JackWong0911's picture
End of training
761b028 verified
{
"best_metric": 0.98,
"best_model_checkpoint": "timesformer-base-finetuned-k400-kinetic400-subset-epoch10real-num_frame_10_myViT_more_data/checkpoint-375",
"epoch": 9.1,
"eval_steps": 500,
"global_step": 1250,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"accuracy": 0.5,
"epoch": 0,
"step": 0
},
{
"accuracy": 0.5,
"epoch": 0.0,
"step": 1
},
{
"accuracy": 0.0,
"epoch": 0.0,
"step": 2
},
{
"accuracy": 0.0,
"epoch": 0.0,
"step": 3
},
{
"accuracy": 0.5,
"epoch": 0.0,
"step": 4
},
{
"accuracy": 0.0,
"epoch": 0.0,
"step": 5
},
{
"accuracy": 0.0,
"epoch": 0.0,
"step": 6
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 7
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 8
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 9
},
{
"epoch": 0.01,
"grad_norm": 21.571460723876953,
"learning_rate": 4.000000000000001e-06,
"loss": 1.6013,
"step": 10
},
{
"accuracy": 0.5,
"epoch": 0.01,
"step": 10
},
{
"accuracy": 0.5,
"epoch": 0.01,
"step": 11
},
{
"accuracy": 1.0,
"epoch": 0.01,
"step": 12
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 13
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 14
},
{
"accuracy": 0.5,
"epoch": 0.01,
"step": 15
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 16
},
{
"accuracy": 1.0,
"epoch": 0.01,
"step": 17
},
{
"accuracy": 0.0,
"epoch": 0.01,
"step": 18
},
{
"accuracy": 0.0,
"epoch": 0.02,
"step": 19
},
{
"epoch": 0.02,
"grad_norm": 23.990596771240234,
"learning_rate": 8.000000000000001e-06,
"loss": 1.4067,
"step": 20
},
{
"accuracy": 0.0,
"epoch": 0.02,
"step": 20
},
{
"accuracy": 1.0,
"epoch": 0.02,
"step": 21
},
{
"accuracy": 0.0,
"epoch": 0.02,
"step": 22
},
{
"accuracy": 1.0,
"epoch": 0.02,
"step": 23
},
{
"accuracy": 0.5,
"epoch": 0.02,
"step": 24
},
{
"accuracy": 0.0,
"epoch": 0.02,
"step": 25
},
{
"accuracy": 0.5,
"epoch": 0.02,
"step": 26
},
{
"accuracy": 0.5,
"epoch": 0.02,
"step": 27
},
{
"accuracy": 1.0,
"epoch": 0.02,
"step": 28
},
{
"accuracy": 0.0,
"epoch": 0.02,
"step": 29
},
{
"epoch": 0.02,
"grad_norm": 21.826385498046875,
"learning_rate": 1.2e-05,
"loss": 1.2524,
"step": 30
},
{
"accuracy": 1.0,
"epoch": 0.02,
"step": 30
},
{
"accuracy": 0.5,
"epoch": 0.02,
"step": 31
},
{
"accuracy": 1.0,
"epoch": 0.03,
"step": 32
},
{
"accuracy": 0.5,
"epoch": 0.03,
"step": 33
},
{
"accuracy": 0.5,
"epoch": 0.03,
"step": 34
},
{
"accuracy": 1.0,
"epoch": 0.03,
"step": 35
},
{
"accuracy": 0.5,
"epoch": 0.03,
"step": 36
},
{
"accuracy": 0.5,
"epoch": 0.03,
"step": 37
},
{
"accuracy": 1.0,
"epoch": 0.03,
"step": 38
},
{
"accuracy": 1.0,
"epoch": 0.03,
"step": 39
},
{
"epoch": 0.03,
"grad_norm": 11.284616470336914,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.9216,
"step": 40
},
{
"accuracy": 0.5,
"epoch": 0.03,
"step": 40
},
{
"accuracy": 1.0,
"epoch": 0.03,
"step": 41
},
{
"accuracy": 1.0,
"epoch": 0.03,
"step": 42
},
{
"accuracy": 0.5,
"epoch": 0.03,
"step": 43
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 44
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 45
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 46
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 47
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 48
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 49
},
{
"epoch": 0.04,
"grad_norm": 5.933394908905029,
"learning_rate": 2e-05,
"loss": 0.5672,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 50
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 51
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 52
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 53
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 54
},
{
"accuracy": 0.5,
"epoch": 0.04,
"step": 55
},
{
"accuracy": 1.0,
"epoch": 0.04,
"step": 56
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 57
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 58
},
{
"accuracy": 0.5,
"epoch": 0.05,
"step": 59
},
{
"epoch": 0.05,
"grad_norm": 16.361772537231445,
"learning_rate": 2.4e-05,
"loss": 0.4843,
"step": 60
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 60
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 61
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 62
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 63
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 64
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 65
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 66
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 67
},
{
"accuracy": 1.0,
"epoch": 0.05,
"step": 68
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 69
},
{
"epoch": 0.06,
"grad_norm": 10.710639953613281,
"learning_rate": 2.8000000000000003e-05,
"loss": 0.337,
"step": 70
},
{
"accuracy": 0.5,
"epoch": 0.06,
"step": 70
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 71
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 72
},
{
"accuracy": 0.5,
"epoch": 0.06,
"step": 73
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 74
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 75
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 76
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 77
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 78
},
{
"accuracy": 0.5,
"epoch": 0.06,
"step": 79
},
{
"epoch": 0.06,
"grad_norm": 22.222740173339844,
"learning_rate": 3.2000000000000005e-05,
"loss": 0.5472,
"step": 80
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 80
},
{
"accuracy": 1.0,
"epoch": 0.06,
"step": 81
},
{
"accuracy": 0.5,
"epoch": 0.07,
"step": 82
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 83
},
{
"accuracy": 0.0,
"epoch": 0.07,
"step": 84
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 85
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 86
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 87
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 88
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 89
},
{
"epoch": 0.07,
"grad_norm": 1.158689260482788,
"learning_rate": 3.6e-05,
"loss": 0.2179,
"step": 90
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 90
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 91
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 92
},
{
"accuracy": 1.0,
"epoch": 0.07,
"step": 93
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 94
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 95
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 96
},
{
"accuracy": 0.5,
"epoch": 0.08,
"step": 97
},
{
"accuracy": 0.5,
"epoch": 0.08,
"step": 98
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 99
},
{
"epoch": 0.08,
"grad_norm": 0.06950497627258301,
"learning_rate": 4e-05,
"loss": 0.3101,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 100
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 101
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 102
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 103
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 104
},
{
"accuracy": 1.0,
"epoch": 0.08,
"step": 105
},
{
"accuracy": 0.5,
"epoch": 0.08,
"step": 106
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 107
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 108
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 109
},
{
"epoch": 0.09,
"grad_norm": 0.23963814973831177,
"learning_rate": 4.4000000000000006e-05,
"loss": 0.1309,
"step": 110
},
{
"accuracy": 0.5,
"epoch": 0.09,
"step": 110
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 111
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 112
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 113
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 114
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 115
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 116
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 117
},
{
"accuracy": 1.0,
"epoch": 0.09,
"step": 118
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 119
},
{
"epoch": 0.1,
"grad_norm": 0.5907143950462341,
"learning_rate": 4.8e-05,
"loss": 0.2067,
"step": 120
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 120
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 121
},
{
"accuracy": 0.5,
"epoch": 0.1,
"step": 122
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 123
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 124
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 0.5,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 0.5,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 0.5,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 0.5,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"epoch": 0.1,
"eval_accuracy": 0.96,
"eval_loss": 0.1522197425365448,
"eval_runtime": 98.5698,
"eval_samples_per_second": 1.015,
"eval_steps_per_second": 0.507,
"step": 125
},
{
"accuracy": 1.0,
"epoch": 0.1,
"step": 125
},
{
"accuracy": 0.5,
"epoch": 1.0,
"step": 126
},
{
"accuracy": 1.0,
"epoch": 1.0,
"step": 127
},
{
"accuracy": 1.0,
"epoch": 1.0,
"step": 128
},
{
"accuracy": 1.0,
"epoch": 1.0,
"step": 129
},
{
"epoch": 1.0,
"grad_norm": 13.441699028015137,
"learning_rate": 4.977777777777778e-05,
"loss": 0.3567,
"step": 130
},
{
"accuracy": 1.0,
"epoch": 1.0,
"step": 130
},
{
"accuracy": 0.5,
"epoch": 1.0,
"step": 131
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 132
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 133
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 134
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 135
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 136
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 137
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 138
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 139
},
{
"epoch": 1.01,
"grad_norm": 6.15524959564209,
"learning_rate": 4.933333333333334e-05,
"loss": 0.3622,
"step": 140
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 140
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 141
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 142
},
{
"accuracy": 1.0,
"epoch": 1.01,
"step": 143
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 144
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 145
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 146
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 147
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 148
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 149
},
{
"epoch": 1.02,
"grad_norm": 0.026683134958148003,
"learning_rate": 4.888888888888889e-05,
"loss": 0.0032,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 150
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 151
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 152
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 153
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 154
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 155
},
{
"accuracy": 1.0,
"epoch": 1.02,
"step": 156
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 157
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 158
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 159
},
{
"epoch": 1.03,
"grad_norm": 0.21031653881072998,
"learning_rate": 4.844444444444445e-05,
"loss": 0.0208,
"step": 160
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 160
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 161
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 162
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 163
