diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,33013 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.99989090679421,
+  "global_step": 5442,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5625e-07,
+      "loss": 2.3487,
+      "step": 1
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125e-07,
+      "loss": 2.4205,
+      "step": 2
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.6875000000000006e-07,
+      "loss": 1.9017,
+      "step": 3
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 6.25e-07,
+      "loss": 2.2767,
+      "step": 4
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 7.8125e-07,
+      "loss": 2.5307,
+      "step": 5
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 9.375000000000001e-07,
+      "loss": 1.9619,
+      "step": 6
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.0937500000000001e-06,
+      "loss": 1.9485,
+      "step": 7
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.25e-06,
+      "loss": 2.1356,
+      "step": 8
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.40625e-06,
+      "loss": 1.9083,
+      "step": 9
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.5625e-06,
+      "loss": 2.2253,
+      "step": 10
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.71875e-06,
+      "loss": 2.1095,
+      "step": 11
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 1.8750000000000003e-06,
+      "loss": 2.6857,
+      "step": 12
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.0312500000000002e-06,
+      "loss": 2.4379,
+      "step": 13
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.1875000000000002e-06,
+      "loss": 2.4168,
+      "step": 14
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.3437500000000002e-06,
+      "loss": 2.1123,
+      "step": 15
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.5e-06,
+      "loss": 1.9643,
+      "step": 16
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.65625e-06,
+      "loss": 2.3382,
+      "step": 17
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.8125e-06,
+      "loss": 2.1628,
+      "step": 18
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 2.96875e-06,
+      "loss": 2.0519,
+      "step": 19
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.125e-06,
+      "loss": 2.0945,
+      "step": 20
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.28125e-06,
+      "loss": 1.5743,
+      "step": 21
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.4375e-06,
+      "loss": 1.7847,
+      "step": 22
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.59375e-06,
+      "loss": 1.6909,
+      "step": 23
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.7500000000000005e-06,
+      "loss": 1.8397,
+      "step": 24
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 3.90625e-06,
+      "loss": 1.5484,
+      "step": 25
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.0625000000000005e-06,
+      "loss": 1.8205,
+      "step": 26
+    },
+    {
+      "epoch": 0.0,
+      "learning_rate": 4.21875e-06,
+      "loss": 1.5234,
+      "step": 27
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.3750000000000005e-06,
+      "loss": 1.7319,
+      "step": 28
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.53125e-06,
+      "loss": 1.607,
+      "step": 29
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.6875000000000004e-06,
+      "loss": 1.5012,
+      "step": 30
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 4.84375e-06,
+      "loss": 1.5322,
+      "step": 31
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5e-06,
+      "loss": 1.5344,
+      "step": 32
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.156250000000001e-06,
+      "loss": 1.5732,
+      "step": 33
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.3125e-06,
+      "loss": 1.5932,
+      "step": 34
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.468750000000001e-06,
+      "loss": 1.702,
+      "step": 35
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.625e-06,
+      "loss": 1.442,
+      "step": 36
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.781250000000001e-06,
+      "loss": 1.4934,
+      "step": 37
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 5.9375e-06,
+      "loss": 1.3733,
+      "step": 38
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.093750000000001e-06,
+      "loss": 1.3046,
+      "step": 39
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.25e-06,
+      "loss": 1.4031,
+      "step": 40
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.406250000000001e-06,
+      "loss": 1.3591,
+      "step": 41
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.5625e-06,
+      "loss": 1.3649,
+      "step": 42
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.718750000000001e-06,
+      "loss": 1.3686,
+      "step": 43
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 6.875e-06,
+      "loss": 1.4616,
+      "step": 44
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.031250000000001e-06,
+      "loss": 1.4921,
+      "step": 45
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.1875e-06,
+      "loss": 1.4335,
+      "step": 46
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.343750000000001e-06,
+      "loss": 1.3393,
+      "step": 47
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.500000000000001e-06,
+      "loss": 1.3964,
+      "step": 48
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.656250000000001e-06,
+      "loss": 1.5092,
+      "step": 49
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.8125e-06,
+      "loss": 1.4013,
+      "step": 50
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 7.96875e-06,
+      "loss": 1.3204,
+      "step": 51
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 1.295,
+      "step": 52
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.281250000000001e-06,
+      "loss": 1.1988,
+      "step": 53
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.4375e-06,
+      "loss": 1.174,
+      "step": 54
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.59375e-06,
+      "loss": 1.4132,
+      "step": 55
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.750000000000001e-06,
+      "loss": 1.1933,
+      "step": 56
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 8.906250000000001e-06,
+      "loss": 1.3416,
+      "step": 57
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.0625e-06,
+      "loss": 1.3337,
+      "step": 58
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.21875e-06,
+      "loss": 1.2899,
+      "step": 59
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 1.3358,
+      "step": 60
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.531250000000001e-06,
+      "loss": 1.3554,
+      "step": 61
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.6875e-06,
+      "loss": 1.2143,
+      "step": 62
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.84375e-06,
+      "loss": 1.3669,
+      "step": 63
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 1e-05,
+      "loss": 1.3674,
+      "step": 64
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.99999914690353e-06,
+      "loss": 1.1833,
+      "step": 65
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999996587614404e-06,
+      "loss": 1.4345,
+      "step": 66
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.9999923221335e-06,
+      "loss": 1.3707,
+      "step": 67
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999986350462271e-06,
+      "loss": 1.33,
+      "step": 68
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999978672602755e-06,
+      "loss": 1.3159,
+      "step": 69
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999969288557573e-06,
+      "loss": 1.4069,
+      "step": 70
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999958198329928e-06,
+      "loss": 1.2977,
+      "step": 71
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999945401923603e-06,
+      "loss": 1.2771,
+      "step": 72
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999930899342967e-06,
+      "loss": 1.2951,
+      "step": 73
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999914690592964e-06,
+      "loss": 1.2666,
+      "step": 74
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999896775679127e-06,
+      "loss": 1.3499,
+      "step": 75
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999877154607573e-06,
+      "loss": 1.2006,
+      "step": 76
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999855827384992e-06,
+      "loss": 1.2266,
+      "step": 77
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999832794018667e-06,
+      "loss": 1.4036,
+      "step": 78
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999808054516454e-06,
+      "loss": 1.408,
+      "step": 79
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999781608886793e-06,
+      "loss": 1.2253,
+      "step": 80
+    },
+    {
+      "epoch": 0.01,
+      "learning_rate": 9.999753457138715e-06,
+      "loss": 1.4919,
+      "step": 81
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999723599281822e-06,
+      "loss": 1.1754,
+      "step": 82
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999692035326304e-06,
+      "loss": 1.2525,
+      "step": 83
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99965876528293e-06,
+      "loss": 1.443,
+      "step": 84
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999623789163056e-06,
+      "loss": 1.3728,
+      "step": 85
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999587106978614e-06,
+      "loss": 1.2785,
+      "step": 86
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999548718742124e-06,
+      "loss": 1.4477,
+      "step": 87
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999508624466685e-06,
+      "loss": 1.4006,
+      "step": 88
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999466824165978e-06,
+      "loss": 1.2411,
+      "step": 89
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999423317854266e-06,
+      "loss": 1.2597,
+      "step": 90
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999378105546398e-06,
+      "loss": 1.2276,
+      "step": 91
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9993311872578e-06,
+      "loss": 1.1935,
+      "step": 92
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99928256300448e-06,
+      "loss": 1.2785,
+      "step": 93
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999232232803036e-06,
+      "loss": 1.3276,
+      "step": 94
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999180196670638e-06,
+      "loss": 1.2304,
+      "step": 95
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999126454625046e-06,
+      "loss": 1.2639,
+      "step": 96
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.999071006684597e-06,
+      "loss": 1.1983,
+      "step": 97
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99901385286821e-06,
+      "loss": 1.3568,
+      "step": 98
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998954993195394e-06,
+      "loss": 1.4421,
+      "step": 99
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998894427686229e-06,
+      "loss": 1.3332,
+      "step": 100
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998832156361385e-06,
+      "loss": 1.2198,
+      "step": 101
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998768179242108e-06,
+      "loss": 1.2298,
+      "step": 102
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998702496350231e-06,
+      "loss": 1.2719,
+      "step": 103
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998635107708173e-06,
+      "loss": 1.2747,
+      "step": 104
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998566013338922e-06,
+      "loss": 1.2436,
+      "step": 105
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998495213266058e-06,
+      "loss": 1.3132,
+      "step": 106
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998422707513742e-06,
+      "loss": 1.3948,
+      "step": 107
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998348496106715e-06,
+      "loss": 1.1647,
+      "step": 108
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.9982725790703e-06,
+      "loss": 1.4002,
+      "step": 109
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998194956430406e-06,
+      "loss": 1.2033,
+      "step": 110
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998115628213515e-06,
+      "loss": 1.0662,
+      "step": 111
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.998034594446702e-06,
+      "loss": 1.1302,
+      "step": 112
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997951855157617e-06,
+      "loss": 1.1937,
+      "step": 113
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997867410374496e-06,
+      "loss": 1.3402,
+      "step": 114
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99778126012615e-06,
+      "loss": 1.2378,
+      "step": 115
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997693404441981e-06,
+      "loss": 1.2016,
+      "step": 116
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997603843351968e-06,
+      "loss": 1.3939,
+      "step": 117
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99751257688667e-06,
+      "loss": 1.1665,
+      "step": 118
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997419605077234e-06,
+      "loss": 1.2711,
+      "step": 119
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997324927955386e-06,
+      "loss": 1.2179,
+      "step": 120
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997228545553428e-06,
+      "loss": 1.2929,
+      "step": 121
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997130457904256e-06,
+      "loss": 1.3267,
+      "step": 122
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.997030665041337e-06,
+      "loss": 1.2535,
+      "step": 123
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996929166998727e-06,
+      "loss": 1.3286,
+      "step": 124
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99682596381106e-06,
+      "loss": 1.2681,
+      "step": 125
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99672105551355e-06,
+      "loss": 1.1596,
+      "step": 126
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996614442142e-06,
+      "loss": 1.1478,
+      "step": 127
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996506123732789e-06,
+      "loss": 1.3649,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "eval_loss": 2.0032918453216553,
+      "eval_runtime": 0.1746,
+      "eval_samples_per_second": 17.178,
+      "eval_steps_per_second": 17.178,
+      "step": 128
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996396100322877e-06,
+      "loss": 1.3803,
+      "step": 129
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996284371949815e-06,
+      "loss": 1.3704,
+      "step": 130
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996170938651721e-06,
+      "loss": 1.3524,
+      "step": 131
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.996055800467309e-06,
+      "loss": 1.2616,
+      "step": 132
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.995938957435865e-06,
+      "loss": 1.262,
+      "step": 133
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99582040959726e-06,
+      "loss": 1.436,
+      "step": 134
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99570015699195e-06,
+      "loss": 1.3177,
+      "step": 135
+    },
+    {
+      "epoch": 0.02,
+      "learning_rate": 9.99557819966097e-06,
+      "loss": 1.227,
+      "step": 136
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995454537645932e-06,
+      "loss": 1.2056,
+      "step": 137
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995329170989038e-06,
+      "loss": 1.1711,
+      "step": 138
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.995202099733067e-06,
+      "loss": 1.1775,
+      "step": 139
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99507332392138e-06,
+      "loss": 1.1503,
+      "step": 140
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994942843597922e-06,
+      "loss": 1.3519,
+      "step": 141
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994810658807217e-06,
+      "loss": 1.1814,
+      "step": 142
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994676769594372e-06,
+      "loss": 1.2688,
+      "step": 143
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994541176005072e-06,
+      "loss": 1.2624,
+      "step": 144
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994403878085592e-06,
+      "loss": 1.2061,
+      "step": 145
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99426487588278e-06,
+      "loss": 1.3677,
+      "step": 146
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.994124169444068e-06,
+      "loss": 1.2201,
+      "step": 147
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993981758817474e-06,
+      "loss": 1.1986,
+      "step": 148
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993837644051592e-06,
+      "loss": 1.2429,
+      "step": 149
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993691825195599e-06,
+      "loss": 1.3322,
+      "step": 150
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993544302299254e-06,
+      "loss": 1.2216,
+      "step": 151
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9933950754129e-06,
+      "loss": 1.1585,
+      "step": 152
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993244144587455e-06,
+      "loss": 1.1419,
+      "step": 153
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.993091509874425e-06,
+      "loss": 1.2124,
+      "step": 154
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992937171325895e-06,
+      "loss": 1.3108,
+      "step": 155
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99278112899453e-06,
+      "loss": 1.2896,
+      "step": 156
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99262338293358e-06,
+      "loss": 1.2077,
+      "step": 157
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992463933196869e-06,
+      "loss": 1.2569,
+      "step": 158
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992302779838814e-06,
+      "loss": 1.1888,
+      "step": 159
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.992139922914403e-06,
+      "loss": 1.2634,
+      "step": 160
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991975362479208e-06,
+      "loss": 1.2318,
+      "step": 161
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991809098589386e-06,
+      "loss": 1.1974,
+      "step": 162
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991641131301672e-06,
+      "loss": 1.0981,
+      "step": 163
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991471460673381e-06,
+      "loss": 1.1262,
+      "step": 164
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.991300086762414e-06,
+      "loss": 1.277,
+      "step": 165
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.99112700962725e-06,
+      "loss": 1.2583,
+      "step": 166
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990952229326947e-06,
+      "loss": 1.3115,
+      "step": 167
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990775745921148e-06,
+      "loss": 1.1602,
+      "step": 168
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990597559470079e-06,
+      "loss": 1.3407,
+      "step": 169
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990417670034539e-06,
+      "loss": 1.3764,
+      "step": 170
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990236077675916e-06,
+      "loss": 1.1541,
+      "step": 171
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.990052782456177e-06,
+      "loss": 1.3556,
+      "step": 172
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989867784437868e-06,
+      "loss": 1.2175,
+      "step": 173
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989681083684118e-06,
+      "loss": 1.3626,
+      "step": 174
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989492680258636e-06,
+      "loss": 1.3878,
+      "step": 175
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989302574225714e-06,
+      "loss": 1.3249,
+      "step": 176
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.989110765650221e-06,
+      "loss": 1.3019,
+      "step": 177
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988917254597613e-06,
+      "loss": 1.3881,
+      "step": 178
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988722041133919e-06,
+      "loss": 1.101,
+      "step": 179
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988525125325756e-06,
+      "loss": 1.4316,
+      "step": 180
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98832650724032e-06,
+      "loss": 1.3495,
+      "step": 181
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.988126186945385e-06,
+      "loss": 1.0971,
+      "step": 182
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987924164509311e-06,
+      "loss": 1.1758,
+      "step": 183
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987720440001031e-06,
+      "loss": 1.2761,
+      "step": 184
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.98751501349007e-06,
+      "loss": 1.3643,
+      "step": 185
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987307885046523e-06,
+      "loss": 1.1788,
+      "step": 186
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.987099054741071e-06,
+      "loss": 1.2674,
+      "step": 187
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986888522644975e-06,
+      "loss": 1.4476,
+      "step": 188
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.986676288830078e-06,
+      "loss": 1.2227,
+      "step": 189
+    },
+    {
+      "epoch": 0.03,
+      "learning_rate": 9.9864623533688e-06,
+      "loss": 1.125,
+      "step": 190
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.986246716334148e-06,
+      "loss": 1.3198,
+      "step": 191
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.9860293777997e-06,
+      "loss": 1.1579,
+      "step": 192
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985810337839626e-06,
+      "loss": 1.37,
+      "step": 193
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985589596528667e-06,
+      "loss": 1.2942,
+      "step": 194
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985367153942149e-06,
+      "loss": 1.3338,
+      "step": 195
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.985143010155979e-06,
+      "loss": 1.277,
+      "step": 196
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.984917165246644e-06,
+      "loss": 1.2787,
+      "step": 197
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.984689619291208e-06,
+      "loss": 1.2278,
+      "step": 198
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.984460372367323e-06,
+      "loss": 1.3554,
+      "step": 199
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.984229424553214e-06,
+      "loss": 1.2357,
+      "step": 200
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98399677592769e-06,
+      "loss": 1.2859,
+      "step": 201
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98376242657014e-06,
+      "loss": 1.3347,
+      "step": 202
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.983526376560531e-06,
+      "loss": 1.2119,
+      "step": 203
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.983288625979415e-06,
+      "loss": 1.2442,
+      "step": 204
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.983049174907921e-06,
+      "loss": 1.1375,
+      "step": 205
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.982808023427758e-06,
+      "loss": 1.3956,
+      "step": 206
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.982565171621217e-06,
+      "loss": 1.3042,
+      "step": 207
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98232061957117e-06,
+      "loss": 1.1745,
+      "step": 208
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.982074367361062e-06,
+      "loss": 1.0813,
+      "step": 209
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981826415074929e-06,
+      "loss": 1.1328,
+      "step": 210
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.98157676279738e-06,
+      "loss": 1.2563,
+      "step": 211
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981325410613608e-06,
+      "loss": 1.2073,
+      "step": 212
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.981072358609383e-06,
+      "loss": 1.3918,
+      "step": 213
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.980817606871054e-06,
+      "loss": 1.2203,
+      "step": 214
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.980561155485553e-06,
+      "loss": 1.091,
+      "step": 215
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.980303004540393e-06,
+      "loss": 1.288,
+      "step": 216
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.980043154123664e-06,
+      "loss": 1.2238,
+      "step": 217
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.979781604324037e-06,
+      "loss": 1.1775,
+      "step": 218
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.979518355230762e-06,
+      "loss": 1.1186,
+      "step": 219
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.979253406933672e-06,
+      "loss": 1.204,
+      "step": 220
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.978986759523175e-06,
+      "loss": 1.2566,
+      "step": 221
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.978718413090263e-06,
+      "loss": 1.2656,
+      "step": 222
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.978448367726505e-06,
+      "loss": 1.1091,
+      "step": 223
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.978176623524051e-06,
+      "loss": 1.3222,
+      "step": 224
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.977903180575632e-06,
+      "loss": 1.2493,
+      "step": 225
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.977628038974557e-06,
+      "loss": 1.2082,
+      "step": 226
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.977351198814712e-06,
+      "loss": 1.1433,
+      "step": 227
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97707266019057e-06,
+      "loss": 1.2517,
+      "step": 228
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.976792423197177e-06,
+      "loss": 1.0908,
+      "step": 229
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97651048793016e-06,
+      "loss": 1.3443,
+      "step": 230
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.976226854485728e-06,
+      "loss": 1.3809,
+      "step": 231
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.975941522960667e-06,
+      "loss": 1.2198,
+      "step": 232
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.975654493452341e-06,
+      "loss": 1.2102,
+      "step": 233
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.975365766058697e-06,
+      "loss": 1.2113,
+      "step": 234
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.975075340878261e-06,
+      "loss": 1.1277,
+      "step": 235
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974783218010136e-06,
+      "loss": 1.2481,
+      "step": 236
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974489397554006e-06,
+      "loss": 1.2354,
+      "step": 237
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.974193879610135e-06,
+      "loss": 1.1492,
+      "step": 238
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973896664279363e-06,
+      "loss": 1.1257,
+      "step": 239
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973597751663113e-06,
+      "loss": 1.2693,
+      "step": 240
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.973297141863385e-06,
+      "loss": 1.163,
+      "step": 241
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.97299483498276e-06,
+      "loss": 1.2689,
+      "step": 242
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972690831124392e-06,
+      "loss": 1.1369,
+      "step": 243
+    },
+    {
+      "epoch": 0.04,
+      "learning_rate": 9.972385130392025e-06,
+      "loss": 1.0375,
+      "step": 244
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.972077732889971e-06,
+      "loss": 1.1315,
+      "step": 245
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.97176863872313e-06,
+      "loss": 1.3403,
+      "step": 246
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.971457847996974e-06,
+      "loss": 1.3094,
+      "step": 247
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.971145360817558e-06,
+      "loss": 1.1861,
+      "step": 248
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.970831177291513e-06,
+      "loss": 1.2027,
+      "step": 249
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.970515297526054e-06,
+      "loss": 1.2339,
+      "step": 250
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.970197721628967e-06,
+      "loss": 1.3565,
+      "step": 251
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.969878449708624e-06,
+      "loss": 1.2314,
+      "step": 252
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.969557481873973e-06,
+      "loss": 1.2321,
+      "step": 253
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96923481823454e-06,
+      "loss": 1.1791,
+      "step": 254
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.968910458900429e-06,
+      "loss": 1.1928,
+      "step": 255
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.968584403982326e-06,
+      "loss": 1.2271,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "eval_loss": 1.9943183660507202,
+      "eval_runtime": 0.1725,
+      "eval_samples_per_second": 17.388,
+      "eval_steps_per_second": 17.388,
+      "step": 256
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.968256653591493e-06,
+      "loss": 1.1122,
+      "step": 257
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96792720783977e-06,
+      "loss": 1.2764,
+      "step": 258
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.967596066839575e-06,
+      "loss": 1.2484,
+      "step": 259
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.967263230703912e-06,
+      "loss": 1.0206,
+      "step": 260
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.966928699546352e-06,
+      "loss": 1.2469,
+      "step": 261
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.966592473481053e-06,
+      "loss": 1.1074,
+      "step": 262
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.966254552622747e-06,
+      "loss": 1.331,
+      "step": 263
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.965914937086745e-06,
+      "loss": 1.1653,
+      "step": 264
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.965573626988939e-06,
+      "loss": 1.3106,
+      "step": 265
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.965230622445795e-06,
+      "loss": 1.182,
+      "step": 266
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.964885923574359e-06,
+      "loss": 1.2449,
+      "step": 267
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.964539530492258e-06,
+      "loss": 1.1948,
+      "step": 268
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.964191443317694e-06,
+      "loss": 1.158,
+      "step": 269
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.963841662169446e-06,
+      "loss": 1.2662,
+      "step": 270
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.963490187166876e-06,
+      "loss": 1.1561,
+      "step": 271
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.963137018429918e-06,
+      "loss": 1.2165,
+      "step": 272
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.962782156079088e-06,
+      "loss": 1.2459,
+      "step": 273
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96242560023548e-06,
+      "loss": 1.2151,
+      "step": 274
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.96206735102076e-06,
+      "loss": 1.1049,
+      "step": 275
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.961707408557181e-06,
+      "loss": 1.3813,
+      "step": 276
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.961345772967569e-06,
+      "loss": 1.0569,
+      "step": 277
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.960982444375325e-06,
+      "loss": 1.2343,
+      "step": 278
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.960617422904432e-06,
+      "loss": 1.5035,
+      "step": 279
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.960250708679453e-06,
+      "loss": 1.3168,
+      "step": 280
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95988230182552e-06,
+      "loss": 1.2427,
+      "step": 281
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95951220246835e-06,
+      "loss": 1.3658,
+      "step": 282
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.959140410734233e-06,
+      "loss": 1.1789,
+      "step": 283
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.958766926750043e-06,
+      "loss": 1.2084,
+      "step": 284
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.958391750643225e-06,
+      "loss": 1.1834,
+      "step": 285
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.958014882541802e-06,
+      "loss": 1.2876,
+      "step": 286
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.957636322574378e-06,
+      "loss": 1.1416,
+      "step": 287
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95725607087013e-06,
+      "loss": 1.2216,
+      "step": 288
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.956874127558818e-06,
+      "loss": 1.1877,
+      "step": 289
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.956490492770771e-06,
+      "loss": 1.2436,
+      "step": 290
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.956105166636905e-06,
+      "loss": 1.2196,
+      "step": 291
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955718149288706e-06,
+      "loss": 1.1365,
+      "step": 292
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.955329440858239e-06,
+      "loss": 1.0796,
+      "step": 293
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954939041478147e-06,
+      "loss": 1.3289,
+      "step": 294
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954546951281648e-06,
+      "loss": 1.3235,
+      "step": 295
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.954153170402538e-06,
+      "loss": 1.2042,
+      "step": 296
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953757698975193e-06,
+      "loss": 1.0532,
+      "step": 297
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.953360537134562e-06,
+      "loss": 1.1647,
+      "step": 298
+    },
+    {
+      "epoch": 0.05,
+      "learning_rate": 9.95296168501617e-06,
+      "loss": 1.2023,
+      "step": 299
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.952561142756123e-06,
+      "loss": 1.0819,
+      "step": 300
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.952158910491102e-06,
+      "loss": 1.3255,
+      "step": 301
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.95175498835836e-06,
+      "loss": 1.35,
+      "step": 302
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.951349376495735e-06,
+      "loss": 1.309,
+      "step": 303
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.950942075041637e-06,
+      "loss": 1.1653,
+      "step": 304
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.95053308413505e-06,
+      "loss": 1.1453,
+      "step": 305
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.95012240391554e-06,
+      "loss": 1.3098,
+      "step": 306
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.949710034523248e-06,
+      "loss": 1.137,
+      "step": 307
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.949295976098887e-06,
+      "loss": 1.3166,
+      "step": 308
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.948880228783752e-06,
+      "loss": 1.3482,
+      "step": 309
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.948462792719713e-06,
+      "loss": 1.1751,
+      "step": 310
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.948043668049212e-06,
+      "loss": 1.2357,
+      "step": 311
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.947622854915273e-06,
+      "loss": 1.1679,
+      "step": 312
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.947200353461492e-06,
+      "loss": 1.2995,
+      "step": 313
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.946776163832046e-06,
+      "loss": 1.3115,
+      "step": 314
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.946350286171682e-06,
+      "loss": 1.226,
+      "step": 315
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.945922720625727e-06,
+      "loss": 1.2025,
+      "step": 316
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.945493467340082e-06,
+      "loss": 1.1366,
+      "step": 317
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.945062526461224e-06,
+      "loss": 1.093,
+      "step": 318
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.944629898136209e-06,
+      "loss": 1.4244,
+      "step": 319
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.944195582512664e-06,
+      "loss": 1.2347,
+      "step": 320
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.943759579738798e-06,
+      "loss": 1.1013,
+      "step": 321
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.943321889963388e-06,
+      "loss": 1.1889,
+      "step": 322
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.942882513335792e-06,
+      "loss": 1.1637,
+      "step": 323
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.942441450005944e-06,
+      "loss": 1.2257,
+      "step": 324
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.94199870012435e-06,
+      "loss": 1.2404,
+      "step": 325
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.941554263842092e-06,
+      "loss": 1.179,
+      "step": 326
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.941108141310832e-06,
+      "loss": 1.4025,
+      "step": 327
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940660332682803e-06,
+      "loss": 1.3699,
+      "step": 328
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.940210838110813e-06,
+      "loss": 1.3229,
+      "step": 329
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.939759657748248e-06,
+      "loss": 1.1384,
+      "step": 330
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93930679174907e-06,
+      "loss": 1.0775,
+      "step": 331
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.938852240267813e-06,
+      "loss": 1.3563,
+      "step": 332
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.938396003459584e-06,
+      "loss": 1.1739,
+      "step": 333
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.937938081480076e-06,
+      "loss": 1.2072,
+      "step": 334
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.937478474485542e-06,
+      "loss": 1.237,
+      "step": 335
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.937017182632824e-06,
+      "loss": 1.1009,
+      "step": 336
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.936554206079328e-06,
+      "loss": 1.2993,
+      "step": 337
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93608954498304e-06,
+      "loss": 1.2509,
+      "step": 338
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.935623199502524e-06,
+      "loss": 1.1498,
+      "step": 339
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93515516979691e-06,
+      "loss": 1.2761,
+      "step": 340
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.934685456025912e-06,
+      "loss": 1.2147,
+      "step": 341
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.934214058349813e-06,
+      "loss": 1.2565,
+      "step": 342
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.933740976929472e-06,
+      "loss": 1.2739,
+      "step": 343
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.933266211926321e-06,
+      "loss": 1.2681,
+      "step": 344
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93278976350237e-06,
+      "loss": 1.1676,
+      "step": 345
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.932311631820202e-06,
+      "loss": 1.2592,
+      "step": 346
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931831817042972e-06,
+      "loss": 1.2322,
+      "step": 347
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.931350319334412e-06,
+      "loss": 1.2968,
+      "step": 348
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.93086713885883e-06,
+      "loss": 1.2289,
+      "step": 349
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.930382275781103e-06,
+      "loss": 1.3838,
+      "step": 350
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929895730266687e-06,
+      "loss": 1.1643,
+      "step": 351
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.929407502481607e-06,
+      "loss": 1.0962,
+      "step": 352
+    },
+    {
+      "epoch": 0.06,
+      "learning_rate": 9.928917592592468e-06,
+      "loss": 1.277,
+      "step": 353
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.928426000766447e-06,
+      "loss": 1.2447,
+      "step": 354
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.927932727171291e-06,
+      "loss": 1.243,
+      "step": 355
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.927437771975325e-06,
+      "loss": 1.2032,
+      "step": 356
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.926941135347449e-06,
+      "loss": 1.1589,
+      "step": 357
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.926442817457132e-06,
+      "loss": 1.1836,
+      "step": 358
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.925942818474419e-06,
+      "loss": 1.2515,
+      "step": 359
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.925441138569932e-06,
+      "loss": 1.2564,
+      "step": 360
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.924937777914859e-06,
+      "loss": 1.118,
+      "step": 361
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.924432736680972e-06,
+      "loss": 1.1105,
+      "step": 362
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.923926015040604e-06,
+      "loss": 1.0903,
+      "step": 363
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.923417613166673e-06,
+      "loss": 1.3449,
+      "step": 364
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.922907531232663e-06,
+      "loss": 1.184,
+      "step": 365
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.922395769412633e-06,
+      "loss": 1.0246,
+      "step": 366
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.92188232788122e-06,
+      "loss": 1.1647,
+      "step": 367
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.921367206813625e-06,
+      "loss": 1.2424,
+      "step": 368
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.920850406385632e-06,
+      "loss": 1.282,
+      "step": 369
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.920331926773588e-06,
+      "loss": 1.2042,
+      "step": 370
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.919811768154422e-06,
+      "loss": 1.1675,
+      "step": 371
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.919289930705628e-06,
+      "loss": 1.0577,
+      "step": 372
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.918766414605283e-06,
+      "loss": 1.1929,
+      "step": 373
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.918241220032027e-06,
+      "loss": 1.2109,
+      "step": 374
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.91771434716508e-06,
+      "loss": 1.1739,
+      "step": 375
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.917185796184227e-06,
+      "loss": 1.1546,
+      "step": 376
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.916655567269833e-06,
+      "loss": 1.1033,
+      "step": 377
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.91612366060283e-06,
+      "loss": 1.2188,
+      "step": 378
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.915590076364728e-06,
+      "loss": 1.2263,
+      "step": 379
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.915054814737604e-06,
+      "loss": 1.1871,
+      "step": 380
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.914517875904112e-06,
+      "loss": 1.2888,
+      "step": 381
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.913979260047475e-06,
+      "loss": 1.2153,
+      "step": 382
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.91343896735149e-06,
+      "loss": 1.1982,
+      "step": 383
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.912896998000526e-06,
+      "loss": 1.0836,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "eval_loss": 1.9874013662338257,
+      "eval_runtime": 0.1767,
+      "eval_samples_per_second": 16.978,
+      "eval_steps_per_second": 16.978,
+      "step": 384
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.912353352179523e-06,
+      "loss": 1.0993,
+      "step": 385
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.911808030073995e-06,
+      "loss": 1.2902,
+      "step": 386
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.911261031870026e-06,
+      "loss": 1.0622,
+      "step": 387
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.910712357754271e-06,
+      "loss": 0.9384,
+      "step": 388
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.910162007913964e-06,
+      "loss": 1.2797,
+      "step": 389
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.909609982536901e-06,
+      "loss": 1.413,
+      "step": 390
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.909056281811457e-06,
+      "loss": 1.3284,
+      "step": 391
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.908500905926574e-06,
+      "loss": 1.0413,
+      "step": 392
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.907943855071769e-06,
+      "loss": 1.1696,
+      "step": 393
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.907385129437129e-06,
+      "loss": 1.3052,
+      "step": 394
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.906824729213314e-06,
+      "loss": 1.2073,
+      "step": 395
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90626265459155e-06,
+      "loss": 1.1689,
+      "step": 396
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.905698905763644e-06,
+      "loss": 1.082,
+      "step": 397
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.905133482921965e-06,
+      "loss": 1.1321,
+      "step": 398
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90456638625946e-06,
+      "loss": 1.169,
+      "step": 399
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903997615969641e-06,
+      "loss": 1.1359,
+      "step": 400
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.903427172246596e-06,
+      "loss": 1.1723,
+      "step": 401
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.902855055284983e-06,
+      "loss": 1.2014,
+      "step": 402
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.90228126528003e-06,
+      "loss": 1.2018,
+      "step": 403
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.901705802427534e-06,
+      "loss": 1.1701,
+      "step": 404
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.901128666923869e-06,
+      "loss": 1.1606,
+      "step": 405
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.900549858965975e-06,
+      "loss": 1.2487,
+      "step": 406
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899969378751362e-06,
+      "loss": 1.2719,
+      "step": 407
+    },
+    {
+      "epoch": 0.07,
+      "learning_rate": 9.899387226478112e-06,
+      "loss": 1.3184,
+      "step": 408
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.89880340234488e-06,
+      "loss": 1.2064,
+      "step": 409
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.898217906550889e-06,
+      "loss": 1.3109,
+      "step": 410
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.897630739295928e-06,
+      "loss": 1.0763,
+      "step": 411
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.897041900780368e-06,
+      "loss": 1.2172,
+      "step": 412
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.896451391205142e-06,
+      "loss": 1.1539,
+      "step": 413
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.895859210771749e-06,
+      "loss": 1.1968,
+      "step": 414
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.89526535968227e-06,
+      "loss": 1.2816,
+      "step": 415
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.894669838139351e-06,
+      "loss": 1.2586,
+      "step": 416
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.8940726463462e-06,
+      "loss": 1.3442,
+      "step": 417
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.893473784506607e-06,
+      "loss": 1.1204,
+      "step": 418
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.892873252824924e-06,
+      "loss": 1.1924,
+      "step": 419
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.892271051506077e-06,
+      "loss": 1.1771,
+      "step": 420
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.89166718075556e-06,
+      "loss": 1.1361,
+      "step": 421
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.89106164077944e-06,
+      "loss": 1.1304,
+      "step": 422
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.890454431784346e-06,
+      "loss": 1.1506,
+      "step": 423
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.889845553977482e-06,
+      "loss": 1.143,
+      "step": 424
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.889235007566624e-06,
+      "loss": 1.2452,
+      "step": 425
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.888622792760111e-06,
+      "loss": 1.042,
+      "step": 426
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.888008909766855e-06,
+      "loss": 1.0547,
+      "step": 427
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.887393358796335e-06,
+      "loss": 1.2618,
+      "step": 428
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.886776140058603e-06,
+      "loss": 1.3724,
+      "step": 429
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.886157253764279e-06,
+      "loss": 1.0937,
+      "step": 430
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.885536700124548e-06,
+      "loss": 1.1696,
+      "step": 431
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.884914479351167e-06,
+      "loss": 1.2896,
+      "step": 432
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.884290591656463e-06,
+      "loss": 1.1333,
+      "step": 433
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.88366503725333e-06,
+      "loss": 1.1234,
+      "step": 434
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.883037816355231e-06,
+      "loss": 1.3469,
+      "step": 435
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.8824089291762e-06,
+      "loss": 1.1605,
+      "step": 436
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.881778375930836e-06,
+      "loss": 1.288,
+      "step": 437
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.881146156834307e-06,
+      "loss": 1.2204,
+      "step": 438
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.880512272102352e-06,
+      "loss": 1.2686,
+      "step": 439
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.879876721951276e-06,
+      "loss": 1.1412,
+      "step": 440
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.879239506597955e-06,
+      "loss": 1.0662,
+      "step": 441
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87860062625983e-06,
+      "loss": 1.2362,
+      "step": 442
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.877960081154911e-06,
+      "loss": 1.2802,
+      "step": 443
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.877317871501778e-06,
+      "loss": 1.1033,
+      "step": 444
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87667399751958e-06,
+      "loss": 1.1731,
+      "step": 445
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.876028459428026e-06,
+      "loss": 1.406,
+      "step": 446
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.875381257447402e-06,
+      "loss": 1.1509,
+      "step": 447
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.874732391798559e-06,
+      "loss": 1.3243,
+      "step": 448
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.874081862702912e-06,
+      "loss": 1.0963,
+      "step": 449
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.873429670382448e-06,
+      "loss": 1.4594,
+      "step": 450
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.872775815059723e-06,
+      "loss": 1.131,
+      "step": 451
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.872120296957855e-06,
+      "loss": 1.1474,
+      "step": 452
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.87146311630053e-06,
+      "loss": 1.2028,
+      "step": 453
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.870804273312007e-06,
+      "loss": 1.1645,
+      "step": 454
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.870143768217109e-06,
+      "loss": 1.2461,
+      "step": 455
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.869481601241224e-06,
+      "loss": 1.1625,
+      "step": 456
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868817772610306e-06,
+      "loss": 1.1008,
+      "step": 457
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.868152282550884e-06,
+      "loss": 1.2499,
+      "step": 458
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.867485131290048e-06,
+      "loss": 1.1696,
+      "step": 459
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866816319055455e-06,
+      "loss": 1.2545,
+      "step": 460
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.866145846075327e-06,
+      "loss": 1.1747,
+      "step": 461
+    },
+    {
+      "epoch": 0.08,
+      "learning_rate": 9.86547371257846e-06,
+      "loss": 1.1763,
+      "step": 462
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.864799918794208e-06,
+      "loss": 1.2282,
+      "step": 463
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.864124464952499e-06,
+      "loss": 1.2583,
+      "step": 464
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.863447351283818e-06,
+      "loss": 1.0579,
+      "step": 465
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.862768578019228e-06,
+      "loss": 1.1317,
+      "step": 466
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.862088145390351e-06,
+      "loss": 1.3153,
+      "step": 467
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.861406053629376e-06,
+      "loss": 1.2593,
+      "step": 468
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.860722302969061e-06,
+      "loss": 1.2299,
+      "step": 469
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.860036893642726e-06,
+      "loss": 1.2467,
+      "step": 470
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.859349825884259e-06,
+      "loss": 1.1674,
+      "step": 471
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.858661099928115e-06,
+      "loss": 1.2724,
+      "step": 472
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.857970716009316e-06,
+      "loss": 1.1508,
+      "step": 473
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.857278674363442e-06,
+      "loss": 1.1998,
+      "step": 474
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.85658497522665e-06,
+      "loss": 1.2597,
+      "step": 475
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.855889618835654e-06,
+      "loss": 1.0848,
+      "step": 476
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.855192605427738e-06,
+      "loss": 1.155,
+      "step": 477
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.854493935240747e-06,
+      "loss": 1.0797,
+      "step": 478
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.853793608513097e-06,
+      "loss": 1.2856,
+      "step": 479
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.853091625483766e-06,
+      "loss": 1.17,
+      "step": 480
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.852387986392298e-06,
+      "loss": 1.1828,
+      "step": 481
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.851682691478799e-06,
+      "loss": 1.1367,
+      "step": 482
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.850975740983947e-06,
+      "loss": 1.2671,
+      "step": 483
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.850267135148978e-06,
+      "loss": 1.2266,
+      "step": 484
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.849556874215696e-06,
+      "loss": 1.2196,
+      "step": 485
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.84884495842647e-06,
+      "loss": 1.1669,
+      "step": 486
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.848131388024234e-06,
+      "loss": 1.2728,
+      "step": 487
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.847416163252486e-06,
+      "loss": 1.1219,
+      "step": 488
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.846699284355285e-06,
+      "loss": 1.3738,
+      "step": 489
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.84598075157726e-06,
+      "loss": 1.0972,
+      "step": 490
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.845260565163602e-06,
+      "loss": 1.3069,
+      "step": 491
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.844538725360067e-06,
+      "loss": 1.1606,
+      "step": 492
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.843815232412972e-06,
+      "loss": 1.1903,
+      "step": 493
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.843090086569205e-06,
+      "loss": 1.1645,
+      "step": 494
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.842363288076212e-06,
+      "loss": 1.0969,
+      "step": 495
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.841634837182002e-06,
+      "loss": 1.3699,
+      "step": 496
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.840904734135153e-06,
+      "loss": 1.2451,
+      "step": 497
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.840172979184803e-06,
+      "loss": 1.263,
+      "step": 498
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.839439572580658e-06,
+      "loss": 1.3488,
+      "step": 499
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.838704514572982e-06,
+      "loss": 1.1082,
+      "step": 500
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.837967805412605e-06,
+      "loss": 1.2005,
+      "step": 501
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.837229445350922e-06,
+      "loss": 1.2147,
+      "step": 502
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.83648943463989e-06,
+      "loss": 1.3005,
+      "step": 503
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.835747773532027e-06,
+      "loss": 1.2041,
+      "step": 504
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.835004462280419e-06,
+      "loss": 1.1842,
+      "step": 505
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.834259501138711e-06,
+      "loss": 1.0365,
+      "step": 506
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.833512890361115e-06,
+      "loss": 1.1635,
+      "step": 507
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.832764630202398e-06,
+      "loss": 1.1709,
+      "step": 508
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.8320147209179e-06,
+      "loss": 1.3285,
+      "step": 509
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.831263162763517e-06,
+      "loss": 1.0945,
+      "step": 510
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.830509955995711e-06,
+      "loss": 1.2316,
+      "step": 511
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.829755100871503e-06,
+      "loss": 1.2273,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "eval_loss": 1.9788621664047241,
+      "eval_runtime": 0.1782,
+      "eval_samples_per_second": 16.835,
+      "eval_steps_per_second": 16.835,
+      "step": 512
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828998597648482e-06,
+      "loss": 1.0562,
+      "step": 513
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.828240446584794e-06,
+      "loss": 1.3395,
+      "step": 514
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.827480647939148e-06,
+      "loss": 1.1123,
+      "step": 515
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.826719201970819e-06,
+      "loss": 1.1232,
+      "step": 516
+    },
+    {
+      "epoch": 0.09,
+      "learning_rate": 9.82595610893964e-06,
+      "loss": 1.0183,
+      "step": 517
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.82519136910601e-06,
+      "loss": 1.07,
+      "step": 518
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.824424982730887e-06,
+      "loss": 1.1131,
+      "step": 519
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.823656950075791e-06,
+      "loss": 1.1796,
+      "step": 520
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.822887271402804e-06,
+      "loss": 1.0786,
+      "step": 521
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.82211594697457e-06,
+      "loss": 1.1776,
+      "step": 522
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.821342977054295e-06,
+      "loss": 1.1121,
+      "step": 523
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.820568361905746e-06,
+      "loss": 1.3111,
+      "step": 524
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.819792101793254e-06,
+      "loss": 1.3338,
+      "step": 525
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.819014196981705e-06,
+      "loss": 1.3345,
+      "step": 526
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.818234647736554e-06,
+      "loss": 1.2476,
+      "step": 527
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.81745345432381e-06,
+      "loss": 1.0871,
+      "step": 528
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.816670617010048e-06,
+      "loss": 1.2906,
+      "step": 529
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.8158861360624e-06,
+      "loss": 1.2217,
+      "step": 530
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.815100011748564e-06,
+      "loss": 1.2398,
+      "step": 531
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.814312244336794e-06,
+      "loss": 1.3247,
+      "step": 532
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.81352283409591e-06,
+      "loss": 1.1417,
+      "step": 533
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.812731781295286e-06,
+      "loss": 1.2482,
+      "step": 534
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.81193908620486e-06,
+      "loss": 1.1619,
+      "step": 535
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.811144749095132e-06,
+      "loss": 1.2336,
+      "step": 536
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.810348770237159e-06,
+      "loss": 1.21,
+      "step": 537
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.809551149902559e-06,
+      "loss": 1.0133,
+      "step": 538
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.808751888363513e-06,
+      "loss": 1.0645,
+      "step": 539
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.807950985892759e-06,
+      "loss": 1.2109,
+      "step": 540
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.807148442763592e-06,
+      "loss": 1.2364,
+      "step": 541
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.806344259249878e-06,
+      "loss": 1.2379,
+      "step": 542
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.805538435626031e-06,
+      "loss": 1.1058,
+      "step": 543
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.804730972167029e-06,
+      "loss": 1.2281,
+      "step": 544
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.80392186914841e-06,
+      "loss": 1.2082,
+      "step": 545
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.803111126846272e-06,
+      "loss": 1.2742,
+      "step": 546
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.802298745537272e-06,
+      "loss": 1.259,
+      "step": 547
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.801484725498624e-06,
+      "loss": 1.1801,
+      "step": 548
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.800669067008105e-06,
+      "loss": 1.2201,
+      "step": 549
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.799851770344048e-06,
+      "loss": 1.1275,
+      "step": 550
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.799032835785347e-06,
+      "loss": 1.0648,
+      "step": 551
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.798212263611453e-06,
+      "loss": 1.1341,
+      "step": 552
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.797390054102376e-06,
+      "loss": 1.2951,
+      "step": 553
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.796566207538689e-06,
+      "loss": 1.3428,
+      "step": 554
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.795740724201518e-06,
+      "loss": 1.1056,
+      "step": 555
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.794913604372547e-06,
+      "loss": 1.2243,
+      "step": 556
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.794084848334027e-06,
+      "loss": 1.1073,
+      "step": 557
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.793254456368757e-06,
+      "loss": 1.0988,
+      "step": 558
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.792422428760102e-06,
+      "loss": 1.1826,
+      "step": 559
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.791588765791981e-06,
+      "loss": 1.178,
+      "step": 560
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.79075346774887e-06,
+      "loss": 1.2852,
+      "step": 561
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.789916534915807e-06,
+      "loss": 1.2444,
+      "step": 562
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.789077967578385e-06,
+      "loss": 1.1365,
+      "step": 563
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.788237766022755e-06,
+      "loss": 1.1793,
+      "step": 564
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.787395930535629e-06,
+      "loss": 1.0514,
+      "step": 565
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.78655246140427e-06,
+      "loss": 1.2697,
+      "step": 566
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.785707358916502e-06,
+      "loss": 1.1041,
+      "step": 567
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.784860623360712e-06,
+      "loss": 1.1833,
+      "step": 568
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.784012255025833e-06,
+      "loss": 1.2487,
+      "step": 569
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.783162254201365e-06,
+      "loss": 1.2727,
+      "step": 570
+    },
+    {
+      "epoch": 0.1,
+      "learning_rate": 9.782310621177357e-06,
+      "loss": 1.2421,
+      "step": 571
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.781457356244421e-06,
+      "loss": 1.2321,
+      "step": 572
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.780602459693726e-06,
+      "loss": 1.1893,
+      "step": 573
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.779745931816995e-06,
+      "loss": 1.1413,
+      "step": 574
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.778887772906505e-06,
+      "loss": 1.1723,
+      "step": 575
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.778027983255097e-06,
+      "loss": 1.1376,
+      "step": 576
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.777166563156162e-06,
+      "loss": 1.1978,
+      "step": 577
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.77630351290365e-06,
+      "loss": 1.1321,
+      "step": 578
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.775438832792069e-06,
+      "loss": 1.1286,
+      "step": 579
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.774572523116479e-06,
+      "loss": 1.2222,
+      "step": 580
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.773704584172498e-06,
+      "loss": 0.9631,
+      "step": 581
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.772835016256304e-06,
+      "loss": 1.2081,
+      "step": 582
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.771963819664623e-06,
+      "loss": 1.1471,
+      "step": 583
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.77109099469474e-06,
+      "loss": 1.1148,
+      "step": 584
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.770216541644502e-06,
+      "loss": 1.0429,
+      "step": 585
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.769340460812303e-06,
+      "loss": 1.205,
+      "step": 586
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.768462752497095e-06,
+      "loss": 1.1608,
+      "step": 587
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.767583416998385e-06,
+      "loss": 1.102,
+      "step": 588
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.76670245461624e-06,
+      "loss": 1.2193,
+      "step": 589
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.765819865651274e-06,
+      "loss": 1.2733,
+      "step": 590
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.764935650404663e-06,
+      "loss": 1.0495,
+      "step": 591
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.764049809178136e-06,
+      "loss": 1.2624,
+      "step": 592
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.763162342273974e-06,
+      "loss": 1.1924,
+      "step": 593
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.762273249995016e-06,
+      "loss": 1.1971,
+      "step": 594
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.761382532644655e-06,
+      "loss": 1.163,
+      "step": 595
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.76049019052684e-06,
+      "loss": 1.1592,
+      "step": 596
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.759596223946066e-06,
+      "loss": 1.1514,
+      "step": 597
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.758700633207396e-06,
+      "loss": 1.0963,
+      "step": 598
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.757803418616439e-06,
+      "loss": 1.2363,
+      "step": 599
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.756904580479354e-06,
+      "loss": 1.1679,
+      "step": 600
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.756004119102865e-06,
+      "loss": 0.9515,
+      "step": 601
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.755102034794241e-06,
+      "loss": 1.0976,
+      "step": 602
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.754198327861309e-06,
+      "loss": 1.3785,
+      "step": 603
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.75329299861245e-06,
+      "loss": 0.9393,
+      "step": 604
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.752386047356594e-06,
+      "loss": 1.1562,
+      "step": 605
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.751477474403232e-06,
+      "loss": 1.1601,
+      "step": 606
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.7505672800624e-06,
+      "loss": 1.1506,
+      "step": 607
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.749655464644693e-06,
+      "loss": 1.2045,
+      "step": 608
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.74874202846126e-06,
+      "loss": 1.3764,
+      "step": 609
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.747826971823797e-06,
+      "loss": 1.051,
+      "step": 610
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.74691029504456e-06,
+      "loss": 1.1007,
+      "step": 611
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.745991998436351e-06,
+      "loss": 1.1642,
+      "step": 612
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.74507208231253e-06,
+      "loss": 1.1971,
+      "step": 613
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.744150546987008e-06,
+      "loss": 1.2896,
+      "step": 614
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.743227392774249e-06,
+      "loss": 1.097,
+      "step": 615
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.742302619989268e-06,
+      "loss": 1.0493,
+      "step": 616
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.741376228947633e-06,
+      "loss": 1.2386,
+      "step": 617
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.740448219965463e-06,
+      "loss": 1.2414,
+      "step": 618
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.739518593359434e-06,
+      "loss": 1.2299,
+      "step": 619
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.73858734944677e-06,
+      "loss": 1.1072,
+      "step": 620
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.737654488545244e-06,
+      "loss": 1.2822,
+      "step": 621
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.736720010973184e-06,
+      "loss": 1.2942,
+      "step": 622
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.735783917049474e-06,
+      "loss": 1.0701,
+      "step": 623
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.734846207093545e-06,
+      "loss": 1.1265,
+      "step": 624
+    },
+    {
+      "epoch": 0.11,
+      "learning_rate": 9.733906881425375e-06,
+      "loss": 1.2189,
+      "step": 625
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.732965940365503e-06,
+      "loss": 1.2947,
+      "step": 626
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.732023384235011e-06,
+      "loss": 1.2699,
+      "step": 627
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.73107921335554e-06,
+      "loss": 1.294,
+      "step": 628
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.730133428049273e-06,
+      "loss": 1.2954,
+      "step": 629
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.729186028638951e-06,
+      "loss": 1.0476,
+      "step": 630
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.72823701544786e-06,
+      "loss": 1.1817,
+      "step": 631
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.727286388799843e-06,
+      "loss": 1.1931,
+      "step": 632
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.726334149019292e-06,
+      "loss": 1.2757,
+      "step": 633
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.725380296431144e-06,
+      "loss": 1.1768,
+      "step": 634
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.724424831360893e-06,
+      "loss": 1.1821,
+      "step": 635
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.723467754134579e-06,
+      "loss": 1.1751,
+      "step": 636
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.722509065078793e-06,
+      "loss": 1.294,
+      "step": 637
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.72154876452068e-06,
+      "loss": 1.2715,
+      "step": 638
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.72058685278793e-06,
+      "loss": 1.2494,
+      "step": 639
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.719623330208783e-06,
+      "loss": 1.1129,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "eval_loss": 1.9725112915039062,
+      "eval_runtime": 0.1721,
+      "eval_samples_per_second": 17.427,
+      "eval_steps_per_second": 17.427,
+      "step": 640
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.71865819711203e-06,
+      "loss": 1.1986,
+      "step": 641
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.717691453827015e-06,
+      "loss": 1.1916,
+      "step": 642
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.716723100683624e-06,
+      "loss": 1.1713,
+      "step": 643
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.715753138012298e-06,
+      "loss": 1.1103,
+      "step": 644
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.714781566144029e-06,
+      "loss": 1.2135,
+      "step": 645
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.713808385410349e-06,
+      "loss": 1.1303,
+      "step": 646
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.712833596143348e-06,
+      "loss": 1.045,
+      "step": 647
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.711857198675662e-06,
+      "loss": 1.263,
+      "step": 648
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.710879193340474e-06,
+      "loss": 1.1976,
+      "step": 649
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.70989958047152e-06,
+      "loss": 1.1732,
+      "step": 650
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.708918360403078e-06,
+      "loss": 1.0703,
+      "step": 651
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.707935533469981e-06,
+      "loss": 1.2952,
+      "step": 652
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.706951100007604e-06,
+      "loss": 1.1874,
+      "step": 653
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.70596506035188e-06,
+      "loss": 1.1592,
+      "step": 654
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.704977414839277e-06,
+      "loss": 1.1793,
+      "step": 655
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.703988163806822e-06,
+      "loss": 1.2384,
+      "step": 656
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.702997307592082e-06,
+      "loss": 1.2349,
+      "step": 657
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.70200484653318e-06,
+      "loss": 1.1499,
+      "step": 658
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.701010780968778e-06,
+      "loss": 1.0834,
+      "step": 659
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.70001511123809e-06,
+      "loss": 1.1833,
+      "step": 660
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.699017837680882e-06,
+      "loss": 1.2679,
+      "step": 661
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.698018960637454e-06,
+      "loss": 1.1336,
+      "step": 662
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.697018480448668e-06,
+      "loss": 1.1439,
+      "step": 663
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.696016397455925e-06,
+      "loss": 1.2105,
+      "step": 664
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.695012712001171e-06,
+      "loss": 1.1967,
+      "step": 665
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.694007424426907e-06,
+      "loss": 1.1219,
+      "step": 666
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.693000535076173e-06,
+      "loss": 1.0841,
+      "step": 667
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.691992044292557e-06,
+      "loss": 1.1799,
+      "step": 668
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.6909819524202e-06,
+      "loss": 1.1061,
+      "step": 669
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68997025980378e-06,
+      "loss": 1.0878,
+      "step": 670
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.688956966788528e-06,
+      "loss": 1.2933,
+      "step": 671
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.687942073720217e-06,
+      "loss": 1.1747,
+      "step": 672
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.686925580945167e-06,
+      "loss": 1.1201,
+      "step": 673
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68590748881025e-06,
+      "loss": 1.2542,
+      "step": 674
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68488779766287e-06,
+      "loss": 1.1379,
+      "step": 675
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.683866507850991e-06,
+      "loss": 1.0578,
+      "step": 676
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.682843619723115e-06,
+      "loss": 1.2595,
+      "step": 677
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68181913362829e-06,
+      "loss": 1.2359,
+      "step": 678
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.68079304991611e-06,
+      "loss": 1.2446,
+      "step": 679
+    },
+    {
+      "epoch": 0.12,
+      "learning_rate": 9.679765368936716e-06,
+      "loss": 1.1974,
+      "step": 680
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.678736091040791e-06,
+      "loss": 1.2122,
+      "step": 681
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.677705216579564e-06,
+      "loss": 1.1707,
+      "step": 682
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.676672745904813e-06,
+      "loss": 1.2939,
+      "step": 683
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.675638679368854e-06,
+      "loss": 1.1258,
+      "step": 684
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.674603017324547e-06,
+      "loss": 1.2899,
+      "step": 685
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.673565760125304e-06,
+      "loss": 1.1645,
+      "step": 686
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.672526908125077e-06,
+      "loss": 1.1888,
+      "step": 687
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.671486461678363e-06,
+      "loss": 1.188,
+      "step": 688
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.670444421140199e-06,
+      "loss": 1.2617,
+      "step": 689
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.669400786866172e-06,
+      "loss": 1.3123,
+      "step": 690
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.66835555921241e-06,
+      "loss": 1.3963,
+      "step": 691
