Training in progress, step 779, 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 114106856
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f7e94715efa8923f83c90e8a4b164c445876d355075323cb068661e7c275ac8a
|
| 3 |
size 114106856
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 58562836
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1dcb761caed78bf48b94ced600453a1e977fa69914ed3c89ca938afcc739adab
|
| 3 |
size 58562836
|
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:f152be734e491ea28e2f66d9760c5b92c8c1c304987e832953ad75be8eaa1da9
|
| 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 sha256:63400ae03f6cd4101f3500643a1b7255f746b2b1cc2a30342cfacd4858d6273c
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -4102,6 +4102,1372 @@
|
|
| 4102 |
"learning_rate": 1.47164856854565e-05,
|
| 4103 |
"loss": 3.4533,
|
| 4104 |
"step": 585
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 4105 |
}
|
| 4106 |
],
|
| 4107 |
"logging_steps": 1,
|
|
@@ -4116,12 +5482,12 @@
|
|
| 4116 |
"should_evaluate": false,
|
| 4117 |
"should_log": false,
|
| 4118 |
"should_save": true,
|
| 4119 |
-
"should_training_stop":
|
| 4120 |
},
|
| 4121 |
"attributes": {}
|
| 4122 |
}
|
| 4123 |
},
|
| 4124 |
-
"total_flos":
|
| 4125 |
"train_batch_size": 4,
|
| 4126 |
"trial_name": null,
|
| 4127 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 1.0009630818619584,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 779,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 4102 |
"learning_rate": 1.47164856854565e-05,
|
| 4103 |
"loss": 3.4533,
|
| 4104 |
"step": 585
|
| 4105 |
+
},
|
| 4106 |
+
{
|
| 4107 |
+
"epoch": 0.7524879614767255,
|
| 4108 |
+
"grad_norm": 6.916558742523193,
|
| 4109 |
+
"learning_rate": 1.4572981813093507e-05,
|
| 4110 |
+
"loss": 3.6141,
|
| 4111 |
+
"step": 586
|
| 4112 |
+
},
|
| 4113 |
+
{
|
| 4114 |
+
"epoch": 0.7537720706260033,
|
| 4115 |
+
"grad_norm": 5.969259738922119,
|
| 4116 |
+
"learning_rate": 1.4430061589841121e-05,
|
| 4117 |
+
"loss": 2.0467,
|
| 4118 |
+
"step": 587
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 0.755056179775281,
|
| 4122 |
+
"grad_norm": 5.210697174072266,
|
| 4123 |
+
"learning_rate": 1.4287727370265558e-05,
|
| 4124 |
+
"loss": 1.8851,
|
| 4125 |
+
"step": 588
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 0.7563402889245586,
|
| 4129 |
+
"grad_norm": 7.276077747344971,
|
| 4130 |
+
"learning_rate": 1.4145981499278876e-05,
|
| 4131 |
+
"loss": 3.3363,
|
| 4132 |
+
"step": 589
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 0.7576243980738363,
|
| 4136 |
+
"grad_norm": 9.250455856323242,
|
| 4137 |
+
"learning_rate": 1.4004826312100216e-05,
|
| 4138 |
+
"loss": 4.4776,
|
| 4139 |
+
"step": 590
|
| 4140 |
+
},
|
| 4141 |
+
{
|
| 4142 |
+
"epoch": 0.758908507223114,
|
| 4143 |
+
"grad_norm": 6.841946125030518,
|
| 4144 |
+
"learning_rate": 1.386426413421738e-05,
|
| 4145 |
+
"loss": 2.9102,
|
| 4146 |
+
"step": 591
|
| 4147 |
+
},
|
| 4148 |
+
{
|
| 4149 |
+
"epoch": 0.7601926163723917,
|
| 4150 |
+
"grad_norm": 6.715099334716797,
|
| 4151 |
+
"learning_rate": 1.3724297281348592e-05,
|
| 4152 |
+
"loss": 2.9798,
|
| 4153 |
+
"step": 592
|
| 4154 |
+
},
|
| 4155 |
+
{
|
| 4156 |
+
"epoch": 0.7614767255216693,
|
| 4157 |
+
"grad_norm": 7.211644649505615,
|
| 4158 |
+
"learning_rate": 1.3584928059404205e-05,
|
| 4159 |
+
"loss": 3.9018,
|
| 4160 |
+
"step": 593
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 0.762760834670947,
|
| 4164 |
+
"grad_norm": 6.415737628936768,
|
| 4165 |
+
"learning_rate": 1.3446158764448841e-05,
|
| 4166 |
+
"loss": 2.4854,
|
| 4167 |
+
"step": 594
|
| 4168 |
+
},
|
| 4169 |
+
{
|
| 4170 |
+
"epoch": 0.7640449438202247,
|
| 4171 |
+
"grad_norm": 8.730350494384766,
|
| 4172 |
+
"learning_rate": 1.3307991682663462e-05,
|
| 4173 |
+
"loss": 3.7137,
|
| 4174 |
+
"step": 595
|
| 4175 |
+
},
|
| 4176 |
+
{
|
| 4177 |
+
"epoch": 0.7653290529695024,
|
| 4178 |
+
"grad_norm": 10.936077117919922,
|
| 4179 |
+
"learning_rate": 1.3170429090307823e-05,
|
| 4180 |
+
"loss": 4.2927,
|
| 4181 |
+
"step": 596
|
| 4182 |
+
},
|
| 4183 |
+
{
|
| 4184 |
+
"epoch": 0.7666131621187801,
|
| 4185 |
+
"grad_norm": 10.573296546936035,
|
| 4186 |
+
"learning_rate": 1.3033473253682848e-05,
|
| 4187 |
+
"loss": 4.0148,
|
| 4188 |
+
"step": 597
|
| 4189 |
+
},
|
| 4190 |
+
{
|
| 4191 |
+
"epoch": 0.7678972712680577,
|
| 4192 |
+
"grad_norm": 7.918160915374756,
|
| 4193 |
+
"learning_rate": 1.2897126429093354e-05,
|
| 4194 |
+
"loss": 3.0252,
|
| 4195 |
+
"step": 598
|
| 4196 |
+
},
|
| 4197 |
+
{
|
| 4198 |
+
"epoch": 0.7691813804173355,
|
| 4199 |
+
"grad_norm": 8.217205047607422,
|
| 4200 |
+
"learning_rate": 1.2761390862810907e-05,
|
| 4201 |
+
"loss": 3.4848,
|
| 4202 |
+
"step": 599
|
| 4203 |
+
},
|
| 4204 |
+
{
|
| 4205 |
+
"epoch": 0.7704654895666132,
|
| 4206 |
+
"grad_norm": 9.806517601013184,
|
| 4207 |
+
"learning_rate": 1.2626268791036767e-05,
|
| 4208 |
+
"loss": 4.4466,
|
| 4209 |
+
"step": 600
|
| 4210 |
+
},
|
| 4211 |
+
{
|
| 4212 |
+
"epoch": 0.7717495987158909,
|
| 4213 |
+
"grad_norm": 11.450250625610352,
|
| 4214 |
+
"learning_rate": 1.2491762439865035e-05,
|
| 4215 |
+
"loss": 4.424,
|
| 4216 |
+
"step": 601
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 0.7730337078651686,
|
| 4220 |
+
"grad_norm": 9.043597221374512,
|
| 4221 |
+
"learning_rate": 1.235787402524603e-05,
|
| 4222 |
+
"loss": 4.3972,
|
| 4223 |
+
"step": 602
|
| 4224 |
+
},
|
| 4225 |
+
{
|
| 4226 |
+
"epoch": 0.7743178170144462,
|
| 4227 |
+
"grad_norm": 7.3961005210876465,
|
| 4228 |
+
"learning_rate": 1.2224605752949786e-05,
|
| 4229 |
+
"loss": 3.427,
|
| 4230 |
+
"step": 603
|
| 4231 |
+
},
|
| 4232 |
+
{
|
| 4233 |
+
"epoch": 0.7756019261637239,
|
