Training in progress, step 1500, 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 28348936
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2cb15242cc7cfc229266a4e3a90e8815b604118499a28582005f3abe6be3f6d1
|
| 3 |
size 28348936
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14714324
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:74bf4d264496d0bc7b2505629bfacf458fe8c4c843d890a86ce662861b7fc79c
|
| 3 |
size 14714324
|
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 sha256:733b5c5c0a3916e70f852e663c0513c05bab3aa0f71a30256e61fa740397a848
|
| 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:fe4e625cf5525cf235a8c7e4c6984ba05d1b5436f9e969799dcec0119a3a6c2a
|
| 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": 375,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7914,6 +7914,2639 @@
|
|
| 7914 |
"eval_samples_per_second": 18.578,
|
| 7915 |
"eval_steps_per_second": 9.291,
|
| 7916 |
"step": 1125
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7917 |
}
|
| 7918 |
],
|
| 7919 |
"logging_steps": 1,
|
|
@@ -7928,12 +10561,12 @@
|
|
| 7928 |
"should_evaluate": false,
|
| 7929 |
"should_log": false,
|
| 7930 |
"should_save": true,
|
| 7931 |
-
"should_training_stop":
|
| 7932 |
},
|
| 7933 |
"attributes": {}
|
| 7934 |
}
|
| 7935 |
},
|
| 7936 |
-
"total_flos":
|
| 7937 |
"train_batch_size": 2,
|
| 7938 |
"trial_name": null,
|
| 7939 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.5067781578613961,
|
| 5 |
"eval_steps": 375,
|
| 6 |
+
"global_step": 1500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7914 |
"eval_samples_per_second": 18.578,
|
| 7915 |
"eval_steps_per_second": 9.291,
|
| 7916 |
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.38042147050128805,
|
| 7920 |
+
"grad_norm": 0.2514338195323944,
|
| 7921 |
+
"learning_rate": 2.9513310198183065e-05,
|
| 7922 |
+
"loss": 1.5087,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.380759322606529,
|
| 7927 |
+
"grad_norm": 0.252075731754303,
|
| 7928 |
+
"learning_rate": 2.936390617691097e-05,
|
| 7929 |
+
"loss": 1.5065,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.3810971747117699,
|
| 7934 |
+
"grad_norm": 0.24784700572490692,
|
| 7935 |
+
"learning_rate": 2.9214816173045356e-05,
|
| 7936 |
+
"loss": 1.4838,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.38143502681701086,
|
| 7941 |
+
"grad_norm": 0.2557724118232727,
|
| 7942 |
+
"learning_rate": 2.906604084937572e-05,
|
| 7943 |
+
"loss": 1.4298,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.38177287892225176,
|
| 7948 |
+
"grad_norm": 0.24667373299598694,
|
| 7949 |
+
"learning_rate": 2.8917580867292526e-05,
|
| 7950 |
+
"loss": 1.378,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.3821107310274927,
|
| 7955 |
+
"grad_norm": 0.25280389189720154,
|
| 7956 |
+
"learning_rate": 2.8769436886784408e-05,
|
| 7957 |
+
"loss": 1.4796,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.38244858313273367,
|
| 7962 |
+
"grad_norm": 0.25293001532554626,
|
| 7963 |
+
"learning_rate": 2.862160956643517e-05,
|
| 7964 |
+
"loss": 1.3806,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.38278643523797456,
|
| 7969 |
+
"grad_norm": 0.2621861398220062,
|
| 7970 |
+
"learning_rate": 2.847409956342092e-05,
|
| 7971 |
+
"loss": 1.4621,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.3831242873432155,
|
| 7976 |
+
"grad_norm": 0.254467248916626,
|
| 7977 |
+
"learning_rate": 2.8326907533507074e-05,
|
| 7978 |
+
"loss": 1.4475,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.3834621394484564,
|
| 7983 |
+
"grad_norm": 0.25853973627090454,
|
| 7984 |
+
"learning_rate": 2.8180034131045464e-05,
|
| 7985 |
+
"loss": 1.4008,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.3837999915536974,
|
| 7990 |
+
"grad_norm": 0.2530171573162079,
|
| 7991 |
+
"learning_rate": 2.8033480008971546e-05,
|
| 7992 |
+
"loss": 1.4372,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.3841378436589383,
|
| 7997 |
+
"grad_norm": 0.2639373540878296,
|
| 7998 |
+
"learning_rate": 2.7887245818801277e-05,
|
| 7999 |
+
"loss": 1.4753,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.3844756957641792,
|
| 8004 |
+
"grad_norm": 0.270391047000885,
|
| 8005 |
+
"learning_rate": 2.7741332210628345e-05,
|
| 8006 |
+
"loss": 1.5221,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.3848135478694202,
|
| 8011 |
+
"grad_norm": 0.2627699375152588,
|
| 8012 |
+
"learning_rate": 2.759573983312138e-05,
|
| 8013 |
+
"loss": 1.3277,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.3851513999746611,
|
| 8018 |
+
"grad_norm": 0.2647635340690613,
|
| 8019 |
+
"learning_rate": 2.7450469333520855e-05,
|
| 8020 |
+
"loss": 1.4227,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.38548925207990203,
|
| 8025 |
+
"grad_norm": 0.264097660779953,
|
| 8026 |
+
"learning_rate": 2.730552135763632e-05,
|
| 8027 |
+
"loss": 1.4693,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.38582710418514293,
|
| 8032 |
+
"grad_norm": 0.27580371499061584,
|
| 8033 |
+
"learning_rate": 2.7160896549843562e-05,
|
| 8034 |
+
"loss": 1.4946,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.3861649562903839,
|
| 8039 |
+
"grad_norm": 0.2730904817581177,
|
| 8040 |
+
"learning_rate": 2.701659555308169e-05,
|
| 8041 |
+
"loss": 1.481,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.38650280839562484,
|
| 8046 |
+
"grad_norm": 0.27228155732154846,
|
| 8047 |
+
"learning_rate": 2.6872619008850274e-05,
|
| 8048 |
+
"loss": 1.4353,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.38684066050086574,
|
| 8053 |
+
"grad_norm": 0.296337753534317,
|
| 8054 |
+
"learning_rate": 2.672896755720654e-05,
|
| 8055 |
+
"loss": 1.4298,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.3871785126061067,
|
| 8060 |
+
"grad_norm": 0.28648385405540466,
|
| 8061 |
+
"learning_rate": 2.6585641836762433e-05,
|
| 8062 |
+
"loss": 1.485,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.3875163647113476,
|
| 8067 |
+
"grad_norm": 0.2983736991882324,
|
| 8068 |
+
"learning_rate": 2.6442642484681944e-05,
|
| 8069 |
+
"loss": 1.4592,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.38785421681658855,
|
| 8074 |
+
"grad_norm": 0.299578458070755,
|
| 8075 |
+
"learning_rate": 2.6299970136678077e-05,
|
| 8076 |
+
"loss": 1.3252,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.38819206892182945,
|
| 8081 |
+
"grad_norm": 0.33620649576187134,
|
| 8082 |
+
"learning_rate": 2.6157625427010156e-05,
|
| 8083 |
+
"loss": 1.4812,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.3885299210270704,
|
| 8088 |
+
"grad_norm": 0.5138651728630066,
|
| 8089 |
+
"learning_rate": 2.6015608988480955e-05,
|
| 8090 |
+
"loss": 1.2072,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.38886777313231136,
|
| 8095 |
+
"grad_norm": 0.19787752628326416,
|
| 8096 |
+
"learning_rate": 2.5873921452433915e-05,
|
| 8097 |
+
"loss": 1.506,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.38920562523755226,
|
| 8102 |
+
"grad_norm": 0.209125816822052,
|
| 8103 |
+
"learning_rate": 2.57325634487503e-05,
|
| 8104 |
+
"loss": 1.5611,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.3895434773427932,
|
| 8109 |
+
"grad_norm": 0.20977284014225006,
|
| 8110 |
+
"learning_rate": 2.5591535605846383e-05,
|
| 8111 |
+
"loss": 1.5057,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.3898813294480341,
|
| 8116 |
+
"grad_norm": 0.21223272383213043,
|
| 8117 |
+
"learning_rate": 2.5450838550670808e-05,
|
| 8118 |
+
"loss": 1.4882,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.39021918155327506,
|
| 8123 |
+
"grad_norm": 0.19779768586158752,
|
| 8124 |
+
"learning_rate": 2.5310472908701555e-05,
|
| 8125 |
+
"loss": 1.4438,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.39055703365851596,
|
| 8130 |
+
"grad_norm": 0.20631931722164154,
|
| 8131 |
+
"learning_rate": 2.5170439303943294e-05,
|
| 8132 |
+
"loss": 1.4559,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.3908948857637569,
|
| 8137 |
+
"grad_norm": 0.21120774745941162,
|
| 8138 |
+
"learning_rate": 2.503073835892471e-05,
|
| 8139 |
+
"loss": 1.5819,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.39123273786899787,
|
| 8144 |
+
"grad_norm": 0.21454963088035583,
|
| 8145 |
+
"learning_rate": 2.4891370694695517e-05,
|
| 8146 |
+
"loss": 1.4856,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.39157058997423877,
|
| 8151 |
+
"grad_norm": 0.21086949110031128,
|
| 8152 |
+
"learning_rate": 2.4752336930823837e-05,
|
| 8153 |
+
"loss": 1.4149,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.3919084420794797,
|
| 8158 |
+
"grad_norm": 0.21786537766456604,
|
| 8159 |
+
"learning_rate": 2.4613637685393432e-05,
|
| 8160 |
+
"loss": 1.4088,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.3922462941847206,
|
| 8165 |
+
"grad_norm": 0.2270769625902176,
|
| 8166 |
+
"learning_rate": 2.4475273575000936e-05,
|
| 8167 |
+
"loss": 1.4738,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.3925841462899616,
|
| 8172 |
+
"grad_norm": 0.21777530014514923,
|
| 8173 |
+
"learning_rate": 2.4337245214753103e-05,
|
| 8174 |
+
"loss": 1.342,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.3929219983952025,
|
| 8179 |
+
"grad_norm": 0.21940767765045166,
|
| 8180 |
+
"learning_rate": 2.4199553218264093e-05,
|
| 8181 |
+
"loss": 1.4082,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.39325985050044343,
|
| 8186 |
+
"grad_norm": 0.23166020214557648,
|
| 8187 |
+
"learning_rate": 2.4062198197652752e-05,
|
| 8188 |
+
"loss": 1.451,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.3935977026056844,
|
| 8193 |
+
"grad_norm": 0.2314266413450241,
|
| 8194 |
+
"learning_rate": 2.3925180763539844e-05,
|
| 8195 |
+
"loss": 1.4501,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.3939355547109253,
|
| 8200 |
+
"grad_norm": 0.2357269525527954,
|
| 8201 |
+
"learning_rate": 2.3788501525045438e-05,
|
| 8202 |
+
"loss": 1.4474,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.39427340681616624,
|
| 8207 |
+
"grad_norm": 0.23468124866485596,
|
| 8208 |
+
