Training in progress, step 9521, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 368988278
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3b0f8d65cedaf44e4ac5bad3713ac7c3f605e9a0d2835b275497af127aebea82
|
3 |
size 368988278
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1107079290
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:5e7067080a90c90e2510ef46294c006830456dcdecf0678da97a2d600c67f826
|
3 |
size 1107079290
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1000
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:178f30be8493d5268a297ee3466f58f6af4b6a964eca4182cdc4fc0db2805cb6
|
3 |
size 1000
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 2500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5281,6 +5281,1420 @@
|
|
5281 |
"eval_samples_per_second": 1527.965,
|
5282 |
"eval_steps_per_second": 47.752,
|
5283 |
"step": 7500
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5284 |
}
|
5285 |
],
|
5286 |
"logging_steps": 10,
|
@@ -5295,12 +6709,12 @@
|
|
5295 |
"should_evaluate": false,
|
5296 |
"should_log": false,
|
5297 |
"should_save": true,
|
5298 |
-
"should_training_stop":
|
5299 |
},
|
5300 |
"attributes": {}
|
5301 |
}
|
5302 |
},
|
5303 |
-
"total_flos": 2.
|
5304 |
"train_batch_size": 4,
|
5305 |
"trial_name": null,
|
5306 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.9999491282051282,
|
5 |
"eval_steps": 2500,
|
6 |
+
"global_step": 9521,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5281 |
"eval_samples_per_second": 1527.965,
|
5282 |
"eval_steps_per_second": 47.752,
|
5283 |
"step": 7500
|
5284 |
+
},
|
5285 |
+
{
|
5286 |
+
"epoch": 0.788742564102564,
|
5287 |
+
"grad_norm": 78.0625,
|
5288 |
+
"learning_rate": 9.98459487179487e-07,
|
5289 |
+
"loss": 132.7844,
|
5290 |
+
"step": 7510
|
5291 |
+
},
|
5292 |
+
{
|
5293 |
+
"epoch": 0.7897928205128205,
|
5294 |
+
"grad_norm": 76.1875,
|
5295 |
+
"learning_rate": 9.984574358974358e-07,
|
5296 |
+
"loss": 132.6412,
|
5297 |
+
"step": 7520
|
5298 |
+
},
|
5299 |
+
{
|
5300 |
+
"epoch": 0.7908430769230769,
|
5301 |
+
"grad_norm": 79.4375,
|
5302 |
+
"learning_rate": 9.984553846153847e-07,
|
5303 |
+
"loss": 133.3601,
|
5304 |
+
"step": 7530
|
5305 |
+
},
|
5306 |
+
{
|
5307 |
+
"epoch": 0.7918933333333333,
|
5308 |
+
"grad_norm": 74.9375,
|
5309 |
+
"learning_rate": 9.984533333333332e-07,
|
5310 |
+
"loss": 133.7195,
|
5311 |
+
"step": 7540
|
5312 |
+
},
|
5313 |
+
{
|
5314 |
+
"epoch": 0.7929435897435897,
|
5315 |
+
"grad_norm": 76.1875,
|
5316 |
+
"learning_rate": 9.98451282051282e-07,
|
5317 |
+
"loss": 133.1729,
|
5318 |
+
"step": 7550
|
5319 |
+
},
|
5320 |
+
{
|
5321 |
+
"epoch": 0.7939938461538462,
|
5322 |
+
"grad_norm": 79.125,
|
5323 |
+
"learning_rate": 9.984492307692308e-07,
|
5324 |
+
"loss": 133.8151,
|
5325 |
+
"step": 7560
|
5326 |
+
},
|
5327 |
+
{
|
5328 |
+
"epoch": 0.7950441025641025,
|
5329 |
+
"grad_norm": 76.8125,
|
5330 |
+
"learning_rate": 9.984471794871795e-07,
|
5331 |
+
"loss": 134.1982,
|
5332 |
+
"step": 7570
|
5333 |
+
},
|
5334 |
+
{
|
5335 |
+
"epoch": 0.796094358974359,
|
5336 |
+
"grad_norm": 80.0,
|
5337 |
+
"learning_rate": 9.984451282051281e-07,
|
5338 |
+
"loss": 133.2489,
|
5339 |
+
"step": 7580
|
5340 |
+
},
|
5341 |
+
{
|
5342 |
+
"epoch": 0.7971446153846153,
|
5343 |
+
"grad_norm": 75.0,
|
5344 |
+
"learning_rate": 9.984430769230768e-07,
|
5345 |
+
"loss": 134.1791,
|
5346 |
+
"step": 7590
|
5347 |
+
},
|
5348 |
+
{
|
5349 |
+
"epoch": 0.7981948717948718,
|
5350 |
+
"grad_norm": 77.375,
|
5351 |
+
"learning_rate": 9.984410256410255e-07,
|
5352 |
+
"loss": 134.0118,
|
5353 |
+
"step": 7600
|
5354 |
+
},
|
5355 |
+
{
|
5356 |
+
"epoch": 0.7992451282051282,
|
5357 |
+
"grad_norm": 82.3125,
|
5358 |
+
"learning_rate": 9.984389743589742e-07,
|
5359 |
+
"loss": 132.9163,
|
5360 |
+
"step": 7610
|
5361 |
+
},
|
5362 |
+
{
|
5363 |
+
"epoch": 0.8002953846153846,
|
5364 |
+
"grad_norm": 78.25,
|
5365 |
+
"learning_rate": 9.984369230769231e-07,
|
5366 |
+
"loss": 133.4907,
|
5367 |
+
"step": 7620
|
5368 |
+
},
|
5369 |
+
{
|
5370 |
+
"epoch": 0.801345641025641,
|
5371 |
+
"grad_norm": 72.375,
|
5372 |
+
"learning_rate": 9.984348717948718e-07,
|
5373 |
+
"loss": 134.2763,
|
5374 |
+
"step": 7630
|
5375 |
+
},
|
5376 |
+
{
|
5377 |
+
"epoch": 0.8023958974358975,
|
5378 |
+
"grad_norm": 77.0625,
|
5379 |
+
"learning_rate": 9.984328205128205e-07,
|
5380 |
+
"loss": 133.4557,
|
5381 |
+
"step": 7640
|
5382 |
+
},
|
5383 |
+
{
|
5384 |
+
"epoch": 0.8034461538461538,
|
5385 |
+
"grad_norm": 75.4375,
|
5386 |
+
"learning_rate": 9.984307692307692e-07,
|
5387 |
+
"loss": 133.6842,
|
5388 |
+
"step": 7650
|
5389 |
+
},
|
5390 |
+
{
|
5391 |
+
"epoch": 0.8044964102564103,
|
5392 |
+
"grad_norm": 74.75,
|
5393 |
+
"learning_rate": 9.98428717948718e-07,
|
5394 |
+
"loss": 133.6221,
|
5395 |
+
"step": 7660
|
5396 |
+
},
|
5397 |
+
{
|
5398 |
+
"epoch": 0.8055466666666666,
|
5399 |
+
"grad_norm": 74.1875,
|
5400 |
+
"learning_rate": 9.984266666666666e-07,
|
5401 |
+
"loss": 133.9818,
|
5402 |
+
"step": 7670
|
5403 |
+
},
|
5404 |
+
{
|
5405 |
+
"epoch": 0.8065969230769231,
|
5406 |
+
"grad_norm": 80.9375,
|
5407 |
+
"learning_rate": 9.984246153846153e-07,
|
5408 |
+
"loss": 134.001,
|
5409 |
+
"step": 7680
|
5410 |
+
},
|
5411 |
+
{
|
5412 |
+
"epoch": 0.8076471794871795,
|
5413 |
+
"grad_norm": 73.4375,
|
5414 |
+
