Commit ·
ba51f12
1
Parent(s): 6c87579
Upload trainer_state.json
Browse files- trainer_state.json +331 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,331 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 29.997802197802198,
|
| 5 |
+
"global_step": 6810,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.32,
|
| 12 |
+
"learning_rate": 0.00017999999999999998,
|
| 13 |
+
"loss": 11.6291,
|
| 14 |
+
"step": 300
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.32,
|
| 18 |
+
"eval_loss": 3.028644323348999,
|
| 19 |
+
"eval_runtime": 643.9816,
|
| 20 |
+
"eval_samples_per_second": 8.087,
|
| 21 |
+
"eval_wer": 1.0,
|
| 22 |
+
"step": 300
|
| 23 |
+
},
|
| 24 |
+
{
|
| 25 |
+
"epoch": 2.64,
|
| 26 |
+
"learning_rate": 0.0002952456418383518,
|
| 27 |
+
"loss": 3.0796,
|
| 28 |
+
"step": 600
|
| 29 |
+
},
|
| 30 |
+
{
|
| 31 |
+
"epoch": 2.64,
|
| 32 |
+
"eval_loss": 2.9614171981811523,
|
| 33 |
+
"eval_runtime": 650.4972,
|
| 34 |
+
"eval_samples_per_second": 8.006,
|
| 35 |
+
"eval_wer": 1.0,
|
| 36 |
+
"step": 600
|
| 37 |
+
},
|
| 38 |
+
{
|
| 39 |
+
"epoch": 3.96,
|
| 40 |
+
"learning_rate": 0.0002809825673534073,
|
| 41 |
+
"loss": 1.6756,
|
| 42 |
+
"step": 900
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 3.96,
|
| 46 |
+
"eval_loss": 0.6643654108047485,
|
| 47 |
+
"eval_runtime": 668.5248,
|
| 48 |
+
"eval_samples_per_second": 7.79,
|
| 49 |
+
"eval_wer": 0.5919852387016294,
|
| 50 |
+
"step": 900
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 5.29,
|
| 54 |
+
"learning_rate": 0.00026671949286846274,
|
| 55 |
+
"loss": 0.5169,
|
| 56 |
+
"step": 1200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 5.29,
|
| 60 |
+
"eval_loss": 0.48998841643333435,
|
| 61 |
+
"eval_runtime": 670.8355,
|
| 62 |
+
"eval_samples_per_second": 7.763,
|
| 63 |
+
"eval_wer": 0.4740247927440942,
|
| 64 |
+
"step": 1200
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 6.61,
|
| 68 |
+
"learning_rate": 0.0002524564183835182,
|
| 69 |
+
"loss": 0.3643,
|
| 70 |
+
"step": 1500
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 6.61,
|
| 74 |
+
"eval_loss": 0.4465296268463135,
|
| 75 |
+
"eval_runtime": 685.664,
|
| 76 |
+
"eval_samples_per_second": 7.596,
|
| 77 |
+
"eval_wer": 0.43660178279061307,
|
| 78 |
+
"step": 1500
|
| 79 |
+
},
|
| 80 |
+
{
|
| 81 |
+
"epoch": 7.93,
|
| 82 |
+
"learning_rate": 0.00023819334389857367,
|
| 83 |
+
"loss": 0.277,
|
| 84 |
+
"step": 1800
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 7.93,
|
| 88 |
+
"eval_loss": 0.4225097894668579,
|
| 89 |
+
"eval_runtime": 675.9712,
|
| 90 |
+
"eval_samples_per_second": 7.704,
|
| 91 |
+
"eval_wer": 0.4157592453026326,
|
| 92 |
+
"step": 1800
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 9.25,
|
| 96 |
+
"learning_rate": 0.00022393026941362913,
|
| 97 |
+
"loss": 0.2271,
|
| 98 |
+
"step": 2100
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 9.25,
|
| 102 |
+
"eval_loss": 0.4406384825706482,
|
| 103 |
+
"eval_runtime": 685.3056,
|
| 104 |
+
"eval_samples_per_second": 7.6,
|
| 105 |
+
"eval_wer": 0.39260375789391616,
|
| 106 |
+
"step": 2100
|
| 107 |
+
},
|
| 108 |
+
{
|
| 109 |
+
"epoch": 10.57,
|
| 110 |
+
"learning_rate": 0.0002096671949286846,
|
| 111 |
+
"loss": 0.1895,
|
| 112 |
+
"step": 2400
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 10.57,
