Commit
•
7cffce7
1
Parent(s):
c44e918
Upload folder using huggingface_hub
Browse files- data/model.safetensors +1 -1
- data/optimizer.pt +1 -1
- data/rng_state.pth +1 -1
- data/scheduler.pt +1 -1
- data/trainer_state.json +2334 -3
data/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 576008736
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:02a118a503a23e518774fed3f4dec339d8da14145ceac0e9fffee2ae001ad0f2
|
| 3 |
size 576008736
|
data/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1152256984
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:95a066061ad59333db8588202a134c688a5fa8bf89d8941dea7651093313aa50
|
| 3 |
size 1152256984
|
data/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 sha256:0aea246e6f4af30d76fc739911915f34c10cbfe67d00ea529a3053e734d22782
|
| 3 |
size 14244
|
data/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 sha256:485d39d405243b01acdfe502b2e8c0eb6d4698b82fde169faed5a021088dcbc6
|
| 3 |
size 1064
|
data/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 0.
|
| 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,
|
|
@@ -14007,6 +14007,2337 @@
|
|
| 14007 |
"learning_rate": 2.974807414080627e-05,
|
| 14008 |
"loss": 1.9742,
|
| 14009 |
"step": 30000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 14010 |
}
|
| 14011 |
],
|
| 14012 |
"logging_steps": 15,
|
|
@@ -14026,7 +16357,7 @@
|
|
| 14026 |
"attributes": {}
|
| 14027 |
}
|
| 14028 |
},
|
| 14029 |
-
"total_flos":
|
| 14030 |
"train_batch_size": 4,
|
| 14031 |
"trial_name": null,
|
| 14032 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.06815729925728017,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 35000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 14007 |
"learning_rate": 2.974807414080627e-05,
|
| 14008 |
"loss": 1.9742,
|
| 14009 |
"step": 30000
|
| 14010 |
+
},
|
| 14011 |
+
{
|
| 14012 |
+
"epoch": 0.05844975249163613,
|
| 14013 |
+
"grad_norm": 4.261213302612305,
|
| 14014 |
+
"learning_rate": 2.9747822860414652e-05,
|
| 14015 |
+
"loss": 1.747,
|
| 14016 |
+
"step": 30015
|
| 14017 |
+
},
|
| 14018 |
+
{
|
| 14019 |
+
"epoch": 0.05847896276274639,
|
| 14020 |
+
"grad_norm": 2.775156259536743,
|
| 14021 |
+
"learning_rate": 2.9747571455829544e-05,
|
| 14022 |
+
"loss": 1.8235,
|
| 14023 |
+
"step": 30030
|
| 14024 |
+
},
|
| 14025 |
+
{
|
| 14026 |
+
"epoch": 0.05850817303385665,
|
| 14027 |
+
"grad_norm": 4.271034240722656,
|
| 14028 |
+
"learning_rate": 2.9747319927053066e-05,
|
| 14029 |
+
"loss": 1.834,
|
| 14030 |
+
"step": 30045
|
| 14031 |
+
},
|
| 14032 |
+
{
|
| 14033 |
+
"epoch": 0.05853738330496691,
|
| 14034 |
+
"grad_norm": 3.628793239593506,
|
| 14035 |
+
"learning_rate": 2.9747068274087327e-05,
|
| 14036 |
+
"loss": 2.0451,
|
| 14037 |
+
"step": 30060
|
| 14038 |
+
},
|
| 14039 |
+
{
|
| 14040 |
+
"epoch": 0.05856659357607718,
|
| 14041 |
+
"grad_norm": 2.388171434402466,
|
| 14042 |
+
"learning_rate": 2.9746816496934452e-05,
|
| 14043 |
+
"loss": 1.9012,
|
| 14044 |
+
"step": 30075
|
| 14045 |
+
},
|
| 14046 |
+
{
|
| 14047 |
+
"epoch": 0.05859580384718744,
|
| 14048 |
+
"grad_norm": 5.260224342346191,
|
| 14049 |
+
"learning_rate": 2.9746564595596566e-05,
|
| 14050 |
+
"loss": 1.8394,
|
| 14051 |
+
"step": 30090
|
| 14052 |
+
},
|
| 14053 |
+
{
|
| 14054 |
+
"epoch": 0.0586250141182977,
|
| 14055 |
+
"grad_norm": 3.47929310798645,
|
| 14056 |
+
"learning_rate": 2.9746312570075786e-05,
|
| 14057 |
+
"loss": 1.7858,
|
| 14058 |
+
"step": 30105
|
| 14059 |
+
},
|
| 14060 |
+
{
|
| 14061 |
+
"epoch": 0.05865422438940797,
|
| 14062 |
+
"grad_norm": 2.082864761352539,
|
| 14063 |
+
"learning_rate": 2.974606042037423e-05,
|
| 14064 |
+
"loss": 1.8976,
|
| 14065 |
+
"step": 30120
|
| 14066 |
+
},
|
| 14067 |
+
{
|
| 14068 |
+
"epoch": 0.05868343466051823,
|
| 14069 |
+
"grad_norm": 2.003638982772827,
|
| 14070 |
+
"learning_rate": 2.9745808146494027e-05,
|
| 14071 |
+
"loss": 1.7789,
|
| 14072 |
+
"step": 30135
|
| 14073 |
+
},
|
| 14074 |
+
{
|
| 14075 |
+
"epoch": 0.05871264493162849,
|
| 14076 |
+
"grad_norm": 3.8583061695098877,
|
| 14077 |
+
"learning_rate": 2.9745555748437298e-05,
|
| 14078 |
+
"loss": 1.9207,
|
| 14079 |
+
"step": 30150
|
| 14080 |
+
},
|
| 14081 |
+
{
|
| 14082 |
+
"epoch": 0.05874185520273875,
|
| 14083 |
+
"grad_norm": 4.0564703941345215,
|
| 14084 |
+
"learning_rate": 2.9745303226206172e-05,
|
| 14085 |
+
"loss": 1.9422,
|
| 14086 |
+
"step": 30165
|
| 14087 |
+
},
|
| 14088 |
+
{
|
| 14089 |
+
"epoch": 0.05877106547384902,
|
| 14090 |
+
"grad_norm": 1.928350806236267,
|
| 14091 |
+
"learning_rate": 2.974505057980277e-05,
|
| 14092 |
+
"loss": 1.8624,
|
| 14093 |
+
"step": 30180
|
| 14094 |
+
},
|
| 14095 |
+
{
|
| 14096 |
+
"epoch": 0.05880027574495928,
|
| 14097 |
+
"grad_norm": 2.1937801837921143,
|
| 14098 |
+
"learning_rate": 2.974479780922923e-05,
|
| 14099 |
+
"loss": 1.9617,
|
| 14100 |
+
"step": 30195
|
| 14101 |
+
},
|
| 14102 |
+
{
|
| 14103 |
+
"epoch": 0.05882948601606954,
|
| 14104 |
+
"grad_norm": 4.145959854125977,
|
| 14105 |
+
"learning_rate": 2.974454491448767e-05,
|
| 14106 |
+
"loss": 1.8212,
|
| 14107 |
+
"step": 30210
|
| 14108 |
+
},
|
| 14109 |
+
{
|
| 14110 |
+
"epoch": 0.05885869628717981,
|
| 14111 |
+
"grad_norm": 4.780306816101074,
|
| 14112 |
+
"learning_rate": 2.9744291895580222e-05,
|
| 14113 |
+
"loss": 1.7573,
|
| 14114 |
+
"step": 30225
|
| 14115 |
+
},
|
| 14116 |
+
{
|
| 14117 |
+
"epoch": 0.05888790655829007,
|
| 14118 |
+
"grad_norm": 3.6125333309173584,
|
| 14119 |
+
"learning_rate": 2.9744038752509016e-05,
|
| 14120 |
+
"loss": 1.9406,
|
| 14121 |
+
"step": 30240
|
| 14122 |
+
},
|
| 14123 |
+
{
|
| 14124 |
+
"epoch": 0.05891711682940033,
|
| 14125 |
+
"grad_norm": 2.673916816711426,
|
| 14126 |
+
"learning_rate": 2.974378548527619e-05,
|
| 14127 |
+
"loss": 1.8403,
|
| 14128 |
+
"step": 30255
|
| 14129 |
+
},
|
| 14130 |
+
{
|
| 14131 |
+
"epoch": 0.05894632710051059,
|
| 14132 |
+
"grad_norm": 2.3269338607788086,
|
| 14133 |
+
"learning_rate": 2.9743532093883877e-05,
|
| 14134 |
+
"loss": 2.1496,
|
| 14135 |
+
"step": 30270
|
| 14136 |
+
},
|
| 14137 |
+
{
|
| 14138 |
+
"epoch": 0.05897553737162086,
|
| 14139 |
+
"grad_norm": 4.606401443481445,
|
| 14140 |
+
"learning_rate": 2.9743278578334205e-05,
|
| 14141 |
+
"loss": 1.7679,
|
| 14142 |
+
"step": 30285
|
| 14143 |
+
},
|
| 14144 |
+
{
|
| 14145 |
+
"epoch": 0.05900474764273112,
|
| 14146 |
+
"grad_norm": 4.217224597930908,
|
| 14147 |
+
"learning_rate": 2.9743024938629303e-05,
|
| 14148 |
+
"loss": 1.8091,
|
| 14149 |
+
"step": 30300
|
| 14150 |
+
},
|
| 14151 |
+
{
|
| 14152 |
+
"epoch": 0.05903395791384138,
|
| 14153 |
+
"grad_norm": 2.7839462757110596,
|
| 14154 |
+
"learning_rate": 2.974277117477132e-05,
|
| 14155 |
+
"loss": 1.9216,
|
| 14156 |
+
"step": 30315
|
| 14157 |
+
},
|
| 14158 |
+
{
|
| 14159 |
+
"epoch": 0.05906316818495165,
|
| 14160 |
+
"grad_norm": 2.616410970687866,
|
| 14161 |
+
"learning_rate": 2.9742517286762386e-05,
|
| 14162 |
+
"loss": 1.7535,
|
| 14163 |
+
"step": 30330
|
| 14164 |
+
},
|
| 14165 |
+
{
|
| 14166 |
+
"epoch": 0.05909237845606191,
|
| 14167 |
+
"grad_norm": 2.646209239959717,
|
| 14168 |
+
"learning_rate": 2.974226327460464e-05,
|
| 14169 |
+
"loss": 1.7763,
|
| 14170 |
+
"step": 30345
|
| 14171 |
+
},
|
| 14172 |
+
{
|
| 14173 |
+
"epoch": 0.05912158872717217,
|
| 14174 |
+
"grad_norm": 4.532891750335693,
|
| 14175 |
+
"learning_rate": 2.9742009138300225e-05,
|
| 14176 |
+
"loss": 1.9677,
|
| 14177 |
+
"step": 30360
|
| 14178 |
+
},
|
| 14179 |
+
{
|
| 14180 |
+
"epoch": 0.05915079899828243,
|
| 14181 |
+
"grad_norm": 3.891831874847412,
|
| 14182 |
+
"learning_rate": 2.9741754877851277e-05,
|
| 14183 |
+
"loss": 1.8411,
|
| 14184 |
+
"step": 30375
|
| 14185 |
+
},
|
| 14186 |
+
{
|
| 14187 |
+
"epoch": 0.0591800092693927,
|
| 14188 |
+
"grad_norm": 2.234100580215454,
|
| 14189 |
+
"learning_rate": 2.9741500493259933e-05,
|
| 14190 |
+
"loss": 1.77,
|
| 14191 |
+
"step": 30390
|
| 14192 |
+
},
|
| 14193 |
+
{
|
| 14194 |
+
"epoch": 0.05920921954050296,
|
| 14195 |
+
"grad_norm": 3.4290971755981445,
|
| 14196 |
+
"learning_rate": 2.9741245984528342e-05,
|
| 14197 |
+
"loss": 1.9234,
|
| 14198 |
+
"step": 30405
|
| 14199 |
+
},
|
| 14200 |
+
{
|
| 14201 |
+
"epoch": 0.05923842981161322,
|
| 14202 |
+
"grad_norm": 2.175567150115967,
|
| 14203 |
+
"learning_rate": 2.974099135165865e-05,
|
| 14204 |
+
"loss": 2.1054,
|
| 14205 |
+
"step": 30420
|
| 14206 |
+
},
|
| 14207 |
+
{
|
| 14208 |
+
"epoch": 0.05926764008272349,
|
| 14209 |
+
"grad_norm": 2.412052631378174,
|
| 14210 |
+
"learning_rate": 2.9740736594652992e-05,
|
| 14211 |
+
"loss": 1.8607,
|
| 14212 |
+
"step": 30435
|
| 14213 |
+
},
|
| 14214 |
+
{
|
| 14215 |
+
"epoch": 0.05929685035383375,
|
| 14216 |
+
"grad_norm": 2.27789044380188,
|
| 14217 |
+
"learning_rate": 2.974048171351352e-05,
|
| 14218 |
+
"loss": 1.9491,
|
| 14219 |
+
"step": 30450
|
| 14220 |
+
},
|
| 14221 |
+
{
|
| 14222 |
+
"epoch": 0.05932606062494401,
|
| 14223 |
+
"grad_norm": 3.9195199012756348,
|
| 14224 |
+
"learning_rate": 2.9740226708242375e-05,
|
| 14225 |
+
"loss": 1.8516,
|
| 14226 |
+
"step": 30465
|
| 14227 |
+
},
|
| 14228 |
+
{
|
| 14229 |
+
"epoch": 0.059355270896054274,
|
| 14230 |
+
"grad_norm": 2.3743503093719482,
|
| 14231 |
+
"learning_rate": 2.973997157884171e-05,
|
| 14232 |
+
"loss": 2.1744,
|
| 14233 |
+
"step": 30480
|
| 14234 |
+
},
|
| 14235 |
+
{
|
| 14236 |
+
"epoch": 0.05938448116716454,
|
| 14237 |
+
"grad_norm": 4.110390663146973,
|
| 14238 |
+
"learning_rate": 2.9739716325313676e-05,
|
| 14239 |
+
"loss": 1.8803,
|
| 14240 |
+
"step": 30495
|
| 14241 |
+
},
|
| 14242 |
+
{
|
| 14243 |
+
"epoch": 0.0594136914382748,
|
| 14244 |
+
"grad_norm": 3.738240957260132,
|
| 14245 |
+
"learning_rate": 2.973946094766041e-05,
|
| 14246 |
+
"loss": 1.7978,
|
| 14247 |
+
"step": 30510
|
| 14248 |
+
},
|
| 14249 |
+
{
|
| 14250 |
+
"epoch": 0.059442901709385064,
|
| 14251 |
+
"grad_norm": 2.840871572494507,
|
| 14252 |
+
"learning_rate": 2.9739205445884075e-05,
|
| 14253 |
+
"loss": 1.8413,
|
| 14254 |
+
"step": 30525
|
| 14255 |
+
},
|
