Upload folder using huggingface_hub
Browse files- model-00001-of-00002.safetensors +1 -1
- model-00002-of-00002.safetensors +1 -1
- optimizer.pt +3 -0
- rng_state_0.pth +1 -1
- rng_state_1.pth +1 -1
- rng_state_2.pth +1 -1
- rng_state_3.pth +1 -1
- scheduler.pt +1 -1
- trainer_state.json +2769 -3
model-00001-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 4965799096
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f1a6b10529c49f267dab1303f60c029426a06dd66fca8c15811e26ede178fcee
|
| 3 |
size 4965799096
|
model-00002-of-00002.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1459729952
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:94b4d2fbf9759aadec878bab5277561c46f206ca7bf049bc2b8f706adee8d133
|
| 3 |
size 1459729952
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c9193fa8cf70010c71b5a9976745a677384e893f350b844cef300e95cc242a16
|
| 3 |
+
size 6527220350
|
rng_state_0.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:308f94f9a5c24e1bad5c393d56ae7af7782600f4e791d9c6ac35b22fff2105b6
|
| 3 |
size 15024
|
rng_state_1.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:b056f3c23cb32dc77a2ec9e7651e0b64e4440e21f0fdf969b86bfc56a1cbdf06
|
| 3 |
size 15024
|
rng_state_2.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3f8a05714bc528f4885a2816181652f2303b3e8150f89b56aaee6bec56aa520
|
| 3 |
size 15024
|
rng_state_3.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4f755bd3c330281961e5c03af9d10ce8c1e1678619d384f6f1fd5fd7dce2ff50
|
| 3 |
size 15024
|
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:d581c751b3c27570ea48b8a98b97e22f9e4131f91b7e98afc959d50fcaa099d7
|
| 3 |
size 1064
|
trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch": 1.
|
| 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,
|
|
@@ -5523,6 +5523,2772 @@
|
|
| 5523 |
"learning_rate": 6.156259737516275e-06,
|
| 5524 |
"loss": 0.5108,
|
| 5525 |
"step": 788
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5526 |
}
|
| 5527 |
],
|
| 5528 |
"logging_steps": 1,
|
|
@@ -5542,7 +8308,7 @@
|
|
| 5542 |
"attributes": {}
|
| 5543 |
}
|
| 5544 |
},
|
| 5545 |
-
"total_flos":
|
| 5546 |
"train_batch_size": 2,
|
| 5547 |
"trial_name": null,
|
| 5548 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.499047619047619,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 1182,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5523 |
"learning_rate": 6.156259737516275e-06,
|
| 5524 |
"loss": 0.5108,
|
| 5525 |
"step": 788
|
| 5526 |
+
},
|
| 5527 |
+
{
|
| 5528 |
+
"epoch": 1.0006349206349205,
|
| 5529 |
+
"grad_norm": 0.6875,
|
| 5530 |
+
"learning_rate": 6.153909247219202e-06,
|
| 5531 |
+
"loss": 0.6055,
|
| 5532 |
+
"step": 789
|
| 5533 |
+
},
|
| 5534 |
+
{
|
| 5535 |
+
"epoch": 1.0006349206349205,
|
| 5536 |
+
"eval_loss": 1.1893014907836914,
|
| 5537 |
+
"eval_runtime": 99.4944,
|
| 5538 |
+
"eval_samples_per_second": 43.57,
|
| 5539 |
+
"eval_steps_per_second": 5.448,
|
| 5540 |
+
"step": 789
|
| 5541 |
+
},
|
| 5542 |
+
{
|
| 5543 |
+
"epoch": 1.0012698412698413,
|
| 5544 |
+
"grad_norm": 0.96875,
|
| 5545 |
+
"learning_rate": 6.151555937527625e-06,
|
| 5546 |
+
"loss": 1.1263,
|
| 5547 |
+
"step": 790
|
| 5548 |
+
},
|
| 5549 |
+
{
|
| 5550 |
+
"epoch": 1.0025396825396826,
|
| 5551 |
+
"grad_norm": 1.0390625,
|
| 5552 |
+
"learning_rate": 6.149199810941588e-06,
|
| 5553 |
+
"loss": 1.1561,
|
| 5554 |
+
"step": 791
|
| 5555 |
+
},
|
| 5556 |
+
{
|
| 5557 |
+
"epoch": 1.0038095238095237,
|
| 5558 |
+
"grad_norm": 0.90234375,
|
| 5559 |
+
"learning_rate": 6.146840869964141e-06,
|
| 5560 |
+
"loss": 1.0985,
|
| 5561 |
+
"step": 792
|
| 5562 |
+
},
|
| 5563 |
+
{
|
| 5564 |
+
"epoch": 1.005079365079365,
|
| 5565 |
+
"grad_norm": 0.97265625,
|
| 5566 |
+
"learning_rate": 6.144479117101314e-06,
|
| 5567 |
+
"loss": 1.1212,
|
| 5568 |
+
"step": 793
|
| 5569 |
+
},
|
| 5570 |
+
{
|
| 5571 |
+
"epoch": 1.0063492063492063,
|
| 5572 |
+
"grad_norm": 0.90234375,
|
| 5573 |
+
"learning_rate": 6.142114554862129e-06,
|
| 5574 |
+
"loss": 1.0649,
|
| 5575 |
+
"step": 794
|
| 5576 |
+
},
|
| 5577 |
+
{
|
| 5578 |
+
"epoch": 1.0076190476190476,
|
| 5579 |
+
"grad_norm": 0.9453125,
|
| 5580 |
+
"learning_rate": 6.139747185758589e-06,
|
| 5581 |
+
"loss": 1.0868,
|
| 5582 |
+
"step": 795
|
| 5583 |
+
},
|
| 5584 |
+
{
|
| 5585 |
+
"epoch": 1.008888888888889,
|
| 5586 |
+
"grad_norm": 0.88671875,
|
| 5587 |
+
"learning_rate": 6.137377012305684e-06,
|
| 5588 |
+
"loss": 1.105,
|
| 5589 |
+
"step": 796
|
| 5590 |
+
},
|
| 5591 |
+
{
|
| 5592 |
+
"epoch": 1.0101587301587303,
|
| 5593 |
+
"grad_norm": 0.9140625,
|
| 5594 |
+
"learning_rate": 6.135004037021378e-06,
|
| 5595 |
+
"loss": 1.0531,
|
| 5596 |
+
"step": 797
|
| 5597 |
+
},
|
| 5598 |
+
{
|
| 5599 |
+
"epoch": 1.0114285714285713,
|
| 5600 |
+
"grad_norm": 0.94921875,
|
| 5601 |
+
"learning_rate": 6.1326282624266155e-06,
|
| 5602 |
+
"loss": 1.1147,
|
| 5603 |
+
"step": 798
|
| 5604 |
+
},
|
| 5605 |
+
{
|
| 5606 |
+
"epoch": 1.0126984126984127,
|
| 5607 |
+
"grad_norm": 0.9453125,
|
| 5608 |
+
"learning_rate": 6.130249691045313e-06,
|
| 5609 |
+
"loss": 1.1736,
|
| 5610 |
+
"step": 799
|
| 5611 |
+
},
|
| 5612 |
+
{
|
| 5613 |
+
"epoch": 1.013968253968254,
|
| 5614 |
+
"grad_norm": 0.91015625,
|
| 5615 |
+
"learning_rate": 6.127868325404357e-06,
|
| 5616 |
+
"loss": 1.1453,
|
| 5617 |
+
"step": 800
|
| 5618 |
+
},
|
| 5619 |
+
{
|
| 5620 |
+
"epoch": 1.0152380952380953,
|
| 5621 |
+
"grad_norm": 0.90234375,
|
| 5622 |
+
"learning_rate": 6.1254841680336046e-06,
|
| 5623 |
+
"loss": 1.1401,
|
| 5624 |
+
"step": 801
|
| 5625 |
+
},
|
| 5626 |
+
{
|
| 5627 |
+
"epoch": 1.0165079365079366,
|
| 5628 |
+
"grad_norm": 0.90625,
|
| 5629 |
+
"learning_rate": 6.123097221465878e-06,
|
| 5630 |
+
"loss": 1.1181,
|
| 5631 |
+
"step": 802
|
| 5632 |
+
},
|
| 5633 |
+
{
|
| 5634 |
+
"epoch": 1.0177777777777777,
|
| 5635 |
+
"grad_norm": 0.93359375,
|
| 5636 |
+
"learning_rate": 6.120707488236962e-06,
|
| 5637 |
+
"loss": 1.1782,
|
| 5638 |
+
"step": 803
|
| 5639 |
+
},
|
| 5640 |
+
{
|
| 5641 |
+
"epoch": 1.019047619047619,
|
| 5642 |
+
"grad_norm": 0.99609375,
|
| 5643 |
+
"learning_rate": 6.118314970885604e-06,
|
| 5644 |
+
"loss": 1.1901,
|
| 5645 |
+
"step": 804
|
| 5646 |
+
},
|
| 5647 |
+
{
|
| 5648 |
+
"epoch": 1.0203174603174603,
|
| 5649 |
+
"grad_norm": 0.90234375,
|
| 5650 |
+
"learning_rate": 6.115919671953506e-06,
|
| 5651 |
+
"loss": 1.1247,
|
| 5652 |
+
"step": 805
|
| 5653 |
+
},
|
| 5654 |
+
{
|
| 5655 |
+
"epoch": 1.0215873015873016,
|
| 5656 |
+
"grad_norm": 0.92578125,
|
| 5657 |
+
"learning_rate": 6.113521593985326e-06,
|
| 5658 |
+
"loss": 1.2278,
|
| 5659 |
+
"step": 806
|
| 5660 |
+
},
|
| 5661 |
+
{
|
| 5662 |
+
"epoch": 1.022857142857143,
|
| 5663 |
+
"grad_norm": 0.88671875,
|
| 5664 |
+
"learning_rate": 6.111120739528676e-06,
|
| 5665 |
+
"loss": 1.1343,
|
| 5666 |
+
"step": 807
|
| 5667 |
+
},
|
| 5668 |
+
{
|
| 5669 |
+
"epoch": 1.0241269841269842,
|
| 5670 |
+
"grad_norm": 0.9921875,
|
| 5671 |
+
"learning_rate": 6.108717111134116e-06,
|
| 5672 |
+
"loss": 1.2013,
|
| 5673 |
+
"step": 808
|
| 5674 |
+
},
|
| 5675 |
+
{
|
| 5676 |
+
"epoch": 1.0253968253968253,
|
| 5677 |
+
"grad_norm": 0.94140625,
|
| 5678 |
+
"learning_rate": 6.106310711355153e-06,
|
| 5679 |
+
"loss": 1.1276,
|
| 5680 |
+
"step": 809
|
| 5681 |
+
},
|
| 5682 |
+
{
|
| 5683 |
+
"epoch": 1.0266666666666666,
|
| 5684 |
+
"grad_norm": 1.0078125,
|
| 5685 |
+
"learning_rate": 6.10390154274824e-06,
|
| 5686 |
+
"loss": 1.1422,
|
| 5687 |
+
"step": 810
|
| 5688 |
+
},
|
| 5689 |
+
{
|
| 5690 |
+
"epoch": 1.027936507936508,
|
| 5691 |
+
"grad_norm": 0.9609375,
|
| 5692 |
+
"learning_rate": 6.101489607872768e-06,
|
| 5693 |
+
"loss": 1.0979,
|
| 5694 |
+
"step": 811
|
| 5695 |
+
},
|
| 5696 |
+
{
|
| 5697 |
+
"epoch": 1.0292063492063492,
|
| 5698 |
+
"grad_norm": 0.91015625,
|
| 5699 |
+
"learning_rate": 6.099074909291069e-06,
|
| 5700 |
+
"loss": 1.0752,
|
| 5701 |
+
"step": 812
|
| 5702 |
+
},
|
| 5703 |
+
{
|
| 5704 |
+
"epoch": 1.0304761904761905,
|
| 5705 |
+
"grad_norm": 0.95703125,
|
| 5706 |
+
"learning_rate": 6.096657449568412e-06,
|
| 5707 |
+
"loss": 1.1214,
|
| 5708 |
+
"step": 813
|
| 5709 |
+
},
|
| 5710 |
+
{
|
| 5711 |
+
"epoch": 1.0317460317460316,
|
| 5712 |
+
"grad_norm": 0.9140625,
|
| 5713 |
+
"learning_rate": 6.094237231272998e-06,
|
| 5714 |
+
"loss": 1.1449,
|
| 5715 |
+
"step": 814
|
| 5716 |
+
},
|
| 5717 |
+
{
|
| 5718 |
+
"epoch": 1.033015873015873,
|
| 5719 |
+
"grad_norm": 0.97265625,
|
| 5720 |
+
"learning_rate": 6.091814256975957e-06,
|
| 5721 |
+
"loss": 1.153,
|
| 5722 |
+
"step": 815
|
| 5723 |
+
},
|
| 5724 |
+
{
|
| 5725 |
+
"epoch": 1.0342857142857143,
|
| 5726 |
+
"grad_norm": 0.99609375,
|
| 5727 |
+
"learning_rate": 6.089388529251351e-06,
|
| 5728 |
+
"loss": 1.159,
|
| 5729 |
+
"step": 816
|
| 5730 |
+
},
|
| 5731 |
+
{
|
| 5732 |
+
"epoch": 1.0355555555555556,
|
| 5733 |
+
"grad_norm": 0.9296875,
|
| 5734 |
+
"learning_rate": 6.086960050676163e-06,
|
| 5735 |
+
"loss": 1.1684,
|
| 5736 |
+
"step": 817
|
| 5737 |
+
},
|
| 5738 |
+
{
|
| 5739 |
+
"epoch": 1.0368253968253969,
|
| 5740 |
+
"grad_norm": 0.9140625,
|
| 5741 |
+
"learning_rate": 6.084528823830299e-06,
|
| 5742 |
+
"loss": 1.1217,
|
| 5743 |
+
"step": 818
|
| 5744 |
+
},
|
| 5745 |
+
{
|
| 5746 |
+
"epoch": 1.0380952380952382,
|
| 5747 |
+
"grad_norm": 0.984375,
|
| 5748 |
+
"learning_rate": 6.082094851296589e-06,
|
| 5749 |
+
"loss": 1.1292,
|
| 5750 |
+
"step": 819
|
| 5751 |
+
},
|
| 5752 |
+
{
|
| 5753 |
+
"epoch": 1.0393650793650793,
|
| 5754 |
+
"grad_norm": 0.97265625,
|
| 5755 |
+
"learning_rate": 6.079658135660774e-06,
|
| 5756 |
+
"loss": 1.1013,
|
| 5757 |
+
"step": 820
|
| 5758 |
+
},
|
| 5759 |
+
{
|
| 5760 |
+
"epoch": 1.0406349206349206,
|
| 5761 |
+
"grad_norm": 0.98046875,
|
| 5762 |
+
"learning_rate": 6.077218679511512e-06,
|
| 5763 |
+
"loss": 1.1311,
|
| 5764 |
+
"step": 821
|
| 5765 |
+
},
|
| 5766 |
+
{
|
| 5767 |
+
"epoch": 1.041904761904762,
|
| 5768 |
+
"grad_norm": 1.015625,
|
| 5769 |
+
"learning_rate": 6.074776485440372e-06,
|
| 5770 |
+
"loss": 1.197,
|
| 5771 |
+
"step": 822
|
| 5772 |
+
},
|
| 5773 |
+
{
|
| 5774 |
+
"epoch": 1.0431746031746032,
|
| 5775 |
+
"grad_norm": 0.94140625,
|
| 5776 |
+
"learning_rate": 6.072331556041833e-06,
|
| 5777 |
+
"loss": 1.054,
|
| 5778 |
+
"step": 823
|
| 5779 |
+
},
|
| 5780 |
+
{
|
| 5781 |
+
"epoch": 1.0444444444444445,
|
| 5782 |
+
"grad_norm": 1.0390625,
|
| 5783 |
+
"learning_rate": 6.0698838939132786e-06,
|
| 5784 |
+
"loss": 1.1083,
|
| 5785 |
+
"step": 824
|
| 5786 |
+
},
|
| 5787 |
+
{
|
| 5788 |
+
"epoch": 1.0457142857142858,
|
| 5789 |
+
"grad_norm": 0.98046875,
|
| 5790 |
+
"learning_rate": 6.0674335016549945e-06,
|
| 5791 |
+
"loss": 1.1574,
|
| 5792 |
+
"step": 825
|
| 5793 |
+
},
|
| 5794 |
+
{
|
| 5795 |
+
"epoch": 1.046984126984127,
|
| 5796 |
+
"grad_norm": 0.91796875,
|
| 5797 |
+
"learning_rate": 6.064980381870168e-06,
|
| 5798 |
+
"loss": 1.1341,
|
| 5799 |
+
"step": 826
|
| 5800 |
+
},
|
| 5801 |
+
{
|
| 5802 |
+
"epoch": 1.0482539682539682,
