Training in progress, step 1437, 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 30026872
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:062e2f8ce317c796405adeab0219eb61c16b8226c87287e7bac473baf87e5ee9
|
| 3 |
size 30026872
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 15611732
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5d9c021df4ffdfe77f9733a45841065bdb97497a128c9d10929c84c0bf2accd4
|
| 3 |
size 15611732
|
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 sha256:42ff6c0ef3bc465d439974c3c2c6f55ed96cc7913337d0d0ac9aca4d6bc6024a
|
| 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 sha256:44982f8089d8acc9695fc1bb4a2c4c9d4026da90ed4ed55893f2bfa0b6ecf12c
|
| 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": 360,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7599,6 +7599,2505 @@
|
|
| 7599 |
"eval_samples_per_second": 48.789,
|
| 7600 |
"eval_steps_per_second": 24.396,
|
| 7601 |
"step": 1080
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7602 |
}
|
| 7603 |
],
|
| 7604 |
"logging_steps": 1,
|
|
@@ -7613,12 +10112,12 @@
|
|
| 7613 |
"should_evaluate": false,
|
| 7614 |
"should_log": false,
|
| 7615 |
"should_save": true,
|
| 7616 |
-
"should_training_stop":
|
| 7617 |
},
|
| 7618 |
"attributes": {}
|
| 7619 |
}
|
| 7620 |
},
|
| 7621 |
-
"total_flos":
|
| 7622 |
"train_batch_size": 2,
|
| 7623 |
"trial_name": null,
|
| 7624 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.030573329716446728,
|
| 5 |
"eval_steps": 360,
|
| 6 |
+
"global_step": 1437,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7599 |
"eval_samples_per_second": 48.789,
|
| 7600 |
"eval_steps_per_second": 24.396,
|
| 7601 |
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.02299914364890669,
|
| 7605 |
+
"grad_norm": 0.41674280166625977,
|
| 7606 |
+
"learning_rate": 2.917266421714292e-05,
|
| 7607 |
+
"loss": 1.8273,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.023020419452467196,
|
| 7612 |
+
"grad_norm": 0.39788734912872314,
|
| 7613 |
+
"learning_rate": 2.9017421092606924e-05,
|
| 7614 |
+
"loss": 1.8648,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.0230416952560277,
|
| 7619 |
+
"grad_norm": 0.4401637315750122,
|
| 7620 |
+
"learning_rate": 2.8862522003681192e-05,
|
| 7621 |
+
"loss": 2.0145,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.023062971059588207,
|
| 7626 |
+
"grad_norm": 0.4017985463142395,
|
| 7627 |
+
"learning_rate": 2.870796770112466e-05,
|
| 7628 |
+
"loss": 1.8623,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.02308424686314871,
|
| 7633 |
+
"grad_norm": 0.41117116808891296,
|
| 7634 |
+
"learning_rate": 2.8553758934025097e-05,
|
| 7635 |
+
"loss": 1.8354,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.02310552266670922,
|
| 7640 |
+
"grad_norm": 0.40373316407203674,
|
| 7641 |
+
"learning_rate": 2.839989644979555e-05,
|
| 7642 |
+
"loss": 2.104,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.023126798470269723,
|
| 7647 |
+
"grad_norm": 0.35661011934280396,
|
| 7648 |
+
"learning_rate": 2.8246380994170752e-05,
|
| 7649 |
+
"loss": 1.9508,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.02314807427383023,
|
| 7654 |
+
"grad_norm": 0.3797127902507782,
|
| 7655 |
+
"learning_rate": 2.809321331120346e-05,
|
| 7656 |
+
"loss": 1.8893,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.023169350077390734,
|
| 7661 |
+
"grad_norm": 0.37274572253227234,
|
| 7662 |
+
"learning_rate": 2.7940394143260872e-05,
|
| 7663 |
+
"loss": 1.8823,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.023190625880951242,
|
| 7668 |
+
"grad_norm": 0.37364649772644043,
|
| 7669 |
+
"learning_rate": 2.7787924231021034e-05,
|
| 7670 |
+
"loss": 1.995,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.023211901684511746,
|
| 7675 |
+
"grad_norm": 0.4065852761268616,
|
| 7676 |
+
"learning_rate": 2.7635804313469138e-05,
|
| 7677 |
+
"loss": 1.8606,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.023233177488072253,
|
| 7682 |
+
"grad_norm": 0.36767449975013733,
|
| 7683 |
+
"learning_rate": 2.7484035127894157e-05,
|
| 7684 |
+
"loss": 1.8317,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.023254453291632757,
|
| 7689 |
+
"grad_norm": 0.3782156705856323,
|
| 7690 |
+
"learning_rate": 2.733261740988503e-05,
|
| 7691 |
+
"loss": 1.8395,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.023275729095193265,
|
| 7696 |
+
"grad_norm": 0.4104126989841461,
|
| 7697 |
+
"learning_rate": 2.718155189332733e-05,
|
| 7698 |
+
"loss": 1.9346,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.02329700489875377,
|
| 7703 |
+
"grad_norm": 0.3796551525592804,
|
| 7704 |
+
"learning_rate": 2.7030839310399546e-05,
|
| 7705 |
+
"loss": 1.9242,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.023318280702314276,
|
| 7710 |
+
"grad_norm": 0.42018982768058777,
|
| 7711 |
+
"learning_rate": 2.6880480391569528e-05,
|
| 7712 |
+
"loss": 1.7642,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.02333955650587478,
|
| 7717 |
+
"grad_norm": 0.387173593044281,
|
| 7718 |
+
"learning_rate": 2.67304758655911e-05,
|
| 7719 |
+
"loss": 1.8935,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.023360832309435288,
|
| 7724 |
+
"grad_norm": 0.38176941871643066,
|
| 7725 |
+
"learning_rate": 2.6580826459500353e-05,
|
| 7726 |
+
"loss": 1.9855,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.023382108112995792,
|
| 7731 |
+
"grad_norm": 0.3577720522880554,
|
| 7732 |
+
"learning_rate": 2.6431532898612277e-05,
|
| 7733 |
+
"loss": 1.6802,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.0234033839165563,
|
| 7738 |
+
"grad_norm": 0.40479302406311035,
|
| 7739 |
+
"learning_rate": 2.6282595906517036e-05,
|
| 7740 |
+
"loss": 2.0189,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.023424659720116803,
|
| 7745 |
+
"grad_norm": 0.43480184674263,
|
| 7746 |
+
"learning_rate": 2.613401620507674e-05,
|
| 7747 |
+
"loss": 2.1037,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.02344593552367731,
|
| 7752 |
+
"grad_norm": 0.4054833948612213,
|
| 7753 |
+
"learning_rate": 2.5985794514421647e-05,
|
| 7754 |
+
"loss": 1.8234,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.023467211327237815,
|
| 7759 |
+
"grad_norm": 0.3937501311302185,
|
| 7760 |
+
"learning_rate": 2.5837931552946937e-05,
|
| 7761 |
+
"loss": 1.8975,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.023488487130798322,
|
| 7766 |
+
"grad_norm": 0.41668424010276794,
|
| 7767 |
+
"learning_rate": 2.5690428037309078e-05,
|
| 7768 |
+
"loss": 1.8508,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.023509762934358826,
|
| 7773 |
+
"grad_norm": 0.3966442942619324,
|
| 7774 |
+
"learning_rate": 2.554328468242242e-05,
|
| 7775 |
+
"loss": 1.8631,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.023531038737919334,
|
| 7780 |
+
"grad_norm": 0.39415624737739563,
|
| 7781 |
+
"learning_rate": 2.539650220145564e-05,
|
| 7782 |
+
"loss": 1.9255,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.023552314541479838,
|
| 7787 |
+
"grad_norm": 0.39599108695983887,
|
| 7788 |
+
"learning_rate": 2.5250081305828332e-05,
|
| 7789 |
+
"loss": 1.9263,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.023573590345040345,
|
| 7794 |
+
"grad_norm": 0.3804737329483032,
|
| 7795 |
+
"learning_rate": 2.5104022705207654e-05,
|
| 7796 |
+
"loss": 1.789,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.02359486614860085,
|
| 7801 |
+
"grad_norm": 0.404774010181427,
|
| 7802 |
+
"learning_rate": 2.4958327107504763e-05,
|
| 7803 |
+
"loss": 1.8815,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.023616141952161357,
|
| 7808 |
+
"grad_norm": 0.4321078956127167,
|
| 7809 |
+
"learning_rate": 2.4812995218871437e-05,
|
| 7810 |
+
"loss": 1.7646,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.02363741775572186,
|
| 7815 |
+
"grad_norm": 0.42398884892463684,
|
| 7816 |
+
"learning_rate": 2.4668027743696654e-05,
|
| 7817 |
+
"loss": 1.9746,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.02365869355928237,
|
| 7822 |
+
"grad_norm": 0.4005934000015259,
|
| 7823 |
+
"learning_rate": 2.4523425384603094e-05,
|
| 7824 |
+
"loss": 1.7124,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.023679969362842872,
|
| 7829 |
+
"grad_norm": 0.38693100214004517,
|
| 7830 |
+
"learning_rate": 2.43791888424439e-05,
|
| 7831 |
+
"loss": 1.8516,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.02370124516640338,
|
| 7836 |
+
"grad_norm": 0.3936989903450012,
|
| 7837 |
+
"learning_rate": 2.4235318816299167e-05,
|
| 7838 |
+
"loss": 1.8451,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.023722520969963884,
|
| 7843 |
+
"grad_norm": 0.3978916108608246,
|
| 7844 |
+
"learning_rate": 2.4091816003472522e-05,
|
| 7845 |
+
"loss": 1.706,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.02374379677352439,
|
| 7850 |
+
"grad_norm": 0.3996455669403076,
|
| 7851 |
+
"learning_rate": 2.3948681099487856e-05,
|
| 7852 |
+
"loss": 1.9042,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.023765072577084895,
|
| 7857 |
+
"grad_norm": 0.36959224939346313,
|
| 7858 |
+
"learning_rate": 2.3805914798085848e-05,
|
| 7859 |
+
"loss": 1.897,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.023786348380645403,
|
| 7864 |
+
"grad_norm": 0.41218438744544983,
|
| 7865 |
+
"learning_rate": 2.3663517791220667e-05,
