w06618pm commited on
Commit ·
7157452
1
Parent(s): ef8b14e
update to tuned model
Browse files- {checkpoint-492 → checkpoint-16000}/config.json +0 -0
- {checkpoint-492 → checkpoint-16000}/optimizer.pt +1 -1
- {checkpoint-492 → checkpoint-16000}/pytorch_model.bin +1 -1
- {checkpoint-492 → checkpoint-16000}/rng_state.pth +1 -1
- {checkpoint-492 → checkpoint-16000}/scheduler.pt +1 -1
- {checkpoint-492 → checkpoint-16000}/special_tokens_map.json +0 -0
- {checkpoint-492 → checkpoint-16000}/tokenizer.json +0 -0
- {checkpoint-492 → checkpoint-16000}/tokenizer_config.json +0 -0
- checkpoint-16000/trainer_state.json +1226 -0
- checkpoint-492/trainer_state.json +0 -97
- pytorch_model.bin +1 -1
{checkpoint-492 → checkpoint-16000}/config.json
RENAMED
|
File without changes
|
{checkpoint-492 → checkpoint-16000}/optimizer.pt
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1033299706
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:7bea97b55325c43b8f3dad204e399582a0cbb3ad5099895d9357b02b15c266c9
|
| 3 |
size 1033299706
|
{checkpoint-492 → checkpoint-16000}/pytorch_model.bin
RENAMED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 516640282
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:829b58ddaceb0fa2c29313fe2d068d2e14ad9c0149c357ae8eb72674d2eddc91
|
| 3 |
size 516640282
|
{checkpoint-492 → checkpoint-16000}/rng_state.pth
RENAMED
|
@@ -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:b1a97db8e41139aa1239ba7fb79ddeb0af5998c6305a440c1fe182e6ad02f2f5
|
| 3 |
size 14244
|
{checkpoint-492 → checkpoint-16000}/scheduler.pt
RENAMED
|
@@ -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:e2e07fd269933d93775fcbd2b65b42a65580350ef79eda4f62b5567059bd0986
|
| 3 |
size 1064
|
{checkpoint-492 → checkpoint-16000}/special_tokens_map.json
RENAMED
|
File without changes
|
{checkpoint-492 → checkpoint-16000}/tokenizer.json
RENAMED
|
File without changes
|
{checkpoint-492 → checkpoint-16000}/tokenizer_config.json
RENAMED
|
File without changes
|
checkpoint-16000/trainer_state.json
ADDED
|
@@ -0,0 +1,1226 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 4000,
|
| 3 |
+
"best_metric": 0.47247838973999023,
|
| 4 |
+
"best_model_checkpoint": "mamba_nli_ensemble/checkpoint-4000",
|
| 5 |
+
"epoch": 2.619515389652914,
|
| 6 |
+
"eval_steps": 2000,
|
| 7 |
+
"global_step": 16000,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.016371971185330715,
|
| 14 |
+
"grad_norm": 29.85440444946289,
|
| 15 |
+
"learning_rate": 1.6371971185330716e-06,
|
| 16 |
+
"loss": 0.7262,
|
| 17 |
+
"step": 100
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.03274394237066143,
|
| 21 |
+
"grad_norm": 63.15359878540039,
|
| 22 |
+
"learning_rate": 3.2743942370661432e-06,
|
| 23 |
+
"loss": 0.69,
|
| 24 |
+
"step": 200
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04911591355599214,
|
| 28 |
+
"grad_norm": 9.709274291992188,
|
| 29 |
+
"learning_rate": 4.911591355599214e-06,
|
| 30 |
+
"loss": 0.6864,
|
| 31 |
+
"step": 300
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.06548788474132286,
|
| 35 |
+
"grad_norm": 12.61182689666748,
|
| 36 |
+
"learning_rate": 6.5487884741322864e-06,
|
| 37 |
+
"loss": 0.6999,
|
| 38 |
+
"step": 400
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08185985592665357,
|
| 42 |
+
"grad_norm": 8.176433563232422,
|
| 43 |
+
"learning_rate": 8.185985592665357e-06,
|
| 44 |
+
"loss": 0.6842,
|
| 45 |
+
"step": 500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.09823182711198428,
|
| 49 |
+
"grad_norm": 2.774385690689087,
|
| 50 |
+
"learning_rate": 9.823182711198428e-06,
|
| 51 |
+
"loss": 0.6686,
|
| 52 |
+
"step": 600
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.114603798297315,
|
| 56 |
+
"grad_norm": 5.177670955657959,
|
| 57 |
+
"learning_rate": 1.14603798297315e-05,
|
| 58 |
+
"loss": 0.6835,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.13097576948264572,
|
| 63 |
+
"grad_norm": 3.342498302459717,
|
| 64 |
+
"learning_rate": 1.3097576948264573e-05,
|
| 65 |
+
"loss": 0.6631,
|
| 66 |
+
"step": 800
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.14734774066797643,
|
| 70 |
+
"grad_norm": 4.506181716918945,
|
| 71 |
+
"learning_rate": 1.4734774066797644e-05,
|
| 72 |
+
"loss": 0.6665,
|
| 73 |
+
"step": 900
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.16371971185330714,
|
| 77 |
+
"grad_norm": 6.701014995574951,
|
| 78 |
+
"learning_rate": 1.6371971185330713e-05,
|
| 79 |
+
"loss": 0.6644,
|
| 80 |
+
"step": 1000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.18009168303863785,
|
| 84 |
+
"grad_norm": 5.5269246101379395,
|
| 85 |
+
"learning_rate": 1.8009168303863786e-05,
|
| 86 |
+
"loss": 0.651,
|
| 87 |
+
"step": 1100
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.19646365422396855,
|
| 91 |
+
"grad_norm": 2.700180768966675,
|
| 92 |
+
"learning_rate": 1.9646365422396855e-05,
|
| 93 |
+
"loss": 0.6115,
|
| 94 |
+
"step": 1200
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.2128356254092993,
|
| 98 |
+
"grad_norm": 10.621367454528809,
|
| 99 |
+
"learning_rate": 2.128356254092993e-05,
|
| 100 |
+
"loss": 0.6403,
|
| 101 |
+
"step": 1300
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.22920759659463,
|
| 105 |
+
"grad_norm": 2.2622969150543213,
|
| 106 |
+
"learning_rate": 2.2920759659463e-05,
|
| 107 |
+
"loss": 0.5914,
|
| 108 |
+
"step": 1400
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.2455795677799607,
|
| 112 |
+
"grad_norm": 4.627520561218262,
|
| 113 |
+
"learning_rate": 2.4557956777996073e-05,
|
| 114 |
+
"loss": 0.6149,
|
| 115 |
+
"step": 1500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.26195153896529144,
