diff --git "a/checkpoint-800/trainer_state.json" "b/checkpoint-800/trainer_state.json"
--- "a/checkpoint-800/trainer_state.json"
+++ "b/checkpoint-800/trainer_state.json"
@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.4103236794471741,
+  "best_metric": 0.43044719099998474,
   "best_model_checkpoint": "experts/mistralic-expert-3/checkpoint-800",
-  "epoch": 0.315893385982231,
+  "epoch": 0.1579466929911155,
   "eval_steps": 200,
   "global_step": 800,
   "is_hyper_param_search": false,
@@ -11,773 +11,5093 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.5036,
+      "loss": 0.3844,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.4476,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.715,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.6904,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.4288,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.38,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.5455,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.6589,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.6253,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.473,
       "step": 10
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.3938,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7852,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.3725,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.5708,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.4799,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.4818,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.5886,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.4483,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.3668,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.4814,
+      "loss": 0.4126,
       "step": 20
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.4493,
-      "step": 30
+      "loss": 0.5653,
+      "step": 21
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.4445,
-      "step": 40
+      "loss": 0.592,
+      "step": 22
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.4996,
-      "step": 50
+      "loss": 0.5745,
+      "step": 23
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.4473,
-      "step": 60
+      "loss": 0.4949,
+      "step": 24
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.4851,
-      "step": 70
+      "loss": 0.433,
+      "step": 25
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7357,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5013,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
       "learning_rate": 0.0002,
       "loss": 0.4328,
-      "step": 80
+      "step": 28
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4486,
-      "step": 90
+      "loss": 0.4199,
+      "step": 29
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4732,
-      "step": 100
+      "loss": 0.5718,
+      "step": 30
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4512,
-      "step": 110
+      "loss": 0.6628,
+      "step": 31
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4638,
-      "step": 120
+      "loss": 0.4337,
+      "step": 32
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4559,
-      "step": 130
+      "loss": 0.3818,
+      "step": 33
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4411,
-      "step": 140
+      "loss": 0.4064,
+      "step": 34
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4187,
-      "step": 150
+      "loss": 0.4428,
+      "step": 35
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4495,
-      "step": 160
+      "loss": 0.3465,
+      "step": 36
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4172,
-      "step": 170
+      "loss": 0.5089,
+      "step": 37
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4359,
-      "step": 180
+      "loss": 0.4848,
+      "step": 38
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4444,
-      "step": 190
+      "loss": 0.3647,
+      "step": 39
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4625,
-      "step": 200
+      "loss": 0.4391,
+      "step": 40
     },
     {
-      "epoch": 0.08,
-      "eval_loss": 0.43659043312072754,
-      "eval_runtime": 125.7967,
-      "eval_samples_per_second": 7.949,
-      "eval_steps_per_second": 3.975,
-      "step": 200
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4559,
+      "step": 41
     },
     {
-      "epoch": 0.08,
-      "mmlu_eval_accuracy": 0.59599123107443,
-      "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
-      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
-      "mmlu_eval_accuracy_astronomy": 0.6875,
-      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
-      "mmlu_eval_accuracy_college_biology": 0.625,
-      "mmlu_eval_accuracy_college_chemistry": 0.375,
-      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
-      "mmlu_eval_accuracy_college_medicine": 0.5909090909090909,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
-      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
-      "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
-      "mmlu_eval_accuracy_electrical_engineering": 0.625,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
-      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
-      "mmlu_eval_accuracy_global_facts": 0.4,
-      "mmlu_eval_accuracy_high_school_biology": 0.625,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
-      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5581395348837209,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
-      "mmlu_eval_accuracy_high_school_physics": 0.058823529411764705,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
-      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
-      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
-      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
-      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
-      "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
-      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
-      "mmlu_eval_accuracy_management": 0.8181818181818182,
-      "mmlu_eval_accuracy_marketing": 0.92,
-      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
-      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
-      "mmlu_eval_accuracy_moral_scenarios": 0.33,
-      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
-      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
-      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
-      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
-      "mmlu_eval_accuracy_professional_law": 0.40588235294117647,
-      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
-      "mmlu_eval_accuracy_professional_psychology": 0.6086956521739131,
-      "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.6666666666666666,
-      "mmlu_eval_accuracy_sociology": 0.9545454545454546,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
-      "mmlu_eval_accuracy_virology": 0.5555555555555556,
-      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
-      "mmlu_loss": 1.5606910698407623,
-      "step": 200
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4008,
+      "step": 42
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4345,
-      "step": 210
+      "loss": 0.4791,
+      "step": 43
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4352,
-      "step": 220
+      "loss": 0.4348,
+      "step": 44
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4573,
-      "step": 230
+      "loss": 0.4618,
+      "step": 45
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.3892,
-      "step": 240
+      "loss": 0.5419,
+      "step": 46
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.3906,
-      "step": 250
+      "loss": 0.4007,
+      "step": 47
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4336,
-      "step": 260
+      "loss": 0.6463,
+      "step": 48
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4311,
-      "step": 270
+      "loss": 0.568,
+      "step": 49
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4118,
-      "step": 280
+      "loss": 0.5005,
+      "step": 50
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4112,
-      "step": 290
+      "loss": 0.4113,
+      "step": 51
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4579,
-      "step": 300
+      "loss": 0.4875,
+      "step": 52
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.397,
-      "step": 310
+      "loss": 0.3864,
+      "step": 53
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.39,
-      "step": 320
+      "loss": 0.429,
+      "step": 54
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4383,
-      "step": 330
+      "loss": 0.4502,
+      "step": 55
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4338,
-      "step": 340
+      "loss": 0.43,
+      "step": 56
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.4496,
-      "step": 350
+      "loss": 0.3315,
+      "step": 57
