Upload trainer_state.json
Browse files- trainer_state.json +289 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,289 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 2500,
|
| 3 |
+
"best_metric": 0.07224167883396149,
|
| 4 |
+
"best_model_checkpoint": "C:\\Users\\Daniel\\Desktop\\GitHub\\NLPinitiative\\models\\dlsmallw\\NLPinitiative-Multilabel-Regression\\checkpoint-2500",
|
| 5 |
+
"epoch": 2.7203482045701852,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.544069640914037,
|
| 14 |
+
"grad_norm": 0.5715875625610352,
|
| 15 |
+
"learning_rate": 1.637286906057309e-05,
|
| 16 |
+
"loss": 0.0289,
|
| 17 |
+
"step": 500
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.544069640914037,
|
| 21 |
+
"eval_loss": 0.011422036215662956,
|
| 22 |
+
"eval_mae_per_cat": [
|
| 23 |
+
0.0535765178501606,
|
| 24 |
+
0.059359531849622726,
|
| 25 |
+
0.05057797580957413,
|
| 26 |
+
0.031064841896295547,
|
| 27 |
+
0.06608528643846512,
|
| 28 |
+
0.015492343343794346
|
| 29 |
+
],
|
| 30 |
+
"eval_mean_mae": 0.0460260808467865,
|
| 31 |
+
"eval_mean_pearson": 0.9106294274330139,
|
| 32 |
+
"eval_mean_r2": 0.6782520413398743,
|
| 33 |
+
"eval_mean_rmse": 0.09810739755630493,
|
| 34 |
+
"eval_pearson_per_cat": [
|
| 35 |
+
0.9014391899108887,
|
| 36 |
+
0.923698365688324,
|
| 37 |
+
0.8990985155105591,
|
| 38 |
+
0.9024374485015869,
|
| 39 |
+
0.9264736175537109,
|
| 40 |
+
NaN
|
| 41 |
+
],
|
| 42 |
+
"eval_r2_per_cat": [
|
| 43 |
+
0.7889423370361328,
|
| 44 |
+
0.8516361713409424,
|
| 45 |
+
0.7870068550109863,
|
| 46 |
+
0.8109977841377258,
|
| 47 |
+
0.8309295177459717,
|
| 48 |
+
0.0
|
| 49 |
+
],
|
| 50 |
+
"eval_rmse_per_cat": [
|
| 51 |
+
0.11551735550165176,
|
| 52 |
+
0.1352391391992569,
|
| 53 |
+
0.10884607583284378,
|
| 54 |
+
0.0673733726143837,
|
| 55 |
+
0.1418396383523941,
|
| 56 |
+
0.019828785210847855
|
| 57 |
+
],
|
| 58 |
+
"eval_runtime": 100.9453,
|
| 59 |
+
"eval_samples_per_second": 31.225,
|
| 60 |
+
"eval_steps_per_second": 3.903,
|
| 61 |
+
"step": 500
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 1.088139281828074,
|
| 65 |
+
"grad_norm": 0.3862350583076477,
|
| 66 |
+
"learning_rate": 1.2745738121146174e-05,
|
| 67 |
+
"loss": 0.0108,
|
| 68 |
+
"step": 1000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 1.088139281828074,
|
| 72 |
+
"eval_loss": 0.008166518062353134,
|
| 73 |
+
"eval_mae_per_cat": [
|
| 74 |
+
0.0379190668463707,
|
| 75 |
+
0.0694158673286438,
|
| 76 |
+
0.032430507242679596,
|
| 77 |
+
0.025210702791810036,
|
| 78 |
+
0.03987590968608856,
|
| 79 |
+
0.0133794154971838
|
| 80 |
+
],
|
| 81 |
+
"eval_mean_mae": 0.03637191280722618,
|
| 82 |
+
"eval_mean_pearson": 0.9361262321472168,
|
| 83 |
+
"eval_mean_r2": 0.7225764393806458,
|
| 84 |
+
"eval_mean_rmse": 0.0826454609632492,
|
| 85 |
+
"eval_pearson_per_cat": [
|
| 86 |
+
0.9333711266517639,
|
| 87 |
+
0.9291742444038391,
|
| 88 |
+
0.9329087138175964,
|
| 89 |
+
0.9267426133155823,
|
| 90 |
+
0.9584344625473022,
|
| 91 |
+
NaN
|
| 92 |
+
],
|
| 93 |
+
"eval_r2_per_cat": [
|
| 94 |
+
0.8588278889656067,
|
| 95 |
+
0.8505284190177917,