| 4234 |
+
"grad_norm": 6.560033798217773,
|
| 4235 |
+
"learning_rate": 1.2091959818529636e-05,
|
| 4236 |
+
"loss": 2.9121,
|
| 4237 |
+
"step": 604
|
| 4238 |
+
},
|
| 4239 |
+
{
|
| 4240 |
+
"epoch": 0.7768860353130016,
|
| 4241 |
+
"grad_norm": 6.344895839691162,
|
| 4242 |
+
"learning_rate": 1.1959938407286097e-05,
|
| 4243 |
+
"loss": 2.7389,
|
| 4244 |
+
"step": 605
|
| 4245 |
+
},
|
| 4246 |
+
{
|
| 4247 |
+
"epoch": 0.7781701444622793,
|
| 4248 |
+
"grad_norm": 6.930209159851074,
|
| 4249 |
+
"learning_rate": 1.1828543694230909e-05,
|
| 4250 |
+
"loss": 2.8555,
|
| 4251 |
+
"step": 606
|
| 4252 |
+
},
|
| 4253 |
+
{
|
| 4254 |
+
"epoch": 0.779454253611557,
|
| 4255 |
+
"grad_norm": 8.534188270568848,
|
| 4256 |
+
"learning_rate": 1.1697777844051105e-05,
|
| 4257 |
+
"loss": 3.666,
|
| 4258 |
+
"step": 607
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 0.7807383627608346,
|
| 4262 |
+
"grad_norm": 7.170388698577881,
|
| 4263 |
+
"learning_rate": 1.1567643011073392e-05,
|
| 4264 |
+
"loss": 3.4359,
|
| 4265 |
+
"step": 608
|
| 4266 |
+
},
|
| 4267 |
+
{
|
| 4268 |
+
"epoch": 0.7820224719101123,
|
| 4269 |
+
"grad_norm": 9.166104316711426,
|
| 4270 |
+
"learning_rate": 1.143814133922872e-05,
|
| 4271 |
+
"loss": 4.8905,
|
| 4272 |
+
"step": 609
|
| 4273 |
+
},
|
| 4274 |
+
{
|
| 4275 |
+
"epoch": 0.78330658105939,
|
| 4276 |
+
"grad_norm": 6.033200740814209,
|
| 4277 |
+
"learning_rate": 1.1309274962016852e-05,
|
| 4278 |
+
"loss": 2.4617,
|
| 4279 |
+
"step": 610
|
| 4280 |
+
},
|
| 4281 |
+
{
|
| 4282 |
+
"epoch": 0.7845906902086678,
|
| 4283 |
+
"grad_norm": 6.07381534576416,
|
| 4284 |
+
"learning_rate": 1.118104600247129e-05,
|
| 4285 |
+
"loss": 2.538,
|
| 4286 |
+
"step": 611
|
| 4287 |
+
},
|
| 4288 |
+
{
|
| 4289 |
+
"epoch": 0.7858747993579455,
|
| 4290 |
+
"grad_norm": 5.244448184967041,
|
| 4291 |
+
"learning_rate": 1.105345657312427e-05,
|
| 4292 |
+
"loss": 2.54,
|
| 4293 |
+
"step": 612
|
| 4294 |
+
},
|
| 4295 |
+
{
|
| 4296 |
+
"epoch": 0.7871589085072231,
|
| 4297 |
+
"grad_norm": 5.321899890899658,
|
| 4298 |
+
"learning_rate": 1.0926508775971994e-05,
|
| 4299 |
+
"loss": 2.1132,
|
| 4300 |
+
"step": 613
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 0.7884430176565008,
|
| 4304 |
+
"grad_norm": 6.152393817901611,
|
| 4305 |
+
"learning_rate": 1.0800204702439937e-05,
|
| 4306 |
+
"loss": 2.3172,
|
| 4307 |
+
"step": 614
|
| 4308 |
+
},
|
| 4309 |
+
{
|
| 4310 |
+
"epoch": 0.7897271268057785,
|
| 4311 |
+
"grad_norm": 6.6167755126953125,
|
| 4312 |
+
"learning_rate": 1.0674546433348454e-05,
|
| 4313 |
+
"loss": 3.0088,
|
| 4314 |
+
"step": 615
|
| 4315 |
+
},
|
| 4316 |
+
{
|
| 4317 |
+
"epoch": 0.7910112359550562,
|
| 4318 |
+
"grad_norm": 6.641218185424805,
|
| 4319 |
+
"learning_rate": 1.0549536038878432e-05,
|
| 4320 |
+
"loss": 2.767,
|
| 4321 |
+
"step": 616
|
| 4322 |
+
},
|
| 4323 |
+
{
|
| 4324 |
+
"epoch": 0.7922953451043339,
|
| 4325 |
+
"grad_norm": 6.559841632843018,
|
| 4326 |
+
"learning_rate": 1.0425175578537299e-05,
|
| 4327 |
+
"loss": 2.868,
|
| 4328 |
+
"step": 617
|
| 4329 |
+
},
|
| 4330 |
+
{
|
| 4331 |
+
"epoch": 0.7935794542536115,
|
| 4332 |
+
"grad_norm": 5.474334716796875,
|
| 4333 |
+
"learning_rate": 1.0301467101124957e-05,
|
| 4334 |
+
"loss": 2.4926,
|
| 4335 |
+
"step": 618
|
| 4336 |
+
},
|
| 4337 |
+
{
|
| 4338 |
+
"epoch": 0.7948635634028892,
|
| 4339 |
+
"grad_norm": 6.981647491455078,
|
| 4340 |
+
"learning_rate": 1.0178412644700092e-05,
|
| 4341 |
+
"loss": 2.4924,
|
| 4342 |
+
"step": 619
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 0.7961476725521669,
|
| 4346 |
+
"grad_norm": 7.217685222625732,
|
| 4347 |
+
"learning_rate": 1.0056014236546646e-05,
|
| 4348 |
+
"loss": 2.7151,
|
| 4349 |
+
"step": 620
|
| 4350 |
+
},
|
| 4351 |
+
{
|
| 4352 |
+
"epoch": 0.7974317817014446,
|
| 4353 |
+
"grad_norm": 6.473022937774658,
|
| 4354 |
+
"learning_rate": 9.934273893140334e-06,
|
| 4355 |
+
"loss": 2.993,
|
| 4356 |
+
"step": 621
|
| 4357 |
+
},
|
| 4358 |
+
{
|
| 4359 |
+
"epoch": 0.7987158908507224,
|
| 4360 |
+
"grad_norm": 7.541585922241211,
|
| 4361 |
+
"learning_rate": 9.813193620115447e-06,
|
| 4362 |
+
"loss": 3.0707,
|
| 4363 |
+
"step": 622
|
| 4364 |
+
},
|
| 4365 |
+
{
|
| 4366 |
+
"epoch": 0.8,
|
| 4367 |
+
"grad_norm": 7.679656982421875,
|
| 4368 |
+
"learning_rate": 9.692775412231863e-06,
|
| 4369 |
+
"loss": 2.8438,
|
| 4370 |
+
"step": 623
|
| 4371 |
+
},
|
| 4372 |
+
{
|
| 4373 |
+
"epoch": 0.8012841091492777,
|
| 4374 |
+
"grad_norm": 8.271646499633789,
|
| 4375 |
+
"learning_rate": 9.573021253342112e-06,
|
| 4376 |
+
"loss": 3.336,
|
| 4377 |
+
"step": 624
|
| 4378 |
+
},
|
| 4379 |
+
{
|
| 4380 |
+
"epoch": 0.8025682182985554,
|
| 4381 |
+
"grad_norm": 6.968911170959473,
|
| 4382 |
+
"learning_rate": 9.453933116358715e-06,
|
| 4383 |
+
"loss": 3.3099,
|
| 4384 |
+
"step": 625
|
| 4385 |
+
},
|
| 4386 |
+
{
|
| 4387 |
+
"epoch": 0.8038523274478331,
|
| 4388 |
+
"grad_norm": 8.06809139251709,
|
| 4389 |
+
"learning_rate": 9.335512963221732e-06,
|
| 4390 |
+
"loss": 3.7685,
|
| 4391 |
+
"step": 626
|
| 4392 |
+
},
|
| 4393 |
+
{
|
| 4394 |
+
"epoch": 0.8051364365971108,
|
| 4395 |
+
"grad_norm": 5.143070697784424,
|
| 4396 |
+
"learning_rate": 9.21776274486636e-06,
|
| 4397 |
+
"loss": 2.0179,
|
| 4398 |
+
"step": 627
|
| 4399 |
+
},
|
| 4400 |
+
{
|
| 4401 |
+
"epoch": 0.8064205457463884,
|
| 4402 |
+
"grad_norm": 9.07247543334961,
|
| 4403 |
+
"learning_rate": 9.100684401190828e-06,
|
| 4404 |
+
"loss": 4.0971,
|
| 4405 |
+
"step": 628
|
| 4406 |
+
},
|
| 4407 |
+
{
|
| 4408 |
+
"epoch": 0.8077046548956661,
|
| 4409 |
+
"grad_norm": 7.609170436859131,
|
| 4410 |
+
"learning_rate": 8.984279861024453e-06,
|
| 4411 |
+
"loss": 3.4357,
|
| 4412 |
