wesley7137's picture
Upload 14 files
9947c3a
{
"best_metric": null,
"best_model_checkpoint": null,
"epoch": 0.0060759493670886075,
"eval_steps": 500,
"global_step": 300,
"is_hyper_param_search": false,
"is_local_process_zero": true,
"is_world_process_zero": true,
"log_history": [
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.0959,
"step": 1
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.1546,
"step": 2
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.3479,
"step": 3
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.2503,
"step": 4
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.3416,
"step": 5
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.3484,
"step": 6
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.3347,
"step": 7
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.2724,
"step": 8
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.2079,
"step": 9
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.2547,
"step": 10
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.2018,
"step": 11
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.2583,
"step": 12
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.284,
"step": 13
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.0542,
"step": 14
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.1568,
"step": 15
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.017,
"step": 16
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 1.0164,
"step": 17
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.9156,
"step": 18
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.9252,
"step": 19
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.7876,
"step": 20
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.7165,
"step": 21
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6807,
"step": 22
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6941,
"step": 23
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6218,
"step": 24
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6979,
"step": 25
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6974,
"step": 26
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6624,
"step": 27
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6455,
"step": 28
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.701,
"step": 29
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6802,
"step": 30
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6353,
"step": 31
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5297,
"step": 32
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5491,
"step": 33
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6545,
"step": 34
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6168,
"step": 35
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6217,
"step": 36
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6234,
"step": 37
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5422,
"step": 38
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.7242,
"step": 39
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6204,
"step": 40
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5836,
"step": 41
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5765,
"step": 42
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5622,
"step": 43
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5649,
"step": 44
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6327,
"step": 45
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.488,
"step": 46
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5458,
"step": 47
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5693,
"step": 48
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5141,
"step": 49
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5334,
"step": 50
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5639,
"step": 51
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6323,
"step": 52
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6546,
"step": 53
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6939,
"step": 54
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6168,
"step": 55
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.544,
"step": 56
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5988,
"step": 57
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5091,
"step": 58
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5366,
"step": 59
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5133,
"step": 60
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5831,
"step": 61
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6005,
"step": 62
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5529,
"step": 63
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5784,
"step": 64
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5777,
"step": 65
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5436,
"step": 66
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5509,
"step": 67
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6578,
"step": 68
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5672,
"step": 69
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.507,
"step": 70
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6225,
"step": 71
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6094,
"step": 72
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5362,
"step": 73
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5388,
"step": 74
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5537,
"step": 75
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5853,
"step": 76
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4963,
"step": 77
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.55,
"step": 78
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5461,
"step": 79
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.541,
"step": 80
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.622,
"step": 81
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5707,
"step": 82
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5854,
"step": 83
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5205,
"step": 84
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4645,
"step": 85
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5519,
"step": 86
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5284,
"step": 87
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4823,
"step": 88
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5102,
"step": 89
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5052,
"step": 90
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5103,
"step": 91
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4865,
"step": 92
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5144,
"step": 93
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4552,
"step": 94
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.502,
"step": 95
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4691,
"step": 96
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4939,
"step": 97
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5012,
"step": 98
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4274,
"step": 99
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4208,
"step": 100
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5306,
"step": 101
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5154,
"step": 102
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6008,
"step": 103
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5512,
"step": 104
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4762,
"step": 105
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6241,
"step": 106
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5532,
"step": 107
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4581,
"step": 108
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5168,
"step": 109
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5216,
"step": 110
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4855,
"step": 111
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4711,
"step": 112
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4754,
"step": 113
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4509,
"step": 114
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4848,
"step": 115
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.513,
"step": 116
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5992,
"step": 117
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5308,
"step": 118
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.582,
"step": 119
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5188,
"step": 120
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5254,
"step": 121
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.498,
"step": 122
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5578,
"step": 123
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4419,
"step": 124
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5088,
"step": 125
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4544,
"step": 126
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5729,
"step": 127
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5504,
"step": 128
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5765,
"step": 129
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4918,
"step": 130
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4546,
"step": 131
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4365,
"step": 132
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4332,
"step": 133
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4494,
"step": 134
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5548,
"step": 135
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5012,
"step": 136
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5329,
"step": 137
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4324,
"step": 138
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5084,
"step": 139
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.551,
"step": 140
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4932,
"step": 141
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4554,
"step": 142
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4614,
"step": 143
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.494,
"step": 144
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4955,
"step": 145
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5002,
"step": 146
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4597,
"step": 147
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5241,
"step": 148
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.3929,
"step": 149
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.3445,
"step": 150
