Upload folder using huggingface_hub
Browse files- optimizer.pt +1 -1
- pytorch_model.bin +1 -1
- rng_state.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +1402 -2
optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1909053417
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1da777fdc2f589e2fd22bba7e2fef4a2ed107c1214679527001ae57564ee62c4
|
| 3 |
size 1909053417
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 969281034
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5c9fd6559fcb670e2401c547414ff82836bdc646bb72642a2eec3867cef32b65
|
| 3 |
size 969281034
|
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:f85273902d9a34de6c102fe53d722bd3322efbe0f1ea96148f22d32665f5895c
|
| 3 |
size 14244
|
scheduler.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1064
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9b4faebb8a687e5379473d389ae7f00051e86251b0e805e330f53a52ba86cf88
|
| 3 |
size 1064
|
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,
|
|
@@ -9807,6 +9807,1406 @@
|
|
| 9807 |
"learning_rate": 9.99474835923667e-06,
|
| 9808 |
"loss": 117.067,
|
| 9809 |
"step": 14000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 9810 |
}
|
| 9811 |
],
|
| 9812 |
"logging_steps": 10,
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.13235719899077636,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 16000,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 9807 |
"learning_rate": 9.99474835923667e-06,
|
| 9808 |
"loss": 117.067,
|
| 9809 |
"step": 14000
|
| 9810 |
+
},
|
| 9811 |
+
{
|
| 9812 |
+
"epoch": 0.11589527236629855,
|
| 9813 |
+
"grad_norm": 870.8054809570312,
|
| 9814 |
+
"learning_rate": 9.994683536271437e-06,
|
| 9815 |
+
"loss": 177.513,
|
| 9816 |
+
"step": 14010
|
| 9817 |
+
},
|
| 9818 |
+
{
|
| 9819 |
+
"epoch": 0.11597799561566778,
|
| 9820 |
+
"grad_norm": 1938.4117431640625,
|
| 9821 |
+
"learning_rate": 9.994618315902161e-06,
|
| 9822 |
+
"loss": 147.8295,
|
| 9823 |
+
"step": 14020
|
| 9824 |
+
},
|
| 9825 |
+
{
|
| 9826 |
+
"epoch": 0.11606071886503702,
|
| 9827 |
+
"grad_norm": 1145.3619384765625,
|
| 9828 |
+
"learning_rate": 9.994552698134023e-06,
|
| 9829 |
+
"loss": 126.2492,
|
| 9830 |
+
"step": 14030
|
| 9831 |
+
},
|
| 9832 |
+
{
|
| 9833 |
+
"epoch": 0.11614344211440625,
|
| 9834 |
+
"grad_norm": 704.4757080078125,
|
| 9835 |
+
"learning_rate": 9.994486682972253e-06,
|
| 9836 |
+
"loss": 183.3489,
|
| 9837 |
+
"step": 14040
|
| 9838 |
+
},
|
| 9839 |
+
{
|
| 9840 |
+
"epoch": 0.11622616536377549,
|
| 9841 |
+
"grad_norm": 799.8057250976562,
|
| 9842 |
+
"learning_rate": 9.994420270422096e-06,
|
| 9843 |
+
"loss": 155.8286,
|
| 9844 |
+
"step": 14050
|
| 9845 |
+
},
|
| 9846 |
+
{
|
| 9847 |
+
"epoch": 0.11630888861314473,
|
| 9848 |
+
"grad_norm": 2168.39794921875,
|
| 9849 |
+
"learning_rate": 9.994353460488842e-06,
|
| 9850 |
+
"loss": 165.6206,
|
| 9851 |
+
"step": 14060
|
| 9852 |
+
},
|
| 9853 |
+
{
|
| 9854 |
+
"epoch": 0.11639161186251396,
|
| 9855 |
+
"grad_norm": 1048.3438720703125,
|
| 9856 |
+
"learning_rate": 9.994286253177803e-06,
|
| 9857 |
+
"loss": 196.4472,
|
| 9858 |
+
"step": 14070
|
| 9859 |
+
},
|
| 9860 |
+
{
|
| 9861 |
+
"epoch": 0.1164743351118832,
|
| 9862 |
+
"grad_norm": 1240.358642578125,
|
| 9863 |
+
"learning_rate": 9.994218648494327e-06,
|
| 9864 |
+
"loss": 169.1644,
|
| 9865 |
+
"step": 14080
|
| 9866 |
+
},
|
| 9867 |
+
{
|
| 9868 |
+
"epoch": 0.11655705836125244,
|
| 9869 |
+
"grad_norm": 1450.99169921875,
|
| 9870 |
+
"learning_rate": 9.994150646443793e-06,
|
| 9871 |
+
"loss": 119.286,
|
| 9872 |
+
"step": 14090
|
| 9873 |
+
},
|
| 9874 |
+
{
|
| 9875 |
+
"epoch": 0.11663978161062166,
|
| 9876 |
+
"grad_norm": 1026.7149658203125,
|
| 9877 |
+
"learning_rate": 9.994082247031613e-06,
|
| 9878 |
+
"loss": 166.7578,
|
| 9879 |
+
"step": 14100
|
| 9880 |
+
},
|
| 9881 |
+
{
|
| 9882 |
+
"epoch": 0.1167225048599909,
|
| 9883 |
+
"grad_norm": 1081.6656494140625,
|
| 9884 |
+
"learning_rate": 9.99401345026323e-06,
|
| 9885 |
+
"loss": 141.0757,
|
| 9886 |
+
"step": 14110
|
| 9887 |
+
},
|
| 9888 |
+
{
|
| 9889 |
+
"epoch": 0.11680522810936013,
|
| 9890 |
+
"grad_norm": 746.9979248046875,
|
| 9891 |
+
"learning_rate": 9.993944256144115e-06,
|
| 9892 |
+
"loss": 124.9759,
|
| 9893 |
+
"step": 14120
|
| 9894 |
+
},
|
| 9895 |
+
{
|
| 9896 |
+
"epoch": 0.11688795135872937,
|
| 9897 |
+
"grad_norm": 891.9210815429688,
|
| 9898 |
+
"learning_rate": 9.993874664679774e-06,
|
| 9899 |
+
"loss": 150.3685,
|
| 9900 |
+
"step": 14130
|
| 9901 |
+
},
|
| 9902 |
+
{
|
| 9903 |
+
"epoch": 0.11697067460809861,
|
| 9904 |
+
"grad_norm": 1401.4002685546875,
|
| 9905 |
+
"learning_rate": 9.993804675875744e-06,
|
| 9906 |
+
"loss": 168.8493,
|
| 9907 |
+
"step": 14140
|
| 9908 |
+
},
|
| 9909 |
+
{
|
| 9910 |
+
"epoch": 0.11705339785746784,
|
| 9911 |
+
"grad_norm": 1588.7640380859375,
|
| 9912 |
+
"learning_rate": 9.993734289737596e-06,
|
| 9913 |
+
"loss": 141.4464,
|
| 9914 |
+
"step": 14150
|
| 9915 |
+
},
|
| 9916 |
+
{
|
| 9917 |
+
"epoch": 0.11713612110683708,
|
| 9918 |
+
"grad_norm": 1859.66552734375,
|
| 9919 |
+
"learning_rate": 9.993663506270928e-06,
|
| 9920 |
+
"loss": 162.024,
|
| 9921 |
+
"step": 14160
|
| 9922 |
+
},
|
| 9923 |
+
{
|
| 9924 |
+
"epoch": 0.11721884435620632,
|
| 9925 |
+
"grad_norm": 1133.1839599609375,
|
| 9926 |
+
"learning_rate": 9.993592325481373e-06,
|
| 9927 |
+
"loss": 166.6096,
|
| 9928 |
+
"step": 14170
|
| 9929 |
+
},
|
| 9930 |
+
{
|
| 9931 |
+
"epoch": 0.11730156760557554,
|
| 9932 |
+
"grad_norm": 1811.849365234375,
|
| 9933 |
+
"learning_rate": 9.993520747374594e-06,
|
| 9934 |
+
"loss": 127.2197,
|
| 9935 |
+
"step": 14180
|
| 9936 |
+
},
|
| 9937 |
+
{
|
| 9938 |
+
"epoch": 0.11738429085494478,
|
| 9939 |
+
"grad_norm": 909.2362060546875,
|
| 9940 |
+
"learning_rate": 9.993448771956285e-06,
|
| 9941 |
+
"loss": 189.4919,
|
| 9942 |
+
"step": 14190
|
| 9943 |
+
},
|
| 9944 |
+
{
|
| 9945 |
+
"epoch": 0.11746701410431402,
|
| 9946 |
+
"grad_norm": 1350.44140625,
|
| 9947 |
+
"learning_rate": 9.993376399232175e-06,
|
| 9948 |
+
"loss": 142.4382,
