Commit
•
1aa6bfe
1
Parent(s):
ffa8e4a
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": 0.7640449438202247,
|
| 13 |
+
"eval_f1": 0.0,
|
| 14 |
+
"eval_loss": 1.1711010932922363,
|
| 15 |
+
"eval_precision": 0.0,
|
| 16 |
+
"eval_recall": 0.0,
|
| 17 |
+
"eval_runtime": 0.0905,
|
| 18 |
+
"eval_samples_per_second": 88.431,
|
| 19 |
+
"eval_steps_per_second": 11.054,
|
| 20 |
+
"step": 7
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 1.94,
|
| 24 |
+
"eval_accuracy": 0.7640449438202247,
|
| 25 |
+
"eval_f1": 0.0,
|
| 26 |
+
"eval_loss": 0.8495532274246216,
|
| 27 |
+
"eval_precision": 0.0,
|
| 28 |
+
"eval_recall": 0.0,
|
| 29 |
+
"eval_runtime": 0.0887,
|
| 30 |
+
"eval_samples_per_second": 90.205,
|
| 31 |
+
"eval_steps_per_second": 11.276,
|
| 32 |
+
"step": 15
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.97,
|
| 36 |
+
"eval_accuracy": 0.846441947565543,
|
| 37 |
+
"eval_f1": 0.36,
|
| 38 |
+
"eval_loss": 0.6479682922363281,
|
| 39 |
+
"eval_precision": 0.34177215189873417,
|
| 40 |
+
"eval_recall": 0.38028169014084506,
|
| 41 |
+
"eval_runtime": 0.0624,
|
| 42 |
+
"eval_samples_per_second": 128.122,
|
| 43 |
+
"eval_steps_per_second": 16.015,
|
| 44 |
+
"step": 23
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 4.0,
|
| 48 |
+
"eval_accuracy": 0.8614232209737828,
|
| 49 |
+
"eval_f1": 0.4657534246575343,
|
| 50 |
+
"eval_loss": 0.5528039932250977,
|
| 51 |
+
"eval_precision": 0.4533333333333333,
|
| 52 |
+
"eval_recall": 0.4788732394366197,
|
| 53 |
+
"eval_runtime": 0.0609,
|
| 54 |
+
"eval_samples_per_second": 131.435,
|
| 55 |
+
"eval_steps_per_second": 16.429,
|
| 56 |
+
"step": 31
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 4.9,
|
| 60 |
+
"eval_accuracy": 0.8876404494382022,
|
| 61 |
+
"eval_f1": 0.543046357615894,
|
| 62 |
+
"eval_loss": 0.5004148483276367,
|
| 63 |
+
"eval_precision": 0.5125,
|
| 64 |
+
"eval_recall": 0.5774647887323944,
|
| 65 |
+
"eval_runtime": 0.0771,
|
| 66 |
+
"eval_samples_per_second": 103.737,
|
| 67 |
+
"eval_steps_per_second": 12.967,
|
| 68 |
+
"step": 38
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 5.94,
|
| 72 |
+
"eval_accuracy": 0.9038701622971286,
|
| 73 |
+
"eval_f1": 0.632258064516129,
|
| 74 |
+
"eval_loss": 0.46185028553009033,
|
| 75 |
+
"eval_precision": 0.5833333333333334,
|
| 76 |
+
"eval_recall": 0.6901408450704225,
|
| 77 |
+
"eval_runtime": 0.0979,
|
| 78 |
+
"eval_samples_per_second": 81.705,
|
| 79 |
+
"eval_steps_per_second": 10.213,
|
| 80 |
+
"step": 46
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 6.97,
|
| 84 |
+
"eval_accuracy": 0.8976279650436954,
|
| 85 |
+
"eval_f1": 0.6143790849673202,
|
| 86 |
+
"eval_loss": 0.4409206807613373,
|
| 87 |
+
"eval_precision": 0.573170731707317,
|
| 88 |
+
"eval_recall": 0.6619718309859155,
|
| 89 |
+
"eval_runtime": 0.0613,
|
| 90 |
+
"eval_samples_per_second": 130.58,
|
| 91 |
+
