Training in progress, step 1348, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 27024
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:40ee21f947c861f283979fc9002276136bca96cc9df1a7eca3efe53eff88d4aa
|
3 |
size 27024
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 64038
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:245d0993b8b552a2d4001ecfb38d11a8ea2c01c327875647fe1afab8460b08a4
|
3 |
size 64038
|
last-checkpoint/rng_state.pth
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 14244
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:df61d42dbf64890c146fc1b345e5825548bdc732960cf7184a20a2a496707abb
|
3 |
size 14244
|
last-checkpoint/scheduler.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 1064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:dbe6b681f982230cf148c02eb0c76204db7bc5c5c9dc78ffffd92b93ee47cc19
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 337,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7116,6 +7116,2373 @@
|
|
7116 |
"eval_samples_per_second": 284.832,
|
7117 |
"eval_steps_per_second": 142.559,
|
7118 |
"step": 1011
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7119 |
}
|
7120 |
],
|
7121 |
"logging_steps": 1,
|
@@ -7130,12 +9497,12 @@
|
|
7130 |
"should_evaluate": false,
|
7131 |
"should_log": false,
|
7132 |
"should_save": true,
|
7133 |
-
"should_training_stop":
|
7134 |
},
|
7135 |
"attributes": {}
|
7136 |
}
|
7137 |
},
|
7138 |
-
"total_flos":
|
7139 |
"train_batch_size": 2,
|
7140 |
"trial_name": null,
|
7141 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.5708840656431975,
|
5 |
"eval_steps": 337,
|
6 |
+
"global_step": 1348,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7116 |
"eval_samples_per_second": 284.832,
|
7117 |
"eval_steps_per_second": 142.559,
|
7118 |
"step": 1011
|
7119 |
+
},
|
7120 |
+
{
|
7121 |
+
"epoch": 0.4285865537321334,
|
7122 |
+
"grad_norm": 0.03415974974632263,
|
7123 |
+
"learning_rate": 2.9538800445457946e-05,
|
7124 |
+
"loss": 10.3323,
|
7125 |
+
"step": 1012
|
7126 |
+
},
|
7127 |
+
{
|
7128 |
+
"epoch": 0.4290100582318687,
|
7129 |
+
"grad_norm": 0.039172153919935226,
|
7130 |
+
"learning_rate": 2.9372384098526784e-05,
|
7131 |
+
"loss": 10.3347,
|
7132 |
+
"step": 1013
|
7133 |
+
},
|
7134 |
+
{
|
7135 |
+
"epoch": 0.42943356273160405,
|
7136 |
+
"grad_norm": 0.031853485852479935,
|
7137 |
+
"learning_rate": 2.9206357121038285e-05,
|
7138 |
+
"loss": 10.3338,
|
7139 |
+
"step": 1014
|
7140 |
+
},
|
7141 |
+
{
|
7142 |
+
"epoch": 0.4298570672313393,
|
7143 |
+
"grad_norm": 0.04923943430185318,
|
7144 |
+
"learning_rate": 2.904072042829775e-05,
|
7145 |
+
"loss": 10.3323,
|
7146 |
+
"step": 1015
|
7147 |
+
},
|
7148 |
+
{
|
7149 |
+
"epoch": 0.43028057173107465,
|
7150 |
+
"grad_norm": 0.03674182668328285,
|
7151 |
+
"learning_rate": 2.8875474933458847e-05,
|
7152 |
+
"loss": 10.3334,
|
7153 |
+
"step": 1016
|
7154 |
+
},
|
7155 |
+
{
|
7156 |
+
"epoch": 0.43070407623081,
|
7157 |
+
"grad_norm": 0.030546877533197403,
|
7158 |
+
"learning_rate": 2.871062154751858e-05,
|
7159 |
+
"loss": 10.3296,
|
7160 |
+
"step": 1017
|
7161 |
+
},
|
7162 |
+
{
|
7163 |
+
"epoch": 0.43112758073054525,
|
7164 |
+
"grad_norm": 0.030613403767347336,
|
7165 |
+
"learning_rate": 2.8546161179312248e-05,
|
7166 |
+
"loss": 10.3354,
|
7167 |
+
"step": 1018
|
7168 |
+
},
|
7169 |
+
{
|
7170 |
+
"epoch": 0.4315510852302806,
|
7171 |
+
"grad_norm": 0.030776720494031906,
|
7172 |
+
"learning_rate": 2.8382094735508457e-05,
|
7173 |
+
"loss": 10.3303,
|
7174 |
+
"step": 1019
|
7175 |
+
},
|
7176 |
+
{
|
7177 |
+
"epoch": 0.4319745897300159,
|
7178 |
+
"grad_norm": 0.03810757398605347,
|
7179 |
+
"learning_rate": 2.821842312060409e-05,
|
7180 |
+
"loss": 10.3334,
|
7181 |
+
"step": 1020
|
7182 |
+
},
|
7183 |
+
{
|
7184 |
+
"epoch": 0.4323980942297512,
|
7185 |
+
"grad_norm": 0.030035821720957756,
|
7186 |
+
"learning_rate": 2.8055147236919442e-05,
|
7187 |
+
"loss": 10.3345,
|
7188 |
+
"step": 1021
|
7189 |
+
},
|
7190 |
+
{
|
7191 |
+
"epoch": 0.4328215987294865,
|
7192 |
+
"grad_norm": 0.03650267794728279,
|
7193 |
+
"learning_rate": 2.789226798459298e-05,
|
7194 |
+
"loss": 10.3299,
|
7195 |
+
"step": 1022
|
7196 |
+
},
|
7197 |
+
{
|
7198 |
+
"epoch": 0.43324510322922183,
|
7199 |
+
"grad_norm": 0.030346672981977463,
|
7200 |
+
"learning_rate": 2.7729786261576617e-05,
|
7201 |
+
"loss": 10.334,
|
7202 |
+
"step": 1023
|
7203 |
+
},
|
7204 |
+
{
|
7205 |
+
"epoch": 0.4336686077289571,
|
7206 |
+
"grad_norm": 0.0330539271235466,
|
7207 |
+
"learning_rate": 2.7567702963630803e-05,
|
7208 |
+
"loss": 10.3316,
|
7209 |
+
"step": 1024
|
7210 |
+
},
|
7211 |
+
{
|
7212 |
+
"epoch": 0.43409211222869243,
|
7213 |
+
"grad_norm": 0.03174733370542526,
|
7214 |
+
"learning_rate": 2.740601898431925e-05,
|
7215 |
+
"loss": 10.3278,
|
7216 |
+
"step": 1025
|
7217 |
+
},
|
7218 |
+
{
|
7219 |
+
"epoch": 0.43451561672842776,
|
7220 |
+
"grad_norm": 0.03628386929631233,
|
7221 |
+
"learning_rate": 2.7244735215004446e-05,
|
7222 |
+
"loss": 10.3274,
|
7223 |
+
"step": 1026
|
7224 |
+
},
|
7225 |
+
{
|
7226 |
+
"epoch": 0.43493912122816303,
|
7227 |
+
"grad_norm": 0.024906015023589134,
|
7228 |
+
"learning_rate": 2.7083852544842436e-05,
|
7229 |
+
"loss": 10.3332,
|
7230 |
+
"step": 1027
|
7231 |
+
},
|
7232 |
+
{
|
7233 |
+
"epoch": 0.43536262572789836,
|
7234 |
+
"grad_norm": 0.043956976383924484,
|
7235 |
+
"learning_rate": 2.692337186077791e-05,
|
7236 |
+
"loss": 10.3266,
|
7237 |
+
"step": 1028
|
7238 |
+
},
|
7239 |
+
{
|
7240 |
+
"epoch": 0.4357861302276337,
|
7241 |
+
"grad_norm": 0.032996706664562225,
|
7242 |
+
"learning_rate": 2.67632940475396e-05,
|
7243 |
+
"loss": 10.3346,
|
7244 |
+
"step": 1029
|
7245 |
+
},
|
7246 |
+
{
|
7247 |
+
"epoch": 0.43620963472736896,
|
7248 |
+
"grad_norm": 0.044276829808950424,
|
7249 |
+
"learning_rate": 2.6603619987635086e-05,
|
7250 |
+
"loss": 10.3274,
|
7251 |
+
"step": 1030
|
7252 |
+
},
|
7253 |
+
{
|
7254 |
+
"epoch": 0.4366331392271043,
|
7255 |
+
"grad_norm": 0.038449618965387344,
|
7256 |
+
"learning_rate": 2.64443505613461e-05,
|
7257 |
+
"loss": 10.3341,
|
7258 |
+
"step": 1031
|
7259 |
+
},
|
7260 |
+
{
|
7261 |
+
"epoch": 0.4370566437268396,
|
7262 |
+
"grad_norm": 0.03220584616065025,
|
7263 |
+
"learning_rate": 2.6285486646723634e-05,
|
7264 |
+
"loss": 10.3324,
|
7265 |
+
"step": 1032
|
7266 |
+
},
|
7267 |
+
{
|
7268 |
+
"epoch": 0.4374801482265749,
|
7269 |
+
"grad_norm": 0.03746611624956131,
|
7270 |
+
"learning_rate": 2.612702911958308e-05,
|
7271 |
+
"loss": 10.3354,
|
7272 |
+
"step": 1033
|
7273 |
+
},
|
7274 |
+
{
|
7275 |
+
"epoch": 0.4379036527263102,
|
7276 |
+
"grad_norm": 0.04333876073360443,
|
7277 |
+
"learning_rate": 2.5968978853499425e-05,
|
7278 |
+
"loss": 10.329,
|
7279 |
+
"step": 1034
|
7280 |
+
},
|
7281 |
+
{
|
7282 |
+
"epoch": 0.43832715722604554,
|
7283 |
+
"grad_norm": 0.03539913892745972,
|
7284 |
+
"learning_rate": 2.581133671980246e-05,
|
7285 |
+
"loss": 10.3324,
|
7286 |
+
"step": 1035
|
7287 |
+
},
|
7288 |
+
{
|
7289 |
+
"epoch": 0.4387506617257808,
|
7290 |
+
"grad_norm": 0.04690808430314064,
|
7291 |
+
"learning_rate": 2.565410358757189e-05,
|
7292 |
+
"loss": 10.3316,
|
7293 |
+
"step": 1036
|
7294 |
+
},
|
7295 |
+
{
|
7296 |
+
"epoch": 0.43917416622551614,
|
7297 |
+
"grad_norm": 0.038458049297332764,
|
7298 |
+
"learning_rate": 2.5497280323632654e-05,
|
7299 |
+
"loss": 10.3431,
|
7300 |
+
"step": 1037
|
7301 |
+
},
|
7302 |
+
{
|
7303 |
+
"epoch": 0.43959767072525147,
|
7304 |
+
"grad_norm": 0.03451355919241905,
|
7305 |
+
"learning_rate": 2.534086779255005e-05,
|
7306 |
+
"loss": 10.3296,
|
7307 |
+
"step": 1038
|
7308 |
+
},
|
7309 |
+
{
|
7310 |
+
"epoch": 0.44002117522498674,
|
7311 |
+
"grad_norm": 0.03873763233423233,
|
7312 |
+
"learning_rate": 2.5184866856625023e-05,
|
7313 |
+
"loss": 10.3273,
|
7314 |
+
"step": 1039
|
7315 |
+
},
|
7316 |
+
{
|
7317 |
+
"epoch": 0.44044467972472207,
|
7318 |
+
"grad_norm": 0.044388849288225174,
|
7319 |
+
"learning_rate": 2.5029278375889387e-05,
|
7320 |
+
"loss": 10.3324,
|
7321 |
+
"step": 1040
|
7322 |
+
},
|
7323 |
+
{
|
7324 |
+
"epoch": 0.4408681842244574,
|
7325 |
+
"grad_norm": 0.03534289821982384,
|
7326 |
+
"learning_rate": 2.4874103208101183e-05,
|
7327 |
+
"loss": 10.3343,
|
7328 |
+
"step": 1041
|
7329 |
+
},
|
7330 |
+
{
|
7331 |
+
"epoch": 0.4412916887241927,
|
7332 |
+
"grad_norm": 0.0375693254172802,
|
7333 |
+
"learning_rate": 2.4719342208739693e-05,
|
7334 |
+
"loss": 10.3323,
|
7335 |
+
"step": 1042
|
7336 |
+
},
|
7337 |
+
{
|
7338 |
+
"epoch": 0.441715193223928,
|
7339 |
+
"grad_norm": 0.03341260179877281,
|
7340 |
+
"learning_rate": 2.456499623100098e-05,
|
7341 |
+
"loss": 10.3318,
|
7342 |
+
"step": 1043
|
7343 |
+
},
|
7344 |
+
{
|
7345 |
+
"epoch": 0.4421386977236633,
|
7346 |
+
"grad_norm": 0.04234972223639488,
|
7347 |
+
"learning_rate": 2.4411066125793203e-05,
|
7348 |
+
"loss": 10.3319,
|
7349 |
+
"step": 1044
|
7350 |
+
},
|
7351 |
+
{
|
7352 |
+
"epoch": 0.44256220222339865,
|
7353 |
+
"grad_norm": 0.031914252787828445,
|
7354 |
+
"learning_rate": 2.4257552741731592e-05,