| 14256 |
+
{
|
| 14257 |
+
"epoch": 0.059472111980495325,
|
| 14258 |
+
"grad_norm": 2.931602954864502,
|
| 14259 |
+
"learning_rate": 2.9738949819986814e-05,
|
| 14260 |
+
"loss": 1.9035,
|
| 14261 |
+
"step": 30540
|
| 14262 |
+
},
|
| 14263 |
+
{
|
| 14264 |
+
"epoch": 0.05950132225160559,
|
| 14265 |
+
"grad_norm": 2.814918041229248,
|
| 14266 |
+
"learning_rate": 2.9738694069970787e-05,
|
| 14267 |
+
"loss": 1.9169,
|
| 14268 |
+
"step": 30555
|
| 14269 |
+
},
|
| 14270 |
+
{
|
| 14271 |
+
"epoch": 0.059530532522715854,
|
| 14272 |
+
"grad_norm": 4.824680805206299,
|
| 14273 |
+
"learning_rate": 2.9738438195838145e-05,
|
| 14274 |
+
"loss": 1.8696,
|
| 14275 |
+
"step": 30570
|
| 14276 |
+
},
|
| 14277 |
+
{
|
| 14278 |
+
"epoch": 0.059559742793826115,
|
| 14279 |
+
"grad_norm": 2.1868393421173096,
|
| 14280 |
+
"learning_rate": 2.9738182197591038e-05,
|
| 14281 |
+
"loss": 1.8737,
|
| 14282 |
+
"step": 30585
|
| 14283 |
+
},
|
| 14284 |
+
{
|
| 14285 |
+
"epoch": 0.05958895306493638,
|
| 14286 |
+
"grad_norm": 4.172234535217285,
|
| 14287 |
+
"learning_rate": 2.973792607523163e-05,
|
| 14288 |
+
"loss": 1.8148,
|
| 14289 |
+
"step": 30600
|
| 14290 |
+
},
|
| 14291 |
+
{
|
| 14292 |
+
"epoch": 0.059618163336046644,
|
| 14293 |
+
"grad_norm": 4.052221775054932,
|
| 14294 |
+
"learning_rate": 2.9737669828762074e-05,
|
| 14295 |
+
"loss": 1.8824,
|
| 14296 |
+
"step": 30615
|
| 14297 |
+
},
|
| 14298 |
+
{
|
| 14299 |
+
"epoch": 0.059647373607156905,
|
| 14300 |
+
"grad_norm": 3.6041007041931152,
|
| 14301 |
+
"learning_rate": 2.9737413458184522e-05,
|
| 14302 |
+
"loss": 1.8255,
|
| 14303 |
+
"step": 30630
|
| 14304 |
+
},
|
| 14305 |
+
{
|
| 14306 |
+
"epoch": 0.059676583878267166,
|
| 14307 |
+
"grad_norm": 4.06122350692749,
|
| 14308 |
+
"learning_rate": 2.9737156963501143e-05,
|
| 14309 |
+
"loss": 1.951,
|
| 14310 |
+
"step": 30645
|
| 14311 |
+
},
|
| 14312 |
+
{
|
| 14313 |
+
"epoch": 0.059705794149377434,
|
| 14314 |
+
"grad_norm": 2.3008453845977783,
|
| 14315 |
+
"learning_rate": 2.973690034471409e-05,
|
| 14316 |
+
"loss": 1.758,
|
| 14317 |
+
"step": 30660
|
| 14318 |
+
},
|
| 14319 |
+
{
|
| 14320 |
+
"epoch": 0.059735004420487695,
|
| 14321 |
+
"grad_norm": 3.7912003993988037,
|
| 14322 |
+
"learning_rate": 2.9736643601825532e-05,
|
| 14323 |
+
"loss": 1.8688,
|
| 14324 |
+
"step": 30675
|
| 14325 |
+
},
|
| 14326 |
+
{
|
| 14327 |
+
"epoch": 0.059764214691597956,
|
| 14328 |
+
"grad_norm": 3.7779648303985596,
|
| 14329 |
+
"learning_rate": 2.9736386734837616e-05,
|
| 14330 |
+
"loss": 1.7613,
|
| 14331 |
+
"step": 30690
|
| 14332 |
+
},
|
| 14333 |
+
{
|
| 14334 |
+
"epoch": 0.05979342496270822,
|
| 14335 |
+
"grad_norm": 4.632726669311523,
|
| 14336 |
+
"learning_rate": 2.9736129743752523e-05,
|
| 14337 |
+
"loss": 1.6212,
|
| 14338 |
+
"step": 30705
|
| 14339 |
+
},
|
| 14340 |
+
{
|
| 14341 |
+
"epoch": 0.059822635233818484,
|
| 14342 |
+
"grad_norm": 4.628555774688721,
|
| 14343 |
+
"learning_rate": 2.9735872628572404e-05,
|
| 14344 |
+
"loss": 1.8643,
|
| 14345 |
+
"step": 30720
|
| 14346 |
+
},
|
| 14347 |
+
{
|
| 14348 |
+
"epoch": 0.059851845504928745,
|
| 14349 |
+
"grad_norm": 3.8072681427001953,
|
| 14350 |
+
"learning_rate": 2.9735615389299434e-05,
|
| 14351 |
+
"loss": 1.7805,
|
| 14352 |
+
"step": 30735
|
| 14353 |
+
},
|
| 14354 |
+
{
|
| 14355 |
+
"epoch": 0.059881055776039006,
|
| 14356 |
+
"grad_norm": 3.9826717376708984,
|
| 14357 |
+
"learning_rate": 2.973535802593577e-05,
|
| 14358 |
+
"loss": 1.8594,
|
| 14359 |
+
"step": 30750
|
| 14360 |
+
},
|
| 14361 |
+
{
|
| 14362 |
+
"epoch": 0.059910266047149274,
|
| 14363 |
+
"grad_norm": 3.04962420463562,
|
| 14364 |
+
"learning_rate": 2.9735100538483587e-05,
|
| 14365 |
+
"loss": 1.9104,
|
| 14366 |
+
"step": 30765
|
| 14367 |
+
},
|
| 14368 |
+
{
|
| 14369 |
+
"epoch": 0.059939476318259535,
|
| 14370 |
+
"grad_norm": 2.8500261306762695,
|
| 14371 |
+
"learning_rate": 2.9734842926945048e-05,
|
| 14372 |
+
"loss": 1.717,
|
| 14373 |
+
"step": 30780
|
| 14374 |
+
},
|
| 14375 |
+
{
|
| 14376 |
+
"epoch": 0.059968686589369796,
|
| 14377 |
+
"grad_norm": 1.5368775129318237,
|
| 14378 |
+
"learning_rate": 2.9734585191322324e-05,
|
| 14379 |
+
"loss": 1.8905,
|
| 14380 |
+
"step": 30795
|
| 14381 |
+
},
|
| 14382 |
+
{
|
| 14383 |
+
"epoch": 0.059997896860480064,
|
| 14384 |
+
"grad_norm": 3.886584997177124,
|
| 14385 |
+
"learning_rate": 2.9734327331617588e-05,
|
| 14386 |
+
"loss": 1.804,
|
| 14387 |
+
"step": 30810
|
| 14388 |
+
},
|
| 14389 |
+
{
|
| 14390 |
+
"epoch": 0.060027107131590325,
|
| 14391 |
+
"grad_norm": 3.792663812637329,
|
| 14392 |
+
"learning_rate": 2.9734069347833005e-05,
|
| 14393 |
+
"loss": 1.9667,
|
| 14394 |
+
"step": 30825
|
| 14395 |
+
},
|
| 14396 |
+
{
|
| 14397 |
+
"epoch": 0.060056317402700586,
|
| 14398 |
+
"grad_norm": 3.7504067420959473,
|
| 14399 |
+
"learning_rate": 2.9733811239970756e-05,
|
| 14400 |
+
"loss": 1.9053,
|
| 14401 |
+
"step": 30840
|
| 14402 |
+
},
|
| 14403 |
+
{
|
| 14404 |
+
"epoch": 0.06008552767381085,
|
| 14405 |
+
"grad_norm": 2.617370367050171,
|
| 14406 |
+
"learning_rate": 2.973355300803301e-05,
|
| 14407 |
+
"loss": 1.8853,
|
| 14408 |
+
"step": 30855
|
| 14409 |
+
},
|
| 14410 |
+
{
|
| 14411 |
+
"epoch": 0.060114737944921115,
|
| 14412 |
+
"grad_norm": 3.2594456672668457,
|
| 14413 |
+
"learning_rate": 2.973329465202194e-05,
|
| 14414 |
+
"loss": 1.7045,
|
| 14415 |
+
"step": 30870
|
| 14416 |
+
},
|
| 14417 |
+
{
|
| 14418 |
+
"epoch": 0.060143948216031376,
|
| 14419 |
+
"grad_norm": 1.8166978359222412,
|
| 14420 |
+
"learning_rate": 2.9733036171939723e-05,
|
| 14421 |
+
"loss": 1.8431,
|
| 14422 |
+
"step": 30885
|
| 14423 |
+
},
|
| 14424 |
+
{
|
| 14425 |
+
"epoch": 0.06017315848714164,
|
| 14426 |
+
"grad_norm": 2.9280359745025635,
|
| 14427 |
+
"learning_rate": 2.973277756778854e-05,
|
| 14428 |
+
"loss": 1.9088,
|
| 14429 |
+
"step": 30900
|
| 14430 |
+
},
|
| 14431 |
+
{
|
| 14432 |
+
"epoch": 0.060202368758251905,
|
| 14433 |
+
"grad_norm": 4.446992874145508,
|
| 14434 |
+
"learning_rate": 2.973251883957056e-05,
|
| 14435 |
+
"loss": 1.7744,
|
| 14436 |
+
"step": 30915
|
| 14437 |
+
},
|
| 14438 |
+
{
|
| 14439 |
+
"epoch": 0.060231579029362166,
|
| 14440 |
+
"grad_norm": 4.279955863952637,
|
| 14441 |
+
"learning_rate": 2.9732259987287967e-05,
|
| 14442 |
+
"loss": 1.8309,
|
| 14443 |
+
"step": 30930
|
| 14444 |
+
},
|
| 14445 |
+
{
|
| 14446 |
+
"epoch": 0.06026078930047243,
|
| 14447 |
+
"grad_norm": 2.3727381229400635,
|
| 14448 |
+
"learning_rate": 2.9732001010942946e-05,
|
| 14449 |
+
"loss": 1.9247,
|
| 14450 |
+
"step": 30945
|
| 14451 |
+
},
|
| 14452 |
+
{
|
| 14453 |
+
"epoch": 0.06028999957158269,
|
| 14454 |
+
"grad_norm": 2.6014645099639893,
|
| 14455 |
+
"learning_rate": 2.9731741910537668e-05,
|
| 14456 |
+
"loss": 1.7653,
|
| 14457 |
+
"step": 30960
|
| 14458 |
+
},
|
| 14459 |
+
{
|
| 14460 |
+
"epoch": 0.060319209842692956,
|
| 14461 |
+
"grad_norm": 2.387563467025757,
|
| 14462 |
+
"learning_rate": 2.973148268607432e-05,
|
| 14463 |
+
"loss": 1.7814,
|
| 14464 |
+
"step": 30975
|
| 14465 |
+
},
|
| 14466 |
+
{
|
| 14467 |
+
"epoch": 0.06034842011380322,
|
| 14468 |
+
"grad_norm": 3.4683475494384766,
|
| 14469 |
+
"learning_rate": 2.9731223337555088e-05,
|
| 14470 |
+
"loss": 1.8522,
|
| 14471 |
+
"step": 30990
|
| 14472 |
+
},
|
| 14473 |
+
{
|
| 14474 |
+
"epoch": 0.06037763038491348,
|
| 14475 |
+
"grad_norm": 2.352069854736328,
|
| 14476 |
+
"learning_rate": 2.973096386498215e-05,
|
| 14477 |
+
"loss": 2.0015,
|
| 14478 |
+
"step": 31005
|
| 14479 |
+
},
|
| 14480 |
+
{
|
| 14481 |
+
"epoch": 0.06040684065602374,
|
| 14482 |
+
"grad_norm": 3.2257015705108643,
|
| 14483 |
+
"learning_rate": 2.9730704268357694e-05,
|
| 14484 |
+
"loss": 1.8176,
|
| 14485 |
+
"step": 31020
|
| 14486 |
+
},
|
| 14487 |
+
{
|
| 14488 |
+
"epoch": 0.06043605092713401,
|
| 14489 |
+
"grad_norm": 2.4939286708831787,
|
| 14490 |
+
"learning_rate": 2.97304445476839e-05,
|
| 14491 |
+
"loss": 1.7388,
|
| 14492 |
+
"step": 31035
|
| 14493 |
+
},
|
| 14494 |
+
{
|
| 14495 |
+
"epoch": 0.06046526119824427,
|
| 14496 |
+
"grad_norm": 3.216533660888672,
|
| 14497 |
+
"learning_rate": 2.9730184702962968e-05,
|
| 14498 |
+
"loss": 1.859,
|
| 14499 |
+
"step": 31050
|
| 14500 |
+
},
|
| 14501 |
+
{
|
| 14502 |
+
"epoch": 0.06049447146935453,
|
| 14503 |
+
"grad_norm": 3.4212570190429688,
|
| 14504 |
+
"learning_rate": 2.9729924734197077e-05,
|
| 14505 |
+
"loss": 1.7748,
|
| 14506 |
+
"step": 31065
|
| 14507 |
+
},
|
| 14508 |
+
{
|
| 14509 |
+
"epoch": 0.060523681740464796,
|
| 14510 |
+
"grad_norm": 3.6867172718048096,
|
| 14511 |
+
"learning_rate": 2.9729664641388417e-05,
|
| 14512 |
+
"loss": 2.0345,
|
| 14513 |
+
"step": 31080
|
| 14514 |
+
},
|
| 14515 |
+
{
|
| 14516 |
+
"epoch": 0.06055289201157506,
|
| 14517 |
+
"grad_norm": 3.243898630142212,
|
| 14518 |
+
"learning_rate": 2.9729404424539183e-05,
|
| 14519 |
+
"loss": 1.9114,
|
| 14520 |
+
"step": 31095
|
| 14521 |
+
},
|
| 14522 |
+
{
|
| 14523 |
+
"epoch": 0.06058210228268532,
|
| 14524 |
+
"grad_norm": 2.8866801261901855,
|
| 14525 |
+
"learning_rate": 2.9729144083651554e-05,
|
| 14526 |
+
"loss": 1.9833,
|
| 14527 |
+
"step": 31110
|
| 14528 |
+
},
|
| 14529 |
+
{
|
| 14530 |
+
"epoch": 0.06061131255379558,
|
| 14531 |
+
"grad_norm": 3.0051145553588867,
|
| 14532 |
+
"learning_rate": 2.9728883618727738e-05,
|
| 14533 |
+
"loss": 2.026,
|
| 14534 |
+
"step": 31125
|
| 14535 |
+
},
|
| 14536 |
+
{
|
| 14537 |
+
"epoch": 0.06064052282490585,
|
| 14538 |
+
"grad_norm": 3.0416595935821533,
|
| 14539 |
+
"learning_rate": 2.972862302976992e-05,
|
| 14540 |
+
"loss": 1.8541,
|
| 14541 |
+
"step": 31140
|
| 14542 |
+
},
|
| 14543 |
+
{
|
| 14544 |
+
"epoch": 0.06066973309601611,
|
| 14545 |
+
"grad_norm": 2.835458517074585,
|
| 14546 |
+
"learning_rate": 2.9728362316780294e-05,
|
| 14547 |
+
"loss": 1.8636,
|
| 14548 |
+
"step": 31155
|
| 14549 |
+
},
|
| 14550 |
+
{
|
| 14551 |
+
"epoch": 0.06069894336712637,
|
| 14552 |
+
"grad_norm": 2.40006947517395,
|
| 14553 |
+
"learning_rate": 2.972810147976106e-05,