"learning_rate": 2.3652161089786086e-05,
|
| 8209 |
+
"loss": 1.5431,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.39461125892140714,
|
| 8214 |
+
"grad_norm": 0.22679342329502106,
|
| 8215 |
+
"learning_rate": 2.351616006387214e-05,
|
| 8216 |
+
"loss": 1.4276,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.3949491110266481,
|
| 8221 |
+
"grad_norm": 0.2353004813194275,
|
| 8222 |
+
"learning_rate": 2.3380499051905137e-05,
|
| 8223 |
+
"loss": 1.5205,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.395286963131889,
|
| 8228 |
+
"grad_norm": 0.23783273994922638,
|
| 8229 |
+
"learning_rate": 2.324517865697501e-05,
|
| 8230 |
+
"loss": 1.3945,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.39562481523712995,
|
| 8235 |
+
"grad_norm": 0.24038663506507874,
|
| 8236 |
+
"learning_rate": 2.3110199480657525e-05,
|
| 8237 |
+
"loss": 1.4852,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.3959626673423709,
|
| 8242 |
+
"grad_norm": 0.23865370452404022,
|
| 8243 |
+
"learning_rate": 2.2975562123011495e-05,
|
| 8244 |
+
"loss": 1.4707,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.3963005194476118,
|
| 8249 |
+
"grad_norm": 0.24839963018894196,
|
| 8250 |
+
"learning_rate": 2.2841267182576143e-05,
|
| 8251 |
+
"loss": 1.369,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.39663837155285275,
|
| 8256 |
+
"grad_norm": 0.2446480244398117,
|
| 8257 |
+
"learning_rate": 2.2707315256368433e-05,
|
| 8258 |
+
"loss": 1.4312,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.39697622365809365,
|
| 8263 |
+
"grad_norm": 0.23917041718959808,
|
| 8264 |
+
"learning_rate": 2.2573706939880555e-05,
|
| 8265 |
+
"loss": 1.3773,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.3973140757633346,
|
| 8270 |
+
"grad_norm": 0.2481280118227005,
|
| 8271 |
+
"learning_rate": 2.2440442827077045e-05,
|
| 8272 |
+
"loss": 1.4452,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.3976519278685755,
|
| 8277 |
+
"grad_norm": 0.24594064056873322,
|
| 8278 |
+
"learning_rate": 2.230752351039228e-05,
|
| 8279 |
+
"loss": 1.4282,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.39798977997381646,
|
| 8284 |
+
"grad_norm": 0.25568026304244995,
|
| 8285 |
+
"learning_rate": 2.2174949580727832e-05,
|
| 8286 |
+
"loss": 1.5368,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.3983276320790574,
|
| 8291 |
+
"grad_norm": 0.25840944051742554,
|
| 8292 |
+
"learning_rate": 2.2042721627449846e-05,
|
| 8293 |
+
"loss": 1.4432,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.3986654841842983,
|
| 8298 |
+
"grad_norm": 0.2656974792480469,
|
| 8299 |
+
"learning_rate": 2.1910840238386398e-05,
|
| 8300 |
+
"loss": 1.4133,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.39900333628953927,
|
| 8305 |
+
"grad_norm": 0.2548368275165558,
|
| 8306 |
+
"learning_rate": 2.1779305999824884e-05,
|
| 8307 |
+
"loss": 1.447,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.39934118839478017,
|
| 8312 |
+
"grad_norm": 0.25699570775032043,
|
| 8313 |
+
"learning_rate": 2.164811949650942e-05,
|
| 8314 |
+
"loss": 1.4677,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.3996790405000211,
|
| 8319 |
+
"grad_norm": 0.2600949704647064,
|
| 8320 |
+
"learning_rate": 2.1517281311638217e-05,
|
| 8321 |
+
"loss": 1.3731,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.400016892605262,
|
| 8326 |
+
"grad_norm": 0.2724064290523529,
|
| 8327 |
+
"learning_rate": 2.1386792026861103e-05,
|
| 8328 |
+
"loss": 1.4862,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.400354744710503,
|
| 8333 |
+
"grad_norm": 0.26320120692253113,
|
| 8334 |
+
"learning_rate": 2.125665222227675e-05,
|
| 8335 |
+
"loss": 1.4296,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.40069259681574393,
|
| 8340 |
+
"grad_norm": 0.2696899473667145,
|
| 8341 |
+
"learning_rate": 2.112686247643024e-05,
|
| 8342 |
+
"loss": 1.4762,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.40103044892098483,
|
| 8347 |
+
"grad_norm": 0.2627149522304535,
|
| 8348 |
+
"learning_rate": 2.09974233663104e-05,
|
| 8349 |
+
"loss": 1.4566,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.4013683010262258,
|
| 8354 |
+
"grad_norm": 0.25996828079223633,
|
| 8355 |
+
"learning_rate": 2.0868335467347366e-05,
|
| 8356 |
+
"loss": 1.3959,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.4017061531314667,
|
| 8361 |
+
"grad_norm": 0.2771797478199005,
|
| 8362 |
+
"learning_rate": 2.073959935340988e-05,
|
| 8363 |
+
"loss": 1.5015,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.40204400523670764,
|
| 8368 |
+
"grad_norm": 0.274434357881546,
|
| 8369 |
+
"learning_rate": 2.06112155968028e-05,
|
| 8370 |
+
"loss": 1.4836,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.40238185734194853,
|
| 8375 |
+
"grad_norm": 0.2731093764305115,
|
| 8376 |
+
"learning_rate": 2.0483184768264596e-05,
|
| 8377 |
+
"loss": 1.4847,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.4027197094471895,
|
| 8382 |
+
"grad_norm": 0.2809274196624756,
|
| 8383 |
+
"learning_rate": 2.035550743696468e-05,
|
| 8384 |
+
"loss": 1.5361,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.40305756155243044,
|
| 8389 |
+
"grad_norm": 0.2769002914428711,
|
| 8390 |
+
"learning_rate": 2.022818417050113e-05,
|
| 8391 |
+
"loss": 1.4824,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.40339541365767134,
|
| 8396 |
+
"grad_norm": 0.2797887325286865,
|
| 8397 |
+
"learning_rate": 2.0101215534897855e-05,
|
| 8398 |
+
"loss": 1.4485,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.4037332657629123,
|
| 8403 |
+
"grad_norm": 0.28619834780693054,
|
| 8404 |
+
"learning_rate": 1.99746020946023e-05,
|
| 8405 |
+
"loss": 1.4364,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.4040711178681532,
|
| 8410 |
+
"grad_norm": 0.2799307703971863,
|
| 8411 |
+
"learning_rate": 1.9848344412482854e-05,
|
| 8412 |
+
"loss": 1.4896,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.40440896997339415,
|
| 8417 |
+
"grad_norm": 0.2994498610496521,
|
| 8418 |
+
"learning_rate": 1.9722443049826344e-05,
|
| 8419 |
+
"loss": 1.4138,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.4047468220786351,
|
| 8424 |
+
"grad_norm": 0.30917614698410034,
|
| 8425 |
+
"learning_rate": 1.9596898566335576e-05,
|
| 8426 |
+
"loss": 1.4878,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.405084674183876,
|
| 8431 |
+
"grad_norm": 0.3226688802242279,
|
| 8432 |
+
"learning_rate": 1.9471711520126824e-05,
|
| 8433 |
+
"loss": 1.3941,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.40542252628911696,
|
| 8438 |
+
"grad_norm": 0.4267667829990387,
|
| 8439 |
+
"learning_rate": 1.9346882467727325e-05,
|
| 8440 |
+
"loss": 1.4196,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.40576037839435786,
|
| 8445 |
+
"grad_norm": 0.20707213878631592,
|
| 8446 |
+
"learning_rate": 1.9222411964072884e-05,
|
| 8447 |
+
"loss": 1.51,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.4060982304995988,
|
| 8452 |
+
"grad_norm": 0.21047216653823853,
|
| 8453 |
+
"learning_rate": 1.9098300562505266e-05,
|
| 8454 |
+
"loss": 1.5016,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.4064360826048397,
|
| 8459 |
+
"grad_norm": 0.20799440145492554,
|
| 8460 |
+
"learning_rate": 1.8974548814769944e-05,
|
| 8461 |
+
"loss": 1.4707,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.40677393471008066,
|
| 8466 |
+
"grad_norm": 0.20645122230052948,
|
| 8467 |
+
"learning_rate": 1.8851157271013442e-05,
|
| 8468 |
+
"loss": 1.4439,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.4071117868153216,
|
| 8473 |
+
"grad_norm": 0.2075306624174118,
|
| 8474 |
+
"learning_rate": 1.872812647978095e-05,
|
| 8475 |
+
"loss": 1.4627,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.4074496389205625,
|
| 8480 |
+
"grad_norm": 0.20466361939907074,
|
| 8481 |
+
"learning_rate": 1.8605456988014015e-05,
|
| 8482 |
+
"loss": 1.3827,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.4077874910258035,
|
| 8487 |
+
"grad_norm": 0.20638106763362885,
|
| 8488 |
+
"learning_rate": 1.8483149341047923e-05,
|
| 8489 |
+
"loss": 1.4561,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.40812534313104437,
|
| 8494 |
+
"grad_norm": 0.21052102744579315,
|
| 8495 |
+
"learning_rate": 1.8361204082609352e-05,
|
| 8496 |
+
"loss": 1.5199,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.4084631952362853,
|
| 8501 |
+
"grad_norm": 0.21788109838962555,
|
| 8502 |
+
"learning_rate": 1.8239621754813995e-05,
|
| 8503 |
+
"loss": 1.4561,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.4088010473415262,
|
| 8508 |
+
"grad_norm": 0.22226934134960175,
|
| 8509 |
+
"learning_rate": 1.811840289816409e-05,
|
| 8510 |
+
"loss": 1.4802,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.4091388994467672,
|
| 8515 |
+
"grad_norm": 0.2264350950717926,
|
| 8516 |
+
"learning_rate": 1.799754805154603e-05,
|
| 8517 |
+
"loss": 1.5255,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.40947675155200813,
|
| 8522 |
+
"grad_norm": 0.22181668877601624,
|
| 8523 |
+
"learning_rate": 1.787705775222802e-05,
|
| 8524 |
+
"loss": 1.3917,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.40981460365724903,
|
| 8529 |
+
"grad_norm": 0.23737621307373047,
|
| 8530 |
+
"learning_rate": 1.775693253585763e-05,
|
| 8531 |
+
"loss": 1.4871,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.41015245576249,
|
| 8536 |
+
"grad_norm": 0.23132474720478058,
|
| 8537 |
+
"learning_rate": 1.763717293645939e-05,
|
| 8538 |
+
"loss": 1.4626,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.4104903078677309,
|
| 8543 |
+
"grad_norm": 0.22525909543037415,
|
| 8544 |
+
"learning_rate": 1.7517779486432495e-05,
|
| 8545 |
+
"loss": 1.3503,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.41082815997297184,
|
| 8550 |
+
"grad_norm": 0.22630459070205688,
|
| 8551 |
+
"learning_rate": 1.7398752716548395e-05,
|
| 8552 |
+
"loss": 1.4196,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.41116601207821274,
|
| 8557 |
+
"grad_norm": 0.234827920794487,
|
| 8558 |
+
