Upload Dril-Instruct LoRA

adapter_config.json

@@ -0,0 +1,17 @@
+{
+  "base_model_name_or_path": "/workspace/vicuna-13b-cocktail",
+  "bias": "none",
+  "fan_in_fan_out": false,
+  "inference_mode": true,
+  "init_lora_weights": true,
+  "lora_alpha": 16,
+  "lora_dropout": 0.05,
+  "modules_to_save": null,
+  "peft_type": "LORA",
+  "r": 8,
+  "target_modules": [
+    "q_proj",
+    "v_proj"
+  ],
+  "task_type": "CAUSAL_LM"
+}

adapter_model.bin

@@ -0,0 +1,3 @@
+version https://git-lfs.github.com/spec/v1
+oid sha256:d732699342e0884e7f33e3d52aa74881c0899b2eec9430350f758577d42bb060
+size 13162317

trainer_state.json

@@ -0,0 +1,4453 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 1.0,
+  "global_step": 738,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 8.695652173913044e-07,
+      "loss": 4.0786,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.7391304347826088e-06,
+      "loss": 4.3694,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.6086956521739132e-06,
+      "loss": 4.4697,
+      "step": 3
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 3.4782608695652175e-06,
+      "loss": 4.5119,
+      "step": 4
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.347826086956522e-06,
+      "loss": 3.981,
+      "step": 5
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.2173913043478265e-06,
+      "loss": 4.5717,
+      "step": 6
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.086956521739132e-06,
+      "loss": 3.9509,
+      "step": 7
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.956521739130435e-06,
+      "loss": 3.7814,
+      "step": 8
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.82608695652174e-06,
+      "loss": 4.9508,
+      "step": 9
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.695652173913044e-06,
+      "loss": 4.4459,
+      "step": 10
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.565217391304349e-06,
+      "loss": 4.0627,
+      "step": 11
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.0434782608695653e-05,
+      "loss": 3.7246,
+      "step": 12
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.1304347826086957e-05,
+      "loss": 4.578,
+      "step": 13
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.2173913043478263e-05,
+      "loss": 3.7387,
+      "step": 14
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.3043478260869566e-05,
+      "loss": 4.6955,
+      "step": 15
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.391304347826087e-05,
+      "loss": 4.4181,
+      "step": 16
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.4782608695652174e-05,
+      "loss": 4.0946,
+      "step": 17
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 1.565217391304348e-05,
+      "loss": 4.8493,
+      "step": 18
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.6521739130434785e-05,
+      "loss": 4.7478,
+      "step": 19
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.739130434782609e-05,
+      "loss": 4.9121,
+      "step": 20
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.8260869565217393e-05,
+      "loss": 4.0576,
+      "step": 21
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9130434782608697e-05,
+      "loss": 4.0833,
+      "step": 22
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 2e-05,
+      "loss": 3.7284,
+      "step": 23
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999903471186634e-05,
+      "loss": 4.0887,
+      "step": 24
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 1.9999613886610097e-05,
+      "loss": 3.9048,
+      "step": 25
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9999131251861037e-05,
+      "loss": 3.8901,
+      "step": 26
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999845557625709e-05,
+      "loss": 3.9268,
+      "step": 27
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9997586872842683e-05,
+      "loss": 5.0328,
+      "step": 28
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9996525158388804e-05,
+      "loss": 4.6897,
+      "step": 29
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999527045339266e-05,
+      "loss": 4.2548,
+      "step": 30
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.9993822782077282e-05,
+      "loss": 4.0835,
+      "step": 31
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999218217239108e-05,
+      "loss": 3.5524,
+      "step": 32
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 1.999034865600726e-05,
+      "loss": 4.23,
+      "step": 33
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.998832226832327e-05,
+      "loss": 4.3294,
+      "step": 34
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9986103048460056e-05,
+      "loss": 4.1848,
+      "step": 35
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9983691039261358e-05,
+      "loss": 4.2649,
+      "step": 36
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9981086287292853e-05,
+      "loss": 4.2958,
+      "step": 37
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.997828884284126e-05,
+      "loss": 4.2612,
+      "step": 38
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9975298759913382e-05,
+      "loss": 4.5491,
+      "step": 39
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 1.9972116096235047e-05,
+      "loss": 4.3637,
+      "step": 40
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.996874091325001e-05,
+      "loss": 4.363,
+      "step": 41
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9965173276118747e-05,
+      "loss": 3.4844,
+      "step": 42
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9961413253717214e-05,
+      "loss": 4.8517,
+      "step": 43
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9957460918635513e-05,
+      "loss": 4.6956,
+      "step": 44
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.995331634717649e-05,
+      "loss": 4.6778,
+      "step": 45
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.9948979619354253e-05,
+      "loss": 3.6159,
+      "step": 46
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 1.994445081889264e-05,
+      "loss": 4.1196,
+      "step": 47
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99397300332236e-05,
+      "loss": 4.3408,
+      "step": 48
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.99348173534855e-05,
+      "loss": 4.0532,
+      "step": 49
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9929712874521375e-05,
+      "loss": 4.1374,
+      "step": 50
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9924416694877077e-05,
+      "loss": 3.8111,
+      "step": 51
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9918928916799395e-05,
+      "loss": 4.6571,
+      "step": 52
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9913249646234072e-05,
+      "loss": 4.2369,
+      "step": 53
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.9907378992823755e-05,
+      "loss": 4.2798,
+      "step": 54
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 1.990131706990589e-05,
+      "loss": 4.7844,
+      "step": 55
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9895063994510512e-05,
+      "loss": 4.7413,
+      "step": 56
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9888619887358024e-05,
+      "loss": 3.906,
+      "step": 57
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.988198487285682e-05,
+      "loss": 4.5495,
+      "step": 58
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9875159079100917e-05,
+      "loss": 3.8774,
+      "step": 59
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9868142637867474e-05,
+      "loss": 4.8845,
+      "step": 60
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9860935684614235e-05,
+      "loss": 5.0222,
+      "step": 61
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 1.9853538358476933e-05,
+      "loss": 4.2982,
+      "step": 62
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9845950802266584e-05,
+      "loss": 4.0971,
+      "step": 63
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.983817316246676e-05,
+      "loss": 3.5684,
+      "step": 64
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9830205589230713e-05,
+      "loss": 3.8654,
+      "step": 65
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9822048236378536e-05,
+      "loss": 3.3175,
+      "step": 66
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9813701261394136e-05,
+      "loss": 3.857,
+      "step": 67
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.980516482542224e-05,
+      "loss": 4.0755,
+      "step": 68
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9796439093265245e-05,
+      "loss": 4.1953,
+      "step": 69
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 1.9787524233380076e-05,
+      "loss": 3.7893,
+      "step": 70
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9778420417874894e-05,
+      "loss": 4.7063,
+      "step": 71
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9769127822505805e-05,
+      "loss": 3.9717,
+      "step": 72
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9759646626673445e-05,
+      "loss": 3.7575,
+      "step": 73
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9749977013419536e-05,
+      "loss": 3.5247,
+      "step": 74
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9740119169423337e-05,
+      "loss": 4.0718,
+      "step": 75
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.973007328499804e-05,
+      "loss": 4.7772,
+      "step": 76
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 1.9719839554087108e-05,
+      "loss": 3.8932,
+      "step": 77
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9709418174260523e-05,
+      "loss": 3.7054,
+      "step": 78
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9698809346710965e-05,
+      "loss": 4.3931,
+      "step": 79
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9688013276249947e-05,
+      "loss": 4.2247,
+      "step": 80
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9677030171303842e-05,
+      "loss": 4.0688,
+      "step": 81
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.966586024390986e-05,
+      "loss": 4.1457,
+      "step": 82
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.9654503709711984e-05,
+      "loss": 3.7592,
+      "step": 83
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 1.964296078795675e-05,
+      "loss": 3.8759,
+      "step": 84
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9631231701489083e-05,
+      "loss": 3.349,
+      "step": 85
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9619316676747928e-05,
+      "loss": 4.5228,
+      "step": 86
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9607215943761933e-05,
+      "loss": 4.0959,
+      "step": 87
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9594929736144978e-05,
+      "loss": 3.8604,
+      "step": 88
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9582458291091664e-05,
