Training in progress, step 1194, 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 27024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:120382d75a76dc9aae3ecacf2b16e667269b012ea8dc72d3830a79a45451be4c
|
| 3 |
size 27024
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 64038
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fcbd73225251960c958ae2469a30aedc9dd24fb6c97a26c4cf272dd03d0488fc
|
| 3 |
size 64038
|
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 sha256:8d95029bb41dde4a999ba2c809499dfbab9677cd9a713289e762e38ab1141bf4
|
| 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:b824446a00432ccc1e96d92ab06963b8ded78c8458a01706f53dabd26621c714
|
| 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": 398,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -5603,6 +5603,2800 @@
|
|
| 5603 |
"eval_samples_per_second": 355.912,
|
| 5604 |
"eval_steps_per_second": 178.025,
|
| 5605 |
"step": 796
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 5606 |
}
|
| 5607 |
],
|
| 5608 |
"logging_steps": 1,
|
|
@@ -5622,7 +8416,7 @@
|
|
| 5622 |
"attributes": {}
|
| 5623 |
}
|
| 5624 |
},
|
| 5625 |
-
"total_flos":
|
| 5626 |
"train_batch_size": 2,
|
| 5627 |
"trial_name": null,
|
| 5628 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.19524159921510914,
|
| 5 |
"eval_steps": 398,
|
| 6 |
+
"global_step": 1194,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 5603 |
"eval_samples_per_second": 355.912,
|
| 5604 |
"eval_steps_per_second": 178.025,
|
| 5605 |
"step": 796
|
| 5606 |
+
},
|
| 5607 |
+
{
|
| 5608 |
+
"epoch": 0.13032458507072195,
|
| 5609 |
+
"grad_norm": 0.08599609136581421,
|
| 5610 |
+
"learning_rate": 0.00010059650139677765,
|
| 5611 |
+
"loss": 10.34,
|
| 5612 |
+
"step": 797
|
| 5613 |
+
},
|
| 5614 |
+
{
|
| 5615 |
+
"epoch": 0.13048810399803779,
|
| 5616 |
+
"grad_norm": 0.060881245881319046,
|
| 5617 |
+
"learning_rate": 0.00010039766890801677,
|
| 5618 |
+
"loss": 10.3459,
|
| 5619 |
+
"step": 798
|
| 5620 |
+
},
|
| 5621 |
+
{
|
| 5622 |
+
"epoch": 0.13065162292535362,
|
| 5623 |
+
"grad_norm": 0.06616215407848358,
|
| 5624 |
+
"learning_rate": 0.00010019883484705851,
|
| 5625 |
+
"loss": 10.3435,
|
| 5626 |
+
"step": 799
|
| 5627 |
+
},
|
| 5628 |
+
{
|
| 5629 |
+
"epoch": 0.13081514185266946,
|
| 5630 |
+
"grad_norm": 0.07355780899524689,
|
| 5631 |
+
"learning_rate": 0.0001,
|
| 5632 |
+
"loss": 10.3455,
|
| 5633 |
+
"step": 800
|
| 5634 |
+
},
|
| 5635 |
+
{
|
| 5636 |
+
"epoch": 0.1309786607799853,
|
| 5637 |
+
"grad_norm": 0.09209233522415161,
|
| 5638 |
+
"learning_rate": 9.980116515294151e-05,
|
| 5639 |
+
"loss": 10.346,
|
| 5640 |
+
"step": 801
|
| 5641 |
+
},
|
| 5642 |
+
{
|
| 5643 |
+
"epoch": 0.13114217970730113,
|
| 5644 |
+
"grad_norm": 0.1081072986125946,
|
| 5645 |
+
"learning_rate": 9.960233109198327e-05,
|
| 5646 |
+
"loss": 10.3553,
|
| 5647 |
+
"step": 802
|
| 5648 |
+
},
|
| 5649 |
+
{
|
| 5650 |
+
"epoch": 0.13130569863461697,
|
| 5651 |
+
"grad_norm": 0.06916413456201553,
|
| 5652 |
+
"learning_rate": 9.940349860322237e-05,
|
| 5653 |
+
"loss": 10.3381,
|
| 5654 |
+
"step": 803
|
| 5655 |
+
},
|
| 5656 |
+
{
|
| 5657 |
+
"epoch": 0.1314692175619328,
|
| 5658 |
+
"grad_norm": 0.047468386590480804,
|
| 5659 |
+
"learning_rate": 9.920466847274979e-05,
|
| 5660 |
+
"loss": 10.3494,
|
| 5661 |
+
"step": 804
|
| 5662 |
+
},
|
| 5663 |
+
{
|
| 5664 |
+
"epoch": 0.13163273648924864,
|
| 5665 |
+
"grad_norm": 0.11021769791841507,
|
| 5666 |
+
"learning_rate": 9.900584148664704e-05,
|
| 5667 |
+
"loss": 10.3441,
|
| 5668 |
+
"step": 805
|
| 5669 |
+
},
|
| 5670 |
+
{
|
| 5671 |
+
"epoch": 0.13179625541656448,
|
| 5672 |
+
"grad_norm": 0.07369951158761978,
|
| 5673 |
+
"learning_rate": 9.880701843098335e-05,
|
| 5674 |
+
"loss": 10.3417,
|
| 5675 |
+
"step": 806
|
| 5676 |
+
},
|
| 5677 |
+
{
|
| 5678 |
+
"epoch": 0.13195977434388032,
|
| 5679 |
+
"grad_norm": 0.06742331385612488,
|
| 5680 |
+
"learning_rate": 9.860820009181232e-05,
|
| 5681 |
+
"loss": 10.3512,
|
| 5682 |
+
"step": 807
|
| 5683 |
+
},
|
| 5684 |
+
{
|
| 5685 |
+
"epoch": 0.13212329327119615,
|
| 5686 |
+
"grad_norm": 0.08016981184482574,
|
| 5687 |
+
"learning_rate": 9.840938725516889e-05,
|
| 5688 |
+
"loss": 10.3463,
|
| 5689 |
+
"step": 808
|
| 5690 |
+
},
|
| 5691 |
+
{
|
| 5692 |
