Training in progress, step 1673, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2503003904
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:696b61b107c7830e174c74fc2c1e98a8ac2eb60432a2780ac20743074c60bfef
|
| 3 |
size 2503003904
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 5006244836
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:183ba56959c3f98243460a5cf43908fb35573ed3c49e7e641501af88f84532ad
|
| 3 |
size 5006244836
|
last-checkpoint/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:a9aa8a834991d1a099287d763aaf65662c5909034fe39dff582c0258f97c0051
|
| 3 |
size 14244
|
last-checkpoint/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:e526b4ee743444ec6815869e1af216b1753a6adb990a1534692135db761d6817
|
| 3 |
size 1064
|
last-checkpoint/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,
|
|
@@ -10045,6 +10045,1679 @@
|
|
| 10045 |
"learning_rate": 1.8280335791817733e-05,
|
| 10046 |
"loss": 0.875,
|
| 10047 |
"step": 1434
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 10048 |
}
|
| 10049 |
],
|
| 10050 |
"logging_steps": 1,
|
|
@@ -10064,7 +11737,7 @@
|
|
| 10064 |
"attributes": {}
|
| 10065 |
}
|
| 10066 |
},
|
| 10067 |
-
"total_flos":
|
| 10068 |
"train_batch_size": 8,
|
| 10069 |
"trial_name": null,
|
| 10070 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.8553169734151329,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1673,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 10045 |
"learning_rate": 1.8280335791817733e-05,
|
| 10046 |
"loss": 0.875,
|
| 10047 |
"step": 1434
|
| 10048 |
+
},
|
| 10049 |
+
{
|
| 10050 |
+
"epoch": 0.733640081799591,
|
| 10051 |
+
"grad_norm": 3.518186569213867,
|
| 10052 |
+
"learning_rate": 1.821495874296003e-05,
|
| 10053 |
+
"loss": 0.7896,
|
| 10054 |
+
"step": 1435
|
| 10055 |
+
},
|
| 10056 |
+
{
|
| 10057 |
+
"epoch": 0.7341513292433538,
|
| 10058 |
+
"grad_norm": 4.46856164932251,
|
| 10059 |
+
"learning_rate": 1.8149672762244624e-05,
|
| 10060 |
+
"loss": 1.0133,
|
| 10061 |
+
"step": 1436
|
| 10062 |
+
},
|
| 10063 |
+
{
|
| 10064 |
+
"epoch": 0.7346625766871165,
|
| 10065 |
+
"grad_norm": 3.835460662841797,
|
| 10066 |
+
"learning_rate": 1.808447803672404e-05,
|
| 10067 |
+
"loss": 0.837,
|
| 10068 |
+
"step": 1437
|
| 10069 |
+
},
|
| 10070 |
+
{
|
| 10071 |
+
"epoch": 0.7351738241308794,
|
| 10072 |
+
"grad_norm": 4.0243916511535645,
|
| 10073 |
+
"learning_rate": 1.801937475318939e-05,
|
| 10074 |
+
"loss": 0.8681,
|
| 10075 |
+
"step": 1438
|
| 10076 |
+
},
|
| 10077 |
+
{
|
| 10078 |
+
"epoch": 0.7356850715746421,
|
| 10079 |
+
"grad_norm": 4.4127960205078125,
|
| 10080 |
+
"learning_rate": 1.7954363098169768e-05,
|
| 10081 |
+
"loss": 0.9832,
|
| 10082 |
+
"step": 1439
|
| 10083 |
+
},
|
| 10084 |
+
{
|
| 10085 |
+
"epoch": 0.7361963190184049,
|
| 10086 |
+
"grad_norm": 4.0299248695373535,
|
| 10087 |
+
"learning_rate": 1.7889443257931737e-05,
|
| 10088 |
+
"loss": 0.8392,
|
| 10089 |
+
"step": 1440
|
| 10090 |
+
},
|
| 10091 |
+
{
|
| 10092 |
+
"epoch": 0.7367075664621677,
|
| 10093 |
+
"grad_norm": 4.006964206695557,
|
| 10094 |
+
"learning_rate": 1.782461541847879e-05,
|
| 10095 |
+
"loss": 0.7381,
|
| 10096 |
+
"step": 1441
|
| 10097 |
+
},
|
| 10098 |
+
{
|
| 10099 |
+
"epoch": 0.7372188139059305,
|
| 10100 |
+
"grad_norm": 4.23606014251709,
|
| 10101 |
+
"learning_rate": 1.7759879765550887e-05,
|
| 10102 |
+
"loss": 0.8984,
|
| 10103 |
+
"step": 1442
|
| 10104 |
+
},
|
| 10105 |
+
{
|
| 10106 |
+
"epoch": 0.7377300613496932,
|
| 10107 |
+
"grad_norm": 4.232422351837158,
|
| 10108 |
+
"learning_rate": 1.769523648462379e-05,
|
| 10109 |
+
"loss": 0.8403,
|
| 10110 |
+
"step": 1443
|
| 10111 |
+
},
|
| 10112 |
+
{
|
| 10113 |
+
"epoch": 0.7382413087934561,
|
| 10114 |
+
"grad_norm": 4.366641044616699,
|
| 10115 |
+
"learning_rate": 1.7630685760908622e-05,
|
| 10116 |
+
"loss": 0.8691,
|
| 10117 |
+
"step": 1444
|
| 10118 |
+
},
|
| 10119 |
+
{
|
| 10120 |
+
"epoch": 0.7387525562372188,
|
| 10121 |
+
"grad_norm": 4.925327777862549,
|
| 10122 |
+
"learning_rate": 1.7566227779351357e-05,
|
| 10123 |
+
"loss": 0.9101,
|
| 10124 |
+
"step": 1445
|
| 10125 |
+
},
|
| 10126 |
+
{
|
| 10127 |
+
"epoch": 0.7392638036809815,
|
| 10128 |
+
"grad_norm": 5.099564552307129,
|
| 10129 |
+
"learning_rate": 1.750186272463219e-05,
|
| 10130 |
+
"loss": 0.7557,
|
| 10131 |
+
"step": 1446
|
| 10132 |
+
},
|
| 10133 |
+
{
|
| 10134 |
+
"epoch": 0.7397750511247444,
|
| 10135 |
+
"grad_norm": 5.312896728515625,
|
| 10136 |
+
"learning_rate": 1.7437590781165138e-05,
|
| 10137 |
+
"loss": 0.6865,
|
| 10138 |
+
"step": 1447
|
| 10139 |
+
},
|
| 10140 |
+
{
|
| 10141 |
+
"epoch": 0.7402862985685071,
|
| 10142 |
+
"grad_norm": 5.26222038269043,
|
| 10143 |
+
"learning_rate": 1.7373412133097372e-05,
|
| 10144 |
+
"loss": 0.7639,
|
| 10145 |
+
"step": 1448
|
| 10146 |
+
},
|
| 10147 |
+
{
|
| 10148 |
+
"epoch": 0.74079754601227,
|
| 10149 |
+
"grad_norm": 4.816617012023926,
|
| 10150 |
+
"learning_rate": 1.7309326964308838e-05,
|
| 10151 |
+
"loss": 0.4676,
|
| 10152 |
+
"step": 1449
|
| 10153 |
+
},
|
| 10154 |
+
{
|
| 10155 |
+
"epoch": 0.7413087934560327,
|
| 10156 |
+
"grad_norm": 6.375493049621582,
|
| 10157 |
+
"learning_rate": 1.7245335458411542e-05,
|
| 10158 |
+
"loss": 0.4741,
|
| 10159 |
+
"step": 1450
|
| 10160 |
+
},
|
| 10161 |
+
{
|
| 10162 |
+
"epoch": 0.7418200408997955,
|
| 10163 |
+
"grad_norm": 1.9384807348251343,
|
| 10164 |
+
"learning_rate": 1.7181437798749256e-05,
|
| 10165 |
+
"loss": 0.8033,
|
| 10166 |
+
"step": 1451
|
| 10167 |
+
},
|
| 10168 |
+
{
|
| 10169 |
+
"epoch": 0.7423312883435583,
|
| 10170 |
+
"grad_norm": 2.3475661277770996,
|
| 10171 |
+
"learning_rate": 1.7117634168396774e-05,
|
| 10172 |
+
"loss": 0.9676,
|
| 10173 |
+
"step": 1452
|
| 10174 |
+
},
|
| 10175 |
+
{
|
| 10176 |
+
"epoch": 0.7428425357873211,
|
| 10177 |
+
"grad_norm": 2.614190101623535,
|
| 10178 |
+
"learning_rate": 1.705392475015956e-05,
|
| 10179 |
+
"loss": 1.0618,
|
| 10180 |
+
"step": 1453
|
| 10181 |
+
},
|
| 10182 |
+
{
|
| 10183 |
+
"epoch": 0.7433537832310838,
|
| 10184 |
+
"grad_norm": 2.739084243774414,
|
| 10185 |
+
"learning_rate": 1.6990309726573095e-05,
|
| 10186 |
+
"loss": 0.9619,
|
| 10187 |
+
"step": 1454
|
| 10188 |
+
},
|
| 10189 |
+
{
|
| 10190 |
+
"epoch": 0.7438650306748467,
|
| 10191 |
+
"grad_norm": 2.832260847091675,
|
| 10192 |
+
"learning_rate": 1.6926789279902412e-05,
|
| 10193 |
+
"loss": 0.9322,
|
| 10194 |
+
"step": 1455
|
| 10195 |
+
},
|
| 10196 |
+
{
|
| 10197 |
+
"epoch": 0.7443762781186094,
|
| 10198 |
+
"grad_norm": 2.8056933879852295,
|
| 10199 |
+
"learning_rate": 1.6863363592141618e-05,
|
| 10200 |
+
"loss": 0.9499,
|
| 10201 |
+
"step": 1456
|
| 10202 |
+
},
|
| 10203 |
+
{
|
| 10204 |
+
"epoch": 0.7448875255623721,
|
| 10205 |
+
"grad_norm": 2.8965537548065186,
|
| 10206 |
+
"learning_rate": 1.6800032845013247e-05,
|
| 10207 |
+
"loss": 0.907,
|
| 10208 |
+
"step": 1457
|
| 10209 |
+
},
|
| 10210 |
+
{
|
| 10211 |
+
"epoch": 0.745398773006135,
|
| 10212 |
+
"grad_norm": 3.008425235748291,
|
| 10213 |
+
"learning_rate": 1.673679721996789e-05,
|
| 10214 |
+
"loss": 0.9932,
|
| 10215 |
+
"step": 1458
|
| 10216 |
+
},
|
| 10217 |
+
{
|
| 10218 |