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.578,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4223,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4504,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.445,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5778,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4022,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4877,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5165,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5367,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4669,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.639,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.3895,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5085,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.3571,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4391,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5531,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.3872,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.4402,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4963,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4507,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.3566,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4096,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.3198,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6501,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4595,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.3907,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4045,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4506,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.702,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5446,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4881,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6571,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5235,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4005,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4138,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6961,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5523,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5507,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4296,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5833,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.3338,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4244,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6051,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5386,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5212,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5893,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.2868,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.3358,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.475,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.417,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.3283,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.517,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.373,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4229,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4553,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4169,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4359,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6103,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6083,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6512,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.584,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.3836,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.3245,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5128,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4698,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5391,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.406,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4633,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6352,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6486,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4463,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4328,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5154,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5053,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.485,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6043,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.49,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4202,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.621,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4248,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4826,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4296,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4634,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4633,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3996,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4175,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3721,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3758,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.542,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5827,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4624,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4605,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.526,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3986,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5239,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3477,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5489,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3926,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5175,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5478,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3951,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5371,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.2852,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4754,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3596,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4046,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5032,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5228,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4347,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.487,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3746,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5577,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3328,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4275,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4712,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4113,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.3732,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5678,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4048,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4925,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4892,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4925,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4083,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4729,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4136,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3813,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5984,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4421,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5827,
+      "step": 186
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.471,
+      "step": 187
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.436,
+      "step": 188
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4074,
+      "step": 189
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4846,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3325,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4094,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4312,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4903,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5195,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5583,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6267,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5788,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4445,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5825,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.4535829424858093,
+      "eval_runtime": 126.5186,
+      "eval_samples_per_second": 7.904,
+      "eval_steps_per_second": 3.952,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "mmlu_eval_accuracy": 0.587308195877975,
+      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
+      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
+      "mmlu_eval_accuracy_astronomy": 0.8125,
+      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
+      "mmlu_eval_accuracy_college_biology": 0.625,
+      "mmlu_eval_accuracy_college_chemistry": 0.25,
+      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_medicine": 0.5909090909090909,
+      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
+      "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
+      "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
+      "mmlu_eval_accuracy_electrical_engineering": 0.625,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
+      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
+      "mmlu_eval_accuracy_global_facts": 0.4,
+      "mmlu_eval_accuracy_high_school_biology": 0.5625,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
+      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5813953488372093,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
+      "mmlu_eval_accuracy_high_school_physics": 0.058823529411764705,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
+      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
+      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
+      "mmlu_eval_accuracy_management": 0.8181818181818182,
+      "mmlu_eval_accuracy_marketing": 0.92,
+      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
+      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
+      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
+      "mmlu_eval_accuracy_moral_scenarios": 0.29,
+      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
+      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
+      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
+      "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
+      "mmlu_eval_accuracy_professional_law": 0.4,
+      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
+      "mmlu_eval_accuracy_professional_psychology": 0.5942028985507246,
+      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_security_studies": 0.6666666666666666,
+      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
+      "mmlu_eval_accuracy_virology": 0.5555555555555556,
+      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