|
| 116 |
+
"eval_loss": 0.4478096067905426,
|
| 117 |
+
"eval_runtime": 689.1958,
|
| 118 |
+
"eval_samples_per_second": 7.557,
|
| 119 |
+
"eval_wer": 0.3899529613555446,
|
| 120 |
+
"step": 2400
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 11.89,
|
| 124 |
+
"learning_rate": 0.0001954041204437401,
|
| 125 |
+
"loss": 0.1653,
|
| 126 |
+
"step": 2700
|
| 127 |
+
},
|
| 128 |
+
{
|
| 129 |
+
"epoch": 11.89,
|
| 130 |
+
"eval_loss": 0.4374055862426758,
|
| 131 |
+
"eval_runtime": 693.8794,
|
| 132 |
+
"eval_samples_per_second": 7.506,
|
| 133 |
+
"eval_wer": 0.3792977987993451,
|
| 134 |
+
"step": 2700
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 13.22,
|
| 138 |
+
"learning_rate": 0.00018114104595879556,
|
| 139 |
+
"loss": 0.1366,
|
| 140 |
+
"step": 3000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 13.22,
|
| 144 |
+
"eval_loss": 0.4575331211090088,
|
| 145 |
+
"eval_runtime": 703.7067,
|
| 146 |
+
"eval_samples_per_second": 7.401,
|
| 147 |
+
"eval_wer": 0.37355440629954,
|
| 148 |
+
"step": 3000
|
| 149 |
+
},
|
| 150 |
+
{
|
| 151 |
+
"epoch": 14.54,
|
| 152 |
+
"learning_rate": 0.000166877971473851,
|
| 153 |
+
"loss": 0.1219,
|
| 154 |
+
"step": 3300
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 14.54,
|
| 158 |
+
"eval_loss": 0.4847738444805145,
|
| 159 |
+
"eval_runtime": 705.6195,
|
| 160 |
+
"eval_samples_per_second": 7.381,
|
| 161 |
+
"eval_wer": 0.36983809350554847,
|
| 162 |
+
"step": 3300
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 15.86,
|
| 166 |
+
"learning_rate": 0.00015261489698890646,
|
| 167 |
+
"loss": 0.1102,
|
| 168 |
+
"step": 3600
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 15.86,
|
| 172 |
+
"eval_loss": 0.4509291350841522,
|
| 173 |
+
"eval_runtime": 713.2815,
|
| 174 |
+
"eval_samples_per_second": 7.301,
|
| 175 |
+
"eval_wer": 0.35738974505574467,
|
| 176 |
+
"step": 3600
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 17.18,
|
| 180 |
+
"learning_rate": 0.00013835182250396196,
|
| 181 |
+
"loss": 0.1017,
|
| 182 |
+
"step": 3900
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 17.18,
|
| 186 |
+
"eval_loss": 0.45042145252227783,
|
| 187 |
+
"eval_runtime": 714.3616,
|
| 188 |
+
"eval_samples_per_second": 7.29,
|
| 189 |
+
"eval_wer": 0.34715039372125056,
|
| 190 |
+
"step": 3900
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 18.5,
|
| 194 |
+
"learning_rate": 0.00012408874801901742,
|
| 195 |
+
"loss": 0.0937,
|
| 196 |
+
"step": 4200
|
| 197 |
+
},
|
| 198 |
+
{
|
| 199 |
+
"epoch": 18.5,
|
| 200 |
+
"eval_loss": 0.4522610008716583,
|
| 201 |
+
"eval_runtime": 719.9965,
|
| 202 |
+
"eval_samples_per_second": 7.233,
|
| 203 |
+
"eval_wer": 0.3469164999090413,
|
| 204 |
+
"step": 4200
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 19.82,
|
| 208 |
+
"learning_rate": 0.00010982567353407289,
|
| 209 |
+
"loss": 0.0829,
|
| 210 |
+
"step": 4500
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 19.82,
|
| 214 |
+
"eval_loss": 0.4488724172115326,
|
| 215 |
+
"eval_runtime": 720.1162,
|
| 216 |
+
"eval_samples_per_second": 7.232,
|
| 217 |
+
"eval_wer": 0.3452012786195067,
|
| 218 |
+
"step": 4500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 21.15,
|
| 222 |
+
"learning_rate": 9.556259904912835e-05,