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.667308738535586e-06,
+      "loss": 1.1308,
+      "step": 692
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.666260325192913e-06,
+      "loss": 1.1335,
+      "step": 693
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.665210319542154e-06,
+      "loss": 1.2027,
+      "step": 694
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.664158721941607e-06,
+      "loss": 1.217,
+      "step": 695
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.66310553275012e-06,
+      "loss": 1.4966,
+      "step": 696
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.662050752327081e-06,
+      "loss": 1.1211,
+      "step": 697
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.660994381032423e-06,
+      "loss": 1.0462,
+      "step": 698
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.659936419226622e-06,
+      "loss": 1.0901,
+      "step": 699
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.65887686727069e-06,
+      "loss": 1.1676,
+      "step": 700
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.657815725526192e-06,
+      "loss": 1.1442,
+      "step": 701
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.656752994355229e-06,
+      "loss": 1.2455,
+      "step": 702
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.655688674120445e-06,
+      "loss": 1.1288,
+      "step": 703
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.654622765185029e-06,
+      "loss": 1.1952,
+      "step": 704
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.653555267912708e-06,
+      "loss": 1.1719,
+      "step": 705
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.652486182667755e-06,
+      "loss": 1.2047,
+      "step": 706
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.651415509814984e-06,
+      "loss": 1.085,
+      "step": 707
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.650343249719745e-06,
+      "loss": 1.3102,
+      "step": 708
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.64926940274794e-06,
+      "loss": 0.9858,
+      "step": 709
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.648193969266006e-06,
+      "loss": 1.2251,
+      "step": 710
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.64711694964092e-06,
+      "loss": 1.2755,
+      "step": 711
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.646038344240205e-06,
+      "loss": 1.2336,
+      "step": 712
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.644958153431921e-06,
+      "loss": 1.2837,
+      "step": 713
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.643876377584674e-06,
+      "loss": 0.9452,
+      "step": 714
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.642793017067601e-06,
+      "loss": 1.1756,
+      "step": 715
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.641708072250394e-06,
+      "loss": 1.1764,
+      "step": 716
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.640621543503274e-06,
+      "loss": 1.0501,
+      "step": 717
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.639533431197008e-06,
+      "loss": 1.2403,
+      "step": 718
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.638443735702899e-06,
+      "loss": 1.074,
+      "step": 719
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.637352457392796e-06,
+      "loss": 1.2838,
+      "step": 720
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.636259596639085e-06,
+      "loss": 1.0937,
+      "step": 721
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.635165153814695e-06,
+      "loss": 1.224,
+      "step": 722
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.634069129293084e-06,
+      "loss": 1.32,
+      "step": 723
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.632971523448267e-06,
+      "loss": 1.0648,
+      "step": 724
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.631872336654783e-06,
+      "loss": 1.2193,
+      "step": 725
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.63077156928772e-06,
+      "loss": 1.1954,
+      "step": 726
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.629669221722702e-06,
+      "loss": 1.2532,
+      "step": 727
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62856529433589e-06,
+      "loss": 1.1944,
+      "step": 728
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62745978750399e-06,
+      "loss": 1.1801,
+      "step": 729
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.626352701604243e-06,
+      "loss": 1.0922,
+      "step": 730
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.62524403701443e-06,
+      "loss": 1.1069,
+      "step": 731
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.624133794112867e-06,
+      "loss": 1.1554,
+      "step": 732
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.623021973278412e-06,
+      "loss": 1.2774,
+      "step": 733
+    },
+    {
+      "epoch": 0.13,
+      "learning_rate": 9.621908574890466e-06,
+      "loss": 1.2153,
+      "step": 734
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.620793599328958e-06,
+      "loss": 1.0126,
+      "step": 735
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.619677046974364e-06,
+      "loss": 1.2216,
+      "step": 736
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.618558918207694e-06,
+      "loss": 1.2215,
+      "step": 737
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.617439213410497e-06,
+      "loss": 1.1747,
+      "step": 738
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.616317932964858e-06,
+      "loss": 1.1234,
+      "step": 739
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.615195077253403e-06,
+      "loss": 1.1915,
+      "step": 740
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.614070646659292e-06,
+      "loss": 1.2641,
+      "step": 741
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.612944641566225e-06,
+      "loss": 1.2645,
+      "step": 742
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.611817062358437e-06,
+      "loss": 1.0011,
+      "step": 743
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.610687909420704e-06,
+      "loss": 1.2235,
+      "step": 744
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.609557183138336e-06,
+      "loss": 1.1234,
+      "step": 745
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.608424883897178e-06,
+      "loss": 1.092,
+      "step": 746
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.607291012083617e-06,
+      "loss": 1.1642,
+      "step": 747
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.606155568084572e-06,
+      "loss": 1.1912,
+      "step": 748
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.605018552287504e-06,
+      "loss": 1.2359,
+      "step": 749
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.603879965080401e-06,
+      "loss": 1.1144,
+      "step": 750
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.602739806851797e-06,
+      "loss": 1.2746,
+      "step": 751
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.601598077990754e-06,
+      "loss": 1.1832,
+      "step": 752
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.600454778886881e-06,
+      "loss": 1.3005,
+      "step": 753
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.599309909930308e-06,
+      "loss": 1.1032,
+      "step": 754
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.598163471511713e-06,
+      "loss": 1.2435,
+      "step": 755
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.597015464022303e-06,
+      "loss": 1.1723,
+      "step": 756
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.595865887853825e-06,
+      "loss": 1.055,
+      "step": 757
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.594714743398557e-06,
+      "loss": 1.2356,
+      "step": 758
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.593562031049313e-06,
+      "loss": 1.0038,
+      "step": 759
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.592407751199442e-06,
+      "loss": 1.0857,
+      "step": 760
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.591251904242833e-06,
+      "loss": 1.1175,
+      "step": 761
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.590094490573904e-06,
+      "loss": 1.2137,
+      "step": 762
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.588935510587606e-06,
+      "loss": 1.2036,
+      "step": 763
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.587774964679432e-06,
+      "loss": 1.2052,
+      "step": 764
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.586612853245406e-06,
+      "loss": 1.2179,
+      "step": 765
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.585449176682078e-06,
+      "loss": 1.0121,
+      "step": 766
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.584283935386547e-06,
+      "loss": 1.1783,
+      "step": 767
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.583117129756433e-06,
+      "loss": 1.1631,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "eval_loss": 1.9853423833847046,
+      "eval_runtime": 0.1644,
+      "eval_samples_per_second": 18.245,
+      "eval_steps_per_second": 18.245,
+      "step": 768
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.581948760189898e-06,
+      "loss": 1.21,
+      "step": 769
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.580778827085634e-06,
+      "loss": 1.1196,
+      "step": 770
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.579607330842868e-06,
+      "loss": 1.2569,
+      "step": 771
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.578434271861358e-06,
+      "loss": 1.0701,
+      "step": 772
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.5772596505414e-06,
+      "loss": 1.161,
+      "step": 773
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.576083467283816e-06,
+      "loss": 1.2651,
+      "step": 774
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.574905722489969e-06,
+      "loss": 1.2839,
+      "step": 775
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.573726416561748e-06,
+      "loss": 1.0668,
+      "step": 776
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.57254554990158e-06,
+      "loss": 1.2288,
+      "step": 777
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.57136312291242e-06,
+      "loss": 1.1031,
+      "step": 778
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.570179135997761e-06,
+      "loss": 1.2326,
+      "step": 779
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.568993589561623e-06,
+      "loss": 1.2105,
+      "step": 780
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.56780648400856e-06,
+      "loss": 1.1025,
+      "step": 781
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.566617819743657e-06,
+      "loss": 1.2447,
+      "step": 782
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.565427597172535e-06,
+      "loss": 1.114,
+      "step": 783
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.564235816701341e-06,
+      "loss": 1.0283,
+      "step": 784
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.563042478736758e-06,
+      "loss": 0.9774,
+      "step": 785
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.561847583686e-06,
+      "loss": 1.1171,
+      "step": 786
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.560651131956809e-06,
+      "loss": 1.0755,
+      "step": 787
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.559453123957463e-06,
+      "loss": 1.1637,
+      "step": 788
+    },
+    {
+      "epoch": 0.14,
+      "learning_rate": 9.558253560096766e-06,
+      "loss": 1.3531,
+      "step": 789
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.557052440784058e-06,
+      "loss": 1.2359,
+      "step": 790
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.555849766429206e-06,
+      "loss": 1.3055,
+      "step": 791
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.554645537442608e-06,
+      "loss": 1.3776,
+      "step": 792
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.553439754235195e-06,
+      "loss": 1.2302,
+      "step": 793
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.552232417218425e-06,
+      "loss": 1.0811,
+      "step": 794
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.55102352680429e-06,
+      "loss": 1.2839,
+      "step": 795
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.549813083405308e-06,
+      "loss": 1.1843,
+      "step": 796
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.54860108743453e-06,
+      "loss": 1.2062,
+      "step": 797
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.547387539305538e-06,
+      "loss": 1.0709,
+      "step": 798
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.546172439432438e-06,
+      "loss": 1.244,
+      "step": 799
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.54495578822987e-06,
+      "loss": 1.2108,
+      "step": 800
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.543737586113002e-06,
+      "loss": 1.083,
+      "step": 801
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.542517833497534e-06,
+      "loss": 1.1109,
+      "step": 802
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.54129653079969e-06,
+      "loss": 1.1094,
+      "step": 803
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.540073678436226e-06,
+      "loss": 1.1897,
+      "step": 804
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.538849276824426e-06,
+      "loss": 1.0547,
+      "step": 805
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.537623326382106e-06,
+      "loss": 1.1057,
+      "step": 806
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.536395827527605e-06,
+      "loss": 1.2911,
+      "step": 807
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.535166780679795e-06,
+      "loss": 1.2297,
+      "step": 808
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.53393618625807e-06,
+      "loss": 1.1094,
+      "step": 809
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.532704044682359e-06,
+      "loss": 1.1041,
+      "step": 810
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.531470356373119e-06,
+      "loss": 1.3639,
+      "step": 811
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.530235121751328e-06,
+      "loss": 1.1091,
+      "step": 812
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.528998341238496e-06,
+      "loss": 1.234,
+      "step": 813
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.527760015256662e-06,
+      "loss": 0.9869,
+      "step": 814
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.52652014422839e-06,
+      "loss": 1.2305,
+      "step": 815
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.525278728576772e-06,
+      "loss": 1.1316,
+      "step": 816
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.524035768725427e-06,
+      "loss": 1.3053,
+      "step": 817
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.5227912650985e-06,
+      "loss": 1.2198,
+      "step": 818
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.521545218120665e-06,
+      "loss": 1.0899,
+      "step": 819
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.52029762821712e-06,
+      "loss": 1.3593,
+      "step": 820
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.519048495813591e-06,
+      "loss": 1.1266,
+      "step": 821
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.51779782133633e-06,
+      "loss": 1.1517,
+      "step": 822
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.516545605212118e-06,
+      "loss": 1.1243,
+      "step": 823
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.515291847868256e-06,
+      "loss": 1.0986,
+      "step": 824
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.514036549732574e-06,
+      "loss": 1.1492,
+      "step": 825
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.512779711233433e-06,
+      "loss": 1.0136,
+      "step": 826
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.511521332799712e-06,
+      "loss": 1.0719,
+      "step": 827
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.510261414860814e-06,
+      "loss": 1.1904,
+      "step": 828
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.50899995784668e-06,
+      "loss": 1.0976,
+      "step": 829
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.507736962187761e-06,
+      "loss": 1.2416,
+      "step": 830
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.506472428315044e-06,
+      "loss": 1.121,
+      "step": 831
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.505206356660032e-06,
+      "loss": 1.1269,
+      "step": 832
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.503938747654764e-06,
+      "loss": 1.0577,
+      "step": 833
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.502669601731792e-06,
+      "loss": 1.1561,
+      "step": 834
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.5013989193242e-06,
+      "loss": 1.2525,
+      "step": 835
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.500126700865593e-06,
+      "loss": 1.0074,
+      "step": 836
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4988529467901e-06,
+      "loss": 1.2007,
+      "step": 837
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.497577657532377e-06,
+      "loss": 1.1101,
+      "step": 838
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.4963008335276e-06,
+      "loss": 1.0094,
+      "step": 839
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.495022475211472e-06,
+      "loss": 1.1945,
+      "step": 840
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49374258302022e-06,
+      "loss": 1.2689,
+      "step": 841
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.49246115739059e-06,
+      "loss": 1.2131,
+      "step": 842
+    },
+    {
+      "epoch": 0.15,
+      "learning_rate": 9.491178198759854e-06,
+      "loss": 1.1851,
+      "step": 843
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.489893707565804e-06,
+      "loss": 1.1308,
+      "step": 844
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.488607684246766e-06,
+      "loss": 1.0792,
+      "step": 845
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.487320129241574e-06,
+      "loss": 1.0238,
+      "step": 846
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.486031042989594e-06,
+      "loss": 1.0993,
+      "step": 847
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.484740425930713e-06,
+      "loss": 1.331,
+      "step": 848
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.483448278505334e-06,
+      "loss": 1.2391,
+      "step": 849
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.482154601154393e-06,
+      "loss": 1.2799,
+      "step": 850
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.480859394319343e-06,
+      "loss": 1.3826,
+      "step": 851
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.479562658442157e-06,
+      "loss": 1.1754,
+      "step": 852
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.478264393965329e-06,
+      "loss": 1.1491,
+      "step": 853
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.476964601331878e-06,
+      "loss": 1.2103,
+      "step": 854
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.475663280985345e-06,
+      "loss": 1.1871,
+      "step": 855
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.474360433369792e-06,
+      "loss": 1.0699,
+      "step": 856
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.473056058929797e-06,
+      "loss": 1.18,
+      "step": 857
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.471750158110466e-06,
+      "loss": 1.2841,
+      "step": 858
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.47044273135742e-06,
+      "loss": 1.1626,
+      "step": 859
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.469133779116806e-06,
+      "loss": 1.0775,
+      "step": 860
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.467823301835289e-06,
+      "loss": 1.0143,
+      "step": 861
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.46651129996005e-06,
+      "loss": 1.1655,
+      "step": 862
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.465197773938802e-06,
+      "loss": 1.1764,
+      "step": 863
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.463882724219763e-06,
+      "loss": 1.2637,
+      "step": 864
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.462566151251687e-06,
+      "loss": 1.2789,
+      "step": 865
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.461248055483833e-06,
+      "loss": 1.1683,
+      "step": 866
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.459928437365988e-06,
+      "loss": 1.2251,
+      "step": 867
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.458607297348457e-06,
+      "loss": 1.2166,
+      "step": 868
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.457284635882063e-06,
+      "loss": 1.0964,
+      "step": 869
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.45596045341815e-06,
+      "loss": 1.2982,
+      "step": 870
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.454634750408583e-06,
+      "loss": 1.1412,
+      "step": 871
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.453307527305737e-06,
+      "loss": 1.1008,
+      "step": 872
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.451978784562516e-06,
+      "loss": 1.1208,
+      "step": 873
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.450648522632337e-06,
+      "loss": 1.0698,
+      "step": 874
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.449316741969137e-06,
+      "loss": 1.0126,
+      "step": 875
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.44798344302737e-06,
+      "loss": 1.1195,
+      "step": 876
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.44664862626201e-06,
+      "loss": 1.1245,
+      "step": 877
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.445312292128548e-06,
+      "loss": 1.1837,
+      "step": 878
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.443974441082994e-06,
+      "loss": 1.1908,
+      "step": 879
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.442635073581871e-06,
+      "loss": 1.1995,
+      "step": 880
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.441294190082226e-06,
+      "loss": 1.252,
+      "step": 881
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.43995179104162e-06,
+      "loss": 1.1758,
+      "step": 882
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.438607876918129e-06,
+      "loss": 1.0988,
+      "step": 883
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.43726244817035e-06,
+      "loss": 1.2129,
+      "step": 884
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.435915505257396e-06,
+      "loss": 1.2464,
+      "step": 885
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.434567048638894e-06,
+      "loss": 1.1594,
+      "step": 886
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.43321707877499e-06,
+      "loss": 1.2506,
+      "step": 887
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.431865596126346e-06,
+      "loss": 1.109,
+      "step": 888
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.430512601154141e-06,
+      "loss": 1.1847,
+      "step": 889
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.429158094320068e-06,
+      "loss": 1.1232,
+      "step": 890
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.427802076086337e-06,
+      "loss": 1.1769,
+      "step": 891
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.426444546915674e-06,
+      "loss": 1.1898,
+      "step": 892
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.425085507271321e-06,
+      "loss": 1.0565,
+      "step": 893
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.423724957617035e-06,
+      "loss": 1.1636,
+      "step": 894
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.422362898417085e-06,
+      "loss": 1.2764,
+      "step": 895
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.420999330136262e-06,
+      "loss": 1.0226,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "eval_loss": 1.9780372381210327,
+      "eval_runtime": 0.1728,
+      "eval_samples_per_second": 17.361,
+      "eval_steps_per_second": 17.361,
+      "step": 896
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.419634253239866e-06,
+      "loss": 1.1937,
+      "step": 897
+    },
+    {
+      "epoch": 0.16,
+      "learning_rate": 9.418267668193715e-06,
+      "loss": 1.0421,
+      "step": 898
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.416899575464139e-06,
+      "loss": 1.1805,
+      "step": 899
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.415529975517985e-06,
+      "loss": 1.2307,
+      "step": 900
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.414158868822615e-06,
+      "loss": 1.1098,
+      "step": 901
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.412786255845898e-06,
+      "loss": 1.103,
+      "step": 902
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.41141213705623e-06,
+      "loss": 1.1775,
+      "step": 903
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.410036512922509e-06,
+      "loss": 1.4165,
+      "step": 904
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.40865938391415e-06,
+      "loss": 1.2691,
+      "step": 905
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.407280750501086e-06,
+      "loss": 1.0102,
+      "step": 906
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.405900613153756e-06,
+      "loss": 1.2469,
+      "step": 907
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.404518972343122e-06,
+      "loss": 1.1704,
+      "step": 908
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.403135828540646e-06,
+      "loss": 1.2512,
+      "step": 909
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.401751182218315e-06,
+      "loss": 1.089,
+      "step": 910
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.400365033848622e-06,
+      "loss": 1.213,
+      "step": 911
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.398977383904574e-06,
+      "loss": 1.3818,
+      "step": 912
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.397588232859693e-06,
+      "loss": 1.1193,
+      "step": 913
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.39619758118801e-06,
+      "loss": 1.0925,
+      "step": 914
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.394805429364066e-06,
+      "loss": 1.1466,
+      "step": 915
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.39341177786292e-06,
+      "loss": 1.0696,
+      "step": 916
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.39201662716014e-06,
+      "loss": 1.2952,
+      "step": 917
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.390619977731801e-06,
+      "loss": 1.1582,
+      "step": 918
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.3892218300545e-06,
+      "loss": 1.162,
+      "step": 919
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.387822184605332e-06,
+      "loss": 1.267,
+      "step": 920
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.386421041861918e-06,
+      "loss": 1.1487,
+      "step": 921
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.385018402302374e-06,
+      "loss": 1.1385,
+      "step": 922
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.383614266405343e-06,
+      "loss": 1.3603,
+      "step": 923
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.382208634649963e-06,
+      "loss": 1.0672,
+      "step": 924
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.380801507515894e-06,
+      "loss": 1.1954,
+      "step": 925
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.379392885483298e-06,
+      "loss": 1.3153,
+      "step": 926
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.377982769032856e-06,
+      "loss": 1.0408,
+      "step": 927
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.376571158645752e-06,
+      "loss": 1.1787,
+      "step": 928
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.375158054803682e-06,
+      "loss": 0.997,
+      "step": 929
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.37374345798885e-06,
+      "loss": 1.1793,
+      "step": 930
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.372327368683974e-06,
+      "loss": 1.2285,
+      "step": 931
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.370909787372278e-06,
+      "loss": 1.0854,
+      "step": 932
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.369490714537492e-06,
+      "loss": 1.0963,
+      "step": 933
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.368070150663862e-06,
+      "loss": 1.0182,
+      "step": 934
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.366648096236139e-06,
+      "loss": 1.2129,
+      "step": 935
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.36522455173958e-06,
+      "loss": 1.2195,
+      "step": 936
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.363799517659955e-06,
+      "loss": 1.2944,
+      "step": 937
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.362372994483542e-06,
+      "loss": 1.091,
+      "step": 938
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.360944982697125e-06,
+      "loss": 1.132,
+      "step": 939
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.359515482787996e-06,
+      "loss": 1.0921,
+      "step": 940
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.358084495243953e-06,
+      "loss": 1.1509,
+      "step": 941
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.356652020553309e-06,
+      "loss": 1.1298,
+      "step": 942
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.355218059204878e-06,
+      "loss": 1.1487,
+      "step": 943
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.353782611687983e-06,
+      "loss": 1.1769,
+      "step": 944
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.352345678492452e-06,
+      "loss": 1.2512,
+      "step": 945
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.350907260108624e-06,
+      "loss": 1.1559,
+      "step": 946
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.349467357027342e-06,
+      "loss": 1.0064,
+      "step": 947
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.348025969739958e-06,
+      "loss": 1.1113,
+      "step": 948
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.346583098738329e-06,
+      "loss": 1.2552,
+      "step": 949
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.345138744514817e-06,
+      "loss": 1.1534,
+      "step": 950
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.343692907562289e-06,
+      "loss": 1.1502,
+      "step": 951
+    },
+    {
+      "epoch": 0.17,
+      "learning_rate": 9.342245588374127e-06,
+      "loss": 1.1743,
+      "step": 952
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.340796787444204e-06,
+      "loss": 1.3048,
+      "step": 953
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.339346505266913e-06,
+      "loss": 1.153,
+      "step": 954
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.337894742337145e-06,
+      "loss": 1.0057,
+      "step": 955
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.336441499150295e-06,
+      "loss": 1.2049,
+      "step": 956
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.334986776202268e-06,
+      "loss": 1.4413,
+      "step": 957
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.333530573989471e-06,
+      "loss": 1.3571,
+      "step": 958
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.332072893008817e-06,
+      "loss": 1.1735,
+      "step": 959
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.330613733757722e-06,
+      "loss": 1.1311,
+      "step": 960
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.329153096734107e-06,
+      "loss": 1.0931,
+      "step": 961
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.3276909824364e-06,
+      "loss": 1.1143,
+      "step": 962
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.32622739136353e-06,
+      "loss": 1.1184,
+      "step": 963
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.324762324014927e-06,
+      "loss": 1.0661,
+      "step": 964
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.323295780890534e-06,
+      "loss": 1.268,
+      "step": 965
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.321827762490789e-06,
+      "loss": 1.2296,
+      "step": 966
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.320358269316638e-06,
+      "loss": 1.1834,
+      "step": 967
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.318887301869526e-06,
+      "loss": 1.2064,
+      "step": 968
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.317414860651407e-06,
+      "loss": 1.1859,
+      "step": 969
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.315940946164734e-06,
+      "loss": 1.1059,
+      "step": 970
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.314465558912464e-06,
+      "loss": 1.151,
+      "step": 971
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.312988699398053e-06,
+      "loss": 1.1309,
+      "step": 972
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.311510368125467e-06,
+      "loss": 1.1158,
+      "step": 973
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.310030565599167e-06,
+      "loss": 1.0795,
+      "step": 974
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.308549292324118e-06,
+      "loss": 1.2535,
+      "step": 975
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.307066548805789e-06,
+      "loss": 1.2098,
+      "step": 976
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.305582335550149e-06,
+      "loss": 1.1127,
+      "step": 977
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.304096653063668e-06,
+      "loss": 1.1195,
+      "step": 978
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.30260950185332e-06,
+      "loss": 1.1579,
+      "step": 979
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.301120882426578e-06,
+      "loss": 1.3273,
+      "step": 980
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.299630795291415e-06,
+      "loss": 1.3567,
+      "step": 981
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.298139240956306e-06,
+      "loss": 1.1072,
+      "step": 982
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.29664621993023e-06,
+      "loss": 1.1769,
+      "step": 983
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.29515173272266e-06,
+      "loss": 1.1271,
+      "step": 984
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.293655779843574e-06,
+      "loss": 1.118,
+      "step": 985
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.29215836180345e-06,
+      "loss": 1.1606,
+      "step": 986
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.290659479113262e-06,
+      "loss": 1.2188,
+      "step": 987
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.289159132284488e-06,
+      "loss": 1.0263,
+      "step": 988
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.287657321829106e-06,
+      "loss": 1.1782,
+      "step": 989
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.28615404825959e-06,
+      "loss": 1.1889,
+      "step": 990
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.284649312088917e-06,
+      "loss": 1.1398,
+      "step": 991
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.283143113830556e-06,
+      "loss": 1.2103,
+      "step": 992
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.281635453998482e-06,
+      "loss": 1.3194,
+      "step": 993
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.280126333107173e-06,
+      "loss": 1.1361,
+      "step": 994
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.278615751671592e-06,
+      "loss": 1.0853,
+      "step": 995
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.277103710207209e-06,
+      "loss": 1.24,
+      "step": 996
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.275590209229992e-06,
+      "loss": 1.072,
+      "step": 997
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.274075249256404e-06,
+      "loss": 1.184,
+      "step": 998
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.272558830803411e-06,
+      "loss": 1.1347,
+      "step": 999
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.271040954388472e-06,
+      "loss": 1.2494,
+      "step": 1000
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.269521620529546e-06,
+      "loss": 1.2061,
+      "step": 1001
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.268000829745086e-06,
+      "loss": 1.105,
+      "step": 1002
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.266478582554047e-06,
+      "loss": 1.0552,
+      "step": 1003
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.264954879475876e-06,
+      "loss": 1.0738,
+      "step": 1004
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.263429721030521e-06,
+      "loss": 1.0808,
+      "step": 1005
+    },
+    {
+      "epoch": 0.18,
+      "learning_rate": 9.261903107738425e-06,
+      "loss": 1.1511,
+      "step": 1006
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.260375040120526e-06,
+      "loss": 1.2108,
+      "step": 1007
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.258845518698262e-06,
+      "loss": 1.0012,
+      "step": 1008
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.257314543993561e-06,
+      "loss": 1.1725,
+      "step": 1009
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.255782116528855e-06,
+      "loss": 1.1908,
+      "step": 1010
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.254248236827064e-06,
+      "loss": 1.0271,
+      "step": 1011
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.25271290541161e-06,
+      "loss": 1.2108,
+      "step": 1012
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.251176122806404e-06,
+      "loss": 1.2448,
+      "step": 1013
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.249637889535856e-06,
+      "loss": 1.0906,
+      "step": 1014
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.248098206124873e-06,
+      "loss": 1.1657,
+      "step": 1015
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.246557073098855e-06,
+      "loss": 1.3352,
+      "step": 1016
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.24501449098369e-06,
+      "loss": 1.1872,
+      "step": 1017
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.243470460305773e-06,
+      "loss": 1.1686,
+      "step": 1018
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.241924981591986e-06,
+      "loss": 1.1393,
+      "step": 1019
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.2403780553697e-06,
+      "loss": 1.1803,
+      "step": 1020
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.238829682166795e-06,
+      "loss": 1.2385,
+      "step": 1021
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.237279862511627e-06,
+      "loss": 1.1163,
+      "step": 1022
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.235728596933063e-06,
+      "loss": 1.2127,
+      "step": 1023
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.234175885960446e-06,
+      "loss": 1.1515,
+      "step": 1024
+    },
+    {
+      "epoch": 0.19,
+      "eval_loss": 1.973323941230774,
+      "eval_runtime": 0.1809,
+      "eval_samples_per_second": 16.588,
+      "eval_steps_per_second": 16.588,
+      "step": 1024
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.232621730123626e-06,
+      "loss": 0.9936,
+      "step": 1025
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.23106612995294e-06,
+      "loss": 1.1898,
+      "step": 1026
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.229509085979221e-06,
+      "loss": 1.2879,
+      "step": 1027
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.227950598733786e-06,
+      "loss": 1.1412,
+      "step": 1028
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.226390668748458e-06,
+      "loss": 1.2992,
+      "step": 1029
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.224829296555542e-06,
+      "loss": 1.23,
+      "step": 1030
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.223266482687838e-06,
+      "loss": 1.0043,
+      "step": 1031
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.22170222767864e-06,
+      "loss": 1.2171,
+      "step": 1032
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.22013653206173e-06,
+      "loss": 1.1545,
+      "step": 1033
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.218569396371385e-06,
+      "loss": 1.1218,
+      "step": 1034
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.217000821142376e-06,
+      "loss": 1.1813,
+      "step": 1035
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.215430806909956e-06,
+      "loss": 1.0986,
+      "step": 1036
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.213859354209876e-06,
+      "loss": 1.1738,
+      "step": 1037
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.212286463578376e-06,
+      "loss": 0.968,
+      "step": 1038
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.210712135552186e-06,
+      "loss": 1.1096,
+      "step": 1039
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.209136370668533e-06,
+      "loss": 1.1723,
+      "step": 1040
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.20755916946512e-06,
+      "loss": 1.1816,
+      "step": 1041
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.205980532480157e-06,
+      "loss": 1.1398,
+      "step": 1042
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.204400460252331e-06,
+      "loss": 1.2291,
+      "step": 1043
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.202818953320827e-06,
+      "loss": 1.1687,
+      "step": 1044
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.201236012225314e-06,
+      "loss": 1.021,
+      "step": 1045
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.199651637505951e-06,
+      "loss": 1.062,
+      "step": 1046
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.198065829703392e-06,
+      "loss": 1.1824,
+      "step": 1047
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.196478589358773e-06,
+      "loss": 1.173,
+      "step": 1048
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.19488991701372e-06,
+      "loss": 1.1325,
+      "step": 1049
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.193299813210355e-06,
+      "loss": 1.1869,
+      "step": 1050
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.191708278491276e-06,
+      "loss": 1.1216,
+      "step": 1051
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.190115313399581e-06,
+      "loss": 1.2929,
+      "step": 1052
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.18852091847885e-06,
+      "loss": 1.181,
+      "step": 1053
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.186925094273151e-06,
+      "loss": 1.0956,
+      "step": 1054
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.18532784132704e-06,
+      "loss": 1.1724,
+      "step": 1055
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.183729160185566e-06,
+      "loss": 1.2618,
+      "step": 1056
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.182129051394255e-06,
+      "loss": 1.1928,
+      "step": 1057
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.180527515499129e-06,
+      "loss": 1.1123,
+      "step": 1058
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.178924553046693e-06,
+      "loss": 1.127,
+      "step": 1059
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.177320164583939e-06,
+      "loss": 1.0333,
+      "step": 1060
+    },
+    {
+      "epoch": 0.19,
+      "learning_rate": 9.175714350658348e-06,
+      "loss": 1.006,
+      "step": 1061
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.174107111817885e-06,
+      "loss": 1.1504,
+      "step": 1062
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.172498448611e-06,
+      "loss": 1.1131,
+      "step": 1063
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.170888361586633e-06,
+      "loss": 1.218,
+      "step": 1064
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.169276851294209e-06,
+      "loss": 1.1726,
+      "step": 1065
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.167663918283635e-06,
+      "loss": 1.1107,
+      "step": 1066
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.166049563105309e-06,
+      "loss": 1.0941,
+      "step": 1067
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.164433786310106e-06,
+      "loss": 1.2442,
+      "step": 1068
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.162816588449394e-06,
+      "loss": 1.2596,
+      "step": 1069
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.161197970075026e-06,
+      "loss": 1.0113,
+      "step": 1070
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.159577931739335e-06,
+      "loss": 1.0041,
+      "step": 1071
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.15795647399514e-06,
+      "loss": 1.2963,
+      "step": 1072
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.156333597395745e-06,
+      "loss": 1.184,
+      "step": 1073
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.154709302494939e-06,
+      "loss": 1.0844,
+      "step": 1074
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.153083589846994e-06,
+      "loss": 1.172,
+      "step": 1075
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.151456460006667e-06,
+      "loss": 1.1861,
+      "step": 1076
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.149827913529193e-06,
+      "loss": 1.053,
+      "step": 1077
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.1481979509703e-06,
+      "loss": 1.2195,
+      "step": 1078
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.14656657288619e-06,
+      "loss": 1.2834,
+      "step": 1079
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.144933779833558e-06,
+      "loss": 1.1295,
+      "step": 1080
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.14329957236957e-06,
+      "loss": 1.1936,
+      "step": 1081
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.141663951051883e-06,
+      "loss": 1.0675,
+      "step": 1082
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.140026916438634e-06,
+      "loss": 1.2226,
+      "step": 1083
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.138388469088441e-06,
+      "loss": 1.0673,
+      "step": 1084
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.13674860956041e-06,
+      "loss": 1.1223,
+      "step": 1085
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.13510733841412e-06,
+      "loss": 1.1238,
+      "step": 1086
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.133464656209636e-06,
+      "loss": 1.2757,
+      "step": 1087
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.131820563507507e-06,
+      "loss": 0.9722,
+      "step": 1088
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.13017506086876e-06,
+      "loss": 1.1134,
+      "step": 1089
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.128528148854903e-06,
+      "loss": 1.1991,
+      "step": 1090
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.126879828027927e-06,
+      "loss": 1.2049,
+      "step": 1091
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.1252300989503e-06,
+      "loss": 0.9775,
+      "step": 1092
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.123578962184977e-06,
+      "loss": 1.1543,
+      "step": 1093
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.12192641829539e-06,
+      "loss": 1.203,
+      "step": 1094
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.120272467845448e-06,
+      "loss": 1.1464,
+      "step": 1095
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.118617111399543e-06,
+      "loss": 1.1438,
+      "step": 1096
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.116960349522545e-06,
+      "loss": 1.2065,
+      "step": 1097
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.115302182779811e-06,
+      "loss": 1.1193,
+      "step": 1098
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.113642611737166e-06,
+      "loss": 1.0264,
+      "step": 1099
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.11198163696092e-06,
+      "loss": 1.0741,
+      "step": 1100
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.110319259017866e-06,
+      "loss": 1.1187,
+      "step": 1101
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.108655478475266e-06,
+      "loss": 1.286,
+      "step": 1102
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.106990295900869e-06,
+      "loss": 1.1807,
+      "step": 1103
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.1053237118629e-06,
+      "loss": 1.1463,
+      "step": 1104
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.10365572693006e-06,
+      "loss": 1.3005,
+      "step": 1105
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.101986341671531e-06,
+      "loss": 1.1846,
+      "step": 1106
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.10031555665697e-06,
+      "loss": 1.119,
+      "step": 1107
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.098643372456518e-06,
+      "loss": 1.1362,
+      "step": 1108
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.096969789640785e-06,
+      "loss": 1.137,
+      "step": 1109
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.09529480878086e-06,
+      "loss": 1.3932,
+      "step": 1110
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.093618430448317e-06,
+      "loss": 1.0722,
+      "step": 1111
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.091940655215194e-06,
+      "loss": 1.0448,
+      "step": 1112
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.090261483654019e-06,
+      "loss": 1.0919,
+      "step": 1113
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.088580916337788e-06,
+      "loss": 1.1418,
+      "step": 1114
+    },
+    {
+      "epoch": 0.2,
+      "learning_rate": 9.086898953839974e-06,
+      "loss": 1.2664,
+      "step": 1115
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.085215596734527e-06,
+      "loss": 1.0764,
+      "step": 1116
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.083530845595875e-06,
+      "loss": 1.0379,
+      "step": 1117
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.081844700998921e-06,
+      "loss": 1.2222,
+      "step": 1118
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.080157163519042e-06,
+      "loss": 1.2278,
+      "step": 1119
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.07846823373209e-06,
+      "loss": 1.2582,
+      "step": 1120
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.076777912214392e-06,
+      "loss": 1.055,
+      "step": 1121
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.075086199542753e-06,
+      "loss": 1.1465,
+      "step": 1122
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.07339309629445e-06,
+      "loss": 1.128,
+      "step": 1123
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.071698603047236e-06,
+      "loss": 1.0443,
+      "step": 1124
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.070002720379336e-06,
+      "loss": 1.144,
+      "step": 1125
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.068305448869452e-06,
+      "loss": 1.2614,
+      "step": 1126
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.066606789096756e-06,
+      "loss": 1.0741,
+      "step": 1127
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.0649067416409e-06,
+      "loss": 1.1749,
+      "step": 1128
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.063205307082004e-06,
+      "loss": 1.2126,
+      "step": 1129
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.061502486000661e-06,
+      "loss": 1.1707,
+      "step": 1130
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.059798278977942e-06,
+      "loss": 1.2344,
+      "step": 1131
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.058092686595388e-06,
+      "loss": 1.0615,
+      "step": 1132
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.056385709435012e-06,
+      "loss": 1.2301,
+      "step": 1133
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.054677348079301e-06,
+      "loss": 1.1806,
+      "step": 1134
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.052967603111214e-06,
+      "loss": 1.1292,
+      "step": 1135
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.051256475114182e-06,
+      "loss": 1.1181,
+      "step": 1136
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.049543964672106e-06,
+      "loss": 1.1342,
+      "step": 1137
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.047830072369364e-06,
+      "loss": 1.1179,
+      "step": 1138
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.046114798790799e-06,
+      "loss": 1.1798,
+      "step": 1139
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.04439814452173e-06,
+      "loss": 1.1197,
+      "step": 1140
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.042680110147945e-06,
+      "loss": 1.4524,
+      "step": 1141
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.040960696255705e-06,
+      "loss": 1.0201,
+      "step": 1142
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.03923990343174e-06,
+      "loss": 1.2206,
+      "step": 1143
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.03751773226325e-06,
+      "loss": 1.2346,
+      "step": 1144
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.035794183337907e-06,
+      "loss": 1.2248,
+      "step": 1145
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.034069257243853e-06,
+      "loss": 1.194,
+      "step": 1146
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.032342954569697e-06,
+      "loss": 1.015,
+      "step": 1147
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.030615275904522e-06,
+      "loss": 1.2091,
+      "step": 1148
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.028886221837877e-06,
+      "loss": 1.0505,
+      "step": 1149
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.027155792959786e-06,
+      "loss": 1.1083,
+      "step": 1150
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.025423989860733e-06,
+      "loss": 1.2897,
+      "step": 1151
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.02369081313168e-06,
+      "loss": 1.3467,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "eval_loss": 1.9454926252365112,
+      "eval_runtime": 0.1699,
+      "eval_samples_per_second": 17.655,
+      "eval_steps_per_second": 17.655,
+      "step": 1152
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.021956263364052e-06,
+      "loss": 1.1168,
+      "step": 1153
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.020220341149743e-06,
+      "loss": 1.1191,
+      "step": 1154
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.018483047081119e-06,
+      "loss": 1.0124,
+      "step": 1155
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.01674438175101e-06,
+      "loss": 1.2893,
+      "step": 1156
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.015004345752718e-06,
+      "loss": 1.2312,
+      "step": 1157
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.013262939680008e-06,
+      "loss": 1.07,
+      "step": 1158
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.011520164127116e-06,
+      "loss": 1.2274,
+      "step": 1159
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.009776019688745e-06,
+      "loss": 1.225,
+      "step": 1160
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.008030506960062e-06,
+      "loss": 1.1016,
+      "step": 1161
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.006283626536707e-06,
+      "loss": 1.0846,
+      "step": 1162
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.00453537901478e-06,
+      "loss": 1.1734,
+      "step": 1163
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.00278576499085e-06,
+      "loss": 1.1193,
+      "step": 1164
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 9.001034785061956e-06,
+      "loss": 1.1315,
+      "step": 1165
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.999282439825598e-06,
+      "loss": 1.1787,
+      "step": 1166
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.997528729879744e-06,
+      "loss": 1.2521,
+      "step": 1167
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.995773655822826e-06,
+      "loss": 1.1984,
+      "step": 1168
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.994017218253745e-06,
+      "loss": 1.2599,
+      "step": 1169
+    },
+    {
+      "epoch": 0.21,
+      "learning_rate": 8.992259417771867e-06,
+      "loss": 1.2133,
+      "step": 1170
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.990500254977017e-06,
+      "loss": 1.0163,
+      "step": 1171
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.988739730469492e-06,
+      "loss": 1.2006,
+      "step": 1172
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.98697784485005e-06,
+      "loss": 1.1516,
+      "step": 1173
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.985214598719916e-06,
+      "loss": 1.1765,
+      "step": 1174
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.983449992680774e-06,
+      "loss": 1.1095,
+      "step": 1175
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.981684027334778e-06,
+      "loss": 1.1437,
+      "step": 1176
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.979916703284545e-06,
+      "loss": 1.2332,
+      "step": 1177
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.978148021133151e-06,
+      "loss": 1.2517,
+      "step": 1178
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.976377981484142e-06,
+      "loss": 1.1347,
+      "step": 1179
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.974606584941521e-06,
+      "loss": 1.0865,
+      "step": 1180
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.972833832109758e-06,
+      "loss": 1.1033,
+      "step": 1181
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.971059723593785e-06,
+      "loss": 1.1895,
+      "step": 1182
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.969284259998996e-06,
+      "loss": 1.1941,
+      "step": 1183
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.967507441931247e-06,
+      "loss": 1.0575,
+      "step": 1184
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.965729269996857e-06,
+      "loss": 1.1399,
+      "step": 1185
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.963949744802608e-06,
+      "loss": 1.1625,
+      "step": 1186
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.962168866955741e-06,
+      "loss": 1.1302,
+      "step": 1187
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.960386637063962e-06,
+      "loss": 1.2807,
+      "step": 1188
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.958603055735434e-06,
+      "loss": 1.019,
+      "step": 1189
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.95681812357879e-06,
+      "loss": 1.0751,
+      "step": 1190
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.95503184120311e-06,
+      "loss": 1.1899,
+      "step": 1191
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.953244209217946e-06,
+      "loss": 1.1057,
+      "step": 1192
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.951455228233308e-06,
+      "loss": 1.1401,
+      "step": 1193
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.949664898859664e-06,
+      "loss": 1.1357,
+      "step": 1194
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.947873221707944e-06,
+      "loss": 1.0816,
+      "step": 1195
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.946080197389535e-06,
+      "loss": 1.1367,
+      "step": 1196
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.944285826516289e-06,
+      "loss": 1.1829,
+      "step": 1197
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.942490109700515e-06,
+      "loss": 1.1521,
+      "step": 1198
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.940693047554977e-06,
+      "loss": 1.2629,
+      "step": 1199
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.938894640692906e-06,
+      "loss": 1.0794,
+      "step": 1200
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.937094889727987e-06,
+      "loss": 1.2061,
+      "step": 1201
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.935293795274364e-06,
+      "loss": 1.158,
+      "step": 1202
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.933491357946637e-06,
+      "loss": 1.1467,
+      "step": 1203
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.931687578359874e-06,
+      "loss": 1.2367,
+      "step": 1204
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.929882457129588e-06,
+      "loss": 1.1968,
+      "step": 1205
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.928075994871757e-06,
+      "loss": 1.0191,
+      "step": 1206
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.926268192202818e-06,
+      "loss": 1.245,
+      "step": 1207
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.924459049739662e-06,
+      "loss": 1.1004,
+      "step": 1208
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.922648568099639e-06,
+      "loss": 1.1064,
+      "step": 1209
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.920836747900553e-06,
+      "loss": 0.9679,
+      "step": 1210
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.919023589760667e-06,
+      "loss": 1.2367,
+      "step": 1211
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.917209094298704e-06,
+      "loss": 1.3258,
+      "step": 1212
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.915393262133836e-06,
+      "loss": 1.1511,
+      "step": 1213
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.913576093885698e-06,
+      "loss": 1.1531,
+      "step": 1214
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.911757590174374e-06,
+      "loss": 1.0551,
+      "step": 1215
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.909937751620412e-06,
+      "loss": 1.1836,
+      "step": 1216
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.90811657884481e-06,
+      "loss": 1.076,
+      "step": 1217
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.906294072469022e-06,
+      "loss": 1.1302,
+      "step": 1218
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.904470233114955e-06,
+      "loss": 1.1973,
+      "step": 1219
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.902645061404977e-06,
+      "loss": 1.1994,
+      "step": 1220
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.900818557961906e-06,
+      "loss": 1.1874,
+      "step": 1221
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.898990723409014e-06,
+      "loss": 1.0153,
+      "step": 1222
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.89716155837003e-06,
+      "loss": 1.117,
+      "step": 1223
+    },
+    {
+      "epoch": 0.22,
+      "learning_rate": 8.895331063469137e-06,
+      "loss": 0.9954,
+      "step": 1224
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.893499239330968e-06,
+      "loss": 1.0404,
+      "step": 1225
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.89166608658061e-06,
+      "loss": 1.0914,
+      "step": 1226
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.889831605843613e-06,
+      "loss": 1.1284,
+      "step": 1227
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.887995797745965e-06,
+      "loss": 1.3283,
+      "step": 1228
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.886158662914119e-06,
+      "loss": 1.0938,
+      "step": 1229
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.884320201974976e-06,
+      "loss": 1.1358,
+      "step": 1230
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.882480415555887e-06,
+      "loss": 1.1591,
+      "step": 1231
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.880639304284661e-06,
+      "loss": 1.18,
+      "step": 1232
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.878796868789554e-06,
+      "loss": 1.2439,
+      "step": 1233
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.87695310969928e-06,
+      "loss": 1.0689,
+      "step": 1234
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.875108027642996e-06,
+      "loss": 1.1046,
+      "step": 1235
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.873261623250317e-06,
+      "loss": 1.1454,
+      "step": 1236
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.871413897151307e-06,
+      "loss": 1.2152,
+      "step": 1237
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.869564849976484e-06,
+      "loss": 1.2605,
+      "step": 1238
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.867714482356811e-06,
+      "loss": 1.2258,
+      "step": 1239
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.865862794923706e-06,
+      "loss": 1.2078,
+      "step": 1240
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.864009788309035e-06,
+      "loss": 1.1548,
+      "step": 1241
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.862155463145118e-06,
+      "loss": 1.107,
+      "step": 1242
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.860299820064722e-06,
+      "loss": 1.1667,
+      "step": 1243
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.858442859701063e-06,
+      "loss": 1.1688,
+      "step": 1244
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.856584582687805e-06,
+      "loss": 1.2989,
+      "step": 1245
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.85472498965907e-06,
+      "loss": 1.0521,
+      "step": 1246
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.852864081249417e-06,
+      "loss": 1.2097,
+      "step": 1247
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.851001858093862e-06,
+      "loss": 1.1657,
+      "step": 1248
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.84913832082787e-06,
+      "loss": 1.0532,
+      "step": 1249
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.847273470087347e-06,
+      "loss": 1.1334,
+      "step": 1250
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.845407306508654e-06,
+      "loss": 1.2275,
+      "step": 1251
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.843539830728597e-06,
+      "loss": 1.1363,
+      "step": 1252
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.841671043384437e-06,
+      "loss": 1.2504,
+      "step": 1253
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.839800945113868e-06,
+      "loss": 1.1959,
+      "step": 1254
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.837929536555043e-06,
+      "loss": 1.0539,
+      "step": 1255
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.83605681834656e-06,
+      "loss": 1.156,
+      "step": 1256
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.834182791127462e-06,
+      "loss": 1.1139,
+      "step": 1257
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.832307455537237e-06,
+      "loss": 1.1563,
+      "step": 1258
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.830430812215827e-06,
+      "loss": 1.0765,
+      "step": 1259
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.82855286180361e-06,
+      "loss": 1.1941,
+      "step": 1260
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.826673604941418e-06,
+      "loss": 1.0199,
+      "step": 1261
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.824793042270525e-06,
+      "loss": 1.1309,
+      "step": 1262
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.822911174432652e-06,
+      "loss": 1.4063,
+      "step": 1263
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.821028002069965e-06,
+      "loss": 1.2552,
+      "step": 1264
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.819143525825073e-06,
+      "loss": 1.1962,
+      "step": 1265
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.817257746341037e-06,
+      "loss": 1.0757,
+      "step": 1266
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.815370664261352e-06,
+      "loss": 1.1788,
+      "step": 1267
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.813482280229967e-06,
+      "loss": 1.315,
+      "step": 1268
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.81159259489127e-06,
+      "loss": 1.1239,
+      "step": 1269
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.809701608890094e-06,
+      "loss": 1.2263,
+      "step": 1270
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.807809322871718e-06,
+      "loss": 1.0887,
+      "step": 1271
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.805915737481862e-06,
+      "loss": 1.1031,
+      "step": 1272
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.80402085336669e-06,
+      "loss": 1.0093,
+      "step": 1273
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.802124671172811e-06,
+      "loss": 1.1347,
+      "step": 1274
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.800227191547275e-06,
+      "loss": 1.0279,
+      "step": 1275
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.798328415137573e-06,
+      "loss": 1.1263,
+      "step": 1276
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.796428342591645e-06,
+      "loss": 1.1927,
+      "step": 1277
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.794526974557867e-06,
+      "loss": 1.0967,
+      "step": 1278
+    },
+    {
+      "epoch": 0.23,
+      "learning_rate": 8.792624311685057e-06,
+      "loss": 1.1165,
+      "step": 1279
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.790720354622481e-06,
+      "loss": 1.2586,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "eval_loss": 1.9353574514389038,
+      "eval_runtime": 0.1829,
+      "eval_samples_per_second": 16.402,
+      "eval_steps_per_second": 16.402,
+      "step": 1280
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.78881510401984e-06,
+      "loss": 1.1454,
+      "step": 1281
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.786908560527279e-06,
+      "loss": 1.2587,
+      "step": 1282
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.785000724795385e-06,
+      "loss": 1.0694,
+      "step": 1283
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.783091597475185e-06,
+      "loss": 1.1955,
+      "step": 1284
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.781181179218148e-06,
+      "loss": 1.193,
+      "step": 1285
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.77926947067618e-06,
+      "loss": 1.2286,
+      "step": 1286
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.777356472501632e-06,
+      "loss": 1.1414,
+      "step": 1287
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.775442185347291e-06,
+      "loss": 1.2514,
+      "step": 1288
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.773526609866388e-06,
+      "loss": 1.0515,
+      "step": 1289
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.771609746712588e-06,
+      "loss": 1.1606,
+      "step": 1290
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.769691596540003e-06,
+      "loss": 1.1778,
+      "step": 1291
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.767772160003175e-06,
+      "loss": 1.2355,
+      "step": 1292
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.765851437757094e-06,
+      "loss": 1.2002,
+      "step": 1293
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.76392943045718e-06,
+      "loss": 1.1131,
+      "step": 1294
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.7620061387593e-06,
+      "loss": 1.109,
+      "step": 1295
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.760081563319755e-06,
+      "loss": 1.2525,
+      "step": 1296
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.758155704795284e-06,
+      "loss": 1.1097,
+      "step": 1297
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.756228563843063e-06,
+      "loss": 1.1183,
+      "step": 1298
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.754300141120707e-06,
+      "loss": 1.1301,
+      "step": 1299
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.752370437286269e-06,
+      "loss": 1.1994,
+      "step": 1300
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.75043945299824e-06,
+      "loss": 1.2964,
+      "step": 1301
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.748507188915542e-06,
+      "loss": 1.316,
+      "step": 1302
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.746573645697542e-06,
+      "loss": 1.2307,
+      "step": 1303
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.744638824004039e-06,
+      "loss": 1.173,
+      "step": 1304
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.742702724495266e-06,
+      "loss": 1.1562,
+      "step": 1305
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.7407653478319e-06,
+      "loss": 1.0764,
+      "step": 1306
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.738826694675045e-06,
+      "loss": 1.0734,
+      "step": 1307
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.736886765686244e-06,
+      "loss": 1.0171,
+      "step": 1308
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.734945561527477e-06,
+      "loss": 1.1531,
+      "step": 1309
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.733003082861156e-06,
+      "loss": 1.1868,
+      "step": 1310
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.731059330350132e-06,
+      "loss": 1.1908,
+      "step": 1311
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.729114304657686e-06,
+      "loss": 1.1,
+      "step": 1312
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.727168006447538e-06,
+      "loss": 1.2471,
+      "step": 1313
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.725220436383841e-06,
+      "loss": 1.2164,
+      "step": 1314
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.723271595131177e-06,
+      "loss": 1.1786,
+      "step": 1315
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.721321483354569e-06,
+      "loss": 1.1139,
+      "step": 1316
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.71937010171947e-06,
+      "loss": 1.1072,
+      "step": 1317
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.717417450891766e-06,
+      "loss": 1.1519,
+      "step": 1318
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.715463531537776e-06,
+      "loss": 1.3356,
+      "step": 1319
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.713508344324254e-06,
+      "loss": 1.1117,
+      "step": 1320
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.711551889918385e-06,
+      "loss": 1.1748,
+      "step": 1321
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.709594168987788e-06,
+      "loss": 1.0843,
+      "step": 1322
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.70763518220051e-06,
+      "loss": 1.1982,
+      "step": 1323
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.705674930225036e-06,
+      "loss": 1.182,
+      "step": 1324
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.703713413730278e-06,
+      "loss": 1.0449,
+      "step": 1325
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.701750633385581e-06,
+      "loss": 1.082,
+      "step": 1326
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.699786589860723e-06,
+      "loss": 1.1452,
+      "step": 1327
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.69782128382591e-06,
+      "loss": 1.1128,
+      "step": 1328
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.69585471595178e-06,
+      "loss": 1.1786,
+      "step": 1329
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.6938868869094e-06,
+      "loss": 1.0928,
+      "step": 1330
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.691917797370274e-06,
+      "loss": 1.1558,
+      "step": 1331
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.689947448006329e-06,
+      "loss": 1.166,
+      "step": 1332
+    },
+    {
+      "epoch": 0.24,
+      "learning_rate": 8.687975839489923e-06,
+      "loss": 1.3106,
+      "step": 1333
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.686002972493846e-06,
+      "loss": 1.1428,
+      "step": 1334
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.684028847691317e-06,
+      "loss": 1.1748,
+      "step": 1335
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.682053465755981e-06,
+      "loss": 1.0939,
+      "step": 1336
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.680076827361918e-06,
+      "loss": 1.1873,
+      "step": 1337
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.678098933183632e-06,
+      "loss": 1.0457,
+      "step": 1338
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.676119783896057e-06,
+      "loss": 0.9656,
+      "step": 1339
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.674139380174554e-06,
+      "loss": 1.0414,
+      "step": 1340
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.672157722694915e-06,
+      "loss": 1.0717,
+      "step": 1341
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.670174812133354e-06,
+      "loss": 1.0814,
+      "step": 1342
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.66819064916652e-06,
+      "loss": 1.1628,
+      "step": 1343
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.666205234471487e-06,
+      "loss": 1.208,
+      "step": 1344
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.664218568725753e-06,
+      "loss": 1.2397,
+      "step": 1345
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.662230652607244e-06,
+      "loss": 1.2005,
+      "step": 1346
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.660241486794317e-06,
+      "loss": 1.2386,
+      "step": 1347
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.658251071965751e-06,
+      "loss": 1.2298,
+      "step": 1348
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.65625940880075e-06,
+      "loss": 1.1423,
+      "step": 1349
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.65426649797895e-06,
+      "loss": 1.204,
+      "step": 1350
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.652272340180404e-06,
+      "loss": 1.2072,
+      "step": 1351
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.6502769360856e-06,
+      "loss": 1.0607,
+      "step": 1352
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.648280286375446e-06,
+      "loss": 1.148,
+      "step": 1353
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.646282391731276e-06,
+      "loss": 1.1338,
+      "step": 1354
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.644283252834849e-06,
+      "loss": 1.264,
+      "step": 1355
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.642282870368345e-06,
+      "loss": 1.1752,
+      "step": 1356
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.640281245014376e-06,
+      "loss": 1.0822,
+      "step": 1357
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.63827837745597e-06,
+      "loss": 1.1769,
+      "step": 1358
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.636274268376588e-06,
+      "loss": 1.3295,
+      "step": 1359
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.634268918460106e-06,
+      "loss": 1.0206,
+      "step": 1360
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.632262328390823e-06,
+      "loss": 1.0874,
+      "step": 1361
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.630254498853471e-06,
+      "loss": 1.1243,
+      "step": 1362
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.628245430533198e-06,
+      "loss": 1.1988,
+      "step": 1363
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.626235124115575e-06,
+      "loss": 1.1402,
+      "step": 1364
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.624223580286593e-06,
+      "loss": 1.119,
+      "step": 1365
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.622210799732672e-06,
+      "loss": 0.9747,
+      "step": 1366
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.620196783140649e-06,
+      "loss": 1.1066,
+      "step": 1367
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.618181531197785e-06,
+      "loss": 1.2252,
+      "step": 1368
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.616165044591761e-06,
+      "loss": 1.2677,
+      "step": 1369
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.614147324010681e-06,
+      "loss": 1.3468,
+      "step": 1370
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.612128370143068e-06,
+      "loss": 1.0986,
+      "step": 1371
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.610108183677867e-06,
+      "loss": 1.1546,
+      "step": 1372
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.608086765304444e-06,
+      "loss": 1.1388,
+      "step": 1373
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.606064115712585e-06,
+      "loss": 0.9252,
+      "step": 1374
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.604040235592497e-06,