"learning_rate": 9.98422564102564e-07,
|
5415 |
+
"loss": 134.0589,
|
5416 |
+
"step": 7690
|
5417 |
+
},
|
5418 |
+
{
|
5419 |
+
"epoch": 0.8086974358974359,
|
5420 |
+
"grad_norm": 74.3125,
|
5421 |
+
"learning_rate": 9.984205128205127e-07,
|
5422 |
+
"loss": 133.7496,
|
5423 |
+
"step": 7700
|
5424 |
+
},
|
5425 |
+
{
|
5426 |
+
"epoch": 0.8097476923076923,
|
5427 |
+
"grad_norm": 74.3125,
|
5428 |
+
"learning_rate": 9.984184615384616e-07,
|
5429 |
+
"loss": 133.3738,
|
5430 |
+
"step": 7710
|
5431 |
+
},
|
5432 |
+
{
|
5433 |
+
"epoch": 0.8107979487179487,
|
5434 |
+
"grad_norm": 72.0,
|
5435 |
+
"learning_rate": 9.984164102564103e-07,
|
5436 |
+
"loss": 133.1667,
|
5437 |
+
"step": 7720
|
5438 |
+
},
|
5439 |
+
{
|
5440 |
+
"epoch": 0.8118482051282051,
|
5441 |
+
"grad_norm": 75.875,
|
5442 |
+
"learning_rate": 9.98414358974359e-07,
|
5443 |
+
"loss": 132.9567,
|
5444 |
+
"step": 7730
|
5445 |
+
},
|
5446 |
+
{
|
5447 |
+
"epoch": 0.8128984615384616,
|
5448 |
+
"grad_norm": 73.3125,
|
5449 |
+
"learning_rate": 9.984123076923077e-07,
|
5450 |
+
"loss": 134.7816,
|
5451 |
+
"step": 7740
|
5452 |
+
},
|
5453 |
+
{
|
5454 |
+
"epoch": 0.8139487179487179,
|
5455 |
+
"grad_norm": 75.875,
|
5456 |
+
"learning_rate": 9.984102564102564e-07,
|
5457 |
+
"loss": 133.8593,
|
5458 |
+
"step": 7750
|
5459 |
+
},
|
5460 |
+
{
|
5461 |
+
"epoch": 0.8149989743589744,
|
5462 |
+
"grad_norm": 78.5625,
|
5463 |
+
"learning_rate": 9.98408205128205e-07,
|
5464 |
+
"loss": 133.8173,
|
5465 |
+
"step": 7760
|
5466 |
+
},
|
5467 |
+
{
|
5468 |
+
"epoch": 0.8160492307692307,
|
5469 |
+
"grad_norm": 72.25,
|
5470 |
+
"learning_rate": 9.984061538461538e-07,
|
5471 |
+
"loss": 133.5827,
|
5472 |
+
"step": 7770
|
5473 |
+
},
|
5474 |
+
{
|
5475 |
+
"epoch": 0.8170994871794872,
|
5476 |
+
"grad_norm": 77.6875,
|
5477 |
+
"learning_rate": 9.984041025641027e-07,
|
5478 |
+
"loss": 132.9576,
|
5479 |
+
"step": 7780
|
5480 |
+
},
|
5481 |
+
{
|
5482 |
+
"epoch": 0.8181497435897436,
|
5483 |
+
"grad_norm": 71.6875,
|
5484 |
+
"learning_rate": 9.984020512820511e-07,
|
5485 |
+
"loss": 133.3698,
|
5486 |
+
"step": 7790
|
5487 |
+
},
|
5488 |
+
{
|
5489 |
+
"epoch": 0.8192,
|
5490 |
+
"grad_norm": 72.8125,
|
5491 |
+
"learning_rate": 9.983999999999998e-07,
|
5492 |
+
"loss": 133.9683,
|
5493 |
+
"step": 7800
|
5494 |
+
},
|
5495 |
+
{
|
5496 |
+
"epoch": 0.8202502564102564,
|
5497 |
+
"grad_norm": 73.25,
|
5498 |
+
"learning_rate": 9.983979487179487e-07,
|
5499 |
+
"loss": 133.8763,
|
5500 |
+
"step": 7810
|
5501 |
+
},
|
5502 |
+
{
|
5503 |
+
"epoch": 0.8213005128205129,
|
5504 |
+
"grad_norm": 71.875,
|
5505 |
+
"learning_rate": 9.983958974358974e-07,
|
5506 |
+
"loss": 132.3762,
|
5507 |
+
"step": 7820
|
5508 |
+
},
|
5509 |
+
{
|
5510 |
+
"epoch": 0.8223507692307692,
|
5511 |
+
"grad_norm": 77.3125,
|
5512 |
+
"learning_rate": 9.983938461538461e-07,
|
5513 |
+
"loss": 133.6236,
|
5514 |
+
"step": 7830
|
5515 |
+
},
|
5516 |
+
{
|
5517 |
+
"epoch": 0.8234010256410257,
|
5518 |
+
"grad_norm": 83.8125,
|
5519 |
+
"learning_rate": 9.983917948717948e-07,
|
5520 |
+
"loss": 132.3988,
|
5521 |
+
"step": 7840
|
5522 |
+
},
|
5523 |
+
{
|
5524 |
+
"epoch": 0.824451282051282,
|
5525 |
+
"grad_norm": 73.375,
|
5526 |
+
"learning_rate": 9.983897435897435e-07,
|
5527 |
+
"loss": 133.4139,
|
5528 |
+
"step": 7850
|
5529 |
+
},
|
5530 |
+
{
|
5531 |
+
"epoch": 0.8255015384615385,
|
5532 |
+
"grad_norm": 79.3125,
|
5533 |
+
"learning_rate": 9.983876923076922e-07,
|
5534 |
+
"loss": 134.3546,
|
5535 |
+
"step": 7860
|
5536 |
+
},
|
5537 |
+
{
|
5538 |
+
"epoch": 0.8265517948717949,
|
5539 |
+
"grad_norm": 78.4375,
|
5540 |
+
"learning_rate": 9.98385641025641e-07,
|
5541 |
+
"loss": 134.4989,
|
5542 |
+
"step": 7870
|
5543 |
+
},
|
5544 |
+
{
|
5545 |
+
"epoch": 0.8276020512820513,
|
5546 |
+
"grad_norm": 78.0,
|
5547 |
+
"learning_rate": 9.983835897435898e-07,
|
5548 |
+
"loss": 133.5446,
|
5549 |
+
"step": 7880
|
5550 |
+
},
|
5551 |
+
{
|
5552 |
+
"epoch": 0.8286523076923077,
|
5553 |
+
"grad_norm": 77.5,
|
5554 |
+
"learning_rate": 9.983815384615383e-07,
|
5555 |
+
"loss": 133.1525,
|
5556 |
+
"step": 7890
|
5557 |
+
},
|
5558 |
+
{
|
5559 |
+
"epoch": 0.8297025641025642,
|
5560 |
+
"grad_norm": 84.1875,
|
5561 |
+
"learning_rate": 9.983794871794872e-07,
|
5562 |
+
"loss": 133.6593,
|
5563 |
+
"step": 7900
|
5564 |
+
},
|
5565 |
+
{
|
5566 |
+
"epoch": 0.8307528205128205,
|
5567 |
+
"grad_norm": 74.1875,
|
5568 |
+
"learning_rate": 9.983774358974359e-07,
|
5569 |
+
"loss": 133.9224,
|
5570 |
+
"step": 7910
|
5571 |
+
},
|
5572 |
+
{
|
5573 |
+
"epoch": 0.831803076923077,
|
5574 |
+
"grad_norm": 76.0,
|
5575 |
+
"learning_rate": 9.983753846153846e-07,
|
5576 |
+
"loss": 134.1779,
|
5577 |
+
"step": 7920
|
5578 |
+
},
|
5579 |
+
{
|
5580 |
+
"epoch": 0.8328533333333333,
|
5581 |
+
"grad_norm": 73.8125,
|
5582 |
+
"learning_rate": 9.983733333333333e-07,
|
5583 |
+
"loss": 135.2825,
|
5584 |
+
"step": 7930
|
5585 |
+
},
|
5586 |
+
{
|
5587 |
+
"epoch": 0.8339035897435897,
|
5588 |
+
"grad_norm": 72.875,
|
5589 |
+
"learning_rate": 9.98371282051282e-07,
|
5590 |
+
"loss": 133.2882,
|
5591 |
+
"step": 7940
|
5592 |
+
},
|
5593 |
+
{
|
5594 |
+
"epoch": 0.8349538461538462,
|
5595 |
+
"grad_norm": 72.4375,
|
5596 |
+
"learning_rate": 9.983692307692307e-07,
|
5597 |
+
"loss": 134.7268,
|
5598 |
+
"step": 7950
