Training in progress, step 1400
Browse files- last-checkpoint/optimizer.pt +1 -1
- last-checkpoint/pytorch_model.bin +1 -1
- last-checkpoint/rng_state.pth +1 -1
- last-checkpoint/scheduler.pt +1 -1
- last-checkpoint/trainer_state.json +1218 -5
- pytorch_model.bin +1 -1
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 4736616809
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:cc8c482b86deee0fb4a3d206019c2f944600a1beaf33e642652365d949baf544
|
3 |
size 4736616809
|
last-checkpoint/pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2368281769
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:34e19f0a62fca8a6230a310e58105a9559ae0ec18159a3bf81179a7291941a60
|
3 |
size 2368281769
|
last-checkpoint/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 sha256:994fa97f10fcf531c2808fae8d33d622a1ea9689dd2ac2b8450b104451919c77
|
3 |
size 14575
|
last-checkpoint/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 sha256:66959129b2fef45a48ae80b3ff5396af5b977514f39900ca7b3a83bdae9745ff
|
3 |
size 627
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,8 +1,8 @@
|
|
1 |
{
|
2 |
-
"best_metric": 2.
|
3 |
-
"best_model_checkpoint": "output/checkpoint-
|
4 |
-
"epoch":
|
5 |
-
"global_step":
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
@@ -7284,11 +7284,1224 @@
|
|
7284 |
"eval_samples_per_second": 0.054,
|
7285 |
"eval_steps_per_second": 0.054,
|
7286 |
"step": 1200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7287 |
}
|
7288 |
],
|
7289 |
"max_steps": 2000,
|
7290 |
"num_train_epochs": 9223372036854775807,
|
7291 |
-
"total_flos":
|
7292 |
"trial_name": null,
|
7293 |
"trial_params": null
|
7294 |
}
|
|
|
1 |
{
|
2 |
+
"best_metric": 2.1156013011932373,
|
3 |
+
"best_model_checkpoint": "output/checkpoint-1400",
|
4 |
+
"epoch": 1.08650390625,
|
5 |
+
"global_step": 1400,
|
6 |
"is_hyper_param_search": false,
|
7 |
"is_local_process_zero": true,
|
8 |
"is_world_process_zero": true,
|
|
|
7284 |
"eval_samples_per_second": 0.054,
|
7285 |
"eval_steps_per_second": 0.054,
|
7286 |
"step": 1200
|
7287 |
+
},
|
7288 |
+
{
|
7289 |
+
"epoch": 0.6,
|
7290 |
+
"learning_rate": 0.0003551111111111111,
|
7291 |
+
"loss": 2.2215,
|
7292 |
+
"step": 1201
|
7293 |
+
},
|
7294 |
+
{
|
7295 |
+
"epoch": 0.6,
|
7296 |
+
"learning_rate": 0.0003546666666666667,
|
7297 |
+
"loss": 2.189,
|
7298 |
+
"step": 1202
|
7299 |
+
},
|
7300 |
+
{
|
7301 |
+
"epoch": 0.6,
|
7302 |
+
"learning_rate": 0.00035422222222222224,
|
7303 |
+
"loss": 2.2094,
|
7304 |
+
"step": 1203
|
7305 |
+
},
|
7306 |
+
{
|
7307 |
+
"epoch": 0.6,
|
7308 |
+
"learning_rate": 0.0003537777777777778,
|
7309 |
+
"loss": 2.2869,
|
7310 |
+
"step": 1204
|
7311 |
+
},
|
7312 |
+
{
|
7313 |
+
"epoch": 0.6,
|
7314 |
+
"learning_rate": 0.0003533333333333333,
|
7315 |
+
"loss": 2.2525,
|
7316 |
+
"step": 1205
|
7317 |
+
},
|
7318 |
+
{
|
7319 |
+
"epoch": 0.6,
|
7320 |
+
"learning_rate": 0.0003528888888888889,
|
7321 |
+
"loss": 2.1887,
|
7322 |
+
"step": 1206
|
7323 |
+
},
|
7324 |
+
{
|
7325 |
+
"epoch": 0.6,
|
7326 |
+
"learning_rate": 0.00035244444444444445,
|
7327 |
+
"loss": 2.2738,
|
7328 |
+
"step": 1207
|
7329 |
+
},
|
7330 |
+
{
|
7331 |
+
"epoch": 0.6,
|
7332 |
+
"learning_rate": 0.00035200000000000005,
|
7333 |
+
"loss": 2.2358,
|
7334 |
+
"step": 1208
|
7335 |
+
},
|
7336 |
+
{
|
7337 |
+
"epoch": 0.6,
|
7338 |
+
"learning_rate": 0.0003515555555555556,
|
7339 |
+
"loss": 2.2106,
|
7340 |
+
"step": 1209
|
7341 |
+
},
|
7342 |
+
{
|
7343 |
+
"epoch": 0.6,
|
7344 |
+
"learning_rate": 0.0003511111111111111,
|
7345 |
+
"loss": 2.2539,
|
7346 |
+
"step": 1210
|
7347 |
+
},
|
7348 |
+
{
|
7349 |
+
"epoch": 0.61,
|
7350 |
+
"learning_rate": 0.0003506666666666667,
|
7351 |
+
"loss": 2.2395,
|
7352 |
+
"step": 1211
|
7353 |
+
},
|
7354 |
+
{
|
7355 |
+
"epoch": 0.61,
|
7356 |
+
"learning_rate": 0.00035022222222222225,
|
7357 |
+
"loss": 2.146,
|
7358 |
+
"step": 1212
|
7359 |
+
},
|
7360 |
+
{
|
7361 |
+
"epoch": 0.61,
|
7362 |
+
"learning_rate": 0.0003497777777777778,
|
7363 |
+
"loss": 2.2555,
|
7364 |
+
"step": 1213
|
7365 |
+
},
|
7366 |
+
{
|
7367 |
+
"epoch": 0.61,
|
7368 |
+
"learning_rate": 0.00034933333333333333,
|
7369 |
+
"loss": 2.2447,
|
7370 |
+
"step": 1214
|
7371 |
+
},
|