|
| 119 |
+
"grad_norm": 5.510996341705322,
|
| 120 |
+
"learning_rate": 2.6195153896529146e-05,
|
| 121 |
+
"loss": 0.6149,
|
| 122 |
+
"step": 1600
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.2783235101506221,
|
| 126 |
+
"grad_norm": 6.97050142288208,
|
| 127 |
+
"learning_rate": 2.7832351015062215e-05,
|
| 128 |
+
"loss": 0.6256,
|
| 129 |
+
"step": 1700
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.29469548133595286,
|
| 133 |
+
"grad_norm": 2.5861706733703613,
|
| 134 |
+
"learning_rate": 2.9469548133595288e-05,
|
| 135 |
+
"loss": 0.5884,
|
| 136 |
+
"step": 1800
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.31106745252128354,
|
| 140 |
+
"grad_norm": 12.085710525512695,
|
| 141 |
+
"learning_rate": 3.110674525212836e-05,
|
| 142 |
+
"loss": 0.5986,
|
| 143 |
+
"step": 1900
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.3274394237066143,
|
| 147 |
+
"grad_norm": 9.400935173034668,
|
| 148 |
+
"learning_rate": 3.2743942370661426e-05,
|
| 149 |
+
"loss": 0.6357,
|
| 150 |
+
"step": 2000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.3274394237066143,
|
| 154 |
+
"eval_accuracy": 0.7325408618127786,
|
| 155 |
+
"eval_loss": 0.5683333277702332,
|
| 156 |
+
"eval_runtime": 5.8436,
|
| 157 |
+
"eval_samples_per_second": 115.17,
|
| 158 |
+
"eval_steps_per_second": 7.359,
|
| 159 |
+
"step": 2000
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.343811394891945,
|
| 163 |
+
"grad_norm": 4.135758399963379,
|
| 164 |
+
"learning_rate": 3.43811394891945e-05,
|
| 165 |
+
"loss": 0.5574,
|
| 166 |
+
"step": 2100
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.3601833660772757,
|
| 170 |
+
"grad_norm": 8.234903335571289,
|
| 171 |
+
"learning_rate": 3.601833660772757e-05,
|
| 172 |
+
"loss": 0.5793,
|
| 173 |
+
"step": 2200
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.3765553372626064,
|
| 177 |
+
"grad_norm": 11.848236083984375,
|
| 178 |
+
"learning_rate": 3.765553372626065e-05,
|
| 179 |
+
"loss": 0.5812,
|
| 180 |
+
"step": 2300
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.3929273084479371,
|
| 184 |
+
"grad_norm": 15.417450904846191,
|
| 185 |
+
"learning_rate": 3.929273084479371e-05,
|
| 186 |
+
"loss": 0.5165,
|
| 187 |
+
"step": 2400
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.40929927963326784,
|
| 191 |
+
"grad_norm": 170.8882293701172,
|
| 192 |
+
"learning_rate": 4.0929927963326786e-05,
|
| 193 |
+
"loss": 0.6833,
|
| 194 |
+
"step": 2500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.4256712508185986,
|
| 198 |
+
"grad_norm": 256.7959289550781,
|
| 199 |
+
"learning_rate": 4.256712508185986e-05,
|
| 200 |
+
"loss": 0.6222,
|
| 201 |
+
"step": 2600
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.44204322200392926,
|
| 205 |
+
"grad_norm": 7.283181667327881,
|
| 206 |
+
"learning_rate": 4.4204322200392925e-05,
|
| 207 |
+
"loss": 0.6134,
|
| 208 |
+
"step": 2700
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.45841519318926,
|
| 212 |
+
"grad_norm": 3.3215067386627197,
|
| 213 |
+
"learning_rate": 4.5841519318926e-05,
|
| 214 |
+
"loss": 0.5355,
|
| 215 |
+
"step": 2800
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.4747871643745907,
|
| 219 |
+
"grad_norm": 6.261496543884277,
|
| 220 |
+
"learning_rate": 4.747871643745907e-05,
|
| 221 |
+
"loss": 0.5777,
|
| 222 |
+
"step": 2900
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.4911591355599214,
|
| 226 |
+
"grad_norm": 5.140758991241455,
|
| 227 |
+
"learning_rate": 4.9115913555992146e-05,
|
| 228 |
+
"loss": 0.586,
|
| 229 |
+
"step": 3000
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.5075311067452521,
|
| 233 |
+
"grad_norm": 4.261934280395508,
|
| 234 |
+
"learning_rate": 4.999965445760666e-05,
|
| 235 |
+
"loss": 0.6135,
|
| 236 |
+
"step": 3100
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.5239030779305829,
|
| 240 |
+
"grad_norm": 3.795130968093872,
|
| 241 |
+
"learning_rate": 4.999651917405523e-05,
|
| 242 |
+
"loss": 0.6042,
|
| 243 |
+
"step": 3200
|
| 244 |
+
},
|
| 245 |
+
{
|
| 246 |
+
"epoch": 0.5402750491159135,
|
| 247 |
+
"grad_norm": 5.0412116050720215,
|
| 248 |
+
"learning_rate": 4.999011837711028e-05,
|
| 249 |
+
"loss": 0.5806,
|
| 250 |
+
"step": 3300
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"epoch": 0.5566470203012442,
|
| 254 |
+
"grad_norm": 10.38508415222168,
|
| 255 |
+
"learning_rate": 4.998045290296376e-05,
|
| 256 |
+
"loss": 0.4947,
|
| 257 |
+
"step": 3400
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.573018991486575,
|
| 261 |
+
"grad_norm": 38.63774108886719,
|
| 262 |
+
"learning_rate": 4.9967524014300896e-05,
|
| 263 |
+
"loss": 0.5181,
|
| 264 |
+
"step": 3500
|
| 265 |
+
},
|
| 266 |
+
{
|
| 267 |
+
"epoch": 0.5893909626719057,
|
| 268 |
+
"grad_norm": 26.06133460998535,
|
| 269 |
+
"learning_rate": 4.995133340013522e-05,
|
| 270 |
+
"loss": 0.5092,
|
| 271 |
+
"step": 3600
|
| 272 |
+
},
|
| 273 |
+
{
|
| 274 |
+
"epoch": 0.6057629338572365,
|
| 275 |
+
"grad_norm": 17.468584060668945,
|
| 276 |
+
"learning_rate": 4.993188317558791e-05,
|
| 277 |
+
"loss": 0.6309,
|
| 278 |
+
"step": 3700
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.6221349050425671,
|
| 282 |
+
"grad_norm": 2.848097562789917,
|
| 283 |
+
"learning_rate": 4.9909175881611514e-05,
|
| 284 |
+
"loss": 0.496,
|
| 285 |
+
"step": 3800
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.6385068762278978,
|
| 289 |
+
"grad_norm": 13.13325309753418,
|
| 290 |
+
"learning_rate": 4.9883214484657957e-05,
|
| 291 |
+
"loss": 0.5044,
|
| 292 |
+