+
"step": 629
|
| 4413 |
+
},
|
| 4414 |
+
{
|
| 4415 |
+
"epoch": 0.8089887640449438,
|
| 4416 |
+
"grad_norm": 6.985535144805908,
|
| 4417 |
+
"learning_rate": 8.868551042095851e-06,
|
| 4418 |
+
"loss": 3.1968,
|
| 4419 |
+
"step": 630
|
| 4420 |
+
},
|
| 4421 |
+
{
|
| 4422 |
+
"epoch": 0.8102728731942215,
|
| 4423 |
+
"grad_norm": 7.812018394470215,
|
| 4424 |
+
"learning_rate": 8.75349985100134e-06,
|
| 4425 |
+
"loss": 3.5456,
|
| 4426 |
+
"step": 631
|
| 4427 |
+
},
|
| 4428 |
+
{
|
| 4429 |
+
"epoch": 0.8115569823434992,
|
| 4430 |
+
"grad_norm": 8.600101470947266,
|
| 4431 |
+
"learning_rate": 8.639128183173518e-06,
|
| 4432 |
+
"loss": 3.4957,
|
| 4433 |
+
"step": 632
|
| 4434 |
+
},
|
| 4435 |
+
{
|
| 4436 |
+
"epoch": 0.8128410914927768,
|
| 4437 |
+
"grad_norm": 8.320466995239258,
|
| 4438 |
+
"learning_rate": 8.525437922850032e-06,
|
| 4439 |
+
"loss": 2.7815,
|
| 4440 |
+
"step": 633
|
| 4441 |
+
},
|
| 4442 |
+
{
|
| 4443 |
+
"epoch": 0.8141252006420546,
|
| 4444 |
+
"grad_norm": 7.145605564117432,
|
| 4445 |
+
"learning_rate": 8.412430943042615e-06,
|
| 4446 |
+
"loss": 3.4794,
|
| 4447 |
+
"step": 634
|
| 4448 |
+
},
|
| 4449 |
+
{
|
| 4450 |
+
"epoch": 0.8154093097913323,
|
| 4451 |
+
"grad_norm": 5.847816467285156,
|
| 4452 |
+
"learning_rate": 8.30010910550611e-06,
|
| 4453 |
+
"loss": 2.7875,
|
| 4454 |
+
"step": 635
|
| 4455 |
+
},
|
| 4456 |
+
{
|
| 4457 |
+
"epoch": 0.81669341894061,
|
| 4458 |
+
"grad_norm": 7.271595001220703,
|
| 4459 |
+
"learning_rate": 8.188474260707858e-06,
|
| 4460 |
+
"loss": 3.5987,
|
| 4461 |
+
"step": 636
|
| 4462 |
+
},
|
| 4463 |
+
{
|
| 4464 |
+
"epoch": 0.8179775280898877,
|
| 4465 |
+
"grad_norm": 8.630016326904297,
|
| 4466 |
+
"learning_rate": 8.077528247797234e-06,
|
| 4467 |
+
"loss": 3.823,
|
| 4468 |
+
"step": 637
|
| 4469 |
+
},
|
| 4470 |
+
{
|
| 4471 |
+
"epoch": 0.8192616372391653,
|
| 4472 |
+
"grad_norm": 9.090782165527344,
|
| 4473 |
+
"learning_rate": 7.967272894575312e-06,
|
| 4474 |
+
"loss": 4.3018,
|
| 4475 |
+
"step": 638
|
| 4476 |
+
},
|
| 4477 |
+
{
|
| 4478 |
+
"epoch": 0.820545746388443,
|
| 4479 |
+
"grad_norm": 9.1552152633667,
|
| 4480 |
+
"learning_rate": 7.857710017464737e-06,
|
| 4481 |
+
"loss": 4.3033,
|
| 4482 |
+
"step": 639
|
| 4483 |
+
},
|
| 4484 |
+
{
|
| 4485 |
+
"epoch": 0.8218298555377207,
|
| 4486 |
+
"grad_norm": 8.0325288772583,
|
| 4487 |
+
"learning_rate": 7.748841421479874e-06,
|
| 4488 |
+
"loss": 3.1759,
|
| 4489 |
+
"step": 640
|
| 4490 |
+
},
|
| 4491 |
+
{
|
| 4492 |
+
"epoch": 0.8231139646869984,
|
| 4493 |
+
"grad_norm": 11.123332023620605,
|
| 4494 |
+
"learning_rate": 7.640668900196984e-06,
|
| 4495 |
+
"loss": 4.3347,
|
| 4496 |
+
"step": 641
|
| 4497 |
+
},
|
| 4498 |
+
{
|
| 4499 |
+
"epoch": 0.8243980738362761,
|
| 4500 |
+
"grad_norm": 6.174449920654297,
|
| 4501 |
+
"learning_rate": 7.533194235724728e-06,
|
| 4502 |
+
"loss": 2.8528,
|
| 4503 |
+
"step": 642
|
| 4504 |
+
},
|
| 4505 |
+
{
|
| 4506 |
+
"epoch": 0.8256821829855537,
|
| 4507 |
+
"grad_norm": 7.3835554122924805,
|
| 4508 |
+
"learning_rate": 7.426419198674772e-06,
|
| 4509 |
+
"loss": 3.7275,
|
| 4510 |
+
"step": 643
|
| 4511 |
+
},
|
| 4512 |
+
{
|
| 4513 |
+
"epoch": 0.8269662921348314,
|
| 4514 |
+
"grad_norm": 7.971776962280273,
|
| 4515 |
+
"learning_rate": 7.320345548132679e-06,
|
| 4516 |
+
"loss": 3.6571,
|
| 4517 |
+
"step": 644
|
| 4518 |
+
},
|
| 4519 |
+
{
|
| 4520 |
+
"epoch": 0.8282504012841091,
|
| 4521 |
+
"grad_norm": 7.133792400360107,
|
| 4522 |
+
"learning_rate": 7.214975031628857e-06,
|
| 4523 |
+
"loss": 3.1931,
|
| 4524 |
+
"step": 645
|
| 4525 |
+
},
|
| 4526 |
+
{
|
| 4527 |
+
"epoch": 0.8295345104333869,
|
| 4528 |
+
"grad_norm": 6.824871063232422,
|
| 4529 |
+
"learning_rate": 7.110309385109803e-06,
|
| 4530 |
+
"loss": 2.9311,
|
| 4531 |
+
"step": 646
|
| 4532 |
+
},
|
| 4533 |
+
{
|
| 4534 |
+
"epoch": 0.8308186195826646,
|
| 4535 |
+
"grad_norm": 9.289607048034668,
|
| 4536 |
+
"learning_rate": 7.006350332909495e-06,
|
| 4537 |
+
"loss": 4.1846,
|
| 4538 |
+
"step": 647
|
| 4539 |
+
},
|
| 4540 |
+
{
|
| 4541 |
+
"epoch": 0.8321027287319422,
|
| 4542 |
+
"grad_norm": 7.833906173706055,
|
| 4543 |
+
"learning_rate": 6.9030995877210236e-06,
|
| 4544 |
+
"loss": 4.1159,
|
| 4545 |
+
"step": 648
|
| 4546 |
+
},
|
| 4547 |
+
{
|
| 4548 |
+
"epoch": 0.8333868378812199,
|
| 4549 |
+
"grad_norm": 8.37000846862793,
|
| 4550 |
+
"learning_rate": 6.800558850568295e-06,
|
| 4551 |
+
"loss": 4.2116,
|
| 4552 |
+
"step": 649
|
| 4553 |
+
},
|
| 4554 |
+
{
|
| 4555 |
+
"epoch": 0.8346709470304976,
|
| 4556 |
+
"grad_norm": 10.978693008422852,
|
| 4557 |
+
"learning_rate": 6.698729810778065e-06,
|
| 4558 |
+
"loss": 3.4674,
|
| 4559 |
+
"step": 650
|
| 4560 |
+
},
|
| 4561 |
+
{
|
| 4562 |
+
"epoch": 0.8359550561797753,
|
| 4563 |
+
"grad_norm": 7.171913146972656,
|
| 4564 |
+
"learning_rate": 6.5976141459521355e-06,
|
| 4565 |
+
"loss": 3.5828,
|
| 4566 |
+
"step": 651
|
| 4567 |
+
},
|
| 4568 |
+
{
|
| 4569 |
+
"epoch": 0.837239165329053,
|
| 4570 |
+
"grad_norm": 41.70615005493164,
|
| 4571 |
+
"learning_rate": 6.497213521939638e-06,
|
| 4572 |
+
"loss": 5.1122,
|
| 4573 |
+
"step": 652
|
| 4574 |
+
},
|
| 4575 |
+
{
|
| 4576 |
+
"epoch": 0.8385232744783306,
|
| 4577 |
+
"grad_norm": 13.99992847442627,
|
| 4578 |
+
"learning_rate": 6.397529592809614e-06,
|
| 4579 |
+
"loss": 4.6054,
|
| 4580 |
+
"step": 653
|
| 4581 |
+
},
|
| 4582 |
+
{
|
| 4583 |
+
"epoch": 0.8398073836276083,
|
| 4584 |
+
"grad_norm": 8.354331016540527,
|
| 4585 |
+
"learning_rate": 6.298564000823848e-06,
|
| 4586 |
+
"loss": 3.5946,
|
| 4587 |
+
"step": 654
|
| 4588 |
+
},
|
| 4589 |
+
{
|
| 4590 |
+
"epoch": 0.841091492776886,
|
| 4591 |
+
"grad_norm": 7.070506572723389,
