Training in progress, step 1191, 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 125040
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4bedb0111ed18bc25163bbb8e8a61ae926892e8dba7a9674d7fd998082f99bbf
|
| 3 |
size 125040
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 162868
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8cd50ae86362ef7e08351632f4862c08016e8a82522d1baabfc21db012381cae
|
| 3 |
size 162868
|
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 sha256:8e4811efa6b667c9df043dc56d9d4b0af46926d1f1888ffc35e091c15f5c56d6
|
| 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 sha256:4dfb2490db991f40bc8a28ece8171c1b460ff28d33e0f3d99d0a22dae8c82b0a
|
| 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": 397,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5589,6 +5589,2793 @@
|
|
| 5589 |
"eval_samples_per_second": 369.403,
|
| 5590 |
"eval_steps_per_second": 184.745,
|
| 5591 |
"step": 794
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5592 |
}
|
| 5593 |
],
|
| 5594 |
"logging_steps": 1,
|
|
@@ -5608,7 +8395,7 @@
|
|
| 5608 |
"attributes": {}
|
| 5609 |
}
|
| 5610 |
},
|
| 5611 |
-
"total_flos":
|
| 5612 |
"train_batch_size": 2,
|
| 5613 |
"trial_name": null,
|
| 5614 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.11813132315016862,
|
| 5 |
"eval_steps": 397,
|
| 6 |
+
"global_step": 1191,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5589 |
"eval_samples_per_second": 369.403,
|
| 5590 |
"eval_steps_per_second": 184.745,
|
| 5591 |
"step": 794
|
| 5592 |
+
},
|
| 5593 |
+
{
|
| 5594 |
+
"epoch": 0.07885340210275739,
|
| 5595 |
+
"grad_norm": 1.8805289268493652,
|
| 5596 |
+
"learning_rate": 0.00010049866343387581,
|
| 5597 |
+
"loss": 43.2374,
|
| 5598 |
+
"step": 795
|
| 5599 |
+
},
|
| 5600 |
+
{
|
| 5601 |
+
"epoch": 0.07895258877206904,
|
| 5602 |
+
"grad_norm": 1.8383066654205322,
|
| 5603 |
+
"learning_rate": 0.00010029919885393563,
|
| 5604 |
+
"loss": 42.98,
|
| 5605 |
+
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.07905177544138067,
|
| 5609 |
+
"grad_norm": 1.9478559494018555,
|
| 5610 |
+
"learning_rate": 0.0001000997330835804,
|
| 5611 |
+
"loss": 43.2858,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.07915096211069232,
|
| 5616 |
+
"grad_norm": 2.4916789531707764,
|
| 5617 |
+
"learning_rate": 9.990026691641963e-05,
|
| 5618 |
+
"loss": 43.3788,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.07925014878000397,
|
| 5623 |
+
"grad_norm": 1.8996487855911255,
|
| 5624 |
+
"learning_rate": 9.970080114606439e-05,
|
| 5625 |
+
"loss": 43.4716,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.0793493354493156,
|
| 5630 |
+
"grad_norm": 2.1742310523986816,
|
| 5631 |
+
"learning_rate": 9.950133656612421e-05,
|
| 5632 |
+
"loss": 43.3278,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.07944852211862725,
|
| 5637 |
+
"grad_norm": 2.1892952919006348,
|
| 5638 |
+
"learning_rate": 9.930187397020386e-05,
|
| 5639 |
+
"loss": 43.3183,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.0795477087879389,
|
| 5644 |
+
"grad_norm": 1.887717366218567,
|
| 5645 |
+
"learning_rate": 9.910241415190021e-05,
|
| 5646 |
+
"loss": 43.376,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.07964689545725055,
|
| 5651 |
+
"grad_norm": 2.2955996990203857,
|
| 5652 |
+
"learning_rate": 9.890295790479911e-05,
|
| 5653 |
+
"loss": 42.9281,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.07974608212656219,
|
| 5658 |
+
"grad_norm": 1.925470232963562,
|
| 5659 |
+
"learning_rate": 9.870350602247218e-05,
|
| 5660 |
+
"loss": 43.444,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.07984526879587384,
|
| 5665 |
+
"grad_norm": 1.6774808168411255,
|
| 5666 |
+
"learning_rate": 9.850405929847366e-05,
|
| 5667 |
+
"loss": 43.3017,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.07994445546518548,
|
| 5672 |
+
"grad_norm": 1.67777419090271,
|
| 5673 |
+
"learning_rate": 9.830461852633727e-05,
|
| 5674 |
+
"loss": 43.4312,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.08004364213449712,
|
| 5679 |
+
"grad_norm": 1.737864375114441,
|
| 5680 |
+
"learning_rate": 9.810518449957309e-05,
|
| 5681 |
+
"loss": 43.4589,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.08014282880380877,
|
| 5686 |
+
"grad_norm": 2.016724109649658,
|
| 5687 |
+
"learning_rate": 9.790575801166432e-05,
|
| 5688 |
+
"loss": 43.5872,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.08024201547312042,
|
| 5693 |
+
"grad_norm": 1.742182731628418,
|
| 5694 |
+
"learning_rate": 9.770633985606419e-05,
|
| 5695 |
+
"loss": 43.2205,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.08034120214243205,
|
| 5700 |
+
"grad_norm": 1.6217352151870728,
|
| 5701 |
+
"learning_rate": 9.750693082619273e-05,
|
| 5702 |
+
"loss": 43.1472,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.0804403888117437,
|
| 5707 |
+
"grad_norm": 1.8912020921707153,
|
| 5708 |
+
"learning_rate": 9.730753171543374e-05,
|
| 5709 |
+
"loss": 43.5462,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.08053957548105535,
|
| 5714 |
+
"grad_norm": 2.008028984069824,
|
| 5715 |
+
"learning_rate": 9.71081433171315e-05,
|
| 5716 |
+
"loss": 43.5617,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.08063876215036699,
|
| 5721 |
+
"grad_norm": 1.7227747440338135,
|
| 5722 |
+
"learning_rate": 9.690876642458764e-05,
|
| 5723 |
+
"loss": 43.4189,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.08073794881967863,
|
| 5728 |
+
"grad_norm": 1.9663054943084717,
|
| 5729 |
+
"learning_rate": 9.670940183105812e-05,
|
| 5730 |
+
"loss": 43.4703,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.08083713548899028,
|
| 5735 |
+
"grad_norm": 2.0552761554718018,
|
| 5736 |
+
"learning_rate": 9.651005032974994e-05,
|
| 5737 |
+
"loss": 43.2265,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.08093632215830192,
|
| 5742 |
+
"grad_norm": 1.7139379978179932,
|
| 5743 |
+
"learning_rate": 9.631071271381785e-05,
|
| 5744 |
+
"loss": 43.575,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.08103550882761357,
|
| 5749 |
+
"grad_norm": 1.73064386844635,
|
| 5750 |
+
"learning_rate": 9.611138977636153e-05,
|
| 5751 |
+
"loss": 43.5118,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.08113469549692522,
|
| 5756 |
+
"grad_norm": 1.8253397941589355,
|
| 5757 |
+
"learning_rate": 9.591208231042218e-05,
|
| 5758 |
+
"loss": 43.1822,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.08123388216623685,
|
| 5763 |
+
"grad_norm": 1.912844181060791,
|
| 5764 |
+
"learning_rate": 9.571279110897952e-05,
|
| 5765 |
+
"loss": 43.384,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.0813330688355485,
|
| 5770 |
+
"grad_norm": 2.275552749633789,
|
| 5771 |
+
"learning_rate": 9.551351696494854e-05,
|
| 5772 |
+
"loss": 43.3521,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.08143225550486015,
|
| 5777 |
+
"grad_norm": 2.37941312789917,
|
| 5778 |
+
"learning_rate": 9.531426067117623e-05,
|
| 5779 |
+
"loss": 43.3171,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.0815314421741718,
|
| 5784 |
+
"grad_norm": 1.811157464981079,
|
| 5785 |
+
"learning_rate": 9.511502302043868e-05,
|
| 5786 |
+
"loss": 43.2281,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.08163062884348343,
|
| 5791 |
+
"grad_norm": 2.058664083480835,
|
| 5792 |
+
"learning_rate": 9.491580480543784e-05,
|
| 5793 |
+
"loss": 43.1878,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.08172981551279508,
|
| 5798 |
+
"grad_norm": 1.7571479082107544,
|
| 5799 |
+
"learning_rate": 9.471660681879825e-05,
|
| 5800 |
+
"loss": 43.5305,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.08182900218210673,
|
| 5805 |
+
"grad_norm": 2.2421343326568604,
|
| 5806 |
+
"learning_rate": 9.451742985306398e-05,
|
| 5807 |
+
"loss": 43.1806,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.08192818885141837,
|
| 5812 |
+
"grad_norm": 2.145615339279175,
|
| 5813 |
+
"learning_rate": 9.431827470069558e-05,
|
| 5814 |
+
"loss": 42.9589,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.08202737552073001,
|
| 5819 |
+
"grad_norm": 1.8979140520095825,
|
| 5820 |
+
"learning_rate": 9.411914215406654e-05,
|
| 5821 |
+
"loss": 43.2689,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.08212656219004166,
|
| 5826 |
+
"grad_norm": 1.7577365636825562,
|
| 5827 |
+
"learning_rate": 9.39200330054607e-05,
|
| 5828 |
+
"loss": 43.523,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.0822257488593533,
|
| 5833 |
+
"grad_norm": 1.659989356994629,
|
| 5834 |
+
"learning_rate": 9.372094804706867e-05,
|
| 5835 |
+
"loss": 43.6545,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.08232493552866495,
|
| 5840 |
+
"grad_norm": 2.152495861053467,
|
| 5841 |
+
"learning_rate": 9.352188807098481e-05,
|
| 5842 |
+
"loss": 43.4871,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.0824241221979766,
|
| 5847 |
+
"grad_norm": 1.8303964138031006,
|
| 5848 |
+
"learning_rate": 9.332285386920416e-05,
|
| 5849 |
+
"loss": 43.4556,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.08252330886728823,
|
| 5854 |
+
"grad_norm": 1.996830701828003,
|
| 5855 |
+
"learning_rate": 9.312384623361909e-05,
|
| 5856 |
+
"loss": 43.0785,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.08262249553659988,
|
| 5861 |
+
"grad_norm": 1.5665048360824585,
|
| 5862 |
+
"learning_rate": 9.292486595601637e-05,
|
| 5863 |
+
"loss": 43.5144,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.08272168220591153,
|
| 5868 |
+
"grad_norm": 2.127319097518921,
|
| 5869 |
+
"learning_rate": 9.272591382807391e-05,
|
| 5870 |
+
"loss": 43.462,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.08282086887522316,
|
| 5875 |
+
"grad_norm": 2.1389379501342773,
|
| 5876 |
+
"learning_rate": 9.252699064135758e-05,
|
| 5877 |
+
"loss": 43.3599,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.08292005554453481,
|
| 5882 |
+
"grad_norm": 2.174877643585205,
|
| 5883 |
+
"learning_rate": 9.232809718731814e-05,
|
| 5884 |
+
"loss": 43.3102,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.08301924221384646,
|
| 5889 |
+
"grad_norm": 1.9296512603759766,
|
| 5890 |
+
"learning_rate": 9.212923425728806e-05,
|
| 5891 |
+
"loss": 43.4743,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.08311842888315811,
|
| 5896 |
+
"grad_norm": 1.6677355766296387,
|
| 5897 |
+
"learning_rate": 9.193040264247829e-05,
|
| 5898 |
+
"loss": 43.4941,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.08321761555246975,
|
| 5903 |
+
"grad_norm": 1.9605714082717896,
|
| 5904 |
+
"learning_rate": 9.173160313397528e-05,
|
| 5905 |
+
"loss": 43.3157,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.0833168022217814,
|
| 5910 |
+
"grad_norm": 1.9276679754257202,
|
| 5911 |
+