"step": 3900
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 0.6548788474132285,
|
| 296 |
+
"grad_norm": 10.311493873596191,
|
| 297 |
+
"learning_rate": 4.9854002376291046e-05,
|
| 298 |
+
"loss": 0.6015,
|
| 299 |
+
"step": 4000
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 0.6548788474132285,
|
| 303 |
+
"eval_accuracy": 0.7934621099554234,
|
| 304 |
+
"eval_loss": 0.47247838973999023,
|
| 305 |
+
"eval_runtime": 5.8787,
|
| 306 |
+
"eval_samples_per_second": 114.482,
|
| 307 |
+
"eval_steps_per_second": 7.315,
|
| 308 |
+
"step": 4000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.6712508185985593,
|
| 312 |
+
"grad_norm": 47.73946762084961,
|
| 313 |
+
"learning_rate": 4.9821543372743355e-05,
|
| 314 |
+
"loss": 0.4682,
|
| 315 |
+
"step": 4100
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.68762278978389,
|
| 319 |
+
"grad_norm": 14.78221321105957,
|
| 320 |
+
"learning_rate": 4.9785841714417734e-05,
|
| 321 |
+
"loss": 0.5595,
|
| 322 |
+
"step": 4200
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.7039947609692206,
|
| 326 |
+
"grad_norm": 11.297691345214844,
|
| 327 |
+
"learning_rate": 4.97469020653333e-05,
|
| 328 |
+
"loss": 0.5608,
|
| 329 |
+
"step": 4300
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 0.7203667321545514,
|
| 333 |
+
"grad_norm": 29.37626075744629,
|
| 334 |
+
"learning_rate": 4.970472951251617e-05,
|
| 335 |
+
"loss": 0.5414,
|
| 336 |
+
"step": 4400
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 0.7367387033398821,
|
| 340 |
+
"grad_norm": 11.616437911987305,
|
| 341 |
+
"learning_rate": 4.9659329565334854e-05,
|
| 342 |
+
"loss": 0.5016,
|
| 343 |
+
"step": 4500
|
| 344 |
+
},
|
| 345 |
+
{
|
| 346 |
+
"epoch": 0.7531106745252129,
|
| 347 |
+
"grad_norm": 8.013294219970703,
|
| 348 |
+
"learning_rate": 4.9610708154780585e-05,
|
| 349 |
+
"loss": 0.5775,
|
| 350 |
+
"step": 4600
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.7694826457105436,
|
| 354 |
+
"grad_norm": 5.327792167663574,
|
| 355 |
+
"learning_rate": 4.955887163269243e-05,
|
| 356 |
+
"loss": 0.5126,
|
| 357 |
+
"step": 4700
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.7858546168958742,
|
| 361 |
+
"grad_norm": 1.6188075542449951,
|
| 362 |
+
"learning_rate": 4.950382677092754e-05,
|
| 363 |
+
"loss": 0.5971,
|
| 364 |
+
"step": 4800
|
| 365 |
+
},
|
| 366 |
+
{
|
| 367 |
+
"epoch": 0.802226588081205,
|
| 368 |
+
"grad_norm": 6.731476306915283,
|
| 369 |
+
"learning_rate": 4.944558076047649e-05,
|
| 370 |
+
"loss": 0.5159,
|
| 371 |
+
"step": 4900
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.8185985592665357,
|
| 375 |
+
"grad_norm": 1.4254143238067627,
|
| 376 |
+
"learning_rate": 4.9384141210523804e-05,
|
| 377 |
+
"loss": 0.5246,
|
| 378 |
+
"step": 5000
|
| 379 |
+
},
|
| 380 |
+
{
|
| 381 |
+
"epoch": 0.8349705304518664,
|
| 382 |
+
"grad_norm": 44.3812370300293,
|
| 383 |
+
"learning_rate": 4.931951614745395e-05,
|
| 384 |
+
"loss": 0.5282,
|
| 385 |
+
"step": 5100
|
| 386 |
+
},
|
| 387 |
+
{
|
| 388 |
+
"epoch": 0.8513425016371972,
|
| 389 |
+
"grad_norm": 2.9070582389831543,
|
| 390 |
+
"learning_rate": 4.925171401380278e-05,
|
| 391 |
+
"loss": 0.5517,
|
| 392 |
+
"step": 5200
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.8677144728225278,
|
| 396 |
+
"grad_norm": 33.969303131103516,
|
| 397 |
+
"learning_rate": 4.918074366715457e-05,
|
| 398 |
+
"loss": 0.5576,
|
| 399 |
+
"step": 5300
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.8840864440078585,
|
| 403 |
+
"grad_norm": 1.7158643007278442,
|
| 404 |
+
"learning_rate": 4.910661437898493e-05,
|
| 405 |
+
"loss": 0.6597,
|
| 406 |
+
"step": 5400
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9004584151931893,
|
| 410 |
+
"grad_norm": 42.425262451171875,
|
| 411 |
+
"learning_rate": 4.902933583344954e-05,
|
| 412 |
+
"loss": 0.5421,
|
| 413 |
+
"step": 5500
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.91683038637852,
|
| 417 |
+
"grad_norm": 9.340117454528809,
|
| 418 |
+
"learning_rate": 4.8948918126119056e-05,
|
| 419 |
+
"loss": 0.5591,
|
| 420 |
+
"step": 5600
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9332023575638507,
|
| 424 |
+
"grad_norm": 1.2870229482650757,
|
| 425 |
+
"learning_rate": 4.886537176266024e-05,
|
| 426 |
+
"loss": 0.5126,
|
| 427 |
+
"step": 5700
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9495743287491814,
|
| 431 |
+
"grad_norm": 36.61764907836914,
|
| 432 |
+
"learning_rate": 4.877870765746347e-05,
|
| 433 |
+
"loss": 0.5015,
|
| 434 |
+
"step": 5800
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9659462999345121,
|
| 438 |
+
"grad_norm": 13.943790435791016,
|
| 439 |
+
"learning_rate": 4.8688937132216966e-05,
|
| 440 |
+
"loss": 0.4911,
|
| 441 |
+
"step": 5900
|
| 442 |
+
},
|
| 443 |
+
{
|
| 444 |
+
"epoch": 0.9823182711198428,
|
| 445 |
+
"grad_norm": 18.877832412719727,
|
| 446 |
+
"learning_rate": 4.859607191442768e-05,
|
| 447 |
+
"loss": 0.5838,
|
| 448 |
+
"step": 6000
|
| 449 |
+
},
|
| 450 |
+
{
|
| 451 |
+
"epoch": 0.9823182711198428,
|
| 452 |
+
"eval_accuracy": 0.787518573551263,
|
| 453 |
+
"eval_loss": 0.5758853554725647,
|
| 454 |
+
"eval_runtime": 5.9463,
|
| 455 |
+
"eval_samples_per_second": 113.18,
|
| 456 |
+
"eval_steps_per_second": 7.231,
|
| 457 |
+
"step": 6000
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.9986902423051736,
|
| 461 |
+
"grad_norm": 24.715587615966797,
|
| 462 |
+
"learning_rate": 4.850012413588926e-05,
|
| 463 |
+
"loss": 0.4975,
|
| 464 |
+
"step": 6100
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.0150622134905043,
|