|
5599 |
+
},
|
5600 |
+
{
|
5601 |
+
"epoch": 0.8360041025641025,
|
5602 |
+
"grad_norm": 78.8125,
|
5603 |
+
"learning_rate": 9.983671794871794e-07,
|
5604 |
+
"loss": 133.8474,
|
5605 |
+
"step": 7960
|
5606 |
+
},
|
5607 |
+
{
|
5608 |
+
"epoch": 0.837054358974359,
|
5609 |
+
"grad_norm": 74.5,
|
5610 |
+
"learning_rate": 9.983651282051283e-07,
|
5611 |
+
"loss": 131.9032,
|
5612 |
+
"step": 7970
|
5613 |
+
},
|
5614 |
+
{
|
5615 |
+
"epoch": 0.8381046153846153,
|
5616 |
+
"grad_norm": 77.5625,
|
5617 |
+
"learning_rate": 9.98363076923077e-07,
|
5618 |
+
"loss": 134.0065,
|
5619 |
+
"step": 7980
|
5620 |
+
},
|
5621 |
+
{
|
5622 |
+
"epoch": 0.8391548717948718,
|
5623 |
+
"grad_norm": 78.25,
|
5624 |
+
"learning_rate": 9.983610256410257e-07,
|
5625 |
+
"loss": 133.8622,
|
5626 |
+
"step": 7990
|
5627 |
+
},
|
5628 |
+
{
|
5629 |
+
"epoch": 0.8402051282051282,
|
5630 |
+
"grad_norm": 76.3125,
|
5631 |
+
"learning_rate": 9.983589743589743e-07,
|
5632 |
+
"loss": 133.5396,
|
5633 |
+
"step": 8000
|
5634 |
+
},
|
5635 |
+
{
|
5636 |
+
"epoch": 0.8412553846153846,
|
5637 |
+
"grad_norm": 83.25,
|
5638 |
+
"learning_rate": 9.98356923076923e-07,
|
5639 |
+
"loss": 134.0301,
|
5640 |
+
"step": 8010
|
5641 |
+
},
|
5642 |
+
{
|
5643 |
+
"epoch": 0.842305641025641,
|
5644 |
+
"grad_norm": 77.5625,
|
5645 |
+
"learning_rate": 9.983548717948717e-07,
|
5646 |
+
"loss": 133.8874,
|
5647 |
+
"step": 8020
|
5648 |
+
},
|
5649 |
+
{
|
5650 |
+
"epoch": 0.8433558974358975,
|
5651 |
+
"grad_norm": 76.25,
|
5652 |
+
"learning_rate": 9.983528205128204e-07,
|
5653 |
+
"loss": 133.7713,
|
5654 |
+
"step": 8030
|
5655 |
+
},
|
5656 |
+
{
|
5657 |
+
"epoch": 0.8444061538461538,
|
5658 |
+
"grad_norm": 85.25,
|
5659 |
+
"learning_rate": 9.983507692307691e-07,
|
5660 |
+
"loss": 132.7225,
|
5661 |
+
"step": 8040
|
5662 |
+
},
|
5663 |
+
{
|
5664 |
+
"epoch": 0.8454564102564103,
|
5665 |
+
"grad_norm": 77.9375,
|
5666 |
+
"learning_rate": 9.983487179487178e-07,
|
5667 |
+
"loss": 132.4973,
|
5668 |
+
"step": 8050
|
5669 |
+
},
|
5670 |
+
{
|
5671 |
+
"epoch": 0.8465066666666666,
|
5672 |
+
"grad_norm": 77.5,
|
5673 |
+
"learning_rate": 9.983466666666667e-07,
|
5674 |
+
"loss": 134.7827,
|
5675 |
+
"step": 8060
|
5676 |
+
},
|
5677 |
+
{
|
5678 |
+
"epoch": 0.8475569230769231,
|
5679 |
+
"grad_norm": 78.9375,
|
5680 |
+
"learning_rate": 9.983446153846154e-07,
|
5681 |
+
"loss": 134.2993,
|
5682 |
+
"step": 8070
|
5683 |
+
},
|
5684 |
+
{
|
5685 |
+
"epoch": 0.8486071794871795,
|
5686 |
+
"grad_norm": 75.875,
|
5687 |
+
"learning_rate": 9.983425641025641e-07,
|
5688 |
+
"loss": 132.517,
|
5689 |
+
"step": 8080
|
5690 |
+
},
|
5691 |
+
{
|
5692 |
+
"epoch": 0.8496574358974359,
|
5693 |
+
"grad_norm": 72.8125,
|
5694 |
+
"learning_rate": 9.983405128205128e-07,
|
5695 |
+
"loss": 132.6084,
|
5696 |
+
"step": 8090
|
5697 |
+
},
|
5698 |
+
{
|
5699 |
+
"epoch": 0.8507076923076923,
|
5700 |
+
"grad_norm": 75.1875,
|
5701 |
+
"learning_rate": 9.983384615384615e-07,
|
5702 |
+
"loss": 133.5499,
|
5703 |
+
"step": 8100
|
5704 |
+
},
|
5705 |
+
{
|
5706 |
+
"epoch": 0.8517579487179487,
|
5707 |
+
"grad_norm": 76.0,
|
5708 |
+
"learning_rate": 9.983364102564102e-07,
|
5709 |
+
"loss": 131.2289,
|
5710 |
+
"step": 8110
|
5711 |
+
},
|
5712 |
+
{
|
5713 |
+
"epoch": 0.8528082051282051,
|
5714 |
+
"grad_norm": 74.625,
|
5715 |
+
"learning_rate": 9.983343589743589e-07,
|
5716 |
+
"loss": 133.3548,
|
5717 |
+
"step": 8120
|
5718 |
+
},
|
5719 |
+
{
|
5720 |
+
"epoch": 0.8538584615384616,
|
5721 |
+
"grad_norm": 80.9375,
|
5722 |
+
"learning_rate": 9.983323076923078e-07,
|
5723 |
+
"loss": 132.2297,
|
5724 |
+
"step": 8130
|
5725 |
+
},
|
5726 |
+
{
|
5727 |
+
"epoch": 0.8549087179487179,
|
5728 |
+
"grad_norm": 73.125,
|
5729 |
+
"learning_rate": 9.983302564102563e-07,
|
5730 |
+
"loss": 133.2755,
|
5731 |
+
"step": 8140
|
5732 |
+
},
|
5733 |
+
{
|
5734 |
+
"epoch": 0.8559589743589744,
|
5735 |
+
"grad_norm": 75.3125,
|
5736 |
+
"learning_rate": 9.98328205128205e-07,
|
5737 |
+
"loss": 134.9403,
|
5738 |
+
"step": 8150
|
5739 |
+
},
|
5740 |
+
{
|
5741 |
+
"epoch": 0.8570092307692307,
|
5742 |
+
"grad_norm": 73.75,
|
5743 |
+
"learning_rate": 9.983261538461539e-07,
|
5744 |
+
"loss": 133.4854,
|
5745 |
+
"step": 8160
|
5746 |
+
},
|
5747 |
+
{
|
5748 |
+
"epoch": 0.8580594871794872,
|
5749 |
+
"grad_norm": 75.4375,
|
5750 |
+
"learning_rate": 9.983241025641026e-07,
|
5751 |
+
"loss": 133.9221,
|
5752 |
+
"step": 8170
|
5753 |
+
},
|
5754 |
+
{
|
5755 |
+
"epoch": 0.8591097435897436,
|
5756 |
+
"grad_norm": 73.0625,
|
5757 |
+
"learning_rate": 9.983220512820513e-07,
|
5758 |
+
"loss": 133.0202,
|
5759 |
+
"step": 8180
|
5760 |
+
},
|
5761 |
+
{
|
5762 |
+
"epoch": 0.86016,
|
5763 |
+
"grad_norm": 77.8125,
|
5764 |
+
"learning_rate": 9.9832e-07,
|
5765 |
+
"loss": 133.9612,
|
5766 |
+
"step": 8190
|
5767 |
+
},
|
5768 |
+
{
|
5769 |
+
"epoch": 0.8612102564102564,
|
5770 |
+
"grad_norm": 74.75,
|
5771 |
+
"learning_rate": 9.983179487179486e-07,
|
5772 |
+
"loss": 132.866,
|
5773 |
+
"step": 8200
|
5774 |
+
},
|
5775 |
+
{
|
5776 |
+
"epoch": 0.8622605128205129,
|
5777 |
+
"grad_norm": 78.6875,
|
5778 |
+
"learning_rate": 9.983158974358973e-07,
|
5779 |
+
"loss": 131.9501,
|
5780 |
+
"step": 8210
|
5781 |
+
},
|
5782 |
+
{
|
5783 |
+
"epoch": 0.8633107692307692,
|
5784 |
+
"grad_norm": 72.5,