7372 |
+
{
|
7373 |
+
"epoch": 0.61,
|
7374 |
+
"learning_rate": 0.0003488888888888889,
|
7375 |
+
"loss": 2.3177,
|
7376 |
+
"step": 1215
|
7377 |
+
},
|
7378 |
+
{
|
7379 |
+
"epoch": 0.61,
|
7380 |
+
"learning_rate": 0.00034844444444444446,
|
7381 |
+
"loss": 2.2334,
|
7382 |
+
"step": 1216
|
7383 |
+
},
|
7384 |
+
{
|
7385 |
+
"epoch": 0.61,
|
7386 |
+
"learning_rate": 0.000348,
|
7387 |
+
"loss": 2.1884,
|
7388 |
+
"step": 1217
|
7389 |
+
},
|
7390 |
+
{
|
7391 |
+
"epoch": 0.61,
|
7392 |
+
"learning_rate": 0.0003475555555555556,
|
7393 |
+
"loss": 2.2108,
|
7394 |
+
"step": 1218
|
7395 |
+
},
|
7396 |
+
{
|
7397 |
+
"epoch": 0.61,
|
7398 |
+
"learning_rate": 0.00034711111111111113,
|
7399 |
+
"loss": 2.2498,
|
7400 |
+
"step": 1219
|
7401 |
+
},
|
7402 |
+
{
|
7403 |
+
"epoch": 0.61,
|
7404 |
+
"learning_rate": 0.00034666666666666667,
|
7405 |
+
"loss": 2.242,
|
7406 |
+
"step": 1220
|
7407 |
+
},
|
7408 |
+
{
|
7409 |
+
"epoch": 0.61,
|
7410 |
+
"learning_rate": 0.0003462222222222222,
|
7411 |
+
"loss": 2.2292,
|
7412 |
+
"step": 1221
|
7413 |
+
},
|
7414 |
+
{
|
7415 |
+
"epoch": 0.61,
|
7416 |
+
"learning_rate": 0.0003457777777777778,
|
7417 |
+
"loss": 2.2706,
|
7418 |
+
"step": 1222
|
7419 |
+
},
|
7420 |
+
{
|
7421 |
+
"epoch": 0.61,
|
7422 |
+
"learning_rate": 0.00034533333333333334,
|
7423 |
+
"loss": 2.2784,
|
7424 |
+
"step": 1223
|
7425 |
+
},
|
7426 |
+
{
|
7427 |
+
"epoch": 0.61,
|
7428 |
+
"learning_rate": 0.0003448888888888889,
|
7429 |
+
"loss": 2.1807,
|
7430 |
+
"step": 1224
|
7431 |
+
},
|
7432 |
+
{
|
7433 |
+
"epoch": 0.61,
|
7434 |
+
"learning_rate": 0.0003444444444444445,
|
7435 |
+
"loss": 2.211,
|
7436 |
+
"step": 1225
|
7437 |
+
},
|
7438 |
+
{
|
7439 |
+
"epoch": 0.61,
|
7440 |
+
"learning_rate": 0.000344,
|
7441 |
+
"loss": 2.2172,
|
7442 |
+
"step": 1226
|
7443 |
+
},
|
7444 |
+
{
|
7445 |
+
"epoch": 1.0,
|
7446 |
+
"learning_rate": 0.0003435555555555556,
|
7447 |
+
"loss": 2.1684,
|
7448 |
+
"step": 1227
|
7449 |
+
},
|
7450 |
+
{
|
7451 |
+
"epoch": 1.0,
|
7452 |
+
"learning_rate": 0.0003431111111111111,
|
7453 |
+
"loss": 2.2307,
|
7454 |
+
"step": 1228
|
7455 |
+
},
|
7456 |
+
{
|
7457 |
+
"epoch": 1.0,
|
7458 |
+
"learning_rate": 0.0003426666666666667,
|
7459 |
+
"loss": 2.2313,
|
7460 |
+
"step": 1229
|
7461 |
+
},
|
7462 |
+
{
|
7463 |
+
"epoch": 1.0,
|
7464 |
+
"learning_rate": 0.0003422222222222222,
|
7465 |
+
"loss": 2.1823,
|
7466 |
+
"step": 1230
|
7467 |
+
},
|
7468 |
+
{
|
7469 |
+
"epoch": 1.0,
|
7470 |
+
"learning_rate": 0.0003417777777777778,
|
7471 |
+
"loss": 2.1682,
|
7472 |
+
"step": 1231
|
7473 |
+
},
|
7474 |
+
{
|
7475 |
+
"epoch": 1.0,
|
7476 |
+
"learning_rate": 0.00034133333333333335,
|
7477 |
+
"loss": 2.2691,
|
7478 |
+
"step": 1232
|
7479 |
+
},
|
7480 |
+
{
|
7481 |
+
"epoch": 1.0,
|
7482 |
+
"learning_rate": 0.0003408888888888889,
|
7483 |
+
"loss": 2.1713,
|
7484 |
+
"step": 1233
|
7485 |
+
},
|
7486 |
+
{
|
7487 |
+
"epoch": 1.0,
|
7488 |
+
"learning_rate": 0.0003404444444444445,
|
7489 |
+
"loss": 2.1428,
|
7490 |
+
"step": 1234
|
7491 |
+
},
|
7492 |
+
{
|
7493 |
+
"epoch": 1.0,
|
7494 |
+
"learning_rate": 0.00034,
|
7495 |
+
"loss": 2.2978,
|
7496 |
+
"step": 1235
|
7497 |
+
},
|
7498 |
+
{
|
7499 |
+
"epoch": 1.0,
|
7500 |
+
"learning_rate": 0.00033955555555555556,
|
7501 |
+
"loss": 2.2355,
|
7502 |
+
"step": 1236
|
7503 |
+
},
|
7504 |
+
{
|
7505 |
+
"epoch": 1.01,
|
7506 |
+
"learning_rate": 0.0003391111111111111,
|
7507 |
+
"loss": 2.2244,
|
7508 |
+
"step": 1237
|
7509 |
+
},
|
7510 |
+
{
|
7511 |
+
"epoch": 1.01,
|
7512 |
+
"learning_rate": 0.0003386666666666667,
|
7513 |
+
"loss": 2.2772,
|
7514 |
+
"step": 1238
|
7515 |
+
},
|
7516 |
+
{
|
7517 |
+
"epoch": 1.01,
|
7518 |
+
"learning_rate": 0.00033822222222222223,
|
7519 |
+
"loss": 2.2602,
|
7520 |
+
"step": 1239
|
7521 |
+
},
|
7522 |
+
{
|
7523 |
+
"epoch": 1.01,
|
7524 |
+
"learning_rate": 0.00033777777777777777,
|
7525 |
+
"loss": 2.2692,
|
7526 |
+
"step": 1240
|
7527 |
+
},
|
7528 |
+
{
|
7529 |
+
"epoch": 1.01,
|
7530 |
+
"learning_rate": 0.00033733333333333337,
|
7531 |
+
"loss": 2.2571,
|
7532 |
+
"step": 1241
|
7533 |
+
},
|
7534 |
+
{
|
7535 |
+
"epoch": 1.01,
|
7536 |
+
"learning_rate": 0.0003368888888888889,
