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 100059752
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:337aad9699541f09b0ef59416f279a8c8e1c00bb56d6de273ee8cedfd66fac2a
|
| 3 |
size 100059752
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 51244404
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f97630553a15c8ea5ea812100ace5730b92b0062f5a91b142400d50d4fde46b1
|
| 3 |
size 51244404
|
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 sha256:704043e3aa550a9df4c1acfabdf4456ddcbbece3d2cbeacb4181873f95fd932f
|
| 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": 28.136,
|
| 7600 |
"eval_steps_per_second": 14.068,
|
| 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": 2.
|
| 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": 0.12125303238055057,
|
| 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": 28.136,
|
| 7600 |
"eval_steps_per_second": 14.068,
|
| 7601 |
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.091214006961291,
|
| 7605 |
+
"grad_norm": 2.8509836196899414,
|
| 7606 |
+
"learning_rate": 2.917266421714292e-05,
|
| 7607 |
+
"loss": 1.4007,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.09129838624617656,
|
| 7612 |
+
"grad_norm": 3.5944178104400635,
|
| 7613 |
+
"learning_rate": 2.9017421092606924e-05,
|
| 7614 |
+
"loss": 1.8301,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.09138276553106213,
|
| 7619 |
+
"grad_norm": 3.2917916774749756,
|
| 7620 |
+
"learning_rate": 2.8862522003681192e-05,
|
| 7621 |
+
"loss": 1.917,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.09146714481594768,
|
| 7626 |
+
"grad_norm": 3.365098237991333,
|
| 7627 |
+
"learning_rate": 2.870796770112466e-05,
|
| 7628 |
+
"loss": 1.3544,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.09155152410083324,
|
| 7633 |
+
"grad_norm": 3.2107977867126465,
|
| 7634 |
+
"learning_rate": 2.8553758934025097e-05,
|
| 7635 |
+
"loss": 1.735,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.09163590338571881,
|
| 7640 |
+
"grad_norm": 2.7982823848724365,
|
| 7641 |
+
"learning_rate": 2.839989644979555e-05,
|
| 7642 |
+
"loss": 1.296,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.09172028267060436,
|
| 7647 |
+
"grad_norm": 4.238430976867676,
|
| 7648 |
+
"learning_rate": 2.8246380994170752e-05,
|
| 7649 |
+
"loss": 1.6262,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.09180466195548993,
|
| 7654 |
+
"grad_norm": 3.106069326400757,
|
| 7655 |
+
"learning_rate": 2.809321331120346e-05,
|
| 7656 |
+
"loss": 1.4831,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.09188904124037549,
|
| 7661 |
+
"grad_norm": 3.0739190578460693,
|
| 7662 |
+
"learning_rate": 2.7940394143260872e-05,
|
| 7663 |
+
"loss": 1.5681,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.09197342052526104,
|
| 7668 |
+
"grad_norm": 3.5869503021240234,
|
| 7669 |
+
"learning_rate": 2.7787924231021034e-05,
|
| 7670 |
+
"loss": 1.7067,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.09205779981014661,
|
| 7675 |
+
"grad_norm": 3.4459173679351807,
|
| 7676 |
+
"learning_rate": 2.7635804313469138e-05,
|
| 7677 |
+
"loss": 1.9242,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.09214217909503217,
|
| 7682 |
+
"grad_norm": 3.8406589031219482,
|
| 7683 |
+
"learning_rate": 2.7484035127894157e-05,
|
| 7684 |
+
"loss": 1.625,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.09222655837991774,
|
| 7689 |
+
"grad_norm": 2.9050941467285156,
|
| 7690 |
+
"learning_rate": 2.733261740988503e-05,
|
| 7691 |
+
"loss": 1.6394,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.09231093766480329,
|
| 7696 |
+
"grad_norm": 3.867025375366211,
|
| 7697 |
+
"learning_rate": 2.718155189332733e-05,
|
| 7698 |
+
"loss": 2.0617,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.09239531694968885,
|
| 7703 |
+
"grad_norm": 3.619429588317871,
|
| 7704 |
+
"learning_rate": 2.7030839310399546e-05,
|
| 7705 |
+
"loss": 1.7085,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.09247969623457442,
|
| 7710 |
+
"grad_norm": 3.1269400119781494,
|
| 7711 |
+
"learning_rate": 2.6880480391569528e-05,
|
| 7712 |
+
"loss": 1.6982,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.09256407551945997,
|
| 7717 |
+
"grad_norm": 4.308102607727051,
|
| 7718 |
+
"learning_rate": 2.67304758655911e-05,
|
| 7719 |
+
"loss": 1.762,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.09264845480434554,
|
| 7724 |
+
"grad_norm": 3.488555908203125,
|
| 7725 |
+
"learning_rate": 2.6580826459500353e-05,
|
| 7726 |
+
"loss": 1.5595,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.0927328340892311,
|
| 7731 |
+
"grad_norm": 2.2705681324005127,
|
| 7732 |
+
"learning_rate": 2.6431532898612277e-05,
|
| 7733 |
+
"loss": 1.3037,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.09281721337411665,
|
| 7738 |
+
"grad_norm": 3.364454746246338,
|
| 7739 |
+
"learning_rate": 2.6282595906517036e-05,
|
| 7740 |
+
"loss": 1.5788,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.09290159265900222,
|
| 7745 |
+
"grad_norm": 3.739013433456421,
|
| 7746 |
+
"learning_rate": 2.613401620507674e-05,
|
| 7747 |
+
"loss": 1.8872,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.09298597194388777,
|
| 7752 |
+
"grad_norm": 3.7140753269195557,
|
| 7753 |
+
"learning_rate": 2.5985794514421647e-05,
|
| 7754 |
+
"loss": 1.5118,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.09307035122877333,
|
| 7759 |
+
"grad_norm": 3.4308595657348633,
|
| 7760 |
+
"learning_rate": 2.5837931552946937e-05,
|
| 7761 |
+
"loss": 2.1785,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.0931547305136589,
|
| 7766 |
+
"grad_norm": 3.217737913131714,
|
| 7767 |
+
"learning_rate": 2.5690428037309078e-05,
|
| 7768 |
+
"loss": 1.5528,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.09323910979854445,
|
| 7773 |
+
"grad_norm": 3.415586233139038,
|
| 7774 |
+
"learning_rate": 2.554328468242242e-05,
|
| 7775 |
+
"loss": 1.6042,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.09332348908343002,
|
| 7780 |
+
"grad_norm": 3.3001484870910645,
|
| 7781 |
+
"learning_rate": 2.539650220145564e-05,
|
| 7782 |
+
"loss": 1.6438,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.09340786836831558,
|
| 7787 |
+
"grad_norm": 2.35314679145813,
|
| 7788 |
+
"learning_rate": 2.5250081305828332e-05,
|
| 7789 |
+
"loss": 1.1008,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.09349224765320113,
|
| 7794 |
+
"grad_norm": 3.022174596786499,
|
| 7795 |
+
"learning_rate": 2.5104022705207654e-05,
|
| 7796 |
+
"loss": 1.5253,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.0935766269380867,
|
| 7801 |
+
"grad_norm": 3.619746685028076,
|
| 7802 |
+
"learning_rate": 2.4958327107504763e-05,
|
| 7803 |
+
"loss": 1.8555,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.09366100622297226,
|
| 7808 |
+
"grad_norm": 2.6715095043182373,
|
| 7809 |
+
"learning_rate": 2.4812995218871437e-05,
|
| 7810 |
+
"loss": 1.2395,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.09374538550785783,
|
| 7815 |
+
"grad_norm": 2.9282262325286865,
|
| 7816 |
+
"learning_rate": 2.4668027743696654e-05,
|
| 7817 |
+
"loss": 1.1605,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.09382976479274338,
|
| 7822 |
+
"grad_norm": 3.5312087535858154,
|
| 7823 |
+
"learning_rate": 2.4523425384603094e-05,
|
| 7824 |
+
"loss": 1.8562,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.09391414407762894,
|
| 7829 |
+
"grad_norm": 3.0037548542022705,
|
| 7830 |
+
"learning_rate": 2.43791888424439e-05,
|
| 7831 |
+
"loss": 1.2937,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.0939985233625145,
|
| 7836 |
+
"grad_norm": 4.0344367027282715,
|
| 7837 |
+
"learning_rate": 2.4235318816299167e-05,
|
| 7838 |
+
"loss": 1.7686,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.09408290264740006,
|
| 7843 |
+
"grad_norm": 4.171757221221924,
|
| 7844 |
+
"learning_rate": 2.4091816003472522e-05,
|
| 7845 |
+
"loss": 1.9811,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.09416728193228563,
|
| 7850 |
+
"grad_norm": 3.382312059402466,
|
| 7851 |
+
"learning_rate": 2.3948681099487856e-05,
|
| 7852 |
+
"loss": 1.8441,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.09425166121717118,
|
| 7857 |
+
"grad_norm": 3.240097761154175,
|
| 7858 |
+
"learning_rate": 2.3805914798085848e-05,
|
| 7859 |
+
"loss": 1.295,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.09433604050205674,
|
| 7864 |
+
"grad_norm": 3.080533266067505,
|
| 7865 |
+
"learning_rate": 2.3663517791220667e-05,
|
| 7866 |
+
"loss": 1.4059,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.09442041978694231,
|
| 7871 |
+
"grad_norm": 3.4532551765441895,
|
| 7872 |
+
"learning_rate": 2.352149076905663e-05,
|
| 7873 |
+
"loss": 1.3967,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.09450479907182786,
|
| 7878 |
+
"grad_norm": 2.838005304336548,
|
| 7879 |
+
"learning_rate": 2.3379834419964786e-05,
|
| 7880 |
+
"loss": 1.3161,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.09458917835671343,
|
| 7885 |
+
"grad_norm": 3.1617472171783447,
|
| 7886 |
+
"learning_rate": 2.3238549430519663e-05,
|
| 7887 |
+
