Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +316 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,316 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 3.764705882352941,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 48,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.08,
|
| 13 |
+
"learning_rate": 9.989294616193018e-06,
|
| 14 |
+
"loss": 2.3554,
|
| 15 |
+
"step": 1
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.16,
|
| 19 |
+
"learning_rate": 9.957224306869053e-06,
|
| 20 |
+
"loss": 2.1848,
|
| 21 |
+
"step": 2
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.24,
|
| 25 |
+
"learning_rate": 9.903926402016153e-06,
|
| 26 |
+
"loss": 1.9353,
|
| 27 |
+
"step": 3
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.31,
|
| 31 |
+
"learning_rate": 9.829629131445342e-06,
|
| 32 |
+
"loss": 1.6338,
|
| 33 |
+
"step": 4
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.39,
|
| 37 |
+
"learning_rate": 9.73465064747553e-06,
|
| 38 |
+
"loss": 1.5467,
|
| 39 |
+
"step": 5
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.47,
|
| 43 |
+
"learning_rate": 9.619397662556434e-06,
|
| 44 |
+
"loss": 1.6413,
|
| 45 |
+
"step": 6
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.55,
|
| 49 |
+
"learning_rate": 9.484363707663443e-06,
|
| 50 |
+
"loss": 1.5138,
|
| 51 |
+
"step": 7
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.63,
|
| 55 |
+
"learning_rate": 9.330127018922195e-06,
|
| 56 |
+
"loss": 1.3815,
|
| 57 |
+
"step": 8
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.71,
|
| 61 |
+
"learning_rate": 9.157348061512728e-06,
|
| 62 |
+
"loss": 1.4179,
|
| 63 |
+
"step": 9
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.78,
|
| 67 |
+
"learning_rate": 8.966766701456177e-06,
|
| 68 |
+
"loss": 1.2832,
|
| 69 |
+
"step": 10
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.86,
|
| 73 |
+
"learning_rate": 8.759199037394888e-06,
|
| 74 |
+
"loss": 1.2613,
|
| 75 |
+
"step": 11
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.94,
|
| 79 |
+
"learning_rate": 8.535533905932739e-06,
|
| 80 |
+
"loss": 1.246,
|
| 81 |
+
"step": 12
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 1.02,
|
| 85 |
+
"learning_rate": 8.296729075500345e-06,
|
| 86 |
+
"loss": 1.3114,
|
| 87 |
+
"step": 13
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 1.1,
|
| 91 |
+
"learning_rate": 8.043807145043604e-06,
|
| 92 |
+
"loss": 1.1663,
|
| 93 |
+
"step": 14
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 1.18,
|
| 97 |
+
"learning_rate": 7.777851165098012e-06,
|
| 98 |
+
"loss": 1.1314,
|
| 99 |
+
"step": 15
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 1.25,
|
| 103 |
+
"learning_rate": 7.500000000000001e-06,
|
| 104 |
+
"loss": 1.0967,
|
| 105 |
+
"step": 16
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 1.33,
|
| 109 |
+
"learning_rate": 7.211443451095007e-06,
|
| 110 |
+
"loss": 1.1373,
|
| 111 |
+
"step": 17
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 1.41,
|
| 115 |
+
"learning_rate": 6.913417161825449e-06,
|
| 116 |
+
"loss": 1.0439,
|
| 117 |
+
"step": 18
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 1.49,
|
| 121 |
+
"learning_rate": 6.607197326515808e-06,
|
| 122 |
+
"loss": 1.0292,
|
| 123 |
+
"step": 19
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 1.57,
|
| 127 |
+
"learning_rate": 6.294095225512604e-06,
|
| 128 |
+
"loss": 1.0374,
|
| 129 |
+
"step": 20
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 1.65,
|
| 133 |
+
"learning_rate": 5.975451610080643e-06,
|
| 134 |
+
"loss": 1.0239,
|
| 135 |
+
"step": 21
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 1.73,
|
| 139 |
+
"learning_rate": 5.65263096110026e-06,
|
| 140 |
+
"loss": 0.9936,
|
| 141 |
+
"step": 22
|
| 142 |
+
},
|
| 143 |
+
{
|
| 144 |
+
"epoch": 1.8,
|
| 145 |
+
"learning_rate": 5.327015646150716e-06,
|
| 146 |
+
"loss": 1.0088,
|
| 147 |
+
"step": 23
|
| 148 |
+
},
|
| 149 |
+
{
|
| 150 |
+
"epoch": 1.88,
|
| 151 |
+
"learning_rate": 5e-06,
|
| 152 |
+
"loss": 1.0079,
|
| 153 |
+
"step": 24
|
| 154 |
+
},
|
| 155 |
+
{
|
| 156 |
+
"epoch": 1.96,
|
| 157 |
+
"learning_rate": 4.672984353849285e-06,
|
| 158 |
+
"loss": 1.0041,
|
| 159 |
+
"step": 25
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 2.04,