+
"epoch": 0.7459100204498977,
|
| 10219 |
+
"grad_norm": 2.81937837600708,
|
| 10220 |
+
"learning_rate": 1.6673656898183572e-05,
|
| 10221 |
+
"loss": 0.9586,
|
| 10222 |
+
"step": 1459
|
| 10223 |
+
},
|
| 10224 |
+
{
|
| 10225 |
+
"epoch": 0.7464212678936605,
|
| 10226 |
+
"grad_norm": 2.9769203662872314,
|
| 10227 |
+
"learning_rate": 1.6610612060565234e-05,
|
| 10228 |
+
"loss": 0.9148,
|
| 10229 |
+
"step": 1460
|
| 10230 |
+
},
|
| 10231 |
+
{
|
| 10232 |
+
"epoch": 0.7469325153374233,
|
| 10233 |
+
"grad_norm": 3.3971211910247803,
|
| 10234 |
+
"learning_rate": 1.6547662887744265e-05,
|
| 10235 |
+
"loss": 0.906,
|
| 10236 |
+
"step": 1461
|
| 10237 |
+
},
|
| 10238 |
+
{
|
| 10239 |
+
"epoch": 0.7474437627811861,
|
| 10240 |
+
"grad_norm": 2.9512698650360107,
|
| 10241 |
+
"learning_rate": 1.648480956007799e-05,
|
| 10242 |
+
"loss": 0.9082,
|
| 10243 |
+
"step": 1462
|
| 10244 |
+
},
|
| 10245 |
+
{
|
| 10246 |
+
"epoch": 0.7479550102249489,
|
| 10247 |
+
"grad_norm": 3.2894976139068604,
|
| 10248 |
+
"learning_rate": 1.6422052257649078e-05,
|
| 10249 |
+
"loss": 0.9537,
|
| 10250 |
+
"step": 1463
|
| 10251 |
+
},
|
| 10252 |
+
{
|
| 10253 |
+
"epoch": 0.7484662576687117,
|
| 10254 |
+
"grad_norm": 2.9925243854522705,
|
| 10255 |
+
"learning_rate": 1.6359391160265125e-05,
|
| 10256 |
+
"loss": 0.9018,
|
| 10257 |
+
"step": 1464
|
| 10258 |
+
},
|
| 10259 |
+
{
|
| 10260 |
+
"epoch": 0.7489775051124744,
|
| 10261 |
+
"grad_norm": 3.4137401580810547,
|
| 10262 |
+
"learning_rate": 1.629682644745802e-05,
|
| 10263 |
+
"loss": 0.9413,
|
| 10264 |
+
"step": 1465
|
| 10265 |
+
},
|
| 10266 |
+
{
|
| 10267 |
+
"epoch": 0.7494887525562373,
|
| 10268 |
+
"grad_norm": 3.152463436126709,
|
| 10269 |
+
"learning_rate": 1.6234358298483575e-05,
|
| 10270 |
+
"loss": 0.9389,
|
| 10271 |
+
"step": 1466
|
| 10272 |
+
},
|
| 10273 |
+
{
|
| 10274 |
+
"epoch": 0.75,
|
| 10275 |
+
"grad_norm": 3.088907480239868,
|
| 10276 |
+
"learning_rate": 1.6171986892320884e-05,
|
| 10277 |
+
"loss": 0.8818,
|
| 10278 |
+
"step": 1467
|
| 10279 |
+
},
|
| 10280 |
+
{
|
| 10281 |
+
"epoch": 0.7505112474437627,
|
| 10282 |
+
"grad_norm": 2.9647679328918457,
|
| 10283 |
+
"learning_rate": 1.6109712407671867e-05,
|
| 10284 |
+
"loss": 0.8208,
|
| 10285 |
+
"step": 1468
|
| 10286 |
+
},
|
| 10287 |
+
{
|
| 10288 |
+
"epoch": 0.7510224948875256,
|
| 10289 |
+
"grad_norm": 3.121256113052368,
|
| 10290 |
+
"learning_rate": 1.6047535022960757e-05,
|
| 10291 |
+
"loss": 0.8292,
|
| 10292 |
+
"step": 1469
|
| 10293 |
+
},
|
| 10294 |
+
{
|
| 10295 |
+
"epoch": 0.7515337423312883,
|
| 10296 |
+
"grad_norm": 3.0089125633239746,
|
| 10297 |
+
"learning_rate": 1.5985454916333577e-05,
|
| 10298 |
+
"loss": 0.8211,
|
| 10299 |
+
"step": 1470
|
| 10300 |
+
},
|
| 10301 |
+
{
|
| 10302 |
+
"epoch": 0.7520449897750511,
|
| 10303 |
+
"grad_norm": 3.4958393573760986,
|
| 10304 |
+
"learning_rate": 1.592347226565766e-05,
|
| 10305 |
+
"loss": 0.9042,
|
| 10306 |
+
"step": 1471
|
| 10307 |
+
},
|
| 10308 |
+
{
|
| 10309 |
+
"epoch": 0.7525562372188139,
|
| 10310 |
+
"grad_norm": 3.168567419052124,
|
| 10311 |
+
"learning_rate": 1.586158724852108e-05,
|
| 10312 |
+
"loss": 0.783,
|
| 10313 |
+
"step": 1472
|
| 10314 |
+
},
|
| 10315 |
+
{
|
| 10316 |
+
"epoch": 0.7530674846625767,
|
| 10317 |
+
"grad_norm": 3.506746530532837,
|
| 10318 |
+
"learning_rate": 1.579980004223222e-05,
|
| 10319 |
+
"loss": 0.8124,
|
| 10320 |
+
"step": 1473
|
| 10321 |
+
},
|
| 10322 |
+
{
|
| 10323 |
+
"epoch": 0.7535787321063395,
|
| 10324 |
+
"grad_norm": 3.678039789199829,
|
| 10325 |
+
"learning_rate": 1.573811082381918e-05,
|
| 10326 |
+
"loss": 0.9267,
|
| 10327 |
+
"step": 1474
|
| 10328 |
+
},
|
| 10329 |
+
{
|
| 10330 |
+
"epoch": 0.7540899795501023,
|
| 10331 |
+
"grad_norm": 3.4833285808563232,
|
| 10332 |
+
"learning_rate": 1.567651977002935e-05,
|
| 10333 |
+
"loss": 0.8201,
|
| 10334 |
+
"step": 1475
|
| 10335 |
+
},
|
| 10336 |
+
{
|
| 10337 |
+
"epoch": 0.754601226993865,
|
| 10338 |
+
"grad_norm": 3.767958879470825,
|
| 10339 |
+
"learning_rate": 1.561502705732883e-05,
|
| 10340 |
+
"loss": 0.9726,
|
| 10341 |
+
"step": 1476
|
| 10342 |
+
},
|
| 10343 |
+
{
|
| 10344 |
+
"epoch": 0.7551124744376279,
|
| 10345 |
+
"grad_norm": 3.3126537799835205,
|
| 10346 |
+
"learning_rate": 1.5553632861901995e-05,
|
| 10347 |
+
"loss": 0.7471,
|
| 10348 |
+
"step": 1477
|
| 10349 |
+
},
|
| 10350 |
+
{
|
| 10351 |
+
"epoch": 0.7556237218813906,
|
| 10352 |
+
"grad_norm": 3.590383529663086,
|
| 10353 |
+
"learning_rate": 1.5492337359650937e-05,
|
| 10354 |
+
"loss": 0.8543,
|
| 10355 |
+
"step": 1478
|
| 10356 |
+
},
|
| 10357 |
+
{
|
| 10358 |
+
"epoch": 0.7561349693251533,
|
| 10359 |
+
"grad_norm": 3.4559550285339355,
|
| 10360 |
+
"learning_rate": 1.5431140726194974e-05,
|
| 10361 |
+
"loss": 0.8802,
|
| 10362 |
+
"step": 1479
|
| 10363 |
+
},
|
| 10364 |
+
{
|
| 10365 |
+
"epoch": 0.7566462167689162,
|
| 10366 |
+
"grad_norm": 3.7260870933532715,
|
| 10367 |
+
"learning_rate": 1.5370043136870148e-05,
|
| 10368 |
+
"loss": 0.9178,
|
| 10369 |
+
"step": 1480
|
| 10370 |
+
},
|
| 10371 |
+
{
|
| 10372 |
+
"epoch": 0.7571574642126789,
|
| 10373 |
+
"grad_norm": 4.135682106018066,
|
| 10374 |
+
"learning_rate": 1.5309044766728775e-05,
|
| 10375 |
+
"loss": 0.8823,
|
| 10376 |
+
"step": 1481
|
| 10377 |
+
},
|
| 10378 |
+
{
|
| 10379 |
+
"epoch": 0.7576687116564417,
|
| 10380 |
+
"grad_norm": 3.654975175857544,
|
| 10381 |
+
"learning_rate": 1.5248145790538837e-05,
|
| 10382 |
+
"loss": 0.8367,
|
| 10383 |
+
"step": 1482
|
| 10384 |
+
},
|
| 10385 |
+
{
|
| 10386 |
+
"epoch": 0.7581799591002045,
|
| 10387 |
+
"grad_norm": 3.653458833694458,
|
| 10388 |
+
"learning_rate": 1.5187346382783552e-05,
|
| 10389 |
+
"loss": 0.8005,
|
| 10390 |
+
"step": 1483
|
| 10391 |
+
},
|
| 10392 |
+
{
|
| 10393 |
+
"epoch": 0.7586912065439673,
|
| 10394 |
+
"grad_norm": 3.8800647258758545,
|
| 10395 |
+
"learning_rate": 1.5126646717660897e-05,
|
| 10396 |
+
"loss": 0.9004,
|
| 10397 |
+
"step": 1484
|
| 10398 |
+
},
|
| 10399 |
+
{
|
| 10400 |
+
"epoch": 0.75920245398773,
|
| 10401 |
+
"grad_norm": 3.79221773147583,
|
| 10402 |
+
"learning_rate": 1.5066046969083026e-05,
|
| 10403 |
+
"loss": 0.7986,
|
| 10404 |
+
"step": 1485
|
| 10405 |
+
},
|
| 10406 |
+
{
|
| 10407 |
+
"epoch": 0.7597137014314929,
|
| 10408 |
+
"grad_norm": 4.0721025466918945,
|
| 10409 |
+
"learning_rate": 1.5005547310675872e-05,
|
| 10410 |
+
"loss": 0.9721,
|
| 10411 |
+
"step": 1486
|
| 10412 |
+
},
|
| 10413 |
+
{
|
| 10414 |
+
"epoch": 0.7602249488752556,
|
| 10415 |
+
"grad_norm": 3.911992311477661,
|
| 10416 |
+
"learning_rate": 1.4945147915778535e-05,
|
| 10417 |
+
"loss": 0.8429,
|
| 10418 |
+
"step": 1487
|
| 10419 |
+
},
|
| 10420 |
+
{
|
| 10421 |
+
"epoch": 0.7607361963190185,
|
| 10422 |
+
"grad_norm": 4.218154430389404,
|
| 10423 |
+
"learning_rate": 1.4884848957442931e-05,
|
| 10424 |
+
"loss": 0.7957,
|
| 10425 |
+
"step": 1488
|
| 10426 |
+
},
|
| 10427 |
+
{
|
| 10428 |
+
"epoch": 0.7612474437627812,
|
| 10429 |
+
"grad_norm": 3.988151788711548,
|
| 10430 |
+
"learning_rate": 1.4824650608433099e-05,
|
| 10431 |
+
"loss": 0.8568,
|
| 10432 |
+
"step": 1489
|
