diff --git "a/checkpoint-1200/trainer_state.json" "b/checkpoint-1200/trainer_state.json"
--- "a/checkpoint-1200/trainer_state.json"
+++ "b/checkpoint-1200/trainer_state.json"
@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.7805766463279724,
+  "best_metric": 0.8088108897209167,
   "best_model_checkpoint": "experts/mistralic-expert-0/checkpoint-1200",
-  "epoch": 0.43053188626782674,
+  "epoch": 0.21527559761402879,
   "eval_steps": 200,
   "global_step": 1200,
   "is_hyper_param_search": false,
@@ -11,1155 +11,7635 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.8779,
+      "loss": 0.5264,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.829,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.2202,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7436,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.9333,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8569,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.0676,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.681,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.0742,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.9134,
       "step": 10
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7617,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8938,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7277,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8794,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.6638,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.9485,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8506,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7225,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.8476,
+      "loss": 0.9012,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.2047,
       "step": 20
     },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.6835,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8377,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.0735,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.6917,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8773,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7732,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.9571,
+      "step": 27
+    },
     {
       "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8747,
+      "loss": 0.8992,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6006,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7753,
       "step": 30
     },
     {
       "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8234,
-      "step": 40
+      "loss": 0.8963,
+      "step": 31
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8913,
-      "step": 50
+      "loss": 0.7482,
+      "step": 32
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8766,
-      "step": 60
+      "loss": 0.9586,
+      "step": 33
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7964,
-      "step": 70
+      "loss": 0.9084,
+      "step": 34
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8225,
-      "step": 80
+      "loss": 0.7464,
+      "step": 35
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7962,
-      "step": 90
+      "loss": 0.8636,
+      "step": 36
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8032,
-      "step": 100
+      "loss": 0.7518,
+      "step": 37
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.759,
-      "step": 110
+      "loss": 0.8126,
+      "step": 38
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8303,
-      "step": 120
+      "loss": 0.8262,
+      "step": 39
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7749,
-      "step": 130
+      "loss": 0.744,
+      "step": 40
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8305,
-      "step": 140
+      "loss": 0.5329,
+      "step": 41
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7629,
-      "step": 150
+      "loss": 0.7157,
+      "step": 42
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7844,
-      "step": 160
+      "loss": 0.6823,
+      "step": 43
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8626,
-      "step": 170
+      "loss": 0.6387,
+      "step": 44
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8084,
-      "step": 180
+      "loss": 0.9449,
+      "step": 45
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8335,
-      "step": 190
+      "loss": 0.9559,
+      "step": 46
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8401,
-      "step": 200
+      "loss": 0.879,
+      "step": 47
     },
     {
-      "epoch": 0.07,
-      "eval_loss": 0.7983039617538452,
-      "eval_runtime": 119.4534,
-      "eval_samples_per_second": 8.371,
-      "eval_steps_per_second": 4.186,
-      "step": 200
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.9656,
+      "step": 48
     },
     {
-      "epoch": 0.07,
-      "mmlu_eval_accuracy": 0.5960849689925285,
-      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
-      "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": 0.5909090909090909,
-      "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": 0.5,
-      "mmlu_eval_accuracy_electrical_engineering": 0.5,
-      "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": 0.5625,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
-      "mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
-      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
-      "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": 0.4782608695652174,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
-      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
-      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
-      "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
-      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
-      "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
-      "mmlu_eval_accuracy_management": 0.9090909090909091,
-      "mmlu_eval_accuracy_marketing": 0.88,
-      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7906976744186046,
-      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
-      "mmlu_eval_accuracy_moral_scenarios": 0.31,
-      "mmlu_eval_accuracy_nutrition": 0.7575757575757576,
-      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
-      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
-      "mmlu_eval_accuracy_professional_accounting": 0.5806451612903226,
-      "mmlu_eval_accuracy_professional_law": 0.4,
-      "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": 0.6296296296296297,
-      "mmlu_eval_accuracy_sociology": 0.8636363636363636,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
-      "mmlu_eval_accuracy_virology": 0.5,
-      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
-      "mmlu_loss": 1.2690573895335509,
-      "step": 200
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8203,
+      "step": 49
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8515,
-      "step": 210
+      "loss": 0.9455,
+      "step": 50
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8544,
-      "step": 220
+      "loss": 1.0853,
+      "step": 51
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7662,
-      "step": 230
+      "loss": 0.784,
+      "step": 52
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8548,
-      "step": 240
+      "loss": 0.9081,
+      "step": 53
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.8204,
-      "step": 250
+      "loss": 0.942,
+      "step": 54
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7702,
-      "step": 260
+      "loss": 0.7809,
+      "step": 55
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7709,
-      "step": 270
+      "loss": 0.7745,
+      "step": 56
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.01,
       "learning_rate": 0.0002,
-      "loss": 0.7584,
-      "step": 280
+      "loss": 0.6383,
+      "step": 57
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.2732,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8463,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5995,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8898,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.728,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6242,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.0022,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.909,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.9811,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8681,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.87,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6002,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7864,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.9208,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6811,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.9932,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7263,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8073,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6364,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8184,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7538,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5079,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8374,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.2674,
+      "step": 81
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6385,
+      "step": 82
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8123,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.0097,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8071,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7729,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8635,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8271,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8811,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8141,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8279,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.013,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.0494,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8627,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8335,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9933,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9113,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5713,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.13,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.724,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9892,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9359,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8802,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.2836,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8343,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6938,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.2152,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.0564,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8405,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9903,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5582,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.072,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7454,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7892,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9739,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8423,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8456,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7781,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9245,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7775,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8333,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9091,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7332,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.0164,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7347,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8395,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.1135,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.0103,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7041,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7141,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.785,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7197,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6388,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7803,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6539,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6457,
+      "step": 136
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7665,
+      "step": 137
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8242,
+      "step": 138
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7452,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7132,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7869,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8337,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6804,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 1.1292,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9949,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9154,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8191,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7942,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.74,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9031,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8092,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6302,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6542,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8605,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6395,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7311,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8241,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8396,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9689,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7742,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.984,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6432,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6375,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4838,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8588,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 1.249,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8495,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7619,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9125,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7421,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7001,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.749,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8362,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6294,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.4927,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7113,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8189,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9025,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7475,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5455,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8963,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7447,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 1.1451,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8427,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 1.0241,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6195,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8103,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7378,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6879,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.679,
