Training in progress, step 1099, 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 80792096
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:2b5013b3b98299a02e1a1064d79cdcbbf953b32f057f7227ec6f66e9425151e8
|
3 |
size 80792096
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 41460084
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:785b47bba2f2b41c201771bb12d0ac1163e96682bc12bf2d077a0764e62d6f33
|
3 |
size 41460084
|
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:cfc4a264850458b77faa6649b6383744f11fc708a51b406678a774874a223c17
|
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 sha256:47d824ff435dfab48400f8d61469d75b87bb5e0fd7d8ed8b38a440a6169896cc
|
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":
|
5 |
"eval_steps": 275,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -5814,6 +5814,1924 @@
|
|
5814 |
"eval_samples_per_second": 9.011,
|
5815 |
"eval_steps_per_second": 4.508,
|
5816 |
"step": 825
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5817 |
}
|
5818 |
],
|
5819 |
"logging_steps": 1,
|
@@ -5828,12 +7746,12 @@
|
|
5828 |
"should_evaluate": false,
|
5829 |
"should_log": false,
|
5830 |
"should_save": true,
|
5831 |
-
"should_training_stop":
|
5832 |
},
|
5833 |
"attributes": {}
|
5834 |
}
|
5835 |
},
|
5836 |
-
"total_flos":
|
5837 |
"train_batch_size": 2,
|
5838 |
"trial_name": null,
|
5839 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 1.0006826715212198,
|
5 |
"eval_steps": 275,
|
6 |
+
"global_step": 1099,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5814 |
"eval_samples_per_second": 9.011,
|
5815 |
"eval_steps_per_second": 4.508,
|
5816 |
"step": 825
|
5817 |
+
},
|
5818 |
+
{
|
5819 |
+
"epoch": 0.7518489020366367,
|
5820 |
+
"grad_norm": 0.4428364336490631,
|
5821 |
+
"learning_rate": 2.944247910759097e-05,
|
5822 |
+
"loss": 2.1631,
|
5823 |
+
"step": 826
|
5824 |
+
},
|
5825 |
+
{
|
5826 |
+
"epoch": 0.7527591307315963,
|
5827 |
+
"grad_norm": 0.43727412819862366,
|
5828 |
+
"learning_rate": 2.9238342996745817e-05,
|
5829 |
+
"loss": 2.1495,
|
5830 |
+
"step": 827
|
5831 |
+
},
|
5832 |
+
{
|
5833 |
+
"epoch": 0.7536693594265559,
|
5834 |
+
"grad_norm": 0.45698437094688416,
|
5835 |
+
"learning_rate": 2.9034795786069436e-05,
|
5836 |
+
"loss": 2.1497,
|
5837 |
+
"step": 828
|
5838 |
+
},
|
5839 |
+
{
|
5840 |
+
"epoch": 0.7545795881215155,
|
5841 |
+
"grad_norm": 0.423408567905426,
|
5842 |
+
"learning_rate": 2.8831839169543996e-05,
|
5843 |
+
"loss": 1.9607,
|
5844 |
+
"step": 829
|
5845 |
+
},
|
5846 |
+
{
|
5847 |
+
"epoch": 0.7554898168164751,
|
5848 |
+
"grad_norm": 0.4666843116283417,
|
5849 |
+
"learning_rate": 2.862947483623659e-05,
|
5850 |
+
"loss": 2.1271,
|
5851 |
+
"step": 830
|
5852 |
+
},
|
5853 |
+
{
|
5854 |
+
"epoch": 0.7564000455114347,
|
5855 |
+
"grad_norm": 0.46026623249053955,
|
5856 |
+
"learning_rate": 2.8427704470285144e-05,
|
5857 |
+
"loss": 2.1943,
|
5858 |
+
"step": 831
|
5859 |
+
},
|
5860 |
+
{
|
5861 |
+
"epoch": 0.7573102742063944,
|
5862 |
+
"grad_norm": 0.459824800491333,
|
5863 |
+
"learning_rate": 2.8226529750884402e-05,
|
5864 |
+
"loss": 2.0793,
|
5865 |
+
"step": 832
|
5866 |
+
},
|
5867 |
+
{
|
5868 |
+
"epoch": 0.758220502901354,
|
5869 |
+
"grad_norm": 0.4680033326148987,
|
5870 |
+
"learning_rate": 2.8025952352271958e-05,
|
5871 |
+
"loss": 2.1652,
|
5872 |
+
"step": 833
|
5873 |
+
},
|
5874 |
+
{
|
5875 |
+
"epoch": 0.7591307315963136,
|
5876 |
+
"grad_norm": 0.47054019570350647,
|
5877 |
+
"learning_rate": 2.7825973943714335e-05,
|
5878 |
+
"loss": 2.1526,
|
5879 |
+
"step": 834
|
5880 |
+
},
|
5881 |
+
{
|
5882 |
+
"epoch": 0.7600409602912732,
|
5883 |
+
"grad_norm": 0.465668648481369,
|
5884 |
+
"learning_rate": 2.7626596189492983e-05,
|
5885 |
+
"loss": 2.0476,
|
5886 |
+
"step": 835
|
5887 |
+
},
|
5888 |
+
{
|
5889 |
+
"epoch": 0.7609511889862328,
|
5890 |
+
"grad_norm": 0.47716715931892395,
|
5891 |
+
"learning_rate": 2.7427820748890685e-05,
|
5892 |
+
"loss": 2.1511,
|
5893 |
+
"step": 836
|
5894 |
+
},
|
5895 |
+
{
|
5896 |
+
"epoch": 0.7618614176811924,
|
5897 |
+
"grad_norm": 0.468532532453537,
|
5898 |
+
"learning_rate": 2.7229649276177503e-05,
|
5899 |
+
"loss": 2.1065,
|
5900 |
+
"step": 837
|
5901 |
+
},
|
5902 |
+
{
|
5903 |
+
"epoch": 0.762771646376152,
|
5904 |
+
"grad_norm": 0.4883919358253479,
|
5905 |
+
"learning_rate": 2.7032083420597e-05,
|
5906 |
+
"loss": 2.2111,
|
5907 |
+
"step": 838
|
5908 |
+
},
|
5909 |
+
{
|
5910 |
+
"epoch": 0.7636818750711116,
|
5911 |
+
"grad_norm": 0.5037781000137329,
|
5912 |
+
"learning_rate": 2.683512482635281e-05,
|
5913 |
+
"loss": 2.1824,
|
5914 |
+
"step": 839
|
5915 |
+
},
|
5916 |
+
{
|
5917 |
+
"epoch": 0.7645921037660712,
|
5918 |
+
"grad_norm": 0.5097115635871887,
|
5919 |
+
"learning_rate": 2.6638775132594553e-05,
|
5920 |
+
"loss": 2.2818,
|
5921 |
+
"step": 840
|
5922 |
+
},
|
5923 |
+
{
|
5924 |
+
"epoch": 0.7655023324610308,
|
5925 |
+
"grad_norm": 0.5003491640090942,
|
5926 |
+
"learning_rate": 2.6443035973404496e-05,
|
5927 |
+
"loss": 2.112,
|
5928 |
+
"step": 841
|
5929 |
+
},
|
5930 |
+
{
|
5931 |
+
"epoch": 0.7664125611559904,
|
5932 |
+
"grad_norm": 0.5198303461074829,
|
5933 |
+
"learning_rate": 2.624790897778391e-05,
|
5934 |
+
"loss": 2.1864,
|
5935 |
+
"step": 842
|
5936 |
+
},
|
5937 |
+
{
|
5938 |
+
"epoch": 0.76732278985095,
|
5939 |
+
"grad_norm": 0.5522372722625732,
|
5940 |
+
"learning_rate": 2.605339576963929e-05,
|
5941 |
+
"loss": 2.3857,
|
5942 |
+
"step": 843
|
5943 |
+
},
|
5944 |
+
{
|
5945 |
+
"epoch": 0.7682330185459096,
|
5946 |
+
"grad_norm": 0.5393935441970825,
|
5947 |
+
"learning_rate": 2.585949796776912e-05,
|
5948 |
+
"loss": 2.2549,
|
5949 |
+
"step": 844
|
5950 |
+
},
|
5951 |
+
{
|
5952 |
+
"epoch": 0.7691432472408692,
|
5953 |
+
"grad_norm": 0.573794424533844,
|
5954 |
+
"learning_rate": 2.5666217185850262e-05,
|
5955 |
+
"loss": 2.3236,
|
5956 |
+
"step": 845
|
5957 |
+
},
|
5958 |
+
{
|
5959 |
+
"epoch": 0.7700534759358288,
|
5960 |
+
"grad_norm": 0.5689824819564819,
|
5961 |
+
"learning_rate": 2.5473555032424533e-05,
|
5962 |
+
"loss": 2.1463,
|
5963 |
+
"step": 846
|
5964 |
+
},
|
5965 |
+
{
|
5966 |
+
"epoch": 0.7709637046307884,
|
5967 |
+
"grad_norm": 0.6545232534408569,
|
5968 |
+
"learning_rate": 2.528151311088537e-05,
|
5969 |
+
"loss": 2.3964,
|
5970 |
+
"step": 847
|
5971 |
+
},
|
5972 |
+
{
|
5973 |
+
"epoch": 0.771873933325748,
|
5974 |
+
"grad_norm": 0.7613667845726013,
|
5975 |
+
"learning_rate": 2.50900930194644e-05,
|
5976 |
+
"loss": 2.7365,
|
5977 |
+
"step": 848
|
5978 |
+
},
|
5979 |
+
{
|
5980 |
+
"epoch": 0.7727841620207077,
|
5981 |
+
"grad_norm": 0.8592699766159058,
|
5982 |
+
"learning_rate": 2.4899296351218227e-05,
|
5983 |
+
"loss": 2.3281,
|
5984 |
+
"step": 849
|
5985 |
+
},
|
5986 |
+
{
|
5987 |
+
"epoch": 0.7736943907156673,
|
5988 |
+
"grad_norm": 1.6767549514770508,
|
5989 |
+
"learning_rate": 2.4709124694015116e-05,
|
5990 |
+
"loss": 2.4329,
|
5991 |
+
"step": 850
|
5992 |
+
},
|
5993 |
+
{
|
5994 |
+
"epoch": 0.774604619410627,
|
5995 |
+
"grad_norm": 0.42721498012542725,
|
5996 |
+
"learning_rate": 2.451957963052185e-05,
|
5997 |
+
"loss": 2.4287,
|
5998 |
+
"step": 851
|
5999 |
+
},
|
6000 |
+
{
|
6001 |
+
"epoch": 0.7755148481055866,
|
6002 |
+
"grad_norm": 0.4105132818222046,
|
6003 |
+
"learning_rate": 2.433066273819037e-05,
|
6004 |
+
"loss": 2.3069,
|
6005 |
+
"step": 852
|
6006 |
+
},
|
6007 |
+
{
|
6008 |
+
"epoch": 0.7764250768005462,
|
6009 |
+
"grad_norm": 0.42730897665023804,
|
6010 |
+
"learning_rate": 2.4142375589244957e-05,
