1.26e-05,
+      "loss": 18.7622,
+      "step": 44
+    },
+    {
+      "epoch": 0.072,
+      "grad_norm": 14.507094383239746,
+      "learning_rate": 1.2899999999999998e-05,
+      "loss": 24.0105,
+      "step": 45
+    },
+    {
+      "epoch": 0.0736,
+      "grad_norm": 11.43386173248291,
+      "learning_rate": 1.3199999999999997e-05,
+      "loss": 19.3601,
+      "step": 46
+    },
+    {
+      "epoch": 0.0752,
+      "grad_norm": 11.63315200805664,
+      "learning_rate": 1.3499999999999998e-05,
+      "loss": 19.0576,
+      "step": 47
+    },
+    {
+      "epoch": 0.0768,
+      "grad_norm": 12.388842582702637,
+      "learning_rate": 1.3799999999999998e-05,
+      "loss": 19.2927,
+      "step": 48
+    },
+    {
+      "epoch": 0.0784,
+      "grad_norm": 12.819602966308594,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 19.936,
+      "step": 49
+    },
+    {
+      "epoch": 0.08,
+      "grad_norm": NaN,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 17.9793,
+      "step": 50
+    },
+    {
+      "epoch": 0.0816,
+      "grad_norm": 65.82632446289062,
+      "learning_rate": 1.44e-05,
+      "loss": 64.1187,
+      "step": 51
+    },
+    {
+      "epoch": 0.0832,
+      "grad_norm": 26.94750213623047,
+      "learning_rate": 1.47e-05,
+      "loss": 33.92,
+      "step": 52
+    },
+    {
+      "epoch": 0.0848,
+      "grad_norm": 24.70115852355957,
+      "learning_rate": 1.4999999999999999e-05,
+      "loss": 34.5358,
+      "step": 53
+    },
+    {
+      "epoch": 0.0864,
+      "grad_norm": 18.708255767822266,
+      "learning_rate": 1.53e-05,
+      "loss": 25.5629,
+      "step": 54
+    },
+    {
+      "epoch": 0.088,
+      "grad_norm": 21.256839752197266,
+      "learning_rate": 1.5599999999999996e-05,
+      "loss": 27.8492,
+      "step": 55
+    },
+    {
+      "epoch": 0.0896,
+      "grad_norm": 17.251680374145508,
+      "learning_rate": 1.5899999999999997e-05,
+      "loss": 22.1176,
+      "step": 56
+    },
+    {
+      "epoch": 0.0912,
+      "grad_norm": 31.902353286743164,
+      "learning_rate": 1.6199999999999997e-05,
+      "loss": 26.8606,
+      "step": 57
+    },
+    {
+      "epoch": 0.0928,
+      "grad_norm": 20.44807243347168,
+      "learning_rate": 1.6499999999999998e-05,
+      "loss": 23.1075,
+      "step": 58
+    },
+    {
+      "epoch": 0.0944,
+      "grad_norm": 17.042905807495117,
+      "learning_rate": 1.68e-05,
+      "loss": 21.0526,
+      "step": 59
+    },
+    {
+      "epoch": 0.096,
+      "grad_norm": 18.13207244873047,
+      "learning_rate": 1.71e-05,
+      "loss": 20.8373,
+      "step": 60
+    },
+    {
+      "epoch": 0.0976,
+      "grad_norm": 17.839736938476562,
+      "learning_rate": 1.74e-05,
+      "loss": 21.2095,
+      "step": 61
+    },
+    {
+      "epoch": 0.0992,
+      "grad_norm": 18.765409469604492,
+      "learning_rate": 1.7699999999999997e-05,
+      "loss": 21.5782,
+      "step": 62
+    },
+    {
+      "epoch": 0.1008,
+      "grad_norm": 18.407758712768555,
+      "learning_rate": 1.7999999999999997e-05,
+      "loss": 21.009,
+      "step": 63
+    },
+    {
+      "epoch": 0.1024,
+      "grad_norm": 38.16777038574219,
+      "learning_rate": 1.8299999999999998e-05,
+      "loss": 21.9343,
+      "step": 64
+    },
+    {
+      "epoch": 0.104,
+      "grad_norm": 21.490079879760742,
+      "learning_rate": 1.8599999999999998e-05,
+      "loss": 22.2912,
+      "step": 65
+    },
+    {
+      "epoch": 0.1056,
+      "grad_norm": 22.932668685913086,
+      "learning_rate": 1.89e-05,
+      "loss": 20.3447,
+      "step": 66
+    },
+    {
+      "epoch": 0.1072,
+      "grad_norm": 26.10978126525879,
+      "learning_rate": 1.92e-05,
+      "loss": 23.8814,
+      "step": 67
+    },
+    {
+      "epoch": 0.1088,
+      "grad_norm": 18.075897216796875,
+      "learning_rate": 1.95e-05,
+      "loss": 18.7612,
+      "step": 68
+    },
+    {
+      "epoch": 0.1104,
+      "grad_norm": 18.735963821411133,
+      "learning_rate": 1.98e-05,
+      "loss": 18.8199,
+      "step": 69
+    },
+    {
+      "epoch": 0.112,
+      "grad_norm": 22.017709732055664,
+      "learning_rate": 2.01e-05,
+      "loss": 20.6774,
+      "step": 70
+    },
+    {
+      "epoch": 0.1136,
+      "grad_norm": 38.08246994018555,
+      "learning_rate": 2.04e-05,
+      "loss": 30.1672,
+      "step": 71
+    },
+    {
+      "epoch": 0.1152,
+      "grad_norm": 22.627145767211914,
+      "learning_rate": 2.07e-05,
+      "loss": 20.0791,
+      "step": 72
+    },
+    {
+      "epoch": 0.1168,
+      "grad_norm": 30.097496032714844,
+      "learning_rate": 2.1e-05,
+      "loss": 23.1779,
+      "step": 73
+    },
+    {
+      "epoch": 0.1184,
+      "grad_norm": 22.679004669189453,
+      "learning_rate": 2.1299999999999996e-05,
+      "loss": 19.4401,
+      "step": 74
+    },
+    {
+      "epoch": 0.12,
+      "grad_norm": 24.041168212890625,
+      "learning_rate": 2.1599999999999996e-05,
+      "loss": 19.3537,
+      "step": 75
+    },
+    {
+      "epoch": 0.1216,
+      "grad_norm": 26.076839447021484,
+      "learning_rate": 2.1899999999999997e-05,
+      "loss": 20.2714,
+      "step": 76
+    },
+    {
+      "epoch": 0.1232,
+      "grad_norm": 25.533342361450195,
+      "learning_rate": 2.2199999999999998e-05,
+      "loss": 19.6379,
+      "step": 77
+    },
+    {
+      "epoch": 0.1248,
+      "grad_norm": 23.547039031982422,
+      "learning_rate": 2.2499999999999998e-05,
+      "loss": 18.0205,
+      "step": 78
+    },
+    {
+      "epoch": 0.1264,
+      "grad_norm": 23.1020565032959,
+      "learning_rate": 2.28e-05,
+      "loss": 17.6504,
+      "step": 79
+    },
+    {
+      "epoch": 0.128,
+      "grad_norm": 28.12115478515625,
+      "learning_rate": 2.31e-05,
+      "loss": 19.9713,
+      "step": 80
+    },
+    {
+      "epoch": 0.1296,
+      "grad_norm": 24.276756286621094,
+      "learning_rate": 2.34e-05,
+      "loss": 17.293,
+      "step": 81
+    },
+    {
+      "epoch": 0.1312,
+      "grad_norm": 21.40571403503418,
+      "learning_rate": 2.3699999999999997e-05,
+      "loss": 15.7224,
+      "step": 82
+    },
+    {
+      "epoch": 0.1328,
+      "grad_norm": 24.049413681030273,
+      "learning_rate": 2.3999999999999997e-05,
+      "loss": 16.4272,
+      "step": 83
+    },
+    {
+      "epoch": 0.1344,
+      "grad_norm": 27.002574920654297,
+      "learning_rate": 2.4299999999999998e-05,
+      "loss": 17.5712,
+      "step": 84
+    },
+    {
+      "epoch": 0.136,
+      "grad_norm": 31.374860763549805,
+      "learning_rate": 2.4599999999999998e-05,
+      "loss": 19.1111,
+      "step": 85
+    },
+    {
+      "epoch": 0.1376,
+      "grad_norm": 28.4317684173584,
+      "learning_rate": 2.49e-05,
+      "loss": 17.1805,
+      "step": 86
+    },
+    {
+      "epoch": 0.1392,
+      "grad_norm": 32.47872543334961,
+      "learning_rate": 2.52e-05,
+      "loss": 18.6232,
+      "step": 87
+    },
+    {
+      "epoch": 0.1408,
+      "grad_norm": 28.26717185974121,
+      "learning_rate": 2.55e-05,
+      "loss": 16.3721,
+      "step": 88
+    },
+    {
+      "epoch": 0.1424,
+      "grad_norm": 29.546110153198242,
+      "learning_rate": 2.5799999999999997e-05,
+      "loss": 16.6354,
+      "step": 89
+    },
+    {
+      "epoch": 0.144,
+      "grad_norm": 32.13431930541992,
+      "learning_rate": 2.6099999999999997e-05,
+      "loss": 16.9808,
+      "step": 90
+    },
+    {
+      "epoch": 0.1456,
+      "grad_norm": 36.368682861328125,
+      "learning_rate": 2.6399999999999995e-05,
+      "loss": 17.5935,
+      "step": 91
+    },
+    {
+      "epoch": 0.1472,
+      "grad_norm": 28.789241790771484,
+      "learning_rate": 2.6699999999999995e-05,
+      "loss": 14.6106,
+      "step": 92
+    },
+    {
+      "epoch": 0.1488,
+      "grad_norm": 30.914873123168945,
+      "learning_rate": 2.6999999999999996e-05,
+      "loss": 14.5306,
+      "step": 93
+    },
+    {
+      "epoch": 0.1504,
+      "grad_norm": 31.848777770996094,
+      "learning_rate": 2.7299999999999996e-05,
+      "loss": 14.7256,
+      "step": 94
+    },
+    {
+      "epoch": 0.152,
+      "grad_norm": 43.699851989746094,
+      "learning_rate": 2.7599999999999997e-05,
+      "loss": 16.3519,
+      "step": 95
+    },
+    {
+      "epoch": 0.1536,
+      "grad_norm": 39.825836181640625,
+      "learning_rate": 2.7899999999999997e-05,
+      "loss": 15.9264,
+      "step": 96
+    },
+    {
+      "epoch": 0.1552,
+      "grad_norm": 32.04133224487305,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 13.4326,
+      "step": 97
+    },
+    {
+      "epoch": 0.1568,
+      "grad_norm": 39.8133659362793,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 14.3427,
+      "step": 98
+    },
+    {
+      "epoch": 0.1584,
+      "grad_norm": 36.54108810424805,
+      "learning_rate": 2.88e-05,
+      "loss": 13.5702,
+      "step": 99
+    },
+    {
+      "epoch": 0.16,
+      "grad_norm": NaN,
+      "learning_rate": 2.88e-05,
+      "loss": 14.6284,
+      "step": 100
+    },
+    {
+      "epoch": 0.1616,
+      "grad_norm": NaN,
+      "learning_rate": 2.88e-05,
+      "loss": 25.1519,
+      "step": 101
+    },
+    {
+      "epoch": 0.1632,
+      "grad_norm": 146.48204040527344,
+      "learning_rate": 2.91e-05,
+      "loss": 26.3907,
+      "step": 102
+    },
+    {
+      "epoch": 0.1648,
+      "grad_norm": 56.64327621459961,
+      "learning_rate": 2.94e-05,
+      "loss": 17.1257,
+      "step": 103
+    },
+    {
+      "epoch": 0.1664,
+      "grad_norm": 87.11421966552734,
+      "learning_rate": 2.97e-05,
+      "loss": 18.9131,
+      "step": 104
+    },
+    {
+      "epoch": 0.168,
+      "grad_norm": 57.33453369140625,
+      "learning_rate": 2.9999999999999997e-05,
+      "loss": 15.569,
+      "step": 105
+    },
+    {
+      "epoch": 0.1696,
+      "grad_norm": 46.68961715698242,
+      "learning_rate": 3.0299999999999998e-05,
+      "loss": 13.1879,
+      "step": 106
+    },
+    {
+      "epoch": 0.1712,
+      "grad_norm": 59.92051315307617,
+      "learning_rate": 3.06e-05,
+      "loss": 14.6438,
+      "step": 107
+    },
+    {
+      "epoch": 0.1728,
+      "grad_norm": 57.27889633178711,
+      "learning_rate": 3.09e-05,
+      "loss": 14.3459,
+      "step": 108
+    },
+    {
+      "epoch": 0.1744,
+      "grad_norm": 51.79650115966797,
+      "learning_rate": 3.119999999999999e-05,
+      "loss": 12.7327,
+      "step": 109
+    },
+    {
+      "epoch": 0.176,
+      "grad_norm": 45.90155029296875,
+      "learning_rate": 3.149999999999999e-05,
+      "loss": 10.6128,
+      "step": 110
+    },
+    {
+      "epoch": 0.1776,
+      "grad_norm": 46.71571350097656,
+      "learning_rate": 3.1799999999999994e-05,
+      "loss": 11.1419,
+      "step": 111
+    },
+    {
+      "epoch": 0.1792,
+      "grad_norm": 54.34334182739258,
+      "learning_rate": 3.2099999999999994e-05,
+      "loss": 11.3816,
+      "step": 112
+    },
+    {
+      "epoch": 0.1808,
+      "grad_norm": 44.52326965332031,
+      "learning_rate": 3.2399999999999995e-05,
+      "loss": 9.99,
+      "step": 113
+    },
+    {
+      "epoch": 0.1824,
+      "grad_norm": 51.66781997680664,
+      "learning_rate": 3.2699999999999995e-05,
+      "loss": 10.7379,
+      "step": 114
+    },
+    {
+      "epoch": 0.184,
+      "grad_norm": 44.501441955566406,
+      "learning_rate": 3.2999999999999996e-05,
+      "loss": 9.5815,
+      "step": 115
+    },
+    {
+      "epoch": 0.1856,
+      "grad_norm": 42.56653594970703,
+      "learning_rate": 3.3299999999999996e-05,
+      "loss": 9.265,
+      "step": 116
+    },
+    {
+      "epoch": 0.1872,
+      "grad_norm": 40.3764762878418,
+      "learning_rate": 3.36e-05,
+      "loss": 8.6506,
+      "step": 117
+    },
+    {
+      "epoch": 0.1888,
+      "grad_norm": 40.9335823059082,
+      "learning_rate": 3.39e-05,
+      "loss": 8.5488,
+      "step": 118
+    },
+    {
+      "epoch": 0.1904,
+      "grad_norm": 42.77170181274414,
+      "learning_rate": 3.42e-05,
+      "loss": 8.5519,
+      "step": 119
+    },
+    {
+      "epoch": 0.192,
+      "grad_norm": 39.98923873901367,
+      "learning_rate": 3.45e-05,
+      "loss": 8.0759,
+      "step": 120
+    },
+    {
+      "epoch": 0.1936,
+      "grad_norm": 33.6198844909668,
+      "learning_rate": 3.48e-05,
+      "loss": 7.3643,
+      "step": 121
+    },
+    {
+      "epoch": 0.1952,
+      "grad_norm": 36.385223388671875,
+      "learning_rate": 3.51e-05,
+      "loss": 7.4889,
+      "step": 122
+    },
+    {
+      "epoch": 0.1968,
+      "grad_norm": 34.519630432128906,
+      "learning_rate": 3.539999999999999e-05,
+      "loss": 7.0802,
+      "step": 123
+    },
+    {
+      "epoch": 0.1984,
+      "grad_norm": 31.894567489624023,
+      "learning_rate": 3.5699999999999994e-05,
+      "loss": 6.8235,
+      "step": 124
+    },
+    {
+      "epoch": 0.2,
+      "grad_norm": 34.02376174926758,
+      "learning_rate": 3.5999999999999994e-05,
+      "loss": 6.9291,
+      "step": 125
+    },
+    {
+      "epoch": 0.2016,
+      "grad_norm": 28.912235260009766,
+      "learning_rate": 3.6299999999999995e-05,
+      "loss": 6.4299,
+      "step": 126
+    },
+    {
+      "epoch": 0.2032,
+      "grad_norm": 28.84571647644043,
+      "learning_rate": 3.6599999999999995e-05,
+      "loss": 6.3964,
+      "step": 127
+    },
+    {
+      "epoch": 0.2048,
+      "grad_norm": 27.383811950683594,
+      "learning_rate": 3.6899999999999996e-05,
+      "loss": 6.2274,
+      "step": 128
+    },
+    {
+      "epoch": 0.2064,
+      "grad_norm": 29.78708267211914,
+      "learning_rate": 3.7199999999999996e-05,
+      "loss": 6.0653,
+      "step": 129
+    },
+    {
+      "epoch": 0.208,
+      "grad_norm": 28.544218063354492,
+      "learning_rate": 3.75e-05,
+      "loss": 6.2053,
+      "step": 130
+    },
+    {
+      "epoch": 0.2096,
+      "grad_norm": 22.004798889160156,
+      "learning_rate": 3.78e-05,
+      "loss": 5.7141,
+      "step": 131
+    },
+    {
+      "epoch": 0.2112,
+      "grad_norm": 20.87236785888672,
+      "learning_rate": 3.81e-05,
+      "loss": 5.6288,
+      "step": 132
+    },
+    {
+      "epoch": 0.2128,
+      "grad_norm": 21.300033569335938,
+      "learning_rate": 3.84e-05,
+      "loss": 5.5556,
+      "step": 133
+    },
+    {
+      "epoch": 0.2144,
+      "grad_norm": 19.316028594970703,
+      "learning_rate": 3.87e-05,
+      "loss": 5.4672,
+      "step": 134
+    },
+    {
+      "epoch": 0.216,
+      "grad_norm": 19.701513290405273,
+      "learning_rate": 3.9e-05,
+      "loss": 5.4904,
+      "step": 135
+    },
+    {
+      "epoch": 0.2176,
+      "grad_norm": 13.995134353637695,
+      "learning_rate": 3.93e-05,
+      "loss": 5.1972,
+      "step": 136
+    },
+    {
+      "epoch": 0.2192,
+      "grad_norm": 14.500862121582031,
+      "learning_rate": 3.96e-05,
+      "loss": 5.1985,
+      "step": 137
+    },
+    {
+      "epoch": 0.2208,
+      "grad_norm": 10.669529914855957,
+      "learning_rate": 3.99e-05,
+      "loss": 5.0034,
+      "step": 138
+    },
+    {
+      "epoch": 0.2224,
+      "grad_norm": 8.051897048950195,
+      "learning_rate": 4.02e-05,
+      "loss": 4.9248,
+      "step": 139
+    },
+    {
+      "epoch": 0.224,
+      "grad_norm": 11.402167320251465,
+      "learning_rate": 4.05e-05,
+      "loss": 5.0594,
+      "step": 140
+    },
+    {
+      "epoch": 0.2256,
+      "grad_norm": 10.03395938873291,
+      "learning_rate": 4.08e-05,
+      "loss": 5.037,
+      "step": 141
+    },
+    {
+      "epoch": 0.2272,
+      "grad_norm": 8.426224708557129,
+      "learning_rate": 4.11e-05,
+      "loss": 4.9605,
+      "step": 142
+    },
+    {
+      "epoch": 0.2288,
+      "grad_norm": 4.501130104064941,
+      "learning_rate": 4.14e-05,
+      "loss": 4.7972,
+      "step": 143
+    },
+    {
+      "epoch": 0.2304,
+      "grad_norm": 4.365025520324707,
+      "learning_rate": 4.17e-05,
+      "loss": 4.8065,
+      "step": 144
+    },
+    {
+      "epoch": 0.232,
+      "grad_norm": 4.868807792663574,
+      "learning_rate": 4.2e-05,
+      "loss": 4.8179,
+      "step": 145
+    },
+    {
+      "epoch": 0.2336,
+      "grad_norm": 3.3663101196289062,
+      "learning_rate": 4.229999999999999e-05,
+      "loss": 4.7889,
+      "step": 146
+    },
+    {
+      "epoch": 0.2352,
+      "grad_norm": 3.2163665294647217,
+      "learning_rate": 4.259999999999999e-05,
+      "loss": 4.702,
+      "step": 147
+    },
+    {
+      "epoch": 0.2368,
+      "grad_norm": 3.1945879459381104,
+      "learning_rate": 4.289999999999999e-05,
+      "loss": 4.7035,
+      "step": 148
+    },
+    {
+      "epoch": 0.2384,
+      "grad_norm": 4.427632808685303,
+      "learning_rate": 4.319999999999999e-05,
+      "loss": 4.7546,
+      "step": 149
+    },
+    {
+      "epoch": 0.24,
+      "grad_norm": 5.550114631652832,
+      "learning_rate": 4.3499999999999993e-05,
+      "loss": 4.9709,
+      "step": 150
+    },
+    {
+      "epoch": 0.2416,
+      "grad_norm": 53.31546401977539,
+      "learning_rate": 4.3799999999999994e-05,
+      "loss": 5.9402,
+      "step": 151
+    },
+    {
+      "epoch": 0.2432,
+      "grad_norm": 7.296631336212158,
+      "learning_rate": 4.4099999999999995e-05,
+      "loss": 4.5481,
+      "step": 152
+    },
+    {
+      "epoch": 0.2448,
+      "grad_norm": 5.802248954772949,
+      "learning_rate": 4.4399999999999995e-05,
+      "loss": 4.4998,
+      "step": 153
+    },
+    {
+      "epoch": 0.2464,
+      "grad_norm": 3.7592978477478027,
+      "learning_rate": 4.4699999999999996e-05,
+      "loss": 4.4536,
+      "step": 154
+    },
+    {
+      "epoch": 0.248,
+      "grad_norm": 3.3045403957366943,
+      "learning_rate": 4.4999999999999996e-05,
+      "loss": 4.391,
+      "step": 155
+    },
+    {
+      "epoch": 0.2496,
+      "grad_norm": 5.204708099365234,
+      "learning_rate": 4.5299999999999997e-05,
+      "loss": 4.459,
+      "step": 156
+    },
+    {
+      "epoch": 0.2512,
+      "grad_norm": 4.284322261810303,
+      "learning_rate": 4.56e-05,
+      "loss": 4.3071,
+      "step": 157
+    },
+    {
+      "epoch": 0.2528,
+      "grad_norm": 3.191441059112549,
+      "learning_rate": 4.59e-05,
+      "loss": 4.3388,
+      "step": 158
+    },
+    {
+      "epoch": 0.2544,
+      "grad_norm": 3.909501552581787,
+      "learning_rate": 4.62e-05,
+      "loss": 4.2491,
+      "step": 159
+    },
+    {
+      "epoch": 0.256,
+      "grad_norm": 3.5245823860168457,
+      "learning_rate": 4.65e-05,
+      "loss": 4.3493,
+      "step": 160
+    },
+    {
+      "epoch": 0.2576,
+      "grad_norm": 2.955253839492798,
+      "learning_rate": 4.68e-05,
+      "loss": 4.2599,
+      "step": 161
+    },
+    {
+      "epoch": 0.2592,
+      "grad_norm": 3.2109367847442627,
+      "learning_rate": 4.709999999999999e-05,
+      "loss": 4.161,
+      "step": 162
+    },
+    {
+      "epoch": 0.2608,
+      "grad_norm": 2.6348652839660645,
+      "learning_rate": 4.7399999999999993e-05,
+      "loss": 4.1815,
+      "step": 163
+    },
+    {
+      "epoch": 0.2624,
+      "grad_norm": 2.593337297439575,
+      "learning_rate": 4.7699999999999994e-05,
+      "loss": 4.0801,
+      "step": 164
+    },
+    {
+      "epoch": 0.264,
+      "grad_norm": 3.77801251411438,
+      "learning_rate": 4.7999999999999994e-05,