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 164
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 165
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 166
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 167
},
{
"accuracy": 1.0,
"epoch": 1.03,
"step": 168
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 169
},
{
"epoch": 1.04,
"grad_norm": 0.016481030732393265,
"learning_rate": 4.8e-05,
"loss": 0.019,
"step": 170
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 170
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 171
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 172
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 173
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 174
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 175
},
{
"accuracy": 0.5,
"epoch": 1.04,
"step": 176
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 177
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 178
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 179
},
{
"epoch": 1.04,
"grad_norm": 0.015820760279893875,
"learning_rate": 4.755555555555556e-05,
"loss": 0.1675,
"step": 180
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 180
},
{
"accuracy": 1.0,
"epoch": 1.04,
"step": 181
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 182
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 183
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 184
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 185
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 186
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 187
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 188
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 189
},
{
"epoch": 1.05,
"grad_norm": 0.009809507988393307,
"learning_rate": 4.711111111111111e-05,
"loss": 0.0075,
"step": 190
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 190
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 191
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 192
},
{
"accuracy": 1.0,
"epoch": 1.05,
"step": 193
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 194
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 195
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 196
},
{
"accuracy": 0.5,
"epoch": 1.06,
"step": 197
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 198
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 199
},
{
"epoch": 1.06,
"grad_norm": 0.019570348784327507,
"learning_rate": 4.666666666666667e-05,
"loss": 0.2133,
"step": 200
},
{
"accuracy": 0.5,
"epoch": 1.06,
"step": 200
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 201
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 202
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 203
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 204
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 205
},
{
"accuracy": 1.0,
"epoch": 1.06,
"step": 206
},
{
"accuracy": 0.5,
"epoch": 1.07,
"step": 207
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 208
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 209
},
{
"epoch": 1.07,
"grad_norm": 0.02470504306256771,
"learning_rate": 4.6222222222222224e-05,
"loss": 0.559,
"step": 210
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 210
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 211
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 212
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 213
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 214
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 215
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 216
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 217
},
{
"accuracy": 1.0,
"epoch": 1.07,
"step": 218
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 219
},
{
"epoch": 1.08,
"grad_norm": 0.07938097417354584,
"learning_rate": 4.577777777777778e-05,
"loss": 0.0044,
"step": 220
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 220
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 221
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 222
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 223
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 224
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 225
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 226
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 227
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 228
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 229
},
{
"epoch": 1.08,
"grad_norm": 0.013527393341064453,
"learning_rate": 4.5333333333333335e-05,
"loss": 0.0216,
"step": 230
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 230
},
{
"accuracy": 1.0,
"epoch": 1.08,
"step": 231
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 232
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 233
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 234
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 235
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 236
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 237
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 238
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 239
},
{
"epoch": 1.09,
"grad_norm": 0.05658406391739845,
"learning_rate": 4.4888888888888894e-05,
"loss": 0.0093,
"step": 240
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 240
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 241
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 242
},
{
"accuracy": 1.0,
"epoch": 1.09,
"step": 243
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 244
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 245
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 246
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 247
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 248
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 249
},
{
"epoch": 1.1,
"grad_norm": 0.015246179886162281,
"learning_rate": 4.4444444444444447e-05,
"loss": 0.3735,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 0.5,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"epoch": 1.1,
"eval_accuracy": 0.75,
"eval_loss": 1.3725593090057373,
"eval_runtime": 93.929,
"eval_samples_per_second": 1.065,
"eval_steps_per_second": 0.532,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 1.1,
"step": 250
},
{
"accuracy": 1.0,
"epoch": 2.0,
"step": 251
},
{
"accuracy": 1.0,
"epoch": 2.0,
"step": 252
},
{
"accuracy": 0.5,
"epoch": 2.0,
"step": 253
},
{
"accuracy": 0.5,
"epoch": 2.0,
"step": 254
},
{
"accuracy": 1.0,
"epoch": 2.0,
"step": 255
},
{
"accuracy": 1.0,
"epoch": 2.0,
"step": 256
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 257
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 258
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 259
},
{
"epoch": 2.01,
"grad_norm": 0.014890051446855068,
"learning_rate": 4.4000000000000006e-05,
"loss": 0.4016,
"step": 260
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 260
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 261
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 262
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 263
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 264
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 265
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 266
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 267
},
{
"accuracy": 1.0,
"epoch": 2.01,
"step": 268
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 269
},
{
"epoch": 2.02,
"grad_norm": 0.023178286850452423,
"learning_rate": 4.355555555555556e-05,
"loss": 0.0041,
"step": 270
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 270
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 271
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 272
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 273
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 274
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 275
},
{
"accuracy": 0.5,
"epoch": 2.02,
"step": 276
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 277
},
{
"accuracy": 0.5,
"epoch": 2.02,
"step": 278
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 279
},
{
"epoch": 2.02,
"grad_norm": 0.10281816124916077,
"learning_rate": 4.311111111111111e-05,
"loss": 0.2151,
"step": 280
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 280
},
{
"accuracy": 1.0,
"epoch": 2.02,
"step": 281
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 282
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 283
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 284
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 285
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 286
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 287
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 288
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 289
},
{
"epoch": 2.03,
"grad_norm": 0.013549063354730606,
"learning_rate": 4.266666666666667e-05,
"loss": 0.0053,
"step": 290
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 290
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 291
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 292
},
{
"accuracy": 1.0,
"epoch": 2.03,
"step": 293
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 294
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 295
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 296
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 297
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 298
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 299
},
{
"epoch": 2.04,
"grad_norm": 0.029035884886980057,
"learning_rate": 4.222222222222222e-05,
"loss": 0.0006,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 300
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 301
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 302
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 303
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 304
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 305
},
{
"accuracy": 1.0,
"epoch": 2.04,
"step": 306
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 307
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 308
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 309
},
{
"epoch": 2.05,
"grad_norm": 0.015959421172738075,
"learning_rate": 4.177777777777778e-05,
"loss": 0.0006,
"step": 310
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 310
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 311
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 312
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 313
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 314
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 315
},
{
"accuracy": 0.5,
"epoch": 2.05,
"step": 316
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 317
},
{
"accuracy": 1.0,
"epoch": 2.05,
"step": 318
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 319
},
{
"epoch": 2.06,
"grad_norm": 0.11569980531930923,
"learning_rate": 4.133333333333333e-05,
"loss": 0.0776,
"step": 320
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 320
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 321
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 322
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 323
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 324
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 325
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 326