+
"epoch": 0.132286812198512,
|
| 5693 |
+
"grad_norm": 0.07211664319038391,
|
| 5694 |
+
"learning_rate": 9.821058070706632e-05,
|
| 5695 |
+
"loss": 10.3449,
|
| 5696 |
+
"step": 809
|
| 5697 |
+
},
|
| 5698 |
+
{
|
| 5699 |
+
"epoch": 0.13245033112582782,
|
| 5700 |
+
"grad_norm": 0.07631465047597885,
|
| 5701 |
+
"learning_rate": 9.801178123349298e-05,
|
| 5702 |
+
"loss": 10.347,
|
| 5703 |
+
"step": 810
|
| 5704 |
+
},
|
| 5705 |
+
{
|
| 5706 |
+
"epoch": 0.13261385005314366,
|
| 5707 |
+
"grad_norm": 0.0600840263068676,
|
| 5708 |
+
"learning_rate": 9.781298962040922e-05,
|
| 5709 |
+
"loss": 10.3396,
|
| 5710 |
+
"step": 811
|
| 5711 |
+
},
|
| 5712 |
+
{
|
| 5713 |
+
"epoch": 0.1327773689804595,
|
| 5714 |
+
"grad_norm": 0.1150350272655487,
|
| 5715 |
+
"learning_rate": 9.761420665374437e-05,
|
| 5716 |
+
"loss": 10.3351,
|
| 5717 |
+
"step": 812
|
| 5718 |
+
},
|
| 5719 |
+
{
|
| 5720 |
+
"epoch": 0.13294088790777533,
|
| 5721 |
+
"grad_norm": 0.0911078006029129,
|
| 5722 |
+
"learning_rate": 9.741543311939356e-05,
|
| 5723 |
+
"loss": 10.3427,
|
| 5724 |
+
"step": 813
|
| 5725 |
+
},
|
| 5726 |
+
{
|
| 5727 |
+
"epoch": 0.13310440683509117,
|
| 5728 |
+
"grad_norm": 0.08850537985563278,
|
| 5729 |
+
"learning_rate": 9.721666980321459e-05,
|
| 5730 |
+
"loss": 10.348,
|
| 5731 |
+
"step": 814
|
| 5732 |
+
},
|
| 5733 |
+
{
|
| 5734 |
+
"epoch": 0.133267925762407,
|
| 5735 |
+
"grad_norm": 0.07637340575456619,
|
| 5736 |
+
"learning_rate": 9.701791749102495e-05,
|
| 5737 |
+
"loss": 10.3475,
|
| 5738 |
+
"step": 815
|
| 5739 |
+
},
|
| 5740 |
+
{
|
| 5741 |
+
"epoch": 0.13343144468972284,
|
| 5742 |
+
"grad_norm": 0.0791415199637413,
|
| 5743 |
+
"learning_rate": 9.681917696859854e-05,
|
| 5744 |
+
"loss": 10.3489,
|
| 5745 |
+
"step": 816
|
| 5746 |
+
},
|
| 5747 |
+
{
|
| 5748 |
+
"epoch": 0.13359496361703868,
|
| 5749 |
+
"grad_norm": 0.10085449367761612,
|
| 5750 |
+
"learning_rate": 9.662044902166275e-05,
|
| 5751 |
+
"loss": 10.3404,
|
| 5752 |
+
"step": 817
|
| 5753 |
+
},
|
| 5754 |
+
{
|
| 5755 |
+
"epoch": 0.13375848254435452,
|
| 5756 |
+
"grad_norm": 0.06288623064756393,
|
| 5757 |
+
"learning_rate": 9.642173443589507e-05,
|
| 5758 |
+
"loss": 10.3518,
|
| 5759 |
+
"step": 818
|
| 5760 |
+
},
|
| 5761 |
+
{
|
| 5762 |
+
"epoch": 0.13392200147167035,
|
| 5763 |
+
"grad_norm": 0.06273672729730606,
|
| 5764 |
+
"learning_rate": 9.622303399692038e-05,
|
| 5765 |
+
"loss": 10.3437,
|
| 5766 |
+
"step": 819
|
| 5767 |
+
},
|
| 5768 |
+
{
|
| 5769 |
+
"epoch": 0.1340855203989862,
|
| 5770 |
+
"grad_norm": 0.07121649384498596,
|
| 5771 |
+
"learning_rate": 9.602434849030745e-05,
|
| 5772 |
+
"loss": 10.3537,
|
| 5773 |
+
"step": 820
|
| 5774 |
+
},
|
| 5775 |
+
{
|
| 5776 |
+
"epoch": 0.13424903932630203,
|
| 5777 |
+
"grad_norm": 0.0698346421122551,
|
| 5778 |
+
"learning_rate": 9.582567870156618e-05,
|
| 5779 |
+
"loss": 10.3466,
|
| 5780 |
+
"step": 821
|
| 5781 |
+
},
|
| 5782 |
+
{
|
| 5783 |
+
"epoch": 0.13441255825361786,
|
| 5784 |
+
"grad_norm": 0.05925373360514641,
|
| 5785 |
+
"learning_rate": 9.562702541614416e-05,
|
| 5786 |
+
"loss": 10.3394,
|
| 5787 |
+
"step": 822
|
| 5788 |
+
},
|
| 5789 |
+
{
|
| 5790 |
+
"epoch": 0.1345760771809337,
|
| 5791 |
+
"grad_norm": 0.057984985411167145,
|
| 5792 |
+
"learning_rate": 9.54283894194239e-05,
|
| 5793 |
+
"loss": 10.3468,
|
| 5794 |
+
"step": 823
|
| 5795 |
+
},
|
| 5796 |
+
{
|
| 5797 |
+
"epoch": 0.13473959610824954,
|
| 5798 |
+
"grad_norm": 0.0611608549952507,
|
| 5799 |
+
"learning_rate": 9.522977149671942e-05,
|
| 5800 |
+
"loss": 10.3511,
|
| 5801 |
+
"step": 824
|
| 5802 |
+
},
|
| 5803 |
+
{
|
| 5804 |
+
"epoch": 0.13490311503556537,
|
| 5805 |
+
"grad_norm": 0.09152734279632568,
|
| 5806 |
+
"learning_rate": 9.503117243327337e-05,
|
| 5807 |
+
"loss": 10.3458,
|
| 5808 |
+
"step": 825
|
| 5809 |
+
},
|
| 5810 |
+
{
|
| 5811 |
+
"epoch": 0.1350666339628812,
|
| 5812 |
+
"grad_norm": 0.07080190628767014,
|
| 5813 |
+
"learning_rate": 9.483259301425381e-05,
|
| 5814 |
+
"loss": 10.3435,
|
| 5815 |
+
"step": 826
|
| 5816 |
+
},
|
| 5817 |
+
{
|
| 5818 |
+
"epoch": 0.13523015289019705,
|
| 5819 |
+
"grad_norm": 0.067464180290699,
|
| 5820 |
+
"learning_rate": 9.463403402475108e-05,
|
| 5821 |
+
"loss": 10.3399,
|
| 5822 |
+
"step": 827