|
| 4592 |
+
"learning_rate": 6.2003183764096695e-06,
|
| 4593 |
+
"loss": 3.7046,
|
| 4594 |
+
"step": 655
|
| 4595 |
+
},
|
| 4596 |
+
{
|
| 4597 |
+
"epoch": 0.8423756019261637,
|
| 4598 |
+
"grad_norm": 5.7081098556518555,
|
| 4599 |
+
"learning_rate": 6.102794338133194e-06,
|
| 4600 |
+
"loss": 2.483,
|
| 4601 |
+
"step": 656
|
| 4602 |
+
},
|
| 4603 |
+
{
|
| 4604 |
+
"epoch": 0.8436597110754415,
|
| 4605 |
+
"grad_norm": 9.342689514160156,
|
| 4606 |
+
"learning_rate": 6.005993492672657e-06,
|
| 4607 |
+
"loss": 2.9348,
|
| 4608 |
+
"step": 657
|
| 4609 |
+
},
|
| 4610 |
+
{
|
| 4611 |
+
"epoch": 0.8449438202247191,
|
| 4612 |
+
"grad_norm": 7.0432448387146,
|
| 4613 |
+
"learning_rate": 5.909917434791884e-06,
|
| 4614 |
+
"loss": 2.8985,
|
| 4615 |
+
"step": 658
|
| 4616 |
+
},
|
| 4617 |
+
{
|
| 4618 |
+
"epoch": 0.8462279293739968,
|
| 4619 |
+
"grad_norm": 6.8054914474487305,
|
| 4620 |
+
"learning_rate": 5.814567747314048e-06,
|
| 4621 |
+
"loss": 2.8271,
|
| 4622 |
+
"step": 659
|
| 4623 |
+
},
|
| 4624 |
+
{
|
| 4625 |
+
"epoch": 0.8475120385232745,
|
| 4626 |
+
"grad_norm": 8.824766159057617,
|
| 4627 |
+
"learning_rate": 5.719946001095616e-06,
|
| 4628 |
+
"loss": 2.989,
|
| 4629 |
+
"step": 660
|
| 4630 |
+
},
|
| 4631 |
+
{
|
| 4632 |
+
"epoch": 0.8487961476725522,
|
| 4633 |
+
"grad_norm": 9.073323249816895,
|
| 4634 |
+
"learning_rate": 5.626053755000421e-06,
|
| 4635 |
+
"loss": 4.1554,
|
| 4636 |
+
"step": 661
|
| 4637 |
+
},
|
| 4638 |
+
{
|
| 4639 |
+
"epoch": 0.8500802568218299,
|
| 4640 |
+
"grad_norm": 7.761005878448486,
|
| 4641 |
+
"learning_rate": 5.532892555874059e-06,
|
| 4642 |
+
"loss": 3.0684,
|
| 4643 |
+
"step": 662
|
| 4644 |
+
},
|
| 4645 |
+
{
|
| 4646 |
+
"epoch": 0.8513643659711075,
|
| 4647 |
+
"grad_norm": 6.79564905166626,
|
| 4648 |
+
"learning_rate": 5.440463938518303e-06,
|
| 4649 |
+
"loss": 3.2882,
|
| 4650 |
+
"step": 663
|
| 4651 |
+
},
|
| 4652 |
+
{
|
| 4653 |
+
"epoch": 0.8526484751203852,
|
| 4654 |
+
"grad_norm": 6.906652927398682,
|
| 4655 |
+
"learning_rate": 5.348769425665884e-06,
|
| 4656 |
+
"loss": 2.5132,
|
| 4657 |
+
"step": 664
|
| 4658 |
+
},
|
| 4659 |
+
{
|
| 4660 |
+
"epoch": 0.8539325842696629,
|
| 4661 |
+
"grad_norm": 9.796558380126953,
|
| 4662 |
+
"learning_rate": 5.257810527955409e-06,
|
| 4663 |
+
"loss": 3.4539,
|
| 4664 |
+
"step": 665
|
| 4665 |
+
},
|
| 4666 |
+
{
|
| 4667 |
+
"epoch": 0.8552166934189406,
|
| 4668 |
+
"grad_norm": 6.042295932769775,
|
| 4669 |
+
"learning_rate": 5.167588743906432e-06,
|
| 4670 |
+
"loss": 3.2793,
|
| 4671 |
+
"step": 666
|
| 4672 |
+
},
|
| 4673 |
+
{
|
| 4674 |
+
"epoch": 0.8565008025682183,
|
| 4675 |
+
"grad_norm": 5.436481475830078,
|
| 4676 |
+
"learning_rate": 5.078105559894791e-06,
|
| 4677 |
+
"loss": 2.2979,
|
| 4678 |
+
"step": 667
|
| 4679 |
+
},
|
| 4680 |
+
{
|
| 4681 |
+
"epoch": 0.8577849117174959,
|
| 4682 |
+
"grad_norm": 6.132842540740967,
|
| 4683 |
+
"learning_rate": 4.989362450128132e-06,
|
| 4684 |
+
"loss": 3.2762,
|
| 4685 |
+
"step": 668
|
| 4686 |
+
},
|
| 4687 |
+
{
|
| 4688 |
+
"epoch": 0.8590690208667737,
|
| 4689 |
+
"grad_norm": 19.214025497436523,
|
| 4690 |
+
"learning_rate": 4.901360876621597e-06,
|
| 4691 |
+
"loss": 4.0499,
|
| 4692 |
+
"step": 669
|
| 4693 |
+
},
|
| 4694 |
+
{
|
| 4695 |
+
"epoch": 0.8603531300160514,
|
| 4696 |
+
"grad_norm": 8.764758110046387,
|
| 4697 |
+
"learning_rate": 4.814102289173733e-06,
|
| 4698 |
+
"loss": 4.1739,
|
| 4699 |
+
"step": 670
|
| 4700 |
+
},
|
| 4701 |
+
{
|
| 4702 |
+
"epoch": 0.8616372391653291,
|
| 4703 |
+
"grad_norm": 6.873488426208496,
|
| 4704 |
+
"learning_rate": 4.727588125342669e-06,
|
| 4705 |
+
"loss": 3.2478,
|
| 4706 |
+
"step": 671
|
| 4707 |
+
},
|
| 4708 |
+
{
|
| 4709 |
+
"epoch": 0.8629213483146068,
|
| 4710 |
+
"grad_norm": 7.347663402557373,
|
| 4711 |
+
"learning_rate": 4.641819810422343e-06,
|
| 4712 |
+
"loss": 3.4422,
|
| 4713 |
+
"step": 672
|
| 4714 |
+
},
|
| 4715 |
+
{
|
| 4716 |
+
"epoch": 0.8642054574638844,
|
| 4717 |
+
"grad_norm": 7.408380031585693,
|
| 4718 |
+
"learning_rate": 4.556798757419068e-06,
|
| 4719 |
+
"loss": 2.9899,
|
| 4720 |
+
"step": 673
|
| 4721 |
+
},
|
| 4722 |
+
{
|
| 4723 |
+
"epoch": 0.8654895666131621,
|
| 4724 |
+
"grad_norm": 8.367959976196289,
|
| 4725 |
+
"learning_rate": 4.4725263670282905e-06,
|
| 4726 |
+
"loss": 3.5648,
|
| 4727 |
+
"step": 674
|
| 4728 |
+
},
|
| 4729 |
+
{
|
| 4730 |
+
"epoch": 0.8667736757624398,
|
| 4731 |
+
"grad_norm": 7.556360721588135,
|
| 4732 |
+
"learning_rate": 4.389004027611404e-06,
|
| 4733 |
+
"loss": 2.9127,
|
| 4734 |
+
"step": 675
|
| 4735 |
+
},
|
| 4736 |
+
{
|
| 4737 |
+
"epoch": 0.8680577849117175,
|
| 4738 |
+
"grad_norm": 6.504035949707031,
|
| 4739 |
+
"learning_rate": 4.3062331151730085e-06,
|
| 4740 |
+
"loss": 3.0646,
|
| 4741 |
+
"step": 676
|
| 4742 |
+
},
|
| 4743 |
+
{
|
| 4744 |
+
"epoch": 0.8693418940609952,
|
| 4745 |
+
"grad_norm": 7.055627346038818,
|
| 4746 |
+
"learning_rate": 4.224214993338149e-06,
|
| 4747 |
+
"loss": 2.8481,
|
| 4748 |
+
"step": 677
|
| 4749 |
+
},
|
| 4750 |
+
{
|
| 4751 |
+
"epoch": 0.8706260032102728,
|
| 4752 |
+
"grad_norm": 7.115128993988037,
|
| 4753 |
+
"learning_rate": 4.142951013329871e-06,
|
| 4754 |
+
"loss": 4.4346,
|
| 4755 |
+
"step": 678
|
| 4756 |
+
},
|
| 4757 |
+
{
|
| 4758 |
+
"epoch": 0.8719101123595505,
|
| 4759 |
+
"grad_norm": 8.57867431640625,
|
| 4760 |
+
"learning_rate": 4.062442513947007e-06,
|
| 4761 |
+
"loss": 3.6861,
|
| 4762 |
+
"step": 679
|
| 4763 |
+
},
|
| 4764 |
+
{
|
| 4765 |
+
"epoch": 0.8731942215088283,
|
| 4766 |
+
"grad_norm": 6.359896659851074,
|
| 4767 |
+
"learning_rate": 3.982690821542035e-06,
|
| 4768 |
+
"loss": 2.2618,
|
| 4769 |
+
"step": 680
|
| 4770 |
