Commit
·
17c8a36
1
Parent(s):
550f712
(1) classification crawler_total 70e 0.5
Browse files- config.json +26 -1
- optimizer.pt +2 -2
- pytorch_model.bin +2 -2
- rng_state.pth +1 -1
- scaler.pt +3 -0
- scheduler.pt +1 -1
- trainer_state.json +718 -3115
- training_args.bin +2 -2
config.json
CHANGED
|
@@ -2,18 +2,43 @@
|
|
| 2 |
"_name_or_path": "distilbert-base-uncased",
|
| 3 |
"activation": "gelu",
|
| 4 |
"architectures": [
|
| 5 |
-
"
|
| 6 |
],
|
| 7 |
"attention_dropout": 0.1,
|
| 8 |
"dim": 768,
|
| 9 |
"dropout": 0.1,
|
| 10 |
"hidden_dim": 3072,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 11 |
"initializer_range": 0.02,
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 12 |
"max_position_embeddings": 512,
|
| 13 |
"model_type": "distilbert",
|
| 14 |
"n_heads": 12,
|
| 15 |
"n_layers": 6,
|
| 16 |
"pad_token_id": 0,
|
|
|
|
| 17 |
"qa_dropout": 0.1,
|
| 18 |
"seq_classif_dropout": 0.2,
|
| 19 |
"sinusoidal_pos_embds": false,
|
|
|
|
| 2 |
"_name_or_path": "distilbert-base-uncased",
|
| 3 |
"activation": "gelu",
|
| 4 |
"architectures": [
|
| 5 |
+
"DistilBertForSequenceClassification"
|
| 6 |
],
|
| 7 |
"attention_dropout": 0.1,
|
| 8 |
"dim": 768,
|
| 9 |
"dropout": 0.1,
|
| 10 |
"hidden_dim": 3072,
|
| 11 |
+
"id2label": {
|
| 12 |
+
"0": "LABEL_0",
|
| 13 |
+
"1": "LABEL_1",
|
| 14 |
+
"2": "LABEL_2",
|
| 15 |
+
"3": "LABEL_3",
|
| 16 |
+
"4": "LABEL_4",
|
| 17 |
+
"5": "LABEL_5",
|
| 18 |
+
"6": "LABEL_6",
|
| 19 |
+
"7": "LABEL_7",
|
| 20 |
+
"8": "LABEL_8",
|
| 21 |
+
"9": "LABEL_9"
|
| 22 |
+
},
|
| 23 |
"initializer_range": 0.02,
|
| 24 |
+
"label2id": {
|
| 25 |
+
"LABEL_0": 0,
|
| 26 |
+
"LABEL_1": 1,
|
| 27 |
+
"LABEL_2": 2,
|
| 28 |
+
"LABEL_3": 3,
|
| 29 |
+
"LABEL_4": 4,
|
| 30 |
+
"LABEL_5": 5,
|
| 31 |
+
"LABEL_6": 6,
|
| 32 |
+
"LABEL_7": 7,
|
| 33 |
+
"LABEL_8": 8,
|
| 34 |
+
"LABEL_9": 9
|
| 35 |
+
},
|
| 36 |
"max_position_embeddings": 512,
|
| 37 |
"model_type": "distilbert",
|
| 38 |
"n_heads": 12,
|
| 39 |
"n_layers": 6,
|
| 40 |
"pad_token_id": 0,
|
| 41 |
+
"problem_type": "single_label_classification",
|
| 42 |
"qa_dropout": 0.1,
|
| 43 |
"seq_classif_dropout": 0.2,
|
| 44 |
"sinusoidal_pos_embds": false,
|
optimizer.pt
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:f6fa49861293b096f4ec4a7287f7759ac99a2062f0f275e63f7280fc15e0f9f6
|
| 3 |
+
size 535750213
|
pytorch_model.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 sha256:491feb260883d9c03042e7f92de7b291e3680759cf403faee560c936244e4b35
|
| 3 |
+
size 267880109
|
rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14575
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3899805085e9d2ac6cbf41d1e231c8019961099f94de919796ff76ec1a2a6d27
|
| 3 |
size 14575
|
scaler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f164eca888cfa14782f70fc9a282dd4e6c1593faf867ae17dfa46d264aa2942
|
| 3 |
+
size 557
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 627
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:772dd81534f16ccad499eda8bef2dc60508234c8daa4ecae9a5aa2c25f5ac971
|
| 3 |
size 627
|
trainer_state.json
CHANGED
|
@@ -1,3388 +1,991 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 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.
|
| 12 |
-
"learning_rate": 4.
|
| 13 |
-
"loss": 1.
|
| 14 |
-
"step":
|
| 15 |
-
},
|
| 16 |
-
{
|
| 17 |
-
"epoch": 0.01,
|
| 18 |
-
"learning_rate": 4.9964458850457065e-05,
|
| 19 |
-
"loss": 1.6614,
|
| 20 |
-
"step": 1000
|
| 21 |
-
},
|
| 22 |
-
{
|
| 23 |
-
"epoch": 0.02,
|
| 24 |
-
"learning_rate": 4.99466882756856e-05,
|
| 25 |
-
"loss": 1.6338,
|
| 26 |
-
"step": 1500
|
| 27 |
-
},
|
| 28 |
-
{
|
| 29 |
-
"epoch": 0.03,
|
| 30 |
-
"learning_rate": 4.992891770091412e-05,
|
| 31 |
-
"loss": 1.6058,
|
| 32 |
-
"step": 2000
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.04,
|
| 36 |
-
"learning_rate": 4.9911147126142646e-05,
|
| 37 |
-
"loss": 1.5887,
|
| 38 |
-
"step": 2500
|
| 39 |
-
},
|
| 40 |
-
{
|
| 41 |
-
"epoch": 0.04,
|
| 42 |
-
"learning_rate": 4.989337655137118e-05,
|
| 43 |
-
"loss": 1.5595,
|
| 44 |
-
"step": 3000
|
| 45 |
-
},
|
| 46 |
-
{
|
| 47 |
-
"epoch": 0.05,
|
| 48 |
-
"learning_rate": 4.987560597659971e-05,
|
| 49 |
-
"loss": 1.5627,
|
| 50 |
-
"step": 3500
|
| 51 |
-
},
|
| 52 |
-
{
|
| 53 |
-
"epoch": 0.06,
|
| 54 |
-
"learning_rate": 4.985783540182824e-05,
|
| 55 |
-
"loss": 1.5473,
|
| 56 |
-
"step": 4000
|
| 57 |
-
},
|
| 58 |
-
{
|
| 59 |
-
"epoch": 0.06,
|
| 60 |
-
"learning_rate": 4.984006482705677e-05,
|
| 61 |
-
"loss": 1.5343,
|
| 62 |
-
"step": 4500
|
| 63 |
-
},
|
| 64 |
-
{
|
| 65 |
-
"epoch": 0.07,
|
| 66 |
-
"learning_rate": 4.98222942522853e-05,
|
| 67 |
-
"loss": 1.5355,
|
| 68 |
-
"step": 5000
|
| 69 |
-
},
|
| 70 |
-
{
|
| 71 |
-
"epoch": 0.08,
|
| 72 |
-
"learning_rate": 4.980452367751383e-05,
|
| 73 |
-
"loss": 1.5204,
|
| 74 |
-
"step": 5500
|
| 75 |
-
},
|
| 76 |
-
{
|
| 77 |
-
"epoch": 0.09,
|
| 78 |
-
"learning_rate": 4.978675310274236e-05,
|
| 79 |
-
"loss": 1.5188,
|
| 80 |
-
"step": 6000
|
| 81 |
-
},
|
| 82 |
-
{
|
| 83 |
-
"epoch": 0.09,
|
| 84 |
-
"learning_rate": 4.9768982527970884e-05,
|
| 85 |
-
"loss": 1.5112,
|
| 86 |
-
"step": 6500
|
| 87 |
-
},
|
| 88 |
-
{
|
| 89 |
-
"epoch": 0.1,
|
| 90 |
-
"learning_rate": 4.9751211953199416e-05,
|
| 91 |
-
"loss": 1.5038,
|
| 92 |
-
"step": 7000
|
| 93 |
-
},
|
| 94 |
-
{
|
| 95 |
-
"epoch": 0.11,
|
| 96 |
-
"learning_rate": 4.973344137842795e-05,
|
| 97 |
-
"loss": 1.5054,
|
| 98 |
-
"step": 7500
|
| 99 |
-
},
|
| 100 |
-
{
|
| 101 |
-
"epoch": 0.11,
|
| 102 |
-
"learning_rate": 4.971567080365648e-05,
|
| 103 |
-
"loss": 1.4997,
|
| 104 |
-
"step": 8000
|
| 105 |
-
},
|
| 106 |
-
{
|
| 107 |
-
"epoch": 0.12,
|
| 108 |
-
"learning_rate": 4.969790022888501e-05,
|
| 109 |
-
"loss": 1.4958,
|
| 110 |
-
"step": 8500
|
| 111 |
-
},
|
| 112 |
-
{
|
| 113 |
-
"epoch": 0.13,
|
| 114 |
-
"learning_rate": 4.9680129654113535e-05,
|
| 115 |
-
"loss": 1.496,
|
| 116 |
-
"step": 9000
|
| 117 |
-
},
|
| 118 |
-
{
|
| 119 |
-
"epoch": 0.14,
|
| 120 |
-
"learning_rate": 4.9662359079342066e-05,
|
| 121 |
-
"loss": 1.477,
|
| 122 |
-
"step": 9500
|
| 123 |
-
},
|
| 124 |
-
{
|
| 125 |
-
"epoch": 0.14,
|
| 126 |
-
"learning_rate": 4.964458850457059e-05,
|
| 127 |
-
"loss": 1.4939,
|
| 128 |
-
"step": 10000
|
| 129 |
-
},
|
| 130 |
-
{
|
| 131 |
-
"epoch": 0.15,
|
| 132 |
-
"learning_rate": 4.962681792979912e-05,
|
| 133 |
-
"loss": 1.4796,
|
| 134 |
-
"step": 10500
|
| 135 |
-
},
|
| 136 |
-
{
|
| 137 |
-
"epoch": 0.16,
|
| 138 |
-
"learning_rate": 4.9609047355027654e-05,
|
| 139 |
-
"loss": 1.4795,
|
| 140 |
-
"step": 11000
|
| 141 |
-
},
|
| 142 |
-
{
|
| 143 |
-
"epoch": 0.16,
|
| 144 |
-
"learning_rate": 4.9591276780256185e-05,
|
| 145 |
-
"loss": 1.4754,
|
| 146 |
-
"step": 11500
|
| 147 |
-
},
|
| 148 |
-
{
|
| 149 |
-
"epoch": 0.17,
|
| 150 |
-
"learning_rate": 4.957350620548471e-05,
|
| 151 |
-
"loss": 1.4789,
|
| 152 |
-
"step": 12000
|
| 153 |
-
},
|
| 154 |
-
{
|
| 155 |
-
"epoch": 0.18,
|
| 156 |
-
"learning_rate": 4.955573563071324e-05,
|
| 157 |
-
"loss": 1.4704,
|
| 158 |
-
"step": 12500
|
| 159 |
-
},
|
| 160 |
-
{
|
| 161 |
-
"epoch": 0.18,
|
| 162 |
-
"learning_rate": 4.953796505594177e-05,
|
| 163 |
-
"loss": 1.47,
|
| 164 |
-
"step": 13000
|
| 165 |
-
},
|
| 166 |
-
{
|
| 167 |
-
"epoch": 0.19,
|
| 168 |
-
"learning_rate": 4.9520194481170304e-05,
|
| 169 |
-
"loss": 1.4679,
|
| 170 |
-
"step": 13500
|
| 171 |
-
},
|
| 172 |
-
{
|
| 173 |
-
"epoch": 0.2,
|
| 174 |
-
"learning_rate": 4.950242390639883e-05,
|
| 175 |
-
"loss": 1.4573,
|
| 176 |
-
"step": 14000
|
| 177 |
-
},
|
| 178 |
-
{
|
| 179 |
-
"epoch": 0.21,
|
| 180 |
-
"learning_rate": 4.948465333162736e-05,
|
| 181 |
-
"loss": 1.4639,
|
| 182 |
-
"step": 14500
|
| 183 |
-
},
|
| 184 |
-
{
|
| 185 |
-
"epoch": 0.21,
|
| 186 |
-
"learning_rate": 4.946688275685589e-05,
|
| 187 |
-
"loss": 1.4587,
|
| 188 |
-
"step": 15000
|
| 189 |
-
},
|
| 190 |
-
{
|
| 191 |
-
"epoch": 0.22,
|
| 192 |
-
"learning_rate": 4.9449112182084416e-05,
|
| 193 |
-
"loss": 1.4619,
|
| 194 |
-
"step": 15500
|
| 195 |
-
},
|
| 196 |
-
{
|
| 197 |
-
"epoch": 0.23,
|
| 198 |
-
"learning_rate": 4.943134160731295e-05,
|
| 199 |
-
"loss": 1.4531,
|
| 200 |
-
"step": 16000
|
| 201 |
-
},
|
| 202 |
-
{
|
| 203 |
-
"epoch": 0.23,
|
| 204 |
-
"learning_rate": 4.941357103254148e-05,
|
| 205 |
-
"loss": 1.4501,
|
| 206 |
-
"step": 16500
|
| 207 |
-
},
|
| 208 |
-
{
|
| 209 |
-
"epoch": 0.24,
|
| 210 |
-
"learning_rate": 4.939580045777001e-05,
|
| 211 |
-
"loss": 1.4524,
|
| 212 |
-
"step": 17000
|
| 213 |
-
},
|
| 214 |
-
{
|
| 215 |
-
"epoch": 0.25,
|
| 216 |
-
"learning_rate": 4.937802988299854e-05,
|
| 217 |
-
"loss": 1.4455,
|
| 218 |
-
"step": 17500
|
| 219 |
-
},
|
| 220 |
-
{
|
| 221 |
-
"epoch": 0.26,
|
| 222 |
-
"learning_rate": 4.936025930822707e-05,
|
| 223 |
-
"loss": 1.4502,
|
| 224 |
-
"step": 18000
|
| 225 |
-
},
|
| 226 |
-
{
|
| 227 |
-
"epoch": 0.26,
|
| 228 |
-
"learning_rate": 4.93424887334556e-05,
|
| 229 |
-
"loss": 1.439,
|
| 230 |
-
"step": 18500
|
| 231 |
-
},
|
| 232 |
-
{
|
| 233 |
-
"epoch": 0.27,
|
| 234 |
-
"learning_rate": 4.932471815868412e-05,
|
| 235 |
-
"loss": 1.4436,
|
| 236 |
-
"step": 19000
|
| 237 |
-
},
|
| 238 |
-
{
|
| 239 |
-
"epoch": 0.28,
|
| 240 |
-
"learning_rate": 4.9306947583912654e-05,
|
| 241 |
-
"loss": 1.4311,
|
| 242 |
-
"step": 19500
|
| 243 |
-
},
|
| 244 |
-
{
|
| 245 |
-
"epoch": 0.28,
|
| 246 |
-
"learning_rate": 4.9289177009141186e-05,
|
| 247 |
-
"loss": 1.4368,
|
| 248 |
-
"step": 20000
|
| 249 |
-
},
|
| 250 |
-
{
|
| 251 |
-
"epoch": 0.29,
|
| 252 |
-
"learning_rate": 4.927140643436972e-05,
|
| 253 |
-
"loss": 1.4368,
|
| 254 |
-
"step": 20500
|
| 255 |
-
},
|
| 256 |
-
{
|
| 257 |
-
"epoch": 0.3,
|
| 258 |
-
"learning_rate": 4.925363585959825e-05,
|
| 259 |
-
"loss": 1.4312,
|
| 260 |
-
"step": 21000
|
| 261 |
-
},
|
| 262 |
-
{
|
| 263 |
-
"epoch": 0.31,
|
| 264 |
-
"learning_rate": 4.923586528482677e-05,
|
| 265 |
-
"loss": 1.4318,
|
| 266 |
-
"step": 21500
|
| 267 |
-
},
|
| 268 |
-
{
|
| 269 |
-
"epoch": 0.31,
|
| 270 |
-
"learning_rate": 4.9218094710055305e-05,
|
| 271 |
-
"loss": 1.4297,
|
| 272 |
-
"step": 22000
|
| 273 |
-
},
|
| 274 |
-
{
|
| 275 |
-
"epoch": 0.32,
|
| 276 |
-
"learning_rate": 4.920032413528383e-05,
|
| 277 |
-
"loss": 1.4248,
|
| 278 |
-
"step": 22500
|
| 279 |
-
},
|
| 280 |
-
{
|
| 281 |
-
"epoch": 0.33,
|
| 282 |
-
"learning_rate": 4.918255356051236e-05,
|
| 283 |
-
"loss": 1.4221,
|
| 284 |
-
"step": 23000
|
| 285 |
-
},
|
| 286 |
-
{
|
| 287 |
-
"epoch": 0.33,
|
| 288 |
-
"learning_rate": 4.916478298574089e-05,
|
| 289 |
-
"loss": 1.4265,
|
| 290 |
-
"step": 23500
|
| 291 |
-
},
|
| 292 |
-
{
|
| 293 |
-
"epoch": 0.34,
|
| 294 |
-
"learning_rate": 4.9147012410969424e-05,
|
| 295 |
-
"loss": 1.4034,
|
| 296 |
-
"step": 24000
|
| 297 |
-
},
|
| 298 |
-
{
|
| 299 |
-
"epoch": 0.35,
|
| 300 |
-
"learning_rate": 4.9129241836197955e-05,
|
| 301 |
-
"loss": 1.4262,
|
| 302 |
-
"step": 24500
|
| 303 |
-
},
|
| 304 |
-
{
|
| 305 |
-
"epoch": 0.36,
|
| 306 |
-
"learning_rate": 4.911147126142649e-05,
|
| 307 |
-
"loss": 1.4214,
|
| 308 |
-
"step": 25000
|
| 309 |
-
},
|
| 310 |
-
{
|
| 311 |
-
"epoch": 0.36,
|
| 312 |
-
"learning_rate": 4.909370068665501e-05,
|
| 313 |
-
"loss": 1.4041,
|
| 314 |
-
"step": 25500
|
| 315 |
-
},
|
| 316 |
-
{
|
| 317 |
-
"epoch": 0.37,
|
| 318 |
-
"learning_rate": 4.9075930111883536e-05,
|
| 319 |
-
"loss": 1.4201,
|
| 320 |
-
"step": 26000
|
| 321 |
-
},
|
| 322 |
-
{
|
| 323 |
-
"epoch": 0.38,
|
| 324 |
-
"learning_rate": 4.905815953711207e-05,
|
| 325 |
-
"loss": 1.4174,
|
| 326 |
-
"step": 26500
|
| 327 |
-
},
|
| 328 |
-
{
|
| 329 |
-
"epoch": 0.38,
|
| 330 |
-
"learning_rate": 4.90403889623406e-05,
|
| 331 |
-
"loss": 1.4066,
|
| 332 |
-
"step": 27000
|
| 333 |
-
},
|
| 334 |
-
{
|
| 335 |
-
"epoch": 0.39,
|
| 336 |
-
"learning_rate": 4.902261838756913e-05,
|
| 337 |
-
"loss": 1.4199,
|
| 338 |
-
"step": 27500
|
| 339 |
-
},
|
| 340 |
-
{
|
| 341 |
-
"epoch": 0.4,
|
| 342 |
-
"learning_rate": 4.900484781279766e-05,
|
| 343 |
-
"loss": 1.4143,
|
| 344 |
-
"step": 28000
|
| 345 |
-
},
|
| 346 |
-
{
|
| 347 |
-
"epoch": 0.41,
|
| 348 |
-
"learning_rate": 4.898707723802619e-05,
|
| 349 |
-
"loss": 1.3948,
|
| 350 |
-
"step": 28500
|
| 351 |
-
},
|
| 352 |
-
{
|
| 353 |
-
"epoch": 0.41,
|
| 354 |
-
"learning_rate": 4.896930666325472e-05,
|
| 355 |
-
"loss": 1.4169,
|
| 356 |
-
"step": 29000
|
| 357 |
-
},
|
| 358 |
-
{
|
| 359 |
-
"epoch": 0.42,
|
| 360 |
-
"learning_rate": 4.895153608848325e-05,
|
| 361 |
-
"loss": 1.4115,
|
| 362 |
-
"step": 29500
|
| 363 |
-
},
|
| 364 |
-
{
|
| 365 |
-
"epoch": 0.43,
|
| 366 |
-
"learning_rate": 4.8933765513711774e-05,
|
| 367 |
-
"loss": 1.3961,
|
| 368 |
-
"step": 30000
|
| 369 |
-
},
|
| 370 |
-
{
|
| 371 |
-
"epoch": 0.43,
|
| 372 |
-
"learning_rate": 4.8915994938940305e-05,
|
| 373 |
-
"loss": 1.4105,
|
| 374 |
-
"step": 30500
|
| 375 |
-
},
|
| 376 |
-
{
|
| 377 |
-
"epoch": 0.44,
|
| 378 |
-
"learning_rate": 4.889822436416884e-05,
|
| 379 |
-
"loss": 1.4136,
|
| 380 |
-
"step": 31000
|
| 381 |
-
},
|
| 382 |
-
{
|
| 383 |
-
"epoch": 0.45,
|
| 384 |
-
"learning_rate": 4.888045378939737e-05,
|
| 385 |
-
"loss": 1.3963,
|
| 386 |
-
"step": 31500
|
| 387 |
-
},
|
| 388 |
-
{
|
| 389 |
-
"epoch": 0.45,
|
| 390 |
-
"learning_rate": 4.88626832146259e-05,
|
| 391 |
-
"loss": 1.3984,
|
| 392 |
-
"step": 32000
|
| 393 |
-
},
|
| 394 |
-
{
|
| 395 |
-
"epoch": 0.46,
|
| 396 |
-
"learning_rate": 4.8844912639854424e-05,
|
| 397 |
-
"loss": 1.4033,
|
| 398 |
-
"step": 32500
|
| 399 |
-
},
|
| 400 |
-
{
|
| 401 |
-
"epoch": 0.47,
|
| 402 |
-
"learning_rate": 4.8827142065082956e-05,
|
| 403 |
-
"loss": 1.4039,
|
| 404 |
-
"step": 33000
|
| 405 |
-
},
|
| 406 |
-
{
|
| 407 |
-
"epoch": 0.48,
|
| 408 |
-
"learning_rate": 4.880937149031149e-05,
|
| 409 |
-
"loss": 1.399,
|
| 410 |
-
"step": 33500
|
| 411 |
-
},
|
| 412 |
-
{
|
| 413 |
-
"epoch": 0.48,
|
| 414 |
-
"learning_rate": 4.879160091554001e-05,
|
| 415 |
-
"loss": 1.4116,
|
| 416 |
-
"step": 34000
|
| 417 |
-
},
|
| 418 |
-
{
|
| 419 |
-
"epoch": 0.49,
|
| 420 |
-
"learning_rate": 4.8773830340768543e-05,
|
| 421 |
-
"loss": 1.3874,
|
| 422 |
-
"step": 34500
|
| 423 |
-
},
|
| 424 |
-
{
|
| 425 |
-
"epoch": 0.5,
|
| 426 |
-
"learning_rate": 4.8756059765997075e-05,
|
| 427 |
-
"loss": 1.3959,
|
| 428 |
-
"step": 35000
|
| 429 |
-
},
|
| 430 |
-
{
|
| 431 |
-
"epoch": 0.5,
|
| 432 |
-
"learning_rate": 4.8738289191225606e-05,
|
| 433 |
-
"loss": 1.3928,
|
| 434 |
-
"step": 35500
|
| 435 |
-
},
|
| 436 |
-
{
|
| 437 |
-
"epoch": 0.51,
|
| 438 |
-
"learning_rate": 4.872051861645413e-05,
