Upload 2 files
Browse files- trainer_state.json +418 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,418 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 2.390516519546509,
|
| 3 |
+
"best_model_checkpoint": "MIReAD_3e-05/checkpoint-13806",
|
| 4 |
+
"epoch": 6.0,
|
| 5 |
+
"global_step": 27612,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.11,
|
| 12 |
+
"learning_rate": 2.94567579313342e-05,
|
| 13 |
+
"loss": 4.5385,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.22,
|
| 18 |
+
"learning_rate": 2.8913515862668408e-05,
|
| 19 |
+
"loss": 3.745,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.33,
|
| 24 |
+
"learning_rate": 2.8370273794002608e-05,
|
| 25 |
+
"loss": 3.3652,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.43,
|
| 30 |
+
"learning_rate": 2.782703172533681e-05,
|
| 31 |
+
"loss": 3.1532,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.54,
|
| 36 |
+
"learning_rate": 2.728378965667101e-05,
|
| 37 |
+
"loss": 2.9768,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.65,
|
| 42 |
+
"learning_rate": 2.6740547588005218e-05,
|
| 43 |
+
"loss": 2.8561,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.76,
|
| 48 |
+
"learning_rate": 2.6197305519339418e-05,
|
| 49 |
+
"loss": 2.7393,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.87,
|
| 54 |
+
"learning_rate": 2.565406345067362e-05,
|
| 55 |
+
"loss": 2.6236,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.98,
|
| 60 |
+
"learning_rate": 2.511082138200782e-05,
|
| 61 |
+
"loss": 2.6074,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 1.0,
|
| 66 |
+
"eval_accuracy": 0.3415979708306912,
|
| 67 |
+
"eval_f1": 0.28057933815352265,
|
| 68 |
+
"eval_loss": 2.6715340614318848,
|
| 69 |
+
"eval_precision": 0.3280031388570244,
|
| 70 |
+
"eval_recall": 0.30555570676656096,
|
| 71 |
+
"eval_runtime": 128.8869,
|
| 72 |
+
"eval_samples_per_second": 122.355,
|
| 73 |
+
"eval_steps_per_second": 7.65,
|
| 74 |
+
"step": 4602
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 1.09,
|
| 78 |
+
"learning_rate": 2.456757931334203e-05,
|
| 79 |
+
"loss": 2.4343,
|
| 80 |
+
"step": 5000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 1.2,
|
| 84 |
+
"learning_rate": 2.402433724467623e-05,
|
| 85 |
+
"loss": 2.2922,
|
| 86 |
+
"step": 5500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 1.3,
|
| 90 |
+
"learning_rate": 2.3481095176010432e-05,
|
| 91 |
+
"loss": 2.2894,
|
| 92 |
+
"step": 6000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 1.41,
|
| 96 |
+
"learning_rate": 2.2937853107344632e-05,
|
| 97 |
+
"loss": 2.2635,
|
| 98 |
+
"step": 6500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 1.52,
|
| 102 |
+
"learning_rate": 2.239461103867884e-05,
|
| 103 |
+
"loss": 2.238,
|
| 104 |
+
"step": 7000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 1.63,
|
| 108 |
+
"learning_rate": 2.185136897001304e-05,
|
| 109 |
+
"loss": 2.2425,
|
| 110 |
+
"step": 7500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 1.74,
|
| 114 |
+
"learning_rate": 2.1308126901347242e-05,
|
| 115 |
+
"loss": 2.1702,
|
| 116 |
+
"step": 8000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 1.85,
|
| 120 |
+
"learning_rate": 2.0764884832681442e-05,
|
| 121 |
+
"loss": 2.1909,
|
| 122 |
+
"step": 8500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 1.96,
|
| 126 |
+
"learning_rate": 2.0221642764015646e-05,
|
| 127 |
+
"loss": 2.1209,
|
| 128 |
+
"step": 9000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 2.0,
|
| 132 |
+
"eval_accuracy": 0.3929613189600507,
|
| 133 |
+
"eval_f1": 0.3569292122063319,
|
| 134 |
+
"eval_loss": 2.4266042709350586,
|
| 135 |
+
"eval_precision": 0.37263788729318065,
|
| 136 |
+
"eval_recall": 0.3854876449553655,
|
| 137 |
+
"eval_runtime": 127.8545,
|
| 138 |
+
"eval_samples_per_second": 123.343,
|
| 139 |
+
"eval_steps_per_second": 7.712,
|
| 140 |
+
"step": 9204
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 2.06,
|
| 144 |
+
"learning_rate": 1.967840069534985e-05,
|
| 145 |
+