|
7355 |
+
"loss": 10.3361,
|
7356 |
+
"step": 1045
|
7357 |
+
},
|
7358 |
+
{
|
7359 |
+
"epoch": 0.4429857067231339,
|
7360 |
+
"grad_norm": 0.05003447085618973,
|
7361 |
+
"learning_rate": 2.41044569251342e-05,
|
7362 |
+
"loss": 10.3313,
|
7363 |
+
"step": 1046
|
7364 |
+
},
|
7365 |
+
{
|
7366 |
+
"epoch": 0.44340921122286925,
|
7367 |
+
"grad_norm": 0.03364928439259529,
|
7368 |
+
"learning_rate": 2.3951779520016937e-05,
|
7369 |
+
"loss": 10.33,
|
7370 |
+
"step": 1047
|
7371 |
+
},
|
7372 |
+
{
|
7373 |
+
"epoch": 0.4438327157226046,
|
7374 |
+
"grad_norm": 0.028291532769799232,
|
7375 |
+
"learning_rate": 2.379952136808903e-05,
|
7376 |
+
"loss": 10.3336,
|
7377 |
+
"step": 1048
|
7378 |
+
},
|
7379 |
+
{
|
7380 |
+
"epoch": 0.44425622022233985,
|
7381 |
+
"grad_norm": 0.042799290269613266,
|
7382 |
+
"learning_rate": 2.3647683308748392e-05,
|
7383 |
+
"loss": 10.3348,
|
7384 |
+
"step": 1049
|
7385 |
+
},
|
7386 |
+
{
|
7387 |
+
"epoch": 0.4446797247220752,
|
7388 |
+
"grad_norm": 0.042522724717855453,
|
7389 |
+
"learning_rate": 2.3496266179076864e-05,
|
7390 |
+
"loss": 10.3288,
|
7391 |
+
"step": 1050
|
7392 |
+
},
|
7393 |
+
{
|
7394 |
+
"epoch": 0.4451032292218105,
|
7395 |
+
"grad_norm": 0.02918383479118347,
|
7396 |
+
"learning_rate": 2.3345270813835886e-05,
|
7397 |
+
"loss": 10.3361,
|
7398 |
+
"step": 1051
|
7399 |
+
},
|
7400 |
+
{
|
7401 |
+
"epoch": 0.4455267337215458,
|
7402 |
+
"grad_norm": 0.046009406447410583,
|
7403 |
+
"learning_rate": 2.319469804546156e-05,
|
7404 |
+
"loss": 10.3349,
|
7405 |
+
"step": 1052
|
7406 |
+
},
|
7407 |
+
{
|
7408 |
+
"epoch": 0.4459502382212811,
|
7409 |
+
"grad_norm": 0.03431849181652069,
|
7410 |
+
"learning_rate": 2.3044548704060288e-05,
|
7411 |
+
"loss": 10.3283,
|
7412 |
+
"step": 1053
|
7413 |
+
},
|
7414 |
+
{
|
7415 |
+
"epoch": 0.44637374272101643,
|
7416 |
+
"grad_norm": 0.03582574054598808,
|
7417 |
+
"learning_rate": 2.2894823617404104e-05,
|
7418 |
+
"loss": 10.3314,
|
7419 |
+
"step": 1054
|
7420 |
+
},
|
7421 |
+
{
|
7422 |
+
"epoch": 0.4467972472207517,
|
7423 |
+
"grad_norm": 0.02972414344549179,
|
7424 |
+
"learning_rate": 2.2745523610926122e-05,
|
7425 |
+
"loss": 10.3289,
|
7426 |
+
"step": 1055
|
7427 |
+
},
|
7428 |
+
{
|
7429 |
+
"epoch": 0.44722075172048703,
|
7430 |
+
"grad_norm": 0.03548819199204445,
|
7431 |
+
"learning_rate": 2.2596649507716018e-05,
|
7432 |
+
"loss": 10.3299,
|
7433 |
+
"step": 1056
|
7434 |
+
},
|
7435 |
+
{
|
7436 |
+
"epoch": 0.44764425622022236,
|
7437 |
+
"grad_norm": 0.04241335019469261,
|
7438 |
+
"learning_rate": 2.244820212851544e-05,
|
7439 |
+
"loss": 10.3308,
|
7440 |
+
"step": 1057
|
7441 |
+
},
|
7442 |
+
{
|
7443 |
+
"epoch": 0.44806776071995763,
|
7444 |
+
"grad_norm": 0.033176884055137634,
|
7445 |
+
"learning_rate": 2.2300182291713513e-05,
|
7446 |
+
"loss": 10.3351,
|
7447 |
+
"step": 1058
|
7448 |
+
},
|
7449 |
+
{
|
7450 |
+
"epoch": 0.44849126521969296,
|
7451 |
+
"grad_norm": 0.032935190945863724,
|
7452 |
+
"learning_rate": 2.2152590813342345e-05,
|
7453 |
+
"loss": 10.3356,
|
7454 |
+
"step": 1059
|
7455 |
+
},
|
7456 |
+
{
|
7457 |
+
"epoch": 0.4489147697194283,
|
7458 |
+
"grad_norm": 0.030969172716140747,
|
7459 |
+
"learning_rate": 2.2005428507072467e-05,
|
7460 |
+
"loss": 10.3307,
|
7461 |
+
"step": 1060
|
7462 |
+
},
|
7463 |
+
{
|
7464 |
+
"epoch": 0.44933827421916356,
|
7465 |
+
"grad_norm": 0.036834247410297394,
|
7466 |
+
"learning_rate": 2.1858696184208484e-05,
|
7467 |
+
"loss": 10.3324,
|
7468 |
+
"step": 1061
|
7469 |
+
},
|
7470 |
+
{
|
7471 |
+
"epoch": 0.4497617787188989,
|
7472 |
+
"grad_norm": 0.038617976009845734,
|
7473 |
+
"learning_rate": 2.1712394653684344e-05,
|
7474 |
+
"loss": 10.3371,
|
7475 |
+
"step": 1062
|
7476 |
+
},
|
7477 |
+
{
|
7478 |
+
"epoch": 0.4501852832186342,
|
7479 |
+
"grad_norm": 0.026445934548974037,
|
7480 |
+
"learning_rate": 2.15665247220592e-05,
|
7481 |
+
"loss": 10.3334,
|
7482 |
+
"step": 1063
|
7483 |
+
},
|
7484 |
+
{
|
7485 |
+
"epoch": 0.4506087877183695,
|
7486 |
+
"grad_norm": 0.04230870306491852,
|
7487 |
+
"learning_rate": 2.1421087193512756e-05,
|
7488 |
+
"loss": 10.3261,
|
7489 |
+
"step": 1064
|
7490 |
+
},
|
7491 |
+
{
|
7492 |
+
"epoch": 0.4510322922181048,
|
7493 |
+
"grad_norm": 0.03189300373196602,
|
7494 |
+
"learning_rate": 2.1276082869840765e-05,
|
7495 |
+
"loss": 10.3297,
|
7496 |
+
"step": 1065
|
7497 |
+
},
|
7498 |
+
{
|
7499 |
+
"epoch": 0.45145579671784014,
|
7500 |
+
"grad_norm": 0.03367699310183525,
|
7501 |
+
"learning_rate": 2.113151255045095e-05,
|
7502 |
+
"loss": 10.3308,
|
7503 |
+
"step": 1066
|
7504 |
+
},
|
7505 |
+
{
|
7506 |
+
"epoch": 0.4518793012175754,
|
7507 |
+
"grad_norm": 0.032475464046001434,
|
7508 |
+
"learning_rate": 2.0987377032358114e-05,
|
7509 |
+
"loss": 10.339,
|
7510 |
+
"step": 1067
|
7511 |
+
},
|
7512 |
+
{
|
7513 |
+
"epoch": 0.45230280571731074,
|
7514 |
+
"grad_norm": 0.04436371102929115,
|
7515 |
+
"learning_rate": 2.084367711018024e-05,
|
7516 |
+
"loss": 10.3301,
|
7517 |
+
"step": 1068
|
7518 |
+
},
|
7519 |
+
{
|
7520 |
+
"epoch": 0.45272631021704607,
|
7521 |
+
"grad_norm": 0.037988126277923584,
|
7522 |
+
"learning_rate": 2.070041357613376e-05,
|
7523 |
+
"loss": 10.3309,
|
7524 |
+
"step": 1069
|
7525 |
+
},
|
7526 |
+
{
|
7527 |
+
"epoch": 0.45314981471678134,
|
7528 |
+
"grad_norm": 0.03870435804128647,
|
7529 |
+
"learning_rate": 2.0557587220029228e-05,
|
7530 |
+
"loss": 10.3353,
|
7531 |
+
"step": 1070
|
7532 |
+
},
|
7533 |
+
{
|
7534 |
+
"epoch": 0.45357331921651667,
|
7535 |
+
"grad_norm": 0.03660368546843529,
|
7536 |
+
"learning_rate": 2.0415198829267212e-05,
|
7537 |
+
"loss": 10.3317,
|
7538 |
+
"step": 1071
|
7539 |
+
},
|
7540 |
+
{
|
7541 |
+
"epoch": 0.453996823716252,
|
7542 |
+
"grad_norm": 0.03593965247273445,
|
7543 |
+
"learning_rate": 2.0273249188833654e-05,
|
7544 |
+
"loss": 10.3343,
|
7545 |
+
"step": 1072
|
7546 |
+
},
|
7547 |
+
{
|
7548 |
+
"epoch": 0.45442032821598727,
|
7549 |
+
"grad_norm": 0.03798775374889374,
|
7550 |
+
"learning_rate": 2.013173908129573e-05,
|
7551 |
+
"loss": 10.329,
|
7552 |
+
"step": 1073
|
7553 |
+
},
|
7554 |
+
{
|
7555 |
+
"epoch": 0.4548438327157226,
|
7556 |
+
"grad_norm": 0.030165789648890495,
|
7557 |
+
"learning_rate": 1.9990669286797438e-05,
|
7558 |
+
"loss": 10.3325,
|
7559 |
+
"step": 1074
|
7560 |
+
},
|
7561 |
+
{
|
7562 |
+
"epoch": 0.4552673372154579,
|
7563 |
+
"grad_norm": 0.029242129996418953,
|
7564 |
+
"learning_rate": 1.985004058305535e-05,
|
7565 |
+
"loss": 10.3337,
|
7566 |
+
"step": 1075
|
7567 |
+
},
|
7568 |
+
{
|
7569 |
+
"epoch": 0.45569084171519325,
|
7570 |
+
"grad_norm": 0.029076050966978073,
|
7571 |
+
"learning_rate": 1.9709853745354313e-05,
|
7572 |
+
"loss": 10.3347,
|
7573 |
+
"step": 1076
|
7574 |
+
},
|
7575 |
+
{
|
7576 |
+
"epoch": 0.4561143462149285,
|
7577 |
+
"grad_norm": 0.039899520576000214,
|
7578 |
+
"learning_rate": 1.9570109546543126e-05,
|
7579 |
+
"loss": 10.3334,
|
7580 |
+
"step": 1077
|
7581 |
+
},
|
7582 |
+
{
|
7583 |
+
"epoch": 0.45653785071466385,
|
7584 |
+
"grad_norm": 0.03501451388001442,
|
7585 |
+
"learning_rate": 1.943080875703045e-05,
|
7586 |
+
"loss": 10.325,
|
7587 |
+
"step": 1078
|
7588 |
+
},
|
7589 |
+
{
|
7590 |
+
"epoch": 0.4569613552143992,
|
7591 |
+
"grad_norm": 0.029382554814219475,
|
7592 |
+
"learning_rate": 1.929195214478028e-05,
|
7593 |
+
"loss": 10.336,
|
7594 |
+
"step": 1079
|
7595 |
+
},
|
7596 |
+
{
|
7597 |
+
"epoch": 0.45738485971413445,
|
7598 |
+
"grad_norm": 0.03819538280367851,
|
7599 |
+
"learning_rate": 1.915354047530791e-05,
|
7600 |
+
"loss": 10.3329,
|
7601 |
+
"step": 1080
|
7602 |
+
},
|
7603 |
+
{
|
7604 |
+
"epoch": 0.4578083642138698,
|
7605 |
+
"grad_norm": 0.03543626144528389,
|
7606 |
+
"learning_rate": 1.901557451167578e-05,
|
7607 |
+
"loss": 10.3326,
|
7608 |
+
"step": 1081
|
7609 |
+
},
|
7610 |
+
{
|
7611 |
+
"epoch": 0.4582318687136051,
|
7612 |
+
"grad_norm": 0.04363977536559105,
|
7613 |
+
"learning_rate": 1.887805501448896e-05,
|
7614 |
+
"loss": 10.3289,
|
7615 |
+
"step": 1082
|
7616 |
+
},
|
7617 |
+
{
|
7618 |
+
"epoch": 0.4586553732133404,
|
7619 |
+
"grad_norm": 0.03918329253792763,
|
7620 |
+
"learning_rate": 1.8740982741891377e-05,
|
7621 |
+
"loss": 10.3276,
|
7622 |
+
"step": 1083
|
7623 |
+
},
|
7624 |
+
{
|
7625 |
+
"epoch": 0.4590788777130757,
|
7626 |
+
"grad_norm": 0.029666945338249207,
|
7627 |
+
"learning_rate": 1.860435844956121e-05,
|
7628 |
+
"loss": 10.3307,
|
7629 |
+
"step": 1084
|
7630 |
+
},
|
7631 |
+
{
|
7632 |
+
"epoch": 0.45950238221281103,
|
7633 |
+
"grad_norm": 0.035329993814229965,
|
7634 |
+
"learning_rate": 1.8468182890707007e-05,
|
7635 |
+
"loss": 10.3336,
|
7636 |
+
"step": 1085
|
7637 |
+
},
|
7638 |
+
{
|
7639 |
+
"epoch": 0.4599258867125463,
|
7640 |
+
"grad_norm": 0.040378130972385406,
|
7641 |
+
"learning_rate": 1.833245681606356e-05,
|
7642 |
+
"loss": 10.3296,
|
7643 |
+
"step": 1086
|
7644 |
+
},
|
7645 |
+
{
|
7646 |
+
"epoch": 0.46034939121228163,
|
7647 |
+
"grad_norm": 0.04233788326382637,
|
7648 |
+
"learning_rate": 1.8197180973887428e-05,
