Upload folder using huggingface_hub
Browse files- adapter_model.safetensors +1 -1
- optimizer.pt +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1753 -3
adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 3158328
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:283812852aa9943044c46ea9563e51d91830ead59f1e2d9af039a36083b2edcc
|
3 |
size 3158328
|
optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1687034
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ecff4a590405220eacbe2c607e93c31ace86b6339d20cc4eb7cbdd3905a68247
|
3 |
size 1687034
|
rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:6c1cd05962e3ca96a7b68f357adf12ab59e5a3549eefa364fd4ce7ea00f59257
|
3 |
size 14244
|
scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:4fb75d6e1818a81bcc3d53f94d5a831be0e0f045ebba32887861936e527dc932
|
3 |
size 1064
|
trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch":
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5257,6 +5257,1756 @@
|
|
5257 |
"learning_rate": 0.00010229508196721311,
|
5258 |
"loss": 0.0448,
|
5259 |
"step": 15000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5260 |
}
|
5261 |
],
|
5262 |
"logging_steps": 20,
|
@@ -5276,7 +7026,7 @@
|
|
5276 |
"attributes": {}
|
5277 |
}
|
5278 |
},
|
5279 |
-
"total_flos":
|
5280 |
"train_batch_size": 1,
|
5281 |
"trial_name": null,
|
5282 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 97.62050030506407,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 20000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5257 |
"learning_rate": 0.00010229508196721311,
|
5258 |
"loss": 0.0448,
|
5259 |
"step": 15000
|
5260 |
+
},
|
5261 |
+
{
|
5262 |
+
"epoch": 73.31299572910311,
|
5263 |
+
"grad_norm": 1.9551867246627808,
|
5264 |
+
"learning_rate": 0.00010216393442622951,
|
5265 |
+
"loss": 0.0468,
|
5266 |
+
"step": 15020
|
5267 |
+
},
|
5268 |
+
{
|
5269 |
+
"epoch": 73.41061622940818,
|
5270 |
+
"grad_norm": 2.8374905586242676,
|
5271 |
+
"learning_rate": 0.00010203278688524592,
|
5272 |
+
"loss": 0.0473,
|
5273 |
+
"step": 15040
|
5274 |
+
},
|
5275 |
+
{
|
5276 |
+
"epoch": 73.50823672971325,
|
5277 |
+
"grad_norm": 1.2752137184143066,
|
5278 |
+
"learning_rate": 0.00010190163934426229,
|
5279 |
+
"loss": 0.0487,
|
5280 |
+
"step": 15060
|
5281 |
+
},
|
5282 |
+
{
|
5283 |
+
"epoch": 73.6058572300183,
|
5284 |
+
"grad_norm": 1.7868965864181519,
|
5285 |
+
"learning_rate": 0.0001017704918032787,
|
5286 |
+
"loss": 0.0479,
|
5287 |
+
"step": 15080
|
5288 |
+
},
|
5289 |
+
{
|
5290 |
+
"epoch": 73.70347773032337,
|
5291 |
+
"grad_norm": 1.7490606307983398,
|
5292 |
+
"learning_rate": 0.00010163934426229508,
|
5293 |
+
"loss": 0.0466,
|
5294 |
+
"step": 15100
|
5295 |
+
},
|
5296 |
+
{
|
5297 |
+
"epoch": 73.80109823062843,
|
5298 |
+
"grad_norm": 2.4176886081695557,
|
5299 |
+
"learning_rate": 0.00010150819672131149,
|
5300 |
+
"loss": 0.0539,
|
5301 |
+
"step": 15120
|
5302 |
+
},
|
5303 |
+
{
|
5304 |
+
"epoch": 73.8987187309335,
|
5305 |
+
"grad_norm": 1.9388647079467773,
|
5306 |
+
"learning_rate": 0.00010137704918032786,
|
5307 |
+
"loss": 0.0527,
|
5308 |
+
"step": 15140
|
5309 |
+
},
|
5310 |
+
{
|
5311 |
+
"epoch": 73.99633923123857,
|
5312 |
+
"grad_norm": 1.9763060808181763,
|
5313 |
+
"learning_rate": 0.00010124590163934427,
|
5314 |
+
"loss": 0.0512,
|
5315 |
+
"step": 15160
|
5316 |
+
},
|
5317 |
+
{
|
5318 |
+
"epoch": 74.09395973154362,
|
5319 |
+
"grad_norm": 1.1881762742996216,
|
5320 |
+
"learning_rate": 0.00010111475409836067,
|
5321 |
+
"loss": 0.0406,
|
5322 |
+
"step": 15180
|
5323 |
+
},
|
5324 |
+
{
|
5325 |
+
"epoch": 74.19158023184869,
|
5326 |
+
"grad_norm": 1.5949184894561768,
|
5327 |
+
"learning_rate": 0.00010098360655737706,
|
5328 |
+
"loss": 0.0446,
|
5329 |
+
"step": 15200
|
5330 |
+
},
|
5331 |
+
{
|
5332 |
+
"epoch": 74.28920073215376,
|
5333 |
+
"grad_norm": 1.3817694187164307,
|
5334 |
+
"learning_rate": 0.00010085245901639345,
|
5335 |
+
"loss": 0.0468,
|
5336 |
+
"step": 15220
|
5337 |
+
},
|
5338 |
+
{
|
5339 |
+
"epoch": 74.38682123245881,
|
5340 |
+
"grad_norm": 1.4940481185913086,
|
5341 |
+
"learning_rate": 0.00010072131147540984,
|
5342 |
+
"loss": 0.0528,
|
5343 |
+
"step": 15240
|
5344 |
+
},
|
5345 |
+
{
|
5346 |
+
"epoch": 74.48444173276388,
|
5347 |
+
"grad_norm": 1.5146404504776,
|
5348 |
+
"learning_rate": 0.00010059016393442624,
|
5349 |
+
"loss": 0.0503,
|
5350 |
+
"step": 15260
|
5351 |
+
},
|
5352 |
+
{
|
5353 |
+
"epoch": 74.58206223306894,
|
5354 |
+
"grad_norm": 1.4908167123794556,
|
5355 |
+
"learning_rate": 0.00010045901639344261,
|
5356 |
+
"loss": 0.0474,
|
5357 |
+
"step": 15280
|
5358 |
+
},
|
5359 |
+
{
|
5360 |
+
"epoch": 74.67968273337401,
|
5361 |
+
"grad_norm": 1.9127382040023804,
|
5362 |
+
"learning_rate": 0.00010032786885245902,
|
5363 |
+
"loss": 0.0475,
|
5364 |
+
"step": 15300
|
5365 |
+
},
|
5366 |
+
{
|
5367 |
+
"epoch": 74.77730323367908,
|
5368 |
+
"grad_norm": 1.9283052682876587,
|
5369 |
+
"learning_rate": 0.00010019672131147542,
|
5370 |
+
"loss": 0.0482,
|
5371 |
+
"step": 15320
|
5372 |
+
},
|
5373 |
+
{
|
5374 |
+
"epoch": 74.87492373398413,
|
5375 |
+
"grad_norm": 1.434121012687683,
|
5376 |
+
"learning_rate": 0.00010006557377049181,
|
5377 |
+
"loss": 0.0484,
|
5378 |
+
"step": 15340
|
5379 |
+
},
|
5380 |
+
{
|
5381 |
+
"epoch": 74.9725442342892,
|
5382 |
+
"grad_norm": 1.2398452758789062,
|
5383 |
+
"learning_rate": 9.99344262295082e-05,
|
5384 |
+
"loss": 0.0474,
|
5385 |
+
"step": 15360
|
5386 |
+
},
|
5387 |
+
{
|
5388 |
+
"epoch": 75.07016473459427,
|
5389 |
+
"grad_norm": 1.463592529296875,
|
5390 |
+
"learning_rate": 9.980327868852459e-05,
|
5391 |
+
"loss": 0.0435,
|
5392 |
+
"step": 15380
|
5393 |
+
},
|
5394 |
+
{
|
5395 |
+
"epoch": 75.16778523489933,
|
5396 |
+
"grad_norm": 1.5821679830551147,
|
5397 |
+
"learning_rate": 9.967213114754099e-05,
|
5398 |
+
"loss": 0.0447,
|
5399 |
+
"step": 15400
|
5400 |
+
},
|
5401 |
+
{
|
5402 |
+
"epoch": 75.2654057352044,
|
5403 |
+
"grad_norm": 1.525728702545166,
|
5404 |
+
"learning_rate": 9.954098360655738e-05,
|
5405 |
+
"loss": 0.0442,
|
5406 |
+
"step": 15420
|
5407 |
+
},
|
5408 |
+
{
|
5409 |
+
"epoch": 75.36302623550945,
|
5410 |
+
"grad_norm": 1.2761436700820923,
|
5411 |
+
"learning_rate": 9.940983606557378e-05,
|
5412 |
+
"loss": 0.0464,
|
5413 |
+
"step": 15440
|
5414 |
+
},
|
5415 |
+
{
|
5416 |
+
"epoch": 75.46064673581452,
|
5417 |
+
"grad_norm": 1.4206050634384155,
|
5418 |
+
"learning_rate": 9.927868852459017e-05,
|
5419 |
+
"loss": 0.0436,
|
5420 |
+
"step": 15460
|
5421 |
+
},
|
5422 |
+
{
|
5423 |
+
"epoch": 75.55826723611959,
|
5424 |
+
"grad_norm": 2.7865281105041504,
|
5425 |
+
"learning_rate": 9.914754098360656e-05,
|
5426 |
+
"loss": 0.0505,
|
5427 |
+
"step": 15480
|
5428 |
+
},
|
5429 |
+
{
|
5430 |
+
"epoch": 75.65588773642465,
|
5431 |
+
"grad_norm": 1.249638557434082,
|
5432 |
+
