Training in progress, step 3500, 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 1673342072
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4a8ad065d2bf6fe719db00bfc86d9aca4be6f0bca9b3557a8ed7a64539963b55
|
| 3 |
size 1673342072
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 194745274
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:1f0a8c27caac8d4d654fe68e5b9789c63eb21a703a3bf567823bc1d52feb39dc
|
| 3 |
size 194745274
|
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:e9b1df5fb25513ffea2c862b1a01785af03a097fa89b0a8f449bf99d0346d6e0
|
| 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 sha256:4288febd9081786b52bdf01d6f570b7ebf36526320fd40daa36c3cb1500b408c
|
| 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": 900,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -21031,6 +21031,3506 @@
|
|
| 21031 |
"learning_rate": 6.6815144766147e-05,
|
| 21032 |
"loss": 0.8101,
|
| 21033 |
"step": 3000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 21034 |
}
|
| 21035 |
],
|
| 21036 |
"logging_steps": 1,
|
|
@@ -21050,7 +24550,7 @@
|
|
| 21050 |
"attributes": {}
|
| 21051 |
}
|
| 21052 |
},
|
| 21053 |
-
"total_flos": 3.
|
| 21054 |
"train_batch_size": 1,
|
| 21055 |
"trial_name": null,
|
| 21056 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.7777777777777778,
|
| 5 |
"eval_steps": 900,
|
| 6 |
+
"global_step": 3500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 21031 |
"learning_rate": 6.6815144766147e-05,
|
| 21032 |
"loss": 0.8101,
|
| 21033 |
"step": 3000
|
| 21034 |
+
},
|
| 21035 |
+
{
|
| 21036 |
+
"epoch": 0.6668888888888889,
|
| 21037 |
+
"grad_norm": 0.8144944906234741,
|
| 21038 |
+
"learning_rate": 6.67706013363029e-05,
|
| 21039 |
+
"loss": 1.9887,
|
| 21040 |
+
"step": 3001
|
| 21041 |
+
},
|
| 21042 |
+
{
|
| 21043 |
+
"epoch": 0.6671111111111111,
|
| 21044 |
+
"grad_norm": 0.8721863627433777,
|
| 21045 |
+
"learning_rate": 6.67260579064588e-05,
|
| 21046 |
+
"loss": 2.3995,
|
| 21047 |
+
"step": 3002
|
| 21048 |
+
},
|
| 21049 |
+
{
|
| 21050 |
+
"epoch": 0.6673333333333333,
|
| 21051 |
+
"grad_norm": 0.801108717918396,
|
| 21052 |
+
"learning_rate": 6.66815144766147e-05,
|
| 21053 |
+
"loss": 2.1919,
|
| 21054 |
+
"step": 3003
|
| 21055 |
+
},
|
| 21056 |
+
{
|
| 21057 |
+
"epoch": 0.6675555555555556,
|
| 21058 |
+
"grad_norm": 0.6709057688713074,
|
| 21059 |
+
"learning_rate": 6.66369710467706e-05,
|
| 21060 |
+
"loss": 1.0861,
|
| 21061 |
+
"step": 3004
|
| 21062 |
+
},
|
| 21063 |
+
{
|
| 21064 |
+
"epoch": 0.6677777777777778,
|
| 21065 |
+
"grad_norm": 0.05186731740832329,
|
| 21066 |
+
"learning_rate": 6.659242761692652e-05,
|
| 21067 |
+
"loss": 0.0109,
|
| 21068 |
+
"step": 3005
|
| 21069 |
+
},
|
| 21070 |
+
{
|
| 21071 |
+
"epoch": 0.668,
|
| 21072 |
+
"grad_norm": 0.6289195418357849,
|
| 21073 |
+
"learning_rate": 6.654788418708241e-05,
|
| 21074 |
+
"loss": 1.1656,
|
| 21075 |
+
"step": 3006
|
| 21076 |
+
},
|
| 21077 |
+
{
|
| 21078 |
+
"epoch": 0.6682222222222223,
|
| 21079 |
+
"grad_norm": 0.5143423080444336,
|
| 21080 |
+
"learning_rate": 6.650334075723831e-05,
|
| 21081 |
+
"loss": 1.1429,
|
| 21082 |
+
"step": 3007
|
| 21083 |
+
},
|
| 21084 |
+
{
|
| 21085 |
+
"epoch": 0.6684444444444444,
|
| 21086 |
+
"grad_norm": 0.7924249768257141,
|
| 21087 |
+
"learning_rate": 6.645879732739422e-05,
|
| 21088 |
+
"loss": 1.8575,
|
| 21089 |
+
"step": 3008
|
| 21090 |
+
},
|
| 21091 |
+
{
|
| 21092 |
+
"epoch": 0.6686666666666666,
|
| 21093 |
+
"grad_norm": 0.09778264164924622,
|
| 21094 |
+
"learning_rate": 6.641425389755011e-05,
|
| 21095 |
+
"loss": 0.0128,
|
| 21096 |
+
"step": 3009
|
| 21097 |
+
},
|
| 21098 |
+
{
|
| 21099 |
+
"epoch": 0.6688888888888889,
|
| 21100 |
+
"grad_norm": 0.10178276896476746,
|
| 21101 |
+
"learning_rate": 6.636971046770602e-05,
|
| 21102 |
+
"loss": 0.013,
|
| 21103 |
+
"step": 3010
|
| 21104 |
+
},
|
| 21105 |
+
{
|
| 21106 |
+
"epoch": 0.6691111111111111,
|
| 21107 |
+
"grad_norm": 0.09697845578193665,
|
| 21108 |
+
"learning_rate": 6.632516703786192e-05,
|
| 21109 |
+
"loss": 0.0126,
|
| 21110 |
+
"step": 3011
|
| 21111 |
+
},
|
| 21112 |
+
{
|
| 21113 |
+
"epoch": 0.6693333333333333,
|
| 21114 |
+
"grad_norm": 1.0395288467407227,
|
| 21115 |
+
"learning_rate": 6.628062360801783e-05,
|
| 21116 |
+
"loss": 2.0803,
|
| 21117 |
+
"step": 3012
|
| 21118 |
+
},
|
| 21119 |
+
{
|
| 21120 |
+
"epoch": 0.6695555555555556,
|
| 21121 |
+
"grad_norm": 0.8418979048728943,
|
| 21122 |
+
"learning_rate": 6.623608017817372e-05,
|
| 21123 |
+
"loss": 2.1971,
|
| 21124 |
+
"step": 3013
|
| 21125 |
+
},
|
| 21126 |
+
{
|
| 21127 |
+
"epoch": 0.6697777777777778,
|
| 21128 |
+
"grad_norm": 0.9855999946594238,
|
| 21129 |
+
"learning_rate": 6.619153674832962e-05,
|
| 21130 |
+
"loss": 1.9663,
|
| 21131 |
+
"step": 3014
|
| 21132 |
+
},
|
| 21133 |
+
{
|
| 21134 |
+
"epoch": 0.67,
|
| 21135 |
+
"grad_norm": 0.9182018637657166,
|
| 21136 |
+
"learning_rate": 6.614699331848553e-05,
|
| 21137 |
+
"loss": 2.1056,
|
| 21138 |
+
"step": 3015
|
| 21139 |
+
},
|
| 21140 |
+
{
|
| 21141 |
+
"epoch": 0.6702222222222223,
|
| 21142 |
+
"grad_norm": 0.9927064180374146,
|
| 21143 |
+
"learning_rate": 6.610244988864142e-05,
|
| 21144 |
+
"loss": 1.8567,
|
| 21145 |
+
"step": 3016
|
| 21146 |
+
},
|
| 21147 |
+
{
|
| 21148 |
+
"epoch": 0.6704444444444444,
|
| 21149 |
+
"grad_norm": 0.8513967394828796,
|
| 21150 |
+
"learning_rate": 6.605790645879733e-05,
|
| 21151 |
+
"loss": 1.8663,
|
| 21152 |
+
"step": 3017
|
| 21153 |
+
},
|
| 21154 |
+
{
|
| 21155 |
+
"epoch": 0.6706666666666666,
|
| 21156 |
+
"grad_norm": 0.6829978227615356,
|
| 21157 |
+
"learning_rate": 6.601336302895323e-05,
|
| 21158 |
+
"loss": 0.9329,
|
| 21159 |
+
"step": 3018
|
| 21160 |
+
},
|
| 21161 |
+
{
|
| 21162 |
+
"epoch": 0.6708888888888889,
|
| 21163 |
+
"grad_norm": 1.0144321918487549,
|
| 21164 |
+
"learning_rate": 6.596881959910914e-05,
|
| 21165 |
+
"loss": 2.3309,
|
| 21166 |
+
"step": 3019
|
| 21167 |
+
},
|
| 21168 |
+
{
|
| 21169 |
+
"epoch": 0.6711111111111111,
|
| 21170 |
+
"grad_norm": 0.9434064030647278,
|
| 21171 |
+
"learning_rate": 6.592427616926503e-05,
|
| 21172 |
+
"loss": 1.841,
|
| 21173 |
+
"step": 3020
|
| 21174 |
+
},
|
| 21175 |
+
{
|
| 21176 |
+
"epoch": 0.6713333333333333,
|
| 21177 |
+
"grad_norm": 0.9861494302749634,
|
| 21178 |
+
"learning_rate": 6.587973273942093e-05,
|
| 21179 |
+
"loss": 2.0507,
|
| 21180 |
+
"step": 3021
|
| 21181 |
+
},
|
| 21182 |
+
{
|
| 21183 |
+
"epoch": 0.6715555555555556,
|
| 21184 |
+
"grad_norm": 1.0820823907852173,
|
| 21185 |
+
"learning_rate": 6.583518930957684e-05,
|
| 21186 |
+
"loss": 1.5992,
|
| 21187 |
+
"step": 3022
|
| 21188 |
+
},
|
| 21189 |
+
{
|
| 21190 |
+
"epoch": 0.6717777777777778,
|
| 21191 |
+
"grad_norm": 0.6171742677688599,
|
| 21192 |
+
"learning_rate": 6.579064587973273e-05,
|
| 21193 |
+
"loss": 0.876,
|
| 21194 |
+
"step": 3023
|
| 21195 |
+
},
|
| 21196 |
+
{
|
| 21197 |
+
"epoch": 0.672,
|
| 21198 |
+
"grad_norm": 0.07197222858667374,
|
| 21199 |
+
"learning_rate": 6.574610244988865e-05,
|
| 21200 |
+
"loss": 0.0164,
|
| 21201 |
+
"step": 3024
|
| 21202 |
+
},
|
| 21203 |
+
{
|
| 21204 |
+
"epoch": 0.6722222222222223,
|
| 21205 |
+
"grad_norm": 0.7509397268295288,
|
| 21206 |
+
"learning_rate": 6.570155902004454e-05,
|
| 21207 |
+
"loss": 0.9635,
|
| 21208 |
+
"step": 3025
|
| 21209 |
+
},
|
| 21210 |
+
{
|
| 21211 |
+
"epoch": 0.6724444444444444,
|
| 21212 |
+
"grad_norm": 0.9577994346618652,
|
| 21213 |
+
"learning_rate": 6.565701559020045e-05,
|
| 21214 |
+
"loss": 1.7846,
|
| 21215 |
+
"step": 3026
|
| 21216 |
+
},
|
| 21217 |
+
{
|
| 21218 |
+
"epoch": 0.6726666666666666,
|
| 21219 |
+
"grad_norm": 0.9909307360649109,
|
| 21220 |
+
"learning_rate": 6.561247216035635e-05,
|
| 21221 |
+
"loss": 1.7811,
|
| 21222 |
+
"step": 3027
|
| 21223 |
+
},
|
| 21224 |
+
{
|
| 21225 |
+
"epoch": 0.6728888888888889,
|
| 21226 |
+
"grad_norm": 1.076392412185669,
|
| 21227 |
+
"learning_rate": 6.556792873051225e-05,
|
| 21228 |
+
"loss": 1.6848,
|
| 21229 |
+
"step": 3028
|
| 21230 |
+
},
|
| 21231 |
+
{
|
| 21232 |
+
"epoch": 0.6731111111111111,
|
| 21233 |
+
"grad_norm": 0.9113189578056335,
|
| 21234 |
+
"learning_rate": 6.552338530066815e-05,
|
| 21235 |
+
"loss": 1.7005,
|
| 21236 |
+
"step": 3029
|
| 21237 |
+
},
|
| 21238 |
+
{
|
| 21239 |
+
"epoch": 0.6733333333333333,
|
| 21240 |
+
"grad_norm": 1.2456274032592773,
|
| 21241 |
+
"learning_rate": 6.547884187082406e-05,
|
| 21242 |
+
"loss": 1.7526,
|
| 21243 |
+
"step": 3030
|
| 21244 |
+
},
|
| 21245 |
+
{
|
| 21246 |
+
"epoch": 0.6735555555555556,
|
| 21247 |
+
"grad_norm": 1.0734461545944214,
|
| 21248 |
+
"learning_rate": 6.543429844097996e-05,
|
| 21249 |
+
"loss": 1.9902,
|
| 21250 |
+
"step": 3031
|
| 21251 |
+
},
|
| 21252 |
+
{
|
| 21253 |
+
"epoch": 0.6737777777777778,
|
| 21254 |
+
"grad_norm": 0.7110247015953064,
|
| 21255 |
+
"learning_rate": 6.538975501113585e-05,
|
| 21256 |
+
"loss": 0.8453,
|
| 21257 |
+
"step": 3032
|
| 21258 |
+
},
|
| 21259 |
+
{
|
| 21260 |
+
"epoch": 0.674,
|
| 21261 |
+
"grad_norm": 0.059201423078775406,
|
| 21262 |
+
"learning_rate": 6.534521158129177e-05,
|
| 21263 |
+
"loss": 0.0163,
|
| 21264 |
+
"step": 3033
|
| 21265 |
+
},
|
| 21266 |
+
{
|
| 21267 |
+
"epoch": 0.6742222222222222,
|
| 21268 |
+
"grad_norm": 0.7983320355415344,
|
| 21269 |
+
"learning_rate": 6.530066815144766e-05,
|
| 21270 |
+
"loss": 0.9296,
|
| 21271 |
+
"step": 3034
|
| 21272 |
+
},
|
| 21273 |
+
{
|
| 21274 |
+
"epoch": 0.6744444444444444,
|
| 21275 |
+
"grad_norm": 0.07402991503477097,
|
| 21276 |
+
"learning_rate": 6.525612472160356e-05,
|
| 21277 |
+
"loss": 0.0193,
|
| 21278 |
+
"step": 3035
|
| 21279 |
+
},
|
| 21280 |
+
{
|
| 21281 |
+
"epoch": 0.6746666666666666,
|
| 21282 |
+
"grad_norm": 0.07244686037302017,
|
| 21283 |
+
"learning_rate": 6.521158129175947e-05,
|
| 21284 |
+
"loss": 0.0193,
|
| 21285 |
+
"step": 3036
|
| 21286 |
+
},
|
| 21287 |
+
{
|
| 21288 |
+
"epoch": 0.6748888888888889,
|
| 21289 |
+
"grad_norm": 1.0383340120315552,
|
| 21290 |
+
"learning_rate": 6.516703786191537e-05,
|
| 21291 |
+
"loss": 1.4567,
|
| 21292 |
+
"step": 3037
|
| 21293 |
+
},
|
| 21294 |
+
{
|
| 21295 |
+
"epoch": 0.6751111111111111,
|
| 21296 |
+
"grad_norm": 1.0180835723876953,
|
| 21297 |
+
"learning_rate": 6.512249443207127e-05,
|
| 21298 |
+
"loss": 1.8275,
|
| 21299 |
+
"step": 3038
|
| 21300 |
+
},
|
| 21301 |
+
{
|
| 21302 |
+
"epoch": 0.6753333333333333,
|
| 21303 |
+
"grad_norm": 1.225290298461914,
|
| 21304 |
+
"learning_rate": 6.507795100222718e-05,
|
| 21305 |
+
"loss": 1.7056,
|
| 21306 |
+
"step": 3039
|
| 21307 |
+
},
|
| 21308 |
+
{
|
| 21309 |
+
"epoch": 0.6755555555555556,
|
| 21310 |
+
"grad_norm": 0.8802182674407959,
|
| 21311 |
+
"learning_rate": 6.503340757238308e-05,
|
| 21312 |
+
"loss": 1.0935,
|
| 21313 |
+
"step": 3040
|
| 21314 |
+
},
|
| 21315 |
+
{
|
| 21316 |
+
"epoch": 0.6757777777777778,
|
| 21317 |
+
"grad_norm": 1.0758693218231201,
|
| 21318 |
+
"learning_rate": 6.498886414253897e-05,
|
| 21319 |
+
"loss": 1.7778,
|
| 21320 |
+
"step": 3041
|
| 21321 |
+
},
|
| 21322 |
+
{
|
| 21323 |
+
"epoch": 0.676,
|
| 21324 |
+
"grad_norm": 1.1325352191925049,
|
| 21325 |
+
"learning_rate": 6.494432071269488e-05,
|
| 21326 |
+
"loss": 1.5706,
|
| 21327 |
+
"step": 3042
|
| 21328 |
+
},
|
| 21329 |
+
{
|
| 21330 |
+
"epoch": 0.6762222222222222,
|
| 21331 |
+
"grad_norm": 1.0380780696868896,
|
| 21332 |
+
"learning_rate": 6.489977728285078e-05,
|
| 21333 |
+
"loss": 1.535,
|
| 21334 |
+
"step": 3043
|
| 21335 |
+
},
|
| 21336 |
+
{
|
| 21337 |
+
"epoch": 0.6764444444444444,
|
| 21338 |
+
"grad_norm": 0.9906545281410217,
|
| 21339 |
+
"learning_rate": 6.485523385300668e-05,
|
| 21340 |
+
"loss": 1.4007,
|
| 21341 |
+
"step": 3044
|
| 21342 |
+
},
|
| 21343 |
+
{
|
| 21344 |
+
"epoch": 0.6766666666666666,
|
| 21345 |
+
"grad_norm": 0.17783640325069427,
|
| 21346 |
+
"learning_rate": 6.48106904231626e-05,
|
| 21347 |
+
"loss": 0.0312,
|
| 21348 |
+
"step": 3045
|
| 21349 |
+
},
|
| 21350 |
+
{
|
| 21351 |
+
"epoch": 0.6768888888888889,
|
| 21352 |
+
"grad_norm": 0.9812122583389282,
|
| 21353 |
+
"learning_rate": 6.476614699331849e-05,
|
| 21354 |
+
"loss": 1.2594,
|
| 21355 |
+
"step": 3046
|
| 21356 |
+
},
|
| 21357 |
+
{
|
| 21358 |
+
"epoch": 0.6771111111111111,
|
| 21359 |
+
"grad_norm": 1.060013771057129,
|
| 21360 |
+
"learning_rate": 6.472160356347439e-05,
|
| 21361 |
+
"loss": 1.1064,
|
| 21362 |
+
"step": 3047
|
| 21363 |
+
},
|
| 21364 |
+
{
|
| 21365 |
+
"epoch": 0.6773333333333333,
|
| 21366 |
+
"grad_norm": 0.6272473335266113,
|
| 21367 |
+
"learning_rate": 6.46770601336303e-05,
|
| 21368 |
+
"loss": 0.4809,
|
| 21369 |
+
"step": 3048
|
| 21370 |
+
},
|
| 21371 |
+
{
|
| 21372 |
+
"epoch": 0.6775555555555556,
|
| 21373 |
+
"grad_norm": 0.9412599802017212,
|
| 21374 |
+
"learning_rate": 6.463251670378619e-05,
|
| 21375 |
+
"loss": 0.9192,
|
| 21376 |
+
"step": 3049
|
| 21377 |
+
},
|
| 21378 |
+
{
|
| 21379 |
+
"epoch": 0.6777777777777778,
|
| 21380 |
+
"grad_norm": 1.0236815214157104,
|
| 21381 |
+
"learning_rate": 6.45879732739421e-05,
|
| 21382 |
+
"loss": 0.8751,
|
| 21383 |
+
"step": 3050
|
| 21384 |
+
},
|
| 21385 |
+
{
|
| 21386 |
+
"epoch": 0.678,
|
| 21387 |
+
"grad_norm": 0.05509922653436661,
|
| 21388 |
+
"learning_rate": 6.4543429844098e-05,
|
| 21389 |
+
"loss": 0.0112,
|
| 21390 |
+
"step": 3051
|
| 21391 |
+
},
|
| 21392 |
+
{
|
| 21393 |
+
"epoch": 0.6782222222222222,
|
| 21394 |
+
"grad_norm": 0.8858595490455627,
|
| 21395 |
+
"learning_rate": 6.44988864142539e-05,
|
| 21396 |
+
"loss": 2.2657,
|
| 21397 |
+
"step": 3052
|
| 21398 |
+
},
|
| 21399 |
+
{
|
| 21400 |
+
"epoch": 0.6784444444444444,
|
| 21401 |
+
"grad_norm": 0.5961353182792664,
|
| 21402 |
+
"learning_rate": 6.44543429844098e-05,
|
| 21403 |
+
"loss": 1.2772,
|
| 21404 |
+
"step": 3053
|
| 21405 |
+
},
|
| 21406 |
+
{
|
| 21407 |
+
"epoch": 0.6786666666666666,
|
| 21408 |
+
"grad_norm": 0.05628953129053116,
|
| 21409 |
+
"learning_rate": 6.44097995545657e-05,
|
| 21410 |
+
"loss": 0.011,
|
| 21411 |
+
"step": 3054
|
| 21412 |
+
},
|
| 21413 |
+
{
|
| 21414 |
+
"epoch": 0.6788888888888889,
|
| 21415 |
+
"grad_norm": 0.8051088452339172,
|
| 21416 |
+
"learning_rate": 6.436525612472161e-05,
|
| 21417 |
+
"loss": 2.2465,
|
| 21418 |
+
"step": 3055
|
| 21419 |
+
},
|
| 21420 |
+
{
|
| 21421 |
+
"epoch": 0.6791111111111111,
|
| 21422 |
+
"grad_norm": 0.9172492027282715,
|
| 21423 |
+
"learning_rate": 6.43207126948775e-05,
|
| 21424 |
+
"loss": 2.1671,
|
| 21425 |
+
"step": 3056
|
| 21426 |
+
},
|
| 21427 |
+
{
|
| 21428 |
+
"epoch": 0.6793333333333333,
|
| 21429 |
+
"grad_norm": 0.622685968875885,
|
| 21430 |
+
"learning_rate": 6.427616926503342e-05,
|
| 21431 |
+
"loss": 1.1392,
|
| 21432 |
+