| 468 |
+
"grad_norm": 13.0919828414917,
|
| 469 |
+
"learning_rate": 4.840110633109716e-05,
|
| 470 |
+
"loss": 0.511,
|
| 471 |
+
"step": 6200
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 1.031434184675835,
|
| 475 |
+
"grad_norm": 0.9892318844795227,
|
| 476 |
+
"learning_rate": 4.829903143561113e-05,
|
| 477 |
+
"loss": 0.539,
|
| 478 |
+
"step": 6300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0478061558611658,
|
| 482 |
+
"grad_norm": 18.864852905273438,
|
| 483 |
+
"learning_rate": 4.819391278436539e-05,
|
| 484 |
+
"loss": 0.4764,
|
| 485 |
+
"step": 6400
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.0641781270464965,
|
| 489 |
+
"grad_norm": 0.43955135345458984,
|
| 490 |
+
"learning_rate": 4.8085764109926494e-05,
|
| 491 |
+
"loss": 0.3564,
|
| 492 |
+
"step": 6500
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.080550098231827,
|
| 496 |
+
"grad_norm": 0.3193933963775635,
|
| 497 |
+
"learning_rate": 4.7974599540699386e-05,
|
| 498 |
+
"loss": 0.4985,
|
| 499 |
+
"step": 6600
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.0969220694171578,
|
| 503 |
+
"grad_norm": 17.164764404296875,
|
| 504 |
+
"learning_rate": 4.7860433599081654e-05,
|
| 505 |
+
"loss": 0.4675,
|
| 506 |
+
"step": 6700
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1132940406024885,
|
| 510 |
+
"grad_norm": 0.31629478931427,
|
| 511 |
+
"learning_rate": 4.774328119956633e-05,
|
| 512 |
+
"loss": 0.5588,
|
| 513 |
+
"step": 6800
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.1296660117878192,
|
| 517 |
+
"grad_norm": 8.383326530456543,
|
| 518 |
+
"learning_rate": 4.762315764679353e-05,
|
| 519 |
+
"loss": 0.4681,
|
| 520 |
+
"step": 6900
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 1.14603798297315,
|
| 524 |
+
"grad_norm": 0.6111562252044678,
|
| 525 |
+
"learning_rate": 4.750007863355102e-05,
|
| 526 |
+
"loss": 0.4538,
|
| 527 |
+
"step": 7000
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 1.1624099541584807,
|
| 531 |
+
"grad_norm": 12.500487327575684,
|
| 532 |
+
"learning_rate": 4.737406023872416e-05,
|
| 533 |
+
"loss": 0.445,
|
| 534 |
+
"step": 7100
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 1.1787819253438114,
|
| 538 |
+
"grad_norm": 0.2397201806306839,
|
| 539 |
+
"learning_rate": 4.7245118925195374e-05,
|
| 540 |
+
"loss": 0.4252,
|
| 541 |
+
"step": 7200
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 1.1951538965291422,
|
| 545 |
+
"grad_norm": 22.245431900024414,
|
| 546 |
+
"learning_rate": 4.7113271537693454e-05,
|
| 547 |
+
"loss": 0.4567,
|
| 548 |
+
"step": 7300
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 1.211525867714473,
|
| 552 |
+
"grad_norm": 0.6519126892089844,
|
| 553 |
+
"learning_rate": 4.6978535300593e-05,
|
| 554 |
+
"loss": 0.5072,
|
| 555 |
+
"step": 7400
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 1.2278978388998034,
|
| 559 |
+
"grad_norm": 0.5901318192481995,
|
| 560 |
+
"learning_rate": 4.684092781566422e-05,
|
| 561 |
+
"loss": 0.3924,
|
| 562 |
+
"step": 7500
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 1.2442698100851342,
|
| 566 |
+
"grad_norm": 12.497309684753418,
|
| 567 |
+
"learning_rate": 4.67004670597735e-05,
|
| 568 |
+
"loss": 0.5874,
|
| 569 |
+
"step": 7600
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 1.260641781270465,
|
| 573 |
+
"grad_norm": 2.344989538192749,
|
| 574 |
+
"learning_rate": 4.6557171382534915e-05,
|
| 575 |
+
"loss": 0.4327,
|
| 576 |
+
"step": 7700
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 1.2770137524557956,
|
| 580 |
+
"grad_norm": 17.146575927734375,
|
| 581 |
+
"learning_rate": 4.6411059503913e-05,
|
| 582 |
+
"loss": 0.5129,
|
| 583 |
+
"step": 7800
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 1.2933857236411264,
|
| 587 |
+
"grad_norm": 1.8164324760437012,
|
| 588 |
+
"learning_rate": 4.62621505117773e-05,
|
| 589 |
+
"loss": 0.5208,
|
| 590 |
+
"step": 7900
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 1.309757694826457,
|
| 594 |
+
"grad_norm": 13.502025604248047,
|
| 595 |
+
"learning_rate": 4.611046385940868e-05,
|
| 596 |
+
"loss": 0.4884,
|
| 597 |
+
"step": 8000
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 1.309757694826457,
|
| 601 |
+
"eval_accuracy": 0.7815750371471025,
|
| 602 |
+
"eval_loss": 0.6778917908668518,
|
| 603 |
+
"eval_runtime": 5.9221,
|
| 604 |
+
"eval_samples_per_second": 113.642,
|
| 605 |
+
"eval_steps_per_second": 7.261,
|
| 606 |
+
"step": 8000
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 1.3261296660117878,
|
| 610 |
+
"grad_norm": 8.141022682189941,
|
| 611 |
+
"learning_rate": 4.5956019362958006e-05,
|
| 612 |
+
"loss": 0.4438,
|
| 613 |
+
"step": 8100
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 1.3425016371971186,
|
| 617 |
+
"grad_norm": 0.91554856300354,
|
| 618 |
+
"learning_rate": 4.5798837198857356e-05,
|
| 619 |
+
"loss": 0.4617,
|
| 620 |
+
"step": 8200
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 1.3588736083824493,
|
| 624 |
+
"grad_norm": 0.6429493427276611,
|
| 625 |
+
"learning_rate": 4.563893790118426e-05,
|
| 626 |
+
"loss": 0.4947,
|
| 627 |
+
"step": 8300
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 1.37524557956778,
|
| 631 |
+
"grad_norm": 0.22166290879249573,
|
| 632 |
+
"learning_rate": 4.547634235897906e-05,
|
| 633 |
+
"loss": 0.4057,
|
| 634 |
+
"step": 8400
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 1.3916175507531108,
|
| 638 |
+
"grad_norm": 9.768654823303223,
|
| 639 |
+
"learning_rate": 4.5311071813516106e-05,
|
| 640 |
+
"loss": 0.3861,
|
| 641 |
+
"step": 8500