+      "loss": 1.1278,
+      "step": 1375
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.602015125634804e-06,
+      "loss": 1.1702,
+      "step": 1376
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.599988786530554e-06,
+      "loss": 1.1383,
+      "step": 1377
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.597961218971208e-06,
+      "loss": 1.1627,
+      "step": 1378
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.595932423648654e-06,
+      "loss": 1.1716,
+      "step": 1379
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.593902401255197e-06,
+      "loss": 1.1861,
+      "step": 1380
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.591871152483553e-06,
+      "loss": 1.192,
+      "step": 1381
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.589838678026866e-06,
+      "loss": 1.0703,
+      "step": 1382
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.587804978578696e-06,
+      "loss": 1.0986,
+      "step": 1383
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.585770054833018e-06,
+      "loss": 1.2086,
+      "step": 1384
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.583733907484226e-06,
+      "loss": 1.1316,
+      "step": 1385
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.581696537227134e-06,
+      "loss": 1.0585,
+      "step": 1386
+    },
+    {
+      "epoch": 0.25,
+      "learning_rate": 8.579657944756967e-06,
+      "loss": 1.2208,
+      "step": 1387
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.577618130769378e-06,
+      "loss": 1.0016,
+      "step": 1388
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.575577095960425e-06,
+      "loss": 1.218,
+      "step": 1389
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.573534841026589e-06,
+      "loss": 1.1655,
+      "step": 1390
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.571491366664769e-06,
+      "loss": 1.1138,
+      "step": 1391
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.569446673572272e-06,
+      "loss": 1.0484,
+      "step": 1392
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.567400762446833e-06,
+      "loss": 1.019,
+      "step": 1393
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.56535363398659e-06,
+      "loss": 1.0438,
+      "step": 1394
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.563305288890104e-06,
+      "loss": 1.1662,
+      "step": 1395
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.56125572785635e-06,
+      "loss": 1.2165,
+      "step": 1396
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.559204951584718e-06,
+      "loss": 1.142,
+      "step": 1397
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.55715296077501e-06,
+      "loss": 1.1356,
+      "step": 1398
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.555099756127446e-06,
+      "loss": 1.1237,
+      "step": 1399
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.553045338342658e-06,
+      "loss": 1.0823,
+      "step": 1400
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.550989708121694e-06,
+      "loss": 1.0845,
+      "step": 1401
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.548932866166012e-06,
+      "loss": 1.0241,
+      "step": 1402
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.546874813177486e-06,
+      "loss": 1.0915,
+      "step": 1403
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.544815549858405e-06,
+      "loss": 1.0886,
+      "step": 1404
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.542755076911468e-06,
+      "loss": 1.0806,
+      "step": 1405
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.540693395039787e-06,
+      "loss": 1.177,
+      "step": 1406
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.538630504946891e-06,
+      "loss": 1.0537,
+      "step": 1407
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.536566407336714e-06,
+      "loss": 1.1514,
+      "step": 1408
+    },
+    {
+      "epoch": 0.26,
+      "eval_loss": 1.933529257774353,
+      "eval_runtime": 0.1685,
+      "eval_samples_per_second": 17.808,
+      "eval_steps_per_second": 17.808,
+      "step": 1408
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.534501102913606e-06,
+      "loss": 1.1917,
+      "step": 1409
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.53243459238233e-06,
+      "loss": 1.1188,
+      "step": 1410
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.53036687644806e-06,
+      "loss": 1.0967,
+      "step": 1411
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.528297955816378e-06,
+      "loss": 1.2864,
+      "step": 1412
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.52622783119328e-06,
+      "loss": 1.2306,
+      "step": 1413
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.524156503285175e-06,
+      "loss": 0.9782,
+      "step": 1414
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.522083972798876e-06,
+      "loss": 0.9745,
+      "step": 1415
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.520010240441614e-06,
+      "loss": 1.173,
+      "step": 1416
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.517935306921024e-06,
+      "loss": 1.139,
+      "step": 1417
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.515859172945153e-06,
+      "loss": 1.0427,
+      "step": 1418
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.513781839222463e-06,
+      "loss": 1.3312,
+      "step": 1419
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.511703306461814e-06,
+      "loss": 1.1693,
+      "step": 1420
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.509623575372485e-06,
+      "loss": 1.1244,
+      "step": 1421
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.50754264666416e-06,
+      "loss": 1.1068,
+      "step": 1422
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.505460521046933e-06,
+      "loss": 1.0992,
+      "step": 1423
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.503377199231303e-06,
+      "loss": 1.1605,
+      "step": 1424
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.501292681928183e-06,
+      "loss": 1.0563,
+      "step": 1425
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.499206969848887e-06,
+      "loss": 1.2788,
+      "step": 1426
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.497120063705145e-06,
+      "loss": 1.0613,
+      "step": 1427
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.495031964209085e-06,
+      "loss": 0.9976,
+      "step": 1428
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.492942672073253e-06,
+      "loss": 1.0741,
+      "step": 1429
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.490852188010591e-06,
+      "loss": 0.9904,
+      "step": 1430
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.488760512734455e-06,
+      "loss": 1.0379,
+      "step": 1431
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.486667646958604e-06,
+      "loss": 1.0532,
+      "step": 1432
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.484573591397207e-06,
+      "loss": 1.0079,
+      "step": 1433
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.482478346764835e-06,
+      "loss": 1.2861,
+      "step": 1434
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.480381913776466e-06,
+      "loss": 1.2081,
+      "step": 1435
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.478284293147482e-06,
+      "loss": 1.1179,
+      "step": 1436
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.476185485593678e-06,
+      "loss": 1.0653,
+      "step": 1437
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.474085491831242e-06,
+      "loss": 1.142,
+      "step": 1438
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.471984312576776e-06,
+      "loss": 1.2089,
+      "step": 1439
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.469881948547283e-06,
+      "loss": 1.1374,
+      "step": 1440
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.467778400460171e-06,
+      "loss": 1.1276,
+      "step": 1441
+    },
+    {
+      "epoch": 0.26,
+      "learning_rate": 8.465673669033251e-06,
+      "loss": 1.1515,
+      "step": 1442
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.463567754984738e-06,
+      "loss": 1.0463,
+      "step": 1443
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.461460659033252e-06,
+      "loss": 0.9998,
+      "step": 1444
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.459352381897819e-06,
+      "loss": 1.1001,
+      "step": 1445
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.457242924297857e-06,
+      "loss": 1.1116,
+      "step": 1446
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.4551322869532e-06,
+      "loss": 1.1351,
+      "step": 1447
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.453020470584075e-06,
+      "loss": 1.0523,
+      "step": 1448
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.450907475911121e-06,
+      "loss": 1.3758,
+      "step": 1449
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.448793303655367e-06,
+      "loss": 0.9992,
+      "step": 1450
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.446677954538254e-06,
+      "loss": 1.1932,
+      "step": 1451
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.44456142928162e-06,
+      "loss": 1.0723,
+      "step": 1452
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.442443728607706e-06,
+      "loss": 1.1839,
+      "step": 1453
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.440324853239151e-06,
+      "loss": 1.1015,
+      "step": 1454
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.438204803898996e-06,
+      "loss": 1.1718,
+      "step": 1455
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.436083581310688e-06,
+      "loss": 1.0555,
+      "step": 1456
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.433961186198068e-06,
+      "loss": 1.0529,
+      "step": 1457
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.431837619285377e-06,
+      "loss": 1.1243,
+      "step": 1458
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.42971288129726e-06,
+      "loss": 1.211,
+      "step": 1459
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.427586972958761e-06,
+      "loss": 1.2906,
+      "step": 1460
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.425459894995318e-06,
+      "loss": 1.0708,
+      "step": 1461
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.423331648132778e-06,
+      "loss": 1.0734,
+      "step": 1462
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.421202233097374e-06,
+      "loss": 1.1798,
+      "step": 1463
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.419071650615749e-06,
+      "loss": 1.0888,
+      "step": 1464
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.416939901414938e-06,
+      "loss": 1.1227,
+      "step": 1465
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.41480698622238e-06,
+      "loss": 1.2342,
+      "step": 1466
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.412672905765902e-06,
+      "loss": 1.2526,
+      "step": 1467
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.41053766077374e-06,
+      "loss": 1.1679,
+      "step": 1468
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.408401251974516e-06,
+      "loss": 1.0532,
+      "step": 1469
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.406263680097261e-06,
+      "loss": 1.1764,
+      "step": 1470
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.404124945871394e-06,
+      "loss": 1.1991,
+      "step": 1471
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.401985050026736e-06,
+      "loss": 1.1469,
+      "step": 1472
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.399843993293499e-06,
+      "loss": 1.2261,
+      "step": 1473
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.397701776402296e-06,
+      "loss": 1.1319,
+      "step": 1474
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.395558400084132e-06,
+      "loss": 1.043,
+      "step": 1475
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.393413865070413e-06,
+      "loss": 1.2922,
+      "step": 1476
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.391268172092935e-06,
+      "loss": 1.2452,
+      "step": 1477
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.389121321883894e-06,
+      "loss": 1.2972,
+      "step": 1478
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.386973315175873e-06,
+      "loss": 1.1313,
+      "step": 1479
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.384824152701857e-06,
+      "loss": 1.2354,
+      "step": 1480
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.382673835195224e-06,
+      "loss": 1.1395,
+      "step": 1481
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.380522363389747e-06,
+      "loss": 1.064,
+      "step": 1482
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.378369738019588e-06,
+      "loss": 1.1031,
+      "step": 1483
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.376215959819307e-06,
+      "loss": 1.1331,
+      "step": 1484
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.374061029523859e-06,
+      "loss": 1.1079,
+      "step": 1485
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.371904947868584e-06,
+      "loss": 1.0019,
+      "step": 1486
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.369747715589223e-06,
+      "loss": 0.9721,
+      "step": 1487
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.367589333421908e-06,
+      "loss": 1.1236,
+      "step": 1488
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.36542980210316e-06,
+      "loss": 1.1874,
+      "step": 1489
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.363269122369896e-06,
+      "loss": 1.1765,
+      "step": 1490
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.361107294959423e-06,
+      "loss": 1.1426,
+      "step": 1491
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.35894432060944e-06,
+      "loss": 1.193,
+      "step": 1492
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.356780200058037e-06,
+      "loss": 1.1282,
+      "step": 1493
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.354614934043697e-06,
+      "loss": 1.1391,
+      "step": 1494
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.352448523305289e-06,
+      "loss": 1.031,
+      "step": 1495
+    },
+    {
+      "epoch": 0.27,
+      "learning_rate": 8.35028096858208e-06,
+      "loss": 1.2779,
+      "step": 1496
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.348112270613718e-06,
+      "loss": 1.1322,
+      "step": 1497
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.345942430140252e-06,
+      "loss": 1.2438,
+      "step": 1498
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.343771447902112e-06,
+      "loss": 1.2601,
+      "step": 1499
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.341599324640121e-06,
+      "loss": 1.094,
+      "step": 1500
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.339426061095493e-06,
+      "loss": 1.1998,
+      "step": 1501
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.337251658009829e-06,
+      "loss": 1.2104,
+      "step": 1502
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.335076116125119e-06,
+      "loss": 1.1855,
+      "step": 1503
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.33289943618374e-06,
+      "loss": 1.225,
+      "step": 1504
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.33072161892846e-06,
+      "loss": 1.0299,
+      "step": 1505
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.328542665102439e-06,
+      "loss": 1.1085,
+      "step": 1506
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.326362575449212e-06,
+      "loss": 1.2137,
+      "step": 1507
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.324181350712717e-06,
+      "loss": 1.1572,
+      "step": 1508
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.321998991637267e-06,
+      "loss": 1.2668,
+      "step": 1509
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.319815498967569e-06,
+      "loss": 1.2423,
+      "step": 1510
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.317630873448717e-06,
+      "loss": 1.2694,
+      "step": 1511
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.315445115826187e-06,
+      "loss": 1.0894,
+      "step": 1512
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.313258226845842e-06,
+      "loss": 1.237,
+      "step": 1513
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.311070207253938e-06,
+      "loss": 1.0368,
+      "step": 1514
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.308881057797108e-06,
+      "loss": 1.0086,
+      "step": 1515
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.306690779222375e-06,
+      "loss": 1.0937,
+      "step": 1516
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.304499372277147e-06,
+      "loss": 1.0677,
+      "step": 1517
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.302306837709217e-06,
+      "loss": 1.1304,
+      "step": 1518
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.300113176266763e-06,
+      "loss": 1.079,
+      "step": 1519
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.297918388698342e-06,
+      "loss": 1.2514,
+      "step": 1520
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.295722475752907e-06,
+      "loss": 1.207,
+      "step": 1521
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.293525438179786e-06,
+      "loss": 0.9842,
+      "step": 1522
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.29132727672869e-06,
+      "loss": 1.1219,
+      "step": 1523
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.289127992149719e-06,
+      "loss": 1.2749,
+      "step": 1524
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.286927585193356e-06,
+      "loss": 1.1057,
+      "step": 1525
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.28472605661046e-06,
+      "loss": 1.1133,
+      "step": 1526
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.28252340715228e-06,
+      "loss": 1.2734,
+      "step": 1527
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.280319637570447e-06,
+      "loss": 1.1862,
+      "step": 1528
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.278114748616966e-06,
+      "loss": 1.2852,
+      "step": 1529
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.275908741044238e-06,
+      "loss": 1.2415,
+      "step": 1530
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.273701615605033e-06,
+      "loss": 1.0325,
+      "step": 1531
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.271493373052507e-06,
+      "loss": 1.156,
+      "step": 1532
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.2692840141402e-06,
+      "loss": 1.067,
+      "step": 1533
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.26707353962203e-06,
+      "loss": 0.9896,
+      "step": 1534
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.264861950252297e-06,
+      "loss": 0.9887,
+      "step": 1535
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.262649246785677e-06,
+      "loss": 1.1377,
+      "step": 1536
+    },
+    {
+      "epoch": 0.28,
+      "eval_loss": 1.9336408376693726,
+      "eval_runtime": 0.1851,
+      "eval_samples_per_second": 16.204,
+      "eval_steps_per_second": 16.204,
+      "step": 1536
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.260435429977233e-06,
+      "loss": 0.9453,
+      "step": 1537
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.258220500582406e-06,
+      "loss": 1.0652,
+      "step": 1538
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.25600445935701e-06,
+      "loss": 1.1697,
+      "step": 1539
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.253787307057248e-06,
+      "loss": 1.0342,
+      "step": 1540
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.2515690444397e-06,
+      "loss": 1.1149,
+      "step": 1541
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.249349672261317e-06,
+      "loss": 1.1181,
+      "step": 1542
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.247129191279437e-06,
+      "loss": 1.1578,
+      "step": 1543
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.244907602251776e-06,
+      "loss": 1.0431,
+      "step": 1544
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.242684905936423e-06,
+      "loss": 1.0974,
+      "step": 1545
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.24046110309185e-06,
+      "loss": 1.2151,
+      "step": 1546
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.238236194476902e-06,
+      "loss": 1.2638,
+      "step": 1547
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.236010180850804e-06,
+      "loss": 1.3553,
+      "step": 1548
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.23378306297316e-06,
+      "loss": 1.0854,
+      "step": 1549
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.231554841603946e-06,
+      "loss": 1.0977,
+      "step": 1550
+    },
+    {
+      "epoch": 0.28,
+      "learning_rate": 8.229325517503516e-06,
+      "loss": 1.1108,
+      "step": 1551
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.227095091432608e-06,
+      "loss": 1.1219,
+      "step": 1552
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.224863564152324e-06,
+      "loss": 1.0806,
+      "step": 1553
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.222630936424146e-06,
+      "loss": 1.0953,
+      "step": 1554
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.220397209009937e-06,
+      "loss": 1.0235,
+      "step": 1555
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.218162382671928e-06,
+      "loss": 1.2872,
+      "step": 1556
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.21592645817273e-06,
+      "loss": 1.2397,
+      "step": 1557
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.213689436275325e-06,
+      "loss": 1.2167,
+      "step": 1558
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.21145131774307e-06,
+      "loss": 1.1143,
+      "step": 1559
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.209212103339704e-06,
+      "loss": 1.2879,
+      "step": 1560
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.206971793829324e-06,
+      "loss": 1.1084,
+      "step": 1561
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.204730389976418e-06,
+      "loss": 1.0507,
+      "step": 1562
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.202487892545834e-06,
+      "loss": 1.1179,
+      "step": 1563
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.200244302302801e-06,
+      "loss": 1.1618,
+      "step": 1564
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.19799962001292e-06,
+      "loss": 1.0906,
+      "step": 1565
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.195753846442162e-06,
+      "loss": 1.1582,
+      "step": 1566
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.193506982356869e-06,
+      "loss": 1.2627,
+      "step": 1567
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.191259028523762e-06,
+      "loss": 1.0267,
+      "step": 1568
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.189009985709925e-06,
+      "loss": 1.0263,
+      "step": 1569
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.186759854682824e-06,
+      "loss": 1.1096,
+      "step": 1570
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.184508636210285e-06,
+      "loss": 1.0955,
+      "step": 1571
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.182256331060514e-06,
+      "loss": 1.1933,
+      "step": 1572
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.180002940002083e-06,
+      "loss": 1.1356,
+      "step": 1573
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.177748463803937e-06,
+      "loss": 1.062,
+      "step": 1574
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.175492903235389e-06,
+      "loss": 1.2051,
+      "step": 1575
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.173236259066125e-06,
+      "loss": 1.2043,
+      "step": 1576
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.170978532066195e-06,
+      "loss": 1.1831,
+      "step": 1577
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.168719723006028e-06,
+      "loss": 1.1721,
+      "step": 1578
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.166459832656415e-06,
+      "loss": 1.1675,
+      "step": 1579
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.164198861788515e-06,
+      "loss": 1.0634,
+      "step": 1580
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.161936811173862e-06,
+      "loss": 1.2922,
+      "step": 1581
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.159673681584355e-06,
+      "loss": 1.2437,
+      "step": 1582
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.157409473792257e-06,
+      "loss": 1.2164,
+      "step": 1583
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.155144188570208e-06,
+      "loss": 1.0739,
+      "step": 1584
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.152877826691209e-06,
+      "loss": 1.0681,
+      "step": 1585
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.150610388928627e-06,
+      "loss": 1.1532,
+      "step": 1586
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.148341876056204e-06,
+      "loss": 1.26,
+      "step": 1587
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.146072288848042e-06,
+      "loss": 1.1345,
+      "step": 1588
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.14380162807861e-06,
+      "loss": 1.2035,
+      "step": 1589
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.14152989452275e-06,
+      "loss": 1.1652,
+      "step": 1590
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.139257088955658e-06,
+      "loss": 0.9818,
+      "step": 1591
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.13698321215291e-06,
+      "loss": 1.0777,
+      "step": 1592
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.134708264890439e-06,
+      "loss": 1.1355,
+      "step": 1593
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.132432247944541e-06,
+      "loss": 1.1862,
+      "step": 1594
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.130155162091885e-06,
+      "loss": 1.1868,
+      "step": 1595
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.127877008109496e-06,
+      "loss": 1.2416,
+      "step": 1596
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.125597786774773e-06,
+      "loss": 1.2422,
+      "step": 1597
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.12331749886547e-06,
+      "loss": 1.0598,
+      "step": 1598
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.121036145159713e-06,
+      "loss": 1.1164,
+      "step": 1599
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.118753726435987e-06,
+      "loss": 1.1803,
+      "step": 1600
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.116470243473138e-06,
+      "loss": 1.1047,
+      "step": 1601
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.114185697050383e-06,
+      "loss": 1.225,
+      "step": 1602
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.111900087947295e-06,
+      "loss": 1.2573,
+      "step": 1603
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.109613416943811e-06,
+      "loss": 1.1051,
+      "step": 1604
+    },
+    {
+      "epoch": 0.29,
+      "learning_rate": 8.107325684820235e-06,
+      "loss": 1.1682,
+      "step": 1605
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.105036892357226e-06,
+      "loss": 1.2551,
+      "step": 1606
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.10274704033581e-06,
+      "loss": 1.0066,
+      "step": 1607
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.100456129537373e-06,
+      "loss": 1.1313,
+      "step": 1608
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.098164160743663e-06,
+      "loss": 1.0896,
+      "step": 1609
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.095871134736784e-06,
+      "loss": 1.213,
+      "step": 1610
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.093577052299208e-06,
+      "loss": 1.1998,
+      "step": 1611
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.091281914213766e-06,
+      "loss": 1.0233,
+      "step": 1612
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.088985721263646e-06,
+      "loss": 1.0763,
+      "step": 1613
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.086688474232397e-06,
+      "loss": 1.2324,
+      "step": 1614
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.084390173903929e-06,
+      "loss": 1.2297,
+      "step": 1615
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.08209082106251e-06,
+      "loss": 1.0204,
+      "step": 1616
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.079790416492769e-06,
+      "loss": 1.2153,
+      "step": 1617
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.077488960979694e-06,
+      "loss": 1.0989,
+      "step": 1618
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.075186455308627e-06,
+      "loss": 1.1133,
+      "step": 1619
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.072882900265276e-06,
+      "loss": 1.1601,
+      "step": 1620
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.070578296635699e-06,
+      "loss": 1.0393,
+      "step": 1621
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.068272645206316e-06,
+      "loss": 1.1285,
+      "step": 1622
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.065965946763907e-06,
+      "loss": 1.136,
+      "step": 1623
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.063658202095604e-06,
+      "loss": 1.0964,
+      "step": 1624
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.0613494119889e-06,
+      "loss": 1.2038,
+      "step": 1625
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.059039577231642e-06,
+      "loss": 1.2138,
+      "step": 1626
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.056728698612038e-06,
+      "loss": 0.9866,
+      "step": 1627
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.054416776918644e-06,
+      "loss": 1.1403,
+      "step": 1628
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.052103812940381e-06,
+      "loss": 0.9844,
+      "step": 1629
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.049789807466519e-06,
+      "loss": 0.9609,
+      "step": 1630
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.047474761286688e-06,
+      "loss": 1.1613,
+      "step": 1631
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.04515867519087e-06,
+      "loss": 1.0685,
+      "step": 1632
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.042841549969403e-06,
+      "loss": 1.1674,
+      "step": 1633
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.040523386412978e-06,
+      "loss": 1.0277,
+      "step": 1634
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.038204185312643e-06,
+      "loss": 1.1598,
+      "step": 1635
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.035883947459802e-06,
+      "loss": 1.0497,
+      "step": 1636
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.033562673646204e-06,
+      "loss": 1.0638,
+      "step": 1637
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.031240364663962e-06,
+      "loss": 1.1976,
+      "step": 1638
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.028917021305535e-06,
+      "loss": 1.2848,
+      "step": 1639
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.02659264436374e-06,
+      "loss": 1.18,
+      "step": 1640
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.02426723463174e-06,
+      "loss": 1.1612,
+      "step": 1641
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.021940792903056e-06,
+      "loss": 1.167,
+      "step": 1642
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.019613319971562e-06,
+      "loss": 1.1993,
+      "step": 1643
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.01728481663148e-06,
+      "loss": 1.2103,
+      "step": 1644
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.014955283677384e-06,
+      "loss": 1.1352,
+      "step": 1645
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.012624721904203e-06,
+      "loss": 1.2168,
+      "step": 1646
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.010293132107212e-06,
+      "loss": 1.0116,
+      "step": 1647
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.00796051508204e-06,
+      "loss": 1.1468,
+      "step": 1648
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.005626871624668e-06,
+      "loss": 1.0707,
+      "step": 1649
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.003292202531425e-06,
+      "loss": 1.0834,
+      "step": 1650
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 8.000956508598986e-06,
+      "loss": 1.0919,
+      "step": 1651
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.998619790624387e-06,
+      "loss": 1.1349,
+      "step": 1652
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.996282049405e-06,
+      "loss": 1.0675,
+      "step": 1653
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.993943285738554e-06,
+      "loss": 1.3079,
+      "step": 1654
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.991603500423126e-06,
+      "loss": 0.9664,
+      "step": 1655
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.989262694257144e-06,
+      "loss": 1.2953,
+      "step": 1656
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.986920868039376e-06,
+      "loss": 1.0669,
+      "step": 1657
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.984578022568947e-06,
+      "loss": 1.107,
+      "step": 1658
+    },
+    {
+      "epoch": 0.3,
+      "learning_rate": 7.982234158645325e-06,
+      "loss": 1.1225,
+      "step": 1659
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.979889277068329e-06,
+      "loss": 1.2046,
+      "step": 1660
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.977543378638119e-06,
+      "loss": 1.074,
+      "step": 1661
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.975196464155209e-06,
+      "loss": 1.0881,
+      "step": 1662
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.972848534420456e-06,
+      "loss": 1.0889,
+      "step": 1663
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.970499590235065e-06,
+      "loss": 1.0284,
+      "step": 1664
+    },
+    {
+      "epoch": 0.31,
+      "eval_loss": 1.9469109773635864,
+      "eval_runtime": 0.1809,
+      "eval_samples_per_second": 16.58,
+      "eval_steps_per_second": 16.58,
+      "step": 1664
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.968149632400585e-06,
+      "loss": 1.1236,
+      "step": 1665
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.965798661718915e-06,
+      "loss": 1.1712,
+      "step": 1666
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.963446678992293e-06,
+      "loss": 1.3402,
+      "step": 1667
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.961093685023308e-06,
+      "loss": 1.1589,
+      "step": 1668
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.958739680614894e-06,
+      "loss": 1.1589,
+      "step": 1669
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.956384666570327e-06,
+      "loss": 1.1101,
+      "step": 1670
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.954028643693227e-06,
+      "loss": 1.1814,
+      "step": 1671
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.951671612787562e-06,
+      "loss": 1.0922,
+      "step": 1672
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.94931357465764e-06,
+      "loss": 1.0951,
+      "step": 1673
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.946954530108117e-06,
+      "loss": 1.1203,
+      "step": 1674
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.94459447994399e-06,
+      "loss": 1.1377,
+      "step": 1675
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.942233424970596e-06,
+      "loss": 1.109,
+      "step": 1676
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.939871365993622e-06,
+      "loss": 1.1949,
+      "step": 1677
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.93750830381909e-06,
+      "loss": 1.2133,
+      "step": 1678
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.93514423925337e-06,
+      "loss": 1.1003,
+      "step": 1679
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.932779173103174e-06,
+      "loss": 1.2611,
+      "step": 1680
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.930413106175548e-06,
+      "loss": 1.2104,
+      "step": 1681
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.928046039277892e-06,
+      "loss": 1.0418,
+      "step": 1682
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.925677973217937e-06,
+      "loss": 1.0399,
+      "step": 1683
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.923308908803758e-06,
+      "loss": 1.1005,
+      "step": 1684
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.920938846843774e-06,
+      "loss": 1.1159,
+      "step": 1685
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.918567788146738e-06,
+      "loss": 1.2147,
+      "step": 1686
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.91619573352175e-06,
+      "loss": 1.049,
+      "step": 1687
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.913822683778245e-06,
+      "loss": 1.0919,
+      "step": 1688
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.911448639725999e-06,
+      "loss": 1.119,
+      "step": 1689
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.909073602175127e-06,
+      "loss": 1.0458,
+      "step": 1690
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.906697571936086e-06,
+      "loss": 1.0139,
+      "step": 1691
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.904320549819667e-06,
+      "loss": 1.0598,
+      "step": 1692
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.901942536637002e-06,
+      "loss": 0.9614,
+      "step": 1693
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.899563533199561e-06,
+      "loss": 0.996,
+      "step": 1694
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.89718354031915e-06,
+      "loss": 1.143,
+      "step": 1695
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.89480255880792e-06,
+      "loss": 1.05,
+      "step": 1696
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.892420589478347e-06,
+      "loss": 1.1629,
+      "step": 1697
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.890037633143254e-06,
+      "loss": 0.9967,
+      "step": 1698
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.887653690615798e-06,
+      "loss": 1.1495,
+      "step": 1699
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.885268762709472e-06,
+      "loss": 1.1829,
+      "step": 1700
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.882882850238104e-06,
+      "loss": 1.0054,
+      "step": 1701
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.880495954015861e-06,
+      "loss": 1.1328,
+      "step": 1702
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.878108074857245e-06,
+      "loss": 1.1736,
+      "step": 1703
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.87571921357709e-06,
+      "loss": 1.1856,
+      "step": 1704
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.873329370990567e-06,
+      "loss": 1.0946,
+      "step": 1705
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.870938547913188e-06,
+      "loss": 1.1273,
+      "step": 1706
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.868546745160788e-06,
+      "loss": 1.2269,
+      "step": 1707
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.866153963549543e-06,
+      "loss": 1.0365,
+      "step": 1708
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.863760203895967e-06,
+      "loss": 1.1747,
+      "step": 1709
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.8613654670169e-06,
+      "loss": 1.2714,
+      "step": 1710
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.858969753729518e-06,
+      "loss": 1.1813,
+      "step": 1711
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.856573064851332e-06,
+      "loss": 1.0956,
+      "step": 1712
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.854175401200186e-06,
+      "loss": 1.0886,
+      "step": 1713
+    },
+    {
+      "epoch": 0.31,
+      "learning_rate": 7.851776763594252e-06,
+      "loss": 1.1998,
+      "step": 1714
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.84937715285204e-06,
+      "loss": 1.1169,
+      "step": 1715
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.846976569792391e-06,
+      "loss": 1.1539,
+      "step": 1716
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.844575015234473e-06,
+      "loss": 1.0814,
+      "step": 1717
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.84217248999779e-06,
+      "loss": 1.0847,
+      "step": 1718
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.83976899490218e-06,
+      "loss": 1.1442,
+      "step": 1719
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.837364530767804e-06,
+      "loss": 1.1054,
+      "step": 1720
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.83495909841516e-06,
+      "loss": 1.18,
+      "step": 1721
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.832552698665075e-06,
+      "loss": 1.2115,
+      "step": 1722
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.830145332338703e-06,
+      "loss": 1.1323,
+      "step": 1723
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.827737000257533e-06,
+      "loss": 1.1132,
+      "step": 1724
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.825327703243377e-06,
+      "loss": 1.0984,
+      "step": 1725
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.822917442118387e-06,
+      "loss": 1.1608,
+      "step": 1726
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.820506217705029e-06,
+      "loss": 1.128,
+      "step": 1727
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.81809403082611e-06,
+      "loss": 1.1294,
+      "step": 1728
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.815680882304762e-06,
+      "loss": 1.2537,
+      "step": 1729
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.813266772964443e-06,
+      "loss": 1.2137,
+      "step": 1730
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.810851703628939e-06,
+      "loss": 1.0684,
+      "step": 1731
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.808435675122368e-06,
+      "loss": 1.113,
+      "step": 1732
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.80601868826917e-06,
+      "loss": 1.2134,
+      "step": 1733
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.803600743894114e-06,
+      "loss": 1.1345,
+      "step": 1734
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.801181842822296e-06,
+      "loss": 1.0148,
+      "step": 1735
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.798761985879143e-06,
+      "loss": 1.1753,
+      "step": 1736
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.796341173890395e-06,
+      "loss": 1.104,
+      "step": 1737
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.793919407682132e-06,
+      "loss": 1.1982,
+      "step": 1738
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.791496688080754e-06,
+      "loss": 1.0378,
+      "step": 1739
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.789073015912985e-06,
+      "loss": 1.1184,
+      "step": 1740
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.786648392005876e-06,
+      "loss": 1.1786,
+      "step": 1741
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.784222817186802e-06,
+      "loss": 1.2271,
+      "step": 1742
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.781796292283463e-06,
+      "loss": 1.2524,
+      "step": 1743
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.779368818123882e-06,
+      "loss": 1.1528,
+      "step": 1744
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.77694039553641e-06,
+      "loss": 1.0935,
+      "step": 1745
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.774511025349716e-06,
+      "loss": 1.0467,
+      "step": 1746
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.772080708392793e-06,
+      "loss": 1.2124,
+      "step": 1747
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.769649445494963e-06,
+      "loss": 1.2037,
+      "step": 1748
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.767217237485862e-06,
+      "loss": 1.3125,
+      "step": 1749
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.764784085195458e-06,
+      "loss": 1.0581,
+      "step": 1750
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.762349989454035e-06,
+      "loss": 1.063,
+      "step": 1751
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.759914951092198e-06,
+      "loss": 1.2063,
+      "step": 1752
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.757478970940879e-06,
+      "loss": 1.1076,
+      "step": 1753
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.755042049831327e-06,
+      "loss": 1.1951,
+      "step": 1754
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.752604188595115e-06,
+      "loss": 1.2109,
+      "step": 1755
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.750165388064133e-06,
+      "loss": 1.1064,
+      "step": 1756
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.747725649070594e-06,
+      "loss": 1.0119,
+      "step": 1757
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.745284972447034e-06,
+      "loss": 1.0973,
+      "step": 1758
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.742843359026303e-06,
+      "loss": 1.0947,
+      "step": 1759
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.740400809641575e-06,
+      "loss": 1.041,
+      "step": 1760
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.737957325126341e-06,
+      "loss": 1.249,
+      "step": 1761
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.735512906314415e-06,
+      "loss": 1.1227,
+      "step": 1762
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.733067554039924e-06,
+      "loss": 1.2309,
+      "step": 1763
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.730621269137319e-06,
+      "loss": 1.0065,
+      "step": 1764
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.728174052441365e-06,
+      "loss": 1.1345,
+      "step": 1765
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.725725904787146e-06,
+      "loss": 1.0021,
+      "step": 1766
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.723276827010066e-06,
+      "loss": 1.3234,
+      "step": 1767
+    },
+    {
+      "epoch": 0.32,
+      "learning_rate": 7.720826819945845e-06,
+      "loss": 1.0713,
+      "step": 1768
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.718375884430523e-06,
+      "loss": 1.2016,
+      "step": 1769
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.715924021300445e-06,
+      "loss": 1.24,
+      "step": 1770
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.713471231392288e-06,
+      "loss": 0.94,
+      "step": 1771
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.71101751554304e-06,
+      "loss": 1.1412,
+      "step": 1772
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.708562874589999e-06,
+      "loss": 1.1077,
+      "step": 1773
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.706107309370784e-06,
+      "loss": 1.0738,
+      "step": 1774
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.703650820723332e-06,
+      "loss": 1.0588,
+      "step": 1775
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.701193409485887e-06,
+      "loss": 1.2681,
+      "step": 1776
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.698735076497015e-06,
+      "loss": 1.2175,
+      "step": 1777
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.696275822595596e-06,
+      "loss": 1.023,
+      "step": 1778
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.693815648620817e-06,
+      "loss": 1.1112,
+      "step": 1779
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.69135455541219e-06,
+      "loss": 1.0855,
+      "step": 1780
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.688892543809531e-06,
+      "loss": 1.1036,
+      "step": 1781
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.686429614652978e-06,
+      "loss": 1.1234,
+      "step": 1782
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.68396576878297e-06,
+      "loss": 1.3142,
+      "step": 1783
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.681501007040276e-06,
+      "loss": 1.173,
+      "step": 1784
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.679035330265959e-06,
+      "loss": 1.1309,
+      "step": 1785
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.676568739301405e-06,
+      "loss": 1.0691,
+      "step": 1786
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.674101234988313e-06,
+      "loss": 1.1329,
+      "step": 1787
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.67163281816869e-06,
+      "loss": 1.2962,
+      "step": 1788
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.669163489684854e-06,
+      "loss": 1.1467,
+      "step": 1789
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.666693250379436e-06,
+      "loss": 1.2346,
+      "step": 1790
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.664222101095375e-06,
+      "loss": 1.0766,
+      "step": 1791
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.661750042675925e-06,
+      "loss": 1.2157,
+      "step": 1792
+    },
+    {
+      "epoch": 0.33,
+      "eval_loss": 1.9402960538864136,
+      "eval_runtime": 0.1792,
+      "eval_samples_per_second": 16.744,
+      "eval_steps_per_second": 16.744,
+      "step": 1792
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.659277075964646e-06,
+      "loss": 1.0328,
+      "step": 1793
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.656803201805412e-06,
+      "loss": 1.1317,
+      "step": 1794
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.654328421042401e-06,
+      "loss": 1.1646,
+      "step": 1795
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.651852734520105e-06,
+      "loss": 1.1903,
+      "step": 1796
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.649376143083324e-06,
+      "loss": 1.0401,
+      "step": 1797
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.64689864757717e-06,
+      "loss": 1.186,
+      "step": 1798
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.644420248847054e-06,
+      "loss": 1.0354,
+      "step": 1799
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.641940947738703e-06,
+      "loss": 1.0232,
+      "step": 1800
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.639460745098156e-06,
+      "loss": 1.1562,
+      "step": 1801
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.636979641771745e-06,
+      "loss": 0.9914,
+      "step": 1802
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.634497638606123e-06,
+      "loss": 1.0437,
+      "step": 1803
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.632014736448245e-06,
+      "loss": 1.372,
+      "step": 1804
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.629530936145373e-06,
+      "loss": 1.1778,
+      "step": 1805
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6270462385450735e-06,
+      "loss": 1.1979,
+      "step": 1806
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.624560644495224e-06,
+      "loss": 1.2114,
+      "step": 1807
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.622074154844005e-06,
+      "loss": 1.1776,
+      "step": 1808
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6195867704399e-06,
+      "loss": 1.0357,
+      "step": 1809
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.617098492131703e-06,
+      "loss": 1.1725,
+      "step": 1810
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.614609320768509e-06,
+      "loss": 1.0659,
+      "step": 1811
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6121192571997204e-06,
+      "loss": 1.2178,
+      "step": 1812
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.609628302275043e-06,
+      "loss": 1.1912,
+      "step": 1813
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6071364568444864e-06,
+      "loss": 1.0433,
+      "step": 1814
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.604643721758364e-06,
+      "loss": 1.1541,
+      "step": 1815
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.6021500978672935e-06,
+      "loss": 1.2954,
+      "step": 1816
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.599655586022196e-06,
+      "loss": 1.2246,
+      "step": 1817
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.597160187074296e-06,
+      "loss": 1.2252,
+      "step": 1818
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.594663901875118e-06,
+      "loss": 1.0107,
+      "step": 1819
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.5921667312764915e-06,
+      "loss": 1.0549,
+      "step": 1820
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.589668676130548e-06,
+      "loss": 1.1066,
+      "step": 1821
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.58716973728972e-06,
+      "loss": 1.075,
+      "step": 1822
+    },
+    {
+      "epoch": 0.33,
+      "learning_rate": 7.584669915606741e-06,
+      "loss": 1.0704,
+      "step": 1823
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.582169211934648e-06,
+      "loss": 1.2433,
+      "step": 1824
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.579667627126778e-06,
+      "loss": 1.1875,
+      "step": 1825
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.577165162036766e-06,
+      "loss": 1.0869,
+      "step": 1826
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.574661817518551e-06,
+      "loss": 1.0997,
+      "step": 1827
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.572157594426373e-06,
+      "loss": 1.0773,
+      "step": 1828
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.5696524936147656e-06,
+      "loss": 1.1856,
+      "step": 1829
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.567146515938566e-06,
+      "loss": 1.1614,
+      "step": 1830
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.5646396622529135e-06,
+      "loss": 1.2015,
+      "step": 1831
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.562131933413242e-06,
+      "loss": 1.1556,
+      "step": 1832
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.559623330275285e-06,
+      "loss": 1.1549,
+      "step": 1833
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.557113853695075e-06,
+      "loss": 1.0223,
+      "step": 1834
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.554603504528942e-06,
+      "loss": 1.1575,
+      "step": 1835
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.552092283633514e-06,
+      "loss": 1.2685,
+      "step": 1836
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.549580191865717e-06,
+      "loss": 0.9698,
+      "step": 1837
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.547067230082774e-06,
+      "loss": 1.1616,
+      "step": 1838
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.544553399142203e-06,
+      "loss": 1.2361,
+      "step": 1839
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.542038699901819e-06,
+      "loss": 1.1759,
+      "step": 1840
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.539523133219739e-06,
+      "loss": 1.0599,
+      "step": 1841
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.537006699954368e-06,
+      "loss": 1.1038,
+      "step": 1842
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.534489400964411e-06,
+      "loss": 1.0241,
+      "step": 1843
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.531971237108867e-06,
+      "loss": 1.15,
+      "step": 1844
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.529452209247032e-06,
+      "loss": 1.1238,
+      "step": 1845
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.5269323182384936e-06,
+      "loss": 1.0301,
+      "step": 1846
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.524411564943137e-06,
+      "loss": 1.2052,
+      "step": 1847
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.521889950221142e-06,
+      "loss": 1.1022,
+      "step": 1848
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.519367474932978e-06,
+      "loss": 1.1549,
+      "step": 1849
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.516844139939412e-06,
+      "loss": 1.1856,
+      "step": 1850
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.5143199461015045e-06,
+      "loss": 1.1257,
+      "step": 1851
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.511794894280606e-06,
+      "loss": 1.0327,
+      "step": 1852
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.509268985338362e-06,
+      "loss": 1.0948,
+      "step": 1853
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.506742220136711e-06,
+      "loss": 1.1296,
+      "step": 1854
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.504214599537882e-06,
+      "loss": 1.0963,
+      "step": 1855
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.501686124404397e-06,
+      "loss": 1.0471,
+      "step": 1856
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.499156795599069e-06,
+      "loss": 1.0405,
+      "step": 1857
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.496626613985004e-06,
+      "loss": 1.0753,
+      "step": 1858
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4940955804255955e-06,
+      "loss": 1.1011,
+      "step": 1859
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.491563695784532e-06,
+      "loss": 1.362,
+      "step": 1860
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.489030960925787e-06,
+      "loss": 1.0996,
+      "step": 1861
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.48649737671363e-06,
+      "loss": 1.1622,
+      "step": 1862
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.483962944012618e-06,
+      "loss": 1.2161,
+      "step": 1863
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.481427663687597e-06,
+      "loss": 1.1142,
+      "step": 1864
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.478891536603701e-06,
+      "loss": 1.1248,
+      "step": 1865
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.4763545636263535e-06,
+      "loss": 1.393,
+      "step": 1866
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.47381674562127e-06,
+      "loss": 1.215,
+      "step": 1867
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.471278083454454e-06,
+      "loss": 1.2577,
+      "step": 1868
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.46873857799219e-06,
+      "loss": 0.9778,
+      "step": 1869
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.466198230101057e-06,
+      "loss": 1.194,
+      "step": 1870
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.463657040647921e-06,
+      "loss": 1.0647,
+      "step": 1871
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.461115010499933e-06,
+      "loss": 1.0976,
+      "step": 1872
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.458572140524533e-06,
+      "loss": 1.2443,
+      "step": 1873
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.456028431589444e-06,
+      "loss": 1.1225,
+      "step": 1874
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.45348388456268e-06,
+      "loss": 1.2456,
+      "step": 1875
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.450938500312535e-06,
+      "loss": 1.074,
+      "step": 1876
+    },
+    {
+      "epoch": 0.34,
+      "learning_rate": 7.448392279707597e-06,
+      "loss": 1.1158,
+      "step": 1877
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.445845223616733e-06,
+      "loss": 1.1181,
+      "step": 1878
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.443297332909096e-06,
+      "loss": 1.2966,
+      "step": 1879
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.440748608454125e-06,
+      "loss": 1.0513,
+      "step": 1880
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.438199051121544e-06,
+      "loss": 1.0854,
+      "step": 1881
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.435648661781358e-06,
+      "loss": 1.0633,
+      "step": 1882
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.433097441303861e-06,
+      "loss": 1.1409,
+      "step": 1883
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.430545390559626e-06,
+      "loss": 1.1725,
+      "step": 1884
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.427992510419512e-06,
+      "loss": 0.9471,
+      "step": 1885
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4254388017546595e-06,
+      "loss": 1.1388,
+      "step": 1886
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.422884265436494e-06,
+      "loss": 1.0695,
+      "step": 1887
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.420328902336721e-06,
+      "loss": 1.0969,
+      "step": 1888
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.417772713327328e-06,
+      "loss": 1.1778,
+      "step": 1889
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.4152156992805865e-06,
+      "loss": 1.1322,
+      "step": 1890
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.412657861069049e-06,
+      "loss": 1.0968,
+      "step": 1891
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.410099199565547e-06,
+      "loss": 1.0983,
+      "step": 1892
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.407539715643195e-06,
+      "loss": 1.2472,
+      "step": 1893
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.404979410175389e-06,
+      "loss": 1.1829,
+      "step": 1894
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.402418284035802e-06,
+      "loss": 0.9524,
+      "step": 1895
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.39985633809839e-06,
+      "loss": 1.2215,
+      "step": 1896
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.397293573237389e-06,
+      "loss": 1.0687,
+      "step": 1897
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.394729990327312e-06,
+      "loss": 1.0983,
+      "step": 1898
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.39216559024295e-06,
+      "loss": 1.1995,
+      "step": 1899
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.38960037385938e-06,
+      "loss": 1.1756,
+      "step": 1900
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.387034342051951e-06,
+      "loss": 1.0551,
+      "step": 1901
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.3844674956962904e-06,
+      "loss": 1.1983,
+      "step": 1902
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.381899835668307e-06,
+      "loss": 1.0254,
+      "step": 1903
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.379331362844185e-06,
+      "loss": 1.2319,
+      "step": 1904
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.376762078100385e-06,
+      "loss": 1.0528,
+      "step": 1905
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.37419198231365e-06,
+      "loss": 1.1909,
+      "step": 1906
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.371621076360993e-06,
+      "loss": 1.1574,
+      "step": 1907
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.369049361119706e-06,
+      "loss": 1.2814,
+      "step": 1908
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.3664768374673575e-06,
+      "loss": 1.0998,
+      "step": 1909
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.363903506281793e-06,
+      "loss": 1.1734,
+      "step": 1910
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.361329368441132e-06,
+      "loss": 1.1604,
+      "step": 1911
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.3587544248237705e-06,
+      "loss": 1.1591,
+      "step": 1912
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.356178676308377e-06,
+      "loss": 1.0755,
+      "step": 1913
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.353602123773897e-06,
+      "loss": 1.1272,
+      "step": 1914
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.351024768099551e-06,
+      "loss": 1.2043,
+      "step": 1915
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.348446610164828e-06,
+      "loss": 1.0437,
+      "step": 1916
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.345867650849501e-06,
+      "loss": 1.3038,
+      "step": 1917
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.343287891033605e-06,
+      "loss": 1.0386,
+      "step": 1918
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.340707331597457e-06,
+      "loss": 1.1548,
+      "step": 1919
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.338125973421641e-06,
+      "loss": 1.1553,
+      "step": 1920
+    },
+    {
+      "epoch": 0.35,
+      "eval_loss": 1.9376972913742065,
+      "eval_runtime": 0.1775,
+      "eval_samples_per_second": 16.905,
+      "eval_steps_per_second": 16.905,
+      "step": 1920
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.335543817387019e-06,
+      "loss": 1.1716,
+      "step": 1921
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.33296086437472e-06,
+      "loss": 1.0718,
+      "step": 1922
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.330377115266147e-06,
+      "loss": 1.3024,
+      "step": 1923
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.3277925709429775e-06,
+      "loss": 1.0767,
+      "step": 1924
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.325207232287155e-06,
+      "loss": 1.079,
+      "step": 1925
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.322621100180898e-06,
+      "loss": 1.0768,
+      "step": 1926
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.320034175506695e-06,
+      "loss": 1.112,
+      "step": 1927
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.317446459147304e-06,
+      "loss": 1.0517,
+      "step": 1928
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.314857951985752e-06,
+      "loss": 1.2402,
+      "step": 1929
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.312268654905341e-06,
+      "loss": 1.2059,
+      "step": 1930
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.309678568789636e-06,
+      "loss": 1.1856,
+      "step": 1931
+    },
+    {
+      "epoch": 0.35,
+      "learning_rate": 7.3070876945224765e-06,
+      "loss": 1.0378,
+      "step": 1932
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.3044960329879675e-06,
+      "loss": 1.2691,
+      "step": 1933
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.301903585070484e-06,
+      "loss": 1.3588,
+      "step": 1934
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.299310351654671e-06,
+      "loss": 1.3278,
+      "step": 1935
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.296716333625436e-06,
+      "loss": 1.0296,
+      "step": 1936
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.294121531867963e-06,
+      "loss": 1.2506,
+      "step": 1937
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.291525947267694e-06,
+      "loss": 1.0166,
+      "step": 1938
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.288929580710345e-06,
+      "loss": 1.1434,
+      "step": 1939
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.286332433081897e-06,
+      "loss": 1.127,
+      "step": 1940
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.2837345052685945e-06,
+      "loss": 1.1184,
+      "step": 1941
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.281135798156953e-06,
+      "loss": 1.0523,
+      "step": 1942
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.27853631263375e-06,
+      "loss": 1.2102,
+      "step": 1943
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.275936049586032e-06,
+      "loss": 1.178,
+      "step": 1944
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.273335009901108e-06,
+      "loss": 1.2264,
+      "step": 1945
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.270733194466554e-06,
+      "loss": 1.0678,
+      "step": 1946
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.268130604170208e-06,
+      "loss": 1.1062,
+      "step": 1947
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.265527239900176e-06,
+      "loss": 1.0253,
+      "step": 1948
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.262923102544826e-06,
+      "loss": 1.1847,
+      "step": 1949
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.26031819299279e-06,
+      "loss": 1.0787,
+      "step": 1950
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.257712512132963e-06,
+      "loss": 1.1701,
+      "step": 1951
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.255106060854504e-06,
+      "loss": 1.1257,
+      "step": 1952
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.252498840046836e-06,
+      "loss": 1.0084,
+      "step": 1953
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.249890850599642e-06,
+      "loss": 1.1784,
+      "step": 1954
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.247282093402869e-06,
+      "loss": 1.0063,
+      "step": 1955
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.244672569346727e-06,
+      "loss": 1.0613,
+      "step": 1956
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.242062279321684e-06,
+      "loss": 1.0633,
+      "step": 1957
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.2394512242184735e-06,
+      "loss": 1.1269,
+      "step": 1958
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.2368394049280865e-06,
+      "loss": 1.213,
+      "step": 1959
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.234226822341779e-06,
+      "loss": 1.1988,
+      "step": 1960
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.231613477351063e-06,
+      "loss": 1.221,
+      "step": 1961
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.228999370847712e-06,
+      "loss": 1.0237,
+      "step": 1962
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.226384503723764e-06,
+      "loss": 1.1568,
+      "step": 1963
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.223768876871507e-06,
+      "loss": 1.0771,
+      "step": 1964
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.221152491183497e-06,
+      "loss": 1.1978,
+      "step": 1965
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.218535347552547e-06,
+      "loss": 1.0007,
+      "step": 1966
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.2159174468717254e-06,
+      "loss": 1.1335,
+      "step": 1967
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.21329879003436e-06,
+      "loss": 1.1237,
+      "step": 1968
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.210679377934039e-06,
+      "loss": 1.1094,
+      "step": 1969
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.208059211464609e-06,
+      "loss": 1.1568,
+      "step": 1970
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.205438291520167e-06,
+      "loss": 1.1065,
+      "step": 1971
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.2028166189950755e-06,
+      "loss": 1.1924,
+      "step": 1972
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.20019419478395e-06,
+      "loss": 1.0567,
+      "step": 1973
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.197571019781662e-06,
+      "loss": 1.3201,
+      "step": 1974
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.19494709488334e-06,
+      "loss": 1.0471,
+      "step": 1975
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.19232242098437e-06,
+      "loss": 1.296,
+      "step": 1976
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.189696998980391e-06,
+      "loss": 1.2111,
+      "step": 1977
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.187070829767297e-06,
+      "loss": 1.0393,
+      "step": 1978
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.184443914241239e-06,
+      "loss": 1.127,
+      "step": 1979
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.181816253298623e-06,
+      "loss": 1.1475,
+      "step": 1980
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.179187847836109e-06,
+      "loss": 1.1506,
+      "step": 1981
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1765586987506075e-06,
+      "loss": 1.1549,
+      "step": 1982
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.173928806939287e-06,
+      "loss": 0.9734,
+      "step": 1983
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1712981732995705e-06,
+      "loss": 1.2443,
+      "step": 1984
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.168666798729127e-06,
+      "loss": 1.0181,
+      "step": 1985
+    },
+    {
+      "epoch": 0.36,
+      "learning_rate": 7.1660346841258866e-06,
+      "loss": 1.1042,
+      "step": 1986
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1634018303880275e-06,
+      "loss": 1.1077,
+      "step": 1987
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.16076823841398e-06,
+      "loss": 1.3274,
+      "step": 1988
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.158133909102427e-06,
+      "loss": 1.1974,
+      "step": 1989
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.155498843352307e-06,
+      "loss": 1.1738,
+      "step": 1990
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.152863042062802e-06,
+      "loss": 1.0794,
+      "step": 1991
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.150226506133351e-06,
+      "loss": 1.1083,
+      "step": 1992
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.147589236463642e-06,
+      "loss": 1.1271,
+      "step": 1993
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.144951233953611e-06,
+      "loss": 1.0724,
+      "step": 1994
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.142312499503448e-06,
+      "loss": 1.0027,
+      "step": 1995
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.139673034013593e-06,
+      "loss": 0.9527,
+      "step": 1996
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1370328383847295e-06,
+      "loss": 1.1101,
+      "step": 1997
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.134391913517797e-06,
+      "loss": 1.0559,
+      "step": 1998
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.131750260313979e-06,
+      "loss": 1.1163,
+      "step": 1999
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.129107879674711e-06,
+      "loss": 1.1099,
+      "step": 2000
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.1264647725016756e-06,
+      "loss": 1.0887,
+      "step": 2001
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.123820939696801e-06,
+      "loss": 1.0488,
+      "step": 2002
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.121176382162266e-06,
+      "loss": 1.1675,
+      "step": 2003
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.118531100800496e-06,
+      "loss": 1.1485,
+      "step": 2004
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.115885096514163e-06,
+      "loss": 1.108,
+      "step": 2005
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.113238370206186e-06,
+      "loss": 1.1985,
+      "step": 2006
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.110590922779728e-06,
+      "loss": 1.1909,
+      "step": 2007
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.107942755138203e-06,
+      "loss": 1.1544,
+      "step": 2008
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.105293868185268e-06,
+      "loss": 1.23,
+      "step": 2009
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.102644262824822e-06,
+      "loss": 1.3049,
+      "step": 2010
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.099993939961017e-06,
+      "loss": 1.3543,
+      "step": 2011
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.097342900498243e-06,
+      "loss": 1.1809,
+      "step": 2012
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.094691145341137e-06,
+      "loss": 1.1485,
+      "step": 2013
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.092038675394582e-06,
+      "loss": 1.0688,
+      "step": 2014
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.0893854915637e-06,
+      "loss": 1.0368,
+      "step": 2015
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.086731594753861e-06,
+      "loss": 1.1612,
+      "step": 2016
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.084076985870678e-06,
+      "loss": 1.1127,
+      "step": 2017
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.081421665820006e-06,
+      "loss": 1.0161,
+      "step": 2018
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.078765635507943e-06,
+      "loss": 1.1394,
+      "step": 2019
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.0761088958408255e-06,
+      "loss": 1.1439,
+      "step": 2020
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.073451447725239e-06,
+      "loss": 1.1821,
+      "step": 2021
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.0707932920680065e-06,
+      "loss": 1.1126,
+      "step": 2022
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.068134429776194e-06,
+      "loss": 1.1558,
+      "step": 2023
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.0654748617571066e-06,
+      "loss": 1.194,
+      "step": 2024
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.062814588918293e-06,
+      "loss": 1.0223,
+      "step": 2025
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.060153612167541e-06,
+      "loss": 1.0728,
+      "step": 2026
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.057491932412876e-06,
+      "loss": 1.1017,
+      "step": 2027
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.054829550562567e-06,
+      "loss": 1.2297,
+      "step": 2028
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.052166467525124e-06,
+      "loss": 1.1474,
+      "step": 2029
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.049502684209288e-06,
+      "loss": 1.1885,
+      "step": 2030
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.046838201524051e-06,
+      "loss": 1.1459,
+      "step": 2031
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.044173020378634e-06,
+      "loss": 1.2933,
+      "step": 2032
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.041507141682501e-06,
+      "loss": 1.0779,
+      "step": 2033
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.038840566345351e-06,
+      "loss": 1.1499,
+      "step": 2034
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.036173295277124e-06,
+      "loss": 1.1358,
+      "step": 2035
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.033505329387995e-06,
+      "loss": 1.041,
+      "step": 2036
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.030836669588377e-06,
+      "loss": 1.1773,
+      "step": 2037
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.028167316788918e-06,
+      "loss": 1.1019,
+      "step": 2038
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.02549727190051e-06,
+      "loss": 1.1307,
+      "step": 2039
+    },
+    {
+      "epoch": 0.37,
+      "learning_rate": 7.022826535834268e-06,
+      "loss": 1.1365,
+      "step": 2040
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.020155109501554e-06,
+      "loss": 1.1165,
+      "step": 2041
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.017482993813962e-06,
+      "loss": 1.1032,
+      "step": 2042
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.01481018968332e-06,
+      "loss": 1.1059,
+      "step": 2043
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.012136698021692e-06,
+      "loss": 1.0182,
+      "step": 2044
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.009462519741377e-06,
+      "loss": 1.1043,
+      "step": 2045
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.006787655754907e-06,
+      "loss": 1.1896,
+      "step": 2046
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.004112106975049e-06,
+      "loss": 1.1627,
+      "step": 2047
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 7.001435874314805e-06,
+      "loss": 1.0985,
+      "step": 2048
+    },
+    {
+      "epoch": 0.38,
+      "eval_loss": 1.9464306831359863,
+      "eval_runtime": 0.1764,
+      "eval_samples_per_second": 17.01,
+      "eval_steps_per_second": 17.01,
+      "step": 2048
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.998758958687407e-06,
+      "loss": 1.1312,
+      "step": 2049
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9960813610063225e-06,
+      "loss": 1.1652,
+      "step": 2050
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.993403082185251e-06,
+      "loss": 1.1152,
+      "step": 2051
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.990724123138126e-06,
+      "loss": 1.0349,
+      "step": 2052