+      "mmlu_loss": 1.586101784096063,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4679,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6958,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4558,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4753,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6579,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.489,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4094,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4796,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4285,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.57,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4275,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.395,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3844,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3804,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.462,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3336,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4949,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3938,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3869,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4721,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6252,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7146,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3661,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4267,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4813,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4902,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.3782,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4939,
+      "step": 228
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4697,
+      "step": 229
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3897,
+      "step": 230
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4336,
+      "step": 231
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5118,
+      "step": 232
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5036,
+      "step": 233
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3383,
+      "step": 234
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3323,
+      "step": 235
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6961,
+      "step": 236
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4029,
+      "step": 237
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4,
+      "step": 238
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4076,
+      "step": 239
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5963,
+      "step": 240
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6931,
+      "step": 241
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4609,
+      "step": 242
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5343,
+      "step": 243
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4595,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4553,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5125,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3143,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5566,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4033,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5293,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5176,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3558,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4485,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4648,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5255,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4495,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4256,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4648,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3821,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.489,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4883,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4805,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3878,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4923,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.401,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5671,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4653,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3678,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3257,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.481,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3072,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.3776,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4294,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5227,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4617,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5209,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6255,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4825,
+      "step": 278
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3516,
+      "step": 279
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5352,
+      "step": 280
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4315,
+      "step": 281
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5186,
+      "step": 282
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3842,
+      "step": 283
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.361,
+      "step": 284
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4,
+      "step": 285
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4162,
+      "step": 286
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4525,
+      "step": 287
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.398,
+      "step": 288
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4367,
+      "step": 289
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5637,
+      "step": 290
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4982,
+      "step": 291
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.456,
+      "step": 292
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4848,
+      "step": 293
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4005,
+      "step": 294
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4541,
+      "step": 295
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4483,
+      "step": 296
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4353,
+      "step": 297
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3983,
+      "step": 298
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3249,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4995,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5284,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5122,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4249,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4273,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3416,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4184,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.495,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5097,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3626,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5343,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4966,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4583,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.451,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6254,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4295,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.556,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4661,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.44,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4236,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3557,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3709,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3196,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.373,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3542,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.4292,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5218,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3554,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3298,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.3046,
+      "step": 329
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4406,
+      "step": 330
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6105,
+      "step": 331
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4693,
+      "step": 332
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4602,
+      "step": 333
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4868,
+      "step": 334
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3725,
+      "step": 335
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6264,
+      "step": 336
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5627,
+      "step": 337
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4265,
+      "step": 338
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5247,
+      "step": 339
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4599,
+      "step": 340
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3831,
+      "step": 341
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.353,
+      "step": 342
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4889,
+      "step": 343
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3849,
+      "step": 344
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.402,
+      "step": 345
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5243,
+      "step": 346
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.494,
+      "step": 347
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4092,
+      "step": 348
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4419,
+      "step": 349
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4857,
+      "step": 350
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3485,
+      "step": 351
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4136,
+      "step": 352
