Training in progress, step 500
Browse files- config.json +1 -1
- last-checkpoint/config.json +1 -1
- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scaler.pt +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +5 -683
- last-checkpoint/training_args.bin +2 -2
- pytorch_model.bin +1 -1
- runs/Oct14_12-15-12_e3717350dd21/1665750223.3983731/events.out.tfevents.1665750223.e3717350dd21.24.1 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750242.5320206/events.out.tfevents.1665750242.e3717350dd21.24.2 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750248.424454/events.out.tfevents.1665750248.e3717350dd21.24.3 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750250.8963432/events.out.tfevents.1665750250.e3717350dd21.24.4 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750252.3554833/events.out.tfevents.1665750252.e3717350dd21.24.5 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750253.167781/events.out.tfevents.1665750253.e3717350dd21.24.6 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750253.8073034/events.out.tfevents.1665750253.e3717350dd21.24.7 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750254.2377386/events.out.tfevents.1665750254.e3717350dd21.24.8 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750255.023941/events.out.tfevents.1665750255.e3717350dd21.24.9 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750255.9294362/events.out.tfevents.1665750255.e3717350dd21.24.10 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/1665750257.3678532/events.out.tfevents.1665750257.e3717350dd21.24.11 +3 -0
- runs/Oct14_12-15-12_e3717350dd21/events.out.tfevents.1665750223.e3717350dd21.24.0 +3 -0
- training_args.bin +2 -2
config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "
|
| 3 |
"activation_function": "gelu_new",
|
| 4 |
"architectures": [
|
| 5 |
"GPTNeoForCausalLM"
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "cedpsam/EleutherAI_gpt-neo-125M-stablediffionprompts",
|
| 3 |
"activation_function": "gelu_new",
|
| 4 |
"architectures": [
|
| 5 |
"GPTNeoForCausalLM"
|
last-checkpoint/config.json
CHANGED
|
@@ -1,5 +1,5 @@
|
|
| 1 |
{
|
| 2 |
-
"_name_or_path": "
|
| 3 |
"activation_function": "gelu_new",
|
| 4 |
"architectures": [
|
| 5 |
"GPTNeoForCausalLM"
|
| 1 |
{
|
| 2 |
+
"_name_or_path": "cedpsam/EleutherAI_gpt-neo-125M-stablediffionprompts",
|
| 3 |
"activation_function": "gelu_new",
|
| 4 |
"architectures": [
|
| 5 |
"GPTNeoForCausalLM"
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1001681601
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5befd02eeceef15599300b21e2ad4c744b7c43d7ac572f30756beb54fb3c5ab0
|
| 3 |
size 1001681601
|
last-checkpoint/pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 551185105
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e323cff0bca0bc2e6934298b8e92f0bc0a28c258605b1e175656618d1fc524e7
|
| 3 |
size 551185105
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14503
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:62851f63d26f31f0098aaf22bfc590da9d5dd492f2a149b0c502c13be7100a3d
|
| 3 |
size 14503
|
last-checkpoint/scaler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 559
