diff --git "a/checkpoint-1000/trainer_state.json" "b/checkpoint-1000/trainer_state.json"
--- "a/checkpoint-1000/trainer_state.json"
+++ "b/checkpoint-1000/trainer_state.json"
@@ -1,7 +1,7 @@
 {
-  "best_metric": 0.692868709564209,
+  "best_metric": 0.7207491993904114,
   "best_model_checkpoint": "experts/mistralic-expert-9/checkpoint-1000",
-  "epoch": 0.48128985681626757,
+  "epoch": 0.2406594067745623,
   "eval_steps": 200,
   "global_step": 1000,
   "is_hyper_param_search": false,
@@ -11,964 +11,6364 @@
     {
       "epoch": 0.0,
       "learning_rate": 0.0002,
-      "loss": 0.7899,
+      "loss": 0.7829,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7089,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.652,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.0244,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 1.0769,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7341,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7328,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8798,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.6666,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8604,
       "step": 10
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8468,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.9661,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8119,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8655,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.5448,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.9035,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8323,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.6687,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.8933,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 0.0002,
+      "loss": 0.7238,
+      "step": 20
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7591,
+      "step": 21
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5128,
+      "step": 22
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.5734,
+      "step": 23
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.0502,
+      "step": 24
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.002,
+      "step": 25
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.798,
+      "step": 26
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7186,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.9112,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5919,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8125,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6634,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.0096,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.672,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7521,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5946,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.0072,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.939,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6419,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.0206,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6814,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8587,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6549,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7792,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7181,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.0225,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.761,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6052,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6887,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8332,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7334,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7373,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7164,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.6855,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.9398,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.687,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7259,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 1.0526,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7918,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8921,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.5324,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.8227,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 0.0002,
+      "loss": 0.7512,
+      "step": 62
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9117,
+      "step": 63
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7862,
+      "step": 64
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.9442,
+      "step": 65
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8861,
+      "step": 66
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6102,
+      "step": 67
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7185,
+      "step": 68
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8237,
+      "step": 69
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8127,
+      "step": 70
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7026,
+      "step": 71
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6567,
+      "step": 72
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7449,
+      "step": 73
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.0395,
+      "step": 74
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8453,
+      "step": 75
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.0422,
+      "step": 76
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4983,
+      "step": 77
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6663,
+      "step": 78
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5482,
+      "step": 79
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5009,
+      "step": 80
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.4778,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7916,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5971,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8567,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 1.0076,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7959,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8249,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7941,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.696,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6833,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.5947,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6119,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6495,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6445,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7122,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8692,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8587,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.805,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6772,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.8503,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7101,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.6113,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 0.0002,
+      "loss": 0.7861,
+      "step": 103
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6994,
+      "step": 104
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6874,
+      "step": 105
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8741,
+      "step": 106
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6024,
+      "step": 107
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7844,
+      "step": 108
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8048,
+      "step": 109
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8304,
+      "step": 110
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5302,
+      "step": 111
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.846,
+      "step": 112
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6849,
+      "step": 113
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7676,
+      "step": 114
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7423,
+      "step": 115
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7003,
+      "step": 116
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5765,
+      "step": 117
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8719,
+      "step": 118
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9269,
+      "step": 119
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7549,
+      "step": 120
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7172,
+      "step": 121
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7196,
+      "step": 122
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5534,
+      "step": 123
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6364,
+      "step": 124
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6204,
+      "step": 125
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 1.1158,
+      "step": 126
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8617,
+      "step": 127
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.673,
+      "step": 128
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5937,
+      "step": 129
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6191,
+      "step": 130
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.807,
+      "step": 131
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.5617,
+      "step": 132
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9628,
+      "step": 133
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.9681,
+      "step": 134
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8927,
+      "step": 135
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7657,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8324,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7427,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8727,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8147,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.6597,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.8342,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7009,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7304,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 0.0002,
+      "loss": 0.7918,
+      "step": 145
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7891,
+      "step": 146
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7903,
+      "step": 147
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7538,
+      "step": 148
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8071,
+      "step": 149
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6163,
+      "step": 150
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7015,
+      "step": 151
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6356,
+      "step": 152
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6419,
+      "step": 153
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.628,
+      "step": 154
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8761,
+      "step": 155
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9747,
+      "step": 156
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7915,
+      "step": 157
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5916,
+      "step": 158
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9699,
+      "step": 159
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.668,
+      "step": 160
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8413,
+      "step": 161
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9765,
+      "step": 162
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7141,
+      "step": 163
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6775,
+      "step": 164
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6601,
+      "step": 165
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6567,
+      "step": 166
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8919,
+      "step": 167
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6437,
+      "step": 168
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.5891,
+      "step": 169
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6067,
+      "step": 170
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7152,
+      "step": 171
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7658,
+      "step": 172
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.659,
+      "step": 173
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7332,
+      "step": 174
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9071,
+      "step": 175
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.4358,
+      "step": 176
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7819,
+      "step": 177
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.6895,
+      "step": 178
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9041,
+      "step": 179
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 1.1299,
+      "step": 180
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 1.0473,
+      "step": 181
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9748,
+      "step": 182
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.9271,
+      "step": 183
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.7196,
+      "step": 184
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8309,
+      "step": 185
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 0.0002,
+      "loss": 0.8785,
+      "step": 186
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6239,
+      "step": 187
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8092,
+      "step": 188
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7551,
+      "step": 189
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8115,
+      "step": 190
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7745,
+      "step": 191
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 1.0405,
+      "step": 192
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7509,
+      "step": 193
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8147,
+      "step": 194
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6865,
