Training in progress, step 1472, 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 144805440
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:199cdffee86eb37719656ecd5540378d5c9749f011284f740afd8b8bde9f425e
|
3 |
size 144805440
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 74292308
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:d1c5c452f5bc6e446e4ebb411c36774631978965e1fd9ae5181b64d85dfed0f3
|
3 |
size 74292308
|
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:c6b81d6d271fca7a7f5d3e20c893651caf32f09b2ab3622300e7e002882bb168
|
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:ed371e0f8322f8f9632fd410d79b3b44054937f0f8ef4428d1d8675a960ce27d
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 500,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -7735,6 +7735,2582 @@
|
|
7735 |
"learning_rate": 1.4739429690090533e-05,
|
7736 |
"loss": 1.6115,
|
7737 |
"step": 1104
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
7738 |
}
|
7739 |
],
|
7740 |
"logging_steps": 1,
|
@@ -7749,12 +10325,12 @@
|
|
7749 |
"should_evaluate": false,
|
7750 |
"should_log": false,
|
7751 |
"should_save": true,
|
7752 |
-
"should_training_stop":
|
7753 |
},
|
7754 |
"attributes": {}
|
7755 |
}
|
7756 |
},
|
7757 |
-
"total_flos": 1.
|
7758 |
"train_batch_size": 4,
|
7759 |
"trial_name": null,
|
7760 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.09684688389229731,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1472,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
7735 |
"learning_rate": 1.4739429690090533e-05,
|
7736 |
"loss": 1.6115,
|
7737 |
"step": 1104
|
7738 |
+
},
|
7739 |
+
{
|
7740 |
+
"epoch": 0.07270095563925852,
|
7741 |
+
"grad_norm": 0.5422717928886414,
|
7742 |
+
"learning_rate": 1.4663594444569667e-05,
|
7743 |
+
"loss": 1.5951,
|
7744 |
+
"step": 1105
|
7745 |
+
},
|
7746 |
+
{
|
7747 |
+
"epoch": 0.07276674835929405,
|
7748 |
+
"grad_norm": 0.5676501989364624,
|
7749 |
+
"learning_rate": 1.4587921253782849e-05,
|
7750 |
+
"loss": 1.6266,
|
7751 |
+
"step": 1106
|
7752 |
+
},
|
7753 |
+
{
|
7754 |
+
"epoch": 0.07283254107932957,
|
7755 |
+
"grad_norm": 0.5662168264389038,
|
7756 |
+
"learning_rate": 1.4512410464771514e-05,
|
7757 |
+
"loss": 1.6486,
|
7758 |
+
"step": 1107
|
7759 |
+
},
|
7760 |
+
{
|
7761 |
+
"epoch": 0.0728983337993651,
|
7762 |
+
"grad_norm": 0.5962639451026917,
|
7763 |
+
"learning_rate": 1.4437062423832426e-05,
|
7764 |
+
"loss": 1.8598,
|
7765 |
+
"step": 1108
|
7766 |
+
},
|
7767 |
+
{
|
7768 |
+
"epoch": 0.07296412651940062,
|
7769 |
+
"grad_norm": 0.6214326620101929,
|
7770 |
+
"learning_rate": 1.4361877476515889e-05,
|
7771 |
+
"loss": 1.5881,
|
7772 |
+
"step": 1109
|
7773 |
+
},
|
7774 |
+
{
|
7775 |
+
"epoch": 0.07302991923943615,
|
7776 |
+
"grad_norm": 0.5972443222999573,
|
7777 |
+
"learning_rate": 1.428685596762429e-05,
|
7778 |
+
"loss": 1.6234,
|
7779 |
+
"step": 1110
|
7780 |
+
},
|
7781 |
+
{
|
7782 |
+
"epoch": 0.07309571195947169,
|
7783 |
+
"grad_norm": 0.5804590582847595,
|
7784 |
+
"learning_rate": 1.4211998241210484e-05,
|
7785 |
+
"loss": 1.7503,
|
7786 |
+
"step": 1111
|
7787 |
+
},
|
7788 |
+
{
|
7789 |
+
"epoch": 0.07316150467950722,
|
7790 |
+
"grad_norm": 0.6213569045066833,
|
7791 |
+
"learning_rate": 1.413730464057616e-05,
|
7792 |
+
"loss": 1.6576,
|
7793 |
+
"step": 1112
|
7794 |
+
},
|
7795 |
+
{
|
7796 |
+
"epoch": 0.07322729739954274,
|
7797 |
+
"grad_norm": 0.6171157360076904,
|
7798 |
+
"learning_rate": 1.406277550827037e-05,
|
7799 |
+
"loss": 1.6889,
|
7800 |
+
"step": 1113
|
7801 |
+
},
|
7802 |
+
{
|
7803 |
+
"epoch": 0.07329309011957827,
|
7804 |
+
"grad_norm": 0.63736891746521,
|
7805 |
+
"learning_rate": 1.3988411186087885e-05,
|
7806 |
+
"loss": 1.5542,
|
7807 |
+
"step": 1114
|
7808 |
+
},
|
7809 |
+
{
|
7810 |
+
"epoch": 0.0733588828396138,
|
7811 |
+
"grad_norm": 0.6111243367195129,
|
7812 |
+
"learning_rate": 1.3914212015067651e-05,
|
7813 |
+
"loss": 1.6043,
|
7814 |
+
"step": 1115
|
7815 |
+
},
|
7816 |
+
{
|
7817 |
+
"epoch": 0.07342467555964932,
|
7818 |
+
"grad_norm": 0.6354307532310486,
|
7819 |
+
"learning_rate": 1.3840178335491222e-05,
|
7820 |
+
"loss": 1.5426,
|
7821 |
+
"step": 1116
|
7822 |
+
},
|
7823 |
+
{
|
7824 |
+
"epoch": 0.07349046827968485,
|
7825 |
+
"grad_norm": 0.6276394128799438,
|
7826 |
+
"learning_rate": 1.376631048688119e-05,
|
7827 |
+
"loss": 1.5725,
|
7828 |
+
"step": 1117
|
7829 |
+
},
|
7830 |
+
{
|
7831 |
+
"epoch": 0.07355626099972037,
|
7832 |
+
"grad_norm": 0.6149314641952515,
|
7833 |
+
"learning_rate": 1.3692608807999652e-05,
|
7834 |
+
"loss": 1.6483,
|
7835 |
+
"step": 1118
|
7836 |
+
},
|
7837 |
+
{
|
7838 |
+
"epoch": 0.07362205371975591,
|
7839 |
+
"grad_norm": 0.6823477745056152,
|
7840 |
+
"learning_rate": 1.3619073636846625e-05,
|
7841 |
+
"loss": 1.6028,
|
7842 |
+
"step": 1119
|
7843 |
+
},
|
7844 |
+
{
|
7845 |
+
"epoch": 0.07368784643979144,
|
7846 |
+
"grad_norm": 0.6506856083869934,
|
7847 |
+
"learning_rate": 1.3545705310658529e-05,
|
7848 |
+
"loss": 1.5915,
|
7849 |
+
"step": 1120
|
7850 |
+
},
|
7851 |
+
{
|
7852 |
+
"epoch": 0.07375363915982697,
|
7853 |
+
"grad_norm": 0.6432639956474304,
|
7854 |
+
"learning_rate": 1.3472504165906613e-05,
|
7855 |
+
"loss": 1.5872,
|
7856 |
+
"step": 1121
|
7857 |
+
},
|
7858 |
+
{
|
7859 |
+
"epoch": 0.0738194318798625,
|
7860 |
+
"grad_norm": 0.6213419437408447,
|
7861 |
+
"learning_rate": 1.3399470538295434e-05,
|
7862 |
+
"loss": 1.5422,
|
7863 |
+
"step": 1122
|
7864 |
+
},
|
7865 |
+
{
|
7866 |
+
"epoch": 0.07388522459989802,
|
7867 |
+
"grad_norm": 0.6513727903366089,
|
7868 |
+
"learning_rate": 1.3326604762761258e-05,
|
7869 |
+
"loss": 1.6088,
|
7870 |
+
"step": 1123
|
7871 |
+
},
|
7872 |
+
{
|
7873 |
+
"epoch": 0.07395101731993355,
|
7874 |
+
"grad_norm": 0.6472317576408386,
|
7875 |
+
"learning_rate": 1.3253907173470648e-05,
|
7876 |
+
"loss": 1.5638,
|
7877 |
+
"step": 1124
|
7878 |
+
},
|
7879 |
+
{
|
7880 |
+
"epoch": 0.07401681003996907,
|
7881 |
+
"grad_norm": 0.6375536322593689,
|
7882 |
+
"learning_rate": 1.3181378103818814e-05,
|
7883 |
+
"loss": 1.4387,
|
7884 |
+
"step": 1125
|
7885 |
+
},
|
7886 |
+
{
|
7887 |
+
"epoch": 0.07408260276000461,
|
7888 |
+
"grad_norm": 0.6542819738388062,
|
7889 |
+
"learning_rate": 1.3109017886428122e-05,
|
7890 |
+
"loss": 1.5019,
|
7891 |
+
"step": 1126
|
7892 |
+
},
|
7893 |
+
{
|
7894 |
+
"epoch": 0.07414839548004014,
|
7895 |
+
"grad_norm": 0.6725215911865234,
|
7896 |
+
"learning_rate": 1.30368268531466e-05,
|
7897 |
+
"loss": 1.5493,
|
7898 |
+
"step": 1127
|
7899 |
+
},
|
7900 |
+
{
|
7901 |
+
"epoch": 0.07421418820007566,
|
7902 |
+
"grad_norm": 0.695514976978302,
|
7903 |
+
"learning_rate": 1.2964805335046332e-05,
|
7904 |
+
"loss": 1.7503,
|
7905 |
+
"step": 1128
|
7906 |
+
},
|
7907 |
+
{
|
7908 |
+
"epoch": 0.07427998092011119,
|
7909 |
+
"grad_norm": 0.6565685868263245,
|
7910 |
+
"learning_rate": 1.2892953662422047e-05,
|
7911 |
+
"loss": 1.4902,
|
7912 |
+
"step": 1129
|
7913 |
+
},
|
7914 |
+
{
|
7915 |
+
"epoch": 0.07434577364014672,
|
7916 |
+
"grad_norm": 0.7066967487335205,
|
7917 |
+
"learning_rate": 1.2821272164789544e-05,
|
7918 |
+
"loss": 1.5197,
|
7919 |
+
"step": 1130
|
7920 |
+
},
|
7921 |
+
{
|
7922 |
+
"epoch": 0.07441156636018224,
|
7923 |
+
"grad_norm": 0.686930239200592,
|
7924 |
+
"learning_rate": 1.2749761170884179e-05,
|
7925 |
+
"loss": 1.596,
|
7926 |
+
"step": 1131
|
7927 |
+
},
|
7928 |
+
{
|
7929 |
+
"epoch": 0.07447735908021777,
|
7930 |
+
"grad_norm": 0.7322065830230713,
|
7931 |
+
"learning_rate": 1.2678421008659375e-05,
|
7932 |
+
"loss": 1.5857,
|
7933 |
+
"step": 1132
|
7934 |
+
},
|
7935 |
+
{
|
7936 |
+
"epoch": 0.0745431518002533,
|
7937 |
+
"grad_norm": 0.6980546116828918,
|
7938 |
+
"learning_rate": 1.2607252005285109e-05,
|
7939 |
+
"loss": 1.5733,
|
7940 |
+
"step": 1133
|
7941 |
+
},
|
7942 |
+
{
|
7943 |
+
"epoch": 0.07460894452028884,
|
7944 |
+
"grad_norm": 0.7424272298812866,
|
7945 |
+
"learning_rate": 1.2536254487146415e-05,
|
7946 |
+
"loss": 1.6048,
|
7947 |
+
"step": 1134
|
7948 |
+
},
|
7949 |
+
{
|
7950 |
+
"epoch": 0.07467473724032436,
|
7951 |
+
"grad_norm": 0.708510160446167,
|
7952 |
+
"learning_rate": 1.2465428779841882e-05,
|
7953 |
+
"loss": 1.5323,
|
7954 |
+
"step": 1135
|
7955 |
+
},
|
7956 |
+
{
|
7957 |
+
"epoch": 0.07474052996035989,
|
7958 |
+
"grad_norm": 0.7060134410858154,
|
7959 |
+
"learning_rate": 1.2394775208182174e-05,
|
7960 |
+
"loss": 1.5616,
|
7961 |
+
"step": 1136
|
7962 |
+
},
|
7963 |
+
{
|
7964 |
+
"epoch": 0.07480632268039542,
|
7965 |
+
"grad_norm": 0.7687208652496338,
|
7966 |
+
"learning_rate": 1.2324294096188526e-05,
|
7967 |
+
"loss": 1.6314,
|
7968 |
+
"step": 1137
|
7969 |
+
},
|
7970 |
+
{
|
7971 |
+
"epoch": 0.07487211540043094,
|
7972 |
+
"grad_norm": 0.7260802388191223,
|
7973 |
+
"learning_rate": 1.2253985767091274e-05,
|
7974 |
+
"loss": 1.5503,
|
7975 |
+
"step": 1138
|
7976 |
+
},
|
7977 |
+
{
|
7978 |
+
"epoch": 0.07493790812046647,
|
7979 |
+
"grad_norm": 0.7328141331672668,
|
7980 |
+
"learning_rate": 1.2183850543328312e-05,
|
7981 |
+
"loss": 1.5422,
|
7982 |
+
"step": 1139
|
7983 |
+
},
|
7984 |
+
{
|
7985 |
+
"epoch": 0.075003700840502,
|
7986 |
+
"grad_norm": 0.7931233048439026,
|
7987 |
+
"learning_rate": 1.2113888746543738e-05,
|
7988 |
+
"loss": 1.4909,
|
7989 |
+
"step": 1140
|
7990 |
+
},
|
7991 |
+
{
|
7992 |
+
"epoch": 0.07506949356053752,
|
7993 |
+
"grad_norm": 0.7329046130180359,
|
7994 |
+
"learning_rate": 1.2044100697586263e-05,
|
7995 |
+
"loss": 1.4206,
|
7996 |
+
"step": 1141
|
7997 |
+
},
|
7998 |
+
{
|
7999 |
+
"epoch": 0.07513528628057306,
|
8000 |
+
"grad_norm": 0.8105408549308777,
|
8001 |
+
"learning_rate": 1.1974486716507783e-05,
|
8002 |
+
"loss": 1.6199,
|
8003 |
+
"step": 1142
|
8004 |
+
},
|
8005 |
+
{
|
8006 |
+
"epoch": 0.07520107900060859,
|
8007 |
+
"grad_norm": 0.8885856866836548,
|
8008 |
+
"learning_rate": 1.1905047122561924e-05,
|
8009 |
+
"loss": 1.7629,
|
8010 |
+
"step": 1143
|
8011 |
+
},
|
8012 |
+
{
|
8013 |
+
"epoch": 0.07526687172064411,
|
8014 |
+
"grad_norm": 0.8523672223091125,
|
8015 |
+
"learning_rate": 1.1835782234202525e-05,