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3978f33a9eccdea52f77a26979e6a0077a50cd239694fba6bc25cd0891e8edf1
|
| 3 |
size 559
|
last-checkpoint/scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 623
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0d5a0f188c3720b88efe77da2bf8bc1f175f481de9e3f20b0aaa889683ccd7ea
|
| 3 |
size 623
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,700 +1,22 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 5 |
-
"global_step":
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.01,
|
| 12 |
-
"learning_rate": 4.9321739602268106e-05,
|
| 13 |
-
"loss": 0.0506,
|
| 14 |
-
"step": 500
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.01,
|
| 18 |
-
"learning_rate": 4.864347920453621e-05,
|
| 19 |
-
"loss": 0.0442,
|
| 20 |
-
"step": 1000
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.02,
|
| 24 |
-
"learning_rate": 4.796521880680431e-05,
|
| 25 |
-
"loss": 0.0454,
|
| 26 |
-
"step": 1500
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.03,
|
| 30 |
-
"learning_rate": 4.7286958409072415e-05,
|
| 31 |
-
"loss": 0.0447,
|
| 32 |
-
"step": 2000
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.03,
|
| 36 |
-
"learning_rate": 4.660869801134052e-05,
|
| 37 |
-
"loss": 0.0455,
|
| 38 |
-
"step": 2500
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.04,
|
| 42 |
-
"learning_rate": 4.5930437613608615e-05,
|
| 43 |
-
"loss": 0.046,
|
| 44 |
-
"step": 3000
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.05,
|
| 48 |
-
"learning_rate": 4.5252177215876725e-05,
|
| 49 |
-
"loss": 0.0452,
|
| 50 |
-
"step": 3500
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.05,
|
| 54 |
-
"learning_rate": 4.457391681814482e-05,
|
| 55 |
-
"loss": 0.0441,
|
| 56 |
-
"step": 4000
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.06,
|
| 60 |
-
"learning_rate": 4.389565642041293e-05,
|
| 61 |
-
"loss": 0.0459,
|
| 62 |
-
"step": 4500
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.07,
|
| 66 |
-
"learning_rate": 4.321739602268103e-05,
|
| 67 |
-
"loss": 0.0433,
|
| 68 |
-
"step": 5000
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.07,
|
| 72 |
-
"learning_rate": 4.25404921457446e-05,
|
| 73 |
-
"loss": 0.0423,
|
| 74 |
-
"step": 5500
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.08,
|
| 78 |
-
"learning_rate": 4.18622317480127e-05,
|
| 79 |
-
"loss": 0.0417,
|
| 80 |
-
"step": 6000
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.09,
|
| 84 |
-
"learning_rate": 4.1183971350280804e-05,
|
| 85 |
-
"loss": 0.0438,
|
| 86 |
-
"step": 6500
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.09,
|
| 90 |
-
"learning_rate": 4.05057109525489e-05,
|
| 91 |
-
"loss": 0.043,
|
| 92 |
-
"step": 7000
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 0.1,
|
| 96 |
-
"learning_rate": 3.9828807075612476e-05,
|
| 97 |
-
"loss": 0.0442,
|
| 98 |
-
"step": 7500
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.11,
|
| 102 |
-
"learning_rate": 3.915054667788057e-05,
|
| 103 |
-
"loss": 0.0424,
|
| 104 |
-
"step": 8000
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 0.12,
|