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.988044484779109e-06,
+      "loss": 1.1265,
+      "step": 2053
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.985364168022598e-06,
+      "loss": 0.9773,
+      "step": 2054
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.98268317378322e-06,
+      "loss": 1.1337,
+      "step": 2055
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.980001502975835e-06,
+      "loss": 1.1865,
+      "step": 2056
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.977319156515529e-06,
+      "loss": 1.0417,
+      "step": 2057
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.974636135317626e-06,
+      "loss": 1.1581,
+      "step": 2058
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.971952440297674e-06,
+      "loss": 1.0717,
+      "step": 2059
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.969268072371456e-06,
+      "loss": 1.2609,
+      "step": 2060
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.966583032454977e-06,
+      "loss": 1.0889,
+      "step": 2061
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.963897321464481e-06,
+      "loss": 1.1241,
+      "step": 2062
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.961210940316434e-06,
+      "loss": 1.1489,
+      "step": 2063
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.958523889927533e-06,
+      "loss": 1.1209,
+      "step": 2064
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.955836171214705e-06,
+      "loss": 1.1962,
+      "step": 2065
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.953147785095101e-06,
+      "loss": 0.9101,
+      "step": 2066
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9504587324861025e-06,
+      "loss": 1.1761,
+      "step": 2067
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.94776901430532e-06,
+      "loss": 1.1612,
+      "step": 2068
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.945078631470588e-06,
+      "loss": 1.1347,
+      "step": 2069
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.942387584899967e-06,
+      "loss": 1.1374,
+      "step": 2070
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.939695875511748e-06,
+      "loss": 0.9618,
+      "step": 2071
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.937003504224447e-06,
+      "loss": 1.1936,
+      "step": 2072
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.934310471956802e-06,
+      "loss": 1.0562,
+      "step": 2073
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9316167796277825e-06,
+      "loss": 1.1676,
+      "step": 2074
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9289224281565794e-06,
+      "loss": 1.0017,
+      "step": 2075
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.926227418462608e-06,
+      "loss": 1.1234,
+      "step": 2076
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.92353175146551e-06,
+      "loss": 1.0458,
+      "step": 2077
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.920835428085152e-06,
+      "loss": 1.0539,
+      "step": 2078
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.918138449241624e-06,
+      "loss": 1.1966,
+      "step": 2079
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9154408158552376e-06,
+      "loss": 1.1953,
+      "step": 2080
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.91274252884653e-06,
+      "loss": 1.0496,
+      "step": 2081
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.9100435891362626e-06,
+      "loss": 1.0963,
+      "step": 2082
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.907343997645415e-06,
+      "loss": 1.1084,
+      "step": 2083
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.904643755295193e-06,
+      "loss": 1.1466,
+      "step": 2084
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.901942863007025e-06,
+      "loss": 1.1127,
+      "step": 2085
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.899241321702558e-06,
+      "loss": 0.9882,
+      "step": 2086
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.896539132303662e-06,
+      "loss": 1.0418,
+      "step": 2087
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.8938362957324286e-06,
+      "loss": 1.1304,
+      "step": 2088
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.891132812911172e-06,
+      "loss": 1.0269,
+      "step": 2089
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.888428684762421e-06,
+      "loss": 1.1252,
+      "step": 2090
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.885723912208933e-06,
+      "loss": 1.1417,
+      "step": 2091
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.883018496173677e-06,
+      "loss": 1.0366,
+      "step": 2092
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.880312437579846e-06,
+      "loss": 1.1602,
+      "step": 2093
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.877605737350852e-06,
+      "loss": 1.1588,
+      "step": 2094
+    },
+    {
+      "epoch": 0.38,
+      "learning_rate": 6.874898396410328e-06,
+      "loss": 1.2235,
+      "step": 2095
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.872190415682119e-06,
+      "loss": 1.3263,
+      "step": 2096
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.869481796090295e-06,
+      "loss": 1.0569,
+      "step": 2097
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.866772538559141e-06,
+      "loss": 1.074,
+      "step": 2098
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.864062644013161e-06,
+      "loss": 1.1603,
+      "step": 2099
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.861352113377075e-06,
+      "loss": 1.2573,
+      "step": 2100
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.85864094757582e-06,
+      "loss": 1.161,
+      "step": 2101
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.855929147534551e-06,
+      "loss": 1.2911,
+      "step": 2102
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8532167141786386e-06,
+      "loss": 0.986,
+      "step": 2103
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8505036484336694e-06,
+      "loss": 1.2026,
+      "step": 2104
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.847789951225448e-06,
+      "loss": 0.9248,
+      "step": 2105
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.845075623479991e-06,
+      "loss": 1.1674,
+      "step": 2106
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.842360666123531e-06,
+      "loss": 1.1949,
+      "step": 2107
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8396450800825185e-06,
+      "loss": 1.1739,
+      "step": 2108
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.836928866283614e-06,
+      "loss": 1.0553,
+      "step": 2109
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.834212025653696e-06,
+      "loss": 1.2104,
+      "step": 2110
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8314945591198535e-06,
+      "loss": 1.0911,
+      "step": 2111
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8287764676093945e-06,
+      "loss": 1.1787,
+      "step": 2112
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8260577520498324e-06,
+      "loss": 1.1654,
+      "step": 2113
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8233384133689005e-06,
+      "loss": 1.1751,
+      "step": 2114
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.820618452494543e-06,
+      "loss": 1.2359,
+      "step": 2115
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8178978703549116e-06,
+      "loss": 1.2252,
+      "step": 2116
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.815176667878377e-06,
+      "loss": 1.1041,
+      "step": 2117
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.812454845993519e-06,
+      "loss": 1.1288,
+      "step": 2118
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.809732405629126e-06,
+      "loss": 1.1843,
+      "step": 2119
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.807009347714201e-06,
+      "loss": 1.1592,
+      "step": 2120
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.804285673177956e-06,
+      "loss": 1.0916,
+      "step": 2121
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.8015613829498154e-06,
+      "loss": 1.212,
+      "step": 2122
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.79883647795941e-06,
+      "loss": 1.0951,
+      "step": 2123
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.796110959136581e-06,
+      "loss": 1.1709,
+      "step": 2124
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.793384827411387e-06,
+      "loss": 1.0076,
+      "step": 2125
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.790658083714083e-06,
+      "loss": 1.0669,
+      "step": 2126
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.787930728975142e-06,
+      "loss": 1.0209,
+      "step": 2127
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.7852027641252424e-06,
+      "loss": 1.1745,
+      "step": 2128
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.78247419009527e-06,
+      "loss": 1.1281,
+      "step": 2129
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.77974500781632e-06,
+      "loss": 1.1159,
+      "step": 2130
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.777015218219697e-06,
+      "loss": 1.1615,
+      "step": 2131
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.774284822236907e-06,
+      "loss": 1.0939,
+      "step": 2132
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.771553820799669e-06,
+      "loss": 1.0918,
+      "step": 2133
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.768822214839905e-06,
+      "loss": 1.1745,
+      "step": 2134
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.766090005289745e-06,
+      "loss": 1.0635,
+      "step": 2135
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.763357193081523e-06,
+      "loss": 1.2103,
+      "step": 2136
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.760623779147781e-06,
+      "loss": 1.1519,
+      "step": 2137
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.757889764421266e-06,
+      "loss": 1.124,
+      "step": 2138
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.755155149834928e-06,
+      "loss": 1.2823,
+      "step": 2139
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.752419936321924e-06,
+      "loss": 1.1989,
+      "step": 2140
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.7496841248156135e-06,
+      "loss": 1.186,
+      "step": 2141
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.746947716249561e-06,
+      "loss": 1.0864,
+      "step": 2142
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.744210711557535e-06,
+      "loss": 1.0543,
+      "step": 2143
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.7414731116735065e-06,
+      "loss": 1.0836,
+      "step": 2144
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.738734917531653e-06,
+      "loss": 1.183,
+      "step": 2145
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.735996130066349e-06,
+      "loss": 1.2102,
+      "step": 2146
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.7332567502121746e-06,
+      "loss": 1.2179,
+      "step": 2147
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.730516778903915e-06,
+      "loss": 1.1192,
+      "step": 2148
+    },
+    {
+      "epoch": 0.39,
+      "learning_rate": 6.7277762170765495e-06,
+      "loss": 1.1393,
+      "step": 2149
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.725035065665266e-06,
+      "loss": 1.0991,
+      "step": 2150
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.722293325605453e-06,
+      "loss": 1.0023,
+      "step": 2151
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.719550997832694e-06,
+      "loss": 1.044,
+      "step": 2152
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.71680808328278e-06,
+      "loss": 1.1963,
+      "step": 2153
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.714064582891697e-06,
+      "loss": 1.3668,
+      "step": 2154
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.711320497595637e-06,
+      "loss": 1.1205,
+      "step": 2155
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7085758283309845e-06,
+      "loss": 1.0011,
+      "step": 2156
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.705830576034327e-06,
+      "loss": 1.0124,
+      "step": 2157
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.7030847416424515e-06,
+      "loss": 1.2322,
+      "step": 2158
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.700338326092341e-06,
+      "loss": 1.1929,
+      "step": 2159
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.697591330321181e-06,
+      "loss": 1.0888,
+      "step": 2160
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6948437552663515e-06,
+      "loss": 1.0849,
+      "step": 2161
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.692095601865429e-06,
+      "loss": 0.9986,
+      "step": 2162
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.689346871056193e-06,
+      "loss": 1.0519,
+      "step": 2163
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6865975637766125e-06,
+      "loss": 1.1967,
+      "step": 2164
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.683847680964861e-06,
+      "loss": 1.1206,
+      "step": 2165
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.681097223559303e-06,
+      "loss": 1.1124,
+      "step": 2166
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.678346192498501e-06,
+      "loss": 1.0445,
+      "step": 2167
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6755945887212125e-06,
+      "loss": 1.0418,
+      "step": 2168
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.672842413166391e-06,
+      "loss": 0.9713,
+      "step": 2169
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.670089666773184e-06,
+      "loss": 1.2333,
+      "step": 2170
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.667336350480937e-06,
+      "loss": 1.1086,
+      "step": 2171
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.664582465229187e-06,
+      "loss": 1.1031,
+      "step": 2172
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.661828011957665e-06,
+      "loss": 0.9607,
+      "step": 2173
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.659072991606298e-06,
+      "loss": 1.1398,
+      "step": 2174
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6563174051152046e-06,
+      "loss": 1.155,
+      "step": 2175
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6535612534246965e-06,
+      "loss": 1.0056,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4,
+      "eval_loss": 1.9448951482772827,
+      "eval_runtime": 0.1704,
+      "eval_samples_per_second": 17.601,
+      "eval_steps_per_second": 17.601,
+      "step": 2176
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.650804537475279e-06,
+      "loss": 1.1345,
+      "step": 2177
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6480472582076525e-06,
+      "loss": 1.0121,
+      "step": 2178
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.645289416562705e-06,
+      "loss": 1.1059,
+      "step": 2179
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.642531013481518e-06,
+      "loss": 1.1649,
+      "step": 2180
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.639772049905367e-06,
+      "loss": 1.18,
+      "step": 2181
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6370125267757145e-06,
+      "loss": 1.0481,
+      "step": 2182
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.634252445034218e-06,
+      "loss": 1.1262,
+      "step": 2183
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.631491805622725e-06,
+      "loss": 1.1432,
+      "step": 2184
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.628730609483268e-06,
+      "loss": 1.1777,
+      "step": 2185
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.625968857558078e-06,
+      "loss": 1.1166,
+      "step": 2186
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.623206550789568e-06,
+      "loss": 0.9841,
+      "step": 2187
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.620443690120347e-06,
+      "loss": 1.0441,
+      "step": 2188
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.617680276493207e-06,
+      "loss": 1.1619,
+      "step": 2189
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.614916310851133e-06,
+      "loss": 1.1025,
+      "step": 2190
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.612151794137296e-06,
+      "loss": 1.0963,
+      "step": 2191
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.609386727295055e-06,
+      "loss": 1.1398,
+      "step": 2192
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.606621111267958e-06,
+      "loss": 1.0398,
+      "step": 2193
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.603854946999741e-06,
+      "loss": 1.2648,
+      "step": 2194
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.6010882354343254e-06,
+      "loss": 1.0997,
+      "step": 2195
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.5983209775158195e-06,
+      "loss": 1.0331,
+      "step": 2196
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.595553174188518e-06,
+      "loss": 0.9272,
+      "step": 2197
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.592784826396905e-06,
+      "loss": 1.1015,
+      "step": 2198
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.590015935085644e-06,
+      "loss": 1.0255,
+      "step": 2199
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.587246501199589e-06,
+      "loss": 1.0538,
+      "step": 2200
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.58447652568378e-06,
+      "loss": 1.0879,
+      "step": 2201
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.581706009483435e-06,
+      "loss": 1.2234,
+      "step": 2202
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.578934953543963e-06,
+      "loss": 1.0707,
+      "step": 2203
+    },
+    {
+      "epoch": 0.4,
+      "learning_rate": 6.576163358810959e-06,
+      "loss": 1.2227,
+      "step": 2204
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.573391226230191e-06,
+      "loss": 1.0395,
+      "step": 2205
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.570618556747622e-06,
+      "loss": 1.0692,
+      "step": 2206
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.567845351309396e-06,
+      "loss": 1.2314,
+      "step": 2207
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.565071610861832e-06,
+      "loss": 1.1499,
+      "step": 2208
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.562297336351439e-06,
+      "loss": 1.0889,
+      "step": 2209
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.559522528724911e-06,
+      "loss": 1.1778,
+      "step": 2210
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.556747188929115e-06,
+      "loss": 1.0988,
+      "step": 2211
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5539713179111045e-06,
+      "loss": 1.099,
+      "step": 2212
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5511949166181145e-06,
+      "loss": 0.9827,
+      "step": 2213
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.54841798599756e-06,
+      "loss": 1.0375,
+      "step": 2214
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.545640526997038e-06,
+      "loss": 1.1236,
+      "step": 2215
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.542862540564323e-06,
+      "loss": 1.2138,
+      "step": 2216
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.540084027647373e-06,
+      "loss": 1.2293,
+      "step": 2217
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.537304989194321e-06,
+      "loss": 1.2182,
+      "step": 2218
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5345254261534845e-06,
+      "loss": 1.0633,
+      "step": 2219
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.531745339473357e-06,
+      "loss": 0.974,
+      "step": 2220
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.528964730102612e-06,
+      "loss": 1.195,
+      "step": 2221
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5261835989901005e-06,
+      "loss": 1.2687,
+      "step": 2222
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.523401947084851e-06,
+      "loss": 1.0781,
+      "step": 2223
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.5206197753360715e-06,
+      "loss": 1.0596,
+      "step": 2224
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.517837084693144e-06,
+      "loss": 1.0957,
+      "step": 2225
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.515053876105632e-06,
+      "loss": 1.1663,
+      "step": 2226
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.512270150523276e-06,
+      "loss": 1.0686,
+      "step": 2227
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.509485908895985e-06,
+      "loss": 1.1606,
+      "step": 2228
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.506701152173853e-06,
+      "loss": 1.173,
+      "step": 2229
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.503915881307147e-06,
+      "loss": 1.0764,
+      "step": 2230
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.501130097246308e-06,
+      "loss": 0.9948,
+      "step": 2231
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.498343800941953e-06,
+      "loss": 1.1171,
+      "step": 2232
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.495556993344872e-06,
+      "loss": 1.0069,
+      "step": 2233
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.492769675406036e-06,
+      "loss": 1.1917,
+      "step": 2234
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.489981848076581e-06,
+      "loss": 1.2538,
+      "step": 2235
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.4871935123078225e-06,
+      "loss": 1.1624,
+      "step": 2236
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.48440466905125e-06,
+      "loss": 1.2244,
+      "step": 2237
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.481615319258522e-06,
+      "loss": 1.1231,
+      "step": 2238
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.478825463881473e-06,
+      "loss": 1.1103,
+      "step": 2239
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.4760351038721105e-06,
+      "loss": 0.9812,
+      "step": 2240
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.4732442401826115e-06,
+      "loss": 1.2153,
+      "step": 2241
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.470452873765326e-06,
+      "loss": 1.0092,
+      "step": 2242
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.467661005572779e-06,
+      "loss": 1.1941,
+      "step": 2243
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.464868636557661e-06,
+      "loss": 1.2007,
+      "step": 2244
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.462075767672835e-06,
+      "loss": 1.0821,
+      "step": 2245
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.459282399871338e-06,
+      "loss": 1.2185,
+      "step": 2246
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.456488534106376e-06,
+      "loss": 1.0818,
+      "step": 2247
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.45369417133132e-06,
+      "loss": 1.198,
+      "step": 2248
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.450899312499718e-06,
+      "loss": 1.1112,
+      "step": 2249
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.448103958565281e-06,
+      "loss": 0.9292,
+      "step": 2250
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.445308110481894e-06,
+      "loss": 1.1262,
+      "step": 2251
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.442511769203605e-06,
+      "loss": 1.1628,
+      "step": 2252
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.439714935684636e-06,
+      "loss": 1.1502,
+      "step": 2253
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.436917610879376e-06,
+      "loss": 1.2021,
+      "step": 2254
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.434119795742378e-06,
+      "loss": 1.2082,
+      "step": 2255
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.431321491228362e-06,
+      "loss": 1.2574,
+      "step": 2256
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.4285226982922225e-06,
+      "loss": 1.1605,
+      "step": 2257
+    },
+    {
+      "epoch": 0.41,
+      "learning_rate": 6.425723417889013e-06,
+      "loss": 1.1305,
+      "step": 2258
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.422923650973956e-06,
+      "loss": 1.0898,
+      "step": 2259
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.420123398502441e-06,
+      "loss": 1.0566,
+      "step": 2260
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.41732266143002e-06,
+      "loss": 1.32,
+      "step": 2261
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.414521440712415e-06,
+      "loss": 1.0063,
+      "step": 2262
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.4117197373055086e-06,
+      "loss": 1.1432,
+      "step": 2263
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.408917552165353e-06,
+      "loss": 1.1276,
+      "step": 2264
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.406114886248161e-06,
+      "loss": 1.193,
+      "step": 2265
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.403311740510308e-06,
+      "loss": 1.2305,
+      "step": 2266
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.400508115908337e-06,
+      "loss": 1.1831,
+      "step": 2267
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.397704013398952e-06,
+      "loss": 1.1133,
+      "step": 2268
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.394899433939024e-06,
+      "loss": 1.0276,
+      "step": 2269
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.39209437848558e-06,
+      "loss": 1.2936,
+      "step": 2270
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.389288847995815e-06,
+      "loss": 0.9829,
+      "step": 2271
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.3864828434270845e-06,
+      "loss": 1.1005,
+      "step": 2272
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.383676365736906e-06,
+      "loss": 0.9373,
+      "step": 2273
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.3808694158829555e-06,
+      "loss": 1.2096,
+      "step": 2274
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.378061994823075e-06,
+      "loss": 1.4552,
+      "step": 2275
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.375254103515263e-06,
+      "loss": 1.0839,
+      "step": 2276
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.372445742917683e-06,
+      "loss": 1.1296,
+      "step": 2277
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.369636913988654e-06,
+      "loss": 0.9533,
+      "step": 2278
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.366827617686656e-06,
+      "loss": 1.1196,
+      "step": 2279
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.364017854970332e-06,
+      "loss": 1.0644,
+      "step": 2280
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.36120762679848e-06,
+      "loss": 0.998,
+      "step": 2281
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.358396934130058e-06,
+      "loss": 1.094,
+      "step": 2282
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.355585777924185e-06,
+      "loss": 1.295,
+      "step": 2283
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.352774159140132e-06,
+      "loss": 1.1486,
+      "step": 2284
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.349962078737333e-06,
+      "loss": 1.0762,
+      "step": 2285
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.347149537675381e-06,
+      "loss": 1.1386,
+      "step": 2286
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.344336536914022e-06,
+      "loss": 1.0525,
+      "step": 2287
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.34152307741316e-06,
+      "loss": 1.1473,
+      "step": 2288
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.338709160132856e-06,
+      "loss": 1.1812,
+      "step": 2289
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.3358947860333275e-06,
+      "loss": 1.0392,
+      "step": 2290
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.3330799560749464e-06,
+      "loss": 1.0917,
+      "step": 2291
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.330264671218244e-06,
+      "loss": 1.0915,
+      "step": 2292
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.327448932423902e-06,
+      "loss": 1.0251,
+      "step": 2293
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.3246327406527585e-06,
+      "loss": 1.176,
+      "step": 2294
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.321816096865808e-06,
+      "loss": 0.9686,
+      "step": 2295
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.318999002024198e-06,
+      "loss": 1.1007,
+      "step": 2296
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.316181457089232e-06,
+      "loss": 1.2085,
+      "step": 2297
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.313363463022361e-06,
+      "loss": 1.07,
+      "step": 2298
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.3105450207851955e-06,
+      "loss": 1.1336,
+      "step": 2299
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.307726131339497e-06,
+      "loss": 0.9865,
+      "step": 2300
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.304906795647179e-06,
+      "loss": 1.1915,
+      "step": 2301
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.302087014670306e-06,
+      "loss": 1.2114,
+      "step": 2302
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2992667893711e-06,
+      "loss": 1.1279,
+      "step": 2303
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.296446120711928e-06,
+      "loss": 1.0619,
+      "step": 2304
+    },
+    {
+      "epoch": 0.42,
+      "eval_loss": 1.9496101140975952,
+      "eval_runtime": 0.1737,
+      "eval_samples_per_second": 17.272,
+      "eval_steps_per_second": 17.272,
+      "step": 2304
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2936250096553085e-06,
+      "loss": 1.189,
+      "step": 2305
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.29080345716392e-06,
+      "loss": 1.3117,
+      "step": 2306
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.2879814642005785e-06,
+      "loss": 1.2111,
+      "step": 2307
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.285159031728261e-06,
+      "loss": 1.0479,
+      "step": 2308
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.282336160710089e-06,
+      "loss": 1.0655,
+      "step": 2309
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.279512852109335e-06,
+      "loss": 1.0399,
+      "step": 2310
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.276689106889421e-06,
+      "loss": 1.1988,
+      "step": 2311
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.273864926013917e-06,
+      "loss": 1.2827,
+      "step": 2312
+    },
+    {
+      "epoch": 0.42,
+      "learning_rate": 6.271040310446545e-06,
+      "loss": 1.0486,
+      "step": 2313
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2682152611511706e-06,
+      "loss": 1.1378,
+      "step": 2314
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2653897790918095e-06,
+      "loss": 1.1645,
+      "step": 2315
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.262563865232627e-06,
+      "loss": 1.0691,
+      "step": 2316
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.259737520537932e-06,
+      "loss": 1.1615,
+      "step": 2317
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.256910745972183e-06,
+      "loss": 1.1244,
+      "step": 2318
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.254083542499985e-06,
+      "loss": 1.1286,
+      "step": 2319
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2512559110860895e-06,
+      "loss": 1.1139,
+      "step": 2320
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.248427852695391e-06,
+      "loss": 1.1191,
+      "step": 2321
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.245599368292935e-06,
+      "loss": 1.1928,
+      "step": 2322
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.242770458843909e-06,
+      "loss": 1.0872,
+      "step": 2323
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.239941125313645e-06,
+      "loss": 1.1527,
+      "step": 2324
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.237111368667619e-06,
+      "loss": 0.9564,
+      "step": 2325
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.234281189871458e-06,
+      "loss": 1.0388,
+      "step": 2326
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.231450589890924e-06,
+      "loss": 1.0782,
+      "step": 2327
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2286195696919295e-06,
+      "loss": 1.1506,
+      "step": 2328
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2257881302405265e-06,
+      "loss": 1.1964,
+      "step": 2329
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.222956272502912e-06,
+      "loss": 1.0886,
+      "step": 2330
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.220123997445424e-06,
+      "loss": 1.0922,
+      "step": 2331
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.2172913060345464e-06,
+      "loss": 1.1366,
+      "step": 2332
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.214458199236902e-06,
+      "loss": 1.1909,
+      "step": 2333
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.211624678019252e-06,
+      "loss": 0.9621,
+      "step": 2334
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.20879074334851e-06,
+      "loss": 1.0913,
+      "step": 2335
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.205956396191719e-06,
+      "loss": 1.0484,
+      "step": 2336
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.20312163751607e-06,
+      "loss": 1.1561,
+      "step": 2337
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.20028646828889e-06,
+      "loss": 1.0801,
+      "step": 2338
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.197450889477651e-06,
+      "loss": 1.1281,
+      "step": 2339
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.194614902049959e-06,
+      "loss": 1.1076,
+      "step": 2340
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.191778506973565e-06,
+      "loss": 1.172,
+      "step": 2341
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1889417052163535e-06,
+      "loss": 1.1888,
+      "step": 2342
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.186104497746353e-06,
+      "loss": 1.0408,
+      "step": 2343
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.183266885531728e-06,
+      "loss": 1.1716,
+      "step": 2344
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.18042886954078e-06,
+      "loss": 1.068,
+      "step": 2345
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1775904507419516e-06,
+      "loss": 0.9668,
+      "step": 2346
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.174751630103819e-06,
+      "loss": 1.0561,
+      "step": 2347
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.171912408595098e-06,
+      "loss": 1.0515,
+      "step": 2348
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.16907278718464e-06,
+      "loss": 1.1029,
+      "step": 2349
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.166232766841435e-06,
+      "loss": 1.169,
+      "step": 2350
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.163392348534605e-06,
+      "loss": 1.0554,
+      "step": 2351
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.160551533233412e-06,
+      "loss": 1.1318,
+      "step": 2352
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.157710321907252e-06,
+      "loss": 1.1141,
+      "step": 2353
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1548687155256545e-06,
+      "loss": 0.9872,
+      "step": 2354
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.152026715058288e-06,
+      "loss": 1.2876,
+      "step": 2355
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1491843214749506e-06,
+      "loss": 0.9815,
+      "step": 2356
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.146341535745576e-06,
+      "loss": 1.1389,
+      "step": 2357
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1434983588402346e-06,
+      "loss": 1.2051,
+      "step": 2358
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.140654791729126e-06,
+      "loss": 1.1081,
+      "step": 2359
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.137810835382586e-06,
+      "loss": 1.1216,
+      "step": 2360
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.134966490771083e-06,
+      "loss": 1.2251,
+      "step": 2361
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.132121758865215e-06,
+      "loss": 1.0766,
+      "step": 2362
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.129276640635718e-06,
+      "loss": 1.0821,
+      "step": 2363
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.126431137053451e-06,
+      "loss": 1.2808,
+      "step": 2364
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.123585249089414e-06,
+      "loss": 1.1704,
+      "step": 2365
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.120738977714733e-06,
+      "loss": 1.044,
+      "step": 2366
+    },
+    {
+      "epoch": 0.43,
+      "learning_rate": 6.1178923239006625e-06,
+      "loss": 0.9843,
+      "step": 2367
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.115045288618594e-06,
+      "loss": 1.0496,
+      "step": 2368
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1121978728400445e-06,
+      "loss": 1.092,
+      "step": 2369
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.109350077536664e-06,
+      "loss": 1.1222,
+      "step": 2370
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.1065019036802265e-06,
+      "loss": 1.1357,
+      "step": 2371
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.103653352242643e-06,
+      "loss": 1.0317,
+      "step": 2372
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.100804424195945e-06,
+      "loss": 1.1118,
+      "step": 2373
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0979551205123e-06,
+      "loss": 1.0324,
+      "step": 2374
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.095105442163998e-06,
+      "loss": 1.1234,
+      "step": 2375
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.092255390123461e-06,
+      "loss": 1.0267,
+      "step": 2376
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0894049653632355e-06,
+      "loss": 1.0171,
+      "step": 2377
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.086554168855998e-06,
+      "loss": 1.1536,
+      "step": 2378
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.083703001574548e-06,
+      "loss": 1.0683,
+      "step": 2379
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.080851464491817e-06,
+      "loss": 1.313,
+      "step": 2380
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.077999558580856e-06,
+      "loss": 1.1425,
+      "step": 2381
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.075147284814846e-06,
+      "loss": 1.0869,
+      "step": 2382
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.072294644167096e-06,
+      "loss": 1.2629,
+      "step": 2383
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0694416376110335e-06,
+      "loss": 1.2422,
+      "step": 2384
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.066588266120216e-06,
+      "loss": 1.2402,
+      "step": 2385
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.063734530668324e-06,
+      "loss": 0.9728,
+      "step": 2386
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.060880432229161e-06,
+      "loss": 1.1733,
+      "step": 2387
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.058025971776658e-06,
+      "loss": 1.0608,
+      "step": 2388
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.055171150284865e-06,
+      "loss": 1.153,
+      "step": 2389
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.052315968727958e-06,
+      "loss": 1.1606,
+      "step": 2390
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.049460428080233e-06,
+      "loss": 1.0382,
+      "step": 2391
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.046604529316113e-06,
+      "loss": 1.2141,
+      "step": 2392
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.043748273410143e-06,
+      "loss": 1.1439,
+      "step": 2393
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.040891661336982e-06,
+      "loss": 1.1595,
+      "step": 2394
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0380346940714204e-06,
+      "loss": 1.1686,
+      "step": 2395
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.035177372588364e-06,
+      "loss": 0.9802,
+      "step": 2396
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.032319697862844e-06,
+      "loss": 1.2509,
+      "step": 2397
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.029461670870005e-06,
+      "loss": 1.093,
+      "step": 2398
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.026603292585118e-06,
+      "loss": 1.1874,
+      "step": 2399
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.023744563983574e-06,
+      "loss": 1.0707,
+      "step": 2400
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.02088548604088e-06,
+      "loss": 1.2025,
+      "step": 2401
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.018026059732661e-06,
+      "loss": 1.0436,
+      "step": 2402
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.0151662860346685e-06,
+      "loss": 1.1208,
+      "step": 2403
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.012306165922763e-06,
+      "loss": 0.9845,
+      "step": 2404
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.00944570037293e-06,
+      "loss": 1.1725,
+      "step": 2405
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.006584890361271e-06,
+      "loss": 1.136,
+      "step": 2406
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.003723736864007e-06,
+      "loss": 1.0272,
+      "step": 2407
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 6.000862240857469e-06,
+      "loss": 1.1619,
+      "step": 2408
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9980004033181116e-06,
+      "loss": 1.1575,
+      "step": 2409
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9951382252225076e-06,
+      "loss": 1.2129,
+      "step": 2410
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.992275707547338e-06,
+      "loss": 1.2018,
+      "step": 2411
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.989412851269407e-06,
+      "loss": 1.0466,
+      "step": 2412
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.9865496573656305e-06,
+      "loss": 1.1981,
+      "step": 2413
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.983686126813042e-06,
+      "loss": 1.1511,
+      "step": 2414
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.980822260588786e-06,
+      "loss": 1.1528,
+      "step": 2415
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.977958059670127e-06,
+      "loss": 1.086,
+      "step": 2416
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.975093525034441e-06,
+      "loss": 0.9768,
+      "step": 2417
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.972228657659215e-06,
+      "loss": 1.2457,
+      "step": 2418
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.969363458522054e-06,
+      "loss": 1.0925,
+      "step": 2419
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.966497928600676e-06,
+      "loss": 1.1811,
+      "step": 2420
+    },
+    {
+      "epoch": 0.44,
+      "learning_rate": 5.963632068872906e-06,
+      "loss": 1.0969,
+      "step": 2421
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.96076588031669e-06,
+      "loss": 1.1627,
+      "step": 2422
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.95789936391008e-06,
+      "loss": 1.117,
+      "step": 2423
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.955032520631245e-06,
+      "loss": 1.1287,
+      "step": 2424
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.952165351458459e-06,
+      "loss": 1.0382,
+      "step": 2425
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.949297857370112e-06,
+      "loss": 1.1505,
+      "step": 2426
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.946430039344705e-06,
+      "loss": 1.1726,
+      "step": 2427
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.943561898360846e-06,
+      "loss": 0.9848,
+      "step": 2428
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.940693435397256e-06,
+      "loss": 1.1419,
+      "step": 2429
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.937824651432767e-06,
+      "loss": 1.1214,
+      "step": 2430
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.934955547446315e-06,
+      "loss": 1.1277,
+      "step": 2431
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.932086124416952e-06,
+      "loss": 1.0898,
+      "step": 2432
+    },
+    {
+      "epoch": 0.45,
+      "eval_loss": 1.943060278892517,
+      "eval_runtime": 0.169,
+      "eval_samples_per_second": 17.753,
+      "eval_steps_per_second": 17.753,
+      "step": 2432
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.929216383323836e-06,
+      "loss": 1.2001,
+      "step": 2433
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.926346325146234e-06,
+      "loss": 1.151,
+      "step": 2434
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.923475950863517e-06,
+      "loss": 1.0949,
+      "step": 2435
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.92060526145517e-06,
+      "loss": 1.1949,
+      "step": 2436
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.917734257900785e-06,
+      "loss": 1.0519,
+      "step": 2437
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.9148629411800555e-06,
+      "loss": 1.074,
+      "step": 2438
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.911991312272785e-06,
+      "loss": 1.0618,
+      "step": 2439
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.909119372158889e-06,
+      "loss": 1.1663,
+      "step": 2440
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.90624712181838e-06,
+      "loss": 1.2065,
+      "step": 2441
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.903374562231383e-06,
+      "loss": 1.1766,
+      "step": 2442
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.900501694378123e-06,
+      "loss": 1.1576,
+      "step": 2443
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8976285192389386e-06,
+      "loss": 1.1212,
+      "step": 2444
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.894755037794263e-06,
+      "loss": 1.1649,
+      "step": 2445
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.891881251024641e-06,
+      "loss": 1.1039,
+      "step": 2446
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8890071599107215e-06,
+      "loss": 1.1769,
+      "step": 2447
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.886132765433251e-06,
+      "loss": 1.1665,
+      "step": 2448
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.883258068573088e-06,
+      "loss": 1.1377,
+      "step": 2449
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.880383070311187e-06,
+      "loss": 1.1252,
+      "step": 2450
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.87750777162861e-06,
+      "loss": 0.9687,
+      "step": 2451
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.874632173506518e-06,
+      "loss": 1.0293,
+      "step": 2452
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8717562769261785e-06,
+      "loss": 1.1298,
+      "step": 2453
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.868880082868958e-06,
+      "loss": 1.1086,
+      "step": 2454
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8660035923163225e-06,
+      "loss": 1.0005,
+      "step": 2455
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.863126806249843e-06,
+      "loss": 1.229,
+      "step": 2456
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.860249725651192e-06,
+      "loss": 1.0469,
+      "step": 2457
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.857372351502138e-06,
+      "loss": 1.0809,
+      "step": 2458
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.854494684784553e-06,
+      "loss": 1.2306,
+      "step": 2459
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.851616726480406e-06,
+      "loss": 1.1266,
+      "step": 2460
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.848738477571771e-06,
+      "loss": 1.1107,
+      "step": 2461
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.845859939040814e-06,
+      "loss": 1.0542,
+      "step": 2462
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.842981111869807e-06,
+      "loss": 1.1279,
+      "step": 2463
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.840101997041115e-06,
+      "loss": 1.1354,
+      "step": 2464
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.837222595537201e-06,
+      "loss": 1.1183,
+      "step": 2465
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.83434290834063e-06,
+      "loss": 1.2295,
+      "step": 2466
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.831462936434064e-06,
+      "loss": 1.1812,
+      "step": 2467
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.828582680800259e-06,
+      "loss": 1.207,
+      "step": 2468
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8257021424220675e-06,
+      "loss": 1.1381,
+      "step": 2469
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.822821322282443e-06,
+      "loss": 1.1359,
+      "step": 2470
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.819940221364433e-06,
+      "loss": 1.1419,
+      "step": 2471
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.817058840651177e-06,
+      "loss": 1.1371,
+      "step": 2472
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.814177181125915e-06,
+      "loss": 1.1631,
+      "step": 2473
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.811295243771983e-06,
+      "loss": 1.1581,
+      "step": 2474
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.808413029572806e-06,
+      "loss": 1.067,
+      "step": 2475
+    },
+    {
+      "epoch": 0.45,
+      "learning_rate": 5.8055305395119066e-06,
+      "loss": 1.0994,
+      "step": 2476
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.802647774572903e-06,
+      "loss": 1.1586,
+      "step": 2477
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.799764735739506e-06,
+      "loss": 1.151,
+      "step": 2478
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.796881423995518e-06,
+      "loss": 1.0178,
+      "step": 2479
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7939978403248385e-06,
+      "loss": 1.0028,
+      "step": 2480
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.791113985711457e-06,
+      "loss": 1.1979,
+      "step": 2481
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.788229861139455e-06,
+      "loss": 1.0341,
+      "step": 2482
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.785345467593006e-06,
+      "loss": 1.2041,
+      "step": 2483
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.782460806056381e-06,
+      "loss": 1.0525,
+      "step": 2484
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.779575877513932e-06,
+      "loss": 1.0745,
+      "step": 2485
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.776690682950111e-06,
+      "loss": 1.189,
+      "step": 2486
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7738052233494575e-06,
+      "loss": 1.0692,
+      "step": 2487
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.770919499696603e-06,
+      "loss": 1.0585,
+      "step": 2488
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.768033512976266e-06,
+      "loss": 1.1319,
+      "step": 2489
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.765147264173257e-06,
+      "loss": 1.2323,
+      "step": 2490
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.762260754272474e-06,
+      "loss": 1.0506,
+      "step": 2491
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.759373984258908e-06,
+      "loss": 1.1805,
+      "step": 2492
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7564869551176335e-06,
+      "loss": 1.135,
+      "step": 2493
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.75359966783382e-06,
+      "loss": 1.0796,
+      "step": 2494
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.750712123392719e-06,
+      "loss": 1.02,
+      "step": 2495
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.747824322779671e-06,
+      "loss": 0.9823,
+      "step": 2496
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.744936266980108e-06,
+      "loss": 1.1771,
+      "step": 2497
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7420479569795455e-06,
+      "loss": 1.1839,
+      "step": 2498
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.739159393763583e-06,
+      "loss": 1.0663,
+      "step": 2499
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.736270578317912e-06,
+      "loss": 1.1446,
+      "step": 2500
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.73338151162831e-06,
+      "loss": 1.1529,
+      "step": 2501
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.730492194680633e-06,
+      "loss": 1.1378,
+      "step": 2502
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.727602628460833e-06,
+      "loss": 1.2185,
+      "step": 2503
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7247128139549375e-06,
+      "loss": 0.9957,
+      "step": 2504
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.721822752149065e-06,
+      "loss": 1.1289,
+      "step": 2505
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.718932444029415e-06,
+      "loss": 1.1622,
+      "step": 2506
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.716041890582271e-06,
+      "loss": 1.0899,
+      "step": 2507
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.713151092794005e-06,
+      "loss": 1.2407,
+      "step": 2508
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7102600516510645e-06,
+      "loss": 0.9845,
+      "step": 2509
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.707368768139986e-06,
+      "loss": 1.219,
+      "step": 2510
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.7044772432473894e-06,
+      "loss": 1.2822,
+      "step": 2511
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.701585477959971e-06,
+      "loss": 1.0129,
+      "step": 2512
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.698693473264514e-06,
+      "loss": 1.0671,
+      "step": 2513
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.695801230147884e-06,
+      "loss": 1.0196,
+      "step": 2514
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6929087495970215e-06,
+      "loss": 1.2334,
+      "step": 2515
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.690016032598956e-06,
+      "loss": 1.1426,
+      "step": 2516
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.687123080140794e-06,
+      "loss": 1.2242,
+      "step": 2517
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6842298932097224e-06,
+      "loss": 1.1576,
+      "step": 2518
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.681336472793006e-06,
+      "loss": 1.134,
+      "step": 2519
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.678442819877994e-06,
+      "loss": 1.088,
+      "step": 2520
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.675548935452112e-06,
+      "loss": 1.2103,
+      "step": 2521
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.672654820502863e-06,
+      "loss": 1.0696,
+      "step": 2522
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.669760476017834e-06,
+      "loss": 1.0765,
+      "step": 2523
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.666865902984685e-06,
+      "loss": 1.1776,
+      "step": 2524
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6639711023911555e-06,
+      "loss": 1.2016,
+      "step": 2525
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.661076075225064e-06,
+      "loss": 1.0705,
+      "step": 2526
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.6581808224743075e-06,
+      "loss": 1.0564,
+      "step": 2527
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.655285345126853e-06,
+      "loss": 1.172,
+      "step": 2528
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.652389644170754e-06,
+      "loss": 0.9412,
+      "step": 2529
+    },
+    {
+      "epoch": 0.46,
+      "learning_rate": 5.649493720594133e-06,
+      "loss": 1.0926,
+      "step": 2530
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.646597575385191e-06,
+      "loss": 1.1484,
+      "step": 2531
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6437012095322055e-06,
+      "loss": 1.0333,
+      "step": 2532
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.640804624023527e-06,
+      "loss": 1.1048,
+      "step": 2533
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.637907819847583e-06,
+      "loss": 1.0578,
+      "step": 2534
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.635010797992875e-06,
+      "loss": 1.0876,
+      "step": 2535
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.632113559447978e-06,
+      "loss": 1.0242,
+      "step": 2536
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6292161052015436e-06,
+      "loss": 1.0807,
+      "step": 2537
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6263184362422914e-06,
+      "loss": 1.0445,
+      "step": 2538
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.62342055355902e-06,
+      "loss": 1.1666,
+      "step": 2539
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6205224581406e-06,
+      "loss": 1.1122,
+      "step": 2540
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6176241509759725e-06,
+      "loss": 1.1667,
+      "step": 2541
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.614725633054149e-06,
+      "loss": 1.1278,
+      "step": 2542
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.611826905364219e-06,
+      "loss": 1.2003,
+      "step": 2543
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.6089279688953405e-06,
+      "loss": 1.2437,
+      "step": 2544
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.606028824636739e-06,
+      "loss": 1.1456,
+      "step": 2545
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.603129473577717e-06,
+      "loss": 1.0556,
+      "step": 2546
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.600229916707647e-06,
+      "loss": 1.225,
+      "step": 2547
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.597330155015966e-06,
+      "loss": 1.1566,
+      "step": 2548
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.594430189492184e-06,
+      "loss": 1.1689,
+      "step": 2549
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.591530021125884e-06,
+      "loss": 1.0688,
+      "step": 2550
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.588629650906715e-06,
+      "loss": 1.2636,
+      "step": 2551
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.585729079824395e-06,
+      "loss": 1.1616,
+      "step": 2552
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5828283088687084e-06,
+      "loss": 1.1928,
+      "step": 2553
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.579927339029514e-06,
+      "loss": 1.1562,
+      "step": 2554
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.577026171296732e-06,
+      "loss": 0.9817,
+      "step": 2555
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.574124806660354e-06,
+      "loss": 1.034,
+      "step": 2556
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.571223246110438e-06,
+      "loss": 1.0507,
+      "step": 2557
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.568321490637106e-06,
+      "loss": 1.1845,
+      "step": 2558
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.565419541230551e-06,
+      "loss": 1.1366,
+      "step": 2559
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.562517398881031e-06,
+      "loss": 1.1569,
+      "step": 2560
+    },
+    {
+      "epoch": 0.47,
+      "eval_loss": 1.9410213232040405,
+      "eval_runtime": 0.1734,
+      "eval_samples_per_second": 17.303,
+      "eval_steps_per_second": 17.303,
+      "step": 2560
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.559615064578866e-06,
+      "loss": 1.2183,
+      "step": 2561
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.556712539314446e-06,
+      "loss": 1.1217,
+      "step": 2562
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5538098240782245e-06,
+      "loss": 1.0418,
+      "step": 2563
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5509069198607226e-06,
+      "loss": 1.1641,
+      "step": 2564
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.548003827652518e-06,
+      "loss": 0.9861,
+      "step": 2565
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.545100548444262e-06,
+      "loss": 1.2427,
+      "step": 2566
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.542197083226665e-06,
+      "loss": 1.0536,
+      "step": 2567
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.539293432990498e-06,
+      "loss": 1.2539,
+      "step": 2568
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5363895987266024e-06,
+      "loss": 1.0566,
+      "step": 2569
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.533485581425877e-06,
+      "loss": 1.1882,
+      "step": 2570
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.530581382079285e-06,
+      "loss": 1.0858,
+      "step": 2571
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.52767700167785e-06,
+      "loss": 1.1948,
+      "step": 2572
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.52477244121266e-06,
+      "loss": 1.2267,
+      "step": 2573
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.521867701674863e-06,
+      "loss": 1.0648,
+      "step": 2574
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.518962784055668e-06,
+      "loss": 1.0837,
+      "step": 2575
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.516057689346344e-06,
+      "loss": 1.1537,
+      "step": 2576
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.513152418538223e-06,
+      "loss": 1.111,
+      "step": 2577
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.510246972622696e-06,
+      "loss": 1.065,
+      "step": 2578
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.50734135259121e-06,
+      "loss": 1.1121,
+      "step": 2579
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.5044355594352785e-06,
+      "loss": 1.1053,
+      "step": 2580
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.501529594146467e-06,
+      "loss": 1.088,
+      "step": 2581
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.498623457716405e-06,
+      "loss": 1.0851,
+      "step": 2582
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.495717151136779e-06,
+      "loss": 1.1283,
+      "step": 2583
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.492810675399331e-06,
+      "loss": 1.2047,
+      "step": 2584
+    },
+    {
+      "epoch": 0.47,
+      "learning_rate": 5.489904031495864e-06,
+      "loss": 1.0097,
+      "step": 2585
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.486997220418237e-06,
+      "loss": 1.1503,
+      "step": 2586
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4840902431583655e-06,
+      "loss": 1.1715,
+      "step": 2587
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4811831007082214e-06,
+      "loss": 0.9919,
+      "step": 2588
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.478275794059837e-06,
+      "loss": 1.2156,
+      "step": 2589
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4753683242052945e-06,
+      "loss": 1.144,
+      "step": 2590
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.4724606921367355e-06,
+      "loss": 1.1229,
+      "step": 2591
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.469552898846356e-06,
+      "loss": 1.2491,
+      "step": 2592
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.46664494532641e-06,
+      "loss": 1.1589,
+      "step": 2593
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.463736832569201e-06,
+      "loss": 1.0854,
+      "step": 2594
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.460828561567089e-06,
+      "loss": 1.1399,
+      "step": 2595
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.457920133312487e-06,
+      "loss": 1.157,
+      "step": 2596
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.455011548797868e-06,
+      "loss": 1.0905,
+      "step": 2597
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.452102809015749e-06,
+      "loss": 1.0608,
+      "step": 2598
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.449193914958706e-06,
+      "loss": 1.0248,
+      "step": 2599
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.446284867619366e-06,
+      "loss": 1.072,
+      "step": 2600
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.443375667990406e-06,
+      "loss": 1.0686,
+      "step": 2601
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.440466317064559e-06,
+      "loss": 1.1772,
+      "step": 2602
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.437556815834608e-06,
+      "loss": 1.0238,
+      "step": 2603
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.434647165293386e-06,
+      "loss": 1.0351,
+      "step": 2604
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.431737366433779e-06,
+      "loss": 1.2024,
+      "step": 2605
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.428827420248722e-06,
+      "loss": 1.1387,
+      "step": 2606
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.425917327731203e-06,
+      "loss": 1.0666,
+      "step": 2607
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.423007089874253e-06,
+      "loss": 1.0646,
+      "step": 2608
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.420096707670963e-06,
+      "loss": 1.0962,
+      "step": 2609
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.417186182114466e-06,
+      "loss": 1.0593,
+      "step": 2610
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.414275514197943e-06,
+      "loss": 1.1663,
+      "step": 2611
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.411364704914629e-06,
+      "loss": 1.1954,
+      "step": 2612
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.408453755257804e-06,
+      "loss": 1.2963,
+      "step": 2613
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.405542666220796e-06,
+      "loss": 0.9024,
+      "step": 2614
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.402631438796979e-06,
+      "loss": 1.0732,
+      "step": 2615
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3997200739797795e-06,
+      "loss": 1.1304,
+      "step": 2616
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.396808572762666e-06,
+      "loss": 1.1796,
+      "step": 2617
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.393896936139155e-06,
+      "loss": 1.2967,
+      "step": 2618
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.390985165102809e-06,
+      "loss": 1.0461,
+      "step": 2619
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.3880732606472375e-06,
+      "loss": 1.0245,
+      "step": 2620
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.385161223766094e-06,
+      "loss": 1.139,
+      "step": 2621
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.382249055453078e-06,
+      "loss": 1.2052,
+      "step": 2622
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.379336756701935e-06,
+      "loss": 1.0884,
+      "step": 2623
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.376424328506451e-06,
+      "loss": 1.019,
+      "step": 2624
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.37351177186046e-06,
+      "loss": 1.2485,
+      "step": 2625
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.370599087757841e-06,
+      "loss": 1.033,
+      "step": 2626
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.36768627719251e-06,
+      "loss": 1.2212,
+      "step": 2627
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.364773341158434e-06,
+      "loss": 1.0873,
+      "step": 2628
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.361860280649617e-06,
+      "loss": 1.0735,
+      "step": 2629
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.358947096660109e-06,
+      "loss": 1.1401,
+      "step": 2630
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.356033790184e-06,
+      "loss": 1.1329,
+      "step": 2631
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.353120362215423e-06,
+      "loss": 1.0607,
+      "step": 2632
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.350206813748553e-06,
+      "loss": 1.089,
+      "step": 2633
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.347293145777602e-06,
+      "loss": 1.0021,
+      "step": 2634
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.344379359296829e-06,
+      "loss": 1.0331,
+      "step": 2635
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.341465455300531e-06,
+      "loss": 1.1943,
+      "step": 2636
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.33855143478304e-06,
+      "loss": 0.9818,
+      "step": 2637
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.335637298738737e-06,
+      "loss": 1.2201,
+      "step": 2638
+    },
+    {
+      "epoch": 0.48,
+      "learning_rate": 5.332723048162035e-06,
+      "loss": 1.0714,
+      "step": 2639
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.329808684047391e-06,
+      "loss": 1.1498,
+      "step": 2640
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.326894207389296e-06,
+      "loss": 1.0606,
+      "step": 2641
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.323979619182282e-06,
+      "loss": 1.0411,
+      "step": 2642
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3210649204209216e-06,
+      "loss": 1.1723,
+      "step": 2643
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.31815011209982e-06,
+      "loss": 1.1572,
+      "step": 2644
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.315235195213624e-06,
+      "loss": 1.1539,
+      "step": 2645
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3123201707570146e-06,
+      "loss": 1.1412,
+      "step": 2646
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.30940503972471e-06,
+      "loss": 1.1598,
+      "step": 2647
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3064898031114665e-06,
+      "loss": 1.1267,
+      "step": 2648
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.303574461912076e-06,
+      "loss": 1.2279,
+      "step": 2649
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.3006590171213635e-06,
+      "loss": 1.069,
+      "step": 2650
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.297743469734192e-06,
+      "loss": 1.1315,
+      "step": 2651
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2948278207454595e-06,
+      "loss": 1.1262,
+      "step": 2652
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2919120711501e-06,
+      "loss": 1.1647,
+      "step": 2653
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.288996221943073e-06,
+      "loss": 1.0513,
+      "step": 2654
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.286080274119385e-06,
+      "loss": 0.9128,
+      "step": 2655
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.28316422867407e-06,
+      "loss": 1.167,
+      "step": 2656
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.28024808660219e-06,
+      "loss": 1.1351,
+      "step": 2657
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.277331848898849e-06,
+      "loss": 1.0024,
+      "step": 2658
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.274415516559181e-06,
+      "loss": 1.2081,
+      "step": 2659
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.271499090578348e-06,
+      "loss": 1.1498,
+      "step": 2660
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.268582571951549e-06,
+      "loss": 1.1516,
+      "step": 2661
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.265665961674011e-06,
+      "loss": 1.0907,
+      "step": 2662
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.262749260740997e-06,
+      "loss": 1.1873,
+      "step": 2663
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.259832470147796e-06,
+      "loss": 1.1402,
+      "step": 2664
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.256915590889727e-06,
+      "loss": 1.172,
+      "step": 2665
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.253998623962146e-06,
+      "loss": 1.0562,
+      "step": 2666
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.251081570360433e-06,
+      "loss": 0.9452,
+      "step": 2667
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.248164431079999e-06,
+      "loss": 1.2356,
+      "step": 2668
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.245247207116287e-06,
+      "loss": 1.2396,
+      "step": 2669
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.242329899464761e-06,
+      "loss": 1.1368,
+      "step": 2670
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.239412509120924e-06,
+      "loss": 1.0757,
+      "step": 2671
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.236495037080298e-06,
+      "loss": 1.0475,
+      "step": 2672
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.233577484338442e-06,
+      "loss": 1.1168,
+      "step": 2673
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.230659851890934e-06,
+      "loss": 1.1976,
+      "step": 2674
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.227742140733381e-06,
+      "loss": 1.0119,
+      "step": 2675
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.224824351861426e-06,
+      "loss": 1.0828,
+      "step": 2676
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2219064862707235e-06,
+      "loss": 1.131,
+      "step": 2677
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.2189885449569646e-06,
+      "loss": 1.0486,
+      "step": 2678
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.216070528915865e-06,
+      "loss": 1.2725,
+      "step": 2679
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.213152439143162e-06,
+      "loss": 1.155,
+      "step": 2680
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.210234276634622e-06,
+      "loss": 1.063,
+      "step": 2681
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.207316042386033e-06,
+      "loss": 1.2634,
+      "step": 2682
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.204397737393212e-06,
+      "loss": 1.1427,
+      "step": 2683
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.201479362651995e-06,
+      "loss": 1.1409,
+      "step": 2684
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.198560919158244e-06,
+      "loss": 1.1132,
+      "step": 2685
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.195642407907846e-06,
+      "loss": 1.1547,
+      "step": 2686
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.192723829896708e-06,
+      "loss": 1.046,
+      "step": 2687
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.189805186120762e-06,
+      "loss": 1.2441,
+      "step": 2688
+    },
+    {
+      "epoch": 0.49,
+      "eval_loss": 1.9336870908737183,
+      "eval_runtime": 0.1833,
+      "eval_samples_per_second": 16.369,
+      "eval_steps_per_second": 16.369,
+      "step": 2688
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.186886477575964e-06,
+      "loss": 1.2329,
+      "step": 2689
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.183967705258286e-06,
+      "loss": 1.118,
+      "step": 2690
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.1810488701637286e-06,
+      "loss": 1.0264,
+      "step": 2691
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.178129973288312e-06,
+      "loss": 1.037,
+      "step": 2692
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.175211015628072e-06,
+      "loss": 1.096,
+      "step": 2693
+    },
+    {
+      "epoch": 0.49,
+      "learning_rate": 5.172291998179073e-06,
+      "loss": 1.1662,
+      "step": 2694
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.169372921937396e-06,
+      "loss": 1.0905,
+      "step": 2695
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.166453787899143e-06,
+      "loss": 1.071,
+      "step": 2696
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.163534597060433e-06,
+      "loss": 1.1083,
+      "step": 2697
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.160615350417409e-06,
+      "loss": 1.126,
+      "step": 2698
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.157696048966229e-06,
+      "loss": 1.0655,
+      "step": 2699
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.154776693703071e-06,
+      "loss": 1.103,
+      "step": 2700
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1518572856241325e-06,
+      "loss": 1.2807,
+      "step": 2701
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.14893782572563e-06,
+      "loss": 1.2001,
+      "step": 2702
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.146018315003792e-06,
+      "loss": 1.1833,
+      "step": 2703
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1430987544548705e-06,
+      "loss": 1.0909,
+      "step": 2704
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.140179145075132e-06,
+      "loss": 1.154,
+      "step": 2705
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.137259487860862e-06,
+      "loss": 1.1658,
+      "step": 2706
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.134339783808355e-06,
+      "loss": 1.0254,
+      "step": 2707
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.131420033913931e-06,
+      "loss": 1.1563,
+      "step": 2708
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.128500239173921e-06,
+      "loss": 1.0711,
+      "step": 2709
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.12558040058467e-06,
+      "loss": 1.1329,
+      "step": 2710
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.122660519142539e-06,
+      "loss": 1.2225,
+      "step": 2711
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1197405958439085e-06,
+      "loss": 1.0251,
+      "step": 2712
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.116820631685164e-06,
+      "loss": 1.2218,
+      "step": 2713
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.113900627662712e-06,
+      "loss": 1.0089,
+      "step": 2714
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.110980584772971e-06,
+      "loss": 1.1614,
+      "step": 2715
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.108060504012372e-06,
+      "loss": 1.1369,
+      "step": 2716
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.1051403863773585e-06,
+      "loss": 1.1047,
+      "step": 2717
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.102220232864389e-06,
+      "loss": 1.134,
+      "step": 2718
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0993000444699315e-06,
+      "loss": 1.2097,
+      "step": 2719
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.096379822190466e-06,
+      "loss": 1.0933,
+      "step": 2720
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.093459567022485e-06,
+      "loss": 1.0147,
+      "step": 2721
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.090539279962495e-06,
+      "loss": 1.1024,
+      "step": 2722
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0876189620070085e-06,
+      "loss": 1.0509,
+      "step": 2723
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0846986141525515e-06,
+      "loss": 1.0917,
+      "step": 2724
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.081778237395658e-06,
+      "loss": 1.1049,
+      "step": 2725
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.078857832732874e-06,
+      "loss": 1.0518,
+      "step": 2726
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0759374011607554e-06,
+      "loss": 1.0185,
+      "step": 2727
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.073016943675865e-06,
+      "loss": 1.1367,
+      "step": 2728
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.070096461274776e-06,
+      "loss": 1.1554,
+      "step": 2729
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.067175954954068e-06,
+      "loss": 1.1735,
+      "step": 2730
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.064255425710332e-06,
+      "loss": 1.1802,
+      "step": 2731
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.061334874540168e-06,