"step": 3057
|
| 21433 |
+
},
|
| 21434 |
+
{
|
| 21435 |
+
"epoch": 0.6795555555555556,
|
| 21436 |
+
"grad_norm": 1.047365665435791,
|
| 21437 |
+
"learning_rate": 6.423162583518931e-05,
|
| 21438 |
+
"loss": 2.1024,
|
| 21439 |
+
"step": 3058
|
| 21440 |
+
},
|
| 21441 |
+
{
|
| 21442 |
+
"epoch": 0.6797777777777778,
|
| 21443 |
+
"grad_norm": 0.6596314311027527,
|
| 21444 |
+
"learning_rate": 6.418708240534522e-05,
|
| 21445 |
+
"loss": 1.2222,
|
| 21446 |
+
"step": 3059
|
| 21447 |
+
},
|
| 21448 |
+
{
|
| 21449 |
+
"epoch": 0.68,
|
| 21450 |
+
"grad_norm": 0.058744728565216064,
|
| 21451 |
+
"learning_rate": 6.414253897550112e-05,
|
| 21452 |
+
"loss": 0.0111,
|
| 21453 |
+
"step": 3060
|
| 21454 |
+
},
|
| 21455 |
+
{
|
| 21456 |
+
"epoch": 0.6802222222222222,
|
| 21457 |
+
"grad_norm": 0.9326651692390442,
|
| 21458 |
+
"learning_rate": 6.409799554565701e-05,
|
| 21459 |
+
"loss": 2.1789,
|
| 21460 |
+
"step": 3061
|
| 21461 |
+
},
|
| 21462 |
+
{
|
| 21463 |
+
"epoch": 0.6804444444444444,
|
| 21464 |
+
"grad_norm": 0.9691800475120544,
|
| 21465 |
+
"learning_rate": 6.405345211581292e-05,
|
| 21466 |
+
"loss": 2.2824,
|
| 21467 |
+
"step": 3062
|
| 21468 |
+
},
|
| 21469 |
+
{
|
| 21470 |
+
"epoch": 0.6806666666666666,
|
| 21471 |
+
"grad_norm": 0.925193190574646,
|
| 21472 |
+
"learning_rate": 6.400890868596882e-05,
|
| 21473 |
+
"loss": 2.431,
|
| 21474 |
+
"step": 3063
|
| 21475 |
+
},
|
| 21476 |
+
{
|
| 21477 |
+
"epoch": 0.6808888888888889,
|
| 21478 |
+
"grad_norm": 0.9088225364685059,
|
| 21479 |
+
"learning_rate": 6.396436525612473e-05,
|
| 21480 |
+
"loss": 2.0492,
|
| 21481 |
+
"step": 3064
|
| 21482 |
+
},
|
| 21483 |
+
{
|
| 21484 |
+
"epoch": 0.6811111111111111,
|
| 21485 |
+
"grad_norm": 0.8372054696083069,
|
| 21486 |
+
"learning_rate": 6.391982182628062e-05,
|
| 21487 |
+
"loss": 2.2126,
|
| 21488 |
+
"step": 3065
|
| 21489 |
+
},
|
| 21490 |
+
{
|
| 21491 |
+
"epoch": 0.6813333333333333,
|
| 21492 |
+
"grad_norm": 0.8477223515510559,
|
| 21493 |
+
"learning_rate": 6.387527839643653e-05,
|
| 21494 |
+
"loss": 1.933,
|
| 21495 |
+
"step": 3066
|
| 21496 |
+
},
|
| 21497 |
+
{
|
| 21498 |
+
"epoch": 0.6815555555555556,
|
| 21499 |
+
"grad_norm": 0.6637649536132812,
|
| 21500 |
+
"learning_rate": 6.383073496659243e-05,
|
| 21501 |
+
"loss": 1.0526,
|
| 21502 |
+
"step": 3067
|
| 21503 |
+
},
|
| 21504 |
+
{
|
| 21505 |
+
"epoch": 0.6817777777777778,
|
| 21506 |
+
"grad_norm": 0.9227988719940186,
|
| 21507 |
+
"learning_rate": 6.378619153674832e-05,
|
| 21508 |
+
"loss": 1.9949,
|
| 21509 |
+
"step": 3068
|
| 21510 |
+
},
|
| 21511 |
+
{
|
| 21512 |
+
"epoch": 0.682,
|
| 21513 |
+
"grad_norm": 0.9380735754966736,
|
| 21514 |
+
"learning_rate": 6.374164810690424e-05,
|
| 21515 |
+
"loss": 1.8282,
|
| 21516 |
+
"step": 3069
|
| 21517 |
+
},
|
| 21518 |
+
{
|
| 21519 |
+
"epoch": 0.6822222222222222,
|
| 21520 |
+
"grad_norm": 0.992690920829773,
|
| 21521 |
+
"learning_rate": 6.369710467706013e-05,
|
| 21522 |
+
"loss": 1.7263,
|
| 21523 |
+
"step": 3070
|
| 21524 |
+
},
|
| 21525 |
+
{
|
| 21526 |
+
"epoch": 0.6824444444444444,
|
| 21527 |
+
"grad_norm": 0.9732444286346436,
|
| 21528 |
+
"learning_rate": 6.365256124721604e-05,
|
| 21529 |
+
"loss": 2.0818,
|
| 21530 |
+
"step": 3071
|
| 21531 |
+
},
|
| 21532 |
+
{
|
| 21533 |
+
"epoch": 0.6826666666666666,
|
| 21534 |
+
"grad_norm": 0.07128032296895981,
|
| 21535 |
+
"learning_rate": 6.360801781737195e-05,
|
| 21536 |
+
"loss": 0.0167,
|
| 21537 |
+
"step": 3072
|
| 21538 |
+
},
|
| 21539 |
+
{
|
| 21540 |
+
"epoch": 0.6828888888888889,
|
| 21541 |
+
"grad_norm": 0.9442581534385681,
|
| 21542 |
+
"learning_rate": 6.356347438752784e-05,
|
| 21543 |
+
"loss": 1.9286,
|
| 21544 |
+
"step": 3073
|
| 21545 |
+
},
|
| 21546 |
+
{
|
| 21547 |
+
"epoch": 0.6831111111111111,
|
| 21548 |
+
"grad_norm": 0.13606970012187958,
|
| 21549 |
+
"learning_rate": 6.351893095768374e-05,
|
| 21550 |
+
"loss": 0.0205,
|
| 21551 |
+
"step": 3074
|
| 21552 |
+
},
|
| 21553 |
+
{
|
| 21554 |
+
"epoch": 0.6833333333333333,
|
| 21555 |
+
"grad_norm": 0.648127555847168,
|
| 21556 |
+
"learning_rate": 6.347438752783965e-05,
|
| 21557 |
+
"loss": 0.8501,
|
| 21558 |
+
"step": 3075
|
| 21559 |
+
},
|
| 21560 |
+
{
|
| 21561 |
+
"epoch": 0.6835555555555556,
|
| 21562 |
+
"grad_norm": 0.9328134655952454,
|
| 21563 |
+
"learning_rate": 6.342984409799555e-05,
|
| 21564 |
+
"loss": 1.6076,
|
| 21565 |
+
"step": 3076
|
| 21566 |
+
},
|
| 21567 |
+
{
|
| 21568 |
+
"epoch": 0.6837777777777778,
|
| 21569 |
+
"grad_norm": 1.0561175346374512,
|
| 21570 |
+
"learning_rate": 6.338530066815144e-05,
|
| 21571 |
+
"loss": 1.765,
|
| 21572 |
+
"step": 3077
|
| 21573 |
+
},
|
| 21574 |
+
{
|
| 21575 |
+
"epoch": 0.684,
|
| 21576 |
+
"grad_norm": 1.1816853284835815,
|
| 21577 |
+
"learning_rate": 6.334075723830736e-05,
|
| 21578 |
+
"loss": 2.1397,
|
| 21579 |
+
"step": 3078
|
| 21580 |
+
},
|
| 21581 |
+
{
|
| 21582 |
+
"epoch": 0.6842222222222222,
|
| 21583 |
+
"grad_norm": 1.151865839958191,
|
| 21584 |
+
"learning_rate": 6.329621380846326e-05,
|
| 21585 |
+
"loss": 1.8881,
|
| 21586 |
+
"step": 3079
|
| 21587 |
+
},
|
| 21588 |
+
{
|
| 21589 |
+
"epoch": 0.6844444444444444,
|
| 21590 |
+
"grad_norm": 0.07445438951253891,
|
| 21591 |
+
"learning_rate": 6.325167037861916e-05,
|
| 21592 |
+
"loss": 0.0177,
|
| 21593 |
+
"step": 3080
|
| 21594 |
+
},
|
| 21595 |
+
{
|
| 21596 |
+
"epoch": 0.6846666666666666,
|
| 21597 |
+
"grad_norm": 0.07042374461889267,
|
| 21598 |
+
"learning_rate": 6.320712694877507e-05,
|
| 21599 |
+
"loss": 0.0171,
|
| 21600 |
+
"step": 3081
|
| 21601 |
+
},
|
| 21602 |
+
{
|
| 21603 |
+
"epoch": 0.6848888888888889,
|
| 21604 |
+
"grad_norm": 0.6800836324691772,
|
| 21605 |
+
"learning_rate": 6.316258351893096e-05,
|
| 21606 |
+
"loss": 0.7686,
|
| 21607 |
+
"step": 3082
|
| 21608 |
+
},
|
| 21609 |
+
{
|
| 21610 |
+
"epoch": 0.6851111111111111,
|
| 21611 |
+
"grad_norm": 0.7797111868858337,
|
| 21612 |
+
"learning_rate": 6.311804008908686e-05,
|
| 21613 |
+
"loss": 0.9435,
|
| 21614 |
+
"step": 3083
|
| 21615 |
+
},
|
| 21616 |
+
{
|
| 21617 |
+
"epoch": 0.6853333333333333,
|
| 21618 |
+
"grad_norm": 0.07849026471376419,
|
| 21619 |
+
"learning_rate": 6.307349665924277e-05,
|
| 21620 |
+
"loss": 0.0194,
|
| 21621 |
+
"step": 3084
|
| 21622 |
+
},
|
| 21623 |
+
{
|
| 21624 |
+
"epoch": 0.6855555555555556,
|
| 21625 |
+
"grad_norm": 0.07662785053253174,
|
| 21626 |
+
"learning_rate": 6.302895322939867e-05,
|
| 21627 |
+
"loss": 0.0191,
|
| 21628 |
+
"step": 3085
|
| 21629 |
+
},
|
| 21630 |
+
{
|
| 21631 |
+
"epoch": 0.6857777777777778,
|
| 21632 |
+
"grad_norm": 0.0744476169347763,
|
| 21633 |
+
"learning_rate": 6.298440979955457e-05,
|
| 21634 |
+
"loss": 0.0183,
|
| 21635 |
+
"step": 3086
|
| 21636 |
+
},
|
| 21637 |
+
{
|
| 21638 |
+
"epoch": 0.686,
|
| 21639 |
+
"grad_norm": 0.0717550590634346,
|
| 21640 |
+
"learning_rate": 6.293986636971047e-05,
|
| 21641 |
+
"loss": 0.0184,
|
| 21642 |
+
"step": 3087
|
| 21643 |
+
},
|
| 21644 |
+
{
|
| 21645 |
+
"epoch": 0.6862222222222222,
|
| 21646 |
+
"grad_norm": 1.115823745727539,
|
| 21647 |
+
"learning_rate": 6.289532293986638e-05,
|
| 21648 |
+
"loss": 1.871,
|
| 21649 |
+
"step": 3088
|
| 21650 |
+
},
|
| 21651 |
+
{
|
| 21652 |
+
"epoch": 0.6864444444444444,
|
| 21653 |
+
"grad_norm": 0.9394058585166931,
|
| 21654 |
+
"learning_rate": 6.285077951002227e-05,
|
| 21655 |
+
"loss": 1.2818,
|
| 21656 |
+
"step": 3089
|
| 21657 |
+
},
|
| 21658 |
+
{
|
| 21659 |
+
"epoch": 0.6866666666666666,
|
| 21660 |
+
"grad_norm": 0.7753637433052063,
|
| 21661 |
+
"learning_rate": 6.280623608017817e-05,
|
| 21662 |
+
"loss": 0.9434,
|
| 21663 |
+
"step": 3090
|
| 21664 |
+
},
|
| 21665 |
+
{
|
| 21666 |
+
"epoch": 0.6868888888888889,
|
| 21667 |
+
"grad_norm": 0.7117932438850403,
|
| 21668 |
+
"learning_rate": 6.276169265033408e-05,
|
| 21669 |
+
"loss": 0.7796,
|
| 21670 |
+
"step": 3091
|
| 21671 |
+
},
|
| 21672 |
+
{
|
| 21673 |
+
"epoch": 0.6871111111111111,
|
| 21674 |
+
"grad_norm": 1.1098551750183105,
|
| 21675 |
+
"learning_rate": 6.271714922048998e-05,
|
| 21676 |
+
"loss": 1.3902,
|
| 21677 |
+
"step": 3092
|
| 21678 |
+
},
|
| 21679 |
+
{
|
| 21680 |
+
"epoch": 0.6873333333333334,
|
| 21681 |
+
"grad_norm": 1.1206355094909668,
|
| 21682 |
+
"learning_rate": 6.267260579064588e-05,
|
| 21683 |
+
"loss": 1.4423,
|
| 21684 |
+
"step": 3093
|
| 21685 |
+
},
|
| 21686 |
+
{
|
| 21687 |
+
"epoch": 0.6875555555555556,
|
| 21688 |
+
"grad_norm": 1.0141700506210327,
|
| 21689 |
+
"learning_rate": 6.262806236080178e-05,
|
| 21690 |
+
"loss": 1.587,
|
| 21691 |
+
"step": 3094
|
| 21692 |
+
},
|
| 21693 |
+
{
|
| 21694 |
+
"epoch": 0.6877777777777778,
|
| 21695 |
+
"grad_norm": 1.2595239877700806,
|
| 21696 |
+
"learning_rate": 6.258351893095769e-05,
|
| 21697 |
+
"loss": 1.5666,
|
| 21698 |
+
"step": 3095
|
| 21699 |
+
},
|
| 21700 |
+
{
|
| 21701 |
+
"epoch": 0.688,
|
| 21702 |
+
"grad_norm": 0.9674675464630127,
|
| 21703 |
+
"learning_rate": 6.253897550111358e-05,
|
| 21704 |
+
"loss": 0.7225,
|
| 21705 |
+
"step": 3096
|
| 21706 |
+
},
|
| 21707 |
+
{
|
| 21708 |
+
"epoch": 0.6882222222222222,
|
| 21709 |
+
"grad_norm": 0.1801719069480896,
|
| 21710 |
+
"learning_rate": 6.24944320712695e-05,
|
| 21711 |
+
"loss": 0.0303,
|
| 21712 |
+
"step": 3097
|
| 21713 |
+
},
|
| 21714 |
+
{
|
| 21715 |
+
"epoch": 0.6884444444444444,
|
| 21716 |
+
"grad_norm": 1.2141374349594116,
|
| 21717 |
+
"learning_rate": 6.244988864142539e-05,
|
| 21718 |
+
"loss": 1.0987,
|
| 21719 |
+
"step": 3098
|
| 21720 |
+
},
|
| 21721 |
+
{
|
| 21722 |
+
"epoch": 0.6886666666666666,
|
| 21723 |
+
"grad_norm": 1.031459093093872,
|
| 21724 |
+
"learning_rate": 6.24053452115813e-05,
|
| 21725 |
+
"loss": 1.0589,
|
| 21726 |
+
"step": 3099
|
| 21727 |
+
},
|
| 21728 |
+
{
|
| 21729 |
+
"epoch": 0.6888888888888889,
|
| 21730 |
+
"grad_norm": 0.8709812760353088,
|
| 21731 |
+
"learning_rate": 6.23608017817372e-05,
|
| 21732 |
+
"loss": 0.5863,
|
| 21733 |
+
"step": 3100
|
| 21734 |
+
},
|
| 21735 |
+
{
|
| 21736 |
+
"epoch": 0.6891111111111111,
|
| 21737 |
+
"grad_norm": 0.607231616973877,
|
| 21738 |
+
"learning_rate": 6.231625835189309e-05,
|
| 21739 |
+
"loss": 1.0338,
|
| 21740 |
+
"step": 3101
|
| 21741 |
+
},
|
| 21742 |
+
{
|
| 21743 |
+
"epoch": 0.6893333333333334,
|
| 21744 |
+
"grad_norm": 0.5942530035972595,
|
| 21745 |
+
"learning_rate": 6.2271714922049e-05,
|
| 21746 |
+
"loss": 1.0758,
|
| 21747 |
+
"step": 3102
|
| 21748 |
+
},
|
| 21749 |
+
{
|
| 21750 |
+
"epoch": 0.6895555555555556,
|
| 21751 |
+
"grad_norm": 0.8875899910926819,
|
| 21752 |
+
"learning_rate": 6.22271714922049e-05,
|
| 21753 |
+
"loss": 2.1266,
|
| 21754 |
+
"step": 3103
|
| 21755 |
+
},
|
| 21756 |
+
{
|
| 21757 |
+
"epoch": 0.6897777777777778,
|
| 21758 |
+
"grad_norm": 0.8262476325035095,
|
| 21759 |
+
"learning_rate": 6.218262806236081e-05,
|
| 21760 |
+
"loss": 2.2635,
|
| 21761 |
+
"step": 3104
|
| 21762 |
+
},
|
| 21763 |
+
{
|
| 21764 |
+
"epoch": 0.69,
|
| 21765 |
+
"grad_norm": 0.0524749718606472,
|
| 21766 |
+
"learning_rate": 6.21380846325167e-05,
|
| 21767 |
+
"loss": 0.0106,
|
| 21768 |
+
"step": 3105
|
| 21769 |
+
},
|
| 21770 |
+
{
|
| 21771 |
+
"epoch": 0.6902222222222222,
|
| 21772 |
+
"grad_norm": 0.5803321003913879,
|
| 21773 |
+
"learning_rate": 6.20935412026726e-05,
|
| 21774 |
+
"loss": 1.0884,
|
| 21775 |
+
"step": 3106
|
| 21776 |
+
},
|
| 21777 |
+
{
|
| 21778 |
+
"epoch": 0.6904444444444444,
|
| 21779 |
+
"grad_norm": 1.0942132472991943,
|
| 21780 |
+
"learning_rate": 6.204899777282851e-05,
|
| 21781 |
+
"loss": 2.2134,
|
| 21782 |
+
"step": 3107
|
| 21783 |
+
},
|
| 21784 |
+
{
|
| 21785 |
+
"epoch": 0.6906666666666667,
|
| 21786 |
+
"grad_norm": 0.07692880928516388,
|
| 21787 |
+
"learning_rate": 6.20044543429844e-05,
|
| 21788 |
+
"loss": 0.0116,
|
| 21789 |
+
"step": 3108
|
| 21790 |
+
},
|
| 21791 |
+
{
|
| 21792 |
+
"epoch": 0.6908888888888889,
|
| 21793 |
+
"grad_norm": 0.07232845574617386,
|
| 21794 |
+
"learning_rate": 6.195991091314032e-05,
|
| 21795 |
+
"loss": 0.0117,
|
| 21796 |
+
"step": 3109
|
| 21797 |
+
},
|
| 21798 |
+
{
|
| 21799 |
+
"epoch": 0.6911111111111111,
|
| 21800 |
+
"grad_norm": 0.8545564413070679,
|
| 21801 |
+
"learning_rate": 6.191536748329621e-05,
|
| 21802 |
+
"loss": 2.0441,
|
| 21803 |
+
"step": 3110
|
| 21804 |
+
},
|
| 21805 |
+
{
|
| 21806 |
+
"epoch": 0.6913333333333334,
|
| 21807 |
+
"grad_norm": 0.8483017086982727,
|
| 21808 |
+
"learning_rate": 6.187082405345212e-05,
|
| 21809 |
+
"loss": 1.9926,
|
| 21810 |
+
"step": 3111
|
| 21811 |
+
},
|
| 21812 |
+
{
|
| 21813 |
+
"epoch": 0.6915555555555556,
|
| 21814 |
+
"grad_norm": 0.8519989848136902,
|
| 21815 |
+
"learning_rate": 6.182628062360802e-05,
|
| 21816 |
+
"loss": 2.1871,
|
| 21817 |
+
"step": 3112
|
| 21818 |
+
},
|
| 21819 |
+
{
|
| 21820 |
+
"epoch": 0.6917777777777778,
|
| 21821 |
+
"grad_norm": 0.8962295055389404,
|
| 21822 |
+
"learning_rate": 6.178173719376392e-05,
|
| 21823 |
+
"loss": 1.9079,
|
| 21824 |
+
"step": 3113
|
| 21825 |
+
},
|
| 21826 |
+
{
|
| 21827 |
+
"epoch": 0.692,
|
| 21828 |
+
"grad_norm": 0.908099353313446,
|
| 21829 |
+
"learning_rate": 6.173719376391982e-05,
|
| 21830 |
+
"loss": 2.2056,
|
| 21831 |
+
"step": 3114
|
| 21832 |
+
},
|
| 21833 |
+
{
|
| 21834 |
+
"epoch": 0.6922222222222222,
|
| 21835 |
+
"grad_norm": 0.9471180438995361,
|
| 21836 |
+
"learning_rate": 6.169265033407573e-05,
|
| 21837 |
+
"loss": 1.6695,
|
| 21838 |
+
"step": 3115
|
| 21839 |
+
},
|
| 21840 |
+
{
|
| 21841 |
+
"epoch": 0.6924444444444444,
|
| 21842 |
+
"grad_norm": 0.9277594685554504,
|
| 21843 |
+
"learning_rate": 6.164810690423163e-05,
|
| 21844 |
+
"loss": 2.0879,
|
| 21845 |
+
"step": 3116
|
| 21846 |
+
},
|
| 21847 |
+
{
|
| 21848 |
+
"epoch": 0.6926666666666667,
|
| 21849 |
+
"grad_norm": 0.6673265695571899,
|
| 21850 |
+
"learning_rate": 6.160356347438752e-05,
|
| 21851 |
+
"loss": 0.9213,
|
| 21852 |
+
"step": 3117
|
| 21853 |
+
},
|
| 21854 |
+
{
|
| 21855 |
+
"epoch": 0.6928888888888889,
|
| 21856 |
+
"grad_norm": 0.12814414501190186,
|
| 21857 |
+
"learning_rate": 6.155902004454344e-05,
|
| 21858 |
+
"loss": 0.0211,
|
| 21859 |
+
"step": 3118
|
| 21860 |
+
},
|
| 21861 |
+
{
|
| 21862 |
+
"epoch": 0.6931111111111111,
|
| 21863 |
+
"grad_norm": 0.9539985656738281,
|
| 21864 |
+
"learning_rate": 6.151447661469933e-05,
|
| 21865 |
+
"loss": 1.9212,
|
| 21866 |
+
"step": 3119
|
| 21867 |
+
},
|
| 21868 |
+
{
|
| 21869 |
+
"epoch": 0.6933333333333334,
|
| 21870 |
+
"grad_norm": 0.927853524684906,
|
| 21871 |
+
"learning_rate": 6.146993318485523e-05,
|
| 21872 |
+
"loss": 1.6054,
|
| 21873 |
+
"step": 3120
|
| 21874 |
+
},
|
| 21875 |
+
{
|
| 21876 |
+
"epoch": 0.6935555555555556,
|
| 21877 |
+
"grad_norm": 0.6636569499969482,