| 10433 |
+
},
|
| 10434 |
+
{
|
| 10435 |
+
"epoch": 0.7617586912065439,
|
| 10436 |
+
"grad_norm": 4.378921985626221,
|
| 10437 |
+
"learning_rate": 1.4764553041224926e-05,
|
| 10438 |
+
"loss": 0.7963,
|
| 10439 |
+
"step": 1490
|
| 10440 |
+
},
|
| 10441 |
+
{
|
| 10442 |
+
"epoch": 0.7622699386503068,
|
| 10443 |
+
"grad_norm": 4.452109336853027,
|
| 10444 |
+
"learning_rate": 1.4704556428005478e-05,
|
| 10445 |
+
"loss": 0.835,
|
| 10446 |
+
"step": 1491
|
| 10447 |
+
},
|
| 10448 |
+
{
|
| 10449 |
+
"epoch": 0.7627811860940695,
|
| 10450 |
+
"grad_norm": 6.035761833190918,
|
| 10451 |
+
"learning_rate": 1.4644660940672627e-05,
|
| 10452 |
+
"loss": 0.9188,
|
| 10453 |
+
"step": 1492
|
| 10454 |
+
},
|
| 10455 |
+
{
|
| 10456 |
+
"epoch": 0.7632924335378323,
|
| 10457 |
+
"grad_norm": 4.4452996253967285,
|
| 10458 |
+
"learning_rate": 1.4584866750834464e-05,
|
| 10459 |
+
"loss": 0.7396,
|
| 10460 |
+
"step": 1493
|
| 10461 |
+
},
|
| 10462 |
+
{
|
| 10463 |
+
"epoch": 0.7638036809815951,
|
| 10464 |
+
"grad_norm": 4.476807594299316,
|
| 10465 |
+
"learning_rate": 1.4525174029808858e-05,
|
| 10466 |
+
"loss": 0.7628,
|
| 10467 |
+
"step": 1494
|
| 10468 |
+
},
|
| 10469 |
+
{
|
| 10470 |
+
"epoch": 0.7643149284253579,
|
| 10471 |
+
"grad_norm": 4.541752338409424,
|
| 10472 |
+
"learning_rate": 1.4465582948622986e-05,
|
| 10473 |
+
"loss": 0.6893,
|
| 10474 |
+
"step": 1495
|
| 10475 |
+
},
|
| 10476 |
+
{
|
| 10477 |
+
"epoch": 0.7648261758691206,
|
| 10478 |
+
"grad_norm": 4.8742899894714355,
|
| 10479 |
+
"learning_rate": 1.4406093678012766e-05,
|
| 10480 |
+
"loss": 0.7884,
|
| 10481 |
+
"step": 1496
|
| 10482 |
+
},
|
| 10483 |
+
{
|
| 10484 |
+
"epoch": 0.7653374233128835,
|
| 10485 |
+
"grad_norm": 5.151451110839844,
|
| 10486 |
+
"learning_rate": 1.4346706388422493e-05,
|
| 10487 |
+
"loss": 0.683,
|
| 10488 |
+
"step": 1497
|
| 10489 |
+
},
|
| 10490 |
+
{
|
| 10491 |
+
"epoch": 0.7658486707566462,
|
| 10492 |
+
"grad_norm": 5.411539554595947,
|
| 10493 |
+
"learning_rate": 1.4287421250004157e-05,
|
| 10494 |
+
"loss": 0.79,
|
| 10495 |
+
"step": 1498
|
| 10496 |
+
},
|
| 10497 |
+
{
|
| 10498 |
+
"epoch": 0.766359918200409,
|
| 10499 |
+
"grad_norm": 5.503897666931152,
|
| 10500 |
+
"learning_rate": 1.4228238432617186e-05,
|
| 10501 |
+
"loss": 0.556,
|
| 10502 |
+
"step": 1499
|
| 10503 |
+
},
|
| 10504 |
+
{
|
| 10505 |
+
"epoch": 0.7668711656441718,
|
| 10506 |
+
"grad_norm": 6.515267372131348,
|
| 10507 |
+
"learning_rate": 1.4169158105827768e-05,
|
| 10508 |
+
"loss": 0.393,
|
| 10509 |
+
"step": 1500
|
| 10510 |
+
},
|
| 10511 |
+
{
|
| 10512 |
+
"epoch": 0.7673824130879345,
|
| 10513 |
+
"grad_norm": 2.1720798015594482,
|
| 10514 |
+
"learning_rate": 1.4110180438908505e-05,
|
| 10515 |
+
"loss": 0.7376,
|
| 10516 |
+
"step": 1501
|
| 10517 |
+
},
|
| 10518 |
+
{
|
| 10519 |
+
"epoch": 0.7678936605316974,
|
| 10520 |
+
"grad_norm": 2.3930561542510986,
|
| 10521 |
+
"learning_rate": 1.4051305600837799e-05,
|
| 10522 |
+
"loss": 1.0402,
|
| 10523 |
+
"step": 1502
|
| 10524 |
+
},
|
| 10525 |
+
{
|
| 10526 |
+
"epoch": 0.7684049079754601,
|
| 10527 |
+
"grad_norm": 2.3610873222351074,
|
| 10528 |
+
"learning_rate": 1.3992533760299498e-05,
|
| 10529 |
+
"loss": 0.9187,
|
| 10530 |
+
"step": 1503
|
| 10531 |
+
},
|
| 10532 |
+
{
|
| 10533 |
+
"epoch": 0.7689161554192229,
|
| 10534 |
+
"grad_norm": 2.4556972980499268,
|
| 10535 |
+
"learning_rate": 1.3933865085682312e-05,
|
| 10536 |
+
"loss": 0.9221,
|
| 10537 |
+
"step": 1504
|
| 10538 |
+
},
|
| 10539 |
+
{
|
| 10540 |
+
"epoch": 0.7694274028629857,
|
| 10541 |
+
"grad_norm": 2.7687621116638184,
|
| 10542 |
+
"learning_rate": 1.387529974507935e-05,
|
| 10543 |
+
"loss": 1.0278,
|
| 10544 |
+
"step": 1505
|
| 10545 |
+
},
|
| 10546 |
+
{
|
| 10547 |
+
"epoch": 0.7699386503067485,
|
| 10548 |
+
"grad_norm": 2.8164188861846924,
|
| 10549 |
+
"learning_rate": 1.3816837906287722e-05,
|
| 10550 |
+
"loss": 0.9995,
|
| 10551 |
+
"step": 1506
|
| 10552 |
+
},
|
| 10553 |
+
{
|
| 10554 |
+
"epoch": 0.7704498977505112,
|
| 10555 |
+
"grad_norm": 2.5187416076660156,
|
| 10556 |
+
"learning_rate": 1.3758479736807928e-05,
|
| 10557 |
+
"loss": 0.813,
|
| 10558 |
+
"step": 1507
|
| 10559 |
+
},
|
| 10560 |
+
{
|
| 10561 |
+
"epoch": 0.7709611451942741,
|
| 10562 |
+
"grad_norm": 2.9586594104766846,
|
| 10563 |
+
"learning_rate": 1.3700225403843469e-05,
|
| 10564 |
+
"loss": 0.9001,
|
| 10565 |
+
"step": 1508
|
| 10566 |
+
},
|
| 10567 |
+
{
|
| 10568 |
+
"epoch": 0.7714723926380368,
|
| 10569 |
+
"grad_norm": 3.797866106033325,
|
| 10570 |
+
"learning_rate": 1.3642075074300325e-05,
|
| 10571 |
+
"loss": 0.9324,
|
| 10572 |
+
"step": 1509
|
| 10573 |
+
},
|
| 10574 |
+
{
|
| 10575 |
+
"epoch": 0.7719836400817995,
|
| 10576 |
+
"grad_norm": 2.9937219619750977,
|
| 10577 |
+
"learning_rate": 1.3584028914786539e-05,
|
| 10578 |
+
"loss": 0.922,
|
| 10579 |
+
"step": 1510
|
| 10580 |
+
},
|
| 10581 |
+
{
|
| 10582 |
+
"epoch": 0.7724948875255624,
|
| 10583 |
+
"grad_norm": 2.9317002296447754,
|
| 10584 |
+
"learning_rate": 1.3526087091611623e-05,
|
| 10585 |
+
"loss": 0.9387,
|
| 10586 |
+
"step": 1511
|
| 10587 |
+
},
|
| 10588 |
+
{
|
| 10589 |
+
"epoch": 0.7730061349693251,
|
| 10590 |
+
"grad_norm": 3.1819257736206055,
|
| 10591 |
+
"learning_rate": 1.3468249770786223e-05,
|
| 10592 |
+
"loss": 0.8964,
|
| 10593 |
+
"step": 1512
|
| 10594 |
+
},
|
| 10595 |
+
{
|
| 10596 |
+
"epoch": 0.773517382413088,
|
| 10597 |
+
"grad_norm": 2.7915236949920654,
|
| 10598 |
+
"learning_rate": 1.341051711802151e-05,
|
| 10599 |
+
"loss": 0.9163,
|
| 10600 |
+
"step": 1513
|
| 10601 |
+
},
|
| 10602 |
+
{
|
| 10603 |
+
"epoch": 0.7740286298568507,
|
| 10604 |
+
"grad_norm": 2.946099281311035,
|
| 10605 |
+
"learning_rate": 1.3352889298728832e-05,
|
| 10606 |
+
"loss": 0.9546,
|
| 10607 |
+
"step": 1514
|
| 10608 |
+
},
|
| 10609 |
+
{
|
| 10610 |
+
"epoch": 0.7745398773006135,
|
| 10611 |
+
"grad_norm": 3.0270745754241943,
|
| 10612 |
+
"learning_rate": 1.3295366478019112e-05,
|
| 10613 |
+
"loss": 0.9431,
|
| 10614 |
+
"step": 1515
|
| 10615 |
+
},
|
| 10616 |
+
{
|
| 10617 |
+
"epoch": 0.7750511247443763,
|
| 10618 |
+
"grad_norm": 3.3205618858337402,
|
| 10619 |
+
"learning_rate": 1.3237948820702495e-05,
|
| 10620 |
+
"loss": 0.9058,
|
| 10621 |
+
"step": 1516
|
| 10622 |
+
},
|
| 10623 |
+
{
|
| 10624 |
+
"epoch": 0.7755623721881391,
|
| 10625 |
+
"grad_norm": 3.179041624069214,
|
| 10626 |
+
"learning_rate": 1.3180636491287773e-05,
|
| 10627 |
+
"loss": 0.8804,
|
| 10628 |
+
"step": 1517
|
| 10629 |
+
},
|
| 10630 |
+
{
|
| 10631 |
+
"epoch": 0.7760736196319018,
|
| 10632 |
+
"grad_norm": 3.682973623275757,
|
| 10633 |
+
"learning_rate": 1.3123429653981995e-05,
|
| 10634 |
+
"loss": 0.985,
|
| 10635 |
+
"step": 1518
|
| 10636 |
+
},
|
| 10637 |
+
{
|
| 10638 |
+
"epoch": 0.7765848670756647,
|
| 10639 |
+
"grad_norm": 3.1981019973754883,
|
| 10640 |
+
"learning_rate": 1.3066328472689932e-05,
|
| 10641 |
+
"loss": 0.8148,
|
| 10642 |
+
"step": 1519
|
| 10643 |
+
},
|
| 10644 |
+
{
|
| 10645 |
+
"epoch": 0.7770961145194274,
|
| 10646 |
+
"grad_norm": 3.3223016262054443,
|
| 10647 |
+
"learning_rate": 1.300933311101365e-05,
|
| 10648 |
