Upload trainer_state.json
Browse files- trainer_state.json +508 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,508 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 88.58308022828253,
|
| 3 |
+
"best_model_checkpoint": "output/bert/checkpoint-20500",
|
| 4 |
+
"epoch": 1.8525212362190493,
|
| 5 |
+
"global_step": 20500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.05,
|
| 12 |
+
"learning_rate": 2.9322248328212544e-05,
|
| 13 |
+
"loss": 2.6168,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.05,
|
| 18 |
+
"eval_exact_match": 60.0,
|
| 19 |
+
"eval_f1": 71.21519064840525,
|
| 20 |
+
"step": 500
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.09,
|
| 24 |
+
"learning_rate": 2.8644496656425085e-05,
|
| 25 |
+
"loss": 1.6041,
|
| 26 |
+
"step": 1000
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.09,
|
| 30 |
+
"eval_exact_match": 67.11447492904446,
|
| 31 |
+
"eval_f1": 77.66743510634613,
|
| 32 |
+
"step": 1000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.14,
|
| 36 |
+
"learning_rate": 2.7966744984637632e-05,
|
| 37 |
+
"loss": 1.4496,
|
| 38 |
+
"step": 1500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.14,
|
| 42 |
+
"eval_exact_match": 69.66887417218543,
|
| 43 |
+
"eval_f1": 79.43837393741833,
|
| 44 |
+
"step": 1500
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.18,
|
| 48 |
+
"learning_rate": 2.7288993312850172e-05,
|
| 49 |
+
"loss": 1.3674,
|
| 50 |
+
"step": 2000
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.18,
|
| 54 |
+
"eval_exact_match": 72.5922421948912,
|
| 55 |
+
"eval_f1": 81.81276872830003,
|
| 56 |
+
"step": 2000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.23,
|
| 60 |
+
"learning_rate": 2.6611241641062716e-05,
|
| 61 |
+
"loss": 1.3504,
|
| 62 |
+
"step": 2500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.23,
|
| 66 |
+
"eval_exact_match": 73.52885525070955,
|
| 67 |
+
"eval_f1": 82.34072045279031,
|
| 68 |
+
"step": 2500
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.27,
|
| 72 |
+
"learning_rate": 2.5933489969275256e-05,
|
| 73 |
+
"loss": 1.3002,
|
| 74 |
+
"step": 3000
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.27,
|
| 78 |
+
"eval_exact_match": 74.13434247871334,
|
| 79 |
+
"eval_f1": 83.19942219951206,
|
| 80 |
+
"step": 3000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.32,
|
| 84 |
+
"learning_rate": 2.5255738297487804e-05,
|
| 85 |
+
"loss": 1.2499,
|
| 86 |
+
"step": 3500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.32,
|
| 90 |
+
"eval_exact_match": 74.12488174077578,
|
| 91 |
+
"eval_f1": 83.16211440023876,
|
| 92 |
+
"step": 3500
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.36,
|
| 96 |
+
"learning_rate": 2.4577986625700344e-05,
|
| 97 |
+
"loss": 1.1953,
|
| 98 |
+
"step": 4000
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.36,
|
| 102 |
+
"eval_exact_match": 75.37369914853359,
|
| 103 |
+
"eval_f1": 84.23388802584633,
|
| 104 |
+
"step": 4000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.41,
|
| 108 |
+
"learning_rate": 2.3900234953912888e-05,
|
| 109 |
+
"loss": 1.1952,
|
| 110 |
+
"step": 4500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.41,
|
| 114 |
+
"eval_exact_match": 76.14001892147587,
|
| 115 |
+
"eval_f1": 84.76099418267141,
|
| 116 |
+
"step": 4500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.45,
|
| 120 |
+
"learning_rate": 2.3222483282125428e-05,
|
| 121 |
+
"loss": 1.2084,
|
| 122 |
+
"step": 5000
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.45,
|
| 126 |
+
"eval_exact_match": 77.11447492904446,
|
| 127 |
+
"eval_f1": 85.29225320990346,