"learning_rate": 9.901639344262295e-05,
|
5433 |
+
"loss": 0.0461,
|
5434 |
+
"step": 15500
|
5435 |
+
},
|
5436 |
+
{
|
5437 |
+
"epoch": 75.75350823672972,
|
5438 |
+
"grad_norm": 1.9104782342910767,
|
5439 |
+
"learning_rate": 9.888524590163934e-05,
|
5440 |
+
"loss": 0.0465,
|
5441 |
+
"step": 15520
|
5442 |
+
},
|
5443 |
+
{
|
5444 |
+
"epoch": 75.85112873703477,
|
5445 |
+
"grad_norm": 1.3011826276779175,
|
5446 |
+
"learning_rate": 9.875409836065574e-05,
|
5447 |
+
"loss": 0.0488,
|
5448 |
+
"step": 15540
|
5449 |
+
},
|
5450 |
+
{
|
5451 |
+
"epoch": 75.94874923733984,
|
5452 |
+
"grad_norm": 2.3834664821624756,
|
5453 |
+
"learning_rate": 9.862295081967213e-05,
|
5454 |
+
"loss": 0.0519,
|
5455 |
+
"step": 15560
|
5456 |
+
},
|
5457 |
+
{
|
5458 |
+
"epoch": 76.04636973764491,
|
5459 |
+
"grad_norm": 1.3640445470809937,
|
5460 |
+
"learning_rate": 9.849180327868854e-05,
|
5461 |
+
"loss": 0.0449,
|
5462 |
+
"step": 15580
|
5463 |
+
},
|
5464 |
+
{
|
5465 |
+
"epoch": 76.14399023794996,
|
5466 |
+
"grad_norm": 1.3013081550598145,
|
5467 |
+
"learning_rate": 9.836065573770493e-05,
|
5468 |
+
"loss": 0.0399,
|
5469 |
+
"step": 15600
|
5470 |
+
},
|
5471 |
+
{
|
5472 |
+
"epoch": 76.24161073825503,
|
5473 |
+
"grad_norm": 2.726545810699463,
|
5474 |
+
"learning_rate": 9.822950819672132e-05,
|
5475 |
+
"loss": 0.0427,
|
5476 |
+
"step": 15620
|
5477 |
+
},
|
5478 |
+
{
|
5479 |
+
"epoch": 76.3392312385601,
|
5480 |
+
"grad_norm": 1.2742412090301514,
|
5481 |
+
"learning_rate": 9.80983606557377e-05,
|
5482 |
+
"loss": 0.0439,
|
5483 |
+
"step": 15640
|
5484 |
+
},
|
5485 |
+
{
|
5486 |
+
"epoch": 76.43685173886516,
|
5487 |
+
"grad_norm": 1.67985999584198,
|
5488 |
+
"learning_rate": 9.796721311475411e-05,
|
5489 |
+
"loss": 0.0483,
|
5490 |
+
"step": 15660
|
5491 |
+
},
|
5492 |
+
{
|
5493 |
+
"epoch": 76.53447223917023,
|
5494 |
+
"grad_norm": 1.9676116704940796,
|
5495 |
+
"learning_rate": 9.78360655737705e-05,
|
5496 |
+
"loss": 0.0442,
|
5497 |
+
"step": 15680
|
5498 |
+
},
|
5499 |
+
{
|
5500 |
+
"epoch": 76.63209273947528,
|
5501 |
+
"grad_norm": 2.992687463760376,
|
5502 |
+
"learning_rate": 9.770491803278689e-05,
|
5503 |
+
"loss": 0.0429,
|
5504 |
+
"step": 15700
|
5505 |
+
},
|
5506 |
+
{
|
5507 |
+
"epoch": 76.72971323978035,
|
5508 |
+
"grad_norm": 1.0530943870544434,
|
5509 |
+
"learning_rate": 9.757377049180329e-05,
|
5510 |
+
"loss": 0.0488,
|
5511 |
+
"step": 15720
|
5512 |
+
},
|
5513 |
+
{
|
5514 |
+
"epoch": 76.82733374008542,
|
5515 |
+
"grad_norm": 1.5308395624160767,
|
5516 |
+
"learning_rate": 9.744262295081968e-05,
|
5517 |
+
"loss": 0.0442,
|
5518 |
+
"step": 15740
|
5519 |
+
},
|
5520 |
+
{
|
5521 |
+
"epoch": 76.92495424039048,
|
5522 |
+
"grad_norm": 1.6086573600769043,
|
5523 |
+
"learning_rate": 9.731147540983607e-05,
|
5524 |
+
"loss": 0.0504,
|
5525 |
+
"step": 15760
|
5526 |
+
},
|
5527 |
+
{
|
5528 |
+
"epoch": 77.02257474069555,
|
5529 |
+
"grad_norm": 0.7401487827301025,
|
5530 |
+
"learning_rate": 9.718032786885246e-05,
|
5531 |
+
"loss": 0.0469,
|
5532 |
+
"step": 15780
|
5533 |
+
},
|
5534 |
+
{
|
5535 |
+
"epoch": 77.12019524100062,
|
5536 |
+
"grad_norm": 1.6960521936416626,
|
5537 |
+
"learning_rate": 9.704918032786886e-05,
|
5538 |
+
"loss": 0.041,
|
5539 |
+
"step": 15800
|
5540 |
+
},
|
5541 |
+
{
|
5542 |
+
"epoch": 77.21781574130567,
|
5543 |
+
"grad_norm": 1.638244390487671,
|
5544 |
+
"learning_rate": 9.691803278688525e-05,
|
5545 |
+
"loss": 0.0381,
|
5546 |
+
"step": 15820
|
5547 |
+
},
|
5548 |
+
{
|
5549 |
+
"epoch": 77.31543624161074,
|
5550 |
+
"grad_norm": 1.749157190322876,
|
5551 |
+
"learning_rate": 9.678688524590165e-05,
|
5552 |
+
"loss": 0.0425,
|
5553 |
+
"step": 15840
|
5554 |
+
},
|
5555 |
+
{
|
5556 |
+
"epoch": 77.4130567419158,
|
5557 |
+
"grad_norm": 1.9708433151245117,
|
5558 |
+
"learning_rate": 9.665573770491804e-05,
|
5559 |
+
"loss": 0.046,
|
5560 |
+
"step": 15860
|
5561 |
+
},
|
5562 |
+
{
|
5563 |
+
"epoch": 77.51067724222086,
|
5564 |
+
"grad_norm": 1.3981733322143555,
|
5565 |
+
"learning_rate": 9.652459016393443e-05,
|
5566 |
+
"loss": 0.0489,
|
5567 |
+
"step": 15880
|
5568 |
+
},
|
5569 |
+
{
|
5570 |
+
"epoch": 77.60829774252593,
|
5571 |
+
"grad_norm": 2.0385096073150635,
|
5572 |
+
"learning_rate": 9.639344262295082e-05,
|
5573 |
+
"loss": 0.0442,
|
5574 |
+
"step": 15900
|
5575 |
+
},
|
5576 |
+
{
|
5577 |
+
"epoch": 77.70591824283099,
|
5578 |
+
"grad_norm": 1.5670779943466187,
|
5579 |
+
"learning_rate": 9.626229508196721e-05,
|
5580 |
+
"loss": 0.0412,
|
5581 |
+
"step": 15920
|
5582 |
+
},
|
5583 |
+
{
|
5584 |
+
"epoch": 77.80353874313606,
|
5585 |
+
"grad_norm": 1.3687881231307983,
|
5586 |
+
"learning_rate": 9.613114754098361e-05,
|
5587 |
+
"loss": 0.0469,
|
5588 |
+
"step": 15940
|
5589 |
+
},
|
5590 |
+
{
|
5591 |
+
"epoch": 77.90115924344113,
|
5592 |
+
"grad_norm": 1.388791799545288,
|
5593 |
+
"learning_rate": 9.6e-05,
|
5594 |
+
"loss": 0.0456,
|
5595 |
+
"step": 15960
|
5596 |
+
},
|
5597 |
+
{
|
5598 |
+
"epoch": 77.99877974374618,
|
5599 |
+
"grad_norm": 1.5988264083862305,
|
5600 |
+
"learning_rate": 9.58688524590164e-05,
|
5601 |
+
"loss": 0.048,
|
5602 |
+
"step": 15980
|
5603 |
+
},
|
5604 |
+
{
|
5605 |
+
"epoch": 78.09640024405125,
|
5606 |
+
"grad_norm": 2.0128464698791504,
|
5607 |
+
"learning_rate": 9.57377049180328e-05,
|
5608 |
+
"loss": 0.0406,
|
5609 |
+
"step": 16000
|
5610 |
+
},
|
5611 |
+
{
|
5612 |
+
"epoch": 78.19402074435631,
|
5613 |
+
"grad_norm": 1.9334806203842163,
|
5614 |
+
"learning_rate": 9.560655737704918e-05,
|
5615 |
+
"loss": 0.0402,
|
5616 |
+
"step": 16020
|
5617 |
+
},
|
5618 |
+
{
|
5619 |
+
"epoch": 78.29164124466138,
|
5620 |
+
"grad_norm": 2.799999952316284,
|
5621 |
+
"learning_rate": 9.547540983606557e-05,
|
5622 |
+
"loss": 0.0415,
|
5623 |
+
"step": 16040
|
5624 |
+
},
|
5625 |
+
{
|
5626 |
+
"epoch": 78.38926174496645,
|
5627 |
+
"grad_norm": 1.2322758436203003,
|
5628 |
+
"learning_rate": 9.534426229508198e-05,
|
5629 |
+
"loss": 0.0396,
|
5630 |
+
"step": 16060
|
5631 |
+
},
|
5632 |
+
{
|
5633 |
+
"epoch": 78.4868822452715,
|
5634 |
+
"grad_norm": 2.0646209716796875,
|
5635 |
+
"learning_rate": 9.521311475409837e-05,
|
5636 |
+
"loss": 0.0449,
|
5637 |
+
"step": 16080
|
5638 |
+
},
|
5639 |
+
{
|
5640 |
+
"epoch": 78.58450274557657,
|
5641 |
+
"grad_norm": 1.3239318132400513,
|
5642 |
+
"learning_rate": 9.508196721311476e-05,
|
5643 |
+
"loss": 0.0473,
|
5644 |
+
"step": 16100
|
5645 |
+
},
|
5646 |
+
{
|
5647 |
+
"epoch": 78.68212324588164,
|
5648 |
+
"grad_norm": 1.0266462564468384,
|
5649 |
+
"learning_rate": 9.495081967213116e-05,
|
5650 |
+
"loss": 0.047,
|
5651 |
+
"step": 16120
|
5652 |
+
},
|
5653 |
+
{
|
5654 |
+
"epoch": 78.7797437461867,
|
5655 |
+
"grad_norm": 1.3568265438079834,
|
5656 |
+
"learning_rate": 9.481967213114755e-05,
|
5657 |
+
"loss": 0.0472,
|
5658 |
+
"step": 16140