|
| 14554 |
+
"loss": 1.9535,
|
| 14555 |
+
"step": 31170
|
| 14556 |
+
},
|
| 14557 |
+
{
|
| 14558 |
+
"epoch": 0.06072815363823664,
|
| 14559 |
+
"grad_norm": 2.7326056957244873,
|
| 14560 |
+
"learning_rate": 2.9727840518714415e-05,
|
| 14561 |
+
"loss": 1.7814,
|
| 14562 |
+
"step": 31185
|
| 14563 |
+
},
|
| 14564 |
+
{
|
| 14565 |
+
"epoch": 0.0607573639093469,
|
| 14566 |
+
"grad_norm": 4.403411388397217,
|
| 14567 |
+
"learning_rate": 2.972757943364255e-05,
|
| 14568 |
+
"loss": 1.7441,
|
| 14569 |
+
"step": 31200
|
| 14570 |
+
},
|
| 14571 |
+
{
|
| 14572 |
+
"epoch": 0.06078657418045716,
|
| 14573 |
+
"grad_norm": 2.259716272354126,
|
| 14574 |
+
"learning_rate": 2.9727318224547667e-05,
|
| 14575 |
+
"loss": 1.8447,
|
| 14576 |
+
"step": 31215
|
| 14577 |
+
},
|
| 14578 |
+
{
|
| 14579 |
+
"epoch": 0.06081578445156742,
|
| 14580 |
+
"grad_norm": 2.397702217102051,
|
| 14581 |
+
"learning_rate": 2.9727056891431966e-05,
|
| 14582 |
+
"loss": 1.7677,
|
| 14583 |
+
"step": 31230
|
| 14584 |
+
},
|
| 14585 |
+
{
|
| 14586 |
+
"epoch": 0.06084499472267769,
|
| 14587 |
+
"grad_norm": 3.840811014175415,
|
| 14588 |
+
"learning_rate": 2.9726795434297646e-05,
|
| 14589 |
+
"loss": 1.7271,
|
| 14590 |
+
"step": 31245
|
| 14591 |
+
},
|
| 14592 |
+
{
|
| 14593 |
+
"epoch": 0.06087420499378795,
|
| 14594 |
+
"grad_norm": 2.695345163345337,
|
| 14595 |
+
"learning_rate": 2.9726533853146913e-05,
|
| 14596 |
+
"loss": 1.9575,
|
| 14597 |
+
"step": 31260
|
| 14598 |
+
},
|
| 14599 |
+
{
|
| 14600 |
+
"epoch": 0.06090341526489821,
|
| 14601 |
+
"grad_norm": 2.330676555633545,
|
| 14602 |
+
"learning_rate": 2.972627214798197e-05,
|
| 14603 |
+
"loss": 1.9488,
|
| 14604 |
+
"step": 31275
|
| 14605 |
+
},
|
| 14606 |
+
{
|
| 14607 |
+
"epoch": 0.06093262553600848,
|
| 14608 |
+
"grad_norm": 3.987534999847412,
|
| 14609 |
+
"learning_rate": 2.9726010318805014e-05,
|
| 14610 |
+
"loss": 1.8628,
|
| 14611 |
+
"step": 31290
|
| 14612 |
+
},
|
| 14613 |
+
{
|
| 14614 |
+
"epoch": 0.06096183580711874,
|
| 14615 |
+
"grad_norm": 3.307753086090088,
|
| 14616 |
+
"learning_rate": 2.9725748365618252e-05,
|
| 14617 |
+
"loss": 1.9931,
|
| 14618 |
+
"step": 31305
|
| 14619 |
+
},
|
| 14620 |
+
{
|
| 14621 |
+
"epoch": 0.060991046078229,
|
| 14622 |
+
"grad_norm": 4.703588485717773,
|
| 14623 |
+
"learning_rate": 2.9725486288423894e-05,
|
| 14624 |
+
"loss": 1.8416,
|
| 14625 |
+
"step": 31320
|
| 14626 |
+
},
|
| 14627 |
+
{
|
| 14628 |
+
"epoch": 0.06102025634933926,
|
| 14629 |
+
"grad_norm": 2.01497220993042,
|
| 14630 |
+
"learning_rate": 2.9725224087224146e-05,
|
| 14631 |
+
"loss": 1.897,
|
| 14632 |
+
"step": 31335
|
| 14633 |
+
},
|
| 14634 |
+
{
|
| 14635 |
+
"epoch": 0.06104946662044953,
|
| 14636 |
+
"grad_norm": 2.9679431915283203,
|
| 14637 |
+
"learning_rate": 2.9724961762021215e-05,
|
| 14638 |
+
"loss": 1.951,
|
| 14639 |
+
"step": 31350
|
| 14640 |
+
},
|
| 14641 |
+
{
|
| 14642 |
+
"epoch": 0.06107867689155979,
|
| 14643 |
+
"grad_norm": 2.5453622341156006,
|
| 14644 |
+
"learning_rate": 2.972469931281731e-05,
|
| 14645 |
+
"loss": 1.8823,
|
| 14646 |
+
"step": 31365
|
| 14647 |
+
},
|
| 14648 |
+
{
|
| 14649 |
+
"epoch": 0.06110788716267005,
|
| 14650 |
+
"grad_norm": 4.135504245758057,
|
| 14651 |
+
"learning_rate": 2.9724436739614643e-05,
|
| 14652 |
+
"loss": 1.9061,
|
| 14653 |
+
"step": 31380
|
| 14654 |
+
},
|
| 14655 |
+
{
|
| 14656 |
+
"epoch": 0.06113709743378032,
|
| 14657 |
+
"grad_norm": 1.99344003200531,
|
| 14658 |
+
"learning_rate": 2.9724174042415417e-05,
|
| 14659 |
+
"loss": 1.986,
|
| 14660 |
+
"step": 31395
|
| 14661 |
+
},
|
| 14662 |
+
{
|
| 14663 |
+
"epoch": 0.06116630770489058,
|
| 14664 |
+
"grad_norm": 4.097934722900391,
|
| 14665 |
+
"learning_rate": 2.9723911221221857e-05,
|
| 14666 |
+
"loss": 1.8222,
|
| 14667 |
+
"step": 31410
|
| 14668 |
+
},
|
| 14669 |
+
{
|
| 14670 |
+
"epoch": 0.06119551797600084,
|
| 14671 |
+
"grad_norm": 4.836818695068359,
|
| 14672 |
+
"learning_rate": 2.9723648276036165e-05,
|
| 14673 |
+
"loss": 1.9038,
|
| 14674 |
+
"step": 31425
|
| 14675 |
+
},
|
| 14676 |
+
{
|
| 14677 |
+
"epoch": 0.0612247282471111,
|
| 14678 |
+
"grad_norm": 2.4198572635650635,
|
| 14679 |
+
"learning_rate": 2.972338520686056e-05,
|
| 14680 |
+
"loss": 1.9228,
|
| 14681 |
+
"step": 31440
|
| 14682 |
+
},
|
| 14683 |
+
{
|
| 14684 |
+
"epoch": 0.06125393851822137,
|
| 14685 |
+
"grad_norm": 4.546353816986084,
|
| 14686 |
+
"learning_rate": 2.9723122013697265e-05,
|
| 14687 |
+
"loss": 1.6837,
|
| 14688 |
+
"step": 31455
|
| 14689 |
+
},
|
| 14690 |
+
{
|
| 14691 |
+
"epoch": 0.06128314878933163,
|
| 14692 |
+
"grad_norm": 3.9125819206237793,
|
| 14693 |
+
"learning_rate": 2.972285869654848e-05,
|
| 14694 |
+
"loss": 1.9057,
|
| 14695 |
+
"step": 31470
|
| 14696 |
+
},
|
| 14697 |
+
{
|
| 14698 |
+
"epoch": 0.06131235906044189,
|
| 14699 |
+
"grad_norm": 2.7409276962280273,
|
| 14700 |
+
"learning_rate": 2.9722595255416435e-05,
|
| 14701 |
+
"loss": 1.6805,
|
| 14702 |
+
"step": 31485
|
| 14703 |
+
},
|
| 14704 |
+
{
|
| 14705 |
+
"epoch": 0.06134156933155216,
|
| 14706 |
+
"grad_norm": 3.4497761726379395,
|
| 14707 |
+
"learning_rate": 2.9722331690303344e-05,
|
| 14708 |
+
"loss": 1.7006,
|
| 14709 |
+
"step": 31500
|
| 14710 |
+
},
|
| 14711 |
+
{
|
| 14712 |
+
"epoch": 0.06137077960266242,
|
| 14713 |
+
"grad_norm": 3.8101327419281006,
|
| 14714 |
+
"learning_rate": 2.9722068001211427e-05,
|
| 14715 |
+
"loss": 1.8445,
|
| 14716 |
+
"step": 31515
|
| 14717 |
+
},
|
| 14718 |
+
{
|
| 14719 |
+
"epoch": 0.06139998987377268,
|
| 14720 |
+
"grad_norm": 4.296467304229736,
|
| 14721 |
+
"learning_rate": 2.9721804188142906e-05,
|
| 14722 |
+
"loss": 1.9453,
|
| 14723 |
+
"step": 31530
|
| 14724 |
+
},
|
| 14725 |
+
{
|
| 14726 |
+
"epoch": 0.06142920014488294,
|
| 14727 |
+
"grad_norm": 4.05743408203125,
|
| 14728 |
+
"learning_rate": 2.9721540251100003e-05,
|
| 14729 |
+
"loss": 1.6627,
|
| 14730 |
+
"step": 31545
|
| 14731 |
+
},
|
| 14732 |
+
{
|
| 14733 |
+
"epoch": 0.06145841041599321,
|
| 14734 |
+
"grad_norm": 1.896415114402771,
|
| 14735 |
+
"learning_rate": 2.9721276190084934e-05,
|
| 14736 |
+
"loss": 2.046,
|
| 14737 |
+
"step": 31560
|
| 14738 |
+
},
|
| 14739 |
+
{
|
| 14740 |
+
"epoch": 0.06148762068710347,
|
| 14741 |
+
"grad_norm": 3.3542966842651367,
|
| 14742 |
+
"learning_rate": 2.9721012005099933e-05,
|
| 14743 |
+
"loss": 1.7947,
|
| 14744 |
+
"step": 31575
|
| 14745 |
+
},
|
| 14746 |
+
{
|
| 14747 |
+
"epoch": 0.06151683095821373,
|
| 14748 |
+
"grad_norm": 2.018230676651001,
|
| 14749 |
+
"learning_rate": 2.9720747696147214e-05,
|
| 14750 |
+
"loss": 1.8102,
|
| 14751 |
+
"step": 31590
|
| 14752 |
+
},
|
| 14753 |
+
{
|
| 14754 |
+
"epoch": 0.06154604122932399,
|
| 14755 |
+
"grad_norm": 2.8734261989593506,
|
| 14756 |
+
"learning_rate": 2.9720483263229012e-05,
|
| 14757 |
+
"loss": 1.9102,
|
| 14758 |
+
"step": 31605
|
| 14759 |
+
},
|
| 14760 |
+
{
|
| 14761 |
+
"epoch": 0.06157525150043426,
|
| 14762 |
+
"grad_norm": 2.8697266578674316,
|
| 14763 |
+
"learning_rate": 2.972021870634755e-05,
|
| 14764 |
+
"loss": 1.9179,
|
| 14765 |
+
"step": 31620
|
| 14766 |
+
},
|
| 14767 |
+
{
|
| 14768 |
+
"epoch": 0.06160446177154452,
|
| 14769 |
+
"grad_norm": 2.5872962474823,
|
| 14770 |
+
"learning_rate": 2.9719954025505054e-05,
|
| 14771 |
+
"loss": 1.8555,
|
| 14772 |
+
"step": 31635
|
| 14773 |
+
},
|
| 14774 |
+
{
|
| 14775 |
+
"epoch": 0.06163367204265478,
|
| 14776 |
+
"grad_norm": 3.8212106227874756,
|
| 14777 |
+
"learning_rate": 2.9719689220703758e-05,
|
| 14778 |
+
"loss": 1.7516,
|
| 14779 |
+
"step": 31650
|
| 14780 |
+
},
|
| 14781 |
+
{
|
| 14782 |
+
"epoch": 0.06166288231376505,
|
| 14783 |
+
"grad_norm": 2.7884206771850586,
|
| 14784 |
+
"learning_rate": 2.9719424291945885e-05,
|
| 14785 |
+
"loss": 1.8213,
|
| 14786 |
+
"step": 31665
|
| 14787 |
+
},
|
| 14788 |
+
{
|
| 14789 |
+
"epoch": 0.06169209258487531,
|
| 14790 |
+
"grad_norm": 2.5221076011657715,
|
| 14791 |
+
"learning_rate": 2.971915923923367e-05,
|
| 14792 |
+
"loss": 1.6586,
|
| 14793 |
+
"step": 31680
|
| 14794 |
+
},
|
| 14795 |
+
{
|
| 14796 |
+
"epoch": 0.06172130285598557,
|
| 14797 |
+
"grad_norm": 2.4005963802337646,
|
| 14798 |
+
"learning_rate": 2.971889406256935e-05,
|
| 14799 |
+
"loss": 2.0582,
|
| 14800 |
+
"step": 31695
|
| 14801 |
+
},
|
| 14802 |
+
{
|
| 14803 |
+
"epoch": 0.061750513127095834,
|
| 14804 |
+
"grad_norm": 2.02565860748291,
|
| 14805 |
+
"learning_rate": 2.9718628761955146e-05,
|
| 14806 |
+
"loss": 1.8941,
|
| 14807 |
+
"step": 31710
|
| 14808 |
+
},
|
| 14809 |
+
{
|
| 14810 |
+
"epoch": 0.0617797233982061,
|
| 14811 |
+
"grad_norm": 3.2867822647094727,
|
| 14812 |
+
"learning_rate": 2.9718363337393303e-05,
|
| 14813 |
+
"loss": 1.7149,
|
| 14814 |
+
"step": 31725
|
| 14815 |
+
},
|
| 14816 |
+
{
|
| 14817 |
+
"epoch": 0.06180893366931636,
|
| 14818 |
+
"grad_norm": 2.4237873554229736,
|
| 14819 |
+
"learning_rate": 2.9718097788886054e-05,
|
| 14820 |
+
"loss": 1.8685,
|
| 14821 |
+
"step": 31740
|
| 14822 |
+
},
|
| 14823 |
+
{
|
| 14824 |
+
"epoch": 0.061838143940426624,
|
| 14825 |
+
"grad_norm": 2.000887155532837,
|
| 14826 |
+
"learning_rate": 2.9717832116435632e-05,
|
| 14827 |
+
"loss": 1.7489,
|
| 14828 |
+
"step": 31755
|
| 14829 |
+
},
|
| 14830 |
+
{
|
| 14831 |
+
"epoch": 0.06186735421153689,
|
| 14832 |
+
"grad_norm": 2.2382023334503174,
|
| 14833 |
+
"learning_rate": 2.9717566320044276e-05,
|
| 14834 |
+
"loss": 1.8845,
|
| 14835 |
+
"step": 31770
|
| 14836 |
+
},
|
| 14837 |
+
{
|
| 14838 |
+
"epoch": 0.06189656448264715,
|
| 14839 |
+
"grad_norm": 3.9158990383148193,
|
| 14840 |
+
"learning_rate": 2.9717300399714222e-05,
|
| 14841 |
+
"loss": 1.689,
|
| 14842 |
+
"step": 31785
|
| 14843 |
+
},
|
| 14844 |
+
{
|
| 14845 |
+
"epoch": 0.061925774753757414,
|
| 14846 |
+
"grad_norm": 4.004560470581055,
|
| 14847 |
+
"learning_rate": 2.9717034355447713e-05,
|
| 14848 |
+
"loss": 1.9279,
|
| 14849 |
+
"step": 31800
|
| 14850 |
+
},
|
| 14851 |
+
{
|
| 14852 |
+
"epoch": 0.061954985024867675,
|
| 14853 |
+