"loss": 1.7358,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.09467355764159899,
|
| 7892 |
+
"grad_norm": 3.414036989212036,
|
| 7893 |
+
"learning_rate": 2.3097636485495832e-05,
|
| 7894 |
+
"loss": 1.5305,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.09475793692648454,
|
| 7899 |
+
"grad_norm": 3.2192142009735107,
|
| 7900 |
+
"learning_rate": 2.2957096267864676e-05,
|
| 7901 |
+
"loss": 1.6985,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.09484231621137011,
|
| 7906 |
+
"grad_norm": 3.579663038253784,
|
| 7907 |
+
"learning_rate": 2.281692945879109e-05,
|
| 7908 |
+
"loss": 1.4448,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.09492669549625567,
|
| 7913 |
+
"grad_norm": 3.8103861808776855,
|
| 7914 |
+
"learning_rate": 2.2677136737630123e-05,
|
| 7915 |
+
"loss": 1.3404,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.09501107478114124,
|
| 7920 |
+
"grad_norm": 4.645077705383301,
|
| 7921 |
+
"learning_rate": 2.2537718781923756e-05,
|
| 7922 |
+
"loss": 2.1606,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.09509545406602679,
|
| 7927 |
+
"grad_norm": 3.2441060543060303,
|
| 7928 |
+
"learning_rate": 2.2398676267397457e-05,
|
| 7929 |
+
"loss": 1.4034,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.09517983335091235,
|
| 7934 |
+
"grad_norm": 3.3004987239837646,
|
| 7935 |
+
"learning_rate": 2.2260009867957167e-05,
|
| 7936 |
+
"loss": 2.0345,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.09526421263579792,
|
| 7941 |
+
"grad_norm": 3.6183228492736816,
|
| 7942 |
+
"learning_rate": 2.212172025568574e-05,
|
| 7943 |
+
"loss": 1.6312,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.09534859192068347,
|
| 7948 |
+
"grad_norm": 4.666016578674316,
|
| 7949 |
+
"learning_rate": 2.1983808100839943e-05,
|
| 7950 |
+
"loss": 1.5116,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.09543297120556904,
|
| 7955 |
+
"grad_norm": 2.3226547241210938,
|
| 7956 |
+
"learning_rate": 2.184627407184705e-05,
|
| 7957 |
+
"loss": 1.1591,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.0955173504904546,
|
| 7962 |
+
"grad_norm": 3.2942864894866943,
|
| 7963 |
+
"learning_rate": 2.1709118835301688e-05,
|
| 7964 |
+
"loss": 1.4112,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.09560172977534015,
|
| 7969 |
+
"grad_norm": 2.986431360244751,
|
| 7970 |
+
"learning_rate": 2.1572343055962485e-05,
|
| 7971 |
+
"loss": 1.4647,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.09568610906022572,
|
| 7976 |
+
"grad_norm": 2.305901050567627,
|
| 7977 |
+
"learning_rate": 2.1435947396749024e-05,
|
| 7978 |
+
"loss": 1.0954,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.09577048834511127,
|
| 7983 |
+
"grad_norm": 3.060244560241699,
|
| 7984 |
+
"learning_rate": 2.1299932518738498e-05,
|
| 7985 |
+
"loss": 1.2653,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.09585486762999683,
|
| 7990 |
+
"grad_norm": 3.3505055904388428,
|
| 7991 |
+
"learning_rate": 2.1164299081162498e-05,
|
| 7992 |
+
"loss": 1.6929,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.0959392469148824,
|
| 7997 |
+
"grad_norm": 3.0686864852905273,
|
| 7998 |
+
"learning_rate": 2.1029047741403975e-05,
|
| 7999 |
+
"loss": 1.5366,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.09602362619976795,
|
| 8004 |
+
"grad_norm": 3.4930968284606934,
|
| 8005 |
+
"learning_rate": 2.089417915499381e-05,
|
| 8006 |
+
"loss": 1.7365,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.09610800548465352,
|
| 8011 |
+
"grad_norm": 3.2524631023406982,
|
| 8012 |
+
"learning_rate": 2.075969397560785e-05,
|
| 8013 |
+
"loss": 1.4176,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.09619238476953908,
|
| 8018 |
+
"grad_norm": 3.2085278034210205,
|
| 8019 |
+
"learning_rate": 2.062559285506368e-05,
|
| 8020 |
+
"loss": 1.6044,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.09627676405442463,
|
| 8025 |
+
"grad_norm": 3.602881669998169,
|
| 8026 |
+
"learning_rate": 2.049187644331738e-05,
|
| 8027 |
+
"loss": 1.6562,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.0963611433393102,
|
| 8032 |
+
"grad_norm": 3.3112502098083496,
|
| 8033 |
+
"learning_rate": 2.0358545388460516e-05,
|
| 8034 |
+
"loss": 1.6472,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.09644552262419576,
|
| 8039 |
+
"grad_norm": 3.7688615322113037,
|
| 8040 |
+
"learning_rate": 2.022560033671683e-05,
|
| 8041 |
+
"loss": 1.9564,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.09652990190908133,
|
| 8046 |
+
"grad_norm": 3.2180838584899902,
|
| 8047 |
+
"learning_rate": 2.0093041932439226e-05,
|
| 8048 |
+
"loss": 1.5863,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.09661428119396688,
|
| 8053 |
+
"grad_norm": 2.982928991317749,
|
| 8054 |
+
"learning_rate": 1.996087081810666e-05,
|
| 8055 |
+
"loss": 1.2345,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.09669866047885244,
|
| 8060 |
+
"grad_norm": 3.0494682788848877,
|
| 8061 |
+
"learning_rate": 1.9829087634320952e-05,
|
| 8062 |
+
"loss": 1.5807,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.096783039763738,
|
| 8067 |
+
"grad_norm": 3.110319137573242,
|
| 8068 |
+
"learning_rate": 1.9697693019803776e-05,
|
| 8069 |
+
"loss": 1.577,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.09686741904862356,
|
| 8074 |
+
"grad_norm": 3.3014461994171143,
|
| 8075 |
+
"learning_rate": 1.9566687611393373e-05,
|
| 8076 |
+
"loss": 1.8186,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.09695179833350913,
|
| 8081 |
+
"grad_norm": 2.3945372104644775,
|
| 8082 |
+
"learning_rate": 1.943607204404173e-05,
|
| 8083 |
+
"loss": 0.9261,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.09703617761839468,
|
| 8088 |
+
"grad_norm": 2.712374687194824,
|
| 8089 |
+
"learning_rate": 1.9305846950811312e-05,
|
| 8090 |
+
"loss": 1.4963,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.09712055690328024,
|
| 8095 |
+
"grad_norm": 3.8724498748779297,
|
| 8096 |
+
"learning_rate": 1.9176012962872015e-05,
|
| 8097 |
+
"loss": 1.6943,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.09720493618816581,
|
| 8102 |
+
"grad_norm": 3.7097952365875244,
|
| 8103 |
+
"learning_rate": 1.90465707094982e-05,
|
| 8104 |
+
"loss": 1.5224,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.09728931547305136,
|
| 8109 |
+
"grad_norm": 4.344565391540527,
|
| 8110 |
+
"learning_rate": 1.891752081806557e-05,
|
| 8111 |
+
"loss": 1.4517,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.09737369475793693,
|
| 8116 |
+
"grad_norm": 2.983797550201416,
|
| 8117 |
+
"learning_rate": 1.8788863914048105e-05,
|
| 8118 |
+
"loss": 1.8816,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.09745807404282249,
|
| 8123 |
+
"grad_norm": 4.158752918243408,
|
| 8124 |
+
"learning_rate": 1.8660600621015113e-05,
|
| 8125 |
+
"loss": 1.8136,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.09754245332770804,
|
| 8130 |
+
"grad_norm": 3.6239845752716064,
|
| 8131 |
+
"learning_rate": 1.8532731560628148e-05,
|
| 8132 |
+
"loss": 1.9316,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.09762683261259361,
|
| 8137 |
+
"grad_norm": 2.62990665435791,
|
| 8138 |
+
"learning_rate": 1.8405257352638028e-05,
|
| 8139 |
+
"loss": 1.4178,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.09771121189747917,
|
| 8144 |
+
"grad_norm": 3.1281239986419678,
|
| 8145 |
+
"learning_rate": 1.827817861488178e-05,
|
| 8146 |
+
"loss": 1.3878,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.09779559118236474,
|
| 8151 |
+
"grad_norm": 2.6983275413513184,
|
| 8152 |
+
"learning_rate": 1.8151495963279695e-05,
|
| 8153 |
+
"loss": 1.6486,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.09787997046725029,
|
| 8158 |
+
"grad_norm": 3.463437795639038,
|
| 8159 |
+
"learning_rate": 1.8025210011832327e-05,
|
| 8160 |
+
"loss": 1.7887,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.09796434975213585,
|
| 8165 |
+
"grad_norm": 3.5641868114471436,
|
| 8166 |
+
"learning_rate": 1.789932137261756e-05,
|
| 8167 |
+
"loss": 1.7378,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.09804872903702141,
|
| 8172 |
+
"grad_norm": 2.3257901668548584,
|
| 8173 |
+
"learning_rate": 1.777383065578755e-05,
|
| 8174 |
+
"loss": 1.1249,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.09813310832190697,
|
| 8179 |
+
"grad_norm": 4.11950159072876,
|
| 8180 |
+
"learning_rate": 1.764873846956585e-05,
|
| 8181 |
+
"loss": 1.6153,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.09821748760679254,
|
| 8186 |
+
"grad_norm": 3.400200366973877,
|
| 8187 |
+
"learning_rate": 1.7524045420244416e-05,
|
| 8188 |
+
"loss": 1.5002,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.0983018668916781,
|
| 8193 |
+
"grad_norm": 2.9896585941314697,
|
| 8194 |
+
"learning_rate": 1.7399752112180645e-05,
|
| 8195 |
+
"loss": 1.5213,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.09838624617656365,
|
| 8200 |
+
"grad_norm": 3.7832117080688477,
|
| 8201 |
+
"learning_rate": 1.727585914779455e-05,
|
| 8202 |
+
"loss": 1.7518,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.09847062546144922,
|
| 8207 |
+