+      "loss": 1.0092,
+      "step": 2732
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.058414302440175e-06,
+      "loss": 1.3243,
+      "step": 2733
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.055493710406969e-06,
+      "loss": 1.1171,
+      "step": 2734
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0525730994371694e-06,
+      "loss": 0.9379,
+      "step": 2735
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0496524705273965e-06,
+      "loss": 1.0971,
+      "step": 2736
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.046731824674285e-06,
+      "loss": 1.1099,
+      "step": 2737
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.043811162874472e-06,
+      "loss": 0.9944,
+      "step": 2738
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.040890486124601e-06,
+      "loss": 1.0405,
+      "step": 2739
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.037969795421317e-06,
+      "loss": 1.0597,
+      "step": 2740
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.035049091761274e-06,
+      "loss": 1.1108,
+      "step": 2741
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.0321283761411285e-06,
+      "loss": 1.2704,
+      "step": 2742
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.029207649557542e-06,
+      "loss": 1.0794,
+      "step": 2743
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.026286913007176e-06,
+      "loss": 1.1292,
+      "step": 2744
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.023366167486703e-06,
+      "loss": 1.278,
+      "step": 2745
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.020445413992792e-06,
+      "loss": 0.9647,
+      "step": 2746
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.017524653522115e-06,
+      "loss": 1.2117,
+      "step": 2747
+    },
+    {
+      "epoch": 0.5,
+      "learning_rate": 5.014603887071351e-06,
+      "loss": 1.0571,
+      "step": 2748
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.011683115637177e-06,
+      "loss": 0.9492,
+      "step": 2749
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.008762340216273e-06,
+      "loss": 1.2084,
+      "step": 2750
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.0058415618053215e-06,
+      "loss": 1.1284,
+      "step": 2751
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5.002920781401003e-06,
+      "loss": 1.1779,
+      "step": 2752
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 5e-06,
+      "loss": 1.1372,
+      "step": 2753
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9970792185989994e-06,
+      "loss": 1.1104,
+      "step": 2754
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.994158438194679e-06,
+      "loss": 1.3204,
+      "step": 2755
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.991237659783728e-06,
+      "loss": 1.122,
+      "step": 2756
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.988316884362825e-06,
+      "loss": 1.0868,
+      "step": 2757
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.98539611292865e-06,
+      "loss": 1.113,
+      "step": 2758
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.982475346477886e-06,
+      "loss": 0.9515,
+      "step": 2759
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.979554586007212e-06,
+      "loss": 1.125,
+      "step": 2760
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.976633832513298e-06,
+      "loss": 1.2003,
+      "step": 2761
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.973713086992825e-06,
+      "loss": 1.1236,
+      "step": 2762
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9707923504424615e-06,
+      "loss": 1.09,
+      "step": 2763
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.967871623858872e-06,
+      "loss": 1.0378,
+      "step": 2764
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.964950908238728e-06,
+      "loss": 0.9681,
+      "step": 2765
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.962030204578683e-06,
+      "loss": 1.1876,
+      "step": 2766
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9591095138754e-06,
+      "loss": 1.0853,
+      "step": 2767
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.956188837125529e-06,
+      "loss": 1.0167,
+      "step": 2768
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9532681753257154e-06,
+      "loss": 1.0755,
+      "step": 2769
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.950347529472605e-06,
+      "loss": 1.208,
+      "step": 2770
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.947426900562834e-06,
+      "loss": 1.1028,
+      "step": 2771
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.944506289593032e-06,
+      "loss": 1.21,
+      "step": 2772
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.941585697559826e-06,
+      "loss": 1.1438,
+      "step": 2773
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.938665125459835e-06,
+      "loss": 1.1991,
+      "step": 2774
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.935744574289668e-06,
+      "loss": 1.0943,
+      "step": 2775
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.932824045045933e-06,
+      "loss": 1.0951,
+      "step": 2776
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.929903538725227e-06,
+      "loss": 1.0886,
+      "step": 2777
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.926983056324136e-06,
+      "loss": 1.0753,
+      "step": 2778
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.924062598839246e-06,
+      "loss": 1.0894,
+      "step": 2779
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.9211421672671275e-06,
+      "loss": 1.1161,
+      "step": 2780
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.918221762604343e-06,
+      "loss": 1.0861,
+      "step": 2781
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.915301385847451e-06,
+      "loss": 1.236,
+      "step": 2782
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.912381037992994e-06,
+      "loss": 1.105,
+      "step": 2783
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.909460720037506e-06,
+      "loss": 1.1805,
+      "step": 2784
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.906540432977516e-06,
+      "loss": 1.1758,
+      "step": 2785
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.903620177809537e-06,
+      "loss": 1.2458,
+      "step": 2786
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.90069995553007e-06,
+      "loss": 1.1491,
+      "step": 2787
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.897779767135613e-06,
+      "loss": 0.9286,
+      "step": 2788
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8948596136226415e-06,
+      "loss": 1.0901,
+      "step": 2789
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.891939495987629e-06,
+      "loss": 1.1122,
+      "step": 2790
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.889019415227031e-06,
+      "loss": 1.1224,
+      "step": 2791
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.886099372337289e-06,
+      "loss": 1.039,
+      "step": 2792
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.883179368314838e-06,
+      "loss": 1.0921,
+      "step": 2793
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.880259404156094e-06,
+      "loss": 1.139,
+      "step": 2794
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.877339480857461e-06,
+      "loss": 1.1567,
+      "step": 2795
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8744195994153315e-06,
+      "loss": 1.0823,
+      "step": 2796
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8714997608260814e-06,
+      "loss": 1.153,
+      "step": 2797
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.8685799660860685e-06,
+      "loss": 1.0166,
+      "step": 2798
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.865660216191646e-06,
+      "loss": 1.028,
+      "step": 2799
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.862740512139142e-06,
+      "loss": 1.0859,
+      "step": 2800
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.859820854924869e-06,
+      "loss": 1.0082,
+      "step": 2801
+    },
+    {
+      "epoch": 0.51,
+      "learning_rate": 4.856901245545131e-06,
+      "loss": 1.1386,
+      "step": 2802
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.85398168499621e-06,
+      "loss": 1.1239,
+      "step": 2803
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.851062174274372e-06,
+      "loss": 1.1231,
+      "step": 2804
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.848142714375869e-06,
+      "loss": 1.0772,
+      "step": 2805
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.845223306296932e-06,
+      "loss": 1.1552,
+      "step": 2806
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.842303951033773e-06,
+      "loss": 1.0685,
+      "step": 2807
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.839384649582594e-06,
+      "loss": 1.1985,
+      "step": 2808
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.836465402939568e-06,
+      "loss": 1.2203,
+      "step": 2809
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.833546212100859e-06,
+      "loss": 1.1233,
+      "step": 2810
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.830627078062606e-06,
+      "loss": 1.0602,
+      "step": 2811
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8277080018209275e-06,
+      "loss": 1.0651,
+      "step": 2812
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8247889843719294e-06,
+      "loss": 1.1758,
+      "step": 2813
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.821870026711691e-06,
+      "loss": 1.304,
+      "step": 2814
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.818951129836271e-06,
+      "loss": 1.1937,
+      "step": 2815
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.816032294741715e-06,
+      "loss": 1.1933,
+      "step": 2816
+    },
+    {
+      "epoch": 0.52,
+      "eval_loss": 1.9383550882339478,
+      "eval_runtime": 0.1902,
+      "eval_samples_per_second": 15.772,
+      "eval_steps_per_second": 15.772,
+      "step": 2816
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8131135224240385e-06,
+      "loss": 0.9927,
+      "step": 2817
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.810194813879238e-06,
+      "loss": 1.0161,
+      "step": 2818
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.807276170103293e-06,
+      "loss": 1.0502,
+      "step": 2819
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.804357592092157e-06,
+      "loss": 1.2328,
+      "step": 2820
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.8014390808417566e-06,
+      "loss": 1.2613,
+      "step": 2821
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.798520637348006e-06,
+      "loss": 1.1905,
+      "step": 2822
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.79560226260679e-06,
+      "loss": 1.1068,
+      "step": 2823
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.792683957613968e-06,
+      "loss": 1.1193,
+      "step": 2824
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.78976572336538e-06,
+      "loss": 1.2108,
+      "step": 2825
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.78684756085684e-06,
+      "loss": 1.0588,
+      "step": 2826
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.783929471084136e-06,
+      "loss": 1.2527,
+      "step": 2827
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.781011455043037e-06,
+      "loss": 1.0802,
+      "step": 2828
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.77809351372928e-06,
+      "loss": 1.1554,
+      "step": 2829
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.775175648138576e-06,
+      "loss": 1.089,
+      "step": 2830
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7722578592666195e-06,
+      "loss": 1.018,
+      "step": 2831
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.769340148109067e-06,
+      "loss": 1.2068,
+      "step": 2832
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.766422515661559e-06,
+      "loss": 1.0337,
+      "step": 2833
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.763504962919703e-06,
+      "loss": 1.175,
+      "step": 2834
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.760587490879077e-06,
+      "loss": 1.0561,
+      "step": 2835
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.75767010053524e-06,
+      "loss": 1.1434,
+      "step": 2836
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.754752792883716e-06,
+      "loss": 1.245,
+      "step": 2837
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.751835568920001e-06,
+      "loss": 1.1218,
+      "step": 2838
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.748918429639568e-06,
+      "loss": 0.8746,
+      "step": 2839
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.7460013760378555e-06,
+      "loss": 1.0678,
+      "step": 2840
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.743084409110273e-06,
+      "loss": 1.1871,
+      "step": 2841
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.740167529852207e-06,
+      "loss": 1.0564,
+      "step": 2842
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.737250739259005e-06,
+      "loss": 1.0163,
+      "step": 2843
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.734334038325989e-06,
+      "loss": 1.3158,
+      "step": 2844
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.731417428048452e-06,
+      "loss": 1.2351,
+      "step": 2845
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.728500909421654e-06,
+      "loss": 1.084,
+      "step": 2846
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.72558448344082e-06,
+      "loss": 1.081,
+      "step": 2847
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.722668151101152e-06,
+      "loss": 1.1317,
+      "step": 2848
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.719751913397813e-06,
+      "loss": 1.0984,
+      "step": 2849
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.716835771325932e-06,
+      "loss": 0.9871,
+      "step": 2850
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.713919725880616e-06,
+      "loss": 1.0865,
+      "step": 2851
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.711003778056929e-06,
+      "loss": 1.1212,
+      "step": 2852
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.708087928849903e-06,
+      "loss": 1.0601,
+      "step": 2853
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.705172179254541e-06,
+      "loss": 1.0847,
+      "step": 2854
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.702256530265808e-06,
+      "loss": 1.2792,
+      "step": 2855
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.699340982878638e-06,
+      "loss": 0.9929,
+      "step": 2856
+    },
+    {
+      "epoch": 0.52,
+      "learning_rate": 4.696425538087927e-06,
+      "loss": 1.1372,
+      "step": 2857
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6935101968885335e-06,
+      "loss": 1.1757,
+      "step": 2858
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.690594960275292e-06,
+      "loss": 1.1577,
+      "step": 2859
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.687679829242988e-06,
+      "loss": 1.1019,
+      "step": 2860
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.684764804786377e-06,
+      "loss": 1.2286,
+      "step": 2861
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.681849887900181e-06,
+      "loss": 1.0067,
+      "step": 2862
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.678935079579081e-06,
+      "loss": 1.049,
+      "step": 2863
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.676020380817718e-06,
+      "loss": 1.0229,
+      "step": 2864
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.673105792610706e-06,
+      "loss": 1.1036,
+      "step": 2865
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.670191315952612e-06,
+      "loss": 1.1203,
+      "step": 2866
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.667276951837966e-06,
+      "loss": 1.17,
+      "step": 2867
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.664362701261265e-06,
+      "loss": 1.1156,
+      "step": 2868
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.661448565216962e-06,
+      "loss": 1.1884,
+      "step": 2869
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.658534544699471e-06,
+      "loss": 1.1184,
+      "step": 2870
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.655620640703172e-06,
+      "loss": 1.1411,
+      "step": 2871
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6527068542224e-06,
+      "loss": 1.0558,
+      "step": 2872
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.649793186251449e-06,
+      "loss": 1.1222,
+      "step": 2873
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.646879637784579e-06,
+      "loss": 1.1277,
+      "step": 2874
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.643966209816001e-06,
+      "loss": 1.0273,
+      "step": 2875
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.641052903339892e-06,
+      "loss": 1.1224,
+      "step": 2876
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.638139719350385e-06,
+      "loss": 1.1023,
+      "step": 2877
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.635226658841567e-06,
+      "loss": 1.1168,
+      "step": 2878
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.632313722807492e-06,
+      "loss": 1.125,
+      "step": 2879
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.6294009122421625e-06,
+      "loss": 0.9739,
+      "step": 2880
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.626488228139541e-06,
+      "loss": 1.317,
+      "step": 2881
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.623575671493551e-06,
+      "loss": 1.1052,
+      "step": 2882
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.620663243298068e-06,
+      "loss": 1.0942,
+      "step": 2883
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.617750944546922e-06,
+      "loss": 1.0588,
+      "step": 2884
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.614838776233908e-06,
+      "loss": 1.0464,
+      "step": 2885
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.611926739352764e-06,
+      "loss": 1.1044,
+      "step": 2886
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.609014834897191e-06,
+      "loss": 1.1979,
+      "step": 2887
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.606103063860847e-06,
+      "loss": 1.0916,
+      "step": 2888
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.603191427237336e-06,
+      "loss": 1.1351,
+      "step": 2889
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.600279926020221e-06,
+      "loss": 1.2239,
+      "step": 2890
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.597368561203023e-06,
+      "loss": 1.1488,
+      "step": 2891
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5944573337792076e-06,
+      "loss": 1.2019,
+      "step": 2892
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.591546244742198e-06,
+      "loss": 1.0745,
+      "step": 2893
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.588635295085373e-06,
+      "loss": 0.9994,
+      "step": 2894
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5857244858020595e-06,
+      "loss": 0.9171,
+      "step": 2895
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.582813817885536e-06,
+      "loss": 1.0689,
+      "step": 2896
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.579903292329038e-06,
+      "loss": 1.0994,
+      "step": 2897
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5769929101257466e-06,
+      "loss": 0.9988,
+      "step": 2898
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.574082672268799e-06,
+      "loss": 1.1612,
+      "step": 2899
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.571172579751279e-06,
+      "loss": 1.2018,
+      "step": 2900
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5682626335662214e-06,
+      "loss": 1.1561,
+      "step": 2901
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.565352834706615e-06,
+      "loss": 1.0991,
+      "step": 2902
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.562443184165395e-06,
+      "loss": 1.0558,
+      "step": 2903
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.559533682935442e-06,
+      "loss": 1.1018,
+      "step": 2904
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.556624332009596e-06,
+      "loss": 1.2451,
+      "step": 2905
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5537151323806375e-06,
+      "loss": 1.1592,
+      "step": 2906
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.550806085041295e-06,
+      "loss": 0.9637,
+      "step": 2907
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.547897190984252e-06,
+      "loss": 1.2222,
+      "step": 2908
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.5449884512021345e-06,
+      "loss": 1.0645,
+      "step": 2909
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.542079866687513e-06,
+      "loss": 1.0448,
+      "step": 2910
+    },
+    {
+      "epoch": 0.53,
+      "learning_rate": 4.539171438432913e-06,
+      "loss": 1.0256,
+      "step": 2911
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.536263167430802e-06,
+      "loss": 1.0939,
+      "step": 2912
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.533355054673591e-06,
+      "loss": 1.0403,
+      "step": 2913
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.5304471011536445e-06,
+      "loss": 1.2008,
+      "step": 2914
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.527539307863267e-06,
+      "loss": 1.1767,
+      "step": 2915
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.524631675794707e-06,
+      "loss": 1.2451,
+      "step": 2916
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.521724205940165e-06,
+      "loss": 1.3826,
+      "step": 2917
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.51881689929178e-06,
+      "loss": 1.215,
+      "step": 2918
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.515909756841636e-06,
+      "loss": 1.1666,
+      "step": 2919
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.513002779581765e-06,
+      "loss": 1.1163,
+      "step": 2920
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.510095968504135e-06,
+      "loss": 1.0696,
+      "step": 2921
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.50718932460067e-06,
+      "loss": 1.0329,
+      "step": 2922
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.504282848863222e-06,
+      "loss": 1.2007,
+      "step": 2923
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.501376542283594e-06,
+      "loss": 1.0305,
+      "step": 2924
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.498470405853533e-06,
+      "loss": 1.3269,
+      "step": 2925
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.495564440564724e-06,
+      "loss": 1.0696,
+      "step": 2926
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.492658647408791e-06,
+      "loss": 1.0628,
+      "step": 2927
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.489753027377306e-06,
+      "loss": 1.0696,
+      "step": 2928
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.486847581461778e-06,
+      "loss": 1.095,
+      "step": 2929
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.483942310653656e-06,
+      "loss": 1.2111,
+      "step": 2930
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.481037215944334e-06,
+      "loss": 1.1203,
+      "step": 2931
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.478132298325139e-06,
+      "loss": 1.1725,
+      "step": 2932
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.475227558787341e-06,
+      "loss": 1.2821,
+      "step": 2933
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.472322998322151e-06,
+      "loss": 1.1264,
+      "step": 2934
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.469418617920718e-06,
+      "loss": 1.2142,
+      "step": 2935
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.466514418574124e-06,
+      "loss": 1.0188,
+      "step": 2936
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.463610401273399e-06,
+      "loss": 1.0917,
+      "step": 2937
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.460706567009504e-06,
+      "loss": 1.1552,
+      "step": 2938
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.457802916773338e-06,
+      "loss": 1.1448,
+      "step": 2939
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.45489945155574e-06,
+      "loss": 1.1026,
+      "step": 2940
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.451996172347482e-06,
+      "loss": 0.9998,
+      "step": 2941
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.44909308013928e-06,
+      "loss": 1.1135,
+      "step": 2942
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.446190175921776e-06,
+      "loss": 1.1719,
+      "step": 2943
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.443287460685554e-06,
+      "loss": 1.0921,
+      "step": 2944
+    },
+    {
+      "epoch": 0.54,
+      "eval_loss": 1.935546875,
+      "eval_runtime": 0.1878,
+      "eval_samples_per_second": 15.97,
+      "eval_steps_per_second": 15.97,
+      "step": 2944
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.440384935421136e-06,
+      "loss": 1.215,
+      "step": 2945
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.437482601118972e-06,
+      "loss": 1.2271,
+      "step": 2946
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.43458045876945e-06,
+      "loss": 1.3151,
+      "step": 2947
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.431678509362895e-06,
+      "loss": 1.1691,
+      "step": 2948
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.428776753889565e-06,
+      "loss": 1.1428,
+      "step": 2949
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.425875193339646e-06,
+      "loss": 1.2036,
+      "step": 2950
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.422973828703269e-06,
+      "loss": 1.1207,
+      "step": 2951
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.420072660970488e-06,
+      "loss": 1.0551,
+      "step": 2952
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4171716911312915e-06,
+      "loss": 1.0527,
+      "step": 2953
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.4142709201756075e-06,
+      "loss": 1.2728,
+      "step": 2954
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.411370349093288e-06,
+      "loss": 1.0732,
+      "step": 2955
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.408469978874116e-06,
+      "loss": 1.1567,
+      "step": 2956
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.405569810507818e-06,
+      "loss": 1.1856,
+      "step": 2957
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.402669844984038e-06,
+      "loss": 1.0723,
+      "step": 2958
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.399770083292355e-06,
+      "loss": 1.1349,
+      "step": 2959
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.396870526422284e-06,
+      "loss": 1.0345,
+      "step": 2960
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.393971175363263e-06,
+      "loss": 1.1257,
+      "step": 2961
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.391072031104662e-06,
+      "loss": 1.191,
+      "step": 2962
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.388173094635783e-06,
+      "loss": 1.1471,
+      "step": 2963
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.385274366945852e-06,
+      "loss": 1.183,
+      "step": 2964
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.38237584902403e-06,
+      "loss": 1.1358,
+      "step": 2965
+    },
+    {
+      "epoch": 0.54,
+      "learning_rate": 4.379477541859402e-06,
+      "loss": 1.1799,
+      "step": 2966
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.37657944644098e-06,
+      "loss": 1.1954,
+      "step": 2967
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.37368156375771e-06,
+      "loss": 0.9977,
+      "step": 2968
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.370783894798459e-06,
+      "loss": 1.1407,
+      "step": 2969
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3678864405520225e-06,
+      "loss": 1.1618,
+      "step": 2970
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.364989202007126e-06,
+      "loss": 1.0908,
+      "step": 2971
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3620921801524195e-06,
+      "loss": 1.1861,
+      "step": 2972
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3591953759764735e-06,
+      "loss": 1.0788,
+      "step": 2973
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.356298790467796e-06,
+      "loss": 1.031,
+      "step": 2974
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.353402424614811e-06,
+      "loss": 1.0045,
+      "step": 2975
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.350506279405868e-06,
+      "loss": 0.8676,
+      "step": 2976
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.347610355829248e-06,
+      "loss": 1.1263,
+      "step": 2977
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.344714654873148e-06,
+      "loss": 1.0978,
+      "step": 2978
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.341819177525695e-06,
+      "loss": 1.0589,
+      "step": 2979
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.338923924774937e-06,
+      "loss": 1.1585,
+      "step": 2980
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.336028897608846e-06,
+      "loss": 1.2218,
+      "step": 2981
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3331340970153165e-06,
+      "loss": 1.2202,
+      "step": 2982
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.3302395239821685e-06,
+      "loss": 1.3376,
+      "step": 2983
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.327345179497138e-06,
+      "loss": 1.0585,
+      "step": 2984
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.32445106454789e-06,
+      "loss": 0.9935,
+      "step": 2985
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.321557180122008e-06,
+      "loss": 0.9937,
+      "step": 2986
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.318663527206995e-06,
+      "loss": 1.193,
+      "step": 2987
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.315770106790279e-06,
+      "loss": 1.1094,
+      "step": 2988
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.312876919859208e-06,
+      "loss": 1.2384,
+      "step": 2989
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.309983967401043e-06,
+      "loss": 1.0411,
+      "step": 2990
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.307091250402979e-06,
+      "loss": 1.0971,
+      "step": 2991
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.304198769852119e-06,
+      "loss": 1.1425,
+      "step": 2992
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.301306526735486e-06,
+      "loss": 1.1658,
+      "step": 2993
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.298414522040031e-06,
+      "loss": 1.0581,
+      "step": 2994
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.295522756752612e-06,
+      "loss": 1.2187,
+      "step": 2995
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.292631231860014e-06,
+      "loss": 0.9095,
+      "step": 2996
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.289739948348937e-06,
+      "loss": 1.1915,
+      "step": 2997
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.286848907205997e-06,
+      "loss": 1.0234,
+      "step": 2998
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.28395810941773e-06,
+      "loss": 1.1003,
+      "step": 2999
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.281067555970587e-06,
+      "loss": 1.1811,
+      "step": 3000
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.278177247850936e-06,
+      "loss": 1.1002,
+      "step": 3001
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.275287186045063e-06,
+      "loss": 1.0046,
+      "step": 3002
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2723973715391685e-06,
+      "loss": 1.1536,
+      "step": 3003
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.269507805319368e-06,
+      "loss": 1.112,
+      "step": 3004
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.266618488371692e-06,
+      "loss": 1.178,
+      "step": 3005
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.263729421682089e-06,
+      "loss": 1.1339,
+      "step": 3006
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.260840606236418e-06,
+      "loss": 1.2335,
+      "step": 3007
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.257952043020457e-06,
+      "loss": 1.3153,
+      "step": 3008
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.255063733019893e-06,
+      "loss": 1.0651,
+      "step": 3009
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.252175677220328e-06,
+      "loss": 1.1785,
+      "step": 3010
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.249287876607282e-06,
+      "loss": 1.1511,
+      "step": 3011
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.2464003321661816e-06,
+      "loss": 1.0395,
+      "step": 3012
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.243513044882366e-06,
+      "loss": 1.2827,
+      "step": 3013
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.240626015741094e-06,
+      "loss": 1.1021,
+      "step": 3014
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.237739245727527e-06,
+      "loss": 1.0259,
+      "step": 3015
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.234852735826744e-06,
+      "loss": 1.1048,
+      "step": 3016
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.231966487023735e-06,
+      "loss": 1.2174,
+      "step": 3017
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.229080500303398e-06,
+      "loss": 1.3703,
+      "step": 3018
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.226194776650542e-06,
+      "loss": 1.0926,
+      "step": 3019
+    },
+    {
+      "epoch": 0.55,
+      "learning_rate": 4.22330931704989e-06,
+      "loss": 0.9748,
+      "step": 3020
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2204241224860685e-06,
+      "loss": 1.0195,
+      "step": 3021
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.21753919394362e-06,
+      "loss": 1.0953,
+      "step": 3022
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2146545324069946e-06,
+      "loss": 1.134,
+      "step": 3023
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.2117701388605465e-06,
+      "loss": 1.1609,
+      "step": 3024
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.208886014288544e-06,
+      "loss": 1.0549,
+      "step": 3025
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.206002159675162e-06,
+      "loss": 1.0519,
+      "step": 3026
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.203118576004483e-06,
+      "loss": 1.1363,
+      "step": 3027
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.200235264260496e-06,
+      "loss": 1.0816,
+      "step": 3028
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.197352225427099e-06,
+      "loss": 0.9954,
+      "step": 3029
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.194469460488094e-06,
+      "loss": 1.1259,
+      "step": 3030
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.191586970427196e-06,
+      "loss": 1.1329,
+      "step": 3031
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.188704756228018e-06,
+      "loss": 1.1932,
+      "step": 3032
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.185822818874085e-06,
+      "loss": 1.3612,
+      "step": 3033
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.182941159348824e-06,
+      "loss": 1.0757,
+      "step": 3034
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.180059778635569e-06,
+      "loss": 1.083,
+      "step": 3035
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.177178677717558e-06,
+      "loss": 1.1461,
+      "step": 3036
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.174297857577934e-06,
+      "loss": 1.1544,
+      "step": 3037
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.171417319199743e-06,
+      "loss": 1.1285,
+      "step": 3038
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.168537063565937e-06,
+      "loss": 0.9847,
+      "step": 3039
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.165657091659371e-06,
+      "loss": 1.06,
+      "step": 3040
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.162777404462801e-06,
+      "loss": 1.0814,
+      "step": 3041
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.159898002958888e-06,
+      "loss": 1.166,
+      "step": 3042
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.157018888130195e-06,
+      "loss": 1.02,
+      "step": 3043
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.154140060959187e-06,
+      "loss": 1.0742,
+      "step": 3044
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1512615224282306e-06,
+      "loss": 1.2083,
+      "step": 3045
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.148383273519595e-06,
+      "loss": 1.1063,
+      "step": 3046
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.145505315215449e-06,
+      "loss": 1.1133,
+      "step": 3047
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.142627648497863e-06,
+      "loss": 1.0783,
+      "step": 3048
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.1397502743488086e-06,
+      "loss": 0.9802,
+      "step": 3049
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.136873193750158e-06,
+      "loss": 1.1002,
+      "step": 3050
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.133996407683679e-06,
+      "loss": 1.0922,
+      "step": 3051
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.131119917131044e-06,
+      "loss": 1.1616,
+      "step": 3052
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.128243723073822e-06,
+      "loss": 1.0267,
+      "step": 3053
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.125367826493483e-06,
+      "loss": 0.9804,
+      "step": 3054
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.122492228371392e-06,
+      "loss": 1.2589,
+      "step": 3055
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.119616929688814e-06,
+      "loss": 1.108,
+      "step": 3056
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.116741931426914e-06,
+      "loss": 1.0622,
+      "step": 3057
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.11386723456675e-06,
+      "loss": 1.2423,
+      "step": 3058
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.110992840089279e-06,
+      "loss": 0.9807,
+      "step": 3059
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.10811874897536e-06,
+      "loss": 1.1978,
+      "step": 3060
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.105244962205738e-06,
+      "loss": 1.1398,
+      "step": 3061
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.102371480761062e-06,
+      "loss": 1.0149,
+      "step": 3062
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.099498305621878e-06,
+      "loss": 1.1683,
+      "step": 3063
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.096625437768619e-06,
+      "loss": 1.0571,
+      "step": 3064
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.093752878181621e-06,
+      "loss": 1.0038,
+      "step": 3065
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.090880627841113e-06,
+      "loss": 1.0782,
+      "step": 3066
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.088008687727215e-06,
+      "loss": 1.1094,
+      "step": 3067
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.085137058819947e-06,
+      "loss": 1.0433,
+      "step": 3068
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.082265742099216e-06,
+      "loss": 1.1432,
+      "step": 3069
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.079394738544831e-06,
+      "loss": 1.1026,
+      "step": 3070
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.076524049136484e-06,
+      "loss": 1.0712,
+      "step": 3071
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.073653674853768e-06,
+      "loss": 1.1993,
+      "step": 3072
+    },
+    {
+      "epoch": 0.56,
+      "eval_loss": 1.9350656270980835,
+      "eval_runtime": 0.1838,
+      "eval_samples_per_second": 16.324,
+      "eval_steps_per_second": 16.324,
+      "step": 3072
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0707836166761634e-06,
+      "loss": 1.067,
+      "step": 3073
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.0679138755830484e-06,
+      "loss": 1.1902,
+      "step": 3074
+    },
+    {
+      "epoch": 0.56,
+      "learning_rate": 4.065044452553687e-06,
+      "loss": 1.0825,
+      "step": 3075
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.062175348567236e-06,
+      "loss": 1.1069,
+      "step": 3076
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.059306564602746e-06,
+      "loss": 1.0431,
+      "step": 3077
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0564381016391565e-06,
+      "loss": 1.1697,
+      "step": 3078
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.053569960655297e-06,
+      "loss": 1.1766,
+      "step": 3079
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.05070214262989e-06,
+      "loss": 1.1635,
+      "step": 3080
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.047834648541543e-06,
+      "loss": 1.1189,
+      "step": 3081
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.044967479368756e-06,
+      "loss": 1.1026,
+      "step": 3082
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.042100636089921e-06,
+      "loss": 1.1795,
+      "step": 3083
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.039234119683312e-06,
+      "loss": 1.2132,
+      "step": 3084
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0363679311270954e-06,
+      "loss": 1.0672,
+      "step": 3085
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.033502071399326e-06,
+      "loss": 1.311,
+      "step": 3086
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.030636541477947e-06,
+      "loss": 1.1977,
+      "step": 3087
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.027771342340787e-06,
+      "loss": 1.1151,
+      "step": 3088
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0249064749655606e-06,
+      "loss": 1.1016,
+      "step": 3089
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.022041940329874e-06,
+      "loss": 1.209,
+      "step": 3090
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.019177739411215e-06,
+      "loss": 0.9797,
+      "step": 3091
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.01631387318696e-06,
+      "loss": 1.1062,
+      "step": 3092
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.013450342634371e-06,
+      "loss": 1.0497,
+      "step": 3093
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.0105871487305945e-06,
+      "loss": 1.1916,
+      "step": 3094
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.007724292452663e-06,
+      "loss": 1.1414,
+      "step": 3095
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.004861774777494e-06,
+      "loss": 1.1156,
+      "step": 3096
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 4.001999596681889e-06,
+      "loss": 1.1491,
+      "step": 3097
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.999137759142533e-06,
+      "loss": 1.0831,
+      "step": 3098
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9962762631359955e-06,
+      "loss": 1.1048,
+      "step": 3099
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.99341510963873e-06,
+      "loss": 1.2516,
+      "step": 3100
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.990554299627071e-06,
+      "loss": 1.1156,
+      "step": 3101
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.987693834077239e-06,
+      "loss": 1.1641,
+      "step": 3102
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.984833713965334e-06,
+      "loss": 0.9911,
+      "step": 3103
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.981973940267341e-06,
+      "loss": 1.0278,
+      "step": 3104
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.979114513959123e-06,
+      "loss": 1.108,
+      "step": 3105
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.976255436016427e-06,
+      "loss": 1.1325,
+      "step": 3106
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.973396707414883e-06,
+      "loss": 1.0474,
+      "step": 3107
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.970538329129997e-06,
+      "loss": 0.9819,
+      "step": 3108
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9676803021371576e-06,
+      "loss": 1.2116,
+      "step": 3109
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9648226274116366e-06,
+      "loss": 1.1337,
+      "step": 3110
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.961965305928581e-06,
+      "loss": 1.019,
+      "step": 3111
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.959108338663019e-06,
+      "loss": 0.9579,
+      "step": 3112
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.95625172658986e-06,
+      "loss": 1.1266,
+      "step": 3113
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.953395470683888e-06,
+      "loss": 1.1128,
+      "step": 3114
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.950539571919768e-06,
+      "loss": 1.1928,
+      "step": 3115
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.9476840312720455e-06,
+      "loss": 1.2178,
+      "step": 3116
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.944828849715137e-06,
+      "loss": 1.1228,
+      "step": 3117
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.941974028223344e-06,
+      "loss": 1.1258,
+      "step": 3118
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.939119567770839e-06,
+      "loss": 1.1472,
+      "step": 3119
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.936265469331678e-06,
+      "loss": 1.0744,
+      "step": 3120
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.933411733879785e-06,
+      "loss": 1.2122,
+      "step": 3121
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.930558362388967e-06,
+      "loss": 1.0914,
+      "step": 3122
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.927705355832905e-06,
+      "loss": 1.0804,
+      "step": 3123
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.924852715185155e-06,
+      "loss": 1.1361,
+      "step": 3124
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.922000441419146e-06,
+      "loss": 1.0814,
+      "step": 3125
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.919148535508185e-06,
+      "loss": 1.1671,
+      "step": 3126
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.916296998425453e-06,
+      "loss": 1.0932,
+      "step": 3127
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.913445831144004e-06,
+      "loss": 1.1282,
+      "step": 3128
+    },
+    {
+      "epoch": 0.57,
+      "learning_rate": 3.910595034636765e-06,
+      "loss": 1.1964,
+      "step": 3129
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.907744609876541e-06,
+      "loss": 1.1442,
+      "step": 3130
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.904894557836004e-06,
+      "loss": 1.174,
+      "step": 3131
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.902044879487702e-06,
+      "loss": 1.1552,
+      "step": 3132
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8991955758040575e-06,
+      "loss": 1.2704,
+      "step": 3133
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.89634664775736e-06,
+      "loss": 0.9649,
+      "step": 3134
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.893498096319774e-06,
+      "loss": 1.2727,
+      "step": 3135
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.890649922463339e-06,
+      "loss": 1.0796,
+      "step": 3136
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.887802127159956e-06,
+      "loss": 1.0963,
+      "step": 3137
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.884954711381408e-06,
+      "loss": 1.2024,
+      "step": 3138
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.882107676099338e-06,
+      "loss": 1.0304,
+      "step": 3139
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8792610222852685e-06,
+      "loss": 1.1263,
+      "step": 3140
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.876414750910588e-06,
+      "loss": 1.022,
+      "step": 3141
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.87356886294655e-06,
+      "loss": 1.1795,
+      "step": 3142
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.870723359364284e-06,
+      "loss": 1.0338,
+      "step": 3143
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.867878241134786e-06,
+      "loss": 1.1437,
+      "step": 3144
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.865033509228919e-06,
+      "loss": 1.0225,
+      "step": 3145
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.862189164617415e-06,
+      "loss": 1.0518,
+      "step": 3146
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.859345208270876e-06,
+      "loss": 1.0711,
+      "step": 3147
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.856501641159768e-06,
+      "loss": 1.1164,
+      "step": 3148
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.853658464254425e-06,
+      "loss": 1.1248,
+      "step": 3149
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.850815678525052e-06,
+      "loss": 1.1367,
+      "step": 3150
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.847973284941714e-06,
+      "loss": 1.0701,
+      "step": 3151
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.845131284474346e-06,
+      "loss": 1.0795,
+      "step": 3152
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.842289678092751e-06,
+      "loss": 1.2572,
+      "step": 3153
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.83944846676659e-06,
+      "loss": 1.202,
+      "step": 3154
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.836607651465397e-06,
+      "loss": 1.1694,
+      "step": 3155
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8337672331585685e-06,
+      "loss": 1.1567,
+      "step": 3156
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.830927212815361e-06,
+      "loss": 1.1634,
+      "step": 3157
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.828087591404904e-06,
+      "loss": 1.0603,
+      "step": 3158
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8252483698961836e-06,
+      "loss": 1.1562,
+      "step": 3159
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.822409549258049e-06,
+      "loss": 1.1206,
+      "step": 3160
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8195711304592205e-06,
+      "loss": 1.0671,
+      "step": 3161
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.816733114468273e-06,
+      "loss": 1.1576,
+      "step": 3162
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8138955022536474e-06,
+      "loss": 1.0401,
+      "step": 3163
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.811058294783648e-06,
+      "loss": 1.104,
+      "step": 3164
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8082214930264373e-06,
+      "loss": 1.2153,
+      "step": 3165
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.8053850979500415e-06,
+      "loss": 1.1496,
+      "step": 3166
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.802549110522351e-06,
+      "loss": 1.0962,
+      "step": 3167
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.799713531711111e-06,
+      "loss": 1.1512,
+      "step": 3168
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7968783624839316e-06,
+      "loss": 1.0629,
+      "step": 3169
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7940436038082828e-06,
+      "loss": 0.9858,
+      "step": 3170
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7912092566514923e-06,
+      "loss": 0.9762,
+      "step": 3171
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7883753219807483e-06,
+      "loss": 1.0697,
+      "step": 3172
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7855418007631017e-06,
+      "loss": 1.1389,
+      "step": 3173
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7827086939654557e-06,
+      "loss": 1.2036,
+      "step": 3174
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7798760025545766e-06,
+      "loss": 1.0887,
+      "step": 3175
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7770437274970905e-06,
+      "loss": 1.2527,
+      "step": 3176
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7742118697594743e-06,
+      "loss": 1.1905,
+      "step": 3177
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.771380430308072e-06,
+      "loss": 1.0469,
+      "step": 3178
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7685494101090782e-06,
+      "loss": 1.2419,
+      "step": 3179
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7657188101285437e-06,
+      "loss": 1.1079,
+      "step": 3180
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.762888631332382e-06,
+      "loss": 1.1375,
+      "step": 3181
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7600588746863587e-06,
+      "loss": 1.0859,
+      "step": 3182
+    },
+    {
+      "epoch": 0.58,
+      "learning_rate": 3.7572295411560926e-06,
+      "loss": 1.0597,
+      "step": 3183
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.754400631707067e-06,
+      "loss": 1.1033,
+      "step": 3184
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.75157214730461e-06,
+      "loss": 1.1635,
+      "step": 3185
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7487440889139126e-06,
+      "loss": 1.1064,
+      "step": 3186
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7459164575000174e-06,
+      "loss": 1.1273,
+      "step": 3187
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7430892540278187e-06,
+      "loss": 1.3145,
+      "step": 3188
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.74026247946207e-06,
+      "loss": 1.1538,
+      "step": 3189
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.737436134767376e-06,
+      "loss": 1.0939,
+      "step": 3190
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7346102209081925e-06,
+      "loss": 0.9945,
+      "step": 3191
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.731784738848831e-06,
+      "loss": 1.0738,
+      "step": 3192
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7289596895534572e-06,
+      "loss": 1.0436,
+      "step": 3193
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.726135073986083e-06,
+      "loss": 1.2036,
+      "step": 3194
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7233108931105806e-06,
+      "loss": 1.1346,
+      "step": 3195
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7204871478906673e-06,
+      "loss": 1.0049,
+      "step": 3196
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.717663839289912e-06,
+      "loss": 1.0098,
+      "step": 3197
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.714840968271741e-06,
+      "loss": 1.1924,
+      "step": 3198
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.712018535799423e-06,
+      "loss": 1.1374,
+      "step": 3199
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.709196542836082e-06,
+      "loss": 1.223,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "eval_loss": 1.9349209070205688,
+      "eval_runtime": 0.1708,
+      "eval_samples_per_second": 17.56,
+      "eval_steps_per_second": 17.56,
+      "step": 3200
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7063749903446923e-06,
+      "loss": 1.2219,
+      "step": 3201
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7035538792880756e-06,
+      "loss": 1.2218,
+      "step": 3202
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.7007332106289008e-06,
+      "loss": 1.2104,
+      "step": 3203
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6979129853296946e-06,
+      "loss": 1.0688,
+      "step": 3204
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6950932043528226e-06,
+      "loss": 1.0827,
+      "step": 3205
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6922738686605043e-06,
+      "loss": 1.3436,
+      "step": 3206
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.689454979214806e-06,
+      "loss": 1.4697,
+      "step": 3207
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6866365369776403e-06,
+      "loss": 1.0688,
+      "step": 3208
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.68381854291077e-06,
+      "loss": 1.1007,
+      "step": 3209
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.681000997975803e-06,
+      "loss": 1.1823,
+      "step": 3210
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6781839031341916e-06,
+      "loss": 1.1524,
+      "step": 3211
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6753672593472428e-06,
+      "loss": 1.0726,
+      "step": 3212
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6725510675761013e-06,
+      "loss": 1.1982,
+      "step": 3213
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6697353287817572e-06,
+      "loss": 1.2695,
+      "step": 3214
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.666920043925055e-06,
+      "loss": 1.2775,
+      "step": 3215
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6641052139666755e-06,
+      "loss": 1.1034,
+      "step": 3216
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6612908398671452e-06,
+      "loss": 1.158,
+      "step": 3217
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6584769225868424e-06,
+      "loss": 1.232,
+      "step": 3218
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6556634630859803e-06,
+      "loss": 1.208,
+      "step": 3219
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6528504623246197e-06,
+      "loss": 1.1697,
+      "step": 3220
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.650037921262668e-06,
+      "loss": 1.0771,
+      "step": 3221
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6472258408598716e-06,
+      "loss": 0.9873,
+      "step": 3222
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.644414222075817e-06,
+      "loss": 1.0951,
+      "step": 3223
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.641603065869943e-06,
+      "loss": 1.0589,
+      "step": 3224
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.638792373201522e-06,
+      "loss": 1.2102,
+      "step": 3225
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6359821450296685e-06,
+      "loss": 0.9911,
+      "step": 3226
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6331723823133448e-06,
+      "loss": 1.039,
+      "step": 3227
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6303630860113483e-06,
+      "loss": 1.1692,
+      "step": 3228
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6275542570823187e-06,
+      "loss": 1.1628,
+      "step": 3229
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.624745896484738e-06,
+      "loss": 1.0544,
+      "step": 3230
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.621938005176926e-06,
+      "loss": 1.1014,
+      "step": 3231
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.619130584117046e-06,
+      "loss": 1.1675,
+      "step": 3232
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.616323634263097e-06,
+      "loss": 1.182,
+      "step": 3233
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.613517156572916e-06,
+      "loss": 1.023,
+      "step": 3234
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.610711152004186e-06,
+      "loss": 1.1224,
+      "step": 3235
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.6079056215144228e-06,
+      "loss": 1.111,
+      "step": 3236
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.605100566060977e-06,
+      "loss": 1.0443,
+      "step": 3237
+    },
+    {
+      "epoch": 0.59,
+      "learning_rate": 3.602295986601049e-06,
+      "loss": 1.0021,
+      "step": 3238
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.599491884091666e-06,
+      "loss": 1.0274,
+      "step": 3239
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5966882594896936e-06,
+      "loss": 1.1742,
+      "step": 3240
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5938851137518414e-06,
+      "loss": 1.1915,
+      "step": 3241
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5910824478346482e-06,
+      "loss": 1.057,
+      "step": 3242
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5882802626944914e-06,
+      "loss": 0.9058,
+      "step": 3243
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.585478559287587e-06,
+      "loss": 1.1916,
+      "step": 3244
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5826773385699826e-06,
+      "loss": 1.2113,
+      "step": 3245
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.579876601497561e-06,
+      "loss": 1.0311,
+      "step": 3246
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5770763490260464e-06,
+      "loss": 1.2561,
+      "step": 3247
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5742765821109903e-06,
+      "loss": 1.1526,
+      "step": 3248
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5714773017077788e-06,
+      "loss": 1.0893,
+      "step": 3249
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.568678508771639e-06,
+      "loss": 1.0683,
+      "step": 3250
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5658802042576236e-06,
+      "loss": 0.9336,
+      "step": 3251
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5630823891206257e-06,
+      "loss": 1.0538,
+      "step": 3252
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.560285064315365e-06,
+      "loss": 1.0821,
+      "step": 3253
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.557488230796395e-06,
+      "loss": 1.0337,
+      "step": 3254
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5546918895181078e-06,
+      "loss": 1.3288,
+      "step": 3255
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5518960414347204e-06,
+      "loss": 1.0228,
+      "step": 3256
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5491006875002827e-06,
+      "loss": 1.0765,
+      "step": 3257
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.54630582866868e-06,
+      "loss": 1.2661,
+      "step": 3258
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.543511465893626e-06,
+      "loss": 1.0791,
+      "step": 3259
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5407176001286615e-06,
+      "loss": 1.3044,
+      "step": 3260
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5379242323271655e-06,
+      "loss": 1.134,
+      "step": 3261
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5351313634423424e-06,
+      "loss": 1.1387,
+      "step": 3262
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5323389944272225e-06,
+      "loss": 1.0587,
+      "step": 3263
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5295471262346746e-06,
+      "loss": 1.0939,
+      "step": 3264
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5267557598173914e-06,
+      "loss": 1.0794,
+      "step": 3265
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.523964896127891e-06,
+      "loss": 1.2538,
+      "step": 3266
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.521174536118529e-06,
+      "loss": 1.041,
+      "step": 3267
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.518384680741481e-06,
+      "loss": 1.1413,
+      "step": 3268
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.515595330948751e-06,
+      "loss": 1.2056,
+      "step": 3269
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.512806487692179e-06,
+      "loss": 1.1516,
+      "step": 3270
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5100181519234194e-06,
+      "loss": 1.0036,
+      "step": 3271
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.5072303245939653e-06,
+      "loss": 1.1098,
+      "step": 3272
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.504443006655128e-06,
+      "loss": 1.0079,
+      "step": 3273
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.501656199058048e-06,
+      "loss": 1.1295,
+      "step": 3274
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4988699027536933e-06,
+      "loss": 1.1148,
+      "step": 3275
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4960841186928544e-06,
+      "loss": 1.1613,
+      "step": 3276
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.493298847826147e-06,
+      "loss": 1.376,
+      "step": 3277
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4905140911040163e-06,
+      "loss": 1.1579,
+      "step": 3278
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.487729849476727e-06,
+      "loss": 1.0897,
+      "step": 3279
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4849461238943677e-06,
+      "loss": 1.1571,
+      "step": 3280
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.482162915306857e-06,
+      "loss": 1.3265,
+      "step": 3281
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.479380224663932e-06,
+      "loss": 1.0484,
+      "step": 3282
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.47659805291515e-06,
+      "loss": 1.2424,
+      "step": 3283
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4738164010099007e-06,
+      "loss": 1.1303,
+      "step": 3284
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4710352698973897e-06,
+      "loss": 1.1804,
+      "step": 3285
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4682546605266433e-06,
+      "loss": 1.1232,
+      "step": 3286
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4654745738465168e-06,
+      "loss": 1.1294,
+      "step": 3287
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.462695010805681e-06,
+      "loss": 1.1412,
+      "step": 3288
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4599159723526287e-06,
+      "loss": 1.1573,
+      "step": 3289
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.457137459435679e-06,
+      "loss": 1.1109,
+      "step": 3290
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.4543594730029652e-06,
+      "loss": 1.1882,
+      "step": 3291
+    },
+    {
+      "epoch": 0.6,
+      "learning_rate": 3.451582014002441e-06,
+      "loss": 1.0084,
+      "step": 3292
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4488050833818876e-06,
+      "loss": 1.0473,
+      "step": 3293
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.446028682088896e-06,
+      "loss": 1.1715,
+      "step": 3294
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4432528110708862e-06,
+      "loss": 1.0352,
+      "step": 3295
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.440477471275091e-06,
+      "loss": 0.9368,
+      "step": 3296
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.43770266364856e-06,
+      "loss": 1.2348,
+      "step": 3297
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4349283891381696e-06,
+      "loss": 0.9617,
+      "step": 3298
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4321546486906073e-06,
+      "loss": 1.1407,
+      "step": 3299
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4293814432523776e-06,
+      "loss": 1.093,
+      "step": 3300
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4266087737698105e-06,
+      "loss": 1.1009,
+      "step": 3301
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4238366411890446e-06,
+      "loss": 1.0395,
+      "step": 3302
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.421065046456037e-06,
+      "loss": 1.2463,
+      "step": 3303
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.418293990516567e-06,
+      "loss": 1.0549,
+      "step": 3304
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4155234743162236e-06,
+      "loss": 0.9537,
+      "step": 3305
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.412753498800411e-06,
+      "loss": 1.212,
+      "step": 3306
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4099840649143574e-06,
+      "loss": 1.1632,
+      "step": 3307
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4072151736030977e-06,
+      "loss": 1.0425,
+      "step": 3308
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.4044468258114816e-06,
+      "loss": 1.1155,
+      "step": 3309
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.401679022484182e-06,
+      "loss": 1.1574,
+      "step": 3310
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3989117645656766e-06,
+      "loss": 1.1582,
+      "step": 3311
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3961450530002594e-06,
+      "loss": 1.1746,
+      "step": 3312
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3933788887320428e-06,
+      "loss": 1.0833,
+      "step": 3313
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.390613272704948e-06,
+      "loss": 1.0131,
+      "step": 3314
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.387848205862706e-06,
+      "loss": 1.1491,
+      "step": 3315
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.385083689148869e-06,
+      "loss": 1.1391,
+      "step": 3316
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.382319723506793e-06,
+      "loss": 1.0217,
+      "step": 3317
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.379556309879654e-06,
+      "loss": 1.0604,
+      "step": 3318
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3767934492104333e-06,
+      "loss": 1.0938,
+      "step": 3319
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3740311424419224e-06,
+      "loss": 1.1521,
+      "step": 3320
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.371269390516733e-06,
+      "loss": 1.0821,
+      "step": 3321
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.368508194377278e-06,
+      "loss": 1.1154,
+      "step": 3322
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.365747554965782e-06,
+      "loss": 1.1738,
+      "step": 3323
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3629874732242863e-06,
+      "loss": 1.0648,
+      "step": 3324
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.360227950094635e-06,
+      "loss": 1.0427,
+      "step": 3325
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3574689865184816e-06,
+      "loss": 1.0447,
+      "step": 3326
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3547105834372962e-06,
+      "loss": 1.2292,
+      "step": 3327
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.351952741792349e-06,
+      "loss": 1.1958,
+      "step": 3328
+    },
+    {
+      "epoch": 0.61,
+      "eval_loss": 1.931814193725586,
+      "eval_runtime": 0.1839,
+      "eval_samples_per_second": 16.317,
+      "eval_steps_per_second": 16.317,
+      "step": 3328
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3491954625247207e-06,
+      "loss": 1.0824,
+      "step": 3329
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3464387465753056e-06,
+      "loss": 1.1911,
+      "step": 3330
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3436825948847984e-06,
+      "loss": 1.1211,
+      "step": 3331
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.340927008393703e-06,
+      "loss": 1.1943,
+      "step": 3332
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3381719880423365e-06,
+      "loss": 1.2754,
+      "step": 3333
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.335417534770815e-06,
+      "loss": 0.9216,
+      "step": 3334
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3326636495190635e-06,
+      "loss": 1.3278,
+      "step": 3335
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3299103332268174e-06,
+      "loss": 0.9834,
+      "step": 3336
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3271575868336127e-06,
+      "loss": 1.2111,
+      "step": 3337
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.324405411278789e-06,
+      "loss": 1.0446,
+      "step": 3338
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3216538075015012e-06,
+      "loss": 1.1167,
+      "step": 3339
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3189027764406973e-06,
+      "loss": 1.2984,
+      "step": 3340
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3161523190351396e-06,
+      "loss": 1.1414,
+      "step": 3341
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3134024362233887e-06,
+      "loss": 1.0954,
+      "step": 3342
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3106531289438083e-06,
+      "loss": 1.2469,
+      "step": 3343
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3079043981345717e-06,
+      "loss": 1.1154,
+      "step": 3344
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3051562447336506e-06,
+      "loss": 1.1047,
+      "step": 3345
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.3024086696788193e-06,
+      "loss": 1.0935,
+      "step": 3346
+    },
+    {
+      "epoch": 0.61,
+      "learning_rate": 3.299661673907659e-06,
+      "loss": 1.2663,
+      "step": 3347
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.29691525835755e-06,
+      "loss": 1.3761,
+      "step": 3348
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2941694239656733e-06,
+      "loss": 1.0545,
+      "step": 3349
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.291424171669017e-06,
+      "loss": 0.9778,
+      "step": 3350
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.288679502404365e-06,
+      "loss": 0.972,
+      "step": 3351
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2859354171083023e-06,
+      "loss": 1.0228,
+      "step": 3352
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2831919167172214e-06,
+      "loss": 1.094,
+      "step": 3353
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2804490021673085e-06,
+      "loss": 1.0894,
+      "step": 3354
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2777066743945486e-06,
+      "loss": 1.1669,
+      "step": 3355
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2749649343347356e-06,
+      "loss": 0.9927,
+      "step": 3356
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2722237829234534e-06,
+      "loss": 1.331,
+      "step": 3357
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.269483221096087e-06,
+      "loss": 1.0776,
+      "step": 3358
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2667432497878267e-06,
+      "loss": 1.0533,
+      "step": 3359
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2640038699336518e-06,
+      "loss": 1.0289,
+      "step": 3360
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2612650824683484e-06,
+      "loss": 1.0591,
+      "step": 3361
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2585268883264943e-06,
+      "loss": 1.1203,
+      "step": 3362
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.255789288442466e-06,
+      "loss": 1.0871,
+      "step": 3363
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2530522837504408e-06,
+      "loss": 1.0144,
+      "step": 3364
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2503158751843895e-06,
+      "loss": 1.1171,
+      "step": 3365
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2475800636780772e-06,
+      "loss": 1.0448,
+      "step": 3366
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2448448501650737e-06,
+      "loss": 1.1446,
+      "step": 3367
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.242110235578736e-06,
+      "loss": 1.0643,
+      "step": 3368
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2393762208522193e-06,
+      "loss": 1.2256,
+      "step": 3369
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2366428069184786e-06,
+      "loss": 1.0775,
+      "step": 3370
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.233909994710257e-06,
+      "loss": 1.0114,
+      "step": 3371
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2311777851600954e-06,
+      "loss": 1.1044,
+      "step": 3372
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.228446179200332e-06,
+      "loss": 1.2202,
+      "step": 3373
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2257151777630947e-06,
+      "loss": 1.1088,
+      "step": 3374
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2229847817803037e-06,
+      "loss": 1.2777,
+      "step": 3375
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2202549921836805e-06,
+      "loss": 1.1763,
+      "step": 3376
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2175258099047324e-06,
+      "loss": 1.183,
+      "step": 3377
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2147972358747592e-06,
+      "loss": 1.194,
+      "step": 3378
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2120692710248597e-06,
+      "loss": 1.0399,
+      "step": 3379
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2093419162859197e-06,
+      "loss": 1.2905,
+      "step": 3380
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.206615172588614e-06,
+      "loss": 1.1155,
+      "step": 3381
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.2038890408634192e-06,
+      "loss": 0.939,
+      "step": 3382
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.201163522040591e-06,
+      "loss": 1.1589,
+      "step": 3383
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1984386170501862e-06,
+      "loss": 1.1985,
+      "step": 3384
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1957143268220453e-06,
+      "loss": 1.1074,
+      "step": 3385
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1929906522857997e-06,
+      "loss": 1.1066,
+      "step": 3386
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.190267594370875e-06,
+      "loss": 0.9897,
+      "step": 3387
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1875451540064833e-06,
+      "loss": 1.0971,
+      "step": 3388
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1848233321216233e-06,
+      "loss": 1.1013,
+      "step": 3389
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1821021296450893e-06,
+      "loss": 1.1918,
+      "step": 3390
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1793815475054602e-06,
+      "loss": 0.9488,
+      "step": 3391
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1766615866311003e-06,
+      "loss": 1.1351,
+      "step": 3392
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1739422479501692e-06,
+      "loss": 1.1167,
+      "step": 3393
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.171223532390608e-06,
+      "loss": 1.0502,
+      "step": 3394
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1685054408801465e-06,
+      "loss": 1.0355,
+      "step": 3395
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.165787974346306e-06,
+      "loss": 1.1,
+      "step": 3396
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1630711337163883e-06,
+      "loss": 1.1105,
+      "step": 3397
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1603549199174827e-06,
+      "loss": 1.0322,
+      "step": 3398
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.15763933387647e-06,
+      "loss": 1.1044,
+      "step": 3399
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.1549243765200117e-06,
+      "loss": 1.1935,
+      "step": 3400
+    },
+    {
+      "epoch": 0.62,
+      "learning_rate": 3.152210048774553e-06,
+      "loss": 1.1341,
+      "step": 3401
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1494963515663314e-06,
+      "loss": 1.0139,
+      "step": 3402
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1467832858213635e-06,
+      "loss": 1.0446,
+      "step": 3403
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.14407085246545e-06,
+      "loss": 1.0823,
+      "step": 3404
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.141359052424182e-06,
+      "loss": 1.093,
+      "step": 3405
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.138647886622926e-06,
+      "loss": 1.2348,
+      "step": 3406
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.13593735598684e-06,
+      "loss": 1.0748,
+      "step": 3407
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1332274614408602e-06,
+      "loss": 1.1047,
+      "step": 3408
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1305182039097053e-06,
+      "loss": 1.1755,
+      "step": 3409
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1278095843178825e-06,
+      "loss": 1.0833,
+      "step": 3410
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1251016035896746e-06,
+      "loss": 1.0902,
+      "step": 3411