|
5659 |
+
},
|
5660 |
+
{
|
5661 |
+
"epoch": 78.87736424649177,
|
5662 |
+
"grad_norm": 1.8407416343688965,
|
5663 |
+
"learning_rate": 9.468852459016394e-05,
|
5664 |
+
"loss": 0.0489,
|
5665 |
+
"step": 16160
|
5666 |
+
},
|
5667 |
+
{
|
5668 |
+
"epoch": 78.97498474679682,
|
5669 |
+
"grad_norm": 3.1566367149353027,
|
5670 |
+
"learning_rate": 9.455737704918033e-05,
|
5671 |
+
"loss": 0.0447,
|
5672 |
+
"step": 16180
|
5673 |
+
},
|
5674 |
+
{
|
5675 |
+
"epoch": 79.07260524710189,
|
5676 |
+
"grad_norm": 1.4140985012054443,
|
5677 |
+
"learning_rate": 9.442622950819673e-05,
|
5678 |
+
"loss": 0.0397,
|
5679 |
+
"step": 16200
|
5680 |
+
},
|
5681 |
+
{
|
5682 |
+
"epoch": 79.17022574740696,
|
5683 |
+
"grad_norm": 1.774674654006958,
|
5684 |
+
"learning_rate": 9.429508196721312e-05,
|
5685 |
+
"loss": 0.0394,
|
5686 |
+
"step": 16220
|
5687 |
+
},
|
5688 |
+
{
|
5689 |
+
"epoch": 79.26784624771201,
|
5690 |
+
"grad_norm": 1.8382582664489746,
|
5691 |
+
"learning_rate": 9.416393442622952e-05,
|
5692 |
+
"loss": 0.0391,
|
5693 |
+
"step": 16240
|
5694 |
+
},
|
5695 |
+
{
|
5696 |
+
"epoch": 79.36546674801708,
|
5697 |
+
"grad_norm": 1.465022087097168,
|
5698 |
+
"learning_rate": 9.403278688524591e-05,
|
5699 |
+
"loss": 0.0427,
|
5700 |
+
"step": 16260
|
5701 |
+
},
|
5702 |
+
{
|
5703 |
+
"epoch": 79.46308724832215,
|
5704 |
+
"grad_norm": 1.5447912216186523,
|
5705 |
+
"learning_rate": 9.39016393442623e-05,
|
5706 |
+
"loss": 0.0435,
|
5707 |
+
"step": 16280
|
5708 |
+
},
|
5709 |
+
{
|
5710 |
+
"epoch": 79.56070774862721,
|
5711 |
+
"grad_norm": 1.5630700588226318,
|
5712 |
+
"learning_rate": 9.377049180327869e-05,
|
5713 |
+
"loss": 0.0426,
|
5714 |
+
"step": 16300
|
5715 |
+
},
|
5716 |
+
{
|
5717 |
+
"epoch": 79.65832824893228,
|
5718 |
+
"grad_norm": 2.1536877155303955,
|
5719 |
+
"learning_rate": 9.363934426229508e-05,
|
5720 |
+
"loss": 0.0436,
|
5721 |
+
"step": 16320
|
5722 |
+
},
|
5723 |
+
{
|
5724 |
+
"epoch": 79.75594874923733,
|
5725 |
+
"grad_norm": 1.5898195505142212,
|
5726 |
+
"learning_rate": 9.350819672131148e-05,
|
5727 |
+
"loss": 0.0488,
|
5728 |
+
"step": 16340
|
5729 |
+
},
|
5730 |
+
{
|
5731 |
+
"epoch": 79.8535692495424,
|
5732 |
+
"grad_norm": 1.5002379417419434,
|
5733 |
+
"learning_rate": 9.337704918032787e-05,
|
5734 |
+
"loss": 0.0453,
|
5735 |
+
"step": 16360
|
5736 |
+
},
|
5737 |
+
{
|
5738 |
+
"epoch": 79.95118974984747,
|
5739 |
+
"grad_norm": 1.7117916345596313,
|
5740 |
+
"learning_rate": 9.324590163934427e-05,
|
5741 |
+
"loss": 0.0483,
|
5742 |
+
"step": 16380
|
5743 |
+
},
|
5744 |
+
{
|
5745 |
+
"epoch": 80.04881025015253,
|
5746 |
+
"grad_norm": 1.672136664390564,
|
5747 |
+
"learning_rate": 9.311475409836066e-05,
|
5748 |
+
"loss": 0.0444,
|
5749 |
+
"step": 16400
|
5750 |
+
},
|
5751 |
+
{
|
5752 |
+
"epoch": 80.1464307504576,
|
5753 |
+
"grad_norm": 1.2889288663864136,
|
5754 |
+
"learning_rate": 9.298360655737705e-05,
|
5755 |
+
"loss": 0.0377,
|
5756 |
+
"step": 16420
|
5757 |
+
},
|
5758 |
+
{
|
5759 |
+
"epoch": 80.24405125076267,
|
5760 |
+
"grad_norm": 1.602657675743103,
|
5761 |
+
"learning_rate": 9.285245901639344e-05,
|
5762 |
+
"loss": 0.0431,
|
5763 |
+
"step": 16440
|
5764 |
+
},
|
5765 |
+
{
|
5766 |
+
"epoch": 80.34167175106772,
|
5767 |
+
"grad_norm": 1.673161506652832,
|
5768 |
+
"learning_rate": 9.272131147540985e-05,
|
5769 |
+
"loss": 0.0392,
|
5770 |
+
"step": 16460
|
5771 |
+
},
|
5772 |
+
{
|
5773 |
+
"epoch": 80.43929225137279,
|
5774 |
+
"grad_norm": 1.5590755939483643,
|
5775 |
+
"learning_rate": 9.259016393442623e-05,
|
5776 |
+
"loss": 0.0401,
|
5777 |
+
"step": 16480
|
5778 |
+
},
|
5779 |
+
{
|
5780 |
+
"epoch": 80.53691275167785,
|
5781 |
+
"grad_norm": 1.3792154788970947,
|
5782 |
+
"learning_rate": 9.245901639344264e-05,
|
5783 |
+
"loss": 0.0434,
|
5784 |
+
"step": 16500
|
5785 |
+
},
|
5786 |
+
{
|
5787 |
+
"epoch": 80.63453325198292,
|
5788 |
+
"grad_norm": 1.0110834836959839,
|
5789 |
+
"learning_rate": 9.232786885245903e-05,
|
5790 |
+
"loss": 0.0455,
|
5791 |
+
"step": 16520
|
5792 |
+
},
|
5793 |
+
{
|
5794 |
+
"epoch": 80.73215375228799,
|
5795 |
+
"grad_norm": 1.3876726627349854,
|
5796 |
+
"learning_rate": 9.21967213114754e-05,
|
5797 |
+
"loss": 0.0457,
|
5798 |
+
"step": 16540
|
5799 |
+
},
|
5800 |
+
{
|
5801 |
+
"epoch": 80.82977425259304,
|
5802 |
+
"grad_norm": 2.110999822616577,
|
5803 |
+
"learning_rate": 9.20655737704918e-05,
|
5804 |
+
"loss": 0.0467,
|
5805 |
+
"step": 16560
|
5806 |
+
},
|
5807 |
+
{
|
5808 |
+
"epoch": 80.92739475289811,
|
5809 |
+
"grad_norm": 1.9339549541473389,
|
5810 |
+
"learning_rate": 9.19344262295082e-05,
|
5811 |
+
"loss": 0.0432,
|
5812 |
+
"step": 16580
|
5813 |
+
},
|
5814 |
+
{
|
5815 |
+
"epoch": 81.02501525320318,
|
5816 |
+
"grad_norm": 1.1022757291793823,
|
5817 |
+
"learning_rate": 9.18032786885246e-05,
|
5818 |
+
"loss": 0.0454,
|
5819 |
+
"step": 16600
|
5820 |
+
},
|
5821 |
+
{
|
5822 |
+
"epoch": 81.12263575350823,
|
5823 |
+
"grad_norm": 1.7495752573013306,
|
5824 |
+
"learning_rate": 9.167213114754099e-05,
|
5825 |
+
"loss": 0.0375,
|
5826 |
+
"step": 16620
|
5827 |
+
},
|
5828 |
+
{
|
5829 |
+
"epoch": 81.2202562538133,
|
5830 |
+
"grad_norm": 1.6110937595367432,
|
5831 |
+
"learning_rate": 9.154098360655739e-05,
|
5832 |
+
"loss": 0.0399,
|
5833 |
+
"step": 16640
|
5834 |
+
},
|
5835 |
+
{
|
5836 |
+
"epoch": 81.31787675411836,
|
5837 |
+
"grad_norm": 1.7473942041397095,
|
5838 |
+
"learning_rate": 9.140983606557378e-05,
|
5839 |
+
"loss": 0.0399,
|
5840 |
+
"step": 16660
|
5841 |
+
},
|
5842 |
+
{
|
5843 |
+
"epoch": 81.41549725442343,
|
5844 |
+
"grad_norm": 1.6311129331588745,
|
5845 |
+
"learning_rate": 9.127868852459017e-05,
|
5846 |
+
"loss": 0.0397,
|
5847 |
+
"step": 16680
|
5848 |
+
},
|
5849 |
+
{
|
5850 |
+
"epoch": 81.5131177547285,
|
5851 |
+
"grad_norm": 2.0772838592529297,
|
5852 |
+
"learning_rate": 9.114754098360656e-05,
|
5853 |
+
"loss": 0.042,
|
5854 |
+
"step": 16700
|
5855 |
+
},
|
5856 |
+
{
|
5857 |
+
"epoch": 81.61073825503355,
|
5858 |
+
"grad_norm": 1.5818514823913574,
|
5859 |
+
"learning_rate": 9.101639344262296e-05,
|
5860 |
+
"loss": 0.0419,
|
5861 |
+
"step": 16720
|
5862 |
+
},
|
5863 |
+
{
|
5864 |
+
"epoch": 81.70835875533862,
|
5865 |
+
"grad_norm": 1.0751603841781616,
|
5866 |
+
"learning_rate": 9.088524590163935e-05,
|
5867 |
+
"loss": 0.0447,
|
5868 |
+
"step": 16740
|
5869 |
+
},
|
5870 |
+
{
|
5871 |
+
"epoch": 81.80597925564369,
|
5872 |
+
"grad_norm": 1.2915217876434326,
|
5873 |
+
"learning_rate": 9.075409836065574e-05,
|
5874 |
+
"loss": 0.0425,
|
5875 |
+
"step": 16760
|
5876 |
+
},
|
5877 |
+
{
|
5878 |
+
"epoch": 81.90359975594875,
|
5879 |
+
"grad_norm": 1.4814674854278564,
|
5880 |
+
"learning_rate": 9.062295081967214e-05,
|
5881 |
+
"loss": 0.0455,
|
5882 |
+
"step": 16780
|
5883 |
+
},
|
5884 |
+
{
|
5885 |
+
"epoch": 82.00122025625382,