"grad_norm": 3.106523036956787,
|
| 8208 |
+
"learning_rate": 1.715236712756575e-05,
|
| 8209 |
+
"loss": 1.1785,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.09855500474633477,
|
| 8214 |
+
"grad_norm": 2.8863906860351562,
|
| 8215 |
+
"learning_rate": 1.702927665003059e-05,
|
| 8216 |
+
"loss": 1.3234,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.09863938403122033,
|
| 8221 |
+
"grad_norm": 2.6570043563842773,
|
| 8222 |
+
"learning_rate": 1.690658831177917e-05,
|
| 8223 |
+
"loss": 1.1568,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.0987237633161059,
|
| 8228 |
+
"grad_norm": 2.9703497886657715,
|
| 8229 |
+
"learning_rate": 1.6784302707452614e-05,
|
| 8230 |
+
"loss": 1.541,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.09880814260099145,
|
| 8235 |
+
"grad_norm": 3.341186046600342,
|
| 8236 |
+
"learning_rate": 1.6662420429740044e-05,
|
| 8237 |
+
"loss": 1.64,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.09889252188587702,
|
| 8242 |
+
"grad_norm": 3.65065598487854,
|
| 8243 |
+
"learning_rate": 1.6540942069375764e-05,
|
| 8244 |
+
"loss": 2.1576,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.09897690117076258,
|
| 8249 |
+
"grad_norm": 2.9629430770874023,
|
| 8250 |
+
"learning_rate": 1.6419868215136337e-05,
|
| 8251 |
+
"loss": 1.2974,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.09906128045564813,
|
| 8256 |
+
"grad_norm": 3.012216567993164,
|
| 8257 |
+
"learning_rate": 1.6299199453837877e-05,
|
| 8258 |
+
"loss": 1.4309,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.0991456597405337,
|
| 8263 |
+
"grad_norm": 2.6822774410247803,
|
| 8264 |
+
"learning_rate": 1.617893637033301e-05,
|
| 8265 |
+
"loss": 1.4372,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.09923003902541926,
|
| 8270 |
+
"grad_norm": 3.7954866886138916,
|
| 8271 |
+
"learning_rate": 1.6059079547508195e-05,
|
| 8272 |
+
"loss": 1.6665,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.09931441831030483,
|
| 8277 |
+
"grad_norm": 3.383831024169922,
|
| 8278 |
+
"learning_rate": 1.5939629566280846e-05,
|
| 8279 |
+
"loss": 1.6414,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.09939879759519038,
|
| 8284 |
+
"grad_norm": 3.5251333713531494,
|
| 8285 |
+
"learning_rate": 1.582058700559652e-05,
|
| 8286 |
+
"loss": 1.5191,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.09948317688007594,
|
| 8291 |
+
"grad_norm": 2.910231351852417,
|
| 8292 |
+
"learning_rate": 1.570195244242605e-05,
|
| 8293 |
+
"loss": 1.2948,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.0995675561649615,
|
| 8298 |
+
"grad_norm": 2.6657936573028564,
|
| 8299 |
+
"learning_rate": 1.5583726451762826e-05,
|
| 8300 |
+
"loss": 1.2141,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.09965193544984706,
|
| 8305 |
+
"grad_norm": 2.812673807144165,
|
| 8306 |
+
"learning_rate": 1.546590960662e-05,
|
| 8307 |
+
"loss": 1.3571,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.09973631473473263,
|
| 8312 |
+
"grad_norm": 2.73283052444458,
|
| 8313 |
+
"learning_rate": 1.5348502478027705e-05,
|
| 8314 |
+
"loss": 1.2243,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.09982069401961818,
|
| 8319 |
+
"grad_norm": 3.947868585586548,
|
| 8320 |
+
"learning_rate": 1.5231505635030252e-05,
|
| 8321 |
+
"loss": 2.1057,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.09990507330450374,
|
| 8326 |
+
"grad_norm": 3.3415377140045166,
|
| 8327 |
+
"learning_rate": 1.5114919644683412e-05,
|
| 8328 |
+
"loss": 1.5468,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.09998945258938931,
|
| 8333 |
+
"grad_norm": 3.676685333251953,
|
| 8334 |
+
"learning_rate": 1.4998745072051612e-05,
|
| 8335 |
+
"loss": 1.6158,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.10007383187427486,
|
| 8340 |
+
"grad_norm": 3.642958879470825,
|
| 8341 |
+
"learning_rate": 1.488298248020531e-05,
|
| 8342 |
+
"loss": 2.1226,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.10015821115916043,
|
| 8347 |
+
"grad_norm": 4.086678981781006,
|
| 8348 |
+
"learning_rate": 1.4767632430218081e-05,
|
| 8349 |
+
"loss": 1.9566,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.10024259044404599,
|
| 8354 |
+
"grad_norm": 2.6297688484191895,
|
| 8355 |
+
"learning_rate": 1.4652695481164124e-05,
|
| 8356 |
+
"loss": 1.5169,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.10032696972893154,
|
| 8361 |
+
"grad_norm": 3.743318796157837,
|
| 8362 |
+
"learning_rate": 1.4538172190115384e-05,
|
| 8363 |
+
"loss": 1.7759,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.10041134901381711,
|
| 8368 |
+
"grad_norm": 2.8840904235839844,
|
| 8369 |
+
"learning_rate": 1.4424063112138875e-05,
|
| 8370 |
+
"loss": 1.272,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.10049572829870267,
|
| 8375 |
+
"grad_norm": 3.2887556552886963,
|
| 8376 |
+
"learning_rate": 1.4310368800294071e-05,
|
| 8377 |
+
"loss": 1.3963,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.10058010758358824,
|
| 8382 |
+
"grad_norm": 3.0418412685394287,
|
| 8383 |
+
"learning_rate": 1.4197089805630148e-05,
|
| 8384 |
+
"loss": 1.4471,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.10066448686847379,
|
| 8389 |
+
"grad_norm": 3.2652742862701416,
|
| 8390 |
+
"learning_rate": 1.4084226677183354e-05,
|
| 8391 |
+
"loss": 1.2907,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.10074886615335935,
|
| 8396 |
+
"grad_norm": 2.910196304321289,
|
| 8397 |
+
"learning_rate": 1.3971779961974329e-05,
|
| 8398 |
+
"loss": 1.8535,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.10083324543824491,
|
| 8403 |
+
"grad_norm": 2.948594331741333,
|
| 8404 |
+
"learning_rate": 1.385975020500544e-05,
|
| 8405 |
+
"loss": 1.3961,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.10091762472313047,
|
| 8410 |
+
"grad_norm": 3.0885612964630127,
|
| 8411 |
+
"learning_rate": 1.3748137949258132e-05,
|
| 8412 |
+
"loss": 1.4577,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.10100200400801604,
|
| 8417 |
+
"grad_norm": 3.5346765518188477,
|
| 8418 |
+
"learning_rate": 1.36369437356904e-05,
|
| 8419 |
+
"loss": 1.205,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.1010863832929016,
|
| 8424 |
+
"grad_norm": 3.4392857551574707,
|
| 8425 |
+
"learning_rate": 1.3526168103234028e-05,
|
| 8426 |
+
"loss": 1.6209,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.10117076257778715,
|
| 8431 |
+
"grad_norm": 3.295177936553955,
|
| 8432 |
+
"learning_rate": 1.3415811588792093e-05,
|
| 8433 |
+
"loss": 1.4999,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.10125514186267272,
|
| 8438 |
+
"grad_norm": 3.7289631366729736,
|
| 8439 |
+
"learning_rate": 1.3305874727236201e-05,
|
| 8440 |
+
"loss": 1.7763,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.10133952114755827,
|
| 8445 |
+
"grad_norm": 3.2035927772521973,
|
| 8446 |
+
"learning_rate": 1.3196358051404146e-05,
|
| 8447 |
+
"loss": 1.3206,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.10142390043244383,
|
| 8452 |
+
"grad_norm": 3.132443904876709,
|
| 8453 |
+
"learning_rate": 1.308726209209703e-05,
|
| 8454 |
+
"loss": 1.5426,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.1015082797173294,
|
| 8459 |
+
"grad_norm": 3.4239304065704346,
|
| 8460 |
+
"learning_rate": 1.2978587378076945e-05,
|
| 8461 |
+
"loss": 1.4321,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.10159265900221495,
|
| 8466 |
+
"grad_norm": 3.284937858581543,
|
| 8467 |
+
"learning_rate": 1.287033443606429e-05,
|
| 8468 |
+
"loss": 1.3087,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.10167703828710052,
|
| 8473 |
+
"grad_norm": 3.1208696365356445,
|
| 8474 |
+
"learning_rate": 1.2762503790735248e-05,
|
| 8475 |
+
"loss": 1.5838,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.10176141757198608,
|
| 8480 |
+
"grad_norm": 3.225219488143921,
|
| 8481 |
+
"learning_rate": 1.2655095964719154e-05,
|
| 8482 |
+
"loss": 1.6095,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.10184579685687163,
|
| 8487 |
+
"grad_norm": 2.9957473278045654,
|
| 8488 |
+
"learning_rate": 1.2548111478596125e-05,
|
| 8489 |
+
"loss": 1.5329,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.1019301761417572,
|
| 8494 |
+
"grad_norm": 3.315237045288086,
|
| 8495 |
+
"learning_rate": 1.2441550850894434e-05,
|
| 8496 |
+
"loss": 1.5265,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.10201455542664276,
|
| 8501 |
+
"grad_norm": 3.1461193561553955,
|
| 8502 |
+
"learning_rate": 1.2335414598087957e-05,
|
| 8503 |
+
"loss": 1.4827,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.10209893471152832,
|
| 8508 |
+
"grad_norm": 2.751537561416626,
|
| 8509 |
+
"learning_rate": 1.2229703234593814e-05,
|
| 8510 |
+
"loss": 1.1396,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.10218331399641388,
|
| 8515 |
+
"grad_norm": 3.5919241905212402,
|
| 8516 |
+
"learning_rate": 1.2124417272769695e-05,
|
| 8517 |
+
"loss": 1.341,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.10226769328129943,
|
| 8522 |
+
"grad_norm": 3.240518569946289,
|
| 8523 |
+
"learning_rate": 1.2019557222911538e-05,
|
| 8524 |
+