+      "step": 190
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6441,
+      "step": 191
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9548,
+      "step": 192
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8897,
+      "step": 193
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6438,
+      "step": 194
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8215,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5883,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6588,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9593,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.869,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9656,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "eval_loss": 0.819197952747345,
+      "eval_runtime": 120.2102,
+      "eval_samples_per_second": 8.319,
+      "eval_steps_per_second": 4.159,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "mmlu_eval_accuracy": 0.598886069565307,
+      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
+      "mmlu_eval_accuracy_anatomy": 0.5,
+      "mmlu_eval_accuracy_astronomy": 0.8125,
+      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
+      "mmlu_eval_accuracy_college_biology": 0.6875,
+      "mmlu_eval_accuracy_college_chemistry": 0.25,
+      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_medicine": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+      "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
+      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
+      "mmlu_eval_accuracy_econometrics": 0.5,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.4878048780487805,
+      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
+      "mmlu_eval_accuracy_global_facts": 0.4,
+      "mmlu_eval_accuracy_high_school_biology": 0.59375,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
+      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.6538461538461539,
+      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
+      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
+      "mmlu_eval_accuracy_high_school_us_history": 0.8181818181818182,
+      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.5,
+      "mmlu_eval_accuracy_international_law": 1.0,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
+      "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
+      "mmlu_eval_accuracy_management": 0.9090909090909091,
+      "mmlu_eval_accuracy_marketing": 0.88,
+      "mmlu_eval_accuracy_medical_genetics": 1.0,
+      "mmlu_eval_accuracy_miscellaneous": 0.7325581395348837,
+      "mmlu_eval_accuracy_moral_disputes": 0.631578947368421,
+      "mmlu_eval_accuracy_moral_scenarios": 0.36,
+      "mmlu_eval_accuracy_nutrition": 0.7878787878787878,
+      "mmlu_eval_accuracy_philosophy": 0.7058823529411765,
+      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
+      "mmlu_eval_accuracy_professional_accounting": 0.5806451612903226,
+      "mmlu_eval_accuracy_professional_law": 0.4235294117647059,
+      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
+      "mmlu_eval_accuracy_professional_psychology": 0.6231884057971014,
+      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_security_studies": 0.6296296296296297,
+      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
+      "mmlu_eval_accuracy_virology": 0.4444444444444444,
+      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
+      "mmlu_loss": 1.1897795199958836,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8691,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6466,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8482,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7146,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.827,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.786,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7581,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6526,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7893,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6353,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6407,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8251,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6403,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7787,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8152,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9103,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7739,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.875,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4974,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8326,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7268,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8219,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8256,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8176,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8006,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7994,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7726,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9446,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8695,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8484,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 1.0515,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6001,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8815,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7518,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7089,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 1.0574,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.732,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 1.0021,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8112,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8332,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6818,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 1.0446,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6754,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8357,
+      "step": 244
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5055,
+      "step": 245
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7735,
+      "step": 246
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8911,
+      "step": 247
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.685,
+      "step": 248
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6654,
+      "step": 249
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7916,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6564,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6345,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8852,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8113,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5569,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6621,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8455,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7835,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7047,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8994,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.9058,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7971,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.9496,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7594,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7886,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6538,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6809,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8414,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8776,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8006,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6896,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8714,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7161,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7769,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7746,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7707,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8898,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.4789,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7176,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.9641,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6647,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.9271,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6533,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8499,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6656,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8976,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7257,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7333,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7764,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8432,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6872,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7407,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7672,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7499,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6575,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8137,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8963,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8385,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7339,
+      "step": 299
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5774,
+      "step": 300
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7619,
+      "step": 301
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7487,
+      "step": 302
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5988,
+      "step": 303
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8261,
+      "step": 304
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7291,
+      "step": 305
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7985,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.834,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6934,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5891,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5622,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8949,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9118,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5318,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8857,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6523,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5213,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.0887,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8514,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8407,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6914,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7363,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6723,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9866,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6549,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.1392,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8666,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9718,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6719,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7478,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9245,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7089,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.819,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9994,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8888,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9954,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6815,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.881,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.89,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.687,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8041,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8261,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.937,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7216,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7856,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5445,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8157,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9742,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9235,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7944,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9934,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9002,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7322,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6911,
+      "step": 353
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.0813,
+      "step": 354
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6879,
+      "step": 355
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8331,
+      "step": 356
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5794,
+      "step": 357
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9645,
+      "step": 358
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.022,
+      "step": 359
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7269,
+      "step": 360
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6452,
