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:a61eceda4f3e583f262e8ebf73b2c9e19f6b690f584bb98977ea218db783977a
|
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:e093ed1c707300997497dbc1e6b17c4b6c332b015791188ac5b2a741423a2c5d
|
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 sha256:40904a91aa9d0b18232b04f286c69d79a167a7891f9c1462d77da5a10280249d
|
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:b907aed6e1e57922cb480424961aba116ee5b766ca9857e1376e5fbbce10214f
|
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,
|
@@ -3507,6 +3507,1756 @@
|
|
3507 |
"learning_rate": 0.00013508196721311478,
|
3508 |
"loss": 0.0949,
|
3509 |
"step": 10000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
3510 |
}
|
3511 |
],
|
3512 |
"logging_steps": 20,
|
@@ -3526,7 +5276,7 @@
|
|
3526 |
"attributes": {}
|
3527 |
}
|
3528 |
},
|
3529 |
-
"total_flos": 1.
|
3530 |
"train_batch_size": 1,
|
3531 |
"trial_name": null,
|
3532 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 73.21537522879805,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 15000,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
3507 |
"learning_rate": 0.00013508196721311478,
|
3508 |
"loss": 0.0949,
|
3509 |
"step": 10000
|
3510 |
+
},
|
3511 |
+
{
|
3512 |
+
"epoch": 48.9078706528371,
|
3513 |
+
"grad_norm": 2.472350835800171,
|
3514 |
+
"learning_rate": 0.00013495081967213114,
|
3515 |
+
"loss": 0.0868,
|
3516 |
+
"step": 10020
|
3517 |
+
},
|
3518 |
+
{
|
3519 |
+
"epoch": 49.00549115314216,
|
3520 |
+
"grad_norm": 2.2226836681365967,
|
3521 |
+
"learning_rate": 0.00013481967213114756,
|
3522 |
+
"loss": 0.0984,
|
3523 |
+
"step": 10040
|
3524 |
+
},
|
3525 |
+
{
|
3526 |
+
"epoch": 49.10311165344722,
|
3527 |
+
"grad_norm": 1.5109550952911377,
|
3528 |
+
"learning_rate": 0.00013468852459016395,
|
3529 |
+
"loss": 0.0608,
|
3530 |
+
"step": 10060
|
3531 |
+
},
|
3532 |
+
{
|
3533 |
+
"epoch": 49.20073215375229,
|
3534 |
+
"grad_norm": 2.532590627670288,
|
3535 |
+
"learning_rate": 0.00013455737704918034,
|
3536 |
+
"loss": 0.0695,
|
3537 |
+
"step": 10080
|
3538 |
+
},
|
3539 |
+
{
|
3540 |
+
"epoch": 49.298352654057354,
|
3541 |
+
"grad_norm": 2.6609067916870117,
|
3542 |
+
"learning_rate": 0.00013442622950819673,
|
3543 |
+
"loss": 0.082,
|
3544 |
+
"step": 10100
|
3545 |
+
},
|
3546 |
+
{
|
3547 |
+
"epoch": 49.395973154362416,
|
3548 |
+
"grad_norm": 2.5846164226531982,
|
3549 |
+
"learning_rate": 0.00013429508196721311,
|
3550 |
+
"loss": 0.08,
|
3551 |
+
"step": 10120
|
3552 |
+
},
|
3553 |
+
{
|
3554 |
+
"epoch": 49.49359365466748,
|
3555 |
+
"grad_norm": 2.4516260623931885,
|
3556 |
+
"learning_rate": 0.00013416393442622953,
|
3557 |
+
"loss": 0.0796,
|
3558 |
+
"step": 10140
|
3559 |
+
},
|
3560 |
+
{
|
3561 |
+
"epoch": 49.59121415497255,
|
3562 |
+
"grad_norm": 2.357771158218384,
|
3563 |
+
"learning_rate": 0.0001340327868852459,
|
3564 |
+
"loss": 0.0823,
|
3565 |
+
"step": 10160
|
3566 |
+
},
|
3567 |
+
{
|
3568 |
+
"epoch": 49.68883465527761,
|
3569 |
+
"grad_norm": 2.575505018234253,
|
3570 |
+
"learning_rate": 0.0001339016393442623,
|
3571 |
+
"loss": 0.0899,
|
3572 |
+
"step": 10180
|
3573 |
+
},
|
3574 |
+
{
|
3575 |
+
"epoch": 49.78645515558267,
|
3576 |
+
"grad_norm": 2.2076785564422607,
|
3577 |
+
"learning_rate": 0.0001337704918032787,
|
3578 |
+
"loss": 0.0822,
|
3579 |
+
"step": 10200
|
3580 |
+
},
|
3581 |
+
{
|
3582 |
+
"epoch": 49.884075655887735,
|
3583 |
+
"grad_norm": 2.361694097518921,
|
3584 |
+
"learning_rate": 0.0001336393442622951,
|
3585 |
+
"loss": 0.0913,
|
3586 |
+
"step": 10220
|
3587 |
+
},
|
3588 |
+
{
|
3589 |
+
"epoch": 49.981696156192804,
|
3590 |
+
"grad_norm": 2.3063478469848633,
|
3591 |
+
"learning_rate": 0.00013350819672131148,
|
3592 |
+
"loss": 0.0919,
|
3593 |
+
"step": 10240
|
3594 |
+
},
|
3595 |
+
{
|
3596 |
+
"epoch": 50.079316656497866,
|
3597 |
+
"grad_norm": 1.9585156440734863,
|
3598 |
+
"learning_rate": 0.00013337704918032787,
|
3599 |
+
"loss": 0.0767,
|
3600 |
+
"step": 10260
|
3601 |
+
},
|
3602 |
+
{
|
3603 |
+
"epoch": 50.17693715680293,
|
3604 |
+
"grad_norm": 3.1279983520507812,
|
3605 |
+
"learning_rate": 0.00013324590163934428,
|
3606 |
+
"loss": 0.0755,
|
3607 |
+
"step": 10280
|
3608 |
+
},
|
3609 |
+
{
|
3610 |
+
"epoch": 50.27455765710799,
|
3611 |
+
"grad_norm": 2.0095512866973877,
|
3612 |
+
"learning_rate": 0.00013311475409836067,
|
3613 |
+
"loss": 0.0735,
|
3614 |
+
"step": 10300
|
3615 |
+
},
|
3616 |
+
{
|
3617 |
+
"epoch": 50.37217815741306,
|
3618 |
+
"grad_norm": 3.011828899383545,
|
3619 |
+
"learning_rate": 0.00013298360655737706,
|
3620 |
+
"loss": 0.0766,
|
3621 |
+
"step": 10320
|
3622 |
+
},
|
3623 |
+
{
|
3624 |
+
"epoch": 50.46979865771812,
|
3625 |
+
"grad_norm": 2.0386946201324463,
|
3626 |
+
"learning_rate": 0.00013285245901639345,
|
3627 |
+
"loss": 0.0738,
|
3628 |
+
"step": 10340
|
3629 |
+
},
|
3630 |
+
{
|
3631 |
+
"epoch": 50.567419158023185,
|
3632 |
+
"grad_norm": 2.0549607276916504,
|
3633 |
+
"learning_rate": 0.00013272131147540984,
|
3634 |
+
"loss": 0.089,
|
3635 |
+
"step": 10360
|
3636 |
+
},
|
3637 |
+
{
|
3638 |
+
"epoch": 50.66503965832825,
|
3639 |
+
"grad_norm": 2.816814422607422,
|
3640 |
+
"learning_rate": 0.00013259016393442623,
|
3641 |
+
"loss": 0.0849,
|
3642 |
+
"step": 10380
|
3643 |
+
},
|
3644 |
+
{
|
3645 |
+
"epoch": 50.76266015863331,
|
3646 |
+
"grad_norm": 2.2058887481689453,
|
3647 |
+
"learning_rate": 0.00013245901639344262,
|
3648 |
+
"loss": 0.0842,
|
3649 |
+
"step": 10400
|
3650 |
+
},
|
3651 |
+
{
|
3652 |
+
"epoch": 50.86028065893838,
|
3653 |
+
"grad_norm": 2.2805745601654053,
|
3654 |
+
"learning_rate": 0.00013232786885245904,
|
3655 |
+
"loss": 0.0888,
|
3656 |
+
"step": 10420
|
3657 |
+
},
|
3658 |
+
{
|
3659 |
+
"epoch": 50.95790115924344,
|
3660 |
+
"grad_norm": 2.602515697479248,
|
3661 |
+
"learning_rate": 0.00013219672131147543,
|
3662 |
+
"loss": 0.0867,
|
3663 |
+
"step": 10440
|
3664 |
+
},
|
3665 |
+
{
|
3666 |
+
"epoch": 51.0555216595485,
|
3667 |
+
"grad_norm": 1.9022042751312256,
|
3668 |
+
"learning_rate": 0.0001320655737704918,
|
3669 |
+
"loss": 0.0748,
|
3670 |
+
"step": 10460
|
3671 |
+
},
|
3672 |
+
{
|
3673 |
+
"epoch": 51.153142159853566,
|
3674 |
+
"grad_norm": 2.4302797317504883,
|
3675 |
+
"learning_rate": 0.0001319344262295082,
|
3676 |
+
"loss": 0.0693,
|
3677 |
+
"step": 10480
|
3678 |
+
},
|
3679 |
+
{
|
3680 |
+
"epoch": 51.250762660158635,
|
3681 |
+
"grad_norm": 1.5306838750839233,
|
3682 |
+
"learning_rate": 0.0001318032786885246,
|
3683 |
+
"loss": 0.0748,
|
3684 |
+
"step": 10500
|
3685 |
+
},
|
3686 |
+
{
|
3687 |
+
"epoch": 51.3483831604637,
|
3688 |
+
"grad_norm": 2.035921812057495,
|
3689 |
+
"learning_rate": 0.000131672131147541,
|
3690 |
+
"loss": 0.0728,
|
3691 |
+
"step": 10520
|
3692 |
+
},
|
3693 |
+
{
|
3694 |
+
"epoch": 51.44600366076876,
|
3695 |