|
8016 |
+
"loss": 1.505,
|
8017 |
+
"step": 1144
|
8018 |
+
},
|
8019 |
+
{
|
8020 |
+
"epoch": 0.07533266444067964,
|
8021 |
+
"grad_norm": 0.8462732434272766,
|
8022 |
+
"learning_rate": 1.1766692369082255e-05,
|
8023 |
+
"loss": 1.5047,
|
8024 |
+
"step": 1145
|
8025 |
+
},
|
8026 |
+
{
|
8027 |
+
"epoch": 0.07539845716071517,
|
8028 |
+
"grad_norm": 0.8410788178443909,
|
8029 |
+
"learning_rate": 1.1697777844051105e-05,
|
8030 |
+
"loss": 1.5502,
|
8031 |
+
"step": 1146
|
8032 |
+
},
|
8033 |
+
{
|
8034 |
+
"epoch": 0.07546424988075069,
|
8035 |
+
"grad_norm": 0.8786489367485046,
|
8036 |
+
"learning_rate": 1.1629038975154943e-05,
|
8037 |
+
"loss": 1.4928,
|
8038 |
+
"step": 1147
|
8039 |
+
},
|
8040 |
+
{
|
8041 |
+
"epoch": 0.07553004260078622,
|
8042 |
+
"grad_norm": 0.9621903896331787,
|
8043 |
+
"learning_rate": 1.156047607763407e-05,
|
8044 |
+
"loss": 1.4004,
|
8045 |
+
"step": 1148
|
8046 |
+
},
|
8047 |
+
{
|
8048 |
+
"epoch": 0.07559583532082174,
|
8049 |
+
"grad_norm": 1.068969488143921,
|
8050 |
+
"learning_rate": 1.1492089465921768e-05,
|
8051 |
+
"loss": 1.5004,
|
8052 |
+
"step": 1149
|
8053 |
+
},
|
8054 |
+
{
|
8055 |
+
"epoch": 0.07566162804085728,
|
8056 |
+
"grad_norm": 1.3039774894714355,
|
8057 |
+
"learning_rate": 1.1423879453642878e-05,
|
8058 |
+
"loss": 1.5061,
|
8059 |
+
"step": 1150
|
8060 |
+
},
|
8061 |
+
{
|
8062 |
+
"epoch": 0.07572742076089281,
|
8063 |
+
"grad_norm": 0.5217190384864807,
|
8064 |
+
"learning_rate": 1.135584635361232e-05,
|
8065 |
+
"loss": 1.8168,
|
8066 |
+
"step": 1151
|
8067 |
+
},
|
8068 |
+
{
|
8069 |
+
"epoch": 0.07579321348092834,
|
8070 |
+
"grad_norm": 0.5663981437683105,
|
8071 |
+
"learning_rate": 1.128799047783371e-05,
|
8072 |
+
"loss": 1.7287,
|
8073 |
+
"step": 1152
|
8074 |
+
},
|
8075 |
+
{
|
8076 |
+
"epoch": 0.07585900620096386,
|
8077 |
+
"grad_norm": 0.5568148493766785,
|
8078 |
+
"learning_rate": 1.122031213749789e-05,
|
8079 |
+
"loss": 1.611,
|
8080 |
+
"step": 1153
|
8081 |
+
},
|
8082 |
+
{
|
8083 |
+
"epoch": 0.07592479892099939,
|
8084 |
+
"grad_norm": 0.6089836359024048,
|
8085 |
+
"learning_rate": 1.115281164298153e-05,
|
8086 |
+
"loss": 1.5694,
|
8087 |
+
"step": 1154
|
8088 |
+
},
|
8089 |
+
{
|
8090 |
+
"epoch": 0.07599059164103492,
|
8091 |
+
"grad_norm": 0.5871589183807373,
|
8092 |
+
"learning_rate": 1.1085489303845637e-05,
|
8093 |
+
"loss": 1.6517,
|
8094 |
+
"step": 1155
|
8095 |
+
},
|
8096 |
+
{
|
8097 |
+
"epoch": 0.07605638436107044,
|
8098 |
+
"grad_norm": 0.5930696725845337,
|
8099 |
+
"learning_rate": 1.101834542883427e-05,
|
8100 |
+
"loss": 1.6933,
|
8101 |
+
"step": 1156
|
8102 |
+
},
|
8103 |
+
{
|
8104 |
+
"epoch": 0.07612217708110598,
|
8105 |
+
"grad_norm": 0.5683667659759521,
|
8106 |
+
"learning_rate": 1.0951380325872979e-05,
|
8107 |
+
"loss": 1.6172,
|
8108 |
+
"step": 1157
|
8109 |
+
},
|
8110 |
+
{
|
8111 |
+
"epoch": 0.07618796980114151,
|
8112 |
+
"grad_norm": 0.5930812954902649,
|
8113 |
+
"learning_rate": 1.088459430206748e-05,
|
8114 |
+
"loss": 1.7436,
|
8115 |
+
"step": 1158
|
8116 |
+
},
|
8117 |
+
{
|
8118 |
+
"epoch": 0.07625376252117703,
|
8119 |
+
"grad_norm": 0.5984984040260315,
|
8120 |
+
"learning_rate": 1.0817987663702229e-05,
|
8121 |
+
"loss": 1.8179,
|
8122 |
+
"step": 1159
|
8123 |
+
},
|
8124 |
+
{
|
8125 |
+
"epoch": 0.07631955524121256,
|
8126 |
+
"grad_norm": 0.6047726273536682,
|
8127 |
+
"learning_rate": 1.0751560716238967e-05,
|
8128 |
+
"loss": 1.5642,
|
8129 |
+
"step": 1160
|
8130 |
+
},
|
8131 |
+
{
|
8132 |
+
"epoch": 0.07638534796124809,
|
8133 |
+
"grad_norm": 0.6473960280418396,
|
8134 |
+
"learning_rate": 1.0685313764315413e-05,
|
8135 |
+
"loss": 1.7192,
|
8136 |
+
"step": 1161
|
8137 |
+
},
|
8138 |
+
{
|
8139 |
+
"epoch": 0.07645114068128361,
|
8140 |
+
"grad_norm": 0.6276887059211731,
|
8141 |
+
"learning_rate": 1.0619247111743797e-05,
|
8142 |
+
"loss": 1.532,
|
8143 |
+
"step": 1162
|
8144 |
+
},
|
8145 |
+
{
|
8146 |
+
"epoch": 0.07651693340131914,
|
8147 |
+
"grad_norm": 0.6259212493896484,
|
8148 |
+
"learning_rate": 1.055336106150948e-05,
|
8149 |
+
"loss": 1.5419,
|
8150 |
+
"step": 1163
|
8151 |
+
},
|
8152 |
+
{
|
8153 |
+
"epoch": 0.07658272612135467,
|
8154 |
+
"grad_norm": 0.6881235241889954,
|
8155 |
+
"learning_rate": 1.048765591576959e-05,
|
8156 |
+
"loss": 1.7839,
|
8157 |
+
"step": 1164
|
8158 |
+
},
|
8159 |
+
{
|
8160 |
+
"epoch": 0.0766485188413902,
|
8161 |
+
"grad_norm": 0.6311882138252258,
|
8162 |
+
"learning_rate": 1.0422131975851584e-05,
|
8163 |
+
"loss": 1.5377,
|
8164 |
+
"step": 1165
|
8165 |
+
},
|
8166 |
+
{
|
8167 |
+
"epoch": 0.07671431156142573,
|
8168 |
+
"grad_norm": 0.6650845408439636,
|
8169 |
+
"learning_rate": 1.0356789542251938e-05,
|
8170 |
+
"loss": 1.7409,
|
8171 |
+
"step": 1166
|
8172 |
+
},
|
8173 |
+
{
|
8174 |
+
"epoch": 0.07678010428146126,
|
8175 |
+
"grad_norm": 0.6411380767822266,
|
8176 |
+
"learning_rate": 1.0291628914634694e-05,
|
8177 |
+
"loss": 1.6915,
|
8178 |
+
"step": 1167
|
8179 |
+
},
|
8180 |
+
{
|
8181 |
+
"epoch": 0.07684589700149679,
|
8182 |
+
"grad_norm": 0.6582992672920227,
|
8183 |
+
"learning_rate": 1.022665039183015e-05,
|
8184 |
+
"loss": 1.5401,
|
8185 |
+
"step": 1168
|
8186 |
+
},
|
8187 |
+
{
|
8188 |
+
"epoch": 0.07691168972153231,
|
8189 |
+
"grad_norm": 0.6362737417221069,
|
8190 |
+
"learning_rate": 1.0161854271833443e-05,
|
8191 |
+
"loss": 1.5978,
|
8192 |
+
"step": 1169
|
8193 |
+
},
|
8194 |
+
{
|
8195 |
+
"epoch": 0.07697748244156784,
|
8196 |
+
"grad_norm": 0.6066959500312805,
|
8197 |
+
"learning_rate": 1.009724085180322e-05,
|
8198 |
+
"loss": 1.6114,
|
8199 |
+
"step": 1170
|
8200 |
+
},
|
8201 |
+
{
|
8202 |
+
"epoch": 0.07704327516160336,
|
8203 |
+
"grad_norm": 0.6722879409790039,
|
8204 |
+
"learning_rate": 1.0032810428060218e-05,
|
8205 |
+
"loss": 1.597,
|
8206 |
+
"step": 1171
|
8207 |
+
},
|
8208 |
+
{
|
8209 |
+
"epoch": 0.07710906788163889,
|
8210 |
+
"grad_norm": 0.6590387225151062,
|
8211 |
+
"learning_rate": 9.96856329608597e-06,
|
8212 |
+
"loss": 1.5352,
|
8213 |
+
"step": 1172
|
8214 |
+
},
|
8215 |
+
{
|
8216 |
+
"epoch": 0.07717486060167443,
|
8217 |
+
"grad_norm": 0.6803810000419617,
|
8218 |
+
"learning_rate": 9.90449975052144e-06,
|
8219 |
+
"loss": 1.5224,
|
8220 |
+
"step": 1173
|
8221 |
+
},
|
8222 |
+
{
|
8223 |
+
"epoch": 0.07724065332170996,
|
8224 |
+
"grad_norm": 0.6694747805595398,
|
8225 |
+
"learning_rate": 9.840620085165626e-06,
|
8226 |
+
"loss": 1.7062,
|
8227 |
+
"step": 1174
|
8228 |
+
},
|
8229 |
+
{
|
8230 |
+
"epoch": 0.07730644604174548,
|
8231 |
+
"grad_norm": 0.672380268573761,
|
8232 |
+
"learning_rate": 9.776924592974256e-06,
|
8233 |
+
"loss": 1.6693,
|
8234 |
+
"step": 1175
|
8235 |
+
},
|
8236 |
+
{
|
8237 |
+
"epoch": 0.07737223876178101,
|
8238 |
+
"grad_norm": 0.6489235162734985,
|
8239 |
+
"learning_rate": 9.713413566058405e-06,
|
8240 |
+
"loss": 1.5843,
|
8241 |
+
"step": 1176
|
8242 |
+
},
|
8243 |
+
{
|
8244 |
+
"epoch": 0.07743803148181654,
|
8245 |
+
"grad_norm": 0.6905193328857422,
|
8246 |
+
"learning_rate": 9.650087295683202e-06,
|
8247 |
+
"loss": 1.593,
|
8248 |
+
"step": 1177
|
8249 |
+
},
|
8250 |
+
{
|
8251 |
+
"epoch": 0.07750382420185206,
|
8252 |
+
"grad_norm": 0.7134239673614502,
|
8253 |
+
"learning_rate": 9.586946072266478e-06,
|
8254 |
+
"loss": 1.5109,
|
8255 |
+
"step": 1178
|
8256 |
+
},
|
8257 |
+
{
|
8258 |
+
"epoch": 0.07756961692188759,
|
8259 |
+
"grad_norm": 0.6946320533752441,
|
8260 |
+
"learning_rate": 9.523990185377413e-06,
|
8261 |
+
"loss": 1.5549,
|
8262 |
+
"step": 1179
|
8263 |
+
},
|
8264 |
+
{
|
8265 |
+
"epoch": 0.07763540964192311,
|
8266 |
+
"grad_norm": 0.6910642981529236,
|
8267 |
+
"learning_rate": 9.461219923735227e-06,
|
8268 |
+
"loss": 1.7573,
|
8269 |
+
"step": 1180
|
8270 |
+
},
|
8271 |
+
{
|
8272 |
+
"epoch": 0.07770120236195865,
|
8273 |
+
"grad_norm": 0.6611814498901367,
|
8274 |
+
"learning_rate": 9.398635575207854e-06,
|
8275 |
+
"loss": 1.5795,
|
8276 |
+
"step": 1181
|
8277 |
+
},
|
8278 |
+
{
|
8279 |
+
"epoch": 0.07776699508199418,
|
8280 |
+
"grad_norm": 0.6925032734870911,
|
8281 |
+
"learning_rate": 9.336237426810624e-06,
|
8282 |
+
"loss": 1.6087,
|
8283 |
+
"step": 1182
|
8284 |
+
},
|
8285 |
+
{
|
8286 |
+
"epoch": 0.07783278780202971,
|
8287 |
+
"grad_norm": 0.7224239706993103,
|
8288 |
+
"learning_rate": 9.274025764704936e-06,
|
8289 |
+
"loss": 1.5171,
|
8290 |
+
"step": 1183
|
8291 |
+
},
|
8292 |
+
{
|
8293 |
+
"epoch": 0.07789858052206523,
|
8294 |
+
"grad_norm": 0.696713387966156,
|
8295 |
+
"learning_rate": 9.212000874196953e-06,
|
8296 |
+
"loss": 1.4585,
|
8297 |
+
"step": 1184
|
8298 |
+
},
|
8299 |
+
{
|
8300 |
+
"epoch": 0.07796437324210076,
|
8301 |
+
"grad_norm": 0.7397194504737854,
|
8302 |
+
"learning_rate": 9.150163039736297e-06,
|
8303 |
+
"loss": 1.551,
|
8304 |
+
"step": 1185
|
8305 |
+
},
|
8306 |
+
{
|
8307 |
+
"epoch": 0.07803016596213629,
|
8308 |
+
"grad_norm": 0.696258008480072,
|
8309 |
+
"learning_rate": 9.08851254491475e-06,
|
8310 |
+
"loss": 1.5841,
|
8311 |
+
"step": 1186
|
8312 |
+
},
|
8313 |
+
{
|
8314 |
+
"epoch": 0.07809595868217181,
|
8315 |
+
"grad_norm": 0.7256150841712952,
|
8316 |
+
"learning_rate": 9.027049672464916e-06,
|
8317 |
+
"loss": 1.6002,
|
8318 |
+
"step": 1187
|
8319 |
+
},
|
8320 |
+
{
|
8321 |
+
"epoch": 0.07816175140220735,
|
8322 |
+
"grad_norm": 0.7567223906517029,
|
8323 |
+
"learning_rate": 8.965774704258956e-06,
|
8324 |
+
"loss": 1.6436,
|
8325 |
+
"step": 1188
|
8326 |
+
},
|
8327 |
+
{
|
8328 |
+
"epoch": 0.07822754412224288,
|
8329 |
+
"grad_norm": 0.7757462859153748,
|
8330 |
+
"learning_rate": 8.90468792130733e-06,
|
8331 |
+
"loss": 1.7119,
|
8332 |
+
"step": 1189
|
8333 |
+
},
|
8334 |
+
{
|
8335 |
+
"epoch": 0.0782933368422784,
|
8336 |
+
"grad_norm": 0.7670205235481262,
|
8337 |
+
"learning_rate": 8.843789603757446e-06,
|
8338 |
+
"loss": 1.6326,
|
8339 |
+
"step": 1190
|
8340 |
+
},
|
8341 |
+
{
|
8342 |
+
"epoch": 0.07835912956231393,
|
8343 |
+
"grad_norm": 0.8732237219810486,
|
8344 |
+
"learning_rate": 8.783080030892394e-06,
|
8345 |
+
"loss": 1.415,
|
8346 |
+
"step": 1191
|
8347 |