|
| 9949 |
+
"step": 14200
|
| 9950 |
+
},
|
| 9951 |
+
{
|
| 9952 |
+
"epoch": 0.11754973735368325,
|
| 9953 |
+
"grad_norm": 1765.2679443359375,
|
| 9954 |
+
"learning_rate": 9.993303629208023e-06,
|
| 9955 |
+
"loss": 148.8411,
|
| 9956 |
+
"step": 14210
|
| 9957 |
+
},
|
| 9958 |
+
{
|
| 9959 |
+
"epoch": 0.11763246060305249,
|
| 9960 |
+
"grad_norm": 2343.818359375,
|
| 9961 |
+
"learning_rate": 9.993230461889616e-06,
|
| 9962 |
+
"loss": 212.7168,
|
| 9963 |
+
"step": 14220
|
| 9964 |
+
},
|
| 9965 |
+
{
|
| 9966 |
+
"epoch": 0.11771518385242172,
|
| 9967 |
+
"grad_norm": 683.275146484375,
|
| 9968 |
+
"learning_rate": 9.993156897282776e-06,
|
| 9969 |
+
"loss": 148.4446,
|
| 9970 |
+
"step": 14230
|
| 9971 |
+
},
|
| 9972 |
+
{
|
| 9973 |
+
"epoch": 0.11779790710179096,
|
| 9974 |
+
"grad_norm": 998.6349487304688,
|
| 9975 |
+
"learning_rate": 9.99308293539336e-06,
|
| 9976 |
+
"loss": 117.4103,
|
| 9977 |
+
"step": 14240
|
| 9978 |
+
},
|
| 9979 |
+
{
|
| 9980 |
+
"epoch": 0.1178806303511602,
|
| 9981 |
+
"grad_norm": 823.5191040039062,
|
| 9982 |
+
"learning_rate": 9.993008576227248e-06,
|
| 9983 |
+
"loss": 130.8048,
|
| 9984 |
+
"step": 14250
|
| 9985 |
+
},
|
| 9986 |
+
{
|
| 9987 |
+
"epoch": 0.11796335360052942,
|
| 9988 |
+
"grad_norm": 1420.8035888671875,
|
| 9989 |
+
"learning_rate": 9.992933819790358e-06,
|
| 9990 |
+
"loss": 163.5295,
|
| 9991 |
+
"step": 14260
|
| 9992 |
+
},
|
| 9993 |
+
{
|
| 9994 |
+
"epoch": 0.11804607684989867,
|
| 9995 |
+
"grad_norm": 1167.037109375,
|
| 9996 |
+
"learning_rate": 9.992858666088638e-06,
|
| 9997 |
+
"loss": 164.6194,
|
| 9998 |
+
"step": 14270
|
| 9999 |
+
},
|
| 10000 |
+
{
|
| 10001 |
+
"epoch": 0.1181288000992679,
|
| 10002 |
+
"grad_norm": 1515.985107421875,
|
| 10003 |
+
"learning_rate": 9.992783115128072e-06,
|
| 10004 |
+
"loss": 163.406,
|
| 10005 |
+
"step": 14280
|
| 10006 |
+
},
|
| 10007 |
+
{
|
| 10008 |
+
"epoch": 0.11821152334863713,
|
| 10009 |
+
"grad_norm": 1549.8917236328125,
|
| 10010 |
+
"learning_rate": 9.992707166914662e-06,
|
| 10011 |
+
"loss": 168.7726,
|
| 10012 |
+
"step": 14290
|
| 10013 |
+
},
|
| 10014 |
+
{
|
| 10015 |
+
"epoch": 0.11829424659800637,
|
| 10016 |
+
"grad_norm": 1190.7861328125,
|
| 10017 |
+
"learning_rate": 9.992630821454458e-06,
|
| 10018 |
+
"loss": 140.9276,
|
| 10019 |
+
"step": 14300
|
| 10020 |
+
},
|
| 10021 |
+
{
|
| 10022 |
+
"epoch": 0.1183769698473756,
|
| 10023 |
+
"grad_norm": 1568.7037353515625,
|
| 10024 |
+
"learning_rate": 9.992554078753534e-06,
|
| 10025 |
+
"loss": 147.5554,
|
| 10026 |
+
"step": 14310
|
| 10027 |
+
},
|
| 10028 |
+
{
|
| 10029 |
+
"epoch": 0.11845969309674484,
|
| 10030 |
+
"grad_norm": 1000.02880859375,
|
| 10031 |
+
"learning_rate": 9.992476938817994e-06,
|
| 10032 |
+
"loss": 180.1213,
|
| 10033 |
+
"step": 14320
|
| 10034 |
+
},
|
| 10035 |
+
{
|
| 10036 |
+
"epoch": 0.11854241634611408,
|
| 10037 |
+
"grad_norm": 1296.5947265625,
|
| 10038 |
+
"learning_rate": 9.992399401653976e-06,
|
| 10039 |
+
"loss": 137.781,
|
| 10040 |
+
"step": 14330
|
| 10041 |
+
},
|
| 10042 |
+
{
|
| 10043 |
+
"epoch": 0.1186251395954833,
|
| 10044 |
+
"grad_norm": 1144.0504150390625,
|
| 10045 |
+
"learning_rate": 9.99232146726765e-06,
|
| 10046 |
+
"loss": 140.0204,
|
| 10047 |
+
"step": 14340
|
| 10048 |
+
},
|
| 10049 |
+
{
|
| 10050 |
+
"epoch": 0.11870786284485255,
|
| 10051 |
+
"grad_norm": 1281.876708984375,
|
| 10052 |
+
"learning_rate": 9.992243135665217e-06,
|
| 10053 |
+
"loss": 154.8919,
|
| 10054 |
+
"step": 14350
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 0.11879058609422179,
|
| 10058 |
+
"grad_norm": 871.8610229492188,
|
| 10059 |
+
"learning_rate": 9.992164406852908e-06,
|
| 10060 |
+
"loss": 186.1516,
|
| 10061 |
+
"step": 14360
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 0.11887330934359101,
|
| 10065 |
+
"grad_norm": 1534.3536376953125,
|
| 10066 |
+
"learning_rate": 9.992085280836988e-06,
|
| 10067 |
+
"loss": 160.6092,
|
| 10068 |
+
"step": 14370
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 0.11895603259296025,
|
| 10072 |
+
"grad_norm": 988.0948486328125,
|
| 10073 |
+
"learning_rate": 9.992005757623753e-06,
|
| 10074 |
+
"loss": 203.4977,
|
| 10075 |
+
"step": 14380
|
| 10076 |
+
},
|
| 10077 |
+
{
|
| 10078 |
+
"epoch": 0.11903875584232948,
|
| 10079 |
+
"grad_norm": 1357.1668701171875,
|
| 10080 |
+
"learning_rate": 9.991925837219532e-06,
|
| 10081 |
+
"loss": 160.5042,
|
| 10082 |
+
"step": 14390
|
| 10083 |
+
},
|
| 10084 |
+
{
|
| 10085 |
+
"epoch": 0.11912147909169872,
|
| 10086 |
+
"grad_norm": 636.4329223632812,
|
| 10087 |
+
"learning_rate": 9.991845519630679e-06,
|
| 10088 |
+
"loss": 137.9073,
|
| 10089 |
+
"step": 14400
|
| 10090 |
+
},
|
| 10091 |
+
{
|
| 10092 |
+
"epoch": 0.11920420234106796,
|
| 10093 |
+
"grad_norm": 1027.252197265625,
|
| 10094 |
+
"learning_rate": 9.991764804863588e-06,
|
| 10095 |
+
"loss": 122.9028,
|
| 10096 |
+
"step": 14410
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 0.11928692559043719,
|
| 10100 |
+
"grad_norm": 2356.058837890625,
|
| 10101 |
+
"learning_rate": 9.991683692924682e-06,
|
| 10102 |
+
"loss": 155.1582,
|
| 10103 |
+
"step": 14420
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 0.11936964883980643,
|
| 10107 |
+
"grad_norm": 743.142578125,
|
| 10108 |
+
"learning_rate": 9.991602183820412e-06,
|
| 10109 |
+
"loss": 142.9998,
|
| 10110 |
+
"step": 14430
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 0.11945237208917567,
|
| 10114 |
+
"grad_norm": 1039.77978515625,
|
| 10115 |
+
"learning_rate": 9.991520277557266e-06,
|
| 10116 |
+
"loss": 156.9646,
|
| 10117 |
+
"step": 14440
|
| 10118 |
+
},
|
| 10119 |
+
{
|
| 10120 |
+
"epoch": 0.1195350953385449,
|
| 10121 |
+
"grad_norm": 1624.250732421875,
|
| 10122 |
+
"learning_rate": 9.991437974141759e-06,
|
| 10123 |
+
"loss": 165.5059,
|
| 10124 |
+
"step": 14450
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 0.11961781858791413,
|
| 10128 |
+
"grad_norm": 589.4429321289062,
|
| 10129 |
+
"learning_rate": 9.99135527358044e-06,
|
| 10130 |
+
"loss": 119.63,
|