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 1.4079895219384415,
|
| 645 |
+
"grad_norm": 0.4799940586090088,
|
| 646 |
+
"learning_rate": 4.514314785552871e-05,
|
| 647 |
+
"loss": 0.4809,
|
| 648 |
+
"step": 8600
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.424361493123772,
|
| 652 |
+
"grad_norm": 11.279892921447754,
|
| 653 |
+
"learning_rate": 4.4972592422388634e-05,
|
| 654 |
+
"loss": 0.4472,
|
| 655 |
+
"step": 8700
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.4407334643091028,
|
| 659 |
+
"grad_norm": 0.24561847746372223,
|
| 660 |
+
"learning_rate": 4.479942779524022e-05,
|
| 661 |
+
"loss": 0.5542,
|
| 662 |
+
"step": 8800
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.4571054354944335,
|
| 666 |
+
"grad_norm": 0.3300779163837433,
|
| 667 |
+
"learning_rate": 4.462367659608955e-05,
|
| 668 |
+
"loss": 0.427,
|
| 669 |
+
"step": 8900
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 1.4734774066797642,
|
| 673 |
+
"grad_norm": 25.95767593383789,
|
| 674 |
+
"learning_rate": 4.4445361784849195e-05,
|
| 675 |
+
"loss": 0.4385,
|
| 676 |
+
"step": 9000
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 1.489849377865095,
|
| 680 |
+
"grad_norm": 7.935421466827393,
|
| 681 |
+
"learning_rate": 4.4264506656338745e-05,
|
| 682 |
+
"loss": 0.4963,
|
| 683 |
+
"step": 9100
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 1.5062213490504257,
|
| 687 |
+
"grad_norm": 0.8363397717475891,
|
| 688 |
+
"learning_rate": 4.4081134837241585e-05,
|
| 689 |
+
"loss": 0.4049,
|
| 690 |
+
"step": 9200
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 1.5225933202357562,
|
| 694 |
+
"grad_norm": 0.5877137780189514,
|
| 695 |
+
"learning_rate": 4.389527028301836e-05,
|
| 696 |
+
"loss": 0.4596,
|
| 697 |
+
"step": 9300
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 1.538965291421087,
|
| 701 |
+
"grad_norm": 7.968419075012207,
|
| 702 |
+
"learning_rate": 4.370693727477745e-05,
|
| 703 |
+
"loss": 0.5687,
|
| 704 |
+
"step": 9400
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 1.5553372626064177,
|
| 708 |
+
"grad_norm": 26.651159286499023,
|
| 709 |
+
"learning_rate": 4.351616041610292e-05,
|
| 710 |
+
"loss": 0.5284,
|
| 711 |
+
"step": 9500
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 1.5717092337917484,
|
| 715 |
+
"grad_norm": 7.4741129875183105,
|
| 716 |
+
"learning_rate": 4.3322964629840344e-05,
|
| 717 |
+
"loss": 0.4048,
|
| 718 |
+
"step": 9600
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 1.5880812049770792,
|
| 722 |
+
"grad_norm": 0.2695929706096649,
|
| 723 |
+
"learning_rate": 4.312737515484091e-05,
|
| 724 |
+
"loss": 0.4803,
|
| 725 |
+
"step": 9700
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 1.60445317616241,
|
| 729 |
+
"grad_norm": 17.771469116210938,
|
| 730 |
+
"learning_rate": 4.2929417542664244e-05,
|
| 731 |
+
"loss": 0.5075,
|
| 732 |
+
"step": 9800
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.6208251473477406,
|
| 736 |
+
"grad_norm": 15.19389820098877,
|
| 737 |
+
"learning_rate": 4.272911765424039e-05,
|
| 738 |
+
"loss": 0.4512,
|
| 739 |
+
"step": 9900
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 1.6371971185330714,
|
| 743 |
+
"grad_norm": 0.9662685394287109,
|
| 744 |
+
"learning_rate": 4.2526501656491405e-05,
|
| 745 |
+
"loss": 0.4284,
|
| 746 |
+
"step": 10000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 1.6371971185330714,
|
| 750 |
+
"eval_accuracy": 0.8053491827637445,
|
| 751 |
+
"eval_loss": 0.6046619415283203,
|
| 752 |
+
"eval_runtime": 5.9328,
|
| 753 |
+
"eval_samples_per_second": 113.438,
|
| 754 |
+
"eval_steps_per_second": 7.248,
|
| 755 |
+
"step": 10000
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.653569089718402,
|
| 759 |
+
"grad_norm": 6.945915222167969,
|
| 760 |
+
"learning_rate": 4.232159601891287e-05,
|
| 761 |
+
"loss": 0.4298,
|
| 762 |
+
"step": 10100
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.6699410609037328,
|
| 766 |
+
"grad_norm": 0.882854700088501,
|
| 767 |
+
"learning_rate": 4.2114427510116036e-05,
|
| 768 |
+
"loss": 0.4362,
|
| 769 |
+
"step": 10200
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 1.6863130320890636,
|
| 773 |
+
"grad_norm": 19.653860092163086,
|
| 774 |
+
"learning_rate": 4.1905023194330726e-05,
|
| 775 |
+
"loss": 0.4404,
|
| 776 |
+
"step": 10300
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 1.7026850032743943,
|
| 780 |
+
"grad_norm": 26.812660217285156,
|
| 781 |
+
"learning_rate": 4.169341042786977e-05,
|
| 782 |
+
"loss": 0.513,
|
| 783 |
+
"step": 10400
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.719056974459725,
|
| 787 |
+
"grad_norm": 0.45196932554244995,
|
| 788 |
+
"learning_rate": 4.147961685555517e-05,
|
| 789 |
+
"loss": 0.3751,
|
| 790 |
+
"step": 10500
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 1.7354289456450558,
|
| 794 |
+
"grad_norm": 14.361971855163574,
|
| 795 |
+
"learning_rate": 4.12636704071066e-05,
|
| 796 |
+
"loss": 0.4847,
|
| 797 |
+
"step": 10600
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 1.7518009168303865,
|
| 801 |
+
"grad_norm": 1.6658482551574707,
|
| 802 |
+
"learning_rate": 4.104559929349277e-05,
|
| 803 |
+
"loss": 0.3794,
|
| 804 |
+
"step": 10700
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 1.768172888015717,
|
| 808 |
+
"grad_norm": 37.13115692138672,
|
| 809 |
+
"learning_rate": 4.08254320032459e-05,
|
| 810 |
+
"loss": 0.423,
|
| 811 |
+
"step": 10800
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 1.7845448592010478,
|
| 815 |
+
"grad_norm": 0.23938187956809998,
|
| 816 |
+