| 108 |
-
"learning_rate": 3.8472286280148676e-05,
|
| 109 |
-
"loss": 0.0434,
|
| 110 |
-
"step": 8500
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.12,
|
| 114 |
-
"learning_rate": 3.779402588241678e-05,
|
| 115 |
-
"loss": 0.0418,
|
| 116 |
-
"step": 9000
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.13,
|
| 120 |
-
"learning_rate": 3.711576548468488e-05,
|
| 121 |
-
"loss": 0.0404,
|
| 122 |
-
"step": 9500
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.14,
|
| 126 |
-
"learning_rate": 3.6438861607748445e-05,
|
| 127 |
-
"loss": 0.0405,
|
| 128 |
-
"step": 10000
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.14,
|
| 132 |
-
"learning_rate": 3.5760601210016555e-05,
|
| 133 |
-
"loss": 0.0404,
|
| 134 |
-
"step": 10500
|
| 135 |
-
},
|
| 136 |
-
{
|
| 137 |
-
"epoch": 0.15,
|
| 138 |
-
"learning_rate": 3.508234081228465e-05,
|
| 139 |
-
"loss": 0.0429,
|
| 140 |
-
"step": 11000
|
| 141 |
-
},
|
| 142 |
-
{
|
| 143 |
-
"epoch": 0.16,
|
| 144 |
-
"learning_rate": 3.440408041455276e-05,
|
| 145 |
-
"loss": 0.0411,
|
| 146 |
-
"step": 11500
|
| 147 |
-
},
|
| 148 |
-
{
|
| 149 |
-
"epoch": 0.16,
|
| 150 |
-
"learning_rate": 3.372582001682086e-05,
|
| 151 |
-
"loss": 0.0407,
|
| 152 |
-
"step": 12000
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"epoch": 0.17,
|
| 156 |
-
"learning_rate": 3.304755961908896e-05,
|
| 157 |
-
"loss": 0.0412,
|
| 158 |
-
"step": 12500
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.18,
|
| 162 |
-
"learning_rate": 3.2369299221357064e-05,
|
| 163 |
-
"loss": 0.0395,
|
| 164 |
-
"step": 13000
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.18,
|
| 168 |
-
"learning_rate": 3.169103882362517e-05,
|
| 169 |
-
"loss": 0.0414,
|
| 170 |
-
"step": 13500
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 0.19,
|
| 174 |
-
"learning_rate": 3.101277842589327e-05,
|
| 175 |
-
"loss": 0.0405,
|
| 176 |
-
"step": 14000
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 0.2,
|
| 180 |
-
"learning_rate": 3.0334518028161374e-05,
|
| 181 |
-
"loss": 0.0409,
|
| 182 |
-
"step": 14500
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"epoch": 0.2,
|
| 186 |
-
"learning_rate": 2.9656257630429474e-05,
|
| 187 |
-
"loss": 0.0424,
|
| 188 |
-
"step": 15000
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 0.21,
|
| 192 |
-
"learning_rate": 2.8979353753493043e-05,
|
| 193 |
-
"loss": 0.0432,
|
| 194 |
-
"step": 15500
|
| 195 |
-
},
|
| 196 |
-
{
|
| 197 |
-
"epoch": 0.22,
|
| 198 |
-
"learning_rate": 2.8301093355761143e-05,
|
| 199 |
-
"loss": 0.0409,
|
| 200 |
-
"step": 16000
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 0.22,
|
| 204 |
-
"learning_rate": 2.762283295802925e-05,
|
| 205 |
-
"loss": 0.0416,
|
| 206 |
-
"step": 16500
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.23,
|
| 210 |
-
"learning_rate": 2.694457256029735e-05,
|
| 211 |
-
"loss": 0.0408,
|
| 212 |
-
"step": 17000
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.24,
|
| 216 |
-
"learning_rate": 2.6266312162565452e-05,
|
| 217 |
-
"loss": 0.0393,
|
| 218 |
-
"step": 17500
|
| 219 |
-
},
|
| 220 |
-
{
|
| 221 |
-
"epoch": 0.24,