|
5886 |
+
"grad_norm": 1.1471761465072632,
|
5887 |
+
"learning_rate": 9.049180327868852e-05,
|
5888 |
+
"loss": 0.0418,
|
5889 |
+
"step": 16800
|
5890 |
+
},
|
5891 |
+
{
|
5892 |
+
"epoch": 82.09884075655887,
|
5893 |
+
"grad_norm": 1.663212537765503,
|
5894 |
+
"learning_rate": 9.036065573770492e-05,
|
5895 |
+
"loss": 0.0348,
|
5896 |
+
"step": 16820
|
5897 |
+
},
|
5898 |
+
{
|
5899 |
+
"epoch": 82.19646125686394,
|
5900 |
+
"grad_norm": 2.423266887664795,
|
5901 |
+
"learning_rate": 9.022950819672131e-05,
|
5902 |
+
"loss": 0.0367,
|
5903 |
+
"step": 16840
|
5904 |
+
},
|
5905 |
+
{
|
5906 |
+
"epoch": 82.29408175716901,
|
5907 |
+
"grad_norm": 1.5083626508712769,
|
5908 |
+
"learning_rate": 9.009836065573771e-05,
|
5909 |
+
"loss": 0.038,
|
5910 |
+
"step": 16860
|
5911 |
+
},
|
5912 |
+
{
|
5913 |
+
"epoch": 82.39170225747407,
|
5914 |
+
"grad_norm": 1.2111772298812866,
|
5915 |
+
"learning_rate": 8.99672131147541e-05,
|
5916 |
+
"loss": 0.0404,
|
5917 |
+
"step": 16880
|
5918 |
+
},
|
5919 |
+
{
|
5920 |
+
"epoch": 82.48932275777914,
|
5921 |
+
"grad_norm": 2.1727235317230225,
|
5922 |
+
"learning_rate": 8.98360655737705e-05,
|
5923 |
+
"loss": 0.0434,
|
5924 |
+
"step": 16900
|
5925 |
+
},
|
5926 |
+
{
|
5927 |
+
"epoch": 82.5869432580842,
|
5928 |
+
"grad_norm": 1.3636887073516846,
|
5929 |
+
"learning_rate": 8.97049180327869e-05,
|
5930 |
+
"loss": 0.0404,
|
5931 |
+
"step": 16920
|
5932 |
+
},
|
5933 |
+
{
|
5934 |
+
"epoch": 82.68456375838926,
|
5935 |
+
"grad_norm": 2.5128073692321777,
|
5936 |
+
"learning_rate": 8.957377049180328e-05,
|
5937 |
+
"loss": 0.0442,
|
5938 |
+
"step": 16940
|
5939 |
+
},
|
5940 |
+
{
|
5941 |
+
"epoch": 82.78218425869433,
|
5942 |
+
"grad_norm": 1.3435509204864502,
|
5943 |
+
"learning_rate": 8.944262295081967e-05,
|
5944 |
+
"loss": 0.044,
|
5945 |
+
"step": 16960
|
5946 |
+
},
|
5947 |
+
{
|
5948 |
+
"epoch": 82.87980475899938,
|
5949 |
+
"grad_norm": 2.397655487060547,
|
5950 |
+
"learning_rate": 8.931147540983606e-05,
|
5951 |
+
"loss": 0.0447,
|
5952 |
+
"step": 16980
|
5953 |
+
},
|
5954 |
+
{
|
5955 |
+
"epoch": 82.97742525930445,
|
5956 |
+
"grad_norm": 1.5072938203811646,
|
5957 |
+
"learning_rate": 8.918032786885247e-05,
|
5958 |
+
"loss": 0.0411,
|
5959 |
+
"step": 17000
|
5960 |
+
},
|
5961 |
+
{
|
5962 |
+
"epoch": 83.07504575960952,
|
5963 |
+
"grad_norm": 1.9065088033676147,
|
5964 |
+
"learning_rate": 8.904918032786886e-05,
|
5965 |
+
"loss": 0.036,
|
5966 |
+
"step": 17020
|
5967 |
+
},
|
5968 |
+
{
|
5969 |
+
"epoch": 83.17266625991458,
|
5970 |
+
"grad_norm": 1.4267394542694092,
|
5971 |
+
"learning_rate": 8.891803278688526e-05,
|
5972 |
+
"loss": 0.035,
|
5973 |
+
"step": 17040
|
5974 |
+
},
|
5975 |
+
{
|
5976 |
+
"epoch": 83.27028676021965,
|
5977 |
+
"grad_norm": 1.677395224571228,
|
5978 |
+
"learning_rate": 8.878688524590163e-05,
|
5979 |
+
"loss": 0.0425,
|
5980 |
+
"step": 17060
|
5981 |
+
},
|
5982 |
+
{
|
5983 |
+
"epoch": 83.36790726052472,
|
5984 |
+
"grad_norm": 1.5407129526138306,
|
5985 |
+
"learning_rate": 8.865573770491804e-05,
|
5986 |
+
"loss": 0.0362,
|
5987 |
+
"step": 17080
|
5988 |
+
},
|
5989 |
+
{
|
5990 |
+
"epoch": 83.46552776082977,
|
5991 |
+
"grad_norm": 1.1154581308364868,
|
5992 |
+
"learning_rate": 8.852459016393443e-05,
|
5993 |
+
"loss": 0.0403,
|
5994 |
+
"step": 17100
|
5995 |
+
},
|
5996 |
+
{
|
5997 |
+
"epoch": 83.56314826113484,
|
5998 |
+
"grad_norm": 1.6043624877929688,
|
5999 |
+
"learning_rate": 8.839344262295083e-05,
|
6000 |
+
"loss": 0.0413,
|
6001 |
+
"step": 17120
|
6002 |
+
},
|
6003 |
+
{
|
6004 |
+
"epoch": 83.6607687614399,
|
6005 |
+
"grad_norm": 1.4952675104141235,
|
6006 |
+
"learning_rate": 8.826229508196722e-05,
|
6007 |
+
"loss": 0.0397,
|
6008 |
+
"step": 17140
|
6009 |
+
},
|
6010 |
+
{
|
6011 |
+
"epoch": 83.75838926174497,
|
6012 |
+
"grad_norm": 1.4162040948867798,
|
6013 |
+
"learning_rate": 8.813114754098362e-05,
|
6014 |
+
"loss": 0.0435,
|
6015 |
+
"step": 17160
|
6016 |
+
},
|
6017 |
+
{
|
6018 |
+
"epoch": 83.85600976205004,
|
6019 |
+
"grad_norm": 1.616909384727478,
|
6020 |
+
"learning_rate": 8.800000000000001e-05,
|
6021 |
+
"loss": 0.0455,
|
6022 |
+
"step": 17180
|
6023 |
+
},
|
6024 |
+
{
|
6025 |
+
"epoch": 83.95363026235509,
|
6026 |
+
"grad_norm": 1.3712050914764404,
|
6027 |
+
"learning_rate": 8.786885245901639e-05,
|
6028 |
+
"loss": 0.0442,
|
6029 |
+
"step": 17200
|
6030 |
+
},
|
6031 |
+
{
|
6032 |
+
"epoch": 84.05125076266016,
|
6033 |
+
"grad_norm": 1.0349308252334595,
|
6034 |
+
"learning_rate": 8.773770491803279e-05,
|
6035 |
+
"loss": 0.0368,
|
6036 |
+
"step": 17220
|
6037 |
+
},
|
6038 |
+
{
|
6039 |
+
"epoch": 84.14887126296523,
|
6040 |
+
"grad_norm": 1.427959680557251,
|
6041 |
+
"learning_rate": 8.760655737704918e-05,
|
6042 |
+
"loss": 0.0318,
|
6043 |
+
"step": 17240
|
6044 |
+
},
|
6045 |
+
{
|
6046 |
+
"epoch": 84.24649176327028,
|
6047 |
+
"grad_norm": 1.1814582347869873,
|
6048 |
+
"learning_rate": 8.747540983606558e-05,
|
6049 |
+
"loss": 0.0386,
|
6050 |
+
"step": 17260
|
6051 |
+
},
|
6052 |
+
{
|
6053 |
+
"epoch": 84.34411226357535,
|
6054 |
+
"grad_norm": 1.0893490314483643,
|
6055 |
+
"learning_rate": 8.734426229508197e-05,
|
6056 |
+
"loss": 0.0398,
|
6057 |
+
"step": 17280
|
6058 |
+
},
|
6059 |
+
{
|
6060 |
+
"epoch": 84.44173276388041,
|
6061 |
+
"grad_norm": 1.2510510683059692,
|
6062 |
+
"learning_rate": 8.721311475409837e-05,
|
6063 |
+
"loss": 0.0425,
|
6064 |
+
"step": 17300
|
6065 |
+
},
|
6066 |
+
{
|
6067 |
+
"epoch": 84.53935326418548,
|
6068 |
+
"grad_norm": 0.9642720818519592,
|
6069 |
+
"learning_rate": 8.708196721311475e-05,
|
6070 |
+
"loss": 0.0384,
|
6071 |
+
"step": 17320
|
6072 |
+
},
|
6073 |
+
{
|
6074 |
+
"epoch": 84.63697376449055,
|
6075 |
+
"grad_norm": 1.2420659065246582,
|
6076 |
+
"learning_rate": 8.695081967213115e-05,
|
6077 |
+
"loss": 0.038,
|
6078 |
+
"step": 17340
|
6079 |
+
},
|
6080 |
+
{
|
6081 |
+
"epoch": 84.7345942647956,
|
6082 |
+
"grad_norm": 1.0976778268814087,
|
6083 |
+
"learning_rate": 8.681967213114754e-05,
|
6084 |
+
"loss": 0.0428,
|
6085 |
+
"step": 17360
|
6086 |
+
},
|
6087 |
+
{
|
6088 |
+
"epoch": 84.83221476510067,
|
6089 |
+
"grad_norm": 1.282233476638794,
|
6090 |
+
"learning_rate": 8.668852459016393e-05,
|
6091 |
+
"loss": 0.0418,
|
6092 |
+
"step": 17380
|
6093 |
+
},
|
6094 |
+
{
|
6095 |
+
"epoch": 84.92983526540573,
|
6096 |
+
"grad_norm": 2.0834672451019287,
|
6097 |
+
"learning_rate": 8.655737704918033e-05,
|
6098 |
+
"loss": 0.0441,
|
6099 |
+
"step": 17400
|
6100 |
+
},
|
6101 |
+
{
|
6102 |
+
"epoch": 85.0274557657108,
|
6103 |
+
"grad_norm": 1.0747671127319336,
|
6104 |
+
"learning_rate": 8.642622950819672e-05,
|
6105 |
+
"loss": 0.0415,
|
6106 |
+
"step": 17420
|
6107 |
+
},
|
6108 |
+
{
|
6109 |
+
"epoch": 85.12507626601587,
|
6110 |
+
"grad_norm": 1.6715941429138184,
|
6111 |
+