+
"grad_norm": 1.8543293476104736,
|
3696 |
+
"learning_rate": 0.00013154098360655737,
|
3697 |
+
"loss": 0.0752,
|
3698 |
+
"step": 10540
|
3699 |
+
},
|
3700 |
+
{
|
3701 |
+
"epoch": 51.54362416107382,
|
3702 |
+
"grad_norm": 2.405411720275879,
|
3703 |
+
"learning_rate": 0.0001314098360655738,
|
3704 |
+
"loss": 0.0775,
|
3705 |
+
"step": 10560
|
3706 |
+
},
|
3707 |
+
{
|
3708 |
+
"epoch": 51.64124466137889,
|
3709 |
+
"grad_norm": 2.767051935195923,
|
3710 |
+
"learning_rate": 0.00013127868852459018,
|
3711 |
+
"loss": 0.075,
|
3712 |
+
"step": 10580
|
3713 |
+
},
|
3714 |
+
{
|
3715 |
+
"epoch": 51.738865161683954,
|
3716 |
+
"grad_norm": 2.200438976287842,
|
3717 |
+
"learning_rate": 0.00013114754098360654,
|
3718 |
+
"loss": 0.084,
|
3719 |
+
"step": 10600
|
3720 |
+
},
|
3721 |
+
{
|
3722 |
+
"epoch": 51.836485661989016,
|
3723 |
+
"grad_norm": 2.7286131381988525,
|
3724 |
+
"learning_rate": 0.00013101639344262296,
|
3725 |
+
"loss": 0.0863,
|
3726 |
+
"step": 10620
|
3727 |
+
},
|
3728 |
+
{
|
3729 |
+
"epoch": 51.93410616229408,
|
3730 |
+
"grad_norm": 1.557892084121704,
|
3731 |
+
"learning_rate": 0.00013088524590163935,
|
3732 |
+
"loss": 0.0832,
|
3733 |
+
"step": 10640
|
3734 |
+
},
|
3735 |
+
{
|
3736 |
+
"epoch": 52.03172666259915,
|
3737 |
+
"grad_norm": 1.6213957071304321,
|
3738 |
+
"learning_rate": 0.00013075409836065576,
|
3739 |
+
"loss": 0.0835,
|
3740 |
+
"step": 10660
|
3741 |
+
},
|
3742 |
+
{
|
3743 |
+
"epoch": 52.12934716290421,
|
3744 |
+
"grad_norm": 1.9538192749023438,
|
3745 |
+
"learning_rate": 0.00013062295081967213,
|
3746 |
+
"loss": 0.0632,
|
3747 |
+
"step": 10680
|
3748 |
+
},
|
3749 |
+
{
|
3750 |
+
"epoch": 52.22696766320927,
|
3751 |
+
"grad_norm": 1.5224443674087524,
|
3752 |
+
"learning_rate": 0.00013049180327868854,
|
3753 |
+
"loss": 0.0691,
|
3754 |
+
"step": 10700
|
3755 |
+
},
|
3756 |
+
{
|
3757 |
+
"epoch": 52.324588163514335,
|
3758 |
+
"grad_norm": 2.0320827960968018,
|
3759 |
+
"learning_rate": 0.00013036065573770493,
|
3760 |
+
"loss": 0.0777,
|
3761 |
+
"step": 10720
|
3762 |
+
},
|
3763 |
+
{
|
3764 |
+
"epoch": 52.422208663819404,
|
3765 |
+
"grad_norm": 2.9247395992279053,
|
3766 |
+
"learning_rate": 0.00013022950819672132,
|
3767 |
+
"loss": 0.0738,
|
3768 |
+
"step": 10740
|
3769 |
+
},
|
3770 |
+
{
|
3771 |
+
"epoch": 52.51982916412447,
|
3772 |
+
"grad_norm": 2.0458271503448486,
|
3773 |
+
"learning_rate": 0.0001300983606557377,
|
3774 |
+
"loss": 0.0716,
|
3775 |
+
"step": 10760
|
3776 |
+
},
|
3777 |
+
{
|
3778 |
+
"epoch": 52.61744966442953,
|
3779 |
+
"grad_norm": 2.1657023429870605,
|
3780 |
+
"learning_rate": 0.0001299672131147541,
|
3781 |
+
"loss": 0.0798,
|
3782 |
+
"step": 10780
|
3783 |
+
},
|
3784 |
+
{
|
3785 |
+
"epoch": 52.71507016473459,
|
3786 |
+
"grad_norm": 2.819549798965454,
|
3787 |
+
"learning_rate": 0.00012983606557377052,
|
3788 |
+
"loss": 0.0869,
|
3789 |
+
"step": 10800
|
3790 |
+
},
|
3791 |
+
{
|
3792 |
+
"epoch": 52.81269066503966,
|
3793 |
+
"grad_norm": 2.455775737762451,
|
3794 |
+
"learning_rate": 0.00012970491803278688,
|
3795 |
+
"loss": 0.0784,
|
3796 |
+
"step": 10820
|
3797 |
+
},
|
3798 |
+
{
|
3799 |
+
"epoch": 52.91031116534472,
|
3800 |
+
"grad_norm": 2.803501844406128,
|
3801 |
+
"learning_rate": 0.0001295737704918033,
|
3802 |
+
"loss": 0.0802,
|
3803 |
+
"step": 10840
|
3804 |
+
},
|
3805 |
+
{
|
3806 |
+
"epoch": 53.007931665649785,
|
3807 |
+
"grad_norm": 2.2946598529815674,
|
3808 |
+
"learning_rate": 0.00012944262295081968,
|
3809 |
+
"loss": 0.0767,
|
3810 |
+
"step": 10860
|
3811 |
+
},
|
3812 |
+
{
|
3813 |
+
"epoch": 53.10555216595485,
|
3814 |
+
"grad_norm": 1.9233254194259644,
|
3815 |
+
"learning_rate": 0.00012931147540983607,
|
3816 |
+
"loss": 0.0652,
|
3817 |
+
"step": 10880
|
3818 |
+
},
|
3819 |
+
{
|
3820 |
+
"epoch": 53.20317266625992,
|
3821 |
+
"grad_norm": 1.8651479482650757,
|
3822 |
+
"learning_rate": 0.00012918032786885246,
|
3823 |
+
"loss": 0.0675,
|
3824 |
+
"step": 10900
|
3825 |
+
},
|
3826 |
+
{
|
3827 |
+
"epoch": 53.30079316656498,
|
3828 |
+
"grad_norm": 2.216233253479004,
|
3829 |
+
"learning_rate": 0.00012904918032786885,
|
3830 |
+
"loss": 0.0716,
|
3831 |
+
"step": 10920
|
3832 |
+
},
|
3833 |
+
{
|
3834 |
+
"epoch": 53.39841366687004,
|
3835 |
+
"grad_norm": 2.591519355773926,
|
3836 |
+
"learning_rate": 0.00012891803278688527,
|
3837 |
+
"loss": 0.073,
|
3838 |
+
"step": 10940
|
3839 |
+
},
|
3840 |
+
{
|
3841 |
+
"epoch": 53.496034167175104,
|
3842 |
+
"grad_norm": 2.490187406539917,
|
3843 |
+
"learning_rate": 0.00012878688524590166,
|
3844 |
+
"loss": 0.0715,
|
3845 |
+
"step": 10960
|
3846 |
+
},
|
3847 |
+
{
|
3848 |
+
"epoch": 53.59365466748017,
|
3849 |
+
"grad_norm": 2.2473397254943848,
|
3850 |
+
"learning_rate": 0.00012865573770491802,
|
3851 |
+
"loss": 0.0715,
|
3852 |
+
"step": 10980
|
3853 |
+
},
|
3854 |
+
{
|
3855 |
+
"epoch": 53.691275167785236,
|
3856 |
+
"grad_norm": 1.9002201557159424,
|
3857 |
+
"learning_rate": 0.00012852459016393444,
|
3858 |
+
"loss": 0.0762,
|
3859 |
+
"step": 11000
|
3860 |
+
},
|
3861 |
+
{
|
3862 |
+
"epoch": 53.7888956680903,
|
3863 |
+
"grad_norm": 2.4991517066955566,
|
3864 |
+
"learning_rate": 0.00012839344262295083,
|
3865 |
+
"loss": 0.0869,
|
3866 |
+
"step": 11020
|
3867 |
+
},
|
3868 |
+
{
|
3869 |
+
"epoch": 53.88651616839536,
|
3870 |
+
"grad_norm": 2.5355913639068604,
|
3871 |
+
"learning_rate": 0.00012826229508196722,
|
3872 |
+
"loss": 0.0801,
|
3873 |
+
"step": 11040
|
3874 |
+
},
|
3875 |
+
{
|
3876 |
+
"epoch": 53.98413666870043,
|
3877 |
+
"grad_norm": 1.895386815071106,
|
3878 |
+
"learning_rate": 0.0001281311475409836,
|
3879 |
+
"loss": 0.0747,
|
3880 |
+
"step": 11060
|
3881 |
+
},
|
3882 |
+
{
|
3883 |
+
"epoch": 54.08175716900549,
|
3884 |
+
"grad_norm": 2.150261163711548,
|
3885 |
+
"learning_rate": 0.00012800000000000002,
|
3886 |
+
"loss": 0.0646,
|
3887 |
+
"step": 11080
|
3888 |
+
},
|
3889 |
+
{
|
3890 |
+
"epoch": 54.179377669310554,
|
3891 |
+
"grad_norm": 4.12515926361084,
|
3892 |
+
"learning_rate": 0.0001278688524590164,
|
3893 |
+
"loss": 0.0693,
|
3894 |
+
"step": 11100
|
3895 |
+
},
|
3896 |
+
{
|
3897 |
+
"epoch": 54.27699816961562,
|
3898 |
+
"grad_norm": 1.3180917501449585,
|
3899 |
+
"learning_rate": 0.00012773770491803277,
|
3900 |
+
"loss": 0.069,
|
3901 |
+
"step": 11120
|
3902 |
+
},
|
3903 |
+
{
|
3904 |
+
"epoch": 54.374618669920686,
|
3905 |
+
"grad_norm": 1.714634895324707,
|
3906 |
+
"learning_rate": 0.0001276065573770492,
|
3907 |
+
"loss": 0.0701,
|
3908 |
+
"step": 11140
|
3909 |
+
},
|
3910 |
+
{
|
3911 |
+
"epoch": 54.47223917022575,
|
3912 |
+
"grad_norm": 1.93068265914917,
|
3913 |
+
"learning_rate": 0.00012747540983606558,
|
3914 |
+
"loss": 0.0682,
|
3915 |
+
"step": 11160
|
3916 |
+
},
|
3917 |
+
{
|
3918 |
+
"epoch": 54.56985967053081,
|
3919 |
+
"grad_norm": 1.9187488555908203,