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 327
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 328
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 329
},
{
"epoch": 2.06,
"grad_norm": 0.06833364814519882,
"learning_rate": 4.088888888888889e-05,
"loss": 0.0016,
"step": 330
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 330
},
{
"accuracy": 1.0,
"epoch": 2.06,
"step": 331
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 332
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 333
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 334
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 335
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 336
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 337
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 338
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 339
},
{
"epoch": 2.07,
"grad_norm": 0.05876513943076134,
"learning_rate": 4.0444444444444444e-05,
"loss": 0.0298,
"step": 340
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 340
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 341
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 342
},
{
"accuracy": 1.0,
"epoch": 2.07,
"step": 343
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 344
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 345
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 346
},
{
"accuracy": 0.5,
"epoch": 2.08,
"step": 347
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 348
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 349
},
{
"epoch": 2.08,
"grad_norm": 0.008067545481026173,
"learning_rate": 4e-05,
"loss": 0.2723,
"step": 350
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 350
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 351
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 352
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 353
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 354
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 355
},
{
"accuracy": 1.0,
"epoch": 2.08,
"step": 356
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 357
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 358
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 359
},
{
"epoch": 2.09,
"grad_norm": 0.006565236486494541,
"learning_rate": 3.9555555555555556e-05,
"loss": 0.0004,
"step": 360
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 360
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 361
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 362
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 363
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 364
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 365
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 366
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 367
},
{
"accuracy": 1.0,
"epoch": 2.09,
"step": 368
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 369
},
{
"epoch": 2.1,
"grad_norm": 0.007222988177090883,
"learning_rate": 3.9111111111111115e-05,
"loss": 0.0037,
"step": 370
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 370
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 371
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 372
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 373
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 374
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 0.5,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 0.5,
"epoch": 2.1,
"step": 375
},
{
"epoch": 2.1,
"eval_accuracy": 0.98,
"eval_loss": 0.11935045570135117,
"eval_runtime": 94.0926,
"eval_samples_per_second": 1.063,
"eval_steps_per_second": 0.531,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 2.1,
"step": 375
},
{
"accuracy": 1.0,
"epoch": 3.0,
"step": 376
},
{
"accuracy": 1.0,
"epoch": 3.0,
"step": 377
},
{
"accuracy": 1.0,
"epoch": 3.0,
"step": 378
},
{
"accuracy": 1.0,
"epoch": 3.0,
"step": 379
},
{
"epoch": 3.0,
"grad_norm": 0.007090215105563402,
"learning_rate": 3.866666666666667e-05,
"loss": 0.0003,
"step": 380
},
{
"accuracy": 1.0,
"epoch": 3.0,
"step": 380
},
{
"accuracy": 1.0,
"epoch": 3.0,
"step": 381
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 382
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 383
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 384
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 385
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 386
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 387
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 388
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 389
},
{
"epoch": 3.01,
"grad_norm": 0.018512064591050148,
"learning_rate": 3.8222222222222226e-05,
"loss": 0.004,
"step": 390
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 390
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 391
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 392
},
{
"accuracy": 1.0,
"epoch": 3.01,
"step": 393
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 394
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 395
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 396
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 397
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 398
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 399
},
{
"epoch": 3.02,
"grad_norm": 0.0037549559492617846,
"learning_rate": 3.777777777777778e-05,
"loss": 0.001,
"step": 400
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 400
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 401
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 402
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 403
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 404
},
{
"accuracy": 1.0,
"epoch": 3.02,
"step": 405
},
{
"accuracy": 0.5,
"epoch": 3.02,
"step": 406
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 407
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 408
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 409
},
{
"epoch": 3.03,
"grad_norm": 0.035387154668569565,
"learning_rate": 3.733333333333334e-05,
"loss": 0.1121,
"step": 410
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 410
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 411
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 412
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 413
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 414
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 415
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 416
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 417
},
{
"accuracy": 1.0,
"epoch": 3.03,
"step": 418
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 419
},
{
"epoch": 3.04,
"grad_norm": 0.01542646624147892,
"learning_rate": 3.688888888888889e-05,
"loss": 0.0004,
"step": 420
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 420
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 421
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 422
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 423
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 424
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 425
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 426
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 427
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 428
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 429
},
{
"epoch": 3.04,
"grad_norm": 0.005868628621101379,
"learning_rate": 3.644444444444445e-05,
"loss": 0.0034,
"step": 430
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 430
},
{
"accuracy": 1.0,
"epoch": 3.04,
"step": 431
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 432
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 433
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 434
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 435
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 436
},
{
"accuracy": 0.5,
"epoch": 3.05,
"step": 437
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 438
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 439
},
{
"epoch": 3.05,
"grad_norm": 0.0035853590816259384,
"learning_rate": 3.6e-05,
"loss": 0.1673,
"step": 440
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 440
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 441
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 442
},
{
"accuracy": 1.0,
"epoch": 3.05,
"step": 443
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 444
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 445
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 446
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 447
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 448
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 449
},
{
"epoch": 3.06,
"grad_norm": 0.020550154149532318,
"learning_rate": 3.555555555555556e-05,
"loss": 0.0003,
"step": 450
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 450
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 451
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 452
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 453
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 454
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 455
},
{
"accuracy": 1.0,
"epoch": 3.06,
"step": 456
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 457
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 458
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 459
},
{
"epoch": 3.07,
"grad_norm": 0.015121286734938622,
"learning_rate": 3.511111111111111e-05,
"loss": 0.0137,
"step": 460
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 460
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 461
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 462
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 463
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 464
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 465
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 466
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 467
},
{
"accuracy": 1.0,
"epoch": 3.07,
"step": 468
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 469
},
{
"epoch": 3.08,
"grad_norm": 0.0074648368172347546,
"learning_rate": 3.466666666666667e-05,
"loss": 0.0004,
"step": 470
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 470
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 471
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 472
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 473
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 474
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 475
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 476
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 477
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 478
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 479
},
{
"epoch": 3.08,
"grad_norm": 0.01709677092730999,
"learning_rate": 3.4222222222222224e-05,
"loss": 0.0004,
"step": 480
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 480
},
{
"accuracy": 1.0,
"epoch": 3.08,
"step": 481
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 482
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 483
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 484
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 485
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 486
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 487