"loss": 1.6799,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.102352072566185,
|
| 8529 |
+
"grad_norm": 3.2598378658294678,
|
| 8530 |
+
"learning_rate": 1.1915123593250955e-05,
|
| 8531 |
+
"loss": 1.6434,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.10243645185107056,
|
| 8536 |
+
"grad_norm": 3.110367774963379,
|
| 8537 |
+
"learning_rate": 1.1811116889952833e-05,
|
| 8538 |
+
"loss": 1.4139,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.10252083113595613,
|
| 8543 |
+
"grad_norm": 2.530456304550171,
|
| 8544 |
+
"learning_rate": 1.1707537617112829e-05,
|
| 8545 |
+
"loss": 1.2431,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.10260521042084168,
|
| 8550 |
+
"grad_norm": 2.9934744834899902,
|
| 8551 |
+
"learning_rate": 1.160438627675493e-05,
|
| 8552 |
+
"loss": 1.4463,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.10268958970572724,
|
| 8557 |
+
"grad_norm": 3.5974788665771484,
|
| 8558 |
+
"learning_rate": 1.150166336882904e-05,
|
| 8559 |
+
"loss": 1.1886,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.10277396899061281,
|
| 8564 |
+
"grad_norm": 3.687152624130249,
|
| 8565 |
+
"learning_rate": 1.139936939120858e-05,
|
| 8566 |
+
"loss": 1.8659,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.10285834827549836,
|
| 8571 |
+
"grad_norm": 3.481801748275757,
|
| 8572 |
+
"learning_rate": 1.1297504839688034e-05,
|
| 8573 |
+
"loss": 1.9201,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.10294272756038393,
|
| 8578 |
+
"grad_norm": 3.397308588027954,
|
| 8579 |
+
"learning_rate": 1.1196070207980569e-05,
|
| 8580 |
+
"loss": 1.7231,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.10302710684526949,
|
| 8585 |
+
"grad_norm": 3.4461331367492676,
|
| 8586 |
+
"learning_rate": 1.1095065987715591e-05,
|
| 8587 |
+
"loss": 1.621,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.10311148613015504,
|
| 8592 |
+
"grad_norm": 2.756736993789673,
|
| 8593 |
+
"learning_rate": 1.0994492668436463e-05,
|
| 8594 |
+
"loss": 1.1307,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.10319586541504061,
|
| 8599 |
+
"grad_norm": 2.8683910369873047,
|
| 8600 |
+
"learning_rate": 1.0894350737598058e-05,
|
| 8601 |
+
"loss": 1.4611,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.10328024469992617,
|
| 8606 |
+
"grad_norm": 3.279130697250366,
|
| 8607 |
+
"learning_rate": 1.0794640680564372e-05,
|
| 8608 |
+
"loss": 1.4581,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.10336462398481173,
|
| 8613 |
+
"grad_norm": 3.0329201221466064,
|
| 8614 |
+
"learning_rate": 1.0695362980606249e-05,
|
| 8615 |
+
"loss": 1.4492,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.10344900326969729,
|
| 8620 |
+
"grad_norm": 3.720149278640747,
|
| 8621 |
+
"learning_rate": 1.0596518118899035e-05,
|
| 8622 |
+
"loss": 1.5345,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.10353338255458285,
|
| 8627 |
+
"grad_norm": 2.57875919342041,
|
| 8628 |
+
"learning_rate": 1.0498106574520105e-05,
|
| 8629 |
+
"loss": 1.0971,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.10361776183946841,
|
| 8634 |
+
"grad_norm": 3.4571704864501953,
|
| 8635 |
+
"learning_rate": 1.0400128824446753e-05,
|
| 8636 |
+
"loss": 1.429,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.10370214112435397,
|
| 8641 |
+
"grad_norm": 3.012058973312378,
|
| 8642 |
+
"learning_rate": 1.0302585343553739e-05,
|
| 8643 |
+
"loss": 1.3748,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.10378652040923954,
|
| 8648 |
+
"grad_norm": 3.26574444770813,
|
| 8649 |
+
"learning_rate": 1.0205476604611042e-05,
|
| 8650 |
+
"loss": 1.5926,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.1038708996941251,
|
| 8655 |
+
"grad_norm": 2.778719425201416,
|
| 8656 |
+
"learning_rate": 1.0108803078281503e-05,
|
| 8657 |
+
"loss": 1.3482,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.10395527897901065,
|
| 8662 |
+
"grad_norm": 2.932727098464966,
|
| 8663 |
+
"learning_rate": 1.0012565233118597e-05,
|
| 8664 |
+
"loss": 1.0628,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.10403965826389622,
|
| 8669 |
+
"grad_norm": 3.3466224670410156,
|
| 8670 |
+
"learning_rate": 9.916763535564221e-06,
|
| 8671 |
+
"loss": 1.7522,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.10412403754878177,
|
| 8676 |
+
"grad_norm": 3.4813435077667236,
|
| 8677 |
+
"learning_rate": 9.821398449946306e-06,
|
| 8678 |
+
"loss": 1.6653,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.10420841683366733,
|
| 8683 |
+
"grad_norm": 2.8772964477539062,
|
| 8684 |
+
"learning_rate": 9.726470438476653e-06,
|
| 8685 |
+
"loss": 1.2617,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.1042927961185529,
|
| 8690 |
+
"grad_norm": 3.323155403137207,
|
| 8691 |
+
"learning_rate": 9.631979961248694e-06,
|
| 8692 |
+
"loss": 1.6244,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.10437717540343845,
|
| 8697 |
+
"grad_norm": 2.973639965057373,
|
| 8698 |
+
"learning_rate": 9.537927476235175e-06,
|
| 8699 |
+
"loss": 1.5841,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.10446155468832402,
|
| 8704 |
+
"grad_norm": 3.7626113891601562,
|
| 8705 |
+
"learning_rate": 9.444313439286034e-06,
|
| 8706 |
+
"loss": 1.8155,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.10454593397320958,
|
| 8711 |
+
"grad_norm": 2.8624961376190186,
|
| 8712 |
+
"learning_rate": 9.351138304126161e-06,
|
| 8713 |
+
"loss": 1.3577,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.10463031325809513,
|
| 8718 |
+
"grad_norm": 2.4197492599487305,
|
| 8719 |
+
"learning_rate": 9.258402522353182e-06,
|
| 8720 |
+
"loss": 1.0776,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.1047146925429807,
|
| 8725 |
+
"grad_norm": 2.837425947189331,
|
| 8726 |
+
"learning_rate": 9.166106543435316e-06,
|
| 8727 |
+
"loss": 1.3218,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.10479907182786626,
|
| 8732 |
+
"grad_norm": 3.0032570362091064,
|
| 8733 |
+
"learning_rate": 9.074250814709095e-06,
|
| 8734 |
+
"loss": 1.3004,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.10488345111275182,
|
| 8739 |
+
"grad_norm": 2.6306238174438477,
|
| 8740 |
+
"learning_rate": 8.982835781377318e-06,
|
| 8741 |
+
"loss": 1.1977,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.10496783039763738,
|
| 8746 |
+
"grad_norm": 3.238788604736328,
|
| 8747 |
+
"learning_rate": 8.891861886506824e-06,
|
| 8748 |
+
"loss": 1.815,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.10505220968252293,
|
| 8753 |
+
"grad_norm": 3.901887893676758,
|
| 8754 |
+
"learning_rate": 8.801329571026373e-06,
|
| 8755 |
+
"loss": 1.7613,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.1051365889674085,
|
| 8760 |
+
"grad_norm": 2.458286762237549,
|
| 8761 |
+
"learning_rate": 8.711239273724447e-06,
|
| 8762 |
+
"loss": 1.4532,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.10522096825229406,
|
| 8767 |
+
"grad_norm": 2.9087612628936768,
|
| 8768 |
+
"learning_rate": 8.621591431247244e-06,
|
| 8769 |
+
"loss": 1.111,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.10530534753717963,
|
| 8774 |
+
"grad_norm": 3.07930588722229,
|
| 8775 |
+
"learning_rate": 8.532386478096398e-06,
|
| 8776 |
+
"loss": 1.7926,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.10538972682206518,
|
| 8781 |
+
"grad_norm": 3.5351290702819824,
|
| 8782 |
+
"learning_rate": 8.443624846627063e-06,
|
| 8783 |
+
"loss": 1.7935,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.10547410610695074,
|
| 8788 |
+
"grad_norm": 5.294930934906006,
|
| 8789 |
+
"learning_rate": 8.35530696704564e-06,
|
| 8790 |
+
"loss": 1.6532,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.10555848539183631,
|
| 8795 |
+
"grad_norm": 3.089399814605713,
|
| 8796 |
+
"learning_rate": 8.267433267407842e-06,
|
| 8797 |
+
"loss": 1.4275,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.10564286467672186,
|
| 8802 |
+
"grad_norm": 2.679612636566162,
|
| 8803 |
+
"learning_rate": 8.180004173616484e-06,
|
| 8804 |
+
"loss": 1.0819,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.10572724396160743,
|
| 8809 |
+
"grad_norm": 2.798283815383911,
|
| 8810 |
+
"learning_rate": 8.09302010941949e-06,
|
| 8811 |
+
"loss": 1.4165,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.10581162324649299,
|
| 8816 |
+
"grad_norm": 3.1109769344329834,
|
| 8817 |
+
"learning_rate": 8.006481496407858e-06,
|
| 8818 |
+
"loss": 1.1018,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.10589600253137854,
|
| 8823 |
+
"grad_norm": 3.1817078590393066,
|
| 8824 |
+
"learning_rate": 7.920388754013597e-06,
|
| 8825 |
+
"loss": 1.3661,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.10598038181626411,
|
| 8830 |
+
"grad_norm": 3.6453311443328857,
|
| 8831 |
+
"learning_rate": 7.834742299507658e-06,
|
| 8832 |
+
"loss": 2.0217,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.10606476110114967,
|
| 8837 |
+
"grad_norm": 2.8965976238250732,
|
| 8838 |
+
"learning_rate": 7.749542547997968e-06,
|
| 8839 |
+
"loss": 1.1904,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.10614914038603523,
|
| 8844 |
+
"grad_norm": 3.768522024154663,
|
| 8845 |
+