|
| 222 |
-
"learning_rate": 2.5588051764833552e-05,
|
| 223 |
-
"loss": 0.0378,
|
| 224 |
-
"step": 18000
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 0.25,
|
| 228 |
-
"learning_rate": 2.490979136710166e-05,
|
| 229 |
-
"loss": 0.0428,
|
| 230 |
-
"step": 18500
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"epoch": 0.26,
|
| 234 |
-
"learning_rate": 2.4232887490165228e-05,
|
| 235 |
-
"loss": 0.0418,
|
| 236 |
-
"step": 19000
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 0.26,
|
| 240 |
-
"learning_rate": 2.3554627092433328e-05,
|
| 241 |
-
"loss": 0.0392,
|
| 242 |
-
"step": 19500
|
| 243 |
-
},
|
| 244 |
-
{
|
| 245 |
-
"epoch": 0.27,
|
| 246 |
-
"learning_rate": 2.287636669470143e-05,
|
| 247 |
-
"loss": 0.0403,
|
| 248 |
-
"step": 20000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.28,
|
| 252 |
-
"learning_rate": 2.2198106296969534e-05,
|
| 253 |
-
"loss": 0.0418,
|
| 254 |
-
"step": 20500
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.28,
|
| 258 |
-
"learning_rate": 2.1519845899237638e-05,
|
| 259 |
-
"loss": 0.0392,
|
| 260 |
-
"step": 21000
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.29,
|
| 264 |
-
"learning_rate": 2.0842942022301203e-05,
|
| 265 |
-
"loss": 0.0405,
|
| 266 |
-
"step": 21500
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 0.3,
|
| 270 |
-
"learning_rate": 2.0164681624569307e-05,
|
| 271 |
-
"loss": 0.038,
|
| 272 |
-
"step": 22000
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 0.31,
|
| 276 |
-
"learning_rate": 1.948642122683741e-05,
|
| 277 |
-
"loss": 0.0373,
|
| 278 |
-
"step": 22500
|
| 279 |
-
},
|
| 280 |
-
{
|
| 281 |
-
"epoch": 0.31,
|
| 282 |
-
"learning_rate": 1.8808160829105513e-05,
|
| 283 |
-
"loss": 0.0437,
|
| 284 |
-
"step": 23000
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 0.32,
|
| 288 |
-
"learning_rate": 1.8129900431373613e-05,
|
| 289 |
-
"loss": 0.0395,
|
| 290 |
-
"step": 23500
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.33,
|
| 294 |
-
"learning_rate": 1.7451640033641716e-05,
|
| 295 |
-
"loss": 0.0376,
|
| 296 |
-
"step": 24000
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 0.33,
|
| 300 |
-
"learning_rate": 1.677337963590982e-05,
|
| 301 |
-
"loss": 0.0394,
|
| 302 |
-
"step": 24500
|
| 303 |
-
},
|
| 304 |
-
{
|
| 305 |
-
"epoch": 0.34,
|
| 306 |
-
"learning_rate": 1.6095119238177923e-05,
|
| 307 |
-
"loss": 0.0424,
|
| 308 |
-
"step": 25000
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 0.35,
|
| 312 |
-
"learning_rate": 1.5416858840446026e-05,
|
| 313 |
-
"loss": 0.0425,
|
| 314 |
-
"step": 25500
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"epoch": 0.35,
|
| 318 |
-
"learning_rate": 1.4738598442714127e-05,
|
| 319 |
-
"loss": 0.0401,
|
| 320 |
-
"step": 26000
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 0.36,
|
| 324 |
-
"learning_rate": 1.4061694565777695e-05,
|
| 325 |
-
"loss": 0.0397,
|
| 326 |
-
"step": 26500
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 0.37,
|
| 330 |
-
"learning_rate": 1.3383434168045796e-05,
|
| 331 |
-
"loss": 0.0423,
|
| 332 |
-
"step": 27000