|
6011 |
+
"loss": 2.3786,
|
6012 |
+
"step": 853
|
6013 |
+
},
|
6014 |
+
{
|
6015 |
+
"epoch": 0.7773353054955058,
|
6016 |
+
"grad_norm": 0.41130363941192627,
|
6017 |
+
"learning_rate": 2.3954719750668907e-05,
|
6018 |
+
"loss": 2.1378,
|
6019 |
+
"step": 854
|
6020 |
+
},
|
6021 |
+
{
|
6022 |
+
"epoch": 0.7782455341904654,
|
6023 |
+
"grad_norm": 0.41424882411956787,
|
6024 |
+
"learning_rate": 2.3767696784191463e-05,
|
6025 |
+
"loss": 2.2526,
|
6026 |
+
"step": 855
|
6027 |
+
},
|
6028 |
+
{
|
6029 |
+
"epoch": 0.779155762885425,
|
6030 |
+
"grad_norm": 0.42201122641563416,
|
6031 |
+
"learning_rate": 2.3581308246275103e-05,
|
6032 |
+
"loss": 2.3379,
|
6033 |
+
"step": 856
|
6034 |
+
},
|
6035 |
+
{
|
6036 |
+
"epoch": 0.7800659915803846,
|
6037 |
+
"grad_norm": 0.4150107204914093,
|
6038 |
+
"learning_rate": 2.339555568810221e-05,
|
6039 |
+
"loss": 2.1722,
|
6040 |
+
"step": 857
|
6041 |
+
},
|
6042 |
+
{
|
6043 |
+
"epoch": 0.7809762202753442,
|
6044 |
+
"grad_norm": 0.432271271944046,
|
6045 |
+
"learning_rate": 2.321044065556246e-05,
|
6046 |
+
"loss": 2.4875,
|
6047 |
+
"step": 858
|
6048 |
+
},
|
6049 |
+
{
|
6050 |
+
"epoch": 0.7818864489703038,
|
6051 |
+
"grad_norm": 0.4255993068218231,
|
6052 |
+
"learning_rate": 2.302596468923981e-05,
|
6053 |
+
"loss": 2.3043,
|
6054 |
+
"step": 859
|
6055 |
+
},
|
6056 |
+
{
|
6057 |
+
"epoch": 0.7827966776652634,
|
6058 |
+
"grad_norm": 0.4193149507045746,
|
6059 |
+
"learning_rate": 2.284212932439972e-05,
|
6060 |
+
"loss": 2.3238,
|
6061 |
+
"step": 860
|
6062 |
+
},
|
6063 |
+
{
|
6064 |
+
"epoch": 0.783706906360223,
|
6065 |
+
"grad_norm": 0.3979727625846863,
|
6066 |
+
"learning_rate": 2.265893609097637e-05,
|
6067 |
+
"loss": 2.0908,
|
6068 |
+
"step": 861
|
6069 |
+
},
|
6070 |
+
{
|
6071 |
+
"epoch": 0.7846171350551826,
|
6072 |
+
"grad_norm": 0.42481502890586853,
|
6073 |
+
"learning_rate": 2.247638651355991e-05,
|
6074 |
+
"loss": 2.3404,
|
6075 |
+
"step": 862
|
6076 |
+
},
|
6077 |
+
{
|
6078 |
+
"epoch": 0.7855273637501422,
|
6079 |
+
"grad_norm": 0.41932496428489685,
|
6080 |
+
"learning_rate": 2.229448211138382e-05,
|
6081 |
+
"loss": 2.3529,
|
6082 |
+
"step": 863
|
6083 |
+
},
|
6084 |
+
{
|
6085 |
+
"epoch": 0.7864375924451018,
|
6086 |
+
"grad_norm": 0.4188045263290405,
|
6087 |
+
"learning_rate": 2.211322439831218e-05,
|
6088 |
+
"loss": 2.1973,
|
6089 |
+
"step": 864
|
6090 |
+
},
|
6091 |
+
{
|
6092 |
+
"epoch": 0.7873478211400614,
|
6093 |
+
"grad_norm": 0.41575223207473755,
|
6094 |
+
"learning_rate": 2.1932614882827197e-05,
|
6095 |
+
"loss": 2.2664,
|
6096 |
+
"step": 865
|
6097 |
+
},
|
6098 |
+
{
|
6099 |
+
"epoch": 0.788258049835021,
|
6100 |
+
"grad_norm": 0.41103559732437134,
|
6101 |
+
"learning_rate": 2.1752655068016515e-05,
|
6102 |
+
"loss": 2.1176,
|
6103 |
+
"step": 866
|
6104 |
+
},
|
6105 |
+
{
|
6106 |
+
"epoch": 0.7891682785299806,
|
6107 |
+
"grad_norm": 0.3994426727294922,
|
6108 |
+
"learning_rate": 2.1573346451560794e-05,
|
6109 |
+
"loss": 2.0824,
|
6110 |
+
"step": 867
|
6111 |
+
},
|
6112 |
+
{
|
6113 |
+
"epoch": 0.7900785072249402,
|
6114 |
+
"grad_norm": 0.40480148792266846,
|
6115 |
+
"learning_rate": 2.139469052572127e-05,
|
6116 |
+
"loss": 1.9797,
|
6117 |
+
"step": 868
|
6118 |
+
},
|
6119 |
+
{
|
6120 |
+
"epoch": 0.7909887359198998,
|
6121 |
+
"grad_norm": 0.4224672317504883,
|
6122 |
+
"learning_rate": 2.1216688777327154e-05,
|
6123 |
+
"loss": 2.0783,
|
6124 |
+
"step": 869
|
6125 |
+
},
|
6126 |
+
{
|
6127 |
+
"epoch": 0.7918989646148595,
|
6128 |
+
"grad_norm": 0.4260886013507843,
|
6129 |
+
"learning_rate": 2.1039342687763586e-05,
|
6130 |
+
"loss": 2.203,
|
6131 |
+
"step": 870
|
6132 |
+
},
|
6133 |
+
{
|
6134 |
+
"epoch": 0.7928091933098191,
|
6135 |
+
"grad_norm": 0.41183462738990784,
|
6136 |
+
"learning_rate": 2.0862653732958915e-05,
|
6137 |
+
"loss": 1.9724,
|
6138 |
+
"step": 871
|
6139 |
+
},
|
6140 |
+
{
|
6141 |
+
"epoch": 0.7937194220047787,
|
6142 |
+
"grad_norm": 0.43447592854499817,
|
6143 |
+
"learning_rate": 2.0686623383372715e-05,
|
6144 |
+
"loss": 2.1632,
|
6145 |
+
"step": 872
|
6146 |
+
},
|
6147 |
+
{
|
6148 |
+
"epoch": 0.7946296506997383,
|
6149 |
+
"grad_norm": 0.4297522008419037,
|
6150 |
+
"learning_rate": 2.051125310398353e-05,
|
6151 |
+
"loss": 2.0486,
|
6152 |
+
"step": 873
|
6153 |
+
},
|
6154 |
+
{
|
6155 |
+
"epoch": 0.795539879394698,
|
6156 |
+
"grad_norm": 0.45072224736213684,
|
6157 |
+
"learning_rate": 2.03365443542764e-05,
|
6158 |
+
"loss": 2.1973,
|
6159 |
+
"step": 874
|
6160 |
+
},
|
6161 |
+
{
|
6162 |
+
"epoch": 0.7964501080896575,
|
6163 |
+
"grad_norm": 0.4462050199508667,
|
6164 |
+
"learning_rate": 2.016249858823106e-05,
|
6165 |
+
"loss": 2.0274,
|
6166 |
+
"step": 875
|
6167 |
+
},
|
6168 |
+
{
|
6169 |
+
"epoch": 0.7973603367846172,
|
6170 |
+
"grad_norm": 0.4606810212135315,
|
6171 |
+
"learning_rate": 1.998911725430963e-05,
|
6172 |
+
"loss": 2.1616,
|
6173 |
+
"step": 876
|
6174 |
+
},
|
6175 |
+
{
|
6176 |
+
"epoch": 0.7982705654795768,
|
6177 |
+
"grad_norm": 0.44487303495407104,
|
6178 |
+
"learning_rate": 1.981640179544466e-05,
|
6179 |
+
"loss": 2.323,
|
6180 |
+
"step": 877
|
6181 |
+
},
|
6182 |
+
{
|
6183 |
+
"epoch": 0.7991807941745364,
|
6184 |
+
"grad_norm": 0.45202627778053284,
|
6185 |
+
"learning_rate": 1.964435364902705e-05,
|
6186 |
+
"loss": 2.1361,
|
6187 |
+
"step": 878
|
6188 |
+
},
|
6189 |
+
{
|
6190 |
+
"epoch": 0.800091022869496,
|
6191 |
+
"grad_norm": 0.44588690996170044,
|
6192 |
+
"learning_rate": 1.947297424689414e-05,
|
6193 |
+
"loss": 2.1173,
|
6194 |
+
"step": 879
|
6195 |
+
},
|
6196 |
+
{
|
6197 |
+
"epoch": 0.8010012515644556,
|
6198 |
+
"grad_norm": 0.46819573640823364,
|
6199 |
+
"learning_rate": 1.93022650153178e-05,
|
6200 |
+
"loss": 2.0187,
|
6201 |
+
"step": 880
|
6202 |
+
},
|
6203 |
+
{
|
6204 |
+
"epoch": 0.8019114802594152,
|
6205 |
+
"grad_norm": 0.44944408535957336,
|
6206 |
+
"learning_rate": 1.913222737499243e-05,
|
6207 |
+
"loss": 2.0103,
|
6208 |
+
"step": 881
|
6209 |
+
},
|
6210 |
+
{
|
6211 |
+
"epoch": 0.8028217089543748,
|
6212 |
+
"grad_norm": 0.44194296002388,
|
6213 |
+
"learning_rate": 1.8962862741023423e-05,
|
6214 |
+
"loss": 1.9489,
|
6215 |
+
"step": 882
|
6216 |
+
},
|
6217 |
+
{
|
6218 |
+
"epoch": 0.8037319376493344,
|
6219 |
+
"grad_norm": 0.4707835614681244,
|
6220 |
+
"learning_rate": 1.879417252291502e-05,
|
6221 |
+
"loss": 2.1982,
|
6222 |
+
"step": 883
|
6223 |
+
},
|
6224 |
+
{
|
6225 |
+
"epoch": 0.804642166344294,
|
6226 |
+
"grad_norm": 0.4707585573196411,
|
6227 |
+
"learning_rate": 1.8626158124558858e-05,
|
6228 |
+
"loss": 2.1049,
|
6229 |
+
"step": 884
|
6230 |
+
},
|
6231 |
+
{
|
6232 |
+
"epoch": 0.8055523950392536,
|
6233 |
+
"grad_norm": 0.4964425265789032,
|
6234 |
+
"learning_rate": 1.8458820944222255e-05,
|
6235 |
+
"loss": 2.2127,
|
6236 |
+
"step": 885
|
6237 |
+
},
|
6238 |
+
{
|
6239 |
+
"epoch": 0.8064626237342132,
|
6240 |
+
"grad_norm": 0.4742617607116699,
|
6241 |
+
"learning_rate": 1.829216237453637e-05,
|
6242 |
+
"loss": 2.1019,
|
6243 |
+
"step": 886
|
6244 |
+
},
|
6245 |
+
{
|
6246 |
+
"epoch": 0.8073728524291728,
|
6247 |
+
"grad_norm": 0.49655184149742126,
|
6248 |
+
"learning_rate": 1.8126183802484865e-05,
|
6249 |
+
"loss": 2.2403,
|
6250 |
+
"step": 887
|
6251 |
+
},
|
6252 |
+
{
|
6253 |
+
"epoch": 0.8082830811241324,
|
6254 |
+
"grad_norm": 0.4954749643802643,
|
6255 |
+
"learning_rate": 1.7960886609392214e-05,
|
6256 |
+
"loss": 2.0321,
|
6257 |
+
"step": 888
|
6258 |
+
},
|
6259 |
+
{
|
6260 |
+