|
3920 |
+
"learning_rate": 0.000127344262295082,
|
3921 |
+
"loss": 0.0778,
|
3922 |
+
"step": 11180
|
3923 |
+
},
|
3924 |
+
{
|
3925 |
+
"epoch": 54.66748017083587,
|
3926 |
+
"grad_norm": 2.3723368644714355,
|
3927 |
+
"learning_rate": 0.00012721311475409836,
|
3928 |
+
"loss": 0.0769,
|
3929 |
+
"step": 11200
|
3930 |
+
},
|
3931 |
+
{
|
3932 |
+
"epoch": 54.76510067114094,
|
3933 |
+
"grad_norm": 1.7806050777435303,
|
3934 |
+
"learning_rate": 0.00012708196721311477,
|
3935 |
+
"loss": 0.0729,
|
3936 |
+
"step": 11220
|
3937 |
+
},
|
3938 |
+
{
|
3939 |
+
"epoch": 54.862721171446005,
|
3940 |
+
"grad_norm": 1.839896321296692,
|
3941 |
+
"learning_rate": 0.00012695081967213116,
|
3942 |
+
"loss": 0.0711,
|
3943 |
+
"step": 11240
|
3944 |
+
},
|
3945 |
+
{
|
3946 |
+
"epoch": 54.96034167175107,
|
3947 |
+
"grad_norm": 3.055255889892578,
|
3948 |
+
"learning_rate": 0.00012681967213114753,
|
3949 |
+
"loss": 0.077,
|
3950 |
+
"step": 11260
|
3951 |
+
},
|
3952 |
+
{
|
3953 |
+
"epoch": 55.05796217205613,
|
3954 |
+
"grad_norm": 1.504418134689331,
|
3955 |
+
"learning_rate": 0.00012668852459016394,
|
3956 |
+
"loss": 0.0711,
|
3957 |
+
"step": 11280
|
3958 |
+
},
|
3959 |
+
{
|
3960 |
+
"epoch": 55.1555826723612,
|
3961 |
+
"grad_norm": 3.407508373260498,
|
3962 |
+
"learning_rate": 0.00012655737704918033,
|
3963 |
+
"loss": 0.0762,
|
3964 |
+
"step": 11300
|
3965 |
+
},
|
3966 |
+
{
|
3967 |
+
"epoch": 55.25320317266626,
|
3968 |
+
"grad_norm": 1.7444705963134766,
|
3969 |
+
"learning_rate": 0.00012642622950819675,
|
3970 |
+
"loss": 0.0676,
|
3971 |
+
"step": 11320
|
3972 |
+
},
|
3973 |
+
{
|
3974 |
+
"epoch": 55.35082367297132,
|
3975 |
+
"grad_norm": 1.7877843379974365,
|
3976 |
+
"learning_rate": 0.0001262950819672131,
|
3977 |
+
"loss": 0.0611,
|
3978 |
+
"step": 11340
|
3979 |
+
},
|
3980 |
+
{
|
3981 |
+
"epoch": 55.448444173276386,
|
3982 |
+
"grad_norm": 2.1501216888427734,
|
3983 |
+
"learning_rate": 0.0001261639344262295,
|
3984 |
+
"loss": 0.0702,
|
3985 |
+
"step": 11360
|
3986 |
+
},
|
3987 |
+
{
|
3988 |
+
"epoch": 55.546064673581455,
|
3989 |
+
"grad_norm": 1.4569560289382935,
|
3990 |
+
"learning_rate": 0.00012603278688524592,
|
3991 |
+
"loss": 0.0651,
|
3992 |
+
"step": 11380
|
3993 |
+
},
|
3994 |
+
{
|
3995 |
+
"epoch": 55.64368517388652,
|
3996 |
+
"grad_norm": 1.5365070104599,
|
3997 |
+
"learning_rate": 0.0001259016393442623,
|
3998 |
+
"loss": 0.0708,
|
3999 |
+
"step": 11400
|
4000 |
+
},
|
4001 |
+
{
|
4002 |
+
"epoch": 55.74130567419158,
|
4003 |
+
"grad_norm": 2.2428839206695557,
|
4004 |
+
"learning_rate": 0.0001257704918032787,
|
4005 |
+
"loss": 0.0694,
|
4006 |
+
"step": 11420
|
4007 |
+
},
|
4008 |
+
{
|
4009 |
+
"epoch": 55.83892617449664,
|
4010 |
+
"grad_norm": 2.054964542388916,
|
4011 |
+
"learning_rate": 0.00012563934426229508,
|
4012 |
+
"loss": 0.0754,
|
4013 |
+
"step": 11440
|
4014 |
+
},
|
4015 |
+
{
|
4016 |
+
"epoch": 55.93654667480171,
|
4017 |
+
"grad_norm": 3.42903733253479,
|
4018 |
+
"learning_rate": 0.0001255081967213115,
|
4019 |
+
"loss": 0.0752,
|
4020 |
+
"step": 11460
|
4021 |
+
},
|
4022 |
+
{
|
4023 |
+
"epoch": 56.034167175106774,
|
4024 |
+
"grad_norm": 1.6482346057891846,
|
4025 |
+
"learning_rate": 0.00012537704918032786,
|
4026 |
+
"loss": 0.068,
|
4027 |
+
"step": 11480
|
4028 |
+
},
|
4029 |
+
{
|
4030 |
+
"epoch": 56.131787675411836,
|
4031 |
+
"grad_norm": 2.0840466022491455,
|
4032 |
+
"learning_rate": 0.00012524590163934425,
|
4033 |
+
"loss": 0.0624,
|
4034 |
+
"step": 11500
|
4035 |
+
},
|
4036 |
+
{
|
4037 |
+
"epoch": 56.2294081757169,
|
4038 |
+
"grad_norm": 2.7554879188537598,
|
4039 |
+
"learning_rate": 0.00012511475409836067,
|
4040 |
+
"loss": 0.0592,
|
4041 |
+
"step": 11520
|
4042 |
+
},
|
4043 |
+
{
|
4044 |
+
"epoch": 56.32702867602197,
|
4045 |
+
"grad_norm": 2.8247487545013428,
|
4046 |
+
"learning_rate": 0.00012498360655737706,
|
4047 |
+
"loss": 0.0661,
|
4048 |
+
"step": 11540
|
4049 |
+
},
|
4050 |
+
{
|
4051 |
+
"epoch": 56.42464917632703,
|
4052 |
+
"grad_norm": 2.2485480308532715,
|
4053 |
+
"learning_rate": 0.00012485245901639345,
|
4054 |
+
"loss": 0.0712,
|
4055 |
+
"step": 11560
|
4056 |
+
},
|
4057 |
+
{
|
4058 |
+
"epoch": 56.52226967663209,
|
4059 |
+
"grad_norm": 2.1642680168151855,
|
4060 |
+
"learning_rate": 0.00012472131147540984,
|
4061 |
+
"loss": 0.0752,
|
4062 |
+
"step": 11580
|
4063 |
+
},
|
4064 |
+
{
|
4065 |
+
"epoch": 56.619890176937155,
|
4066 |
+
"grad_norm": 1.827771544456482,
|
4067 |
+
"learning_rate": 0.00012459016393442625,
|
4068 |
+
"loss": 0.0707,
|
4069 |
+
"step": 11600
|
4070 |
+
},
|
4071 |
+
{
|
4072 |
+
"epoch": 56.717510677242224,
|
4073 |
+
"grad_norm": 1.9134409427642822,
|
4074 |
+
"learning_rate": 0.00012445901639344262,
|
4075 |
+
"loss": 0.0696,
|
4076 |
+
"step": 11620
|
4077 |
+
},
|
4078 |
+
{
|
4079 |
+
"epoch": 56.815131177547286,
|
4080 |
+
"grad_norm": 2.3444812297821045,
|
4081 |
+
"learning_rate": 0.000124327868852459,
|
4082 |
+
"loss": 0.065,
|
4083 |
+
"step": 11640
|
4084 |
+
},
|
4085 |
+
{
|
4086 |
+
"epoch": 56.91275167785235,
|
4087 |
+
"grad_norm": 2.0354135036468506,
|
4088 |
+
"learning_rate": 0.00012419672131147542,
|
4089 |
+
"loss": 0.0746,
|
4090 |
+
"step": 11660
|
4091 |
+
},
|
4092 |
+
{
|
4093 |
+
"epoch": 57.01037217815741,
|
4094 |
+
"grad_norm": 1.5275322198867798,
|
4095 |
+
"learning_rate": 0.0001240655737704918,
|
4096 |
+
"loss": 0.0756,
|
4097 |
+
"step": 11680
|
4098 |
+
},
|
4099 |
+
{
|
4100 |
+
"epoch": 57.10799267846248,
|
4101 |
+
"grad_norm": 2.310797929763794,
|
4102 |
+
"learning_rate": 0.0001239344262295082,
|
4103 |
+
"loss": 0.057,
|
4104 |
+
"step": 11700
|
4105 |
+
},
|
4106 |
+
{
|
4107 |
+
"epoch": 57.20561317876754,
|
4108 |
+
"grad_norm": 1.903204083442688,
|
4109 |
+
"learning_rate": 0.0001238032786885246,
|
4110 |
+
"loss": 0.0657,
|
4111 |
+
"step": 11720
|
4112 |
+
},
|
4113 |
+
{
|
4114 |
+
"epoch": 57.303233679072605,
|
4115 |
+
"grad_norm": 2.115591049194336,
|
4116 |
+
"learning_rate": 0.000123672131147541,
|
4117 |
+
"loss": 0.0624,
|
4118 |
+
"step": 11740
|
4119 |
+
},
|
4120 |
+
{
|
4121 |
+
"epoch": 57.40085417937767,
|
4122 |
+
"grad_norm": 2.1732656955718994,
|
4123 |
+
"learning_rate": 0.0001235409836065574,
|
4124 |
+
"loss": 0.0636,
|
4125 |
+
"step": 11760
|
4126 |
+
},
|
4127 |
+
{
|
4128 |
+
"epoch": 57.49847467968274,
|
4129 |
+
"grad_norm": 2.2554845809936523,
|
4130 |
+
"learning_rate": 0.00012340983606557376,
|
4131 |
+
"loss": 0.0691,
|
4132 |
+
"step": 11780
|
4133 |
+
},
|
4134 |
+
{
|
4135 |
+
"epoch": 57.5960951799878,
|
4136 |
+
"grad_norm": 2.994563341140747,
|
4137 |
+
"learning_rate": 0.00012327868852459017,
|
4138 |
+
"loss": 0.0724,
|
4139 |
+
"step": 11800
|
4140 |
+
},
|
4141 |
+
{
|
4142 |
+
"epoch": 57.69371568029286,
|
4143 |
+
"grad_norm": 2.824934482574463,
|
4144 |
+