+      "step": 361
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.1992,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7354,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.795,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8158,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.589,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6763,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8345,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7371,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7679,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 1.0635,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9093,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9003,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8673,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9434,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7443,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.83,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8583,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7994,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8135,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 1.0633,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.942,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8432,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9047,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 1.0228,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6272,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.954,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.93,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.57,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7213,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.864,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6982,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.773,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8143,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6744,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7048,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6799,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7449,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8773,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9398,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 0.8146390914916992,
+      "eval_runtime": 119.9921,
+      "eval_samples_per_second": 8.334,
+      "eval_steps_per_second": 4.167,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "mmlu_eval_accuracy": 0.5840454378937088,
+      "mmlu_eval_accuracy_abstract_algebra": 0.45454545454545453,
+      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
+      "mmlu_eval_accuracy_astronomy": 0.6875,
+      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+      "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": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
+      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
+      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+      "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
+      "mmlu_eval_accuracy_conceptual_physics": 0.46153846153846156,
+      "mmlu_eval_accuracy_econometrics": 0.5,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5,
+      "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": 0.5625,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.45454545454545453,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
+      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
+      "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": 0.11764705882352941,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
+      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
+      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
+      "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": 1.0,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
+      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
+      "mmlu_eval_accuracy_management": 0.9090909090909091,
+      "mmlu_eval_accuracy_marketing": 0.8,
+      "mmlu_eval_accuracy_medical_genetics": 1.0,
+      "mmlu_eval_accuracy_miscellaneous": 0.7790697674418605,
+      "mmlu_eval_accuracy_moral_disputes": 0.5789473684210527,
+      "mmlu_eval_accuracy_moral_scenarios": 0.36,
+      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
+      "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
+      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+      "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
+      "mmlu_eval_accuracy_professional_law": 0.3764705882352941,
+      "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": 0.6296296296296297,
+      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
+      "mmlu_eval_accuracy_virology": 0.3888888888888889,
+      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
+      "mmlu_loss": 1.2478325493030697,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8952,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6421,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9271,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8853,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6216,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9962,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5766,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.764,
+      "step": 408
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7164,
+      "step": 409
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8913,
+      "step": 410
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8564,
+      "step": 411
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6592,
+      "step": 412
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6042,
+      "step": 413
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8746,
+      "step": 414
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9486,
+      "step": 415
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8353,
+      "step": 416
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9095,
+      "step": 417
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6699,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 1.1071,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.983,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5656,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.675,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8796,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9583,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7425,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7113,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8578,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7023,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.977,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7728,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.701,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5901,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8082,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7467,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6743,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.967,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9099,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 1.1272,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6706,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.843,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8487,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7673,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7837,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8561,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9924,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.636,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8629,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5707,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8511,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4633,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9191,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7401,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8498,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7354,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7749,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9347,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5879,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9931,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7215,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9971,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 1.0129,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 1.0919,
+      "step": 462
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 1.1395,
+      "step": 463
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8581,
+      "step": 464
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7299,
+      "step": 465
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.749,
+      "step": 466
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9947,
+      "step": 467
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7373,
+      "step": 468
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7911,
+      "step": 469
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8132,
+      "step": 470
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6938,
+      "step": 471
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6997,
+      "step": 472
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.646,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.1959,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7402,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6113,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7298,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.688,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.0373,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9351,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8599,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7164,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.818,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.0831,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9243,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8152,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6372,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7632,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6665,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9449,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7717,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9482,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5271,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8563,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6312,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.0792,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.736,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.0319,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9515,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8827,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.4783,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8084,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8713,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7058,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6528,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6743,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7397,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6268,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.0668,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7789,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8351,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7186,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9317,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5945,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7909,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7789,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9191,
+      "step": 517
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5431,
+      "step": 518
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6632,
+      "step": 519
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6069,
+      "step": 520
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.525,
+      "step": 521
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8939,
+      "step": 522
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7299,
+      "step": 523
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8089,
+      "step": 524
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7111,
+      "step": 525
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9181,
+      "step": 526
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.768,
+      "step": 527
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6898,
+      "step": 528
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7758,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.95,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8559,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9042,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8297,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 1.0169,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8675,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4781,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.631,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6513,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6692,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6927,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8807,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8045,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6948,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 1.2426,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 1.0208,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6158,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6377,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7656,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9128,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7771,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5964,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7297,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.682,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7522,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6163,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9973,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7769,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9604,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6804,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.561,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9593,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.789,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7274,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9649,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8848,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6318,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5717,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7621,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7481,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4239,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.77,
