Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +282 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,282 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.8571252943582888,
|
| 3 |
+
"best_model_checkpoint": "result/my-unsup-simcse-roberta-large/SupCon-42-0.05-1e-7-0.07-0.00",
|
| 4 |
+
"epoch": 1.0,
|
| 5 |
+
"global_step": 3907,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.03,
|
| 12 |
+
"eval_avg_sts": 0.7853114840394502,
|
| 13 |
+
"eval_sickr_spearman": 0.7560937831878156,
|
| 14 |
+
"eval_stsb_spearman": 0.8145291848910846,
|
| 15 |
+
"step": 125
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.06,
|
| 19 |
+
"eval_avg_sts": 0.7984241433982234,
|
| 20 |
+
"eval_sickr_spearman": 0.766604333053315,
|
| 21 |
+
"eval_stsb_spearman": 0.8302439537431319,
|
| 22 |
+
"step": 250
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.1,
|
| 26 |
+
"eval_avg_sts": 0.7997989872061093,
|
| 27 |
+
"eval_sickr_spearman": 0.7659919845410805,
|
| 28 |
+
"eval_stsb_spearman": 0.833605989871138,
|
| 29 |
+
"step": 375
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.13,
|
| 33 |
+
"learning_rate": 8.72024571282314e-06,
|
| 34 |
+
"loss": 0.2799,
|
| 35 |
+
"step": 500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.13,
|
| 39 |
+
"eval_avg_sts": 0.7994693836795467,
|
| 40 |
+
"eval_sickr_spearman": 0.75997018126048,
|
| 41 |
+
"eval_stsb_spearman": 0.8389685860986134,
|
| 42 |
+
"step": 500
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.16,
|
| 46 |
+
"eval_avg_sts": 0.7985232581506884,
|
| 47 |
+
"eval_sickr_spearman": 0.7565649202618221,
|
| 48 |
+
"eval_stsb_spearman": 0.8404815960395546,
|
| 49 |
+
"step": 625
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.19,
|
| 53 |
+
"eval_avg_sts": 0.8024224801920968,
|
| 54 |
+
"eval_sickr_spearman": 0.7631343261299185,
|
| 55 |
+
"eval_stsb_spearman": 0.8417106342542752,
|
| 56 |
+
"step": 750
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.22,
|
| 60 |
+
"eval_avg_sts": 0.806486354250185,
|
| 61 |
+
"eval_sickr_spearman": 0.7684197645943713,
|
| 62 |
+
"eval_stsb_spearman": 0.8445529439059987,
|
| 63 |
+
"step": 875
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.26,
|
| 67 |
+
"learning_rate": 7.440491425646277e-06,
|
| 68 |
+
"loss": 0.0008,
|
| 69 |
+
"step": 1000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.26,
|
| 73 |
+
"eval_avg_sts": 0.8101191717236411,
|
| 74 |
+
"eval_sickr_spearman": 0.7713727167107866,
|
| 75 |
+
"eval_stsb_spearman": 0.8488656267364958,
|
| 76 |
+
"step": 1000
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.29,
|
| 80 |
+
"eval_avg_sts": 0.8104516414525988,
|
| 81 |
+
"eval_sickr_spearman": 0.7705375038878791,
|
| 82 |
+
"eval_stsb_spearman": 0.8503657790173186,
|
| 83 |
+
"step": 1125
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.32,
|
| 87 |
+
"eval_avg_sts": 0.8107275737129924,
|
| 88 |
+
"eval_sickr_spearman": 0.768677691609094,
|
| 89 |
+
"eval_stsb_spearman": 0.8527774558168908,
|
| 90 |
+
"step": 1250
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.35,
|
| 94 |
+
"eval_avg_sts": 0.8078135414107745,
|
| 95 |
+
"eval_sickr_spearman": 0.7649218035699379,
|
| 96 |
+
"eval_stsb_spearman": 0.8507052792516113,
|
| 97 |
+
"step": 1375
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.38,
|
| 101 |
+
"learning_rate": 6.160737138469415e-06,
|
| 102 |
+
"loss": 0.0007,
|
| 103 |
+
"step": 1500
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.38,
|
| 107 |
+
"eval_avg_sts": 0.8084864773058751,
|
| 108 |
+
"eval_sickr_spearman": 0.7651023044491423,
|
| 109 |
+
"eval_stsb_spearman": 0.8518706501626078,
|
| 110 |
+
"step": 1500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.42,
|
| 114 |
+
"eval_avg_sts": 0.8064083907884463,
|
| 115 |
+
"eval_sickr_spearman": 0.7633883145943233,
|
| 116 |
+
"eval_stsb_spearman": 0.8494284669825691,
|
| 117 |
+
"step": 1625
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.45,
|
| 121 |
+
"eval_avg_sts": 0.8064072242587522,
|
| 122 |
+
"eval_sickr_spearman": 0.762559682032313,
|
| 123 |
+
"eval_stsb_spearman": 0.8502547664851915,
|
| 124 |
+
"step": 1750
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.48,
|
| 128 |
+
"eval_avg_sts": 0.8089447046609223,
|
| 129 |
+
"eval_sickr_spearman": 0.7644766032907081,
|
| 130 |
+
"eval_stsb_spearman": 0.8534128060311365,
|
| 131 |
+
"step": 1875
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.51,
|
| 135 |
+
"learning_rate": 4.8809828512925526e-06,
|
| 136 |
+
"loss": 0.0008,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.51,
|
| 141 |
+
"eval_avg_sts": 0.8115395380996886,
|
| 142 |
+