|
7649 |
+
"loss": 10.3312,
|
7650 |
+
"step": 1087
|
7651 |
+
},
|
7652 |
+
{
|
7653 |
+
"epoch": 0.46077289571201696,
|
7654 |
+
"grad_norm": 0.03670990467071533,
|
7655 |
+
"learning_rate": 1.806235610995327e-05,
|
7656 |
+
"loss": 10.3303,
|
7657 |
+
"step": 1088
|
7658 |
+
},
|
7659 |
+
{
|
7660 |
+
"epoch": 0.46119640021175223,
|
7661 |
+
"grad_norm": 0.03234660625457764,
|
7662 |
+
"learning_rate": 1.7927982967549384e-05,
|
7663 |
+
"loss": 10.3355,
|
7664 |
+
"step": 1089
|
7665 |
+
},
|
7666 |
+
{
|
7667 |
+
"epoch": 0.46161990471148756,
|
7668 |
+
"grad_norm": 0.042892660945653915,
|
7669 |
+
"learning_rate": 1.7794062287473735e-05,
|
7670 |
+
"loss": 10.331,
|
7671 |
+
"step": 1090
|
7672 |
+
},
|
7673 |
+
{
|
7674 |
+
"epoch": 0.4620434092112229,
|
7675 |
+
"grad_norm": 0.04852224513888359,
|
7676 |
+
"learning_rate": 1.7660594808029908e-05,
|
7677 |
+
"loss": 10.3361,
|
7678 |
+
"step": 1091
|
7679 |
+
},
|
7680 |
+
{
|
7681 |
+
"epoch": 0.46246691371095816,
|
7682 |
+
"grad_norm": 0.036822058260440826,
|
7683 |
+
"learning_rate": 1.7527581265022965e-05,
|
7684 |
+
"loss": 10.3364,
|
7685 |
+
"step": 1092
|
7686 |
+
},
|
7687 |
+
{
|
7688 |
+
"epoch": 0.4628904182106935,
|
7689 |
+
"grad_norm": 0.03043217770755291,
|
7690 |
+
"learning_rate": 1.7395022391755434e-05,
|
7691 |
+
"loss": 10.335,
|
7692 |
+
"step": 1093
|
7693 |
+
},
|
7694 |
+
{
|
7695 |
+
"epoch": 0.4633139227104288,
|
7696 |
+
"grad_norm": 0.027736082673072815,
|
7697 |
+
"learning_rate": 1.7262918919023243e-05,
|
7698 |
+
"loss": 10.3335,
|
7699 |
+
"step": 1094
|
7700 |
+
},
|
7701 |
+
{
|
7702 |
+
"epoch": 0.4637374272101641,
|
7703 |
+
"grad_norm": 0.03186174854636192,
|
7704 |
+
"learning_rate": 1.713127157511172e-05,
|
7705 |
+
"loss": 10.3365,
|
7706 |
+
"step": 1095
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 0.4641609317098994,
|
7710 |
+
"grad_norm": 0.03788574039936066,
|
7711 |
+
"learning_rate": 1.700008108579154e-05,
|
7712 |
+
"loss": 10.3317,
|
7713 |
+
"step": 1096
|
7714 |
+
},
|
7715 |
+
{
|
7716 |
+
"epoch": 0.46458443620963474,
|
7717 |
+
"grad_norm": 0.047464434057474136,
|
7718 |
+
"learning_rate": 1.6869348174314738e-05,
|
7719 |
+
"loss": 10.3307,
|
7720 |
+
"step": 1097
|
7721 |
+
},
|
7722 |
+
{
|
7723 |
+
"epoch": 0.46500794070937,
|
7724 |
+
"grad_norm": 0.03223862871527672,
|
7725 |
+
"learning_rate": 1.673907356141079e-05,
|
7726 |
+
"loss": 10.3337,
|
7727 |
+
"step": 1098
|
7728 |
+
},
|
7729 |
+
{
|
7730 |
+
"epoch": 0.46543144520910534,
|
7731 |
+
"grad_norm": 0.02775878831744194,
|
7732 |
+
"learning_rate": 1.6609257965282453e-05,
|
7733 |
+
"loss": 10.3376,
|
7734 |
+
"step": 1099
|
7735 |
+
},
|
7736 |
+
{
|
7737 |
+
"epoch": 0.46585494970884067,
|
7738 |
+
"grad_norm": 0.0346621610224247,
|
7739 |
+
"learning_rate": 1.647990210160204e-05,
|
7740 |
+
"loss": 10.334,
|
7741 |
+
"step": 1100
|
7742 |
+
},
|
7743 |
+
{
|
7744 |
+
"epoch": 0.46627845420857594,
|
7745 |
+
"grad_norm": 0.03867461159825325,
|
7746 |
+
"learning_rate": 1.6351006683507297e-05,
|
7747 |
+
"loss": 10.3321,
|
7748 |
+
"step": 1101
|
7749 |
+
},
|
7750 |
+
{
|
7751 |
+
"epoch": 0.46670195870831127,
|
7752 |
+
"grad_norm": 0.033736009150743484,
|
7753 |
+
"learning_rate": 1.622257242159756e-05,
|
7754 |
+
"loss": 10.329,
|
7755 |
+
"step": 1102
|
7756 |
+
},
|
7757 |
+
{
|
7758 |
+
"epoch": 0.4671254632080466,
|
7759 |
+
"grad_norm": 0.03446945920586586,
|
7760 |
+
"learning_rate": 1.6094600023929884e-05,
|
7761 |
+
"loss": 10.3281,
|
7762 |
+
"step": 1103
|
7763 |
+
},
|
7764 |
+
{
|
7765 |
+
"epoch": 0.46754896770778187,
|
7766 |
+
"grad_norm": 0.03439204394817352,
|
7767 |
+
"learning_rate": 1.59670901960149e-05,
|
7768 |
+
"loss": 10.3339,
|
7769 |
+
"step": 1104
|
7770 |
+
},
|
7771 |
+
{
|
7772 |
+
"epoch": 0.4679724722075172,
|
7773 |
+
"grad_norm": 0.03250345215201378,
|
7774 |
+
"learning_rate": 1.5840043640813274e-05,
|
7775 |
+
"loss": 10.3308,
|
7776 |
+
"step": 1105
|
7777 |
+
},
|
7778 |
+
{
|
7779 |
+
"epoch": 0.4683959767072525,
|
7780 |
+
"grad_norm": 0.030219173058867455,
|
7781 |
+
"learning_rate": 1.5713461058731572e-05,
|
7782 |
+
"loss": 10.333,
|
7783 |
+
"step": 1106
|
7784 |
+
},
|
7785 |
+
{
|
7786 |
+
"epoch": 0.4688194812069878,
|
7787 |
+
"grad_norm": 0.031828220933675766,
|
7788 |
+
"learning_rate": 1.558734314761844e-05,
|
7789 |
+
"loss": 10.3353,
|
7790 |
+
"step": 1107
|
7791 |
+
},
|
7792 |
+
{
|
7793 |
+
"epoch": 0.4692429857067231,
|
7794 |
+
"grad_norm": 0.047410812228918076,
|
7795 |
+
"learning_rate": 1.546169060276088e-05,
|
7796 |
+
"loss": 10.3289,
|
7797 |
+
"step": 1108
|
7798 |
+
},
|
7799 |
+
{
|
7800 |
+
"epoch": 0.46966649020645845,
|
7801 |
+
"grad_norm": 0.036803584545850754,
|
7802 |
+
"learning_rate": 1.53365041168803e-05,
|
7803 |
+
"loss": 10.3358,
|
7804 |
+
"step": 1109
|
7805 |
+
},
|
7806 |
+
{
|
7807 |
+
"epoch": 0.4700899947061937,
|
7808 |
+
"grad_norm": 0.03534479811787605,
|
7809 |
+
"learning_rate": 1.5211784380128714e-05,
|
7810 |
+
"loss": 10.33,
|
7811 |
+
"step": 1110
|
7812 |
+
},
|
7813 |
+
{
|
7814 |
+
"epoch": 0.47051349920592905,
|
7815 |
+
"grad_norm": 0.036183904856443405,
|
7816 |
+
"learning_rate": 1.5087532080084976e-05,
|
7817 |
+
"loss": 10.3289,
|
7818 |
+
"step": 1111
|
7819 |
+
},
|
7820 |
+
{
|
7821 |
+
"epoch": 0.4709370037056644,
|
7822 |
+
"grad_norm": 0.033738043159246445,
|
7823 |
+
"learning_rate": 1.4963747901750936e-05,
|
7824 |
+
"loss": 10.3303,
|
7825 |
+
"step": 1112
|
7826 |
+
},
|
7827 |
+
{
|
7828 |
+
"epoch": 0.4713605082053997,
|
7829 |
+
"grad_norm": 0.03870893269777298,
|
7830 |
+
"learning_rate": 1.4840432527547732e-05,
|
7831 |
+
"loss": 10.3364,
|
7832 |
+
"step": 1113
|
7833 |
+
},
|
7834 |
+
{
|
7835 |
+
"epoch": 0.471784012705135,
|
7836 |
+
"grad_norm": 0.04043989256024361,
|
7837 |
+
"learning_rate": 1.4717586637311943e-05,
|
7838 |
+
"loss": 10.3316,
|
7839 |
+
"step": 1114
|
7840 |
+
},
|
7841 |
+
{
|
7842 |
+
"epoch": 0.4722075172048703,
|
7843 |
+
"grad_norm": 0.03024929389357567,
|
7844 |
+
"learning_rate": 1.4595210908291935e-05,
|
7845 |
+
"loss": 10.3364,
|
7846 |
+
"step": 1115
|
7847 |
+
},
|
7848 |
+
{
|
7849 |
+
"epoch": 0.47263102170460564,
|
7850 |
+
"grad_norm": 0.04411826282739639,
|
7851 |
+
"learning_rate": 1.447330601514405e-05,
|
7852 |
+
"loss": 10.3331,
|
7853 |
+
"step": 1116
|
7854 |
+
},
|
7855 |
+
{
|
7856 |
+
"epoch": 0.4730545262043409,
|
7857 |
+
"grad_norm": 0.03368929401040077,
|
7858 |
+
"learning_rate": 1.4351872629928908e-05,
|
7859 |
+
"loss": 10.3323,
|
7860 |
+
"step": 1117
|
7861 |
+
},
|
7862 |
+
{
|
7863 |
+
"epoch": 0.47347803070407624,
|
7864 |
+
"grad_norm": 0.038087401539087296,
|
7865 |
+
"learning_rate": 1.423091142210774e-05,
|
7866 |
+
"loss": 10.3295,
|
7867 |
+
"step": 1118
|
7868 |
+
},
|
7869 |
+
{
|
7870 |
+
"epoch": 0.47390153520381156,
|
7871 |
+
"grad_norm": 0.03507355973124504,
|
7872 |
+
"learning_rate": 1.4110423058538624e-05,
|
7873 |
+
"loss": 10.3273,
|
7874 |
+
"step": 1119
|
7875 |
+
},
|
7876 |
+
{
|
7877 |
+
"epoch": 0.47432503970354684,
|
7878 |
+
"grad_norm": 0.03440206125378609,
|
7879 |
+
"learning_rate": 1.3990408203472938e-05,
|
7880 |
+
"loss": 10.3336,
|
7881 |
+
"step": 1120
|
7882 |
+
},
|
7883 |
+
{
|
7884 |
+
"epoch": 0.47474854420328216,
|
7885 |
+
"grad_norm": 0.03201809525489807,
|
7886 |
+
"learning_rate": 1.387086751855149e-05,
|
7887 |
+
"loss": 10.3323,
|
7888 |
+
"step": 1121
|
7889 |
+
},
|
7890 |
+
{
|
7891 |
+
"epoch": 0.4751720487030175,
|
7892 |
+
"grad_norm": 0.02803219109773636,
|
7893 |
+
"learning_rate": 1.3751801662801056e-05,
|
7894 |
+
"loss": 10.3343,
|
7895 |
+
"step": 1122
|
7896 |
+
},
|
7897 |
+
{
|
7898 |
+
"epoch": 0.47559555320275276,
|
7899 |
+
"grad_norm": 0.03642897307872772,
|
7900 |
+
"learning_rate": 1.3633211292630742e-05,
|
7901 |
+
"loss": 10.3309,
|
7902 |
+
"step": 1123
|
7903 |
+
},
|
7904 |
+
{
|
7905 |
+
"epoch": 0.4760190577024881,
|
7906 |
+
"grad_norm": 0.04547721892595291,
|
7907 |
+
"learning_rate": 1.3515097061828164e-05,
|
7908 |
+
"loss": 10.3248,
|
7909 |
+
"step": 1124
|
7910 |
+
},
|
7911 |
+
{
|
7912 |
+
"epoch": 0.4764425622022234,
|
7913 |
+
"grad_norm": 0.03152972459793091,
|
7914 |
+
"learning_rate": 1.339745962155613e-05,
|
7915 |
+
"loss": 10.3396,
|
7916 |
+
"step": 1125
|
7917 |
+
},
|
7918 |
+
{
|
7919 |
+
"epoch": 0.4768660667019587,
|
7920 |
+
"grad_norm": 0.028171587735414505,
|
7921 |
+
"learning_rate": 1.3280299620348846e-05,
|
7922 |
+
"loss": 10.33,
|
7923 |
+
"step": 1126
|
7924 |
+
},
|
7925 |
+
{
|
7926 |
+
"epoch": 0.477289571201694,
|
7927 |
+
"grad_norm": 0.03410959243774414,
|
7928 |
+
"learning_rate": 1.3163617704108321e-05,
|
7929 |
+
"loss": 10.3344,
|
7930 |
+
"step": 1127
|
7931 |
+
},
|
7932 |
+
{
|
7933 |
+
"epoch": 0.47771307570142935,
|
7934 |
+
"grad_norm": 0.030304502695798874,
|
7935 |
+
"learning_rate": 1.304741451610103e-05,
|
7936 |
+
"loss": 10.3309,
|
7937 |
+
"step": 1128
|
7938 |
+
},
|
7939 |
+
{
|
7940 |
+
"epoch": 0.4781365802011646,
|
7941 |
+
"grad_norm": 0.03257643058896065,
|
7942 |
+
"learning_rate": 1.2931690696954135e-05,
|
7943 |