"learning_rate": 8.629508196721313e-05,
|
6112 |
+
"loss": 0.0366,
|
6113 |
+
"step": 17440
|
6114 |
+
},
|
6115 |
+
{
|
6116 |
+
"epoch": 85.22269676632092,
|
6117 |
+
"grad_norm": 1.115537166595459,
|
6118 |
+
"learning_rate": 8.61639344262295e-05,
|
6119 |
+
"loss": 0.0326,
|
6120 |
+
"step": 17460
|
6121 |
+
},
|
6122 |
+
{
|
6123 |
+
"epoch": 85.32031726662599,
|
6124 |
+
"grad_norm": 1.118455410003662,
|
6125 |
+
"learning_rate": 8.60327868852459e-05,
|
6126 |
+
"loss": 0.0384,
|
6127 |
+
"step": 17480
|
6128 |
+
},
|
6129 |
+
{
|
6130 |
+
"epoch": 85.41793776693106,
|
6131 |
+
"grad_norm": 1.6129100322723389,
|
6132 |
+
"learning_rate": 8.59016393442623e-05,
|
6133 |
+
"loss": 0.0376,
|
6134 |
+
"step": 17500
|
6135 |
+
},
|
6136 |
+
{
|
6137 |
+
"epoch": 85.51555826723612,
|
6138 |
+
"grad_norm": 2.3377175331115723,
|
6139 |
+
"learning_rate": 8.57704918032787e-05,
|
6140 |
+
"loss": 0.0388,
|
6141 |
+
"step": 17520
|
6142 |
+
},
|
6143 |
+
{
|
6144 |
+
"epoch": 85.61317876754119,
|
6145 |
+
"grad_norm": 1.2189383506774902,
|
6146 |
+
"learning_rate": 8.563934426229509e-05,
|
6147 |
+
"loss": 0.0405,
|
6148 |
+
"step": 17540
|
6149 |
+
},
|
6150 |
+
{
|
6151 |
+
"epoch": 85.71079926784624,
|
6152 |
+
"grad_norm": 1.24251389503479,
|
6153 |
+
"learning_rate": 8.550819672131149e-05,
|
6154 |
+
"loss": 0.0426,
|
6155 |
+
"step": 17560
|
6156 |
+
},
|
6157 |
+
{
|
6158 |
+
"epoch": 85.80841976815131,
|
6159 |
+
"grad_norm": 1.1355634927749634,
|
6160 |
+
"learning_rate": 8.537704918032787e-05,
|
6161 |
+
"loss": 0.0407,
|
6162 |
+
"step": 17580
|
6163 |
+
},
|
6164 |
+
{
|
6165 |
+
"epoch": 85.90604026845638,
|
6166 |
+
"grad_norm": 1.5692590475082397,
|
6167 |
+
"learning_rate": 8.524590163934426e-05,
|
6168 |
+
"loss": 0.0438,
|
6169 |
+
"step": 17600
|
6170 |
+
},
|
6171 |
+
{
|
6172 |
+
"epoch": 86.00366076876143,
|
6173 |
+
"grad_norm": 0.9955014586448669,
|
6174 |
+
"learning_rate": 8.511475409836066e-05,
|
6175 |
+
"loss": 0.0414,
|
6176 |
+
"step": 17620
|
6177 |
+
},
|
6178 |
+
{
|
6179 |
+
"epoch": 86.1012812690665,
|
6180 |
+
"grad_norm": 1.3448798656463623,
|
6181 |
+
"learning_rate": 8.498360655737705e-05,
|
6182 |
+
"loss": 0.0307,
|
6183 |
+
"step": 17640
|
6184 |
+
},
|
6185 |
+
{
|
6186 |
+
"epoch": 86.19890176937157,
|
6187 |
+
"grad_norm": 2.263423442840576,
|
6188 |
+
"learning_rate": 8.485245901639345e-05,
|
6189 |
+
"loss": 0.0354,
|
6190 |
+
"step": 17660
|
6191 |
+
},
|
6192 |
+
{
|
6193 |
+
"epoch": 86.29652226967663,
|
6194 |
+
"grad_norm": 2.385218620300293,
|
6195 |
+
"learning_rate": 8.472131147540984e-05,
|
6196 |
+
"loss": 0.0373,
|
6197 |
+
"step": 17680
|
6198 |
+
},
|
6199 |
+
{
|
6200 |
+
"epoch": 86.3941427699817,
|
6201 |
+
"grad_norm": 1.5091354846954346,
|
6202 |
+
"learning_rate": 8.459016393442624e-05,
|
6203 |
+
"loss": 0.0366,
|
6204 |
+
"step": 17700
|
6205 |
+
},
|
6206 |
+
{
|
6207 |
+
"epoch": 86.49176327028675,
|
6208 |
+
"grad_norm": 1.300026297569275,
|
6209 |
+
"learning_rate": 8.445901639344262e-05,
|
6210 |
+
"loss": 0.0423,
|
6211 |
+
"step": 17720
|
6212 |
+
},
|
6213 |
+
{
|
6214 |
+
"epoch": 86.58938377059182,
|
6215 |
+
"grad_norm": 1.4994693994522095,
|
6216 |
+
"learning_rate": 8.432786885245902e-05,
|
6217 |
+
"loss": 0.0406,
|
6218 |
+
"step": 17740
|
6219 |
+
},
|
6220 |
+
{
|
6221 |
+
"epoch": 86.68700427089689,
|
6222 |
+
"grad_norm": 1.3377670049667358,
|
6223 |
+
"learning_rate": 8.419672131147541e-05,
|
6224 |
+
"loss": 0.0414,
|
6225 |
+
"step": 17760
|
6226 |
+
},
|
6227 |
+
{
|
6228 |
+
"epoch": 86.78462477120195,
|
6229 |
+
"grad_norm": 0.9744328260421753,
|
6230 |
+
"learning_rate": 8.406557377049181e-05,
|
6231 |
+
"loss": 0.0391,
|
6232 |
+
"step": 17780
|
6233 |
+
},
|
6234 |
+
{
|
6235 |
+
"epoch": 86.88224527150702,
|
6236 |
+
"grad_norm": 1.1345022916793823,
|
6237 |
+
"learning_rate": 8.39344262295082e-05,
|
6238 |
+
"loss": 0.0404,
|
6239 |
+
"step": 17800
|
6240 |
+
},
|
6241 |
+
{
|
6242 |
+
"epoch": 86.97986577181209,
|
6243 |
+
"grad_norm": 1.7516177892684937,
|
6244 |
+
"learning_rate": 8.380327868852459e-05,
|
6245 |
+
"loss": 0.0439,
|
6246 |
+
"step": 17820
|
6247 |
+
},
|
6248 |
+
{
|
6249 |
+
"epoch": 87.07748627211714,
|
6250 |
+
"grad_norm": 1.2439887523651123,
|
6251 |
+
"learning_rate": 8.367213114754098e-05,
|
6252 |
+
"loss": 0.0335,
|
6253 |
+
"step": 17840
|
6254 |
+
},
|
6255 |
+
{
|
6256 |
+
"epoch": 87.17510677242221,
|
6257 |
+
"grad_norm": 1.586146354675293,
|
6258 |
+
"learning_rate": 8.354098360655737e-05,
|
6259 |
+
"loss": 0.031,
|
6260 |
+
"step": 17860
|
6261 |
+
},
|
6262 |
+
{
|
6263 |
+
"epoch": 87.27272727272727,
|
6264 |
+
"grad_norm": 1.3947432041168213,
|
6265 |
+
"learning_rate": 8.340983606557377e-05,
|
6266 |
+
"loss": 0.0385,
|
6267 |
+
"step": 17880
|
6268 |
+
},
|
6269 |
+
{
|
6270 |
+
"epoch": 87.37034777303234,
|
6271 |
+
"grad_norm": 0.9418930411338806,
|
6272 |
+
"learning_rate": 8.327868852459016e-05,
|
6273 |
+
"loss": 0.0364,
|
6274 |
+
"step": 17900
|
6275 |
+
},
|
6276 |
+
{
|
6277 |
+
"epoch": 87.4679682733374,
|
6278 |
+
"grad_norm": 0.8570541739463806,
|
6279 |
+
"learning_rate": 8.314754098360657e-05,
|
6280 |
+
"loss": 0.0378,
|
6281 |
+
"step": 17920
|
6282 |
+
},
|
6283 |
+
{
|
6284 |
+
"epoch": 87.56558877364246,
|
6285 |
+
"grad_norm": 1.2564805746078491,
|
6286 |
+
"learning_rate": 8.301639344262296e-05,
|
6287 |
+
"loss": 0.0373,
|
6288 |
+
"step": 17940
|
6289 |
+
},
|
6290 |
+
{
|
6291 |
+
"epoch": 87.66320927394753,
|
6292 |
+
"grad_norm": 1.062628149986267,
|
6293 |
+
"learning_rate": 8.288524590163935e-05,
|
6294 |
+
"loss": 0.0372,
|
6295 |
+
"step": 17960
|
6296 |
+
},
|
6297 |
+
{
|
6298 |
+
"epoch": 87.7608297742526,
|
6299 |
+
"grad_norm": 1.4381693601608276,
|
6300 |
+
"learning_rate": 8.275409836065573e-05,
|
6301 |
+
"loss": 0.0417,
|
6302 |
+
"step": 17980
|
6303 |
+
},
|
6304 |
+
{
|
6305 |
+
"epoch": 87.85845027455765,
|
6306 |
+
"grad_norm": 1.6939609050750732,
|
6307 |
+
"learning_rate": 8.262295081967214e-05,
|
6308 |
+
"loss": 0.0414,
|
6309 |
+
"step": 18000
|
6310 |
+
},
|
6311 |
+
{
|
6312 |
+
"epoch": 87.95607077486272,
|
6313 |
+
"grad_norm": 1.5280972719192505,
|
6314 |
+
"learning_rate": 8.249180327868853e-05,
|
6315 |
+
"loss": 0.0411,
|
6316 |
+
"step": 18020
|
6317 |
+
},
|
6318 |
+
{
|
6319 |
+
"epoch": 88.05369127516778,
|
6320 |
+
"grad_norm": 1.0062308311462402,
|
6321 |
+
"learning_rate": 8.236065573770492e-05,
|
6322 |
+
"loss": 0.0354,
|
6323 |
+
"step": 18040
|
6324 |
+
},
|
6325 |
+
{
|
6326 |
+
"epoch": 88.15131177547285,
|
6327 |
+
"grad_norm": 2.354576826095581,
|
6328 |
+
"learning_rate": 8.222950819672132e-05,
|
6329 |
+
"loss": 0.0318,
|
6330 |
+
"step": 18060
|
6331 |
+
},
|
6332 |
+
{
|
6333 |
+
"epoch": 88.24893227577792,
|
6334 |
+
"grad_norm": 1.3578128814697266,
|
6335 |
+
"learning_rate": 8.209836065573771e-05,
|
6336 |
+
"loss": 0.0355,
|
6337 |
+
"step": 18080