|
| 21878 |
+
"learning_rate": 6.142538975501115e-05,
|
| 21879 |
+
"loss": 0.9331,
|
| 21880 |
+
"step": 3121
|
| 21881 |
+
},
|
| 21882 |
+
{
|
| 21883 |
+
"epoch": 0.6937777777777778,
|
| 21884 |
+
"grad_norm": 0.07317844778299332,
|
| 21885 |
+
"learning_rate": 6.138084632516704e-05,
|
| 21886 |
+
"loss": 0.0166,
|
| 21887 |
+
"step": 3122
|
| 21888 |
+
},
|
| 21889 |
+
{
|
| 21890 |
+
"epoch": 0.694,
|
| 21891 |
+
"grad_norm": 0.07253949344158173,
|
| 21892 |
+
"learning_rate": 6.133630289532294e-05,
|
| 21893 |
+
"loss": 0.0165,
|
| 21894 |
+
"step": 3123
|
| 21895 |
+
},
|
| 21896 |
+
{
|
| 21897 |
+
"epoch": 0.6942222222222222,
|
| 21898 |
+
"grad_norm": 0.07455820590257645,
|
| 21899 |
+
"learning_rate": 6.129175946547885e-05,
|
| 21900 |
+
"loss": 0.0168,
|
| 21901 |
+
"step": 3124
|
| 21902 |
+
},
|
| 21903 |
+
{
|
| 21904 |
+
"epoch": 0.6944444444444444,
|
| 21905 |
+
"grad_norm": 0.7180811762809753,
|
| 21906 |
+
"learning_rate": 6.124721603563475e-05,
|
| 21907 |
+
"loss": 1.3197,
|
| 21908 |
+
"step": 3125
|
| 21909 |
+
},
|
| 21910 |
+
{
|
| 21911 |
+
"epoch": 0.6946666666666667,
|
| 21912 |
+
"grad_norm": 1.0325121879577637,
|
| 21913 |
+
"learning_rate": 6.120267260579064e-05,
|
| 21914 |
+
"loss": 1.8146,
|
| 21915 |
+
"step": 3126
|
| 21916 |
+
},
|
| 21917 |
+
{
|
| 21918 |
+
"epoch": 0.6948888888888889,
|
| 21919 |
+
"grad_norm": 1.0472650527954102,
|
| 21920 |
+
"learning_rate": 6.115812917594655e-05,
|
| 21921 |
+
"loss": 1.8477,
|
| 21922 |
+
"step": 3127
|
| 21923 |
+
},
|
| 21924 |
+
{
|
| 21925 |
+
"epoch": 0.6951111111111111,
|
| 21926 |
+
"grad_norm": 1.3057109117507935,
|
| 21927 |
+
"learning_rate": 6.111358574610246e-05,
|
| 21928 |
+
"loss": 1.6522,
|
| 21929 |
+
"step": 3128
|
| 21930 |
+
},
|
| 21931 |
+
{
|
| 21932 |
+
"epoch": 0.6953333333333334,
|
| 21933 |
+
"grad_norm": 0.9642925262451172,
|
| 21934 |
+
"learning_rate": 6.106904231625835e-05,
|
| 21935 |
+
"loss": 1.9227,
|
| 21936 |
+
"step": 3129
|
| 21937 |
+
},
|
| 21938 |
+
{
|
| 21939 |
+
"epoch": 0.6955555555555556,
|
| 21940 |
+
"grad_norm": 0.9852336049079895,
|
| 21941 |
+
"learning_rate": 6.102449888641426e-05,
|
| 21942 |
+
"loss": 1.8925,
|
| 21943 |
+
"step": 3130
|
| 21944 |
+
},
|
| 21945 |
+
{
|
| 21946 |
+
"epoch": 0.6957777777777778,
|
| 21947 |
+
"grad_norm": 0.0911262258887291,
|
| 21948 |
+
"learning_rate": 6.097995545657016e-05,
|
| 21949 |
+
"loss": 0.0171,
|
| 21950 |
+
"step": 3131
|
| 21951 |
+
},
|
| 21952 |
+
{
|
| 21953 |
+
"epoch": 0.696,
|
| 21954 |
+
"grad_norm": 0.6741465330123901,
|
| 21955 |
+
"learning_rate": 6.093541202672606e-05,
|
| 21956 |
+
"loss": 0.8653,
|
| 21957 |
+
"step": 3132
|
| 21958 |
+
},
|
| 21959 |
+
{
|
| 21960 |
+
"epoch": 0.6962222222222222,
|
| 21961 |
+
"grad_norm": 0.9752011895179749,
|
| 21962 |
+
"learning_rate": 6.089086859688197e-05,
|
| 21963 |
+
"loss": 1.4045,
|
| 21964 |
+
"step": 3133
|
| 21965 |
+
},
|
| 21966 |
+
{
|
| 21967 |
+
"epoch": 0.6964444444444444,
|
| 21968 |
+
"grad_norm": 0.07129085063934326,
|
| 21969 |
+
"learning_rate": 6.084632516703787e-05,
|
| 21970 |
+
"loss": 0.0192,
|
| 21971 |
+
"step": 3134
|
| 21972 |
+
},
|
| 21973 |
+
{
|
| 21974 |
+
"epoch": 0.6966666666666667,
|
| 21975 |
+
"grad_norm": 0.0695280209183693,
|
| 21976 |
+
"learning_rate": 6.0801781737193766e-05,
|
| 21977 |
+
"loss": 0.0186,
|
| 21978 |
+
"step": 3135
|
| 21979 |
+
},
|
| 21980 |
+
{
|
| 21981 |
+
"epoch": 0.6968888888888889,
|
| 21982 |
+
"grad_norm": 0.07262587547302246,
|
| 21983 |
+
"learning_rate": 6.075723830734967e-05,
|
| 21984 |
+
"loss": 0.0183,
|
| 21985 |
+
"step": 3136
|
| 21986 |
+
},
|
| 21987 |
+
{
|
| 21988 |
+
"epoch": 0.6971111111111111,
|
| 21989 |
+
"grad_norm": 0.9831186532974243,
|
| 21990 |
+
"learning_rate": 6.071269487750557e-05,
|
| 21991 |
+
"loss": 1.4122,
|
| 21992 |
+
"step": 3137
|
| 21993 |
+
},
|
| 21994 |
+
{
|
| 21995 |
+
"epoch": 0.6973333333333334,
|
| 21996 |
+
"grad_norm": 0.9442914724349976,
|
| 21997 |
+
"learning_rate": 6.066815144766147e-05,
|
| 21998 |
+
"loss": 1.452,
|
| 21999 |
+
"step": 3138
|
| 22000 |
+
},
|
| 22001 |
+
{
|
| 22002 |
+
"epoch": 0.6975555555555556,
|
| 22003 |
+
"grad_norm": 1.1144623756408691,
|
| 22004 |
+
"learning_rate": 6.062360801781738e-05,
|
| 22005 |
+
"loss": 1.6665,
|
| 22006 |
+
"step": 3139
|
| 22007 |
+
},
|
| 22008 |
+
{
|
| 22009 |
+
"epoch": 0.6977777777777778,
|
| 22010 |
+
"grad_norm": 1.0614639520645142,
|
| 22011 |
+
"learning_rate": 6.057906458797328e-05,
|
| 22012 |
+
"loss": 1.5621,
|
| 22013 |
+
"step": 3140
|
| 22014 |
+
},
|
| 22015 |
+
{
|
| 22016 |
+
"epoch": 0.698,
|
| 22017 |
+
"grad_norm": 1.2415484189987183,
|
| 22018 |
+
"learning_rate": 6.053452115812918e-05,
|
| 22019 |
+
"loss": 1.5338,
|
| 22020 |
+
"step": 3141
|
| 22021 |
+
},
|
| 22022 |
+
{
|
| 22023 |
+
"epoch": 0.6982222222222222,
|
| 22024 |
+
"grad_norm": 1.146238923072815,
|
| 22025 |
+
"learning_rate": 6.048997772828508e-05,
|
| 22026 |
+
"loss": 1.8936,
|
| 22027 |
+
"step": 3142
|
| 22028 |
+
},
|
| 22029 |
+
{
|
| 22030 |
+
"epoch": 0.6984444444444444,
|
| 22031 |
+
"grad_norm": 1.1693158149719238,
|
| 22032 |
+
"learning_rate": 6.044543429844098e-05,
|
| 22033 |
+
"loss": 1.7386,
|
| 22034 |
+
"step": 3143
|
| 22035 |
+
},
|
| 22036 |
+
{
|
| 22037 |
+
"epoch": 0.6986666666666667,
|
| 22038 |
+
"grad_norm": 1.2340409755706787,
|
| 22039 |
+
"learning_rate": 6.040089086859688e-05,
|
| 22040 |
+
"loss": 1.4781,
|
| 22041 |
+
"step": 3144
|
| 22042 |
+
},
|
| 22043 |
+
{
|
| 22044 |
+
"epoch": 0.6988888888888889,
|
| 22045 |
+
"grad_norm": 1.0042845010757446,
|
| 22046 |
+
"learning_rate": 6.035634743875279e-05,
|
| 22047 |
+
"loss": 1.3662,
|
| 22048 |
+
"step": 3145
|
| 22049 |
+
},
|
| 22050 |
+
{
|
| 22051 |
+
"epoch": 0.6991111111111111,
|
| 22052 |
+
"grad_norm": 0.18454298377037048,
|
| 22053 |
+
"learning_rate": 6.031180400890869e-05,
|
| 22054 |
+
"loss": 0.0276,
|
| 22055 |
+
"step": 3146
|
| 22056 |
+
},
|
| 22057 |
+
{
|
| 22058 |
+
"epoch": 0.6993333333333334,
|
| 22059 |
+
"grad_norm": 1.1719262599945068,
|
| 22060 |
+
"learning_rate": 6.026726057906459e-05,
|
| 22061 |
+
"loss": 1.0601,
|
| 22062 |
+
"step": 3147
|
| 22063 |
+
},
|
| 22064 |
+
{
|
| 22065 |
+
"epoch": 0.6995555555555556,
|
| 22066 |
+
"grad_norm": 0.9232467412948608,
|
| 22067 |
+
"learning_rate": 6.0222717149220495e-05,
|
| 22068 |
+
"loss": 0.8415,
|
| 22069 |
+
"step": 3148
|
| 22070 |
+
},
|
| 22071 |
+
{
|
| 22072 |
+
"epoch": 0.6997777777777778,
|
| 22073 |
+
"grad_norm": 0.194104865193367,
|
| 22074 |
+
"learning_rate": 6.0178173719376394e-05,
|
| 22075 |
+
"loss": 0.0401,
|
| 22076 |
+
"step": 3149
|
| 22077 |
+
},
|
| 22078 |
+
{
|
| 22079 |
+
"epoch": 0.7,
|
| 22080 |
+
"grad_norm": 0.7421103119850159,
|
| 22081 |
+
"learning_rate": 6.013363028953229e-05,
|
| 22082 |
+
"loss": 0.6284,
|
| 22083 |
+
"step": 3150
|
| 22084 |
+
},
|
| 22085 |
+
{
|
| 22086 |
+
"epoch": 0.7002222222222222,
|
| 22087 |
+
"grad_norm": 0.7694705724716187,
|
| 22088 |
+
"learning_rate": 6.0089086859688204e-05,
|
| 22089 |
+
"loss": 1.1152,
|
| 22090 |
+
"step": 3151
|
| 22091 |
+
},
|
| 22092 |
+
{
|
| 22093 |
+
"epoch": 0.7004444444444444,
|
| 22094 |
+
"grad_norm": 0.703349232673645,
|
| 22095 |
+
"learning_rate": 6.00445434298441e-05,
|
| 22096 |
+
"loss": 1.1683,
|
| 22097 |
+
"step": 3152
|
| 22098 |
+
},
|
| 22099 |
+
{
|
| 22100 |
+
"epoch": 0.7006666666666667,
|
| 22101 |
+
"grad_norm": 0.05406121537089348,
|
| 22102 |
+
"learning_rate": 6e-05,
|
| 22103 |
+
"loss": 0.0105,
|
| 22104 |
+
"step": 3153
|
| 22105 |
+
},
|
| 22106 |
+
{
|
| 22107 |
+
"epoch": 0.7008888888888889,
|
| 22108 |
+
"grad_norm": 0.5842484831809998,
|
| 22109 |
+
"learning_rate": 5.995545657015591e-05,
|
| 22110 |
+
"loss": 1.15,
|
| 22111 |
+
"step": 3154
|
| 22112 |
+
},
|
| 22113 |
+
{
|
| 22114 |
+
"epoch": 0.7011111111111111,
|
| 22115 |
+
"grad_norm": 0.05234431475400925,
|
| 22116 |
+
"learning_rate": 5.9910913140311805e-05,
|
| 22117 |
+
"loss": 0.0105,
|
| 22118 |
+
"step": 3155
|
| 22119 |
+
},
|
| 22120 |
+
{
|
| 22121 |
+
"epoch": 0.7013333333333334,
|
| 22122 |
+
"grad_norm": 0.5893082618713379,
|
| 22123 |
+
"learning_rate": 5.9866369710467704e-05,
|
| 22124 |
+
"loss": 1.1913,
|
| 22125 |
+
"step": 3156
|
| 22126 |
+
},
|
| 22127 |
+
{
|
| 22128 |
+
"epoch": 0.7015555555555556,
|
| 22129 |
+
"grad_norm": 0.5218148231506348,
|
| 22130 |
+
"learning_rate": 5.9821826280623616e-05,
|
| 22131 |
+
"loss": 0.9835,
|
| 22132 |
+
"step": 3157
|
| 22133 |
+
},
|
| 22134 |
+
{
|
| 22135 |
+
"epoch": 0.7017777777777777,
|
| 22136 |
+
"grad_norm": 0.5484596490859985,
|
| 22137 |
+
"learning_rate": 5.9777282850779515e-05,
|
| 22138 |
+
"loss": 0.9247,
|
| 22139 |
+
"step": 3158
|
| 22140 |
+
},
|
| 22141 |
+
{
|
| 22142 |
+
"epoch": 0.702,
|
| 22143 |
+
"grad_norm": 0.6557696461677551,
|
| 22144 |
+
"learning_rate": 5.973273942093541e-05,
|
| 22145 |
+
"loss": 1.1391,
|
| 22146 |
+
"step": 3159
|
| 22147 |
+
},
|
| 22148 |
+
{
|
| 22149 |
+
"epoch": 0.7022222222222222,
|
| 22150 |
+
"grad_norm": 0.5898274779319763,
|
| 22151 |
+
"learning_rate": 5.9688195991091325e-05,
|
| 22152 |
+
"loss": 1.2284,
|
| 22153 |
+
"step": 3160
|
| 22154 |
+
},
|
| 22155 |
+
{
|
| 22156 |
+
"epoch": 0.7024444444444444,
|
| 22157 |
+
"grad_norm": 0.09231838583946228,
|
| 22158 |
+
"learning_rate": 5.9643652561247224e-05,
|
| 22159 |
+
"loss": 0.0125,
|
| 22160 |
+
"step": 3161
|
| 22161 |
+
},
|
| 22162 |
+
{
|
| 22163 |
+
"epoch": 0.7026666666666667,
|
| 22164 |
+
"grad_norm": 1.012488842010498,
|
| 22165 |
+
"learning_rate": 5.9599109131403116e-05,
|
| 22166 |
+
"loss": 2.0515,
|
| 22167 |
+
"step": 3162
|
| 22168 |
+
},
|
| 22169 |
+
{
|
| 22170 |
+
"epoch": 0.7028888888888889,
|
| 22171 |
+
"grad_norm": 0.9501926302909851,
|
| 22172 |
+
"learning_rate": 5.9554565701559014e-05,
|
| 22173 |
+
"loss": 2.3767,
|
| 22174 |
+
"step": 3163
|
| 22175 |
+
},
|
| 22176 |
+
{
|
| 22177 |
+
"epoch": 0.7031111111111111,
|
| 22178 |
+
"grad_norm": 0.9576533436775208,
|
| 22179 |
+
"learning_rate": 5.9510022271714927e-05,
|
| 22180 |
+
"loss": 2.3394,
|
| 22181 |
+
"step": 3164
|
| 22182 |
+
},
|
| 22183 |
+
{
|
| 22184 |
+
"epoch": 0.7033333333333334,
|
| 22185 |
+
"grad_norm": 0.944797694683075,
|
| 22186 |
+
"learning_rate": 5.9465478841870825e-05,
|
| 22187 |
+
"loss": 1.9659,
|
| 22188 |
+
"step": 3165
|
| 22189 |
+
},
|
| 22190 |
+
{
|
| 22191 |
+
"epoch": 0.7035555555555556,
|
| 22192 |
+
"grad_norm": 0.8810012340545654,
|
| 22193 |
+
"learning_rate": 5.9420935412026724e-05,
|
| 22194 |
+
"loss": 1.8656,
|
| 22195 |
+
"step": 3166
|
| 22196 |
+
},
|
| 22197 |
+
{
|
| 22198 |
+
"epoch": 0.7037777777777777,
|
| 22199 |
+
"grad_norm": 0.6439220309257507,
|
| 22200 |
+
"learning_rate": 5.9376391982182636e-05,
|
| 22201 |
+
"loss": 0.8335,
|
| 22202 |
+
"step": 3167
|
| 22203 |
+
},
|
| 22204 |
+
{
|
| 22205 |
+
"epoch": 0.704,
|
| 22206 |
+
"grad_norm": 0.9962994456291199,
|
| 22207 |
+
"learning_rate": 5.9331848552338534e-05,
|
| 22208 |
+
"loss": 2.0233,
|
| 22209 |
+
"step": 3168
|
| 22210 |
+
},
|
| 22211 |
+
{
|
| 22212 |
+
"epoch": 0.7042222222222222,
|
| 22213 |
+
"grad_norm": 0.9703332185745239,
|
| 22214 |
+
"learning_rate": 5.928730512249443e-05,
|
| 22215 |
+
"loss": 1.9519,
|
| 22216 |
+
"step": 3169
|
| 22217 |
+
},
|
| 22218 |
+
{
|
| 22219 |
+
"epoch": 0.7044444444444444,
|
| 22220 |
+
"grad_norm": 1.0500884056091309,
|
| 22221 |
+
"learning_rate": 5.924276169265034e-05,
|
| 22222 |
+
"loss": 1.578,
|
| 22223 |
+
"step": 3170
|
| 22224 |
+
},
|
| 22225 |
+
{
|
| 22226 |
+
"epoch": 0.7046666666666667,
|
| 22227 |
+
"grad_norm": 0.9718672037124634,
|
| 22228 |
+
"learning_rate": 5.919821826280624e-05,
|
| 22229 |
+
"loss": 2.0021,
|
| 22230 |
+
"step": 3171
|
| 22231 |
+
},
|
| 22232 |
+
{
|
| 22233 |
+
"epoch": 0.7048888888888889,
|
| 22234 |
+
"grad_norm": 0.07014777511358261,
|
| 22235 |
+
"learning_rate": 5.9153674832962136e-05,
|
| 22236 |
+
"loss": 0.0162,
|
| 22237 |
+
"step": 3172
|
| 22238 |
+
},
|
| 22239 |
+
{
|
| 22240 |
+
"epoch": 0.7051111111111111,
|
| 22241 |
+
"grad_norm": 0.07737057656049728,
|
| 22242 |
+
"learning_rate": 5.910913140311805e-05,
|
| 22243 |
+
"loss": 0.0156,
|
| 22244 |
+
"step": 3173
|
| 22245 |
+
},
|
| 22246 |
+
{
|
| 22247 |
+
"epoch": 0.7053333333333334,
|
| 22248 |
+
"grad_norm": 0.12630076706409454,
|
| 22249 |
+
"learning_rate": 5.9064587973273946e-05,
|
| 22250 |
+
"loss": 0.0213,
|
| 22251 |
+
"step": 3174
|
| 22252 |
+
},
|
| 22253 |
+
{
|
| 22254 |
+
"epoch": 0.7055555555555556,
|
| 22255 |
+
"grad_norm": 0.7619150876998901,
|
| 22256 |
+
"learning_rate": 5.9020044543429845e-05,
|
| 22257 |
+
"loss": 0.9373,
|
| 22258 |
+
"step": 3175
|
| 22259 |
+
},
|
| 22260 |
+
{
|
| 22261 |
+
"epoch": 0.7057777777777777,
|
| 22262 |
+
"grad_norm": 1.0807890892028809,
|
| 22263 |
+
"learning_rate": 5.897550111358575e-05,
|
| 22264 |
+
"loss": 2.0725,
|
| 22265 |
+
"step": 3176
|
| 22266 |
+
},
|
| 22267 |
+
{
|
| 22268 |
+
"epoch": 0.706,
|
| 22269 |
+
"grad_norm": 0.9409441351890564,
|
| 22270 |
+
"learning_rate": 5.893095768374165e-05,
|
| 22271 |
+
"loss": 1.6597,
|
| 22272 |
+
"step": 3177
|
| 22273 |
+
},
|
| 22274 |
+
{
|
| 22275 |
+
"epoch": 0.7062222222222222,
|
| 22276 |
+
"grad_norm": 1.096917986869812,
|
| 22277 |
+
"learning_rate": 5.888641425389755e-05,
|
| 22278 |
+
"loss": 1.9767,
|
| 22279 |
+
"step": 3178
|
| 22280 |
+
},
|
| 22281 |
+
{
|
| 22282 |
+
"epoch": 0.7064444444444444,
|
| 22283 |
+
"grad_norm": 1.091698408126831,
|
| 22284 |
+
"learning_rate": 5.884187082405346e-05,
|
| 22285 |
+
"loss": 1.7166,
|
| 22286 |
+
"step": 3179
|
| 22287 |
+
},
|
| 22288 |
+
{
|
| 22289 |
+
"epoch": 0.7066666666666667,
|
| 22290 |
+
"grad_norm": 1.0211970806121826,
|
| 22291 |
+
"learning_rate": 5.879732739420936e-05,
|
| 22292 |
+
"loss": 1.6798,
|
| 22293 |
+
"step": 3180
|
| 22294 |
+
},
|
| 22295 |
+
{
|
| 22296 |
+
"epoch": 0.7068888888888889,
|
| 22297 |
+
"grad_norm": 0.6886789202690125,
|
| 22298 |
+
"learning_rate": 5.875278396436526e-05,
|
| 22299 |
+
"loss": 1.0461,
|
| 22300 |
+
"step": 3181
|
| 22301 |
+
},
|
| 22302 |
+
{
|
| 22303 |
+
"epoch": 0.7071111111111111,
|
| 22304 |
+
"grad_norm": 0.05880124494433403,
|
| 22305 |
+
"learning_rate": 5.870824053452116e-05,
|
| 22306 |
+
"loss": 0.0178,
|
| 22307 |
+
"step": 3182
|
| 22308 |
+
},
|
| 22309 |
+
{
|
| 22310 |
+
"epoch": 0.7073333333333334,
|
| 22311 |
+
"grad_norm": 0.060819823294878006,
|
| 22312 |
+
"learning_rate": 5.866369710467706e-05,
|
| 22313 |
+
"loss": 0.0178,
|
| 22314 |
+
"step": 3183
|
| 22315 |
+
},
|
| 22316 |
+
{
|
| 22317 |
+
"epoch": 0.7075555555555556,
|
| 22318 |
+
"grad_norm": 0.724615752696991,
|
| 22319 |
+
"learning_rate": 5.861915367483296e-05,
|
| 22320 |
+
"loss": 1.0519,
|
| 22321 |
+