"learning_rate": 9.153283652273768e-05,
|
| 5912 |
+
"loss": 43.3372,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.08341598889109304,
|
| 5917 |
+
"grad_norm": 1.7903660535812378,
|
| 5918 |
+
"learning_rate": 9.13341035995933e-05,
|
| 5919 |
+
"loss": 43.3393,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.08351517556040468,
|
| 5924 |
+
"grad_norm": 2.195932388305664,
|
| 5925 |
+
"learning_rate": 9.113540515523587e-05,
|
| 5926 |
+
"loss": 43.3919,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.08361436222971633,
|
| 5931 |
+
"grad_norm": 1.932118535041809,
|
| 5932 |
+
"learning_rate": 9.093674198022201e-05,
|
| 5933 |
+
"loss": 43.6243,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.08371354889902798,
|
| 5938 |
+
"grad_norm": 1.9004766941070557,
|
| 5939 |
+
"learning_rate": 9.073811486496788e-05,
|
| 5940 |
+
"loss": 43.429,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.08381273556833961,
|
| 5945 |
+
"grad_norm": 2.0523078441619873,
|
| 5946 |
+
"learning_rate": 9.05395245997463e-05,
|
| 5947 |
+
"loss": 43.3915,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.08391192223765126,
|
| 5952 |
+
"grad_norm": 1.7547565698623657,
|
| 5953 |
+
"learning_rate": 9.034097197468342e-05,
|
| 5954 |
+
"loss": 43.6595,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.08401110890696291,
|
| 5959 |
+
"grad_norm": 2.1675822734832764,
|
| 5960 |
+
"learning_rate": 9.014245777975565e-05,
|
| 5961 |
+
"loss": 43.3933,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.08411029557627454,
|
| 5966 |
+
"grad_norm": 1.886845350265503,
|
| 5967 |
+
"learning_rate": 8.994398280478652e-05,
|
| 5968 |
+
"loss": 43.4644,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.08420948224558619,
|
| 5973 |
+
"grad_norm": 1.8034923076629639,
|
| 5974 |
+
"learning_rate": 8.974554783944342e-05,
|
| 5975 |
+
"loss": 43.1188,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.08430866891489784,
|
| 5980 |
+
"grad_norm": 1.7743672132492065,
|
| 5981 |
+
"learning_rate": 8.954715367323468e-05,
|
| 5982 |
+
"loss": 43.4643,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.08440785558420948,
|
| 5987 |
+
"grad_norm": 1.9661450386047363,
|
| 5988 |
+
"learning_rate": 8.93488010955062e-05,
|
| 5989 |
+
"loss": 43.5252,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.08450704225352113,
|
| 5994 |
+
"grad_norm": 1.9290202856063843,
|
| 5995 |
+
"learning_rate": 8.91504908954385e-05,
|
| 5996 |
+
"loss": 43.4773,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.08460622892283277,
|
| 6001 |
+
"grad_norm": 1.6719213724136353,
|
| 6002 |
+
"learning_rate": 8.89522238620434e-05,
|
| 6003 |
+
"loss": 43.3466,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.08470541559214441,
|
| 6008 |
+
"grad_norm": 2.064358949661255,
|
| 6009 |
+
"learning_rate": 8.875400078416113e-05,
|
| 6010 |
+
"loss": 43.3472,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.08480460226145606,
|
| 6015 |
+
"grad_norm": 2.1467957496643066,
|
| 6016 |
+
"learning_rate": 8.855582245045683e-05,
|
| 6017 |
+
"loss": 43.0823,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.0849037889307677,
|
| 6022 |
+
"grad_norm": 1.9114947319030762,
|
| 6023 |
+
"learning_rate": 8.835768964941773e-05,
|
| 6024 |
+
"loss": 43.2485,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.08500297560007936,
|
| 6029 |
+
"grad_norm": 1.9376516342163086,
|
| 6030 |
+
"learning_rate": 8.81596031693499e-05,
|
| 6031 |
+
"loss": 43.4138,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.08510216226939099,
|
| 6036 |
+
"grad_norm": 1.9953455924987793,
|
| 6037 |
+
"learning_rate": 8.796156379837513e-05,
|
| 6038 |
+
"loss": 42.9396,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.08520134893870264,
|
| 6043 |
+
"grad_norm": 2.0541064739227295,
|
| 6044 |
+
"learning_rate": 8.776357232442778e-05,
|
| 6045 |
+
"loss": 43.2261,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.08530053560801429,
|
| 6050 |
+
"grad_norm": 2.20963716506958,
|
| 6051 |
+
"learning_rate": 8.756562953525152e-05,
|
| 6052 |
+
"loss": 43.3041,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.08539972227732592,
|
| 6057 |
+
"grad_norm": 1.735040307044983,
|
| 6058 |
+
"learning_rate": 8.736773621839645e-05,
|
| 6059 |
+
"loss": 43.1483,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.08549890894663757,
|
| 6064 |
+
"grad_norm": 1.861799955368042,
|
| 6065 |
+
"learning_rate": 8.716989316121578e-05,
|
| 6066 |
+
"loss": 43.4874,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.08559809561594922,
|
| 6071 |
+
"grad_norm": 1.737962007522583,
|
| 6072 |
+
"learning_rate": 8.697210115086284e-05,
|
| 6073 |
+
"loss": 43.541,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.08569728228526086,
|
| 6078 |
+
"grad_norm": 2.2596657276153564,
|
| 6079 |
+
"learning_rate": 8.677436097428775e-05,
|
| 6080 |
+
"loss": 43.1737,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.0857964689545725,
|
| 6085 |
+
"grad_norm": 1.6303043365478516,
|
| 6086 |
+
"learning_rate": 8.657667341823448e-05,
|
| 6087 |
+
"loss": 43.3128,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.08589565562388415,
|
| 6092 |
+
"grad_norm": 2.3954904079437256,
|
| 6093 |
+
"learning_rate": 8.637903926923751e-05,
|
| 6094 |
+
"loss": 43.2482,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.08599484229319579,
|
| 6099 |
+
"grad_norm": 1.8219658136367798,
|
| 6100 |
+
"learning_rate": 8.618145931361902e-05,
|
| 6101 |
+
"loss": 43.2866,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.08609402896250744,
|
| 6106 |
+
"grad_norm": 2.017775535583496,
|
| 6107 |
+
"learning_rate": 8.598393433748541e-05,
|
| 6108 |
+
"loss": 43.3168,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.08619321563181909,
|
| 6113 |
+
"grad_norm": 2.4712929725646973,
|
| 6114 |
+
"learning_rate": 8.578646512672446e-05,
|
| 6115 |
+
"loss": 43.2597,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.08629240230113072,
|
| 6120 |
+
"grad_norm": 2.1750926971435547,
|
| 6121 |
+
"learning_rate": 8.558905246700201e-05,
|
| 6122 |
+
"loss": 43.5202,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.08639158897044237,
|
| 6127 |
+
"grad_norm": 2.4658377170562744,
|
| 6128 |
+
"learning_rate": 8.539169714375885e-05,
|
| 6129 |
+
"loss": 43.1393,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.08649077563975402,
|
| 6134 |
+
"grad_norm": 2.1692118644714355,
|
| 6135 |
+
"learning_rate": 8.519439994220775e-05,
|
| 6136 |
+
"loss": 43.0212,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.08658996230906567,
|
| 6141 |
+
"grad_norm": 2.0546863079071045,
|
| 6142 |
+
"learning_rate": 8.49971616473302e-05,
|
| 6143 |
+
"loss": 43.3599,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.0866891489783773,
|
| 6148 |
+
"grad_norm": 2.119807720184326,
|
| 6149 |
+
"learning_rate": 8.479998304387329e-05,
|
| 6150 |
+
"loss": 43.2814,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.08678833564768895,
|
| 6155 |
+
"grad_norm": 1.5778454542160034,
|
| 6156 |
+
"learning_rate": 8.460286491634663e-05,
|
| 6157 |
+
"loss": 43.3305,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.0868875223170006,
|
| 6162 |
+
"grad_norm": 1.8098981380462646,
|
| 6163 |
+
"learning_rate": 8.440580804901927e-05,
|
| 6164 |
+
"loss": 43.4353,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.08698670898631224,
|
| 6169 |
+
"grad_norm": 1.7996573448181152,
|
| 6170 |
+
"learning_rate": 8.420881322591642e-05,
|
| 6171 |
+
"loss": 43.1245,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.08708589565562389,
|
| 6176 |
+
"grad_norm": 1.94331693649292,
|
| 6177 |
+
"learning_rate": 8.401188123081653e-05,
|
| 6178 |
+
"loss": 43.387,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.08718508232493553,
|
| 6183 |
+
"grad_norm": 1.9635634422302246,
|
| 6184 |
+
"learning_rate": 8.381501284724801e-05,
|
| 6185 |
+
"loss": 43.2979,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.08728426899424717,
|
| 6190 |
+
"grad_norm": 1.813352108001709,
|
| 6191 |
+
"learning_rate": 8.361820885848624e-05,
|
| 6192 |
+
"loss": 43.1953,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.08738345566355882,
|
| 6197 |
+
"grad_norm": 1.7070103883743286,
|
| 6198 |
+
"learning_rate": 8.342147004755036e-05,
|
| 6199 |
+
"loss": 43.3006,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.08748264233287047,
|
| 6204 |
+
"grad_norm": 1.6127233505249023,
|
| 6205 |
+
"learning_rate": 8.32247971972002e-05,
|
| 6206 |
+
"loss": 43.3668,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.0875818290021821,
|
| 6211 |
+
"grad_norm": 1.8241714239120483,
|
| 6212 |
+
"learning_rate": 8.302819108993312e-05,
|
| 6213 |
+
"loss": 43.39,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.08768101567149375,
|
| 6218 |
+
"grad_norm": 2.063602924346924,
|
| 6219 |
+
"learning_rate": 8.283165250798099e-05,
|
| 6220 |
+
"loss": 43.2349,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.0877802023408054,
|
| 6225 |
+
"grad_norm": 1.9642107486724854,
|
| 6226 |
+
"learning_rate": 8.263518223330697e-05,
|
| 6227 |
+
"loss": 43.2532,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.08787938901011703,
|
| 6232 |
+
"grad_norm": 1.7925834655761719,
|
| 6233 |
+
"learning_rate": 8.243878104760249e-05,
|
| 6234 |
+
"loss": 43.2591,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.08797857567942868,
|
| 6239 |
+
"grad_norm": 2.3461196422576904,
|
| 6240 |
+
"learning_rate": 8.224244973228408e-05,
|
| 6241 |
+
"loss": 43.2511,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.08807776234874033,
|
| 6246 |
+
"grad_norm": 1.7713696956634521,
|
| 6247 |
+
"learning_rate": 8.204618906849021e-05,
|
| 6248 |
+
"loss": 43.084,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.08817694901805197,
|
| 6253 |
+
"grad_norm": 2.40630841255188,
|
| 6254 |
+
"learning_rate": 8.184999983707837e-05,
|
| 6255 |
+
"loss": 43.1874,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.08827613568736362,
|
| 6260 |
+
"grad_norm": 1.782840371131897,
|
| 6261 |
+
"learning_rate": 8.165388281862178e-05,
|
| 6262 |
+
"loss": 43.4851,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.08837532235667526,
|
| 6267 |
+
"grad_norm": 1.8510971069335938,
|
| 6268 |
+
"learning_rate": 8.145783879340634e-05,
|
| 6269 |
+
"loss": 43.2571,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.08847450902598691,
|
| 6274 |
+
"grad_norm": 1.772338628768921,
|
| 6275 |
+
"learning_rate": 8.126186854142752e-05,
|
| 6276 |
+
"loss": 43.5884,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.08857369569529855,
|
| 6281 |
+
"grad_norm": 2.018734931945801,
|
| 6282 |
+
"learning_rate": 8.10659728423874e-05,
|
| 6283 |
+
"loss": 42.9666,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.0886728823646102,
