Upload trainer_state.json
Browse files- trainer_state.json +247 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,247 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 5.0,
|
| 5 |
+
"global_step": 11815,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.42,
|
| 12 |
+
"learning_rate": 1.9153618281845116e-05,
|
| 13 |
+
"loss": 0.0365,
|
| 14 |
+
"step": 1000
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.42,
|
| 18 |
+
"eval_EG_f1": 0.9898869720404521,
|
| 19 |
+
"eval_ET_f1": 0.9967868618350589,
|
| 20 |
+
"eval_TE_f1": 0.988691822402095,
|
| 21 |
+
"eval_loss": 0.0057431175373494625,
|
| 22 |
+
"eval_overall_accuracy": 0.998336295769641,
|
| 23 |
+
"eval_overall_f1": 0.9917886469118172,
|
| 24 |
+
"eval_overall_precision": 0.9914346895074947,
|
| 25 |
+
"eval_overall_recall": 0.9921428571428571,
|
| 26 |
+
"eval_runtime": 105.4808,
|
| 27 |
+
"eval_samples_per_second": 39.818,
|
| 28 |
+
"eval_steps_per_second": 2.493,
|
| 29 |
+
"step": 1000
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.85,
|
| 33 |
+
"learning_rate": 1.8307236563690226e-05,
|
| 34 |
+
"loss": 0.0079,
|
| 35 |
+
"step": 2000
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.85,
|
| 39 |
+
"eval_EG_f1": 0.9944080904223678,
|
| 40 |
+
"eval_ET_f1": 0.9988095238095238,
|
| 41 |
+
"eval_TE_f1": 0.9936957297490187,
|
| 42 |
+
"eval_loss": 0.003063528100028634,
|
| 43 |
+
"eval_overall_accuracy": 0.9991894774262353,
|
| 44 |
+
"eval_overall_f1": 0.9956369982547992,
|
| 45 |
+
"eval_overall_precision": 0.995163336504916,
|
| 46 |
+
"eval_overall_recall": 0.9961111111111111,
|
| 47 |
+
"eval_runtime": 103.5359,
|
| 48 |
+
"eval_samples_per_second": 40.566,
|
| 49 |
+
"eval_steps_per_second": 2.54,
|
| 50 |
+
"step": 2000
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 1.27,
|
| 54 |
+
"learning_rate": 1.7460854845535337e-05,
|
| 55 |
+
"loss": 0.0043,
|
| 56 |
+
"step": 3000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 1.27,
|
| 60 |
+
"eval_EG_f1": 0.997381575815282,
|
| 61 |
+
"eval_ET_f1": 0.9983337300642704,
|
| 62 |
+
"eval_TE_f1": 0.9963073257891601,
|
| 63 |
+
"eval_loss": 0.0027184109203517437,
|
| 64 |
+
"eval_overall_accuracy": 0.999431212228937,
|
| 65 |
+
"eval_overall_f1": 0.9973411643319179,
|
| 66 |
+
"eval_overall_precision": 0.997380744503532,
|
| 67 |
+
"eval_overall_recall": 0.9973015873015874,
|
| 68 |
+
"eval_runtime": 102.7681,
|
| 69 |
+
"eval_samples_per_second": 40.869,
|
| 70 |
+
"eval_steps_per_second": 2.559,
|
| 71 |
+
"step": 3000
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 1.69,
|
| 75 |
+
"learning_rate": 1.661447312738045e-05,
|
| 76 |
+
"loss": 0.0021,
|
| 77 |
+
"step": 4000
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 1.69,
|
| 81 |
+
"eval_EG_f1": 0.9942870745060699,
|
| 82 |
+
"eval_ET_f1": 0.9983333333333333,
|
| 83 |
+
"eval_TE_f1": 0.9942857142857143,
|
| 84 |
+
"eval_loss": 0.00385509361512959,
|
| 85 |
+
"eval_overall_accuracy": 0.9992179168147884,
|
| 86 |
+