"learning_rate": 7.664789912427362e-06,
|
| 8846 |
+
"loss": 2.0709,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.10623351967092079,
|
| 8851 |
+
"grad_norm": 2.6715152263641357,
|
| 8852 |
+
"learning_rate": 7.580484803571641e-06,
|
| 8853 |
+
"loss": 1.4802,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.10631789895580634,
|
| 8858 |
+
"grad_norm": 3.661702871322632,
|
| 8859 |
+
"learning_rate": 7.496627630037523e-06,
|
| 8860 |
+
"loss": 1.5005,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.10640227824069191,
|
| 8865 |
+
"grad_norm": 2.863940477371216,
|
| 8866 |
+
"learning_rate": 7.413218798260702e-06,
|
| 8867 |
+
"loss": 1.3796,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.10648665752557747,
|
| 8872 |
+
"grad_norm": 3.5947470664978027,
|
| 8873 |
+
"learning_rate": 7.33025871250389e-06,
|
| 8874 |
+
"loss": 1.7538,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.10657103681046304,
|
| 8879 |
+
"grad_norm": 3.535916566848755,
|
| 8880 |
+
"learning_rate": 7.247747774854796e-06,
|
| 8881 |
+
"loss": 1.3744,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.10665541609534859,
|
| 8886 |
+
"grad_norm": 3.8513424396514893,
|
| 8887 |
+
"learning_rate": 7.165686385224224e-06,
|
| 8888 |
+
"loss": 1.9813,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.10673979538023415,
|
| 8893 |
+
"grad_norm": 2.9487640857696533,
|
| 8894 |
+
"learning_rate": 7.084074941344155e-06,
|
| 8895 |
+
"loss": 1.0871,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.10682417466511972,
|
| 8900 |
+
"grad_norm": 3.1807610988616943,
|
| 8901 |
+
"learning_rate": 7.002913838765779e-06,
|
| 8902 |
+
"loss": 1.5686,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.10690855395000527,
|
| 8907 |
+
"grad_norm": 2.9313714504241943,
|
| 8908 |
+
"learning_rate": 6.922203470857536e-06,
|
| 8909 |
+
"loss": 1.4564,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.10699293323489083,
|
| 8914 |
+
"grad_norm": 2.8799333572387695,
|
| 8915 |
+
"learning_rate": 6.8419442288033544e-06,
|
| 8916 |
+
"loss": 1.437,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.1070773125197764,
|
| 8921 |
+
"grad_norm": 3.8256032466888428,
|
| 8922 |
+
"learning_rate": 6.76213650160058e-06,
|
| 8923 |
+
"loss": 1.4882,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.10716169180466195,
|
| 8928 |
+
"grad_norm": 3.028865337371826,
|
| 8929 |
+
"learning_rate": 6.68278067605822e-06,
|
| 8930 |
+
"loss": 1.199,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.10724607108954752,
|
| 8935 |
+
"grad_norm": 3.519663095474243,
|
| 8936 |
+
"learning_rate": 6.603877136795022e-06,
|
| 8937 |
+
"loss": 1.5687,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.10733045037443308,
|
| 8942 |
+
"grad_norm": 2.9813034534454346,
|
| 8943 |
+
"learning_rate": 6.5254262662376e-06,
|
| 8944 |
+
"loss": 1.189,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.10741482965931863,
|
| 8949 |
+
"grad_norm": 3.2407150268554688,
|
| 8950 |
+
"learning_rate": 6.447428444618575e-06,
|
| 8951 |
+
"loss": 1.8209,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.1074992089442042,
|
| 8956 |
+
"grad_norm": 3.319223642349243,
|
| 8957 |
+
"learning_rate": 6.369884049974784e-06,
|
| 8958 |
+
"loss": 1.807,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.10758358822908975,
|
| 8963 |
+
"grad_norm": 3.135756731033325,
|
| 8964 |
+
"learning_rate": 6.292793458145374e-06,
|
| 8965 |
+
"loss": 1.914,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.10766796751397532,
|
| 8970 |
+
"grad_norm": 3.1744773387908936,
|
| 8971 |
+
"learning_rate": 6.216157042770032e-06,
|
| 8972 |
+
"loss": 1.6731,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.10775234679886088,
|
| 8977 |
+
"grad_norm": 3.0961568355560303,
|
| 8978 |
+
"learning_rate": 6.139975175287171e-06,
|
| 8979 |
+
"loss": 1.5665,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.10783672608374643,
|
| 8984 |
+
"grad_norm": 3.098339557647705,
|
| 8985 |
+
"learning_rate": 6.064248224932134e-06,
|
| 8986 |
+
"loss": 1.1993,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.107921105368632,
|
| 8991 |
+
"grad_norm": 3.588793992996216,
|
| 8992 |
+
"learning_rate": 5.988976558735327e-06,
|
| 8993 |
+
"loss": 1.6308,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.10800548465351756,
|
| 8998 |
+
"grad_norm": 3.139782190322876,
|
| 8999 |
+
"learning_rate": 5.9141605415205525e-06,
|
| 9000 |
+
"loss": 1.6942,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.10808986393840313,
|
| 9005 |
+
"grad_norm": 3.4625742435455322,
|
| 9006 |
+
"learning_rate": 5.839800535903217e-06,
|
| 9007 |
+
"loss": 1.2796,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.10817424322328868,
|
| 9012 |
+
"grad_norm": 3.1997220516204834,
|
| 9013 |
+
"learning_rate": 5.76589690228847e-06,
|
| 9014 |
+
"loss": 1.5741,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.10825862250817424,
|
| 9019 |
+
"grad_norm": 2.9986748695373535,
|
| 9020 |
+
"learning_rate": 5.692449998869598e-06,
|
| 9021 |
+
"loss": 1.1792,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.1083430017930598,
|
| 9026 |
+
"grad_norm": 3.4218549728393555,
|
| 9027 |
+
"learning_rate": 5.61946018162618e-06,
|
| 9028 |
+
"loss": 1.6594,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.10842738107794536,
|
| 9033 |
+
"grad_norm": 2.927212715148926,
|
| 9034 |
+
"learning_rate": 5.546927804322455e-06,
|
| 9035 |
+
"loss": 1.4203,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.10851176036283093,
|
| 9040 |
+
"grad_norm": 3.106448173522949,
|
| 9041 |
+
"learning_rate": 5.474853218505515e-06,
|
| 9042 |
+
"loss": 1.574,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.10859613964771649,
|
| 9047 |
+
"grad_norm": 3.5189123153686523,
|
| 9048 |
+
"learning_rate": 5.403236773503695e-06,
|
| 9049 |
+
"loss": 2.0111,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.10868051893260204,
|
| 9054 |
+
"grad_norm": 3.392515182495117,
|
| 9055 |
+
"learning_rate": 5.332078816424802e-06,
|
| 9056 |
+
"loss": 1.5574,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.10876489821748761,
|
| 9061 |
+
"grad_norm": 2.9669435024261475,
|
| 9062 |
+
"learning_rate": 5.261379692154433e-06,
|
| 9063 |
+
"loss": 1.27,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.10884927750237317,
|
| 9068 |
+
"grad_norm": 2.9603822231292725,
|
| 9069 |
+
"learning_rate": 5.191139743354378e-06,
|
| 9070 |
+
"loss": 1.2592,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.10893365678725873,
|
| 9075 |
+
"grad_norm": 3.3389413356781006,
|
| 9076 |
+
"learning_rate": 5.121359310460893e-06,
|
| 9077 |
+
"loss": 1.6887,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.10901803607214429,
|
| 9082 |
+
"grad_norm": 3.289600133895874,
|
| 9083 |
+
"learning_rate": 5.052038731683073e-06,
|
| 9084 |
+
"loss": 1.6037,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.10910241535702984,
|
| 9089 |
+
"grad_norm": 3.2429816722869873,
|
| 9090 |
+
"learning_rate": 4.983178343001204e-06,
|
| 9091 |
+
"loss": 1.2427,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.10918679464191541,
|
| 9096 |
+
"grad_norm": 2.521946430206299,
|
| 9097 |
+
"learning_rate": 4.914778478165183e-06,
|
| 9098 |
+
"loss": 1.3839,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.10927117392680097,
|
| 9103 |
+
"grad_norm": 2.9969964027404785,
|
| 9104 |
+
"learning_rate": 4.846839468692777e-06,
|
| 9105 |
+
"loss": 1.4985,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.10935555321168654,
|
| 9110 |
+
"grad_norm": 3.4893572330474854,
|
| 9111 |
+
"learning_rate": 4.779361643868174e-06,
|
| 9112 |
+
"loss": 1.5088,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.10943993249657209,
|
| 9117 |
+
"grad_norm": 2.998934507369995,
|
| 9118 |
+
"learning_rate": 4.712345330740253e-06,
|
| 9119 |
+
"loss": 1.7704,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.10952431178145765,
|
| 9124 |
+
"grad_norm": 3.3882455825805664,
|
| 9125 |
+
"learning_rate": 4.645790854121091e-06,
|
| 9126 |
+
"loss": 1.6373,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.10960869106634322,
|
| 9131 |
+
"grad_norm": 2.9345991611480713,
|
| 9132 |
+
"learning_rate": 4.579698536584365e-06,
|
| 9133 |
+
"loss": 1.1513,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.10969307035122877,
|
| 9138 |
+
"grad_norm": 3.5486810207366943,
|
| 9139 |
+
"learning_rate": 4.5140686984637136e-06,
|
| 9140 |
+
"loss": 1.9755,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.10977744963611433,
|
| 9145 |
+
"grad_norm": 3.2446093559265137,
|
| 9146 |
+
"learning_rate": 4.4489016578513234e-06,
|
| 9147 |
+
"loss": 1.7986,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.1098618289209999,
|
| 9152 |
+
"grad_norm": 3.037705421447754,
|
| 9153 |
+
"learning_rate": 4.384197730596262e-06,
|
| 9154 |
+
"loss": 1.5818,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.10994620820588545,
|
| 9159 |
+
"grad_norm": 2.904578924179077,
|
| 9160 |
+
"learning_rate": 4.319957230303029e-06,
|
| 9161 |
+
"loss": 1.309,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.11003058749077102,
|
| 9166 |
