Commit
·
385bd88
1
Parent(s):
ce7b497
Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +253 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,253 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 18.06451612903226,
|
| 5 |
+
"global_step": 140,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.9,
|
| 12 |
+
"eval_accuracy": 0.7432239657631954,
|
| 13 |
+
"eval_f1": 0.0,
|
| 14 |
+
"eval_loss": 1.1863211393356323,
|
| 15 |
+
"eval_precision": 0.0,
|
| 16 |
+
"eval_recall": 0.0,
|
| 17 |
+
"eval_runtime": 0.0695,
|
| 18 |
+
"eval_samples_per_second": 115.178,
|
| 19 |
+
"eval_steps_per_second": 14.397,
|
| 20 |
+
"step": 7
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.94,
|
| 24 |
+
"eval_accuracy": 0.7931526390870185,
|
| 25 |
+
"eval_f1": 0.25925925925925924,
|
| 26 |
+
"eval_loss": 0.7849462628364563,
|
| 27 |
+
"eval_precision": 0.34146341463414637,
|
| 28 |
+
"eval_recall": 0.208955223880597,
|
| 29 |
+
"eval_runtime": 0.113,
|
| 30 |
+
"eval_samples_per_second": 70.799,
|
| 31 |
+
"eval_steps_per_second": 8.85,
|
| 32 |
+
"step": 15
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.97,
|
| 36 |
+
"eval_accuracy": 0.8487874465049928,
|
| 37 |
+
"eval_f1": 0.484375,
|
| 38 |
+
"eval_loss": 0.5964732766151428,
|
| 39 |
+
"eval_precision": 0.5081967213114754,
|
| 40 |
+
"eval_recall": 0.4626865671641791,
|
| 41 |
+
"eval_runtime": 0.0576,
|
| 42 |
+
"eval_samples_per_second": 139.001,
|
| 43 |
+
"eval_steps_per_second": 17.375,
|
| 44 |
+
"step": 23
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.0,
|
| 48 |
+
"eval_accuracy": 0.8944365192582026,
|
| 49 |
+
"eval_f1": 0.6277372262773724,
|
| 50 |
+
"eval_loss": 0.5038930773735046,
|
| 51 |
+
"eval_precision": 0.6142857142857143,
|
| 52 |
+
"eval_recall": 0.6417910447761194,
|
| 53 |
+
"eval_runtime": 0.0605,
|
| 54 |
+
"eval_samples_per_second": 132.311,
|
| 55 |
+
"eval_steps_per_second": 16.539,
|
| 56 |
+
"step": 31
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 4.9,
|
| 60 |
+
"eval_accuracy": 0.891583452211127,
|
| 61 |
+
"eval_f1": 0.6293706293706294,
|
| 62 |
+
"eval_loss": 0.45688581466674805,
|
| 63 |
+
"eval_precision": 0.5921052631578947,
|
| 64 |
+
"eval_recall": 0.6716417910447762,
|
| 65 |
+
"eval_runtime": 0.0797,
|
| 66 |
+
"eval_samples_per_second": 100.319,
|
| 67 |
+
"eval_steps_per_second": 12.54,
|
| 68 |
+
"step": 38
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 5.94,
|
| 72 |
+
"eval_accuracy": 0.8887303851640513,
|
| 73 |
+
"eval_f1": 0.5899280575539567,
|
| 74 |
+
"eval_loss": 0.43453264236450195,
|
| 75 |
+
"eval_precision": 0.5694444444444444,
|
| 76 |
+
"eval_recall": 0.6119402985074627,
|
| 77 |
+
"eval_runtime": 0.0617,
|
| 78 |
+
"eval_samples_per_second": 129.645,
|
| 79 |
+
"eval_steps_per_second": 16.206,
|
| 80 |
+
"step": 46
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 6.97,
|
| 84 |
+
"eval_accuracy": 0.9072753209700428,
|
| 85 |
+
"eval_f1": 0.6225165562913907,
|
| 86 |
+
"eval_loss": 0.3776801824569702,
|
| 87 |
+
"eval_precision": 0.5595238095238095,