"loss": 1.9419,
|
| 146 |
+
"step": 9500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 2.17,
|
| 150 |
+
"learning_rate": 1.913515862668405e-05,
|
| 151 |
+
"loss": 1.7981,
|
| 152 |
+
"step": 10000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 2.28,
|
| 156 |
+
"learning_rate": 1.8591916558018253e-05,
|
| 157 |
+
"loss": 1.82,
|
| 158 |
+
"step": 10500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 2.39,
|
| 162 |
+
"learning_rate": 1.8048674489352456e-05,
|
| 163 |
+
"loss": 1.7695,
|
| 164 |
+
"step": 11000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 2.5,
|
| 168 |
+
"learning_rate": 1.750543242068666e-05,
|
| 169 |
+
"loss": 1.7817,
|
| 170 |
+
"step": 11500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 2.61,
|
| 174 |
+
"learning_rate": 1.696219035202086e-05,
|
| 175 |
+
"loss": 1.7977,
|
| 176 |
+
"step": 12000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 2.72,
|
| 180 |
+
"learning_rate": 1.6418948283355063e-05,
|
| 181 |
+
"loss": 1.7337,
|
| 182 |
+
"step": 12500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 2.82,
|
| 186 |
+
"learning_rate": 1.5875706214689266e-05,
|
| 187 |
+
"loss": 1.7563,
|
| 188 |
+
"step": 13000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 2.93,
|
| 192 |
+
"learning_rate": 1.533246414602347e-05,
|
| 193 |
+
"loss": 1.7413,
|
| 194 |
+
"step": 13500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 3.0,
|
| 198 |
+
"eval_accuracy": 0.40786303107165506,
|
| 199 |
+
"eval_f1": 0.3830566278260409,
|
| 200 |
+
"eval_loss": 2.390516519546509,
|
| 201 |
+
"eval_precision": 0.40575163850397916,
|
| 202 |
+
"eval_recall": 0.39817650401283294,
|
| 203 |
+
"eval_runtime": 129.2884,
|
| 204 |
+
"eval_samples_per_second": 121.975,
|
| 205 |
+
"eval_steps_per_second": 7.626,
|
| 206 |
+
"step": 13806
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 3.04,
|
| 210 |
+
"learning_rate": 1.478922207735767e-05,
|
| 211 |
+
"loss": 1.6035,
|
| 212 |
+
"step": 14000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 3.15,
|
| 216 |
+
"learning_rate": 1.4245980008691873e-05,
|
| 217 |
+
"loss": 1.4123,
|
| 218 |
+
"step": 14500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 3.26,
|
| 222 |
+
"learning_rate": 1.3702737940026075e-05,
|
| 223 |
+
"loss": 1.3897,
|
| 224 |
+
"step": 15000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 3.37,
|
| 228 |
+
"learning_rate": 1.3159495871360279e-05,
|
| 229 |
+
"loss": 1.3592,
|
| 230 |
+
"step": 15500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 3.48,
|
| 234 |
+
"learning_rate": 1.261625380269448e-05,
|
| 235 |
+
"loss": 1.3673,
|
| 236 |
+
"step": 16000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 3.59,
|
| 240 |
+
"learning_rate": 1.2073011734028684e-05,
|
| 241 |
+
"loss": 1.3792,
|
| 242 |
+
"step": 16500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 3.69,
|
| 246 |
+
"learning_rate": 1.1529769665362886e-05,
|
| 247 |
+
"loss": 1.3385,
|
| 248 |
+
"step": 17000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 3.8,
|
| 252 |
+
"learning_rate": 1.0986527596697089e-05,
|
| 253 |
+
"loss": 1.3258,
|
| 254 |
+
"step": 17500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 3.91,
|
| 258 |
+
"learning_rate": 1.044328552803129e-05,
|
| 259 |
+
"loss": 1.3415,
|
| 260 |
+
"step": 18000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 4.0,
|
| 264 |
+
"eval_accuracy": 0.4099556119213697,
|
| 265 |
+
"eval_f1": 0.3978990080253614,
|
| 266 |
+
"eval_loss": 2.46903133392334,
|
| 267 |
+
"eval_precision": 0.4214347585145809,
|
| 268 |
+
"eval_recall": 0.40462921661844276,
|
| 269 |
+
"eval_runtime": 129.7625,
|
| 270 |
+
"eval_samples_per_second": 121.53,
|
| 271 |
+
"eval_steps_per_second": 7.598,
|
| 272 |
+
"step": 18408
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 4.02,
|
| 276 |
+
"learning_rate": 9.900043459365494e-06,
|
| 277 |
+
"loss": 1.2695,
|
| 278 |
+
"step": 18500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 4.13,
|
| 282 |
+
"learning_rate": 9.356801390699696e-06,