|
| 223 |
+
"loss": 0.0752,
|
| 224 |
+
"step": 4800
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 21.15,
|
| 228 |
+
"eval_loss": 0.4762090742588043,
|
| 229 |
+
"eval_runtime": 730.1963,
|
| 230 |
+
"eval_samples_per_second": 7.132,
|
| 231 |
+
"eval_wer": 0.33800254684373293,
|
| 232 |
+
"step": 4800
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 22.47,
|
| 236 |
+
"learning_rate": 8.129952456418383e-05,
|
| 237 |
+
"loss": 0.0698,
|
| 238 |
+
"step": 5100
|
| 239 |
+
},
|
| 240 |
+
{
|
| 241 |
+
"epoch": 22.47,
|
| 242 |
+
"eval_loss": 0.480989009141922,
|
| 243 |
+
"eval_runtime": 727.7194,
|
| 244 |
+
"eval_samples_per_second": 7.157,
|
| 245 |
+
"eval_wer": 0.33597546713791937,
|
| 246 |
+
"step": 5100
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 23.79,
|
| 250 |
+
"learning_rate": 6.70364500792393e-05,
|
| 251 |
+
"loss": 0.0642,
|
| 252 |
+
"step": 5400
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 23.79,
|
| 256 |
+
"eval_loss": 0.49236759543418884,
|
| 257 |
+
"eval_runtime": 737.7861,
|
| 258 |
+
"eval_samples_per_second": 7.059,
|
| 259 |
+
"eval_wer": 0.3343382104524546,
|
| 260 |
+
"step": 5400
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 25.11,
|
| 264 |
+
"learning_rate": 5.2773375594294765e-05,
|
| 265 |
+
"loss": 0.058,
|
| 266 |
+
"step": 5700
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 25.11,
|
| 270 |
+
"eval_loss": 0.4960506558418274,
|
| 271 |
+
"eval_runtime": 740.7836,
|
| 272 |
+
"eval_samples_per_second": 7.03,
|
| 273 |
+
"eval_wer": 0.3302840510408275,
|
| 274 |
+
"step": 5700
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 26.43,
|
| 278 |
+
"learning_rate": 3.851030110935023e-05,
|
| 279 |
+
"loss": 0.0553,
|
| 280 |
+
"step": 6000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 26.43,
|
| 284 |
+
"eval_loss": 0.4915299415588379,
|
| 285 |
+
"eval_runtime": 748.2447,
|
| 286 |
+
"eval_samples_per_second": 6.96,
|
| 287 |
+
"eval_wer": 0.32506042256815404,
|
| 288 |
+
"step": 6000
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 27.75,
|
| 292 |
+
"learning_rate": 2.4247226624405704e-05,
|
| 293 |
+
"loss": 0.0545,
|
| 294 |
+
"step": 6300
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 27.75,
|
| 298 |
+
"eval_loss": 0.5011223554611206,
|
| 299 |
+
"eval_runtime": 762.4791,
|
| 300 |
+
"eval_samples_per_second": 6.83,
|
| 301 |
+
"eval_wer": 0.3210062631565269,
|
| 302 |
+
"step": 6300
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 29.07,
|
| 306 |
+
"learning_rate": 9.984152139461172e-06,
|
| 307 |
+
"loss": 0.0523,
|
| 308 |
+
"step": 6600
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 29.07,
|
| 312 |
+
"eval_loss": 0.49885958433151245,
|
| 313 |
+
"eval_runtime": 743.9824,
|
| 314 |
+
"eval_samples_per_second": 7.0,
|
| 315 |
+
"eval_wer": 0.3229034018555576,
|
| 316 |
+
"step": 6600
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 30.0,
|
| 320 |
+
"step": 6810,
|
| 321 |
+
"total_flos": 4.998078272539758e+19,
|
| 322 |
+
"train_runtime": 61300.1017,
|
| 323 |
+
"train_samples_per_second": 0.111
|
| 324 |
+
}
|
| 325 |
+
],
|
| 326 |
+
"max_steps": 6810,
|
| 327 |
+
"num_train_epochs": 30,
|
| 328 |
+
"total_flos": 4.998078272539758e+19,
|
| 329 |
+
"trial_name": null,
|
| 330 |
+
"trial_params": null
|
| 331 |
+
}
|