"learning_rate": 0.00012314754098360656,
|
4145 |
+
"loss": 0.07,
|
4146 |
+
"step": 11820
|
4147 |
+
},
|
4148 |
+
{
|
4149 |
+
"epoch": 57.79133618059792,
|
4150 |
+
"grad_norm": 2.435310125350952,
|
4151 |
+
"learning_rate": 0.00012301639344262295,
|
4152 |
+
"loss": 0.0753,
|
4153 |
+
"step": 11840
|
4154 |
+
},
|
4155 |
+
{
|
4156 |
+
"epoch": 57.88895668090299,
|
4157 |
+
"grad_norm": 2.140200138092041,
|
4158 |
+
"learning_rate": 0.00012288524590163934,
|
4159 |
+
"loss": 0.0762,
|
4160 |
+
"step": 11860
|
4161 |
+
},
|
4162 |
+
{
|
4163 |
+
"epoch": 57.986577181208055,
|
4164 |
+
"grad_norm": 2.013983726501465,
|
4165 |
+
"learning_rate": 0.00012275409836065573,
|
4166 |
+
"loss": 0.067,
|
4167 |
+
"step": 11880
|
4168 |
+
},
|
4169 |
+
{
|
4170 |
+
"epoch": 58.08419768151312,
|
4171 |
+
"grad_norm": 1.8855304718017578,
|
4172 |
+
"learning_rate": 0.00012262295081967215,
|
4173 |
+
"loss": 0.0606,
|
4174 |
+
"step": 11900
|
4175 |
+
},
|
4176 |
+
{
|
4177 |
+
"epoch": 58.18181818181818,
|
4178 |
+
"grad_norm": 2.3649916648864746,
|
4179 |
+
"learning_rate": 0.0001224918032786885,
|
4180 |
+
"loss": 0.0626,
|
4181 |
+
"step": 11920
|
4182 |
+
},
|
4183 |
+
{
|
4184 |
+
"epoch": 58.27943868212325,
|
4185 |
+
"grad_norm": 2.566420078277588,
|
4186 |
+
"learning_rate": 0.00012236065573770493,
|
4187 |
+
"loss": 0.059,
|
4188 |
+
"step": 11940
|
4189 |
+
},
|
4190 |
+
{
|
4191 |
+
"epoch": 58.37705918242831,
|
4192 |
+
"grad_norm": 2.3648924827575684,
|
4193 |
+
"learning_rate": 0.00012222950819672132,
|
4194 |
+
"loss": 0.0644,
|
4195 |
+
"step": 11960
|
4196 |
+
},
|
4197 |
+
{
|
4198 |
+
"epoch": 58.474679682733374,
|
4199 |
+
"grad_norm": 1.7778581380844116,
|
4200 |
+
"learning_rate": 0.00012209836065573773,
|
4201 |
+
"loss": 0.0691,
|
4202 |
+
"step": 11980
|
4203 |
+
},
|
4204 |
+
{
|
4205 |
+
"epoch": 58.572300183038436,
|
4206 |
+
"grad_norm": 1.8090124130249023,
|
4207 |
+
"learning_rate": 0.0001219672131147541,
|
4208 |
+
"loss": 0.0685,
|
4209 |
+
"step": 12000
|
4210 |
+
},
|
4211 |
+
{
|
4212 |
+
"epoch": 58.669920683343506,
|
4213 |
+
"grad_norm": 2.2762486934661865,
|
4214 |
+
"learning_rate": 0.0001218360655737705,
|
4215 |
+
"loss": 0.0634,
|
4216 |
+
"step": 12020
|
4217 |
+
},
|
4218 |
+
{
|
4219 |
+
"epoch": 58.76754118364857,
|
4220 |
+
"grad_norm": 2.279273509979248,
|
4221 |
+
"learning_rate": 0.0001217049180327869,
|
4222 |
+
"loss": 0.0681,
|
4223 |
+
"step": 12040
|
4224 |
+
},
|
4225 |
+
{
|
4226 |
+
"epoch": 58.86516168395363,
|
4227 |
+
"grad_norm": 1.5724430084228516,
|
4228 |
+
"learning_rate": 0.00012157377049180328,
|
4229 |
+
"loss": 0.0712,
|
4230 |
+
"step": 12060
|
4231 |
+
},
|
4232 |
+
{
|
4233 |
+
"epoch": 58.96278218425869,
|
4234 |
+
"grad_norm": 1.7655867338180542,
|
4235 |
+
"learning_rate": 0.00012144262295081968,
|
4236 |
+
"loss": 0.0711,
|
4237 |
+
"step": 12080
|
4238 |
+
},
|
4239 |
+
{
|
4240 |
+
"epoch": 59.060402684563755,
|
4241 |
+
"grad_norm": 1.0987945795059204,
|
4242 |
+
"learning_rate": 0.00012131147540983607,
|
4243 |
+
"loss": 0.0614,
|
4244 |
+
"step": 12100
|
4245 |
+
},
|
4246 |
+
{
|
4247 |
+
"epoch": 59.158023184868824,
|
4248 |
+
"grad_norm": 1.433910608291626,
|
4249 |
+
"learning_rate": 0.00012118032786885247,
|
4250 |
+
"loss": 0.0522,
|
4251 |
+
"step": 12120
|
4252 |
+
},
|
4253 |
+
{
|
4254 |
+
"epoch": 59.25564368517389,
|
4255 |
+
"grad_norm": 1.4189090728759766,
|
4256 |
+
"learning_rate": 0.00012104918032786885,
|
4257 |
+
"loss": 0.0603,
|
4258 |
+
"step": 12140
|
4259 |
+
},
|
4260 |
+
{
|
4261 |
+
"epoch": 59.35326418547895,
|
4262 |
+
"grad_norm": 1.915209174156189,
|
4263 |
+
"learning_rate": 0.00012091803278688525,
|
4264 |
+
"loss": 0.0631,
|
4265 |
+
"step": 12160
|
4266 |
+
},
|
4267 |
+
{
|
4268 |
+
"epoch": 59.45088468578401,
|
4269 |
+
"grad_norm": 2.0415852069854736,
|
4270 |
+
"learning_rate": 0.00012078688524590165,
|
4271 |
+
"loss": 0.065,
|
4272 |
+
"step": 12180
|
4273 |
+
},
|
4274 |
+
{
|
4275 |
+
"epoch": 59.54850518608908,
|
4276 |
+
"grad_norm": 1.8674949407577515,
|
4277 |
+
"learning_rate": 0.00012065573770491804,
|
4278 |
+
"loss": 0.0636,
|
4279 |
+
"step": 12200
|
4280 |
+
},
|
4281 |
+
{
|
4282 |
+
"epoch": 59.64612568639414,
|
4283 |
+
"grad_norm": 3.768343687057495,
|
4284 |
+
"learning_rate": 0.00012052459016393443,
|
4285 |
+
"loss": 0.0666,
|
4286 |
+
"step": 12220
|
4287 |
+
},
|
4288 |
+
{
|
4289 |
+
"epoch": 59.743746186699205,
|
4290 |
+
"grad_norm": 1.8944804668426514,
|
4291 |
+
"learning_rate": 0.00012039344262295082,
|
4292 |
+
"loss": 0.0698,
|
4293 |
+
"step": 12240
|
4294 |
+
},
|
4295 |
+
{
|
4296 |
+
"epoch": 59.84136668700427,
|
4297 |
+
"grad_norm": 2.296515464782715,
|
4298 |
+
"learning_rate": 0.00012026229508196722,
|
4299 |
+
"loss": 0.0715,
|
4300 |
+
"step": 12260
|
4301 |
+
},
|
4302 |
+
{
|
4303 |
+
"epoch": 59.93898718730934,
|
4304 |
+
"grad_norm": 1.957593560218811,
|
4305 |
+
"learning_rate": 0.0001201311475409836,
|
4306 |
+
"loss": 0.0681,
|
4307 |
+
"step": 12280
|
4308 |
+
},
|
4309 |
+
{
|
4310 |
+
"epoch": 60.0366076876144,
|
4311 |
+
"grad_norm": 1.8776521682739258,
|
4312 |
+
"learning_rate": 0.00012,
|
4313 |
+
"loss": 0.0676,
|
4314 |
+
"step": 12300
|
4315 |
+
},
|
4316 |
+
{
|
4317 |
+
"epoch": 60.13422818791946,
|
4318 |
+
"grad_norm": 1.3199673891067505,
|
4319 |
+
"learning_rate": 0.0001198688524590164,
|
4320 |
+
"loss": 0.0515,
|
4321 |
+
"step": 12320
|
4322 |
+
},
|
4323 |
+
{
|
4324 |
+
"epoch": 60.231848688224524,
|
4325 |
+
"grad_norm": 1.8994742631912231,
|
4326 |
+
"learning_rate": 0.0001197377049180328,
|
4327 |
+
"loss": 0.0566,
|
4328 |
+
"step": 12340
|
4329 |
+
},
|
4330 |
+
{
|
4331 |
+
"epoch": 60.32946918852959,
|
4332 |
+
"grad_norm": 1.8091648817062378,
|
4333 |
+
"learning_rate": 0.00011960655737704917,
|
4334 |
+
"loss": 0.0628,
|
4335 |
+
"step": 12360
|
4336 |
+
},
|
4337 |
+
{
|
4338 |
+
"epoch": 60.427089688834656,
|
4339 |
+
"grad_norm": 1.7513519525527954,
|
4340 |
+
"learning_rate": 0.00011947540983606557,
|
4341 |
+
"loss": 0.0652,
|
4342 |
+
"step": 12380
|
4343 |
+
},
|
4344 |
+
{
|
4345 |
+
"epoch": 60.52471018913972,
|
4346 |
+
"grad_norm": 1.8451545238494873,
|
4347 |
+
"learning_rate": 0.00011934426229508198,
|
4348 |
+
"loss": 0.063,
|
4349 |
+
"step": 12400
|
4350 |
+
},
|
4351 |
+
{
|
4352 |
+
"epoch": 60.62233068944478,
|
4353 |
+
"grad_norm": 1.6873937845230103,
|
4354 |
+
"learning_rate": 0.00011921311475409838,
|
4355 |
+
"loss": 0.0666,
|
4356 |
+
"step": 12420
|
4357 |
+
},
|
4358 |
+
{
|
4359 |
+
"epoch": 60.71995118974985,
|
4360 |
+
"grad_norm": 2.2382214069366455,
|
4361 |
+
"learning_rate": 0.00011908196721311476,
|
4362 |
+
"loss": 0.0593,
|
4363 |
+
"step": 12440
|
4364 |
+
},
|
4365 |
+
{
|
4366 |
+
"epoch": 60.81757169005491,
|
4367 |
+
"grad_norm": 1.4879833459854126,
|
4368 |
+
"learning_rate": 0.00011895081967213116,