+      "step": 571
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.728,
+      "step": 572
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6687,
+      "step": 573
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 1.0965,
+      "step": 574
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8505,
+      "step": 575
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7421,
+      "step": 576
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7896,
+      "step": 577
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6582,
+      "step": 578
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8823,
+      "step": 579
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6455,
+      "step": 580
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9429,
+      "step": 581
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8702,
+      "step": 582
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9492,
+      "step": 583
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7109,
+      "step": 584
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6594,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6978,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.682,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.137,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7552,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0373,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7309,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9709,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0436,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6451,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7466,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9419,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7417,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7272,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7524,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6241,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "eval_loss": 0.8122884035110474,
+      "eval_runtime": 120.1018,
+      "eval_samples_per_second": 8.326,
+      "eval_steps_per_second": 4.163,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "mmlu_eval_accuracy": 0.5775488236324069,
+      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
+      "mmlu_eval_accuracy_astronomy": 0.625,
+      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+      "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": 0.2727272727272727,
+      "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": 0.45454545454545453,
+      "mmlu_eval_accuracy_conceptual_physics": 0.5,
+      "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
+      "mmlu_eval_accuracy_electrical_engineering": 0.75,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
+      "mmlu_eval_accuracy_formal_logic": 0.21428571428571427,
+      "mmlu_eval_accuracy_global_facts": 0.4,
+      "mmlu_eval_accuracy_high_school_biology": 0.53125,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
+      "mmlu_eval_accuracy_high_school_european_history": 0.8333333333333334,
+      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5348837209302325,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
+      "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": 0.6923076923076923,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
+      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
+      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
+      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
+      "mmlu_eval_accuracy_management": 0.8181818181818182,
+      "mmlu_eval_accuracy_marketing": 0.88,
+      "mmlu_eval_accuracy_medical_genetics": 1.0,
+      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
+      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
+      "mmlu_eval_accuracy_moral_scenarios": 0.31,
+      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
+      "mmlu_eval_accuracy_philosophy": 0.7058823529411765,
+      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
+      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
+      "mmlu_eval_accuracy_professional_law": 0.40588235294117647,
+      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
+      "mmlu_eval_accuracy_professional_psychology": 0.5797101449275363,
+      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_security_studies": 0.6296296296296297,
+      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
+      "mmlu_eval_accuracy_virology": 0.5,
+      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
+      "mmlu_loss": 1.155314199373554,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6356,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6931,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6434,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7497,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0151,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7799,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4129,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6497,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8673,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8972,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7414,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8824,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8458,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5602,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.1145,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.751,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.91,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6223,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9187,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8892,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7446,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8344,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8258,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0045,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7711,
+      "step": 625
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9652,
+      "step": 626
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.977,
+      "step": 627
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.975,
+      "step": 628
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9071,
+      "step": 629
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0192,
+      "step": 630
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8535,
+      "step": 631
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7861,
+      "step": 632
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6559,
+      "step": 633
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8987,
+      "step": 634
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9118,
+      "step": 635
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9167,
+      "step": 636
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9752,
+      "step": 637
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.9444,
+      "step": 638
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0146,
+      "step": 639
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7298,
+      "step": 640
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.97,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8458,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6694,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5434,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5527,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.0355,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6512,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9389,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6727,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5559,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6009,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.2096,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7077,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8533,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.0063,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6545,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6986,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8059,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5942,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9657,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8902,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8335,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7972,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5325,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6723,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7587,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.2527,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9213,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7374,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6813,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7271,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8961,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8716,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4987,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8746,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.2122,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6908,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5627,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.843,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.4994,
+      "step": 680
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6439,
+      "step": 681
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7117,
+      "step": 682
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9306,
+      "step": 683
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7992,
+      "step": 684
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9131,
+      "step": 685
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8099,
+      "step": 686
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.5372,
+      "step": 687
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8114,
+      "step": 688
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8681,
+      "step": 689
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.001,
+      "step": 690
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8004,
+      "step": 691
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.0169,
+      "step": 692
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9716,
+      "step": 693
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8436,
+      "step": 694
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8383,
+      "step": 695
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9273,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.805,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.591,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7372,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6843,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 1.0615,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.4435,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 1.0096,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9595,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8328,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6296,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.536,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 1.1206,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7323,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7574,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8945,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6381,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9567,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7999,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8028,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5545,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8406,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9554,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9843,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9497,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 1.0865,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9369,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8747,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5984,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.725,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7256,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8308,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 1.0842,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7355,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6329,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7386,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8152,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8898,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9807,
+      "step": 734
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.548,
+      "step": 735
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6874,
+      "step": 736
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9956,
+      "step": 737
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7104,
+      "step": 738