+
"grad_norm": 3.392925977706909,
|
| 9167 |
+
"learning_rate": 4.2561804683299844e-06,
|
| 9168 |
+
"loss": 1.9903,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.11011496677565658,
|
| 9173 |
+
"grad_norm": 3.4092817306518555,
|
| 9174 |
+
"learning_rate": 4.1928677537878655e-06,
|
| 9175 |
+
"loss": 1.6592,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.11019934606054213,
|
| 9180 |
+
"grad_norm": 4.5724639892578125,
|
| 9181 |
+
"learning_rate": 4.130019393538265e-06,
|
| 9182 |
+
"loss": 1.4045,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.1102837253454277,
|
| 9187 |
+
"grad_norm": 3.0653493404388428,
|
| 9188 |
+
"learning_rate": 4.067635692192195e-06,
|
| 9189 |
+
"loss": 0.996,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.11036810463031325,
|
| 9194 |
+
"grad_norm": 3.045896053314209,
|
| 9195 |
+
"learning_rate": 4.005716952108552e-06,
|
| 9196 |
+
"loss": 1.4012,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.11045248391519882,
|
| 9201 |
+
"grad_norm": 3.0745668411254883,
|
| 9202 |
+
"learning_rate": 3.944263473392706e-06,
|
| 9203 |
+
"loss": 1.363,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.11053686320008438,
|
| 9208 |
+
"grad_norm": 3.7745227813720703,
|
| 9209 |
+
"learning_rate": 3.883275553894972e-06,
|
| 9210 |
+
"loss": 2.0488,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.11062124248496993,
|
| 9215 |
+
"grad_norm": 3.5060205459594727,
|
| 9216 |
+
"learning_rate": 3.822753489209219e-06,
|
| 9217 |
+
"loss": 1.4832,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.1107056217698555,
|
| 9222 |
+
"grad_norm": 2.9425277709960938,
|
| 9223 |
+
"learning_rate": 3.7626975726714588e-06,
|
| 9224 |
+
"loss": 1.0551,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.11079000105474106,
|
| 9229 |
+
"grad_norm": 2.8311192989349365,
|
| 9230 |
+
"learning_rate": 3.7031080953583873e-06,
|
| 9231 |
+
"loss": 1.1704,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.11087438033962663,
|
| 9236 |
+
"grad_norm": 3.563190460205078,
|
| 9237 |
+
"learning_rate": 3.643985346085965e-06,
|
| 9238 |
+
"loss": 1.9308,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.11095875962451218,
|
| 9243 |
+
"grad_norm": 2.653465509414673,
|
| 9244 |
+
"learning_rate": 3.5853296114080392e-06,
|
| 9245 |
+
"loss": 1.51,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.11104313890939774,
|
| 9250 |
+
"grad_norm": 3.5574042797088623,
|
| 9251 |
+
"learning_rate": 3.527141175614945e-06,
|
| 9252 |
+
"loss": 1.6356,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.1111275181942833,
|
| 9257 |
+
"grad_norm": 2.8941783905029297,
|
| 9258 |
+
"learning_rate": 3.469420320732153e-06,
|
| 9259 |
+
"loss": 1.1934,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.11121189747916886,
|
| 9264 |
+
"grad_norm": 2.820608139038086,
|
| 9265 |
+
"learning_rate": 3.4121673265188335e-06,
|
| 9266 |
+
"loss": 1.4181,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.11129627676405443,
|
| 9271 |
+
"grad_norm": 3.2258927822113037,
|
| 9272 |
+
"learning_rate": 3.355382470466606e-06,
|
| 9273 |
+
"loss": 1.29,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.11138065604893999,
|
| 9278 |
+
"grad_norm": 2.660257577896118,
|
| 9279 |
+
"learning_rate": 3.2990660277981032e-06,
|
| 9280 |
+
"loss": 1.287,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.11146503533382554,
|
| 9285 |
+
"grad_norm": 3.393606185913086,
|
| 9286 |
+
"learning_rate": 3.2432182714656534e-06,
|
| 9287 |
+
"loss": 1.3975,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.11154941461871111,
|
| 9292 |
+
"grad_norm": 3.444636583328247,
|
| 9293 |
+
"learning_rate": 3.1878394721500225e-06,
|
| 9294 |
+
"loss": 1.8969,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.11163379390359666,
|
| 9299 |
+
"grad_norm": 3.307429552078247,
|
| 9300 |
+
"learning_rate": 3.1329298982590184e-06,
|
| 9301 |
+
"loss": 1.7586,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.11171817318848223,
|
| 9306 |
+
"grad_norm": 2.8406436443328857,
|
| 9307 |
+
"learning_rate": 3.078489815926222e-06,
|
| 9308 |
+
"loss": 1.3457,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.11180255247336779,
|
| 9313 |
+
"grad_norm": 3.238516092300415,
|
| 9314 |
+
"learning_rate": 3.024519489009736e-06,
|
| 9315 |
+
"loss": 1.7479,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.11188693175825334,
|
| 9320 |
+
"grad_norm": 3.127741813659668,
|
| 9321 |
+
"learning_rate": 2.9710191790908394e-06,
|
| 9322 |
+
"loss": 1.8992,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.11197131104313891,
|
| 9327 |
+
"grad_norm": 3.0072107315063477,
|
| 9328 |
+
"learning_rate": 2.9179891454727437e-06,
|
| 9329 |
+
"loss": 1.4593,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.11205569032802447,
|
| 9334 |
+
"grad_norm": 3.2357969284057617,
|
| 9335 |
+
"learning_rate": 2.865429645179374e-06,
|
| 9336 |
+
"loss": 1.6217,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.11214006961291004,
|
| 9341 |
+
"grad_norm": 2.916038751602173,
|
| 9342 |
+
"learning_rate": 2.813340932954067e-06,
|
| 9343 |
+
"loss": 1.3307,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.11222444889779559,
|
| 9348 |
+
"grad_norm": 3.672311782836914,
|
| 9349 |
+
"learning_rate": 2.7617232612583977e-06,
|
| 9350 |
+
"loss": 1.8352,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.11230882818268115,
|
| 9355 |
+
"grad_norm": 3.160975933074951,
|
| 9356 |
+
"learning_rate": 2.7105768802708544e-06,
|
| 9357 |
+
"loss": 1.5768,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.11239320746756672,
|
| 9362 |
+
"grad_norm": 4.55344820022583,
|
| 9363 |
+
"learning_rate": 2.659902037885775e-06,
|
| 9364 |
+
"loss": 2.1345,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.11247758675245227,
|
| 9369 |
+
"grad_norm": 3.365166425704956,
|
| 9370 |
+
"learning_rate": 2.6096989797119943e-06,
|
| 9371 |
+
"loss": 1.7261,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.11256196603733783,
|
| 9376 |
+
"grad_norm": 3.8819565773010254,
|
| 9377 |
+
"learning_rate": 2.55996794907174e-06,
|
| 9378 |
+
"loss": 1.7748,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.1126463453222234,
|
| 9383 |
+
"grad_norm": 2.803469181060791,
|
| 9384 |
+
"learning_rate": 2.5107091869994714e-06,
|
| 9385 |
+
"loss": 1.1567,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.11273072460710895,
|
| 9390 |
+
"grad_norm": 2.9851107597351074,
|
| 9391 |
+
"learning_rate": 2.4619229322406347e-06,
|
| 9392 |
+
"loss": 1.1736,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.11281510389199452,
|
| 9397 |
+
"grad_norm": 2.5405960083007812,
|
| 9398 |
+
"learning_rate": 2.41360942125054e-06,
|
| 9399 |
+
"loss": 1.2245,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.11289948317688007,
|
| 9404 |
+
"grad_norm": 3.213590383529663,
|
| 9405 |
+
"learning_rate": 2.3657688881932536e-06,
|
| 9406 |
+
"loss": 1.5082,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.11298386246176563,
|
| 9411 |
+
"grad_norm": 3.265373468399048,
|
| 9412 |
+
"learning_rate": 2.31840156494042e-06,
|
| 9413 |
+
"loss": 1.6259,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.1130682417466512,
|
| 9418 |
+
"grad_norm": 3.7356579303741455,
|
| 9419 |
+
"learning_rate": 2.2715076810701284e-06,
|
| 9420 |
+
"loss": 1.817,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.11315262103153675,
|
| 9425 |
+
"grad_norm": 3.2784945964813232,
|
| 9426 |
+
"learning_rate": 2.2250874638658604e-06,
|
| 9427 |
+
"loss": 1.0401,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.11323700031642232,
|
| 9432 |
+
"grad_norm": 3.730741500854492,
|
| 9433 |
+
"learning_rate": 2.1791411383152995e-06,
|
| 9434 |
+
"loss": 1.8904,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.11332137960130788,
|
| 9439 |
+
"grad_norm": 3.1985487937927246,
|
| 9440 |
+
"learning_rate": 2.1336689271093224e-06,
|
| 9441 |
+
"loss": 1.5335,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.11340575888619343,
|
| 9446 |
+
"grad_norm": 3.5334551334381104,
|
| 9447 |
+
"learning_rate": 2.088671050640889e-06,
|
| 9448 |
+
"loss": 1.6912,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.113490138171079,
|
| 9453 |
+
"grad_norm": 2.56057071685791,
|
| 9454 |
+
"learning_rate": 2.044147727003942e-06,
|
| 9455 |
+
"loss": 0.9783,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.11357451745596456,
|
| 9460 |
+
"grad_norm": 3.8409194946289062,
|
| 9461 |
+
"learning_rate": 2.00009917199242e-06,
|
| 9462 |
+
"loss": 1.7168,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.11365889674085013,
|
| 9467 |
+
"grad_norm": 3.3304567337036133,
|
| 9468 |
+
"learning_rate": 1.956525599099146e-06,
|
| 9469 |
+
"loss": 1.6863,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.11374327602573568,
|
| 9474 |
+
"grad_norm": 3.584198236465454,
|
| 9475 |
+
"learning_rate": 1.9134272195148075e-06,
|
| 9476 |
+
"loss": 1.5168,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.11382765531062124,
|
| 9481 |
+
"grad_norm": 3.4981017112731934,
|
| 9482 |
+
"learning_rate": 1.8708042421269777e-06,
|
| 9483 |
+