+
"loss": 10.3346,
|
7944 |
+
"step": 1129
|
7945 |
+
},
|
7946 |
+
{
|
7947 |
+
"epoch": 0.47856008470089995,
|
7948 |
+
"grad_norm": 0.04555933550000191,
|
7949 |
+
"learning_rate": 1.2816446884652066e-05,
|
7950 |
+
"loss": 10.3302,
|
7951 |
+
"step": 1130
|
7952 |
+
},
|
7953 |
+
{
|
7954 |
+
"epoch": 0.4789835892006353,
|
7955 |
+
"grad_norm": 0.0384778194129467,
|
7956 |
+
"learning_rate": 1.2701683714532975e-05,
|
7957 |
+
"loss": 10.3317,
|
7958 |
+
"step": 1131
|
7959 |
+
},
|
7960 |
+
{
|
7961 |
+
"epoch": 0.47940709370037055,
|
7962 |
+
"grad_norm": 0.03637570142745972,
|
7963 |
+
"learning_rate": 1.2587401819285239e-05,
|
7964 |
+
"loss": 10.3295,
|
7965 |
+
"step": 1132
|
7966 |
+
},
|
7967 |
+
{
|
7968 |
+
"epoch": 0.4798305982001059,
|
7969 |
+
"grad_norm": 0.04053565487265587,
|
7970 |
+
"learning_rate": 1.2473601828943949e-05,
|
7971 |
+
"loss": 10.3293,
|
7972 |
+
"step": 1133
|
7973 |
+
},
|
7974 |
+
{
|
7975 |
+
"epoch": 0.4802541026998412,
|
7976 |
+
"grad_norm": 0.042270079255104065,
|
7977 |
+
"learning_rate": 1.236028437088751e-05,
|
7978 |
+
"loss": 10.3271,
|
7979 |
+
"step": 1134
|
7980 |
+
},
|
7981 |
+
{
|
7982 |
+
"epoch": 0.4806776071995765,
|
7983 |
+
"grad_norm": 0.04081670939922333,
|
7984 |
+
"learning_rate": 1.2247450069834076e-05,
|
7985 |
+
"loss": 10.3365,
|
7986 |
+
"step": 1135
|
7987 |
+
},
|
7988 |
+
{
|
7989 |
+
"epoch": 0.4811011116993118,
|
7990 |
+
"grad_norm": 0.03796311840415001,
|
7991 |
+
"learning_rate": 1.2135099547838192e-05,
|
7992 |
+
"loss": 10.333,
|
7993 |
+
"step": 1136
|
7994 |
+
},
|
7995 |
+
{
|
7996 |
+
"epoch": 0.48152461619904713,
|
7997 |
+
"grad_norm": 0.02851458452641964,
|
7998 |
+
"learning_rate": 1.2023233424287328e-05,
|
7999 |
+
"loss": 10.3304,
|
8000 |
+
"step": 1137
|
8001 |
+
},
|
8002 |
+
{
|
8003 |
+
"epoch": 0.4819481206987824,
|
8004 |
+
"grad_norm": 0.03447718173265457,
|
8005 |
+
"learning_rate": 1.1911852315898463e-05,
|
8006 |
+
"loss": 10.3316,
|
8007 |
+
"step": 1138
|
8008 |
+
},
|
8009 |
+
{
|
8010 |
+
"epoch": 0.48237162519851773,
|
8011 |
+
"grad_norm": 0.037812747061252594,
|
8012 |
+
"learning_rate": 1.1800956836714682e-05,
|
8013 |
+
"loss": 10.3288,
|
8014 |
+
"step": 1139
|
8015 |
+
},
|
8016 |
+
{
|
8017 |
+
"epoch": 0.48279512969825306,
|
8018 |
+
"grad_norm": 0.03977108374238014,
|
8019 |
+
"learning_rate": 1.1690547598101864e-05,
|
8020 |
+
"loss": 10.3303,
|
8021 |
+
"step": 1140
|
8022 |
+
},
|
8023 |
+
{
|
8024 |
+
"epoch": 0.48321863419798833,
|
8025 |
+
"grad_norm": 0.031228644773364067,
|
8026 |
+
"learning_rate": 1.1580625208745145e-05,
|
8027 |
+
"loss": 10.3294,
|
8028 |
+
"step": 1141
|
8029 |
+
},
|
8030 |
+
{
|
8031 |
+
"epoch": 0.48364213869772366,
|
8032 |
+
"grad_norm": 0.0270911306142807,
|
8033 |
+
"learning_rate": 1.1471190274645704e-05,
|
8034 |
+
"loss": 10.3322,
|
8035 |
+
"step": 1142
|
8036 |
+
},
|
8037 |
+
{
|
8038 |
+
"epoch": 0.484065643197459,
|
8039 |
+
"grad_norm": 0.03246387094259262,
|
8040 |
+
"learning_rate": 1.1362243399117478e-05,
|
8041 |
+
"loss": 10.3306,
|
8042 |
+
"step": 1143
|
8043 |
+
},
|
8044 |
+
{
|
8045 |
+
"epoch": 0.48448914769719426,
|
8046 |
+
"grad_norm": 0.03161618486046791,
|
8047 |
+
"learning_rate": 1.1253785182783572e-05,
|
8048 |
+
"loss": 10.335,
|
8049 |
+
"step": 1144
|
8050 |
+
},
|
8051 |
+
{
|
8052 |
+
"epoch": 0.4849126521969296,
|
8053 |
+
"grad_norm": 0.03287721052765846,
|
8054 |
+
"learning_rate": 1.1145816223573259e-05,
|
8055 |
+
"loss": 10.3312,
|
8056 |
+
"step": 1145
|
8057 |
+
},
|
8058 |
+
{
|
8059 |
+
"epoch": 0.4853361566966649,
|
8060 |
+
"grad_norm": 0.029835056513547897,
|
8061 |
+
"learning_rate": 1.1038337116718467e-05,
|
8062 |
+
"loss": 10.3309,
|
8063 |
+
"step": 1146
|
8064 |
+
},
|
8065 |
+
{
|
8066 |
+
"epoch": 0.48575966119640024,
|
8067 |
+
"grad_norm": 0.03465202450752258,
|
8068 |
+
"learning_rate": 1.0931348454750601e-05,
|
8069 |
+
"loss": 10.3336,
|
8070 |
+
"step": 1147
|
8071 |
+
},
|
8072 |
+
{
|
8073 |
+
"epoch": 0.4861831656961355,
|
8074 |
+
"grad_norm": 0.03778757527470589,
|
8075 |
+
"learning_rate": 1.0824850827497246e-05,
|
8076 |
+
"loss": 10.3342,
|
8077 |
+
"step": 1148
|
8078 |
+
},
|
8079 |
+
{
|
8080 |
+
"epoch": 0.48660667019587084,
|
8081 |
+
"grad_norm": 0.03788898512721062,
|
8082 |
+
"learning_rate": 1.07188448220789e-05,
|
8083 |
+
"loss": 10.3338,
|
8084 |
+
"step": 1149
|
8085 |
+
},
|
8086 |
+
{
|
8087 |
+
"epoch": 0.48703017469560617,
|
8088 |
+
"grad_norm": 0.03392605856060982,
|
8089 |
+
"learning_rate": 1.061333102290576e-05,
|
8090 |
+
"loss": 10.3314,
|
8091 |
+
"step": 1150
|
8092 |
+
},
|
8093 |
+
{
|
8094 |
+
"epoch": 0.48745367919534144,
|
8095 |
+
"grad_norm": 0.03181210905313492,
|
8096 |
+
"learning_rate": 1.0508310011674516e-05,
|
8097 |
+
"loss": 10.3347,
|
8098 |
+
"step": 1151
|
8099 |
+
},
|
8100 |
+
{
|
8101 |
+
"epoch": 0.48787718369507677,
|
8102 |
+
"grad_norm": 0.03807486966252327,
|
8103 |
+
"learning_rate": 1.0403782367365088e-05,
|
8104 |
+
"loss": 10.3334,
|
8105 |
+
"step": 1152
|
8106 |
+
},
|
8107 |
+
{
|
8108 |
+
"epoch": 0.4883006881948121,
|
8109 |
+
"grad_norm": 0.04221343249082565,
|
8110 |
+
"learning_rate": 1.0299748666237485e-05,
|
8111 |
+
"loss": 10.33,
|
8112 |
+
"step": 1153
|
8113 |
+
},
|
8114 |
+
{
|
8115 |
+
"epoch": 0.48872419269454737,
|
8116 |
+
"grad_norm": 0.03662874549627304,
|
8117 |
+
"learning_rate": 1.0196209481828633e-05,
|
8118 |
+
"loss": 10.3337,
|
8119 |
+
"step": 1154
|
8120 |
+
},
|
8121 |
+
{
|
8122 |
+
"epoch": 0.4891476971942827,
|
8123 |
+
"grad_norm": 0.03761863335967064,
|
8124 |
+
"learning_rate": 1.0093165384949155e-05,
|
8125 |
+
"loss": 10.3363,
|
8126 |
+
"step": 1155
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 0.489571201694018,
|
8130 |
+
"grad_norm": 0.03691156208515167,
|
8131 |
+
"learning_rate": 9.990616943680265e-06,
|
8132 |
+
"loss": 10.3355,
|
8133 |
+
"step": 1156
|
8134 |
+
},
|
8135 |
+
{
|
8136 |
+
"epoch": 0.4899947061937533,
|
8137 |
+
"grad_norm": 0.03406470641493797,
|
8138 |
+
"learning_rate": 9.888564723370664e-06,
|
8139 |
+
"loss": 10.3348,
|
8140 |
+
"step": 1157
|
8141 |
+
},
|
8142 |
+
{
|
8143 |
+
"epoch": 0.4904182106934886,
|
8144 |
+
"grad_norm": 0.03452722728252411,
|
8145 |
+
"learning_rate": 9.787009286633363e-06,
|
8146 |
+
"loss": 10.3332,
|
8147 |
+
"step": 1158
|
8148 |
+
},
|
8149 |
+
{
|
8150 |
+
"epoch": 0.49084171519322395,
|
8151 |
+
"grad_norm": 0.03500404581427574,
|
8152 |
+
"learning_rate": 9.685951193342602e-06,
|
8153 |
+
"loss": 10.3328,
|
8154 |
+
"step": 1159
|
8155 |
+
},
|
8156 |
+
{
|
8157 |
+
"epoch": 0.4912652196929592,
|
8158 |
+
"grad_norm": 0.034697335213422775,
|
8159 |
+
"learning_rate": 9.585391000630828e-06,
|
8160 |
+
"loss": 10.3292,
|
8161 |
+
"step": 1160
|
8162 |
+
},
|
8163 |
+
{
|
8164 |
+
"epoch": 0.49168872419269455,
|
8165 |
+
"grad_norm": 0.028287572786211967,
|
8166 |
+
"learning_rate": 9.485329262885457e-06,
|
8167 |
+
"loss": 10.3337,
|
8168 |
+
"step": 1161
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 0.4921122286924299,
|
8172 |
+
"grad_norm": 0.0407349169254303,
|
8173 |
+
"learning_rate": 9.385766531746054e-06,
|
8174 |
+
"loss": 10.3314,
|
8175 |
+
"step": 1162
|
8176 |
+
},
|
8177 |
+
{
|
8178 |
+
"epoch": 0.49253573319216515,
|
8179 |
+
"grad_norm": 0.03521955758333206,
|
8180 |
+
"learning_rate": 9.28670335610109e-06,
|
8181 |
+
"loss": 10.3313,
|
8182 |
+
"step": 1163
|
8183 |
+
},
|
8184 |
+
{
|
8185 |
+
"epoch": 0.4929592376919005,
|
8186 |
+
"grad_norm": 0.038377124816179276,
|
8187 |
+
"learning_rate": 9.188140282084967e-06,
|
8188 |
+
"loss": 10.3295,
|
8189 |
+
"step": 1164
|
8190 |
+
},
|
8191 |
+
{
|
8192 |
+
"epoch": 0.4933827421916358,
|
8193 |
+
"grad_norm": 0.037929970771074295,
|
8194 |
+
"learning_rate": 9.090077853075118e-06,
|
8195 |
+
"loss": 10.331,
|
8196 |
+
"step": 1165
|
8197 |
+
},
|
8198 |
+
{
|
8199 |
+
"epoch": 0.4938062466913711,
|
8200 |
+
"grad_norm": 0.03767012432217598,
|
8201 |
+
"learning_rate": 8.992516609688862e-06,
|
8202 |
+
"loss": 10.3305,
|
8203 |
+
"step": 1166
|
8204 |
+
},
|
8205 |
+
{
|
8206 |
+
"epoch": 0.4942297511911064,
|
8207 |
+
"grad_norm": 0.04114054888486862,
|
8208 |
+
"learning_rate": 8.89545708978049e-06,
|
8209 |
+
"loss": 10.3327,
|
8210 |
+
"step": 1167
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 0.49465325569084173,
|
8214 |
+
"grad_norm": 0.03139737620949745,
|
8215 |
+
"learning_rate": 8.798899828438333e-06,
|
8216 |
+
"loss": 10.3342,
|
8217 |
+
"step": 1168
|
8218 |
+
},
|
8219 |
+
{
|
8220 |
+
"epoch": 0.495076760190577,
|
8221 |
+
"grad_norm": 0.0350373312830925,
|
8222 |
+
"learning_rate": 8.70284535798168e-06,
|
8223 |
+
"loss": 10.3335,
|
8224 |
+
"step": 1169
|
8225 |
+
},
|
8226 |
+
{
|
8227 |
+
"epoch": 0.49550026469031233,
|
8228 |
+
"grad_norm": 0.03645787015557289,
|
8229 |
+
"learning_rate": 8.607294207958073e-06,
|
8230 |
+
"loss": 10.3285,
|
8231 |
+
"step": 1170
|
8232 |
+
},
|
8233 |
+
{
|
8234 |
+
"epoch": 0.49592376919004766,
|
8235 |
+
"grad_norm": 0.04092005640268326,
|
8236 |
+
"learning_rate": 8.512246905140165e-06,
|
8237 |
+