|
| 7866 |
+
"loss": 1.8196,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.023807624184205907,
|
| 7871 |
+
"grad_norm": 0.40864235162734985,
|
| 7872 |
+
"learning_rate": 2.352149076905663e-05,
|
| 7873 |
+
"loss": 1.818,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.023828899987766414,
|
| 7878 |
+
"grad_norm": 0.36998435854911804,
|
| 7879 |
+
"learning_rate": 2.3379834419964786e-05,
|
| 7880 |
+
"loss": 1.8445,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.02385017579132692,
|
| 7885 |
+
"grad_norm": 0.38013777136802673,
|
| 7886 |
+
"learning_rate": 2.3238549430519663e-05,
|
| 7887 |
+
"loss": 1.8842,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.023871451594887426,
|
| 7892 |
+
"grad_norm": 0.37000346183776855,
|
| 7893 |
+
"learning_rate": 2.3097636485495832e-05,
|
| 7894 |
+
"loss": 2.0156,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.02389272739844793,
|
| 7899 |
+
"grad_norm": 0.35548990964889526,
|
| 7900 |
+
"learning_rate": 2.2957096267864676e-05,
|
| 7901 |
+
"loss": 1.8225,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.023914003202008437,
|
| 7906 |
+
"grad_norm": 0.4264899492263794,
|
| 7907 |
+
"learning_rate": 2.281692945879109e-05,
|
| 7908 |
+
"loss": 1.7632,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.02393527900556894,
|
| 7913 |
+
"grad_norm": 0.36815860867500305,
|
| 7914 |
+
"learning_rate": 2.2677136737630123e-05,
|
| 7915 |
+
"loss": 1.8928,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.02395655480912945,
|
| 7920 |
+
"grad_norm": 0.3817468583583832,
|
| 7921 |
+
"learning_rate": 2.2537718781923756e-05,
|
| 7922 |
+
"loss": 1.765,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.023977830612689953,
|
| 7927 |
+
"grad_norm": 0.38322868943214417,
|
| 7928 |
+
"learning_rate": 2.2398676267397457e-05,
|
| 7929 |
+
"loss": 1.6741,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.02399910641625046,
|
| 7934 |
+
"grad_norm": 0.4257258176803589,
|
| 7935 |
+
"learning_rate": 2.2260009867957167e-05,
|
| 7936 |
+
"loss": 1.9132,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.024020382219810964,
|
| 7941 |
+
"grad_norm": 0.3691489100456238,
|
| 7942 |
+
"learning_rate": 2.212172025568574e-05,
|
| 7943 |
+
"loss": 1.8471,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.024041658023371472,
|
| 7948 |
+
"grad_norm": 0.37367841601371765,
|
| 7949 |
+
"learning_rate": 2.1983808100839943e-05,
|
| 7950 |
+
"loss": 1.8165,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.024062933826931976,
|
| 7955 |
+
"grad_norm": 0.3934177756309509,
|
| 7956 |
+
"learning_rate": 2.184627407184705e-05,
|
| 7957 |
+
"loss": 1.7747,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.024084209630492483,
|
| 7962 |
+
"grad_norm": 0.4731823205947876,
|
| 7963 |
+
"learning_rate": 2.1709118835301688e-05,
|
| 7964 |
+
"loss": 1.8267,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.024105485434052987,
|
| 7969 |
+
"grad_norm": 0.38958320021629333,
|
| 7970 |
+
"learning_rate": 2.1572343055962485e-05,
|
| 7971 |
+
"loss": 1.9303,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.02412676123761349,
|
| 7976 |
+
"grad_norm": 0.3958386480808258,
|
| 7977 |
+
"learning_rate": 2.1435947396749024e-05,
|
| 7978 |
+
"loss": 1.897,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.024148037041174,
|
| 7983 |
+
"grad_norm": 0.41613322496414185,
|
| 7984 |
+
"learning_rate": 2.1299932518738498e-05,
|
| 7985 |
+
"loss": 1.7671,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.024169312844734503,
|
| 7990 |
+
"grad_norm": 0.4169005751609802,
|
| 7991 |
+
"learning_rate": 2.1164299081162498e-05,
|
| 7992 |
+
"loss": 2.1049,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.02419058864829501,
|
| 7997 |
+
"grad_norm": 0.373477041721344,
|
| 7998 |
+
"learning_rate": 2.1029047741403975e-05,
|
| 7999 |
+
"loss": 1.7534,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.024211864451855514,
|
| 8004 |
+
"grad_norm": 0.37139496207237244,
|
| 8005 |
+
"learning_rate": 2.089417915499381e-05,
|
| 8006 |
+
"loss": 1.894,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.024233140255416022,
|
| 8011 |
+
"grad_norm": 0.3943324685096741,
|
| 8012 |
+
"learning_rate": 2.075969397560785e-05,
|
| 8013 |
+
"loss": 2.0342,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.024254416058976526,
|
| 8018 |
+
"grad_norm": 0.41412708163261414,
|
| 8019 |
+
"learning_rate": 2.062559285506368e-05,
|
| 8020 |
+
"loss": 1.9508,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.024275691862537033,
|
| 8025 |
+
"grad_norm": 0.3780665397644043,
|
| 8026 |
+
"learning_rate": 2.049187644331738e-05,
|
| 8027 |
+
"loss": 1.9173,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.024296967666097537,
|
| 8032 |
+
"grad_norm": 0.42113977670669556,
|
| 8033 |
+
"learning_rate": 2.0358545388460516e-05,
|
| 8034 |
+
"loss": 1.7598,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.024318243469658045,
|
| 8039 |
+
"grad_norm": 0.4007102847099304,
|
| 8040 |
+
"learning_rate": 2.022560033671683e-05,
|
| 8041 |
+
"loss": 2.0653,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.02433951927321855,
|
| 8046 |
+
"grad_norm": 0.36581388115882874,
|
| 8047 |
+
"learning_rate": 2.0093041932439226e-05,
|
| 8048 |
+
"loss": 1.7624,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.024360795076779056,
|
| 8053 |
+
"grad_norm": 0.41151005029678345,
|
| 8054 |
+
"learning_rate": 1.996087081810666e-05,
|
| 8055 |
+
"loss": 1.9184,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.02438207088033956,
|
| 8060 |
+
"grad_norm": 0.39749282598495483,
|
| 8061 |
+
"learning_rate": 1.9829087634320952e-05,
|
| 8062 |
+
"loss": 1.7034,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.024403346683900068,
|
| 8067 |
+
"grad_norm": 0.37557074427604675,
|
| 8068 |
+
"learning_rate": 1.9697693019803776e-05,
|
| 8069 |
+
"loss": 1.8761,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.024424622487460572,
|
| 8074 |
+
"grad_norm": 0.3893011510372162,
|
| 8075 |
+
"learning_rate": 1.9566687611393373e-05,
|
| 8076 |
+
"loss": 1.8786,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.02444589829102108,
|
| 8081 |
+
"grad_norm": 0.3731049597263336,
|
| 8082 |
+
"learning_rate": 1.943607204404173e-05,
|
| 8083 |
+
"loss": 1.8652,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.024467174094581583,
|
| 8088 |
+
"grad_norm": 0.3560284674167633,
|
| 8089 |
+
"learning_rate": 1.9305846950811312e-05,
|
| 8090 |
+
"loss": 2.0102,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.02448844989814209,
|
| 8095 |
+
"grad_norm": 0.392687052488327,
|
| 8096 |
+
"learning_rate": 1.9176012962872015e-05,
|
| 8097 |
+
"loss": 1.8759,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.024509725701702595,
|
| 8102 |
+
"grad_norm": 0.3997262418270111,
|
| 8103 |
+
"learning_rate": 1.90465707094982e-05,
|
| 8104 |
+
"loss": 1.9648,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.024531001505263102,
|
| 8109 |
+
"grad_norm": 0.3895493149757385,
|
| 8110 |
+
"learning_rate": 1.891752081806557e-05,
|
| 8111 |
+
"loss": 1.8322,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.024552277308823606,
|
| 8116 |
+
"grad_norm": 0.41220536828041077,
|
| 8117 |
+
"learning_rate": 1.8788863914048105e-05,
|
| 8118 |
+
"loss": 1.9151,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.024573553112384114,
|
| 8123 |
+
"grad_norm": 0.4090578854084015,
|
| 8124 |
+
"learning_rate": 1.8660600621015113e-05,
|
| 8125 |
+
"loss": 1.7724,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.024594828915944618,
|
| 8130 |
+
"grad_norm": 0.4076254367828369,
|
| 8131 |
+
"learning_rate": 1.8532731560628148e-05,
|
| 8132 |
+
"loss": 2.2509,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.024616104719505125,
|
| 8137 |
+
"grad_norm": 0.41099920868873596,
|
| 8138 |
+
"learning_rate": 1.8405257352638028e-05,
|
| 8139 |
+
"loss": 1.7472,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.02463738052306563,
|
| 8144 |
+
"grad_norm": 0.4149285852909088,
|
| 8145 |
+
"learning_rate": 1.827817861488178e-05,
|
| 8146 |
+
"loss": 1.831,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.024658656326626137,
|
| 8151 |
+
"grad_norm": 0.3463163375854492,
|
| 8152 |
+
"learning_rate": 1.8151495963279695e-05,
|
| 8153 |
+
"loss": 1.8381,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.02467993213018664,
|
| 8158 |
+
"grad_norm": 0.3854790925979614,
|
| 8159 |
+
"learning_rate": 1.8025210011832327e-05,
|
| 8160 |
+
"loss": 1.7098,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.02470120793374715,
|
| 8165 |
+
"grad_norm": 0.42400407791137695,
|
| 8166 |
+
"learning_rate": 1.789932137261756e-05,
|
| 8167 |
+
"loss": 1.838,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.024722483737307652,
|
| 8172 |
+
"grad_norm": 0.37702676653862,
|
| 8173 |
+
"learning_rate": 1.777383065578755e-05,
|
| 8174 |
+
"loss": 2.0821,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.02474375954086816,
|
| 8179 |
+
"grad_norm": 0.3950926959514618,
|
| 8180 |
+
"learning_rate": 1.764873846956585e-05,
|
| 8181 |
+
"loss": 1.8544,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.024765035344428664,
|
| 8186 |
+
"grad_norm": 0.41955599188804626,
|
| 8187 |
+
"learning_rate": 1.7524045420244416e-05,
|
| 8188 |
+
