diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,6683 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 2.912,
+  "eval_steps": 78,
+  "global_step": 1092,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 5e-06,
+      "loss": 1.5679,
+      "step": 1
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 0.6564,
+      "step": 2
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.991095280498665e-06,
+      "loss": 1.3112,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.98219056099733e-06,
+      "loss": 1.1069,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.973285841495993e-06,
+      "loss": 1.3761,
+      "step": 5
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.96438112199466e-06,
+      "loss": 1.4913,
+      "step": 6
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.955476402493323e-06,
+      "loss": 0.7124,
+      "step": 7
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.946571682991988e-06,
+      "loss": 1.3899,
+      "step": 8
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.93766696349065e-06,
+      "loss": 1.3987,
+      "step": 9
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.928762243989314e-06,
+      "loss": 1.2584,
+      "step": 10
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.919857524487978e-06,
+      "loss": 0.6098,
+      "step": 11
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.910952804986643e-06,
+      "loss": 0.6515,
+      "step": 12
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.902048085485308e-06,
+      "loss": 1.2805,
+      "step": 13
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.893143365983973e-06,
+      "loss": 1.2644,
+      "step": 14
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.884238646482637e-06,
+      "loss": 0.5699,
+      "step": 15
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.875333926981301e-06,
+      "loss": 0.6603,
+      "step": 16
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.866429207479965e-06,
+      "loss": 1.2595,
+      "step": 17
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.85752448797863e-06,
+      "loss": 1.3991,
+      "step": 18
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.848619768477293e-06,
+      "loss": 1.2662,
+      "step": 19
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.83971504897596e-06,
+      "loss": 0.6415,
+      "step": 20
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.830810329474622e-06,
+      "loss": 1.4047,
+      "step": 21
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.821905609973286e-06,
+      "loss": 0.6753,
+      "step": 22
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.81300089047195e-06,
+      "loss": 1.1443,
+      "step": 23
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.804096170970614e-06,
+      "loss": 1.2767,
+      "step": 24
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.795191451469279e-06,
+      "loss": 1.1256,
+      "step": 25
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.786286731967943e-06,
+      "loss": 1.2777,
+      "step": 26
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.777382012466609e-06,
+      "loss": 1.195,
+      "step": 27
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.768477292965273e-06,
+      "loss": 0.6548,
+      "step": 28
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.759572573463937e-06,
+      "loss": 1.2147,
+      "step": 29
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.750667853962601e-06,
+      "loss": 1.1264,
+      "step": 30
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.741763134461265e-06,
+      "loss": 0.5284,
+      "step": 31
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.73285841495993e-06,
+      "loss": 0.6118,
+      "step": 32
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.723953695458594e-06,
+      "loss": 0.9468,
+      "step": 33
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.715048975957258e-06,
+      "loss": 1.0446,
+      "step": 34
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.706144256455922e-06,
+      "loss": 1.1151,
+      "step": 35
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.697239536954586e-06,
+      "loss": 1.0386,
+      "step": 36
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.68833481745325e-06,
+      "loss": 0.5454,
+      "step": 37
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.679430097951915e-06,
+      "loss": 1.0753,
+      "step": 38
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.670525378450579e-06,
+      "loss": 1.095,
+      "step": 39
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.661620658949243e-06,
+      "loss": 0.9714,
+      "step": 40
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.652715939447909e-06,
+      "loss": 1.2478,
+      "step": 41
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.643811219946573e-06,
+      "loss": 0.852,
+      "step": 42
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.634906500445237e-06,
+      "loss": 1.0796,
+      "step": 43
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.626001780943901e-06,
+      "loss": 1.0243,
+      "step": 44
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.617097061442565e-06,
+      "loss": 0.8925,
+      "step": 45
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.60819234194123e-06,
+      "loss": 1.0763,
+      "step": 46
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.599287622439894e-06,
+      "loss": 0.4674,
+      "step": 47
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.590382902938558e-06,
+      "loss": 0.59,
+      "step": 48
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.581478183437222e-06,
+      "loss": 1.1625,
+      "step": 49
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.572573463935886e-06,
+      "loss": 1.0378,
+      "step": 50
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56366874443455e-06,
+      "loss": 1.0207,
+      "step": 51
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.554764024933215e-06,
+      "loss": 0.5295,
+      "step": 52
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.545859305431879e-06,
+      "loss": 1.0134,
+      "step": 53
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.536954585930543e-06,
+      "loss": 0.8273,
+      "step": 54
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.528049866429209e-06,
+      "loss": 1.1349,
+      "step": 55
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.519145146927873e-06,
+      "loss": 0.415,
+      "step": 56
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.510240427426537e-06,
+      "loss": 1.1188,
+      "step": 57
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.501335707925201e-06,
+      "loss": 1.0772,
+      "step": 58
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.492430988423866e-06,
+      "loss": 1.1848,
+      "step": 59
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.48352626892253e-06,
+      "loss": 0.5526,
+      "step": 60
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.474621549421194e-06,
+      "loss": 0.9804,
+      "step": 61
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.465716829919858e-06,
+      "loss": 0.433,
+      "step": 62
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.456812110418522e-06,
+      "loss": 0.9639,
+      "step": 63
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.447907390917186e-06,
+      "loss": 1.0203,
+      "step": 64
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.43900267141585e-06,
+      "loss": 1.1368,
+      "step": 65
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.430097951914515e-06,
+      "loss": 0.8997,
+      "step": 66
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.421193232413179e-06,
+      "loss": 0.4708,
+      "step": 67
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.412288512911843e-06,
+      "loss": 0.9834,
+      "step": 68
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.403383793410509e-06,
+      "loss": 0.9815,
+      "step": 69
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.394479073909173e-06,
+      "loss": 0.9501,
+      "step": 70
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.385574354407837e-06,
+      "loss": 0.8634,
+      "step": 71
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.376669634906502e-06,
+      "loss": 0.5108,
+      "step": 72
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.367764915405166e-06,
+      "loss": 0.9672,
+      "step": 73
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.35886019590383e-06,
+      "loss": 1.0061,
+      "step": 74
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.349955476402494e-06,
+      "loss": 0.5873,
+      "step": 75
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.341050756901158e-06,
+      "loss": 0.9315,
+      "step": 76
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.332146037399822e-06,
+      "loss": 0.9999,
+      "step": 77
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.323241317898487e-06,
+      "loss": 0.939,
+      "step": 78
+    },
+    {
+      "epoch": 0.21,
+      "eval_runtime": 2902.2089,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 78
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.31433659839715e-06,
+      "loss": 0.5028,
+      "step": 79
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.305431878895815e-06,
+      "loss": 0.5495,
+      "step": 80
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.296527159394479e-06,
+      "loss": 0.8755,
+      "step": 81
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.287622439893143e-06,
+      "loss": 0.8868,
+      "step": 82
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.27871772039181e-06,
+      "loss": 0.9817,
+      "step": 83
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 9.269813000890473e-06,
+      "loss": 0.9276,
+      "step": 84
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.260908281389138e-06,
+      "loss": 0.5936,
+      "step": 85
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.252003561887802e-06,
+      "loss": 0.9501,
+      "step": 86
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.243098842386466e-06,
+      "loss": 0.9134,
+      "step": 87
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 9.23419412288513e-06,
+      "loss": 1.0789,
+      "step": 88
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.225289403383794e-06,
+      "loss": 1.0623,
+      "step": 89
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.216384683882458e-06,
+      "loss": 0.9853,
+      "step": 90
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 9.207479964381123e-06,
+      "loss": 0.9017,
+      "step": 91
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.198575244879787e-06,
+      "loss": 0.9913,
+      "step": 92
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.189670525378451e-06,
+      "loss": 0.889,
+      "step": 93
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.180765805877115e-06,
+      "loss": 1.0237,
+      "step": 94
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 9.17186108637578e-06,
+      "loss": 0.9583,
+      "step": 95
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.162956366874443e-06,
+      "loss": 0.5038,
+      "step": 96
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.15405164737311e-06,
+      "loss": 0.901,
+      "step": 97
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.145146927871773e-06,
+      "loss": 1.09,
+      "step": 98
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 9.136242208370438e-06,
+      "loss": 1.1057,
+      "step": 99
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.127337488869102e-06,
+      "loss": 0.5086,
+      "step": 100
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.118432769367766e-06,
+      "loss": 1.0832,
+      "step": 101
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.10952804986643e-06,
+      "loss": 0.3827,
+      "step": 102
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 9.100623330365094e-06,
+      "loss": 0.986,
+      "step": 103
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.091718610863759e-06,
+      "loss": 0.5014,
+      "step": 104
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.082813891362423e-06,
+      "loss": 0.9803,
+      "step": 105
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 9.073909171861087e-06,
+      "loss": 0.9462,
+      "step": 106
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.065004452359751e-06,
+      "loss": 0.9408,
+      "step": 107
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.056099732858415e-06,
+      "loss": 0.9063,
+      "step": 108
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.04719501335708e-06,
+      "loss": 0.5223,
+      "step": 109
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 9.038290293855744e-06,
+      "loss": 1.0592,
+      "step": 110
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.02938557435441e-06,
+      "loss": 0.9392,
+      "step": 111
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.020480854853074e-06,
+      "loss": 0.9839,
+      "step": 112
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.011576135351738e-06,
+      "loss": 1.0379,
+      "step": 113
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 9.002671415850402e-06,
+      "loss": 1.0347,
+      "step": 114
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.993766696349066e-06,
+      "loss": 0.4347,
+      "step": 115
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.98486197684773e-06,
+      "loss": 0.9698,
+      "step": 116
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.975957257346395e-06,
+      "loss": 0.4782,
+      "step": 117
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 8.967052537845059e-06,
+      "loss": 0.9048,
+      "step": 118
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.958147818343723e-06,
+      "loss": 1.047,
+      "step": 119
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.949243098842387e-06,
+      "loss": 1.0747,
+      "step": 120
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 8.940338379341051e-06,
+      "loss": 0.9291,
+      "step": 121
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.931433659839715e-06,
+      "loss": 0.9185,
+      "step": 122
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.92252894033838e-06,
+      "loss": 0.3955,
+      "step": 123
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.913624220837044e-06,
+      "loss": 0.9323,
+      "step": 124
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 8.90471950133571e-06,
+      "loss": 1.0434,
+      "step": 125
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.895814781834374e-06,
+      "loss": 0.9057,
+      "step": 126
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.886910062333038e-06,
+      "loss": 1.0899,
+      "step": 127
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.878005342831702e-06,
+      "loss": 1.0595,
+      "step": 128
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 8.869100623330366e-06,
+      "loss": 0.8791,
+      "step": 129
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.86019590382903e-06,
+      "loss": 0.8141,
+      "step": 130
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.851291184327695e-06,
+      "loss": 1.0871,
+      "step": 131
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.842386464826359e-06,
+      "loss": 1.0,
+      "step": 132
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 8.833481745325023e-06,
+      "loss": 0.3848,
+      "step": 133
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.824577025823687e-06,
+      "loss": 0.9309,
+      "step": 134
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.815672306322351e-06,
+      "loss": 0.8992,
+      "step": 135
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 8.806767586821016e-06,
+      "loss": 0.9731,
+      "step": 136
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.79786286731968e-06,
+      "loss": 0.7418,
+      "step": 137
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.788958147818344e-06,
+      "loss": 1.0392,
+      "step": 138
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.78005342831701e-06,
+      "loss": 0.9117,
+      "step": 139
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 8.771148708815674e-06,
+      "loss": 0.884,
+      "step": 140