|
| 6288 |
+
"grad_norm": 2.0410776138305664,
|
| 6289 |
+
"learning_rate": 8.087015247569124e-05,
|
| 6290 |
+
"loss": 43.2727,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.08877206903392185,
|
| 6295 |
+
"grad_norm": 1.8451515436172485,
|
| 6296 |
+
"learning_rate": 8.067440822044469e-05,
|
| 6297 |
+
"loss": 43.2777,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.08887125570323348,
|
| 6302 |
+
"grad_norm": 1.7834467887878418,
|
| 6303 |
+
"learning_rate": 8.047874085545058e-05,
|
| 6304 |
+
"loss": 42.9418,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.08897044237254513,
|
| 6309 |
+
"grad_norm": 2.3698744773864746,
|
| 6310 |
+
"learning_rate": 8.028315115920576e-05,
|
| 6311 |
+
"loss": 43.3835,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.08906962904185678,
|
| 6316 |
+
"grad_norm": 2.0287153720855713,
|
| 6317 |
+
"learning_rate": 8.00876399098982e-05,
|
| 6318 |
+
"loss": 43.3926,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.08916881571116841,
|
| 6323 |
+
"grad_norm": 1.825823426246643,
|
| 6324 |
+
"learning_rate": 7.989220788540355e-05,
|
| 6325 |
+
"loss": 43.1946,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.08926800238048006,
|
| 6330 |
+
"grad_norm": 1.9353700876235962,
|
| 6331 |
+
"learning_rate": 7.96968558632824e-05,
|
| 6332 |
+
"loss": 43.1476,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.08936718904979171,
|
| 6337 |
+
"grad_norm": 1.78703773021698,
|
| 6338 |
+
"learning_rate": 7.950158462077697e-05,
|
| 6339 |
+
"loss": 43.2263,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.08946637571910335,
|
| 6344 |
+
"grad_norm": 1.9925075769424438,
|
| 6345 |
+
"learning_rate": 7.930639493480814e-05,
|
| 6346 |
+
"loss": 43.3162,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.089565562388415,
|
| 6351 |
+
"grad_norm": 1.8245471715927124,
|
| 6352 |
+
"learning_rate": 7.911128758197224e-05,
|
| 6353 |
+
"loss": 43.4107,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.08966474905772664,
|
| 6358 |
+
"grad_norm": 1.942383885383606,
|
| 6359 |
+
"learning_rate": 7.891626333853812e-05,
|
| 6360 |
+
"loss": 43.3753,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.08976393572703828,
|
| 6365 |
+
"grad_norm": 1.8528326749801636,
|
| 6366 |
+
"learning_rate": 7.872132298044382e-05,
|
| 6367 |
+
"loss": 43.3682,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.08986312239634993,
|
| 6372 |
+
"grad_norm": 2.216946840286255,
|
| 6373 |
+
"learning_rate": 7.852646728329368e-05,
|
| 6374 |
+
"loss": 43.195,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.08996230906566158,
|
| 6379 |
+
"grad_norm": 1.8748693466186523,
|
| 6380 |
+
"learning_rate": 7.833169702235531e-05,
|
| 6381 |
+
"loss": 43.4997,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.09006149573497323,
|
| 6386 |
+
"grad_norm": 1.8855562210083008,
|
| 6387 |
+
"learning_rate": 7.813701297255628e-05,
|
| 6388 |
+
"loss": 43.4364,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.09016068240428486,
|
| 6393 |
+
"grad_norm": 1.8406039476394653,
|
| 6394 |
+
"learning_rate": 7.794241590848123e-05,
|
| 6395 |
+
"loss": 43.2181,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.09025986907359651,
|
| 6400 |
+
"grad_norm": 2.068734645843506,
|
| 6401 |
+
"learning_rate": 7.774790660436858e-05,
|
| 6402 |
+
"loss": 43.4028,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.09035905574290816,
|
| 6407 |
+
"grad_norm": 2.1496853828430176,
|
| 6408 |
+
"learning_rate": 7.755348583410772e-05,
|
| 6409 |
+
"loss": 43.3901,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.0904582424122198,
|
| 6414 |
+
"grad_norm": 1.993266224861145,
|
| 6415 |
+
"learning_rate": 7.735915437123574e-05,
|
| 6416 |
+
"loss": 43.3823,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.09055742908153144,
|
| 6421 |
+
"grad_norm": 1.862154245376587,
|
| 6422 |
+
"learning_rate": 7.716491298893442e-05,
|
| 6423 |
+
"loss": 43.0743,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.09065661575084309,
|
| 6428 |
+
"grad_norm": 2.08268141746521,
|
| 6429 |
+
"learning_rate": 7.697076246002713e-05,
|
| 6430 |
+
"loss": 43.413,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.09075580242015473,
|
| 6435 |
+
"grad_norm": 1.8610597848892212,
|
| 6436 |
+
"learning_rate": 7.677670355697577e-05,
|
| 6437 |
+
"loss": 43.3784,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.09085498908946638,
|
| 6442 |
+
"grad_norm": 1.7491427659988403,
|
| 6443 |
+
"learning_rate": 7.658273705187761e-05,
|
| 6444 |
+
"loss": 43.4197,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.09095417575877802,
|
| 6449 |
+
"grad_norm": 1.8177448511123657,
|
| 6450 |
+
"learning_rate": 7.638886371646246e-05,
|
| 6451 |
+
"loss": 43.4021,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.09105336242808966,
|
| 6456 |
+
"grad_norm": 1.9328161478042603,
|
| 6457 |
+
"learning_rate": 7.619508432208931e-05,
|
| 6458 |
+
"loss": 43.1076,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.09115254909740131,
|
| 6463 |
+
"grad_norm": 1.773515224456787,
|
| 6464 |
+
"learning_rate": 7.600139963974341e-05,
|
| 6465 |
+
"loss": 43.2534,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.09125173576671296,
|
| 6470 |
+
"grad_norm": 1.7070053815841675,
|
| 6471 |
+
"learning_rate": 7.580781044003324e-05,
|
| 6472 |
+
"loss": 43.316,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.09135092243602459,
|
| 6477 |
+
"grad_norm": 1.9303302764892578,
|
| 6478 |
+
"learning_rate": 7.561431749318731e-05,
|
| 6479 |
+
"loss": 42.9603,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.09145010910533624,
|
| 6484 |
+
"grad_norm": 1.822993516921997,
|
| 6485 |
+
"learning_rate": 7.542092156905123e-05,
|
| 6486 |
+
"loss": 43.3826,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.09154929577464789,
|
| 6491 |
+
"grad_norm": 1.8518813848495483,
|
| 6492 |
+
"learning_rate": 7.522762343708454e-05,
|
| 6493 |
+
"loss": 43.4774,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.09164848244395953,
|
| 6498 |
+
"grad_norm": 1.886132836341858,
|
| 6499 |
+
"learning_rate": 7.503442386635774e-05,
|
| 6500 |
+
"loss": 43.3404,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.09174766911327117,
|
| 6505 |
+
"grad_norm": 2.080753803253174,
|
| 6506 |
+
"learning_rate": 7.484132362554915e-05,
|
| 6507 |
+
"loss": 43.3287,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.09184685578258282,
|
| 6512 |
+
"grad_norm": 1.8040423393249512,
|
| 6513 |
+
"learning_rate": 7.464832348294195e-05,
|
| 6514 |
+
"loss": 43.3848,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.09194604245189447,
|
| 6519 |
+
"grad_norm": 1.7399113178253174,
|
| 6520 |
+
"learning_rate": 7.445542420642097e-05,
|
| 6521 |
+
"loss": 43.41,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.0920452291212061,
|
| 6526 |
+
"grad_norm": 1.636152982711792,
|
| 6527 |
+
"learning_rate": 7.426262656346978e-05,
|
| 6528 |
+
"loss": 43.4037,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.09214441579051776,
|
| 6533 |
+
"grad_norm": 1.8403246402740479,
|
| 6534 |
+
"learning_rate": 7.406993132116759e-05,
|
| 6535 |
+
"loss": 43.3664,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.0922436024598294,
|
| 6540 |
+
"grad_norm": 1.7476787567138672,
|
| 6541 |
+
"learning_rate": 7.387733924618617e-05,
|
| 6542 |
+
"loss": 43.4288,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.09234278912914104,
|
| 6547 |
+
"grad_norm": 1.964263916015625,
|
| 6548 |
+
"learning_rate": 7.368485110478685e-05,
|
| 6549 |
+
"loss": 43.2634,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.09244197579845269,
|
| 6554 |
+
"grad_norm": 2.1043291091918945,
|
| 6555 |
+
"learning_rate": 7.349246766281741e-05,
|
| 6556 |
+
"loss": 43.4181,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.09254116246776434,
|
| 6561 |
+
"grad_norm": 2.108569860458374,
|
| 6562 |
+
"learning_rate": 7.330018968570907e-05,
|
| 6563 |
+
"loss": 43.3852,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.09264034913707597,
|
| 6568 |
+
"grad_norm": 2.0387933254241943,
|
| 6569 |
+
"learning_rate": 7.310801793847344e-05,
|
| 6570 |
+
"loss": 43.065,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.09273953580638762,
|
| 6575 |
+
"grad_norm": 1.9925363063812256,
|
| 6576 |
+
"learning_rate": 7.291595318569951e-05,
|
| 6577 |
+
"loss": 43.3083,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.09283872247569927,
|
| 6582 |
+
"grad_norm": 1.754214882850647,
|
| 6583 |
+
"learning_rate": 7.272399619155048e-05,
|
| 6584 |
+
"loss": 43.5388,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.0929379091450109,
|
| 6589 |
+
"grad_norm": 1.985870361328125,
|
| 6590 |
+
"learning_rate": 7.2532147719761e-05,
|
| 6591 |
+
"loss": 43.2456,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.09303709581432255,
|
| 6596 |
+
"grad_norm": 2.0967764854431152,
|
| 6597 |
+
"learning_rate": 7.234040853363371e-05,
|
| 6598 |
+
"loss": 43.2949,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.0931362824836342,
|
| 6603 |
+
"grad_norm": 2.0468029975891113,
|
| 6604 |
+
"learning_rate": 7.214877939603659e-05,
|
| 6605 |
+
"loss": 43.4696,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.09323546915294584,
|
| 6610 |
+
"grad_norm": 1.9126473665237427,
|
| 6611 |
+
"learning_rate": 7.195726106939974e-05,
|
| 6612 |
+
"loss": 43.5634,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.09333465582225749,
|
| 6617 |
+
"grad_norm": 2.130176544189453,
|
| 6618 |
+
"learning_rate": 7.176585431571235e-05,
|
| 6619 |
+
"loss": 43.058,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.09343384249156914,
|
| 6624 |
+
"grad_norm": 2.085319995880127,
|
| 6625 |
+
"learning_rate": 7.157455989651976e-05,
|
| 6626 |
+
"loss": 43.3287,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.09353302916088078,
|
| 6631 |
+
"grad_norm": 1.8544113636016846,
|
| 6632 |
+
"learning_rate": 7.138337857292034e-05,
|
| 6633 |
+
"loss": 43.3071,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.09363221583019242,
|
| 6638 |
+
"grad_norm": 1.7780606746673584,
|
| 6639 |
+
"learning_rate": 7.119231110556242e-05,
|
| 6640 |
+
"loss": 43.4521,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.09373140249950407,
|
| 6645 |
+
"grad_norm": 1.8853678703308105,
|
| 6646 |
+
"learning_rate": 7.100135825464139e-05,
|
| 6647 |
+
"loss": 43.394,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.09383058916881572,
|
| 6652 |
+
"grad_norm": 2.073620557785034,
|
| 6653 |
+
"learning_rate": 7.081052077989667e-05,
|
| 6654 |
+
"loss": 42.9639,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.09392977583812735,
|
| 6659 |
+
"grad_norm": 2.062309980392456,
|
| 6660 |
+