|
6338 |
+
},
|
6339 |
+
{
|
6340 |
+
"epoch": 88.34655277608297,
|
6341 |
+
"grad_norm": 1.0361335277557373,
|
6342 |
+
"learning_rate": 8.19672131147541e-05,
|
6343 |
+
"loss": 0.0332,
|
6344 |
+
"step": 18100
|
6345 |
+
},
|
6346 |
+
{
|
6347 |
+
"epoch": 88.44417327638804,
|
6348 |
+
"grad_norm": 1.3577088117599487,
|
6349 |
+
"learning_rate": 8.183606557377049e-05,
|
6350 |
+
"loss": 0.0372,
|
6351 |
+
"step": 18120
|
6352 |
+
},
|
6353 |
+
{
|
6354 |
+
"epoch": 88.54179377669311,
|
6355 |
+
"grad_norm": 1.5548292398452759,
|
6356 |
+
"learning_rate": 8.170491803278689e-05,
|
6357 |
+
"loss": 0.0384,
|
6358 |
+
"step": 18140
|
6359 |
+
},
|
6360 |
+
{
|
6361 |
+
"epoch": 88.63941427699817,
|
6362 |
+
"grad_norm": 0.9935953617095947,
|
6363 |
+
"learning_rate": 8.157377049180328e-05,
|
6364 |
+
"loss": 0.0364,
|
6365 |
+
"step": 18160
|
6366 |
+
},
|
6367 |
+
{
|
6368 |
+
"epoch": 88.73703477730324,
|
6369 |
+
"grad_norm": 1.093231201171875,
|
6370 |
+
"learning_rate": 8.144262295081968e-05,
|
6371 |
+
"loss": 0.0369,
|
6372 |
+
"step": 18180
|
6373 |
+
},
|
6374 |
+
{
|
6375 |
+
"epoch": 88.83465527760829,
|
6376 |
+
"grad_norm": 1.0061986446380615,
|
6377 |
+
"learning_rate": 8.131147540983607e-05,
|
6378 |
+
"loss": 0.0415,
|
6379 |
+
"step": 18200
|
6380 |
+
},
|
6381 |
+
{
|
6382 |
+
"epoch": 88.93227577791336,
|
6383 |
+
"grad_norm": 1.2928169965744019,
|
6384 |
+
"learning_rate": 8.118032786885246e-05,
|
6385 |
+
"loss": 0.0405,
|
6386 |
+
"step": 18220
|
6387 |
+
},
|
6388 |
+
{
|
6389 |
+
"epoch": 89.02989627821843,
|
6390 |
+
"grad_norm": 1.4209178686141968,
|
6391 |
+
"learning_rate": 8.104918032786885e-05,
|
6392 |
+
"loss": 0.0382,
|
6393 |
+
"step": 18240
|
6394 |
+
},
|
6395 |
+
{
|
6396 |
+
"epoch": 89.12751677852349,
|
6397 |
+
"grad_norm": 1.7799944877624512,
|
6398 |
+
"learning_rate": 8.091803278688524e-05,
|
6399 |
+
"loss": 0.0338,
|
6400 |
+
"step": 18260
|
6401 |
+
},
|
6402 |
+
{
|
6403 |
+
"epoch": 89.22513727882856,
|
6404 |
+
"grad_norm": 0.9708789587020874,
|
6405 |
+
"learning_rate": 8.078688524590164e-05,
|
6406 |
+
"loss": 0.0362,
|
6407 |
+
"step": 18280
|
6408 |
+
},
|
6409 |
+
{
|
6410 |
+
"epoch": 89.32275777913362,
|
6411 |
+
"grad_norm": 1.9509353637695312,
|
6412 |
+
"learning_rate": 8.065573770491803e-05,
|
6413 |
+
"loss": 0.0339,
|
6414 |
+
"step": 18300
|
6415 |
+
},
|
6416 |
+
{
|
6417 |
+
"epoch": 89.42037827943868,
|
6418 |
+
"grad_norm": 1.2407267093658447,
|
6419 |
+
"learning_rate": 8.052459016393444e-05,
|
6420 |
+
"loss": 0.0368,
|
6421 |
+
"step": 18320
|
6422 |
+
},
|
6423 |
+
{
|
6424 |
+
"epoch": 89.51799877974375,
|
6425 |
+
"grad_norm": 1.3398857116699219,
|
6426 |
+
"learning_rate": 8.039344262295082e-05,
|
6427 |
+
"loss": 0.0368,
|
6428 |
+
"step": 18340
|
6429 |
+
},
|
6430 |
+
{
|
6431 |
+
"epoch": 89.6156192800488,
|
6432 |
+
"grad_norm": 1.7856793403625488,
|
6433 |
+
"learning_rate": 8.026229508196721e-05,
|
6434 |
+
"loss": 0.0421,
|
6435 |
+
"step": 18360
|
6436 |
+
},
|
6437 |
+
{
|
6438 |
+
"epoch": 89.71323978035387,
|
6439 |
+
"grad_norm": 0.8198001980781555,
|
6440 |
+
"learning_rate": 8.01311475409836e-05,
|
6441 |
+
"loss": 0.0343,
|
6442 |
+
"step": 18380
|
6443 |
+
},
|
6444 |
+
{
|
6445 |
+
"epoch": 89.81086028065894,
|
6446 |
+
"grad_norm": 1.2887758016586304,
|
6447 |
+
"learning_rate": 8e-05,
|
6448 |
+
"loss": 0.037,
|
6449 |
+
"step": 18400
|
6450 |
+
},
|
6451 |
+
{
|
6452 |
+
"epoch": 89.908480780964,
|
6453 |
+
"grad_norm": 1.6205062866210938,
|
6454 |
+
"learning_rate": 7.98688524590164e-05,
|
6455 |
+
"loss": 0.0395,
|
6456 |
+
"step": 18420
|
6457 |
+
},
|
6458 |
+
{
|
6459 |
+
"epoch": 90.00610128126907,
|
6460 |
+
"grad_norm": 1.4648716449737549,
|
6461 |
+
"learning_rate": 7.97377049180328e-05,
|
6462 |
+
"loss": 0.0387,
|
6463 |
+
"step": 18440
|
6464 |
+
},
|
6465 |
+
{
|
6466 |
+
"epoch": 90.10372178157414,
|
6467 |
+
"grad_norm": 0.8236134648323059,
|
6468 |
+
"learning_rate": 7.960655737704919e-05,
|
6469 |
+
"loss": 0.0317,
|
6470 |
+
"step": 18460
|
6471 |
+
},
|
6472 |
+
{
|
6473 |
+
"epoch": 90.20134228187919,
|
6474 |
+
"grad_norm": 1.4608750343322754,
|
6475 |
+
"learning_rate": 7.947540983606558e-05,
|
6476 |
+
"loss": 0.0347,
|
6477 |
+
"step": 18480
|
6478 |
+
},
|
6479 |
+
{
|
6480 |
+
"epoch": 90.29896278218426,
|
6481 |
+
"grad_norm": 0.9114894866943359,
|
6482 |
+
"learning_rate": 7.934426229508197e-05,
|
6483 |
+
"loss": 0.0347,
|
6484 |
+
"step": 18500
|
6485 |
+
},
|
6486 |
+
{
|
6487 |
+
"epoch": 90.39658328248932,
|
6488 |
+
"grad_norm": 0.7654961347579956,
|
6489 |
+
"learning_rate": 7.921311475409836e-05,
|
6490 |
+
"loss": 0.0324,
|
6491 |
+
"step": 18520
|
6492 |
+
},
|
6493 |
+
{
|
6494 |
+
"epoch": 90.49420378279439,
|
6495 |
+
"grad_norm": 1.1844931840896606,
|
6496 |
+
"learning_rate": 7.908196721311476e-05,
|
6497 |
+
"loss": 0.0352,
|
6498 |
+
"step": 18540
|
6499 |
+
},
|
6500 |
+
{
|
6501 |
+
"epoch": 90.59182428309946,
|
6502 |
+
"grad_norm": 1.6406841278076172,
|
6503 |
+
"learning_rate": 7.895081967213115e-05,
|
6504 |
+
"loss": 0.0352,
|
6505 |
+
"step": 18560
|
6506 |
+
},
|
6507 |
+
{
|
6508 |
+
"epoch": 90.68944478340451,
|
6509 |
+
"grad_norm": 1.5059739351272583,
|
6510 |
+
"learning_rate": 7.881967213114755e-05,
|
6511 |
+
"loss": 0.0386,
|
6512 |
+
"step": 18580
|
6513 |
+
},
|
6514 |
+
{
|
6515 |
+
"epoch": 90.78706528370958,
|
6516 |
+
"grad_norm": 0.7421912550926208,
|
6517 |
+
"learning_rate": 7.868852459016394e-05,
|
6518 |
+
"loss": 0.0381,
|
6519 |
+
"step": 18600
|
6520 |
+
},
|
6521 |
+
{
|
6522 |
+
"epoch": 90.88468578401465,
|
6523 |
+
"grad_norm": 1.2397936582565308,
|
6524 |
+
"learning_rate": 7.855737704918033e-05,
|
6525 |
+
"loss": 0.0365,
|
6526 |
+
"step": 18620
|
6527 |
+
},
|
6528 |
+
{
|
6529 |
+
"epoch": 90.9823062843197,
|
6530 |
+
"grad_norm": 1.0268443822860718,
|
6531 |
+
"learning_rate": 7.842622950819672e-05,
|
6532 |
+
"loss": 0.0402,
|
6533 |
+
"step": 18640
|
6534 |
+
},
|
6535 |
+
{
|
6536 |
+
"epoch": 91.07992678462477,
|
6537 |
+
"grad_norm": 0.87629234790802,
|
6538 |
+
"learning_rate": 7.829508196721311e-05,
|
6539 |
+
"loss": 0.0308,
|
6540 |
+
"step": 18660
|
6541 |
+
},
|
6542 |
+
{
|
6543 |
+
"epoch": 91.17754728492983,
|
6544 |
+
"grad_norm": 2.2132532596588135,
|
6545 |
+
"learning_rate": 7.816393442622951e-05,
|
6546 |
+
"loss": 0.031,
|
6547 |
+
"step": 18680
|
6548 |
+
},
|
6549 |
+
{
|
6550 |
+
"epoch": 91.2751677852349,
|
6551 |
+
"grad_norm": 1.2565170526504517,
|
6552 |
+
"learning_rate": 7.80327868852459e-05,
|
6553 |
+
"loss": 0.034,
|
6554 |
+
"step": 18700
|
6555 |
+
},
|
6556 |
+
{
|
6557 |
+
"epoch": 91.37278828553997,
|
6558 |
+
"grad_norm": 1.191254734992981,
|
6559 |
+
"learning_rate": 7.79016393442623e-05,
|
6560 |
+
"loss": 0.0348,
|
6561 |
+
"step": 18720
|
6562 |
+
},
|
6563 |
+
{
|
6564 |
+
"epoch": 91.47040878584502,
|
6565 |
+