+
},
|
8348 |
+
{
|
8349 |
+
"epoch": 0.07842492228234946,
|
8350 |
+
"grad_norm": 0.8682703375816345,
|
8351 |
+
"learning_rate": 8.72255948112966e-06,
|
8352 |
+
"loss": 1.629,
|
8353 |
+
"step": 1192
|
8354 |
+
},
|
8355 |
+
{
|
8356 |
+
"epoch": 0.07849071500238498,
|
8357 |
+
"grad_norm": 0.7908897399902344,
|
8358 |
+
"learning_rate": 8.662228232019876e-06,
|
8359 |
+
"loss": 1.3307,
|
8360 |
+
"step": 1193
|
8361 |
+
},
|
8362 |
+
{
|
8363 |
+
"epoch": 0.07855650772242051,
|
8364 |
+
"grad_norm": 0.8436971306800842,
|
8365 |
+
"learning_rate": 8.602086560245537e-06,
|
8366 |
+
"loss": 1.5682,
|
8367 |
+
"step": 1194
|
8368 |
+
},
|
8369 |
+
{
|
8370 |
+
"epoch": 0.07862230044245604,
|
8371 |
+
"grad_norm": 0.8712666630744934,
|
8372 |
+
"learning_rate": 8.542134741619711e-06,
|
8373 |
+
"loss": 1.5686,
|
8374 |
+
"step": 1195
|
8375 |
+
},
|
8376 |
+
{
|
8377 |
+
"epoch": 0.07868809316249158,
|
8378 |
+
"grad_norm": 0.8938828110694885,
|
8379 |
+
"learning_rate": 8.48237305108479e-06,
|
8380 |
+
"loss": 1.5938,
|
8381 |
+
"step": 1196
|
8382 |
+
},
|
8383 |
+
{
|
8384 |
+
"epoch": 0.0787538858825271,
|
8385 |
+
"grad_norm": 0.8834605813026428,
|
8386 |
+
"learning_rate": 8.422801762711247e-06,
|
8387 |
+
"loss": 1.4887,
|
8388 |
+
"step": 1197
|
8389 |
+
},
|
8390 |
+
{
|
8391 |
+
"epoch": 0.07881967860256263,
|
8392 |
+
"grad_norm": 1.0207496881484985,
|
8393 |
+
"learning_rate": 8.363421149696332e-06,
|
8394 |
+
"loss": 1.576,
|
8395 |
+
"step": 1198
|
8396 |
+
},
|
8397 |
+
{
|
8398 |
+
"epoch": 0.07888547132259816,
|
8399 |
+
"grad_norm": 1.0710493326187134,
|
8400 |
+
"learning_rate": 8.304231484362868e-06,
|
8401 |
+
"loss": 1.6295,
|
8402 |
+
"step": 1199
|
8403 |
+
},
|
8404 |
+
{
|
8405 |
+
"epoch": 0.07895126404263368,
|
8406 |
+
"grad_norm": 1.221801996231079,
|
8407 |
+
"learning_rate": 8.245233038157962e-06,
|
8408 |
+
"loss": 1.6023,
|
8409 |
+
"step": 1200
|
8410 |
+
},
|
8411 |
+
{
|
8412 |
+
"epoch": 0.07901705676266921,
|
8413 |
+
"grad_norm": 0.517112672328949,
|
8414 |
+
"learning_rate": 8.186426081651804e-06,
|
8415 |
+
"loss": 1.6261,
|
8416 |
+
"step": 1201
|
8417 |
+
},
|
8418 |
+
{
|
8419 |
+
"epoch": 0.07908284948270473,
|
8420 |
+
"grad_norm": 0.5203869938850403,
|
8421 |
+
"learning_rate": 8.127810884536403e-06,
|
8422 |
+
"loss": 1.6803,
|
8423 |
+
"step": 1202
|
8424 |
+
},
|
8425 |
+
{
|
8426 |
+
"epoch": 0.07914864220274026,
|
8427 |
+
"grad_norm": 0.526978611946106,
|
8428 |
+
"learning_rate": 8.069387715624294e-06,
|
8429 |
+
"loss": 1.6658,
|
8430 |
+
"step": 1203
|
8431 |
+
},
|
8432 |
+
{
|
8433 |
+
"epoch": 0.0792144349227758,
|
8434 |
+
"grad_norm": 0.5750939249992371,
|
8435 |
+
"learning_rate": 8.011156842847412e-06,
|
8436 |
+
"loss": 1.6741,
|
8437 |
+
"step": 1204
|
8438 |
+
},
|
8439 |
+
{
|
8440 |
+
"epoch": 0.07928022764281133,
|
8441 |
+
"grad_norm": 0.5610478520393372,
|
8442 |
+
"learning_rate": 7.95311853325582e-06,
|
8443 |
+
"loss": 1.6639,
|
8444 |
+
"step": 1205
|
8445 |
+
},
|
8446 |
+
{
|
8447 |
+
"epoch": 0.07934602036284685,
|
8448 |
+
"grad_norm": 0.5532097816467285,
|
8449 |
+
"learning_rate": 7.89527305301645e-06,
|
8450 |
+
"loss": 1.6807,
|
8451 |
+
"step": 1206
|
8452 |
+
},
|
8453 |
+
{
|
8454 |
+
"epoch": 0.07941181308288238,
|
8455 |
+
"grad_norm": 0.5609330534934998,
|
8456 |
+
"learning_rate": 7.83762066741191e-06,
|
8457 |
+
"loss": 1.701,
|
8458 |
+
"step": 1207
|
8459 |
+
},
|
8460 |
+
{
|
8461 |
+
"epoch": 0.0794776058029179,
|
8462 |
+
"grad_norm": 0.5949382781982422,
|
8463 |
+
"learning_rate": 7.780161640839257e-06,
|
8464 |
+
"loss": 1.7165,
|
8465 |
+
"step": 1208
|
8466 |
+
},
|
8467 |
+
{
|
8468 |
+
"epoch": 0.07954339852295343,
|
8469 |
+
"grad_norm": 0.5886333584785461,
|
8470 |
+
"learning_rate": 7.722896236808807e-06,
|
8471 |
+
"loss": 1.5563,
|
8472 |
+
"step": 1209
|
8473 |
+
},
|
8474 |
+
{
|
8475 |
+
"epoch": 0.07960919124298896,
|
8476 |
+
"grad_norm": 0.6132276654243469,
|
8477 |
+
"learning_rate": 7.665824717942915e-06,
|
8478 |
+
"loss": 1.7714,
|
8479 |
+
"step": 1210
|
8480 |
+
},
|
8481 |
+
{
|
8482 |
+
"epoch": 0.0796749839630245,
|
8483 |
+
"grad_norm": 0.6169953942298889,
|
8484 |
+
"learning_rate": 7.60894734597476e-06,
|
8485 |
+
"loss": 1.6546,
|
8486 |
+
"step": 1211
|
8487 |
+
},
|
8488 |
+
{
|
8489 |
+
"epoch": 0.07974077668306002,
|
8490 |
+
"grad_norm": 0.6119076609611511,
|
8491 |
+
"learning_rate": 7.552264381747148e-06,
|
8492 |
+
"loss": 1.6585,
|
8493 |
+
"step": 1212
|
8494 |
+
},
|
8495 |
+
{
|
8496 |
+
"epoch": 0.07980656940309555,
|
8497 |
+
"grad_norm": 0.6173310279846191,
|
8498 |
+
"learning_rate": 7.495776085211331e-06,
|
8499 |
+
"loss": 1.4996,
|
8500 |
+
"step": 1213
|
8501 |
+
},
|
8502 |
+
{
|
8503 |
+
"epoch": 0.07987236212313108,
|
8504 |
+
"grad_norm": 0.590090811252594,
|
8505 |
+
"learning_rate": 7.439482715425805e-06,
|
8506 |
+
"loss": 1.5205,
|
8507 |
+
"step": 1214
|
8508 |
+
},
|
8509 |
+
{
|
8510 |
+
"epoch": 0.0799381548431666,
|
8511 |
+
"grad_norm": 0.6101630926132202,
|
8512 |
+
"learning_rate": 7.383384530555104e-06,
|
8513 |
+
"loss": 1.681,
|
8514 |
+
"step": 1215
|
8515 |
+
},
|
8516 |
+
{
|
8517 |
+
"epoch": 0.08000394756320213,
|
8518 |
+
"grad_norm": 0.6025815010070801,
|
8519 |
+
"learning_rate": 7.327481787868646e-06,
|
8520 |
+
"loss": 1.5154,
|
8521 |
+
"step": 1216
|
8522 |
+
},
|
8523 |
+
{
|
8524 |
+
"epoch": 0.08006974028323766,
|
8525 |
+
"grad_norm": 0.6469805240631104,
|
8526 |
+
"learning_rate": 7.271774743739545e-06,
|
8527 |
+
"loss": 1.7022,
|
8528 |
+
"step": 1217
|
8529 |
+
},
|
8530 |
+
{
|
8531 |
+
"epoch": 0.08013553300327318,
|
8532 |
+
"grad_norm": 0.8975655436515808,
|
8533 |
+
"learning_rate": 7.216263653643435e-06,
|
8534 |
+
"loss": 1.8082,
|
8535 |
+
"step": 1218
|
8536 |
+
},
|
8537 |
+
{
|
8538 |
+
"epoch": 0.08020132572330872,
|
8539 |
+
"grad_norm": 0.625879168510437,
|
8540 |
+
"learning_rate": 7.16094877215725e-06,
|
8541 |
+
"loss": 1.472,
|
8542 |
+
"step": 1219
|
8543 |
+
},
|
8544 |
+
{
|
8545 |
+
"epoch": 0.08026711844334425,
|
8546 |
+
"grad_norm": 0.6286988258361816,
|
8547 |
+
"learning_rate": 7.105830352958142e-06,
|
8548 |
+
"loss": 1.5377,
|
8549 |
+
"step": 1220
|
8550 |
+
},
|
8551 |
+
{
|
8552 |
+
"epoch": 0.08033291116337977,
|
8553 |
+
"grad_norm": 0.6680283546447754,
|
8554 |
+
"learning_rate": 7.050908648822291e-06,
|
8555 |
+
"loss": 1.6344,
|
8556 |
+
"step": 1221
|
8557 |
+
},
|
8558 |
+
{
|
8559 |
+
"epoch": 0.0803987038834153,
|
8560 |
+
"grad_norm": 0.650658130645752,
|
8561 |
+
"learning_rate": 6.996183911623688e-06,
|
8562 |
+
"loss": 1.6916,
|
8563 |
+
"step": 1222
|
8564 |
+
},
|
8565 |
+
{
|
8566 |
+
"epoch": 0.08046449660345083,
|
8567 |
+
"grad_norm": 0.6374256014823914,
|
8568 |
+
"learning_rate": 6.941656392333046e-06,
|
8569 |
+
"loss": 1.5798,
|
8570 |
+
"step": 1223
|
8571 |
+
},
|
8572 |
+
{
|
8573 |
+
"epoch": 0.08053028932348635,
|
8574 |
+
"grad_norm": 0.6473804116249084,
|
8575 |
+
"learning_rate": 6.887326341016636e-06,
|
8576 |
+
"loss": 1.433,
|
8577 |
+
"step": 1224
|
8578 |
+
},
|
8579 |
+
{
|
8580 |
+
"epoch": 0.08059608204352188,
|
8581 |
+
"grad_norm": 0.6380757689476013,
|
8582 |
+
"learning_rate": 6.833194006835081e-06,
|
8583 |
+
"loss": 1.5023,
|
8584 |
+
"step": 1225
|
8585 |
+
},
|
8586 |
+
{
|
8587 |
+
"epoch": 0.0806618747635574,
|
8588 |
+
"grad_norm": 0.6916465759277344,
|
8589 |
+
"learning_rate": 6.779259638042318e-06,
|
8590 |
+
"loss": 1.6726,
|
8591 |
+
"step": 1226
|
8592 |
+
},
|
8593 |
+
{
|
8594 |
+
"epoch": 0.08072766748359295,
|
8595 |
+
"grad_norm": 0.6720482707023621,
|
8596 |
+
"learning_rate": 6.725523481984375e-06,
|
8597 |
+
"loss": 1.7212,
|
8598 |
+
"step": 1227
|
8599 |
+
},
|
8600 |
+
{
|
8601 |
+
"epoch": 0.08079346020362847,
|
8602 |
+
"grad_norm": 0.6906450986862183,
|
8603 |
+
"learning_rate": 6.671985785098278e-06,
|
8604 |
+
"loss": 1.4751,
|
8605 |
+
"step": 1228
|
8606 |
+
},
|
8607 |
+
{
|
8608 |
+
"epoch": 0.080859252923664,
|
8609 |
+
"grad_norm": 0.685138463973999,
|
8610 |
+
"learning_rate": 6.618646792910893e-06,
|
8611 |
+
"loss": 1.5658,
|
8612 |
+
"step": 1229
|
8613 |
+
},
|
8614 |
+
{
|
8615 |
+
"epoch": 0.08092504564369953,
|
8616 |
+
"grad_norm": 0.7090581059455872,
|
8617 |
+
"learning_rate": 6.565506750037836e-06,
|
8618 |
+
"loss": 1.5718,
|
8619 |
+
"step": 1230
|
8620 |
+
},
|
8621 |
+
{
|
8622 |
+
"epoch": 0.08099083836373505,
|
8623 |
+
"grad_norm": 0.6716530919075012,
|
8624 |
+
"learning_rate": 6.512565900182305e-06,
|
8625 |
+
"loss": 1.4331,
|
8626 |
+
"step": 1231
|
8627 |
+
},
|
8628 |
+
{
|
8629 |
+
"epoch": 0.08105663108377058,
|
8630 |
+
"grad_norm": 0.6960655450820923,
|
8631 |
+
"learning_rate": 6.459824486134014e-06,
|
8632 |
+
"loss": 1.6364,
|
8633 |
+
"step": 1232
|
8634 |
+
},
|
8635 |
+
{
|
8636 |
+
"epoch": 0.0811224238038061,
|
8637 |
+
"grad_norm": 0.7236138582229614,
|
8638 |
+
"learning_rate": 6.407282749768029e-06,
|
8639 |
+
"loss": 1.5542,
|
8640 |
+
"step": 1233
|
8641 |
+
},
|
8642 |
+
{
|
8643 |
+
"epoch": 0.08118821652384163,
|
8644 |
+
"grad_norm": 0.6596976518630981,
|
8645 |
+
"learning_rate": 6.354940932043713e-06,
|
8646 |
+
"loss": 1.5349,
|
8647 |
+
"step": 1234
|
8648 |
+
},
|
8649 |
+
{
|
8650 |
+
"epoch": 0.08125400924387717,
|
8651 |
+
"grad_norm": 0.6876804232597351,
|
8652 |
+
"learning_rate": 6.302799273003546e-06,
|
8653 |
+
"loss": 1.442,
|
8654 |
+
"step": 1235
|
8655 |
+
},
|
8656 |
+
{
|
8657 |
+
"epoch": 0.0813198019639127,
|
8658 |
+
"grad_norm": 0.710265576839447,
|
8659 |
+
"learning_rate": 6.2508580117720985e-06,
|
8660 |
+
"loss": 1.515,
|
8661 |
+
"step": 1236
|
8662 |
+
},
|
8663 |
+
{
|
8664 |
+
"epoch": 0.08138559468394822,
|
8665 |
+
"grad_norm": 0.7719532251358032,
|
8666 |
+
"learning_rate": 6.199117386554926e-06,
|
8667 |
+
"loss": 1.7442,
|
8668 |
+
"step": 1237
|
8669 |
+
},
|
8670 |
+
{
|
8671 |
+
"epoch": 0.08145138740398375,
|
8672 |
+
"grad_norm": 0.7328771948814392,
|
8673 |
+
"learning_rate": 6.147577634637414e-06,
|
8674 |
+
"loss": 1.569,
|
8675 |
+
"step": 1238
|
8676 |
+
},
|
8677 |
+
{
|
8678 |
+
"epoch": 0.08151718012401928,
|
8679 |
+
"grad_norm": 0.7373148202896118,
|