|
5785 |
+
"learning_rate": 9.98313846153846e-07,
|
5786 |
+
"loss": 133.986,
|
5787 |
+
"step": 8220
|
5788 |
+
},
|
5789 |
+
{
|
5790 |
+
"epoch": 0.8643610256410257,
|
5791 |
+
"grad_norm": 77.5625,
|
5792 |
+
"learning_rate": 9.98311794871795e-07,
|
5793 |
+
"loss": 134.4183,
|
5794 |
+
"step": 8230
|
5795 |
+
},
|
5796 |
+
{
|
5797 |
+
"epoch": 0.865411282051282,
|
5798 |
+
"grad_norm": 79.1875,
|
5799 |
+
"learning_rate": 9.983097435897434e-07,
|
5800 |
+
"loss": 133.7371,
|
5801 |
+
"step": 8240
|
5802 |
+
},
|
5803 |
+
{
|
5804 |
+
"epoch": 0.8664615384615385,
|
5805 |
+
"grad_norm": 83.9375,
|
5806 |
+
"learning_rate": 9.983076923076923e-07,
|
5807 |
+
"loss": 131.5742,
|
5808 |
+
"step": 8250
|
5809 |
+
},
|
5810 |
+
{
|
5811 |
+
"epoch": 0.8675117948717949,
|
5812 |
+
"grad_norm": 78.125,
|
5813 |
+
"learning_rate": 9.98305641025641e-07,
|
5814 |
+
"loss": 133.3313,
|
5815 |
+
"step": 8260
|
5816 |
+
},
|
5817 |
+
{
|
5818 |
+
"epoch": 0.8685620512820513,
|
5819 |
+
"grad_norm": 81.25,
|
5820 |
+
"learning_rate": 9.983035897435897e-07,
|
5821 |
+
"loss": 133.0588,
|
5822 |
+
"step": 8270
|
5823 |
+
},
|
5824 |
+
{
|
5825 |
+
"epoch": 0.8696123076923077,
|
5826 |
+
"grad_norm": 74.875,
|
5827 |
+
"learning_rate": 9.983015384615384e-07,
|
5828 |
+
"loss": 132.9553,
|
5829 |
+
"step": 8280
|
5830 |
+
},
|
5831 |
+
{
|
5832 |
+
"epoch": 0.8706625641025642,
|
5833 |
+
"grad_norm": 76.25,
|
5834 |
+
"learning_rate": 9.982994871794871e-07,
|
5835 |
+
"loss": 132.9017,
|
5836 |
+
"step": 8290
|
5837 |
+
},
|
5838 |
+
{
|
5839 |
+
"epoch": 0.8717128205128205,
|
5840 |
+
"grad_norm": 80.25,
|
5841 |
+
"learning_rate": 9.982974358974358e-07,
|
5842 |
+
"loss": 132.7887,
|
5843 |
+
"step": 8300
|
5844 |
+
},
|
5845 |
+
{
|
5846 |
+
"epoch": 0.872763076923077,
|
5847 |
+
"grad_norm": 82.5,
|
5848 |
+
"learning_rate": 9.982953846153845e-07,
|
5849 |
+
"loss": 131.9362,
|
5850 |
+
"step": 8310
|
5851 |
+
},
|
5852 |
+
{
|
5853 |
+
"epoch": 0.8738133333333333,
|
5854 |
+
"grad_norm": 74.6875,
|
5855 |
+
"learning_rate": 9.982933333333334e-07,
|
5856 |
+
"loss": 133.4346,
|
5857 |
+
"step": 8320
|
5858 |
+
},
|
5859 |
+
{
|
5860 |
+
"epoch": 0.8748635897435898,
|
5861 |
+
"grad_norm": 74.4375,
|
5862 |
+
"learning_rate": 9.98291282051282e-07,
|
5863 |
+
"loss": 133.0578,
|
5864 |
+
"step": 8330
|
5865 |
+
},
|
5866 |
+
{
|
5867 |
+
"epoch": 0.8759138461538462,
|
5868 |
+
"grad_norm": 74.75,
|
5869 |
+
"learning_rate": 9.982892307692308e-07,
|
5870 |
+
"loss": 133.309,
|
5871 |
+
"step": 8340
|
5872 |
+
},
|
5873 |
+
{
|
5874 |
+
"epoch": 0.8769641025641025,
|
5875 |
+
"grad_norm": 78.125,
|
5876 |
+
"learning_rate": 9.982871794871795e-07,
|
5877 |
+
"loss": 134.0469,
|
5878 |
+
"step": 8350
|
5879 |
+
},
|
5880 |
+
{
|
5881 |
+
"epoch": 0.878014358974359,
|
5882 |
+
"grad_norm": 82.125,
|
5883 |
+
"learning_rate": 9.982851282051282e-07,
|
5884 |
+
"loss": 131.7594,
|
5885 |
+
"step": 8360
|
5886 |
+
},
|
5887 |
+
{
|
5888 |
+
"epoch": 0.8790646153846153,
|
5889 |
+
"grad_norm": 80.9375,
|
5890 |
+
"learning_rate": 9.982830769230769e-07,
|
5891 |
+
"loss": 134.0729,
|
5892 |
+
"step": 8370
|
5893 |
+
},
|
5894 |
+
{
|
5895 |
+
"epoch": 0.8801148717948718,
|
5896 |
+
"grad_norm": 80.75,
|
5897 |
+
"learning_rate": 9.982810256410256e-07,
|
5898 |
+
"loss": 133.4013,
|
5899 |
+
"step": 8380
|
5900 |
+
},
|
5901 |
+
{
|
5902 |
+
"epoch": 0.8811651282051282,
|
5903 |
+
"grad_norm": 74.6875,
|
5904 |
+
"learning_rate": 9.982789743589743e-07,
|
5905 |
+
"loss": 132.9633,
|
5906 |
+
"step": 8390
|
5907 |
+
},
|
5908 |
+
{
|
5909 |
+
"epoch": 0.8822153846153846,
|
5910 |
+
"grad_norm": 74.3125,
|
5911 |
+
"learning_rate": 9.98276923076923e-07,
|
5912 |
+
"loss": 133.8356,
|
5913 |
+
"step": 8400
|
5914 |
+
},
|
5915 |
+
{
|
5916 |
+
"epoch": 0.883265641025641,
|
5917 |
+
"grad_norm": 75.0,
|
5918 |
+
"learning_rate": 9.982748717948719e-07,
|
5919 |
+
"loss": 133.2425,
|
5920 |
+
"step": 8410
|
5921 |
+
},
|
5922 |
+
{
|
5923 |
+
"epoch": 0.8843158974358974,
|
5924 |
+
"grad_norm": 73.25,
|
5925 |
+
"learning_rate": 9.982728205128206e-07,
|
5926 |
+
"loss": 132.7144,
|
5927 |
+
"step": 8420
|
5928 |
+
},
|
5929 |
+
{
|
5930 |
+
"epoch": 0.8853661538461538,
|
5931 |
+
"grad_norm": 78.625,
|
5932 |
+
"learning_rate": 9.98270769230769e-07,
|
5933 |
+
"loss": 134.2052,
|
5934 |
+
"step": 8430
|
5935 |
+
},
|
5936 |
+
{
|
5937 |
+
"epoch": 0.8864164102564103,
|
5938 |
+
"grad_norm": 71.0,
|
5939 |
+
"learning_rate": 9.98268717948718e-07,
|
5940 |
+
"loss": 133.7353,
|
5941 |
+
"step": 8440
|
5942 |
+
},
|
5943 |
+
{
|
5944 |
+
"epoch": 0.8874666666666666,
|
5945 |
+
"grad_norm": 77.3125,
|
5946 |
+
"learning_rate": 9.982666666666666e-07,
|
5947 |
+
"loss": 134.3944,
|
5948 |
+
"step": 8450
|
5949 |
+
},
|
5950 |
+
{
|
5951 |
+
"epoch": 0.8885169230769231,
|
5952 |
+
"grad_norm": 79.5,
|
5953 |
+
"learning_rate": 9.982646153846153e-07,
|
5954 |
+
"loss": 134.6475,
|
5955 |
+
"step": 8460
|
5956 |
+
},
|
5957 |
+
{
|
5958 |
+
"epoch": 0.8895671794871794,
|
5959 |
+
"grad_norm": 78.375,
|
5960 |
+
"learning_rate": 9.98262564102564e-07,
|
5961 |
+
"loss": 133.3185,
|
5962 |
+
"step": 8470
|
5963 |
+
},
|
5964 |
+
{
|
5965 |
+
"epoch": 0.8906174358974359,
|
5966 |
+
"grad_norm": 83.3125,
|
5967 |
+
"learning_rate": 9.98260512820513e-07,
|
5968 |
+
"loss": 133.362,
|
5969 |
+