"learning_rate": 7.061979944060856e-05,
|
| 6661 |
+
"loss": 43.4094,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.094028962507439,
|
| 6666 |
+
"grad_norm": 2.0947256088256836,
|
| 6667 |
+
"learning_rate": 7.042919499559537e-05,
|
| 6668 |
+
"loss": 43.3735,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.09412814917675065,
|
| 6673 |
+
"grad_norm": 2.1636717319488525,
|
| 6674 |
+
"learning_rate": 7.023870820321017e-05,
|
| 6675 |
+
"loss": 43.5101,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.09422733584606229,
|
| 6680 |
+
"grad_norm": 1.8934404850006104,
|
| 6681 |
+
"learning_rate": 7.004833982133808e-05,
|
| 6682 |
+
"loss": 43.4439,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.09432652251537393,
|
| 6687 |
+
"grad_norm": 1.883296012878418,
|
| 6688 |
+
"learning_rate": 6.98580906073931e-05,
|
| 6689 |
+
"loss": 43.1449,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.09442570918468558,
|
| 6694 |
+
"grad_norm": 2.0616371631622314,
|
| 6695 |
+
"learning_rate": 6.966796131831501e-05,
|
| 6696 |
+
"loss": 43.2397,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.09452489585399722,
|
| 6701 |
+
"grad_norm": 2.622436046600342,
|
| 6702 |
+
"learning_rate": 6.947795271056651e-05,
|
| 6703 |
+
"loss": 43.0966,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.09462408252330887,
|
| 6708 |
+
"grad_norm": 2.2651469707489014,
|
| 6709 |
+
"learning_rate": 6.928806554013016e-05,
|
| 6710 |
+
"loss": 42.9688,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.09472326919262052,
|
| 6715 |
+
"grad_norm": 1.7404958009719849,
|
| 6716 |
+
"learning_rate": 6.909830056250527e-05,
|
| 6717 |
+
"loss": 43.4282,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.09482245586193215,
|
| 6722 |
+
"grad_norm": 1.9390512704849243,
|
| 6723 |
+
"learning_rate": 6.890865853270509e-05,
|
| 6724 |
+
"loss": 43.2355,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.0949216425312438,
|
| 6729 |
+
"grad_norm": 1.7587411403656006,
|
| 6730 |
+
"learning_rate": 6.871914020525363e-05,
|
| 6731 |
+
"loss": 43.5188,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.09502082920055545,
|
| 6736 |
+
"grad_norm": 1.802949070930481,
|
| 6737 |
+
"learning_rate": 6.85297463341828e-05,
|
| 6738 |
+
"loss": 43.4377,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.09512001586986708,
|
| 6743 |
+
"grad_norm": 2.380028486251831,
|
| 6744 |
+
"learning_rate": 6.834047767302929e-05,
|
| 6745 |
+
"loss": 43.4229,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.09521920253917873,
|
| 6750 |
+
"grad_norm": 2.1051738262176514,
|
| 6751 |
+
"learning_rate": 6.815133497483157e-05,
|
| 6752 |
+
"loss": 43.2247,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.09531838920849038,
|
| 6757 |
+
"grad_norm": 2.068763017654419,
|
| 6758 |
+
"learning_rate": 6.796231899212704e-05,
|
| 6759 |
+
"loss": 43.0241,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.09541757587780203,
|
| 6764 |
+
"grad_norm": 1.9503624439239502,
|
| 6765 |
+
"learning_rate": 6.77734304769489e-05,
|
| 6766 |
+
"loss": 43.1053,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.09551676254711366,
|
| 6771 |
+
"grad_norm": 2.1576602458953857,
|
| 6772 |
+
"learning_rate": 6.75846701808232e-05,
|
| 6773 |
+
"loss": 43.2077,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.09561594921642531,
|
| 6778 |
+
"grad_norm": 2.2297677993774414,
|
| 6779 |
+
"learning_rate": 6.739603885476582e-05,
|
| 6780 |
+
"loss": 43.4295,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.09571513588573696,
|
| 6785 |
+
"grad_norm": 1.9149441719055176,
|
| 6786 |
+
"learning_rate": 6.720753724927958e-05,
|
| 6787 |
+
"loss": 43.3554,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.0958143225550486,
|
| 6792 |
+
"grad_norm": 1.9192272424697876,
|
| 6793 |
+
"learning_rate": 6.701916611435109e-05,
|
| 6794 |
+
"loss": 43.2641,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.09591350922436025,
|
| 6799 |
+
"grad_norm": 1.752898097038269,
|
| 6800 |
+
"learning_rate": 6.683092619944796e-05,
|
| 6801 |
+
"loss": 43.2443,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.0960126958936719,
|
| 6806 |
+
"grad_norm": 1.7976934909820557,
|
| 6807 |
+
"learning_rate": 6.664281825351562e-05,
|
| 6808 |
+
"loss": 43.1963,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.09611188256298353,
|
| 6813 |
+
"grad_norm": 1.8018180131912231,
|
| 6814 |
+
"learning_rate": 6.64548430249745e-05,
|
| 6815 |
+
"loss": 43.321,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.09621106923229518,
|
| 6820 |
+
"grad_norm": 2.098780870437622,
|
| 6821 |
+
"learning_rate": 6.626700126171702e-05,
|
| 6822 |
+
"loss": 43.426,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.09631025590160683,
|
| 6827 |
+
"grad_norm": 1.8992630243301392,
|
| 6828 |
+
"learning_rate": 6.607929371110448e-05,
|
| 6829 |
+
"loss": 42.978,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.09640944257091846,
|
| 6834 |
+
"grad_norm": 1.8057256937026978,
|
| 6835 |
+
"learning_rate": 6.58917211199643e-05,
|
| 6836 |
+
"loss": 43.2817,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.09650862924023011,
|
| 6841 |
+
"grad_norm": 1.7181857824325562,
|
| 6842 |
+
"learning_rate": 6.570428423458687e-05,
|
| 6843 |
+
"loss": 43.4864,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.09660781590954176,
|
| 6848 |
+
"grad_norm": 1.844660758972168,
|
| 6849 |
+
"learning_rate": 6.551698380072266e-05,
|
| 6850 |
+
"loss": 43.1988,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.0967070025788534,
|
| 6855 |
+
"grad_norm": 1.6959925889968872,
|
| 6856 |
+
"learning_rate": 6.532982056357928e-05,
|
| 6857 |
+
"loss": 43.2842,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.09680618924816504,
|
| 6862 |
+
"grad_norm": 2.1474428176879883,
|
| 6863 |
+
"learning_rate": 6.51427952678185e-05,
|
| 6864 |
+
"loss": 42.8695,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.0969053759174767,
|
| 6869 |
+
"grad_norm": 1.9975131750106812,
|
| 6870 |
+
"learning_rate": 6.495590865755315e-05,
|
| 6871 |
+
"loss": 43.0511,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.09700456258678833,
|
| 6876 |
+
"grad_norm": 2.1720874309539795,
|
| 6877 |
+
"learning_rate": 6.476916147634437e-05,
|
| 6878 |
+
"loss": 42.9337,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.09710374925609998,
|
| 6883 |
+
"grad_norm": 1.5775563716888428,
|
| 6884 |
+
"learning_rate": 6.458255446719854e-05,
|
| 6885 |
+
"loss": 43.4077,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.09720293592541163,
|
| 6890 |
+
"grad_norm": 1.7421826124191284,
|
| 6891 |
+
"learning_rate": 6.439608837256432e-05,
|
| 6892 |
+
"loss": 43.3882,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.09730212259472328,
|
| 6897 |
+
"grad_norm": 1.8827160596847534,
|
| 6898 |
+
"learning_rate": 6.420976393432977e-05,
|
| 6899 |
+
"loss": 43.192,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.09740130926403491,
|
| 6904 |
+
"grad_norm": 2.0593631267547607,
|
| 6905 |
+
"learning_rate": 6.402358189381934e-05,
|
| 6906 |
+
"loss": 43.2,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.09750049593334656,
|
| 6911 |
+
"grad_norm": 2.002188205718994,
|
| 6912 |
+
"learning_rate": 6.383754299179079e-05,
|
| 6913 |
+
"loss": 43.2412,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.09759968260265821,
|
| 6918 |
+
"grad_norm": 1.59535551071167,
|
| 6919 |
+
"learning_rate": 6.365164796843252e-05,
|
| 6920 |
+
"loss": 43.5838,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.09769886927196984,
|
| 6925 |
+
"grad_norm": 2.0612454414367676,
|
| 6926 |
+
"learning_rate": 6.34658975633605e-05,
|
| 6927 |
+
"loss": 42.9227,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.09779805594128149,
|
| 6932 |
+
"grad_norm": 1.842847466468811,
|
| 6933 |
+
"learning_rate": 6.328029251561521e-05,
|
| 6934 |
+
"loss": 43.4525,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.09789724261059314,
|
| 6939 |
+
"grad_norm": 1.7609597444534302,
|
| 6940 |
+
"learning_rate": 6.30948335636589e-05,
|
| 6941 |
+
"loss": 43.1777,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.09799642927990478,
|
| 6946 |
+
"grad_norm": 2.0873639583587646,
|
| 6947 |
+
"learning_rate": 6.290952144537241e-05,
|
| 6948 |
+
"loss": 43.4048,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.09809561594921642,
|
| 6953 |
+
"grad_norm": 2.211897134780884,
|
| 6954 |
+
"learning_rate": 6.27243568980525e-05,
|
| 6955 |
+
"loss": 43.2654,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.09819480261852807,
|
| 6960 |
+
"grad_norm": 1.9050241708755493,
|
| 6961 |
+
"learning_rate": 6.25393406584088e-05,
|
| 6962 |
+
"loss": 43.2461,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.09829398928783971,
|
| 6967 |
+
"grad_norm": 1.964715838432312,
|
| 6968 |
+
"learning_rate": 6.23544734625608e-05,
|
| 6969 |
+
"loss": 43.3799,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.09839317595715136,
|
| 6974 |
+
"grad_norm": 2.0911576747894287,
|
| 6975 |
+
"learning_rate": 6.216975604603503e-05,
|
| 6976 |
+
"loss": 42.9,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.098492362626463,
|
| 6981 |
+
"grad_norm": 2.2503297328948975,
|
| 6982 |
+
"learning_rate": 6.198518914376212e-05,
|
| 6983 |
+
"loss": 43.4324,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.09859154929577464,
|
| 6988 |
+
"grad_norm": 1.845924735069275,
|
| 6989 |
+
"learning_rate": 6.180077349007376e-05,
|
| 6990 |
+
"loss": 43.5936,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.09869073596508629,
|
| 6995 |
+
"grad_norm": 2.30470609664917,
|
| 6996 |
+
"learning_rate": 6.161650981869998e-05,
|
| 6997 |
+
"loss": 43.4711,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.09878992263439794,
|
| 7002 |
+
"grad_norm": 2.0436346530914307,
|
| 7003 |
+
"learning_rate": 6.143239886276606e-05,
|
| 7004 |
+
"loss": 43.1616,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.09888910930370959,
|
| 7009 |
+
"grad_norm": 2.165714979171753,
|
| 7010 |
+
"learning_rate": 6.12484413547897e-05,
|
| 7011 |
+
"loss": 43.3748,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.09898829597302122,
|
| 7016 |
+
"grad_norm": 1.9041638374328613,
|
| 7017 |
+
"learning_rate": 6.10646380266781e-05,
|
| 7018 |
+
"loss": 43.108,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.09908748264233287,
|
| 7023 |
+
"grad_norm": 2.148827075958252,
|
| 7024 |
+
"learning_rate": 6.0880989609724936e-05,
|
| 7025 |
+
"loss": 43.2891,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.09918666931164452,
|
| 7030 |
+
"grad_norm": 2.6612515449523926,
|
| 7031 |
+
"learning_rate": 6.069749683460765e-05,
|
| 7032 |
+
"loss": 43.5092,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.09928585598095616,