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.556,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3647,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5793,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4485,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3916,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5084,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5333,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5562,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5308,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4851,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3877,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3881,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4768,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5751,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5152,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4679,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3968,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4654,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4658,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5175,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3991,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.421,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5892,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3395,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4653,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4325,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.3811,
+      "step": 379
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4245,
+      "step": 380
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5546,
+      "step": 381
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4962,
+      "step": 382
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.398,
+      "step": 383
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5522,
+      "step": 384
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3749,
+      "step": 385
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4272,
+      "step": 386
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3686,
+      "step": 387
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4304,
+      "step": 388
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4199,
+      "step": 389
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4634,
+      "step": 390
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5483,
+      "step": 391
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5617,
+      "step": 392
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4602,
+      "step": 393
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.35,
+      "step": 394
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5024,
+      "step": 395
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5558,
+      "step": 396
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3453,
+      "step": 397
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4801,
+      "step": 398
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4665,
+      "step": 399
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6868,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "eval_loss": 0.44599705934524536,
+      "eval_runtime": 126.2866,
+      "eval_samples_per_second": 7.918,
+      "eval_steps_per_second": 3.959,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "mmlu_eval_accuracy": 0.5914391476332194,
+      "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
+      "mmlu_eval_accuracy_anatomy": 0.5,
+      "mmlu_eval_accuracy_astronomy": 0.625,
+      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
+      "mmlu_eval_accuracy_college_biology": 0.625,
+      "mmlu_eval_accuracy_college_chemistry": 0.375,
+      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
+      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
+      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
+      "mmlu_eval_accuracy_conceptual_physics": 0.5,
+      "mmlu_eval_accuracy_econometrics": 0.5,
+      "mmlu_eval_accuracy_electrical_engineering": 0.6875,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
+      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
+      "mmlu_eval_accuracy_global_facts": 0.3,
+      "mmlu_eval_accuracy_high_school_biology": 0.53125,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
+      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.627906976744186,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
+      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
+      "mmlu_eval_accuracy_high_school_psychology": 0.85,
+      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.5,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+      "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
+      "mmlu_eval_accuracy_management": 0.8181818181818182,
+      "mmlu_eval_accuracy_marketing": 0.92,
+      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
+      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
+      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
+      "mmlu_eval_accuracy_moral_scenarios": 0.33,
+      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
+      "mmlu_eval_accuracy_philosophy": 0.7058823529411765,
+      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+      "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
+      "mmlu_eval_accuracy_professional_law": 0.4,
+      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
+      "mmlu_eval_accuracy_professional_psychology": 0.5942028985507246,
+      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_security_studies": 0.6666666666666666,
+      "mmlu_eval_accuracy_sociology": 0.9545454545454546,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
+      "mmlu_eval_accuracy_virology": 0.5555555555555556,
+      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
+      "mmlu_loss": 1.5952020311635717,
+      "step": 400
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4167,
+      "step": 401
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4688,
+      "step": 402
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5566,
+      "step": 403
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3849,
+      "step": 404
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4983,
+      "step": 405
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.387,
+      "step": 406
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4051,
+      "step": 407
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.605,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3609,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.329,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4074,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3769,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3089,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4549,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4936,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.403,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3767,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3885,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4688,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8412,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3976,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3743,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5031,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5841,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5269,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4606,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3955,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4122,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4445,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.3949,
+      "step": 430
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3021,
+      "step": 431
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4357,
+      "step": 432
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3879,
+      "step": 433
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4184,
+      "step": 434
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.402,
+      "step": 435
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4365,
+      "step": 436
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.499,
+      "step": 437
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5597,
+      "step": 438
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.579,
+      "step": 439
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5909,
+      "step": 440
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4426,
+      "step": 441
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.611,
+      "step": 442
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4294,
+      "step": 443
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4749,
+      "step": 444
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4137,
+      "step": 445
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.2984,
+      "step": 446
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4904,
+      "step": 447
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5013,
+      "step": 448
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5519,
+      "step": 449
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3451,
+      "step": 450
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5129,
+      "step": 451
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5881,
+      "step": 452
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6934,
+      "step": 453
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4664,
+      "step": 454
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3699,
+      "step": 455
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4982,
+      "step": 456
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4642,
+      "step": 457
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6157,
+      "step": 458
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4153,
+      "step": 459
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3707,
+      "step": 460
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4899,
+      "step": 461
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3927,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3162,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3328,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4015,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4255,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3613,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4838,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3725,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3931,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5642,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4039,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4343,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5121,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.306,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3682,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5239,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3601,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4352,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.311,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.3291,