|
| 5803 |
+
"grad_norm": 0.91796875,
|
| 5804 |
+
"learning_rate": 6.062524537164885e-06,
|
| 5805 |
+
"loss": 1.1482,
|
| 5806 |
+
"step": 827
|
| 5807 |
+
},
|
| 5808 |
+
{
|
| 5809 |
+
"epoch": 1.0495238095238095,
|
| 5810 |
+
"grad_norm": 0.9375,
|
| 5811 |
+
"learning_rate": 6.060065970148123e-06,
|
| 5812 |
+
"loss": 1.0957,
|
| 5813 |
+
"step": 828
|
| 5814 |
+
},
|
| 5815 |
+
{
|
| 5816 |
+
"epoch": 1.0507936507936508,
|
| 5817 |
+
"grad_norm": 1.0078125,
|
| 5818 |
+
"learning_rate": 6.057604683431756e-06,
|
| 5819 |
+
"loss": 1.1756,
|
| 5820 |
+
"step": 829
|
| 5821 |
+
},
|
| 5822 |
+
{
|
| 5823 |
+
"epoch": 1.0520634920634921,
|
| 5824 |
+
"grad_norm": 1.0234375,
|
| 5825 |
+
"learning_rate": 6.055140679630543e-06,
|
| 5826 |
+
"loss": 1.199,
|
| 5827 |
+
"step": 830
|
| 5828 |
+
},
|
| 5829 |
+
{
|
| 5830 |
+
"epoch": 1.0533333333333332,
|
| 5831 |
+
"grad_norm": 0.9140625,
|
| 5832 |
+
"learning_rate": 6.052673961362132e-06,
|
| 5833 |
+
"loss": 1.1214,
|
| 5834 |
+
"step": 831
|
| 5835 |
+
},
|
| 5836 |
+
{
|
| 5837 |
+
"epoch": 1.0546031746031745,
|
| 5838 |
+
"grad_norm": 0.94921875,
|
| 5839 |
+
"learning_rate": 6.050204531247056e-06,
|
| 5840 |
+
"loss": 1.1826,
|
| 5841 |
+
"step": 832
|
| 5842 |
+
},
|
| 5843 |
+
{
|
| 5844 |
+
"epoch": 1.0558730158730159,
|
| 5845 |
+
"grad_norm": 1.0234375,
|
| 5846 |
+
"learning_rate": 6.047732391908725e-06,
|
| 5847 |
+
"loss": 1.0719,
|
| 5848 |
+
"step": 833
|
| 5849 |
+
},
|
| 5850 |
+
{
|
| 5851 |
+
"epoch": 1.0571428571428572,
|
| 5852 |
+
"grad_norm": 1.046875,
|
| 5853 |
+
"learning_rate": 6.04525754597343e-06,
|
| 5854 |
+
"loss": 1.1604,
|
| 5855 |
+
"step": 834
|
| 5856 |
+
},
|
| 5857 |
+
{
|
| 5858 |
+
"epoch": 1.0584126984126985,
|
| 5859 |
+
"grad_norm": 1.0078125,
|
| 5860 |
+
"learning_rate": 6.042779996070335e-06,
|
| 5861 |
+
"loss": 1.1072,
|
| 5862 |
+
"step": 835
|
| 5863 |
+
},
|
| 5864 |
+
{
|
| 5865 |
+
"epoch": 1.0596825396825398,
|
| 5866 |
+
"grad_norm": 1.015625,
|
| 5867 |
+
"learning_rate": 6.040299744831482e-06,
|
| 5868 |
+
"loss": 1.1305,
|
| 5869 |
+
"step": 836
|
| 5870 |
+
},
|
| 5871 |
+
{
|
| 5872 |
+
"epoch": 1.0609523809523809,
|
| 5873 |
+
"grad_norm": 0.97265625,
|
| 5874 |
+
"learning_rate": 6.0378167948917755e-06,
|
| 5875 |
+
"loss": 1.0534,
|
| 5876 |
+
"step": 837
|
| 5877 |
+
},
|
| 5878 |
+
{
|
| 5879 |
+
"epoch": 1.0622222222222222,
|
| 5880 |
+
"grad_norm": 0.96875,
|
| 5881 |
+
"learning_rate": 6.035331148888992e-06,
|
| 5882 |
+
"loss": 1.1919,
|
| 5883 |
+
"step": 838
|
| 5884 |
+
},
|
| 5885 |
+
{
|
| 5886 |
+
"epoch": 1.0634920634920635,
|
| 5887 |
+
"grad_norm": 0.99609375,
|
| 5888 |
+
"learning_rate": 6.032842809463771e-06,
|
| 5889 |
+
"loss": 1.1266,
|
| 5890 |
+
"step": 839
|
| 5891 |
+
},
|
| 5892 |
+
{
|
| 5893 |
+
"epoch": 1.0647619047619048,
|
| 5894 |
+
"grad_norm": 1.0,
|
| 5895 |
+
"learning_rate": 6.030351779259612e-06,
|
| 5896 |
+
"loss": 1.1957,
|
| 5897 |
+
"step": 840
|
| 5898 |
+
},
|
| 5899 |
+
{
|
| 5900 |
+
"epoch": 1.066031746031746,
|
| 5901 |
+
"grad_norm": 0.9609375,
|
| 5902 |
+
"learning_rate": 6.027858060922874e-06,
|
| 5903 |
+
"loss": 1.1445,
|
| 5904 |
+
"step": 841
|
| 5905 |
+
},
|
| 5906 |
+
{
|
| 5907 |
+
"epoch": 1.0673015873015874,
|
| 5908 |
+
"grad_norm": 0.9453125,
|
| 5909 |
+
"learning_rate": 6.025361657102773e-06,
|
| 5910 |
+
"loss": 1.0895,
|
| 5911 |
+
"step": 842
|
| 5912 |
+
},
|
| 5913 |
+
{
|
| 5914 |
+
"epoch": 1.0685714285714285,
|
| 5915 |
+
"grad_norm": 1.015625,
|
| 5916 |
+
"learning_rate": 6.022862570451376e-06,
|
| 5917 |
+
"loss": 1.2063,
|
| 5918 |
+
"step": 843
|
| 5919 |
+
},
|
| 5920 |
+
{
|
| 5921 |
+
"epoch": 1.0698412698412698,
|
| 5922 |
+
"grad_norm": 0.9609375,
|
| 5923 |
+
"learning_rate": 6.020360803623601e-06,
|
| 5924 |
+
"loss": 1.0889,
|
| 5925 |
+
"step": 844
|
| 5926 |
+
},
|
| 5927 |
+
{
|
| 5928 |
+
"epoch": 1.0711111111111111,
|
| 5929 |
+
"grad_norm": 0.9609375,
|
| 5930 |
+
"learning_rate": 6.017856359277215e-06,
|
| 5931 |
+
"loss": 1.138,
|
| 5932 |
+
"step": 845
|
| 5933 |
+
},
|
| 5934 |
+
{
|
| 5935 |
+
"epoch": 1.0723809523809524,
|
| 5936 |
+
"grad_norm": 1.0234375,
|
| 5937 |
+
"learning_rate": 6.015349240072824e-06,
|
| 5938 |
+
"loss": 1.1788,
|
| 5939 |
+
"step": 846
|
| 5940 |
+
},
|
| 5941 |
+
{
|
| 5942 |
+
"epoch": 1.0736507936507937,
|
| 5943 |
+
"grad_norm": 0.90234375,
|
| 5944 |
+
"learning_rate": 6.012839448673882e-06,
|
| 5945 |
+
"loss": 1.1004,
|
| 5946 |
+
"step": 847
|
| 5947 |
+
},
|
| 5948 |
+
{
|
| 5949 |
+
"epoch": 1.0749206349206348,
|
| 5950 |
+
"grad_norm": 1.0,
|
| 5951 |
+
"learning_rate": 6.010326987746679e-06,
|
| 5952 |
+
"loss": 1.1197,
|
| 5953 |
+
"step": 848
|
| 5954 |
+
},
|
| 5955 |
+
{
|
| 5956 |
+
"epoch": 1.0761904761904761,
|
| 5957 |
+
"grad_norm": 0.9609375,
|
| 5958 |
+
"learning_rate": 6.007811859960339e-06,
|
| 5959 |
+
"loss": 1.0873,
|
| 5960 |
+
"step": 849
|
| 5961 |
+
},
|
| 5962 |
+
{
|
| 5963 |
+
"epoch": 1.0774603174603175,
|
| 5964 |
+
"grad_norm": 1.0078125,
|
| 5965 |
+
"learning_rate": 6.005294067986824e-06,
|
| 5966 |
+
"loss": 1.1637,
|
| 5967 |
+
"step": 850
|
| 5968 |
+
},
|
| 5969 |
+
{
|
| 5970 |
+
"epoch": 1.0787301587301588,
|
| 5971 |
+
"grad_norm": 0.890625,
|
| 5972 |
+
"learning_rate": 6.002773614500922e-06,
|
| 5973 |
+
"loss": 1.1315,
|
| 5974 |
+
"step": 851
|
| 5975 |
+
},
|
| 5976 |
+
{
|
| 5977 |
+
"epoch": 1.08,
|
| 5978 |
+
"grad_norm": 0.96875,
|
| 5979 |
+
"learning_rate": 6.000250502180251e-06,
|
| 5980 |
+
"loss": 1.1133,
|
| 5981 |
+
"step": 852
|
| 5982 |
+
},
|
| 5983 |
+
{
|
| 5984 |
+
"epoch": 1.0812698412698412,
|
| 5985 |
+
"grad_norm": 0.9375,
|
| 5986 |
+
"learning_rate": 5.9977247337052515e-06,
|
| 5987 |
+
"loss": 1.1943,
|
| 5988 |
+
"step": 853
|
| 5989 |
+
},
|
| 5990 |
+
{
|
| 5991 |
+
"epoch": 1.0825396825396825,
|
| 5992 |
+
"grad_norm": 0.984375,
|
| 5993 |
+
"learning_rate": 5.995196311759189e-06,
|
| 5994 |
+
"loss": 1.133,
|
| 5995 |
+
"step": 854
|
| 5996 |
+
},
|
| 5997 |
+
{
|
| 5998 |
+
"epoch": 1.0838095238095238,
|
| 5999 |
+
"grad_norm": 0.9765625,
|
| 6000 |
+
"learning_rate": 5.992665239028144e-06,
|
| 6001 |
+
"loss": 1.1914,
|
| 6002 |
+
"step": 855
|
| 6003 |
+
},
|
| 6004 |
+
{
|
| 6005 |
+
"epoch": 1.085079365079365,
|
| 6006 |
+
"grad_norm": 0.9375,
|
| 6007 |
+
"learning_rate": 5.990131518201016e-06,
|
| 6008 |
+
"loss": 1.1349,
|
| 6009 |
+
"step": 856
|
| 6010 |
+
},
|
| 6011 |
+
{
|
| 6012 |
+
"epoch": 1.0863492063492064,
|
| 6013 |
+
"grad_norm": 0.9609375,
|
| 6014 |
+
"learning_rate": 5.987595151969516e-06,
|
| 6015 |
+
"loss": 1.1804,
|
| 6016 |
+
"step": 857
|
| 6017 |
+
},
|
| 6018 |
+
{
|
| 6019 |
+
"epoch": 1.0876190476190477,
|
| 6020 |
+
"grad_norm": 0.98046875,
|
| 6021 |
+
"learning_rate": 5.985056143028167e-06,
|
| 6022 |
+
"loss": 1.1553,
|
| 6023 |
+
"step": 858
|
| 6024 |
+
},
|
| 6025 |
+
{
|
| 6026 |
+
"epoch": 1.0888888888888888,
|
| 6027 |
+
"grad_norm": 0.94140625,
|
| 6028 |
+
"learning_rate": 5.982514494074297e-06,
|
| 6029 |
+
"loss": 1.0719,
|
| 6030 |
+
"step": 859
|
| 6031 |
+
},
|
| 6032 |
+
{
|
| 6033 |
+
"epoch": 1.09015873015873,
|
| 6034 |
+
"grad_norm": 1.0078125,
|
| 6035 |
+
"learning_rate": 5.979970207808042e-06,
|
| 6036 |
+
"loss": 1.1532,
|
| 6037 |
+
"step": 860
|
| 6038 |
+
},
|
| 6039 |
+
{
|
| 6040 |
+
"epoch": 1.0914285714285714,
|
| 6041 |
+
"grad_norm": 0.8984375,
|
| 6042 |
+
"learning_rate": 5.9774232869323365e-06,
|
| 6043 |
+
"loss": 1.0941,
|
| 6044 |
+
"step": 861
|
| 6045 |
+
},
|
| 6046 |
+
{
|
| 6047 |
+
"epoch": 1.0926984126984127,
|
| 6048 |
+
"grad_norm": 0.921875,
|
| 6049 |
+
"learning_rate": 5.974873734152916e-06,
|
| 6050 |
+
"loss": 1.1314,
|
| 6051 |
+
"step": 862
|
| 6052 |
+
},
|
| 6053 |
+
{
|
| 6054 |
+
"epoch": 1.093968253968254,
|
| 6055 |
+
"grad_norm": 0.97265625,
|
| 6056 |
+
"learning_rate": 5.972321552178312e-06,
|
| 6057 |
+
"loss": 1.165,
|
| 6058 |
+
"step": 863
|
| 6059 |
+
},
|
| 6060 |
+
{
|
| 6061 |
+
"epoch": 1.0952380952380953,
|
| 6062 |
+
"grad_norm": 0.95703125,
|
| 6063 |
+
"learning_rate": 5.969766743719847e-06,
|
| 6064 |
+
"loss": 1.1825,
|
| 6065 |
+
"step": 864
|
| 6066 |
+
},
|
| 6067 |
+
{
|
| 6068 |
+
"epoch": 1.0965079365079364,
|
| 6069 |
+
"grad_norm": 1.0234375,
|
| 6070 |
+
"learning_rate": 5.967209311491636e-06,
|
| 6071 |
+
"loss": 1.1202,
|
| 6072 |
+
"step": 865
|
| 6073 |
+
},
|
| 6074 |
+
{
|
| 6075 |
+
"epoch": 1.0977777777777777,
|
| 6076 |
+
"grad_norm": 1.0078125,
|
| 6077 |
+
"learning_rate": 5.964649258210583e-06,
|
| 6078 |
+
"loss": 1.139,
|
| 6079 |
+
"step": 866
|
| 6080 |
+
},
|
| 6081 |
+
{
|
| 6082 |
+
"epoch": 1.099047619047619,
|
| 6083 |
+
"grad_norm": 0.8828125,
|
| 6084 |
+
"learning_rate": 5.962086586596369e-06,
|
| 6085 |
+
"loss": 1.0856,
|
| 6086 |
+
"step": 867
|
| 6087 |
+
},
|
| 6088 |
+
{
|
| 6089 |
+
"epoch": 1.1003174603174604,
|
| 6090 |
+
"grad_norm": 0.9609375,
|
| 6091 |
+
"learning_rate": 5.959521299371465e-06,
|
| 6092 |
+
"loss": 1.1494,
|
| 6093 |
+
"step": 868
|
| 6094 |
+
},
|
| 6095 |
+
{
|
| 6096 |
+
"epoch": 1.1015873015873017,
|
| 6097 |
+
"grad_norm": 0.91796875,
|
| 6098 |
+
"learning_rate": 5.956953399261118e-06,
|
| 6099 |
+
"loss": 1.0633,
|
| 6100 |
+
"step": 869
|
| 6101 |
+
},
|
| 6102 |
+
{
|
| 6103 |
+
"epoch": 1.1028571428571428,
|
| 6104 |
+
"grad_norm": 0.90234375,
|
| 6105 |
+
"learning_rate": 5.9543828889933484e-06,
|
| 6106 |
+
"loss": 1.1601,
|
| 6107 |
+
"step": 870
|
| 6108 |
+
},
|
| 6109 |
+
{
|
| 6110 |
+
"epoch": 1.104126984126984,
|
| 6111 |
+
"grad_norm": 0.97265625,
|
| 6112 |
+
"learning_rate": 5.951809771298954e-06,
|
| 6113 |
+
"loss": 1.1452,
|
| 6114 |
+
"step": 871
|
| 6115 |
+
},
|
| 6116 |
+
{
|
| 6117 |
+
"epoch": 1.1053968253968254,
|
| 6118 |
+
"grad_norm": 0.921875,
|
| 6119 |
+
"learning_rate": 5.949234048911497e-06,
|
| 6120 |
+
"loss": 1.0889,
|
| 6121 |
+
"step": 872
|
| 6122 |
+
},
|
| 6123 |
+
{
|
| 6124 |
+
"epoch": 1.1066666666666667,
|
| 6125 |
+
"grad_norm": 1.0078125,
|
| 6126 |
+
"learning_rate": 5.94665572456731e-06,
|
| 6127 |
+
"loss": 1.1215,
|
| 6128 |
+
"step": 873
|
| 6129 |
+
},
|
| 6130 |
+
{
|
| 6131 |
+
"epoch": 1.107936507936508,
|
| 6132 |
+
"grad_norm": 1.0,
|
| 6133 |
+
"learning_rate": 5.94407480100549e-06,
|
| 6134 |
+
"loss": 1.1259,
|
| 6135 |
+
"step": 874
|
| 6136 |
+
},
|
| 6137 |
+
{
|
| 6138 |
+
"epoch": 1.1092063492063493,
|
| 6139 |
+
"grad_norm": 0.94140625,
|
| 6140 |
+
"learning_rate": 5.941491280967898e-06,
|
| 6141 |
+
"loss": 1.0588,
|
| 6142 |
+
"step": 875
|
| 6143 |
+
},
|
| 6144 |
+
{
|
| 6145 |
+
"epoch": 1.1104761904761904,
|
| 6146 |
+
"grad_norm": 0.9921875,
|
| 6147 |
+
"learning_rate": 5.938905167199145e-06,
|
| 6148 |
+
"loss": 1.1346,
|
| 6149 |
+
"step": 876
|
| 6150 |
+
},
|
| 6151 |
+
{
|
| 6152 |
+
"epoch": 1.1117460317460317,
|
| 6153 |
+
"grad_norm": 0.9765625,
|
| 6154 |
+
"learning_rate": 5.936316462446605e-06,
|
| 6155 |
+
"loss": 1.0855,
|
| 6156 |
+
"step": 877
|
| 6157 |
+
},
|
| 6158 |
+
{
|
| 6159 |
+
"epoch": 1.113015873015873,
|
| 6160 |
+
"grad_norm": 0.93359375,