+      "step": 195
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6052,
+      "step": 196
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7591,
+      "step": 197
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8932,
+      "step": 198
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.618,
+      "step": 199
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6768,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 0.7412691712379456,
+      "eval_runtime": 134.0567,
+      "eval_samples_per_second": 7.46,
+      "eval_steps_per_second": 3.73,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "mmlu_eval_accuracy": 0.5472120959224429,
+      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
+      "mmlu_eval_accuracy_anatomy": 0.5,
+      "mmlu_eval_accuracy_astronomy": 0.5625,
+      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.4482758620689655,
+      "mmlu_eval_accuracy_college_biology": 0.625,
+      "mmlu_eval_accuracy_college_chemistry": 0.125,
+      "mmlu_eval_accuracy_college_computer_science": 0.2727272727272727,
+      "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
+      "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": 0.5384615384615384,
+      "mmlu_eval_accuracy_econometrics": 0.4166666666666667,
+      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.4146341463414634,
+      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
+      "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": 0.7777777777777778,
+      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5581395348837209,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5,
+      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
+      "mmlu_eval_accuracy_high_school_psychology": 0.7666666666666667,
+      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_world_history": 0.5769230769230769,
+      "mmlu_eval_accuracy_human_aging": 0.8695652173913043,
+      "mmlu_eval_accuracy_human_sexuality": 0.4166666666666667,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.36363636363636365,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6111111111111112,
+      "mmlu_eval_accuracy_machine_learning": 0.5454545454545454,
+      "mmlu_eval_accuracy_management": 0.8181818181818182,
+      "mmlu_eval_accuracy_marketing": 0.88,
+      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
+      "mmlu_eval_accuracy_miscellaneous": 0.7209302325581395,
+      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
+      "mmlu_eval_accuracy_moral_scenarios": 0.29,
+      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
+      "mmlu_eval_accuracy_philosophy": 0.5882352941176471,
+      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+      "mmlu_eval_accuracy_professional_accounting": 0.6451612903225806,
+      "mmlu_eval_accuracy_professional_law": 0.3352941176470588,
+      "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": 0.5555555555555556,
+      "mmlu_eval_accuracy_sociology": 0.8636363636363636,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.9090909090909091,
+      "mmlu_eval_accuracy_virology": 0.6111111111111112,
+      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
+      "mmlu_loss": 1.4129784670184238,
+      "step": 200
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7578,
+      "step": 201
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6877,
+      "step": 202
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8905,
+      "step": 203
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7928,
+      "step": 204
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.5133,
+      "step": 205
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7105,
+      "step": 206
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7523,
+      "step": 207
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6032,
+      "step": 208
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.817,
+      "step": 209
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7819,
+      "step": 210
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 1.1033,
+      "step": 211
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.9117,
+      "step": 212
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8523,
+      "step": 213
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.756,
+      "step": 214
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8372,
+      "step": 215
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.9113,
+      "step": 216
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7574,
+      "step": 217
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7985,
+      "step": 218
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6361,
+      "step": 219
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8262,
+      "step": 220
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8068,
+      "step": 221
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7409,
+      "step": 222
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.6916,
+      "step": 223
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.9771,
+      "step": 224
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7918,
+      "step": 225
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7358,
+      "step": 226
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.7509,
+      "step": 227
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 0.0002,
+      "loss": 0.8409,
+      "step": 228
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6204,
+      "step": 229
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7744,
+      "step": 230
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7642,
+      "step": 231
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7039,
+      "step": 232
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7506,
+      "step": 233
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8357,
+      "step": 234
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.173,
+      "step": 235
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7943,
+      "step": 236
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8563,
+      "step": 237
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.815,
+      "step": 238
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6474,
+      "step": 239
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8408,
+      "step": 240
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9206,
+      "step": 241
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7378,
+      "step": 242
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7542,
+      "step": 243
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7606,
+      "step": 244
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6862,
+      "step": 245
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6237,
+      "step": 246
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5309,
+      "step": 247
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6157,
+      "step": 248
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5884,
+      "step": 249
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7231,
+      "step": 250
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.9123,
+      "step": 251
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5973,
+      "step": 252
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7957,
+      "step": 253
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6449,
+      "step": 254
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6614,
+      "step": 255
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6381,
+      "step": 256
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.6559,
+      "step": 257
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.833,
+      "step": 258
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.0383,
+      "step": 259
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7461,
+      "step": 260
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8383,
+      "step": 261
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 1.0236,
+      "step": 262
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7598,
+      "step": 263
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.5967,
+      "step": 264
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.7097,
+      "step": 265
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.709,
+      "step": 266
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.682,
+      "step": 267
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8359,
+      "step": 268
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.68,
+      "step": 269
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 0.0002,
+      "loss": 0.8403,
+      "step": 270
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6073,
+      "step": 271
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6612,
+      "step": 272
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5965,
+      "step": 273
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9327,
+      "step": 274
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6308,
+      "step": 275
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7039,
+      "step": 276
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5699,
+      "step": 277
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8565,
+      "step": 278
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7331,
+      "step": 279
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7344,
+      "step": 280
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7181,
+      "step": 281
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.619,
+      "step": 282
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9035,
+      "step": 283
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8835,
+      "step": 284
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6835,
+      "step": 285
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6284,
+      "step": 286
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7191,
+      "step": 287
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9702,
+      "step": 288
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 1.0929,
+      "step": 289
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9881,
+      "step": 290
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8347,
+      "step": 291
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4855,
+      "step": 292
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6767,
+      "step": 293
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9992,
+      "step": 294
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7851,
+      "step": 295
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5559,
+      "step": 296
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7087,
+      "step": 297
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8397,
+      "step": 298
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6175,
+      "step": 299
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.4592,
+      "step": 300
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6005,
+      "step": 301
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6443,
+      "step": 302
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.9034,
+      "step": 303
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8548,
+      "step": 304
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7751,
+      "step": 305
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.5758,
+      "step": 306
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7318,
+      "step": 307
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8113,
+      "step": 308
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.7886,
+      "step": 309
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.8755,
+      "step": 310
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 0.0002,
+      "loss": 0.6733,
+      "step": 311
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8691,
+      "step": 312
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7446,
+      "step": 313
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7937,
+      "step": 314
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8296,
+      "step": 315
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7961,
+      "step": 316
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6128,
+      "step": 317
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7549,
+      "step": 318
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6768,
+      "step": 319
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7734,
+      "step": 320
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4684,
+      "step": 321
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9625,
+      "step": 322
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7095,
+      "step": 323
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7619,
+      "step": 324
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6717,
+      "step": 325
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6586,
+      "step": 326
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6572,
+      "step": 327
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.4487,
+      "step": 328
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5975,
+      "step": 329
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6366,
+      "step": 330
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6788,
+      "step": 331
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8823,
+      "step": 332
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7312,
+      "step": 333
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6469,
+      "step": 334
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5019,
+      "step": 335
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.913,
+      "step": 336
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7615,
+      "step": 337
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5602,
+      "step": 338
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.8117,
+      "step": 339
