Upload trainer_state.json
Browse files- trainer_state.json +205 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,205 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 9.0,
|
| 5 |
+
"global_step": 31725,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"learning_rate": 1.8e-05,
|
| 13 |
+
"loss": 0.0026,
|
| 14 |
+
"step": 3525
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.0,
|
| 18 |
+
"eval_EG_f1": 0.9988826815642459,
|
| 19 |
+
"eval_ET_f1": 0.9997551420176297,
|
| 20 |
+
"eval_TE_f1": 0.9980463980463979,
|
| 21 |
+
"eval_loss": 0.000998740317299962,
|
| 22 |
+
"eval_overall_accuracy": 0.9997666459240822,
|
| 23 |
+
"eval_overall_f1": 0.9988922736084187,
|
| 24 |
+
"eval_overall_precision": 0.9988231221876082,
|
| 25 |
+
"eval_overall_recall": 0.998961434604999,
|
| 26 |
+
"eval_runtime": 142.8836,
|
| 27 |
+
"eval_samples_per_second": 43.854,
|
| 28 |
+
"eval_steps_per_second": 2.743,
|
| 29 |
+
"step": 3525
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 2.0,
|
| 33 |
+
"learning_rate": 1.6000000000000003e-05,
|
| 34 |
+
"loss": 0.0012,
|
| 35 |
+
"step": 7050
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 2.0,
|
| 39 |
+
"eval_EG_f1": 0.998084902649218,
|
| 40 |
+
"eval_ET_f1": 0.9993876301285977,
|
| 41 |
+
"eval_TE_f1": 0.9978038067349927,
|
| 42 |
+
"eval_loss": 0.001965524861589074,
|
| 43 |
+
"eval_overall_accuracy": 0.9996499688861232,
|
| 44 |
+
"eval_overall_f1": 0.9983733084138026,
|
| 45 |
+
"eval_overall_precision": 0.9981314878892733,
|
| 46 |
+
"eval_overall_recall": 0.9986152461399986,
|
| 47 |
+
"eval_runtime": 140.6221,
|
| 48 |
+
"eval_samples_per_second": 44.559,
|
| 49 |
+
"eval_steps_per_second": 2.788,
|
| 50 |
+
"step": 7050
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.0,
|
| 54 |
+
"learning_rate": 1.4e-05,
|
| 55 |
+
"loss": 0.0007,
|
| 56 |
+
"step": 10575
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 3.0,
|
| 60 |
+
"eval_EG_f1": 0.9988031596585016,
|
| 61 |
+
"eval_ET_f1": 0.998775710088149,
|
| 62 |
+
"eval_TE_f1": 0.9981682745145928,
|
| 63 |
+
"eval_loss": 0.0016128044808283448,
|
| 64 |
+
"eval_overall_accuracy": 0.9997666459240822,
|
| 65 |
+
"eval_overall_f1": 0.9986154378677742,
|
| 66 |
+
"eval_overall_precision": 0.9984771924967121,
|
| 67 |
+
"eval_overall_recall": 0.9987537215259987,
|
| 68 |
+
"eval_runtime": 141.5414,
|
| 69 |
+
"eval_samples_per_second": 44.27,
|
| 70 |
+
"eval_steps_per_second": 2.77,
|
| 71 |
+
"step": 10575
|
| 72 |
+
},
|
| 73 |
+
{
|
| 74 |
+
"epoch": 4.0,
|
| 75 |
+
"learning_rate": 1.2e-05,
|
| 76 |
+
"loss": 0.0008,
|
| 77 |
+
"step": 14100
|
| 78 |
+
},
|
| 79 |
+
{
|
| 80 |
+
"epoch": 4.0,
|
| 81 |
+
"eval_EG_f1": 0.9992020427705075,
|
| 82 |
+
"eval_ET_f1": 0.9997550820475141,
|
| 83 |
+
"eval_TE_f1": 0.9992673992673993,
|
| 84 |
+
"eval_loss": 0.0006879049469716847,
|
| 85 |
+
"eval_overall_accuracy": 0.9999092511926986,
|
| 86 |
+
"eval_overall_f1": 0.9993769039047354,
|
| 87 |
+
"eval_overall_precision": 0.9993077189338871,
|
| 88 |
+
"eval_overall_recall": 0.9994460984559994,
|
| 89 |
+
"eval_runtime": 141.9363,
|
| 90 |
+
"eval_samples_per_second": 44.147,
|
| 91 |
+
"eval_steps_per_second": 2.762,
|
| 92 |
+
"step": 14100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 5.0,
|
| 96 |
+
"learning_rate": 1e-05,
|
| 97 |
+
"loss": 0.0004,
|
| 98 |
+
"step": 17625
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 5.0,
|
| 102 |
+
"eval_EG_f1": 0.9988828598787105,
|
| 103 |
+
"eval_ET_f1": 0.9996326680543651,
|
| 104 |
+
"eval_TE_f1": 0.9989012330606765,
|
| 105 |
+
"eval_loss": 0.0006999427569098771,
|
| 106 |
+
"eval_overall_accuracy": 0.9998055382700685,
|