|
| 7037 |
+
"grad_norm": 1.887052059173584,
|
| 7038 |
+
"learning_rate": 6.051416043138438e-05,
|
| 7039 |
+
"loss": 43.315,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.0993850426502678,
|
| 7044 |
+
"grad_norm": 1.5838396549224854,
|
| 7045 |
+
"learning_rate": 6.0330981129491134e-05,
|
| 7046 |
+
"loss": 43.4339,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.09948422931957945,
|
| 7051 |
+
"grad_norm": 1.859200119972229,
|
| 7052 |
+
"learning_rate": 6.014795965773884e-05,
|
| 7053 |
+
"loss": 43.3558,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.09958341598889109,
|
| 7058 |
+
"grad_norm": 1.7264127731323242,
|
| 7059 |
+
"learning_rate": 5.9965096744310526e-05,
|
| 7060 |
+
"loss": 43.4768,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.09968260265820274,
|
| 7065 |
+
"grad_norm": 1.9149467945098877,
|
| 7066 |
+
"learning_rate": 5.978239311675826e-05,
|
| 7067 |
+
"loss": 43.5295,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.09978178932751439,
|
| 7072 |
+
"grad_norm": 1.9356281757354736,
|
| 7073 |
+
"learning_rate": 5.9599849502000485e-05,
|
| 7074 |
+
"loss": 43.241,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.09988097599682602,
|
| 7079 |
+
"grad_norm": 1.800052523612976,
|
| 7080 |
+
"learning_rate": 5.941746662631893e-05,
|
| 7081 |
+
"loss": 43.4511,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.09998016266613767,
|
| 7086 |
+
"grad_norm": 1.773478627204895,
|
| 7087 |
+
"learning_rate": 5.923524521535582e-05,
|
| 7088 |
+
"loss": 43.336,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.10007934933544932,
|
| 7093 |
+
"grad_norm": 2.0657424926757812,
|
| 7094 |
+
"learning_rate": 5.9053185994110974e-05,
|
| 7095 |
+
"loss": 43.2339,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.10017853600476095,
|
| 7100 |
+
"grad_norm": 2.2843918800354004,
|
| 7101 |
+
"learning_rate": 5.887128968693887e-05,
|
| 7102 |
+
"loss": 43.5656,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.1002777226740726,
|
| 7107 |
+
"grad_norm": 2.3814151287078857,
|
| 7108 |
+
"learning_rate": 5.868955701754584e-05,
|
| 7109 |
+
"loss": 43.1899,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.10037690934338425,
|
| 7114 |
+
"grad_norm": 2.003471612930298,
|
| 7115 |
+
"learning_rate": 5.8507988708987146e-05,
|
| 7116 |
+
"loss": 43.2224,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.10047609601269589,
|
| 7121 |
+
"grad_norm": 1.667802095413208,
|
| 7122 |
+
"learning_rate": 5.832658548366412e-05,
|
| 7123 |
+
"loss": 43.3213,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.10057528268200754,
|
| 7128 |
+
"grad_norm": 2.0781381130218506,
|
| 7129 |
+
"learning_rate": 5.8145348063321295e-05,
|
| 7130 |
+
"loss": 43.2473,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.10067446935131918,
|
| 7135 |
+
"grad_norm": 1.962714433670044,
|
| 7136 |
+
"learning_rate": 5.796427716904347e-05,
|
| 7137 |
+
"loss": 43.3516,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.10077365602063083,
|
| 7142 |
+
"grad_norm": 1.6675232648849487,
|
| 7143 |
+
"learning_rate": 5.7783373521252894e-05,
|
| 7144 |
+
"loss": 43.2493,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.10087284268994247,
|
| 7149 |
+
"grad_norm": 1.6517949104309082,
|
| 7150 |
+
"learning_rate": 5.7602637839706476e-05,
|
| 7151 |
+
"loss": 43.4813,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.10097202935925412,
|
| 7156 |
+
"grad_norm": 1.887805700302124,
|
| 7157 |
+
"learning_rate": 5.7422070843492734e-05,
|
| 7158 |
+
"loss": 43.0628,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.10107121602856577,
|
| 7163 |
+
"grad_norm": 2.185135841369629,
|
| 7164 |
+
"learning_rate": 5.724167325102915e-05,
|
| 7165 |
+
"loss": 42.8941,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.1011704026978774,
|
| 7170 |
+
"grad_norm": 1.701960802078247,
|
| 7171 |
+
"learning_rate": 5.7061445780059074e-05,
|
| 7172 |
+
"loss": 43.4726,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.10126958936718905,
|
| 7177 |
+
"grad_norm": 1.8178409337997437,
|
| 7178 |
+
"learning_rate": 5.6881389147649176e-05,
|
| 7179 |
+
"loss": 43.1844,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.1013687760365007,
|
| 7184 |
+
"grad_norm": 2.3397371768951416,
|
| 7185 |
+
"learning_rate": 5.670150407018624e-05,
|
| 7186 |
+
"loss": 43.4218,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.10146796270581233,
|
| 7191 |
+
"grad_norm": 1.9263267517089844,
|
| 7192 |
+
"learning_rate": 5.6521791263374534e-05,
|
| 7193 |
+
"loss": 43.5893,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.10156714937512398,
|
| 7198 |
+
"grad_norm": 2.1690573692321777,
|
| 7199 |
+
"learning_rate": 5.634225144223302e-05,
|
| 7200 |
+
"loss": 42.8077,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.10166633604443563,
|
| 7205 |
+
"grad_norm": 1.6065139770507812,
|
| 7206 |
+
"learning_rate": 5.616288532109225e-05,
|
| 7207 |
+
"loss": 43.3785,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.10176552271374727,
|
| 7212 |
+
"grad_norm": 2.0147528648376465,
|
| 7213 |
+
"learning_rate": 5.598369361359184e-05,
|
| 7214 |
+
"loss": 43.3484,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.10186470938305892,
|
| 7219 |
+
"grad_norm": 2.2999625205993652,
|
| 7220 |
+
"learning_rate": 5.5804677032677354e-05,
|
| 7221 |
+
"loss": 43.3572,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.10196389605237056,
|
| 7226 |
+
"grad_norm": 1.7449270486831665,
|
| 7227 |
+
"learning_rate": 5.562583629059757e-05,
|
| 7228 |
+
"loss": 43.4006,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.1020630827216822,
|
| 7233 |
+
"grad_norm": 1.7739657163619995,
|
| 7234 |
+
"learning_rate": 5.54471720989018e-05,
|
| 7235 |
+
"loss": 43.3843,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.10216226939099385,
|
| 7240 |
+
"grad_norm": 1.581351399421692,
|
| 7241 |
+
"learning_rate": 5.526868516843673e-05,
|
| 7242 |
+
"loss": 43.4641,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.1022614560603055,
|
| 7247 |
+
"grad_norm": 1.8967163562774658,
|
| 7248 |
+
"learning_rate": 5.509037620934399e-05,
|
| 7249 |
+
"loss": 43.3712,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.10236064272961715,
|
| 7254 |
+
"grad_norm": 1.862441897392273,
|
| 7255 |
+
"learning_rate": 5.491224593105695e-05,
|
| 7256 |
+
"loss": 43.4708,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.10245982939892878,
|
| 7261 |
+
"grad_norm": 2.014580011367798,
|
| 7262 |
+
"learning_rate": 5.47342950422981e-05,
|
| 7263 |
+
"loss": 43.2605,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.10255901606824043,
|
| 7268 |
+
"grad_norm": 2.7726058959960938,
|
| 7269 |
+
"learning_rate": 5.4556524251076304e-05,
|
| 7270 |
+
"loss": 43.2066,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.10265820273755208,
|
| 7275 |
+
"grad_norm": 1.7870392799377441,
|
| 7276 |
+
"learning_rate": 5.43789342646837e-05,
|
| 7277 |
+
"loss": 43.245,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.10275738940686371,
|
| 7282 |
+
"grad_norm": 1.6118180751800537,
|
| 7283 |
+
"learning_rate": 5.420152578969326e-05,
|
| 7284 |
+
"loss": 43.3048,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.10285657607617536,
|
| 7289 |
+
"grad_norm": 1.8771209716796875,
|
| 7290 |
+
"learning_rate": 5.4024299531955645e-05,
|
| 7291 |
+
"loss": 43.3119,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.10295576274548701,
|
| 7296 |
+
"grad_norm": 1.8395864963531494,
|
| 7297 |
+
"learning_rate": 5.38472561965965e-05,
|
| 7298 |
+
"loss": 43.2732,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.10305494941479865,
|
| 7303 |
+
"grad_norm": 1.6536940336227417,
|
| 7304 |
+
"learning_rate": 5.3670396488013854e-05,
|
| 7305 |
+
"loss": 43.0749,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.1031541360841103,
|
| 7310 |
+
"grad_norm": 2.0966672897338867,
|
| 7311 |
+
"learning_rate": 5.349372110987496e-05,
|
| 7312 |
+
"loss": 43.0109,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.10325332275342194,
|
| 7317 |
+
"grad_norm": 1.9550343751907349,
|
| 7318 |
+
"learning_rate": 5.331723076511379e-05,
|
| 7319 |
+
"loss": 43.3826,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.10335250942273358,
|
| 7324 |
+
"grad_norm": 2.0896692276000977,
|
| 7325 |
+
"learning_rate": 5.3140926155928136e-05,
|
| 7326 |
+
"loss": 43.2346,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.10345169609204523,
|
| 7331 |
+
"grad_norm": 1.7114324569702148,
|
| 7332 |
+
"learning_rate": 5.2964807983776744e-05,
|
| 7333 |
+
"loss": 43.1971,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.10355088276135688,
|
| 7338 |
+
"grad_norm": 1.8694396018981934,
|
| 7339 |
+
"learning_rate": 5.278887694937662e-05,
|
| 7340 |
+
"loss": 43.2731,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.10365006943066851,
|
| 7345 |
+
"grad_norm": 1.9449472427368164,
|
| 7346 |
+
"learning_rate": 5.261313375270014e-05,
|
| 7347 |
+
"loss": 43.2565,
|
| 7348 |
+
"step": 1045
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.10374925609998016,
|
| 7352 |
+
"grad_norm": 2.180279493331909,
|
| 7353 |
+
"learning_rate": 5.243757909297247e-05,
|
| 7354 |
+
"loss": 43.3411,
|
| 7355 |
+
"step": 1046
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.10384844276929181,
|
| 7359 |
+
"grad_norm": 2.0333480834960938,
|
| 7360 |
+
"learning_rate": 5.226221366866859e-05,
|
| 7361 |
+
"loss": 43.0145,
|
| 7362 |
+
"step": 1047
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.10394762943860344,
|
| 7366 |
+
"grad_norm": 2.140620708465576,
|
| 7367 |
+
"learning_rate": 5.208703817751053e-05,
|
| 7368 |
+
"loss": 42.9723,
|
| 7369 |
+
"step": 1048
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.1040468161079151,
|
| 7373 |
+
"grad_norm": 1.8201910257339478,
|
| 7374 |
+
"learning_rate": 5.1912053316464694e-05,
|
| 7375 |
+
"loss": 43.3619,
|
| 7376 |
+
"step": 1049
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.10414600277722674,
|
| 7380 |
+
"grad_norm": 1.8055152893066406,
|
| 7381 |
+
"learning_rate": 5.1737259781738936e-05,
|
| 7382 |
+
"loss": 43.4053,
|
| 7383 |
+
"step": 1050
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.10424518944653839,
|
| 7387 |
+
"grad_norm": 2.1512773036956787,
|
| 7388 |
+
"learning_rate": 5.156265826877999e-05,
|
| 7389 |
+
"loss": 42.7926,
|
| 7390 |
+
"step": 1051
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.10434437611585003,
|
| 7394 |
+
"grad_norm": 1.7757370471954346,
|
| 7395 |
+
"learning_rate": 5.138824947227063e-05,
|
| 7396 |
+
"loss": 43.2739,
|
| 7397 |
+
"step": 1052
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.10444356278516168,
|
| 7401 |
+
"grad_norm": 1.9325311183929443,
|