+
"loss": 0.8839,
|
| 10649 |
+
"step": 1520
|
| 10650 |
+
},
|
| 10651 |
+
{
|
| 10652 |
+
"epoch": 0.7776073619631901,
|
| 10653 |
+
"grad_norm": 3.214383602142334,
|
| 10654 |
+
"learning_rate": 1.2952443732252057e-05,
|
| 10655 |
+
"loss": 0.9364,
|
| 10656 |
+
"step": 1521
|
| 10657 |
+
},
|
| 10658 |
+
{
|
| 10659 |
+
"epoch": 0.778118609406953,
|
| 10660 |
+
"grad_norm": 3.4616458415985107,
|
| 10661 |
+
"learning_rate": 1.2895660499400348e-05,
|
| 10662 |
+
"loss": 0.8707,
|
| 10663 |
+
"step": 1522
|
| 10664 |
+
},
|
| 10665 |
+
{
|
| 10666 |
+
"epoch": 0.7786298568507157,
|
| 10667 |
+
"grad_norm": 3.4007458686828613,
|
| 10668 |
+
"learning_rate": 1.283898357514966e-05,
|
| 10669 |
+
"loss": 0.9236,
|
| 10670 |
+
"step": 1523
|
| 10671 |
+
},
|
| 10672 |
+
{
|
| 10673 |
+
"epoch": 0.7791411042944786,
|
| 10674 |
+
"grad_norm": 3.370945930480957,
|
| 10675 |
+
"learning_rate": 1.2782413121886483e-05,
|
| 10676 |
+
"loss": 0.8499,
|
| 10677 |
+
"step": 1524
|
| 10678 |
+
},
|
| 10679 |
+
{
|
| 10680 |
+
"epoch": 0.7796523517382413,
|
| 10681 |
+
"grad_norm": 3.3444669246673584,
|
| 10682 |
+
"learning_rate": 1.2725949301692314e-05,
|
| 10683 |
+
"loss": 0.8872,
|
| 10684 |
+
"step": 1525
|
| 10685 |
+
},
|
| 10686 |
+
{
|
| 10687 |
+
"epoch": 0.7801635991820041,
|
| 10688 |
+
"grad_norm": 3.51122784614563,
|
| 10689 |
+
"learning_rate": 1.2669592276343084e-05,
|
| 10690 |
+
"loss": 0.911,
|
| 10691 |
+
"step": 1526
|
| 10692 |
+
},
|
| 10693 |
+
{
|
| 10694 |
+
"epoch": 0.7806748466257669,
|
| 10695 |
+
"grad_norm": 3.6887714862823486,
|
| 10696 |
+
"learning_rate": 1.2613342207308764e-05,
|
| 10697 |
+
"loss": 0.9304,
|
| 10698 |
+
"step": 1527
|
| 10699 |
+
},
|
| 10700 |
+
{
|
| 10701 |
+
"epoch": 0.7811860940695297,
|
| 10702 |
+
"grad_norm": 3.856081485748291,
|
| 10703 |
+
"learning_rate": 1.2557199255752867e-05,
|
| 10704 |
+
"loss": 0.9293,
|
| 10705 |
+
"step": 1528
|
| 10706 |
+
},
|
| 10707 |
+
{
|
| 10708 |
+
"epoch": 0.7816973415132924,
|
| 10709 |
+
"grad_norm": 3.9948816299438477,
|
| 10710 |
+
"learning_rate": 1.2501163582532038e-05,
|
| 10711 |
+
"loss": 0.837,
|
| 10712 |
+
"step": 1529
|
| 10713 |
+
},
|
| 10714 |
+
{
|
| 10715 |
+
"epoch": 0.7822085889570553,
|
| 10716 |
+
"grad_norm": 3.6386806964874268,
|
| 10717 |
+
"learning_rate": 1.24452353481955e-05,
|
| 10718 |
+
"loss": 0.8118,
|
| 10719 |
+
"step": 1530
|
| 10720 |
+
},
|
| 10721 |
+
{
|
| 10722 |
+
"epoch": 0.782719836400818,
|
| 10723 |
+
"grad_norm": 3.716519355773926,
|
| 10724 |
+
"learning_rate": 1.2389414712984715e-05,
|
| 10725 |
+
"loss": 0.7942,
|
| 10726 |
+
"step": 1531
|
| 10727 |
+
},
|
| 10728 |
+
{
|
| 10729 |
+
"epoch": 0.7832310838445807,
|
| 10730 |
+
"grad_norm": 3.621734142303467,
|
| 10731 |
+
"learning_rate": 1.2333701836832812e-05,
|
| 10732 |
+
"loss": 0.7823,
|
| 10733 |
+
"step": 1532
|
| 10734 |
+
},
|
| 10735 |
+
{
|
| 10736 |
+
"epoch": 0.7837423312883436,
|
| 10737 |
+
"grad_norm": 3.675961494445801,
|
| 10738 |
+
"learning_rate": 1.227809687936417e-05,
|
| 10739 |
+
"loss": 0.8813,
|
| 10740 |
+
"step": 1533
|
| 10741 |
+
},
|
| 10742 |
+
{
|
| 10743 |
+
"epoch": 0.7842535787321063,
|
| 10744 |
+
"grad_norm": 4.009909629821777,
|
| 10745 |
+
"learning_rate": 1.2222599999894018e-05,
|
| 10746 |
+
"loss": 0.9671,
|
| 10747 |
+
"step": 1534
|
| 10748 |
+
},
|
| 10749 |
+
{
|
| 10750 |
+
"epoch": 0.7847648261758691,
|
| 10751 |
+
"grad_norm": 3.6856648921966553,
|
| 10752 |
+
"learning_rate": 1.2167211357427878e-05,
|
| 10753 |
+
"loss": 0.7759,
|
| 10754 |
+
"step": 1535
|
| 10755 |
+
},
|
| 10756 |
+
{
|
| 10757 |
+
"epoch": 0.7852760736196319,
|
| 10758 |
+
"grad_norm": 4.252887725830078,
|
| 10759 |
+
"learning_rate": 1.2111931110661212e-05,
|
| 10760 |
+
"loss": 0.8245,
|
| 10761 |
+
"step": 1536
|
| 10762 |
+
},
|
| 10763 |
+
{
|
| 10764 |
+
"epoch": 0.7857873210633947,
|
| 10765 |
+
"grad_norm": 4.382503986358643,
|
| 10766 |
+
"learning_rate": 1.2056759417978835e-05,
|
| 10767 |
+
"loss": 1.0605,
|
| 10768 |
+
"step": 1537
|
| 10769 |
+
},
|
| 10770 |
+
{
|
| 10771 |
+
"epoch": 0.7862985685071575,
|
| 10772 |
+
"grad_norm": 4.000999927520752,
|
| 10773 |
+
"learning_rate": 1.2001696437454624e-05,
|
| 10774 |
+
"loss": 0.8371,
|
| 10775 |
+
"step": 1538
|
| 10776 |
+
},
|
| 10777 |
+
{
|
| 10778 |
+
"epoch": 0.7868098159509203,
|
| 10779 |
+
"grad_norm": 4.442836761474609,
|
| 10780 |
+
"learning_rate": 1.1946742326850912e-05,
|
| 10781 |
+
"loss": 0.8388,
|
| 10782 |
+
"step": 1539
|
| 10783 |
+
},
|
| 10784 |
+
{
|
| 10785 |
+
"epoch": 0.787321063394683,
|
| 10786 |
+
"grad_norm": 4.2866530418396,
|
| 10787 |
+
"learning_rate": 1.1891897243618182e-05,
|
| 10788 |
+
"loss": 0.7989,
|
| 10789 |
+
"step": 1540
|
| 10790 |
+
},
|
| 10791 |
+
{
|
| 10792 |
+
"epoch": 0.7878323108384458,
|
| 10793 |
+
"grad_norm": 4.245499610900879,
|
| 10794 |
+
"learning_rate": 1.183716134489446e-05,
|
| 10795 |
+
"loss": 0.9314,
|
| 10796 |
+
"step": 1541
|
| 10797 |
+
},
|
| 10798 |
+
{
|
| 10799 |
+
"epoch": 0.7883435582822086,
|
| 10800 |
+
"grad_norm": 4.35511589050293,
|
| 10801 |
+
"learning_rate": 1.1782534787505017e-05,
|
| 10802 |
+
"loss": 0.7689,
|
| 10803 |
+
"step": 1542
|
| 10804 |
+
},
|
| 10805 |
+
{
|
| 10806 |
+
"epoch": 0.7888548057259713,
|
| 10807 |
+
"grad_norm": 4.421148777008057,
|
| 10808 |
+
"learning_rate": 1.1728017727961794e-05,
|
| 10809 |
+
"loss": 0.7314,
|
| 10810 |
+
"step": 1543
|
| 10811 |
+
},
|
| 10812 |
+
{
|
| 10813 |
+
"epoch": 0.7893660531697342,
|
| 10814 |
+
"grad_norm": 4.419631481170654,
|
| 10815 |
+
"learning_rate": 1.1673610322463014e-05,
|
| 10816 |
+
"loss": 0.7021,
|
| 10817 |
+
"step": 1544
|
| 10818 |
+
},
|
| 10819 |
+
{
|
| 10820 |
+
"epoch": 0.7898773006134969,
|
| 10821 |
+
"grad_norm": 4.954163074493408,
|
| 10822 |
+
"learning_rate": 1.1619312726892762e-05,
|
| 10823 |
+
"loss": 0.7313,
|
| 10824 |
+
"step": 1545
|
| 10825 |
+
},
|
| 10826 |
+
{
|
| 10827 |
+
"epoch": 0.7903885480572597,
|
| 10828 |
+
"grad_norm": 4.796021461486816,
|
| 10829 |
+
"learning_rate": 1.1565125096820473e-05,
|
| 10830 |
+
"loss": 0.6311,
|
| 10831 |
+
"step": 1546
|
| 10832 |
+
},
|
| 10833 |
+
{
|
| 10834 |
+
"epoch": 0.7908997955010225,
|
| 10835 |
+
"grad_norm": 4.741822242736816,
|
| 10836 |
+
"learning_rate": 1.1511047587500523e-05,
|
| 10837 |
+
"loss": 0.5983,
|
| 10838 |
+
"step": 1547
|
| 10839 |
+
},
|
| 10840 |
+
{
|
| 10841 |
+
"epoch": 0.7914110429447853,
|
| 10842 |
+
"grad_norm": 6.696097373962402,
|
| 10843 |
+
"learning_rate": 1.1457080353871769e-05,
|
| 10844 |
+
"loss": 0.9485,
|
| 10845 |
+
"step": 1548
|
| 10846 |
+
},
|
| 10847 |
+
{
|
| 10848 |
+
"epoch": 0.7919222903885481,
|
| 10849 |
+
"grad_norm": 5.633409023284912,
|
| 10850 |
+
"learning_rate": 1.1403223550557146e-05,
|
| 10851 |
+
"loss": 0.6431,
|
| 10852 |
+
"step": 1549
|
| 10853 |
+
},
|
| 10854 |
+
{
|
| 10855 |
+
"epoch": 0.7924335378323109,
|
| 10856 |
+
"grad_norm": 7.090452194213867,
|
| 10857 |
+
"learning_rate": 1.134947733186315e-05,
|
| 10858 |
+
"loss": 0.4912,
|
| 10859 |
+
"step": 1550
|
| 10860 |
+
},
|
| 10861 |
+
{
|
| 10862 |
+
"epoch": 0.7929447852760736,
|
| 10863 |
+
"grad_norm": 1.965378761291504,
|
| 10864 |
+