"grad_norm": 2.9310123920440674,
|
| 14854 |
+
"learning_rate": 2.9716768187246986e-05,
|
| 14855 |
+
"loss": 1.771,
|
| 14856 |
+
"step": 31815
|
| 14857 |
+
},
|
| 14858 |
+
{
|
| 14859 |
+
"epoch": 0.06198419529597794,
|
| 14860 |
+
"grad_norm": 2.5769150257110596,
|
| 14861 |
+
"learning_rate": 2.9716501895114287e-05,
|
| 14862 |
+
"loss": 1.9365,
|
| 14863 |
+
"step": 31830
|
| 14864 |
+
},
|
| 14865 |
+
{
|
| 14866 |
+
"epoch": 0.0620134055670882,
|
| 14867 |
+
"grad_norm": 3.767500162124634,
|
| 14868 |
+
"learning_rate": 2.9716235479051858e-05,
|
| 14869 |
+
"loss": 1.8358,
|
| 14870 |
+
"step": 31845
|
| 14871 |
+
},
|
| 14872 |
+
{
|
| 14873 |
+
"epoch": 0.062042615838198464,
|
| 14874 |
+
"grad_norm": 4.410051345825195,
|
| 14875 |
+
"learning_rate": 2.9715968939061932e-05,
|
| 14876 |
+
"loss": 1.8382,
|
| 14877 |
+
"step": 31860
|
| 14878 |
+
},
|
| 14879 |
+
{
|
| 14880 |
+
"epoch": 0.06207182610930873,
|
| 14881 |
+
"grad_norm": 2.5900702476501465,
|
| 14882 |
+
"learning_rate": 2.9715702275146767e-05,
|
| 14883 |
+
"loss": 1.8717,
|
| 14884 |
+
"step": 31875
|
| 14885 |
+
},
|
| 14886 |
+
{
|
| 14887 |
+
"epoch": 0.06210103638041899,
|
| 14888 |
+
"grad_norm": 3.7013213634490967,
|
| 14889 |
+
"learning_rate": 2.9715435487308608e-05,
|
| 14890 |
+
"loss": 1.8794,
|
| 14891 |
+
"step": 31890
|
| 14892 |
+
},
|
| 14893 |
+
{
|
| 14894 |
+
"epoch": 0.062130246651529254,
|
| 14895 |
+
"grad_norm": 2.48606276512146,
|
| 14896 |
+
"learning_rate": 2.9715168575549688e-05,
|
| 14897 |
+
"loss": 1.9964,
|
| 14898 |
+
"step": 31905
|
| 14899 |
+
},
|
| 14900 |
+
{
|
| 14901 |
+
"epoch": 0.062159456922639515,
|
| 14902 |
+
"grad_norm": 2.358415365219116,
|
| 14903 |
+
"learning_rate": 2.9714901539872267e-05,
|
| 14904 |
+
"loss": 1.902,
|
| 14905 |
+
"step": 31920
|
| 14906 |
+
},
|
| 14907 |
+
{
|
| 14908 |
+
"epoch": 0.06218866719374978,
|
| 14909 |
+
"grad_norm": 3.140083074569702,
|
| 14910 |
+
"learning_rate": 2.9714634380278594e-05,
|
| 14911 |
+
"loss": 1.8696,
|
| 14912 |
+
"step": 31935
|
| 14913 |
+
},
|
| 14914 |
+
{
|
| 14915 |
+
"epoch": 0.062217877464860044,
|
| 14916 |
+
"grad_norm": 3.2135777473449707,
|
| 14917 |
+
"learning_rate": 2.9714367096770913e-05,
|
| 14918 |
+
"loss": 2.0002,
|
| 14919 |
+
"step": 31950
|
| 14920 |
+
},
|
| 14921 |
+
{
|
| 14922 |
+
"epoch": 0.062247087735970305,
|
| 14923 |
+
"grad_norm": 3.2304131984710693,
|
| 14924 |
+
"learning_rate": 2.9714099689351477e-05,
|
| 14925 |
+
"loss": 1.8721,
|
| 14926 |
+
"step": 31965
|
| 14927 |
+
},
|
| 14928 |
+
{
|
| 14929 |
+
"epoch": 0.06227629800708057,
|
| 14930 |
+
"grad_norm": 2.206209659576416,
|
| 14931 |
+
"learning_rate": 2.9713832158022535e-05,
|
| 14932 |
+
"loss": 1.9788,
|
| 14933 |
+
"step": 31980
|
| 14934 |
+
},
|
| 14935 |
+
{
|
| 14936 |
+
"epoch": 0.062305508278190834,
|
| 14937 |
+
"grad_norm": 3.055828332901001,
|
| 14938 |
+
"learning_rate": 2.9713564502786348e-05,
|
| 14939 |
+
"loss": 1.7976,
|
| 14940 |
+
"step": 31995
|
| 14941 |
+
},
|
| 14942 |
+
{
|
| 14943 |
+
"epoch": 0.062334718549301095,
|
| 14944 |
+
"grad_norm": 3.891845226287842,
|
| 14945 |
+
"learning_rate": 2.9713296723645165e-05,
|
| 14946 |
+
"loss": 1.7772,
|
| 14947 |
+
"step": 32010
|
| 14948 |
+
},
|
| 14949 |
+
{
|
| 14950 |
+
"epoch": 0.062363928820411356,
|
| 14951 |
+
"grad_norm": 1.8532415628433228,
|
| 14952 |
+
"learning_rate": 2.9713028820601238e-05,
|
| 14953 |
+
"loss": 1.9817,
|
| 14954 |
+
"step": 32025
|
| 14955 |
+
},
|
| 14956 |
+
{
|
| 14957 |
+
"epoch": 0.062393139091521624,
|
| 14958 |
+
"grad_norm": 1.650211215019226,
|
| 14959 |
+
"learning_rate": 2.9712760793656826e-05,
|
| 14960 |
+
"loss": 2.1243,
|
| 14961 |
+
"step": 32040
|
| 14962 |
+
},
|
| 14963 |
+
{
|
| 14964 |
+
"epoch": 0.062422349362631885,
|
| 14965 |
+
"grad_norm": 2.663621187210083,
|
| 14966 |
+
"learning_rate": 2.971249264281419e-05,
|
| 14967 |
+
"loss": 2.0554,
|
| 14968 |
+
"step": 32055
|
| 14969 |
+
},
|
| 14970 |
+
{
|
| 14971 |
+
"epoch": 0.062451559633742146,
|
| 14972 |
+
"grad_norm": 3.2437026500701904,
|
| 14973 |
+
"learning_rate": 2.9712224368075578e-05,
|
| 14974 |
+
"loss": 1.765,
|
| 14975 |
+
"step": 32070
|
| 14976 |
+
},
|
| 14977 |
+
{
|
| 14978 |
+
"epoch": 0.06248076990485241,
|
| 14979 |
+
"grad_norm": 4.013802528381348,
|
| 14980 |
+
"learning_rate": 2.971195596944326e-05,
|
| 14981 |
+
"loss": 1.8195,
|
| 14982 |
+
"step": 32085
|
| 14983 |
+
},
|
| 14984 |
+
{
|
| 14985 |
+
"epoch": 0.06250998017596267,
|
| 14986 |
+
"grad_norm": 4.0295610427856445,
|
| 14987 |
+
"learning_rate": 2.971168744691949e-05,
|
| 14988 |
+
"loss": 2.036,
|
| 14989 |
+
"step": 32100
|
| 14990 |
+
},
|
| 14991 |
+
{
|
| 14992 |
+
"epoch": 0.06253919044707294,
|
| 14993 |
+
"grad_norm": 2.830064296722412,
|
| 14994 |
+
"learning_rate": 2.9711418800506533e-05,
|
| 14995 |
+
"loss": 1.8499,
|
| 14996 |
+
"step": 32115
|
| 14997 |
+
},
|
| 14998 |
+
{
|
| 14999 |
+
"epoch": 0.0625684007181832,
|
| 15000 |
+
"grad_norm": 3.896538019180298,
|
| 15001 |
+
"learning_rate": 2.9711150030206643e-05,
|
| 15002 |
+
"loss": 2.0089,
|
| 15003 |
+
"step": 32130
|
| 15004 |
+
},
|
| 15005 |
+
{
|
| 15006 |
+
"epoch": 0.06259761098929346,
|
| 15007 |
+
"grad_norm": 3.6802754402160645,
|
| 15008 |
+
"learning_rate": 2.97108811360221e-05,
|
| 15009 |
+
"loss": 1.9039,
|
| 15010 |
+
"step": 32145
|
| 15011 |
+
},
|
| 15012 |
+
{
|
| 15013 |
+
"epoch": 0.06262682126040373,
|
| 15014 |
+
"grad_norm": 2.8547956943511963,
|
| 15015 |
+
"learning_rate": 2.9710612117955148e-05,
|
| 15016 |
+
"loss": 1.9065,
|
| 15017 |
+
"step": 32160
|
| 15018 |
+
},
|
| 15019 |
+
{
|
| 15020 |
+
"epoch": 0.06265603153151399,
|
| 15021 |
+
"grad_norm": 2.0819807052612305,
|
| 15022 |
+
"learning_rate": 2.9710342976008066e-05,
|
| 15023 |
+
"loss": 1.7786,
|
| 15024 |
+
"step": 32175
|
| 15025 |
+
},
|
| 15026 |
+
{
|
| 15027 |
+
"epoch": 0.06268524180262425,
|
| 15028 |
+
"grad_norm": 2.173609972000122,
|
| 15029 |
+
"learning_rate": 2.9710073710183118e-05,
|
| 15030 |
+
"loss": 1.8011,
|
| 15031 |
+
"step": 32190
|
| 15032 |
+
},
|
| 15033 |
+
{
|
| 15034 |
+
"epoch": 0.06271445207373451,
|
| 15035 |
+
"grad_norm": 2.2233803272247314,
|
| 15036 |
+
"learning_rate": 2.970980432048257e-05,
|
| 15037 |
+
"loss": 1.8308,
|
| 15038 |
+
"step": 32205
|
| 15039 |
+
},
|
| 15040 |
+
{
|
| 15041 |
+
"epoch": 0.06274366234484478,
|
| 15042 |
+
"grad_norm": 2.227229356765747,
|
| 15043 |
+
"learning_rate": 2.9709534806908692e-05,
|
| 15044 |
+
"loss": 1.9227,
|
| 15045 |
+
"step": 32220
|
| 15046 |
+
},
|
| 15047 |
+
{
|
| 15048 |
+
"epoch": 0.06277287261595504,
|
| 15049 |
+
"grad_norm": 3.8464114665985107,
|
| 15050 |
+
"learning_rate": 2.970926516946375e-05,
|
| 15051 |
+
"loss": 1.9808,
|
| 15052 |
+
"step": 32235
|
| 15053 |
+
},
|
| 15054 |
+
{
|
| 15055 |
+
"epoch": 0.0628020828870653,
|
| 15056 |
+
"grad_norm": 4.511869430541992,
|
| 15057 |
+
"learning_rate": 2.9708995408150018e-05,
|
| 15058 |
+
"loss": 1.9856,
|
| 15059 |
+
"step": 32250
|
| 15060 |
+
},
|
| 15061 |
+
{
|
| 15062 |
+
"epoch": 0.06283129315817557,
|
| 15063 |
+
"grad_norm": 2.0209896564483643,
|
| 15064 |
+
"learning_rate": 2.9708725522969767e-05,
|
| 15065 |
+
"loss": 1.8766,
|
| 15066 |
+
"step": 32265
|
| 15067 |
+
},
|
| 15068 |
+
{
|
| 15069 |
+
"epoch": 0.06286050342928583,
|
| 15070 |
+
"grad_norm": 4.369567394256592,
|
| 15071 |
+
"learning_rate": 2.9708455513925273e-05,
|
| 15072 |
+
"loss": 1.8103,
|
| 15073 |
+
"step": 32280
|
| 15074 |
+
},
|
| 15075 |
+
{
|
| 15076 |
+
"epoch": 0.06288971370039609,
|
| 15077 |
+
"grad_norm": 4.502676486968994,
|
| 15078 |
+
"learning_rate": 2.97081853810188e-05,
|
| 15079 |
+
"loss": 1.8489,
|
| 15080 |
+
"step": 32295
|
| 15081 |
+
},
|
| 15082 |
+
{
|
| 15083 |
+
"epoch": 0.06291892397150635,
|
| 15084 |
+
"grad_norm": 2.288998603820801,
|
| 15085 |
+
"learning_rate": 2.9707915124252637e-05,
|
| 15086 |
+
"loss": 1.9542,
|
| 15087 |
+
"step": 32310
|
| 15088 |
+
},
|
| 15089 |
+
{
|
| 15090 |
+
"epoch": 0.06294813424261662,
|
| 15091 |
+
"grad_norm": 2.6359355449676514,
|
| 15092 |
+
"learning_rate": 2.9707644743629046e-05,
|
| 15093 |
+
"loss": 1.7883,
|
| 15094 |
+
"step": 32325
|
| 15095 |
+
},
|
| 15096 |
+
{
|
| 15097 |
+
"epoch": 0.06297734451372689,
|
| 15098 |
+
"grad_norm": 4.356678485870361,
|
| 15099 |
+
"learning_rate": 2.970737423915031e-05,
|
| 15100 |
+
"loss": 1.8937,
|
| 15101 |
+
"step": 32340
|
| 15102 |
+
},
|
| 15103 |
+
{
|
| 15104 |
+
"epoch": 0.06300655478483715,
|
| 15105 |
+
"grad_norm": 3.2190353870391846,
|
| 15106 |
+
"learning_rate": 2.970710361081871e-05,
|
| 15107 |
+
"loss": 1.9758,
|
| 15108 |
+
"step": 32355
|
| 15109 |
+
},
|
| 15110 |
+
{
|
| 15111 |
+
"epoch": 0.06303576505594741,
|
| 15112 |
+
"grad_norm": 2.522303819656372,
|
| 15113 |
+
"learning_rate": 2.970683285863652e-05,
|
| 15114 |
+
"loss": 1.7315,
|
| 15115 |
+
"step": 32370
|
| 15116 |
+
},
|
| 15117 |
+
{
|
| 15118 |
+
"epoch": 0.06306497532705767,
|
| 15119 |
+
"grad_norm": 2.500927448272705,
|
| 15120 |
+
"learning_rate": 2.9706561982606023e-05,
|
| 15121 |
+
"loss": 1.9004,
|
| 15122 |
+
"step": 32385
|
| 15123 |
+
},
|
| 15124 |
+
{
|
| 15125 |
+
"epoch": 0.06309418559816793,
|
| 15126 |
+
"grad_norm": 3.656970262527466,
|
| 15127 |
+
"learning_rate": 2.97062909827295e-05,
|
| 15128 |
+
"loss": 1.9246,
|
| 15129 |
+
"step": 32400
|
| 15130 |
+
},
|
| 15131 |
+
{
|
| 15132 |
+
"epoch": 0.06312339586927819,
|
| 15133 |
+
"grad_norm": 4.1837158203125,
|
| 15134 |
+
"learning_rate": 2.970601985900923e-05,
|
| 15135 |
+
"loss": 1.6453,
|
| 15136 |
+
"step": 32415
|
| 15137 |
+
},
|
| 15138 |
+
{
|
| 15139 |
+
"epoch": 0.06315260614038845,
|
| 15140 |
+
"grad_norm": 5.141902446746826,
|
| 15141 |
+
"learning_rate": 2.9705748611447498e-05,
|
| 15142 |
+
"loss": 1.7215,
|
| 15143 |
+
"step": 32430
|
| 15144 |
+
},
|
| 15145 |
+
{
|
| 15146 |
+
"epoch": 0.06318181641149873,
|
| 15147 |
+
"grad_norm": 5.384919166564941,
|
| 15148 |
+
"learning_rate": 2.9705477240046595e-05,
|
| 15149 |
+
"loss": 2.0638,
|
| 15150 |
+
"step": 32445
|
| 15151 |
+
},
|
| 15152 |
+
{
|