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 488
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 489
},
{
"epoch": 3.09,
"grad_norm": 0.01061286497861147,
"learning_rate": 3.377777777777778e-05,
"loss": 0.0007,
"step": 490
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 490
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 491
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 492
},
{
"accuracy": 1.0,
"epoch": 3.09,
"step": 493
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 494
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 495
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 496
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 497
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 498
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 499
},
{
"epoch": 3.1,
"grad_norm": 0.05872837081551552,
"learning_rate": 3.3333333333333335e-05,
"loss": 0.0005,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 0.5,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 0.5,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 0.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 0.5,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"epoch": 3.1,
"eval_accuracy": 0.95,
"eval_loss": 0.25089970231056213,
"eval_runtime": 93.918,
"eval_samples_per_second": 1.065,
"eval_steps_per_second": 0.532,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 3.1,
"step": 500
},
{
"accuracy": 1.0,
"epoch": 4.0,
"step": 501
},
{
"accuracy": 1.0,
"epoch": 4.0,
"step": 502
},
{
"accuracy": 1.0,
"epoch": 4.0,
"step": 503
},
{
"accuracy": 1.0,
"epoch": 4.0,
"step": 504
},
{
"accuracy": 1.0,
"epoch": 4.0,
"step": 505
},
{
"accuracy": 1.0,
"epoch": 4.0,
"step": 506
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 507
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 508
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 509
},
{
"epoch": 4.01,
"grad_norm": 0.003928630147129297,
"learning_rate": 3.2888888888888894e-05,
"loss": 0.0013,
"step": 510
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 510
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 511
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 512
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 513
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 514
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 515
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 516
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 517
},
{
"accuracy": 1.0,
"epoch": 4.01,
"step": 518
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 519
},
{
"epoch": 4.02,
"grad_norm": 0.002800439950078726,
"learning_rate": 3.2444444444444446e-05,
"loss": 0.0002,
"step": 520
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 520
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 521
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 522
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 523
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 524
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 525
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 526
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 527
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 528
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 529
},
{
"epoch": 4.02,
"grad_norm": 0.009118727408349514,
"learning_rate": 3.2000000000000005e-05,
"loss": 0.0003,
"step": 530
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 530
},
{
"accuracy": 1.0,
"epoch": 4.02,
"step": 531
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 532
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 533
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 534
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 535
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 536
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 537
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 538
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 539
},
{
"epoch": 4.03,
"grad_norm": 0.0068476758897304535,
"learning_rate": 3.155555555555556e-05,
"loss": 0.0004,
"step": 540
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 540
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 541
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 542
},
{
"accuracy": 1.0,
"epoch": 4.03,
"step": 543
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 544
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 545
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 546
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 547
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 548
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 549
},
{
"epoch": 4.04,
"grad_norm": 0.004228397272527218,
"learning_rate": 3.111111111111111e-05,
"loss": 0.0006,
"step": 550
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 550
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 551
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 552
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 553
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 554
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 555
},
{
"accuracy": 1.0,
"epoch": 4.04,
"step": 556
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 557
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 558
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 559
},
{
"epoch": 4.05,
"grad_norm": 0.0036121259909123182,
"learning_rate": 3.066666666666667e-05,
"loss": 0.0002,
"step": 560
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 560
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 561
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 562
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 563
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 564
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 565
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 566
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 567
},
{
"accuracy": 1.0,
"epoch": 4.05,
"step": 568
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 569
},
{
"epoch": 4.06,
"grad_norm": 0.002266103867441416,
"learning_rate": 3.0222222222222225e-05,
"loss": 0.0002,
"step": 570
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 570
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 571
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 572
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 573
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 574
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 575
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 576
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 577
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 578
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 579
},
{
"epoch": 4.06,
"grad_norm": 0.014209027402102947,
"learning_rate": 2.9777777777777777e-05,
"loss": 0.0065,
"step": 580
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 580
},
{
"accuracy": 1.0,
"epoch": 4.06,
"step": 581
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 582
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 583
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 584
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 585
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 586
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 587
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 588
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 589
},
{
"epoch": 4.07,
"grad_norm": 0.0025224019773304462,
"learning_rate": 2.9333333333333336e-05,
"loss": 0.0003,
"step": 590
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 590
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 591
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 592
},
{
"accuracy": 1.0,
"epoch": 4.07,
"step": 593
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 594
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 595
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 596
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 597
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 598
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 599
},
{
"epoch": 4.08,
"grad_norm": 0.0034769491758197546,
"learning_rate": 2.8888888888888888e-05,
"loss": 0.0003,
"step": 600
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 600
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 601
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 602
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 603
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 604
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 605
},
{
"accuracy": 1.0,
"epoch": 4.08,
"step": 606
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 607
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 608
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 609
},
{
"epoch": 4.09,
"grad_norm": 0.004542391747236252,
"learning_rate": 2.8444444444444447e-05,
"loss": 0.0002,
"step": 610
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 610
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 611
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 612
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 613
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 614
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 615
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 616
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 617
},
{
"accuracy": 1.0,
"epoch": 4.09,
"step": 618
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 619
},
{
"epoch": 4.1,
"grad_norm": 0.005689011886715889,
"learning_rate": 2.8000000000000003e-05,
"loss": 0.0003,
"step": 620
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 620
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 621
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 622
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 623
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 624
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 0.5,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 0.5,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 0.5,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 0.5,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"epoch": 4.1,
"eval_accuracy": 0.96,
"eval_loss": 0.29478344321250916,
"eval_runtime": 94.0648,
"eval_samples_per_second": 1.063,
"eval_steps_per_second": 0.532,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 4.1,
"step": 625
},
{
"accuracy": 1.0,
"epoch": 5.0,
"step": 626
},
{
"accuracy": 1.0,
"epoch": 5.0,
"step": 627
},
{
"accuracy": 1.0,
"epoch": 5.0,
"step": 628
},
{
"accuracy": 1.0,
"epoch": 5.0,
"step": 629
},
{
"epoch": 5.0,
"grad_norm": 0.0021017505787312984,
"learning_rate": 2.7555555555555555e-05,
"loss": 0.0008,
"step": 630
},
{
"accuracy": 1.0,
"epoch": 5.0,
"step": 630
},
{
"accuracy": 1.0,
"epoch": 5.0,
"step": 631
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 632
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 633
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 634
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 635
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 636
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 637
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 638
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 639
},
{
"epoch": 5.01,