+
},
|
| 4771 |
+
{
|
| 4772 |
+
"epoch": 0.874478330658106,
|
| 4773 |
+
"grad_norm": 12.26174545288086,
|
| 4774 |
+
"learning_rate": 3.903697249999289e-06,
|
| 4775 |
+
"loss": 4.1746,
|
| 4776 |
+
"step": 681
|
| 4777 |
+
},
|
| 4778 |
+
{
|
| 4779 |
+
"epoch": 0.8757624398073837,
|
| 4780 |
+
"grad_norm": 7.479450225830078,
|
| 4781 |
+
"learning_rate": 3.8254631007133165e-06,
|
| 4782 |
+
"loss": 3.7854,
|
| 4783 |
+
"step": 682
|
| 4784 |
+
},
|
| 4785 |
+
{
|
| 4786 |
+
"epoch": 0.8770465489566613,
|
| 4787 |
+
"grad_norm": 8.125298500061035,
|
| 4788 |
+
"learning_rate": 3.7479896625674027e-06,
|
| 4789 |
+
"loss": 4.0026,
|
| 4790 |
+
"step": 683
|
| 4791 |
+
},
|
| 4792 |
+
{
|
| 4793 |
+
"epoch": 0.878330658105939,
|
| 4794 |
+
"grad_norm": 7.174749374389648,
|
| 4795 |
+
"learning_rate": 3.671278211912338e-06,
|
| 4796 |
+
"loss": 2.8277,
|
| 4797 |
+
"step": 684
|
| 4798 |
+
},
|
| 4799 |
+
{
|
| 4800 |
+
"epoch": 0.8796147672552167,
|
| 4801 |
+
"grad_norm": 7.114166736602783,
|
| 4802 |
+
"learning_rate": 3.595330012545445e-06,
|
| 4803 |
+
"loss": 3.7241,
|
| 4804 |
+
"step": 685
|
| 4805 |
+
},
|
| 4806 |
+
{
|
| 4807 |
+
"epoch": 0.8808988764044944,
|
| 4808 |
+
"grad_norm": 6.707666397094727,
|
| 4809 |
+
"learning_rate": 3.520146315689693e-06,
|
| 4810 |
+
"loss": 2.3535,
|
| 4811 |
+
"step": 686
|
| 4812 |
+
},
|
| 4813 |
+
{
|
| 4814 |
+
"epoch": 0.8821829855537721,
|
| 4815 |
+
"grad_norm": 7.33445930480957,
|
| 4816 |
+
"learning_rate": 3.445728359973094e-06,
|
| 4817 |
+
"loss": 3.2334,
|
| 4818 |
+
"step": 687
|
| 4819 |
+
},
|
| 4820 |
+
{
|
| 4821 |
+
"epoch": 0.8834670947030497,
|
| 4822 |
+
"grad_norm": 8.785782814025879,
|
| 4823 |
+
"learning_rate": 3.372077371408361e-06,
|
| 4824 |
+
"loss": 4.1721,
|
| 4825 |
+
"step": 688
|
| 4826 |
+
},
|
| 4827 |
+
{
|
| 4828 |
+
"epoch": 0.8847512038523274,
|
| 4829 |
+
"grad_norm": 6.775081634521484,
|
| 4830 |
+
"learning_rate": 3.299194563372604e-06,
|
| 4831 |
+
"loss": 2.6997,
|
| 4832 |
+
"step": 689
|
| 4833 |
+
},
|
| 4834 |
+
{
|
| 4835 |
+
"epoch": 0.8860353130016051,
|
| 4836 |
+
"grad_norm": 6.534388542175293,
|
| 4837 |
+
"learning_rate": 3.22708113658744e-06,
|
| 4838 |
+
"loss": 2.798,
|
| 4839 |
+
"step": 690
|
| 4840 |
+
},
|
| 4841 |
+
{
|
| 4842 |
+
"epoch": 0.8873194221508828,
|
| 4843 |
+
"grad_norm": 7.509188175201416,
|
| 4844 |
+
"learning_rate": 3.1557382790991687e-06,
|
| 4845 |
+
"loss": 3.993,
|
| 4846 |
+
"step": 691
|
| 4847 |
+
},
|
| 4848 |
+
{
|
| 4849 |
+
"epoch": 0.8886035313001606,
|
| 4850 |
+
"grad_norm": 6.793269157409668,
|
| 4851 |
+
"learning_rate": 3.085167166259162e-06,
|
| 4852 |
+
"loss": 2.6374,
|
| 4853 |
+
"step": 692
|
| 4854 |
+
},
|
| 4855 |
+
{
|
| 4856 |
+
"epoch": 0.8898876404494382,
|
| 4857 |
+
"grad_norm": 7.3369340896606445,
|
| 4858 |
+
"learning_rate": 3.0153689607045845e-06,
|
| 4859 |
+
"loss": 2.9641,
|
| 4860 |
+
"step": 693
|
| 4861 |
+
},
|
| 4862 |
+
{
|
| 4863 |
+
"epoch": 0.8911717495987159,
|
| 4864 |
+
"grad_norm": 8.02035140991211,
|
| 4865 |
+
"learning_rate": 2.9463448123391634e-06,
|
| 4866 |
+
"loss": 4.3613,
|
| 4867 |
+
"step": 694
|
| 4868 |
+
},
|
| 4869 |
+
{
|
| 4870 |
+
"epoch": 0.8924558587479936,
|
| 4871 |
+
"grad_norm": 6.194982051849365,
|
| 4872 |
+
"learning_rate": 2.878095858314278e-06,
|
| 4873 |
+
"loss": 2.8135,
|
| 4874 |
+
"step": 695
|
| 4875 |
+
},
|
| 4876 |
+
{
|
| 4877 |
+
"epoch": 0.8937399678972713,
|
| 4878 |
+
"grad_norm": 8.533773422241211,
|
| 4879 |
+
"learning_rate": 2.8106232230102448e-06,
|
| 4880 |
+
"loss": 3.7117,
|
| 4881 |
+
"step": 696
|
| 4882 |
+
},
|
| 4883 |
+
{
|
| 4884 |
+
"epoch": 0.895024077046549,
|
| 4885 |
+
"grad_norm": 7.960330009460449,
|
| 4886 |
+
"learning_rate": 2.743928018017744e-06,
|
| 4887 |
+
"loss": 2.6817,
|
| 4888 |
+
"step": 697
|
| 4889 |
+
},
|
| 4890 |
+
{
|
| 4891 |
+
"epoch": 0.8963081861958266,
|
| 4892 |
+
"grad_norm": 8.209793090820312,
|
| 4893 |
+
"learning_rate": 2.6780113421195298e-06,
|
| 4894 |
+
"loss": 2.9455,
|
| 4895 |
+
"step": 698
|
| 4896 |
+
},
|
| 4897 |
+
{
|
| 4898 |
+
"epoch": 0.8975922953451043,
|
| 4899 |
+
"grad_norm": 7.603484153747559,
|
| 4900 |
+
"learning_rate": 2.6128742812723704e-06,
|
| 4901 |
+
"loss": 2.6488,
|
| 4902 |
+
"step": 699
|
| 4903 |
+
},
|
| 4904 |
+
{
|
| 4905 |
+
"epoch": 0.898876404494382,
|
| 4906 |
+
"grad_norm": 7.542942523956299,
|
| 4907 |
+
"learning_rate": 2.5485179085890767e-06,
|
| 4908 |
+
"loss": 3.0132,
|
| 4909 |
+
"step": 700
|
| 4910 |
+
},
|
| 4911 |
+
{
|
| 4912 |
+
"epoch": 0.9001605136436597,
|
| 4913 |
+
"grad_norm": 8.356619834899902,
|
| 4914 |
+
"learning_rate": 2.4849432843208785e-06,
|
| 4915 |
+
"loss": 3.2918,
|
| 4916 |
+
"step": 701
|
| 4917 |
+
},
|
| 4918 |
+
{
|
| 4919 |
+
"epoch": 0.9014446227929374,
|
| 4920 |
+
"grad_norm": 8.363799095153809,
|
| 4921 |
+
"learning_rate": 2.4221514558399548e-06,
|
| 4922 |
+
"loss": 2.7588,
|
| 4923 |
+
"step": 702
|
| 4924 |
+
},
|
| 4925 |
+
{
|
| 4926 |
+
"epoch": 0.902728731942215,
|
| 4927 |
+
"grad_norm": 5.096377372741699,
|
| 4928 |
+
"learning_rate": 2.3601434576221546e-06,
|
| 4929 |
+
"loss": 2.2989,
|
| 4930 |
+
"step": 703
|
| 4931 |
+
},
|
| 4932 |
+
{
|
| 4933 |
+
"epoch": 0.9040128410914928,
|
| 4934 |
+
"grad_norm": 6.197229385375977,
|
| 4935 |
+
"learning_rate": 2.2989203112299684e-06,
|
| 4936 |
+
"loss": 3.1503,
|
| 4937 |
+
"step": 704
|
| 4938 |
+
},
|
| 4939 |
+
{
|
| 4940 |
+
"epoch": 0.9052969502407705,
|
| 4941 |
+
"grad_norm": 6.665518760681152,
|
| 4942 |
+
"learning_rate": 2.238483025295707e-06,
|
| 4943 |
+
"loss": 3.6999,
|
| 4944 |
+
"step": 705
|
| 4945 |
+
},
|
| 4946 |
+
{
|
| 4947 |
+
"epoch": 0.9065810593900482,
|
| 4948 |
+
"grad_norm": 5.712645053863525,
|