|
4369 |
+
"loss": 0.0718,
|
4370 |
+
"step": 12460
|
4371 |
+
},
|
4372 |
+
{
|
4373 |
+
"epoch": 60.915192190359974,
|
4374 |
+
"grad_norm": 1.8051949739456177,
|
4375 |
+
"learning_rate": 0.00011881967213114755,
|
4376 |
+
"loss": 0.0659,
|
4377 |
+
"step": 12480
|
4378 |
+
},
|
4379 |
+
{
|
4380 |
+
"epoch": 61.01281269066504,
|
4381 |
+
"grad_norm": 2.070537567138672,
|
4382 |
+
"learning_rate": 0.00011868852459016392,
|
4383 |
+
"loss": 0.0647,
|
4384 |
+
"step": 12500
|
4385 |
+
},
|
4386 |
+
{
|
4387 |
+
"epoch": 61.110433190970106,
|
4388 |
+
"grad_norm": 2.195227861404419,
|
4389 |
+
"learning_rate": 0.00011855737704918033,
|
4390 |
+
"loss": 0.0516,
|
4391 |
+
"step": 12520
|
4392 |
+
},
|
4393 |
+
{
|
4394 |
+
"epoch": 61.20805369127517,
|
4395 |
+
"grad_norm": 2.3762006759643555,
|
4396 |
+
"learning_rate": 0.00011842622950819673,
|
4397 |
+
"loss": 0.056,
|
4398 |
+
"step": 12540
|
4399 |
+
},
|
4400 |
+
{
|
4401 |
+
"epoch": 61.30567419158023,
|
4402 |
+
"grad_norm": 1.836585521697998,
|
4403 |
+
"learning_rate": 0.00011829508196721313,
|
4404 |
+
"loss": 0.0624,
|
4405 |
+
"step": 12560
|
4406 |
+
},
|
4407 |
+
{
|
4408 |
+
"epoch": 61.40329469188529,
|
4409 |
+
"grad_norm": 1.6185029745101929,
|
4410 |
+
"learning_rate": 0.00011816393442622951,
|
4411 |
+
"loss": 0.0591,
|
4412 |
+
"step": 12580
|
4413 |
+
},
|
4414 |
+
{
|
4415 |
+
"epoch": 61.50091519219036,
|
4416 |
+
"grad_norm": 2.116314649581909,
|
4417 |
+
"learning_rate": 0.00011803278688524591,
|
4418 |
+
"loss": 0.0615,
|
4419 |
+
"step": 12600
|
4420 |
+
},
|
4421 |
+
{
|
4422 |
+
"epoch": 61.598535692495425,
|
4423 |
+
"grad_norm": 2.873701333999634,
|
4424 |
+
"learning_rate": 0.0001179016393442623,
|
4425 |
+
"loss": 0.0624,
|
4426 |
+
"step": 12620
|
4427 |
+
},
|
4428 |
+
{
|
4429 |
+
"epoch": 61.69615619280049,
|
4430 |
+
"grad_norm": 1.8966227769851685,
|
4431 |
+
"learning_rate": 0.0001177704918032787,
|
4432 |
+
"loss": 0.0643,
|
4433 |
+
"step": 12640
|
4434 |
+
},
|
4435 |
+
{
|
4436 |
+
"epoch": 61.79377669310555,
|
4437 |
+
"grad_norm": 2.343330144882202,
|
4438 |
+
"learning_rate": 0.00011763934426229508,
|
4439 |
+
"loss": 0.0637,
|
4440 |
+
"step": 12660
|
4441 |
+
},
|
4442 |
+
{
|
4443 |
+
"epoch": 61.89139719341062,
|
4444 |
+
"grad_norm": 2.882538318634033,
|
4445 |
+
"learning_rate": 0.00011750819672131148,
|
4446 |
+
"loss": 0.0678,
|
4447 |
+
"step": 12680
|
4448 |
+
},
|
4449 |
+
{
|
4450 |
+
"epoch": 61.98901769371568,
|
4451 |
+
"grad_norm": 3.441783905029297,
|
4452 |
+
"learning_rate": 0.00011737704918032789,
|
4453 |
+
"loss": 0.0667,
|
4454 |
+
"step": 12700
|
4455 |
+
},
|
4456 |
+
{
|
4457 |
+
"epoch": 62.08663819402074,
|
4458 |
+
"grad_norm": 1.9193288087844849,
|
4459 |
+
"learning_rate": 0.00011724590163934426,
|
4460 |
+
"loss": 0.0512,
|
4461 |
+
"step": 12720
|
4462 |
+
},
|
4463 |
+
{
|
4464 |
+
"epoch": 62.184258694325806,
|
4465 |
+
"grad_norm": 1.326977252960205,
|
4466 |
+
"learning_rate": 0.00011711475409836066,
|
4467 |
+
"loss": 0.0602,
|
4468 |
+
"step": 12740
|
4469 |
+
},
|
4470 |
+
{
|
4471 |
+
"epoch": 62.281879194630875,
|
4472 |
+
"grad_norm": 1.6590323448181152,
|
4473 |
+
"learning_rate": 0.00011698360655737705,
|
4474 |
+
"loss": 0.0564,
|
4475 |
+
"step": 12760
|
4476 |
+
},
|
4477 |
+
{
|
4478 |
+
"epoch": 62.37949969493594,
|
4479 |
+
"grad_norm": 1.5968765020370483,
|
4480 |
+
"learning_rate": 0.00011685245901639346,
|
4481 |
+
"loss": 0.0591,
|
4482 |
+
"step": 12780
|
4483 |
+
},
|
4484 |
+
{
|
4485 |
+
"epoch": 62.477120195241,
|
4486 |
+
"grad_norm": 1.674349069595337,
|
4487 |
+
"learning_rate": 0.00011672131147540983,
|
4488 |
+
"loss": 0.0603,
|
4489 |
+
"step": 12800
|
4490 |
+
},
|
4491 |
+
{
|
4492 |
+
"epoch": 62.57474069554606,
|
4493 |
+
"grad_norm": 2.2285475730895996,
|
4494 |
+
"learning_rate": 0.00011659016393442623,
|
4495 |
+
"loss": 0.0578,
|
4496 |
+
"step": 12820
|
4497 |
+
},
|
4498 |
+
{
|
4499 |
+
"epoch": 62.67236119585113,
|
4500 |
+
"grad_norm": 1.4100092649459839,
|
4501 |
+
"learning_rate": 0.00011645901639344264,
|
4502 |
+
"loss": 0.0652,
|
4503 |
+
"step": 12840
|
4504 |
+
},
|
4505 |
+
{
|
4506 |
+
"epoch": 62.769981696156194,
|
4507 |
+
"grad_norm": 2.5096209049224854,
|
4508 |
+
"learning_rate": 0.00011632786885245903,
|
4509 |
+
"loss": 0.0551,
|
4510 |
+
"step": 12860
|
4511 |
+
},
|
4512 |
+
{
|
4513 |
+
"epoch": 62.867602196461256,
|
4514 |
+
"grad_norm": 3.1331403255462646,
|
4515 |
+
"learning_rate": 0.0001161967213114754,
|
4516 |
+
"loss": 0.0662,
|
4517 |
+
"step": 12880
|
4518 |
+
},
|
4519 |
+
{
|
4520 |
+
"epoch": 62.96522269676632,
|
4521 |
+
"grad_norm": 1.8345986604690552,
|
4522 |
+
"learning_rate": 0.0001160655737704918,
|
4523 |
+
"loss": 0.0671,
|
4524 |
+
"step": 12900
|
4525 |
+
},
|
4526 |
+
{
|
4527 |
+
"epoch": 63.06284319707139,
|
4528 |
+
"grad_norm": 1.7236154079437256,
|
4529 |
+
"learning_rate": 0.00011593442622950821,
|
4530 |
+
"loss": 0.0553,
|
4531 |
+
"step": 12920
|
4532 |
+
},
|
4533 |
+
{
|
4534 |
+
"epoch": 63.16046369737645,
|
4535 |
+
"grad_norm": 1.664116621017456,
|
4536 |
+
"learning_rate": 0.00011580327868852458,
|
4537 |
+
"loss": 0.0554,
|
4538 |
+
"step": 12940
|
4539 |
+
},
|
4540 |
+
{
|
4541 |
+
"epoch": 63.25808419768151,
|
4542 |
+
"grad_norm": 1.4789968729019165,
|
4543 |
+
"learning_rate": 0.00011567213114754099,
|
4544 |
+
"loss": 0.0562,
|
4545 |
+
"step": 12960
|
4546 |
+
},
|
4547 |
+
{
|
4548 |
+
"epoch": 63.355704697986575,
|
4549 |
+
"grad_norm": 2.4718666076660156,
|
4550 |
+
"learning_rate": 0.00011554098360655739,
|
4551 |
+
"loss": 0.0625,
|
4552 |
+
"step": 12980
|
4553 |
+
},
|
4554 |
+
{
|
4555 |
+
"epoch": 63.453325198291644,
|
4556 |
+
"grad_norm": 2.611966133117676,
|
4557 |
+
"learning_rate": 0.00011540983606557378,
|
4558 |
+
"loss": 0.0564,
|
4559 |
+
"step": 13000
|
4560 |
+
},
|
4561 |
+
{
|
4562 |
+
"epoch": 63.550945698596706,
|
4563 |
+
"grad_norm": 2.4028069972991943,
|
4564 |
+
"learning_rate": 0.00011527868852459016,
|
4565 |
+
"loss": 0.0629,
|
4566 |
+
"step": 13020
|
4567 |
+
},
|
4568 |
+
{
|
4569 |
+
"epoch": 63.64856619890177,
|
4570 |
+
"grad_norm": 1.251297950744629,
|
4571 |
+
"learning_rate": 0.00011514754098360656,
|
4572 |
+
"loss": 0.0608,
|
4573 |
+
"step": 13040
|
4574 |
+
},
|
4575 |
+
{
|
4576 |
+
"epoch": 63.74618669920683,
|
4577 |
+
"grad_norm": 2.8238635063171387,
|
4578 |
+
"learning_rate": 0.00011501639344262296,
|
4579 |
+
"loss": 0.0619,
|
4580 |
+
"step": 13060
|
4581 |
+
},
|
4582 |
+
{
|
4583 |
+
"epoch": 63.8438071995119,
|
4584 |
+
"grad_norm": 2.705125570297241,
|
4585 |
+
"learning_rate": 0.00011488524590163936,
|
4586 |
+
"loss": 0.0631,
|
4587 |
+
"step": 13080
|
4588 |
+
},
|
4589 |
+
{
|
4590 |
+
"epoch": 63.94142769981696,
|
4591 |
+
"grad_norm": 1.7994593381881714,
|
4592 |
+
"learning_rate": 0.00011475409836065574,
|
4593 |