"step": 3184
|
| 22322 |
+
},
|
| 22323 |
+
{
|
| 22324 |
+
"epoch": 0.7077777777777777,
|
| 22325 |
+
"grad_norm": 0.8110787868499756,
|
| 22326 |
+
"learning_rate": 5.857461024498887e-05,
|
| 22327 |
+
"loss": 0.9285,
|
| 22328 |
+
"step": 3185
|
| 22329 |
+
},
|
| 22330 |
+
{
|
| 22331 |
+
"epoch": 0.708,
|
| 22332 |
+
"grad_norm": 0.06422421336174011,
|
| 22333 |
+
"learning_rate": 5.853006681514477e-05,
|
| 22334 |
+
"loss": 0.0176,
|
| 22335 |
+
"step": 3186
|
| 22336 |
+
},
|
| 22337 |
+
{
|
| 22338 |
+
"epoch": 0.7082222222222222,
|
| 22339 |
+
"grad_norm": 0.067098468542099,
|
| 22340 |
+
"learning_rate": 5.848552338530067e-05,
|
| 22341 |
+
"loss": 0.0175,
|
| 22342 |
+
"step": 3187
|
| 22343 |
+
},
|
| 22344 |
+
{
|
| 22345 |
+
"epoch": 0.7084444444444444,
|
| 22346 |
+
"grad_norm": 0.06487097591161728,
|
| 22347 |
+
"learning_rate": 5.8440979955456574e-05,
|
| 22348 |
+
"loss": 0.0172,
|
| 22349 |
+
"step": 3188
|
| 22350 |
+
},
|
| 22351 |
+
{
|
| 22352 |
+
"epoch": 0.7086666666666667,
|
| 22353 |
+
"grad_norm": 0.06535470485687256,
|
| 22354 |
+
"learning_rate": 5.839643652561247e-05,
|
| 22355 |
+
"loss": 0.0174,
|
| 22356 |
+
"step": 3189
|
| 22357 |
+
},
|
| 22358 |
+
{
|
| 22359 |
+
"epoch": 0.7088888888888889,
|
| 22360 |
+
"grad_norm": 1.172293782234192,
|
| 22361 |
+
"learning_rate": 5.835189309576837e-05,
|
| 22362 |
+
"loss": 1.5616,
|
| 22363 |
+
"step": 3190
|
| 22364 |
+
},
|
| 22365 |
+
{
|
| 22366 |
+
"epoch": 0.7091111111111111,
|
| 22367 |
+
"grad_norm": 1.1036264896392822,
|
| 22368 |
+
"learning_rate": 5.830734966592428e-05,
|
| 22369 |
+
"loss": 1.6201,
|
| 22370 |
+
"step": 3191
|
| 22371 |
+
},
|
| 22372 |
+
{
|
| 22373 |
+
"epoch": 0.7093333333333334,
|
| 22374 |
+
"grad_norm": 0.7746077179908752,
|
| 22375 |
+
"learning_rate": 5.826280623608018e-05,
|
| 22376 |
+
"loss": 0.8634,
|
| 22377 |
+
"step": 3192
|
| 22378 |
+
},
|
| 22379 |
+
{
|
| 22380 |
+
"epoch": 0.7095555555555556,
|
| 22381 |
+
"grad_norm": 0.9545249342918396,
|
| 22382 |
+
"learning_rate": 5.821826280623608e-05,
|
| 22383 |
+
"loss": 1.4337,
|
| 22384 |
+
"step": 3193
|
| 22385 |
+
},
|
| 22386 |
+
{
|
| 22387 |
+
"epoch": 0.7097777777777777,
|
| 22388 |
+
"grad_norm": 1.0250579118728638,
|
| 22389 |
+
"learning_rate": 5.8173719376391986e-05,
|
| 22390 |
+
"loss": 1.6208,
|
| 22391 |
+
"step": 3194
|
| 22392 |
+
},
|
| 22393 |
+
{
|
| 22394 |
+
"epoch": 0.71,
|
| 22395 |
+
"grad_norm": 1.0089478492736816,
|
| 22396 |
+
"learning_rate": 5.8129175946547884e-05,
|
| 22397 |
+
"loss": 1.2085,
|
| 22398 |
+
"step": 3195
|
| 22399 |
+
},
|
| 22400 |
+
{
|
| 22401 |
+
"epoch": 0.7102222222222222,
|
| 22402 |
+
"grad_norm": 1.1248306035995483,
|
| 22403 |
+
"learning_rate": 5.808463251670378e-05,
|
| 22404 |
+
"loss": 1.5717,
|
| 22405 |
+
"step": 3196
|
| 22406 |
+
},
|
| 22407 |
+
{
|
| 22408 |
+
"epoch": 0.7104444444444444,
|
| 22409 |
+
"grad_norm": 0.6627147793769836,
|
| 22410 |
+
"learning_rate": 5.8040089086859695e-05,
|
| 22411 |
+
"loss": 0.6573,
|
| 22412 |
+
"step": 3197
|
| 22413 |
+
},
|
| 22414 |
+
{
|
| 22415 |
+
"epoch": 0.7106666666666667,
|
| 22416 |
+
"grad_norm": 1.230597972869873,
|
| 22417 |
+
"learning_rate": 5.7995545657015594e-05,
|
| 22418 |
+
"loss": 1.2766,
|
| 22419 |
+
"step": 3198
|
| 22420 |
+
},
|
| 22421 |
+
{
|
| 22422 |
+
"epoch": 0.7108888888888889,
|
| 22423 |
+
"grad_norm": 0.1396600902080536,
|
| 22424 |
+
"learning_rate": 5.795100222717149e-05,
|
| 22425 |
+
"loss": 0.0322,
|
| 22426 |
+
"step": 3199
|
| 22427 |
+
},
|
| 22428 |
+
{
|
| 22429 |
+
"epoch": 0.7111111111111111,
|
| 22430 |
+
"grad_norm": 0.9573265314102173,
|
| 22431 |
+
"learning_rate": 5.79064587973274e-05,
|
| 22432 |
+
"loss": 0.507,
|
| 22433 |
+
"step": 3200
|
| 22434 |
+
},
|
| 22435 |
+
{
|
| 22436 |
+
"epoch": 0.7113333333333334,
|
| 22437 |
+
"grad_norm": 0.8519662618637085,
|
| 22438 |
+
"learning_rate": 5.7861915367483296e-05,
|
| 22439 |
+
"loss": 1.1341,
|
| 22440 |
+
"step": 3201
|
| 22441 |
+
},
|
| 22442 |
+
{
|
| 22443 |
+
"epoch": 0.7115555555555556,
|
| 22444 |
+
"grad_norm": 0.04184136167168617,
|
| 22445 |
+
"learning_rate": 5.7817371937639195e-05,
|
| 22446 |
+
"loss": 0.0093,
|
| 22447 |
+
"step": 3202
|
| 22448 |
+
},
|
| 22449 |
+
{
|
| 22450 |
+
"epoch": 0.7117777777777777,
|
| 22451 |
+
"grad_norm": 0.6974391341209412,
|
| 22452 |
+
"learning_rate": 5.777282850779511e-05,
|
| 22453 |
+
"loss": 1.3839,
|
| 22454 |
+
"step": 3203
|
| 22455 |
+
},
|
| 22456 |
+
{
|
| 22457 |
+
"epoch": 0.712,
|
| 22458 |
+
"grad_norm": 0.8318896293640137,
|
| 22459 |
+
"learning_rate": 5.7728285077951005e-05,
|
| 22460 |
+
"loss": 2.0999,
|
| 22461 |
+
"step": 3204
|
| 22462 |
+
},
|
| 22463 |
+
{
|
| 22464 |
+
"epoch": 0.7122222222222222,
|
| 22465 |
+
"grad_norm": 0.5589978694915771,
|
| 22466 |
+
"learning_rate": 5.7683741648106904e-05,
|
| 22467 |
+
"loss": 1.1487,
|
| 22468 |
+
"step": 3205
|
| 22469 |
+
},
|
| 22470 |
+
{
|
| 22471 |
+
"epoch": 0.7124444444444444,
|
| 22472 |
+
"grad_norm": 0.07977552711963654,
|
| 22473 |
+
"learning_rate": 5.7639198218262816e-05,
|
| 22474 |
+
"loss": 0.0123,
|
| 22475 |
+
"step": 3206
|
| 22476 |
+
},
|
| 22477 |
+
{
|
| 22478 |
+
"epoch": 0.7126666666666667,
|
| 22479 |
+
"grad_norm": 0.07479345053434372,
|
| 22480 |
+
"learning_rate": 5.7594654788418715e-05,
|
| 22481 |
+
"loss": 0.0123,
|
| 22482 |
+
"step": 3207
|
| 22483 |
+
},
|
| 22484 |
+
{
|
| 22485 |
+
"epoch": 0.7128888888888889,
|
| 22486 |
+
"grad_norm": 0.7520397305488586,
|
| 22487 |
+
"learning_rate": 5.7550111358574607e-05,
|
| 22488 |
+
"loss": 1.8669,
|
| 22489 |
+
"step": 3208
|
| 22490 |
+
},
|
| 22491 |
+
{
|
| 22492 |
+
"epoch": 0.7131111111111111,
|
| 22493 |
+
"grad_norm": 0.891527533531189,
|
| 22494 |
+
"learning_rate": 5.750556792873052e-05,
|
| 22495 |
+
"loss": 1.8476,
|
| 22496 |
+
"step": 3209
|
| 22497 |
+
},
|
| 22498 |
+
{
|
| 22499 |
+
"epoch": 0.7133333333333334,
|
| 22500 |
+
"grad_norm": 0.870412290096283,
|
| 22501 |
+
"learning_rate": 5.746102449888642e-05,
|
| 22502 |
+
"loss": 1.9461,
|
| 22503 |
+
"step": 3210
|
| 22504 |
+
},
|
| 22505 |
+
{
|
| 22506 |
+
"epoch": 0.7135555555555556,
|
| 22507 |
+
"grad_norm": 0.9231261610984802,
|
| 22508 |
+
"learning_rate": 5.7416481069042316e-05,
|
| 22509 |
+
"loss": 2.1436,
|
| 22510 |
+
"step": 3211
|
| 22511 |
+
},
|
| 22512 |
+
{
|
| 22513 |
+
"epoch": 0.7137777777777777,
|
| 22514 |
+
"grad_norm": 0.804538369178772,
|
| 22515 |
+
"learning_rate": 5.737193763919823e-05,
|
| 22516 |
+
"loss": 1.6058,
|
| 22517 |
+
"step": 3212
|
| 22518 |
+
},
|
| 22519 |
+
{
|
| 22520 |
+
"epoch": 0.714,
|
| 22521 |
+
"grad_norm": 0.9710292220115662,
|
| 22522 |
+
"learning_rate": 5.7327394209354127e-05,
|
| 22523 |
+
"loss": 1.0738,
|
| 22524 |
+
"step": 3213
|
| 22525 |
+
},
|
| 22526 |
+
{
|
| 22527 |
+
"epoch": 0.7142222222222222,
|
| 22528 |
+
"grad_norm": 0.9411685466766357,
|
| 22529 |
+
"learning_rate": 5.7282850779510025e-05,
|
| 22530 |
+
"loss": 2.0708,
|
| 22531 |
+
"step": 3214
|
| 22532 |
+
},
|
| 22533 |
+
{
|
| 22534 |
+
"epoch": 0.7144444444444444,
|
| 22535 |
+
"grad_norm": 0.9712237119674683,
|
| 22536 |
+
"learning_rate": 5.723830734966593e-05,
|
| 22537 |
+
"loss": 2.1416,
|
| 22538 |
+
"step": 3215
|
| 22539 |
+
},
|
| 22540 |
+
{
|
| 22541 |
+
"epoch": 0.7146666666666667,
|
| 22542 |
+
"grad_norm": 0.6982542872428894,
|
| 22543 |
+
"learning_rate": 5.719376391982183e-05,
|
| 22544 |
+
"loss": 0.7926,
|
| 22545 |
+
"step": 3216
|
| 22546 |
+
},
|
| 22547 |
+
{
|
| 22548 |
+
"epoch": 0.7148888888888889,
|
| 22549 |
+
"grad_norm": 0.7483058571815491,
|
| 22550 |
+
"learning_rate": 5.714922048997773e-05,
|
| 22551 |
+
"loss": 0.8696,
|
| 22552 |
+
"step": 3217
|
| 22553 |
+
},
|
| 22554 |
+
{
|
| 22555 |
+
"epoch": 0.7151111111111111,
|
| 22556 |
+
"grad_norm": 0.6382774114608765,
|
| 22557 |
+
"learning_rate": 5.710467706013364e-05,
|
| 22558 |
+
"loss": 0.8758,
|
| 22559 |
+
"step": 3218
|
| 22560 |
+
},
|
| 22561 |
+
{
|
| 22562 |
+
"epoch": 0.7153333333333334,
|
| 22563 |
+
"grad_norm": 0.09534616768360138,
|
| 22564 |
+
"learning_rate": 5.706013363028954e-05,
|
| 22565 |
+
"loss": 0.0179,
|
| 22566 |
+
"step": 3219
|
| 22567 |
+
},
|
| 22568 |
+
{
|
| 22569 |
+
"epoch": 0.7155555555555555,
|
| 22570 |
+
"grad_norm": 0.9931474328041077,
|
| 22571 |
+
"learning_rate": 5.701559020044544e-05,
|
| 22572 |
+
"loss": 1.7448,
|
| 22573 |
+
"step": 3220
|
| 22574 |
+
},
|
| 22575 |
+
{
|
| 22576 |
+
"epoch": 0.7157777777777777,
|
| 22577 |
+
"grad_norm": 1.051207184791565,
|
| 22578 |
+
"learning_rate": 5.697104677060134e-05,
|
| 22579 |
+
"loss": 1.8485,
|
| 22580 |
+
"step": 3221
|
| 22581 |
+
},
|
| 22582 |
+
{
|
| 22583 |
+
"epoch": 0.716,
|
| 22584 |
+
"grad_norm": 0.9426413178443909,
|
| 22585 |
+
"learning_rate": 5.692650334075724e-05,
|
| 22586 |
+
"loss": 1.6347,
|
| 22587 |
+
"step": 3222
|
| 22588 |
+
},
|
| 22589 |
+
{
|
| 22590 |
+
"epoch": 0.7162222222222222,
|
| 22591 |
+
"grad_norm": 0.919272243976593,
|
| 22592 |
+
"learning_rate": 5.688195991091314e-05,
|
| 22593 |
+
"loss": 1.7151,
|
| 22594 |
+
"step": 3223
|
| 22595 |
+
},
|
| 22596 |
+
{
|
| 22597 |
+
"epoch": 0.7164444444444444,
|
| 22598 |
+
"grad_norm": 0.9655510783195496,
|
| 22599 |
+
"learning_rate": 5.683741648106905e-05,
|
| 22600 |
+
"loss": 1.6615,
|
| 22601 |
+
"step": 3224
|
| 22602 |
+
},
|
| 22603 |
+
{
|
| 22604 |
+
"epoch": 0.7166666666666667,
|
| 22605 |
+
"grad_norm": 1.2728337049484253,
|
| 22606 |
+
"learning_rate": 5.679287305122495e-05,
|
| 22607 |
+
"loss": 1.8277,
|
| 22608 |
+
"step": 3225
|
| 22609 |
+
},
|
| 22610 |
+
{
|
| 22611 |
+
"epoch": 0.7168888888888889,
|
| 22612 |
+
"grad_norm": 0.7086578011512756,
|
| 22613 |
+
"learning_rate": 5.674832962138085e-05,
|
| 22614 |
+
"loss": 0.8322,
|
| 22615 |
+
"step": 3226
|
| 22616 |
+
},
|
| 22617 |
+
{
|
| 22618 |
+
"epoch": 0.7171111111111111,
|
| 22619 |
+
"grad_norm": 0.06795133650302887,
|
| 22620 |
+
"learning_rate": 5.6703786191536754e-05,
|
| 22621 |
+
"loss": 0.017,
|
| 22622 |
+
"step": 3227
|
| 22623 |
+
},
|
| 22624 |
+
{
|
| 22625 |
+
"epoch": 0.7173333333333334,
|
| 22626 |
+
"grad_norm": 0.06331969052553177,
|
| 22627 |
+
"learning_rate": 5.665924276169265e-05,
|
| 22628 |
+
"loss": 0.0171,
|
| 22629 |
+
"step": 3228
|
| 22630 |
+
},
|
| 22631 |
+
{
|
| 22632 |
+
"epoch": 0.7175555555555555,
|
| 22633 |
+
"grad_norm": 0.0663456916809082,
|
| 22634 |
+
"learning_rate": 5.661469933184855e-05,
|
| 22635 |
+
"loss": 0.0173,
|
| 22636 |
+
"step": 3229
|
| 22637 |
+
},
|
| 22638 |
+
{
|
| 22639 |
+
"epoch": 0.7177777777777777,
|
| 22640 |
+
"grad_norm": 0.8989565968513489,
|
| 22641 |
+
"learning_rate": 5.6570155902004463e-05,
|
| 22642 |
+
"loss": 1.6765,
|
| 22643 |
+
"step": 3230
|
| 22644 |
+
},
|
| 22645 |
+
{
|
| 22646 |
+
"epoch": 0.718,
|
| 22647 |
+
"grad_norm": 0.7637456059455872,
|
| 22648 |
+
"learning_rate": 5.652561247216036e-05,
|
| 22649 |
+
"loss": 0.7514,
|
| 22650 |
+
"step": 3231
|
| 22651 |
+
},
|
| 22652 |
+
{
|
| 22653 |
+
"epoch": 0.7182222222222222,
|
| 22654 |
+
"grad_norm": 0.08078856021165848,
|
| 22655 |
+
"learning_rate": 5.648106904231626e-05,
|
| 22656 |
+
"loss": 0.0163,
|
| 22657 |
+
"step": 3232
|
| 22658 |
+
},
|
| 22659 |
+
{
|
| 22660 |
+
"epoch": 0.7184444444444444,
|
| 22661 |
+
"grad_norm": 0.8078843951225281,
|
| 22662 |
+
"learning_rate": 5.643652561247216e-05,
|
| 22663 |
+
"loss": 0.8599,
|
| 22664 |
+
"step": 3233
|
| 22665 |
+
},
|
| 22666 |
+
{
|
| 22667 |
+
"epoch": 0.7186666666666667,
|
| 22668 |
+
"grad_norm": 1.0271605253219604,
|
| 22669 |
+
"learning_rate": 5.6391982182628065e-05,
|
| 22670 |
+
"loss": 1.415,
|
| 22671 |
+
"step": 3234
|
| 22672 |
+
},
|
| 22673 |
+
{
|
| 22674 |
+
"epoch": 0.7188888888888889,
|
| 22675 |
+
"grad_norm": 1.2213661670684814,
|
| 22676 |
+
"learning_rate": 5.634743875278396e-05,
|
| 22677 |
+
"loss": 1.8322,
|
| 22678 |
+
"step": 3235
|
| 22679 |
+
},
|
| 22680 |
+
{
|
| 22681 |
+
"epoch": 0.7191111111111111,
|
| 22682 |
+
"grad_norm": 1.0940077304840088,
|
| 22683 |
+
"learning_rate": 5.630289532293986e-05,
|
| 22684 |
+
"loss": 1.5287,
|
| 22685 |
+
"step": 3236
|
| 22686 |
+
},
|
| 22687 |
+
{
|
| 22688 |
+
"epoch": 0.7193333333333334,
|
| 22689 |
+
"grad_norm": 1.0005013942718506,
|
| 22690 |
+
"learning_rate": 5.6258351893095774e-05,
|
| 22691 |
+
"loss": 1.6254,
|
| 22692 |
+
"step": 3237
|
| 22693 |
+
},
|
| 22694 |
+
{
|
| 22695 |
+
"epoch": 0.7195555555555555,
|
| 22696 |
+
"grad_norm": 0.8303656578063965,
|
| 22697 |
+
"learning_rate": 5.621380846325167e-05,
|
| 22698 |
+
"loss": 1.027,
|
| 22699 |
+
"step": 3238
|
| 22700 |
+
},
|
| 22701 |
+
{
|
| 22702 |
+
"epoch": 0.7197777777777777,
|
| 22703 |
+
"grad_norm": 0.704897403717041,
|
| 22704 |
+
"learning_rate": 5.616926503340757e-05,
|
| 22705 |
+
"loss": 0.7988,
|
| 22706 |
+
"step": 3239
|
| 22707 |
+
},
|
| 22708 |
+
{
|
| 22709 |
+
"epoch": 0.72,
|
| 22710 |
+
"grad_norm": 1.0700993537902832,
|
| 22711 |
+
"learning_rate": 5.6124721603563476e-05,
|
| 22712 |
+
"loss": 1.7471,
|
| 22713 |
+
"step": 3240
|
| 22714 |
+
},
|
| 22715 |
+
{
|
| 22716 |
+
"epoch": 0.7202222222222222,
|
| 22717 |
+
"grad_norm": 1.1328794956207275,
|
| 22718 |
+
"learning_rate": 5.6080178173719375e-05,
|
| 22719 |
+
"loss": 1.2742,
|
| 22720 |
+
"step": 3241
|
| 22721 |
+
},
|
| 22722 |
+
{
|
| 22723 |
+
"epoch": 0.7204444444444444,
|
| 22724 |
+
"grad_norm": 0.9732044339179993,
|
| 22725 |
+
"learning_rate": 5.6035634743875274e-05,
|
| 22726 |
+
"loss": 1.3644,
|
| 22727 |
+
"step": 3242
|
| 22728 |
+
},
|
| 22729 |
+
{
|
| 22730 |
+
"epoch": 0.7206666666666667,
|
| 22731 |
+
"grad_norm": 1.174729347229004,
|
| 22732 |
+
"learning_rate": 5.5991091314031186e-05,
|
| 22733 |
+
"loss": 1.5483,
|
| 22734 |
+
"step": 3243
|
| 22735 |
+
},
|
| 22736 |
+
{
|
| 22737 |
+
"epoch": 0.7208888888888889,
|
| 22738 |
+
"grad_norm": 0.78294837474823,
|
| 22739 |
+
"learning_rate": 5.5946547884187084e-05,
|
| 22740 |
+
"loss": 0.6034,
|
| 22741 |
+
"step": 3244
|
| 22742 |
+
},
|
| 22743 |
+
{
|
| 22744 |
+
"epoch": 0.7211111111111111,
|
| 22745 |
+
"grad_norm": 0.8941324949264526,
|
| 22746 |
+
"learning_rate": 5.590200445434298e-05,
|
| 22747 |
+
"loss": 1.0701,
|
| 22748 |
+
"step": 3245
|
| 22749 |
+
},
|
| 22750 |
+
{
|
| 22751 |
+
"epoch": 0.7213333333333334,
|
| 22752 |
+
"grad_norm": 1.1886690855026245,
|
| 22753 |
+
"learning_rate": 5.5857461024498895e-05,
|
| 22754 |
+
"loss": 1.3666,
|
| 22755 |
+
"step": 3246
|
| 22756 |
+
},
|
| 22757 |
+
{
|
| 22758 |
+
"epoch": 0.7215555555555555,
|
| 22759 |
+
"grad_norm": 1.0057522058486938,
|
| 22760 |
+
"learning_rate": 5.581291759465479e-05,
|
| 22761 |
+
"loss": 1.1996,
|
| 22762 |
+
"step": 3247
|
| 22763 |
+
},
|
| 22764 |
+
{
|
| 22765 |
+
"epoch": 0.7217777777777777,
|
| 22766 |
+