"loss": 1.5405,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.1139120345955068,
|
| 9488 |
+
"grad_norm": 3.4696459770202637,
|
| 9489 |
+
"learning_rate": 1.8286568735190411e-06,
|
| 9490 |
+
"loss": 1.3721,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.11399641388039236,
|
| 9495 |
+
"grad_norm": 3.3460185527801514,
|
| 9496 |
+
"learning_rate": 1.7869853179692365e-06,
|
| 9497 |
+
"loss": 1.4007,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.11408079316527793,
|
| 9502 |
+
"grad_norm": 3.1510324478149414,
|
| 9503 |
+
"learning_rate": 1.7457897774496245e-06,
|
| 9504 |
+
"loss": 1.4591,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.11416517245016349,
|
| 9509 |
+
"grad_norm": 3.922882318496704,
|
| 9510 |
+
"learning_rate": 1.7050704516251682e-06,
|
| 9511 |
+
"loss": 1.9073,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.11424955173504904,
|
| 9516 |
+
"grad_norm": 2.8829705715179443,
|
| 9517 |
+
"learning_rate": 1.6648275378527201e-06,
|
| 9518 |
+
"loss": 1.399,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.11433393101993461,
|
| 9523 |
+
"grad_norm": 3.3844122886657715,
|
| 9524 |
+
"learning_rate": 1.6250612311800695e-06,
|
| 9525 |
+
"loss": 1.471,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.11441831030482016,
|
| 9530 |
+
"grad_norm": 3.1320078372955322,
|
| 9531 |
+
"learning_rate": 1.5857717243449976e-06,
|
| 9532 |
+
"loss": 1.4049,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.11450268958970573,
|
| 9537 |
+
"grad_norm": 3.1687233448028564,
|
| 9538 |
+
"learning_rate": 1.54695920777439e-06,
|
| 9539 |
+
"loss": 1.5405,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.11458706887459129,
|
| 9544 |
+
"grad_norm": 3.051271915435791,
|
| 9545 |
+
"learning_rate": 1.5086238695832145e-06,
|
| 9546 |
+
"loss": 1.6626,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.11467144815947684,
|
| 9551 |
+
"grad_norm": 3.041841506958008,
|
| 9552 |
+
"learning_rate": 1.4707658955737113e-06,
|
| 9553 |
+
"loss": 1.6603,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.11475582744436241,
|
| 9558 |
+
"grad_norm": 3.717074394226074,
|
| 9559 |
+
"learning_rate": 1.4333854692344494e-06,
|
| 9560 |
+
"loss": 1.8031,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.11484020672924797,
|
| 9565 |
+
"grad_norm": 3.747009038925171,
|
| 9566 |
+
"learning_rate": 1.396482771739438e-06,
|
| 9567 |
+
"loss": 1.7961,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.11492458601413354,
|
| 9572 |
+
"grad_norm": 3.030385732650757,
|
| 9573 |
+
"learning_rate": 1.360057981947227e-06,
|
| 9574 |
+
"loss": 1.6847,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.11500896529901909,
|
| 9579 |
+
"grad_norm": 3.3725221157073975,
|
| 9580 |
+
"learning_rate": 1.324111276400064e-06,
|
| 9581 |
+
"loss": 1.9146,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.11509334458390465,
|
| 9586 |
+
"grad_norm": 3.4748051166534424,
|
| 9587 |
+
"learning_rate": 1.2886428293230502e-06,
|
| 9588 |
+
"loss": 1.929,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.11517772386879022,
|
| 9593 |
+
"grad_norm": 3.4699349403381348,
|
| 9594 |
+
"learning_rate": 1.253652812623296e-06,
|
| 9595 |
+
"loss": 1.7753,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.11526210315367577,
|
| 9600 |
+
"grad_norm": 3.4791486263275146,
|
| 9601 |
+
"learning_rate": 1.2191413958890452e-06,
|
| 9602 |
+
"loss": 1.4113,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.11534648243856133,
|
| 9607 |
+
"grad_norm": 3.4561052322387695,
|
| 9608 |
+
"learning_rate": 1.1851087463889076e-06,
|
| 9609 |
+
"loss": 1.7275,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.1154308617234469,
|
| 9614 |
+
"grad_norm": 3.3334875106811523,
|
| 9615 |
+
"learning_rate": 1.1515550290709832e-06,
|
| 9616 |
+
"loss": 1.6052,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.11551524100833245,
|
| 9621 |
+
"grad_norm": 2.4955997467041016,
|
| 9622 |
+
"learning_rate": 1.1184804065621502e-06,
|
| 9623 |
+
"loss": 1.1223,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.11559962029321802,
|
| 9628 |
+
"grad_norm": 3.3001463413238525,
|
| 9629 |
+
"learning_rate": 1.0858850391671894e-06,
|
| 9630 |
+
"loss": 1.7515,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.11568399957810357,
|
| 9635 |
+
"grad_norm": 3.1258296966552734,
|
| 9636 |
+
"learning_rate": 1.0537690848680837e-06,
|
| 9637 |
+
"loss": 1.5765,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.11576837886298913,
|
| 9642 |
+
"grad_norm": 2.9277150630950928,
|
| 9643 |
+
"learning_rate": 1.0221326993231861e-06,
|
| 9644 |
+
"loss": 1.185,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.1158527581478747,
|
| 9649 |
+
"grad_norm": 2.9128518104553223,
|
| 9650 |
+
"learning_rate": 9.909760358664867e-07,
|
| 9651 |
+
"loss": 1.3728,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.11593713743276025,
|
| 9656 |
+
"grad_norm": 3.100315809249878,
|
| 9657 |
+
"learning_rate": 9.602992455069126e-07,
|
| 9658 |
+
"loss": 1.5857,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.11602151671764582,
|
| 9663 |
+
"grad_norm": 2.6965436935424805,
|
| 9664 |
+
"learning_rate": 9.301024769275302e-07,
|
| 9665 |
+
"loss": 1.5054,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.11610589600253138,
|
| 9670 |
+
"grad_norm": 3.325709342956543,
|
| 9671 |
+
"learning_rate": 9.003858764848771e-07,
|
| 9672 |
+
"loss": 1.7131,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.11619027528741693,
|
| 9677 |
+
"grad_norm": 3.0008556842803955,
|
| 9678 |
+
"learning_rate": 8.711495882082087e-07,
|
| 9679 |
+
"loss": 1.0944,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.1162746545723025,
|
| 9684 |
+
"grad_norm": 2.860565423965454,
|
| 9685 |
+
"learning_rate": 8.423937537988536e-07,
|
| 9686 |
+
"loss": 1.1299,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.11635903385718806,
|
| 9691 |
+
"grad_norm": 3.720402240753174,
|
| 9692 |
+
"learning_rate": 8.14118512629447e-07,
|
| 9693 |
+
"loss": 1.9886,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.11644341314207363,
|
| 9698 |
+
"grad_norm": 3.104672431945801,
|
| 9699 |
+
"learning_rate": 7.863240017433438e-07,
|
| 9700 |
+
"loss": 1.5153,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.11652779242695918,
|
| 9705 |
+
"grad_norm": 3.108832597732544,
|
| 9706 |
+
"learning_rate": 7.590103558538952e-07,
|
| 9707 |
+
"loss": 1.5678,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.11661217171184474,
|
| 9712 |
+
"grad_norm": 2.9974732398986816,
|
| 9713 |
+
"learning_rate": 7.321777073438397e-07,
|
| 9714 |
+
"loss": 1.4503,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.1166965509967303,
|
| 9719 |
+
"grad_norm": 3.0287883281707764,
|
| 9720 |
+
"learning_rate": 7.058261862646021e-07,
|
| 9721 |
+
"loss": 1.3268,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.11678093028161586,
|
| 9726 |
+
"grad_norm": 2.951352119445801,
|
| 9727 |
+
"learning_rate": 6.799559203357175e-07,
|
| 9728 |
+
"loss": 1.5485,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.11686530956650143,
|
| 9733 |
+
"grad_norm": 4.005697250366211,
|
| 9734 |
+
"learning_rate": 6.545670349441757e-07,
|
| 9735 |
+
"loss": 1.9356,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.11694968885138698,
|
| 9740 |
+
"grad_norm": 3.2043464183807373,
|
| 9741 |
+
"learning_rate": 6.296596531438547e-07,
|
| 9742 |
+
"loss": 1.7386,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.11703406813627254,
|
| 9747 |
+
"grad_norm": 3.1487245559692383,
|
| 9748 |
+
"learning_rate": 6.052338956548776e-07,
|
| 9749 |
+
"loss": 1.1601,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.11711844742115811,
|
| 9754 |
+
"grad_norm": 3.383782386779785,
|
| 9755 |
+
"learning_rate": 5.812898808630785e-07,
|
| 9756 |
+
"loss": 1.4245,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.11720282670604366,
|
| 9761 |
+
"grad_norm": 3.1075830459594727,
|
| 9762 |
+
"learning_rate": 5.578277248193486e-07,
|
| 9763 |
+
"loss": 1.8653,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.11728720599092923,
|
| 9768 |
+
"grad_norm": 2.613389253616333,
|
| 9769 |
+
"learning_rate": 5.348475412391806e-07,
|
| 9770 |
+
"loss": 1.4329,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.11737158527581479,
|
| 9775 |
+
"grad_norm": 3.2017884254455566,
|
| 9776 |
+
"learning_rate": 5.123494415020247e-07,
|
| 9777 |
+
"loss": 1.6501,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.11745596456070034,
|
| 9782 |
+
"grad_norm": 3.469419240951538,
|
| 9783 |
+
"learning_rate": 4.903335346507997e-07,
|
| 9784 |
+
"loss": 1.6442,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.11754034384558591,
|
| 9789 |
+
"grad_norm": 3.5213255882263184,
|
| 9790 |
+
"learning_rate": 4.6879992739136127e-07,
|
| 9791 |
+
"loss": 1.9345,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.11762472313047147,
|
| 9796 |
+
"grad_norm": 2.5828282833099365,
|
| 9797 |
+
"learning_rate": 4.4774872409193426e-07,
|
| 9798 |
+
"loss": 1.3535,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.11770910241535704,
|
| 9803 |
+
"grad_norm": 3.743889808654785,
|