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.762243989314338e-06,
+      "loss": 0.5381,
+      "step": 141
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.753339269813002e-06,
+      "loss": 0.9654,
+      "step": 142
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.744434550311666e-06,
+      "loss": 1.0507,
+      "step": 143
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 8.73552983081033e-06,
+      "loss": 1.1323,
+      "step": 144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.726625111308995e-06,
+      "loss": 0.8715,
+      "step": 145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.717720391807659e-06,
+      "loss": 1.0715,
+      "step": 146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.708815672306323e-06,
+      "loss": 0.3935,
+      "step": 147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 8.699910952804987e-06,
+      "loss": 0.8251,
+      "step": 148
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.691006233303652e-06,
+      "loss": 0.8783,
+      "step": 149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.682101513802316e-06,
+      "loss": 1.0448,
+      "step": 150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 8.67319679430098e-06,
+      "loss": 0.9143,
+      "step": 151
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.664292074799646e-06,
+      "loss": 1.2174,
+      "step": 152
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.65538735529831e-06,
+      "loss": 0.9514,
+      "step": 153
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.646482635796974e-06,
+      "loss": 0.9791,
+      "step": 154
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 8.637577916295638e-06,
+      "loss": 0.4886,
+      "step": 155
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.628673196794302e-06,
+      "loss": 0.9256,
+      "step": 156
+    },
+    {
+      "epoch": 0.42,
+      "eval_runtime": 2894.9614,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 156
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.619768477292965e-06,
+      "loss": 0.7765,
+      "step": 157
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.610863757791629e-06,
+      "loss": 1.0918,
+      "step": 158
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 8.601959038290295e-06,
+      "loss": 0.9926,
+      "step": 159
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.593054318788959e-06,
+      "loss": 1.0111,
+      "step": 160
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.584149599287623e-06,
+      "loss": 0.9597,
+      "step": 161
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.575244879786287e-06,
+      "loss": 1.0429,
+      "step": 162
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 8.566340160284952e-06,
+      "loss": 0.9598,
+      "step": 163
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.557435440783616e-06,
+      "loss": 0.9508,
+      "step": 164
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.54853072128228e-06,
+      "loss": 1.0247,
+      "step": 165
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 8.539626001780946e-06,
+      "loss": 0.5291,
+      "step": 166
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.53072128227961e-06,
+      "loss": 0.9168,
+      "step": 167
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.521816562778274e-06,
+      "loss": 1.0452,
+      "step": 168
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.512911843276937e-06,
+      "loss": 0.4452,
+      "step": 169
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 8.504007123775601e-06,
+      "loss": 0.5363,
+      "step": 170
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.495102404274265e-06,
+      "loss": 0.8212,
+      "step": 171
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.48619768477293e-06,
+      "loss": 1.0423,
+      "step": 172
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.477292965271595e-06,
+      "loss": 0.95,
+      "step": 173
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 8.46838824577026e-06,
+      "loss": 0.4479,
+      "step": 174
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.459483526268923e-06,
+      "loss": 0.9053,
+      "step": 175
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.450578806767588e-06,
+      "loss": 0.97,
+      "step": 176
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.441674087266252e-06,
+      "loss": 0.9242,
+      "step": 177
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 8.432769367764916e-06,
+      "loss": 0.9432,
+      "step": 178
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.42386464826358e-06,
+      "loss": 1.0624,
+      "step": 179
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.414959928762246e-06,
+      "loss": 0.3553,
+      "step": 180
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 8.406055209260909e-06,
+      "loss": 0.827,
+      "step": 181
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.397150489759573e-06,
+      "loss": 0.8575,
+      "step": 182
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.388245770258237e-06,
+      "loss": 1.0311,
+      "step": 183
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.379341050756901e-06,
+      "loss": 1.0357,
+      "step": 184
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 8.370436331255565e-06,
+      "loss": 0.8676,
+      "step": 185
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.36153161175423e-06,
+      "loss": 0.906,
+      "step": 186
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.352626892252895e-06,
+      "loss": 0.9272,
+      "step": 187
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.34372217275156e-06,
+      "loss": 1.0062,
+      "step": 188
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 8.334817453250224e-06,
+      "loss": 0.8497,
+      "step": 189
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.325912733748888e-06,
+      "loss": 1.253,
+      "step": 190
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.317008014247552e-06,
+      "loss": 0.946,
+      "step": 191
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.308103294746216e-06,
+      "loss": 0.4112,
+      "step": 192
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 8.29919857524488e-06,
+      "loss": 0.7953,
+      "step": 193
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.290293855743544e-06,
+      "loss": 0.4192,
+      "step": 194
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.281389136242209e-06,
+      "loss": 0.9343,
+      "step": 195
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 8.272484416740873e-06,
+      "loss": 0.8515,
+      "step": 196
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.263579697239537e-06,
+      "loss": 1.0043,
+      "step": 197
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.254674977738201e-06,
+      "loss": 1.044,
+      "step": 198
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.245770258236865e-06,
+      "loss": 0.8394,
+      "step": 199
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 8.23686553873553e-06,
+      "loss": 0.8923,
+      "step": 200
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.227960819234195e-06,
+      "loss": 0.8973,
+      "step": 201
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.21905609973286e-06,
+      "loss": 0.4128,
+      "step": 202
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.210151380231524e-06,
+      "loss": 1.0187,
+      "step": 203
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 8.201246660730188e-06,
+      "loss": 0.7363,
+      "step": 204
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.192341941228852e-06,
+      "loss": 1.0541,
+      "step": 205
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.183437221727516e-06,
+      "loss": 0.8073,
+      "step": 206
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.17453250222618e-06,
+      "loss": 1.1757,
+      "step": 207
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.165627782724845e-06,
+      "loss": 0.4661,
+      "step": 208
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.156723063223509e-06,
+      "loss": 1.0298,
+      "step": 209
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.147818343722173e-06,
+      "loss": 0.2968,
+      "step": 210
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.138913624220837e-06,
+      "loss": 0.9361,
+      "step": 211
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.130008904719501e-06,
+      "loss": 0.8532,
+      "step": 212
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.121104185218166e-06,
+      "loss": 0.3431,
+      "step": 213
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.11219946571683e-06,
+      "loss": 0.9224,
+      "step": 214
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.103294746215496e-06,
+      "loss": 0.9166,
+      "step": 215
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.09439002671416e-06,
+      "loss": 1.0915,
+      "step": 216
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.085485307212824e-06,
+      "loss": 0.8984,
+      "step": 217
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.076580587711488e-06,
+      "loss": 1.0331,
+      "step": 218
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.067675868210152e-06,
+      "loss": 0.943,
+      "step": 219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.058771148708816e-06,
+      "loss": 0.9597,
+      "step": 220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.04986642920748e-06,
+      "loss": 1.0531,
+      "step": 221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.040961709706145e-06,
+      "loss": 0.9519,
+      "step": 222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 8.032056990204809e-06,
+      "loss": 0.8728,
+      "step": 223
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.023152270703473e-06,
+      "loss": 1.043,
+      "step": 224
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.014247551202137e-06,
+      "loss": 0.5001,
+      "step": 225
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 8.005342831700801e-06,
+      "loss": 0.4669,
+      "step": 226
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.996438112199466e-06,
+      "loss": 1.1133,
+      "step": 227
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.98753339269813e-06,
+      "loss": 0.9002,
+      "step": 228
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.978628673196796e-06,
+      "loss": 0.0,
+      "step": 229
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.96972395369546e-06,
+      "loss": 1.0035,
+      "step": 230
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.960819234194124e-06,
+      "loss": 0.8859,
+      "step": 231
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.951914514692788e-06,
+      "loss": 0.5216,
+      "step": 232
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.943009795191452e-06,
+      "loss": 0.9377,
+      "step": 233
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 7.934105075690117e-06,
+      "loss": 1.0615,
+      "step": 234
+    },
+    {
+      "epoch": 0.62,
+      "eval_runtime": 2902.0945,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 234
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.92520035618878e-06,
+      "loss": 0.9937,
+      "step": 235
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.916295636687445e-06,
+      "loss": 0.4319,
+      "step": 236
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.907390917186109e-06,
+      "loss": 0.9159,
+      "step": 237
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 7.898486197684773e-06,
+      "loss": 0.4726,
+      "step": 238
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.889581478183437e-06,
+      "loss": 0.4521,
+      "step": 239
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.880676758682102e-06,
+      "loss": 0.9275,
+      "step": 240
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 7.871772039180766e-06,
+      "loss": 1.0599,
+      "step": 241
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.86286731967943e-06,
+      "loss": 0.8194,
+      "step": 242
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.853962600178096e-06,
+      "loss": 0.405,
+      "step": 243
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.84505788067676e-06,
+      "loss": 0.9001,
+      "step": 244
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 7.836153161175424e-06,
+      "loss": 0.8087,
+      "step": 245
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.827248441674088e-06,
+      "loss": 1.0342,
+      "step": 246
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.818343722172753e-06,
+      "loss": 0.9034,
+      "step": 247
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.809439002671417e-06,
+      "loss": 1.0013,
+      "step": 248
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 7.800534283170081e-06,
+      "loss": 0.8647,
+      "step": 249
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.791629563668745e-06,
+      "loss": 0.8553,
+      "step": 250
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.78272484416741e-06,
+      "loss": 0.9419,
+      "step": 251
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.773820124666073e-06,
+      "loss": 1.0152,
+      "step": 252
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 7.764915405164738e-06,
+      "loss": 0.5669,
+      "step": 253
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.756010685663402e-06,
+      "loss": 0.9791,
+      "step": 254
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.747105966162066e-06,
+      "loss": 0.8199,
+      "step": 255
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 7.73820124666073e-06,
+      "loss": 0.9894,
+      "step": 256
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.729296527159396e-06,
+      "loss": 1.0476,
+      "step": 257
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.72039180765806e-06,
+      "loss": 0.793,
+      "step": 258
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.711487088156724e-06,
+      "loss": 0.9452,
+      "step": 259
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 7.702582368655389e-06,
+      "loss": 0.9307,
+      "step": 260
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.693677649154053e-06,
+      "loss": 0.3762,
+      "step": 261
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.684772929652717e-06,
+      "loss": 0.8075,
+      "step": 262
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.675868210151381e-06,
+      "loss": 0.9438,
+      "step": 263
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 7.666963490650045e-06,
+      "loss": 0.9377,
+      "step": 264
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.65805877114871e-06,
+      "loss": 0.9443,
+      "step": 265
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.649154051647374e-06,
+      "loss": 0.6893,
+      "step": 266
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.640249332146038e-06,
+      "loss": 0.8966,
+      "step": 267
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 7.631344612644702e-06,
+      "loss": 0.4491,
+      "step": 268
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.622439893143366e-06,
+      "loss": 0.7882,
+      "step": 269
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.61353517364203e-06,
+      "loss": 1.0144,
+      "step": 270
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 7.604630454140695e-06,
+      "loss": 0.8236,
+      "step": 271
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.5957257346393595e-06,
+      "loss": 0.8152,
+      "step": 272
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.586821015138024e-06,
+      "loss": 0.9237,
+      "step": 273
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.577916295636688e-06,
+      "loss": 0.9811,
+      "step": 274
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 7.569011576135352e-06,
+      "loss": 0.9554,
+      "step": 275
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.560106856634016e-06,
+      "loss": 0.9198,
+      "step": 276
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.55120213713268e-06,
+      "loss": 0.4373,
+      "step": 277
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.542297417631345e-06,
+      "loss": 0.9393,
+      "step": 278
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 7.5333926981300095e-06,