+      "step": 481
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4456,
+      "step": 482
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3515,
+      "step": 483
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.397,
+      "step": 484
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4148,
+      "step": 485
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4179,
+      "step": 486
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3313,
+      "step": 487
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4845,
+      "step": 488
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3413,
+      "step": 489
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5205,
+      "step": 490
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4326,
+      "step": 491
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6914,
+      "step": 492
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.2394,
+      "step": 493
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4248,
+      "step": 494
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.35,
+      "step": 495
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3944,
+      "step": 496
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3573,
+      "step": 497
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3848,
+      "step": 498
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4969,
+      "step": 499
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4478,
+      "step": 500
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.2628,
+      "step": 501
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3685,
+      "step": 502
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5894,
+      "step": 503
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3814,
+      "step": 504
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5015,
+      "step": 505
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7351,
+      "step": 506
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3227,
+      "step": 507
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5138,
+      "step": 508
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4135,
+      "step": 509
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4907,
+      "step": 510
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3639,
+      "step": 511
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3524,
+      "step": 512
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.498,
+      "step": 513
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4158,
+      "step": 514
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3911,
+      "step": 515
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6107,
+      "step": 516
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6056,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5007,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4991,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3278,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3798,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3834,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5246,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4098,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3271,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.486,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3848,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3121,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.3917,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6843,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.416,
+      "step": 531
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5641,
+      "step": 532
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4789,
+      "step": 533
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6327,
+      "step": 534
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3822,
+      "step": 535
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5069,
+      "step": 536
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.315,
+      "step": 537
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4768,
+      "step": 538
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3834,
+      "step": 539
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3832,
+      "step": 540
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4364,
+      "step": 541
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4825,
+      "step": 542
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.2895,
+      "step": 543
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4579,
+      "step": 544
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3002,
+      "step": 545
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3306,
+      "step": 546
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.457,
+      "step": 547
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4108,
+      "step": 548
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5136,
+      "step": 549
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3233,
+      "step": 550
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5221,
+      "step": 551
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5275,
+      "step": 552
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3832,
+      "step": 553
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4867,
+      "step": 554
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3573,
+      "step": 555
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.574,
+      "step": 556
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4703,
+      "step": 557
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.445,
+      "step": 558
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4047,
+      "step": 559
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3206,
+      "step": 560
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6012,
+      "step": 561
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3947,
+      "step": 562
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4637,
+      "step": 563
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3375,
+      "step": 564
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4335,
+      "step": 565
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3998,
+      "step": 566
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.498,
+      "step": 567
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7256,
+      "step": 568
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3629,
+      "step": 569
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3133,
+      "step": 570
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4598,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3066,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5406,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3697,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3452,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5807,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.398,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.3154,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4314,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4825,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5535,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5473,
+      "step": 582
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3913,
+      "step": 583
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3965,
+      "step": 584
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4827,
+      "step": 585
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.693,
+      "step": 586
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4705,
+      "step": 587
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3516,
+      "step": 588
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5253,
+      "step": 589
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5529,
+      "step": 590
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4611,
+      "step": 591
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3539,
+      "step": 592
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4487,
+      "step": 593
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3538,
+      "step": 594
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3981,
+      "step": 595
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3618,
+      "step": 596
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4358,
+      "step": 597
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5432,
+      "step": 598
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3794,
+      "step": 599
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3944,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 0.4361969232559204,
+      "eval_runtime": 126.3308,
+      "eval_samples_per_second": 7.916,
+      "eval_steps_per_second": 3.958,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "mmlu_eval_accuracy": 0.5976142582746435,
+      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+      "mmlu_eval_accuracy_astronomy": 0.6875,
+      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
+      "mmlu_eval_accuracy_college_biology": 0.625,
+      "mmlu_eval_accuracy_college_chemistry": 0.375,
+      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+      "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
+      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
+      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
+      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
+      "mmlu_eval_accuracy_econometrics": 0.4166666666666667,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.5121951219512195,