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5394,
+      "step": 739
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7863,
+      "step": 740
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9701,
+      "step": 741
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6934,
+      "step": 742
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.645,
+      "step": 743
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7102,
+      "step": 744
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7523,
+      "step": 745
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9494,
+      "step": 746
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.881,
+      "step": 747
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7908,
+      "step": 748
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6808,
+      "step": 749
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8387,
+      "step": 750
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8542,
+      "step": 751
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9962,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6898,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.77,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8921,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 1.0765,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7215,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8062,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7367,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7853,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8128,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6854,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 1.0334,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6495,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8573,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7062,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6943,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9436,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8105,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8338,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8709,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7766,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8329,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9291,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8712,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5845,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8188,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4564,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9452,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9851,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7869,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6011,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7872,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9304,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8963,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 1.0458,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6291,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6855,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4715,
+      "step": 789
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6295,
+      "step": 790
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6677,
+      "step": 791
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5447,
+      "step": 792
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7778,
+      "step": 793
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7773,
+      "step": 794
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7345,
+      "step": 795
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8149,
+      "step": 796
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6858,
+      "step": 797
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8848,
+      "step": 798
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8424,
+      "step": 799
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6627,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.8102672696113586,
+      "eval_runtime": 120.0261,
+      "eval_samples_per_second": 8.332,
+      "eval_steps_per_second": 4.166,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "mmlu_eval_accuracy": 0.5822729953967423,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.35714285714285715,
+      "mmlu_eval_accuracy_astronomy": 0.5625,
+      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
+      "mmlu_eval_accuracy_college_biology": 0.6875,
+      "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": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
+      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+      "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+      "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": 0.5,
+      "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": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5813953488372093,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.6153846153846154,
+      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
+      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+      "mmlu_eval_accuracy_high_school_us_history": 0.8181818181818182,
+      "mmlu_eval_accuracy_high_school_world_history": 0.6153846153846154,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
+      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
+      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+      "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
+      "mmlu_eval_accuracy_management": 0.9090909090909091,
+      "mmlu_eval_accuracy_marketing": 0.88,
+      "mmlu_eval_accuracy_medical_genetics": 1.0,
+      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
+      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
+      "mmlu_eval_accuracy_moral_scenarios": 0.38,
+      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
+      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
+      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+      "mmlu_eval_accuracy_professional_accounting": 0.5806451612903226,
+      "mmlu_eval_accuracy_professional_law": 0.4,
+      "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": 0.5555555555555556,
+      "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": 1.24667573788303,
+      "step": 800
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7349,
+      "step": 801
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8926,
+      "step": 802
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8905,
+      "step": 803
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6114,
+      "step": 804
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9096,
+      "step": 805
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8915,
+      "step": 806
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6685,
+      "step": 807
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.736,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6852,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.0443,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6871,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9018,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8111,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9451,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8058,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.0257,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8692,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6026,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7625,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6635,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7603,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9596,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7564,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.0032,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.831,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7135,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8368,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6973,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7404,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8406,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9669,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9976,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8777,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6272,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7334,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9365,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8458,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8249,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8132,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9527,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7713,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7706,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7875,
+      "step": 843
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6717,
+      "step": 844
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7642,
+      "step": 845
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.75,
+      "step": 846
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6557,
+      "step": 847
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8093,
+      "step": 848
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.1117,
+      "step": 849
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8126,
+      "step": 850
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6141,
+      "step": 851
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.216,
+      "step": 852
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.72,
+      "step": 853
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9137,
+      "step": 854
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.029,
+      "step": 855
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6722,
+      "step": 856
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6809,
+      "step": 857
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6048,
+      "step": 858
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7776,
+      "step": 859
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6477,
+      "step": 860
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.5878,
+      "step": 861
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8157,
+      "step": 862
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.0297,
+      "step": 863
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9082,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8636,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9638,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6518,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9081,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7937,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7313,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7721,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.0205,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.5584,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.0349,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.0456,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9087,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.0519,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8639,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7572,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9824,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.5098,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9407,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6657,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6396,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8067,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6769,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.5499,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9525,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7992,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8248,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8118,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8176,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.0351,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7835,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9283,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8568,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.939,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6814,
+      "step": 898
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7237,
+      "step": 899
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7062,
+      "step": 900
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7082,
+      "step": 901
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.0502,
+      "step": 902
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7895,
+      "step": 903
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7635,
+      "step": 904
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8228,
+      "step": 905
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8439,
+      "step": 906
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8659,
+      "step": 907
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.3819,
+      "step": 908
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7967,
+      "step": 909
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7517,
+      "step": 910
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.1273,
+      "step": 911
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.0413,
+      "step": 912
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6612,
+      "step": 913
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.4956,
+      "step": 914
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7759,
+      "step": 915
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.848,
+      "step": 916
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8788,
+      "step": 917
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 1.0118,
+      "step": 918