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.37,
|
| 336 |
-
"learning_rate": 1.27051737703139e-05,
|
| 337 |
-
"loss": 0.0393,
|
| 338 |
-
"step": 27500
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 0.38,
|
| 342 |
-
"learning_rate": 1.2026913372582003e-05,
|
| 343 |
-
"loss": 0.0407,
|
| 344 |
-
"step": 28000
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 0.39,
|
| 348 |
-
"learning_rate": 1.1348652974850104e-05,
|
| 349 |
-
"loss": 0.0405,
|
| 350 |
-
"step": 28500
|
| 351 |
-
},
|
| 352 |
-
{
|
| 353 |
-
"epoch": 0.39,
|
| 354 |
-
"learning_rate": 1.0671749097913672e-05,
|
| 355 |
-
"loss": 0.0383,
|
| 356 |
-
"step": 29000
|
| 357 |
-
},
|
| 358 |
-
{
|
| 359 |
-
"epoch": 0.4,
|
| 360 |
-
"learning_rate": 9.993488700181775e-06,
|
| 361 |
-
"loss": 0.0405,
|
| 362 |
-
"step": 29500
|
| 363 |
-
},
|
| 364 |
-
{
|
| 365 |
-
"epoch": 0.41,
|
| 366 |
-
"learning_rate": 9.315228302449877e-06,
|
| 367 |
-
"loss": 0.0407,
|
| 368 |
-
"step": 30000
|
| 369 |
-
},
|
| 370 |
-
{
|
| 371 |
-
"epoch": 0.41,
|
| 372 |
-
"learning_rate": 8.63696790471798e-06,
|
| 373 |
-
"loss": 0.0432,
|
| 374 |
-
"step": 30500
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.42,
|
| 378 |
-
"learning_rate": 7.960064027781547e-06,
|
| 379 |
-
"loss": 0.0409,
|
| 380 |
-
"step": 31000
|
| 381 |
-
},
|
| 382 |
-
{
|
| 383 |
-
"epoch": 0.43,
|
| 384 |
-
"learning_rate": 7.28180363004965e-06,
|
| 385 |
-
"loss": 0.0408,
|
| 386 |
-
"step": 31500
|
| 387 |
-
},
|
| 388 |
-
{
|
| 389 |
-
"epoch": 0.43,
|
| 390 |
-
"learning_rate": 6.603543232317752e-06,
|
| 391 |
-
"loss": 0.0429,
|
| 392 |
-
"step": 32000
|
| 393 |
-
},
|
| 394 |
-
{
|
| 395 |
-
"epoch": 0.44,
|
| 396 |
-
"learning_rate": 5.925282834585855e-06,
|
| 397 |
-
"loss": 0.0398,
|
| 398 |
-
"step": 32500
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 0.45,
|
| 402 |
-
"learning_rate": 5.247022436853957e-06,
|
| 403 |
-
"loss": 0.0379,
|
| 404 |
-
"step": 33000
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 0.45,
|
| 408 |
-
"learning_rate": 4.568762039122059e-06,
|
| 409 |
-
"loss": 0.0409,
|
| 410 |
-
"step": 33500
|
| 411 |
-
},
|
| 412 |
-
{
|
| 413 |
-
"epoch": 0.46,
|
| 414 |
-
"learning_rate": 3.890501641390163e-06,
|
| 415 |
-
"loss": 0.0394,
|
| 416 |
-
"step": 34000
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.47,
|
| 420 |
-
"learning_rate": 3.2122412436582654e-06,
|
| 421 |
-
"loss": 0.0392,
|
| 422 |
-
"step": 34500
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 0.47,
|
| 426 |
-
"learning_rate": 2.5339808459263683e-06,
|
| 427 |
-
"loss": 0.0383,
|
| 428 |
-
"step": 35000
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 0.48,
|
| 432 |
-
"learning_rate": 1.8570769689899347e-06,
|
| 433 |
-
"loss": 0.0394,
|
| 434 |
-
"step": 35500
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 0.49,
|
| 438 |
-
"learning_rate": 1.1788165712580375e-06,
|
| 439 |
-
"loss": 0.041,
|
| 440 |
-
"step": 36000
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 0.5,
|
| 444 |
-
"learning_rate": 5.005561735261402e-07,
|
| 445 |
-
"loss": 0.0407,
|
| 446 |
-
"step": 36500
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 0.5,