| 15153 |
+
"epoch": 0.06321102668260899,
|
| 15154 |
+
"grad_norm": 2.812014579772949,
|
| 15155 |
+
"learning_rate": 2.9705205744808795e-05,
|
| 15156 |
+
"loss": 1.8133,
|
| 15157 |
+
"step": 32460
|
| 15158 |
+
},
|
| 15159 |
+
{
|
| 15160 |
+
"epoch": 0.06324023695371925,
|
| 15161 |
+
"grad_norm": 4.618744850158691,
|
| 15162 |
+
"learning_rate": 2.970493412573639e-05,
|
| 15163 |
+
"loss": 1.7221,
|
| 15164 |
+
"step": 32475
|
| 15165 |
+
},
|
| 15166 |
+
{
|
| 15167 |
+
"epoch": 0.06326944722482951,
|
| 15168 |
+
"grad_norm": 2.0948140621185303,
|
| 15169 |
+
"learning_rate": 2.9704662382831665e-05,
|
| 15170 |
+
"loss": 1.7292,
|
| 15171 |
+
"step": 32490
|
| 15172 |
+
},
|
| 15173 |
+
{
|
| 15174 |
+
"epoch": 0.06329865749593977,
|
| 15175 |
+
"grad_norm": 2.2021284103393555,
|
| 15176 |
+
"learning_rate": 2.970439051609691e-05,
|
| 15177 |
+
"loss": 1.6685,
|
| 15178 |
+
"step": 32505
|
| 15179 |
+
},
|
| 15180 |
+
{
|
| 15181 |
+
"epoch": 0.06332786776705003,
|
| 15182 |
+
"grad_norm": 4.773664474487305,
|
| 15183 |
+
"learning_rate": 2.9704118525534414e-05,
|
| 15184 |
+
"loss": 1.8922,
|
| 15185 |
+
"step": 32520
|
| 15186 |
+
},
|
| 15187 |
+
{
|
| 15188 |
+
"epoch": 0.06335707803816029,
|
| 15189 |
+
"grad_norm": 2.716444253921509,
|
| 15190 |
+
"learning_rate": 2.970384641114647e-05,
|
| 15191 |
+
"loss": 1.816,
|
| 15192 |
+
"step": 32535
|
| 15193 |
+
},
|
| 15194 |
+
{
|
| 15195 |
+
"epoch": 0.06338628830927057,
|
| 15196 |
+
"grad_norm": 4.086103916168213,
|
| 15197 |
+
"learning_rate": 2.9703574172935366e-05,
|
| 15198 |
+
"loss": 1.8084,
|
| 15199 |
+
"step": 32550
|
| 15200 |
+
},
|
| 15201 |
+
{
|
| 15202 |
+
"epoch": 0.06341549858038083,
|
| 15203 |
+
"grad_norm": 2.4854936599731445,
|
| 15204 |
+
"learning_rate": 2.9703301810903397e-05,
|
| 15205 |
+
"loss": 1.7871,
|
| 15206 |
+
"step": 32565
|
| 15207 |
+
},
|
| 15208 |
+
{
|
| 15209 |
+
"epoch": 0.06344470885149109,
|
| 15210 |
+
"grad_norm": 1.8130111694335938,
|
| 15211 |
+
"learning_rate": 2.9703029325052857e-05,
|
| 15212 |
+
"loss": 1.8795,
|
| 15213 |
+
"step": 32580
|
| 15214 |
+
},
|
| 15215 |
+
{
|
| 15216 |
+
"epoch": 0.06347391912260135,
|
| 15217 |
+
"grad_norm": 3.8686211109161377,
|
| 15218 |
+
"learning_rate": 2.970275671538604e-05,
|
| 15219 |
+
"loss": 1.9867,
|
| 15220 |
+
"step": 32595
|
| 15221 |
+
},
|
| 15222 |
+
{
|
| 15223 |
+
"epoch": 0.06350312939371161,
|
| 15224 |
+
"grad_norm": 2.5253164768218994,
|
| 15225 |
+
"learning_rate": 2.9702483981905233e-05,
|
| 15226 |
+
"loss": 1.875,
|
| 15227 |
+
"step": 32610
|
| 15228 |
+
},
|
| 15229 |
+
{
|
| 15230 |
+
"epoch": 0.06353233966482187,
|
| 15231 |
+
"grad_norm": 3.0097036361694336,
|
| 15232 |
+
"learning_rate": 2.9702211124612745e-05,
|
| 15233 |
+
"loss": 1.9521,
|
| 15234 |
+
"step": 32625
|
| 15235 |
+
},
|
| 15236 |
+
{
|
| 15237 |
+
"epoch": 0.06356154993593213,
|
| 15238 |
+
"grad_norm": 3.3433878421783447,
|
| 15239 |
+
"learning_rate": 2.9701938143510873e-05,
|
| 15240 |
+
"loss": 1.806,
|
| 15241 |
+
"step": 32640
|
| 15242 |
+
},
|
| 15243 |
+
{
|
| 15244 |
+
"epoch": 0.06359076020704241,
|
| 15245 |
+
"grad_norm": 1.4688777923583984,
|
| 15246 |
+
"learning_rate": 2.970166503860191e-05,
|
| 15247 |
+
"loss": 1.8248,
|
| 15248 |
+
"step": 32655
|
| 15249 |
+
},
|
| 15250 |
+
{
|
| 15251 |
+
"epoch": 0.06361997047815267,
|
| 15252 |
+
"grad_norm": 2.365694522857666,
|
| 15253 |
+
"learning_rate": 2.9701391809888156e-05,
|
| 15254 |
+
"loss": 1.7152,
|
| 15255 |
+
"step": 32670
|
| 15256 |
+
},
|
| 15257 |
+
{
|
| 15258 |
+
"epoch": 0.06364918074926293,
|
| 15259 |
+
"grad_norm": 2.6821203231811523,
|
| 15260 |
+
"learning_rate": 2.9701118457371915e-05,
|
| 15261 |
+
"loss": 1.858,
|
| 15262 |
+
"step": 32685
|
| 15263 |
+
},
|
| 15264 |
+
{
|
| 15265 |
+
"epoch": 0.06367839102037319,
|
| 15266 |
+
"grad_norm": 1.842124342918396,
|
| 15267 |
+
"learning_rate": 2.9700844981055486e-05,
|
| 15268 |
+
"loss": 1.7689,
|
| 15269 |
+
"step": 32700
|
| 15270 |
+
},
|
| 15271 |
+
{
|
| 15272 |
+
"epoch": 0.06370760129148345,
|
| 15273 |
+
"grad_norm": 2.2430474758148193,
|
| 15274 |
+
"learning_rate": 2.9700571380941178e-05,
|
| 15275 |
+
"loss": 2.0056,
|
| 15276 |
+
"step": 32715
|
| 15277 |
+
},
|
| 15278 |
+
{
|
| 15279 |
+
"epoch": 0.06373681156259371,
|
| 15280 |
+
"grad_norm": 3.1308460235595703,
|
| 15281 |
+
"learning_rate": 2.970029765703129e-05,
|
| 15282 |
+
"loss": 1.9163,
|
| 15283 |
+
"step": 32730
|
| 15284 |
+
},
|
| 15285 |
+
{
|
| 15286 |
+
"epoch": 0.06376602183370397,
|
| 15287 |
+
"grad_norm": 2.6456706523895264,
|
| 15288 |
+
"learning_rate": 2.9700023809328122e-05,
|
| 15289 |
+
"loss": 1.8002,
|
| 15290 |
+
"step": 32745
|
| 15291 |
+
},
|
| 15292 |
+
{
|
| 15293 |
+
"epoch": 0.06379523210481425,
|
| 15294 |
+
"grad_norm": 3.659024953842163,
|
| 15295 |
+
"learning_rate": 2.969974983783399e-05,
|
| 15296 |
+
"loss": 1.9222,
|
| 15297 |
+
"step": 32760
|
| 15298 |
+
},
|
| 15299 |
+
{
|
| 15300 |
+
"epoch": 0.06382444237592451,
|
| 15301 |
+
"grad_norm": 2.3271381855010986,
|
| 15302 |
+
"learning_rate": 2.96994757425512e-05,
|
| 15303 |
+
"loss": 1.7778,
|
| 15304 |
+
"step": 32775
|
| 15305 |
+
},
|
| 15306 |
+
{
|
| 15307 |
+
"epoch": 0.06385365264703477,
|
| 15308 |
+
"grad_norm": 3.411533832550049,
|
| 15309 |
+
"learning_rate": 2.9699201523482057e-05,
|
| 15310 |
+
"loss": 1.778,
|
| 15311 |
+
"step": 32790
|
| 15312 |
+
},
|
| 15313 |
+
{
|
| 15314 |
+
"epoch": 0.06388286291814503,
|
| 15315 |
+
"grad_norm": 2.649052143096924,
|
| 15316 |
+
"learning_rate": 2.9698927180628866e-05,
|
| 15317 |
+
"loss": 1.8199,
|
| 15318 |
+
"step": 32805
|
| 15319 |
+
},
|
| 15320 |
+
{
|
| 15321 |
+
"epoch": 0.06391207318925529,
|
| 15322 |
+
"grad_norm": 2.8163363933563232,
|
| 15323 |
+
"learning_rate": 2.9698652713993947e-05,
|
| 15324 |
+
"loss": 1.8579,
|
| 15325 |
+
"step": 32820
|
| 15326 |
+
},
|
| 15327 |
+
{
|
| 15328 |
+
"epoch": 0.06394128346036555,
|
| 15329 |
+
"grad_norm": 3.319955348968506,
|
| 15330 |
+
"learning_rate": 2.9698378123579603e-05,
|
| 15331 |
+
"loss": 1.9385,
|
| 15332 |
+
"step": 32835
|
| 15333 |
+
},
|
| 15334 |
+
{
|
| 15335 |
+
"epoch": 0.06397049373147581,
|
| 15336 |
+
"grad_norm": 2.217761754989624,
|
| 15337 |
+
"learning_rate": 2.969810340938815e-05,
|
| 15338 |
+
"loss": 2.0007,
|
| 15339 |
+
"step": 32850
|
| 15340 |
+
},
|
| 15341 |
+
{
|
| 15342 |
+
"epoch": 0.06399970400258609,
|
| 15343 |
+
"grad_norm": 4.801873683929443,
|
| 15344 |
+
"learning_rate": 2.96978285714219e-05,
|
| 15345 |
+
"loss": 1.9876,
|
| 15346 |
+
"step": 32865
|
| 15347 |
+
},
|
| 15348 |
+
{
|
| 15349 |
+
"epoch": 0.06402891427369635,
|
| 15350 |
+
"grad_norm": 3.5146069526672363,
|
| 15351 |
+
"learning_rate": 2.969755360968317e-05,
|
| 15352 |
+
"loss": 1.8267,
|
| 15353 |
+
"step": 32880
|
| 15354 |
+
},
|
| 15355 |
+
{
|
| 15356 |
+
"epoch": 0.06405812454480661,
|
| 15357 |
+
"grad_norm": 3.407545328140259,
|
| 15358 |
+
"learning_rate": 2.9697278524174275e-05,
|
| 15359 |
+
"loss": 1.743,
|
| 15360 |
+
"step": 32895
|
| 15361 |
+
},
|
| 15362 |
+
{
|
| 15363 |
+
"epoch": 0.06408733481591687,
|
| 15364 |
+
"grad_norm": 1.410946011543274,
|
| 15365 |
+
"learning_rate": 2.969700331489753e-05,
|
| 15366 |
+
"loss": 1.9281,
|
| 15367 |
+
"step": 32910
|
| 15368 |
+
},
|
| 15369 |
+
{
|
| 15370 |
+
"epoch": 0.06411654508702713,
|
| 15371 |
+
"grad_norm": 2.602207899093628,
|
| 15372 |
+
"learning_rate": 2.9696727981855253e-05,
|
| 15373 |
+
"loss": 1.8532,
|
| 15374 |
+
"step": 32925
|
| 15375 |
+
},
|
| 15376 |
+
{
|
| 15377 |
+
"epoch": 0.0641457553581374,
|
| 15378 |
+
"grad_norm": 4.358025550842285,
|
| 15379 |
+
"learning_rate": 2.969645252504976e-05,
|
| 15380 |
+
"loss": 1.7457,
|
| 15381 |
+
"step": 32940
|
| 15382 |
+
},
|
| 15383 |
+
{
|
| 15384 |
+
"epoch": 0.06417496562924765,
|
| 15385 |
+
"grad_norm": 3.8209457397460938,
|
| 15386 |
+
"learning_rate": 2.9696176944483373e-05,
|
| 15387 |
+
"loss": 2.0736,
|
| 15388 |
+
"step": 32955
|
| 15389 |
+
},
|
| 15390 |
+
{
|
| 15391 |
+
"epoch": 0.06420417590035793,
|
| 15392 |
+
"grad_norm": 3.317567825317383,
|
| 15393 |
+
"learning_rate": 2.9695901240158415e-05,
|
| 15394 |
+
"loss": 1.9322,
|
| 15395 |
+
"step": 32970
|
| 15396 |
+
},
|
| 15397 |
+
{
|
| 15398 |
+
"epoch": 0.06423338617146819,
|
| 15399 |
+
"grad_norm": 4.668582439422607,
|
| 15400 |
+
"learning_rate": 2.9695625412077208e-05,
|
| 15401 |
+
"loss": 2.0152,
|
| 15402 |
+
"step": 32985
|
| 15403 |
+
},
|
| 15404 |
+
{
|
| 15405 |
+
"epoch": 0.06426259644257845,
|
| 15406 |
+
"grad_norm": 2.6789300441741943,
|
| 15407 |
+
"learning_rate": 2.969534946024207e-05,
|
| 15408 |
+
"loss": 1.7129,
|
| 15409 |
+
"step": 33000
|
| 15410 |
+
},
|
| 15411 |
+
{
|
| 15412 |
+
"epoch": 0.06429180671368871,
|
| 15413 |
+
"grad_norm": 2.073744297027588,
|
| 15414 |
+
"learning_rate": 2.9695073384655326e-05,
|
| 15415 |
+
"loss": 1.6952,
|
| 15416 |
+
"step": 33015
|
| 15417 |
+
},
|
| 15418 |
+
{
|
| 15419 |
+
"epoch": 0.06432101698479897,
|
| 15420 |
+
"grad_norm": 3.3017473220825195,
|
| 15421 |
+
"learning_rate": 2.9694797185319304e-05,
|
| 15422 |
+
"loss": 1.8158,
|
| 15423 |
+
"step": 33030
|
| 15424 |
+
},
|
| 15425 |
+
{
|
| 15426 |
+
"epoch": 0.06435022725590923,
|
| 15427 |
+
"grad_norm": 4.441155433654785,
|
| 15428 |
+
"learning_rate": 2.969452086223633e-05,
|
| 15429 |
+
"loss": 1.9204,
|
| 15430 |
+
"step": 33045
|
| 15431 |
+
},
|
| 15432 |
+
{
|
| 15433 |
+
"epoch": 0.0643794375270195,
|
| 15434 |
+
"grad_norm": 2.9764716625213623,
|
| 15435 |
+
"learning_rate": 2.9694244415408727e-05,
|
| 15436 |
+
"loss": 1.8952,
|
| 15437 |
+
"step": 33060
|
| 15438 |
+
},
|
| 15439 |
+
{
|
| 15440 |
+
"epoch": 0.06440864779812977,
|
| 15441 |
+
"grad_norm": 4.1646575927734375,
|
| 15442 |
+
"learning_rate": 2.9693967844838827e-05,
|
| 15443 |
+
"loss": 1.7692,
|
| 15444 |
+
"step": 33075
|
| 15445 |
+
},
|
| 15446 |
+
{
|
| 15447 |
+
"epoch": 0.06443785806924003,
|
| 15448 |
+
"grad_norm": 4.056210994720459,
|
| 15449 |
+
"learning_rate": 2.9693691150528955e-05,
|
| 15450 |
+
"loss": 1.855,
|