|
7537 |
+
"loss": 2.1604,
|
7538 |
+
"step": 1242
|
7539 |
+
},
|
7540 |
+
{
|
7541 |
+
"epoch": 1.01,
|
7542 |
+
"learning_rate": 0.0003364444444444445,
|
7543 |
+
"loss": 2.2203,
|
7544 |
+
"step": 1243
|
7545 |
+
},
|
7546 |
+
{
|
7547 |
+
"epoch": 1.01,
|
7548 |
+
"learning_rate": 0.000336,
|
7549 |
+
"loss": 2.233,
|
7550 |
+
"step": 1244
|
7551 |
+
},
|
7552 |
+
{
|
7553 |
+
"epoch": 1.01,
|
7554 |
+
"learning_rate": 0.0003355555555555556,
|
7555 |
+
"loss": 2.2651,
|
7556 |
+
"step": 1245
|
7557 |
+
},
|
7558 |
+
{
|
7559 |
+
"epoch": 1.01,
|
7560 |
+
"learning_rate": 0.0003351111111111111,
|
7561 |
+
"loss": 2.1877,
|
7562 |
+
"step": 1246
|
7563 |
+
},
|
7564 |
+
{
|
7565 |
+
"epoch": 1.01,
|
7566 |
+
"learning_rate": 0.0003346666666666667,
|
7567 |
+
"loss": 2.2141,
|
7568 |
+
"step": 1247
|
7569 |
+
},
|
7570 |
+
{
|
7571 |
+
"epoch": 1.01,
|
7572 |
+
"learning_rate": 0.00033422222222222224,
|
7573 |
+
"loss": 2.2046,
|
7574 |
+
"step": 1248
|
7575 |
+
},
|
7576 |
+
{
|
7577 |
+
"epoch": 1.01,
|
7578 |
+
"learning_rate": 0.0003337777777777778,
|
7579 |
+
"loss": 2.188,
|
7580 |
+
"step": 1249
|
7581 |
+
},
|
7582 |
+
{
|
7583 |
+
"epoch": 1.01,
|
7584 |
+
"learning_rate": 0.0003333333333333334,
|
7585 |
+
"loss": 2.1651,
|
7586 |
+
"step": 1250
|
7587 |
+
},
|
7588 |
+
{
|
7589 |
+
"epoch": 1.01,
|
7590 |
+
"learning_rate": 0.0003328888888888889,
|
7591 |
+
"loss": 2.1521,
|
7592 |
+
"step": 1251
|
7593 |
+
},
|
7594 |
+
{
|
7595 |
+
"epoch": 1.01,
|
7596 |
+
"learning_rate": 0.00033244444444444445,
|
7597 |
+
"loss": 2.1967,
|
7598 |
+
"step": 1252
|
7599 |
+
},
|
7600 |
+
{
|
7601 |
+
"epoch": 1.01,
|
7602 |
+
"learning_rate": 0.000332,
|
7603 |
+
"loss": 2.1108,
|
7604 |
+
"step": 1253
|
7605 |
+
},
|
7606 |
+
{
|
7607 |
+
"epoch": 1.01,
|
7608 |
+
"learning_rate": 0.0003315555555555556,
|
7609 |
+
"loss": 2.1332,
|
7610 |
+
"step": 1254
|
7611 |
+
},
|
7612 |
+
{
|
7613 |
+
"epoch": 1.01,
|
7614 |
+
"learning_rate": 0.0003311111111111111,
|
7615 |
+
"loss": 2.2851,
|
7616 |
+
"step": 1255
|
7617 |
+
},
|
7618 |
+
{
|
7619 |
+
"epoch": 1.01,
|
7620 |
+
"learning_rate": 0.00033066666666666666,
|
7621 |
+
"loss": 2.254,
|
7622 |
+
"step": 1256
|
7623 |
+
},
|
7624 |
+
{
|
7625 |
+
"epoch": 1.02,
|
7626 |
+
"learning_rate": 0.00033022222222222226,
|
7627 |
+
"loss": 2.17,
|
7628 |
+
"step": 1257
|
7629 |
+
},
|
7630 |
+
{
|
7631 |
+
"epoch": 1.02,
|
7632 |
+
"learning_rate": 0.0003297777777777778,
|
7633 |
+
"loss": 2.1728,
|
7634 |
+
"step": 1258
|
7635 |
+
},
|
7636 |
+
{
|
7637 |
+
"epoch": 1.02,
|
7638 |
+
"learning_rate": 0.00032933333333333333,
|
7639 |
+
"loss": 2.154,
|
7640 |
+
"step": 1259
|
7641 |
+
},
|
7642 |
+
{
|
7643 |
+
"epoch": 1.02,
|
7644 |
+
"learning_rate": 0.00032888888888888887,
|
7645 |
+
"loss": 2.2122,
|
7646 |
+
"step": 1260
|
7647 |
+
},
|
7648 |
+
{
|
7649 |
+
"epoch": 1.02,
|
7650 |
+
"learning_rate": 0.00032844444444444447,
|
7651 |
+
"loss": 2.2088,
|
7652 |
+
"step": 1261
|
7653 |
+
},
|
7654 |
+
{
|
7655 |
+
"epoch": 1.02,
|
7656 |
+
"learning_rate": 0.000328,
|
7657 |
+
"loss": 2.2363,
|
7658 |
+
"step": 1262
|
7659 |
+
},
|
7660 |
+
{
|
7661 |
+
"epoch": 1.02,
|
7662 |
+
"learning_rate": 0.0003275555555555556,
|
7663 |
+
"loss": 2.1167,
|
7664 |
+
"step": 1263
|
7665 |
+
},
|
7666 |
+
{
|
7667 |
+
"epoch": 1.02,
|
7668 |
+
"learning_rate": 0.00032711111111111114,
|
7669 |
+
"loss": 2.0969,
|
7670 |
+
"step": 1264
|
7671 |
+
},
|
7672 |
+
{
|
7673 |
+
"epoch": 1.02,
|
7674 |
+
"learning_rate": 0.0003266666666666667,
|
7675 |
+
"loss": 2.197,
|
7676 |
+
"step": 1265
|
7677 |
+
},
|
7678 |
+
{
|
7679 |
+
"epoch": 1.02,
|
7680 |
+
"learning_rate": 0.00032622222222222227,
|
7681 |
+
"loss": 2.1962,
|
7682 |
+
"step": 1266
|
7683 |
+
},
|
7684 |
+
{
|
7685 |
+
"epoch": 1.02,
|
7686 |
+
"learning_rate": 0.0003257777777777778,
|
7687 |
+
"loss": 2.1862,
|
7688 |
+
"step": 1267
|
7689 |
+
},
|
7690 |
+
{
|
7691 |
+
"epoch": 1.02,
|
7692 |
+
"learning_rate": 0.00032533333333333334,
|
7693 |
+
"loss": 2.1808,
|
7694 |
+
"step": 1268
|
7695 |
+
},
|
7696 |
+
{
|
7697 |
+
"epoch": 1.02,
|
7698 |
+
"learning_rate": 0.0003248888888888889,
|
7699 |
+
"loss": 2.2553,
|
7700 |
+
"step": 1269
|
7701 |
+
},
|
7702 |