|
| 128 |
+
"step": 5000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.5,
|
| 132 |
+
"learning_rate": 2.2544731610337975e-05,
|
| 133 |
+
"loss": 1.1546,
|
| 134 |
+
"step": 5500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.5,
|
| 138 |
+
"eval_exact_match": 76.65089877010406,
|
| 139 |
+
"eval_f1": 85.47014269865286,
|
| 140 |
+
"step": 5500
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.54,
|
| 144 |
+
"learning_rate": 2.1866979938550515e-05,
|
| 145 |
+
"loss": 1.1018,
|
| 146 |
+
"step": 6000
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.54,
|
| 150 |
+
"eval_exact_match": 76.9441816461684,
|
| 151 |
+
"eval_f1": 85.33429382849097,
|
| 152 |
+
"step": 6000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.59,
|
| 156 |
+
"learning_rate": 2.118922826676306e-05,
|
| 157 |
+
"loss": 1.0937,
|
| 158 |
+
"step": 6500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.59,
|
| 162 |
+
"eval_exact_match": 77.07663197729423,
|
| 163 |
+
"eval_f1": 85.52813865025963,
|
| 164 |
+
"step": 6500
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.63,
|
| 168 |
+
"learning_rate": 2.05114765949756e-05,
|
| 169 |
+
"loss": 1.0422,
|
| 170 |
+
"step": 7000
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.63,
|
| 174 |
+
"eval_exact_match": 78.0794701986755,
|
| 175 |
+
"eval_f1": 86.03681982738262,
|
| 176 |
+
"step": 7000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.68,
|
| 180 |
+
"learning_rate": 1.9833724923188147e-05,
|
| 181 |
+
"loss": 1.116,
|
| 182 |
+
"step": 7500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.68,
|
| 186 |
+
"eval_exact_match": 77.95648060548723,
|
| 187 |
+
"eval_f1": 86.26399611648696,
|
| 188 |
+
"step": 7500
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.72,
|
| 192 |
+
"learning_rate": 1.9155973251400687e-05,
|
| 193 |
+
"loss": 1.1176,
|
| 194 |
+
"step": 8000
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.72,
|
| 198 |
+
"eval_exact_match": 78.59035004730369,
|
| 199 |
+
"eval_f1": 86.54207970028193,
|
| 200 |
+
"step": 8000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.77,
|
| 204 |
+
"learning_rate": 1.847822157961323e-05,
|
| 205 |
+
"loss": 1.1029,
|
| 206 |
+
"step": 8500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.77,
|
| 210 |
+
"eval_exact_match": 78.66603595080416,
|
| 211 |
+
"eval_f1": 86.56865525427538,
|
| 212 |
+
"step": 8500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.81,
|
| 216 |
+
"learning_rate": 1.780046990782577e-05,
|
| 217 |
+
"loss": 1.0594,
|
| 218 |
+
"step": 9000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.81,
|
| 222 |
+
"eval_exact_match": 78.9120151371807,
|
| 223 |
+
"eval_f1": 86.80153948101524,
|
| 224 |
+
"step": 9000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.86,
|
| 228 |
+
"learning_rate": 1.7122718236038318e-05,
|
| 229 |
+
"loss": 1.0266,
|
| 230 |
+
"step": 9500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.86,
|
| 234 |
+
"eval_exact_match": 78.90255439924314,
|
| 235 |
+
"eval_f1": 86.69867179951433,
|
| 236 |
+
"step": 9500
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.9,
|
| 240 |
+
"learning_rate": 1.644496656425086e-05,
|
| 241 |
+
"loss": 1.063,
|
| 242 |
+
"step": 10000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.9,
|
| 246 |
+
"eval_exact_match": 79.5837275307474,
|
| 247 |
+
"eval_f1": 87.50572394546504,
|
| 248 |
+
"step": 10000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.95,
|
| 252 |
+
"learning_rate": 1.5767214892463402e-05,
|
| 253 |
+
"loss": 1.0353,
|
| 254 |
+
"step": 10500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.95,