8680 |
+
"learning_rate": 6.096238992383752e-06,
|
8681 |
+
"loss": 1.5404,
|
8682 |
+
"step": 1239
|
8683 |
+
},
|
8684 |
+
{
|
8685 |
+
"epoch": 0.0815829728440548,
|
8686 |
+
"grad_norm": 0.8095849752426147,
|
8687 |
+
"learning_rate": 6.045101695235844e-06,
|
8688 |
+
"loss": 1.5851,
|
8689 |
+
"step": 1240
|
8690 |
+
},
|
8691 |
+
{
|
8692 |
+
"epoch": 0.08164876556409033,
|
8693 |
+
"grad_norm": 0.7829951047897339,
|
8694 |
+
"learning_rate": 5.994165977712174e-06,
|
8695 |
+
"loss": 1.5866,
|
8696 |
+
"step": 1241
|
8697 |
+
},
|
8698 |
+
{
|
8699 |
+
"epoch": 0.08171455828412587,
|
8700 |
+
"grad_norm": 0.7352398633956909,
|
8701 |
+
"learning_rate": 5.943432073406796e-06,
|
8702 |
+
"loss": 1.5109,
|
8703 |
+
"step": 1242
|
8704 |
+
},
|
8705 |
+
{
|
8706 |
+
"epoch": 0.0817803510041614,
|
8707 |
+
"grad_norm": 0.7571715116500854,
|
8708 |
+
"learning_rate": 5.892900214988245e-06,
|
8709 |
+
"loss": 1.5424,
|
8710 |
+
"step": 1243
|
8711 |
+
},
|
8712 |
+
{
|
8713 |
+
"epoch": 0.08184614372419692,
|
8714 |
+
"grad_norm": 0.806236207485199,
|
8715 |
+
"learning_rate": 5.842570634198452e-06,
|
8716 |
+
"loss": 1.6222,
|
8717 |
+
"step": 1244
|
8718 |
+
},
|
8719 |
+
{
|
8720 |
+
"epoch": 0.08191193644423245,
|
8721 |
+
"grad_norm": 0.894612729549408,
|
8722 |
+
"learning_rate": 5.792443561851685e-06,
|
8723 |
+
"loss": 1.4945,
|
8724 |
+
"step": 1245
|
8725 |
+
},
|
8726 |
+
{
|
8727 |
+
"epoch": 0.08197772916426797,
|
8728 |
+
"grad_norm": 0.8263707756996155,
|
8729 |
+
"learning_rate": 5.742519227833509e-06,
|
8730 |
+
"loss": 1.5567,
|
8731 |
+
"step": 1246
|
8732 |
+
},
|
8733 |
+
{
|
8734 |
+
"epoch": 0.0820435218843035,
|
8735 |
+
"grad_norm": 0.8109045624732971,
|
8736 |
+
"learning_rate": 5.692797861099719e-06,
|
8737 |
+
"loss": 1.4979,
|
8738 |
+
"step": 1247
|
8739 |
+
},
|
8740 |
+
{
|
8741 |
+
"epoch": 0.08210931460433903,
|
8742 |
+
"grad_norm": 1.0256260633468628,
|
8743 |
+
"learning_rate": 5.643279689675279e-06,
|
8744 |
+
"loss": 1.528,
|
8745 |
+
"step": 1248
|
8746 |
+
},
|
8747 |
+
{
|
8748 |
+
"epoch": 0.08217510732437455,
|
8749 |
+
"grad_norm": 0.9840426445007324,
|
8750 |
+
"learning_rate": 5.593964940653296e-06,
|
8751 |
+
"loss": 1.6114,
|
8752 |
+
"step": 1249
|
8753 |
+
},
|
8754 |
+
{
|
8755 |
+
"epoch": 0.08224090004441009,
|
8756 |
+
"grad_norm": 1.197351098060608,
|
8757 |
+
"learning_rate": 5.544853840193981e-06,
|
8758 |
+
"loss": 1.2792,
|
8759 |
+
"step": 1250
|
8760 |
+
},
|
8761 |
+
{
|
8762 |
+
"epoch": 0.08230669276444562,
|
8763 |
+
"grad_norm": 0.5352635383605957,
|
8764 |
+
"learning_rate": 5.495946613523567e-06,
|
8765 |
+
"loss": 1.7328,
|
8766 |
+
"step": 1251
|
8767 |
+
},
|
8768 |
+
{
|
8769 |
+
"epoch": 0.08237248548448114,
|
8770 |
+
"grad_norm": 0.5567160248756409,
|
8771 |
+
"learning_rate": 5.4472434849333396e-06,
|
8772 |
+
"loss": 1.6491,
|
8773 |
+
"step": 1252
|
8774 |
+
},
|
8775 |
+
{
|
8776 |
+
"epoch": 0.08243827820451667,
|
8777 |
+
"grad_norm": 0.6188229918479919,
|
8778 |
+
"learning_rate": 5.398744677778594e-06,
|
8779 |
+
"loss": 1.6258,
|
8780 |
+
"step": 1253
|
8781 |
+
},
|
8782 |
+
{
|
8783 |
+
"epoch": 0.0825040709245522,
|
8784 |
+
"grad_norm": 0.565366804599762,
|
8785 |
+
"learning_rate": 5.3504504144775535e-06,
|
8786 |
+
"loss": 1.7164,
|
8787 |
+
"step": 1254
|
8788 |
+
},
|
8789 |
+
{
|
8790 |
+
"epoch": 0.08256986364458772,
|
8791 |
+
"grad_norm": 0.5582414269447327,
|
8792 |
+
"learning_rate": 5.302360916510424e-06,
|
8793 |
+
"loss": 1.6116,
|
8794 |
+
"step": 1255
|
8795 |
+
},
|
8796 |
+
{
|
8797 |
+
"epoch": 0.08263565636462325,
|
8798 |
+
"grad_norm": 0.5899804830551147,
|
8799 |
+
"learning_rate": 5.25447640441834e-06,
|
8800 |
+
"loss": 1.6297,
|
8801 |
+
"step": 1256
|
8802 |
+
},
|
8803 |
+
{
|
8804 |
+
"epoch": 0.08270144908465878,
|
8805 |
+
"grad_norm": 0.6136707067489624,
|
8806 |
+
"learning_rate": 5.206797097802341e-06,
|
8807 |
+
"loss": 1.8456,
|
8808 |
+
"step": 1257
|
8809 |
+
},
|
8810 |
+
{
|
8811 |
+
"epoch": 0.08276724180469432,
|
8812 |
+
"grad_norm": 0.5709074139595032,
|
8813 |
+
"learning_rate": 5.1593232153223984e-06,
|
8814 |
+
"loss": 1.5469,
|
8815 |
+
"step": 1258
|
8816 |
+
},
|
8817 |
+
{
|
8818 |
+
"epoch": 0.08283303452472984,
|
8819 |
+
"grad_norm": 0.5794378519058228,
|
8820 |
+
"learning_rate": 5.112054974696395e-06,
|
8821 |
+
"loss": 1.594,
|
8822 |
+
"step": 1259
|
8823 |
+
},
|
8824 |
+
{
|
8825 |
+
"epoch": 0.08289882724476537,
|
8826 |
+
"grad_norm": 0.6078215837478638,
|
8827 |
+
"learning_rate": 5.064992592699136e-06,
|
8828 |
+
"loss": 1.7557,
|
8829 |
+
"step": 1260
|
8830 |
+
},
|
8831 |
+
{
|
8832 |
+
"epoch": 0.0829646199648009,
|
8833 |
+
"grad_norm": 0.6244221925735474,
|
8834 |
+
"learning_rate": 5.018136285161329e-06,
|
8835 |
+
"loss": 1.6409,
|
8836 |
+
"step": 1261
|
8837 |
+
},
|
8838 |
+
{
|
8839 |
+
"epoch": 0.08303041268483642,
|
8840 |
+
"grad_norm": 0.6333777904510498,
|
8841 |
+
"learning_rate": 4.9714862669686335e-06,
|
8842 |
+
"loss": 1.6453,
|
8843 |
+
"step": 1262
|
8844 |
+
},
|
8845 |
+
{
|
8846 |
+
"epoch": 0.08309620540487195,
|
8847 |
+
"grad_norm": 0.6196883320808411,
|
8848 |
+
"learning_rate": 4.925042752060638e-06,
|
8849 |
+
"loss": 1.539,
|
8850 |
+
"step": 1263
|
8851 |
+
},
|
8852 |
+
{
|
8853 |
+
"epoch": 0.08316199812490747,
|
8854 |
+
"grad_norm": 0.6235030293464661,
|
8855 |
+
"learning_rate": 4.87880595342991e-06,
|
8856 |
+
"loss": 1.6992,
|
8857 |
+
"step": 1264
|
8858 |
+
},
|
8859 |
+
{
|
8860 |
+
"epoch": 0.083227790844943,
|
8861 |
+
"grad_norm": 0.6125441789627075,
|
8862 |
+
"learning_rate": 4.832776083120982e-06,
|
8863 |
+
"loss": 1.5375,
|
8864 |
+
"step": 1265
|
8865 |
+
},
|
8866 |
+
{
|
8867 |
+
"epoch": 0.08329358356497854,
|
8868 |
+
"grad_norm": 0.6090467572212219,
|
8869 |
+
"learning_rate": 4.7869533522294395e-06,
|
8870 |
+
"loss": 1.5203,
|
8871 |
+
"step": 1266
|
8872 |
+
},
|
8873 |
+
{
|
8874 |
+
"epoch": 0.08335937628501407,
|
8875 |
+
"grad_norm": 0.6368080377578735,
|
8876 |
+
"learning_rate": 4.741337970900866e-06,
|
8877 |
+
"loss": 1.6075,
|
8878 |
+
"step": 1267
|
8879 |
+
},
|
8880 |
+
{
|
8881 |
+
"epoch": 0.0834251690050496,
|
8882 |
+
"grad_norm": 0.6555090546607971,
|
8883 |
+
"learning_rate": 4.695930148329958e-06,
|
8884 |
+
"loss": 1.666,
|
8885 |
+
"step": 1268
|
8886 |
+
},
|
8887 |
+
{
|
8888 |
+
"epoch": 0.08349096172508512,
|
8889 |
+
"grad_norm": 0.6124194264411926,
|
8890 |
+
"learning_rate": 4.650730092759542e-06,
|
8891 |
+
"loss": 1.5002,
|
8892 |
+
"step": 1269
|
8893 |
+
},
|
8894 |
+
{
|
8895 |
+
"epoch": 0.08355675444512065,
|
8896 |
+
"grad_norm": 0.6861993670463562,
|
8897 |
+
"learning_rate": 4.605738011479604e-06,
|
8898 |
+
"loss": 1.8197,
|
8899 |
+
"step": 1270
|
8900 |
+
},
|
8901 |
+
{
|
8902 |
+
"epoch": 0.08362254716515617,
|
8903 |
+
"grad_norm": 0.6512139439582825,
|
8904 |
+
"learning_rate": 4.560954110826337e-06,
|
8905 |
+
"loss": 1.5302,
|
8906 |
+
"step": 1271
|
8907 |
+
},
|
8908 |
+
{
|
8909 |
+
"epoch": 0.0836883398851917,
|
8910 |
+
"grad_norm": 0.6771849393844604,
|
8911 |
+
"learning_rate": 4.516378596181237e-06,
|
8912 |
+
"loss": 1.5859,
|
8913 |
+
"step": 1272
|
8914 |
+
},
|
8915 |
+
{
|
8916 |
+
"epoch": 0.08375413260522724,
|
8917 |
+
"grad_norm": 0.6660013198852539,
|
8918 |
+
"learning_rate": 4.472011671970083e-06,
|
8919 |
+
"loss": 1.6946,
|
8920 |
+
"step": 1273
|
8921 |
+
},
|
8922 |
+
{
|
8923 |
+
"epoch": 0.08381992532526276,
|
8924 |
+
"grad_norm": 0.6350880861282349,
|
8925 |
+
"learning_rate": 4.427853541662091e-06,
|
8926 |
+
"loss": 1.5967,
|
8927 |
+
"step": 1274
|
8928 |
+
},
|
8929 |
+
{
|
8930 |
+
"epoch": 0.08388571804529829,
|
8931 |
+
"grad_norm": 0.6567019820213318,
|
8932 |
+
"learning_rate": 4.383904407768907e-06,
|
8933 |
+
"loss": 1.5232,
|
8934 |
+
"step": 1275
|
8935 |
+
},
|
8936 |
+
{
|
8937 |
+
"epoch": 0.08395151076533382,
|
8938 |
+
"grad_norm": 0.6577147245407104,
|
8939 |
+
"learning_rate": 4.340164471843722e-06,
|
8940 |
+
"loss": 1.5343,
|
8941 |
+
"step": 1276
|
8942 |
+
},
|
8943 |
+
{
|
8944 |
+
"epoch": 0.08401730348536934,
|
8945 |
+
"grad_norm": 0.7278039455413818,
|
8946 |
+
"learning_rate": 4.296633934480337e-06,
|
8947 |
+
"loss": 1.6968,
|
8948 |
+
"step": 1277
|
8949 |
+
},
|
8950 |
+
{
|
8951 |
+
"epoch": 0.08408309620540487,
|
8952 |
+
"grad_norm": 0.7200085520744324,
|
8953 |
+
"learning_rate": 4.253312995312231e-06,
|
8954 |
+
"loss": 1.5763,
|
8955 |
+
"step": 1278
|
8956 |
+
},
|
8957 |
+
{
|
8958 |
+
"epoch": 0.0841488889254404,
|
8959 |
+
"grad_norm": 0.6737844944000244,
|
8960 |
+
"learning_rate": 4.210201853011652e-06,
|
8961 |
+
"loss": 1.5614,
|
8962 |
+
"step": 1279
|
8963 |
+
},
|
8964 |
+
{
|
8965 |
+
"epoch": 0.08421468164547592,
|
8966 |
+
"grad_norm": 0.7159338593482971,
|
8967 |
+
"learning_rate": 4.167300705288718e-06,
|
8968 |
+
"loss": 1.7838,
|
8969 |
+
"step": 1280
|
8970 |
+
},
|
8971 |
+
{
|
8972 |
+
"epoch": 0.08428047436551146,
|
8973 |
+
"grad_norm": 0.7275623679161072,
|
8974 |
+
"learning_rate": 4.12460974889049e-06,
|
8975 |
+
"loss": 1.7268,
|
8976 |
+
"step": 1281
|
8977 |
+
},
|
8978 |
+
{
|
8979 |
+
"epoch": 0.08434626708554699,
|
8980 |
+
"grad_norm": 0.6852192878723145,
|
8981 |
+
"learning_rate": 4.082129179600097e-06,
|
8982 |
+
"loss": 1.4123,
|
8983 |
+
"step": 1282
|
8984 |
+
},
|
8985 |
+
{
|
8986 |
+
"epoch": 0.08441205980558251,
|
8987 |
+
"grad_norm": 0.7260058522224426,
|
8988 |
+
"learning_rate": 4.039859192235779e-06,
|
8989 |
+
"loss": 1.5387,
|
8990 |
+
"step": 1283
|
8991 |
+
},
|
8992 |
+
{
|
8993 |
+
"epoch": 0.08447785252561804,
|
8994 |
+
"grad_norm": 0.7276716828346252,
|
8995 |
+
"learning_rate": 3.99779998065008e-06,
|
8996 |
+
"loss": 1.6186,
|
8997 |
+
"step": 1284
|
8998 |
+
},
|
8999 |
+
{
|
9000 |
+
"epoch": 0.08454364524565357,
|
9001 |
+
"grad_norm": 0.8000338077545166,
|
9002 |
+
"learning_rate": 3.955951737728902e-06,
|
9003 |
+
"loss": 1.544,
|
9004 |
+
"step": 1285
|
9005 |
+
},
|
9006 |
+
{
|
9007 |
+
"epoch": 0.0846094379656891,
|
9008 |
+
"grad_norm": 0.7734615206718445,
|
9009 |
+
"learning_rate": 3.914314655390633e-06,
|
9010 |
+
"loss": 1.628,
|
9011 |
+