+
{
|
7703 |
+
"epoch": 1.02,
|
7704 |
+
"learning_rate": 0.0003244444444444445,
|
7705 |
+
"loss": 2.1425,
|
7706 |
+
"step": 1270
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 1.02,
|
7710 |
+
"learning_rate": 0.000324,
|
7711 |
+
"loss": 2.1588,
|
7712 |
+
"step": 1271
|
7713 |
+
},
|
7714 |
+
{
|
7715 |
+
"epoch": 1.02,
|
7716 |
+
"learning_rate": 0.00032355555555555555,
|
7717 |
+
"loss": 2.1203,
|
7718 |
+
"step": 1272
|
7719 |
+
},
|
7720 |
+
{
|
7721 |
+
"epoch": 1.02,
|
7722 |
+
"learning_rate": 0.00032311111111111115,
|
7723 |
+
"loss": 2.188,
|
7724 |
+
"step": 1273
|
7725 |
+
},
|
7726 |
+
{
|
7727 |
+
"epoch": 1.02,
|
7728 |
+
"learning_rate": 0.0003226666666666667,
|
7729 |
+
"loss": 2.1015,
|
7730 |
+
"step": 1274
|
7731 |
+
},
|
7732 |
+
{
|
7733 |
+
"epoch": 1.02,
|
7734 |
+
"learning_rate": 0.0003222222222222222,
|
7735 |
+
"loss": 2.1881,
|
7736 |
+
"step": 1275
|
7737 |
+
},
|
7738 |
+
{
|
7739 |
+
"epoch": 1.02,
|
7740 |
+
"learning_rate": 0.00032177777777777776,
|
7741 |
+
"loss": 2.2133,
|
7742 |
+
"step": 1276
|
7743 |
+
},
|
7744 |
+
{
|
7745 |
+
"epoch": 1.03,
|
7746 |
+
"learning_rate": 0.00032133333333333336,
|
7747 |
+
"loss": 2.181,
|
7748 |
+
"step": 1277
|
7749 |
+
},
|
7750 |
+
{
|
7751 |
+
"epoch": 1.03,
|
7752 |
+
"learning_rate": 0.0003208888888888889,
|
7753 |
+
"loss": 2.1,
|
7754 |
+
"step": 1278
|
7755 |
+
},
|
7756 |
+
{
|
7757 |
+
"epoch": 1.03,
|
7758 |
+
"learning_rate": 0.00032044444444444443,
|
7759 |
+
"loss": 2.1577,
|
7760 |
+
"step": 1279
|
7761 |
+
},
|
7762 |
+
{
|
7763 |
+
"epoch": 1.03,
|
7764 |
+
"learning_rate": 0.00032,
|
7765 |
+
"loss": 2.2327,
|
7766 |
+
"step": 1280
|
7767 |
+
},
|
7768 |
+
{
|
7769 |
+
"epoch": 1.03,
|
7770 |
+
"learning_rate": 0.00031955555555555556,
|
7771 |
+
"loss": 2.182,
|
7772 |
+
"step": 1281
|
7773 |
+
},
|
7774 |
+
{
|
7775 |
+
"epoch": 1.03,
|
7776 |
+
"learning_rate": 0.00031911111111111116,
|
7777 |
+
"loss": 2.1862,
|
7778 |
+
"step": 1282
|
7779 |
+
},
|
7780 |
+
{
|
7781 |
+
"epoch": 1.03,
|
7782 |
+
"learning_rate": 0.00031866666666666664,
|
7783 |
+
"loss": 2.1597,
|
7784 |
+
"step": 1283
|
7785 |
+
},
|
7786 |
+
{
|
7787 |
+
"epoch": 1.03,
|
7788 |
+
"learning_rate": 0.00031822222222222224,
|
7789 |
+
"loss": 2.2375,
|
7790 |
+
"step": 1284
|
7791 |
+
},
|
7792 |
+
{
|
7793 |
+
"epoch": 1.03,
|
7794 |
+
"learning_rate": 0.0003177777777777778,
|
7795 |
+
"loss": 2.1721,
|
7796 |
+
"step": 1285
|
7797 |
+
},
|
7798 |
+
{
|
7799 |
+
"epoch": 1.03,
|
7800 |
+
"learning_rate": 0.00031733333333333337,
|
7801 |
+
"loss": 2.1676,
|
7802 |
+
"step": 1286
|
7803 |
+
},
|
7804 |
+
{
|
7805 |
+
"epoch": 1.03,
|
7806 |
+
"learning_rate": 0.0003168888888888889,
|
7807 |
+
"loss": 2.1929,
|
7808 |
+
"step": 1287
|
7809 |
+
},
|
7810 |
+
{
|
7811 |
+
"epoch": 1.03,
|
7812 |
+
"learning_rate": 0.00031644444444444444,
|
7813 |
+
"loss": 2.1102,
|
7814 |
+
"step": 1288
|
7815 |
+
},
|
7816 |
+
{
|
7817 |
+
"epoch": 1.03,
|
7818 |
+
"learning_rate": 0.00031600000000000004,
|
7819 |
+
"loss": 2.199,
|
7820 |
+
"step": 1289
|
7821 |
+
},
|
7822 |
+
{
|
7823 |
+
"epoch": 1.03,
|
7824 |
+
"learning_rate": 0.0003155555555555556,
|
7825 |
+
"loss": 2.2004,
|
7826 |
+
"step": 1290
|
7827 |
+
},
|
7828 |
+
{
|
7829 |
+
"epoch": 1.03,
|
7830 |
+
"learning_rate": 0.0003151111111111111,
|
7831 |
+
"loss": 2.2512,
|
7832 |
+
"step": 1291
|
7833 |
+
},
|
7834 |
+
{
|
7835 |
+
"epoch": 1.03,
|
7836 |
+
"learning_rate": 0.00031466666666666665,
|
7837 |
+
"loss": 2.1994,
|
7838 |
+
"step": 1292
|
7839 |
+
},
|
7840 |
+
{
|
7841 |
+
"epoch": 1.03,
|
7842 |
+
"learning_rate": 0.00031422222222222225,
|
7843 |
+
"loss": 2.1508,
|
7844 |
+
"step": 1293
|
7845 |
+
},
|
7846 |
+
{
|
7847 |
+
"epoch": 1.03,
|
7848 |
+
"learning_rate": 0.00031377777777777784,
|
7849 |
+
"loss": 2.1607,
|
7850 |
+
"step": 1294
|
7851 |
+
},
|
7852 |
+
{
|
7853 |
+
"epoch": 1.03,
|
7854 |
+
"learning_rate": 0.0003133333333333333,
|
7855 |
+
"loss": 2.2171,
|
7856 |
+
"step": 1295
|
7857 |
+
},
|
7858 |
+
{
|
7859 |
+
"epoch": 1.03,
|
7860 |
+
"learning_rate": 0.0003128888888888889,
|
7861 |
+
"loss": 2.2,
|
7862 |
+
"step": 1296
|
7863 |
+
},
|
7864 |
+
{
|
7865 |
+
"epoch": 1.04,
|
7866 |
+