"grad_norm": 0.019864145666360855,
"learning_rate": 2.7111111111111114e-05,
"loss": 0.0002,
"step": 640
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 640
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 641
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 642
},
{
"accuracy": 1.0,
"epoch": 5.01,
"step": 643
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 644
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 645
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 646
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 647
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 648
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 649
},
{
"epoch": 5.02,
"grad_norm": 0.002870849333703518,
"learning_rate": 2.6666666666666667e-05,
"loss": 0.0001,
"step": 650
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 650
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 651
},
{
"accuracy": 0.5,
"epoch": 5.02,
"step": 652
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 653
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 654
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 655
},
{
"accuracy": 1.0,
"epoch": 5.02,
"step": 656
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 657
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 658
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 659
},
{
"epoch": 5.03,
"grad_norm": 0.0022918384056538343,
"learning_rate": 2.6222222222222226e-05,
"loss": 0.1137,
"step": 660
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 660
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 661
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 662
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 663
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 664
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 665
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 666
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 667
},
{
"accuracy": 1.0,
"epoch": 5.03,
"step": 668
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 669
},
{
"epoch": 5.04,
"grad_norm": 0.015529915690422058,
"learning_rate": 2.5777777777777778e-05,
"loss": 0.0002,
"step": 670
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 670
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 671
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 672
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 673
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 674
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 675
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 676
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 677
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 678
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 679
},
{
"epoch": 5.04,
"grad_norm": 0.007806820794939995,
"learning_rate": 2.5333333333333337e-05,
"loss": 0.0003,
"step": 680
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 680
},
{
"accuracy": 1.0,
"epoch": 5.04,
"step": 681
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 682
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 683
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 684
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 685
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 686
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 687
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 688
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 689
},
{
"epoch": 5.05,
"grad_norm": 0.01139283087104559,
"learning_rate": 2.488888888888889e-05,
"loss": 0.0008,
"step": 690
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 690
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 691
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 692
},
{
"accuracy": 1.0,
"epoch": 5.05,
"step": 693
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 694
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 695
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 696
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 697
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 698
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 699
},
{
"epoch": 5.06,
"grad_norm": 0.004373813513666391,
"learning_rate": 2.4444444444444445e-05,
"loss": 0.0002,
"step": 700
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 700
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 701
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 702
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 703
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 704
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 705
},
{
"accuracy": 1.0,
"epoch": 5.06,
"step": 706
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 707
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 708
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 709
},
{
"epoch": 5.07,
"grad_norm": 0.009814570657908916,
"learning_rate": 2.4e-05,
"loss": 0.0002,
"step": 710
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 710
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 711
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 712
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 713
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 714
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 715
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 716
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 717
},
{
"accuracy": 1.0,
"epoch": 5.07,
"step": 718
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 719
},
{
"epoch": 5.08,
"grad_norm": 0.0022362421732395887,
"learning_rate": 2.3555555555555556e-05,
"loss": 0.0002,
"step": 720
},
{
"accuracy": 0.5,
"epoch": 5.08,
"step": 720
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 721
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 722
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 723
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 724
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 725
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 726
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 727
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 728
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 729
},
{
"epoch": 5.08,
"grad_norm": 1.4584314823150635,
"learning_rate": 2.3111111111111112e-05,
"loss": 0.2522,
"step": 730
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 730
},
{
"accuracy": 1.0,
"epoch": 5.08,
"step": 731
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 732
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 733
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 734
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 735
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 736
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 737
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 738
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 739
},
{
"epoch": 5.09,
"grad_norm": 0.0015705085825175047,
"learning_rate": 2.2666666666666668e-05,
"loss": 0.0004,
"step": 740
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 740
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 741
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 742
},
{
"accuracy": 1.0,
"epoch": 5.09,
"step": 743
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 744
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 745
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 746
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 747
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 748
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 749
},
{
"epoch": 5.1,
"grad_norm": 0.004298639018088579,
"learning_rate": 2.2222222222222223e-05,
"loss": 0.0007,
"step": 750
},
{
"accuracy": 0.5,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 0.5,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 0.5,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 0.5,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 0.5,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 0.5,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 0.5,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 0.5,
"epoch": 5.1,
"step": 750
},
{
"epoch": 5.1,
"eval_accuracy": 0.92,
"eval_loss": 0.44825905561447144,
"eval_runtime": 94.39,
"eval_samples_per_second": 1.059,
"eval_steps_per_second": 0.53,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 5.1,
"step": 750
},
{
"accuracy": 1.0,
"epoch": 6.0,
"step": 751
},
{
"accuracy": 1.0,
"epoch": 6.0,
"step": 752
},
{
"accuracy": 1.0,
"epoch": 6.0,
"step": 753
},
{
"accuracy": 1.0,
"epoch": 6.0,
"step": 754
},
{
"accuracy": 1.0,
"epoch": 6.0,
"step": 755
},
{
"accuracy": 1.0,
"epoch": 6.0,
"step": 756
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 757
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 758
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 759
},
{
"epoch": 6.01,
"grad_norm": 0.004659948870539665,
"learning_rate": 2.177777777777778e-05,
"loss": 0.0002,
"step": 760
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 760
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 761
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 762
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 763
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 764
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 765
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 766
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 767
},
{
"accuracy": 1.0,
"epoch": 6.01,
"step": 768
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 769
},
{
"epoch": 6.02,
"grad_norm": 0.003489469876512885,
"learning_rate": 2.1333333333333335e-05,
"loss": 0.0002,
"step": 770
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 770
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 771
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 772
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 773
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 774
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 775
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 776
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 777
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 778
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 779
},
{
"epoch": 6.02,
"grad_norm": 0.0028468689415603876,
"learning_rate": 2.088888888888889e-05,
"loss": 0.0001,
"step": 780
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 780
},
{
"accuracy": 1.0,
"epoch": 6.02,
"step": 781
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 782
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 783
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 784
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 785
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 786
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 787
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 788
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 789
},
{
"epoch": 6.03,
"grad_norm": 0.005153105594217777,
"learning_rate": 2.0444444444444446e-05,
"loss": 0.0002,
"step": 790
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 790
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 791
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 792
},
{
"accuracy": 1.0,
"epoch": 6.03,
"step": 793
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 794
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 795