"eval_overall_f1": 0.9956352670422981,
|
| 87 |
+
"eval_overall_precision": 0.9955562609109665,
|
| 88 |
+
"eval_overall_recall": 0.9957142857142857,
|
| 89 |
+
"eval_runtime": 103.0813,
|
| 90 |
+
"eval_samples_per_second": 40.745,
|
| 91 |
+
"eval_steps_per_second": 2.551,
|
| 92 |
+
"step": 4000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 2.12,
|
| 96 |
+
"learning_rate": 1.5768091409225562e-05,
|
| 97 |
+
"loss": 0.002,
|
| 98 |
+
"step": 5000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 2.12,
|
| 102 |
+
"eval_EG_f1": 0.9958348208972985,
|
| 103 |
+
"eval_ET_f1": 0.9990476190476191,
|
| 104 |
+
"eval_TE_f1": 0.9957142857142857,
|
| 105 |
+
"eval_loss": 0.0032494200859218836,
|
| 106 |
+
"eval_overall_accuracy": 0.9993174546747245,
|
| 107 |
+
"eval_overall_f1": 0.996865452525493,
|
| 108 |
+
"eval_overall_precision": 0.9967468063159565,
|
| 109 |
+
"eval_overall_recall": 0.996984126984127,
|
| 110 |
+
"eval_runtime": 102.2205,
|
| 111 |
+
"eval_samples_per_second": 41.088,
|
| 112 |
+
"eval_steps_per_second": 2.573,
|
| 113 |
+
"step": 5000
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 2.54,
|
| 117 |
+
"learning_rate": 1.4921709691070674e-05,
|
| 118 |
+
"loss": 0.0013,
|
| 119 |
+
"step": 6000
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 2.54,
|
| 123 |
+
"eval_EG_f1": 0.9952403617325083,
|
| 124 |
+
"eval_ET_f1": 0.998928698964409,
|
| 125 |
+
"eval_TE_f1": 0.994880342897964,
|
| 126 |
+
"eval_loss": 0.003066167002543807,
|
| 127 |
+
"eval_overall_accuracy": 0.9992605758976182,
|
| 128 |
+
"eval_overall_f1": 0.9963497857482939,
|
| 129 |
+
"eval_overall_precision": 0.9961916851793081,
|
| 130 |
+
"eval_overall_recall": 0.9965079365079365,
|
| 131 |
+
"eval_runtime": 102.0329,
|
| 132 |
+
"eval_samples_per_second": 41.163,
|
| 133 |
+
"eval_steps_per_second": 2.578,
|
| 134 |
+
"step": 6000
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 2.96,
|
| 138 |
+
"learning_rate": 1.4075327972915787e-05,
|
| 139 |
+
"loss": 0.002,
|
| 140 |
+
"step": 7000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 2.96,
|
| 144 |
+
"eval_EG_f1": 0.9971428571428571,
|
| 145 |
+
"eval_ET_f1": 0.9990476190476191,
|
| 146 |
+
"eval_TE_f1": 0.9971428571428571,
|
| 147 |
+
"eval_loss": 0.00179725990165025,
|
| 148 |
+
"eval_overall_accuracy": 0.9995591894774263,
|
| 149 |
+
"eval_overall_f1": 0.9977777777777778,
|
| 150 |
+
"eval_overall_precision": 0.9977777777777778,
|
| 151 |
+
"eval_overall_recall": 0.9977777777777778,
|
| 152 |
+
"eval_runtime": 104.5567,
|
| 153 |
+
"eval_samples_per_second": 40.17,
|
| 154 |
+
"eval_steps_per_second": 2.515,
|
| 155 |
+
"step": 7000
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 3.39,
|
| 159 |
+
"learning_rate": 1.3228946254760899e-05,
|
| 160 |
+
"loss": 0.0009,
|
| 161 |
+
"step": 8000
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 3.39,
|
| 165 |
+
"eval_EG_f1": 0.9976196143775291,
|
| 166 |
+
"eval_ET_f1": 0.998690632067611,
|
| 167 |
+
"eval_TE_f1": 0.996904761904762,
|