"step": 1286
|
9012 |
+
},
|
9013 |
+
{
|
9014 |
+
"epoch": 0.08467523068572462,
|
9015 |
+
"grad_norm": 0.742764949798584,
|
9016 |
+
"learning_rate": 3.872888924585255e-06,
|
9017 |
+
"loss": 1.5357,
|
9018 |
+
"step": 1287
|
9019 |
+
},
|
9020 |
+
{
|
9021 |
+
"epoch": 0.08474102340576015,
|
9022 |
+
"grad_norm": 0.7545043230056763,
|
9023 |
+
"learning_rate": 3.831674735293506e-06,
|
9024 |
+
"loss": 1.6103,
|
9025 |
+
"step": 1288
|
9026 |
+
},
|
9027 |
+
{
|
9028 |
+
"epoch": 0.08480681612579569,
|
9029 |
+
"grad_norm": 0.7599538564682007,
|
9030 |
+
"learning_rate": 3.790672276525936e-06,
|
9031 |
+
"loss": 1.5277,
|
9032 |
+
"step": 1289
|
9033 |
+
},
|
9034 |
+
{
|
9035 |
+
"epoch": 0.08487260884583121,
|
9036 |
+
"grad_norm": 0.7628370523452759,
|
9037 |
+
"learning_rate": 3.749881736322136e-06,
|
9038 |
+
"loss": 1.5536,
|
9039 |
+
"step": 1290
|
9040 |
+
},
|
9041 |
+
{
|
9042 |
+
"epoch": 0.08493840156586674,
|
9043 |
+
"grad_norm": 0.7717123031616211,
|
9044 |
+
"learning_rate": 3.7093033017497946e-06,
|
9045 |
+
"loss": 1.6771,
|
9046 |
+
"step": 1291
|
9047 |
+
},
|
9048 |
+
{
|
9049 |
+
"epoch": 0.08500419428590227,
|
9050 |
+
"grad_norm": 0.7848740220069885,
|
9051 |
+
"learning_rate": 3.668937158903901e-06,
|
9052 |
+
"loss": 1.4859,
|
9053 |
+
"step": 1292
|
9054 |
+
},
|
9055 |
+
{
|
9056 |
+
"epoch": 0.08506998700593779,
|
9057 |
+
"grad_norm": 0.8771865367889404,
|
9058 |
+
"learning_rate": 3.6287834929058295e-06,
|
9059 |
+
"loss": 1.5999,
|
9060 |
+
"step": 1293
|
9061 |
+
},
|
9062 |
+
{
|
9063 |
+
"epoch": 0.08513577972597332,
|
9064 |
+
"grad_norm": 0.8170807361602783,
|
9065 |
+
"learning_rate": 3.5888424879025495e-06,
|
9066 |
+
"loss": 1.4485,
|
9067 |
+
"step": 1294
|
9068 |
+
},
|
9069 |
+
{
|
9070 |
+
"epoch": 0.08520157244600884,
|
9071 |
+
"grad_norm": 0.8586212396621704,
|
9072 |
+
"learning_rate": 3.5491143270657446e-06,
|
9073 |
+
"loss": 1.6085,
|
9074 |
+
"step": 1295
|
9075 |
+
},
|
9076 |
+
{
|
9077 |
+
"epoch": 0.08526736516604437,
|
9078 |
+
"grad_norm": 0.8808199763298035,
|
9079 |
+
"learning_rate": 3.5095991925909845e-06,
|
9080 |
+
"loss": 1.4775,
|
9081 |
+
"step": 1296
|
9082 |
+
},
|
9083 |
+
{
|
9084 |
+
"epoch": 0.08533315788607991,
|
9085 |
+
"grad_norm": 0.8514795303344727,
|
9086 |
+
"learning_rate": 3.470297265696887e-06,
|
9087 |
+
"loss": 1.4806,
|
9088 |
+
"step": 1297
|
9089 |
+
},
|
9090 |
+
{
|
9091 |
+
"epoch": 0.08539895060611544,
|
9092 |
+
"grad_norm": 0.9353112578392029,
|
9093 |
+
"learning_rate": 3.4312087266242963e-06,
|
9094 |
+
"loss": 1.4445,
|
9095 |
+
"step": 1298
|
9096 |
+
},
|
9097 |
+
{
|
9098 |
+
"epoch": 0.08546474332615096,
|
9099 |
+
"grad_norm": 0.9689911007881165,
|
9100 |
+
"learning_rate": 3.3923337546354297e-06,
|
9101 |
+
"loss": 1.2047,
|
9102 |
+
"step": 1299
|
9103 |
+
},
|
9104 |
+
{
|
9105 |
+
"epoch": 0.08553053604618649,
|
9106 |
+
"grad_norm": 1.2559093236923218,
|
9107 |
+
"learning_rate": 3.3536725280130744e-06,
|
9108 |
+
"loss": 1.2073,
|
9109 |
+
"step": 1300
|
9110 |
+
},
|
9111 |
+
{
|
9112 |
+
"epoch": 0.08559632876622202,
|
9113 |
+
"grad_norm": 0.5083854794502258,
|
9114 |
+
"learning_rate": 3.315225224059809e-06,
|
9115 |
+
"loss": 1.5963,
|
9116 |
+
"step": 1301
|
9117 |
+
},
|
9118 |
+
{
|
9119 |
+
"epoch": 0.08566212148625754,
|
9120 |
+
"grad_norm": 0.5269922018051147,
|
9121 |
+
"learning_rate": 3.2769920190971027e-06,
|
9122 |
+
"loss": 1.5835,
|
9123 |
+
"step": 1302
|
9124 |
+
},
|
9125 |
+
{
|
9126 |
+
"epoch": 0.08572791420629307,
|
9127 |
+
"grad_norm": 0.5758770704269409,
|
9128 |
+
"learning_rate": 3.2389730884645807e-06,
|
9129 |
+
"loss": 1.6247,
|
9130 |
+
"step": 1303
|
9131 |
+
},
|
9132 |
+
{
|
9133 |
+
"epoch": 0.08579370692632861,
|
9134 |
+
"grad_norm": 0.5545147061347961,
|
9135 |
+
"learning_rate": 3.2011686065191895e-06,
|
9136 |
+
"loss": 1.6925,
|
9137 |
+
"step": 1304
|
9138 |
+
},
|
9139 |
+
{
|
9140 |
+
"epoch": 0.08585949964636413,
|
9141 |
+
"grad_norm": 0.5974170565605164,
|
9142 |
+
"learning_rate": 3.163578746634388e-06,
|
9143 |
+
"loss": 1.7166,
|
9144 |
+
"step": 1305
|
9145 |
+
},
|
9146 |
+
{
|
9147 |
+
"epoch": 0.08592529236639966,
|
9148 |
+
"grad_norm": 0.5378652215003967,
|
9149 |
+
"learning_rate": 3.1262036811993856e-06,
|
9150 |
+
"loss": 1.5484,
|
9151 |
+
"step": 1306
|
9152 |
+
},
|
9153 |
+
{
|
9154 |
+
"epoch": 0.08599108508643519,
|
9155 |
+
"grad_norm": 0.6244592666625977,
|
9156 |
+
"learning_rate": 3.0890435816183226e-06,
|
9157 |
+
"loss": 1.8135,
|
9158 |
+
"step": 1307
|
9159 |
+
},
|
9160 |
+
{
|
9161 |
+
"epoch": 0.08605687780647071,
|
9162 |
+
"grad_norm": 0.5631170272827148,
|
9163 |
+
"learning_rate": 3.0520986183095014e-06,
|
9164 |
+
"loss": 1.6121,
|
9165 |
+
"step": 1308
|
9166 |
+
},
|
9167 |
+
{
|
9168 |
+
"epoch": 0.08612267052650624,
|
9169 |
+
"grad_norm": 0.5755576491355896,
|
9170 |
+
"learning_rate": 3.0153689607045845e-06,
|
9171 |
+
"loss": 1.5121,
|
9172 |
+
"step": 1309
|
9173 |
+
},
|
9174 |
+
{
|
9175 |
+
"epoch": 0.08618846324654177,
|
9176 |
+
"grad_norm": 0.6194965839385986,
|
9177 |
+
"learning_rate": 2.9788547772478416e-06,
|
9178 |
+
"loss": 1.5921,
|
9179 |
+
"step": 1310
|
9180 |
+
},
|
9181 |
+
{
|
9182 |
+
"epoch": 0.08625425596657729,
|
9183 |
+
"grad_norm": 0.6144405603408813,
|
9184 |
+
"learning_rate": 2.9425562353953604e-06,
|
9185 |
+
"loss": 1.683,
|
9186 |
+
"step": 1311
|
9187 |
+
},
|
9188 |
+
{
|
9189 |
+
"epoch": 0.08632004868661283,
|
9190 |
+
"grad_norm": 0.6393880844116211,
|
9191 |
+
"learning_rate": 2.9064735016142873e-06,
|
9192 |
+
"loss": 1.6299,
|
9193 |
+
"step": 1312
|
9194 |
+
},
|
9195 |
+
{
|
9196 |
+
"epoch": 0.08638584140664836,
|
9197 |
+
"grad_norm": 0.6104136109352112,
|
9198 |
+
"learning_rate": 2.870606741382059e-06,
|
9199 |
+
"loss": 1.6589,
|
9200 |
+
"step": 1313
|
9201 |
+
},
|
9202 |
+
{
|
9203 |
+
"epoch": 0.08645163412668389,
|
9204 |
+
"grad_norm": 0.6379032135009766,
|
9205 |
+
"learning_rate": 2.83495611918565e-06,
|
9206 |
+
"loss": 1.5917,
|
9207 |
+
"step": 1314
|
9208 |
+
},
|
9209 |
+
{
|
9210 |
+
"epoch": 0.08651742684671941,
|
9211 |
+
"grad_norm": 0.6215010285377502,
|
9212 |
+
"learning_rate": 2.7995217985208098e-06,
|
9213 |
+
"loss": 1.5812,
|
9214 |
+
"step": 1315
|
9215 |
+
},
|
9216 |
+
{
|
9217 |
+
"epoch": 0.08658321956675494,
|
9218 |
+
"grad_norm": 0.6086938977241516,
|
9219 |
+
"learning_rate": 2.7643039418913e-06,
|
9220 |
+
"loss": 1.5549,
|
9221 |
+
"step": 1316
|
9222 |
+
},
|
9223 |
+
{
|
9224 |
+
"epoch": 0.08664901228679046,
|
9225 |
+
"grad_norm": 0.6653648018836975,
|
9226 |
+
"learning_rate": 2.729302710808196e-06,
|
9227 |
+
"loss": 1.4754,
|
9228 |
+
"step": 1317
|
9229 |
+
},
|
9230 |
+
{
|
9231 |
+
"epoch": 0.08671480500682599,
|
9232 |
+
"grad_norm": 0.6537653803825378,
|
9233 |
+
"learning_rate": 2.6945182657891034e-06,
|
9234 |
+
"loss": 1.7067,
|
9235 |
+
"step": 1318
|
9236 |
+
},
|
9237 |
+
{
|
9238 |
+
"epoch": 0.08678059772686152,
|
9239 |
+
"grad_norm": 0.6637395620346069,
|
9240 |
+
"learning_rate": 2.6599507663574384e-06,
|
9241 |
+
"loss": 1.5416,
|
9242 |
+
"step": 1319
|
9243 |
+
},
|
9244 |
+
{
|
9245 |
+
"epoch": 0.08684639044689706,
|
9246 |
+
"grad_norm": 0.6458157300949097,
|
9247 |
+
"learning_rate": 2.6256003710416864e-06,
|
9248 |
+
"loss": 1.5356,
|
9249 |
+
"step": 1320
|
9250 |
+
},
|
9251 |
+
{
|
9252 |
+
"epoch": 0.08691218316693258,
|
9253 |
+
"grad_norm": 0.6820713877677917,
|
9254 |
+
"learning_rate": 2.5914672373746674e-06,
|
9255 |
+
"loss": 1.797,
|
9256 |
+
"step": 1321
|
9257 |
+
},
|
9258 |
+
{
|
9259 |
+
"epoch": 0.08697797588696811,
|
9260 |
+
"grad_norm": 0.6089844107627869,
|
9261 |
+
"learning_rate": 2.5575515218928592e-06,
|
9262 |
+
"loss": 1.5584,
|
9263 |
+
"step": 1322
|
9264 |
+
},
|
9265 |
+
{
|
9266 |
+
"epoch": 0.08704376860700364,
|
9267 |
+
"grad_norm": 0.7012182474136353,
|
9268 |
+
"learning_rate": 2.5238533801356324e-06,
|
9269 |
+
"loss": 1.6888,
|
9270 |
+
"step": 1323
|
9271 |
+
},
|
9272 |
+
{
|
9273 |
+
"epoch": 0.08710956132703916,
|
9274 |
+
"grad_norm": 0.6670122742652893,
|
9275 |
+
"learning_rate": 2.490372966644544e-06,
|
9276 |
+
"loss": 1.6455,
|
9277 |
+
"step": 1324
|
9278 |
+
},
|
9279 |
+
{
|
9280 |
+
"epoch": 0.08717535404707469,
|
9281 |
+
"grad_norm": 0.6817238330841064,
|
9282 |
+
"learning_rate": 2.457110434962645e-06,
|
9283 |
+
"loss": 1.575,
|
9284 |
+
"step": 1325
|
9285 |
+
},
|
9286 |
+
{
|
9287 |
+
"epoch": 0.08724114676711021,
|
9288 |
+
"grad_norm": 0.6881915926933289,
|
9289 |
+
"learning_rate": 2.424065937633768e-06,
|
9290 |
+
"loss": 1.7438,
|
9291 |
+
"step": 1326
|
9292 |
+
},
|
9293 |
+
{
|
9294 |
+
"epoch": 0.08730693948714574,
|
9295 |
+
"grad_norm": 0.6820628643035889,
|
9296 |
+
"learning_rate": 2.3912396262018357e-06,
|
9297 |
+
"loss": 1.5352,
|
9298 |
+
"step": 1327
|
9299 |
+
},
|
9300 |
+
{
|
9301 |
+
"epoch": 0.08737273220718128,
|
9302 |
+
"grad_norm": 0.7019688487052917,
|
9303 |
+
"learning_rate": 2.3586316512101416e-06,
|
9304 |
+
"loss": 1.6225,
|
9305 |
+
"step": 1328
|
9306 |
+
},
|
9307 |
+
{
|
9308 |
+
"epoch": 0.0874385249272168,
|
9309 |
+
"grad_norm": 0.6856455206871033,
|
9310 |
+
"learning_rate": 2.3262421622006868e-06,
|
9311 |
+
"loss": 1.7121,
|
9312 |
+
"step": 1329
|
9313 |
+
},
|
9314 |
+
{
|
9315 |
+
"epoch": 0.08750431764725233,
|
9316 |
+
"grad_norm": 0.7237346768379211,
|
9317 |
+
"learning_rate": 2.29407130771348e-06,
|
9318 |
+
"loss": 1.6317,
|
9319 |
+
"step": 1330
|
9320 |
+
},
|
9321 |
+
{
|
9322 |
+
"epoch": 0.08757011036728786,
|
9323 |
+
"grad_norm": 0.728233814239502,
|
9324 |
+
"learning_rate": 2.26211923528587e-06,
|
9325 |
+
"loss": 1.5668,
|
9326 |
+
"step": 1331
|
9327 |
+
},
|
9328 |
+
{
|
9329 |
+
"epoch": 0.08763590308732339,
|
9330 |
+
"grad_norm": 0.6574517488479614,
|
9331 |
+
"learning_rate": 2.2303860914518306e-06,
|
9332 |
+
"loss": 1.496,
|
9333 |
+
"step": 1332
|
9334 |
+
},
|
9335 |
+
{
|
9336 |
+
"epoch": 0.08770169580735891,
|
9337 |
+
"grad_norm": 0.7108151912689209,
|
9338 |
+
"learning_rate": 2.1988720217413494e-06,
|
9339 |
+
"loss": 1.4753,
|
9340 |
+
"step": 1333
|
9341 |
+
},
|
9342 |
+
{
|
9343 |
+