| 4949 |
+
"learning_rate": 2.178832595504854e-06,
|
| 4950 |
+
"loss": 2.3308,
|
| 4951 |
+
"step": 706
|
| 4952 |
+
},
|
| 4953 |
+
{
|
| 4954 |
+
"epoch": 0.9078651685393259,
|
| 4955 |
+
"grad_norm": 8.328536033630371,
|
| 4956 |
+
"learning_rate": 2.1199700045797077e-06,
|
| 4957 |
+
"loss": 4.041,
|
| 4958 |
+
"step": 707
|
| 4959 |
+
},
|
| 4960 |
+
{
|
| 4961 |
+
"epoch": 0.9091492776886035,
|
| 4962 |
+
"grad_norm": 7.73337984085083,
|
| 4963 |
+
"learning_rate": 2.0618962222631432e-06,
|
| 4964 |
+
"loss": 3.2939,
|
| 4965 |
+
"step": 708
|
| 4966 |
+
},
|
| 4967 |
+
{
|
| 4968 |
+
"epoch": 0.9104333868378812,
|
| 4969 |
+
"grad_norm": 5.95501184463501,
|
| 4970 |
+
"learning_rate": 2.0046122053026694e-06,
|
| 4971 |
+
"loss": 3.0356,
|
| 4972 |
+
"step": 709
|
| 4973 |
+
},
|
| 4974 |
+
{
|
| 4975 |
+
"epoch": 0.9117174959871589,
|
| 4976 |
+
"grad_norm": 7.161213397979736,
|
| 4977 |
+
"learning_rate": 1.9481188974346696e-06,
|
| 4978 |
+
"loss": 3.4678,
|
| 4979 |
+
"step": 710
|
| 4980 |
+
},
|
| 4981 |
+
{
|
| 4982 |
+
"epoch": 0.9130016051364366,
|
| 4983 |
+
"grad_norm": 5.240903377532959,
|
| 4984 |
+
"learning_rate": 1.8924172293688147e-06,
|
| 4985 |
+
"loss": 2.3559,
|
| 4986 |
+
"step": 711
|
| 4987 |
+
},
|
| 4988 |
+
{
|
| 4989 |
+
"epoch": 0.9142857142857143,
|
| 4990 |
+
"grad_norm": 7.932479381561279,
|
| 4991 |
+
"learning_rate": 1.8375081187727683e-06,
|
| 4992 |
+
"loss": 3.328,
|
| 4993 |
+
"step": 712
|
| 4994 |
+
},
|
| 4995 |
+
{
|
| 4996 |
+
"epoch": 0.9155698234349919,
|
| 4997 |
+
"grad_norm": 7.0898051261901855,
|
| 4998 |
+
"learning_rate": 1.7833924702570725e-06,
|
| 4999 |
+
"loss": 3.2187,
|
| 5000 |
+
"step": 713
|
| 5001 |
+
},
|
| 5002 |
+
{
|
| 5003 |
+
"epoch": 0.9168539325842696,
|
| 5004 |
+
"grad_norm": 6.591296195983887,
|
| 5005 |
+
"learning_rate": 1.7300711753601983e-06,
|
| 5006 |
+
"loss": 2.724,
|
| 5007 |
+
"step": 714
|
| 5008 |
+
},
|
| 5009 |
+
{
|
| 5010 |
+
"epoch": 0.9181380417335474,
|
| 5011 |
+
"grad_norm": 6.064736843109131,
|
| 5012 |
+
"learning_rate": 1.6775451125338959e-06,
|
| 5013 |
+
"loss": 2.4792,
|
| 5014 |
+
"step": 715
|
| 5015 |
+
},
|
| 5016 |
+
{
|
| 5017 |
+
"epoch": 0.9194221508828251,
|
| 5018 |
+
"grad_norm": 6.841074466705322,
|
| 5019 |
+
"learning_rate": 1.6258151471287396e-06,
|
| 5020 |
+
"loss": 3.3866,
|
| 5021 |
+
"step": 716
|
| 5022 |
+
},
|
| 5023 |
+
{
|
| 5024 |
+
"epoch": 0.9207062600321028,
|
| 5025 |
+
"grad_norm": 7.046119689941406,
|
| 5026 |
+
"learning_rate": 1.5748821313798124e-06,
|
| 5027 |
+
"loss": 3.0957,
|
| 5028 |
+
"step": 717
|
| 5029 |
+
},
|
| 5030 |
+
{
|
| 5031 |
+
"epoch": 0.9219903691813804,
|
| 5032 |
+
"grad_norm": 9.995711326599121,
|
| 5033 |
+
"learning_rate": 1.5247469043927155e-06,
|
| 5034 |
+
"loss": 3.2111,
|
| 5035 |
+
"step": 718
|
| 5036 |
+
},
|
| 5037 |
+
{
|
| 5038 |
+
"epoch": 0.9232744783306581,
|
| 5039 |
+
"grad_norm": 7.515043258666992,
|
| 5040 |
+
"learning_rate": 1.4754102921297364e-06,
|
| 5041 |
+
"loss": 2.9457,
|
| 5042 |
+
"step": 719
|
| 5043 |
+
},
|
| 5044 |
+
{
|
| 5045 |
+
"epoch": 0.9245585874799358,
|
| 5046 |
+
"grad_norm": 6.644869804382324,
|
| 5047 |
+
"learning_rate": 1.4268731073962094e-06,
|
| 5048 |
+
"loss": 3.0214,
|
| 5049 |
+
"step": 720
|
| 5050 |
+
},
|
| 5051 |
+
{
|
| 5052 |
+
"epoch": 0.9258426966292135,
|
| 5053 |
+
"grad_norm": 6.054432392120361,
|
| 5054 |
+
"learning_rate": 1.3791361498271705e-06,
|
| 5055 |
+
"loss": 2.3658,
|
| 5056 |
+
"step": 721
|
| 5057 |
+
},
|
| 5058 |
+
{
|
| 5059 |
+
"epoch": 0.9271268057784912,
|
| 5060 |
+
"grad_norm": 6.487936019897461,
|
| 5061 |
+
"learning_rate": 1.3322002058741678e-06,
|
| 5062 |
+
"loss": 2.7158,
|
| 5063 |
+
"step": 722
|
| 5064 |
+
},
|
| 5065 |
+
{
|
| 5066 |
+
"epoch": 0.9284109149277688,
|
| 5067 |
+
"grad_norm": 9.501916885375977,
|
| 5068 |
+
"learning_rate": 1.2860660487922616e-06,
|
| 5069 |
+
"loss": 3.9914,
|
| 5070 |
+
"step": 723
|
| 5071 |
+
},
|
| 5072 |
+
{
|
| 5073 |
+
"epoch": 0.9296950240770465,
|
| 5074 |
+
"grad_norm": 6.846279621124268,
|
| 5075 |
+
"learning_rate": 1.2407344386273611e-06,
|
| 5076 |
+
"loss": 2.7458,
|
| 5077 |
+
"step": 724
|
| 5078 |
+
},
|
| 5079 |
+
{
|
| 5080 |
+
"epoch": 0.9309791332263242,
|
| 5081 |
+
"grad_norm": 7.539971828460693,
|
| 5082 |
+
"learning_rate": 1.196206122203647e-06,
|
| 5083 |
+
"loss": 3.3559,
|
| 5084 |
+
"step": 725
|
| 5085 |
+
},
|
| 5086 |
+
{
|
| 5087 |
+
"epoch": 0.9322632423756019,
|
| 5088 |
+
"grad_norm": 8.702104568481445,
|
| 5089 |
+
"learning_rate": 1.1524818331112851e-06,
|
| 5090 |
+
"loss": 3.0192,
|
| 5091 |
+
"step": 726
|
| 5092 |
+
},
|
| 5093 |
+
{
|
| 5094 |
+
"epoch": 0.9335473515248797,
|
| 5095 |
+
"grad_norm": 7.85031270980835,
|
| 5096 |
+
"learning_rate": 1.1095622916943494e-06,
|
| 5097 |
+
"loss": 3.9229,
|
| 5098 |
+
"step": 727
|
| 5099 |
+
},
|
| 5100 |
+
{
|
| 5101 |
+
"epoch": 0.9348314606741573,
|
| 5102 |
+
"grad_norm": 6.479403972625732,
|
| 5103 |
+
"learning_rate": 1.0674482050389455e-06,
|
| 5104 |
+
"loss": 2.739,
|
| 5105 |
+
"step": 728
|
| 5106 |
+
},
|
| 5107 |
+
{
|
| 5108 |
+
"epoch": 0.936115569823435,
|
| 5109 |
+
"grad_norm": 7.294396877288818,
|
| 5110 |
+
"learning_rate": 1.0261402669615505e-06,
|
| 5111 |
+
"loss": 3.6074,
|
| 5112 |
+
"step": 729
|
| 5113 |
+
},
|
| 5114 |
+
{
|
| 5115 |
+
"epoch": 0.9373996789727127,
|
| 5116 |
+
"grad_norm": 7.156967639923096,
|
| 5117 |
+
"learning_rate": 9.856391579976032e-07,
|
| 5118 |
+
"loss": 3.3275,
|
| 5119 |
+
"step": 730
|
| 5120 |
+
},
|
| 5121 |
+
{
|
| 5122 |
+
"epoch": 0.9386837881219904,
|
| 5123 |
+
"grad_norm": 7.538506031036377,
|
| 5124 |
+
"learning_rate": 9.459455453902866e-07,
|
| 5125 |
+
"loss": 3.2643,
|
| 5126 |
+
"step": 731
|