"learning_rate": 0.00031244444444444446,
|
7867 |
+
"loss": 2.1907,
|
7868 |
+
"step": 1297
|
7869 |
+
},
|
7870 |
+
{
|
7871 |
+
"epoch": 1.04,
|
7872 |
+
"learning_rate": 0.00031200000000000005,
|
7873 |
+
"loss": 2.2026,
|
7874 |
+
"step": 1298
|
7875 |
+
},
|
7876 |
+
{
|
7877 |
+
"epoch": 1.04,
|
7878 |
+
"learning_rate": 0.00031155555555555553,
|
7879 |
+
"loss": 2.176,
|
7880 |
+
"step": 1299
|
7881 |
+
},
|
7882 |
+
{
|
7883 |
+
"epoch": 1.04,
|
7884 |
+
"learning_rate": 0.0003111111111111111,
|
7885 |
+
"loss": 2.2456,
|
7886 |
+
"step": 1300
|
7887 |
+
},
|
7888 |
+
{
|
7889 |
+
"epoch": 1.04,
|
7890 |
+
"learning_rate": 0.00031066666666666666,
|
7891 |
+
"loss": 2.2479,
|
7892 |
+
"step": 1301
|
7893 |
+
},
|
7894 |
+
{
|
7895 |
+
"epoch": 1.04,
|
7896 |
+
"learning_rate": 0.00031022222222222226,
|
7897 |
+
"loss": 2.1432,
|
7898 |
+
"step": 1302
|
7899 |
+
},
|
7900 |
+
{
|
7901 |
+
"epoch": 1.04,
|
7902 |
+
"learning_rate": 0.0003097777777777778,
|
7903 |
+
"loss": 2.178,
|
7904 |
+
"step": 1303
|
7905 |
+
},
|
7906 |
+
{
|
7907 |
+
"epoch": 1.04,
|
7908 |
+
"learning_rate": 0.00030933333333333334,
|
7909 |
+
"loss": 2.1422,
|
7910 |
+
"step": 1304
|
7911 |
+
},
|
7912 |
+
{
|
7913 |
+
"epoch": 1.04,
|
7914 |
+
"learning_rate": 0.00030888888888888893,
|
7915 |
+
"loss": 2.1744,
|
7916 |
+
"step": 1305
|
7917 |
+
},
|
7918 |
+
{
|
7919 |
+
"epoch": 1.04,
|
7920 |
+
"learning_rate": 0.00030844444444444447,
|
7921 |
+
"loss": 2.2191,
|
7922 |
+
"step": 1306
|
7923 |
+
},
|
7924 |
+
{
|
7925 |
+
"epoch": 1.04,
|
7926 |
+
"learning_rate": 0.000308,
|
7927 |
+
"loss": 2.2665,
|
7928 |
+
"step": 1307
|
7929 |
+
},
|
7930 |
+
{
|
7931 |
+
"epoch": 1.04,
|
7932 |
+
"learning_rate": 0.00030755555555555554,
|
7933 |
+
"loss": 2.2301,
|
7934 |
+
"step": 1308
|
7935 |
+
},
|
7936 |
+
{
|
7937 |
+
"epoch": 1.04,
|
7938 |
+
"learning_rate": 0.00030711111111111114,
|
7939 |
+
"loss": 2.2328,
|
7940 |
+
"step": 1309
|
7941 |
+
},
|
7942 |
+
{
|
7943 |
+
"epoch": 1.04,
|
7944 |
+
"learning_rate": 0.0003066666666666667,
|
7945 |
+
"loss": 2.2027,
|
7946 |
+
"step": 1310
|
7947 |
+
},
|
7948 |
+
{
|
7949 |
+
"epoch": 1.04,
|
7950 |
+
"learning_rate": 0.0003062222222222222,
|
7951 |
+
"loss": 2.1684,
|
7952 |
+
"step": 1311
|
7953 |
+
},
|
7954 |
+
{
|
7955 |
+
"epoch": 1.04,
|
7956 |
+
"learning_rate": 0.0003057777777777778,
|
7957 |
+
"loss": 2.2214,
|
7958 |
+
"step": 1312
|
7959 |
+
},
|
7960 |
+
{
|
7961 |
+
"epoch": 1.04,
|
7962 |
+
"learning_rate": 0.00030533333333333335,
|
7963 |
+
"loss": 2.1323,
|
7964 |
+
"step": 1313
|
7965 |
+
},
|
7966 |
+
{
|
7967 |
+
"epoch": 1.04,
|
7968 |
+
"learning_rate": 0.00030488888888888894,
|
7969 |
+
"loss": 2.1567,
|
7970 |
+
"step": 1314
|
7971 |
+
},
|
7972 |
+
{
|
7973 |
+
"epoch": 1.04,
|
7974 |
+
"learning_rate": 0.0003044444444444444,
|
7975 |
+
"loss": 2.1194,
|
7976 |
+
"step": 1315
|
7977 |
+
},
|
7978 |
+
{
|
7979 |
+
"epoch": 1.04,
|
7980 |
+
"learning_rate": 0.000304,
|
7981 |
+
"loss": 2.1019,
|
7982 |
+
"step": 1316
|
7983 |
+
},
|
7984 |
+
{
|
7985 |
+
"epoch": 1.05,
|
7986 |
+
"learning_rate": 0.00030355555555555556,
|
7987 |
+
"loss": 2.15,
|
7988 |
+
"step": 1317
|
7989 |
+
},
|
7990 |
+
{
|
7991 |
+
"epoch": 1.05,
|
7992 |
+
"learning_rate": 0.00030311111111111115,
|
7993 |
+
"loss": 2.1954,
|
7994 |
+
"step": 1318
|
7995 |
+
},
|
7996 |
+
{
|
7997 |
+
"epoch": 1.05,
|
7998 |
+
"learning_rate": 0.0003026666666666667,
|
7999 |
+
"loss": 2.1297,
|
8000 |
+
"step": 1319
|
8001 |
+
},
|
8002 |
+
{
|
8003 |
+
"epoch": 1.05,
|
8004 |
+
"learning_rate": 0.0003022222222222222,
|
8005 |
+
"loss": 2.1343,
|
8006 |
+
"step": 1320
|
8007 |
+
},
|
8008 |
+
{
|
8009 |
+
"epoch": 1.05,
|
8010 |
+
"learning_rate": 0.0003017777777777778,
|
8011 |
+
"loss": 2.1918,
|
8012 |
+
"step": 1321
|
8013 |
+
},
|
8014 |
+
{
|
8015 |
+
"epoch": 1.05,
|
8016 |
+
"learning_rate": 0.00030133333333333336,
|
8017 |
+
"loss": 2.2325,
|
8018 |
+
"step": 1322
|
8019 |
+
},
|
8020 |
+
{
|
8021 |
+
"epoch": 1.05,
|
8022 |
+
"learning_rate": 0.0003008888888888889,
|
8023 |
+
"loss": 2.2012,
|
8024 |
+
"step": 1323
|
8025 |
+
},
|
8026 |
+
{
|
8027 |
+
"epoch": 1.05,
|
8028 |
+
"learning_rate": 0.00030044444444444443,
|
8029 |
+
"loss": 2.1262,