+      "loss": 4.1635,
+      "step": 165
+    },
+    {
+      "epoch": 0.2656,
+      "grad_norm": 2.2782046794891357,
+      "learning_rate": 4.8299999999999995e-05,
+      "loss": 4.0901,
+      "step": 166
+    },
+    {
+      "epoch": 0.2672,
+      "grad_norm": 1.8724416494369507,
+      "learning_rate": 4.8599999999999995e-05,
+      "loss": 4.0705,
+      "step": 167
+    },
+    {
+      "epoch": 0.2688,
+      "grad_norm": 1.7749762535095215,
+      "learning_rate": 4.8899999999999996e-05,
+      "loss": 4.0042,
+      "step": 168
+    },
+    {
+      "epoch": 0.2704,
+      "grad_norm": 1.4947004318237305,
+      "learning_rate": 4.9199999999999997e-05,
+      "loss": 4.0104,
+      "step": 169
+    },
+    {
+      "epoch": 0.272,
+      "grad_norm": 3.1837706565856934,
+      "learning_rate": 4.95e-05,
+      "loss": 4.0004,
+      "step": 170
+    },
+    {
+      "epoch": 0.2736,
+      "grad_norm": 1.914663553237915,
+      "learning_rate": 4.98e-05,
+      "loss": 3.9682,
+      "step": 171
+    },
+    {
+      "epoch": 0.2752,
+      "grad_norm": 1.708173394203186,
+      "learning_rate": 5.01e-05,
+      "loss": 3.9609,
+      "step": 172
+    },
+    {
+      "epoch": 0.2768,
+      "grad_norm": 4.2649407386779785,
+      "learning_rate": 5.04e-05,
+      "loss": 3.9917,
+      "step": 173
+    },
+    {
+      "epoch": 0.2784,
+      "grad_norm": 1.626754641532898,
+      "learning_rate": 5.07e-05,
+      "loss": 3.9464,
+      "step": 174
+    },
+    {
+      "epoch": 0.28,
+      "grad_norm": 1.2504980564117432,
+      "learning_rate": 5.1e-05,
+      "loss": 3.9311,
+      "step": 175
+    },
+    {
+      "epoch": 0.2816,
+      "grad_norm": 1.5902968645095825,
+      "learning_rate": 5.13e-05,
+      "loss": 3.9733,
+      "step": 176
+    },
+    {
+      "epoch": 0.2832,
+      "grad_norm": 1.160041093826294,
+      "learning_rate": 5.1599999999999994e-05,
+      "loss": 3.9358,
+      "step": 177
+    },
+    {
+      "epoch": 0.2848,
+      "grad_norm": 1.2470163106918335,
+      "learning_rate": 5.1899999999999994e-05,
+      "loss": 3.9375,
+      "step": 178
+    },
+    {
+      "epoch": 0.2864,
+      "grad_norm": 1.0665581226348877,
+      "learning_rate": 5.2199999999999995e-05,
+      "loss": 3.9701,
+      "step": 179
+    },
+    {
+      "epoch": 0.288,
+      "grad_norm": 1.6660319566726685,
+      "learning_rate": 5.2499999999999995e-05,
+      "loss": 3.8957,
+      "step": 180
+    },
+    {
+      "epoch": 0.2896,
+      "grad_norm": 1.1934682130813599,
+      "learning_rate": 5.279999999999999e-05,
+      "loss": 3.9527,
+      "step": 181
+    },
+    {
+      "epoch": 0.2912,
+      "grad_norm": 2.6113617420196533,
+      "learning_rate": 5.309999999999999e-05,
+      "loss": 3.9198,
+      "step": 182
+    },
+    {
+      "epoch": 0.2928,
+      "grad_norm": 1.4902148246765137,
+      "learning_rate": 5.339999999999999e-05,
+      "loss": 3.8525,
+      "step": 183
+    },
+    {
+      "epoch": 0.2944,
+      "grad_norm": 1.645369529724121,
+      "learning_rate": 5.369999999999999e-05,
+      "loss": 3.8462,
+      "step": 184
+    },
+    {
+      "epoch": 0.296,
+      "grad_norm": 1.7309008836746216,
+      "learning_rate": 5.399999999999999e-05,
+      "loss": 3.9076,
+      "step": 185
+    },
+    {
+      "epoch": 0.2976,
+      "grad_norm": 1.9582569599151611,
+      "learning_rate": 5.429999999999999e-05,
+      "loss": 3.9207,
+      "step": 186
+    },
+    {
+      "epoch": 0.2992,
+      "grad_norm": 1.795342206954956,
+      "learning_rate": 5.459999999999999e-05,
+      "loss": 3.9195,
+      "step": 187
+    },
+    {
+      "epoch": 0.3008,
+      "grad_norm": 0.6605049967765808,
+      "learning_rate": 5.489999999999999e-05,
+      "loss": 3.8484,
+      "step": 188
+    },
+    {
+      "epoch": 0.3024,
+      "grad_norm": 1.3379170894622803,
+      "learning_rate": 5.519999999999999e-05,
+      "loss": 3.8709,
+      "step": 189
+    },
+    {
+      "epoch": 0.304,
+      "grad_norm": 2.7950329780578613,
+      "learning_rate": 5.5499999999999994e-05,
+      "loss": 3.8463,
+      "step": 190
+    },
+    {
+      "epoch": 0.3056,
+      "grad_norm": 4.116920471191406,
+      "learning_rate": 5.5799999999999994e-05,
+      "loss": 4.0707,
+      "step": 191
+    },
+    {
+      "epoch": 0.3072,
+      "grad_norm": 1.450844168663025,
+      "learning_rate": 5.6099999999999995e-05,
+      "loss": 3.9311,
+      "step": 192
+    },
+    {
+      "epoch": 0.3088,
+      "grad_norm": 3.2481865882873535,
+      "learning_rate": 5.6399999999999995e-05,
+      "loss": 4.0222,
+      "step": 193
+    },
+    {
+      "epoch": 0.3104,
+      "grad_norm": 1.377772331237793,
+      "learning_rate": 5.6699999999999996e-05,
+      "loss": 3.9005,
+      "step": 194
+    },
+    {
+      "epoch": 0.312,
+      "grad_norm": 2.0002799034118652,
+      "learning_rate": 5.6999999999999996e-05,
+      "loss": 3.9021,
+      "step": 195
+    },
+    {
+      "epoch": 0.3136,
+      "grad_norm": 2.1365325450897217,
+      "learning_rate": 5.73e-05,
+      "loss": 3.8869,
+      "step": 196
+    },
+    {
+      "epoch": 0.3152,
+      "grad_norm": 2.5995476245880127,
+      "learning_rate": 5.76e-05,
+      "loss": 3.8563,
+      "step": 197
+    },
+    {
+      "epoch": 0.3168,
+      "grad_norm": 3.046848773956299,
+      "learning_rate": 5.79e-05,
+      "loss": 3.9452,
+      "step": 198
+    },
+    {
+      "epoch": 0.3184,
+      "grad_norm": 2.9868597984313965,
+      "learning_rate": 5.82e-05,
+      "loss": 3.9164,
+      "step": 199
+    },
+    {
+      "epoch": 0.32,
+      "grad_norm": NaN,
+      "learning_rate": 5.82e-05,
+      "loss": 3.8911,
+      "step": 200
+    },
+    {
+      "epoch": 0.3216,
+      "grad_norm": 22.723224639892578,
+      "learning_rate": 5.85e-05,
+      "loss": 4.3327,
+      "step": 201
+    },
+    {
+      "epoch": 0.3232,
+      "grad_norm": 14.148520469665527,
+      "learning_rate": 5.88e-05,
+      "loss": 4.0732,
+      "step": 202
+    },
+    {
+      "epoch": 0.3248,
+      "grad_norm": 12.10940170288086,
+      "learning_rate": 5.91e-05,
+      "loss": 4.0389,
+      "step": 203
+    },
+    {
+      "epoch": 0.3264,
+      "grad_norm": 7.240406513214111,
+      "learning_rate": 5.94e-05,
+      "loss": 3.938,
+      "step": 204
+    },
+    {
+      "epoch": 0.328,
+      "grad_norm": 4.678879261016846,
+      "learning_rate": 5.97e-05,
+      "loss": 3.911,
+      "step": 205
+    },
+    {
+      "epoch": 0.3296,
+      "grad_norm": 2.724951982498169,
+      "learning_rate": 5.9999999999999995e-05,
+      "loss": 3.9436,
+      "step": 206
+    },
+    {
+      "epoch": 0.3312,
+      "grad_norm": 4.7506279945373535,
+      "learning_rate": 6.0299999999999995e-05,
+      "loss": 3.8465,
+      "step": 207
+    },
+    {
+      "epoch": 0.3328,
+      "grad_norm": 2.6390953063964844,
+      "learning_rate": 6.0599999999999996e-05,
+      "loss": 3.932,
+      "step": 208
+    },
+    {
+      "epoch": 0.3344,
+      "grad_norm": 3.661578893661499,
+      "learning_rate": 6.0899999999999996e-05,
+      "loss": 3.8498,
+      "step": 209
+    },
+    {
+      "epoch": 0.336,
+      "grad_norm": 2.446004867553711,
+      "learning_rate": 6.12e-05,
+      "loss": 3.9245,
+      "step": 210
+    },
+    {
+      "epoch": 0.3376,
+      "grad_norm": 1.197083592414856,
+      "learning_rate": 6.149999999999999e-05,
+      "loss": 4.0049,
+      "step": 211
+    },
+    {
+      "epoch": 0.3392,
+      "grad_norm": 3.957880735397339,
+      "learning_rate": 6.18e-05,
+      "loss": 3.8129,
+      "step": 212
+    },
+    {
+      "epoch": 0.3408,
+      "grad_norm": 2.243058681488037,
+      "learning_rate": 6.209999999999999e-05,
+      "loss": 3.8419,
+      "step": 213
+    },
+    {
+      "epoch": 0.3424,
+      "grad_norm": 0.8457456827163696,
+      "learning_rate": 6.239999999999999e-05,
+      "loss": 3.8015,
+      "step": 214
+    },
+    {
+      "epoch": 0.344,
+      "grad_norm": 2.7040092945098877,
+      "learning_rate": 6.269999999999999e-05,
+      "loss": 3.7757,
+      "step": 215
+    },
+    {
+      "epoch": 0.3456,
+      "grad_norm": 2.867565155029297,
+      "learning_rate": 6.299999999999999e-05,
+      "loss": 3.748,
+      "step": 216
+    },
+    {
+      "epoch": 0.3472,
+      "grad_norm": 9.108602523803711,
+      "learning_rate": 6.33e-05,
+      "loss": 4.0417,
+      "step": 217
+    },
+    {
+      "epoch": 0.3488,
+      "grad_norm": 2.7541725635528564,
+      "learning_rate": 6.359999999999999e-05,
+      "loss": 3.7908,
+      "step": 218
+    },
+    {
+      "epoch": 0.3504,
+      "grad_norm": 1.1848869323730469,
+      "learning_rate": 6.39e-05,
+      "loss": 3.7637,
+      "step": 219
+    },
+    {
+      "epoch": 0.352,
+      "grad_norm": 0.6500396132469177,
+      "learning_rate": 6.419999999999999e-05,
+      "loss": 3.8055,
+      "step": 220
+    },
+    {
+      "epoch": 0.3536,
+      "grad_norm": 2.706550359725952,
+      "learning_rate": 6.45e-05,
+      "loss": 3.7637,
+      "step": 221
+    },
+    {
+      "epoch": 0.3552,
+      "grad_norm": 5.064160346984863,
+      "learning_rate": 6.479999999999999e-05,
+      "loss": 3.7861,
+      "step": 222
+    },
+    {
+      "epoch": 0.3568,
+      "grad_norm": 3.20385479927063,
+      "learning_rate": 6.51e-05,
+      "loss": 3.7752,
+      "step": 223
+    },
+    {
+      "epoch": 0.3584,
+      "grad_norm": 2.3726119995117188,
+      "learning_rate": 6.539999999999999e-05,
+      "loss": 3.7934,
+      "step": 224
+    },
+    {
+      "epoch": 0.36,
+      "grad_norm": 1.985705852508545,
+      "learning_rate": 6.57e-05,
+      "loss": 3.8806,
+      "step": 225
+    },
+    {
+      "epoch": 0.3616,
+      "grad_norm": 0.669208288192749,
+      "learning_rate": 6.599999999999999e-05,
+      "loss": 3.7576,
+      "step": 226
+    },
+    {
+      "epoch": 0.3632,
+      "grad_norm": 1.7072322368621826,
+      "learning_rate": 6.63e-05,
+      "loss": 3.7382,
+      "step": 227
+    },
+    {
+      "epoch": 0.3648,
+      "grad_norm": 2.339816093444824,
+      "learning_rate": 6.659999999999999e-05,
+      "loss": 3.8001,
+      "step": 228
+    },
+    {
+      "epoch": 0.3664,
+      "grad_norm": 0.6553944945335388,
+      "learning_rate": 6.69e-05,
+      "loss": 3.7473,
+      "step": 229
+    },
+    {
+      "epoch": 0.368,
+      "grad_norm": 1.8117849826812744,
+      "learning_rate": 6.72e-05,
+      "loss": 3.79,
+      "step": 230
+    },
+    {
+      "epoch": 0.3696,
+      "grad_norm": 1.0229136943817139,
+      "learning_rate": 6.75e-05,
+      "loss": 3.7968,
+      "step": 231
+    },
+    {
+      "epoch": 0.3712,
+      "grad_norm": 1.6037867069244385,
+      "learning_rate": 6.78e-05,
+      "loss": 3.7716,
+      "step": 232
+    },
+    {
+      "epoch": 0.3728,
+      "grad_norm": 1.716901183128357,
+      "learning_rate": 6.81e-05,
+      "loss": 3.8464,
+      "step": 233
+    },
+    {
+      "epoch": 0.3744,
+      "grad_norm": 9.919891357421875,
+      "learning_rate": 6.84e-05,
+      "loss": 3.8106,
+      "step": 234
+    },
+    {
+      "epoch": 0.376,
+      "grad_norm": 1.2543926239013672,
+      "learning_rate": 6.87e-05,
+      "loss": 3.7871,
+      "step": 235
+    },
+    {
+      "epoch": 0.3776,
+      "grad_norm": 5.111069202423096,
+      "learning_rate": 6.9e-05,
+      "loss": 3.9449,
+      "step": 236
+    },
+    {
+      "epoch": 0.3792,
+      "grad_norm": 0.940678060054779,
+      "learning_rate": 6.93e-05,
+      "loss": 3.784,
+      "step": 237
+    },
+    {
+      "epoch": 0.3808,
+      "grad_norm": 0.9248812794685364,
+      "learning_rate": 6.96e-05,
+      "loss": 3.8137,
+      "step": 238
+    },
+    {
+      "epoch": 0.3824,
+      "grad_norm": 0.8821243643760681,
+      "learning_rate": 6.989999999999999e-05,
+      "loss": 3.7626,
+      "step": 239
+    },
+    {
+      "epoch": 0.384,
+      "grad_norm": 1.0918103456497192,
+      "learning_rate": 7.02e-05,
+      "loss": 3.7819,
+      "step": 240
+    },
+    {
+      "epoch": 0.3856,
+      "grad_norm": 0.6585227251052856,
+      "learning_rate": 7.049999999999999e-05,
+      "loss": 3.7891,
+      "step": 241
+    },
+    {
+      "epoch": 0.3872,
+      "grad_norm": 3.0343358516693115,
+      "learning_rate": 7.079999999999999e-05,
+      "loss": 3.7803,
+      "step": 242
+    },
+    {
+      "epoch": 0.3888,
+      "grad_norm": 2.1487510204315186,
+      "learning_rate": 7.11e-05,
+      "loss": 3.8404,
+      "step": 243
+    },
+    {
+      "epoch": 0.3904,
+      "grad_norm": 1.0203007459640503,
+      "learning_rate": 7.139999999999999e-05,
+      "loss": 3.7602,
+      "step": 244
+    },
+    {
+      "epoch": 0.392,
+      "grad_norm": 0.8433353900909424,
+      "learning_rate": 7.17e-05,
+      "loss": 3.7826,
+      "step": 245
+    },
+    {
+      "epoch": 0.3936,
+      "grad_norm": 2.8857128620147705,
+      "learning_rate": 7.199999999999999e-05,
+      "loss": 3.7436,
+      "step": 246
+    },
+    {
+      "epoch": 0.3952,
+      "grad_norm": 6.611523628234863,
+      "learning_rate": 7.23e-05,
+      "loss": 4.0391,
+      "step": 247
+    },
+    {
+      "epoch": 0.3968,
+      "grad_norm": 0.7234116196632385,
+      "learning_rate": 7.259999999999999e-05,
+      "loss": 3.8167,
+      "step": 248
+    },
+    {
+      "epoch": 0.3984,
+      "grad_norm": 0.973664402961731,
+      "learning_rate": 7.29e-05,
+      "loss": 3.8963,
+      "step": 249
+    },
+    {
+      "epoch": 0.4,
+      "grad_norm": 1.6993762254714966,
+      "learning_rate": 7.319999999999999e-05,
+      "loss": 3.9033,
+      "step": 250
+    },
+    {
+      "epoch": 0.4016,
+      "grad_norm": 17.571664810180664,
+      "learning_rate": 7.35e-05,
+      "loss": 4.2596,
+      "step": 251
+    },
+    {
+      "epoch": 0.4032,
+      "grad_norm": 11.271060943603516,
+      "learning_rate": 7.379999999999999e-05,
+      "loss": 4.1186,
+      "step": 252
+    },
+    {
+      "epoch": 0.4048,
+      "grad_norm": 8.646568298339844,
+      "learning_rate": 7.41e-05,
+      "loss": 4.013,
+      "step": 253
+    },
+    {
+      "epoch": 0.4064,
+      "grad_norm": 2.02486252784729,
+      "learning_rate": 7.439999999999999e-05,
+      "loss": 3.8974,
+      "step": 254
+    },
+    {
+      "epoch": 0.408,
+      "grad_norm": 3.4109764099121094,
+      "learning_rate": 7.47e-05,
+      "loss": 3.8285,
+      "step": 255
+    },
+    {
+      "epoch": 0.4096,
+      "grad_norm": 3.8505735397338867,
+      "learning_rate": 7.5e-05,
+      "loss": 3.9306,
+      "step": 256
+    },
+    {
+      "epoch": 0.4112,
+      "grad_norm": 7.018677234649658,
+      "learning_rate": 7.529999999999999e-05,
+      "loss": 3.8432,
+      "step": 257
+    },
+    {
+      "epoch": 0.4128,
+      "grad_norm": 4.351247310638428,
+      "learning_rate": 7.56e-05,
+      "loss": 3.8534,
+      "step": 258
+    },
+    {
+      "epoch": 0.4144,
+      "grad_norm": 5.365427494049072,
+      "learning_rate": 7.589999999999999e-05,
+      "loss": 3.8408,
+      "step": 259
+    },
+    {
+      "epoch": 0.416,
+      "grad_norm": 3.984861135482788,
+      "learning_rate": 7.62e-05,
+      "loss": 3.7589,
+      "step": 260
+    },
+    {
+      "epoch": 0.4176,
+      "grad_norm": 1.2847763299942017,
+      "learning_rate": 7.649999999999999e-05,
+      "loss": 3.748,
+      "step": 261
+    },
+    {
+      "epoch": 0.4192,
+      "grad_norm": 4.559200286865234,
+      "learning_rate": 7.68e-05,
+      "loss": 3.8104,
+      "step": 262
+    },
+    {
+      "epoch": 0.4208,
+      "grad_norm": 4.230029106140137,
+      "learning_rate": 7.709999999999999e-05,
+      "loss": 3.7676,
+      "step": 263
+    },
+    {
+      "epoch": 0.4224,
+      "grad_norm": 6.13962984085083,
+      "learning_rate": 7.74e-05,
+      "loss": 3.7336,
+      "step": 264
+    },
+    {
+      "epoch": 0.424,
+      "grad_norm": 4.625703811645508,
+      "learning_rate": 7.769999999999999e-05,
+      "loss": 3.744,
+      "step": 265
+    },
+    {
+      "epoch": 0.4256,
+      "grad_norm": 4.050301551818848,
+      "learning_rate": 7.8e-05,
+      "loss": 3.7662,
+      "step": 266
+    },
+    {
+      "epoch": 0.4272,
+      "grad_norm": 3.0125648975372314,
+      "learning_rate": 7.829999999999999e-05,
+      "loss": 3.7469,
+      "step": 267
+    },
+    {
+      "epoch": 0.4288,
+      "grad_norm": 0.6710224747657776,
+      "learning_rate": 7.86e-05,
+      "loss": 3.7913,
+      "step": 268
+    },
+    {
+      "epoch": 0.4304,
+      "grad_norm": 0.7062709927558899,
+      "learning_rate": 7.89e-05,
+      "loss": 3.7765,
+      "step": 269
+    },
+    {
+      "epoch": 0.432,
+      "grad_norm": 5.028995990753174,
+      "learning_rate": 7.92e-05,
+      "loss": 3.7567,
+      "step": 270
+    },
+    {
+      "epoch": 0.4336,
+      "grad_norm": 4.44848108291626,
+      "learning_rate": 7.95e-05,
+      "loss": 3.7673,
+      "step": 271
+    },
+    {
+      "epoch": 0.4352,
+      "grad_norm": 4.467078685760498,
+      "learning_rate": 7.98e-05,
+      "loss": 3.7462,
+      "step": 272
+    },
+    {
+      "epoch": 0.4368,
+      "grad_norm": 3.1866374015808105,
+      "learning_rate": 8.01e-05,
+      "loss": 3.715,
+      "step": 273
+    },
+    {
+      "epoch": 0.4384,
+      "grad_norm": 2.605476140975952,
+      "learning_rate": 8.04e-05,
+      "loss": 3.6984,
+      "step": 274
+    },
+    {
+      "epoch": 0.44,
+      "grad_norm": 0.6094714999198914,
+      "learning_rate": 8.07e-05,
+      "loss": 3.7325,
+      "step": 275
+    },
+    {
+      "epoch": 0.4416,
+      "grad_norm": 6.599428653717041,
+      "learning_rate": 8.1e-05,
+      "loss": 3.8765,
+      "step": 276
+    },
+    {
+      "epoch": 0.4432,
+      "grad_norm": 2.6780223846435547,
+      "learning_rate": 8.13e-05,
+      "loss": 3.7577,
+      "step": 277
+    },
+    {
+      "epoch": 0.4448,
+      "grad_norm": 3.663605213165283,
+      "learning_rate": 8.16e-05,
+      "loss": 3.8035,
+      "step": 278
+    },
+    {
+      "epoch": 0.4464,
+      "grad_norm": 2.812157392501831,
+      "learning_rate": 8.19e-05,
+      "loss": 3.749,
+      "step": 279
+    },
+    {
+      "epoch": 0.448,
+      "grad_norm": 2.2692039012908936,
+      "learning_rate": 8.22e-05,
+      "loss": 3.7149,
+      "step": 280
+    },
+    {
+      "epoch": 0.4496,
+      "grad_norm": 1.1938503980636597,
+      "learning_rate": 8.25e-05,
+      "loss": 3.7246,
+      "step": 281
+    },
+    {
+      "epoch": 0.4512,
+      "grad_norm": 1.3016897439956665,
+      "learning_rate": 8.28e-05,
+      "loss": 3.6932,
+      "step": 282
+    },
+    {
+      "epoch": 0.4528,
+      "grad_norm": 2.0602409839630127,
+      "learning_rate": 8.31e-05,
+      "loss": 3.7896,
+      "step": 283
+    },
+    {
+      "epoch": 0.4544,
+      "grad_norm": 1.2453322410583496,
+      "learning_rate": 8.34e-05,
+      "loss": 3.7896,
+      "step": 284
+    },
+    {
+      "epoch": 0.456,
+      "grad_norm": 0.7699930667877197,
+      "learning_rate": 8.37e-05,
+      "loss": 3.7406,
+      "step": 285
+    },
+    {
+      "epoch": 0.4576,
+      "grad_norm": 0.9949842691421509,
+      "learning_rate": 8.4e-05,
+      "loss": 3.7951,
+      "step": 286
+    },
+    {
+      "epoch": 0.4592,
+      "grad_norm": 1.2708395719528198,
+      "learning_rate": 8.43e-05,