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9246,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8681,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6518,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8517,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8253,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7183,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6525,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7671,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6801,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9189,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9451,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8237,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 1.0214,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9972,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9342,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8488,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8945,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9811,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9587,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7897,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.793,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7522,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7604,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 1.0987,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6051,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7157,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5994,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7723,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.997,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7381,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7236,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 1.1055,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8061,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8286,
+      "step": 952
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9728,
+      "step": 953
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6722,
+      "step": 954
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6802,
+      "step": 955
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7981,
+      "step": 956
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7531,
+      "step": 957
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8957,
+      "step": 958
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8334,
+      "step": 959
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7717,
+      "step": 960
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8412,
+      "step": 961
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5571,
+      "step": 962
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6741,
+      "step": 963
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.605,
+      "step": 964
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8041,
+      "step": 965
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5841,
+      "step": 966
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7498,
+      "step": 967
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9486,
+      "step": 968
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 1.0114,
+      "step": 969
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6379,
+      "step": 970
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6591,
+      "step": 971
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8938,
+      "step": 972
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6101,
+      "step": 973
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8045,
+      "step": 974
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7324,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7788,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.617,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6762,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 1.0155,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7408,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8656,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7208,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9636,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7963,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.5973,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7131,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.5622,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7594,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8865,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8393,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.685,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9986,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.5276,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8268,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9159,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7032,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.745,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6313,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 1.0211,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7744,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "eval_loss": 0.8099237084388733,
+      "eval_runtime": 120.1218,
+      "eval_samples_per_second": 8.325,
+      "eval_steps_per_second": 4.162,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "mmlu_eval_accuracy": 0.5733945104551192,
+      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+      "mmlu_eval_accuracy_astronomy": 0.625,
+      "mmlu_eval_accuracy_business_ethics": 0.7272727272727273,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
+      "mmlu_eval_accuracy_college_biology": 0.5,
+      "mmlu_eval_accuracy_college_chemistry": 0.25,
+      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_medicine": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
+      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
+      "mmlu_eval_accuracy_conceptual_physics": 0.5,
+      "mmlu_eval_accuracy_econometrics": 0.3333333333333333,
+      "mmlu_eval_accuracy_electrical_engineering": 0.625,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
+      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
+      "mmlu_eval_accuracy_global_facts": 0.2,
+      "mmlu_eval_accuracy_high_school_biology": 0.5,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_european_history": 0.8888888888888888,
+      "mmlu_eval_accuracy_high_school_geography": 0.9090909090909091,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5116279069767442,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
+      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8,
+      "mmlu_eval_accuracy_high_school_statistics": 0.5217391304347826,
+      "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": 0.6666666666666666,
+      "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": 0.45454545454545453,
+      "mmlu_eval_accuracy_management": 0.9090909090909091,
+      "mmlu_eval_accuracy_marketing": 0.84,
+      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
+      "mmlu_eval_accuracy_miscellaneous": 0.7093023255813954,
+      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
+      "mmlu_eval_accuracy_moral_scenarios": 0.29,
+      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
+      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
+      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
+      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
+      "mmlu_eval_accuracy_professional_law": 0.4117647058823529,
+      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
+      "mmlu_eval_accuracy_professional_psychology": 0.5942028985507246,
+      "mmlu_eval_accuracy_public_relations": 0.3333333333333333,
+      "mmlu_eval_accuracy_security_studies": 0.5925925925925926,
+      "mmlu_eval_accuracy_sociology": 0.8636363636363636,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
+      "mmlu_eval_accuracy_virology": 0.5555555555555556,
+      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
+      "mmlu_loss": 1.233580312475212,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9884,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6461,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7407,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9994,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.4772,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6857,
+      "step": 1006
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7761,
+      "step": 1007
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.5209,
+      "step": 1008
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.5138,
+      "step": 1009
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6956,
+      "step": 1010
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8676,
+      "step": 1011
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.647,
+      "step": 1012
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9174,
+      "step": 1013
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7508,
+      "step": 1014
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.5988,
+      "step": 1015
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7933,
+      "step": 1016
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7837,
+      "step": 1017
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9667,
+      "step": 1018
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7731,
+      "step": 1019
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8228,
+      "step": 1020
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6162,
+      "step": 1021
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9746,
+      "step": 1022
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 1.0127,
+      "step": 1023
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 1.1577,
+      "step": 1024
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7684,
+      "step": 1025
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.5803,
+      "step": 1026
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8425,
+      "step": 1027
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8765,
+      "step": 1028
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 1.4075,
+      "step": 1029
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8019,
+      "step": 1030
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7039,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6683,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7524,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.0455,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.9615,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7047,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7846,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.106,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.2221,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8795,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7115,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.9716,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.1039,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.764,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8904,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8369,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.9216,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6592,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.938,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6673,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.821,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8687,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6174,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6096,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8766,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7078,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7375,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7975,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.741,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6121,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.715,
+      "step": 1061
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.66,
+      "step": 1062
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.877,
+      "step": 1063
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.697,
+      "step": 1064
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7048,
+      "step": 1065
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.0353,
+      "step": 1066
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6105,
+      "step": 1067
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7002,
+      "step": 1068
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5125,
+      "step": 1069
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5618,
+      "step": 1070
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5144,
+      "step": 1071
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8734,
+      "step": 1072
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8636,
+      "step": 1073
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6496,
+      "step": 1074
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.1773,
+      "step": 1075
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.825,
+      "step": 1076
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.9007,
+      "step": 1077
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7527,
+      "step": 1078
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8812,
+      "step": 1079
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.881,
+      "step": 1080
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8933,
+      "step": 1081
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5951,
+      "step": 1082
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6243,
+      "step": 1083
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6716,
+      "step": 1084
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.9126,
+      "step": 1085