| 15451 |
+
"step": 33090
|
| 15452 |
+
},
|
| 15453 |
+
{
|
| 15454 |
+
"epoch": 0.06446706834035029,
|
| 15455 |
+
"grad_norm": 2.6098761558532715,
|
| 15456 |
+
"learning_rate": 2.9693414332481443e-05,
|
| 15457 |
+
"loss": 1.8284,
|
| 15458 |
+
"step": 33105
|
| 15459 |
+
},
|
| 15460 |
+
{
|
| 15461 |
+
"epoch": 0.06449627861146055,
|
| 15462 |
+
"grad_norm": 4.3388166427612305,
|
| 15463 |
+
"learning_rate": 2.9693137390698622e-05,
|
| 15464 |
+
"loss": 1.8155,
|
| 15465 |
+
"step": 33120
|
| 15466 |
+
},
|
| 15467 |
+
{
|
| 15468 |
+
"epoch": 0.06452548888257081,
|
| 15469 |
+
"grad_norm": 3.3735921382904053,
|
| 15470 |
+
"learning_rate": 2.9692860325182825e-05,
|
| 15471 |
+
"loss": 1.7765,
|
| 15472 |
+
"step": 33135
|
| 15473 |
+
},
|
| 15474 |
+
{
|
| 15475 |
+
"epoch": 0.06455469915368107,
|
| 15476 |
+
"grad_norm": 3.9341137409210205,
|
| 15477 |
+
"learning_rate": 2.9692583135936385e-05,
|
| 15478 |
+
"loss": 1.7622,
|
| 15479 |
+
"step": 33150
|
| 15480 |
+
},
|
| 15481 |
+
{
|
| 15482 |
+
"epoch": 0.06458390942479134,
|
| 15483 |
+
"grad_norm": 5.620511531829834,
|
| 15484 |
+
"learning_rate": 2.9692305822961637e-05,
|
| 15485 |
+
"loss": 1.7926,
|
| 15486 |
+
"step": 33165
|
| 15487 |
+
},
|
| 15488 |
+
{
|
| 15489 |
+
"epoch": 0.06461311969590161,
|
| 15490 |
+
"grad_norm": 1.8949609994888306,
|
| 15491 |
+
"learning_rate": 2.9692028386260915e-05,
|
| 15492 |
+
"loss": 1.9158,
|
| 15493 |
+
"step": 33180
|
| 15494 |
+
},
|
| 15495 |
+
{
|
| 15496 |
+
"epoch": 0.06464232996701187,
|
| 15497 |
+
"grad_norm": 2.183438539505005,
|
| 15498 |
+
"learning_rate": 2.9691750825836557e-05,
|
| 15499 |
+
"loss": 1.9586,
|
| 15500 |
+
"step": 33195
|
| 15501 |
+
},
|
| 15502 |
+
{
|
| 15503 |
+
"epoch": 0.06467154023812213,
|
| 15504 |
+
"grad_norm": 2.05364727973938,
|
| 15505 |
+
"learning_rate": 2.9691473141690894e-05,
|
| 15506 |
+
"loss": 1.8952,
|
| 15507 |
+
"step": 33210
|
| 15508 |
+
},
|
| 15509 |
+
{
|
| 15510 |
+
"epoch": 0.0647007505092324,
|
| 15511 |
+
"grad_norm": 2.479680299758911,
|
| 15512 |
+
"learning_rate": 2.9691195333826277e-05,
|
| 15513 |
+
"loss": 1.7908,
|
| 15514 |
+
"step": 33225
|
| 15515 |
+
},
|
| 15516 |
+
{
|
| 15517 |
+
"epoch": 0.06472996078034265,
|
| 15518 |
+
"grad_norm": 2.0384159088134766,
|
| 15519 |
+
"learning_rate": 2.9690917402245034e-05,
|
| 15520 |
+
"loss": 1.8231,
|
| 15521 |
+
"step": 33240
|
| 15522 |
+
},
|
| 15523 |
+
{
|
| 15524 |
+
"epoch": 0.06475917105145292,
|
| 15525 |
+
"grad_norm": 2.3003928661346436,
|
| 15526 |
+
"learning_rate": 2.9690639346949503e-05,
|
| 15527 |
+
"loss": 1.6675,
|
| 15528 |
+
"step": 33255
|
| 15529 |
+
},
|
| 15530 |
+
{
|
| 15531 |
+
"epoch": 0.06478838132256318,
|
| 15532 |
+
"grad_norm": 2.9308221340179443,
|
| 15533 |
+
"learning_rate": 2.9690361167942042e-05,
|
| 15534 |
+
"loss": 1.9136,
|
| 15535 |
+
"step": 33270
|
| 15536 |
+
},
|
| 15537 |
+
{
|
| 15538 |
+
"epoch": 0.06481759159367345,
|
| 15539 |
+
"grad_norm": 2.3866639137268066,
|
| 15540 |
+
"learning_rate": 2.9690082865224975e-05,
|
| 15541 |
+
"loss": 1.8992,
|
| 15542 |
+
"step": 33285
|
| 15543 |
+
},
|
| 15544 |
+
{
|
| 15545 |
+
"epoch": 0.06484680186478371,
|
| 15546 |
+
"grad_norm": 4.399453163146973,
|
| 15547 |
+
"learning_rate": 2.968980443880066e-05,
|
| 15548 |
+
"loss": 1.9369,
|
| 15549 |
+
"step": 33300
|
| 15550 |
+
},
|
| 15551 |
+
{
|
| 15552 |
+
"epoch": 0.06487601213589397,
|
| 15553 |
+
"grad_norm": 3.06842041015625,
|
| 15554 |
+
"learning_rate": 2.9689525888671436e-05,
|
| 15555 |
+
"loss": 1.9624,
|
| 15556 |
+
"step": 33315
|
| 15557 |
+
},
|
| 15558 |
+
{
|
| 15559 |
+
"epoch": 0.06490522240700423,
|
| 15560 |
+
"grad_norm": 1.9636744260787964,
|
| 15561 |
+
"learning_rate": 2.9689247214839647e-05,
|
| 15562 |
+
"loss": 1.7785,
|
| 15563 |
+
"step": 33330
|
| 15564 |
+
},
|
| 15565 |
+
{
|
| 15566 |
+
"epoch": 0.0649344326781145,
|
| 15567 |
+
"grad_norm": 3.882155418395996,
|
| 15568 |
+
"learning_rate": 2.968896841730764e-05,
|
| 15569 |
+
"loss": 1.8328,
|
| 15570 |
+
"step": 33345
|
| 15571 |
+
},
|
| 15572 |
+
{
|
| 15573 |
+
"epoch": 0.06496364294922476,
|
| 15574 |
+
"grad_norm": 3.773200511932373,
|
| 15575 |
+
"learning_rate": 2.9688689496077764e-05,
|
| 15576 |
+
"loss": 1.9322,
|
| 15577 |
+
"step": 33360
|
| 15578 |
+
},
|
| 15579 |
+
{
|
| 15580 |
+
"epoch": 0.06499285322033502,
|
| 15581 |
+
"grad_norm": 2.331101417541504,
|
| 15582 |
+
"learning_rate": 2.968841045115237e-05,
|
| 15583 |
+
"loss": 1.7726,
|
| 15584 |
+
"step": 33375
|
| 15585 |
+
},
|
| 15586 |
+
{
|
| 15587 |
+
"epoch": 0.06502206349144529,
|
| 15588 |
+
"grad_norm": 3.3869986534118652,
|
| 15589 |
+
"learning_rate": 2.9688131282533802e-05,
|
| 15590 |
+
"loss": 1.9489,
|
| 15591 |
+
"step": 33390
|
| 15592 |
+
},
|
| 15593 |
+
{
|
| 15594 |
+
"epoch": 0.06505127376255555,
|
| 15595 |
+
"grad_norm": 2.627638578414917,
|
| 15596 |
+
"learning_rate": 2.968785199022442e-05,
|
| 15597 |
+
"loss": 1.7626,
|
| 15598 |
+
"step": 33405
|
| 15599 |
+
},
|
| 15600 |
+
{
|
| 15601 |
+
"epoch": 0.06508048403366581,
|
| 15602 |
+
"grad_norm": 2.031043291091919,
|
| 15603 |
+
"learning_rate": 2.9687572574226566e-05,
|
| 15604 |
+
"loss": 1.8948,
|
| 15605 |
+
"step": 33420
|
| 15606 |
+
},
|
| 15607 |
+
{
|
| 15608 |
+
"epoch": 0.06510969430477608,
|
| 15609 |
+
"grad_norm": 1.6995463371276855,
|
| 15610 |
+
"learning_rate": 2.96872930345426e-05,
|
| 15611 |
+
"loss": 1.8642,
|
| 15612 |
+
"step": 33435
|
| 15613 |
+
},
|
| 15614 |
+
{
|
| 15615 |
+
"epoch": 0.06513890457588634,
|
| 15616 |
+
"grad_norm": 3.292022943496704,
|
| 15617 |
+
"learning_rate": 2.9687013371174873e-05,
|
| 15618 |
+
"loss": 1.8344,
|
| 15619 |
+
"step": 33450
|
| 15620 |
+
},
|
| 15621 |
+
{
|
| 15622 |
+
"epoch": 0.0651681148469966,
|
| 15623 |
+
"grad_norm": 2.9283857345581055,
|
| 15624 |
+
"learning_rate": 2.968673358412574e-05,
|
| 15625 |
+
"loss": 2.0218,
|
| 15626 |
+
"step": 33465
|
| 15627 |
+
},
|
| 15628 |
+
{
|
| 15629 |
+
"epoch": 0.06519732511810686,
|
| 15630 |
+
"grad_norm": 3.2172300815582275,
|
| 15631 |
+
"learning_rate": 2.968645367339756e-05,
|
| 15632 |
+
"loss": 1.8049,
|
| 15633 |
+
"step": 33480
|
| 15634 |
+
},
|
| 15635 |
+
{
|
| 15636 |
+
"epoch": 0.06522653538921712,
|
| 15637 |
+
"grad_norm": 2.1529486179351807,
|
| 15638 |
+
"learning_rate": 2.9686173638992687e-05,
|
| 15639 |
+
"loss": 1.8755,
|
| 15640 |
+
"step": 33495
|
| 15641 |
+
},
|
| 15642 |
+
{
|
| 15643 |
+
"epoch": 0.0652557456603274,
|
| 15644 |
+
"grad_norm": 2.5349905490875244,
|
| 15645 |
+
"learning_rate": 2.9685893480913477e-05,
|
| 15646 |
+
"loss": 2.0173,
|
| 15647 |
+
"step": 33510
|
| 15648 |
+
},
|
| 15649 |
+
{
|
| 15650 |
+
"epoch": 0.06528495593143765,
|
| 15651 |
+
"grad_norm": 3.646724224090576,
|
| 15652 |
+
"learning_rate": 2.9685613199162296e-05,
|
| 15653 |
+
"loss": 1.9269,
|
| 15654 |
+
"step": 33525
|
| 15655 |
+
},
|
| 15656 |
+
{
|
| 15657 |
+
"epoch": 0.06531416620254792,
|
| 15658 |
+
"grad_norm": 3.3442795276641846,
|
| 15659 |
+
"learning_rate": 2.96853327937415e-05,
|
| 15660 |
+
"loss": 1.7363,
|
| 15661 |
+
"step": 33540
|
| 15662 |
+
},
|
| 15663 |
+
{
|
| 15664 |
+
"epoch": 0.06534337647365818,
|
| 15665 |
+
"grad_norm": 3.301544189453125,
|
| 15666 |
+
"learning_rate": 2.9685052264653452e-05,
|
| 15667 |
+
"loss": 1.9225,
|
| 15668 |
+
"step": 33555
|
| 15669 |
+
},
|
| 15670 |
+
{
|
| 15671 |
+
"epoch": 0.06537258674476844,
|
| 15672 |
+
"grad_norm": 1.919858455657959,
|
| 15673 |
+
"learning_rate": 2.9684771611900512e-05,
|
| 15674 |
+
"loss": 1.7969,
|
| 15675 |
+
"step": 33570
|
| 15676 |
+
},
|
| 15677 |
+
{
|
| 15678 |
+
"epoch": 0.0654017970158787,
|
| 15679 |
+
"grad_norm": 3.6086807250976562,
|
| 15680 |
+
"learning_rate": 2.9684490835485045e-05,
|
| 15681 |
+
"loss": 1.8075,
|
| 15682 |
+
"step": 33585
|
| 15683 |
+
},
|
| 15684 |
+
{
|
| 15685 |
+
"epoch": 0.06543100728698896,
|
| 15686 |
+
"grad_norm": 1.659013032913208,
|
| 15687 |
+
"learning_rate": 2.9684209935409418e-05,
|
| 15688 |
+
"loss": 1.7631,
|
| 15689 |
+
"step": 33600
|
| 15690 |
+
},
|
| 15691 |
+
{
|
| 15692 |
+
"epoch": 0.06546021755809923,
|
| 15693 |
+
"grad_norm": 4.616360187530518,
|
| 15694 |
+
"learning_rate": 2.968392891167599e-05,
|
| 15695 |
+
"loss": 1.817,
|
| 15696 |
+
"step": 33615
|
| 15697 |
+
},
|
| 15698 |
+
{
|
| 15699 |
+
"epoch": 0.0654894278292095,
|
| 15700 |
+
"grad_norm": 2.684124708175659,
|
| 15701 |
+
"learning_rate": 2.9683647764287136e-05,
|
| 15702 |
+
"loss": 1.8655,
|
| 15703 |
+
"step": 33630
|
| 15704 |
+
},
|
| 15705 |
+
{
|
| 15706 |
+
"epoch": 0.06551863810031976,
|
| 15707 |
+
"grad_norm": 2.5763068199157715,
|
| 15708 |
+
"learning_rate": 2.9683366493245213e-05,
|
| 15709 |
+
"loss": 1.851,
|
| 15710 |
+
"step": 33645
|
| 15711 |
+
},
|
| 15712 |
+
{
|
| 15713 |
+
"epoch": 0.06554784837143002,
|
| 15714 |
+
"grad_norm": 4.1813859939575195,
|
| 15715 |
+
"learning_rate": 2.96830850985526e-05,
|
| 15716 |
+
"loss": 1.8571,
|
| 15717 |
+
"step": 33660
|
| 15718 |
+
},
|
| 15719 |
+
{
|
| 15720 |
+
"epoch": 0.06557705864254028,
|
| 15721 |
+
"grad_norm": 3.5396945476531982,
|
| 15722 |
+
"learning_rate": 2.968280358021166e-05,
|
| 15723 |
+
"loss": 1.8414,
|
| 15724 |
+
"step": 33675
|
| 15725 |
+
},
|
| 15726 |
+
{
|
| 15727 |
+
"epoch": 0.06560626891365054,
|
| 15728 |
+
"grad_norm": 3.497781276702881,
|
| 15729 |
+
"learning_rate": 2.968252193822477e-05,
|
| 15730 |
+
"loss": 2.0295,
|
| 15731 |
+
"step": 33690
|
| 15732 |
+
},
|
| 15733 |
+
{
|
| 15734 |
+
"epoch": 0.0656354791847608,
|
| 15735 |
+
"grad_norm": 2.3261985778808594,
|
| 15736 |
+
"learning_rate": 2.9682240172594294e-05,
|
| 15737 |
+
"loss": 1.8092,
|
| 15738 |
+
"step": 33705
|
| 15739 |
+
},
|
| 15740 |
+
{
|
| 15741 |
+
"epoch": 0.06566468945587108,
|
| 15742 |
+
"grad_norm": 2.558701753616333,
|
| 15743 |
+
"learning_rate": 2.968195828332261e-05,
|
| 15744 |
+
"loss": 1.9174,
|
| 15745 |
+
"step": 33720
|
| 15746 |
+
},
|
| 15747 |
+
{
|
| 15748 |
+
"epoch": 0.06569389972698134,
|
| 15749 |
+
"grad_norm": 3.210392713546753,
|
| 15750 |
+