+      "loss": 3.7142,
+      "step": 287
+    },
+    {
+      "epoch": 0.4608,
+      "grad_norm": 1.6578696966171265,
+      "learning_rate": 8.459999999999998e-05,
+      "loss": 3.7042,
+      "step": 288
+    },
+    {
+      "epoch": 0.4624,
+      "grad_norm": 0.9027276635169983,
+      "learning_rate": 8.489999999999999e-05,
+      "loss": 3.6875,
+      "step": 289
+    },
+    {
+      "epoch": 0.464,
+      "grad_norm": 1.3110026121139526,
+      "learning_rate": 8.519999999999998e-05,
+      "loss": 3.7219,
+      "step": 290
+    },
+    {
+      "epoch": 0.4656,
+      "grad_norm": 0.9840269088745117,
+      "learning_rate": 8.549999999999999e-05,
+      "loss": 3.7555,
+      "step": 291
+    },
+    {
+      "epoch": 0.4672,
+      "grad_norm": 1.4040346145629883,
+      "learning_rate": 8.579999999999998e-05,
+      "loss": 3.7981,
+      "step": 292
+    },
+    {
+      "epoch": 0.4688,
+      "grad_norm": 1.0543975830078125,
+      "learning_rate": 8.609999999999999e-05,
+      "loss": 3.7075,
+      "step": 293
+    },
+    {
+      "epoch": 0.4704,
+      "grad_norm": 0.9345111846923828,
+      "learning_rate": 8.639999999999999e-05,
+      "loss": 3.677,
+      "step": 294
+    },
+    {
+      "epoch": 0.472,
+      "grad_norm": 1.007042646408081,
+      "learning_rate": 8.669999999999998e-05,
+      "loss": 3.7533,
+      "step": 295
+    },
+    {
+      "epoch": 0.4736,
+      "grad_norm": 1.7284626960754395,
+      "learning_rate": 8.699999999999999e-05,
+      "loss": 3.6897,
+      "step": 296
+    },
+    {
+      "epoch": 0.4752,
+      "grad_norm": 2.507981538772583,
+      "learning_rate": 8.729999999999998e-05,
+      "loss": 3.8189,
+      "step": 297
+    },
+    {
+      "epoch": 0.4768,
+      "grad_norm": 2.26454496383667,
+      "learning_rate": 8.759999999999999e-05,
+      "loss": 3.8095,
+      "step": 298
+    },
+    {
+      "epoch": 0.4784,
+      "grad_norm": 1.5712822675704956,
+      "learning_rate": 8.789999999999998e-05,
+      "loss": 3.8321,
+      "step": 299
+    },
+    {
+      "epoch": 0.48,
+      "grad_norm": 1.8837485313415527,
+      "learning_rate": 8.819999999999999e-05,
+      "loss": 3.9162,
+      "step": 300
+    },
+    {
+      "epoch": 0.4816,
+      "grad_norm": 14.770750999450684,
+      "learning_rate": 8.849999999999998e-05,
+      "loss": 4.0945,
+      "step": 301
+    },
+    {
+      "epoch": 0.4832,
+      "grad_norm": 14.976526260375977,
+      "learning_rate": 8.879999999999999e-05,
+      "loss": 4.0664,
+      "step": 302
+    },
+    {
+      "epoch": 0.4848,
+      "grad_norm": 9.496882438659668,
+      "learning_rate": 8.909999999999998e-05,
+      "loss": 4.0088,
+      "step": 303
+    },
+    {
+      "epoch": 0.4864,
+      "grad_norm": 13.879667282104492,
+      "learning_rate": 8.939999999999999e-05,
+      "loss": 4.1825,
+      "step": 304
+    },
+    {
+      "epoch": 0.488,
+      "grad_norm": 6.049519062042236,
+      "learning_rate": 8.969999999999998e-05,
+      "loss": 3.883,
+      "step": 305
+    },
+    {
+      "epoch": 0.4896,
+      "grad_norm": 1.975704312324524,
+      "learning_rate": 8.999999999999999e-05,
+      "loss": 3.8272,
+      "step": 306
+    },
+    {
+      "epoch": 0.4912,
+      "grad_norm": 5.8130598068237305,
+      "learning_rate": 9.029999999999999e-05,
+      "loss": 3.8001,
+      "step": 307
+    },
+    {
+      "epoch": 0.4928,
+      "grad_norm": 3.5878612995147705,
+      "learning_rate": 9.059999999999999e-05,
+      "loss": 3.8353,
+      "step": 308
+    },
+    {
+      "epoch": 0.4944,
+      "grad_norm": 4.1221513748168945,
+      "learning_rate": 9.089999999999999e-05,
+      "loss": 3.9413,
+      "step": 309
+    },
+    {
+      "epoch": 0.496,
+      "grad_norm": 5.489438533782959,
+      "learning_rate": 9.12e-05,
+      "loss": 3.7904,
+      "step": 310
+    },
+    {
+      "epoch": 0.4976,
+      "grad_norm": 1.6620556116104126,
+      "learning_rate": 9.149999999999999e-05,
+      "loss": 3.7937,
+      "step": 311
+    },
+    {
+      "epoch": 0.4992,
+      "grad_norm": 1.13966703414917,
+      "learning_rate": 9.18e-05,
+      "loss": 3.7336,
+      "step": 312
+    },
+    {
+      "epoch": 0.5008,
+      "grad_norm": 1.5242611169815063,
+      "learning_rate": 9.209999999999999e-05,
+      "loss": 3.7161,
+      "step": 313
+    },
+    {
+      "epoch": 0.5024,
+      "grad_norm": 3.9798734188079834,
+      "learning_rate": 9.24e-05,
+      "loss": 3.7,
+      "step": 314
+    },
+    {
+      "epoch": 0.504,
+      "grad_norm": 7.939405918121338,
+      "learning_rate": 9.269999999999999e-05,
+      "loss": 3.7537,
+      "step": 315
+    },
+    {
+      "epoch": 0.5056,
+      "grad_norm": 4.0154709815979,
+      "learning_rate": 9.3e-05,
+      "loss": 3.708,
+      "step": 316
+    },
+    {
+      "epoch": 0.5072,
+      "grad_norm": 4.138357639312744,
+      "learning_rate": 9.329999999999999e-05,
+      "loss": 3.7166,
+      "step": 317
+    },
+    {
+      "epoch": 0.5088,
+      "grad_norm": 8.64471435546875,
+      "learning_rate": 9.36e-05,
+      "loss": 4.0627,
+      "step": 318
+    },
+    {
+      "epoch": 0.5104,
+      "grad_norm": 0.7231702208518982,
+      "learning_rate": 9.389999999999999e-05,
+      "loss": 3.7427,
+      "step": 319
+    },
+    {
+      "epoch": 0.512,
+      "grad_norm": 2.371631622314453,
+      "learning_rate": 9.419999999999999e-05,
+      "loss": 3.7203,
+      "step": 320
+    },
+    {
+      "epoch": 0.5136,
+      "grad_norm": 4.284900188446045,
+      "learning_rate": 9.449999999999999e-05,
+      "loss": 3.7413,
+      "step": 321
+    },
+    {
+      "epoch": 0.5152,
+      "grad_norm": 3.0372443199157715,
+      "learning_rate": 9.479999999999999e-05,
+      "loss": 3.6992,
+      "step": 322
+    },
+    {
+      "epoch": 0.5168,
+      "grad_norm": 1.9789845943450928,
+      "learning_rate": 9.51e-05,
+      "loss": 3.7184,
+      "step": 323
+    },
+    {
+      "epoch": 0.5184,
+      "grad_norm": 1.624227523803711,
+      "learning_rate": 9.539999999999999e-05,
+      "loss": 3.6749,
+      "step": 324
+    },
+    {
+      "epoch": 0.52,
+      "grad_norm": 1.5696678161621094,
+      "learning_rate": 9.57e-05,
+      "loss": 3.7581,
+      "step": 325
+    },
+    {
+      "epoch": 0.5216,
+      "grad_norm": 2.7740790843963623,
+      "learning_rate": 9.599999999999999e-05,
+      "loss": 3.6773,
+      "step": 326
+    },
+    {
+      "epoch": 0.5232,
+      "grad_norm": 2.1769227981567383,
+      "learning_rate": 9.63e-05,
+      "loss": 3.6821,
+      "step": 327
+    },
+    {
+      "epoch": 0.5248,
+      "grad_norm": 3.454484224319458,
+      "learning_rate": 9.659999999999999e-05,
+      "loss": 3.6977,
+      "step": 328
+    },
+    {
+      "epoch": 0.5264,
+      "grad_norm": 1.035311222076416,
+      "learning_rate": 9.69e-05,
+      "loss": 3.6816,
+      "step": 329
+    },
+    {
+      "epoch": 0.528,
+      "grad_norm": 1.0064358711242676,
+      "learning_rate": 9.719999999999999e-05,
+      "loss": 3.7105,
+      "step": 330
+    },
+    {
+      "epoch": 0.5296,
+      "grad_norm": 2.302251100540161,
+      "learning_rate": 9.75e-05,
+      "loss": 3.7867,
+      "step": 331
+    },
+    {
+      "epoch": 0.5312,
+      "grad_norm": 2.8935694694519043,
+      "learning_rate": 9.779999999999999e-05,
+      "loss": 3.7186,
+      "step": 332
+    },
+    {
+      "epoch": 0.5328,
+      "grad_norm": 4.943471908569336,
+      "learning_rate": 9.81e-05,
+      "loss": 3.6946,
+      "step": 333
+    },
+    {
+      "epoch": 0.5344,
+      "grad_norm": 2.8398258686065674,
+      "learning_rate": 9.839999999999999e-05,
+      "loss": 3.7091,
+      "step": 334
+    },
+    {
+      "epoch": 0.536,
+      "grad_norm": 3.0762977600097656,
+      "learning_rate": 9.87e-05,
+      "loss": 3.7574,
+      "step": 335
+    },
+    {
+      "epoch": 0.5376,
+      "grad_norm": 1.8813797235488892,
+      "learning_rate": 9.9e-05,
+      "loss": 3.7118,
+      "step": 336
+    },
+    {
+      "epoch": 0.5392,
+      "grad_norm": 0.8849917054176331,
+      "learning_rate": 9.93e-05,
+      "loss": 3.77,
+      "step": 337
+    },
+    {
+      "epoch": 0.5408,
+      "grad_norm": 1.4980673789978027,
+      "learning_rate": 9.96e-05,
+      "loss": 3.6983,
+      "step": 338
+    },
+    {
+      "epoch": 0.5424,
+      "grad_norm": 1.593652367591858,
+      "learning_rate": 9.99e-05,
+      "loss": 3.6434,
+      "step": 339
+    },
+    {
+      "epoch": 0.544,
+      "grad_norm": 1.0899137258529663,
+      "learning_rate": 0.0001002,
+      "loss": 3.6421,
+      "step": 340
+    },
+    {
+      "epoch": 0.5456,
+      "grad_norm": 0.6649819016456604,
+      "learning_rate": 0.0001005,
+      "loss": 3.6891,
+      "step": 341
+    },
+    {
+      "epoch": 0.5472,
+      "grad_norm": 1.5821571350097656,
+      "learning_rate": 0.0001008,
+      "loss": 3.794,
+      "step": 342
+    },
+    {
+      "epoch": 0.5488,
+      "grad_norm": 2.7996299266815186,
+      "learning_rate": 0.0001011,
+      "loss": 3.6779,
+      "step": 343
+    },
+    {
+      "epoch": 0.5504,
+      "grad_norm": 2.5575685501098633,
+      "learning_rate": 0.0001014,
+      "loss": 3.7879,
+      "step": 344
+    },
+    {
+      "epoch": 0.552,
+      "grad_norm": 1.2596614360809326,
+      "learning_rate": 0.00010169999999999999,
+      "loss": 3.7208,
+      "step": 345
+    },
+    {
+      "epoch": 0.5536,
+      "grad_norm": 0.8037718534469604,
+      "learning_rate": 0.000102,
+      "loss": 3.6811,
+      "step": 346
+    },
+    {
+      "epoch": 0.5552,
+      "grad_norm": 1.3349064588546753,
+      "learning_rate": 0.00010229999999999999,
+      "loss": 3.6959,
+      "step": 347
+    },
+    {
+      "epoch": 0.5568,
+      "grad_norm": 1.269572138786316,
+      "learning_rate": 0.0001026,
+      "loss": 3.7447,
+      "step": 348
+    },
+    {
+      "epoch": 0.5584,
+      "grad_norm": 1.022356629371643,
+      "learning_rate": 0.0001029,
+      "loss": 3.6698,
+      "step": 349
+    },
+    {
+      "epoch": 0.56,
+      "grad_norm": 1.5080015659332275,
+      "learning_rate": 0.00010319999999999999,
+      "loss": 3.7888,
+      "step": 350
+    },
+    {
+      "epoch": 0.5616,
+      "grad_norm": 39.285152435302734,
+      "learning_rate": 0.00010349999999999998,
+      "loss": 5.0541,
+      "step": 351
+    },
+    {
+      "epoch": 0.5632,
+      "grad_norm": 9.576460838317871,
+      "learning_rate": 0.00010379999999999999,
+      "loss": 4.014,
+      "step": 352
+    },
+    {
+      "epoch": 0.5648,
+      "grad_norm": 6.356123447418213,
+      "learning_rate": 0.00010409999999999998,
+      "loss": 3.9312,
+      "step": 353
+    },
+    {
+      "epoch": 0.5664,
+      "grad_norm": 2.139646291732788,
+      "learning_rate": 0.00010439999999999999,
+      "loss": 3.7221,
+      "step": 354
+    },
+    {
+      "epoch": 0.568,
+      "grad_norm": 1.8156242370605469,
+      "learning_rate": 0.00010469999999999998,
+      "loss": 3.8395,
+      "step": 355
+    },
+    {
+      "epoch": 0.5696,
+      "grad_norm": 4.785361289978027,
+      "learning_rate": 0.00010499999999999999,
+      "loss": 3.719,
+      "step": 356
+    },
+    {
+      "epoch": 0.5712,
+      "grad_norm": 3.600017786026001,
+      "learning_rate": 0.00010529999999999998,
+      "loss": 3.7428,
+      "step": 357
+    },
+    {
+      "epoch": 0.5728,
+      "grad_norm": 2.4187095165252686,
+      "learning_rate": 0.00010559999999999998,
+      "loss": 3.6832,
+      "step": 358
+    },
+    {
+      "epoch": 0.5744,
+      "grad_norm": 1.5843887329101562,
+      "learning_rate": 0.00010589999999999999,
+      "loss": 3.6504,
+      "step": 359
+    },
+    {
+      "epoch": 0.576,
+      "grad_norm": 1.5045654773712158,
+      "learning_rate": 0.00010619999999999998,
+      "loss": 3.6816,
+      "step": 360
+    },
+    {
+      "epoch": 0.5776,
+      "grad_norm": 2.58827543258667,
+      "learning_rate": 0.00010649999999999999,
+      "loss": 3.642,
+      "step": 361
+    },
+    {
+      "epoch": 0.5792,
+      "grad_norm": 2.5386884212493896,
+      "learning_rate": 0.00010679999999999998,
+      "loss": 3.6578,
+      "step": 362
+    },
+    {
+      "epoch": 0.5808,
+      "grad_norm": 2.9344706535339355,
+      "learning_rate": 0.00010709999999999999,
+      "loss": 3.6694,
+      "step": 363
+    },
+    {
+      "epoch": 0.5824,
+      "grad_norm": 2.340221643447876,
+      "learning_rate": 0.00010739999999999998,
+      "loss": 3.6504,
+      "step": 364
+    },
+    {
+      "epoch": 0.584,
+      "grad_norm": 2.816999912261963,
+      "learning_rate": 0.00010769999999999999,
+      "loss": 3.6654,
+      "step": 365
+    },
+    {
+      "epoch": 0.5856,
+      "grad_norm": 1.5071390867233276,
+      "learning_rate": 0.00010799999999999998,
+      "loss": 3.6406,
+      "step": 366
+    },
+    {
+      "epoch": 0.5872,
+      "grad_norm": 0.7593219876289368,
+      "learning_rate": 0.00010829999999999999,
+      "loss": 3.6683,
+      "step": 367
+    },
+    {
+      "epoch": 0.5888,
+      "grad_norm": 2.646967887878418,
+      "learning_rate": 0.00010859999999999998,
+      "loss": 3.6844,
+      "step": 368
+    },
+    {
+      "epoch": 0.5904,
+      "grad_norm": 2.8628735542297363,
+      "learning_rate": 0.00010889999999999999,
+      "loss": 3.6159,
+      "step": 369
+    },
+    {
+      "epoch": 0.592,
+      "grad_norm": 2.4796457290649414,
+      "learning_rate": 0.00010919999999999998,
+      "loss": 3.6951,
+      "step": 370
+    },
+    {
+      "epoch": 0.5936,
+      "grad_norm": 1.1962988376617432,
+      "learning_rate": 0.00010949999999999999,
+      "loss": 3.5725,
+      "step": 371
+    },
+    {
+      "epoch": 0.5952,
+      "grad_norm": 0.9031145572662354,
+      "learning_rate": 0.00010979999999999999,
+      "loss": 3.5847,
+      "step": 372
+    },
+    {
+      "epoch": 0.5968,
+      "grad_norm": 2.3092095851898193,
+      "learning_rate": 0.00011009999999999999,
+      "loss": 3.6181,
+      "step": 373
+    },
+    {
+      "epoch": 0.5984,
+      "grad_norm": 1.1398162841796875,
+      "learning_rate": 0.00011039999999999999,
+      "loss": 3.6774,
+      "step": 374
+    },
+    {
+      "epoch": 0.6,
+      "grad_norm": 1.0708636045455933,
+      "learning_rate": 0.0001107,
+      "loss": 3.5872,
+      "step": 375
+    },
+    {
+      "epoch": 0.6016,
+      "grad_norm": 0.873319149017334,
+      "learning_rate": 0.00011099999999999999,
+      "loss": 3.6512,
+      "step": 376
+    },
+    {
+      "epoch": 0.6032,
+      "grad_norm": 0.8766927123069763,
+      "learning_rate": 0.0001113,
+      "loss": 3.6434,
+      "step": 377
+    },
+    {
+      "epoch": 0.6048,
+      "grad_norm": 0.7983845472335815,
+      "learning_rate": 0.00011159999999999999,
+      "loss": 3.5572,
+      "step": 378
+    },
+    {
+      "epoch": 0.6064,
+      "grad_norm": 6.645163059234619,
+      "learning_rate": 0.0001119,
+      "loss": 3.7195,
+      "step": 379
+    },
+    {
+      "epoch": 0.608,
+      "grad_norm": 4.845424652099609,
+      "learning_rate": 0.00011219999999999999,
+      "loss": 3.7342,
+      "step": 380
+    },
+    {
+      "epoch": 0.6096,
+      "grad_norm": 0.7528507709503174,
+      "learning_rate": 0.0001125,
+      "loss": 3.5633,
+      "step": 381
+    },
+    {
+      "epoch": 0.6112,
+      "grad_norm": 0.69414883852005,
+      "learning_rate": 0.00011279999999999999,
+      "loss": 3.6274,
+      "step": 382
+    },
+    {
+      "epoch": 0.6128,
+      "grad_norm": 2.0266685485839844,
+      "learning_rate": 0.00011309999999999998,
+      "loss": 3.5491,
+      "step": 383
+    },
+    {
+      "epoch": 0.6144,
+      "grad_norm": 0.7420620918273926,
+      "learning_rate": 0.00011339999999999999,
+      "loss": 3.6293,
+      "step": 384
+    },
+    {
+      "epoch": 0.616,
+      "grad_norm": 1.1609050035476685,
+      "learning_rate": 0.00011369999999999999,
+      "loss": 3.5036,
+      "step": 385
+    },
+    {
+      "epoch": 0.6176,
+      "grad_norm": 1.6696407794952393,
+      "learning_rate": 0.00011399999999999999,
+      "loss": 3.5909,
+      "step": 386
+    },
+    {
+      "epoch": 0.6192,
+      "grad_norm": 1.1470685005187988,
+      "learning_rate": 0.00011429999999999999,
+      "loss": 3.6343,
+      "step": 387
+    },
+    {
+      "epoch": 0.6208,
+      "grad_norm": 1.3622722625732422,
+      "learning_rate": 0.0001146,
+      "loss": 3.5755,
+      "step": 388
+    },
+    {
+      "epoch": 0.6224,
+      "grad_norm": 1.2317267656326294,
+      "learning_rate": 0.00011489999999999999,
+      "loss": 3.5538,
+      "step": 389
+    },
+    {
+      "epoch": 0.624,
+      "grad_norm": 1.1414676904678345,
+      "learning_rate": 0.0001152,
+      "loss": 3.5555,
+      "step": 390
+    },
+    {
+      "epoch": 0.6256,
+      "grad_norm": 1.998960018157959,
+      "learning_rate": 0.00011549999999999999,
+      "loss": 3.5465,
+      "step": 391
+    },
+    {
+      "epoch": 0.6272,
+      "grad_norm": 1.4650264978408813,
+      "learning_rate": 0.0001158,
+      "loss": 3.5777,
+      "step": 392
+    },
+    {
+      "epoch": 0.6288,
+      "grad_norm": 1.5700796842575073,
+      "learning_rate": 0.00011609999999999999,
+      "loss": 3.6034,
+      "step": 393
+    },
+    {
+      "epoch": 0.6304,
+      "grad_norm": 2.38299298286438,
+      "learning_rate": 0.0001164,
+      "loss": 3.5582,
+      "step": 394
+    },
+    {
+      "epoch": 0.632,
+      "grad_norm": 1.2898112535476685,
+      "learning_rate": 0.00011669999999999999,
+      "loss": 3.4869,
+      "step": 395
+    },
+    {
+      "epoch": 0.6336,
+      "grad_norm": 1.2601486444473267,
+      "learning_rate": 0.000117,
+      "loss": 3.5423,
+      "step": 396
+    },
+    {
+      "epoch": 0.6352,
+      "grad_norm": 1.907885193824768,
+      "learning_rate": 0.00011729999999999999,
+      "loss": 3.5019,
+      "step": 397
+    },
+    {
+      "epoch": 0.6368,
+      "grad_norm": 1.2280569076538086,
+      "learning_rate": 0.0001176,
+      "loss": 3.6299,
+      "step": 398
+    },
+    {
+      "epoch": 0.6384,
+      "grad_norm": 2.214331865310669,
+      "learning_rate": 0.00011789999999999999,
+      "loss": 3.5979,
+      "step": 399
+    },
+    {
+      "epoch": 0.64,
+      "grad_norm": NaN,
+      "learning_rate": 0.00011789999999999999,
+      "loss": 3.4218,
+      "step": 400
+    },
+    {
+      "epoch": 0.6416,
+      "grad_norm": 38.77112579345703,
+      "learning_rate": 0.0001182,
+      "loss": 4.5562,
+      "step": 401
+    },
+    {
+      "epoch": 0.6432,
+      "grad_norm": 8.848291397094727,
+      "learning_rate": 0.0001185,
+      "loss": 3.7699,
+      "step": 402
+    },
+    {
+      "epoch": 0.6448,
+      "grad_norm": 6.007197856903076,
+      "learning_rate": 0.0001188,
+      "loss": 3.539,
+      "step": 403
+    },
+    {
+      "epoch": 0.6464,
+      "grad_norm": 3.0180368423461914,
+      "learning_rate": 0.0001191,
+      "loss": 3.5382,
+      "step": 404
+    },
+    {
+      "epoch": 0.648,
+      "grad_norm": 2.071746587753296,
+      "learning_rate": 0.0001194,
+      "loss": 3.5857,
+      "step": 405
+    },
+    {
+      "epoch": 0.6496,
+      "grad_norm": 4.427801132202148,
+      "learning_rate": 0.0001197,
+      "loss": 3.526,
+      "step": 406
+    },
+    {