|
| 163 |
+
"learning_rate": 4.347369038899744e-06,
|
| 164 |
+
"loss": 0.9283,
|
| 165 |
+
"step": 26
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 2.12,
|
| 169 |
+
"learning_rate": 4.02454838991936e-06,
|
| 170 |
+
"loss": 0.9017,
|
| 171 |
+
"step": 27
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 2.2,
|
| 175 |
+
"learning_rate": 3.705904774487396e-06,
|
| 176 |
+
"loss": 0.9113,
|
| 177 |
+
"step": 28
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 2.27,
|
| 181 |
+
"learning_rate": 3.3928026734841935e-06,
|
| 182 |
+
"loss": 0.8751,
|
| 183 |
+
"step": 29
|
| 184 |
+
},
|
| 185 |
+
{
|
| 186 |
+
"epoch": 2.35,
|
| 187 |
+
"learning_rate": 3.0865828381745515e-06,
|
| 188 |
+
"loss": 0.8379,
|
| 189 |
+
"step": 30
|
| 190 |
+
},
|
| 191 |
+
{
|
| 192 |
+
"epoch": 2.43,
|
| 193 |
+
"learning_rate": 2.7885565489049948e-06,
|
| 194 |
+
"loss": 0.8646,
|
| 195 |
+
"step": 31
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 2.51,
|
| 199 |
+
"learning_rate": 2.5000000000000015e-06,
|
| 200 |
+
"loss": 0.8262,
|
| 201 |
+
"step": 32
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 2.59,
|
| 205 |
+
"learning_rate": 2.2221488349019903e-06,
|
| 206 |
+
"loss": 0.8025,
|
| 207 |
+
"step": 33
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 2.67,
|
| 211 |
+
"learning_rate": 1.956192854956397e-06,
|
| 212 |
+
"loss": 0.8409,
|
| 213 |
+
"step": 34
|
| 214 |
+
},
|
| 215 |
+
{
|
| 216 |
+
"epoch": 2.75,
|
| 217 |
+
"learning_rate": 1.7032709244996559e-06,
|
| 218 |
+
"loss": 0.8191,
|
| 219 |
+
"step": 35
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 2.82,
|
| 223 |
+
"learning_rate": 1.4644660940672628e-06,
|
| 224 |
+
"loss": 0.8319,
|
| 225 |
+
"step": 36
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 2.9,
|
| 229 |
+
"learning_rate": 1.2408009626051137e-06,
|
| 230 |
+
"loss": 0.785,
|
| 231 |
+
"step": 37
|
| 232 |
+
},
|
| 233 |
+
{
|
| 234 |
+
"epoch": 2.98,
|
| 235 |
+
"learning_rate": 1.0332332985438248e-06,
|
| 236 |
+
"loss": 0.8475,
|
| 237 |
+
"step": 38
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 3.06,
|
| 241 |
+
"learning_rate": 8.426519384872733e-07,
|
| 242 |
+
"loss": 0.7509,
|
| 243 |
+
"step": 39
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 3.14,
|
| 247 |
+
"learning_rate": 6.698729810778065e-07,
|
| 248 |
+
"loss": 0.7825,
|
| 249 |
+
"step": 40
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 3.22,
|
| 253 |
+
"learning_rate": 5.156362923365587e-07,
|
| 254 |
+
"loss": 0.8104,
|
| 255 |
+
"step": 41
|
| 256 |
+
},
|
| 257 |
+
{
|
| 258 |
+
"epoch": 3.29,
|
| 259 |
+
"learning_rate": 3.8060233744356634e-07,
|
| 260 |
+
"loss": 0.7583,
|
| 261 |
+
"step": 42
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 3.37,
|
| 265 |
+
"learning_rate": 2.653493525244721e-07,
|
| 266 |
+
"loss": 0.7456,
|
| 267 |
+
"step": 43
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 3.45,
|
| 271 |
+
"learning_rate": 1.7037086855465902e-07,
|
| 272 |
+
"loss": 0.7221,
|
| 273 |
+
"step": 44
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 3.53,
|
| 277 |
+
"learning_rate": 9.607359798384785e-08,
|
| 278 |
+
"loss": 0.7034,
|
| 279 |
+
"step": 45
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 3.61,
|
| 283 |
+
"learning_rate": 4.2775693130948094e-08,
|
| 284 |
+
"loss": 0.7386,
|
| 285 |
+
"step": 46
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 3.69,
|
| 289 |
+
"learning_rate": 1.0705383806982606e-08,
|
| 290 |
+
"loss": 0.7359,
|
| 291 |
+
"step": 47
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 3.76,
|
| 295 |
+
"learning_rate": 0.0,
|
| 296 |
+
"loss": 0.7549,
|
| 297 |
+
"step": 48
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 3.76,
|
| 301 |
+
"step": 48,
|
| 302 |
+
"total_flos": 495910423363584.0,
|
| 303 |
+
"train_loss": 1.082658023883899,
|
| 304 |
+
"train_runtime": 22627.0931,
|
| 305 |
+
"train_samples_per_second": 4.596,
|
| 306 |
+
"train_steps_per_second": 0.002
|
| 307 |
+
}
|
| 308 |
+
],
|
| 309 |
+
"logging_steps": 1.0,
|
| 310 |
+
"max_steps": 48,
|
| 311 |
+
"num_train_epochs": 4,
|
| 312 |
+
"save_steps": 500,
|
| 313 |
+
"total_flos": 495910423363584.0,
|
| 314 |
+
"trial_name": null,
|
| 315 |
+
"trial_params": null
|
| 316 |
+
}
|