| 9804 |
+
"learning_rate": 4.271800267827031e-07,
|
| 9805 |
+
"loss": 1.4343,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.11779348170024259,
|
| 9810 |
+
"grad_norm": 3.10267972946167,
|
| 9811 |
+
"learning_rate": 4.0709393515521165e-07,
|
| 9812 |
+
"loss": 1.6854,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.11787786098512815,
|
| 9817 |
+
"grad_norm": 2.8783440589904785,
|
| 9818 |
+
"learning_rate": 3.874905465619749e-07,
|
| 9819 |
+
"loss": 1.3887,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.11796224027001372,
|
| 9824 |
+
"grad_norm": 3.2154781818389893,
|
| 9825 |
+
"learning_rate": 3.683699560159015e-07,
|
| 9826 |
+
"loss": 1.7844,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.11804661955489927,
|
| 9831 |
+
"grad_norm": 3.865391731262207,
|
| 9832 |
+
"learning_rate": 3.4973225618996077e-07,
|
| 9833 |
+
"loss": 1.7215,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.11813099883978483,
|
| 9838 |
+
"grad_norm": 2.8252804279327393,
|
| 9839 |
+
"learning_rate": 3.3157753741661635e-07,
|
| 9840 |
+
"loss": 1.6918,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.1182153781246704,
|
| 9845 |
+
"grad_norm": 3.2213222980499268,
|
| 9846 |
+
"learning_rate": 3.13905887687449e-07,
|
| 9847 |
+
"loss": 1.439,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.11829975740955595,
|
| 9852 |
+
"grad_norm": 3.1654062271118164,
|
| 9853 |
+
"learning_rate": 2.967173926527345e-07,
|
| 9854 |
+
"loss": 1.591,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.11838413669444152,
|
| 9859 |
+
"grad_norm": 3.003859519958496,
|
| 9860 |
+
"learning_rate": 2.8001213562101056e-07,
|
| 9861 |
+
"loss": 1.6151,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.11846851597932707,
|
| 9866 |
+
"grad_norm": 2.7857260704040527,
|
| 9867 |
+
"learning_rate": 2.6379019755866654e-07,
|
| 9868 |
+
"loss": 1.4298,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.11855289526421263,
|
| 9873 |
+
"grad_norm": 4.761999607086182,
|
| 9874 |
+
"learning_rate": 2.4805165708954306e-07,
|
| 9875 |
+
"loss": 1.7377,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.1186372745490982,
|
| 9880 |
+
"grad_norm": 3.1942343711853027,
|
| 9881 |
+
"learning_rate": 2.3279659049461055e-07,
|
| 9882 |
+
"loss": 1.1933,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.11872165383398375,
|
| 9887 |
+
"grad_norm": 4.278934955596924,
|
| 9888 |
+
"learning_rate": 2.1802507171152508e-07,
|
| 9889 |
+
"loss": 2.1788,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.11880603311886932,
|
| 9894 |
+
"grad_norm": 3.102963447570801,
|
| 9895 |
+
"learning_rate": 2.0373717233431732e-07,
|
| 9896 |
+
"loss": 1.7477,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.11889041240375488,
|
| 9901 |
+
"grad_norm": 3.8555305004119873,
|
| 9902 |
+
"learning_rate": 1.8993296161301522e-07,
|
| 9903 |
+
"loss": 1.8359,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.11897479168864043,
|
| 9908 |
+
"grad_norm": 2.72951602935791,
|
| 9909 |
+
"learning_rate": 1.7661250645333304e-07,
|
| 9910 |
+
"loss": 1.6447,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.119059170973526,
|
| 9915 |
+
"grad_norm": 2.7467477321624756,
|
| 9916 |
+
"learning_rate": 1.6377587141633843e-07,
|
| 9917 |
+
"loss": 1.1554,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.11914355025841156,
|
| 9922 |
+
"grad_norm": 2.9201080799102783,
|
| 9923 |
+
"learning_rate": 1.514231187181303e-07,
|
| 9924 |
+
"loss": 1.3712,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.11922792954329713,
|
| 9929 |
+
"grad_norm": 2.7157042026519775,
|
| 9930 |
+
"learning_rate": 1.395543082295614e-07,
|
| 9931 |
+
"loss": 1.4105,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.11931230882818268,
|
| 9936 |
+
"grad_norm": 3.224670648574829,
|
| 9937 |
+
"learning_rate": 1.2816949747592732e-07,
|
| 9938 |
+
"loss": 1.6758,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.11939668811306824,
|
| 9943 |
+
"grad_norm": 4.1305952072143555,
|
| 9944 |
+
"learning_rate": 1.172687416366669e-07,
|
| 9945 |
+
"loss": 1.8614,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.1194810673979538,
|
| 9950 |
+
"grad_norm": 3.448305606842041,
|
| 9951 |
+
"learning_rate": 1.0685209354515114e-07,
|
| 9952 |
+
"loss": 1.8458,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.11956544668283936,
|
| 9957 |
+
"grad_norm": 3.2459452152252197,
|
| 9958 |
+
"learning_rate": 9.691960368838349e-08,
|
| 9959 |
+
"loss": 1.5464,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.11964982596772493,
|
| 9964 |
+
"grad_norm": 3.1071956157684326,
|
| 9965 |
+
"learning_rate": 8.747132020677784e-08,
|
| 9966 |
+
"loss": 1.824,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.11973420525261048,
|
| 9971 |
+
"grad_norm": 3.374622106552124,
|
| 9972 |
+
"learning_rate": 7.850728889391424e-08,
|
| 9973 |
+
"loss": 1.3654,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.11981858453749604,
|
| 9978 |
+
"grad_norm": 3.738436222076416,
|
| 9979 |
+
"learning_rate": 7.002755319628352e-08,
|
| 9980 |
+
"loss": 2.1948,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.11990296382238161,
|
| 9985 |
+
"grad_norm": 3.6799709796905518,
|
| 9986 |
+
"learning_rate": 6.203215421316521e-08,
|
| 9987 |
+
"loss": 1.7075,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.11998734310726716,
|
| 9992 |
+
"grad_norm": 3.5823380947113037,
|
| 9993 |
+
"learning_rate": 5.452113069633891e-08,
|
| 9994 |
+
"loss": 1.4088,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.12007172239215273,
|
| 9999 |
+
"grad_norm": 3.7232697010040283,
|
| 10000 |
+
"learning_rate": 4.749451904995095e-08,
|
| 10001 |
+
"loss": 1.7351,
|
| 10002 |
+
"step": 1423
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 0.12015610167703829,
|
| 10006 |
+
"grad_norm": 3.699223279953003,
|
| 10007 |
+
"learning_rate": 4.095235333030356e-08,
|
| 10008 |
+
"loss": 1.7289,
|
| 10009 |
+
"step": 1424
|
| 10010 |
+
},
|
| 10011 |
+
{
|
| 10012 |
+
"epoch": 0.12024048096192384,
|
| 10013 |
+
"grad_norm": 3.006974220275879,
|
| 10014 |
+
"learning_rate": 3.48946652456994e-08,
|
| 10015 |
+
"loss": 1.8416,
|
| 10016 |
+
"step": 1425
|
| 10017 |
+
},
|
| 10018 |
+
{
|
| 10019 |
+
"epoch": 0.12032486024680941,
|
| 10020 |
+
"grad_norm": 3.424286365509033,
|
| 10021 |
+
"learning_rate": 2.9321484156330515e-08,
|
| 10022 |
+
"loss": 1.3221,
|
| 10023 |
+
"step": 1426
|
| 10024 |
+
},
|
| 10025 |
+
{
|
| 10026 |
+
"epoch": 0.12040923953169497,
|
| 10027 |
+
"grad_norm": 3.1630194187164307,
|
| 10028 |
+
"learning_rate": 2.4232837074056326e-08,
|
| 10029 |
+
"loss": 1.4155,
|
| 10030 |
+
"step": 1427
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 0.12049361881658054,
|
| 10034 |
+
"grad_norm": 3.1406900882720947,
|
| 10035 |
+
"learning_rate": 1.9628748662348096e-08,
|
| 10036 |
+
"loss": 1.5631,
|
| 10037 |
+
"step": 1428
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 0.12057799810146609,
|
| 10041 |
+
"grad_norm": 3.563826084136963,
|
| 10042 |
+
"learning_rate": 1.5509241236111305e-08,
|
| 10043 |
+
"loss": 1.5566,
|
| 10044 |
+
"step": 1429
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 0.12066237738635165,
|
| 10048 |
+
"grad_norm": 3.097853899002075,
|
| 10049 |
+
"learning_rate": 1.1874334761619033e-08,
|
| 10050 |
+
"loss": 1.4279,
|
| 10051 |
+
"step": 1430
|
| 10052 |
+
},
|
| 10053 |
+
{
|
| 10054 |
+
"epoch": 0.12074675667123722,
|
| 10055 |
+
"grad_norm": 3.372969627380371,
|
| 10056 |
+
"learning_rate": 8.724046856389833e-09,
|
| 10057 |
+
"loss": 1.5542,
|
| 10058 |
+
"step": 1431
|
| 10059 |
+
},
|
| 10060 |
+
{
|
| 10061 |
+
"epoch": 0.12083113595612277,
|
| 10062 |
+
"grad_norm": 3.480236768722534,
|
| 10063 |
+
"learning_rate": 6.058392789132228e-09,
|
| 10064 |
+
"loss": 1.7213,
|
| 10065 |
+
"step": 1432
|
| 10066 |
+
},
|
| 10067 |
+
{
|
| 10068 |
+
"epoch": 0.12091551524100833,
|
| 10069 |
+
"grad_norm": 4.48645544052124,
|
| 10070 |
+
"learning_rate": 3.877385479622575e-09,
|
| 10071 |
+
"loss": 2.0068,
|
| 10072 |
+
"step": 1433
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 0.1209998945258939,
|
| 10076 |
+
"grad_norm": 3.350323438644409,
|
| 10077 |
+
"learning_rate": 2.1810354986828708e-09,
|
| 10078 |
+
"loss": 1.366,
|
| 10079 |
+
"step": 1434
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 0.12108427381077945,
|
| 10083 |
+
"grad_norm": 3.5823779106140137,
|
| 10084 |
+
"learning_rate": 9.693510681141326e-10,
|
| 10085 |
+
"loss": 1.591,
|
| 10086 |
+
"step": 1435
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 0.12116865309566502,
|
| 10090 |
+
"grad_norm": 2.748819589614868,
|
| 10091 |
+
"learning_rate": 2.423380606630943e-10,
|
| 10092 |
+
"loss": 1.3537,
|
| 10093 |
+
"step": 1436
|
| 10094 |
+
},
|
| 10095 |
+
{
|
| 10096 |
+
"epoch": 0.12125303238055057,
|
| 10097 |
+
"grad_norm": 3.025970458984375,
|
| 10098 |
+
"learning_rate": 0.0,
|
| 10099 |
+
"loss": 1.483,
|
| 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": 2.7792046921089024e+17,
|
| 10121 |
"train_batch_size": 2,
|
| 10122 |
"trial_name": null,
|
| 10123 |
"trial_params": null
|