"eval_steps_per_second": 16.323,
|
| 92 |
+
"step": 54
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 8.0,
|
| 96 |
+
"eval_accuracy": 0.9038701622971286,
|
| 97 |
+
"eval_f1": 0.6540880503144655,
|
| 98 |
+
"eval_loss": 0.41902562975883484,
|
| 99 |
+
"eval_precision": 0.5909090909090909,
|
| 100 |
+
"eval_recall": 0.7323943661971831,
|
| 101 |
+
"eval_runtime": 0.0632,
|
| 102 |
+
"eval_samples_per_second": 126.495,
|
| 103 |
+
"eval_steps_per_second": 15.812,
|
| 104 |
+
"step": 62
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 8.9,
|
| 108 |
+
"eval_accuracy": 0.9088639200998752,
|
| 109 |
+
"eval_f1": 0.6075949367088608,
|
| 110 |
+
"eval_loss": 0.41402581334114075,
|
| 111 |
+
"eval_precision": 0.5517241379310345,
|
| 112 |
+
"eval_recall": 0.676056338028169,
|
| 113 |
+
"eval_runtime": 0.08,
|
| 114 |
+
"eval_samples_per_second": 99.991,
|
| 115 |
+
"eval_steps_per_second": 12.499,
|
| 116 |
+
"step": 69
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 9.94,
|
| 120 |
+
"eval_accuracy": 0.9101123595505618,
|
| 121 |
+
"eval_f1": 0.6463414634146342,
|
| 122 |
+
"eval_loss": 0.4209658205509186,
|
| 123 |
+
"eval_precision": 0.5698924731182796,
|
| 124 |
+
"eval_recall": 0.7464788732394366,
|
| 125 |
+
"eval_runtime": 0.0753,
|
| 126 |
+
"eval_samples_per_second": 106.269,
|
| 127 |
+
"eval_steps_per_second": 13.284,
|
| 128 |
+
"step": 77
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 10.97,
|
| 132 |
+
"eval_accuracy": 0.9176029962546817,
|
| 133 |
+
"eval_f1": 0.6257668711656442,
|
| 134 |
+
"eval_loss": 0.4074121415615082,
|
| 135 |
+
"eval_precision": 0.5543478260869565,
|
| 136 |
+
"eval_recall": 0.7183098591549296,
|
| 137 |
+
"eval_runtime": 0.0641,
|
| 138 |
+
"eval_samples_per_second": 124.724,
|
| 139 |
+
"eval_steps_per_second": 15.591,
|
| 140 |
+
"step": 85
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 12.0,
|
| 144 |
+
"eval_accuracy": 0.9138576779026217,
|
| 145 |
+
"eval_f1": 0.6296296296296295,
|
| 146 |
+
"eval_loss": 0.4206610321998596,
|
| 147 |
+
"eval_precision": 0.5604395604395604,
|
| 148 |
+
"eval_recall": 0.7183098591549296,
|
| 149 |
+
"eval_runtime": 0.0628,
|
| 150 |
+
"eval_samples_per_second": 127.441,
|
| 151 |
+
"eval_steps_per_second": 15.93,
|
| 152 |
+
"step": 93
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 12.9,
|
| 156 |
+
"eval_accuracy": 0.9138576779026217,
|
| 157 |
+
"eval_f1": 0.6380368098159509,
|
| 158 |
+
"eval_loss": 0.41813310980796814,
|
| 159 |
+
"eval_precision": 0.5652173913043478,
|
| 160 |
+
"eval_recall": 0.7323943661971831,
|
| 161 |
+
"eval_runtime": 0.0843,
|
| 162 |
+
"eval_samples_per_second": 94.887,
|
| 163 |
+
"eval_steps_per_second": 11.861,
|
| 164 |
+
"step": 100
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 13.94,
|
| 168 |
+
"eval_accuracy": 0.9151061173533084,
|
| 169 |
+
"eval_f1": 0.6503067484662577,
|