"loss": 2.0615,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.02478631114798917,
|
| 8193 |
+
"grad_norm": 0.38509541749954224,
|
| 8194 |
+
"learning_rate": 1.7399752112180645e-05,
|
| 8195 |
+
"loss": 1.8497,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.024807586951549675,
|
| 8200 |
+
"grad_norm": 0.3934166431427002,
|
| 8201 |
+
"learning_rate": 1.727585914779455e-05,
|
| 8202 |
+
"loss": 1.8509,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.024828862755110183,
|
| 8207 |
+
"grad_norm": 0.42457836866378784,
|
| 8208 |
+
"learning_rate": 1.715236712756575e-05,
|
| 8209 |
+
"loss": 2.2264,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.024850138558670687,
|
| 8214 |
+
"grad_norm": 0.3627967834472656,
|
| 8215 |
+
"learning_rate": 1.702927665003059e-05,
|
| 8216 |
+
"loss": 1.8636,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.024871414362231194,
|
| 8221 |
+
"grad_norm": 0.42242783308029175,
|
| 8222 |
+
"learning_rate": 1.690658831177917e-05,
|
| 8223 |
+
"loss": 1.8916,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.0248926901657917,
|
| 8228 |
+
"grad_norm": 0.3954100012779236,
|
| 8229 |
+
"learning_rate": 1.6784302707452614e-05,
|
| 8230 |
+
"loss": 1.8362,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.024913965969352206,
|
| 8235 |
+
"grad_norm": 0.3579208552837372,
|
| 8236 |
+
"learning_rate": 1.6662420429740044e-05,
|
| 8237 |
+
"loss": 1.874,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.02493524177291271,
|
| 8242 |
+
"grad_norm": 0.40131744742393494,
|
| 8243 |
+
"learning_rate": 1.6540942069375764e-05,
|
| 8244 |
+
"loss": 1.9105,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.024956517576473217,
|
| 8249 |
+
"grad_norm": 0.44038259983062744,
|
| 8250 |
+
"learning_rate": 1.6419868215136337e-05,
|
| 8251 |
+
"loss": 1.7521,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.02497779338003372,
|
| 8256 |
+
"grad_norm": 0.384378045797348,
|
| 8257 |
+
"learning_rate": 1.6299199453837877e-05,
|
| 8258 |
+
"loss": 1.8899,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.02499906918359423,
|
| 8263 |
+
"grad_norm": 0.3976133167743683,
|
| 8264 |
+
"learning_rate": 1.617893637033301e-05,
|
| 8265 |
+
"loss": 1.6707,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.025020344987154733,
|
| 8270 |
+
"grad_norm": 0.38938310742378235,
|
| 8271 |
+
"learning_rate": 1.6059079547508195e-05,
|
| 8272 |
+
"loss": 2.0859,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.02504162079071524,
|
| 8277 |
+
"grad_norm": 0.4013230502605438,
|
| 8278 |
+
"learning_rate": 1.5939629566280846e-05,
|
| 8279 |
+
"loss": 1.9707,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.025062896594275744,
|
| 8284 |
+
"grad_norm": 0.3887784779071808,
|
| 8285 |
+
"learning_rate": 1.582058700559652e-05,
|
| 8286 |
+
"loss": 1.8225,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.025084172397836252,
|
| 8291 |
+
"grad_norm": 0.380338191986084,
|
| 8292 |
+
"learning_rate": 1.570195244242605e-05,
|
| 8293 |
+
"loss": 1.8805,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.025105448201396756,
|
| 8298 |
+
"grad_norm": 0.385150671005249,
|
| 8299 |
+
"learning_rate": 1.5583726451762826e-05,
|
| 8300 |
+
"loss": 1.8983,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.025126724004957263,
|
| 8305 |
+
"grad_norm": 0.4519411325454712,
|
| 8306 |
+
"learning_rate": 1.546590960662e-05,
|
| 8307 |
+
"loss": 1.8334,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.025147999808517767,
|
| 8312 |
+
"grad_norm": 0.42050448060035706,
|
| 8313 |
+
"learning_rate": 1.5348502478027705e-05,
|
| 8314 |
+
"loss": 1.9352,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.025169275612078275,
|
| 8319 |
+
"grad_norm": 0.3789636194705963,
|
| 8320 |
+
"learning_rate": 1.5231505635030252e-05,
|
| 8321 |
+
"loss": 1.6573,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.02519055141563878,
|
| 8326 |
+
"grad_norm": 0.4297623932361603,
|
| 8327 |
+
"learning_rate": 1.5114919644683412e-05,
|
| 8328 |
+
"loss": 1.9876,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.025211827219199286,
|
| 8333 |
+
"grad_norm": 0.41687914729118347,
|
| 8334 |
+
"learning_rate": 1.4998745072051612e-05,
|
| 8335 |
+
"loss": 1.7652,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.02523310302275979,
|
| 8340 |
+
"grad_norm": 0.39616456627845764,
|
| 8341 |
+
"learning_rate": 1.488298248020531e-05,
|
| 8342 |
+
"loss": 1.9961,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.025254378826320298,
|
| 8347 |
+
"grad_norm": 0.38997429609298706,
|
| 8348 |
+
"learning_rate": 1.4767632430218081e-05,
|
| 8349 |
+
"loss": 1.8353,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.025275654629880802,
|
| 8354 |
+
"grad_norm": 0.428004652261734,
|
| 8355 |
+
"learning_rate": 1.4652695481164124e-05,
|
| 8356 |
+
"loss": 2.0361,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.02529693043344131,
|
| 8361 |
+
"grad_norm": 0.3818623423576355,
|
| 8362 |
+
"learning_rate": 1.4538172190115384e-05,
|
| 8363 |
+
"loss": 2.0433,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.025318206237001813,
|
| 8368 |
+
"grad_norm": 0.36803102493286133,
|
| 8369 |
+
"learning_rate": 1.4424063112138875e-05,
|
| 8370 |
+
"loss": 1.8027,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.02533948204056232,
|
| 8375 |
+
"grad_norm": 0.36837559938430786,
|
| 8376 |
+
"learning_rate": 1.4310368800294071e-05,
|
| 8377 |
+
"loss": 1.8881,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.025360757844122825,
|
| 8382 |
+
"grad_norm": 0.3547351658344269,
|
| 8383 |
+
"learning_rate": 1.4197089805630148e-05,
|
| 8384 |
+
"loss": 1.8507,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.025382033647683332,
|
| 8389 |
+
"grad_norm": 0.46106603741645813,
|
| 8390 |
+
"learning_rate": 1.4084226677183354e-05,
|
| 8391 |
+
"loss": 2.1002,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.025403309451243836,
|
| 8396 |
+
"grad_norm": 0.43937787413597107,
|
| 8397 |
+
"learning_rate": 1.3971779961974329e-05,
|
| 8398 |
+
"loss": 1.8439,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.025424585254804344,
|
| 8403 |
+
"grad_norm": 0.3855731189250946,
|
| 8404 |
+
"learning_rate": 1.385975020500544e-05,
|
| 8405 |
+
"loss": 1.7561,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.025445861058364848,
|
| 8410 |
+
"grad_norm": 0.3939302861690521,
|
| 8411 |
+
"learning_rate": 1.3748137949258132e-05,
|
| 8412 |
+
"loss": 1.9178,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.025467136861925355,
|
| 8417 |
+
"grad_norm": 0.38290202617645264,
|
| 8418 |
+
"learning_rate": 1.36369437356904e-05,
|
| 8419 |
+
"loss": 1.8738,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.02548841266548586,
|
| 8424 |
+
"grad_norm": 0.41021549701690674,
|
| 8425 |
+
"learning_rate": 1.3526168103234028e-05,
|
| 8426 |
+
"loss": 2.1128,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.025509688469046367,
|
| 8431 |
+
"grad_norm": 0.3867902457714081,
|
| 8432 |
+
"learning_rate": 1.3415811588792093e-05,
|
| 8433 |
+
"loss": 1.9701,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.02553096427260687,
|
| 8438 |
+
"grad_norm": 0.4187905788421631,
|
| 8439 |
+
"learning_rate": 1.3305874727236201e-05,
|
| 8440 |
+
"loss": 1.8594,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.02555224007616738,
|
| 8445 |
+
"grad_norm": 0.40430817008018494,
|
| 8446 |
+
"learning_rate": 1.3196358051404146e-05,
|
| 8447 |
+
"loss": 1.8805,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.025573515879727882,
|
| 8452 |
+
"grad_norm": 0.37520286440849304,
|
| 8453 |
+
"learning_rate": 1.308726209209703e-05,
|
| 8454 |
+
"loss": 1.7832,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.02559479168328839,
|
| 8459 |
+
"grad_norm": 0.40683072805404663,
|
| 8460 |
+
"learning_rate": 1.2978587378076945e-05,
|
| 8461 |
+
"loss": 2.0032,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.025616067486848894,
|
| 8466 |
+
"grad_norm": 0.41131818294525146,
|
| 8467 |
+
"learning_rate": 1.287033443606429e-05,
|
| 8468 |
+
"loss": 1.8312,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.0256373432904094,
|
| 8473 |
+
"grad_norm": 0.40292322635650635,
|
| 8474 |
+
"learning_rate": 1.2762503790735248e-05,
|
| 8475 |
+
"loss": 1.8747,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.025658619093969905,
|
| 8480 |
+
"grad_norm": 0.38722479343414307,
|
| 8481 |
+
"learning_rate": 1.2655095964719154e-05,
|
| 8482 |
+
"loss": 1.8538,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.025679894897530413,
|
| 8487 |
+
"grad_norm": 0.3931411802768707,
|
| 8488 |
+
"learning_rate": 1.2548111478596125e-05,
|
| 8489 |
+
"loss": 1.8604,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.025701170701090917,
|
| 8494 |
+
"grad_norm": 0.398651123046875,
|
| 8495 |
+
"learning_rate": 1.2441550850894434e-05,
|
| 8496 |
+
"loss": 1.7429,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.02572244650465142,
|
| 8501 |
+
"grad_norm": 0.4010455906391144,
|
| 8502 |
+
"learning_rate": 1.2335414598087957e-05,
|
| 8503 |
+
"loss": 1.814,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.02574372230821193,
|
| 8508 |
+
"grad_norm": 0.3841938376426697,
|
| 8509 |
+
"learning_rate": 1.2229703234593814e-05,
|
| 8510 |
+
"loss": 1.6558,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.025764998111772432,