+      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
+      "mmlu_eval_accuracy_global_facts": 0.3,
+      "mmlu_eval_accuracy_high_school_biology": 0.5625,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
+      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.627906976744186,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.6538461538461539,
+      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
+      "mmlu_eval_accuracy_high_school_psychology": 0.85,
+      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+      "mmlu_eval_accuracy_high_school_us_history": 0.8181818181818182,
+      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.5,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
+      "mmlu_eval_accuracy_management": 0.9090909090909091,
+      "mmlu_eval_accuracy_marketing": 0.88,
+      "mmlu_eval_accuracy_medical_genetics": 1.0,
+      "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
+      "mmlu_eval_accuracy_moral_disputes": 0.6052631578947368,
+      "mmlu_eval_accuracy_moral_scenarios": 0.45,
+      "mmlu_eval_accuracy_nutrition": 0.7878787878787878,
+      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
+      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
+      "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
+      "mmlu_eval_accuracy_professional_law": 0.38823529411764707,
+      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
+      "mmlu_eval_accuracy_professional_psychology": 0.5942028985507246,
+      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_security_studies": 0.7407407407407407,
+      "mmlu_eval_accuracy_sociology": 0.9545454545454546,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
+      "mmlu_eval_accuracy_virology": 0.5555555555555556,
+      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
+      "mmlu_loss": 1.4924844036136533,
+      "step": 600
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.2598,
+      "step": 601
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3026,
+      "step": 602
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6204,
+      "step": 603
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4893,
+      "step": 604
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4428,
+      "step": 605
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4569,
+      "step": 606
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4773,
+      "step": 607
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3947,
+      "step": 608
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.413,
+      "step": 609
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5188,
+      "step": 610
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3409,
+      "step": 611
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.456,
+      "step": 612
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4172,
+      "step": 613
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3585,
+      "step": 614
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3622,
+      "step": 615
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3184,
+      "step": 616
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.327,
+      "step": 617
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5281,
+      "step": 618
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3763,
+      "step": 619
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4348,
+      "step": 620
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3816,
+      "step": 621
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5117,
+      "step": 622
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4613,
+      "step": 623
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3679,
+      "step": 624
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4469,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4865,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3539,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.3416,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4377,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5071,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5304,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4644,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.471,
+      "step": 633
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.377,
+      "step": 634
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3438,
+      "step": 635
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3404,
+      "step": 636
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.2779,
+      "step": 637
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4611,
+      "step": 638
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.289,
+      "step": 639
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.36,
+      "step": 640
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3991,
+      "step": 641
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3704,
+      "step": 642
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4606,
+      "step": 643
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5336,
+      "step": 644
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6027,
+      "step": 645
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6684,
+      "step": 646
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4774,
+      "step": 647
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3756,
+      "step": 648
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5438,
+      "step": 649
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3847,
+      "step": 650
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4466,
+      "step": 651
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3292,
+      "step": 652
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4387,
+      "step": 653
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4272,
+      "step": 654
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.667,
+      "step": 655
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4929,
+      "step": 656
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3796,
+      "step": 657
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.377,
+      "step": 658
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3663,
+      "step": 659
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4507,
+      "step": 660
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3757,
+      "step": 661
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5356,
+      "step": 662
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5435,
+      "step": 663
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4153,
+      "step": 664
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3077,
+      "step": 665
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.2956,
+      "step": 666
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7522,
+      "step": 667
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.315,
+      "step": 668
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3322,
+      "step": 669
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4188,
+      "step": 670
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3014,
+      "step": 671
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.491,
+      "step": 672
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6646,
+      "step": 673
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3574,
+      "step": 674
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4224,
+      "step": 675
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3751,
+      "step": 676
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4506,
+      "step": 677
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4523,
+      "step": 678
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4275,
+      "step": 679
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5313,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5943,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.3883,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4622,
+      "step": 683
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3109,
+      "step": 684
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.467,
+      "step": 685
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5738,
+      "step": 686
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5819,
+      "step": 687
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5742,
+      "step": 688
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4333,
+      "step": 689
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7384,
+      "step": 690
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3619,
+      "step": 691
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5534,
+      "step": 692
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.495,
+      "step": 693
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4099,
+      "step": 694
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6178,
+      "step": 695
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3037,
+      "step": 696
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4958,
+      "step": 697
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4404,
+      "step": 698
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3397,
+      "step": 699
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4728,
+      "step": 700
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3766,
+      "step": 701
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3133,
+      "step": 702
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.2881,
+      "step": 703
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3153,
+      "step": 704
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4919,
+      "step": 705
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4825,
+      "step": 706
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3302,
+      "step": 707
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4342,
+      "step": 708
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4476,
+      "step": 709
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.481,