"epoch": 0.8091933098190921,
|
6261 |
+
"grad_norm": 0.4694468379020691,
|
6262 |
+
"learning_rate": 1.7796272170912253e-05,
|
6263 |
+
"loss": 1.817,
|
6264 |
+
"step": 889
|
6265 |
+
},
|
6266 |
+
{
|
6267 |
+
"epoch": 0.8101035385140517,
|
6268 |
+
"grad_norm": 0.5026715397834778,
|
6269 |
+
"learning_rate": 1.763234185701673e-05,
|
6270 |
+
"loss": 2.2038,
|
6271 |
+
"step": 890
|
6272 |
+
},
|
6273 |
+
{
|
6274 |
+
"epoch": 0.8110137672090113,
|
6275 |
+
"grad_norm": 0.5050073862075806,
|
6276 |
+
"learning_rate": 1.7469097031983893e-05,
|
6277 |
+
"loss": 2.0861,
|
6278 |
+
"step": 891
|
6279 |
+
},
|
6280 |
+
{
|
6281 |
+
"epoch": 0.8119239959039709,
|
6282 |
+
"grad_norm": 0.5078185796737671,
|
6283 |
+
"learning_rate": 1.730653905438714e-05,
|
6284 |
+
"loss": 2.0672,
|
6285 |
+
"step": 892
|
6286 |
+
},
|
6287 |
+
{
|
6288 |
+
"epoch": 0.8128342245989305,
|
6289 |
+
"grad_norm": 0.525215744972229,
|
6290 |
+
"learning_rate": 1.7144669277083712e-05,
|
6291 |
+
"loss": 2.1502,
|
6292 |
+
"step": 893
|
6293 |
+
},
|
6294 |
+
{
|
6295 |
+
"epoch": 0.8137444532938901,
|
6296 |
+
"grad_norm": 0.5429519414901733,
|
6297 |
+
"learning_rate": 1.6983489047203483e-05,
|
6298 |
+
"loss": 2.0935,
|
6299 |
+
"step": 894
|
6300 |
+
},
|
6301 |
+
{
|
6302 |
+
"epoch": 0.8146546819888497,
|
6303 |
+
"grad_norm": 0.5544317960739136,
|
6304 |
+
"learning_rate": 1.6822999706137567e-05,
|
6305 |
+
"loss": 2.0943,
|
6306 |
+
"step": 895
|
6307 |
+
},
|
6308 |
+
{
|
6309 |
+
"epoch": 0.8155649106838093,
|
6310 |
+
"grad_norm": 0.6273201107978821,
|
6311 |
+
"learning_rate": 1.6663202589527473e-05,
|
6312 |
+
"loss": 2.3608,
|
6313 |
+
"step": 896
|
6314 |
+
},
|
6315 |
+
{
|
6316 |
+
"epoch": 0.8164751393787689,
|
6317 |
+
"grad_norm": 0.7101454734802246,
|
6318 |
+
"learning_rate": 1.6504099027253706e-05,
|
6319 |
+
"loss": 2.4168,
|
6320 |
+
"step": 897
|
6321 |
+
},
|
6322 |
+
{
|
6323 |
+
"epoch": 0.8173853680737285,
|
6324 |
+
"grad_norm": 0.7550842761993408,
|
6325 |
+
"learning_rate": 1.634569034342476e-05,
|
6326 |
+
"loss": 2.5798,
|
6327 |
+
"step": 898
|
6328 |
+
},
|
6329 |
+
{
|
6330 |
+
"epoch": 0.8182955967686881,
|
6331 |
+
"grad_norm": 0.8533863425254822,
|
6332 |
+
"learning_rate": 1.6187977856366253e-05,
|
6333 |
+
"loss": 2.5575,
|
6334 |
+
"step": 899
|
6335 |
+
},
|
6336 |
+
{
|
6337 |
+
"epoch": 0.8192058254636477,
|
6338 |
+
"grad_norm": 1.34774911403656,
|
6339 |
+
"learning_rate": 1.6030962878609725e-05,
|
6340 |
+
"loss": 2.4134,
|
6341 |
+
"step": 900
|
6342 |
+
},
|
6343 |
+
{
|
6344 |
+
"epoch": 0.8201160541586073,
|
6345 |
+
"grad_norm": 0.4584032893180847,
|
6346 |
+
"learning_rate": 1.587464671688187e-05,
|
6347 |
+
"loss": 2.4781,
|
6348 |
+
"step": 901
|
6349 |
+
},
|
6350 |
+
{
|
6351 |
+
"epoch": 0.8210262828535669,
|
6352 |
+
"grad_norm": 0.43342748284339905,
|
6353 |
+
"learning_rate": 1.5719030672093717e-05,
|
6354 |
+
"loss": 2.3685,
|
6355 |
+
"step": 902
|
6356 |
+
},
|
6357 |
+
{
|
6358 |
+
"epoch": 0.8219365115485265,
|
6359 |
+
"grad_norm": 0.4225307106971741,
|
6360 |
+
"learning_rate": 1.5564116039329545e-05,
|
6361 |
+
"loss": 2.2022,
|
6362 |
+
"step": 903
|
6363 |
+
},
|
6364 |
+
{
|
6365 |
+
"epoch": 0.8228467402434861,
|
6366 |
+
"grad_norm": 0.43026039004325867,
|
6367 |
+
"learning_rate": 1.5409904107836358e-05,
|
6368 |
+
"loss": 2.2817,
|
6369 |
+
"step": 904
|
6370 |
+
},
|
6371 |
+
{
|
6372 |
+
"epoch": 0.8237569689384457,
|
6373 |
+
"grad_norm": 0.4114493131637573,
|
6374 |
+
"learning_rate": 1.5256396161013075e-05,
|
6375 |
+
"loss": 2.3298,
|
6376 |
+
"step": 905
|
6377 |
+
},
|
6378 |
+
{
|
6379 |
+
"epoch": 0.8246671976334053,
|
6380 |
+
"grad_norm": 0.42313718795776367,
|
6381 |
+
"learning_rate": 1.5103593476399791e-05,
|
6382 |
+
"loss": 2.3211,
|
6383 |
+
"step": 906
|
6384 |
+
},
|
6385 |
+
{
|
6386 |
+
"epoch": 0.825577426328365,
|
6387 |
+
"grad_norm": 0.4246841371059418,
|
6388 |
+
"learning_rate": 1.495149732566723e-05,
|
6389 |
+
"loss": 2.2385,
|
6390 |
+
"step": 907
|
6391 |
+
},
|
6392 |
+
{
|
6393 |
+
"epoch": 0.8264876550233247,
|
6394 |
+
"grad_norm": 0.4131985008716583,
|
6395 |
+
"learning_rate": 1.4800108974606119e-05,
|
6396 |
+
"loss": 2.2873,
|
6397 |
+
"step": 908
|
6398 |
+
},
|
6399 |
+
{
|
6400 |
+
"epoch": 0.8273978837182843,
|
6401 |
+
"grad_norm": 0.42265599966049194,
|
6402 |
+
"learning_rate": 1.4649429683116644e-05,
|
6403 |
+
"loss": 2.1486,
|
6404 |
+
"step": 909
|
6405 |
+
},
|
6406 |
+
{
|
6407 |
+
"epoch": 0.8283081124132439,
|
6408 |
+
"grad_norm": 0.4338424801826477,
|
6409 |
+
"learning_rate": 1.4499460705197998e-05,
|
6410 |
+
"loss": 2.2365,
|
6411 |
+
"step": 910
|
6412 |
+
},
|
6413 |
+
{
|
6414 |
+
"epoch": 0.8292183411082035,
|
6415 |
+
"grad_norm": 0.4278540015220642,
|
6416 |
+
"learning_rate": 1.4350203288937936e-05,
|
6417 |
+
"loss": 2.36,
|
6418 |
+
"step": 911
|
6419 |
+
},
|
6420 |
+
{
|
6421 |
+
"epoch": 0.8301285698031631,
|
6422 |
+
"grad_norm": 0.41379448771476746,
|
6423 |
+
"learning_rate": 1.4201658676502294e-05,
|
6424 |
+
"loss": 2.184,
|
6425 |
+
"step": 912
|
6426 |
+
},
|
6427 |
+
{
|
6428 |
+
"epoch": 0.8310387984981227,
|
6429 |
+
"grad_norm": 0.42351198196411133,
|
6430 |
+
"learning_rate": 1.4053828104124867e-05,
|
6431 |
+
"loss": 2.2505,
|
6432 |
+
"step": 913
|
6433 |
+
},
|
6434 |
+
{
|
6435 |
+
"epoch": 0.8319490271930823,
|
6436 |
+
"grad_norm": 0.40783679485321045,
|
6437 |
+
"learning_rate": 1.3906712802096933e-05,
|
6438 |
+
"loss": 2.0255,
|
6439 |
+
"step": 914
|
6440 |
+
},
|
6441 |
+
{
|
6442 |
+
"epoch": 0.8328592558880419,
|
6443 |
+
"grad_norm": 0.4174416661262512,
|
6444 |
+
"learning_rate": 1.3760313994757001e-05,
|
6445 |
+
"loss": 2.2376,
|
6446 |
+
"step": 915
|
6447 |
+
},
|
6448 |
+
{
|
6449 |
+
"epoch": 0.8337694845830015,
|
6450 |
+
"grad_norm": 0.41884645819664,
|
6451 |
+
"learning_rate": 1.361463290048085e-05,
|
6452 |
+
"loss": 2.0206,
|
6453 |
+
"step": 916
|
6454 |
+
},
|
6455 |
+
{
|
6456 |
+
"epoch": 0.8346797132779611,
|
6457 |
+
"grad_norm": 0.399498850107193,
|
6458 |
+
"learning_rate": 1.3469670731671046e-05,
|
6459 |
+
"loss": 2.063,
|
6460 |
+
"step": 917
|
6461 |
+
},
|
6462 |
+
{
|
6463 |
+
"epoch": 0.8355899419729207,
|
6464 |
+
"grad_norm": 0.40431055426597595,
|
6465 |
+
"learning_rate": 1.3325428694747177e-05,
|
6466 |
+
"loss": 2.0053,
|
6467 |
+
"step": 918
|
6468 |
+
},
|
6469 |
+
{
|
6470 |
+
"epoch": 0.8365001706678803,
|
6471 |
+
"grad_norm": 0.40479356050491333,
|
6472 |
+
"learning_rate": 1.3181907990135622e-05,
|
6473 |
+
"loss": 2.0693,
|
6474 |
+
"step": 919
|
6475 |
+
},
|
6476 |
+
{
|
6477 |
+
"epoch": 0.8374103993628399,
|
6478 |
+
"grad_norm": 0.4056653678417206,
|
6479 |
+
"learning_rate": 1.3039109812259598e-05,
|
6480 |
+
"loss": 2.0361,
|
6481 |
+
"step": 920
|
6482 |
+
},
|
6483 |
+
{
|
6484 |
+
"epoch": 0.8383206280577995,
|
6485 |
+
"grad_norm": 0.4257088005542755,
|
6486 |
+
"learning_rate": 1.2897035349529263e-05,
|
6487 |
+
"loss": 2.0589,
|
6488 |
+
"step": 921
|
6489 |
+
},
|
6490 |
+
{
|
6491 |
+
"epoch": 0.8392308567527591,
|
6492 |
+
"grad_norm": 0.43024080991744995,
|
6493 |
+
"learning_rate": 1.2755685784331783e-05,
|
6494 |
+
"loss": 2.0419,
|
6495 |
+
"step": 922
|
6496 |
+
},
|
6497 |
+
{
|
6498 |
+
"epoch": 0.8401410854477187,
|
6499 |
+
"grad_norm": 0.42889195680618286,
|
6500 |
+
"learning_rate": 1.2615062293021507e-05,
|
6501 |
+
"loss": 2.0515,
|
6502 |
+
"step": 923
|
6503 |
+
},
|
6504 |
+
{
|
6505 |
+
"epoch": 0.8410513141426783,
|
6506 |
+
"grad_norm": 0.4491952061653137,
|
6507 |
+