"learning_rate": 4.060319729874007e-05,
|
| 817 |
+
"loss": 0.4097,
|
| 818 |
+
"step": 10900
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.8009168303863785,
|
| 822 |
+
"grad_norm": 4.112476825714111,
|
| 823 |
+
"learning_rate": 4.0378924212433715e-05,
|
| 824 |
+
"loss": 0.472,
|
| 825 |
+
"step": 11000
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.8172888015717092,
|
| 829 |
+
"grad_norm": 14.257973670959473,
|
| 830 |
+
"learning_rate": 4.0152642043076884e-05,
|
| 831 |
+
"loss": 0.4984,
|
| 832 |
+
"step": 11100
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 1.83366077275704,
|
| 836 |
+
"grad_norm": 13.519189834594727,
|
| 837 |
+
"learning_rate": 3.992438035188366e-05,
|
| 838 |
+
"loss": 0.3799,
|
| 839 |
+
"step": 11200
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 1.8500327439423707,
|
| 843 |
+
"grad_norm": 3.054865837097168,
|
| 844 |
+
"learning_rate": 3.969416895867034e-05,
|
| 845 |
+
"loss": 0.4403,
|
| 846 |
+
"step": 11300
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.8664047151277012,
|
| 850 |
+
"grad_norm": 2.4153125286102295,
|
| 851 |
+
"learning_rate": 3.946203793795982e-05,
|
| 852 |
+
"loss": 0.4938,
|
| 853 |
+
"step": 11400
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.882776686313032,
|
| 857 |
+
"grad_norm": 0.34407490491867065,
|
| 858 |
+
"learning_rate": 3.922801761505264e-05,
|
| 859 |
+
"loss": 0.4565,
|
| 860 |
+
"step": 11500
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.8991486574983627,
|
| 864 |
+
"grad_norm": 19.420068740844727,
|
| 865 |
+
"learning_rate": 3.8992138562065415e-05,
|
| 866 |
+
"loss": 0.4581,
|
| 867 |
+
"step": 11600
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.9155206286836934,
|
| 871 |
+
"grad_norm": 23.287689208984375,
|
| 872 |
+
"learning_rate": 3.875443159393689e-05,
|
| 873 |
+
"loss": 0.3939,
|
| 874 |
+
"step": 11700
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.9318925998690242,
|
| 878 |
+
"grad_norm": 31.722305297851562,
|
| 879 |
+
"learning_rate": 3.8514927764402274e-05,
|
| 880 |
+
"loss": 0.5128,
|
| 881 |
+
"step": 11800
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 1.948264571054355,
|
| 885 |
+
"grad_norm": 175.8199920654297,
|
| 886 |
+
"learning_rate": 3.8273658361936505e-05,
|
| 887 |
+
"loss": 0.397,
|
| 888 |
+
"step": 11900
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 1.9646365422396856,
|
| 892 |
+
"grad_norm": 9.348654747009277,
|
| 893 |
+
"learning_rate": 3.803065490566667e-05,
|
| 894 |
+
"loss": 0.4633,
|
| 895 |
+
"step": 12000
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 1.9646365422396856,
|
| 899 |
+
"eval_accuracy": 0.8172362555720654,
|
| 900 |
+
"eval_loss": 0.6007680296897888,
|
| 901 |
+
"eval_runtime": 5.8584,
|
| 902 |
+
"eval_samples_per_second": 114.878,
|
| 903 |
+
"eval_steps_per_second": 7.34,
|
| 904 |
+
"step": 12000
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.9810085134250164,
|
| 908 |
+
"grad_norm": 0.6896190047264099,
|
| 909 |
+
"learning_rate": 3.7785949141254475e-05,
|
| 910 |
+
"loss": 0.3702,
|
| 911 |
+
"step": 12100
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.9973804846103471,
|
| 915 |
+
"grad_norm": 3.107114315032959,
|
| 916 |
+
"learning_rate": 3.753957303674897e-05,
|
| 917 |
+
"loss": 0.4142,
|
| 918 |
+
"step": 12200
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 2.013752455795678,
|
| 922 |
+
"grad_norm": 0.11377547681331635,
|
| 923 |
+
"learning_rate": 3.7291558778410314e-05,
|
| 924 |
+
"loss": 0.2155,
|
| 925 |
+
"step": 12300
|
| 926 |
+
},
|
| 927 |
+
{
|
| 928 |
+
"epoch": 2.0301244269810086,
|
| 929 |
+
"grad_norm": 1.9735820293426514,
|
| 930 |
+
"learning_rate": 3.704193876650499e-05,
|
| 931 |
+
"loss": 0.2937,
|
| 932 |
+
"step": 12400
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 2.0464963981663393,
|
| 936 |
+
"grad_norm": 0.07780765742063522,
|
| 937 |
+
"learning_rate": 3.6790745611073065e-05,
|
| 938 |
+
"loss": 0.189,
|
| 939 |
+
"step": 12500
|
| 940 |
+
},
|
| 941 |
+
{
|
| 942 |
+
"epoch": 2.06286836935167,
|
| 943 |
+
"grad_norm": 10.954134941101074,
|
| 944 |
+
"learning_rate": 3.65380121276681e-05,
|
| 945 |
+
"loss": 0.2428,
|
| 946 |
+
"step": 12600
|
| 947 |
+
},
|
| 948 |
+
{
|
| 949 |
+
"epoch": 2.079240340537001,
|
| 950 |
+
"grad_norm": 0.05607511103153229,
|
| 951 |
+
"learning_rate": 3.6283771333070127e-05,
|
| 952 |
+
"loss": 0.2512,
|
| 953 |
+
"step": 12700
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 2.0956123117223315,
|
| 957 |
+
"grad_norm": 16.878358840942383,
|
| 958 |
+
"learning_rate": 3.6028056440972374e-05,
|
| 959 |
+
"loss": 0.226,
|
| 960 |
+
"step": 12800
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 2.1119842829076623,
|
| 964 |
+
"grad_norm": 1.4940121173858643,
|
| 965 |
+
"learning_rate": 3.5770900857642306e-05,
|
| 966 |
+
"loss": 0.1941,
|
| 967 |
+
"step": 12900
|
| 968 |
+
},
|
| 969 |
+
{
|
| 970 |
+
"epoch": 2.128356254092993,
|
| 971 |
+
"grad_norm": 28.74452781677246,
|
| 972 |
+
"learning_rate": 3.551233817755745e-05,
|
| 973 |
+
"loss": 0.1594,
|
| 974 |
+
"step": 13000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 2.1447282252783237,
|
| 978 |
+
"grad_norm": 0.10467255115509033,
|
| 979 |
+
"learning_rate": 3.525240217901665e-05,
|
| 980 |
+
"loss": 0.3147,
|
| 981 |
+
"step": 13100
|
| 982 |
+
},
|
| 983 |
+
{
|
| 984 |
+
"epoch": 2.161100196463654,
|
| 985 |
+
"grad_norm": 125.1854248046875,
|
| 986 |
+
"learning_rate": 3.499112681972734e-05,
|
| 987 |
+
"loss": 0.3013,
|
| 988 |
+
"step": 13200
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 2.1774721676489848,