| 5127 |
+
},
|
| 5128 |
+
{
|
| 5129 |
+
"epoch": 0.9399678972712681,
|
| 5130 |
+
"grad_norm": 8.024343490600586,
|
| 5131 |
+
"learning_rate": 9.070600830795251e-07,
|
| 5132 |
+
"loss": 2.7308,
|
| 5133 |
+
"step": 732
|
| 5134 |
+
},
|
| 5135 |
+
{
|
| 5136 |
+
"epoch": 0.9412520064205457,
|
| 5137 |
+
"grad_norm": 9.4879150390625,
|
| 5138 |
+
"learning_rate": 8.68983411691221e-07,
|
| 5139 |
+
"loss": 4.2723,
|
| 5140 |
+
"step": 733
|
| 5141 |
+
},
|
| 5142 |
+
{
|
| 5143 |
+
"epoch": 0.9425361155698234,
|
| 5144 |
+
"grad_norm": 5.0413713455200195,
|
| 5145 |
+
"learning_rate": 8.317161585266964e-07,
|
| 5146 |
+
"loss": 2.3525,
|
| 5147 |
+
"step": 734
|
| 5148 |
+
},
|
| 5149 |
+
{
|
| 5150 |
+
"epoch": 0.9438202247191011,
|
| 5151 |
+
"grad_norm": 8.031405448913574,
|
| 5152 |
+
"learning_rate": 7.952589375523567e-07,
|
| 5153 |
+
"loss": 3.5764,
|
| 5154 |
+
"step": 735
|
| 5155 |
+
},
|
| 5156 |
+
{
|
| 5157 |
+
"epoch": 0.9451043338683788,
|
| 5158 |
+
"grad_norm": 8.198379516601562,
|
| 5159 |
+
"learning_rate": 7.596123493895991e-07,
|
| 5160 |
+
"loss": 4.5045,
|
| 5161 |
+
"step": 736
|
| 5162 |
+
},
|
| 5163 |
+
{
|
| 5164 |
+
"epoch": 0.9463884430176565,
|
| 5165 |
+
"grad_norm": 8.198650360107422,
|
| 5166 |
+
"learning_rate": 7.247769813048644e-07,
|
| 5167 |
+
"loss": 3.1831,
|
| 5168 |
+
"step": 737
|
| 5169 |
+
},
|
| 5170 |
+
{
|
| 5171 |
+
"epoch": 0.9476725521669341,
|
| 5172 |
+
"grad_norm": 6.928763389587402,
|
| 5173 |
+
"learning_rate": 6.907534072000177e-07,
|
| 5174 |
+
"loss": 3.261,
|
| 5175 |
+
"step": 738
|
| 5176 |
+
},
|
| 5177 |
+
{
|
| 5178 |
+
"epoch": 0.9489566613162119,
|
| 5179 |
+
"grad_norm": 6.926176071166992,
|
| 5180 |
+
"learning_rate": 6.57542187602872e-07,
|
| 5181 |
+
"loss": 3.2459,
|
| 5182 |
+
"step": 739
|
| 5183 |
+
},
|
| 5184 |
+
{
|
| 5185 |
+
"epoch": 0.9502407704654896,
|
| 5186 |
+
"grad_norm": 7.637943267822266,
|
| 5187 |
+
"learning_rate": 6.251438696579293e-07,
|
| 5188 |
+
"loss": 3.2871,
|
| 5189 |
+
"step": 740
|
| 5190 |
+
},
|
| 5191 |
+
{
|
| 5192 |
+
"epoch": 0.9515248796147673,
|
| 5193 |
+
"grad_norm": 6.943028450012207,
|
| 5194 |
+
"learning_rate": 5.935589871174208e-07,
|
| 5195 |
+
"loss": 3.2108,
|
| 5196 |
+
"step": 741
|
| 5197 |
+
},
|
| 5198 |
+
{
|
| 5199 |
+
"epoch": 0.952808988764045,
|
| 5200 |
+
"grad_norm": 7.411741256713867,
|
| 5201 |
+
"learning_rate": 5.627880603324532e-07,
|
| 5202 |
+
"loss": 2.5244,
|
| 5203 |
+
"step": 742
|
| 5204 |
+
},
|
| 5205 |
+
{
|
| 5206 |
+
"epoch": 0.9540930979133226,
|
| 5207 |
+
"grad_norm": 8.225117683410645,
|
| 5208 |
+
"learning_rate": 5.328315962444874e-07,
|
| 5209 |
+
"loss": 3.3371,
|
| 5210 |
+
"step": 743
|
| 5211 |
+
},
|
| 5212 |
+
{
|
| 5213 |
+
"epoch": 0.9553772070626003,
|
| 5214 |
+
"grad_norm": 8.055129051208496,
|
| 5215 |
+
"learning_rate": 5.036900883769624e-07,
|
| 5216 |
+
"loss": 3.4518,
|
| 5217 |
+
"step": 744
|
| 5218 |
+
},
|
| 5219 |
+
{
|
| 5220 |
+
"epoch": 0.956661316211878,
|
| 5221 |
+
"grad_norm": 7.234530925750732,
|
| 5222 |
+
"learning_rate": 4.753640168271456e-07,
|
| 5223 |
+
"loss": 3.6194,
|
| 5224 |
+
"step": 745
|
| 5225 |
+
},
|
| 5226 |
+
{
|
| 5227 |
+
"epoch": 0.9579454253611557,
|
| 5228 |
+
"grad_norm": 7.3645920753479,
|
| 5229 |
+
"learning_rate": 4.4785384825826173e-07,
|
| 5230 |
+
"loss": 2.6069,
|
| 5231 |
+
"step": 746
|
| 5232 |
+
},
|
| 5233 |
+
{
|
| 5234 |
+
"epoch": 0.9592295345104334,
|
| 5235 |
+
"grad_norm": 8.200539588928223,
|
| 5236 |
+
"learning_rate": 4.2116003589179887e-07,
|
| 5237 |
+
"loss": 3.3361,
|
| 5238 |
+
"step": 747
|
| 5239 |
+
},
|
| 5240 |
+
{
|
| 5241 |
+
"epoch": 0.960513643659711,
|
| 5242 |
+
"grad_norm": 7.350730895996094,
|
| 5243 |
+
"learning_rate": 3.9528301950000347e-07,
|
| 5244 |
+
"loss": 3.3126,
|
| 5245 |
+
"step": 748
|
| 5246 |
+
},
|
| 5247 |
+
{
|
| 5248 |
+
"epoch": 0.9617977528089887,
|
| 5249 |
+
"grad_norm": 9.777095794677734,
|
| 5250 |
+
"learning_rate": 3.702232253986804e-07,
|
| 5251 |
+
"loss": 3.9927,
|
| 5252 |
+
"step": 749
|
| 5253 |
+
},
|
| 5254 |
+
{
|
| 5255 |
+
"epoch": 0.9630818619582665,
|
| 5256 |
+
"grad_norm": 7.824852466583252,
|
| 5257 |
+
"learning_rate": 3.459810664401486e-07,
|
| 5258 |
+
"loss": 2.5065,
|
| 5259 |
+
"step": 750
|
| 5260 |
+
},
|
| 5261 |
+
{
|
| 5262 |
+
"epoch": 0.9643659711075442,
|
| 5263 |
+
"grad_norm": 14.027108192443848,
|
| 5264 |
+
"learning_rate": 3.2255694200643006e-07,
|
| 5265 |
+
"loss": 4.6722,
|
| 5266 |
+
"step": 751
|
| 5267 |
+
},
|
| 5268 |
+
{
|
| 5269 |
+
"epoch": 0.9656500802568219,
|
| 5270 |
+
"grad_norm": 8.372529983520508,
|
| 5271 |
+
"learning_rate": 2.9995123800270476e-07,
|
| 5272 |
+
"loss": 3.2234,
|
| 5273 |
+
"step": 752
|
| 5274 |
+
},
|
| 5275 |
+
{
|
| 5276 |
+
"epoch": 0.9669341894060995,
|
| 5277 |
+
"grad_norm": 6.728063583374023,
|
| 5278 |
+
"learning_rate": 2.78164326850916e-07,
|
| 5279 |
+
"loss": 3.0884,
|
| 5280 |
+
"step": 753
|
| 5281 |
+
},
|
| 5282 |
+
{
|
| 5283 |
+
"epoch": 0.9682182985553772,
|
| 5284 |
+
"grad_norm": 6.5443902015686035,
|
| 5285 |
+
"learning_rate": 2.571965674836418e-07,
|
| 5286 |
+
"loss": 3.1012,
|
| 5287 |
+
"step": 754
|
| 5288 |
+
},
|
| 5289 |
+
{
|
| 5290 |
+
"epoch": 0.9695024077046549,
|
| 5291 |
+
"grad_norm": 5.537796497344971,
|
| 5292 |
+
"learning_rate": 2.3704830533821108e-07,
|
| 5293 |
+
"loss": 2.5468,
|
| 5294 |
+
"step": 755
|
| 5295 |
+
},
|
| 5296 |
+
{
|
| 5297 |
+
"epoch": 0.9707865168539326,
|
| 5298 |
+
"grad_norm": 5.125178813934326,
|
| 5299 |
+
"learning_rate": 2.177198723509688e-07,
|
| 5300 |
+
"loss": 2.5329,
|
| 5301 |
+
"step": 756
|
| 5302 |
+
},
|
| 5303 |
+
{
|
| 5304 |
+
"epoch": 0.9720706260032103,
|
| 5305 |
+
"grad_norm": 5.382153034210205,
|