+      "loss": 4.0226,
+      "step": 89
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9569801849372757e-05,
+      "loss": 3.4955,
+      "step": 90
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.9556960655330512e-05,
+      "loss": 3.3423,
+      "step": 91
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 1.954393495687398e-05,
+      "loss": 3.3918,
+      "step": 92
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9530725005474195e-05,
+      "loss": 4.1001,
+      "step": 93
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9517331056159353e-05,
+      "loss": 3.6566,
+      "step": 94
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9503753367509855e-05,
+      "loss": 3.2956,
+      "step": 95
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9489992201653337e-05,
+      "loss": 3.6158,
+      "step": 96
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.9476047824259602e-05,
+      "loss": 3.6402,
+      "step": 97
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.946192050453549e-05,
+      "loss": 4.1444,
+      "step": 98
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 1.944761051521968e-05,
+      "loss": 3.8668,
+      "step": 99
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9433118132577432e-05,
+      "loss": 3.8939,
+      "step": 100
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.941844363639525e-05,
+      "loss": 3.9215,
+      "step": 101
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9403587309975467e-05,
+      "loss": 4.1247,
+      "step": 102
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.93885494401308e-05,
+      "loss": 3.3528,
+      "step": 103
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9373330317178797e-05,
+      "loss": 3.9051,
+      "step": 104
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9357930234936228e-05,
+      "loss": 4.2422,
+      "step": 105
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.9342349490713427e-05,
+      "loss": 4.1208,
+      "step": 106
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 1.932658838530855e-05,
+      "loss": 3.6984,
+      "step": 107
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9310647223001752e-05,
+      "loss": 3.7027,
+      "step": 108
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.929452631154933e-05,
+      "loss": 4.2205,
+      "step": 109
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9278225962177776e-05,
+      "loss": 3.9275,
+      "step": 110
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9261746489577767e-05,
+      "loss": 4.4105,
+      "step": 111
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9245088211898086e-05,
+      "loss": 4.6844,
+      "step": 112
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.9228251450739495e-05,
+      "loss": 3.4644,
+      "step": 113
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 1.92112365311485e-05,
+      "loss": 4.2983,
+      "step": 114
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.919404378161111e-05,
+      "loss": 4.2432,
+      "step": 115
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9176673534046465e-05,
+      "loss": 2.5393,
+      "step": 116
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9159126123800437e-05,
+      "loss": 3.6555,
+      "step": 117
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9141401889639167e-05,
+      "loss": 3.7163,
+      "step": 118
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.9123501173742514e-05,
+      "loss": 3.0856,
+      "step": 119
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.910542432169745e-05,
+      "loss": 3.9067,
+      "step": 120
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 1.908717168249139e-05,
+      "loss": 2.8562,
+      "step": 121
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9068743608505454e-05,
+      "loss": 3.5617,
+      "step": 122
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.905014045550767e-05,
+      "loss": 2.6561,
+      "step": 123
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.903136258264609e-05,
+      "loss": 4.0248,
+      "step": 124
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.9012410352441866e-05,
+      "loss": 3.5109,
+      "step": 125
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.899328413078227e-05,
+      "loss": 3.968,
+      "step": 126
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8973984286913584e-05,
+      "loss": 3.7493,
+      "step": 127
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8954511193434024e-05,
+      "loss": 3.9637,
+      "step": 128
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 1.8934865226286507e-05,
+      "loss": 4.1919,
+      "step": 129
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.891504676475141e-05,
+      "loss": 3.5211,
+      "step": 130
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8895056191439252e-05,
+      "loss": 3.5958,
+      "step": 131
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8874893892283296e-05,
+      "loss": 3.1333,
+      "step": 132
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8854560256532098e-05,
+      "loss": 3.1487,
+      "step": 133
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8834055676742018e-05,
+      "loss": 4.7189,
+      "step": 134
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8813380548769594e-05,
+      "loss": 3.6627,
+      "step": 135
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 1.8792535271763944e-05,
+      "loss": 3.5609,
+      "step": 136
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8771520248159044e-05,
+      "loss": 3.9891,
+      "step": 137
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8750335883665948e-05,
+      "loss": 4.0442,
+      "step": 138
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8728982587264965e-05,
+      "loss": 3.9044,
+      "step": 139
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8707460771197773e-05,
+      "loss": 3.9895,
+      "step": 140
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8685770850959444e-05,
+      "loss": 3.5426,
+      "step": 141
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8663913245290433e-05,
+      "loss": 3.7832,
+      "step": 142
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 1.8641888376168483e-05,
+      "loss": 3.1022,
+      "step": 143
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8619696668800494e-05,
+      "loss": 3.3351,
+      "step": 144
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.85973385516143e-05,
+      "loss": 3.3754,
+      "step": 145
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8574814456250406e-05,
+      "loss": 3.1684,
+      "step": 146
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.855212481755365e-05,
+      "loss": 3.731,
+      "step": 147
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.852927007356481e-05,
+      "loss": 2.823,
+      "step": 148
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8506250665512156e-05,
+      "loss": 3.1365,
+      "step": 149
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.848306703780291e-05,
+      "loss": 3.4664,
+      "step": 150
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 1.8459719638014693e-05,
+      "loss": 3.1022,
+      "step": 151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.843620891688686e-05,
+      "loss": 3.3208,
+      "step": 152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8412535328311813e-05,
+      "loss": 3.3757,
+      "step": 153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8388699329326237e-05,
+      "loss": 3.7115,
+      "step": 154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8364701380102267e-05,
+      "loss": 3.7887,
+      "step": 155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8340541943938623e-05,
+      "loss": 3.075,
+      "step": 156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.831622148725164e-05,
+      "loss": 3.6212,
+      "step": 157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 1.8291740479566286e-05,
+      "loss": 3.5491,
+      "step": 158
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8267099393507083e-05,
+      "loss": 4.183,
+      "step": 159
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8242298704788988e-05,
+      "loss": 3.432,
+      "step": 160
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.821733889220822e-05,
+      "loss": 3.6803,
+      "step": 161
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.819222043763299e-05,
+      "loss": 3.1544,
+      "step": 162
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.816694382599422e-05,
+      "loss": 3.0101,
+      "step": 163
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.814150954527618e-05,
+      "loss": 3.3107,
+      "step": 164
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8115918086507054e-05,
+      "loss": 3.275,
+      "step": 165
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 1.8090169943749477e-05,
+      "loss": 3.8981,
+      "step": 166
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8064265614090976e-05,
+      "loss": 3.0138,
+      "step": 167
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.8038205597634392e-05,
+      "loss": 3.428,
+      "step": 168
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.801199039748822e-05,
+      "loss": 3.2076,
+      "step": 169
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7985620519756897e-05,
+      "loss": 3.4325,
+      "step": 170
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7959096473531016e-05,
+      "loss": 4.1224,
+      "step": 171
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7932418770877523e-05,
+      "loss": 3.4242,
+      "step": 172
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 1.7905587926829815e-05,
+      "loss": 3.2022,
+      "step": 173
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7878604459377795e-05,
+      "loss": 2.7946,
+      "step": 174
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7851468889457883e-05,
+      "loss": 3.1358,
+      "step": 175
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7824181740942958e-05,
+      "loss": 3.0715,
+      "step": 176
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7796743540632226e-05,
+      "loss": 3.7963,
+      "step": 177
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.776915481824107e-05,
+      "loss": 3.3344,
+      "step": 178
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7741416106390828e-05,
+      "loss": 4.0142,
+      "step": 179
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 1.7713527940598473e-05,
+      "loss": 3.6756,
+      "step": 180
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7685490859266324e-05,
+      "loss": 3.426,
+      "step": 181
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7657305403671618e-05,
+      "loss": 2.8344,