|
| 992 |
+
"grad_norm": 0.17910659313201904,
|
| 993 |
+
"learning_rate": 3.4728546232369303e-05,
|
| 994 |
+
"loss": 0.163,
|
| 995 |
+
"step": 13300
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 2.1938441388343155,
|
| 999 |
+
"grad_norm": 0.015883993357419968,
|
| 1000 |
+
"learning_rate": 3.4464694720135695e-05,
|
| 1001 |
+
"loss": 0.258,
|
| 1002 |
+
"step": 13400
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 2.2102161100196462,
|
| 1006 |
+
"grad_norm": 0.01641729474067688,
|
| 1007 |
+
"learning_rate": 3.4199606752251634e-05,
|
| 1008 |
+
"loss": 0.138,
|
| 1009 |
+
"step": 13500
|
| 1010 |
+
},
|
| 1011 |
+
{
|
| 1012 |
+
"epoch": 2.226588081204977,
|
| 1013 |
+
"grad_norm": 0.05746021494269371,
|
| 1014 |
+
"learning_rate": 3.3933316959471265e-05,
|
| 1015 |
+
"loss": 0.2195,
|
| 1016 |
+
"step": 13600
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 2.2429600523903077,
|
| 1020 |
+
"grad_norm": 0.017622973769903183,
|
| 1021 |
+
"learning_rate": 3.3665860129553584e-05,
|
| 1022 |
+
"loss": 0.1332,
|
| 1023 |
+
"step": 13700
|
| 1024 |
+
},
|
| 1025 |
+
{
|
| 1026 |
+
"epoch": 2.2593320235756384,
|
| 1027 |
+
"grad_norm": 0.048372671008110046,
|
| 1028 |
+
"learning_rate": 3.3397271202717834e-05,
|
| 1029 |
+
"loss": 0.2528,
|
| 1030 |
+
"step": 13800
|
| 1031 |
+
},
|
| 1032 |
+
{
|
| 1033 |
+
"epoch": 2.275703994760969,
|
| 1034 |
+
"grad_norm": 0.06655670702457428,
|
| 1035 |
+
"learning_rate": 3.312758526707895e-05,
|
| 1036 |
+
"loss": 0.2649,
|
| 1037 |
+
"step": 13900
|
| 1038 |
+
},
|
| 1039 |
+
{
|
| 1040 |
+
"epoch": 2.2920759659463,
|
| 1041 |
+
"grad_norm": 0.11124414205551147,
|
| 1042 |
+
"learning_rate": 3.285683755406373e-05,
|
| 1043 |
+
"loss": 0.2191,
|
| 1044 |
+
"step": 14000
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 2.2920759659463,
|
| 1048 |
+
"eval_accuracy": 0.8231797919762258,
|
| 1049 |
+
"eval_loss": 0.8087013363838196,
|
| 1050 |
+
"eval_runtime": 5.8945,
|
| 1051 |
+
"eval_samples_per_second": 114.174,
|
| 1052 |
+
"eval_steps_per_second": 7.295,
|
| 1053 |
+
"step": 14000
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.3084479371316307,
|
| 1057 |
+
"grad_norm": 0.172523632645607,
|
| 1058 |
+
"learning_rate": 3.258506343380815e-05,
|
| 1059 |
+
"loss": 0.2064,
|
| 1060 |
+
"step": 14100
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.3248199083169614,
|
| 1064 |
+
"grad_norm": 0.041510824114084244,
|
| 1065 |
+
"learning_rate": 3.23122984105368e-05,
|
| 1066 |
+
"loss": 0.2191,
|
| 1067 |
+
"step": 14200
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.341191879502292,
|
| 1071 |
+
"grad_norm": 0.022055430337786674,
|
| 1072 |
+
"learning_rate": 3.203857811792451e-05,
|
| 1073 |
+
"loss": 0.2203,
|
| 1074 |
+
"step": 14300
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.357563850687623,
|
| 1078 |
+
"grad_norm": 0.09686886519193649,
|
| 1079 |
+
"learning_rate": 3.176393831444131e-05,
|
| 1080 |
+
"loss": 0.271,
|
| 1081 |
+
"step": 14400
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.3739358218729536,
|
| 1085 |
+
"grad_norm": 0.0746513158082962,
|
| 1086 |
+
"learning_rate": 3.148841487868095e-05,
|
| 1087 |
+
"loss": 0.2665,
|
| 1088 |
+
"step": 14500
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.3903077930582843,
|
| 1092 |
+
"grad_norm": 0.01806914061307907,
|
| 1093 |
+
"learning_rate": 3.121204380467379e-05,
|
| 1094 |
+
"loss": 0.1947,
|
| 1095 |
+
"step": 14600
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.406679764243615,
|
| 1099 |
+
"grad_norm": 0.018033646047115326,
|
| 1100 |
+
"learning_rate": 3.093486119718455e-05,
|
| 1101 |
+
"loss": 0.1888,
|
| 1102 |
+
"step": 14700
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.423051735428946,
|
| 1106 |
+
"grad_norm": 0.01413845457136631,
|
| 1107 |
+
"learning_rate": 3.065690326699564e-05,
|
| 1108 |
+
"loss": 0.1409,
|
| 1109 |
+
"step": 14800
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 2.4394237066142765,
|
| 1113 |
+
"grad_norm": 0.12208285927772522,
|
| 1114 |
+
"learning_rate": 3.0378206326176674e-05,
|
| 1115 |
+
"loss": 0.2763,
|
| 1116 |
+
"step": 14900
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 2.455795677799607,
|
| 1120 |
+
"grad_norm": 0.08330941945314407,
|
| 1121 |
+
"learning_rate": 3.0098806783340644e-05,
|
| 1122 |
+
"loss": 0.2277,
|
| 1123 |
+
"step": 15000
|
| 1124 |
+
},
|
| 1125 |
+
{
|
| 1126 |
+
"epoch": 2.4721676489849376,
|
| 1127 |
+
"grad_norm": 0.10370787978172302,
|
| 1128 |
+
"learning_rate": 2.9818741138887584e-05,
|
| 1129 |
+
"loss": 0.2776,
|
| 1130 |
+
"step": 15100
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 2.4885396201702683,
|
| 1134 |
+
"grad_norm": 0.2352355420589447,
|
| 1135 |
+
"learning_rate": 2.9538045980236194e-05,
|
| 1136 |
+
"loss": 0.2935,
|
| 1137 |
+
"step": 15200
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 2.504911591355599,
|
| 1141 |
+
"grad_norm": 2.1543736457824707,
|
| 1142 |
+
"learning_rate": 2.925675797704411e-05,
|
| 1143 |
+
"loss": 0.2717,
|
| 1144 |
+
"step": 15300
|
| 1145 |
+
},
|
| 1146 |
+
{
|
| 1147 |
+
"epoch": 2.52128356254093,
|
| 1148 |
+
"grad_norm": 45.37192916870117,
|
| 1149 |
+
"learning_rate": 2.89749138764174e-05,
|
| 1150 |
+
"loss": 0.1947,
|
| 1151 |
+
"step": 15400
|
| 1152 |
+
},
|
| 1153 |
+
{
|
| 1154 |
+
"epoch": 2.5376555337262605,
|
| 1155 |
+
"grad_norm": 0.01007983461022377,
|
| 1156 |
+
"learning_rate": 2.8692550498110017e-05,
|
| 1157 |
+
"loss": 0.1994,
|
| 1158 |
+
"step": 15500
|
| 1159 |
+
},
|
| 1160 |
+
{
|
| 1161 |
+