+      "loss": 0.8901,
+      "step": 279
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.524487978628674e-06,
+      "loss": 0.7799,
+      "step": 280
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.515583259127338e-06,
+      "loss": 0.8058,
+      "step": 281
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.506678539626002e-06,
+      "loss": 0.9184,
+      "step": 282
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 7.497773820124666e-06,
+      "loss": 0.7998,
+      "step": 283
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.48886910062333e-06,
+      "loss": 1.0055,
+      "step": 284
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.4799643811219954e-06,
+      "loss": 0.9932,
+      "step": 285
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 7.47105966162066e-06,
+      "loss": 0.9412,
+      "step": 286
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.462154942119324e-06,
+      "loss": 0.4469,
+      "step": 287
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.453250222617988e-06,
+      "loss": 0.4806,
+      "step": 288
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.444345503116652e-06,
+      "loss": 0.3946,
+      "step": 289
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 7.435440783615316e-06,
+      "loss": 0.8185,
+      "step": 290
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.4265360641139805e-06,
+      "loss": 0.9209,
+      "step": 291
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.4176313446126455e-06,
+      "loss": 0.4973,
+      "step": 292
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.40872662511131e-06,
+      "loss": 0.7342,
+      "step": 293
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 7.399821905609974e-06,
+      "loss": 0.8185,
+      "step": 294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.390917186108638e-06,
+      "loss": 0.4528,
+      "step": 295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.382012466607302e-06,
+      "loss": 0.902,
+      "step": 296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.373107747105966e-06,
+      "loss": 0.8621,
+      "step": 297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 7.3642030276046306e-06,
+      "loss": 0.9926,
+      "step": 298
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.355298308103296e-06,
+      "loss": 0.9885,
+      "step": 299
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.34639358860196e-06,
+      "loss": 0.9643,
+      "step": 300
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 7.337488869100624e-06,
+      "loss": 1.0397,
+      "step": 301
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.328584149599288e-06,
+      "loss": 0.7156,
+      "step": 302
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.319679430097952e-06,
+      "loss": 1.0714,
+      "step": 303
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.3107747105966165e-06,
+      "loss": 1.0542,
+      "step": 304
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 7.301869991095281e-06,
+      "loss": 0.9136,
+      "step": 305
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.292965271593946e-06,
+      "loss": 0.7983,
+      "step": 306
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.28406055209261e-06,
+      "loss": 0.9695,
+      "step": 307
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.275155832591274e-06,
+      "loss": 0.8712,
+      "step": 308
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.266251113089938e-06,
+      "loss": 0.9916,
+      "step": 309
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.257346393588602e-06,
+      "loss": 0.3765,
+      "step": 310
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2484416740872665e-06,
+      "loss": 0.9512,
+      "step": 311
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.239536954585931e-06,
+      "loss": 0.9954,
+      "step": 312
+    },
+    {
+      "epoch": 0.83,
+      "eval_runtime": 3122.2257,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 312
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.230632235084596e-06,
+      "loss": 0.9674,
+      "step": 313
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.22172751558326e-06,
+      "loss": 0.4962,
+      "step": 314
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.212822796081924e-06,
+      "loss": 0.9569,
+      "step": 315
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 7.203918076580588e-06,
+      "loss": 0.8976,
+      "step": 316
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.1950133570792524e-06,
+      "loss": 0.9225,
+      "step": 317
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.186108637577917e-06,
+      "loss": 0.8356,
+      "step": 318
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.177203918076581e-06,
+      "loss": 0.8768,
+      "step": 319
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 7.168299198575246e-06,
+      "loss": 0.8348,
+      "step": 320
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.15939447907391e-06,
+      "loss": 1.0096,
+      "step": 321
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.150489759572574e-06,
+      "loss": 0.8797,
+      "step": 322
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.141585040071238e-06,
+      "loss": 0.8958,
+      "step": 323
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 7.1326803205699025e-06,
+      "loss": 0.5527,
+      "step": 324
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.123775601068567e-06,
+      "loss": 0.8017,
+      "step": 325
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.114870881567231e-06,
+      "loss": 0.8284,
+      "step": 326
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.105966162065896e-06,
+      "loss": 0.7174,
+      "step": 327
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 7.09706144256456e-06,
+      "loss": 0.8545,
+      "step": 328
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.088156723063224e-06,
+      "loss": 0.7732,
+      "step": 329
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.079252003561888e-06,
+      "loss": 0.7843,
+      "step": 330
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.0703472840605526e-06,
+      "loss": 0.7947,
+      "step": 331
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.061442564559217e-06,
+      "loss": 0.8644,
+      "step": 332
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.052537845057881e-06,
+      "loss": 0.8802,
+      "step": 333
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.043633125556546e-06,
+      "loss": 0.9204,
+      "step": 334
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 7.03472840605521e-06,
+      "loss": 0.3897,
+      "step": 335
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.025823686553874e-06,
+      "loss": 0.9818,
+      "step": 336
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.0169189670525385e-06,
+      "loss": 0.9283,
+      "step": 337
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 7.008014247551203e-06,
+      "loss": 0.6939,
+      "step": 338
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 6.999109528049867e-06,
+      "loss": 0.392,
+      "step": 339
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.990204808548531e-06,
+      "loss": 0.8212,
+      "step": 340
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.981300089047196e-06,
+      "loss": 0.9621,
+      "step": 341
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.97239536954586e-06,
+      "loss": 0.911,
+      "step": 342
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 6.963490650044524e-06,
+      "loss": 0.7779,
+      "step": 343
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.9545859305431885e-06,
+      "loss": 0.8816,
+      "step": 344
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.945681211041853e-06,
+      "loss": 0.3727,
+      "step": 345
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 6.936776491540517e-06,
+      "loss": 0.9448,
+      "step": 346
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.927871772039181e-06,
+      "loss": 0.9095,
+      "step": 347
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.918967052537846e-06,
+      "loss": 0.8677,
+      "step": 348
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.91006233303651e-06,
+      "loss": 0.8424,
+      "step": 349
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 6.9011576135351744e-06,
+      "loss": 0.959,
+      "step": 350
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.892252894033839e-06,
+      "loss": 1.0696,
+      "step": 351
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.883348174532503e-06,
+      "loss": 0.9292,
+      "step": 352
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.874443455031167e-06,
+      "loss": 0.9112,
+      "step": 353
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 6.865538735529831e-06,
+      "loss": 0.8184,
+      "step": 354
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.856634016028496e-06,
+      "loss": 0.9305,
+      "step": 355
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.84772929652716e-06,
+      "loss": 0.8871,
+      "step": 356
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.8388245770258245e-06,
+      "loss": 0.8366,
+      "step": 357
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.829919857524489e-06,
+      "loss": 0.923,
+      "step": 358
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.821015138023153e-06,
+      "loss": 0.8915,
+      "step": 359
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.812110418521817e-06,
+      "loss": 0.9932,
+      "step": 360
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 6.803205699020481e-06,
+      "loss": 0.8932,
+      "step": 361
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.794300979519146e-06,
+      "loss": 0.7949,
+      "step": 362
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.78539626001781e-06,
+      "loss": 0.9908,
+      "step": 363
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.776491540516475e-06,
+      "loss": 0.8642,
+      "step": 364
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.767586821015139e-06,
+      "loss": 1.0203,
+      "step": 365
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.758682101513803e-06,
+      "loss": 0.9003,
+      "step": 366
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.749777382012467e-06,
+      "loss": 0.8528,
+      "step": 367
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.740872662511131e-06,
+      "loss": 0.4638,
+      "step": 368
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.731967943009796e-06,
+      "loss": 0.9339,
+      "step": 369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.7230632235084605e-06,
+      "loss": 0.4146,
+      "step": 370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.714158504007125e-06,
+      "loss": 0.85,
+      "step": 371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.705253784505789e-06,
+      "loss": 0.8628,
+      "step": 372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.696349065004453e-06,
+      "loss": 0.785,
+      "step": 373
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.687444345503117e-06,
+      "loss": 0.9177,
+      "step": 374
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.678539626001781e-06,
+      "loss": 0.9702,
+      "step": 375
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.669634906500446e-06,
+      "loss": 0.9265,
+      "step": 376
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.6607301869991106e-06,
+      "loss": 0.8732,
+      "step": 377
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.651825467497775e-06,
+      "loss": 0.765,
+      "step": 378
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.642920747996439e-06,
+      "loss": 0.8754,
+      "step": 379
+    },
+    {
+      "epoch": 1.01,
+      "learning_rate": 6.634016028495103e-06,
+      "loss": 0.8784,
+      "step": 380
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.625111308993767e-06,
+      "loss": 0.5156,
+      "step": 381
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.616206589492431e-06,
+      "loss": 0.8685,
+      "step": 382
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.6073018699910965e-06,
+      "loss": 0.4085,
+      "step": 383
+    },
+    {
+      "epoch": 1.02,
+      "learning_rate": 6.598397150489761e-06,
+      "loss": 0.4314,
+      "step": 384
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.589492430988425e-06,
+      "loss": 0.8535,
+      "step": 385
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.580587711487089e-06,
+      "loss": 0.7977,
+      "step": 386
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.571682991985753e-06,
+      "loss": 0.8364,
+      "step": 387
+    },
+    {
+      "epoch": 1.03,
+      "learning_rate": 6.5627782724844165e-06,
+      "loss": 0.9467,
+      "step": 388
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.553873552983081e-06,
+      "loss": 0.7582,
+      "step": 389
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.5449688334817465e-06,
+      "loss": 0.9325,
+      "step": 390
+    },
+    {
+      "epoch": 1.04,
+      "eval_runtime": 3122.5407,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 390
+    },
+    {
+      "epoch": 1.04,
+      "learning_rate": 6.536064113980411e-06,
+      "loss": 0.8624,
+      "step": 391
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.527159394479075e-06,
+      "loss": 0.92,
+      "step": 392
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.518254674977739e-06,
+      "loss": 0.928,
+      "step": 393
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.509349955476402e-06,
+      "loss": 0.9179,
+      "step": 394
+    },
+    {
+      "epoch": 1.05,
+      "learning_rate": 6.5004452359750666e-06,
+      "loss": 0.8678,
+      "step": 395
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.491540516473731e-06,
+      "loss": 1.0186,
+      "step": 396
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.482635796972397e-06,
+      "loss": 0.8375,
+      "step": 397
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.473731077471061e-06,
+      "loss": 0.95,
+      "step": 398
+    },
+    {
+      "epoch": 1.06,
+      "learning_rate": 6.464826357969725e-06,
+      "loss": 0.7644,
+      "step": 399
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.455921638468389e-06,
+      "loss": 0.8695,
+      "step": 400
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.4470169189670525e-06,
+      "loss": 0.5194,
+      "step": 401
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.438112199465717e-06,
+      "loss": 0.0,
+      "step": 402
+    },
+    {
+      "epoch": 1.07,
+      "learning_rate": 6.429207479964381e-06,
+      "loss": 0.7814,
+      "step": 403
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.420302760463047e-06,
+      "loss": 0.8846,
+      "step": 404
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.411398040961711e-06,
+      "loss": 0.804,
+      "step": 405
+    },
+    {
+      "epoch": 1.08,
+      "learning_rate": 6.402493321460375e-06,
+      "loss": 0.8727,
+      "step": 406
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.393588601959038e-06,
+      "loss": 0.9715,
+      "step": 407
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.3846838824577025e-06,
+      "loss": 0.4305,
+      "step": 408
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.375779162956367e-06,
+      "loss": 0.8358,
+      "step": 409
+    },
+    {
+      "epoch": 1.09,
+      "learning_rate": 6.366874443455031e-06,
+      "loss": 0.0,
+      "step": 410
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.357969723953697e-06,
+      "loss": 0.4067,
+      "step": 411
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.349065004452361e-06,
+      "loss": 0.9305,
+      "step": 412
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.340160284951024e-06,
+      "loss": 0.4258,
+      "step": 413
+    },
+    {
+      "epoch": 1.1,
+      "learning_rate": 6.3312555654496884e-06,
+      "loss": 0.8784,
+      "step": 414
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.322350845948353e-06,
+      "loss": 0.8755,
+      "step": 415
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.313446126447017e-06,
+      "loss": 0.8634,
+      "step": 416