|
| 8515 |
+
"grad_norm": 0.4107387363910675,
|
| 8516 |
+
"learning_rate": 1.2124417272769695e-05,
|
| 8517 |
+
"loss": 1.9101,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.02578627391533294,
|
| 8522 |
+
"grad_norm": 0.4160413146018982,
|
| 8523 |
+
"learning_rate": 1.2019557222911538e-05,
|
| 8524 |
+
"loss": 1.917,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.025807549718893444,
|
| 8529 |
+
"grad_norm": 0.3943398892879486,
|
| 8530 |
+
"learning_rate": 1.1915123593250955e-05,
|
| 8531 |
+
"loss": 1.9204,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.02582882552245395,
|
| 8536 |
+
"grad_norm": 0.3777497410774231,
|
| 8537 |
+
"learning_rate": 1.1811116889952833e-05,
|
| 8538 |
+
"loss": 1.7994,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.025850101326014455,
|
| 8543 |
+
"grad_norm": 0.3632175028324127,
|
| 8544 |
+
"learning_rate": 1.1707537617112829e-05,
|
| 8545 |
+
"loss": 2.1236,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.025871377129574963,
|
| 8550 |
+
"grad_norm": 0.4010590612888336,
|
| 8551 |
+
"learning_rate": 1.160438627675493e-05,
|
| 8552 |
+
"loss": 1.6692,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.025892652933135467,
|
| 8557 |
+
"grad_norm": 0.38821929693222046,
|
| 8558 |
+
"learning_rate": 1.150166336882904e-05,
|
| 8559 |
+
"loss": 1.7917,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.025913928736695974,
|
| 8564 |
+
"grad_norm": 0.41523030400276184,
|
| 8565 |
+
"learning_rate": 1.139936939120858e-05,
|
| 8566 |
+
"loss": 1.7513,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.02593520454025648,
|
| 8571 |
+
"grad_norm": 0.3896010220050812,
|
| 8572 |
+
"learning_rate": 1.1297504839688034e-05,
|
| 8573 |
+
"loss": 1.9406,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.025956480343816986,
|
| 8578 |
+
"grad_norm": 0.376023530960083,
|
| 8579 |
+
"learning_rate": 1.1196070207980569e-05,
|
| 8580 |
+
"loss": 1.9,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.02597775614737749,
|
| 8585 |
+
"grad_norm": 0.4502955675125122,
|
| 8586 |
+
"learning_rate": 1.1095065987715591e-05,
|
| 8587 |
+
"loss": 1.8199,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.025999031950937997,
|
| 8592 |
+
"grad_norm": 0.3819735050201416,
|
| 8593 |
+
"learning_rate": 1.0994492668436463e-05,
|
| 8594 |
+
"loss": 1.799,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.0260203077544985,
|
| 8599 |
+
"grad_norm": 0.3974403142929077,
|
| 8600 |
+
"learning_rate": 1.0894350737598058e-05,
|
| 8601 |
+
"loss": 1.8671,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.02604158355805901,
|
| 8606 |
+
"grad_norm": 0.41351979970932007,
|
| 8607 |
+
"learning_rate": 1.0794640680564372e-05,
|
| 8608 |
+
"loss": 1.7076,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.026062859361619513,
|
| 8613 |
+
"grad_norm": 0.42260563373565674,
|
| 8614 |
+
"learning_rate": 1.0695362980606249e-05,
|
| 8615 |
+
"loss": 1.9297,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.02608413516518002,
|
| 8620 |
+
"grad_norm": 0.41318872570991516,
|
| 8621 |
+
"learning_rate": 1.0596518118899035e-05,
|
| 8622 |
+
"loss": 1.9712,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.026105410968740524,
|
| 8627 |
+
"grad_norm": 0.41095852851867676,
|
| 8628 |
+
"learning_rate": 1.0498106574520105e-05,
|
| 8629 |
+
"loss": 1.8985,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.026126686772301032,
|
| 8634 |
+
"grad_norm": 0.38839998841285706,
|
| 8635 |
+
"learning_rate": 1.0400128824446753e-05,
|
| 8636 |
+
"loss": 2.0522,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.026147962575861536,
|
| 8641 |
+
"grad_norm": 0.3980291485786438,
|
| 8642 |
+
"learning_rate": 1.0302585343553739e-05,
|
| 8643 |
+
"loss": 1.959,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.026169238379422043,
|
| 8648 |
+
"grad_norm": 0.4165712594985962,
|
| 8649 |
+
"learning_rate": 1.0205476604611042e-05,
|
| 8650 |
+
"loss": 2.1,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.026190514182982547,
|
| 8655 |
+
"grad_norm": 0.3855735659599304,
|
| 8656 |
+
"learning_rate": 1.0108803078281503e-05,
|
| 8657 |
+
"loss": 1.9172,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.026211789986543055,
|
| 8662 |
+
"grad_norm": 0.38896241784095764,
|
| 8663 |
+
"learning_rate": 1.0012565233118597e-05,
|
| 8664 |
+
"loss": 1.9471,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.02623306579010356,
|
| 8669 |
+
"grad_norm": 0.3784373700618744,
|
| 8670 |
+
"learning_rate": 9.916763535564221e-06,
|
| 8671 |
+
"loss": 2.0471,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.026254341593664066,
|
| 8676 |
+
"grad_norm": 0.447631299495697,
|
| 8677 |
+
"learning_rate": 9.821398449946306e-06,
|
| 8678 |
+
"loss": 1.924,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.02627561739722457,
|
| 8683 |
+
"grad_norm": 0.4306599199771881,
|
| 8684 |
+
"learning_rate": 9.726470438476653e-06,
|
| 8685 |
+
"loss": 1.8301,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.026296893200785078,
|
| 8690 |
+
"grad_norm": 0.378780722618103,
|
| 8691 |
+
"learning_rate": 9.631979961248694e-06,
|
| 8692 |
+
"loss": 1.9342,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.026318169004345582,
|
| 8697 |
+
"grad_norm": 0.3726535439491272,
|
| 8698 |
+
"learning_rate": 9.537927476235175e-06,
|
| 8699 |
+
"loss": 1.8574,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.02633944480790609,
|
| 8704 |
+
"grad_norm": 0.3970372676849365,
|
| 8705 |
+
"learning_rate": 9.444313439286034e-06,
|
| 8706 |
+
"loss": 1.7524,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.026360720611466593,
|
| 8711 |
+
"grad_norm": 0.43677589297294617,
|
| 8712 |
+
"learning_rate": 9.351138304126161e-06,
|
| 8713 |
+
"loss": 1.6698,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.0263819964150271,
|
| 8718 |
+
"grad_norm": 0.37834641337394714,
|
| 8719 |
+
"learning_rate": 9.258402522353182e-06,
|
| 8720 |
+
"loss": 1.8125,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.026403272218587605,
|
| 8725 |
+
"grad_norm": 0.39560702443122864,
|
| 8726 |
+
"learning_rate": 9.166106543435316e-06,
|
| 8727 |
+
"loss": 1.9334,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.026424548022148112,
|
| 8732 |
+
"grad_norm": 0.4083397090435028,
|
| 8733 |
+
"learning_rate": 9.074250814709095e-06,
|
| 8734 |
+
"loss": 1.6393,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.026445823825708616,
|
| 8739 |
+
"grad_norm": 0.4539376497268677,
|
| 8740 |
+
"learning_rate": 8.982835781377318e-06,
|
| 8741 |
+
"loss": 1.7088,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.026467099629269124,
|
| 8746 |
+
"grad_norm": 0.38352295756340027,
|
| 8747 |
+
"learning_rate": 8.891861886506824e-06,
|
| 8748 |
+
"loss": 1.8229,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.026488375432829628,
|
| 8753 |
+
"grad_norm": 0.39906784892082214,
|
| 8754 |
+
"learning_rate": 8.801329571026373e-06,
|
| 8755 |
+
"loss": 1.7655,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.026509651236390135,
|
| 8760 |
+
"grad_norm": 0.37908226251602173,
|
| 8761 |
+
"learning_rate": 8.711239273724447e-06,
|
| 8762 |
+
"loss": 1.8801,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.02653092703995064,
|
| 8767 |
+
"grad_norm": 0.4054863154888153,
|
| 8768 |
+
"learning_rate": 8.621591431247244e-06,
|
| 8769 |
+
"loss": 1.7628,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.026552202843511147,
|
| 8774 |
+
"grad_norm": 0.39226725697517395,
|
| 8775 |
+
"learning_rate": 8.532386478096398e-06,
|
| 8776 |
+
"loss": 1.8788,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.02657347864707165,
|
| 8781 |
+
"grad_norm": 0.4020155966281891,
|
| 8782 |
+
"learning_rate": 8.443624846627063e-06,
|
| 8783 |
+
"loss": 1.8104,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.02659475445063216,
|
| 8788 |
+
"grad_norm": 0.4017484486103058,
|
| 8789 |
+
"learning_rate": 8.35530696704564e-06,
|
| 8790 |
+
"loss": 1.9193,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.026616030254192662,
|
| 8795 |
+
"grad_norm": 0.3993646204471588,
|
| 8796 |
+
"learning_rate": 8.267433267407842e-06,
|
| 8797 |
+
"loss": 1.9756,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.02663730605775317,
|
| 8802 |
+
"grad_norm": 0.3720335066318512,
|
| 8803 |
+
"learning_rate": 8.180004173616484e-06,
|
| 8804 |
+
"loss": 1.8116,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.026658581861313674,
|
| 8809 |
+
"grad_norm": 0.39052093029022217,
|
| 8810 |
+
"learning_rate": 8.09302010941949e-06,
|
| 8811 |
+
"loss": 1.7284,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.02667985766487418,
|
| 8816 |
+
"grad_norm": 0.40721672773361206,
|
| 8817 |
+
"learning_rate": 8.006481496407858e-06,
|
| 8818 |
+
"loss": 1.9753,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.026701133468434685,
|
| 8823 |
+
"grad_norm": 0.40760159492492676,
|
| 8824 |
+
"learning_rate": 7.920388754013597e-06,
|
| 8825 |
+
"loss": 2.0382,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.026722409271995193,
|
| 8830 |
+
"grad_norm": 0.45113295316696167,
|
| 8831 |
+
"learning_rate": 7.834742299507658e-06,
|
| 8832 |
+
"loss": 2.0173,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.026743685075555697,
|
| 8837 |
+
"grad_norm": 0.42447155714035034,