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 796
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 797
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 798
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 799
},
{
"epoch": 6.04,
"grad_norm": 0.002051904797554016,
"learning_rate": 2e-05,
"loss": 0.0002,
"step": 800
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 800
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 801
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 802
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 803
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 804
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 805
},
{
"accuracy": 1.0,
"epoch": 6.04,
"step": 806
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 807
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 808
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 809
},
{
"epoch": 6.05,
"grad_norm": 0.004386757500469685,
"learning_rate": 1.9555555555555557e-05,
"loss": 0.0001,
"step": 810
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 810
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 811
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 812
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 813
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 814
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 815
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 816
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 817
},
{
"accuracy": 1.0,
"epoch": 6.05,
"step": 818
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 819
},
{
"epoch": 6.06,
"grad_norm": 0.0025042714551091194,
"learning_rate": 1.9111111111111113e-05,
"loss": 0.0001,
"step": 820
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 820
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 821
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 822
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 823
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 824
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 825
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 826
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 827
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 828
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 829
},
{
"epoch": 6.06,
"grad_norm": 0.003163014305755496,
"learning_rate": 1.866666666666667e-05,
"loss": 0.0183,
"step": 830
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 830
},
{
"accuracy": 1.0,
"epoch": 6.06,
"step": 831
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 832
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 833
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 834
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 835
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 836
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 837
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 838
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 839
},
{
"epoch": 6.07,
"grad_norm": 0.00828773993998766,
"learning_rate": 1.8222222222222224e-05,
"loss": 0.0002,
"step": 840
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 840
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 841
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 842
},
{
"accuracy": 1.0,
"epoch": 6.07,
"step": 843
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 844
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 845
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 846
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 847
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 848
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 849
},
{
"epoch": 6.08,
"grad_norm": 0.0025350828655064106,
"learning_rate": 1.777777777777778e-05,
"loss": 0.0001,
"step": 850
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 850
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 851
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 852
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 853
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 854
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 855
},
{
"accuracy": 1.0,
"epoch": 6.08,
"step": 856
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 857
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 858
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 859
},
{
"epoch": 6.09,
"grad_norm": 0.00199713371694088,
"learning_rate": 1.7333333333333336e-05,
"loss": 0.0009,
"step": 860
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 860
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 861
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 862
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 863
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 864
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 865
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 866
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 867
},
{
"accuracy": 1.0,
"epoch": 6.09,
"step": 868
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 869
},
{
"epoch": 6.1,
"grad_norm": 0.0026183256413787603,
"learning_rate": 1.688888888888889e-05,
"loss": 0.0018,
"step": 870
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 870
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 871
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 872
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 873
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 874
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 0.5,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 0.5,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 0.5,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 0.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 0.5,
"epoch": 6.1,
"step": 875
},
{
"epoch": 6.1,
"eval_accuracy": 0.94,
"eval_loss": 0.31350982189178467,
"eval_runtime": 94.3815,
"eval_samples_per_second": 1.06,
"eval_steps_per_second": 0.53,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 6.1,
"step": 875
},
{
"accuracy": 1.0,
"epoch": 7.0,
"step": 876
},
{
"accuracy": 1.0,
"epoch": 7.0,
"step": 877
},
{
"accuracy": 1.0,
"epoch": 7.0,
"step": 878
},
{
"accuracy": 1.0,
"epoch": 7.0,
"step": 879
},
{
"epoch": 7.0,
"grad_norm": 0.0016529739368706942,
"learning_rate": 1.6444444444444447e-05,
"loss": 0.0003,
"step": 880
},
{
"accuracy": 1.0,
"epoch": 7.0,
"step": 880
},
{
"accuracy": 1.0,
"epoch": 7.0,
"step": 881
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 882
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 883
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 884
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 885
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 886
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 887
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 888
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 889
},
{
"epoch": 7.01,
"grad_norm": 0.02236793003976345,
"learning_rate": 1.6000000000000003e-05,
"loss": 0.0007,
"step": 890
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 890
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 891
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 892
},
{
"accuracy": 1.0,
"epoch": 7.01,
"step": 893
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 894
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 895
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 896
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 897
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 898
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 899
},
{
"epoch": 7.02,
"grad_norm": 0.018501687794923782,
"learning_rate": 1.5555555555555555e-05,
"loss": 0.0001,
"step": 900
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 900
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 901
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 902
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 903
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 904
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 905
},
{
"accuracy": 1.0,
"epoch": 7.02,
"step": 906
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 907
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 908
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 909
},
{
"epoch": 7.03,
"grad_norm": 0.0014730874681845307,
"learning_rate": 1.5111111111111112e-05,
"loss": 0.0001,
"step": 910
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 910
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 911
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 912
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 913
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 914
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 915
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 916
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 917
},
{
"accuracy": 1.0,
"epoch": 7.03,
"step": 918
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 919
},
{
"epoch": 7.04,
"grad_norm": 0.0027739210054278374,
"learning_rate": 1.4666666666666668e-05,
"loss": 0.0001,
"step": 920
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 920
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 921
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 922
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 923
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 924
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 925
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 926
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 927
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 928
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 929
},
{
"epoch": 7.04,
"grad_norm": 0.0033802660182118416,
"learning_rate": 1.4222222222222224e-05,
"loss": 0.0001,
"step": 930
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 930
},
{
"accuracy": 1.0,
"epoch": 7.04,
"step": 931
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 932
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 933
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 934
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 935
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 936
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 937
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 938
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 939
},
{
"epoch": 7.05,
"grad_norm": 0.0022476555313915014,
"learning_rate": 1.3777777777777778e-05,
"loss": 0.0001,
"step": 940
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 940
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 941
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 942
},
{
"accuracy": 1.0,
"epoch": 7.05,
"step": 943
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 944
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 945
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 946
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 947
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 948
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 949
},
{
"epoch": 7.06,
"grad_norm": 0.004386269487440586,
"learning_rate": 1.3333333333333333e-05,
"loss": 0.0001,
"step": 950
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 950
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 951
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 952