+      "step": 710
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.727,
+      "step": 711
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7297,
+      "step": 712
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.51,
+      "step": 713
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4425,
+      "step": 714
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4809,
+      "step": 715
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6989,
+      "step": 716
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4121,
+      "step": 717
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5026,
+      "step": 718
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5721,
+      "step": 719
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3401,
+      "step": 720
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5194,
+      "step": 721
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3838,
+      "step": 722
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4309,
+      "step": 723
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6054,
+      "step": 724
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5241,
+      "step": 725
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4957,
+      "step": 726
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3414,
+      "step": 727
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6675,
+      "step": 728
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.2908,
+      "step": 729
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4647,
+      "step": 730
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5197,
+      "step": 731
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5692,
+      "step": 732
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.3174,
+      "step": 733
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.423,
+      "step": 734
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.3684,
+      "step": 735
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4731,
+      "step": 736
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4372,
+      "step": 737
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4442,
+      "step": 738
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4673,
+      "step": 739
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.5194,
+      "step": 740
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4319,
+      "step": 741
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4329,
+      "step": 742
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4507,
+      "step": 743
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.3279,
+      "step": 744
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.5465,
+      "step": 745
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.3553,
+      "step": 746
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.3262,
+      "step": 747
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4106,
+      "step": 748
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.2919,
+      "step": 749
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.5699,
+      "step": 750
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4896,
+      "step": 751
+    },
+    {
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.449,
-      "step": 360
+      "loss": 0.4367,
+      "step": 752
     },
     {
       "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4515,
-      "step": 370
+      "loss": 0.4654,
+      "step": 753
     },
     {
       "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4175,
-      "step": 380
+      "loss": 0.4172,
+      "step": 754
     },
     {
       "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.424,
-      "step": 390
+      "loss": 0.3156,
+      "step": 755
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.449,
-      "step": 400
+      "loss": 0.361,
+      "step": 756
     },
     {
-      "epoch": 0.16,
-      "eval_loss": 0.42175617814064026,
-      "eval_runtime": 125.6452,
-      "eval_samples_per_second": 7.959,
-      "eval_steps_per_second": 3.979,
-      "step": 400
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4922,
+      "step": 757
     },
     {
-      "epoch": 0.16,
-      "mmlu_eval_accuracy": 0.5963882087549879,
-      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
-      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
-      "mmlu_eval_accuracy_astronomy": 0.8125,
-      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
-      "mmlu_eval_accuracy_college_biology": 0.5625,
-      "mmlu_eval_accuracy_college_chemistry": 0.25,
-      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
-      "mmlu_eval_accuracy_conceptual_physics": 0.5,
-      "mmlu_eval_accuracy_econometrics": 0.5,
-      "mmlu_eval_accuracy_electrical_engineering": 0.6875,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
-      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
-      "mmlu_eval_accuracy_global_facts": 0.3,
-      "mmlu_eval_accuracy_high_school_biology": 0.59375,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
-      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5581395348837209,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.6153846153846154,
-      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
-      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
-      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
-      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
-      "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
-      "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
-      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
-      "mmlu_eval_accuracy_management": 0.9090909090909091,
-      "mmlu_eval_accuracy_marketing": 0.92,
-      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
-      "mmlu_eval_accuracy_moral_disputes": 0.5789473684210527,
-      "mmlu_eval_accuracy_moral_scenarios": 0.36,
-      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
-      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
-      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
-      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
-      "mmlu_eval_accuracy_professional_law": 0.4235294117647059,
-      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
-      "mmlu_eval_accuracy_professional_psychology": 0.6086956521739131,
-      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
-      "mmlu_eval_accuracy_security_studies": 0.7407407407407407,
-      "mmlu_eval_accuracy_sociology": 0.9545454545454546,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
-      "mmlu_eval_accuracy_virology": 0.5555555555555556,
-      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
-      "mmlu_loss": 1.4338379978511102,
-      "step": 400
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.44,
+      "step": 758
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.435,
-      "step": 410
+      "loss": 0.4128,
+      "step": 759
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.403,
-      "step": 420
+      "loss": 0.491,
+      "step": 760
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4161,
-      "step": 430
+      "loss": 0.4232,
+      "step": 761
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4319,
-      "step": 440
+      "loss": 0.4672,
+      "step": 762
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.391,
-      "step": 450
+      "loss": 0.5472,
+      "step": 763
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4213,
-      "step": 460
+      "loss": 0.3946,
+      "step": 764
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4129,
-      "step": 470
+      "loss": 0.5746,
+      "step": 765
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4105,
-      "step": 480
+      "loss": 0.4713,
+      "step": 766
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.3956,
-      "step": 490
+      "loss": 0.3048,
+      "step": 767
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.3874,
-      "step": 500
+      "loss": 0.4637,
+      "step": 768
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4208,
-      "step": 510
+      "loss": 0.4536,
+      "step": 769
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4048,
-      "step": 520
+      "loss": 0.3456,
+      "step": 770
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4252,
-      "step": 530
+      "loss": 0.2991,
+      "step": 771
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.452,
-      "step": 540
+      "loss": 0.3439,
+      "step": 772
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4386,
-      "step": 550
+      "loss": 0.5462,
+      "step": 773
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.43,
-      "step": 560
+      "loss": 0.4574,
+      "step": 774
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4565,
-      "step": 570
+      "loss": 0.3621,
+      "step": 775
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.435,
-      "step": 580
+      "loss": 0.428,
+      "step": 776
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.3766,
-      "step": 590
+      "loss": 0.5527,
+      "step": 777
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4068,
-      "step": 600
+      "loss": 0.3384,
+      "step": 778
     },
     {
-      "epoch": 0.24,
-      "eval_loss": 0.4149494469165802,
-      "eval_runtime": 125.7,
-      "eval_samples_per_second": 7.955,
-      "eval_steps_per_second": 3.978,
-      "step": 600
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.3637,
+      "step": 779
     },
     {
-      "epoch": 0.24,
-      "mmlu_eval_accuracy": 0.5931101560244482,
-      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
-      "mmlu_eval_accuracy_anatomy": 0.5,
-      "mmlu_eval_accuracy_astronomy": 0.8125,
-      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
-      "mmlu_eval_accuracy_college_biology": 0.625,
-      "mmlu_eval_accuracy_college_chemistry": 0.375,
-      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
-      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
-      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
-      "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