"loss": 10.332,
|
8238 |
+
"step": 1171
|
8239 |
+
},
|
8240 |
+
{
|
8241 |
+
"epoch": 0.49634727368978293,
|
8242 |
+
"grad_norm": 0.03972132131457329,
|
8243 |
+
"learning_rate": 8.417703973522917e-06,
|
8244 |
+
"loss": 10.3336,
|
8245 |
+
"step": 1172
|
8246 |
+
},
|
8247 |
+
{
|
8248 |
+
"epoch": 0.49677077818951826,
|
8249 |
+
"grad_norm": 0.02949652262032032,
|
8250 |
+
"learning_rate": 8.323665934320713e-06,
|
8251 |
+
"loss": 10.3329,
|
8252 |
+
"step": 1173
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 0.4971942826892536,
|
8256 |
+
"grad_norm": 0.04814364016056061,
|
8257 |
+
"learning_rate": 8.23013330596445e-06,
|
8258 |
+
"loss": 10.3317,
|
8259 |
+
"step": 1174
|
8260 |
+
},
|
8261 |
+
{
|
8262 |
+
"epoch": 0.49761778718898886,
|
8263 |
+
"grad_norm": 0.0334940031170845,
|
8264 |
+
"learning_rate": 8.13710660409871e-06,
|
8265 |
+
"loss": 10.3367,
|
8266 |
+
"step": 1175
|
8267 |
+
},
|
8268 |
+
{
|
8269 |
+
"epoch": 0.4980412916887242,
|
8270 |
+
"grad_norm": 0.03809863701462746,
|
8271 |
+
"learning_rate": 8.044586341578886e-06,
|
8272 |
+
"loss": 10.3347,
|
8273 |
+
"step": 1176
|
8274 |
+
},
|
8275 |
+
{
|
8276 |
+
"epoch": 0.4984647961884595,
|
8277 |
+
"grad_norm": 0.03746895492076874,
|
8278 |
+
"learning_rate": 7.952573028468457e-06,
|
8279 |
+
"loss": 10.3362,
|
8280 |
+
"step": 1177
|
8281 |
+
},
|
8282 |
+
{
|
8283 |
+
"epoch": 0.4988883006881948,
|
8284 |
+
"grad_norm": 0.024187074974179268,
|
8285 |
+
"learning_rate": 7.861067172035962e-06,
|
8286 |
+
"loss": 10.3327,
|
8287 |
+
"step": 1178
|
8288 |
+
},
|
8289 |
+
{
|
8290 |
+
"epoch": 0.4993118051879301,
|
8291 |
+
"grad_norm": 0.03394331783056259,
|
8292 |
+
"learning_rate": 7.770069276752422e-06,
|
8293 |
+
"loss": 10.3268,
|
8294 |
+
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.49973530968766544,
|
8298 |
+
"grad_norm": 0.0327443964779377,
|
8299 |
+
"learning_rate": 7.679579844288509e-06,
|
8300 |
+
"loss": 10.332,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.5001588141874007,
|
8305 |
+
"grad_norm": 0.027774417772889137,
|
8306 |
+
"learning_rate": 7.589599373511602e-06,
|
8307 |
+
"loss": 10.329,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.5005823186871361,
|
8312 |
+
"grad_norm": 0.03464759886264801,
|
8313 |
+
"learning_rate": 7.500128360483338e-06,
|
8314 |
+
"loss": 10.3334,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.5010058231868714,
|
8319 |
+
"grad_norm": 0.03733719512820244,
|
8320 |
+
"learning_rate": 7.411167298456634e-06,
|
8321 |
+
"loss": 10.3307,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.5014293276866066,
|
8326 |
+
"grad_norm": 0.033785175532102585,
|
8327 |
+
"learning_rate": 7.32271667787302e-06,
|
8328 |
+
"loss": 10.3362,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.501852832186342,
|
8333 |
+
"grad_norm": 0.038209252059459686,
|
8334 |
+
"learning_rate": 7.234776986360059e-06,
|
8335 |
+
"loss": 10.3309,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.5022763366860773,
|
8340 |
+
"grad_norm": 0.03651139885187149,
|
8341 |
+
"learning_rate": 7.147348708728507e-06,
|
8342 |
+
"loss": 10.335,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.5026998411858126,
|
8347 |
+
"grad_norm": 0.03249209746718407,
|
8348 |
+
"learning_rate": 7.060432326969713e-06,
|
8349 |
+
"loss": 10.3326,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.503123345685548,
|
8354 |
+
"grad_norm": 0.049712520092725754,
|
8355 |
+
"learning_rate": 6.974028320252934e-06,
|
8356 |
+
"loss": 10.3269,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.5035468501852832,
|
8361 |
+
"grad_norm": 0.03345096856355667,
|
8362 |
+
"learning_rate": 6.888137164922725e-06,
|
8363 |
+
"loss": 10.3273,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.5039703546850185,
|
8368 |
+
"grad_norm": 0.028842521831393242,
|
8369 |
+
"learning_rate": 6.802759334496289e-06,
|
8370 |
+
"loss": 10.3299,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.5043938591847539,
|
8375 |
+
"grad_norm": 0.02980581857264042,
|
8376 |
+
"learning_rate": 6.717895299660892e-06,
|
8377 |
+
"loss": 10.3337,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.5048173636844892,
|
8382 |
+
"grad_norm": 0.032008688896894455,
|
8383 |
+
"learning_rate": 6.633545528271212e-06,
|
8384 |
+
"loss": 10.3275,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.5052408681842244,
|
8389 |
+
"grad_norm": 0.03007701225578785,
|
8390 |
+
"learning_rate": 6.549710485346827e-06,
|
8391 |
+
"loss": 10.3319,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.5056643726839598,
|
8396 |
+
"grad_norm": 0.03393697366118431,
|
8397 |
+
"learning_rate": 6.466390633069608e-06,
|
8398 |
+
"loss": 10.3292,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.5060878771836951,
|
8403 |
+
"grad_norm": 0.04486103355884552,
|
8404 |
+
"learning_rate": 6.383586430781197e-06,
|
8405 |
+
"loss": 10.3289,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.5065113816834304,
|
8410 |
+
"grad_norm": 0.03052888996899128,
|
8411 |
+
"learning_rate": 6.301298334980421e-06,
|
8412 |
+
"loss": 10.3374,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.5069348861831657,
|
8417 |
+
"grad_norm": 0.030694812536239624,
|
8418 |
+
"learning_rate": 6.219526799320919e-06,
|
8419 |
+
"loss": 10.3308,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.507358390682901,
|
8424 |
+
"grad_norm": 0.03446760028600693,
|
8425 |
+
"learning_rate": 6.138272274608403e-06,
|
8426 |
+
"loss": 10.3346,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.5077818951826363,
|
8431 |
+
"grad_norm": 0.033587660640478134,
|
8432 |
+
"learning_rate": 6.057535208798371e-06,
|
8433 |
+
"loss": 10.3337,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.5082053996823717,
|
8438 |
+
"grad_norm": 0.03484556823968887,
|
8439 |
+
"learning_rate": 5.977316046993642e-06,
|
8440 |
+
"loss": 10.3311,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.5086289041821069,
|
8445 |
+
"grad_norm": 0.03142661601305008,
|
8446 |
+
"learning_rate": 5.897615231441689e-06,
|
8447 |
+
"loss": 10.3335,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.5090524086818422,
|
8452 |
+
"grad_norm": 0.03492956608533859,
|
8453 |
+
"learning_rate": 5.81843320153248e-06,
|
8454 |
+
"loss": 10.3298,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.5094759131815776,
|
8459 |
+
"grad_norm": 0.035875819623470306,
|
8460 |
+
"learning_rate": 5.739770393795851e-06,
|
8461 |
+
"loss": 10.3339,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.5098994176813129,
|
8466 |
+
"grad_norm": 0.028575167059898376,
|
8467 |
+
"learning_rate": 5.6616272418991926e-06,
|
8468 |
+
"loss": 10.3306,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.5103229221810481,
|
8473 |
+
"grad_norm": 0.034280769526958466,
|
8474 |
+
"learning_rate": 5.584004176645052e-06,
|
8475 |
+
"loss": 10.3339,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.5107464266807835,
|
8480 |
+
"grad_norm": 0.03369034081697464,
|
8481 |
+
"learning_rate": 5.5069016259686635e-06,
|
8482 |
+
"loss": 10.3293,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.5111699311805188,
|
8487 |
+
"grad_norm": 0.03932506591081619,
|
8488 |
+
"learning_rate": 5.430320014935797e-06,
|
8489 |
+
"loss": 10.3339,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.5115934356802541,
|
8494 |
+
"grad_norm": 0.04464678466320038,
|
8495 |
+
"learning_rate": 5.354259765740177e-06,
|
8496 |
+
"loss": 10.3316,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.5120169401799894,
|
8501 |
+
"grad_norm": 0.033909354358911514,
|
8502 |
+
"learning_rate": 5.278721297701339e-06,
|
8503 |
+
"loss": 10.3317,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.5124404446797247,
|
8508 |
+
"grad_norm": 0.02771197073161602,
|
8509 |
+
"learning_rate": 5.203705027262184e-06,
|
8510 |
+
"loss": 10.3337,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.51286394917946,
|
8515 |
+
"grad_norm": 0.03711957111954689,
|
8516 |
+
"learning_rate": 5.129211367986786e-06,
|
8517 |
+
"loss": 10.3374,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.5132874536791954,
|
8522 |
+
"grad_norm": 0.04035378247499466,
|
8523 |
+
"learning_rate": 5.055240730558042e-06,
|
8524 |
+
"loss": 10.3278,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.5137109581789306,
|
8529 |
+
"grad_norm": 0.037376079708337784,
|
8530 |
+
"learning_rate": 4.981793522775457e-06,
|
8531 |
+
"loss": 10.3354,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.5141344626786659,
|
8536 |
+
"grad_norm": 0.033283621072769165,
|
8537 |
+
"learning_rate": 4.908870149552835e-06,
|
8538 |
+
"loss": 10.3304,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.5145579671784013,
|
8543 |
+
"grad_norm": 0.04279647022485733,
|
8544 |
+
"learning_rate": 4.836471012916144e-06,
|
8545 |
+
"loss": 10.3317,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.5149814716781366,
|
8550 |
+
"grad_norm": 0.026392200961709023,
|
8551 |
+
"learning_rate": 4.764596512001162e-06,
|
8552 |
+
"loss": 10.3338,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.5154049761778718,
|
8557 |
+
"grad_norm": 0.038188233971595764,
|