|
| 5823 |
+
},
|
| 5824 |
+
{
|
| 5825 |
+
"epoch": 0.13539367181751288,
|
| 5826 |
+
"grad_norm": 0.10405400395393372,
|
| 5827 |
+
"learning_rate": 9.443549624977487e-05,
|
| 5828 |
+
"loss": 10.3379,
|
| 5829 |
+
"step": 828
|
| 5830 |
+
},
|
| 5831 |
+
{
|
| 5832 |
+
"epoch": 0.13555719074482872,
|
| 5833 |
+
"grad_norm": 0.0796792209148407,
|
| 5834 |
+
"learning_rate": 9.423698047425095e-05,
|
| 5835 |
+
"loss": 10.3417,
|
| 5836 |
+
"step": 829
|
| 5837 |
+
},
|
| 5838 |
+
{
|
| 5839 |
+
"epoch": 0.13572070967214456,
|
| 5840 |
+
"grad_norm": 0.06306777894496918,
|
| 5841 |
+
"learning_rate": 9.403848748301802e-05,
|
| 5842 |
+
"loss": 10.3452,
|
| 5843 |
+
"step": 830
|
| 5844 |
+
},
|
| 5845 |
+
{
|
| 5846 |
+
"epoch": 0.1358842285994604,
|
| 5847 |
+
"grad_norm": 0.07225450873374939,
|
| 5848 |
+
"learning_rate": 9.384001806082483e-05,
|
| 5849 |
+
"loss": 10.3481,
|
| 5850 |
+
"step": 831
|
| 5851 |
+
},
|
| 5852 |
+
{
|
| 5853 |
+
"epoch": 0.13604774752677623,
|
| 5854 |
+
"grad_norm": 0.043913766741752625,
|
| 5855 |
+
"learning_rate": 9.364157299232692e-05,
|
| 5856 |
+
"loss": 10.3442,
|
| 5857 |
+
"step": 832
|
| 5858 |
+
},
|
| 5859 |
+
{
|
| 5860 |
+
"epoch": 0.13621126645409207,
|
| 5861 |
+
"grad_norm": 0.08979449421167374,
|
| 5862 |
+
"learning_rate": 9.344315306208348e-05,
|
| 5863 |
+
"loss": 10.3396,
|
| 5864 |
+
"step": 833
|
| 5865 |
+
},
|
| 5866 |
+
{
|
| 5867 |
+
"epoch": 0.1363747853814079,
|
| 5868 |
+
"grad_norm": 0.13488726317882538,
|
| 5869 |
+
"learning_rate": 9.324475905455439e-05,
|
| 5870 |
+
"loss": 10.3556,
|
| 5871 |
+
"step": 834
|
| 5872 |
+
},
|
| 5873 |
+
{
|
| 5874 |
+
"epoch": 0.13653830430872374,
|
| 5875 |
+
"grad_norm": 0.0586860291659832,
|
| 5876 |
+
"learning_rate": 9.304639175409698e-05,
|
| 5877 |
+
"loss": 10.3479,
|
| 5878 |
+
"step": 835
|
| 5879 |
+
},
|
| 5880 |
+
{
|
| 5881 |
+
"epoch": 0.13670182323603958,
|
| 5882 |
+
"grad_norm": 0.10774936527013779,
|
| 5883 |
+
"learning_rate": 9.284805194496305e-05,
|
| 5884 |
+
"loss": 10.3387,
|
| 5885 |
+
"step": 836
|
| 5886 |
+
},
|
| 5887 |
+
{
|
| 5888 |
+
"epoch": 0.1368653421633554,
|
| 5889 |
+
"grad_norm": 0.08792544156312943,
|
| 5890 |
+
"learning_rate": 9.264974041129572e-05,
|
| 5891 |
+
"loss": 10.3466,
|
| 5892 |
+
"step": 837
|
| 5893 |
+
},
|
| 5894 |
+
{
|
| 5895 |
+
"epoch": 0.13702886109067125,
|
| 5896 |
+
"grad_norm": 0.06584392488002777,
|
| 5897 |
+
"learning_rate": 9.245145793712623e-05,
|
| 5898 |
+
"loss": 10.3492,
|
| 5899 |
+
"step": 838
|
| 5900 |
+
},
|
| 5901 |
+
{
|
| 5902 |
+
"epoch": 0.1371923800179871,
|
| 5903 |
+
"grad_norm": 0.08722234517335892,
|
| 5904 |
+
"learning_rate": 9.225320530637104e-05,
|
| 5905 |
+
"loss": 10.3575,
|
| 5906 |
+
"step": 839
|
| 5907 |
+
},
|
| 5908 |
+
{
|
| 5909 |
+
"epoch": 0.13735589894530292,
|
| 5910 |
+
"grad_norm": 0.07480508834123611,
|
| 5911 |
+
"learning_rate": 9.205498330282856e-05,
|
| 5912 |
+
"loss": 10.3442,
|
| 5913 |
+
"step": 840
|
| 5914 |
+
},
|
| 5915 |
+
{
|
| 5916 |
+
"epoch": 0.13751941787261876,
|
| 5917 |
+
"grad_norm": 0.11431939899921417,
|
| 5918 |
+
"learning_rate": 9.185679271017613e-05,
|
| 5919 |
+
"loss": 10.3433,
|
| 5920 |
+
"step": 841
|
| 5921 |
+
},
|
| 5922 |
+
{
|
| 5923 |
+
"epoch": 0.1376829367999346,
|
| 5924 |
+
"grad_norm": 0.08573208004236221,
|
| 5925 |
+
"learning_rate": 9.165863431196693e-05,
|
| 5926 |
+
"loss": 10.3478,
|
| 5927 |
+
"step": 842
|
| 5928 |
+
},
|
| 5929 |
+
{
|
| 5930 |
+
"epoch": 0.13784645572725043,
|
| 5931 |
+
"grad_norm": 0.06913722306489944,
|
| 5932 |
+
"learning_rate": 9.146050889162679e-05,
|
| 5933 |
+
"loss": 10.3498,
|
| 5934 |
+
"step": 843
|
| 5935 |
+
},
|
| 5936 |
+
{
|
| 5937 |
+
"epoch": 0.13800997465456627,
|
| 5938 |
+
"grad_norm": 0.11297217756509781,
|
| 5939 |
+
"learning_rate": 9.126241723245122e-05,
|
| 5940 |
+
"loss": 10.3358,
|
| 5941 |
+
"step": 844
|
| 5942 |
+
},
|
| 5943 |
+
{
|
| 5944 |
+
"epoch": 0.1381734935818821,
|
| 5945 |
+
"grad_norm": 0.06213637441396713,
|
| 5946 |
+
"learning_rate": 9.106436011760229e-05,
|
| 5947 |
+
"loss": 10.3378,
|
| 5948 |
+
"step": 845
|
| 5949 |
+
},
|
| 5950 |
+
{
|
| 5951 |
+
"epoch": 0.13833701250919794,
|
| 5952 |
+
"grad_norm": 0.09207039326429367,
|
| 5953 |
+
"learning_rate": 9.086633833010537e-05,
|