| 170 |
+
"eval_loss": 0.42074155807495117,
|
| 171 |
+
"eval_precision": 0.5760869565217391,
|
| 172 |
+
"eval_recall": 0.7464788732394366,
|
| 173 |
+
"eval_runtime": 0.0639,
|
| 174 |
+
"eval_samples_per_second": 125.197,
|
| 175 |
+
"eval_steps_per_second": 15.65,
|
| 176 |
+
"step": 108
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 14.97,
|
| 180 |
+
"eval_accuracy": 0.9176029962546817,
|
| 181 |
+
"eval_f1": 0.6625,
|
| 182 |
+
"eval_loss": 0.4188980460166931,
|
| 183 |
+
"eval_precision": 0.5955056179775281,
|
| 184 |
+
"eval_recall": 0.7464788732394366,
|
| 185 |
+
"eval_runtime": 0.0909,
|
| 186 |
+
"eval_samples_per_second": 87.963,
|
| 187 |
+
"eval_steps_per_second": 10.995,
|
| 188 |
+
"step": 116
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 16.0,
|
| 192 |
+
"eval_accuracy": 0.920099875156055,
|
| 193 |
+
"eval_f1": 0.6666666666666666,
|
| 194 |
+
"eval_loss": 0.4166664779186249,
|
| 195 |
+
"eval_precision": 0.6022727272727273,
|
| 196 |
+
"eval_recall": 0.7464788732394366,
|
| 197 |
+
"eval_runtime": 0.0772,
|
| 198 |
+
"eval_samples_per_second": 103.585,
|
| 199 |
+
"eval_steps_per_second": 12.948,
|
| 200 |
+
"step": 124
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 16.9,
|
| 204 |
+
"eval_accuracy": 0.9176029962546817,
|
| 205 |
+
"eval_f1": 0.6583850931677019,
|
| 206 |
+
"eval_loss": 0.4165511429309845,
|
| 207 |
+
"eval_precision": 0.5888888888888889,
|
| 208 |
+
"eval_recall": 0.7464788732394366,
|
| 209 |
+
"eval_runtime": 0.0861,
|
| 210 |
+
"eval_samples_per_second": 92.942,
|
| 211 |
+
"eval_steps_per_second": 11.618,
|
| 212 |
+
"step": 131
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 17.94,
|
| 216 |
+
"eval_accuracy": 0.9176029962546817,
|
| 217 |
+
"eval_f1": 0.6583850931677019,
|
| 218 |
+
"eval_loss": 0.41603589057922363,
|
| 219 |
+
"eval_precision": 0.5888888888888889,
|
| 220 |
+
"eval_recall": 0.7464788732394366,
|
| 221 |
+
"eval_runtime": 0.076,
|
| 222 |
+
"eval_samples_per_second": 105.309,
|
| 223 |
+
"eval_steps_per_second": 13.164,
|
| 224 |
+
"step": 139
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 18.06,
|
| 228 |
+
"eval_accuracy": 0.9176029962546817,
|
| 229 |
+
"eval_f1": 0.6583850931677019,
|
| 230 |
+
"eval_loss": 0.4160560369491577,
|
| 231 |
+
"eval_precision": 0.5888888888888889,
|
| 232 |
+
"eval_recall": 0.7464788732394366,
|
| 233 |
+
"eval_runtime": 0.0925,
|
| 234 |
+
"eval_samples_per_second": 86.509,
|
| 235 |
+
"eval_steps_per_second": 10.814,
|
| 236 |
+
"step": 140
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 18.06,
|
| 240 |
+
"step": 140,
|
| 241 |
+
"total_flos": 142823207204352.0,
|
| 242 |
+
"train_loss": 0.3585296630859375,
|
| 243 |
+
"train_runtime": 632.6175,
|
| 244 |
+
"train_samples_per_second": 3.825,
|
| 245 |
+
"train_steps_per_second": 0.221
|
| 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 |
+
}
|