|
| 88 |
+
"eval_recall": 0.7014925373134329,
|
| 89 |
+
"eval_runtime": 0.0586,
|
| 90 |
+
"eval_samples_per_second": 136.463,
|
| 91 |
+
"eval_steps_per_second": 17.058,
|
| 92 |
+
"step": 54
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 8.0,
|
| 96 |
+
"eval_accuracy": 0.9115549215406562,
|
| 97 |
+
"eval_f1": 0.6533333333333333,
|
| 98 |
+
"eval_loss": 0.379366397857666,
|
| 99 |
+
"eval_precision": 0.5903614457831325,
|
| 100 |
+
"eval_recall": 0.7313432835820896,
|
| 101 |
+
"eval_runtime": 0.0869,
|
| 102 |
+
"eval_samples_per_second": 92.01,
|
| 103 |
+
"eval_steps_per_second": 11.501,
|
| 104 |
+
"step": 62
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 8.9,
|
| 108 |
+
"eval_accuracy": 0.920114122681883,
|
| 109 |
+
"eval_f1": 0.7200000000000001,
|
| 110 |
+
"eval_loss": 0.3528553247451782,
|
| 111 |
+
"eval_precision": 0.6506024096385542,
|
| 112 |
+
"eval_recall": 0.8059701492537313,
|
| 113 |
+
"eval_runtime": 0.0586,
|
| 114 |
+
"eval_samples_per_second": 136.489,
|
| 115 |
+
"eval_steps_per_second": 17.061,
|
| 116 |
+
"step": 69
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 9.94,
|
| 120 |
+
"eval_accuracy": 0.920114122681883,
|
| 121 |
+
"eval_f1": 0.7172413793103448,
|
| 122 |
+
"eval_loss": 0.38467955589294434,
|
| 123 |
+
"eval_precision": 0.6666666666666666,
|
| 124 |
+
"eval_recall": 0.7761194029850746,
|
| 125 |
+
"eval_runtime": 0.0815,
|
| 126 |
+
"eval_samples_per_second": 98.125,
|
| 127 |
+
"eval_steps_per_second": 12.266,
|
| 128 |
+
"step": 77
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 10.97,
|
| 132 |
+
"eval_accuracy": 0.920114122681883,
|
| 133 |
+
"eval_f1": 0.7346938775510204,
|
| 134 |
+
"eval_loss": 0.3774794638156891,
|
| 135 |
+
"eval_precision": 0.675,
|
| 136 |
+
"eval_recall": 0.8059701492537313,
|
| 137 |
+
"eval_runtime": 0.0582,
|
| 138 |
+
"eval_samples_per_second": 137.398,
|
| 139 |
+
"eval_steps_per_second": 17.175,
|
| 140 |
+
"step": 85
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 12.0,
|
| 144 |
+
"eval_accuracy": 0.920114122681883,
|
| 145 |
+
"eval_f1": 0.7464788732394366,
|
| 146 |
+
"eval_loss": 0.3588719367980957,
|
| 147 |
+
"eval_precision": 0.7066666666666667,
|
| 148 |
+
"eval_recall": 0.7910447761194029,
|
| 149 |
+
"eval_runtime": 0.0607,
|
| 150 |
+
"eval_samples_per_second": 131.854,
|
| 151 |
+
"eval_steps_per_second": 16.482,
|
| 152 |
+
"step": 93
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 12.9,
|
| 156 |
+
"eval_accuracy": 0.9243937232524965,
|
| 157 |
+
"eval_f1": 0.7692307692307693,
|
| 158 |
+
"eval_loss": 0.358673632144928,
|
| 159 |
+
"eval_precision": 0.7236842105263158,
|
| 160 |
+
"eval_recall": 0.8208955223880597,
|
| 161 |
+
"eval_runtime": 0.0777,
|
| 162 |
+
"eval_samples_per_second": 102.924,
|
| 163 |
+
"eval_steps_per_second": 12.866,
|
| 164 |
+
"step": 100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 13.94,
|
| 168 |
+
"eval_accuracy": 0.9144079885877318,
|
| 169 |
+
"eval_f1": 0.6973684210526316,
|
| 170 |
+