"step": 8480
|
5970 |
+
},
|
5971 |
+
{
|
5972 |
+
"epoch": 0.8916676923076923,
|
5973 |
+
"grad_norm": 75.875,
|
5974 |
+
"learning_rate": 9.982584615384614e-07,
|
5975 |
+
"loss": 132.8726,
|
5976 |
+
"step": 8490
|
5977 |
+
},
|
5978 |
+
{
|
5979 |
+
"epoch": 0.8927179487179487,
|
5980 |
+
"grad_norm": 73.5625,
|
5981 |
+
"learning_rate": 9.9825641025641e-07,
|
5982 |
+
"loss": 132.8631,
|
5983 |
+
"step": 8500
|
5984 |
+
},
|
5985 |
+
{
|
5986 |
+
"epoch": 0.8937682051282051,
|
5987 |
+
"grad_norm": 73.4375,
|
5988 |
+
"learning_rate": 9.98254358974359e-07,
|
5989 |
+
"loss": 132.5936,
|
5990 |
+
"step": 8510
|
5991 |
+
},
|
5992 |
+
{
|
5993 |
+
"epoch": 0.8948184615384616,
|
5994 |
+
"grad_norm": 80.75,
|
5995 |
+
"learning_rate": 9.982523076923077e-07,
|
5996 |
+
"loss": 133.115,
|
5997 |
+
"step": 8520
|
5998 |
+
},
|
5999 |
+
{
|
6000 |
+
"epoch": 0.8958687179487179,
|
6001 |
+
"grad_norm": 78.25,
|
6002 |
+
"learning_rate": 9.982502564102564e-07,
|
6003 |
+
"loss": 132.0712,
|
6004 |
+
"step": 8530
|
6005 |
+
},
|
6006 |
+
{
|
6007 |
+
"epoch": 0.8969189743589744,
|
6008 |
+
"grad_norm": 71.3125,
|
6009 |
+
"learning_rate": 9.98248205128205e-07,
|
6010 |
+
"loss": 134.2035,
|
6011 |
+
"step": 8540
|
6012 |
+
},
|
6013 |
+
{
|
6014 |
+
"epoch": 0.8979692307692307,
|
6015 |
+
"grad_norm": 75.8125,
|
6016 |
+
"learning_rate": 9.982461538461538e-07,
|
6017 |
+
"loss": 134.253,
|
6018 |
+
"step": 8550
|
6019 |
+
},
|
6020 |
+
{
|
6021 |
+
"epoch": 0.8990194871794872,
|
6022 |
+
"grad_norm": 76.125,
|
6023 |
+
"learning_rate": 9.982441025641025e-07,
|
6024 |
+
"loss": 132.0401,
|
6025 |
+
"step": 8560
|
6026 |
+
},
|
6027 |
+
{
|
6028 |
+
"epoch": 0.9000697435897436,
|
6029 |
+
"grad_norm": 77.625,
|
6030 |
+
"learning_rate": 9.982420512820512e-07,
|
6031 |
+
"loss": 133.6438,
|
6032 |
+
"step": 8570
|
6033 |
+
},
|
6034 |
+
{
|
6035 |
+
"epoch": 0.90112,
|
6036 |
+
"grad_norm": 74.8125,
|
6037 |
+
"learning_rate": 9.9824e-07,
|
6038 |
+
"loss": 134.649,
|
6039 |
+
"step": 8580
|
6040 |
+
},
|
6041 |
+
{
|
6042 |
+
"epoch": 0.9021702564102564,
|
6043 |
+
"grad_norm": 78.8125,
|
6044 |
+
"learning_rate": 9.982379487179486e-07,
|
6045 |
+
"loss": 133.015,
|
6046 |
+
"step": 8590
|
6047 |
+
},
|
6048 |
+
{
|
6049 |
+
"epoch": 0.9032205128205129,
|
6050 |
+
"grad_norm": 79.5,
|
6051 |
+
"learning_rate": 9.982358974358975e-07,
|
6052 |
+
"loss": 133.7255,
|
6053 |
+
"step": 8600
|
6054 |
+
},
|
6055 |
+
{
|
6056 |
+
"epoch": 0.9042707692307692,
|
6057 |
+
"grad_norm": 76.375,
|
6058 |
+
"learning_rate": 9.982338461538462e-07,
|
6059 |
+
"loss": 133.1744,
|
6060 |
+
"step": 8610
|
6061 |
+
},
|
6062 |
+
{
|
6063 |
+
"epoch": 0.9053210256410257,
|
6064 |
+
"grad_norm": 76.8125,
|
6065 |
+
"learning_rate": 9.982317948717949e-07,
|
6066 |
+
"loss": 133.6467,
|
6067 |
+
"step": 8620
|
6068 |
+
},
|
6069 |
+
{
|
6070 |
+
"epoch": 0.906371282051282,
|
6071 |
+
"grad_norm": 75.5,
|
6072 |
+
"learning_rate": 9.982297435897435e-07,
|
6073 |
+
"loss": 133.2113,
|
6074 |
+
"step": 8630
|
6075 |
+
},
|
6076 |
+
{
|
6077 |
+
"epoch": 0.9074215384615385,
|
6078 |
+
"grad_norm": 90.875,
|
6079 |
+
"learning_rate": 9.982276923076922e-07,
|
6080 |
+
"loss": 134.016,
|
6081 |
+
"step": 8640
|
6082 |
+
},
|
6083 |
+
{
|
6084 |
+
"epoch": 0.9084717948717949,
|
6085 |
+
"grad_norm": 75.625,
|
6086 |
+
"learning_rate": 9.98225641025641e-07,
|
6087 |
+
"loss": 132.7446,
|
6088 |
+
"step": 8650
|
6089 |
+
},
|
6090 |
+
{
|
6091 |
+
"epoch": 0.9095220512820513,
|
6092 |
+
"grad_norm": 83.25,
|
6093 |
+
"learning_rate": 9.982235897435896e-07,
|
6094 |
+
"loss": 134.9909,
|
6095 |
+
"step": 8660
|
6096 |
+
},
|
6097 |
+
{
|
6098 |
+
"epoch": 0.9105723076923077,
|
6099 |
+
"grad_norm": 84.3125,
|
6100 |
+
"learning_rate": 9.982215384615385e-07,
|
6101 |
+
"loss": 133.9167,
|
6102 |
+
"step": 8670
|
6103 |
+
},
|
6104 |
+
{
|
6105 |
+
"epoch": 0.9116225641025641,
|
6106 |
+
"grad_norm": 77.375,
|
6107 |
+
"learning_rate": 9.98219487179487e-07,
|
6108 |
+
"loss": 133.2151,
|
6109 |
+
"step": 8680
|
6110 |
+
},
|
6111 |
+
{
|
6112 |
+
"epoch": 0.9126728205128205,
|
6113 |
+
"grad_norm": 74.375,
|
6114 |
+
"learning_rate": 9.98217435897436e-07,
|
6115 |
+
"loss": 132.7321,
|
6116 |
+
"step": 8690
|
6117 |
+
},
|
6118 |
+
{
|
6119 |
+
"epoch": 0.913723076923077,
|
6120 |
+
"grad_norm": 76.25,
|
6121 |
+
"learning_rate": 9.982153846153846e-07,
|
6122 |
+
"loss": 133.1595,
|
6123 |
+
"step": 8700
|
6124 |
+
},
|
6125 |
+
{
|
6126 |
+
"epoch": 0.9147733333333333,
|
6127 |
+
"grad_norm": 74.125,
|
6128 |
+
"learning_rate": 9.982133333333333e-07,
|
6129 |
+
"loss": 133.1293,
|
6130 |
+
"step": 8710
|
6131 |
+
},
|
6132 |
+
{
|
6133 |
+
"epoch": 0.9158235897435898,
|
6134 |
+
"grad_norm": 75.5625,
|
6135 |
+
"learning_rate": 9.98211282051282e-07,
|
6136 |
+
"loss": 133.3949,
|
6137 |
+
"step": 8720
|
6138 |
+
},
|
6139 |
+
{
|
6140 |
+
"epoch": 0.9168738461538462,
|
6141 |
+
"grad_norm": 77.0,
|
6142 |
+
"learning_rate": 9.982092307692307e-07,
|
6143 |
+
"loss": 133.7157,
|
6144 |
+
"step": 8730
|
6145 |
+
},
|
6146 |
+
{
|
6147 |
+
"epoch": 0.9179241025641026,
|
6148 |
+
"grad_norm": 69.6875,
|
6149 |
+
"learning_rate": 9.982071794871794e-07,
|
6150 |
+
"loss": 131.9628,
|
6151 |
+
"step": 8740
|
6152 |
+
},
|
6153 |
+
{
|
6154 |
+
"epoch": 0.918974358974359,
|
6155 |
+
"grad_norm": 76.75,