+      "step": 182
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.762897211795607e-05,
+      "loss": 3.6437,
+      "step": 183
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.760049154911537e-05,
+      "loss": 3.3443,
+      "step": 184
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7571864246988623e-05,
+      "loss": 3.3062,
+      "step": 185
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7543090764247726e-05,
+      "loss": 3.6169,
+      "step": 186
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.751417165638671e-05,
+      "loss": 3.2821,
+      "step": 187
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 1.7485107481711014e-05,
+      "loss": 3.2372,
+      "step": 188
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7455898801326685e-05,
+      "loss": 3.3073,
+      "step": 189
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.742654617912958e-05,
+      "loss": 3.5097,
+      "step": 190
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7397050181794463e-05,
+      "loss": 3.1931,
+      "step": 191
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.736741137876405e-05,
+      "loss": 3.6169,
+      "step": 192
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.733763034223804e-05,
+      "loss": 3.6615,
+      "step": 193
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.730770764716206e-05,
+      "loss": 3.6591,
+      "step": 194
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 1.7277643871216558e-05,
+      "loss": 3.5028,
+      "step": 195
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.724743959480565e-05,
+      "loss": 3.8692,
+      "step": 196
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.721709540104594e-05,
+      "loss": 2.9761,
+      "step": 197
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7186611875755227e-05,
+      "loss": 3.679,
+      "step": 198
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.715598960744121e-05,
+      "loss": 3.4565,
+      "step": 199
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.712522918729014e-05,
+      "loss": 2.9959,
+      "step": 200
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7094331209155394e-05,
+      "loss": 3.0216,
+      "step": 201
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 1.7063296269545988e-05,
+      "loss": 3.5365,
+      "step": 202
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7032124967615112e-05,
+      "loss": 3.1137,
+      "step": 203
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.7000817905148523e-05,
+      "loss": 3.1813,
+      "step": 204
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.696937568655294e-05,
+      "loss": 3.0031,
+      "step": 205
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6937798918844363e-05,
+      "loss": 3.2111,
+      "step": 206
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6906088211636387e-05,
+      "loss": 3.6658,
+      "step": 207
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6874244177128395e-05,
+      "loss": 3.5334,
+      "step": 208
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6842267430093762e-05,
+      "loss": 3.7355,
+      "step": 209
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 1.6810158587867973e-05,
+      "loss": 2.9213,
+      "step": 210
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6777918270336718e-05,
+      "loss": 3.631,
+      "step": 211
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6745547099923917e-05,
+      "loss": 4.1871,
+      "step": 212
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6713045701579705e-05,
+      "loss": 3.7414,
+      "step": 213
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.6680414702768358e-05,
+      "loss": 2.785,
+      "step": 214
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.66476547334562e-05,
+      "loss": 3.4182,
+      "step": 215
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.661476642609943e-05,
+      "loss": 3.2041,
+      "step": 216
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 1.658175041563189e-05,
+      "loss": 4.2476,
+      "step": 217
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6548607339452853e-05,
+      "loss": 2.9944,
+      "step": 218
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6515337837414677e-05,
+      "loss": 3.4829,
+      "step": 219
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6481942551810476e-05,
+      "loss": 3.174,
+      "step": 220
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6448422127361707e-05,
+      "loss": 3.6737,
+      "step": 221
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.641477721120573e-05,
+      "loss": 3.3298,
+      "step": 222
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.638100845288331e-05,
+      "loss": 2.6232,
+      "step": 223
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.6347116504326082e-05,
+      "loss": 3.5669,
+      "step": 224
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 1.631310201984396e-05,
+      "loss": 3.6168,
+      "step": 225
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.627896565611251e-05,
+      "loss": 3.1156,
+      "step": 226
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6244708072160267e-05,
+      "loss": 3.6188,
+      "step": 227
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6210329929356017e-05,
+      "loss": 3.7165,
+      "step": 228
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6175831891396034e-05,
+      "loss": 2.9527,
+      "step": 229
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.614121462429124e-05,
+      "loss": 3.1555,
+      "step": 230
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6106478796354382e-05,
+      "loss": 3.2923,
+      "step": 231
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 1.6071625078187113e-05,
+      "loss": 3.0248,
+      "step": 232
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.6036654142667043e-05,
+      "loss": 3.3092,
+      "step": 233
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.600156666493475e-05,
+      "loss": 3.3198,
+      "step": 234
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.596636332238076e-05,
+      "loss": 3.1906,
+      "step": 235
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.593104479463244e-05,
+      "loss": 3.1719,
+      "step": 236
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5895611763540914e-05,
+      "loss": 3.9507,
+      "step": 237
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5860064913167863e-05,
+      "loss": 3.1961,
+      "step": 238
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 1.5824404929772347e-05,
+      "loss": 4.026,
+      "step": 239
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5788632501797545e-05,
+      "loss": 3.6545,
+      "step": 240
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.575274831985746e-05,
+      "loss": 3.2192,
+      "step": 241
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5716753076723594e-05,
+      "loss": 3.2661,
+      "step": 242
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.568064746731156e-05,
+      "loss": 3.4022,
+      "step": 243
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5644432188667695e-05,
+      "loss": 3.7246,
+      "step": 244
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5608107939955565e-05,
+      "loss": 2.8461,
+      "step": 245
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5571675422442504e-05,
+      "loss": 3.579,
+      "step": 246
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 1.5535135339486044e-05,
+      "loss": 3.3706,
+      "step": 247
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.549848839652035e-05,
+      "loss": 3.1511,
+      "step": 248
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5461735301042615e-05,
+      "loss": 2.9157,
+      "step": 249
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.542487676259937e-05,
+      "loss": 2.8705,
+      "step": 250
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5387913492772816e-05,
+      "loss": 3.3106,
+      "step": 251
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5350846205167065e-05,
+      "loss": 3.4992,
+      "step": 252
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.5313675615394373e-05,
+      "loss": 3.6425,
+      "step": 253
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 1.527640244106133e-05,
+      "loss": 3.4484,
+      "step": 254
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5239027401754987e-05,
+      "loss": 2.9688,
+      "step": 255
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5201551219028988e-05,
+      "loss": 3.0834,
+      "step": 256
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5163974616389621e-05,
+      "loss": 3.4041,
+      "step": 257
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5126298319281859e-05,
+      "loss": 3.1486,
+      "step": 258
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.508852305507535e-05,
+      "loss": 3.495,
+      "step": 259
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5050649553050383e-05,
+      "loss": 3.3416,
+      "step": 260
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 1.5012678544383797e-05,
+      "loss": 3.2129,
+      "step": 261
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4974610762134875e-05,
+      "loss": 2.4619,
+      "step": 262
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4936446941231186e-05,
+      "loss": 3.2066,
+      "step": 263
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4898187818454401e-05,
+      "loss": 3.5822,
+      "step": 264
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.485983413242606e-05,
+      "loss": 2.577,
+      "step": 265
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4821386623593332e-05,
+      "loss": 3.5614,
+      "step": 266
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4782846034214689e-05,
+      "loss": 3.3427,
+      "step": 267
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4744213108345605e-05,
+      "loss": 2.935,
+      "step": 268
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 1.4705488591824182e-05,
+      "loss": 2.9306,
+      "step": 269
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4666673232256738e-05,
+      "loss": 2.5384,
+      "step": 270
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4627767779003402e-05,
+      "loss": 3.3218,
+      "step": 271
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4588772983163612e-05,
+      "loss": 3.1161,
+      "step": 272
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4549689597561652e-05,
+      "loss": 3.5516,
+      "step": 273
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4510518376732081e-05,
+      "loss": 3.4072,
+      "step": 274
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4471260076905198e-05,
+      "loss": 3.3651,
+      "step": 275