| 10131 |
+
"step": 14460
|
| 10132 |
+
},
|
| 10133 |
+
{
|
| 10134 |
+
"epoch": 0.11970054183728338,
|
| 10135 |
+
"grad_norm": 1199.8302001953125,
|
| 10136 |
+
"learning_rate": 9.991272175879888e-06,
|
| 10137 |
+
"loss": 175.7935,
|
| 10138 |
+
"step": 14470
|
| 10139 |
+
},
|
| 10140 |
+
{
|
| 10141 |
+
"epoch": 0.1197832650866526,
|
| 10142 |
+
"grad_norm": 1233.6771240234375,
|
| 10143 |
+
"learning_rate": 9.991188681046718e-06,
|
| 10144 |
+
"loss": 192.3081,
|
| 10145 |
+
"step": 14480
|
| 10146 |
+
},
|
| 10147 |
+
{
|
| 10148 |
+
"epoch": 0.11986598833602184,
|
| 10149 |
+
"grad_norm": 1209.5980224609375,
|
| 10150 |
+
"learning_rate": 9.991104789087568e-06,
|
| 10151 |
+
"loss": 139.1357,
|
| 10152 |
+
"step": 14490
|
| 10153 |
+
},
|
| 10154 |
+
{
|
| 10155 |
+
"epoch": 0.11994871158539107,
|
| 10156 |
+
"grad_norm": 1937.0008544921875,
|
| 10157 |
+
"learning_rate": 9.991020500009118e-06,
|
| 10158 |
+
"loss": 138.8762,
|
| 10159 |
+
"step": 14500
|
| 10160 |
+
},
|
| 10161 |
+
{
|
| 10162 |
+
"epoch": 0.12003143483476031,
|
| 10163 |
+
"grad_norm": 545.6674194335938,
|
| 10164 |
+
"learning_rate": 9.990935813818073e-06,
|
| 10165 |
+
"loss": 139.6014,
|
| 10166 |
+
"step": 14510
|
| 10167 |
+
},
|
| 10168 |
+
{
|
| 10169 |
+
"epoch": 0.12011415808412955,
|
| 10170 |
+
"grad_norm": 650.5000610351562,
|
| 10171 |
+
"learning_rate": 9.99085073052117e-06,
|
| 10172 |
+
"loss": 195.5592,
|
| 10173 |
+
"step": 14520
|
| 10174 |
+
},
|
| 10175 |
+
{
|
| 10176 |
+
"epoch": 0.12019688133349878,
|
| 10177 |
+
"grad_norm": 566.500732421875,
|
| 10178 |
+
"learning_rate": 9.990765250125179e-06,
|
| 10179 |
+
"loss": 154.2413,
|
| 10180 |
+
"step": 14530
|
| 10181 |
+
},
|
| 10182 |
+
{
|
| 10183 |
+
"epoch": 0.12027960458286802,
|
| 10184 |
+
"grad_norm": 1750.378662109375,
|
| 10185 |
+
"learning_rate": 9.990679372636902e-06,
|
| 10186 |
+
"loss": 161.8778,
|
| 10187 |
+
"step": 14540
|
| 10188 |
+
},
|
| 10189 |
+
{
|
| 10190 |
+
"epoch": 0.12036232783223726,
|
| 10191 |
+
"grad_norm": 1349.2432861328125,
|
| 10192 |
+
"learning_rate": 9.99059309806317e-06,
|
| 10193 |
+
"loss": 172.9573,
|
| 10194 |
+
"step": 14550
|
| 10195 |
+
},
|
| 10196 |
+
{
|
| 10197 |
+
"epoch": 0.12044505108160648,
|
| 10198 |
+
"grad_norm": 975.1179809570312,
|
| 10199 |
+
"learning_rate": 9.990506426410851e-06,
|
| 10200 |
+
"loss": 128.1013,
|
| 10201 |
+
"step": 14560
|
| 10202 |
+
},
|
| 10203 |
+
{
|
| 10204 |
+
"epoch": 0.12052777433097572,
|
| 10205 |
+
"grad_norm": 3406.892333984375,
|
| 10206 |
+
"learning_rate": 9.990419357686839e-06,
|
| 10207 |
+
"loss": 177.5993,
|
| 10208 |
+
"step": 14570
|
| 10209 |
+
},
|
| 10210 |
+
{
|
| 10211 |
+
"epoch": 0.12061049758034495,
|
| 10212 |
+
"grad_norm": 428.1646423339844,
|
| 10213 |
+
"learning_rate": 9.99033189189806e-06,
|
| 10214 |
+
"loss": 128.9519,
|
| 10215 |
+
"step": 14580
|
| 10216 |
+
},
|
| 10217 |
+
{
|
| 10218 |
+
"epoch": 0.12069322082971419,
|
| 10219 |
+
"grad_norm": 996.7293701171875,
|
| 10220 |
+
"learning_rate": 9.990244029051475e-06,
|
| 10221 |
+
"loss": 149.3833,
|
| 10222 |
+
"step": 14590
|
| 10223 |
+
},
|
| 10224 |
+
{
|
| 10225 |
+
"epoch": 0.12077594407908343,
|
| 10226 |
+
"grad_norm": 1458.9307861328125,
|
| 10227 |
+
"learning_rate": 9.990155769154077e-06,
|
| 10228 |
+
"loss": 160.3518,
|
| 10229 |
+
"step": 14600
|
| 10230 |
+
},
|
| 10231 |
+
{
|
| 10232 |
+
"epoch": 0.12085866732845266,
|
| 10233 |
+
"grad_norm": 714.3155517578125,
|
| 10234 |
+
"learning_rate": 9.990067112212884e-06,
|
| 10235 |
+
"loss": 128.5733,
|
| 10236 |
+
"step": 14610
|
| 10237 |
+
},
|
| 10238 |
+
{
|
| 10239 |
+
"epoch": 0.1209413905778219,
|
| 10240 |
+
"grad_norm": 629.7801513671875,
|
| 10241 |
+
"learning_rate": 9.989978058234952e-06,
|
| 10242 |
+
"loss": 141.1231,
|
| 10243 |
+
"step": 14620
|
| 10244 |
+
},
|
| 10245 |
+
{
|
| 10246 |
+
"epoch": 0.12102411382719114,
|
| 10247 |
+
"grad_norm": 1056.1544189453125,
|
| 10248 |
+
"learning_rate": 9.989888607227369e-06,
|
| 10249 |
+
"loss": 173.8705,
|
| 10250 |
+
"step": 14630
|
| 10251 |
+
},
|
| 10252 |
+
{
|
| 10253 |
+
"epoch": 0.12110683707656036,
|
| 10254 |
+
"grad_norm": 1272.472412109375,
|
| 10255 |
+
"learning_rate": 9.989798759197247e-06,
|
| 10256 |
+
"loss": 146.9385,
|
| 10257 |
+
"step": 14640
|
| 10258 |
+
},
|
| 10259 |
+
{
|
| 10260 |
+
"epoch": 0.1211895603259296,
|
| 10261 |
+
"grad_norm": 1054.3629150390625,
|
| 10262 |
+
"learning_rate": 9.989708514151739e-06,
|
| 10263 |
+
"loss": 164.1719,
|
| 10264 |
+
"step": 14650
|
| 10265 |
+
},
|
| 10266 |
+
{
|
| 10267 |
+
"epoch": 0.12127228357529883,
|
| 10268 |
+
"grad_norm": 659.7613525390625,
|
| 10269 |
+
"learning_rate": 9.989617872098026e-06,
|
| 10270 |
+
"loss": 149.6539,
|
| 10271 |
+
"step": 14660
|
| 10272 |
+
},
|
| 10273 |
+
{
|
| 10274 |
+
"epoch": 0.12135500682466807,
|
| 10275 |
+
"grad_norm": 815.8479614257812,
|
| 10276 |
+
"learning_rate": 9.989526833043316e-06,
|
| 10277 |
+
"loss": 140.1702,
|
| 10278 |
+
"step": 14670
|
| 10279 |
+
},
|
| 10280 |
+
{
|
| 10281 |
+
"epoch": 0.12143773007403731,
|
| 10282 |
+
"grad_norm": 1148.9129638671875,
|
| 10283 |
+
"learning_rate": 9.989435396994856e-06,
|
| 10284 |
+
"loss": 125.2471,
|
| 10285 |
+
"step": 14680
|
| 10286 |
+
},
|
| 10287 |
+
{
|
| 10288 |
+
"epoch": 0.12152045332340654,
|
| 10289 |
+
"grad_norm": 3006.224609375,
|
| 10290 |
+
"learning_rate": 9.989343563959919e-06,
|
| 10291 |
+
"loss": 150.3076,
|
| 10292 |
+
"step": 14690
|
| 10293 |
+
},
|
| 10294 |
+
{
|
| 10295 |
+
"epoch": 0.12160317657277578,
|
| 10296 |
+
"grad_norm": 1365.89892578125,
|
| 10297 |
+
"learning_rate": 9.989251333945813e-06,
|
| 10298 |
+
"loss": 179.2145,
|
| 10299 |
+
"step": 14700
|
| 10300 |
+
},
|
| 10301 |
+
{
|
| 10302 |
+
"epoch": 0.12168589982214502,
|
| 10303 |
+
"grad_norm": 982.8682861328125,
|
| 10304 |
+
"learning_rate": 9.989158706959875e-06,
|
| 10305 |
+
"loss": 137.9394,
|
| 10306 |
+
"step": 14710
|
| 10307 |
+
},
|
| 10308 |
+
{
|
| 10309 |
+
"epoch": 0.12176862307151425,
|
| 10310 |
+
"grad_norm": 1103.151123046875,
|
| 10311 |