-      "mmlu_eval_accuracy_electrical_engineering": 0.625,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
-      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
-      "mmlu_eval_accuracy_global_facts": 0.3,
-      "mmlu_eval_accuracy_high_school_biology": 0.5625,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
-      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5581395348837209,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
-      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
-      "mmlu_eval_accuracy_high_school_psychology": 0.9,
-      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
-      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
-      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
-      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
-      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
-      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
-      "mmlu_eval_accuracy_management": 0.8181818181818182,
-      "mmlu_eval_accuracy_marketing": 0.92,
-      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
-      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
-      "mmlu_eval_accuracy_moral_scenarios": 0.35,
-      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
-      "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
-      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
-      "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
-      "mmlu_eval_accuracy_professional_law": 0.37058823529411766,
-      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
-      "mmlu_eval_accuracy_professional_psychology": 0.6086956521739131,
-      "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.7037037037037037,
-      "mmlu_eval_accuracy_sociology": 0.8636363636363636,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
-      "mmlu_eval_accuracy_virology": 0.5,
-      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
-      "mmlu_loss": 1.5072792833452124,
-      "step": 600
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.4353,
+      "step": 780
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4224,
-      "step": 610
+      "loss": 0.276,
+      "step": 781
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4038,
-      "step": 620
+      "loss": 0.4257,
+      "step": 782
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.4231,
-      "step": 630
+      "loss": 0.5785,
+      "step": 783
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.3935,
-      "step": 640
+      "loss": 0.7029,
+      "step": 784
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.15,
       "learning_rate": 0.0002,
-      "loss": 0.3924,
-      "step": 650
+      "loss": 0.488,
+      "step": 785
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.3875,
-      "step": 660
+      "loss": 0.3961,
+      "step": 786
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.4067,
-      "step": 670
+      "loss": 0.5306,
+      "step": 787
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.437,
-      "step": 680
+      "loss": 0.8475,
+      "step": 788
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.4481,
-      "step": 690
+      "loss": 0.4883,
+      "step": 789
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.3913,
-      "step": 700
+      "loss": 0.3324,
+      "step": 790
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.4536,
-      "step": 710
+      "loss": 0.3676,
+      "step": 791
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.4141,
-      "step": 720
+      "loss": 0.4738,
+      "step": 792
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.4181,
-      "step": 730
+      "loss": 0.3699,
+      "step": 793
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.3946,
-      "step": 740
+      "loss": 0.416,
+      "step": 794
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.3701,
-      "step": 750
+      "loss": 0.5871,
+      "step": 795
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.4371,
-      "step": 760
+      "loss": 0.4392,
+      "step": 796
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.3872,
-      "step": 770
+      "loss": 0.4721,
+      "step": 797
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.4033,
-      "step": 780
+      "loss": 0.4375,
+      "step": 798
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.3858,
-      "step": 790
+      "loss": 0.4206,
+      "step": 799
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.16,
       "learning_rate": 0.0002,
-      "loss": 0.3983,
+      "loss": 0.3377,
       "step": 800
     },
     {
-      "epoch": 0.32,
-      "eval_loss": 0.4103236794471741,
-      "eval_runtime": 125.6915,
-      "eval_samples_per_second": 7.956,
-      "eval_steps_per_second": 3.978,
+      "epoch": 0.16,
+      "eval_loss": 0.43044719099998474,
+      "eval_runtime": 126.0386,
+      "eval_samples_per_second": 7.934,
+      "eval_steps_per_second": 3.967,
       "step": 800
     },
     {
-      "epoch": 0.32,
-      "mmlu_eval_accuracy": 0.5903333803276282,
-      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
-      "mmlu_eval_accuracy_anatomy": 0.5,
+      "epoch": 0.16,
+      "mmlu_eval_accuracy": 0.5967065990739967,
+      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
       "mmlu_eval_accuracy_astronomy": 0.6875,
       "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
-      "mmlu_eval_accuracy_college_biology": 0.625,
-      "mmlu_eval_accuracy_college_chemistry": 0.25,
-      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
-      "mmlu_eval_accuracy_college_mathematics": 0.6363636363636364,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
+      "mmlu_eval_accuracy_college_biology": 0.5625,
+      "mmlu_eval_accuracy_college_chemistry": 0.375,
+      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
       "mmlu_eval_accuracy_college_medicine": 0.5909090909090909,
       "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
-      "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
-      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
-      "mmlu_eval_accuracy_econometrics": 0.5,
-      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
-      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
+      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
+      "mmlu_eval_accuracy_conceptual_physics": 0.5,
+      "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
+      "mmlu_eval_accuracy_electrical_engineering": 0.6875,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
+      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
       "mmlu_eval_accuracy_global_facts": 0.3,
-      "mmlu_eval_accuracy_high_school_biology": 0.625,
+      "mmlu_eval_accuracy_high_school_biology": 0.5625,
       "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
       "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
       "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
       "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
       "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
       "mmlu_eval_accuracy_high_school_macroeconomics": 0.5581395348837209,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.5384615384615384,
-      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
+      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
       "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
-      "mmlu_eval_accuracy_high_school_statistics": 0.30434782608695654,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
       "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
+      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
+      "mmlu_eval_accuracy_human_sexuality": 0.5,
       "mmlu_eval_accuracy_international_law": 0.8461538461538461,
-      "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
-      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
-      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
+      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
       "mmlu_eval_accuracy_management": 0.9090909090909091,
-      "mmlu_eval_accuracy_marketing": 0.96,
+      "mmlu_eval_accuracy_marketing": 0.92,
       "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7093023255813954,
+      "mmlu_eval_accuracy_miscellaneous": 0.7674418604651163,
       "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
-      "mmlu_eval_accuracy_moral_scenarios": 0.42,
-      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
-      "mmlu_eval_accuracy_philosophy": 0.7058823529411765,
+      "mmlu_eval_accuracy_moral_scenarios": 0.34,
+      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
+      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
       "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
-      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
-      "mmlu_eval_accuracy_professional_law": 0.38823529411764707,
-      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
-      "mmlu_eval_accuracy_professional_psychology": 0.5652173913043478,
-      "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.6666666666666666,
-      "mmlu_eval_accuracy_sociology": 0.9545454545454546,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
+      "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
+      "mmlu_eval_accuracy_professional_law": 0.40588235294117647,
+      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
+      "mmlu_eval_accuracy_professional_psychology": 0.5942028985507246,
+      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+      "mmlu_eval_accuracy_security_studies": 0.7037037037037037,
+      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
       "mmlu_eval_accuracy_virology": 0.5555555555555556,
-      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
-      "mmlu_loss": 1.4657139905750285,
+      "mmlu_eval_accuracy_world_religions": 0.9473684210526315,
+      "mmlu_loss": 1.47119294208117,
       "step": 800
     }
   ],
-  "logging_steps": 10,
-  "max_steps": 7596,
+  "logging_steps": 1,
+  "max_steps": 15195,
   "num_train_epochs": 3,
   "save_steps": 200,
-  "total_flos": 3.737597728456704e+17,
+  "total_flos": 1.473988082934743e+17,
   "trial_name": null,
   "trial_params": null
 }