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.709,
+      "step": 340
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.5675,
+      "step": 341
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6191,
+      "step": 342
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7517,
+      "step": 343
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9037,
+      "step": 344
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7339,
+      "step": 345
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7992,
+      "step": 346
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.9029,
+      "step": 347
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7139,
+      "step": 348
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7248,
+      "step": 349
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6524,
+      "step": 350
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.6524,
+      "step": 351
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.7101,
+      "step": 352
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 0.0002,
+      "loss": 0.847,
+      "step": 353
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.0317,
+      "step": 354
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 1.0285,
+      "step": 355
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7928,
+      "step": 356
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.626,
+      "step": 357
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.669,
+      "step": 358
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5556,
+      "step": 359
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9189,
+      "step": 360
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6982,
+      "step": 361
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5555,
+      "step": 362
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5817,
+      "step": 363
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7223,
+      "step": 364
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8566,
+      "step": 365
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6357,
+      "step": 366
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7608,
+      "step": 367
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6095,
+      "step": 368
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7246,
+      "step": 369
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8225,
+      "step": 370
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8044,
+      "step": 371
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5546,
+      "step": 372
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8517,
+      "step": 373
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7983,
+      "step": 374
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5217,
+      "step": 375
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5817,
+      "step": 376
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7111,
+      "step": 377
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8437,
+      "step": 378
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7774,
+      "step": 379
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7384,
+      "step": 380
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8405,
+      "step": 381
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6979,
+      "step": 382
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7938,
+      "step": 383
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6571,
+      "step": 384
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.9623,
+      "step": 385
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7444,
+      "step": 386
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.52,
+      "step": 387
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6312,
+      "step": 388
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7149,
+      "step": 389
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7371,
+      "step": 390
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.8836,
+      "step": 391
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.6378,
+      "step": 392
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.7229,
+      "step": 393
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 0.0002,
+      "loss": 0.5179,
+      "step": 394
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6813,
+      "step": 395
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6754,
+      "step": 396
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6578,
+      "step": 397
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6735,
+      "step": 398
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9044,
+      "step": 399
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6774,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "eval_loss": 0.7310672402381897,
+      "eval_runtime": 134.0222,
+      "eval_samples_per_second": 7.461,
+      "eval_steps_per_second": 3.731,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "mmlu_eval_accuracy": 0.5847840681976317,
+      "mmlu_eval_accuracy_abstract_algebra": 0.09090909090909091,
+      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
+      "mmlu_eval_accuracy_astronomy": 0.6875,
+      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.6206896551724138,
+      "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": 0.5454545454545454,
+      "mmlu_eval_accuracy_college_medicine": 0.6818181818181818,
+      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
+      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
+      "mmlu_eval_accuracy_conceptual_physics": 0.5,
+      "mmlu_eval_accuracy_econometrics": 0.5,
+      "mmlu_eval_accuracy_electrical_engineering": 0.6875,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.36585365853658536,
+      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
+      "mmlu_eval_accuracy_global_facts": 0.4,
+      "mmlu_eval_accuracy_high_school_biology": 0.65625,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
+      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.8095238095238095,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.3793103448275862,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.6538461538461539,
+      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
+      "mmlu_eval_accuracy_high_school_psychology": 0.85,
+      "mmlu_eval_accuracy_high_school_statistics": 0.34782608695652173,
+      "mmlu_eval_accuracy_high_school_us_history": 0.6818181818181818,
+      "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": 0.8461538461538461,
+      "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": 0.9090909090909091,
+      "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
+      "mmlu_eval_accuracy_moral_disputes": 0.5789473684210527,
+      "mmlu_eval_accuracy_moral_scenarios": 0.32,
+      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
+      "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
+      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
+      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
+      "mmlu_eval_accuracy_professional_law": 0.4176470588235294,
+      "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": 0.6666666666666666,
+      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
+      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
+      "mmlu_eval_accuracy_virology": 0.5555555555555556,
+      "mmlu_eval_accuracy_world_religions": 0.7368421052631579,
+      "mmlu_loss": 1.0065513858790498,
+      "step": 400
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6637,
+      "step": 401
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6684,
+      "step": 402
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6889,
+      "step": 403
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 1.195,
+      "step": 404
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6737,
+      "step": 405
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4768,
+      "step": 406
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.4604,
+      "step": 407
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8305,
+      "step": 408
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8372,
+      "step": 409
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7863,
+      "step": 410
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9165,
+      "step": 411
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7748,
+      "step": 412
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6335,
+      "step": 413
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.83,
+      "step": 414
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5789,
+      "step": 415
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7305,
+      "step": 416
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.749,
+      "step": 417
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8968,
+      "step": 418
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5709,
+      "step": 419
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7709,
+      "step": 420
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 1.0298,
+      "step": 421
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9443,
+      "step": 422
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7035,
+      "step": 423
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8436,
+      "step": 424
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7382,
+      "step": 425
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6467,
+      "step": 426
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6999,
+      "step": 427
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8816,
+      "step": 428
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5861,
+      "step": 429
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7559,
+      "step": 430
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6926,
+      "step": 431
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.5772,
+      "step": 432
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.9274,
+      "step": 433
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.8483,
+      "step": 434
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.7913,
+      "step": 435
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 0.0002,
+      "loss": 0.6109,
+      "step": 436
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0144,
+      "step": 437
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7146,
+      "step": 438
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7158,
+      "step": 439
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0971,
+      "step": 440
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5999,
+      "step": 441
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5126,
+      "step": 442
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6365,
+      "step": 443
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8523,
+      "step": 444
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7006,
+      "step": 445
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8474,
+      "step": 446
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6532,
+      "step": 447
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6658,
+      "step": 448
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8371,
+      "step": 449
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8525,
+      "step": 450
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6598,
+      "step": 451
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6407,
+      "step": 452
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7606,
+      "step": 453
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8324,
+      "step": 454
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6646,
+      "step": 455
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 1.0582,
+      "step": 456
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5981,
+      "step": 457
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6321,
+      "step": 458
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.551,
+      "step": 459
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8432,
+      "step": 460
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8593,
+      "step": 461
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8263,
+      "step": 462
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8807,
+      "step": 463
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.779,
+      "step": 464
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6577,
+      "step": 465
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7737,
+      "step": 466
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7346,
+      "step": 467
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6813,
+      "step": 468
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6482,
+      "step": 469
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.5621,
+      "step": 470
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7516,
+      "step": 471
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6925,
+      "step": 472
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.6258,
+      "step": 473
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.4763,
+      "step": 474
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7208,
+      "step": 475
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.8338,
+      "step": 476
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 0.0002,
+      "loss": 0.7907,
+      "step": 477
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6233,
+      "step": 478
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.0559,
+      "step": 479
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6443,
+      "step": 480
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7538,
+      "step": 481
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.752,
+      "step": 482
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6144,
+      "step": 483
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6633,
+      "step": 484