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1223942626491484e-06,
+      "loss": 1.0506,
+      "step": 3412
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1196875624201554e-06,
+      "loss": 1.1823,
+      "step": 3413
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.116981503826326e-06,
+      "loss": 0.9569,
+      "step": 3414
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1142760877910684e-06,
+      "loss": 1.0134,
+      "step": 3415
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1115713152375794e-06,
+      "loss": 1.3297,
+      "step": 3416
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1088671870888297e-06,
+      "loss": 1.0053,
+      "step": 3417
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1061637042675714e-06,
+      "loss": 0.9791,
+      "step": 3418
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1034608676963395e-06,
+      "loss": 1.213,
+      "step": 3419
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.1007586782974444e-06,
+      "loss": 1.1629,
+      "step": 3420
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.098057136992976e-06,
+      "loss": 1.1352,
+      "step": 3421
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.095356244704808e-06,
+      "loss": 1.3035,
+      "step": 3422
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.092656002354587e-06,
+      "loss": 1.1877,
+      "step": 3423
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0899564108637387e-06,
+      "loss": 1.0691,
+      "step": 3424
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.087257471153471e-06,
+      "loss": 1.2608,
+      "step": 3425
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0845591841447633e-06,
+      "loss": 1.114,
+      "step": 3426
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.081861550758377e-06,
+      "loss": 1.1266,
+      "step": 3427
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.07916457191485e-06,
+      "loss": 1.1832,
+      "step": 3428
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.076468248534491e-06,
+      "loss": 1.1277,
+      "step": 3429
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.073772581537394e-06,
+      "loss": 1.0568,
+      "step": 3430
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.071077571843424e-06,
+      "loss": 1.0397,
+      "step": 3431
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0683832203722187e-06,
+      "loss": 1.1125,
+      "step": 3432
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0656895280431985e-06,
+      "loss": 1.0045,
+      "step": 3433
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0629964957755555e-06,
+      "loss": 1.1488,
+      "step": 3434
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0603041244882525e-06,
+      "loss": 1.1214,
+      "step": 3435
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.057612415100034e-06,
+      "loss": 1.025,
+      "step": 3436
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.054921368529415e-06,
+      "loss": 0.9753,
+      "step": 3437
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0522309856946808e-06,
+      "loss": 1.0836,
+      "step": 3438
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.049541267513898e-06,
+      "loss": 1.0644,
+      "step": 3439
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0468522149049006e-06,
+      "loss": 1.1354,
+      "step": 3440
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0441638287852958e-06,
+      "loss": 1.0587,
+      "step": 3441
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.041476110072468e-06,
+      "loss": 0.9159,
+      "step": 3442
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0387890596835677e-06,
+      "loss": 1.0054,
+      "step": 3443
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.03610267853552e-06,
+      "loss": 1.1639,
+      "step": 3444
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.033416967545024e-06,
+      "loss": 1.0324,
+      "step": 3445
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0307319276285475e-06,
+      "loss": 1.1248,
+      "step": 3446
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.028047559702326e-06,
+      "loss": 1.1286,
+      "step": 3447
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0253638646823757e-06,
+      "loss": 1.0812,
+      "step": 3448
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.022680843484471e-06,
+      "loss": 1.0491,
+      "step": 3449
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.019998497024167e-06,
+      "loss": 1.1952,
+      "step": 3450
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0173168262167817e-06,
+      "loss": 1.1951,
+      "step": 3451
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0146358319774023e-06,
+      "loss": 1.0494,
+      "step": 3452
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0119555152208924e-06,
+      "loss": 1.1139,
+      "step": 3453
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.0092758768618767e-06,
+      "loss": 1.1023,
+      "step": 3454
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.006596917814749e-06,
+      "loss": 1.2054,
+      "step": 3455
+    },
+    {
+      "epoch": 0.63,
+      "learning_rate": 3.003918638993679e-06,
+      "loss": 1.1548,
+      "step": 3456
+    },
+    {
+      "epoch": 0.63,
+      "eval_loss": 1.9343700408935547,
+      "eval_runtime": 0.1873,
+      "eval_samples_per_second": 16.02,
+      "eval_steps_per_second": 16.02,
+      "step": 3456
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 3.001241041312595e-06,
+      "loss": 1.0619,
+      "step": 3457
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.998564125685196e-06,
+      "loss": 1.1072,
+      "step": 3458
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.995887893024952e-06,
+      "loss": 1.045,
+      "step": 3459
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9932123442450956e-06,
+      "loss": 1.2377,
+      "step": 3460
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.990537480258624e-06,
+      "loss": 1.1181,
+      "step": 3461
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.98786330197831e-06,
+      "loss": 1.1293,
+      "step": 3462
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9851898103166822e-06,
+      "loss": 1.1285,
+      "step": 3463
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.982517006186039e-06,
+      "loss": 1.169,
+      "step": 3464
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9798448904984474e-06,
+      "loss": 1.1656,
+      "step": 3465
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9771734641657346e-06,
+      "loss": 1.1222,
+      "step": 3466
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.974502728099492e-06,
+      "loss": 1.1285,
+      "step": 3467
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.971832683211082e-06,
+      "loss": 1.2048,
+      "step": 3468
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.969163330411626e-06,
+      "loss": 1.1262,
+      "step": 3469
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.966494670612006e-06,
+      "loss": 1.0589,
+      "step": 3470
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9638267047228776e-06,
+      "loss": 1.0992,
+      "step": 3471
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.961159433654649e-06,
+      "loss": 1.1168,
+      "step": 3472
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9584928583175e-06,
+      "loss": 1.0452,
+      "step": 3473
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.955826979621367e-06,
+      "loss": 1.2709,
+      "step": 3474
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.953161798475949e-06,
+      "loss": 1.1413,
+      "step": 3475
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9504973157907125e-06,
+      "loss": 1.065,
+      "step": 3476
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9478335324748798e-06,
+      "loss": 1.0584,
+      "step": 3477
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9451704494374333e-06,
+      "loss": 1.2678,
+      "step": 3478
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9425080675871263e-06,
+      "loss": 1.0399,
+      "step": 3479
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9398463878324623e-06,
+      "loss": 1.1498,
+      "step": 3480
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.937185411081708e-06,
+      "loss": 1.1086,
+      "step": 3481
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9345251382428943e-06,
+      "loss": 1.2057,
+      "step": 3482
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9318655702238074e-06,
+      "loss": 1.2304,
+      "step": 3483
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9292067079319935e-06,
+      "loss": 1.1165,
+      "step": 3484
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.926548552274762e-06,
+      "loss": 1.2284,
+      "step": 3485
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9238911041591765e-06,
+      "loss": 1.1835,
+      "step": 3486
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9212343644920595e-06,
+      "loss": 0.9956,
+      "step": 3487
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9185783341799944e-06,
+      "loss": 1.0418,
+      "step": 3488
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9159230141293226e-06,
+      "loss": 1.0211,
+      "step": 3489
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.913268405246139e-06,
+      "loss": 1.0983,
+      "step": 3490
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9106145084363014e-06,
+      "loss": 1.1365,
+      "step": 3491
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9079613246054204e-06,
+      "loss": 1.1022,
+      "step": 3492
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9053088546588626e-06,
+      "loss": 0.9836,
+      "step": 3493
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.902657099501758e-06,
+      "loss": 1.1468,
+      "step": 3494
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.9000060600389844e-06,
+      "loss": 1.2191,
+      "step": 3495
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.897355737175177e-06,
+      "loss": 0.9589,
+      "step": 3496
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8947061318147334e-06,
+      "loss": 1.1138,
+      "step": 3497
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8920572448617986e-06,
+      "loss": 1.0409,
+      "step": 3498
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8894090772202722e-06,
+      "loss": 1.0187,
+      "step": 3499
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.886761629793816e-06,
+      "loss": 1.1514,
+      "step": 3500
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.884114903485839e-06,
+      "loss": 1.2893,
+      "step": 3501
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8814688991995044e-06,
+      "loss": 1.1226,
+      "step": 3502
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.878823617837735e-06,
+      "loss": 1.1328,
+      "step": 3503
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8761790603032013e-06,
+      "loss": 1.188,
+      "step": 3504
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8735352274983252e-06,
+      "loss": 1.0708,
+      "step": 3505
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.87089212032529e-06,
+      "loss": 1.2532,
+      "step": 3506
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8682497396860205e-06,
+      "loss": 1.036,
+      "step": 3507
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.865608086482203e-06,
+      "loss": 0.9988,
+      "step": 3508
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8629671616152718e-06,
+      "loss": 0.931,
+      "step": 3509
+    },
+    {
+      "epoch": 0.64,
+      "learning_rate": 2.8603269659864074e-06,
+      "loss": 1.1643,
+      "step": 3510
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8576875004965516e-06,
+      "loss": 1.1508,
+      "step": 3511
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8550487660463897e-06,
+      "loss": 1.1167,
+      "step": 3512
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8524107635363584e-06,
+      "loss": 1.1009,
+      "step": 3513
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.84977349386665e-06,
+      "loss": 1.0413,
+      "step": 3514
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8471369579371995e-06,
+      "loss": 1.1335,
+      "step": 3515
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.844501156647693e-06,
+      "loss": 1.2425,
+      "step": 3516
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8418660908975727e-06,
+      "loss": 0.996,
+      "step": 3517
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.839231761586022e-06,
+      "loss": 0.996,
+      "step": 3518
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.836598169611974e-06,
+      "loss": 1.2061,
+      "step": 3519
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8339653158741142e-06,
+      "loss": 1.1195,
+      "step": 3520
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.831333201270875e-06,
+      "loss": 1.1752,
+      "step": 3521
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8287018267004308e-06,
+      "loss": 1.0656,
+      "step": 3522
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.826071193060713e-06,
+      "loss": 1.0997,
+      "step": 3523
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.823441301249392e-06,
+      "loss": 1.096,
+      "step": 3524
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.820812152163892e-06,
+      "loss": 1.0101,
+      "step": 3525
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8181837467013773e-06,
+      "loss": 1.0891,
+      "step": 3526
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8155560857587603e-06,
+      "loss": 1.2096,
+      "step": 3527
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.812929170232703e-06,
+      "loss": 1.0026,
+      "step": 3528
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8103030010196107e-06,
+      "loss": 1.1382,
+      "step": 3529
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.8076775790156298e-06,
+      "loss": 1.0621,
+      "step": 3530
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.80505290511666e-06,
+      "loss": 1.0197,
+      "step": 3531
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.802428980218339e-06,
+      "loss": 1.011,
+      "step": 3532
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7998058052160498e-06,
+      "loss": 1.2477,
+      "step": 3533
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.797183381004925e-06,
+      "loss": 1.2667,
+      "step": 3534
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7945617084798343e-06,
+      "loss": 1.0407,
+      "step": 3535
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7919407885353922e-06,
+      "loss": 1.1543,
+      "step": 3536
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.789320622065961e-06,
+      "loss": 1.2439,
+      "step": 3537
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7867012099656417e-06,
+      "loss": 1.066,
+      "step": 3538
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7840825531282754e-06,
+      "loss": 1.0713,
+      "step": 3539
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.781464652447454e-06,
+      "loss": 1.1687,
+      "step": 3540
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.778847508816504e-06,
+      "loss": 1.0582,
+      "step": 3541
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7762311231284933e-06,
+      "loss": 1.0237,
+      "step": 3542
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7736154962762385e-06,
+      "loss": 1.2372,
+      "step": 3543
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7710006291522875e-06,
+      "loss": 1.0914,
+      "step": 3544
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7683865226489383e-06,
+      "loss": 1.0576,
+      "step": 3545
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.765773177658222e-06,
+      "loss": 1.1984,
+      "step": 3546
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7631605950719126e-06,
+      "loss": 1.052,
+      "step": 3547
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7605487757815273e-06,
+      "loss": 1.089,
+      "step": 3548
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.757937720678317e-06,
+      "loss": 1.1181,
+      "step": 3549
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7553274306532728e-06,
+      "loss": 1.1245,
+      "step": 3550
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.752717906597131e-06,
+      "loss": 1.0316,
+      "step": 3551
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7501091494003592e-06,
+      "loss": 1.0709,
+      "step": 3552
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.747501159953164e-06,
+      "loss": 1.1816,
+      "step": 3553
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7448939391454964e-06,
+      "loss": 1.2019,
+      "step": 3554
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7422874878670386e-06,
+      "loss": 1.1016,
+      "step": 3555
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7396818070072108e-06,
+      "loss": 1.1391,
+      "step": 3556
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7370768974551747e-06,
+      "loss": 1.0039,
+      "step": 3557
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7344727600998255e-06,
+      "loss": 1.1884,
+      "step": 3558
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7318693958297917e-06,
+      "loss": 1.0286,
+      "step": 3559
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.729266805533447e-06,
+      "loss": 1.1295,
+      "step": 3560
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.726664990098891e-06,
+      "loss": 1.1197,
+      "step": 3561
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7240639504139676e-06,
+      "loss": 1.1968,
+      "step": 3562
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7214636873662505e-06,
+      "loss": 1.0254,
+      "step": 3563
+    },
+    {
+      "epoch": 0.65,
+      "learning_rate": 2.7188642018430468e-06,
+      "loss": 0.9769,
+      "step": 3564
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.716265494731406e-06,
+      "loss": 1.3527,
+      "step": 3565
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7136675669181043e-06,
+      "loss": 1.0744,
+      "step": 3566
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.711070419289654e-06,
+      "loss": 1.0807,
+      "step": 3567
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.708474052732306e-06,
+      "loss": 1.0471,
+      "step": 3568
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.705878468132038e-06,
+      "loss": 1.0889,
+      "step": 3569
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.7032836663745633e-06,
+      "loss": 1.1386,
+      "step": 3570
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.70068964834533e-06,
+      "loss": 1.0912,
+      "step": 3571
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6980964149295166e-06,
+      "loss": 1.0958,
+      "step": 3572
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.695503967012032e-06,
+      "loss": 0.9948,
+      "step": 3573
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6929123054775234e-06,
+      "loss": 1.0654,
+      "step": 3574
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6903214312103643e-06,
+      "loss": 1.2633,
+      "step": 3575
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6877313450946587e-06,
+      "loss": 1.1792,
+      "step": 3576
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6851420480142475e-06,
+      "loss": 1.2412,
+      "step": 3577
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6825535408526983e-06,
+      "loss": 1.1555,
+      "step": 3578
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6799658244933047e-06,
+      "loss": 1.073,
+      "step": 3579
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.677378899819102e-06,
+      "loss": 1.1488,
+      "step": 3580
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6747927677128438e-06,
+      "loss": 1.096,
+      "step": 3581
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.672207429057023e-06,
+      "loss": 1.1291,
+      "step": 3582
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.669622884733853e-06,
+      "loss": 1.13,
+      "step": 3583
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6670391356252795e-06,
+      "loss": 1.0292,
+      "step": 3584
+    },
+    {
+      "epoch": 0.66,
+      "eval_loss": 1.9287360906600952,
+      "eval_runtime": 0.1775,
+      "eval_samples_per_second": 16.906,
+      "eval_steps_per_second": 16.906,
+      "step": 3584
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.664456182612981e-06,
+      "loss": 1.1606,
+      "step": 3585
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6618740265783595e-06,
+      "loss": 1.2017,
+      "step": 3586
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6592926684025433e-06,
+      "loss": 1.2561,
+      "step": 3587
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6567121089663953e-06,
+      "loss": 1.1578,
+      "step": 3588
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6541323491505013e-06,
+      "loss": 1.1007,
+      "step": 3589
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.651553389835171e-06,
+      "loss": 1.0171,
+      "step": 3590
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6489752319004502e-06,
+      "loss": 1.1155,
+      "step": 3591
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6463978762261037e-06,
+      "loss": 1.1023,
+      "step": 3592
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6438213236916223e-06,
+      "loss": 1.1438,
+      "step": 3593
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.64124557517623e-06,
+      "loss": 1.0518,
+      "step": 3594
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6386706315588682e-06,
+      "loss": 1.0888,
+      "step": 3595
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6360964937182064e-06,
+      "loss": 1.2381,
+      "step": 3596
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.633523162532643e-06,
+      "loss": 1.1099,
+      "step": 3597
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6309506388802953e-06,
+      "loss": 1.0335,
+      "step": 3598
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.628378923639007e-06,
+      "loss": 1.0314,
+      "step": 3599
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6258080176863505e-06,
+      "loss": 1.152,
+      "step": 3600
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.623237921899615e-06,
+      "loss": 1.2653,
+      "step": 3601
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6206686371558153e-06,
+      "loss": 1.195,
+      "step": 3602
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.618100164331694e-06,
+      "loss": 1.1148,
+      "step": 3603
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6155325043037095e-06,
+      "loss": 1.1382,
+      "step": 3604
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.61296565794805e-06,
+      "loss": 1.1146,
+      "step": 3605
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6103996261406208e-06,
+      "loss": 1.0527,
+      "step": 3606
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6078344097570484e-06,
+      "loss": 1.0269,
+      "step": 3607
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6052700096726892e-06,
+      "loss": 1.2482,
+      "step": 3608
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.6027064267626125e-06,
+      "loss": 1.0119,
+      "step": 3609
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.600143661901609e-06,
+      "loss": 0.9993,
+      "step": 3610
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.597581715964198e-06,
+      "loss": 1.1562,
+      "step": 3611
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5950205898246117e-06,
+      "loss": 1.0197,
+      "step": 3612
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5924602843568043e-06,
+      "loss": 1.1003,
+      "step": 3613
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.589900800434453e-06,
+      "loss": 1.0224,
+      "step": 3614
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.587342138930952e-06,
+      "loss": 1.0,
+      "step": 3615
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.584784300719413e-06,
+      "loss": 1.0265,
+      "step": 3616
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.582227286672672e-06,
+      "loss": 0.955,
+      "step": 3617
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5796710976632803e-06,
+      "loss": 1.0349,
+      "step": 3618
+    },
+    {
+      "epoch": 0.66,
+      "learning_rate": 2.5771157345635056e-06,
+      "loss": 0.9523,
+      "step": 3619
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.574561198245341e-06,
+      "loss": 0.9914,
+      "step": 3620
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5720074895804895e-06,
+      "loss": 1.0384,
+      "step": 3621
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5694546094403742e-06,
+      "loss": 1.0921,
+      "step": 3622
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5669025586961405e-06,
+      "loss": 1.0946,
+      "step": 3623
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5643513382186415e-06,
+      "loss": 1.0398,
+      "step": 3624
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.561800948878457e-06,
+      "loss": 1.1046,
+      "step": 3625
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.559251391545876e-06,
+      "loss": 1.0176,
+      "step": 3626
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5567026670909036e-06,
+      "loss": 1.1343,
+      "step": 3627
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.554154776383267e-06,
+      "loss": 1.1534,
+      "step": 3628
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.551607720292403e-06,
+      "loss": 1.0348,
+      "step": 3629
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5490614996874634e-06,
+      "loss": 1.0184,
+      "step": 3630
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5465161154373208e-06,
+      "loss": 1.0366,
+      "step": 3631
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.543971568410557e-06,
+      "loss": 1.2482,
+      "step": 3632
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.541427859475467e-06,
+      "loss": 1.1121,
+      "step": 3633
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5388849895000667e-06,
+      "loss": 1.0561,
+      "step": 3634
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5363429593520792e-06,
+      "loss": 1.0161,
+      "step": 3635
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5338017698989427e-06,
+      "loss": 1.0554,
+      "step": 3636
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5312614220078107e-06,
+      "loss": 1.0445,
+      "step": 3637
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.528721916545548e-06,
+      "loss": 1.2397,
+      "step": 3638
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5261832543787284e-06,
+      "loss": 1.2989,
+      "step": 3639
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.523645436373646e-06,
+      "loss": 1.1537,
+      "step": 3640
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.521108463396301e-06,
+      "loss": 1.0312,
+      "step": 3641
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.518572336312404e-06,
+      "loss": 0.9116,
+      "step": 3642
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5160370559873825e-06,
+      "loss": 1.1575,
+      "step": 3643
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.51350262328637e-06,
+      "loss": 1.0404,
+      "step": 3644
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5109690390742126e-06,
+      "loss": 1.1339,
+      "step": 3645
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.508436304215469e-06,
+      "loss": 1.1584,
+      "step": 3646
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.5059044195744036e-06,
+      "loss": 1.1279,
+      "step": 3647
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.503373386014996e-06,
+      "loss": 1.1567,
+      "step": 3648
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.500843204400931e-06,
+      "loss": 1.1388,
+      "step": 3649
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.498313875595602e-06,
+      "loss": 1.1913,
+      "step": 3650
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4957854004621177e-06,
+      "loss": 1.1312,
+      "step": 3651
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4932577798632902e-06,
+      "loss": 1.1414,
+      "step": 3652
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4907310146616375e-06,
+      "loss": 1.043,
+      "step": 3653
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4882051057193943e-06,
+      "loss": 0.937,
+      "step": 3654
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.485680053898497e-06,
+      "loss": 1.1474,
+      "step": 3655
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4831558600605876e-06,
+      "loss": 1.0614,
+      "step": 3656
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.480632525067022e-06,
+      "loss": 1.1504,
+      "step": 3657
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.478110049778859e-06,
+      "loss": 1.001,
+      "step": 3658
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4755884350568615e-06,
+      "loss": 1.1819,
+      "step": 3659
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4730676817615064e-06,
+      "loss": 1.1933,
+      "step": 3660
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.470547790752969e-06,
+      "loss": 1.1058,
+      "step": 3661
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4680287628911327e-06,
+      "loss": 1.2072,
+      "step": 3662
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4655105990355893e-06,
+      "loss": 1.0862,
+      "step": 3663
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.462993300045633e-06,
+      "loss": 1.1523,
+      "step": 3664
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.460476866780261e-06,
+      "loss": 1.2469,
+      "step": 3665
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4579613000981807e-06,
+      "loss": 1.211,
+      "step": 3666
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4554466008577995e-06,
+      "loss": 1.0871,
+      "step": 3667
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4529327699172266e-06,
+      "loss": 1.1128,
+      "step": 3668
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4504198081342832e-06,
+      "loss": 1.0807,
+      "step": 3669
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.447907716366485e-06,
+      "loss": 1.2094,
+      "step": 3670
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.445396495471058e-06,
+      "loss": 1.0349,
+      "step": 3671
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.442886146304926e-06,
+      "loss": 1.1381,
+      "step": 3672
+    },
+    {
+      "epoch": 0.67,
+      "learning_rate": 2.4403766697247145e-06,
+      "loss": 1.2932,
+      "step": 3673
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4378680665867583e-06,
+      "loss": 0.9989,
+      "step": 3674
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.435360337747087e-06,
+      "loss": 1.2833,
+      "step": 3675
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.432853484061433e-06,
+      "loss": 1.1363,
+      "step": 3676
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.430347506385236e-06,
+      "loss": 1.1934,
+      "step": 3677
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4278424055736288e-06,
+      "loss": 1.2138,
+      "step": 3678
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.425338182481448e-06,
+      "loss": 1.1813,
+      "step": 3679
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4228348379632344e-06,
+      "loss": 1.0742,
+      "step": 3680
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4203323728732237e-06,
+      "loss": 0.9791,
+      "step": 3681
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4178307880653514e-06,
+      "loss": 1.1392,
+      "step": 3682
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.415330084393259e-06,
+      "loss": 1.1876,
+      "step": 3683
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.412830262710281e-06,
+      "loss": 1.1258,
+      "step": 3684
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.410331323869452e-06,
+      "loss": 1.0982,
+      "step": 3685
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.407833268723509e-06,
+      "loss": 1.2037,
+      "step": 3686
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4053360981248835e-06,
+      "loss": 1.3141,
+      "step": 3687
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.402839812925704e-06,
+      "loss": 1.0278,
+      "step": 3688
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.4003444139778036e-06,
+      "loss": 1.064,
+      "step": 3689
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3978499021327056e-06,
+      "loss": 1.0255,
+      "step": 3690
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.395356278241636e-06,
+      "loss": 1.0496,
+      "step": 3691
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.392863543155515e-06,
+      "loss": 1.0761,
+      "step": 3692
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.390371697724957e-06,
+      "loss": 0.9852,
+      "step": 3693
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.38788074280028e-06,
+      "loss": 1.1232,
+      "step": 3694
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.385390679231492e-06,
+      "loss": 1.1997,
+      "step": 3695
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.382901507868297e-06,
+      "loss": 1.1685,
+      "step": 3696
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3804132295601007e-06,
+      "loss": 1.2486,
+      "step": 3697
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3779258451559966e-06,
+      "loss": 1.0561,
+      "step": 3698
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3754393555047756e-06,
+      "loss": 0.9601,
+      "step": 3699
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3729537614549264e-06,
+      "loss": 1.0104,
+      "step": 3700
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3704690638546293e-06,
+      "loss": 1.3353,
+      "step": 3701
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.367985263551755e-06,
+      "loss": 1.1017,
+      "step": 3702
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3655023613938776e-06,
+      "loss": 1.1274,
+      "step": 3703
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3630203582282568e-06,
+      "loss": 1.144,
+      "step": 3704
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3605392549018457e-06,
+      "loss": 1.1439,
+      "step": 3705
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.358059052261296e-06,
+      "loss": 1.168,
+      "step": 3706
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3555797511529473e-06,
+      "loss": 1.1375,
+      "step": 3707
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3531013524228306e-06,
+      "loss": 0.9913,
+      "step": 3708
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.350623856916675e-06,
+      "loss": 1.1723,
+      "step": 3709
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.348147265479896e-06,
+      "loss": 1.1041,
+      "step": 3710
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3456715789575994e-06,
+      "loss": 1.0639,
+      "step": 3711
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3431967981945896e-06,
+      "loss": 1.0799,
+      "step": 3712
+    },
+    {
+      "epoch": 0.68,
+      "eval_loss": 1.930917739868164,
+      "eval_runtime": 0.1814,
+      "eval_samples_per_second": 16.538,
+      "eval_steps_per_second": 16.538,
+      "step": 3712
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.340722924035353e-06,
+      "loss": 1.1933,
+      "step": 3713
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3382499573240754e-06,
+      "loss": 1.0429,
+      "step": 3714
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.335777898904626e-06,
+      "loss": 1.1055,
+      "step": 3715
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3333067496205643e-06,
+      "loss": 1.0867,
+      "step": 3716
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3308365103151465e-06,
+      "loss": 1.0149,
+      "step": 3717
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.328367181831311e-06,
+      "loss": 1.2845,
+      "step": 3718
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3258987650116866e-06,
+      "loss": 1.1361,
+      "step": 3719
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3234312606985955e-06,
+      "loss": 1.14,
+      "step": 3720
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3209646697340438e-06,
+      "loss": 1.0094,
+      "step": 3721
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.318498992959726e-06,
+      "loss": 1.1642,
+      "step": 3722
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.31603423121703e-06,
+      "loss": 1.1879,
+      "step": 3723
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.313570385347024e-06,
+      "loss": 1.1016,
+      "step": 3724
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3111074561904678e-06,
+      "loss": 1.0702,
+      "step": 3725
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3086454445878103e-06,
+      "loss": 1.1438,
+      "step": 3726
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.3061843513791838e-06,
+      "loss": 1.0032,
+      "step": 3727
+    },
+    {
+      "epoch": 0.68,
+      "learning_rate": 2.303724177404405e-06,
+      "loss": 1.1045,
+      "step": 3728
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.3012649235029853e-06,
+      "loss": 1.1172,
+      "step": 3729
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.298806590514115e-06,
+      "loss": 1.2016,
+      "step": 3730
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.296349179276669e-06,
+      "loss": 1.066,
+      "step": 3731
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2938926906292164e-06,
+      "loss": 1.1057,
+      "step": 3732
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.291437125410003e-06,
+      "loss": 1.0934,
+      "step": 3733
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2889824844569607e-06,
+      "loss": 1.0501,
+      "step": 3734
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2865287686077116e-06,
+      "loss": 0.9894,
+      "step": 3735
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.284075978699554e-06,
+      "loss": 1.0577,
+      "step": 3736
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.281624115569479e-06,
+      "loss": 1.0952,
+      "step": 3737
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.279173180054155e-06,
+      "loss": 1.2744,
+      "step": 3738
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2767231729899333e-06,
+      "loss": 1.0954,
+      "step": 3739
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2742740952128544e-06,
+      "loss": 1.149,
+      "step": 3740
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2718259475586366e-06,
+      "loss": 1.0536,
+      "step": 3741
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2693787308626814e-06,
+      "loss": 1.0535,
+      "step": 3742
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.266932445960076e-06,
+      "loss": 1.0754,
+      "step": 3743
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.264487093685586e-06,
+      "loss": 0.9194,
+      "step": 3744
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.262042674873658e-06,
+      "loss": 1.0692,
+      "step": 3745
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.259599190358426e-06,
+      "loss": 1.0376,
+      "step": 3746
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2571566409736983e-06,
+      "loss": 1.0744,
+      "step": 3747
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2547150275529666e-06,
+      "loss": 1.1212,
+      "step": 3748
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2522743509294064e-06,
+      "loss": 0.9922,
+      "step": 3749
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.249834611935869e-06,
+      "loss": 1.2477,
+      "step": 3750
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.247395811404886e-06,
+      "loss": 1.0256,
+      "step": 3751
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2449579501686737e-06,
+      "loss": 1.0999,
+      "step": 3752
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.242521029059122e-06,
+      "loss": 1.0016,
+      "step": 3753
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.240085048907802e-06,
+      "loss": 1.0554,
+      "step": 3754
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2376500105459664e-06,
+      "loss": 1.0206,
+      "step": 3755
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2352159148045417e-06,
+      "loss": 1.2553,
+      "step": 3756
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.232782762514138e-06,
+      "loss": 1.1172,
+      "step": 3757
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.230350554505039e-06,
+      "loss": 1.1907,
+      "step": 3758
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2279192916072075e-06,
+      "loss": 1.081,
+      "step": 3759
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2254889746502856e-06,
+      "loss": 1.1643,
+      "step": 3760
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2230596044635917e-06,
+      "loss": 1.1048,
+      "step": 3761
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2206311818761174e-06,
+      "loss": 1.0209,
+      "step": 3762
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2182037077165376e-06,
+      "loss": 1.0817,
+      "step": 3763
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2157771828131995e-06,
+      "loss": 1.1658,
+      "step": 3764
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2133516079941246e-06,
+      "loss": 1.1235,
+      "step": 3765
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.210926984087016e-06,
+      "loss": 1.0844,
+      "step": 3766
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.208503311919248e-06,
+      "loss": 1.0171,
+      "step": 3767
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2060805923178684e-06,
+      "loss": 1.2894,
+      "step": 3768
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2036588261096066e-06,
+      "loss": 1.184,
+      "step": 3769
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.2012380141208604e-06,
+      "loss": 1.1651,
+      "step": 3770
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1988181571777034e-06,
+      "loss": 1.2374,
+      "step": 3771
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1963992561058873e-06,
+      "loss": 1.1576,
+      "step": 3772
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1939813117308325e-06,
+      "loss": 1.2836,
+      "step": 3773
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1915643248776325e-06,
+      "loss": 1.1315,
+      "step": 3774
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.189148296371062e-06,
+      "loss": 0.9764,
+      "step": 3775
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1867332270355594e-06,
+      "loss": 1.0331,
+      "step": 3776
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.184319117695239e-06,
+      "loss": 1.1182,
+      "step": 3777
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.181905969173891e-06,
+      "loss": 1.0758,
+      "step": 3778
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1794937822949713e-06,
+      "loss": 1.11,
+      "step": 3779
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.177082557881615e-06,
+      "loss": 1.0183,
+      "step": 3780
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.174672296756623e-06,
+      "loss": 0.9531,
+      "step": 3781
+    },
+    {
+      "epoch": 0.69,
+      "learning_rate": 2.1722629997424672e-06,
+      "loss": 1.1344,
+      "step": 3782
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.169854667661297e-06,
+      "loss": 1.2083,
+      "step": 3783
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1674473013349263e-06,
+      "loss": 0.9729,
+      "step": 3784
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1650409015848395e-06,
+      "loss": 1.1687,
+      "step": 3785
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.162635469232196e-06,
+      "loss": 1.091,
+      "step": 3786
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1602310050978215e-06,
+      "loss": 1.0737,
+      "step": 3787
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1578275100022094e-06,
+      "loss": 1.2258,
+      "step": 3788
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1554249847655286e-06,
+      "loss": 1.14,
+      "step": 3789
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.153023430207612e-06,
+      "loss": 1.1244,
+      "step": 3790
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.15062284714796e-06,
+      "loss": 0.9964,
+      "step": 3791
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.148223236405749e-06,
+      "loss": 1.0484,
+      "step": 3792
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1458245987998167e-06,
+      "loss": 1.168,
+      "step": 3793
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.143426935148668e-06,
+      "loss": 1.039,
+      "step": 3794
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1410302462704828e-06,
+      "loss": 1.031,
+      "step": 3795
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.138634532983102e-06,
+      "loss": 1.2904,
+      "step": 3796
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1362397961040334e-06,
+      "loss": 1.0311,
+      "step": 3797
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.133846036450457e-06,
+      "loss": 0.9741,
+      "step": 3798
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1314532548392147e-06,
+      "loss": 1.184,
+      "step": 3799
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1290614520868137e-06,
+      "loss": 0.9839,
+      "step": 3800
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1266706290094334e-06,
+      "loss": 1.1619,
+      "step": 3801
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.124280786422911e-06,
+      "loss": 1.0609,
+      "step": 3802
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.121891925142756e-06,
+      "loss": 1.159,
+      "step": 3803
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.11950404598414e-06,
+      "loss": 1.1381,
+      "step": 3804
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.117117149761896e-06,
+      "loss": 1.0353,
+      "step": 3805
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.114731237290529e-06,
+      "loss": 1.0893,
+      "step": 3806
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1123463093842034e-06,
+      "loss": 1.0445,
+      "step": 3807
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.109962366856746e-06,
+      "loss": 1.0919,
+      "step": 3808
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.107579410521654e-06,
+      "loss": 0.9398,
+      "step": 3809
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1051974411920826e-06,
+      "loss": 1.0552,
+      "step": 3810
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.1028164596808494e-06,
+      "loss": 1.1242,
+      "step": 3811
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.10043646680044e-06,
+      "loss": 1.0051,
+      "step": 3812
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0980574633629996e-06,
+      "loss": 1.134,
+      "step": 3813
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0956794501803336e-06,
+      "loss": 1.0972,
+      "step": 3814
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.093302428063915e-06,
+      "loss": 1.0953,
+      "step": 3815
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0909263978248734e-06,
+      "loss": 0.8699,
+      "step": 3816
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0885513602740016e-06,
+      "loss": 1.2186,
+      "step": 3817
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0861773162217565e-06,
+      "loss": 1.1835,
+      "step": 3818
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0838042664782516e-06,
+      "loss": 0.9316,
+      "step": 3819
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.081432211853262e-06,
+      "loss": 1.0647,
+      "step": 3820
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0790611531562277e-06,
+      "loss": 1.2532,
+      "step": 3821
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.076691091196242e-06,
+      "loss": 1.0394,
+      "step": 3822
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.074322026782064e-06,
+      "loss": 0.99,
+      "step": 3823
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0719539607221097e-06,
+      "loss": 1.0518,
+      "step": 3824
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0695868938244513e-06,
+      "loss": 1.0504,
+      "step": 3825
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.067220826896828e-06,
+      "loss": 1.0903,
+      "step": 3826
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0648557607466307e-06,
+      "loss": 1.1042,
+      "step": 3827
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0624916961809107e-06,
+      "loss": 1.0825,
+      "step": 3828
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0601286340063797e-06,
+      "loss": 1.0801,
+      "step": 3829
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.057766575029405e-06,
+      "loss": 1.0738,
+      "step": 3830
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.055405520056011e-06,
+      "loss": 1.0731,
+      "step": 3831
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0530454698918835e-06,
+      "loss": 1.026,
+      "step": 3832
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.050686425342361e-06,
+      "loss": 1.1443,
+      "step": 3833
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.048328387212439e-06,
+      "loss": 1.0428,
+      "step": 3834
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.045971356306774e-06,
+      "loss": 1.1166,
+      "step": 3835
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0436153334296755e-06,
+      "loss": 1.1891,
+      "step": 3836
+    },
+    {
+      "epoch": 0.7,
+      "learning_rate": 2.0412603193851065e-06,
+      "loss": 1.0902,
+      "step": 3837
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0389063149766924e-06,
+      "loss": 1.2253,
+      "step": 3838
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.036553321007709e-06,
+      "loss": 1.1131,
+      "step": 3839
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.034201338281086e-06,
+      "loss": 1.1186,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "eval_loss": 1.9316970109939575,
+      "eval_runtime": 0.1775,
+      "eval_samples_per_second": 16.898,
+      "eval_steps_per_second": 16.898,
+      "step": 3840
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0318503675994156e-06,
+      "loss": 1.0242,
+      "step": 3841
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0295004097649363e-06,
+      "loss": 1.0481,
+      "step": 3842
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0271514655795436e-06,
+      "loss": 1.0274,
+      "step": 3843
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.024803535844792e-06,
+      "loss": 1.0258,
+      "step": 3844
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.022456621361881e-06,
+      "loss": 1.1613,
+      "step": 3845
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0201107229316724e-06,
+      "loss": 1.0622,
+      "step": 3846
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0177658413546756e-06,
+      "loss": 1.1763,
+      "step": 3847
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0154219774310527e-06,
+      "loss": 0.9421,
+      "step": 3848
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0130791319606248e-06,
+      "loss": 1.1632,
+      "step": 3849
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.0107373057428583e-06,
+      "loss": 1.1413,
+      "step": 3850
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.008396499576873e-06,
+      "loss": 0.9842,
+      "step": 3851
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.006056714261447e-06,
+      "loss": 1.085,
+      "step": 3852
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.003717950595002e-06,
+      "loss": 1.2553,
+      "step": 3853
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 2.001380209375614e-06,
+      "loss": 0.974,
+      "step": 3854
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.999043491401013e-06,
+      "loss": 1.1967,
+      "step": 3855
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9967077974685766e-06,
+      "loss": 1.0712,
+      "step": 3856
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9943731283753307e-06,
+      "loss": 1.1118,
+      "step": 3857
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.99203948491796e-06,
+      "loss": 1.1641,
+      "step": 3858
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9897068678927895e-06,
+      "loss": 0.9783,
+      "step": 3859
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.987375278095798e-06,
+      "loss": 1.0258,
+      "step": 3860
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.985044716322617e-06,
+      "loss": 1.1095,
+      "step": 3861
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9827151833685222e-06,
+      "loss": 1.1427,
+      "step": 3862
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9803866800284384e-06,
+      "loss": 1.0861,
+      "step": 3863
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9780592070969447e-06,
+      "loss": 1.2913,
+      "step": 3864
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.975732765368262e-06,
+      "loss": 1.1877,
+      "step": 3865
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9734073556362616e-06,
+      "loss": 1.2546,
+      "step": 3866
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9710829786944657e-06,
+      "loss": 0.9571,
+      "step": 3867
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9687596353360378e-06,
+      "loss": 1.0593,
+      "step": 3868
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.966437326353796e-06,
+      "loss": 1.0474,
+      "step": 3869
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9641160525402005e-06,
+      "loss": 1.1602,
+      "step": 3870
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9617958146873565e-06,
+      "loss": 1.1205,
+      "step": 3871
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9594766135870237e-06,
+      "loss": 1.1329,
+      "step": 3872
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9571584500306003e-06,
+      "loss": 1.1829,
+      "step": 3873
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9548413248091315e-06,
+      "loss": 1.0899,
+      "step": 3874
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.952525238713313e-06,
+      "loss": 0.9477,
+      "step": 3875
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.950210192533482e-06,
+      "loss": 1.0267,
+      "step": 3876
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.947896187059619e-06,
+      "loss": 1.0662,
+      "step": 3877
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9455832230813565e-06,
+      "loss": 1.027,
+      "step": 3878
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.943271301387964e-06,
+      "loss": 1.1078,
+      "step": 3879
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.940960422768357e-06,
+      "loss": 1.1473,
+      "step": 3880
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9386505880111004e-06,
+      "loss": 1.1714,
+      "step": 3881
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9363417979043973e-06,
+      "loss": 1.1472,
+      "step": 3882
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.934034053236094e-06,
+      "loss": 0.9628,
+      "step": 3883
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.931727354793685e-06,
+      "loss": 1.0245,
+      "step": 3884
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9294217033643035e-06,
+      "loss": 1.2568,
+      "step": 3885
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9271170997347253e-06,
+      "loss": 1.0315,
+      "step": 3886
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9248135446913735e-06,
+      "loss": 1.1823,
+      "step": 3887
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9225110390203078e-06,
+      "loss": 1.2049,
+      "step": 3888
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9202095835072304e-06,
+      "loss": 1.1228,
+      "step": 3889
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.9179091789374905e-06,
+      "loss": 1.0605,
+      "step": 3890
+    },
+    {
+      "epoch": 0.71,
+      "learning_rate": 1.915609826096071e-06,
+      "loss": 1.2233,
+      "step": 3891
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.913311525767604e-06,
+      "loss": 1.0719,
+      "step": 3892
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9110142787363557e-06,
+      "loss": 1.095,
+      "step": 3893
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9087180857862337e-06,
+      "loss": 1.2171,
+      "step": 3894
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9064229477007916e-06,
+      "loss": 0.9841,
+      "step": 3895
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9041288652632173e-06,
+      "loss": 1.1253,
+      "step": 3896
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.9018358392563385e-06,
+      "loss": 1.1072,
+      "step": 3897
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8995438704626279e-06,
+      "loss": 1.0423,
+      "step": 3898
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8972529596641909e-06,
+      "loss": 1.0688,
+      "step": 3899
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8949631076427739e-06,
+      "loss": 1.2736,
+      "step": 3900
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.892674315179766e-06,
+      "loss": 1.0644,
+      "step": 3901
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8903865830561901e-06,
+      "loss": 0.949,
+      "step": 3902
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.888099912052706e-06,
+      "loss": 1.1027,
+      "step": 3903
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8858143029496184e-06,
+      "loss": 1.0521,
+      "step": 3904
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8835297565268634e-06,
+      "loss": 1.2731,
+      "step": 3905
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8812462735640146e-06,
+      "loss": 1.2209,
+      "step": 3906
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8789638548402883e-06,
+      "loss": 1.0922,
+      "step": 3907
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8766825011345314e-06,
+      "loss": 1.1753,
+      "step": 3908
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8744022132252283e-06,
+      "loss": 1.0644,
+      "step": 3909
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8721229918905053e-06,
+      "loss": 1.0638,
+      "step": 3910
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8698448379081163e-06,
+      "loss": 1.0386,
+      "step": 3911
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8675677520554597e-06,
+      "loss": 1.1518,
+      "step": 3912
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8652917351095628e-06,
+      "loss": 1.062,
+      "step": 3913
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8630167878470889e-06,
+      "loss": 1.1062,
+      "step": 3914
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8607429110443415e-06,
+      "loss": 1.0778,
+      "step": 3915
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8584701054772524e-06,
+      "loss": 1.0517,
+      "step": 3916
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.85619837192139e-06,
+      "loss": 1.1616,
+      "step": 3917
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8539277111519598e-06,
+      "loss": 1.2413,
+      "step": 3918
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.851658123943798e-06,
+      "loss": 1.0633,
+      "step": 3919
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8493896110713733e-06,
+      "loss": 1.0958,
+      "step": 3920
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8471221733087934e-06,
+      "loss": 1.2269,
+      "step": 3921
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8448558114297939e-06,
+      "loss": 1.3,
+      "step": 3922
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8425905262077432e-06,
+      "loss": 1.1161,
+      "step": 3923
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8403263184156466e-06,
+      "loss": 1.0798,
+      "step": 3924
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.838063188826139e-06,
+      "loss": 1.1058,
+      "step": 3925
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8358011382114848e-06,
+      "loss": 1.155,
+      "step": 3926
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8335401673435864e-06,
+      "loss": 1.0657,
+      "step": 3927
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.831280276993973e-06,
+      "loss": 1.238,
+      "step": 3928
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8290214679338047e-06,
+      "loss": 1.2102,
+      "step": 3929
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8267637409338773e-06,
+      "loss": 1.111,
+      "step": 3930
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8245070967646128e-06,
+      "loss": 1.0464,
+      "step": 3931
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8222515361960636e-06,
+      "loss": 0.9933,
+      "step": 3932
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8199970599979179e-06,
+      "loss": 0.9146,
+      "step": 3933
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8177436689394855e-06,
+      "loss": 1.0887,
+      "step": 3934
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8154913637897154e-06,
+      "loss": 1.1995,
+      "step": 3935
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8132401453171777e-06,
+      "loss": 1.0548,
+      "step": 3936
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8109900142900737e-06,
+      "loss": 1.3024,
+      "step": 3937
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8087409714762394e-06,
+      "loss": 1.0458,
+      "step": 3938
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8064930176431322e-06,
+      "loss": 1.0051,
+      "step": 3939
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8042461535578392e-06,
+      "loss": 1.0724,
+      "step": 3940
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.8020003799870806e-06,
+      "loss": 1.0519,
+      "step": 3941
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7997556976971992e-06,
+      "loss": 1.1633,
+      "step": 3942
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7975121074541663e-06,
+      "loss": 1.0809,
+      "step": 3943
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7952696100235834e-06,
+      "loss": 0.9191,
+      "step": 3944
+    },
+    {
+      "epoch": 0.72,
+      "learning_rate": 1.7930282061706766e-06,
+      "loss": 1.0081,
+      "step": 3945
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7907878966602975e-06,
+      "loss": 1.0884,
+      "step": 3946
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7885486822569298e-06,
+      "loss": 1.0808,
+      "step": 3947
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.786310563724677e-06,
+      "loss": 1.0534,
+      "step": 3948
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7840735418272715e-06,
+      "loss": 1.1158,
+      "step": 3949
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7818376173280732e-06,
+      "loss": 1.1129,
+      "step": 3950
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7796027909900653e-06,
+      "loss": 1.0802,
+      "step": 3951
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7773690635758545e-06,
+      "loss": 1.106,
+      "step": 3952
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7751364358476787e-06,
+      "loss": 1.1044,
+      "step": 3953
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7729049085673944e-06,
+      "loss": 1.2939,
+      "step": 3954
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7706744824964833e-06,
+      "loss": 1.1159,
+      "step": 3955
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7684451583960565e-06,
+      "loss": 1.0199,
+      "step": 3956
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7662169370268417e-06,
+      "loss": 1.0615,
+      "step": 3957
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7639898191491972e-06,
+      "loss": 1.3287,
+      "step": 3958
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7617638055231008e-06,
+      "loss": 1.0251,
+      "step": 3959
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7595388969081518e-06,
+      "loss": 1.2002,
+      "step": 3960
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.757315094063578e-06,
+      "loss": 1.1642,
+      "step": 3961
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.755092397748226e-06,
+      "loss": 1.1869,
+      "step": 3962
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7528708087205626e-06,
+      "loss": 1.1543,
+      "step": 3963
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7506503277386845e-06,
+      "loss": 0.9541,
+      "step": 3964
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7484309555603024e-06,
+      "loss": 1.1099,
+      "step": 3965
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7462126929427515e-06,
+      "loss": 1.0625,
+      "step": 3966
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7439955406429908e-06,
+      "loss": 1.0515,
+      "step": 3967
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7417794994175975e-06,
+      "loss": 1.0386,
+      "step": 3968
+    },
+    {
+      "epoch": 0.73,
+      "eval_loss": 1.9325541257858276,
+      "eval_runtime": 0.1694,
+      "eval_samples_per_second": 17.71,
+      "eval_steps_per_second": 17.71,
+      "step": 3968
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7395645700227677e-06,
+      "loss": 1.0119,
+      "step": 3969
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7373507532143242e-06,
+      "loss": 1.0695,
+      "step": 3970
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.735138049747706e-06,
+      "loss": 0.9752,
+      "step": 3971
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7329264603779705e-06,
+      "loss": 1.2204,
+      "step": 3972
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.730715985859801e-06,
+      "loss": 1.0688,
+      "step": 3973
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7285066269474948e-06,
+      "loss": 1.2043,
+      "step": 3974
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7262983843949689e-06,
+      "loss": 1.0728,
+      "step": 3975
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.724091258955764e-06,
+      "loss": 1.1565,
+      "step": 3976
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7218852513830335e-06,
+      "loss": 1.1277,
+      "step": 3977
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7196803624295555e-06,
+      "loss": 1.1183,
+      "step": 3978
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.717476592847721e-06,
+      "loss": 1.1984,
+      "step": 3979
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7152739433895404e-06,
+      "loss": 1.1797,
+      "step": 3980
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7130724148066458e-06,
+      "loss": 1.1857,
+      "step": 3981
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7108720078502815e-06,
+      "loss": 1.1974,
+      "step": 3982
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7086727232713103e-06,
+      "loss": 1.1505,
+      "step": 3983
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7064745618202161e-06,
+      "loss": 1.0732,
+      "step": 3984
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7042775242470943e-06,
+      "loss": 0.9908,
+      "step": 3985
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.7020816113016575e-06,
+      "loss": 1.1073,
+      "step": 3986
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6998868237332393e-06,
+      "loss": 1.1034,
+      "step": 3987
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.697693162290785e-06,
+      "loss": 1.238,
+      "step": 3988
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6955006277228536e-06,
+      "loss": 1.1249,
+      "step": 3989
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6933092207776264e-06,
+      "loss": 1.1793,
+      "step": 3990
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.691118942202894e-06,
+      "loss": 1.0884,
+      "step": 3991
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.688929792746063e-06,
+      "loss": 1.3342,
+      "step": 3992
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6867417731541586e-06,
+      "loss": 1.0579,
+      "step": 3993
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6845548841738158e-06,
+      "loss": 0.9633,
+      "step": 3994
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6823691265512842e-06,
+      "loss": 1.1481,
+      "step": 3995
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6801845010324314e-06,
+      "loss": 1.1845,
+      "step": 3996
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6780010083627346e-06,
+      "loss": 1.1973,
+      "step": 3997
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.675818649287284e-06,
+      "loss": 1.2939,
+      "step": 3998
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.6736374245507886e-06,
+      "loss": 1.3282,
+      "step": 3999
+    },
+    {
+      "epoch": 0.73,
+      "learning_rate": 1.671457334897562e-06,
+      "loss": 1.0251,
+      "step": 4000
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6692783810715396e-06,
+      "loss": 0.972,
+      "step": 4001
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6671005638162619e-06,
+      "loss": 1.1376,
+      "step": 4002
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6649238838748822e-06,
+      "loss": 1.1086,
+      "step": 4003
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6627483419901725e-06,
+      "loss": 1.0851,
+      "step": 4004
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6605739389045083e-06,
+      "loss": 1.0843,
+      "step": 4005
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6584006753598792e-06,
+      "loss": 1.1765,
+      "step": 4006
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6562285520978898e-06,
+      "loss": 1.0471,
+      "step": 4007
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6540575698597506e-06,
+      "loss": 1.1642,
+      "step": 4008
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6518877293862829e-06,
+      "loss": 1.0987,
+      "step": 4009
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6497190314179228e-06,
+      "loss": 1.0858,
+      "step": 4010
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6475514766947132e-06,
+      "loss": 1.2609,
+      "step": 4011
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6453850659563047e-06,
+      "loss": 1.0752,
+      "step": 4012
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.643219799941964e-06,
+      "loss": 1.0602,
+      "step": 4013
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6410556793905613e-06,
+      "loss": 1.1353,
+      "step": 4014
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6388927050405773e-06,
+      "loss": 1.0973,
+      "step": 4015
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6367308776301049e-06,
+      "loss": 1.178,
+      "step": 4016
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.634570197896842e-06,
+      "loss": 1.1463,
+      "step": 4017
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6324106665780936e-06,
+      "loss": 1.0909,
+      "step": 4018
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6302522844107788e-06,
+      "loss": 1.0421,
+      "step": 4019
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6280950521314186e-06,
+      "loss": 1.1945,
+      "step": 4020
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.625938970476143e-06,
+      "loss": 1.0393,
+      "step": 4021
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6237840401806937e-06,
+      "loss": 1.2288,
+      "step": 4022
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6216302619804119e-06,
+      "loss": 1.1832,
+      "step": 4023
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6194776366102539e-06,
+      "loss": 1.0944,
+      "step": 4024
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6173261648047766e-06,
+      "loss": 1.1181,
+      "step": 4025
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.615175847298144e-06,
+      "loss": 1.1752,
+      "step": 4026
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6130266848241295e-06,
+      "loss": 1.104,
+      "step": 4027
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6108786781161096e-06,
+      "loss": 0.9678,
+      "step": 4028
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6087318279070652e-06,
+      "loss": 0.9937,
+      "step": 4029
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.606586134929588e-06,
+      "loss": 1.0702,
+      "step": 4030
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.604441599915869e-06,
+      "loss": 1.1694,
+      "step": 4031
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.6022982235977057e-06,
+      "loss": 1.0694,
+      "step": 4032
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.600156006706503e-06,
+      "loss": 1.0361,
+      "step": 4033
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5980149499732666e-06,
+      "loss": 0.9352,
+      "step": 4034
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5958750541286067e-06,
+      "loss": 1.0714,
+      "step": 4035
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5937363199027406e-06,
+      "loss": 1.1775,
+      "step": 4036
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5915987480254858e-06,
+      "loss": 1.1645,
+      "step": 4037
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5894623392262627e-06,
+      "loss": 1.1925,
+      "step": 4038
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5873270942340996e-06,
+      "loss": 1.2784,
+      "step": 4039
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5851930137776233e-06,
+      "loss": 1.1585,
+      "step": 4040
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5830600985850624e-06,
+      "loss": 0.9717,
+      "step": 4041
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5809283493842525e-06,
+      "loss": 1.1204,
+      "step": 4042
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5787977669026267e-06,
+      "loss": 1.0804,
+      "step": 4043
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5766683518672243e-06,
+      "loss": 1.1007,
+      "step": 4044
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5745401050046826e-06,
+      "loss": 1.2012,
+      "step": 4045
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5724130270412402e-06,
+      "loss": 0.9829,
+      "step": 4046
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5702871187027408e-06,
+      "loss": 1.0411,
+      "step": 4047
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5681623807146245e-06,
+      "loss": 1.1428,
+      "step": 4048
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5660388138019333e-06,
+      "loss": 0.971,
+      "step": 4049
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5639164186893129e-06,
+      "loss": 1.249,
+      "step": 4050
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5617951961010053e-06,
+      "loss": 1.2528,
+      "step": 4051
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.5596751467608506e-06,
+      "loss": 1.2003,
+      "step": 4052
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.557556271392296e-06,
+      "loss": 1.1774,
+      "step": 4053
+    },
+    {
+      "epoch": 0.74,
+      "learning_rate": 1.555438570718381e-06,
+      "loss": 1.1462,
+      "step": 4054
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5533220454617459e-06,
+      "loss": 1.2381,
+      "step": 4055
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5512066963446338e-06,
+      "loss": 0.9994,
+      "step": 4056
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5490925240888816e-06,
+      "loss": 1.2394,
+      "step": 4057
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5469795294159244e-06,
+      "loss": 1.0998,
+      "step": 4058
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.544867713046802e-06,
+      "loss": 1.0095,
+      "step": 4059
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5427570757021453e-06,
+      "loss": 1.084,
+      "step": 4060
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5406476181021834e-06,
+      "loss": 1.1747,
+      "step": 4061
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.538539340966748e-06,
+      "loss": 1.0613,
+      "step": 4062
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5364322450152635e-06,
+      "loss": 1.1258,
+      "step": 4063
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5343263309667505e-06,
+      "loss": 1.087,
+      "step": 4064
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.532221599539831e-06,
+      "loss": 1.1045,
+      "step": 4065
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.530118051452717e-06,
+      "loss": 1.0605,
+      "step": 4066
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.528015687423225e-06,
+      "loss": 1.1761,
+      "step": 4067
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5259145081687592e-06,
+      "loss": 1.14,
+      "step": 4068
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5238145144063232e-06,
+      "loss": 1.1858,
+      "step": 4069
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5217157068525179e-06,
+      "loss": 1.0779,
+      "step": 4070
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5196180862235366e-06,
+      "loss": 1.1391,
+      "step": 4071
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5175216532351662e-06,
+      "loss": 1.1598,
+      "step": 4072
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5154264086027942e-06,
+      "loss": 1.1517,
+      "step": 4073
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5133323530413967e-06,
+      "loss": 1.0166,
+      "step": 4074
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5112394872655457e-06,
+      "loss": 0.9364,
+      "step": 4075
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5091478119894099e-06,
+      "loss": 1.1554,
+      "step": 4076
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5070573279267487e-06,
+      "loss": 1.2521,
+      "step": 4077
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.504968035790914e-06,
+      "loss": 1.102,
+      "step": 4078
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5028799362948565e-06,
+      "loss": 1.0742,
+      "step": 4079
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.5007930301511143e-06,
+      "loss": 1.2315,
+      "step": 4080
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4987073180718187e-06,
+      "loss": 1.2308,
+      "step": 4081
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4966228007686984e-06,
+      "loss": 1.1144,
+      "step": 4082
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.49453947895307e-06,
+      "loss": 1.1418,
+      "step": 4083
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4924573533358406e-06,
+      "loss": 1.1516,
+      "step": 4084
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4903764246275165e-06,
+      "loss": 1.1764,
+      "step": 4085
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4882966935381882e-06,
+      "loss": 1.2144,
+      "step": 4086
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4862181607775388e-06,
+      "loss": 1.1534,
+      "step": 4087
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4841408270548474e-06,
+      "loss": 1.2137,
+      "step": 4088
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4820646930789766e-06,
+      "loss": 1.0871,
+      "step": 4089
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4799897595583879e-06,
+      "loss": 1.3021,
+      "step": 4090