|
8030 |
+
"step": 1324
|
8031 |
+
},
|
8032 |
+
{
|
8033 |
+
"epoch": 1.05,
|
8034 |
+
"learning_rate": 0.00030000000000000003,
|
8035 |
+
"loss": 2.1162,
|
8036 |
+
"step": 1325
|
8037 |
+
},
|
8038 |
+
{
|
8039 |
+
"epoch": 1.05,
|
8040 |
+
"learning_rate": 0.00029955555555555557,
|
8041 |
+
"loss": 2.1746,
|
8042 |
+
"step": 1326
|
8043 |
+
},
|
8044 |
+
{
|
8045 |
+
"epoch": 1.05,
|
8046 |
+
"learning_rate": 0.0002991111111111111,
|
8047 |
+
"loss": 2.1809,
|
8048 |
+
"step": 1327
|
8049 |
+
},
|
8050 |
+
{
|
8051 |
+
"epoch": 1.05,
|
8052 |
+
"learning_rate": 0.0002986666666666667,
|
8053 |
+
"loss": 2.268,
|
8054 |
+
"step": 1328
|
8055 |
+
},
|
8056 |
+
{
|
8057 |
+
"epoch": 1.05,
|
8058 |
+
"learning_rate": 0.00029822222222222224,
|
8059 |
+
"loss": 2.2361,
|
8060 |
+
"step": 1329
|
8061 |
+
},
|
8062 |
+
{
|
8063 |
+
"epoch": 1.05,
|
8064 |
+
"learning_rate": 0.0002977777777777778,
|
8065 |
+
"loss": 2.1774,
|
8066 |
+
"step": 1330
|
8067 |
+
},
|
8068 |
+
{
|
8069 |
+
"epoch": 1.05,
|
8070 |
+
"learning_rate": 0.0002973333333333333,
|
8071 |
+
"loss": 2.1574,
|
8072 |
+
"step": 1331
|
8073 |
+
},
|
8074 |
+
{
|
8075 |
+
"epoch": 1.05,
|
8076 |
+
"learning_rate": 0.0002968888888888889,
|
8077 |
+
"loss": 2.1475,
|
8078 |
+
"step": 1332
|
8079 |
+
},
|
8080 |
+
{
|
8081 |
+
"epoch": 1.05,
|
8082 |
+
"learning_rate": 0.0002964444444444445,
|
8083 |
+
"loss": 2.1711,
|
8084 |
+
"step": 1333
|
8085 |
+
},
|
8086 |
+
{
|
8087 |
+
"epoch": 1.05,
|
8088 |
+
"learning_rate": 0.000296,
|
8089 |
+
"loss": 2.2065,
|
8090 |
+
"step": 1334
|
8091 |
+
},
|
8092 |
+
{
|
8093 |
+
"epoch": 1.05,
|
8094 |
+
"learning_rate": 0.0002955555555555556,
|
8095 |
+
"loss": 2.1746,
|
8096 |
+
"step": 1335
|
8097 |
+
},
|
8098 |
+
{
|
8099 |
+
"epoch": 1.05,
|
8100 |
+
"learning_rate": 0.0002951111111111111,
|
8101 |
+
"loss": 2.2004,
|
8102 |
+
"step": 1336
|
8103 |
+
},
|
8104 |
+
{
|
8105 |
+
"epoch": 1.06,
|
8106 |
+
"learning_rate": 0.0002946666666666667,
|
8107 |
+
"loss": 2.0863,
|
8108 |
+
"step": 1337
|
8109 |
+
},
|
8110 |
+
{
|
8111 |
+
"epoch": 1.06,
|
8112 |
+
"learning_rate": 0.00029422222222222225,
|
8113 |
+
"loss": 2.2187,
|
8114 |
+
"step": 1338
|
8115 |
+
},
|
8116 |
+
{
|
8117 |
+
"epoch": 1.06,
|
8118 |
+
"learning_rate": 0.0002937777777777778,
|
8119 |
+
"loss": 2.1756,
|
8120 |
+
"step": 1339
|
8121 |
+
},
|
8122 |
+
{
|
8123 |
+
"epoch": 1.06,
|
8124 |
+
"learning_rate": 0.0002933333333333333,
|
8125 |
+
"loss": 2.1669,
|
8126 |
+
"step": 1340
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 1.06,
|
8130 |
+
"learning_rate": 0.0002928888888888889,
|
8131 |
+
"loss": 2.0371,
|
8132 |
+
"step": 1341
|
8133 |
+
},
|
8134 |
+
{
|
8135 |
+
"epoch": 1.06,
|
8136 |
+
"learning_rate": 0.00029244444444444446,
|
8137 |
+
"loss": 2.2079,
|
8138 |
+
"step": 1342
|
8139 |
+
},
|
8140 |
+
{
|
8141 |
+
"epoch": 1.06,
|
8142 |
+
"learning_rate": 0.000292,
|
8143 |
+
"loss": 2.138,
|
8144 |
+
"step": 1343
|
8145 |
+
},
|
8146 |
+
{
|
8147 |
+
"epoch": 1.06,
|
8148 |
+
"learning_rate": 0.0002915555555555556,
|
8149 |
+
"loss": 2.1843,
|
8150 |
+
"step": 1344
|
8151 |
+
},
|
8152 |
+
{
|
8153 |
+
"epoch": 1.06,
|
8154 |
+
"learning_rate": 0.00029111111111111113,
|
8155 |
+
"loss": 2.0759,
|
8156 |
+
"step": 1345
|
8157 |
+
},
|
8158 |
+
{
|
8159 |
+
"epoch": 1.06,
|
8160 |
+
"learning_rate": 0.00029066666666666667,
|
8161 |
+
"loss": 2.1002,
|
8162 |
+
"step": 1346
|
8163 |
+
},
|
8164 |
+
{
|
8165 |
+
"epoch": 1.06,
|
8166 |
+
"learning_rate": 0.0002902222222222222,
|
8167 |
+
"loss": 2.1626,
|
8168 |
+
"step": 1347
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 1.06,
|
8172 |
+
"learning_rate": 0.0002897777777777778,
|
8173 |
+
"loss": 2.2071,
|
8174 |
+
"step": 1348
|
8175 |
+
},
|
8176 |
+
{
|
8177 |
+
"epoch": 1.06,
|
8178 |
+
"learning_rate": 0.0002893333333333334,
|
8179 |
+
"loss": 2.1772,
|
8180 |
+
"step": 1349
|
8181 |
+
},
|
8182 |
+
{
|
8183 |
+
"epoch": 1.06,
|
8184 |
+
"learning_rate": 0.0002888888888888889,
|
8185 |
+
"loss": 2.1284,
|
8186 |
+
"step": 1350
|
8187 |
+
},
|
8188 |
+
{
|
8189 |
+
"epoch": 1.06,
|
8190 |
+
"learning_rate": 0.00028844444444444447,
|
8191 |
+
"loss": 2.1774,
|
8192 |
+
"step": 1351
|
8193 |
+
},
|
8194 |