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8243,
+      "step": 485
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.708,
+      "step": 486
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6421,
+      "step": 487
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6249,
+      "step": 488
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5536,
+      "step": 489
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7006,
+      "step": 490
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5877,
+      "step": 491
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.638,
+      "step": 492
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5669,
+      "step": 493
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5445,
+      "step": 494
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6924,
+      "step": 495
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8554,
+      "step": 496
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7551,
+      "step": 497
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8665,
+      "step": 498
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7166,
+      "step": 499
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6724,
+      "step": 500
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8199,
+      "step": 501
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7342,
+      "step": 502
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 1.0662,
+      "step": 503
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6987,
+      "step": 504
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.781,
+      "step": 505
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8262,
+      "step": 506
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7811,
+      "step": 507
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6871,
+      "step": 508
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9329,
+      "step": 509
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6715,
+      "step": 510
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7512,
+      "step": 511
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6138,
+      "step": 512
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.6114,
+      "step": 513
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9257,
+      "step": 514
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.7834,
+      "step": 515
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.5774,
+      "step": 516
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.66,
+      "step": 517
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.8416,
+      "step": 518
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 0.0002,
+      "loss": 0.9217,
+      "step": 519
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8815,
+      "step": 520
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.589,
+      "step": 521
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.843,
+      "step": 522
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8886,
+      "step": 523
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7577,
+      "step": 524
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.749,
+      "step": 525
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6833,
+      "step": 526
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9316,
+      "step": 527
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7154,
+      "step": 528
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6197,
+      "step": 529
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6586,
+      "step": 530
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7995,
+      "step": 531
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6799,
+      "step": 532
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7078,
+      "step": 533
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8105,
+      "step": 534
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8578,
+      "step": 535
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8691,
+      "step": 536
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8643,
+      "step": 537
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6235,
+      "step": 538
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.8933,
+      "step": 539
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7673,
+      "step": 540
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9723,
+      "step": 541
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7709,
+      "step": 542
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6197,
+      "step": 543
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.758,
+      "step": 544
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.649,
+      "step": 545
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7469,
+      "step": 546
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7217,
+      "step": 547
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6997,
+      "step": 548
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6798,
+      "step": 549
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.6293,
+      "step": 550
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7419,
+      "step": 551
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.747,
+      "step": 552
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.796,
+      "step": 553
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7065,
+      "step": 554
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9859,
+      "step": 555
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.7094,
+      "step": 556
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.724,
+      "step": 557
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 1.0979,
+      "step": 558
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.9745,
+      "step": 559
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 0.0002,
+      "loss": 0.5665,
+      "step": 560
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6547,
+      "step": 561
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8175,
+      "step": 562
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6868,
+      "step": 563
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5751,
+      "step": 564
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8083,
+      "step": 565
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8233,
+      "step": 566
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 1.1842,
+      "step": 567
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 1.0463,
+      "step": 568
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8303,
+      "step": 569
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6358,
+      "step": 570
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.577,
+      "step": 571
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9896,
+      "step": 572
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8385,
+      "step": 573
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7269,
+      "step": 574
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7405,
+      "step": 575
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6531,
+      "step": 576
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5727,
+      "step": 577
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8353,
+      "step": 578
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8271,
+      "step": 579
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6666,
+      "step": 580
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6781,
+      "step": 581
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9931,
+      "step": 582
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6298,
+      "step": 583
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8553,
+      "step": 584
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.4881,
+      "step": 585
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7209,
+      "step": 586
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7328,
+      "step": 587
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9516,
+      "step": 588
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7076,
+      "step": 589
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.9411,
+      "step": 590
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5285,
+      "step": 591
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7789,
+      "step": 592
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6815,
+      "step": 593
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7924,
+      "step": 594
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.8039,
+      "step": 595
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7215,
+      "step": 596
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.5125,
+      "step": 597
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.774,
+      "step": 598
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7092,
+      "step": 599
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 1.0468,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 0.7286259531974792,
+      "eval_runtime": 134.2534,
+      "eval_samples_per_second": 7.449,
+      "eval_steps_per_second": 3.724,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "mmlu_eval_accuracy": 0.5932734327709844,
+      "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": 0.75,
+      "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": 0.6818181818181818,
+      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
+      "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": 0.14285714285714285,
+      "mmlu_eval_accuracy_global_facts": 0.5,
+      "mmlu_eval_accuracy_high_school_biology": 0.71875,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_european_history": 0.7222222222222222,
+      "mmlu_eval_accuracy_high_school_geography": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6511627906976745,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.27586206896551724,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.6153846153846154,
+      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
+      "mmlu_eval_accuracy_high_school_psychology": 0.85,
+      "mmlu_eval_accuracy_high_school_statistics": 0.5217391304347826,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
+      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+      "mmlu_eval_accuracy_logical_fallacies": 0.7222222222222222,
+      "mmlu_eval_accuracy_machine_learning": 0.45454545454545453,
+      "mmlu_eval_accuracy_management": 0.8181818181818182,
+      "mmlu_eval_accuracy_marketing": 0.84,
+      "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": 0.31,
+      "mmlu_eval_accuracy_nutrition": 0.8484848484848485,
+      "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
+      "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
+      "mmlu_eval_accuracy_professional_accounting": 0.6129032258064516,
+      "mmlu_eval_accuracy_professional_law": 0.3941176470588235,
+      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
+      "mmlu_eval_accuracy_professional_psychology": 0.6086956521739131,
+      "mmlu_eval_accuracy_public_relations": 0.5833333333333334,
+      "mmlu_eval_accuracy_security_studies": 0.6296296296296297,
+      "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": 0.8421052631578947,
+      "mmlu_loss": 1.2674363679191774,
+      "step": 600
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.6655,
+      "step": 601
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 0.0002,
+      "loss": 0.7498,
+      "step": 602
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8131,
+      "step": 603
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6184,
+      "step": 604
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6295,
+      "step": 605
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6954,
+      "step": 606
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.653,
+      "step": 607
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6874,
+      "step": 608
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7817,
+      "step": 609
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.618,
+      "step": 610
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6373,
+      "step": 611
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6485,
+      "step": 612
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.5799,
+      "step": 613
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6766,
+      "step": 614
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8426,
+      "step": 615
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8768,
+      "step": 616
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9871,
+      "step": 617
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7622,
+      "step": 618
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7522,
+      "step": 619
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.763,
+      "step": 620
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6306,
+      "step": 621
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.826,
+      "step": 622
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6045,
+      "step": 623
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8409,
+      "step": 624
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8099,
+      "step": 625
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8063,
+      "step": 626
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7533,
+      "step": 627
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7232,
+      "step": 628
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7204,
+      "step": 629
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.0063,
+      "step": 630
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7806,
+      "step": 631
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6797,
+      "step": 632
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9628,
+      "step": 633
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.8439,
+      "step": 634
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.725,
+      "step": 635