"epoch": 2.5540275049115913,
|
| 1162 |
+
"grad_norm": 45.349517822265625,
|
| 1163 |
+
"learning_rate": 2.8409704729713694e-05,
|
| 1164 |
+
"loss": 0.3004,
|
| 1165 |
+
"step": 15600
|
| 1166 |
+
},
|
| 1167 |
+
{
|
| 1168 |
+
"epoch": 2.570399476096922,
|
| 1169 |
+
"grad_norm": 0.0555860809981823,
|
| 1170 |
+
"learning_rate": 2.812641352183897e-05,
|
| 1171 |
+
"loss": 0.235,
|
| 1172 |
+
"step": 15700
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 2.5867714472822527,
|
| 1176 |
+
"grad_norm": 19.971233367919922,
|
| 1177 |
+
"learning_rate": 2.784271388328802e-05,
|
| 1178 |
+
"loss": 0.2062,
|
| 1179 |
+
"step": 15800
|
| 1180 |
+
},
|
| 1181 |
+
{
|
| 1182 |
+
"epoch": 2.6031434184675835,
|
| 1183 |
+
"grad_norm": 0.2190929502248764,
|
| 1184 |
+
"learning_rate": 2.755864287621992e-05,
|
| 1185 |
+
"loss": 0.324,
|
| 1186 |
+
"step": 15900
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 2.619515389652914,
|
| 1190 |
+
"grad_norm": 0.01323526818305254,
|
| 1191 |
+
"learning_rate": 2.7274237611308816e-05,
|
| 1192 |
+
"loss": 0.2028,
|
| 1193 |
+
"step": 16000
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 2.619515389652914,
|
| 1197 |
+
"eval_accuracy": 0.8231797919762258,
|
| 1198 |
+
"eval_loss": 0.8805943727493286,
|
| 1199 |
+
"eval_runtime": 5.9036,
|
| 1200 |
+
"eval_samples_per_second": 113.998,
|
| 1201 |
+
"eval_steps_per_second": 7.284,
|
| 1202 |
+
"step": 16000
|
| 1203 |
+
}
|
| 1204 |
+
],
|
| 1205 |
+
"logging_steps": 100,
|
| 1206 |
+
"max_steps": 30540,
|
| 1207 |
+
"num_input_tokens_seen": 0,
|
| 1208 |
+
"num_train_epochs": 5,
|
| 1209 |
+
"save_steps": 2000,
|
| 1210 |
+
"stateful_callbacks": {
|
| 1211 |
+
"TrainerControl": {
|
| 1212 |
+
"args": {
|
| 1213 |
+
"should_epoch_stop": false,
|
| 1214 |
+
"should_evaluate": false,
|
| 1215 |
+
"should_log": false,
|
| 1216 |
+
"should_save": true,
|
| 1217 |
+
"should_training_stop": false
|
| 1218 |
+
},
|
| 1219 |
+
"attributes": {}
|
| 1220 |
+
}
|
| 1221 |
+
},
|
| 1222 |
+
"total_flos": 0.0,
|
| 1223 |
+
"train_batch_size": 4,
|
| 1224 |
+
"trial_name": null,
|
| 1225 |
+
"trial_params": null
|
| 1226 |
+
}
|
checkpoint-492/trainer_state.json
DELETED
|
@@ -1,97 +0,0 @@
|
|
| 1 |
-
{
|
| 2 |
-
"best_metric": 0.6848214864730835,
|
| 3 |
-
"best_model_checkpoint": "mamba_nli_ensemble/checkpoint-492",
|
| 4 |
-
"epoch": 0.08055009823182711,
|
| 5 |
-
"eval_steps": 123,
|
| 6 |
-
"global_step": 492,
|
| 7 |
-
"is_hyper_param_search": false,
|
| 8 |
-
"is_local_process_zero": true,
|
| 9 |
-
"is_world_process_zero": true,
|
| 10 |
-
"log_history": [
|
| 11 |
-
{
|
| 12 |
-
"epoch": 0.016371971185330715,
|
| 13 |
-
"grad_norm": 14.62908935546875,
|
| 14 |
-
"learning_rate": 4.7131750465091315e-05,
|
| 15 |
-
"loss": 0.5889,
|
| 16 |
-
"step": 100
|
| 17 |
-
},
|
| 18 |
-
{
|
| 19 |
-
"epoch": 0.02013752455795678,
|
| 20 |
-
"eval_accuracy": 0.5074626865671642,
|
| 21 |
-
"eval_loss": 0.8629121780395508,
|
| 22 |
-
"eval_runtime": 0.6645,
|
| 23 |
-
"eval_samples_per_second": 100.835,
|
| 24 |
-
"eval_steps_per_second": 7.525,
|
| 25 |
-
"step": 123
|
| 26 |
-
},
|
| 27 |
-
{
|
| 28 |
-
"epoch": 0.03274394237066143,
|
| 29 |
-
"grad_norm": 14.226408004760742,
|
| 30 |
-
"learning_rate": 3.842639695026885e-05,
|
| 31 |
-
"loss": 0.6064,
|
| 32 |
-
"step": 200
|
| 33 |
-
},
|
| 34 |
-
{
|
| 35 |
-
"epoch": 0.04027504911591356,
|
| 36 |
-
"eval_accuracy": 0.4626865671641791,
|
| 37 |
-
"eval_loss": 0.7645987868309021,
|
| 38 |
-
"eval_runtime": 0.9197,
|
| 39 |
-
"eval_samples_per_second": 72.848,
|
| 40 |
-
"eval_steps_per_second": 5.436,
|
| 41 |
-
"step": 246
|
| 42 |
-
},
|
| 43 |
-
{
|
| 44 |
-
"epoch": 0.04911591355599214,
|
| 45 |
-
"grad_norm": 6.1402587890625,
|
| 46 |
-
"learning_rate": 2.6169867903169525e-05,
|
| 47 |
-
"loss": 0.7895,
|
| 48 |
-
"step": 300
|
| 49 |
-
},
|
| 50 |
-
{
|
| 51 |
-
"epoch": 0.06041257367387033,
|
| 52 |
-
"eval_accuracy": 0.5223880597014925,
|
| 53 |
-
"eval_loss": 0.6901785731315613,
|
| 54 |
-
"eval_runtime": 0.6307,
|
| 55 |
-
"eval_samples_per_second": 106.227,
|
| 56 |
-
"eval_steps_per_second": 7.927,
|
| 57 |
-
"step": 369
|
| 58 |
-
},
|
| 59 |
-
{
|
| 60 |
-
"epoch": 0.06548788474132286,
|
| 61 |
-
"grad_norm": 6.895239353179932,
|
| 62 |
-
"learning_rate": 1.3603918071840487e-05,
|
| 63 |
-
"loss": 0.7144,
|
| 64 |
-
"step": 400
|
| 65 |
-
},
|
| 66 |
-
{
|
| 67 |
-
"epoch": 0.08055009823182711,
|
| 68 |
-
"eval_accuracy": 0.5522388059701493,
|
| 69 |
-
"eval_loss": 0.6848214864730835,
|
| 70 |
-
"eval_runtime": 0.6479,
|
| 71 |
-
"eval_samples_per_second": 103.414,
|
| 72 |
-
"eval_steps_per_second": 7.717,
|
| 73 |
-
"step": 492
|
| 74 |
-
}
|
| 75 |
-
],
|
| 76 |
-
"logging_steps": 100,
|
| 77 |
-
"max_steps": 611,
|
| 78 |
-
"num_input_tokens_seen": 0,
|
| 79 |
-
"num_train_epochs": 1,
|
| 80 |
-
"save_steps": 123,
|
| 81 |
-
"stateful_callbacks": {
|
| 82 |
-
"TrainerControl": {
|
| 83 |
-
"args": {
|
| 84 |
-
"should_epoch_stop": false,
|
| 85 |
-
"should_evaluate": false,
|
| 86 |
-
"should_log": false,
|
| 87 |
-
"should_save": true,
|
| 88 |
-
"should_training_stop": false
|
| 89 |
-
},
|
| 90 |
-
"attributes": {}
|
| 91 |
-
}
|
| 92 |
-
},
|
| 93 |
-
"total_flos": 0.0,
|
| 94 |
-
"train_batch_size": 4,
|
| 95 |
-
"trial_name": null,
|
| 96 |
-
"trial_params": null
|
| 97 |
-
}
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
pytorch_model.bin
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 516640282
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c239fef244a712098b8e4629d9d994c594ef095b2f967efec482942b3f225373
|
| 3 |
size 516640282
|