|
| 6161 |
+
"learning_rate": 5.933725169460404e-06,
|
| 6162 |
+
"loss": 1.1648,
|
| 6163 |
+
"step": 878
|
| 6164 |
+
},
|
| 6165 |
+
{
|
| 6166 |
+
"epoch": 1.1142857142857143,
|
| 6167 |
+
"grad_norm": 0.9609375,
|
| 6168 |
+
"learning_rate": 5.931131290993411e-06,
|
| 6169 |
+
"loss": 1.1728,
|
| 6170 |
+
"step": 879
|
| 6171 |
+
},
|
| 6172 |
+
{
|
| 6173 |
+
"epoch": 1.1155555555555556,
|
| 6174 |
+
"grad_norm": 0.9765625,
|
| 6175 |
+
"learning_rate": 5.928534829801252e-06,
|
| 6176 |
+
"loss": 1.082,
|
| 6177 |
+
"step": 880
|
| 6178 |
+
},
|
| 6179 |
+
{
|
| 6180 |
+
"epoch": 1.116825396825397,
|
| 6181 |
+
"grad_norm": 0.9765625,
|
| 6182 |
+
"learning_rate": 5.9259357886422866e-06,
|
| 6183 |
+
"loss": 1.1417,
|
| 6184 |
+
"step": 881
|
| 6185 |
+
},
|
| 6186 |
+
{
|
| 6187 |
+
"epoch": 1.118095238095238,
|
| 6188 |
+
"grad_norm": 0.94921875,
|
| 6189 |
+
"learning_rate": 5.923334170277623e-06,
|
| 6190 |
+
"loss": 1.1807,
|
| 6191 |
+
"step": 882
|
| 6192 |
+
},
|
| 6193 |
+
{
|
| 6194 |
+
"epoch": 1.1193650793650793,
|
| 6195 |
+
"grad_norm": 0.8984375,
|
| 6196 |
+
"learning_rate": 5.9207299774711025e-06,
|
| 6197 |
+
"loss": 1.1927,
|
| 6198 |
+
"step": 883
|
| 6199 |
+
},
|
| 6200 |
+
{
|
| 6201 |
+
"epoch": 1.1206349206349207,
|
| 6202 |
+
"grad_norm": 0.9609375,
|
| 6203 |
+
"learning_rate": 5.918123212989304e-06,
|
| 6204 |
+
"loss": 1.1026,
|
| 6205 |
+
"step": 884
|
| 6206 |
+
},
|
| 6207 |
+
{
|
| 6208 |
+
"epoch": 1.121904761904762,
|
| 6209 |
+
"grad_norm": 0.921875,
|
| 6210 |
+
"learning_rate": 5.915513879601538e-06,
|
| 6211 |
+
"loss": 1.1803,
|
| 6212 |
+
"step": 885
|
| 6213 |
+
},
|
| 6214 |
+
{
|
| 6215 |
+
"epoch": 1.1231746031746033,
|
| 6216 |
+
"grad_norm": 0.94140625,
|
| 6217 |
+
"learning_rate": 5.912901980079843e-06,
|
| 6218 |
+
"loss": 1.1667,
|
| 6219 |
+
"step": 886
|
| 6220 |
+
},
|
| 6221 |
+
{
|
| 6222 |
+
"epoch": 1.1244444444444444,
|
| 6223 |
+
"grad_norm": 0.93359375,
|
| 6224 |
+
"learning_rate": 5.910287517198983e-06,
|
| 6225 |
+
"loss": 1.1347,
|
| 6226 |
+
"step": 887
|
| 6227 |
+
},
|
| 6228 |
+
{
|
| 6229 |
+
"epoch": 1.1257142857142857,
|
| 6230 |
+
"grad_norm": 0.91796875,
|
| 6231 |
+
"learning_rate": 5.9076704937364475e-06,
|
| 6232 |
+
"loss": 1.1578,
|
| 6233 |
+
"step": 888
|
| 6234 |
+
},
|
| 6235 |
+
{
|
| 6236 |
+
"epoch": 1.126984126984127,
|
| 6237 |
+
"grad_norm": 0.96875,
|
| 6238 |
+
"learning_rate": 5.9050509124724456e-06,
|
| 6239 |
+
"loss": 1.1774,
|
| 6240 |
+
"step": 889
|
| 6241 |
+
},
|
| 6242 |
+
{
|
| 6243 |
+
"epoch": 1.1282539682539683,
|
| 6244 |
+
"grad_norm": 0.93359375,
|
| 6245 |
+
"learning_rate": 5.902428776189903e-06,
|
| 6246 |
+
"loss": 1.1567,
|
| 6247 |
+
"step": 890
|
| 6248 |
+
},
|
| 6249 |
+
{
|
| 6250 |
+
"epoch": 1.1295238095238096,
|
| 6251 |
+
"grad_norm": 0.94140625,
|
| 6252 |
+
"learning_rate": 5.899804087674461e-06,
|
| 6253 |
+
"loss": 1.1014,
|
| 6254 |
+
"step": 891
|
| 6255 |
+
},
|
| 6256 |
+
{
|
| 6257 |
+
"epoch": 1.1307936507936507,
|
| 6258 |
+
"grad_norm": 0.94921875,
|
| 6259 |
+
"learning_rate": 5.897176849714469e-06,
|
| 6260 |
+
"loss": 1.1083,
|
| 6261 |
+
"step": 892
|
| 6262 |
+
},
|
| 6263 |
+
{
|
| 6264 |
+
"epoch": 1.132063492063492,
|
| 6265 |
+
"grad_norm": 0.9296875,
|
| 6266 |
+
"learning_rate": 5.8945470651009875e-06,
|
| 6267 |
+
"loss": 1.1097,
|
| 6268 |
+
"step": 893
|
| 6269 |
+
},
|
| 6270 |
+
{
|
| 6271 |
+
"epoch": 1.1333333333333333,
|
| 6272 |
+
"grad_norm": 0.9453125,
|
| 6273 |
+
"learning_rate": 5.891914736627785e-06,
|
| 6274 |
+
"loss": 1.096,
|
| 6275 |
+
"step": 894
|
| 6276 |
+
},
|
| 6277 |
+
{
|
| 6278 |
+
"epoch": 1.1346031746031746,
|
| 6279 |
+
"grad_norm": 0.95703125,
|
| 6280 |
+
"learning_rate": 5.8892798670913275e-06,
|
| 6281 |
+
"loss": 1.1589,
|
| 6282 |
+
"step": 895
|
| 6283 |
+
},
|
| 6284 |
+
{
|
| 6285 |
+
"epoch": 1.135873015873016,
|
| 6286 |
+
"grad_norm": 0.98046875,
|
| 6287 |
+
"learning_rate": 5.886642459290782e-06,
|
| 6288 |
+
"loss": 1.1019,
|
| 6289 |
+
"step": 896
|
| 6290 |
+
},
|
| 6291 |
+
{
|
| 6292 |
+
"epoch": 1.1371428571428572,
|
| 6293 |
+
"grad_norm": 0.921875,
|
| 6294 |
+
"learning_rate": 5.884002516028012e-06,
|
| 6295 |
+
"loss": 1.1257,
|
| 6296 |
+
"step": 897
|
| 6297 |
+
},
|
| 6298 |
+
{
|
| 6299 |
+
"epoch": 1.1384126984126983,
|
| 6300 |
+
"grad_norm": 0.9140625,
|
| 6301 |
+
"learning_rate": 5.881360040107578e-06,
|
| 6302 |
+
"loss": 1.1422,
|
| 6303 |
+
"step": 898
|
| 6304 |
+
},
|
| 6305 |
+
{
|
| 6306 |
+
"epoch": 1.1396825396825396,
|
| 6307 |
+
"grad_norm": 0.90625,
|
| 6308 |
+
"learning_rate": 5.878715034336726e-06,
|
| 6309 |
+
"loss": 1.1903,
|
| 6310 |
+
"step": 899
|
| 6311 |
+
},
|
| 6312 |
+
{
|
| 6313 |
+
"epoch": 1.140952380952381,
|
| 6314 |
+
"grad_norm": 0.92578125,
|
| 6315 |
+
"learning_rate": 5.8760675015253935e-06,
|
| 6316 |
+
"loss": 1.1277,
|
| 6317 |
+
"step": 900
|
| 6318 |
+
},
|
| 6319 |
+
{
|
| 6320 |
+
"epoch": 1.1422222222222222,
|
| 6321 |
+
"grad_norm": 0.9453125,
|
| 6322 |
+
"learning_rate": 5.8734174444862e-06,
|
| 6323 |
+
"loss": 1.1103,
|
| 6324 |
+
"step": 901
|
| 6325 |
+
},
|
| 6326 |
+
{
|
| 6327 |
+
"epoch": 1.1434920634920636,
|
| 6328 |
+
"grad_norm": 0.9296875,
|
| 6329 |
+
"learning_rate": 5.8707648660344454e-06,
|
| 6330 |
+
"loss": 1.1382,
|
| 6331 |
+
"step": 902
|
| 6332 |
+
},
|
| 6333 |
+
{
|
| 6334 |
+
"epoch": 1.1447619047619049,
|
| 6335 |
+
"grad_norm": 0.91015625,
|
| 6336 |
+
"learning_rate": 5.868109768988114e-06,
|
| 6337 |
+
"loss": 1.1512,
|
| 6338 |
+
"step": 903
|
| 6339 |
+
},
|
| 6340 |
+
{
|
| 6341 |
+
"epoch": 1.146031746031746,
|
| 6342 |
+
"grad_norm": 0.9375,
|
| 6343 |
+
"learning_rate": 5.865452156167861e-06,
|
| 6344 |
+
"loss": 1.1509,
|
| 6345 |
+
"step": 904
|
| 6346 |
+
},
|
| 6347 |
+
{
|
| 6348 |
+
"epoch": 1.1473015873015873,
|
| 6349 |
+
"grad_norm": 0.96484375,
|
| 6350 |
+
"learning_rate": 5.862792030397015e-06,
|
| 6351 |
+
"loss": 1.1152,
|
| 6352 |
+
"step": 905
|
| 6353 |
+
},
|
| 6354 |
+
{
|
| 6355 |
+
"epoch": 1.1485714285714286,
|
| 6356 |
+
"grad_norm": 0.91015625,
|
| 6357 |
+
"learning_rate": 5.8601293945015745e-06,
|
| 6358 |
+
"loss": 1.142,
|
| 6359 |
+
"step": 906
|
| 6360 |
+
},
|
| 6361 |
+
{
|
| 6362 |
+
"epoch": 1.1498412698412699,
|
| 6363 |
+
"grad_norm": 0.9765625,
|
| 6364 |
+
"learning_rate": 5.857464251310206e-06,
|
| 6365 |
+
"loss": 1.0839,
|
| 6366 |
+
"step": 907
|
| 6367 |
+
},
|
| 6368 |
+
{
|
| 6369 |
+
"epoch": 1.1511111111111112,
|
| 6370 |
+
"grad_norm": 0.9921875,
|
| 6371 |
+
"learning_rate": 5.854796603654236e-06,
|
| 6372 |
+
"loss": 1.2105,
|
| 6373 |
+
"step": 908
|
| 6374 |
+
},
|
| 6375 |
+
{
|
| 6376 |
+
"epoch": 1.1523809523809523,
|
| 6377 |
+
"grad_norm": 1.0078125,
|
| 6378 |
+
"learning_rate": 5.852126454367657e-06,
|
| 6379 |
+
"loss": 1.1147,
|
| 6380 |
+
"step": 909
|
| 6381 |
+
},
|
| 6382 |
+
{
|
| 6383 |
+
"epoch": 1.1536507936507936,
|
| 6384 |
+
"grad_norm": 0.90234375,
|
| 6385 |
+
"learning_rate": 5.8494538062871145e-06,
|
| 6386 |
+
"loss": 1.1292,
|
| 6387 |
+
"step": 910
|
| 6388 |
+
},
|
| 6389 |
+
{
|
| 6390 |
+
"epoch": 1.154920634920635,
|
| 6391 |
+
"grad_norm": 0.91015625,
|
| 6392 |
+
"learning_rate": 5.8467786622519105e-06,
|
| 6393 |
+
"loss": 1.1353,
|
| 6394 |
+
"step": 911
|
| 6395 |
+
},
|
| 6396 |
+
{
|
| 6397 |
+
"epoch": 1.1561904761904762,
|
| 6398 |
+
"grad_norm": 0.94140625,
|
| 6399 |
+
"learning_rate": 5.844101025103999e-06,
|
| 6400 |
+
"loss": 1.0897,
|
| 6401 |
+
"step": 912
|
| 6402 |
+
},
|
| 6403 |
+
{
|
| 6404 |
+
"epoch": 1.1574603174603175,
|
| 6405 |
+
"grad_norm": 0.97265625,
|
| 6406 |
+
"learning_rate": 5.841420897687981e-06,
|
| 6407 |
+
"loss": 1.1639,
|
| 6408 |
+
"step": 913
|
| 6409 |
+
},
|
| 6410 |
+
{
|
| 6411 |
+
"epoch": 1.1587301587301586,
|
| 6412 |
+
"grad_norm": 0.94921875,
|
| 6413 |
+
"learning_rate": 5.838738282851105e-06,
|
| 6414 |
+
"loss": 1.1634,
|
| 6415 |
+
"step": 914
|
| 6416 |
+
},
|
| 6417 |
+
{
|
| 6418 |
+
"epoch": 1.16,
|
| 6419 |
+
"grad_norm": 0.953125,
|
| 6420 |
+
"learning_rate": 5.836053183443262e-06,
|
| 6421 |
+
"loss": 1.1569,
|
| 6422 |
+
"step": 915
|
| 6423 |
+
},
|
| 6424 |
+
{
|
| 6425 |
+
"epoch": 1.1612698412698412,
|
| 6426 |
+
"grad_norm": 1.0390625,
|
| 6427 |
+
"learning_rate": 5.833365602316978e-06,
|
| 6428 |
+
"loss": 1.2137,
|
| 6429 |
+
"step": 916
|
| 6430 |
+
},
|
| 6431 |
+
{
|
| 6432 |
+
"epoch": 1.1625396825396825,
|
| 6433 |
+
"grad_norm": 0.8671875,
|
| 6434 |
+
"learning_rate": 5.830675542327421e-06,
|
| 6435 |
+
"loss": 1.111,
|
| 6436 |
+
"step": 917
|
| 6437 |
+
},
|
| 6438 |
+
{
|
| 6439 |
+
"epoch": 1.1638095238095238,
|
| 6440 |
+
"grad_norm": 0.90234375,
|
| 6441 |
+
"learning_rate": 5.827983006332391e-06,
|
| 6442 |
+
"loss": 1.0904,
|
| 6443 |
+
"step": 918
|
| 6444 |
+
},
|
| 6445 |
+
{
|
| 6446 |
+
"epoch": 1.1650793650793652,
|
| 6447 |
+
"grad_norm": 0.95703125,
|
| 6448 |
+
"learning_rate": 5.825287997192318e-06,
|
| 6449 |
+
"loss": 1.1414,
|
| 6450 |
+
"step": 919
|
| 6451 |
+
},
|
| 6452 |
+
{
|
| 6453 |
+
"epoch": 1.1663492063492065,
|
| 6454 |
+
"grad_norm": 0.8671875,
|
| 6455 |
+
"learning_rate": 5.822590517770259e-06,
|
| 6456 |
+
"loss": 1.1094,
|
| 6457 |
+
"step": 920
|
| 6458 |
+
},
|
| 6459 |
+
{
|
| 6460 |
+
"epoch": 1.1676190476190476,
|
| 6461 |
+
"grad_norm": 0.92578125,
|
| 6462 |
+
"learning_rate": 5.819890570931895e-06,
|
| 6463 |
+
"loss": 1.128,
|
| 6464 |
+
"step": 921
|
| 6465 |
+
},
|
| 6466 |
+
{
|
| 6467 |
+
"epoch": 1.1688888888888889,
|
| 6468 |
+
"grad_norm": 1.0078125,
|
| 6469 |
+
"learning_rate": 5.817188159545529e-06,
|
| 6470 |
+
"loss": 1.1724,
|
| 6471 |
+
"step": 922
|
| 6472 |
+
},
|
| 6473 |
+
{
|
| 6474 |
+
"epoch": 1.1701587301587302,
|
| 6475 |
+
"grad_norm": 1.0234375,
|
| 6476 |
+
"learning_rate": 5.814483286482081e-06,
|
| 6477 |
+
"loss": 1.0582,
|
| 6478 |
+
"step": 923
|
| 6479 |
+
},
|
| 6480 |
+
{
|
| 6481 |
+
"epoch": 1.1714285714285715,
|
| 6482 |
+
"grad_norm": 0.91796875,
|
| 6483 |
+
"learning_rate": 5.811775954615088e-06,
|
| 6484 |
+
"loss": 1.0837,
|
| 6485 |
+
"step": 924
|
| 6486 |
+
},
|
| 6487 |
+
{
|
| 6488 |
+
"epoch": 1.1726984126984128,
|
| 6489 |
+
"grad_norm": 0.89453125,
|
| 6490 |
+
"learning_rate": 5.809066166820699e-06,
|
| 6491 |
+
"loss": 1.0938,
|
| 6492 |
+
"step": 925
|
| 6493 |
+
},
|
| 6494 |
+
{
|
| 6495 |
+
"epoch": 1.1739682539682539,
|
| 6496 |
+
"grad_norm": 0.94140625,
|
| 6497 |
+
"learning_rate": 5.806353925977671e-06,
|
| 6498 |
+
"loss": 1.1441,
|
| 6499 |
+
"step": 926
|
| 6500 |
+
},
|
| 6501 |
+
{
|
| 6502 |
+
"epoch": 1.1752380952380952,
|
| 6503 |
+
"grad_norm": 1.03125,
|
| 6504 |
+
"learning_rate": 5.803639234967367e-06,
|
| 6505 |
+
"loss": 1.1908,
|
| 6506 |
+
"step": 927
|
| 6507 |
+
},
|
| 6508 |
+
{
|
| 6509 |
+
"epoch": 1.1765079365079365,
|
| 6510 |
+
"grad_norm": 0.9921875,
|
| 6511 |
+
"learning_rate": 5.800922096673753e-06,
|
| 6512 |
+
"loss": 1.1109,
|
| 6513 |
+
"step": 928
|
| 6514 |
+
},
|
| 6515 |
+
{
|
| 6516 |
+
"epoch": 1.1777777777777778,
|
| 6517 |
+