+
"loss": 0.0664,
|
4594 |
+
"step": 13100
|
4595 |
+
},
|
4596 |
+
{
|
4597 |
+
"epoch": 64.03904820012202,
|
4598 |
+
"grad_norm": 3.055631399154663,
|
4599 |
+
"learning_rate": 0.00011462295081967214,
|
4600 |
+
"loss": 0.0561,
|
4601 |
+
"step": 13120
|
4602 |
+
},
|
4603 |
+
{
|
4604 |
+
"epoch": 64.1366687004271,
|
4605 |
+
"grad_norm": 1.7646046876907349,
|
4606 |
+
"learning_rate": 0.00011449180327868853,
|
4607 |
+
"loss": 0.0542,
|
4608 |
+
"step": 13140
|
4609 |
+
},
|
4610 |
+
{
|
4611 |
+
"epoch": 64.23428920073215,
|
4612 |
+
"grad_norm": 1.33475923538208,
|
4613 |
+
"learning_rate": 0.00011436065573770491,
|
4614 |
+
"loss": 0.0535,
|
4615 |
+
"step": 13160
|
4616 |
+
},
|
4617 |
+
{
|
4618 |
+
"epoch": 64.33190970103722,
|
4619 |
+
"grad_norm": 1.4737622737884521,
|
4620 |
+
"learning_rate": 0.00011422950819672131,
|
4621 |
+
"loss": 0.0527,
|
4622 |
+
"step": 13180
|
4623 |
+
},
|
4624 |
+
{
|
4625 |
+
"epoch": 64.42953020134229,
|
4626 |
+
"grad_norm": 2.2712624073028564,
|
4627 |
+
"learning_rate": 0.00011409836065573771,
|
4628 |
+
"loss": 0.0563,
|
4629 |
+
"step": 13200
|
4630 |
+
},
|
4631 |
+
{
|
4632 |
+
"epoch": 64.52715070164734,
|
4633 |
+
"grad_norm": 1.3113006353378296,
|
4634 |
+
"learning_rate": 0.00011396721311475412,
|
4635 |
+
"loss": 0.0529,
|
4636 |
+
"step": 13220
|
4637 |
+
},
|
4638 |
+
{
|
4639 |
+
"epoch": 64.62477120195241,
|
4640 |
+
"grad_norm": 1.557981014251709,
|
4641 |
+
"learning_rate": 0.00011383606557377049,
|
4642 |
+
"loss": 0.0625,
|
4643 |
+
"step": 13240
|
4644 |
+
},
|
4645 |
+
{
|
4646 |
+
"epoch": 64.72239170225747,
|
4647 |
+
"grad_norm": 1.8564097881317139,
|
4648 |
+
"learning_rate": 0.00011370491803278688,
|
4649 |
+
"loss": 0.0638,
|
4650 |
+
"step": 13260
|
4651 |
+
},
|
4652 |
+
{
|
4653 |
+
"epoch": 64.82001220256254,
|
4654 |
+
"grad_norm": 1.741255521774292,
|
4655 |
+
"learning_rate": 0.00011357377049180329,
|
4656 |
+
"loss": 0.0591,
|
4657 |
+
"step": 13280
|
4658 |
+
},
|
4659 |
+
{
|
4660 |
+
"epoch": 64.9176327028676,
|
4661 |
+
"grad_norm": 1.85543692111969,
|
4662 |
+
"learning_rate": 0.00011344262295081969,
|
4663 |
+
"loss": 0.0628,
|
4664 |
+
"step": 13300
|
4665 |
+
},
|
4666 |
+
{
|
4667 |
+
"epoch": 65.01525320317266,
|
4668 |
+
"grad_norm": 1.9299827814102173,
|
4669 |
+
"learning_rate": 0.00011331147540983606,
|
4670 |
+
"loss": 0.0553,
|
4671 |
+
"step": 13320
|
4672 |
+
},
|
4673 |
+
{
|
4674 |
+
"epoch": 65.11287370347773,
|
4675 |
+
"grad_norm": 1.5189353227615356,
|
4676 |
+
"learning_rate": 0.00011318032786885247,
|
4677 |
+
"loss": 0.0502,
|
4678 |
+
"step": 13340
|
4679 |
+
},
|
4680 |
+
{
|
4681 |
+
"epoch": 65.2104942037828,
|
4682 |
+
"grad_norm": 1.8111690282821655,
|
4683 |
+
"learning_rate": 0.00011304918032786887,
|
4684 |
+
"loss": 0.0514,
|
4685 |
+
"step": 13360
|
4686 |
+
},
|
4687 |
+
{
|
4688 |
+
"epoch": 65.30811470408786,
|
4689 |
+
"grad_norm": 2.5628137588500977,
|
4690 |
+
"learning_rate": 0.00011291803278688525,
|
4691 |
+
"loss": 0.0479,
|
4692 |
+
"step": 13380
|
4693 |
+
},
|
4694 |
+
{
|
4695 |
+
"epoch": 65.40573520439293,
|
4696 |
+
"grad_norm": 1.4838755130767822,
|
4697 |
+
"learning_rate": 0.00011278688524590164,
|
4698 |
+
"loss": 0.0591,
|
4699 |
+
"step": 13400
|
4700 |
+
},
|
4701 |
+
{
|
4702 |
+
"epoch": 65.50335570469798,
|
4703 |
+
"grad_norm": 1.512497067451477,
|
4704 |
+
"learning_rate": 0.00011265573770491804,
|
4705 |
+
"loss": 0.0554,
|
4706 |
+
"step": 13420
|
4707 |
+
},
|
4708 |
+
{
|
4709 |
+
"epoch": 65.60097620500305,
|
4710 |
+
"grad_norm": 2.546452283859253,
|
4711 |
+
"learning_rate": 0.00011252459016393444,
|
4712 |
+
"loss": 0.0567,
|
4713 |
+
"step": 13440
|
4714 |
+
},
|
4715 |
+
{
|
4716 |
+
"epoch": 65.69859670530812,
|
4717 |
+
"grad_norm": 3.046600580215454,
|
4718 |
+
"learning_rate": 0.00011239344262295082,
|
4719 |
+
"loss": 0.0614,
|
4720 |
+
"step": 13460
|
4721 |
+
},
|
4722 |
+
{
|
4723 |
+
"epoch": 65.79621720561317,
|
4724 |
+
"grad_norm": 2.1981520652770996,
|
4725 |
+
"learning_rate": 0.00011226229508196722,
|
4726 |
+
"loss": 0.0631,
|
4727 |
+
"step": 13480
|
4728 |
+
},
|
4729 |
+
{
|
4730 |
+
"epoch": 65.89383770591824,
|
4731 |
+
"grad_norm": 2.0264618396759033,
|
4732 |
+
"learning_rate": 0.00011213114754098362,
|
4733 |
+
"loss": 0.0613,
|
4734 |
+
"step": 13500
|
4735 |
+
},
|
4736 |
+
{
|
4737 |
+
"epoch": 65.99145820622331,
|
4738 |
+
"grad_norm": 5.12288236618042,
|
4739 |
+
"learning_rate": 0.00011200000000000001,
|
4740 |
+
"loss": 0.06,
|
4741 |
+
"step": 13520
|
4742 |
+
},
|
4743 |
+
{
|
4744 |
+
"epoch": 66.08907870652837,
|
4745 |
+
"grad_norm": 1.6284067630767822,
|
4746 |
+
"learning_rate": 0.00011186885245901639,
|
4747 |
+
"loss": 0.0492,
|
4748 |
+
"step": 13540
|
4749 |
+
},
|
4750 |
+
{
|
4751 |
+
"epoch": 66.18669920683344,
|
4752 |
+
"grad_norm": 1.5839916467666626,
|
4753 |
+
"learning_rate": 0.00011173770491803279,
|
4754 |
+
"loss": 0.0522,
|
4755 |
+
"step": 13560
|
4756 |
+
},
|
4757 |
+
{
|
4758 |
+
"epoch": 66.2843197071385,
|
4759 |
+
"grad_norm": 2.381410598754883,
|
4760 |
+
"learning_rate": 0.0001116065573770492,
|
4761 |
+
"loss": 0.054,
|
4762 |
+
"step": 13580
|
4763 |
+
},
|
4764 |
+
{
|
4765 |
+
"epoch": 66.38194020744356,
|
4766 |
+
"grad_norm": 2.1883935928344727,
|
4767 |
+
"learning_rate": 0.00011147540983606557,
|
4768 |
+
"loss": 0.0539,
|
4769 |
+
"step": 13600
|
4770 |
+
},
|
4771 |
+
{
|
4772 |
+
"epoch": 66.47956070774863,
|
4773 |
+
"grad_norm": 1.3095104694366455,
|
4774 |
+
"learning_rate": 0.00011134426229508197,
|
4775 |
+
"loss": 0.0548,
|
4776 |
+
"step": 13620
|
4777 |
+
},
|
4778 |
+
{
|
4779 |
+
"epoch": 66.57718120805369,
|
4780 |
+
"grad_norm": 3.0383543968200684,
|
4781 |
+
"learning_rate": 0.00011121311475409838,
|
4782 |
+
"loss": 0.0559,
|
4783 |
+
"step": 13640
|
4784 |
+
},
|
4785 |
+
{
|
4786 |
+
"epoch": 66.67480170835876,
|
4787 |
+
"grad_norm": 1.848026990890503,
|
4788 |
+
"learning_rate": 0.00011108196721311476,
|
4789 |
+
"loss": 0.0582,
|
4790 |
+
"step": 13660
|
4791 |
+
},
|
4792 |
+
{
|
4793 |
+
"epoch": 66.77242220866383,
|
4794 |
+
"grad_norm": 3.3856444358825684,
|
4795 |
+
"learning_rate": 0.00011095081967213114,
|
4796 |
+
"loss": 0.0583,
|
4797 |
+
"step": 13680
|
4798 |
+
},
|
4799 |
+
{
|
4800 |
+
"epoch": 66.87004270896888,
|
4801 |
+
"grad_norm": 2.1376798152923584,
|
4802 |
+
"learning_rate": 0.00011081967213114754,
|
4803 |
+
"loss": 0.0542,
|
4804 |
+
"step": 13700
|
4805 |
+
},
|
4806 |
+
{
|
4807 |
+
"epoch": 66.96766320927395,
|
4808 |
+
"grad_norm": 1.6410188674926758,
|
4809 |
+
"learning_rate": 0.00011068852459016395,
|
4810 |
+
"loss": 0.0648,
|
4811 |
+
"step": 13720
|
4812 |
+
},
|
4813 |
+
{
|
4814 |
+
"epoch": 67.065283709579,
|
4815 |
+
"grad_norm": 2.1899428367614746,
|
4816 |
+
"learning_rate": 0.00011055737704918035,
|
4817 |
+
"loss": 0.0518,
|
4818 |