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.304541406945681e-06,
+      "loss": 0.9332,
+      "step": 417
+    },
+    {
+      "epoch": 1.11,
+      "learning_rate": 6.295636687444347e-06,
+      "loss": 0.7862,
+      "step": 418
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.28673196794301e-06,
+      "loss": 0.9407,
+      "step": 419
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.277827248441674e-06,
+      "loss": 0.9119,
+      "step": 420
+    },
+    {
+      "epoch": 1.12,
+      "learning_rate": 6.2689225289403385e-06,
+      "loss": 0.9594,
+      "step": 421
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.260017809439003e-06,
+      "loss": 0.8862,
+      "step": 422
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.251113089937667e-06,
+      "loss": 0.8584,
+      "step": 423
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.242208370436333e-06,
+      "loss": 0.7163,
+      "step": 424
+    },
+    {
+      "epoch": 1.13,
+      "learning_rate": 6.233303650934996e-06,
+      "loss": 0.9448,
+      "step": 425
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.22439893143366e-06,
+      "loss": 0.7041,
+      "step": 426
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.215494211932324e-06,
+      "loss": 0.7988,
+      "step": 427
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.206589492430989e-06,
+      "loss": 0.8867,
+      "step": 428
+    },
+    {
+      "epoch": 1.14,
+      "learning_rate": 6.197684772929653e-06,
+      "loss": 0.918,
+      "step": 429
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.188780053428317e-06,
+      "loss": 0.952,
+      "step": 430
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.179875333926982e-06,
+      "loss": 0.7854,
+      "step": 431
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.170970614425646e-06,
+      "loss": 0.79,
+      "step": 432
+    },
+    {
+      "epoch": 1.15,
+      "learning_rate": 6.16206589492431e-06,
+      "loss": 0.5203,
+      "step": 433
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.1531611754229745e-06,
+      "loss": 0.9377,
+      "step": 434
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.144256455921639e-06,
+      "loss": 0.5002,
+      "step": 435
+    },
+    {
+      "epoch": 1.16,
+      "learning_rate": 6.135351736420303e-06,
+      "loss": 1.0624,
+      "step": 436
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.126447016918967e-06,
+      "loss": 0.7539,
+      "step": 437
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.117542297417632e-06,
+      "loss": 0.8116,
+      "step": 438
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.108637577916296e-06,
+      "loss": 0.7941,
+      "step": 439
+    },
+    {
+      "epoch": 1.17,
+      "learning_rate": 6.09973285841496e-06,
+      "loss": 0.9041,
+      "step": 440
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.0908281389136245e-06,
+      "loss": 0.8227,
+      "step": 441
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.081923419412289e-06,
+      "loss": 0.6632,
+      "step": 442
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.073018699910953e-06,
+      "loss": 0.8247,
+      "step": 443
+    },
+    {
+      "epoch": 1.18,
+      "learning_rate": 6.064113980409617e-06,
+      "loss": 0.4879,
+      "step": 444
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.055209260908282e-06,
+      "loss": 0.8061,
+      "step": 445
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.046304541406946e-06,
+      "loss": 0.3525,
+      "step": 446
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.0373998219056104e-06,
+      "loss": 0.8479,
+      "step": 447
+    },
+    {
+      "epoch": 1.19,
+      "learning_rate": 6.028495102404275e-06,
+      "loss": 0.4194,
+      "step": 448
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.019590382902939e-06,
+      "loss": 0.8421,
+      "step": 449
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.010685663401603e-06,
+      "loss": 0.8906,
+      "step": 450
+    },
+    {
+      "epoch": 1.2,
+      "learning_rate": 6.001780943900267e-06,
+      "loss": 0.9139,
+      "step": 451
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.992876224398932e-06,
+      "loss": 0.9562,
+      "step": 452
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.983971504897596e-06,
+      "loss": 0.4751,
+      "step": 453
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.9750667853962605e-06,
+      "loss": 0.7749,
+      "step": 454
+    },
+    {
+      "epoch": 1.21,
+      "learning_rate": 5.966162065894925e-06,
+      "loss": 0.8086,
+      "step": 455
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.957257346393589e-06,
+      "loss": 0.4612,
+      "step": 456
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.948352626892253e-06,
+      "loss": 0.9931,
+      "step": 457
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.939447907390917e-06,
+      "loss": 0.4716,
+      "step": 458
+    },
+    {
+      "epoch": 1.22,
+      "learning_rate": 5.930543187889582e-06,
+      "loss": 0.7453,
+      "step": 459
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.921638468388246e-06,
+      "loss": 0.7605,
+      "step": 460
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.912733748886911e-06,
+      "loss": 1.014,
+      "step": 461
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.903829029385575e-06,
+      "loss": 0.8695,
+      "step": 462
+    },
+    {
+      "epoch": 1.23,
+      "learning_rate": 5.894924309884239e-06,
+      "loss": 0.7938,
+      "step": 463
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.886019590382903e-06,
+      "loss": 0.9953,
+      "step": 464
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.877114870881567e-06,
+      "loss": 1.0152,
+      "step": 465
+    },
+    {
+      "epoch": 1.24,
+      "learning_rate": 5.868210151380232e-06,
+      "loss": 0.4923,
+      "step": 466
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.8593054318788965e-06,
+      "loss": 0.8597,
+      "step": 467
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.850400712377561e-06,
+      "loss": 0.7741,
+      "step": 468
+    },
+    {
+      "epoch": 1.25,
+      "eval_runtime": 3066.3472,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 468
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.841495992876225e-06,
+      "loss": 0.8426,
+      "step": 469
+    },
+    {
+      "epoch": 1.25,
+      "learning_rate": 5.832591273374889e-06,
+      "loss": 0.3766,
+      "step": 470
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.823686553873553e-06,
+      "loss": 0.8344,
+      "step": 471
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.814781834372217e-06,
+      "loss": 0.391,
+      "step": 472
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.805877114870882e-06,
+      "loss": 0.7629,
+      "step": 473
+    },
+    {
+      "epoch": 1.26,
+      "learning_rate": 5.7969723953695466e-06,
+      "loss": 0.602,
+      "step": 474
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.788067675868211e-06,
+      "loss": 0.9718,
+      "step": 475
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.779162956366875e-06,
+      "loss": 0.8287,
+      "step": 476
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.770258236865539e-06,
+      "loss": 0.7907,
+      "step": 477
+    },
+    {
+      "epoch": 1.27,
+      "learning_rate": 5.761353517364203e-06,
+      "loss": 0.4556,
+      "step": 478
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.7524487978628674e-06,
+      "loss": 0.7257,
+      "step": 479
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.7435440783615325e-06,
+      "loss": 0.939,
+      "step": 480
+    },
+    {
+      "epoch": 1.28,
+      "learning_rate": 5.734639358860197e-06,
+      "loss": 0.9592,
+      "step": 481
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.725734639358861e-06,
+      "loss": 0.7791,
+      "step": 482
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.716829919857525e-06,
+      "loss": 0.8606,
+      "step": 483
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.707925200356189e-06,
+      "loss": 0.373,
+      "step": 484
+    },
+    {
+      "epoch": 1.29,
+      "learning_rate": 5.699020480854853e-06,
+      "loss": 0.5349,
+      "step": 485
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6901157613535175e-06,
+      "loss": 0.7222,
+      "step": 486
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.6812110418521825e-06,
+      "loss": 0.4575,
+      "step": 487
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.672306322350847e-06,
+      "loss": 0.9973,
+      "step": 488
+    },
+    {
+      "epoch": 1.3,
+      "learning_rate": 5.663401602849511e-06,
+      "loss": 0.9032,
+      "step": 489
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.654496883348175e-06,
+      "loss": 0.8622,
+      "step": 490
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.645592163846839e-06,
+      "loss": 0.9274,
+      "step": 491
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.636687444345503e-06,
+      "loss": 0.8456,
+      "step": 492
+    },
+    {
+      "epoch": 1.31,
+      "learning_rate": 5.627782724844168e-06,
+      "loss": 0.8908,
+      "step": 493
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.618878005342833e-06,
+      "loss": 0.4209,
+      "step": 494
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.609973285841497e-06,
+      "loss": 0.8654,
+      "step": 495
+    },
+    {
+      "epoch": 1.32,
+      "learning_rate": 5.601068566340161e-06,
+      "loss": 0.9551,
+      "step": 496
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.592163846838825e-06,
+      "loss": 0.7966,
+      "step": 497
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.583259127337489e-06,
+      "loss": 0.8372,
+      "step": 498
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.5743544078361535e-06,
+      "loss": 0.4467,
+      "step": 499
+    },
+    {
+      "epoch": 1.33,
+      "learning_rate": 5.565449688334818e-06,
+      "loss": 0.8214,
+      "step": 500
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.556544968833483e-06,
+      "loss": 0.8515,
+      "step": 501
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.547640249332147e-06,
+      "loss": 0.8392,
+      "step": 502
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.538735529830811e-06,
+      "loss": 0.5566,
+      "step": 503
+    },
+    {
+      "epoch": 1.34,
+      "learning_rate": 5.529830810329475e-06,
+      "loss": 0.9163,
+      "step": 504
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.520926090828139e-06,
+      "loss": 0.7711,
+      "step": 505
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.5120213713268036e-06,
+      "loss": 0.8306,
+      "step": 506
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.503116651825468e-06,
+      "loss": 0.9389,
+      "step": 507
+    },
+    {
+      "epoch": 1.35,
+      "learning_rate": 5.494211932324133e-06,
+      "loss": 0.926,
+      "step": 508
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.485307212822797e-06,
+      "loss": 0.9432,
+      "step": 509
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.476402493321461e-06,
+      "loss": 0.8445,
+      "step": 510
+    },
+    {
+      "epoch": 1.36,
+      "learning_rate": 5.467497773820125e-06,
+      "loss": 0.8631,
+      "step": 511
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.4585930543187895e-06,
+      "loss": 0.861,
+      "step": 512
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.449688334817454e-06,
+      "loss": 0.7042,
+      "step": 513
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.440783615316118e-06,
+      "loss": 0.7012,
+      "step": 514
+    },
+    {
+      "epoch": 1.37,
+      "learning_rate": 5.431878895814783e-06,
+      "loss": 0.8749,
+      "step": 515
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.422974176313447e-06,
+      "loss": 0.9296,
+      "step": 516
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.414069456812111e-06,
+      "loss": 0.7968,
+      "step": 517
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.405164737310775e-06,
+      "loss": 0.7622,
+      "step": 518
+    },
+    {
+      "epoch": 1.38,
+      "learning_rate": 5.3962600178094395e-06,
+      "loss": 0.8212,
+      "step": 519
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.387355298308104e-06,
+      "loss": 0.3467,
+      "step": 520
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.378450578806768e-06,
+      "loss": 0.6776,
+      "step": 521
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.369545859305433e-06,
+      "loss": 0.8917,
+      "step": 522
+    },
+    {
+      "epoch": 1.39,
+      "learning_rate": 5.360641139804097e-06,
+      "loss": 0.4889,
+      "step": 523
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.351736420302761e-06,
+      "loss": 0.9016,
+      "step": 524
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.3428317008014254e-06,
+      "loss": 0.8406,
+      "step": 525
+    },
+    {
+      "epoch": 1.4,
+      "learning_rate": 5.33392698130009e-06,
+      "loss": 1.0101,
+      "step": 526
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.325022261798754e-06,
+      "loss": 0.77,
+      "step": 527
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.316117542297418e-06,
+      "loss": 0.3997,
+      "step": 528
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.307212822796083e-06,
+      "loss": 0.8681,
+      "step": 529
+    },
+    {
+      "epoch": 1.41,
+      "learning_rate": 5.298308103294747e-06,
+      "loss": 0.8111,
+      "step": 530
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.289403383793411e-06,
+      "loss": 0.9534,
+      "step": 531
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.2804986642920755e-06,
+      "loss": 0.8531,
+      "step": 532
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.27159394479074e-06,
+      "loss": 0.7534,
+      "step": 533
+    },
+    {
+      "epoch": 1.42,
+      "learning_rate": 5.262689225289404e-06,
+      "loss": 0.3682,
+      "step": 534
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.253784505788068e-06,
+      "loss": 1.0476,
+      "step": 535
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.244879786286733e-06,
+      "loss": 0.818,
+      "step": 536
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.235975066785397e-06,
+      "loss": 0.7976,
+      "step": 537
+    },
+    {
+      "epoch": 1.43,
+      "learning_rate": 5.227070347284061e-06,
+      "loss": 0.832,
+      "step": 538
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.2181656277827256e-06,
+      "loss": 0.4572,
+      "step": 539
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.20926090828139e-06,
+      "loss": 0.7371,
+      "step": 540
+    },
+    {
+      "epoch": 1.44,
+      "learning_rate": 5.200356188780054e-06,
+      "loss": 0.743,
+      "step": 541
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.191451469278718e-06,
+      "loss": 0.7998,
+      "step": 542
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.182546749777383e-06,
+      "loss": 0.8813,
+      "step": 543
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.173642030276047e-06,
+      "loss": 0.7341,
+      "step": 544
+    },
+    {
+      "epoch": 1.45,
+      "learning_rate": 5.1647373107747115e-06,
+      "loss": 0.8094,
+      "step": 545
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.155832591273376e-06,
+      "loss": 0.9034,
+      "step": 546
+    },
+    {
+      "epoch": 1.46,
+      "eval_runtime": 3122.8577,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 546
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.14692787177204e-06,
+      "loss": 0.4552,
+      "step": 547
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.138023152270704e-06,
+      "loss": 0.8386,
+      "step": 548
+    },
+    {
+      "epoch": 1.46,
+      "learning_rate": 5.129118432769367e-06,
+      "loss": 0.862,
+      "step": 549
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.120213713268033e-06,
+      "loss": 0.3847,
+      "step": 550
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.111308993766697e-06,
+      "loss": 0.7881,
+      "step": 551
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.1024042742653615e-06,
+      "loss": 0.4372,