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 1.4431915455992416e-05,
+      "loss": 3.5782,
+      "step": 276
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4392485273571652e-05,
+      "loss": 3.2503,
+      "step": 277
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.435297029087265e-05,
+      "loss": 2.7281,
+      "step": 278
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.431337127076229e-05,
+      "loss": 2.694,
+      "step": 279
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4273688977729852e-05,
+      "loss": 4.0722,
+      "step": 280
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4233924177872269e-05,
+      "loss": 3.4334,
+      "step": 281
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4194077638879333e-05,
+      "loss": 3.1692,
+      "step": 282
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4154150130018867e-05,
+      "loss": 3.1848,
+      "step": 283
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 1.4114142422121879e-05,
+      "loss": 3.5117,
+      "step": 284
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4074055287567685e-05,
+      "loss": 3.4279,
+      "step": 285
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.4033889500268991e-05,
+      "loss": 3.3445,
+      "step": 286
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3993645835656955e-05,
+      "loss": 2.9628,
+      "step": 287
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3953325070666215e-05,
+      "loss": 3.2026,
+      "step": 288
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3912927983719888e-05,
+      "loss": 3.4085,
+      "step": 289
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3872455354714552e-05,
+      "loss": 3.4118,
+      "step": 290
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 1.3831907965005173e-05,
+      "loss": 3.8913,
+      "step": 291
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3791286597390035e-05,
+      "loss": 3.8467,
+      "step": 292
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.375059203609562e-05,
+      "loss": 3.1046,
+      "step": 293
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.370982506676147e-05,
+      "loss": 3.6762,
+      "step": 294
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3668986476425024e-05,
+      "loss": 3.5088,
+      "step": 295
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.362807705350641e-05,
+      "loss": 3.3849,
+      "step": 296
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3587097587793243e-05,
+      "loss": 3.01,
+      "step": 297
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 1.3546048870425356e-05,
+      "loss": 2.9031,
+      "step": 298
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3504931693879553e-05,
+      "loss": 3.8562,
+      "step": 299
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3463746851954275e-05,
+      "loss": 3.1888,
+      "step": 300
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.342249513975432e-05,
+      "loss": 2.8173,
+      "step": 301
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3381177353675441e-05,
+      "loss": 3.5295,
+      "step": 302
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3339794291389015e-05,
+      "loss": 2.9573,
+      "step": 303
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3298346751826624e-05,
+      "loss": 3.1344,
+      "step": 304
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3256835535164622e-05,
+      "loss": 3.4805,
+      "step": 305
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 1.3215261442808718e-05,
+      "loss": 3.3684,
+      "step": 306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3173625277378464e-05,
+      "loss": 3.3554,
+      "step": 307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3131927842691793e-05,
+      "loss": 2.7138,
+      "step": 308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3090169943749475e-05,
+      "loss": 3.09,
+      "step": 309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3048352386719605e-05,
+      "loss": 3.4217,
+      "step": 310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.3006475978922013e-05,
+      "loss": 3.1616,
+      "step": 311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2964541528812689e-05,
+      "loss": 3.1437,
+      "step": 312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 1.2922549845968174e-05,
+      "loss": 2.6185,
+      "step": 313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2880501741069931e-05,
+      "loss": 3.2304,
+      "step": 314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2838398025888695e-05,
+      "loss": 3.7043,
+      "step": 315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2796239513268796e-05,
+      "loss": 3.3408,
+      "step": 316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.275402701711248e-05,
+      "loss": 3.4695,
+      "step": 317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2711761352364172e-05,
+      "loss": 2.7639,
+      "step": 318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2669443334994768e-05,
+      "loss": 3.1568,
+      "step": 319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.262707378198587e-05,
+      "loss": 3.1008,
+      "step": 320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 1.2584653511314012e-05,
+      "loss": 2.7824,
+      "step": 321
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2542183341934873e-05,
+      "loss": 3.6959,
+      "step": 322
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2499664093767458e-05,
+      "loss": 3.1506,
+      "step": 323
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.245709658767829e-05,
+      "loss": 3.3476,
+      "step": 324
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.241448164546553e-05,
+      "loss": 3.5336,
+      "step": 325
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2371820089843145e-05,
+      "loss": 3.0568,
+      "step": 326
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2329112744425e-05,
+      "loss": 3.5106,
+      "step": 327
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 1.2286360433708976e-05,
+      "loss": 3.5715,
+      "step": 328
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2243563983061029e-05,
+      "loss": 3.9286,
+      "step": 329
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2200724218699284e-05,
+      "loss": 3.0521,
+      "step": 330
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2157841967678064e-05,
+      "loss": 3.1919,
+      "step": 331
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2114918057871928e-05,
+      "loss": 3.7024,
+      "step": 332
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.2071953317959692e-05,
+      "loss": 3.2707,
+      "step": 333
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202894857740843e-05,
+      "loss": 3.0606,
+      "step": 334
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 1.202894857740843e-05,
+      "loss": 3.387,
+      "step": 335
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1985904666457455e-05,
+      "loss": 3.3803,
+      "step": 336
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.19428224161023e-05,
+      "loss": 3.5345,
+      "step": 337
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1899702658078663e-05,
+      "loss": 3.0052,
+      "step": 338
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1856546224846354e-05,
+      "loss": 3.2168,
+      "step": 339
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.181335394957324e-05,
+      "loss": 3.1219,
+      "step": 340
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1770126666119133e-05,
+      "loss": 3.1971,
+      "step": 341
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1726865209019709e-05,
+      "loss": 3.327,
+      "step": 342
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 1.1683570413470384e-05,
+      "loss": 2.6641,
+      "step": 343
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1640243115310219e-05,
+      "loss": 3.785,
+      "step": 344
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1596884151005743e-05,
+      "loss": 3.7546,
+      "step": 345
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.155349435763483e-05,
+      "loss": 3.3265,
+      "step": 346
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1510074572870533e-05,
+      "loss": 2.7378,
+      "step": 347
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1466625634964911e-05,
+      "loss": 3.5108,
+      "step": 348
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1423148382732854e-05,
+      "loss": 3.4042,
+      "step": 349
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 1.1379643655535869e-05,
+      "loss": 3.1124,
+      "step": 350
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1336112293265896e-05,
+      "loss": 3.2475,
+      "step": 351
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1292555136329082e-05,
+      "loss": 3.8716,
+      "step": 352
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1248973025629567e-05,
+      "loss": 3.2482,
+      "step": 353
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1205366802553231e-05,
+      "loss": 2.9432,
+      "step": 354
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1161737308951473e-05,
+      "loss": 2.6907,
+      "step": 355
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.111808538712494e-05,
+      "loss": 3.4467,
+      "step": 356
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 1.1074411879807271e-05,
+      "loss": 3.504,
+      "step": 357
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.1030717630148839e-05,
+      "loss": 2.9749,
+      "step": 358
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0987003481700456e-05,
+      "loss": 3.735,
+      "step": 359
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0943270278397097e-05,
+      "loss": 2.9007,
+      "step": 360
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0899518864541607e-05,
+      "loss": 3.5406,
+      "step": 361
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.08557500847884e-05,
+      "loss": 3.1011,
+      "step": 362
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0811964784127145e-05,
+      "loss": 3.3565,
+      "step": 363
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.076816380786647e-05,
+      "loss": 3.1954,
+      "step": 364
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 1.0724348001617626e-05,
+      "loss": 3.4458,
+      "step": 365
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0680518211278169e-05,
+      "loss": 3.1999,
+      "step": 366
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.063667528301563e-05,
+      "loss": 3.1261,
+      "step": 367
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0592820063251177e-05,
+      "loss": 3.1734,
+      "step": 368
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0548953398643276e-05,