+
"step": 13740
|
4819 |
+
},
|
4820 |
+
{
|
4821 |
+
"epoch": 67.16290420988408,
|
4822 |
+
"grad_norm": 1.662888526916504,
|
4823 |
+
"learning_rate": 0.00011042622950819672,
|
4824 |
+
"loss": 0.0481,
|
4825 |
+
"step": 13760
|
4826 |
+
},
|
4827 |
+
{
|
4828 |
+
"epoch": 67.26052471018915,
|
4829 |
+
"grad_norm": 1.8363763093948364,
|
4830 |
+
"learning_rate": 0.00011029508196721311,
|
4831 |
+
"loss": 0.0521,
|
4832 |
+
"step": 13780
|
4833 |
+
},
|
4834 |
+
{
|
4835 |
+
"epoch": 67.3581452104942,
|
4836 |
+
"grad_norm": 1.5296106338500977,
|
4837 |
+
"learning_rate": 0.00011016393442622952,
|
4838 |
+
"loss": 0.0569,
|
4839 |
+
"step": 13800
|
4840 |
+
},
|
4841 |
+
{
|
4842 |
+
"epoch": 67.45576571079927,
|
4843 |
+
"grad_norm": 3.030101776123047,
|
4844 |
+
"learning_rate": 0.00011003278688524589,
|
4845 |
+
"loss": 0.0542,
|
4846 |
+
"step": 13820
|
4847 |
+
},
|
4848 |
+
{
|
4849 |
+
"epoch": 67.55338621110434,
|
4850 |
+
"grad_norm": 1.4792683124542236,
|
4851 |
+
"learning_rate": 0.0001099016393442623,
|
4852 |
+
"loss": 0.0571,
|
4853 |
+
"step": 13840
|
4854 |
+
},
|
4855 |
+
{
|
4856 |
+
"epoch": 67.6510067114094,
|
4857 |
+
"grad_norm": 1.8390675783157349,
|
4858 |
+
"learning_rate": 0.0001097704918032787,
|
4859 |
+
"loss": 0.0562,
|
4860 |
+
"step": 13860
|
4861 |
+
},
|
4862 |
+
{
|
4863 |
+
"epoch": 67.74862721171446,
|
4864 |
+
"grad_norm": 1.6094483137130737,
|
4865 |
+
"learning_rate": 0.0001096393442622951,
|
4866 |
+
"loss": 0.0529,
|
4867 |
+
"step": 13880
|
4868 |
+
},
|
4869 |
+
{
|
4870 |
+
"epoch": 67.84624771201952,
|
4871 |
+
"grad_norm": 2.030127763748169,
|
4872 |
+
"learning_rate": 0.00010950819672131148,
|
4873 |
+
"loss": 0.0572,
|
4874 |
+
"step": 13900
|
4875 |
+
},
|
4876 |
+
{
|
4877 |
+
"epoch": 67.94386821232459,
|
4878 |
+
"grad_norm": 1.7216037511825562,
|
4879 |
+
"learning_rate": 0.00010937704918032787,
|
4880 |
+
"loss": 0.0618,
|
4881 |
+
"step": 13920
|
4882 |
+
},
|
4883 |
+
{
|
4884 |
+
"epoch": 68.04148871262966,
|
4885 |
+
"grad_norm": 2.081637382507324,
|
4886 |
+
"learning_rate": 0.00010924590163934427,
|
4887 |
+
"loss": 0.0524,
|
4888 |
+
"step": 13940
|
4889 |
+
},
|
4890 |
+
{
|
4891 |
+
"epoch": 68.13910921293471,
|
4892 |
+
"grad_norm": 1.6611350774765015,
|
4893 |
+
"learning_rate": 0.00010911475409836067,
|
4894 |
+
"loss": 0.0463,
|
4895 |
+
"step": 13960
|
4896 |
+
},
|
4897 |
+
{
|
4898 |
+
"epoch": 68.23672971323978,
|
4899 |
+
"grad_norm": 1.8240830898284912,
|
4900 |
+
"learning_rate": 0.00010898360655737705,
|
4901 |
+
"loss": 0.0528,
|
4902 |
+
"step": 13980
|
4903 |
+
},
|
4904 |
+
{
|
4905 |
+
"epoch": 68.33435021354484,
|
4906 |
+
"grad_norm": 2.299060344696045,
|
4907 |
+
"learning_rate": 0.00010885245901639345,
|
4908 |
+
"loss": 0.0525,
|
4909 |
+
"step": 14000
|
4910 |
+
},
|
4911 |
+
{
|
4912 |
+
"epoch": 68.4319707138499,
|
4913 |
+
"grad_norm": 2.1000702381134033,
|
4914 |
+
"learning_rate": 0.00010872131147540985,
|
4915 |
+
"loss": 0.0534,
|
4916 |
+
"step": 14020
|
4917 |
+
},
|
4918 |
+
{
|
4919 |
+
"epoch": 68.52959121415498,
|
4920 |
+
"grad_norm": 1.4629288911819458,
|
4921 |
+
"learning_rate": 0.00010859016393442623,
|
4922 |
+
"loss": 0.05,
|
4923 |
+
"step": 14040
|
4924 |
+
},
|
4925 |
+
{
|
4926 |
+
"epoch": 68.62721171446003,
|
4927 |
+
"grad_norm": 1.571041464805603,
|
4928 |
+
"learning_rate": 0.00010845901639344262,
|
4929 |
+
"loss": 0.0535,
|
4930 |
+
"step": 14060
|
4931 |
+
},
|
4932 |
+
{
|
4933 |
+
"epoch": 68.7248322147651,
|
4934 |
+
"grad_norm": 1.6722811460494995,
|
4935 |
+
"learning_rate": 0.00010832786885245902,
|
4936 |
+
"loss": 0.0498,
|
4937 |
+
"step": 14080
|
4938 |
+
},
|
4939 |
+
{
|
4940 |
+
"epoch": 68.82245271507017,
|
4941 |
+
"grad_norm": 1.8763340711593628,
|
4942 |
+
"learning_rate": 0.00010819672131147543,
|
4943 |
+
"loss": 0.0547,
|
4944 |
+
"step": 14100
|
4945 |
+
},
|
4946 |
+
{
|
4947 |
+
"epoch": 68.92007321537523,
|
4948 |
+
"grad_norm": 2.936068534851074,
|
4949 |
+
"learning_rate": 0.0001080655737704918,
|
4950 |
+
"loss": 0.055,
|
4951 |
+
"step": 14120
|
4952 |
+
},
|
4953 |
+
{
|
4954 |
+
"epoch": 69.0176937156803,
|
4955 |
+
"grad_norm": 2.24238920211792,
|
4956 |
+
"learning_rate": 0.0001079344262295082,
|
4957 |
+
"loss": 0.059,
|
4958 |
+
"step": 14140
|
4959 |
+
},
|
4960 |
+
{
|
4961 |
+
"epoch": 69.11531421598535,
|
4962 |
+
"grad_norm": 1.4389880895614624,
|
4963 |
+
"learning_rate": 0.00010780327868852461,
|
4964 |
+
"loss": 0.0453,
|
4965 |
+
"step": 14160
|
4966 |
+
},
|
4967 |
+
{
|
4968 |
+
"epoch": 69.21293471629042,
|
4969 |
+
"grad_norm": 1.5990976095199585,
|
4970 |
+
"learning_rate": 0.00010767213114754098,
|
4971 |
+
"loss": 0.0483,
|
4972 |
+
"step": 14180
|
4973 |
+
},
|
4974 |
+
{
|
4975 |
+
"epoch": 69.31055521659549,
|
4976 |
+
"grad_norm": 1.6288248300552368,
|
4977 |
+
"learning_rate": 0.00010754098360655737,
|
4978 |
+
"loss": 0.0492,
|
4979 |
+
"step": 14200
|
4980 |
+
},
|
4981 |
+
{
|
4982 |
+
"epoch": 69.40817571690054,
|
4983 |
+
"grad_norm": 1.6476655006408691,
|
4984 |
+
"learning_rate": 0.00010740983606557378,
|
4985 |
+
"loss": 0.0485,
|
4986 |
+
"step": 14220
|
4987 |
+
},
|
4988 |
+
{
|
4989 |
+
"epoch": 69.50579621720561,
|
4990 |
+
"grad_norm": 1.6670396327972412,
|
4991 |
+
"learning_rate": 0.00010727868852459018,
|
4992 |
+
"loss": 0.0573,
|
4993 |
+
"step": 14240
|
4994 |
+
},
|
4995 |
+
{
|
4996 |
+
"epoch": 69.60341671751068,
|
4997 |
+
"grad_norm": 1.6438359022140503,
|
4998 |
+
"learning_rate": 0.00010714754098360655,
|
4999 |
+
"loss": 0.0511,
|
5000 |
+
"step": 14260
|
5001 |
+
},
|
5002 |
+
{
|
5003 |
+
"epoch": 69.70103721781574,
|
5004 |
+
"grad_norm": 2.0703489780426025,
|
5005 |
+
"learning_rate": 0.00010701639344262296,
|
5006 |
+
"loss": 0.056,
|
5007 |
+
"step": 14280
|
5008 |
+
},
|
5009 |
+
{
|
5010 |
+
"epoch": 69.79865771812081,
|
5011 |
+
"grad_norm": 1.2886505126953125,
|
5012 |
+
"learning_rate": 0.00010688524590163935,
|
5013 |
+
"loss": 0.0518,
|
5014 |
+
"step": 14300
|
5015 |
+
},
|
5016 |
+
{
|
5017 |
+
"epoch": 69.89627821842586,
|
5018 |
+
"grad_norm": 1.869315505027771,
|
5019 |
+
"learning_rate": 0.00010675409836065575,
|
5020 |
+
"loss": 0.0613,
|
5021 |
+
"step": 14320
|
5022 |
+
},
|
5023 |
+
{
|
5024 |
+
"epoch": 69.99389871873093,
|
5025 |
+
"grad_norm": 1.5631581544876099,
|
5026 |
+
"learning_rate": 0.00010662295081967212,
|
5027 |
+
"loss": 0.0501,
|
5028 |
+
"step": 14340
|
5029 |
+
},
|
5030 |
+
{
|
5031 |
+
"epoch": 70.091519219036,
|
5032 |
+
"grad_norm": 2.5238254070281982,
|
5033 |
+
"learning_rate": 0.00010649180327868853,
|
5034 |
+
"loss": 0.0429,
|
5035 |
+
"step": 14360
|
5036 |
+
},
|
5037 |
+
{
|
5038 |
+
"epoch": 70.18913971934106,
|
5039 |
+
"grad_norm": 1.3227728605270386,
|
5040 |
+
"learning_rate": 0.00010636065573770493,
|
5041 |
+
"loss": 0.0474,
|
5042 |
+
"step": 14380
|
5043 |
+
},