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 953
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 954
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 955
},
{
"accuracy": 1.0,
"epoch": 7.06,
"step": 956
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 957
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 958
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 959
},
{
"epoch": 7.07,
"grad_norm": 0.0022108792327344418,
"learning_rate": 1.2888888888888889e-05,
"loss": 0.0001,
"step": 960
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 960
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 961
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 962
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 963
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 964
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 965
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 966
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 967
},
{
"accuracy": 1.0,
"epoch": 7.07,
"step": 968
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 969
},
{
"epoch": 7.08,
"grad_norm": 0.0013348901411518455,
"learning_rate": 1.2444444444444445e-05,
"loss": 0.0001,
"step": 970
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 970
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 971
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 972
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 973
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 974
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 975
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 976
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 977
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 978
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 979
},
{
"epoch": 7.08,
"grad_norm": 0.002387480577453971,
"learning_rate": 1.2e-05,
"loss": 0.0002,
"step": 980
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 980
},
{
"accuracy": 1.0,
"epoch": 7.08,
"step": 981
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 982
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 983
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 984
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 985
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 986
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 987
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 988
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 989
},
{
"epoch": 7.09,
"grad_norm": 0.0019740264397114515,
"learning_rate": 1.1555555555555556e-05,
"loss": 0.0002,
"step": 990
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 990
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 991
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 992
},
{
"accuracy": 1.0,
"epoch": 7.09,
"step": 993
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 994
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 995
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 996
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 997
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 998
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 999
},
{
"epoch": 7.1,
"grad_norm": 0.001346210832707584,
"learning_rate": 1.1111111111111112e-05,
"loss": 0.0001,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 0.5,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 0.5,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 0.5,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 0.0,
"epoch": 7.1,
"step": 1000
},
{
"epoch": 7.1,
"eval_accuracy": 0.95,
"eval_loss": 0.3059749901294708,
"eval_runtime": 93.8808,
"eval_samples_per_second": 1.065,
"eval_steps_per_second": 0.533,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 7.1,
"step": 1000
},
{
"accuracy": 1.0,
"epoch": 8.0,
"step": 1001
},
{
"accuracy": 1.0,
"epoch": 8.0,
"step": 1002
},
{
"accuracy": 1.0,
"epoch": 8.0,
"step": 1003
},
{
"accuracy": 1.0,
"epoch": 8.0,
"step": 1004
},
{
"accuracy": 1.0,
"epoch": 8.0,
"step": 1005
},
{
"accuracy": 1.0,
"epoch": 8.0,
"step": 1006
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1007
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1008
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1009
},
{
"epoch": 8.01,
"grad_norm": 0.0010541501687839627,
"learning_rate": 1.0666666666666667e-05,
"loss": 0.0001,
"step": 1010
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1010
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1011
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1012
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1013
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1014
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1015
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1016
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1017
},
{
"accuracy": 1.0,
"epoch": 8.01,
"step": 1018
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1019
},
{
"epoch": 8.02,
"grad_norm": 0.001377376145683229,
"learning_rate": 1.0222222222222223e-05,
"loss": 0.0001,
"step": 1020
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1020
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1021
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1022
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1023
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1024
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1025
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1026
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1027
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1028
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1029
},
{
"epoch": 8.02,
"grad_norm": 0.0018889001803472638,
"learning_rate": 9.777777777777779e-06,
"loss": 0.0001,
"step": 1030
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1030
},
{
"accuracy": 1.0,
"epoch": 8.02,
"step": 1031
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1032
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1033
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1034
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1035
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1036
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1037
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1038
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1039
},
{
"epoch": 8.03,
"grad_norm": 0.003292502835392952,
"learning_rate": 9.333333333333334e-06,
"loss": 0.0001,
"step": 1040
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1040
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1041
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1042
},
{
"accuracy": 1.0,
"epoch": 8.03,
"step": 1043
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1044
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1045
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1046
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1047
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1048
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1049
},
{
"epoch": 8.04,
"grad_norm": 0.001774812932126224,
"learning_rate": 8.88888888888889e-06,
"loss": 0.0001,
"step": 1050
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1050
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1051
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1052
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1053
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1054
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1055
},
{
"accuracy": 1.0,
"epoch": 8.04,
"step": 1056
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1057
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1058
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1059
},
{
"epoch": 8.05,
"grad_norm": 0.0028727883473038673,
"learning_rate": 8.444444444444446e-06,
"loss": 0.0001,
"step": 1060
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1060
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1061
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1062
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1063
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1064
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1065
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1066
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1067
},
{
"accuracy": 1.0,
"epoch": 8.05,
"step": 1068
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1069
},
{
"epoch": 8.06,
"grad_norm": 0.012458299286663532,
"learning_rate": 8.000000000000001e-06,
"loss": 0.0001,
"step": 1070
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1070
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1071
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1072
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1073
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1074
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1075
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1076
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1077
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1078
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1079
},
{
"epoch": 8.06,
"grad_norm": 0.0028815751429647207,
"learning_rate": 7.555555555555556e-06,
"loss": 0.0001,
"step": 1080
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1080
},
{
"accuracy": 1.0,
"epoch": 8.06,
"step": 1081
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1082
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1083
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1084
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1085
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1086
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1087
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1088
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1089
},
{
"epoch": 8.07,
"grad_norm": 0.0020726954098790884,
"learning_rate": 7.111111111111112e-06,
"loss": 0.0001,
"step": 1090
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1090
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1091
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1092
},
{
"accuracy": 1.0,
"epoch": 8.07,
"step": 1093
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1094
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1095
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1096
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1097
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1098
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1099
},
{
"epoch": 8.08,
"grad_norm": 0.0026790346018970013,
"learning_rate": 6.666666666666667e-06,
"loss": 0.0001,
"step": 1100
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1100
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1101
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1102
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1103
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1104
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1105
},
{
"accuracy": 1.0,
"epoch": 8.08,
"step": 1106
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1107
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1108
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1109
},
{
"epoch": 8.09,
"grad_norm": 0.005608059000223875,
"learning_rate": 6.222222222222222e-06,
"loss": 0.0001,
"step": 1110