"learning_rate": 1.2475166045910159e-05,
|
6508 |
+
"loss": 2.2535,
|
6509 |
+
"step": 924
|
6510 |
+
},
|
6511 |
+
{
|
6512 |
+
"epoch": 0.8419615428376379,
|
6513 |
+
"grad_norm": 0.43797358870506287,
|
6514 |
+
"learning_rate": 1.2335998207257137e-05,
|
6515 |
+
"loss": 2.1338,
|
6516 |
+
"step": 925
|
6517 |
+
},
|
6518 |
+
{
|
6519 |
+
"epoch": 0.8428717715325975,
|
6520 |
+
"grad_norm": 0.4491622745990753,
|
6521 |
+
"learning_rate": 1.2197559935259795e-05,
|
6522 |
+
"loss": 2.2059,
|
6523 |
+
"step": 926
|
6524 |
+
},
|
6525 |
+
{
|
6526 |
+
"epoch": 0.8437820002275572,
|
6527 |
+
"grad_norm": 0.43628188967704773,
|
6528 |
+
"learning_rate": 1.20598523820438e-05,
|
6529 |
+
"loss": 1.8784,
|
6530 |
+
"step": 927
|
6531 |
+
},
|
6532 |
+
{
|
6533 |
+
"epoch": 0.8446922289225168,
|
6534 |
+
"grad_norm": 0.45739004015922546,
|
6535 |
+
"learning_rate": 1.1922876693653585e-05,
|
6536 |
+
"loss": 2.0433,
|
6537 |
+
"step": 928
|
6538 |
+
},
|
6539 |
+
{
|
6540 |
+
"epoch": 0.8456024576174764,
|
6541 |
+
"grad_norm": 0.44873446226119995,
|
6542 |
+
"learning_rate": 1.1786634010042719e-05,
|
6543 |
+
"loss": 1.9578,
|
6544 |
+
"step": 929
|
6545 |
+
},
|
6546 |
+
{
|
6547 |
+
"epoch": 0.846512686312436,
|
6548 |
+
"grad_norm": 0.43957433104515076,
|
6549 |
+
"learning_rate": 1.1651125465064516e-05,
|
6550 |
+
"loss": 2.0078,
|
6551 |
+
"step": 930
|
6552 |
+
},
|
6553 |
+
{
|
6554 |
+
"epoch": 0.8474229150073956,
|
6555 |
+
"grad_norm": 0.4639342129230499,
|
6556 |
+
"learning_rate": 1.1516352186462586e-05,
|
6557 |
+
"loss": 2.0714,
|
6558 |
+
"step": 931
|
6559 |
+
},
|
6560 |
+
{
|
6561 |
+
"epoch": 0.8483331437023552,
|
6562 |
+
"grad_norm": 0.44638022780418396,
|
6563 |
+
"learning_rate": 1.13823152958614e-05,
|
6564 |
+
"loss": 1.8991,
|
6565 |
+
"step": 932
|
6566 |
+
},
|
6567 |
+
{
|
6568 |
+
"epoch": 0.8492433723973148,
|
6569 |
+
"grad_norm": 0.4596819579601288,
|
6570 |
+
"learning_rate": 1.1249015908756998e-05,
|
6571 |
+
"loss": 1.9595,
|
6572 |
+
"step": 933
|
6573 |
+
},
|
6574 |
+
{
|
6575 |
+
"epoch": 0.8501536010922744,
|
6576 |
+
"grad_norm": 0.47656434774398804,
|
6577 |
+
"learning_rate": 1.1116455134507664e-05,
|
6578 |
+
"loss": 2.0788,
|
6579 |
+
"step": 934
|
6580 |
+
},
|
6581 |
+
{
|
6582 |
+
"epoch": 0.851063829787234,
|
6583 |
+
"grad_norm": 0.4645254611968994,
|
6584 |
+
"learning_rate": 1.098463407632474e-05,
|
6585 |
+
"loss": 2.0703,
|
6586 |
+
"step": 935
|
6587 |
+
},
|
6588 |
+
{
|
6589 |
+
"epoch": 0.8519740584821937,
|
6590 |
+
"grad_norm": 0.4659541845321655,
|
6591 |
+
"learning_rate": 1.0853553831263418e-05,
|
6592 |
+
"loss": 2.0804,
|
6593 |
+
"step": 936
|
6594 |
+
},
|
6595 |
+
{
|
6596 |
+
"epoch": 0.8528842871771533,
|
6597 |
+
"grad_norm": 0.4771886467933655,
|
6598 |
+
"learning_rate": 1.0723215490213634e-05,
|
6599 |
+
"loss": 2.1124,
|
6600 |
+
"step": 937
|
6601 |
+
},
|
6602 |
+
{
|
6603 |
+
"epoch": 0.8537945158721129,
|
6604 |
+
"grad_norm": 0.49211612343788147,
|
6605 |
+
"learning_rate": 1.0593620137890948e-05,
|
6606 |
+
"loss": 2.2221,
|
6607 |
+
"step": 938
|
6608 |
+
},
|
6609 |
+
{
|
6610 |
+
"epoch": 0.8547047445670725,
|
6611 |
+
"grad_norm": 0.5174618363380432,
|
6612 |
+
"learning_rate": 1.0464768852827545e-05,
|
6613 |
+
"loss": 2.1684,
|
6614 |
+
"step": 939
|
6615 |
+
},
|
6616 |
+
{
|
6617 |
+
"epoch": 0.8556149732620321,
|
6618 |
+
"grad_norm": 0.5098733305931091,
|
6619 |
+
"learning_rate": 1.0336662707363287e-05,
|
6620 |
+
"loss": 2.103,
|
6621 |
+
"step": 940
|
6622 |
+
},
|
6623 |
+
{
|
6624 |
+
"epoch": 0.8565252019569917,
|
6625 |
+
"grad_norm": 0.5197715163230896,
|
6626 |
+
"learning_rate": 1.0209302767636664e-05,
|
6627 |
+
"loss": 2.2107,
|
6628 |
+
"step": 941
|
6629 |
+
},
|
6630 |
+
{
|
6631 |
+
"epoch": 0.8574354306519513,
|
6632 |
+
"grad_norm": 0.547512412071228,
|
6633 |
+
"learning_rate": 1.0082690093576163e-05,
|
6634 |
+
"loss": 2.2448,
|
6635 |
+
"step": 942
|
6636 |
+
},
|
6637 |
+
{
|
6638 |
+
"epoch": 0.8583456593469109,
|
6639 |
+
"grad_norm": 0.5418568849563599,
|
6640 |
+
"learning_rate": 9.95682573889114e-06,
|
6641 |
+
"loss": 2.2423,
|
6642 |
+
"step": 943
|
6643 |
+
},
|
6644 |
+
{
|
6645 |
+
"epoch": 0.8592558880418705,
|
6646 |
+
"grad_norm": 0.5369839072227478,
|
6647 |
+
"learning_rate": 9.831710751063283e-06,
|
6648 |
+
"loss": 1.9788,
|
6649 |
+
"step": 944
|
6650 |
+
},
|
6651 |
+
{
|
6652 |
+
"epoch": 0.8601661167368301,
|
6653 |
+
"grad_norm": 0.573844313621521,
|
6654 |
+
"learning_rate": 9.707346171337894e-06,
|
6655 |
+
"loss": 2.2906,
|
6656 |
+
"step": 945
|
6657 |
+
},
|
6658 |
+
{
|
6659 |
+
"epoch": 0.8610763454317898,
|
6660 |
+
"grad_norm": 0.6142247915267944,
|
6661 |
+
"learning_rate": 9.583733034714981e-06,
|
6662 |
+
"loss": 2.3744,
|
6663 |
+
"step": 946
|
6664 |
+
},
|
6665 |
+
{
|
6666 |
+
"epoch": 0.8619865741267494,
|
6667 |
+
"grad_norm": 0.6646602153778076,
|
6668 |
+
"learning_rate": 9.460872369940955e-06,
|
6669 |
+
"loss": 2.4641,
|
6670 |
+
"step": 947
|
6671 |
+
},
|
6672 |
+
{
|
6673 |
+
"epoch": 0.862896802821709,
|
6674 |
+
"grad_norm": 0.727783739566803,
|
6675 |
+
"learning_rate": 9.338765199499854e-06,
|
6676 |
+
"loss": 2.4612,
|
6677 |
+
"step": 948
|
6678 |
+
},
|
6679 |
+
{
|
6680 |
+
"epoch": 0.8638070315166686,
|
6681 |
+
"grad_norm": 0.851578950881958,
|
6682 |
+
"learning_rate": 9.217412539604942e-06,
|
6683 |
+
"loss": 2.6441,
|
6684 |
+
"step": 949
|
6685 |
+
},
|
6686 |
+
{
|
6687 |
+
"epoch": 0.8647172602116282,
|
6688 |
+
"grad_norm": 1.461125373840332,
|
6689 |
+
"learning_rate": 9.096815400190172e-06,
|
6690 |
+
"loss": 2.4248,
|
6691 |
+
"step": 950
|
6692 |
+
},
|
6693 |
+
{
|
6694 |
+
"epoch": 0.8656274889065878,
|
6695 |
+
"grad_norm": 0.4408392608165741,
|
6696 |
+
"learning_rate": 8.97697478490188e-06,
|
6697 |
+
"loss": 2.5431,
|
6698 |
+
"step": 951
|
6699 |
+
},
|
6700 |
+
{
|
6701 |
+
"epoch": 0.8665377176015474,
|
6702 |
+
"grad_norm": 0.4102359414100647,
|
6703 |
+
"learning_rate": 8.857891691090337e-06,
|
6704 |
+
"loss": 2.3448,
|
6705 |
+
"step": 952
|
6706 |
+
},
|
6707 |
+
{
|
6708 |
+
"epoch": 0.867447946296507,
|
6709 |
+
"grad_norm": 0.4374777674674988,
|
6710 |
+
"learning_rate": 8.739567109801494e-06,
|
6711 |
+
"loss": 2.3647,
|
6712 |
+
"step": 953
|
6713 |
+
},
|
6714 |
+
{
|
6715 |
+
"epoch": 0.8683581749914666,
|
6716 |
+
"grad_norm": 0.40009114146232605,
|
6717 |
+
"learning_rate": 8.62200202576875e-06,
|
6718 |
+
"loss": 2.2401,
|
6719 |
+
"step": 954
|
6720 |
+
},
|
6721 |
+
{
|
6722 |
+
"epoch": 0.8692684036864262,
|
6723 |
+
"grad_norm": 0.42013484239578247,
|
6724 |
+
"learning_rate": 8.505197417404687e-06,
|
6725 |
+
"loss": 2.1772,
|
6726 |
+
"step": 955
|
6727 |
+
},
|
6728 |
+
{
|
6729 |
+
"epoch": 0.8701786323813858,
|
6730 |
+
"grad_norm": 0.43588119745254517,
|
6731 |
+
"learning_rate": 8.38915425679304e-06,
|
6732 |
+
"loss": 2.4605,
|
6733 |
+
"step": 956
|
6734 |
+
},
|
6735 |
+
{
|
6736 |
+
"epoch": 0.8710888610763454,
|
6737 |
+
"grad_norm": 0.4295041561126709,
|
6738 |
+
"learning_rate": 8.273873509680519e-06,
|
6739 |
+
"loss": 2.4302,
|
6740 |
+
"step": 957
|
6741 |
+
},
|
6742 |
+
{
|
6743 |
+
"epoch": 0.871999089771305,
|
6744 |
+
"grad_norm": 0.4430733621120453,
|
6745 |
+
"learning_rate": 8.15935613546872e-06,
|
6746 |
+
"loss": 2.3014,
|
6747 |
+
"step": 958
|
6748 |
+
},
|
6749 |
+
{
|
6750 |
+
"epoch": 0.8729093184662646,
|
6751 |
+
"grad_norm": 0.4275224804878235,
|
6752 |
+
"learning_rate": 8.045603087206388e-06,
|
6753 |
+
"loss": 2.251,
|
6754 |
+
"step": 959
|
6755 |
+
},
|
6756 |
+
{
|
6757 |
+