"grad_norm": 0.88671875,
|
| 6518 |
+
"learning_rate": 5.798202513983397e-06,
|
| 6519 |
+
"loss": 1.1131,
|
| 6520 |
+
"step": 929
|
| 6521 |
+
},
|
| 6522 |
+
{
|
| 6523 |
+
"epoch": 1.1790476190476191,
|
| 6524 |
+
"grad_norm": 0.91796875,
|
| 6525 |
+
"learning_rate": 5.79548048978546e-06,
|
| 6526 |
+
"loss": 1.1502,
|
| 6527 |
+
"step": 930
|
| 6528 |
+
},
|
| 6529 |
+
{
|
| 6530 |
+
"epoch": 1.1803174603174602,
|
| 6531 |
+
"grad_norm": 0.96875,
|
| 6532 |
+
"learning_rate": 5.7927560269717e-06,
|
| 6533 |
+
"loss": 1.1954,
|
| 6534 |
+
"step": 931
|
| 6535 |
+
},
|
| 6536 |
+
{
|
| 6537 |
+
"epoch": 1.1815873015873015,
|
| 6538 |
+
"grad_norm": 1.0078125,
|
| 6539 |
+
"learning_rate": 5.790029128436464e-06,
|
| 6540 |
+
"loss": 1.1165,
|
| 6541 |
+
"step": 932
|
| 6542 |
+
},
|
| 6543 |
+
{
|
| 6544 |
+
"epoch": 1.1828571428571428,
|
| 6545 |
+
"grad_norm": 0.94921875,
|
| 6546 |
+
"learning_rate": 5.787299797076686e-06,
|
| 6547 |
+
"loss": 1.1306,
|
| 6548 |
+
"step": 933
|
| 6549 |
+
},
|
| 6550 |
+
{
|
| 6551 |
+
"epoch": 1.1841269841269841,
|
| 6552 |
+
"grad_norm": 0.9375,
|
| 6553 |
+
"learning_rate": 5.784568035791887e-06,
|
| 6554 |
+
"loss": 1.1082,
|
| 6555 |
+
"step": 934
|
| 6556 |
+
},
|
| 6557 |
+
{
|
| 6558 |
+
"epoch": 1.1853968253968254,
|
| 6559 |
+
"grad_norm": 0.9296875,
|
| 6560 |
+
"learning_rate": 5.781833847484169e-06,
|
| 6561 |
+
"loss": 1.1388,
|
| 6562 |
+
"step": 935
|
| 6563 |
+
},
|
| 6564 |
+
{
|
| 6565 |
+
"epoch": 1.1866666666666668,
|
| 6566 |
+
"grad_norm": 0.98046875,
|
| 6567 |
+
"learning_rate": 5.7790972350582095e-06,
|
| 6568 |
+
"loss": 1.1739,
|
| 6569 |
+
"step": 936
|
| 6570 |
+
},
|
| 6571 |
+
{
|
| 6572 |
+
"epoch": 1.1879365079365078,
|
| 6573 |
+
"grad_norm": 0.93359375,
|
| 6574 |
+
"learning_rate": 5.776358201421263e-06,
|
| 6575 |
+
"loss": 1.1154,
|
| 6576 |
+
"step": 937
|
| 6577 |
+
},
|
| 6578 |
+
{
|
| 6579 |
+
"epoch": 1.1892063492063492,
|
| 6580 |
+
"grad_norm": 0.95703125,
|
| 6581 |
+
"learning_rate": 5.773616749483157e-06,
|
| 6582 |
+
"loss": 1.1866,
|
| 6583 |
+
"step": 938
|
| 6584 |
+
},
|
| 6585 |
+
{
|
| 6586 |
+
"epoch": 1.1904761904761905,
|
| 6587 |
+
"grad_norm": 0.9609375,
|
| 6588 |
+
"learning_rate": 5.770872882156288e-06,
|
| 6589 |
+
"loss": 1.2014,
|
| 6590 |
+
"step": 939
|
| 6591 |
+
},
|
| 6592 |
+
{
|
| 6593 |
+
"epoch": 1.1917460317460318,
|
| 6594 |
+
"grad_norm": 0.8984375,
|
| 6595 |
+
"learning_rate": 5.768126602355618e-06,
|
| 6596 |
+
"loss": 1.0875,
|
| 6597 |
+
"step": 940
|
| 6598 |
+
},
|
| 6599 |
+
{
|
| 6600 |
+
"epoch": 1.193015873015873,
|
| 6601 |
+
"grad_norm": 0.9375,
|
| 6602 |
+
"learning_rate": 5.765377912998671e-06,
|
| 6603 |
+
"loss": 1.1382,
|
| 6604 |
+
"step": 941
|
| 6605 |
+
},
|
| 6606 |
+
{
|
| 6607 |
+
"epoch": 1.1942857142857144,
|
| 6608 |
+
"grad_norm": 0.890625,
|
| 6609 |
+
"learning_rate": 5.762626817005533e-06,
|
| 6610 |
+
"loss": 1.0788,
|
| 6611 |
+
"step": 942
|
| 6612 |
+
},
|
| 6613 |
+
{
|
| 6614 |
+
"epoch": 1.1955555555555555,
|
| 6615 |
+
"grad_norm": 1.0625,
|
| 6616 |
+
"learning_rate": 5.759873317298842e-06,
|
| 6617 |
+
"loss": 1.1058,
|
| 6618 |
+
"step": 943
|
| 6619 |
+
},
|
| 6620 |
+
{
|
| 6621 |
+
"epoch": 1.1968253968253968,
|
| 6622 |
+
"grad_norm": 0.96875,
|
| 6623 |
+
"learning_rate": 5.757117416803798e-06,
|
| 6624 |
+
"loss": 1.1358,
|
| 6625 |
+
"step": 944
|
| 6626 |
+
},
|
| 6627 |
+
{
|
| 6628 |
+
"epoch": 1.198095238095238,
|
| 6629 |
+
"grad_norm": 0.9609375,
|
| 6630 |
+
"learning_rate": 5.754359118448141e-06,
|
| 6631 |
+
"loss": 1.1381,
|
| 6632 |
+
"step": 945
|
| 6633 |
+
},
|
| 6634 |
+
{
|
| 6635 |
+
"epoch": 1.1993650793650794,
|
| 6636 |
+
"grad_norm": 0.9140625,
|
| 6637 |
+
"learning_rate": 5.751598425162169e-06,
|
| 6638 |
+
"loss": 1.2095,
|
| 6639 |
+
"step": 946
|
| 6640 |
+
},
|
| 6641 |
+
{
|
| 6642 |
+
"epoch": 1.2006349206349207,
|
| 6643 |
+
"grad_norm": 1.0390625,
|
| 6644 |
+
"learning_rate": 5.7488353398787165e-06,
|
| 6645 |
+
"loss": 1.1923,
|
| 6646 |
+
"step": 947
|
| 6647 |
+
},
|
| 6648 |
+
{
|
| 6649 |
+
"epoch": 1.2019047619047618,
|
| 6650 |
+
"grad_norm": 0.90625,
|
| 6651 |
+
"learning_rate": 5.746069865533161e-06,
|
| 6652 |
+
"loss": 1.0759,
|
| 6653 |
+
"step": 948
|
| 6654 |
+
},
|
| 6655 |
+
{
|
| 6656 |
+
"epoch": 1.2031746031746031,
|
| 6657 |
+
"grad_norm": 0.87109375,
|
| 6658 |
+
"learning_rate": 5.743302005063419e-06,
|
| 6659 |
+
"loss": 1.0627,
|
| 6660 |
+
"step": 949
|
| 6661 |
+
},
|
| 6662 |
+
{
|
| 6663 |
+
"epoch": 1.2044444444444444,
|
| 6664 |
+
"grad_norm": 0.94921875,
|
| 6665 |
+
"learning_rate": 5.740531761409945e-06,
|
| 6666 |
+
"loss": 1.1598,
|
| 6667 |
+
"step": 950
|
| 6668 |
+
},
|
| 6669 |
+
{
|
| 6670 |
+
"epoch": 1.2057142857142857,
|
| 6671 |
+
"grad_norm": 0.8671875,
|
| 6672 |
+
"learning_rate": 5.7377591375157175e-06,
|
| 6673 |
+
"loss": 1.1227,
|
| 6674 |
+
"step": 951
|
| 6675 |
+
},
|
| 6676 |
+
{
|
| 6677 |
+
"epoch": 1.206984126984127,
|
| 6678 |
+
"grad_norm": 0.9765625,
|
| 6679 |
+
"learning_rate": 5.73498413632625e-06,
|
| 6680 |
+
"loss": 1.1589,
|
| 6681 |
+
"step": 952
|
| 6682 |
+
},
|
| 6683 |
+
{
|
| 6684 |
+
"epoch": 1.2082539682539681,
|
| 6685 |
+
"grad_norm": 0.90234375,
|
| 6686 |
+
"learning_rate": 5.73220676078958e-06,
|
| 6687 |
+
"loss": 1.0832,
|
| 6688 |
+
"step": 953
|
| 6689 |
+
},
|
| 6690 |
+
{
|
| 6691 |
+
"epoch": 1.2095238095238094,
|
| 6692 |
+
"grad_norm": 0.94921875,
|
| 6693 |
+
"learning_rate": 5.7294270138562675e-06,
|
| 6694 |
+
"loss": 1.2069,
|
| 6695 |
+
"step": 954
|
| 6696 |
+
},
|
| 6697 |
+
{
|
| 6698 |
+
"epoch": 1.2107936507936508,
|
| 6699 |
+
"grad_norm": 0.96484375,
|
| 6700 |
+
"learning_rate": 5.7266448984793905e-06,
|
| 6701 |
+
"loss": 1.2382,
|
| 6702 |
+
"step": 955
|
| 6703 |
+
},
|
| 6704 |
+
{
|
| 6705 |
+
"epoch": 1.212063492063492,
|
| 6706 |
+
"grad_norm": 0.9375,
|
| 6707 |
+
"learning_rate": 5.723860417614543e-06,
|
| 6708 |
+
"loss": 1.1499,
|
| 6709 |
+
"step": 956
|
| 6710 |
+
},
|
| 6711 |
+
{
|
| 6712 |
+
"epoch": 1.2133333333333334,
|
| 6713 |
+
"grad_norm": 0.92578125,
|
| 6714 |
+
"learning_rate": 5.721073574219835e-06,
|
| 6715 |
+
"loss": 1.1811,
|
| 6716 |
+
"step": 957
|
| 6717 |
+
},
|
| 6718 |
+
{
|
| 6719 |
+
"epoch": 1.2146031746031747,
|
| 6720 |
+
"grad_norm": 0.87109375,
|
| 6721 |
+
"learning_rate": 5.718284371255883e-06,
|
| 6722 |
+
"loss": 1.0129,
|
| 6723 |
+
"step": 958
|
| 6724 |
+
},
|
| 6725 |
+
{
|
| 6726 |
+
"epoch": 1.215873015873016,
|
| 6727 |
+
"grad_norm": 0.95703125,
|
| 6728 |
+
"learning_rate": 5.715492811685813e-06,
|
| 6729 |
+
"loss": 1.1847,
|
| 6730 |
+
"step": 959
|
| 6731 |
+
},
|
| 6732 |
+
{
|
| 6733 |
+
"epoch": 1.217142857142857,
|
| 6734 |
+
"grad_norm": 0.85546875,
|
| 6735 |
+
"learning_rate": 5.712698898475249e-06,
|
| 6736 |
+
"loss": 1.0698,
|
| 6737 |
+
"step": 960
|
| 6738 |
+
},
|
| 6739 |
+
{
|
| 6740 |
+
"epoch": 1.2184126984126984,
|
| 6741 |
+
"grad_norm": 0.9375,
|
| 6742 |
+
"learning_rate": 5.709902634592324e-06,
|
| 6743 |
+
"loss": 1.0597,
|
| 6744 |
+
"step": 961
|
| 6745 |
+
},
|
| 6746 |
+
{
|
| 6747 |
+
"epoch": 1.2196825396825397,
|
| 6748 |
+
"grad_norm": 0.94140625,
|
| 6749 |
+
"learning_rate": 5.707104023007664e-06,
|
| 6750 |
+
"loss": 1.0922,
|
| 6751 |
+
"step": 962
|
| 6752 |
+
},
|
| 6753 |
+
{
|
| 6754 |
+
"epoch": 1.220952380952381,
|
| 6755 |
+
"grad_norm": 0.90625,
|
| 6756 |
+
"learning_rate": 5.7043030666943846e-06,
|
| 6757 |
+
"loss": 1.1159,
|
| 6758 |
+
"step": 963
|
| 6759 |
+
},
|
| 6760 |
+
{
|
| 6761 |
+
"epoch": 1.2222222222222223,
|
| 6762 |
+
"grad_norm": 0.90625,
|
| 6763 |
+
"learning_rate": 5.701499768628099e-06,
|
| 6764 |
+
"loss": 1.1535,
|
| 6765 |
+
"step": 964
|
| 6766 |
+
},
|
| 6767 |
+
{
|
| 6768 |
+
"epoch": 1.2234920634920634,
|
| 6769 |
+
"grad_norm": 0.95703125,
|
| 6770 |
+
"learning_rate": 5.6986941317869065e-06,
|
| 6771 |
+
"loss": 1.0798,
|
| 6772 |
+
"step": 965
|
| 6773 |
+
},
|
| 6774 |
+
{
|
| 6775 |
+
"epoch": 1.2247619047619047,
|
| 6776 |
+
"grad_norm": 1.0625,
|
| 6777 |
+
"learning_rate": 5.695886159151388e-06,
|
| 6778 |
+
"loss": 1.1546,
|
| 6779 |
+
"step": 966
|
| 6780 |
+
},
|
| 6781 |
+
{
|
| 6782 |
+
"epoch": 1.226031746031746,
|
| 6783 |
+
"grad_norm": 0.9140625,
|
| 6784 |
+
"learning_rate": 5.69307585370461e-06,
|
| 6785 |
+
"loss": 1.0339,
|
| 6786 |
+
"step": 967
|
| 6787 |
+
},
|
| 6788 |
+
{
|
| 6789 |
+
"epoch": 1.2273015873015873,
|
| 6790 |
+
"grad_norm": 0.94921875,
|
| 6791 |
+
"learning_rate": 5.690263218432114e-06,
|
| 6792 |
+
"loss": 1.1029,
|
| 6793 |
+
"step": 968
|
| 6794 |
+
},
|
| 6795 |
+
{
|
| 6796 |
+
"epoch": 1.2285714285714286,
|
| 6797 |
+
"grad_norm": 0.95703125,
|
| 6798 |
+
"learning_rate": 5.687448256321919e-06,
|
| 6799 |
+
"loss": 1.0723,
|
| 6800 |
+
"step": 969
|
| 6801 |
+
},
|
| 6802 |
+
{
|
| 6803 |
+
"epoch": 1.2298412698412697,
|
| 6804 |
+
"grad_norm": 0.96875,
|
| 6805 |
+
"learning_rate": 5.684630970364515e-06,
|
| 6806 |
+
"loss": 1.0463,
|
| 6807 |
+
"step": 970
|
| 6808 |
+
},
|
| 6809 |
+
{
|
| 6810 |
+
"epoch": 1.231111111111111,
|
| 6811 |
+
"grad_norm": 1.046875,
|
| 6812 |
+
"learning_rate": 5.6818113635528595e-06,
|
| 6813 |
+
"loss": 1.1948,
|
| 6814 |
+
"step": 971
|
| 6815 |
+
},
|
| 6816 |
+
{
|
| 6817 |
+
"epoch": 1.2323809523809524,
|
| 6818 |
+
"grad_norm": 1.0,
|
| 6819 |
+
"learning_rate": 5.6789894388823764e-06,
|
| 6820 |
+
"loss": 1.1977,
|
| 6821 |
+
"step": 972
|
| 6822 |
+
},
|
| 6823 |
+
{
|
| 6824 |
+
"epoch": 1.2336507936507937,
|
| 6825 |
+
"grad_norm": 0.875,
|
| 6826 |
+
"learning_rate": 5.676165199350954e-06,
|
| 6827 |
+
"loss": 1.0944,
|
| 6828 |
+
"step": 973
|
| 6829 |
+
},
|
| 6830 |
+
{
|
| 6831 |
+
"epoch": 1.234920634920635,
|
| 6832 |
+
"grad_norm": 0.91796875,
|
| 6833 |
+
"learning_rate": 5.6733386479589375e-06,
|
| 6834 |
+
"loss": 1.1806,
|
| 6835 |
+
"step": 974
|
| 6836 |
+
},
|
| 6837 |
+
{
|
| 6838 |
+
"epoch": 1.2361904761904763,
|
| 6839 |
+
"grad_norm": 1.0625,
|
| 6840 |
+
"learning_rate": 5.670509787709128e-06,
|
| 6841 |
+
"loss": 1.169,
|
| 6842 |
+
"step": 975
|
| 6843 |
+
},
|
| 6844 |
+
{
|
| 6845 |
+
"epoch": 1.2374603174603174,
|
| 6846 |
+
"grad_norm": 0.92578125,
|
| 6847 |
+
"learning_rate": 5.667678621606781e-06,
|
| 6848 |
+
"loss": 1.161,
|
| 6849 |
+
"step": 976
|
| 6850 |
+
},
|
| 6851 |
+
{
|
| 6852 |
+
"epoch": 1.2387301587301587,
|
| 6853 |
+
"grad_norm": 0.93359375,
|
| 6854 |
+
"learning_rate": 5.664845152659599e-06,
|
| 6855 |
+
"loss": 1.1933,
|
| 6856 |
+
"step": 977
|
| 6857 |
+
},
|
| 6858 |
+
{
|
| 6859 |
+
"epoch": 1.24,
|
| 6860 |
+
"grad_norm": 0.93359375,
|
| 6861 |
+
"learning_rate": 5.6620093838777345e-06,
|
| 6862 |
+
"loss": 1.1191,
|
| 6863 |
+
"step": 978
|
| 6864 |
+
},
|
| 6865 |
+
{
|
| 6866 |
+
"epoch": 1.2412698412698413,
|
| 6867 |
+
"grad_norm": 0.95703125,
|
| 6868 |
+
"learning_rate": 5.659171318273782e-06,
|
| 6869 |
+
"loss": 1.1189,
|
| 6870 |
+
"step": 979
|
| 6871 |
+
},
|
| 6872 |
+
{
|
| 6873 |
+
"epoch": 1.2425396825396826,
|
| 6874 |
+
"grad_norm": 0.96875,
|
| 6875 |