"grad_norm": 0.839670717716217,
|
| 22767 |
+
"learning_rate": 5.5768374164810685e-05,
|
| 22768 |
+
"loss": 0.6609,
|
| 22769 |
+
"step": 3248
|
| 22770 |
+
},
|
| 22771 |
+
{
|
| 22772 |
+
"epoch": 0.722,
|
| 22773 |
+
"grad_norm": 1.1767035722732544,
|
| 22774 |
+
"learning_rate": 5.57238307349666e-05,
|
| 22775 |
+
"loss": 1.0941,
|
| 22776 |
+
"step": 3249
|
| 22777 |
+
},
|
| 22778 |
+
{
|
| 22779 |
+
"epoch": 0.7222222222222222,
|
| 22780 |
+
"grad_norm": 1.2154204845428467,
|
| 22781 |
+
"learning_rate": 5.5679287305122496e-05,
|
| 22782 |
+
"loss": 0.7415,
|
| 22783 |
+
"step": 3250
|
| 22784 |
+
},
|
| 22785 |
+
{
|
| 22786 |
+
"epoch": 0.7224444444444444,
|
| 22787 |
+
"grad_norm": 0.5861397385597229,
|
| 22788 |
+
"learning_rate": 5.5634743875278395e-05,
|
| 22789 |
+
"loss": 1.1437,
|
| 22790 |
+
"step": 3251
|
| 22791 |
+
},
|
| 22792 |
+
{
|
| 22793 |
+
"epoch": 0.7226666666666667,
|
| 22794 |
+
"grad_norm": 0.041759125888347626,
|
| 22795 |
+
"learning_rate": 5.559020044543431e-05,
|
| 22796 |
+
"loss": 0.0097,
|
| 22797 |
+
"step": 3252
|
| 22798 |
+
},
|
| 22799 |
+
{
|
| 22800 |
+
"epoch": 0.7228888888888889,
|
| 22801 |
+
"grad_norm": 0.7977886199951172,
|
| 22802 |
+
"learning_rate": 5.5545657015590205e-05,
|
| 22803 |
+
"loss": 2.1774,
|
| 22804 |
+
"step": 3253
|
| 22805 |
+
},
|
| 22806 |
+
{
|
| 22807 |
+
"epoch": 0.7231111111111111,
|
| 22808 |
+
"grad_norm": 0.571662425994873,
|
| 22809 |
+
"learning_rate": 5.5501113585746104e-05,
|
| 22810 |
+
"loss": 1.1858,
|
| 22811 |
+
"step": 3254
|
| 22812 |
+
},
|
| 22813 |
+
{
|
| 22814 |
+
"epoch": 0.7233333333333334,
|
| 22815 |
+
"grad_norm": 0.7104848027229309,
|
| 22816 |
+
"learning_rate": 5.545657015590201e-05,
|
| 22817 |
+
"loss": 1.0467,
|
| 22818 |
+
"step": 3255
|
| 22819 |
+
},
|
| 22820 |
+
{
|
| 22821 |
+
"epoch": 0.7235555555555555,
|
| 22822 |
+
"grad_norm": 0.8153942823410034,
|
| 22823 |
+
"learning_rate": 5.541202672605791e-05,
|
| 22824 |
+
"loss": 1.976,
|
| 22825 |
+
"step": 3256
|
| 22826 |
+
},
|
| 22827 |
+
{
|
| 22828 |
+
"epoch": 0.7237777777777777,
|
| 22829 |
+
"grad_norm": 0.08071549981832504,
|
| 22830 |
+
"learning_rate": 5.5367483296213806e-05,
|
| 22831 |
+
"loss": 0.0125,
|
| 22832 |
+
"step": 3257
|
| 22833 |
+
},
|
| 22834 |
+
{
|
| 22835 |
+
"epoch": 0.724,
|
| 22836 |
+
"grad_norm": 0.12843948602676392,
|
| 22837 |
+
"learning_rate": 5.532293986636972e-05,
|
| 22838 |
+
"loss": 0.0124,
|
| 22839 |
+
"step": 3258
|
| 22840 |
+
},
|
| 22841 |
+
{
|
| 22842 |
+
"epoch": 0.7242222222222222,
|
| 22843 |
+
"grad_norm": 0.07995433360338211,
|
| 22844 |
+
"learning_rate": 5.527839643652562e-05,
|
| 22845 |
+
"loss": 0.0122,
|
| 22846 |
+
"step": 3259
|
| 22847 |
+
},
|
| 22848 |
+
{
|
| 22849 |
+
"epoch": 0.7244444444444444,
|
| 22850 |
+
"grad_norm": 0.0713566243648529,
|
| 22851 |
+
"learning_rate": 5.5233853006681516e-05,
|
| 22852 |
+
"loss": 0.0119,
|
| 22853 |
+
"step": 3260
|
| 22854 |
+
},
|
| 22855 |
+
{
|
| 22856 |
+
"epoch": 0.7246666666666667,
|
| 22857 |
+
"grad_norm": 0.07306591421365738,
|
| 22858 |
+
"learning_rate": 5.518930957683742e-05,
|
| 22859 |
+
"loss": 0.0118,
|
| 22860 |
+
"step": 3261
|
| 22861 |
+
},
|
| 22862 |
+
{
|
| 22863 |
+
"epoch": 0.7248888888888889,
|
| 22864 |
+
"grad_norm": 0.8607704043388367,
|
| 22865 |
+
"learning_rate": 5.514476614699332e-05,
|
| 22866 |
+
"loss": 2.1437,
|
| 22867 |
+
"step": 3262
|
| 22868 |
+
},
|
| 22869 |
+
{
|
| 22870 |
+
"epoch": 0.7251111111111112,
|
| 22871 |
+
"grad_norm": 0.8772170543670654,
|
| 22872 |
+
"learning_rate": 5.510022271714922e-05,
|
| 22873 |
+
"loss": 1.9092,
|
| 22874 |
+
"step": 3263
|
| 22875 |
+
},
|
| 22876 |
+
{
|
| 22877 |
+
"epoch": 0.7253333333333334,
|
| 22878 |
+
"grad_norm": 0.9902425408363342,
|
| 22879 |
+
"learning_rate": 5.505567928730513e-05,
|
| 22880 |
+
"loss": 2.1999,
|
| 22881 |
+
"step": 3264
|
| 22882 |
+
},
|
| 22883 |
+
{
|
| 22884 |
+
"epoch": 0.7255555555555555,
|
| 22885 |
+
"grad_norm": 0.926304817199707,
|
| 22886 |
+
"learning_rate": 5.501113585746103e-05,
|
| 22887 |
+
"loss": 2.0622,
|
| 22888 |
+
"step": 3265
|
| 22889 |
+
},
|
| 22890 |
+
{
|
| 22891 |
+
"epoch": 0.7257777777777777,
|
| 22892 |
+
"grad_norm": 0.8717379570007324,
|
| 22893 |
+
"learning_rate": 5.496659242761693e-05,
|
| 22894 |
+
"loss": 1.6769,
|
| 22895 |
+
"step": 3266
|
| 22896 |
+
},
|
| 22897 |
+
{
|
| 22898 |
+
"epoch": 0.726,
|
| 22899 |
+
"grad_norm": 1.0354970693588257,
|
| 22900 |
+
"learning_rate": 5.492204899777283e-05,
|
| 22901 |
+
"loss": 1.9093,
|
| 22902 |
+
"step": 3267
|
| 22903 |
+
},
|
| 22904 |
+
{
|
| 22905 |
+
"epoch": 0.7262222222222222,
|
| 22906 |
+
"grad_norm": 0.9445512890815735,
|
| 22907 |
+
"learning_rate": 5.487750556792873e-05,
|
| 22908 |
+
"loss": 1.9806,
|
| 22909 |
+
"step": 3268
|
| 22910 |
+
},
|
| 22911 |
+
{
|
| 22912 |
+
"epoch": 0.7264444444444444,
|
| 22913 |
+
"grad_norm": 0.9720260500907898,
|
| 22914 |
+
"learning_rate": 5.483296213808463e-05,
|
| 22915 |
+
"loss": 1.793,
|
| 22916 |
+
"step": 3269
|
| 22917 |
+
},
|
| 22918 |
+
{
|
| 22919 |
+
"epoch": 0.7266666666666667,
|
| 22920 |
+
"grad_norm": 0.932304859161377,
|
| 22921 |
+
"learning_rate": 5.478841870824054e-05,
|
| 22922 |
+
"loss": 1.8463,
|
| 22923 |
+
"step": 3270
|
| 22924 |
+
},
|
| 22925 |
+
{
|
| 22926 |
+
"epoch": 0.7268888888888889,
|
| 22927 |
+
"grad_norm": 0.9925035238265991,
|
| 22928 |
+
"learning_rate": 5.474387527839644e-05,
|
| 22929 |
+
"loss": 1.9726,
|
| 22930 |
+
"step": 3271
|
| 22931 |
+
},
|
| 22932 |
+
{
|
| 22933 |
+
"epoch": 0.7271111111111112,
|
| 22934 |
+
"grad_norm": 0.5608296990394592,
|
| 22935 |
+
"learning_rate": 5.469933184855234e-05,
|
| 22936 |
+
"loss": 0.7764,
|
| 22937 |
+
"step": 3272
|
| 22938 |
+
},
|
| 22939 |
+
{
|
| 22940 |
+
"epoch": 0.7273333333333334,
|
| 22941 |
+
"grad_norm": 0.6601234674453735,
|
| 22942 |
+
"learning_rate": 5.4654788418708245e-05,
|
| 22943 |
+
"loss": 0.8271,
|
| 22944 |
+
"step": 3273
|
| 22945 |
+
},
|
| 22946 |
+
{
|
| 22947 |
+
"epoch": 0.7275555555555555,
|
| 22948 |
+
"grad_norm": 0.6779617071151733,
|
| 22949 |
+
"learning_rate": 5.461024498886414e-05,
|
| 22950 |
+
"loss": 0.9032,
|
| 22951 |
+
"step": 3274
|
| 22952 |
+
},
|
| 22953 |
+
{
|
| 22954 |
+
"epoch": 0.7277777777777777,
|
| 22955 |
+
"grad_norm": 0.9753432869911194,
|
| 22956 |
+
"learning_rate": 5.456570155902004e-05,
|
| 22957 |
+
"loss": 1.7793,
|
| 22958 |
+
"step": 3275
|
| 22959 |
+
},
|
| 22960 |
+
{
|
| 22961 |
+
"epoch": 0.728,
|
| 22962 |
+
"grad_norm": 0.9676978588104248,
|
| 22963 |
+
"learning_rate": 5.4521158129175954e-05,
|
| 22964 |
+
"loss": 1.6972,
|
| 22965 |
+
"step": 3276
|
| 22966 |
+
},
|
| 22967 |
+
{
|
| 22968 |
+
"epoch": 0.7282222222222222,
|
| 22969 |
+
"grad_norm": 1.093235969543457,
|
| 22970 |
+
"learning_rate": 5.447661469933185e-05,
|
| 22971 |
+
"loss": 2.0882,
|
| 22972 |
+
"step": 3277
|
| 22973 |
+
},
|
| 22974 |
+
{
|
| 22975 |
+
"epoch": 0.7284444444444444,
|
| 22976 |
+
"grad_norm": 1.0347819328308105,
|
| 22977 |
+
"learning_rate": 5.443207126948775e-05,
|
| 22978 |
+
"loss": 2.039,
|
| 22979 |
+
"step": 3278
|
| 22980 |
+
},
|
| 22981 |
+
{
|
| 22982 |
+
"epoch": 0.7286666666666667,
|
| 22983 |
+
"grad_norm": 0.071097731590271,
|
| 22984 |
+
"learning_rate": 5.4387527839643657e-05,
|
| 22985 |
+
"loss": 0.0174,
|
| 22986 |
+
"step": 3279
|
| 22987 |
+
},
|
| 22988 |
+
{
|
| 22989 |
+
"epoch": 0.7288888888888889,
|
| 22990 |
+
"grad_norm": 0.9010851383209229,
|
| 22991 |
+
"learning_rate": 5.4342984409799555e-05,
|
| 22992 |
+
"loss": 1.0428,
|
| 22993 |
+
"step": 3280
|
| 22994 |
+
},
|
| 22995 |
+
{
|
| 22996 |
+
"epoch": 0.7291111111111112,
|
| 22997 |
+
"grad_norm": 0.07293925434350967,
|
| 22998 |
+
"learning_rate": 5.4298440979955454e-05,
|
| 22999 |
+
"loss": 0.0173,
|
| 23000 |
+
"step": 3281
|
| 23001 |
+
},
|
| 23002 |
+
{
|
| 23003 |
+
"epoch": 0.7293333333333333,
|
| 23004 |
+
"grad_norm": 1.1432619094848633,
|
| 23005 |
+
"learning_rate": 5.4253897550111366e-05,
|
| 23006 |
+
"loss": 1.8629,
|
| 23007 |
+
"step": 3282
|
| 23008 |
+
},
|
| 23009 |
+
{
|
| 23010 |
+
"epoch": 0.7295555555555555,
|
| 23011 |
+
"grad_norm": 1.1886756420135498,
|
| 23012 |
+
"learning_rate": 5.4209354120267264e-05,
|
| 23013 |
+
"loss": 1.6837,
|
| 23014 |
+
"step": 3283
|
| 23015 |
+
},
|
| 23016 |
+
{
|
| 23017 |
+
"epoch": 0.7297777777777777,
|
| 23018 |
+
"grad_norm": 1.0832699537277222,
|
| 23019 |
+
"learning_rate": 5.416481069042316e-05,
|
| 23020 |
+
"loss": 1.536,
|
| 23021 |
+
"step": 3284
|
| 23022 |
+
},
|
| 23023 |
+
{
|
| 23024 |
+
"epoch": 0.73,
|
| 23025 |
+
"grad_norm": 0.6643537878990173,
|
| 23026 |
+
"learning_rate": 5.412026726057907e-05,
|
| 23027 |
+
"loss": 0.7855,
|
| 23028 |
+
"step": 3285
|
| 23029 |
+
},
|
| 23030 |
+
{
|
| 23031 |
+
"epoch": 0.7302222222222222,
|
| 23032 |
+
"grad_norm": 1.0094225406646729,
|
| 23033 |
+
"learning_rate": 5.407572383073497e-05,
|
| 23034 |
+
"loss": 1.4139,
|
| 23035 |
+
"step": 3286
|
| 23036 |
+
},
|
| 23037 |
+
{
|
| 23038 |
+
"epoch": 0.7304444444444445,
|
| 23039 |
+
"grad_norm": 1.14029860496521,
|
| 23040 |
+
"learning_rate": 5.4031180400890866e-05,
|
| 23041 |
+
"loss": 1.5854,
|
| 23042 |
+
"step": 3287
|
| 23043 |
+
},
|
| 23044 |
+
{
|
| 23045 |
+
"epoch": 0.7306666666666667,
|
| 23046 |
+
"grad_norm": 0.9698799848556519,
|
| 23047 |
+
"learning_rate": 5.398663697104678e-05,
|
| 23048 |
+
"loss": 1.4863,
|
| 23049 |
+
"step": 3288
|
| 23050 |
+
},
|
| 23051 |
+
{
|
| 23052 |
+
"epoch": 0.7308888888888889,
|
| 23053 |
+
"grad_norm": 1.1054226160049438,
|
| 23054 |
+
"learning_rate": 5.3942093541202676e-05,
|
| 23055 |
+
"loss": 1.3324,
|
| 23056 |
+
"step": 3289
|
| 23057 |
+
},
|
| 23058 |
+
{
|
| 23059 |
+
"epoch": 0.7311111111111112,
|
| 23060 |
+
"grad_norm": 1.1010569334030151,
|
| 23061 |
+
"learning_rate": 5.3897550111358575e-05,
|
| 23062 |
+
"loss": 1.4656,
|
| 23063 |
+
"step": 3290
|
| 23064 |
+
},
|
| 23065 |
+
{
|
| 23066 |
+
"epoch": 0.7313333333333333,
|
| 23067 |
+
"grad_norm": 1.315499186515808,
|
| 23068 |
+
"learning_rate": 5.385300668151449e-05,
|
| 23069 |
+
"loss": 1.4048,
|
| 23070 |
+
"step": 3291
|
| 23071 |
+
},
|
| 23072 |
+
{
|
| 23073 |
+
"epoch": 0.7315555555555555,
|
| 23074 |
+
"grad_norm": 1.108127474784851,
|
| 23075 |
+
"learning_rate": 5.3808463251670386e-05,
|
| 23076 |
+
"loss": 1.327,
|
| 23077 |
+
"step": 3292
|
| 23078 |
+
},
|
| 23079 |
+
{
|
| 23080 |
+
"epoch": 0.7317777777777777,
|
| 23081 |
+
"grad_norm": 0.6661926507949829,
|
| 23082 |
+
"learning_rate": 5.376391982182628e-05,
|
| 23083 |
+
"loss": 0.6108,
|
| 23084 |
+
"step": 3293
|
| 23085 |
+
},
|
| 23086 |
+
{
|
| 23087 |
+
"epoch": 0.732,
|
| 23088 |
+
"grad_norm": 0.9805776476860046,
|
| 23089 |
+
"learning_rate": 5.371937639198219e-05,
|
| 23090 |
+
"loss": 1.1752,
|
| 23091 |
+
"step": 3294
|
| 23092 |
+
},
|
| 23093 |
+
{
|
| 23094 |
+
"epoch": 0.7322222222222222,
|
| 23095 |
+
"grad_norm": 1.0693986415863037,
|
| 23096 |
+
"learning_rate": 5.367483296213809e-05,
|
| 23097 |
+
"loss": 1.3078,
|
| 23098 |
+
"step": 3295
|
| 23099 |
+
},
|
| 23100 |
+
{
|
| 23101 |
+
"epoch": 0.7324444444444445,
|
| 23102 |
+
"grad_norm": 1.078148603439331,
|
| 23103 |
+
"learning_rate": 5.363028953229399e-05,
|
| 23104 |
+
"loss": 1.2446,
|
| 23105 |
+
"step": 3296
|
| 23106 |
+
},
|
| 23107 |
+
{
|
| 23108 |
+
"epoch": 0.7326666666666667,
|
| 23109 |
+
"grad_norm": 1.1625440120697021,
|
| 23110 |
+
"learning_rate": 5.35857461024499e-05,
|
| 23111 |
+
"loss": 1.2387,
|
| 23112 |
+
"step": 3297
|
| 23113 |
+
},
|
| 23114 |
+
{
|
| 23115 |
+
"epoch": 0.7328888888888889,
|
| 23116 |
+
"grad_norm": 1.1278488636016846,
|
| 23117 |
+
"learning_rate": 5.35412026726058e-05,
|
| 23118 |
+
"loss": 1.1962,
|
| 23119 |
+
"step": 3298
|
| 23120 |
+
},
|
| 23121 |
+
{
|
| 23122 |
+
"epoch": 0.7331111111111112,
|
| 23123 |
+
"grad_norm": 1.182511806488037,
|
| 23124 |
+
"learning_rate": 5.3496659242761696e-05,
|
| 23125 |
+
"loss": 1.1573,
|
| 23126 |
+
"step": 3299
|
| 23127 |
+
},
|
| 23128 |
+
{
|
| 23129 |
+
"epoch": 0.7333333333333333,
|
| 23130 |
+
"grad_norm": 1.1381057500839233,
|
| 23131 |
+
"learning_rate": 5.34521158129176e-05,
|
| 23132 |
+
"loss": 0.7817,
|
| 23133 |
+
"step": 3300
|
| 23134 |
+
},
|
| 23135 |
+
{
|
| 23136 |
+
"epoch": 0.7335555555555555,
|
| 23137 |
+
"grad_norm": 0.5531929135322571,
|
| 23138 |
+
"learning_rate": 5.34075723830735e-05,
|
| 23139 |
+
"loss": 0.8331,
|
| 23140 |
+
"step": 3301
|
| 23141 |
+
},
|
| 23142 |
+
{
|
| 23143 |
+
"epoch": 0.7337777777777778,
|
| 23144 |
+
"grad_norm": 0.8333101868629456,
|
| 23145 |
+
"learning_rate": 5.33630289532294e-05,
|
| 23146 |
+
"loss": 1.9768,
|
| 23147 |
+
"step": 3302
|
| 23148 |
+
},
|
| 23149 |
+
{
|
| 23150 |
+
"epoch": 0.734,
|
| 23151 |
+
"grad_norm": 0.6918635964393616,
|
| 23152 |
+
"learning_rate": 5.331848552338531e-05,
|
| 23153 |
+
"loss": 1.0828,
|
| 23154 |
+
"step": 3303
|
| 23155 |
+
},
|
| 23156 |
+
{
|
| 23157 |
+
"epoch": 0.7342222222222222,
|
| 23158 |
+
"grad_norm": 0.9859722256660461,
|
| 23159 |
+
"learning_rate": 5.327394209354121e-05,
|
| 23160 |
+
"loss": 2.2754,
|
| 23161 |
+
"step": 3304
|
| 23162 |
+
},
|
| 23163 |
+
{
|
| 23164 |
+
"epoch": 0.7344444444444445,
|
| 23165 |
+
"grad_norm": 0.6960622072219849,
|
| 23166 |
+
"learning_rate": 5.322939866369711e-05,
|
| 23167 |
+
"loss": 1.0663,
|
| 23168 |
+
"step": 3305
|
| 23169 |
+
},
|
| 23170 |
+
{
|
| 23171 |
+
"epoch": 0.7346666666666667,
|
| 23172 |
+
"grad_norm": 1.1575109958648682,
|
| 23173 |
+
"learning_rate": 5.3184855233853006e-05,
|
| 23174 |
+
"loss": 2.2622,
|
| 23175 |
+
"step": 3306
|
| 23176 |
+
},
|
| 23177 |
+
{
|
| 23178 |
+
"epoch": 0.7348888888888889,
|
| 23179 |
+
"grad_norm": 0.5985379219055176,
|
| 23180 |
+
"learning_rate": 5.314031180400891e-05,
|
| 23181 |
+
"loss": 1.0319,
|
| 23182 |
+
"step": 3307
|
| 23183 |
+
},
|
| 23184 |
+
{
|
| 23185 |
+
"epoch": 0.7351111111111112,
|
| 23186 |
+
"grad_norm": 0.06290951371192932,
|
| 23187 |
+
"learning_rate": 5.309576837416481e-05,
|
| 23188 |
+
"loss": 0.0109,
|
| 23189 |
+
"step": 3308
|
| 23190 |
+
},
|
| 23191 |
+
{
|
| 23192 |
+
"epoch": 0.7353333333333333,
|
| 23193 |
+
"grad_norm": 0.06811843812465668,
|
| 23194 |
+
"learning_rate": 5.305122494432071e-05,
|
| 23195 |
+
"loss": 0.0109,
|
| 23196 |
+
"step": 3309
|
| 23197 |
+
},
|
| 23198 |
+
{
|
| 23199 |
+
"epoch": 0.7355555555555555,
|
| 23200 |
+
"grad_norm": 0.06429023295640945,
|
| 23201 |
+
"learning_rate": 5.300668151447662e-05,
|
| 23202 |
+
"loss": 0.0107,
|
| 23203 |
+
"step": 3310
|
| 23204 |
+
},
|
| 23205 |
+
{
|
| 23206 |
+
"epoch": 0.7357777777777778,
|
| 23207 |
+
"grad_norm": 0.06323552876710892,
|
| 23208 |
+
"learning_rate": 5.296213808463252e-05,
|
| 23209 |
+