|
| 439 |
-
"loss": 1.3811,
|
| 440 |
-
"step": 36000
|
| 441 |
-
},
|
| 442 |
-
{
|
| 443 |
-
"epoch": 0.52,
|
| 444 |
-
"learning_rate": 4.870274804168266e-05,
|
| 445 |
-
"loss": 1.3922,
|
| 446 |
-
"step": 36500
|
| 447 |
-
},
|
| 448 |
-
{
|
| 449 |
-
"epoch": 0.53,
|
| 450 |
-
"learning_rate": 4.8684977466911194e-05,
|
| 451 |
-
"loss": 1.3861,
|
| 452 |
-
"step": 37000
|
| 453 |
-
},
|
| 454 |
-
{
|
| 455 |
-
"epoch": 0.53,
|
| 456 |
-
"learning_rate": 4.866720689213972e-05,
|
| 457 |
-
"loss": 1.3875,
|
| 458 |
-
"step": 37500
|
| 459 |
-
},
|
| 460 |
-
{
|
| 461 |
-
"epoch": 0.54,
|
| 462 |
-
"learning_rate": 4.864943631736825e-05,
|
| 463 |
-
"loss": 1.3842,
|
| 464 |
-
"step": 38000
|
| 465 |
-
},
|
| 466 |
-
{
|
| 467 |
-
"epoch": 0.55,
|
| 468 |
-
"learning_rate": 4.863166574259678e-05,
|
| 469 |
-
"loss": 1.3796,
|
| 470 |
-
"step": 38500
|
| 471 |
-
},
|
| 472 |
-
{
|
| 473 |
-
"epoch": 0.55,
|
| 474 |
-
"learning_rate": 4.861389516782531e-05,
|
| 475 |
-
"loss": 1.3883,
|
| 476 |
-
"step": 39000
|
| 477 |
-
},
|
| 478 |
-
{
|
| 479 |
-
"epoch": 0.56,
|
| 480 |
-
"learning_rate": 4.859612459305384e-05,
|
| 481 |
-
"loss": 1.3722,
|
| 482 |
-
"step": 39500
|
| 483 |
-
},
|
| 484 |
-
{
|
| 485 |
-
"epoch": 0.57,
|
| 486 |
-
"learning_rate": 4.857835401828237e-05,
|
| 487 |
-
"loss": 1.3799,
|
| 488 |
-
"step": 40000
|
| 489 |
-
},
|
| 490 |
-
{
|
| 491 |
-
"epoch": 0.58,
|
| 492 |
-
"learning_rate": 4.85605834435109e-05,
|
| 493 |
-
"loss": 1.3999,
|
| 494 |
-
"step": 40500
|
| 495 |
-
},
|
| 496 |
-
{
|
| 497 |
-
"epoch": 0.58,
|
| 498 |
-
"learning_rate": 4.854281286873943e-05,
|
| 499 |
-
"loss": 1.3674,
|
| 500 |
-
"step": 41000
|
| 501 |
-
},
|
| 502 |
-
{
|
| 503 |
-
"epoch": 0.59,
|
| 504 |
-
"learning_rate": 4.852504229396796e-05,
|
| 505 |
-
"loss": 1.3881,
|
| 506 |
-
"step": 41500
|
| 507 |
-
},
|
| 508 |
-
{
|
| 509 |
-
"epoch": 0.6,
|
| 510 |
-
"learning_rate": 4.850727171919649e-05,
|
| 511 |
-
"loss": 1.3812,
|
| 512 |
-
"step": 42000
|
| 513 |
-
},
|
| 514 |
-
{
|
| 515 |
-
"epoch": 0.6,
|
| 516 |
-
"learning_rate": 4.848950114442502e-05,
|
| 517 |
-
"loss": 1.3749,
|
| 518 |
-
"step": 42500
|
| 519 |
-
},
|
| 520 |
-
{
|
| 521 |
-
"epoch": 0.61,
|
| 522 |
-
"learning_rate": 4.8471730569653544e-05,
|
| 523 |
-
"loss": 1.3803,
|
| 524 |
-
"step": 43000
|
| 525 |
-
},
|
| 526 |
-
{
|
| 527 |
-
"epoch": 0.62,
|
| 528 |
-
"learning_rate": 4.8453959994882076e-05,
|
| 529 |
-
"loss": 1.3856,
|
| 530 |
-
"step": 43500
|
| 531 |
-
},
|
| 532 |
-
{
|
| 533 |
-
"epoch": 0.63,
|
| 534 |
-
"learning_rate": 4.843618942011061e-05,
|
| 535 |
-
"loss": 1.381,
|
| 536 |
-
"step": 44000
|
| 537 |
-
},
|
| 538 |
-
{
|
| 539 |
-
"epoch": 0.63,
|
| 540 |
-
"learning_rate": 4.841841884533914e-05,
|
| 541 |
-
"loss": 1.3817,
|
| 542 |
-
"step": 44500
|
| 543 |
-
},
|
| 544 |
-
{
|
| 545 |
-
"epoch": 0.64,
|
| 546 |
-
"learning_rate": 4.840064827056767e-05,
|
| 547 |
-
"loss": 1.3679,
|
| 548 |
-
"step": 45000
|
| 549 |
-
},
|
| 550 |
-
{
|
| 551 |
-
"epoch": 0.65,
|
| 552 |
-
"learning_rate": 4.8382877695796195e-05,
|
| 553 |
-
"loss": 1.3849,
|
| 554 |
-
"step": 45500
|
| 555 |
-
},
|
| 556 |
-
{
|
| 557 |
-
"epoch": 0.65,
|
| 558 |
-
"learning_rate": 4.836510712102472e-05,
|
| 559 |
-
"loss": 1.3758,
|
| 560 |
-
"step": 46000
|
| 561 |
-
},
|
| 562 |
-
{
|
| 563 |
-
"epoch": 0.66,
|
| 564 |
-
"learning_rate": 4.834733654625325e-05,
|
| 565 |
-
"loss": 1.3616,
|
| 566 |
-
"step": 46500
|
| 567 |
-
},
|
| 568 |
-
{
|
| 569 |
-
"epoch": 0.67,
|
| 570 |
-
"learning_rate": 4.832956597148178e-05,
|
| 571 |
-
"loss": 1.364,
|
| 572 |
-
"step": 47000
|
| 573 |
-
},
|
| 574 |
-
{
|
| 575 |
-
"epoch": 0.68,
|
| 576 |
-
"learning_rate": 4.8311795396710314e-05,
|
| 577 |
-
"loss": 1.3727,
|
| 578 |
-
"step": 47500
|
| 579 |
-
},
|
| 580 |
-
{
|
| 581 |
-
"epoch": 0.68,
|
| 582 |
-
"learning_rate": 4.8294024821938845e-05,
|
| 583 |
-
"loss": 1.3612,
|
| 584 |
-
"step": 48000
|
| 585 |
-
},
|
| 586 |
-
{
|
| 587 |
-
"epoch": 0.69,
|
| 588 |
-
"learning_rate": 4.8276254247167377e-05,
|
| 589 |
-
"loss": 1.3567,
|
| 590 |
-
"step": 48500
|
| 591 |
-
},
|
| 592 |
-
{
|
| 593 |
-
"epoch": 0.7,
|
| 594 |
-
"learning_rate": 4.82584836723959e-05,
|
| 595 |
-
"loss": 1.367,
|
| 596 |
-
"step": 49000
|
| 597 |
-
},
|
| 598 |
-
{
|
| 599 |
-
"epoch": 0.7,
|
| 600 |
-
"learning_rate": 4.8240713097624426e-05,
|
| 601 |
-
"loss": 1.3671,
|
| 602 |
-
"step": 49500
|
| 603 |
-
},
|
| 604 |
-
{
|
| 605 |
-
"epoch": 0.71,
|
| 606 |
-
"learning_rate": 4.822294252285296e-05,
|
| 607 |
-
"loss": 1.3736,
|
| 608 |
-
"step": 50000
|
| 609 |
-
},
|
| 610 |
-
{
|
| 611 |
-
"epoch": 0.72,
|
| 612 |
-
"learning_rate": 4.820517194808149e-05,
|
| 613 |
-
"loss": 1.3661,
|
| 614 |
-
"step": 50500
|
| 615 |
-
},
|
| 616 |
-
{
|
| 617 |
-
"epoch": 0.73,
|
| 618 |
-
"learning_rate": 4.818740137331002e-05,
|
| 619 |
-
"loss": 1.3604,
|
| 620 |
-
"step": 51000
|
| 621 |
-
},
|
| 622 |
-
{
|
| 623 |
-
"epoch": 0.73,
|
| 624 |
-
"learning_rate": 4.816963079853855e-05,
|
| 625 |
-
"loss": 1.3534,
|
| 626 |
-
"step": 51500
|
| 627 |
-
},
|
| 628 |
-
{
|
| 629 |
-
"epoch": 0.74,
|
| 630 |
-
"learning_rate": 4.815186022376708e-05,
|
| 631 |
-
"loss": 1.3636,
|
| 632 |
-
"step": 52000
|
| 633 |
-
},
|
| 634 |
-
{
|
| 635 |
-
"epoch": 0.75,
|
| 636 |
-
"learning_rate": 4.8134089648995615e-05,
|
| 637 |
-
"loss": 1.361,
|
| 638 |
-
"step": 52500
|
| 639 |
-
},
|
| 640 |
-
{
|
| 641 |
-
"epoch": 0.75,
|
| 642 |
-
"learning_rate": 4.811631907422414e-05,
|
| 643 |
-
"loss": 1.3541,
|
| 644 |
-
"step": 53000
|
| 645 |
-
},
|
| 646 |
-
{
|
| 647 |
-
"epoch": 0.76,
|
| 648 |
-
"learning_rate": 4.8098548499452664e-05,
|
| 649 |
-
"loss": 1.3642,
|
| 650 |
-
"step": 53500
|
| 651 |
-
},
|
| 652 |
-
{
|
| 653 |
-
"epoch": 0.77,
|
| 654 |
-
"learning_rate": 4.8080777924681195e-05,
|
| 655 |
-
"loss": 1.3548,
|
| 656 |
-
"step": 54000
|
| 657 |
-
},
|
| 658 |
-
{
|
| 659 |
-
"epoch": 0.77,
|
| 660 |
-
"learning_rate": 4.806300734990973e-05,
|
| 661 |
-
"loss": 1.36,
|
| 662 |
-
"step": 54500
|
| 663 |
-
},
|
| 664 |
-
{
|
| 665 |
-
"epoch": 0.78,
|
| 666 |
-
"learning_rate": 4.804523677513826e-05,
|
| 667 |
-
"loss": 1.3545,
|
| 668 |
-
"step": 55000
|
| 669 |
-
},
|
| 670 |
-
{
|
| 671 |
-
"epoch": 0.79,
|
| 672 |
-
"learning_rate": 4.802746620036679e-05,
|
| 673 |
-
"loss": 1.3543,
|
| 674 |
-
"step": 55500
|
| 675 |
-
},
|
| 676 |
-
{
|
| 677 |
-
"epoch": 0.8,
|
| 678 |
-
"learning_rate": 4.800969562559532e-05,
|
| 679 |
-
"loss": 1.3426,
|
| 680 |
-
"step": 56000
|
| 681 |
-
},
|
| 682 |
-
{
|
| 683 |
-
"epoch": 0.8,
|
| 684 |
-
"learning_rate": 4.7991925050823846e-05,
|
| 685 |
-
"loss": 1.3596,
|
| 686 |
-
"step": 56500
|
| 687 |
-
},
|
| 688 |
-
{
|
| 689 |
-
"epoch": 0.81,
|
| 690 |
-
"learning_rate": 4.797415447605238e-05,
|
| 691 |
-
"loss": 1.3629,
|
| 692 |
-
"step": 57000
|
| 693 |
-
},
|
| 694 |
-
{
|
| 695 |
-
"epoch": 0.82,
|
| 696 |
-
"learning_rate": 4.79563839012809e-05,
|
| 697 |
-
"loss": 1.3657,
|
| 698 |
-
"step": 57500
|
| 699 |
-
},
|
| 700 |
-
{
|
| 701 |
-
"epoch": 0.82,
|
| 702 |
-
"learning_rate": 4.7938613326509433e-05,
|
| 703 |
-
"loss": 1.3602,
|
| 704 |
-
"step": 58000
|
| 705 |
-
},
|
| 706 |
-
{
|
| 707 |
-
"epoch": 0.83,
|
| 708 |
-
"learning_rate": 4.7920842751737965e-05,
|
| 709 |
-
"loss": 1.3518,
|
| 710 |
-
"step": 58500
|
| 711 |
-
},
|
| 712 |
-
{
|
| 713 |
-
"epoch": 0.84,
|
| 714 |
-
"learning_rate": 4.7903072176966496e-05,
|
| 715 |
-
"loss": 1.3622,
|
| 716 |
-
"step": 59000
|
| 717 |
-
},
|
| 718 |
-
{
|
| 719 |
-
"epoch": 0.85,
|
| 720 |
-
"learning_rate": 4.788530160219503e-05,
|
| 721 |
-
"loss": 1.3472,
|
| 722 |
-
"step": 59500
|
| 723 |
-
},
|
| 724 |
-
{
|
| 725 |
-
"epoch": 0.85,
|
| 726 |
-
"learning_rate": 4.786753102742355e-05,
|
| 727 |
-
"loss": 1.3618,
|
| 728 |
-
"step": 60000
|
| 729 |
-
},
|
| 730 |
-
{
|
| 731 |
-
"epoch": 0.86,
|
| 732 |
-
"learning_rate": 4.7849760452652084e-05,
|
| 733 |
-
"loss": 1.3623,
|
| 734 |
-
"step": 60500
|
| 735 |
-
},
|
| 736 |
-
{
|
| 737 |
-
"epoch": 0.87,
|
| 738 |
-
"learning_rate": 4.783198987788061e-05,
|
| 739 |
-
"loss": 1.339,
|
| 740 |
-
"step": 61000
|
| 741 |
-
},
|
| 742 |
-
{
|
| 743 |
-
"epoch": 0.87,
|
| 744 |
-
"learning_rate": 4.781421930310914e-05,
|
| 745 |
-
"loss": 1.3478,
|
| 746 |
-
"step": 61500
|
| 747 |
-
},
|
| 748 |
-
{
|
| 749 |
-
"epoch": 0.88,
|
| 750 |
-
"learning_rate": 4.779644872833767e-05,
|
| 751 |
-
"loss": 1.354,
|
| 752 |
-
"step": 62000
|
| 753 |
-
},
|
| 754 |
-
{
|
| 755 |
-
"epoch": 0.89,
|
| 756 |
-
"learning_rate": 4.77786781535662e-05,
|
| 757 |
-
"loss": 1.3461,
|
| 758 |
-
"step": 62500
|
| 759 |
-
},
|
| 760 |
-
{
|
| 761 |
-
"epoch": 0.9,
|
| 762 |
-
"learning_rate": 4.7760907578794734e-05,
|
| 763 |
-
"loss": 1.3504,
|
| 764 |
-
"step": 63000
|
| 765 |
-
},
|
| 766 |
-
{
|
| 767 |
-
"epoch": 0.9,
|
| 768 |
-
"learning_rate": 4.774313700402326e-05,
|
| 769 |
-
"loss": 1.3519,
|
| 770 |
-
"step": 63500
|
| 771 |
-
},
|
| 772 |
-
{
|
| 773 |
-
"epoch": 0.91,
|
| 774 |
-
"learning_rate": 4.772536642925179e-05,
|
| 775 |
-
"loss": 1.3517,
|
| 776 |
-
"step": 64000
|
| 777 |
-
},
|
| 778 |
-
{
|
| 779 |
-
"epoch": 0.92,
|
| 780 |
-
"learning_rate": 4.770759585448032e-05,
|
| 781 |
-
"loss": 1.3475,
|
| 782 |
-
"step": 64500
|
| 783 |
-
},
|
| 784 |
-
{
|
| 785 |
-
"epoch": 0.92,
|
| 786 |
-
"learning_rate": 4.7689825279708847e-05,
|
| 787 |
-
"loss": 1.3307,
|
| 788 |
-
"step": 65000
|
| 789 |
-
},
|
| 790 |
-
{
|
| 791 |
-
"epoch": 0.93,
|
| 792 |
-
"learning_rate": 4.767205470493738e-05,
|
| 793 |
-
"loss": 1.3612,
|
| 794 |
-
"step": 65500
|
| 795 |
-
},
|
| 796 |
-
{
|
| 797 |
-
"epoch": 0.94,
|
| 798 |
-
"learning_rate": 4.765428413016591e-05,
|
| 799 |
-
"loss": 1.3482,
|
| 800 |
-
"step": 66000
|
| 801 |
-
},
|
| 802 |
-
{
|
| 803 |
-
"epoch": 0.95,
|
| 804 |
-
"learning_rate": 4.7636513555394434e-05,
|
| 805 |
-
"loss": 1.3432,
|
| 806 |
-
"step": 66500
|
| 807 |
-
},
|
| 808 |
-
{
|
| 809 |
-
"epoch": 0.95,
|
| 810 |
-
"learning_rate": 4.7618742980622966e-05,
|
| 811 |
-
"loss": 1.3466,
|
| 812 |
-
"step": 67000
|
| 813 |
-
},
|
| 814 |
-
{
|
| 815 |
-
"epoch": 0.96,
|
| 816 |
-
"learning_rate": 4.76009724058515e-05,
|
| 817 |
-
"loss": 1.3462,
|
| 818 |
-
"step": 67500
|
| 819 |
-
},
|
| 820 |
-
{
|
| 821 |
-
"epoch": 0.97,
|
| 822 |
-
"learning_rate": 4.758320183108003e-05,
|
| 823 |
-
"loss": 1.3408,
|
| 824 |
-
"step": 68000
|
| 825 |
-
},
|
| 826 |
-
{
|
| 827 |
-
"epoch": 0.97,
|
| 828 |
-
"learning_rate": 4.756543125630856e-05,
|
| 829 |
-
"loss": 1.3259,
|
| 830 |
-
"step": 68500
|
| 831 |
-
},
|
| 832 |
-
{
|
| 833 |
-
"epoch": 0.98,
|
| 834 |
-
"learning_rate": 4.7547660681537085e-05,
|
| 835 |
-
"loss": 1.3419,
|
| 836 |
-
"step": 69000
|
| 837 |
-
},
|
| 838 |
-
{
|
| 839 |
-
"epoch": 0.99,
|
| 840 |
-
"learning_rate": 4.7529890106765616e-05,
|
| 841 |
-
"loss": 1.3326,
|
| 842 |
-
"step": 69500
|
| 843 |
-
},
|
| 844 |
-
{
|
| 845 |
-
"epoch": 1.0,
|
| 846 |
-
"learning_rate": 4.751211953199414e-05,
|
| 847 |
-
"loss": 1.3475,
|
| 848 |
-
"step": 70000
|
| 849 |
-
},
|
| 850 |
-
{
|
| 851 |
-
"epoch": 1.0,
|
| 852 |
-
"learning_rate": 4.749434895722267e-05,
|
| 853 |
-
"loss": 1.3372,
|
| 854 |
-
"step": 70500
|
| 855 |
-
},
|
| 856 |
-
{
|
| 857 |
-
"epoch": 1.01,
|
| 858 |
-
"learning_rate": 4.7476578382451204e-05,
|
| 859 |
-
"loss": 1.2898,
|
| 860 |
-
"step": 71000
|
| 861 |
-
},
|
| 862 |
-
{
|
| 863 |
-
"epoch": 1.02,
|
| 864 |
-
"learning_rate": 4.7458807807679735e-05,
|
| 865 |
-
"loss": 1.2904,
|
| 866 |
-
"step": 71500
|
| 867 |
-
},
|
| 868 |
-
{
|
| 869 |
-
"epoch": 1.02,
|
| 870 |
-
"learning_rate": 4.7441037232908266e-05,
|
| 871 |
-
"loss": 1.2765,
|
| 872 |
-
"step": 72000
|
| 873 |
-
},
|
| 874 |
-
{
|
| 875 |
-
"epoch": 1.03,
|
| 876 |
-
"learning_rate": 4.74232666581368e-05,
|
| 877 |
-
"loss": 1.2742,
|
| 878 |
-
"step": 72500
|
| 879 |
-
},
|
| 880 |
-
{
|
| 881 |
-
"epoch": 1.04,
|
| 882 |
-
"learning_rate": 4.740549608336532e-05,
|
| 883 |
-
"loss": 1.2786,
|
| 884 |
-
"step": 73000
|
| 885 |
-
},
|
| 886 |
-
{
|
| 887 |
-
"epoch": 1.04,
|
| 888 |
-
"learning_rate": 4.738772550859385e-05,
|
| 889 |
-
"loss": 1.2712,
|
| 890 |
-
"step": 73500
|
| 891 |
-
},
|
| 892 |
-
{
|
| 893 |
-
"epoch": 1.05,
|
| 894 |
-
"learning_rate": 4.736995493382238e-05,
|
| 895 |
-
"loss": 1.2725,
|
| 896 |
-
"step": 74000
|
| 897 |
-
},
|
| 898 |
-
{
|
| 899 |
-
"epoch": 1.06,
|
| 900 |
-
"learning_rate": 4.735218435905091e-05,
|
| 901 |
-
"loss": 1.2768,
|
| 902 |
-
"step": 74500
|
| 903 |
-
},
|
| 904 |
-
{
|
| 905 |
-
"epoch": 1.07,
|
| 906 |
-
"learning_rate": 4.733441378427944e-05,
|
| 907 |
-
"loss": 1.2833,
|
| 908 |
-
"step": 75000
|
| 909 |
-
},
|
| 910 |
-
{
|
| 911 |
-
"epoch": 1.07,
|
| 912 |
-
"learning_rate": 4.731664320950797e-05,
|
| 913 |
-
"loss": 1.2744,
|
| 914 |
-
"step": 75500
|
| 915 |
-
},
|
| 916 |
-
{
|
| 917 |
-
"epoch": 1.08,
|
| 918 |
-
"learning_rate": 4.7298872634736505e-05,
|
| 919 |
-
"loss": 1.2556,
|
| 920 |
-
"step": 76000
|
| 921 |
-
},
|
| 922 |
-
{
|
| 923 |
-
"epoch": 1.09,
|
| 924 |
-
"learning_rate": 4.728110205996503e-05,
|
| 925 |
-
"loss": 1.2811,
|
| 926 |
-
"step": 76500
|
| 927 |
-
},
|
| 928 |
-
{
|
| 929 |
-
"epoch": 1.09,
|
| 930 |
-
"learning_rate": 4.7263331485193554e-05,
|
| 931 |
-
"loss": 1.2795,
|
| 932 |
-
"step": 77000
|
| 933 |
-
},
|
| 934 |
-
{
|
| 935 |
-
"epoch": 1.1,
|
| 936 |
-
"learning_rate": 4.7245560910422085e-05,
|
| 937 |
-
"loss": 1.2865,
|
| 938 |
-
"step": 77500
|
| 939 |
-
},
|
| 940 |
-
{
|
| 941 |
-
"epoch": 1.11,
|
| 942 |
-
"learning_rate": 4.722779033565062e-05,
|
| 943 |
-
"loss": 1.2924,
|
| 944 |
-
"step": 78000
|
| 945 |
-
},
|
| 946 |
-
{
|
| 947 |
-
"epoch": 1.12,
|
| 948 |
-
"learning_rate": 4.721001976087915e-05,
|
| 949 |
-
"loss": 1.2786,
|
| 950 |
-
"step": 78500
|
| 951 |
-
},
|
| 952 |
-
{
|
| 953 |
-
"epoch": 1.12,
|
| 954 |
-
"learning_rate": 4.719224918610768e-05,
|
| 955 |
-
"loss": 1.2756,
|
| 956 |
-
"step": 79000
|
| 957 |
-
},
|
| 958 |
-
{
|
| 959 |
-
"epoch": 1.13,
|
| 960 |
-
"learning_rate": 4.717447861133621e-05,
|
| 961 |
-
"loss": 1.2816,
|
| 962 |
-
"step": 79500
|
| 963 |
-
},
|
| 964 |
-
{
|
| 965 |
-
"epoch": 1.14,
|
| 966 |
-
"learning_rate": 4.715670803656474e-05,
|
| 967 |
-
"loss": 1.2823,
|
| 968 |
-
"step": 80000
|
| 969 |
-
},
|
| 970 |
-
{
|
| 971 |
-
"epoch": 1.14,
|
| 972 |
-
"learning_rate": 4.713893746179327e-05,
|
| 973 |
-
"loss": 1.2742,
|
| 974 |
-
"step": 80500
|
| 975 |
-
},
|
| 976 |
-
{
|
| 977 |
-
"epoch": 1.15,
|
| 978 |
-
"learning_rate": 4.712116688702179e-05,
|
| 979 |
-
"loss": 1.2783,
|
| 980 |
-
"step": 81000