| 7402 |
+
"learning_rate": 5.121403408612672e-05,
|
| 7403 |
+
"loss": 43.719,
|
| 7404 |
+
"step": 1053
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.10454274945447332,
|
| 7408 |
+
"grad_norm": 1.8399518728256226,
|
| 7409 |
+
"learning_rate": 5.1040012803494795e-05,
|
| 7410 |
+
"loss": 43.3349,
|
| 7411 |
+
"step": 1054
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.10464193612378496,
|
| 7415 |
+
"grad_norm": 1.9976588487625122,
|
| 7416 |
+
"learning_rate": 5.086618631674888e-05,
|
| 7417 |
+
"loss": 43.2846,
|
| 7418 |
+
"step": 1055
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.10474112279309661,
|
| 7422 |
+
"grad_norm": 2.160475730895996,
|
| 7423 |
+
"learning_rate": 5.069255531748817e-05,
|
| 7424 |
+
"loss": 43.3701,
|
| 7425 |
+
"step": 1056
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.10484030946240826,
|
| 7429 |
+
"grad_norm": 1.8825490474700928,
|
| 7430 |
+
"learning_rate": 5.0519120496534044e-05,
|
| 7431 |
+
"loss": 43.1186,
|
| 7432 |
+
"step": 1057
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.10493949613171989,
|
| 7436 |
+
"grad_norm": 1.802302360534668,
|
| 7437 |
+
"learning_rate": 5.034588254392725e-05,
|
| 7438 |
+
"loss": 43.2876,
|
| 7439 |
+
"step": 1058
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.10503868280103154,
|
| 7443 |
+
"grad_norm": 1.9771673679351807,
|
| 7444 |
+
"learning_rate": 5.017284214892542e-05,
|
| 7445 |
+
"loss": 43.0377,
|
| 7446 |
+
"step": 1059
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.10513786947034319,
|
| 7450 |
+
"grad_norm": 1.850211262702942,
|
| 7451 |
+
"learning_rate": 5.000000000000002e-05,
|
| 7452 |
+
"loss": 43.2103,
|
| 7453 |
+
"step": 1060
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.10523705613965482,
|
| 7457 |
+
"grad_norm": 2.223870277404785,
|
| 7458 |
+
"learning_rate": 4.98273567848338e-05,
|
| 7459 |
+
"loss": 43.4367,
|
| 7460 |
+
"step": 1061
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.10533624280896647,
|
| 7464 |
+
"grad_norm": 2.4837911128997803,
|
| 7465 |
+
"learning_rate": 4.96549131903181e-05,
|
| 7466 |
+
"loss": 43.1383,
|
| 7467 |
+
"step": 1062
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.10543542947827812,
|
| 7471 |
+
"grad_norm": 2.060779333114624,
|
| 7472 |
+
"learning_rate": 4.9482669902549894e-05,
|
| 7473 |
+
"loss": 43.4964,
|
| 7474 |
+
"step": 1063
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.10553461614758976,
|
| 7478 |
+
"grad_norm": 1.7346091270446777,
|
| 7479 |
+
"learning_rate": 4.9310627606829354e-05,
|
| 7480 |
+
"loss": 43.3321,
|
| 7481 |
+
"step": 1064
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.1056338028169014,
|
| 7485 |
+
"grad_norm": 2.1066203117370605,
|
| 7486 |
+
"learning_rate": 4.913878698765686e-05,
|
| 7487 |
+
"loss": 43.3465,
|
| 7488 |
+
"step": 1065
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.10573298948621306,
|
| 7492 |
+
"grad_norm": 1.863369345664978,
|
| 7493 |
+
"learning_rate": 4.896714872873038e-05,
|
| 7494 |
+
"loss": 43.3925,
|
| 7495 |
+
"step": 1066
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.1058321761555247,
|
| 7499 |
+
"grad_norm": 1.7801071405410767,
|
| 7500 |
+
"learning_rate": 4.8795713512942865e-05,
|
| 7501 |
+
"loss": 43.3664,
|
| 7502 |
+
"step": 1067
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.10593136282483634,
|
| 7506 |
+
"grad_norm": 1.668254017829895,
|
| 7507 |
+
"learning_rate": 4.862448202237928e-05,
|
| 7508 |
+
"loss": 43.3477,
|
| 7509 |
+
"step": 1068
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.10603054949414799,
|
| 7513 |
+
"grad_norm": 1.7274221181869507,
|
| 7514 |
+
"learning_rate": 4.845345493831419e-05,
|
| 7515 |
+
"loss": 43.2328,
|
| 7516 |
+
"step": 1069
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.10612973616345964,
|
| 7520 |
+
"grad_norm": 1.7936841249465942,
|
| 7521 |
+
"learning_rate": 4.8282632941208725e-05,
|
| 7522 |
+
"loss": 43.4864,
|
| 7523 |
+
"step": 1070
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.10622892283277127,
|
| 7527 |
+
"grad_norm": 1.9316316843032837,
|
| 7528 |
+
"learning_rate": 4.811201671070822e-05,
|
| 7529 |
+
"loss": 43.3535,
|
| 7530 |
+
"step": 1071
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.10632810950208292,
|
| 7534 |
+
"grad_norm": 2.2845723628997803,
|
| 7535 |
+
"learning_rate": 4.794160692563917e-05,
|
| 7536 |
+
"loss": 43.3578,
|
| 7537 |
+
"step": 1072
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.10642729617139457,
|
| 7541 |
+
"grad_norm": 2.1877341270446777,
|
| 7542 |
+
"learning_rate": 4.777140426400674e-05,
|
| 7543 |
+
"loss": 43.105,
|
| 7544 |
+
"step": 1073
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.1065264828407062,
|
| 7548 |
+
"grad_norm": 2.0506670475006104,
|
| 7549 |
+
"learning_rate": 4.7601409402992106e-05,
|
| 7550 |
+
"loss": 43.7111,
|
| 7551 |
+
"step": 1074
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.10662566951001785,
|
| 7555 |
+
"grad_norm": 1.8831859827041626,
|
| 7556 |
+
"learning_rate": 4.743162301894952e-05,
|
| 7557 |
+
"loss": 43.3874,
|
| 7558 |
+
"step": 1075
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.1067248561793295,
|
| 7562 |
+
"grad_norm": 1.7879894971847534,
|
| 7563 |
+
"learning_rate": 4.7262045787403954e-05,
|
| 7564 |
+
"loss": 43.307,
|
| 7565 |
+
"step": 1076
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.10682404284864114,
|
| 7569 |
+
"grad_norm": 2.1504244804382324,
|
| 7570 |
+
"learning_rate": 4.709267838304807e-05,
|
| 7571 |
+
"loss": 43.4333,
|
| 7572 |
+
"step": 1077
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.10692322951795279,
|
| 7576 |
+
"grad_norm": 2.150182008743286,
|
| 7577 |
+
"learning_rate": 4.692352147973973e-05,
|
| 7578 |
+
"loss": 43.1122,
|
| 7579 |
+
"step": 1078
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.10702241618726444,
|
| 7583 |
+
"grad_norm": 1.8888157606124878,
|
| 7584 |
+
"learning_rate": 4.675457575049939e-05,
|
| 7585 |
+
"loss": 43.4367,
|
| 7586 |
+
"step": 1079
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.10712160285657607,
|
| 7590 |
+
"grad_norm": 1.8539072275161743,
|
| 7591 |
+
"learning_rate": 4.658584186750713e-05,
|
| 7592 |
+
"loss": 43.4915,
|
| 7593 |
+
"step": 1080
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.10722078952588772,
|
| 7597 |
+
"grad_norm": 2.0341169834136963,
|
| 7598 |
+
"learning_rate": 4.6417320502100316e-05,
|
| 7599 |
+
"loss": 43.3262,
|
| 7600 |
+
"step": 1081
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.10731997619519937,
|
| 7604 |
+
"grad_norm": 2.1516928672790527,
|
| 7605 |
+
"learning_rate": 4.62490123247708e-05,
|
| 7606 |
+
"loss": 43.4068,
|
| 7607 |
+
"step": 1082
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.107419162864511,
|
| 7611 |
+
"grad_norm": 1.8792285919189453,
|
| 7612 |
+
"learning_rate": 4.6080918005162e-05,
|
| 7613 |
+
"loss": 43.2522,
|
| 7614 |
+
"step": 1083
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.10751834953382265,
|
| 7618 |
+
"grad_norm": 2.136021614074707,
|
| 7619 |
+
"learning_rate": 4.591303821206673e-05,
|
| 7620 |
+
"loss": 43.4792,
|
| 7621 |
+
"step": 1084
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.1076175362031343,
|
| 7625 |
+
"grad_norm": 2.0001168251037598,
|
| 7626 |
+
"learning_rate": 4.574537361342407e-05,
|
| 7627 |
+
"loss": 43.1491,
|
| 7628 |
+
"step": 1085
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.10771672287244595,
|
| 7632 |
+
"grad_norm": 2.2928881645202637,
|
| 7633 |
+
"learning_rate": 4.557792487631703e-05,
|
| 7634 |
+
"loss": 43.3719,
|
| 7635 |
+
"step": 1086
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.10781590954175758,
|
| 7639 |
+
"grad_norm": 1.7387434244155884,
|
| 7640 |
+
"learning_rate": 4.541069266696984e-05,
|
| 7641 |
+
"loss": 43.6456,
|
| 7642 |
+
"step": 1087
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.10791509621106923,
|
| 7646 |
+
"grad_norm": 2.0293469429016113,
|
| 7647 |
+
"learning_rate": 4.524367765074499e-05,
|
| 7648 |
+
"loss": 43.3561,
|
| 7649 |
+
"step": 1088
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.10801428288038088,
|
| 7653 |
+
"grad_norm": 1.9844273328781128,
|
| 7654 |
+
"learning_rate": 4.5076880492141085e-05,
|
| 7655 |
+
"loss": 42.8144,
|
| 7656 |
+
"step": 1089
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.10811346954969252,
|
| 7660 |
+
"grad_norm": 2.1764378547668457,
|
| 7661 |
+
"learning_rate": 4.491030185478976e-05,
|
| 7662 |
+
"loss": 43.4372,
|
| 7663 |
+
"step": 1090
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.10821265621900417,
|
| 7667 |
+
"grad_norm": 1.9110809564590454,
|
| 7668 |
+
"learning_rate": 4.4743942401453344e-05,
|
| 7669 |
+
"loss": 43.4611,
|
| 7670 |
+
"step": 1091
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.10831184288831582,
|
| 7674 |
+
"grad_norm": 2.146286964416504,
|
| 7675 |
+
"learning_rate": 4.457780279402212e-05,
|
| 7676 |
+
"loss": 43.29,
|
| 7677 |
+
"step": 1092
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.10841102955762745,
|
| 7681 |
+
"grad_norm": 1.9756395816802979,
|
| 7682 |
+
"learning_rate": 4.441188369351157e-05,
|
| 7683 |
+
"loss": 43.3144,
|
| 7684 |
+
"step": 1093
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.1085102162269391,
|
| 7688 |
+
"grad_norm": 1.7883363962173462,
|
| 7689 |
+
"learning_rate": 4.424618576005986e-05,
|
| 7690 |
+
"loss": 43.301,
|
| 7691 |
+
"step": 1094
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.10860940289625075,
|
| 7695 |
+
"grad_norm": 2.0152227878570557,
|
| 7696 |
+
"learning_rate": 4.4080709652925336e-05,
|
| 7697 |
+
"loss": 43.241,
|
| 7698 |
+
"step": 1095
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.10870858956556238,
|
| 7702 |
+
"grad_norm": 1.8434028625488281,
|
| 7703 |
+
"learning_rate": 4.391545603048358e-05,
|
| 7704 |
+
"loss": 43.2776,
|
| 7705 |
+
"step": 1096
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.10880777623487403,
|
| 7709 |
+
"grad_norm": 2.1915762424468994,
|
| 7710 |
+
"learning_rate": 4.3750425550225185e-05,
|
| 7711 |
+
"loss": 43.4547,
|
| 7712 |
+
"step": 1097
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.10890696290418568,
|
| 7716 |
+
"grad_norm": 1.9623425006866455,
|
| 7717 |
+
"learning_rate": 4.358561886875273e-05,
|
| 7718 |
+
"loss": 43.1283,
|
| 7719 |
+
"step": 1098
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.10900614957349732,
|
| 7723 |
+
"grad_norm": 1.6339510679244995,
|
| 7724 |
+
"learning_rate": 4.3421036641778556e-05,
|
| 7725 |
+
"loss": 43.2629,
|
| 7726 |
+
"step": 1099
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.10910533624280896,
|
| 7730 |
+
"grad_norm": 2.0017924308776855,
|
| 7731 |
+