"learning_rate": 1.1295841851779488e-05,
|
| 10865 |
+
"loss": 0.9309,
|
| 10866 |
+
"step": 1551
|
| 10867 |
+
},
|
| 10868 |
+
{
|
| 10869 |
+
"epoch": 0.7934560327198364,
|
| 10870 |
+
"grad_norm": 2.6269702911376953,
|
| 10871 |
+
"learning_rate": 1.1242317263978525e-05,
|
| 10872 |
+
"loss": 1.0566,
|
| 10873 |
+
"step": 1552
|
| 10874 |
+
},
|
| 10875 |
+
{
|
| 10876 |
+
"epoch": 0.7939672801635992,
|
| 10877 |
+
"grad_norm": 2.5522866249084473,
|
| 10878 |
+
"learning_rate": 1.118890372181497e-05,
|
| 10879 |
+
"loss": 0.9996,
|
| 10880 |
+
"step": 1553
|
| 10881 |
+
},
|
| 10882 |
+
{
|
| 10883 |
+
"epoch": 0.7944785276073619,
|
| 10884 |
+
"grad_norm": 2.6898603439331055,
|
| 10885 |
+
"learning_rate": 1.1135601378325316e-05,
|
| 10886 |
+
"loss": 0.8959,
|
| 10887 |
+
"step": 1554
|
| 10888 |
+
},
|
| 10889 |
+
{
|
| 10890 |
+
"epoch": 0.7949897750511248,
|
| 10891 |
+
"grad_norm": 2.5996921062469482,
|
| 10892 |
+
"learning_rate": 1.1082410386227527e-05,
|
| 10893 |
+
"loss": 0.9355,
|
| 10894 |
+
"step": 1555
|
| 10895 |
+
},
|
| 10896 |
+
{
|
| 10897 |
+
"epoch": 0.7955010224948875,
|
| 10898 |
+
"grad_norm": 2.5710082054138184,
|
| 10899 |
+
"learning_rate": 1.102933089792042e-05,
|
| 10900 |
+
"loss": 0.9206,
|
| 10901 |
+
"step": 1556
|
| 10902 |
+
},
|
| 10903 |
+
{
|
| 10904 |
+
"epoch": 0.7960122699386503,
|
| 10905 |
+
"grad_norm": 2.8860228061676025,
|
| 10906 |
+
"learning_rate": 1.0976363065483464e-05,
|
| 10907 |
+
"loss": 0.9861,
|
| 10908 |
+
"step": 1557
|
| 10909 |
+
},
|
| 10910 |
+
{
|
| 10911 |
+
"epoch": 0.7965235173824131,
|
| 10912 |
+
"grad_norm": 3.0384457111358643,
|
| 10913 |
+
"learning_rate": 1.092350704067614e-05,
|
| 10914 |
+
"loss": 1.0165,
|
| 10915 |
+
"step": 1558
|
| 10916 |
+
},
|
| 10917 |
+
{
|
| 10918 |
+
"epoch": 0.7970347648261759,
|
| 10919 |
+
"grad_norm": 2.8058857917785645,
|
| 10920 |
+
"learning_rate": 1.0870762974937598e-05,
|
| 10921 |
+
"loss": 0.8881,
|
| 10922 |
+
"step": 1559
|
| 10923 |
+
},
|
| 10924 |
+
{
|
| 10925 |
+
"epoch": 0.7975460122699386,
|
| 10926 |
+
"grad_norm": 2.571446418762207,
|
| 10927 |
+
"learning_rate": 1.0818131019386252e-05,
|
| 10928 |
+
"loss": 0.8571,
|
| 10929 |
+
"step": 1560
|
| 10930 |
+
},
|
| 10931 |
+
{
|
| 10932 |
+
"epoch": 0.7980572597137015,
|
| 10933 |
+
"grad_norm": 3.1314709186553955,
|
| 10934 |
+
"learning_rate": 1.0765611324819247e-05,
|
| 10935 |
+
"loss": 0.8383,
|
| 10936 |
+
"step": 1561
|
| 10937 |
+
},
|
| 10938 |
+
{
|
| 10939 |
+
"epoch": 0.7985685071574642,
|
| 10940 |
+
"grad_norm": 2.915090799331665,
|
| 10941 |
+
"learning_rate": 1.0713204041712145e-05,
|
| 10942 |
+
"loss": 0.8493,
|
| 10943 |
+
"step": 1562
|
| 10944 |
+
},
|
| 10945 |
+
{
|
| 10946 |
+
"epoch": 0.799079754601227,
|
| 10947 |
+
"grad_norm": 3.2132315635681152,
|
| 10948 |
+
"learning_rate": 1.066090932021837e-05,
|
| 10949 |
+
"loss": 0.9181,
|
| 10950 |
+
"step": 1563
|
| 10951 |
+
},
|
| 10952 |
+
{
|
| 10953 |
+
"epoch": 0.7995910020449898,
|
| 10954 |
+
"grad_norm": 3.073258399963379,
|
| 10955 |
+
"learning_rate": 1.060872731016892e-05,
|
| 10956 |
+
"loss": 0.8863,
|
| 10957 |
+
"step": 1564
|
| 10958 |
+
},
|
| 10959 |
+
{
|
| 10960 |
+
"epoch": 0.8001022494887525,
|
| 10961 |
+
"grad_norm": 3.104003667831421,
|
| 10962 |
+
"learning_rate": 1.0556658161071792e-05,
|
| 10963 |
+
"loss": 0.9108,
|
| 10964 |
+
"step": 1565
|
| 10965 |
+
},
|
| 10966 |
+
{
|
| 10967 |
+
"epoch": 0.8006134969325154,
|
| 10968 |
+
"grad_norm": 3.0709762573242188,
|
| 10969 |
+
"learning_rate": 1.0504702022111661e-05,
|
| 10970 |
+
"loss": 0.8055,
|
| 10971 |
+
"step": 1566
|
| 10972 |
+
},
|
| 10973 |
+
{
|
| 10974 |
+
"epoch": 0.8011247443762781,
|
| 10975 |
+
"grad_norm": 3.2626142501831055,
|
| 10976 |
+
"learning_rate": 1.0452859042149382e-05,
|
| 10977 |
+
"loss": 0.8732,
|
| 10978 |
+
"step": 1567
|
| 10979 |
+
},
|
| 10980 |
+
{
|
| 10981 |
+
"epoch": 0.8016359918200409,
|
| 10982 |
+
"grad_norm": 3.4502737522125244,
|
| 10983 |
+
"learning_rate": 1.040112936972164e-05,
|
| 10984 |
+
"loss": 0.9525,
|
| 10985 |
+
"step": 1568
|
| 10986 |
+
},
|
| 10987 |
+
{
|
| 10988 |
+
"epoch": 0.8021472392638037,
|
| 10989 |
+
"grad_norm": 3.2807888984680176,
|
| 10990 |
+
"learning_rate": 1.0349513153040436e-05,
|
| 10991 |
+
"loss": 0.937,
|
| 10992 |
+
"step": 1569
|
| 10993 |
+
},
|
| 10994 |
+
{
|
| 10995 |
+
"epoch": 0.8026584867075665,
|
| 10996 |
+
"grad_norm": 3.3237321376800537,
|
| 10997 |
+
"learning_rate": 1.0298010539992748e-05,
|
| 10998 |
+
"loss": 0.8712,
|
| 10999 |
+
"step": 1570
|
| 11000 |
+
},
|
| 11001 |
+
{
|
| 11002 |
+
"epoch": 0.8031697341513292,
|
| 11003 |
+
"grad_norm": 3.9305286407470703,
|
| 11004 |
+
"learning_rate": 1.0246621678140023e-05,
|
| 11005 |
+
"loss": 0.9547,
|
| 11006 |
+
"step": 1571
|
| 11007 |
+
},
|
| 11008 |
+
{
|
| 11009 |
+
"epoch": 0.803680981595092,
|
| 11010 |
+
"grad_norm": 3.3259196281433105,
|
| 11011 |
+
"learning_rate": 1.0195346714717813e-05,
|
| 11012 |
+
"loss": 0.8361,
|
| 11013 |
+
"step": 1572
|
| 11014 |
+
},
|
| 11015 |
+
{
|
| 11016 |
+
"epoch": 0.8041922290388548,
|
| 11017 |
+
"grad_norm": 3.3074426651000977,
|
| 11018 |
+
"learning_rate": 1.0144185796635359e-05,
|
| 11019 |
+
"loss": 0.7841,
|
| 11020 |
+
"step": 1573
|
| 11021 |
+
},
|
| 11022 |
+
{
|
| 11023 |
+
"epoch": 0.8047034764826176,
|
| 11024 |
+
"grad_norm": 3.421103000640869,
|
| 11025 |
+
"learning_rate": 1.00931390704751e-05,
|
| 11026 |
+
"loss": 0.8898,
|
| 11027 |
+
"step": 1574
|
| 11028 |
+
},
|
| 11029 |
+
{
|
| 11030 |
+
"epoch": 0.8052147239263804,
|
| 11031 |
+
"grad_norm": 3.6833465099334717,
|
| 11032 |
+
"learning_rate": 1.0042206682492372e-05,
|
| 11033 |
+
"loss": 0.9386,
|
| 11034 |
+
"step": 1575
|
| 11035 |
+
},
|
| 11036 |
+
{
|
| 11037 |
+
"epoch": 0.8057259713701431,
|
| 11038 |
+
"grad_norm": 3.3095312118530273,
|
| 11039 |
+
"learning_rate": 9.991388778614824e-06,
|
| 11040 |
+
"loss": 0.8398,
|
| 11041 |
+
"step": 1576
|
| 11042 |
+
},
|
| 11043 |
+
{
|
| 11044 |
+
"epoch": 0.806237218813906,
|
| 11045 |
+
"grad_norm": 3.4535927772521973,
|
| 11046 |
+
"learning_rate": 9.940685504442183e-06,
|
| 11047 |
+
"loss": 0.86,
|
| 11048 |
+
"step": 1577
|
| 11049 |
+
},
|
| 11050 |
+
{
|
| 11051 |
+
"epoch": 0.8067484662576687,
|
| 11052 |
+
"grad_norm": 3.7982585430145264,
|
| 11053 |
+
"learning_rate": 9.89009700524568e-06,
|
| 11054 |
+
"loss": 0.9577,
|
| 11055 |
+
"step": 1578
|
| 11056 |
+
},
|
| 11057 |
+
{
|
| 11058 |
+
"epoch": 0.8072597137014315,
|
| 11059 |
+
"grad_norm": 3.5163521766662598,
|
| 11060 |
+
"learning_rate": 9.83962342596776e-06,
|
| 11061 |
+
"loss": 0.7962,
|
| 11062 |
+
"step": 1579
|
| 11063 |
+
},
|
| 11064 |
+
{
|
| 11065 |
+
"epoch": 0.8077709611451943,
|
| 11066 |
+
"grad_norm": 3.4356038570404053,
|
| 11067 |
+
"learning_rate": 9.789264911221546e-06,
|
| 11068 |
+
"loss": 0.8193,
|
| 11069 |
+
"step": 1580
|
| 11070 |
+
},
|
| 11071 |
+
{
|
| 11072 |
+
"epoch": 0.808282208588957,
|
| 11073 |
+
"grad_norm": 3.920682907104492,
|
| 11074 |
+
"learning_rate": 9.739021605290549e-06,
|
| 11075 |
+
"loss": 0.8903,
|
| 11076 |
+
"step": 1581
|
| 11077 |
+
},
|
| 11078 |
+
{
|
| 11079 |
+
"epoch": 0.8087934560327198,
|
| 11080 |
+