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.58,
"step": 151
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5182,
"step": 152
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5274,
"step": 153
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5099,
"step": 154
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5458,
"step": 155
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.474,
"step": 156
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4769,
"step": 157
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4472,
"step": 158
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.514,
"step": 159
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4605,
"step": 160
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4921,
"step": 161
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5079,
"step": 162
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5588,
"step": 163
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4637,
"step": 164
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4842,
"step": 165
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4781,
"step": 166
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5443,
"step": 167
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5679,
"step": 168
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.523,
"step": 169
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5176,
"step": 170
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4599,
"step": 171
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5255,
"step": 172
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5217,
"step": 173
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4861,
"step": 174
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5303,
"step": 175
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5442,
"step": 176
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4619,
"step": 177
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4551,
"step": 178
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5021,
"step": 179
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4832,
"step": 180
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5725,
"step": 181
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4902,
"step": 182
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.506,
"step": 183
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5055,
"step": 184
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5262,
"step": 185
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5572,
"step": 186
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4542,
"step": 187
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5356,
"step": 188
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.493,
"step": 189
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4752,
"step": 190
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4753,
"step": 191
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4706,
"step": 192
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4993,
"step": 193
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4781,
"step": 194
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4874,
"step": 195
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.549,
"step": 196
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4599,
"step": 197
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.3982,
"step": 198
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.36,
"step": 199
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.3572,
"step": 200
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4734,
"step": 201
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5174,
"step": 202
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5445,
"step": 203
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5521,
"step": 204
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4273,
"step": 205
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4393,
"step": 206
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4361,
"step": 207
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4453,
"step": 208
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4597,
"step": 209
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5559,
"step": 210
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5015,
"step": 211
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.6537,
"step": 212
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5598,
"step": 213
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4147,
"step": 214
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4752,
"step": 215
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5194,
"step": 216
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4395,
"step": 217
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5379,
"step": 218
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4989,
"step": 219
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4861,
"step": 220
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4913,
"step": 221
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5293,
"step": 222
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5284,
"step": 223
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5284,
"step": 224
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4723,
"step": 225
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4811,
"step": 226
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4969,
"step": 227
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.632,
"step": 228
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.507,
"step": 229
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5252,
"step": 230
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5073,
"step": 231
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5142,
"step": 232
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.472,
"step": 233
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5791,
"step": 234
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5082,
"step": 235
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4911,
"step": 236
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4799,
"step": 237
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4584,
"step": 238
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.5571,
"step": 239
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4438,
"step": 240
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.491,
"step": 241
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4601,
"step": 242
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.442,
"step": 243
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4082,
"step": 244
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4723,
"step": 245
},
{
"epoch": 0.0,
"learning_rate": 0.0002,
"loss": 0.4398,
"step": 246
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4463,
"step": 247
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.3909,
"step": 248
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4419,
"step": 249
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.3694,
"step": 250
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5897,
"step": 251
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5437,
"step": 252
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4756,
"step": 253
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4621,
"step": 254
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4806,
"step": 255
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4978,
"step": 256
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.3948,
"step": 257
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4398,
"step": 258
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4413,
"step": 259
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4838,
"step": 260
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4948,
"step": 261
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4863,
"step": 262
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5742,
"step": 263
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4775,
"step": 264
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4741,
"step": 265
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4886,
"step": 266
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5147,
"step": 267
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4474,
"step": 268
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4613,
"step": 269
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4768,
"step": 270
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5597,
"step": 271
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5458,
"step": 272
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5718,
"step": 273
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4413,
"step": 274
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4806,
"step": 275
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5187,
"step": 276
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5801,
"step": 277
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4721,
"step": 278
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5101,
"step": 279
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4714,
"step": 280
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4709,
"step": 281
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5251,
"step": 282
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4749,
"step": 283
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4328,
"step": 284
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.516,
"step": 285
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4037,
"step": 286
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4403,
"step": 287
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4025,
"step": 288
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.482,
"step": 289
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4305,
"step": 290
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4409,
"step": 291
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4818,
"step": 292
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4425,
"step": 293
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.472,
"step": 294
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.5274,
"step": 295
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4769,
"step": 296
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4344,
"step": 297
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4103,
"step": 298
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.4382,
"step": 299
},
{
"epoch": 0.01,
"learning_rate": 0.0002,
"loss": 0.3984,
"step": 300
}
],
"logging_steps": 1,
"max_steps": 10000,
"num_input_tokens_seen": 0,
"num_train_epochs": 1,
"save_steps": 50,
"total_flos": 3.3099085906182144e+16,
"train_batch_size": 4,
"trial_name": null,
"trial_params": null
}