+      "epoch": 0.6512,
+      "grad_norm": 5.680927753448486,
+      "learning_rate": 0.00011999999999999999,
+      "loss": 3.4296,
+      "step": 407
+    },
+    {
+      "epoch": 0.6528,
+      "grad_norm": 2.7837042808532715,
+      "learning_rate": 0.0001203,
+      "loss": 3.4657,
+      "step": 408
+    },
+    {
+      "epoch": 0.6544,
+      "grad_norm": 4.605573654174805,
+      "learning_rate": 0.00012059999999999999,
+      "loss": 3.4279,
+      "step": 409
+    },
+    {
+      "epoch": 0.656,
+      "grad_norm": 1.96554696559906,
+      "learning_rate": 0.0001209,
+      "loss": 3.3618,
+      "step": 410
+    },
+    {
+      "epoch": 0.6576,
+      "grad_norm": 5.76222038269043,
+      "learning_rate": 0.00012119999999999999,
+      "loss": 3.4163,
+      "step": 411
+    },
+    {
+      "epoch": 0.6592,
+      "grad_norm": 4.640344619750977,
+      "learning_rate": 0.0001215,
+      "loss": 3.4429,
+      "step": 412
+    },
+    {
+      "epoch": 0.6608,
+      "grad_norm": 4.301933288574219,
+      "learning_rate": 0.00012179999999999999,
+      "loss": 3.3255,
+      "step": 413
+    },
+    {
+      "epoch": 0.6624,
+      "grad_norm": 3.781334638595581,
+      "learning_rate": 0.00012209999999999999,
+      "loss": 3.3261,
+      "step": 414
+    },
+    {
+      "epoch": 0.664,
+      "grad_norm": 3.663053035736084,
+      "learning_rate": 0.0001224,
+      "loss": 3.2962,
+      "step": 415
+    },
+    {
+      "epoch": 0.6656,
+      "grad_norm": 3.2776567935943604,
+      "learning_rate": 0.00012269999999999997,
+      "loss": 3.333,
+      "step": 416
+    },
+    {
+      "epoch": 0.6672,
+      "grad_norm": 1.000927209854126,
+      "learning_rate": 0.00012299999999999998,
+      "loss": 3.2321,
+      "step": 417
+    },
+    {
+      "epoch": 0.6688,
+      "grad_norm": 1.561220407485962,
+      "learning_rate": 0.0001233,
+      "loss": 3.3121,
+      "step": 418
+    },
+    {
+      "epoch": 0.6704,
+      "grad_norm": 0.8714520931243896,
+      "learning_rate": 0.0001236,
+      "loss": 3.3394,
+      "step": 419
+    },
+    {
+      "epoch": 0.672,
+      "grad_norm": 1.1457229852676392,
+      "learning_rate": 0.00012389999999999998,
+      "loss": 3.1645,
+      "step": 420
+    },
+    {
+      "epoch": 0.6736,
+      "grad_norm": 2.054020881652832,
+      "learning_rate": 0.00012419999999999998,
+      "loss": 3.2115,
+      "step": 421
+    },
+    {
+      "epoch": 0.6752,
+      "grad_norm": 3.8146936893463135,
+      "learning_rate": 0.0001245,
+      "loss": 3.3334,
+      "step": 422
+    },
+    {
+      "epoch": 0.6768,
+      "grad_norm": 2.3825631141662598,
+      "learning_rate": 0.00012479999999999997,
+      "loss": 3.2264,
+      "step": 423
+    },
+    {
+      "epoch": 0.6784,
+      "grad_norm": 1.282517671585083,
+      "learning_rate": 0.00012509999999999998,
+      "loss": 3.3354,
+      "step": 424
+    },
+    {
+      "epoch": 0.68,
+      "grad_norm": 1.5535123348236084,
+      "learning_rate": 0.00012539999999999999,
+      "loss": 3.1819,
+      "step": 425
+    },
+    {
+      "epoch": 0.6816,
+      "grad_norm": 1.8400110006332397,
+      "learning_rate": 0.0001257,
+      "loss": 3.1989,
+      "step": 426
+    },
+    {
+      "epoch": 0.6832,
+      "grad_norm": 1.3851298093795776,
+      "learning_rate": 0.00012599999999999997,
+      "loss": 3.2336,
+      "step": 427
+    },
+    {
+      "epoch": 0.6848,
+      "grad_norm": 1.884459376335144,
+      "learning_rate": 0.00012629999999999998,
+      "loss": 3.2123,
+      "step": 428
+    },
+    {
+      "epoch": 0.6864,
+      "grad_norm": 1.7640012502670288,
+      "learning_rate": 0.0001266,
+      "loss": 3.0558,
+      "step": 429
+    },
+    {
+      "epoch": 0.688,
+      "grad_norm": 2.564265489578247,
+      "learning_rate": 0.0001269,
+      "loss": 3.0314,
+      "step": 430
+    },
+    {
+      "epoch": 0.6896,
+      "grad_norm": 1.8793052434921265,
+      "learning_rate": 0.00012719999999999997,
+      "loss": 3.0916,
+      "step": 431
+    },
+    {
+      "epoch": 0.6912,
+      "grad_norm": 1.3174560070037842,
+      "learning_rate": 0.00012749999999999998,
+      "loss": 3.0977,
+      "step": 432
+    },
+    {
+      "epoch": 0.6928,
+      "grad_norm": 0.9135323166847229,
+      "learning_rate": 0.0001278,
+      "loss": 3.1459,
+      "step": 433
+    },
+    {
+      "epoch": 0.6944,
+      "grad_norm": 1.05746591091156,
+      "learning_rate": 0.0001281,
+      "loss": 3.1823,
+      "step": 434
+    },
+    {
+      "epoch": 0.696,
+      "grad_norm": 1.2425645589828491,
+      "learning_rate": 0.00012839999999999998,
+      "loss": 2.9603,
+      "step": 435
+    },
+    {
+      "epoch": 0.6976,
+      "grad_norm": 1.2454054355621338,
+      "learning_rate": 0.00012869999999999998,
+      "loss": 2.9414,
+      "step": 436
+    },
+    {
+      "epoch": 0.6992,
+      "grad_norm": 0.9464673399925232,
+      "learning_rate": 0.000129,
+      "loss": 3.1432,
+      "step": 437
+    },
+    {
+      "epoch": 0.7008,
+      "grad_norm": 1.5856995582580566,
+      "learning_rate": 0.0001293,
+      "loss": 3.0063,
+      "step": 438
+    },
+    {
+      "epoch": 0.7024,
+      "grad_norm": 1.043485403060913,
+      "learning_rate": 0.00012959999999999998,
+      "loss": 2.9146,
+      "step": 439
+    },
+    {
+      "epoch": 0.704,
+      "grad_norm": 1.240867257118225,
+      "learning_rate": 0.00012989999999999999,
+      "loss": 2.9979,
+      "step": 440
+    },
+    {
+      "epoch": 0.7056,
+      "grad_norm": 1.7289670705795288,
+      "learning_rate": 0.0001302,
+      "loss": 2.8699,
+      "step": 441
+    },
+    {
+      "epoch": 0.7072,
+      "grad_norm": 1.728317141532898,
+      "learning_rate": 0.0001305,
+      "loss": 2.7802,
+      "step": 442
+    },
+    {
+      "epoch": 0.7088,
+      "grad_norm": 0.960502028465271,
+      "learning_rate": 0.00013079999999999998,
+      "loss": 2.9141,
+      "step": 443
+    },
+    {
+      "epoch": 0.7104,
+      "grad_norm": 2.093698501586914,
+      "learning_rate": 0.0001311,
+      "loss": 3.1318,
+      "step": 444
+    },
+    {
+      "epoch": 0.712,
+      "grad_norm": 1.6515812873840332,
+      "learning_rate": 0.0001314,
+      "loss": 2.9467,
+      "step": 445
+    },
+    {
+      "epoch": 0.7136,
+      "grad_norm": 1.4129968881607056,
+      "learning_rate": 0.00013169999999999998,
+      "loss": 2.7909,
+      "step": 446
+    },
+    {
+      "epoch": 0.7152,
+      "grad_norm": 1.5885038375854492,
+      "learning_rate": 0.00013199999999999998,
+      "loss": 3.1262,
+      "step": 447
+    },
+    {
+      "epoch": 0.7168,
+      "grad_norm": 1.222842812538147,
+      "learning_rate": 0.0001323,
+      "loss": 3.2029,
+      "step": 448
+    },
+    {
+      "epoch": 0.7184,
+      "grad_norm": 1.3282477855682373,
+      "learning_rate": 0.0001326,
+      "loss": 3.0013,
+      "step": 449
+    },
+    {
+      "epoch": 0.72,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001326,
+      "loss": 3.4352,
+      "step": 450
+    },
+    {
+      "epoch": 0.7216,
+      "grad_norm": 27.799081802368164,
+      "learning_rate": 0.00013289999999999998,
+      "loss": 4.3657,
+      "step": 451
+    },
+    {
+      "epoch": 0.7232,
+      "grad_norm": 5.403924465179443,
+      "learning_rate": 0.00013319999999999999,
+      "loss": 2.9244,
+      "step": 452
+    },
+    {
+      "epoch": 0.7248,
+      "grad_norm": 3.8071448802948,
+      "learning_rate": 0.0001335,
+      "loss": 2.927,
+      "step": 453
+    },
+    {
+      "epoch": 0.7264,
+      "grad_norm": 3.504509210586548,
+      "learning_rate": 0.0001338,
+      "loss": 2.9719,
+      "step": 454
+    },
+    {
+      "epoch": 0.728,
+      "grad_norm": 3.500847578048706,
+      "learning_rate": 0.00013409999999999998,
+      "loss": 3.0386,
+      "step": 455
+    },
+    {
+      "epoch": 0.7296,
+      "grad_norm": 3.5392863750457764,
+      "learning_rate": 0.0001344,
+      "loss": 2.9468,
+      "step": 456
+    },
+    {
+      "epoch": 0.7312,
+      "grad_norm": 5.1045732498168945,
+      "learning_rate": 0.0001347,
+      "loss": 2.7885,
+      "step": 457
+    },
+    {
+      "epoch": 0.7328,
+      "grad_norm": 6.027789115905762,
+      "learning_rate": 0.000135,
+      "loss": 2.9067,
+      "step": 458
+    },
+    {
+      "epoch": 0.7344,
+      "grad_norm": 5.094452381134033,
+      "learning_rate": 0.00013529999999999998,
+      "loss": 2.8379,
+      "step": 459
+    },
+    {
+      "epoch": 0.736,
+      "grad_norm": 2.6457953453063965,
+      "learning_rate": 0.0001356,
+      "loss": 2.8325,
+      "step": 460
+    },
+    {
+      "epoch": 0.7376,
+      "grad_norm": 1.5734143257141113,
+      "learning_rate": 0.0001359,
+      "loss": 2.8004,
+      "step": 461
+    },
+    {
+      "epoch": 0.7392,
+      "grad_norm": 2.7408978939056396,
+      "learning_rate": 0.0001362,
+      "loss": 2.5294,
+      "step": 462
+    },
+    {
+      "epoch": 0.7408,
+      "grad_norm": 3.2462551593780518,
+      "learning_rate": 0.00013649999999999998,
+      "loss": 2.5774,
+      "step": 463
+    },
+    {
+      "epoch": 0.7424,
+      "grad_norm": 5.122827529907227,
+      "learning_rate": 0.0001368,
+      "loss": 2.5412,
+      "step": 464
+    },
+    {
+      "epoch": 0.744,
+      "grad_norm": 6.828001976013184,
+      "learning_rate": 0.0001371,
+      "loss": 2.5768,
+      "step": 465
+    },
+    {
+      "epoch": 0.7456,
+      "grad_norm": 5.996628761291504,
+      "learning_rate": 0.0001374,
+      "loss": 2.5803,
+      "step": 466
+    },
+    {
+      "epoch": 0.7472,
+      "grad_norm": 3.842134714126587,
+      "learning_rate": 0.00013769999999999999,
+      "loss": 2.3747,
+      "step": 467
+    },
+    {
+      "epoch": 0.7488,
+      "grad_norm": 1.4524292945861816,
+      "learning_rate": 0.000138,
+      "loss": 2.4186,
+      "step": 468
+    },
+    {
+      "epoch": 0.7504,
+      "grad_norm": 1.6084707975387573,
+      "learning_rate": 0.0001383,
+      "loss": 2.3012,
+      "step": 469
+    },
+    {
+      "epoch": 0.752,
+      "grad_norm": 2.121351718902588,
+      "learning_rate": 0.0001386,
+      "loss": 2.442,
+      "step": 470
+    },
+    {
+      "epoch": 0.7536,
+      "grad_norm": 1.5034464597702026,
+      "learning_rate": 0.0001389,
+      "loss": 2.2728,
+      "step": 471
+    },
+    {
+      "epoch": 0.7552,
+      "grad_norm": 1.2867931127548218,
+      "learning_rate": 0.0001392,
+      "loss": 2.3669,
+      "step": 472
+    },
+    {
+      "epoch": 0.7568,
+      "grad_norm": 1.8455201387405396,
+      "learning_rate": 0.0001395,
+      "loss": 2.3831,
+      "step": 473
+    },
+    {
+      "epoch": 0.7584,
+      "grad_norm": 1.4569259881973267,
+      "learning_rate": 0.00013979999999999998,
+      "loss": 2.6884,
+      "step": 474
+    },
+    {
+      "epoch": 0.76,
+      "grad_norm": 1.9550684690475464,
+      "learning_rate": 0.0001401,
+      "loss": 2.4852,
+      "step": 475
+    },
+    {
+      "epoch": 0.7616,
+      "grad_norm": 2.876927137374878,
+      "learning_rate": 0.0001404,
+      "loss": 2.5227,
+      "step": 476
+    },
+    {
+      "epoch": 0.7632,
+      "grad_norm": 1.2651807069778442,
+      "learning_rate": 0.00014069999999999998,
+      "loss": 2.5994,
+      "step": 477
+    },
+    {
+      "epoch": 0.7648,
+      "grad_norm": 1.26189386844635,
+      "learning_rate": 0.00014099999999999998,
+      "loss": 2.2933,
+      "step": 478
+    },
+    {
+      "epoch": 0.7664,
+      "grad_norm": 1.3137550354003906,
+      "learning_rate": 0.0001413,
+      "loss": 2.3087,
+      "step": 479
+    },
+    {
+      "epoch": 0.768,
+      "grad_norm": 1.7220642566680908,
+      "learning_rate": 0.00014159999999999997,
+      "loss": 2.4592,
+      "step": 480
+    },
+    {
+      "epoch": 0.7696,
+      "grad_norm": 1.3261381387710571,
+      "learning_rate": 0.00014189999999999998,
+      "loss": 2.0056,
+      "step": 481
+    },
+    {
+      "epoch": 0.7712,
+      "grad_norm": 2.571230173110962,
+      "learning_rate": 0.0001422,
+      "loss": 2.2005,
+      "step": 482
+    },
+    {
+      "epoch": 0.7728,
+      "grad_norm": 1.9342719316482544,
+      "learning_rate": 0.0001425,
+      "loss": 2.444,
+      "step": 483
+    },
+    {
+      "epoch": 0.7744,
+      "grad_norm": 1.9060297012329102,
+      "learning_rate": 0.00014279999999999997,
+      "loss": 2.5657,
+      "step": 484
+    },
+    {
+      "epoch": 0.776,
+      "grad_norm": 1.7057262659072876,
+      "learning_rate": 0.00014309999999999998,
+      "loss": 2.2488,
+      "step": 485
+    },
+    {
+      "epoch": 0.7776,
+      "grad_norm": 1.5254745483398438,
+      "learning_rate": 0.0001434,
+      "loss": 2.3053,
+      "step": 486
+    },
+    {
+      "epoch": 0.7792,
+      "grad_norm": 1.2841426134109497,
+      "learning_rate": 0.00014369999999999997,
+      "loss": 2.7327,
+      "step": 487
+    },
+    {
+      "epoch": 0.7808,
+      "grad_norm": 1.2939062118530273,
+      "learning_rate": 0.00014399999999999998,
+      "loss": 2.1748,
+      "step": 488
+    },
+    {
+      "epoch": 0.7824,
+      "grad_norm": 1.041858434677124,
+      "learning_rate": 0.00014429999999999998,
+      "loss": 2.2685,
+      "step": 489
+    },
+    {
+      "epoch": 0.784,
+      "grad_norm": 1.1529954671859741,
+      "learning_rate": 0.0001446,
+      "loss": 2.6499,
+      "step": 490
+    },
+    {
+      "epoch": 0.7856,
+      "grad_norm": 1.2997585535049438,
+      "learning_rate": 0.00014489999999999997,
+      "loss": 2.4287,
+      "step": 491
+    },
+    {
+      "epoch": 0.7872,
+      "grad_norm": 1.8214664459228516,
+      "learning_rate": 0.00014519999999999998,
+      "loss": 2.4024,
+      "step": 492
+    },
+    {
+      "epoch": 0.7888,
+      "grad_norm": 2.8641598224639893,
+      "learning_rate": 0.00014549999999999999,
+      "loss": 2.3568,
+      "step": 493
+    },
+    {
+      "epoch": 0.7904,
+      "grad_norm": 2.793945789337158,
+      "learning_rate": 0.0001458,
+      "loss": 2.635,
+      "step": 494
+    },
+    {
+      "epoch": 0.792,
+      "grad_norm": 1.2558726072311401,
+      "learning_rate": 0.00014609999999999997,
+      "loss": 2.4789,
+      "step": 495
+    },
+    {
+      "epoch": 0.7936,
+      "grad_norm": 1.8537378311157227,
+      "learning_rate": 0.00014639999999999998,
+      "loss": 2.0977,
+      "step": 496
+    },
+    {
+      "epoch": 0.7952,
+      "grad_norm": 1.3181400299072266,
+      "learning_rate": 0.0001467,
+      "loss": 2.168,
+      "step": 497
+    },
+    {
+      "epoch": 0.7968,
+      "grad_norm": 9.861762046813965,
+      "learning_rate": 0.000147,
+      "loss": 3.4399,
+      "step": 498
+    },
+    {
+      "epoch": 0.7984,
+      "grad_norm": 2.7572944164276123,
+      "learning_rate": 0.00014729999999999998,
+      "loss": 2.4976,
+      "step": 499
+    },
+    {
+      "epoch": 0.8,
+      "grad_norm": 3.072735071182251,
+      "learning_rate": 0.00014759999999999998,
+      "loss": 3.0006,
+      "step": 500
+    },
+    {
+      "epoch": 0.8016,
+      "grad_norm": 3.723292350769043,
+      "learning_rate": 0.0001479,
+      "loss": 2.4371,
+      "step": 501
+    },
+    {
+      "epoch": 0.8032,
+      "grad_norm": 5.342506408691406,
+      "learning_rate": 0.0001482,
+      "loss": 3.0689,
+      "step": 502
+    },
+    {
+      "epoch": 0.8048,
+      "grad_norm": 5.763881683349609,
+      "learning_rate": 0.00014849999999999998,
+      "loss": 2.8854,
+      "step": 503
+    },
+    {
+      "epoch": 0.8064,
+      "grad_norm": 1.8335249423980713,
+      "learning_rate": 0.00014879999999999998,
+      "loss": 2.4936,
+      "step": 504
+    },
+    {
+      "epoch": 0.808,
+      "grad_norm": 2.8503644466400146,
+      "learning_rate": 0.0001491,
+      "loss": 2.4671,
+      "step": 505
+    },
+    {
+      "epoch": 0.8096,
+      "grad_norm": 5.93911600112915,
+      "learning_rate": 0.0001494,
+      "loss": 2.2154,
+      "step": 506
+    },
+    {
+      "epoch": 0.8112,
+      "grad_norm": 4.656365871429443,
+      "learning_rate": 0.00014969999999999998,
+      "loss": 2.2564,
+      "step": 507
+    },
+    {
+      "epoch": 0.8128,
+      "grad_norm": 4.47904109954834,
+      "learning_rate": 0.00015,
+      "loss": 2.4022,
+      "step": 508
+    },
+    {
+      "epoch": 0.8144,
+      "grad_norm": 2.0499017238616943,
+      "learning_rate": 0.0001503,
+      "loss": 1.9681,
+      "step": 509
+    },
+    {
+      "epoch": 0.816,
+      "grad_norm": 1.0935138463974,
+      "learning_rate": 0.00015059999999999997,
+      "loss": 2.0941,
+      "step": 510
+    },
+    {
+      "epoch": 0.8176,
+      "grad_norm": 2.3944854736328125,
+      "learning_rate": 0.00015089999999999998,
+      "loss": 2.0268,
+      "step": 511
+    },
+    {
+      "epoch": 0.8192,
+      "grad_norm": 6.021939277648926,
+      "learning_rate": 0.0001512,
+      "loss": 2.3145,
+      "step": 512
+    },
+    {
+      "epoch": 0.8208,
+      "grad_norm": 5.291767120361328,
+      "learning_rate": 0.0001515,
+      "loss": 2.038,
+      "step": 513
+    },
+    {
+      "epoch": 0.8224,
+      "grad_norm": 4.051759719848633,
+      "learning_rate": 0.00015179999999999998,
+      "loss": 1.8124,
+      "step": 514
+    },
+    {
+      "epoch": 0.824,
+      "grad_norm": 6.387513637542725,
+      "learning_rate": 0.00015209999999999998,
+      "loss": 2.114,
+      "step": 515
+    },
+    {
+      "epoch": 0.8256,
+      "grad_norm": 3.993975877761841,
+      "learning_rate": 0.0001524,
+      "loss": 1.9412,
+      "step": 516
+    },
+    {
+      "epoch": 0.8272,
+      "grad_norm": 2.036212682723999,
+      "learning_rate": 0.0001527,
+      "loss": 1.8678,
+      "step": 517
+    },
+    {
+      "epoch": 0.8288,
+      "grad_norm": 1.404420256614685,
+      "learning_rate": 0.00015299999999999998,
+      "loss": 2.1287,
+      "step": 518
+    },
+    {
+      "epoch": 0.8304,
+      "grad_norm": 1.0048662424087524,
+      "learning_rate": 0.00015329999999999999,
+      "loss": 1.9134,
+      "step": 519
+    },
+    {
+      "epoch": 0.832,
+      "grad_norm": 2.347856044769287,
+      "learning_rate": 0.0001536,
+      "loss": 1.8799,
+      "step": 520
+    },
+    {
+      "epoch": 0.8336,
+      "grad_norm": 3.0598201751708984,
+      "learning_rate": 0.0001539,
+      "loss": 1.9441,
+      "step": 521
+    },
+    {
+      "epoch": 0.8352,
+      "grad_norm": 2.636126756668091,
+      "learning_rate": 0.00015419999999999998,
+      "loss": 1.7355,
+      "step": 522
+    },
+    {
+      "epoch": 0.8368,
+      "grad_norm": 1.8599352836608887,
+      "learning_rate": 0.0001545,
+      "loss": 1.9851,
+      "step": 523
+    },
+    {
+      "epoch": 0.8384,
+      "grad_norm": 0.9748109579086304,
+      "learning_rate": 0.0001548,
+      "loss": 1.7774,
+      "step": 524
+    },
+    {
+      "epoch": 0.84,
+      "grad_norm": 1.414323091506958,
+      "learning_rate": 0.0001551,
+      "loss": 2.1997,
+      "step": 525
+    },
+    {
+      "epoch": 0.8416,
+      "grad_norm": 2.8852648735046387,
+      "learning_rate": 0.00015539999999999998,
+      "loss": 2.0179,
+      "step": 526
+    },
+    {
+      "epoch": 0.8432,
+      "grad_norm": 2.0136239528656006,
+      "learning_rate": 0.0001557,
+      "loss": 1.9451,
+      "step": 527
+    },
+    {