|
| 8838 |
+
"learning_rate": 7.749542547997968e-06,
|
| 8839 |
+
"loss": 2.0214,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.026764960879116204,
|
| 8844 |
+
"grad_norm": 0.3985137641429901,
|
| 8845 |
+
"learning_rate": 7.664789912427362e-06,
|
| 8846 |
+
"loss": 1.7741,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.02678623668267671,
|
| 8851 |
+
"grad_norm": 0.38606393337249756,
|
| 8852 |
+
"learning_rate": 7.580484803571641e-06,
|
| 8853 |
+
"loss": 1.8675,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.026807512486237216,
|
| 8858 |
+
"grad_norm": 0.39193248748779297,
|
| 8859 |
+
"learning_rate": 7.496627630037523e-06,
|
| 8860 |
+
"loss": 1.7698,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.02682878828979772,
|
| 8865 |
+
"grad_norm": 0.39374077320098877,
|
| 8866 |
+
"learning_rate": 7.413218798260702e-06,
|
| 8867 |
+
"loss": 1.8508,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.026850064093358227,
|
| 8872 |
+
"grad_norm": 0.40037915110588074,
|
| 8873 |
+
"learning_rate": 7.33025871250389e-06,
|
| 8874 |
+
"loss": 2.0289,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.02687133989691873,
|
| 8879 |
+
"grad_norm": 0.3833630681037903,
|
| 8880 |
+
"learning_rate": 7.247747774854796e-06,
|
| 8881 |
+
"loss": 1.8062,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.02689261570047924,
|
| 8886 |
+
"grad_norm": 0.37605446577072144,
|
| 8887 |
+
"learning_rate": 7.165686385224224e-06,
|
| 8888 |
+
"loss": 1.7292,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.026913891504039743,
|
| 8893 |
+
"grad_norm": 0.3856199085712433,
|
| 8894 |
+
"learning_rate": 7.084074941344155e-06,
|
| 8895 |
+
"loss": 1.879,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.02693516730760025,
|
| 8900 |
+
"grad_norm": 0.3812181353569031,
|
| 8901 |
+
"learning_rate": 7.002913838765779e-06,
|
| 8902 |
+
"loss": 1.8499,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.026956443111160754,
|
| 8907 |
+
"grad_norm": 0.39195212721824646,
|
| 8908 |
+
"learning_rate": 6.922203470857536e-06,
|
| 8909 |
+
"loss": 2.0196,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.026977718914721262,
|
| 8914 |
+
"grad_norm": 0.337797075510025,
|
| 8915 |
+
"learning_rate": 6.8419442288033544e-06,
|
| 8916 |
+
"loss": 1.8914,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.026998994718281766,
|
| 8921 |
+
"grad_norm": 0.3863051235675812,
|
| 8922 |
+
"learning_rate": 6.76213650160058e-06,
|
| 8923 |
+
"loss": 1.8511,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.027020270521842273,
|
| 8928 |
+
"grad_norm": 0.3731456398963928,
|
| 8929 |
+
"learning_rate": 6.68278067605822e-06,
|
| 8930 |
+
"loss": 1.7359,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.027041546325402777,
|
| 8935 |
+
"grad_norm": 0.3744986057281494,
|
| 8936 |
+
"learning_rate": 6.603877136795022e-06,
|
| 8937 |
+
"loss": 2.1574,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.027062822128963285,
|
| 8942 |
+
"grad_norm": 0.39129239320755005,
|
| 8943 |
+
"learning_rate": 6.5254262662376e-06,
|
| 8944 |
+
"loss": 1.914,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.02708409793252379,
|
| 8949 |
+
"grad_norm": 0.40502652525901794,
|
| 8950 |
+
"learning_rate": 6.447428444618575e-06,
|
| 8951 |
+
"loss": 1.7117,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.027105373736084296,
|
| 8956 |
+
"grad_norm": 0.36531540751457214,
|
| 8957 |
+
"learning_rate": 6.369884049974784e-06,
|
| 8958 |
+
"loss": 1.8006,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.0271266495396448,
|
| 8963 |
+
"grad_norm": 0.4066818654537201,
|
| 8964 |
+
"learning_rate": 6.292793458145374e-06,
|
| 8965 |
+
"loss": 1.6577,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.027147925343205308,
|
| 8970 |
+
"grad_norm": 0.39166662096977234,
|
| 8971 |
+
"learning_rate": 6.216157042770032e-06,
|
| 8972 |
+
"loss": 1.9577,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.027169201146765812,
|
| 8977 |
+
"grad_norm": 0.3947905898094177,
|
| 8978 |
+
"learning_rate": 6.139975175287171e-06,
|
| 8979 |
+
"loss": 1.7232,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.02719047695032632,
|
| 8984 |
+
"grad_norm": 0.39538541436195374,
|
| 8985 |
+
"learning_rate": 6.064248224932134e-06,
|
| 8986 |
+
"loss": 2.0029,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.027211752753886823,
|
| 8991 |
+
"grad_norm": 0.39565232396125793,
|
| 8992 |
+
"learning_rate": 5.988976558735327e-06,
|
| 8993 |
+
"loss": 1.9474,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.02723302855744733,
|
| 8998 |
+
"grad_norm": 0.37842199206352234,
|
| 8999 |
+
"learning_rate": 5.9141605415205525e-06,
|
| 9000 |
+
"loss": 1.87,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.027254304361007835,
|
| 9005 |
+
"grad_norm": 0.40924105048179626,
|
| 9006 |
+
"learning_rate": 5.839800535903217e-06,
|
| 9007 |
+
"loss": 1.7493,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.027275580164568342,
|
| 9012 |
+
"grad_norm": 0.3788970410823822,
|
| 9013 |
+
"learning_rate": 5.76589690228847e-06,
|
| 9014 |
+
"loss": 1.772,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.027296855968128846,
|
| 9019 |
+
"grad_norm": 0.37743717432022095,
|
| 9020 |
+
"learning_rate": 5.692449998869598e-06,
|
| 9021 |
+
"loss": 1.8087,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.02731813177168935,
|
| 9026 |
+
"grad_norm": 0.4205155074596405,
|
| 9027 |
+
"learning_rate": 5.61946018162618e-06,
|
| 9028 |
+
"loss": 1.8244,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.027339407575249858,
|
| 9033 |
+
"grad_norm": 0.3725995421409607,
|
| 9034 |
+
"learning_rate": 5.546927804322455e-06,
|
| 9035 |
+
"loss": 1.8394,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.027360683378810362,
|
| 9040 |
+
"grad_norm": 0.3655053377151489,
|
| 9041 |
+
"learning_rate": 5.474853218505515e-06,
|
| 9042 |
+
"loss": 1.6932,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.02738195918237087,
|
| 9047 |
+
"grad_norm": 0.389007031917572,
|
| 9048 |
+
"learning_rate": 5.403236773503695e-06,
|
| 9049 |
+
"loss": 2.1557,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.027403234985931373,
|
| 9054 |
+
"grad_norm": 0.40871235728263855,
|
| 9055 |
+
"learning_rate": 5.332078816424802e-06,
|
| 9056 |
+
"loss": 1.7485,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.02742451078949188,
|
| 9061 |
+
"grad_norm": 0.42805495858192444,
|
| 9062 |
+
"learning_rate": 5.261379692154433e-06,
|
| 9063 |
+
"loss": 2.0794,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.027445786593052385,
|
| 9068 |
+
"grad_norm": 0.42444759607315063,
|
| 9069 |
+
"learning_rate": 5.191139743354378e-06,
|
| 9070 |
+
"loss": 2.0183,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.027467062396612892,
|
| 9075 |
+
"grad_norm": 0.34589648246765137,
|
| 9076 |
+
"learning_rate": 5.121359310460893e-06,
|
| 9077 |
+
"loss": 1.6517,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.027488338200173396,
|
| 9082 |
+
"grad_norm": 0.3890903890132904,
|
| 9083 |
+
"learning_rate": 5.052038731683073e-06,
|
| 9084 |
+
"loss": 2.0088,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.027509614003733904,
|
| 9089 |
+
"grad_norm": 0.38702985644340515,
|
| 9090 |
+
"learning_rate": 4.983178343001204e-06,
|
| 9091 |
+
"loss": 1.9748,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.027530889807294408,
|
| 9096 |
+
"grad_norm": 0.37256234884262085,
|
| 9097 |
+
"learning_rate": 4.914778478165183e-06,
|
| 9098 |
+
"loss": 1.7918,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.027552165610854915,
|
| 9103 |
+
"grad_norm": 0.49755898118019104,
|
| 9104 |
+
"learning_rate": 4.846839468692777e-06,
|
| 9105 |
+
"loss": 1.9706,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.02757344141441542,
|
| 9110 |
+
"grad_norm": 0.4190734326839447,
|
| 9111 |
+
"learning_rate": 4.779361643868174e-06,
|
| 9112 |
+
"loss": 1.8438,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.027594717217975927,
|
| 9117 |
+
"grad_norm": 0.3469494879245758,
|
| 9118 |
+
"learning_rate": 4.712345330740253e-06,
|
| 9119 |
+
"loss": 2.0153,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.02761599302153643,
|
| 9124 |
+
"grad_norm": 0.36997222900390625,
|
| 9125 |
+
"learning_rate": 4.645790854121091e-06,
|
| 9126 |
+
"loss": 1.9133,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.027637268825096938,
|
| 9131 |
+
"grad_norm": 0.41857630014419556,
|
| 9132 |
+
"learning_rate": 4.579698536584365e-06,
|
| 9133 |
+
"loss": 2.1018,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.027658544628657442,
|
| 9138 |
+
"grad_norm": 0.41414695978164673,
|
| 9139 |
+
"learning_rate": 4.5140686984637136e-06,
|
| 9140 |
+
"loss": 2.0969,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.02767982043221795,
|
| 9145 |
+
"grad_norm": 0.40217119455337524,
|
| 9146 |
+
"learning_rate": 4.4489016578513234e-06,
|
| 9147 |
+
"loss": 1.9087,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.027701096235778454,
|
| 9152 |
+
"grad_norm": 0.3664983808994293,
|
| 9153 |
+
"learning_rate": 4.384197730596262e-06,
|
| 9154 |
+
"loss": 2.0693,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.02772237203933896,
|
| 9159 |
+
"grad_norm": 0.35403603315353394,
|
| 9160 |
+
"learning_rate": 4.319957230303029e-06,