"learning_rate": 4.3256679524121834e-05,
|
| 7732 |
+
"loss": 43.2816,
|
| 7733 |
+
"step": 1100
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.10920452291212061,
|
| 7737 |
+
"grad_norm": 1.7236039638519287,
|
| 7738 |
+
"learning_rate": 4.3092548169706127e-05,
|
| 7739 |
+
"loss": 43.5544,
|
| 7740 |
+
"step": 1101
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.10930370958143226,
|
| 7744 |
+
"grad_norm": 1.7514629364013672,
|
| 7745 |
+
"learning_rate": 4.2928643231556844e-05,
|
| 7746 |
+
"loss": 43.4454,
|
| 7747 |
+
"step": 1102
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.1094028962507439,
|
| 7751 |
+
"grad_norm": 2.013662576675415,
|
| 7752 |
+
"learning_rate": 4.2764965361798427e-05,
|
| 7753 |
+
"loss": 43.4465,
|
| 7754 |
+
"step": 1103
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.10950208292005555,
|
| 7758 |
+
"grad_norm": 2.235226631164551,
|
| 7759 |
+
"learning_rate": 4.2601515211652035e-05,
|
| 7760 |
+
"loss": 43.5502,
|
| 7761 |
+
"step": 1104
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.1096012695893672,
|
| 7765 |
+
"grad_norm": 1.858008623123169,
|
| 7766 |
+
"learning_rate": 4.2438293431432665e-05,
|
| 7767 |
+
"loss": 43.1428,
|
| 7768 |
+
"step": 1105
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.10970045625867883,
|
| 7772 |
+
"grad_norm": 2.1349453926086426,
|
| 7773 |
+
"learning_rate": 4.227530067054676e-05,
|
| 7774 |
+
"loss": 43.5154,
|
| 7775 |
+
"step": 1106
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.10979964292799048,
|
| 7779 |
+
"grad_norm": 2.008498191833496,
|
| 7780 |
+
"learning_rate": 4.211253757748961e-05,
|
| 7781 |
+
"loss": 43.4369,
|
| 7782 |
+
"step": 1107
|
| 7783 |
+
},
|
| 7784 |
+
{
|
| 7785 |
+
"epoch": 0.10989882959730213,
|
| 7786 |
+
"grad_norm": 2.134763240814209,
|
| 7787 |
+
"learning_rate": 4.195000479984265e-05,
|
| 7788 |
+
"loss": 43.1325,
|
| 7789 |
+
"step": 1108
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 0.10999801626661376,
|
| 7793 |
+
"grad_norm": 1.965054988861084,
|
| 7794 |
+
"learning_rate": 4.1787702984271074e-05,
|
| 7795 |
+
"loss": 43.3593,
|
| 7796 |
+
"step": 1109
|
| 7797 |
+
},
|
| 7798 |
+
{
|
| 7799 |
+
"epoch": 0.11009720293592541,
|
| 7800 |
+
"grad_norm": 1.7480825185775757,
|
| 7801 |
+
"learning_rate": 4.1625632776521037e-05,
|
| 7802 |
+
"loss": 43.1167,
|
| 7803 |
+
"step": 1110
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 0.11019638960523706,
|
| 7807 |
+
"grad_norm": 1.6889071464538574,
|
| 7808 |
+
"learning_rate": 4.146379482141723e-05,
|
| 7809 |
+
"loss": 43.574,
|
| 7810 |
+
"step": 1111
|
| 7811 |
+
},
|
| 7812 |
+
{
|
| 7813 |
+
"epoch": 0.1102955762745487,
|
| 7814 |
+
"grad_norm": 2.013655662536621,
|
| 7815 |
+
"learning_rate": 4.1302189762860385e-05,
|
| 7816 |
+
"loss": 43.1924,
|
| 7817 |
+
"step": 1112
|
| 7818 |
+
},
|
| 7819 |
+
{
|
| 7820 |
+
"epoch": 0.11039476294386034,
|
| 7821 |
+
"grad_norm": 1.8140257596969604,
|
| 7822 |
+
"learning_rate": 4.1140818243824465e-05,
|
| 7823 |
+
"loss": 43.4124,
|
| 7824 |
+
"step": 1113
|
| 7825 |
+
},
|
| 7826 |
+
{
|
| 7827 |
+
"epoch": 0.110493949613172,
|
| 7828 |
+
"grad_norm": 1.9268773794174194,
|
| 7829 |
+
"learning_rate": 4.097968090635439e-05,
|
| 7830 |
+
"loss": 43.236,
|
| 7831 |
+
"step": 1114
|
| 7832 |
+
},
|
| 7833 |
+
{
|
| 7834 |
+
"epoch": 0.11059313628248363,
|
| 7835 |
+
"grad_norm": 1.7209376096725464,
|
| 7836 |
+
"learning_rate": 4.081877839156325e-05,
|
| 7837 |
+
"loss": 43.2984,
|
| 7838 |
+
"step": 1115
|
| 7839 |
+
},
|
| 7840 |
+
{
|
| 7841 |
+
"epoch": 0.11069232295179528,
|
| 7842 |
+
"grad_norm": 2.013707399368286,
|
| 7843 |
+
"learning_rate": 4.065811133962987e-05,
|
| 7844 |
+
"loss": 43.0707,
|
| 7845 |
+
"step": 1116
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 0.11079150962110693,
|
| 7849 |
+
"grad_norm": 2.080244302749634,
|
| 7850 |
+
"learning_rate": 4.049768038979631e-05,
|
| 7851 |
+
"loss": 43.2184,
|
| 7852 |
+
"step": 1117
|
| 7853 |
+
},
|
| 7854 |
+
{
|
| 7855 |
+
"epoch": 0.11089069629041856,
|
| 7856 |
+
"grad_norm": 1.7076959609985352,
|
| 7857 |
+
"learning_rate": 4.033748618036515e-05,
|
| 7858 |
+
"loss": 43.2103,
|
| 7859 |
+
"step": 1118
|
| 7860 |
+
},
|
| 7861 |
+
{
|
| 7862 |
+
"epoch": 0.11098988295973021,
|
| 7863 |
+
"grad_norm": 2.194852828979492,
|
| 7864 |
+
"learning_rate": 4.0177529348697185e-05,
|
| 7865 |
+
"loss": 43.0531,
|
| 7866 |
+
"step": 1119
|
| 7867 |
+
},
|
| 7868 |
+
{
|
| 7869 |
+
"epoch": 0.11108906962904186,
|
| 7870 |
+
"grad_norm": 1.9565582275390625,
|
| 7871 |
+
"learning_rate": 4.001781053120863e-05,
|
| 7872 |
+
"loss": 43.1544,
|
| 7873 |
+
"step": 1120
|
| 7874 |
+
},
|
| 7875 |
+
{
|
| 7876 |
+
"epoch": 0.11118825629835351,
|
| 7877 |
+
"grad_norm": 1.8571580648422241,
|
| 7878 |
+
"learning_rate": 3.985833036336883e-05,
|
| 7879 |
+
"loss": 43.1887,
|
| 7880 |
+
"step": 1121
|
| 7881 |
+
},
|
| 7882 |
+
{
|
| 7883 |
+
"epoch": 0.11128744296766514,
|
| 7884 |
+
"grad_norm": 1.8049238920211792,
|
| 7885 |
+
"learning_rate": 3.969908947969756e-05,
|
| 7886 |
+
"loss": 43.2068,
|
| 7887 |
+
"step": 1122
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 0.11138662963697679,
|
| 7891 |
+
"grad_norm": 1.9766581058502197,
|
| 7892 |
+
"learning_rate": 3.954008851376252e-05,
|
| 7893 |
+
"loss": 43.5265,
|
| 7894 |
+
"step": 1123
|
| 7895 |
+
},
|
| 7896 |
+
{
|
| 7897 |
+
"epoch": 0.11148581630628844,
|
| 7898 |
+
"grad_norm": 2.094628095626831,
|
| 7899 |
+
"learning_rate": 3.9381328098176985e-05,
|
| 7900 |
+
"loss": 43.0592,
|
| 7901 |
+
"step": 1124
|
| 7902 |
+
},
|
| 7903 |
+
{
|
| 7904 |
+
"epoch": 0.11158500297560008,
|
| 7905 |
+
"grad_norm": 2.1410720348358154,
|
| 7906 |
+
"learning_rate": 3.9222808864597004e-05,
|
| 7907 |
+
"loss": 42.795,
|
| 7908 |
+
"step": 1125
|
| 7909 |
+
},
|
| 7910 |
+
{
|
| 7911 |
+
"epoch": 0.11168418964491172,
|
| 7912 |
+
"grad_norm": 2.1086578369140625,
|
| 7913 |
+
"learning_rate": 3.90645314437192e-05,
|
| 7914 |
+
"loss": 43.2222,
|
| 7915 |
+
"step": 1126
|
| 7916 |
+
},
|
| 7917 |
+
{
|
| 7918 |
+
"epoch": 0.11178337631422337,
|
| 7919 |
+
"grad_norm": 2.397113561630249,
|
| 7920 |
+
"learning_rate": 3.8906496465277976e-05,
|
| 7921 |
+
"loss": 42.9432,
|
| 7922 |
+
"step": 1127
|
| 7923 |
+
},
|
| 7924 |
+
{
|
| 7925 |
+
"epoch": 0.11188256298353501,
|
| 7926 |
+
"grad_norm": 1.9690929651260376,
|
| 7927 |
+
"learning_rate": 3.8748704558043135e-05,
|
| 7928 |
+
"loss": 43.4933,
|
| 7929 |
+
"step": 1128
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 0.11198174965284666,
|
| 7933 |
+
"grad_norm": 2.4064979553222656,
|
| 7934 |
+
"learning_rate": 3.859115634981748e-05,
|
| 7935 |
+
"loss": 43.374,
|
| 7936 |
+
"step": 1129
|
| 7937 |
+
},
|
| 7938 |
+
{
|
| 7939 |
+
"epoch": 0.1120809363221583,
|
| 7940 |
+
"grad_norm": 2.03690242767334,
|
| 7941 |
+
"learning_rate": 3.843385246743417e-05,
|
| 7942 |
+
"loss": 43.3578,
|
| 7943 |
+
"step": 1130
|
| 7944 |
+
},
|
| 7945 |
+
{
|
| 7946 |
+
"epoch": 0.11218012299146994,
|
| 7947 |
+
"grad_norm": 1.9198079109191895,
|
| 7948 |
+
"learning_rate": 3.827679353675421e-05,
|
| 7949 |
+
"loss": 43.3936,
|
| 7950 |
+
"step": 1131
|
| 7951 |
+
},
|
| 7952 |
+
{
|
| 7953 |
+
"epoch": 0.11227930966078159,
|
| 7954 |
+
"grad_norm": 1.8240329027175903,
|
| 7955 |
+
"learning_rate": 3.811998018266416e-05,
|
| 7956 |
+
"loss": 43.3825,
|
| 7957 |
+
"step": 1132
|
| 7958 |
+
},
|
| 7959 |
+
{
|
| 7960 |
+
"epoch": 0.11237849633009324,
|
| 7961 |
+
"grad_norm": 1.9616767168045044,
|
| 7962 |
+
"learning_rate": 3.796341302907328e-05,
|
| 7963 |
+
"loss": 42.9891,
|
| 7964 |
+
"step": 1133
|
| 7965 |
+
},
|
| 7966 |
+
{
|
| 7967 |
+
"epoch": 0.11247768299940487,
|
| 7968 |
+
"grad_norm": 1.9595699310302734,
|
| 7969 |
+
"learning_rate": 3.7807092698911494e-05,
|
| 7970 |
+
"loss": 42.9743,
|
| 7971 |
+
"step": 1134
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 0.11257686966871652,
|
| 7975 |
+
"grad_norm": 1.847184658050537,
|
| 7976 |
+
"learning_rate": 3.7651019814126654e-05,
|
| 7977 |
+
"loss": 43.2845,
|
| 7978 |
+
"step": 1135
|
| 7979 |
+
},
|
| 7980 |
+
{
|
| 7981 |
+
"epoch": 0.11267605633802817,
|
| 7982 |
+
"grad_norm": 2.2341554164886475,
|
| 7983 |
+
"learning_rate": 3.7495194995682e-05,
|
| 7984 |
+
"loss": 42.8106,
|
| 7985 |
+
"step": 1136
|
| 7986 |
+
},
|
| 7987 |
+
{
|
| 7988 |
+
"epoch": 0.11277524300733982,
|
| 7989 |
+
"grad_norm": 2.246288299560547,
|
| 7990 |
+
"learning_rate": 3.733961886355398e-05,
|
| 7991 |
+
"loss": 43.5755,
|
| 7992 |
+
"step": 1137
|
| 7993 |
+
},
|
| 7994 |
+
{
|
| 7995 |
+
"epoch": 0.11287442967665146,
|
| 7996 |
+
"grad_norm": 2.117567300796509,
|
| 7997 |
+
"learning_rate": 3.718429203672936e-05,
|
| 7998 |
+
"loss": 43.3262,
|
| 7999 |
+
"step": 1138
|
| 8000 |
+
},
|
| 8001 |
+
{
|
| 8002 |
+
"epoch": 0.1129736163459631,
|
| 8003 |
+
"grad_norm": 1.8588794469833374,
|
| 8004 |
+
"learning_rate": 3.702921513320318e-05,
|
| 8005 |
+
"loss": 43.42,
|
| 8006 |
+
"step": 1139
|
| 8007 |
+
},
|
| 8008 |
+
{
|
| 8009 |
+
"epoch": 0.11307280301527475,
|
| 8010 |
+
"grad_norm": 2.050076723098755,
|
| 8011 |
+
"learning_rate": 3.687438876997612e-05,
|
| 8012 |
+
"loss": 43.5885,
|
| 8013 |
+
"step": 1140
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 0.11317198968458639,
|
| 8017 |
+
"grad_norm": 1.8713175058364868,
|
| 8018 |
+
"learning_rate": 3.671981356305191e-05,
|
| 8019 |
+
"loss": 43.1833,
|
| 8020 |
+
"step": 1141
|
| 8021 |
+
},
|
| 8022 |
+
{
|
| 8023 |
+
"epoch": 0.11327117635389804,
|
| 8024 |
+
"grad_norm": 1.745923638343811,
|
| 8025 |
+
"learning_rate": 3.656549012743515e-05,
|
| 8026 |
+
"loss": 43.3071,
|
| 8027 |
+
"step": 1142
|
| 8028 |
+
},
|
| 8029 |
+
{
|
| 8030 |
+
"epoch": 0.11337036302320969,
|
| 8031 |
+
"grad_norm": 2.048849105834961,
|
| 8032 |
+
"learning_rate": 3.641141907712865e-05,
|
| 8033 |
+
"loss": 43.2482,
|
| 8034 |
+
"step": 1143
|
| 8035 |
+
},
|
| 8036 |
+
{
|
| 8037 |
+
"epoch": 0.11346954969252132,
|
| 8038 |
+
"grad_norm": 2.088815450668335,
|
| 8039 |
+
"learning_rate": 3.6257601025131026e-05,
|
| 8040 |
+
"loss": 43.5718,
|
| 8041 |
+
"step": 1144
|
| 8042 |
+
},
|
| 8043 |
+
{
|
| 8044 |
+
"epoch": 0.11356873636183297,
|
| 8045 |
+
"grad_norm": 1.8028887510299683,
|
| 8046 |
+
"learning_rate": 3.610403658343443e-05,
|
| 8047 |
+
"loss": 43.3762,
|
| 8048 |
+
"step": 1145
|
| 8049 |
+
},
|
| 8050 |
+
{
|
| 8051 |
+
"epoch": 0.11366792303114462,
|
| 8052 |
+
"grad_norm": 2.095184803009033,
|
| 8053 |
+
"learning_rate": 3.59507263630218e-05,
|
| 8054 |
+
"loss": 43.4513,
|
| 8055 |
+
"step": 1146
|
| 8056 |
+
},
|
| 8057 |
+
{
|
| 8058 |
+
"epoch": 0.11376710970045625,
|
| 8059 |
+
"grad_norm": 2.1411571502685547,