|
| 981 |
-
},
|
| 982 |
-
{
|
| 983 |
-
"epoch": 1.16,
|
| 984 |
-
"learning_rate": 4.710339631225032e-05,
|
| 985 |
-
"loss": 1.2796,
|
| 986 |
-
"step": 81500
|
| 987 |
-
},
|
| 988 |
-
{
|
| 989 |
-
"epoch": 1.17,
|
| 990 |
-
"learning_rate": 4.7085625737478855e-05,
|
| 991 |
-
"loss": 1.271,
|
| 992 |
-
"step": 82000
|
| 993 |
-
},
|
| 994 |
-
{
|
| 995 |
-
"epoch": 1.17,
|
| 996 |
-
"learning_rate": 4.7067855162707386e-05,
|
| 997 |
-
"loss": 1.265,
|
| 998 |
-
"step": 82500
|
| 999 |
-
},
|
| 1000 |
-
{
|
| 1001 |
-
"epoch": 1.18,
|
| 1002 |
-
"learning_rate": 4.705008458793592e-05,
|
| 1003 |
-
"loss": 1.2875,
|
| 1004 |
-
"step": 83000
|
| 1005 |
-
},
|
| 1006 |
-
{
|
| 1007 |
-
"epoch": 1.19,
|
| 1008 |
-
"learning_rate": 4.703231401316444e-05,
|
| 1009 |
-
"loss": 1.2732,
|
| 1010 |
-
"step": 83500
|
| 1011 |
-
},
|
| 1012 |
-
{
|
| 1013 |
-
"epoch": 1.19,
|
| 1014 |
-
"learning_rate": 4.7014543438392974e-05,
|
| 1015 |
-
"loss": 1.2698,
|
| 1016 |
-
"step": 84000
|
| 1017 |
-
},
|
| 1018 |
-
{
|
| 1019 |
-
"epoch": 1.2,
|
| 1020 |
-
"learning_rate": 4.6996772863621505e-05,
|
| 1021 |
-
"loss": 1.2909,
|
| 1022 |
-
"step": 84500
|
| 1023 |
-
},
|
| 1024 |
-
{
|
| 1025 |
-
"epoch": 1.21,
|
| 1026 |
-
"learning_rate": 4.697900228885003e-05,
|
| 1027 |
-
"loss": 1.2706,
|
| 1028 |
-
"step": 85000
|
| 1029 |
-
},
|
| 1030 |
-
{
|
| 1031 |
-
"epoch": 1.22,
|
| 1032 |
-
"learning_rate": 4.696123171407856e-05,
|
| 1033 |
-
"loss": 1.2891,
|
| 1034 |
-
"step": 85500
|
| 1035 |
-
},
|
| 1036 |
-
{
|
| 1037 |
-
"epoch": 1.22,
|
| 1038 |
-
"learning_rate": 4.694346113930709e-05,
|
| 1039 |
-
"loss": 1.2761,
|
| 1040 |
-
"step": 86000
|
| 1041 |
-
},
|
| 1042 |
-
{
|
| 1043 |
-
"epoch": 1.23,
|
| 1044 |
-
"learning_rate": 4.6925690564535624e-05,
|
| 1045 |
-
"loss": 1.2864,
|
| 1046 |
-
"step": 86500
|
| 1047 |
-
},
|
| 1048 |
-
{
|
| 1049 |
-
"epoch": 1.24,
|
| 1050 |
-
"learning_rate": 4.690791998976415e-05,
|
| 1051 |
-
"loss": 1.279,
|
| 1052 |
-
"step": 87000
|
| 1053 |
-
},
|
| 1054 |
-
{
|
| 1055 |
-
"epoch": 1.24,
|
| 1056 |
-
"learning_rate": 4.689014941499268e-05,
|
| 1057 |
-
"loss": 1.276,
|
| 1058 |
-
"step": 87500
|
| 1059 |
-
},
|
| 1060 |
-
{
|
| 1061 |
-
"epoch": 1.25,
|
| 1062 |
-
"learning_rate": 4.687237884022121e-05,
|
| 1063 |
-
"loss": 1.2674,
|
| 1064 |
-
"step": 88000
|
| 1065 |
-
},
|
| 1066 |
-
{
|
| 1067 |
-
"epoch": 1.26,
|
| 1068 |
-
"learning_rate": 4.6854608265449736e-05,
|
| 1069 |
-
"loss": 1.2738,
|
| 1070 |
-
"step": 88500
|
| 1071 |
-
},
|
| 1072 |
-
{
|
| 1073 |
-
"epoch": 1.27,
|
| 1074 |
-
"learning_rate": 4.683683769067827e-05,
|
| 1075 |
-
"loss": 1.2683,
|
| 1076 |
-
"step": 89000
|
| 1077 |
-
},
|
| 1078 |
-
{
|
| 1079 |
-
"epoch": 1.27,
|
| 1080 |
-
"learning_rate": 4.68190671159068e-05,
|
| 1081 |
-
"loss": 1.2658,
|
| 1082 |
-
"step": 89500
|
| 1083 |
-
},
|
| 1084 |
-
{
|
| 1085 |
-
"epoch": 1.28,
|
| 1086 |
-
"learning_rate": 4.680129654113533e-05,
|
| 1087 |
-
"loss": 1.276,
|
| 1088 |
-
"step": 90000
|
| 1089 |
-
},
|
| 1090 |
-
{
|
| 1091 |
-
"epoch": 1.29,
|
| 1092 |
-
"learning_rate": 4.6783525966363855e-05,
|
| 1093 |
-
"loss": 1.2703,
|
| 1094 |
-
"step": 90500
|
| 1095 |
-
},
|
| 1096 |
-
{
|
| 1097 |
-
"epoch": 1.29,
|
| 1098 |
-
"learning_rate": 4.676575539159239e-05,
|
| 1099 |
-
"loss": 1.276,
|
| 1100 |
-
"step": 91000
|
| 1101 |
-
},
|
| 1102 |
-
{
|
| 1103 |
-
"epoch": 1.3,
|
| 1104 |
-
"learning_rate": 4.674798481682092e-05,
|
| 1105 |
-
"loss": 1.2564,
|
| 1106 |
-
"step": 91500
|
| 1107 |
-
},
|
| 1108 |
-
{
|
| 1109 |
-
"epoch": 1.31,
|
| 1110 |
-
"learning_rate": 4.673021424204945e-05,
|
| 1111 |
-
"loss": 1.2743,
|
| 1112 |
-
"step": 92000
|
| 1113 |
-
},
|
| 1114 |
-
{
|
| 1115 |
-
"epoch": 1.32,
|
| 1116 |
-
"learning_rate": 4.6712443667277974e-05,
|
| 1117 |
-
"loss": 1.2737,
|
| 1118 |
-
"step": 92500
|
| 1119 |
-
},
|
| 1120 |
-
{
|
| 1121 |
-
"epoch": 1.32,
|
| 1122 |
-
"learning_rate": 4.6694673092506506e-05,
|
| 1123 |
-
"loss": 1.2653,
|
| 1124 |
-
"step": 93000
|
| 1125 |
-
},
|
| 1126 |
-
{
|
| 1127 |
-
"epoch": 1.33,
|
| 1128 |
-
"learning_rate": 4.667690251773504e-05,
|
| 1129 |
-
"loss": 1.2644,
|
| 1130 |
-
"step": 93500
|
| 1131 |
-
},
|
| 1132 |
-
{
|
| 1133 |
-
"epoch": 1.34,
|
| 1134 |
-
"learning_rate": 4.665913194296356e-05,
|
| 1135 |
-
"loss": 1.2695,
|
| 1136 |
-
"step": 94000
|
| 1137 |
-
},
|
| 1138 |
-
{
|
| 1139 |
-
"epoch": 1.34,
|
| 1140 |
-
"learning_rate": 4.6641361368192094e-05,
|
| 1141 |
-
"loss": 1.2794,
|
| 1142 |
-
"step": 94500
|
| 1143 |
-
},
|
| 1144 |
-
{
|
| 1145 |
-
"epoch": 1.35,
|
| 1146 |
-
"learning_rate": 4.6623590793420625e-05,
|
| 1147 |
-
"loss": 1.2765,
|
| 1148 |
-
"step": 95000
|
| 1149 |
-
},
|
| 1150 |
-
{
|
| 1151 |
-
"epoch": 1.36,
|
| 1152 |
-
"learning_rate": 4.6605820218649156e-05,
|
| 1153 |
-
"loss": 1.2737,
|
| 1154 |
-
"step": 95500
|
| 1155 |
-
},
|
| 1156 |
-
{
|
| 1157 |
-
"epoch": 1.36,
|
| 1158 |
-
"learning_rate": 4.658804964387769e-05,
|
| 1159 |
-
"loss": 1.2639,
|
| 1160 |
-
"step": 96000
|
| 1161 |
-
},
|
| 1162 |
-
{
|
| 1163 |
-
"epoch": 1.37,
|
| 1164 |
-
"learning_rate": 4.657027906910621e-05,
|
| 1165 |
-
"loss": 1.2762,
|
| 1166 |
-
"step": 96500
|
| 1167 |
-
},
|
| 1168 |
-
{
|
| 1169 |
-
"epoch": 1.38,
|
| 1170 |
-
"learning_rate": 4.6552508494334744e-05,
|
| 1171 |
-
"loss": 1.26,
|
| 1172 |
-
"step": 97000
|
| 1173 |
-
},
|
| 1174 |
-
{
|
| 1175 |
-
"epoch": 1.39,
|
| 1176 |
-
"learning_rate": 4.653473791956327e-05,
|
| 1177 |
-
"loss": 1.2733,
|
| 1178 |
-
"step": 97500
|
| 1179 |
-
},
|
| 1180 |
-
{
|
| 1181 |
-
"epoch": 1.39,
|
| 1182 |
-
"learning_rate": 4.65169673447918e-05,
|
| 1183 |
-
"loss": 1.2685,
|
| 1184 |
-
"step": 98000
|
| 1185 |
-
},
|
| 1186 |
-
{
|
| 1187 |
-
"epoch": 1.4,
|
| 1188 |
-
"learning_rate": 4.649919677002033e-05,
|
| 1189 |
-
"loss": 1.2654,
|
| 1190 |
-
"step": 98500
|
| 1191 |
-
},
|
| 1192 |
-
{
|
| 1193 |
-
"epoch": 1.41,
|
| 1194 |
-
"learning_rate": 4.648142619524886e-05,
|
| 1195 |
-
"loss": 1.2741,
|
| 1196 |
-
"step": 99000
|
| 1197 |
-
},
|
| 1198 |
-
{
|
| 1199 |
-
"epoch": 1.41,
|
| 1200 |
-
"learning_rate": 4.6463655620477394e-05,
|
| 1201 |
-
"loss": 1.2772,
|
| 1202 |
-
"step": 99500
|
| 1203 |
-
},
|
| 1204 |
-
{
|
| 1205 |
-
"epoch": 1.42,
|
| 1206 |
-
"learning_rate": 4.644588504570592e-05,
|
| 1207 |
-
"loss": 1.274,
|
| 1208 |
-
"step": 100000
|
| 1209 |
-
},
|
| 1210 |
-
{
|
| 1211 |
-
"epoch": 1.43,
|
| 1212 |
-
"learning_rate": 4.642811447093445e-05,
|
| 1213 |
-
"loss": 1.2505,
|
| 1214 |
-
"step": 100500
|
| 1215 |
-
},
|
| 1216 |
-
{
|
| 1217 |
-
"epoch": 1.44,
|
| 1218 |
-
"learning_rate": 4.6410343896162975e-05,
|
| 1219 |
-
"loss": 1.2666,
|
| 1220 |
-
"step": 101000
|
| 1221 |
-
},
|
| 1222 |
-
{
|
| 1223 |
-
"epoch": 1.44,
|
| 1224 |
-
"learning_rate": 4.639257332139151e-05,
|
| 1225 |
-
"loss": 1.2571,
|
| 1226 |
-
"step": 101500
|
| 1227 |
-
},
|
| 1228 |
-
{
|
| 1229 |
-
"epoch": 1.45,
|
| 1230 |
-
"learning_rate": 4.637480274662004e-05,
|
| 1231 |
-
"loss": 1.2551,
|
| 1232 |
-
"step": 102000
|
| 1233 |
-
},
|
| 1234 |
-
{
|
| 1235 |
-
"epoch": 1.46,
|
| 1236 |
-
"learning_rate": 4.635703217184857e-05,
|
| 1237 |
-
"loss": 1.2546,
|
| 1238 |
-
"step": 102500
|
| 1239 |
-
},
|
| 1240 |
-
{
|
| 1241 |
-
"epoch": 1.46,
|
| 1242 |
-
"learning_rate": 4.63392615970771e-05,
|
| 1243 |
-
"loss": 1.271,
|
| 1244 |
-
"step": 103000
|
| 1245 |
-
},
|
| 1246 |
-
{
|
| 1247 |
-
"epoch": 1.47,
|
| 1248 |
-
"learning_rate": 4.632149102230563e-05,
|
| 1249 |
-
"loss": 1.2672,
|
| 1250 |
-
"step": 103500
|
| 1251 |
-
},
|
| 1252 |
-
{
|
| 1253 |
-
"epoch": 1.48,
|
| 1254 |
-
"learning_rate": 4.630372044753416e-05,
|
| 1255 |
-
"loss": 1.2502,
|
| 1256 |
-
"step": 104000
|
| 1257 |
-
},
|
| 1258 |
-
{
|
| 1259 |
-
"epoch": 1.49,
|
| 1260 |
-
"learning_rate": 4.628594987276268e-05,
|
| 1261 |
-
"loss": 1.2706,
|
| 1262 |
-
"step": 104500
|
| 1263 |
-
},
|
| 1264 |
-
{
|
| 1265 |
-
"epoch": 1.49,
|
| 1266 |
-
"learning_rate": 4.626817929799121e-05,
|
| 1267 |
-
"loss": 1.2685,
|
| 1268 |
-
"step": 105000
|
| 1269 |
-
},
|
| 1270 |
-
{
|
| 1271 |
-
"epoch": 1.5,
|
| 1272 |
-
"learning_rate": 4.6250408723219745e-05,
|
| 1273 |
-
"loss": 1.2456,
|
| 1274 |
-
"step": 105500
|
| 1275 |
-
},
|
| 1276 |
-
{
|
| 1277 |
-
"epoch": 1.51,
|
| 1278 |
-
"learning_rate": 4.6232638148448276e-05,
|
| 1279 |
-
"loss": 1.2636,
|
| 1280 |
-
"step": 106000
|
| 1281 |
-
},
|
| 1282 |
-
{
|
| 1283 |
-
"epoch": 1.51,
|
| 1284 |
-
"learning_rate": 4.621486757367681e-05,
|
| 1285 |
-
"loss": 1.2736,
|
| 1286 |
-
"step": 106500
|
| 1287 |
-
},
|
| 1288 |
-
{
|
| 1289 |
-
"epoch": 1.52,
|
| 1290 |
-
"learning_rate": 4.619709699890534e-05,
|
| 1291 |
-
"loss": 1.2667,
|
| 1292 |
-
"step": 107000
|
| 1293 |
-
},
|
| 1294 |
-
{
|
| 1295 |
-
"epoch": 1.53,
|
| 1296 |
-
"learning_rate": 4.6179326424133864e-05,
|
| 1297 |
-
"loss": 1.2524,
|
| 1298 |
-
"step": 107500
|
| 1299 |
-
},
|
| 1300 |
-
{
|
| 1301 |
-
"epoch": 1.54,
|
| 1302 |
-
"learning_rate": 4.6161555849362395e-05,
|
| 1303 |
-
"loss": 1.2742,
|
| 1304 |
-
"step": 108000
|
| 1305 |
-
},
|
| 1306 |
-
{
|
| 1307 |
-
"epoch": 1.54,
|
| 1308 |
-
"learning_rate": 4.614378527459092e-05,
|
| 1309 |
-
"loss": 1.2392,
|
| 1310 |
-
"step": 108500
|
| 1311 |
-
},
|
| 1312 |
-
{
|
| 1313 |
-
"epoch": 1.55,
|
| 1314 |
-
"learning_rate": 4.612601469981945e-05,
|
| 1315 |
-
"loss": 1.2555,
|
| 1316 |
-
"step": 109000
|
| 1317 |
-
},
|
| 1318 |
-
{
|
| 1319 |
-
"epoch": 1.56,
|
| 1320 |
-
"learning_rate": 4.610824412504798e-05,
|
| 1321 |
-
"loss": 1.2543,
|
| 1322 |
-
"step": 109500
|
| 1323 |
-
},
|
| 1324 |
-
{
|
| 1325 |
-
"epoch": 1.56,
|
| 1326 |
-
"learning_rate": 4.6090473550276514e-05,
|
| 1327 |
-
"loss": 1.2676,
|
| 1328 |
-
"step": 110000
|
| 1329 |
-
},
|
| 1330 |
-
{
|
| 1331 |
-
"epoch": 1.57,
|
| 1332 |
-
"learning_rate": 4.6072702975505046e-05,
|
| 1333 |
-
"loss": 1.2501,
|
| 1334 |
-
"step": 110500
|
| 1335 |
-
},
|
| 1336 |
-
{
|
| 1337 |
-
"epoch": 1.58,
|
| 1338 |
-
"learning_rate": 4.605493240073357e-05,
|
| 1339 |
-
"loss": 1.2603,
|
| 1340 |
-
"step": 111000
|
| 1341 |
-
},
|
| 1342 |
-
{
|
| 1343 |
-
"epoch": 1.59,
|
| 1344 |
-
"learning_rate": 4.60371618259621e-05,
|
| 1345 |
-
"loss": 1.2625,
|
| 1346 |
-
"step": 111500
|
| 1347 |
-
},
|
| 1348 |
-
{
|
| 1349 |
-
"epoch": 1.59,
|
| 1350 |
-
"learning_rate": 4.601939125119063e-05,
|
| 1351 |
-
"loss": 1.2649,
|
| 1352 |
-
"step": 112000
|
| 1353 |
-
},
|
| 1354 |
-
{
|
| 1355 |
-
"epoch": 1.6,
|
| 1356 |
-
"learning_rate": 4.600162067641916e-05,
|
| 1357 |
-
"loss": 1.2574,
|
| 1358 |
-
"step": 112500
|
| 1359 |
-
},
|
| 1360 |
-
{
|
| 1361 |
-
"epoch": 1.61,
|
| 1362 |
-
"learning_rate": 4.598385010164769e-05,
|
| 1363 |
-
"loss": 1.265,
|
| 1364 |
-
"step": 113000
|
| 1365 |
-
},
|
| 1366 |
-
{
|
| 1367 |
-
"epoch": 1.61,
|
| 1368 |
-
"learning_rate": 4.596607952687622e-05,
|
| 1369 |
-
"loss": 1.2608,
|
| 1370 |
-
"step": 113500
|
| 1371 |
-
},
|
| 1372 |
-
{
|
| 1373 |
-
"epoch": 1.62,
|
| 1374 |
-
"learning_rate": 4.594830895210475e-05,
|
| 1375 |
-
"loss": 1.259,
|
| 1376 |
-
"step": 114000
|
| 1377 |
-
},
|
| 1378 |
-
{
|
| 1379 |
-
"epoch": 1.63,
|
| 1380 |
-
"learning_rate": 4.593053837733328e-05,
|
| 1381 |
-
"loss": 1.2666,
|
| 1382 |
-
"step": 114500
|
| 1383 |
-
},
|
| 1384 |
-
{
|
| 1385 |
-
"epoch": 1.63,
|
| 1386 |
-
"learning_rate": 4.591276780256181e-05,
|
| 1387 |
-
"loss": 1.2678,
|
| 1388 |
-
"step": 115000
|
| 1389 |
-
},
|
| 1390 |
-
{
|
| 1391 |
-
"epoch": 1.64,
|
| 1392 |
-
"learning_rate": 4.589499722779034e-05,
|
| 1393 |
-
"loss": 1.2744,
|
| 1394 |
-
"step": 115500
|
| 1395 |
-
},
|
| 1396 |
-
{
|
| 1397 |
-
"epoch": 1.65,
|
| 1398 |
-
"learning_rate": 4.5877226653018864e-05,
|
| 1399 |
-
"loss": 1.2672,
|
| 1400 |
-
"step": 116000
|
| 1401 |
-
},
|
| 1402 |
-
{
|
| 1403 |
-
"epoch": 1.66,
|
| 1404 |
-
"learning_rate": 4.5859456078247396e-05,
|
| 1405 |
-
"loss": 1.265,
|
| 1406 |
-
"step": 116500
|
| 1407 |
-
},
|
| 1408 |
-
{
|
| 1409 |
-
"epoch": 1.66,
|
| 1410 |
-
"learning_rate": 4.584168550347593e-05,
|
| 1411 |
-
"loss": 1.273,
|
| 1412 |
-
"step": 117000
|
| 1413 |
-
},
|
| 1414 |
-
{
|
| 1415 |
-
"epoch": 1.67,
|
| 1416 |
-
"learning_rate": 4.582391492870446e-05,
|
| 1417 |
-
"loss": 1.2635,
|
| 1418 |
-
"step": 117500
|
| 1419 |
-
},
|
| 1420 |
-
{
|
| 1421 |
-
"epoch": 1.68,
|
| 1422 |
-
"learning_rate": 4.5806144353932983e-05,
|
| 1423 |
-
"loss": 1.268,
|
| 1424 |
-
"step": 118000
|
| 1425 |
-
},
|
| 1426 |
-
{
|
| 1427 |
-
"epoch": 1.68,
|
| 1428 |
-
"learning_rate": 4.5788373779161515e-05,
|
| 1429 |
-
"loss": 1.2722,
|
| 1430 |
-
"step": 118500
|
| 1431 |
-
},
|
| 1432 |
-
{
|
| 1433 |
-
"epoch": 1.69,
|
| 1434 |
-
"learning_rate": 4.5770603204390046e-05,
|
| 1435 |
-
"loss": 1.2534,
|
| 1436 |
-
"step": 119000
|
| 1437 |
-
},
|
| 1438 |
-
{
|
| 1439 |
-
"epoch": 1.7,
|
| 1440 |
-
"learning_rate": 4.575283262961858e-05,
|
| 1441 |
-
"loss": 1.2545,
|
| 1442 |
-
"step": 119500
|
| 1443 |
-
},
|
| 1444 |
-
{
|
| 1445 |
-
"epoch": 1.71,
|
| 1446 |
-
"learning_rate": 4.57350620548471e-05,
|
| 1447 |
-
"loss": 1.2465,
|
| 1448 |
-
"step": 120000
|
| 1449 |
-
},
|
| 1450 |
-
{
|
| 1451 |
-
"epoch": 1.71,
|
| 1452 |
-
"learning_rate": 4.5717291480075634e-05,
|
| 1453 |
-
"loss": 1.2459,
|
| 1454 |
-
"step": 120500
|
| 1455 |
-
},
|
| 1456 |
-
{
|
| 1457 |
-
"epoch": 1.72,
|
| 1458 |
-
"learning_rate": 4.569952090530416e-05,
|
| 1459 |
-
"loss": 1.2655,
|
| 1460 |
-
"step": 121000
|
| 1461 |
-
},
|
| 1462 |
-
{
|
| 1463 |
-
"epoch": 1.73,
|
| 1464 |
-
"learning_rate": 4.568175033053269e-05,
|
| 1465 |
-
"loss": 1.2582,
|
| 1466 |
-
"step": 121500
|
| 1467 |
-
},
|
| 1468 |
-
{
|
| 1469 |
-
"epoch": 1.73,
|
| 1470 |
-
"learning_rate": 4.566397975576122e-05,
|
| 1471 |
-
"loss": 1.2516,
|
| 1472 |
-
"step": 122000
|
| 1473 |
-
},
|
| 1474 |
-
{
|
| 1475 |
-
"epoch": 1.74,
|
| 1476 |
-
"learning_rate": 4.564620918098975e-05,
|
| 1477 |
-
"loss": 1.2683,
|
| 1478 |
-
"step": 122500
|
| 1479 |
-
},
|
| 1480 |
-
{
|
| 1481 |
-
"epoch": 1.75,
|
| 1482 |
-
"learning_rate": 4.5628438606218284e-05,
|
| 1483 |
-
"loss": 1.2683,
|
| 1484 |
-
"step": 123000
|
| 1485 |
-
},
|
| 1486 |
-
{
|
| 1487 |
-
"epoch": 1.76,
|
| 1488 |
-
"learning_rate": 4.5610668031446816e-05,
|
| 1489 |
-
"loss": 1.2487,
|
| 1490 |
-
"step": 123500
|
| 1491 |
-
},
|
| 1492 |
-
{
|
| 1493 |
-
"epoch": 1.76,
|
| 1494 |
-
"learning_rate": 4.559289745667534e-05,
|
| 1495 |
-
"loss": 1.2609,
|
| 1496 |
-
"step": 124000
|
| 1497 |
-
},
|
| 1498 |
-
{
|
| 1499 |
-
"epoch": 1.77,
|
| 1500 |
-
"learning_rate": 4.5575126881903865e-05,
|
| 1501 |
-
"loss": 1.2684,
|
| 1502 |
-