"loss": 0.0107,
|
| 23210 |
+
"step": 3311
|
| 23211 |
+
},
|
| 23212 |
+
{
|
| 23213 |
+
"epoch": 0.736,
|
| 23214 |
+
"grad_norm": 0.6487092971801758,
|
| 23215 |
+
"learning_rate": 5.291759465478842e-05,
|
| 23216 |
+
"loss": 0.9286,
|
| 23217 |
+
"step": 3312
|
| 23218 |
+
},
|
| 23219 |
+
{
|
| 23220 |
+
"epoch": 0.7362222222222222,
|
| 23221 |
+
"grad_norm": 0.8638578653335571,
|
| 23222 |
+
"learning_rate": 5.2873051224944324e-05,
|
| 23223 |
+
"loss": 1.8427,
|
| 23224 |
+
"step": 3313
|
| 23225 |
+
},
|
| 23226 |
+
{
|
| 23227 |
+
"epoch": 0.7364444444444445,
|
| 23228 |
+
"grad_norm": 0.9095218181610107,
|
| 23229 |
+
"learning_rate": 5.282850779510022e-05,
|
| 23230 |
+
"loss": 2.0546,
|
| 23231 |
+
"step": 3314
|
| 23232 |
+
},
|
| 23233 |
+
{
|
| 23234 |
+
"epoch": 0.7366666666666667,
|
| 23235 |
+
"grad_norm": 0.87845379114151,
|
| 23236 |
+
"learning_rate": 5.278396436525612e-05,
|
| 23237 |
+
"loss": 1.9648,
|
| 23238 |
+
"step": 3315
|
| 23239 |
+
},
|
| 23240 |
+
{
|
| 23241 |
+
"epoch": 0.7368888888888889,
|
| 23242 |
+
"grad_norm": 0.8854038119316101,
|
| 23243 |
+
"learning_rate": 5.273942093541203e-05,
|
| 23244 |
+
"loss": 1.8114,
|
| 23245 |
+
"step": 3316
|
| 23246 |
+
},
|
| 23247 |
+
{
|
| 23248 |
+
"epoch": 0.7371111111111112,
|
| 23249 |
+
"grad_norm": 0.5725350379943848,
|
| 23250 |
+
"learning_rate": 5.269487750556793e-05,
|
| 23251 |
+
"loss": 1.0721,
|
| 23252 |
+
"step": 3317
|
| 23253 |
+
},
|
| 23254 |
+
{
|
| 23255 |
+
"epoch": 0.7373333333333333,
|
| 23256 |
+
"grad_norm": 0.6683716177940369,
|
| 23257 |
+
"learning_rate": 5.265033407572383e-05,
|
| 23258 |
+
"loss": 0.9192,
|
| 23259 |
+
"step": 3318
|
| 23260 |
+
},
|
| 23261 |
+
{
|
| 23262 |
+
"epoch": 0.7375555555555555,
|
| 23263 |
+
"grad_norm": 0.9927780628204346,
|
| 23264 |
+
"learning_rate": 5.2605790645879735e-05,
|
| 23265 |
+
"loss": 1.8748,
|
| 23266 |
+
"step": 3319
|
| 23267 |
+
},
|
| 23268 |
+
{
|
| 23269 |
+
"epoch": 0.7377777777777778,
|
| 23270 |
+
"grad_norm": 0.8612250685691833,
|
| 23271 |
+
"learning_rate": 5.2561247216035634e-05,
|
| 23272 |
+
"loss": 1.8307,
|
| 23273 |
+
"step": 3320
|
| 23274 |
+
},
|
| 23275 |
+
{
|
| 23276 |
+
"epoch": 0.738,
|
| 23277 |
+
"grad_norm": 0.9024035930633545,
|
| 23278 |
+
"learning_rate": 5.251670378619153e-05,
|
| 23279 |
+
"loss": 1.8448,
|
| 23280 |
+
"step": 3321
|
| 23281 |
+
},
|
| 23282 |
+
{
|
| 23283 |
+
"epoch": 0.7382222222222222,
|
| 23284 |
+
"grad_norm": 0.969914436340332,
|
| 23285 |
+
"learning_rate": 5.2472160356347445e-05,
|
| 23286 |
+
"loss": 1.866,
|
| 23287 |
+
"step": 3322
|
| 23288 |
+
},
|
| 23289 |
+
{
|
| 23290 |
+
"epoch": 0.7384444444444445,
|
| 23291 |
+
"grad_norm": 0.6315984129905701,
|
| 23292 |
+
"learning_rate": 5.242761692650334e-05,
|
| 23293 |
+
"loss": 0.9124,
|
| 23294 |
+
"step": 3323
|
| 23295 |
+
},
|
| 23296 |
+
{
|
| 23297 |
+
"epoch": 0.7386666666666667,
|
| 23298 |
+
"grad_norm": 0.07167524099349976,
|
| 23299 |
+
"learning_rate": 5.238307349665924e-05,
|
| 23300 |
+
"loss": 0.0158,
|
| 23301 |
+
"step": 3324
|
| 23302 |
+
},
|
| 23303 |
+
{
|
| 23304 |
+
"epoch": 0.7388888888888889,
|
| 23305 |
+
"grad_norm": 0.07736406475305557,
|
| 23306 |
+
"learning_rate": 5.233853006681515e-05,
|
| 23307 |
+
"loss": 0.0161,
|
| 23308 |
+
"step": 3325
|
| 23309 |
+
},
|
| 23310 |
+
{
|
| 23311 |
+
"epoch": 0.7391111111111112,
|
| 23312 |
+
"grad_norm": 0.07857107371091843,
|
| 23313 |
+
"learning_rate": 5.2293986636971046e-05,
|
| 23314 |
+
"loss": 0.0164,
|
| 23315 |
+
"step": 3326
|
| 23316 |
+
},
|
| 23317 |
+
{
|
| 23318 |
+
"epoch": 0.7393333333333333,
|
| 23319 |
+
"grad_norm": 0.0633215382695198,
|
| 23320 |
+
"learning_rate": 5.2249443207126944e-05,
|
| 23321 |
+
"loss": 0.0173,
|
| 23322 |
+
"step": 3327
|
| 23323 |
+
},
|
| 23324 |
+
{
|
| 23325 |
+
"epoch": 0.7395555555555555,
|
| 23326 |
+
"grad_norm": 0.7630808353424072,
|
| 23327 |
+
"learning_rate": 5.2204899777282857e-05,
|
| 23328 |
+
"loss": 0.9757,
|
| 23329 |
+
"step": 3328
|
| 23330 |
+
},
|
| 23331 |
+
{
|
| 23332 |
+
"epoch": 0.7397777777777778,
|
| 23333 |
+
"grad_norm": 0.8969722986221313,
|
| 23334 |
+
"learning_rate": 5.2160356347438755e-05,
|
| 23335 |
+
"loss": 1.6171,
|
| 23336 |
+
"step": 3329
|
| 23337 |
+
},
|
| 23338 |
+
{
|
| 23339 |
+
"epoch": 0.74,
|
| 23340 |
+
"grad_norm": 0.9955383539199829,
|
| 23341 |
+
"learning_rate": 5.2115812917594654e-05,
|
| 23342 |
+
"loss": 1.6627,
|
| 23343 |
+
"step": 3330
|
| 23344 |
+
},
|
| 23345 |
+
{
|
| 23346 |
+
"epoch": 0.7402222222222222,
|
| 23347 |
+
"grad_norm": 1.0531073808670044,
|
| 23348 |
+
"learning_rate": 5.2071269487750566e-05,
|
| 23349 |
+
"loss": 1.7925,
|
| 23350 |
+
"step": 3331
|
| 23351 |
+
},
|
| 23352 |
+
{
|
| 23353 |
+
"epoch": 0.7404444444444445,
|
| 23354 |
+
"grad_norm": 1.1096101999282837,
|
| 23355 |
+
"learning_rate": 5.202672605790646e-05,
|
| 23356 |
+
"loss": 1.4716,
|
| 23357 |
+
"step": 3332
|
| 23358 |
+
},
|
| 23359 |
+
{
|
| 23360 |
+
"epoch": 0.7406666666666667,
|
| 23361 |
+
"grad_norm": 0.06471211463212967,
|
| 23362 |
+
"learning_rate": 5.1982182628062356e-05,
|
| 23363 |
+
"loss": 0.0184,
|
| 23364 |
+
"step": 3333
|
| 23365 |
+
},
|
| 23366 |
+
{
|
| 23367 |
+
"epoch": 0.7408888888888889,
|
| 23368 |
+
"grad_norm": 0.07156452536582947,
|
| 23369 |
+
"learning_rate": 5.193763919821827e-05,
|
| 23370 |
+
"loss": 0.0183,
|
| 23371 |
+
"step": 3334
|
| 23372 |
+
},
|
| 23373 |
+
{
|
| 23374 |
+
"epoch": 0.7411111111111112,
|
| 23375 |
+
"grad_norm": 0.7111669182777405,
|
| 23376 |
+
"learning_rate": 5.189309576837417e-05,
|
| 23377 |
+
"loss": 0.8435,
|
| 23378 |
+
"step": 3335
|
| 23379 |
+
},
|
| 23380 |
+
{
|
| 23381 |
+
"epoch": 0.7413333333333333,
|
| 23382 |
+
"grad_norm": 0.784017026424408,
|
| 23383 |
+
"learning_rate": 5.1848552338530066e-05,
|
| 23384 |
+
"loss": 0.9037,
|
| 23385 |
+
"step": 3336
|
| 23386 |
+
},
|
| 23387 |
+
{
|
| 23388 |
+
"epoch": 0.7415555555555555,
|
| 23389 |
+
"grad_norm": 0.07464414834976196,
|
| 23390 |
+
"learning_rate": 5.180400890868598e-05,
|
| 23391 |
+
"loss": 0.0173,
|
| 23392 |
+
"step": 3337
|
| 23393 |
+
},
|
| 23394 |
+
{
|
| 23395 |
+
"epoch": 0.7417777777777778,
|
| 23396 |
+
"grad_norm": 0.7238468527793884,
|
| 23397 |
+
"learning_rate": 5.1759465478841876e-05,
|
| 23398 |
+
"loss": 0.8807,
|
| 23399 |
+
"step": 3338
|
| 23400 |
+
},
|
| 23401 |
+
{
|
| 23402 |
+
"epoch": 0.742,
|
| 23403 |
+
"grad_norm": 0.07420375943183899,
|
| 23404 |
+
"learning_rate": 5.1714922048997775e-05,
|
| 23405 |
+
"loss": 0.0192,
|
| 23406 |
+
"step": 3339
|
| 23407 |
+
},
|
| 23408 |
+
{
|
| 23409 |
+
"epoch": 0.7422222222222222,
|
| 23410 |
+
"grad_norm": 0.07133994251489639,
|
| 23411 |
+
"learning_rate": 5.167037861915368e-05,
|
| 23412 |
+
"loss": 0.0189,
|
| 23413 |
+
"step": 3340
|
| 23414 |
+
},
|
| 23415 |
+
{
|
| 23416 |
+
"epoch": 0.7424444444444445,
|
| 23417 |
+
"grad_norm": 0.0961189940571785,
|
| 23418 |
+
"learning_rate": 5.162583518930958e-05,
|
| 23419 |
+
"loss": 0.0194,
|
| 23420 |
+
"step": 3341
|
| 23421 |
+
},
|
| 23422 |
+
{
|
| 23423 |
+
"epoch": 0.7426666666666667,
|
| 23424 |
+
"grad_norm": 1.0209311246871948,
|
| 23425 |
+
"learning_rate": 5.158129175946548e-05,
|
| 23426 |
+
"loss": 1.7523,
|
| 23427 |
+
"step": 3342
|
| 23428 |
+
},
|
| 23429 |
+
{
|
| 23430 |
+
"epoch": 0.7428888888888889,
|
| 23431 |
+
"grad_norm": 1.067814588546753,
|
| 23432 |
+
"learning_rate": 5.153674832962139e-05,
|
| 23433 |
+
"loss": 1.7394,
|
| 23434 |
+
"step": 3343
|
| 23435 |
+
},
|
| 23436 |
+
{
|
| 23437 |
+
"epoch": 0.7431111111111111,
|
| 23438 |
+
"grad_norm": 0.10426237434148788,
|
| 23439 |
+
"learning_rate": 5.149220489977729e-05,
|
| 23440 |
+
"loss": 0.0266,
|
| 23441 |
+
"step": 3344
|
| 23442 |
+
},
|
| 23443 |
+
{
|
| 23444 |
+
"epoch": 0.7433333333333333,
|
| 23445 |
+
"grad_norm": 1.1256235837936401,
|
| 23446 |
+
"learning_rate": 5.144766146993319e-05,
|
| 23447 |
+
"loss": 1.5493,
|
| 23448 |
+
"step": 3345
|
| 23449 |
+
},
|
| 23450 |
+
{
|
| 23451 |
+
"epoch": 0.7435555555555555,
|
| 23452 |
+
"grad_norm": 1.0838463306427002,
|
| 23453 |
+
"learning_rate": 5.140311804008909e-05,
|
| 23454 |
+
"loss": 1.508,
|
| 23455 |
+
"step": 3346
|
| 23456 |
+
},
|
| 23457 |
+
{
|
| 23458 |
+
"epoch": 0.7437777777777778,
|
| 23459 |
+
"grad_norm": 1.0034325122833252,
|
| 23460 |
+
"learning_rate": 5.135857461024499e-05,
|
| 23461 |
+
"loss": 1.3716,
|
| 23462 |
+
"step": 3347
|
| 23463 |
+
},
|
| 23464 |
+
{
|
| 23465 |
+
"epoch": 0.744,
|
| 23466 |
+
"grad_norm": 1.1057904958724976,
|
| 23467 |
+
"learning_rate": 5.131403118040089e-05,
|
| 23468 |
+
"loss": 0.9587,
|
| 23469 |
+
"step": 3348
|
| 23470 |
+
},
|
| 23471 |
+
{
|
| 23472 |
+
"epoch": 0.7442222222222222,
|
| 23473 |
+
"grad_norm": 0.19667142629623413,
|
| 23474 |
+
"learning_rate": 5.12694877505568e-05,
|
| 23475 |
+
"loss": 0.0377,
|
| 23476 |
+
"step": 3349
|
| 23477 |
+
},
|
| 23478 |
+
{
|
| 23479 |
+
"epoch": 0.7444444444444445,
|
| 23480 |
+
"grad_norm": 1.0404895544052124,
|
| 23481 |
+
"learning_rate": 5.12249443207127e-05,
|
| 23482 |
+
"loss": 1.0799,
|
| 23483 |
+
"step": 3350
|
| 23484 |
+
},
|
| 23485 |
+
{
|
| 23486 |
+
"epoch": 0.7446666666666667,
|
| 23487 |
+
"grad_norm": 0.8521629571914673,
|
| 23488 |
+
"learning_rate": 5.11804008908686e-05,
|
| 23489 |
+
"loss": 2.0826,
|
| 23490 |
+
"step": 3351
|
| 23491 |
+
},
|
| 23492 |
+
{
|
| 23493 |
+
"epoch": 0.7448888888888889,
|
| 23494 |
+
"grad_norm": 0.046493686735630035,
|
| 23495 |
+
"learning_rate": 5.1135857461024504e-05,
|
| 23496 |
+
"loss": 0.0101,
|
| 23497 |
+
"step": 3352
|
| 23498 |
+
},
|
| 23499 |
+
{
|
| 23500 |
+
"epoch": 0.7451111111111111,
|
| 23501 |
+
"grad_norm": 0.04533799737691879,
|
| 23502 |
+
"learning_rate": 5.10913140311804e-05,
|
| 23503 |
+
"loss": 0.0102,
|
| 23504 |
+
"step": 3353
|
| 23505 |
+
},
|
| 23506 |
+
{
|
| 23507 |
+
"epoch": 0.7453333333333333,
|
| 23508 |
+
"grad_norm": 0.6256393194198608,
|
| 23509 |
+
"learning_rate": 5.10467706013363e-05,
|
| 23510 |
+
"loss": 1.2161,
|
| 23511 |
+
"step": 3354
|
| 23512 |
+
},
|
| 23513 |
+
{
|
| 23514 |
+
"epoch": 0.7455555555555555,
|
| 23515 |
+
"grad_norm": 0.5878841280937195,
|
| 23516 |
+
"learning_rate": 5.100222717149221e-05,
|
| 23517 |
+
"loss": 1.1603,
|
| 23518 |
+
"step": 3355
|
| 23519 |
+
},
|
| 23520 |
+
{
|
| 23521 |
+
"epoch": 0.7457777777777778,
|
| 23522 |
+
"grad_norm": 0.04651748016476631,
|
| 23523 |
+
"learning_rate": 5.095768374164811e-05,
|
| 23524 |
+
"loss": 0.01,
|
| 23525 |
+
"step": 3356
|
| 23526 |
+
},
|
| 23527 |
+
{
|
| 23528 |
+
"epoch": 0.746,
|
| 23529 |
+
"grad_norm": 0.03794243186712265,
|
| 23530 |
+
"learning_rate": 5.091314031180401e-05,
|
| 23531 |
+
"loss": 0.01,
|
| 23532 |
+
"step": 3357
|
| 23533 |
+
},
|
| 23534 |
+
{
|
| 23535 |
+
"epoch": 0.7462222222222222,
|
| 23536 |
+
"grad_norm": 0.04922659322619438,
|
| 23537 |
+
"learning_rate": 5.0868596881959916e-05,
|
| 23538 |
+
"loss": 0.0097,
|
| 23539 |
+
"step": 3358
|
| 23540 |
+
},
|
| 23541 |
+
{
|
| 23542 |
+
"epoch": 0.7464444444444445,
|
| 23543 |
+
"grad_norm": 0.8625622391700745,
|
| 23544 |
+
"learning_rate": 5.0824053452115814e-05,
|
| 23545 |
+
"loss": 1.8859,
|
| 23546 |
+
"step": 3359
|
| 23547 |
+
},
|
| 23548 |
+
{
|
| 23549 |
+
"epoch": 0.7466666666666667,
|
| 23550 |
+
"grad_norm": 0.8704177141189575,
|
| 23551 |
+
"learning_rate": 5.077951002227171e-05,
|
| 23552 |
+
"loss": 1.9087,
|
| 23553 |
+
"step": 3360
|
| 23554 |
+
},
|
| 23555 |
+
{
|
| 23556 |
+
"epoch": 0.7468888888888889,
|
| 23557 |
+
"grad_norm": 0.9514003992080688,
|
| 23558 |
+
"learning_rate": 5.0734966592427625e-05,
|
| 23559 |
+
"loss": 2.1152,
|
| 23560 |
+
"step": 3361
|
| 23561 |
+
},
|
| 23562 |
+
{
|
| 23563 |
+
"epoch": 0.7471111111111111,
|
| 23564 |
+
"grad_norm": 0.9952490925788879,
|
| 23565 |
+
"learning_rate": 5.0690423162583524e-05,
|
| 23566 |
+
"loss": 2.237,
|
| 23567 |
+
"step": 3362
|
| 23568 |
+
},
|
| 23569 |
+
{
|
| 23570 |
+
"epoch": 0.7473333333333333,
|
| 23571 |
+
"grad_norm": 1.0425519943237305,
|
| 23572 |
+
"learning_rate": 5.064587973273942e-05,
|
| 23573 |
+
"loss": 2.1412,
|
| 23574 |
+
"step": 3363
|
| 23575 |
+
},
|
| 23576 |
+
{
|
| 23577 |
+
"epoch": 0.7475555555555555,
|
| 23578 |
+
"grad_norm": 0.7753322124481201,
|
| 23579 |
+
"learning_rate": 5.060133630289533e-05,
|
| 23580 |
+
"loss": 1.7639,
|
| 23581 |
+
"step": 3364
|
| 23582 |
+
},
|
| 23583 |
+
{
|
| 23584 |
+
"epoch": 0.7477777777777778,
|
| 23585 |
+
"grad_norm": 0.9439111351966858,
|
| 23586 |
+
"learning_rate": 5.0556792873051226e-05,
|
| 23587 |
+
"loss": 1.7622,
|
| 23588 |
+
"step": 3365
|
| 23589 |
+
},
|
| 23590 |
+
{
|
| 23591 |
+
"epoch": 0.748,
|
| 23592 |
+
"grad_norm": 0.9274625778198242,
|
| 23593 |
+
"learning_rate": 5.0512249443207125e-05,
|
| 23594 |
+
"loss": 2.1017,
|
| 23595 |
+
"step": 3366
|
| 23596 |
+
},
|
| 23597 |
+
{
|
| 23598 |
+
"epoch": 0.7482222222222222,
|
| 23599 |
+
"grad_norm": 0.9550508856773376,
|
| 23600 |
+
"learning_rate": 5.046770601336304e-05,
|
| 23601 |
+
"loss": 1.8416,
|
| 23602 |
+
"step": 3367
|
| 23603 |
+
},
|
| 23604 |
+
{
|
| 23605 |
+
"epoch": 0.7484444444444445,
|
| 23606 |
+
"grad_norm": 0.8628423810005188,
|
| 23607 |
+
"learning_rate": 5.0423162583518935e-05,
|
| 23608 |
+
"loss": 1.9227,
|
| 23609 |
+
"step": 3368
|
| 23610 |
+
},
|
| 23611 |
+
{
|
| 23612 |
+
"epoch": 0.7486666666666667,
|
| 23613 |
+
"grad_norm": 1.0649088621139526,
|
| 23614 |
+
"learning_rate": 5.0378619153674834e-05,
|
| 23615 |
+
"loss": 2.1865,
|
| 23616 |
+
"step": 3369
|
| 23617 |
+
},
|
| 23618 |
+
{
|
| 23619 |
+
"epoch": 0.7488888888888889,
|
| 23620 |
+
"grad_norm": 0.9452845454216003,
|
| 23621 |
+
"learning_rate": 5.033407572383074e-05,
|
| 23622 |
+
"loss": 1.9341,
|
| 23623 |
+
"step": 3370
|
| 23624 |
+
},
|
| 23625 |
+
{
|
| 23626 |
+
"epoch": 0.7491111111111111,
|
| 23627 |
+
"grad_norm": 0.9852356910705566,
|
| 23628 |
+
"learning_rate": 5.028953229398664e-05,
|
| 23629 |
+
"loss": 1.6767,
|
| 23630 |
+
"step": 3371
|
| 23631 |
+
},
|
| 23632 |
+
{
|
| 23633 |
+
"epoch": 0.7493333333333333,
|
| 23634 |
+
"grad_norm": 0.9458546042442322,
|
| 23635 |
+
"learning_rate": 5.0244988864142536e-05,
|
| 23636 |
+
"loss": 1.786,
|
| 23637 |
+
"step": 3372
|
| 23638 |
+
},
|
| 23639 |
+
{
|
| 23640 |
+
"epoch": 0.7495555555555555,
|
| 23641 |
+
"grad_norm": 0.07178652286529541,
|
| 23642 |
+
"learning_rate": 5.020044543429845e-05,
|
| 23643 |
+
"loss": 0.015,
|
| 23644 |
+
"step": 3373
|
| 23645 |
+
},
|
| 23646 |
+
{
|
| 23647 |
+
"epoch": 0.7497777777777778,
|
| 23648 |
+
"grad_norm": 0.07055787742137909,
|
| 23649 |
+
"learning_rate": 5.015590200445435e-05,
|
| 23650 |
+
"loss": 0.0153,
|
| 23651 |
+
"step": 3374
|
| 23652 |
+
},
|
| 23653 |
+
{
|
| 23654 |
+
"epoch": 0.75,