+
"learning_rate": 5.656330958862774e-06,
|
| 6876 |
+
"loss": 1.0603,
|
| 6877 |
+
"step": 980
|
| 6878 |
+
},
|
| 6879 |
+
{
|
| 6880 |
+
"epoch": 1.243809523809524,
|
| 6881 |
+
"grad_norm": 1.0234375,
|
| 6882 |
+
"learning_rate": 5.653488308662181e-06,
|
| 6883 |
+
"loss": 1.1747,
|
| 6884 |
+
"step": 981
|
| 6885 |
+
},
|
| 6886 |
+
{
|
| 6887 |
+
"epoch": 1.245079365079365,
|
| 6888 |
+
"grad_norm": 0.984375,
|
| 6889 |
+
"learning_rate": 5.650643370691909e-06,
|
| 6890 |
+
"loss": 1.1333,
|
| 6891 |
+
"step": 982
|
| 6892 |
+
},
|
| 6893 |
+
{
|
| 6894 |
+
"epoch": 1.2463492063492063,
|
| 6895 |
+
"grad_norm": 0.9765625,
|
| 6896 |
+
"learning_rate": 5.647796147974291e-06,
|
| 6897 |
+
"loss": 1.166,
|
| 6898 |
+
"step": 983
|
| 6899 |
+
},
|
| 6900 |
+
{
|
| 6901 |
+
"epoch": 1.2476190476190476,
|
| 6902 |
+
"grad_norm": 0.95703125,
|
| 6903 |
+
"learning_rate": 5.644946643534091e-06,
|
| 6904 |
+
"loss": 1.1551,
|
| 6905 |
+
"step": 984
|
| 6906 |
+
},
|
| 6907 |
+
{
|
| 6908 |
+
"epoch": 1.248888888888889,
|
| 6909 |
+
"grad_norm": 0.9765625,
|
| 6910 |
+
"learning_rate": 5.642094860398495e-06,
|
| 6911 |
+
"loss": 1.1008,
|
| 6912 |
+
"step": 985
|
| 6913 |
+
},
|
| 6914 |
+
{
|
| 6915 |
+
"epoch": 1.2501587301587302,
|
| 6916 |
+
"grad_norm": 0.9296875,
|
| 6917 |
+
"learning_rate": 5.639240801597108e-06,
|
| 6918 |
+
"loss": 1.0894,
|
| 6919 |
+
"step": 986
|
| 6920 |
+
},
|
| 6921 |
+
{
|
| 6922 |
+
"epoch": 1.2514285714285713,
|
| 6923 |
+
"grad_norm": 0.95703125,
|
| 6924 |
+
"learning_rate": 5.636384470161957e-06,
|
| 6925 |
+
"loss": 1.1374,
|
| 6926 |
+
"step": 987
|
| 6927 |
+
},
|
| 6928 |
+
{
|
| 6929 |
+
"epoch": 1.2526984126984126,
|
| 6930 |
+
"grad_norm": 0.9375,
|
| 6931 |
+
"learning_rate": 5.633525869127478e-06,
|
| 6932 |
+
"loss": 1.1105,
|
| 6933 |
+
"step": 988
|
| 6934 |
+
},
|
| 6935 |
+
{
|
| 6936 |
+
"epoch": 1.253968253968254,
|
| 6937 |
+
"grad_norm": 0.9296875,
|
| 6938 |
+
"learning_rate": 5.630665001530522e-06,
|
| 6939 |
+
"loss": 1.0807,
|
| 6940 |
+
"step": 989
|
| 6941 |
+
},
|
| 6942 |
+
{
|
| 6943 |
+
"epoch": 1.2552380952380953,
|
| 6944 |
+
"grad_norm": 0.90625,
|
| 6945 |
+
"learning_rate": 5.627801870410348e-06,
|
| 6946 |
+
"loss": 1.1284,
|
| 6947 |
+
"step": 990
|
| 6948 |
+
},
|
| 6949 |
+
{
|
| 6950 |
+
"epoch": 1.2565079365079366,
|
| 6951 |
+
"grad_norm": 0.94140625,
|
| 6952 |
+
"learning_rate": 5.624936478808617e-06,
|
| 6953 |
+
"loss": 1.1448,
|
| 6954 |
+
"step": 991
|
| 6955 |
+
},
|
| 6956 |
+
{
|
| 6957 |
+
"epoch": 1.2577777777777777,
|
| 6958 |
+
"grad_norm": 0.87890625,
|
| 6959 |
+
"learning_rate": 5.622068829769393e-06,
|
| 6960 |
+
"loss": 1.0997,
|
| 6961 |
+
"step": 992
|
| 6962 |
+
},
|
| 6963 |
+
{
|
| 6964 |
+
"epoch": 1.259047619047619,
|
| 6965 |
+
"grad_norm": 0.9921875,
|
| 6966 |
+
"learning_rate": 5.61919892633914e-06,
|
| 6967 |
+
"loss": 1.1034,
|
| 6968 |
+
"step": 993
|
| 6969 |
+
},
|
| 6970 |
+
{
|
| 6971 |
+
"epoch": 1.2603174603174603,
|
| 6972 |
+
"grad_norm": 0.95703125,
|
| 6973 |
+
"learning_rate": 5.616326771566714e-06,
|
| 6974 |
+
"loss": 1.1347,
|
| 6975 |
+
"step": 994
|
| 6976 |
+
},
|
| 6977 |
+
{
|
| 6978 |
+
"epoch": 1.2615873015873016,
|
| 6979 |
+
"grad_norm": 0.96484375,
|
| 6980 |
+
"learning_rate": 5.613452368503362e-06,
|
| 6981 |
+
"loss": 1.0823,
|
| 6982 |
+
"step": 995
|
| 6983 |
+
},
|
| 6984 |
+
{
|
| 6985 |
+
"epoch": 1.262857142857143,
|
| 6986 |
+
"grad_norm": 0.87890625,
|
| 6987 |
+
"learning_rate": 5.610575720202723e-06,
|
| 6988 |
+
"loss": 1.0923,
|
| 6989 |
+
"step": 996
|
| 6990 |
+
},
|
| 6991 |
+
{
|
| 6992 |
+
"epoch": 1.2641269841269842,
|
| 6993 |
+
"grad_norm": 0.90234375,
|
| 6994 |
+
"learning_rate": 5.60769682972082e-06,
|
| 6995 |
+
"loss": 1.0534,
|
| 6996 |
+
"step": 997
|
| 6997 |
+
},
|
| 6998 |
+
{
|
| 6999 |
+
"epoch": 1.2653968253968255,
|
| 7000 |
+
"grad_norm": 0.94921875,
|
| 7001 |
+
"learning_rate": 5.604815700116055e-06,
|
| 7002 |
+
"loss": 1.1278,
|
| 7003 |
+
"step": 998
|
| 7004 |
+
},
|
| 7005 |
+
{
|
| 7006 |
+
"epoch": 1.2666666666666666,
|
| 7007 |
+
"grad_norm": 0.8828125,
|
| 7008 |
+
"learning_rate": 5.601932334449214e-06,
|
| 7009 |
+
"loss": 1.037,
|
| 7010 |
+
"step": 999
|
| 7011 |
+
},
|
| 7012 |
+
{
|
| 7013 |
+
"epoch": 1.267936507936508,
|
| 7014 |
+
"grad_norm": 1.0390625,
|
| 7015 |
+
"learning_rate": 5.599046735783456e-06,
|
| 7016 |
+
"loss": 1.087,
|
| 7017 |
+
"step": 1000
|
| 7018 |
+
},
|
| 7019 |
+
{
|
| 7020 |
+
"epoch": 1.2692063492063492,
|
| 7021 |
+
"grad_norm": 0.984375,
|
| 7022 |
+
"learning_rate": 5.596158907184309e-06,
|
| 7023 |
+
"loss": 1.0948,
|
| 7024 |
+
"step": 1001
|
| 7025 |
+
},
|
| 7026 |
+
{
|
| 7027 |
+
"epoch": 1.2704761904761905,
|
| 7028 |
+
"grad_norm": 0.95703125,
|
| 7029 |
+
"learning_rate": 5.593268851719676e-06,
|
| 7030 |
+
"loss": 1.1269,
|
| 7031 |
+
"step": 1002
|
| 7032 |
+
},
|
| 7033 |
+
{
|
| 7034 |
+
"epoch": 1.2717460317460318,
|
| 7035 |
+
"grad_norm": 0.953125,
|
| 7036 |
+
"learning_rate": 5.590376572459821e-06,
|
| 7037 |
+
"loss": 1.1055,
|
| 7038 |
+
"step": 1003
|
| 7039 |
+
},
|
| 7040 |
+
{
|
| 7041 |
+
"epoch": 1.273015873015873,
|
| 7042 |
+
"grad_norm": 0.93359375,
|
| 7043 |
+
"learning_rate": 5.587482072477371e-06,
|
| 7044 |
+
"loss": 1.0811,
|
| 7045 |
+
"step": 1004
|
| 7046 |
+
},
|
| 7047 |
+
{
|
| 7048 |
+
"epoch": 1.2742857142857142,
|
| 7049 |
+
"grad_norm": 0.95703125,
|
| 7050 |
+
"learning_rate": 5.584585354847317e-06,
|
| 7051 |
+
"loss": 1.0866,
|
| 7052 |
+
"step": 1005
|
| 7053 |
+
},
|
| 7054 |
+
{
|
| 7055 |
+
"epoch": 1.2755555555555556,
|
| 7056 |
+
"grad_norm": 0.9375,
|
| 7057 |
+
"learning_rate": 5.581686422646998e-06,
|
| 7058 |
+
"loss": 1.0588,
|
| 7059 |
+
"step": 1006
|
| 7060 |
+
},
|
| 7061 |
+
{
|
| 7062 |
+
"epoch": 1.2768253968253969,
|
| 7063 |
+
"grad_norm": 1.0078125,
|
| 7064 |
+
"learning_rate": 5.578785278956114e-06,
|
| 7065 |
+
"loss": 1.0968,
|
| 7066 |
+
"step": 1007
|
| 7067 |
+
},
|
| 7068 |
+
{
|
| 7069 |
+
"epoch": 1.2780952380952382,
|
| 7070 |
+
"grad_norm": 0.90625,
|
| 7071 |
+
"learning_rate": 5.575881926856708e-06,
|
| 7072 |
+
"loss": 1.0973,
|
| 7073 |
+
"step": 1008
|
| 7074 |
+
},
|
| 7075 |
+
{
|
| 7076 |
+
"epoch": 1.2793650793650793,
|
| 7077 |
+
"grad_norm": 0.95703125,
|
| 7078 |
+
"learning_rate": 5.572976369433172e-06,
|
| 7079 |
+
"loss": 1.1661,
|
| 7080 |
+
"step": 1009
|
| 7081 |
+
},
|
| 7082 |
+
{
|
| 7083 |
+
"epoch": 1.2806349206349206,
|
| 7084 |
+
"grad_norm": 0.953125,
|
| 7085 |
+
"learning_rate": 5.5700686097722384e-06,
|
| 7086 |
+
"loss": 1.1793,
|
| 7087 |
+
"step": 1010
|
| 7088 |
+
},
|
| 7089 |
+
{
|
| 7090 |
+
"epoch": 1.2819047619047619,
|
| 7091 |
+
"grad_norm": 0.97265625,
|
| 7092 |
+
"learning_rate": 5.5671586509629845e-06,
|
| 7093 |
+
"loss": 1.1664,
|
| 7094 |
+
"step": 1011
|
| 7095 |
+
},
|
| 7096 |
+
{
|
| 7097 |
+
"epoch": 1.2831746031746032,
|
| 7098 |
+
"grad_norm": 1.0078125,
|
| 7099 |
+
"learning_rate": 5.564246496096818e-06,
|
| 7100 |
+
"loss": 1.1553,
|
| 7101 |
+
"step": 1012
|
| 7102 |
+
},
|
| 7103 |
+
{
|
| 7104 |
+
"epoch": 1.2844444444444445,
|
| 7105 |
+
"grad_norm": 0.97265625,
|
| 7106 |
+
"learning_rate": 5.5613321482674834e-06,
|
| 7107 |
+
"loss": 1.0554,
|
| 7108 |
+
"step": 1013
|
| 7109 |
+
},
|
| 7110 |
+
{
|
| 7111 |
+
"epoch": 1.2857142857142856,
|
| 7112 |
+
"grad_norm": 0.94921875,
|
| 7113 |
+
"learning_rate": 5.558415610571052e-06,
|
| 7114 |
+
"loss": 1.1176,
|
| 7115 |
+
"step": 1014
|
| 7116 |
+
},
|
| 7117 |
+
{
|
| 7118 |
+
"epoch": 1.2869841269841271,
|
| 7119 |
+
"grad_norm": 0.9296875,
|
| 7120 |
+
"learning_rate": 5.555496886105924e-06,
|
| 7121 |
+
"loss": 1.1676,
|
| 7122 |
+
"step": 1015
|
| 7123 |
+
},
|
| 7124 |
+
{
|
| 7125 |
+
"epoch": 1.2882539682539682,
|
| 7126 |
+
"grad_norm": 0.95703125,
|
| 7127 |
+
"learning_rate": 5.552575977972822e-06,
|
| 7128 |
+
"loss": 1.1585,
|
| 7129 |
+
"step": 1016
|
| 7130 |
+
},
|
| 7131 |
+
{
|
| 7132 |
+
"epoch": 1.2895238095238095,
|
| 7133 |
+
"grad_norm": 0.97265625,
|
| 7134 |
+
"learning_rate": 5.549652889274788e-06,
|
| 7135 |
+
"loss": 1.127,
|
| 7136 |
+
"step": 1017
|
| 7137 |
+
},
|
| 7138 |
+
{
|
| 7139 |
+
"epoch": 1.2907936507936508,
|
| 7140 |
+
"grad_norm": 1.109375,
|
| 7141 |
+
"learning_rate": 5.546727623117179e-06,
|
| 7142 |
+
"loss": 1.2112,
|
| 7143 |
+
"step": 1018
|
| 7144 |
+
},
|
| 7145 |
+
{
|
| 7146 |
+
"epoch": 1.2920634920634921,
|
| 7147 |
+
"grad_norm": 1.015625,
|
| 7148 |
+
"learning_rate": 5.543800182607669e-06,
|
| 7149 |
+
"loss": 1.1219,
|
| 7150 |
+
"step": 1019
|
| 7151 |
+
},
|
| 7152 |
+
{
|
| 7153 |
+
"epoch": 1.2933333333333334,
|
| 7154 |
+
"grad_norm": 0.94140625,
|
| 7155 |
+
"learning_rate": 5.54087057085624e-06,
|
| 7156 |
+
"loss": 1.2062,
|
| 7157 |
+
"step": 1020
|
| 7158 |
+
},
|
| 7159 |
+
{
|
| 7160 |
+
"epoch": 1.2946031746031745,
|
| 7161 |
+
"grad_norm": 1.109375,
|
| 7162 |
+
"learning_rate": 5.537938790975179e-06,
|
| 7163 |
+
"loss": 1.1131,
|
| 7164 |
+
"step": 1021
|
| 7165 |
+
},
|
| 7166 |
+
{
|
| 7167 |
+
"epoch": 1.2958730158730158,
|
| 7168 |
+
"grad_norm": 0.9765625,
|
| 7169 |
+
"learning_rate": 5.535004846079079e-06,
|
| 7170 |
+
"loss": 1.0721,
|
| 7171 |
+
"step": 1022
|
| 7172 |
+
},
|
| 7173 |
+
{
|
| 7174 |
+
"epoch": 1.2971428571428572,
|
| 7175 |
+
"grad_norm": 0.98828125,
|
| 7176 |
+
"learning_rate": 5.532068739284832e-06,
|
| 7177 |
+
"loss": 1.1498,
|
| 7178 |
+
"step": 1023
|
| 7179 |
+
},
|
| 7180 |
+
{
|
| 7181 |
+
"epoch": 1.2984126984126985,
|
| 7182 |
+
"grad_norm": 0.9375,
|
| 7183 |
+
"learning_rate": 5.5291304737116254e-06,
|
| 7184 |
+
"loss": 1.1301,
|
| 7185 |
+
"step": 1024
|
| 7186 |
+
},
|
| 7187 |
+
{
|
| 7188 |
+
"epoch": 1.2996825396825398,
|
| 7189 |
+
"grad_norm": 1.0234375,
|
| 7190 |
+
"learning_rate": 5.526190052480942e-06,
|
| 7191 |
+
"loss": 1.1593,
|
| 7192 |
+
"step": 1025
|
| 7193 |
+
},
|
| 7194 |
+
{
|
| 7195 |
+
"epoch": 1.3009523809523809,
|
| 7196 |
+
"grad_norm": 0.9921875,
|
| 7197 |
+
"learning_rate": 5.523247478716555e-06,
|
| 7198 |
+
"loss": 1.0929,
|
| 7199 |
+
"step": 1026
|
| 7200 |
+
},
|
| 7201 |
+
{
|
| 7202 |
+
"epoch": 1.3022222222222222,
|
| 7203 |
+
"grad_norm": 0.94140625,
|
| 7204 |
+
"learning_rate": 5.5203027555445205e-06,
|
| 7205 |
+
"loss": 1.1383,
|
| 7206 |
+
"step": 1027
|
| 7207 |
+
},
|
| 7208 |
+
{
|
| 7209 |
+
"epoch": 1.3034920634920635,
|
| 7210 |
+
"grad_norm": 0.9296875,
|
| 7211 |
+
"learning_rate": 5.517355886093182e-06,
|
| 7212 |
+
"loss": 1.1199,
|
| 7213 |
+
"step": 1028
|
| 7214 |
+
},
|
| 7215 |
+
{
|
| 7216 |
+
"epoch": 1.3047619047619048,
|
| 7217 |
+
"grad_norm": 0.95703125,
|
| 7218 |
+
"learning_rate": 5.514406873493163e-06,
|
| 7219 |
+
"loss": 1.0592,
|
| 7220 |
+
"step": 1029
|
| 7221 |
+
},
|
| 7222 |
+
{
|
| 7223 |
+
"epoch": 1.306031746031746,
|
| 7224 |
+
"grad_norm": 0.8359375,
|
| 7225 |
+
"learning_rate": 5.511455720877359e-06,
|
| 7226 |
+
"loss": 1.0237,
|
| 7227 |
+
"step": 1030
|
| 7228 |
+
},
|
| 7229 |
+
{
|
| 7230 |
+
"epoch": 1.3073015873015872,
|
| 7231 |