|
6156 |
+
"learning_rate": 9.98205128205128e-07,
|
6157 |
+
"loss": 133.9219,
|
6158 |
+
"step": 8750
|
6159 |
+
},
|
6160 |
+
{
|
6161 |
+
"epoch": 0.9200246153846153,
|
6162 |
+
"grad_norm": 77.0,
|
6163 |
+
"learning_rate": 9.98203076923077e-07,
|
6164 |
+
"loss": 133.5647,
|
6165 |
+
"step": 8760
|
6166 |
+
},
|
6167 |
+
{
|
6168 |
+
"epoch": 0.9210748717948718,
|
6169 |
+
"grad_norm": 73.3125,
|
6170 |
+
"learning_rate": 9.982010256410257e-07,
|
6171 |
+
"loss": 133.8773,
|
6172 |
+
"step": 8770
|
6173 |
+
},
|
6174 |
+
{
|
6175 |
+
"epoch": 0.9221251282051282,
|
6176 |
+
"grad_norm": 81.875,
|
6177 |
+
"learning_rate": 9.981989743589742e-07,
|
6178 |
+
"loss": 133.6441,
|
6179 |
+
"step": 8780
|
6180 |
+
},
|
6181 |
+
{
|
6182 |
+
"epoch": 0.9231753846153846,
|
6183 |
+
"grad_norm": 77.375,
|
6184 |
+
"learning_rate": 9.98196923076923e-07,
|
6185 |
+
"loss": 134.1967,
|
6186 |
+
"step": 8790
|
6187 |
+
},
|
6188 |
+
{
|
6189 |
+
"epoch": 0.924225641025641,
|
6190 |
+
"grad_norm": 75.75,
|
6191 |
+
"learning_rate": 9.981948717948718e-07,
|
6192 |
+
"loss": 133.6943,
|
6193 |
+
"step": 8800
|
6194 |
+
},
|
6195 |
+
{
|
6196 |
+
"epoch": 0.9252758974358974,
|
6197 |
+
"grad_norm": 74.5,
|
6198 |
+
"learning_rate": 9.981928205128205e-07,
|
6199 |
+
"loss": 131.9684,
|
6200 |
+
"step": 8810
|
6201 |
+
},
|
6202 |
+
{
|
6203 |
+
"epoch": 0.9263261538461538,
|
6204 |
+
"grad_norm": 77.9375,
|
6205 |
+
"learning_rate": 9.981907692307692e-07,
|
6206 |
+
"loss": 132.8939,
|
6207 |
+
"step": 8820
|
6208 |
+
},
|
6209 |
+
{
|
6210 |
+
"epoch": 0.9273764102564103,
|
6211 |
+
"grad_norm": 82.5625,
|
6212 |
+
"learning_rate": 9.981887179487178e-07,
|
6213 |
+
"loss": 132.9447,
|
6214 |
+
"step": 8830
|
6215 |
+
},
|
6216 |
+
{
|
6217 |
+
"epoch": 0.9284266666666666,
|
6218 |
+
"grad_norm": 78.5625,
|
6219 |
+
"learning_rate": 9.981866666666665e-07,
|
6220 |
+
"loss": 133.4321,
|
6221 |
+
"step": 8840
|
6222 |
+
},
|
6223 |
+
{
|
6224 |
+
"epoch": 0.9294769230769231,
|
6225 |
+
"grad_norm": 78.875,
|
6226 |
+
"learning_rate": 9.981846153846152e-07,
|
6227 |
+
"loss": 133.6094,
|
6228 |
+
"step": 8850
|
6229 |
+
},
|
6230 |
+
{
|
6231 |
+
"epoch": 0.9305271794871794,
|
6232 |
+
"grad_norm": 71.25,
|
6233 |
+
"learning_rate": 9.981825641025641e-07,
|
6234 |
+
"loss": 132.4788,
|
6235 |
+
"step": 8860
|
6236 |
+
},
|
6237 |
+
{
|
6238 |
+
"epoch": 0.9315774358974359,
|
6239 |
+
"grad_norm": 73.9375,
|
6240 |
+
"learning_rate": 9.981805128205128e-07,
|
6241 |
+
"loss": 132.1331,
|
6242 |
+
"step": 8870
|
6243 |
+
},
|
6244 |
+
{
|
6245 |
+
"epoch": 0.9326276923076923,
|
6246 |
+
"grad_norm": 74.6875,
|
6247 |
+
"learning_rate": 9.981784615384615e-07,
|
6248 |
+
"loss": 132.9293,
|
6249 |
+
"step": 8880
|
6250 |
+
},
|
6251 |
+
{
|
6252 |
+
"epoch": 0.9336779487179487,
|
6253 |
+
"grad_norm": 78.0625,
|
6254 |
+
"learning_rate": 9.981764102564102e-07,
|
6255 |
+
"loss": 134.1451,
|
6256 |
+
"step": 8890
|
6257 |
+
},
|
6258 |
+
{
|
6259 |
+
"epoch": 0.9347282051282051,
|
6260 |
+
"grad_norm": 77.75,
|
6261 |
+
"learning_rate": 9.98174358974359e-07,
|
6262 |
+
"loss": 132.7121,
|
6263 |
+
"step": 8900
|
6264 |
+
},
|
6265 |
+
{
|
6266 |
+
"epoch": 0.9357784615384616,
|
6267 |
+
"grad_norm": 72.75,
|
6268 |
+
"learning_rate": 9.981723076923076e-07,
|
6269 |
+
"loss": 132.0682,
|
6270 |
+
"step": 8910
|
6271 |
+
},
|
6272 |
+
{
|
6273 |
+
"epoch": 0.9368287179487179,
|
6274 |
+
"grad_norm": 78.75,
|
6275 |
+
"learning_rate": 9.981702564102563e-07,
|
6276 |
+
"loss": 132.9574,
|
6277 |
+
"step": 8920
|
6278 |
+
},
|
6279 |
+
{
|
6280 |
+
"epoch": 0.9378789743589744,
|
6281 |
+
"grad_norm": 81.625,
|
6282 |
+
"learning_rate": 9.98168205128205e-07,
|
6283 |
+
"loss": 133.9366,
|
6284 |
+
"step": 8930
|
6285 |
+
},
|
6286 |
+
{
|
6287 |
+
"epoch": 0.9389292307692307,
|
6288 |
+
"grad_norm": 74.125,
|
6289 |
+
"learning_rate": 9.981661538461537e-07,
|
6290 |
+
"loss": 134.1782,
|
6291 |
+
"step": 8940
|
6292 |
+
},
|
6293 |
+
{
|
6294 |
+
"epoch": 0.9399794871794872,
|
6295 |
+
"grad_norm": 80.4375,
|
6296 |
+
"learning_rate": 9.981641025641026e-07,
|
6297 |
+
"loss": 134.4583,
|
6298 |
+
"step": 8950
|
6299 |
+
},
|
6300 |
+
{
|
6301 |
+
"epoch": 0.9410297435897436,
|
6302 |
+
"grad_norm": 79.8125,
|
6303 |
+
"learning_rate": 9.981620512820513e-07,
|
6304 |
+
"loss": 134.201,
|
6305 |
+
"step": 8960
|
6306 |
+
},
|
6307 |
+
{
|
6308 |
+
"epoch": 0.94208,
|
6309 |
+
"grad_norm": 77.0,
|
6310 |
+
"learning_rate": 9.9816e-07,
|
6311 |
+
"loss": 133.9063,
|
6312 |
+
"step": 8970
|
6313 |
+
},
|
6314 |
+
{
|
6315 |
+
"epoch": 0.9431302564102564,
|
6316 |
+
"grad_norm": 74.8125,
|
6317 |
+
"learning_rate": 9.981579487179487e-07,
|
6318 |
+
"loss": 133.6804,
|
6319 |
+
"step": 8980
|
6320 |
+
},
|
6321 |
+
{
|
6322 |
+
"epoch": 0.9441805128205129,
|
6323 |
+
"grad_norm": 73.375,
|
6324 |
+
"learning_rate": 9.981558974358974e-07,
|
6325 |
+
"loss": 133.0501,
|
6326 |
+
"step": 8990
|
6327 |
+
},
|
6328 |
+
{
|
6329 |
+
"epoch": 0.9452307692307692,
|
6330 |
+
"grad_norm": 71.5,
|
6331 |
+
"learning_rate": 9.98153846153846e-07,
|
6332 |
+
"loss": 133.0835,
|
6333 |
+
"step": 9000
|
6334 |
+
},
|
6335 |
+
{
|
6336 |
+
"epoch": 0.9462810256410257,
|
6337 |
+
"grad_norm": 76.5625,
|
6338 |
+
"learning_rate": 9.981517948717948e-07,
|
6339 |
+
"loss": 133.7177,
|
6340 |
+
"step": 9010
|