+      "step": 552
+    },
+    {
+      "epoch": 1.47,
+      "learning_rate": 5.093499554764026e-06,
+      "loss": 0.4659,
+      "step": 553
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.08459483526269e-06,
+      "loss": 0.4164,
+      "step": 554
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.075690115761353e-06,
+      "loss": 0.8909,
+      "step": 555
+    },
+    {
+      "epoch": 1.48,
+      "learning_rate": 5.066785396260017e-06,
+      "loss": 0.8409,
+      "step": 556
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.057880676758683e-06,
+      "loss": 0.9546,
+      "step": 557
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.0489759572573474e-06,
+      "loss": 0.398,
+      "step": 558
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.040071237756012e-06,
+      "loss": 0.9555,
+      "step": 559
+    },
+    {
+      "epoch": 1.49,
+      "learning_rate": 5.031166518254676e-06,
+      "loss": 0.4583,
+      "step": 560
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.022261798753339e-06,
+      "loss": 0.8357,
+      "step": 561
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.013357079252003e-06,
+      "loss": 0.7141,
+      "step": 562
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 5.0044523597506675e-06,
+      "loss": 0.9004,
+      "step": 563
+    },
+    {
+      "epoch": 1.5,
+      "learning_rate": 4.9955476402493325e-06,
+      "loss": 0.7932,
+      "step": 564
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.986642920747997e-06,
+      "loss": 0.9521,
+      "step": 565
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.977738201246662e-06,
+      "loss": 0.9708,
+      "step": 566
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.968833481745325e-06,
+      "loss": 0.4273,
+      "step": 567
+    },
+    {
+      "epoch": 1.51,
+      "learning_rate": 4.959928762243989e-06,
+      "loss": 0.7986,
+      "step": 568
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.951024042742654e-06,
+      "loss": 0.7576,
+      "step": 569
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.942119323241318e-06,
+      "loss": 0.9655,
+      "step": 570
+    },
+    {
+      "epoch": 1.52,
+      "learning_rate": 4.9332146037399826e-06,
+      "loss": 0.8743,
+      "step": 571
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.924309884238647e-06,
+      "loss": 0.8293,
+      "step": 572
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.915405164737311e-06,
+      "loss": 0.8109,
+      "step": 573
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.906500445235975e-06,
+      "loss": 0.837,
+      "step": 574
+    },
+    {
+      "epoch": 1.53,
+      "learning_rate": 4.897595725734639e-06,
+      "loss": 0.4545,
+      "step": 575
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.888691006233304e-06,
+      "loss": 0.8971,
+      "step": 576
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.8797862867319685e-06,
+      "loss": 0.0,
+      "step": 577
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.870881567230633e-06,
+      "loss": 0.8086,
+      "step": 578
+    },
+    {
+      "epoch": 1.54,
+      "learning_rate": 4.861976847729297e-06,
+      "loss": 0.7675,
+      "step": 579
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.853072128227961e-06,
+      "loss": 0.8417,
+      "step": 580
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.844167408726625e-06,
+      "loss": 0.3466,
+      "step": 581
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.835262689225289e-06,
+      "loss": 0.7503,
+      "step": 582
+    },
+    {
+      "epoch": 1.55,
+      "learning_rate": 4.826357969723954e-06,
+      "loss": 0.9633,
+      "step": 583
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.8174532502226185e-06,
+      "loss": 0.8579,
+      "step": 584
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.808548530721283e-06,
+      "loss": 0.8327,
+      "step": 585
+    },
+    {
+      "epoch": 1.56,
+      "learning_rate": 4.799643811219947e-06,
+      "loss": 0.4118,
+      "step": 586
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.790739091718611e-06,
+      "loss": 1.0814,
+      "step": 587
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.781834372217275e-06,
+      "loss": 0.8208,
+      "step": 588
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.772929652715939e-06,
+      "loss": 0.6655,
+      "step": 589
+    },
+    {
+      "epoch": 1.57,
+      "learning_rate": 4.7640249332146044e-06,
+      "loss": 0.3753,
+      "step": 590
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.755120213713269e-06,
+      "loss": 0.7219,
+      "step": 591
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.746215494211933e-06,
+      "loss": 0.9511,
+      "step": 592
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.737310774710597e-06,
+      "loss": 0.9182,
+      "step": 593
+    },
+    {
+      "epoch": 1.58,
+      "learning_rate": 4.728406055209261e-06,
+      "loss": 0.903,
+      "step": 594
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.719501335707925e-06,
+      "loss": 0.3432,
+      "step": 595
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7105966162065895e-06,
+      "loss": 0.8602,
+      "step": 596
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.7016918967052545e-06,
+      "loss": 0.8359,
+      "step": 597
+    },
+    {
+      "epoch": 1.59,
+      "learning_rate": 4.692787177203919e-06,
+      "loss": 0.8604,
+      "step": 598
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.683882457702583e-06,
+      "loss": 0.8918,
+      "step": 599
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.674977738201247e-06,
+      "loss": 0.8516,
+      "step": 600
+    },
+    {
+      "epoch": 1.6,
+      "learning_rate": 4.666073018699911e-06,
+      "loss": 0.871,
+      "step": 601
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.657168299198575e-06,
+      "loss": 0.7887,
+      "step": 602
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.6482635796972396e-06,
+      "loss": 0.8715,
+      "step": 603
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.639358860195905e-06,
+      "loss": 0.8054,
+      "step": 604
+    },
+    {
+      "epoch": 1.61,
+      "learning_rate": 4.630454140694569e-06,
+      "loss": 0.9374,
+      "step": 605
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.621549421193233e-06,
+      "loss": 0.931,
+      "step": 606
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.612644701691897e-06,
+      "loss": 0.8014,
+      "step": 607
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.603739982190561e-06,
+      "loss": 0.4344,
+      "step": 608
+    },
+    {
+      "epoch": 1.62,
+      "learning_rate": 4.5948352626892255e-06,
+      "loss": 0.7468,
+      "step": 609
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.58593054318789e-06,
+      "loss": 0.8325,
+      "step": 610
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.577025823686555e-06,
+      "loss": 0.7677,
+      "step": 611
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.568121104185219e-06,
+      "loss": 0.7887,
+      "step": 612
+    },
+    {
+      "epoch": 1.63,
+      "learning_rate": 4.559216384683883e-06,
+      "loss": 0.7649,
+      "step": 613
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.550311665182547e-06,
+      "loss": 0.7567,
+      "step": 614
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.541406945681211e-06,
+      "loss": 0.8182,
+      "step": 615
+    },
+    {
+      "epoch": 1.64,
+      "learning_rate": 4.5325022261798755e-06,
+      "loss": 0.7293,
+      "step": 616
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.52359750667854e-06,
+      "loss": 0.7722,
+      "step": 617
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.514692787177205e-06,
+      "loss": 0.7997,
+      "step": 618
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.505788067675869e-06,
+      "loss": 0.3808,
+      "step": 619
+    },
+    {
+      "epoch": 1.65,
+      "learning_rate": 4.496883348174533e-06,
+      "loss": 0.7773,
+      "step": 620
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.487978628673197e-06,
+      "loss": 0.9762,
+      "step": 621
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.4790739091718614e-06,
+      "loss": 0.4097,
+      "step": 622
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.470169189670526e-06,
+      "loss": 0.7206,
+      "step": 623
+    },
+    {
+      "epoch": 1.66,
+      "learning_rate": 4.46126447016919e-06,
+      "loss": 0.8525,
+      "step": 624
+    },
+    {
+      "epoch": 1.66,
+      "eval_runtime": 3122.4132,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 624
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.452359750667855e-06,
+      "loss": 0.9353,
+      "step": 625
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.443455031166519e-06,
+      "loss": 0.8174,
+      "step": 626
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.434550311665183e-06,
+      "loss": 0.8533,
+      "step": 627
+    },
+    {
+      "epoch": 1.67,
+      "learning_rate": 4.425645592163847e-06,
+      "loss": 1.0085,
+      "step": 628
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.4167408726625115e-06,
+      "loss": 0.7917,
+      "step": 629
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.407836153161176e-06,
+      "loss": 0.8576,
+      "step": 630
+    },
+    {
+      "epoch": 1.68,
+      "learning_rate": 4.39893143365984e-06,
+      "loss": 0.8541,
+      "step": 631
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.390026714158505e-06,
+      "loss": 0.775,
+      "step": 632
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.381121994657169e-06,
+      "loss": 0.9133,
+      "step": 633
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.372217275155833e-06,
+      "loss": 0.7304,
+      "step": 634
+    },
+    {
+      "epoch": 1.69,
+      "learning_rate": 4.363312555654497e-06,
+      "loss": 0.7185,
+      "step": 635
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.354407836153162e-06,
+      "loss": 0.7191,
+      "step": 636
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.345503116651826e-06,
+      "loss": 0.4596,
+      "step": 637
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.33659839715049e-06,
+      "loss": 0.7959,
+      "step": 638
+    },
+    {
+      "epoch": 1.7,
+      "learning_rate": 4.327693677649155e-06,
+      "loss": 0.8166,
+      "step": 639
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.318788958147819e-06,
+      "loss": 0.7797,
+      "step": 640
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3098842386464825e-06,
+      "loss": 0.3484,
+      "step": 641
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.3009795191451475e-06,
+      "loss": 0.4098,
+      "step": 642
+    },
+    {
+      "epoch": 1.71,
+      "learning_rate": 4.292074799643812e-06,
+      "loss": 0.8647,
+      "step": 643
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.283170080142476e-06,
+      "loss": 0.7583,
+      "step": 644
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.27426536064114e-06,
+      "loss": 0.7828,
+      "step": 645
+    },
+    {
+      "epoch": 1.72,
+      "learning_rate": 4.265360641139805e-06,
+      "loss": 0.8213,
+      "step": 646
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.256455921638468e-06,
+      "loss": 0.7223,
+      "step": 647
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2475512021371325e-06,
+      "loss": 0.8994,
+      "step": 648
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.2386464826357976e-06,
+      "loss": 0.7186,
+      "step": 649
+    },
+    {
+      "epoch": 1.73,
+      "learning_rate": 4.229741763134462e-06,
+      "loss": 0.7235,
+      "step": 650
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.220837043633126e-06,
+      "loss": 0.7809,
+      "step": 651
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.21193232413179e-06,
+      "loss": 0.9377,
+      "step": 652
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.203027604630454e-06,
+      "loss": 0.4418,
+      "step": 653
+    },
+    {
+      "epoch": 1.74,
+      "learning_rate": 4.1941228851291184e-06,
+      "loss": 0.6694,
+      "step": 654
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.185218165627783e-06,
+      "loss": 0.7216,
+      "step": 655
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.176313446126448e-06,
+      "loss": 0.4918,
+      "step": 656
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.167408726625112e-06,
+      "loss": 0.8457,
+      "step": 657
+    },
+    {
+      "epoch": 1.75,
+      "learning_rate": 4.158504007123776e-06,
+      "loss": 0.7652,
+      "step": 658
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.14959928762244e-06,
+      "loss": 0.8186,
+      "step": 659
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.140694568121104e-06,
+      "loss": 0.8579,
+      "step": 660
+    },
+    {
+      "epoch": 1.76,
+      "learning_rate": 4.1317898486197685e-06,
+      "loss": 0.8741,
+      "step": 661
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.122885129118433e-06,
+      "loss": 0.7352,
+      "step": 662
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.113980409617098e-06,
+      "loss": 0.8883,
+      "step": 663
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.105075690115762e-06,
+      "loss": 0.6495,
+      "step": 664
+    },
+    {
+      "epoch": 1.77,
+      "learning_rate": 4.096170970614426e-06,
+      "loss": 0.7221,
+      "step": 665
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.08726625111309e-06,
+      "loss": 1.0745,
+      "step": 666
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.078361531611754e-06,
+      "loss": 0.8604,
+      "step": 667
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.0694568121104186e-06,
+      "loss": 0.8786,
+      "step": 668
+    },
+    {
+      "epoch": 1.78,
+      "learning_rate": 4.060552092609083e-06,
+      "loss": 0.809,
+      "step": 669
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.051647373107748e-06,
+      "loss": 0.9381,
+      "step": 670
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.042742653606412e-06,
+      "loss": 0.7242,
+      "step": 671
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.033837934105076e-06,
+      "loss": 0.6726,
+      "step": 672
+    },
+    {
+      "epoch": 1.79,
+      "learning_rate": 4.02493321460374e-06,
+      "loss": 0.9276,
+      "step": 673
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.0160284951024045e-06,
+      "loss": 0.8467,
+      "step": 674
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 4.007123775601069e-06,
+      "loss": 0.7156,
+      "step": 675
+    },
+    {
+      "epoch": 1.8,
+      "learning_rate": 3.998219056099733e-06,
+      "loss": 0.9258,
+      "step": 676
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.989314336598398e-06,
+      "loss": 0.8715,
+      "step": 677
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.980409617097062e-06,
+      "loss": 0.7564,
+      "step": 678
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.971504897595726e-06,
+      "loss": 0.8953,
+      "step": 679
+    },
+    {
+      "epoch": 1.81,
+      "learning_rate": 3.96260017809439e-06,
+      "loss": 0.8331,
+      "step": 680
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.9536954585930545e-06,
+      "loss": 0.7388,
+      "step": 681
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.944790739091719e-06,
+      "loss": 0.8656,
+      "step": 682
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.935886019590383e-06,
+      "loss": 0.9194,
+      "step": 683
+    },
+    {
+      "epoch": 1.82,
+      "learning_rate": 3.926981300089048e-06,
+      "loss": 0.7962,
+      "step": 684
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.918076580587712e-06,
+      "loss": 0.8706,
+      "step": 685
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.909171861086376e-06,
+      "loss": 0.3566,
+      "step": 686
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.9002671415850404e-06,
+      "loss": 0.7434,
+      "step": 687
+    },
+    {
+      "epoch": 1.83,
+      "learning_rate": 3.891362422083705e-06,
+      "loss": 0.918,
+      "step": 688
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.882457702582369e-06,