|
| 23655 |
+
"grad_norm": 0.6104269027709961,
|
| 23656 |
+
"learning_rate": 5.0111358574610246e-05,
|
| 23657 |
+
"loss": 0.8618,
|
| 23658 |
+
"step": 3375
|
| 23659 |
+
},
|
| 23660 |
+
{
|
| 23661 |
+
"epoch": 0.7502222222222222,
|
| 23662 |
+
"grad_norm": 0.6599386930465698,
|
| 23663 |
+
"learning_rate": 5.006681514476616e-05,
|
| 23664 |
+
"loss": 0.8642,
|
| 23665 |
+
"step": 3376
|
| 23666 |
+
},
|
| 23667 |
+
{
|
| 23668 |
+
"epoch": 0.7504444444444445,
|
| 23669 |
+
"grad_norm": 0.6750035881996155,
|
| 23670 |
+
"learning_rate": 5.0022271714922056e-05,
|
| 23671 |
+
"loss": 0.8647,
|
| 23672 |
+
"step": 3377
|
| 23673 |
+
},
|
| 23674 |
+
{
|
| 23675 |
+
"epoch": 0.7506666666666667,
|
| 23676 |
+
"grad_norm": 0.9692963361740112,
|
| 23677 |
+
"learning_rate": 4.997772828507795e-05,
|
| 23678 |
+
"loss": 1.8036,
|
| 23679 |
+
"step": 3378
|
| 23680 |
+
},
|
| 23681 |
+
{
|
| 23682 |
+
"epoch": 0.7508888888888889,
|
| 23683 |
+
"grad_norm": 1.0836691856384277,
|
| 23684 |
+
"learning_rate": 4.9933184855233854e-05,
|
| 23685 |
+
"loss": 2.039,
|
| 23686 |
+
"step": 3379
|
| 23687 |
+
},
|
| 23688 |
+
{
|
| 23689 |
+
"epoch": 0.7511111111111111,
|
| 23690 |
+
"grad_norm": 0.06479348987340927,
|
| 23691 |
+
"learning_rate": 4.988864142538976e-05,
|
| 23692 |
+
"loss": 0.0173,
|
| 23693 |
+
"step": 3380
|
| 23694 |
+
},
|
| 23695 |
+
{
|
| 23696 |
+
"epoch": 0.7513333333333333,
|
| 23697 |
+
"grad_norm": 0.06957981735467911,
|
| 23698 |
+
"learning_rate": 4.984409799554566e-05,
|
| 23699 |
+
"loss": 0.0166,
|
| 23700 |
+
"step": 3381
|
| 23701 |
+
},
|
| 23702 |
+
{
|
| 23703 |
+
"epoch": 0.7515555555555555,
|
| 23704 |
+
"grad_norm": 0.666901707649231,
|
| 23705 |
+
"learning_rate": 4.979955456570156e-05,
|
| 23706 |
+
"loss": 0.7578,
|
| 23707 |
+
"step": 3382
|
| 23708 |
+
},
|
| 23709 |
+
{
|
| 23710 |
+
"epoch": 0.7517777777777778,
|
| 23711 |
+
"grad_norm": 1.0305155515670776,
|
| 23712 |
+
"learning_rate": 4.975501113585747e-05,
|
| 23713 |
+
"loss": 1.6703,
|
| 23714 |
+
"step": 3383
|
| 23715 |
+
},
|
| 23716 |
+
{
|
| 23717 |
+
"epoch": 0.752,
|
| 23718 |
+
"grad_norm": 0.9969210624694824,
|
| 23719 |
+
"learning_rate": 4.971046770601337e-05,
|
| 23720 |
+
"loss": 1.7831,
|
| 23721 |
+
"step": 3384
|
| 23722 |
+
},
|
| 23723 |
+
{
|
| 23724 |
+
"epoch": 0.7522222222222222,
|
| 23725 |
+
"grad_norm": 0.068308025598526,
|
| 23726 |
+
"learning_rate": 4.9665924276169265e-05,
|
| 23727 |
+
"loss": 0.018,
|
| 23728 |
+
"step": 3385
|
| 23729 |
+
},
|
| 23730 |
+
{
|
| 23731 |
+
"epoch": 0.7524444444444445,
|
| 23732 |
+
"grad_norm": 0.06835668534040451,
|
| 23733 |
+
"learning_rate": 4.962138084632517e-05,
|
| 23734 |
+
"loss": 0.0171,
|
| 23735 |
+
"step": 3386
|
| 23736 |
+
},
|
| 23737 |
+
{
|
| 23738 |
+
"epoch": 0.7526666666666667,
|
| 23739 |
+
"grad_norm": 0.562114417552948,
|
| 23740 |
+
"learning_rate": 4.957683741648107e-05,
|
| 23741 |
+
"loss": 0.8015,
|
| 23742 |
+
"step": 3387
|
| 23743 |
+
},
|
| 23744 |
+
{
|
| 23745 |
+
"epoch": 0.7528888888888889,
|
| 23746 |
+
"grad_norm": 0.9326373338699341,
|
| 23747 |
+
"learning_rate": 4.9532293986636975e-05,
|
| 23748 |
+
"loss": 1.7364,
|
| 23749 |
+
"step": 3388
|
| 23750 |
+
},
|
| 23751 |
+
{
|
| 23752 |
+
"epoch": 0.7531111111111111,
|
| 23753 |
+
"grad_norm": 1.0560567378997803,
|
| 23754 |
+
"learning_rate": 4.948775055679288e-05,
|
| 23755 |
+
"loss": 1.3854,
|
| 23756 |
+
"step": 3389
|
| 23757 |
+
},
|
| 23758 |
+
{
|
| 23759 |
+
"epoch": 0.7533333333333333,
|
| 23760 |
+
"grad_norm": 1.0617526769638062,
|
| 23761 |
+
"learning_rate": 4.944320712694878e-05,
|
| 23762 |
+
"loss": 1.3826,
|
| 23763 |
+
"step": 3390
|
| 23764 |
+
},
|
| 23765 |
+
{
|
| 23766 |
+
"epoch": 0.7535555555555555,
|
| 23767 |
+
"grad_norm": 0.6773163080215454,
|
| 23768 |
+
"learning_rate": 4.939866369710468e-05,
|
| 23769 |
+
"loss": 0.9724,
|
| 23770 |
+
"step": 3391
|
| 23771 |
+
},
|
| 23772 |
+
{
|
| 23773 |
+
"epoch": 0.7537777777777778,
|
| 23774 |
+
"grad_norm": 0.8919631838798523,
|
| 23775 |
+
"learning_rate": 4.935412026726058e-05,
|
| 23776 |
+
"loss": 1.4029,
|
| 23777 |
+
"step": 3392
|
| 23778 |
+
},
|
| 23779 |
+
{
|
| 23780 |
+
"epoch": 0.754,
|
| 23781 |
+
"grad_norm": 1.0007896423339844,
|
| 23782 |
+
"learning_rate": 4.930957683741648e-05,
|
| 23783 |
+
"loss": 1.3675,
|
| 23784 |
+
"step": 3393
|
| 23785 |
+
},
|
| 23786 |
+
{
|
| 23787 |
+
"epoch": 0.7542222222222222,
|
| 23788 |
+
"grad_norm": 1.1181669235229492,
|
| 23789 |
+
"learning_rate": 4.9265033407572387e-05,
|
| 23790 |
+
"loss": 1.5695,
|
| 23791 |
+
"step": 3394
|
| 23792 |
+
},
|
| 23793 |
+
{
|
| 23794 |
+
"epoch": 0.7544444444444445,
|
| 23795 |
+
"grad_norm": 1.058223843574524,
|
| 23796 |
+
"learning_rate": 4.922048997772829e-05,
|
| 23797 |
+
"loss": 1.444,
|
| 23798 |
+
"step": 3395
|
| 23799 |
+
},
|
| 23800 |
+
{
|
| 23801 |
+
"epoch": 0.7546666666666667,
|
| 23802 |
+
"grad_norm": 1.0917662382125854,
|
| 23803 |
+
"learning_rate": 4.917594654788419e-05,
|
| 23804 |
+
"loss": 1.5776,
|
| 23805 |
+
"step": 3396
|
| 23806 |
+
},
|
| 23807 |
+
{
|
| 23808 |
+
"epoch": 0.7548888888888889,
|
| 23809 |
+
"grad_norm": 1.2129132747650146,
|
| 23810 |
+
"learning_rate": 4.913140311804009e-05,
|
| 23811 |
+
"loss": 1.5378,
|
| 23812 |
+
"step": 3397
|
| 23813 |
+
},
|
| 23814 |
+
{
|
| 23815 |
+
"epoch": 0.7551111111111111,
|
| 23816 |
+
"grad_norm": 0.7757513523101807,
|
| 23817 |
+
"learning_rate": 4.908685968819599e-05,
|
| 23818 |
+
"loss": 0.7143,
|
| 23819 |
+
"step": 3398
|
| 23820 |
+
},
|
| 23821 |
+
{
|
| 23822 |
+
"epoch": 0.7553333333333333,
|
| 23823 |
+
"grad_norm": 1.0675660371780396,
|
| 23824 |
+
"learning_rate": 4.904231625835189e-05,
|
| 23825 |
+
"loss": 1.2625,
|
| 23826 |
+
"step": 3399
|
| 23827 |
+
},
|
| 23828 |
+
{
|
| 23829 |
+
"epoch": 0.7555555555555555,
|
| 23830 |
+
"grad_norm": 0.7911191582679749,
|
| 23831 |
+
"learning_rate": 4.89977728285078e-05,
|
| 23832 |
+
"loss": 0.6726,
|
| 23833 |
+
"step": 3400
|
| 23834 |
+
},
|
| 23835 |
+
{
|
| 23836 |
+
"epoch": 0.7557777777777778,
|
| 23837 |
+
"grad_norm": 0.936028003692627,
|
| 23838 |
+
"learning_rate": 4.89532293986637e-05,
|
| 23839 |
+
"loss": 2.5741,
|
| 23840 |
+
"step": 3401
|
| 23841 |
+
},
|
| 23842 |
+
{
|
| 23843 |
+
"epoch": 0.756,
|
| 23844 |
+
"grad_norm": 0.04625101387500763,
|
| 23845 |
+
"learning_rate": 4.89086859688196e-05,
|
| 23846 |
+
"loss": 0.0098,
|
| 23847 |
+
"step": 3402
|
| 23848 |
+
},
|
| 23849 |
+
{
|
| 23850 |
+
"epoch": 0.7562222222222222,
|
| 23851 |
+
"grad_norm": 0.5739651918411255,
|
| 23852 |
+
"learning_rate": 4.886414253897551e-05,
|
| 23853 |
+
"loss": 1.0021,
|
| 23854 |
+
"step": 3403
|
| 23855 |
+
},
|
| 23856 |
+
{
|
| 23857 |
+
"epoch": 0.7564444444444445,
|
| 23858 |
+
"grad_norm": 0.874405562877655,
|
| 23859 |
+
"learning_rate": 4.8819599109131406e-05,
|
| 23860 |
+
"loss": 2.1036,
|
| 23861 |
+
"step": 3404
|
| 23862 |
+
},
|
| 23863 |
+
{
|
| 23864 |
+
"epoch": 0.7566666666666667,
|
| 23865 |
+
"grad_norm": 0.5654922723770142,
|
| 23866 |
+
"learning_rate": 4.8775055679287305e-05,
|
| 23867 |
+
"loss": 0.9892,
|
| 23868 |
+
"step": 3405
|
| 23869 |
+
},
|
| 23870 |
+
{
|
| 23871 |
+
"epoch": 0.7568888888888889,
|
| 23872 |
+
"grad_norm": 0.6591737866401672,
|
| 23873 |
+
"learning_rate": 4.873051224944321e-05,
|
| 23874 |
+
"loss": 0.9575,
|
| 23875 |
+
"step": 3406
|
| 23876 |
+
},
|
| 23877 |
+
{
|
| 23878 |
+
"epoch": 0.7571111111111111,
|
| 23879 |
+
"grad_norm": 0.05461383983492851,
|
| 23880 |
+
"learning_rate": 4.868596881959911e-05,
|
| 23881 |
+
"loss": 0.011,
|
| 23882 |
+
"step": 3407
|
| 23883 |
+
},
|
| 23884 |
+
{
|
| 23885 |
+
"epoch": 0.7573333333333333,
|
| 23886 |
+
"grad_norm": 0.0622735358774662,
|
| 23887 |
+
"learning_rate": 4.8641425389755014e-05,
|
| 23888 |
+
"loss": 0.0112,
|
| 23889 |
+
"step": 3408
|
| 23890 |
+
},
|
| 23891 |
+
{
|
| 23892 |
+
"epoch": 0.7575555555555555,
|
| 23893 |
+
"grad_norm": 0.059408292174339294,
|
| 23894 |
+
"learning_rate": 4.859688195991092e-05,
|
| 23895 |
+
"loss": 0.011,
|
| 23896 |
+
"step": 3409
|
| 23897 |
+
},
|
| 23898 |
+
{
|
| 23899 |
+
"epoch": 0.7577777777777778,
|
| 23900 |
+
"grad_norm": 0.6495372653007507,
|
| 23901 |
+
"learning_rate": 4.855233853006682e-05,
|
| 23902 |
+
"loss": 0.8378,
|
| 23903 |
+
"step": 3410
|
| 23904 |
+
},
|
| 23905 |
+
{
|
| 23906 |
+
"epoch": 0.758,
|
| 23907 |
+
"grad_norm": 0.9061746001243591,
|
| 23908 |
+
"learning_rate": 4.850779510022272e-05,
|
| 23909 |
+
"loss": 2.2088,
|
| 23910 |
+
"step": 3411
|
| 23911 |
+
},
|
| 23912 |
+
{
|
| 23913 |
+
"epoch": 0.7582222222222222,
|
| 23914 |
+
"grad_norm": 0.8633875846862793,
|
| 23915 |
+
"learning_rate": 4.846325167037862e-05,
|
| 23916 |
+
"loss": 1.9511,
|
| 23917 |
+
"step": 3412
|
| 23918 |
+
},
|
| 23919 |
+
{
|
| 23920 |
+
"epoch": 0.7584444444444445,
|
| 23921 |
+
"grad_norm": 1.055767297744751,
|
| 23922 |
+
"learning_rate": 4.841870824053452e-05,
|
| 23923 |
+
"loss": 1.9755,
|
| 23924 |
+
"step": 3413
|
| 23925 |
+
},
|
| 23926 |
+
{
|
| 23927 |
+
"epoch": 0.7586666666666667,
|
| 23928 |
+
"grad_norm": 0.8679887056350708,
|
| 23929 |
+
"learning_rate": 4.8374164810690426e-05,
|
| 23930 |
+
"loss": 1.8676,
|
| 23931 |
+
"step": 3414
|
| 23932 |
+
},
|
| 23933 |
+
{
|
| 23934 |
+
"epoch": 0.7588888888888888,
|
| 23935 |
+
"grad_norm": 0.9158828258514404,
|
| 23936 |
+
"learning_rate": 4.832962138084633e-05,
|
| 23937 |
+
"loss": 2.0772,
|
| 23938 |
+
"step": 3415
|
| 23939 |
+
},
|
| 23940 |
+
{
|
| 23941 |
+
"epoch": 0.7591111111111111,
|
| 23942 |
+
"grad_norm": 0.6672974228858948,
|
| 23943 |
+
"learning_rate": 4.828507795100223e-05,
|
| 23944 |
+
"loss": 1.1813,
|
| 23945 |
+
"step": 3416
|
| 23946 |
+
},
|
| 23947 |
+
{
|
| 23948 |
+
"epoch": 0.7593333333333333,
|
| 23949 |
+
"grad_norm": 0.9546223282814026,
|
| 23950 |
+
"learning_rate": 4.824053452115813e-05,
|
| 23951 |
+
"loss": 1.9467,
|
| 23952 |
+
"step": 3417
|
| 23953 |
+
},
|
| 23954 |
+
{
|
| 23955 |
+
"epoch": 0.7595555555555555,
|
| 23956 |
+
"grad_norm": 1.0391935110092163,
|
| 23957 |
+
"learning_rate": 4.8195991091314034e-05,
|
| 23958 |
+
"loss": 2.041,
|
| 23959 |
+
"step": 3418
|
| 23960 |
+
},
|
| 23961 |
+
{
|
| 23962 |
+
"epoch": 0.7597777777777778,
|
| 23963 |
+
"grad_norm": 1.0147621631622314,
|
| 23964 |
+
"learning_rate": 4.815144766146993e-05,
|
| 23965 |
+
"loss": 2.0473,
|
| 23966 |
+
"step": 3419
|
| 23967 |
+
},
|
| 23968 |
+
{
|
| 23969 |
+
"epoch": 0.76,
|
| 23970 |
+
"grad_norm": 0.6334058046340942,
|
| 23971 |
+
"learning_rate": 4.810690423162584e-05,
|
| 23972 |
+
"loss": 0.8882,
|
| 23973 |
+
"step": 3420
|
| 23974 |
+
},
|
| 23975 |
+
{
|
| 23976 |
+
"epoch": 0.7602222222222222,
|
| 23977 |
+
"grad_norm": 0.06809257715940475,
|
| 23978 |
+
"learning_rate": 4.806236080178174e-05,
|
| 23979 |
+
"loss": 0.0153,
|
| 23980 |
+
"step": 3421
|
| 23981 |
+
},
|
| 23982 |
+
{
|
| 23983 |
+
"epoch": 0.7604444444444445,
|
| 23984 |
+
"grad_norm": 0.06833475828170776,
|
| 23985 |
+
"learning_rate": 4.801781737193764e-05,
|
| 23986 |
+
"loss": 0.0156,
|
| 23987 |
+
"step": 3422
|
| 23988 |
+
},
|
| 23989 |
+
{
|
| 23990 |
+
"epoch": 0.7606666666666667,
|
| 23991 |
+
"grad_norm": 0.09722508490085602,
|
| 23992 |
+
"learning_rate": 4.797327394209355e-05,
|
| 23993 |
+
"loss": 0.0179,
|
| 23994 |
+
"step": 3423
|
| 23995 |
+
},
|
| 23996 |
+
{
|
| 23997 |
+
"epoch": 0.7608888888888888,
|
| 23998 |
+
"grad_norm": 0.92330402135849,
|
| 23999 |
+
"learning_rate": 4.7928730512249446e-05,
|
| 24000 |
+
"loss": 1.7349,
|
| 24001 |
+
"step": 3424
|
| 24002 |
+
},
|
| 24003 |
+
{
|
| 24004 |
+
"epoch": 0.7611111111111111,
|
| 24005 |
+
"grad_norm": 1.0066584348678589,
|
| 24006 |
+
"learning_rate": 4.7884187082405344e-05,
|
| 24007 |
+
"loss": 1.6615,
|
| 24008 |
+
"step": 3425
|
| 24009 |
+
},
|
| 24010 |
+
{
|
| 24011 |
+
"epoch": 0.7613333333333333,
|
| 24012 |
+
"grad_norm": 0.9122890830039978,
|
| 24013 |
+
"learning_rate": 4.783964365256125e-05,
|
| 24014 |
+
"loss": 1.9283,
|
| 24015 |
+
"step": 3426
|
| 24016 |
+
},
|
| 24017 |
+
{
|
| 24018 |
+
"epoch": 0.7615555555555555,
|
| 24019 |
+
"grad_norm": 1.0834369659423828,
|
| 24020 |
+
"learning_rate": 4.7795100222717155e-05,
|
| 24021 |
+
"loss": 1.8457,
|
| 24022 |
+
"step": 3427
|
| 24023 |
+
},
|
| 24024 |
+
{
|
| 24025 |
+
"epoch": 0.7617777777777778,
|
| 24026 |
+
"grad_norm": 0.9122326970100403,
|
| 24027 |
+
"learning_rate": 4.7750556792873054e-05,
|
| 24028 |
+
"loss": 1.5779,
|
| 24029 |
+
"step": 3428
|
| 24030 |
+
},
|
| 24031 |
+
{
|
| 24032 |
+
"epoch": 0.762,
|
| 24033 |
+
"grad_norm": 0.6459372639656067,
|
| 24034 |
+
"learning_rate": 4.770601336302896e-05,
|
| 24035 |
+
"loss": 0.8526,
|
| 24036 |
+
"step": 3429
|
| 24037 |
+
},
|
| 24038 |
+
{
|
| 24039 |
+
"epoch": 0.7622222222222222,
|
| 24040 |
+
"grad_norm": 0.06661590933799744,
|
| 24041 |
+
"learning_rate": 4.766146993318486e-05,
|
| 24042 |
+
"loss": 0.018,
|
| 24043 |
+
"step": 3430
|
| 24044 |
+
},
|
| 24045 |
+
{
|
| 24046 |
+
"epoch": 0.7624444444444445,
|
| 24047 |
+
"grad_norm": 0.06595264375209808,
|
| 24048 |
+
"learning_rate": 4.7616926503340756e-05,
|
| 24049 |
+
"loss": 0.0176,
|
| 24050 |
+
"step": 3431
|
| 24051 |
+
},
|
| 24052 |
+
{
|
| 24053 |
+
"epoch": 0.7626666666666667,
|
| 24054 |
+
"grad_norm": 0.06258884072303772,
|
| 24055 |
+
"learning_rate": 4.757238307349666e-05,
|
| 24056 |
+
"loss": 0.0175,
|
| 24057 |
+
"step": 3432
|
| 24058 |
+
},
|
| 24059 |
+
{
|
| 24060 |
+
"epoch": 0.7628888888888888,
|
| 24061 |
+
"grad_norm": 0.9908372163772583,
|
| 24062 |
+
"learning_rate": 4.752783964365256e-05,
|
| 24063 |
+
"loss": 1.5601,
|
| 24064 |
+
"step": 3433
|
| 24065 |
+
},
|
| 24066 |
+
{
|
| 24067 |
+
"epoch": 0.7631111111111111,
|
| 24068 |
+
"grad_norm": 1.1008018255233765,
|
| 24069 |
+
"learning_rate": 4.7483296213808465e-05,
|
| 24070 |
+
"loss": 1.9175,
|
| 24071 |
+
"step": 3434
|
| 24072 |
+
},
|
| 24073 |
+
{
|
| 24074 |
+
"epoch": 0.7633333333333333,
|
| 24075 |
+
"grad_norm": 0.06766713410615921,
|
| 24076 |
+
"learning_rate": 4.743875278396437e-05,
|
| 24077 |
+
"loss": 0.0185,
|
| 24078 |
+
"step": 3435
|
| 24079 |
+
},
|
| 24080 |
+
{
|
| 24081 |
+
"epoch": 0.7635555555555555,
|
| 24082 |
+
"grad_norm": 0.06862013787031174,
|
| 24083 |
+
"learning_rate": 4.739420935412027e-05,
|
| 24084 |
+
"loss": 0.018,
|
| 24085 |
+
"step": 3436
|
| 24086 |
+
},
|
| 24087 |
+
{
|
| 24088 |
+
"epoch": 0.7637777777777778,
|
| 24089 |
+
"grad_norm": 0.995215654373169,
|
| 24090 |
+
"learning_rate": 4.734966592427617e-05,
|
| 24091 |
+
"loss": 1.6609,
|
| 24092 |
+
"step": 3437
|
| 24093 |
+
},
|
| 24094 |
+
{
|
| 24095 |
+
"epoch": 0.764,
|
| 24096 |
+
"grad_norm": 1.1150976419448853,
|
| 24097 |
+
"learning_rate": 4.730512249443207e-05,