+
"learning_rate": 9.989065683009477e-06,
|
| 10312 |
+
"loss": 150.3043,
|
| 10313 |
+
"step": 14720
|
| 10314 |
+
},
|
| 10315 |
+
{
|
| 10316 |
+
"epoch": 0.12185134632088349,
|
| 10317 |
+
"grad_norm": 1068.439208984375,
|
| 10318 |
+
"learning_rate": 9.988972262102018e-06,
|
| 10319 |
+
"loss": 115.0475,
|
| 10320 |
+
"step": 14730
|
| 10321 |
+
},
|
| 10322 |
+
{
|
| 10323 |
+
"epoch": 0.12193406957025273,
|
| 10324 |
+
"grad_norm": 1475.9112548828125,
|
| 10325 |
+
"learning_rate": 9.988878444244937e-06,
|
| 10326 |
+
"loss": 162.3183,
|
| 10327 |
+
"step": 14740
|
| 10328 |
+
},
|
| 10329 |
+
{
|
| 10330 |
+
"epoch": 0.12201679281962195,
|
| 10331 |
+
"grad_norm": 1475.7916259765625,
|
| 10332 |
+
"learning_rate": 9.988784229445689e-06,
|
| 10333 |
+
"loss": 132.9056,
|
| 10334 |
+
"step": 14750
|
| 10335 |
+
},
|
| 10336 |
+
{
|
| 10337 |
+
"epoch": 0.12209951606899119,
|
| 10338 |
+
"grad_norm": 1580.335205078125,
|
| 10339 |
+
"learning_rate": 9.988689617711777e-06,
|
| 10340 |
+
"loss": 180.2133,
|
| 10341 |
+
"step": 14760
|
| 10342 |
+
},
|
| 10343 |
+
{
|
| 10344 |
+
"epoch": 0.12218223931836042,
|
| 10345 |
+
"grad_norm": 1742.8638916015625,
|
| 10346 |
+
"learning_rate": 9.988594609050726e-06,
|
| 10347 |
+
"loss": 170.6644,
|
| 10348 |
+
"step": 14770
|
| 10349 |
+
},
|
| 10350 |
+
{
|
| 10351 |
+
"epoch": 0.12226496256772966,
|
| 10352 |
+
"grad_norm": 778.4093017578125,
|
| 10353 |
+
"learning_rate": 9.988499203470097e-06,
|
| 10354 |
+
"loss": 163.6835,
|
| 10355 |
+
"step": 14780
|
| 10356 |
+
},
|
| 10357 |
+
{
|
| 10358 |
+
"epoch": 0.1223476858170989,
|
| 10359 |
+
"grad_norm": 908.4758911132812,
|
| 10360 |
+
"learning_rate": 9.988403400977482e-06,
|
| 10361 |
+
"loss": 143.079,
|
| 10362 |
+
"step": 14790
|
| 10363 |
+
},
|
| 10364 |
+
{
|
| 10365 |
+
"epoch": 0.12243040906646813,
|
| 10366 |
+
"grad_norm": 1540.624755859375,
|
| 10367 |
+
"learning_rate": 9.9883072015805e-06,
|
| 10368 |
+
"loss": 160.3763,
|
| 10369 |
+
"step": 14800
|
| 10370 |
+
},
|
| 10371 |
+
{
|
| 10372 |
+
"epoch": 0.12251313231583737,
|
| 10373 |
+
"grad_norm": 919.294677734375,
|
| 10374 |
+
"learning_rate": 9.98821060528681e-06,
|
| 10375 |
+
"loss": 165.4283,
|
| 10376 |
+
"step": 14810
|
| 10377 |
+
},
|
| 10378 |
+
{
|
| 10379 |
+
"epoch": 0.12259585556520661,
|
| 10380 |
+
"grad_norm": 865.2339477539062,
|
| 10381 |
+
"learning_rate": 9.988113612104093e-06,
|
| 10382 |
+
"loss": 128.951,
|
| 10383 |
+
"step": 14820
|
| 10384 |
+
},
|
| 10385 |
+
{
|
| 10386 |
+
"epoch": 0.12267857881457583,
|
| 10387 |
+
"grad_norm": 2098.492919921875,
|
| 10388 |
+
"learning_rate": 9.988016222040067e-06,
|
| 10389 |
+
"loss": 151.1649,
|
| 10390 |
+
"step": 14830
|
| 10391 |
+
},
|
| 10392 |
+
{
|
| 10393 |
+
"epoch": 0.12276130206394507,
|
| 10394 |
+
"grad_norm": 847.621337890625,
|
| 10395 |
+
"learning_rate": 9.987918435102484e-06,
|
| 10396 |
+
"loss": 121.6645,
|
| 10397 |
+
"step": 14840
|
| 10398 |
+
},
|
| 10399 |
+
{
|
| 10400 |
+
"epoch": 0.1228440253133143,
|
| 10401 |
+
"grad_norm": 1472.7208251953125,
|
| 10402 |
+
"learning_rate": 9.987820251299121e-06,
|
| 10403 |
+
"loss": 140.8588,
|
| 10404 |
+
"step": 14850
|
| 10405 |
+
},
|
| 10406 |
+
{
|
| 10407 |
+
"epoch": 0.12292674856268354,
|
| 10408 |
+
"grad_norm": 1310.5726318359375,
|
| 10409 |
+
"learning_rate": 9.987721670637794e-06,
|
| 10410 |
+
"loss": 132.6207,
|
| 10411 |
+
"step": 14860
|
| 10412 |
+
},
|
| 10413 |
+
{
|
| 10414 |
+
"epoch": 0.12300947181205278,
|
| 10415 |
+
"grad_norm": 788.9578247070312,
|
| 10416 |
+
"learning_rate": 9.987622693126342e-06,
|
| 10417 |
+
"loss": 139.2334,
|
| 10418 |
+
"step": 14870
|
| 10419 |
+
},
|
| 10420 |
+
{
|
| 10421 |
+
"epoch": 0.12309219506142201,
|
| 10422 |
+
"grad_norm": 1761.3287353515625,
|
| 10423 |
+
"learning_rate": 9.987523318772644e-06,
|
| 10424 |
+
"loss": 156.3363,
|
| 10425 |
+
"step": 14880
|
| 10426 |
+
},
|
| 10427 |
+
{
|
| 10428 |
+
"epoch": 0.12317491831079125,
|
| 10429 |
+
"grad_norm": 1024.44140625,
|
| 10430 |
+
"learning_rate": 9.987423547584605e-06,
|
| 10431 |
+
"loss": 167.0266,
|
| 10432 |
+
"step": 14890
|
| 10433 |
+
},
|
| 10434 |
+
{
|
| 10435 |
+
"epoch": 0.12325764156016049,
|
| 10436 |
+
"grad_norm": 891.505126953125,
|
| 10437 |
+
"learning_rate": 9.987323379570161e-06,
|
| 10438 |
+
"loss": 144.4436,
|
| 10439 |
+
"step": 14900
|
| 10440 |
+
},
|
| 10441 |
+
{
|
| 10442 |
+
"epoch": 0.12334036480952972,
|
| 10443 |
+
"grad_norm": 1373.43359375,
|
| 10444 |
+
"learning_rate": 9.987222814737287e-06,
|
| 10445 |
+
"loss": 139.1032,
|
| 10446 |
+
"step": 14910
|
| 10447 |
+
},
|
| 10448 |
+
{
|
| 10449 |
+
"epoch": 0.12342308805889896,
|
| 10450 |
+
"grad_norm": 2005.6689453125,
|
| 10451 |
+
"learning_rate": 9.987121853093982e-06,
|
| 10452 |
+
"loss": 179.2018,
|
| 10453 |
+
"step": 14920
|
| 10454 |
+
},
|
| 10455 |
+
{
|
| 10456 |
+
"epoch": 0.12350581130826818,
|
| 10457 |
+
"grad_norm": 1276.0216064453125,
|
| 10458 |
+
"learning_rate": 9.987020494648279e-06,
|
| 10459 |
+
"loss": 157.102,
|
| 10460 |
+
"step": 14930
|
| 10461 |
+
},
|
| 10462 |
+
{
|
| 10463 |
+
"epoch": 0.12358853455763742,
|
| 10464 |
+
"grad_norm": 1294.2474365234375,
|
| 10465 |
+
"learning_rate": 9.986918739408241e-06,
|
| 10466 |
+
"loss": 176.7196,
|
| 10467 |
+
"step": 14940
|
| 10468 |
+
},
|
| 10469 |
+
{
|
| 10470 |
+
"epoch": 0.12367125780700666,
|
| 10471 |
+
"grad_norm": 1715.0291748046875,
|
| 10472 |
+
"learning_rate": 9.986816587381966e-06,
|
| 10473 |
+
"loss": 150.2139,
|
| 10474 |
+
"step": 14950
|
| 10475 |
+
},
|
| 10476 |
+
{
|
| 10477 |
+
"epoch": 0.12375398105637589,
|
| 10478 |
+
"grad_norm": 1020.9821166992188,
|
| 10479 |
+
"learning_rate": 9.986714038577582e-06,
|
| 10480 |
+
"loss": 145.5669,
|
| 10481 |
+
"step": 14960
|
| 10482 |
+
},
|
| 10483 |
+
{
|
| 10484 |
+
"epoch": 0.12383670430574513,
|
| 10485 |
+
"grad_norm": 969.8027954101562,
|
| 10486 |
+
"learning_rate": 9.986611093003249e-06,
|
| 10487 |
+
"loss": 129.0563,
|
| 10488 |
+
"step": 14970
|
| 10489 |
+
},
|
| 10490 |
+
{
|
| 10491 |
+
"epoch": 0.12391942755511437,