"grad_norm": 3.7626466751098633,
|
| 11081 |
+
"learning_rate": 9.688893652128151e-06,
|
| 11082 |
+
"loss": 0.8662,
|
| 11083 |
+
"step": 1582
|
| 11084 |
+
},
|
| 11085 |
+
{
|
| 11086 |
+
"epoch": 0.8093047034764826,
|
| 11087 |
+
"grad_norm": 4.124518394470215,
|
| 11088 |
+
"learning_rate": 9.638881195357224e-06,
|
| 11089 |
+
"loss": 0.8829,
|
| 11090 |
+
"step": 1583
|
| 11091 |
+
},
|
| 11092 |
+
{
|
| 11093 |
+
"epoch": 0.8098159509202454,
|
| 11094 |
+
"grad_norm": 3.74548602104187,
|
| 11095 |
+
"learning_rate": 9.588984378269783e-06,
|
| 11096 |
+
"loss": 0.7028,
|
| 11097 |
+
"step": 1584
|
| 11098 |
+
},
|
| 11099 |
+
{
|
| 11100 |
+
"epoch": 0.8103271983640081,
|
| 11101 |
+
"grad_norm": 3.769134998321533,
|
| 11102 |
+
"learning_rate": 9.539203343826469e-06,
|
| 11103 |
+
"loss": 0.9024,
|
| 11104 |
+
"step": 1585
|
| 11105 |
+
},
|
| 11106 |
+
{
|
| 11107 |
+
"epoch": 0.810838445807771,
|
| 11108 |
+
"grad_norm": 3.955554723739624,
|
| 11109 |
+
"learning_rate": 9.489538234656214e-06,
|
| 11110 |
+
"loss": 0.8195,
|
| 11111 |
+
"step": 1586
|
| 11112 |
+
},
|
| 11113 |
+
{
|
| 11114 |
+
"epoch": 0.8113496932515337,
|
| 11115 |
+
"grad_norm": 3.704868793487549,
|
| 11116 |
+
"learning_rate": 9.439989193055788e-06,
|
| 11117 |
+
"loss": 0.7934,
|
| 11118 |
+
"step": 1587
|
| 11119 |
+
},
|
| 11120 |
+
{
|
| 11121 |
+
"epoch": 0.8118609406952966,
|
| 11122 |
+
"grad_norm": 3.6619386672973633,
|
| 11123 |
+
"learning_rate": 9.39055636098945e-06,
|
| 11124 |
+
"loss": 0.868,
|
| 11125 |
+
"step": 1588
|
| 11126 |
+
},
|
| 11127 |
+
{
|
| 11128 |
+
"epoch": 0.8123721881390593,
|
| 11129 |
+
"grad_norm": 4.098423480987549,
|
| 11130 |
+
"learning_rate": 9.341239880088465e-06,
|
| 11131 |
+
"loss": 0.8754,
|
| 11132 |
+
"step": 1589
|
| 11133 |
+
},
|
| 11134 |
+
{
|
| 11135 |
+
"epoch": 0.8128834355828221,
|
| 11136 |
+
"grad_norm": 4.314637184143066,
|
| 11137 |
+
"learning_rate": 9.292039891650784e-06,
|
| 11138 |
+
"loss": 0.8408,
|
| 11139 |
+
"step": 1590
|
| 11140 |
+
},
|
| 11141 |
+
{
|
| 11142 |
+
"epoch": 0.8133946830265849,
|
| 11143 |
+
"grad_norm": 4.415641784667969,
|
| 11144 |
+
"learning_rate": 9.24295653664053e-06,
|
| 11145 |
+
"loss": 0.8727,
|
| 11146 |
+
"step": 1591
|
| 11147 |
+
},
|
| 11148 |
+
{
|
| 11149 |
+
"epoch": 0.8139059304703476,
|
| 11150 |
+
"grad_norm": 4.379350185394287,
|
| 11151 |
+
"learning_rate": 9.193989955687715e-06,
|
| 11152 |
+
"loss": 0.8209,
|
| 11153 |
+
"step": 1592
|
| 11154 |
+
},
|
| 11155 |
+
{
|
| 11156 |
+
"epoch": 0.8144171779141104,
|
| 11157 |
+
"grad_norm": 4.419501304626465,
|
| 11158 |
+
"learning_rate": 9.145140289087755e-06,
|
| 11159 |
+
"loss": 0.9432,
|
| 11160 |
+
"step": 1593
|
| 11161 |
+
},
|
| 11162 |
+
{
|
| 11163 |
+
"epoch": 0.8149284253578732,
|
| 11164 |
+
"grad_norm": 4.199825763702393,
|
| 11165 |
+
"learning_rate": 9.096407676801077e-06,
|
| 11166 |
+
"loss": 0.7578,
|
| 11167 |
+
"step": 1594
|
| 11168 |
+
},
|
| 11169 |
+
{
|
| 11170 |
+
"epoch": 0.815439672801636,
|
| 11171 |
+
"grad_norm": 5.105489253997803,
|
| 11172 |
+
"learning_rate": 9.047792258452742e-06,
|
| 11173 |
+
"loss": 0.8932,
|
| 11174 |
+
"step": 1595
|
| 11175 |
+
},
|
| 11176 |
+
{
|
| 11177 |
+
"epoch": 0.8159509202453987,
|
| 11178 |
+
"grad_norm": 5.17318058013916,
|
| 11179 |
+
"learning_rate": 8.999294173332058e-06,
|
| 11180 |
+
"loss": 0.9014,
|
| 11181 |
+
"step": 1596
|
| 11182 |
+
},
|
| 11183 |
+
{
|
| 11184 |
+
"epoch": 0.8164621676891616,
|
| 11185 |
+
"grad_norm": 4.764847755432129,
|
| 11186 |
+
"learning_rate": 8.950913560392132e-06,
|
| 11187 |
+
"loss": 0.7204,
|
| 11188 |
+
"step": 1597
|
| 11189 |
+
},
|
| 11190 |
+
{
|
| 11191 |
+
"epoch": 0.8169734151329243,
|
| 11192 |
+
"grad_norm": 5.115185737609863,
|
| 11193 |
+
"learning_rate": 8.902650558249499e-06,
|
| 11194 |
+
"loss": 0.7984,
|
| 11195 |
+
"step": 1598
|
| 11196 |
+
},
|
| 11197 |
+
{
|
| 11198 |
+
"epoch": 0.8174846625766872,
|
| 11199 |
+
"grad_norm": 5.964483261108398,
|
| 11200 |
+
"learning_rate": 8.854505305183752e-06,
|
| 11201 |
+
"loss": 0.709,
|
| 11202 |
+
"step": 1599
|
| 11203 |
+
},
|
| 11204 |
+
{
|
| 11205 |
+
"epoch": 0.8179959100204499,
|
| 11206 |
+
"grad_norm": 5.853104591369629,
|
| 11207 |
+
"learning_rate": 8.80647793913708e-06,
|
| 11208 |
+
"loss": 0.4204,
|
| 11209 |
+
"step": 1600
|
| 11210 |
+
},
|
| 11211 |
+
{
|
| 11212 |
+
"epoch": 0.8185071574642127,
|
| 11213 |
+
"grad_norm": 1.909232258796692,
|
| 11214 |
+
"learning_rate": 8.758568597713946e-06,
|
| 11215 |
+
"loss": 0.9726,
|
| 11216 |
+
"step": 1601
|
| 11217 |
+
},
|
| 11218 |
+
{
|
| 11219 |
+
"epoch": 0.8190184049079755,
|
| 11220 |
+
"grad_norm": 2.361161231994629,
|
| 11221 |
+
"learning_rate": 8.710777418180615e-06,
|
| 11222 |
+
"loss": 0.939,
|
| 11223 |
+
"step": 1602
|
| 11224 |
+
},
|
| 11225 |
+
{
|
| 11226 |
+
"epoch": 0.8195296523517382,
|
| 11227 |
+
"grad_norm": 2.425341844558716,
|
| 11228 |
+
"learning_rate": 8.663104537464866e-06,
|
| 11229 |
+
"loss": 0.8918,
|
| 11230 |
+
"step": 1603
|
| 11231 |
+
},
|
| 11232 |
+
{
|
| 11233 |
+
"epoch": 0.820040899795501,
|
| 11234 |
+
"grad_norm": 2.4736742973327637,
|
| 11235 |
+
"learning_rate": 8.615550092155478e-06,
|
| 11236 |
+
"loss": 0.9242,
|
| 11237 |
+
"step": 1604
|
| 11238 |
+
},
|
| 11239 |
+
{
|
| 11240 |
+
"epoch": 0.8205521472392638,
|
| 11241 |
+
"grad_norm": 2.6613001823425293,
|
| 11242 |
+
"learning_rate": 8.568114218501922e-06,
|
| 11243 |
+
"loss": 0.9088,
|
| 11244 |
+
"step": 1605
|
| 11245 |
+
},
|
| 11246 |
+
{
|
| 11247 |
+
"epoch": 0.8210633946830266,
|
| 11248 |
+
"grad_norm": 2.5696263313293457,
|
| 11249 |
+
"learning_rate": 8.520797052413931e-06,
|
| 11250 |
+
"loss": 0.8776,
|
| 11251 |
+
"step": 1606
|
| 11252 |
+
},
|
| 11253 |
+
{
|
| 11254 |
+
"epoch": 0.8215746421267893,
|
| 11255 |
+
"grad_norm": 2.944065570831299,
|
| 11256 |
+
"learning_rate": 8.473598729461163e-06,
|
| 11257 |
+
"loss": 0.9622,
|
| 11258 |
+
"step": 1607
|
| 11259 |
+
},
|
| 11260 |
+
{
|
| 11261 |
+
"epoch": 0.8220858895705522,
|
| 11262 |
+
"grad_norm": 2.8112123012542725,
|
| 11263 |
+
"learning_rate": 8.426519384872733e-06,
|
| 11264 |
+
"loss": 0.905,
|
| 11265 |
+
"step": 1608
|
| 11266 |
+
},
|
| 11267 |
+
{
|
| 11268 |
+
"epoch": 0.8225971370143149,
|
| 11269 |
+
"grad_norm": 2.9660046100616455,
|
| 11270 |
+
"learning_rate": 8.379559153536909e-06,
|
| 11271 |
+
"loss": 0.9772,
|
| 11272 |
+
"step": 1609
|
| 11273 |
+
},
|
| 11274 |
+
{
|
| 11275 |
+
"epoch": 0.8231083844580777,
|
| 11276 |
+
"grad_norm": 2.898458242416382,
|
| 11277 |
+
"learning_rate": 8.332718170000647e-06,
|
| 11278 |
+
"loss": 0.9601,
|
| 11279 |
+
"step": 1610
|
| 11280 |
+
},
|
| 11281 |
+
{
|
| 11282 |
+
"epoch": 0.8236196319018405,
|
| 11283 |
+
"grad_norm": 2.934843063354492,
|
| 11284 |
+
"learning_rate": 8.285996568469245e-06,
|
| 11285 |
+
"loss": 0.9621,
|
| 11286 |
+
"step": 1611
|
| 11287 |
+
},
|
| 11288 |
+
{
|
| 11289 |
+
"epoch": 0.8241308793456033,
|
| 11290 |
+
"grad_norm": 2.774526834487915,
|
| 11291 |
+
"learning_rate": 8.239394482805996e-06,
|
| 11292 |
+
"loss": 0.8366,
|
| 11293 |
+
"step": 1612
|
| 11294 |
+
},
|
| 11295 |
+
{
|
| 11296 |
+
"epoch": 0.8246421267893661,
|