"eval_sickr_spearman": 0.7682088119968551,
|
| 143 |
+
"eval_stsb_spearman": 0.8548702642025222,
|
| 144 |
+
"step": 2000
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.54,
|
| 148 |
+
"eval_avg_sts": 0.8048101519918642,
|
| 149 |
+
"eval_sickr_spearman": 0.7595115803039489,
|
| 150 |
+
"eval_stsb_spearman": 0.8501087236797796,
|
| 151 |
+
"step": 2125
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.58,
|
| 155 |
+
"eval_avg_sts": 0.8036394579221384,
|
| 156 |
+
"eval_sickr_spearman": 0.7555471412223486,
|
| 157 |
+
"eval_stsb_spearman": 0.8517317746219284,
|
| 158 |
+
"step": 2250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.61,
|
| 162 |
+
"eval_avg_sts": 0.8038357750120058,
|
| 163 |
+
"eval_sickr_spearman": 0.7555232217338324,
|
| 164 |
+
"eval_stsb_spearman": 0.8521483282901793,
|
| 165 |
+
"step": 2375
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.64,
|
| 169 |
+
"learning_rate": 3.60122856411569e-06,
|
| 170 |
+
"loss": 0.0006,
|
| 171 |
+
"step": 2500
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.64,
|
| 175 |
+
"eval_avg_sts": 0.8054629187086522,
|
| 176 |
+
"eval_sickr_spearman": 0.7540965058967146,
|
| 177 |
+
"eval_stsb_spearman": 0.8568293315205897,
|
| 178 |
+
"step": 2500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.67,
|
| 182 |
+
"eval_avg_sts": 0.8072241773989005,
|
| 183 |
+
"eval_sickr_spearman": 0.7604686960623063,
|
| 184 |
+
"eval_stsb_spearman": 0.8539796587354948,
|
| 185 |
+
"step": 2625
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.7,
|
| 189 |
+
"eval_avg_sts": 0.8080836122408098,
|
| 190 |
+
"eval_sickr_spearman": 0.7615256204494518,
|
| 191 |
+
"eval_stsb_spearman": 0.8546416040321678,
|
| 192 |
+
"step": 2750
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.74,
|
| 196 |
+
"eval_avg_sts": 0.8075903071060964,
|
| 197 |
+
"eval_sickr_spearman": 0.7616180803197202,
|
| 198 |
+
"eval_stsb_spearman": 0.8535625338924724,
|
| 199 |
+
"step": 2875
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.77,
|
| 203 |
+
"learning_rate": 2.321474276938828e-06,
|
| 204 |
+
"loss": 0.0006,
|
| 205 |
+
"step": 3000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.77,
|
| 209 |
+
"eval_avg_sts": 0.8082750370832301,
|
| 210 |
+
"eval_sickr_spearman": 0.7632299560528819,
|
| 211 |
+
"eval_stsb_spearman": 0.8533201181135783,
|
| 212 |
+
"step": 3000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.8,
|
| 216 |
+
"eval_avg_sts": 0.8117576966383484,
|
| 217 |
+
"eval_sickr_spearman": 0.7675114004039738,
|
| 218 |
+
"eval_stsb_spearman": 0.856003992872723,
|
| 219 |
+
"step": 3125
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.83,
|
| 223 |
+
"eval_avg_sts": 0.8114894072718958,
|
| 224 |
+
"eval_sickr_spearman": 0.766471257266614,
|
| 225 |
+
"eval_stsb_spearman": 0.8565075572771775,
|
| 226 |
+
"step": 3250
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.86,
|
| 230 |
+
"eval_avg_sts": 0.8119784627217479,
|
| 231 |
+
"eval_sickr_spearman": 0.7669392058925415,
|
| 232 |
+
"eval_stsb_spearman": 0.8570177195509542,
|
| 233 |
+
"step": 3375
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.9,
|
| 237 |
+
"learning_rate": 1.0417199897619657e-06,
|
| 238 |
+
"loss": 0.0006,
|
| 239 |
+
"step": 3500
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.9,
|
| 243 |
+
"eval_avg_sts": 0.8121041618893012,
|
| 244 |
+
"eval_sickr_spearman": 0.7670830294203138,
|
| 245 |
+
"eval_stsb_spearman": 0.8571252943582888,
|
| 246 |
+
"step": 3500
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.93,
|
| 250 |
+
"eval_avg_sts": 0.812099226388967,
|
| 251 |
+
"eval_sickr_spearman": 0.7671061620011403,
|
| 252 |
+
"eval_stsb_spearman": 0.8570922907767938,
|
| 253 |
+
"step": 3625
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.96,
|
| 257 |
+
"eval_avg_sts": 0.8112897053907395,
|
| 258 |
+
"eval_sickr_spearman": 0.7658678721749643,
|
| 259 |
+
"eval_stsb_spearman": 0.8567115386065146,
|
| 260 |
+
"step": 3750
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.99,
|
| 264 |
+
"eval_avg_sts": 0.8109815241029494,
|
| 265 |
+
"eval_sickr_spearman": 0.7654719998369115,
|
| 266 |
+
"eval_stsb_spearman": 0.8564910483689874,
|
| 267 |
+
"step": 3875
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 1.0,
|
| 271 |
+
"step": 3907,
|
| 272 |
+
"total_flos": 137373784221646848,
|
| 273 |
+
"train_runtime": 6612.0652,
|
| 274 |
+
"train_samples_per_second": 0.591
|
| 275 |
+
}
|
| 276 |
+
],
|
| 277 |
+
"max_steps": 3907,
|
| 278 |
+
"num_train_epochs": 1,
|
| 279 |
+
"total_flos": 137373784221646848,
|
| 280 |
+
"trial_name": null,
|
| 281 |
+
"trial_params": null
|
| 282 |
+
}
|