+      "epoch": 0.8448,
+      "grad_norm": 2.07312273979187,
+      "learning_rate": 0.000156,
+      "loss": 1.7522,
+      "step": 528
+    },
+    {
+      "epoch": 0.8464,
+      "grad_norm": 1.4143507480621338,
+      "learning_rate": 0.0001563,
+      "loss": 1.6561,
+      "step": 529
+    },
+    {
+      "epoch": 0.848,
+      "grad_norm": 3.017238140106201,
+      "learning_rate": 0.00015659999999999998,
+      "loss": 1.913,
+      "step": 530
+    },
+    {
+      "epoch": 0.8496,
+      "grad_norm": 0.9368352293968201,
+      "learning_rate": 0.0001569,
+      "loss": 1.8592,
+      "step": 531
+    },
+    {
+      "epoch": 0.8512,
+      "grad_norm": 1.308072566986084,
+      "learning_rate": 0.0001572,
+      "loss": 2.2341,
+      "step": 532
+    },
+    {
+      "epoch": 0.8528,
+      "grad_norm": 2.2798593044281006,
+      "learning_rate": 0.00015749999999999998,
+      "loss": 1.9506,
+      "step": 533
+    },
+    {
+      "epoch": 0.8544,
+      "grad_norm": 2.6132118701934814,
+      "learning_rate": 0.0001578,
+      "loss": 1.9343,
+      "step": 534
+    },
+    {
+      "epoch": 0.856,
+      "grad_norm": 1.162194848060608,
+      "learning_rate": 0.0001581,
+      "loss": 1.9341,
+      "step": 535
+    },
+    {
+      "epoch": 0.8576,
+      "grad_norm": 1.3427730798721313,
+      "learning_rate": 0.0001584,
+      "loss": 1.7395,
+      "step": 536
+    },
+    {
+      "epoch": 0.8592,
+      "grad_norm": 2.1670310497283936,
+      "learning_rate": 0.00015869999999999998,
+      "loss": 2.3282,
+      "step": 537
+    },
+    {
+      "epoch": 0.8608,
+      "grad_norm": 1.257582187652588,
+      "learning_rate": 0.000159,
+      "loss": 2.09,
+      "step": 538
+    },
+    {
+      "epoch": 0.8624,
+      "grad_norm": 1.4573386907577515,
+      "learning_rate": 0.0001593,
+      "loss": 1.8402,
+      "step": 539
+    },
+    {
+      "epoch": 0.864,
+      "grad_norm": 1.3384615182876587,
+      "learning_rate": 0.0001596,
+      "loss": 1.7193,
+      "step": 540
+    },
+    {
+      "epoch": 0.8656,
+      "grad_norm": 2.220402479171753,
+      "learning_rate": 0.00015989999999999998,
+      "loss": 1.5656,
+      "step": 541
+    },
+    {
+      "epoch": 0.8672,
+      "grad_norm": 2.4653773307800293,
+      "learning_rate": 0.0001602,
+      "loss": 2.0628,
+      "step": 542
+    },
+    {
+      "epoch": 0.8688,
+      "grad_norm": 1.280678391456604,
+      "learning_rate": 0.0001605,
+      "loss": 1.8363,
+      "step": 543
+    },
+    {
+      "epoch": 0.8704,
+      "grad_norm": 2.4655933380126953,
+      "learning_rate": 0.0001608,
+      "loss": 1.7545,
+      "step": 544
+    },
+    {
+      "epoch": 0.872,
+      "grad_norm": 1.506415605545044,
+      "learning_rate": 0.00016109999999999999,
+      "loss": 1.8381,
+      "step": 545
+    },
+    {
+      "epoch": 0.8736,
+      "grad_norm": 1.1475555896759033,
+      "learning_rate": 0.0001614,
+      "loss": 1.829,
+      "step": 546
+    },
+    {
+      "epoch": 0.8752,
+      "grad_norm": 1.4434545040130615,
+      "learning_rate": 0.0001617,
+      "loss": 1.8184,
+      "step": 547
+    },
+    {
+      "epoch": 0.8768,
+      "grad_norm": 1.8260152339935303,
+      "learning_rate": 0.000162,
+      "loss": 1.9946,
+      "step": 548
+    },
+    {
+      "epoch": 0.8784,
+      "grad_norm": 1.8104926347732544,
+      "learning_rate": 0.0001623,
+      "loss": 2.031,
+      "step": 549
+    },
+    {
+      "epoch": 0.88,
+      "grad_norm": 2.094877243041992,
+      "learning_rate": 0.0001626,
+      "loss": 2.2711,
+      "step": 550
+    },
+    {
+      "epoch": 0.8816,
+      "grad_norm": 23.733247756958008,
+      "learning_rate": 0.0001629,
+      "loss": 4.8174,
+      "step": 551
+    },
+    {
+      "epoch": 0.8832,
+      "grad_norm": 12.243576049804688,
+      "learning_rate": 0.0001632,
+      "loss": 3.3335,
+      "step": 552
+    },
+    {
+      "epoch": 0.8848,
+      "grad_norm": 3.982137441635132,
+      "learning_rate": 0.0001635,
+      "loss": 2.3234,
+      "step": 553
+    },
+    {
+      "epoch": 0.8864,
+      "grad_norm": 2.5422203540802,
+      "learning_rate": 0.0001638,
+      "loss": 2.1768,
+      "step": 554
+    },
+    {
+      "epoch": 0.888,
+      "grad_norm": 2.649517059326172,
+      "learning_rate": 0.0001641,
+      "loss": 2.022,
+      "step": 555
+    },
+    {
+      "epoch": 0.8896,
+      "grad_norm": 4.723710536956787,
+      "learning_rate": 0.0001644,
+      "loss": 1.8511,
+      "step": 556
+    },
+    {
+      "epoch": 0.8912,
+      "grad_norm": 2.3035788536071777,
+      "learning_rate": 0.0001647,
+      "loss": 1.9528,
+      "step": 557
+    },
+    {
+      "epoch": 0.8928,
+      "grad_norm": 3.8410518169403076,
+      "learning_rate": 0.000165,
+      "loss": 1.9104,
+      "step": 558
+    },
+    {
+      "epoch": 0.8944,
+      "grad_norm": 3.0108225345611572,
+      "learning_rate": 0.0001653,
+      "loss": 1.7834,
+      "step": 559
+    },
+    {
+      "epoch": 0.896,
+      "grad_norm": 1.3487671613693237,
+      "learning_rate": 0.0001656,
+      "loss": 1.883,
+      "step": 560
+    },
+    {
+      "epoch": 0.8976,
+      "grad_norm": 1.061733365058899,
+      "learning_rate": 0.0001659,
+      "loss": 1.5688,
+      "step": 561
+    },
+    {
+      "epoch": 0.8992,
+      "grad_norm": 2.0784027576446533,
+      "learning_rate": 0.0001662,
+      "loss": 1.6914,
+      "step": 562
+    },
+    {
+      "epoch": 0.9008,
+      "grad_norm": 6.085043907165527,
+      "learning_rate": 0.0001665,
+      "loss": 2.3407,
+      "step": 563
+    },
+    {
+      "epoch": 0.9024,
+      "grad_norm": 1.459148645401001,
+      "learning_rate": 0.0001668,
+      "loss": 1.7104,
+      "step": 564
+    },
+    {
+      "epoch": 0.904,
+      "grad_norm": 1.9622076749801636,
+      "learning_rate": 0.0001671,
+      "loss": 1.5955,
+      "step": 565
+    },
+    {
+      "epoch": 0.9056,
+      "grad_norm": 1.2756608724594116,
+      "learning_rate": 0.0001674,
+      "loss": 1.4071,
+      "step": 566
+    },
+    {
+      "epoch": 0.9072,
+      "grad_norm": 0.940319299697876,
+      "learning_rate": 0.0001677,
+      "loss": 1.6557,
+      "step": 567
+    },
+    {
+      "epoch": 0.9088,
+      "grad_norm": 0.9497667551040649,
+      "learning_rate": 0.000168,
+      "loss": 1.774,
+      "step": 568
+    },
+    {
+      "epoch": 0.9104,
+      "grad_norm": 1.1930807828903198,
+      "learning_rate": 0.0001683,
+      "loss": 1.8378,
+      "step": 569
+    },
+    {
+      "epoch": 0.912,
+      "grad_norm": 1.7330429553985596,
+      "learning_rate": 0.0001686,
+      "loss": 1.6816,
+      "step": 570
+    },
+    {
+      "epoch": 0.9136,
+      "grad_norm": 0.9604584574699402,
+      "learning_rate": 0.00016889999999999996,
+      "loss": 1.6782,
+      "step": 571
+    },
+    {
+      "epoch": 0.9152,
+      "grad_norm": 0.9503042101860046,
+      "learning_rate": 0.00016919999999999997,
+      "loss": 1.5947,
+      "step": 572
+    },
+    {
+      "epoch": 0.9168,
+      "grad_norm": 1.1088024377822876,
+      "learning_rate": 0.00016949999999999997,
+      "loss": 1.6978,
+      "step": 573
+    },
+    {
+      "epoch": 0.9184,
+      "grad_norm": 1.118318796157837,
+      "learning_rate": 0.00016979999999999998,
+      "loss": 1.656,
+      "step": 574
+    },
+    {
+      "epoch": 0.92,
+      "grad_norm": 1.5163230895996094,
+      "learning_rate": 0.00017009999999999996,
+      "loss": 1.6588,
+      "step": 575
+    },
+    {
+      "epoch": 0.9216,
+      "grad_norm": 1.4612356424331665,
+      "learning_rate": 0.00017039999999999997,
+      "loss": 1.9119,
+      "step": 576
+    },
+    {
+      "epoch": 0.9232,
+      "grad_norm": 1.2807903289794922,
+      "learning_rate": 0.00017069999999999998,
+      "loss": 1.4299,
+      "step": 577
+    },
+    {
+      "epoch": 0.9248,
+      "grad_norm": 1.049907922744751,
+      "learning_rate": 0.00017099999999999998,
+      "loss": 1.3226,
+      "step": 578
+    },
+    {
+      "epoch": 0.9264,
+      "grad_norm": 1.0162078142166138,
+      "learning_rate": 0.00017129999999999996,
+      "loss": 1.8021,
+      "step": 579
+    },
+    {
+      "epoch": 0.928,
+      "grad_norm": 1.3673537969589233,
+      "learning_rate": 0.00017159999999999997,
+      "loss": 1.6087,
+      "step": 580
+    },
+    {
+      "epoch": 0.9296,
+      "grad_norm": 1.2779172658920288,
+      "learning_rate": 0.00017189999999999998,
+      "loss": 1.6225,
+      "step": 581
+    },
+    {
+      "epoch": 0.9312,
+      "grad_norm": 1.2135735750198364,
+      "learning_rate": 0.00017219999999999998,
+      "loss": 1.5889,
+      "step": 582
+    },
+    {
+      "epoch": 0.9328,
+      "grad_norm": 1.45180344581604,
+      "learning_rate": 0.00017249999999999996,
+      "loss": 2.1697,
+      "step": 583
+    },
+    {
+      "epoch": 0.9344,
+      "grad_norm": 1.1630367040634155,
+      "learning_rate": 0.00017279999999999997,
+      "loss": 1.79,
+      "step": 584
+    },
+    {
+      "epoch": 0.936,
+      "grad_norm": 2.428530693054199,
+      "learning_rate": 0.00017309999999999998,
+      "loss": 1.5455,
+      "step": 585
+    },
+    {
+      "epoch": 0.9376,
+      "grad_norm": 1.3975725173950195,
+      "learning_rate": 0.00017339999999999996,
+      "loss": 1.7658,
+      "step": 586
+    },
+    {
+      "epoch": 0.9392,
+      "grad_norm": 1.242210865020752,
+      "learning_rate": 0.00017369999999999997,
+      "loss": 1.8039,
+      "step": 587
+    },
+    {
+      "epoch": 0.9408,
+      "grad_norm": 1.071577787399292,
+      "learning_rate": 0.00017399999999999997,
+      "loss": 1.7215,
+      "step": 588
+    },
+    {
+      "epoch": 0.9424,
+      "grad_norm": 1.208039402961731,
+      "learning_rate": 0.00017429999999999998,
+      "loss": 1.8733,
+      "step": 589
+    },
+    {
+      "epoch": 0.944,
+      "grad_norm": 1.5233865976333618,
+      "learning_rate": 0.00017459999999999996,
+      "loss": 1.4408,
+      "step": 590
+    },
+    {
+      "epoch": 0.9456,
+      "grad_norm": 1.411783218383789,
+      "learning_rate": 0.00017489999999999997,
+      "loss": 1.8393,
+      "step": 591
+    },
+    {
+      "epoch": 0.9472,
+      "grad_norm": 1.629401683807373,
+      "learning_rate": 0.00017519999999999998,
+      "loss": 1.679,
+      "step": 592
+    },
+    {
+      "epoch": 0.9488,
+      "grad_norm": 1.487720012664795,
+      "learning_rate": 0.00017549999999999998,
+      "loss": 1.9937,
+      "step": 593
+    },
+    {
+      "epoch": 0.9504,
+      "grad_norm": 1.7428632974624634,
+      "learning_rate": 0.00017579999999999996,
+      "loss": 1.8585,
+      "step": 594
+    },
+    {
+      "epoch": 0.952,
+      "grad_norm": 1.5290313959121704,
+      "learning_rate": 0.00017609999999999997,
+      "loss": 1.758,
+      "step": 595
+    },
+    {
+      "epoch": 0.9536,
+      "grad_norm": 1.4210582971572876,
+      "learning_rate": 0.00017639999999999998,
+      "loss": 1.6403,
+      "step": 596
+    },
+    {
+      "epoch": 0.9552,
+      "grad_norm": 1.487386703491211,
+      "learning_rate": 0.00017669999999999999,
+      "loss": 2.0706,
+      "step": 597
+    },
+    {
+      "epoch": 0.9568,
+      "grad_norm": 1.789679765701294,
+      "learning_rate": 0.00017699999999999997,
+      "loss": 2.0324,
+      "step": 598
+    },
+    {
+      "epoch": 0.9584,
+      "grad_norm": 3.552408456802368,
+      "learning_rate": 0.00017729999999999997,
+      "loss": 2.4765,
+      "step": 599
+    },
+    {
+      "epoch": 0.96,
+      "grad_norm": 2.6970980167388916,
+      "learning_rate": 0.00017759999999999998,
+      "loss": 2.49,
+      "step": 600
+    },
+    {
+      "epoch": 0.9616,
+      "grad_norm": 6.3989667892456055,
+      "learning_rate": 0.0001779,
+      "loss": 2.2124,
+      "step": 601
+    },
+    {
+      "epoch": 0.9632,
+      "grad_norm": 3.559483528137207,
+      "learning_rate": 0.00017819999999999997,
+      "loss": 1.9114,
+      "step": 602
+    },
+    {
+      "epoch": 0.9648,
+      "grad_norm": 2.688811779022217,
+      "learning_rate": 0.00017849999999999997,
+      "loss": 1.7274,
+      "step": 603
+    },
+    {
+      "epoch": 0.9664,
+      "grad_norm": 1.4167048931121826,
+      "learning_rate": 0.00017879999999999998,
+      "loss": 1.5342,
+      "step": 604
+    },
+    {
+      "epoch": 0.968,
+      "grad_norm": 1.0234233140945435,
+      "learning_rate": 0.0001791,
+      "loss": 1.6476,
+      "step": 605
+    },
+    {
+      "epoch": 0.9696,
+      "grad_norm": 2.3607473373413086,
+      "learning_rate": 0.00017939999999999997,
+      "loss": 1.8034,
+      "step": 606
+    },
+    {
+      "epoch": 0.9712,
+      "grad_norm": 1.8193793296813965,
+      "learning_rate": 0.00017969999999999998,
+      "loss": 1.2502,
+      "step": 607
+    },
+    {
+      "epoch": 0.9728,
+      "grad_norm": 2.5050389766693115,
+      "learning_rate": 0.00017999999999999998,
+      "loss": 1.7518,
+      "step": 608
+    },
+    {
+      "epoch": 0.9744,
+      "grad_norm": 1.852980375289917,
+      "learning_rate": 0.00018029999999999996,
+      "loss": 2.2657,
+      "step": 609
+    },
+    {
+      "epoch": 0.976,
+      "grad_norm": 1.1846544742584229,
+      "learning_rate": 0.00018059999999999997,
+      "loss": 1.6213,
+      "step": 610
+    },
+    {
+      "epoch": 0.9776,
+      "grad_norm": 1.1806446313858032,
+      "learning_rate": 0.00018089999999999998,
+      "loss": 1.5566,
+      "step": 611
+    },
+    {
+      "epoch": 0.9792,
+      "grad_norm": 0.9722961187362671,
+      "learning_rate": 0.00018119999999999999,
+      "loss": 1.4004,
+      "step": 612
+    },
+    {
+      "epoch": 0.9808,
+      "grad_norm": 1.2534488439559937,
+      "learning_rate": 0.00018149999999999997,
+      "loss": 1.9613,
+      "step": 613
+    },
+    {
+      "epoch": 0.9824,
+      "grad_norm": 1.55427885055542,
+      "learning_rate": 0.00018179999999999997,
+      "loss": 1.3668,
+      "step": 614
+    },
+    {
+      "epoch": 0.984,
+      "grad_norm": 1.8559104204177856,
+      "learning_rate": 0.00018209999999999998,
+      "loss": 1.153,
+      "step": 615
+    },
+    {
+      "epoch": 0.9856,
+      "grad_norm": 1.3127942085266113,
+      "learning_rate": 0.0001824,
+      "loss": 1.6635,
+      "step": 616
+    },
+    {
+      "epoch": 0.9872,
+      "grad_norm": 1.3206202983856201,
+      "learning_rate": 0.00018269999999999997,
+      "loss": 1.5436,
+      "step": 617
+    },
+    {
+      "epoch": 0.9888,
+      "grad_norm": 1.0405744314193726,
+      "learning_rate": 0.00018299999999999998,
+      "loss": 1.6072,
+      "step": 618
+    },
+    {
+      "epoch": 0.9904,
+      "grad_norm": 1.1208364963531494,
+      "learning_rate": 0.00018329999999999998,
+      "loss": 1.3522,
+      "step": 619
+    },
+    {
+      "epoch": 0.992,
+      "grad_norm": 1.4611485004425049,
+      "learning_rate": 0.0001836,
+      "loss": 1.8288,
+      "step": 620
+    },
+    {
+      "epoch": 0.9936,
+      "grad_norm": 2.102464199066162,
+      "learning_rate": 0.00018389999999999997,
+      "loss": 2.2311,
+      "step": 621
+    },
+    {
+      "epoch": 0.9952,
+      "grad_norm": 1.3121858835220337,
+      "learning_rate": 0.00018419999999999998,
+      "loss": 1.6955,
+      "step": 622
+    },
+    {
+      "epoch": 0.9968,
+      "grad_norm": 1.732784390449524,
+      "learning_rate": 0.00018449999999999999,
+      "loss": 2.066,
+      "step": 623
+    },
+    {
+      "epoch": 0.9984,
+      "grad_norm": 1.474577784538269,
+      "learning_rate": 0.0001848,
+      "loss": 1.7517,
+      "step": 624
+    },
+    {
+      "epoch": 1.0,
+      "grad_norm": NaN,
+      "learning_rate": 0.0001848,
+      "loss": 2.8279,
+      "step": 625
+    },
+    {
+      "epoch": 1.0016,
+      "grad_norm": 21.10396385192871,
+      "learning_rate": 0.00018509999999999997,
+      "loss": 4.359,
+      "step": 626
+    },
+    {
+      "epoch": 1.0032,
+      "grad_norm": 2.5289759635925293,
+      "learning_rate": 0.00018539999999999998,
+      "loss": 1.5803,
+      "step": 627
+    },
+    {
+      "epoch": 1.0048,
+      "grad_norm": 17.63152503967285,
+      "learning_rate": 0.0001857,
+      "loss": 4.1665,
+      "step": 628
+    },
+    {
+      "epoch": 1.0064,
+      "grad_norm": 1.2565017938613892,
+      "learning_rate": 0.000186,
+      "loss": 1.5211,
+      "step": 629
+    },
+    {
+      "epoch": 1.008,
+      "grad_norm": 0.9237573146820068,
+      "learning_rate": 0.00018629999999999997,
+      "loss": 1.6206,
+      "step": 630
+    },
+    {
+      "epoch": 1.0096,
+      "grad_norm": 1.304307222366333,
+      "learning_rate": 0.00018659999999999998,
+      "loss": 1.7635,
+      "step": 631
+    },
+    {
+      "epoch": 1.0112,
+      "grad_norm": 2.240795850753784,
+      "learning_rate": 0.0001869,
+      "loss": 1.4183,
+      "step": 632
+    },
+    {
+      "epoch": 1.0128,
+      "grad_norm": 1.2945712804794312,
+      "learning_rate": 0.0001872,
+      "loss": 1.8278,
+      "step": 633
+    },
+    {
+      "epoch": 1.0144,
+      "grad_norm": 2.4284050464630127,
+      "learning_rate": 0.00018749999999999998,
+      "loss": 1.8362,
+      "step": 634
+    },
+    {
+      "epoch": 1.016,
+      "grad_norm": 1.5324746370315552,
+      "learning_rate": 0.00018779999999999998,
+      "loss": 1.3312,
+      "step": 635
+    },
+    {
+      "epoch": 1.0176,
+      "grad_norm": 0.9457862973213196,
+      "learning_rate": 0.0001881,
+      "loss": 1.5771,
+      "step": 636
+    },
+    {
+      "epoch": 1.0192,
+      "grad_norm": 1.761409878730774,
+      "learning_rate": 0.00018839999999999997,
+      "loss": 1.3939,
+      "step": 637
+    },
+    {
+      "epoch": 1.0208,
+      "grad_norm": 2.4509124755859375,
+      "learning_rate": 0.00018869999999999998,
+      "loss": 1.305,
+      "step": 638
+    },
+    {
+      "epoch": 1.0224,
+      "grad_norm": 1.434770941734314,
+      "learning_rate": 0.00018899999999999999,
+      "loss": 1.2198,
+      "step": 639
+    },
+    {
+      "epoch": 1.024,
+      "grad_norm": 1.683680772781372,
+      "learning_rate": 0.0001893,
+      "loss": 1.4401,
+      "step": 640
+    },
+    {
+      "epoch": 1.0256,
+      "grad_norm": 1.468677282333374,
+      "learning_rate": 0.00018959999999999997,
+      "loss": 1.3005,
+      "step": 641
+    },
+    {
+      "epoch": 1.0272,
+      "grad_norm": 2.2306525707244873,
+      "learning_rate": 0.00018989999999999998,
+      "loss": 1.6153,
+      "step": 642
+    },
+    {
+      "epoch": 1.0288,
+      "grad_norm": 4.796661853790283,
+      "learning_rate": 0.0001902,
+      "loss": 2.2058,
+      "step": 643
+    },
+    {
+      "epoch": 1.0304,
+      "grad_norm": 1.139748454093933,
+      "learning_rate": 0.0001905,
+      "loss": 1.3829,
+      "step": 644
+    },
+    {
+      "epoch": 1.032,
+      "grad_norm": 1.9971469640731812,
+      "learning_rate": 0.00019079999999999998,
+      "loss": 1.5598,
+      "step": 645
+    },
+    {
+      "epoch": 1.0336,
+      "grad_norm": 2.224128007888794,
+      "learning_rate": 0.00019109999999999998,
+      "loss": 1.3654,
+      "step": 646
+    },
+    {
+      "epoch": 1.0352,
+      "grad_norm": 2.5159313678741455,
+      "learning_rate": 0.0001914,
+      "loss": 1.6379,
+      "step": 647
+    },
+    {
+      "epoch": 1.0368,