| 5954 |
+
"loss": 10.3402,
|
| 5955 |
+
"step": 846
|
| 5956 |
+
},
|
| 5957 |
+
{
|
| 5958 |
+
"epoch": 0.13850053143651378,
|
| 5959 |
+
"grad_norm": 0.09865938127040863,
|
| 5960 |
+
"learning_rate": 9.066835265284632e-05,
|
| 5961 |
+
"loss": 10.3447,
|
| 5962 |
+
"step": 847
|
| 5963 |
+
},
|
| 5964 |
+
{
|
| 5965 |
+
"epoch": 0.13866405036382962,
|
| 5966 |
+
"grad_norm": 0.06385111808776855,
|
| 5967 |
+
"learning_rate": 9.047040386856807e-05,
|
| 5968 |
+
"loss": 10.3582,
|
| 5969 |
+
"step": 848
|
| 5970 |
+
},
|
| 5971 |
+
{
|
| 5972 |
+
"epoch": 0.13882756929114545,
|
| 5973 |
+
"grad_norm": 0.06984174996614456,
|
| 5974 |
+
"learning_rate": 9.027249275986783e-05,
|
| 5975 |
+
"loss": 10.3414,
|
| 5976 |
+
"step": 849
|
| 5977 |
+
},
|
| 5978 |
+
{
|
| 5979 |
+
"epoch": 0.1389910882184613,
|
| 5980 |
+
"grad_norm": 0.048934344202280045,
|
| 5981 |
+
"learning_rate": 9.007462010919386e-05,
|
| 5982 |
+
"loss": 10.3432,
|
| 5983 |
+
"step": 850
|
| 5984 |
+
},
|
| 5985 |
+
{
|
| 5986 |
+
"epoch": 0.13915460714577713,
|
| 5987 |
+
"grad_norm": 0.049415282905101776,
|
| 5988 |
+
"learning_rate": 8.987678669884224e-05,
|
| 5989 |
+
"loss": 10.3454,
|
| 5990 |
+
"step": 851
|
| 5991 |
+
},
|
| 5992 |
+
{
|
| 5993 |
+
"epoch": 0.13931812607309296,
|
| 5994 |
+
"grad_norm": 0.08953525871038437,
|
| 5995 |
+
"learning_rate": 8.9678993310954e-05,
|
| 5996 |
+
"loss": 10.3427,
|
| 5997 |
+
"step": 852
|
| 5998 |
+
},
|
| 5999 |
+
{
|
| 6000 |
+
"epoch": 0.1394816450004088,
|
| 6001 |
+
"grad_norm": 0.0782846063375473,
|
| 6002 |
+
"learning_rate": 8.948124072751195e-05,
|
| 6003 |
+
"loss": 10.3393,
|
| 6004 |
+
"step": 853
|
| 6005 |
+
},
|
| 6006 |
+
{
|
| 6007 |
+
"epoch": 0.13964516392772464,
|
| 6008 |
+
"grad_norm": 0.09727578610181808,
|
| 6009 |
+
"learning_rate": 8.928352973033759e-05,
|
| 6010 |
+
"loss": 10.35,
|
| 6011 |
+
"step": 854
|
| 6012 |
+
},
|
| 6013 |
+
{
|
| 6014 |
+
"epoch": 0.13980868285504047,
|
| 6015 |
+
"grad_norm": 0.05756600573658943,
|
| 6016 |
+
"learning_rate": 8.908586110108794e-05,
|
| 6017 |
+
"loss": 10.3507,
|
| 6018 |
+
"step": 855
|
| 6019 |
+
},
|
| 6020 |
+
{
|
| 6021 |
+
"epoch": 0.1399722017823563,
|
| 6022 |
+
"grad_norm": 0.0850805938243866,
|
| 6023 |
+
"learning_rate": 8.888823562125259e-05,
|
| 6024 |
+
"loss": 10.3456,
|
| 6025 |
+
"step": 856
|
| 6026 |
+
},
|
| 6027 |
+
{
|
| 6028 |
+
"epoch": 0.14013572070967215,
|
| 6029 |
+
"grad_norm": 0.052331432700157166,
|
| 6030 |
+
"learning_rate": 8.869065407215046e-05,
|
| 6031 |
+
"loss": 10.3478,
|
| 6032 |
+
"step": 857
|
| 6033 |
+
},
|
| 6034 |
+
{
|
| 6035 |
+
"epoch": 0.14029923963698798,
|
| 6036 |
+
"grad_norm": 0.09885098040103912,
|
| 6037 |
+
"learning_rate": 8.849311723492689e-05,
|
| 6038 |
+
"loss": 10.3437,
|
| 6039 |
+
"step": 858
|
| 6040 |
+
},
|
| 6041 |
+
{
|
| 6042 |
+
"epoch": 0.14046275856430382,
|
| 6043 |
+
"grad_norm": 0.0896768644452095,
|
| 6044 |
+
"learning_rate": 8.829562589055037e-05,
|
| 6045 |
+
"loss": 10.3512,
|
| 6046 |
+
"step": 859
|
| 6047 |
+
},
|
| 6048 |
+
{
|
| 6049 |
+
"epoch": 0.14062627749161966,
|
| 6050 |
+
"grad_norm": 0.07669004797935486,
|
| 6051 |
+
"learning_rate": 8.809818081980953e-05,
|
| 6052 |
+
"loss": 10.3417,
|
| 6053 |
+
"step": 860
|
| 6054 |
+
},
|
| 6055 |
+
{
|
| 6056 |
+
"epoch": 0.1407897964189355,
|
| 6057 |
+
"grad_norm": 0.06714767962694168,
|
| 6058 |
+
"learning_rate": 8.790078280331011e-05,
|
| 6059 |
+
"loss": 10.3475,
|
| 6060 |
+
"step": 861
|
| 6061 |
+
},
|
| 6062 |
+
{
|
| 6063 |
+
"epoch": 0.14095331534625133,
|
| 6064 |
+
"grad_norm": 0.059197574853897095,
|
| 6065 |
+
"learning_rate": 8.77034326214718e-05,
|
| 6066 |
+
"loss": 10.3503,
|
| 6067 |
+
"step": 862
|
| 6068 |
+
},
|
| 6069 |
+
{
|
| 6070 |
+
"epoch": 0.14111683427356717,
|
| 6071 |
+
"grad_norm": 0.07281272858381271,
|
| 6072 |
+
"learning_rate": 8.750613105452511e-05,
|
| 6073 |
+
"loss": 10.3434,
|
| 6074 |
+
"step": 863
|
| 6075 |
+
},
|
| 6076 |
+
{
|
| 6077 |
+
"epoch": 0.141280353200883,
|
| 6078 |
+
"grad_norm": 0.06779444217681885,
|
| 6079 |
+
"learning_rate": 8.730887888250849e-05,
|
| 6080 |
+
"loss": 10.3378,
|
| 6081 |
+
"step": 864
|
| 6082 |
+
},
|
| 6083 |
+
{
|
| 6084 |
+
"epoch": 0.14144387212819884,
|
| 6085 |
+
"grad_norm": 0.07162304222583771,
|