+
"grad_norm": 0.95703125,
|
| 7232 |
+
"learning_rate": 5.508502431380948e-06,
|
| 7233 |
+
"loss": 1.1415,
|
| 7234 |
+
"step": 1031
|
| 7235 |
+
},
|
| 7236 |
+
{
|
| 7237 |
+
"epoch": 1.3085714285714285,
|
| 7238 |
+
"grad_norm": 0.94921875,
|
| 7239 |
+
"learning_rate": 5.5055470081413684e-06,
|
| 7240 |
+
"loss": 1.1732,
|
| 7241 |
+
"step": 1032
|
| 7242 |
+
},
|
| 7243 |
+
{
|
| 7244 |
+
"epoch": 1.3098412698412698,
|
| 7245 |
+
"grad_norm": 1.015625,
|
| 7246 |
+
"learning_rate": 5.5025894542983315e-06,
|
| 7247 |
+
"loss": 1.1345,
|
| 7248 |
+
"step": 1033
|
| 7249 |
+
},
|
| 7250 |
+
{
|
| 7251 |
+
"epoch": 1.3111111111111111,
|
| 7252 |
+
"grad_norm": 1.0,
|
| 7253 |
+
"learning_rate": 5.499629772993809e-06,
|
| 7254 |
+
"loss": 1.0832,
|
| 7255 |
+
"step": 1034
|
| 7256 |
+
},
|
| 7257 |
+
{
|
| 7258 |
+
"epoch": 1.3123809523809524,
|
| 7259 |
+
"grad_norm": 0.99609375,
|
| 7260 |
+
"learning_rate": 5.496667967372035e-06,
|
| 7261 |
+
"loss": 1.2329,
|
| 7262 |
+
"step": 1035
|
| 7263 |
+
},
|
| 7264 |
+
{
|
| 7265 |
+
"epoch": 1.3136507936507935,
|
| 7266 |
+
"grad_norm": 1.0078125,
|
| 7267 |
+
"learning_rate": 5.4937040405795e-06,
|
| 7268 |
+
"loss": 1.1604,
|
| 7269 |
+
"step": 1036
|
| 7270 |
+
},
|
| 7271 |
+
{
|
| 7272 |
+
"epoch": 1.314920634920635,
|
| 7273 |
+
"grad_norm": 1.0703125,
|
| 7274 |
+
"learning_rate": 5.490737995764946e-06,
|
| 7275 |
+
"loss": 1.133,
|
| 7276 |
+
"step": 1037
|
| 7277 |
+
},
|
| 7278 |
+
{
|
| 7279 |
+
"epoch": 1.3161904761904761,
|
| 7280 |
+
"grad_norm": 1.09375,
|
| 7281 |
+
"learning_rate": 5.487769836079367e-06,
|
| 7282 |
+
"loss": 1.0956,
|
| 7283 |
+
"step": 1038
|
| 7284 |
+
},
|
| 7285 |
+
{
|
| 7286 |
+
"epoch": 1.3174603174603174,
|
| 7287 |
+
"grad_norm": 0.97265625,
|
| 7288 |
+
"learning_rate": 5.484799564676002e-06,
|
| 7289 |
+
"loss": 1.0835,
|
| 7290 |
+
"step": 1039
|
| 7291 |
+
},
|
| 7292 |
+
{
|
| 7293 |
+
"epoch": 1.3187301587301588,
|
| 7294 |
+
"grad_norm": 0.88671875,
|
| 7295 |
+
"learning_rate": 5.481827184710336e-06,
|
| 7296 |
+
"loss": 1.0909,
|
| 7297 |
+
"step": 1040
|
| 7298 |
+
},
|
| 7299 |
+
{
|
| 7300 |
+
"epoch": 1.32,
|
| 7301 |
+
"grad_norm": 0.9296875,
|
| 7302 |
+
"learning_rate": 5.478852699340092e-06,
|
| 7303 |
+
"loss": 1.0682,
|
| 7304 |
+
"step": 1041
|
| 7305 |
+
},
|
| 7306 |
+
{
|
| 7307 |
+
"epoch": 1.3212698412698414,
|
| 7308 |
+
"grad_norm": 1.0234375,
|
| 7309 |
+
"learning_rate": 5.47587611172523e-06,
|
| 7310 |
+
"loss": 1.1619,
|
| 7311 |
+
"step": 1042
|
| 7312 |
+
},
|
| 7313 |
+
{
|
| 7314 |
+
"epoch": 1.3225396825396825,
|
| 7315 |
+
"grad_norm": 1.078125,
|
| 7316 |
+
"learning_rate": 5.4728974250279445e-06,
|
| 7317 |
+
"loss": 1.1885,
|
| 7318 |
+
"step": 1043
|
| 7319 |
+
},
|
| 7320 |
+
{
|
| 7321 |
+
"epoch": 1.3238095238095238,
|
| 7322 |
+
"grad_norm": 1.0234375,
|
| 7323 |
+
"learning_rate": 5.4699166424126575e-06,
|
| 7324 |
+
"loss": 1.1089,
|
| 7325 |
+
"step": 1044
|
| 7326 |
+
},
|
| 7327 |
+
{
|
| 7328 |
+
"epoch": 1.325079365079365,
|
| 7329 |
+
"grad_norm": 0.94921875,
|
| 7330 |
+
"learning_rate": 5.4669337670460205e-06,
|
| 7331 |
+
"loss": 1.1159,
|
| 7332 |
+
"step": 1045
|
| 7333 |
+
},
|
| 7334 |
+
{
|
| 7335 |
+
"epoch": 1.3263492063492064,
|
| 7336 |
+
"grad_norm": 0.9453125,
|
| 7337 |
+
"learning_rate": 5.463948802096907e-06,
|
| 7338 |
+
"loss": 1.153,
|
| 7339 |
+
"step": 1046
|
| 7340 |
+
},
|
| 7341 |
+
{
|
| 7342 |
+
"epoch": 1.3276190476190477,
|
| 7343 |
+
"grad_norm": 0.99609375,
|
| 7344 |
+
"learning_rate": 5.460961750736408e-06,
|
| 7345 |
+
"loss": 1.0732,
|
| 7346 |
+
"step": 1047
|
| 7347 |
+
},
|
| 7348 |
+
{
|
| 7349 |
+
"epoch": 1.3288888888888888,
|
| 7350 |
+
"grad_norm": 0.921875,
|
| 7351 |
+
"learning_rate": 5.457972616137836e-06,
|
| 7352 |
+
"loss": 1.0756,
|
| 7353 |
+
"step": 1048
|
| 7354 |
+
},
|
| 7355 |
+
{
|
| 7356 |
+
"epoch": 1.33015873015873,
|
| 7357 |
+
"grad_norm": 0.9765625,
|
| 7358 |
+
"learning_rate": 5.454981401476713e-06,
|
| 7359 |
+
"loss": 1.1332,
|
| 7360 |
+
"step": 1049
|
| 7361 |
+
},
|
| 7362 |
+
{
|
| 7363 |
+
"epoch": 1.3314285714285714,
|
| 7364 |
+
"grad_norm": 1.03125,
|
| 7365 |
+
"learning_rate": 5.451988109930771e-06,
|
| 7366 |
+
"loss": 1.1398,
|
| 7367 |
+
"step": 1050
|
| 7368 |
+
},
|
| 7369 |
+
{
|
| 7370 |
+
"epoch": 1.3326984126984127,
|
| 7371 |
+
"grad_norm": 0.91796875,
|
| 7372 |
+
"learning_rate": 5.44899274467995e-06,
|
| 7373 |
+
"loss": 1.1576,
|
| 7374 |
+
"step": 1051
|
| 7375 |
+
},
|
| 7376 |
+
{
|
| 7377 |
+
"epoch": 1.333968253968254,
|
| 7378 |
+
"grad_norm": 0.9296875,
|
| 7379 |
+
"learning_rate": 5.4459953089063925e-06,
|
| 7380 |
+
"loss": 1.0711,
|
| 7381 |
+
"step": 1052
|
| 7382 |
+
},
|
| 7383 |
+
{
|
| 7384 |
+
"epoch": 1.3352380952380951,
|
| 7385 |
+
"grad_norm": 0.9453125,
|
| 7386 |
+
"learning_rate": 5.442995805794438e-06,
|
| 7387 |
+
"loss": 1.1551,
|
| 7388 |
+
"step": 1053
|
| 7389 |
+
},
|
| 7390 |
+
{
|
| 7391 |
+
"epoch": 1.3365079365079366,
|
| 7392 |
+
"grad_norm": 0.9140625,
|
| 7393 |
+
"learning_rate": 5.439994238530627e-06,
|
| 7394 |
+
"loss": 1.0791,
|
| 7395 |
+
"step": 1054
|
| 7396 |
+
},
|
| 7397 |
+
{
|
| 7398 |
+
"epoch": 1.3377777777777777,
|
| 7399 |
+
"grad_norm": 0.98828125,
|
| 7400 |
+
"learning_rate": 5.4369906103036895e-06,
|
| 7401 |
+
"loss": 1.2093,
|
| 7402 |
+
"step": 1055
|
| 7403 |
+
},
|
| 7404 |
+
{
|
| 7405 |
+
"epoch": 1.339047619047619,
|
| 7406 |
+
"grad_norm": 0.92578125,
|
| 7407 |
+
"learning_rate": 5.433984924304546e-06,
|
| 7408 |
+
"loss": 1.1535,
|
| 7409 |
+
"step": 1056
|
| 7410 |
+
},
|
| 7411 |
+
{
|
| 7412 |
+
"epoch": 1.3403174603174604,
|
| 7413 |
+
"grad_norm": 0.9609375,
|
| 7414 |
+
"learning_rate": 5.430977183726302e-06,
|
| 7415 |
+
"loss": 1.113,
|
| 7416 |
+
"step": 1057
|
| 7417 |
+
},
|
| 7418 |
+
{
|
| 7419 |
+
"epoch": 1.3415873015873017,
|
| 7420 |
+
"grad_norm": 0.90625,
|
| 7421 |
+
"learning_rate": 5.427967391764247e-06,
|
| 7422 |
+
"loss": 1.1166,
|
| 7423 |
+
"step": 1058
|
| 7424 |
+
},
|
| 7425 |
+
{
|
| 7426 |
+
"epoch": 1.342857142857143,
|
| 7427 |
+
"grad_norm": 1.015625,
|
| 7428 |
+
"learning_rate": 5.42495555161585e-06,
|
| 7429 |
+
"loss": 1.2128,
|
| 7430 |
+
"step": 1059
|
| 7431 |
+
},
|
| 7432 |
+
{
|
| 7433 |
+
"epoch": 1.344126984126984,
|
| 7434 |
+
"grad_norm": 0.95703125,
|
| 7435 |
+
"learning_rate": 5.421941666480755e-06,
|
| 7436 |
+
"loss": 1.0608,
|
| 7437 |
+
"step": 1060
|
| 7438 |
+
},
|
| 7439 |
+
{
|
| 7440 |
+
"epoch": 1.3453968253968254,
|
| 7441 |
+
"grad_norm": 0.984375,
|
| 7442 |
+
"learning_rate": 5.418925739560777e-06,
|
| 7443 |
+
"loss": 1.1802,
|
| 7444 |
+
"step": 1061
|
| 7445 |
+
},
|
| 7446 |
+
{
|
| 7447 |
+
"epoch": 1.3466666666666667,
|
| 7448 |
+
"grad_norm": 0.8984375,
|
| 7449 |
+
"learning_rate": 5.415907774059904e-06,
|
| 7450 |
+
"loss": 1.0853,
|
| 7451 |
+
"step": 1062
|
| 7452 |
+
},
|
| 7453 |
+
{
|
| 7454 |
+
"epoch": 1.347936507936508,
|
| 7455 |
+
"grad_norm": 0.91796875,
|
| 7456 |
+
"learning_rate": 5.412887773184288e-06,
|
| 7457 |
+
"loss": 1.1111,
|
| 7458 |
+
"step": 1063
|
| 7459 |
+
},
|
| 7460 |
+
{
|
| 7461 |
+
"epoch": 1.3492063492063493,
|
| 7462 |
+
"grad_norm": 0.98828125,
|
| 7463 |
+
"learning_rate": 5.409865740142242e-06,
|
| 7464 |
+
"loss": 1.0664,
|
| 7465 |
+
"step": 1064
|
| 7466 |
+
},
|
| 7467 |
+
{
|
| 7468 |
+
"epoch": 1.3504761904761904,
|
| 7469 |
+
"grad_norm": 1.0078125,
|
| 7470 |
+
"learning_rate": 5.406841678144237e-06,
|
| 7471 |
+
"loss": 1.0606,
|
| 7472 |
+
"step": 1065
|
| 7473 |
+
},
|
| 7474 |
+
{
|
| 7475 |
+
"epoch": 1.3517460317460317,
|
| 7476 |
+
"grad_norm": 0.921875,
|
| 7477 |
+
"learning_rate": 5.403815590402905e-06,
|
| 7478 |
+
"loss": 1.1376,
|
| 7479 |
+
"step": 1066
|
| 7480 |
+
},
|
| 7481 |
+
{
|
| 7482 |
+
"epoch": 1.353015873015873,
|
| 7483 |
+
"grad_norm": 0.96484375,
|
| 7484 |
+
"learning_rate": 5.400787480133023e-06,
|
| 7485 |
+
"loss": 1.1475,
|
| 7486 |
+
"step": 1067
|
| 7487 |
+
},
|
| 7488 |
+
{
|
| 7489 |
+
"epoch": 1.3542857142857143,
|
| 7490 |
+
"grad_norm": 1.0078125,
|
| 7491 |
+
"learning_rate": 5.397757350551521e-06,
|
| 7492 |
+
"loss": 1.021,
|
| 7493 |
+
"step": 1068
|
| 7494 |
+
},
|
| 7495 |
+
{
|
| 7496 |
+
"epoch": 1.3555555555555556,
|
| 7497 |
+
"grad_norm": 1.0,
|
| 7498 |
+
"learning_rate": 5.394725204877474e-06,
|
| 7499 |
+
"loss": 1.141,
|
| 7500 |
+
"step": 1069
|
| 7501 |
+
},
|
| 7502 |
+
{
|
| 7503 |
+
"epoch": 1.3568253968253967,
|
| 7504 |
+
"grad_norm": 1.015625,
|
| 7505 |
+
"learning_rate": 5.391691046332097e-06,
|
| 7506 |
+
"loss": 1.1207,
|
| 7507 |
+
"step": 1070
|
| 7508 |
+
},
|
| 7509 |
+
{
|
| 7510 |
+
"epoch": 1.358095238095238,
|
| 7511 |
+
"grad_norm": 0.99609375,
|
| 7512 |
+
"learning_rate": 5.3886548781387455e-06,
|
| 7513 |
+
"loss": 1.1566,
|
| 7514 |
+
"step": 1071
|
| 7515 |
+
},
|
| 7516 |
+
{
|
| 7517 |
+
"epoch": 1.3593650793650793,
|
| 7518 |
+
"grad_norm": 0.96484375,
|
| 7519 |
+
"learning_rate": 5.385616703522907e-06,
|
| 7520 |
+
"loss": 1.1329,
|
| 7521 |
+
"step": 1072
|
| 7522 |
+
},
|
| 7523 |
+
{
|
| 7524 |
+
"epoch": 1.3606349206349206,
|
| 7525 |
+
"grad_norm": 0.94140625,
|
| 7526 |
+
"learning_rate": 5.382576525712205e-06,
|
| 7527 |
+
"loss": 1.1949,
|
| 7528 |
+
"step": 1073
|
| 7529 |
+
},
|
| 7530 |
+
{
|
| 7531 |
+
"epoch": 1.361904761904762,
|
| 7532 |
+
"grad_norm": 0.96875,
|
| 7533 |
+
"learning_rate": 5.379534347936386e-06,
|
| 7534 |
+
"loss": 1.1444,
|
| 7535 |
+
"step": 1074
|
| 7536 |
+
},
|
| 7537 |
+
{
|
| 7538 |
+
"epoch": 1.363174603174603,
|
| 7539 |
+
"grad_norm": 0.953125,
|
| 7540 |
+
"learning_rate": 5.376490173427324e-06,
|
| 7541 |
+
"loss": 1.1498,
|
| 7542 |
+
"step": 1075
|
| 7543 |
+
},
|
| 7544 |
+
{
|
| 7545 |
+
"epoch": 1.3644444444444446,
|
| 7546 |
+
"grad_norm": 1.0,
|
| 7547 |
+
"learning_rate": 5.373444005419014e-06,
|
| 7548 |
+
"loss": 1.0996,
|
| 7549 |
+
"step": 1076
|
| 7550 |
+
},
|
| 7551 |
+
{
|
| 7552 |
+
"epoch": 1.3657142857142857,
|
| 7553 |
+
"grad_norm": 0.9453125,
|
| 7554 |
+
"learning_rate": 5.37039584714757e-06,
|
| 7555 |
+
"loss": 1.1395,
|
| 7556 |
+
"step": 1077
|
| 7557 |
+
},
|
| 7558 |
+
{
|
| 7559 |
+
"epoch": 1.366984126984127,
|
| 7560 |
+
"grad_norm": 1.046875,
|
| 7561 |
+
"learning_rate": 5.367345701851217e-06,
|
| 7562 |
+
"loss": 1.1675,
|
| 7563 |
+
"step": 1078
|
| 7564 |
+
},
|
| 7565 |
+
{
|
| 7566 |
+
"epoch": 1.3682539682539683,
|
| 7567 |
+
"grad_norm": 0.96484375,
|
| 7568 |
+
"learning_rate": 5.364293572770295e-06,
|
| 7569 |
+
"loss": 1.1776,
|
| 7570 |
+
"step": 1079
|
| 7571 |
+
},
|
| 7572 |
+
{
|
| 7573 |
+
"epoch": 1.3695238095238096,
|
| 7574 |
+
"grad_norm": 0.94140625,
|
| 7575 |
+
"learning_rate": 5.3612394631472475e-06,
|
| 7576 |
+
"loss": 1.1552,
|
| 7577 |
+
"step": 1080
|
| 7578 |
+
},
|
| 7579 |
+
{
|
| 7580 |
+
"epoch": 1.370793650793651,
|
| 7581 |
+
"grad_norm": 0.890625,
|
| 7582 |
+
"learning_rate": 5.358183376226626e-06,
|
| 7583 |
+
"loss": 1.0682,
|
| 7584 |
+
"step": 1081
|
| 7585 |
+
},
|
| 7586 |
+
{
|
| 7587 |
+
"epoch": 1.372063492063492,