+      "loss": 3.8427,
+      "step": 369
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0505076136071342e-05,
+      "loss": 3.2633,
+      "step": 370
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0461189122619394e-05,
+      "loss": 3.4789,
+      "step": 371
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 1.0417293205559694e-05,
+      "loss": 3.3846,
+      "step": 372
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0373389232336404e-05,
+      "loss": 3.1423,
+      "step": 373
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0329478050549208e-05,
+      "loss": 2.7868,
+      "step": 374
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0285560507936962e-05,
+      "loss": 3.5709,
+      "step": 375
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0241637452361323e-05,
+      "loss": 2.9785,
+      "step": 376
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0197709731790375e-05,
+      "loss": 3.6323,
+      "step": 377
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0153778194282269e-05,
+      "loss": 3.3213,
+      "step": 378
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0109843687968837e-05,
+      "loss": 3.0757,
+      "step": 379
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 1.0065907061039234e-05,
+      "loss": 3.288,
+      "step": 380
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 1.0021969161723555e-05,
+      "loss": 3.1818,
+      "step": 381
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.97803083827645e-06,
+      "loss": 3.6509,
+      "step": 382
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.934092938960769e-06,
+      "loss": 2.9391,
+      "step": 383
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.890156312031165e-06,
+      "loss": 3.9642,
+      "step": 384
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.846221805717734e-06,
+      "loss": 3.3998,
+      "step": 385
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.802290268209627e-06,
+      "loss": 3.7466,
+      "step": 386
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 9.75836254763868e-06,
+      "loss": 3.7435,
+      "step": 387
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.71443949206304e-06,
+      "loss": 3.3039,
+      "step": 388
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.670521949450793e-06,
+      "loss": 3.0855,
+      "step": 389
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.6266107676636e-06,
+      "loss": 3.0572,
+      "step": 390
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.58270679444031e-06,
+      "loss": 2.9945,
+      "step": 391
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.538810877380611e-06,
+      "loss": 3.2003,
+      "step": 392
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.49492386392866e-06,
+      "loss": 2.9528,
+      "step": 393
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 9.451046601356725e-06,
+      "loss": 2.9552,
+      "step": 394
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.407179936748827e-06,
+      "loss": 3.9002,
+      "step": 395
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.363324716984375e-06,
+      "loss": 3.3311,
+      "step": 396
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.319481788721833e-06,
+      "loss": 2.5855,
+      "step": 397
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.275651998382377e-06,
+      "loss": 2.6173,
+      "step": 398
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.231836192133532e-06,
+      "loss": 3.3126,
+      "step": 399
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.188035215872858e-06,
+      "loss": 3.1054,
+      "step": 400
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.144249915211605e-06,
+      "loss": 3.0286,
+      "step": 401
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 9.100481135458393e-06,
+      "loss": 3.237,
+      "step": 402
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.056729721602904e-06,
+      "loss": 3.1178,
+      "step": 403
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 9.012996518299547e-06,
+      "loss": 3.096,
+      "step": 404
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.969282369851163e-06,
+      "loss": 2.5678,
+      "step": 405
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.92558812019273e-06,
+      "loss": 3.2872,
+      "step": 406
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.881914612875062e-06,
+      "loss": 3.6379,
+      "step": 407
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.838262691048529e-06,
+      "loss": 3.2067,
+      "step": 408
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 8.79463319744677e-06,
+      "loss": 3.1514,
+      "step": 409
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.751026974370438e-06,
+      "loss": 3.5105,
+      "step": 410
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.70744486367092e-06,
+      "loss": 3.6947,
+      "step": 411
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.66388770673411e-06,
+      "loss": 3.0926,
+      "step": 412
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.620356344464135e-06,
+      "loss": 2.4111,
+      "step": 413
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.576851617267151e-06,
+      "loss": 3.2351,
+      "step": 414
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.533374365035089e-06,
+      "loss": 3.1198,
+      "step": 415
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 8.489925427129469e-06,
+      "loss": 3.0652,
+      "step": 416
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.446505642365174e-06,
+      "loss": 3.0873,
+      "step": 417
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.40311584899426e-06,
+      "loss": 2.4897,
+      "step": 418
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.359756884689785e-06,
+      "loss": 3.0725,
+      "step": 419
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.316429586529616e-06,
+      "loss": 2.6962,
+      "step": 420
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.273134790980295e-06,
+      "loss": 3.4296,
+      "step": 421
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.22987333388087e-06,
+      "loss": 3.106,
+      "step": 422
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.186646050426763e-06,
+      "loss": 2.7377,
+      "step": 423
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 8.143453775153646e-06,
+      "loss": 3.4143,
+      "step": 424
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.100297341921342e-06,
+      "loss": 3.28,
+      "step": 425
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.057177583897704e-06,
+      "loss": 3.3969,
+      "step": 426
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 8.014095333542548e-06,
+      "loss": 3.4838,
+      "step": 427
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.971051422591571e-06,
+      "loss": 3.4206,
+      "step": 428
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.928046682040311e-06,
+      "loss": 2.6433,
+      "step": 429
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.885081942128074e-06,
+      "loss": 3.307,
+      "step": 430
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 7.84215803232194e-06,
+      "loss": 3.0421,
+      "step": 431
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.79927578130072e-06,
+      "loss": 3.3433,
+      "step": 432
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.756436016938973e-06,
+      "loss": 3.0555,
+      "step": 433
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.713639566291028e-06,
+      "loss": 2.9357,
+      "step": 434
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.670887255575003e-06,
+      "loss": 3.4509,
+      "step": 435
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.628179910156859e-06,
+      "loss": 3.6262,
+      "step": 436
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.585518354534473e-06,
+      "loss": 3.4636,
+      "step": 437
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.542903412321714e-06,
+      "loss": 2.9969,
+      "step": 438
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 7.500335906232544e-06,
+      "loss": 2.6747,
+      "step": 439
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4578166580651335e-06,
+      "loss": 3.6768,
+      "step": 440
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.4153464886859925e-06,
+      "loss": 3.7074,
+      "step": 441
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.372926218014131e-06,
+      "loss": 3.9206,
+      "step": 442
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.330556665005235e-06,
+      "loss": 2.9609,
+      "step": 443
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2882386476358304e-06,
+      "loss": 3.1899,
+      "step": 444
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.2459729828875256e-06,
+      "loss": 3.1902,
+      "step": 445
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 7.203760486731204e-06,
+      "loss": 3.2165,
+      "step": 446
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.161601974111308e-06,
+      "loss": 3.67,
+      "step": 447
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.119498258930073e-06,
+      "loss": 3.0986,
+      "step": 448
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.0774501540318305e-06,
+      "loss": 3.0567,
+      "step": 449
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 7.035458471187312e-06,
+      "loss": 3.2137,
+      "step": 450
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.993524021077989e-06,
+      "loss": 3.1834,
+      "step": 451
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.951647613280397e-06,
+      "loss": 3.3187,
+      "step": 452
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 6.909830056250527e-06,
+      "loss": 3.4788,
+      "step": 453
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.868072157308213e-06,
+      "loss": 2.7824,
+      "step": 454
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.826374722621536e-06,
+      "loss": 3.4156,
+      "step": 455
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.784738557191284e-06,
+      "loss": 3.5492,
+      "step": 456
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.7431644648353785e-06,
+      "loss": 3.0707,
+      "step": 457
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.701653248173382e-06,
+      "loss": 3.5231,
+      "step": 458
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.660205708610987e-06,
+      "loss": 3.2012,
+      "step": 459
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.618822646324563e-06,
+      "loss": 3.2893,
+      "step": 460
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 6.577504860245684e-06,
+      "loss": 3.3058,
+      "step": 461
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.536253148045726e-06,
+      "loss": 2.6357,