"epoch": 0.08776748852739444,
|
9344 |
+
"grad_norm": 0.7035221457481384,
|
9345 |
+
"learning_rate": 2.1675771706797132e-06,
|
9346 |
+
"loss": 1.4783,
|
9347 |
+
"step": 1334
|
9348 |
+
},
|
9349 |
+
{
|
9350 |
+
"epoch": 0.08783328124742998,
|
9351 |
+
"grad_norm": 0.713604211807251,
|
9352 |
+
"learning_rate": 2.136501681786862e-06,
|
9353 |
+
"loss": 1.6723,
|
9354 |
+
"step": 1335
|
9355 |
+
},
|
9356 |
+
{
|
9357 |
+
"epoch": 0.0878990739674655,
|
9358 |
+
"grad_norm": 0.7584853768348694,
|
9359 |
+
"learning_rate": 2.10564569757673e-06,
|
9360 |
+
"loss": 1.457,
|
9361 |
+
"step": 1336
|
9362 |
+
},
|
9363 |
+
{
|
9364 |
+
"epoch": 0.08796486668750103,
|
9365 |
+
"grad_norm": 0.7580429911613464,
|
9366 |
+
"learning_rate": 2.0750093595565733e-06,
|
9367 |
+
"loss": 1.597,
|
9368 |
+
"step": 1337
|
9369 |
+
},
|
9370 |
+
{
|
9371 |
+
"epoch": 0.08803065940753656,
|
9372 |
+
"grad_norm": 0.7131192684173584,
|
9373 |
+
"learning_rate": 2.0445928082263645e-06,
|
9374 |
+
"loss": 1.5909,
|
9375 |
+
"step": 1338
|
9376 |
+
},
|
9377 |
+
{
|
9378 |
+
"epoch": 0.08809645212757208,
|
9379 |
+
"grad_norm": 0.7904130816459656,
|
9380 |
+
"learning_rate": 2.01439618307811e-06,
|
9381 |
+
"loss": 1.5464,
|
9382 |
+
"step": 1339
|
9383 |
+
},
|
9384 |
+
{
|
9385 |
+
"epoch": 0.08816224484760761,
|
9386 |
+
"grad_norm": 0.8404092192649841,
|
9387 |
+
"learning_rate": 1.984419622595224e-06,
|
9388 |
+
"loss": 1.616,
|
9389 |
+
"step": 1340
|
9390 |
+
},
|
9391 |
+
{
|
9392 |
+
"epoch": 0.08822803756764314,
|
9393 |
+
"grad_norm": 0.7262335419654846,
|
9394 |
+
"learning_rate": 1.954663264251888e-06,
|
9395 |
+
"loss": 1.5096,
|
9396 |
+
"step": 1341
|
9397 |
+
},
|
9398 |
+
{
|
9399 |
+
"epoch": 0.08829383028767866,
|
9400 |
+
"grad_norm": 0.8143447041511536,
|
9401 |
+
"learning_rate": 1.925127244512426e-06,
|
9402 |
+
"loss": 1.4777,
|
9403 |
+
"step": 1342
|
9404 |
+
},
|
9405 |
+
{
|
9406 |
+
"epoch": 0.0883596230077142,
|
9407 |
+
"grad_norm": 0.8100689053535461,
|
9408 |
+
"learning_rate": 1.895811698830685e-06,
|
9409 |
+
"loss": 1.4577,
|
9410 |
+
"step": 1343
|
9411 |
+
},
|
9412 |
+
{
|
9413 |
+
"epoch": 0.08842541572774973,
|
9414 |
+
"grad_norm": 0.8460294008255005,
|
9415 |
+
"learning_rate": 1.8667167616493896e-06,
|
9416 |
+
"loss": 1.5934,
|
9417 |
+
"step": 1344
|
9418 |
+
},
|
9419 |
+
{
|
9420 |
+
"epoch": 0.08849120844778526,
|
9421 |
+
"grad_norm": 0.8502865433692932,
|
9422 |
+
"learning_rate": 1.8378425663995559e-06,
|
9423 |
+
"loss": 1.435,
|
9424 |
+
"step": 1345
|
9425 |
+
},
|
9426 |
+
{
|
9427 |
+
"epoch": 0.08855700116782078,
|
9428 |
+
"grad_norm": 0.9002560377120972,
|
9429 |
+
"learning_rate": 1.8091892454998594e-06,
|
9430 |
+
"loss": 1.459,
|
9431 |
+
"step": 1346
|
9432 |
+
},
|
9433 |
+
{
|
9434 |
+
"epoch": 0.08862279388785631,
|
9435 |
+
"grad_norm": 0.8573265075683594,
|
9436 |
+
"learning_rate": 1.7807569303560367e-06,
|
9437 |
+
"loss": 1.5029,
|
9438 |
+
"step": 1347
|
9439 |
+
},
|
9440 |
+
{
|
9441 |
+
"epoch": 0.08868858660789183,
|
9442 |
+
"grad_norm": 0.8953090310096741,
|
9443 |
+
"learning_rate": 1.7525457513602683e-06,
|
9444 |
+
"loss": 1.4322,
|
9445 |
+
"step": 1348
|
9446 |
+
},
|
9447 |
+
{
|
9448 |
+
"epoch": 0.08875437932792736,
|
9449 |
+
"grad_norm": 0.9949759244918823,
|
9450 |
+
"learning_rate": 1.7245558378906013e-06,
|
9451 |
+
"loss": 1.4967,
|
9452 |
+
"step": 1349
|
9453 |
+
},
|
9454 |
+
{
|
9455 |
+
"epoch": 0.08882017204796289,
|
9456 |
+
"grad_norm": 1.1277623176574707,
|
9457 |
+
"learning_rate": 1.6967873183103556e-06,
|
9458 |
+
"loss": 1.0865,
|
9459 |
+
"step": 1350
|
9460 |
+
},
|
9461 |
+
{
|
9462 |
+
"epoch": 0.08888596476799843,
|
9463 |
+
"grad_norm": 0.5030365586280823,
|
9464 |
+
"learning_rate": 1.6692403199675078e-06,
|
9465 |
+
"loss": 1.6924,
|
9466 |
+
"step": 1351
|
9467 |
+
},
|
9468 |
+
{
|
9469 |
+
"epoch": 0.08895175748803395,
|
9470 |
+
"grad_norm": 0.5737792253494263,
|
9471 |
+
"learning_rate": 1.641914969194147e-06,
|
9472 |
+
"loss": 1.7576,
|
9473 |
+
"step": 1352
|
9474 |
+
},
|
9475 |
+
{
|
9476 |
+
"epoch": 0.08901755020806948,
|
9477 |
+
"grad_norm": 0.5396728515625,
|
9478 |
+
"learning_rate": 1.6148113913058427e-06,
|
9479 |
+
"loss": 1.4578,
|
9480 |
+
"step": 1353
|
9481 |
+
},
|
9482 |
+
{
|
9483 |
+
"epoch": 0.089083342928105,
|
9484 |
+
"grad_norm": 0.535679280757904,
|
9485 |
+
"learning_rate": 1.587929710601127e-06,
|
9486 |
+
"loss": 1.5822,
|
9487 |
+
"step": 1354
|
9488 |
+
},
|
9489 |
+
{
|
9490 |
+
"epoch": 0.08914913564814053,
|
9491 |
+
"grad_norm": 0.5770007371902466,
|
9492 |
+
"learning_rate": 1.5612700503608968e-06,
|
9493 |
+
"loss": 1.7699,
|
9494 |
+
"step": 1355
|
9495 |
+
},
|
9496 |
+
{
|
9497 |
+
"epoch": 0.08921492836817606,
|
9498 |
+
"grad_norm": 0.5972631573677063,
|
9499 |
+
"learning_rate": 1.5348325328478408e-06,
|
9500 |
+
"loss": 1.6394,
|
9501 |
+
"step": 1356
|
9502 |
+
},
|
9503 |
+
{
|
9504 |
+
"epoch": 0.08928072108821158,
|
9505 |
+
"grad_norm": 0.5818490386009216,
|
9506 |
+
"learning_rate": 1.5086172793059017e-06,
|
9507 |
+
"loss": 1.7062,
|
9508 |
+
"step": 1357
|
9509 |
+
},
|
9510 |
+
{
|
9511 |
+
"epoch": 0.08934651380824711,
|
9512 |
+
"grad_norm": 0.6017389893531799,
|
9513 |
+
"learning_rate": 1.4826244099596986e-06,
|
9514 |
+
"loss": 1.6664,
|
9515 |
+
"step": 1358
|
9516 |
+
},
|
9517 |
+
{
|
9518 |
+
"epoch": 0.08941230652828265,
|
9519 |
+
"grad_norm": 0.6041622161865234,
|
9520 |
+
"learning_rate": 1.4568540440139777e-06,
|
9521 |
+
"loss": 1.7427,
|
9522 |
+
"step": 1359
|
9523 |
+
},
|
9524 |
+
{
|
9525 |
+
"epoch": 0.08947809924831818,
|
9526 |
+
"grad_norm": 0.8328604698181152,
|
9527 |
+
"learning_rate": 1.4313062996530847e-06,
|
9528 |
+
"loss": 1.5239,
|
9529 |
+
"step": 1360
|
9530 |
+
},
|
9531 |
+
{
|
9532 |
+
"epoch": 0.0895438919683537,
|
9533 |
+
"grad_norm": 0.6048279404640198,
|
9534 |
+
"learning_rate": 1.4059812940404093e-06,
|
9535 |
+
"loss": 1.7337,
|
9536 |
+
"step": 1361
|
9537 |
+
},
|
9538 |
+
{
|
9539 |
+
"epoch": 0.08960968468838923,
|
9540 |
+
"grad_norm": 0.6227853894233704,
|
9541 |
+
"learning_rate": 1.3808791433178369e-06,
|
9542 |
+
"loss": 1.451,
|
9543 |
+
"step": 1362
|
9544 |
+
},
|
9545 |
+
{
|
9546 |
+
"epoch": 0.08967547740842476,
|
9547 |
+
"grad_norm": 0.5862622857093811,
|
9548 |
+
"learning_rate": 1.3559999626052477e-06,
|
9549 |
+
"loss": 1.4699,
|
9550 |
+
"step": 1363
|
9551 |
+
},
|
9552 |
+
{
|
9553 |
+
"epoch": 0.08974127012846028,
|
9554 |
+
"grad_norm": 0.6687149405479431,
|
9555 |
+
"learning_rate": 1.33134386599994e-06,
|
9556 |
+
"loss": 1.5708,
|
9557 |
+
"step": 1364
|
9558 |
+
},
|
9559 |
+
{
|
9560 |
+
"epoch": 0.08980706284849581,
|
9561 |
+
"grad_norm": 0.65779709815979,
|
9562 |
+
"learning_rate": 1.3069109665761693e-06,
|
9563 |
+
"loss": 1.6006,
|
9564 |
+
"step": 1365
|
9565 |
+
},
|
9566 |
+
{
|
9567 |
+
"epoch": 0.08987285556853135,
|
9568 |
+
"grad_norm": 0.6536720991134644,
|
9569 |
+
"learning_rate": 1.2827013763845707e-06,
|
9570 |
+
"loss": 1.6428,
|
9571 |
+
"step": 1366
|
9572 |
+
},
|
9573 |
+
{
|
9574 |
+
"epoch": 0.08993864828856687,
|
9575 |
+
"grad_norm": 0.6218163967132568,
|
9576 |
+
"learning_rate": 1.2587152064516827e-06,
|
9577 |
+
"loss": 1.5861,
|
9578 |
+
"step": 1367
|
9579 |
+
},
|
9580 |
+
{
|
9581 |
+
"epoch": 0.0900044410086024,
|
9582 |
+
"grad_norm": 0.6612845063209534,
|
9583 |
+
"learning_rate": 1.2349525667794293e-06,
|
9584 |
+
"loss": 1.7162,
|
9585 |
+
"step": 1368
|
9586 |
+
},
|
9587 |
+
{
|
9588 |
+
"epoch": 0.09007023372863793,
|
9589 |
+
"grad_norm": 0.6771432757377625,
|
9590 |
+
"learning_rate": 1.211413566344599e-06,
|
9591 |
+
"loss": 1.7318,
|
9592 |
+
"step": 1369
|
9593 |
+
},
|
9594 |
+
{
|
9595 |
+
"epoch": 0.09013602644867345,
|
9596 |
+
"grad_norm": 0.6161140203475952,
|
9597 |
+
"learning_rate": 1.1880983130983626e-06,
|
9598 |
+
"loss": 1.5076,
|
9599 |
+
"step": 1370
|
9600 |
+
},
|
9601 |
+
{
|
9602 |
+
"epoch": 0.09020181916870898,
|
9603 |
+
"grad_norm": 0.6894911527633667,
|
9604 |
+
"learning_rate": 1.1650069139657826e-06,
|
9605 |
+
"loss": 1.7694,
|
9606 |
+
"step": 1371
|
9607 |
+
},
|
9608 |
+
{
|
9609 |
+
"epoch": 0.0902676118887445,
|
9610 |
+
"grad_norm": 0.6760029196739197,
|
9611 |
+
"learning_rate": 1.1421394748453108e-06,
|
9612 |
+
"loss": 1.6852,
|
9613 |
+
"step": 1372
|
9614 |
+
},
|
9615 |
+
{
|
9616 |
+
"epoch": 0.09033340460878003,
|
9617 |
+
"grad_norm": 0.716984748840332,
|
9618 |
+
"learning_rate": 1.1194961006082972e-06,
|
9619 |
+
"loss": 1.4935,
|
9620 |
+
"step": 1373
|
9621 |
+
},
|
9622 |
+
{
|
9623 |
+
"epoch": 0.09039919732881557,
|
9624 |
+
"grad_norm": 0.6679751873016357,
|
9625 |
+
"learning_rate": 1.0970768950985199e-06,
|
9626 |
+
"loss": 1.7724,
|
9627 |
+
"step": 1374
|
9628 |
+
},
|
9629 |
+
{
|
9630 |
+
"epoch": 0.0904649900488511,
|
9631 |
+
"grad_norm": 0.6891523599624634,
|
9632 |
+
"learning_rate": 1.074881961131724e-06,
|
9633 |
+
"loss": 1.5886,
|
9634 |
+
"step": 1375
|
9635 |
+
},
|
9636 |
+
{
|
9637 |
+
"epoch": 0.09053078276888663,
|
9638 |
+
"grad_norm": 0.6760563254356384,
|
9639 |
+
"learning_rate": 1.0529114004951047e-06,
|
9640 |
+
"loss": 1.632,
|
9641 |
+
"step": 1376
|
9642 |
+
},
|
9643 |
+
{
|
9644 |
+
"epoch": 0.09059657548892215,
|
9645 |
+
"grad_norm": 0.6902234554290771,
|
9646 |
+
"learning_rate": 1.0311653139468969e-06,
|
9647 |
+
"loss": 1.6434,
|
9648 |
+
"step": 1377
|
9649 |
+
},
|
9650 |
+
{
|
9651 |
+
"epoch": 0.09066236820895768,
|
9652 |
+
"grad_norm": 0.6432000398635864,
|
9653 |
+
"learning_rate": 1.0096438012158539e-06,
|
9654 |
+
"loss": 1.5294,
|
9655 |
+
"step": 1378
|
9656 |
+
},
|
9657 |
+
{
|
9658 |
+
"epoch": 0.0907281609289932,
|
9659 |
+
"grad_norm": 0.6482113599777222,
|
9660 |
+
"learning_rate": 9.883469610008577e-07,
|
9661 |
+
"loss": 1.5759,
|
9662 |
+
"step": 1379
|
9663 |
+
},
|
9664 |
+
{
|
9665 |
+
"epoch": 0.09079395364902873,
|
9666 |
+
"grad_norm": 0.7423402070999146,
|
9667 |
+
"learning_rate": 9.672748909703934e-07,
|
9668 |
+
"loss": 1.697,
|
9669 |
+
"step": 1380
|
9670 |
+
},
|
9671 |
+
{
|
9672 |
+
"epoch": 0.09085974636906426,
|
9673 |
+
"grad_norm": 0.7108086943626404,