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7112,
+      "step": 636
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6966,
+      "step": 637
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 1.1543,
+      "step": 638
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9062,
+      "step": 639
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.9612,
+      "step": 640
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6007,
+      "step": 641
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.7896,
+      "step": 642
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6859,
+      "step": 643
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 0.0002,
+      "loss": 0.6354,
+      "step": 644
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8166,
+      "step": 645
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7521,
+      "step": 646
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6965,
+      "step": 647
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7455,
+      "step": 648
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6067,
+      "step": 649
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7609,
+      "step": 650
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9064,
+      "step": 651
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8769,
+      "step": 652
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6537,
+      "step": 653
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6742,
+      "step": 654
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7951,
+      "step": 655
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7765,
+      "step": 656
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9696,
+      "step": 657
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6123,
+      "step": 658
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7679,
+      "step": 659
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7243,
+      "step": 660
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7163,
+      "step": 661
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.5492,
+      "step": 662
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6985,
+      "step": 663
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7374,
+      "step": 664
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8563,
+      "step": 665
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6179,
+      "step": 666
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.4717,
+      "step": 667
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.686,
+      "step": 668
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7899,
+      "step": 669
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7996,
+      "step": 670
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7587,
+      "step": 671
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.749,
+      "step": 672
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7249,
+      "step": 673
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.8679,
+      "step": 674
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9212,
+      "step": 675
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6461,
+      "step": 676
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6228,
+      "step": 677
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.9143,
+      "step": 678
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.5956,
+      "step": 679
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6185,
+      "step": 680
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7049,
+      "step": 681
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6064,
+      "step": 682
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.7436,
+      "step": 683
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6832,
+      "step": 684
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 0.0002,
+      "loss": 0.6248,
+      "step": 685
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6944,
+      "step": 686
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6402,
+      "step": 687
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.755,
+      "step": 688
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5987,
+      "step": 689
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8574,
+      "step": 690
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9799,
+      "step": 691
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.658,
+      "step": 692
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.701,
+      "step": 693
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5968,
+      "step": 694
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8025,
+      "step": 695
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7759,
+      "step": 696
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5402,
+      "step": 697
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.4972,
+      "step": 698
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5578,
+      "step": 699
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7579,
+      "step": 700
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 1.0019,
+      "step": 701
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8391,
+      "step": 702
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6358,
+      "step": 703
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6885,
+      "step": 704
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6366,
+      "step": 705
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8087,
+      "step": 706
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6881,
+      "step": 707
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8975,
+      "step": 708
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7963,
+      "step": 709
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6991,
+      "step": 710
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7675,
+      "step": 711
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.727,
+      "step": 712
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.9688,
+      "step": 713
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7187,
+      "step": 714
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6465,
+      "step": 715
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7982,
+      "step": 716
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6121,
+      "step": 717
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.6841,
+      "step": 718
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5609,
+      "step": 719
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7194,
+      "step": 720
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.749,
+      "step": 721
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.908,
+      "step": 722
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7651,
+      "step": 723
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.7169,
+      "step": 724
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.5898,
+      "step": 725
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 1.0248,
+      "step": 726
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 0.0002,
+      "loss": 0.8546,
+      "step": 727
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8233,
+      "step": 728
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6465,
+      "step": 729
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8397,
+      "step": 730
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8623,
+      "step": 731
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6888,
+      "step": 732
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8152,
+      "step": 733
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8455,
+      "step": 734
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9103,
+      "step": 735
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.693,
+      "step": 736
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6127,
+      "step": 737
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.547,
+      "step": 738
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6516,
+      "step": 739
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.775,
+      "step": 740
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9026,
+      "step": 741
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8331,
+      "step": 742
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6344,
+      "step": 743
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6924,
+      "step": 744
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6213,
+      "step": 745
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6672,
+      "step": 746
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7349,
+      "step": 747
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.62,
+      "step": 748
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7824,
+      "step": 749
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8701,
+      "step": 750
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8007,
+      "step": 751
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6076,
+      "step": 752
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7324,
+      "step": 753
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8675,
+      "step": 754
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6253,
+      "step": 755
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6267,
+      "step": 756
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6996,
+      "step": 757
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.7982,
+      "step": 758
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8651,
+      "step": 759
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6264,
+      "step": 760
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8223,
+      "step": 761
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8483,
+      "step": 762
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.8448,
+      "step": 763
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6566,
+      "step": 764
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.6491,
+      "step": 765
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.635,
+      "step": 766
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 0.9089,
+      "step": 767
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 0.0002,
+      "loss": 1.1495,
+      "step": 768
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7183,
+      "step": 769
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5471,
+      "step": 770
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.0757,
+      "step": 771
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.799,
+      "step": 772
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7035,
+      "step": 773
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7346,
+      "step": 774
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7813,
+      "step": 775
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7777,
+      "step": 776
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7744,
+      "step": 777
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7948,
+      "step": 778
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7664,
+      "step": 779
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5548,
+      "step": 780
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7654,
+      "step": 781
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6196,
+      "step": 782
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6921,
+      "step": 783
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.713,
+      "step": 784
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8963,
+      "step": 785
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5201,
+      "step": 786
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6061,
+      "step": 787
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8025,
+      "step": 788
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8181,
+      "step": 789
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6752,
+      "step": 790
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7352,
+      "step": 791
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7279,
+      "step": 792
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7355,
+      "step": 793
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5628,
+      "step": 794
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.5691,
+      "step": 795
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6719,
+      "step": 796
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8278,
+      "step": 797
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6019,
+      "step": 798
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.769,
+      "step": 799
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8177,
+      "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 0.7239950299263,
+      "eval_runtime": 134.4496,
+      "eval_samples_per_second": 7.438,
+      "eval_steps_per_second": 3.719,
+      "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "mmlu_eval_accuracy": 0.5753427883544671,
+      "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": 0.5517241379310345,
+      "mmlu_eval_accuracy_college_biology": 0.75,
+      "mmlu_eval_accuracy_college_chemistry": 0.25,
+      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
+      "mmlu_eval_accuracy_college_mathematics": 0.18181818181818182,
+      "mmlu_eval_accuracy_college_medicine": 0.6818181818181818,
+      "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
+      "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
+      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
+      "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": 0.07142857142857142,
+      "mmlu_eval_accuracy_global_facts": 0.4,
+      "mmlu_eval_accuracy_high_school_biology": 0.71875,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.3181818181818182,
+      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