| 5306 |
+
"learning_rate": 1.9921158695184738e-07,
|
| 5307 |
+
"loss": 2.6962,
|
| 5308 |
+
"step": 757
|
| 5309 |
+
},
|
| 5310 |
+
{
|
| 5311 |
+
"epoch": 0.9733547351524879,
|
| 5312 |
+
"grad_norm": 8.632124900817871,
|
| 5313 |
+
"learning_rate": 1.8152375405909305e-07,
|
| 5314 |
+
"loss": 4.1156,
|
| 5315 |
+
"step": 758
|
| 5316 |
+
},
|
| 5317 |
+
{
|
| 5318 |
+
"epoch": 0.9746388443017656,
|
| 5319 |
+
"grad_norm": 8.175101280212402,
|
| 5320 |
+
"learning_rate": 1.6465666507425315e-07,
|
| 5321 |
+
"loss": 3.4293,
|
| 5322 |
+
"step": 759
|
| 5323 |
+
},
|
| 5324 |
+
{
|
| 5325 |
+
"epoch": 0.9759229534510433,
|
| 5326 |
+
"grad_norm": 6.334762096405029,
|
| 5327 |
+
"learning_rate": 1.4861059787736885e-07,
|
| 5328 |
+
"loss": 3.3443,
|
| 5329 |
+
"step": 760
|
| 5330 |
+
},
|
| 5331 |
+
{
|
| 5332 |
+
"epoch": 0.977207062600321,
|
| 5333 |
+
"grad_norm": 7.494119644165039,
|
| 5334 |
+
"learning_rate": 1.333858168224178e-07,
|
| 5335 |
+
"loss": 2.7613,
|
| 5336 |
+
"step": 761
|
| 5337 |
+
},
|
| 5338 |
+
{
|
| 5339 |
+
"epoch": 0.9784911717495988,
|
| 5340 |
+
"grad_norm": 10.954715728759766,
|
| 5341 |
+
"learning_rate": 1.1898257273292857e-07,
|
| 5342 |
+
"loss": 2.9989,
|
| 5343 |
+
"step": 762
|
| 5344 |
+
},
|
| 5345 |
+
{
|
| 5346 |
+
"epoch": 0.9797752808988764,
|
| 5347 |
+
"grad_norm": 6.305700302124023,
|
| 5348 |
+
"learning_rate": 1.0540110289786742e-07,
|
| 5349 |
+
"loss": 3.4268,
|
| 5350 |
+
"step": 763
|
| 5351 |
+
},
|
| 5352 |
+
{
|
| 5353 |
+
"epoch": 0.9810593900481541,
|
| 5354 |
+
"grad_norm": 8.201444625854492,
|
| 5355 |
+
"learning_rate": 9.264163106774137e-08,
|
| 5356 |
+
"loss": 3.5135,
|
| 5357 |
+
"step": 764
|
| 5358 |
+
},
|
| 5359 |
+
{
|
| 5360 |
+
"epoch": 0.9823434991974318,
|
| 5361 |
+
"grad_norm": 7.041131019592285,
|
| 5362 |
+
"learning_rate": 8.07043674508623e-08,
|
| 5363 |
+
"loss": 3.0868,
|
| 5364 |
+
"step": 765
|
| 5365 |
+
},
|
| 5366 |
+
{
|
| 5367 |
+
"epoch": 0.9836276083467095,
|
| 5368 |
+
"grad_norm": 6.509793281555176,
|
| 5369 |
+
"learning_rate": 6.958950870994963e-08,
|
| 5370 |
+
"loss": 3.5145,
|
| 5371 |
+
"step": 766
|
| 5372 |
+
},
|
| 5373 |
+
{
|
| 5374 |
+
"epoch": 0.9849117174959872,
|
| 5375 |
+
"grad_norm": 7.05448055267334,
|
| 5376 |
+
"learning_rate": 5.929723795884967e-08,
|
| 5377 |
+
"loss": 2.7421,
|
| 5378 |
+
"step": 767
|
| 5379 |
+
},
|
| 5380 |
+
{
|
| 5381 |
+
"epoch": 0.9861958266452648,
|
| 5382 |
+
"grad_norm": 6.827080249786377,
|
| 5383 |
+
"learning_rate": 4.982772475951025e-08,
|
| 5384 |
+
"loss": 2.4727,
|
| 5385 |
+
"step": 768
|
| 5386 |
+
},
|
| 5387 |
+
{
|
| 5388 |
+
"epoch": 0.9874799357945425,
|
| 5389 |
+
"grad_norm": 9.33382511138916,
|
| 5390 |
+
"learning_rate": 4.1181125119221787e-08,
|
| 5391 |
+
"loss": 5.2741,
|
| 5392 |
+
"step": 769
|
| 5393 |
+
},
|
| 5394 |
+
{
|
| 5395 |
+
"epoch": 0.9887640449438202,
|
| 5396 |
+
"grad_norm": 6.9964447021484375,
|
| 5397 |
+
"learning_rate": 3.3357581488030475e-08,
|
| 5398 |
+
"loss": 2.8354,
|
| 5399 |
+
"step": 770
|
| 5400 |
+
},
|
| 5401 |
+
{
|
| 5402 |
+
"epoch": 0.9900481540930979,
|
| 5403 |
+
"grad_norm": 7.825475215911865,
|
| 5404 |
+
"learning_rate": 2.6357222756384636e-08,
|
| 5405 |
+
"loss": 2.8096,
|
| 5406 |
+
"step": 771
|
| 5407 |
+
},
|
| 5408 |
+
{
|
| 5409 |
+
"epoch": 0.9913322632423756,
|
| 5410 |
+
"grad_norm": 6.928106784820557,
|
| 5411 |
+
"learning_rate": 2.0180164253008614e-08,
|
| 5412 |
+
"loss": 3.5979,
|
| 5413 |
+
"step": 772
|
| 5414 |
+
},
|
| 5415 |
+
{
|
| 5416 |
+
"epoch": 0.9926163723916533,
|
| 5417 |
+
"grad_norm": 7.775272369384766,
|
| 5418 |
+
"learning_rate": 1.482650774303207e-08,
|
| 5419 |
+
"loss": 2.3856,
|
| 5420 |
+
"step": 773
|
| 5421 |
+
},
|
| 5422 |
+
{
|
| 5423 |
+
"epoch": 0.993900481540931,
|
| 5424 |
+
"grad_norm": 10.018780708312988,
|
| 5425 |
+
"learning_rate": 1.0296341426274669e-08,
|
| 5426 |
+
"loss": 4.6775,
|
| 5427 |
+
"step": 774
|
| 5428 |
+
},
|
| 5429 |
+
{
|
| 5430 |
+
"epoch": 0.9951845906902087,
|
| 5431 |
+
"grad_norm": 8.294422149658203,
|
| 5432 |
+
"learning_rate": 6.589739935819461e-09,
|
| 5433 |
+
"loss": 3.1598,
|
| 5434 |
+
"step": 775
|
| 5435 |
+
},
|
| 5436 |
+
{
|
| 5437 |
+
"epoch": 0.9964686998394864,
|
| 5438 |
+
"grad_norm": 8.607256889343262,
|
| 5439 |
+
"learning_rate": 3.7067643367749705e-09,
|
| 5440 |
+
"loss": 3.5999,
|
| 5441 |
+
"step": 776
|
| 5442 |
+
},
|
| 5443 |
+
{
|
| 5444 |
+
"epoch": 0.9977528089887641,
|
| 5445 |
+
"grad_norm": 7.560439586639404,
|
| 5446 |
+
"learning_rate": 1.6474621252704493e-09,
|
| 5447 |
+
"loss": 3.439,
|
| 5448 |
+
"step": 777
|
| 5449 |
+
},
|
| 5450 |
+
{
|
| 5451 |
+
"epoch": 0.9990369181380417,
|
| 5452 |
+
"grad_norm": 11.580015182495117,
|
| 5453 |
+
"learning_rate": 4.118672276620661e-10,
|
| 5454 |
+
"loss": 4.1616,
|
| 5455 |
+
"step": 778
|
| 5456 |
+
},
|
| 5457 |
+
{
|
| 5458 |
+
"epoch": 0.9990369181380417,
|
| 5459 |
+
"eval_loss": 0.8082237243652344,
|
| 5460 |
+
"eval_runtime": 52.1681,
|
| 5461 |
+
"eval_samples_per_second": 12.575,
|
| 5462 |
+
"eval_steps_per_second": 3.144,
|
| 5463 |
+
"step": 778
|
| 5464 |
+
},
|
| 5465 |
+
{
|
| 5466 |
+
"epoch": 1.0009630818619584,
|
| 5467 |
+
"grad_norm": 9.079451560974121,
|
| 5468 |
+
"learning_rate": 0.0,
|
| 5469 |
+
"loss": 4.5426,
|
| 5470 |
+
"step": 779
|
| 5471 |
}
|
| 5472 |
],
|
| 5473 |
"logging_steps": 1,
|
|
|
|
| 5482 |
"should_evaluate": false,
|
| 5483 |
"should_log": false,
|
| 5484 |
"should_save": true,
|
| 5485 |
+
"should_training_stop": true
|
| 5486 |
},
|
| 5487 |
"attributes": {}
|
| 5488 |
}
|
| 5489 |
},
|
| 5490 |
+
"total_flos": 9.195152422522061e+17,
|
| 5491 |
"train_batch_size": 4,
|
| 5492 |
"trial_name": null,
|
| 5493 |
"trial_params": null
|