| 6086 |
+
"learning_rate": 8.711167688526493e-05,
|
| 6087 |
+
"loss": 10.3525,
|
| 6088 |
+
"step": 865
|
| 6089 |
+
},
|
| 6090 |
+
{
|
| 6091 |
+
"epoch": 0.14160739105551468,
|
| 6092 |
+
"grad_norm": 0.08897604048252106,
|
| 6093 |
+
"learning_rate": 8.69145258424392e-05,
|
| 6094 |
+
"loss": 10.343,
|
| 6095 |
+
"step": 866
|
| 6096 |
+
},
|
| 6097 |
+
{
|
| 6098 |
+
"epoch": 0.1417709099828305,
|
| 6099 |
+
"grad_norm": 0.09015105664730072,
|
| 6100 |
+
"learning_rate": 8.671742653347456e-05,
|
| 6101 |
+
"loss": 10.3454,
|
| 6102 |
+
"step": 867
|
| 6103 |
+
},
|
| 6104 |
+
{
|
| 6105 |
+
"epoch": 0.14193442891014635,
|
| 6106 |
+
"grad_norm": 0.06099049746990204,
|
| 6107 |
+
"learning_rate": 8.652037973760969e-05,
|
| 6108 |
+
"loss": 10.3512,
|
| 6109 |
+
"step": 868
|
| 6110 |
+
},
|
| 6111 |
+
{
|
| 6112 |
+
"epoch": 0.14209794783746219,
|
| 6113 |
+
"grad_norm": 0.0777360200881958,
|
| 6114 |
+
"learning_rate": 8.632338623387577e-05,
|
| 6115 |
+
"loss": 10.333,
|
| 6116 |
+
"step": 869
|
| 6117 |
+
},
|
| 6118 |
+
{
|
| 6119 |
+
"epoch": 0.14226146676477802,
|
| 6120 |
+
"grad_norm": 0.061629846692085266,
|
| 6121 |
+
"learning_rate": 8.612644680109319e-05,
|
| 6122 |
+
"loss": 10.3486,
|
| 6123 |
+
"step": 870
|
| 6124 |
+
},
|
| 6125 |
+
{
|
| 6126 |
+
"epoch": 0.14242498569209386,
|
| 6127 |
+
"grad_norm": 0.07192935049533844,
|
| 6128 |
+
"learning_rate": 8.59295622178686e-05,
|
| 6129 |
+
"loss": 10.3455,
|
| 6130 |
+
"step": 871
|
| 6131 |
+
},
|
| 6132 |
+
{
|
| 6133 |
+
"epoch": 0.1425885046194097,
|
| 6134 |
+
"grad_norm": 0.0606074221432209,
|
| 6135 |
+
"learning_rate": 8.573273326259184e-05,
|
| 6136 |
+
"loss": 10.3434,
|
| 6137 |
+
"step": 872
|
| 6138 |
+
},
|
| 6139 |
+
{
|
| 6140 |
+
"epoch": 0.14275202354672553,
|
| 6141 |
+
"grad_norm": 0.06332432478666306,
|
| 6142 |
+
"learning_rate": 8.553596071343274e-05,
|
| 6143 |
+
"loss": 10.342,
|
| 6144 |
+
"step": 873
|
| 6145 |
+
},
|
| 6146 |
+
{
|
| 6147 |
+
"epoch": 0.14291554247404137,
|
| 6148 |
+
"grad_norm": 0.06337670981884003,
|
| 6149 |
+
"learning_rate": 8.533924534833822e-05,
|
| 6150 |
+
"loss": 10.3499,
|
| 6151 |
+
"step": 874
|
| 6152 |
+
},
|
| 6153 |
+
{
|
| 6154 |
+
"epoch": 0.1430790614013572,
|
| 6155 |
+
"grad_norm": 0.06023447588086128,
|
| 6156 |
+
"learning_rate": 8.514258794502905e-05,
|
| 6157 |
+
"loss": 10.3513,
|
| 6158 |
+
"step": 875
|
| 6159 |
+
},
|
| 6160 |
+
{
|
| 6161 |
+
"epoch": 0.14324258032867304,
|
| 6162 |
+
"grad_norm": 0.1311032921075821,
|
| 6163 |
+
"learning_rate": 8.494598928099688e-05,
|
| 6164 |
+
"loss": 10.3421,
|
| 6165 |
+
"step": 876
|
| 6166 |
+
},
|
| 6167 |
+
{
|
| 6168 |
+
"epoch": 0.14340609925598888,
|
| 6169 |
+
"grad_norm": 0.07438237965106964,
|
| 6170 |
+
"learning_rate": 8.474945013350112e-05,
|
| 6171 |
+
"loss": 10.3464,
|
| 6172 |
+
"step": 877
|
| 6173 |
+
},
|
| 6174 |
+
{
|
| 6175 |
+
"epoch": 0.14356961818330471,
|
| 6176 |
+
"grad_norm": 0.09203974902629852,
|
| 6177 |
+
"learning_rate": 8.455297127956589e-05,
|
| 6178 |
+
"loss": 10.3408,
|
| 6179 |
+
"step": 878
|
| 6180 |
+
},
|
| 6181 |
+
{
|
| 6182 |
+
"epoch": 0.14373313711062055,
|
| 6183 |
+
"grad_norm": 0.07816965878009796,
|
| 6184 |
+
"learning_rate": 8.435655349597689e-05,
|
| 6185 |
+
"loss": 10.347,
|
| 6186 |
+
"step": 879
|
| 6187 |
+
},
|
| 6188 |
+
{
|
| 6189 |
+
"epoch": 0.1438966560379364,
|
| 6190 |
+
"grad_norm": 0.058189962059259415,
|
| 6191 |
+
"learning_rate": 8.416019755927851e-05,
|
| 6192 |
+
"loss": 10.3411,
|
| 6193 |
+
"step": 880
|
| 6194 |
+
},
|
| 6195 |
+
{
|
| 6196 |
+
"epoch": 0.14406017496525222,
|
| 6197 |
+
"grad_norm": 0.09259826689958572,
|
| 6198 |
+
"learning_rate": 8.396390424577043e-05,
|
| 6199 |
+
"loss": 10.3413,
|
| 6200 |
+
"step": 881
|
| 6201 |
+
},
|
| 6202 |
+
{
|
| 6203 |
+
"epoch": 0.14422369389256806,
|
| 6204 |
+
"grad_norm": 0.06901207566261292,
|
| 6205 |
+
"learning_rate": 8.376767433150491e-05,
|
| 6206 |
+
"loss": 10.3458,
|
| 6207 |
+
"step": 882
|
| 6208 |
+
},
|
| 6209 |
+
{
|
| 6210 |
+
"epoch": 0.1443872128198839,
|
| 6211 |
+
"grad_norm": 0.06953845173120499,
|
| 6212 |
+
"learning_rate": 8.357150859228351e-05,
|
| 6213 |
+
"loss": 10.3409,
|
| 6214 |
+
"step": 883
|
| 6215 |
+
},
|
| 6216 |
+
{
|
| 6217 |
+
"epoch": 0.14455073174719973,
|