"learning_rate": 1.728009315594843e-05,
|
| 8559 |
+
"loss": 1.489,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.4115038641834537,
|
| 8564 |
+
"grad_norm": 0.23307843506336212,
|
| 8565 |
+
"learning_rate": 1.716180133214149e-05,
|
| 8566 |
+
"loss": 1.4527,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.41184171628869465,
|
| 8571 |
+
"grad_norm": 0.22875098884105682,
|
| 8572 |
+
"learning_rate": 1.704387777100165e-05,
|
| 8573 |
+
"loss": 1.413,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.41217956839393555,
|
| 8578 |
+
"grad_norm": 0.2398001253604889,
|
| 8579 |
+
"learning_rate": 1.6926322996765897e-05,
|
| 8580 |
+
"loss": 1.4631,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.4125174204991765,
|
| 8585 |
+
"grad_norm": 0.23415181040763855,
|
| 8586 |
+
"learning_rate": 1.6809137532031704e-05,
|
| 8587 |
+
"loss": 1.3652,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.4128552726044174,
|
| 8592 |
+
"grad_norm": 0.24499595165252686,
|
| 8593 |
+
"learning_rate": 1.6692321897754758e-05,
|
| 8594 |
+
"loss": 1.487,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.41319312470965835,
|
| 8599 |
+
"grad_norm": 0.2394736111164093,
|
| 8600 |
+
"learning_rate": 1.65758766132467e-05,
|
| 8601 |
+
"loss": 1.4251,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.41353097681489925,
|
| 8606 |
+
"grad_norm": 0.24736353754997253,
|
| 8607 |
+
"learning_rate": 1.6459802196172668e-05,
|
| 8608 |
+
"loss": 1.4362,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.4138688289201402,
|
| 8613 |
+
"grad_norm": 0.26526200771331787,
|
| 8614 |
+
"learning_rate": 1.634409916254914e-05,
|
| 8615 |
+
"loss": 1.5483,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.41420668102538116,
|
| 8620 |
+
"grad_norm": 0.25085341930389404,
|
| 8621 |
+
"learning_rate": 1.622876802674158e-05,
|
| 8622 |
+
"loss": 1.428,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.41454453313062206,
|
| 8627 |
+
"grad_norm": 0.2499944567680359,
|
| 8628 |
+
"learning_rate": 1.6113809301462125e-05,
|
| 8629 |
+
"loss": 1.416,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.414882385235863,
|
| 8634 |
+
"grad_norm": 0.2523176968097687,
|
| 8635 |
+
"learning_rate": 1.599922349776738e-05,
|
| 8636 |
+
"loss": 1.4418,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.4152202373411039,
|
| 8641 |
+
"grad_norm": 0.2542124092578888,
|
| 8642 |
+
"learning_rate": 1.5885011125056047e-05,
|
| 8643 |
+
"loss": 1.4081,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.41555808944634487,
|
| 8648 |
+
"grad_norm": 0.2558896243572235,
|
| 8649 |
+
"learning_rate": 1.5771172691066794e-05,
|
| 8650 |
+
"loss": 1.4147,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.41589594155158577,
|
| 8655 |
+
"grad_norm": 0.2498987764120102,
|
| 8656 |
+
"learning_rate": 1.565770870187585e-05,
|
| 8657 |
+
"loss": 1.4303,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.4162337936568267,
|
| 8662 |
+
"grad_norm": 0.2590921223163605,
|
| 8663 |
+
"learning_rate": 1.5544619661894864e-05,
|
| 8664 |
+
"loss": 1.4951,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.4165716457620677,
|
| 8669 |
+
"grad_norm": 0.2667752802371979,
|
| 8670 |
+
"learning_rate": 1.543190607386861e-05,
|
| 8671 |
+
"loss": 1.4777,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.4169094978673086,
|
| 8676 |
+
"grad_norm": 0.26036345958709717,
|
| 8677 |
+
"learning_rate": 1.5319568438872745e-05,
|
| 8678 |
+
"loss": 1.4496,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.41724734997254953,
|
| 8683 |
+
"grad_norm": 0.2599640190601349,
|
| 8684 |
+
"learning_rate": 1.520760725631164e-05,
|
| 8685 |
+
"loss": 1.482,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.41758520207779043,
|
| 8690 |
+
"grad_norm": 0.27320942282676697,
|
| 8691 |
+
"learning_rate": 1.5096023023916094e-05,
|
| 8692 |
+
"loss": 1.4464,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.4179230541830314,
|
| 8697 |
+
"grad_norm": 0.2667524814605713,
|
| 8698 |
+
"learning_rate": 1.498481623774115e-05,
|
| 8699 |
+
"loss": 1.4874,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.4182609062882723,
|
| 8704 |
+
"grad_norm": 0.2656776010990143,
|
| 8705 |
+
"learning_rate": 1.4873987392163947e-05,
|
| 8706 |
+
"loss": 1.4351,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.41859875839351324,
|
| 8711 |
+
"grad_norm": 0.2779434323310852,
|
| 8712 |
+
"learning_rate": 1.4763536979881354e-05,
|
| 8713 |
+
"loss": 1.5735,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.4189366104987542,
|
| 8718 |
+
"grad_norm": 0.26216769218444824,
|
| 8719 |
+
"learning_rate": 1.4653465491908003e-05,
|
| 8720 |
+
"loss": 1.3866,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.4192744626039951,
|
| 8725 |
+
"grad_norm": 0.2756368815898895,
|
| 8726 |
+
"learning_rate": 1.4543773417573925e-05,
|
| 8727 |
+
"loss": 1.4764,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.41961231470923605,
|
| 8732 |
+
"grad_norm": 0.27156350016593933,
|
| 8733 |
+
"learning_rate": 1.4434461244522458e-05,
|
| 8734 |
+
"loss": 1.4217,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.41995016681447694,
|
| 8739 |
+
"grad_norm": 0.28269901871681213,
|
| 8740 |
+
"learning_rate": 1.4325529458708065e-05,
|
| 8741 |
+
"loss": 1.463,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.4202880189197179,
|
| 8746 |
+
"grad_norm": 0.2808777093887329,
|
| 8747 |
+
"learning_rate": 1.4216978544394177e-05,
|
| 8748 |
+
"loss": 1.4329,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.4206258710249588,
|
| 8753 |
+
"grad_norm": 0.2859387695789337,
|
| 8754 |
+
"learning_rate": 1.4108808984151023e-05,
|
| 8755 |
+
"loss": 1.5526,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.42096372313019975,
|
| 8760 |
+
"grad_norm": 0.28296807408332825,
|
| 8761 |
+
"learning_rate": 1.4001021258853509e-05,
|
| 8762 |
+
"loss": 1.4057,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.4213015752354407,
|
| 8767 |
+
"grad_norm": 0.28596076369285583,
|
| 8768 |
+
"learning_rate": 1.3893615847679065e-05,
|
| 8769 |
+
"loss": 1.4185,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.4216394273406816,
|
| 8774 |
+
"grad_norm": 0.2941034734249115,
|
| 8775 |
+
"learning_rate": 1.3786593228105494e-05,
|
| 8776 |
+
"loss": 1.3956,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.42197727944592256,
|
| 8781 |
+
"grad_norm": 0.3489391803741455,
|
| 8782 |
+
"learning_rate": 1.3679953875908957e-05,
|
| 8783 |
+
"loss": 1.4171,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.42231513155116346,
|
| 8788 |
+
"grad_norm": 0.4632200300693512,
|
| 8789 |
+
"learning_rate": 1.3573698265161683e-05,
|
| 8790 |
+
"loss": 1.4153,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.4226529836564044,
|
| 8795 |
+
"grad_norm": 0.18695835769176483,
|
| 8796 |
+
"learning_rate": 1.3467826868229994e-05,
|
| 8797 |
+
"loss": 1.4056,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.42299083576164537,
|
| 8802 |
+
"grad_norm": 0.20138730108737946,
|
| 8803 |
+
"learning_rate": 1.3362340155772146e-05,
|
| 8804 |
+
"loss": 1.4695,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.42332868786688627,
|
| 8809 |
+
"grad_norm": 0.2032420039176941,
|
| 8810 |
+
"learning_rate": 1.3257238596736266e-05,
|
| 8811 |
+
"loss": 1.4015,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.4236665399721272,
|
| 8816 |
+
"grad_norm": 0.202083021402359,
|
| 8817 |
+
"learning_rate": 1.3152522658358245e-05,
|
| 8818 |
+
"loss": 1.4081,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.4240043920773681,
|
| 8823 |
+
"grad_norm": 0.20888225734233856,
|
| 8824 |
+
"learning_rate": 1.3048192806159721e-05,
|
| 8825 |
+
"loss": 1.5185,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.4243422441826091,
|
| 8830 |
+
"grad_norm": 0.2090698480606079,
|
| 8831 |
+
"learning_rate": 1.2944249503945894e-05,
|
| 8832 |
+
"loss": 1.4583,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.42468009628785,
|
| 8837 |
+
"grad_norm": 0.20110248029232025,
|
| 8838 |
+
"learning_rate": 1.2840693213803545e-05,
|
| 8839 |
+
"loss": 1.4213,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.4250179483930909,
|
| 8844 |
+
"grad_norm": 0.20445598661899567,
|
| 8845 |
+
"learning_rate": 1.2737524396099032e-05,
|
| 8846 |
+
"loss": 1.3881,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.4253558004983319,
|
| 8851 |
+
"grad_norm": 0.20724329352378845,
|
| 8852 |
+
"learning_rate": 1.2634743509476088e-05,
|
| 8853 |
+
"loss": 1.3841,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.4256936526035728,
|
| 8858 |
+
"grad_norm": 0.22444471716880798,
|
| 8859 |
+
"learning_rate": 1.2532351010853916e-05,
|
| 8860 |
+
"loss": 1.4913,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.42603150470881374,
|
| 8865 |
+
"grad_norm": 0.21949529647827148,
|
| 8866 |
+
"learning_rate": 1.243034735542512e-05,
|
| 8867 |
+
"loss": 1.3538,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.42636935681405463,
|
| 8872 |
+
"grad_norm": 0.22307273745536804,
|
| 8873 |
+
"learning_rate": 1.2328732996653669e-05,
|
| 8874 |
+
"loss": 1.4007,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.4267072089192956,
|
| 8879 |
+
"grad_norm": 0.2236008495092392,
|
| 8880 |
+
"learning_rate": 1.2227508386272878e-05,
|
| 8881 |
+
"loss": 1.4903,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.4270450610245365,
|
| 8886 |
+
"grad_norm": 0.22181101143360138,
|
| 8887 |
+
"learning_rate": 1.212667397428342e-05,
|
| 8888 |
+
"loss": 1.3518,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.42738291312977744,
|
| 8893 |
+
"grad_norm": 0.2172001749277115,
|
| 8894 |
+
"learning_rate": 1.2026230208951306e-05,
|
| 8895 |
+
"loss": 1.4029,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.4277207652350184,
|
| 8900 |
+
"grad_norm": 0.2310638725757599,
|
| 8901 |
+
"learning_rate": 1.1926177536805905e-05,
|
| 8902 |
+
"loss": 1.4582,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.4280586173402593,
|
| 8907 |
+
"grad_norm": 0.24587257206439972,
|
| 8908 |
+