| 107 |
+
"eval_overall_f1": 0.9991000346140533,
|
| 108 |
+
"eval_overall_precision": 0.9989617221568492,
|
| 109 |
+
"eval_overall_recall": 0.9992383853769993,
|
| 110 |
+
"eval_runtime": 233.5972,
|
| 111 |
+
"eval_samples_per_second": 26.824,
|
| 112 |
+
"eval_steps_per_second": 1.678,
|
| 113 |
+
"step": 17625
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 6.0,
|
| 117 |
+
"learning_rate": 8.000000000000001e-06,
|
| 118 |
+
"loss": 0.0004,
|
| 119 |
+
"step": 21150
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 6.0,
|
| 123 |
+
"eval_EG_f1": 0.999680817108203,
|
| 124 |
+
"eval_ET_f1": 0.9995101640950281,
|
| 125 |
+
"eval_TE_f1": 0.9996336549029186,
|
| 126 |
+
"eval_loss": 0.0005218852893449366,
|
| 127 |
+
"eval_overall_accuracy": 0.9998962870773699,
|
| 128 |
+
"eval_overall_f1": 0.9996192058711532,
|
| 129 |
+
"eval_overall_precision": 0.9995846026031571,
|
| 130 |
+
"eval_overall_recall": 0.9996538115349997,
|
| 131 |
+
"eval_runtime": 141.8762,
|
| 132 |
+
"eval_samples_per_second": 44.165,
|
| 133 |
+
"eval_steps_per_second": 2.763,
|
| 134 |
+
"step": 21150
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 7.0,
|
| 138 |
+
"learning_rate": 6e-06,
|
| 139 |
+
"loss": 0.0,
|
| 140 |
+
"step": 24675
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 7.0,
|
| 144 |
+
"eval_EG_f1": 0.999361634216406,
|
| 145 |
+
"eval_ET_f1": 0.9997550820475141,
|
| 146 |
+
"eval_TE_f1": 0.9993892756809576,
|
| 147 |
+
"eval_loss": 0.0003737392835319042,
|
| 148 |
+
"eval_overall_accuracy": 0.9999351794233562,
|
| 149 |
+
"eval_overall_f1": 0.9994806993249091,
|
| 150 |
+
"eval_overall_precision": 0.999515302589669,
|
| 151 |
+
"eval_overall_recall": 0.9994460984559994,
|
| 152 |
+
"eval_runtime": 244.7883,
|
| 153 |
+
"eval_samples_per_second": 25.598,
|
| 154 |
+
"eval_steps_per_second": 1.601,
|
| 155 |
+
"step": 24675
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 8.0,
|
| 159 |
+
"learning_rate": 4.000000000000001e-06,
|
| 160 |
+
"loss": 0.0001,
|
| 161 |
+
"step": 28200
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 8.0,
|
| 165 |
+
"eval_EG_f1": 0.9995212256623045,
|
| 166 |
+
"eval_ET_f1": 0.9997550820475141,
|
| 167 |
+
"eval_TE_f1": 0.9997557997557998,
|
| 168 |
+
"eval_loss": 0.00023818030604161322,
|
| 169 |
+
"eval_overall_accuracy": 0.9999481435386849,
|
| 170 |
+
"eval_overall_f1": 0.9996538355026309,
|
| 171 |
+
"eval_overall_precision": 0.9995846313603323,
|
| 172 |
+
"eval_overall_recall": 0.9997230492279997,
|
| 173 |
+
"eval_runtime": 365.8307,
|
| 174 |
+
"eval_samples_per_second": 17.128,
|
| 175 |
+
"eval_steps_per_second": 1.072,
|
| 176 |
+
"step": 28200
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 9.0,
|
| 180 |
+
"learning_rate": 2.0000000000000003e-06,
|
| 181 |
+
"loss": 0.0001,
|
| 182 |
+
"step": 31725
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 9.0,
|
| 186 |
+
"eval_EG_f1": 0.9988828598787105,
|
| 187 |
+
"eval_ET_f1": 0.9997550820475141,
|
| 188 |
+
"eval_TE_f1": 0.9992677568952892,
|
| 189 |
+
"eval_loss": 0.001098370412364602,
|
| 190 |
+
"eval_overall_accuracy": 0.9997925741547397,
|
| 191 |
+
"eval_overall_f1": 0.9992385435414648,
|
| 192 |
+
"eval_overall_precision": 0.9990310748148661,
|
| 193 |
+
"eval_overall_recall": 0.9994460984559994,
|
| 194 |
+
"eval_runtime": 243.6202,
|
| 195 |
+
"eval_samples_per_second": 25.72,
|
| 196 |
+
"eval_steps_per_second": 1.609,
|
| 197 |
+
"step": 31725
|
| 198 |
+
}
|
| 199 |
+
],
|
| 200 |
+
"max_steps": 35250,
|
| 201 |
+
"num_train_epochs": 10,
|
| 202 |
+
"total_flos": 7174982911185900.0,
|
| 203 |
+
"trial_name": null,
|
| 204 |
+
"trial_params": null
|
| 205 |
+
}
|