"grad_norm": 0.711301326751709,
|
6566 |
+
"learning_rate": 7.77704918032787e-05,
|
6567 |
+
"loss": 0.0342,
|
6568 |
+
"step": 18740
|
6569 |
+
},
|
6570 |
+
{
|
6571 |
+
"epoch": 91.5680292861501,
|
6572 |
+
"grad_norm": 0.8898797035217285,
|
6573 |
+
"learning_rate": 7.763934426229508e-05,
|
6574 |
+
"loss": 0.0348,
|
6575 |
+
"step": 18760
|
6576 |
+
},
|
6577 |
+
{
|
6578 |
+
"epoch": 91.66564978645516,
|
6579 |
+
"grad_norm": 0.8842754364013672,
|
6580 |
+
"learning_rate": 7.750819672131147e-05,
|
6581 |
+
"loss": 0.0343,
|
6582 |
+
"step": 18780
|
6583 |
+
},
|
6584 |
+
{
|
6585 |
+
"epoch": 91.76327028676022,
|
6586 |
+
"grad_norm": 1.7899420261383057,
|
6587 |
+
"learning_rate": 7.737704918032788e-05,
|
6588 |
+
"loss": 0.0395,
|
6589 |
+
"step": 18800
|
6590 |
+
},
|
6591 |
+
{
|
6592 |
+
"epoch": 91.86089078706529,
|
6593 |
+
"grad_norm": 1.710175633430481,
|
6594 |
+
"learning_rate": 7.724590163934426e-05,
|
6595 |
+
"loss": 0.0387,
|
6596 |
+
"step": 18820
|
6597 |
+
},
|
6598 |
+
{
|
6599 |
+
"epoch": 91.95851128737034,
|
6600 |
+
"grad_norm": 1.8949339389801025,
|
6601 |
+
"learning_rate": 7.711475409836067e-05,
|
6602 |
+
"loss": 0.0409,
|
6603 |
+
"step": 18840
|
6604 |
+
},
|
6605 |
+
{
|
6606 |
+
"epoch": 92.05613178767541,
|
6607 |
+
"grad_norm": 1.6512871980667114,
|
6608 |
+
"learning_rate": 7.698360655737706e-05,
|
6609 |
+
"loss": 0.0353,
|
6610 |
+
"step": 18860
|
6611 |
+
},
|
6612 |
+
{
|
6613 |
+
"epoch": 92.15375228798048,
|
6614 |
+
"grad_norm": 1.1428453922271729,
|
6615 |
+
"learning_rate": 7.685245901639345e-05,
|
6616 |
+
"loss": 0.031,
|
6617 |
+
"step": 18880
|
6618 |
+
},
|
6619 |
+
{
|
6620 |
+
"epoch": 92.25137278828554,
|
6621 |
+
"grad_norm": 1.9489351511001587,
|
6622 |
+
"learning_rate": 7.672131147540984e-05,
|
6623 |
+
"loss": 0.0328,
|
6624 |
+
"step": 18900
|
6625 |
+
},
|
6626 |
+
{
|
6627 |
+
"epoch": 92.3489932885906,
|
6628 |
+
"grad_norm": 1.148255467414856,
|
6629 |
+
"learning_rate": 7.659016393442622e-05,
|
6630 |
+
"loss": 0.0334,
|
6631 |
+
"step": 18920
|
6632 |
+
},
|
6633 |
+
{
|
6634 |
+
"epoch": 92.44661378889568,
|
6635 |
+
"grad_norm": 1.2118985652923584,
|
6636 |
+
"learning_rate": 7.645901639344263e-05,
|
6637 |
+
"loss": 0.0319,
|
6638 |
+
"step": 18940
|
6639 |
+
},
|
6640 |
+
{
|
6641 |
+
"epoch": 92.54423428920073,
|
6642 |
+
"grad_norm": 0.9434294700622559,
|
6643 |
+
"learning_rate": 7.632786885245902e-05,
|
6644 |
+
"loss": 0.0377,
|
6645 |
+
"step": 18960
|
6646 |
+
},
|
6647 |
+
{
|
6648 |
+
"epoch": 92.6418547895058,
|
6649 |
+
"grad_norm": 1.136614441871643,
|
6650 |
+
"learning_rate": 7.619672131147542e-05,
|
6651 |
+
"loss": 0.0358,
|
6652 |
+
"step": 18980
|
6653 |
+
},
|
6654 |
+
{
|
6655 |
+
"epoch": 92.73947528981085,
|
6656 |
+
"grad_norm": 1.6465675830841064,
|
6657 |
+
"learning_rate": 7.606557377049181e-05,
|
6658 |
+
"loss": 0.0335,
|
6659 |
+
"step": 19000
|
6660 |
+
},
|
6661 |
+
{
|
6662 |
+
"epoch": 92.83709579011592,
|
6663 |
+
"grad_norm": 1.623631477355957,
|
6664 |
+
"learning_rate": 7.59344262295082e-05,
|
6665 |
+
"loss": 0.0377,
|
6666 |
+
"step": 19020
|
6667 |
+
},
|
6668 |
+
{
|
6669 |
+
"epoch": 92.934716290421,
|
6670 |
+
"grad_norm": 1.2394886016845703,
|
6671 |
+
"learning_rate": 7.580327868852459e-05,
|
6672 |
+
"loss": 0.0378,
|
6673 |
+
"step": 19040
|
6674 |
+
},
|
6675 |
+
{
|
6676 |
+
"epoch": 93.03233679072605,
|
6677 |
+
"grad_norm": 1.0836046934127808,
|
6678 |
+
"learning_rate": 7.567213114754099e-05,
|
6679 |
+
"loss": 0.0374,
|
6680 |
+
"step": 19060
|
6681 |
+
},
|
6682 |
+
{
|
6683 |
+
"epoch": 93.12995729103112,
|
6684 |
+
"grad_norm": 1.2980749607086182,
|
6685 |
+
"learning_rate": 7.554098360655738e-05,
|
6686 |
+
"loss": 0.0299,
|
6687 |
+
"step": 19080
|
6688 |
+
},
|
6689 |
+
{
|
6690 |
+
"epoch": 93.22757779133617,
|
6691 |
+
"grad_norm": 0.7896084785461426,
|
6692 |
+
"learning_rate": 7.540983606557377e-05,
|
6693 |
+
"loss": 0.0338,
|
6694 |
+
"step": 19100
|
6695 |
+
},
|
6696 |
+
{
|
6697 |
+
"epoch": 93.32519829164124,
|
6698 |
+
"grad_norm": 1.4562427997589111,
|
6699 |
+
"learning_rate": 7.527868852459017e-05,
|
6700 |
+
"loss": 0.0325,
|
6701 |
+
"step": 19120
|
6702 |
+
},
|
6703 |
+
{
|
6704 |
+
"epoch": 93.42281879194631,
|
6705 |
+
"grad_norm": 1.0484211444854736,
|
6706 |
+
"learning_rate": 7.514754098360656e-05,
|
6707 |
+
"loss": 0.0319,
|
6708 |
+
"step": 19140
|
6709 |
+
},
|
6710 |
+
{
|
6711 |
+
"epoch": 93.52043929225137,
|
6712 |
+
"grad_norm": 1.2180018424987793,
|
6713 |
+
"learning_rate": 7.501639344262295e-05,
|
6714 |
+
"loss": 0.0358,
|
6715 |
+
"step": 19160
|
6716 |
+
},
|
6717 |
+
{
|
6718 |
+
"epoch": 93.61805979255644,
|
6719 |
+
"grad_norm": 1.5725558996200562,
|
6720 |
+
"learning_rate": 7.488524590163934e-05,
|
6721 |
+
"loss": 0.0371,
|
6722 |
+
"step": 19180
|
6723 |
+
},
|
6724 |
+
{
|
6725 |
+
"epoch": 93.7156802928615,
|
6726 |
+
"grad_norm": 0.962448000907898,
|
6727 |
+
"learning_rate": 7.475409836065574e-05,
|
6728 |
+
"loss": 0.0329,
|
6729 |
+
"step": 19200
|
6730 |
+
},
|
6731 |
+
{
|
6732 |
+
"epoch": 93.81330079316656,
|
6733 |
+
"grad_norm": 1.2971309423446655,
|
6734 |
+
"learning_rate": 7.462295081967213e-05,
|
6735 |
+
"loss": 0.0388,
|
6736 |
+
"step": 19220
|
6737 |
+
},
|
6738 |
+
{
|
6739 |
+
"epoch": 93.91092129347163,
|
6740 |
+
"grad_norm": 1.621389627456665,
|
6741 |
+
"learning_rate": 7.449180327868854e-05,
|
6742 |
+
"loss": 0.035,
|
6743 |
+
"step": 19240
|
6744 |
+
},
|
6745 |
+
{
|
6746 |
+
"epoch": 94.00854179377669,
|
6747 |
+
"grad_norm": 1.1524312496185303,
|
6748 |
+
"learning_rate": 7.436065573770493e-05,
|
6749 |
+
"loss": 0.036,
|
6750 |
+
"step": 19260
|
6751 |
+
},
|
6752 |
+
{
|
6753 |
+
"epoch": 94.10616229408176,
|
6754 |
+
"grad_norm": 1.0361751317977905,
|
6755 |
+
"learning_rate": 7.422950819672131e-05,
|
6756 |
+
"loss": 0.0311,
|
6757 |
+
"step": 19280
|
6758 |
+
},
|
6759 |
+
{
|
6760 |
+
"epoch": 94.20378279438683,
|
6761 |
+
"grad_norm": 1.2280365228652954,
|
6762 |
+
"learning_rate": 7.40983606557377e-05,
|
6763 |
+
"loss": 0.032,
|
6764 |
+
"step": 19300
|
6765 |
+
},
|
6766 |
+
{
|
6767 |
+
"epoch": 94.30140329469188,
|
6768 |
+
"grad_norm": 1.39139723777771,
|
6769 |
+
"learning_rate": 7.39672131147541e-05,
|
6770 |
+
"loss": 0.0312,
|
6771 |
+
"step": 19320
|
6772 |
+
},
|
6773 |
+
{
|
6774 |
+
"epoch": 94.39902379499695,
|
6775 |
+
"grad_norm": 2.101762533187866,
|
6776 |
+
"learning_rate": 7.38360655737705e-05,
|
6777 |
+
"loss": 0.0304,
|
6778 |
+
"step": 19340
|
6779 |
+
},
|
6780 |
+
{
|
6781 |
+
"epoch": 94.49664429530202,
|
6782 |
+
"grad_norm": 1.5655161142349243,
|
6783 |
+
"learning_rate": 7.370491803278689e-05,
|
6784 |
+
"loss": 0.0332,
|
6785 |
+
"step": 19360
|
6786 |
+
},
|
6787 |
+
{
|
6788 |
+
"epoch": 94.59426479560707,
|
6789 |
+
"grad_norm": 1.1902984380722046,
|
6790 |
+
"learning_rate": 7.357377049180329e-05,