6341 |
+
},
|
6342 |
+
{
|
6343 |
+
"epoch": 0.947331282051282,
|
6344 |
+
"grad_norm": 70.625,
|
6345 |
+
"learning_rate": 9.981497435897437e-07,
|
6346 |
+
"loss": 132.7033,
|
6347 |
+
"step": 9020
|
6348 |
+
},
|
6349 |
+
{
|
6350 |
+
"epoch": 0.9483815384615385,
|
6351 |
+
"grad_norm": 81.25,
|
6352 |
+
"learning_rate": 9.981476923076921e-07,
|
6353 |
+
"loss": 132.286,
|
6354 |
+
"step": 9030
|
6355 |
+
},
|
6356 |
+
{
|
6357 |
+
"epoch": 0.9494317948717949,
|
6358 |
+
"grad_norm": 82.125,
|
6359 |
+
"learning_rate": 9.98145641025641e-07,
|
6360 |
+
"loss": 132.4269,
|
6361 |
+
"step": 9040
|
6362 |
+
},
|
6363 |
+
{
|
6364 |
+
"epoch": 0.9504820512820513,
|
6365 |
+
"grad_norm": 76.875,
|
6366 |
+
"learning_rate": 9.981435897435897e-07,
|
6367 |
+
"loss": 132.6721,
|
6368 |
+
"step": 9050
|
6369 |
+
},
|
6370 |
+
{
|
6371 |
+
"epoch": 0.9515323076923077,
|
6372 |
+
"grad_norm": 72.5,
|
6373 |
+
"learning_rate": 9.981415384615384e-07,
|
6374 |
+
"loss": 132.3392,
|
6375 |
+
"step": 9060
|
6376 |
+
},
|
6377 |
+
{
|
6378 |
+
"epoch": 0.9525825641025641,
|
6379 |
+
"grad_norm": 82.25,
|
6380 |
+
"learning_rate": 9.981394871794871e-07,
|
6381 |
+
"loss": 133.6953,
|
6382 |
+
"step": 9070
|
6383 |
+
},
|
6384 |
+
{
|
6385 |
+
"epoch": 0.9536328205128205,
|
6386 |
+
"grad_norm": 81.0625,
|
6387 |
+
"learning_rate": 9.981374358974358e-07,
|
6388 |
+
"loss": 133.0798,
|
6389 |
+
"step": 9080
|
6390 |
+
},
|
6391 |
+
{
|
6392 |
+
"epoch": 0.954683076923077,
|
6393 |
+
"grad_norm": 75.0625,
|
6394 |
+
"learning_rate": 9.981353846153845e-07,
|
6395 |
+
"loss": 134.3506,
|
6396 |
+
"step": 9090
|
6397 |
+
},
|
6398 |
+
{
|
6399 |
+
"epoch": 0.9557333333333333,
|
6400 |
+
"grad_norm": 82.4375,
|
6401 |
+
"learning_rate": 9.981333333333332e-07,
|
6402 |
+
"loss": 135.1331,
|
6403 |
+
"step": 9100
|
6404 |
+
},
|
6405 |
+
{
|
6406 |
+
"epoch": 0.9567835897435898,
|
6407 |
+
"grad_norm": 76.375,
|
6408 |
+
"learning_rate": 9.981312820512821e-07,
|
6409 |
+
"loss": 131.6688,
|
6410 |
+
"step": 9110
|
6411 |
+
},
|
6412 |
+
{
|
6413 |
+
"epoch": 0.9578338461538461,
|
6414 |
+
"grad_norm": 72.25,
|
6415 |
+
"learning_rate": 9.981292307692308e-07,
|
6416 |
+
"loss": 134.2536,
|
6417 |
+
"step": 9120
|
6418 |
+
},
|
6419 |
+
{
|
6420 |
+
"epoch": 0.9588841025641026,
|
6421 |
+
"grad_norm": 75.375,
|
6422 |
+
"learning_rate": 9.981271794871793e-07,
|
6423 |
+
"loss": 132.993,
|
6424 |
+
"step": 9130
|
6425 |
+
},
|
6426 |
+
{
|
6427 |
+
"epoch": 0.959934358974359,
|
6428 |
+
"grad_norm": 79.625,
|
6429 |
+
"learning_rate": 9.981251282051282e-07,
|
6430 |
+
"loss": 132.7532,
|
6431 |
+
"step": 9140
|
6432 |
+
},
|
6433 |
+
{
|
6434 |
+
"epoch": 0.9609846153846154,
|
6435 |
+
"grad_norm": 72.8125,
|
6436 |
+
"learning_rate": 9.98123076923077e-07,
|
6437 |
+
"loss": 134.5466,
|
6438 |
+
"step": 9150
|
6439 |
+
},
|
6440 |
+
{
|
6441 |
+
"epoch": 0.9620348717948718,
|
6442 |
+
"grad_norm": 73.625,
|
6443 |
+
"learning_rate": 9.981210256410256e-07,
|
6444 |
+
"loss": 133.4606,
|
6445 |
+
"step": 9160
|
6446 |
+
},
|
6447 |
+
{
|
6448 |
+
"epoch": 0.9630851282051283,
|
6449 |
+
"grad_norm": 69.25,
|
6450 |
+
"learning_rate": 9.981189743589743e-07,
|
6451 |
+
"loss": 132.9202,
|
6452 |
+
"step": 9170
|
6453 |
+
},
|
6454 |
+
{
|
6455 |
+
"epoch": 0.9641353846153846,
|
6456 |
+
"grad_norm": 77.625,
|
6457 |
+
"learning_rate": 9.98116923076923e-07,
|
6458 |
+
"loss": 133.554,
|
6459 |
+
"step": 9180
|
6460 |
+
},
|
6461 |
+
{
|
6462 |
+
"epoch": 0.965185641025641,
|
6463 |
+
"grad_norm": 72.0,
|
6464 |
+
"learning_rate": 9.981148717948717e-07,
|
6465 |
+
"loss": 133.4544,
|
6466 |
+
"step": 9190
|
6467 |
+
},
|
6468 |
+
{
|
6469 |
+
"epoch": 0.9662358974358974,
|
6470 |
+
"grad_norm": 72.75,
|
6471 |
+
"learning_rate": 9.981128205128204e-07,
|
6472 |
+
"loss": 132.8837,
|
6473 |
+
"step": 9200
|
6474 |
+
},
|
6475 |
+
{
|
6476 |
+
"epoch": 0.9672861538461538,
|
6477 |
+
"grad_norm": 75.1875,
|
6478 |
+
"learning_rate": 9.981107692307693e-07,
|
6479 |
+
"loss": 133.7232,
|
6480 |
+
"step": 9210
|
6481 |
+
},
|
6482 |
+
{
|
6483 |
+
"epoch": 0.9683364102564103,
|
6484 |
+
"grad_norm": 69.6875,
|
6485 |
+
"learning_rate": 9.98108717948718e-07,
|
6486 |
+
"loss": 132.7222,
|
6487 |
+
"step": 9220
|
6488 |
+
},
|
6489 |
+
{
|
6490 |
+
"epoch": 0.9693866666666666,
|
6491 |
+
"grad_norm": 76.3125,
|
6492 |
+
"learning_rate": 9.981066666666667e-07,
|
6493 |
+
"loss": 134.062,
|
6494 |
+
"step": 9230
|
6495 |
+
},
|
6496 |
+
{
|
6497 |
+
"epoch": 0.9704369230769231,
|
6498 |
+
"grad_norm": 73.8125,
|
6499 |
+
"learning_rate": 9.981046153846154e-07,
|
6500 |
+
"loss": 133.7961,
|
6501 |
+
"step": 9240
|
6502 |
+
},
|
6503 |
+
{
|
6504 |
+
"epoch": 0.9714871794871794,
|
6505 |
+
"grad_norm": 76.25,
|
6506 |
+
"learning_rate": 9.98102564102564e-07,
|
6507 |
+
"loss": 133.53,
|
6508 |
+
"step": 9250
|
6509 |
+
},
|
6510 |
+
{
|
6511 |
+
"epoch": 0.9725374358974359,
|
6512 |
+
"grad_norm": 76.8125,
|
6513 |
+
"learning_rate": 9.981005128205127e-07,
|
6514 |
+
"loss": 134.952,
|
6515 |
+
"step": 9260
|
6516 |
+
},
|
6517 |
+
{
|
6518 |
+
"epoch": 0.9735876923076923,
|
6519 |
+
"grad_norm": 74.1875,
|
6520 |
+
"learning_rate": 9.980984615384614e-07,
|
6521 |
+
"loss": 132.9375,
|
6522 |
+
"step": 9270
|
6523 |
+
},
|
6524 |
+
{
|
6525 |
+
"epoch": 0.9746379487179487,
|
6526 |
+
"grad_norm": 87.25,