8558 |
+
"learning_rate": 4.693247043051441e-06,
|
8559 |
+
"loss": 10.3363,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.5158284806776072,
|
8564 |
+
"grad_norm": 0.03593307361006737,
|
8565 |
+
"learning_rate": 4.622422999415965e-06,
|
8566 |
+
"loss": 10.3302,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.5162519851773425,
|
8571 |
+
"grad_norm": 0.03967192396521568,
|
8572 |
+
"learning_rate": 4.5521247715470945e-06,
|
8573 |
+
"loss": 10.33,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.5166754896770778,
|
8578 |
+
"grad_norm": 0.0491623692214489,
|
8579 |
+
"learning_rate": 4.482352746998364e-06,
|
8580 |
+
"loss": 10.3386,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.5170989941768132,
|
8585 |
+
"grad_norm": 0.0371236614882946,
|
8586 |
+
"learning_rate": 4.413107310422326e-06,
|
8587 |
+
"loss": 10.3336,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.5175224986765484,
|
8592 |
+
"grad_norm": 0.027762679383158684,
|
8593 |
+
"learning_rate": 4.344388843568503e-06,
|
8594 |
+
"loss": 10.3282,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.5179460031762837,
|
8599 |
+
"grad_norm": 0.03931552171707153,
|
8600 |
+
"learning_rate": 4.2761977252811945e-06,
|
8601 |
+
"loss": 10.3331,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.5183695076760191,
|
8606 |
+
"grad_norm": 0.047121018171310425,
|
8607 |
+
"learning_rate": 4.2085343314974715e-06,
|
8608 |
+
"loss": 10.3297,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.5187930121757544,
|
8613 |
+
"grad_norm": 0.042633168399333954,
|
8614 |
+
"learning_rate": 4.141399035245052e-06,
|
8615 |
+
"loss": 10.3337,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.5192165166754896,
|
8620 |
+
"grad_norm": 0.03988894075155258,
|
8621 |
+
"learning_rate": 4.07479220664021e-06,
|
8622 |
+
"loss": 10.3262,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.519640021175225,
|
8627 |
+
"grad_norm": 0.030842246487736702,
|
8628 |
+
"learning_rate": 4.008714212885856e-06,
|
8629 |
+
"loss": 10.3322,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.5200635256749603,
|
8634 |
+
"grad_norm": 0.04261520504951477,
|
8635 |
+
"learning_rate": 3.943165418269401e-06,
|
8636 |
+
"loss": 10.328,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.5204870301746956,
|
8641 |
+
"grad_norm": 0.030063187703490257,
|
8642 |
+
"learning_rate": 3.87814618416078e-06,
|
8643 |
+
"loss": 10.3345,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.5209105346744309,
|
8648 |
+
"grad_norm": 0.030118783935904503,
|
8649 |
+
"learning_rate": 3.8136568690104957e-06,
|
8650 |
+
"loss": 10.3325,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.5213340391741662,
|
8655 |
+
"grad_norm": 0.03795788437128067,
|
8656 |
+
"learning_rate": 3.7496978283475648e-06,
|
8657 |
+
"loss": 10.3327,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.5217575436739015,
|
8662 |
+
"grad_norm": 0.036961231380701065,
|
8663 |
+
"learning_rate": 3.686269414777643e-06,
|
8664 |
+
"loss": 10.3344,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.5221810481736369,
|
8669 |
+
"grad_norm": 0.0403430350124836,
|
8670 |
+
"learning_rate": 3.623371977981027e-06,
|
8671 |
+
"loss": 10.3324,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.5226045526733721,
|
8676 |
+
"grad_norm": 0.03135257214307785,
|
8677 |
+
"learning_rate": 3.5610058647107538e-06,
|
8678 |
+
"loss": 10.3319,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.5230280571731075,
|
8683 |
+
"grad_norm": 0.0364365391433239,
|
8684 |
+
"learning_rate": 3.499171418790681e-06,
|
8685 |
+
"loss": 10.3343,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.5234515616728428,
|
8690 |
+
"grad_norm": 0.025732390582561493,
|
8691 |
+
"learning_rate": 3.437868981113557e-06,
|
8692 |
+
"loss": 10.3338,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.5238750661725781,
|
8697 |
+
"grad_norm": 0.03495744988322258,
|
8698 |
+
"learning_rate": 3.37709888963923e-06,
|
8699 |
+
"loss": 10.3302,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.5242985706723134,
|
8704 |
+
"grad_norm": 0.032097022980451584,
|
8705 |
+
"learning_rate": 3.3168614793926524e-06,
|
8706 |
+
"loss": 10.3356,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.5247220751720487,
|
8711 |
+
"grad_norm": 0.029357150197029114,
|
8712 |
+
"learning_rate": 3.2571570824621923e-06,
|
8713 |
+
"loss": 10.3304,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.525145579671784,
|
8718 |
+
"grad_norm": 0.03179454430937767,
|
8719 |
+
"learning_rate": 3.197986027997657e-06,
|
8720 |
+
"loss": 10.3311,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.5255690841715194,
|
8725 |
+
"grad_norm": 0.038864728063344955,
|
8726 |
+
"learning_rate": 3.1393486422085618e-06,
|
8727 |
+
"loss": 10.3308,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.5259925886712546,
|
8732 |
+
"grad_norm": 0.027193231508135796,
|
8733 |
+
"learning_rate": 3.08124524836233e-06,
|
8734 |
+
"loss": 10.3314,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.5264160931709899,
|
8739 |
+
"grad_norm": 0.035837847739458084,
|
8740 |
+
"learning_rate": 3.023676166782452e-06,
|
8741 |
+
"loss": 10.3327,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.5268395976707253,
|
8746 |
+
"grad_norm": 0.02682778798043728,
|
8747 |
+
"learning_rate": 2.9666417148468072e-06,
|
8748 |
+
"loss": 10.3325,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.5272631021704606,
|
8753 |
+
"grad_norm": 0.04898487776517868,
|
8754 |
+
"learning_rate": 2.910142206985833e-06,
|
8755 |
+
"loss": 10.3317,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.5276866066701958,
|
8760 |
+
"grad_norm": 0.030211864039301872,
|
8761 |
+
"learning_rate": 2.8541779546808256e-06,
|
8762 |
+
"loss": 10.3292,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.5281101111699312,
|
8767 |
+
"grad_norm": 0.03472064808011055,
|
8768 |
+
"learning_rate": 2.7987492664622307e-06,
|
8769 |
+
"loss": 10.3324,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.5285336156696665,
|
8774 |
+
"grad_norm": 0.03139955550432205,
|
8775 |
+
"learning_rate": 2.743856447907944e-06,
|
8776 |
+
"loss": 10.3309,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.5289571201694018,
|
8781 |
+
"grad_norm": 0.02904195711016655,
|
8782 |
+
"learning_rate": 2.689499801641593e-06,
|
8783 |
+
"loss": 10.332,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.5293806246691372,
|
8788 |
+
"grad_norm": 0.045261383056640625,
|
8789 |
+
"learning_rate": 2.6356796273309116e-06,
|
8790 |
+
"loss": 10.33,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.5298041291688724,
|
8795 |
+
"grad_norm": 0.03183293342590332,
|
8796 |
+
"learning_rate": 2.5823962216860562e-06,
|
8797 |
+
"loss": 10.3297,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.5302276336686077,
|
8802 |
+
"grad_norm": 0.04214952513575554,
|
8803 |
+
"learning_rate": 2.5296498784579846e-06,
|
8804 |
+
"loss": 10.3309,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.5306511381683431,
|
8809 |
+
"grad_norm": 0.03488962724804878,
|
8810 |
+
"learning_rate": 2.4774408884368215e-06,
|
8811 |
+
"loss": 10.3333,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.5310746426680784,
|
8816 |
+
"grad_norm": 0.03279737010598183,
|
8817 |
+
"learning_rate": 2.4257695394503287e-06,
|
8818 |
+
"loss": 10.3278,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.5314981471678136,
|
8823 |
+
"grad_norm": 0.03219415992498398,
|
8824 |
+
"learning_rate": 2.374636116362172e-06,
|
8825 |
+
"loss": 10.3334,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.531921651667549,
|
8830 |
+
"grad_norm": 0.05066683888435364,
|
8831 |
+
"learning_rate": 2.32404090107049e-06,
|
8832 |
+
"loss": 10.3306,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.5323451561672843,
|
8837 |
+
"grad_norm": 0.028979485854506493,
|
8838 |
+
"learning_rate": 2.2739841725062715e-06,
|
8839 |
+
"loss": 10.3319,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.5327686606670196,
|
8844 |
+
"grad_norm": 0.03191670775413513,
|
8845 |
+
"learning_rate": 2.2244662066318146e-06,
|
8846 |
+
"loss": 10.333,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.5331921651667549,
|
8851 |
+
"grad_norm": 0.04911280795931816,
|
8852 |
+
"learning_rate": 2.1754872764392698e-06,
|
8853 |
+
"loss": 10.3313,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.5336156696664902,
|
8858 |
+
"grad_norm": 0.039490871131420135,
|
8859 |
+
"learning_rate": 2.1270476519490435e-06,
|
8860 |
+
"loss": 10.3244,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.5340391741662255,
|
8865 |
+
"grad_norm": 0.03646280616521835,
|
8866 |
+
"learning_rate": 2.079147600208364e-06,
|
8867 |
+
"loss": 10.3303,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.5344626786659609,
|
8872 |
+
"grad_norm": 0.039123885333538055,
|
8873 |
+
"learning_rate": 2.0317873852898518e-06,
|
8874 |
+
"loss": 10.332,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.5348861831656961,
|
8879 |
+
"grad_norm": 0.04183242470026016,
|
8880 |
+
"learning_rate": 1.9849672682898944e-06,
|
8881 |
+