+      "mmlu_eval_accuracy_high_school_geography": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_government_and_politics": 0.8095238095238095,
+      "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": 0.17647058823529413,
+      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
+      "mmlu_eval_accuracy_high_school_statistics": 0.4782608695652174,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
+      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
+      "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
+      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_jurisprudence": 0.5454545454545454,
+      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
+      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
+      "mmlu_eval_accuracy_management": 0.9090909090909091,
+      "mmlu_eval_accuracy_marketing": 0.88,
+      "mmlu_eval_accuracy_medical_genetics": 0.7272727272727273,
+      "mmlu_eval_accuracy_miscellaneous": 0.7325581395348837,
+      "mmlu_eval_accuracy_moral_disputes": 0.5789473684210527,
+      "mmlu_eval_accuracy_moral_scenarios": 0.28,
+      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
+      "mmlu_eval_accuracy_philosophy": 0.6470588235294118,
+      "mmlu_eval_accuracy_prehistory": 0.5714285714285714,
+      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
+      "mmlu_eval_accuracy_professional_law": 0.4294117647058823,
+      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
+      "mmlu_eval_accuracy_professional_psychology": 0.5652173913043478,
+      "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": 0.8181818181818182,
+      "mmlu_eval_accuracy_virology": 0.5,
+      "mmlu_eval_accuracy_world_religions": 0.7894736842105263,
+      "mmlu_loss": 1.0745949354946769,
+      "step": 800
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.0825,
+      "step": 801
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8115,
+      "step": 802
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7761,
+      "step": 803
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.7772,
+      "step": 804
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8492,
+      "step": 805
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 1.1243,
+      "step": 806
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8167,
+      "step": 807
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8575,
+      "step": 808
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.6835,
+      "step": 809
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 0.0002,
+      "loss": 0.8665,
+      "step": 810
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6632,
+      "step": 811
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7311,
+      "step": 812
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8316,
+      "step": 813
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7578,
+      "step": 814
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6769,
+      "step": 815
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8397,
+      "step": 816
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8652,
+      "step": 817
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7459,
+      "step": 818
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8338,
+      "step": 819
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.5916,
+      "step": 820
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.722,
+      "step": 821
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6675,
+      "step": 822
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7488,
+      "step": 823
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8157,
+      "step": 824
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.9181,
+      "step": 825
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6226,
+      "step": 826
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7702,
+      "step": 827
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.814,
+      "step": 828
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.5983,
+      "step": 829
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.9346,
+      "step": 830
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.5142,
+      "step": 831
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7448,
+      "step": 832
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8205,
+      "step": 833
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.5905,
+      "step": 834
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6575,
+      "step": 835
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7603,
+      "step": 836
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7611,
+      "step": 837
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6134,
+      "step": 838
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.9152,
+      "step": 839
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8657,
+      "step": 840
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8002,
+      "step": 841
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.9159,
+      "step": 842
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.815,
+      "step": 843
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6841,
+      "step": 844
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8036,
+      "step": 845
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8004,
+      "step": 846
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.6196,
+      "step": 847
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.8639,
+      "step": 848
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.846,
+      "step": 849
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.7199,
+      "step": 850
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 0.0002,
+      "loss": 0.9038,
+      "step": 851
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6961,
+      "step": 852
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7525,
+      "step": 853
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6338,
+      "step": 854
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6793,
+      "step": 855
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.8711,
+      "step": 856
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6788,
+      "step": 857
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6174,
+      "step": 858
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6804,
+      "step": 859
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7595,
+      "step": 860
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.8655,
+      "step": 861
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.5748,
+      "step": 862
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6817,
+      "step": 863
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6669,
+      "step": 864
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.4497,
+      "step": 865
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7569,
+      "step": 866
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.8027,
+      "step": 867
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.8585,
+      "step": 868
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6116,
+      "step": 869
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7514,
+      "step": 870
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7235,
+      "step": 871
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6929,
+      "step": 872
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7981,
+      "step": 873
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.9363,
+      "step": 874
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.4867,
+      "step": 875
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.8945,
+      "step": 876
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.8476,
+      "step": 877
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7188,
+      "step": 878
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7405,
+      "step": 879
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6898,
+      "step": 880
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7015,
+      "step": 881
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7414,
+      "step": 882
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.8442,
+      "step": 883
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.5488,
+      "step": 884
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 1.0315,
+      "step": 885
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.7131,
+      "step": 886
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 0.6115,
+      "step": 887
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 0.0002,
+      "loss": 1.2947,
+      "step": 888
+    },
+    {
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7897,
-      "step": 20
+      "loss": 0.7414,
+      "step": 889
     },
     {
-      "epoch": 0.01,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7499,
-      "step": 30
+      "loss": 0.7756,
+      "step": 890
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7455,
-      "step": 40
+      "loss": 0.5815,
+      "step": 891
     },
     {
-      "epoch": 0.02,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7387,
-      "step": 50
+      "loss": 0.7633,
+      "step": 892
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.21,
       "learning_rate": 0.0002,
-      "loss": 0.7331,
-      "step": 60
+      "loss": 0.6815,
+      "step": 893
     },
     {
-      "epoch": 0.03,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7296,
-      "step": 70
+      "loss": 0.9169,
+      "step": 894
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.729,
-      "step": 80
+      "loss": 0.5973,
+      "step": 895
     },
     {
-      "epoch": 0.04,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.741,
-      "step": 90
+      "loss": 0.6257,
+      "step": 896
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7803,
-      "step": 100
+      "loss": 0.7479,
+      "step": 897
     },
     {
-      "epoch": 0.05,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7286,
-      "step": 110
+      "loss": 0.6264,
+      "step": 898
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7689,
-      "step": 120
+      "loss": 0.5946,
+      "step": 899
     },
     {
-      "epoch": 0.06,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.6787,
-      "step": 130
+      "loss": 0.842,
+      "step": 900
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7264,
-      "step": 140
+      "loss": 0.8394,
+      "step": 901
     },
     {
-      "epoch": 0.07,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7486,
-      "step": 150
+      "loss": 0.8753,
+      "step": 902
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7315,
-      "step": 160
+      "loss": 0.5577,
+      "step": 903
     },
     {
-      "epoch": 0.08,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.6814,
-      "step": 170
+      "loss": 0.5594,
+      "step": 904
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7104,
-      "step": 180
+      "loss": 0.7981,
+      "step": 905
     },
     {
-      "epoch": 0.09,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7133,
-      "step": 190
+      "loss": 0.5441,
+      "step": 906
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7104,
-      "step": 200
+      "loss": 0.9019,
+      "step": 907
     },
     {
-      "epoch": 0.1,
-      "eval_loss": 0.7200608849525452,
-      "eval_runtime": 135.0227,
-      "eval_samples_per_second": 7.406,
-      "eval_steps_per_second": 3.703,
-      "step": 200
+      "epoch": 0.22,
+      "learning_rate": 0.0002,
+      "loss": 0.8178,
+      "step": 908
     },
     {
-      "epoch": 0.1,
-      "mmlu_eval_accuracy": 0.598134136806777,
-      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
-      "mmlu_eval_accuracy_anatomy": 0.5,
-      "mmlu_eval_accuracy_astronomy": 0.6875,
-      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
-      "mmlu_eval_accuracy_college_biology": 0.5,
-      "mmlu_eval_accuracy_college_chemistry": 0.375,
-      "mmlu_eval_accuracy_college_computer_science": 0.5454545454545454,
-      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_medicine": 0.5909090909090909,
-      "mmlu_eval_accuracy_college_physics": 0.5454545454545454,
-      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
-      "mmlu_eval_accuracy_conceptual_physics": 0.5769230769230769,
-      "mmlu_eval_accuracy_econometrics": 0.5,
-      "mmlu_eval_accuracy_electrical_engineering": 0.6875,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.3902439024390244,
-      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
-      "mmlu_eval_accuracy_global_facts": 0.3,
-      "mmlu_eval_accuracy_high_school_biology": 0.65625,
-      "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": 0.8181818181818182,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5813953488372093,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.4482758620689655,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.6923076923076923,
-      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
-      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
-      "mmlu_eval_accuracy_high_school_world_history": 0.6923076923076923,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
-      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
-      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
-      "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
-      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
-      "mmlu_eval_accuracy_management": 0.9090909090909091,
-      "mmlu_eval_accuracy_marketing": 0.88,
-      "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
-      "mmlu_eval_accuracy_moral_disputes": 0.5789473684210527,
-      "mmlu_eval_accuracy_moral_scenarios": 0.36,
-      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
-      "mmlu_eval_accuracy_philosophy": 0.7352941176470589,
-      "mmlu_eval_accuracy_prehistory": 0.4857142857142857,
-      "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
-      "mmlu_eval_accuracy_professional_law": 0.3941176470588235,
-      "mmlu_eval_accuracy_professional_medicine": 0.6774193548387096,