|
6527 |
+
"learning_rate": 9.980964102564101e-07,
|
6528 |
+
"loss": 132.9781,
|
6529 |
+
"step": 9280
|
6530 |
+
},
|
6531 |
+
{
|
6532 |
+
"epoch": 0.9756882051282051,
|
6533 |
+
"grad_norm": 74.4375,
|
6534 |
+
"learning_rate": 9.980943589743588e-07,
|
6535 |
+
"loss": 132.2021,
|
6536 |
+
"step": 9290
|
6537 |
+
},
|
6538 |
+
{
|
6539 |
+
"epoch": 0.9767384615384616,
|
6540 |
+
"grad_norm": 73.0,
|
6541 |
+
"learning_rate": 9.980923076923077e-07,
|
6542 |
+
"loss": 132.5939,
|
6543 |
+
"step": 9300
|
6544 |
+
},
|
6545 |
+
{
|
6546 |
+
"epoch": 0.9777887179487179,
|
6547 |
+
"grad_norm": 77.5,
|
6548 |
+
"learning_rate": 9.980902564102564e-07,
|
6549 |
+
"loss": 133.525,
|
6550 |
+
"step": 9310
|
6551 |
+
},
|
6552 |
+
{
|
6553 |
+
"epoch": 0.9788389743589744,
|
6554 |
+
"grad_norm": 79.1875,
|
6555 |
+
"learning_rate": 9.980882051282051e-07,
|
6556 |
+
"loss": 132.8209,
|
6557 |
+
"step": 9320
|
6558 |
+
},
|
6559 |
+
{
|
6560 |
+
"epoch": 0.9798892307692307,
|
6561 |
+
"grad_norm": 76.9375,
|
6562 |
+
"learning_rate": 9.980861538461538e-07,
|
6563 |
+
"loss": 133.3684,
|
6564 |
+
"step": 9330
|
6565 |
+
},
|
6566 |
+
{
|
6567 |
+
"epoch": 0.9809394871794872,
|
6568 |
+
"grad_norm": 71.9375,
|
6569 |
+
"learning_rate": 9.980841025641025e-07,
|
6570 |
+
"loss": 130.963,
|
6571 |
+
"step": 9340
|
6572 |
+
},
|
6573 |
+
{
|
6574 |
+
"epoch": 0.9819897435897436,
|
6575 |
+
"grad_norm": 79.9375,
|
6576 |
+
"learning_rate": 9.980820512820512e-07,
|
6577 |
+
"loss": 133.7593,
|
6578 |
+
"step": 9350
|
6579 |
+
},
|
6580 |
+
{
|
6581 |
+
"epoch": 0.98304,
|
6582 |
+
"grad_norm": 77.125,
|
6583 |
+
"learning_rate": 9.9808e-07,
|
6584 |
+
"loss": 132.379,
|
6585 |
+
"step": 9360
|
6586 |
+
},
|
6587 |
+
{
|
6588 |
+
"epoch": 0.9840902564102564,
|
6589 |
+
"grad_norm": 76.75,
|
6590 |
+
"learning_rate": 9.980779487179488e-07,
|
6591 |
+
"loss": 131.4839,
|
6592 |
+
"step": 9370
|
6593 |
+
},
|
6594 |
+
{
|
6595 |
+
"epoch": 0.9851405128205128,
|
6596 |
+
"grad_norm": 73.8125,
|
6597 |
+
"learning_rate": 9.980758974358973e-07,
|
6598 |
+
"loss": 133.2282,
|
6599 |
+
"step": 9380
|
6600 |
+
},
|
6601 |
+
{
|
6602 |
+
"epoch": 0.9861907692307692,
|
6603 |
+
"grad_norm": 73.9375,
|
6604 |
+
"learning_rate": 9.980738461538462e-07,
|
6605 |
+
"loss": 132.6678,
|
6606 |
+
"step": 9390
|
6607 |
+
},
|
6608 |
+
{
|
6609 |
+
"epoch": 0.9872410256410257,
|
6610 |
+
"grad_norm": 76.625,
|
6611 |
+
"learning_rate": 9.980717948717949e-07,
|
6612 |
+
"loss": 132.3676,
|
6613 |
+
"step": 9400
|
6614 |
+
},
|
6615 |
+
{
|
6616 |
+
"epoch": 0.988291282051282,
|
6617 |
+
"grad_norm": 73.125,
|
6618 |
+
"learning_rate": 9.980697435897436e-07,
|
6619 |
+
"loss": 132.7415,
|
6620 |
+
"step": 9410
|
6621 |
+
},
|
6622 |
+
{
|
6623 |
+
"epoch": 0.9893415384615385,
|
6624 |
+
"grad_norm": 74.75,
|
6625 |
+
"learning_rate": 9.980676923076923e-07,
|
6626 |
+
"loss": 134.5323,
|
6627 |
+
"step": 9420
|
6628 |
+
},
|
6629 |
+
{
|
6630 |
+
"epoch": 0.9903917948717949,
|
6631 |
+
"grad_norm": 75.75,
|
6632 |
+
"learning_rate": 9.98065641025641e-07,
|
6633 |
+
"loss": 132.3584,
|
6634 |
+
"step": 9430
|
6635 |
+
},
|
6636 |
+
{
|
6637 |
+
"epoch": 0.9914420512820513,
|
6638 |
+
"grad_norm": 77.5625,
|
6639 |
+
"learning_rate": 9.980635897435897e-07,
|
6640 |
+
"loss": 132.1099,
|
6641 |
+
"step": 9440
|
6642 |
+
},
|
6643 |
+
{
|
6644 |
+
"epoch": 0.9924923076923077,
|
6645 |
+
"grad_norm": 75.5625,
|
6646 |
+
"learning_rate": 9.980615384615384e-07,
|
6647 |
+
"loss": 132.6088,
|
6648 |
+
"step": 9450
|
6649 |
+
},
|
6650 |
+
{
|
6651 |
+
"epoch": 0.9935425641025641,
|
6652 |
+
"grad_norm": 77.75,
|
6653 |
+
"learning_rate": 9.980594871794873e-07,
|
6654 |
+
"loss": 132.0042,
|
6655 |
+
"step": 9460
|
6656 |
+
},
|
6657 |
+
{
|
6658 |
+
"epoch": 0.9945928205128205,
|
6659 |
+
"grad_norm": 76.6875,
|
6660 |
+
"learning_rate": 9.98057435897436e-07,
|
6661 |
+
"loss": 132.9037,
|
6662 |
+
"step": 9470
|
6663 |
+
},
|
6664 |
+
{
|
6665 |
+
"epoch": 0.995643076923077,
|
6666 |
+
"grad_norm": 75.5,
|
6667 |
+
"learning_rate": 9.980553846153844e-07,
|
6668 |
+
"loss": 133.0483,
|
6669 |
+
"step": 9480
|
6670 |
+
},
|
6671 |
+
{
|
6672 |
+
"epoch": 0.9966933333333333,
|
6673 |
+
"grad_norm": 74.8125,
|
6674 |
+
"learning_rate": 9.980533333333333e-07,
|
6675 |
+
"loss": 134.3751,
|
6676 |
+
"step": 9490
|
6677 |
+
},
|
6678 |
+
{
|
6679 |
+
"epoch": 0.9977435897435898,
|
6680 |
+
"grad_norm": 74.625,
|
6681 |
+
"learning_rate": 9.98051282051282e-07,
|
6682 |
+
"loss": 132.6699,
|
6683 |
+
"step": 9500
|
6684 |
+
},
|
6685 |
+
{
|
6686 |
+
"epoch": 0.9987938461538461,
|
6687 |
+
"grad_norm": 79.875,
|
6688 |
+
"learning_rate": 9.980492307692307e-07,
|
6689 |
+
"loss": 132.3946,
|
6690 |
+
"step": 9510
|
6691 |
+
},
|
6692 |
+
{
|
6693 |
+
"epoch": 0.9998441025641026,
|
6694 |
+
"grad_norm": 73.5625,
|
6695 |
+
"learning_rate": 9.980471794871794e-07,
|
6696 |
+
"loss": 133.7389,
|
6697 |
+
"step": 9520
|
6698 |
}
|
6699 |
],
|
6700 |
"logging_steps": 10,
|
|
|
6709 |
"should_evaluate": false,
|
6710 |
"should_log": false,
|
6711 |
"should_save": true,
|
6712 |
+
"should_training_stop": true
|
6713 |
},
|
6714 |
"attributes": {}
|
6715 |
}
|
6716 |
},
|
6717 |
+
"total_flos": 2.6288590936506106e+19,
|
6718 |
"train_batch_size": 4,
|
6719 |
"trial_name": null,
|
6720 |
"trial_params": null
|