"learning_rate": 2.9681676270412092e-05,
|
| 15751 |
+
"loss": 1.8506,
|
| 15752 |
+
"step": 33735
|
| 15753 |
+
},
|
| 15754 |
+
{
|
| 15755 |
+
"epoch": 0.0657231099980916,
|
| 15756 |
+
"grad_norm": 2.7672812938690186,
|
| 15757 |
+
"learning_rate": 2.968139413386511e-05,
|
| 15758 |
+
"loss": 1.9361,
|
| 15759 |
+
"step": 33750
|
| 15760 |
+
},
|
| 15761 |
+
{
|
| 15762 |
+
"epoch": 0.06575232026920186,
|
| 15763 |
+
"grad_norm": 4.212212562561035,
|
| 15764 |
+
"learning_rate": 2.9681111873684046e-05,
|
| 15765 |
+
"loss": 1.7308,
|
| 15766 |
+
"step": 33765
|
| 15767 |
+
},
|
| 15768 |
+
{
|
| 15769 |
+
"epoch": 0.06578153054031212,
|
| 15770 |
+
"grad_norm": 1.9889777898788452,
|
| 15771 |
+
"learning_rate": 2.9680829489871274e-05,
|
| 15772 |
+
"loss": 1.8549,
|
| 15773 |
+
"step": 33780
|
| 15774 |
+
},
|
| 15775 |
+
{
|
| 15776 |
+
"epoch": 0.06581074081142238,
|
| 15777 |
+
"grad_norm": 1.7036499977111816,
|
| 15778 |
+
"learning_rate": 2.9680546982429166e-05,
|
| 15779 |
+
"loss": 1.877,
|
| 15780 |
+
"step": 33795
|
| 15781 |
+
},
|
| 15782 |
+
{
|
| 15783 |
+
"epoch": 0.06583995108253264,
|
| 15784 |
+
"grad_norm": 2.7053897380828857,
|
| 15785 |
+
"learning_rate": 2.9680264351360115e-05,
|
| 15786 |
+
"loss": 1.8077,
|
| 15787 |
+
"step": 33810
|
| 15788 |
+
},
|
| 15789 |
+
{
|
| 15790 |
+
"epoch": 0.06586916135364292,
|
| 15791 |
+
"grad_norm": 2.8420169353485107,
|
| 15792 |
+
"learning_rate": 2.967998159666649e-05,
|
| 15793 |
+
"loss": 1.8568,
|
| 15794 |
+
"step": 33825
|
| 15795 |
+
},
|
| 15796 |
+
{
|
| 15797 |
+
"epoch": 0.06589837162475318,
|
| 15798 |
+
"grad_norm": 2.4401204586029053,
|
| 15799 |
+
"learning_rate": 2.9679698718350673e-05,
|
| 15800 |
+
"loss": 1.8119,
|
| 15801 |
+
"step": 33840
|
| 15802 |
+
},
|
| 15803 |
+
{
|
| 15804 |
+
"epoch": 0.06592758189586344,
|
| 15805 |
+
"grad_norm": 3.5529043674468994,
|
| 15806 |
+
"learning_rate": 2.9679415716415053e-05,
|
| 15807 |
+
"loss": 1.755,
|
| 15808 |
+
"step": 33855
|
| 15809 |
+
},
|
| 15810 |
+
{
|
| 15811 |
+
"epoch": 0.0659567921669737,
|
| 15812 |
+
"grad_norm": 2.672156572341919,
|
| 15813 |
+
"learning_rate": 2.9679132590862004e-05,
|
| 15814 |
+
"loss": 1.9222,
|
| 15815 |
+
"step": 33870
|
| 15816 |
+
},
|
| 15817 |
+
{
|
| 15818 |
+
"epoch": 0.06598600243808396,
|
| 15819 |
+
"grad_norm": 1.9536501169204712,
|
| 15820 |
+
"learning_rate": 2.967884934169392e-05,
|
| 15821 |
+
"loss": 1.9098,
|
| 15822 |
+
"step": 33885
|
| 15823 |
+
},
|
| 15824 |
+
{
|
| 15825 |
+
"epoch": 0.06601521270919422,
|
| 15826 |
+
"grad_norm": 2.567133903503418,
|
| 15827 |
+
"learning_rate": 2.9678565968913177e-05,
|
| 15828 |
+
"loss": 1.924,
|
| 15829 |
+
"step": 33900
|
| 15830 |
+
},
|
| 15831 |
+
{
|
| 15832 |
+
"epoch": 0.06604442298030448,
|
| 15833 |
+
"grad_norm": 2.739780902862549,
|
| 15834 |
+
"learning_rate": 2.967828247252217e-05,
|
| 15835 |
+
"loss": 1.9591,
|
| 15836 |
+
"step": 33915
|
| 15837 |
+
},
|
| 15838 |
+
{
|
| 15839 |
+
"epoch": 0.06607363325141476,
|
| 15840 |
+
"grad_norm": 3.8824591636657715,
|
| 15841 |
+
"learning_rate": 2.9677998852523277e-05,
|
| 15842 |
+
"loss": 1.8305,
|
| 15843 |
+
"step": 33930
|
| 15844 |
+
},
|
| 15845 |
+
{
|
| 15846 |
+
"epoch": 0.06610284352252502,
|
| 15847 |
+
"grad_norm": 3.137244939804077,
|
| 15848 |
+
"learning_rate": 2.967771510891889e-05,
|
| 15849 |
+
"loss": 1.9673,
|
| 15850 |
+
"step": 33945
|
| 15851 |
+
},
|
| 15852 |
+
{
|
| 15853 |
+
"epoch": 0.06613205379363528,
|
| 15854 |
+
"grad_norm": 2.5533084869384766,
|
| 15855 |
+
"learning_rate": 2.9677431241711405e-05,
|
| 15856 |
+
"loss": 1.7838,
|
| 15857 |
+
"step": 33960
|
| 15858 |
+
},
|
| 15859 |
+
{
|
| 15860 |
+
"epoch": 0.06616126406474554,
|
| 15861 |
+
"grad_norm": 3.2840499877929688,
|
| 15862 |
+
"learning_rate": 2.9677147250903203e-05,
|
| 15863 |
+
"loss": 1.7223,
|
| 15864 |
+
"step": 33975
|
| 15865 |
+
},
|
| 15866 |
+
{
|
| 15867 |
+
"epoch": 0.0661904743358558,
|
| 15868 |
+
"grad_norm": 3.5756568908691406,
|
| 15869 |
+
"learning_rate": 2.9676863136496685e-05,
|
| 15870 |
+
"loss": 2.0505,
|
| 15871 |
+
"step": 33990
|
| 15872 |
+
},
|
| 15873 |
+
{
|
| 15874 |
+
"epoch": 0.06621968460696606,
|
| 15875 |
+
"grad_norm": 2.9339518547058105,
|
| 15876 |
+
"learning_rate": 2.967657889849423e-05,
|
| 15877 |
+
"loss": 1.8248,
|
| 15878 |
+
"step": 34005
|
| 15879 |
+
},
|
| 15880 |
+
{
|
| 15881 |
+
"epoch": 0.06624889487807632,
|
| 15882 |
+
"grad_norm": 2.587409496307373,
|
| 15883 |
+
"learning_rate": 2.9676294536898247e-05,
|
| 15884 |
+
"loss": 2.1007,
|
| 15885 |
+
"step": 34020
|
| 15886 |
+
},
|
| 15887 |
+
{
|
| 15888 |
+
"epoch": 0.0662781051491866,
|
| 15889 |
+
"grad_norm": 2.164442777633667,
|
| 15890 |
+
"learning_rate": 2.9676010051711123e-05,
|
| 15891 |
+
"loss": 1.6494,
|
| 15892 |
+
"step": 34035
|
| 15893 |
+
},
|
| 15894 |
+
{
|
| 15895 |
+
"epoch": 0.06630731542029686,
|
| 15896 |
+
"grad_norm": 2.022341012954712,
|
| 15897 |
+
"learning_rate": 2.9675725442935252e-05,
|
| 15898 |
+
"loss": 2.01,
|
| 15899 |
+
"step": 34050
|
| 15900 |
+
},
|
| 15901 |
+
{
|
| 15902 |
+
"epoch": 0.06633652569140712,
|
| 15903 |
+
"grad_norm": 2.1963977813720703,
|
| 15904 |
+
"learning_rate": 2.9675440710573036e-05,
|
| 15905 |
+
"loss": 1.7275,
|
| 15906 |
+
"step": 34065
|
| 15907 |
+
},
|
| 15908 |
+
{
|
| 15909 |
+
"epoch": 0.06636573596251738,
|
| 15910 |
+
"grad_norm": 2.7345070838928223,
|
| 15911 |
+
"learning_rate": 2.967515585462687e-05,
|
| 15912 |
+
"loss": 1.6542,
|
| 15913 |
+
"step": 34080
|
| 15914 |
+
},
|
| 15915 |
+
{
|
| 15916 |
+
"epoch": 0.06639494623362764,
|
| 15917 |
+
"grad_norm": 2.9531948566436768,
|
| 15918 |
+
"learning_rate": 2.9674870875099144e-05,
|
| 15919 |
+
"loss": 1.9653,
|
| 15920 |
+
"step": 34095
|
| 15921 |
+
},
|
| 15922 |
+
{
|
| 15923 |
+
"epoch": 0.0664241565047379,
|
| 15924 |
+
"grad_norm": 2.839843988418579,
|
| 15925 |
+
"learning_rate": 2.9674585771992277e-05,
|
| 15926 |
+
"loss": 1.8574,
|
| 15927 |
+
"step": 34110
|
| 15928 |
+
},
|
| 15929 |
+
{
|
| 15930 |
+
"epoch": 0.06645336677584816,
|
| 15931 |
+
"grad_norm": 2.976874351501465,
|
| 15932 |
+
"learning_rate": 2.967430054530865e-05,
|
| 15933 |
+
"loss": 2.0622,
|
| 15934 |
+
"step": 34125
|
| 15935 |
+
},
|
| 15936 |
+
{
|
| 15937 |
+
"epoch": 0.06648257704695844,
|
| 15938 |
+
"grad_norm": 2.1137330532073975,
|
| 15939 |
+
"learning_rate": 2.967401519505068e-05,
|
| 15940 |
+
"loss": 1.8188,
|
| 15941 |
+
"step": 34140
|
| 15942 |
+
},
|
| 15943 |
+
{
|
| 15944 |
+
"epoch": 0.0665117873180687,
|
| 15945 |
+
"grad_norm": 2.7060563564300537,
|
| 15946 |
+
"learning_rate": 2.9673729721220765e-05,
|
| 15947 |
+
"loss": 1.9227,
|
| 15948 |
+
"step": 34155
|
| 15949 |
+
},
|
| 15950 |
+
{
|
| 15951 |
+
"epoch": 0.06654099758917896,
|
| 15952 |
+
"grad_norm": 2.3645975589752197,
|
| 15953 |
+
"learning_rate": 2.9673444123821306e-05,
|
| 15954 |
+
"loss": 1.8725,
|
| 15955 |
+
"step": 34170
|
| 15956 |
+
},
|
| 15957 |
+
{
|
| 15958 |
+
"epoch": 0.06657020786028922,
|
| 15959 |
+
"grad_norm": 2.382066011428833,
|
| 15960 |
+
"learning_rate": 2.967315840285471e-05,
|
| 15961 |
+
"loss": 1.7719,
|
| 15962 |
+
"step": 34185
|
| 15963 |
+
},
|
| 15964 |
+
{
|
| 15965 |
+
"epoch": 0.06659941813139948,
|
| 15966 |
+
"grad_norm": 1.7729606628417969,
|
| 15967 |
+
"learning_rate": 2.9672872558323385e-05,
|
| 15968 |
+
"loss": 1.8138,
|
| 15969 |
+
"step": 34200
|
| 15970 |
+
},
|
| 15971 |
+
{
|
| 15972 |
+
"epoch": 0.06662862840250974,
|
| 15973 |
+
"grad_norm": 2.7769312858581543,
|
| 15974 |
+
"learning_rate": 2.9672586590229735e-05,
|
| 15975 |
+
"loss": 1.8957,
|
| 15976 |
+
"step": 34215
|
| 15977 |
+
},
|
| 15978 |
+
{
|
| 15979 |
+
"epoch": 0.06665783867362,
|
| 15980 |
+
"grad_norm": 4.402248382568359,
|
| 15981 |
+
"learning_rate": 2.9672300498576173e-05,
|
| 15982 |
+
"loss": 1.9404,
|
| 15983 |
+
"step": 34230
|
| 15984 |
+
},
|
| 15985 |
+
{
|
| 15986 |
+
"epoch": 0.06668704894473028,
|
| 15987 |
+
"grad_norm": 4.229050159454346,
|
| 15988 |
+
"learning_rate": 2.96720142833651e-05,
|
| 15989 |
+
"loss": 1.8112,
|
| 15990 |
+
"step": 34245
|
| 15991 |
+
},
|
| 15992 |
+
{
|
| 15993 |
+
"epoch": 0.06671625921584054,
|
| 15994 |
+
"grad_norm": 4.353719711303711,
|
| 15995 |
+
"learning_rate": 2.9671727944598935e-05,
|
| 15996 |
+
"loss": 1.8216,
|
| 15997 |
+
"step": 34260
|
| 15998 |
+
},
|
| 15999 |
+
{
|
| 16000 |
+
"epoch": 0.0667454694869508,
|
| 16001 |
+
"grad_norm": 3.652221918106079,
|
| 16002 |
+
"learning_rate": 2.9671441482280083e-05,
|
| 16003 |
+
"loss": 1.9569,
|
| 16004 |
+
"step": 34275
|
| 16005 |
+
},
|
| 16006 |
+
{
|
| 16007 |
+
"epoch": 0.06677467975806106,
|
| 16008 |
+
"grad_norm": 2.7157819271087646,
|
| 16009 |
+
"learning_rate": 2.9671154896410962e-05,
|
| 16010 |
+
"loss": 1.7512,
|
| 16011 |
+
"step": 34290
|
| 16012 |
+
},
|
| 16013 |
+
{
|
| 16014 |
+
"epoch": 0.06680389002917132,
|
| 16015 |
+
"grad_norm": 2.719609498977661,
|
| 16016 |
+
"learning_rate": 2.9670868186993982e-05,
|
| 16017 |
+
"loss": 1.9252,
|
| 16018 |
+
"step": 34305
|
| 16019 |
+
},
|
| 16020 |
+
{
|
| 16021 |
+
"epoch": 0.06683310030028158,
|
| 16022 |
+
"grad_norm": 2.834977865219116,
|
| 16023 |
+
"learning_rate": 2.967058135403155e-05,
|
| 16024 |
+
"loss": 1.7615,
|
| 16025 |
+
"step": 34320
|
| 16026 |
+
},
|
| 16027 |
+
{
|
| 16028 |
+
"epoch": 0.06686231057139184,
|
| 16029 |
+
"grad_norm": 2.27976393699646,
|
| 16030 |
+
"learning_rate": 2.9670294397526097e-05,
|
| 16031 |
+
"loss": 1.7578,
|
| 16032 |
+
"step": 34335
|
| 16033 |
+
},
|
| 16034 |
+
{
|
| 16035 |
+
"epoch": 0.06689152084250212,
|
| 16036 |
+
"grad_norm": 1.7607027292251587,
|
| 16037 |
+
"learning_rate": 2.967000731748003e-05,
|
| 16038 |
+
"loss": 1.8367,
|
| 16039 |
+
"step": 34350
|
| 16040 |
+
},
|
| 16041 |
+
{
|
| 16042 |
+
"epoch": 0.06692073111361238,
|
| 16043 |
+
"grad_norm": 3.453352451324463,
|
| 16044 |
+
"learning_rate": 2.9669720113895763e-05,
|
| 16045 |
+
"loss": 1.8645,
|
| 16046 |
+
"step": 34365
|
| 16047 |
+
},
|
| 16048 |
+
{
|
| 16049 |
+
"epoch": 0.06694994138472264,
|
| 16050 |
+