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4779160272011255e-06,
+      "loss": 1.0292,
+      "step": 4091
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4758434967148267e-06,
+      "loss": 1.2684,
+      "step": 4092
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4737721688067208e-06,
+      "loss": 1.3042,
+      "step": 4093
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4717020441836239e-06,
+      "loss": 1.0786,
+      "step": 4094
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4696331235519411e-06,
+      "loss": 1.2585,
+      "step": 4095
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4675654076176703e-06,
+      "loss": 1.1777,
+      "step": 4096
+    },
+    {
+      "epoch": 0.75,
+      "eval_loss": 1.9340654611587524,
+      "eval_runtime": 0.1656,
+      "eval_samples_per_second": 18.112,
+      "eval_steps_per_second": 18.112,
+      "step": 4096
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4654988970863954e-06,
+      "loss": 1.0881,
+      "step": 4097
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.463433592663287e-06,
+      "loss": 1.0081,
+      "step": 4098
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4613694950531103e-06,
+      "loss": 1.2027,
+      "step": 4099
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4593066049602135e-06,
+      "loss": 1.1325,
+      "step": 4100
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4572449230885327e-06,
+      "loss": 1.2456,
+      "step": 4101
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4551844501415964e-06,
+      "loss": 1.0894,
+      "step": 4102
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.453125186822516e-06,
+      "loss": 1.0251,
+      "step": 4103
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.45106713383399e-06,
+      "loss": 1.0476,
+      "step": 4104
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4490102918783083e-06,
+      "loss": 1.2105,
+      "step": 4105
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4469546616573438e-06,
+      "loss": 1.003,
+      "step": 4106
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4449002438725545e-06,
+      "loss": 1.0518,
+      "step": 4107
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.442847039224991e-06,
+      "loss": 1.1396,
+      "step": 4108
+    },
+    {
+      "epoch": 0.75,
+      "learning_rate": 1.4407950484152828e-06,
+      "loss": 1.2709,
+      "step": 4109
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4387442721436506e-06,
+      "loss": 1.088,
+      "step": 4110
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4366947111098972e-06,
+      "loss": 0.9843,
+      "step": 4111
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4346463660134107e-06,
+      "loss": 1.0355,
+      "step": 4112
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4325992375531684e-06,
+      "loss": 1.1628,
+      "step": 4113
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4305533264277283e-06,
+      "loss": 1.2284,
+      "step": 4114
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4285086333352322e-06,
+      "loss": 1.3384,
+      "step": 4115
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4264651589734114e-06,
+      "loss": 1.0541,
+      "step": 4116
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.424422904039577e-06,
+      "loss": 0.9746,
+      "step": 4117
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4223818692306235e-06,
+      "loss": 1.0342,
+      "step": 4118
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4203420552430336e-06,
+      "loss": 0.9292,
+      "step": 4119
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4183034627728692e-06,
+      "loss": 1.0255,
+      "step": 4120
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.416266092515775e-06,
+      "loss": 1.1363,
+      "step": 4121
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.414229945166984e-06,
+      "loss": 1.2949,
+      "step": 4122
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.412195021421306e-06,
+      "loss": 1.1384,
+      "step": 4123
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4101613219731342e-06,
+      "loss": 1.0856,
+      "step": 4124
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.408128847516449e-06,
+      "loss": 1.0364,
+      "step": 4125
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4060975987448066e-06,
+      "loss": 1.1343,
+      "step": 4126
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4040675763513462e-06,
+      "loss": 1.1303,
+      "step": 4127
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4020387810287932e-06,
+      "loss": 1.0994,
+      "step": 4128
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.4000112134694494e-06,
+      "loss": 1.1781,
+      "step": 4129
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3979848743651974e-06,
+      "loss": 1.1178,
+      "step": 4130
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3959597644075052e-06,
+      "loss": 1.1428,
+      "step": 4131
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.393935884287415e-06,
+      "loss": 1.0386,
+      "step": 4132
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3919132346955567e-06,
+      "loss": 0.9808,
+      "step": 4133
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3898918163221342e-06,
+      "loss": 1.1716,
+      "step": 4134
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3878716298569328e-06,
+      "loss": 1.0944,
+      "step": 4135
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3858526759893198e-06,
+      "loss": 1.1383,
+      "step": 4136
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3838349554082397e-06,
+      "loss": 1.105,
+      "step": 4137
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.381818468802215e-06,
+      "loss": 1.1292,
+      "step": 4138
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3798032168593517e-06,
+      "loss": 1.0614,
+      "step": 4139
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.37778920026733e-06,
+      "loss": 1.1053,
+      "step": 4140
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3757764197134077e-06,
+      "loss": 1.0344,
+      "step": 4141
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3737648758844274e-06,
+      "loss": 1.1922,
+      "step": 4142
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3717545694668032e-06,
+      "loss": 1.1422,
+      "step": 4143
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3697455011465288e-06,
+      "loss": 1.2518,
+      "step": 4144
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3677376716091778e-06,
+      "loss": 1.0764,
+      "step": 4145
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3657310815398978e-06,
+      "loss": 1.1076,
+      "step": 4146
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.363725731623413e-06,
+      "loss": 1.0773,
+      "step": 4147
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3617216225440299e-06,
+      "loss": 1.1056,
+      "step": 4148
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.359718754985626e-06,
+      "loss": 1.0508,
+      "step": 4149
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.357717129631656e-06,
+      "loss": 1.1073,
+      "step": 4150
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.355716747165154e-06,
+      "loss": 1.1127,
+      "step": 4151
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3537176082687258e-06,
+      "loss": 1.2192,
+      "step": 4152
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3517197136245546e-06,
+      "loss": 1.0444,
+      "step": 4153
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.349723063914401e-06,
+      "loss": 1.011,
+      "step": 4154
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3477276598195966e-06,
+      "loss": 1.125,
+      "step": 4155
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3457335020210527e-06,
+      "loss": 0.9204,
+      "step": 4156
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.343740591199252e-06,
+      "loss": 1.041,
+      "step": 4157
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3417489280342504e-06,
+      "loss": 1.1188,
+      "step": 4158
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3397585132056833e-06,
+      "loss": 1.2187,
+      "step": 4159
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3377693473927567e-06,
+      "loss": 1.0213,
+      "step": 4160
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.335781431274248e-06,
+      "loss": 1.144,
+      "step": 4161
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3337947655285145e-06,
+      "loss": 1.1105,
+      "step": 4162
+    },
+    {
+      "epoch": 0.76,
+      "learning_rate": 1.3318093508334812e-06,
+      "loss": 1.1164,
+      "step": 4163
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3298251878666468e-06,
+      "loss": 1.15,
+      "step": 4164
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.327842277305088e-06,
+      "loss": 1.276,
+      "step": 4165
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.325860619825448e-06,
+      "loss": 1.1293,
+      "step": 4166
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3238802161039438e-06,
+      "loss": 1.0541,
+      "step": 4167
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.321901066816369e-06,
+      "loss": 1.1831,
+      "step": 4168
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3199231726380829e-06,
+      "loss": 1.1176,
+      "step": 4169
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.317946534244019e-06,
+      "loss": 1.1176,
+      "step": 4170
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3159711523086848e-06,
+      "loss": 1.2406,
+      "step": 4171
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.313997027506156e-06,
+      "loss": 1.0749,
+      "step": 4172
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.312024160510078e-06,
+      "loss": 1.1412,
+      "step": 4173
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.310052551993673e-06,
+      "loss": 1.0797,
+      "step": 4174
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3080822026297263e-06,
+      "loss": 1.0614,
+      "step": 4175
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3061131130906002e-06,
+      "loss": 1.0723,
+      "step": 4176
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3041452840482227e-06,
+      "loss": 1.1559,
+      "step": 4177
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3021787161740912e-06,
+      "loss": 1.1696,
+      "step": 4178
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.3002134101392776e-06,
+      "loss": 1.1435,
+      "step": 4179
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2982493666144191e-06,
+      "loss": 1.15,
+      "step": 4180
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2962865862697216e-06,
+      "loss": 1.1319,
+      "step": 4181
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2943250697749642e-06,
+      "loss": 1.1376,
+      "step": 4182
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2923648177994901e-06,
+      "loss": 1.2492,
+      "step": 4183
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2904058310122126e-06,
+      "loss": 1.1266,
+      "step": 4184
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2884481100816155e-06,
+      "loss": 1.2026,
+      "step": 4185
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2864916556757472e-06,
+      "loss": 1.1638,
+      "step": 4186
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2845364684622247e-06,
+      "loss": 0.9623,
+      "step": 4187
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2825825491082361e-06,
+      "loss": 0.9802,
+      "step": 4188
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2806298982805321e-06,
+      "loss": 1.109,
+      "step": 4189
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2786785166454318e-06,
+      "loss": 1.0146,
+      "step": 4190
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2767284048688245e-06,
+      "loss": 1.0454,
+      "step": 4191
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.274779563616162e-06,
+      "loss": 1.0692,
+      "step": 4192
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2728319935524624e-06,
+      "loss": 1.0886,
+      "step": 4193
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2708856953423149e-06,
+      "loss": 1.0306,
+      "step": 4194
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2689406696498706e-06,
+      "loss": 1.0793,
+      "step": 4195
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2669969171388457e-06,
+      "loss": 1.0929,
+      "step": 4196
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.265054438472526e-06,
+      "loss": 1.1131,
+      "step": 4197
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2631132343137575e-06,
+      "loss": 1.0587,
+      "step": 4198
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.261173305324957e-06,
+      "loss": 1.1676,
+      "step": 4199
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2592346521681014e-06,
+      "loss": 1.1618,
+      "step": 4200
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.257297275504733e-06,
+      "loss": 1.0373,
+      "step": 4201
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2553611759959623e-06,
+      "loss": 1.2005,
+      "step": 4202
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2534263543024588e-06,
+      "loss": 1.1256,
+      "step": 4203
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2514928110844582e-06,
+      "loss": 0.9951,
+      "step": 4204
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2495605470017618e-06,
+      "loss": 1.1925,
+      "step": 4205
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.247629562713732e-06,
+      "loss": 1.1936,
+      "step": 4206
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2456998588792934e-06,
+      "loss": 1.1853,
+      "step": 4207
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2437714361569387e-06,
+      "loss": 1.1031,
+      "step": 4208
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2418442952047177e-06,
+      "loss": 1.0551,
+      "step": 4209
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2399184366802453e-06,
+      "loss": 1.1035,
+      "step": 4210
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2379938612407006e-06,
+      "loss": 0.9716,
+      "step": 4211
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.236070569542821e-06,
+      "loss": 1.2217,
+      "step": 4212
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2341485622429078e-06,
+      "loss": 1.1396,
+      "step": 4213
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2322278399968262e-06,
+      "loss": 1.226,
+      "step": 4214
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2303084034599999e-06,
+      "loss": 1.0097,
+      "step": 4215
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2283902532874125e-06,
+      "loss": 1.157,
+      "step": 4216
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.226473390133614e-06,
+      "loss": 1.0823,
+      "step": 4217
+    },
+    {
+      "epoch": 0.77,
+      "learning_rate": 1.2245578146527104e-06,
+      "loss": 1.1874,
+      "step": 4218
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2226435274983694e-06,
+      "loss": 1.0954,
+      "step": 4219
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2207305293238214e-06,
+      "loss": 1.1413,
+      "step": 4220
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.218818820781853e-06,
+      "loss": 1.1868,
+      "step": 4221
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2169084025248156e-06,
+      "loss": 1.1142,
+      "step": 4222
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.214999275204617e-06,
+      "loss": 1.1064,
+      "step": 4223
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2130914394727222e-06,
+      "loss": 1.026,
+      "step": 4224
+    },
+    {
+      "epoch": 0.78,
+      "eval_loss": 1.935506820678711,
+      "eval_runtime": 0.1849,
+      "eval_samples_per_second": 16.226,
+      "eval_steps_per_second": 16.226,
+      "step": 4224
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.211184895980162e-06,
+      "loss": 1.177,
+      "step": 4225
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2092796453775213e-06,
+      "loss": 1.1922,
+      "step": 4226
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2073756883149433e-06,
+      "loss": 1.1514,
+      "step": 4227
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.205473025442135e-06,
+      "loss": 1.1319,
+      "step": 4228
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2035716574083567e-06,
+      "loss": 1.045,
+      "step": 4229
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.2016715848624273e-06,
+      "loss": 1.1491,
+      "step": 4230
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1997728084527272e-06,
+      "loss": 1.0821,
+      "step": 4231
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.197875328827191e-06,
+      "loss": 1.1233,
+      "step": 4232
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1959791466333109e-06,
+      "loss": 0.9281,
+      "step": 4233
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1940842625181398e-06,
+      "loss": 1.1426,
+      "step": 4234
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1921906771282844e-06,
+      "loss": 1.1087,
+      "step": 4235
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.190298391109907e-06,
+      "loss": 1.1759,
+      "step": 4236
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.188407405108732e-06,
+      "loss": 1.0596,
+      "step": 4237
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1865177197700351e-06,
+      "loss": 1.1504,
+      "step": 4238
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1846293357386484e-06,
+      "loss": 1.0179,
+      "step": 4239
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.182742253658965e-06,
+      "loss": 1.2112,
+      "step": 4240
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1808564741749261e-06,
+      "loss": 1.1204,
+      "step": 4241
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1789719979300363e-06,
+      "loss": 1.2092,
+      "step": 4242
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.177088825567349e-06,
+      "loss": 1.0653,
+      "step": 4243
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.175206957729475e-06,
+      "loss": 1.0971,
+      "step": 4244
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1733263950585827e-06,
+      "loss": 1.1682,
+      "step": 4245
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1714471381963916e-06,
+      "loss": 1.2262,
+      "step": 4246
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1695691877841741e-06,
+      "loss": 1.0585,
+      "step": 4247
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.167692544462763e-06,
+      "loss": 0.9188,
+      "step": 4248
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.16581720887254e-06,
+      "loss": 1.1579,
+      "step": 4249
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1639431816534408e-06,
+      "loss": 0.9788,
+      "step": 4250
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.162070463444958e-06,
+      "loss": 1.1764,
+      "step": 4251
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1601990548861342e-06,
+      "loss": 1.1082,
+      "step": 4252
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1583289566155647e-06,
+      "loss": 1.1356,
+      "step": 4253
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1564601692714022e-06,
+      "loss": 1.216,
+      "step": 4254
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1545926934913477e-06,
+      "loss": 1.1381,
+      "step": 4255
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1527265299126545e-06,
+      "loss": 1.1137,
+      "step": 4256
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1508616791721322e-06,
+      "loss": 0.9448,
+      "step": 4257
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1489981419061392e-06,
+      "loss": 0.9986,
+      "step": 4258
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1471359187505837e-06,
+      "loss": 1.2638,
+      "step": 4259
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1452750103409317e-06,
+      "loss": 1.1863,
+      "step": 4260
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1434154173121952e-06,
+      "loss": 1.1289,
+      "step": 4261
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1415571402989384e-06,
+      "loss": 1.1696,
+      "step": 4262
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1397001799352792e-06,
+      "loss": 1.1732,
+      "step": 4263
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1378445368548813e-06,
+      "loss": 1.0089,
+      "step": 4264
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.135990211690965e-06,
+      "loss": 0.991,
+      "step": 4265
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1341372050762961e-06,
+      "loss": 1.0227,
+      "step": 4266
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1322855176431907e-06,
+      "loss": 1.194,
+      "step": 4267
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1304351500235183e-06,
+      "loss": 1.0733,
+      "step": 4268
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1285861028486945e-06,
+      "loss": 1.0571,
+      "step": 4269
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1267383767496843e-06,
+      "loss": 1.0603,
+      "step": 4270
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1248919723570063e-06,
+      "loss": 1.1848,
+      "step": 4271
+    },
+    {
+      "epoch": 0.78,
+      "learning_rate": 1.1230468903007225e-06,
+      "loss": 1.0547,
+      "step": 4272
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.121203131210446e-06,
+      "loss": 1.0569,
+      "step": 4273
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1193606957153401e-06,
+      "loss": 1.216,
+      "step": 4274
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1175195844441144e-06,
+      "loss": 1.0255,
+      "step": 4275
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1156797980250251e-06,
+      "loss": 1.155,
+      "step": 4276
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.113841337085882e-06,
+      "loss": 1.0546,
+      "step": 4277
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.112004202254036e-06,
+      "loss": 1.0275,
+      "step": 4278
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1101683941563884e-06,
+      "loss": 1.1396,
+      "step": 4279
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1083339134193905e-06,
+      "loss": 1.1493,
+      "step": 4280
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1065007606690352e-06,
+      "loss": 1.0437,
+      "step": 4281
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1046689365308656e-06,
+      "loss": 1.1434,
+      "step": 4282
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.1028384416299715e-06,
+      "loss": 1.2043,
+      "step": 4283
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.101009276590988e-06,
+      "loss": 1.1697,
+      "step": 4284
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0991814420380957e-06,
+      "loss": 1.0895,
+      "step": 4285
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0973549385950249e-06,
+      "loss": 1.125,
+      "step": 4286
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0955297668850456e-06,
+      "loss": 1.1505,
+      "step": 4287
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0937059275309802e-06,
+      "loss": 1.1152,
+      "step": 4288
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0918834211551914e-06,
+      "loss": 1.1926,
+      "step": 4289
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.090062248379588e-06,
+      "loss": 1.2533,
+      "step": 4290
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0882424098256261e-06,
+      "loss": 1.0777,
+      "step": 4291
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0864239061143045e-06,
+      "loss": 1.0632,
+      "step": 4292
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0846067378661646e-06,
+      "loss": 1.0037,
+      "step": 4293
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0827909057012981e-06,
+      "loss": 1.0445,
+      "step": 4294
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0809764102393343e-06,
+      "loss": 1.2908,
+      "step": 4295
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0791632520994487e-06,
+      "loss": 1.3422,
+      "step": 4296
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0773514319003631e-06,
+      "loss": 1.0517,
+      "step": 4297
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0755409502603398e-06,
+      "loss": 1.1665,
+      "step": 4298
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0737318077971826e-06,
+      "loss": 1.0688,
+      "step": 4299
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0719240051282442e-06,
+      "loss": 1.1167,
+      "step": 4300
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0701175428704147e-06,
+      "loss": 1.107,
+      "step": 4301
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.068312421640128e-06,
+      "loss": 1.1764,
+      "step": 4302
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0665086420533632e-06,
+      "loss": 1.0335,
+      "step": 4303
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.064706204725639e-06,
+      "loss": 1.223,
+      "step": 4304
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.062905110272014e-06,
+      "loss": 1.1035,
+      "step": 4305
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0611053593070953e-06,
+      "loss": 1.1315,
+      "step": 4306
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0593069524450233e-06,
+      "loss": 1.0693,
+      "step": 4307
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0575098902994873e-06,
+      "loss": 1.0473,
+      "step": 4308
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.055714173483713e-06,
+      "loss": 1.0747,
+      "step": 4309
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0539198026104664e-06,
+      "loss": 1.2644,
+      "step": 4310
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.052126778292059e-06,
+      "loss": 1.0003,
+      "step": 4311
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0503351011403384e-06,
+      "loss": 1.1625,
+      "step": 4312
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.048544771766693e-06,
+      "loss": 1.2337,
+      "step": 4313
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0467557907820552e-06,
+      "loss": 1.0954,
+      "step": 4314
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0449681587968918e-06,
+      "loss": 1.2931,
+      "step": 4315
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.043181876421212e-06,
+      "loss": 1.1844,
+      "step": 4316
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0413969442645655e-06,
+      "loss": 1.076,
+      "step": 4317
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.03961336293604e-06,
+      "loss": 1.0887,
+      "step": 4318
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0378311330442598e-06,
+      "loss": 1.0734,
+      "step": 4319
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0360502551973938e-06,
+      "loss": 1.1881,
+      "step": 4320
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0342707300031451e-06,
+      "loss": 1.182,
+      "step": 4321
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0324925580687551e-06,
+      "loss": 1.0958,
+      "step": 4322
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0307157400010066e-06,
+      "loss": 1.2127,
+      "step": 4323
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0289402764062173e-06,
+      "loss": 1.1688,
+      "step": 4324
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0271661678902427e-06,
+      "loss": 1.2347,
+      "step": 4325
+    },
+    {
+      "epoch": 0.79,
+      "learning_rate": 1.0253934150584805e-06,
+      "loss": 0.9539,
+      "step": 4326
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0236220185158603e-06,
+      "loss": 1.1103,
+      "step": 4327
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.021851978866849e-06,
+      "loss": 1.085,
+      "step": 4328
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0200832967154567e-06,
+      "loss": 0.9865,
+      "step": 4329
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.018315972665222e-06,
+      "loss": 1.0001,
+      "step": 4330
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0165500073192274e-06,
+      "loss": 1.1654,
+      "step": 4331
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0147854012800868e-06,
+      "loss": 1.06,
+      "step": 4332
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0130221551499508e-06,
+      "loss": 1.3845,
+      "step": 4333
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0112602695305096e-06,
+      "loss": 1.147,
+      "step": 4334
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.009499745022985e-06,
+      "loss": 1.0872,
+      "step": 4335
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0077405822281345e-06,
+      "loss": 1.0711,
+      "step": 4336
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0059827817462554e-06,
+      "loss": 1.0452,
+      "step": 4337
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0042263441771755e-06,
+      "loss": 1.2318,
+      "step": 4338
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.0024712701202576e-06,
+      "loss": 0.9953,
+      "step": 4339
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 1.000717560174404e-06,
+      "loss": 1.0346,
+      "step": 4340
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.989652149380457e-07,
+      "loss": 1.0226,
+      "step": 4341
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.972142350091507e-07,
+      "loss": 0.963,
+      "step": 4342
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.954646209852226e-07,
+      "loss": 1.1598,
+      "step": 4343
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.937163734632954e-07,
+      "loss": 1.1472,
+      "step": 4344
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.919694930399386e-07,
+      "loss": 1.0826,
+      "step": 4345
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.902239803112573e-07,
+      "loss": 1.0809,
+      "step": 4346
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.884798358728858e-07,
+      "loss": 1.0524,
+      "step": 4347
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.867370603199934e-07,
+      "loss": 1.2415,
+      "step": 4348
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.849956542472839e-07,
+      "loss": 1.0222,
+      "step": 4349
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.832556182489916e-07,
+      "loss": 1.1152,
+      "step": 4350
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.815169529188829e-07,
+      "loss": 1.0411,
+      "step": 4351
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.79779658850259e-07,
+      "loss": 1.2816,
+      "step": 4352
+    },
+    {
+      "epoch": 0.8,
+      "eval_loss": 1.931909203529358,
+      "eval_runtime": 0.185,
+      "eval_samples_per_second": 16.219,
+      "eval_steps_per_second": 16.219,
+      "step": 4352
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.780437366359496e-07,
+      "loss": 1.1825,
+      "step": 4353
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.763091868683216e-07,
+      "loss": 1.1075,
+      "step": 4354
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.745760101392683e-07,
+      "loss": 1.1656,
+      "step": 4355
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.72844207040215e-07,
+      "loss": 1.1045,
+      "step": 4356
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.711137781621233e-07,
+      "loss": 1.1225,
+      "step": 4357
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.693847240954795e-07,
+      "loss": 1.1605,
+      "step": 4358
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.676570454303042e-07,
+      "loss": 1.1052,
+      "step": 4359
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.659307427561493e-07,
+      "loss": 1.1041,
+      "step": 4360
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.642058166620943e-07,
+      "loss": 1.172,
+      "step": 4361
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.624822677367508e-07,
+      "loss": 1.1115,
+      "step": 4362
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.607600965682611e-07,
+      "loss": 1.1708,
+      "step": 4363
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.590393037442963e-07,
+      "loss": 1.2657,
+      "step": 4364
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.573198898520552e-07,
+      "loss": 1.216,
+      "step": 4365
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.556018554782715e-07,
+      "loss": 1.0959,
+      "step": 4366
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.53885201209203e-07,
+      "loss": 1.2298,
+      "step": 4367
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.521699276306379e-07,
+      "loss": 0.948,
+      "step": 4368
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.504560353278953e-07,
+      "loss": 1.1567,
+      "step": 4369
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.487435248858206e-07,
+      "loss": 1.0775,
+      "step": 4370
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.470323968887868e-07,
+      "loss": 1.2627,
+      "step": 4371
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.453226519206998e-07,
+      "loss": 1.0654,
+      "step": 4372
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.436142905649892e-07,
+      "loss": 1.0442,
+      "step": 4373
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.419073134046125e-07,
+      "loss": 1.0321,
+      "step": 4374
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.402017210220588e-07,
+      "loss": 1.0863,
+      "step": 4375
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.384975139993396e-07,
+      "loss": 1.1078,
+      "step": 4376
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.367946929179978e-07,
+      "loss": 1.0641,
+      "step": 4377
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.350932583591016e-07,
+      "loss": 1.0669,
+      "step": 4378
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.333932109032439e-07,
+      "loss": 1.1173,
+      "step": 4379
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.316945511305497e-07,
+      "loss": 1.1127,
+      "step": 4380
+    },
+    {
+      "epoch": 0.8,
+      "learning_rate": 9.299972796206653e-07,
+      "loss": 1.3228,
+      "step": 4381
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.283013969527648e-07,
+      "loss": 1.0927,
+      "step": 4382
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.266069037055508e-07,
+      "loss": 1.1518,
+      "step": 4383
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.249138004572483e-07,
+      "loss": 1.0162,
+      "step": 4384
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.232220877856091e-07,
+      "loss": 1.0965,
+      "step": 4385
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.215317662679124e-07,
+      "loss": 1.1058,
+      "step": 4386
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.198428364809608e-07,
+      "loss": 1.0649,
+      "step": 4387
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.181552990010801e-07,
+      "loss": 1.1077,
+      "step": 4388
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.164691544041259e-07,
+      "loss": 1.1181,
+      "step": 4389
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.147844032654757e-07,
+      "loss": 1.3082,
+      "step": 4390
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.131010461600287e-07,
+      "loss": 1.0977,
+      "step": 4391
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.114190836622144e-07,
+      "loss": 1.1011,
+      "step": 4392
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.097385163459827e-07,
+      "loss": 1.0687,
+      "step": 4393
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.080593447848062e-07,
+      "loss": 1.3296,
+      "step": 4394
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.063815695516859e-07,
+      "loss": 1.0603,
+      "step": 4395
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.047051912191401e-07,
+      "loss": 0.9715,
+      "step": 4396
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.03030210359217e-07,
+      "loss": 1.0676,
+      "step": 4397
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 9.013566275434837e-07,
+      "loss": 1.1283,
+      "step": 4398
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.996844433430291e-07,
+      "loss": 1.2155,
+      "step": 4399
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.980136583284704e-07,
+      "loss": 1.1638,
+      "step": 4400
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.963442730699423e-07,
+      "loss": 1.1445,
+      "step": 4401
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.946762881371019e-07,
+      "loss": 1.1472,
+      "step": 4402
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.930097040991326e-07,
+      "loss": 1.08,
+      "step": 4403
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.913445215247363e-07,
+      "loss": 0.9354,
+      "step": 4404
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.896807409821362e-07,
+      "loss": 1.0932,
+      "step": 4405
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.88018363039081e-07,
+      "loss": 1.0211,
+      "step": 4406
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.863573882628362e-07,
+      "loss": 1.0683,
+      "step": 4407
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.846978172201903e-07,
+      "loss": 1.363,
+      "step": 4408
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.830396504774552e-07,
+      "loss": 1.2601,
+      "step": 4409
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.8138288860046e-07,
+      "loss": 1.0677,
+      "step": 4410
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.79727532154554e-07,
+      "loss": 1.1776,
+      "step": 4411
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.780735817046116e-07,
+      "loss": 0.9863,
+      "step": 4412
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.76421037815024e-07,
+      "loss": 1.0592,
+      "step": 4413
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.747699010497008e-07,
+      "loss": 0.9999,
+      "step": 4414
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.731201719720761e-07,
+      "loss": 1.1784,
+      "step": 4415
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.714718511450998e-07,
+      "loss": 1.1184,
+      "step": 4416
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.698249391312424e-07,
+      "loss": 1.181,
+      "step": 4417
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.681794364924951e-07,
+      "loss": 1.1452,
+      "step": 4418
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.66535343790365e-07,
+      "loss": 1.1974,
+      "step": 4419
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.648926615858821e-07,
+      "loss": 1.0963,
+      "step": 4420
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.632513904395922e-07,
+      "loss": 1.2193,
+      "step": 4421
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.616115309115586e-07,
+      "loss": 1.3241,
+      "step": 4422
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.599730835613679e-07,
+      "loss": 1.1535,
+      "step": 4423
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.58336048948119e-07,
+      "loss": 0.921,
+      "step": 4424
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.567004276304313e-07,
+      "loss": 1.0319,
+      "step": 4425
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.550662201664439e-07,
+      "loss": 1.1338,
+      "step": 4426
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.534334271138101e-07,
+      "loss": 1.0753,
+      "step": 4427
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.518020490297013e-07,
+      "loss": 1.135,
+      "step": 4428
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.501720864708079e-07,
+      "loss": 1.0495,
+      "step": 4429
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.485435399933362e-07,
+      "loss": 1.2423,
+      "step": 4430
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.469164101530065e-07,
+      "loss": 1.0941,
+      "step": 4431
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.452906975050617e-07,
+      "loss": 1.0216,
+      "step": 4432
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.436664026042562e-07,
+      "loss": 0.9882,
+      "step": 4433
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.42043526004861e-07,
+      "loss": 1.073,
+      "step": 4434
+    },
+    {
+      "epoch": 0.81,
+      "learning_rate": 8.404220682606662e-07,
+      "loss": 1.0661,
+      "step": 4435
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.388020299249755e-07,
+      "loss": 1.1564,
+      "step": 4436
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.371834115506061e-07,
+      "loss": 1.198,
+      "step": 4437
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.355662136898962e-07,
+      "loss": 1.0795,
+      "step": 4438
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.339504368946949e-07,
+      "loss": 1.2258,
+      "step": 4439
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.323360817163656e-07,
+      "loss": 1.1447,
+      "step": 4440
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.307231487057921e-07,
+      "loss": 1.0504,
+      "step": 4441
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.291116384133663e-07,
+      "loss": 1.226,
+      "step": 4442
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.275015513890005e-07,
+      "loss": 1.1137,
+      "step": 4443
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.258928881821171e-07,
+      "loss": 1.128,
+      "step": 4444
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.242856493416529e-07,
+      "loss": 1.0383,
+      "step": 4445
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.22679835416062e-07,
+      "loss": 1.1065,
+      "step": 4446
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.210754469533088e-07,
+      "loss": 1.1497,
+      "step": 4447
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.194724845008723e-07,
+      "loss": 1.0642,
+      "step": 4448
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.178709486057462e-07,
+      "loss": 1.0312,
+      "step": 4449
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.162708398144365e-07,
+      "loss": 1.0783,
+      "step": 4450
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.146721586729594e-07,
+      "loss": 1.2076,
+      "step": 4451
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.130749057268505e-07,
+      "loss": 1.1465,
+      "step": 4452
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.114790815211521e-07,
+      "loss": 1.2077,
+      "step": 4453
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.098846866004195e-07,
+      "loss": 1.169,
+      "step": 4454
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.082917215087249e-07,
+      "loss": 1.0287,
+      "step": 4455
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.067001867896479e-07,
+      "loss": 1.0369,
+      "step": 4456
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.051100829862802e-07,
+      "loss": 1.1144,
+      "step": 4457
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.035214106412292e-07,
+      "loss": 1.0571,
+      "step": 4458
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.019341702966105e-07,
+      "loss": 1.1071,
+      "step": 4459
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 8.003483624940494e-07,
+      "loss": 1.27,
+      "step": 4460
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.987639877746883e-07,
+      "loss": 1.0791,
+      "step": 4461
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.971810466791735e-07,
+      "loss": 1.095,
+      "step": 4462
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.95599539747669e-07,
+      "loss": 1.0796,
+      "step": 4463
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.940194675198443e-07,
+      "loss": 1.2655,
+      "step": 4464
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.924408305348796e-07,
+      "loss": 1.1371,
+      "step": 4465
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.908636293314697e-07,
+      "loss": 1.2627,
+      "step": 4466
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.892878644478152e-07,
+      "loss": 1.232,
+      "step": 4467
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.87713536421626e-07,
+      "loss": 1.1407,
+      "step": 4468
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.861406457901267e-07,
+      "loss": 1.1249,
+      "step": 4469
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.845691930900467e-07,
+      "loss": 1.1393,
+      "step": 4470
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.829991788576258e-07,
+      "loss": 1.193,
+      "step": 4471
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.814306036286151e-07,
+      "loss": 0.9926,
+      "step": 4472
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.798634679382716e-07,
+      "loss": 1.0903,
+      "step": 4473
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.78297772321362e-07,
+      "loss": 1.0836,
+      "step": 4474
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.767335173121643e-07,
+      "loss": 1.1871,
+      "step": 4475
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.751707034444611e-07,
+      "loss": 1.0506,
+      "step": 4476
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.736093312515435e-07,
+      "loss": 1.0371,
+      "step": 4477
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.72049401266215e-07,
+      "loss": 1.1232,
+      "step": 4478
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.704909140207828e-07,
+      "loss": 1.0536,
+      "step": 4479
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.689338700470605e-07,
+      "loss": 1.135,
+      "step": 4480
+    },
+    {
+      "epoch": 0.82,
+      "eval_loss": 1.9354063272476196,
+      "eval_runtime": 0.1827,
+      "eval_samples_per_second": 16.42,
+      "eval_steps_per_second": 16.42,
+      "step": 4480
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.673782698763754e-07,
+      "loss": 1.1384,
+      "step": 4481
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.658241140395555e-07,
+      "loss": 1.138,
+      "step": 4482
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.642714030669396e-07,
+      "loss": 1.2934,
+      "step": 4483
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.627201374883731e-07,
+      "loss": 1.0754,
+      "step": 4484
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.611703178332064e-07,
+      "loss": 1.0633,
+      "step": 4485
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.596219446303e-07,
+      "loss": 1.2045,
+      "step": 4486
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.580750184080171e-07,
+      "loss": 1.083,
+      "step": 4487
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.565295396942273e-07,
+      "loss": 1.0527,
+      "step": 4488
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.549855090163105e-07,
+      "loss": 1.0703,
+      "step": 4489
+    },
+    {
+      "epoch": 0.82,
+      "learning_rate": 7.534429269011478e-07,
+      "loss": 1.1096,
+      "step": 4490
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.519017938751272e-07,
+      "loss": 1.1419,
+      "step": 4491
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.503621104641445e-07,
+      "loss": 1.0771,
+      "step": 4492
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.48823877193598e-07,
+      "loss": 1.0423,
+      "step": 4493
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.472870945883914e-07,
+      "loss": 1.1067,
+      "step": 4494
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.457517631729367e-07,
+      "loss": 1.2035,
+      "step": 4495
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.442178834711467e-07,
+      "loss": 1.003,
+      "step": 4496
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.426854560064389e-07,
+      "loss": 1.1402,
+      "step": 4497
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.411544813017397e-07,
+      "loss": 1.0652,
+      "step": 4498
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.396249598794753e-07,
+      "loss": 1.3016,
+      "step": 4499
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.380968922615761e-07,
+      "loss": 1.1002,
+      "step": 4500
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.365702789694801e-07,
+      "loss": 0.9795,
+      "step": 4501
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.350451205241255e-07,
+      "loss": 0.9638,
+      "step": 4502
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.335214174459543e-07,
+      "loss": 1.0267,
+      "step": 4503
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.319991702549151e-07,
+      "loss": 1.0971,
+      "step": 4504
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.304783794704561e-07,
+      "loss": 1.0754,
+      "step": 4505
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.289590456115287e-07,
+      "loss": 0.9757,
+      "step": 4506
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2744116919659e-07,
+      "loss": 0.9994,
+      "step": 4507
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.259247507435963e-07,
+      "loss": 1.2219,
+      "step": 4508
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.244097907700104e-07,
+      "loss": 0.9853,
+      "step": 4509
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.228962897927938e-07,
+      "loss": 1.0623,
+      "step": 4510
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.2138424832841e-07,
+      "loss": 1.2032,
+      "step": 4511
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.198736668928291e-07,
+      "loss": 1.2379,
+      "step": 4512
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.183645460015176e-07,
+      "loss": 1.249,
+      "step": 4513
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.168568861694453e-07,
+      "loss": 1.0847,
+      "step": 4514
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.153506879110861e-07,
+      "loss": 1.0395,
+      "step": 4515
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.138459517404112e-07,
+      "loss": 1.1937,
+      "step": 4516
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.123426781708941e-07,
+      "loss": 1.2998,
+      "step": 4517
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.108408677155121e-07,
+      "loss": 0.9112,
+      "step": 4518
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.093405208867399e-07,
+      "loss": 1.1122,
+      "step": 4519
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.07841638196552e-07,
+      "loss": 1.0997,
+      "step": 4520
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.063442201564274e-07,
+      "loss": 1.038,
+      "step": 4521
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.048482672773422e-07,
+      "loss": 1.0994,
+      "step": 4522
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.033537800697715e-07,
+      "loss": 1.0743,
+      "step": 4523
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.018607590436949e-07,
+      "loss": 1.1072,
+      "step": 4524
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 7.003692047085875e-07,
+      "loss": 1.0867,
+      "step": 4525
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.988791175734239e-07,
+      "loss": 1.044,
+      "step": 4526
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.973904981466812e-07,
+      "loss": 1.0715,
+      "step": 4527
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.959033469363324e-07,
+      "loss": 0.9721,
+      "step": 4528
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.944176644498524e-07,
+      "loss": 1.201,
+      "step": 4529
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.929334511942132e-07,
+      "loss": 1.1958,
+      "step": 4530
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.91450707675883e-07,
+      "loss": 1.1004,
+      "step": 4531
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.899694344008351e-07,
+      "loss": 0.9796,
+      "step": 4532
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.884896318745349e-07,
+      "loss": 1.1418,
+      "step": 4533
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.870113006019469e-07,
+      "loss": 1.1043,
+      "step": 4534
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.855344410875375e-07,
+      "loss": 1.1918,
+      "step": 4535
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.840590538352671e-07,
+      "loss": 1.0774,
+      "step": 4536
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.825851393485933e-07,
+      "loss": 0.878,
+      "step": 4537
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.81112698130475e-07,
+      "loss": 1.0793,
+      "step": 4538
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.796417306833647e-07,
+      "loss": 1.0185,
+      "step": 4539
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.78172237509212e-07,
+      "loss": 1.0972,
+      "step": 4540
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.767042191094674e-07,
+      "loss": 1.1186,
+      "step": 4541
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.752376759850743e-07,
+      "loss": 1.1984,
+      "step": 4542
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.737726086364727e-07,
+      "loss": 1.1636,
+      "step": 4543
+    },
+    {
+      "epoch": 0.83,
+      "learning_rate": 6.723090175636015e-07,
+      "loss": 1.078,
+      "step": 4544
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.708469032658943e-07,
+      "loss": 1.0716,
+      "step": 4545
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.69386266242279e-07,
+      "loss": 1.242,
+      "step": 4546
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.679271069911847e-07,
+      "loss": 0.9601,
+      "step": 4547
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.664694260105304e-07,
+      "loss": 1.0877,
+      "step": 4548
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.650132237977331e-07,
+      "loss": 1.1039,
+      "step": 4549
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.635585008497064e-07,
+      "loss": 1.2015,
+      "step": 4550
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.621052576628567e-07,
+      "loss": 1.0454,
+      "step": 4551
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.606534947330878e-07,
+      "loss": 1.1118,
+      "step": 4552
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.592032125557974e-07,
+      "loss": 1.1053,
+      "step": 4553
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.577544116258755e-07,
+      "loss": 0.9794,
+      "step": 4554
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.563070924377119e-07,
+      "loss": 1.2196,
+      "step": 4555
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.548612554851858e-07,
+      "loss": 1.1366,
+      "step": 4556
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.534169012616721e-07,
+      "loss": 1.0611,
+      "step": 4557
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.519740302600425e-07,
+      "loss": 1.1775,
+      "step": 4558
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.505326429726588e-07,
+      "loss": 1.1889,
+      "step": 4559
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.490927398913771e-07,
+      "loss": 1.1359,
+      "step": 4560
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.476543215075498e-07,
+      "loss": 1.1765,
+      "step": 4561
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.462173883120199e-07,
+      "loss": 1.0027,
+      "step": 4562
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.447819407951228e-07,
+      "loss": 1.0272,
+      "step": 4563
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.433479794466919e-07,
+      "loss": 1.0737,
+      "step": 4564
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.419155047560482e-07,
+      "loss": 1.2382,
+      "step": 4565
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.404845172120066e-07,
+      "loss": 1.0299,
+      "step": 4566
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.390550173028775e-07,
+      "loss": 1.1713,
+      "step": 4567
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.376270055164597e-07,
+      "loss": 1.0631,
+      "step": 4568
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.362004823400458e-07,
+      "loss": 1.1074,
+      "step": 4569
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.347754482604218e-07,
+      "loss": 1.1224,
+      "step": 4570
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.333519037638636e-07,
+      "loss": 1.0701,
+      "step": 4571
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.319298493361392e-07,
+      "loss": 1.1974,
+      "step": 4572
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.305092854625095e-07,
+      "loss": 1.0711,
+      "step": 4573
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.290902126277237e-07,
+      "loss": 1.1686,
+      "step": 4574
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.27672631316027e-07,
+      "loss": 1.0086,
+      "step": 4575
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.26256542011151e-07,
+      "loss": 1.2556,
+      "step": 4576
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.248419451963195e-07,
+      "loss": 1.0801,
+      "step": 4577
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.234288413542495e-07,
+      "loss": 1.1015,
+      "step": 4578
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.220172309671457e-07,
+      "loss": 1.2225,
+      "step": 4579
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.206071145167025e-07,
+      "loss": 1.0341,
+      "step": 4580
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.191984924841088e-07,
+      "loss": 1.0834,
+      "step": 4581
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.17791365350039e-07,
+      "loss": 0.9962,
+      "step": 4582
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.163857335946588e-07,
+      "loss": 1.2302,
+      "step": 4583
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.149815976976259e-07,
+      "loss": 1.0891,
+      "step": 4584
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.135789581380841e-07,
+      "loss": 1.1824,
+      "step": 4585
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.121778153946678e-07,
+      "loss": 1.1736,
+      "step": 4586
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.107781699455023e-07,
+      "loss": 1.0762,
+      "step": 4587
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.093800222682006e-07,
+      "loss": 1.2392,
+      "step": 4588
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.079833728398627e-07,
+      "loss": 0.9304,
+      "step": 4589
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.065882221370817e-07,
+      "loss": 1.0533,
+      "step": 4590
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.051945706359364e-07,
+      "loss": 1.2287,
+      "step": 4591
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.038024188119923e-07,
+      "loss": 1.0917,
+      "step": 4592
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.024117671403079e-07,
+      "loss": 1.1547,
+      "step": 4593
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 6.010226160954258e-07,
+      "loss": 1.1341,
+      "step": 4594
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.996349661513789e-07,
+      "loss": 1.159,
+      "step": 4595
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.982488177816865e-07,
+      "loss": 1.1614,
+      "step": 4596
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.968641714593549e-07,
+      "loss": 1.1068,
+      "step": 4597
+    },
+    {
+      "epoch": 0.84,
+      "learning_rate": 5.954810276568807e-07,
+      "loss": 1.0081,
+      "step": 4598
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.940993868462447e-07,
+      "loss": 1.1979,
+      "step": 4599
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.927192494989159e-07,
+      "loss": 1.0524,
+      "step": 4600
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.913406160858515e-07,
+      "loss": 1.0411,
+      "step": 4601
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.89963487077494e-07,
+      "loss": 1.1467,
+      "step": 4602
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.885878629437714e-07,
+      "loss": 1.1077,
+      "step": 4603
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.872137441541026e-07,
+      "loss": 1.0412,
+      "step": 4604
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.858411311773882e-07,
+      "loss": 1.009,
+      "step": 4605
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.844700244820161e-07,
+      "loss": 1.0967,
+      "step": 4606
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.831004245358629e-07,
+      "loss": 1.2519,
+      "step": 4607
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.817323318062873e-07,
+      "loss": 0.9483,
+      "step": 4608
+    },
+    {
+      "epoch": 0.85,
+      "eval_loss": 1.9343419075012207,
+      "eval_runtime": 0.1751,
+      "eval_samples_per_second": 17.133,
+      "eval_steps_per_second": 17.133,
+      "step": 4608
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.803657467601348e-07,
+      "loss": 1.0709,
+      "step": 4609
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.790006698637391e-07,
+      "loss": 1.1601,
+      "step": 4610
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.776371015829158e-07,
+      "loss": 1.2068,
+      "step": 4611
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.762750423829667e-07,
+      "loss": 1.1394,
+      "step": 4612
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.749144927286798e-07,
+      "loss": 1.0327,
+      "step": 4613
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.735554530843268e-07,
+      "loss": 1.0946,
+      "step": 4614
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.721979239136638e-07,
+      "loss": 1.1854,
+      "step": 4615
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.70841905679933e-07,
+      "loss": 1.0768,
+      "step": 4616
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.6948739884586e-07,
+      "loss": 1.0131,
+      "step": 4617
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.681344038736547e-07,
+      "loss": 1.1958,
+      "step": 4618
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.66782921225012e-07,
+      "loss": 1.1822,
+      "step": 4619
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.654329513611078e-07,
+      "loss": 1.0182,
+      "step": 4620
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.640844947426061e-07,
+      "loss": 1.1259,
+      "step": 4621
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.627375518296518e-07,
+      "loss": 1.2315,
+      "step": 4622
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.613921230818719e-07,
+      "loss": 1.0027,
+      "step": 4623
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.60048208958382e-07,
+      "loss": 1.1103,
+      "step": 4624
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.58705809917775e-07,
+      "loss": 1.0931,
+      "step": 4625
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.573649264181296e-07,
+      "loss": 1.0851,
+      "step": 4626
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.560255589170077e-07,
+      "loss": 1.2562,
+      "step": 4627
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.54687707871453e-07,
+      "loss": 1.2144,
+      "step": 4628
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.533513737379909e-07,
+      "loss": 1.0758,
+      "step": 4629
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.520165569726316e-07,
+      "loss": 1.0391,
+      "step": 4630
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.506832580308652e-07,
+      "loss": 1.0967,
+      "step": 4631
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.493514773676644e-07,
+      "loss": 1.1811,
+      "step": 4632
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.480212154374859e-07,
+      "loss": 1.18,
+      "step": 4633
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.466924726942652e-07,
+      "loss": 1.1672,
+      "step": 4634
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.453652495914191e-07,
+      "loss": 1.1162,
+      "step": 4635
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.440395465818504e-07,
+      "loss": 1.1618,
+      "step": 4636
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.427153641179372e-07,
+      "loss": 1.1566,
+      "step": 4637
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.413927026515447e-07,
+      "loss": 1.0773,
+      "step": 4638
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.400715626340142e-07,
+      "loss": 1.1601,
+      "step": 4639
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.387519445161687e-07,
+      "loss": 1.194,
+      "step": 4640
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.37433848748315e-07,
+      "loss": 1.3191,
+      "step": 4641
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.361172757802369e-07,
+      "loss": 1.0911,
+      "step": 4642
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.348022260611996e-07,
+      "loss": 1.1604,
+      "step": 4643
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.334887000399502e-07,
+      "loss": 1.1638,
+      "step": 4644
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.321766981647142e-07,
+      "loss": 1.0083,
+      "step": 4645
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.308662208831949e-07,
+      "loss": 1.0486,
+      "step": 4646
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.295572686425815e-07,
+      "loss": 1.1305,
+      "step": 4647
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.282498418895366e-07,
+      "loss": 1.0976,
+      "step": 4648
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.269439410702043e-07,
+      "loss": 1.2068,
+      "step": 4649
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.2563956663021e-07,
+      "loss": 1.2603,
+      "step": 4650
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.243367190146559e-07,
+      "loss": 1.0779,
+      "step": 4651
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.230353986681231e-07,
+      "loss": 1.1147,
+      "step": 4652
+    },
+    {
+      "epoch": 0.85,
+      "learning_rate": 5.217356060346734e-07,
+      "loss": 1.118,
+      "step": 4653
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.204373415578451e-07,
+      "loss": 1.0794,
+      "step": 4654
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.19140605680658e-07,
+      "loss": 1.0629,
+      "step": 4655
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.178453988456073e-07,
+      "loss": 1.1944,
+      "step": 4656
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.165517214946663e-07,
+      "loss": 1.1489,
+      "step": 4657
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.1525957406929e-07,
+      "loss": 1.0898,
+      "step": 4658
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.139689570104078e-07,
+      "loss": 1.083,
+      "step": 4659
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.126798707584269e-07,
+      "loss": 0.9245,
+      "step": 4660
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.113923157532358e-07,
+      "loss": 1.1938,
+      "step": 4661
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.101062924341959e-07,
+      "loss": 1.1744,
+      "step": 4662
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.088218012401486e-07,
+      "loss": 0.9776,
+      "step": 4663
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.075388426094119e-07,
+      "loss": 1.1728,
+      "step": 4664
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.062574169797818e-07,
+      "loss": 1.037,
+      "step": 4665
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.049775247885275e-07,
+      "loss": 1.08,
+      "step": 4666
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.036991664724005e-07,
+      "loss": 1.1067,
+      "step": 4667
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.02422342467625e-07,
+      "loss": 1.0711,
+      "step": 4668
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 5.011470532099011e-07,
+      "loss": 1.178,
+      "step": 4669
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.998732991344091e-07,
+      "loss": 1.0655,
+      "step": 4670
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.986010806758024e-07,
+      "loss": 1.105,
+      "step": 4671
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.973303982682093e-07,
+      "loss": 1.1642,
+      "step": 4672
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.96061252345238e-07,
+      "loss": 1.1975,
+      "step": 4673
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.947936433399675e-07,
+      "loss": 1.2524,
+      "step": 4674
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.93527571684958e-07,
+      "loss": 1.0895,
+      "step": 4675
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.922630378122406e-07,
+      "loss": 1.1034,
+      "step": 4676
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.910000421533212e-07,
+      "loss": 1.1996,
+      "step": 4677
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.897385851391862e-07,
+      "loss": 1.227,
+      "step": 4678
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.884786672002912e-07,
+      "loss": 1.0636,
+      "step": 4679
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.872202887665678e-07,
+      "loss": 1.042,
+      "step": 4680
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.859634502674259e-07,
+      "loss": 1.2044,
+      "step": 4681
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.847081521317465e-07,
+      "loss": 1.1532,
+      "step": 4682
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.83454394787884e-07,
+      "loss": 1.0446,
+      "step": 4683
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.82202178663671e-07,
+      "loss": 1.1595,
+      "step": 4684
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.80951504186411e-07,
+      "loss": 1.0576,
+      "step": 4685
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.797023717828814e-07,
+      "loss": 1.1284,
+      "step": 4686
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.784547818793367e-07,
+      "loss": 1.2039,
+      "step": 4687
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.772087349015014e-07,
+      "loss": 1.1517,
+      "step": 4688
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7596423127457434e-07,
+      "loss": 1.0926,
+      "step": 4689
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7472127142322913e-07,
+      "loss": 0.9831,
+      "step": 4690
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7347985577161025e-07,
+      "loss": 1.1016,
+      "step": 4691
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.72239984743339e-07,
+      "loss": 1.3405,
+      "step": 4692
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.7100165876150517e-07,
+      "loss": 1.139,
+      "step": 4693
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.697648782486736e-07,
+      "loss": 1.1516,
+      "step": 4694
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.685296436268827e-07,
+      "loss": 1.2376,
+      "step": 4695
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.672959553176415e-07,
+      "loss": 1.1011,
+      "step": 4696
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.66063813741931e-07,
+      "loss": 1.0515,
+      "step": 4697
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.648332193202076e-07,
+      "loss": 1.1864,
+      "step": 4698
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.636041724723961e-07,
+      "loss": 1.1539,
+      "step": 4699
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.6237667361789417e-07,
+      "loss": 1.0802,
+      "step": 4700
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.611507231755741e-07,
+      "loss": 1.0701,
+      "step": 4701
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.599263215637756e-07,
+      "loss": 1.3369,
+      "step": 4702
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5870346920031115e-07,
+      "loss": 1.0932,
+      "step": 4703
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.574821665024676e-07,
+      "loss": 1.151,
+      "step": 4704
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5626241388699866e-07,
+      "loss": 1.169,
+      "step": 4705
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.5504421177013094e-07,
+      "loss": 1.1452,
+      "step": 4706
+    },
+    {
+      "epoch": 0.86,
+      "learning_rate": 4.538275605675635e-07,
+      "loss": 1.0218,
+      "step": 4707
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.526124606944632e-07,
+      "loss": 1.0625,
+      "step": 4708
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5139891256547e-07,
+      "loss": 1.1999,
+      "step": 4709
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.5018691659469317e-07,
+      "loss": 1.2136,
+      "step": 4710
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4897647319571136e-07,
+      "loss": 1.2151,
+      "step": 4711
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4776758278157584e-07,
+      "loss": 0.9835,
+      "step": 4712
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4656024576480716e-07,
+      "loss": 1.0171,
+      "step": 4713
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.45354462557393e-07,
+      "loss": 1.2988,
+      "step": 4714
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4415023357079535e-07,
+      "loss": 1.0742,
+      "step": 4715
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4294755921594323e-07,
+      "loss": 1.0331,
+      "step": 4716
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.4174643990323395e-07,
+      "loss": 1.1258,
+      "step": 4717
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.40546876042538e-07,
+      "loss": 0.9502,
+      "step": 4718
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3934886804319174e-07,
+      "loss": 1.0833,
+      "step": 4719
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3815241631400106e-07,
+      "loss": 1.093,
+      "step": 4720
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3695752126324267e-07,
+      "loss": 1.0777,
+      "step": 4721
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3576418329866086e-07,
+      "loss": 0.9775,
+      "step": 4722
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3457240282746726e-07,
+      "loss": 1.0988,
+      "step": 4723
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.3338218025634483e-07,
+      "loss": 0.9636,
+      "step": 4724
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.32193515991442e-07,
+      "loss": 1.1165,
+      "step": 4725
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.310064104383782e-07,
+      "loss": 1.3553,
+      "step": 4726
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2982086400224e-07,
+      "loss": 1.2252,
+      "step": 4727
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.286368770875793e-07,
+      "loss": 1.0605,
+      "step": 4728
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.274544500984207e-07,
+      "loss": 0.9746,
+      "step": 4729
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.262735834382531e-07,
+      "loss": 0.9616,
+      "step": 4730
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.2509427751003195e-07,
+      "loss": 1.0871,
+      "step": 4731
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.239165327161848e-07,
+      "loss": 1.1719,
+      "step": 4732
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.227403494586019e-07,
+      "loss": 1.0434,
+      "step": 4733
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.215657281386426e-07,
+      "loss": 1.0362,
+      "step": 4734
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.203926691571336e-07,
+      "loss": 1.1125,
+      "step": 4735
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.192211729143675e-07,
+      "loss": 1.024,
+      "step": 4736
+    },
+    {
+      "epoch": 0.87,
+      "eval_loss": 1.9344431161880493,
+      "eval_runtime": 0.1755,
+      "eval_samples_per_second": 17.092,
+      "eval_steps_per_second": 17.092,
+      "step": 4736
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1805123981010274e-07,
+      "loss": 1.0527,
+      "step": 4737
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1688287024356845e-07,
+      "loss": 1.0495,
+      "step": 4738
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.157160646134556e-07,
+      "loss": 0.97,
+      "step": 4739
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.145508233179224e-07,
+      "loss": 0.9871,
+      "step": 4740
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.1338714675459634e-07,
+      "loss": 1.1024,
+      "step": 4741
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.122250353205681e-07,
+      "loss": 1.1406,
+      "step": 4742
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.110644894123933e-07,
+      "loss": 1.1817,
+      "step": 4743
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.099055094260973e-07,
+      "loss": 1.1031,
+      "step": 4744
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.087480957571671e-07,
+      "loss": 1.0217,
+      "step": 4745
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.075922488005579e-07,
+      "loss": 1.0213,
+      "step": 4746
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.064379689506892e-07,
+      "loss": 1.1507,
+      "step": 4747
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.052852566014448e-07,
+      "loss": 1.1165,
+      "step": 4748
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.041341121461762e-07,
+      "loss": 1.1725,
+      "step": 4749
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0298453597769796e-07,
+      "loss": 1.1035,
+      "step": 4750
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0183652848828803e-07,
+      "loss": 1.1695,
+      "step": 4751
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 4.0069009006969306e-07,
+      "loss": 1.1065,
+      "step": 4752
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9954522111312123e-07,
+      "loss": 1.2635,
+      "step": 4753
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9840192200924553e-07,
+      "loss": 1.2987,
+      "step": 4754
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.972601931482051e-07,
+      "loss": 1.1093,
+      "step": 4755
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9612003491960037e-07,
+      "loss": 1.1098,
+      "step": 4756
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9498144771249745e-07,
+      "loss": 1.1731,
+      "step": 4757
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9384443191542775e-07,
+      "loss": 1.1652,
+      "step": 4758
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9270898791638425e-07,
+      "loss": 1.0175,
+      "step": 4759
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9157511610282253e-07,
+      "loss": 0.9525,
+      "step": 4760
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.9044281686166586e-07,
+      "loss": 1.1912,
+      "step": 4761
+    },
+    {
+      "epoch": 0.87,
+      "learning_rate": 3.893120905792963e-07,
+      "loss": 1.1559,
+      "step": 4762
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8818293764156355e-07,
+      "loss": 1.1331,
+      "step": 4763
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.870553584337772e-07,
+      "loss": 1.0215,
+      "step": 4764
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8592935334070937e-07,
+      "loss": 1.0315,
+      "step": 4765
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.8480492274659884e-07,
+      "loss": 1.1521,
+      "step": 4766
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.836820670351432e-07,
+      "loss": 1.0501,
+      "step": 4767
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.825607865895037e-07,