|
| 7588 |
+
"grad_norm": 0.98828125,
|
| 7589 |
+
"learning_rate": 5.355125315255079e-06,
|
| 7590 |
+
"loss": 1.117,
|
| 7591 |
+
"step": 1082
|
| 7592 |
+
},
|
| 7593 |
+
{
|
| 7594 |
+
"epoch": 1.3733333333333333,
|
| 7595 |
+
"grad_norm": 0.95703125,
|
| 7596 |
+
"learning_rate": 5.352065283481355e-06,
|
| 7597 |
+
"loss": 1.2372,
|
| 7598 |
+
"step": 1083
|
| 7599 |
+
},
|
| 7600 |
+
{
|
| 7601 |
+
"epoch": 1.3746031746031746,
|
| 7602 |
+
"grad_norm": 0.91796875,
|
| 7603 |
+
"learning_rate": 5.349003284156292e-06,
|
| 7604 |
+
"loss": 1.1573,
|
| 7605 |
+
"step": 1084
|
| 7606 |
+
},
|
| 7607 |
+
{
|
| 7608 |
+
"epoch": 1.375873015873016,
|
| 7609 |
+
"grad_norm": 0.9921875,
|
| 7610 |
+
"learning_rate": 5.3459393205328255e-06,
|
| 7611 |
+
"loss": 1.1628,
|
| 7612 |
+
"step": 1085
|
| 7613 |
+
},
|
| 7614 |
+
{
|
| 7615 |
+
"epoch": 1.3771428571428572,
|
| 7616 |
+
"grad_norm": 1.0078125,
|
| 7617 |
+
"learning_rate": 5.342873395865971e-06,
|
| 7618 |
+
"loss": 1.1204,
|
| 7619 |
+
"step": 1086
|
| 7620 |
+
},
|
| 7621 |
+
{
|
| 7622 |
+
"epoch": 1.3784126984126983,
|
| 7623 |
+
"grad_norm": 1.03125,
|
| 7624 |
+
"learning_rate": 5.3398055134128295e-06,
|
| 7625 |
+
"loss": 1.124,
|
| 7626 |
+
"step": 1087
|
| 7627 |
+
},
|
| 7628 |
+
{
|
| 7629 |
+
"epoch": 1.3796825396825396,
|
| 7630 |
+
"grad_norm": 0.9375,
|
| 7631 |
+
"learning_rate": 5.336735676432583e-06,
|
| 7632 |
+
"loss": 1.0391,
|
| 7633 |
+
"step": 1088
|
| 7634 |
+
},
|
| 7635 |
+
{
|
| 7636 |
+
"epoch": 1.380952380952381,
|
| 7637 |
+
"grad_norm": 0.98046875,
|
| 7638 |
+
"learning_rate": 5.333663888186488e-06,
|
| 7639 |
+
"loss": 1.1251,
|
| 7640 |
+
"step": 1089
|
| 7641 |
+
},
|
| 7642 |
+
{
|
| 7643 |
+
"epoch": 1.3822222222222222,
|
| 7644 |
+
"grad_norm": 0.94140625,
|
| 7645 |
+
"learning_rate": 5.330590151937877e-06,
|
| 7646 |
+
"loss": 1.1557,
|
| 7647 |
+
"step": 1090
|
| 7648 |
+
},
|
| 7649 |
+
{
|
| 7650 |
+
"epoch": 1.3834920634920636,
|
| 7651 |
+
"grad_norm": 0.9921875,
|
| 7652 |
+
"learning_rate": 5.327514470952147e-06,
|
| 7653 |
+
"loss": 1.1114,
|
| 7654 |
+
"step": 1091
|
| 7655 |
+
},
|
| 7656 |
+
{
|
| 7657 |
+
"epoch": 1.3847619047619046,
|
| 7658 |
+
"grad_norm": 0.94140625,
|
| 7659 |
+
"learning_rate": 5.324436848496766e-06,
|
| 7660 |
+
"loss": 1.1198,
|
| 7661 |
+
"step": 1092
|
| 7662 |
+
},
|
| 7663 |
+
{
|
| 7664 |
+
"epoch": 1.3860317460317462,
|
| 7665 |
+
"grad_norm": 0.94140625,
|
| 7666 |
+
"learning_rate": 5.321357287841262e-06,
|
| 7667 |
+
"loss": 1.0892,
|
| 7668 |
+
"step": 1093
|
| 7669 |
+
},
|
| 7670 |
+
{
|
| 7671 |
+
"epoch": 1.3873015873015873,
|
| 7672 |
+
"grad_norm": 0.91015625,
|
| 7673 |
+
"learning_rate": 5.3182757922572226e-06,
|
| 7674 |
+
"loss": 1.1187,
|
| 7675 |
+
"step": 1094
|
| 7676 |
+
},
|
| 7677 |
+
{
|
| 7678 |
+
"epoch": 1.3885714285714286,
|
| 7679 |
+
"grad_norm": 0.90234375,
|
| 7680 |
+
"learning_rate": 5.315192365018292e-06,
|
| 7681 |
+
"loss": 1.0835,
|
| 7682 |
+
"step": 1095
|
| 7683 |
+
},
|
| 7684 |
+
{
|
| 7685 |
+
"epoch": 1.3898412698412699,
|
| 7686 |
+
"grad_norm": 0.9375,
|
| 7687 |
+
"learning_rate": 5.312107009400163e-06,
|
| 7688 |
+
"loss": 1.1224,
|
| 7689 |
+
"step": 1096
|
| 7690 |
+
},
|
| 7691 |
+
{
|
| 7692 |
+
"epoch": 1.3911111111111112,
|
| 7693 |
+
"grad_norm": 1.0078125,
|
| 7694 |
+
"learning_rate": 5.309019728680581e-06,
|
| 7695 |
+
"loss": 1.1558,
|
| 7696 |
+
"step": 1097
|
| 7697 |
+
},
|
| 7698 |
+
{
|
| 7699 |
+
"epoch": 1.3923809523809525,
|
| 7700 |
+
"grad_norm": 1.015625,
|
| 7701 |
+
"learning_rate": 5.3059305261393355e-06,
|
| 7702 |
+
"loss": 1.2182,
|
| 7703 |
+
"step": 1098
|
| 7704 |
+
},
|
| 7705 |
+
{
|
| 7706 |
+
"epoch": 1.3936507936507936,
|
| 7707 |
+
"grad_norm": 0.99609375,
|
| 7708 |
+
"learning_rate": 5.3028394050582555e-06,
|
| 7709 |
+
"loss": 1.1034,
|
| 7710 |
+
"step": 1099
|
| 7711 |
+
},
|
| 7712 |
+
{
|
| 7713 |
+
"epoch": 1.394920634920635,
|
| 7714 |
+
"grad_norm": 1.0,
|
| 7715 |
+
"learning_rate": 5.29974636872121e-06,
|
| 7716 |
+
"loss": 1.1412,
|
| 7717 |
+
"step": 1100
|
| 7718 |
+
},
|
| 7719 |
+
{
|
| 7720 |
+
"epoch": 1.3961904761904762,
|
| 7721 |
+
"grad_norm": 1.0390625,
|
| 7722 |
+
"learning_rate": 5.296651420414104e-06,
|
| 7723 |
+
"loss": 1.1102,
|
| 7724 |
+
"step": 1101
|
| 7725 |
+
},
|
| 7726 |
+
{
|
| 7727 |
+
"epoch": 1.3974603174603175,
|
| 7728 |
+
"grad_norm": 0.96484375,
|
| 7729 |
+
"learning_rate": 5.293554563424871e-06,
|
| 7730 |
+
"loss": 1.1589,
|
| 7731 |
+
"step": 1102
|
| 7732 |
+
},
|
| 7733 |
+
{
|
| 7734 |
+
"epoch": 1.3987301587301588,
|
| 7735 |
+
"grad_norm": 1.03125,
|
| 7736 |
+
"learning_rate": 5.2904558010434745e-06,
|
| 7737 |
+
"loss": 1.1025,
|
| 7738 |
+
"step": 1103
|
| 7739 |
+
},
|
| 7740 |
+
{
|
| 7741 |
+
"epoch": 1.4,
|
| 7742 |
+
"grad_norm": 1.125,
|
| 7743 |
+
"learning_rate": 5.2873551365619e-06,
|
| 7744 |
+
"loss": 1.1318,
|
| 7745 |
+
"step": 1104
|
| 7746 |
+
},
|
| 7747 |
+
{
|
| 7748 |
+
"epoch": 1.4012698412698412,
|
| 7749 |
+
"grad_norm": 1.1171875,
|
| 7750 |
+
"learning_rate": 5.284252573274153e-06,
|
| 7751 |
+
"loss": 1.125,
|
| 7752 |
+
"step": 1105
|
| 7753 |
+
},
|
| 7754 |
+
{
|
| 7755 |
+
"epoch": 1.4025396825396825,
|
| 7756 |
+
"grad_norm": 0.90625,
|
| 7757 |
+
"learning_rate": 5.281148114476263e-06,
|
| 7758 |
+
"loss": 1.0316,
|
| 7759 |
+
"step": 1106
|
| 7760 |
+
},
|
| 7761 |
+
{
|
| 7762 |
+
"epoch": 1.4038095238095238,
|
| 7763 |
+
"grad_norm": 1.03125,
|
| 7764 |
+
"learning_rate": 5.278041763466265e-06,
|
| 7765 |
+
"loss": 1.1106,
|
| 7766 |
+
"step": 1107
|
| 7767 |
+
},
|
| 7768 |
+
{
|
| 7769 |
+
"epoch": 1.4050793650793652,
|
| 7770 |
+
"grad_norm": 1.1171875,
|
| 7771 |
+
"learning_rate": 5.274933523544206e-06,
|
| 7772 |
+
"loss": 1.2083,
|
| 7773 |
+
"step": 1108
|
| 7774 |
+
},
|
| 7775 |
+
{
|
| 7776 |
+
"epoch": 1.4063492063492062,
|
| 7777 |
+
"grad_norm": 1.078125,
|
| 7778 |
+
"learning_rate": 5.2718233980121435e-06,
|
| 7779 |
+
"loss": 1.0879,
|
| 7780 |
+
"step": 1109
|
| 7781 |
+
},
|
| 7782 |
+
{
|
| 7783 |
+
"epoch": 1.4076190476190475,
|
| 7784 |
+
"grad_norm": 1.046875,
|
| 7785 |
+
"learning_rate": 5.268711390174135e-06,
|
| 7786 |
+
"loss": 1.2132,
|
| 7787 |
+
"step": 1110
|
| 7788 |
+
},
|
| 7789 |
+
{
|
| 7790 |
+
"epoch": 1.4088888888888889,
|
| 7791 |
+
"grad_norm": 1.109375,
|
| 7792 |
+
"learning_rate": 5.265597503336238e-06,
|
| 7793 |
+
"loss": 1.1204,
|
| 7794 |
+
"step": 1111
|
| 7795 |
+
},
|
| 7796 |
+
{
|
| 7797 |
+
"epoch": 1.4101587301587302,
|
| 7798 |
+
"grad_norm": 0.9375,
|
| 7799 |
+
"learning_rate": 5.262481740806509e-06,
|
| 7800 |
+
"loss": 1.048,
|
| 7801 |
+
"step": 1112
|
| 7802 |
+
},
|
| 7803 |
+
{
|
| 7804 |
+
"epoch": 1.4114285714285715,
|
| 7805 |
+
"grad_norm": 0.9921875,
|
| 7806 |
+
"learning_rate": 5.259364105894991e-06,
|
| 7807 |
+
"loss": 1.1192,
|
| 7808 |
+
"step": 1113
|
| 7809 |
+
},
|
| 7810 |
+
{
|
| 7811 |
+
"epoch": 1.4126984126984126,
|
| 7812 |
+
"grad_norm": 1.0078125,
|
| 7813 |
+
"learning_rate": 5.2562446019137225e-06,
|
| 7814 |
+
"loss": 1.1135,
|
| 7815 |
+
"step": 1114
|
| 7816 |
+
},
|
| 7817 |
+
{
|
| 7818 |
+
"epoch": 1.413968253968254,
|
| 7819 |
+
"grad_norm": 1.0625,
|
| 7820 |
+
"learning_rate": 5.253123232176724e-06,
|
| 7821 |
+
"loss": 1.1061,
|
| 7822 |
+
"step": 1115
|
| 7823 |
+
},
|
| 7824 |
+
{
|
| 7825 |
+
"epoch": 1.4152380952380952,
|
| 7826 |
+
"grad_norm": 0.96875,
|
| 7827 |
+
"learning_rate": 5.25e-06,
|
| 7828 |
+
"loss": 1.0388,
|
| 7829 |
+
"step": 1116
|
| 7830 |
+
},
|
| 7831 |
+
{
|
| 7832 |
+
"epoch": 1.4165079365079365,
|
| 7833 |
+
"grad_norm": 0.9375,
|
| 7834 |
+
"learning_rate": 5.246874908701532e-06,
|
| 7835 |
+
"loss": 1.0881,
|
| 7836 |
+
"step": 1117
|
| 7837 |
+
},
|
| 7838 |
+
{
|
| 7839 |
+
"epoch": 1.4177777777777778,
|
| 7840 |
+
"grad_norm": 0.8984375,
|
| 7841 |
+
"learning_rate": 5.2437479616012775e-06,
|
| 7842 |
+
"loss": 1.0944,
|
| 7843 |
+
"step": 1118
|
| 7844 |
+
},
|
| 7845 |
+
{
|
| 7846 |
+
"epoch": 1.4190476190476191,
|
| 7847 |
+
"grad_norm": 1.0234375,
|
| 7848 |
+
"learning_rate": 5.240619162021165e-06,
|
| 7849 |
+
"loss": 1.119,
|
| 7850 |
+
"step": 1119
|
| 7851 |
+
},
|
| 7852 |
+
{
|
| 7853 |
+
"epoch": 1.4203174603174604,
|
| 7854 |
+
"grad_norm": 0.99609375,
|
| 7855 |
+
"learning_rate": 5.237488513285092e-06,
|
| 7856 |
+
"loss": 1.1366,
|
| 7857 |
+
"step": 1120
|
| 7858 |
+
},
|
| 7859 |
+
{
|
| 7860 |
+
"epoch": 1.4215873015873015,
|
| 7861 |
+
"grad_norm": 0.94921875,
|
| 7862 |
+
"learning_rate": 5.234356018718919e-06,
|
| 7863 |
+
"loss": 1.0363,
|
| 7864 |
+
"step": 1121
|
| 7865 |
+
},
|
| 7866 |
+
{
|
| 7867 |
+
"epoch": 1.4228571428571428,
|
| 7868 |
+
"grad_norm": 0.9765625,
|
| 7869 |
+
"learning_rate": 5.231221681650469e-06,
|
| 7870 |
+
"loss": 1.1057,
|
| 7871 |
+
"step": 1122
|
| 7872 |
+
},
|
| 7873 |
+
{
|
| 7874 |
+
"epoch": 1.4241269841269841,
|
| 7875 |
+
"grad_norm": 1.015625,
|
| 7876 |
+
"learning_rate": 5.228085505409519e-06,
|
| 7877 |
+
"loss": 1.1738,
|
| 7878 |
+
"step": 1123
|
| 7879 |
+
},
|
| 7880 |
+
{
|
| 7881 |
+
"epoch": 1.4253968253968254,
|
| 7882 |
+
"grad_norm": 0.875,
|
| 7883 |
+
"learning_rate": 5.224947493327806e-06,
|
| 7884 |
+
"loss": 1.0843,
|
| 7885 |
+
"step": 1124
|
| 7886 |
+
},
|
| 7887 |
+
{
|
| 7888 |
+
"epoch": 1.4266666666666667,
|
| 7889 |
+
"grad_norm": 1.0546875,
|
| 7890 |
+
"learning_rate": 5.221807648739012e-06,
|
| 7891 |
+
"loss": 1.2114,
|
| 7892 |
+
"step": 1125
|
| 7893 |
+
},
|
| 7894 |
+
{
|
| 7895 |
+
"epoch": 1.4279365079365078,
|
| 7896 |
+
"grad_norm": 1.015625,
|
| 7897 |
+
"learning_rate": 5.218665974978767e-06,
|
| 7898 |
+
"loss": 1.1338,
|
| 7899 |
+
"step": 1126
|
| 7900 |
+
},
|
| 7901 |
+
{
|
| 7902 |
+
"epoch": 1.4292063492063491,
|
| 7903 |
+
"grad_norm": 1.015625,
|
| 7904 |
+
"learning_rate": 5.215522475384645e-06,
|
| 7905 |
+
"loss": 1.0466,
|
| 7906 |
+
"step": 1127
|
| 7907 |
+
},
|
| 7908 |
+
{
|
| 7909 |
+
"epoch": 1.4304761904761905,
|
| 7910 |
+
"grad_norm": 0.953125,
|
| 7911 |
+
"learning_rate": 5.21237715329616e-06,
|
| 7912 |
+
"loss": 1.1675,
|
| 7913 |
+
"step": 1128
|
| 7914 |
+
},
|
| 7915 |
+
{
|
| 7916 |
+
"epoch": 1.4317460317460318,
|
| 7917 |
+
"grad_norm": 0.984375,
|
| 7918 |
+
"learning_rate": 5.209230012054761e-06,
|
| 7919 |
+
"loss": 1.2035,
|
| 7920 |
+
"step": 1129
|
| 7921 |
+
},
|
| 7922 |
+
{
|
| 7923 |
+
"epoch": 1.433015873015873,
|
| 7924 |
+
"grad_norm": 0.97265625,
|
| 7925 |
+
"learning_rate": 5.20608105500383e-06,
|
| 7926 |
+
"loss": 1.0926,
|
| 7927 |
+
"step": 1130
|
| 7928 |
+
},
|
| 7929 |
+
{
|
| 7930 |
+
"epoch": 1.4342857142857142,
|
| 7931 |
+
"grad_norm": 0.9765625,
|
| 7932 |
+
"learning_rate": 5.202930285488679e-06,
|
| 7933 |
+
"loss": 1.0956,
|
| 7934 |
+
"step": 1131
|
| 7935 |
+
},
|
| 7936 |
+
{
|
| 7937 |
+
"epoch": 1.4355555555555555,
|
| 7938 |
+
"grad_norm": 0.9765625,
|
| 7939 |
+
"learning_rate": 5.1997777068565426e-06,
|
| 7940 |
+
"loss": 1.1305,
|
| 7941 |
+
"step": 1132
|
| 7942 |
+
},
|
| 7943 |
+
{
|
| 7944 |
+
"epoch": 1.4368253968253968,