+      "step": 462
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.495068306120452e-06,
+      "loss": 3.4267,
+      "step": 463
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.453951129574644e-06,
+      "loss": 3.3638,
+      "step": 464
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.41290241220676e-06,
+      "loss": 3.7637,
+      "step": 465
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.3719229464935915e-06,
+      "loss": 3.0229,
+      "step": 466
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.331013523574978e-06,
+      "loss": 3.0312,
+      "step": 467
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 6.290174933238531e-06,
+      "loss": 3.8664,
+      "step": 468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.249407963904381e-06,
+      "loss": 2.5091,
+      "step": 469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.208713402609968e-06,
+      "loss": 2.9386,
+      "step": 470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.168092034994832e-06,
+      "loss": 3.5734,
+      "step": 471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.127544645285448e-06,
+      "loss": 2.7051,
+      "step": 472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.087072016280111e-06,
+      "loss": 3.2061,
+      "step": 473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.046674929333787e-06,
+      "loss": 3.3821,
+      "step": 474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 6.006354164343047e-06,
+      "loss": 3.1994,
+      "step": 475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 5.96611049973101e-06,
+      "loss": 3.2121,
+      "step": 476
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.925944712432317e-06,
+      "loss": 3.5318,
+      "step": 477
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.885857577878122e-06,
+      "loss": 3.1452,
+      "step": 478
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.845849869981137e-06,
+      "loss": 3.1709,
+      "step": 479
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.8059223611206716e-06,
+      "loss": 2.6462,
+      "step": 480
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.766075822127735e-06,
+      "loss": 3.2483,
+      "step": 481
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.726311022270152e-06,
+      "loss": 3.1185,
+      "step": 482
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 5.686628729237713e-06,
+      "loss": 2.5134,
+      "step": 483
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.647029709127355e-06,
+      "loss": 3.1885,
+      "step": 484
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.6075147264283526e-06,
+      "loss": 3.9264,
+      "step": 485
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.5680845440075885e-06,
+      "loss": 3.221,
+      "step": 486
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.528739923094806e-06,
+      "loss": 3.0646,
+      "step": 487
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4894816232679195e-06,
+      "loss": 3.3041,
+      "step": 488
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.450310402438353e-06,
+      "loss": 3.8586,
+      "step": 489
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 5.4112270168363854e-06,
+      "loss": 3.555,
+      "step": 490
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.3722322209966024e-06,
+      "loss": 2.9065,
+      "step": 491
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.333326767743263e-06,
+      "loss": 3.737,
+      "step": 492
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.294511408175825e-06,
+      "loss": 3.0565,
+      "step": 493
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.2557868916543996e-06,
+      "loss": 3.0359,
+      "step": 494
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.217153965785315e-06,
+      "loss": 3.2647,
+      "step": 495
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.178613376406672e-06,
+      "loss": 4.1204,
+      "step": 496
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.14016586757394e-06,
+      "loss": 3.6044,
+      "step": 497
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 5.1018121815456045e-06,
+      "loss": 3.0304,
+      "step": 498
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.063553058768814e-06,
+      "loss": 3.163,
+      "step": 499
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 5.025389237865128e-06,
+      "loss": 3.1335,
+      "step": 500
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.987321455616206e-06,
+      "loss": 3.3261,
+      "step": 501
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.9493504469496235e-06,
+      "loss": 2.9559,
+      "step": 502
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.911476944924651e-06,
+      "loss": 3.358,
+      "step": 503
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.873701680718146e-06,
+      "loss": 3.3523,
+      "step": 504
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 4.836025383610382e-06,
+      "loss": 3.6517,
+      "step": 505
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.798448780971013e-06,
+      "loss": 3.3581,
+      "step": 506
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7609725982450176e-06,
+      "loss": 3.0971,
+      "step": 507
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.7235975589386715e-06,
+      "loss": 2.7656,
+      "step": 508
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.686324384605629e-06,
+      "loss": 2.3538,
+      "step": 509
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.649153794832939e-06,
+      "loss": 2.6863,
+      "step": 510
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.612086507227186e-06,
+      "loss": 2.8402,
+      "step": 511
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 4.5751232374006304e-06,
+      "loss": 2.9124,
+      "step": 512
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.538264698957387e-06,
+      "loss": 3.5337,
+      "step": 513
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.501511603479653e-06,
+      "loss": 3.4941,
+      "step": 514
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.4648646605139605e-06,
+      "loss": 3.2193,
+      "step": 515
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.428324577557501e-06,
+      "loss": 3.2332,
+      "step": 516
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.391892060044435e-06,
+      "loss": 2.8695,
+      "step": 517
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.355567811332311e-06,
+      "loss": 3.1526,
+      "step": 518
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.319352532688444e-06,
+      "loss": 2.7082,
+      "step": 519
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 4.283246923276411e-06,
+      "loss": 3.2863,
+      "step": 520
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.247251680142542e-06,
+      "loss": 3.1025,
+      "step": 521
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.211367498202456e-06,
+      "loss": 2.8076,
+      "step": 522
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.175595070227655e-06,
+      "loss": 3.4128,
+      "step": 523
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.13993508683214e-06,
+      "loss": 3.5704,
+      "step": 524
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.1043882364590895e-06,
+      "loss": 3.8676,
+      "step": 525
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.068955205367559e-06,
+      "loss": 3.5048,
+      "step": 526
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 4.033636677619242e-06,
+      "loss": 3.0329,
+      "step": 527
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.998433335065251e-06,
+      "loss": 3.3583,
+      "step": 528
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.96334585733296e-06,
+      "loss": 3.2228,
+      "step": 529
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.9283749218128885e-06,
+      "loss": 3.2615,
+      "step": 530
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.893521203645618e-06,
+      "loss": 2.7499,
+      "step": 531
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.858785375708764e-06,
+      "loss": 2.3303,
+      "step": 532
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.824168108603971e-06,
+      "loss": 3.3019,
+      "step": 533
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7896700706439826e-06,
+      "loss": 2.8705,
+      "step": 534
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 3.7552919278397335e-06,
+      "loss": 2.956,
+      "step": 535
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.7210343438874917e-06,
+      "loss": 3.2552,
+      "step": 536
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6868979801560443e-06,
+      "loss": 3.2201,
+      "step": 537
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6528834956739224e-06,
+      "loss": 3.1987,
+      "step": 538
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.6189915471166927e-06,
+      "loss": 3.076,
+      "step": 539
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5852227887942713e-06,
+      "loss": 3.6418,
+      "step": 540
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.5515778726382967e-06,
+      "loss": 2.9409,
+      "step": 541
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 3.518057448189527e-06,
+      "loss": 2.8558,
+      "step": 542
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4846621625853248e-06,
+      "loss": 3.2091,
+      "step": 543
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4513926605471504e-06,
+      "loss": 2.7831,
+      "step": 544
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.4182495843681117e-06,
+      "loss": 3.3716,
+      "step": 545
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.385233573900576e-06,
+      "loss": 3.623,
+      "step": 546
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3523452665438004e-06,
+      "loss": 3.0671,
+      "step": 547
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.3195852972316435e-06,
+      "loss": 3.5666,
+      "step": 548
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 3.2869542984202974e-06,
+      "loss": 4.0739,
+      "step": 549
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.254452900076083e-06,
+      "loss": 3.4124,
+      "step": 550
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.2220817296632845e-06,
+      "loss": 3.1846,
+      "step": 551
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1898414121320277e-06,
+      "loss": 3.2453,
+      "step": 552
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.1577325699062424e-06,
+      "loss": 3.3743,
+      "step": 553
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.125755822871607e-06,
+      "loss": 3.313,
+      "step": 554
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.093911788363617e-06,
+      "loss": 3.1968,
+      "step": 555