"step": 124500
|
| 1503 |
-
},
|
| 1504 |
-
{
|
| 1505 |
-
"epoch": 1.78,
|
| 1506 |
-
"learning_rate": 4.5557356307132397e-05,
|
| 1507 |
-
"loss": 1.2713,
|
| 1508 |
-
"step": 125000
|
| 1509 |
-
},
|
| 1510 |
-
{
|
| 1511 |
-
"epoch": 1.78,
|
| 1512 |
-
"learning_rate": 4.553958573236093e-05,
|
| 1513 |
-
"loss": 1.2664,
|
| 1514 |
-
"step": 125500
|
| 1515 |
-
},
|
| 1516 |
-
{
|
| 1517 |
-
"epoch": 1.79,
|
| 1518 |
-
"learning_rate": 4.552181515758946e-05,
|
| 1519 |
-
"loss": 1.2708,
|
| 1520 |
-
"step": 126000
|
| 1521 |
-
},
|
| 1522 |
-
{
|
| 1523 |
-
"epoch": 1.8,
|
| 1524 |
-
"learning_rate": 4.550404458281799e-05,
|
| 1525 |
-
"loss": 1.2658,
|
| 1526 |
-
"step": 126500
|
| 1527 |
-
},
|
| 1528 |
-
{
|
| 1529 |
-
"epoch": 1.81,
|
| 1530 |
-
"learning_rate": 4.548627400804652e-05,
|
| 1531 |
-
"loss": 1.2674,
|
| 1532 |
-
"step": 127000
|
| 1533 |
-
},
|
| 1534 |
-
{
|
| 1535 |
-
"epoch": 1.81,
|
| 1536 |
-
"learning_rate": 4.546850343327505e-05,
|
| 1537 |
-
"loss": 1.2476,
|
| 1538 |
-
"step": 127500
|
| 1539 |
-
},
|
| 1540 |
-
{
|
| 1541 |
-
"epoch": 1.82,
|
| 1542 |
-
"learning_rate": 4.545073285850358e-05,
|
| 1543 |
-
"loss": 1.265,
|
| 1544 |
-
"step": 128000
|
| 1545 |
-
},
|
| 1546 |
-
{
|
| 1547 |
-
"epoch": 1.83,
|
| 1548 |
-
"learning_rate": 4.54329622837321e-05,
|
| 1549 |
-
"loss": 1.2622,
|
| 1550 |
-
"step": 128500
|
| 1551 |
-
},
|
| 1552 |
-
{
|
| 1553 |
-
"epoch": 1.83,
|
| 1554 |
-
"learning_rate": 4.5415191708960635e-05,
|
| 1555 |
-
"loss": 1.2434,
|
| 1556 |
-
"step": 129000
|
| 1557 |
-
},
|
| 1558 |
-
{
|
| 1559 |
-
"epoch": 1.84,
|
| 1560 |
-
"learning_rate": 4.5397421134189166e-05,
|
| 1561 |
-
"loss": 1.2428,
|
| 1562 |
-
"step": 129500
|
| 1563 |
-
},
|
| 1564 |
-
{
|
| 1565 |
-
"epoch": 1.85,
|
| 1566 |
-
"learning_rate": 4.53796505594177e-05,
|
| 1567 |
-
"loss": 1.261,
|
| 1568 |
-
"step": 130000
|
| 1569 |
-
},
|
| 1570 |
-
{
|
| 1571 |
-
"epoch": 1.86,
|
| 1572 |
-
"learning_rate": 4.536187998464623e-05,
|
| 1573 |
-
"loss": 1.2547,
|
| 1574 |
-
"step": 130500
|
| 1575 |
-
},
|
| 1576 |
-
{
|
| 1577 |
-
"epoch": 1.86,
|
| 1578 |
-
"learning_rate": 4.534410940987476e-05,
|
| 1579 |
-
"loss": 1.2441,
|
| 1580 |
-
"step": 131000
|
| 1581 |
-
},
|
| 1582 |
-
{
|
| 1583 |
-
"epoch": 1.87,
|
| 1584 |
-
"learning_rate": 4.5326338835103285e-05,
|
| 1585 |
-
"loss": 1.2617,
|
| 1586 |
-
"step": 131500
|
| 1587 |
-
},
|
| 1588 |
-
{
|
| 1589 |
-
"epoch": 1.88,
|
| 1590 |
-
"learning_rate": 4.530856826033181e-05,
|
| 1591 |
-
"loss": 1.2587,
|
| 1592 |
-
"step": 132000
|
| 1593 |
-
},
|
| 1594 |
-
{
|
| 1595 |
-
"epoch": 1.88,
|
| 1596 |
-
"learning_rate": 4.529079768556034e-05,
|
| 1597 |
-
"loss": 1.2491,
|
| 1598 |
-
"step": 132500
|
| 1599 |
-
},
|
| 1600 |
-
{
|
| 1601 |
-
"epoch": 1.89,
|
| 1602 |
-
"learning_rate": 4.527302711078887e-05,
|
| 1603 |
-
"loss": 1.2418,
|
| 1604 |
-
"step": 133000
|
| 1605 |
-
},
|
| 1606 |
-
{
|
| 1607 |
-
"epoch": 1.9,
|
| 1608 |
-
"learning_rate": 4.5255256536017404e-05,
|
| 1609 |
-
"loss": 1.2558,
|
| 1610 |
-
"step": 133500
|
| 1611 |
-
},
|
| 1612 |
-
{
|
| 1613 |
-
"epoch": 1.91,
|
| 1614 |
-
"learning_rate": 4.5237485961245936e-05,
|
| 1615 |
-
"loss": 1.2497,
|
| 1616 |
-
"step": 134000
|
| 1617 |
-
},
|
| 1618 |
-
{
|
| 1619 |
-
"epoch": 1.91,
|
| 1620 |
-
"learning_rate": 4.521971538647447e-05,
|
| 1621 |
-
"loss": 1.2672,
|
| 1622 |
-
"step": 134500
|
| 1623 |
-
},
|
| 1624 |
-
{
|
| 1625 |
-
"epoch": 1.92,
|
| 1626 |
-
"learning_rate": 4.520194481170299e-05,
|
| 1627 |
-
"loss": 1.2667,
|
| 1628 |
-
"step": 135000
|
| 1629 |
-
},
|
| 1630 |
-
{
|
| 1631 |
-
"epoch": 1.93,
|
| 1632 |
-
"learning_rate": 4.518417423693152e-05,
|
| 1633 |
-
"loss": 1.2577,
|
| 1634 |
-
"step": 135500
|
| 1635 |
-
},
|
| 1636 |
-
{
|
| 1637 |
-
"epoch": 1.93,
|
| 1638 |
-
"learning_rate": 4.516640366216005e-05,
|
| 1639 |
-
"loss": 1.2504,
|
| 1640 |
-
"step": 136000
|
| 1641 |
-
},
|
| 1642 |
-
{
|
| 1643 |
-
"epoch": 1.94,
|
| 1644 |
-
"learning_rate": 4.514863308738858e-05,
|
| 1645 |
-
"loss": 1.2318,
|
| 1646 |
-
"step": 136500
|
| 1647 |
-
},
|
| 1648 |
-
{
|
| 1649 |
-
"epoch": 1.95,
|
| 1650 |
-
"learning_rate": 4.513086251261711e-05,
|
| 1651 |
-
"loss": 1.2371,
|
| 1652 |
-
"step": 137000
|
| 1653 |
-
},
|
| 1654 |
-
{
|
| 1655 |
-
"epoch": 1.95,
|
| 1656 |
-
"learning_rate": 4.511309193784564e-05,
|
| 1657 |
-
"loss": 1.249,
|
| 1658 |
-
"step": 137500
|
| 1659 |
-
},
|
| 1660 |
-
{
|
| 1661 |
-
"epoch": 1.96,
|
| 1662 |
-
"learning_rate": 4.509532136307417e-05,
|
| 1663 |
-
"loss": 1.2479,
|
| 1664 |
-
"step": 138000
|
| 1665 |
-
},
|
| 1666 |
-
{
|
| 1667 |
-
"epoch": 1.97,
|
| 1668 |
-
"learning_rate": 4.50775507883027e-05,
|
| 1669 |
-
"loss": 1.2432,
|
| 1670 |
-
"step": 138500
|
| 1671 |
-
},
|
| 1672 |
-
{
|
| 1673 |
-
"epoch": 1.98,
|
| 1674 |
-
"learning_rate": 4.505978021353123e-05,
|
| 1675 |
-
"loss": 1.2518,
|
| 1676 |
-
"step": 139000
|
| 1677 |
-
},
|
| 1678 |
-
{
|
| 1679 |
-
"epoch": 1.98,
|
| 1680 |
-
"learning_rate": 4.504200963875976e-05,
|
| 1681 |
-
"loss": 1.2503,
|
| 1682 |
-
"step": 139500
|
| 1683 |
-
},
|
| 1684 |
-
{
|
| 1685 |
-
"epoch": 1.99,
|
| 1686 |
-
"learning_rate": 4.5024239063988286e-05,
|
| 1687 |
-
"loss": 1.2478,
|
| 1688 |
-
"step": 140000
|
| 1689 |
-
},
|
| 1690 |
-
{
|
| 1691 |
-
"epoch": 2.0,
|
| 1692 |
-
"learning_rate": 4.500646848921682e-05,
|
| 1693 |
-
"loss": 1.2414,
|
| 1694 |
-
"step": 140500
|
| 1695 |
-
},
|
| 1696 |
-
{
|
| 1697 |
-
"epoch": 2.0,
|
| 1698 |
-
"learning_rate": 4.498869791444535e-05,
|
| 1699 |
-
"loss": 1.1995,
|
| 1700 |
-
"step": 141000
|
| 1701 |
-
},
|
| 1702 |
-
{
|
| 1703 |
-
"epoch": 2.01,
|
| 1704 |
-
"learning_rate": 4.497092733967387e-05,
|
| 1705 |
-
"loss": 1.1555,
|
| 1706 |
-
"step": 141500
|
| 1707 |
-
},
|
| 1708 |
-
{
|
| 1709 |
-
"epoch": 2.02,
|
| 1710 |
-
"learning_rate": 4.4953156764902405e-05,
|
| 1711 |
-
"loss": 1.1378,
|
| 1712 |
-
"step": 142000
|
| 1713 |
-
},
|
| 1714 |
-
{
|
| 1715 |
-
"epoch": 2.03,
|
| 1716 |
-
"learning_rate": 4.4935386190130936e-05,
|
| 1717 |
-
"loss": 1.1664,
|
| 1718 |
-
"step": 142500
|
| 1719 |
-
},
|
| 1720 |
-
{
|
| 1721 |
-
"epoch": 2.03,
|
| 1722 |
-
"learning_rate": 4.491761561535947e-05,
|
| 1723 |
-
"loss": 1.1552,
|
| 1724 |
-
"step": 143000
|
| 1725 |
-
},
|
| 1726 |
-
{
|
| 1727 |
-
"epoch": 2.04,
|
| 1728 |
-
"learning_rate": 4.489984504058799e-05,
|
| 1729 |
-
"loss": 1.1738,
|
| 1730 |
-
"step": 143500
|
| 1731 |
-
},
|
| 1732 |
-
{
|
| 1733 |
-
"epoch": 2.05,
|
| 1734 |
-
"learning_rate": 4.4882074465816524e-05,
|
| 1735 |
-
"loss": 1.1595,
|
| 1736 |
-
"step": 144000
|
| 1737 |
-
},
|
| 1738 |
-
{
|
| 1739 |
-
"epoch": 2.05,
|
| 1740 |
-
"learning_rate": 4.4864303891045055e-05,
|
| 1741 |
-
"loss": 1.1746,
|
| 1742 |
-
"step": 144500
|
| 1743 |
-
},
|
| 1744 |
-
{
|
| 1745 |
-
"epoch": 2.06,
|
| 1746 |
-
"learning_rate": 4.484653331627358e-05,
|
| 1747 |
-
"loss": 1.1839,
|
| 1748 |
-
"step": 145000
|
| 1749 |
-
},
|
| 1750 |
-
{
|
| 1751 |
-
"epoch": 2.07,
|
| 1752 |
-
"learning_rate": 4.482876274150211e-05,
|
| 1753 |
-
"loss": 1.1628,
|
| 1754 |
-
"step": 145500
|
| 1755 |
-
},
|
| 1756 |
-
{
|
| 1757 |
-
"epoch": 2.08,
|
| 1758 |
-
"learning_rate": 4.481099216673064e-05,
|
| 1759 |
-
"loss": 1.17,
|
| 1760 |
-
"step": 146000
|
| 1761 |
-
},
|
| 1762 |
-
{
|
| 1763 |
-
"epoch": 2.08,
|
| 1764 |
-
"learning_rate": 4.4793221591959174e-05,
|
| 1765 |
-
"loss": 1.1901,
|
| 1766 |
-
"step": 146500
|
| 1767 |
-
},
|
| 1768 |
-
{
|
| 1769 |
-
"epoch": 2.09,
|
| 1770 |
-
"learning_rate": 4.4775451017187706e-05,
|
| 1771 |
-
"loss": 1.1828,
|
| 1772 |
-
"step": 147000
|
| 1773 |
-
},
|
| 1774 |
-
{
|
| 1775 |
-
"epoch": 2.1,
|
| 1776 |
-
"learning_rate": 4.475768044241623e-05,
|
| 1777 |
-
"loss": 1.1702,
|
| 1778 |
-
"step": 147500
|
| 1779 |
-
},
|
| 1780 |
-
{
|
| 1781 |
-
"epoch": 2.1,
|
| 1782 |
-
"learning_rate": 4.473990986764476e-05,
|
| 1783 |
-
"loss": 1.1679,
|
| 1784 |
-
"step": 148000
|
| 1785 |
-
},
|
| 1786 |
-
{
|
| 1787 |
-
"epoch": 2.11,
|
| 1788 |
-
"learning_rate": 4.4722139292873286e-05,
|
| 1789 |
-
"loss": 1.1726,
|
| 1790 |
-
"step": 148500
|
| 1791 |
-
},
|
| 1792 |
-
{
|
| 1793 |
-
"epoch": 2.12,
|
| 1794 |
-
"learning_rate": 4.470436871810182e-05,
|
| 1795 |
-
"loss": 1.1692,
|
| 1796 |
-
"step": 149000
|
| 1797 |
-
},
|
| 1798 |
-
{
|
| 1799 |
-
"epoch": 2.13,
|
| 1800 |
-
"learning_rate": 4.468659814333035e-05,
|
| 1801 |
-
"loss": 1.1662,
|
| 1802 |
-
"step": 149500
|
| 1803 |
-
},
|
| 1804 |
-
{
|
| 1805 |
-
"epoch": 2.13,
|
| 1806 |
-
"learning_rate": 4.466882756855888e-05,
|
| 1807 |
-
"loss": 1.1638,
|
| 1808 |
-
"step": 150000
|
| 1809 |
-
},
|
| 1810 |
-
{
|
| 1811 |
-
"epoch": 2.14,
|
| 1812 |
-
"learning_rate": 4.465105699378741e-05,
|
| 1813 |
-
"loss": 1.1829,
|
| 1814 |
-
"step": 150500
|
| 1815 |
-
},
|
| 1816 |
-
{
|
| 1817 |
-
"epoch": 2.15,
|
| 1818 |
-
"learning_rate": 4.4633286419015944e-05,
|
| 1819 |
-
"loss": 1.187,
|
| 1820 |
-
"step": 151000
|
| 1821 |
-
},
|
| 1822 |
-
{
|
| 1823 |
-
"epoch": 2.15,
|
| 1824 |
-
"learning_rate": 4.461551584424447e-05,
|
| 1825 |
-
"loss": 1.1835,
|
| 1826 |
-
"step": 151500
|
| 1827 |
-
},
|
| 1828 |
-
{
|
| 1829 |
-
"epoch": 2.16,
|
| 1830 |
-
"learning_rate": 4.459774526947299e-05,
|
| 1831 |
-
"loss": 1.1609,
|
| 1832 |
-
"step": 152000
|
| 1833 |
-
},
|
| 1834 |
-
{
|
| 1835 |
-
"epoch": 2.17,
|
| 1836 |
-
"learning_rate": 4.4579974694701525e-05,
|
| 1837 |
-
"loss": 1.1768,
|
| 1838 |
-
"step": 152500
|
| 1839 |
-
},
|
| 1840 |
-
{
|
| 1841 |
-
"epoch": 2.18,
|
| 1842 |
-
"learning_rate": 4.4562204119930056e-05,
|
| 1843 |
-
"loss": 1.1795,
|
| 1844 |
-
"step": 153000
|
| 1845 |
-
},
|
| 1846 |
-
{
|
| 1847 |
-
"epoch": 2.18,
|
| 1848 |
-
"learning_rate": 4.454443354515859e-05,
|
| 1849 |
-
"loss": 1.1621,
|
| 1850 |
-
"step": 153500
|
| 1851 |
-
},
|
| 1852 |
-
{
|
| 1853 |
-
"epoch": 2.19,
|
| 1854 |
-
"learning_rate": 4.452666297038712e-05,
|
| 1855 |
-
"loss": 1.171,
|
| 1856 |
-
"step": 154000
|
| 1857 |
-
},
|
| 1858 |
-
{
|
| 1859 |
-
"epoch": 2.2,
|
| 1860 |
-
"learning_rate": 4.450889239561565e-05,
|
| 1861 |
-
"loss": 1.18,
|
| 1862 |
-
"step": 154500
|
| 1863 |
-
},
|
| 1864 |
-
{
|
| 1865 |
-
"epoch": 2.2,
|
| 1866 |
-
"learning_rate": 4.4491121820844175e-05,
|
| 1867 |
-
"loss": 1.1797,
|
| 1868 |
-
"step": 155000
|
| 1869 |
-
},
|
| 1870 |
-
{
|
| 1871 |
-
"epoch": 2.21,
|
| 1872 |
-
"learning_rate": 4.4473351246072706e-05,
|
| 1873 |
-
"loss": 1.1735,
|
| 1874 |
-
"step": 155500
|
| 1875 |
-
},
|
| 1876 |
-
{
|
| 1877 |
-
"epoch": 2.22,
|
| 1878 |
-
"learning_rate": 4.445558067130123e-05,
|
| 1879 |
-
"loss": 1.1885,
|
| 1880 |
-
"step": 156000
|
| 1881 |
-
},
|
| 1882 |
-
{
|
| 1883 |
-
"epoch": 2.22,
|
| 1884 |
-
"learning_rate": 4.443781009652976e-05,
|
| 1885 |
-
"loss": 1.1779,
|
| 1886 |
-
"step": 156500
|
| 1887 |
-
},
|
| 1888 |
-
{
|
| 1889 |
-
"epoch": 2.23,
|
| 1890 |
-
"learning_rate": 4.4420039521758294e-05,
|
| 1891 |
-
"loss": 1.1769,
|
| 1892 |
-
"step": 157000
|
| 1893 |
-
},
|
| 1894 |
-
{
|
| 1895 |
-
"epoch": 2.24,
|
| 1896 |
-
"learning_rate": 4.4402268946986825e-05,
|
| 1897 |
-
"loss": 1.1496,
|
| 1898 |
-
"step": 157500
|
| 1899 |
-
},
|
| 1900 |
-
{
|
| 1901 |
-
"epoch": 2.25,
|
| 1902 |
-
"learning_rate": 4.438449837221536e-05,
|
| 1903 |
-
"loss": 1.1866,
|
| 1904 |
-
"step": 158000
|
| 1905 |
-
},
|
| 1906 |
-
{
|
| 1907 |
-
"epoch": 2.25,
|
| 1908 |
-
"learning_rate": 4.436672779744388e-05,
|
| 1909 |
-
"loss": 1.1765,
|
| 1910 |
-
"step": 158500
|
| 1911 |
-
},
|
| 1912 |
-
{
|
| 1913 |
-
"epoch": 2.26,
|
| 1914 |
-
"learning_rate": 4.434895722267241e-05,
|
| 1915 |
-
"loss": 1.1634,
|
| 1916 |
-
"step": 159000
|
| 1917 |
-
},
|
| 1918 |
-
{
|
| 1919 |
-
"epoch": 2.27,
|
| 1920 |
-
"learning_rate": 4.433118664790094e-05,
|
| 1921 |
-
"loss": 1.1835,
|
| 1922 |
-
"step": 159500
|
| 1923 |
-
},
|
| 1924 |
-
{
|
| 1925 |
-
"epoch": 2.27,
|
| 1926 |
-
"learning_rate": 4.431341607312947e-05,
|
| 1927 |
-
"loss": 1.1819,
|
| 1928 |
-
"step": 160000
|
| 1929 |
-
},
|
| 1930 |
-
{
|
| 1931 |
-
"epoch": 2.28,
|
| 1932 |
-
"learning_rate": 4.4295645498358e-05,
|
| 1933 |
-
"loss": 1.1869,
|
| 1934 |
-
"step": 160500
|
| 1935 |
-
},
|
| 1936 |
-
{
|
| 1937 |
-
"epoch": 2.29,
|
| 1938 |
-
"learning_rate": 4.427787492358653e-05,
|
| 1939 |
-
"loss": 1.1803,
|
| 1940 |
-
"step": 161000
|
| 1941 |
-
},
|
| 1942 |
-
{
|
| 1943 |
-
"epoch": 2.3,
|
| 1944 |
-
"learning_rate": 4.4260104348815063e-05,
|
| 1945 |
-
"loss": 1.1855,
|
| 1946 |
-
"step": 161500
|
| 1947 |
-
},
|
| 1948 |
-
{
|
| 1949 |
-
"epoch": 2.3,
|
| 1950 |
-
"learning_rate": 4.424233377404359e-05,
|
| 1951 |
-
"loss": 1.1865,
|
| 1952 |
-
"step": 162000
|
| 1953 |
-
},
|
| 1954 |
-
{
|
| 1955 |
-
"epoch": 2.31,
|
| 1956 |
-
"learning_rate": 4.422456319927212e-05,
|
| 1957 |
-
"loss": 1.192,
|
| 1958 |
-
"step": 162500
|
| 1959 |
-
},
|
| 1960 |
-
{
|
| 1961 |
-
"epoch": 2.32,
|
| 1962 |
-
"learning_rate": 4.420679262450065e-05,
|
| 1963 |
-
"loss": 1.1817,
|
| 1964 |
-
"step": 163000
|
| 1965 |
-
},
|
| 1966 |
-
{
|
| 1967 |
-
"epoch": 2.32,
|
| 1968 |
-
"learning_rate": 4.4189022049729176e-05,
|
| 1969 |
-
"loss": 1.1814,
|
| 1970 |
-
"step": 163500
|
| 1971 |
-
},
|
| 1972 |
-
{
|
| 1973 |
-
"epoch": 2.33,
|
| 1974 |
-
"learning_rate": 4.417125147495771e-05,
|
| 1975 |
-
"loss": 1.181,
|
| 1976 |
-
"step": 164000
|
| 1977 |
-
},
|
| 1978 |
-
{
|
| 1979 |
-
"epoch": 2.34,
|
| 1980 |
-
"learning_rate": 4.415348090018624e-05,
|
| 1981 |
-
"loss": 1.1756,
|
| 1982 |
-
"step": 164500
|
| 1983 |
-
},
|
| 1984 |
-
{
|
| 1985 |
-
"epoch": 2.35,
|
| 1986 |
-
"learning_rate": 4.413571032541477e-05,
|
| 1987 |
-
"loss": 1.1834,
|
| 1988 |
-
"step": 165000
|
| 1989 |
-
},
|
| 1990 |
-
{
|
| 1991 |
-
"epoch": 2.35,
|
| 1992 |
-
"learning_rate": 4.4117939750643295e-05,
|
| 1993 |
-
"loss": 1.1745,
|
| 1994 |
-
"step": 165500
|
| 1995 |
-
},
|
| 1996 |
-
{
|
| 1997 |
-
"epoch": 2.36,
|
| 1998 |
-
"learning_rate": 4.4100169175871826e-05,
|
| 1999 |
-
"loss": 1.167,
|
| 2000 |
-
"step": 166000
|
| 2001 |
-
},
|
| 2002 |
-
{
|
| 2003 |
-
"epoch": 2.37,
|
| 2004 |
-
"learning_rate": 4.408239860110036e-05,
|
| 2005 |
-
"loss": 1.1856,
|
| 2006 |
-
"step": 166500
|
| 2007 |
-
},
|
| 2008 |
-
{
|
| 2009 |
-
"epoch": 2.37,
|
| 2010 |
-
"learning_rate": 4.406462802632889e-05,
|
| 2011 |
-
"loss": 1.1859,
|
| 2012 |
-
"step": 167000
|
| 2013 |
-
},
|
| 2014 |
-
{
|
| 2015 |
-
"epoch": 2.38,
|
| 2016 |
-
"learning_rate": 4.4046857451557414e-05,
|
| 2017 |
-
"loss": 1.2023,
|
| 2018 |
-
"step": 167500
|
| 2019 |
-
},
|
| 2020 |
-
{
|
| 2021 |
-
"epoch": 2.39,
|
| 2022 |
-