|
| 258 |
+
"eval_exact_match": 80.10406811731315,
|
| 259 |
+
"eval_f1": 87.58460117305698,
|
| 260 |
+
"step": 10500
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.99,
|
| 264 |
+
"learning_rate": 1.5089463220675944e-05,
|
| 265 |
+
"loss": 1.0198,
|
| 266 |
+
"step": 11000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.99,
|
| 270 |
+
"eval_exact_match": 80.1135288552507,
|
| 271 |
+
"eval_f1": 87.63553208656376,
|
| 272 |
+
"step": 11000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 1.04,
|
| 276 |
+
"learning_rate": 1.4411711548888486e-05,
|
| 277 |
+
"loss": 0.7558,
|
| 278 |
+
"step": 11500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.04,
|
| 282 |
+
"eval_exact_match": 80.00946073793756,
|
| 283 |
+
"eval_f1": 87.60930859917372,
|
| 284 |
+
"step": 11500
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.08,
|
| 288 |
+
"learning_rate": 1.373395987710103e-05,
|
| 289 |
+
"loss": 0.7337,
|
| 290 |
+
"step": 12000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.08,
|
| 294 |
+
"eval_exact_match": 79.92431409649953,
|
| 295 |
+
"eval_f1": 87.57399698225302,
|
| 296 |
+
"step": 12000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.13,
|
| 300 |
+
"learning_rate": 1.3056208205313572e-05,
|
| 301 |
+
"loss": 0.6884,
|
| 302 |
+
"step": 12500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.13,
|
| 306 |
+
"eval_exact_match": 80.2554399243141,
|
| 307 |
+
"eval_f1": 87.80990616116381,
|
| 308 |
+
"step": 12500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.17,
|
| 312 |
+
"learning_rate": 1.2378456533526116e-05,
|
| 313 |
+
"loss": 0.7082,
|
| 314 |
+
"step": 13000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.17,
|
| 318 |
+
"eval_exact_match": 80.17975402081362,
|
| 319 |
+
"eval_f1": 87.80432576834488,
|
| 320 |
+
"step": 13000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 1.22,
|
| 324 |
+
"learning_rate": 1.1700704861738658e-05,
|
| 325 |
+
"loss": 0.6914,
|
| 326 |
+
"step": 13500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 1.22,
|
| 330 |
+
"eval_exact_match": 79.87701040681173,
|
| 331 |
+
"eval_f1": 87.89607002450978,
|
| 332 |
+
"step": 13500
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 1.27,
|
| 336 |
+
"learning_rate": 1.1022953189951202e-05,
|
| 337 |
+
"loss": 0.7051,
|
| 338 |
+
"step": 14000
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 1.27,
|
| 342 |
+
"eval_exact_match": 79.80132450331126,
|
| 343 |
+
"eval_f1": 87.69384875801946,
|
| 344 |
+
"step": 14000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 1.31,
|
| 348 |
+
"learning_rate": 1.0345201518163744e-05,
|
| 349 |
+
"loss": 0.7302,
|
| 350 |
+
"step": 14500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 1.31,
|
| 354 |
+
"eval_exact_match": 80.2081362346263,
|
| 355 |
+
"eval_f1": 88.06697491299282,
|
| 356 |
+
"step": 14500
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 1.36,
|
| 360 |
+
"learning_rate": 9.667449846376287e-06,
|
| 361 |
+
"loss": 0.6824,
|
| 362 |
+
"step": 15000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 1.36,
|
| 366 |
+
"eval_exact_match": 80.50141911069063,
|
| 367 |
+
"eval_f1": 87.97379997606235,
|
| 368 |
+
"step": 15000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 1.4,
|
| 372 |
+
"learning_rate": 8.98969817458883e-06,
|
| 373 |
+
"loss": 0.7169,
|
| 374 |
+
"step": 15500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 1.4,
|
| 378 |
+
"eval_exact_match": 80.35004730368969,
|
| 379 |
+
"eval_f1": 88.00649157738846,
|
| 380 |
+
"step": 15500
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 1.45,