+      "grad_norm": 1.9604592323303223,
+      "learning_rate": 0.0001917,
+      "loss": 1.5734,
+      "step": 648
+    },
+    {
+      "epoch": 1.0384,
+      "grad_norm": 1.4151877164840698,
+      "learning_rate": 0.00019199999999999998,
+      "loss": 1.1349,
+      "step": 649
+    },
+    {
+      "epoch": 1.04,
+      "grad_norm": 1.21165931224823,
+      "learning_rate": 0.00019229999999999999,
+      "loss": 1.7592,
+      "step": 650
+    },
+    {
+      "epoch": 1.0416,
+      "grad_norm": 2.344447135925293,
+      "learning_rate": 0.0001926,
+      "loss": 1.505,
+      "step": 651
+    },
+    {
+      "epoch": 1.0432,
+      "grad_norm": 2.5432910919189453,
+      "learning_rate": 0.0001929,
+      "loss": 1.6673,
+      "step": 652
+    },
+    {
+      "epoch": 1.0448,
+      "grad_norm": 1.5895689725875854,
+      "learning_rate": 0.00019319999999999998,
+      "loss": 1.6617,
+      "step": 653
+    },
+    {
+      "epoch": 1.0464,
+      "grad_norm": 1.7360563278198242,
+      "learning_rate": 0.0001935,
+      "loss": 1.6216,
+      "step": 654
+    },
+    {
+      "epoch": 1.048,
+      "grad_norm": 1.3723790645599365,
+      "learning_rate": 0.0001938,
+      "loss": 1.257,
+      "step": 655
+    },
+    {
+      "epoch": 1.0496,
+      "grad_norm": 0.8750591278076172,
+      "learning_rate": 0.0001941,
+      "loss": 1.4356,
+      "step": 656
+    },
+    {
+      "epoch": 1.0512,
+      "grad_norm": 1.407861590385437,
+      "learning_rate": 0.00019439999999999998,
+      "loss": 1.346,
+      "step": 657
+    },
+    {
+      "epoch": 1.0528,
+      "grad_norm": 1.2812424898147583,
+      "learning_rate": 0.0001947,
+      "loss": 1.2363,
+      "step": 658
+    },
+    {
+      "epoch": 1.0544,
+      "grad_norm": 1.2920845746994019,
+      "learning_rate": 0.000195,
+      "loss": 1.4353,
+      "step": 659
+    },
+    {
+      "epoch": 1.056,
+      "grad_norm": 1.0122877359390259,
+      "learning_rate": 0.00019529999999999998,
+      "loss": 1.5272,
+      "step": 660
+    },
+    {
+      "epoch": 1.0576,
+      "grad_norm": 1.0607578754425049,
+      "learning_rate": 0.00019559999999999998,
+      "loss": 1.1926,
+      "step": 661
+    },
+    {
+      "epoch": 1.0592,
+      "grad_norm": 1.2849078178405762,
+      "learning_rate": 0.0001959,
+      "loss": 1.361,
+      "step": 662
+    },
+    {
+      "epoch": 1.0608,
+      "grad_norm": 2.199488401412964,
+      "learning_rate": 0.0001962,
+      "loss": 1.4892,
+      "step": 663
+    },
+    {
+      "epoch": 1.0624,
+      "grad_norm": 1.7300806045532227,
+      "learning_rate": 0.00019649999999999998,
+      "loss": 1.4795,
+      "step": 664
+    },
+    {
+      "epoch": 1.064,
+      "grad_norm": 1.210700273513794,
+      "learning_rate": 0.00019679999999999999,
+      "loss": 1.6863,
+      "step": 665
+    },
+    {
+      "epoch": 1.0656,
+      "grad_norm": 1.1998845338821411,
+      "learning_rate": 0.0001971,
+      "loss": 1.3863,
+      "step": 666
+    },
+    {
+      "epoch": 1.0672,
+      "grad_norm": 1.5421574115753174,
+      "learning_rate": 0.0001974,
+      "loss": 1.9558,
+      "step": 667
+    },
+    {
+      "epoch": 1.0688,
+      "grad_norm": 2.3596279621124268,
+      "learning_rate": 0.00019769999999999998,
+      "loss": 1.3471,
+      "step": 668
+    },
+    {
+      "epoch": 1.0704,
+      "grad_norm": 1.3288168907165527,
+      "learning_rate": 0.000198,
+      "loss": 1.3686,
+      "step": 669
+    },
+    {
+      "epoch": 1.072,
+      "grad_norm": 1.5977771282196045,
+      "learning_rate": 0.0001983,
+      "loss": 1.6142,
+      "step": 670
+    },
+    {
+      "epoch": 1.0735999999999999,
+      "grad_norm": 1.171886682510376,
+      "learning_rate": 0.0001986,
+      "loss": 1.8817,
+      "step": 671
+    },
+    {
+      "epoch": 1.0752,
+      "grad_norm": 1.4820473194122314,
+      "learning_rate": 0.00019889999999999998,
+      "loss": 1.502,
+      "step": 672
+    },
+    {
+      "epoch": 1.0768,
+      "grad_norm": 1.4286924600601196,
+      "learning_rate": 0.0001992,
+      "loss": 1.9869,
+      "step": 673
+    },
+    {
+      "epoch": 1.0784,
+      "grad_norm": 1.496476650238037,
+      "learning_rate": 0.0001995,
+      "loss": 1.5545,
+      "step": 674
+    },
+    {
+      "epoch": 1.08,
+      "grad_norm": 11.650896072387695,
+      "learning_rate": 0.0001998,
+      "loss": 3.5297,
+      "step": 675
+    },
+    {
+      "epoch": 1.0816,
+      "grad_norm": 10.930564880371094,
+      "learning_rate": 0.00020009999999999998,
+      "loss": 3.5244,
+      "step": 676
+    },
+    {
+      "epoch": 1.0832,
+      "grad_norm": 4.526219367980957,
+      "learning_rate": 0.0002004,
+      "loss": 1.948,
+      "step": 677
+    },
+    {
+      "epoch": 1.0848,
+      "grad_norm": 1.8217471837997437,
+      "learning_rate": 0.0002007,
+      "loss": 1.6832,
+      "step": 678
+    },
+    {
+      "epoch": 1.0864,
+      "grad_norm": 2.5544323921203613,
+      "learning_rate": 0.000201,
+      "loss": 2.3308,
+      "step": 679
+    },
+    {
+      "epoch": 1.088,
+      "grad_norm": 2.732450246810913,
+      "learning_rate": 0.0002013,
+      "loss": 1.7663,
+      "step": 680
+    },
+    {
+      "epoch": 1.0896,
+      "grad_norm": 4.002326488494873,
+      "learning_rate": 0.0002016,
+      "loss": 1.9597,
+      "step": 681
+    },
+    {
+      "epoch": 1.0912,
+      "grad_norm": 2.9579389095306396,
+      "learning_rate": 0.0002019,
+      "loss": 1.6625,
+      "step": 682
+    },
+    {
+      "epoch": 1.0928,
+      "grad_norm": 3.6762917041778564,
+      "learning_rate": 0.0002022,
+      "loss": 1.4949,
+      "step": 683
+    },
+    {
+      "epoch": 1.0944,
+      "grad_norm": 2.8355441093444824,
+      "learning_rate": 0.0002025,
+      "loss": 1.5695,
+      "step": 684
+    },
+    {
+      "epoch": 1.096,
+      "grad_norm": 2.894350290298462,
+      "learning_rate": 0.0002028,
+      "loss": 1.5717,
+      "step": 685
+    },
+    {
+      "epoch": 1.0976,
+      "grad_norm": 4.992308616638184,
+      "learning_rate": 0.0002031,
+      "loss": 1.7573,
+      "step": 686
+    },
+    {
+      "epoch": 1.0992,
+      "grad_norm": 1.175133466720581,
+      "learning_rate": 0.00020339999999999998,
+      "loss": 1.2767,
+      "step": 687
+    },
+    {
+      "epoch": 1.1008,
+      "grad_norm": 0.7449688911437988,
+      "learning_rate": 0.0002037,
+      "loss": 1.4346,
+      "step": 688
+    },
+    {
+      "epoch": 1.1024,
+      "grad_norm": 2.100440740585327,
+      "learning_rate": 0.000204,
+      "loss": 1.5286,
+      "step": 689
+    },
+    {
+      "epoch": 1.104,
+      "grad_norm": 1.06446373462677,
+      "learning_rate": 0.0002043,
+      "loss": 1.4716,
+      "step": 690
+    },
+    {
+      "epoch": 1.1056,
+      "grad_norm": 1.1813894510269165,
+      "learning_rate": 0.00020459999999999999,
+      "loss": 1.729,
+      "step": 691
+    },
+    {
+      "epoch": 1.1072,
+      "grad_norm": 1.2244285345077515,
+      "learning_rate": 0.0002049,
+      "loss": 1.5456,
+      "step": 692
+    },
+    {
+      "epoch": 1.1088,
+      "grad_norm": 1.395920991897583,
+      "learning_rate": 0.0002052,
+      "loss": 1.6253,
+      "step": 693
+    },
+    {
+      "epoch": 1.1104,
+      "grad_norm": 0.8973720073699951,
+      "learning_rate": 0.0002055,
+      "loss": 1.3474,
+      "step": 694
+    },
+    {
+      "epoch": 1.112,
+      "grad_norm": 0.9351361393928528,
+      "learning_rate": 0.0002058,
+      "loss": 1.5375,
+      "step": 695
+    },
+    {
+      "epoch": 1.1136,
+      "grad_norm": 0.9488412737846375,
+      "learning_rate": 0.0002061,
+      "loss": 1.2332,
+      "step": 696
+    },
+    {
+      "epoch": 1.1152,
+      "grad_norm": 0.800336480140686,
+      "learning_rate": 0.00020639999999999998,
+      "loss": 1.3265,
+      "step": 697
+    },
+    {
+      "epoch": 1.1168,
+      "grad_norm": 1.771794319152832,
+      "learning_rate": 0.00020669999999999996,
+      "loss": 1.3347,
+      "step": 698
+    },
+    {
+      "epoch": 1.1184,
+      "grad_norm": 3.4581542015075684,
+      "learning_rate": 0.00020699999999999996,
+      "loss": 1.783,
+      "step": 699
+    },
+    {
+      "epoch": 1.12,
+      "grad_norm": 0.837477445602417,
+      "learning_rate": 0.00020729999999999997,
+      "loss": 1.3295,
+      "step": 700
+    },
+    {
+      "epoch": 1.1216,
+      "grad_norm": 2.1295042037963867,
+      "learning_rate": 0.00020759999999999998,
+      "loss": 1.4163,
+      "step": 701
+    },
+    {
+      "epoch": 1.1232,
+      "grad_norm": 1.0342674255371094,
+      "learning_rate": 0.00020789999999999996,
+      "loss": 1.2891,
+      "step": 702
+    },
+    {
+      "epoch": 1.1248,
+      "grad_norm": 1.1783955097198486,
+      "learning_rate": 0.00020819999999999996,
+      "loss": 1.5386,
+      "step": 703
+    },
+    {
+      "epoch": 1.1264,
+      "grad_norm": 2.5988528728485107,
+      "learning_rate": 0.00020849999999999997,
+      "loss": 1.5942,
+      "step": 704
+    },
+    {
+      "epoch": 1.1280000000000001,
+      "grad_norm": 1.206281065940857,
+      "learning_rate": 0.00020879999999999998,
+      "loss": 1.4828,
+      "step": 705
+    },
+    {
+      "epoch": 1.1296,
+      "grad_norm": 1.60711669921875,
+      "learning_rate": 0.00020909999999999996,
+      "loss": 1.6748,
+      "step": 706
+    },
+    {
+      "epoch": 1.1312,
+      "grad_norm": 1.3890515565872192,
+      "learning_rate": 0.00020939999999999997,
+      "loss": 1.4464,
+      "step": 707
+    },
+    {
+      "epoch": 1.1328,
+      "grad_norm": 1.4788490533828735,
+      "learning_rate": 0.00020969999999999997,
+      "loss": 1.9039,
+      "step": 708
+    },
+    {
+      "epoch": 1.1344,
+      "grad_norm": 0.9197102189064026,
+      "learning_rate": 0.00020999999999999998,
+      "loss": 1.5258,
+      "step": 709
+    },
+    {
+      "epoch": 1.1360000000000001,
+      "grad_norm": 3.082664728164673,
+      "learning_rate": 0.00021029999999999996,
+      "loss": 1.6637,
+      "step": 710
+    },
+    {
+      "epoch": 1.1376,
+      "grad_norm": 1.3979014158248901,
+      "learning_rate": 0.00021059999999999997,
+      "loss": 1.3499,
+      "step": 711
+    },
+    {
+      "epoch": 1.1392,
+      "grad_norm": 2.7370402812957764,
+      "learning_rate": 0.00021089999999999998,
+      "loss": 1.7379,
+      "step": 712
+    },
+    {
+      "epoch": 1.1408,
+      "grad_norm": 1.36969792842865,
+      "learning_rate": 0.00021119999999999996,
+      "loss": 1.3235,
+      "step": 713
+    },
+    {
+      "epoch": 1.1424,
+      "grad_norm": 1.3009356260299683,
+      "learning_rate": 0.00021149999999999996,
+      "loss": 1.2209,
+      "step": 714
+    },
+    {
+      "epoch": 1.144,
+      "grad_norm": 1.0813698768615723,
+      "learning_rate": 0.00021179999999999997,
+      "loss": 1.237,
+      "step": 715
+    },
+    {
+      "epoch": 1.1456,
+      "grad_norm": 1.2386032342910767,
+      "learning_rate": 0.00021209999999999998,
+      "loss": 1.5799,
+      "step": 716
+    },
+    {
+      "epoch": 1.1472,
+      "grad_norm": 1.7847639322280884,
+      "learning_rate": 0.00021239999999999996,
+      "loss": 1.852,
+      "step": 717
+    },
+    {
+      "epoch": 1.1488,
+      "grad_norm": 1.4111274480819702,
+      "learning_rate": 0.00021269999999999997,
+      "loss": 1.7199,
+      "step": 718
+    },
+    {
+      "epoch": 1.1504,
+      "grad_norm": 1.6253108978271484,
+      "learning_rate": 0.00021299999999999997,
+      "loss": 1.6921,
+      "step": 719
+    },
+    {
+      "epoch": 1.152,
+      "grad_norm": 1.3691827058792114,
+      "learning_rate": 0.00021329999999999998,
+      "loss": 1.5512,
+      "step": 720
+    },
+    {
+      "epoch": 1.1536,
+      "grad_norm": 1.0425063371658325,
+      "learning_rate": 0.00021359999999999996,
+      "loss": 1.6953,
+      "step": 721
+    },
+    {
+      "epoch": 1.1552,
+      "grad_norm": 1.5456832647323608,
+      "learning_rate": 0.00021389999999999997,
+      "loss": 1.2235,
+      "step": 722
+    },
+    {
+      "epoch": 1.1568,
+      "grad_norm": 5.289543151855469,
+      "learning_rate": 0.00021419999999999998,
+      "loss": 2.3722,
+      "step": 723
+    },
+    {
+      "epoch": 1.1584,
+      "grad_norm": 3.060047149658203,
+      "learning_rate": 0.00021449999999999998,
+      "loss": 1.7361,
+      "step": 724
+    },
+    {
+      "epoch": 1.16,
+      "grad_norm": 2.2316486835479736,
+      "learning_rate": 0.00021479999999999996,
+      "loss": 2.2835,
+      "step": 725
+    },
+    {
+      "epoch": 1.1616,
+      "grad_norm": 11.319620132446289,
+      "learning_rate": 0.00021509999999999997,
+      "loss": 2.8832,
+      "step": 726
+    },
+    {
+      "epoch": 1.1632,
+      "grad_norm": 4.8169121742248535,
+      "learning_rate": 0.00021539999999999998,
+      "loss": 2.4756,
+      "step": 727
+    },
+    {
+      "epoch": 1.1648,
+      "grad_norm": 4.998867511749268,
+      "learning_rate": 0.00021569999999999998,
+      "loss": 2.5684,
+      "step": 728
+    },
+    {
+      "epoch": 1.1663999999999999,
+      "grad_norm": 1.6369566917419434,
+      "learning_rate": 0.00021599999999999996,
+      "loss": 1.9992,
+      "step": 729
+    },
+    {
+      "epoch": 1.168,
+      "grad_norm": 4.867010593414307,
+      "learning_rate": 0.00021629999999999997,
+      "loss": 1.6888,
+      "step": 730
+    },
+    {
+      "epoch": 1.1696,
+      "grad_norm": 4.665241241455078,
+      "learning_rate": 0.00021659999999999998,
+      "loss": 1.703,
+      "step": 731
+    },
+    {
+      "epoch": 1.1712,
+      "grad_norm": 4.923267364501953,
+      "learning_rate": 0.0002169,
+      "loss": 1.8941,
+      "step": 732
+    },
+    {
+      "epoch": 1.1728,
+      "grad_norm": 5.301763534545898,
+      "learning_rate": 0.00021719999999999997,
+      "loss": 1.8063,
+      "step": 733
+    },
+    {
+      "epoch": 1.1743999999999999,
+      "grad_norm": 5.480170726776123,
+      "learning_rate": 0.00021749999999999997,
+      "loss": 1.6896,
+      "step": 734
+    },
+    {
+      "epoch": 1.176,
+      "grad_norm": 5.136298656463623,
+      "learning_rate": 0.00021779999999999998,
+      "loss": 1.7705,
+      "step": 735
+    },
+    {
+      "epoch": 1.1776,
+      "grad_norm": 2.6885194778442383,
+      "learning_rate": 0.00021809999999999996,
+      "loss": 1.29,
+      "step": 736
+    },
+    {
+      "epoch": 1.1792,
+      "grad_norm": 1.7481634616851807,
+      "learning_rate": 0.00021839999999999997,
+      "loss": 1.8537,
+      "step": 737
+    },
+    {
+      "epoch": 1.1808,
+      "grad_norm": 2.1533167362213135,
+      "learning_rate": 0.00021869999999999998,
+      "loss": 1.3772,
+      "step": 738
+    },
+    {
+      "epoch": 1.1824,
+      "grad_norm": 1.7290595769882202,
+      "learning_rate": 0.00021899999999999998,
+      "loss": 1.2517,
+      "step": 739
+    },
+    {
+      "epoch": 1.184,
+      "grad_norm": 5.765242576599121,
+      "learning_rate": 0.00021929999999999996,
+      "loss": 1.9578,
+      "step": 740
+    },
+    {
+      "epoch": 1.1856,
+      "grad_norm": 5.171415328979492,
+      "learning_rate": 0.00021959999999999997,
+      "loss": 1.6263,
+      "step": 741
+    },
+    {
+      "epoch": 1.1872,
+      "grad_norm": 2.4269332885742188,
+      "learning_rate": 0.00021989999999999998,
+      "loss": 1.5452,
+      "step": 742
+    },
+    {
+      "epoch": 1.1888,
+      "grad_norm": 1.2465523481369019,
+      "learning_rate": 0.00022019999999999999,
+      "loss": 1.2282,
+      "step": 743
+    },
+    {
+      "epoch": 1.1904,
+      "grad_norm": 1.9669184684753418,
+      "learning_rate": 0.00022049999999999997,
+      "loss": 1.1983,
+      "step": 744
+    },
+    {
+      "epoch": 1.192,
+      "grad_norm": 5.07749605178833,
+      "learning_rate": 0.00022079999999999997,
+      "loss": 1.7768,
+      "step": 745
+    },
+    {
+      "epoch": 1.1936,
+      "grad_norm": 1.4985103607177734,
+      "learning_rate": 0.00022109999999999998,
+      "loss": 1.504,
+      "step": 746
+    },
+    {
+      "epoch": 1.1952,
+      "grad_norm": 0.8555597066879272,
+      "learning_rate": 0.0002214,
+      "loss": 1.2868,
+      "step": 747
+    },
+    {
+      "epoch": 1.1968,
+      "grad_norm": 1.0134530067443848,
+      "learning_rate": 0.00022169999999999997,
+      "loss": 1.5408,
+      "step": 748
+    },
+    {
+      "epoch": 1.1984,
+      "grad_norm": 2.2932121753692627,
+      "learning_rate": 0.00022199999999999998,
+      "loss": 1.4542,
+      "step": 749
+    },
+    {
+      "epoch": 1.2,
+      "grad_norm": 2.048572063446045,
+      "learning_rate": 0.00022229999999999998,
+      "loss": 1.6303,
+      "step": 750
+    },
+    {
+      "epoch": 1.2016,
+      "grad_norm": 1.3398712873458862,
+      "learning_rate": 0.0002226,
+      "loss": 1.6202,
+      "step": 751
+    },
+    {
+      "epoch": 1.2032,
+      "grad_norm": 1.2826422452926636,
+      "learning_rate": 0.00022289999999999997,
+      "loss": 1.114,
+      "step": 752
+    },
+    {
+      "epoch": 1.2048,
+      "grad_norm": 0.9887292385101318,
+      "learning_rate": 0.00022319999999999998,
+      "loss": 1.2535,
+      "step": 753
+    },
+    {
+      "epoch": 1.2064,
+      "grad_norm": 0.8880885243415833,
+      "learning_rate": 0.00022349999999999998,
+      "loss": 1.123,
+      "step": 754
+    },
+    {
+      "epoch": 1.208,
+      "grad_norm": 4.056207180023193,
+      "learning_rate": 0.0002238,
+      "loss": 1.7848,
+      "step": 755
+    },
+    {
+      "epoch": 1.2096,
+      "grad_norm": 1.0242630243301392,
+      "learning_rate": 0.00022409999999999997,
+      "loss": 1.6084,
+      "step": 756
+    },
+    {
+      "epoch": 1.2112,
+      "grad_norm": 1.5216087102890015,
+      "learning_rate": 0.00022439999999999998,
+      "loss": 1.0884,
+      "step": 757
+    },
+    {
+      "epoch": 1.2128,
+      "grad_norm": 1.1595895290374756,
+      "learning_rate": 0.0002247,
+      "loss": 1.4366,
+      "step": 758
+    },
+    {
+      "epoch": 1.2144,
+      "grad_norm": 1.78994619846344,
+      "learning_rate": 0.000225,
+      "loss": 1.4678,
+      "step": 759
+    },
+    {
+      "epoch": 1.216,
+      "grad_norm": 1.5748515129089355,
+      "learning_rate": 0.00022529999999999997,
+      "loss": 1.5742,
+      "step": 760
+    },
+    {
+      "epoch": 1.2176,
+      "grad_norm": 1.2527673244476318,
+      "learning_rate": 0.00022559999999999998,
+      "loss": 1.2028,
+      "step": 761
+    },
+    {
+      "epoch": 1.2192,
+      "grad_norm": 1.4951261281967163,
+      "learning_rate": 0.0002259,
+      "loss": 1.8887,
+      "step": 762
+    },
+    {
+      "epoch": 1.2208,
+      "grad_norm": 1.1303513050079346,
+      "learning_rate": 0.00022619999999999997,
+      "loss": 1.6479,
+      "step": 763
+    },
+    {
+      "epoch": 1.2224,
+      "grad_norm": 1.3236031532287598,
+      "learning_rate": 0.00022649999999999998,
+      "loss": 1.703,
+      "step": 764
+    },
+    {
+      "epoch": 1.224,
+      "grad_norm": 1.5853848457336426,
+      "learning_rate": 0.00022679999999999998,
+      "loss": 1.8706,
+      "step": 765
+    },
+    {
+      "epoch": 1.2256,
+      "grad_norm": 2.0144317150115967,
+      "learning_rate": 0.0002271,
+      "loss": 1.495,
+      "step": 766
+    },
+    {
+      "epoch": 1.2272,