"learning_rate": 4.4029086876785945e-05,
|
| 2023 |
-
"loss": 1.1877,
|
| 2024 |
-
"step": 168000
|
| 2025 |
-
},
|
| 2026 |
-
{
|
| 2027 |
-
"epoch": 2.4,
|
| 2028 |
-
"learning_rate": 4.4011316302014477e-05,
|
| 2029 |
-
"loss": 1.1764,
|
| 2030 |
-
"step": 168500
|
| 2031 |
-
},
|
| 2032 |
-
{
|
| 2033 |
-
"epoch": 2.4,
|
| 2034 |
-
"learning_rate": 4.3993545727243e-05,
|
| 2035 |
-
"loss": 1.1848,
|
| 2036 |
-
"step": 169000
|
| 2037 |
-
},
|
| 2038 |
-
{
|
| 2039 |
-
"epoch": 2.41,
|
| 2040 |
-
"learning_rate": 4.397577515247153e-05,
|
| 2041 |
-
"loss": 1.1762,
|
| 2042 |
-
"step": 169500
|
| 2043 |
-
},
|
| 2044 |
-
{
|
| 2045 |
-
"epoch": 2.42,
|
| 2046 |
-
"learning_rate": 4.3958004577700064e-05,
|
| 2047 |
-
"loss": 1.1806,
|
| 2048 |
-
"step": 170000
|
| 2049 |
-
},
|
| 2050 |
-
{
|
| 2051 |
-
"epoch": 2.42,
|
| 2052 |
-
"learning_rate": 4.3940234002928596e-05,
|
| 2053 |
-
"loss": 1.1808,
|
| 2054 |
-
"step": 170500
|
| 2055 |
-
},
|
| 2056 |
-
{
|
| 2057 |
-
"epoch": 2.43,
|
| 2058 |
-
"learning_rate": 4.392246342815712e-05,
|
| 2059 |
-
"loss": 1.1906,
|
| 2060 |
-
"step": 171000
|
| 2061 |
-
},
|
| 2062 |
-
{
|
| 2063 |
-
"epoch": 2.44,
|
| 2064 |
-
"learning_rate": 4.390469285338565e-05,
|
| 2065 |
-
"loss": 1.1898,
|
| 2066 |
-
"step": 171500
|
| 2067 |
-
},
|
| 2068 |
-
{
|
| 2069 |
-
"epoch": 2.45,
|
| 2070 |
-
"learning_rate": 4.388692227861418e-05,
|
| 2071 |
-
"loss": 1.1734,
|
| 2072 |
-
"step": 172000
|
| 2073 |
-
},
|
| 2074 |
-
{
|
| 2075 |
-
"epoch": 2.45,
|
| 2076 |
-
"learning_rate": 4.386915170384271e-05,
|
| 2077 |
-
"loss": 1.173,
|
| 2078 |
-
"step": 172500
|
| 2079 |
-
},
|
| 2080 |
-
{
|
| 2081 |
-
"epoch": 2.46,
|
| 2082 |
-
"learning_rate": 4.385138112907124e-05,
|
| 2083 |
-
"loss": 1.1701,
|
| 2084 |
-
"step": 173000
|
| 2085 |
-
},
|
| 2086 |
-
{
|
| 2087 |
-
"epoch": 2.47,
|
| 2088 |
-
"learning_rate": 4.383361055429977e-05,
|
| 2089 |
-
"loss": 1.1791,
|
| 2090 |
-
"step": 173500
|
| 2091 |
-
},
|
| 2092 |
-
{
|
| 2093 |
-
"epoch": 2.47,
|
| 2094 |
-
"learning_rate": 4.38158399795283e-05,
|
| 2095 |
-
"loss": 1.1675,
|
| 2096 |
-
"step": 174000
|
| 2097 |
-
},
|
| 2098 |
-
{
|
| 2099 |
-
"epoch": 2.48,
|
| 2100 |
-
"learning_rate": 4.3798069404756834e-05,
|
| 2101 |
-
"loss": 1.1716,
|
| 2102 |
-
"step": 174500
|
| 2103 |
-
},
|
| 2104 |
-
{
|
| 2105 |
-
"epoch": 2.49,
|
| 2106 |
-
"learning_rate": 4.378029882998536e-05,
|
| 2107 |
-
"loss": 1.1702,
|
| 2108 |
-
"step": 175000
|
| 2109 |
-
},
|
| 2110 |
-
{
|
| 2111 |
-
"epoch": 2.49,
|
| 2112 |
-
"learning_rate": 4.376252825521388e-05,
|
| 2113 |
-
"loss": 1.1848,
|
| 2114 |
-
"step": 175500
|
| 2115 |
-
},
|
| 2116 |
-
{
|
| 2117 |
-
"epoch": 2.5,
|
| 2118 |
-
"learning_rate": 4.3744757680442414e-05,
|
| 2119 |
-
"loss": 1.1749,
|
| 2120 |
-
"step": 176000
|
| 2121 |
-
},
|
| 2122 |
-
{
|
| 2123 |
-
"epoch": 2.51,
|
| 2124 |
-
"learning_rate": 4.3726987105670946e-05,
|
| 2125 |
-
"loss": 1.1859,
|
| 2126 |
-
"step": 176500
|
| 2127 |
-
},
|
| 2128 |
-
{
|
| 2129 |
-
"epoch": 2.52,
|
| 2130 |
-
"learning_rate": 4.370921653089948e-05,
|
| 2131 |
-
"loss": 1.1865,
|
| 2132 |
-
"step": 177000
|
| 2133 |
-
},
|
| 2134 |
-
{
|
| 2135 |
-
"epoch": 2.52,
|
| 2136 |
-
"learning_rate": 4.369144595612801e-05,
|
| 2137 |
-
"loss": 1.1831,
|
| 2138 |
-
"step": 177500
|
| 2139 |
-
},
|
| 2140 |
-
{
|
| 2141 |
-
"epoch": 2.53,
|
| 2142 |
-
"learning_rate": 4.367367538135654e-05,
|
| 2143 |
-
"loss": 1.1771,
|
| 2144 |
-
"step": 178000
|
| 2145 |
-
},
|
| 2146 |
-
{
|
| 2147 |
-
"epoch": 2.54,
|
| 2148 |
-
"learning_rate": 4.365590480658507e-05,
|
| 2149 |
-
"loss": 1.1927,
|
| 2150 |
-
"step": 178500
|
| 2151 |
-
},
|
| 2152 |
-
{
|
| 2153 |
-
"epoch": 2.54,
|
| 2154 |
-
"learning_rate": 4.3638134231813596e-05,
|
| 2155 |
-
"loss": 1.1855,
|
| 2156 |
-
"step": 179000
|
| 2157 |
-
},
|
| 2158 |
-
{
|
| 2159 |
-
"epoch": 2.55,
|
| 2160 |
-
"learning_rate": 4.362036365704212e-05,
|
| 2161 |
-
"loss": 1.184,
|
| 2162 |
-
"step": 179500
|
| 2163 |
-
},
|
| 2164 |
-
{
|
| 2165 |
-
"epoch": 2.56,
|
| 2166 |
-
"learning_rate": 4.360259308227065e-05,
|
| 2167 |
-
"loss": 1.1894,
|
| 2168 |
-
"step": 180000
|
| 2169 |
-
},
|
| 2170 |
-
{
|
| 2171 |
-
"epoch": 2.57,
|
| 2172 |
-
"learning_rate": 4.3584822507499184e-05,
|
| 2173 |
-
"loss": 1.1639,
|
| 2174 |
-
"step": 180500
|
| 2175 |
-
},
|
| 2176 |
-
{
|
| 2177 |
-
"epoch": 2.57,
|
| 2178 |
-
"learning_rate": 4.3567051932727715e-05,
|
| 2179 |
-
"loss": 1.1661,
|
| 2180 |
-
"step": 181000
|
| 2181 |
-
},
|
| 2182 |
-
{
|
| 2183 |
-
"epoch": 2.58,
|
| 2184 |
-
"learning_rate": 4.354928135795625e-05,
|
| 2185 |
-
"loss": 1.1678,
|
| 2186 |
-
"step": 181500
|
| 2187 |
-
},
|
| 2188 |
-
{
|
| 2189 |
-
"epoch": 2.59,
|
| 2190 |
-
"learning_rate": 4.353151078318478e-05,
|
| 2191 |
-
"loss": 1.1613,
|
| 2192 |
-
"step": 182000
|
| 2193 |
-
},
|
| 2194 |
-
{
|
| 2195 |
-
"epoch": 2.59,
|
| 2196 |
-
"learning_rate": 4.35137402084133e-05,
|
| 2197 |
-
"loss": 1.1801,
|
| 2198 |
-
"step": 182500
|
| 2199 |
-
},
|
| 2200 |
-
{
|
| 2201 |
-
"epoch": 2.6,
|
| 2202 |
-
"learning_rate": 4.349596963364183e-05,
|
| 2203 |
-
"loss": 1.1781,
|
| 2204 |
-
"step": 183000
|
| 2205 |
-
},
|
| 2206 |
-
{
|
| 2207 |
-
"epoch": 2.61,
|
| 2208 |
-
"learning_rate": 4.347819905887036e-05,
|
| 2209 |
-
"loss": 1.1781,
|
| 2210 |
-
"step": 183500
|
| 2211 |
-
},
|
| 2212 |
-
{
|
| 2213 |
-
"epoch": 2.62,
|
| 2214 |
-
"learning_rate": 4.346042848409889e-05,
|
| 2215 |
-
"loss": 1.1761,
|
| 2216 |
-
"step": 184000
|
| 2217 |
-
},
|
| 2218 |
-
{
|
| 2219 |
-
"epoch": 2.62,
|
| 2220 |
-
"learning_rate": 4.344265790932742e-05,
|
| 2221 |
-
"loss": 1.1848,
|
| 2222 |
-
"step": 184500
|
| 2223 |
-
},
|
| 2224 |
-
{
|
| 2225 |
-
"epoch": 2.63,
|
| 2226 |
-
"learning_rate": 4.342488733455595e-05,
|
| 2227 |
-
"loss": 1.1657,
|
| 2228 |
-
"step": 185000
|
| 2229 |
-
},
|
| 2230 |
-
{
|
| 2231 |
-
"epoch": 2.64,
|
| 2232 |
-
"learning_rate": 4.3407116759784485e-05,
|
| 2233 |
-
"loss": 1.1852,
|
| 2234 |
-
"step": 185500
|
| 2235 |
-
},
|
| 2236 |
-
{
|
| 2237 |
-
"epoch": 2.64,
|
| 2238 |
-
"learning_rate": 4.338934618501301e-05,
|
| 2239 |
-
"loss": 1.184,
|
| 2240 |
-
"step": 186000
|
| 2241 |
-
},
|
| 2242 |
-
{
|
| 2243 |
-
"epoch": 2.65,
|
| 2244 |
-
"learning_rate": 4.337157561024154e-05,
|
| 2245 |
-
"loss": 1.173,
|
| 2246 |
-
"step": 186500
|
| 2247 |
-
},
|
| 2248 |
-
{
|
| 2249 |
-
"epoch": 2.66,
|
| 2250 |
-
"learning_rate": 4.3353805035470066e-05,
|
| 2251 |
-
"loss": 1.1754,
|
| 2252 |
-
"step": 187000
|
| 2253 |
-
},
|
| 2254 |
-
{
|
| 2255 |
-
"epoch": 2.67,
|
| 2256 |
-
"learning_rate": 4.33360344606986e-05,
|
| 2257 |
-
"loss": 1.1889,
|
| 2258 |
-
"step": 187500
|
| 2259 |
-
},
|
| 2260 |
-
{
|
| 2261 |
-
"epoch": 2.67,
|
| 2262 |
-
"learning_rate": 4.331826388592713e-05,
|
| 2263 |
-
"loss": 1.1717,
|
| 2264 |
-
"step": 188000
|
| 2265 |
-
},
|
| 2266 |
-
{
|
| 2267 |
-
"epoch": 2.68,
|
| 2268 |
-
"learning_rate": 4.330049331115566e-05,
|
| 2269 |
-
"loss": 1.1783,
|
| 2270 |
-
"step": 188500
|
| 2271 |
-
},
|
| 2272 |
-
{
|
| 2273 |
-
"epoch": 2.69,
|
| 2274 |
-
"learning_rate": 4.328272273638419e-05,
|
| 2275 |
-
"loss": 1.175,
|
| 2276 |
-
"step": 189000
|
| 2277 |
-
},
|
| 2278 |
-
{
|
| 2279 |
-
"epoch": 2.69,
|
| 2280 |
-
"learning_rate": 4.3264952161612716e-05,
|
| 2281 |
-
"loss": 1.1831,
|
| 2282 |
-
"step": 189500
|
| 2283 |
-
},
|
| 2284 |
-
{
|
| 2285 |
-
"epoch": 2.7,
|
| 2286 |
-
"learning_rate": 4.324718158684125e-05,
|
| 2287 |
-
"loss": 1.1832,
|
| 2288 |
-
"step": 190000
|
| 2289 |
-
},
|
| 2290 |
-
{
|
| 2291 |
-
"epoch": 2.71,
|
| 2292 |
-
"learning_rate": 4.322941101206978e-05,
|
| 2293 |
-
"loss": 1.1814,
|
| 2294 |
-
"step": 190500
|
| 2295 |
-
},
|
| 2296 |
-
{
|
| 2297 |
-
"epoch": 2.72,
|
| 2298 |
-
"learning_rate": 4.3211640437298304e-05,
|
| 2299 |
-
"loss": 1.1685,
|
| 2300 |
-
"step": 191000
|
| 2301 |
-
},
|
| 2302 |
-
{
|
| 2303 |
-
"epoch": 2.72,
|
| 2304 |
-
"learning_rate": 4.3193869862526835e-05,
|
| 2305 |
-
"loss": 1.1755,
|
| 2306 |
-
"step": 191500
|
| 2307 |
-
},
|
| 2308 |
-
{
|
| 2309 |
-
"epoch": 2.73,
|
| 2310 |
-
"learning_rate": 4.3176099287755367e-05,
|
| 2311 |
-
"loss": 1.179,
|
| 2312 |
-
"step": 192000
|
| 2313 |
-
},
|
| 2314 |
-
{
|
| 2315 |
-
"epoch": 2.74,
|
| 2316 |
-
"learning_rate": 4.31583287129839e-05,
|
| 2317 |
-
"loss": 1.1853,
|
| 2318 |
-
"step": 192500
|
| 2319 |
-
},
|
| 2320 |
-
{
|
| 2321 |
-
"epoch": 2.74,
|
| 2322 |
-
"learning_rate": 4.314055813821242e-05,
|
| 2323 |
-
"loss": 1.1856,
|
| 2324 |
-
"step": 193000
|
| 2325 |
-
},
|
| 2326 |
-
{
|
| 2327 |
-
"epoch": 2.75,
|
| 2328 |
-
"learning_rate": 4.3122787563440954e-05,
|
| 2329 |
-
"loss": 1.1851,
|
| 2330 |
-
"step": 193500
|
| 2331 |
-
},
|
| 2332 |
-
{
|
| 2333 |
-
"epoch": 2.76,
|
| 2334 |
-
"learning_rate": 4.3105016988669486e-05,
|
| 2335 |
-
"loss": 1.173,
|
| 2336 |
-
"step": 194000
|
| 2337 |
-
},
|
| 2338 |
-
{
|
| 2339 |
-
"epoch": 2.77,
|
| 2340 |
-
"learning_rate": 4.308724641389802e-05,
|
| 2341 |
-
"loss": 1.1873,
|
| 2342 |
-
"step": 194500
|
| 2343 |
-
},
|
| 2344 |
-
{
|
| 2345 |
-
"epoch": 2.77,
|
| 2346 |
-
"learning_rate": 4.306947583912654e-05,
|
| 2347 |
-
"loss": 1.1886,
|
| 2348 |
-
"step": 195000
|
| 2349 |
-
},
|
| 2350 |
-
{
|
| 2351 |
-
"epoch": 2.78,
|
| 2352 |
-
"learning_rate": 4.305170526435507e-05,
|
| 2353 |
-
"loss": 1.1836,
|
| 2354 |
-
"step": 195500
|
| 2355 |
-
},
|
| 2356 |
-
{
|
| 2357 |
-
"epoch": 2.79,
|
| 2358 |
-
"learning_rate": 4.30339346895836e-05,
|
| 2359 |
-
"loss": 1.1723,
|
| 2360 |
-
"step": 196000
|
| 2361 |
-
},
|
| 2362 |
-
{
|
| 2363 |
-
"epoch": 2.79,
|
| 2364 |
-
"learning_rate": 4.301616411481213e-05,
|
| 2365 |
-
"loss": 1.1673,
|
| 2366 |
-
"step": 196500
|
| 2367 |
-
},
|
| 2368 |
-
{
|
| 2369 |
-
"epoch": 2.8,
|
| 2370 |
-
"learning_rate": 4.299839354004066e-05,
|
| 2371 |
-
"loss": 1.1673,
|
| 2372 |
-
"step": 197000
|
| 2373 |
-
},
|
| 2374 |
-
{
|
| 2375 |
-
"epoch": 2.81,
|
| 2376 |
-
"learning_rate": 4.298062296526919e-05,
|
| 2377 |
-
"loss": 1.1607,
|
| 2378 |
-
"step": 197500
|
| 2379 |
-
},
|
| 2380 |
-
{
|
| 2381 |
-
"epoch": 2.81,
|
| 2382 |
-
"learning_rate": 4.2962852390497724e-05,
|
| 2383 |
-
"loss": 1.1785,
|
| 2384 |
-
"step": 198000
|
| 2385 |
-
},
|
| 2386 |
-
{
|
| 2387 |
-
"epoch": 2.82,
|
| 2388 |
-
"learning_rate": 4.294508181572625e-05,
|
| 2389 |
-
"loss": 1.1817,
|
| 2390 |
-
"step": 198500
|
| 2391 |
-
},
|
| 2392 |
-
{
|
| 2393 |
-
"epoch": 2.83,
|
| 2394 |
-
"learning_rate": 4.292731124095478e-05,
|
| 2395 |
-
"loss": 1.1658,
|
| 2396 |
-
"step": 199000
|
| 2397 |
-
},
|
| 2398 |
-
{
|
| 2399 |
-
"epoch": 2.84,
|
| 2400 |
-
"learning_rate": 4.2909540666183304e-05,
|
| 2401 |
-
"loss": 1.1664,
|
| 2402 |
-
"step": 199500
|
| 2403 |
-
},
|
| 2404 |
-
{
|
| 2405 |
-
"epoch": 2.84,
|
| 2406 |
-
"learning_rate": 4.2891770091411836e-05,
|
| 2407 |
-
"loss": 1.1962,
|
| 2408 |
-
"step": 200000
|
| 2409 |
-
},
|
| 2410 |
-
{
|
| 2411 |
-
"epoch": 2.85,
|
| 2412 |
-
"learning_rate": 4.287399951664037e-05,
|
| 2413 |
-
"loss": 1.1754,
|
| 2414 |
-
"step": 200500
|
| 2415 |
-
},
|
| 2416 |
-
{
|
| 2417 |
-
"epoch": 2.86,
|
| 2418 |
-
"learning_rate": 4.28562289418689e-05,
|
| 2419 |
-
"loss": 1.1827,
|
| 2420 |
-
"step": 201000
|
| 2421 |
-
},
|
| 2422 |
-
{
|
| 2423 |
-
"epoch": 2.86,
|
| 2424 |
-
"learning_rate": 4.283845836709743e-05,
|
| 2425 |
-
"loss": 1.1919,
|
| 2426 |
-
"step": 201500
|
| 2427 |
-
},
|
| 2428 |
-
{
|
| 2429 |
-
"epoch": 2.87,
|
| 2430 |
-
"learning_rate": 4.282068779232596e-05,
|
| 2431 |
-
"loss": 1.1833,
|
| 2432 |
-
"step": 202000
|
| 2433 |
-
},
|
| 2434 |
-
{
|
| 2435 |
-
"epoch": 2.88,
|
| 2436 |
-
"learning_rate": 4.2802917217554486e-05,
|
| 2437 |
-
"loss": 1.187,
|
| 2438 |
-
"step": 202500
|
| 2439 |
-
},
|
| 2440 |
-
{
|
| 2441 |
-
"epoch": 2.89,
|
| 2442 |
-
"learning_rate": 4.278514664278301e-05,
|
| 2443 |
-
"loss": 1.1901,
|
| 2444 |
-
"step": 203000
|
| 2445 |
-
},
|
| 2446 |
-
{
|
| 2447 |
-
"epoch": 2.89,
|
| 2448 |
-
"learning_rate": 4.276737606801154e-05,
|
| 2449 |
-
"loss": 1.1696,
|
| 2450 |
-
"step": 203500
|
| 2451 |
-
},
|
| 2452 |
-
{
|
| 2453 |
-
"epoch": 2.9,
|
| 2454 |
-
"learning_rate": 4.2749605493240074e-05,
|
| 2455 |
-
"loss": 1.1834,
|
| 2456 |
-
"step": 204000
|
| 2457 |
-
},
|
| 2458 |
-
{
|
| 2459 |
-
"epoch": 2.91,
|
| 2460 |
-
"learning_rate": 4.2731834918468605e-05,
|
| 2461 |
-
"loss": 1.1886,
|
| 2462 |
-
"step": 204500
|
| 2463 |
-
},
|
| 2464 |
-
{
|
| 2465 |
-
"epoch": 2.91,
|
| 2466 |
-
"learning_rate": 4.271406434369714e-05,
|
| 2467 |
-
"loss": 1.1847,
|
| 2468 |
-
"step": 205000
|
| 2469 |
-
},
|
| 2470 |
-
{
|
| 2471 |
-
"epoch": 2.92,
|
| 2472 |
-
"learning_rate": 4.269629376892567e-05,
|
| 2473 |
-
"loss": 1.1744,
|
| 2474 |
-
"step": 205500
|
| 2475 |
-
},
|
| 2476 |
-
{
|
| 2477 |
-
"epoch": 2.93,
|
| 2478 |
-
"learning_rate": 4.26785231941542e-05,
|
| 2479 |
-
"loss": 1.1869,
|
| 2480 |
-
"step": 206000
|
| 2481 |
-
},
|
| 2482 |
-
{
|
| 2483 |
-
"epoch": 2.94,
|
| 2484 |
-
"learning_rate": 4.2660752619382724e-05,
|
| 2485 |
-
"loss": 1.1751,
|
| 2486 |
-
"step": 206500
|
| 2487 |
-
},
|
| 2488 |
-
{
|
| 2489 |
-
"epoch": 2.94,
|
| 2490 |
-
"learning_rate": 4.264298204461125e-05,
|
| 2491 |
-
"loss": 1.1695,
|
| 2492 |
-
"step": 207000
|
| 2493 |
},
|
| 2494 |
{
|
| 2495 |
-
"epoch":
|
| 2496 |
-
"
|
| 2497 |
-
"
|
| 2498 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2499 |
},
|
| 2500 |
{
|
| 2501 |
-
"epoch":
|
| 2502 |
-
"learning_rate": 4.
|
| 2503 |
-
"loss": 1.
|
| 2504 |
-
"step":
|
| 2505 |
},
|
| 2506 |
{
|
| 2507 |
-
"epoch": 2.
|
| 2508 |
-
"
|
| 2509 |
-
"
|
| 2510 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2511 |
},
|
| 2512 |
{
|
| 2513 |
"epoch": 2.97,
|
| 2514 |
-
"learning_rate": 4.
|
| 2515 |
-
"loss":
|
| 2516 |
-
"step":
|
| 2517 |
-
},
|
| 2518 |
-
{
|
| 2519 |
-
"epoch": 2.98,
|
| 2520 |
-
"learning_rate": 4.2554129170753906e-05,
|
| 2521 |
-
"loss": 1.1814,
|
| 2522 |
-
"step": 209500
|
| 2523 |
-
},
|
| 2524 |
-
{
|
| 2525 |
-
"epoch": 2.99,
|
| 2526 |
-
"learning_rate": 4.253635859598243e-05,
|
| 2527 |
-
"loss": 1.1796,
|
| 2528 |
-
"step": 210000
|
| 2529 |
-
},
|
| 2530 |
-
{
|
| 2531 |
-
"epoch": 2.99,
|
| 2532 |
-
"learning_rate": 4.2518588021210956e-05,
|
| 2533 |
-
"loss": 1.1751,
|
| 2534 |
-
"step": 210500
|
| 2535 |
},
|
| 2536 |
{
|
| 2537 |
"epoch": 3.0,
|
| 2538 |
-
"
|
| 2539 |
-
"
|
| 2540 |
-
"
|
| 2541 |
-
|
| 2542 |
-
|
| 2543 |
-
"
|
| 2544 |
-
"learning_rate": 4.248304687166802e-05,
|
| 2545 |
-
"loss": 1.0614,
|
| 2546 |
-
"step": 211500
|
| 2547 |
},
|
| 2548 |
{
|
| 2549 |
-
"epoch": 3.
|
| 2550 |
-
"learning_rate": 4.