-      "mmlu_eval_accuracy_professional_psychology": 0.5797101449275363,
-      "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.7037037037037037,
-      "mmlu_eval_accuracy_sociology": 0.9090909090909091,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
-      "mmlu_eval_accuracy_virology": 0.5,
-      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
-      "mmlu_loss": 1.0969554948822324,
-      "step": 200
+      "epoch": 0.22,
+      "learning_rate": 0.0002,
+      "loss": 0.577,
+      "step": 909
     },
     {
-      "epoch": 0.1,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7252,
-      "step": 210
+      "loss": 0.6044,
+      "step": 910
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7894,
-      "step": 220
+      "loss": 0.6033,
+      "step": 911
     },
     {
-      "epoch": 0.11,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7148,
-      "step": 230
+      "loss": 0.5683,
+      "step": 912
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.6983,
-      "step": 240
+      "loss": 0.8842,
+      "step": 913
     },
     {
-      "epoch": 0.12,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.688,
-      "step": 250
+      "loss": 0.882,
+      "step": 914
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7462,
-      "step": 260
+      "loss": 0.8556,
+      "step": 915
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7535,
-      "step": 270
+      "loss": 0.9205,
+      "step": 916
     },
     {
-      "epoch": 0.13,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7329,
-      "step": 280
+      "loss": 0.8148,
+      "step": 917
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.766,
-      "step": 290
+      "loss": 0.5743,
+      "step": 918
     },
     {
-      "epoch": 0.14,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7173,
-      "step": 300
+      "loss": 0.6712,
+      "step": 919
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7178,
-      "step": 310
+      "loss": 0.8465,
+      "step": 920
     },
     {
-      "epoch": 0.15,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7773,
-      "step": 320
+      "loss": 0.8824,
+      "step": 921
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7392,
-      "step": 330
+      "loss": 0.6747,
+      "step": 922
     },
     {
-      "epoch": 0.16,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7008,
-      "step": 340
+      "loss": 0.7338,
+      "step": 923
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.6709,
-      "step": 350
+      "loss": 0.7016,
+      "step": 924
     },
     {
-      "epoch": 0.17,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7072,
-      "step": 360
+      "loss": 0.6894,
+      "step": 925
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7495,
-      "step": 370
+      "loss": 0.8086,
+      "step": 926
     },
     {
-      "epoch": 0.18,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7264,
-      "step": 380
+      "loss": 0.7405,
+      "step": 927
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7346,
-      "step": 390
+      "loss": 0.6667,
+      "step": 928
     },
     {
-      "epoch": 0.19,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.6949,
-      "step": 400
+      "loss": 0.6109,
+      "step": 929
     },
     {
-      "epoch": 0.19,
-      "eval_loss": 0.7091138362884521,
-      "eval_runtime": 134.7715,
-      "eval_samples_per_second": 7.42,
-      "eval_steps_per_second": 3.71,
-      "step": 400
+      "epoch": 0.22,
+      "learning_rate": 0.0002,
+      "loss": 0.7632,
+      "step": 930
     },
     {
-      "epoch": 0.19,
-      "mmlu_eval_accuracy": 0.592211571347746,
-      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
-      "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": 0.5,
-      "mmlu_eval_accuracy_college_chemistry": 0.125,
-      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
-      "mmlu_eval_accuracy_college_medicine": 0.6818181818181818,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "mmlu_eval_accuracy_computer_security": 0.45454545454545453,
-      "mmlu_eval_accuracy_conceptual_physics": 0.5384615384615384,
-      "mmlu_eval_accuracy_econometrics": 0.5833333333333334,
-      "mmlu_eval_accuracy_electrical_engineering": 0.75,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
-      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
-      "mmlu_eval_accuracy_global_facts": 0.5,
-      "mmlu_eval_accuracy_high_school_biology": 0.625,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
-      "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
-      "mmlu_eval_accuracy_high_school_geography": 0.8181818181818182,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7142857142857143,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.627906976744186,
-      "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": 0.8,
-      "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
-      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5,
-      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
-      "mmlu_eval_accuracy_jurisprudence": 0.45454545454545453,
-      "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": 0.7441860465116279,
-      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
-      "mmlu_eval_accuracy_moral_scenarios": 0.3,
-      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
-      "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
-      "mmlu_eval_accuracy_prehistory": 0.6,
-      "mmlu_eval_accuracy_professional_accounting": 0.5161290322580645,
-      "mmlu_eval_accuracy_professional_law": 0.3941176470588235,
-      "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": 0.7037037037037037,
-      "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": 1.3008765503569621,
-      "step": 400
+      "epoch": 0.22,
+      "learning_rate": 0.0002,
+      "loss": 0.7689,
+      "step": 931
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7787,
-      "step": 410
+      "loss": 0.821,
+      "step": 932
     },
     {
-      "epoch": 0.2,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7021,
-      "step": 420
+      "loss": 0.6997,
+      "step": 933
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.22,
       "learning_rate": 0.0002,
-      "loss": 0.7274,
-      "step": 430
+      "loss": 0.7497,
+      "step": 934
     },
     {
-      "epoch": 0.21,
+      "epoch": 0.23,
+      "learning_rate": 0.0002,
+      "loss": 0.8191,
+      "step": 935
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002,
+      "loss": 0.681,
+      "step": 936
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002,
+      "loss": 0.6943,
+      "step": 937
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002,
+      "loss": 0.6888,
+      "step": 938
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 0.0002,
+      "loss": 0.8876,
+      "step": 939
+    },
+    {
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7108,
-      "step": 440
+      "loss": 0.6896,
+      "step": 940
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7207,
-      "step": 450
+      "loss": 0.7447,
+      "step": 941
     },
     {
-      "epoch": 0.22,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6881,
-      "step": 460
+      "loss": 1.2093,
+      "step": 942
     },
     {
       "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7196,
-      "step": 470
+      "loss": 0.6784,
+      "step": 943
     },
     {
       "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7283,
-      "step": 480
+      "loss": 0.6701,
+      "step": 944
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6806,
-      "step": 490
+      "loss": 0.6928,
+      "step": 945
     },
     {
-      "epoch": 0.24,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.703,
-      "step": 500
+      "loss": 0.8282,
+      "step": 946
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7188,
-      "step": 510
+      "loss": 0.7377,
+      "step": 947
     },
     {
-      "epoch": 0.25,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6997,
-      "step": 520
+      "loss": 0.5665,
+      "step": 948
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7681,
-      "step": 530
+      "loss": 0.7073,
+      "step": 949
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.733,
-      "step": 540
+      "loss": 1.0227,
+      "step": 950
     },
     {
-      "epoch": 0.26,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7336,
-      "step": 550
+      "loss": 0.8268,
+      "step": 951
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7251,
-      "step": 560
+      "loss": 0.6999,
+      "step": 952
     },
     {
-      "epoch": 0.27,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6643,
-      "step": 570
+      "loss": 0.8245,
+      "step": 953
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6937,
-      "step": 580
+      "loss": 0.8413,
+      "step": 954
     },
     {
-      "epoch": 0.28,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7344,
-      "step": 590
+      "loss": 0.8187,
+      "step": 955
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7257,
-      "step": 600
+      "loss": 0.6356,
+      "step": 956
     },
     {
-      "epoch": 0.29,
-      "eval_loss": 0.7025431394577026,
-      "eval_runtime": 134.7301,
-      "eval_samples_per_second": 7.422,
-      "eval_steps_per_second": 3.711,
-      "step": 600
+      "epoch": 0.23,
+      "learning_rate": 0.0002,
+      "loss": 0.6456,
+      "step": 957
     },
     {
-      "epoch": 0.29,
-      "mmlu_eval_accuracy": 0.5888137105723482,
-      "mmlu_eval_accuracy_abstract_algebra": 0.2727272727272727,
-      "mmlu_eval_accuracy_anatomy": 0.5714285714285714,
-      "mmlu_eval_accuracy_astronomy": 0.6875,
-      "mmlu_eval_accuracy_business_ethics": 0.6363636363636364,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.5862068965517241,
-      "mmlu_eval_accuracy_college_biology": 0.625,
-      "mmlu_eval_accuracy_college_chemistry": 0.125,
-      "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
-      "mmlu_eval_accuracy_conceptual_physics": 0.5,
-      "mmlu_eval_accuracy_econometrics": 0.5,
-      "mmlu_eval_accuracy_electrical_engineering": 0.625,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
-      "mmlu_eval_accuracy_formal_logic": 0.14285714285714285,
-      "mmlu_eval_accuracy_global_facts": 0.3,
-      "mmlu_eval_accuracy_high_school_biology": 0.65625,
-      "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": 0.7727272727272727,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5813953488372093,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.6923076923076923,
-      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8166666666666667,
-      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
-      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
-      "mmlu_eval_accuracy_human_aging": 0.782608695652174,
-      "mmlu_eval_accuracy_human_sexuality": 0.5,
-      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
-      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
-      "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": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
-      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
-      "mmlu_eval_accuracy_moral_scenarios": 0.36,
-      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
-      "mmlu_eval_accuracy_philosophy": 0.6764705882352942,
-      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
-      "mmlu_eval_accuracy_professional_accounting": 0.4838709677419355,
-      "mmlu_eval_accuracy_professional_law": 0.4176470588235294,
-      "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": 0.7777777777777778,
-      "mmlu_eval_accuracy_sociology": 0.9545454545454546,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
-      "mmlu_eval_accuracy_virology": 0.5,
-      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
-      "mmlu_loss": 1.2211437792208426,
-      "step": 600
+      "epoch": 0.23,
+      "learning_rate": 0.0002,
+      "loss": 0.7318,
+      "step": 958
     },
     {
-      "epoch": 0.29,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7102,
-      "step": 610
+      "loss": 0.4114,
+      "step": 959
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6837,
-      "step": 620
+      "loss": 0.8829,
+      "step": 960
     },
     {
-      "epoch": 0.3,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6901,
-      "step": 630
+      "loss": 0.5164,
+      "step": 961
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7344,
-      "step": 640
+      "loss": 0.6171,
+      "step": 962
     },
     {
-      "epoch": 0.31,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6781,
-      "step": 650
+      "loss": 0.666,
+      "step": 963
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7241,
-      "step": 660
+      "loss": 0.61,
+      "step": 964
     },
     {
-      "epoch": 0.32,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7392,
-      "step": 670
+      "loss": 0.8229,
+      "step": 965
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6869,
-      "step": 680
+      "loss": 0.6037,
+      "step": 966
     },
     {
-      "epoch": 0.33,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7297,
-      "step": 690
+      "loss": 0.6619,
+      "step": 967
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7047,
-      "step": 700
+      "loss": 0.7775,
+      "step": 968
     },
     {
-      "epoch": 0.34,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7215,
-      "step": 710
+      "loss": 0.8701,
+      "step": 969
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6823,
-      "step": 720
+      "loss": 0.606,
+      "step": 970
     },
     {
-      "epoch": 0.35,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7411,
-      "step": 730
+      "loss": 0.7611,
+      "step": 971
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6782,
-      "step": 740
+      "loss": 0.6763,
+      "step": 972
     },
     {
-      "epoch": 0.36,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7132,
-      "step": 750
+      "loss": 0.7471,