| 168 |
+
"eval_loss": 0.0014693811535835266,
|
| 169 |
+
"eval_overall_accuracy": 0.999701386420192,
|
| 170 |
+
"eval_overall_f1": 0.9977383644804191,
|
| 171 |
+
"eval_overall_precision": 0.9976196143775291,
|
| 172 |
+
"eval_overall_recall": 0.9978571428571429,
|
| 173 |
+
"eval_runtime": 102.0191,
|
| 174 |
+
"eval_samples_per_second": 41.169,
|
| 175 |
+
"eval_steps_per_second": 2.578,
|
| 176 |
+
"step": 8000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 3.81,
|
| 180 |
+
"learning_rate": 1.238256453660601e-05,
|
| 181 |
+
"loss": 0.0008,
|
| 182 |
+
"step": 9000
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 3.81,
|
| 186 |
+
"eval_EG_f1": 0.997381575815282,
|
| 187 |
+
"eval_ET_f1": 0.9992857142857143,
|
| 188 |
+
"eval_TE_f1": 0.9973809523809524,
|
| 189 |
+
"eval_loss": 0.0014952768106013536,
|
| 190 |
+
"eval_overall_accuracy": 0.9997298258087451,
|
| 191 |
+
"eval_overall_f1": 0.998016030473772,
|
| 192 |
+
"eval_overall_precision": 0.9979368354229488,
|
| 193 |
+
"eval_overall_recall": 0.9980952380952381,
|
| 194 |
+
"eval_runtime": 102.1451,
|
| 195 |
+
"eval_samples_per_second": 41.118,
|
| 196 |
+
"eval_steps_per_second": 2.575,
|
| 197 |
+
"step": 9000
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 4.23,
|
| 201 |
+
"learning_rate": 1.1536182818451122e-05,
|
| 202 |
+
"loss": 0.0008,
|
| 203 |
+
"step": 10000
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 4.23,
|
| 207 |
+
"eval_EG_f1": 0.9976190476190476,
|
| 208 |
+
"eval_ET_f1": 0.9983333333333333,
|
| 209 |
+
"eval_TE_f1": 0.9972615787593762,
|
| 210 |
+
"eval_loss": 0.00143534317612648,
|
| 211 |
+
"eval_overall_accuracy": 0.9997298258087451,
|
| 212 |
+
"eval_overall_f1": 0.9977380054764079,
|
| 213 |
+
"eval_overall_precision": 0.9977776013969363,
|
| 214 |
+
"eval_overall_recall": 0.9976984126984128,
|
| 215 |
+
"eval_runtime": 103.0577,
|
| 216 |
+
"eval_samples_per_second": 40.754,
|
| 217 |
+
"eval_steps_per_second": 2.552,
|
| 218 |
+
"step": 10000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 4.66,
|
| 222 |
+
"learning_rate": 1.0689801100296236e-05,
|
| 223 |
+
"loss": 0.0005,
|
| 224 |
+
"step": 11000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 4.66,
|
| 228 |
+
"eval_EG_f1": 0.9974997023455173,
|
| 229 |
+
"eval_ET_f1": 0.9995238095238095,
|
| 230 |
+
"eval_TE_f1": 0.9973809523809524,
|
| 231 |
+
"eval_loss": 0.0019883255008608103,
|
| 232 |
+
"eval_overall_accuracy": 0.9996729470316388,
|
| 233 |
+
"eval_overall_f1": 0.9981348466208977,
|
| 234 |
+
"eval_overall_precision": 0.9981744582903405,
|
| 235 |
+
"eval_overall_recall": 0.9980952380952381,
|
| 236 |
+
"eval_runtime": 102.2987,
|
| 237 |
+
"eval_samples_per_second": 41.056,
|
| 238 |
+
"eval_steps_per_second": 2.571,
|
| 239 |
+
"step": 11000
|
| 240 |
+
}
|
| 241 |
+
],
|
| 242 |
+
"max_steps": 23630,
|
| 243 |
+
"num_train_epochs": 10,
|
| 244 |
+
"total_flos": 2874592437250464.0,
|
| 245 |
+
"trial_name": null,
|
| 246 |
+
"trial_params": null
|
| 247 |
+
}
|