+
{
|
8195 |
+
"epoch": 1.06,
|
8196 |
+
"learning_rate": 0.000288,
|
8197 |
+
"loss": 2.1344,
|
8198 |
+
"step": 1352
|
8199 |
+
},
|
8200 |
+
{
|
8201 |
+
"epoch": 1.06,
|
8202 |
+
"learning_rate": 0.0002875555555555556,
|
8203 |
+
"loss": 2.1619,
|
8204 |
+
"step": 1353
|
8205 |
+
},
|
8206 |
+
{
|
8207 |
+
"epoch": 1.06,
|
8208 |
+
"learning_rate": 0.0002871111111111111,
|
8209 |
+
"loss": 2.1673,
|
8210 |
+
"step": 1354
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 1.06,
|
8214 |
+
"learning_rate": 0.0002866666666666667,
|
8215 |
+
"loss": 2.1395,
|
8216 |
+
"step": 1355
|
8217 |
+
},
|
8218 |
+
{
|
8219 |
+
"epoch": 1.06,
|
8220 |
+
"learning_rate": 0.0002862222222222222,
|
8221 |
+
"loss": 2.1698,
|
8222 |
+
"step": 1356
|
8223 |
+
},
|
8224 |
+
{
|
8225 |
+
"epoch": 1.07,
|
8226 |
+
"learning_rate": 0.0002857777777777778,
|
8227 |
+
"loss": 2.1331,
|
8228 |
+
"step": 1357
|
8229 |
+
},
|
8230 |
+
{
|
8231 |
+
"epoch": 1.07,
|
8232 |
+
"learning_rate": 0.00028533333333333335,
|
8233 |
+
"loss": 2.0731,
|
8234 |
+
"step": 1358
|
8235 |
+
},
|
8236 |
+
{
|
8237 |
+
"epoch": 1.07,
|
8238 |
+
"learning_rate": 0.0002848888888888889,
|
8239 |
+
"loss": 2.1612,
|
8240 |
+
"step": 1359
|
8241 |
+
},
|
8242 |
+
{
|
8243 |
+
"epoch": 1.07,
|
8244 |
+
"learning_rate": 0.0002844444444444445,
|
8245 |
+
"loss": 2.1686,
|
8246 |
+
"step": 1360
|
8247 |
+
},
|
8248 |
+
{
|
8249 |
+
"epoch": 1.07,
|
8250 |
+
"learning_rate": 0.000284,
|
8251 |
+
"loss": 2.1915,
|
8252 |
+
"step": 1361
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 1.07,
|
8256 |
+
"learning_rate": 0.00028355555555555556,
|
8257 |
+
"loss": 2.1867,
|
8258 |
+
"step": 1362
|
8259 |
+
},
|
8260 |
+
{
|
8261 |
+
"epoch": 1.07,
|
8262 |
+
"learning_rate": 0.0002831111111111111,
|
8263 |
+
"loss": 2.1668,
|
8264 |
+
"step": 1363
|
8265 |
+
},
|
8266 |
+
{
|
8267 |
+
"epoch": 1.07,
|
8268 |
+
"learning_rate": 0.0002826666666666667,
|
8269 |
+
"loss": 2.2493,
|
8270 |
+
"step": 1364
|
8271 |
+
},
|
8272 |
+
{
|
8273 |
+
"epoch": 1.07,
|
8274 |
+
"learning_rate": 0.00028222222222222223,
|
8275 |
+
"loss": 2.0671,
|
8276 |
+
"step": 1365
|
8277 |
+
},
|
8278 |
+
{
|
8279 |
+
"epoch": 1.07,
|
8280 |
+
"learning_rate": 0.00028177777777777777,
|
8281 |
+
"loss": 2.0957,
|
8282 |
+
"step": 1366
|
8283 |
+
},
|
8284 |
+
{
|
8285 |
+
"epoch": 1.07,
|
8286 |
+
"learning_rate": 0.00028133333333333336,
|
8287 |
+
"loss": 2.0841,
|
8288 |
+
"step": 1367
|
8289 |
+
},
|
8290 |
+
{
|
8291 |
+
"epoch": 1.07,
|
8292 |
+
"learning_rate": 0.0002808888888888889,
|
8293 |
+
"loss": 2.1165,
|
8294 |
+
"step": 1368
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 1.07,
|
8298 |
+
"learning_rate": 0.0002804444444444445,
|
8299 |
+
"loss": 2.1172,
|
8300 |
+
"step": 1369
|
8301 |
+
},
|
8302 |
+
{
|
8303 |
+
"epoch": 1.07,
|
8304 |
+
"learning_rate": 0.00028,
|
8305 |
+
"loss": 2.1452,
|
8306 |
+
"step": 1370
|
8307 |
+
},
|
8308 |
+
{
|
8309 |
+
"epoch": 1.07,
|
8310 |
+
"learning_rate": 0.00027955555555555557,
|
8311 |
+
"loss": 2.2362,
|
8312 |
+
"step": 1371
|
8313 |
+
},
|
8314 |
+
{
|
8315 |
+
"epoch": 1.07,
|
8316 |
+
"learning_rate": 0.00027911111111111116,
|
8317 |
+
"loss": 2.1393,
|
8318 |
+
"step": 1372
|
8319 |
+
},
|
8320 |
+
{
|
8321 |
+
"epoch": 1.07,
|
8322 |
+
"learning_rate": 0.0002786666666666667,
|
8323 |
+
"loss": 2.1432,
|
8324 |
+
"step": 1373
|
8325 |
+
},
|
8326 |
+
{
|
8327 |
+
"epoch": 1.07,
|
8328 |
+
"learning_rate": 0.00027822222222222224,
|
8329 |
+
"loss": 2.188,
|
8330 |
+
"step": 1374
|
8331 |
+
},
|
8332 |
+
{
|
8333 |
+
"epoch": 1.07,
|
8334 |
+
"learning_rate": 0.0002777777777777778,
|
8335 |
+
"loss": 2.0608,
|
8336 |
+
"step": 1375
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 1.07,
|
8340 |
+
"learning_rate": 0.00027733333333333337,
|
8341 |
+
"loss": 2.176,
|
8342 |
+
"step": 1376
|
8343 |
+
},
|
8344 |
+
{
|
8345 |
+
"epoch": 1.08,
|
8346 |
+
"learning_rate": 0.0002768888888888889,
|
8347 |
+
"loss": 2.1274,
|
8348 |
+
"step": 1377
|
8349 |
+
},
|
8350 |
+
{
|
8351 |
+
"epoch": 1.08,
|
8352 |
+
"learning_rate": 0.00027644444444444445,
|
8353 |
+
"loss": 2.1743,
|
8354 |
+
"step": 1378
|
8355 |
+
},
|
8356 |
+
{
|
8357 |
+
"epoch": 1.08,
|
8358 |
+