|
| 10492 |
+
"grad_norm": 1188.594482421875,
|
| 10493 |
+
"learning_rate": 9.986507750667157e-06,
|
| 10494 |
+
"loss": 130.371,
|
| 10495 |
+
"step": 14980
|
| 10496 |
+
},
|
| 10497 |
+
{
|
| 10498 |
+
"epoch": 0.1240021508044836,
|
| 10499 |
+
"grad_norm": 2070.416015625,
|
| 10500 |
+
"learning_rate": 9.986404011577525e-06,
|
| 10501 |
+
"loss": 164.5774,
|
| 10502 |
+
"step": 14990
|
| 10503 |
+
},
|
| 10504 |
+
{
|
| 10505 |
+
"epoch": 0.12408487405385284,
|
| 10506 |
+
"grad_norm": 1122.52587890625,
|
| 10507 |
+
"learning_rate": 9.986299875742612e-06,
|
| 10508 |
+
"loss": 187.0694,
|
| 10509 |
+
"step": 15000
|
| 10510 |
+
},
|
| 10511 |
+
{
|
| 10512 |
+
"epoch": 0.12416759730322208,
|
| 10513 |
+
"grad_norm": 1119.6961669921875,
|
| 10514 |
+
"learning_rate": 9.986195343170703e-06,
|
| 10515 |
+
"loss": 180.5289,
|
| 10516 |
+
"step": 15010
|
| 10517 |
+
},
|
| 10518 |
+
{
|
| 10519 |
+
"epoch": 0.1242503205525913,
|
| 10520 |
+
"grad_norm": 1574.2568359375,
|
| 10521 |
+
"learning_rate": 9.986090413870114e-06,
|
| 10522 |
+
"loss": 144.7522,
|
| 10523 |
+
"step": 15020
|
| 10524 |
+
},
|
| 10525 |
+
{
|
| 10526 |
+
"epoch": 0.12433304380196054,
|
| 10527 |
+
"grad_norm": 1006.7045288085938,
|
| 10528 |
+
"learning_rate": 9.985985087849193e-06,
|
| 10529 |
+
"loss": 143.7221,
|
| 10530 |
+
"step": 15030
|
| 10531 |
+
},
|
| 10532 |
+
{
|
| 10533 |
+
"epoch": 0.12441576705132977,
|
| 10534 |
+
"grad_norm": 1300.2181396484375,
|
| 10535 |
+
"learning_rate": 9.98587936511632e-06,
|
| 10536 |
+
"loss": 150.1932,
|
| 10537 |
+
"step": 15040
|
| 10538 |
+
},
|
| 10539 |
+
{
|
| 10540 |
+
"epoch": 0.12449849030069901,
|
| 10541 |
+
"grad_norm": 821.7041625976562,
|
| 10542 |
+
"learning_rate": 9.98577324567991e-06,
|
| 10543 |
+
"loss": 139.0086,
|
| 10544 |
+
"step": 15050
|
| 10545 |
+
},
|
| 10546 |
+
{
|
| 10547 |
+
"epoch": 0.12458121355006825,
|
| 10548 |
+
"grad_norm": 830.7269287109375,
|
| 10549 |
+
"learning_rate": 9.985666729548404e-06,
|
| 10550 |
+
"loss": 146.4651,
|
| 10551 |
+
"step": 15060
|
| 10552 |
+
},
|
| 10553 |
+
{
|
| 10554 |
+
"epoch": 0.12466393679943748,
|
| 10555 |
+
"grad_norm": 1310.355224609375,
|
| 10556 |
+
"learning_rate": 9.985559816730277e-06,
|
| 10557 |
+
"loss": 141.5489,
|
| 10558 |
+
"step": 15070
|
| 10559 |
+
},
|
| 10560 |
+
{
|
| 10561 |
+
"epoch": 0.12474666004880672,
|
| 10562 |
+
"grad_norm": 1190.0335693359375,
|
| 10563 |
+
"learning_rate": 9.985452507234037e-06,
|
| 10564 |
+
"loss": 144.9001,
|
| 10565 |
+
"step": 15080
|
| 10566 |
+
},
|
| 10567 |
+
{
|
| 10568 |
+
"epoch": 0.12482938329817596,
|
| 10569 |
+
"grad_norm": 2714.714599609375,
|
| 10570 |
+
"learning_rate": 9.98534480106822e-06,
|
| 10571 |
+
"loss": 154.9118,
|
| 10572 |
+
"step": 15090
|
| 10573 |
+
},
|
| 10574 |
+
{
|
| 10575 |
+
"epoch": 0.12491210654754518,
|
| 10576 |
+
"grad_norm": 792.223388671875,
|
| 10577 |
+
"learning_rate": 9.985236698241396e-06,
|
| 10578 |
+
"loss": 149.7406,
|
| 10579 |
+
"step": 15100
|
| 10580 |
+
},
|
| 10581 |
+
{
|
| 10582 |
+
"epoch": 0.12499482979691443,
|
| 10583 |
+
"grad_norm": 1287.8345947265625,
|
| 10584 |
+
"learning_rate": 9.985128198762168e-06,
|
| 10585 |
+
"loss": 171.4261,
|
| 10586 |
+
"step": 15110
|
| 10587 |
+
},
|
| 10588 |
+
{
|
| 10589 |
+
"epoch": 0.12507755304628365,
|
| 10590 |
+
"grad_norm": 957.4619140625,
|
| 10591 |
+
"learning_rate": 9.98501930263917e-06,
|
| 10592 |
+
"loss": 234.8733,
|
| 10593 |
+
"step": 15120
|
| 10594 |
+
},
|
| 10595 |
+
{
|
| 10596 |
+
"epoch": 0.1251602762956529,
|
| 10597 |
+
"grad_norm": 862.0460205078125,
|
| 10598 |
+
"learning_rate": 9.984910009881062e-06,
|
| 10599 |
+
"loss": 112.6332,
|
| 10600 |
+
"step": 15130
|
| 10601 |
+
},
|
| 10602 |
+
{
|
| 10603 |
+
"epoch": 0.12524299954502213,
|
| 10604 |
+
"grad_norm": 1100.07080078125,
|
| 10605 |
+
"learning_rate": 9.984800320496542e-06,
|
| 10606 |
+
"loss": 139.7673,
|
| 10607 |
+
"step": 15140
|
| 10608 |
+
},
|
| 10609 |
+
{
|
| 10610 |
+
"epoch": 0.12532572279439136,
|
| 10611 |
+
"grad_norm": 1111.9737548828125,
|
| 10612 |
+
"learning_rate": 9.984690234494338e-06,
|
| 10613 |
+
"loss": 106.7051,
|
| 10614 |
+
"step": 15150
|
| 10615 |
+
},
|
| 10616 |
+
{
|
| 10617 |
+
"epoch": 0.12540844604376059,
|
| 10618 |
+
"grad_norm": 744.8794555664062,
|
| 10619 |
+
"learning_rate": 9.98457975188321e-06,
|
| 10620 |
+
"loss": 142.1312,
|
| 10621 |
+
"step": 15160
|
| 10622 |
+
},
|
| 10623 |
+
{
|
| 10624 |
+
"epoch": 0.12549116929312984,
|
| 10625 |
+
"grad_norm": 923.189697265625,
|
| 10626 |
+
"learning_rate": 9.984468872671945e-06,
|
| 10627 |
+
"loss": 139.3656,
|
| 10628 |
+
"step": 15170
|
| 10629 |
+
},
|
| 10630 |
+
{
|
| 10631 |
+
"epoch": 0.12557389254249907,
|
| 10632 |
+
"grad_norm": 1322.687255859375,
|
| 10633 |
+
"learning_rate": 9.984357596869369e-06,
|
| 10634 |
+
"loss": 148.6495,
|
| 10635 |
+
"step": 15180
|
| 10636 |
+
},
|
| 10637 |
+
{
|
| 10638 |
+
"epoch": 0.1256566157918683,
|
| 10639 |
+
"grad_norm": 882.3487548828125,
|
| 10640 |
+
"learning_rate": 9.984245924484334e-06,
|
| 10641 |
+
"loss": 141.7766,
|
| 10642 |
+
"step": 15190
|
| 10643 |
+
},
|
| 10644 |
+
{
|
| 10645 |
+
"epoch": 0.12573933904123755,
|
| 10646 |
+
"grad_norm": 725.9840698242188,
|
| 10647 |
+
"learning_rate": 9.984133855525723e-06,
|
| 10648 |
+
"loss": 138.5364,
|
| 10649 |
+
"step": 15200
|
| 10650 |
+
},
|
| 10651 |
+
{
|
| 10652 |
+
"epoch": 0.12582206229060677,
|
| 10653 |
+
"grad_norm": 1547.522705078125,
|
| 10654 |
+
"learning_rate": 9.984021390002458e-06,
|
| 10655 |
+
"loss": 136.4458,
|
| 10656 |
+
"step": 15210
|
| 10657 |
+
},
|
| 10658 |
+
{
|
| 10659 |
+
"epoch": 0.125904785539976,
|
| 10660 |
+
"grad_norm": 1425.56494140625,
|
| 10661 |
+
"learning_rate": 9.983908527923486e-06,
|
| 10662 |
+
"loss": 222.0387,
|
| 10663 |
+
"step": 15220
|
| 10664 |
+
},
|
| 10665 |
+
{
|
| 10666 |
+
"epoch": 0.12598750878934525,
|
| 10667 |
+
"grad_norm": 911.7035522460938,
|
| 10668 |
+
"learning_rate": 9.983795269297782e-06,
|
| 10669 |
+
"loss": 169.7902,
|
| 10670 |
+
"step": 15230
|