|
| 96 |
+
0.8572311997413635,
|
| 97 |
+
0.8514397144317627,
|
| 98 |
+
0.9174313545227051,
|
| 99 |
+
0.0
|
| 100 |
+
],
|
| 101 |
+
"eval_rmse_per_cat": [
|
| 102 |
+
0.09447590261697769,
|
| 103 |
+
0.13574306666851044,
|
| 104 |
+
0.08911412954330444,
|
| 105 |
+
0.05973187834024429,
|
| 106 |
+
0.09912225604057312,
|
| 107 |
+
0.01768551953136921
|
| 108 |
+
],
|
| 109 |
+
"eval_runtime": 99.9899,
|
| 110 |
+
"eval_samples_per_second": 31.523,
|
| 111 |
+
"eval_steps_per_second": 3.94,
|
| 112 |
+
"step": 1000
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 1.632208922742111,
|
| 116 |
+
"grad_norm": 0.17565102875232697,
|
| 117 |
+
"learning_rate": 9.11860718171926e-06,
|
| 118 |
+
"loss": 0.0059,
|
| 119 |
+
"step": 1500
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 1.632208922742111,
|
| 123 |
+
"eval_loss": 0.007027682848274708,
|
| 124 |
+
"eval_mae_per_cat": [
|
| 125 |
+
0.02809344045817852,
|
| 126 |
+
0.039010003209114075,
|
| 127 |
+
0.03197040781378746,
|
| 128 |
+
0.01924479752779007,
|
| 129 |
+
0.033096104860305786,
|
| 130 |
+
0.00918651558458805
|
| 131 |
+
],
|
| 132 |
+
"eval_mean_mae": 0.026766879484057426,
|
| 133 |
+
"eval_mean_pearson": 0.9401901483535766,
|
| 134 |
+
"eval_mean_r2": 0.7350945472717285,
|
| 135 |
+
"eval_mean_rmse": 0.07674583047628403,
|
| 136 |
+
"eval_pearson_per_cat": [
|
| 137 |
+
0.9324471950531006,
|
| 138 |
+
0.9467915296554565,
|
| 139 |
+
0.9339689612388611,
|
| 140 |
+
0.9317143559455872,
|
| 141 |
+
0.9560286998748779,
|
| 142 |
+
NaN
|
| 143 |
+
],
|
| 144 |
+
"eval_r2_per_cat": [
|
| 145 |
+
0.8691833019256592,
|
| 146 |
+
0.8961314558982849,
|
| 147 |
+
0.8661218285560608,
|
| 148 |
+
0.8654465675354004,
|
| 149 |
+
0.9136843681335449,
|
| 150 |
+
0.0
|
| 151 |
+
],
|
| 152 |
+
"eval_rmse_per_cat": [
|
| 153 |
+
0.09094487875699997,
|
| 154 |
+
0.11315672099590302,
|
| 155 |
+
0.08629484474658966,
|
| 156 |
+
0.05684630945324898,
|
| 157 |
+
0.10134641826152802,
|
| 158 |
+
0.01188575103878975
|
| 159 |
+
],
|
| 160 |
+
"eval_runtime": 102.2428,
|
| 161 |
+
"eval_samples_per_second": 30.829,
|
| 162 |
+
"eval_steps_per_second": 3.854,
|
| 163 |
+
"step": 1500
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 2.176278563656148,
|
| 167 |
+
"grad_norm": 0.19192808866500854,
|
| 168 |
+
"learning_rate": 5.491476242292347e-06,
|
| 169 |
+
"loss": 0.005,
|
| 170 |
+
"step": 2000
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.176278563656148,
|
| 174 |
+
"eval_loss": 0.006472242530435324,
|
| 175 |
+
"eval_mae_per_cat": [
|
| 176 |
+
0.02706627920269966,
|
| 177 |
+
0.04395332559943199,
|
| 178 |
+
0.024061929434537888,
|
| 179 |
+
0.016889315098524094,
|
| 180 |
+
0.033904846757650375,
|
| 181 |
+
0.007554503157734871
|
| 182 |
+
],
|
| 183 |
+
"eval_mean_mae": 0.025571702048182487,
|
| 184 |
+
"eval_mean_pearson": 0.9477012753486633,
|
| 185 |
+
"eval_mean_r2": 0.7445929050445557,
|
| 186 |
+
"eval_mean_rmse": 0.07305774837732315,
|
| 187 |
+
"eval_pearson_per_cat": [
|
| 188 |
+
0.9431683421134949,
|
| 189 |
+
0.9483680725097656,
|
| 190 |
+