"learning_rate": 1.1826516402637989e-05,
|
| 8909 |
+
"loss": 1.5345,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.42839646944550025,
|
| 8914 |
+
"grad_norm": 0.2233538180589676,
|
| 8915 |
+
"learning_rate": 1.1727247249497685e-05,
|
| 8916 |
+
"loss": 1.3496,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.42873432155074115,
|
| 8921 |
+
"grad_norm": 0.23955301940441132,
|
| 8922 |
+
"learning_rate": 1.1628370518692533e-05,
|
| 8923 |
+
"loss": 1.5096,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.4290721736559821,
|
| 8928 |
+
"grad_norm": 0.24007707834243774,
|
| 8929 |
+
"learning_rate": 1.152988664978556e-05,
|
| 8930 |
+
"loss": 1.4225,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.429410025761223,
|
| 8935 |
+
"grad_norm": 0.24410776793956757,
|
| 8936 |
+
"learning_rate": 1.1431796080593283e-05,
|
| 8937 |
+
"loss": 1.4288,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.42974787786646396,
|
| 8942 |
+
"grad_norm": 0.24154405295848846,
|
| 8943 |
+
"learning_rate": 1.1334099247183783e-05,
|
| 8944 |
+
"loss": 1.418,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.4300857299717049,
|
| 8949 |
+
"grad_norm": 0.24884504079818726,
|
| 8950 |
+
"learning_rate": 1.1236796583874787e-05,
|
| 8951 |
+
"loss": 1.4265,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.4304235820769458,
|
| 8956 |
+
"grad_norm": 0.24534472823143005,
|
| 8957 |
+
"learning_rate": 1.1139888523231678e-05,
|
| 8958 |
+
"loss": 1.4183,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.43076143418218676,
|
| 8963 |
+
"grad_norm": 0.24859485030174255,
|
| 8964 |
+
"learning_rate": 1.1043375496065611e-05,
|
| 8965 |
+
"loss": 1.4682,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.43109928628742766,
|
| 8970 |
+
"grad_norm": 0.25250065326690674,
|
| 8971 |
+
"learning_rate": 1.0947257931431642e-05,
|
| 8972 |
+
"loss": 1.4602,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.4314371383926686,
|
| 8977 |
+
"grad_norm": 0.25797921419143677,
|
| 8978 |
+
"learning_rate": 1.0851536256626705e-05,
|
| 8979 |
+
"loss": 1.4245,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.4317749904979095,
|
| 8984 |
+
"grad_norm": 0.2544008791446686,
|
| 8985 |
+
"learning_rate": 1.0756210897187812e-05,
|
| 8986 |
+
"loss": 1.4412,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.43211284260315047,
|
| 8991 |
+
"grad_norm": 0.2455921173095703,
|
| 8992 |
+
"learning_rate": 1.0661282276890127e-05,
|
| 8993 |
+
"loss": 1.3871,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.4324506947083914,
|
| 8998 |
+
"grad_norm": 0.2514779269695282,
|
| 8999 |
+
"learning_rate": 1.0566750817745074e-05,
|
| 9000 |
+
"loss": 1.4301,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.4327885468136323,
|
| 9005 |
+
"grad_norm": 0.2560863792896271,
|
| 9006 |
+
"learning_rate": 1.0472616939998492e-05,
|
| 9007 |
+
"loss": 1.371,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.4331263989188733,
|
| 9012 |
+
"grad_norm": 0.2695235311985016,
|
| 9013 |
+
"learning_rate": 1.0378881062128731e-05,
|
| 9014 |
+
"loss": 1.4691,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.4334642510241142,
|
| 9019 |
+
"grad_norm": 0.26122233271598816,
|
| 9020 |
+
"learning_rate": 1.0285543600844804e-05,
|
| 9021 |
+
"loss": 1.4662,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.43380210312935513,
|
| 9026 |
+
"grad_norm": 0.2570619285106659,
|
| 9027 |
+
"learning_rate": 1.019260497108453e-05,
|
| 9028 |
+
"loss": 1.4487,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.43413995523459603,
|
| 9033 |
+
"grad_norm": 0.2697749137878418,
|
| 9034 |
+
"learning_rate": 1.010006558601274e-05,
|
| 9035 |
+
"loss": 1.5578,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.434477807339837,
|
| 9040 |
+
"grad_norm": 0.2633000314235687,
|
| 9041 |
+
"learning_rate": 1.000792585701934e-05,
|
| 9042 |
+
"loss": 1.4546,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.43481565944507794,
|
| 9047 |
+
"grad_norm": 0.2652289569377899,
|
| 9048 |
+
"learning_rate": 9.91618619371757e-06,
|
| 9049 |
+
"loss": 1.4218,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.43515351155031884,
|
| 9054 |
+
"grad_norm": 0.26297253370285034,
|
| 9055 |
+
"learning_rate": 9.82484700394215e-06,
|
| 9056 |
+
"loss": 1.3576,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.4354913636555598,
|
| 9061 |
+
"grad_norm": 0.2698511779308319,
|
| 9062 |
+
"learning_rate": 9.73390869374743e-06,
|
| 9063 |
+
"loss": 1.4131,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.4358292157608007,
|
| 9068 |
+
"grad_norm": 0.26395317912101746,
|
| 9069 |
+
"learning_rate": 9.643371667405698e-06,
|
| 9070 |
+
"loss": 1.4107,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.43616706786604165,
|
| 9075 |
+
"grad_norm": 0.27847278118133545,
|
| 9076 |
+
"learning_rate": 9.553236327405246e-06,
|
| 9077 |
+
"loss": 1.4913,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.43650491997128255,
|
| 9082 |
+
"grad_norm": 0.2765458822250366,
|
| 9083 |
+
"learning_rate": 9.463503074448677e-06,
|
| 9084 |
+
"loss": 1.3891,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.4368427720765235,
|
| 9089 |
+
"grad_norm": 0.28723907470703125,
|
| 9090 |
+
"learning_rate": 9.374172307451068e-06,
|
| 9091 |
+
"loss": 1.4339,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.43718062418176445,
|
| 9096 |
+
"grad_norm": 0.2864030599594116,
|
| 9097 |
+
"learning_rate": 9.285244423538197e-06,
|
| 9098 |
+
"loss": 1.4475,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.43751847628700535,
|
| 9103 |
+
"grad_norm": 0.2855558693408966,
|
| 9104 |
+
"learning_rate": 9.196719818044886e-06,
|
| 9105 |
+
"loss": 1.46,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.4378563283922463,
|
| 9110 |
+
"grad_norm": 0.2890433967113495,
|
| 9111 |
+
"learning_rate": 9.108598884513053e-06,
|
| 9112 |
+
"loss": 1.5384,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.4381941804974872,
|
| 9117 |
+
"grad_norm": 0.30390143394470215,
|
| 9118 |
+
"learning_rate": 9.020882014690136e-06,
|
| 9119 |
+
"loss": 1.4686,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.43853203260272816,
|
| 9124 |
+
"grad_norm": 0.3386201560497284,
|
| 9125 |
+
"learning_rate": 8.933569598527247e-06,
|
| 9126 |
+
"loss": 1.4845,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.43886988470796906,
|
| 9131 |
+
"grad_norm": 0.3602554500102997,
|
| 9132 |
+
"learning_rate": 8.846662024177477e-06,
|
| 9133 |
+
"loss": 1.3967,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.43920773681321,
|
| 9138 |
+
"grad_norm": 0.5023772716522217,
|
| 9139 |
+
"learning_rate": 8.760159677994172e-06,
|
| 9140 |
+
"loss": 1.2609,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.43954558891845097,
|
| 9145 |
+
"grad_norm": 0.19685684144496918,
|
| 9146 |
+
"learning_rate": 8.674062944529216e-06,
|
| 9147 |
+
"loss": 1.4367,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.43988344102369187,
|
| 9152 |
+
"grad_norm": 0.20138689875602722,
|
| 9153 |
+
"learning_rate": 8.588372206531292e-06,
|
| 9154 |
+
"loss": 1.4906,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.4402212931289328,
|
| 9159 |
+
"grad_norm": 0.2059500813484192,
|
| 9160 |
+
"learning_rate": 8.503087844944213e-06,
|
| 9161 |
+
"loss": 1.5427,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.4405591452341737,
|
| 9166 |
+
"grad_norm": 0.20476102828979492,
|
| 9167 |
+
"learning_rate": 8.418210238905256e-06,
|
| 9168 |
+
"loss": 1.4225,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.4408969973394147,
|
| 9173 |
+
"grad_norm": 0.20567205548286438,
|
| 9174 |
+
"learning_rate": 8.333739765743398e-06,
|
| 9175 |
+
"loss": 1.4374,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.4412348494446556,
|
| 9180 |
+
"grad_norm": 0.19746853411197662,
|
| 9181 |
+
"learning_rate": 8.249676800977658e-06,
|
| 9182 |
+
"loss": 1.3952,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.44157270154989653,
|
| 9187 |
+
"grad_norm": 0.21089419722557068,
|
| 9188 |
+
"learning_rate": 8.16602171831553e-06,
|
| 9189 |
+
"loss": 1.5382,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.4419105536551375,
|
| 9194 |
+
"grad_norm": 0.21633197367191315,
|
| 9195 |
+
"learning_rate": 8.082774889651168e-06,
|
| 9196 |
+
"loss": 1.4424,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.4422484057603784,
|
| 9201 |
+
"grad_norm": 0.21298642456531525,
|
| 9202 |
+
"learning_rate": 7.999936685063835e-06,
|
| 9203 |
+
"loss": 1.3776,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.44258625786561934,
|
| 9208 |
+
"grad_norm": 0.21993917226791382,
|
| 9209 |
+
"learning_rate": 7.91750747281621e-06,
|
| 9210 |
+
"loss": 1.4607,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.44292410997086024,
|
| 9215 |
+
"grad_norm": 0.21888743340969086,
|
| 9216 |
+
"learning_rate": 7.835487619352811e-06,
|
| 9217 |
+
"loss": 1.4236,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.4432619620761012,
|
| 9222 |
+
"grad_norm": 0.22910566627979279,
|
| 9223 |
+
"learning_rate": 7.753877489298244e-06,
|
| 9224 |
+
"loss": 1.4937,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.44359981418134214,
|
| 9229 |
+
"grad_norm": 0.23162463307380676,
|
| 9230 |
+
"learning_rate": 7.67267744545579e-06,
|
| 9231 |
+
"loss": 1.4616,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.44393766628658304,
|
| 9236 |
+
"grad_norm": 0.23120923340320587,
|
| 9237 |
+
"learning_rate": 7.591887848805545e-06,
|
| 9238 |
+
"loss": 1.5042,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.444275518391824,
|
| 9243 |
+
"grad_norm": 0.2324828952550888,
|
| 9244 |
+
"learning_rate": 7.5115090585029966e-06,
|
| 9245 |
+
"loss": 1.4422,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.4446133704970649,
|
| 9250 |
+
"grad_norm": 0.22663557529449463,
|
| 9251 |
+
"learning_rate": 7.431541431877342e-06,
|
| 9252 |
+
"loss": 1.3462,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.44495122260230585,
|
| 9257 |
+
"grad_norm": 0.22750574350357056,
|
| 9258 |
+