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.062201081155637e-06,
+      "loss": 2.9313,
+      "step": 556
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 3.0306243134470668e-06,
+      "loss": 3.1517,
+      "step": 557
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9991820948514795e-06,
+      "loss": 3.4108,
+      "step": 558
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.9678750323848893e-06,
+      "loss": 3.4752,
+      "step": 559
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.936703730454017e-06,
+      "loss": 3.4397,
+      "step": 560
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.90566879084461e-06,
+      "loss": 2.8048,
+      "step": 561
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8747708127098593e-06,
+      "loss": 2.8351,
+      "step": 562
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.8440103925587904e-06,
+      "loss": 4.0691,
+      "step": 563
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 2.813388124244778e-06,
+      "loss": 3.0715,
+      "step": 564
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7829045989540594e-06,
+      "loss": 3.2249,
+      "step": 565
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7525604051943512e-06,
+      "loss": 3.3365,
+      "step": 566
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.7223561287834467e-06,
+      "loss": 3.0979,
+      "step": 567
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.692292352837942e-06,
+      "loss": 2.9278,
+      "step": 568
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.662369657761963e-06,
+      "loss": 3.3263,
+      "step": 569
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.6325886212359496e-06,
+      "loss": 2.966,
+      "step": 570
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 2.602949818205539e-06,
+      "loss": 3.1721,
+      "step": 571
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5734538208704197e-06,
+      "loss": 3.675,
+      "step": 572
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.5441011986733165e-06,
+      "loss": 3.1926,
+      "step": 573
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.514892518288988e-06,
+      "loss": 2.3935,
+      "step": 574
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.485828343613288e-06,
+      "loss": 2.3375,
+      "step": 575
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.456909235752276e-06,
+      "loss": 3.4595,
+      "step": 576
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.4281357530113804e-06,
+      "loss": 3.0297,
+      "step": 577
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.399508450884631e-06,
+      "loss": 3.4869,
+      "step": 578
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 2.3710278820439313e-06,
+      "loss": 3.2857,
+      "step": 579
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3426945963283853e-06,
+      "loss": 3.1201,
+      "step": 580
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.3145091407336785e-06,
+      "loss": 3.4965,
+      "step": 581
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2864720594015288e-06,
+      "loss": 3.167,
+      "step": 582
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2585838936091753e-06,
+      "loss": 3.2821,
+      "step": 583
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.230845181758928e-06,
+      "loss": 3.2159,
+      "step": 584
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.2032564593677773e-06,
+      "loss": 3.7117,
+      "step": 585
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 2.1758182590570454e-06,
+      "loss": 3.6264,
+      "step": 586
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.148531110542118e-06,
+      "loss": 3.1606,
+      "step": 587
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.1213955406222076e-06,
+      "loss": 2.9482,
+      "step": 588
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.09441207317019e-06,
+      "loss": 3.3531,
+      "step": 589
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0675812291224796e-06,
+      "loss": 3.0972,
+      "step": 590
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.0409035264689857e-06,
+      "loss": 2.8781,
+      "step": 591
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 2.014379480243105e-06,
+      "loss": 3.5874,
+      "step": 592
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.988009602511779e-06,
+      "loss": 3.1568,
+      "step": 593
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.961794402365611e-06,
+      "loss": 3.7799,
+      "step": 594
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.935734385909028e-06,
+      "loss": 2.571,
+      "step": 595
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.9098300562505266e-06,
+      "loss": 2.9241,
+      "step": 596
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8840819134929467e-06,
+      "loss": 3.3022,
+      "step": 597
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8584904547238214e-06,
+      "loss": 3.4095,
+      "step": 598
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8330561740057839e-06,
+      "loss": 3.1118,
+      "step": 599
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.8077795623670135e-06,
+      "loss": 3.2652,
+      "step": 600
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 1.7826611077917843e-06,
+      "loss": 2.8249,
+      "step": 601
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.757701295211014e-06,
+      "loss": 3.2609,
+      "step": 602
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7329006064929232e-06,
+      "loss": 3.333,
+      "step": 603
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.7082595204337183e-06,
+      "loss": 3.1817,
+      "step": 604
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.683778512748362e-06,
+      "loss": 3.1209,
+      "step": 605
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6594580560613782e-06,
+      "loss": 3.3353,
+      "step": 606
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6352986198977327e-06,
+      "loss": 3.2617,
+      "step": 607
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 1.6113006706737667e-06,
+      "loss": 2.9665,
+      "step": 608
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.587464671688187e-06,
+      "loss": 3.2136,
+      "step": 609
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.563791083113142e-06,
+      "loss": 3.4878,
+      "step": 610
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.540280361985308e-06,
+      "loss": 3.8198,
+      "step": 611
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.5169329621970918e-06,
+      "loss": 2.3477,
+      "step": 612
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4937493344878474e-06,
+      "loss": 3.2629,
+      "step": 613
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4707299264351914e-06,
+      "loss": 3.9149,
+      "step": 614
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4478751824463543e-06,
+      "loss": 3.351,
+      "step": 615
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 1.4251855437495976e-06,
+      "loss": 3.1838,
+      "step": 616
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.4026614483857037e-06,
+      "loss": 3.5445,
+      "step": 617
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3803033311995072e-06,
+      "loss": 3.2856,
+      "step": 618
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3581116238315194e-06,
+      "loss": 3.097,
+      "step": 619
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.336086754709569e-06,
+      "loss": 3.5377,
+      "step": 620
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.3142291490405568e-06,
+      "loss": 3.0701,
+      "step": 621
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2925392288022299e-06,
+      "loss": 3.4196,
+      "step": 622
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 1.2710174127350362e-06,
+      "loss": 2.5715,
+      "step": 623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2496641163340562e-06,
+      "loss": 3.4349,
+      "step": 624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2284797518409575e-06,
+      "loss": 3.473,
+      "step": 625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.2074647282360573e-06,
+      "loss": 3.5017,
+      "step": 626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1866194512304075e-06,
+      "loss": 3.6054,
+      "step": 627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.165944323257986e-06,
+      "loss": 3.4716,
+      "step": 628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.1454397434679022e-06,
+      "loss": 3.6019,
+      "step": 629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 1.125106107716708e-06,
+      "loss": 3.0826,
+      "step": 630
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.10494380856075e-06,
+      "loss": 4.1209,
+      "step": 631
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0849532352485903e-06,
+      "loss": 2.9778,
+      "step": 632
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0651347737134965e-06,
+      "loss": 3.4931,
+      "step": 633
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0454888065659775e-06,
+      "loss": 2.9216,
+      "step": 634
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0260157130864178e-06,
+      "loss": 3.285,
+      "step": 635
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 1.0067158692177325e-06,
+      "loss": 3.093,
+      "step": 636
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.87589647558135e-07,
+      "loss": 2.8769,
+      "step": 637
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 9.686374173539147e-07,
+      "loss": 3.002,
+      "step": 638
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.49859544492332e-07,
+      "loss": 3.396,
+      "step": 639
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.312563914945461e-07,
+      "loss": 3.2366,
+      "step": 640
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 9.128283175086106e-07,
+      "loss": 3.1459,
+      "step": 641
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.945756783025528e-07,
+      "loss": 3.1189,
+      "step": 642
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.76498826257488e-07,
+      "loss": 2.4421,
+      "step": 643
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.585981103608343e-07,
+      "loss": 2.7928,
+      "step": 644
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 8.40873876199565e-07,
+      "loss": 3.6139,
+      "step": 645
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.233264659535367e-07,
+      "loss": 3.3497,
+      "step": 646
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 8.059562183888903e-07,
+      "loss": 3.1132,
+      "step": 647
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.887634688515e-07,
+      "loss": 3.0642,
+      "step": 648
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.71748549260507e-07,