+      "loss": 0.7629,
+      "step": 689
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.873552983081033e-06,
+      "loss": 0.7918,
+      "step": 690
+    },
+    {
+      "epoch": 1.84,
+      "learning_rate": 3.864648263579698e-06,
+      "loss": 1.1052,
+      "step": 691
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.855743544078362e-06,
+      "loss": 0.8304,
+      "step": 692
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.846838824577026e-06,
+      "loss": 0.8761,
+      "step": 693
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.8379341050756905e-06,
+      "loss": 0.4308,
+      "step": 694
+    },
+    {
+      "epoch": 1.85,
+      "learning_rate": 3.829029385574355e-06,
+      "loss": 0.9123,
+      "step": 695
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.820124666073019e-06,
+      "loss": 0.4577,
+      "step": 696
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.811219946571683e-06,
+      "loss": 0.818,
+      "step": 697
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.8023152270703476e-06,
+      "loss": 0.8135,
+      "step": 698
+    },
+    {
+      "epoch": 1.86,
+      "learning_rate": 3.793410507569012e-06,
+      "loss": 0.8271,
+      "step": 699
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.784505788067676e-06,
+      "loss": 0.8895,
+      "step": 700
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.77560106856634e-06,
+      "loss": 0.6559,
+      "step": 701
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.7666963490650048e-06,
+      "loss": 0.83,
+      "step": 702
+    },
+    {
+      "epoch": 1.87,
+      "eval_runtime": 2848.8264,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 702
+    },
+    {
+      "epoch": 1.87,
+      "learning_rate": 3.757791629563669e-06,
+      "loss": 0.7539,
+      "step": 703
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.748886910062333e-06,
+      "loss": 0.9391,
+      "step": 704
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.7399821905609977e-06,
+      "loss": 0.9129,
+      "step": 705
+    },
+    {
+      "epoch": 1.88,
+      "learning_rate": 3.731077471059662e-06,
+      "loss": 0.7648,
+      "step": 706
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.722172751558326e-06,
+      "loss": 0.0,
+      "step": 707
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.7132680320569902e-06,
+      "loss": 0.4201,
+      "step": 708
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.704363312555655e-06,
+      "loss": 0.7903,
+      "step": 709
+    },
+    {
+      "epoch": 1.89,
+      "learning_rate": 3.695458593054319e-06,
+      "loss": 0.7851,
+      "step": 710
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.686553873552983e-06,
+      "loss": 0.4806,
+      "step": 711
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.677649154051648e-06,
+      "loss": 0.7905,
+      "step": 712
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.668744434550312e-06,
+      "loss": 0.7273,
+      "step": 713
+    },
+    {
+      "epoch": 1.9,
+      "learning_rate": 3.659839715048976e-06,
+      "loss": 0.9456,
+      "step": 714
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.6509349955476403e-06,
+      "loss": 0.8671,
+      "step": 715
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.642030276046305e-06,
+      "loss": 0.774,
+      "step": 716
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.633125556544969e-06,
+      "loss": 0.7052,
+      "step": 717
+    },
+    {
+      "epoch": 1.91,
+      "learning_rate": 3.6242208370436333e-06,
+      "loss": 0.8426,
+      "step": 718
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.615316117542298e-06,
+      "loss": 0.9182,
+      "step": 719
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.606411398040962e-06,
+      "loss": 0.8033,
+      "step": 720
+    },
+    {
+      "epoch": 1.92,
+      "learning_rate": 3.5975066785396262e-06,
+      "loss": 0.7312,
+      "step": 721
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5886019590382904e-06,
+      "loss": 0.5093,
+      "step": 722
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.579697239536955e-06,
+      "loss": 0.8947,
+      "step": 723
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.570792520035619e-06,
+      "loss": 0.9183,
+      "step": 724
+    },
+    {
+      "epoch": 1.93,
+      "learning_rate": 3.5618878005342833e-06,
+      "loss": 0.39,
+      "step": 725
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.552983081032948e-06,
+      "loss": 0.8709,
+      "step": 726
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.544078361531612e-06,
+      "loss": 0.9282,
+      "step": 727
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5351736420302763e-06,
+      "loss": 0.8685,
+      "step": 728
+    },
+    {
+      "epoch": 1.94,
+      "learning_rate": 3.5262689225289405e-06,
+      "loss": 0.7234,
+      "step": 729
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.517364203027605e-06,
+      "loss": 0.9683,
+      "step": 730
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.5084594835262692e-06,
+      "loss": 0.3982,
+      "step": 731
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.4995547640249334e-06,
+      "loss": 0.7382,
+      "step": 732
+    },
+    {
+      "epoch": 1.95,
+      "learning_rate": 3.490650044523598e-06,
+      "loss": 0.9388,
+      "step": 733
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.481745325022262e-06,
+      "loss": 0.7842,
+      "step": 734
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4728406055209264e-06,
+      "loss": 0.8743,
+      "step": 735
+    },
+    {
+      "epoch": 1.96,
+      "learning_rate": 3.4639358860195905e-06,
+      "loss": 0.7962,
+      "step": 736
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.455031166518255e-06,
+      "loss": 0.3478,
+      "step": 737
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4461264470169193e-06,
+      "loss": 0.4107,
+      "step": 738
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.4372217275155835e-06,
+      "loss": 0.7844,
+      "step": 739
+    },
+    {
+      "epoch": 1.97,
+      "learning_rate": 3.428317008014248e-06,
+      "loss": 0.6875,
+      "step": 740
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4194122885129123e-06,
+      "loss": 0.9117,
+      "step": 741
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4105075690115764e-06,
+      "loss": 0.8503,
+      "step": 742
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.4016028495102406e-06,
+      "loss": 0.7383,
+      "step": 743
+    },
+    {
+      "epoch": 1.98,
+      "learning_rate": 3.392698130008905e-06,
+      "loss": 0.8355,
+      "step": 744
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3837934105075694e-06,
+      "loss": 0.4949,
+      "step": 745
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3748886910062336e-06,
+      "loss": 0.3459,
+      "step": 746
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.365983971504898e-06,
+      "loss": 0.8427,
+      "step": 747
+    },
+    {
+      "epoch": 1.99,
+      "learning_rate": 3.3570792520035623e-06,
+      "loss": 0.7547,
+      "step": 748
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3481745325022265e-06,
+      "loss": 0.7757,
+      "step": 749
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3392698130008907e-06,
+      "loss": 0.3703,
+      "step": 750
+    },
+    {
+      "epoch": 2.0,
+      "learning_rate": 3.3303650934995553e-06,
+      "loss": 0.7,
+      "step": 751
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3214603739982195e-06,
+      "loss": 0.7939,
+      "step": 752
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3125556544968836e-06,
+      "loss": 0.3969,
+      "step": 753
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.3036509349955482e-06,
+      "loss": 0.9467,
+      "step": 754
+    },
+    {
+      "epoch": 2.01,
+      "learning_rate": 3.2947462154942124e-06,
+      "loss": 0.7383,
+      "step": 755
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2858414959928766e-06,
+      "loss": 0.8772,
+      "step": 756
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2769367764915403e-06,
+      "loss": 0.0,
+      "step": 757
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2680320569902054e-06,
+      "loss": 0.3946,
+      "step": 758
+    },
+    {
+      "epoch": 2.02,
+      "learning_rate": 3.2591273374888695e-06,
+      "loss": 0.3985,
+      "step": 759
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.2502226179875333e-06,
+      "loss": 0.8536,
+      "step": 760
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.2413178984861983e-06,
+      "loss": 0.8032,
+      "step": 761
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.2324131789848625e-06,
+      "loss": 0.7619,
+      "step": 762
+    },
+    {
+      "epoch": 2.03,
+      "learning_rate": 3.2235084594835262e-06,
+      "loss": 0.7849,
+      "step": 763
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.2146037399821904e-06,
+      "loss": 0.8194,
+      "step": 764
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.2056990204808554e-06,
+      "loss": 0.426,
+      "step": 765
+    },
+    {
+      "epoch": 2.04,
+      "learning_rate": 3.196794300979519e-06,
+      "loss": 0.7653,
+      "step": 766
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.1878895814781834e-06,
+      "loss": 0.9101,
+      "step": 767
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.1789848619768484e-06,
+      "loss": 0.3441,
+      "step": 768
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.170080142475512e-06,
+      "loss": 0.3691,
+      "step": 769
+    },
+    {
+      "epoch": 2.05,
+      "learning_rate": 3.1611754229741763e-06,
+      "loss": 0.4104,
+      "step": 770
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.1522707034728405e-06,
+      "loss": 0.7852,
+      "step": 771
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.143365983971505e-06,
+      "loss": 0.7601,
+      "step": 772
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.1344612644701693e-06,
+      "loss": 0.7588,
+      "step": 773
+    },
+    {
+      "epoch": 2.06,
+      "learning_rate": 3.1255565449688334e-06,
+      "loss": 0.4328,
+      "step": 774
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.116651825467498e-06,
+      "loss": 0.6969,
+      "step": 775
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.107747105966162e-06,
+      "loss": 0.9033,
+      "step": 776
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.0988423864648264e-06,
+      "loss": 0.6618,
+      "step": 777
+    },
+    {
+      "epoch": 2.07,
+      "learning_rate": 3.089937666963491e-06,
+      "loss": 0.6033,
+      "step": 778
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.081032947462155e-06,
+      "loss": 0.8574,
+      "step": 779
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.0721282279608193e-06,
+      "loss": 0.7092,
+      "step": 780
+    },
+    {
+      "epoch": 2.08,
+      "eval_runtime": 3064.0098,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 780
+    },
+    {
+      "epoch": 2.08,
+      "learning_rate": 3.0632235084594835e-06,
+      "loss": 0.8391,
+      "step": 781
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.054318788958148e-06,
+      "loss": 0.861,
+      "step": 782
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.0454140694568123e-06,
+      "loss": 0.4563,
+      "step": 783
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.0365093499554764e-06,
+      "loss": 0.7271,
+      "step": 784
+    },
+    {
+      "epoch": 2.09,
+      "learning_rate": 3.027604630454141e-06,
+      "loss": 0.9406,
+      "step": 785
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.0186999109528052e-06,
+      "loss": 0.2791,
+      "step": 786
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.0097951914514694e-06,
+      "loss": 0.3954,
+      "step": 787
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 3.0008904719501336e-06,
+      "loss": 0.8004,
+      "step": 788
+    },
+    {
+      "epoch": 2.1,
+      "learning_rate": 2.991985752448798e-06,
+      "loss": 0.3988,
+      "step": 789
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9830810329474623e-06,
+      "loss": 0.8197,
+      "step": 790
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9741763134461265e-06,
+      "loss": 0.9167,
+      "step": 791
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.965271593944791e-06,
+      "loss": 0.6647,
+      "step": 792
+    },
+    {
+      "epoch": 2.11,
+      "learning_rate": 2.9563668744434553e-06,
+      "loss": 0.7018,
+      "step": 793
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9474621549421195e-06,
+      "loss": 0.798,
+      "step": 794
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9385574354407836e-06,
+      "loss": 0.8137,
+      "step": 795
+    },
+    {
+      "epoch": 2.12,
+      "learning_rate": 2.9296527159394482e-06,
+      "loss": 0.8914,
+      "step": 796
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.9207479964381124e-06,
+      "loss": 0.984,
+      "step": 797
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.9118432769367766e-06,
+      "loss": 0.9731,
+      "step": 798
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.902938557435441e-06,
+      "loss": 0.3192,
+      "step": 799
+    },
+    {
+      "epoch": 2.13,
+      "learning_rate": 2.8940338379341054e-06,
+      "loss": 0.3567,
+      "step": 800
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8851291184327695e-06,
+      "loss": 0.4601,
+      "step": 801
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8762243989314337e-06,
+      "loss": 0.8648,
+      "step": 802
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8673196794300983e-06,
+      "loss": 0.7561,
+      "step": 803
+    },
+    {
+      "epoch": 2.14,
+      "learning_rate": 2.8584149599287625e-06,
+      "loss": 0.4967,
+      "step": 804
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8495102404274267e-06,
+      "loss": 0.7973,
+      "step": 805
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8406055209260913e-06,
+      "loss": 0.3771,
+      "step": 806
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8317008014247554e-06,
+      "loss": 0.7678,
+      "step": 807
+    },
+    {
+      "epoch": 2.15,
+      "learning_rate": 2.8227960819234196e-06,
+      "loss": 0.7436,
+      "step": 808
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.813891362422084e-06,
+      "loss": 0.7273,
+      "step": 809
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.8049866429207484e-06,
+      "loss": 0.7536,
+      "step": 810
+    },
+    {
+      "epoch": 2.16,
+      "learning_rate": 2.7960819234194126e-06,
+      "loss": 0.7986,
+      "step": 811
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7871772039180767e-06,
+      "loss": 0.7831,
+      "step": 812
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7782724844167413e-06,
+      "loss": 0.8599,
+      "step": 813
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7693677649154055e-06,
+      "loss": 0.858,
+      "step": 814
+    },
+    {
+      "epoch": 2.17,
+      "learning_rate": 2.7604630454140697e-06,
+      "loss": 0.8767,
+      "step": 815
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.751558325912734e-06,
+      "loss": 0.8197,
+      "step": 816
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.7426536064113985e-06,
+      "loss": 0.8517,
+      "step": 817
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.7337488869100626e-06,
+      "loss": 0.8084,
+      "step": 818
+    },
+    {
+      "epoch": 2.18,
+      "learning_rate": 2.724844167408727e-06,
+      "loss": 0.8419,
+      "step": 819
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.7159394479073914e-06,
+      "loss": 0.7501,
+      "step": 820
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.7070347284060556e-06,
+      "loss": 0.758,
+      "step": 821
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.6981300089047198e-06,
+      "loss": 0.8042,
+      "step": 822
+    },
+    {
+      "epoch": 2.19,
+      "learning_rate": 2.689225289403384e-06,
+      "loss": 0.9276,
+      "step": 823
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.6803205699020485e-06,
+      "loss": 0.9257,
+      "step": 824