"learning_rate": 7.351985324429933e-06,
|
| 9259 |
+
"loss": 1.3832,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.44528907470754675,
|
| 9264 |
+
"grad_norm": 0.2285628318786621,
|
| 9265 |
+
"learning_rate": 7.272841089832694e-06,
|
| 9266 |
+
"loss": 1.4929,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.4456269268127877,
|
| 9271 |
+
"grad_norm": 0.2365717887878418,
|
| 9272 |
+
"learning_rate": 7.194109079926514e-06,
|
| 9273 |
+
"loss": 1.4422,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.44596477891802866,
|
| 9278 |
+
"grad_norm": 0.23710958659648895,
|
| 9279 |
+
"learning_rate": 7.115789644719728e-06,
|
| 9280 |
+
"loss": 1.4292,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.44630263102326956,
|
| 9285 |
+
"grad_norm": 0.2324679046869278,
|
| 9286 |
+
"learning_rate": 7.037883132386547e-06,
|
| 9287 |
+
"loss": 1.4281,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.4466404831285105,
|
| 9292 |
+
"grad_norm": 0.24180293083190918,
|
| 9293 |
+
"learning_rate": 6.960389889265517e-06,
|
| 9294 |
+
"loss": 1.436,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.4469783352337514,
|
| 9299 |
+
"grad_norm": 0.2439158707857132,
|
| 9300 |
+
"learning_rate": 6.883310259857944e-06,
|
| 9301 |
+
"loss": 1.3984,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.44731618733899237,
|
| 9306 |
+
"grad_norm": 0.24821530282497406,
|
| 9307 |
+
"learning_rate": 6.806644586826383e-06,
|
| 9308 |
+
"loss": 1.398,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.44765403944423326,
|
| 9313 |
+
"grad_norm": 0.24612189829349518,
|
| 9314 |
+
"learning_rate": 6.730393210993147e-06,
|
| 9315 |
+
"loss": 1.4308,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.4479918915494742,
|
| 9320 |
+
"grad_norm": 0.24643537402153015,
|
| 9321 |
+
"learning_rate": 6.654556471338746e-06,
|
| 9322 |
+
"loss": 1.462,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.4483297436547152,
|
| 9327 |
+
"grad_norm": 0.25232747197151184,
|
| 9328 |
+
"learning_rate": 6.579134705000412e-06,
|
| 9329 |
+
"loss": 1.4424,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.4486675957599561,
|
| 9334 |
+
"grad_norm": 0.2574306130409241,
|
| 9335 |
+
"learning_rate": 6.504128247270546e-06,
|
| 9336 |
+
"loss": 1.4867,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.449005447865197,
|
| 9341 |
+
"grad_norm": 0.24852421879768372,
|
| 9342 |
+
"learning_rate": 6.429537431595312e-06,
|
| 9343 |
+
"loss": 1.3959,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.4493432999704379,
|
| 9348 |
+
"grad_norm": 0.2500188946723938,
|
| 9349 |
+
"learning_rate": 6.355362589573077e-06,
|
| 9350 |
+
"loss": 1.4143,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.4496811520756789,
|
| 9355 |
+
"grad_norm": 0.2519555985927582,
|
| 9356 |
+
"learning_rate": 6.2816040509530165e-06,
|
| 9357 |
+
"loss": 1.4464,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.4500190041809198,
|
| 9362 |
+
"grad_norm": 0.2661184072494507,
|
| 9363 |
+
"learning_rate": 6.2082621436335475e-06,
|
| 9364 |
+
"loss": 1.4803,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.45035685628616073,
|
| 9369 |
+
"grad_norm": 0.26744377613067627,
|
| 9370 |
+
"learning_rate": 6.135337193660962e-06,
|
| 9371 |
+
"loss": 1.4683,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.4506947083914017,
|
| 9376 |
+
"grad_norm": 0.2750174105167389,
|
| 9377 |
+
"learning_rate": 6.062829525227909e-06,
|
| 9378 |
+
"loss": 1.597,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.4510325604966426,
|
| 9383 |
+
"grad_norm": 0.25979262590408325,
|
| 9384 |
+
"learning_rate": 5.990739460672024e-06,
|
| 9385 |
+
"loss": 1.4443,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.45137041260188354,
|
| 9390 |
+
"grad_norm": 0.2599670886993408,
|
| 9391 |
+
"learning_rate": 5.9190673204744255e-06,
|
| 9392 |
+
"loss": 1.3954,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.45170826470712444,
|
| 9397 |
+
"grad_norm": 0.2586192190647125,
|
| 9398 |
+
"learning_rate": 5.84781342325833e-06,
|
| 9399 |
+
"loss": 1.4169,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.4520461168123654,
|
| 9404 |
+
"grad_norm": 0.25740161538124084,
|
| 9405 |
+
"learning_rate": 5.77697808578761e-06,
|
| 9406 |
+
"loss": 1.4083,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.4523839689176063,
|
| 9411 |
+
"grad_norm": 0.261178195476532,
|
| 9412 |
+
"learning_rate": 5.706561622965467e-06,
|
| 9413 |
+
"loss": 1.3997,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.45272182102284725,
|
| 9418 |
+
"grad_norm": 0.2803891599178314,
|
| 9419 |
+
"learning_rate": 5.636564347832907e-06,
|
| 9420 |
+
"loss": 1.4765,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.4530596731280882,
|
| 9425 |
+
"grad_norm": 0.27532878518104553,
|
| 9426 |
+
"learning_rate": 5.566986571567401e-06,
|
| 9427 |
+
"loss": 1.5001,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.4533975252333291,
|
| 9432 |
+
"grad_norm": 0.2747570872306824,
|
| 9433 |
+
"learning_rate": 5.497828603481569e-06,
|
| 9434 |
+
"loss": 1.4494,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.45373537733857006,
|
| 9439 |
+
"grad_norm": 0.28951317071914673,
|
| 9440 |
+
"learning_rate": 5.429090751021704e-06,
|
| 9441 |
+
"loss": 1.531,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.45407322944381096,
|
| 9446 |
+
"grad_norm": 0.2831999957561493,
|
| 9447 |
+
"learning_rate": 5.3607733197664436e-06,
|
| 9448 |
+
"loss": 1.3793,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.4544110815490519,
|
| 9453 |
+
"grad_norm": 0.294784277677536,
|
| 9454 |
+
"learning_rate": 5.2928766134254345e-06,
|
| 9455 |
+
"loss": 1.4295,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.4547489336542928,
|
| 9460 |
+
"grad_norm": 0.302503377199173,
|
| 9461 |
+
"learning_rate": 5.225400933837954e-06,
|
| 9462 |
+
"loss": 1.5096,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.45508678575953376,
|
| 9467 |
+
"grad_norm": 0.30284932255744934,
|
| 9468 |
+
"learning_rate": 5.158346580971573e-06,
|
| 9469 |
+
"loss": 1.468,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.4554246378647747,
|
| 9474 |
+
"grad_norm": 0.32171115279197693,
|
| 9475 |
+
"learning_rate": 5.091713852920854e-06,
|
| 9476 |
+
"loss": 1.4164,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.4557624899700156,
|
| 9481 |
+
"grad_norm": 0.3458273410797119,
|
| 9482 |
+
"learning_rate": 5.025503045905933e-06,
|
| 9483 |
+
"loss": 1.4508,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.45610034207525657,
|
| 9488 |
+
"grad_norm": 0.45918381214141846,
|
| 9489 |
+
"learning_rate": 4.959714454271369e-06,
|
| 9490 |
+
"loss": 1.1859,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.45643819418049747,
|
| 9495 |
+
"grad_norm": 0.21645964682102203,
|
| 9496 |
+
"learning_rate": 4.8943483704846475e-06,
|
| 9497 |
+
"loss": 1.434,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.4567760462857384,
|
| 9502 |
+
"grad_norm": 0.21095064282417297,
|
| 9503 |
+
"learning_rate": 4.829405085134997e-06,
|
| 9504 |
+
"loss": 1.6274,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.4571138983909793,
|
| 9509 |
+
"grad_norm": 0.20745161175727844,
|
| 9510 |
+
"learning_rate": 4.764884886932086e-06,
|
| 9511 |
+
"loss": 1.5284,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.4574517504962203,
|
| 9516 |
+
"grad_norm": 0.2087661772966385,
|
| 9517 |
+
"learning_rate": 4.700788062704687e-06,
|
| 9518 |
+
"loss": 1.5616,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.45778960260146123,
|
| 9523 |
+
"grad_norm": 0.20495671033859253,
|
| 9524 |
+
"learning_rate": 4.6371148973994525e-06,
|
| 9525 |
+
"loss": 1.3866,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.45812745470670213,
|
| 9530 |
+
"grad_norm": 0.2103399932384491,
|
| 9531 |
+
"learning_rate": 4.573865674079625e-06,
|
| 9532 |
+
"loss": 1.4875,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.4584653068119431,
|
| 9537 |
+
"grad_norm": 0.20977221429347992,
|
| 9538 |
+
"learning_rate": 4.511040673923828e-06,
|
| 9539 |
+
"loss": 1.6612,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.458803158917184,
|
| 9544 |
+
"grad_norm": 0.21798568964004517,
|
| 9545 |
+
"learning_rate": 4.448640176224694e-06,
|
| 9546 |
+
"loss": 1.4265,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.45914101102242494,
|
| 9551 |
+
"grad_norm": 0.211429625749588,
|
| 9552 |
+
"learning_rate": 4.386664458387779e-06,
|
| 9553 |
+
"loss": 1.4075,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.45947886312766584,
|
| 9558 |
+
"grad_norm": 0.21625758707523346,
|
| 9559 |
+
"learning_rate": 4.325113795930203e-06,
|
| 9560 |
+
"loss": 1.4486,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.4598167152329068,
|
| 9565 |
+
"grad_norm": 0.2152903974056244,
|
| 9566 |
+
"learning_rate": 4.263988462479484e-06,
|
| 9567 |
+
"loss": 1.3261,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.46015456733814775,
|
| 9572 |
+
"grad_norm": 0.22299934923648834,
|
| 9573 |
+
"learning_rate": 4.203288729772326e-06,
|
| 9574 |
+
"loss": 1.4135,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.46049241944338865,
|
| 9579 |
+
"grad_norm": 0.2258044183254242,
|
| 9580 |
+
"learning_rate": 4.143014867653383e-06,
|
| 9581 |
+
"loss": 1.4606,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.4608302715486296,
|
| 9586 |
+
"grad_norm": 0.2317265421152115,
|
| 9587 |
+
"learning_rate": 4.083167144074073e-06,
|
| 9588 |
+
"loss": 1.4427,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.4611681236538705,
|
| 9593 |
+
"grad_norm": 0.23432107269763947,
|
| 9594 |
+
"learning_rate": 4.023745825091407e-06,
|
| 9595 |
+
"loss": 1.5321,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.46150597575911145,
|
| 9600 |
+
"grad_norm": 0.22782748937606812,
|
| 9601 |
+
"learning_rate": 3.964751174866765e-06,
|
| 9602 |
+
"loss": 1.4526,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.46184382786435235,
|
| 9607 |
+
"grad_norm": 0.22840560972690582,
|
| 9608 |
+
"learning_rate": 3.906183455664725e-06,
|
| 9609 |