"grad_norm": 2.4427194595336914,
|
| 16051 |
+
"learning_rate": 2.9669432786775727e-05,
|
| 16052 |
+
"loss": 1.8592,
|
| 16053 |
+
"step": 34380
|
| 16054 |
+
},
|
| 16055 |
+
{
|
| 16056 |
+
"epoch": 0.0669791516558329,
|
| 16057 |
+
"grad_norm": 3.492604970932007,
|
| 16058 |
+
"learning_rate": 2.9669145336122335e-05,
|
| 16059 |
+
"loss": 1.8394,
|
| 16060 |
+
"step": 34395
|
| 16061 |
+
},
|
| 16062 |
+
{
|
| 16063 |
+
"epoch": 0.06700836192694316,
|
| 16064 |
+
"grad_norm": 2.3918869495391846,
|
| 16065 |
+
"learning_rate": 2.9668857761938e-05,
|
| 16066 |
+
"loss": 1.7539,
|
| 16067 |
+
"step": 34410
|
| 16068 |
+
},
|
| 16069 |
+
{
|
| 16070 |
+
"epoch": 0.06703757219805342,
|
| 16071 |
+
"grad_norm": 1.826912760734558,
|
| 16072 |
+
"learning_rate": 2.9668570064225156e-05,
|
| 16073 |
+
"loss": 1.8747,
|
| 16074 |
+
"step": 34425
|
| 16075 |
+
},
|
| 16076 |
+
{
|
| 16077 |
+
"epoch": 0.06706678246916369,
|
| 16078 |
+
"grad_norm": 4.5764241218566895,
|
| 16079 |
+
"learning_rate": 2.966828224298622e-05,
|
| 16080 |
+
"loss": 1.8474,
|
| 16081 |
+
"step": 34440
|
| 16082 |
+
},
|
| 16083 |
+
{
|
| 16084 |
+
"epoch": 0.06709599274027396,
|
| 16085 |
+
"grad_norm": 3.9237124919891357,
|
| 16086 |
+
"learning_rate": 2.9667994298223612e-05,
|
| 16087 |
+
"loss": 1.8965,
|
| 16088 |
+
"step": 34455
|
| 16089 |
+
},
|
| 16090 |
+
{
|
| 16091 |
+
"epoch": 0.06712520301138422,
|
| 16092 |
+
"grad_norm": 3.064443826675415,
|
| 16093 |
+
"learning_rate": 2.9667706229939765e-05,
|
| 16094 |
+
"loss": 1.8549,
|
| 16095 |
+
"step": 34470
|
| 16096 |
+
},
|
| 16097 |
+
{
|
| 16098 |
+
"epoch": 0.06715441328249448,
|
| 16099 |
+
"grad_norm": 2.0611674785614014,
|
| 16100 |
+
"learning_rate": 2.96674180381371e-05,
|
| 16101 |
+
"loss": 1.8735,
|
| 16102 |
+
"step": 34485
|
| 16103 |
+
},
|
| 16104 |
+
{
|
| 16105 |
+
"epoch": 0.06718362355360474,
|
| 16106 |
+
"grad_norm": 3.887948513031006,
|
| 16107 |
+
"learning_rate": 2.9667129722818044e-05,
|
| 16108 |
+
"loss": 1.973,
|
| 16109 |
+
"step": 34500
|
| 16110 |
+
},
|
| 16111 |
+
{
|
| 16112 |
+
"epoch": 0.067212833824715,
|
| 16113 |
+
"grad_norm": 2.5983119010925293,
|
| 16114 |
+
"learning_rate": 2.966684128398503e-05,
|
| 16115 |
+
"loss": 1.8455,
|
| 16116 |
+
"step": 34515
|
| 16117 |
+
},
|
| 16118 |
+
{
|
| 16119 |
+
"epoch": 0.06724204409582527,
|
| 16120 |
+
"grad_norm": 3.6295909881591797,
|
| 16121 |
+
"learning_rate": 2.9666552721640474e-05,
|
| 16122 |
+
"loss": 1.738,
|
| 16123 |
+
"step": 34530
|
| 16124 |
+
},
|
| 16125 |
+
{
|
| 16126 |
+
"epoch": 0.06727125436693553,
|
| 16127 |
+
"grad_norm": 2.456125020980835,
|
| 16128 |
+
"learning_rate": 2.966626403578682e-05,
|
| 16129 |
+
"loss": 1.8651,
|
| 16130 |
+
"step": 34545
|
| 16131 |
+
},
|
| 16132 |
+
{
|
| 16133 |
+
"epoch": 0.06730046463804579,
|
| 16134 |
+
"grad_norm": 1.6152453422546387,
|
| 16135 |
+
"learning_rate": 2.966597522642649e-05,
|
| 16136 |
+
"loss": 1.8728,
|
| 16137 |
+
"step": 34560
|
| 16138 |
+
},
|
| 16139 |
+
{
|
| 16140 |
+
"epoch": 0.06732967490915606,
|
| 16141 |
+
"grad_norm": 2.789376974105835,
|
| 16142 |
+
"learning_rate": 2.966568629356193e-05,
|
| 16143 |
+
"loss": 1.8996,
|
| 16144 |
+
"step": 34575
|
| 16145 |
+
},
|
| 16146 |
+
{
|
| 16147 |
+
"epoch": 0.06735888518026632,
|
| 16148 |
+
"grad_norm": 3.9699041843414307,
|
| 16149 |
+
"learning_rate": 2.9665397237195555e-05,
|
| 16150 |
+
"loss": 1.7645,
|
| 16151 |
+
"step": 34590
|
| 16152 |
+
},
|
| 16153 |
+
{
|
| 16154 |
+
"epoch": 0.06738809545137658,
|
| 16155 |
+
"grad_norm": 2.7433955669403076,
|
| 16156 |
+
"learning_rate": 2.966510805732981e-05,
|
| 16157 |
+
"loss": 1.885,
|
| 16158 |
+
"step": 34605
|
| 16159 |
+
},
|
| 16160 |
+
{
|
| 16161 |
+
"epoch": 0.06741730572248684,
|
| 16162 |
+
"grad_norm": 3.254692316055298,
|
| 16163 |
+
"learning_rate": 2.9664818753967123e-05,
|
| 16164 |
+
"loss": 1.7664,
|
| 16165 |
+
"step": 34620
|
| 16166 |
+
},
|
| 16167 |
+
{
|
| 16168 |
+
"epoch": 0.0674465159935971,
|
| 16169 |
+
"grad_norm": 2.647752285003662,
|
| 16170 |
+
"learning_rate": 2.966452932710994e-05,
|
| 16171 |
+
"loss": 1.8037,
|
| 16172 |
+
"step": 34635
|
| 16173 |
+
},
|
| 16174 |
+
{
|
| 16175 |
+
"epoch": 0.06747572626470737,
|
| 16176 |
+
"grad_norm": 3.770737648010254,
|
| 16177 |
+
"learning_rate": 2.966423977676069e-05,
|
| 16178 |
+
"loss": 1.7053,
|
| 16179 |
+
"step": 34650
|
| 16180 |
+
},
|
| 16181 |
+
{
|
| 16182 |
+
"epoch": 0.06750493653581763,
|
| 16183 |
+
"grad_norm": 2.101461410522461,
|
| 16184 |
+
"learning_rate": 2.9663950102921814e-05,
|
| 16185 |
+
"loss": 1.8039,
|
| 16186 |
+
"step": 34665
|
| 16187 |
+
},
|
| 16188 |
+
{
|
| 16189 |
+
"epoch": 0.0675341468069279,
|
| 16190 |
+
"grad_norm": 4.246973514556885,
|
| 16191 |
+
"learning_rate": 2.9663660305595754e-05,
|
| 16192 |
+
"loss": 1.776,
|
| 16193 |
+
"step": 34680
|
| 16194 |
+
},
|
| 16195 |
+
{
|
| 16196 |
+
"epoch": 0.06756335707803816,
|
| 16197 |
+
"grad_norm": 6.2981157302856445,
|
| 16198 |
+
"learning_rate": 2.9663370384784946e-05,
|
| 16199 |
+
"loss": 1.7988,
|
| 16200 |
+
"step": 34695
|
| 16201 |
+
},
|
| 16202 |
+
{
|
| 16203 |
+
"epoch": 0.06759256734914842,
|
| 16204 |
+
"grad_norm": 2.8231287002563477,
|
| 16205 |
+
"learning_rate": 2.9663080340491838e-05,
|
| 16206 |
+
"loss": 1.9157,
|
| 16207 |
+
"step": 34710
|
| 16208 |
+
},
|
| 16209 |
+
{
|
| 16210 |
+
"epoch": 0.06762177762025869,
|
| 16211 |
+
"grad_norm": 2.1489720344543457,
|
| 16212 |
+
"learning_rate": 2.9662790172718867e-05,
|
| 16213 |
+
"loss": 1.7745,
|
| 16214 |
+
"step": 34725
|
| 16215 |
+
},
|
| 16216 |
+
{
|
| 16217 |
+
"epoch": 0.06765098789136895,
|
| 16218 |
+
"grad_norm": 2.714576482772827,
|
| 16219 |
+
"learning_rate": 2.9662499881468475e-05,
|
| 16220 |
+
"loss": 1.8596,
|
| 16221 |
+
"step": 34740
|
| 16222 |
+
},
|
| 16223 |
+
{
|
| 16224 |
+
"epoch": 0.06768019816247921,
|
| 16225 |
+
"grad_norm": 2.1026928424835205,
|
| 16226 |
+
"learning_rate": 2.966220946674311e-05,
|
| 16227 |
+
"loss": 1.9096,
|
| 16228 |
+
"step": 34755
|
| 16229 |
+
},
|
| 16230 |
+
{
|
| 16231 |
+
"epoch": 0.06770940843358947,
|
| 16232 |
+
"grad_norm": 4.316662311553955,
|
| 16233 |
+
"learning_rate": 2.9661918928545215e-05,
|
| 16234 |
+
"loss": 1.8735,
|
| 16235 |
+
"step": 34770
|
| 16236 |
+
},
|
| 16237 |
+
{
|
| 16238 |
+
"epoch": 0.06773861870469974,
|
| 16239 |
+
"grad_norm": 3.814514636993408,
|
| 16240 |
+
"learning_rate": 2.966162826687724e-05,
|
| 16241 |
+
"loss": 1.9701,
|
| 16242 |
+
"step": 34785
|
| 16243 |
+
},
|
| 16244 |
+
{
|
| 16245 |
+
"epoch": 0.06776782897581,
|
| 16246 |
+
"grad_norm": 4.129803657531738,
|
| 16247 |
+
"learning_rate": 2.966133748174164e-05,
|
| 16248 |
+
"loss": 1.9273,
|
| 16249 |
+
"step": 34800
|
| 16250 |
+
},
|
| 16251 |
+
{
|
| 16252 |
+
"epoch": 0.06779703924692027,
|
| 16253 |
+
"grad_norm": 1.9344481229782104,
|
| 16254 |
+
"learning_rate": 2.966104657314084e-05,
|
| 16255 |
+
"loss": 1.8738,
|
| 16256 |
+
"step": 34815
|
| 16257 |
+
},
|
| 16258 |
+
{
|
| 16259 |
+
"epoch": 0.06782624951803053,
|
| 16260 |
+
"grad_norm": 5.156272888183594,
|
| 16261 |
+
"learning_rate": 2.9660755541077314e-05,
|
| 16262 |
+
"loss": 1.9722,
|
| 16263 |
+
"step": 34830
|
| 16264 |
+
},
|
| 16265 |
+
{
|
| 16266 |
+
"epoch": 0.06785545978914079,
|
| 16267 |
+
"grad_norm": 2.4508838653564453,
|
| 16268 |
+
"learning_rate": 2.9660464385553504e-05,
|
| 16269 |
+
"loss": 1.8354,
|
| 16270 |
+
"step": 34845
|
| 16271 |
+
},
|
| 16272 |
+
{
|
| 16273 |
+
"epoch": 0.06788467006025105,
|
| 16274 |
+
"grad_norm": 4.646978855133057,
|
| 16275 |
+
"learning_rate": 2.9660173106571856e-05,
|
| 16276 |
+
"loss": 1.6924,
|
| 16277 |
+
"step": 34860
|
| 16278 |
+
},
|
| 16279 |
+
{
|
| 16280 |
+
"epoch": 0.06791388033136131,
|
| 16281 |
+
"grad_norm": 4.116873741149902,
|
| 16282 |
+
"learning_rate": 2.9659881704134835e-05,
|
| 16283 |
+
"loss": 1.9238,
|
| 16284 |
+
"step": 34875
|
| 16285 |
+
},
|
| 16286 |
+
{
|
| 16287 |
+
"epoch": 0.06794309060247158,
|
| 16288 |
+
"grad_norm": 2.2530601024627686,
|
| 16289 |
+
"learning_rate": 2.965959017824488e-05,
|
| 16290 |
+
"loss": 1.744,
|
| 16291 |
+
"step": 34890
|
| 16292 |
+
},
|
| 16293 |
+
{
|
| 16294 |
+
"epoch": 0.06797230087358185,
|
| 16295 |
+
"grad_norm": 3.1118083000183105,
|
| 16296 |
+
"learning_rate": 2.9659298528904462e-05,
|
| 16297 |
+
"loss": 1.7291,
|
| 16298 |
+
"step": 34905
|
| 16299 |
+
},
|
| 16300 |
+
{
|
| 16301 |
+
"epoch": 0.0680015111446921,
|
| 16302 |
+
"grad_norm": 4.10203218460083,
|
| 16303 |
+
"learning_rate": 2.9659006756116024e-05,
|
| 16304 |
+
"loss": 1.9676,
|
| 16305 |
+
"step": 34920
|
| 16306 |
+
},
|
| 16307 |
+
{
|
| 16308 |
+
"epoch": 0.06803072141580237,
|
| 16309 |
+
"grad_norm": 2.487515926361084,
|
| 16310 |
+
"learning_rate": 2.9658714859882033e-05,
|
| 16311 |
+
"loss": 1.755,
|
| 16312 |
+
"step": 34935
|
| 16313 |
+
},
|
| 16314 |
+
{
|
| 16315 |
+
"epoch": 0.06805993168691263,
|
| 16316 |
+
"grad_norm": 4.722667694091797,
|
| 16317 |
+
"learning_rate": 2.965842284020494e-05,
|
| 16318 |
+
"loss": 1.6132,
|
| 16319 |
+
"step": 34950
|
| 16320 |
+
},
|
| 16321 |
+
{
|
| 16322 |
+
"epoch": 0.06808914195802289,
|
| 16323 |
+
"grad_norm": 3.8074474334716797,
|
| 16324 |
+
"learning_rate": 2.9658130697087206e-05,
|
| 16325 |
+
"loss": 1.8949,
|
| 16326 |
+
"step": 34965
|
| 16327 |
+
},
|
| 16328 |
+
{
|
| 16329 |
+
"epoch": 0.06811835222913315,
|
| 16330 |
+
"grad_norm": 3.2412679195404053,
|
| 16331 |
+
"learning_rate": 2.9657838430531295e-05,
|
| 16332 |
+
"loss": 1.8062,
|
| 16333 |
+
"step": 34980
|
| 16334 |
+
},
|
| 16335 |
+
{
|
| 16336 |
+
"epoch": 0.06814756250024342,
|
| 16337 |
+
"grad_norm": 3.2179923057556152,
|
| 16338 |
+
"learning_rate": 2.9657546040539662e-05,
|
| 16339 |
+
"loss": 1.7753,
|
| 16340 |
+
"step": 34995
|
| 16341 |
}
|
| 16342 |
],
|
| 16343 |
"logging_steps": 15,
|
|
|
|
| 16357 |
"attributes": {}
|
| 16358 |
}
|
| 16359 |
},
|
| 16360 |
+
"total_flos": 5.536910190624768e+16,
|
| 16361 |
"train_batch_size": 4,
|
| 16362 |
"trial_name": null,
|
| 16363 |
"trial_params": null
|