"loss": 10.3297,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.5353096876654314,
|
8886 |
+
"grad_norm": 0.03520303592085838,
|
8887 |
+
"learning_rate": 1.9386875073274636e-06,
|
8888 |
+
"loss": 10.3265,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.5357331921651668,
|
8893 |
+
"grad_norm": 0.0325089730322361,
|
8894 |
+
"learning_rate": 1.8929483575424455e-06,
|
8895 |
+
"loss": 10.3345,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.5361566966649021,
|
8900 |
+
"grad_norm": 0.029976682737469673,
|
8901 |
+
"learning_rate": 1.8477500710944007e-06,
|
8902 |
+
"loss": 10.3292,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.5365802011646373,
|
8907 |
+
"grad_norm": 0.034131329506635666,
|
8908 |
+
"learning_rate": 1.803092897161096e-06,
|
8909 |
+
"loss": 10.3276,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.5370037056643727,
|
8914 |
+
"grad_norm": 0.03793232887983322,
|
8915 |
+
"learning_rate": 1.75897708193713e-06,
|
8916 |
+
"loss": 10.3349,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.537427210164108,
|
8921 |
+
"grad_norm": 0.025969160720705986,
|
8922 |
+
"learning_rate": 1.715402868632643e-06,
|
8923 |
+
"loss": 10.3325,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.5378507146638433,
|
8928 |
+
"grad_norm": 0.04372668266296387,
|
8929 |
+
"learning_rate": 1.6723704974718756e-06,
|
8930 |
+
"loss": 10.33,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.5382742191635786,
|
8935 |
+
"grad_norm": 0.03358982875943184,
|
8936 |
+
"learning_rate": 1.629880205691936e-06,
|
8937 |
+
"loss": 10.3321,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.5386977236633139,
|
8942 |
+
"grad_norm": 0.045495398342609406,
|
8943 |
+
"learning_rate": 1.5879322275414332e-06,
|
8944 |
+
"loss": 10.3334,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.5391212281630492,
|
8949 |
+
"grad_norm": 0.02813423052430153,
|
8950 |
+
"learning_rate": 1.5465267942792127e-06,
|
8951 |
+
"loss": 10.332,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.5395447326627846,
|
8956 |
+
"grad_norm": 0.02770121954381466,
|
8957 |
+
"learning_rate": 1.5056641341730903e-06,
|
8958 |
+
"loss": 10.3296,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.5399682371625198,
|
8963 |
+
"grad_norm": 0.04436861723661423,
|
8964 |
+
"learning_rate": 1.465344472498531e-06,
|
8965 |
+
"loss": 10.3286,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.5403917416622551,
|
8970 |
+
"grad_norm": 0.043747782707214355,
|
8971 |
+
"learning_rate": 1.4255680315375164e-06,
|
8972 |
+
"loss": 10.3332,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.5408152461619905,
|
8977 |
+
"grad_norm": 0.028111323714256287,
|
8978 |
+
"learning_rate": 1.3863350305772017e-06,
|
8979 |
+
"loss": 10.3319,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.5412387506617258,
|
8984 |
+
"grad_norm": 0.03884616121649742,
|
8985 |
+
"learning_rate": 1.3476456859087828e-06,
|
8986 |
+
"loss": 10.3317,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.541662255161461,
|
8991 |
+
"grad_norm": 0.04214450716972351,
|
8992 |
+
"learning_rate": 1.3095002108263199e-06,
|
8993 |
+
"loss": 10.3336,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.5420857596611964,
|
8998 |
+
"grad_norm": 0.0312722884118557,
|
8999 |
+
"learning_rate": 1.2718988156254607e-06,
|
9000 |
+
"loss": 10.3357,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.5425092641609317,
|
9005 |
+
"grad_norm": 0.09322332590818405,
|
9006 |
+
"learning_rate": 1.2348417076023745e-06,
|
9007 |
+
"loss": 10.3333,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.542932768660667,
|
9012 |
+
"grad_norm": 0.04540476202964783,
|
9013 |
+
"learning_rate": 1.198329091052608e-06,
|
9014 |
+
"loss": 10.3309,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.5433562731604024,
|
9019 |
+
"grad_norm": 0.029997704550623894,
|
9020 |
+
"learning_rate": 1.1623611672698765e-06,
|
9021 |
+
"loss": 10.3358,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.5437797776601376,
|
9026 |
+
"grad_norm": 0.0350346714258194,
|
9027 |
+
"learning_rate": 1.1269381345450526e-06,
|
9028 |
+
"loss": 10.3306,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.5442032821598729,
|
9033 |
+
"grad_norm": 0.04271746799349785,
|
9034 |
+
"learning_rate": 1.0920601881650006e-06,
|
9035 |
+
"loss": 10.3313,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.5446267866596083,
|
9040 |
+
"grad_norm": 0.03767610713839531,
|
9041 |
+
"learning_rate": 1.0577275204115444e-06,
|
9042 |
+
"loss": 10.3275,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.5450502911593436,
|
9047 |
+
"grad_norm": 0.02964678965508938,
|
9048 |
+
"learning_rate": 1.0239403205604014e-06,
|
9049 |
+
"loss": 10.3296,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.5454737956590788,
|
9054 |
+
"grad_norm": 0.03278511017560959,
|
9055 |
+
"learning_rate": 9.906987748800944e-07,
|
9056 |
+
"loss": 10.3329,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.5458973001588142,
|
9061 |
+
"grad_norm": 0.05790937691926956,
|
9062 |
+
"learning_rate": 9.580030666309969e-07,
|
9063 |
+
"loss": 10.3372,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.5463208046585495,
|
9068 |
+
"grad_norm": 0.03746120631694794,
|
9069 |
+
"learning_rate": 9.258533760642563e-07,
|
9070 |
+
"loss": 10.3302,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.5467443091582848,
|
9075 |
+
"grad_norm": 0.03203713148832321,
|
9076 |
+
"learning_rate": 8.942498804208498e-07,
|
9077 |
+
"loss": 10.3328,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.5471678136580201,
|
9082 |
+
"grad_norm": 0.032408781349658966,
|
9083 |
+
"learning_rate": 8.631927539305862e-07,
|
9084 |
+
"loss": 10.3328,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.5475913181577554,
|
9089 |
+
"grad_norm": 0.038404081016778946,
|
9090 |
+
"learning_rate": 8.326821678111163e-07,
|
9091 |
+
"loss": 10.3357,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.5480148226574907,
|
9096 |
+
"grad_norm": 0.03704221174120903,
|
9097 |
+
"learning_rate": 8.027182902670571e-07,
|
9098 |
+
"loss": 10.3267,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.5484383271572261,
|
9103 |
+
"grad_norm": 0.02777581661939621,
|
9104 |
+
"learning_rate": 7.733012864890032e-07,
|
9105 |
+
"loss": 10.3331,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.5488618316569613,
|
9110 |
+
"grad_norm": 0.0339139886200428,
|
9111 |
+
"learning_rate": 7.444313186526608e-07,
|
9112 |
+
"loss": 10.3355,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.5492853361566966,
|
9117 |
+
"grad_norm": 0.027996981516480446,
|
9118 |
+
"learning_rate": 7.161085459178929e-07,
|
9119 |
+
"loss": 10.3301,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.549708840656432,
|
9124 |
+
"grad_norm": 0.04270913451910019,
|
9125 |
+
"learning_rate": 6.88333124427909e-07,
|
9126 |
+
"loss": 10.3269,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.5501323451561673,
|
9131 |
+
"grad_norm": 0.0351426862180233,
|
9132 |
+
"learning_rate": 6.611052073083768e-07,
|
9133 |
+
"loss": 10.3306,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.5505558496559025,
|
9138 |
+
"grad_norm": 0.0378975048661232,
|
9139 |
+
"learning_rate": 6.344249446665674e-07,
|
9140 |
+
"loss": 10.3283,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.5509793541556379,
|
9145 |
+
"grad_norm": 0.028754916042089462,
|
9146 |
+
"learning_rate": 6.082924835905446e-07,
|
9147 |
+
"loss": 10.3287,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.5514028586553732,
|
9152 |
+
"grad_norm": 0.0465865433216095,
|
9153 |
+
"learning_rate": 5.827079681483438e-07,
|
9154 |
+
"loss": 10.3325,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.5518263631551085,
|
9159 |
+
"grad_norm": 0.037231337279081345,
|
9160 |
+
"learning_rate": 5.576715393871613e-07,
|
9161 |
+
"loss": 10.3278,
|
9162 |
+
"step": 1303
|
9163 |
+
},
|
9164 |
+
{
|
9165 |
+
"epoch": 0.5522498676548439,
|
9166 |
+
"grad_norm": 0.03710845485329628,
|
9167 |
+
"learning_rate": 5.331833353326432e-07,
|
9168 |
+
"loss": 10.3344,
|
9169 |
+
"step": 1304
|
9170 |
+
},
|
9171 |
+
{
|
9172 |
+
"epoch": 0.5526733721545791,
|
9173 |
+
"grad_norm": 0.02809790149331093,
|
9174 |
+
"learning_rate": 5.092434909880317e-07,
|
9175 |
+
"loss": 10.3321,
|
9176 |
+
"step": 1305
|
9177 |
+
},
|
9178 |
+
{
|
9179 |
+
"epoch": 0.5530968766543145,
|
9180 |
+
"grad_norm": 0.045991264283657074,
|
9181 |
+
"learning_rate": 4.858521383334868e-07,
|
9182 |
+
"loss": 10.3345,
|
9183 |
+
"step": 1306
|
9184 |
+
},
|
9185 |
+
{
|
9186 |
+
"epoch": 0.5535203811540498,
|
9187 |
+
"grad_norm": 0.03640573099255562,
|
9188 |
+
"learning_rate": 4.630094063253321e-07,
|
9189 |
+
"loss": 10.3294,
|
9190 |
+
"step": 1307
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 0.553943885653785,
|
9194 |
+
"grad_norm": 0.029001332819461823,
|
9195 |
+
"learning_rate": 4.4071542089535454e-07,
|
9196 |
+
"loss": 10.3318,
|
9197 |
+
"step": 1308
|
9198 |
+
},
|
9199 |
+
{
|
9200 |
+
"epoch": 0.5543673901535204,
|
9201 |
+
"grad_norm": 0.02934233844280243,
|
9202 |
+
"learning_rate": 4.18970304950117e-07,
|
9203 |
+
"loss": 10.3299,
|