|
6791 |
+
"loss": 0.0308,
|
6792 |
+
"step": 19380
|
6793 |
+
},
|
6794 |
+
{
|
6795 |
+
"epoch": 94.69188529591214,
|
6796 |
+
"grad_norm": 1.3503782749176025,
|
6797 |
+
"learning_rate": 7.344262295081968e-05,
|
6798 |
+
"loss": 0.0342,
|
6799 |
+
"step": 19400
|
6800 |
+
},
|
6801 |
+
{
|
6802 |
+
"epoch": 94.7895057962172,
|
6803 |
+
"grad_norm": 1.6430237293243408,
|
6804 |
+
"learning_rate": 7.331147540983607e-05,
|
6805 |
+
"loss": 0.0375,
|
6806 |
+
"step": 19420
|
6807 |
+
},
|
6808 |
+
{
|
6809 |
+
"epoch": 94.88712629652227,
|
6810 |
+
"grad_norm": 1.4753270149230957,
|
6811 |
+
"learning_rate": 7.318032786885246e-05,
|
6812 |
+
"loss": 0.0361,
|
6813 |
+
"step": 19440
|
6814 |
+
},
|
6815 |
+
{
|
6816 |
+
"epoch": 94.98474679682734,
|
6817 |
+
"grad_norm": 1.6849883794784546,
|
6818 |
+
"learning_rate": 7.304918032786886e-05,
|
6819 |
+
"loss": 0.0377,
|
6820 |
+
"step": 19460
|
6821 |
+
},
|
6822 |
+
{
|
6823 |
+
"epoch": 95.0823672971324,
|
6824 |
+
"grad_norm": 0.9980133175849915,
|
6825 |
+
"learning_rate": 7.291803278688525e-05,
|
6826 |
+
"loss": 0.0286,
|
6827 |
+
"step": 19480
|
6828 |
+
},
|
6829 |
+
{
|
6830 |
+
"epoch": 95.17998779743746,
|
6831 |
+
"grad_norm": 0.911799430847168,
|
6832 |
+
"learning_rate": 7.278688524590165e-05,
|
6833 |
+
"loss": 0.0304,
|
6834 |
+
"step": 19500
|
6835 |
+
},
|
6836 |
+
{
|
6837 |
+
"epoch": 95.27760829774253,
|
6838 |
+
"grad_norm": 1.0748980045318604,
|
6839 |
+
"learning_rate": 7.265573770491804e-05,
|
6840 |
+
"loss": 0.0312,
|
6841 |
+
"step": 19520
|
6842 |
+
},
|
6843 |
+
{
|
6844 |
+
"epoch": 95.37522879804759,
|
6845 |
+
"grad_norm": 1.6856895685195923,
|
6846 |
+
"learning_rate": 7.252459016393443e-05,
|
6847 |
+
"loss": 0.0313,
|
6848 |
+
"step": 19540
|
6849 |
+
},
|
6850 |
+
{
|
6851 |
+
"epoch": 95.47284929835266,
|
6852 |
+
"grad_norm": 1.281265139579773,
|
6853 |
+
"learning_rate": 7.239344262295082e-05,
|
6854 |
+
"loss": 0.0325,
|
6855 |
+
"step": 19560
|
6856 |
+
},
|
6857 |
+
{
|
6858 |
+
"epoch": 95.57046979865771,
|
6859 |
+
"grad_norm": 1.3441650867462158,
|
6860 |
+
"learning_rate": 7.226229508196721e-05,
|
6861 |
+
"loss": 0.0369,
|
6862 |
+
"step": 19580
|
6863 |
+
},
|
6864 |
+
{
|
6865 |
+
"epoch": 95.66809029896278,
|
6866 |
+
"grad_norm": 0.9338245391845703,
|
6867 |
+
"learning_rate": 7.213114754098361e-05,
|
6868 |
+
"loss": 0.0328,
|
6869 |
+
"step": 19600
|
6870 |
+
},
|
6871 |
+
{
|
6872 |
+
"epoch": 95.76571079926785,
|
6873 |
+
"grad_norm": 1.0416558980941772,
|
6874 |
+
"learning_rate": 7.2e-05,
|
6875 |
+
"loss": 0.0313,
|
6876 |
+
"step": 19620
|
6877 |
+
},
|
6878 |
+
{
|
6879 |
+
"epoch": 95.8633312995729,
|
6880 |
+
"grad_norm": 1.187015175819397,
|
6881 |
+
"learning_rate": 7.18688524590164e-05,
|
6882 |
+
"loss": 0.0389,
|
6883 |
+
"step": 19640
|
6884 |
+
},
|
6885 |
+
{
|
6886 |
+
"epoch": 95.96095179987798,
|
6887 |
+
"grad_norm": 1.7248045206069946,
|
6888 |
+
"learning_rate": 7.17377049180328e-05,
|
6889 |
+
"loss": 0.0389,
|
6890 |
+
"step": 19660
|
6891 |
+
},
|
6892 |
+
{
|
6893 |
+
"epoch": 96.05857230018304,
|
6894 |
+
"grad_norm": 1.244112253189087,
|
6895 |
+
"learning_rate": 7.160655737704918e-05,
|
6896 |
+
"loss": 0.0282,
|
6897 |
+
"step": 19680
|
6898 |
+
},
|
6899 |
+
{
|
6900 |
+
"epoch": 96.1561928004881,
|
6901 |
+
"grad_norm": 0.868563711643219,
|
6902 |
+
"learning_rate": 7.147540983606557e-05,
|
6903 |
+
"loss": 0.03,
|
6904 |
+
"step": 19700
|
6905 |
+
},
|
6906 |
+
{
|
6907 |
+
"epoch": 96.25381330079317,
|
6908 |
+
"grad_norm": 0.6851422190666199,
|
6909 |
+
"learning_rate": 7.134426229508198e-05,
|
6910 |
+
"loss": 0.0293,
|
6911 |
+
"step": 19720
|
6912 |
+
},
|
6913 |
+
{
|
6914 |
+
"epoch": 96.35143380109822,
|
6915 |
+
"grad_norm": 1.0482094287872314,
|
6916 |
+
"learning_rate": 7.121311475409837e-05,
|
6917 |
+
"loss": 0.0348,
|
6918 |
+
"step": 19740
|
6919 |
+
},
|
6920 |
+
{
|
6921 |
+
"epoch": 96.4490543014033,
|
6922 |
+
"grad_norm": 0.9821768403053284,
|
6923 |
+
"learning_rate": 7.108196721311475e-05,
|
6924 |
+
"loss": 0.0346,
|
6925 |
+
"step": 19760
|
6926 |
+
},
|
6927 |
+
{
|
6928 |
+
"epoch": 96.54667480170836,
|
6929 |
+
"grad_norm": 1.2944159507751465,
|
6930 |
+
"learning_rate": 7.095081967213116e-05,
|
6931 |
+
"loss": 0.033,
|
6932 |
+
"step": 19780
|
6933 |
+
},
|
6934 |
+
{
|
6935 |
+
"epoch": 96.64429530201342,
|
6936 |
+
"grad_norm": 1.797796368598938,
|
6937 |
+
"learning_rate": 7.081967213114755e-05,
|
6938 |
+
"loss": 0.0313,
|
6939 |
+
"step": 19800
|
6940 |
+
},
|
6941 |
+
{
|
6942 |
+
"epoch": 96.74191580231849,
|
6943 |
+
"grad_norm": 2.1053693294525146,
|
6944 |
+
"learning_rate": 7.068852459016394e-05,
|
6945 |
+
"loss": 0.0321,
|
6946 |
+
"step": 19820
|
6947 |
+
},
|
6948 |
+
{
|
6949 |
+
"epoch": 96.83953630262356,
|
6950 |
+
"grad_norm": 0.6129661202430725,
|
6951 |
+
"learning_rate": 7.055737704918033e-05,
|
6952 |
+
"loss": 0.0343,
|
6953 |
+
"step": 19840
|
6954 |
+
},
|
6955 |
+
{
|
6956 |
+
"epoch": 96.93715680292861,
|
6957 |
+
"grad_norm": 1.184906005859375,
|
6958 |
+
"learning_rate": 7.042622950819673e-05,
|
6959 |
+
"loss": 0.0335,
|
6960 |
+
"step": 19860
|
6961 |
+
},
|
6962 |
+
{
|
6963 |
+
"epoch": 97.03477730323368,
|
6964 |
+
"grad_norm": 1.0629398822784424,
|
6965 |
+
"learning_rate": 7.029508196721312e-05,
|
6966 |
+
"loss": 0.0336,
|
6967 |
+
"step": 19880
|
6968 |
+
},
|
6969 |
+
{
|
6970 |
+
"epoch": 97.13239780353874,
|
6971 |
+
"grad_norm": 1.0311895608901978,
|
6972 |
+
"learning_rate": 7.016393442622952e-05,
|
6973 |
+
"loss": 0.028,
|
6974 |
+
"step": 19900
|
6975 |
+
},
|
6976 |
+
{
|
6977 |
+
"epoch": 97.2300183038438,
|
6978 |
+
"grad_norm": 1.3785455226898193,
|
6979 |
+
"learning_rate": 7.003278688524591e-05,
|
6980 |
+
"loss": 0.0304,
|
6981 |
+
"step": 19920
|
6982 |
+
},
|
6983 |
+
{
|
6984 |
+
"epoch": 97.32763880414888,
|
6985 |
+
"grad_norm": 1.515749216079712,
|
6986 |
+
"learning_rate": 6.99016393442623e-05,
|
6987 |
+
"loss": 0.0314,
|
6988 |
+
"step": 19940
|
6989 |
+
},
|
6990 |
+
{
|
6991 |
+
"epoch": 97.42525930445393,
|
6992 |
+
"grad_norm": 1.5662028789520264,
|
6993 |
+
"learning_rate": 6.977049180327869e-05,
|
6994 |
+
"loss": 0.0302,
|
6995 |
+
"step": 19960
|
6996 |
+
},
|
6997 |
+
{
|
6998 |
+
"epoch": 97.522879804759,
|
6999 |
+
"grad_norm": 1.1467646360397339,
|
7000 |
+
"learning_rate": 6.963934426229508e-05,
|
7001 |
+
"loss": 0.0317,
|
7002 |
+
"step": 19980
|
7003 |
+
},
|
7004 |
+
{
|
7005 |
+
"epoch": 97.62050030506407,
|
7006 |
+
"grad_norm": 0.9327864646911621,
|
7007 |
+
"learning_rate": 6.950819672131148e-05,
|
7008 |
+
"loss": 0.0323,
|
7009 |
+
"step": 20000
|
7010 |
}
|
7011 |
],
|
7012 |
"logging_steps": 20,
|
|
|
7026 |
"attributes": {}
|
7027 |
}
|
7028 |
},
|
7029 |
+
"total_flos": 2.1916893437732045e+17,
|
7030 |
"train_batch_size": 1,
|
7031 |
"trial_name": null,
|
7032 |
"trial_params": null
|