|
| 450 |
-
"learning_rate": 0.0,
|
| 451 |
-
"loss": 0.042,
|
| 452 |
-
"step": 37000
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 0.51,
|
| 456 |
-
"learning_rate": 0.0,
|
| 457 |
-
"loss": 0.0388,
|
| 458 |
-
"step": 37500
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.52,
|
| 462 |
-
"learning_rate": 0.0,
|
| 463 |
-
"loss": 0.0435,
|
| 464 |
-
"step": 38000
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 0.52,
|
| 468 |
-
"learning_rate": 0.0,
|
| 469 |
-
"loss": 0.0419,
|
| 470 |
-
"step": 38500
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 0.53,
|
| 474 |
-
"learning_rate": 0.0,
|
| 475 |
-
"loss": 0.0411,
|
| 476 |
-
"step": 39000
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 0.54,
|
| 480 |
-
"learning_rate": 0.0,
|
| 481 |
-
"loss": 0.0418,
|
| 482 |
-
"step": 39500
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 0.54,
|
| 486 |
-
"learning_rate": 0.0,
|
| 487 |
-
"loss": 0.0396,
|
| 488 |
-
"step": 40000
|
| 489 |
-
},
|
| 490 |
-
{
|
| 491 |
-
"epoch": 0.55,
|
| 492 |
-
"learning_rate": 0.0,
|
| 493 |
-
"loss": 0.042,
|
| 494 |
-
"step": 40500
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 0.56,
|
| 498 |
-
"learning_rate": 0.0,
|
| 499 |
-
"loss": 0.0426,
|
| 500 |
-
"step": 41000
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 0.56,
|
| 504 |
-
"learning_rate": 0.0,
|
| 505 |
-
"loss": 0.0411,
|
| 506 |
-
"step": 41500
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 0.57,
|
| 510 |
-
"learning_rate": 0.0,
|
| 511 |
-
"loss": 0.0394,
|
| 512 |
-
"step": 42000
|
| 513 |
-
},
|
| 514 |
-
{
|
| 515 |
-
"epoch": 0.58,
|
| 516 |
-
"learning_rate": 0.0,
|
| 517 |
-
"loss": 0.0405,
|
| 518 |
-
"step": 42500
|
| 519 |
-
},
|
| 520 |
-
{
|
| 521 |
-
"epoch": 0.58,
|
| 522 |
-
"learning_rate": 0.0,
|
| 523 |
-
"loss": 0.0406,
|
| 524 |
-
"step": 43000
|
| 525 |
-
},
|
| 526 |
-
{
|
| 527 |
-
"epoch": 0.59,
|
| 528 |
-
"learning_rate": 0.0,
|
| 529 |
-
"loss": 0.043,
|
| 530 |
-
"step": 43500
|
| 531 |
-
},
|
| 532 |
-
{
|
| 533 |
-
"epoch": 0.6,
|
| 534 |
-
"learning_rate": 0.0,
|
| 535 |
-
"loss": 0.0409,
|
| 536 |
-
"step": 44000
|
| 537 |
-
},
|
| 538 |
-
{
|
| 539 |
-
"epoch": 0.6,
|
| 540 |
-
"learning_rate": 0.0,
|
| 541 |
-
"loss": 0.0405,
|
| 542 |
-
"step": 44500
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 0.61,
|
| 546 |
-
"learning_rate": 0.0,
|
| 547 |
-
"loss": 0.0395,
|
| 548 |
-
"step": 45000
|
| 549 |
-
},
|
| 550 |
-
{
|
| 551 |
-
"epoch": 0.62,
|
| 552 |
-
"learning_rate": 0.0,
|
| 553 |
-
"loss": 0.0412,
|
| 554 |
-
"step": 45500
|
| 555 |
-
},
|
| 556 |
-
{
|
| 557 |
-
"epoch": 0.62,
|
| 558 |
-
"learning_rate": 0.0,
|
| 559 |
-
"loss": 0.0381,
|
| 560 |
-
"step": 46000
|
| 561 |
-
},
|
| 562 |
-
{
|
| 563 |
-
"epoch": 0.63,
|
| 564 |
-
"learning_rate": 0.0,
|
| 565 |
-
"loss": 0.0409,
|
| 566 |
-
"step": 46500
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 0.64,
|
| 570 |
-
"learning_rate": 0.0,
|
| 571 |
-
"loss": 0.0419,
|
| 572 |
-
"step": 47000
|
| 573 |
-
},
|
| 574 |
-
{
|
| 575 |
-
"epoch": 0.64,
|
| 576 |
"learning_rate": 0.0,
|
| 577 |
-
"loss": 0.