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.6714158504007127e-06,
+      "loss": 0.8359,
+      "step": 825
+    },
+    {
+      "epoch": 2.2,
+      "learning_rate": 2.662511130899377e-06,
+      "loss": 0.8369,
+      "step": 826
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.6536064113980415e-06,
+      "loss": 0.3517,
+      "step": 827
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.6447016918967057e-06,
+      "loss": 0.7807,
+      "step": 828
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.63579697239537e-06,
+      "loss": 1.0409,
+      "step": 829
+    },
+    {
+      "epoch": 2.21,
+      "learning_rate": 2.626892252894034e-06,
+      "loss": 0.788,
+      "step": 830
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.6179875333926986e-06,
+      "loss": 0.7939,
+      "step": 831
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.6090828138913628e-06,
+      "loss": 0.9022,
+      "step": 832
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.600178094390027e-06,
+      "loss": 0.3745,
+      "step": 833
+    },
+    {
+      "epoch": 2.22,
+      "learning_rate": 2.5912733748886916e-06,
+      "loss": 0.3296,
+      "step": 834
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5823686553873557e-06,
+      "loss": 0.8166,
+      "step": 835
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.57346393588602e-06,
+      "loss": 0.3616,
+      "step": 836
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5645592163846837e-06,
+      "loss": 0.3693,
+      "step": 837
+    },
+    {
+      "epoch": 2.23,
+      "learning_rate": 2.5556544968833487e-06,
+      "loss": 0.5391,
+      "step": 838
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.546749777382013e-06,
+      "loss": 0.8638,
+      "step": 839
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5378450578806766e-06,
+      "loss": 0.818,
+      "step": 840
+    },
+    {
+      "epoch": 2.24,
+      "learning_rate": 2.5289403383793416e-06,
+      "loss": 0.8057,
+      "step": 841
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.520035618878006e-06,
+      "loss": 0.6899,
+      "step": 842
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5111308993766696e-06,
+      "loss": 0.8833,
+      "step": 843
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.5022261798753337e-06,
+      "loss": 0.7854,
+      "step": 844
+    },
+    {
+      "epoch": 2.25,
+      "learning_rate": 2.4933214603739983e-06,
+      "loss": 0.9113,
+      "step": 845
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4844167408726625e-06,
+      "loss": 0.8731,
+      "step": 846
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.475512021371327e-06,
+      "loss": 1.0896,
+      "step": 847
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4666073018699913e-06,
+      "loss": 0.8068,
+      "step": 848
+    },
+    {
+      "epoch": 2.26,
+      "learning_rate": 2.4577025823686555e-06,
+      "loss": 0.8529,
+      "step": 849
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4487978628673196e-06,
+      "loss": 0.8225,
+      "step": 850
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4398931433659842e-06,
+      "loss": 0.8531,
+      "step": 851
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4309884238646484e-06,
+      "loss": 0.8665,
+      "step": 852
+    },
+    {
+      "epoch": 2.27,
+      "learning_rate": 2.4220837043633126e-06,
+      "loss": 0.873,
+      "step": 853
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.413178984861977e-06,
+      "loss": 0.8203,
+      "step": 854
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.4042742653606414e-06,
+      "loss": 0.4228,
+      "step": 855
+    },
+    {
+      "epoch": 2.28,
+      "learning_rate": 2.3953695458593055e-06,
+      "loss": 0.3584,
+      "step": 856
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3864648263579697e-06,
+      "loss": 0.9292,
+      "step": 857
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3775601068566343e-06,
+      "loss": 0.8707,
+      "step": 858
+    },
+    {
+      "epoch": 2.29,
+      "eval_runtime": 3121.0977,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 858
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3686553873552985e-06,
+      "loss": 0.7644,
+      "step": 859
+    },
+    {
+      "epoch": 2.29,
+      "learning_rate": 2.3597506678539627e-06,
+      "loss": 0.8001,
+      "step": 860
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3508459483526273e-06,
+      "loss": 0.3181,
+      "step": 861
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3419412288512914e-06,
+      "loss": 0.8617,
+      "step": 862
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3330365093499556e-06,
+      "loss": 0.7716,
+      "step": 863
+    },
+    {
+      "epoch": 2.3,
+      "learning_rate": 2.3241317898486198e-06,
+      "loss": 0.2911,
+      "step": 864
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3152270703472844e-06,
+      "loss": 0.788,
+      "step": 865
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.3063223508459486e-06,
+      "loss": 0.871,
+      "step": 866
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2974176313446127e-06,
+      "loss": 0.7019,
+      "step": 867
+    },
+    {
+      "epoch": 2.31,
+      "learning_rate": 2.2885129118432773e-06,
+      "loss": 0.4516,
+      "step": 868
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.2796081923419415e-06,
+      "loss": 0.9156,
+      "step": 869
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.2707034728406057e-06,
+      "loss": 0.7424,
+      "step": 870
+    },
+    {
+      "epoch": 2.32,
+      "learning_rate": 2.26179875333927e-06,
+      "loss": 0.3835,
+      "step": 871
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2528940338379345e-06,
+      "loss": 0.7874,
+      "step": 872
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2439893143365986e-06,
+      "loss": 0.7946,
+      "step": 873
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.235084594835263e-06,
+      "loss": 0.7744,
+      "step": 874
+    },
+    {
+      "epoch": 2.33,
+      "learning_rate": 2.2261798753339274e-06,
+      "loss": 0.8611,
+      "step": 875
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2172751558325916e-06,
+      "loss": 0.7996,
+      "step": 876
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.2083704363312558e-06,
+      "loss": 0.8738,
+      "step": 877
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.19946571682992e-06,
+      "loss": 0.8718,
+      "step": 878
+    },
+    {
+      "epoch": 2.34,
+      "learning_rate": 2.1905609973285845e-06,
+      "loss": 0.7943,
+      "step": 879
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.1816562778272487e-06,
+      "loss": 0.3798,
+      "step": 880
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.172751558325913e-06,
+      "loss": 0.8158,
+      "step": 881
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.1638468388245775e-06,
+      "loss": 0.4198,
+      "step": 882
+    },
+    {
+      "epoch": 2.35,
+      "learning_rate": 2.1549421193232412e-06,
+      "loss": 0.775,
+      "step": 883
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.146037399821906e-06,
+      "loss": 0.8699,
+      "step": 884
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.13713268032057e-06,
+      "loss": 0.5129,
+      "step": 885
+    },
+    {
+      "epoch": 2.36,
+      "learning_rate": 2.128227960819234e-06,
+      "loss": 0.6772,
+      "step": 886
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.1193232413178988e-06,
+      "loss": 0.8879,
+      "step": 887
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.110418521816563e-06,
+      "loss": 0.7931,
+      "step": 888
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.101513802315227e-06,
+      "loss": 0.7196,
+      "step": 889
+    },
+    {
+      "epoch": 2.37,
+      "learning_rate": 2.0926090828138913e-06,
+      "loss": 0.4805,
+      "step": 890
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.083704363312556e-06,
+      "loss": 0.827,
+      "step": 891
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.07479964381122e-06,
+      "loss": 0.7838,
+      "step": 892
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.0658949243098843e-06,
+      "loss": 0.8366,
+      "step": 893
+    },
+    {
+      "epoch": 2.38,
+      "learning_rate": 2.056990204808549e-06,
+      "loss": 0.7436,
+      "step": 894
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.048085485307213e-06,
+      "loss": 0.8883,
+      "step": 895
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.039180765805877e-06,
+      "loss": 0.7809,
+      "step": 896
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.0302760463045414e-06,
+      "loss": 0.7135,
+      "step": 897
+    },
+    {
+      "epoch": 2.39,
+      "learning_rate": 2.021371326803206e-06,
+      "loss": 0.7674,
+      "step": 898
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.01246660730187e-06,
+      "loss": 0.7347,
+      "step": 899
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 2.0035618878005343e-06,
+      "loss": 0.8982,
+      "step": 900
+    },
+    {
+      "epoch": 2.4,
+      "learning_rate": 1.994657168299199e-06,
+      "loss": 0.4412,
+      "step": 901
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.985752448797863e-06,
+      "loss": 0.6954,
+      "step": 902
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9768477292965273e-06,
+      "loss": 0.8306,
+      "step": 903
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.9679430097951914e-06,
+      "loss": 0.9298,
+      "step": 904
+    },
+    {
+      "epoch": 2.41,
+      "learning_rate": 1.959038290293856e-06,
+      "loss": 0.3944,
+      "step": 905
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9501335707925202e-06,
+      "loss": 0.7185,
+      "step": 906
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.9412288512911844e-06,
+      "loss": 0.7635,
+      "step": 907
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.932324131789849e-06,
+      "loss": 0.8182,
+      "step": 908
+    },
+    {
+      "epoch": 2.42,
+      "learning_rate": 1.923419412288513e-06,
+      "loss": 0.801,
+      "step": 909
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9145146927871773e-06,
+      "loss": 0.8635,
+      "step": 910
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.9056099732858415e-06,
+      "loss": 0.8334,
+      "step": 911
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.896705253784506e-06,
+      "loss": 0.3652,
+      "step": 912
+    },
+    {
+      "epoch": 2.43,
+      "learning_rate": 1.88780053428317e-06,
+      "loss": 0.8984,
+      "step": 913
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8788958147818345e-06,
+      "loss": 0.8127,
+      "step": 914
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.8699910952804989e-06,
+      "loss": 0.8077,
+      "step": 915
+    },
+    {
+      "epoch": 2.44,
+      "learning_rate": 1.861086375779163e-06,
+      "loss": 0.7799,
+      "step": 916
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8521816562778274e-06,
+      "loss": 1.0054,
+      "step": 917
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8432769367764916e-06,
+      "loss": 0.8,
+      "step": 918
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.834372217275156e-06,
+      "loss": 0.6711,
+      "step": 919
+    },
+    {
+      "epoch": 2.45,
+      "learning_rate": 1.8254674977738202e-06,
+      "loss": 0.7397,
+      "step": 920
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.8165627782724845e-06,
+      "loss": 0.8191,
+      "step": 921
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.807658058771149e-06,
+      "loss": 0.8528,
+      "step": 922
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7987533392698131e-06,
+      "loss": 0.8358,
+      "step": 923
+    },
+    {
+      "epoch": 2.46,
+      "learning_rate": 1.7898486197684775e-06,
+      "loss": 0.9309,
+      "step": 924
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7809439002671417e-06,
+      "loss": 0.7102,
+      "step": 925
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.772039180765806e-06,
+      "loss": 0.4013,
+      "step": 926
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7631344612644702e-06,
+      "loss": 0.751,
+      "step": 927
+    },
+    {
+      "epoch": 2.47,
+      "learning_rate": 1.7542297417631346e-06,
+      "loss": 0.3049,
+      "step": 928
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.745325022261799e-06,
+      "loss": 0.7666,
+      "step": 929
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7364203027604632e-06,
+      "loss": 0.3536,
+      "step": 930
+    },
+    {
+      "epoch": 2.48,
+      "learning_rate": 1.7275155832591276e-06,
+      "loss": 0.8382,
+      "step": 931
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7186108637577917e-06,
+      "loss": 0.7132,
+      "step": 932
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7097061442564561e-06,
+      "loss": 0.9158,
+      "step": 933
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.7008014247551203e-06,
+      "loss": 0.8526,
+      "step": 934
+    },
+    {
+      "epoch": 2.49,
+      "learning_rate": 1.6918967052537847e-06,
+      "loss": 0.7391,
+      "step": 935
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.682991985752449e-06,
+      "loss": 0.8723,
+      "step": 936
+    },
+    {
+      "epoch": 2.5,
+      "eval_runtime": 2895.9409,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 936
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6740872662511133e-06,
+      "loss": 0.3292,
+      "step": 937
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6651825467497776e-06,
+      "loss": 0.6765,
+      "step": 938
+    },
+    {
+      "epoch": 2.5,
+      "learning_rate": 1.6562778272484418e-06,
+      "loss": 0.7178,
+      "step": 939
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6473731077471062e-06,
+      "loss": 0.7653,
+      "step": 940
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6384683882457702e-06,
+      "loss": 0.3523,
+      "step": 941
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6295636687444348e-06,
+      "loss": 0.8,
+      "step": 942
+    },
+    {
+      "epoch": 2.51,
+      "learning_rate": 1.6206589492430992e-06,
+      "loss": 0.6903,
+      "step": 943
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6117542297417631e-06,
+      "loss": 0.6835,
+      "step": 944
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.6028495102404277e-06,
+      "loss": 0.7969,
+      "step": 945
+    },
+    {
+      "epoch": 2.52,
+      "learning_rate": 1.5939447907390917e-06,
+      "loss": 0.7438,
+      "step": 946
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.585040071237756e-06,
+      "loss": 0.7823,
+      "step": 947
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5761353517364202e-06,
+      "loss": 0.4139,
+      "step": 948
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.5672306322350846e-06,
+      "loss": 0.7713,
+      "step": 949
+    },
+    {
+      "epoch": 2.53,
+      "learning_rate": 1.558325912733749e-06,
+      "loss": 0.8082,
+      "step": 950
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5494211932324132e-06,
+      "loss": 0.7197,
+      "step": 951
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5405164737310776e-06,
+      "loss": 0.369,
+      "step": 952
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5316117542297418e-06,
+      "loss": 0.9216,
+      "step": 953
+    },
+    {
+      "epoch": 2.54,
+      "learning_rate": 1.5227070347284061e-06,
+      "loss": 0.9584,
+      "step": 954
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5138023152270705e-06,
+      "loss": 0.86,
+      "step": 955
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.5048975957257347e-06,
+      "loss": 0.9044,
+      "step": 956
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.495992876224399e-06,
+      "loss": 0.862,
+      "step": 957
+    },
+    {
+      "epoch": 2.55,
+      "learning_rate": 1.4870881567230633e-06,
+      "loss": 0.9551,
+      "step": 958
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4781834372217276e-06,
+      "loss": 0.8435,
+      "step": 959
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4692787177203918e-06,