|
| 2551 |
-
"loss":
|
| 2552 |
-
"step":
|
| 2553 |
-
},
|
| 2554 |
-
{
|
| 2555 |
-
"epoch": 3.02,
|
| 2556 |
-
"learning_rate": 4.244750572212508e-05,
|
| 2557 |
-
"loss": 1.0748,
|
| 2558 |
-
"step": 212500
|
| 2559 |
-
},
|
| 2560 |
-
{
|
| 2561 |
-
"epoch": 3.03,
|
| 2562 |
-
"learning_rate": 4.2429735147353606e-05,
|
| 2563 |
-
"loss": 1.0761,
|
| 2564 |
-
"step": 213000
|
| 2565 |
-
},
|
| 2566 |
-
{
|
| 2567 |
-
"epoch": 3.04,
|
| 2568 |
-
"learning_rate": 4.241196457258214e-05,
|
| 2569 |
-
"loss": 1.0821,
|
| 2570 |
-
"step": 213500
|
| 2571 |
-
},
|
| 2572 |
-
{
|
| 2573 |
-
"epoch": 3.04,
|
| 2574 |
-
"learning_rate": 4.239419399781067e-05,
|
| 2575 |
-
"loss": 1.0848,
|
| 2576 |
-
"step": 214000
|
| 2577 |
-
},
|
| 2578 |
-
{
|
| 2579 |
-
"epoch": 3.05,
|
| 2580 |
-
"learning_rate": 4.2376423423039194e-05,
|
| 2581 |
-
"loss": 1.1027,
|
| 2582 |
-
"step": 214500
|
| 2583 |
-
},
|
| 2584 |
-
{
|
| 2585 |
-
"epoch": 3.06,
|
| 2586 |
-
"learning_rate": 4.2358652848267725e-05,
|
| 2587 |
-
"loss": 1.1017,
|
| 2588 |
-
"step": 215000
|
| 2589 |
-
},
|
| 2590 |
-
{
|
| 2591 |
-
"epoch": 3.06,
|
| 2592 |
-
"learning_rate": 4.2340882273496256e-05,
|
| 2593 |
-
"loss": 1.078,
|
| 2594 |
-
"step": 215500
|
| 2595 |
-
},
|
| 2596 |
-
{
|
| 2597 |
-
"epoch": 3.07,
|
| 2598 |
-
"learning_rate": 4.232311169872479e-05,
|
| 2599 |
-
"loss": 1.0849,
|
| 2600 |
-
"step": 216000
|
| 2601 |
-
},
|
| 2602 |
-
{
|
| 2603 |
-
"epoch": 3.08,
|
| 2604 |
-
"learning_rate": 4.230534112395331e-05,
|
| 2605 |
-
"loss": 1.0903,
|
| 2606 |
-
"step": 216500
|
| 2607 |
-
},
|
| 2608 |
-
{
|
| 2609 |
-
"epoch": 3.08,
|
| 2610 |
-
"learning_rate": 4.2287570549181844e-05,
|
| 2611 |
-
"loss": 1.0793,
|
| 2612 |
-
"step": 217000
|
| 2613 |
-
},
|
| 2614 |
-
{
|
| 2615 |
-
"epoch": 3.09,
|
| 2616 |
-
"learning_rate": 4.2269799974410375e-05,
|
| 2617 |
-
"loss": 1.0953,
|
| 2618 |
-
"step": 217500
|
| 2619 |
-
},
|
| 2620 |
-
{
|
| 2621 |
-
"epoch": 3.1,
|
| 2622 |
-
"learning_rate": 4.225202939963891e-05,
|
| 2623 |
-
"loss": 1.0847,
|
| 2624 |
-
"step": 218000
|
| 2625 |
-
},
|
| 2626 |
-
{
|
| 2627 |
-
"epoch": 3.11,
|
| 2628 |
-
"learning_rate": 4.223425882486743e-05,
|
| 2629 |
-
"loss": 1.0795,
|
| 2630 |
-
"step": 218500
|
| 2631 |
-
},
|
| 2632 |
-
{
|
| 2633 |
-
"epoch": 3.11,
|
| 2634 |
-
"learning_rate": 4.221648825009596e-05,
|
| 2635 |
-
"loss": 1.0898,
|
| 2636 |
-
"step": 219000
|
| 2637 |
-
},
|
| 2638 |
-
{
|
| 2639 |
-
"epoch": 3.12,
|
| 2640 |
-
"learning_rate": 4.2198717675324494e-05,
|
| 2641 |
-
"loss": 1.0851,
|
| 2642 |
-
"step": 219500
|
| 2643 |
},
|
| 2644 |
{
|
| 2645 |
-
"epoch":
|
| 2646 |
-
"
|
| 2647 |
-
"
|
| 2648 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2649 |
},
|
| 2650 |
{
|
| 2651 |
-
"epoch":
|
| 2652 |
-
"learning_rate": 4.
|
| 2653 |
-
"loss":
|
| 2654 |
-
"step":
|
| 2655 |
},
|
| 2656 |
{
|
| 2657 |
-
"epoch":
|
| 2658 |
-
"
|
| 2659 |
-
"
|
| 2660 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2661 |
},
|
| 2662 |
{
|
| 2663 |
-
"epoch":
|
| 2664 |
-
"learning_rate": 4.
|
| 2665 |
-
"loss":
|
| 2666 |
-
"step":
|
| 2667 |
},
|
| 2668 |
{
|
| 2669 |
-
"epoch":
|
| 2670 |
-
"
|
| 2671 |
-
"
|
| 2672 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2673 |
},
|
| 2674 |
{
|
| 2675 |
-
"epoch":
|
| 2676 |
-
"learning_rate": 4.
|
| 2677 |
-
"loss":
|
| 2678 |
-
"step":
|
| 2679 |
},
|
| 2680 |
{
|
| 2681 |
-
"epoch":
|
| 2682 |
-
"
|
| 2683 |
-
"
|
| 2684 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2685 |
},
|
| 2686 |
{
|
| 2687 |
-
"epoch":
|
| 2688 |
-
"learning_rate": 4.
|
| 2689 |
-
"loss":
|
| 2690 |
-
"step":
|
| 2691 |
},
|
| 2692 |
{
|
| 2693 |
-
"epoch":
|
| 2694 |
-
"
|
| 2695 |
-
"
|
| 2696 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2697 |
},
|
| 2698 |
{
|
| 2699 |
-
"epoch":
|
| 2700 |
-
"learning_rate": 4.
|
| 2701 |
-
"loss":
|
| 2702 |
-
"step":
|
| 2703 |
},
|
| 2704 |
{
|
| 2705 |
-
"epoch":
|
| 2706 |
-
"
|
| 2707 |
-
"
|
| 2708 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2709 |
},
|
| 2710 |
{
|
| 2711 |
-
"epoch":
|
| 2712 |
-
"learning_rate": 4.
|
| 2713 |
-
"loss":
|
| 2714 |
-
"step":
|
| 2715 |
},
|
| 2716 |
{
|
| 2717 |
-
"epoch":
|
| 2718 |
-
"
|
| 2719 |
-
"
|
| 2720 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2721 |
},
|
| 2722 |
{
|
| 2723 |
-
"epoch":
|
| 2724 |
-
"learning_rate": 4.
|
| 2725 |
-
"loss":
|
| 2726 |
-
"step":
|
| 2727 |
},
|
| 2728 |
{
|
| 2729 |
-
"epoch":
|
| 2730 |
-
"
|
| 2731 |
-
"
|
| 2732 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2733 |
},
|
| 2734 |
{
|
| 2735 |
-
"epoch":
|
| 2736 |
-
"learning_rate": 4.
|
| 2737 |
-
"loss":
|
| 2738 |
-
"step":
|
| 2739 |
},
|
| 2740 |
{
|
| 2741 |
-
"epoch":
|
| 2742 |
-
"
|
| 2743 |
-
"
|
| 2744 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2745 |
},
|
| 2746 |
{
|
| 2747 |
-
"epoch":
|
| 2748 |
-
"learning_rate": 4.
|
| 2749 |
-
"loss":
|
| 2750 |
-
"step":
|
| 2751 |
},
|
| 2752 |
{
|
| 2753 |
-
"epoch":
|
| 2754 |
-
"
|
| 2755 |
-
"
|
| 2756 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2757 |
},
|
| 2758 |
{
|
| 2759 |
-
"epoch":
|
| 2760 |
-
"learning_rate":
|
| 2761 |
-
"loss":
|
| 2762 |
-
"step":
|
| 2763 |
},
|
| 2764 |
{
|
| 2765 |
-
"epoch":
|
| 2766 |
-
"
|
| 2767 |
-
"
|
| 2768 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2769 |
},
|
| 2770 |
{
|
| 2771 |
-
"epoch":
|
| 2772 |
-
"learning_rate":
|
| 2773 |
-
"loss":
|
| 2774 |
-
"step":
|
| 2775 |
},
|
| 2776 |
{
|
| 2777 |
-
"epoch":
|
| 2778 |
-
"
|
| 2779 |
-
"
|
| 2780 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2781 |
},
|
| 2782 |
{
|
| 2783 |
-
"epoch":
|
| 2784 |
-
"learning_rate":
|
| 2785 |
-
"loss":
|
| 2786 |
-
"step":
|
| 2787 |
},
|
| 2788 |
{
|
| 2789 |
-
"epoch":
|
| 2790 |
-
"
|
| 2791 |
-
"
|
| 2792 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2793 |
},
|
| 2794 |
{
|
| 2795 |
-
"epoch":
|
| 2796 |
-
"learning_rate":
|
| 2797 |
-
"loss":
|
| 2798 |
-
"step":
|
| 2799 |
},
|
| 2800 |
{
|
| 2801 |
-
"epoch":
|
| 2802 |
-
"
|
| 2803 |
-
"
|
| 2804 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2805 |
},
|
| 2806 |
{
|
| 2807 |
-
"epoch":
|
| 2808 |
-
"learning_rate":
|
| 2809 |
-
"loss":
|
| 2810 |
-
"step":
|
| 2811 |
},
|
| 2812 |
{
|
| 2813 |
-
"epoch":
|
| 2814 |
-
"
|
| 2815 |
-
"
|
| 2816 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2817 |
},
|
| 2818 |
{
|
| 2819 |
-
"epoch":
|
| 2820 |
-
"learning_rate":
|
| 2821 |
-
"loss":
|
| 2822 |
-
"step":
|
| 2823 |
},
|
| 2824 |
{
|
| 2825 |
-
"epoch":
|
| 2826 |
-
"
|
| 2827 |
-
"
|
| 2828 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2829 |
},
|
| 2830 |
{
|
| 2831 |
-
"epoch":
|
| 2832 |
-
"learning_rate":
|
| 2833 |
-
"loss":
|
| 2834 |
-
"step":
|
| 2835 |
},
|
| 2836 |
{
|
| 2837 |
-
"epoch":
|
| 2838 |
-
"
|
| 2839 |
-
"
|
| 2840 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2841 |
},
|
| 2842 |
{
|
| 2843 |
-
"epoch":
|
| 2844 |
-
"learning_rate":
|
| 2845 |
-
"loss":
|
| 2846 |
-
"step":
|
| 2847 |
},
|
| 2848 |
{
|
| 2849 |
-
"epoch":
|
| 2850 |
-
"
|
| 2851 |
-
"
|
| 2852 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2853 |
},
|
| 2854 |
{
|
| 2855 |
-
"epoch":
|
| 2856 |
-
"learning_rate":
|
| 2857 |
-
"loss":
|
| 2858 |
-
"step":
|
| 2859 |
},
|
| 2860 |
{
|
| 2861 |
-
"epoch":
|
| 2862 |
-
"
|
| 2863 |
-
"
|
| 2864 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2865 |
},
|
| 2866 |
{
|
| 2867 |
-
"epoch":
|
| 2868 |
-
"learning_rate":
|
| 2869 |
-
"loss":
|
| 2870 |
-
"step":
|
| 2871 |
},
|
| 2872 |
{
|
| 2873 |
-
"epoch":
|
| 2874 |
-
"
|
| 2875 |
-
"
|
| 2876 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2877 |
},
|
| 2878 |
{
|
| 2879 |
-
"epoch":
|
| 2880 |
-
"learning_rate":
|
| 2881 |
-
"loss":
|
| 2882 |
-
"step":
|
| 2883 |
},
|
| 2884 |
{
|
| 2885 |
-
"epoch":
|
| 2886 |
-
"
|
| 2887 |
-
"
|
| 2888 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2889 |
},
|
| 2890 |
{
|
| 2891 |
-
"epoch":
|
| 2892 |
-
"learning_rate":
|
| 2893 |
-
"loss":
|
| 2894 |
-
"step":
|
| 2895 |
},
|
| 2896 |
{
|
| 2897 |
-
"epoch":
|
| 2898 |
-
"
|
| 2899 |
-
"
|
| 2900 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2901 |
},
|
| 2902 |
{
|
| 2903 |
-
"epoch":
|
| 2904 |
-
"learning_rate":
|
| 2905 |
-
"loss":
|
| 2906 |
-
"step":
|
| 2907 |
},
|
| 2908 |
{
|
| 2909 |
-
"epoch":
|
| 2910 |
-
"
|
| 2911 |
-
"
|
| 2912 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2913 |
},
|
| 2914 |
{
|
| 2915 |
-
"epoch":
|
| 2916 |
-
"learning_rate":
|
| 2917 |
-
"loss":
|
| 2918 |
-
"step":
|
| 2919 |
},
|
| 2920 |
{
|
| 2921 |
-
"epoch":
|
| 2922 |
-
"
|
| 2923 |
-
"
|
| 2924 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2925 |
},
|
| 2926 |
{
|
| 2927 |
-
"epoch":
|
| 2928 |
-
"learning_rate":
|
| 2929 |
-
"loss":
|
| 2930 |
-
"step":
|
| 2931 |
},
|
| 2932 |
{
|
| 2933 |
-
"epoch":
|
| 2934 |
-
"
|
| 2935 |
-
"
|
| 2936 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2937 |
},
|
| 2938 |
{
|
| 2939 |
-
"epoch":
|
| 2940 |
-
"learning_rate":
|
| 2941 |
-
"loss":
|
| 2942 |
-
"step":
|
| 2943 |
},
|
| 2944 |
{
|
| 2945 |
-
"epoch":
|
| 2946 |
-
"
|
| 2947 |
-
"
|
| 2948 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2949 |
},
|
| 2950 |
{
|
| 2951 |
-
"epoch":
|
| 2952 |
-
"learning_rate":
|
| 2953 |
-
"loss":
|
| 2954 |
-
"step":
|
| 2955 |
},
|
| 2956 |
{
|
| 2957 |
-
"epoch":
|
| 2958 |
-
"
|
| 2959 |
-
"
|
| 2960 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2961 |
},
|
| 2962 |
{
|
| 2963 |
-
"epoch":
|
| 2964 |
-
"learning_rate":
|
| 2965 |
-
"loss":
|
| 2966 |
-
"step":
|
| 2967 |
},
|
| 2968 |
{
|
| 2969 |
-
"epoch":
|
| 2970 |
-
"
|
| 2971 |
-
"
|
| 2972 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2973 |
},
|
| 2974 |
{
|
| 2975 |
-
"epoch":
|
| 2976 |
-
"learning_rate":
|
| 2977 |
-
"loss":
|
| 2978 |
-
"step":
|
| 2979 |
},
|
| 2980 |
{
|
| 2981 |
-
"epoch":
|
| 2982 |
-
"
|
| 2983 |
-
"
|
| 2984 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2985 |
},
|
| 2986 |
{
|
| 2987 |
-
"epoch":
|
| 2988 |
-
"learning_rate":
|
| 2989 |
-
"loss":
|
| 2990 |
-
"step":
|
| 2991 |
},
|
| 2992 |
{
|
| 2993 |
-
"epoch":
|
| 2994 |
-
"
|
| 2995 |
-
"
|
| 2996 |
-
"
|
|
|
|
|
|
|
|
|
|
| 2997 |
},
|
| 2998 |
{
|
| 2999 |
-
"epoch":
|
| 3000 |
-
"learning_rate":
|
| 3001 |
-
"loss":
|
| 3002 |
-
"step":
|
| 3003 |
},
|
| 3004 |
{
|
| 3005 |
-
"epoch":
|
| 3006 |
-
"
|
| 3007 |
-
"
|
| 3008 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3009 |
},
|
| 3010 |
{
|
| 3011 |
-
"epoch":
|
| 3012 |
-
"learning_rate":
|
| 3013 |
-
"loss":
|
| 3014 |
-
"step":
|
| 3015 |
},
|
| 3016 |
{
|
| 3017 |
-
"epoch":
|
| 3018 |
-
"
|
| 3019 |
-
"
|
| 3020 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3021 |
},
|
| 3022 |
{
|
| 3023 |
-
"epoch":
|
| 3024 |
-
"learning_rate":
|
| 3025 |
-
"loss":
|
| 3026 |
-
"step":
|
| 3027 |
},
|
| 3028 |
{
|
| 3029 |
-
"epoch":
|
| 3030 |
-
"
|
| 3031 |
-
"
|
| 3032 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3033 |
},
|
| 3034 |
{
|
| 3035 |
-
"epoch":
|
| 3036 |
-
"learning_rate":
|
| 3037 |
-
"loss":
|
| 3038 |
-
"step":
|
| 3039 |
},
|
| 3040 |
{
|
| 3041 |
-
"epoch":
|
| 3042 |
-
"
|
| 3043 |
-
"
|
| 3044 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3045 |
},
|
| 3046 |
{
|
| 3047 |
-
"epoch":
|
| 3048 |
-
"learning_rate":
|
| 3049 |
-
"loss":
|
| 3050 |
-
"step":
|
| 3051 |
},
|
| 3052 |
{
|
| 3053 |
-
"epoch":
|
| 3054 |
-
"
|
| 3055 |
-
"
|
| 3056 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3057 |
},
|
| 3058 |
{
|
| 3059 |
-
"epoch":
|
| 3060 |
-
"learning_rate":
|
| 3061 |
-
"loss":
|
| 3062 |
-
"step":
|
| 3063 |
},
|
| 3064 |
{
|
| 3065 |
-
"epoch":
|
| 3066 |
-
"
|
| 3067 |
-
"
|
| 3068 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3069 |
},
|
| 3070 |
{
|
| 3071 |
-
"epoch":
|
| 3072 |
-
"learning_rate":
|
| 3073 |
-
"loss":
|
| 3074 |
-
"step":
|
| 3075 |
},
|
| 3076 |
{
|
| 3077 |
-
"epoch":
|
| 3078 |
-
"
|
| 3079 |
-
"
|
| 3080 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3081 |
},
|
| 3082 |
{
|
| 3083 |
-
"epoch":
|
| 3084 |
-
"learning_rate":
|
| 3085 |
-
"loss":
|
| 3086 |
-
"step":
|
| 3087 |
},
|
| 3088 |
{
|
| 3089 |
-
"epoch":
|
| 3090 |
-
"
|
| 3091 |
-
"
|
| 3092 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3093 |
},
|
| 3094 |
{
|
| 3095 |
-
"epoch":
|
| 3096 |
-
"learning_rate":
|
| 3097 |
-
"loss":
|
| 3098 |
-
"step":
|
| 3099 |
},
|
| 3100 |
{
|
| 3101 |
-
"epoch":
|
| 3102 |
-
"
|
| 3103 |
-
"
|
| 3104 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3105 |
},
|
| 3106 |
{
|
| 3107 |
-
"epoch":
|
| 3108 |
-
"learning_rate":
|
| 3109 |
-
"loss":
|
| 3110 |
-
"step":
|
| 3111 |
},
|
| 3112 |
{
|
| 3113 |
-
"epoch":
|
| 3114 |
-
"
|
| 3115 |
-
"
|
| 3116 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3117 |
},
|
| 3118 |
{
|
| 3119 |
-
"epoch":
|
| 3120 |
-
"learning_rate":
|
| 3121 |
-
"loss":
|
| 3122 |
-
"step":
|
| 3123 |
},
|
| 3124 |
{
|
| 3125 |
-
"epoch":
|
| 3126 |
-
"
|
| 3127 |
-
"
|
| 3128 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3129 |
},
|
| 3130 |
{
|
| 3131 |
-
"epoch":
|
| 3132 |
-
"learning_rate":
|
| 3133 |
-
"loss":
|
| 3134 |
-
"step":
|
| 3135 |
},
|
| 3136 |
{
|
| 3137 |
-
"epoch":
|
| 3138 |
-
"
|
| 3139 |
-
"
|
| 3140 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3141 |
},
|
| 3142 |
{
|
| 3143 |
-
"epoch":
|
| 3144 |
-
"learning_rate":
|
| 3145 |
-
"loss":
|
| 3146 |
-
"step":
|
| 3147 |
},
|
| 3148 |
{
|
| 3149 |
-
"epoch":
|
| 3150 |
-
"
|
| 3151 |
-
"
|
| 3152 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3153 |
},
|
| 3154 |
{
|
| 3155 |
-
"epoch":
|
| 3156 |
-
"learning_rate":
|
| 3157 |
-
"loss":
|
| 3158 |
-
"step":
|
| 3159 |
},
|
| 3160 |
{
|
| 3161 |
-
"epoch":
|
| 3162 |
-
"
|
| 3163 |
-
"
|
| 3164 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3165 |
},
|
| 3166 |
{
|
| 3167 |
-
"epoch":
|
| 3168 |
-
"learning_rate":
|
| 3169 |
-
"loss":
|
| 3170 |
-
"step":
|
| 3171 |
},
|
| 3172 |
{
|
| 3173 |
-
"epoch":
|
| 3174 |
-
"
|
| 3175 |
-
"
|
| 3176 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3177 |
},
|
| 3178 |
{
|
| 3179 |
-
"epoch":
|
| 3180 |
-
"learning_rate":
|
| 3181 |
-
"loss":
|
| 3182 |
-
"step":
|
| 3183 |
},
|
| 3184 |
{
|
| 3185 |
-
"epoch":
|
| 3186 |
-
"
|
| 3187 |
-
"
|
| 3188 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3189 |
},
|
| 3190 |
{
|
| 3191 |
-
"epoch":
|
| 3192 |
-
"learning_rate":
|
| 3193 |
-
"loss":
|
| 3194 |
-
"step":
|
| 3195 |
},
|
| 3196 |
{
|
| 3197 |
-
"epoch":
|
| 3198 |
-
"
|
| 3199 |
-
"
|
| 3200 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3201 |
},
|
| 3202 |
{
|
| 3203 |
-
"epoch":
|
| 3204 |
-
"learning_rate":
|
| 3205 |
-
"loss":
|
| 3206 |
-
"step":
|
| 3207 |
},
|
| 3208 |
{
|
| 3209 |
-
"epoch":
|
| 3210 |
-
"
|
| 3211 |
-
"
|
| 3212 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3213 |
},
|
| 3214 |
{
|
| 3215 |
-
"epoch":
|
| 3216 |
-
"learning_rate":
|
| 3217 |
-
"loss":
|
| 3218 |
-
"step":
|
| 3219 |
},
|
| 3220 |
{
|
| 3221 |
-
"epoch":
|
| 3222 |
-
"
|
| 3223 |
-
"
|
| 3224 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3225 |
},
|
| 3226 |
{
|
| 3227 |
-
"epoch":
|
| 3228 |
-
"learning_rate":
|
| 3229 |
-
"loss":
|
| 3230 |
-
"step":
|
| 3231 |
},
|
| 3232 |
{
|
| 3233 |
-
"epoch":
|
| 3234 |
-
"
|
| 3235 |
-
"
|
| 3236 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3237 |
},
|
| 3238 |
{
|
| 3239 |
-
"epoch":
|
| 3240 |
-
"learning_rate":
|
| 3241 |
-
"loss":
|
| 3242 |
-
"step":
|
| 3243 |
},
|
| 3244 |
{
|
| 3245 |
-
"epoch":
|
| 3246 |
-
"
|
| 3247 |
-
"
|
| 3248 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3249 |
},
|
| 3250 |
{
|
| 3251 |
-
"epoch":
|
| 3252 |
-
"learning_rate":
|
| 3253 |
-
"loss":
|
| 3254 |
-
"step":
|
| 3255 |
},
|
| 3256 |
{
|
| 3257 |
-
"epoch":
|
| 3258 |
-
"
|
| 3259 |
-
"
|
| 3260 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3261 |
},
|
| 3262 |
{
|
| 3263 |
-
"epoch":
|
| 3264 |
-
"learning_rate":
|
| 3265 |
-
"loss":
|
| 3266 |
-
"step":
|
| 3267 |
},
|
| 3268 |
{
|
| 3269 |
-
"epoch":
|
| 3270 |
-
"
|
| 3271 |
-
"
|
| 3272 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3273 |
},
|
| 3274 |
{
|
| 3275 |
-
"epoch":
|
| 3276 |
-
"learning_rate":
|
| 3277 |
-
"loss":
|
| 3278 |
-
"step":
|
| 3279 |
},
|
| 3280 |
{
|
| 3281 |
-
"epoch":
|
| 3282 |
-
"
|
| 3283 |
-
"
|
| 3284 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3285 |
},
|
| 3286 |
{
|
| 3287 |
-
"epoch":
|
| 3288 |
-
"learning_rate":
|
| 3289 |
-
"loss":
|
| 3290 |
-
"step":
|
| 3291 |
},
|
| 3292 |
{
|
| 3293 |
-
"epoch":
|
| 3294 |
-
"
|
| 3295 |
-
"
|
| 3296 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3297 |
},
|
| 3298 |
{
|
| 3299 |
-
"epoch":
|
| 3300 |
-
"learning_rate":
|
| 3301 |
-
"loss":
|
| 3302 |
-
"step":
|
| 3303 |
},
|
| 3304 |
{
|
| 3305 |
-
"epoch":
|
| 3306 |
-
"
|
| 3307 |
-
"
|
| 3308 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3309 |
},
|
| 3310 |
{
|
| 3311 |
-
"epoch":
|
| 3312 |
-
"learning_rate": 4.