|
| 283 |
+
"loss": 1.0401,
|
| 284 |
+
"step": 19000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 4.24,
|
| 288 |
+
"learning_rate": 8.8135593220339e-06,
|
| 289 |
+
"loss": 1.0362,
|
| 290 |
+
"step": 19500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 4.35,
|
| 294 |
+
"learning_rate": 8.270317253368101e-06,
|
| 295 |
+
"loss": 1.0133,
|
| 296 |
+
"step": 20000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 4.45,
|
| 300 |
+
"learning_rate": 7.727075184702305e-06,
|
| 301 |
+
"loss": 1.0128,
|
| 302 |
+
"step": 20500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 4.56,
|
| 306 |
+
"learning_rate": 7.183833116036506e-06,
|
| 307 |
+
"loss": 0.9873,
|
| 308 |
+
"step": 21000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 4.67,
|
| 312 |
+
"learning_rate": 6.640591047370709e-06,
|
| 313 |
+
"loss": 0.9855,
|
| 314 |
+
"step": 21500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 4.78,
|
| 318 |
+
"learning_rate": 6.0973489787049115e-06,
|
| 319 |
+
"loss": 1.0153,
|
| 320 |
+
"step": 22000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 4.89,
|
| 324 |
+
"learning_rate": 5.554106910039114e-06,
|
| 325 |
+
"loss": 0.9783,
|
| 326 |
+
"step": 22500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 5.0,
|
| 330 |
+
"learning_rate": 5.010864841373317e-06,
|
| 331 |
+
"loss": 0.9837,
|
| 332 |
+
"step": 23000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 5.0,
|
| 336 |
+
"eval_accuracy": 0.4050095117311351,
|
| 337 |
+
"eval_f1": 0.3983089847187763,
|
| 338 |
+
"eval_loss": 2.622011184692383,
|
| 339 |
+
"eval_precision": 0.411294379610805,
|
| 340 |
+
"eval_recall": 0.40211336103710194,
|
| 341 |
+
"eval_runtime": 129.5856,
|
| 342 |
+
"eval_samples_per_second": 121.696,
|
| 343 |
+
"eval_steps_per_second": 7.609,
|
| 344 |
+
"step": 23010
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 5.11,
|
| 348 |
+
"learning_rate": 4.467622772707519e-06,
|
| 349 |
+
"loss": 0.7989,
|
| 350 |
+
"step": 23500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 5.22,
|
| 354 |
+
"learning_rate": 3.924380704041722e-06,
|
| 355 |
+
"loss": 0.776,
|
| 356 |
+
"step": 24000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 5.32,
|
| 360 |
+
"learning_rate": 3.3811386353759236e-06,
|
| 361 |
+
"loss": 0.7795,
|
| 362 |
+
"step": 24500
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 5.43,
|
| 366 |
+
"learning_rate": 2.837896566710126e-06,
|
| 367 |
+
"loss": 0.7417,
|
| 368 |
+
"step": 25000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 5.54,
|
| 372 |
+
"learning_rate": 2.2946544980443283e-06,
|
| 373 |
+
"loss": 0.7566,
|
| 374 |
+
"step": 25500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 5.65,
|
| 378 |
+
"learning_rate": 1.7514124293785311e-06,
|
| 379 |
+
"loss": 0.7571,
|
| 380 |
+
"step": 26000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 5.76,
|
| 384 |
+
"learning_rate": 1.2081703607127337e-06,
|
| 385 |
+
"loss": 0.7666,
|
| 386 |
+
"step": 26500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 5.87,
|
| 390 |
+
"learning_rate": 6.649282920469362e-07,
|
| 391 |
+
"loss": 0.7696,
|
| 392 |
+
"step": 27000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 5.98,
|
| 396 |
+
"learning_rate": 1.2168622338113866e-07,
|
| 397 |
+
"loss": 0.7456,
|
| 398 |
+
"step": 27500
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 6.0,
|
| 402 |
+
"eval_accuracy": 0.40107799619530754,
|
| 403 |
+
"eval_f1": 0.39625451475720025,
|
| 404 |
+
"eval_loss": 2.7117061614990234,
|
| 405 |
+
"eval_precision": 0.4066294033650095,
|
| 406 |
+
"eval_recall": 0.39986711618175386,
|
| 407 |
+
"eval_runtime": 130.357,
|
| 408 |
+
"eval_samples_per_second": 120.975,
|
| 409 |
+
"eval_steps_per_second": 7.564,
|
| 410 |
+
"step": 27612
|
| 411 |
+
}
|
| 412 |
+
],
|
| 413 |
+
"max_steps": 27612,
|
| 414 |
+
"num_train_epochs": 6,
|
| 415 |
+
"total_flos": 1.1642961087465062e+17,
|
| 416 |
+
"trial_name": null,
|
| 417 |
+
"trial_params": null
|
| 418 |
+
}
|