+
"loss": 1.3746,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.4621816799695933,
|
| 9614 |
+
"grad_norm": 0.2276437133550644,
|
| 9615 |
+
"learning_rate": 3.84804292785198e-06,
|
| 9616 |
+
"loss": 1.3901,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.46251953207483426,
|
| 9621 |
+
"grad_norm": 0.2286890745162964,
|
| 9622 |
+
"learning_rate": 3.7903298498960572e-06,
|
| 9623 |
+
"loss": 1.403,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.46285738418007516,
|
| 9628 |
+
"grad_norm": 0.24942342936992645,
|
| 9629 |
+
"learning_rate": 3.7330444783642338e-06,
|
| 9630 |
+
"loss": 1.5007,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.4631952362853161,
|
| 9635 |
+
"grad_norm": 0.24023260176181793,
|
| 9636 |
+
"learning_rate": 3.676187067922421e-06,
|
| 9637 |
+
"loss": 1.4213,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.463533088390557,
|
| 9642 |
+
"grad_norm": 0.2482660710811615,
|
| 9643 |
+
"learning_rate": 3.619757871333973e-06,
|
| 9644 |
+
"loss": 1.4837,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.46387094049579797,
|
| 9649 |
+
"grad_norm": 0.244531512260437,
|
| 9650 |
+
"learning_rate": 3.563757139458579e-06,
|
| 9651 |
+
"loss": 1.4329,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.4642087926010389,
|
| 9656 |
+
"grad_norm": 0.24769660830497742,
|
| 9657 |
+
"learning_rate": 3.5081851212512175e-06,
|
| 9658 |
+
"loss": 1.4859,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.4645466447062798,
|
| 9663 |
+
"grad_norm": 0.2468714863061905,
|
| 9664 |
+
"learning_rate": 3.4530420637609363e-06,
|
| 9665 |
+
"loss": 1.451,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.4648844968115208,
|
| 9670 |
+
"grad_norm": 0.2557370960712433,
|
| 9671 |
+
"learning_rate": 3.3983282121298086e-06,
|
| 9672 |
+
"loss": 1.4479,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.4652223489167617,
|
| 9677 |
+
"grad_norm": 0.24851356446743011,
|
| 9678 |
+
"learning_rate": 3.3440438095919126e-06,
|
| 9679 |
+
"loss": 1.4514,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.46556020102200263,
|
| 9684 |
+
"grad_norm": 0.25202521681785583,
|
| 9685 |
+
"learning_rate": 3.290189097472096e-06,
|
| 9686 |
+
"loss": 1.4244,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.4658980531272435,
|
| 9691 |
+
"grad_norm": 0.25692325830459595,
|
| 9692 |
+
"learning_rate": 3.236764315185037e-06,
|
| 9693 |
+
"loss": 1.4422,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.4662359052324845,
|
| 9698 |
+
"grad_norm": 0.24908629059791565,
|
| 9699 |
+
"learning_rate": 3.1837697002341293e-06,
|
| 9700 |
+
"loss": 1.4451,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.46657375733772544,
|
| 9705 |
+
"grad_norm": 0.261599600315094,
|
| 9706 |
+
"learning_rate": 3.131205488210409e-06,
|
| 9707 |
+
"loss": 1.4316,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.46691160944296634,
|
| 9712 |
+
"grad_norm": 0.2605118751525879,
|
| 9713 |
+
"learning_rate": 3.0790719127915646e-06,
|
| 9714 |
+
"loss": 1.3966,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.4672494615482073,
|
| 9719 |
+
"grad_norm": 0.2556898891925812,
|
| 9720 |
+
"learning_rate": 3.0273692057408265e-06,
|
| 9721 |
+
"loss": 1.4522,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.4675873136534482,
|
| 9726 |
+
"grad_norm": 0.2627972960472107,
|
| 9727 |
+
"learning_rate": 2.976097596905969e-06,
|
| 9728 |
+
"loss": 1.4938,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.46792516575868914,
|
| 9733 |
+
"grad_norm": 1.4937089681625366,
|
| 9734 |
+
"learning_rate": 2.9252573142183326e-06,
|
| 9735 |
+
"loss": 1.4736,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.46826301786393004,
|
| 9740 |
+
"grad_norm": 0.26597166061401367,
|
| 9741 |
+
"learning_rate": 2.874848583691714e-06,
|
| 9742 |
+
"loss": 1.4108,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.468600869969171,
|
| 9747 |
+
"grad_norm": 0.26211485266685486,
|
| 9748 |
+
"learning_rate": 2.8248716294214774e-06,
|
| 9749 |
+
"loss": 1.5116,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.46893872207441195,
|
| 9754 |
+
"grad_norm": 0.2784208357334137,
|
| 9755 |
+
"learning_rate": 2.7753266735834338e-06,
|
| 9756 |
+
"loss": 1.5245,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.46927657417965285,
|
| 9761 |
+
"grad_norm": 0.26530563831329346,
|
| 9762 |
+
"learning_rate": 2.7262139364329643e-06,
|
| 9763 |
+
"loss": 1.4023,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.4696144262848938,
|
| 9768 |
+
"grad_norm": 0.2741868793964386,
|
| 9769 |
+
"learning_rate": 2.677533636303964e-06,
|
| 9770 |
+
"loss": 1.4072,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.4699522783901347,
|
| 9775 |
+
"grad_norm": 0.2684498429298401,
|
| 9776 |
+
"learning_rate": 2.6292859896079213e-06,
|
| 9777 |
+
"loss": 1.45,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.47029013049537566,
|
| 9782 |
+
"grad_norm": 0.2747170627117157,
|
| 9783 |
+
"learning_rate": 2.581471210832931e-06,
|
| 9784 |
+
"loss": 1.3942,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.47062798260061656,
|
| 9789 |
+
"grad_norm": 0.2784693241119385,
|
| 9790 |
+
"learning_rate": 2.5340895125427364e-06,
|
| 9791 |
+
"loss": 1.4305,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.4709658347058575,
|
| 9796 |
+
"grad_norm": 0.28920525312423706,
|
| 9797 |
+
"learning_rate": 2.4871411053757898e-06,
|
| 9798 |
+
"loss": 1.4891,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.47130368681109847,
|
| 9803 |
+
"grad_norm": 0.31437546014785767,
|
| 9804 |
+
"learning_rate": 2.440626198044327e-06,
|
| 9805 |
+
"loss": 1.4908,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.47164153891633936,
|
| 9810 |
+
"grad_norm": 0.287417471408844,
|
| 9811 |
+
"learning_rate": 2.394544997333437e-06,
|
| 9812 |
+
"loss": 1.4764,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.4719793910215803,
|
| 9817 |
+
"grad_norm": 0.28868481516838074,
|
| 9818 |
+
"learning_rate": 2.3488977081001394e-06,
|
| 9819 |
+
"loss": 1.4521,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.4723172431268212,
|
| 9824 |
+
"grad_norm": 0.317518025636673,
|
| 9825 |
+
"learning_rate": 2.3036845332724543e-06,
|
| 9826 |
+
"loss": 1.4094,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.4726550952320622,
|
| 9831 |
+
"grad_norm": 0.33674874901771545,
|
| 9832 |
+
"learning_rate": 2.2589056738485324e-06,
|
| 9833 |
+
"loss": 1.4015,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.47299294733730307,
|
| 9838 |
+
"grad_norm": 0.49115824699401855,
|
| 9839 |
+
"learning_rate": 2.2145613288957478e-06,
|
| 9840 |
+
"loss": 1.3206,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.473330799442544,
|
| 9845 |
+
"grad_norm": 0.20212937891483307,
|
| 9846 |
+
"learning_rate": 2.170651695549786e-06,
|
| 9847 |
+
"loss": 1.5266,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.473668651547785,
|
| 9852 |
+
"grad_norm": 0.21208123862743378,
|
| 9853 |
+
"learning_rate": 2.1271769690138332e-06,
|
| 9854 |
+
"loss": 1.5774,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.4740065036530259,
|
| 9859 |
+
"grad_norm": 0.21065491437911987,
|
| 9860 |
+
"learning_rate": 2.084137342557646e-06,
|
| 9861 |
+
"loss": 1.5756,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.47434435575826683,
|
| 9866 |
+
"grad_norm": 0.20114471018314362,
|
| 9867 |
+
"learning_rate": 2.0415330075166937e-06,
|
| 9868 |
+
"loss": 1.452,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.47468220786350773,
|
| 9873 |
+
"grad_norm": 0.20464257895946503,
|
| 9874 |
+
"learning_rate": 1.9993641532913833e-06,
|
| 9875 |
+
"loss": 1.4825,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.4750200599687487,
|
| 9880 |
+
"grad_norm": 0.21148669719696045,
|
| 9881 |
+
"learning_rate": 1.9576309673461357e-06,
|
| 9882 |
+
"loss": 1.425,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.4753579120739896,
|
| 9887 |
+
"grad_norm": 0.2091313600540161,
|
| 9888 |
+
"learning_rate": 1.916333635208556e-06,
|
| 9889 |
+
"loss": 1.5293,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.47569576417923054,
|
| 9894 |
+
"grad_norm": 0.21089115738868713,
|
| 9895 |
+
"learning_rate": 1.8754723404686425e-06,
|
| 9896 |
+
"loss": 1.5146,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.4760336162844715,
|
| 9901 |
+
"grad_norm": 0.22198493778705597,
|
| 9902 |
+
"learning_rate": 1.8350472647780116e-06,
|
| 9903 |
+
"loss": 1.5253,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.4763714683897124,
|
| 9908 |
+
"grad_norm": 0.22287015616893768,
|
| 9909 |
+
"learning_rate": 1.7950585878489856e-06,
|
| 9910 |
+
"loss": 1.491,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.47670932049495335,
|
| 9915 |
+
"grad_norm": 0.21632078289985657,
|
| 9916 |
+
"learning_rate": 1.7555064874538397e-06,
|
| 9917 |
+
"loss": 1.4399,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.47704717260019425,
|
| 9922 |
+
"grad_norm": 0.22057609260082245,
|
| 9923 |
+
"learning_rate": 1.7163911394240672e-06,
|
| 9924 |
+
"loss": 1.4489,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.4773850247054352,
|
| 9929 |
+
"grad_norm": 0.21960672736167908,
|
| 9930 |
+
"learning_rate": 1.6777127176495043e-06,
|
| 9931 |
+
"loss": 1.4367,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.4777228768106761,
|
| 9936 |
+
"grad_norm": 0.22372935712337494,
|
| 9937 |
+
"learning_rate": 1.6394713940776296e-06,
|
| 9938 |
+
"loss": 1.3677,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.47806072891591705,
|
| 9943 |
+
"grad_norm": 0.2302674651145935,
|
| 9944 |
+
"learning_rate": 1.6016673387127646e-06,
|
| 9945 |
+
"loss": 1.4437,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.478398581021158,
|
| 9950 |
+
"grad_norm": 0.22710569202899933,
|
| 9951 |
+
"learning_rate": 1.5643007196153302e-06,
|
| 9952 |
+
"loss": 1.4453,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.4787364331263989,
|
| 9957 |
+
"grad_norm": 0.2363537847995758,
|
| 9958 |
+
"learning_rate": 1.5273717029010925e-06,
|
| 9959 |
