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": 0.8572538958705256,
|
| 3 |
+
"best_model_checkpoint": "result/my-unsup-simcse-roberta-large/SupCon-48-0.02-1e-5-0.07-0.0",
|
| 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": 0.7782054985417723,
|
| 13 |
+
"eval_sickr_spearman": 0.7378584393091261,
|
| 14 |
+
"eval_stsb_spearman": 0.8185525577744187,
|
| 15 |
+
"step": 125
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.06,
|
| 19 |
+
"eval_avg_sts": 0.7923362848478215,
|
| 20 |
+
"eval_sickr_spearman": 0.7560154924924716,
|
| 21 |
+
"eval_stsb_spearman": 0.8286570772031715,
|
| 22 |
+
"step": 250
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 0.1,
|
| 26 |
+
"eval_avg_sts": 0.7932742301026626,
|
| 27 |
+
"eval_sickr_spearman": 0.7534467891875612,
|
| 28 |
+
"eval_stsb_spearman": 0.833101671017764,
|
| 29 |
+
"step": 375
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 0.13,
|
| 33 |
+
"learning_rate": 8.72024571282314e-06,
|
| 34 |
+
"loss": 0.3301,
|
| 35 |
+
"step": 500
|
| 36 |
+
},
|
| 37 |
+
{
|
| 38 |
+
"epoch": 0.13,
|
| 39 |
+
"eval_avg_sts": 0.794880955659569,
|
| 40 |
+
"eval_sickr_spearman": 0.7557729834613105,
|
| 41 |
+
"eval_stsb_spearman": 0.8339889278578274,
|
| 42 |
+
"step": 500
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.16,
|
| 46 |
+
"eval_avg_sts": 0.7976909480948997,
|
| 47 |
+
"eval_sickr_spearman": 0.7567474864783884,
|
| 48 |
+
"eval_stsb_spearman": 0.8386344097114109,
|
| 49 |
+
"step": 625
|
| 50 |
+
},
|
| 51 |
+
{
|
| 52 |
+
"epoch": 0.19,
|
| 53 |
+
"eval_avg_sts": 0.8008783148847614,
|
| 54 |
+
"eval_sickr_spearman": 0.7603111060184877,
|
| 55 |
+
"eval_stsb_spearman": 0.8414455237510349,
|
| 56 |
+
"step": 750
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.22,
|
| 60 |
+
"eval_avg_sts": 0.7988165156325646,
|
| 61 |
+
"eval_sickr_spearman": 0.757687311040228,
|
| 62 |
+
"eval_stsb_spearman": 0.8399457202249012,
|
| 63 |
+
"step": 875
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.26,
|
| 67 |
+
"learning_rate": 7.440491425646277e-06,
|
| 68 |
+
"loss": 0.0007,
|
| 69 |
+
"step": 1000
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.26,
|
| 73 |
+
"eval_avg_sts": 0.7996021062049683,
|
| 74 |
+
"eval_sickr_spearman": 0.755320914734575,
|
| 75 |
+
"eval_stsb_spearman": 0.8438832976753616,
|
| 76 |
+
"step": 1000
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.29,
|
| 80 |
+
"eval_avg_sts": 0.8008678219858503,
|
| 81 |
+
"eval_sickr_spearman": 0.7550597696360559,
|
| 82 |
+
"eval_stsb_spearman": 0.8466758743356445,
|
| 83 |
+
"step": 1125
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.32,
|
| 87 |
+
"eval_avg_sts": 0.7974316688328598,
|
| 88 |
+
"eval_sickr_spearman": 0.749437296932812,
|
| 89 |
+
"eval_stsb_spearman": 0.8454260407329075,
|
| 90 |
+
"step": 1250
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.35,
|
| 94 |
+
"eval_avg_sts": 0.8022743240819741,
|
| 95 |
+
"eval_sickr_spearman": 0.7575770316313261,
|
| 96 |
+
"eval_stsb_spearman": 0.846971616532622,
|
| 97 |
+
"step": 1375
|
| 98 |
+
},
|
| 99 |
+
{
|
| 100 |
+
"epoch": 0.38,
|
| 101 |
+
"learning_rate": 6.160737138469415e-06,
|
| 102 |
+
"loss": 0.0008,
|
| 103 |
+
"step": 1500
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.38,
|
| 107 |
+
"eval_avg_sts": 0.801759968136895,
|
| 108 |
+
"eval_sickr_spearman": 0.7535204688971674,
|
| 109 |
+
"eval_stsb_spearman": 0.8499994673766226,
|
| 110 |
+
"step": 1500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.42,
|
| 114 |
+
"eval_avg_sts": 0.801312443833835,
|
| 115 |
+
"eval_sickr_spearman": 0.7532925613208435,
|
| 116 |
+
"eval_stsb_spearman": 0.8493323263468264,
|
| 117 |
+
"step": 1625
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.45,
|
| 121 |
+
"eval_avg_sts": 0.8019017992008193,
|
| 122 |
+
"eval_sickr_spearman": 0.7542989569892763,
|
| 123 |
+
"eval_stsb_spearman": 0.8495046414123621,
|
| 124 |
+
"step": 1750
|
| 125 |
+
},
|
| 126 |
+
{
|
| 127 |
+
"epoch": 0.48,
|
| 128 |
+
"eval_avg_sts": 0.8053453778611492,
|
| 129 |
+
"eval_sickr_spearman": 0.759368639746069,
|
| 130 |
+
"eval_stsb_spearman": 0.8513221159762293,
|
| 131 |
+
"step": 1875
|
| 132 |
+
},
|
| 133 |
+
{
|
| 134 |
+
"epoch": 0.51,
|
| 135 |
+
"learning_rate": 4.8809828512925526e-06,
|
| 136 |
+
"loss": 0.0007,
|
| 137 |
+
"step": 2000