"epoch": 0.8738195471612242,
|
6758 |
+
"grad_norm": 0.4218734800815582,
|
6759 |
+
"learning_rate": 7.932615311581126e-06,
|
6760 |
+
"loss": 2.2841,
|
6761 |
+
"step": 960
|
6762 |
+
},
|
6763 |
+
{
|
6764 |
+
"epoch": 0.8747297758561838,
|
6765 |
+
"grad_norm": 0.4275785982608795,
|
6766 |
+
"learning_rate": 7.820393748911791e-06,
|
6767 |
+
"loss": 2.2751,
|
6768 |
+
"step": 961
|
6769 |
+
},
|
6770 |
+
{
|
6771 |
+
"epoch": 0.8756400045511434,
|
6772 |
+
"grad_norm": 0.40714067220687866,
|
6773 |
+
"learning_rate": 7.708939333140642e-06,
|
6774 |
+
"loss": 2.2023,
|
6775 |
+
"step": 962
|
6776 |
+
},
|
6777 |
+
{
|
6778 |
+
"epoch": 0.876550233246103,
|
6779 |
+
"grad_norm": 0.4284750521183014,
|
6780 |
+
"learning_rate": 7.598252991825372e-06,
|
6781 |
+
"loss": 2.1991,
|
6782 |
+
"step": 963
|
6783 |
+
},
|
6784 |
+
{
|
6785 |
+
"epoch": 0.8774604619410626,
|
6786 |
+
"grad_norm": 0.40348193049430847,
|
6787 |
+
"learning_rate": 7.488335646131628e-06,
|
6788 |
+
"loss": 2.1214,
|
6789 |
+
"step": 964
|
6790 |
+
},
|
6791 |
+
{
|
6792 |
+
"epoch": 0.8783706906360224,
|
6793 |
+
"grad_norm": 0.4067203998565674,
|
6794 |
+
"learning_rate": 7.3791882108251945e-06,
|
6795 |
+
"loss": 2.0977,
|
6796 |
+
"step": 965
|
6797 |
+
},
|
6798 |
+
{
|
6799 |
+
"epoch": 0.879280919330982,
|
6800 |
+
"grad_norm": 0.40969371795654297,
|
6801 |
+
"learning_rate": 7.270811594264437e-06,
|
6802 |
+
"loss": 2.1751,
|
6803 |
+
"step": 966
|
6804 |
+
},
|
6805 |
+
{
|
6806 |
+
"epoch": 0.8801911480259416,
|
6807 |
+
"grad_norm": 0.39071908593177795,
|
6808 |
+
"learning_rate": 7.163206698392744e-06,
|
6809 |
+
"loss": 2.0464,
|
6810 |
+
"step": 967
|
6811 |
+
},
|
6812 |
+
{
|
6813 |
+
"epoch": 0.8811013767209012,
|
6814 |
+
"grad_norm": 0.4038424789905548,
|
6815 |
+
"learning_rate": 7.056374418730971e-06,
|
6816 |
+
"loss": 2.1137,
|
6817 |
+
"step": 968
|
6818 |
+
},
|
6819 |
+
{
|
6820 |
+
"epoch": 0.8820116054158608,
|
6821 |
+
"grad_norm": 0.38801443576812744,
|
6822 |
+
"learning_rate": 6.950315644370075e-06,
|
6823 |
+
"loss": 1.883,
|
6824 |
+
"step": 969
|
6825 |
+
},
|
6826 |
+
{
|
6827 |
+
"epoch": 0.8829218341108204,
|
6828 |
+
"grad_norm": 0.3895006477832794,
|
6829 |
+
"learning_rate": 6.845031257963619e-06,
|
6830 |
+
"loss": 2.0169,
|
6831 |
+
"step": 970
|
6832 |
+
},
|
6833 |
+
{
|
6834 |
+
"epoch": 0.88383206280578,
|
6835 |
+
"grad_norm": 0.413171648979187,
|
6836 |
+
"learning_rate": 6.740522135720517e-06,
|
6837 |
+
"loss": 2.2054,
|
6838 |
+
"step": 971
|
6839 |
+
},
|
6840 |
+
{
|
6841 |
+
"epoch": 0.8847422915007396,
|
6842 |
+
"grad_norm": 0.4204188585281372,
|
6843 |
+
"learning_rate": 6.636789147397637e-06,
|
6844 |
+
"loss": 2.1765,
|
6845 |
+
"step": 972
|
6846 |
+
},
|
6847 |
+
{
|
6848 |
+
"epoch": 0.8856525201956992,
|
6849 |
+
"grad_norm": 0.4209098517894745,
|
6850 |
+
"learning_rate": 6.533833156292679e-06,
|
6851 |
+
"loss": 1.9617,
|
6852 |
+
"step": 973
|
6853 |
+
},
|
6854 |
+
{
|
6855 |
+
"epoch": 0.8865627488906588,
|
6856 |
+
"grad_norm": 0.4256611168384552,
|
6857 |
+
"learning_rate": 6.431655019236948e-06,
|
6858 |
+
"loss": 2.108,
|
6859 |
+
"step": 974
|
6860 |
+
},
|
6861 |
+
{
|
6862 |
+
"epoch": 0.8874729775856184,
|
6863 |
+
"grad_norm": 0.43669816851615906,
|
6864 |
+
"learning_rate": 6.3302555865880965e-06,
|
6865 |
+
"loss": 2.0991,
|
6866 |
+
"step": 975
|
6867 |
+
},
|
6868 |
+
{
|
6869 |
+
"epoch": 0.888383206280578,
|
6870 |
+
"grad_norm": 0.44833648204803467,
|
6871 |
+
"learning_rate": 6.229635702223324e-06,
|
6872 |
+
"loss": 2.2335,
|
6873 |
+
"step": 976
|
6874 |
+
},
|
6875 |
+
{
|
6876 |
+
"epoch": 0.8892934349755376,
|
6877 |
+
"grad_norm": 0.45070621371269226,
|
6878 |
+
"learning_rate": 6.129796203532057e-06,
|
6879 |
+
"loss": 2.2487,
|
6880 |
+
"step": 977
|
6881 |
+
},
|
6882 |
+
{
|
6883 |
+
"epoch": 0.8902036636704972,
|
6884 |
+
"grad_norm": 0.4609052538871765,
|
6885 |
+
"learning_rate": 6.030737921409169e-06,
|
6886 |
+
"loss": 2.1147,
|
6887 |
+
"step": 978
|
6888 |
+
},
|
6889 |
+
{
|
6890 |
+
"epoch": 0.8911138923654568,
|
6891 |
+
"grad_norm": 0.4470416307449341,
|
6892 |
+
"learning_rate": 5.932461680248014e-06,
|
6893 |
+
"loss": 2.0615,
|
6894 |
+
"step": 979
|
6895 |
+
},
|
6896 |
+
{
|
6897 |
+
"epoch": 0.8920241210604164,
|
6898 |
+
"grad_norm": 0.42171233892440796,
|
6899 |
+
"learning_rate": 5.834968297933541e-06,
|
6900 |
+
"loss": 2.0669,
|
6901 |
+
"step": 980
|
6902 |
+
},
|
6903 |
+
{
|
6904 |
+
"epoch": 0.892934349755376,
|
6905 |
+
"grad_norm": 0.4385877251625061,
|
6906 |
+
"learning_rate": 5.738258585835532e-06,
|
6907 |
+
"loss": 1.9846,
|
6908 |
+
"step": 981
|
6909 |
+
},
|
6910 |
+
{
|
6911 |
+
"epoch": 0.8938445784503356,
|
6912 |
+
"grad_norm": 0.4574371576309204,
|
6913 |
+
"learning_rate": 5.6423333488018095e-06,
|
6914 |
+
"loss": 2.114,
|
6915 |
+
"step": 982
|
6916 |
+
},
|
6917 |
+
{
|
6918 |
+
"epoch": 0.8947548071452952,
|
6919 |
+
"grad_norm": 0.46896499395370483,
|
6920 |
+
"learning_rate": 5.547193385151561e-06,
|
6921 |
+
"loss": 2.0444,
|
6922 |
+
"step": 983
|
6923 |
+
},
|
6924 |
+
{
|
6925 |
+
"epoch": 0.8956650358402549,
|
6926 |
+
"grad_norm": 0.45737412571907043,
|
6927 |
+
"learning_rate": 5.45283948666866e-06,
|
6928 |
+
"loss": 2.0976,
|
6929 |
+
"step": 984
|
6930 |
+
},
|
6931 |
+
{
|
6932 |
+
"epoch": 0.8965752645352145,
|
6933 |
+
"grad_norm": 0.47739726305007935,
|
6934 |
+
"learning_rate": 5.359272438595153e-06,
|
6935 |
+
"loss": 2.1393,
|
6936 |
+
"step": 985
|
6937 |
+
},
|
6938 |
+
{
|
6939 |
+
"epoch": 0.8974854932301741,
|
6940 |
+
"grad_norm": 0.47124338150024414,
|
6941 |
+
"learning_rate": 5.266493019624663e-06,
|
6942 |
+
"loss": 2.0509,
|
6943 |
+
"step": 986
|
6944 |
+
},
|
6945 |
+
{
|
6946 |
+
"epoch": 0.8983957219251337,
|
6947 |
+
"grad_norm": 0.4660322070121765,
|
6948 |
+
"learning_rate": 5.1745020018958866e-06,
|
6949 |
+
"loss": 1.9704,
|
6950 |
+
"step": 987
|
6951 |
+
},
|
6952 |
+
{
|
6953 |
+
"epoch": 0.8993059506200933,
|
6954 |
+
"grad_norm": 0.48330241441726685,
|
6955 |
+
"learning_rate": 5.083300150986259e-06,
|
6956 |
+
"loss": 2.021,
|
6957 |
+
"step": 988
|
6958 |
+
},
|
6959 |
+
{
|
6960 |
+
"epoch": 0.9002161793150529,
|
6961 |
+
"grad_norm": 0.51470547914505,
|
6962 |
+
"learning_rate": 4.992888225905468e-06,
|
6963 |
+
"loss": 2.1097,
|
6964 |
+
"step": 989
|
6965 |
+
},
|
6966 |
+
{
|
6967 |
+
"epoch": 0.9011264080100125,
|
6968 |
+
"grad_norm": 0.516373336315155,
|
6969 |
+
"learning_rate": 4.903266979089249e-06,
|
6970 |
+
"loss": 2.1694,
|
6971 |
+
"step": 990
|
6972 |
+
},
|
6973 |
+
{
|
6974 |
+
"epoch": 0.9020366367049721,
|
6975 |
+
"grad_norm": 0.5256400108337402,
|
6976 |
+
"learning_rate": 4.8144371563930476e-06,
|
6977 |
+
"loss": 2.2843,
|
6978 |
+
"step": 991
|
6979 |
+
},
|
6980 |
+
{
|
6981 |
+
"epoch": 0.9029468653999317,
|
6982 |
+
"grad_norm": 0.5559744238853455,
|
6983 |
+
"learning_rate": 4.726399497085832e-06,
|
6984 |
+
"loss": 2.2733,
|
6985 |
+
"step": 992
|
6986 |
+
},
|
6987 |
+
{
|
6988 |
+
"epoch": 0.9038570940948913,
|
6989 |
+
"grad_norm": 0.5462202429771423,
|
6990 |
+
"learning_rate": 4.6391547338439536e-06,
|
6991 |
+
"loss": 2.1758,
|
6992 |
+
"step": 993
|
6993 |
+
},
|
6994 |
+
{
|
6995 |
+
"epoch": 0.904767322789851,
|
6996 |
+
"grad_norm": 0.5769087672233582,
|
6997 |
+
"learning_rate": 4.552703592745033e-06,
|
6998 |
+
"loss": 2.2552,
|
6999 |
+
"step": 994
|
7000 |
+
},
|
7001 |
+
{
|
7002 |
+
"epoch": 0.9056775514848106,
|
7003 |
+
"grad_norm": 0.5632253289222717,
|
7004 |
+
"learning_rate": 4.467046793261931e-06,