| 10671 |
+
},
|
| 10672 |
+
{
|
| 10673 |
+
"epoch": 0.12607023203871448,
|
| 10674 |
+
"grad_norm": 1066.0264892578125,
|
| 10675 |
+
"learning_rate": 9.983681614134363e-06,
|
| 10676 |
+
"loss": 122.5573,
|
| 10677 |
+
"step": 15240
|
| 10678 |
+
},
|
| 10679 |
+
{
|
| 10680 |
+
"epoch": 0.1261529552880837,
|
| 10681 |
+
"grad_norm": 1829.2509765625,
|
| 10682 |
+
"learning_rate": 9.98356756244227e-06,
|
| 10683 |
+
"loss": 154.7958,
|
| 10684 |
+
"step": 15250
|
| 10685 |
+
},
|
| 10686 |
+
{
|
| 10687 |
+
"epoch": 0.12623567853745296,
|
| 10688 |
+
"grad_norm": 1402.93408203125,
|
| 10689 |
+
"learning_rate": 9.983453114230575e-06,
|
| 10690 |
+
"loss": 145.442,
|
| 10691 |
+
"step": 15260
|
| 10692 |
+
},
|
| 10693 |
+
{
|
| 10694 |
+
"epoch": 0.1263184017868222,
|
| 10695 |
+
"grad_norm": 990.7800903320312,
|
| 10696 |
+
"learning_rate": 9.98333826950839e-06,
|
| 10697 |
+
"loss": 138.916,
|
| 10698 |
+
"step": 15270
|
| 10699 |
+
},
|
| 10700 |
+
{
|
| 10701 |
+
"epoch": 0.12640112503619141,
|
| 10702 |
+
"grad_norm": 861.1292724609375,
|
| 10703 |
+
"learning_rate": 9.983223028284847e-06,
|
| 10704 |
+
"loss": 152.3527,
|
| 10705 |
+
"step": 15280
|
| 10706 |
+
},
|
| 10707 |
+
{
|
| 10708 |
+
"epoch": 0.12648384828556067,
|
| 10709 |
+
"grad_norm": 887.3511962890625,
|
| 10710 |
+
"learning_rate": 9.983107390569118e-06,
|
| 10711 |
+
"loss": 129.7973,
|
| 10712 |
+
"step": 15290
|
| 10713 |
+
},
|
| 10714 |
+
{
|
| 10715 |
+
"epoch": 0.1265665715349299,
|
| 10716 |
+
"grad_norm": 1043.2373046875,
|
| 10717 |
+
"learning_rate": 9.982991356370404e-06,
|
| 10718 |
+
"loss": 116.1451,
|
| 10719 |
+
"step": 15300
|
| 10720 |
+
},
|
| 10721 |
+
{
|
| 10722 |
+
"epoch": 0.12664929478429912,
|
| 10723 |
+
"grad_norm": 1244.5079345703125,
|
| 10724 |
+
"learning_rate": 9.982874925697937e-06,
|
| 10725 |
+
"loss": 221.0664,
|
| 10726 |
+
"step": 15310
|
| 10727 |
+
},
|
| 10728 |
+
{
|
| 10729 |
+
"epoch": 0.12673201803366838,
|
| 10730 |
+
"grad_norm": 1715.1995849609375,
|
| 10731 |
+
"learning_rate": 9.982758098560978e-06,
|
| 10732 |
+
"loss": 186.7455,
|
| 10733 |
+
"step": 15320
|
| 10734 |
+
},
|
| 10735 |
+
{
|
| 10736 |
+
"epoch": 0.1268147412830376,
|
| 10737 |
+
"grad_norm": 679.9988403320312,
|
| 10738 |
+
"learning_rate": 9.982640874968827e-06,
|
| 10739 |
+
"loss": 171.8672,
|
| 10740 |
+
"step": 15330
|
| 10741 |
+
},
|
| 10742 |
+
{
|
| 10743 |
+
"epoch": 0.12689746453240683,
|
| 10744 |
+
"grad_norm": 595.40625,
|
| 10745 |
+
"learning_rate": 9.98252325493081e-06,
|
| 10746 |
+
"loss": 130.3511,
|
| 10747 |
+
"step": 15340
|
| 10748 |
+
},
|
| 10749 |
+
{
|
| 10750 |
+
"epoch": 0.12698018778177605,
|
| 10751 |
+
"grad_norm": 915.3275146484375,
|
| 10752 |
+
"learning_rate": 9.982405238456281e-06,
|
| 10753 |
+
"loss": 153.7831,
|
| 10754 |
+
"step": 15350
|
| 10755 |
+
},
|
| 10756 |
+
{
|
| 10757 |
+
"epoch": 0.1270629110311453,
|
| 10758 |
+
"grad_norm": 1383.0423583984375,
|
| 10759 |
+
"learning_rate": 9.982286825554636e-06,
|
| 10760 |
+
"loss": 155.1486,
|
| 10761 |
+
"step": 15360
|
| 10762 |
+
},
|
| 10763 |
+
{
|
| 10764 |
+
"epoch": 0.12714563428051454,
|
| 10765 |
+
"grad_norm": 1527.0670166015625,
|
| 10766 |
+
"learning_rate": 9.982168016235292e-06,
|
| 10767 |
+
"loss": 235.3831,
|
| 10768 |
+
"step": 15370
|
| 10769 |
+
},
|
| 10770 |
+
{
|
| 10771 |
+
"epoch": 0.12722835752988376,
|
| 10772 |
+
"grad_norm": 1168.0416259765625,
|
| 10773 |
+
"learning_rate": 9.982048810507706e-06,
|
| 10774 |
+
"loss": 175.3166,
|
| 10775 |
+
"step": 15380
|
| 10776 |
+
},
|
| 10777 |
+
{
|
| 10778 |
+
"epoch": 0.12731108077925302,
|
| 10779 |
+
"grad_norm": 1577.9107666015625,
|
| 10780 |
+
"learning_rate": 9.98192920838136e-06,
|
| 10781 |
+
"loss": 136.4098,
|
| 10782 |
+
"step": 15390
|
| 10783 |
+
},
|
| 10784 |
+
{
|
| 10785 |
+
"epoch": 0.12739380402862224,
|
| 10786 |
+
"grad_norm": 2239.125244140625,
|
| 10787 |
+
"learning_rate": 9.98180920986577e-06,
|
| 10788 |
+
"loss": 162.4811,
|
| 10789 |
+
"step": 15400
|
| 10790 |
+
},
|
| 10791 |
+
{
|
| 10792 |
+
"epoch": 0.12747652727799147,
|
| 10793 |
+
"grad_norm": 1140.6561279296875,
|
| 10794 |
+
"learning_rate": 9.981688814970485e-06,
|
| 10795 |
+
"loss": 159.3877,
|
| 10796 |
+
"step": 15410
|
| 10797 |
+
},
|
| 10798 |
+
{
|
| 10799 |
+
"epoch": 0.12755925052736072,
|
| 10800 |
+
"grad_norm": 929.8948974609375,
|
| 10801 |
+
"learning_rate": 9.981568023705085e-06,
|
| 10802 |
+
"loss": 113.0717,
|
| 10803 |
+
"step": 15420
|
| 10804 |
+
},
|
| 10805 |
+
{
|
| 10806 |
+
"epoch": 0.12764197377672995,
|
| 10807 |
+
"grad_norm": 1146.56396484375,
|
| 10808 |
+
"learning_rate": 9.981446836079178e-06,
|
| 10809 |
+
"loss": 121.9914,
|
| 10810 |
+
"step": 15430
|
| 10811 |
+
},
|
| 10812 |
+
{
|
| 10813 |
+
"epoch": 0.12772469702609918,
|
| 10814 |
+
"grad_norm": 877.8550415039062,
|
| 10815 |
+
"learning_rate": 9.981325252102408e-06,
|
| 10816 |
+
"loss": 173.6141,
|
| 10817 |
+
"step": 15440
|
| 10818 |
+
},
|
| 10819 |
+
{
|
| 10820 |
+
"epoch": 0.12780742027546843,
|
| 10821 |
+
"grad_norm": 969.0079956054688,
|
| 10822 |
+
"learning_rate": 9.98120327178445e-06,
|
| 10823 |
+
"loss": 178.5706,
|
| 10824 |
+
"step": 15450
|
| 10825 |
+
},
|
| 10826 |
+
{
|
| 10827 |
+
"epoch": 0.12789014352483766,
|
| 10828 |
+
"grad_norm": 1263.2391357421875,
|
| 10829 |
+
"learning_rate": 9.981080895135007e-06,
|
| 10830 |
+
"loss": 180.7431,
|
| 10831 |
+
"step": 15460
|
| 10832 |
+
},
|
| 10833 |
+
{
|
| 10834 |
+
"epoch": 0.12797286677420688,
|
| 10835 |
+
"grad_norm": 742.5184326171875,
|
| 10836 |
+
"learning_rate": 9.980958122163818e-06,
|
| 10837 |
+
"loss": 111.0224,
|
| 10838 |
+
"step": 15470
|
| 10839 |
+
},
|
| 10840 |
+
{
|
| 10841 |
+
"epoch": 0.12805559002357614,
|
| 10842 |
+
"grad_norm": 1423.19873046875,
|
| 10843 |
+
"learning_rate": 9.980834952880652e-06,
|
| 10844 |
+
"loss": 128.3473,
|
| 10845 |
+
"step": 15480
|
| 10846 |
+
},
|
| 10847 |
+
{
|
| 10848 |
+
"epoch": 0.12813831327294536,
|
| 10849 |
+
"grad_norm": 1583.2940673828125,
|
| 10850 |
+
"learning_rate": 9.980711387295306e-06,