| 11297 |
+
"grad_norm": 3.195984363555908,
|
| 11298 |
+
"learning_rate": 8.192912046531732e-06,
|
| 11299 |
+
"loss": 0.8993,
|
| 11300 |
+
"step": 1613
|
| 11301 |
+
},
|
| 11302 |
+
{
|
| 11303 |
+
"epoch": 0.8251533742331288,
|
| 11304 |
+
"grad_norm": 2.9150772094726562,
|
| 11305 |
+
"learning_rate": 8.14654939282447e-06,
|
| 11306 |
+
"loss": 0.818,
|
| 11307 |
+
"step": 1614
|
| 11308 |
+
},
|
| 11309 |
+
{
|
| 11310 |
+
"epoch": 0.8256646216768916,
|
| 11311 |
+
"grad_norm": 3.1671900749206543,
|
| 11312 |
+
"learning_rate": 8.10030665451904e-06,
|
| 11313 |
+
"loss": 0.9031,
|
| 11314 |
+
"step": 1615
|
| 11315 |
+
},
|
| 11316 |
+
{
|
| 11317 |
+
"epoch": 0.8261758691206544,
|
| 11318 |
+
"grad_norm": 3.0718796253204346,
|
| 11319 |
+
"learning_rate": 8.054183964106738e-06,
|
| 11320 |
+
"loss": 0.902,
|
| 11321 |
+
"step": 1616
|
| 11322 |
+
},
|
| 11323 |
+
{
|
| 11324 |
+
"epoch": 0.8266871165644172,
|
| 11325 |
+
"grad_norm": 3.13236403465271,
|
| 11326 |
+
"learning_rate": 8.008181453734832e-06,
|
| 11327 |
+
"loss": 0.8966,
|
| 11328 |
+
"step": 1617
|
| 11329 |
+
},
|
| 11330 |
+
{
|
| 11331 |
+
"epoch": 0.8271983640081799,
|
| 11332 |
+
"grad_norm": 3.3213281631469727,
|
| 11333 |
+
"learning_rate": 7.96229925520634e-06,
|
| 11334 |
+
"loss": 0.9051,
|
| 11335 |
+
"step": 1618
|
| 11336 |
+
},
|
| 11337 |
+
{
|
| 11338 |
+
"epoch": 0.8277096114519428,
|
| 11339 |
+
"grad_norm": 3.212383508682251,
|
| 11340 |
+
"learning_rate": 7.916537499979509e-06,
|
| 11341 |
+
"loss": 0.8877,
|
| 11342 |
+
"step": 1619
|
| 11343 |
+
},
|
| 11344 |
+
{
|
| 11345 |
+
"epoch": 0.8282208588957055,
|
| 11346 |
+
"grad_norm": 3.477694272994995,
|
| 11347 |
+
"learning_rate": 7.870896319167548e-06,
|
| 11348 |
+
"loss": 0.8837,
|
| 11349 |
+
"step": 1620
|
| 11350 |
+
},
|
| 11351 |
+
{
|
| 11352 |
+
"epoch": 0.8287321063394683,
|
| 11353 |
+
"grad_norm": 3.3036673069000244,
|
| 11354 |
+
"learning_rate": 7.825375843538163e-06,
|
| 11355 |
+
"loss": 0.8893,
|
| 11356 |
+
"step": 1621
|
| 11357 |
+
},
|
| 11358 |
+
{
|
| 11359 |
+
"epoch": 0.8292433537832311,
|
| 11360 |
+
"grad_norm": 3.0736191272735596,
|
| 11361 |
+
"learning_rate": 7.77997620351324e-06,
|
| 11362 |
+
"loss": 0.7746,
|
| 11363 |
+
"step": 1622
|
| 11364 |
+
},
|
| 11365 |
+
{
|
| 11366 |
+
"epoch": 0.8297546012269938,
|
| 11367 |
+
"grad_norm": 3.514597177505493,
|
| 11368 |
+
"learning_rate": 7.734697529168483e-06,
|
| 11369 |
+
"loss": 0.9255,
|
| 11370 |
+
"step": 1623
|
| 11371 |
+
},
|
| 11372 |
+
{
|
| 11373 |
+
"epoch": 0.8302658486707567,
|
| 11374 |
+
"grad_norm": 3.152017831802368,
|
| 11375 |
+
"learning_rate": 7.689539950232977e-06,
|
| 11376 |
+
"loss": 0.8449,
|
| 11377 |
+
"step": 1624
|
| 11378 |
+
},
|
| 11379 |
+
{
|
| 11380 |
+
"epoch": 0.8307770961145194,
|
| 11381 |
+
"grad_norm": 3.652573823928833,
|
| 11382 |
+
"learning_rate": 7.644503596088865e-06,
|
| 11383 |
+
"loss": 0.884,
|
| 11384 |
+
"step": 1625
|
| 11385 |
+
},
|
| 11386 |
+
{
|
| 11387 |
+
"epoch": 0.8312883435582822,
|
| 11388 |
+
"grad_norm": 3.5253610610961914,
|
| 11389 |
+
"learning_rate": 7.599588595770957e-06,
|
| 11390 |
+
"loss": 0.8573,
|
| 11391 |
+
"step": 1626
|
| 11392 |
+
},
|
| 11393 |
+
{
|
| 11394 |
+
"epoch": 0.831799591002045,
|
| 11395 |
+
"grad_norm": 3.6805419921875,
|
| 11396 |
+
"learning_rate": 7.554795077966409e-06,
|
| 11397 |
+
"loss": 0.7815,
|
| 11398 |
+
"step": 1627
|
| 11399 |
+
},
|
| 11400 |
+
{
|
| 11401 |
+
"epoch": 0.8323108384458078,
|
| 11402 |
+
"grad_norm": 3.5299296379089355,
|
| 11403 |
+
"learning_rate": 7.510123171014255e-06,
|
| 11404 |
+
"loss": 0.9062,
|
| 11405 |
+
"step": 1628
|
| 11406 |
+
},
|
| 11407 |
+
{
|
| 11408 |
+
"epoch": 0.8328220858895705,
|
| 11409 |
+
"grad_norm": 3.8087053298950195,
|
| 11410 |
+
"learning_rate": 7.4655730029051575e-06,
|
| 11411 |
+
"loss": 0.9056,
|
| 11412 |
+
"step": 1629
|
| 11413 |
+
},
|
| 11414 |
+
{
|
| 11415 |
+
"epoch": 0.8333333333333334,
|
| 11416 |
+
"grad_norm": 3.496819496154785,
|
| 11417 |
+
"learning_rate": 7.42114470128093e-06,
|
| 11418 |
+
"loss": 0.7842,
|
| 11419 |
+
"step": 1630
|
| 11420 |
+
},
|
| 11421 |
+
{
|
| 11422 |
+
"epoch": 0.8338445807770961,
|
| 11423 |
+
"grad_norm": 3.998002767562866,
|
| 11424 |
+
"learning_rate": 7.376838393434265e-06,
|
| 11425 |
+
"loss": 0.8565,
|
| 11426 |
+
"step": 1631
|
| 11427 |
+
},
|
| 11428 |
+
{
|
| 11429 |
+
"epoch": 0.8343558282208589,
|
| 11430 |
+
"grad_norm": 3.822242498397827,
|
| 11431 |
+
"learning_rate": 7.332654206308298e-06,
|
| 11432 |
+
"loss": 0.8017,
|
| 11433 |
+
"step": 1632
|
| 11434 |
+
},
|
| 11435 |
+
{
|
| 11436 |
+
"epoch": 0.8348670756646217,
|
| 11437 |
+
"grad_norm": 3.5486857891082764,
|
| 11438 |
+
"learning_rate": 7.288592266496286e-06,
|
| 11439 |
+
"loss": 0.8265,
|
| 11440 |
+
"step": 1633
|
| 11441 |
+
},
|
| 11442 |
+
{
|
| 11443 |
+
"epoch": 0.8353783231083844,
|
| 11444 |
+
"grad_norm": 3.6276445388793945,
|
| 11445 |
+
"learning_rate": 7.2446527002412225e-06,
|
| 11446 |
+
"loss": 0.851,
|
| 11447 |
+
"step": 1634
|
| 11448 |
+
},
|
| 11449 |
+
{
|
| 11450 |
+
"epoch": 0.8358895705521472,
|
| 11451 |
+
"grad_norm": 3.7170212268829346,
|
| 11452 |
+
"learning_rate": 7.20083563343551e-06,
|
| 11453 |
+
"loss": 0.7564,
|
| 11454 |
+
"step": 1635
|
| 11455 |
+
},
|
| 11456 |
+
{
|
| 11457 |
+
"epoch": 0.83640081799591,
|
| 11458 |
+
"grad_norm": 3.684058666229248,
|
| 11459 |
+
"learning_rate": 7.157141191620548e-06,
|
| 11460 |
+
"loss": 0.8008,
|
| 11461 |
+
"step": 1636
|
| 11462 |
+
},
|
| 11463 |
+
{
|
| 11464 |
+
"epoch": 0.8369120654396728,
|
| 11465 |
+
"grad_norm": 3.752002000808716,
|
| 11466 |
+
"learning_rate": 7.1135694999864e-06,
|
| 11467 |
+
"loss": 0.7732,
|
| 11468 |
+
"step": 1637
|
| 11469 |
+
},
|
| 11470 |
+
{
|
| 11471 |
+
"epoch": 0.8374233128834356,
|
| 11472 |
+
"grad_norm": 4.312587738037109,
|
| 11473 |
+
"learning_rate": 7.070120683371462e-06,
|
| 11474 |
+
"loss": 0.808,
|
| 11475 |
+
"step": 1638
|
| 11476 |
+
},
|
| 11477 |
+
{
|
| 11478 |
+
"epoch": 0.8379345603271984,
|
| 11479 |
+
"grad_norm": 4.659461498260498,
|
| 11480 |
+
"learning_rate": 7.026794866262048e-06,
|
| 11481 |
+
"loss": 0.8293,
|
| 11482 |
+
"step": 1639
|
| 11483 |
+
},
|
| 11484 |
+
{
|
| 11485 |
+
"epoch": 0.8384458077709611,
|
| 11486 |
+
"grad_norm": 4.59619665145874,
|
| 11487 |
+
"learning_rate": 6.983592172792086e-06,
|
| 11488 |
+
"loss": 0.851,
|
| 11489 |
+
"step": 1640
|
| 11490 |
+
},
|
| 11491 |
+
{
|
| 11492 |
+
"epoch": 0.838957055214724,
|
| 11493 |
+
"grad_norm": 4.543966293334961,
|
| 11494 |
+
"learning_rate": 6.940512726742715e-06,
|
| 11495 |
+
"loss": 0.8155,
|
| 11496 |
+
"step": 1641
|
| 11497 |
+
},
|
| 11498 |
+
{
|
| 11499 |
+
"epoch": 0.8394683026584867,
|
| 11500 |
+
"grad_norm": 4.165318489074707,
|
| 11501 |
+
"learning_rate": 6.897556651542003e-06,
|
| 11502 |
+
"loss": 0.7734,
|
| 11503 |
+
"step": 1642
|
| 11504 |
+
},
|
| 11505 |
+
{
|
| 11506 |
+
"epoch": 0.8399795501022495,
|
| 11507 |
+
"grad_norm": 4.547415733337402,
|
| 11508 |
+
"learning_rate": 6.854724070264451e-06,
|
| 11509 |
+
"loss": 0.9181,
|
| 11510 |
+
"step": 1643
|
| 11511 |
+
},
|
| 11512 |
+
{
|
| 11513 |
+
"epoch": 0.8404907975460123,