9204 |
+
"step": 1309
|
9205 |
+
},
|
9206 |
+
{
|
9207 |
+
"epoch": 0.5547908946532557,
|
9208 |
+
"grad_norm": 0.03224503621459007,
|
9209 |
+
"learning_rate": 3.977741783702471e-07,
|
9210 |
+
"loss": 10.3285,
|
9211 |
+
"step": 1310
|
9212 |
+
},
|
9213 |
+
{
|
9214 |
+
"epoch": 0.555214399152991,
|
9215 |
+
"grad_norm": 0.03147895634174347,
|
9216 |
+
"learning_rate": 3.771271580098157e-07,
|
9217 |
+
"loss": 10.3325,
|
9218 |
+
"step": 1311
|
9219 |
+
},
|
9220 |
+
{
|
9221 |
+
"epoch": 0.5556379036527264,
|
9222 |
+
"grad_norm": 0.03843318298459053,
|
9223 |
+
"learning_rate": 3.570293576956596e-07,
|
9224 |
+
"loss": 10.3301,
|
9225 |
+
"step": 1312
|
9226 |
+
},
|
9227 |
+
{
|
9228 |
+
"epoch": 0.5560614081524616,
|
9229 |
+
"grad_norm": 0.0349433533847332,
|
9230 |
+
"learning_rate": 3.3748088822679325e-07,
|
9231 |
+
"loss": 10.332,
|
9232 |
+
"step": 1313
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 0.5564849126521969,
|
9236 |
+
"grad_norm": 0.03259619325399399,
|
9237 |
+
"learning_rate": 3.184818573737425e-07,
|
9238 |
+
"loss": 10.3296,
|
9239 |
+
"step": 1314
|
9240 |
+
},
|
9241 |
+
{
|
9242 |
+
"epoch": 0.5569084171519323,
|
9243 |
+
"grad_norm": 0.03497344255447388,
|
9244 |
+
"learning_rate": 3.0003236987802274e-07,
|
9245 |
+
"loss": 10.3314,
|
9246 |
+
"step": 1315
|
9247 |
+
},
|
9248 |
+
{
|
9249 |
+
"epoch": 0.5573319216516676,
|
9250 |
+
"grad_norm": 0.03283681720495224,
|
9251 |
+
"learning_rate": 2.821325274514952e-07,
|
9252 |
+
"loss": 10.3307,
|
9253 |
+
"step": 1316
|
9254 |
+
},
|
9255 |
+
{
|
9256 |
+
"epoch": 0.5577554261514028,
|
9257 |
+
"grad_norm": 0.03914149850606918,
|
9258 |
+
"learning_rate": 2.6478242877583383e-07,
|
9259 |
+
"loss": 10.3321,
|
9260 |
+
"step": 1317
|
9261 |
+
},
|
9262 |
+
{
|
9263 |
+
"epoch": 0.5581789306511382,
|
9264 |
+
"grad_norm": 0.028979448601603508,
|
9265 |
+
"learning_rate": 2.4798216950198127e-07,
|
9266 |
+
"loss": 10.3295,
|
9267 |
+
"step": 1318
|
9268 |
+
},
|
9269 |
+
{
|
9270 |
+
"epoch": 0.5586024351508735,
|
9271 |
+
"grad_norm": 0.0339006632566452,
|
9272 |
+
"learning_rate": 2.317318422496273e-07,
|
9273 |
+
"loss": 10.3326,
|
9274 |
+
"step": 1319
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 0.5590259396506088,
|
9278 |
+
"grad_norm": 0.027926115319132805,
|
9279 |
+
"learning_rate": 2.1603153660668674e-07,
|
9280 |
+
"loss": 10.3305,
|
9281 |
+
"step": 1320
|
9282 |
+
},
|
9283 |
+
{
|
9284 |
+
"epoch": 0.5594494441503441,
|
9285 |
+
"grad_norm": 0.031478822231292725,
|
9286 |
+
"learning_rate": 2.0088133912881113e-07,
|
9287 |
+
"loss": 10.3288,
|
9288 |
+
"step": 1321
|
9289 |
+
},
|
9290 |
+
{
|
9291 |
+
"epoch": 0.5598729486500794,
|
9292 |
+
"grad_norm": 0.03274491801857948,
|
9293 |
+
"learning_rate": 1.862813333389113e-07,
|
9294 |
+
"loss": 10.3361,
|
9295 |
+
"step": 1322
|
9296 |
+
},
|
9297 |
+
{
|
9298 |
+
"epoch": 0.5602964531498147,
|
9299 |
+
"grad_norm": 0.0399165078997612,
|
9300 |
+
"learning_rate": 1.722315997267021e-07,
|
9301 |
+
"loss": 10.3344,
|
9302 |
+
"step": 1323
|
9303 |
+
},
|
9304 |
+
{
|
9305 |
+
"epoch": 0.5607199576495501,
|
9306 |
+
"grad_norm": 0.030695218592882156,
|
9307 |
+
"learning_rate": 1.5873221574822516e-07,
|
9308 |
+
"loss": 10.3298,
|
9309 |
+
"step": 1324
|
9310 |
+
},
|
9311 |
+
{
|
9312 |
+
"epoch": 0.5611434621492853,
|
9313 |
+
"grad_norm": 0.03967565670609474,
|
9314 |
+
"learning_rate": 1.4578325582548237e-07,
|
9315 |
+
"loss": 10.3305,
|
9316 |
+
"step": 1325
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 0.5615669666490206,
|
9320 |
+
"grad_norm": 0.03664049133658409,
|
9321 |
+
"learning_rate": 1.3338479134596958e-07,
|
9322 |
+
"loss": 10.3293,
|
9323 |
+
"step": 1326
|
9324 |
+
},
|
9325 |
+
{
|
9326 |
+
"epoch": 0.561990471148756,
|
9327 |
+
"grad_norm": 0.03802071511745453,
|
9328 |
+
"learning_rate": 1.2153689066233266e-07,
|
9329 |
+
"loss": 10.3305,
|
9330 |
+
"step": 1327
|
9331 |
+
},
|
9332 |
+
{
|
9333 |
+
"epoch": 0.5624139756484913,
|
9334 |
+
"grad_norm": 0.036713242530822754,
|
9335 |
+
"learning_rate": 1.1023961909192304e-07,
|
9336 |
+
"loss": 10.3287,
|
9337 |
+
"step": 1328
|
9338 |
+
},
|
9339 |
+
{
|
9340 |
+
"epoch": 0.5628374801482265,
|
9341 |
+
"grad_norm": 0.04824815317988396,
|
9342 |
+
"learning_rate": 9.949303891653161e-08,
|
9343 |
+
"loss": 10.3353,
|
9344 |
+
"step": 1329
|
9345 |
+
},
|
9346 |
+
{
|
9347 |
+
"epoch": 0.5632609846479619,
|
9348 |
+
"grad_norm": 0.03399055823683739,
|
9349 |
+
"learning_rate": 8.929720938193331e-08,
|
9350 |
+
"loss": 10.3302,
|
9351 |
+
"step": 1330
|
9352 |
+
},
|
9353 |
+
{
|
9354 |
+
"epoch": 0.5636844891476972,
|
9355 |
+
"grad_norm": 0.030519891530275345,
|
9356 |
+
"learning_rate": 7.965218669766516e-08,
|
9357 |
+
"loss": 10.3277,
|
9358 |
+
"step": 1331
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 0.5641079936474325,
|
9362 |
+
"grad_norm": 0.03647278994321823,
|
9363 |
+
"learning_rate": 7.05580240366488e-08,
|
9364 |
+
"loss": 10.3276,
|
9365 |
+
"step": 1332
|
9366 |
+
},
|
9367 |
+
{
|
9368 |
+
"epoch": 0.5645314981471679,
|
9369 |
+
"grad_norm": 0.0370662622153759,
|
9370 |
+
"learning_rate": 6.201477153493506e-08,
|
9371 |
+
"loss": 10.3344,
|
9372 |
+
"step": 1333
|
9373 |
+
},
|
9374 |
+
{
|
9375 |
+
"epoch": 0.5649550026469031,
|
9376 |
+
"grad_norm": 0.038933202624320984,
|
9377 |
+
"learning_rate": 5.402247629139323e-08,
|
9378 |
+
"loss": 10.3313,
|
9379 |
+
"step": 1334
|
9380 |
+
},
|
9381 |
+
{
|
9382 |
+
"epoch": 0.5653785071466384,
|
9383 |
+
"grad_norm": 0.030461156740784645,
|
9384 |
+
"learning_rate": 4.658118236747777e-08,
|
9385 |
+
"loss": 10.3292,
|
9386 |
+
"step": 1335
|
9387 |
+
},
|
9388 |
+
{
|
9389 |
+
"epoch": 0.5658020116463738,
|
9390 |
+
"grad_norm": 0.030602607876062393,
|
9391 |
+
"learning_rate": 3.9690930786995264e-08,
|
9392 |
+
"loss": 10.3294,
|
9393 |
+
"step": 1336
|
9394 |
+
},
|
9395 |
+
{
|
9396 |
+
"epoch": 0.566225516146109,
|
9397 |
+
"grad_norm": 0.03394312039017677,
|
9398 |
+
"learning_rate": 3.335175953581571e-08,
|
9399 |
+
"loss": 10.3342,
|
9400 |
+
"step": 1337
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 0.5666490206458443,
|
9404 |
+
"grad_norm": 0.051167815923690796,
|
9405 |
+
"learning_rate": 2.756370356175042e-08,
|
9406 |
+
"loss": 10.3349,
|
9407 |
+
"step": 1338
|
9408 |
+
},
|
9409 |
+
{
|
9410 |
+
"epoch": 0.5670725251455797,
|
9411 |
+
"grad_norm": 0.03260042518377304,
|
9412 |
+
"learning_rate": 2.232679477430777e-08,
|
9413 |
+
"loss": 10.3333,
|
9414 |
+
"step": 1339
|
9415 |
+
},
|
9416 |
+
{
|
9417 |
+
"epoch": 0.567496029645315,
|
9418 |
+
"grad_norm": 0.040678899735212326,
|
9419 |
+
"learning_rate": 1.7641062044515544e-08,
|
9420 |
+
"loss": 10.3287,
|
9421 |
+
"step": 1340
|
9422 |
+
},
|
9423 |
+
{
|
9424 |
+
"epoch": 0.5679195341450503,
|
9425 |
+
"grad_norm": 0.04332433268427849,
|
9426 |
+
"learning_rate": 1.350653120477663e-08,
|
9427 |
+
"loss": 10.3336,
|
9428 |
+
"step": 1341
|
9429 |
+
},
|
9430 |
+
{
|
9431 |
+
"epoch": 0.5683430386447856,
|
9432 |
+
"grad_norm": 0.03431249037384987,
|
9433 |
+
"learning_rate": 9.923225048724671e-09,
|
9434 |
+
"loss": 10.3331,
|
9435 |
+
"step": 1342
|
9436 |
+
},
|
9437 |
+
{
|
9438 |
+
"epoch": 0.5687665431445209,
|
9439 |
+
"grad_norm": 0.0347750224173069,
|
9440 |
+
"learning_rate": 6.891163331101957e-09,
|
9441 |
+
"loss": 10.3338,
|
9442 |
+
"step": 1343
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 0.5691900476442562,
|
9446 |
+
"grad_norm": 0.031236495822668076,
|
9447 |
+
"learning_rate": 4.410362767626186e-09,
|
9448 |
+
"loss": 10.3311,
|
9449 |
+
"step": 1344
|
9450 |
+
},
|
9451 |
+
{
|
9452 |
+
"epoch": 0.5696135521439916,
|
9453 |
+
"grad_norm": 0.036892782896757126,
|
9454 |
+
"learning_rate": 2.4808370349460596e-09,
|
9455 |
+
"loss": 10.332,
|
9456 |
+
"step": 1345
|
9457 |
+
},
|
9458 |
+
{
|
9459 |
+
"epoch": 0.5700370566437268,
|
9460 |
+
"grad_norm": 0.02656089887022972,
|
9461 |
+
"learning_rate": 1.1025967705080576e-09,
|
9462 |
+
"loss": 10.3323,
|
9463 |
+
"step": 1346
|
9464 |
+
},
|
9465 |
+
{
|
9466 |
+
"epoch": 0.5704605611434621,
|
9467 |
+
"grad_norm": 0.03345981240272522,
|
9468 |
+
"learning_rate": 2.756495725342312e-10,
|
9469 |
+
"loss": 10.3286,
|
9470 |
+
"step": 1347
|
9471 |
+
},
|
9472 |
+
{
|
9473 |
+
"epoch": 0.5708840656431975,
|
9474 |
+
"grad_norm": 0.03345588967204094,
|
9475 |
+
"learning_rate": 0.0,
|
9476 |
+
"loss": 10.3287,
|
9477 |
+
"step": 1348
|
9478 |
+
},
|
9479 |
+
{
|
9480 |
+
"epoch": 0.5708840656431975,
|
9481 |
+
"eval_loss": 10.330697059631348,
|
9482 |
+
"eval_runtime": 3.473,
|
9483 |
+
"eval_samples_per_second": 286.494,
|
9484 |
+
"eval_steps_per_second": 143.391,
|
9485 |
+
"step": 1348
|
9486 |
}
|
9487 |
],
|
9488 |
"logging_steps": 1,
|
|
|
9497 |
"should_evaluate": false,
|
9498 |
"should_log": false,
|
9499 |
"should_save": true,
|
9500 |
+
"should_training_stop": true
|
9501 |
},
|
9502 |
"attributes": {}
|
9503 |
}
|
9504 |
},
|
9505 |
+
"total_flos": 28945837916160.0,
|
9506 |
"train_batch_size": 2,
|
9507 |
"trial_name": null,
|
9508 |
"trial_params": null
|