| 6218 |
+
"grad_norm": 0.12560045719146729,
|
| 6219 |
+
"learning_rate": 8.337540780365401e-05,
|
| 6220 |
+
"loss": 10.3505,
|
| 6221 |
+
"step": 884
|
| 6222 |
+
},
|
| 6223 |
+
{
|
| 6224 |
+
"epoch": 0.14471425067451557,
|
| 6225 |
+
"grad_norm": 0.0934881642460823,
|
| 6226 |
+
"learning_rate": 8.317937274090747e-05,
|
| 6227 |
+
"loss": 10.3389,
|
| 6228 |
+
"step": 885
|
| 6229 |
+
},
|
| 6230 |
+
{
|
| 6231 |
+
"epoch": 0.1448777696018314,
|
| 6232 |
+
"grad_norm": 0.16856247186660767,
|
| 6233 |
+
"learning_rate": 8.298340417907512e-05,
|
| 6234 |
+
"loss": 10.3607,
|
| 6235 |
+
"step": 886
|
| 6236 |
+
},
|
| 6237 |
+
{
|
| 6238 |
+
"epoch": 0.14504128852914724,
|
| 6239 |
+
"grad_norm": 0.08886121213436127,
|
| 6240 |
+
"learning_rate": 8.278750289292521e-05,
|
| 6241 |
+
"loss": 10.3409,
|
| 6242 |
+
"step": 887
|
| 6243 |
+
},
|
| 6244 |
+
{
|
| 6245 |
+
"epoch": 0.14520480745646308,
|
| 6246 |
+
"grad_norm": 0.08185100555419922,
|
| 6247 |
+
"learning_rate": 8.259166965696003e-05,
|
| 6248 |
+
"loss": 10.3348,
|
| 6249 |
+
"step": 888
|
| 6250 |
+
},
|
| 6251 |
+
{
|
| 6252 |
+
"epoch": 0.14536832638377892,
|
| 6253 |
+
"grad_norm": 0.09273926913738251,
|
| 6254 |
+
"learning_rate": 8.239590524541293e-05,
|
| 6255 |
+
"loss": 10.3403,
|
| 6256 |
+
"step": 889
|
| 6257 |
+
},
|
| 6258 |
+
{
|
| 6259 |
+
"epoch": 0.14553184531109475,
|
| 6260 |
+
"grad_norm": 0.06138148531317711,
|
| 6261 |
+
"learning_rate": 8.2200210432245e-05,
|
| 6262 |
+
"loss": 10.3444,
|
| 6263 |
+
"step": 890
|
| 6264 |
+
},
|
| 6265 |
+
{
|
| 6266 |
+
"epoch": 0.1456953642384106,
|
| 6267 |
+
"grad_norm": 0.07617463916540146,
|
| 6268 |
+
"learning_rate": 8.200458599114228e-05,
|
| 6269 |
+
"loss": 10.3451,
|
| 6270 |
+
"step": 891
|
| 6271 |
+
},
|
| 6272 |
+
{
|
| 6273 |
+
"epoch": 0.14585888316572643,
|
| 6274 |
+
"grad_norm": 0.08602487295866013,
|
| 6275 |
+
"learning_rate": 8.180903269551255e-05,
|
| 6276 |
+
"loss": 10.3556,
|
| 6277 |
+
"step": 892
|
| 6278 |
+
},
|
| 6279 |
+
{
|
| 6280 |
+
"epoch": 0.14602240209304226,
|
| 6281 |
+
"grad_norm": 0.07287289947271347,
|
| 6282 |
+
"learning_rate": 8.161355131848231e-05,
|
| 6283 |
+
"loss": 10.3429,
|
| 6284 |
+
"step": 893
|
| 6285 |
+
},
|
| 6286 |
+
{
|
| 6287 |
+
"epoch": 0.1461859210203581,
|
| 6288 |
+
"grad_norm": 0.10442943125963211,
|
| 6289 |
+
"learning_rate": 8.14181426328938e-05,
|
| 6290 |
+
"loss": 10.3359,
|
| 6291 |
+
"step": 894
|
| 6292 |
+
},
|
| 6293 |
+
{
|
| 6294 |
+
"epoch": 0.14634943994767394,
|
| 6295 |
+
"grad_norm": 0.0643426850438118,
|
| 6296 |
+
"learning_rate": 8.122280741130176e-05,
|
| 6297 |
+
"loss": 10.3482,
|
| 6298 |
+
"step": 895
|
| 6299 |
+
},
|
| 6300 |
+
{
|
| 6301 |
+
"epoch": 0.14651295887498977,
|
| 6302 |
+
"grad_norm": 0.11469782143831253,
|
| 6303 |
+
"learning_rate": 8.102754642597055e-05,
|
| 6304 |
+
"loss": 10.331,
|
| 6305 |
+
"step": 896
|
| 6306 |
+
},
|
| 6307 |
+
{
|
| 6308 |
+
"epoch": 0.1466764778023056,
|
| 6309 |
+
"grad_norm": 0.0717199444770813,
|
| 6310 |
+
"learning_rate": 8.083236044887106e-05,
|
| 6311 |
+
"loss": 10.3453,
|
| 6312 |
+
"step": 897
|
| 6313 |
+
},
|
| 6314 |
+
{
|
| 6315 |
+
"epoch": 0.14683999672962145,
|
| 6316 |
+
"grad_norm": 0.0660734623670578,
|
| 6317 |
+
"learning_rate": 8.063725025167754e-05,
|
| 6318 |
+
"loss": 10.3432,
|
| 6319 |
+
"step": 898
|
| 6320 |
+
},
|
| 6321 |
+
{
|
| 6322 |
+
"epoch": 0.14700351565693728,
|
| 6323 |
+
"grad_norm": 0.0714247077703476,
|
| 6324 |
+
"learning_rate": 8.044221660576476e-05,
|
| 6325 |
+
"loss": 10.3418,
|
| 6326 |
+
"step": 899
|
| 6327 |
+
},
|
| 6328 |
+
{
|
| 6329 |
+
"epoch": 0.14716703458425312,
|
| 6330 |
+
"grad_norm": 0.07890897244215012,
|
| 6331 |
+
"learning_rate": 8.024726028220474e-05,
|
| 6332 |
+
"loss": 10.3439,
|
| 6333 |
+
"step": 900
|
| 6334 |
+
},
|
| 6335 |
+
{
|
| 6336 |
+
"epoch": 0.14733055351156896,
|
| 6337 |
+
"grad_norm": 0.05443431809544563,
|
| 6338 |
+
"learning_rate": 8.005238205176387e-05,
|
| 6339 |
+
"loss": 10.3504,
|
| 6340 |
+
"step": 901
|
| 6341 |
+
},
|
| 6342 |
+
{
|
| 6343 |
+
"epoch": 0.1474940724388848,
|
| 6344 |
+
"grad_norm": 0.09214995801448822,
|
| 6345 |
+
"learning_rate": 7.985758268489975e-05,
|
| 6346 |
+
"loss": 10.3455,
|
| 6347 |
+
"step": 902
|
| 6348 |
+
},
|
| 6349 |
+
{
|