|
9674 |
+
"learning_rate": 9.46427687762158e-07,
|
9675 |
+
"loss": 1.6335,
|
9676 |
+
"step": 1381
|
9677 |
+
},
|
9678 |
+
{
|
9679 |
+
"epoch": 0.0909255390890998,
|
9680 |
+
"grad_norm": 0.6593964695930481,
|
9681 |
+
"learning_rate": 9.258054469825972e-07,
|
9682 |
+
"loss": 1.5451,
|
9683 |
+
"step": 1382
|
9684 |
+
},
|
9685 |
+
{
|
9686 |
+
"epoch": 0.09099133180913532,
|
9687 |
+
"grad_norm": 0.7182891964912415,
|
9688 |
+
"learning_rate": 9.054082632064642e-07,
|
9689 |
+
"loss": 1.4827,
|
9690 |
+
"step": 1383
|
9691 |
+
},
|
9692 |
+
{
|
9693 |
+
"epoch": 0.09105712452917085,
|
9694 |
+
"grad_norm": 0.7174695730209351,
|
9695 |
+
"learning_rate": 8.852362299763772e-07,
|
9696 |
+
"loss": 1.652,
|
9697 |
+
"step": 1384
|
9698 |
+
},
|
9699 |
+
{
|
9700 |
+
"epoch": 0.09112291724920638,
|
9701 |
+
"grad_norm": 0.7062637209892273,
|
9702 |
+
"learning_rate": 8.652894398024136e-07,
|
9703 |
+
"loss": 1.478,
|
9704 |
+
"step": 1385
|
9705 |
+
},
|
9706 |
+
{
|
9707 |
+
"epoch": 0.0911887099692419,
|
9708 |
+
"grad_norm": 0.737032949924469,
|
9709 |
+
"learning_rate": 8.455679841616659e-07,
|
9710 |
+
"loss": 1.6173,
|
9711 |
+
"step": 1386
|
9712 |
+
},
|
9713 |
+
{
|
9714 |
+
"epoch": 0.09125450268927743,
|
9715 |
+
"grad_norm": 0.8026255369186401,
|
9716 |
+
"learning_rate": 8.260719534978368e-07,
|
9717 |
+
"loss": 1.5671,
|
9718 |
+
"step": 1387
|
9719 |
+
},
|
9720 |
+
{
|
9721 |
+
"epoch": 0.09132029540931295,
|
9722 |
+
"grad_norm": 0.7752266526222229,
|
9723 |
+
"learning_rate": 8.06801437220811e-07,
|
9724 |
+
"loss": 1.7083,
|
9725 |
+
"step": 1388
|
9726 |
+
},
|
9727 |
+
{
|
9728 |
+
"epoch": 0.0913860881293485,
|
9729 |
+
"grad_norm": 0.7705795764923096,
|
9730 |
+
"learning_rate": 7.877565237062623e-07,
|
9731 |
+
"loss": 1.4577,
|
9732 |
+
"step": 1389
|
9733 |
+
},
|
9734 |
+
{
|
9735 |
+
"epoch": 0.09145188084938402,
|
9736 |
+
"grad_norm": 0.799823522567749,
|
9737 |
+
"learning_rate": 7.689373002952305e-07,
|
9738 |
+
"loss": 1.7553,
|
9739 |
+
"step": 1390
|
9740 |
+
},
|
9741 |
+
{
|
9742 |
+
"epoch": 0.09151767356941955,
|
9743 |
+
"grad_norm": 0.7618536353111267,
|
9744 |
+
"learning_rate": 7.503438532937168e-07,
|
9745 |
+
"loss": 1.4927,
|
9746 |
+
"step": 1391
|
9747 |
+
},
|
9748 |
+
{
|
9749 |
+
"epoch": 0.09158346628945507,
|
9750 |
+
"grad_norm": 0.9011740684509277,
|
9751 |
+
"learning_rate": 7.319762679723174e-07,
|
9752 |
+
"loss": 1.6078,
|
9753 |
+
"step": 1392
|
9754 |
+
},
|
9755 |
+
{
|
9756 |
+
"epoch": 0.0916492590094906,
|
9757 |
+
"grad_norm": 0.8119943737983704,
|
9758 |
+
"learning_rate": 7.138346285658071e-07,
|
9759 |
+
"loss": 1.3945,
|
9760 |
+
"step": 1393
|
9761 |
+
},
|
9762 |
+
{
|
9763 |
+
"epoch": 0.09171505172952613,
|
9764 |
+
"grad_norm": 0.8282358050346375,
|
9765 |
+
"learning_rate": 6.959190182727615e-07,
|
9766 |
+
"loss": 1.6105,
|
9767 |
+
"step": 1394
|
9768 |
+
},
|
9769 |
+
{
|
9770 |
+
"epoch": 0.09178084444956165,
|
9771 |
+
"grad_norm": 0.8864809274673462,
|
9772 |
+
"learning_rate": 6.782295192551691e-07,
|
9773 |
+
"loss": 1.6667,
|
9774 |
+
"step": 1395
|
9775 |
+
},
|
9776 |
+
{
|
9777 |
+
"epoch": 0.09184663716959718,
|
9778 |
+
"grad_norm": 0.8417677879333496,
|
9779 |
+
"learning_rate": 6.607662126380587e-07,
|
9780 |
+
"loss": 1.513,
|
9781 |
+
"step": 1396
|
9782 |
+
},
|
9783 |
+
{
|
9784 |
+
"epoch": 0.09191242988963272,
|
9785 |
+
"grad_norm": 0.8300901651382446,
|
9786 |
+
"learning_rate": 6.43529178509139e-07,
|
9787 |
+
"loss": 1.3415,
|
9788 |
+
"step": 1397
|
9789 |
+
},
|
9790 |
+
{
|
9791 |
+
"epoch": 0.09197822260966824,
|
9792 |
+
"grad_norm": 0.9474268555641174,
|
9793 |
+
"learning_rate": 6.265184959184101e-07,
|
9794 |
+
"loss": 1.56,
|
9795 |
+
"step": 1398
|
9796 |
+
},
|
9797 |
+
{
|
9798 |
+
"epoch": 0.09204401532970377,
|
9799 |
+
"grad_norm": 1.1085734367370605,
|
9800 |
+
"learning_rate": 6.097342428778185e-07,
|
9801 |
+
"loss": 1.4722,
|
9802 |
+
"step": 1399
|
9803 |
+
},
|
9804 |
+
{
|
9805 |
+
"epoch": 0.0921098080497393,
|
9806 |
+
"grad_norm": 1.2595409154891968,
|
9807 |
+
"learning_rate": 5.931764963608866e-07,
|
9808 |
+
"loss": 1.2695,
|
9809 |
+
"step": 1400
|
9810 |
+
},
|
9811 |
+
{
|
9812 |
+
"epoch": 0.09217560076977482,
|
9813 |
+
"grad_norm": 0.4739496409893036,
|
9814 |
+
"learning_rate": 5.768453323023615e-07,
|
9815 |
+
"loss": 1.5488,
|
9816 |
+
"step": 1401
|
9817 |
+
},
|
9818 |
+
{
|
9819 |
+
"epoch": 0.09224139348981035,
|
9820 |
+
"grad_norm": 0.550033688545227,
|
9821 |
+
"learning_rate": 5.60740825597883e-07,
|
9822 |
+
"loss": 1.761,
|
9823 |
+
"step": 1402
|
9824 |
+
},
|
9825 |
+
{
|
9826 |
+
"epoch": 0.09230718620984588,
|
9827 |
+
"grad_norm": 0.5993760824203491,
|
9828 |
+
"learning_rate": 5.448630501036112e-07,
|
9829 |
+
"loss": 1.8681,
|
9830 |
+
"step": 1403
|
9831 |
+
},
|
9832 |
+
{
|
9833 |
+
"epoch": 0.0923729789298814,
|
9834 |
+
"grad_norm": 0.5492662787437439,
|
9835 |
+
"learning_rate": 5.292120786359267e-07,
|
9836 |
+
"loss": 1.6244,
|
9837 |
+
"step": 1404
|
9838 |
+
},
|
9839 |
+
{
|
9840 |
+
"epoch": 0.09243877164991694,
|
9841 |
+
"grad_norm": 0.5324863195419312,
|
9842 |
+
"learning_rate": 5.137879829710424e-07,
|
9843 |
+
"loss": 1.579,
|
9844 |
+
"step": 1405
|
9845 |
+
},
|
9846 |
+
{
|
9847 |
+
"epoch": 0.09250456436995247,
|
9848 |
+
"grad_norm": 0.5814157724380493,
|
9849 |
+
"learning_rate": 4.985908338447476e-07,
|
9850 |
+
"loss": 1.6163,
|
9851 |
+
"step": 1406
|
9852 |
+
},
|
9853 |
+
{
|
9854 |
+
"epoch": 0.092570357089988,
|
9855 |
+
"grad_norm": 0.5995772480964661,
|
9856 |
+
"learning_rate": 4.836207009519977e-07,
|
9857 |
+
"loss": 1.657,
|
9858 |
+
"step": 1407
|
9859 |
+
},
|
9860 |
+
{
|
9861 |
+
"epoch": 0.09263614981002352,
|
9862 |
+
"grad_norm": 0.6003164649009705,
|
9863 |
+
"learning_rate": 4.688776529466754e-07,
|
9864 |
+
"loss": 1.6132,
|
9865 |
+
"step": 1408
|
9866 |
+
},
|
9867 |
+
{
|
9868 |
+
"epoch": 0.09270194253005905,
|
9869 |
+
"grad_norm": 0.5804505348205566,
|
9870 |
+
"learning_rate": 4.543617574412184e-07,
|
9871 |
+
"loss": 1.5816,
|
9872 |
+
"step": 1409
|
9873 |
+
},
|
9874 |
+
{
|
9875 |
+
"epoch": 0.09276773525009457,
|
9876 |
+
"grad_norm": 0.5852599143981934,
|
9877 |
+
"learning_rate": 4.4007308100633136e-07,
|
9878 |
+
"loss": 1.6799,
|
9879 |
+
"step": 1410
|
9880 |
+
},
|
9881 |
+
{
|
9882 |
+
"epoch": 0.0928335279701301,
|
9883 |
+
"grad_norm": 0.6366518139839172,
|
9884 |
+
"learning_rate": 4.2601168917069114e-07,
|
9885 |
+
"loss": 1.7649,
|
9886 |
+
"step": 1411
|
9887 |
+
},
|
9888 |
+
{
|
9889 |
+
"epoch": 0.09289932069016563,
|
9890 |
+
"grad_norm": 0.5954082608222961,
|
9891 |
+
"learning_rate": 4.121776464206251e-07,
|
9892 |
+
"loss": 1.6245,
|
9893 |
+
"step": 1412
|
9894 |
+
},
|
9895 |
+
{
|
9896 |
+
"epoch": 0.09296511341020117,
|
9897 |
+
"grad_norm": 0.6634638905525208,
|
9898 |
+
"learning_rate": 3.9857101619982797e-07,
|
9899 |
+
"loss": 1.7316,
|
9900 |
+
"step": 1413
|
9901 |
+
},
|
9902 |
+
{
|
9903 |
+
"epoch": 0.09303090613023669,
|
9904 |
+
"grad_norm": 0.6314533352851868,
|
9905 |
+
"learning_rate": 3.851918609090677e-07,
|
9906 |
+
"loss": 1.6691,
|
9907 |
+
"step": 1414
|
9908 |
+
},
|
9909 |
+
{
|
9910 |
+
"epoch": 0.09309669885027222,
|
9911 |
+
"grad_norm": 0.6321081519126892,
|
9912 |
+
"learning_rate": 3.720402419058966e-07,
|
9913 |
+
"loss": 1.4713,
|
9914 |
+
"step": 1415
|
9915 |
+
},
|
9916 |
+
{
|
9917 |
+
"epoch": 0.09316249157030775,
|
9918 |
+
"grad_norm": 0.6127361059188843,
|
9919 |
+
"learning_rate": 3.5911621950438514e-07,
|
9920 |
+
"loss": 1.5676,
|
9921 |
+
"step": 1416
|
9922 |
+
},
|
9923 |
+
{
|
9924 |
+
"epoch": 0.09322828429034327,
|
9925 |
+
"grad_norm": 0.6360514760017395,
|
9926 |
+
"learning_rate": 3.464198529748108e-07,
|
9927 |
+
"loss": 1.5381,
|
9928 |
+
"step": 1417
|
9929 |
+
},
|
9930 |
+
{
|
9931 |
+
"epoch": 0.0932940770103788,
|
9932 |
+
"grad_norm": 0.6701735258102417,
|
9933 |
+
"learning_rate": 3.339512005434309e-07,
|
9934 |
+
"loss": 1.6244,
|
9935 |
+
"step": 1418
|
9936 |
+
},
|
9937 |
+
{
|
9938 |
+
"epoch": 0.09335986973041432,
|
9939 |
+
"grad_norm": 0.6812663674354553,
|
9940 |
+
"learning_rate": 3.2171031939217666e-07,
|
9941 |
+
"loss": 1.6387,
|
9942 |
+
"step": 1419
|
9943 |
+
},
|
9944 |
+
{
|
9945 |
+
"epoch": 0.09342566245044986,
|
9946 |
+
"grad_norm": 0.6711384057998657,
|
9947 |
+
"learning_rate": 3.0969726565842074e-07,
|
9948 |
+
"loss": 1.5678,
|
9949 |
+
"step": 1420
|
9950 |
+
},
|
9951 |
+
{
|
9952 |
+
"epoch": 0.09349145517048539,
|
9953 |
+
"grad_norm": 0.7010900974273682,
|
9954 |
+
"learning_rate": 2.979120944346936e-07,
|
9955 |
+
"loss": 1.6543,
|
9956 |
+
"step": 1421
|
9957 |
+
},
|
9958 |
+
{
|
9959 |
+
"epoch": 0.09355724789052092,
|
9960 |
+
"grad_norm": 0.6457873582839966,
|
9961 |
+
"learning_rate": 2.863548597684562e-07,
|
9962 |
+
"loss": 1.6793,
|
9963 |
+
"step": 1422
|
9964 |
+
},
|
9965 |
+
{
|
9966 |
+
"epoch": 0.09362304061055644,
|
9967 |
+
"grad_norm": 0.6923664808273315,
|
9968 |
+
"learning_rate": 2.750256146618335e-07,
|
9969 |
+
"loss": 1.5873,
|
9970 |
+
"step": 1423
|
9971 |
+
},
|
9972 |
+
{
|
9973 |
+
"epoch": 0.09368883333059197,
|
9974 |
+
"grad_norm": 0.6824166178703308,
|
9975 |
+
"learning_rate": 2.639244110713701e-07,
|
9976 |
+
"loss": 1.5315,
|
9977 |
+
"step": 1424
|
9978 |
+
},
|
9979 |
+
{
|
9980 |
+
"epoch": 0.0937546260506275,
|
9981 |
+
"grad_norm": 0.6563824415206909,
|
9982 |
+
"learning_rate": 2.5305129990781387e-07,
|
9983 |
+
"loss": 1.604,
|
9984 |
+
"step": 1425
|
9985 |
+
},
|
9986 |
+
{
|
9987 |
+
"epoch": 0.09382041877066302,
|
9988 |
+
"grad_norm": 0.6801635026931763,
|
9989 |
+
"learning_rate": 2.424063310358604e-07,
|
9990 |
+
"loss": 1.5743,
|
9991 |
+
"step": 1426
|
9992 |
+
},
|
9993 |
+
{
|
9994 |
+
"epoch": 0.09388621149069855,
|
9995 |
+
"grad_norm": 0.7070457935333252,
|
9996 |
+
"learning_rate": 2.319895532739369e-07,
|
9997 |
+
"loss": 1.5733,
|
9998 |
+
"step": 1427
|
9999 |
+
},
|
10000 |
+
{
|
10001 |
+
"epoch": 0.09395200421073409,
|
10002 |
+
"grad_norm": 0.661389172077179,
|
10003 |
+
"learning_rate": 2.218010143939575e-07,
|
10004 |
+
"loss": 1.4381,
|
10005 |