+      "grad_norm": 1.472916841506958,
+      "learning_rate": 0.00022739999999999997,
+      "loss": 1.6167,
+      "step": 767
+    },
+    {
+      "epoch": 1.2288000000000001,
+      "grad_norm": 1.3060656785964966,
+      "learning_rate": 0.00022769999999999998,
+      "loss": 1.5432,
+      "step": 768
+    },
+    {
+      "epoch": 1.2304,
+      "grad_norm": 1.9118512868881226,
+      "learning_rate": 0.00022799999999999999,
+      "loss": 2.3,
+      "step": 769
+    },
+    {
+      "epoch": 1.232,
+      "grad_norm": 1.5411431789398193,
+      "learning_rate": 0.0002283,
+      "loss": 1.3154,
+      "step": 770
+    },
+    {
+      "epoch": 1.2336,
+      "grad_norm": 1.2540593147277832,
+      "learning_rate": 0.00022859999999999997,
+      "loss": 1.3445,
+      "step": 771
+    },
+    {
+      "epoch": 1.2352,
+      "grad_norm": 1.74718177318573,
+      "learning_rate": 0.00022889999999999998,
+      "loss": 1.7902,
+      "step": 772
+    },
+    {
+      "epoch": 1.2368000000000001,
+      "grad_norm": 1.5919808149337769,
+      "learning_rate": 0.0002292,
+      "loss": 1.7406,
+      "step": 773
+    },
+    {
+      "epoch": 1.2384,
+      "grad_norm": 2.1802892684936523,
+      "learning_rate": 0.0002295,
+      "loss": 2.3024,
+      "step": 774
+    },
+    {
+      "epoch": 1.24,
+      "grad_norm": NaN,
+      "learning_rate": 0.0002295,
+      "loss": 2.2887,
+      "step": 775
+    },
+    {
+      "epoch": 1.2416,
+      "grad_norm": 5.746895790100098,
+      "learning_rate": 0.00022979999999999997,
+      "loss": 2.6305,
+      "step": 776
+    },
+    {
+      "epoch": 1.2432,
+      "grad_norm": 5.819034099578857,
+      "learning_rate": 0.00023009999999999998,
+      "loss": 2.3811,
+      "step": 777
+    },
+    {
+      "epoch": 1.2448,
+      "grad_norm": 2.528698444366455,
+      "learning_rate": 0.0002304,
+      "loss": 1.9592,
+      "step": 778
+    },
+    {
+      "epoch": 1.2464,
+      "grad_norm": 4.070464611053467,
+      "learning_rate": 0.0002307,
+      "loss": 2.035,
+      "step": 779
+    },
+    {
+      "epoch": 1.248,
+      "grad_norm": 1.93435800075531,
+      "learning_rate": 0.00023099999999999998,
+      "loss": 1.8043,
+      "step": 780
+    },
+    {
+      "epoch": 1.2496,
+      "grad_norm": 3.285830497741699,
+      "learning_rate": 0.00023129999999999998,
+      "loss": 1.7722,
+      "step": 781
+    },
+    {
+      "epoch": 1.2511999999999999,
+      "grad_norm": 4.356208324432373,
+      "learning_rate": 0.0002316,
+      "loss": 1.6131,
+      "step": 782
+    },
+    {
+      "epoch": 1.2528000000000001,
+      "grad_norm": 5.4774603843688965,
+      "learning_rate": 0.0002319,
+      "loss": 1.5492,
+      "step": 783
+    },
+    {
+      "epoch": 1.2544,
+      "grad_norm": 3.671088218688965,
+      "learning_rate": 0.00023219999999999998,
+      "loss": 1.6548,
+      "step": 784
+    },
+    {
+      "epoch": 1.256,
+      "grad_norm": 3.7231082916259766,
+      "learning_rate": 0.00023249999999999999,
+      "loss": 1.3874,
+      "step": 785
+    },
+    {
+      "epoch": 1.2576,
+      "grad_norm": 1.782421588897705,
+      "learning_rate": 0.0002328,
+      "loss": 1.3857,
+      "step": 786
+    },
+    {
+      "epoch": 1.2591999999999999,
+      "grad_norm": 1.4939918518066406,
+      "learning_rate": 0.00023309999999999997,
+      "loss": 1.7764,
+      "step": 787
+    },
+    {
+      "epoch": 1.2608,
+      "grad_norm": 1.064145565032959,
+      "learning_rate": 0.00023339999999999998,
+      "loss": 1.3059,
+      "step": 788
+    },
+    {
+      "epoch": 1.2624,
+      "grad_norm": 2.0802013874053955,
+      "learning_rate": 0.0002337,
+      "loss": 1.3497,
+      "step": 789
+    },
+    {
+      "epoch": 1.264,
+      "grad_norm": 2.4550795555114746,
+      "learning_rate": 0.000234,
+      "loss": 1.6281,
+      "step": 790
+    },
+    {
+      "epoch": 1.2656,
+      "grad_norm": 1.424136996269226,
+      "learning_rate": 0.00023429999999999998,
+      "loss": 1.3037,
+      "step": 791
+    },
+    {
+      "epoch": 1.2671999999999999,
+      "grad_norm": 2.6355724334716797,
+      "learning_rate": 0.00023459999999999998,
+      "loss": 1.2327,
+      "step": 792
+    },
+    {
+      "epoch": 1.2688,
+      "grad_norm": 1.9551432132720947,
+      "learning_rate": 0.0002349,
+      "loss": 1.1327,
+      "step": 793
+    },
+    {
+      "epoch": 1.2704,
+      "grad_norm": 0.920864462852478,
+      "learning_rate": 0.0002352,
+      "loss": 1.329,
+      "step": 794
+    },
+    {
+      "epoch": 1.272,
+      "grad_norm": 0.8361489176750183,
+      "learning_rate": 0.00023549999999999998,
+      "loss": 1.4342,
+      "step": 795
+    },
+    {
+      "epoch": 1.2736,
+      "grad_norm": 1.4463287591934204,
+      "learning_rate": 0.00023579999999999999,
+      "loss": 1.2214,
+      "step": 796
+    },
+    {
+      "epoch": 1.2752,
+      "grad_norm": 0.8743594884872437,
+      "learning_rate": 0.0002361,
+      "loss": 1.126,
+      "step": 797
+    },
+    {
+      "epoch": 1.2768,
+      "grad_norm": 0.9425063133239746,
+      "learning_rate": 0.0002364,
+      "loss": 1.4563,
+      "step": 798
+    },
+    {
+      "epoch": 1.2784,
+      "grad_norm": 1.1034338474273682,
+      "learning_rate": 0.00023669999999999998,
+      "loss": 1.495,
+      "step": 799
+    },
+    {
+      "epoch": 1.28,
+      "grad_norm": 1.235459804534912,
+      "learning_rate": 0.000237,
+      "loss": 1.2458,
+      "step": 800
+    },
+    {
+      "epoch": 1.2816,
+      "grad_norm": 1.0407472848892212,
+      "learning_rate": 0.0002373,
+      "loss": 1.2275,
+      "step": 801
+    },
+    {
+      "epoch": 1.2832,
+      "grad_norm": 1.0057398080825806,
+      "learning_rate": 0.0002376,
+      "loss": 1.2064,
+      "step": 802
+    },
+    {
+      "epoch": 1.2848,
+      "grad_norm": 1.2582429647445679,
+      "learning_rate": 0.00023789999999999998,
+      "loss": 1.457,
+      "step": 803
+    },
+    {
+      "epoch": 1.2864,
+      "grad_norm": 1.2544838190078735,
+      "learning_rate": 0.0002382,
+      "loss": 1.0123,
+      "step": 804
+    },
+    {
+      "epoch": 1.288,
+      "grad_norm": 1.267555832862854,
+      "learning_rate": 0.0002385,
+      "loss": 1.3412,
+      "step": 805
+    },
+    {
+      "epoch": 1.2896,
+      "grad_norm": 1.585595726966858,
+      "learning_rate": 0.0002388,
+      "loss": 1.3812,
+      "step": 806
+    },
+    {
+      "epoch": 1.2912,
+      "grad_norm": 1.1115787029266357,
+      "learning_rate": 0.00023909999999999998,
+      "loss": 1.2368,
+      "step": 807
+    },
+    {
+      "epoch": 1.2928,
+      "grad_norm": 1.1539804935455322,
+      "learning_rate": 0.0002394,
+      "loss": 1.0813,
+      "step": 808
+    },
+    {
+      "epoch": 1.2944,
+      "grad_norm": 1.2517136335372925,
+      "learning_rate": 0.0002397,
+      "loss": 1.2567,
+      "step": 809
+    },
+    {
+      "epoch": 1.296,
+      "grad_norm": 1.0020838975906372,
+      "learning_rate": 0.00023999999999999998,
+      "loss": 1.3813,
+      "step": 810
+    },
+    {
+      "epoch": 1.2976,
+      "grad_norm": 1.278122067451477,
+      "learning_rate": 0.00024029999999999999,
+      "loss": 1.4291,
+      "step": 811
+    },
+    {
+      "epoch": 1.2992,
+      "grad_norm": 1.1353975534439087,
+      "learning_rate": 0.0002406,
+      "loss": 1.5433,
+      "step": 812
+    },
+    {
+      "epoch": 1.3008,
+      "grad_norm": 1.0918465852737427,
+      "learning_rate": 0.0002409,
+      "loss": 1.4993,
+      "step": 813
+    },
+    {
+      "epoch": 1.3024,
+      "grad_norm": 1.1691175699234009,
+      "learning_rate": 0.00024119999999999998,
+      "loss": 1.442,
+      "step": 814
+    },
+    {
+      "epoch": 1.304,
+      "grad_norm": 2.067641258239746,
+      "learning_rate": 0.0002415,
+      "loss": 1.3307,
+      "step": 815
+    },
+    {
+      "epoch": 1.3056,
+      "grad_norm": 1.2151570320129395,
+      "learning_rate": 0.0002418,
+      "loss": 1.3187,
+      "step": 816
+    },
+    {
+      "epoch": 1.3072,
+      "grad_norm": 1.3653641939163208,
+      "learning_rate": 0.0002421,
+      "loss": 1.1919,
+      "step": 817
+    },
+    {
+      "epoch": 1.3088,
+      "grad_norm": 1.193217158317566,
+      "learning_rate": 0.00024239999999999998,
+      "loss": 1.5955,
+      "step": 818
+    },
+    {
+      "epoch": 1.3104,
+      "grad_norm": 1.2559990882873535,
+      "learning_rate": 0.0002427,
+      "loss": 1.6028,
+      "step": 819
+    },
+    {
+      "epoch": 1.312,
+      "grad_norm": 1.5885244607925415,
+      "learning_rate": 0.000243,
+      "loss": 1.4464,
+      "step": 820
+    },
+    {
+      "epoch": 1.3136,
+      "grad_norm": 1.5733736753463745,
+      "learning_rate": 0.0002433,
+      "loss": 1.743,
+      "step": 821
+    },
+    {
+      "epoch": 1.3152,
+      "grad_norm": 2.05718731880188,
+      "learning_rate": 0.00024359999999999999,
+      "loss": 1.4574,
+      "step": 822
+    },
+    {
+      "epoch": 1.3168,
+      "grad_norm": 1.5828289985656738,
+      "learning_rate": 0.00024389999999999997,
+      "loss": 1.9888,
+      "step": 823
+    },
+    {
+      "epoch": 1.3184,
+      "grad_norm": 1.8411847352981567,
+      "learning_rate": 0.00024419999999999997,
+      "loss": 1.3748,
+      "step": 824
+    },
+    {
+      "epoch": 1.32,
+      "grad_norm": 2.4953489303588867,
+      "learning_rate": 0.0002445,
+      "loss": 2.2596,
+      "step": 825
+    },
+    {
+      "epoch": 1.3216,
+      "grad_norm": 22.735031127929688,
+      "learning_rate": 0.0002448,
+      "loss": 5.1826,
+      "step": 826
+    },
+    {
+      "epoch": 1.3232,
+      "grad_norm": 2.6146559715270996,
+      "learning_rate": 0.00024509999999999994,
+      "loss": 1.7049,
+      "step": 827
+    },
+    {
+      "epoch": 1.3248,
+      "grad_norm": 1.7866498231887817,
+      "learning_rate": 0.00024539999999999995,
+      "loss": 1.6428,
+      "step": 828
+    },
+    {
+      "epoch": 1.3264,
+      "grad_norm": 3.011427640914917,
+      "learning_rate": 0.00024569999999999995,
+      "loss": 1.7732,
+      "step": 829
+    },
+    {
+      "epoch": 1.328,
+      "grad_norm": 1.7671997547149658,
+      "learning_rate": 0.00024599999999999996,
+      "loss": 1.7,
+      "step": 830
+    },
+    {
+      "epoch": 1.3296000000000001,
+      "grad_norm": 3.2919392585754395,
+      "learning_rate": 0.00024629999999999997,
+      "loss": 1.8395,
+      "step": 831
+    },
+    {
+      "epoch": 1.3312,
+      "grad_norm": 3.4365289211273193,
+      "learning_rate": 0.0002466,
+      "loss": 1.4656,
+      "step": 832
+    },
+    {
+      "epoch": 1.3328,
+      "grad_norm": 2.34206485748291,
+      "learning_rate": 0.0002469,
+      "loss": 1.3553,
+      "step": 833
+    },
+    {
+      "epoch": 1.3344,
+      "grad_norm": 1.3108103275299072,
+      "learning_rate": 0.0002472,
+      "loss": 1.6943,
+      "step": 834
+    },
+    {
+      "epoch": 1.336,
+      "grad_norm": 1.1649718284606934,
+      "learning_rate": 0.00024749999999999994,
+      "loss": 1.6788,
+      "step": 835
+    },
+    {
+      "epoch": 1.3376000000000001,
+      "grad_norm": 0.8755460977554321,
+      "learning_rate": 0.00024779999999999995,
+      "loss": 1.064,
+      "step": 836
+    },
+    {
+      "epoch": 1.3392,
+      "grad_norm": 1.0399974584579468,
+      "learning_rate": 0.00024809999999999996,
+      "loss": 1.1169,
+      "step": 837
+    },
+    {
+      "epoch": 1.3408,
+      "grad_norm": 1.590290904045105,
+      "learning_rate": 0.00024839999999999997,
+      "loss": 1.32,
+      "step": 838
+    },
+    {
+      "epoch": 1.3424,
+      "grad_norm": 1.9613844156265259,
+      "learning_rate": 0.0002487,
+      "loss": 0.9774,
+      "step": 839
+    },
+    {
+      "epoch": 1.3439999999999999,
+      "grad_norm": 1.3067349195480347,
+      "learning_rate": 0.000249,
+      "loss": 1.0947,
+      "step": 840
+    },
+    {
+      "epoch": 1.3456000000000001,
+      "grad_norm": 3.495009422302246,
+      "learning_rate": 0.0002493,
+      "loss": 1.4792,
+      "step": 841
+    },
+    {
+      "epoch": 1.3472,
+      "grad_norm": 2.386378526687622,
+      "learning_rate": 0.00024959999999999994,
+      "loss": 1.2189,
+      "step": 842
+    },
+    {
+      "epoch": 1.3488,
+      "grad_norm": 1.2430686950683594,
+      "learning_rate": 0.00024989999999999995,
+      "loss": 1.383,
+      "step": 843
+    },
+    {
+      "epoch": 1.3504,
+      "grad_norm": 1.1015182733535767,
+      "learning_rate": 0.00025019999999999996,
+      "loss": 1.1695,
+      "step": 844
+    },
+    {
+      "epoch": 1.3519999999999999,
+      "grad_norm": 1.2849568128585815,
+      "learning_rate": 0.00025049999999999996,
+      "loss": 1.4384,
+      "step": 845
+    },
+    {
+      "epoch": 1.3536000000000001,
+      "grad_norm": 0.9367717504501343,
+      "learning_rate": 0.00025079999999999997,
+      "loss": 1.2806,
+      "step": 846
+    },
+    {
+      "epoch": 1.3552,
+      "grad_norm": 1.9065661430358887,
+      "learning_rate": 0.0002511,
+      "loss": 1.9186,
+      "step": 847
+    },
+    {
+      "epoch": 1.3568,
+      "grad_norm": 1.8349933624267578,
+      "learning_rate": 0.0002514,
+      "loss": 1.2997,
+      "step": 848
+    },
+    {
+      "epoch": 1.3584,
+      "grad_norm": 2.411646604537964,
+      "learning_rate": 0.0002517,
+      "loss": 1.7798,
+      "step": 849
+    },
+    {
+      "epoch": 1.3599999999999999,
+      "grad_norm": 1.3963836431503296,
+      "learning_rate": 0.00025199999999999995,
+      "loss": 1.2455,
+      "step": 850
+    },
+    {
+      "epoch": 1.3616,
+      "grad_norm": 1.7644349336624146,
+      "learning_rate": 0.00025229999999999995,
+      "loss": 1.5101,
+      "step": 851
+    },
+    {
+      "epoch": 1.3632,
+      "grad_norm": 1.1302613019943237,
+      "learning_rate": 0.00025259999999999996,
+      "loss": 1.3869,
+      "step": 852
+    },
+    {
+      "epoch": 1.3648,
+      "grad_norm": 2.062229633331299,
+      "learning_rate": 0.00025289999999999997,
+      "loss": 1.4488,
+      "step": 853
+    },
+    {
+      "epoch": 1.3664,
+      "grad_norm": 1.800642967224121,
+      "learning_rate": 0.0002532,
+      "loss": 1.0541,
+      "step": 854
+    },
+    {
+      "epoch": 1.3679999999999999,
+      "grad_norm": 3.4561281204223633,
+      "learning_rate": 0.0002535,
+      "loss": 1.4694,
+      "step": 855
+    },
+    {
+      "epoch": 1.3696,
+      "grad_norm": 2.443664073944092,
+      "learning_rate": 0.0002538,
+      "loss": 1.5959,
+      "step": 856
+    },
+    {
+      "epoch": 1.3712,
+      "grad_norm": 1.2733495235443115,
+      "learning_rate": 0.0002541,
+      "loss": 1.3463,
+      "step": 857
+    },
+    {
+      "epoch": 1.3728,
+      "grad_norm": 1.4782954454421997,
+      "learning_rate": 0.00025439999999999995,
+      "loss": 1.3913,
+      "step": 858
+    },
+    {
+      "epoch": 1.3744,
+      "grad_norm": 1.355779767036438,
+      "learning_rate": 0.00025469999999999996,
+      "loss": 1.4062,
+      "step": 859
+    },
+    {
+      "epoch": 1.376,
+      "grad_norm": 0.991340696811676,
+      "learning_rate": 0.00025499999999999996,
+      "loss": 1.0575,
+      "step": 860
+    },
+    {
+      "epoch": 1.3776,
+      "grad_norm": 1.4136706590652466,
+      "learning_rate": 0.00025529999999999997,
+      "loss": 1.4418,
+      "step": 861
+    },
+    {
+      "epoch": 1.3792,
+      "grad_norm": 1.5610219240188599,
+      "learning_rate": 0.0002556,
+      "loss": 1.909,
+      "step": 862
+    },
+    {
+      "epoch": 1.3808,
+      "grad_norm": 1.4637738466262817,
+      "learning_rate": 0.0002559,
+      "loss": 1.4926,
+      "step": 863
+    },
+    {
+      "epoch": 1.3824,
+      "grad_norm": 1.174544095993042,
+      "learning_rate": 0.0002562,
+      "loss": 1.444,
+      "step": 864
+    },
+    {
+      "epoch": 1.384,
+      "grad_norm": 1.295507550239563,
+      "learning_rate": 0.00025649999999999995,
+      "loss": 1.211,
+      "step": 865
+    },
+    {
+      "epoch": 1.3856,
+      "grad_norm": 1.341482162475586,
+      "learning_rate": 0.00025679999999999995,
+      "loss": 1.5489,
+      "step": 866
+    },
+    {
+      "epoch": 1.3872,
+      "grad_norm": 1.5005024671554565,
+      "learning_rate": 0.00025709999999999996,
+      "loss": 1.304,
+      "step": 867
+    },
+    {
+      "epoch": 1.3888,
+      "grad_norm": 1.7614165544509888,
+      "learning_rate": 0.00025739999999999997,
+      "loss": 1.411,
+      "step": 868
+    },
+    {
+      "epoch": 1.3904,
+      "grad_norm": 2.1121573448181152,
+      "learning_rate": 0.0002577,
+      "loss": 1.5675,
+      "step": 869
+    },
+    {
+      "epoch": 1.392,
+      "grad_norm": 1.4016284942626953,
+      "learning_rate": 0.000258,
+      "loss": 1.2976,
+      "step": 870
+    },
+    {
+      "epoch": 1.3936,
+      "grad_norm": 1.4677424430847168,
+      "learning_rate": 0.0002583,
+      "loss": 1.5295,
+      "step": 871
+    },
+    {
+      "epoch": 1.3952,
+      "grad_norm": 1.7327654361724854,
+      "learning_rate": 0.0002586,
+      "loss": 1.6782,
+      "step": 872
+    },
+    {
+      "epoch": 1.3968,
+      "grad_norm": 1.684560775756836,
+      "learning_rate": 0.00025889999999999995,
+      "loss": 2.1258,
+      "step": 873
+    },
+    {
+      "epoch": 1.3984,
+      "grad_norm": 1.1350618600845337,
+      "learning_rate": 0.00025919999999999996,
+      "loss": 1.2016,
+      "step": 874
+    },
+    {
+      "epoch": 1.4,
+      "grad_norm": 1.8442115783691406,
+      "learning_rate": 0.00025949999999999997,
+      "loss": 2.2616,
+      "step": 875
+    },
+    {
+      "epoch": 1.4016,
+      "grad_norm": 15.330982208251953,
+      "learning_rate": 0.00025979999999999997,
+      "loss": 3.7807,
+      "step": 876
+    },
+    {
+      "epoch": 1.4032,
+      "grad_norm": 9.512137413024902,
+      "learning_rate": 0.0002601,
+      "loss": 2.7959,
+      "step": 877
+    },
+    {
+      "epoch": 1.4048,
+      "grad_norm": 4.83724308013916,
+      "learning_rate": 0.0002604,
+      "loss": 2.2513,
+      "step": 878
+    },
+    {
+      "epoch": 1.4064,
+      "grad_norm": 1.0789581537246704,
+      "learning_rate": 0.0002607,
+      "loss": 1.1761,
+      "step": 879
+    },
+    {
+      "epoch": 1.408,
+      "grad_norm": 1.283165693283081,
+      "learning_rate": 0.000261,
+      "loss": 1.3131,
+      "step": 880
+    },
+    {
+      "epoch": 1.4096,
+      "grad_norm": 1.3923134803771973,
+      "learning_rate": 0.00026129999999999995,
+      "loss": 1.4793,
+      "step": 881
+    },
+    {
+      "epoch": 1.4112,
+      "grad_norm": 1.2186331748962402,
+      "learning_rate": 0.00026159999999999996,
+      "loss": 1.4069,
+      "step": 882
+    },
+    {
+      "epoch": 1.4128,
+      "grad_norm": 1.7576051950454712,
+      "learning_rate": 0.00026189999999999997,
+      "loss": 1.6623,
+      "step": 883
+    },
+    {
+      "epoch": 1.4144,
+      "grad_norm": 2.6093623638153076,
+      "learning_rate": 0.0002622,
+      "loss": 0.9214,
+      "step": 884
+    },
+    {
+      "epoch": 1.416,
+      "grad_norm": 1.0863568782806396,
+      "learning_rate": 0.0002625,
+      "loss": 1.4146,
+      "step": 885
+    },
+    {
+      "epoch": 1.4176,
+      "grad_norm": 2.363821506500244,
+      "learning_rate": 0.0002628,