+      "loss": 0.6677,
+      "step": 960
+    },
+    {
+      "epoch": 2.56,
+      "learning_rate": 1.4603739982190562e-06,
+      "loss": 0.6774,
+      "step": 961
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4514692787177206e-06,
+      "loss": 0.5233,
+      "step": 962
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4425645592163848e-06,
+      "loss": 0.4674,
+      "step": 963
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4336598397150492e-06,
+      "loss": 0.3878,
+      "step": 964
+    },
+    {
+      "epoch": 2.57,
+      "learning_rate": 1.4247551202137133e-06,
+      "loss": 0.6754,
+      "step": 965
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.4158504007123777e-06,
+      "loss": 0.4179,
+      "step": 966
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.406945681211042e-06,
+      "loss": 0.7712,
+      "step": 967
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.3980409617097063e-06,
+      "loss": 0.767,
+      "step": 968
+    },
+    {
+      "epoch": 2.58,
+      "learning_rate": 1.3891362422083707e-06,
+      "loss": 0.3717,
+      "step": 969
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3802315227070348e-06,
+      "loss": 0.0,
+      "step": 970
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3713268032056992e-06,
+      "loss": 0.8067,
+      "step": 971
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3624220837043634e-06,
+      "loss": 0.8287,
+      "step": 972
+    },
+    {
+      "epoch": 2.59,
+      "learning_rate": 1.3535173642030278e-06,
+      "loss": 0.4449,
+      "step": 973
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.344612644701692e-06,
+      "loss": 0.754,
+      "step": 974
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.3357079252003564e-06,
+      "loss": 0.8403,
+      "step": 975
+    },
+    {
+      "epoch": 2.6,
+      "learning_rate": 1.3268032056990207e-06,
+      "loss": 0.7849,
+      "step": 976
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.317898486197685e-06,
+      "loss": 0.7059,
+      "step": 977
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.3089937666963493e-06,
+      "loss": 0.3265,
+      "step": 978
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.3000890471950135e-06,
+      "loss": 0.9245,
+      "step": 979
+    },
+    {
+      "epoch": 2.61,
+      "learning_rate": 1.2911843276936779e-06,
+      "loss": 0.9595,
+      "step": 980
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2822796081923418e-06,
+      "loss": 0.6899,
+      "step": 981
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2733748886910064e-06,
+      "loss": 0.7392,
+      "step": 982
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2644701691896708e-06,
+      "loss": 0.918,
+      "step": 983
+    },
+    {
+      "epoch": 2.62,
+      "learning_rate": 1.2555654496883348e-06,
+      "loss": 0.3556,
+      "step": 984
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2466607301869992e-06,
+      "loss": 0.8141,
+      "step": 985
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2377560106856636e-06,
+      "loss": 0.7781,
+      "step": 986
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2288512911843277e-06,
+      "loss": 0.4804,
+      "step": 987
+    },
+    {
+      "epoch": 2.63,
+      "learning_rate": 1.2199465716829921e-06,
+      "loss": 0.8916,
+      "step": 988
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2110418521816563e-06,
+      "loss": 0.8173,
+      "step": 989
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.2021371326803207e-06,
+      "loss": 0.8234,
+      "step": 990
+    },
+    {
+      "epoch": 2.64,
+      "learning_rate": 1.1932324131789849e-06,
+      "loss": 0.3678,
+      "step": 991
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1843276936776492e-06,
+      "loss": 0.7124,
+      "step": 992
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1754229741763136e-06,
+      "loss": 0.7981,
+      "step": 993
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1665182546749778e-06,
+      "loss": 0.9687,
+      "step": 994
+    },
+    {
+      "epoch": 2.65,
+      "learning_rate": 1.1576135351736422e-06,
+      "loss": 0.4893,
+      "step": 995
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1487088156723064e-06,
+      "loss": 0.3521,
+      "step": 996
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1398040961709708e-06,
+      "loss": 0.7366,
+      "step": 997
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.130899376669635e-06,
+      "loss": 0.8278,
+      "step": 998
+    },
+    {
+      "epoch": 2.66,
+      "learning_rate": 1.1219946571682993e-06,
+      "loss": 0.701,
+      "step": 999
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.1130899376669637e-06,
+      "loss": 0.8029,
+      "step": 1000
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.1041852181656279e-06,
+      "loss": 0.4728,
+      "step": 1001
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.0952804986642923e-06,
+      "loss": 0.8116,
+      "step": 1002
+    },
+    {
+      "epoch": 2.67,
+      "learning_rate": 1.0863757791629564e-06,
+      "loss": 0.4585,
+      "step": 1003
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0774710596616206e-06,
+      "loss": 0.7764,
+      "step": 1004
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.068566340160285e-06,
+      "loss": 0.8829,
+      "step": 1005
+    },
+    {
+      "epoch": 2.68,
+      "learning_rate": 1.0596616206589494e-06,
+      "loss": 0.8676,
+      "step": 1006
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0507569011576136e-06,
+      "loss": 0.8485,
+      "step": 1007
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.041852181656278e-06,
+      "loss": 0.7792,
+      "step": 1008
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0329474621549421e-06,
+      "loss": 0.816,
+      "step": 1009
+    },
+    {
+      "epoch": 2.69,
+      "learning_rate": 1.0240427426536065e-06,
+      "loss": 0.7202,
+      "step": 1010
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.0151380231522707e-06,
+      "loss": 0.6699,
+      "step": 1011
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 1.006233303650935e-06,
+      "loss": 0.7855,
+      "step": 1012
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.973285841495995e-07,
+      "loss": 0.3551,
+      "step": 1013
+    },
+    {
+      "epoch": 2.7,
+      "learning_rate": 9.884238646482636e-07,
+      "loss": 0.922,
+      "step": 1014
+    },
+    {
+      "epoch": 2.7,
+      "eval_runtime": 2894.8949,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 1014
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.79519145146928e-07,
+      "loss": 0.7158,
+      "step": 1015
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.706144256455922e-07,
+      "loss": 0.804,
+      "step": 1016
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.617097061442566e-07,
+      "loss": 0.8388,
+      "step": 1017
+    },
+    {
+      "epoch": 2.71,
+      "learning_rate": 9.528049866429208e-07,
+      "loss": 0.6531,
+      "step": 1018
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.43900267141585e-07,
+      "loss": 0.3711,
+      "step": 1019
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.349955476402494e-07,
+      "loss": 0.4492,
+      "step": 1020
+    },
+    {
+      "epoch": 2.72,
+      "learning_rate": 9.260908281389137e-07,
+      "loss": 0.7714,
+      "step": 1021
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.17186108637578e-07,
+      "loss": 0.7517,
+      "step": 1022
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 9.082813891362423e-07,
+      "loss": 0.7932,
+      "step": 1023
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.993766696349066e-07,
+      "loss": 0.8438,
+      "step": 1024
+    },
+    {
+      "epoch": 2.73,
+      "learning_rate": 8.904719501335708e-07,
+      "loss": 0.7475,
+      "step": 1025
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.815672306322351e-07,
+      "loss": 0.319,
+      "step": 1026
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.726625111308995e-07,
+      "loss": 0.8255,
+      "step": 1027
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.637577916295638e-07,
+      "loss": 0.829,
+      "step": 1028
+    },
+    {
+      "epoch": 2.74,
+      "learning_rate": 8.548530721282281e-07,
+      "loss": 0.3917,
+      "step": 1029
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.459483526268923e-07,
+      "loss": 0.7983,
+      "step": 1030
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.370436331255566e-07,
+      "loss": 0.7754,
+      "step": 1031
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.281389136242209e-07,
+      "loss": 0.8675,
+      "step": 1032
+    },
+    {
+      "epoch": 2.75,
+      "learning_rate": 8.192341941228851e-07,
+      "loss": 0.9558,
+      "step": 1033
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.103294746215496e-07,
+      "loss": 0.8757,
+      "step": 1034
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 8.014247551202139e-07,
+      "loss": 0.8634,
+      "step": 1035
+    },
+    {
+      "epoch": 2.76,
+      "learning_rate": 7.92520035618878e-07,
+      "loss": 0.8055,
+      "step": 1036
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.836153161175423e-07,
+      "loss": 0.7909,
+      "step": 1037
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.747105966162066e-07,
+      "loss": 0.8401,
+      "step": 1038
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.658058771148709e-07,
+      "loss": 0.7952,
+      "step": 1039
+    },
+    {
+      "epoch": 2.77,
+      "learning_rate": 7.569011576135353e-07,
+      "loss": 0.893,
+      "step": 1040
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.479964381121995e-07,
+      "loss": 0.7179,
+      "step": 1041
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.390917186108638e-07,
+      "loss": 0.8212,
+      "step": 1042
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.301869991095281e-07,
+      "loss": 0.8152,
+      "step": 1043
+    },
+    {
+      "epoch": 2.78,
+      "learning_rate": 7.212822796081924e-07,
+      "loss": 0.8137,
+      "step": 1044
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.123775601068567e-07,
+      "loss": 0.8554,
+      "step": 1045
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 7.03472840605521e-07,
+      "loss": 0.8571,
+      "step": 1046
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.945681211041853e-07,
+      "loss": 0.8911,
+      "step": 1047
+    },
+    {
+      "epoch": 2.79,
+      "learning_rate": 6.856634016028496e-07,
+      "loss": 0.8199,
+      "step": 1048
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.767586821015139e-07,
+      "loss": 0.9104,
+      "step": 1049
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.678539626001782e-07,
+      "loss": 0.6369,
+      "step": 1050
+    },
+    {
+      "epoch": 2.8,
+      "learning_rate": 6.589492430988425e-07,
+      "loss": 0.917,
+      "step": 1051
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.500445235975067e-07,
+      "loss": 0.8389,
+      "step": 1052
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.411398040961709e-07,
+      "loss": 0.9057,
+      "step": 1053
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.322350845948354e-07,
+      "loss": 0.7687,
+      "step": 1054
+    },
+    {
+      "epoch": 2.81,
+      "learning_rate": 6.233303650934996e-07,
+      "loss": 0.8434,
+      "step": 1055
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.144256455921639e-07,
+      "loss": 0.7658,
+      "step": 1056
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 6.055209260908281e-07,
+      "loss": 0.7009,
+      "step": 1057
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.966162065894924e-07,
+      "loss": 0.6782,
+      "step": 1058
+    },
+    {
+      "epoch": 2.82,
+      "learning_rate": 5.877114870881568e-07,
+      "loss": 0.7424,
+      "step": 1059
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.788067675868211e-07,
+      "loss": 0.9088,
+      "step": 1060
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.699020480854854e-07,
+      "loss": 0.7602,
+      "step": 1061
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.609973285841497e-07,
+      "loss": 0.9159,
+      "step": 1062
+    },
+    {
+      "epoch": 2.83,
+      "learning_rate": 5.520926090828139e-07,
+      "loss": 0.9262,
+      "step": 1063
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.431878895814782e-07,
+      "loss": 0.8386,
+      "step": 1064
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.342831700801425e-07,
+      "loss": 0.3611,
+      "step": 1065
+    },
+    {
+      "epoch": 2.84,
+      "learning_rate": 5.253784505788068e-07,
+      "loss": 0.878,
+      "step": 1066
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.164737310774711e-07,
+      "loss": 0.8796,
+      "step": 1067
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 5.075690115761353e-07,
+      "loss": 0.7574,
+      "step": 1068
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.986642920747997e-07,
+      "loss": 0.4731,
+      "step": 1069
+    },
+    {
+      "epoch": 2.85,
+      "learning_rate": 4.89759572573464e-07,
+      "loss": 0.3839,
+      "step": 1070
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.808548530721283e-07,
+      "loss": 0.7624,
+      "step": 1071
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.719501335707925e-07,
+      "loss": 0.7692,
+      "step": 1072
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.6304541406945685e-07,
+      "loss": 0.8048,
+      "step": 1073
+    },
+    {
+      "epoch": 2.86,
+      "learning_rate": 4.5414069456812114e-07,
+      "loss": 0.8512,
+      "step": 1074
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.452359750667854e-07,
+      "loss": 0.3529,
+      "step": 1075
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.3633125556544975e-07,
+      "loss": 0.8333,
+      "step": 1076
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.2742653606411403e-07,
+      "loss": 0.8391,
+      "step": 1077
+    },
+    {
+      "epoch": 2.87,
+      "learning_rate": 4.185218165627783e-07,
+      "loss": 0.7346,
+      "step": 1078
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.0961709706144254e-07,
+      "loss": 0.4418,
+      "step": 1079
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 4.0071237756010693e-07,
+      "loss": 0.738,
+      "step": 1080
+    },
+    {
+      "epoch": 2.88,
+      "learning_rate": 3.9180765805877116e-07,
+      "loss": 0.8682,
+      "step": 1081
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.8290293855743544e-07,
+      "loss": 0.7277,
+      "step": 1082
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.7399821905609977e-07,
+      "loss": 0.8297,
+      "step": 1083
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.6509349955476405e-07,
+      "loss": 0.7857,
+      "step": 1084
+    },
+    {
+      "epoch": 2.89,
+      "learning_rate": 3.5618878005342833e-07,
+      "loss": 0.8284,
+      "step": 1085
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.4728406055209267e-07,
+      "loss": 0.9317,
+      "step": 1086
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.3837934105075695e-07,
+      "loss": 0.6591,
+      "step": 1087
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2947462154942123e-07,
+      "loss": 0.8111,
+      "step": 1088
+    },
+    {
+      "epoch": 2.9,
+      "learning_rate": 3.2056990204808546e-07,
+      "loss": 0.4733,
+      "step": 1089
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.116651825467498e-07,
+      "loss": 0.835,
+      "step": 1090
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 3.0276046304541407e-07,
+      "loss": 0.9588,
+      "step": 1091
+    },
+    {
+      "epoch": 2.91,
+      "learning_rate": 2.938557435440784e-07,
+      "loss": 0.9509,
+      "step": 1092
+    },
+    {
+      "epoch": 2.91,
+      "eval_runtime": 2950.1226,
+      "eval_samples_per_second": 0.0,
+      "eval_steps_per_second": 0.0,
+      "fake_metric": 0.0,
+      "step": 1092
+    }
+  ],
+  "logging_steps": 1,
+  "max_steps": 1125,
+  "num_train_epochs": 3,
+  "save_steps": 78,
+  "total_flos": 9.237618268766208e+17,
+  "trial_name": null,
+  "trial_params": null
+}