|
7005 |
+
"loss": 2.2402,
|
7006 |
+
"step": 995
|
7007 |
+
},
|
7008 |
+
{
|
7009 |
+
"epoch": 0.9065877801797702,
|
7010 |
+
"grad_norm": 0.610163688659668,
|
7011 |
+
"learning_rate": 4.3821850482567595e-06,
|
7012 |
+
"loss": 2.4484,
|
7013 |
+
"step": 996
|
7014 |
+
},
|
7015 |
+
{
|
7016 |
+
"epoch": 0.9074980088747298,
|
7017 |
+
"grad_norm": 0.6236492395401001,
|
7018 |
+
"learning_rate": 4.298119063974914e-06,
|
7019 |
+
"loss": 2.1914,
|
7020 |
+
"step": 997
|
7021 |
+
},
|
7022 |
+
{
|
7023 |
+
"epoch": 0.9084082375696894,
|
7024 |
+
"grad_norm": 0.7362584471702576,
|
7025 |
+
"learning_rate": 4.214849540039267e-06,
|
7026 |
+
"loss": 2.5582,
|
7027 |
+
"step": 998
|
7028 |
+
},
|
7029 |
+
{
|
7030 |
+
"epoch": 0.909318466264649,
|
7031 |
+
"grad_norm": 0.8780522346496582,
|
7032 |
+
"learning_rate": 4.132377169444279e-06,
|
7033 |
+
"loss": 2.5269,
|
7034 |
+
"step": 999
|
7035 |
+
},
|
7036 |
+
{
|
7037 |
+
"epoch": 0.9102286949596086,
|
7038 |
+
"grad_norm": 1.4850975275039673,
|
7039 |
+
"learning_rate": 4.050702638550275e-06,
|
7040 |
+
"loss": 2.3857,
|
7041 |
+
"step": 1000
|
7042 |
+
},
|
7043 |
+
{
|
7044 |
+
"epoch": 0.9111389236545682,
|
7045 |
+
"grad_norm": 0.4350475072860718,
|
7046 |
+
"learning_rate": 3.969826627077655e-06,
|
7047 |
+
"loss": 2.5653,
|
7048 |
+
"step": 1001
|
7049 |
+
},
|
7050 |
+
{
|
7051 |
+
"epoch": 0.9120491523495278,
|
7052 |
+
"grad_norm": 0.44277626276016235,
|
7053 |
+
"learning_rate": 3.889749808101395e-06,
|
7054 |
+
"loss": 2.3969,
|
7055 |
+
"step": 1002
|
7056 |
+
},
|
7057 |
+
{
|
7058 |
+
"epoch": 0.9129593810444874,
|
7059 |
+
"grad_norm": 0.44005268812179565,
|
7060 |
+
"learning_rate": 3.810472848045266e-06,
|
7061 |
+
"loss": 2.6065,
|
7062 |
+
"step": 1003
|
7063 |
+
},
|
7064 |
+
{
|
7065 |
+
"epoch": 0.9138696097394471,
|
7066 |
+
"grad_norm": 0.41925248503685,
|
7067 |
+
"learning_rate": 3.7319964066763858e-06,
|
7068 |
+
"loss": 2.3878,
|
7069 |
+
"step": 1004
|
7070 |
+
},
|
7071 |
+
{
|
7072 |
+
"epoch": 0.9147798384344067,
|
7073 |
+
"grad_norm": 0.4320535957813263,
|
7074 |
+
"learning_rate": 3.6543211370997587e-06,
|
7075 |
+
"loss": 2.3829,
|
7076 |
+
"step": 1005
|
7077 |
+
},
|
7078 |
+
{
|
7079 |
+
"epoch": 0.9156900671293663,
|
7080 |
+
"grad_norm": 0.43817150592803955,
|
7081 |
+
"learning_rate": 3.5774476857527107e-06,
|
7082 |
+
"loss": 2.3854,
|
7083 |
+
"step": 1006
|
7084 |
+
},
|
7085 |
+
{
|
7086 |
+
"epoch": 0.9166002958243259,
|
7087 |
+
"grad_norm": 0.42210131883621216,
|
7088 |
+
"learning_rate": 3.5013766923996604e-06,
|
7089 |
+
"loss": 2.2874,
|
7090 |
+
"step": 1007
|
7091 |
+
},
|
7092 |
+
{
|
7093 |
+
"epoch": 0.9175105245192855,
|
7094 |
+
"grad_norm": 0.41610825061798096,
|
7095 |
+
"learning_rate": 3.426108790126681e-06,
|
7096 |
+
"loss": 2.3301,
|
7097 |
+
"step": 1008
|
7098 |
+
},
|
7099 |
+
{
|
7100 |
+
"epoch": 0.9184207532142451,
|
7101 |
+
"grad_norm": 0.42343541979789734,
|
7102 |
+
"learning_rate": 3.3516446053363015e-06,
|
7103 |
+
"loss": 2.2083,
|
7104 |
+
"step": 1009
|
7105 |
+
},
|
7106 |
+
{
|
7107 |
+
"epoch": 0.9193309819092047,
|
7108 |
+
"grad_norm": 0.4323045015335083,
|
7109 |
+
"learning_rate": 3.2779847577422697e-06,
|
7110 |
+
"loss": 2.2401,
|
7111 |
+
"step": 1010
|
7112 |
+
},
|
7113 |
+
{
|
7114 |
+
"epoch": 0.9202412106041643,
|
7115 |
+
"grad_norm": 0.4198078513145447,
|
7116 |
+
"learning_rate": 3.2051298603643753e-06,
|
7117 |
+
"loss": 2.0988,
|
7118 |
+
"step": 1011
|
7119 |
+
},
|
7120 |
+
{
|
7121 |
+
"epoch": 0.9211514392991239,
|
7122 |
+
"grad_norm": 0.4277539551258087,
|
7123 |
+
"learning_rate": 3.133080519523368e-06,
|
7124 |
+
"loss": 2.3482,
|
7125 |
+
"step": 1012
|
7126 |
+
},
|
7127 |
+
{
|
7128 |
+
"epoch": 0.9220616679940835,
|
7129 |
+
"grad_norm": 0.42749837040901184,
|
7130 |
+
"learning_rate": 3.0618373348359264e-06,
|
7131 |
+
"loss": 2.3242,
|
7132 |
+
"step": 1013
|
7133 |
+
},
|
7134 |
+
{
|
7135 |
+
"epoch": 0.9229718966890431,
|
7136 |
+
"grad_norm": 0.4157456159591675,
|
7137 |
+
"learning_rate": 2.991400899209651e-06,
|
7138 |
+
"loss": 2.11,
|
7139 |
+
"step": 1014
|
7140 |
+
},
|
7141 |
+
{
|
7142 |
+
"epoch": 0.9238821253840027,
|
7143 |
+
"grad_norm": 0.41514283418655396,
|
7144 |
+
"learning_rate": 2.921771798838069e-06,
|
7145 |
+
"loss": 2.0979,
|
7146 |
+
"step": 1015
|
7147 |
+
},
|
7148 |
+
{
|
7149 |
+
"epoch": 0.9247923540789623,
|
7150 |
+
"grad_norm": 0.4146190285682678,
|
7151 |
+
"learning_rate": 2.852950613195915e-06,
|
7152 |
+
"loss": 2.1057,
|
7153 |
+
"step": 1016
|
7154 |
+
},
|
7155 |
+
{
|
7156 |
+
"epoch": 0.9257025827739219,
|
7157 |
+
"grad_norm": 0.4031788110733032,
|
7158 |
+
"learning_rate": 2.784937915034169e-06,
|
7159 |
+
"loss": 2.1094,
|
7160 |
+
"step": 1017
|
7161 |
+
},
|
7162 |
+
{
|
7163 |
+
"epoch": 0.9266128114688815,
|
7164 |
+
"grad_norm": 0.4135347604751587,
|
7165 |
+
"learning_rate": 2.717734270375272e-06,
|
7166 |
+
"loss": 2.2154,
|
7167 |
+
"step": 1018
|
7168 |
+
},
|
7169 |
+
{
|
7170 |
+
"epoch": 0.9275230401638411,
|
7171 |
+
"grad_norm": 0.40153443813323975,
|
7172 |
+
"learning_rate": 2.6513402385085704e-06,
|
7173 |
+
"loss": 2.0342,
|
7174 |
+
"step": 1019
|
7175 |
+
},
|
7176 |
+
{
|
7177 |
+
"epoch": 0.9284332688588007,
|
7178 |
+
"grad_norm": 0.4011882543563843,
|
7179 |
+
"learning_rate": 2.585756371985493e-06,
|
7180 |
+
"loss": 1.9751,
|
7181 |
+
"step": 1020
|
7182 |
+
},
|
7183 |
+
{
|
7184 |
+
"epoch": 0.9293434975537603,
|
7185 |
+
"grad_norm": 0.4032374322414398,
|
7186 |
+
"learning_rate": 2.520983216615047e-06,
|
7187 |
+
"loss": 2.0868,
|
7188 |
+
"step": 1021
|
7189 |
+
},
|
7190 |
+
{
|
7191 |
+
"epoch": 0.9302537262487199,
|
7192 |
+
"grad_norm": 0.4052782952785492,
|
7193 |
+
"learning_rate": 2.4570213114592954e-06,
|
7194 |
+
"loss": 2.0716,
|
7195 |
+
"step": 1022
|
7196 |
+
},
|
7197 |
+
{
|
7198 |
+
"epoch": 0.9311639549436797,
|
7199 |
+
"grad_norm": 0.4261015057563782,
|
7200 |
+
"learning_rate": 2.393871188828767e-06,
|
7201 |
+
"loss": 2.1153,
|
7202 |
+
"step": 1023
|
7203 |
+
},
|
7204 |
+
{
|
7205 |
+
"epoch": 0.9320741836386393,
|
7206 |
+
"grad_norm": 0.4141393303871155,
|
7207 |
+
"learning_rate": 2.3315333742780942e-06,
|
7208 |
+
"loss": 2.0839,
|
7209 |
+
"step": 1024
|
7210 |
+
},
|
7211 |
+
{
|
7212 |
+
"epoch": 0.9329844123335989,
|
7213 |
+
"grad_norm": 0.41185298562049866,
|
7214 |
+
"learning_rate": 2.270008386601685e-06,
|
7215 |
+
"loss": 2.0686,
|
7216 |
+
"step": 1025
|
7217 |
+
},
|
7218 |
+
{
|
7219 |
+
"epoch": 0.9338946410285585,
|
7220 |
+
"grad_norm": 0.41932380199432373,
|
7221 |
+
"learning_rate": 2.2092967378292915e-06,
|
7222 |
+
"loss": 2.0688,
|
7223 |
+
"step": 1026
|
7224 |
+
},
|
7225 |
+
{
|
7226 |
+
"epoch": 0.9348048697235181,
|
7227 |
+
"grad_norm": 0.430480033159256,
|
7228 |
+
"learning_rate": 2.1493989332218468e-06,
|
7229 |
+
"loss": 2.1202,
|
7230 |
+
"step": 1027
|
7231 |
+
},
|
7232 |
+
{
|
7233 |
+
"epoch": 0.9357150984184777,
|
7234 |
+
"grad_norm": 0.4182969629764557,
|
7235 |
+
"learning_rate": 2.0903154712672237e-06,
|
7236 |
+
"loss": 1.8457,
|
7237 |
+
"step": 1028
|
7238 |
+
},
|
7239 |
+
{
|
7240 |
+
"epoch": 0.9366253271134373,
|
7241 |
+
"grad_norm": 0.44270989298820496,
|
7242 |
+
"learning_rate": 2.032046843676061e-06,
|
7243 |
+
"loss": 2.2296,
|
7244 |
+
"step": 1029
|
7245 |
+
},
|
7246 |
+
{
|
7247 |
+
"epoch": 0.9375355558083969,
|
7248 |
+
"grad_norm": 0.4312398433685303,
|
7249 |
+
"learning_rate": 1.974593535377722e-06,
|
7250 |
+
"loss": 1.9802,
|
7251 |
+
"step": 1030
|
7252 |
+