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.51,
|
| 141 |
+
"eval_avg_sts": 0.8074214655871451,
|
| 142 |
+
"eval_sickr_spearman": 0.7644952393580661,
|
| 143 |
+
"eval_stsb_spearman": 0.8503476918162242,
|
| 144 |
+
"step": 2000
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.54,
|
| 148 |
+
"eval_avg_sts": 0.8043679171172835,
|
| 149 |
+
"eval_sickr_spearman": 0.760397754125482,
|
| 150 |
+
"eval_stsb_spearman": 0.848338080109085,
|
| 151 |
+
"step": 2125
|
| 152 |
+
},
|
| 153 |
+
{
|
| 154 |
+
"epoch": 0.58,
|
| 155 |
+
"eval_avg_sts": 0.8037538137225763,
|
| 156 |
+
"eval_sickr_spearman": 0.7591252181239807,
|
| 157 |
+
"eval_stsb_spearman": 0.8483824093211718,
|
| 158 |
+
"step": 2250
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.61,
|
| 162 |
+
"eval_avg_sts": 0.8087713317932268,
|
| 163 |
+
"eval_sickr_spearman": 0.7665475602914152,
|
| 164 |
+
"eval_stsb_spearman": 0.8509951032950382,
|
| 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": 0.8110685237844888,
|
| 176 |
+
"eval_sickr_spearman": 0.765394333545886,
|
| 177 |
+
"eval_stsb_spearman": 0.8567427140230917,
|
| 178 |
+
"step": 2500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.67,
|
| 182 |
+
"eval_avg_sts": 0.8091393361687496,
|
| 183 |
+
"eval_sickr_spearman": 0.7646847220532393,
|
| 184 |
+
"eval_stsb_spearman": 0.8535939502842598,
|
| 185 |
+
"step": 2625
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.7,
|
| 189 |
+
"eval_avg_sts": 0.8083355725116668,
|
| 190 |
+
"eval_sickr_spearman": 0.7630815879805395,
|
| 191 |
+
"eval_stsb_spearman": 0.8535895570427939,
|
| 192 |
+
"step": 2750
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.74,
|
| 196 |
+
"eval_avg_sts": 0.8064249843239497,
|
| 197 |
+
"eval_sickr_spearman": 0.7595829064895845,
|
| 198 |
+
"eval_stsb_spearman": 0.8532670621583149,
|
| 199 |
+
"step": 2875
|
| 200 |
+
},
|
| 201 |
+
{
|
| 202 |
+
"epoch": 0.77,
|
| 203 |
+
"learning_rate": 2.321474276938828e-06,
|
| 204 |
+
"loss": 0.0007,
|
| 205 |
+
"step": 3000
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.77,
|
| 209 |
+
"eval_avg_sts": 0.8075703149702256,
|
| 210 |
+
"eval_sickr_spearman": 0.7607818588436827,
|
| 211 |
+
"eval_stsb_spearman": 0.8543587710967685,
|
| 212 |
+
"step": 3000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.8,
|
| 216 |
+
"eval_avg_sts": 0.8088746392140569,
|
| 217 |
+
"eval_sickr_spearman": 0.7618876788920923,
|
| 218 |
+
"eval_stsb_spearman": 0.8558615995360215,
|
| 219 |
+
"step": 3125
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.83,
|
| 223 |
+
"eval_avg_sts": 0.8085515592471004,
|
| 224 |
+
"eval_sickr_spearman": 0.7616940174910939,
|
| 225 |
+
"eval_stsb_spearman": 0.855409101003107,
|
| 226 |
+
"step": 3250
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.86,
|
| 230 |
+
"eval_avg_sts": 0.8087552005841584,
|
| 231 |
+
"eval_sickr_spearman": 0.7611086144267661,
|
| 232 |
+
"eval_stsb_spearman": 0.8564017867415507,
|
| 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": 0.8090700871643673,
|
| 244 |
+
"eval_sickr_spearman": 0.7608862784582091,
|
| 245 |
+
"eval_stsb_spearman": 0.8572538958705256,
|
| 246 |
+
"step": 3500
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.93,
|
| 250 |
+
"eval_avg_sts": 0.8087913053132321,
|
| 251 |
+
"eval_sickr_spearman": 0.7607510228765595,
|
| 252 |
+
"eval_stsb_spearman": 0.8568315877499046,
|
| 253 |
+
"step": 3625
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.96,
|
| 257 |
+
"eval_avg_sts": 0.8089499832999565,
|
| 258 |
+
"eval_sickr_spearman": 0.7610159144009905,
|
| 259 |
+
"eval_stsb_spearman": 0.8568840521989225,
|
| 260 |
+
"step": 3750
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.99,
|
| 264 |
+
"eval_avg_sts": 0.8088062391857902,
|
| 265 |
+
"eval_sickr_spearman": 0.7611648108154486,
|
| 266 |
+
"eval_stsb_spearman": 0.8564476675561318,
|
| 267 |
+
"step": 3875
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 1.0,
|
| 271 |
+
"step": 3907,
|
| 272 |
+
"total_flos": 137241565209427968,
|
| 273 |
+
"train_runtime": 6727.1296,
|
| 274 |
+
"train_samples_per_second": 0.581
|
| 275 |
+
}
|
| 276 |
+
],
|
| 277 |
+
"max_steps": 3907,
|
| 278 |
+
"num_train_epochs": 1,
|
| 279 |
+
"total_flos": 137241565209427968,
|
| 280 |
+
"trial_name": null,
|
| 281 |
+
"trial_params": null
|
| 282 |
+
}
|