|
| 3313 |
-
"loss":
|
| 3314 |
-
"step":
|
| 3315 |
},
|
| 3316 |
{
|
| 3317 |
-
"epoch":
|
| 3318 |
-
"
|
| 3319 |
-
"
|
| 3320 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3321 |
},
|
| 3322 |
{
|
| 3323 |
-
"epoch":
|
| 3324 |
-
"learning_rate":
|
| 3325 |
-
"loss":
|
| 3326 |
-
"step":
|
| 3327 |
},
|
| 3328 |
{
|
| 3329 |
-
"epoch":
|
| 3330 |
-
"
|
| 3331 |
-
"
|
| 3332 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3333 |
},
|
| 3334 |
{
|
| 3335 |
-
"epoch":
|
| 3336 |
-
"learning_rate":
|
| 3337 |
-
"loss":
|
| 3338 |
-
"step":
|
| 3339 |
},
|
| 3340 |
{
|
| 3341 |
-
"epoch":
|
| 3342 |
-
"
|
| 3343 |
-
"
|
| 3344 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3345 |
},
|
| 3346 |
{
|
| 3347 |
-
"epoch":
|
| 3348 |
-
"learning_rate":
|
| 3349 |
-
"loss":
|
| 3350 |
-
"step":
|
| 3351 |
},
|
| 3352 |
{
|
| 3353 |
-
"epoch":
|
| 3354 |
-
"
|
| 3355 |
-
"
|
| 3356 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3357 |
},
|
| 3358 |
{
|
| 3359 |
-
"epoch":
|
| 3360 |
-
"learning_rate":
|
| 3361 |
-
"loss":
|
| 3362 |
-
"step":
|
| 3363 |
},
|
| 3364 |
{
|
| 3365 |
-
"epoch":
|
| 3366 |
-
"
|
| 3367 |
-
"
|
| 3368 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3369 |
},
|
| 3370 |
{
|
| 3371 |
-
"epoch":
|
| 3372 |
-
"learning_rate":
|
| 3373 |
-
"loss":
|
| 3374 |
-
"step":
|
| 3375 |
},
|
| 3376 |
{
|
| 3377 |
-
"epoch":
|
| 3378 |
-
"
|
| 3379 |
-
"
|
| 3380 |
-
"
|
|
|
|
|
|
|
|
|
|
| 3381 |
}
|
| 3382 |
],
|
| 3383 |
-
"max_steps":
|
| 3384 |
-
"num_train_epochs":
|
| 3385 |
-
"total_flos": 1.
|
| 3386 |
"trial_name": null,
|
| 3387 |
"trial_params": null
|
| 3388 |
}
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 65.0,
|
| 5 |
+
"global_step": 6760,
|
| 6 |
"is_hyper_param_search": false,
|
| 7 |
"is_local_process_zero": true,
|
| 8 |
"is_world_process_zero": true,
|
| 9 |
"log_history": [
|
| 10 |
{
|
| 11 |
+
"epoch": 0.99,
|
| 12 |
+
"learning_rate": 4.9238165680473374e-05,
|
| 13 |
+
"loss": 1.4565,
|
| 14 |
+
"step": 103
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 15 |
},
|
| 16 |
{
|
| 17 |
+
"epoch": 1.0,
|
| 18 |
+
"eval_f1": 0.5061315988298739,
|
| 19 |
+
"eval_loss": 1.226438045501709,
|
| 20 |
+
"eval_runtime": 2.9031,
|
| 21 |
+
"eval_samples_per_second": 190.487,
|
| 22 |
+
"eval_steps_per_second": 12.056,
|
| 23 |
+
"step": 104
|
| 24 |
},
|
| 25 |
{
|
| 26 |
+
"epoch": 1.98,
|
| 27 |
+
"learning_rate": 4.8483727810650895e-05,
|
| 28 |
+
"loss": 1.08,
|
| 29 |
+
"step": 206
|
| 30 |
},
|
| 31 |
{
|
| 32 |
+
"epoch": 2.0,
|
| 33 |
+
"eval_f1": 0.5585235698125446,
|
| 34 |
+
"eval_loss": 1.1974396705627441,
|
| 35 |
+
"eval_runtime": 2.9416,
|
| 36 |
+
"eval_samples_per_second": 187.991,
|
| 37 |
+
"eval_steps_per_second": 11.898,
|
| 38 |
+
"step": 208
|
| 39 |
},
|
| 40 |
{
|
| 41 |
"epoch": 2.97,
|
| 42 |
+
"learning_rate": 4.772189349112427e-05,
|
| 43 |
+
"loss": 0.8073,
|
| 44 |
+
"step": 309
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 45 |
},
|
| 46 |
{
|
| 47 |
"epoch": 3.0,
|
| 48 |
+
"eval_f1": 0.5539323613883379,
|
| 49 |
+
"eval_loss": 1.276489496231079,
|
| 50 |
+
"eval_runtime": 2.969,
|
| 51 |
+
"eval_samples_per_second": 186.259,
|
| 52 |
+
"eval_steps_per_second": 11.789,
|
| 53 |
+
"step": 312
|
|
|
|
|
|
|
|
|
|
| 54 |
},
|
| 55 |
{
|
| 56 |
+
"epoch": 3.96,
|
| 57 |
+
"learning_rate": 4.696005917159764e-05,
|
| 58 |
+
"loss": 0.5577,
|
| 59 |
+
"step": 412
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 60 |
},
|
| 61 |
{
|
| 62 |
+
"epoch": 4.0,
|
| 63 |
+
"eval_f1": 0.5553847052045314,
|
| 64 |
+
"eval_loss": 1.427822232246399,
|
| 65 |
+
"eval_runtime": 2.9361,
|
| 66 |
+
"eval_samples_per_second": 188.346,
|
| 67 |
+
"eval_steps_per_second": 11.921,
|
| 68 |
+
"step": 416
|
| 69 |
},
|
| 70 |
{
|
| 71 |
+
"epoch": 4.95,
|
| 72 |
+
"learning_rate": 4.619822485207101e-05,
|
| 73 |
+
"loss": 0.3941,
|
| 74 |
+
"step": 515
|
| 75 |
},
|
| 76 |
{
|
| 77 |
+
"epoch": 5.0,
|
| 78 |
+
"eval_f1": 0.5570342860609194,
|
| 79 |
+
"eval_loss": 1.6517128944396973,
|
| 80 |
+
"eval_runtime": 2.9022,
|
| 81 |
+
"eval_samples_per_second": 190.542,
|
| 82 |
+
"eval_steps_per_second": 12.06,
|
| 83 |
+
"step": 520
|
| 84 |
},
|
| 85 |
{
|
| 86 |
+
"epoch": 5.94,
|
| 87 |
+
"learning_rate": 4.543639053254438e-05,
|
| 88 |
+
"loss": 0.2878,
|
| 89 |
+
"step": 618
|
| 90 |
},
|
| 91 |
{
|
| 92 |
+
"epoch": 6.0,
|
| 93 |
+
"eval_f1": 0.5619826716090497,
|
| 94 |
+
"eval_loss": 1.8180437088012695,
|
| 95 |
+
"eval_runtime": 2.855,
|
| 96 |
+
"eval_samples_per_second": 193.697,
|
| 97 |
+
"eval_steps_per_second": 12.259,
|
| 98 |
+
"step": 624
|
| 99 |
},
|
| 100 |
{
|
| 101 |
+
"epoch": 6.93,
|
| 102 |
+
"learning_rate": 4.468195266272189e-05,
|
| 103 |
+
"loss": 0.2337,
|
| 104 |
+
"step": 721
|
| 105 |
},
|
| 106 |
{
|
| 107 |
+
"epoch": 7.0,
|
| 108 |
+
"eval_f1": 0.5674708526030706,
|
| 109 |
+
"eval_loss": 1.9061989784240723,
|
| 110 |
+
"eval_runtime": 2.8641,
|
| 111 |
+
"eval_samples_per_second": 193.077,
|
| 112 |
+
"eval_steps_per_second": 12.22,
|
| 113 |
+
"step": 728
|
| 114 |
},
|
| 115 |
{
|
| 116 |
+
"epoch": 7.92,
|
| 117 |
+
"learning_rate": 4.392011834319526e-05,
|
| 118 |
+
"loss": 0.1743,
|
| 119 |
+
"step": 824
|
| 120 |
},
|
| 121 |
{
|
| 122 |
+
"epoch": 8.0,
|
| 123 |
+
"eval_f1": 0.5571774381839604,
|
| 124 |
+
"eval_loss": 2.166078805923462,
|
| 125 |
+
"eval_runtime": 2.8885,
|
| 126 |
+
"eval_samples_per_second": 191.449,
|
| 127 |
+
"eval_steps_per_second": 12.117,
|
| 128 |
+
"step": 832
|
| 129 |
},
|
| 130 |
{
|
| 131 |
+
"epoch": 8.91,
|
| 132 |
+
"learning_rate": 4.315828402366864e-05,
|
| 133 |
+
"loss": 0.1324,
|
| 134 |
+
"step": 927
|
| 135 |
},
|
| 136 |
{
|
| 137 |
+
"epoch": 9.0,
|
| 138 |
+
"eval_f1": 0.5586999359656486,
|
| 139 |
+
"eval_loss": 2.1434192657470703,
|
| 140 |
+
"eval_runtime": 2.8793,
|
| 141 |
+
"eval_samples_per_second": 192.062,
|
| 142 |
+
"eval_steps_per_second": 12.156,
|
| 143 |
+
"step": 936
|
| 144 |
},
|
| 145 |
{
|
| 146 |
+
"epoch": 9.9,
|
| 147 |
+
"learning_rate": 4.239644970414201e-05,
|
| 148 |
+
"loss": 0.1051,
|
| 149 |
+
"step": 1030
|
| 150 |
},
|
| 151 |
{
|
| 152 |
+
"epoch": 10.0,
|
| 153 |
+
"eval_f1": 0.571861247626083,
|
| 154 |
+
"eval_loss": 2.2514231204986572,
|
| 155 |
+
"eval_runtime": 2.8876,
|
| 156 |
+
"eval_samples_per_second": 191.511,
|
| 157 |
+
"eval_steps_per_second": 12.121,
|
| 158 |
+
"step": 1040
|
| 159 |
},
|
| 160 |
{
|
| 161 |
+
"epoch": 10.89,
|
| 162 |
+
"learning_rate": 4.163461538461539e-05,
|
| 163 |
+
"loss": 0.1016,
|
| 164 |
+
"step": 1133
|
| 165 |
},
|
| 166 |
{
|
| 167 |
+
"epoch": 11.0,
|
| 168 |
+
"eval_f1": 0.5608736700927537,
|
| 169 |
+
"eval_loss": 2.452277898788452,
|
| 170 |
+
"eval_runtime": 2.9662,
|
| 171 |
+
"eval_samples_per_second": 186.434,
|
| 172 |
+
"eval_steps_per_second": 11.8,
|
| 173 |
+
"step": 1144
|
| 174 |
},
|
| 175 |
{
|
| 176 |
+
"epoch": 11.88,
|
| 177 |
+
"learning_rate": 4.0872781065088764e-05,
|
| 178 |
+
"loss": 0.0814,
|
| 179 |
+
"step": 1236
|
| 180 |
},
|
| 181 |
{
|
| 182 |
+
"epoch": 12.0,
|
| 183 |
+
"eval_f1": 0.5643677851728315,
|
| 184 |
+
"eval_loss": 2.5340888500213623,
|
| 185 |
+
"eval_runtime": 2.8457,
|
| 186 |
+
"eval_samples_per_second": 194.329,
|
| 187 |
+
"eval_steps_per_second": 12.299,
|
| 188 |
+
"step": 1248
|
| 189 |
},
|
| 190 |
{
|
| 191 |
+
"epoch": 12.88,
|
| 192 |
+
"learning_rate": 4.0110946745562136e-05,
|
| 193 |
+
"loss": 0.0673,
|
| 194 |
+
"step": 1339
|
| 195 |
},
|
| 196 |
{
|
| 197 |
+
"epoch": 13.0,
|
| 198 |
+
"eval_f1": 0.5738915229311208,
|
| 199 |
+
"eval_loss": 2.6217703819274902,
|
| 200 |
+
"eval_runtime": 2.9035,
|
| 201 |
+
"eval_samples_per_second": 190.46,
|
| 202 |
+
"eval_steps_per_second": 12.054,
|
| 203 |
+
"step": 1352
|
| 204 |
},
|
| 205 |
{
|
| 206 |
+
"epoch": 13.87,
|
| 207 |
+
"learning_rate": 3.934911242603551e-05,
|
| 208 |
+
"loss": 0.0684,
|
| 209 |
+
"step": 1442
|
| 210 |
},
|
| 211 |
{
|
| 212 |
+
"epoch": 14.0,
|
| 213 |
+
"eval_f1": 0.5366433281464598,
|
| 214 |
+
"eval_loss": 2.9552414417266846,
|
| 215 |
+
"eval_runtime": 2.9063,
|
| 216 |
+
"eval_samples_per_second": 190.277,
|
| 217 |
+
"eval_steps_per_second": 12.043,
|
| 218 |
+
"step": 1456
|
| 219 |
},
|
| 220 |
{
|
| 221 |
+
"epoch": 14.86,
|
| 222 |
+
"learning_rate": 3.858727810650888e-05,
|
| 223 |
+
"loss": 0.0466,
|
| 224 |
+
"step": 1545
|
| 225 |
},
|
| 226 |
{
|
| 227 |
+
"epoch": 15.0,
|
| 228 |
+
"eval_f1": 0.5787084254032917,
|
| 229 |
+
"eval_loss": 2.7240512371063232,
|
| 230 |
+
"eval_runtime": 2.8887,
|
| 231 |
+
"eval_samples_per_second": 191.438,
|
| 232 |
+
"eval_steps_per_second": 12.116,
|
| 233 |
+
"step": 1560
|
| 234 |
},
|
| 235 |
{
|
| 236 |
+
"epoch": 15.85,
|
| 237 |
+
"learning_rate": 3.782544378698225e-05,
|
| 238 |
+
"loss": 0.0577,
|
| 239 |
+
"step": 1648
|
| 240 |
},
|
| 241 |
{
|
| 242 |
+
"epoch": 16.0,
|
| 243 |
+
"eval_f1": 0.5666557248979172,
|
| 244 |
+
"eval_loss": 2.821897506713867,
|
| 245 |
+
"eval_runtime": 2.8994,
|
| 246 |
+
"eval_samples_per_second": 190.727,
|
| 247 |
+
"eval_steps_per_second": 12.071,
|
| 248 |
+
"step": 1664
|
| 249 |
},
|
| 250 |
{
|
| 251 |
+
"epoch": 16.84,
|
| 252 |
+
"learning_rate": 3.706360946745562e-05,
|
| 253 |
+
"loss": 0.042,
|
| 254 |
+
"step": 1751
|
| 255 |
},
|
| 256 |
{
|
| 257 |
+
"epoch": 17.0,
|
| 258 |
+
"eval_f1": 0.56033452806457,
|
| 259 |
+
"eval_loss": 2.9155900478363037,
|
| 260 |
+
"eval_runtime": 2.9258,
|
| 261 |
+
"eval_samples_per_second": 189.01,
|
| 262 |
+
"eval_steps_per_second": 11.963,
|
| 263 |
+
"step": 1768
|
| 264 |
},
|
| 265 |
{
|
| 266 |
+
"epoch": 17.83,
|
| 267 |
+
"learning_rate": 3.6301775147928995e-05,
|
| 268 |
+
"loss": 0.0404,
|
| 269 |
+
"step": 1854
|
| 270 |
},
|
| 271 |
{
|
| 272 |
+
"epoch": 18.0,
|
| 273 |
+
"eval_f1": 0.5621979513908701,
|
| 274 |
+
"eval_loss": 2.893630266189575,
|
| 275 |
+
"eval_runtime": 2.9319,
|
| 276 |
+
"eval_samples_per_second": 188.613,
|
| 277 |
+
"eval_steps_per_second": 11.938,
|
| 278 |
+
"step": 1872
|
| 279 |
},
|
| 280 |
{
|
| 281 |
+
"epoch": 18.82,
|
| 282 |
+
"learning_rate": 3.553994082840237e-05,
|
| 283 |
+
"loss": 0.0426,
|
| 284 |
+
"step": 1957
|
| 285 |
},
|
| 286 |
{
|
| 287 |
+
"epoch": 19.0,
|
| 288 |
+
"eval_f1": 0.5766503161850353,
|
| 289 |
+
"eval_loss": 3.00762939453125,
|
| 290 |
+
"eval_runtime": 2.9064,
|
| 291 |
+
"eval_samples_per_second": 190.271,
|
| 292 |
+
"eval_steps_per_second": 12.042,
|
| 293 |
+
"step": 1976
|
| 294 |
},
|
| 295 |
{
|
| 296 |
+
"epoch": 19.81,
|
| 297 |
+
"learning_rate": 3.477810650887574e-05,
|
| 298 |
+
"loss": 0.0361,
|
| 299 |
+
"step": 2060
|
| 300 |
},
|
| 301 |
{
|
| 302 |
+
"epoch": 20.0,
|
| 303 |
+
"eval_f1": 0.5565194377868121,
|
| 304 |
+
"eval_loss": 3.043562173843384,
|
| 305 |
+
"eval_runtime": 2.9099,
|
| 306 |
+
"eval_samples_per_second": 190.043,
|
| 307 |
+
"eval_steps_per_second": 12.028,
|
| 308 |
+
"step": 2080
|
| 309 |
},
|
| 310 |
{
|
| 311 |
+
"epoch": 20.8,
|
| 312 |
+
"learning_rate": 3.401627218934911e-05,
|
| 313 |
+
"loss": 0.039,
|
| 314 |
+
"step": 2163
|
| 315 |
},
|
| 316 |
{
|
| 317 |
+
"epoch": 21.0,
|
| 318 |
+
"eval_f1": 0.5661283975776907,
|
| 319 |
+
"eval_loss": 3.034050226211548,
|
| 320 |
+
"eval_runtime": 2.8472,
|
| 321 |
+
"eval_samples_per_second": 194.228,
|
| 322 |
+
"eval_steps_per_second": 12.293,
|
| 323 |
+
"step": 2184
|
| 324 |
},
|
| 325 |
{
|
| 326 |
+
"epoch": 21.79,
|
| 327 |
+
"learning_rate": 3.325443786982248e-05,
|
| 328 |
+
"loss": 0.0311,
|
| 329 |
+
"step": 2266
|
| 330 |
},
|
| 331 |
{
|
| 332 |
+
"epoch": 22.0,
|
| 333 |
+
"eval_f1": 0.5698598461896062,
|
| 334 |
+
"eval_loss": 3.1546428203582764,
|
| 335 |
+
"eval_runtime": 2.8965,
|
| 336 |
+
"eval_samples_per_second": 190.921,
|
| 337 |
+
"eval_steps_per_second": 12.084,
|
| 338 |
+
"step": 2288
|
| 339 |
},
|
| 340 |
{
|
| 341 |
+
"epoch": 22.78,
|
| 342 |
+
"learning_rate": 3.2492603550295855e-05,
|
| 343 |
+
"loss": 0.0296,
|
| 344 |
+
"step": 2369
|
| 345 |
},
|
| 346 |
{
|
| 347 |
+
"epoch": 23.0,
|
| 348 |
+
"eval_f1": 0.5584145320343268,
|
| 349 |
+
"eval_loss": 3.3160221576690674,
|
| 350 |
+
"eval_runtime": 2.9004,
|
| 351 |
+
"eval_samples_per_second": 190.664,
|
| 352 |
+
"eval_steps_per_second": 12.067,
|
| 353 |
+
"step": 2392
|
| 354 |
},
|
| 355 |
{
|
| 356 |
+
"epoch": 23.77,
|
| 357 |
+
"learning_rate": 3.1730769230769234e-05,
|
| 358 |
+
"loss": 0.03,
|
| 359 |
+
"step": 2472
|
| 360 |
},
|
| 361 |
{
|
| 362 |
+
"epoch": 24.0,
|
| 363 |
+
"eval_f1": 0.5765799312977243,
|
| 364 |
+
"eval_loss": 3.2025678157806396,
|
| 365 |
+
"eval_runtime": 2.837,
|
| 366 |
+
"eval_samples_per_second": 194.921,
|
| 367 |
+
"eval_steps_per_second": 12.337,
|
| 368 |
+
"step": 2496
|
| 369 |
},
|
| 370 |
{
|
| 371 |
+
"epoch": 24.76,
|
| 372 |
+
"learning_rate": 3.0968934911242606e-05,
|
| 373 |
+
"loss": 0.0333,
|
| 374 |
+
"step": 2575
|
| 375 |
},
|
| 376 |
{
|
| 377 |
+
"epoch": 25.0,
|
| 378 |
+
"eval_f1": 0.5689553713820321,
|
| 379 |
+
"eval_loss": 3.211634397506714,
|
| 380 |
+
"eval_runtime": 2.9044,
|
| 381 |
+
"eval_samples_per_second": 190.402,
|
| 382 |
+
"eval_steps_per_second": 12.051,
|
| 383 |
+
"step": 2600
|
| 384 |
},
|
| 385 |
{
|
| 386 |
+
"epoch": 25.75,
|
| 387 |
+
"learning_rate": 3.0207100591715974e-05,
|
| 388 |
+
"loss": 0.0321,
|
| 389 |
+
"step": 2678
|
| 390 |
},
|
| 391 |
{
|
| 392 |
+
"epoch": 26.0,
|
| 393 |
+
"eval_f1": 0.5756108062994573,
|
| 394 |
+
"eval_loss": 3.2678425312042236,
|
| 395 |
+
"eval_runtime": 2.8888,
|
| 396 |
+
"eval_samples_per_second": 191.428,
|
| 397 |
+
"eval_steps_per_second": 12.116,
|
| 398 |
+
"step": 2704
|
| 399 |
},
|
| 400 |
{
|
| 401 |
+
"epoch": 26.74,
|
| 402 |
+
"learning_rate": 2.944526627218935e-05,
|
| 403 |
+
"loss": 0.0263,
|
| 404 |
+
"step": 2781
|
| 405 |
},
|
| 406 |
{
|
| 407 |
+
"epoch": 27.0,
|
| 408 |
+
"eval_f1": 0.5758065273285641,
|
| 409 |
+
"eval_loss": 3.2969822883605957,
|
| 410 |
+
"eval_runtime": 2.9527,
|
| 411 |
+
"eval_samples_per_second": 187.286,
|
| 412 |
+
"eval_steps_per_second": 11.854,
|
| 413 |
+
"step": 2808
|
| 414 |
},
|
| 415 |
{
|
| 416 |
+
"epoch": 27.73,
|
| 417 |
+
"learning_rate": 2.8683431952662725e-05,
|
| 418 |
+
"loss": 0.0281,
|
| 419 |
+
"step": 2884
|
| 420 |
},
|
| 421 |
{
|
| 422 |
+
"epoch": 28.0,
|
| 423 |
+
"eval_f1": 0.5781354966097151,
|
| 424 |
+
"eval_loss": 3.3730037212371826,
|
| 425 |
+
"eval_runtime": 2.8614,
|
| 426 |
+
"eval_samples_per_second": 193.264,
|
| 427 |
+
"eval_steps_per_second": 12.232,
|
| 428 |
+
"step": 2912
|
| 429 |
},
|
| 430 |
{
|
| 431 |
+
"epoch": 28.72,
|
| 432 |
+
"learning_rate": 2.7921597633136097e-05,
|
| 433 |
+
"loss": 0.0282,
|
| 434 |
+
"step": 2987
|
| 435 |
},
|
| 436 |
{
|
| 437 |
+
"epoch": 29.0,
|
| 438 |
+
"eval_f1": 0.5741866124789994,
|
| 439 |
+
"eval_loss": 3.364117383956909,
|
| 440 |
+
"eval_runtime": 2.8696,
|
| 441 |
+
"eval_samples_per_second": 192.707,
|
| 442 |
+
"eval_steps_per_second": 12.197,
|
| 443 |
+
"step": 3016
|
| 444 |
},
|
| 445 |
{
|
| 446 |
+
"epoch": 29.71,
|
| 447 |
+
"learning_rate": 2.7159763313609472e-05,
|
| 448 |
+
"loss": 0.0296,
|
| 449 |
+
"step": 3090
|
| 450 |
},
|
| 451 |
{
|
| 452 |
+
"epoch": 30.0,
|
| 453 |
+
"eval_f1": 0.5771762774162508,
|
| 454 |
+
"eval_loss": 3.3623032569885254,
|
| 455 |
+
"eval_runtime": 2.9567,
|
| 456 |
+
"eval_samples_per_second": 187.031,
|
| 457 |
+
"eval_steps_per_second": 11.837,
|
| 458 |
+
"step": 3120
|
| 459 |
},
|
| 460 |
{
|
| 461 |
+
"epoch": 30.7,
|
| 462 |
+
"learning_rate": 2.6397928994082844e-05,
|
| 463 |
+
"loss": 0.0308,
|
| 464 |
+
"step": 3193
|
| 465 |
},
|
| 466 |
{
|
| 467 |
+
"epoch": 31.0,
|
| 468 |
+
"eval_f1": 0.578537002980747,
|
| 469 |
+
"eval_loss": 3.4039528369903564,
|
| 470 |
+
"eval_runtime": 2.8263,
|
| 471 |
+