"learning_rate": 0.000276,
|
8359 |
+
"loss": 2.0862,
|
8360 |
+
"step": 1379
|
8361 |
+
},
|
8362 |
+
{
|
8363 |
+
"epoch": 1.08,
|
8364 |
+
"learning_rate": 0.0002755555555555556,
|
8365 |
+
"loss": 2.1069,
|
8366 |
+
"step": 1380
|
8367 |
+
},
|
8368 |
+
{
|
8369 |
+
"epoch": 1.08,
|
8370 |
+
"learning_rate": 0.0002751111111111111,
|
8371 |
+
"loss": 2.0847,
|
8372 |
+
"step": 1381
|
8373 |
+
},
|
8374 |
+
{
|
8375 |
+
"epoch": 1.08,
|
8376 |
+
"learning_rate": 0.00027466666666666666,
|
8377 |
+
"loss": 2.1519,
|
8378 |
+
"step": 1382
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 1.08,
|
8382 |
+
"learning_rate": 0.00027422222222222225,
|
8383 |
+
"loss": 2.1348,
|
8384 |
+
"step": 1383
|
8385 |
+
},
|
8386 |
+
{
|
8387 |
+
"epoch": 1.08,
|
8388 |
+
"learning_rate": 0.0002737777777777778,
|
8389 |
+
"loss": 2.1674,
|
8390 |
+
"step": 1384
|
8391 |
+
},
|
8392 |
+
{
|
8393 |
+
"epoch": 1.08,
|
8394 |
+
"learning_rate": 0.00027333333333333333,
|
8395 |
+
"loss": 2.1839,
|
8396 |
+
"step": 1385
|
8397 |
+
},
|
8398 |
+
{
|
8399 |
+
"epoch": 1.08,
|
8400 |
+
"learning_rate": 0.00027288888888888887,
|
8401 |
+
"loss": 2.1126,
|
8402 |
+
"step": 1386
|
8403 |
+
},
|
8404 |
+
{
|
8405 |
+
"epoch": 1.08,
|
8406 |
+
"learning_rate": 0.00027244444444444446,
|
8407 |
+
"loss": 2.1085,
|
8408 |
+
"step": 1387
|
8409 |
+
},
|
8410 |
+
{
|
8411 |
+
"epoch": 1.08,
|
8412 |
+
"learning_rate": 0.00027200000000000005,
|
8413 |
+
"loss": 2.1103,
|
8414 |
+
"step": 1388
|
8415 |
+
},
|
8416 |
+
{
|
8417 |
+
"epoch": 1.08,
|
8418 |
+
"learning_rate": 0.00027155555555555554,
|
8419 |
+
"loss": 2.0676,
|
8420 |
+
"step": 1389
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 1.08,
|
8424 |
+
"learning_rate": 0.00027111111111111113,
|
8425 |
+
"loss": 2.1052,
|
8426 |
+
"step": 1390
|
8427 |
+
},
|
8428 |
+
{
|
8429 |
+
"epoch": 1.08,
|
8430 |
+
"learning_rate": 0.00027066666666666667,
|
8431 |
+
"loss": 2.124,
|
8432 |
+
"step": 1391
|
8433 |
+
},
|
8434 |
+
{
|
8435 |
+
"epoch": 1.08,
|
8436 |
+
"learning_rate": 0.00027022222222222226,
|
8437 |
+
"loss": 2.1178,
|
8438 |
+
"step": 1392
|
8439 |
+
},
|
8440 |
+
{
|
8441 |
+
"epoch": 1.08,
|
8442 |
+
"learning_rate": 0.0002697777777777778,
|
8443 |
+
"loss": 2.0989,
|
8444 |
+
"step": 1393
|
8445 |
+
},
|
8446 |
+
{
|
8447 |
+
"epoch": 1.08,
|
8448 |
+
"learning_rate": 0.00026933333333333334,
|
8449 |
+
"loss": 2.0984,
|
8450 |
+
"step": 1394
|
8451 |
+
},
|
8452 |
+
{
|
8453 |
+
"epoch": 1.08,
|
8454 |
+
"learning_rate": 0.00026888888888888893,
|
8455 |
+
"loss": 2.1175,
|
8456 |
+
"step": 1395
|
8457 |
+
},
|
8458 |
+
{
|
8459 |
+
"epoch": 1.08,
|
8460 |
+
"learning_rate": 0.00026844444444444447,
|
8461 |
+
"loss": 2.1575,
|
8462 |
+
"step": 1396
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 1.09,
|
8466 |
+
"learning_rate": 0.000268,
|
8467 |
+
"loss": 2.1486,
|
8468 |
+
"step": 1397
|
8469 |
+
},
|
8470 |
+
{
|
8471 |
+
"epoch": 1.09,
|
8472 |
+
"learning_rate": 0.00026755555555555555,
|
8473 |
+
"loss": 2.0775,
|
8474 |
+
"step": 1398
|
8475 |
+
},
|
8476 |
+
{
|
8477 |
+
"epoch": 1.09,
|
8478 |
+
"learning_rate": 0.00026711111111111114,
|
8479 |
+
"loss": 2.2145,
|
8480 |
+
"step": 1399
|
8481 |
+
},
|
8482 |
+
{
|
8483 |
+
"epoch": 1.09,
|
8484 |
+
"learning_rate": 0.0002666666666666667,
|
8485 |
+
"loss": 2.1359,
|
8486 |
+
"step": 1400
|
8487 |
+
},
|
8488 |
+
{
|
8489 |
+
"epoch": 1.09,
|
8490 |
+
"eval_gen_len": 1023.0,
|
8491 |
+
"eval_loss": 2.1156013011932373,
|
8492 |
+
"eval_rouge1": 13.7423,
|
8493 |
+
"eval_rouge2": 4.7061,
|
8494 |
+
"eval_rougeL": 8.6172,
|
8495 |
+
"eval_rougeLsum": 8.6284,
|
8496 |
+
"eval_runtime": 9295.5684,
|
8497 |
+
"eval_samples_per_second": 0.054,
|
8498 |
+
"eval_steps_per_second": 0.054,
|
8499 |
+
"step": 1400
|
8500 |
}
|
8501 |
],
|
8502 |
"max_steps": 2000,
|
8503 |
"num_train_epochs": 9223372036854775807,
|
8504 |
+
"total_flos": 1.1246826661088164e+18,
|
8505 |
"trial_name": null,
|
8506 |
"trial_params": null
|
8507 |
}
|
pytorch_model.bin
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 2368281769
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:34e19f0a62fca8a6230a310e58105a9559ae0ec18159a3bf81179a7291941a60
|
3 |
size 2368281769
|