},
|
7253 |
+
{
|
7254 |
+
"epoch": 0.9384457845033565,
|
7255 |
+
"grad_norm": 0.4557861387729645,
|
7256 |
+
"learning_rate": 1.917956024516243e-06,
|
7257 |
+
"loss": 1.9306,
|
7258 |
+
"step": 1031
|
7259 |
+
},
|
7260 |
+
{
|
7261 |
+
"epoch": 0.9393560131983161,
|
7262 |
+
"grad_norm": 0.4993920624256134,
|
7263 |
+
"learning_rate": 1.8621347824462787e-06,
|
7264 |
+
"loss": 2.275,
|
7265 |
+
"step": 1032
|
7266 |
+
},
|
7267 |
+
{
|
7268 |
+
"epoch": 0.9402662418932757,
|
7269 |
+
"grad_norm": 0.4622466266155243,
|
7270 |
+
"learning_rate": 1.8071302737293295e-06,
|
7271 |
+
"loss": 2.1966,
|
7272 |
+
"step": 1033
|
7273 |
+
},
|
7274 |
+
{
|
7275 |
+
"epoch": 0.9411764705882353,
|
7276 |
+
"grad_norm": 0.4710349142551422,
|
7277 |
+
"learning_rate": 1.752942956129744e-06,
|
7278 |
+
"loss": 2.1353,
|
7279 |
+
"step": 1034
|
7280 |
+
},
|
7281 |
+
{
|
7282 |
+
"epoch": 0.9420866992831949,
|
7283 |
+
"grad_norm": 0.4602985680103302,
|
7284 |
+
"learning_rate": 1.6995732806109554e-06,
|
7285 |
+
"loss": 2.0855,
|
7286 |
+
"step": 1035
|
7287 |
+
},
|
7288 |
+
{
|
7289 |
+
"epoch": 0.9429969279781545,
|
7290 |
+
"grad_norm": 0.4736422002315521,
|
7291 |
+
"learning_rate": 1.6470216913317626e-06,
|
7292 |
+
"loss": 2.081,
|
7293 |
+
"step": 1036
|
7294 |
+
},
|
7295 |
+
{
|
7296 |
+
"epoch": 0.9439071566731141,
|
7297 |
+
"grad_norm": 0.4875909984111786,
|
7298 |
+
"learning_rate": 1.5952886256425547e-06,
|
7299 |
+
"loss": 2.1607,
|
7300 |
+
"step": 1037
|
7301 |
+
},
|
7302 |
+
{
|
7303 |
+
"epoch": 0.9448173853680737,
|
7304 |
+
"grad_norm": 0.5093516111373901,
|
7305 |
+
"learning_rate": 1.5443745140817366e-06,
|
7306 |
+
"loss": 2.2025,
|
7307 |
+
"step": 1038
|
7308 |
+
},
|
7309 |
+
{
|
7310 |
+
"epoch": 0.9457276140630333,
|
7311 |
+
"grad_norm": 0.5034651160240173,
|
7312 |
+
"learning_rate": 1.4942797803721543e-06,
|
7313 |
+
"loss": 1.9985,
|
7314 |
+
"step": 1039
|
7315 |
+
},
|
7316 |
+
{
|
7317 |
+
"epoch": 0.9466378427579929,
|
7318 |
+
"grad_norm": 0.5074111819267273,
|
7319 |
+
"learning_rate": 1.4450048414174854e-06,
|
7320 |
+
"loss": 2.1175,
|
7321 |
+
"step": 1040
|
7322 |
+
},
|
7323 |
+
{
|
7324 |
+
"epoch": 0.9475480714529525,
|
7325 |
+
"grad_norm": 0.5183742046356201,
|
7326 |
+
"learning_rate": 1.3965501072988663e-06,
|
7327 |
+
"loss": 2.0718,
|
7328 |
+
"step": 1041
|
7329 |
+
},
|
7330 |
+
{
|
7331 |
+
"epoch": 0.9484583001479122,
|
7332 |
+
"grad_norm": 0.5284718871116638,
|
7333 |
+
"learning_rate": 1.348915981271437e-06,
|
7334 |
+
"loss": 2.1586,
|
7335 |
+
"step": 1042
|
7336 |
+
},
|
7337 |
+
{
|
7338 |
+
"epoch": 0.9493685288428718,
|
7339 |
+
"grad_norm": 0.545464813709259,
|
7340 |
+
"learning_rate": 1.3021028597609675e-06,
|
7341 |
+
"loss": 2.2445,
|
7342 |
+
"step": 1043
|
7343 |
+
},
|
7344 |
+
{
|
7345 |
+
"epoch": 0.9502787575378314,
|
7346 |
+
"grad_norm": 0.5713001489639282,
|
7347 |
+
"learning_rate": 1.2561111323605712e-06,
|
7348 |
+
"loss": 2.1888,
|
7349 |
+
"step": 1044
|
7350 |
+
},
|
7351 |
+
{
|
7352 |
+
"epoch": 0.951188986232791,
|
7353 |
+
"grad_norm": 0.5774447321891785,
|
7354 |
+
"learning_rate": 1.2109411818274852e-06,
|
7355 |
+
"loss": 2.2029,
|
7356 |
+
"step": 1045
|
7357 |
+
},
|
7358 |
+
{
|
7359 |
+
"epoch": 0.9520992149277506,
|
7360 |
+
"grad_norm": 0.6209971308708191,
|
7361 |
+
"learning_rate": 1.1665933840798838e-06,
|
7362 |
+
"loss": 2.1735,
|
7363 |
+
"step": 1046
|
7364 |
+
},
|
7365 |
+
{
|
7366 |
+
"epoch": 0.9530094436227102,
|
7367 |
+
"grad_norm": 0.6675162315368652,
|
7368 |
+
"learning_rate": 1.1230681081936923e-06,
|
7369 |
+
"loss": 2.4231,
|
7370 |
+
"step": 1047
|
7371 |
+
},
|
7372 |
+
{
|
7373 |
+
"epoch": 0.9539196723176698,
|
7374 |
+
"grad_norm": 0.7409051060676575,
|
7375 |
+
"learning_rate": 1.0803657163995895e-06,
|
7376 |
+
"loss": 2.45,
|
7377 |
+
"step": 1048
|
7378 |
+
},
|
7379 |
+
{
|
7380 |
+
"epoch": 0.9548299010126294,
|
7381 |
+
"grad_norm": 0.8856377601623535,
|
7382 |
+
"learning_rate": 1.0384865640799435e-06,
|
7383 |
+
"loss": 2.4769,
|
7384 |
+
"step": 1049
|
7385 |
+
},
|
7386 |
+
{
|
7387 |
+
"epoch": 0.955740129707589,
|
7388 |
+
"grad_norm": 1.4897712469100952,
|
7389 |
+
"learning_rate": 9.974309997658915e-07,
|
7390 |
+
"loss": 2.6822,
|
7391 |
+
"step": 1050
|
7392 |
+
},
|
7393 |
+
{
|
7394 |
+
"epoch": 0.9566503584025486,
|
7395 |
+
"grad_norm": 0.4162799119949341,
|
7396 |
+
"learning_rate": 9.57199365134387e-07,
|
7397 |
+
"loss": 2.493,
|
7398 |
+
"step": 1051
|
7399 |
+
},
|
7400 |
+
{
|
7401 |
+
"epoch": 0.9575605870975082,
|
7402 |
+
"grad_norm": 0.4336461126804352,
|
7403 |
+
"learning_rate": 9.177919950054237e-07,
|
7404 |
+
"loss": 2.4071,
|
7405 |
+
"step": 1052
|
7406 |
+
},
|
7407 |
+
{
|
7408 |
+
"epoch": 0.9584708157924678,
|
7409 |
+
"grad_norm": 0.42869681119918823,
|
7410 |
+
"learning_rate": 8.792092173391831e-07,
|
7411 |
+
"loss": 2.3585,
|
7412 |
+
"step": 1053
|
7413 |
+
},
|
7414 |
+
{
|
7415 |
+
"epoch": 0.9593810444874274,
|
7416 |
+
"grad_norm": 0.4060511887073517,
|
7417 |
+
"learning_rate": 8.41451353233369e-07,
|
7418 |
+
"loss": 2.3244,
|
7419 |
+
"step": 1054
|
7420 |
+
},
|
7421 |
+
{
|
7422 |
+
"epoch": 0.960291273182387,
|
7423 |
+
"grad_norm": 0.42351028323173523,
|
7424 |
+
"learning_rate": 8.04518716920466e-07,
|
7425 |
+
"loss": 2.4194,
|
7426 |
+
"step": 1055
|
7427 |
+
},
|
7428 |
+
{
|
7429 |
+
"epoch": 0.9612015018773467,
|
7430 |
+
"grad_norm": 0.42555585503578186,
|
7431 |
+
"learning_rate": 7.684116157651966e-07,
|
7432 |
+
"loss": 2.368,
|
7433 |
+
"step": 1056
|
7434 |
+
},
|
7435 |
+
{
|
7436 |
+
"epoch": 0.9621117305723063,
|
7437 |
+
"grad_norm": 0.4169003665447235,
|
7438 |
+
"learning_rate": 7.331303502618903e-07,
|
7439 |
+
"loss": 2.3947,
|
7440 |
+
"step": 1057
|
7441 |
+
},
|
7442 |
+
{
|
7443 |
+
"epoch": 0.9630219592672659,
|
7444 |
+
"grad_norm": 0.43154704570770264,
|
7445 |
+
"learning_rate": 6.986752140320518e-07,
|
7446 |
+
"loss": 2.2809,
|
7447 |
+
"step": 1058
|
7448 |
+
},
|
7449 |
+
{
|
7450 |
+
"epoch": 0.9639321879622255,
|
7451 |
+
"grad_norm": 0.40992870926856995,
|
7452 |
+
"learning_rate": 6.650464938218637e-07,
|
7453 |
+
"loss": 2.2827,
|
7454 |
+
"step": 1059
|
7455 |
+
},
|
7456 |
+
{
|
7457 |
+
"epoch": 0.9648424166571851,
|
7458 |
+
"grad_norm": 0.4168105125427246,
|
7459 |
+
"learning_rate": 6.322444694998319e-07,
|
7460 |
+
"loss": 2.3334,
|
7461 |
+
"step": 1060
|
7462 |
+
},
|
7463 |
+
{
|
7464 |
+
"epoch": 0.9657526453521448,
|
7465 |
+
"grad_norm": 0.43256238102912903,
|
7466 |
+
"learning_rate": 6.002694140544329e-07,
|
7467 |
+
"loss": 2.2669,
|
7468 |
+
"step": 1061
|
7469 |
+
},
|
7470 |
+
{
|
7471 |
+
"epoch": 0.9666628740471044,
|
7472 |
+
"grad_norm": 0.42290446162223816,
|
7473 |
+
"learning_rate": 5.691215935918815e-07,
|
7474 |
+
"loss": 2.1027,
|
7475 |
+
"step": 1062
|
7476 |
+
},
|
7477 |
+
{
|
7478 |
+
"epoch": 0.967573102742064,
|
7479 |
+
"grad_norm": 0.4109112024307251,
|
7480 |
+
"learning_rate": 5.388012673338661e-07,
|
7481 |
+
"loss": 2.15,
|
7482 |
+
"step": 1063
|
7483 |
+
},
|
7484 |
+
{
|
7485 |
+
"epoch": 0.9684833314370236,
|
7486 |
+
"grad_norm": 0.4292824864387512,
|
7487 |
+
"learning_rate": 5.093086876154174e-07,
|
7488 |
+
"loss": 2.2394,
|
7489 |
+
"step": 1064
|
7490 |
+
},
|
7491 |
+
{
|
7492 |
+
"epoch": 0.9693935601319832,
|
7493 |
+
"grad_norm": 0.4136911928653717,
|
7494 |
+
"learning_rate": 4.80644099882821e-07,
|
7495 |
+
"loss": 2.1904,
|
7496 |
+
"step": 1065
|
7497 |
+
},
|
7498 |
+
{
|
7499 |
+
"epoch": 0.9703037888269428,
|
7500 |
+