|
| 9161 |
+
"loss": 1.8706,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.027743647842899465,
|
| 9166 |
+
"grad_norm": 0.3749202489852905,
|
| 9167 |
+
"learning_rate": 4.2561804683299844e-06,
|
| 9168 |
+
"loss": 1.9185,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.027764923646459973,
|
| 9173 |
+
"grad_norm": 0.407201886177063,
|
| 9174 |
+
"learning_rate": 4.1928677537878655e-06,
|
| 9175 |
+
"loss": 2.0707,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.027786199450020477,
|
| 9180 |
+
"grad_norm": 0.41097787022590637,
|
| 9181 |
+
"learning_rate": 4.130019393538265e-06,
|
| 9182 |
+
"loss": 1.7245,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.027807475253580984,
|
| 9187 |
+
"grad_norm": 0.42631837725639343,
|
| 9188 |
+
"learning_rate": 4.067635692192195e-06,
|
| 9189 |
+
"loss": 1.8989,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.027828751057141488,
|
| 9194 |
+
"grad_norm": 0.4397512674331665,
|
| 9195 |
+
"learning_rate": 4.005716952108552e-06,
|
| 9196 |
+
"loss": 1.706,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.027850026860701996,
|
| 9201 |
+
"grad_norm": 0.4033921957015991,
|
| 9202 |
+
"learning_rate": 3.944263473392706e-06,
|
| 9203 |
+
"loss": 1.9004,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.0278713026642625,
|
| 9208 |
+
"grad_norm": 0.3860911726951599,
|
| 9209 |
+
"learning_rate": 3.883275553894972e-06,
|
| 9210 |
+
"loss": 1.8106,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.027892578467823007,
|
| 9215 |
+
"grad_norm": 0.39403754472732544,
|
| 9216 |
+
"learning_rate": 3.822753489209219e-06,
|
| 9217 |
+
"loss": 1.9137,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.02791385427138351,
|
| 9222 |
+
"grad_norm": 0.3832218050956726,
|
| 9223 |
+
"learning_rate": 3.7626975726714588e-06,
|
| 9224 |
+
"loss": 2.0144,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.02793513007494402,
|
| 9229 |
+
"grad_norm": 0.38024333119392395,
|
| 9230 |
+
"learning_rate": 3.7031080953583873e-06,
|
| 9231 |
+
"loss": 2.022,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.027956405878504523,
|
| 9236 |
+
"grad_norm": 0.3953205645084381,
|
| 9237 |
+
"learning_rate": 3.643985346085965e-06,
|
| 9238 |
+
"loss": 1.9927,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.02797768168206503,
|
| 9243 |
+
"grad_norm": 0.4071060121059418,
|
| 9244 |
+
"learning_rate": 3.5853296114080392e-06,
|
| 9245 |
+
"loss": 1.8,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.027998957485625534,
|
| 9250 |
+
"grad_norm": 0.388224720954895,
|
| 9251 |
+
"learning_rate": 3.527141175614945e-06,
|
| 9252 |
+
"loss": 1.7697,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.02802023328918604,
|
| 9257 |
+
"grad_norm": 0.40393635630607605,
|
| 9258 |
+
"learning_rate": 3.469420320732153e-06,
|
| 9259 |
+
"loss": 1.7189,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.028041509092746546,
|
| 9264 |
+
"grad_norm": 0.40168148279190063,
|
| 9265 |
+
"learning_rate": 3.4121673265188335e-06,
|
| 9266 |
+
"loss": 1.7975,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.028062784896307053,
|
| 9271 |
+
"grad_norm": 0.39723289012908936,
|
| 9272 |
+
"learning_rate": 3.355382470466606e-06,
|
| 9273 |
+
"loss": 1.73,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.028084060699867557,
|
| 9278 |
+
"grad_norm": 0.38528016209602356,
|
| 9279 |
+
"learning_rate": 3.2990660277981032e-06,
|
| 9280 |
+
"loss": 1.7325,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.028105336503428065,
|
| 9285 |
+
"grad_norm": 0.37540358304977417,
|
| 9286 |
+
"learning_rate": 3.2432182714656534e-06,
|
| 9287 |
+
"loss": 1.8673,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.02812661230698857,
|
| 9292 |
+
"grad_norm": 0.40019020438194275,
|
| 9293 |
+
"learning_rate": 3.1878394721500225e-06,
|
| 9294 |
+
"loss": 2.0248,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.028147888110549076,
|
| 9299 |
+
"grad_norm": 0.431240439414978,
|
| 9300 |
+
"learning_rate": 3.1329298982590184e-06,
|
| 9301 |
+
"loss": 2.0248,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.02816916391410958,
|
| 9306 |
+
"grad_norm": 0.3818899095058441,
|
| 9307 |
+
"learning_rate": 3.078489815926222e-06,
|
| 9308 |
+
"loss": 2.0896,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.028190439717670088,
|
| 9313 |
+
"grad_norm": 0.3970624506473541,
|
| 9314 |
+
"learning_rate": 3.024519489009736e-06,
|
| 9315 |
+
"loss": 1.841,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.02821171552123059,
|
| 9320 |
+
"grad_norm": 0.41433432698249817,
|
| 9321 |
+
"learning_rate": 2.9710191790908394e-06,
|
| 9322 |
+
"loss": 2.0743,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.0282329913247911,
|
| 9327 |
+
"grad_norm": 0.3981134295463562,
|
| 9328 |
+
"learning_rate": 2.9179891454727437e-06,
|
| 9329 |
+
"loss": 1.8804,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.028254267128351603,
|
| 9334 |
+
"grad_norm": 0.40185731649398804,
|
| 9335 |
+
"learning_rate": 2.865429645179374e-06,
|
| 9336 |
+
"loss": 1.7234,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.02827554293191211,
|
| 9341 |
+
"grad_norm": 0.3775731325149536,
|
| 9342 |
+
"learning_rate": 2.813340932954067e-06,
|
| 9343 |
+
"loss": 1.9488,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.028296818735472615,
|
| 9348 |
+
"grad_norm": 0.38112545013427734,
|
| 9349 |
+
"learning_rate": 2.7617232612583977e-06,
|
| 9350 |
+
"loss": 1.8355,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.028318094539033122,
|
| 9355 |
+
"grad_norm": 0.4050240218639374,
|
| 9356 |
+
"learning_rate": 2.7105768802708544e-06,
|
| 9357 |
+
"loss": 1.9147,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.028339370342593626,
|
| 9362 |
+
"grad_norm": 0.4066665470600128,
|
| 9363 |
+
"learning_rate": 2.659902037885775e-06,
|
| 9364 |
+
"loss": 1.7867,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.028360646146154134,
|
| 9369 |
+
"grad_norm": 0.4173521101474762,
|
| 9370 |
+
"learning_rate": 2.6096989797119943e-06,
|
| 9371 |
+
"loss": 1.9073,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.028381921949714638,
|
| 9376 |
+
"grad_norm": 0.4105026125907898,
|
| 9377 |
+
"learning_rate": 2.55996794907174e-06,
|
| 9378 |
+
"loss": 1.7811,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.028403197753275145,
|
| 9383 |
+
"grad_norm": 0.39894595742225647,
|
| 9384 |
+
"learning_rate": 2.5107091869994714e-06,
|
| 9385 |
+
"loss": 1.8126,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.02842447355683565,
|
| 9390 |
+
"grad_norm": 0.3766692578792572,
|
| 9391 |
+
"learning_rate": 2.4619229322406347e-06,
|
| 9392 |
+
"loss": 1.8318,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.028445749360396157,
|
| 9397 |
+
"grad_norm": 0.3781270980834961,
|
| 9398 |
+
"learning_rate": 2.41360942125054e-06,
|
| 9399 |
+
"loss": 1.8062,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.02846702516395666,
|
| 9404 |
+
"grad_norm": 0.3879123032093048,
|
| 9405 |
+
"learning_rate": 2.3657688881932536e-06,
|
| 9406 |
+
"loss": 1.8371,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.028488300967517168,
|
| 9411 |
+
"grad_norm": 0.38924089074134827,
|
| 9412 |
+
"learning_rate": 2.31840156494042e-06,
|
| 9413 |
+
"loss": 2.0485,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.028509576771077672,
|
| 9418 |
+
"grad_norm": 0.4108506739139557,
|
| 9419 |
+
"learning_rate": 2.2715076810701284e-06,
|
| 9420 |
+
"loss": 2.1079,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.02853085257463818,
|
| 9425 |
+
"grad_norm": 0.39283981919288635,
|
| 9426 |
+
"learning_rate": 2.2250874638658604e-06,
|
| 9427 |
+
"loss": 1.8395,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.028552128378198684,
|
| 9432 |
+
"grad_norm": 0.3722197711467743,
|
| 9433 |
+
"learning_rate": 2.1791411383152995e-06,
|
| 9434 |
+
"loss": 1.6722,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.02857340418175919,
|
| 9439 |
+
"grad_norm": 0.38988062739372253,
|
| 9440 |
+
"learning_rate": 2.1336689271093224e-06,
|
| 9441 |
+
"loss": 1.9387,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.028594679985319695,
|
| 9446 |
+
"grad_norm": 0.39571458101272583,
|
| 9447 |
+
"learning_rate": 2.088671050640889e-06,
|
| 9448 |
+
"loss": 1.8288,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.028615955788880203,
|
| 9453 |
+
"grad_norm": 0.3794815242290497,
|
| 9454 |
+
"learning_rate": 2.044147727003942e-06,
|
| 9455 |
+
"loss": 1.8673,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.028637231592440707,
|
| 9460 |
+
"grad_norm": 0.39228445291519165,
|
| 9461 |
+
"learning_rate": 2.00009917199242e-06,
|
| 9462 |
+
"loss": 1.9868,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.028658507396001214,
|
| 9467 |
+
"grad_norm": 0.36269429326057434,
|
| 9468 |
+
"learning_rate": 1.956525599099146e-06,
|
| 9469 |
+
"loss": 2.0216,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.028679783199561718,
|
| 9474 |
+
"grad_norm": 0.4397389888763428,
|
| 9475 |
+
"learning_rate": 1.9134272195148075e-06,
|
| 9476 |
+
"loss": 1.8674,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.028701059003122226,
|
| 9481 |
+
"grad_norm": 0.3761897385120392,
|
| 9482 |
+
"learning_rate": 1.8708042421269777e-06,
|
| 9483 |
+