|
| 384 |
+
"learning_rate": 8.311946502801373e-06,
|
| 385 |
+
"loss": 0.7149,
|
| 386 |
+
"step": 16000
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 1.45,
|
| 390 |
+
"eval_exact_match": 80.74739829706716,
|
| 391 |
+
"eval_f1": 87.99961546836674,
|
| 392 |
+
"step": 16000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 1.49,
|
| 396 |
+
"learning_rate": 7.634194831013915e-06,
|
| 397 |
+
"loss": 0.6726,
|
| 398 |
+
"step": 16500
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 1.49,
|
| 402 |
+
"eval_exact_match": 80.93661305581836,
|
| 403 |
+
"eval_f1": 88.06055493342305,
|
| 404 |
+
"step": 16500
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.54,
|
| 408 |
+
"learning_rate": 6.95644315922646e-06,
|
| 409 |
+
"loss": 0.6761,
|
| 410 |
+
"step": 17000
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 1.54,
|
| 414 |
+
"eval_exact_match": 80.50141911069063,
|
| 415 |
+
"eval_f1": 87.98739385338236,
|
| 416 |
+
"step": 17000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 1.58,
|
| 420 |
+
"learning_rate": 6.278691487439003e-06,
|
| 421 |
+
"loss": 0.712,
|
| 422 |
+
"step": 17500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.58,
|
| 426 |
+
"eval_exact_match": 81.0406811731315,
|
| 427 |
+
"eval_f1": 88.4045499003394,
|
| 428 |
+
"step": 17500
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 1.63,
|
| 432 |
+
"learning_rate": 5.6009398156515455e-06,
|
| 433 |
+
"loss": 0.693,
|
| 434 |
+
"step": 18000
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 1.63,
|
| 438 |
+
"eval_exact_match": 80.69063386944181,
|
| 439 |
+
"eval_f1": 88.19624181405425,
|
| 440 |
+
"step": 18000
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 1.67,
|
| 444 |
+
"learning_rate": 4.923188143864088e-06,
|
| 445 |
+
"loss": 0.6594,
|
| 446 |
+
"step": 18500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.67,
|
| 450 |
+
"eval_exact_match": 80.97445600756859,
|
| 451 |
+
"eval_f1": 88.26284168848002,
|
| 452 |
+
"step": 18500
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 1.72,
|
| 456 |
+
"learning_rate": 4.245436472076631e-06,
|
| 457 |
+
"loss": 0.6743,
|
| 458 |
+
"step": 19000
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.72,
|
| 462 |
+
"eval_exact_match": 80.43519394512772,
|
| 463 |
+
"eval_f1": 88.07199660434318,
|
| 464 |
+
"step": 19000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.76,
|
| 468 |
+
"learning_rate": 3.567684800289174e-06,
|
| 469 |
+
"loss": 0.6876,
|
| 470 |
+
"step": 19500
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.76,
|
| 474 |
+
"eval_exact_match": 80.88930936613056,
|
| 475 |
+
"eval_f1": 88.35346447229631,
|
| 476 |
+
"step": 19500
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 1.81,
|
| 480 |
+
"learning_rate": 2.889933128501717e-06,
|
| 481 |
+
"loss": 0.6913,
|
| 482 |
+
"step": 20000
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.81,
|
| 486 |
+
"eval_exact_match": 81.22989593188268,
|
| 487 |
+
"eval_f1": 88.55310003155216,
|
| 488 |
+
"step": 20000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 1.85,
|
| 492 |
+
"learning_rate": 2.21218145671426e-06,
|
| 493 |
+
"loss": 0.6383,
|
| 494 |
+
"step": 20500
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 1.85,
|
| 498 |
+
"eval_exact_match": 81.35288552507096,
|
| 499 |
+
"eval_f1": 88.58308022828253,
|
| 500 |
+
"step": 20500
|
| 501 |
+
}
|
| 502 |
+
],
|
| 503 |
+
"max_steps": 22132,
|
| 504 |
+
"num_train_epochs": 2,
|
| 505 |
+
"total_flos": 3.213871718825779e+16,
|
| 506 |
+
"trial_name": null,
|
| 507 |
+
"trial_params": null
|
| 508 |
+
}
|