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5971,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.9161,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7937,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6131,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7779,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7351,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7336,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7224,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6206,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6429,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.934,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7564,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8437,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7004,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8927,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7831,
-      "step": 290
+      "loss": 0.6849,
+      "step": 1101
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8402,
-      "step": 300
+      "loss": 0.9785,
+      "step": 1102
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7957,
-      "step": 310
+      "loss": 1.1522,
+      "step": 1103
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8712,
-      "step": 320
+      "loss": 0.7161,
+      "step": 1104
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7969,
-      "step": 330
+      "loss": 0.6762,
+      "step": 1105
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7617,
-      "step": 340
+      "loss": 0.5125,
+      "step": 1106
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8363,
-      "step": 350
+      "loss": 0.9352,
+      "step": 1107
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8466,
-      "step": 360
+      "loss": 0.739,
+      "step": 1108
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8057,
-      "step": 370
+      "loss": 0.8282,
+      "step": 1109
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8214,
-      "step": 380
+      "loss": 1.192,
+      "step": 1110
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.812,
-      "step": 390
+      "loss": 0.8171,
+      "step": 1111
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7665,
-      "step": 400
+      "loss": 1.1589,
+      "step": 1112
     },
     {
-      "epoch": 0.14,
-      "eval_loss": 0.7917793393135071,
-      "eval_runtime": 119.0164,
-      "eval_samples_per_second": 8.402,
-      "eval_steps_per_second": 4.201,
-      "step": 400
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6615,
+      "step": 1113
     },
     {
-      "epoch": 0.14,
-      "mmlu_eval_accuracy": 0.5963053096671925,
-      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
-      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
-      "mmlu_eval_accuracy_astronomy": 0.75,
-      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
-      "mmlu_eval_accuracy_college_biology": 0.5625,
-      "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": 0.5454545454545454,
-      "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": 0.5,
-      "mmlu_eval_accuracy_electrical_engineering": 0.6875,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
-      "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": 0.4090909090909091,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
-      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.627906976744186,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.3448275862068966,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.6923076923076923,
-      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8333333333333334,
-      "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": 0.9230769230769231,
-      "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": 0.7674418604651163,
-      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
-      "mmlu_eval_accuracy_moral_scenarios": 0.31,
-      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
-      "mmlu_eval_accuracy_philosophy": 0.7647058823529411,
-      "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
-      "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
-      "mmlu_eval_accuracy_professional_law": 0.40588235294117647,
-      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
-      "mmlu_eval_accuracy_professional_psychology": 0.6231884057971014,
-      "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.5925925925925926,
-      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
-      "mmlu_eval_accuracy_virology": 0.5,
-      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
-      "mmlu_loss": 1.4978953666848862,
-      "step": 400
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.9842,
+      "step": 1114
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8155,
-      "step": 410
+      "loss": 0.878,
+      "step": 1115
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8196,
-      "step": 420
+      "loss": 0.8462,
+      "step": 1116
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7908,
-      "step": 430
+      "loss": 0.8861,
+      "step": 1117
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.9,
-      "step": 440
+      "loss": 0.6377,
+      "step": 1118
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8223,
-      "step": 450
+      "loss": 1.0122,
+      "step": 1119
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8565,
-      "step": 460
+      "loss": 0.7613,
+      "step": 1120
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8335,
-      "step": 470
+      "loss": 0.8743,
+      "step": 1121
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8257,
-      "step": 480
+      "loss": 0.9053,
+      "step": 1122
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7703,
-      "step": 490
+      "loss": 0.8195,
+      "step": 1123
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7781,
-      "step": 500
+      "loss": 0.6983,
+      "step": 1124
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7646,
-      "step": 510
+      "loss": 0.679,
+      "step": 1125
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8506,
-      "step": 520
+      "loss": 0.831,
+      "step": 1126
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7818,
-      "step": 530
+      "loss": 0.7893,
+      "step": 1127
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7922,
-      "step": 540
+      "loss": 0.9428,
+      "step": 1128
     },
     {
       "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7783,
-      "step": 550
+      "loss": 0.7274,
+      "step": 1129
     },
     {
       "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8566,
-      "step": 560
+      "loss": 0.3981,
+      "step": 1130
     },
     {
       "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8016,
-      "step": 570
+      "loss": 0.8938,
+      "step": 1131
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8165,
-      "step": 580
+      "loss": 0.68,
+      "step": 1132
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7599,
-      "step": 590
+      "loss": 0.762,
+      "step": 1133
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.768,
-      "step": 600
+      "loss": 1.0052,
+      "step": 1134
     },
     {
-      "epoch": 0.22,
-      "eval_loss": 0.7908002734184265,
-      "eval_runtime": 119.1725,
-      "eval_samples_per_second": 8.391,
-      "eval_steps_per_second": 4.196,
-      "step": 600
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.635,
+      "step": 1135
     },
     {
-      "epoch": 0.22,
-      "mmlu_eval_accuracy": 0.5754860642925688,
-      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
-      "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": 0.5625,
-      "mmlu_eval_accuracy_college_chemistry": 0.25,
-      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
-      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
-      "mmlu_eval_accuracy_college_medicine": 0.5909090909090909,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "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": 0.4375,
-      "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": 0.59375,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
-      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
-      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
-      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
-      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
-      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
-      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
-      "mmlu_eval_accuracy_international_law": 1.0,
-      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
-      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
-      "mmlu_eval_accuracy_machine_learning": 0.6363636363636364,
-      "mmlu_eval_accuracy_management": 0.8181818181818182,
-      "mmlu_eval_accuracy_marketing": 0.92,
-      "mmlu_eval_accuracy_medical_genetics": 1.0,
-      "mmlu_eval_accuracy_miscellaneous": 0.7674418604651163,
-      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
-      "mmlu_eval_accuracy_moral_scenarios": 0.36,
-      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
-      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
-      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
-      "mmlu_eval_accuracy_professional_accounting": 0.5806451612903226,
-      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
-      "mmlu_eval_accuracy_professional_medicine": 0.5161290322580645,
-      "mmlu_eval_accuracy_professional_psychology": 0.5507246376811594,
-      "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.5185185185185185,
-      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
-      "mmlu_eval_accuracy_virology": 0.5555555555555556,
-      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
-      "mmlu_loss": 1.22548696083593,
-      "step": 600
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 1.014,
+      "step": 1136
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7712,
-      "step": 610
+      "loss": 0.7173,
+      "step": 1137
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7589,
-      "step": 620
+      "loss": 0.5893,
+      "step": 1138
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8275,
-      "step": 630
+      "loss": 1.1284,
+      "step": 1139
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8018,
-      "step": 640
+      "loss": 0.9059,
+      "step": 1140
     },
     {
-      "epoch": 0.23,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.7783,
-      "step": 650
+      "loss": 0.7536,
+      "step": 1141
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.2,
       "learning_rate": 0.0002,
-      "loss": 0.8121,
-      "step": 660
+      "loss": 0.6125,
+      "step": 1142
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8181,
-      "step": 670
+      "loss": 0.7951,
+      "step": 1143
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7267,
-      "step": 680
+      "loss": 0.741,
+      "step": 1144
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8174,
-      "step": 690
+      "loss": 0.8551,
+      "step": 1145
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7569,
-      "step": 700
+      "loss": 0.6547,
+      "step": 1146
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8414,
-      "step": 710
+      "loss": 0.8885,
+      "step": 1147
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7667,
-      "step": 720
+      "loss": 0.5953,
+      "step": 1148
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7613,
-      "step": 730
+      "loss": 0.6163,
+      "step": 1149
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7888,
-      "step": 740
+      "loss": 0.9619,
+      "step": 1150
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8222,
-      "step": 750
+      "loss": 0.7336,
+      "step": 1151
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8171,
-      "step": 760
+      "loss": 0.7697,
+      "step": 1152
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8108,
-      "step": 770
+      "loss": 0.8312,
+      "step": 1153
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8173,
-      "step": 780
+      "loss": 0.8344,
+      "step": 1154
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7818,
-      "step": 790
+      "loss": 0.7553,
+      "step": 1155
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8018,
-      "step": 800
+      "loss": 1.0345,
+      "step": 1156
     },
     {
-      "epoch": 0.29,
-      "eval_loss": 0.7824653387069702,
-      "eval_runtime": 119.1607,
-      "eval_samples_per_second": 8.392,
-      "eval_steps_per_second": 4.196,
-      "step": 800
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.8302,
+      "step": 1157
     },
     {
-      "epoch": 0.29,
-      "mmlu_eval_accuracy": 0.5920750532808338,
-      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
-      "mmlu_eval_accuracy_anatomy": 0.5,
-      "mmlu_eval_accuracy_astronomy": 0.75,
-      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.6551724137931034,
-      "mmlu_eval_accuracy_college_biology": 0.625,
-      "mmlu_eval_accuracy_college_chemistry": 0.375,
-      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
-      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_medicine": 0.5909090909090909,
-      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
-      "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
-      "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
-      "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": 0.14285714285714285,
-      "mmlu_eval_accuracy_global_facts": 0.4,
-      "mmlu_eval_accuracy_high_school_biology": 0.59375,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
-      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.20689655172413793,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.6538461538461539,
-      "mmlu_eval_accuracy_high_school_physics": 0.23529411764705882,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8833333333333333,
-      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
-      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
-      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5,
-      "mmlu_eval_accuracy_international_law": 1.0,
-      "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
-      "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
-      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
-      "mmlu_eval_accuracy_management": 0.9090909090909091,
-      "mmlu_eval_accuracy_marketing": 0.88,
-      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7325581395348837,
-      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
-      "mmlu_eval_accuracy_moral_scenarios": 0.41,
-      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