|
| 578 |
-
"step":
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 0.65,
|
| 582 |
-
"learning_rate": 0.0,
|
| 583 |
-
"loss": 0.0413,
|
| 584 |
-
"step": 48000
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 0.66,
|
| 588 |
-
"learning_rate": 0.0,
|
| 589 |
-
"loss": 0.0409,
|
| 590 |
-
"step": 48500
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 0.66,
|
| 594 |
-
"learning_rate": 0.0,
|
| 595 |
-
"loss": 0.0442,
|
| 596 |
-
"step": 49000
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 0.67,
|
| 600 |
-
"learning_rate": 0.0,
|
| 601 |
-
"loss": 0.0416,
|
| 602 |
-
"step": 49500
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 0.68,
|
| 606 |
-
"learning_rate": 0.0,
|
| 607 |
-
"loss": 0.0399,
|
| 608 |
-
"step": 50000
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 0.69,
|
| 612 |
-
"learning_rate": 0.0,
|
| 613 |
-
"loss": 0.0416,
|
| 614 |
-
"step": 50500
|
| 615 |
-
},
|
| 616 |
-
{
|
| 617 |
-
"epoch": 0.69,
|
| 618 |
-
"learning_rate": 0.0,
|
| 619 |
-
"loss": 0.0405,
|
| 620 |
-
"step": 51000
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 0.7,
|
| 624 |
-
"learning_rate": 0.0,
|
| 625 |
-
"loss": 0.0426,
|
| 626 |
-
"step": 51500
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.71,
|
| 630 |
-
"learning_rate": 0.0,
|
| 631 |
-
"loss": 0.041,
|
| 632 |
-
"step": 52000
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 0.71,
|
| 636 |
-
"learning_rate": 0.0,
|
| 637 |
-
"loss": 0.0398,
|
| 638 |
-
"step": 52500
|
| 639 |
-
},
|
| 640 |
-
{
|
| 641 |
-
"epoch": 0.72,
|
| 642 |
-
"learning_rate": 0.0,
|
| 643 |
-
"loss": 0.0453,
|
| 644 |
-
"step": 53000
|
| 645 |
-
},
|
| 646 |
-
{
|
| 647 |
-
"epoch": 0.73,
|
| 648 |
-
"learning_rate": 0.0,
|
| 649 |
-
"loss": 0.0413,
|
| 650 |
-
"step": 53500
|
| 651 |
-
},
|
| 652 |
-
{
|
| 653 |
-
"epoch": 0.73,
|
| 654 |
-
"learning_rate": 0.0,
|
| 655 |
-
"loss": 0.0411,
|
| 656 |
-
"step": 54000
|
| 657 |
-
},
|
| 658 |
-
{
|
| 659 |
-
"epoch": 0.74,
|
| 660 |
-
"learning_rate": 0.0,
|
| 661 |
-
"loss": 0.0415,
|
| 662 |
-
"step": 54500
|
| 663 |
-
},
|
| 664 |
-
{
|
| 665 |
-
"epoch": 0.75,
|
| 666 |
-
"learning_rate": 0.0,
|
| 667 |
-
"loss": 0.043,
|
| 668 |
-
"step": 55000
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.75,
|
| 672 |
-
"learning_rate": 0.0,
|
| 673 |
-
"loss": 0.0425,
|
| 674 |
-
"step": 55500
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"epoch": 0.76,
|
| 678 |
-
"learning_rate": 0.0,
|
| 679 |
-
"loss": 0.0406,
|
| 680 |
-
"step": 56000
|
| 681 |
-
},
|
| 682 |
-
{
|
| 683 |
-
"epoch": 0.77,
|
| 684 |
-
"learning_rate": 0.0,
|
| 685 |
-
"loss": 0.0405,
|
| 686 |
-
"step": 56500
|
| 687 |
-
},
|
| 688 |
-
{
|
| 689 |
-
"epoch": 0.77,
|
| 690 |
-
"learning_rate": 0.0,
|
| 691 |
-
"loss": 0.0419,
|
| 692 |
-
"step": 57000
|
| 693 |
}
|
| 694 |
],
|
| 695 |
"max_steps": 73718,
|
| 696 |
"num_train_epochs": 1,
|
| 697 |
-
"total_flos":
|
| 698 |
"trial_name": null,
|
| 699 |
"trial_params": null
|
| 700 |
}
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.006782603977318972,
|
| 5 |
+
"global_step": 500,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
"epoch": 0.01,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
"learning_rate": 0.0,
|
| 13 |
+
"loss": 0.039,
|
| 14 |
+
"step": 500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
}
|
| 16 |
],
|
| 17 |
"max_steps": 73718,
|
| 18 |
"num_train_epochs": 1,
|
| 19 |
+
"total_flos": 524503851466752.0,
|
| 20 |