"loss": 1.8565,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.02872233480668273,
|
| 9488 |
+
"grad_norm": 0.38449329137802124,
|
| 9489 |
+
"learning_rate": 1.8286568735190411e-06,
|
| 9490 |
+
"loss": 1.9071,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.028743610610243237,
|
| 9495 |
+
"grad_norm": 0.39230549335479736,
|
| 9496 |
+
"learning_rate": 1.7869853179692365e-06,
|
| 9497 |
+
"loss": 1.7119,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.02876488641380374,
|
| 9502 |
+
"grad_norm": 0.39381974935531616,
|
| 9503 |
+
"learning_rate": 1.7457897774496245e-06,
|
| 9504 |
+
"loss": 1.8456,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.02878616221736425,
|
| 9509 |
+
"grad_norm": 0.396465539932251,
|
| 9510 |
+
"learning_rate": 1.7050704516251682e-06,
|
| 9511 |
+
"loss": 2.0615,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.028807438020924753,
|
| 9516 |
+
"grad_norm": 0.3799900710582733,
|
| 9517 |
+
"learning_rate": 1.6648275378527201e-06,
|
| 9518 |
+
"loss": 1.8192,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.02882871382448526,
|
| 9523 |
+
"grad_norm": 0.4248625934123993,
|
| 9524 |
+
"learning_rate": 1.6250612311800695e-06,
|
| 9525 |
+
"loss": 1.949,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.028849989628045764,
|
| 9530 |
+
"grad_norm": 0.39222586154937744,
|
| 9531 |
+
"learning_rate": 1.5857717243449976e-06,
|
| 9532 |
+
"loss": 2.0664,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.02887126543160627,
|
| 9537 |
+
"grad_norm": 0.4144866466522217,
|
| 9538 |
+
"learning_rate": 1.54695920777439e-06,
|
| 9539 |
+
"loss": 1.9277,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.028892541235166776,
|
| 9544 |
+
"grad_norm": 0.403522789478302,
|
| 9545 |
+
"learning_rate": 1.5086238695832145e-06,
|
| 9546 |
+
"loss": 1.8818,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.02891381703872728,
|
| 9551 |
+
"grad_norm": 0.3855152130126953,
|
| 9552 |
+
"learning_rate": 1.4707658955737113e-06,
|
| 9553 |
+
"loss": 1.9005,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.028935092842287787,
|
| 9558 |
+
"grad_norm": 0.3692283630371094,
|
| 9559 |
+
"learning_rate": 1.4333854692344494e-06,
|
| 9560 |
+
"loss": 2.0313,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.02895636864584829,
|
| 9565 |
+
"grad_norm": 0.4059112071990967,
|
| 9566 |
+
"learning_rate": 1.396482771739438e-06,
|
| 9567 |
+
"loss": 1.8817,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.0289776444494088,
|
| 9572 |
+
"grad_norm": 0.41557577252388,
|
| 9573 |
+
"learning_rate": 1.360057981947227e-06,
|
| 9574 |
+
"loss": 2.0334,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.028998920252969303,
|
| 9579 |
+
"grad_norm": 0.39747360348701477,
|
| 9580 |
+
"learning_rate": 1.324111276400064e-06,
|
| 9581 |
+
"loss": 1.993,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.02902019605652981,
|
| 9586 |
+
"grad_norm": 0.3804445266723633,
|
| 9587 |
+
"learning_rate": 1.2886428293230502e-06,
|
| 9588 |
+
"loss": 1.955,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.029041471860090314,
|
| 9593 |
+
"grad_norm": 0.4278270900249481,
|
| 9594 |
+
"learning_rate": 1.253652812623296e-06,
|
| 9595 |
+
"loss": 2.013,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.02906274766365082,
|
| 9600 |
+
"grad_norm": 0.3975847065448761,
|
| 9601 |
+
"learning_rate": 1.2191413958890452e-06,
|
| 9602 |
+
"loss": 1.8955,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.029084023467211326,
|
| 9607 |
+
"grad_norm": 0.4011601507663727,
|
| 9608 |
+
"learning_rate": 1.1851087463889076e-06,
|
| 9609 |
+
"loss": 1.8095,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.029105299270771833,
|
| 9614 |
+
"grad_norm": 0.4060905873775482,
|
| 9615 |
+
"learning_rate": 1.1515550290709832e-06,
|
| 9616 |
+
"loss": 2.0343,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.029126575074332337,
|
| 9621 |
+
"grad_norm": 0.3706141412258148,
|
| 9622 |
+
"learning_rate": 1.1184804065621502e-06,
|
| 9623 |
+
"loss": 1.9444,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.029147850877892845,
|
| 9628 |
+
"grad_norm": 0.4209257960319519,
|
| 9629 |
+
"learning_rate": 1.0858850391671894e-06,
|
| 9630 |
+
"loss": 1.8842,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.02916912668145335,
|
| 9635 |
+
"grad_norm": 0.38073280453681946,
|
| 9636 |
+
"learning_rate": 1.0537690848680837e-06,
|
| 9637 |
+
"loss": 1.8503,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.029190402485013856,
|
| 9642 |
+
"grad_norm": 0.4212806522846222,
|
| 9643 |
+
"learning_rate": 1.0221326993231861e-06,
|
| 9644 |
+
"loss": 2.0407,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.02921167828857436,
|
| 9649 |
+
"grad_norm": 0.3912928104400635,
|
| 9650 |
+
"learning_rate": 9.909760358664867e-07,
|
| 9651 |
+
"loss": 2.029,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.029232954092134868,
|
| 9656 |
+
"grad_norm": 0.4083358645439148,
|
| 9657 |
+
"learning_rate": 9.602992455069126e-07,
|
| 9658 |
+
"loss": 1.8164,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.02925422989569537,
|
| 9663 |
+
"grad_norm": 0.42386138439178467,
|
| 9664 |
+
"learning_rate": 9.301024769275302e-07,
|
| 9665 |
+
"loss": 1.7677,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.02927550569925588,
|
| 9670 |
+
"grad_norm": 0.4002361595630646,
|
| 9671 |
+
"learning_rate": 9.003858764848771e-07,
|
| 9672 |
+
"loss": 1.9575,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.029296781502816383,
|
| 9677 |
+
"grad_norm": 0.3702812194824219,
|
| 9678 |
+
"learning_rate": 8.711495882082087e-07,
|
| 9679 |
+
"loss": 1.7899,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.02931805730637689,
|
| 9684 |
+
"grad_norm": 0.41200730204582214,
|
| 9685 |
+
"learning_rate": 8.423937537988536e-07,
|
| 9686 |
+
"loss": 1.9047,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.029339333109937395,
|
| 9691 |
+
"grad_norm": 0.3852335214614868,
|
| 9692 |
+
"learning_rate": 8.14118512629447e-07,
|
| 9693 |
+
"loss": 2.1176,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.029360608913497902,
|
| 9698 |
+
"grad_norm": 0.37616676092147827,
|
| 9699 |
+
"learning_rate": 7.863240017433438e-07,
|
| 9700 |
+
"loss": 1.6359,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.029381884717058406,
|
| 9705 |
+
"grad_norm": 0.37500837445259094,
|
| 9706 |
+
"learning_rate": 7.590103558538952e-07,
|
| 9707 |
+
"loss": 1.8518,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.029403160520618914,
|
| 9712 |
+
"grad_norm": 0.420379638671875,
|
| 9713 |
+
"learning_rate": 7.321777073438397e-07,
|
| 9714 |
+
"loss": 2.0172,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.029424436324179418,
|
| 9719 |
+
"grad_norm": 0.3880142271518707,
|
| 9720 |
+
"learning_rate": 7.058261862646021e-07,
|
| 9721 |
+
"loss": 1.866,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.029445712127739925,
|
| 9726 |
+
"grad_norm": 0.4096420109272003,
|
| 9727 |
+
"learning_rate": 6.799559203357175e-07,
|
| 9728 |
+
"loss": 1.7721,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.02946698793130043,
|
| 9733 |
+
"grad_norm": 0.38439473509788513,
|
| 9734 |
+
"learning_rate": 6.545670349441757e-07,
|
| 9735 |
+
"loss": 1.8267,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.029488263734860937,
|
| 9740 |
+
"grad_norm": 0.39978665113449097,
|
| 9741 |
+
"learning_rate": 6.296596531438547e-07,
|
| 9742 |
+
"loss": 1.7858,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.02950953953842144,
|
| 9747 |
+
"grad_norm": 0.3609836995601654,
|
| 9748 |
+
"learning_rate": 6.052338956548776e-07,
|
| 9749 |
+
"loss": 1.8623,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.029530815341981948,
|
| 9754 |
+
"grad_norm": 0.38551342487335205,
|
| 9755 |
+
"learning_rate": 5.812898808630785e-07,
|
| 9756 |
+
"loss": 2.0972,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.029552091145542452,
|
| 9761 |
+
"grad_norm": 0.4218285083770752,
|
| 9762 |
+
"learning_rate": 5.578277248193486e-07,
|
| 9763 |
+
"loss": 1.9541,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.02957336694910296,
|
| 9768 |
+
"grad_norm": 0.391440749168396,
|
| 9769 |
+
"learning_rate": 5.348475412391806e-07,
|
| 9770 |
+
"loss": 1.9765,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.029594642752663464,
|
| 9775 |
+
"grad_norm": 0.38792580366134644,
|
| 9776 |
+
"learning_rate": 5.123494415020247e-07,
|
| 9777 |
+
"loss": 1.6567,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.02961591855622397,
|
| 9782 |
+
"grad_norm": 0.389170378446579,
|
| 9783 |
+
"learning_rate": 4.903335346507997e-07,
|
| 9784 |
+
"loss": 1.8255,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.029637194359784475,
|
| 9789 |
+
"grad_norm": 0.38838863372802734,
|
| 9790 |
+
"learning_rate": 4.6879992739136127e-07,
|
| 9791 |
+
"loss": 1.962,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.029658470163344983,
|
| 9796 |
+
"grad_norm": 0.3822590112686157,
|
| 9797 |
+
"learning_rate": 4.4774872409193426e-07,
|
| 9798 |
+
"loss": 1.8072,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.029679745966905487,
|
| 9803 |
+
"grad_norm": 0.4434032440185547,
|
| 9804 |
+
"learning_rate": 4.271800267827031e-07,
|
| 9805 |
+
"loss": 2.0363,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.029701021770465994,
|