+      "loss": 1.8021,
+      "step": 886
+    },
+    {
+      "epoch": 1.4192,
+      "grad_norm": 1.1821964979171753,
+      "learning_rate": 0.0002631,
+      "loss": 1.1834,
+      "step": 887
+    },
+    {
+      "epoch": 1.4208,
+      "grad_norm": 1.32361900806427,
+      "learning_rate": 0.00026339999999999995,
+      "loss": 1.192,
+      "step": 888
+    },
+    {
+      "epoch": 1.4224,
+      "grad_norm": 1.3281641006469727,
+      "learning_rate": 0.00026369999999999996,
+      "loss": 1.0575,
+      "step": 889
+    },
+    {
+      "epoch": 1.424,
+      "grad_norm": 1.5585789680480957,
+      "learning_rate": 0.00026399999999999997,
+      "loss": 1.1708,
+      "step": 890
+    },
+    {
+      "epoch": 1.4256,
+      "grad_norm": 2.31046462059021,
+      "learning_rate": 0.0002643,
+      "loss": 1.1794,
+      "step": 891
+    },
+    {
+      "epoch": 1.4272,
+      "grad_norm": 1.7033979892730713,
+      "learning_rate": 0.0002646,
+      "loss": 1.3221,
+      "step": 892
+    },
+    {
+      "epoch": 1.4288,
+      "grad_norm": 2.653367519378662,
+      "learning_rate": 0.0002649,
+      "loss": 1.4937,
+      "step": 893
+    },
+    {
+      "epoch": 1.4304000000000001,
+      "grad_norm": 0.9184427261352539,
+      "learning_rate": 0.0002652,
+      "loss": 1.078,
+      "step": 894
+    },
+    {
+      "epoch": 1.432,
+      "grad_norm": 0.9819865226745605,
+      "learning_rate": 0.0002655,
+      "loss": 1.186,
+      "step": 895
+    },
+    {
+      "epoch": 1.4336,
+      "grad_norm": 1.0578396320343018,
+      "learning_rate": 0.00026579999999999996,
+      "loss": 1.0641,
+      "step": 896
+    },
+    {
+      "epoch": 1.4352,
+      "grad_norm": 1.4637776613235474,
+      "learning_rate": 0.00026609999999999996,
+      "loss": 1.373,
+      "step": 897
+    },
+    {
+      "epoch": 1.4368,
+      "grad_norm": 0.9520303606987,
+      "learning_rate": 0.00026639999999999997,
+      "loss": 1.1442,
+      "step": 898
+    },
+    {
+      "epoch": 1.4384000000000001,
+      "grad_norm": 1.6817363500595093,
+      "learning_rate": 0.0002667,
+      "loss": 1.4782,
+      "step": 899
+    },
+    {
+      "epoch": 1.44,
+      "grad_norm": 2.1572883129119873,
+      "learning_rate": 0.000267,
+      "loss": 1.8022,
+      "step": 900
+    },
+    {
+      "epoch": 1.4416,
+      "grad_norm": 0.8815500736236572,
+      "learning_rate": 0.0002673,
+      "loss": 1.4329,
+      "step": 901
+    },
+    {
+      "epoch": 1.4432,
+      "grad_norm": 1.2165837287902832,
+      "learning_rate": 0.0002676,
+      "loss": 1.1709,
+      "step": 902
+    },
+    {
+      "epoch": 1.4447999999999999,
+      "grad_norm": 2.121063470840454,
+      "learning_rate": 0.0002679,
+      "loss": 1.3277,
+      "step": 903
+    },
+    {
+      "epoch": 1.4464000000000001,
+      "grad_norm": 1.4610421657562256,
+      "learning_rate": 0.00026819999999999996,
+      "loss": 1.5098,
+      "step": 904
+    },
+    {
+      "epoch": 1.448,
+      "grad_norm": 2.9947142601013184,
+      "learning_rate": 0.00026849999999999997,
+      "loss": 1.3525,
+      "step": 905
+    },
+    {
+      "epoch": 1.4496,
+      "grad_norm": 1.1811401844024658,
+      "learning_rate": 0.0002688,
+      "loss": 1.1726,
+      "step": 906
+    },
+    {
+      "epoch": 1.4512,
+      "grad_norm": 1.4365415573120117,
+      "learning_rate": 0.0002691,
+      "loss": 1.2433,
+      "step": 907
+    },
+    {
+      "epoch": 1.4527999999999999,
+      "grad_norm": 1.34075927734375,
+      "learning_rate": 0.0002694,
+      "loss": 1.0205,
+      "step": 908
+    },
+    {
+      "epoch": 1.4544000000000001,
+      "grad_norm": 2.7684597969055176,
+      "learning_rate": 0.0002697,
+      "loss": 1.0883,
+      "step": 909
+    },
+    {
+      "epoch": 1.456,
+      "grad_norm": 1.557430624961853,
+      "learning_rate": 0.00027,
+      "loss": 1.6006,
+      "step": 910
+    },
+    {
+      "epoch": 1.4576,
+      "grad_norm": 1.616085410118103,
+      "learning_rate": 0.00027029999999999996,
+      "loss": 1.4207,
+      "step": 911
+    },
+    {
+      "epoch": 1.4592,
+      "grad_norm": 5.76104211807251,
+      "learning_rate": 0.00027059999999999996,
+      "loss": 1.6985,
+      "step": 912
+    },
+    {
+      "epoch": 1.4607999999999999,
+      "grad_norm": 1.2783349752426147,
+      "learning_rate": 0.00027089999999999997,
+      "loss": 1.1576,
+      "step": 913
+    },
+    {
+      "epoch": 1.4624,
+      "grad_norm": 1.2653543949127197,
+      "learning_rate": 0.0002712,
+      "loss": 1.4433,
+      "step": 914
+    },
+    {
+      "epoch": 1.464,
+      "grad_norm": 1.2063896656036377,
+      "learning_rate": 0.0002715,
+      "loss": 1.5359,
+      "step": 915
+    },
+    {
+      "epoch": 1.4656,
+      "grad_norm": 2.794680118560791,
+      "learning_rate": 0.0002718,
+      "loss": 1.5251,
+      "step": 916
+    },
+    {
+      "epoch": 1.4672,
+      "grad_norm": 3.2242326736450195,
+      "learning_rate": 0.0002721,
+      "loss": 1.9892,
+      "step": 917
+    },
+    {
+      "epoch": 1.4687999999999999,
+      "grad_norm": 1.8846021890640259,
+      "learning_rate": 0.0002724,
+      "loss": 1.8354,
+      "step": 918
+    },
+    {
+      "epoch": 1.4704,
+      "grad_norm": 2.0368640422821045,
+      "learning_rate": 0.00027269999999999996,
+      "loss": 1.5269,
+      "step": 919
+    },
+    {
+      "epoch": 1.472,
+      "grad_norm": 1.5392261743545532,
+      "learning_rate": 0.00027299999999999997,
+      "loss": 1.6935,
+      "step": 920
+    },
+    {
+      "epoch": 1.4736,
+      "grad_norm": 1.5438854694366455,
+      "learning_rate": 0.0002733,
+      "loss": 1.7796,
+      "step": 921
+    },
+    {
+      "epoch": 1.4752,
+      "grad_norm": 1.796651005744934,
+      "learning_rate": 0.0002736,
+      "loss": 1.3982,
+      "step": 922
+    },
+    {
+      "epoch": 1.4768,
+      "grad_norm": 2.2069437503814697,
+      "learning_rate": 0.0002739,
+      "loss": 1.7769,
+      "step": 923
+    },
+    {
+      "epoch": 1.4784,
+      "grad_norm": 1.9683245420455933,
+      "learning_rate": 0.0002742,
+      "loss": 1.4909,
+      "step": 924
+    },
+    {
+      "epoch": 1.48,
+      "grad_norm": 1.911014199256897,
+      "learning_rate": 0.0002745,
+      "loss": 2.0897,
+      "step": 925
+    },
+    {
+      "epoch": 1.4816,
+      "grad_norm": 5.754384994506836,
+      "learning_rate": 0.0002748,
+      "loss": 2.1389,
+      "step": 926
+    },
+    {
+      "epoch": 1.4832,
+      "grad_norm": 1.59812331199646,
+      "learning_rate": 0.00027509999999999996,
+      "loss": 1.5023,
+      "step": 927
+    },
+    {
+      "epoch": 1.4848,
+      "grad_norm": 3.8509624004364014,
+      "learning_rate": 0.00027539999999999997,
+      "loss": 2.156,
+      "step": 928
+    },
+    {
+      "epoch": 1.4864,
+      "grad_norm": 1.6457704305648804,
+      "learning_rate": 0.0002757,
+      "loss": 1.6978,
+      "step": 929
+    },
+    {
+      "epoch": 1.488,
+      "grad_norm": 9.261984825134277,
+      "learning_rate": 0.000276,
+      "loss": 3.222,
+      "step": 930
+    },
+    {
+      "epoch": 1.4896,
+      "grad_norm": 10.191606521606445,
+      "learning_rate": 0.0002763,
+      "loss": 2.7299,
+      "step": 931
+    },
+    {
+      "epoch": 1.4912,
+      "grad_norm": 2.034604072570801,
+      "learning_rate": 0.0002766,
+      "loss": 1.363,
+      "step": 932
+    },
+    {
+      "epoch": 1.4928,
+      "grad_norm": 2.7943766117095947,
+      "learning_rate": 0.0002769,
+      "loss": 1.4561,
+      "step": 933
+    },
+    {
+      "epoch": 1.4944,
+      "grad_norm": 2.739060640335083,
+      "learning_rate": 0.0002772,
+      "loss": 1.5289,
+      "step": 934
+    },
+    {
+      "epoch": 1.496,
+      "grad_norm": 2.6572391986846924,
+      "learning_rate": 0.00027749999999999997,
+      "loss": 1.3459,
+      "step": 935
+    },
+    {
+      "epoch": 1.4976,
+      "grad_norm": 2.4692184925079346,
+      "learning_rate": 0.0002778,
+      "loss": 1.409,
+      "step": 936
+    },
+    {
+      "epoch": 1.4992,
+      "grad_norm": 1.569419264793396,
+      "learning_rate": 0.0002781,
+      "loss": 1.1897,
+      "step": 937
+    },
+    {
+      "epoch": 1.5008,
+      "grad_norm": 0.9803001880645752,
+      "learning_rate": 0.0002784,
+      "loss": 1.1971,
+      "step": 938
+    },
+    {
+      "epoch": 1.5024,
+      "grad_norm": 1.3759132623672485,
+      "learning_rate": 0.0002787,
+      "loss": 1.1573,
+      "step": 939
+    },
+    {
+      "epoch": 1.504,
+      "grad_norm": 1.4470410346984863,
+      "learning_rate": 0.000279,
+      "loss": 1.2441,
+      "step": 940
+    },
+    {
+      "epoch": 1.5056,
+      "grad_norm": 1.9103741645812988,
+      "learning_rate": 0.0002793,
+      "loss": 1.2325,
+      "step": 941
+    },
+    {
+      "epoch": 1.5072,
+      "grad_norm": 1.5558336973190308,
+      "learning_rate": 0.00027959999999999997,
+      "loss": 1.0402,
+      "step": 942
+    },
+    {
+      "epoch": 1.5088,
+      "grad_norm": 2.0115926265716553,
+      "learning_rate": 0.0002799,
+      "loss": 1.2751,
+      "step": 943
+    },
+    {
+      "epoch": 1.5104,
+      "grad_norm": 1.6013593673706055,
+      "learning_rate": 0.0002802,
+      "loss": 1.6149,
+      "step": 944
+    },
+    {
+      "epoch": 1.512,
+      "grad_norm": 1.3492580652236938,
+      "learning_rate": 0.0002805,
+      "loss": 1.2308,
+      "step": 945
+    },
+    {
+      "epoch": 1.5135999999999998,
+      "grad_norm": 1.3978670835494995,
+      "learning_rate": 0.0002808,
+      "loss": 1.1674,
+      "step": 946
+    },
+    {
+      "epoch": 1.5152,
+      "grad_norm": 1.257152795791626,
+      "learning_rate": 0.0002811,
+      "loss": 1.2999,
+      "step": 947
+    },
+    {
+      "epoch": 1.5168,
+      "grad_norm": 1.3785860538482666,
+      "learning_rate": 0.00028139999999999996,
+      "loss": 1.2184,
+      "step": 948
+    },
+    {
+      "epoch": 1.5184,
+      "grad_norm": 2.098989963531494,
+      "learning_rate": 0.00028169999999999996,
+      "loss": 1.4197,
+      "step": 949
+    },
+    {
+      "epoch": 1.52,
+      "grad_norm": 1.411068320274353,
+      "learning_rate": 0.00028199999999999997,
+      "loss": 1.3696,
+      "step": 950
+    },
+    {
+      "epoch": 1.5215999999999998,
+      "grad_norm": 1.1803005933761597,
+      "learning_rate": 0.0002823,
+      "loss": 1.4752,
+      "step": 951
+    },
+    {
+      "epoch": 1.5232,
+      "grad_norm": 1.4621422290802002,
+      "learning_rate": 0.0002826,
+      "loss": 1.2319,
+      "step": 952
+    },
+    {
+      "epoch": 1.5248,
+      "grad_norm": 2.065951108932495,
+      "learning_rate": 0.00028289999999999994,
+      "loss": 1.3185,
+      "step": 953
+    },
+    {
+      "epoch": 1.5264,
+      "grad_norm": 2.077345371246338,
+      "learning_rate": 0.00028319999999999994,
+      "loss": 1.3232,
+      "step": 954
+    },
+    {
+      "epoch": 1.528,
+      "grad_norm": 2.0758562088012695,
+      "learning_rate": 0.00028349999999999995,
+      "loss": 1.1082,
+      "step": 955
+    },
+    {
+      "epoch": 1.5295999999999998,
+      "grad_norm": 2.724622964859009,
+      "learning_rate": 0.00028379999999999996,
+      "loss": 1.3383,
+      "step": 956
+    },
+    {
+      "epoch": 1.5312000000000001,
+      "grad_norm": 1.9979689121246338,
+      "learning_rate": 0.00028409999999999997,
+      "loss": 1.514,
+      "step": 957
+    },
+    {
+      "epoch": 1.5328,
+      "grad_norm": 1.9366734027862549,
+      "learning_rate": 0.0002844,
+      "loss": 1.2723,
+      "step": 958
+    },
+    {
+      "epoch": 1.5344,
+      "grad_norm": 2.198087215423584,
+      "learning_rate": 0.0002847,
+      "loss": 1.4332,
+      "step": 959
+    },
+    {
+      "epoch": 1.536,
+      "grad_norm": 1.5621610879898071,
+      "learning_rate": 0.000285,
+      "loss": 1.3232,
+      "step": 960
+    },
+    {
+      "epoch": 1.5375999999999999,
+      "grad_norm": 1.4429649114608765,
+      "learning_rate": 0.00028529999999999994,
+      "loss": 1.292,
+      "step": 961
+    },
+    {
+      "epoch": 1.5392000000000001,
+      "grad_norm": 1.7527788877487183,
+      "learning_rate": 0.00028559999999999995,
+      "loss": 1.3802,
+      "step": 962
+    },
+    {
+      "epoch": 1.5408,
+      "grad_norm": 2.562932252883911,
+      "learning_rate": 0.00028589999999999996,
+      "loss": 1.5058,
+      "step": 963
+    },
+    {
+      "epoch": 1.5424,
+      "grad_norm": 2.0278782844543457,
+      "learning_rate": 0.00028619999999999996,
+      "loss": 1.4053,
+      "step": 964
+    },
+    {
+      "epoch": 1.544,
+      "grad_norm": 2.133039712905884,
+      "learning_rate": 0.00028649999999999997,
+      "loss": 1.5985,
+      "step": 965
+    },
+    {
+      "epoch": 1.5455999999999999,
+      "grad_norm": 1.7495462894439697,
+      "learning_rate": 0.0002868,
+      "loss": 1.7425,
+      "step": 966
+    },
+    {
+      "epoch": 1.5472000000000001,
+      "grad_norm": 1.314456582069397,
+      "learning_rate": 0.0002871,
+      "loss": 1.1744,
+      "step": 967
+    },
+    {
+      "epoch": 1.5488,
+      "grad_norm": 1.5634371042251587,
+      "learning_rate": 0.00028739999999999994,
+      "loss": 1.199,
+      "step": 968
+    },
+    {
+      "epoch": 1.5504,
+      "grad_norm": 1.5051501989364624,
+      "learning_rate": 0.00028769999999999995,
+      "loss": 1.3761,
+      "step": 969
+    },
+    {
+      "epoch": 1.552,
+      "grad_norm": 1.4913947582244873,
+      "learning_rate": 0.00028799999999999995,
+      "loss": 1.7574,
+      "step": 970
+    },
+    {
+      "epoch": 1.5535999999999999,
+      "grad_norm": 2.0032637119293213,
+      "learning_rate": 0.00028829999999999996,
+      "loss": 1.6131,
+      "step": 971
+    },
+    {
+      "epoch": 1.5552000000000001,
+      "grad_norm": 1.4599378108978271,
+      "learning_rate": 0.00028859999999999997,
+      "loss": 1.5788,
+      "step": 972
+    },
+    {
+      "epoch": 1.5568,
+      "grad_norm": 1.526383638381958,
+      "learning_rate": 0.0002889,
+      "loss": 1.4599,
+      "step": 973
+    },
+    {
+      "epoch": 1.5584,
+      "grad_norm": 3.349010705947876,
+      "learning_rate": 0.0002892,
+      "loss": 1.7604,
+      "step": 974
+    },
+    {
+      "epoch": 1.56,
+      "grad_norm": 3.1439058780670166,
+      "learning_rate": 0.0002895,
+      "loss": 2.0222,
+      "step": 975
+    },
+    {
+      "epoch": 1.5615999999999999,
+      "grad_norm": 5.448111057281494,
+      "learning_rate": 0.00028979999999999994,
+      "loss": 2.2708,
+      "step": 976
+    },
+    {
+      "epoch": 1.5632000000000001,
+      "grad_norm": 3.330211877822876,
+      "learning_rate": 0.00029009999999999995,
+      "loss": 2.3857,
+      "step": 977
+    },
+    {
+      "epoch": 1.5648,
+      "grad_norm": 1.844307541847229,
+      "learning_rate": 0.00029039999999999996,
+      "loss": 1.8033,
+      "step": 978
+    },
+    {
+      "epoch": 1.5664,
+      "grad_norm": 2.17771053314209,
+      "learning_rate": 0.00029069999999999996,
+      "loss": 1.8416,
+      "step": 979
+    },
+    {
+      "epoch": 1.568,
+      "grad_norm": 1.889838695526123,
+      "learning_rate": 0.00029099999999999997,
+      "loss": 1.4765,
+      "step": 980
+    },
+    {
+      "epoch": 1.5695999999999999,
+      "grad_norm": 1.280713677406311,
+      "learning_rate": 0.0002913,
+      "loss": 1.5771,
+      "step": 981
+    },
+    {
+      "epoch": 1.5712000000000002,
+      "grad_norm": 1.2217782735824585,
+      "learning_rate": 0.0002916,
+      "loss": 1.4645,
+      "step": 982
+    },
+    {
+      "epoch": 1.5728,
+      "grad_norm": 1.493486762046814,
+      "learning_rate": 0.0002919,
+      "loss": 1.3598,
+      "step": 983
+    },
+    {
+      "epoch": 1.5744,
+      "grad_norm": 1.8840752840042114,
+      "learning_rate": 0.00029219999999999995,
+      "loss": 1.5287,
+      "step": 984
+    },
+    {
+      "epoch": 1.576,
+      "grad_norm": 1.965975046157837,
+      "learning_rate": 0.00029249999999999995,
+      "loss": 1.2946,
+      "step": 985
+    },
+    {
+      "epoch": 1.5776,
+      "grad_norm": 1.5697219371795654,
+      "learning_rate": 0.00029279999999999996,
+      "loss": 1.0137,
+      "step": 986
+    },
+    {
+      "epoch": 1.5792000000000002,
+      "grad_norm": 1.665776014328003,
+      "learning_rate": 0.00029309999999999997,
+      "loss": 1.1297,
+      "step": 987
+    },
+    {
+      "epoch": 1.5808,
+      "grad_norm": 1.9357331991195679,
+      "learning_rate": 0.0002934,
+      "loss": 1.1271,
+      "step": 988
+    },
+    {
+      "epoch": 1.5824,
+      "grad_norm": 0.922601044178009,
+      "learning_rate": 0.0002937,
+      "loss": 1.3413,
+      "step": 989
+    },
+    {
+      "epoch": 1.584,
+      "grad_norm": 1.0412627458572388,
+      "learning_rate": 0.000294,
+      "loss": 1.1785,
+      "step": 990
+    },
+    {
+      "epoch": 1.5856,
+      "grad_norm": 1.292492151260376,
+      "learning_rate": 0.00029429999999999994,
+      "loss": 1.4531,
+      "step": 991
+    },
+    {
+      "epoch": 1.5872000000000002,
+      "grad_norm": 1.4930530786514282,
+      "learning_rate": 0.00029459999999999995,
+      "loss": 1.4747,
+      "step": 992
+    },
+    {
+      "epoch": 1.5888,
+      "grad_norm": 1.398553729057312,
+      "learning_rate": 0.00029489999999999996,
+      "loss": 1.2275,
+      "step": 993
+    },
+    {
+      "epoch": 1.5904,
+      "grad_norm": 1.1149609088897705,
+      "learning_rate": 0.00029519999999999997,
+      "loss": 1.3404,
+      "step": 994
+    },
+    {
+      "epoch": 1.592,
+      "grad_norm": 2.243300676345825,
+      "learning_rate": 0.00029549999999999997,
+      "loss": 1.4117,
+      "step": 995
+    },
+    {
+      "epoch": 1.5936,
+      "grad_norm": 0.9678653478622437,
+      "learning_rate": 0.0002958,
+      "loss": 1.2791,
+      "step": 996
+    },
+    {
+      "epoch": 1.5952,
+      "grad_norm": 0.9126372337341309,
+      "learning_rate": 0.0002961,
+      "loss": 1.54,
+      "step": 997
+    },
+    {
+      "epoch": 1.5968,
+      "grad_norm": 1.447344422340393,
+      "learning_rate": 0.0002964,
+      "loss": 1.2979,
+      "step": 998
+    },
+    {
+      "epoch": 1.5984,
+      "grad_norm": 2.5969924926757812,
+      "learning_rate": 0.00029669999999999995,
+      "loss": 1.6844,
+      "step": 999
+    },
+    {
+      "epoch": 1.6,
+      "grad_norm": 1.3566679954528809,
+      "learning_rate": 0.00029699999999999996,
+      "loss": 1.2633,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "eval_cer": 0.342703815793579,
+      "eval_loss": 1.7065902948379517,
+      "eval_runtime": 159.4221,
+      "eval_samples_per_second": 19.671,
+      "eval_steps_per_second": 1.229,
+      "eval_wer": 0.5115005185204882,
+      "step": 1000
+    },
+    {
+      "epoch": 1.6,
+      "step": 1000,
+      "total_flos": 6.203691115248614e+18,
+      "train_loss": 4.768421524226666,
+      "train_runtime": 2101.2175,
+      "train_samples_per_second": 15.229,
+      "train_steps_per_second": 0.476
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 1000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 2,
+  "save_steps": 1000,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 6.203691115248614e+18,
+  "train_batch_size": 16,
+  "trial_name": null,
+  "trial_params": null
+}