|
| 8060 |
+
"learning_rate": 3.57976709738648e-05,
|
| 8061 |
+
"loss": 43.4294,
|
| 8062 |
+
"step": 1147
|
| 8063 |
+
},
|
| 8064 |
+
{
|
| 8065 |
+
"epoch": 0.1138662963697679,
|
| 8066 |
+
"grad_norm": 1.7514969110488892,
|
| 8067 |
+
"learning_rate": 3.564487102492102e-05,
|
| 8068 |
+
"loss": 43.4317,
|
| 8069 |
+
"step": 1148
|
| 8070 |
+
},
|
| 8071 |
+
{
|
| 8072 |
+
"epoch": 0.11396548303907955,
|
| 8073 |
+
"grad_norm": 1.6412819623947144,
|
| 8074 |
+
"learning_rate": 3.5492327124131905e-05,
|
| 8075 |
+
"loss": 43.4319,
|
| 8076 |
+
"step": 1149
|
| 8077 |
+
},
|
| 8078 |
+
{
|
| 8079 |
+
"epoch": 0.11406466970839119,
|
| 8080 |
+
"grad_norm": 2.071122407913208,
|
| 8081 |
+
"learning_rate": 3.534003987842005e-05,
|
| 8082 |
+
"loss": 42.9175,
|
| 8083 |
+
"step": 1150
|
| 8084 |
+
},
|
| 8085 |
+
{
|
| 8086 |
+
"epoch": 0.11416385637770284,
|
| 8087 |
+
"grad_norm": 2.3104162216186523,
|
| 8088 |
+
"learning_rate": 3.518800989368691e-05,
|
| 8089 |
+
"loss": 43.5465,
|
| 8090 |
+
"step": 1151
|
| 8091 |
+
},
|
| 8092 |
+
{
|
| 8093 |
+
"epoch": 0.11426304304701448,
|
| 8094 |
+
"grad_norm": 1.548854112625122,
|
| 8095 |
+
"learning_rate": 3.5036237774810476e-05,
|
| 8096 |
+
"loss": 43.4463,
|
| 8097 |
+
"step": 1152
|
| 8098 |
+
},
|
| 8099 |
+
{
|
| 8100 |
+
"epoch": 0.11436222971632612,
|
| 8101 |
+
"grad_norm": 2.01901912689209,
|
| 8102 |
+
"learning_rate": 3.488472412564264e-05,
|
| 8103 |
+
"loss": 43.2515,
|
| 8104 |
+
"step": 1153
|
| 8105 |
+
},
|
| 8106 |
+
{
|
| 8107 |
+
"epoch": 0.11446141638563777,
|
| 8108 |
+
"grad_norm": 2.0464985370635986,
|
| 8109 |
+
"learning_rate": 3.473346954900708e-05,
|
| 8110 |
+
"loss": 43.451,
|
| 8111 |
+
"step": 1154
|
| 8112 |
+
},
|
| 8113 |
+
{
|
| 8114 |
+
"epoch": 0.11456060305494942,
|
| 8115 |
+
"grad_norm": 2.401381254196167,
|
| 8116 |
+
"learning_rate": 3.458247464669657e-05,
|
| 8117 |
+
"loss": 43.2622,
|
| 8118 |
+
"step": 1155
|
| 8119 |
+
},
|
| 8120 |
+
{
|
| 8121 |
+
"epoch": 0.11465978972426107,
|
| 8122 |
+
"grad_norm": 1.8948826789855957,
|
| 8123 |
+
"learning_rate": 3.4431740019470774e-05,
|
| 8124 |
+
"loss": 43.2473,
|
| 8125 |
+
"step": 1156
|
| 8126 |
+
},
|
| 8127 |
+
{
|
| 8128 |
+
"epoch": 0.1147589763935727,
|
| 8129 |
+
"grad_norm": 1.8047832250595093,
|
| 8130 |
+
"learning_rate": 3.4281266267053856e-05,
|
| 8131 |
+
"loss": 43.1144,
|
| 8132 |
+
"step": 1157
|
| 8133 |
+
},
|
| 8134 |
+
{
|
| 8135 |
+
"epoch": 0.11485816306288435,
|
| 8136 |
+
"grad_norm": 1.9705872535705566,
|
| 8137 |
+
"learning_rate": 3.413105398813195e-05,
|
| 8138 |
+
"loss": 43.2361,
|
| 8139 |
+
"step": 1158
|
| 8140 |
+
},
|
| 8141 |
+
{
|
| 8142 |
+
"epoch": 0.114957349732196,
|
| 8143 |
+
"grad_norm": 2.247692584991455,
|
| 8144 |
+
"learning_rate": 3.398110378035098e-05,
|
| 8145 |
+
"loss": 43.4239,
|
| 8146 |
+
"step": 1159
|
| 8147 |
+
},
|
| 8148 |
+
{
|
| 8149 |
+
"epoch": 0.11505653640150763,
|
| 8150 |
+
"grad_norm": 2.4155638217926025,
|
| 8151 |
+
"learning_rate": 3.383141624031408e-05,
|
| 8152 |
+
"loss": 43.2816,
|
| 8153 |
+
"step": 1160
|
| 8154 |
+
},
|
| 8155 |
+
{
|
| 8156 |
+
"epoch": 0.11515572307081928,
|
| 8157 |
+
"grad_norm": 2.051988124847412,
|
| 8158 |
+
"learning_rate": 3.368199196357934e-05,
|
| 8159 |
+
"loss": 43.6693,
|
| 8160 |
+
"step": 1161
|
| 8161 |
+
},
|
| 8162 |
+
{
|
| 8163 |
+
"epoch": 0.11525490974013093,
|
| 8164 |
+
"grad_norm": 2.1297595500946045,
|
| 8165 |
+
"learning_rate": 3.353283154465746e-05,
|
| 8166 |
+
"loss": 42.9023,
|
| 8167 |
+
"step": 1162
|
| 8168 |
+
},
|
| 8169 |
+
{
|
| 8170 |
+
"epoch": 0.11535409640944257,
|
| 8171 |
+
"grad_norm": 2.1139824390411377,
|
| 8172 |
+
"learning_rate": 3.338393557700924e-05,
|
| 8173 |
+
"loss": 43.5984,
|
| 8174 |
+
"step": 1163
|
| 8175 |
+
},
|
| 8176 |
+
{
|
| 8177 |
+
"epoch": 0.11545328307875422,
|
| 8178 |
+
"grad_norm": 1.9533110857009888,
|
| 8179 |
+
"learning_rate": 3.323530465304341e-05,
|
| 8180 |
+
"loss": 43.2917,
|
| 8181 |
+
"step": 1164
|
| 8182 |
+
},
|
| 8183 |
+
{
|
| 8184 |
+
"epoch": 0.11555246974806586,
|
| 8185 |
+
"grad_norm": 1.7646098136901855,
|
| 8186 |
+
"learning_rate": 3.308693936411421e-05,
|
| 8187 |
+
"loss": 43.2659,
|
| 8188 |
+
"step": 1165
|
| 8189 |
+
},
|
| 8190 |
+
{
|
| 8191 |
+
"epoch": 0.1156516564173775,
|
| 8192 |
+
"grad_norm": 1.7004350423812866,
|
| 8193 |
+
"learning_rate": 3.293884030051879e-05,
|
| 8194 |
+
"loss": 43.4244,
|
| 8195 |
+
"step": 1166
|
| 8196 |
+
},
|
| 8197 |
+
{
|
| 8198 |
+
"epoch": 0.11575084308668915,
|
| 8199 |
+
"grad_norm": 2.168750524520874,
|
| 8200 |
+
"learning_rate": 3.279100805149532e-05,
|
| 8201 |
+
"loss": 43.419,
|
| 8202 |
+
"step": 1167
|
| 8203 |
+
},
|
| 8204 |
+
{
|
| 8205 |
+
"epoch": 0.1158500297560008,
|
| 8206 |
+
"grad_norm": 1.7945916652679443,
|
| 8207 |
+
"learning_rate": 3.264344320522024e-05,
|
| 8208 |
+
"loss": 43.3498,
|
| 8209 |
+
"step": 1168
|
| 8210 |
+
},
|
| 8211 |
+
{
|
| 8212 |
+
"epoch": 0.11594921642531243,
|
| 8213 |
+
"grad_norm": 2.001431465148926,
|
| 8214 |
+
"learning_rate": 3.249614634880617e-05,
|
| 8215 |
+
"loss": 43.1847,
|
| 8216 |
+
"step": 1169
|
| 8217 |
+
},
|
| 8218 |
+
{
|
| 8219 |
+
"epoch": 0.11604840309462408,
|
| 8220 |
+
"grad_norm": 1.8130570650100708,
|
| 8221 |
+
"learning_rate": 3.234911806829948e-05,
|
| 8222 |
+
"loss": 43.0508,
|
| 8223 |
+
"step": 1170
|
| 8224 |
+
},
|
| 8225 |
+
{
|
| 8226 |
+
"epoch": 0.11614758976393573,
|
| 8227 |
+
"grad_norm": 1.989986777305603,
|
| 8228 |
+
"learning_rate": 3.220235894867794e-05,
|
| 8229 |
+
"loss": 43.2058,
|
| 8230 |
+
"step": 1171
|
| 8231 |
+
},
|
| 8232 |
+
{
|
| 8233 |
+
"epoch": 0.11624677643324736,
|
| 8234 |
+
"grad_norm": 1.729008674621582,
|
| 8235 |
+
"learning_rate": 3.205586957384838e-05,
|
| 8236 |
+
"loss": 43.4374,
|
| 8237 |
+
"step": 1172
|
| 8238 |
+
},
|
| 8239 |
+
{
|
| 8240 |
+
"epoch": 0.11634596310255901,
|
| 8241 |
+
"grad_norm": 1.8263639211654663,
|
| 8242 |
+
"learning_rate": 3.190965052664443e-05,
|
| 8243 |
+
"loss": 43.0842,
|
| 8244 |
+
"step": 1173
|
| 8245 |
+
},
|
| 8246 |
+
{
|
| 8247 |
+
"epoch": 0.11644514977187066,
|
| 8248 |
+
"grad_norm": 2.0595879554748535,
|
| 8249 |
+
"learning_rate": 3.1763702388824214e-05,
|
| 8250 |
+
"loss": 43.0447,
|
| 8251 |
+
"step": 1174
|
| 8252 |
+
},
|
| 8253 |
+
{
|
| 8254 |
+
"epoch": 0.11654433644118231,
|
| 8255 |
+
"grad_norm": 2.060483455657959,
|
| 8256 |
+
"learning_rate": 3.161802574106799e-05,
|
| 8257 |
+
"loss": 43.0739,
|
| 8258 |
+
"step": 1175
|
| 8259 |
+
},
|
| 8260 |
+
{
|
| 8261 |
+
"epoch": 0.11664352311049395,
|
| 8262 |
+
"grad_norm": 2.0084433555603027,
|
| 8263 |
+
"learning_rate": 3.14726211629758e-05,
|
| 8264 |
+
"loss": 43.5443,
|
| 8265 |
+
"step": 1176
|
| 8266 |
+
},
|
| 8267 |
+
{
|
| 8268 |
+
"epoch": 0.1167427097798056,
|
| 8269 |
+
"grad_norm": 1.6945682764053345,
|
| 8270 |
+
"learning_rate": 3.132748923306522e-05,
|
| 8271 |
+
"loss": 43.4913,
|
| 8272 |
+
"step": 1177
|
| 8273 |
+
},
|
| 8274 |
+
{
|
| 8275 |
+
"epoch": 0.11684189644911724,
|
| 8276 |
+
"grad_norm": 2.0036113262176514,
|
| 8277 |
+
"learning_rate": 3.118263052876904e-05,
|
| 8278 |
+
"loss": 43.3576,
|
| 8279 |
+
"step": 1178
|
| 8280 |
+
},
|
| 8281 |
+
{
|
| 8282 |
+
"epoch": 0.11694108311842888,
|
| 8283 |
+
"grad_norm": 1.7544937133789062,
|
| 8284 |
+
"learning_rate": 3.103804562643302e-05,
|
| 8285 |
+
"loss": 43.5293,
|
| 8286 |
+
"step": 1179
|
| 8287 |
+
},
|
| 8288 |
+
{
|
| 8289 |
+
"epoch": 0.11704026978774053,
|
| 8290 |
+
"grad_norm": 1.9170342683792114,
|
| 8291 |
+
"learning_rate": 3.089373510131354e-05,
|
| 8292 |
+
"loss": 43.0252,
|
| 8293 |
+
"step": 1180
|
| 8294 |
+
},
|
| 8295 |
+
{
|
| 8296 |
+
"epoch": 0.11713945645705218,
|
| 8297 |
+
"grad_norm": 2.420809030532837,
|
| 8298 |
+
"learning_rate": 3.074969952757526e-05,
|
| 8299 |
+
"loss": 42.728,
|
| 8300 |
+
"step": 1181
|
| 8301 |
+
},
|
| 8302 |
+
{
|
| 8303 |
+
"epoch": 0.11723864312636381,
|
| 8304 |
+
"grad_norm": 1.7426315546035767,
|
| 8305 |
+
"learning_rate": 3.060593947828899e-05,
|
| 8306 |
+
"loss": 43.1784,
|
| 8307 |
+
"step": 1182
|
| 8308 |
+
},
|
| 8309 |
+
{
|
| 8310 |
+
"epoch": 0.11733782979567546,
|
| 8311 |
+
"grad_norm": 2.063729763031006,
|
| 8312 |
+
"learning_rate": 3.0462455525429257e-05,
|
| 8313 |
+
"loss": 43.3167,
|
| 8314 |
+
"step": 1183
|
| 8315 |
+
},
|
| 8316 |
+
{
|
| 8317 |
+
"epoch": 0.11743701646498711,
|
| 8318 |
+
"grad_norm": 2.194005250930786,
|
| 8319 |
+
"learning_rate": 3.0319248239872057e-05,
|
| 8320 |
+
"loss": 43.0883,
|
| 8321 |
+
"step": 1184
|
| 8322 |
+
},
|
| 8323 |
+
{
|
| 8324 |
+
"epoch": 0.11753620313429874,
|
| 8325 |
+
"grad_norm": 1.9286993741989136,
|
| 8326 |
+
"learning_rate": 3.0176318191392726e-05,
|
| 8327 |
+
"loss": 43.4406,
|
| 8328 |
+
"step": 1185
|
| 8329 |
+
},
|
| 8330 |
+
{
|
| 8331 |
+
"epoch": 0.1176353898036104,
|
| 8332 |
+
"grad_norm": 1.8343324661254883,
|
| 8333 |
+
"learning_rate": 3.0033665948663448e-05,
|
| 8334 |
+
"loss": 43.4349,
|
| 8335 |
+
"step": 1186
|
| 8336 |
+
},
|
| 8337 |
+
{
|
| 8338 |
+
"epoch": 0.11773457647292204,
|
| 8339 |
+
"grad_norm": 1.93137526512146,
|
| 8340 |
+
"learning_rate": 2.989129207925122e-05,
|
| 8341 |
+
"loss": 43.3723,
|
| 8342 |
+
"step": 1187
|
| 8343 |
+
},
|
| 8344 |
+
{
|
| 8345 |
+
"epoch": 0.11783376314223368,
|
| 8346 |
+
"grad_norm": 2.171217679977417,
|
| 8347 |
+
"learning_rate": 2.9749197149615392e-05,
|
| 8348 |
+
"loss": 43.1872,
|
| 8349 |
+
"step": 1188
|
| 8350 |
+
},
|
| 8351 |
+
{
|
| 8352 |
+
"epoch": 0.11793294981154533,
|
| 8353 |
+
"grad_norm": 2.2938199043273926,
|
| 8354 |
+
"learning_rate": 2.960738172510551e-05,
|
| 8355 |
+
"loss": 43.0836,
|
| 8356 |
+
"step": 1189
|
| 8357 |
+
},
|
| 8358 |
+
{
|
| 8359 |
+
"epoch": 0.11803213648085698,
|
| 8360 |
+
"grad_norm": 1.7842198610305786,
|
| 8361 |
+
"learning_rate": 2.9465846369959127e-05,
|
| 8362 |
+
"loss": 43.1941,
|
| 8363 |
+
"step": 1190
|
| 8364 |
+
},
|
| 8365 |
+
{
|
| 8366 |
+
"epoch": 0.11813132315016862,
|
| 8367 |
+
"grad_norm": 2.3533241748809814,
|
| 8368 |
+
"learning_rate": 2.9324591647299403e-05,
|
| 8369 |
+
"loss": 43.2269,
|
| 8370 |
+
"step": 1191
|
| 8371 |
+
},
|
| 8372 |
+
{
|
| 8373 |
+
"epoch": 0.11813132315016862,
|
| 8374 |
+
"eval_loss": 10.822596549987793,
|
| 8375 |
+
"eval_runtime": 11.1927,
|
| 8376 |
+
"eval_samples_per_second": 379.265,
|
| 8377 |
+
"eval_steps_per_second": 189.677,
|
| 8378 |
+
"step": 1191
|
| 8379 |
}
|
| 8380 |
],
|
| 8381 |
"logging_steps": 1,
|
|
|
|
| 8395 |
"attributes": {}
|
| 8396 |
}
|
| 8397 |
},
|
| 8398 |
+
"total_flos": 11566528856064.0,
|
| 8399 |
"train_batch_size": 2,
|
| 8400 |
"trial_name": null,
|
| 8401 |
"trial_params": null
|