| 6350 |
+
"epoch": 0.14765759136620063,
|
| 6351 |
+
"grad_norm": 0.08428926020860672,
|
| 6352 |
+
"learning_rate": 7.966286295175824e-05,
|
| 6353 |
+
"loss": 10.3499,
|
| 6354 |
+
"step": 903
|
| 6355 |
+
},
|
| 6356 |
+
{
|
| 6357 |
+
"epoch": 0.14782111029351647,
|
| 6358 |
+
"grad_norm": 0.1792377382516861,
|
| 6359 |
+
"learning_rate": 7.946822362217032e-05,
|
| 6360 |
+
"loss": 10.3274,
|
| 6361 |
+
"step": 904
|
| 6362 |
+
},
|
| 6363 |
+
{
|
| 6364 |
+
"epoch": 0.1479846292208323,
|
| 6365 |
+
"grad_norm": 0.06639645993709564,
|
| 6366 |
+
"learning_rate": 7.927366546564911e-05,
|
| 6367 |
+
"loss": 10.3454,
|
| 6368 |
+
"step": 905
|
| 6369 |
+
},
|
| 6370 |
+
{
|
| 6371 |
+
"epoch": 0.14814814814814814,
|
| 6372 |
+
"grad_norm": 0.13054105639457703,
|
| 6373 |
+
"learning_rate": 7.90791892513868e-05,
|
| 6374 |
+
"loss": 10.3522,
|
| 6375 |
+
"step": 906
|
| 6376 |
+
},
|
| 6377 |
+
{
|
| 6378 |
+
"epoch": 0.14831166707546398,
|
| 6379 |
+
"grad_norm": 0.08296360075473785,
|
| 6380 |
+
"learning_rate": 7.888479574825165e-05,
|
| 6381 |
+
"loss": 10.3411,
|
| 6382 |
+
"step": 907
|
| 6383 |
+
},
|
| 6384 |
+
{
|
| 6385 |
+
"epoch": 0.1484751860027798,
|
| 6386 |
+
"grad_norm": 0.0851949006319046,
|
| 6387 |
+
"learning_rate": 7.869048572478488e-05,
|
| 6388 |
+
"loss": 10.3364,
|
| 6389 |
+
"step": 908
|
| 6390 |
+
},
|
| 6391 |
+
{
|
| 6392 |
+
"epoch": 0.14863870493009565,
|
| 6393 |
+
"grad_norm": 0.08287424594163895,
|
| 6394 |
+
"learning_rate": 7.849625994919767e-05,
|
| 6395 |
+
"loss": 10.3472,
|
| 6396 |
+
"step": 909
|
| 6397 |
+
},
|
| 6398 |
+
{
|
| 6399 |
+
"epoch": 0.1488022238574115,
|
| 6400 |
+
"grad_norm": 0.05285567045211792,
|
| 6401 |
+
"learning_rate": 7.83021191893682e-05,
|
| 6402 |
+
"loss": 10.3501,
|
| 6403 |
+
"step": 910
|
| 6404 |
+
},
|
| 6405 |
+
{
|
| 6406 |
+
"epoch": 0.14896574278472732,
|
| 6407 |
+
"grad_norm": 0.06834254413843155,
|
| 6408 |
+
"learning_rate": 7.810806421283841e-05,
|
| 6409 |
+
"loss": 10.3607,
|
| 6410 |
+
"step": 911
|
| 6411 |
+
},
|
| 6412 |
+
{
|
| 6413 |
+
"epoch": 0.14912926171204316,
|
| 6414 |
+
"grad_norm": 0.058842241764068604,
|
| 6415 |
+
"learning_rate": 7.791409578681121e-05,
|
| 6416 |
+
"loss": 10.3407,
|
| 6417 |
+
"step": 912
|
| 6418 |
+
},
|
| 6419 |
+
{
|
| 6420 |
+
"epoch": 0.149292780639359,
|
| 6421 |
+
"grad_norm": 0.11524898558855057,
|
| 6422 |
+
"learning_rate": 7.772021467814723e-05,
|
| 6423 |
+
"loss": 10.3466,
|
| 6424 |
+
"step": 913
|
| 6425 |
+
},
|
| 6426 |
+
{
|
| 6427 |
+
"epoch": 0.14945629956667483,
|
| 6428 |
+
"grad_norm": 0.08459778130054474,
|
| 6429 |
+
"learning_rate": 7.752642165336195e-05,
|
| 6430 |
+
"loss": 10.3529,
|
| 6431 |
+
"step": 914
|
| 6432 |
+
},
|
| 6433 |
+
{
|
| 6434 |
+
"epoch": 0.14961981849399067,
|
| 6435 |
+
"grad_norm": 0.08150472491979599,
|
| 6436 |
+
"learning_rate": 7.733271747862265e-05,
|
| 6437 |
+
"loss": 10.3447,
|
| 6438 |
+
"step": 915
|
| 6439 |
+
},
|
| 6440 |
+
{
|
| 6441 |
+
"epoch": 0.1497833374213065,
|
| 6442 |
+
"grad_norm": 0.07789532095193863,
|
| 6443 |
+
"learning_rate": 7.713910291974527e-05,
|
| 6444 |
+
"loss": 10.3467,
|
| 6445 |
+
"step": 916
|
| 6446 |
+
},
|
| 6447 |
+
{
|
| 6448 |
+
"epoch": 0.14994685634862234,
|
| 6449 |
+
"grad_norm": 0.06128161400556564,
|
| 6450 |
+
"learning_rate": 7.694557874219138e-05,
|
| 6451 |
+
"loss": 10.3517,
|
| 6452 |
+
"step": 917
|
| 6453 |
+
},
|
| 6454 |
+
{
|
| 6455 |
+
"epoch": 0.15011037527593818,
|
| 6456 |
+
"grad_norm": 0.06433887779712677,
|
| 6457 |
+
"learning_rate": 7.675214571106536e-05,
|
| 6458 |
+
"loss": 10.3451,
|
| 6459 |
+
"step": 918
|
| 6460 |
+
},
|
| 6461 |
+
{
|
| 6462 |
+
"epoch": 0.15027389420325402,
|
| 6463 |
+
"grad_norm": 0.13994459807872772,
|
| 6464 |
+
"learning_rate": 7.655880459111123e-05,
|
| 6465 |
+
"loss": 10.3421,
|
| 6466 |
+
"step": 919
|
| 6467 |
+
},
|
| 6468 |
+
{
|
| 6469 |
+
"epoch": 0.15043741313056985,
|
| 6470 |
+
"grad_norm": 0.069017194211483,
|
| 6471 |
+
"learning_rate": 7.636555614670953e-05,
|
| 6472 |
+
"loss": 10.352,
|
| 6473 |
+
"step": 920
|
| 6474 |
+
},
|
| 6475 |
+
{
|
| 6476 |
+
"epoch": 0.1506009320578857,
|
| 6477 |
+
"grad_norm": 0.08271420747041702,
|
| 6478 |
+
"learning_rate": 7.617240114187452e-05,
|
| 6479 |
+
"loss": 10.3448,
|
| 6480 |
+