+
"step": 1428
|
10006 |
+
},
|
10007 |
+
{
|
10008 |
+
"epoch": 0.09401779693076961,
|
10009 |
+
"grad_norm": 0.7122326493263245,
|
10010 |
+
"learning_rate": 2.1184076112114038e-07,
|
10011 |
+
"loss": 1.5635,
|
10012 |
+
"step": 1429
|
10013 |
+
},
|
10014 |
+
{
|
10015 |
+
"epoch": 0.09408358965080514,
|
10016 |
+
"grad_norm": 0.6946954131126404,
|
10017 |
+
"learning_rate": 2.0210883913376334e-07,
|
10018 |
+
"loss": 1.558,
|
10019 |
+
"step": 1430
|
10020 |
+
},
|
10021 |
+
{
|
10022 |
+
"epoch": 0.09414938237084067,
|
10023 |
+
"grad_norm": 0.7263256311416626,
|
10024 |
+
"learning_rate": 1.9260529306296404e-07,
|
10025 |
+
"loss": 1.6198,
|
10026 |
+
"step": 1431
|
10027 |
+
},
|
10028 |
+
{
|
10029 |
+
"epoch": 0.0942151750908762,
|
10030 |
+
"grad_norm": 0.7192728519439697,
|
10031 |
+
"learning_rate": 1.833301664925402e-07,
|
10032 |
+
"loss": 1.6123,
|
10033 |
+
"step": 1432
|
10034 |
+
},
|
10035 |
+
{
|
10036 |
+
"epoch": 0.09428096781091172,
|
10037 |
+
"grad_norm": 0.6984828114509583,
|
10038 |
+
"learning_rate": 1.742835019587441e-07,
|
10039 |
+
"loss": 1.5176,
|
10040 |
+
"step": 1433
|
10041 |
+
},
|
10042 |
+
{
|
10043 |
+
"epoch": 0.09434676053094725,
|
10044 |
+
"grad_norm": 0.7076666951179504,
|
10045 |
+
"learning_rate": 1.6546534095007172e-07,
|
10046 |
+
"loss": 1.5039,
|
10047 |
+
"step": 1434
|
10048 |
+
},
|
10049 |
+
{
|
10050 |
+
"epoch": 0.09441255325098277,
|
10051 |
+
"grad_norm": 0.7117300629615784,
|
10052 |
+
"learning_rate": 1.5687572390711835e-07,
|
10053 |
+
"loss": 1.5169,
|
10054 |
+
"step": 1435
|
10055 |
+
},
|
10056 |
+
{
|
10057 |
+
"epoch": 0.09447834597101831,
|
10058 |
+
"grad_norm": 0.7957559823989868,
|
10059 |
+
"learning_rate": 1.4851469022234e-07,
|
10060 |
+
"loss": 1.6177,
|
10061 |
+
"step": 1436
|
10062 |
+
},
|
10063 |
+
{
|
10064 |
+
"epoch": 0.09454413869105384,
|
10065 |
+
"grad_norm": 0.7519058585166931,
|
10066 |
+
"learning_rate": 1.403822782399089e-07,
|
10067 |
+
"loss": 1.5329,
|
10068 |
+
"step": 1437
|
10069 |
+
},
|
10070 |
+
{
|
10071 |
+
"epoch": 0.09460993141108937,
|
10072 |
+
"grad_norm": 0.7622172236442566,
|
10073 |
+
"learning_rate": 1.324785252555194e-07,
|
10074 |
+
"loss": 1.3998,
|
10075 |
+
"step": 1438
|
10076 |
+
},
|
10077 |
+
{
|
10078 |
+
"epoch": 0.09467572413112489,
|
10079 |
+
"grad_norm": 0.767694354057312,
|
10080 |
+
"learning_rate": 1.2480346751622686e-07,
|
10081 |
+
"loss": 1.5882,
|
10082 |
+
"step": 1439
|
10083 |
+
},
|
10084 |
+
{
|
10085 |
+
"epoch": 0.09474151685116042,
|
10086 |
+
"grad_norm": 0.7719693183898926,
|
10087 |
+
"learning_rate": 1.1735714022027555e-07,
|
10088 |
+
"loss": 1.3252,
|
10089 |
+
"step": 1440
|
10090 |
+
},
|
10091 |
+
{
|
10092 |
+
"epoch": 0.09480730957119594,
|
10093 |
+
"grad_norm": 0.7426503896713257,
|
10094 |
+
"learning_rate": 1.1013957751693782e-07,
|
10095 |
+
"loss": 1.5377,
|
10096 |
+
"step": 1441
|
10097 |
+
},
|
10098 |
+
{
|
10099 |
+
"epoch": 0.09487310229123147,
|
10100 |
+
"grad_norm": 0.7355154156684875,
|
10101 |
+
"learning_rate": 1.0315081250636405e-07,
|
10102 |
+
"loss": 1.424,
|
10103 |
+
"step": 1442
|
10104 |
+
},
|
10105 |
+
{
|
10106 |
+
"epoch": 0.094938895011267,
|
10107 |
+
"grad_norm": 0.8036089539527893,
|
10108 |
+
"learning_rate": 9.63908772394162e-08,
|
10109 |
+
"loss": 1.605,
|
10110 |
+
"step": 1443
|
10111 |
+
},
|
10112 |
+
{
|
10113 |
+
"epoch": 0.09500468773130254,
|
10114 |
+
"grad_norm": 0.878667950630188,
|
10115 |
+
"learning_rate": 8.985980271754013e-08,
|
10116 |
+
"loss": 1.5414,
|
10117 |
+
"step": 1444
|
10118 |
+
},
|
10119 |
+
{
|
10120 |
+
"epoch": 0.09507048045133806,
|
10121 |
+
"grad_norm": 0.9153657555580139,
|
10122 |
+
"learning_rate": 8.355761889260461e-08,
|
10123 |
+
"loss": 1.5492,
|
10124 |
+
"step": 1445
|
10125 |
+
},
|
10126 |
+
{
|
10127 |
+
"epoch": 0.09513627317137359,
|
10128 |
+
"grad_norm": 0.8688125014305115,
|
10129 |
+
"learning_rate": 7.748435466678471e-08,
|
10130 |
+
"loss": 1.525,
|
10131 |
+
"step": 1446
|
10132 |
+
},
|
10133 |
+
{
|
10134 |
+
"epoch": 0.09520206589140912,
|
10135 |
+
"grad_norm": 0.9009868502616882,
|
10136 |
+
"learning_rate": 7.164003789240648e-08,
|
10137 |
+
"loss": 1.4909,
|
10138 |
+
"step": 1447
|
10139 |
+
},
|
10140 |
+
{
|
10141 |
+
"epoch": 0.09526785861144464,
|
10142 |
+
"grad_norm": 0.947225034236908,
|
10143 |
+
"learning_rate": 6.602469537183021e-08,
|
10144 |
+
"loss": 1.4103,
|
10145 |
+
"step": 1448
|
10146 |
+
},
|
10147 |
+
{
|
10148 |
+
"epoch": 0.09533365133148017,
|
10149 |
+
"grad_norm": 1.0838871002197266,
|
10150 |
+
"learning_rate": 6.063835285733955e-08,
|
10151 |
+
"loss": 1.5726,
|
10152 |
+
"step": 1449
|
10153 |
+
},
|
10154 |
+
{
|
10155 |
+
"epoch": 0.0953994440515157,
|
10156 |
+
"grad_norm": 1.4739456176757812,
|
10157 |
+
"learning_rate": 5.5481035050991556e-08,
|
10158 |
+
"loss": 1.0046,
|
10159 |
+
"step": 1450
|
10160 |
+
},
|
10161 |
+
{
|
10162 |
+
"epoch": 0.09546523677155123,
|
10163 |
+
"grad_norm": 0.5145021677017212,
|
10164 |
+
"learning_rate": 5.0552765604544584e-08,
|
10165 |
+
"loss": 1.7385,
|
10166 |
+
"step": 1451
|
10167 |
+
},
|
10168 |
+
{
|
10169 |
+
"epoch": 0.09553102949158676,
|
10170 |
+
"grad_norm": 0.5236654877662659,
|
10171 |
+
"learning_rate": 4.585356711931388e-08,
|
10172 |
+
"loss": 1.7165,
|
10173 |
+
"step": 1452
|
10174 |
+
},
|
10175 |
+
{
|
10176 |
+
"epoch": 0.09559682221162229,
|
10177 |
+
"grad_norm": 0.5186619758605957,
|
10178 |
+
"learning_rate": 4.138346114608283e-08,
|
10179 |
+
"loss": 1.4829,
|
10180 |
+
"step": 1453
|
10181 |
+
},
|
10182 |
+
{
|
10183 |
+
"epoch": 0.09566261493165781,
|
10184 |
+
"grad_norm": 0.5814453959465027,
|
10185 |
+
"learning_rate": 3.7142468185014104e-08,
|
10186 |
+
"loss": 1.6926,
|
10187 |
+
"step": 1454
|
10188 |
+
},
|
10189 |
+
{
|
10190 |
+
"epoch": 0.09572840765169334,
|
10191 |
+
"grad_norm": 0.5442173480987549,
|
10192 |
+
"learning_rate": 3.313060768553866e-08,
|
10193 |
+
"loss": 1.6347,
|
10194 |
+
"step": 1455
|
10195 |
+
},
|
10196 |
+
{
|
10197 |
+
"epoch": 0.09579420037172887,
|
10198 |
+
"grad_norm": 0.5665310025215149,
|
10199 |
+
"learning_rate": 2.9347898046266918e-08,
|
10200 |
+
"loss": 1.5986,
|
10201 |
+
"step": 1456
|
10202 |
+
},
|
10203 |
+
{
|
10204 |
+
"epoch": 0.09585999309176439,
|
10205 |
+
"grad_norm": 0.5988601446151733,
|
10206 |
+
"learning_rate": 2.5794356614922134e-08,
|
10207 |
+
"loss": 1.8659,
|
10208 |
+
"step": 1457
|
10209 |
+
},
|
10210 |
+
{
|
10211 |
+
"epoch": 0.09592578581179992,
|
10212 |
+
"grad_norm": 0.5948224067687988,
|
10213 |
+
"learning_rate": 2.2469999688246035e-08,
|
10214 |
+
"loss": 1.7225,
|
10215 |
+
"step": 1458
|
10216 |
+
},
|
10217 |
+
{
|
10218 |
+
"epoch": 0.09599157853183546,
|
10219 |
+
"grad_norm": 0.6055050492286682,
|
10220 |
+
"learning_rate": 1.937484251192112e-08,
|
10221 |
+
"loss": 1.6482,
|
10222 |
+
"step": 1459
|
10223 |
+
},
|
10224 |
+
{
|
10225 |
+
"epoch": 0.09605737125187098,
|
10226 |
+
"grad_norm": 0.6156303286552429,
|
10227 |
+
"learning_rate": 1.6508899280515134e-08,
|
10228 |
+
"loss": 1.76,
|
10229 |
+
"step": 1460
|
10230 |
+
},
|
10231 |
+
{
|
10232 |
+
"epoch": 0.09612316397190651,
|
10233 |
+
"grad_norm": 0.5953572988510132,
|
10234 |
+
"learning_rate": 1.3872183137397799e-08,
|
10235 |
+
"loss": 1.6664,
|
10236 |
+
"step": 1461
|
10237 |
+
},
|
10238 |
+
{
|
10239 |
+
"epoch": 0.09618895669194204,
|
10240 |
+
"grad_norm": 0.6339824199676514,
|
10241 |
+
"learning_rate": 1.1464706174701967e-08,
|
10242 |
+
"loss": 1.5749,
|
10243 |
+
"step": 1462
|
10244 |
+
},
|
10245 |
+
{
|
10246 |
+
"epoch": 0.09625474941197756,
|
10247 |
+
"grad_norm": 0.598288357257843,
|
10248 |
+
"learning_rate": 9.286479433257e-09,
|
10249 |
+
"loss": 1.5851,
|
10250 |
+
"step": 1463
|
10251 |
+
},
|
10252 |
+
{
|
10253 |
+
"epoch": 0.09632054213201309,
|
10254 |
+
"grad_norm": 0.6078794002532959,
|
10255 |
+
"learning_rate": 7.337512902522159e-09,
|
10256 |
+
"loss": 1.6656,
|
10257 |
+
"step": 1464
|
10258 |
+
},
|
10259 |
+
{
|
10260 |
+
"epoch": 0.09638633485204862,
|
10261 |
+
"grad_norm": 0.6280359029769897,
|
10262 |
+
"learning_rate": 5.6178155205754975e-09,
|
10263 |
+
"loss": 1.6216,
|
10264 |
+
"step": 1465
|
10265 |
+
},
|
10266 |
+
{
|
10267 |
+
"epoch": 0.09645212757208414,
|
10268 |
+
"grad_norm": 0.5920078754425049,
|
10269 |
+
"learning_rate": 4.127395174036153e-09,
|
10270 |
+
"loss": 1.5143,
|
10271 |
+
"step": 1466
|
10272 |
+
},
|
10273 |
+
{
|
10274 |
+
"epoch": 0.09651792029211968,
|
10275 |
+
"grad_norm": 0.6147146224975586,
|
10276 |
+
"learning_rate": 2.866258698064339e-09,
|
10277 |
+
"loss": 1.559,
|
10278 |
+
"step": 1467
|
10279 |
+
},
|
10280 |
+
{
|
10281 |
+
"epoch": 0.09658371301215521,
|
10282 |
+
"grad_norm": 0.6601701974868774,
|
10283 |
+
"learning_rate": 1.8344118763002903e-09,
|
10284 |
+
"loss": 1.593,
|
10285 |
+
"step": 1468
|
10286 |
+
},
|
10287 |
+
{
|
10288 |
+
"epoch": 0.09664950573219074,
|
10289 |
+
"grad_norm": 0.6434551477432251,
|
10290 |
+
"learning_rate": 1.0318594408476045e-09,
|
10291 |
+
"loss": 1.5645,
|
10292 |
+
"step": 1469
|
10293 |
+
},
|
10294 |
+
{
|
10295 |
+
"epoch": 0.09671529845222626,
|
10296 |
+
"grad_norm": 0.6171491146087646,
|
10297 |
+
"learning_rate": 4.586050722621416e-10,
|
10298 |
+
"loss": 1.589,
|
10299 |
+
"step": 1470
|
10300 |
+
},
|
10301 |
+
{
|
10302 |
+
"epoch": 0.09678109117226179,
|
10303 |
+
"grad_norm": 0.6035428047180176,
|
10304 |
+
"learning_rate": 1.1465139951316595e-10,
|
10305 |
+
"loss": 1.5367,
|
10306 |
+
"step": 1471
|
10307 |
+
},
|
10308 |
+
{
|
10309 |
+
"epoch": 0.09684688389229731,
|
10310 |
+
"grad_norm": 0.6608728170394897,
|
10311 |
+
"learning_rate": 0.0,
|
10312 |
+
"loss": 1.517,
|
10313 |
+
"step": 1472
|
10314 |
}
|
10315 |
],
|
10316 |
"logging_steps": 1,
|
|
|
10325 |
"should_evaluate": false,
|
10326 |
"should_log": false,
|
10327 |
"should_save": true,
|
10328 |
+
"should_training_stop": true
|
10329 |
},
|
10330 |
"attributes": {}
|
10331 |
}
|
10332 |
},
|
10333 |
+
"total_flos": 1.9072999168294257e+18,
|
10334 |
"train_batch_size": 4,
|
10335 |
"trial_name": null,
|
10336 |
"trial_params": null
|