"eval_samples_per_second": 195.66,
|
| 472 |
+
"eval_steps_per_second": 12.384,
|
| 473 |
+
"step": 3224
|
| 474 |
},
|
| 475 |
{
|
| 476 |
+
"epoch": 31.69,
|
| 477 |
+
"learning_rate": 2.5636094674556216e-05,
|
| 478 |
+
"loss": 0.0308,
|
| 479 |
+
"step": 3296
|
| 480 |
},
|
| 481 |
{
|
| 482 |
+
"epoch": 32.0,
|
| 483 |
+
"eval_f1": 0.575919412837488,
|
| 484 |
+
"eval_loss": 3.392319679260254,
|
| 485 |
+
"eval_runtime": 2.9375,
|
| 486 |
+
"eval_samples_per_second": 188.254,
|
| 487 |
+
"eval_steps_per_second": 11.915,
|
| 488 |
+
"step": 3328
|
| 489 |
},
|
| 490 |
{
|
| 491 |
+
"epoch": 32.68,
|
| 492 |
+
"learning_rate": 2.4874260355029588e-05,
|
| 493 |
+
"loss": 0.0262,
|
| 494 |
+
"step": 3399
|
| 495 |
},
|
| 496 |
{
|
| 497 |
+
"epoch": 33.0,
|
| 498 |
+
"eval_f1": 0.5563772891428104,
|
| 499 |
+
"eval_loss": 3.4757542610168457,
|
| 500 |
+
"eval_runtime": 2.865,
|
| 501 |
+
"eval_samples_per_second": 193.019,
|
| 502 |
+
"eval_steps_per_second": 12.216,
|
| 503 |
+
"step": 3432
|
| 504 |
},
|
| 505 |
{
|
| 506 |
+
"epoch": 33.67,
|
| 507 |
+
"learning_rate": 2.411242603550296e-05,
|
| 508 |
+
"loss": 0.0319,
|
| 509 |
+
"step": 3502
|
| 510 |
},
|
| 511 |
{
|
| 512 |
+
"epoch": 34.0,
|
| 513 |
+
"eval_f1": 0.5738865992034025,
|
| 514 |
+
"eval_loss": 3.425334930419922,
|
| 515 |
+
"eval_runtime": 2.9109,
|
| 516 |
+
"eval_samples_per_second": 189.974,
|
| 517 |
+
"eval_steps_per_second": 12.024,
|
| 518 |
+
"step": 3536
|
| 519 |
},
|
| 520 |
{
|
| 521 |
+
"epoch": 34.66,
|
| 522 |
+
"learning_rate": 2.3350591715976332e-05,
|
| 523 |
+
"loss": 0.0277,
|
| 524 |
+
"step": 3605
|
| 525 |
},
|
| 526 |
{
|
| 527 |
+
"epoch": 35.0,
|
| 528 |
+
"eval_f1": 0.5785980513801816,
|
| 529 |
+
"eval_loss": 3.4686436653137207,
|
| 530 |
+
"eval_runtime": 2.93,
|
| 531 |
+
"eval_samples_per_second": 188.738,
|
| 532 |
+
"eval_steps_per_second": 11.945,
|
| 533 |
+
"step": 3640
|
| 534 |
},
|
| 535 |
{
|
| 536 |
+
"epoch": 35.65,
|
| 537 |
+
"learning_rate": 2.2588757396449707e-05,
|
| 538 |
+
"loss": 0.0289,
|
| 539 |
+
"step": 3708
|
| 540 |
},
|
| 541 |
{
|
| 542 |
+
"epoch": 36.0,
|
| 543 |
+
"eval_f1": 0.5836924697871717,
|
| 544 |
+
"eval_loss": 3.462078094482422,
|
| 545 |
+
"eval_runtime": 2.8428,
|
| 546 |
+
"eval_samples_per_second": 194.527,
|
| 547 |
+
"eval_steps_per_second": 12.312,
|
| 548 |
+
"step": 3744
|
| 549 |
},
|
| 550 |
{
|
| 551 |
+
"epoch": 36.64,
|
| 552 |
+
"learning_rate": 2.182692307692308e-05,
|
| 553 |
+
"loss": 0.0247,
|
| 554 |
+
"step": 3811
|
| 555 |
},
|
| 556 |
{
|
| 557 |
+
"epoch": 37.0,
|
| 558 |
+
"eval_f1": 0.5734707197245945,
|
| 559 |
+
"eval_loss": 3.481998920440674,
|
| 560 |
+
"eval_runtime": 3.0017,
|
| 561 |
+
"eval_samples_per_second": 184.228,
|
| 562 |
+
"eval_steps_per_second": 11.66,
|
| 563 |
+
"step": 3848
|
| 564 |
},
|
| 565 |
{
|
| 566 |
+
"epoch": 37.63,
|
| 567 |
+
"learning_rate": 2.106508875739645e-05,
|
| 568 |
+
"loss": 0.0303,
|
| 569 |
+
"step": 3914
|
| 570 |
},
|
| 571 |
{
|
| 572 |
+
"epoch": 38.0,
|
| 573 |
+
"eval_f1": 0.5770262969511715,
|
| 574 |
+
"eval_loss": 3.466510772705078,
|
| 575 |
+
"eval_runtime": 2.8587,
|
| 576 |
+
"eval_samples_per_second": 193.442,
|
| 577 |
+
"eval_steps_per_second": 12.243,
|
| 578 |
+
"step": 3952
|
| 579 |
},
|
| 580 |
{
|
| 581 |
+
"epoch": 38.62,
|
| 582 |
+
"learning_rate": 2.0303254437869823e-05,
|
| 583 |
+
"loss": 0.0239,
|
| 584 |
+
"step": 4017
|
| 585 |
},
|
| 586 |
{
|
| 587 |
+
"epoch": 39.0,
|
| 588 |
+
"eval_f1": 0.5666519467364683,
|
| 589 |
+
"eval_loss": 3.5593807697296143,
|
| 590 |
+
"eval_runtime": 2.8222,
|
| 591 |
+
"eval_samples_per_second": 195.946,
|
| 592 |
+
"eval_steps_per_second": 12.402,
|
| 593 |
+
"step": 4056
|
| 594 |
},
|
| 595 |
{
|
| 596 |
+
"epoch": 39.62,
|
| 597 |
+
"learning_rate": 1.9541420118343195e-05,
|
| 598 |
+
"loss": 0.0262,
|
| 599 |
+
"step": 4120
|
| 600 |
},
|
| 601 |
{
|
| 602 |
+
"epoch": 40.0,
|
| 603 |
+
"eval_f1": 0.5808476343157906,
|
| 604 |
+
"eval_loss": 3.5302422046661377,
|
| 605 |
+
"eval_runtime": 2.8598,
|
| 606 |
+
"eval_samples_per_second": 193.368,
|
| 607 |
+
"eval_steps_per_second": 12.238,
|
| 608 |
+
"step": 4160
|
| 609 |
},
|
| 610 |
{
|
| 611 |
+
"epoch": 40.61,
|
| 612 |
+
"learning_rate": 1.8779585798816567e-05,
|
| 613 |
+
"loss": 0.0282,
|
| 614 |
+
"step": 4223
|
| 615 |
},
|
| 616 |
{
|
| 617 |
+
"epoch": 41.0,
|
| 618 |
+
"eval_f1": 0.5835890408164021,
|
| 619 |
+
"eval_loss": 3.4572339057922363,
|
| 620 |
+
"eval_runtime": 2.8566,
|
| 621 |
+
"eval_samples_per_second": 193.584,
|
| 622 |
+
"eval_steps_per_second": 12.252,
|
| 623 |
+
"step": 4264
|
| 624 |
},
|
| 625 |
{
|
| 626 |
+
"epoch": 41.6,
|
| 627 |
+
"learning_rate": 1.8025147928994084e-05,
|
| 628 |
+
"loss": 0.0469,
|
| 629 |
+
"step": 4326
|
| 630 |
},
|
| 631 |
{
|
| 632 |
+
"epoch": 42.0,
|
| 633 |
+
"eval_f1": 0.5685331156394952,
|
| 634 |
+
"eval_loss": 3.609334707260132,
|
| 635 |
+
"eval_runtime": 2.8251,
|
| 636 |
+
"eval_samples_per_second": 195.747,
|
| 637 |
+
"eval_steps_per_second": 12.389,
|
| 638 |
+
"step": 4368
|
| 639 |
},
|
| 640 |
{
|
| 641 |
+
"epoch": 42.59,
|
| 642 |
+
"learning_rate": 1.7263313609467456e-05,
|
| 643 |
+
"loss": 0.0302,
|
| 644 |
+
"step": 4429
|
| 645 |
},
|
| 646 |
{
|
| 647 |
+
"epoch": 43.0,
|
| 648 |
+
"eval_f1": 0.5684067370608473,
|
| 649 |
+
"eval_loss": 3.6115400791168213,
|
| 650 |
+
"eval_runtime": 2.9194,
|
| 651 |
+
"eval_samples_per_second": 189.42,
|
| 652 |
+
"eval_steps_per_second": 11.989,
|
| 653 |
+
"step": 4472
|
| 654 |
},
|
| 655 |
{
|
| 656 |
+
"epoch": 43.58,
|
| 657 |
+
"learning_rate": 1.650147928994083e-05,
|
| 658 |
+
"loss": 0.0289,
|
| 659 |
+
"step": 4532
|
| 660 |
},
|
| 661 |
{
|
| 662 |
+
"epoch": 44.0,
|
| 663 |
+
"eval_f1": 0.5757900647671246,
|
| 664 |
+
"eval_loss": 3.629568099975586,
|
| 665 |
+
"eval_runtime": 2.9036,
|
| 666 |
+
"eval_samples_per_second": 190.453,
|
| 667 |
+
"eval_steps_per_second": 12.054,
|
| 668 |
+
"step": 4576
|
| 669 |
},
|
| 670 |
{
|
| 671 |
+
"epoch": 44.57,
|
| 672 |
+
"learning_rate": 1.5739644970414204e-05,
|
| 673 |
+
"loss": 0.0254,
|
| 674 |
+
"step": 4635
|
| 675 |
},
|
| 676 |
{
|
| 677 |
+
"epoch": 45.0,
|
| 678 |
+
"eval_f1": 0.5689505752768721,
|
| 679 |
+
"eval_loss": 3.7250843048095703,
|
| 680 |
+
"eval_runtime": 2.9726,
|
| 681 |
+
"eval_samples_per_second": 186.035,
|
| 682 |
+
"eval_steps_per_second": 11.774,
|
| 683 |
+
"step": 4680
|
| 684 |
},
|
| 685 |
{
|
| 686 |
+
"epoch": 45.56,
|
| 687 |
+
"learning_rate": 1.4977810650887576e-05,
|
| 688 |
+
"loss": 0.0283,
|
| 689 |
+
"step": 4738
|
| 690 |
},
|
| 691 |
{
|
| 692 |
+
"epoch": 46.0,
|
| 693 |
+
"eval_f1": 0.5592198654774546,
|
| 694 |
+
"eval_loss": 3.726353645324707,
|
| 695 |
+
"eval_runtime": 2.9328,
|
| 696 |
+
"eval_samples_per_second": 188.559,
|
| 697 |
+
"eval_steps_per_second": 11.934,
|
| 698 |
+
"step": 4784
|
| 699 |
},
|
| 700 |
{
|
| 701 |
+
"epoch": 46.55,
|
| 702 |
+
"learning_rate": 1.4215976331360948e-05,
|
| 703 |
+
"loss": 0.0246,
|
| 704 |
+
"step": 4841
|
| 705 |
},
|
| 706 |
{
|
| 707 |
+
"epoch": 47.0,
|
| 708 |
+
"eval_f1": 0.5650157110711802,
|
| 709 |
+
"eval_loss": 3.7832093238830566,
|
| 710 |
+
"eval_runtime": 2.9067,
|
| 711 |
+
"eval_samples_per_second": 190.249,
|
| 712 |
+
"eval_steps_per_second": 12.041,
|
| 713 |
+
"step": 4888
|
| 714 |
},
|
| 715 |
{
|
| 716 |
+
"epoch": 47.54,
|
| 717 |
+
"learning_rate": 1.345414201183432e-05,
|
| 718 |
+
"loss": 0.0311,
|
| 719 |
+
"step": 4944
|
| 720 |
},
|
| 721 |
{
|
| 722 |
+
"epoch": 48.0,
|
| 723 |
+
"eval_f1": 0.5681512072556809,
|
| 724 |
+
"eval_loss": 3.6964025497436523,
|
| 725 |
+
"eval_runtime": 2.9008,
|
| 726 |
+
"eval_samples_per_second": 190.634,
|
| 727 |
+
"eval_steps_per_second": 12.065,
|
| 728 |
+
"step": 4992
|
| 729 |
},
|
| 730 |
{
|
| 731 |
+
"epoch": 48.53,
|
| 732 |
+
"learning_rate": 1.2692307692307691e-05,
|
| 733 |
+
"loss": 0.0268,
|
| 734 |
+
"step": 5047
|
| 735 |
},
|
| 736 |
{
|
| 737 |
+
"epoch": 49.0,
|
| 738 |
+
"eval_f1": 0.5674808111122996,
|
| 739 |
+
"eval_loss": 3.7195167541503906,
|
| 740 |
+
"eval_runtime": 2.8604,
|
| 741 |
+
"eval_samples_per_second": 193.33,
|
| 742 |
+
"eval_steps_per_second": 12.236,
|
| 743 |
+
"step": 5096
|
| 744 |
},
|
| 745 |
{
|
| 746 |
+
"epoch": 49.52,
|
| 747 |
+
"learning_rate": 1.1930473372781067e-05,
|
| 748 |
+
"loss": 0.0293,
|
| 749 |
+
"step": 5150
|
| 750 |
},
|
| 751 |
{
|
| 752 |
+
"epoch": 50.0,
|
| 753 |
+
"eval_f1": 0.5614419693521525,
|
| 754 |
+
"eval_loss": 3.752530097961426,
|
| 755 |
+
"eval_runtime": 2.8761,
|
| 756 |
+
"eval_samples_per_second": 192.275,
|
| 757 |
+
"eval_steps_per_second": 12.169,
|
| 758 |
+
"step": 5200
|
| 759 |
},
|
| 760 |
{
|
| 761 |
+
"epoch": 50.51,
|
| 762 |
+
"learning_rate": 1.1168639053254439e-05,
|
| 763 |
+
"loss": 0.0282,
|
| 764 |
+
"step": 5253
|
| 765 |
},
|
| 766 |
{
|
| 767 |
+
"epoch": 51.0,
|
| 768 |
+
"eval_f1": 0.5655838635083059,
|
| 769 |
+
"eval_loss": 3.7514984607696533,
|
| 770 |
+
"eval_runtime": 2.8609,
|
| 771 |
+
"eval_samples_per_second": 193.296,
|
| 772 |
+
"eval_steps_per_second": 12.234,
|
| 773 |
+
"step": 5304
|
| 774 |
},
|
| 775 |
{
|
| 776 |
+
"epoch": 51.5,
|
| 777 |
+
"learning_rate": 1.040680473372781e-05,
|
| 778 |
+
"loss": 0.0248,
|
| 779 |
+
"step": 5356
|
| 780 |
},
|
| 781 |
{
|
| 782 |
+
"epoch": 52.0,
|
| 783 |
+
"eval_f1": 0.5590951084274065,
|
| 784 |
+
"eval_loss": 3.7639315128326416,
|
| 785 |
+
"eval_runtime": 2.8211,
|
| 786 |
+
"eval_samples_per_second": 196.025,
|
| 787 |
+
"eval_steps_per_second": 12.407,
|
| 788 |
+
"step": 5408
|
| 789 |
},
|
| 790 |
{
|
| 791 |
+
"epoch": 52.49,
|
| 792 |
+
"learning_rate": 9.644970414201183e-06,
|
| 793 |
+
"loss": 0.0257,
|
| 794 |
+
"step": 5459
|
| 795 |
},
|
| 796 |
{
|
| 797 |
+
"epoch": 53.0,
|
| 798 |
+
"eval_f1": 0.5480134247467852,
|
| 799 |
+
"eval_loss": 3.824922800064087,
|
| 800 |
+
"eval_runtime": 2.8475,
|
| 801 |
+
"eval_samples_per_second": 194.205,
|
| 802 |
+
"eval_steps_per_second": 12.291,
|
| 803 |
+
"step": 5512
|
| 804 |
},
|
| 805 |
{
|
| 806 |
+
"epoch": 53.48,
|
| 807 |
+
"learning_rate": 8.883136094674558e-06,
|
| 808 |
+
"loss": 0.0235,
|
| 809 |
+
"step": 5562
|
| 810 |
},
|
| 811 |
{
|
| 812 |
+
"epoch": 54.0,
|
| 813 |
+
"eval_f1": 0.5565796472147394,
|
| 814 |
+
"eval_loss": 3.7871253490448,
|
| 815 |
+
"eval_runtime": 2.9817,
|
| 816 |
+
"eval_samples_per_second": 185.462,
|
| 817 |
+
"eval_steps_per_second": 11.738,
|
| 818 |
+
"step": 5616
|
| 819 |
},
|
| 820 |
{
|
| 821 |
+
"epoch": 54.47,
|
| 822 |
+
"learning_rate": 8.12130177514793e-06,
|
| 823 |
+
"loss": 0.0299,
|
| 824 |
+
"step": 5665
|
| 825 |
},
|
| 826 |
{
|
| 827 |
+
"epoch": 55.0,
|
| 828 |
+
"eval_f1": 0.5574154263000176,
|
| 829 |
+
"eval_loss": 3.788760185241699,
|
| 830 |
+
"eval_runtime": 2.8852,
|
| 831 |
+
"eval_samples_per_second": 191.665,
|
| 832 |
+
"eval_steps_per_second": 12.131,
|
| 833 |
+
"step": 5720
|
| 834 |
},
|
| 835 |
{
|
| 836 |
+
"epoch": 55.46,
|
| 837 |
+
"learning_rate": 7.359467455621302e-06,
|
| 838 |
+
"loss": 0.0277,
|
| 839 |
+
"step": 5768
|
| 840 |
},
|
| 841 |
{
|
| 842 |
+
"epoch": 56.0,
|
| 843 |
+
"eval_f1": 0.563024311843682,
|
| 844 |
+
"eval_loss": 3.7907044887542725,
|
| 845 |
+
"eval_runtime": 2.8658,
|
| 846 |
+
"eval_samples_per_second": 192.962,
|
| 847 |
+
"eval_steps_per_second": 12.213,
|
| 848 |
+
"step": 5824
|
| 849 |
},
|
| 850 |
{
|
| 851 |
+
"epoch": 56.45,
|
| 852 |
+
"learning_rate": 6.597633136094675e-06,
|
| 853 |
+
"loss": 0.0256,
|
| 854 |
+
"step": 5871
|
| 855 |
},
|
| 856 |
{
|
| 857 |
+
"epoch": 57.0,
|
| 858 |
+
"eval_f1": 0.56153234588093,
|
| 859 |
+
"eval_loss": 3.799422264099121,
|
| 860 |
+
"eval_runtime": 2.8666,
|
| 861 |
+
"eval_samples_per_second": 192.912,
|
| 862 |
+
"eval_steps_per_second": 12.21,
|
| 863 |
+
"step": 5928
|
| 864 |
},
|
| 865 |
{
|
| 866 |
+
"epoch": 57.44,
|
| 867 |
+
"learning_rate": 5.8357988165680474e-06,
|
| 868 |
+
"loss": 0.0226,
|
| 869 |
+
"step": 5974
|
| 870 |
},
|
| 871 |
{
|
| 872 |
+
"epoch": 58.0,
|
| 873 |
+
"eval_f1": 0.5555061070073688,
|
| 874 |
+
"eval_loss": 3.811858892440796,
|
| 875 |
+
"eval_runtime": 2.8683,
|
| 876 |
+
"eval_samples_per_second": 192.797,
|
| 877 |
+
"eval_steps_per_second": 12.202,
|
| 878 |
+
"step": 6032
|
| 879 |
},
|
| 880 |
{
|
| 881 |
+
"epoch": 58.43,
|
| 882 |
+
"learning_rate": 5.07396449704142e-06,
|
| 883 |
+
"loss": 0.0284,
|
| 884 |
+
"step": 6077
|
| 885 |
},
|
| 886 |
{
|
| 887 |
+
"epoch": 59.0,
|
| 888 |
+
"eval_f1": 0.5597671150511061,
|
| 889 |
+
"eval_loss": 3.8192451000213623,
|
| 890 |
+
"eval_runtime": 2.8512,
|
| 891 |
+
"eval_samples_per_second": 193.951,
|
| 892 |
+
"eval_steps_per_second": 12.275,
|
| 893 |
+
"step": 6136
|
| 894 |
},
|
| 895 |
{
|
| 896 |
+
"epoch": 59.42,
|
| 897 |
+
"learning_rate": 4.312130177514793e-06,
|
| 898 |
+
"loss": 0.0233,
|
| 899 |
+
"step": 6180
|
| 900 |
},
|
| 901 |
{
|
| 902 |
+
"epoch": 60.0,
|
| 903 |
+
"eval_f1": 0.5584681716027172,
|
| 904 |
+
"eval_loss": 3.823091983795166,
|
| 905 |
+
"eval_runtime": 2.9385,
|
| 906 |
+
"eval_samples_per_second": 188.191,
|
| 907 |
+
"eval_steps_per_second": 11.911,
|
| 908 |
+
"step": 6240
|
| 909 |
},
|
| 910 |
{
|
| 911 |
+
"epoch": 60.41,
|
| 912 |
+
"learning_rate": 3.550295857988166e-06,
|
| 913 |
+
"loss": 0.0266,
|
| 914 |
+
"step": 6283
|
| 915 |
},
|
| 916 |
{
|
| 917 |
+
"epoch": 61.0,
|
| 918 |
+
"eval_f1": 0.5625000576804086,
|
| 919 |
+
"eval_loss": 3.8085415363311768,
|
| 920 |
+
"eval_runtime": 2.9015,
|
| 921 |
+
"eval_samples_per_second": 190.588,
|
| 922 |
+
"eval_steps_per_second": 12.063,
|
| 923 |
+
"step": 6344
|
| 924 |
},
|
| 925 |
{
|
| 926 |
+
"epoch": 61.4,
|
| 927 |
+
"learning_rate": 2.7958579881656803e-06,
|
| 928 |
+
"loss": 0.0267,
|
| 929 |
+
"step": 6386
|
| 930 |
},
|
| 931 |
{
|
| 932 |
+
"epoch": 62.0,
|
| 933 |
+
"eval_f1": 0.5622167257088028,
|
| 934 |
+
"eval_loss": 3.80642032623291,
|
| 935 |
+
"eval_runtime": 2.8514,
|
| 936 |
+
"eval_samples_per_second": 193.94,
|
| 937 |
+
"eval_steps_per_second": 12.275,
|
| 938 |
+
"step": 6448
|
| 939 |
},
|
| 940 |
{
|
| 941 |
+
"epoch": 62.39,
|
| 942 |
+
"learning_rate": 2.034023668639053e-06,
|
| 943 |
+
"loss": 0.0281,
|
| 944 |
+
"step": 6489
|
| 945 |
},
|
| 946 |
{
|
| 947 |
+
"epoch": 63.0,
|
| 948 |
+
"eval_f1": 0.564106811375439,
|
| 949 |
+
"eval_loss": 3.8057875633239746,
|
| 950 |
+
"eval_runtime": 2.8945,
|
| 951 |
+
"eval_samples_per_second": 191.055,
|
| 952 |
+
"eval_steps_per_second": 12.092,
|
| 953 |
+
"step": 6552
|
| 954 |
},
|
| 955 |
{
|
| 956 |
+
"epoch": 63.38,
|
| 957 |
+
"learning_rate": 1.2721893491124261e-06,
|
| 958 |
+
"loss": 0.025,
|
| 959 |
+
"step": 6592
|
| 960 |
},
|
| 961 |
{
|
| 962 |
+
"epoch": 64.0,
|
| 963 |
+
"eval_f1": 0.5644375312998279,
|
| 964 |
+
"eval_loss": 3.807055950164795,
|
| 965 |
+
"eval_runtime": 2.8941,
|
| 966 |
+
"eval_samples_per_second": 191.08,
|
| 967 |
+
"eval_steps_per_second": 12.094,
|
| 968 |
+
"step": 6656
|
| 969 |
},
|
| 970 |
{
|
| 971 |
+
"epoch": 64.38,
|
| 972 |
+
"learning_rate": 5.103550295857988e-07,
|
| 973 |
+
"loss": 0.0226,
|
| 974 |
+
"step": 6695
|
| 975 |
},
|
| 976 |
{
|
| 977 |
+
"epoch": 65.0,
|
| 978 |
+
"eval_f1": 0.5644375312998279,
|
| 979 |
+
"eval_loss": 3.807528018951416,
|
| 980 |
+
"eval_runtime": 2.8626,
|
| 981 |
+
"eval_samples_per_second": 193.181,
|
| 982 |
+
"eval_steps_per_second": 12.227,
|
| 983 |
+
"step": 6760
|
| 984 |
}
|
| 985 |
],
|
| 986 |
+
"max_steps": 6760,
|
| 987 |
+
"num_train_epochs": 65,
|
| 988 |
+
"total_flos": 1.4286659901696e+16,
|
| 989 |
"trial_name": null,
|
| 990 |
"trial_params": null
|
| 991 |
}
|
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 sha256:69dba15562adc963c0e958faa40949482ed3f4a4e7db086eedc1130e4eecc7b6
|
| 3 |
+
size 3707
|