|
| 7945 |
+
"grad_norm": 0.97265625,
|
| 7946 |
+
"learning_rate": 5.196623322456582e-06,
|
| 7947 |
+
"loss": 1.1132,
|
| 7948 |
+
"step": 1133
|
| 7949 |
+
},
|
| 7950 |
+
{
|
| 7951 |
+
"epoch": 1.438095238095238,
|
| 7952 |
+
"grad_norm": 0.9765625,
|
| 7953 |
+
"learning_rate": 5.193467135639873e-06,
|
| 7954 |
+
"loss": 1.0628,
|
| 7955 |
+
"step": 1134
|
| 7956 |
+
},
|
| 7957 |
+
{
|
| 7958 |
+
"epoch": 1.4393650793650794,
|
| 7959 |
+
"grad_norm": 0.9296875,
|
| 7960 |
+
"learning_rate": 5.1903091497594075e-06,
|
| 7961 |
+
"loss": 1.2005,
|
| 7962 |
+
"step": 1135
|
| 7963 |
+
},
|
| 7964 |
+
{
|
| 7965 |
+
"epoch": 1.4406349206349207,
|
| 7966 |
+
"grad_norm": 1.0234375,
|
| 7967 |
+
"learning_rate": 5.187149368170087e-06,
|
| 7968 |
+
"loss": 1.1237,
|
| 7969 |
+
"step": 1136
|
| 7970 |
+
},
|
| 7971 |
+
{
|
| 7972 |
+
"epoch": 1.441904761904762,
|
| 7973 |
+
"grad_norm": 1.015625,
|
| 7974 |
+
"learning_rate": 5.1839877942287245e-06,
|
| 7975 |
+
"loss": 1.099,
|
| 7976 |
+
"step": 1137
|
| 7977 |
+
},
|
| 7978 |
+
{
|
| 7979 |
+
"epoch": 1.443174603174603,
|
| 7980 |
+
"grad_norm": 0.97265625,
|
| 7981 |
+
"learning_rate": 5.180824431294034e-06,
|
| 7982 |
+
"loss": 1.1586,
|
| 7983 |
+
"step": 1138
|
| 7984 |
+
},
|
| 7985 |
+
{
|
| 7986 |
+
"epoch": 1.4444444444444444,
|
| 7987 |
+
"grad_norm": 0.96875,
|
| 7988 |
+
"learning_rate": 5.17765928272663e-06,
|
| 7989 |
+
"loss": 1.1243,
|
| 7990 |
+
"step": 1139
|
| 7991 |
+
},
|
| 7992 |
+
{
|
| 7993 |
+
"epoch": 1.4457142857142857,
|
| 7994 |
+
"grad_norm": 0.875,
|
| 7995 |
+
"learning_rate": 5.1744923518890255e-06,
|
| 7996 |
+
"loss": 1.0623,
|
| 7997 |
+
"step": 1140
|
| 7998 |
+
},
|
| 7999 |
+
{
|
| 8000 |
+
"epoch": 1.446984126984127,
|
| 8001 |
+
"grad_norm": 0.9765625,
|
| 8002 |
+
"learning_rate": 5.171323642145625e-06,
|
| 8003 |
+
"loss": 1.111,
|
| 8004 |
+
"step": 1141
|
| 8005 |
+
},
|
| 8006 |
+
{
|
| 8007 |
+
"epoch": 1.4482539682539683,
|
| 8008 |
+
"grad_norm": 0.94140625,
|
| 8009 |
+
"learning_rate": 5.168153156862726e-06,
|
| 8010 |
+
"loss": 1.0733,
|
| 8011 |
+
"step": 1142
|
| 8012 |
+
},
|
| 8013 |
+
{
|
| 8014 |
+
"epoch": 1.4495238095238094,
|
| 8015 |
+
"grad_norm": 0.921875,
|
| 8016 |
+
"learning_rate": 5.164980899408509e-06,
|
| 8017 |
+
"loss": 1.0906,
|
| 8018 |
+
"step": 1143
|
| 8019 |
+
},
|
| 8020 |
+
{
|
| 8021 |
+
"epoch": 1.4507936507936507,
|
| 8022 |
+
"grad_norm": 0.96875,
|
| 8023 |
+
"learning_rate": 5.1618068731530395e-06,
|
| 8024 |
+
"loss": 1.1173,
|
| 8025 |
+
"step": 1144
|
| 8026 |
+
},
|
| 8027 |
+
{
|
| 8028 |
+
"epoch": 1.452063492063492,
|
| 8029 |
+
"grad_norm": 1.0,
|
| 8030 |
+
"learning_rate": 5.15863108146826e-06,
|
| 8031 |
+
"loss": 1.1727,
|
| 8032 |
+
"step": 1145
|
| 8033 |
+
},
|
| 8034 |
+
{
|
| 8035 |
+
"epoch": 1.4533333333333334,
|
| 8036 |
+
"grad_norm": 1.03125,
|
| 8037 |
+
"learning_rate": 5.155453527727989e-06,
|
| 8038 |
+
"loss": 1.1193,
|
| 8039 |
+
"step": 1146
|
| 8040 |
+
},
|
| 8041 |
+
{
|
| 8042 |
+
"epoch": 1.4546031746031747,
|
| 8043 |
+
"grad_norm": 0.96484375,
|
| 8044 |
+
"learning_rate": 5.15227421530792e-06,
|
| 8045 |
+
"loss": 1.1275,
|
| 8046 |
+
"step": 1147
|
| 8047 |
+
},
|
| 8048 |
+
{
|
| 8049 |
+
"epoch": 1.4558730158730158,
|
| 8050 |
+
"grad_norm": 0.94921875,
|
| 8051 |
+
"learning_rate": 5.14909314758561e-06,
|
| 8052 |
+
"loss": 1.1938,
|
| 8053 |
+
"step": 1148
|
| 8054 |
+
},
|
| 8055 |
+
{
|
| 8056 |
+
"epoch": 1.457142857142857,
|
| 8057 |
+
"grad_norm": 0.953125,
|
| 8058 |
+
"learning_rate": 5.145910327940484e-06,
|
| 8059 |
+
"loss": 1.1307,
|
| 8060 |
+
"step": 1149
|
| 8061 |
+
},
|
| 8062 |
+
{
|
| 8063 |
+
"epoch": 1.4584126984126984,
|
| 8064 |
+
"grad_norm": 1.0703125,
|
| 8065 |
+
"learning_rate": 5.142725759753826e-06,
|
| 8066 |
+
"loss": 1.2017,
|
| 8067 |
+
"step": 1150
|
| 8068 |
+
},
|
| 8069 |
+
{
|
| 8070 |
+
"epoch": 1.4596825396825397,
|
| 8071 |
+
"grad_norm": 1.046875,
|
| 8072 |
+
"learning_rate": 5.139539446408781e-06,
|
| 8073 |
+
"loss": 1.1306,
|
| 8074 |
+
"step": 1151
|
| 8075 |
+
},
|
| 8076 |
+
{
|
| 8077 |
+
"epoch": 1.460952380952381,
|
| 8078 |
+
"grad_norm": 0.9453125,
|
| 8079 |
+
"learning_rate": 5.136351391290346e-06,
|
| 8080 |
+
"loss": 1.102,
|
| 8081 |
+
"step": 1152
|
| 8082 |
+
},
|
| 8083 |
+
{
|
| 8084 |
+
"epoch": 1.462222222222222,
|
| 8085 |
+
"grad_norm": 0.859375,
|
| 8086 |
+
"learning_rate": 5.133161597785365e-06,
|
| 8087 |
+
"loss": 1.1168,
|
| 8088 |
+
"step": 1153
|
| 8089 |
+
},
|
| 8090 |
+
{
|
| 8091 |
+
"epoch": 1.4634920634920636,
|
| 8092 |
+
"grad_norm": 0.9921875,
|
| 8093 |
+
"learning_rate": 5.129970069282535e-06,
|
| 8094 |
+
"loss": 1.1488,
|
| 8095 |
+
"step": 1154
|
| 8096 |
+
},
|
| 8097 |
+
{
|
| 8098 |
+
"epoch": 1.4647619047619047,
|
| 8099 |
+
"grad_norm": 0.95703125,
|
| 8100 |
+
"learning_rate": 5.126776809172392e-06,
|
| 8101 |
+
"loss": 1.0992,
|
| 8102 |
+
"step": 1155
|
| 8103 |
+
},
|
| 8104 |
+
{
|
| 8105 |
+
"epoch": 1.466031746031746,
|
| 8106 |
+
"grad_norm": 0.94921875,
|
| 8107 |
+
"learning_rate": 5.123581820847313e-06,
|
| 8108 |
+
"loss": 1.1016,
|
| 8109 |
+
"step": 1156
|
| 8110 |
+
},
|
| 8111 |
+
{
|
| 8112 |
+
"epoch": 1.4673015873015873,
|
| 8113 |
+
"grad_norm": 0.9765625,
|
| 8114 |
+
"learning_rate": 5.1203851077015106e-06,
|
| 8115 |
+
"loss": 1.12,
|
| 8116 |
+
"step": 1157
|
| 8117 |
+
},
|
| 8118 |
+
{
|
| 8119 |
+
"epoch": 1.4685714285714286,
|
| 8120 |
+
"grad_norm": 0.9375,
|
| 8121 |
+
"learning_rate": 5.1171866731310285e-06,
|
| 8122 |
+
"loss": 1.087,
|
| 8123 |
+
"step": 1158
|
| 8124 |
+
},
|
| 8125 |
+
{
|
| 8126 |
+
"epoch": 1.46984126984127,
|
| 8127 |
+
"grad_norm": 1.0234375,
|
| 8128 |
+
"learning_rate": 5.113986520533743e-06,
|
| 8129 |
+
"loss": 1.1197,
|
| 8130 |
+
"step": 1159
|
| 8131 |
+
},
|
| 8132 |
+
{
|
| 8133 |
+
"epoch": 1.471111111111111,
|
| 8134 |
+
"grad_norm": 1.1015625,
|
| 8135 |
+
"learning_rate": 5.110784653309353e-06,
|
| 8136 |
+
"loss": 1.1768,
|
| 8137 |
+
"step": 1160
|
| 8138 |
+
},
|
| 8139 |
+
{
|
| 8140 |
+
"epoch": 1.4723809523809523,
|
| 8141 |
+
"grad_norm": 0.9765625,
|
| 8142 |
+
"learning_rate": 5.1075810748593765e-06,
|
| 8143 |
+
"loss": 1.1136,
|
| 8144 |
+
"step": 1161
|
| 8145 |
+
},
|
| 8146 |
+
{
|
| 8147 |
+
"epoch": 1.4736507936507937,
|
| 8148 |
+
"grad_norm": 1.0703125,
|
| 8149 |
+
"learning_rate": 5.104375788587154e-06,
|
| 8150 |
+
"loss": 1.1602,
|
| 8151 |
+
"step": 1162
|
| 8152 |
+
},
|
| 8153 |
+
{
|
| 8154 |
+
"epoch": 1.474920634920635,
|
| 8155 |
+
"grad_norm": 1.0703125,
|
| 8156 |
+
"learning_rate": 5.1011687978978394e-06,
|
| 8157 |
+
"loss": 1.1295,
|
| 8158 |
+
"step": 1163
|
| 8159 |
+
},
|
| 8160 |
+
{
|
| 8161 |
+
"epoch": 1.4761904761904763,
|
| 8162 |
+
"grad_norm": 1.0390625,
|
| 8163 |
+
"learning_rate": 5.097960106198393e-06,
|
| 8164 |
+
"loss": 1.2086,
|
| 8165 |
+
"step": 1164
|
| 8166 |
+
},
|
| 8167 |
+
{
|
| 8168 |
+
"epoch": 1.4774603174603174,
|
| 8169 |
+
"grad_norm": 0.94140625,
|
| 8170 |
+
"learning_rate": 5.094749716897588e-06,
|
| 8171 |
+
"loss": 1.1108,
|
| 8172 |
+
"step": 1165
|
| 8173 |
+
},
|
| 8174 |
+
{
|
| 8175 |
+
"epoch": 1.4787301587301587,
|
| 8176 |
+
"grad_norm": 0.96484375,
|
| 8177 |
+
"learning_rate": 5.091537633405996e-06,
|
| 8178 |
+
"loss": 1.1007,
|
| 8179 |
+
"step": 1166
|
| 8180 |
+
},
|
| 8181 |
+
{
|
| 8182 |
+
"epoch": 1.48,
|
| 8183 |
+
"grad_norm": 1.0078125,
|
| 8184 |
+
"learning_rate": 5.088323859135995e-06,
|
| 8185 |
+
"loss": 1.0905,
|
| 8186 |
+
"step": 1167
|
| 8187 |
+
},
|
| 8188 |
+
{
|
| 8189 |
+
"epoch": 1.4812698412698413,
|
| 8190 |
+
"grad_norm": 0.953125,
|
| 8191 |
+
"learning_rate": 5.0851083975017515e-06,
|
| 8192 |
+
"loss": 1.1302,
|
| 8193 |
+
"step": 1168
|
| 8194 |
+
},
|
| 8195 |
+
{
|
| 8196 |
+
"epoch": 1.4825396825396826,
|
| 8197 |
+
"grad_norm": 0.921875,
|
| 8198 |
+
"learning_rate": 5.081891251919227e-06,
|
| 8199 |
+
"loss": 1.104,
|
| 8200 |
+
"step": 1169
|
| 8201 |
+
},
|
| 8202 |
+
{
|
| 8203 |
+
"epoch": 1.4838095238095237,
|
| 8204 |
+
"grad_norm": 0.94140625,
|
| 8205 |
+
"learning_rate": 5.078672425806178e-06,
|
| 8206 |
+
"loss": 1.0844,
|
| 8207 |
+
"step": 1170
|
| 8208 |
+
},
|
| 8209 |
+
{
|
| 8210 |
+
"epoch": 1.485079365079365,
|
| 8211 |
+
"grad_norm": 0.921875,
|
| 8212 |
+
"learning_rate": 5.075451922582138e-06,
|
| 8213 |
+
"loss": 1.0839,
|
| 8214 |
+
"step": 1171
|
| 8215 |
+
},
|
| 8216 |
+
{
|
| 8217 |
+
"epoch": 1.4863492063492063,
|
| 8218 |
+
"grad_norm": 0.9609375,
|
| 8219 |
+
"learning_rate": 5.072229745668424e-06,
|
| 8220 |
+
"loss": 1.1437,
|
| 8221 |
+
"step": 1172
|
| 8222 |
+
},
|
| 8223 |
+
{
|
| 8224 |
+
"epoch": 1.4876190476190476,
|
| 8225 |
+
"grad_norm": 1.015625,
|
| 8226 |
+
"learning_rate": 5.069005898488137e-06,
|
| 8227 |
+
"loss": 1.0498,
|
| 8228 |
+
"step": 1173
|
| 8229 |
+
},
|
| 8230 |
+
{
|
| 8231 |
+
"epoch": 1.488888888888889,
|
| 8232 |
+
"grad_norm": 0.99609375,
|
| 8233 |
+
"learning_rate": 5.065780384466147e-06,
|
| 8234 |
+
"loss": 1.1558,
|
| 8235 |
+
"step": 1174
|
| 8236 |
+
},
|
| 8237 |
+
{
|
| 8238 |
+
"epoch": 1.4901587301587302,
|
| 8239 |
+
"grad_norm": 1.0546875,
|
| 8240 |
+
"learning_rate": 5.062553207029093e-06,
|
| 8241 |
+
"loss": 1.1752,
|
| 8242 |
+
"step": 1175
|
| 8243 |
+
},
|
| 8244 |
+
{
|
| 8245 |
+
"epoch": 1.4914285714285715,
|
| 8246 |
+
"grad_norm": 1.0390625,
|
| 8247 |
+
"learning_rate": 5.059324369605386e-06,
|
| 8248 |
+
"loss": 1.1394,
|
| 8249 |
+
"step": 1176
|
| 8250 |
+
},
|
| 8251 |
+
{
|
| 8252 |
+
"epoch": 1.4926984126984126,
|
| 8253 |
+
"grad_norm": 0.96484375,
|
| 8254 |
+
"learning_rate": 5.056093875625199e-06,
|
| 8255 |
+
"loss": 1.1602,
|
| 8256 |
+
"step": 1177
|
| 8257 |
+
},
|
| 8258 |
+
{
|
| 8259 |
+
"epoch": 1.493968253968254,
|
| 8260 |
+
"grad_norm": 0.953125,
|
| 8261 |
+
"learning_rate": 5.052861728520462e-06,
|
| 8262 |
+
"loss": 1.1289,
|
| 8263 |
+
"step": 1178
|
| 8264 |
+
},
|
| 8265 |
+
{
|
| 8266 |
+
"epoch": 1.4952380952380953,
|
| 8267 |
+
"grad_norm": 0.93359375,
|
| 8268 |
+
"learning_rate": 5.049627931724868e-06,
|
| 8269 |
+
"loss": 1.1,
|
| 8270 |
+
"step": 1179
|
| 8271 |
+
},
|
| 8272 |
+
{
|
| 8273 |
+
"epoch": 1.4965079365079366,
|
| 8274 |
+
"grad_norm": 0.9140625,
|
| 8275 |
+
"learning_rate": 5.046392488673853e-06,
|
| 8276 |
+
"loss": 1.1183,
|
| 8277 |
+
"step": 1180
|
| 8278 |
+
},
|
| 8279 |
+
{
|
| 8280 |
+
"epoch": 1.4977777777777779,
|
| 8281 |
+
"grad_norm": 0.9296875,
|
| 8282 |
+
"learning_rate": 5.043155402804611e-06,
|
| 8283 |
+
"loss": 1.2041,
|
| 8284 |
+
"step": 1181
|
| 8285 |
+
},
|
| 8286 |
+
{
|
| 8287 |
+
"epoch": 1.499047619047619,
|
| 8288 |
+
"grad_norm": 0.96484375,
|
| 8289 |
+
"learning_rate": 5.039916677556075e-06,
|
| 8290 |
+
"loss": 1.1673,
|
| 8291 |
+
"step": 1182
|
| 8292 |
}
|
| 8293 |
],
|
| 8294 |
"logging_steps": 1,
|
|
|
|
| 8308 |
"attributes": {}
|
| 8309 |
}
|
| 8310 |
},
|
| 8311 |
+
"total_flos": 2.6179854689516913e+18,
|
| 8312 |
"train_batch_size": 2,
|
| 8313 |
"trial_name": null,
|
| 8314 |
"trial_params": null
|