0.9359709024429321,
|
| 191 |
+
0.9474267959594727,
|
| 192 |
+
0.9635722637176514,
|
| 193 |
+
NaN
|
| 194 |
+
],
|
| 195 |
+
"eval_r2_per_cat": [
|
| 196 |
+
0.8886928558349609,
|
| 197 |
+
0.8949806690216064,
|
| 198 |
+
0.8702350854873657,
|
| 199 |
+
0.8879827260971069,
|
| 200 |
+
0.9256656765937805,
|
| 201 |
+
0.0
|
| 202 |
+
],
|
| 203 |
+
"eval_rmse_per_cat": [
|
| 204 |
+
0.08388959616422653,
|
| 205 |
+
0.11378183960914612,
|
| 206 |
+
0.08495883643627167,
|
| 207 |
+
0.051867734640836716,
|
| 208 |
+
0.09404993057250977,
|
| 209 |
+
0.009798564948141575
|
| 210 |
+
],
|
| 211 |
+
"eval_runtime": 103.1288,
|
| 212 |
+
"eval_samples_per_second": 30.564,
|
| 213 |
+
"eval_steps_per_second": 3.82,
|
| 214 |
+
"step": 2000
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 2.7203482045701852,
|
| 218 |
+
"grad_norm": 0.12346106767654419,
|
| 219 |
+
"learning_rate": 1.8643453028654336e-06,
|
| 220 |
+
"loss": 0.0034,
|
| 221 |
+
"step": 2500
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 2.7203482045701852,
|
| 225 |
+
"eval_loss": 0.006370748393237591,
|
| 226 |
+
"eval_mae_per_cat": [
|
| 227 |
+
0.025086652487516403,
|
| 228 |
+
0.03752453625202179,
|
| 229 |
+
0.02304081991314888,
|
| 230 |
+
0.013712005689740181,
|
| 231 |
+
0.030296988785266876,
|
| 232 |
+
0.006659933365881443
|
| 233 |
+
],
|
| 234 |
+
"eval_mean_mae": 0.022720156237483025,
|
| 235 |
+
"eval_mean_pearson": 0.9476440191268921,
|
| 236 |
+
"eval_mean_r2": 0.7471505999565125,
|
| 237 |
+
"eval_mean_rmse": 0.07224167883396149,
|
| 238 |
+
"eval_pearson_per_cat": [
|
| 239 |
+
0.9419375658035278,
|
| 240 |
+
0.9502590894699097,
|
| 241 |
+
0.9368577599525452,
|
| 242 |
+
0.948063850402832,
|
| 243 |
+
0.9611018300056458,
|
| 244 |
+
NaN
|
| 245 |
+
],
|
| 246 |
+
"eval_r2_per_cat": [
|
| 247 |
+
0.8866956233978271,
|
| 248 |
+
0.8990016579627991,
|
| 249 |
+
0.8755217790603638,
|
| 250 |
+
0.8987970352172852,
|
| 251 |
+
0.9228875637054443,
|
| 252 |
+
0.0
|
| 253 |
+
],
|
| 254 |
+
"eval_rmse_per_cat": [
|
| 255 |
+
0.08463889360427856,
|
| 256 |
+
0.11158231645822525,
|
| 257 |
+
0.08321022242307663,
|
| 258 |
+
0.04930051788687706,
|
| 259 |
+
0.09579125046730042,
|
| 260 |
+
0.008926868438720703
|
| 261 |
+
],
|
| 262 |
+
"eval_runtime": 104.2098,
|
| 263 |
+
"eval_samples_per_second": 30.247,
|
| 264 |
+
"eval_steps_per_second": 3.781,
|
| 265 |
+
"step": 2500
|
| 266 |
+
}
|
| 267 |
+
],
|
| 268 |
+
"logging_steps": 500,
|
| 269 |
+
"max_steps": 2757,
|
| 270 |
+
"num_input_tokens_seen": 0,
|
| 271 |
+
"num_train_epochs": 3,
|
| 272 |
+
"save_steps": 500,
|
| 273 |
+
"stateful_callbacks": {
|
| 274 |
+
"TrainerControl": {
|
| 275 |
+
"args": {
|
| 276 |
+
"should_epoch_stop": false,
|
| 277 |
+
"should_evaluate": false,
|
| 278 |
+
"should_log": false,
|
| 279 |
+
"should_save": true,
|
| 280 |
+
"should_training_stop": false
|
| 281 |
+
},
|
| 282 |
+
"attributes": {}
|
| 283 |
+
}
|
| 284 |
+
},
|
| 285 |
+
"total_flos": 1315602524160000.0,
|
| 286 |
+
"train_batch_size": 8,
|
| 287 |
+
"trial_name": null,
|
| 288 |
+
"trial_params": null
|
| 289 |
+
}
|