+
"loss": 1.4274,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.47907428523163986,
|
| 9964 |
+
"grad_norm": 0.23679710924625397,
|
| 9965 |
+
"learning_rate": 1.4908804527404286e-06,
|
| 9966 |
+
"loss": 1.5108,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.47941213733688076,
|
| 9971 |
+
"grad_norm": 0.24485650658607483,
|
| 9972 |
+
"learning_rate": 1.4548271313575835e-06,
|
| 9973 |
+
"loss": 1.4371,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.4797499894421217,
|
| 9978 |
+
"grad_norm": 0.23741504549980164,
|
| 9979 |
+
"learning_rate": 1.4192118990299707e-06,
|
| 9980 |
+
"loss": 1.4239,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.4800878415473626,
|
| 9985 |
+
"grad_norm": 0.24628090858459473,
|
| 9986 |
+
"learning_rate": 1.3840349140874619e-06,
|
| 9987 |
+
"loss": 1.521,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.48042569365260357,
|
| 9992 |
+
"grad_norm": 0.2446862906217575,
|
| 9993 |
+
"learning_rate": 1.3492963329116537e-06,
|
| 9994 |
+
"loss": 1.4534,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.4807635457578445,
|
| 9999 |
+
"grad_norm": 0.25101473927497864,
|
| 10000 |
+
"learning_rate": 1.3149963099352014e-06,
|
| 10001 |
+
"loss": 1.4976,
|
| 10002 |
+
"step": 1423
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 0.4811013978630854,
|
| 10006 |
+
"grad_norm": 0.24808958172798157,
|
| 10007 |
+
"learning_rate": 1.2811349976411202e-06,
|
| 10008 |
+
"loss": 1.4483,
|
| 10009 |
+
"step": 1424
|
| 10010 |
+
},
|
| 10011 |
+
{
|
| 10012 |
+
"epoch": 0.4814392499683264,
|
| 10013 |
+
"grad_norm": 0.25407129526138306,
|
| 10014 |
+
"learning_rate": 1.2477125465620853e-06,
|
| 10015 |
+
"loss": 1.5028,
|
| 10016 |
+
"step": 1425
|
| 10017 |
+
},
|
| 10018 |
+
{
|
| 10019 |
+
"epoch": 0.4817771020735673,
|
| 10020 |
+
"grad_norm": 0.2501218914985657,
|
| 10021 |
+
"learning_rate": 1.2147291052798216e-06,
|
| 10022 |
+
"loss": 1.4429,
|
| 10023 |
+
"step": 1426
|
| 10024 |
+
},
|
| 10025 |
+
{
|
| 10026 |
+
"epoch": 0.48211495417880823,
|
| 10027 |
+
"grad_norm": 0.2511563003063202,
|
| 10028 |
+
"learning_rate": 1.1821848204243814e-06,
|
| 10029 |
+
"loss": 1.4083,
|
| 10030 |
+
"step": 1427
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 0.4824528062840492,
|
| 10034 |
+
"grad_norm": 0.25538429617881775,
|
| 10035 |
+
"learning_rate": 1.1500798366735233e-06,
|
| 10036 |
+
"loss": 1.4678,
|
| 10037 |
+
"step": 1428
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 0.4827906583892901,
|
| 10041 |
+
"grad_norm": 0.2472202330827713,
|
| 10042 |
+
"learning_rate": 1.1184142967520794e-06,
|
| 10043 |
+
"loss": 1.3821,
|
| 10044 |
+
"step": 1429
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 0.48312851049453104,
|
| 10048 |
+
"grad_norm": 0.25652962923049927,
|
| 10049 |
+
"learning_rate": 1.0871883414312777e-06,
|
| 10050 |
+
"loss": 1.3967,
|
| 10051 |
+
"step": 1430
|
| 10052 |
+
},
|
| 10053 |
+
{
|
| 10054 |
+
"epoch": 0.48346636259977194,
|
| 10055 |
+
"grad_norm": 0.2588147521018982,
|
| 10056 |
+
"learning_rate": 1.0564021095281652e-06,
|
| 10057 |
+
"loss": 1.4388,
|
| 10058 |
+
"step": 1431
|
| 10059 |
+
},
|
| 10060 |
+
{
|
| 10061 |
+
"epoch": 0.4838042147050129,
|
| 10062 |
+
"grad_norm": 0.27002790570259094,
|
| 10063 |
+
"learning_rate": 1.0260557379049519e-06,
|
| 10064 |
+
"loss": 1.4767,
|
| 10065 |
+
"step": 1432
|
| 10066 |
+
},
|
| 10067 |
+
{
|
| 10068 |
+
"epoch": 0.4841420668102538,
|
| 10069 |
+
"grad_norm": 0.2688363194465637,
|
| 10070 |
+
"learning_rate": 9.96149361468457e-07,
|
| 10071 |
+
"loss": 1.4784,
|
| 10072 |
+
"step": 1433
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 0.48447991891549474,
|
| 10076 |
+
"grad_norm": 0.25430741906166077,
|
| 10077 |
+
"learning_rate": 9.66683113169431e-07,
|
| 10078 |
+
"loss": 1.4098,
|
| 10079 |
+
"step": 1434
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 0.4848177710207357,
|
| 10083 |
+
"grad_norm": 0.2543031573295593,
|
| 10084 |
+
"learning_rate": 9.376571240020227e-07,
|
| 10085 |
+
"loss": 1.4473,
|
| 10086 |
+
"step": 1435
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 0.4851556231259766,
|
| 10090 |
+
"grad_norm": 0.2679702937602997,
|
| 10091 |
+
"learning_rate": 9.090715230031688e-07,
|
| 10092 |
+
"loss": 1.465,
|
| 10093 |
+
"step": 1436
|
| 10094 |
+
},
|
| 10095 |
+
{
|
| 10096 |
+
"epoch": 0.48549347523121755,
|
| 10097 |
+
"grad_norm": 0.27029067277908325,
|
| 10098 |
+
"learning_rate": 8.809264372520609e-07,
|
| 10099 |
+
"loss": 1.4441,
|
| 10100 |
+
"step": 1437
|
| 10101 |
+
},
|
| 10102 |
+
{
|
| 10103 |
+
"epoch": 0.48583132733645845,
|
| 10104 |
+
"grad_norm": 0.27323269844055176,
|
| 10105 |
+
"learning_rate": 8.532219918695128e-07,
|
| 10106 |
+
"loss": 1.3906,
|
| 10107 |
+
"step": 1438
|
| 10108 |
+
},
|
| 10109 |
+
{
|
| 10110 |
+
"epoch": 0.4861691794416994,
|
| 10111 |
+
"grad_norm": 0.2603475749492645,
|
| 10112 |
+
"learning_rate": 8.259583100174606e-07,
|
| 10113 |
+
"loss": 1.3868,
|
| 10114 |
+
"step": 1439
|
| 10115 |
+
},
|
| 10116 |
+
{
|
| 10117 |
+
"epoch": 0.4865070315469403,
|
| 10118 |
+
"grad_norm": 0.2677699327468872,
|
| 10119 |
+
"learning_rate": 7.991355128984079e-07,
|
| 10120 |
+
"loss": 1.4834,
|
| 10121 |
+
"step": 1440
|
| 10122 |
+
},
|
| 10123 |
+
{
|
| 10124 |
+
"epoch": 0.48684488365218126,
|
| 10125 |
+
"grad_norm": 0.27419692277908325,
|
| 10126 |
+
"learning_rate": 7.727537197548707e-07,
|
| 10127 |
+
"loss": 1.5031,
|
| 10128 |
+
"step": 1441
|
| 10129 |
+
},
|
| 10130 |
+
{
|
| 10131 |
+
"epoch": 0.4871827357574222,
|
| 10132 |
+
"grad_norm": 0.28670912981033325,
|
| 10133 |
+
"learning_rate": 7.468130478688218e-07,
|
| 10134 |
+
"loss": 1.4728,
|
| 10135 |
+
"step": 1442
|
| 10136 |
+
},
|
| 10137 |
+
{
|
| 10138 |
+
"epoch": 0.4875205878626631,
|
| 10139 |
+
"grad_norm": 0.2821759879589081,
|
| 10140 |
+
"learning_rate": 7.213136125612586e-07,
|
| 10141 |
+
"loss": 1.471,
|
| 10142 |
+
"step": 1443
|
| 10143 |
+
},
|
| 10144 |
+
{
|
| 10145 |
+
"epoch": 0.48785843996790407,
|
| 10146 |
+
"grad_norm": 0.2898010015487671,
|
| 10147 |
+
"learning_rate": 6.962555271915805e-07,
|
| 10148 |
+
"loss": 1.4316,
|
| 10149 |
+
"step": 1444
|
| 10150 |
+
},
|
| 10151 |
+
{
|
| 10152 |
+
"epoch": 0.48819629207314497,
|
| 10153 |
+
"grad_norm": 0.2888733148574829,
|
| 10154 |
+
"learning_rate": 6.716389031571568e-07,
|
| 10155 |
+
"loss": 1.4511,
|
| 10156 |
+
"step": 1445
|
| 10157 |
+
},
|
| 10158 |
+
{
|
| 10159 |
+
"epoch": 0.4885341441783859,
|
| 10160 |
+
"grad_norm": 0.30311623215675354,
|
| 10161 |
+
"learning_rate": 6.474638498928265e-07,
|
| 10162 |
+
"loss": 1.4233,
|
| 10163 |
+
"step": 1446
|
| 10164 |
+
},
|
| 10165 |
+
{
|
| 10166 |
+
"epoch": 0.4888719962836268,
|
| 10167 |
+
"grad_norm": 0.30089882016181946,
|
| 10168 |
+
"learning_rate": 6.237304748703543e-07,
|
| 10169 |
+
"loss": 1.4354,
|
| 10170 |
+
"step": 1447
|
| 10171 |
+
},
|
| 10172 |
+
{
|
| 10173 |
+
"epoch": 0.4892098483888678,
|
| 10174 |
+
"grad_norm": 0.3312307894229889,
|
| 10175 |
+
"learning_rate": 6.004388835980423e-07,
|
| 10176 |
+
"loss": 1.4539,
|
| 10177 |
+
"step": 1448
|
| 10178 |
+
},
|
| 10179 |
+
{
|
| 10180 |
+
"epoch": 0.48954770049410873,
|
| 10181 |
+
"grad_norm": 0.3693748712539673,
|
| 10182 |
+
"learning_rate": 5.77589179620186e-07,
|
| 10183 |
+
"loss": 1.4094,
|
| 10184 |
+
"step": 1449
|
| 10185 |
+
},
|
| 10186 |
+
{
|
| 10187 |
+
"epoch": 0.4898855525993496,
|
| 10188 |
+
"grad_norm": 0.5018326044082642,
|
| 10189 |
+
"learning_rate": 5.55181464516652e-07,
|
| 10190 |
+
"loss": 1.2821,
|
| 10191 |
+
"step": 1450
|
| 10192 |
+
},
|
| 10193 |
+
{
|
| 10194 |
+
"epoch": 0.4902234047045906,
|
| 10195 |
+
"grad_norm": 0.196216881275177,
|
| 10196 |
+
"learning_rate": 5.332158379024122e-07,
|
| 10197 |
+
"loss": 1.4552,
|
| 10198 |
+
"step": 1451
|
| 10199 |
+
},
|
| 10200 |
+
{
|
| 10201 |
+
"epoch": 0.4905612568098315,
|
| 10202 |
+
"grad_norm": 0.20321056246757507,
|
| 10203 |
+
"learning_rate": 5.116923974270993e-07,
|
| 10204 |
+
"loss": 1.5635,
|
| 10205 |
+
"step": 1452
|
| 10206 |
+
},
|
| 10207 |
+
{
|
| 10208 |
+
"epoch": 0.49089910891507244,
|
| 10209 |
+
"grad_norm": 0.20241235196590424,
|
| 10210 |
+
"learning_rate": 4.906112387745965e-07,
|
| 10211 |
+
"loss": 1.4912,
|
| 10212 |
+
"step": 1453
|
| 10213 |
+
},
|
| 10214 |
+
{
|
| 10215 |
+
"epoch": 0.49123696102031333,
|
| 10216 |
+
"grad_norm": 0.20252864062786102,
|
| 10217 |
+
"learning_rate": 4.6997245566257064e-07,
|
| 10218 |
+
"loss": 1.4624,
|
| 10219 |
+
"step": 1454
|
| 10220 |
+
},
|
| 10221 |
+
{
|
| 10222 |
+
"epoch": 0.4915748131255543,
|
| 10223 |
+
"grad_norm": 0.20856152474880219,
|
| 10224 |
+
"learning_rate": 4.497761398421063e-07,
|
| 10225 |
+
"loss": 1.4994,
|
| 10226 |
+
"step": 1455
|
| 10227 |
+
},
|
| 10228 |
+
{
|
| 10229 |
+
"epoch": 0.49191266523079524,
|
| 10230 |
+
"grad_norm": 0.21053263545036316,
|
| 10231 |
+
"learning_rate": 4.3002238109723927e-07,
|
| 10232 |
+
"loss": 1.5073,
|
| 10233 |
+
"step": 1456
|
| 10234 |
+
},
|
| 10235 |
+
{
|
| 10236 |
+
"epoch": 0.49225051733603614,
|
| 10237 |
+
"grad_norm": 0.20670384168624878,
|
| 10238 |
+
"learning_rate": 4.107112672446123e-07,
|
| 10239 |
+
"loss": 1.4997,
|
| 10240 |
+
"step": 1457
|
| 10241 |
+
},
|
| 10242 |
+
{
|
| 10243 |
+
"epoch": 0.4925883694412771,
|
| 10244 |
+
"grad_norm": 0.20704622566699982,
|
| 10245 |
+
"learning_rate": 3.9184288413306456e-07,
|
| 10246 |
+
"loss": 1.4,
|
| 10247 |
+
"step": 1458
|
| 10248 |
+
},
|
| 10249 |
+
{
|
| 10250 |
+
"epoch": 0.492926221546518,
|
| 10251 |
+
"grad_norm": 0.2098395824432373,
|
| 10252 |
+
"learning_rate": 3.734173156432208e-07,
|
| 10253 |
+
"loss": 1.5099,
|
| 10254 |
+
"step": 1459
|
| 10255 |
+
},
|
| 10256 |
+
{
|
| 10257 |
+
"epoch": 0.49326407365175895,
|
| 10258 |
+
"grad_norm": 0.21044385433197021,
|
| 10259 |
+
"learning_rate": 3.554346436871581e-07,
|
| 10260 |
+
"loss": 1.4575,
|
| 10261 |
+
"step": 1460
|
| 10262 |
+
},
|
| 10263 |
+
{
|
| 10264 |
+
"epoch": 0.49360192575699985,
|