|
| 10851 |
+
"loss": 149.955,
|
| 10852 |
+
"step": 15490
|
| 10853 |
+
},
|
| 10854 |
+
{
|
| 10855 |
+
"epoch": 0.1282210365223146,
|
| 10856 |
+
"grad_norm": 1052.7265625,
|
| 10857 |
+
"learning_rate": 9.980587425417612e-06,
|
| 10858 |
+
"loss": 159.9205,
|
| 10859 |
+
"step": 15500
|
| 10860 |
+
},
|
| 10861 |
+
{
|
| 10862 |
+
"epoch": 0.12830375977168385,
|
| 10863 |
+
"grad_norm": 2138.17724609375,
|
| 10864 |
+
"learning_rate": 9.980463067257437e-06,
|
| 10865 |
+
"loss": 169.7366,
|
| 10866 |
+
"step": 15510
|
| 10867 |
+
},
|
| 10868 |
+
{
|
| 10869 |
+
"epoch": 0.12838648302105307,
|
| 10870 |
+
"grad_norm": 1006.1878662109375,
|
| 10871 |
+
"learning_rate": 9.980338312824672e-06,
|
| 10872 |
+
"loss": 193.1612,
|
| 10873 |
+
"step": 15520
|
| 10874 |
+
},
|
| 10875 |
+
{
|
| 10876 |
+
"epoch": 0.1284692062704223,
|
| 10877 |
+
"grad_norm": 1047.7593994140625,
|
| 10878 |
+
"learning_rate": 9.980213162129244e-06,
|
| 10879 |
+
"loss": 175.5892,
|
| 10880 |
+
"step": 15530
|
| 10881 |
+
},
|
| 10882 |
+
{
|
| 10883 |
+
"epoch": 0.12855192951979152,
|
| 10884 |
+
"grad_norm": 1267.4644775390625,
|
| 10885 |
+
"learning_rate": 9.980087615181111e-06,
|
| 10886 |
+
"loss": 149.4357,
|
| 10887 |
+
"step": 15540
|
| 10888 |
+
},
|
| 10889 |
+
{
|
| 10890 |
+
"epoch": 0.12863465276916078,
|
| 10891 |
+
"grad_norm": 1171.0859375,
|
| 10892 |
+
"learning_rate": 9.979961671990263e-06,
|
| 10893 |
+
"loss": 165.6414,
|
| 10894 |
+
"step": 15550
|
| 10895 |
+
},
|
| 10896 |
+
{
|
| 10897 |
+
"epoch": 0.12871737601853,
|
| 10898 |
+
"grad_norm": 911.0418701171875,
|
| 10899 |
+
"learning_rate": 9.979835332566719e-06,
|
| 10900 |
+
"loss": 155.2462,
|
| 10901 |
+
"step": 15560
|
| 10902 |
+
},
|
| 10903 |
+
{
|
| 10904 |
+
"epoch": 0.12880009926789923,
|
| 10905 |
+
"grad_norm": 1016.1674194335938,
|
| 10906 |
+
"learning_rate": 9.97970859692053e-06,
|
| 10907 |
+
"loss": 142.4974,
|
| 10908 |
+
"step": 15570
|
| 10909 |
+
},
|
| 10910 |
+
{
|
| 10911 |
+
"epoch": 0.12888282251726849,
|
| 10912 |
+
"grad_norm": 653.232421875,
|
| 10913 |
+
"learning_rate": 9.979581465061784e-06,
|
| 10914 |
+
"loss": 155.5012,
|
| 10915 |
+
"step": 15580
|
| 10916 |
+
},
|
| 10917 |
+
{
|
| 10918 |
+
"epoch": 0.1289655457666377,
|
| 10919 |
+
"grad_norm": 1058.5008544921875,
|
| 10920 |
+
"learning_rate": 9.979453937000594e-06,
|
| 10921 |
+
"loss": 101.9423,
|
| 10922 |
+
"step": 15590
|
| 10923 |
+
},
|
| 10924 |
+
{
|
| 10925 |
+
"epoch": 0.12904826901600694,
|
| 10926 |
+
"grad_norm": 820.3455200195312,
|
| 10927 |
+
"learning_rate": 9.979326012747106e-06,
|
| 10928 |
+
"loss": 117.5258,
|
| 10929 |
+
"step": 15600
|
| 10930 |
+
},
|
| 10931 |
+
{
|
| 10932 |
+
"epoch": 0.1291309922653762,
|
| 10933 |
+
"grad_norm": 1352.7105712890625,
|
| 10934 |
+
"learning_rate": 9.9791976923115e-06,
|
| 10935 |
+
"loss": 99.8209,
|
| 10936 |
+
"step": 15610
|
| 10937 |
+
},
|
| 10938 |
+
{
|
| 10939 |
+
"epoch": 0.12921371551474542,
|
| 10940 |
+
"grad_norm": 1026.9713134765625,
|
| 10941 |
+
"learning_rate": 9.979068975703984e-06,
|
| 10942 |
+
"loss": 166.7305,
|
| 10943 |
+
"step": 15620
|
| 10944 |
+
},
|
| 10945 |
+
{
|
| 10946 |
+
"epoch": 0.12929643876411465,
|
| 10947 |
+
"grad_norm": 1245.904296875,
|
| 10948 |
+
"learning_rate": 9.978939862934802e-06,
|
| 10949 |
+
"loss": 126.3938,
|
| 10950 |
+
"step": 15630
|
| 10951 |
+
},
|
| 10952 |
+
{
|
| 10953 |
+
"epoch": 0.1293791620134839,
|
| 10954 |
+
"grad_norm": 1658.7640380859375,
|
| 10955 |
+
"learning_rate": 9.978810354014223e-06,
|
| 10956 |
+
"loss": 135.5493,
|
| 10957 |
+
"step": 15640
|
| 10958 |
+
},
|
| 10959 |
+
{
|
| 10960 |
+
"epoch": 0.12946188526285313,
|
| 10961 |
+
"grad_norm": 1517.82373046875,
|
| 10962 |
+
"learning_rate": 9.978680448952556e-06,
|
| 10963 |
+
"loss": 139.2036,
|
| 10964 |
+
"step": 15650
|
| 10965 |
+
},
|
| 10966 |
+
{
|
| 10967 |
+
"epoch": 0.12954460851222235,
|
| 10968 |
+
"grad_norm": 1112.2469482421875,
|
| 10969 |
+
"learning_rate": 9.978550147760133e-06,
|
| 10970 |
+
"loss": 127.4167,
|
| 10971 |
+
"step": 15660
|
| 10972 |
+
},
|
| 10973 |
+
{
|
| 10974 |
+
"epoch": 0.1296273317615916,
|
| 10975 |
+
"grad_norm": 995.6966552734375,
|
| 10976 |
+
"learning_rate": 9.978419450447325e-06,
|
| 10977 |
+
"loss": 128.1456,
|
| 10978 |
+
"step": 15670
|
| 10979 |
+
},
|
| 10980 |
+
{
|
| 10981 |
+
"epoch": 0.12971005501096083,
|
| 10982 |
+
"grad_norm": 785.5421142578125,
|
| 10983 |
+
"learning_rate": 9.978288357024527e-06,
|
| 10984 |
+
"loss": 142.7447,
|
| 10985 |
+
"step": 15680
|
| 10986 |
+
},
|
| 10987 |
+
{
|
| 10988 |
+
"epoch": 0.12979277826033006,
|
| 10989 |
+
"grad_norm": 3418.008544921875,
|
| 10990 |
+
"learning_rate": 9.978156867502173e-06,
|
| 10991 |
+
"loss": 161.2918,
|
| 10992 |
+
"step": 15690
|
| 10993 |
+
},
|
| 10994 |
+
{
|
| 10995 |
+
"epoch": 0.12987550150969931,
|
| 10996 |
+
"grad_norm": 1545.32568359375,
|
| 10997 |
+
"learning_rate": 9.978024981890724e-06,
|
| 10998 |
+
"loss": 107.0028,
|
| 10999 |
+
"step": 15700
|
| 11000 |
+
},
|
| 11001 |
+
{
|
| 11002 |
+
"epoch": 0.12995822475906854,
|
| 11003 |
+
"grad_norm": 963.7340087890625,
|
| 11004 |
+
"learning_rate": 9.977892700200673e-06,
|
| 11005 |
+
"loss": 140.569,
|
| 11006 |
+
"step": 15710
|
| 11007 |
+
},
|
| 11008 |
+
{
|
| 11009 |
+
"epoch": 0.13004094800843777,
|
| 11010 |
+
"grad_norm": 692.7611694335938,
|
| 11011 |
+
"learning_rate": 9.977760022442545e-06,
|
| 11012 |
+
"loss": 110.664,
|
| 11013 |
+
"step": 15720
|
| 11014 |
+
},
|
| 11015 |
+
{
|
| 11016 |
+
"epoch": 0.130123671257807,
|
| 11017 |
+
"grad_norm": 1015.7322998046875,
|
| 11018 |
+
"learning_rate": 9.977626948626897e-06,
|
| 11019 |
+
"loss": 158.9243,
|
| 11020 |
+
"step": 15730
|
| 11021 |
+
},
|
| 11022 |
+
{
|
| 11023 |
+
"epoch": 0.13020639450717625,
|
| 11024 |
+
"grad_norm": 1334.6917724609375,
|
| 11025 |
+
"learning_rate": 9.977493478764316e-06,
|
| 11026 |
+
"loss": 152.3215,
|
| 11027 |
+
"step": 15740
|
| 11028 |
+
},
|
| 11029 |
+
{
|
| 11030 |
+
"epoch": 0.13028911775654548,
|
| 11031 |
+
"grad_norm": 963.1575927734375,