+      "step": 973
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.7059,
-      "step": 760
+      "loss": 0.7566,
+      "step": 974
     },
     {
-      "epoch": 0.37,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6966,
-      "step": 770
+      "loss": 0.8083,
+      "step": 975
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.23,
       "learning_rate": 0.0002,
-      "loss": 0.6754,
-      "step": 780
+      "loss": 0.6118,
+      "step": 976
     },
     {
-      "epoch": 0.38,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.711,
-      "step": 790
+      "loss": 0.8078,
+      "step": 977
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7929,
-      "step": 800
+      "loss": 0.6769,
+      "step": 978
     },
     {
-      "epoch": 0.39,
-      "eval_loss": 0.6975825428962708,
-      "eval_runtime": 134.776,
-      "eval_samples_per_second": 7.42,
-      "eval_steps_per_second": 3.71,
-      "step": 800
+      "epoch": 0.24,
+      "learning_rate": 0.0002,
+      "loss": 0.8414,
+      "step": 979
     },
     {
-      "epoch": 0.39,
-      "mmlu_eval_accuracy": 0.5952667918688763,
-      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
-      "mmlu_eval_accuracy_anatomy": 0.5,
-      "mmlu_eval_accuracy_astronomy": 0.625,
-      "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": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_mathematics": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
-      "mmlu_eval_accuracy_college_physics": 0.45454545454545453,
-      "mmlu_eval_accuracy_computer_security": 0.6363636363636364,
-      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
-      "mmlu_eval_accuracy_econometrics": 0.5,
-      "mmlu_eval_accuracy_electrical_engineering": 0.5625,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.43902439024390244,
-      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
-      "mmlu_eval_accuracy_global_facts": 0.4,
-      "mmlu_eval_accuracy_high_school_biology": 0.65625,
-      "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": 0.8181818181818182,
-      "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.627906976744186,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.2413793103448276,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.6923076923076923,
-      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
-      "mmlu_eval_accuracy_high_school_psychology": 0.85,
-      "mmlu_eval_accuracy_high_school_statistics": 0.43478260869565216,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
-      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
-      "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5,
-      "mmlu_eval_accuracy_international_law": 0.9230769230769231,
-      "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
-      "mmlu_eval_accuracy_logical_fallacies": 0.8333333333333334,
-      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
-      "mmlu_eval_accuracy_management": 0.8181818181818182,
-      "mmlu_eval_accuracy_marketing": 0.88,
-      "mmlu_eval_accuracy_medical_genetics": 1.0,
-      "mmlu_eval_accuracy_miscellaneous": 0.7441860465116279,
-      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
-      "mmlu_eval_accuracy_moral_scenarios": 0.35,
-      "mmlu_eval_accuracy_nutrition": 0.696969696969697,
-      "mmlu_eval_accuracy_philosophy": 0.6176470588235294,
-      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
-      "mmlu_eval_accuracy_professional_accounting": 0.45161290322580644,
-      "mmlu_eval_accuracy_professional_law": 0.36470588235294116,
-      "mmlu_eval_accuracy_professional_medicine": 0.5806451612903226,
-      "mmlu_eval_accuracy_professional_psychology": 0.6086956521739131,
-      "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.7777777777777778,
-      "mmlu_eval_accuracy_sociology": 0.9545454545454546,
-      "mmlu_eval_accuracy_us_foreign_policy": 0.7272727272727273,
-      "mmlu_eval_accuracy_virology": 0.5555555555555556,
-      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
-      "mmlu_loss": 1.2398327136164236,
-      "step": 800
+      "epoch": 0.24,
+      "learning_rate": 0.0002,
+      "loss": 0.6816,
+      "step": 980
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.6867,
-      "step": 810
+      "loss": 0.7193,
+      "step": 981
     },
     {
-      "epoch": 0.39,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7316,
-      "step": 820
+      "loss": 0.6208,
+      "step": 982
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.6924,
-      "step": 830
+      "loss": 0.8818,
+      "step": 983
     },
     {
-      "epoch": 0.4,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.6699,
-      "step": 840
+      "loss": 0.6667,
+      "step": 984
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.6919,
-      "step": 850
+      "loss": 0.8538,
+      "step": 985
     },
     {
-      "epoch": 0.41,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.6562,
-      "step": 860
+      "loss": 0.6827,
+      "step": 986
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.663,
-      "step": 870
+      "loss": 0.8169,
+      "step": 987
     },
     {
-      "epoch": 0.42,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7364,
-      "step": 880
+      "loss": 0.4983,
+      "step": 988
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.674,
-      "step": 890
+      "loss": 0.5943,
+      "step": 989
     },
     {
-      "epoch": 0.43,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.716,
-      "step": 900
+      "loss": 0.5234,
+      "step": 990
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7211,
-      "step": 910
+      "loss": 0.6884,
+      "step": 991
     },
     {
-      "epoch": 0.44,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.6421,
-      "step": 920
+      "loss": 0.8083,
+      "step": 992
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7586,
-      "step": 930
+      "loss": 0.7298,
+      "step": 993
     },
     {
-      "epoch": 0.45,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.6934,
-      "step": 940
+      "loss": 1.0189,
+      "step": 994
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7189,
-      "step": 950
+      "loss": 0.6473,
+      "step": 995
     },
     {
-      "epoch": 0.46,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7224,
-      "step": 960
+      "loss": 0.9218,
+      "step": 996
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7386,
-      "step": 970
+      "loss": 0.6945,
+      "step": 997
     },
     {
-      "epoch": 0.47,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7,
-      "step": 980
+      "loss": 0.6509,
+      "step": 998
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.716,
-      "step": 990
+      "loss": 0.7727,
+      "step": 999
     },
     {
-      "epoch": 0.48,
+      "epoch": 0.24,
       "learning_rate": 0.0002,
-      "loss": 0.7119,
+      "loss": 0.5368,
       "step": 1000
     },
     {
-      "epoch": 0.48,
-      "eval_loss": 0.692868709564209,
-      "eval_runtime": 134.8648,
-      "eval_samples_per_second": 7.415,
-      "eval_steps_per_second": 3.707,
+      "epoch": 0.24,
+      "eval_loss": 0.7207491993904114,
+      "eval_runtime": 134.4871,
+      "eval_samples_per_second": 7.436,
+      "eval_steps_per_second": 3.718,
       "step": 1000
     },
     {
-      "epoch": 0.48,
-      "mmlu_eval_accuracy": 0.5944760074380516,
-      "mmlu_eval_accuracy_abstract_algebra": 0.36363636363636365,
-      "mmlu_eval_accuracy_anatomy": 0.42857142857142855,
-      "mmlu_eval_accuracy_astronomy": 0.75,
-      "mmlu_eval_accuracy_business_ethics": 0.5454545454545454,
-      "mmlu_eval_accuracy_clinical_knowledge": 0.5172413793103449,
-      "mmlu_eval_accuracy_college_biology": 0.5,
-      "mmlu_eval_accuracy_college_chemistry": 0.375,
+      "epoch": 0.24,
+      "mmlu_eval_accuracy": 0.5688293285403725,
+      "mmlu_eval_accuracy_abstract_algebra": 0.18181818181818182,
+      "mmlu_eval_accuracy_anatomy": 0.5,
+      "mmlu_eval_accuracy_astronomy": 0.5625,
+      "mmlu_eval_accuracy_business_ethics": 0.45454545454545453,
+      "mmlu_eval_accuracy_clinical_knowledge": 0.6551724137931034,
+      "mmlu_eval_accuracy_college_biology": 0.5625,
+      "mmlu_eval_accuracy_college_chemistry": 0.25,
       "mmlu_eval_accuracy_college_computer_science": 0.45454545454545453,
-      "mmlu_eval_accuracy_college_mathematics": 0.5454545454545454,
-      "mmlu_eval_accuracy_college_medicine": 0.6363636363636364,
-      "mmlu_eval_accuracy_college_physics": 0.36363636363636365,
+      "mmlu_eval_accuracy_college_mathematics": 0.09090909090909091,
+      "mmlu_eval_accuracy_college_medicine": 0.5909090909090909,
+      "mmlu_eval_accuracy_college_physics": 0.6363636363636364,
       "mmlu_eval_accuracy_computer_security": 0.5454545454545454,
-      "mmlu_eval_accuracy_conceptual_physics": 0.4230769230769231,
-      "mmlu_eval_accuracy_econometrics": 0.5,
+      "mmlu_eval_accuracy_conceptual_physics": 0.38461538461538464,
+      "mmlu_eval_accuracy_econometrics": 0.6666666666666666,
       "mmlu_eval_accuracy_electrical_engineering": 0.5,
-      "mmlu_eval_accuracy_elementary_mathematics": 0.4878048780487805,
-      "mmlu_eval_accuracy_formal_logic": 0.2857142857142857,
-      "mmlu_eval_accuracy_global_facts": 0.4,
+      "mmlu_eval_accuracy_elementary_mathematics": 0.4634146341463415,
+      "mmlu_eval_accuracy_formal_logic": 0.35714285714285715,
+      "mmlu_eval_accuracy_global_facts": 0.3,
       "mmlu_eval_accuracy_high_school_biology": 0.625,
-      "mmlu_eval_accuracy_high_school_chemistry": 0.4090909090909091,
+      "mmlu_eval_accuracy_high_school_chemistry": 0.36363636363636365,
       "mmlu_eval_accuracy_high_school_computer_science": 0.6666666666666666,
-      "mmlu_eval_accuracy_high_school_european_history": 0.7777777777777778,
+      "mmlu_eval_accuracy_high_school_european_history": 0.6666666666666666,
       "mmlu_eval_accuracy_high_school_geography": 0.8636363636363636,
       "mmlu_eval_accuracy_high_school_government_and_politics": 0.7619047619047619,
-      "mmlu_eval_accuracy_high_school_macroeconomics": 0.6046511627906976,
-      "mmlu_eval_accuracy_high_school_mathematics": 0.3103448275862069,
-      "mmlu_eval_accuracy_high_school_microeconomics": 0.7307692307692307,
-      "mmlu_eval_accuracy_high_school_physics": 0.11764705882352941,
-      "mmlu_eval_accuracy_high_school_psychology": 0.8666666666666667,
+      "mmlu_eval_accuracy_high_school_macroeconomics": 0.5813953488372093,
+      "mmlu_eval_accuracy_high_school_mathematics": 0.13793103448275862,
+      "mmlu_eval_accuracy_high_school_microeconomics": 0.5769230769230769,
+      "mmlu_eval_accuracy_high_school_physics": 0.17647058823529413,
+      "mmlu_eval_accuracy_high_school_psychology": 0.9,
       "mmlu_eval_accuracy_high_school_statistics": 0.391304347826087,
-      "mmlu_eval_accuracy_high_school_us_history": 0.7272727272727273,
-      "mmlu_eval_accuracy_high_school_world_history": 0.6538461538461539,
+      "mmlu_eval_accuracy_high_school_us_history": 0.7727272727272727,
+      "mmlu_eval_accuracy_high_school_world_history": 0.7307692307692307,
       "mmlu_eval_accuracy_human_aging": 0.7391304347826086,
-      "mmlu_eval_accuracy_human_sexuality": 0.5833333333333334,
-      "mmlu_eval_accuracy_international_law": 0.8461538461538461,
+      "mmlu_eval_accuracy_human_sexuality": 0.6666666666666666,
+      "mmlu_eval_accuracy_international_law": 0.7692307692307693,
       "mmlu_eval_accuracy_jurisprudence": 0.6363636363636364,
       "mmlu_eval_accuracy_logical_fallacies": 0.6666666666666666,
-      "mmlu_eval_accuracy_machine_learning": 0.36363636363636365,
-      "mmlu_eval_accuracy_management": 0.9090909090909091,
-      "mmlu_eval_accuracy_marketing": 0.96,
+      "mmlu_eval_accuracy_machine_learning": 0.2727272727272727,
+      "mmlu_eval_accuracy_management": 0.8181818181818182,
+      "mmlu_eval_accuracy_marketing": 0.92,
       "mmlu_eval_accuracy_medical_genetics": 0.9090909090909091,
-      "mmlu_eval_accuracy_miscellaneous": 0.7558139534883721,
-      "mmlu_eval_accuracy_moral_disputes": 0.5263157894736842,
-      "mmlu_eval_accuracy_moral_scenarios": 0.28,
-      "mmlu_eval_accuracy_nutrition": 0.7272727272727273,
-      "mmlu_eval_accuracy_philosophy": 0.7058823529411765,
-      "mmlu_eval_accuracy_prehistory": 0.5428571428571428,
-      "mmlu_eval_accuracy_professional_accounting": 0.5483870967741935,
-      "mmlu_eval_accuracy_professional_law": 0.4117647058823529,
-      "mmlu_eval_accuracy_professional_medicine": 0.6774193548387096,
-      "mmlu_eval_accuracy_professional_psychology": 0.6231884057971014,
+      "mmlu_eval_accuracy_miscellaneous": 0.7209302325581395,
+      "mmlu_eval_accuracy_moral_disputes": 0.5526315789473685,
+      "mmlu_eval_accuracy_moral_scenarios": 0.34,
+      "mmlu_eval_accuracy_nutrition": 0.6666666666666666,
+      "mmlu_eval_accuracy_philosophy": 0.6470588235294118,
+      "mmlu_eval_accuracy_prehistory": 0.5142857142857142,
+      "mmlu_eval_accuracy_professional_accounting": 0.3870967741935484,
+      "mmlu_eval_accuracy_professional_law": 0.4294117647058823,
+      "mmlu_eval_accuracy_professional_medicine": 0.5483870967741935,
+      "mmlu_eval_accuracy_professional_psychology": 0.6666666666666666,
       "mmlu_eval_accuracy_public_relations": 0.5,
-      "mmlu_eval_accuracy_security_studies": 0.6666666666666666,
+      "mmlu_eval_accuracy_security_studies": 0.5555555555555556,
       "mmlu_eval_accuracy_sociology": 0.9090909090909091,
       "mmlu_eval_accuracy_us_foreign_policy": 0.8181818181818182,
       "mmlu_eval_accuracy_virology": 0.5555555555555556,
-      "mmlu_eval_accuracy_world_religions": 0.8947368421052632,
-      "mmlu_loss": 1.171048034264587,
+      "mmlu_eval_accuracy_world_religions": 0.8421052631578947,
+      "mmlu_loss": 1.2292023156003604,
       "step": 1000
     }
   ],
-  "logging_steps": 10,
-  "max_steps": 6231,
+  "logging_steps": 1,
+  "max_steps": 12465,
   "num_train_epochs": 3,
   "save_steps": 200,
-  "total_flos": 5.162070262020833e+17,
+  "total_flos": 2.065451372571525e+17,
   "trial_name": null,
   "trial_params": null
 }