| 9810 |
+
"grad_norm": 0.4102432131767273,
|
| 9811 |
+
"learning_rate": 4.0709393515521165e-07,
|
| 9812 |
+
"loss": 2.1042,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.029722297574026498,
|
| 9817 |
+
"grad_norm": 0.37742215394973755,
|
| 9818 |
+
"learning_rate": 3.874905465619749e-07,
|
| 9819 |
+
"loss": 1.7551,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.029743573377587006,
|
| 9824 |
+
"grad_norm": 0.40014660358428955,
|
| 9825 |
+
"learning_rate": 3.683699560159015e-07,
|
| 9826 |
+
"loss": 2.0128,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.02976484918114751,
|
| 9831 |
+
"grad_norm": 0.43463030457496643,
|
| 9832 |
+
"learning_rate": 3.4973225618996077e-07,
|
| 9833 |
+
"loss": 1.8536,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.029786124984708017,
|
| 9838 |
+
"grad_norm": 0.38287851214408875,
|
| 9839 |
+
"learning_rate": 3.3157753741661635e-07,
|
| 9840 |
+
"loss": 1.7769,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.02980740078826852,
|
| 9845 |
+
"grad_norm": 0.42383190989494324,
|
| 9846 |
+
"learning_rate": 3.13905887687449e-07,
|
| 9847 |
+
"loss": 1.9495,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.02982867659182903,
|
| 9852 |
+
"grad_norm": 0.39634689688682556,
|
| 9853 |
+
"learning_rate": 2.967173926527345e-07,
|
| 9854 |
+
"loss": 1.801,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.029849952395389533,
|
| 9859 |
+
"grad_norm": 0.40020450949668884,
|
| 9860 |
+
"learning_rate": 2.8001213562101056e-07,
|
| 9861 |
+
"loss": 1.5355,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.02987122819895004,
|
| 9866 |
+
"grad_norm": 0.4128990173339844,
|
| 9867 |
+
"learning_rate": 2.6379019755866654e-07,
|
| 9868 |
+
"loss": 1.9336,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.029892504002510544,
|
| 9873 |
+
"grad_norm": 0.4198834300041199,
|
| 9874 |
+
"learning_rate": 2.4805165708954306e-07,
|
| 9875 |
+
"loss": 1.9941,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.02991377980607105,
|
| 9880 |
+
"grad_norm": 0.3714563250541687,
|
| 9881 |
+
"learning_rate": 2.3279659049461055e-07,
|
| 9882 |
+
"loss": 1.8018,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.029935055609631556,
|
| 9887 |
+
"grad_norm": 0.39097511768341064,
|
| 9888 |
+
"learning_rate": 2.1802507171152508e-07,
|
| 9889 |
+
"loss": 1.867,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.029956331413192063,
|
| 9894 |
+
"grad_norm": 0.3634189963340759,
|
| 9895 |
+
"learning_rate": 2.0373717233431732e-07,
|
| 9896 |
+
"loss": 1.8117,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.029977607216752567,
|
| 9901 |
+
"grad_norm": 0.3934725224971771,
|
| 9902 |
+
"learning_rate": 1.8993296161301522e-07,
|
| 9903 |
+
"loss": 1.8086,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.029998883020313075,
|
| 9908 |
+
"grad_norm": 0.3722934126853943,
|
| 9909 |
+
"learning_rate": 1.7661250645333304e-07,
|
| 9910 |
+
"loss": 1.9438,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.03002015882387358,
|
| 9915 |
+
"grad_norm": 0.378426730632782,
|
| 9916 |
+
"learning_rate": 1.6377587141633843e-07,
|
| 9917 |
+
"loss": 1.7342,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.030041434627434086,
|
| 9922 |
+
"grad_norm": 0.3758505582809448,
|
| 9923 |
+
"learning_rate": 1.514231187181303e-07,
|
| 9924 |
+
"loss": 1.8911,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.03006271043099459,
|
| 9929 |
+
"grad_norm": 0.36688435077667236,
|
| 9930 |
+
"learning_rate": 1.395543082295614e-07,
|
| 9931 |
+
"loss": 1.8377,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.030083986234555098,
|
| 9936 |
+
"grad_norm": 0.4031440317630768,
|
| 9937 |
+
"learning_rate": 1.2816949747592732e-07,
|
| 9938 |
+
"loss": 1.7723,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.0301052620381156,
|
| 9943 |
+
"grad_norm": 0.41031911969184875,
|
| 9944 |
+
"learning_rate": 1.172687416366669e-07,
|
| 9945 |
+
"loss": 1.9124,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.03012653784167611,
|
| 9950 |
+
"grad_norm": 0.38181814551353455,
|
| 9951 |
+
"learning_rate": 1.0685209354515114e-07,
|
| 9952 |
+
"loss": 1.8236,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.030147813645236613,
|
| 9957 |
+
"grad_norm": 0.38461101055145264,
|
| 9958 |
+
"learning_rate": 9.691960368838349e-08,
|
| 9959 |
+
"loss": 1.9541,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.03016908944879712,
|
| 9964 |
+
"grad_norm": 0.38355132937431335,
|
| 9965 |
+
"learning_rate": 8.747132020677784e-08,
|
| 9966 |
+
"loss": 1.7808,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.030190365252357625,
|
| 9971 |
+
"grad_norm": 0.39530685544013977,
|
| 9972 |
+
"learning_rate": 7.850728889391424e-08,
|
| 9973 |
+
"loss": 1.9751,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.030211641055918132,
|
| 9978 |
+
"grad_norm": 0.3940103352069855,
|
| 9979 |
+
"learning_rate": 7.002755319628352e-08,
|
| 9980 |
+
"loss": 1.6835,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.030232916859478636,
|
| 9985 |
+
"grad_norm": 0.3707810640335083,
|
| 9986 |
+
"learning_rate": 6.203215421316521e-08,
|
| 9987 |
+
"loss": 1.8657,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.030254192663039144,
|
| 9992 |
+
"grad_norm": 0.4752837121486664,
|
| 9993 |
+
"learning_rate": 5.452113069633891e-08,
|
| 9994 |
+
"loss": 1.895,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.030275468466599648,
|
| 9999 |
+
"grad_norm": 0.40118181705474854,
|
| 10000 |
+
"learning_rate": 4.749451904995095e-08,
|
| 10001 |
+
"loss": 1.9018,
|
| 10002 |
+
"step": 1423
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 0.030296744270160155,
|
| 10006 |
+
"grad_norm": 0.4106770157814026,
|
| 10007 |
+
"learning_rate": 4.095235333030356e-08,
|
| 10008 |
+
"loss": 1.7572,
|
| 10009 |
+
"step": 1424
|
| 10010 |
+
},
|
| 10011 |
+
{
|
| 10012 |
+
"epoch": 0.03031802007372066,
|
| 10013 |
+
"grad_norm": 0.4028623402118683,
|
| 10014 |
+
"learning_rate": 3.48946652456994e-08,
|
| 10015 |
+
"loss": 2.0699,
|
| 10016 |
+
"step": 1425
|
| 10017 |
+
},
|
| 10018 |
+
{
|
| 10019 |
+
"epoch": 0.030339295877281167,
|
| 10020 |
+
"grad_norm": 0.40891626477241516,
|
| 10021 |
+
"learning_rate": 2.9321484156330515e-08,
|
| 10022 |
+
"loss": 1.8153,
|
| 10023 |
+
"step": 1426
|
| 10024 |
+
},
|
| 10025 |
+
{
|
| 10026 |
+
"epoch": 0.03036057168084167,
|
| 10027 |
+
"grad_norm": 0.38749581575393677,
|
| 10028 |
+
"learning_rate": 2.4232837074056326e-08,
|
| 10029 |
+
"loss": 1.7221,
|
| 10030 |
+
"step": 1427
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 0.030381847484402178,
|
| 10034 |
+
"grad_norm": 0.45594656467437744,
|
| 10035 |
+
"learning_rate": 1.9628748662348096e-08,
|
| 10036 |
+
"loss": 2.0707,
|
| 10037 |
+
"step": 1428
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 0.030403123287962682,
|
| 10041 |
+
"grad_norm": 0.3742702901363373,
|
| 10042 |
+
"learning_rate": 1.5509241236111305e-08,
|
| 10043 |
+
"loss": 1.8493,
|
| 10044 |
+
"step": 1429
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 0.03042439909152319,
|
| 10048 |
+
"grad_norm": 0.39443689584732056,
|
| 10049 |
+
"learning_rate": 1.1874334761619033e-08,
|
| 10050 |
+
"loss": 2.1747,
|
| 10051 |
+
"step": 1430
|
| 10052 |
+
},
|
| 10053 |
+
{
|
| 10054 |
+
"epoch": 0.030445674895083694,
|
| 10055 |
+
"grad_norm": 0.35878825187683105,
|
| 10056 |
+
"learning_rate": 8.724046856389833e-09,
|
| 10057 |
+
"loss": 1.774,
|
| 10058 |
+
"step": 1431
|
| 10059 |
+
},
|
| 10060 |
+
{
|
| 10061 |
+
"epoch": 0.0304669506986442,
|
| 10062 |
+
"grad_norm": 0.40466150641441345,
|
| 10063 |
+
"learning_rate": 6.058392789132228e-09,
|
| 10064 |
+
"loss": 2.1955,
|
| 10065 |
+
"step": 1432
|
| 10066 |
+
},
|
| 10067 |
+
{
|
| 10068 |
+
"epoch": 0.030488226502204705,
|
| 10069 |
+
"grad_norm": 0.4434767961502075,
|
| 10070 |
+
"learning_rate": 3.877385479622575e-09,
|
| 10071 |
+
"loss": 2.0186,
|
| 10072 |
+
"step": 1433
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 0.030509502305765213,
|
| 10076 |
+
"grad_norm": 0.3923577666282654,
|
| 10077 |
+
"learning_rate": 2.1810354986828708e-09,
|
| 10078 |
+
"loss": 1.6698,
|
| 10079 |
+
"step": 1434
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 0.030530778109325717,
|
| 10083 |
+
"grad_norm": 0.4112854599952698,
|
| 10084 |
+
"learning_rate": 9.693510681141326e-10,
|
| 10085 |
+
"loss": 1.7564,
|
| 10086 |
+
"step": 1435
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 0.03055205391288622,
|
| 10090 |
+
"grad_norm": 0.4083521068096161,
|
| 10091 |
+
"learning_rate": 2.423380606630943e-10,
|
| 10092 |
+
"loss": 1.8694,
|
| 10093 |
+
"step": 1436
|
| 10094 |
+
},
|
| 10095 |
+
{
|
| 10096 |
+
"epoch": 0.030573329716446728,
|
| 10097 |
+
"grad_norm": 0.3870006501674652,
|
| 10098 |
+
"learning_rate": 0.0,
|
| 10099 |
+
"loss": 1.9612,
|
| 10100 |
+
"step": 1437
|
| 10101 |
}
|
| 10102 |
],
|
| 10103 |
"logging_steps": 1,
|
|
|
|
| 10112 |
"should_evaluate": false,
|
| 10113 |
"should_log": false,
|
| 10114 |
"should_save": true,
|
| 10115 |
+
"should_training_stop": true
|
| 10116 |
},
|
| 10117 |
"attributes": {}
|
| 10118 |
}
|
| 10119 |
},
|
| 10120 |
+
"total_flos": 5.420113505707622e+16,
|
| 10121 |
"train_batch_size": 2,
|
| 10122 |
"trial_name": null,
|
| 10123 |
"trial_params": null
|