"trial_name": null,
|
| 21 |
"trial_params": null
|
| 22 |
}
|
last-checkpoint/training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a11b77e6168fee66eb97009649c3d2e4ac6aa3c0e8ee476e780119c106f76ff
|
| 3 |
+
size 3439
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 551185105
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e323cff0bca0bc2e6934298b8e92f0bc0a28c258605b1e175656618d1fc524e7
|
| 3 |
size 551185105
|
runs/Oct14_12-15-12_e3717350dd21/1665750223.3983731/events.out.tfevents.1665750223.e3717350dd21.24.1
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3cb477ab320609f8bc4e4ac843afbdb02b46c4919b75700833ffcc4e1e56960
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750242.5320206/events.out.tfevents.1665750242.e3717350dd21.24.2
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd54cac92d8b434be2616f7eee51872673fd2e16ef12ae491ab20a3d8bb75c82
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750248.424454/events.out.tfevents.1665750248.e3717350dd21.24.3
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8a951dbb47a732c4482b5c862a2dcd18e072a6f2e9a9ba9e8f724f175337e226
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750250.8963432/events.out.tfevents.1665750250.e3717350dd21.24.4
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c6f7f7702e916b16cbbc20b850121fe3b77e127480ffc978b681e1bcb5984555
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750252.3554833/events.out.tfevents.1665750252.e3717350dd21.24.5
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:13dd5c423316642a19873a572fcac02353304060852535ac05840393bca7360a
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750253.167781/events.out.tfevents.1665750253.e3717350dd21.24.6
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0f899dd3449d27025118534979780e2f8f7b985d7cc3d73c5571f8a5104c00f5
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750253.8073034/events.out.tfevents.1665750253.e3717350dd21.24.7
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9f78ead40efebaeaf09516eec21c25213b6e5b0ddaba5c5f77be75b9cb802c72
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750254.2377386/events.out.tfevents.1665750254.e3717350dd21.24.8
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5c03a49f66bfa3dffba618aa236dca8e5b6aa43b29839c4b8af1729ea6fbe757
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750255.023941/events.out.tfevents.1665750255.e3717350dd21.24.9
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1424ad06b961e2a9788826d3d3eddd6e2a1d7f5d919d7aa368c0b8129fc3de82
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750255.9294362/events.out.tfevents.1665750255.e3717350dd21.24.10
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:18db0fda493f323f5fbbbc9a5590edd1ed6137f0618aa7c79b3c31f6820dc7ca
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/1665750257.3678532/events.out.tfevents.1665750257.e3717350dd21.24.11
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6e974e90e7d0f7a4c95b2d66e462575e35b927b777ac058d32f9dd80ac31f039
|
| 3 |
+
size 5438
|
runs/Oct14_12-15-12_e3717350dd21/events.out.tfevents.1665750223.e3717350dd21.24.0
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:96bfa2ab4f10182289711b3450e9a404b1b5363a02426560db9241fa8dae9c54
|
| 3 |
+
size 45770
|
training_args.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
-
size
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9a11b77e6168fee66eb97009649c3d2e4ac6aa3c0e8ee476e780119c106f76ff
|
| 3 |
+
size 3439
|