|
| 11032 |
+
"learning_rate": 9.977359612865424e-06,
|
| 11033 |
+
"loss": 137.1868,
|
| 11034 |
+
"step": 15750
|
| 11035 |
+
},
|
| 11036 |
+
{
|
| 11037 |
+
"epoch": 0.1303718410059147,
|
| 11038 |
+
"grad_norm": 1332.0909423828125,
|
| 11039 |
+
"learning_rate": 9.97722535094087e-06,
|
| 11040 |
+
"loss": 130.2153,
|
| 11041 |
+
"step": 15760
|
| 11042 |
+
},
|
| 11043 |
+
{
|
| 11044 |
+
"epoch": 0.13045456425528396,
|
| 11045 |
+
"grad_norm": 864.9472045898438,
|
| 11046 |
+
"learning_rate": 9.977090693001336e-06,
|
| 11047 |
+
"loss": 142.6017,
|
| 11048 |
+
"step": 15770
|
| 11049 |
+
},
|
| 11050 |
+
{
|
| 11051 |
+
"epoch": 0.13053728750465318,
|
| 11052 |
+
"grad_norm": 1091.3128662109375,
|
| 11053 |
+
"learning_rate": 9.976955639057539e-06,
|
| 11054 |
+
"loss": 126.0693,
|
| 11055 |
+
"step": 15780
|
| 11056 |
+
},
|
| 11057 |
+
{
|
| 11058 |
+
"epoch": 0.1306200107540224,
|
| 11059 |
+
"grad_norm": 1137.7115478515625,
|
| 11060 |
+
"learning_rate": 9.976820189120223e-06,
|
| 11061 |
+
"loss": 147.4185,
|
| 11062 |
+
"step": 15790
|
| 11063 |
+
},
|
| 11064 |
+
{
|
| 11065 |
+
"epoch": 0.13070273400339166,
|
| 11066 |
+
"grad_norm": 1658.2982177734375,
|
| 11067 |
+
"learning_rate": 9.976684343200164e-06,
|
| 11068 |
+
"loss": 135.441,
|
| 11069 |
+
"step": 15800
|
| 11070 |
+
},
|
| 11071 |
+
{
|
| 11072 |
+
"epoch": 0.1307854572527609,
|
| 11073 |
+
"grad_norm": 1823.3642578125,
|
| 11074 |
+
"learning_rate": 9.976548101308173e-06,
|
| 11075 |
+
"loss": 138.8229,
|
| 11076 |
+
"step": 15810
|
| 11077 |
+
},
|
| 11078 |
+
{
|
| 11079 |
+
"epoch": 0.13086818050213012,
|
| 11080 |
+
"grad_norm": 1511.6375732421875,
|
| 11081 |
+
"learning_rate": 9.976411463455088e-06,
|
| 11082 |
+
"loss": 140.3549,
|
| 11083 |
+
"step": 15820
|
| 11084 |
+
},
|
| 11085 |
+
{
|
| 11086 |
+
"epoch": 0.13095090375149937,
|
| 11087 |
+
"grad_norm": 970.9559326171875,
|
| 11088 |
+
"learning_rate": 9.976274429651783e-06,
|
| 11089 |
+
"loss": 188.4605,
|
| 11090 |
+
"step": 15830
|
| 11091 |
+
},
|
| 11092 |
+
{
|
| 11093 |
+
"epoch": 0.1310336270008686,
|
| 11094 |
+
"grad_norm": 1540.7110595703125,
|
| 11095 |
+
"learning_rate": 9.976136999909156e-06,
|
| 11096 |
+
"loss": 106.2589,
|
| 11097 |
+
"step": 15840
|
| 11098 |
+
},
|
| 11099 |
+
{
|
| 11100 |
+
"epoch": 0.13111635025023782,
|
| 11101 |
+
"grad_norm": 829.7328491210938,
|
| 11102 |
+
"learning_rate": 9.97599917423815e-06,
|
| 11103 |
+
"loss": 166.2885,
|
| 11104 |
+
"step": 15850
|
| 11105 |
+
},
|
| 11106 |
+
{
|
| 11107 |
+
"epoch": 0.13119907349960708,
|
| 11108 |
+
"grad_norm": 0.0,
|
| 11109 |
+
"learning_rate": 9.975860952649724e-06,
|
| 11110 |
+
"loss": 180.9173,
|
| 11111 |
+
"step": 15860
|
| 11112 |
+
},
|
| 11113 |
+
{
|
| 11114 |
+
"epoch": 0.1312817967489763,
|
| 11115 |
+
"grad_norm": 1011.6902465820312,
|
| 11116 |
+
"learning_rate": 9.975722335154876e-06,
|
| 11117 |
+
"loss": 161.2201,
|
| 11118 |
+
"step": 15870
|
| 11119 |
+
},
|
| 11120 |
+
{
|
| 11121 |
+
"epoch": 0.13136451999834553,
|
| 11122 |
+
"grad_norm": 1166.7960205078125,
|
| 11123 |
+
"learning_rate": 9.975583321764638e-06,
|
| 11124 |
+
"loss": 144.3113,
|
| 11125 |
+
"step": 15880
|
| 11126 |
+
},
|
| 11127 |
+
{
|
| 11128 |
+
"epoch": 0.13144724324771476,
|
| 11129 |
+
"grad_norm": 1041.4771728515625,
|
| 11130 |
+
"learning_rate": 9.975443912490073e-06,
|
| 11131 |
+
"loss": 149.5042,
|
| 11132 |
+
"step": 15890
|
| 11133 |
+
},
|
| 11134 |
+
{
|
| 11135 |
+
"epoch": 0.131529966497084,
|
| 11136 |
+
"grad_norm": 2316.5087890625,
|
| 11137 |
+
"learning_rate": 9.975304107342268e-06,
|
| 11138 |
+
"loss": 179.2303,
|
| 11139 |
+
"step": 15900
|
| 11140 |
+
},
|
| 11141 |
+
{
|
| 11142 |
+
"epoch": 0.13161268974645324,
|
| 11143 |
+
"grad_norm": 692.1578369140625,
|
| 11144 |
+
"learning_rate": 9.97516390633235e-06,
|
| 11145 |
+
"loss": 133.4318,
|
| 11146 |
+
"step": 15910
|
| 11147 |
+
},
|
| 11148 |
+
{
|
| 11149 |
+
"epoch": 0.13169541299582246,
|
| 11150 |
+
"grad_norm": 1125.5006103515625,
|
| 11151 |
+
"learning_rate": 9.975023309471473e-06,
|
| 11152 |
+
"loss": 156.1001,
|
| 11153 |
+
"step": 15920
|
| 11154 |
+
},
|
| 11155 |
+
{
|
| 11156 |
+
"epoch": 0.13177813624519172,
|
| 11157 |
+
"grad_norm": 1064.159423828125,
|
| 11158 |
+
"learning_rate": 9.974882316770823e-06,
|
| 11159 |
+
"loss": 147.876,
|
| 11160 |
+
"step": 15930
|
| 11161 |
+
},
|
| 11162 |
+
{
|
| 11163 |
+
"epoch": 0.13186085949456094,
|
| 11164 |
+
"grad_norm": 1456.9761962890625,
|
| 11165 |
+
"learning_rate": 9.974740928241617e-06,
|
| 11166 |
+
"loss": 146.098,
|
| 11167 |
+
"step": 15940
|
| 11168 |
+
},
|
| 11169 |
+
{
|
| 11170 |
+
"epoch": 0.13194358274393017,
|
| 11171 |
+
"grad_norm": 1191.022705078125,
|
| 11172 |
+
"learning_rate": 9.974599143895107e-06,
|
| 11173 |
+
"loss": 139.6693,
|
| 11174 |
+
"step": 15950
|
| 11175 |
+
},
|
| 11176 |
+
{
|
| 11177 |
+
"epoch": 0.13202630599329943,
|
| 11178 |
+
"grad_norm": 2010.4088134765625,
|
| 11179 |
+
"learning_rate": 9.974456963742573e-06,
|
| 11180 |
+
"loss": 152.4677,
|
| 11181 |
+
"step": 15960
|
| 11182 |
+
},
|
| 11183 |
+
{
|
| 11184 |
+
"epoch": 0.13210902924266865,
|
| 11185 |
+
"grad_norm": 1077.85205078125,
|
| 11186 |
+
"learning_rate": 9.97431438779533e-06,
|
| 11187 |
+
"loss": 199.6097,
|
| 11188 |
+
"step": 15970
|
| 11189 |
+
},
|
| 11190 |
+
{
|
| 11191 |
+
"epoch": 0.13219175249203788,
|
| 11192 |
+
"grad_norm": 975.5093994140625,
|
| 11193 |
+
"learning_rate": 9.974171416064719e-06,
|
| 11194 |
+
"loss": 110.029,
|
| 11195 |
+
"step": 15980
|
| 11196 |
+
},
|
| 11197 |
+
{
|
| 11198 |
+
"epoch": 0.13227447574140713,
|
| 11199 |
+
"grad_norm": 1180.7437744140625,
|
| 11200 |
+
"learning_rate": 9.974028048562118e-06,
|
| 11201 |
+
"loss": 136.7102,
|
| 11202 |
+
"step": 15990
|
| 11203 |
+
},
|
| 11204 |
+
{
|
| 11205 |
+
"epoch": 0.13235719899077636,
|
| 11206 |
+
"grad_norm": 1241.7110595703125,
|
| 11207 |
+
"learning_rate": 9.973884285298932e-06,
|
| 11208 |
+
"loss": 154.7749,
|
| 11209 |
+
"step": 16000
|
| 11210 |
}
|
| 11211 |
],
|
| 11212 |
"logging_steps": 10,
|