|
5044 |
+
{
|
5045 |
+
"epoch": 70.28676021964613,
|
5046 |
+
"grad_norm": 2.284006118774414,
|
5047 |
+
"learning_rate": 0.0001062295081967213,
|
5048 |
+
"loss": 0.0513,
|
5049 |
+
"step": 14400
|
5050 |
+
},
|
5051 |
+
{
|
5052 |
+
"epoch": 70.3843807199512,
|
5053 |
+
"grad_norm": 1.3161708116531372,
|
5054 |
+
"learning_rate": 0.00010609836065573771,
|
5055 |
+
"loss": 0.0517,
|
5056 |
+
"step": 14420
|
5057 |
+
},
|
5058 |
+
{
|
5059 |
+
"epoch": 70.48200122025625,
|
5060 |
+
"grad_norm": 1.7371370792388916,
|
5061 |
+
"learning_rate": 0.0001059672131147541,
|
5062 |
+
"loss": 0.0494,
|
5063 |
+
"step": 14440
|
5064 |
+
},
|
5065 |
+
{
|
5066 |
+
"epoch": 70.57962172056132,
|
5067 |
+
"grad_norm": 3.2471296787261963,
|
5068 |
+
"learning_rate": 0.0001058360655737705,
|
5069 |
+
"loss": 0.0501,
|
5070 |
+
"step": 14460
|
5071 |
+
},
|
5072 |
+
{
|
5073 |
+
"epoch": 70.67724222086638,
|
5074 |
+
"grad_norm": 1.9515056610107422,
|
5075 |
+
"learning_rate": 0.00010570491803278688,
|
5076 |
+
"loss": 0.0562,
|
5077 |
+
"step": 14480
|
5078 |
+
},
|
5079 |
+
{
|
5080 |
+
"epoch": 70.77486272117144,
|
5081 |
+
"grad_norm": 1.9817485809326172,
|
5082 |
+
"learning_rate": 0.00010557377049180328,
|
5083 |
+
"loss": 0.0555,
|
5084 |
+
"step": 14500
|
5085 |
+
},
|
5086 |
+
{
|
5087 |
+
"epoch": 70.87248322147651,
|
5088 |
+
"grad_norm": 1.4870388507843018,
|
5089 |
+
"learning_rate": 0.00010544262295081968,
|
5090 |
+
"loss": 0.0529,
|
5091 |
+
"step": 14520
|
5092 |
+
},
|
5093 |
+
{
|
5094 |
+
"epoch": 70.97010372178157,
|
5095 |
+
"grad_norm": 1.6690961122512817,
|
5096 |
+
"learning_rate": 0.00010531147540983609,
|
5097 |
+
"loss": 0.0525,
|
5098 |
+
"step": 14540
|
5099 |
+
},
|
5100 |
+
{
|
5101 |
+
"epoch": 71.06772422208664,
|
5102 |
+
"grad_norm": 1.4471197128295898,
|
5103 |
+
"learning_rate": 0.00010518032786885246,
|
5104 |
+
"loss": 0.0473,
|
5105 |
+
"step": 14560
|
5106 |
+
},
|
5107 |
+
{
|
5108 |
+
"epoch": 71.16534472239171,
|
5109 |
+
"grad_norm": 2.1219794750213623,
|
5110 |
+
"learning_rate": 0.00010504918032786885,
|
5111 |
+
"loss": 0.044,
|
5112 |
+
"step": 14580
|
5113 |
+
},
|
5114 |
+
{
|
5115 |
+
"epoch": 71.26296522269676,
|
5116 |
+
"grad_norm": 2.1904103755950928,
|
5117 |
+
"learning_rate": 0.00010491803278688525,
|
5118 |
+
"loss": 0.0455,
|
5119 |
+
"step": 14600
|
5120 |
+
},
|
5121 |
+
{
|
5122 |
+
"epoch": 71.36058572300183,
|
5123 |
+
"grad_norm": 2.576287269592285,
|
5124 |
+
"learning_rate": 0.00010478688524590163,
|
5125 |
+
"loss": 0.047,
|
5126 |
+
"step": 14620
|
5127 |
+
},
|
5128 |
+
{
|
5129 |
+
"epoch": 71.45820622330689,
|
5130 |
+
"grad_norm": 2.6635916233062744,
|
5131 |
+
"learning_rate": 0.00010465573770491803,
|
5132 |
+
"loss": 0.0461,
|
5133 |
+
"step": 14640
|
5134 |
+
},
|
5135 |
+
{
|
5136 |
+
"epoch": 71.55582672361196,
|
5137 |
+
"grad_norm": 1.234979510307312,
|
5138 |
+
"learning_rate": 0.00010452459016393444,
|
5139 |
+
"loss": 0.0464,
|
5140 |
+
"step": 14660
|
5141 |
+
},
|
5142 |
+
{
|
5143 |
+
"epoch": 71.65344722391703,
|
5144 |
+
"grad_norm": 1.6037148237228394,
|
5145 |
+
"learning_rate": 0.00010439344262295083,
|
5146 |
+
"loss": 0.0522,
|
5147 |
+
"step": 14680
|
5148 |
+
},
|
5149 |
+
{
|
5150 |
+
"epoch": 71.75106772422208,
|
5151 |
+
"grad_norm": 1.990586280822754,
|
5152 |
+
"learning_rate": 0.00010426229508196721,
|
5153 |
+
"loss": 0.0577,
|
5154 |
+
"step": 14700
|
5155 |
+
},
|
5156 |
+
{
|
5157 |
+
"epoch": 71.84868822452715,
|
5158 |
+
"grad_norm": 1.5060235261917114,
|
5159 |
+
"learning_rate": 0.0001041311475409836,
|
5160 |
+
"loss": 0.0582,
|
5161 |
+
"step": 14720
|
5162 |
+
},
|
5163 |
+
{
|
5164 |
+
"epoch": 71.94630872483222,
|
5165 |
+
"grad_norm": 2.0390243530273438,
|
5166 |
+
"learning_rate": 0.00010400000000000001,
|
5167 |
+
"loss": 0.0559,
|
5168 |
+
"step": 14740
|
5169 |
+
},
|
5170 |
+
{
|
5171 |
+
"epoch": 72.04392922513728,
|
5172 |
+
"grad_norm": 1.5087133646011353,
|
5173 |
+
"learning_rate": 0.00010386885245901641,
|
5174 |
+
"loss": 0.0538,
|
5175 |
+
"step": 14760
|
5176 |
+
},
|
5177 |
+
{
|
5178 |
+
"epoch": 72.14154972544235,
|
5179 |
+
"grad_norm": 2.0309317111968994,
|
5180 |
+
"learning_rate": 0.00010373770491803279,
|
5181 |
+
"loss": 0.0455,
|
5182 |
+
"step": 14780
|
5183 |
+
},
|
5184 |
+
{
|
5185 |
+
"epoch": 72.2391702257474,
|
5186 |
+
"grad_norm": 1.643815279006958,
|
5187 |
+
"learning_rate": 0.00010360655737704919,
|
5188 |
+
"loss": 0.0449,
|
5189 |
+
"step": 14800
|
5190 |
+
},
|
5191 |
+
{
|
5192 |
+
"epoch": 72.33679072605247,
|
5193 |
+
"grad_norm": 1.881631851196289,
|
5194 |
+
"learning_rate": 0.00010347540983606558,
|
5195 |
+
"loss": 0.0473,
|
5196 |
+
"step": 14820
|
5197 |
+
},
|
5198 |
+
{
|
5199 |
+
"epoch": 72.43441122635754,
|
5200 |
+
"grad_norm": 1.6459357738494873,
|
5201 |
+
"learning_rate": 0.00010334426229508197,
|
5202 |
+
"loss": 0.048,
|
5203 |
+
"step": 14840
|
5204 |
+
},
|
5205 |
+
{
|
5206 |
+
"epoch": 72.5320317266626,
|
5207 |
+
"grad_norm": 2.058401346206665,
|
5208 |
+
"learning_rate": 0.00010321311475409836,
|
5209 |
+
"loss": 0.0435,
|
5210 |
+
"step": 14860
|
5211 |
+
},
|
5212 |
+
{
|
5213 |
+
"epoch": 72.62965222696766,
|
5214 |
+
"grad_norm": 1.428796410560608,
|
5215 |
+
"learning_rate": 0.00010308196721311476,
|
5216 |
+
"loss": 0.0501,
|
5217 |
+
"step": 14880
|
5218 |
+
},
|
5219 |
+
{
|
5220 |
+
"epoch": 72.72727272727273,
|
5221 |
+
"grad_norm": 1.7262030839920044,
|
5222 |
+
"learning_rate": 0.00010295081967213116,
|
5223 |
+
"loss": 0.0512,
|
5224 |
+
"step": 14900
|
5225 |
+
},
|
5226 |
+
{
|
5227 |
+
"epoch": 72.82489322757779,
|
5228 |
+
"grad_norm": 1.5179914236068726,
|
5229 |
+
"learning_rate": 0.00010281967213114754,
|
5230 |
+
"loss": 0.0524,
|
5231 |
+
"step": 14920
|
5232 |
+
},
|
5233 |
+
{
|
5234 |
+
"epoch": 72.92251372788286,
|
5235 |
+
"grad_norm": 1.5609859228134155,
|
5236 |
+
"learning_rate": 0.00010268852459016394,
|
5237 |
+
"loss": 0.0547,
|
5238 |
+
"step": 14940
|
5239 |
+
},
|
5240 |
+
{
|
5241 |
+
"epoch": 73.02013422818791,
|
5242 |
+
"grad_norm": 1.1576975584030151,
|
5243 |
+
"learning_rate": 0.00010255737704918033,
|
5244 |
+
"loss": 0.0506,
|
5245 |
+
"step": 14960
|
5246 |
+
},
|
5247 |
+
{
|
5248 |
+
"epoch": 73.11775472849298,
|
5249 |
+
"grad_norm": 1.4885412454605103,
|
5250 |
+
"learning_rate": 0.00010242622950819673,
|
5251 |
+
"loss": 0.0424,
|
5252 |
+
"step": 14980
|
5253 |
+
},
|
5254 |
+
{
|
5255 |
+
"epoch": 73.21537522879805,
|
5256 |
+
"grad_norm": 1.0229533910751343,
|
5257 |
+
"learning_rate": 0.00010229508196721311,
|
5258 |
+
"loss": 0.0448,
|
5259 |
+
"step": 15000
|
5260 |
}
|
5261 |
],
|
5262 |
"logging_steps": 20,
|
|
|
5276 |
"attributes": {}
|
5277 |
}
|
5278 |
},
|
5279 |
+
"total_flos": 1.6437155588267213e+17,
|
5280 |
"train_batch_size": 1,
|
5281 |
"trial_name": null,
|
5282 |
"trial_params": null
|