|
| 24098 |
+
"loss": 1.5773,
|
| 24099 |
+
"step": 3438
|
| 24100 |
+
},
|
| 24101 |
+
{
|
| 24102 |
+
"epoch": 0.7642222222222222,
|
| 24103 |
+
"grad_norm": 0.10110121965408325,
|
| 24104 |
+
"learning_rate": 4.726057906458797e-05,
|
| 24105 |
+
"loss": 0.0254,
|
| 24106 |
+
"step": 3439
|
| 24107 |
+
},
|
| 24108 |
+
{
|
| 24109 |
+
"epoch": 0.7644444444444445,
|
| 24110 |
+
"grad_norm": 0.8509777188301086,
|
| 24111 |
+
"learning_rate": 4.721603563474388e-05,
|
| 24112 |
+
"loss": 0.8449,
|
| 24113 |
+
"step": 3440
|
| 24114 |
+
},
|
| 24115 |
+
{
|
| 24116 |
+
"epoch": 0.7646666666666667,
|
| 24117 |
+
"grad_norm": 1.163260579109192,
|
| 24118 |
+
"learning_rate": 4.717149220489978e-05,
|
| 24119 |
+
"loss": 1.6988,
|
| 24120 |
+
"step": 3441
|
| 24121 |
+
},
|
| 24122 |
+
{
|
| 24123 |
+
"epoch": 0.7648888888888888,
|
| 24124 |
+
"grad_norm": 1.1963449716567993,
|
| 24125 |
+
"learning_rate": 4.712694877505568e-05,
|
| 24126 |
+
"loss": 1.6756,
|
| 24127 |
+
"step": 3442
|
| 24128 |
+
},
|
| 24129 |
+
{
|
| 24130 |
+
"epoch": 0.7651111111111111,
|
| 24131 |
+
"grad_norm": 1.1867884397506714,
|
| 24132 |
+
"learning_rate": 4.7082405345211587e-05,
|
| 24133 |
+
"loss": 1.6131,
|
| 24134 |
+
"step": 3443
|
| 24135 |
+
},
|
| 24136 |
+
{
|
| 24137 |
+
"epoch": 0.7653333333333333,
|
| 24138 |
+
"grad_norm": 1.0478819608688354,
|
| 24139 |
+
"learning_rate": 4.7037861915367485e-05,
|
| 24140 |
+
"loss": 1.4666,
|
| 24141 |
+
"step": 3444
|
| 24142 |
+
},
|
| 24143 |
+
{
|
| 24144 |
+
"epoch": 0.7655555555555555,
|
| 24145 |
+
"grad_norm": 1.076615571975708,
|
| 24146 |
+
"learning_rate": 4.6993318485523384e-05,
|
| 24147 |
+
"loss": 1.3148,
|
| 24148 |
+
"step": 3445
|
| 24149 |
+
},
|
| 24150 |
+
{
|
| 24151 |
+
"epoch": 0.7657777777777778,
|
| 24152 |
+
"grad_norm": 0.7551054954528809,
|
| 24153 |
+
"learning_rate": 4.694877505567929e-05,
|
| 24154 |
+
"loss": 0.7423,
|
| 24155 |
+
"step": 3446
|
| 24156 |
+
},
|
| 24157 |
+
{
|
| 24158 |
+
"epoch": 0.766,
|
| 24159 |
+
"grad_norm": 0.7709291577339172,
|
| 24160 |
+
"learning_rate": 4.6904231625835194e-05,
|
| 24161 |
+
"loss": 0.749,
|
| 24162 |
+
"step": 3447
|
| 24163 |
+
},
|
| 24164 |
+
{
|
| 24165 |
+
"epoch": 0.7662222222222222,
|
| 24166 |
+
"grad_norm": 0.9779494404792786,
|
| 24167 |
+
"learning_rate": 4.685968819599109e-05,
|
| 24168 |
+
"loss": 1.1534,
|
| 24169 |
+
"step": 3448
|
| 24170 |
+
},
|
| 24171 |
+
{
|
| 24172 |
+
"epoch": 0.7664444444444445,
|
| 24173 |
+
"grad_norm": 0.7176189422607422,
|
| 24174 |
+
"learning_rate": 4.6815144766147e-05,
|
| 24175 |
+
"loss": 0.5791,
|
| 24176 |
+
"step": 3449
|
| 24177 |
+
},
|
| 24178 |
+
{
|
| 24179 |
+
"epoch": 0.7666666666666667,
|
| 24180 |
+
"grad_norm": 2.1151397228240967,
|
| 24181 |
+
"learning_rate": 4.67706013363029e-05,
|
| 24182 |
+
"loss": 1.2452,
|
| 24183 |
+
"step": 3450
|
| 24184 |
+
},
|
| 24185 |
+
{
|
| 24186 |
+
"epoch": 0.7668888888888888,
|
| 24187 |
+
"grad_norm": 0.04698283597826958,
|
| 24188 |
+
"learning_rate": 4.6726057906458796e-05,
|
| 24189 |
+
"loss": 0.0105,
|
| 24190 |
+
"step": 3451
|
| 24191 |
+
},
|
| 24192 |
+
{
|
| 24193 |
+
"epoch": 0.7671111111111111,
|
| 24194 |
+
"grad_norm": 0.806088387966156,
|
| 24195 |
+
"learning_rate": 4.66815144766147e-05,
|
| 24196 |
+
"loss": 2.1148,
|
| 24197 |
+
"step": 3452
|
| 24198 |
+
},
|
| 24199 |
+
{
|
| 24200 |
+
"epoch": 0.7673333333333333,
|
| 24201 |
+
"grad_norm": 0.9896338582038879,
|
| 24202 |
+
"learning_rate": 4.6636971046770606e-05,
|
| 24203 |
+
"loss": 2.4891,
|
| 24204 |
+
"step": 3453
|
| 24205 |
+
},
|
| 24206 |
+
{
|
| 24207 |
+
"epoch": 0.7675555555555555,
|
| 24208 |
+
"grad_norm": 0.6359859704971313,
|
| 24209 |
+
"learning_rate": 4.6592427616926505e-05,
|
| 24210 |
+
"loss": 0.8817,
|
| 24211 |
+
"step": 3454
|
| 24212 |
+
},
|
| 24213 |
+
{
|
| 24214 |
+
"epoch": 0.7677777777777778,
|
| 24215 |
+
"grad_norm": 0.6366167068481445,
|
| 24216 |
+
"learning_rate": 4.654788418708241e-05,
|
| 24217 |
+
"loss": 1.1016,
|
| 24218 |
+
"step": 3455
|
| 24219 |
+
},
|
| 24220 |
+
{
|
| 24221 |
+
"epoch": 0.768,
|
| 24222 |
+
"grad_norm": 0.6625463366508484,
|
| 24223 |
+
"learning_rate": 4.650334075723831e-05,
|
| 24224 |
+
"loss": 1.2164,
|
| 24225 |
+
"step": 3456
|
| 24226 |
+
},
|
| 24227 |
+
{
|
| 24228 |
+
"epoch": 0.7682222222222223,
|
| 24229 |
+
"grad_norm": 0.6121510863304138,
|
| 24230 |
+
"learning_rate": 4.645879732739421e-05,
|
| 24231 |
+
"loss": 0.8625,
|
| 24232 |
+
"step": 3457
|
| 24233 |
+
},
|
| 24234 |
+
{
|
| 24235 |
+
"epoch": 0.7684444444444445,
|
| 24236 |
+
"grad_norm": 1.0105525255203247,
|
| 24237 |
+
"learning_rate": 4.641425389755011e-05,
|
| 24238 |
+
"loss": 2.7758,
|
| 24239 |
+
"step": 3458
|
| 24240 |
+
},
|
| 24241 |
+
{
|
| 24242 |
+
"epoch": 0.7686666666666667,
|
| 24243 |
+
"grad_norm": 0.8283724188804626,
|
| 24244 |
+
"learning_rate": 4.636971046770602e-05,
|
| 24245 |
+
"loss": 2.0447,
|
| 24246 |
+
"step": 3459
|
| 24247 |
+
},
|
| 24248 |
+
{
|
| 24249 |
+
"epoch": 0.7688888888888888,
|
| 24250 |
+
"grad_norm": 0.06813201308250427,
|
| 24251 |
+
"learning_rate": 4.632516703786192e-05,
|
| 24252 |
+
"loss": 0.0111,
|
| 24253 |
+
"step": 3460
|
| 24254 |
+
},
|
| 24255 |
+
{
|
| 24256 |
+
"epoch": 0.7691111111111111,
|
| 24257 |
+
"grad_norm": 0.0665576308965683,
|
| 24258 |
+
"learning_rate": 4.628062360801782e-05,
|
| 24259 |
+
"loss": 0.011,
|
| 24260 |
+
"step": 3461
|
| 24261 |
+
},
|
| 24262 |
+
{
|
| 24263 |
+
"epoch": 0.7693333333333333,
|
| 24264 |
+
"grad_norm": 0.06463496387004852,
|
| 24265 |
+
"learning_rate": 4.623608017817373e-05,
|
| 24266 |
+
"loss": 0.0111,
|
| 24267 |
+
"step": 3462
|
| 24268 |
+
},
|
| 24269 |
+
{
|
| 24270 |
+
"epoch": 0.7695555555555555,
|
| 24271 |
+
"grad_norm": 0.8166987895965576,
|
| 24272 |
+
"learning_rate": 4.619153674832962e-05,
|
| 24273 |
+
"loss": 2.0366,
|
| 24274 |
+
"step": 3463
|
| 24275 |
+
},
|
| 24276 |
+
{
|
| 24277 |
+
"epoch": 0.7697777777777778,
|
| 24278 |
+
"grad_norm": 0.9549795985221863,
|
| 24279 |
+
"learning_rate": 4.6146993318485525e-05,
|
| 24280 |
+
"loss": 1.8478,
|
| 24281 |
+
"step": 3464
|
| 24282 |
+
},
|
| 24283 |
+
{
|
| 24284 |
+
"epoch": 0.77,
|
| 24285 |
+
"grad_norm": 0.8335583209991455,
|
| 24286 |
+
"learning_rate": 4.610244988864143e-05,
|
| 24287 |
+
"loss": 1.8226,
|
| 24288 |
+
"step": 3465
|
| 24289 |
+
},
|
| 24290 |
+
{
|
| 24291 |
+
"epoch": 0.7702222222222223,
|
| 24292 |
+
"grad_norm": 0.9823237657546997,
|
| 24293 |
+
"learning_rate": 4.605790645879733e-05,
|
| 24294 |
+
"loss": 2.114,
|
| 24295 |
+
"step": 3466
|
| 24296 |
+
},
|
| 24297 |
+
{
|
| 24298 |
+
"epoch": 0.7704444444444445,
|
| 24299 |
+
"grad_norm": 0.9316264986991882,
|
| 24300 |
+
"learning_rate": 4.6013363028953234e-05,
|
| 24301 |
+
"loss": 2.0765,
|
| 24302 |
+
"step": 3467
|
| 24303 |
+
},
|
| 24304 |
+
{
|
| 24305 |
+
"epoch": 0.7706666666666667,
|
| 24306 |
+
"grad_norm": 0.8862332701683044,
|
| 24307 |
+
"learning_rate": 4.596881959910914e-05,
|
| 24308 |
+
"loss": 1.8899,
|
| 24309 |
+
"step": 3468
|
| 24310 |
+
},
|
| 24311 |
+
{
|
| 24312 |
+
"epoch": 0.7708888888888888,
|
| 24313 |
+
"grad_norm": 0.9615729451179504,
|
| 24314 |
+
"learning_rate": 4.592427616926504e-05,
|
| 24315 |
+
"loss": 1.9076,
|
| 24316 |
+
"step": 3469
|
| 24317 |
+
},
|
| 24318 |
+
{
|
| 24319 |
+
"epoch": 0.7711111111111111,
|
| 24320 |
+
"grad_norm": 0.107745461165905,
|
| 24321 |
+
"learning_rate": 4.5879732739420936e-05,
|
| 24322 |
+
"loss": 0.0184,
|
| 24323 |
+
"step": 3470
|
| 24324 |
+
},
|
| 24325 |
+
{
|
| 24326 |
+
"epoch": 0.7713333333333333,
|
| 24327 |
+
"grad_norm": 0.9588910341262817,
|
| 24328 |
+
"learning_rate": 4.5835189309576835e-05,
|
| 24329 |
+
"loss": 1.7689,
|
| 24330 |
+
"step": 3471
|
| 24331 |
+
},
|
| 24332 |
+
{
|
| 24333 |
+
"epoch": 0.7715555555555556,
|
| 24334 |
+
"grad_norm": 0.8808805346488953,
|
| 24335 |
+
"learning_rate": 4.579064587973274e-05,
|
| 24336 |
+
"loss": 2.204,
|
| 24337 |
+
"step": 3472
|
| 24338 |
+
},
|
| 24339 |
+
{
|
| 24340 |
+
"epoch": 0.7717777777777778,
|
| 24341 |
+
"grad_norm": 1.1614326238632202,
|
| 24342 |
+
"learning_rate": 4.5746102449888646e-05,
|
| 24343 |
+
"loss": 1.9956,
|
| 24344 |
+
"step": 3473
|
| 24345 |
+
},
|
| 24346 |
+
{
|
| 24347 |
+
"epoch": 0.772,
|
| 24348 |
+
"grad_norm": 0.8884471654891968,
|
| 24349 |
+
"learning_rate": 4.5701559020044544e-05,
|
| 24350 |
+
"loss": 1.875,
|
| 24351 |
+
"step": 3474
|
| 24352 |
+
},
|
| 24353 |
+
{
|
| 24354 |
+
"epoch": 0.7722222222222223,
|
| 24355 |
+
"grad_norm": 0.9541723728179932,
|
| 24356 |
+
"learning_rate": 4.565701559020045e-05,
|
| 24357 |
+
"loss": 1.8806,
|
| 24358 |
+
"step": 3475
|
| 24359 |
+
},
|
| 24360 |
+
{
|
| 24361 |
+
"epoch": 0.7724444444444445,
|
| 24362 |
+
"grad_norm": 0.07305742055177689,
|
| 24363 |
+
"learning_rate": 4.561247216035635e-05,
|
| 24364 |
+
"loss": 0.0169,
|
| 24365 |
+
"step": 3476
|
| 24366 |
+
},
|
| 24367 |
+
{
|
| 24368 |
+
"epoch": 0.7726666666666666,
|
| 24369 |
+
"grad_norm": 0.6821660399436951,
|
| 24370 |
+
"learning_rate": 4.556792873051225e-05,
|
| 24371 |
+
"loss": 1.079,
|
| 24372 |
+
"step": 3477
|
| 24373 |
+
},
|
| 24374 |
+
{
|
| 24375 |
+
"epoch": 0.7728888888888888,
|
| 24376 |
+
"grad_norm": 0.9830121994018555,
|
| 24377 |
+
"learning_rate": 4.552338530066815e-05,
|
| 24378 |
+
"loss": 1.7285,
|
| 24379 |
+
"step": 3478
|
| 24380 |
+
},
|
| 24381 |
+
{
|
| 24382 |
+
"epoch": 0.7731111111111111,
|
| 24383 |
+
"grad_norm": 0.6831437945365906,
|
| 24384 |
+
"learning_rate": 4.547884187082406e-05,
|
| 24385 |
+
"loss": 0.8884,
|
| 24386 |
+
"step": 3479
|
| 24387 |
+
},
|
| 24388 |
+
{
|
| 24389 |
+
"epoch": 0.7733333333333333,
|
| 24390 |
+
"grad_norm": 1.0051524639129639,
|
| 24391 |
+
"learning_rate": 4.5434298440979956e-05,
|
| 24392 |
+
"loss": 1.8973,
|
| 24393 |
+
"step": 3480
|
| 24394 |
+
},
|
| 24395 |
+
{
|
| 24396 |
+
"epoch": 0.7735555555555556,
|
| 24397 |
+
"grad_norm": 0.9727129340171814,
|
| 24398 |
+
"learning_rate": 4.538975501113586e-05,
|
| 24399 |
+
"loss": 2.042,
|
| 24400 |
+
"step": 3481
|
| 24401 |
+
},
|
| 24402 |
+
{
|
| 24403 |
+
"epoch": 0.7737777777777778,
|
| 24404 |
+
"grad_norm": 1.0296839475631714,
|
| 24405 |
+
"learning_rate": 4.534521158129176e-05,
|
| 24406 |
+
"loss": 1.7367,
|
| 24407 |
+
"step": 3482
|
| 24408 |
+
},
|
| 24409 |
+
{
|
| 24410 |
+
"epoch": 0.774,
|
| 24411 |
+
"grad_norm": 0.9972522258758545,
|
| 24412 |
+
"learning_rate": 4.530066815144766e-05,
|
| 24413 |
+
"loss": 1.8867,
|
| 24414 |
+
"step": 3483
|
| 24415 |
+
},
|
| 24416 |
+
{
|
| 24417 |
+
"epoch": 0.7742222222222223,
|
| 24418 |
+
"grad_norm": 1.0227113962173462,
|
| 24419 |
+
"learning_rate": 4.5256124721603564e-05,
|
| 24420 |
+
"loss": 1.8279,
|
| 24421 |
+
"step": 3484
|
| 24422 |
+
},
|
| 24423 |
+
{
|
| 24424 |
+
"epoch": 0.7744444444444445,
|
| 24425 |
+
"grad_norm": 1.061448335647583,
|
| 24426 |
+
"learning_rate": 4.521158129175947e-05,
|
| 24427 |
+
"loss": 1.9245,
|
| 24428 |
+
"step": 3485
|
| 24429 |
+
},
|
| 24430 |
+
{
|
| 24431 |
+
"epoch": 0.7746666666666666,
|
| 24432 |
+
"grad_norm": 0.6845740675926208,
|
| 24433 |
+
"learning_rate": 4.516703786191537e-05,
|
| 24434 |
+
"loss": 0.9532,
|
| 24435 |
+
"step": 3486
|
| 24436 |
+
},
|
| 24437 |
+
{
|
| 24438 |
+
"epoch": 0.7748888888888888,
|
| 24439 |
+
"grad_norm": 1.010504961013794,
|
| 24440 |
+
"learning_rate": 4.512249443207127e-05,
|
| 24441 |
+
"loss": 1.6469,
|
| 24442 |
+
"step": 3487
|
| 24443 |
+
},
|
| 24444 |
+
{
|
| 24445 |
+
"epoch": 0.7751111111111111,
|
| 24446 |
+
"grad_norm": 1.15483820438385,
|
| 24447 |
+
"learning_rate": 4.507795100222718e-05,
|
| 24448 |
+
"loss": 1.735,
|
| 24449 |
+
"step": 3488
|
| 24450 |
+
},
|
| 24451 |
+
{
|
| 24452 |
+
"epoch": 0.7753333333333333,
|
| 24453 |
+
"grad_norm": 0.06943599879741669,
|
| 24454 |
+
"learning_rate": 4.503340757238308e-05,
|
| 24455 |
+
"loss": 0.0192,
|
| 24456 |
+
"step": 3489
|
| 24457 |
+
},
|
| 24458 |
+
{
|
| 24459 |
+
"epoch": 0.7755555555555556,
|
| 24460 |
+
"grad_norm": 0.06758453696966171,
|
| 24461 |
+
"learning_rate": 4.4988864142538976e-05,
|
| 24462 |
+
"loss": 0.0185,
|
| 24463 |
+
"step": 3490
|
| 24464 |
+
},
|
| 24465 |
+
{
|
| 24466 |
+
"epoch": 0.7757777777777778,
|
| 24467 |
+
"grad_norm": 0.06915237754583359,
|
| 24468 |
+
"learning_rate": 4.494432071269488e-05,
|
| 24469 |
+
"loss": 0.0189,
|
| 24470 |
+
"step": 3491
|
| 24471 |
+
},
|
| 24472 |
+
{
|
| 24473 |
+
"epoch": 0.776,
|
| 24474 |
+
"grad_norm": 0.7292212843894958,
|
| 24475 |
+
"learning_rate": 4.489977728285078e-05,
|
| 24476 |
+
"loss": 0.8598,
|
| 24477 |
+
"step": 3492
|
| 24478 |
+
},
|
| 24479 |
+
{
|
| 24480 |
+
"epoch": 0.7762222222222223,
|
| 24481 |
+
"grad_norm": 0.9773833751678467,
|
| 24482 |
+
"learning_rate": 4.4855233853006685e-05,
|
| 24483 |
+
"loss": 1.5498,
|
| 24484 |
+
"step": 3493
|
| 24485 |
+
},
|
| 24486 |
+
{
|
| 24487 |
+
"epoch": 0.7764444444444445,
|
| 24488 |
+
"grad_norm": 1.0763559341430664,
|
| 24489 |
+
"learning_rate": 4.481069042316259e-05,
|
| 24490 |
+
"loss": 1.6527,
|
| 24491 |
+
"step": 3494
|
| 24492 |
+
},
|
| 24493 |
+
{
|
| 24494 |
+
"epoch": 0.7766666666666666,
|
| 24495 |
+
"grad_norm": 1.0425339937210083,
|
| 24496 |
+
"learning_rate": 4.476614699331849e-05,
|
| 24497 |
+
"loss": 1.272,
|
| 24498 |
+
"step": 3495
|
| 24499 |
+
},
|
| 24500 |
+
{
|
| 24501 |
+
"epoch": 0.7768888888888889,
|
| 24502 |
+
"grad_norm": 1.1225720643997192,
|
| 24503 |
+
"learning_rate": 4.472160356347439e-05,
|
| 24504 |
+
"loss": 1.3202,
|
| 24505 |
+
"step": 3496
|
| 24506 |
+
},
|
| 24507 |
+
{
|
| 24508 |
+
"epoch": 0.7771111111111111,
|
| 24509 |
+
"grad_norm": 1.2557756900787354,
|
| 24510 |
+
"learning_rate": 4.467706013363029e-05,
|
| 24511 |
+
"loss": 1.378,
|
| 24512 |
+
"step": 3497
|
| 24513 |
+
},
|
| 24514 |
+
{
|
| 24515 |
+
"epoch": 0.7773333333333333,
|
| 24516 |
+
"grad_norm": 0.17433112859725952,
|
| 24517 |
+
"learning_rate": 4.463251670378619e-05,
|
| 24518 |
+
"loss": 0.038,
|
| 24519 |
+
"step": 3498
|
| 24520 |
+
},
|
| 24521 |
+
{
|
| 24522 |
+
"epoch": 0.7775555555555556,
|
| 24523 |
+
"grad_norm": 1.008841633796692,
|
| 24524 |
+
"learning_rate": 4.45879732739421e-05,
|
| 24525 |
+
"loss": 0.8367,
|
| 24526 |
+
"step": 3499
|
| 24527 |
+
},
|
| 24528 |
+
{
|
| 24529 |
+
"epoch": 0.7777777777777778,
|
| 24530 |
+
"grad_norm": 0.9712222814559937,
|
| 24531 |
+
"learning_rate": 4.4543429844098e-05,
|
| 24532 |
+
"loss": 0.6577,
|
| 24533 |
+
"step": 3500
|
| 24534 |
}
|
| 24535 |
],
|
| 24536 |
"logging_steps": 1,
|
|
|
|
| 24550 |
"attributes": {}
|
| 24551 |
}
|
| 24552 |
},
|
| 24553 |
+
"total_flos": 3.773762935084032e+16,
|
| 24554 |
"train_batch_size": 1,
|
| 24555 |
"trial_name": null,
|
| 24556 |
"trial_params": null
|