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1110
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1111
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1112
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1113
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1114
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1115
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1116
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1117
},
{
"accuracy": 1.0,
"epoch": 8.09,
"step": 1118
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1119
},
{
"epoch": 8.1,
"grad_norm": 0.0019766788464039564,
"learning_rate": 5.777777777777778e-06,
"loss": 0.0001,
"step": 1120
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1120
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1121
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1122
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1123
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1124
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 0.5,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 0.5,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 0.5,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 0.5,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 0.5,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"epoch": 8.1,
"eval_accuracy": 0.95,
"eval_loss": 0.30995601415634155,
"eval_runtime": 94.9524,
"eval_samples_per_second": 1.053,
"eval_steps_per_second": 0.527,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 8.1,
"step": 1125
},
{
"accuracy": 1.0,
"epoch": 9.0,
"step": 1126
},
{
"accuracy": 1.0,
"epoch": 9.0,
"step": 1127
},
{
"accuracy": 1.0,
"epoch": 9.0,
"step": 1128
},
{
"accuracy": 1.0,
"epoch": 9.0,
"step": 1129
},
{
"epoch": 9.0,
"grad_norm": 0.0019209388410672545,
"learning_rate": 5.333333333333334e-06,
"loss": 0.0001,
"step": 1130
},
{
"accuracy": 1.0,
"epoch": 9.0,
"step": 1130
},
{
"accuracy": 1.0,
"epoch": 9.0,
"step": 1131
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1132
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1133
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1134
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1135
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1136
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1137
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1138
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1139
},
{
"epoch": 9.01,
"grad_norm": 0.013591528870165348,
"learning_rate": 4.888888888888889e-06,
"loss": 0.0002,
"step": 1140
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1140
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1141
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1142
},
{
"accuracy": 1.0,
"epoch": 9.01,
"step": 1143
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1144
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1145
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1146
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1147
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1148
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1149
},
{
"epoch": 9.02,
"grad_norm": 0.006724990904331207,
"learning_rate": 4.444444444444445e-06,
"loss": 0.0001,
"step": 1150
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1150
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1151
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1152
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1153
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1154
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1155
},
{
"accuracy": 1.0,
"epoch": 9.02,
"step": 1156
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1157
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1158
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1159
},
{
"epoch": 9.03,
"grad_norm": 0.0018337396904826164,
"learning_rate": 4.000000000000001e-06,
"loss": 0.0001,
"step": 1160
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1160
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1161
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1162
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1163
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1164
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1165
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1166
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1167
},
{
"accuracy": 1.0,
"epoch": 9.03,
"step": 1168
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1169
},
{
"epoch": 9.04,
"grad_norm": 0.0012047181371599436,
"learning_rate": 3.555555555555556e-06,
"loss": 0.0001,
"step": 1170
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1170
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1171
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1172
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1173
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1174
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1175
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1176
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1177
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1178
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1179
},
{
"epoch": 9.04,
"grad_norm": 0.0011284436332061887,
"learning_rate": 3.111111111111111e-06,
"loss": 0.0001,
"step": 1180
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1180
},
{
"accuracy": 1.0,
"epoch": 9.04,
"step": 1181
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1182
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1183
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1184
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1185
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1186
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1187
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1188
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1189
},
{
"epoch": 9.05,
"grad_norm": 0.0070198653265833855,
"learning_rate": 2.666666666666667e-06,
"loss": 0.0001,
"step": 1190
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1190
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1191
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1192
},
{
"accuracy": 1.0,
"epoch": 9.05,
"step": 1193
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1194
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1195
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1196
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1197
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1198
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1199
},
{
"epoch": 9.06,
"grad_norm": 0.001229066401720047,
"learning_rate": 2.2222222222222225e-06,
"loss": 0.0001,
"step": 1200
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1200
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1201
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1202
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1203
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1204
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1205
},
{
"accuracy": 1.0,
"epoch": 9.06,
"step": 1206
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1207
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1208
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1209
},
{
"epoch": 9.07,
"grad_norm": 0.0028434055857360363,
"learning_rate": 1.777777777777778e-06,
"loss": 0.0002,
"step": 1210
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1210
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1211
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1212
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1213
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1214
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1215
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1216
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1217
},
{
"accuracy": 1.0,
"epoch": 9.07,
"step": 1218
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1219
},
{
"epoch": 9.08,
"grad_norm": 0.0011250131065025926,
"learning_rate": 1.3333333333333334e-06,
"loss": 0.0001,
"step": 1220
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1220
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1221
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1222
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1223
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1224
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1225
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1226
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1227
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1228
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1229
},
{
"epoch": 9.08,
"grad_norm": 0.0014523700810968876,
"learning_rate": 8.88888888888889e-07,
"loss": 0.0001,
"step": 1230
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1230
},
{
"accuracy": 1.0,
"epoch": 9.08,
"step": 1231
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1232
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1233
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1234
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1235
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1236
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1237
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1238
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1239
},
{
"epoch": 9.09,
"grad_norm": 0.0014028283767402172,
"learning_rate": 4.444444444444445e-07,
"loss": 0.0001,
"step": 1240
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1240
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1241
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1242
},
{
"accuracy": 1.0,
"epoch": 9.09,
"step": 1243
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1244
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1245
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1246
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1247
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1248
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1249
},
{
"epoch": 9.1,
"grad_norm": 0.0020675130654126406,
"learning_rate": 0.0,
"loss": 0.0001,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"epoch": 9.1,
"eval_accuracy": 0.95,
"eval_loss": 0.31238114833831787,
"eval_runtime": 94.0669,
"eval_samples_per_second": 1.063,
"eval_steps_per_second": 0.532,
"step": 1250
},
{
"epoch": 9.1,
"step": 1250,
"total_flos": 2.7380471118336e+18,
"train_loss": 0.0946084261487238,
"train_runtime": 3799.4888,
"train_samples_per_second": 0.658,
"train_steps_per_second": 0.329
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"epoch": 9.1,
"eval_accuracy": 0.93,
"eval_loss": 0.26954787969589233,
"eval_runtime": 101.2634,
"eval_samples_per_second": 0.988,
"eval_steps_per_second": 0.494,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 0.5,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"accuracy": 1.0,
"epoch": 9.1,
"step": 1250
},
{
"epoch": 9.1,
"eval_accuracy": 0.93,
"eval_loss": 0.26954784989356995,
"eval_runtime": 100.3813,
"eval_samples_per_second": 0.996,
"eval_steps_per_second": 0.498,
"step": 1250
}
],
"logging_steps": 10,
"max_steps": 1250,
"num_input_tokens_seen": 0,
"num_train_epochs": 9223372036854775807,
"save_steps": 500,
"total_flos": 2.7380471118336e+18,
"train_batch_size": 2,
"trial_name": null,
"trial_params": null
}