"eval_loss": 0.3642039895057678,
|
| 171 |
+
"eval_precision": 0.6235294117647059,
|
| 172 |
+
"eval_recall": 0.7910447761194029,
|
| 173 |
+
"eval_runtime": 0.0603,
|
| 174 |
+
"eval_samples_per_second": 132.64,
|
| 175 |
+
"eval_steps_per_second": 16.58,
|
| 176 |
+
"step": 108
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 14.97,
|
| 180 |
+
"eval_accuracy": 0.9144079885877318,
|
| 181 |
+
"eval_f1": 0.7066666666666667,
|
| 182 |
+
"eval_loss": 0.3640976548194885,
|
| 183 |
+
"eval_precision": 0.6385542168674698,
|
| 184 |
+
"eval_recall": 0.7910447761194029,
|
| 185 |
+
"eval_runtime": 0.1061,
|
| 186 |
+
"eval_samples_per_second": 75.406,
|
| 187 |
+
"eval_steps_per_second": 9.426,
|
| 188 |
+
"step": 116
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 16.0,
|
| 192 |
+
"eval_accuracy": 0.9172610556348074,
|
| 193 |
+
"eval_f1": 0.7534246575342466,
|
| 194 |
+
"eval_loss": 0.3683820962905884,
|
| 195 |
+
"eval_precision": 0.6962025316455697,
|
| 196 |
+
"eval_recall": 0.8208955223880597,
|
| 197 |
+
"eval_runtime": 0.0609,
|
| 198 |
+
"eval_samples_per_second": 131.457,
|
| 199 |
+
"eval_steps_per_second": 16.432,
|
| 200 |
+
"step": 124
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 16.9,
|
| 204 |
+
"eval_accuracy": 0.9172610556348074,
|
| 205 |
+
"eval_f1": 0.7534246575342466,
|
| 206 |
+
"eval_loss": 0.3730718493461609,
|
| 207 |
+
"eval_precision": 0.6962025316455697,
|
| 208 |
+
"eval_recall": 0.8208955223880597,
|
| 209 |
+
"eval_runtime": 0.0806,
|
| 210 |
+
"eval_samples_per_second": 99.214,
|
| 211 |
+
"eval_steps_per_second": 12.402,
|
| 212 |
+
"step": 131
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 17.94,
|
| 216 |
+
"eval_accuracy": 0.9172610556348074,
|
| 217 |
+
"eval_f1": 0.738255033557047,
|
| 218 |
+
"eval_loss": 0.3693277835845947,
|
| 219 |
+
"eval_precision": 0.6707317073170732,
|
| 220 |
+
"eval_recall": 0.8208955223880597,
|
| 221 |
+
"eval_runtime": 0.0628,
|
| 222 |
+
"eval_samples_per_second": 127.442,
|
| 223 |
+
"eval_steps_per_second": 15.93,
|
| 224 |
+
"step": 139
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 18.06,
|
| 228 |
+
"eval_accuracy": 0.9172610556348074,
|
| 229 |
+
"eval_f1": 0.7333333333333333,
|
| 230 |
+
"eval_loss": 0.36928027868270874,
|
| 231 |
+
"eval_precision": 0.6626506024096386,
|
| 232 |
+
"eval_recall": 0.8208955223880597,
|
| 233 |
+
"eval_runtime": 0.0817,
|
| 234 |
+
"eval_samples_per_second": 97.898,
|
| 235 |
+
"eval_steps_per_second": 12.237,
|
| 236 |
+
"step": 140
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 18.06,
|
| 240 |
+
"step": 140,
|
| 241 |
+
"total_flos": 142823207204352.0,
|
| 242 |
+
"train_loss": 0.284281131199428,
|
| 243 |
+
"train_runtime": 653.9512,
|
| 244 |
+
"train_samples_per_second": 3.701,
|
| 245 |
+
"train_steps_per_second": 0.214
|
| 246 |
+
}
|
| 247 |
+
],
|
| 248 |
+
"max_steps": 140,
|
| 249 |
+
"num_train_epochs": 20,
|
| 250 |
+
"total_flos": 142823207204352.0,
|
| 251 |
+
"trial_name": null,
|
| 252 |
+
"trial_params": null
|
| 253 |
+
}
|