+      "loss": 2.4027,
+      "step": 649
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.549117881019141e-07,
+      "loss": 3.2094,
+      "step": 650
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.382535104222366e-07,
+      "loss": 3.1254,
+      "step": 651
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.21774037822226e-07,
+      "loss": 3.1048,
+      "step": 652
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 7.054736884506718e-07,
+      "loss": 3.2661,
+      "step": 653
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.8935277699825e-07,
+      "loss": 3.1209,
+      "step": 654
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.734116146914516e-07,
+      "loss": 2.804,
+      "step": 655
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.576505092865748e-07,
+      "loss": 3.0518,
+      "step": 656
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.420697650637753e-07,
+      "loss": 3.3038,
+      "step": 657
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.266696828212071e-07,
+      "loss": 3.4095,
+      "step": 658
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 6.114505598692011e-07,
+      "loss": 3.1272,
+      "step": 659
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 5.964126900245359e-07,
+      "loss": 2.845,
+      "step": 660
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.815563636047539e-07,
+      "loss": 3.0669,
+      "step": 661
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.668818674225684e-07,
+      "loss": 3.3752,
+      "step": 662
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.523894847803235e-07,
+      "loss": 3.1723,
+      "step": 663
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.380794954645141e-07,
+      "loss": 3.25,
+      "step": 664
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.23952175740402e-07,
+      "loss": 3.3449,
+      "step": 665
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 5.100077983466667e-07,
+      "loss": 2.9806,
+      "step": 666
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 4.962466324901483e-07,
+      "loss": 3.1956,
+      "step": 667
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.826689438406495e-07,
+      "loss": 3.681,
+      "step": 668
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.6927499452580574e-07,
+      "loss": 3.2948,
+      "step": 669
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.5606504312602384e-07,
+      "loss": 3.3863,
+      "step": 670
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.4303934466948804e-07,
+      "loss": 3.1736,
+      "step": 671
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.3019815062724567e-07,
+      "loss": 3.2741,
+      "step": 672
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.1754170890833777e-07,
+      "loss": 3.6696,
+      "step": 673
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 4.0507026385502747e-07,
+      "loss": 2.936,
+      "step": 674
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 3.9278405623806914e-07,
+      "loss": 3.6071,
+      "step": 675
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.806833232520746e-07,
+      "loss": 3.0734,
+      "step": 676
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.687682985109209e-07,
+      "loss": 2.9812,
+      "step": 677
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.5703921204324863e-07,
+      "loss": 3.3496,
+      "step": 678
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.454962902880199e-07,
+      "loss": 2.6775,
+      "step": 679
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.3413975609013713e-07,
+      "loss": 2.7971,
+      "step": 680
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.2296982869616134e-07,
+      "loss": 2.874,
+      "step": 681
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 3.1198672375005403e-07,
+      "loss": 2.9508,
+      "step": 682
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 3.0119065328903517e-07,
+      "loss": 2.9765,
+      "step": 683
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.905818257394799e-07,
+      "loss": 3.471,
+      "step": 684
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.801604459128926e-07,
+      "loss": 3.3773,
+      "step": 685
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.6992671500196134e-07,
+      "loss": 3.0457,
+      "step": 686
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5988083057666534e-07,
+      "loss": 3.1264,
+      "step": 687
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.5002298658046484e-07,
+      "loss": 3.0226,
+      "step": 688
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.4035337332655504e-07,
+      "loss": 3.1171,
+      "step": 689
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 2.308721774941991e-07,
+      "loss": 2.9497,
+      "step": 690
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.2157958212510877e-07,
+      "loss": 3.2643,
+      "step": 691
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.124757666199273e-07,
+      "loss": 3.9361,
+      "step": 692
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 2.035609067347566e-07,
+      "loss": 2.8183,
+      "step": 693
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.9483517457776436e-07,
+      "loss": 3.1843,
+      "step": 694
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.8629873860586567e-07,
+      "loss": 2.8556,
+      "step": 695
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.7795176362146783e-07,
+      "loss": 2.8733,
+      "step": 696
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.6979441076928837e-07,
+      "loss": 3.2043,
+      "step": 697
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.6182683753324435e-07,
+      "loss": 3.5809,
+      "step": 698
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.5404919773341576e-07,
+      "loss": 3.2,
+      "step": 699
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.464616415230702e-07,
+      "loss": 3.0855,
+      "step": 700
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3906431538576626e-07,
+      "loss": 3.263,
+      "step": 701
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.3185736213252808e-07,
+      "loss": 3.1197,
+      "step": 702
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.2484092089908307e-07,
+      "loss": 3.1774,
+      "step": 703
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 1.1801512714318286e-07,
+      "loss": 3.0137,
+      "step": 704
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.113801126419789e-07,
+      "loss": 3.5185,
+      "step": 705
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 1.0493600548948879e-07,
+      "loss": 2.9359,
+      "step": 706
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.8682930094115e-08,
+      "loss": 3.4492,
+      "step": 707
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 9.262100717624678e-08,
+      "loss": 3.4146,
+      "step": 708
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.675035376593088e-08,
+      "loss": 2.922,
+      "step": 709
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 8.107108320060675e-08,
+      "loss": 3.4334,
+      "step": 710
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.558330512292378e-08,
+      "loss": 3.4152,
+      "step": 711
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 7.028712547862526e-08,
+      "loss": 3.407,
+      "step": 712
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.51826465144978e-08,
+      "loss": 2.4502,
+      "step": 713
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 6.026996677640062e-08,
+      "loss": 2.903,
+      "step": 714
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.5549181107362734e-08,
+      "loss": 3.2649,
+      "step": 715
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 5.102038064575099e-08,
+      "loss": 3.4949,
+      "step": 716
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.6683652823513725e-08,
+      "loss": 3.0703,
+      "step": 717
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 4.253908136448881e-08,
+      "loss": 2.8247,
+      "step": 718
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.858674628278825e-08,
+      "loss": 2.8248,
+      "step": 719
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.482672388125719e-08,
+      "loss": 2.9583,
+      "step": 720
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 3.125908674999289e-08,
+      "loss": 3.529,
+      "step": 721
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.7883903764953647e-08,
+      "loss": 3.2435,
+      "step": 722
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.470124008661978e-08,
+      "loss": 3.1003,
+      "step": 723
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 2.171115715874139e-08,
+      "loss": 3.5848,
+      "step": 724
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.8913712707149255e-08,
+      "loss": 2.9616,
+      "step": 725
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.630896073864352e-08,
+      "loss": 3.1862,
+      "step": 726
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3896951539945635e-08,
+      "loss": 3.3656,
+      "step": 727
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1677731676733584e-08,
+      "loss": 3.1796,
+      "step": 728
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.651343992740369e-09,
+      "loss": 2.9224,
+      "step": 729
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.817827608924689e-09,
+      "loss": 3.9175,
+      "step": 730
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.1772179227181926e-09,
+      "loss": 3.1324,
+      "step": 731
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.7295466073427055e-09,
+      "loss": 3.5297,
+      "step": 732
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.474841611197377e-09,
+      "loss": 3.1745,
+      "step": 733
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4131271573191172e-09,
+      "loss": 3.083,
+      "step": 734
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.5444237429140806e-09,
+      "loss": 3.3742,
+      "step": 735
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.687481389657582e-10,
+      "loss": 3.0691,
+      "step": 736
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.861133899063507e-10,
+      "loss": 3.6025,
+      "step": 737
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 9.652881336696951e-11,
+      "loss": 3.413,
+      "step": 738
+    },
+    {
+      "epoch": 1.0,
+      "step": 738,
+      "total_flos": 2.330591592710144e+17,
+      "train_loss": 3.402625040956306,
+      "train_runtime": 2292.4105,
+      "train_samples_per_second": 1.287,
+      "train_steps_per_second": 0.322
+    }
+  ],
+  "max_steps": 738,
+  "num_train_epochs": 1,
+  "total_flos": 2.330591592710144e+17,
+  "trial_name": null,
+  "trial_params": null
+}