"grad_norm": 0.4183078110218048,
|
7501 |
+
"learning_rate": 4.5280774269154115e-07,
|
7502 |
+
"loss": 2.1499,
|
7503 |
+
"step": 1066
|
7504 |
+
},
|
7505 |
+
{
|
7506 |
+
"epoch": 0.9712140175219024,
|
7507 |
+
"grad_norm": 0.41571420431137085,
|
7508 |
+
"learning_rate": 4.2579984770426686e-07,
|
7509 |
+
"loss": 2.0975,
|
7510 |
+
"step": 1067
|
7511 |
+
},
|
7512 |
+
{
|
7513 |
+
"epoch": 0.972124246216862,
|
7514 |
+
"grad_norm": 0.39653652906417847,
|
7515 |
+
"learning_rate": 3.99620639688969e-07,
|
7516 |
+
"loss": 1.9686,
|
7517 |
+
"step": 1068
|
7518 |
+
},
|
7519 |
+
{
|
7520 |
+
"epoch": 0.9730344749118216,
|
7521 |
+
"grad_norm": 0.4133754372596741,
|
7522 |
+
"learning_rate": 3.742703365170241e-07,
|
7523 |
+
"loss": 2.0801,
|
7524 |
+
"step": 1069
|
7525 |
+
},
|
7526 |
+
{
|
7527 |
+
"epoch": 0.9739447036067812,
|
7528 |
+
"grad_norm": 0.41434839367866516,
|
7529 |
+
"learning_rate": 3.497491491614158e-07,
|
7530 |
+
"loss": 1.9311,
|
7531 |
+
"step": 1070
|
7532 |
+
},
|
7533 |
+
{
|
7534 |
+
"epoch": 0.9748549323017408,
|
7535 |
+
"grad_norm": 0.41751164197921753,
|
7536 |
+
"learning_rate": 3.260572816949692e-07,
|
7537 |
+
"loss": 2.1841,
|
7538 |
+
"step": 1071
|
7539 |
+
},
|
7540 |
+
{
|
7541 |
+
"epoch": 0.9757651609967004,
|
7542 |
+
"grad_norm": 0.41821038722991943,
|
7543 |
+
"learning_rate": 3.0319493128866396e-07,
|
7544 |
+
"loss": 2.0719,
|
7545 |
+
"step": 1072
|
7546 |
+
},
|
7547 |
+
{
|
7548 |
+
"epoch": 0.97667538969166,
|
7549 |
+
"grad_norm": 0.4422387480735779,
|
7550 |
+
"learning_rate": 2.8116228820997957e-07,
|
7551 |
+
"loss": 2.2655,
|
7552 |
+
"step": 1073
|
7553 |
+
},
|
7554 |
+
{
|
7555 |
+
"epoch": 0.9775856183866196,
|
7556 |
+
"grad_norm": 0.4252420961856842,
|
7557 |
+
"learning_rate": 2.5995953582130804e-07,
|
7558 |
+
"loss": 2.098,
|
7559 |
+
"step": 1074
|
7560 |
+
},
|
7561 |
+
{
|
7562 |
+
"epoch": 0.9784958470815792,
|
7563 |
+
"grad_norm": 0.4352913498878479,
|
7564 |
+
"learning_rate": 2.395868505784438e-07,
|
7565 |
+
"loss": 2.1038,
|
7566 |
+
"step": 1075
|
7567 |
+
},
|
7568 |
+
{
|
7569 |
+
"epoch": 0.9794060757765388,
|
7570 |
+
"grad_norm": 0.43940603733062744,
|
7571 |
+
"learning_rate": 2.2004440202911814e-07,
|
7572 |
+
"loss": 2.1152,
|
7573 |
+
"step": 1076
|
7574 |
+
},
|
7575 |
+
{
|
7576 |
+
"epoch": 0.9803163044714984,
|
7577 |
+
"grad_norm": 0.43396398425102234,
|
7578 |
+
"learning_rate": 2.0133235281156736e-07,
|
7579 |
+
"loss": 2.0175,
|
7580 |
+
"step": 1077
|
7581 |
+
},
|
7582 |
+
{
|
7583 |
+
"epoch": 0.981226533166458,
|
7584 |
+
"grad_norm": 0.43348294496536255,
|
7585 |
+
"learning_rate": 1.83450858653178e-07,
|
7586 |
+
"loss": 2.0593,
|
7587 |
+
"step": 1078
|
7588 |
+
},
|
7589 |
+
{
|
7590 |
+
"epoch": 0.9821367618614176,
|
7591 |
+
"grad_norm": 0.43314129114151,
|
7592 |
+
"learning_rate": 1.664000683692324e-07,
|
7593 |
+
"loss": 1.9414,
|
7594 |
+
"step": 1079
|
7595 |
+
},
|
7596 |
+
{
|
7597 |
+
"epoch": 0.9830469905563773,
|
7598 |
+
"grad_norm": 0.47113651037216187,
|
7599 |
+
"learning_rate": 1.5018012386162072e-07,
|
7600 |
+
"loss": 2.3204,
|
7601 |
+
"step": 1080
|
7602 |
+
},
|
7603 |
+
{
|
7604 |
+
"epoch": 0.983957219251337,
|
7605 |
+
"grad_norm": 0.43835383653640747,
|
7606 |
+
"learning_rate": 1.3479116011769767e-07,
|
7607 |
+
"loss": 2.0271,
|
7608 |
+
"step": 1081
|
7609 |
+
},
|
7610 |
+
{
|
7611 |
+
"epoch": 0.9848674479462965,
|
7612 |
+
"grad_norm": 0.42517396807670593,
|
7613 |
+
"learning_rate": 1.2023330520911646e-07,
|
7614 |
+
"loss": 1.8366,
|
7615 |
+
"step": 1082
|
7616 |
+
},
|
7617 |
+
{
|
7618 |
+
"epoch": 0.9857776766412562,
|
7619 |
+
"grad_norm": 0.4593140482902527,
|
7620 |
+
"learning_rate": 1.0650668029079658e-07,
|
7621 |
+
"loss": 2.0962,
|
7622 |
+
"step": 1083
|
7623 |
+
},
|
7624 |
+
{
|
7625 |
+
"epoch": 0.9866879053362158,
|
7626 |
+
"grad_norm": 0.4892318546772003,
|
7627 |
+
"learning_rate": 9.361139959993549e-08,
|
7628 |
+
"loss": 2.144,
|
7629 |
+
"step": 1084
|
7630 |
+
},
|
7631 |
+
{
|
7632 |
+
"epoch": 0.9875981340311754,
|
7633 |
+
"grad_norm": 0.48190560936927795,
|
7634 |
+
"learning_rate": 8.154757045497619e-08,
|
7635 |
+
"loss": 2.1081,
|
7636 |
+
"step": 1085
|
7637 |
+
},
|
7638 |
+
{
|
7639 |
+
"epoch": 0.988508362726135,
|
7640 |
+
"grad_norm": 0.4542143940925598,
|
7641 |
+
"learning_rate": 7.0315293254819e-08,
|
7642 |
+
"loss": 1.9731,
|
7643 |
+
"step": 1086
|
7644 |
+
},
|
7645 |
+
{
|
7646 |
+
"epoch": 0.9894185914210946,
|
7647 |
+
"grad_norm": 0.4879550337791443,
|
7648 |
+
"learning_rate": 5.991466147791113e-08,
|
7649 |
+
"loss": 2.0791,
|
7650 |
+
"step": 1087
|
7651 |
+
},
|
7652 |
+
{
|
7653 |
+
"epoch": 0.9903288201160542,
|
7654 |
+
"grad_norm": 0.5068708062171936,
|
7655 |
+
"learning_rate": 5.0345761681491746e-08,
|
7656 |
+
"loss": 2.181,
|
7657 |
+
"step": 1088
|
7658 |
+
},
|
7659 |
+
{
|
7660 |
+
"epoch": 0.9912390488110138,
|
7661 |
+
"grad_norm": 0.5021325945854187,
|
7662 |
+
"learning_rate": 4.1608673500859175e-08,
|
7663 |
+
"loss": 2.1546,
|
7664 |
+
"step": 1089
|
7665 |
+
},
|
7666 |
+
{
|
7667 |
+
"epoch": 0.9921492775059734,
|
7668 |
+
"grad_norm": 0.5045585036277771,
|
7669 |
+
"learning_rate": 3.370346964876036e-08,
|
7670 |
+
"loss": 2.0335,
|
7671 |
+
"step": 1090
|
7672 |
+
},
|
7673 |
+
{
|
7674 |
+
"epoch": 0.993059506200933,
|
7675 |
+
"grad_norm": 0.5066413879394531,
|
7676 |
+
"learning_rate": 2.6630215914702495e-08,
|
7677 |
+
"loss": 2.0639,
|
7678 |
+
"step": 1091
|
7679 |
+
},
|
7680 |
+
{
|
7681 |
+
"epoch": 0.9939697348958926,
|
7682 |
+
"grad_norm": 0.5761440992355347,
|
7683 |
+
"learning_rate": 2.038897116447558e-08,
|
7684 |
+
"loss": 2.3702,
|
7685 |
+
"step": 1092
|
7686 |
+
},
|
7687 |
+
{
|
7688 |
+
"epoch": 0.9948799635908522,
|
7689 |
+
"grad_norm": 0.5774243474006653,
|
7690 |
+
"learning_rate": 1.4979787339619578e-08,
|
7691 |
+
"loss": 2.1865,
|
7692 |
+
"step": 1093
|
7693 |
+
},
|
7694 |
+
{
|
7695 |
+
"epoch": 0.9957901922858118,
|
7696 |
+
"grad_norm": 0.6156805157661438,
|
7697 |
+
"learning_rate": 1.0402709457035808e-08,
|
7698 |
+
"loss": 2.3407,
|
7699 |
+
"step": 1094
|
7700 |
+
},
|
7701 |
+
{
|
7702 |
+
"epoch": 0.9967004209807714,
|
7703 |
+
"grad_norm": 0.6527782082557678,
|
7704 |
+
"learning_rate": 6.657775608553962e-09,
|
7705 |
+
"loss": 2.3115,
|
7706 |
+
"step": 1095
|
7707 |
+
},
|
7708 |
+
{
|
7709 |
+
"epoch": 0.997610649675731,
|
7710 |
+
"grad_norm": 0.7234971523284912,
|
7711 |
+
"learning_rate": 3.745016960665648e-09,
|
7712 |
+
"loss": 2.5377,
|
7713 |
+
"step": 1096
|
7714 |
+
},
|
7715 |
+
{
|
7716 |
+
"epoch": 0.9985208783706906,
|
7717 |
+
"grad_norm": 0.7837737798690796,
|
7718 |
+
"learning_rate": 1.6644577542357375e-09,
|
7719 |
+
"loss": 2.4761,
|
7720 |
+
"step": 1097
|
7721 |
+
},
|
7722 |
+
{
|
7723 |
+
"epoch": 0.9994311070656502,
|
7724 |
+
"grad_norm": 1.1138982772827148,
|
7725 |
+
"learning_rate": 4.1611530431362453e-10,
|
7726 |
+
"loss": 2.4967,
|
7727 |
+
"step": 1098
|
7728 |
+
},
|
7729 |
+
{
|
7730 |
+
"epoch": 1.0006826715212198,
|
7731 |
+
"grad_norm": 3.5543136596679688,
|
7732 |
+
"learning_rate": 0.0,
|
7733 |
+
"loss": 4.7799,
|
7734 |
+
"step": 1099
|
7735 |
}
|
7736 |
],
|
7737 |
"logging_steps": 1,
|
|
|
7746 |
"should_evaluate": false,
|
7747 |
"should_log": false,
|
7748 |
"should_save": true,
|
7749 |
+
"should_training_stop": true
|
7750 |
},
|
7751 |
"attributes": {}
|
7752 |
}
|
7753 |
},
|
7754 |
+
"total_flos": 2.0148357964895355e+18,
|
7755 |
"train_batch_size": 2,
|
7756 |
"trial_name": null,
|
7757 |
"trial_params": null
|