-      "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
-      "mmlu_eval_accuracy_prehistory": 0.6,
-      "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
-      "mmlu_eval_accuracy_professional_law": 0.3588235294117647,
-      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
-      "mmlu_eval_accuracy_professional_psychology": 0.5652173913043478,
-      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
-      "mmlu_eval_accuracy_security_studies": 0.48148148148148145,
-      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
-      "mmlu_eval_accuracy_virology": 0.5,
-      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
-      "mmlu_loss": 1.2784218954666475,
-      "step": 800
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.9394,
+      "step": 1158
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8091,
-      "step": 810
+      "loss": 0.9296,
+      "step": 1159
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7372,
-      "step": 820
+      "loss": 0.9676,
+      "step": 1160
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8131,
-      "step": 830
+      "loss": 0.5795,
+      "step": 1161
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7849,
-      "step": 840
+      "loss": 1.0122,
+      "step": 1162
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7493,
-      "step": 850
+      "loss": 1.0379,
+      "step": 1163
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8526,
-      "step": 860
+      "loss": 0.9732,
+      "step": 1164
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8205,
-      "step": 870
+      "loss": 0.7463,
+      "step": 1165
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8178,
-      "step": 880
+      "loss": 0.6577,
+      "step": 1166
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7811,
-      "step": 890
+      "loss": 0.9141,
+      "step": 1167
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7714,
-      "step": 900
+      "loss": 0.6224,
+      "step": 1168
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7801,
-      "step": 910
+      "loss": 0.6755,
+      "step": 1169
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7892,
-      "step": 920
+      "loss": 0.7448,
+      "step": 1170
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7387,
-      "step": 930
+      "loss": 0.8902,
+      "step": 1171
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7883,
-      "step": 940
+      "loss": 0.6378,
+      "step": 1172
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8275,
-      "step": 950
+      "loss": 0.8829,
+      "step": 1173
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8036,
-      "step": 960
+      "loss": 0.6271,
+      "step": 1174
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8339,
-      "step": 970
+      "loss": 0.829,
+      "step": 1175
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8275,
-      "step": 980
+      "loss": 0.6688,
+      "step": 1176
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.807,
-      "step": 990
+      "loss": 0.7943,
+      "step": 1177
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7536,
-      "step": 1000
+      "loss": 0.7687,
+      "step": 1178
     },
     {
-      "epoch": 0.36,
-      "eval_loss": 0.7809625267982483,
-      "eval_runtime": 119.1269,
-      "eval_samples_per_second": 8.394,
-      "eval_steps_per_second": 4.197,
-      "step": 1000
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.574,
+      "step": 1179
     },
     {
-      "epoch": 0.36,
-      "mmlu_eval_accuracy": 0.5879325165236478,
-      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
-      "mmlu_eval_accuracy_anatomy": 0.5,
-      "mmlu_eval_accuracy_astronomy": 0.8125,
-      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.5517241379310345,
-      "mmlu_eval_accuracy_college_biology": 0.5,
-      "mmlu_eval_accuracy_college_chemistry": 0.375,
-      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
-      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
-      "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
-      "mmlu_eval_accuracy_econometrics": 0.5,
-      "mmlu_eval_accuracy_electrical_engineering": 0.5,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
-      "mmlu_eval_accuracy_formal_logic": 0.07142857142857142,
-      "mmlu_eval_accuracy_global_facts": 0.4,
-      "mmlu_eval_accuracy_high_school_biology": 0.53125,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.5555555555555556,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
-      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6190476190476191,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6511627906976745,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.7692307692307693,
-      "mmlu_eval_accuracy_high_school_physics": 0.29411764705882354,
-      "mmlu_eval_accuracy_high_school_psychology": 0.85,
-      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
-      "mmlu_eval_accuracy_high_school_us_history": 0.6363636363636364,
-      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5,
-      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
-      "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
-      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
-      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
-      "mmlu_eval_accuracy_management": 0.9090909090909091,
-      "mmlu_eval_accuracy_marketing": 0.88,
-      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
-      "mmlu_eval_accuracy_moral_disputes": 0.6052631578947368,
-      "mmlu_eval_accuracy_moral_scenarios": 0.38,
-      "mmlu_eval_accuracy_nutrition": 0.6363636363636364,
-      "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
-      "mmlu_eval_accuracy_prehistory": 0.6571428571428571,
-      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
-      "mmlu_eval_accuracy_professional_law": 0.38235294117647056,
-      "mmlu_eval_accuracy_professional_medicine": 0.6451612903225806,
-      "mmlu_eval_accuracy_professional_psychology": 0.5362318840579711,
-      "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.5925925925925926,
-      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
-      "mmlu_eval_accuracy_virology": 0.5,
-      "mmlu_eval_accuracy_world_religions": 0.9473684210526315,
-      "mmlu_loss": 1.3696499114792902,
-      "step": 1000
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7261,
+      "step": 1180
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7769,
-      "step": 1010
+      "loss": 0.7134,
+      "step": 1181
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7913,
-      "step": 1020
+      "loss": 1.0088,
+      "step": 1182
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7663,
-      "step": 1030
+      "loss": 0.6574,
+      "step": 1183
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8283,
-      "step": 1040
+      "loss": 0.5751,
+      "step": 1184
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7691,
-      "step": 1050
+      "loss": 0.4795,
+      "step": 1185
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8158,
-      "step": 1060
+      "loss": 0.4943,
+      "step": 1186
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8084,
-      "step": 1070
+      "loss": 0.7801,
+      "step": 1187
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8508,
-      "step": 1080
+      "loss": 0.6625,
+      "step": 1188
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.734,
-      "step": 1090
+      "loss": 0.9798,
+      "step": 1189
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7645,
-      "step": 1100
+      "loss": 0.8326,
+      "step": 1190
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8286,
-      "step": 1110
+      "loss": 0.8879,
+      "step": 1191
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7868,
-      "step": 1120
+      "loss": 0.8049,
+      "step": 1192
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8213,
-      "step": 1130
+      "loss": 0.9478,
+      "step": 1193
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.8405,
-      "step": 1140
+      "loss": 0.8146,
+      "step": 1194
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7747,
-      "step": 1150
+      "loss": 0.7259,
+      "step": 1195
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.784,
-      "step": 1160
+      "loss": 0.9866,
+      "step": 1196
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7624,
-      "step": 1170
+      "loss": 0.6743,
+      "step": 1197
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7944,
-      "step": 1180
+      "loss": 0.6869,
+      "step": 1198
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.8409,
-      "step": 1190
+      "loss": 0.8121,
+      "step": 1199
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.8438,
+      "loss": 0.6509,
       "step": 1200
     },
     {
-      "epoch": 0.43,
-      "eval_loss": 0.7805766463279724,
-      "eval_runtime": 119.1262,
-      "eval_samples_per_second": 8.394,
-      "eval_steps_per_second": 4.197,
+      "epoch": 0.22,
+      "eval_loss": 0.8088108897209167,
+      "eval_runtime": 120.0821,
+      "eval_samples_per_second": 8.328,
+      "eval_steps_per_second": 4.164,
       "step": 1200
     },
     {
-      "epoch": 0.43,
-      "mmlu_eval_accuracy": 0.5875597911670384,
+      "epoch": 0.22,
+      "mmlu_eval_accuracy": 0.5466414321349508,
       "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
       "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
-      "mmlu_eval_accuracy_astronomy": 0.8125,
+      "mmlu_eval_accuracy_astronomy": 0.625,
       "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.4827586206896552,
       "mmlu_eval_accuracy_college_biology": 0.5,
-      "mmlu_eval_accuracy_college_chemistry": 0.375,
-      "mmlu_eval_accuracy_college_computer_science": 0.36363636363636365,
-      "mmlu_eval_accuracy_college_mathematics": 0.2727272727272727,
-      "mmlu_eval_accuracy_college_medicine": 0.6818181818181818,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "mmlu_eval_accuracy_computer_security": 0.7272727272727273,
+      "mmlu_eval_accuracy_college_chemistry": 0.25,
+      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
+      "mmlu_eval_accuracy_college_mathematics": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_medicine": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
+      "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": 0.625,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
+      "mmlu_eval_accuracy_econometrics": 0.3333333333333333,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.34146341463414637,
       "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
-      "mmlu_eval_accuracy_global_facts": 0.5,
-      "mmlu_eval_accuracy_high_school_biology": 0.59375,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
+      "mmlu_eval_accuracy_global_facts": 0.3,
+      "mmlu_eval_accuracy_high_school_biology": 0.4375,
+      "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": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
+      "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
       "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5813953488372093,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
-      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
-      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5581395348837209,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.41379310344827586,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5384615384615384,
+      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
+      "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": 0.4166666666666667,
-      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
-      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
-      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
-      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
-      "mmlu_eval_accuracy_management": 0.9090909090909091,
-      "mmlu_eval_accuracy_marketing": 0.8,
-      "mmlu_eval_accuracy_medical_genetics": 1.0,
-      "mmlu_eval_accuracy_miscellaneous": 0.7674418604651163,
-      "mmlu_eval_accuracy_moral_disputes": 0.631578947368421,
-      "mmlu_eval_accuracy_moral_scenarios": 0.35,
-      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+      "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
+      "mmlu_eval_accuracy_management": 0.7272727272727273,
+      "mmlu_eval_accuracy_marketing": 0.84,
+      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
+      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
+      "mmlu_eval_accuracy_moral_disputes": 0.4473684210526316,
+      "mmlu_eval_accuracy_moral_scenarios": 0.29,
+      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
       "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
       "mmlu_eval_accuracy_prehistory": 0.6,
-      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
-      "mmlu_eval_accuracy_professional_law": 0.4117647058823529,
-      "mmlu_eval_accuracy_professional_medicine": 0.6129032258064516,
-      "mmlu_eval_accuracy_professional_psychology": 0.5797101449275363,
-      "mmlu_eval_accuracy_public_relations": 0.5,
+      "mmlu_eval_accuracy_professional_accounting": 0.5806451612903226,
+      "mmlu_eval_accuracy_professional_law": 0.3411764705882353,
+      "mmlu_eval_accuracy_professional_medicine": 0.4838709677419355,
+      "mmlu_eval_accuracy_professional_psychology": 0.5652173913043478,
+      "mmlu_eval_accuracy_public_relations": 0.4166666666666667,
       "mmlu_eval_accuracy_security_studies": 0.6296296296296297,
-      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
+      "mmlu_eval_accuracy_sociology": 0.7727272727272727,
       "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
-      "mmlu_eval_accuracy_virology": 0.3888888888888889,
-      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
-      "mmlu_loss": 1.2426614148187263,
+      "mmlu_eval_accuracy_virology": 0.5,
+      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
+      "mmlu_loss": 1.0881710131081215,
       "step": 1200
     }
   ],
-  "logging_steps": 10,
-  "max_steps": 8361,
+  "logging_steps": 1,
+  "max_steps": 16722,
   "num_train_epochs": 3,
   "save_steps": 200,
-  "total_flos": 5.469317861796741e+17,
+  "total_flos": 1.9460732365386547e+17,
   "trial_name": null,
   "trial_params": null
 }