| 10265 |
+
"grad_norm": 0.2238229513168335,
|
| 10266 |
+
"learning_rate": 3.3789494820803957e-07,
|
| 10267 |
+
"loss": 1.517,
|
| 10268 |
+
"step": 1461
|
| 10269 |
+
},
|
| 10270 |
+
{
|
| 10271 |
+
"epoch": 0.4939397778622408,
|
| 10272 |
+
"grad_norm": 0.21777567267417908,
|
| 10273 |
+
"learning_rate": 3.2079830717972606e-07,
|
| 10274 |
+
"loss": 1.4322,
|
| 10275 |
+
"step": 1462
|
| 10276 |
+
},
|
| 10277 |
+
{
|
| 10278 |
+
"epoch": 0.49427762996748176,
|
| 10279 |
+
"grad_norm": 0.2231573462486267,
|
| 10280 |
+
"learning_rate": 3.041447966064648e-07,
|
| 10281 |
+
"loss": 1.4736,
|
| 10282 |
+
"step": 1463
|
| 10283 |
+
},
|
| 10284 |
+
{
|
| 10285 |
+
"epoch": 0.49461548207272266,
|
| 10286 |
+
"grad_norm": 0.229090616106987,
|
| 10287 |
+
"learning_rate": 2.8793449052254563e-07,
|
| 10288 |
+
"loss": 1.5046,
|
| 10289 |
+
"step": 1464
|
| 10290 |
+
},
|
| 10291 |
+
{
|
| 10292 |
+
"epoch": 0.4949533341779636,
|
| 10293 |
+
"grad_norm": 0.2325662076473236,
|
| 10294 |
+
"learning_rate": 2.721674609919345e-07,
|
| 10295 |
+
"loss": 1.4862,
|
| 10296 |
+
"step": 1465
|
| 10297 |
+
},
|
| 10298 |
+
{
|
| 10299 |
+
"epoch": 0.4952911862832045,
|
| 10300 |
+
"grad_norm": 0.2252163141965866,
|
| 10301 |
+
"learning_rate": 2.568437781080069e-07,
|
| 10302 |
+
"loss": 1.3253,
|
| 10303 |
+
"step": 1466
|
| 10304 |
+
},
|
| 10305 |
+
{
|
| 10306 |
+
"epoch": 0.49562903838844546,
|
| 10307 |
+
"grad_norm": 0.23609252274036407,
|
| 10308 |
+
"learning_rate": 2.4196350999320384e-07,
|
| 10309 |
+
"loss": 1.4784,
|
| 10310 |
+
"step": 1467
|
| 10311 |
+
},
|
| 10312 |
+
{
|
| 10313 |
+
"epoch": 0.49596689049368636,
|
| 10314 |
+
"grad_norm": 0.25220271944999695,
|
| 10315 |
+
"learning_rate": 2.275267227987321e-07,
|
| 10316 |
+
"loss": 1.5808,
|
| 10317 |
+
"step": 1468
|
| 10318 |
+
},
|
| 10319 |
+
{
|
| 10320 |
+
"epoch": 0.4963047425989273,
|
| 10321 |
+
"grad_norm": 0.22764766216278076,
|
| 10322 |
+
"learning_rate": 2.135334807042866e-07,
|
| 10323 |
+
"loss": 1.4093,
|
| 10324 |
+
"step": 1469
|
| 10325 |
+
},
|
| 10326 |
+
{
|
| 10327 |
+
"epoch": 0.49664259470416827,
|
| 10328 |
+
"grad_norm": 0.23635098338127136,
|
| 10329 |
+
"learning_rate": 1.9998384591773944e-07,
|
| 10330 |
+
"loss": 1.4132,
|
| 10331 |
+
"step": 1470
|
| 10332 |
+
},
|
| 10333 |
+
{
|
| 10334 |
+
"epoch": 0.49698044680940917,
|
| 10335 |
+
"grad_norm": 0.23680339753627777,
|
| 10336 |
+
"learning_rate": 1.8687787867489592e-07,
|
| 10337 |
+
"loss": 1.4401,
|
| 10338 |
+
"step": 1471
|
| 10339 |
+
},
|
| 10340 |
+
{
|
| 10341 |
+
"epoch": 0.4973182989146501,
|
| 10342 |
+
"grad_norm": 0.24005019664764404,
|
| 10343 |
+
"learning_rate": 1.7421563723919454e-07,
|
| 10344 |
+
"loss": 1.3511,
|
| 10345 |
+
"step": 1472
|
| 10346 |
+
},
|
| 10347 |
+
{
|
| 10348 |
+
"epoch": 0.497656151019891,
|
| 10349 |
+
"grad_norm": 0.23929578065872192,
|
| 10350 |
+
"learning_rate": 1.6199717790145174e-07,
|
| 10351 |
+
"loss": 1.489,
|
| 10352 |
+
"step": 1473
|
| 10353 |
+
},
|
| 10354 |
+
{
|
| 10355 |
+
"epoch": 0.497994003125132,
|
| 10356 |
+
"grad_norm": 0.2384917438030243,
|
| 10357 |
+
"learning_rate": 1.5022255497962879e-07,
|
| 10358 |
+
"loss": 1.3928,
|
| 10359 |
+
"step": 1474
|
| 10360 |
+
},
|
| 10361 |
+
{
|
| 10362 |
+
"epoch": 0.4983318552303729,
|
| 10363 |
+
"grad_norm": 0.2462180256843567,
|
| 10364 |
+
"learning_rate": 1.3889182081860962e-07,
|
| 10365 |
+
"loss": 1.3983,
|
| 10366 |
+
"step": 1475
|
| 10367 |
+
},
|
| 10368 |
+
{
|
| 10369 |
+
"epoch": 0.49866970733561383,
|
| 10370 |
+
"grad_norm": 0.25475817918777466,
|
| 10371 |
+
"learning_rate": 1.2800502578991235e-07,
|
| 10372 |
+
"loss": 1.5074,
|
| 10373 |
+
"step": 1476
|
| 10374 |
+
},
|
| 10375 |
+
{
|
| 10376 |
+
"epoch": 0.4990075594408548,
|
| 10377 |
+
"grad_norm": 0.24641422927379608,
|
| 10378 |
+
"learning_rate": 1.1756221829148928e-07,
|
| 10379 |
+
"loss": 1.4259,
|
| 10380 |
+
"step": 1477
|
| 10381 |
+
},
|
| 10382 |
+
{
|
| 10383 |
+
"epoch": 0.4993454115460957,
|
| 10384 |
+
"grad_norm": 0.25399869680404663,
|
| 10385 |
+
"learning_rate": 1.0756344474753821e-07,
|
| 10386 |
+
"loss": 1.4728,
|
| 10387 |
+
"step": 1478
|
| 10388 |
+
},
|
| 10389 |
+
{
|
| 10390 |
+
"epoch": 0.49968326365133664,
|
| 10391 |
+
"grad_norm": 0.24991141259670258,
|
| 10392 |
+
"learning_rate": 9.800874960826933e-08,
|
| 10393 |
+
"loss": 1.445,
|
| 10394 |
+
"step": 1479
|
| 10395 |
+
},
|
| 10396 |
+
{
|
| 10397 |
+
"epoch": 0.5000211157565776,
|
| 10398 |
+
"grad_norm": 0.2517636716365814,
|
| 10399 |
+
"learning_rate": 8.889817534969425e-08,
|
| 10400 |
+
"loss": 1.4285,
|
| 10401 |
+
"step": 1480
|
| 10402 |
+
},
|
| 10403 |
+
{
|
| 10404 |
+
"epoch": 0.5003589678618184,
|
| 10405 |
+
"grad_norm": 0.2585083544254303,
|
| 10406 |
+
"learning_rate": 8.023176247348163e-08,
|
| 10407 |
+
"loss": 1.443,
|
| 10408 |
+
"step": 1481
|
| 10409 |
+
},
|
| 10410 |
+
{
|
| 10411 |
+
"epoch": 0.5006968199670594,
|
| 10412 |
+
"grad_norm": 0.258181095123291,
|
| 10413 |
+
"learning_rate": 7.200954950673522e-08,
|
| 10414 |
+
"loss": 1.3778,
|
| 10415 |
+
"step": 1482
|
| 10416 |
+
},
|
| 10417 |
+
{
|
| 10418 |
+
"epoch": 0.5010346720723003,
|
| 10419 |
+
"grad_norm": 0.2598218321800232,
|
| 10420 |
+
"learning_rate": 6.423157300184946e-08,
|
| 10421 |
+
"loss": 1.473,
|
| 10422 |
+
"step": 1483
|
| 10423 |
+
},
|
| 10424 |
+
{
|
| 10425 |
+
"epoch": 0.5013725241775413,
|
| 10426 |
+
"grad_norm": 0.26156502962112427,
|
| 10427 |
+
"learning_rate": 5.6897867536331864e-08,
|
| 10428 |
+
"loss": 1.4907,
|
| 10429 |
+
"step": 1484
|
| 10430 |
+
},
|
| 10431 |
+
{
|
| 10432 |
+
"epoch": 0.5017103762827823,
|
| 10433 |
+
"grad_norm": 0.2644466459751129,
|
| 10434 |
+
"learning_rate": 5.000846571264761e-08,
|
| 10435 |
+
"loss": 1.4444,
|
| 10436 |
+
"step": 1485
|
| 10437 |
+
},
|
| 10438 |
+
{
|
| 10439 |
+
"epoch": 0.5020482283880231,
|
| 10440 |
+
"grad_norm": 0.27398133277893066,
|
| 10441 |
+
"learning_rate": 4.35633981580974e-08,
|
| 10442 |
+
"loss": 1.4513,
|
| 10443 |
+
"step": 1486
|
| 10444 |
+
},
|
| 10445 |
+
{
|
| 10446 |
+
"epoch": 0.502386080493264,
|
| 10447 |
+
"grad_norm": 0.2620060443878174,
|
| 10448 |
+
"learning_rate": 3.756269352462871e-08,
|
| 10449 |
+
"loss": 1.4428,
|
| 10450 |
+
"step": 1487
|
| 10451 |
+
},
|
| 10452 |
+
{
|
| 10453 |
+
"epoch": 0.502723932598505,
|
| 10454 |
+
"grad_norm": 0.2607761025428772,
|
| 10455 |
+
"learning_rate": 3.20063784888025e-08,
|
| 10456 |
+
"loss": 1.4203,
|
| 10457 |
+
"step": 1488
|
| 10458 |
+
},
|
| 10459 |
+
{
|
| 10460 |
+
"epoch": 0.503061784703746,
|
| 10461 |
+
"grad_norm": 0.26292839646339417,
|
| 10462 |
+
"learning_rate": 2.6894477751548964e-08,
|
| 10463 |
+
"loss": 1.4011,
|
| 10464 |
+
"step": 1489
|
| 10465 |
+
},
|
| 10466 |
+
{
|
| 10467 |
+
"epoch": 0.5033996368089869,
|
| 10468 |
+
"grad_norm": 0.27221038937568665,
|
| 10469 |
+
"learning_rate": 2.222701403818972e-08,
|
| 10470 |
+
"loss": 1.4325,
|
| 10471 |
+
"step": 1490
|
| 10472 |
+
},
|
| 10473 |
+
{
|
| 10474 |
+
"epoch": 0.5037374889142278,
|
| 10475 |
+
"grad_norm": 0.2788538634777069,
|
| 10476 |
+
"learning_rate": 1.8004008098226887e-08,
|
| 10477 |
+
"loss": 1.4261,
|
| 10478 |
+
"step": 1491
|
| 10479 |
+
},
|
| 10480 |
+
{
|
| 10481 |
+
"epoch": 0.5040753410194687,
|
| 10482 |
+
"grad_norm": 0.27817338705062866,
|
| 10483 |
+
"learning_rate": 1.4225478705309769e-08,
|
| 10484 |
+
"loss": 1.4214,
|
| 10485 |
+
"step": 1492
|
| 10486 |
+
},
|
| 10487 |
+
{
|
| 10488 |
+
"epoch": 0.5044131931247097,
|
| 10489 |
+
"grad_norm": 0.27153632044792175,
|
| 10490 |
+
"learning_rate": 1.0891442657134932e-08,
|
| 10491 |
+
"loss": 1.4154,
|
| 10492 |
+
"step": 1493
|
| 10493 |
+
},
|
| 10494 |
+
{
|
| 10495 |
+
"epoch": 0.5047510452299506,
|
| 10496 |
+
"grad_norm": 0.28666210174560547,
|
| 10497 |
+
"learning_rate": 8.001914775401798e-09,
|
| 10498 |
+
"loss": 1.4787,
|
| 10499 |
+
"step": 1494
|
| 10500 |
+
},
|
| 10501 |
+
{
|
| 10502 |
+
"epoch": 0.5050888973351915,
|
| 10503 |
+
"grad_norm": 0.28817957639694214,
|
| 10504 |
+
"learning_rate": 5.5569079056794206e-09,
|
| 10505 |
+
"loss": 1.4737,
|
| 10506 |
+
"step": 1495
|
| 10507 |
+
},
|
| 10508 |
+
{
|
| 10509 |
+
"epoch": 0.5054267494404324,
|
| 10510 |
+
"grad_norm": 0.3024436831474304,
|
| 10511 |
+
"learning_rate": 3.5564329174064824e-09,
|
| 10512 |
+
"loss": 1.4834,
|
| 10513 |
+
"step": 1496
|
| 10514 |
+
},
|
| 10515 |
+
{
|
| 10516 |
+
"epoch": 0.5057646015456734,
|
| 10517 |
+
"grad_norm": 0.3066464066505432,
|
| 10518 |
+
"learning_rate": 2.0004987038246824e-09,
|
| 10519 |
+
"loss": 1.4313,
|
| 10520 |
+
"step": 1497
|
| 10521 |
+
},
|
| 10522 |
+
{
|
| 10523 |
+
"epoch": 0.5061024536509143,
|
| 10524 |
+
"grad_norm": 0.3132622539997101,
|
| 10525 |
+
"learning_rate": 8.891121819565306e-10,
|
| 10526 |
+
"loss": 1.4616,
|
| 10527 |
+
"step": 1498
|
| 10528 |
+
},
|
| 10529 |
+
{
|
| 10530 |
+
"epoch": 0.5064403057561553,
|
| 10531 |
+
"grad_norm": 0.3416168689727783,
|
| 10532 |
+
"learning_rate": 2.2227829252763344e-10,
|
| 10533 |
+
"loss": 1.4852,
|
| 10534 |
+
"step": 1499
|
| 10535 |
+
},
|
| 10536 |
+
{
|
| 10537 |
+
"epoch": 0.5067781578613961,
|
| 10538 |
+
"grad_norm": 0.44100168347358704,
|
| 10539 |
+
"learning_rate": 0.0,
|
| 10540 |
+
"loss": 1.3137,
|
| 10541 |
+
"step": 1500
|
| 10542 |
+
},
|
| 10543 |
+
{
|
| 10544 |
+
"epoch": 0.5067781578613961,
|
| 10545 |
+
"eval_loss": 1.447874903678894,
|
| 10546 |
+
"eval_runtime": 268.936,
|
| 10547 |
+
"eval_samples_per_second": 18.536,
|
| 10548 |
+
"eval_steps_per_second": 9.27,
|
| 10549 |
+
"step": 1500
|
| 10550 |
}
|
| 10551 |
],
|
| 10552 |
"logging_steps": 1,
|
|
|
|
| 10561 |
"should_evaluate": false,
|
| 10562 |
"should_log": false,
|
| 10563 |
"should_save": true,
|
| 10564 |
+
"should_training_stop": true
|
| 10565 |
},
|
| 10566 |
"attributes": {}
|
| 10567 |
}
|
| 10568 |
},
|
| 10569 |
+
"total_flos": 5.735797142543401e+17,
|
| 10570 |
"train_batch_size": 2,
|
| 10571 |
"trial_name": null,
|
| 10572 |
"trial_params": null
|