|
| 11514 |
+
"grad_norm": 4.19881010055542,
|
| 11515 |
+
"learning_rate": 6.812015105630842e-06,
|
| 11516 |
+
"loss": 0.7814,
|
| 11517 |
+
"step": 1644
|
| 11518 |
+
},
|
| 11519 |
+
{
|
| 11520 |
+
"epoch": 0.841002044989775,
|
| 11521 |
+
"grad_norm": 4.672346115112305,
|
| 11522 |
+
"learning_rate": 6.769429880007705e-06,
|
| 11523 |
+
"loss": 0.6833,
|
| 11524 |
+
"step": 1645
|
| 11525 |
+
},
|
| 11526 |
+
{
|
| 11527 |
+
"epoch": 0.8415132924335378,
|
| 11528 |
+
"grad_norm": 4.165514945983887,
|
| 11529 |
+
"learning_rate": 6.7269685154070895e-06,
|
| 11530 |
+
"loss": 0.5298,
|
| 11531 |
+
"step": 1646
|
| 11532 |
+
},
|
| 11533 |
+
{
|
| 11534 |
+
"epoch": 0.8420245398773006,
|
| 11535 |
+
"grad_norm": 5.330099105834961,
|
| 11536 |
+
"learning_rate": 6.6846311334861415e-06,
|
| 11537 |
+
"loss": 0.6836,
|
| 11538 |
+
"step": 1647
|
| 11539 |
+
},
|
| 11540 |
+
{
|
| 11541 |
+
"epoch": 0.8425357873210634,
|
| 11542 |
+
"grad_norm": 4.8200860023498535,
|
| 11543 |
+
"learning_rate": 6.642417855546768e-06,
|
| 11544 |
+
"loss": 0.7025,
|
| 11545 |
+
"step": 1648
|
| 11546 |
+
},
|
| 11547 |
+
{
|
| 11548 |
+
"epoch": 0.8430470347648262,
|
| 11549 |
+
"grad_norm": 5.211686611175537,
|
| 11550 |
+
"learning_rate": 6.600328802535355e-06,
|
| 11551 |
+
"loss": 0.5772,
|
| 11552 |
+
"step": 1649
|
| 11553 |
+
},
|
| 11554 |
+
{
|
| 11555 |
+
"epoch": 0.843558282208589,
|
| 11556 |
+
"grad_norm": 5.803190231323242,
|
| 11557 |
+
"learning_rate": 6.558364095042302e-06,
|
| 11558 |
+
"loss": 0.632,
|
| 11559 |
+
"step": 1650
|
| 11560 |
+
},
|
| 11561 |
+
{
|
| 11562 |
+
"epoch": 0.8440695296523517,
|
| 11563 |
+
"grad_norm": 2.0162458419799805,
|
| 11564 |
+
"learning_rate": 6.516523853301804e-06,
|
| 11565 |
+
"loss": 0.7469,
|
| 11566 |
+
"step": 1651
|
| 11567 |
+
},
|
| 11568 |
+
{
|
| 11569 |
+
"epoch": 0.8445807770961146,
|
| 11570 |
+
"grad_norm": 2.1648364067077637,
|
| 11571 |
+
"learning_rate": 6.474808197191401e-06,
|
| 11572 |
+
"loss": 0.9474,
|
| 11573 |
+
"step": 1652
|
| 11574 |
+
},
|
| 11575 |
+
{
|
| 11576 |
+
"epoch": 0.8450920245398773,
|
| 11577 |
+
"grad_norm": 2.424511194229126,
|
| 11578 |
+
"learning_rate": 6.433217246231704e-06,
|
| 11579 |
+
"loss": 0.9164,
|
| 11580 |
+
"step": 1653
|
| 11581 |
+
},
|
| 11582 |
+
{
|
| 11583 |
+
"epoch": 0.84560327198364,
|
| 11584 |
+
"grad_norm": 2.5511527061462402,
|
| 11585 |
+
"learning_rate": 6.391751119586003e-06,
|
| 11586 |
+
"loss": 0.9075,
|
| 11587 |
+
"step": 1654
|
| 11588 |
+
},
|
| 11589 |
+
{
|
| 11590 |
+
"epoch": 0.8461145194274029,
|
| 11591 |
+
"grad_norm": 2.9107165336608887,
|
| 11592 |
+
"learning_rate": 6.350409936059998e-06,
|
| 11593 |
+
"loss": 0.9481,
|
| 11594 |
+
"step": 1655
|
| 11595 |
+
},
|
| 11596 |
+
{
|
| 11597 |
+
"epoch": 0.8466257668711656,
|
| 11598 |
+
"grad_norm": 2.7874770164489746,
|
| 11599 |
+
"learning_rate": 6.3091938141013495e-06,
|
| 11600 |
+
"loss": 0.9119,
|
| 11601 |
+
"step": 1656
|
| 11602 |
+
},
|
| 11603 |
+
{
|
| 11604 |
+
"epoch": 0.8471370143149284,
|
| 11605 |
+
"grad_norm": 2.7134737968444824,
|
| 11606 |
+
"learning_rate": 6.268102871799459e-06,
|
| 11607 |
+
"loss": 0.8897,
|
| 11608 |
+
"step": 1657
|
| 11609 |
+
},
|
| 11610 |
+
{
|
| 11611 |
+
"epoch": 0.8476482617586912,
|
| 11612 |
+
"grad_norm": 2.6800270080566406,
|
| 11613 |
+
"learning_rate": 6.227137226885027e-06,
|
| 11614 |
+
"loss": 0.8625,
|
| 11615 |
+
"step": 1658
|
| 11616 |
+
},
|
| 11617 |
+
{
|
| 11618 |
+
"epoch": 0.848159509202454,
|
| 11619 |
+
"grad_norm": 3.039987564086914,
|
| 11620 |
+
"learning_rate": 6.186296996729796e-06,
|
| 11621 |
+
"loss": 0.9901,
|
| 11622 |
+
"step": 1659
|
| 11623 |
+
},
|
| 11624 |
+
{
|
| 11625 |
+
"epoch": 0.8486707566462167,
|
| 11626 |
+
"grad_norm": 2.883843183517456,
|
| 11627 |
+
"learning_rate": 6.145582298346153e-06,
|
| 11628 |
+
"loss": 0.8557,
|
| 11629 |
+
"step": 1660
|
| 11630 |
+
},
|
| 11631 |
+
{
|
| 11632 |
+
"epoch": 0.8491820040899796,
|
| 11633 |
+
"grad_norm": 3.040339469909668,
|
| 11634 |
+
"learning_rate": 6.104993248386831e-06,
|
| 11635 |
+
"loss": 0.9266,
|
| 11636 |
+
"step": 1661
|
| 11637 |
+
},
|
| 11638 |
+
{
|
| 11639 |
+
"epoch": 0.8496932515337423,
|
| 11640 |
+
"grad_norm": 3.1167523860931396,
|
| 11641 |
+
"learning_rate": 6.064529963144583e-06,
|
| 11642 |
+
"loss": 0.922,
|
| 11643 |
+
"step": 1662
|
| 11644 |
+
},
|
| 11645 |
+
{
|
| 11646 |
+
"epoch": 0.8502044989775052,
|
| 11647 |
+
"grad_norm": 3.086543560028076,
|
| 11648 |
+
"learning_rate": 6.024192558551784e-06,
|
| 11649 |
+
"loss": 0.8847,
|
| 11650 |
+
"step": 1663
|
| 11651 |
+
},
|
| 11652 |
+
{
|
| 11653 |
+
"epoch": 0.8507157464212679,
|
| 11654 |
+
"grad_norm": 3.4161489009857178,
|
| 11655 |
+
"learning_rate": 5.98398115018019e-06,
|
| 11656 |
+
"loss": 0.8947,
|
| 11657 |
+
"step": 1664
|
| 11658 |
+
},
|
| 11659 |
+
{
|
| 11660 |
+
"epoch": 0.8512269938650306,
|
| 11661 |
+
"grad_norm": 3.196218252182007,
|
| 11662 |
+
"learning_rate": 5.943895853240533e-06,
|
| 11663 |
+
"loss": 0.8973,
|
| 11664 |
+
"step": 1665
|
| 11665 |
+
},
|
| 11666 |
+
{
|
| 11667 |
+
"epoch": 0.8517382413087935,
|
| 11668 |
+
"grad_norm": 3.152571201324463,
|
| 11669 |
+
"learning_rate": 5.903936782582253e-06,
|
| 11670 |
+
"loss": 0.8851,
|
| 11671 |
+
"step": 1666
|
| 11672 |
+
},
|
| 11673 |
+
{
|
| 11674 |
+
"epoch": 0.8522494887525562,
|
| 11675 |
+
"grad_norm": 3.339434862136841,
|
| 11676 |
+
"learning_rate": 5.8641040526930925e-06,
|
| 11677 |
+
"loss": 0.9871,
|
| 11678 |
+
"step": 1667
|
| 11679 |
+
},
|
| 11680 |
+
{
|
| 11681 |
+
"epoch": 0.852760736196319,
|
| 11682 |
+
"grad_norm": 3.3059394359588623,
|
| 11683 |
+
"learning_rate": 5.824397777698859e-06,
|
| 11684 |
+
"loss": 0.8634,
|
| 11685 |
+
"step": 1668
|
| 11686 |
+
},
|
| 11687 |
+
{
|
| 11688 |
+
"epoch": 0.8532719836400818,
|
| 11689 |
+
"grad_norm": 3.992922306060791,
|
| 11690 |
+
"learning_rate": 5.784818071362996e-06,
|
| 11691 |
+
"loss": 0.9209,
|
| 11692 |
+
"step": 1669
|
| 11693 |
+
},
|
| 11694 |
+
{
|
| 11695 |
+
"epoch": 0.8537832310838446,
|
| 11696 |
+
"grad_norm": 3.1529664993286133,
|
| 11697 |
+
"learning_rate": 5.74536504708637e-06,
|
| 11698 |
+
"loss": 0.889,
|
| 11699 |
+
"step": 1670
|
| 11700 |
+
},
|
| 11701 |
+
{
|
| 11702 |
+
"epoch": 0.8542944785276073,
|
| 11703 |
+
"grad_norm": 3.0963938236236572,
|
| 11704 |
+
"learning_rate": 5.706038817906845e-06,
|
| 11705 |
+
"loss": 0.8722,
|
| 11706 |
+
"step": 1671
|
| 11707 |
+
},
|
| 11708 |
+
{
|
| 11709 |
+
"epoch": 0.8548057259713702,
|
| 11710 |
+
"grad_norm": 3.323939085006714,
|
| 11711 |
+
"learning_rate": 5.666839496499022e-06,
|
| 11712 |
+
"loss": 0.8879,
|
| 11713 |
+
"step": 1672
|
| 11714 |
+
},
|
| 11715 |
+
{
|
| 11716 |
+
"epoch": 0.8553169734151329,
|
| 11717 |
+
"grad_norm": 3.5231525897979736,
|
| 11718 |
+
"learning_rate": 5.6277671951738716e-06,
|
| 11719 |
+
"loss": 0.7867,
|
| 11720 |
+
"step": 1673
|
| 11721 |
}
|
| 11722 |
],
|
| 11723 |
"logging_steps": 1,
|
|
|
|
| 11737 |
"attributes": {}
|
| 11738 |
}
|
| 11739 |
},
|
| 11740 |
+
"total_flos": 6.396906433916436e+17,
|
| 11741 |
"train_batch_size": 8,
|
| 11742 |
"trial_name": null,
|
| 11743 |
"trial_params": null
|