+      "loss": 1.1167,
+      "step": 4768
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.814410817923064e-07,
+      "loss": 1.0537,
+      "step": 4769
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.803229530256364e-07,
+      "loss": 1.0419,
+      "step": 4770
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7920640067104167e-07,
+      "loss": 0.9761,
+      "step": 4771
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7809142510953523e-07,
+      "loss": 1.0854,
+      "step": 4772
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7697802672158834e-07,
+      "loss": 1.2274,
+      "step": 4773
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7586620588713486e-07,
+      "loss": 1.128,
+      "step": 4774
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7475596298557235e-07,
+      "loss": 1.1549,
+      "step": 4775
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.736472983957579e-07,
+      "loss": 1.2037,
+      "step": 4776
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.725402124960098e-07,
+      "loss": 1.0818,
+      "step": 4777
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.714347056641104e-07,
+      "loss": 1.1343,
+      "step": 4778
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.7033077827729966e-07,
+      "loss": 1.1066,
+      "step": 4779
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6922843071228123e-07,
+      "loss": 1.0378,
+      "step": 4780
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6812766334521854e-07,
+      "loss": 1.0607,
+      "step": 4781
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6702847655173455e-07,
+      "loss": 1.1955,
+      "step": 4782
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.659308707069159e-07,
+      "loss": 1.3908,
+      "step": 4783
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.648348461853074e-07,
+      "loss": 1.3557,
+      "step": 4784
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.637404033609143e-07,
+      "loss": 1.0137,
+      "step": 4785
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6264754260720405e-07,
+      "loss": 1.0294,
+      "step": 4786
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6155626429710153e-07,
+      "loss": 1.2487,
+      "step": 4787
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.6046656880299347e-07,
+      "loss": 1.2117,
+      "step": 4788
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.593784564967273e-07,
+      "loss": 1.0187,
+      "step": 4789
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.582919277496072e-07,
+      "loss": 1.075,
+      "step": 4790
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5720698293239875e-07,
+      "loss": 1.1532,
+      "step": 4791
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5612362241532863e-07,
+      "loss": 1.1463,
+      "step": 4792
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.5504184656807996e-07,
+      "loss": 1.1283,
+      "step": 4793
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.539616557597958e-07,
+      "loss": 1.0825,
+      "step": 4794
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.528830503590808e-07,
+      "loss": 1.1011,
+      "step": 4795
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.518060307339954e-07,
+      "loss": 1.022,
+      "step": 4796
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.507305972520597e-07,
+      "loss": 1.2681,
+      "step": 4797
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.49656750280255e-07,
+      "loss": 1.2264,
+      "step": 4798
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.485844901850177e-07,
+      "loss": 1.0352,
+      "step": 4799
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4751381733224565e-07,
+      "loss": 1.1491,
+      "step": 4800
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4644473208729266e-07,
+      "loss": 0.979,
+      "step": 4801
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4537723481497197e-07,
+      "loss": 1.1054,
+      "step": 4802
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.443113258795555e-07,
+      "loss": 1.1811,
+      "step": 4803
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4324700564477233e-07,
+      "loss": 1.0497,
+      "step": 4804
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.4218427447380853e-07,
+      "loss": 1.1667,
+      "step": 4805
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.411231327293107e-07,
+      "loss": 1.0528,
+      "step": 4806
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.400635807733804e-07,
+      "loss": 1.0443,
+      "step": 4807
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.390056189675772e-07,
+      "loss": 1.0919,
+      "step": 4808
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.379492476729196e-07,
+      "loss": 1.1697,
+      "step": 4809
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.368944672498814e-07,
+      "loss": 1.151,
+      "step": 4810
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.358412780583942e-07,
+      "loss": 1.1739,
+      "step": 4811
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.3478968045784767e-07,
+      "loss": 1.1846,
+      "step": 4812
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.3373967480708757e-07,
+      "loss": 1.0664,
+      "step": 4813
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.326912614644151e-07,
+      "loss": 1.1571,
+      "step": 4814
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.3164444078759075e-07,
+      "loss": 0.9694,
+      "step": 4815
+    },
+    {
+      "epoch": 0.88,
+      "learning_rate": 3.3059921313382815e-07,
+      "loss": 1.0797,
+      "step": 4816
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2955557885980207e-07,
+      "loss": 1.1195,
+      "step": 4817
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2851353832163926e-07,
+      "loss": 1.0543,
+      "step": 4818
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.274730918749236e-07,
+      "loss": 1.0944,
+      "step": 4819
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.264342398746967e-07,
+      "loss": 1.0787,
+      "step": 4820
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.253969826754544e-07,
+      "loss": 1.0986,
+      "step": 4821
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2436132063114855e-07,
+      "loss": 1.1059,
+      "step": 4822
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2332725409518815e-07,
+      "loss": 1.2493,
+      "step": 4823
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.222947834204354e-07,
+      "loss": 1.0898,
+      "step": 4824
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2126390895920954e-07,
+      "loss": 1.0365,
+      "step": 4825
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.2023463106328535e-07,
+      "loss": 1.1216,
+      "step": 4826
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1920695008389127e-07,
+      "loss": 1.0034,
+      "step": 4827
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1818086637171175e-07,
+      "loss": 1.1772,
+      "step": 4828
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1715638027688677e-07,
+      "loss": 1.0391,
+      "step": 4829
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.161334921490106e-07,
+      "loss": 1.1189,
+      "step": 4830
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1511220233713056e-07,
+      "loss": 1.1562,
+      "step": 4831
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1409251118975246e-07,
+      "loss": 1.1023,
+      "step": 4832
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.130744190548324e-07,
+      "loss": 1.2117,
+      "step": 4833
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.1205792627978403e-07,
+      "loss": 1.1787,
+      "step": 4834
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.110430332114733e-07,
+      "loss": 0.9929,
+      "step": 4835
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.100297401962199e-07,
+      "loss": 1.0378,
+      "step": 4836
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0901804757980046e-07,
+      "loss": 1.1892,
+      "step": 4837
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.080079557074428e-07,
+      "loss": 0.9608,
+      "step": 4838
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.069994649238284e-07,
+      "loss": 0.9758,
+      "step": 4839
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0599257557309415e-07,
+      "loss": 1.0183,
+      "step": 4840
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.049872879988297e-07,
+      "loss": 1.0445,
+      "step": 4841
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.039836025440762e-07,
+      "loss": 1.1452,
+      "step": 4842
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0298151955133246e-07,
+      "loss": 1.1401,
+      "step": 4843
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0198103936254605e-07,
+      "loss": 1.2102,
+      "step": 4844
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 3.0098216231912003e-07,
+      "loss": 1.1313,
+      "step": 4845
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9998488876191e-07,
+      "loss": 1.129,
+      "step": 4846
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.989892190312238e-07,
+      "loss": 1.2382,
+      "step": 4847
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.979951534668218e-07,
+      "loss": 1.0682,
+      "step": 4848
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.970026924079189e-07,
+      "loss": 1.1412,
+      "step": 4849
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9601183619318074e-07,
+      "loss": 1.1363,
+      "step": 4850
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9502258516072414e-07,
+      "loss": 1.1276,
+      "step": 4851
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9403493964812236e-07,
+      "loss": 1.0939,
+      "step": 4852
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9304889999239526e-07,
+      "loss": 1.0474,
+      "step": 4853
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.920644665300204e-07,
+      "loss": 1.1852,
+      "step": 4854
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.910816395969229e-07,
+      "loss": 1.187,
+      "step": 4855
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.9010041952848066e-07,
+      "loss": 1.0525,
+      "step": 4856
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.891208066595258e-07,
+      "loss": 1.1771,
+      "step": 4857
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.881428013243387e-07,
+      "loss": 1.1495,
+      "step": 4858
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.871664038566524e-07,
+      "loss": 1.1837,
+      "step": 4859
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8619161458965206e-07,
+      "loss": 1.1226,
+      "step": 4860
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8521843385597326e-07,
+      "loss": 1.037,
+      "step": 4861
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8424686198770147e-07,
+      "loss": 1.1799,
+      "step": 4862
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8327689931637703e-07,
+      "loss": 1.1351,
+      "step": 4863
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8230854617298685e-07,
+      "loss": 1.1704,
+      "step": 4864
+    },
+    {
+      "epoch": 0.89,
+      "eval_loss": 1.9335230588912964,
+      "eval_runtime": 0.1874,
+      "eval_samples_per_second": 16.005,
+      "eval_steps_per_second": 16.005,
+      "step": 4864
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8134180288796997e-07,
+      "loss": 1.0028,
+      "step": 4865
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.8037666979121857e-07,
+      "loss": 1.2754,
+      "step": 4866
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7941314721207145e-07,
+      "loss": 1.152,
+      "step": 4867
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7845123547932005e-07,
+      "loss": 1.1264,
+      "step": 4868
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.774909349212068e-07,
+      "loss": 1.0159,
+      "step": 4869
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.765322458654218e-07,
+      "loss": 1.1287,
+      "step": 4870
+    },
+    {
+      "epoch": 0.89,
+      "learning_rate": 2.7557516863910783e-07,
+      "loss": 1.1202,
+      "step": 4871
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7461970356885694e-07,
+      "loss": 1.2332,
+      "step": 4872
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.7366585098070885e-07,
+      "loss": 0.9706,
+      "step": 4873
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.727136112001566e-07,
+      "loss": 1.2018,
+      "step": 4874
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.717629845521408e-07,
+      "loss": 1.1476,
+      "step": 4875
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.708139713610508e-07,
+      "loss": 1.1701,
+      "step": 4876
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6986657195072817e-07,
+      "loss": 1.1004,
+      "step": 4877
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.689207866444615e-07,
+      "loss": 1.0998,
+      "step": 4878
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.679766157649882e-07,
+      "loss": 1.0571,
+      "step": 4879
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.670340596344978e-07,
+      "loss": 1.1422,
+      "step": 4880
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.660931185746257e-07,
+      "loss": 1.0566,
+      "step": 4881
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.651537929064563e-07,
+      "loss": 1.1386,
+      "step": 4882
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.642160829505258e-07,
+      "loss": 1.0127,
+      "step": 4883
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.632799890268162e-07,
+      "loss": 1.1672,
+      "step": 4884
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.6234551145475795e-07,
+      "loss": 1.2343,
+      "step": 4885
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.614126505532322e-07,
+      "loss": 1.1362,
+      "step": 4886
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.604814066405659e-07,
+      "loss": 1.2018,
+      "step": 4887
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.595517800345365e-07,
+      "loss": 1.0425,
+      "step": 4888
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.586237710523687e-07,
+      "loss": 1.0194,
+      "step": 4889
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.57697380010733e-07,
+      "loss": 1.1225,
+      "step": 4890
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5677260722575146e-07,
+      "loss": 1.0335,
+      "step": 4891
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5584945301299235e-07,
+      "loss": 1.0161,
+      "step": 4892
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.549279176874703e-07,
+      "loss": 1.1556,
+      "step": 4893
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.540080015636498e-07,
+      "loss": 1.193,
+      "step": 4894
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.530897049554415e-07,
+      "loss": 1.0378,
+      "step": 4895
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.5217302817620306e-07,
+      "loss": 1.1156,
+      "step": 4896
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.512579715387409e-07,
+      "loss": 1.1742,
+      "step": 4897
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.503445353553069e-07,
+      "loss": 1.2095,
+      "step": 4898
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4943271993760067e-07,
+      "loss": 1.2734,
+      "step": 4899
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.485225255967694e-07,
+      "loss": 1.0477,
+      "step": 4900
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4761395264340625e-07,
+      "loss": 1.1688,
+      "step": 4901
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4670700138755113e-07,
+      "loss": 1.146,
+      "step": 4902
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4580167213869134e-07,
+      "loss": 1.0182,
+      "step": 4903
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4489796520576035e-07,
+      "loss": 1.225,
+      "step": 4904
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.43995880897136e-07,
+      "loss": 1.1686,
+      "step": 4905
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4309541952064696e-07,
+      "loss": 1.0295,
+      "step": 4906
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.4219658138356306e-07,
+      "loss": 1.0393,
+      "step": 4907
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.412993667926039e-07,
+      "loss": 1.1005,
+      "step": 4908
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.404037760539341e-07,
+      "loss": 1.2151,
+      "step": 4909
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.395098094731618e-07,
+      "loss": 1.2377,
+      "step": 4910
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3861746735534485e-07,
+      "loss": 1.1727,
+      "step": 4911
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3772675000498457e-07,
+      "loss": 1.1472,
+      "step": 4912
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3683765772602674e-07,
+      "loss": 1.0284,
+      "step": 4913
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3595019082186531e-07,
+      "loss": 1.0365,
+      "step": 4914
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3506434959533807e-07,
+      "loss": 1.2638,
+      "step": 4915
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.341801343487271e-07,
+      "loss": 1.1516,
+      "step": 4916
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3329754538376214e-07,
+      "loss": 1.0577,
+      "step": 4917
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3241658300161617e-07,
+      "loss": 1.1756,
+      "step": 4918
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.3153724750290706e-07,
+      "loss": 1.1617,
+      "step": 4919
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.306595391876987e-07,
+      "loss": 1.0717,
+      "step": 4920
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.297834583554992e-07,
+      "loss": 1.0487,
+      "step": 4921
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.289090053052595e-07,
+      "loss": 1.1758,
+      "step": 4922
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.2803618033537865e-07,
+      "loss": 1.0916,
+      "step": 4923
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.271649837436979e-07,
+      "loss": 0.9863,
+      "step": 4924
+    },
+    {
+      "epoch": 0.9,
+      "learning_rate": 2.2629541582750158e-07,
+      "loss": 1.3499,
+      "step": 4925
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2542747688352183e-07,
+      "loss": 1.0507,
+      "step": 4926
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.245611672079312e-07,
+      "loss": 0.9792,
+      "step": 4927
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2369648709634983e-07,
+      "loss": 1.1073,
+      "step": 4928
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.228334368438384e-07,
+      "loss": 0.9822,
+      "step": 4929
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.2197201674490298e-07,
+      "loss": 1.0602,
+      "step": 4930
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.211122270934951e-07,
+      "loss": 0.9842,
+      "step": 4931
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.202540681830062e-07,
+      "loss": 1.1129,
+      "step": 4932
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1939754030627368e-07,
+      "loss": 1.1146,
+      "step": 4933
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1854264375557878e-07,
+      "loss": 1.0729,
+      "step": 4934
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1768937882264417e-07,
+      "loss": 1.1429,
+      "step": 4935
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1683774579863703e-07,
+      "loss": 1.0656,
+      "step": 4936
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1598774497416763e-07,
+      "loss": 1.1531,
+      "step": 4937
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1513937663928953e-07,
+      "loss": 1.1241,
+      "step": 4938
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1429264108349725e-07,
+      "loss": 1.1488,
+      "step": 4939
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.134475385957313e-07,
+      "loss": 1.097,
+      "step": 4940
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1260406946437274e-07,
+      "loss": 1.1528,
+      "step": 4941
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.117622339772446e-07,
+      "loss": 1.1263,
+      "step": 4942
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1092203242161546e-07,
+      "loss": 1.0886,
+      "step": 4943
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.1008346508419376e-07,
+      "loss": 1.1805,
+      "step": 4944
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0924653225113e-07,
+      "loss": 1.1288,
+      "step": 4945
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0841123420801967e-07,
+      "loss": 1.2069,
+      "step": 4946
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0757757123989753e-07,
+      "loss": 1.1046,
+      "step": 4947
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0674554363124268e-07,
+      "loss": 1.1871,
+      "step": 4948
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0591515166597353e-07,
+      "loss": 1.1312,
+      "step": 4949
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.050863956274529e-07,
+      "loss": 1.0829,
+      "step": 4950
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0425927579848403e-07,
+      "loss": 1.1948,
+      "step": 4951
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0343379246131278e-07,
+      "loss": 1.0974,
+      "step": 4952
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.026099458976244e-07,
+      "loss": 1.1047,
+      "step": 4953
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0178773638854841e-07,
+      "loss": 1.1685,
+      "step": 4954
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.0096716421465434e-07,
+      "loss": 1.1485,
+      "step": 4955
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 2.001482296559526e-07,
+      "loss": 1.1775,
+      "step": 4956
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9933093299189521e-07,
+      "loss": 1.1779,
+      "step": 4957
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9851527450137631e-07,
+      "loss": 1.005,
+      "step": 4958
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9770125446272825e-07,
+      "loss": 0.9993,
+      "step": 4959
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.968888731537283e-07,
+      "loss": 1.1714,
+      "step": 4960
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9607813085159076e-07,
+      "loss": 1.2208,
+      "step": 4961
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9526902783297164e-07,
+      "loss": 1.1958,
+      "step": 4962
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9446156437397001e-07,
+      "loss": 1.1122,
+      "step": 4963
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9365574075012272e-07,
+      "loss": 1.0003,
+      "step": 4964
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9285155723640758e-07,
+      "loss": 0.9318,
+      "step": 4965
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9204901410724342e-07,
+      "loss": 1.2532,
+      "step": 4966
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9124811163648837e-07,
+      "loss": 0.9957,
+      "step": 4967
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.9044885009744164e-07,
+      "loss": 1.1567,
+      "step": 4968
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8965122976284223e-07,
+      "loss": 1.0676,
+      "step": 4969
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8885525090486856e-07,
+      "loss": 1.1433,
+      "step": 4970
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8806091379514e-07,
+      "loss": 1.1024,
+      "step": 4971
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.872682187047148e-07,
+      "loss": 1.1027,
+      "step": 4972
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8647716590409038e-07,
+      "loss": 1.0449,
+      "step": 4973
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8568775566320586e-07,
+      "loss": 1.1097,
+      "step": 4974
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.848999882514374e-07,
+      "loss": 1.2579,
+      "step": 4975
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.841138639376011e-07,
+      "loss": 1.1697,
+      "step": 4976
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.833293829899546e-07,
+      "loss": 1.2454,
+      "step": 4977
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.8254654567619267e-07,
+      "loss": 1.0518,
+      "step": 4978
+    },
+    {
+      "epoch": 0.91,
+      "learning_rate": 1.817653522634477e-07,
+      "loss": 1.0349,
+      "step": 4979
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8098580301829539e-07,
+      "loss": 1.1009,
+      "step": 4980
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.8020789820674734e-07,
+      "loss": 1.1857,
+      "step": 4981
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7943163809425401e-07,
+      "loss": 1.1208,
+      "step": 4982
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7865702294570576e-07,
+      "loss": 1.123,
+      "step": 4983
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7788405302543165e-07,
+      "loss": 1.0088,
+      "step": 4984
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.771127285971974e-07,
+      "loss": 1.2082,
+      "step": 4985
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7634304992421025e-07,
+      "loss": 1.2522,
+      "step": 4986
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7557501726911398e-07,
+      "loss": 1.131,
+      "step": 4987
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7480863089398948e-07,
+      "loss": 1.1211,
+      "step": 4988
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.740438910603598e-07,
+      "loss": 1.2049,
+      "step": 4989
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7328079802918175e-07,
+      "loss": 1.0384,
+      "step": 4990
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7251935206085258e-07,
+      "loss": 1.0413,
+      "step": 4991
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7175955341520779e-07,
+      "loss": 0.9585,
+      "step": 4992
+    },
+    {
+      "epoch": 0.92,
+      "eval_loss": 1.9331599473953247,
+      "eval_runtime": 0.1839,
+      "eval_samples_per_second": 16.313,
+      "eval_steps_per_second": 16.313,
+      "step": 4992
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.710014023515183e-07,
+      "loss": 1.0037,
+      "step": 4993
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.7024489912849662e-07,
+      "loss": 1.0098,
+      "step": 4994
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6949004400429013e-07,
+      "loss": 1.0781,
+      "step": 4995
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6873683723648337e-07,
+      "loss": 1.1987,
+      "step": 4996
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6798527908210072e-07,
+      "loss": 1.2055,
+      "step": 4997
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6723536979760314e-07,
+      "loss": 0.9958,
+      "step": 4998
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6648710963888704e-07,
+      "loss": 1.1207,
+      "step": 4999
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6574049886128929e-07,
+      "loss": 1.068,
+      "step": 5000
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6499553771958167e-07,
+      "loss": 1.1974,
+      "step": 5001
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.64252226467973e-07,
+      "loss": 1.185,
+      "step": 5002
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.63510565360111e-07,
+      "loss": 1.2273,
+      "step": 5003
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6277055464907877e-07,
+      "loss": 1.1736,
+      "step": 5004
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.620321945873954e-07,
+      "loss": 1.0817,
+      "step": 5005
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6129548542701877e-07,
+      "loss": 1.1724,
+      "step": 5006
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.6056042741934287e-07,
+      "loss": 1.1022,
+      "step": 5007
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5982702081519696e-07,
+      "loss": 1.3244,
+      "step": 5008
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5909526586484813e-07,
+      "loss": 1.1627,
+      "step": 5009
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5836516281799942e-07,
+      "loss": 1.0536,
+      "step": 5010
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.576367119237898e-07,
+      "loss": 1.1144,
+      "step": 5011
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5690991343079553e-07,
+      "loss": 1.0834,
+      "step": 5012
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5618476758702705e-07,
+      "loss": 1.0347,
+      "step": 5013
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5546127463993422e-07,
+      "loss": 1.0962,
+      "step": 5014
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5473943483639898e-07,
+      "loss": 1.2961,
+      "step": 5015
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.540192484227404e-07,
+      "loss": 1.1859,
+      "step": 5016
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5330071564471582e-07,
+      "loss": 1.1682,
+      "step": 5017
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5258383674751576e-07,
+      "loss": 1.0433,
+      "step": 5018
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5186861197576564e-07,
+      "loss": 1.3215,
+      "step": 5019
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.5115504157352968e-07,
+      "loss": 1.2417,
+      "step": 5020
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.504431257843042e-07,
+      "loss": 1.1071,
+      "step": 5021
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4973286485102268e-07,
+      "loss": 1.2218,
+      "step": 5022
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4902425901605399e-07,
+      "loss": 1.1569,
+      "step": 5023
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4831730852120142e-07,
+      "loss": 1.3177,
+      "step": 5024
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.476120136077036e-07,
+      "loss": 1.0333,
+      "step": 5025
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4690837451623475e-07,
+      "loss": 1.1844,
+      "step": 5026
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4620639148690386e-07,
+      "loss": 1.0744,
+      "step": 5027
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4550606475925378e-07,
+      "loss": 1.1428,
+      "step": 5028
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4480739457226388e-07,
+      "loss": 1.0635,
+      "step": 5029
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4411038116434738e-07,
+      "loss": 1.1916,
+      "step": 5030
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.434150247733507e-07,
+      "loss": 1.1922,
+      "step": 5031
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4272132563655837e-07,
+      "loss": 1.0143,
+      "step": 5032
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4202928399068616e-07,
+      "loss": 1.1969,
+      "step": 5033
+    },
+    {
+      "epoch": 0.92,
+      "learning_rate": 1.4133890007188512e-07,
+      "loss": 1.0837,
+      "step": 5034
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.4065017411574179e-07,
+      "loss": 1.1153,
+      "step": 5035
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3996310635727538e-07,
+      "loss": 1.1037,
+      "step": 5036
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3927769703094007e-07,
+      "loss": 1.0949,
+      "step": 5037
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.385939463706243e-07,
+      "loss": 0.9419,
+      "step": 5038
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3791185460964974e-07,
+      "loss": 1.084,
+      "step": 5039
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3723142198077245e-07,
+      "loss": 1.1866,
+      "step": 5040
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3655264871618278e-07,
+      "loss": 1.1574,
+      "step": 5041
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3587553504750327e-07,
+      "loss": 1.2068,
+      "step": 5042
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3520008120579242e-07,
+      "loss": 1.1282,
+      "step": 5043
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3452628742154138e-07,
+      "loss": 1.1542,
+      "step": 5044
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3385415392467294e-07,
+      "loss": 1.1921,
+      "step": 5045
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3318368094454636e-07,
+      "loss": 1.1676,
+      "step": 5046
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3251486870995257e-07,
+      "loss": 1.0537,
+      "step": 5047
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3184771744911506e-07,
+      "loss": 1.0685,
+      "step": 5048
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.31182227389694e-07,
+      "loss": 1.0784,
+      "step": 5049
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.3051839875877825e-07,
+      "loss": 1.0661,
+      "step": 5050
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2985623178289163e-07,
+      "loss": 1.1885,
+      "step": 5051
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2919572668799285e-07,
+      "loss": 1.2156,
+      "step": 5052
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2853688369947047e-07,
+      "loss": 1.2231,
+      "step": 5053
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2787970304214636e-07,
+      "loss": 1.0198,
+      "step": 5054
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2722418494027778e-07,
+      "loss": 1.0832,
+      "step": 5055
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.265703296175519e-07,
+      "loss": 1.1286,
+      "step": 5056
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2591813729708913e-07,
+      "loss": 1.2491,
+      "step": 5057
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2526760820144314e-07,
+      "loss": 1.0942,
+      "step": 5058
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2461874255259853e-07,
+      "loss": 1.0423,
+      "step": 5059
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2397154057197548e-07,
+      "loss": 1.1796,
+      "step": 5060
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2332600248042225e-07,
+      "loss": 1.2338,
+      "step": 5061
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2268212849822158e-07,
+      "loss": 1.0738,
+      "step": 5062
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2203991884508937e-07,
+      "loss": 1.2977,
+      "step": 5063
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2139937374017086e-07,
+      "loss": 1.1431,
+      "step": 5064
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2076049340204566e-07,
+      "loss": 1.1935,
+      "step": 5065
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.2012327804872438e-07,
+      "loss": 1.0087,
+      "step": 5066
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1948772789764918e-07,
+      "loss": 0.9687,
+      "step": 5067
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1885384316569382e-07,
+      "loss": 1.1233,
+      "step": 5068
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1822162406916527e-07,
+      "loss": 1.1574,
+      "step": 5069
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1759107082380095e-07,
+      "loss": 1.112,
+      "step": 5070
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1696218364476875e-07,
+      "loss": 1.2161,
+      "step": 5071
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1633496274667088e-07,
+      "loss": 1.0566,
+      "step": 5072
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1570940834353838e-07,
+      "loss": 1.1129,
+      "step": 5073
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1508552064883382e-07,
+      "loss": 1.1776,
+      "step": 5074
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1446329987545357e-07,
+      "loss": 1.1531,
+      "step": 5075
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1384274623572223e-07,
+      "loss": 1.0268,
+      "step": 5076
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1322385994139651e-07,
+      "loss": 1.0816,
+      "step": 5077
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1260664120366527e-07,
+      "loss": 1.0664,
+      "step": 5078
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1199109023314613e-07,
+      "loss": 1.0224,
+      "step": 5079
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1137720723988999e-07,
+      "loss": 1.2416,
+      "step": 5080
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1076499243337702e-07,
+      "loss": 1.0764,
+      "step": 5081
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.1015444602251735e-07,
+      "loss": 1.1921,
+      "step": 5082
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0954556821565487e-07,
+      "loss": 1.0009,
+      "step": 5083
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0893835922056173e-07,
+      "loss": 1.0484,
+      "step": 5084
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0833281924443939e-07,
+      "loss": 1.1742,
+      "step": 5085
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0772894849392367e-07,
+      "loss": 1.1004,
+      "step": 5086
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0712674717507754e-07,
+      "loss": 1.2038,
+      "step": 5087
+    },
+    {
+      "epoch": 0.93,
+      "learning_rate": 1.0652621549339493e-07,
+      "loss": 1.1739,
+      "step": 5088
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0592735365380136e-07,
+      "loss": 1.094,
+      "step": 5089
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0533016186065115e-07,
+      "loss": 1.0629,
+      "step": 5090
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0473464031772907e-07,
+      "loss": 1.0117,
+      "step": 5091
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0414078922825033e-07,
+      "loss": 1.1162,
+      "step": 5092
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0354860879485951e-07,
+      "loss": 1.2243,
+      "step": 5093
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0295809921963107e-07,
+      "loss": 1.0801,
+      "step": 5094
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0236926070407106e-07,
+      "loss": 1.134,
+      "step": 5095
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0178209344911316e-07,
+      "loss": 1.179,
+      "step": 5096
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0119659765512046e-07,
+      "loss": 0.9813,
+      "step": 5097
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0061277352188814e-07,
+      "loss": 1.2317,
+      "step": 5098
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 1.0003062124863905e-07,
+      "loss": 1.147,
+      "step": 5099
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.94501410340254e-08,
+      "loss": 1.1866,
+      "step": 5100
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.887133307613095e-08,
+      "loss": 1.1292,
+      "step": 5101
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.82941975724655e-08,
+      "loss": 1.1052,
+      "step": 5102
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.771873471997151e-08,
+      "loss": 1.1803,
+      "step": 5103
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.714494471501801e-08,
+      "loss": 1.1109,
+      "step": 5104
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.657282775340449e-08,
+      "loss": 1.0808,
+      "step": 5105
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.600238403036033e-08,
+      "loss": 1.1963,
+      "step": 5106
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.54336137405415e-08,
+      "loss": 1.1403,
+      "step": 5107
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.486651707803496e-08,
+      "loss": 0.9615,
+      "step": 5108
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.430109423635703e-08,
+      "loss": 1.0948,
+      "step": 5109
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.373734540845003e-08,
+      "loss": 1.1114,
+      "step": 5110
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.317527078668731e-08,
+      "loss": 1.0662,
+      "step": 5111
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.261487056287156e-08,
+      "loss": 1.0631,
+      "step": 5112
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.205614492823145e-08,
+      "loss": 1.1314,
+      "step": 5113
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.149909407342672e-08,
+      "loss": 1.0772,
+      "step": 5114
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.09437181885442e-08,
+      "loss": 1.1438,
+      "step": 5115
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 9.039001746310006e-08,
+      "loss": 1.1963,
+      "step": 5116
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.983799208603705e-08,
+      "loss": 1.0932,
+      "step": 5117
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.928764224572894e-08,
+      "loss": 1.128,
+      "step": 5118
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.873896812997607e-08,
+      "loss": 1.0468,
+      "step": 5119
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.819196992600643e-08,
+      "loss": 1.0307,
+      "step": 5120
+    },
+    {
+      "epoch": 0.94,
+      "eval_loss": 1.934967041015625,
+      "eval_runtime": 0.1713,
+      "eval_samples_per_second": 17.518,
+      "eval_steps_per_second": 17.518,
+      "step": 5120
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.764664782047794e-08,
+      "loss": 1.0744,
+      "step": 5121
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.710300199947508e-08,
+      "loss": 0.8733,
+      "step": 5122
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.656103264851057e-08,
+      "loss": 1.1164,
+      "step": 5123
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.602073995252592e-08,
+      "loss": 1.108,
+      "step": 5124
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.548212409588863e-08,
+      "loss": 1.0901,
+      "step": 5125
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.49451852623967e-08,
+      "loss": 1.1447,
+      "step": 5126
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.440992363527356e-08,
+      "loss": 1.2283,
+      "step": 5127
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.387633939717032e-08,
+      "loss": 1.079,
+      "step": 5128
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.334443273016857e-08,
+      "loss": 1.1386,
+      "step": 5129
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.281420381577366e-08,
+      "loss": 1.0448,
+      "step": 5130
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.228565283492085e-08,
+      "loss": 1.0644,
+      "step": 5131
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.175877996797254e-08,
+      "loss": 1.1022,
+      "step": 5132
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.123358539471716e-08,
+      "loss": 1.0593,
+      "step": 5133
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.07100692943713e-08,
+      "loss": 1.035,
+      "step": 5134
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 8.018823184557988e-08,
+      "loss": 1.1471,
+      "step": 5135
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.96680732264138e-08,
+      "loss": 1.2198,
+      "step": 5136
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.914959361436991e-08,
+      "loss": 1.1321,
+      "step": 5137
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.863279318637507e-08,
+      "loss": 1.0967,
+      "step": 5138
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.811767211878097e-08,
+      "loss": 1.0262,
+      "step": 5139
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.760423058736643e-08,
+      "loss": 1.0779,
+      "step": 5140
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.7092468767338e-08,
+      "loss": 1.0358,
+      "step": 5141
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.65823868333282e-08,
+      "loss": 1.1451,
+      "step": 5142
+    },
+    {
+      "epoch": 0.94,
+      "learning_rate": 7.607398495939666e-08,
+      "loss": 1.0948,
+      "step": 5143
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.556726331903019e-08,
+      "loss": 1.0174,
+      "step": 5144
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.506222208514102e-08,
+      "loss": 0.9958,
+      "step": 5145
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.455886143006963e-08,
+      "loss": 1.0458,
+      "step": 5146
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.405718152558194e-08,
+      "loss": 1.1946,
+      "step": 5147
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.355718254286937e-08,
+      "loss": 1.1547,
+      "step": 5148
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.305886465255208e-08,
+      "loss": 1.1733,
+      "step": 5149
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.256222802467516e-08,
+      "loss": 1.1276,
+      "step": 5150
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.206727282870973e-08,
+      "loss": 1.1478,
+      "step": 5151
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.157399923355401e-08,
+      "loss": 1.0731,
+      "step": 5152
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.10824074075317e-08,
+      "loss": 0.9279,
+      "step": 5153
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.059249751839304e-08,
+      "loss": 1.1285,
+      "step": 5154
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 7.01042697333143e-08,
+      "loss": 1.1452,
+      "step": 5155
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.961772421889723e-08,
+      "loss": 1.115,
+      "step": 5156
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.913286114117068e-08,
+      "loss": 1.2448,
+      "step": 5157
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.864968066558786e-08,
+      "loss": 1.1258,
+      "step": 5158
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.81681829570291e-08,
+      "loss": 1.0955,
+      "step": 5159
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.768836817979962e-08,
+      "loss": 1.1265,
+      "step": 5160
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.721023649763125e-08,
+      "loss": 1.2051,
+      "step": 5161
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.673378807368069e-08,
+      "loss": 1.1041,
+      "step": 5162
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.625902307053012e-08,
+      "loss": 1.1473,
+      "step": 5163
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.578594165018826e-08,
+      "loss": 1.1472,
+      "step": 5164
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.531454397408876e-08,
+      "loss": 1.0369,
+      "step": 5165
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.484483020309018e-08,
+      "loss": 1.0535,
+      "step": 5166
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.437680049747763e-08,
+      "loss": 1.1207,
+      "step": 5167
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.391045501696003e-08,
+      "loss": 1.1933,
+      "step": 5168
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.344579392067341e-08,
+      "loss": 1.1476,
+      "step": 5169
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.29828173671776e-08,
+      "loss": 1.2303,
+      "step": 5170
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.25215255144579e-08,
+      "loss": 1.0199,
+      "step": 5171
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.206191851992505e-08,
+      "loss": 1.0546,
+      "step": 5172
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.160399654041526e-08,
+      "loss": 1.1754,
+      "step": 5173
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.114775973218801e-08,
+      "loss": 1.0809,
+      "step": 5174
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.069320825093039e-08,
+      "loss": 1.1755,
+      "step": 5175
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 6.024034225175168e-08,
+      "loss": 1.0684,
+      "step": 5176
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.978916188918715e-08,
+      "loss": 1.1099,
+      "step": 5177
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.933966731719809e-08,
+      "loss": 1.1945,
+      "step": 5178
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.889185868916847e-08,
+      "loss": 1.1106,
+      "step": 5179
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.8445736157907715e-08,
+      "loss": 1.0927,
+      "step": 5180
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.800129987565128e-08,
+      "loss": 1.2135,
+      "step": 5181
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.755854999405674e-08,
+      "loss": 1.1905,
+      "step": 5182
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.7117486664207686e-08,
+      "loss": 1.0933,
+      "step": 5183
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.667811003661261e-08,
+      "loss": 1.0269,
+      "step": 5184
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.6240420261203243e-08,
+      "loss": 1.0273,
+      "step": 5185
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.580441748733567e-08,
+      "loss": 1.0135,
+      "step": 5186
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.537010186379199e-08,
+      "loss": 1.1329,
+      "step": 5187
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4937473538776984e-08,
+      "loss": 1.1531,
+      "step": 5188
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.4506532659919785e-08,
+      "loss": 0.9926,
+      "step": 5189
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.407727937427498e-08,
+      "loss": 1.1025,
+      "step": 5190
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.364971382831874e-08,
+      "loss": 1.1166,
+      "step": 5191
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.3223836167954346e-08,
+      "loss": 1.2099,
+      "step": 5192
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.2799646538507777e-08,
+      "loss": 1.1798,
+      "step": 5193
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.237714508472769e-08,
+      "loss": 1.0699,
+      "step": 5194
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1956331950788754e-08,
+      "loss": 0.9742,
+      "step": 5195
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.153720728028833e-08,
+      "loss": 1.0406,
+      "step": 5196
+    },
+    {
+      "epoch": 0.95,
+      "learning_rate": 5.1119771216248114e-08,
+      "loss": 1.0975,
+      "step": 5197
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.070402390111362e-08,
+      "loss": 1.1837,
+      "step": 5198
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 5.0289965476753574e-08,
+      "loss": 1.2164,
+      "step": 5199
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.987759608446052e-08,
+      "loss": 0.984,
+      "step": 5200
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.946691586495134e-08,
+      "loss": 1.2499,
+      "step": 5201
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.9057924958365036e-08,
+      "loss": 1.0497,
+      "step": 5202
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.865062350426608e-08,
+      "loss": 1.2588,
+      "step": 5203
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.824501164164108e-08,
+      "loss": 1.1378,
+      "step": 5204
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.784108950890043e-08,
+      "loss": 1.0961,
+      "step": 5205
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.7438857243877756e-08,
+      "loss": 1.0972,
+      "step": 5206
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.703831498383049e-08,
+      "loss": 1.2188,
+      "step": 5207
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.66394628654393e-08,
+      "loss": 1.1345,
+      "step": 5208
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.624230102480698e-08,
+      "loss": 1.0672,
+      "step": 5209
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5846829597461785e-08,
+      "loss": 1.0521,
+      "step": 5210
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5453048718352985e-08,
+      "loss": 1.2101,
+      "step": 5211
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.5060958521854215e-08,
+      "loss": 1.0702,
+      "step": 5212
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.467055914176177e-08,
+      "loss": 1.109,
+      "step": 5213
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.4281850711294095e-08,
+      "loss": 1.2537,
+      "step": 5214
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.389483336309508e-08,
+      "loss": 1.2632,
+      "step": 5215
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.350950722922853e-08,
+      "loss": 1.0603,
+      "step": 5216
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.3125872441183156e-08,
+      "loss": 1.184,
+      "step": 5217
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2743929129870357e-08,
+      "loss": 1.0951,
+      "step": 5218
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.2363677425623085e-08,
+      "loss": 1.0013,
+      "step": 5219
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1985117458198664e-08,
+      "loss": 1.2149,
+      "step": 5220
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.160824935677599e-08,
+      "loss": 1.2058,
+      "step": 5221
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.1233073249957176e-08,
+      "loss": 1.1118,
+      "step": 5222
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0859589265766495e-08,
+      "loss": 1.2942,
+      "step": 5223
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.048779753165144e-08,
+      "loss": 1.0045,
+      "step": 5224
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 4.0117698174481076e-08,
+      "loss": 1.084,
+      "step": 5225
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.974929132054828e-08,
+      "loss": 1.1194,
+      "step": 5226
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.938257709556803e-08,
+      "loss": 1.1371,
+      "step": 5227
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.901755562467635e-08,
+      "loss": 1.1359,
+      "step": 5228
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.865422703243249e-08,
+      "loss": 1.2537,
+      "step": 5229
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.829259144281949e-08,
+      "loss": 1.0875,
+      "step": 5230
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7932648979240874e-08,
+      "loss": 1.1663,
+      "step": 5231
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7574399764522264e-08,
+      "loss": 1.2494,
+      "step": 5232
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.7217843920912546e-08,
+      "loss": 1.0883,
+      "step": 5233
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.6862981570082166e-08,
+      "loss": 1.1075,
+      "step": 5234
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.650981283312427e-08,
+      "loss": 1.2147,
+      "step": 5235
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.615833783055356e-08,
+      "loss": 1.1229,
+      "step": 5236
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.580855668230632e-08,
+      "loss": 0.9915,
+      "step": 5237
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.546046950774207e-08,
+      "loss": 1.1019,
+      "step": 5238
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.511407642564135e-08,
+      "loss": 1.2662,
+      "step": 5239
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.476937755420684e-08,
+      "loss": 1.1199,
+      "step": 5240
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4426373011062776e-08,
+      "loss": 1.2196,
+      "step": 5241
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.4085062913256085e-08,
+      "loss": 1.0439,
+      "step": 5242
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.374544737725416e-08,
+      "loss": 1.1667,
+      "step": 5243
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.3407526518948187e-08,
+      "loss": 1.1934,
+      "step": 5244
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.30713004536487e-08,
+      "loss": 0.9905,
+      "step": 5245
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.273676929608893e-08,
+      "loss": 1.1185,
+      "step": 5246
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.240393316042478e-08,
+      "loss": 1.1023,
+      "step": 5247
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.207279216023207e-08,
+      "loss": 1.1325,
+      "step": 5248
+    },
+    {
+      "epoch": 0.96,
+      "eval_loss": 1.9319905042648315,
+      "eval_runtime": 0.1829,
+      "eval_samples_per_second": 16.403,
+      "eval_steps_per_second": 16.403,
+      "step": 5248
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.174334640850874e-08,
+      "loss": 1.1046,
+      "step": 5249
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.141559601767541e-08,
+      "loss": 1.1369,
+      "step": 5250
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.1089541099572076e-08,
+      "loss": 1.1399,
+      "step": 5251
+    },
+    {
+      "epoch": 0.96,
+      "learning_rate": 3.076518176546139e-08,
+      "loss": 0.9154,
+      "step": 5252
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.044251812602816e-08,
+      "loss": 1.0387,
+      "step": 5253
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 3.0121550291376534e-08,
+      "loss": 1.0213,
+      "step": 5254
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.980227837103389e-08,
+      "loss": 1.0507,
+      "step": 5255
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.948470247394808e-08,
+      "loss": 0.9713,
+      "step": 5256
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.9168822708487422e-08,
+      "loss": 1.2224,
+      "step": 5257
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8854639182443465e-08,
+      "loss": 1.1232,
+      "step": 5258
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8542152003027124e-08,
+      "loss": 1.0563,
+      "step": 5259
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.8231361276870873e-08,
+      "loss": 1.0547,
+      "step": 5260
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.7922267110028765e-08,
+      "loss": 1.2645,
+      "step": 5261
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.761486960797588e-08,
+      "loss": 1.0966,
+      "step": 5262
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.730916887560775e-08,
+      "loss": 1.0335,
+      "step": 5263
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.700516501724204e-08,
+      "loss": 1.1533,
+      "step": 5264
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6702858136615216e-08,
+      "loss": 1.1634,
+      "step": 5265
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6402248336886983e-08,
+      "loss": 0.991,
+      "step": 5266
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.6103335720636947e-08,
+      "loss": 0.9664,
+      "step": 5267
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5806120389865742e-08,
+      "loss": 0.9132,
+      "step": 5268
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.5510602445993904e-08,
+      "loss": 1.1905,
+      "step": 5269
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.521678198986466e-08,
+      "loss": 0.9476,
+      "step": 5270
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4924659121740026e-08,
+      "loss": 1.0875,
+      "step": 5271
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4634233941304153e-08,
+      "loss": 1.1685,
+      "step": 5272
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4345506547660547e-08,
+      "loss": 0.9633,
+      "step": 5273
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.4058477039335392e-08,
+      "loss": 0.9833,
+      "step": 5274
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3773145514273123e-08,
+      "loss": 1.1084,
+      "step": 5275
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3489512069840292e-08,
+      "loss": 1.1722,
+      "step": 5276
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.3207576802823928e-08,
+      "loss": 1.1405,
+      "step": 5277
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2927339809430405e-08,
+      "loss": 1.0781,
+      "step": 5278
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2648801185288228e-08,
+      "loss": 1.1351,
+      "step": 5279
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.23719610254447e-08,
+      "loss": 1.1636,
+      "step": 5280
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.2096819424369255e-08,
+      "loss": 1.2731,
+      "step": 5281
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.182337647595012e-08,
+      "loss": 1.2232,
+      "step": 5282
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1551632273496548e-08,
+      "loss": 1.1481,
+      "step": 5283
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1281586909738805e-08,
+      "loss": 1.2161,
+      "step": 5284
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.1013240476826512e-08,
+      "loss": 1.1103,
+      "step": 5285
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.07465930663292e-08,
+      "loss": 1.1545,
+      "step": 5286
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.048164476923853e-08,
+      "loss": 1.1501,
+      "step": 5287
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 2.0218395675963844e-08,
+      "loss": 1.1649,
+      "step": 5288
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9956845876336616e-08,
+      "loss": 1.2452,
+      "step": 5289
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.9696995459607127e-08,
+      "loss": 1.168,
+      "step": 5290
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.943884451444722e-08,
+      "loss": 1.2334,
+      "step": 5291
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.918239312894754e-08,
+      "loss": 1.1403,
+      "step": 5292
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8927641390618644e-08,
+      "loss": 1.0129,
+      "step": 5293
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.8674589386392662e-08,
+      "loss": 1.0904,
+      "step": 5294
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.842323720261996e-08,
+      "loss": 1.0705,
+      "step": 5295
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.817358492507193e-08,
+      "loss": 1.0524,
+      "step": 5296
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7925632638939315e-08,
+      "loss": 0.9945,
+      "step": 5297
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.767938042883277e-08,
+      "loss": 1.2264,
+      "step": 5298
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7434828378783964e-08,
+      "loss": 1.0724,
+      "step": 5299
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.7191976572242808e-08,
+      "loss": 1.0841,
+      "step": 5300
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6950825092080235e-08,
+      "loss": 1.0366,
+      "step": 5301
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.671137402058598e-08,
+      "loss": 1.2353,
+      "step": 5302
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6473623439469676e-08,
+      "loss": 1.1405,
+      "step": 5303
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.623757342986143e-08,
+      "loss": 1.1948,
+      "step": 5304
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.6003224072310696e-08,
+      "loss": 1.1446,
+      "step": 5305
+    },
+    {
+      "epoch": 0.97,
+      "learning_rate": 1.577057544678684e-08,
+      "loss": 1.1452,
+      "step": 5306
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5539627632677468e-08,
+      "loss": 1.0478,
+      "step": 5307
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.5310380708791762e-08,
+      "loss": 1.0529,
+      "step": 5308
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.50828347533577e-08,
+      "loss": 1.0682,
+      "step": 5309
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4856989844022063e-08,
+      "loss": 1.1391,
+      "step": 5310
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4632846057852645e-08,
+      "loss": 1.1679,
+      "step": 5311
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4410403471334934e-08,
+      "loss": 1.2717,
+      "step": 5312
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.4189662160375984e-08,
+      "loss": 1.1969,
+      "step": 5313
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3970622200300543e-08,
+      "loss": 1.1269,
+      "step": 5314
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3753283665853822e-08,
+      "loss": 1.2089,
+      "step": 5315
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.353764663119983e-08,
+      "loss": 1.1159,
+      "step": 5316
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.3323711169922482e-08,
+      "loss": 1.1708,
+      "step": 5317
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.311147735502505e-08,
+      "loss": 1.1979,
+      "step": 5318
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2900945258929597e-08,
+      "loss": 1.1392,
+      "step": 5319
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2692114953477552e-08,
+      "loss": 1.0323,
+      "step": 5320
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2484986509930242e-08,
+      "loss": 1.0672,
+      "step": 5321
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2279559998968349e-08,
+      "loss": 1.0578,
+      "step": 5322
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.2075835490690247e-08,
+      "loss": 1.3052,
+      "step": 5323
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.187381305461477e-08,
+      "loss": 1.3913,
+      "step": 5324
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.167349275968066e-08,
+      "loss": 1.0799,
+      "step": 5325
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1474874674244352e-08,
+      "loss": 1.0744,
+      "step": 5326
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1277958866082184e-08,
+      "loss": 1.0851,
+      "step": 5327
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.1082745402389294e-08,
+      "loss": 1.0598,
+      "step": 5328
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0889234349779621e-08,
+      "loss": 1.2001,
+      "step": 5329
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0697425774287006e-08,
+      "loss": 1.1266,
+      "step": 5330
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0507319741364653e-08,
+      "loss": 1.0756,
+      "step": 5331
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0318916315882888e-08,
+      "loss": 1.1307,
+      "step": 5332
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 1.0132215562132507e-08,
+      "loss": 1.1283,
+      "step": 5333
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.947217543823661e-09,
+      "loss": 1.2174,
+      "step": 5334
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.763922324084185e-09,
+      "loss": 1.1256,
+      "step": 5335
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.582329965461823e-09,
+      "loss": 0.9551,
+      "step": 5336
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.402440529922562e-09,
+      "loss": 1.0349,
+      "step": 5337
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.224254078851747e-09,
+      "loss": 1.0895,
+      "step": 5338
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 9.047770673053514e-09,
+      "loss": 1.0797,
+      "step": 5339
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.872990372751355e-09,
+      "loss": 1.0259,
+      "step": 5340
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.6999132375859e-09,
+      "loss": 1.1008,
+      "step": 5341
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.52853932661879e-09,
+      "loss": 1.1526,
+      "step": 5342
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.358868698328804e-09,
+      "loss": 1.1332,
+      "step": 5343
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.190901410614071e-09,
+      "loss": 1.1214,
+      "step": 5344
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 8.024637520792077e-09,
+      "loss": 1.1993,
+      "step": 5345
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.860077085597995e-09,
+      "loss": 1.3025,
+      "step": 5346
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.697220161186346e-09,
+      "loss": 1.1257,
+      "step": 5347
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.53606680313046e-09,
+      "loss": 1.0566,
+      "step": 5348
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.376617066421343e-09,
+      "loss": 1.237,
+      "step": 5349
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.218871005470473e-09,
+      "loss": 1.0358,
+      "step": 5350
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 7.0628286741059035e-09,
+      "loss": 1.095,
+      "step": 5351
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.908490125575595e-09,
+      "loss": 1.0063,
+      "step": 5352
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.755855412545753e-09,
+      "loss": 1.0782,
+      "step": 5353
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.604924587101935e-09,
+      "loss": 1.0684,
+      "step": 5354
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.455697700746832e-09,
+      "loss": 1.1587,
+      "step": 5355
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.308174804402489e-09,
+      "loss": 1.16,
+      "step": 5356
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.162355948409749e-09,
+      "loss": 1.071,
+      "step": 5357
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 6.01824118252714e-09,
+      "loss": 1.0935,
+      "step": 5358
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.875830555932549e-09,
+      "loss": 1.1076,
+      "step": 5359
+    },
+    {
+      "epoch": 0.98,
+      "learning_rate": 5.735124117221547e-09,
+      "loss": 1.1212,
+      "step": 5360
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.596121914409059e-09,
+      "loss": 1.0155,
+      "step": 5361
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.4588239949282525e-09,
+      "loss": 1.1084,
+      "step": 5362
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.323230405629986e-09,
+      "loss": 1.0123,
+      "step": 5363
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.189341192783914e-09,
+      "loss": 1.0606,
+      "step": 5364
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 5.0571564020784894e-09,
+      "loss": 1.0552,
+      "step": 5365
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.926676078620407e-09,
+      "loss": 1.0757,
+      "step": 5366
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.797900266934052e-09,
+      "loss": 1.1226,
+      "step": 5367
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.6708290109631625e-09,
+      "loss": 0.9914,
+      "step": 5368
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.545462354069163e-09,
+      "loss": 1.1108,
+      "step": 5369
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.421800339032278e-09,
+      "loss": 1.2371,
+      "step": 5370
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.2998430080504195e-09,
+      "loss": 1.1007,
+      "step": 5371
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.1795904027397415e-09,
+      "loss": 1.0952,
+      "step": 5372
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 4.0610425641363085e-09,
+      "loss": 1.0904,
+      "step": 5373
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.9441995326922056e-09,
+      "loss": 1.0479,
+      "step": 5374
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.829061348278873e-09,
+      "loss": 1.2712,
+      "step": 5375
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.7156280501859933e-09,
+      "loss": 1.0315,
+      "step": 5376
+    },
+    {
+      "epoch": 0.99,
+      "eval_loss": 1.931097149848938,
+      "eval_runtime": 0.1726,
+      "eval_samples_per_second": 17.385,
+      "eval_steps_per_second": 17.385,
+      "step": 5376
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.603899677122047e-09,
+      "loss": 1.2146,
+      "step": 5377
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.493876267212093e-09,
+      "loss": 1.1058,
+      "step": 5378
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.3855578580010982e-09,
+      "loss": 0.9977,
+      "step": 5379
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.2789444864506083e-09,
+      "loss": 1.1816,
+      "step": 5380
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.174036188942076e-09,
+      "loss": 1.0102,
+      "step": 5381
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 3.0708330012740872e-09,
+      "loss": 1.1111,
+      "step": 5382
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.9693349586634725e-09,
+      "loss": 1.0612,
+      "step": 5383
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.8695420957453037e-09,
+      "loss": 1.1782,
+      "step": 5384
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.7714544465723415e-09,
+      "loss": 1.1777,
+      "step": 5385
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.675072044616145e-09,
+      "loss": 0.9021,
+      "step": 5386
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.580394922766516e-09,
+      "loss": 1.0474,
+      "step": 5387
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.4874231133303893e-09,
+      "loss": 1.1123,
+      "step": 5388
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3961566480334988e-09,
+      "loss": 1.1794,
+      "step": 5389
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.3065955580198197e-09,
+      "loss": 1.2236,
+      "step": 5390
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.218739873850462e-09,
+      "loss": 1.015,
+      "step": 5391
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.1325896255058874e-09,
+      "loss": 1.0056,
+      "step": 5392
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 2.0481448423831375e-09,
+      "loss": 1.2574,
+      "step": 5393
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.9654055532986048e-09,
+      "loss": 1.1544,
+      "step": 5394
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8843717864852617e-09,
+      "loss": 1.1427,
+      "step": 5395
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.8050435695965428e-09,
+      "loss": 1.1525,
+      "step": 5396
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.727420929700796e-09,
+      "loss": 1.1378,
+      "step": 5397
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.651503893286277e-09,
+      "loss": 1.1102,
+      "step": 5398
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.577292486259485e-09,
+      "loss": 1.067,
+      "step": 5399
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.5047867339434974e-09,
+      "loss": 1.1396,
+      "step": 5400
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.4339866610796338e-09,
+      "loss": 1.068,
+      "step": 5401
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.3648922918291229e-09,
+      "loss": 1.1806,
+      "step": 5402
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.297503649768106e-09,
+      "loss": 1.1337,
+      "step": 5403
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.2318207578931874e-09,
+      "loss": 1.1955,
+      "step": 5404
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.1678436386169944e-09,
+      "loss": 1.0511,
+      "step": 5405
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.105572313772063e-09,
+      "loss": 1.1419,
+      "step": 5406
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 1.0450068046069517e-09,
+      "loss": 1.0393,
+      "step": 5407
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.861471317895722e-10,
+      "loss": 1.2102,
+      "step": 5408
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 9.28993315404414e-10,
+      "loss": 0.9785,
+      "step": 5409
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.735453749553202e-10,
+      "loss": 1.0649,
+      "step": 5410
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 8.198033293627117e-10,
+      "loss": 1.1691,
+      "step": 5411
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.677671969652522e-10,
+      "loss": 1.1037,
+      "step": 5412
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 7.174369955204041e-10,
+      "loss": 1.0567,
+      "step": 5413
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.688127422022073e-10,
+      "loss": 1.1088,
+      "step": 5414
+    },
+    {
+      "epoch": 0.99,
+      "learning_rate": 6.218944536034998e-10,
+      "loss": 1.2253,
+      "step": 5415
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.766821457348082e-10,
+      "loss": 1.2598,
+      "step": 5416
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.331758340232363e-10,
+      "loss": 1.1633,
+      "step": 5417
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.913755333157966e-10,
+      "loss": 1.0838,
+      "step": 5418
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.5128125787663455e-10,
+      "loss": 1.2784,
+      "step": 5419
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.128930213864735e-10,
+      "loss": 1.1152,
+      "step": 5420
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.7621083694538987e-10,
+      "loss": 1.0377,
+      "step": 5421
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.4123471707059317e-10,
+      "loss": 1.1115,
+      "step": 5422
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.079646736975361e-10,
+      "loss": 1.1117,
+      "step": 5423
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.764007181788042e-10,
+      "loss": 1.0507,
+      "step": 5424
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.465428612857812e-10,
+      "loss": 1.0445,
+      "step": 5425
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1839111320642868e-10,
+      "loss": 1.0555,
+      "step": 5426
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.9194548354806165e-10,
+      "loss": 1.0616,
+      "step": 5427
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.6720598133457278e-10,
+      "loss": 1.0277,
+      "step": 5428
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.44172615008098e-10,
+      "loss": 1.1901,
+      "step": 5429
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.2284539242790606e-10,
+      "loss": 1.1689,
+      "step": 5430
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.0322432087261914e-10,
+      "loss": 1.1523,
+      "step": 5431
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.530940703688206e-11,
+      "loss": 0.9988,
+      "step": 5432
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 6.910065703469304e-11,
+      "loss": 1.1702,
+      "step": 5433
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 5.459807639662806e-11,
+      "loss": 1.1214,
+      "step": 5434
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 4.180167007150626e-11,
+      "loss": 0.9832,
+      "step": 5435
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.07114424263899e-11,
+      "loss": 1.1281,
+      "step": 5436
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 2.1327397244919057e-11,
+      "loss": 1.1004,
+      "step": 5437
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 1.3649537730087148e-11,
+      "loss": 1.1622,
+      "step": 5438
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 7.677866501465403e-12,
+      "loss": 1.1052,
+      "step": 5439
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 3.412385596868184e-12,
+      "loss": 1.308,
+      "step": 5440
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 8.530964717978763e-13,
+      "loss": 1.0614,
+      "step": 5441
+    },
+    {
+      "epoch": 1.0,
+      "learning_rate": 0.0,
+      "loss": 1.0163,
+      "step": 5442
+    },
+    {
+      "epoch": 1.0,
+      "step": 5442,
+      "total_flos": 1.212897999126528e+18,
+      "train_loss": 1.1426390734107912,
+      "train_runtime": 33581.509,
+      "train_samples_per_second": 10.373,
+      "train_steps_per_second": 0.162
+    }
+  ],
+  "max_steps": 5442,
+  "num_train_epochs": 1,
+  "total_flos": 1.212897999126528e+18,
+  "trial_name": null,
+  "trial_params": null
+}