"step": 921
|
| 6481 |
+
},
|
| 6482 |
+
{
|
| 6483 |
+
"epoch": 0.15076445098520153,
|
| 6484 |
+
"grad_norm": 0.09887969493865967,
|
| 6485 |
+
"learning_rate": 7.597934034025092e-05,
|
| 6486 |
+
"loss": 10.3531,
|
| 6487 |
+
"step": 922
|
| 6488 |
+
},
|
| 6489 |
+
{
|
| 6490 |
+
"epoch": 0.15092796991251736,
|
| 6491 |
+
"grad_norm": 0.1021571010351181,
|
| 6492 |
+
"learning_rate": 7.578637450511116e-05,
|
| 6493 |
+
"loss": 10.3412,
|
| 6494 |
+
"step": 923
|
| 6495 |
+
},
|
| 6496 |
+
{
|
| 6497 |
+
"epoch": 0.1510914888398332,
|
| 6498 |
+
"grad_norm": 0.08674758672714233,
|
| 6499 |
+
"learning_rate": 7.559350439935213e-05,
|
| 6500 |
+
"loss": 10.3403,
|
| 6501 |
+
"step": 924
|
| 6502 |
+
},
|
| 6503 |
+
{
|
| 6504 |
+
"epoch": 0.15125500776714904,
|
| 6505 |
+
"grad_norm": 0.08601857721805573,
|
| 6506 |
+
"learning_rate": 7.540073078549221e-05,
|
| 6507 |
+
"loss": 10.3514,
|
| 6508 |
+
"step": 925
|
| 6509 |
+
},
|
| 6510 |
+
{
|
| 6511 |
+
"epoch": 0.15141852669446487,
|
| 6512 |
+
"grad_norm": 0.09482517093420029,
|
| 6513 |
+
"learning_rate": 7.52080544256684e-05,
|
| 6514 |
+
"loss": 10.3469,
|
| 6515 |
+
"step": 926
|
| 6516 |
+
},
|
| 6517 |
+
{
|
| 6518 |
+
"epoch": 0.1515820456217807,
|
| 6519 |
+
"grad_norm": 0.0778004452586174,
|
| 6520 |
+
"learning_rate": 7.501547608163317e-05,
|
| 6521 |
+
"loss": 10.3439,
|
| 6522 |
+
"step": 927
|
| 6523 |
+
},
|
| 6524 |
+
{
|
| 6525 |
+
"epoch": 0.15174556454909655,
|
| 6526 |
+
"grad_norm": 0.05390161648392677,
|
| 6527 |
+
"learning_rate": 7.48229965147514e-05,
|
| 6528 |
+
"loss": 10.3413,
|
| 6529 |
+
"step": 928
|
| 6530 |
+
},
|
| 6531 |
+
{
|
| 6532 |
+
"epoch": 0.15190908347641238,
|
| 6533 |
+
"grad_norm": 0.11087460815906525,
|
| 6534 |
+
"learning_rate": 7.463061648599757e-05,
|
| 6535 |
+
"loss": 10.347,
|
| 6536 |
+
"step": 929
|
| 6537 |
+
},
|
| 6538 |
+
{
|
| 6539 |
+
"epoch": 0.15207260240372822,
|
| 6540 |
+
"grad_norm": 0.16281658411026,
|
| 6541 |
+
"learning_rate": 7.443833675595255e-05,
|
| 6542 |
+
"loss": 10.3528,
|
| 6543 |
+
"step": 930
|
| 6544 |
+
},
|
| 6545 |
+
{
|
| 6546 |
+
"epoch": 0.15223612133104406,
|
| 6547 |
+
"grad_norm": 0.0704844668507576,
|
| 6548 |
+
"learning_rate": 7.424615808480068e-05,
|
| 6549 |
+
"loss": 10.335,
|
| 6550 |
+
"step": 931
|
| 6551 |
+
},
|
| 6552 |
+
{
|
| 6553 |
+
"epoch": 0.1523996402583599,
|
| 6554 |
+
"grad_norm": 0.07584381848573685,
|
| 6555 |
+
"learning_rate": 7.405408123232686e-05,
|
| 6556 |
+
"loss": 10.3593,
|
| 6557 |
+
"step": 932
|
| 6558 |
+
},
|
| 6559 |
+
{
|
| 6560 |
+
"epoch": 0.15256315918567573,
|
| 6561 |
+
"grad_norm": 0.09159449487924576,
|
| 6562 |
+
"learning_rate": 7.386210695791332e-05,
|
| 6563 |
+
"loss": 10.3342,
|
| 6564 |
+
"step": 933
|
| 6565 |
+
},
|
| 6566 |
+
{
|
| 6567 |
+
"epoch": 0.15272667811299157,
|
| 6568 |
+
"grad_norm": 0.06305116415023804,
|
| 6569 |
+
"learning_rate": 7.367023602053679e-05,
|
| 6570 |
+
"loss": 10.3401,
|
| 6571 |
+
"step": 934
|
| 6572 |
+
},
|
| 6573 |
+
{
|
| 6574 |
+
"epoch": 0.1528901970403074,
|
| 6575 |
+
"grad_norm": 0.07074683904647827,
|
| 6576 |
+
"learning_rate": 7.347846917876544e-05,
|
| 6577 |
+
"loss": 10.3353,
|
| 6578 |
+
"step": 935
|
| 6579 |
+
},
|
| 6580 |
+
{
|
| 6581 |
+
"epoch": 0.15305371596762327,
|
| 6582 |
+
"grad_norm": 0.13172592222690582,
|
| 6583 |
+
"learning_rate": 7.328680719075596e-05,
|
| 6584 |
+
"loss": 10.3611,
|
| 6585 |
+
"step": 936
|
| 6586 |
+
},
|
| 6587 |
+
{
|
| 6588 |
+
"epoch": 0.1532172348949391,
|
| 6589 |
+
"grad_norm": 0.07162374258041382,
|
| 6590 |
+
"learning_rate": 7.309525081425044e-05,
|
| 6591 |
+
"loss": 10.3556,
|
| 6592 |
+
"step": 937
|
| 6593 |
+
},
|
| 6594 |
+
{
|
| 6595 |
+
"epoch": 0.15338075382225494,
|
| 6596 |
+
"grad_norm": 0.061027929186820984,
|
| 6597 |
+
"learning_rate": 7.29038008065734e-05,
|
| 6598 |
+
"loss": 10.3508,
|
| 6599 |
+
"step": 938
|
| 6600 |
+
},
|
| 6601 |
+
{
|
| 6602 |
+
"epoch": 0.15354427274957078,
|
| 6603 |
+
"grad_norm": 0.09178778529167175,
|
| 6604 |
+
"learning_rate": 7.271245792462891e-05,
|
| 6605 |
+
"loss": 10.3429,
|
| 6606 |
+
"step": 939
|
| 6607 |
+
},
|
| 6608 |
+
{
|
| 6609 |
+
"epoch": 0.1537077916768866,
|
| 6610 |
+
"grad_norm": 0.09651417285203934,
|
| 6611 |
+
"learning_rate": 7.252122292489747e-05,
|
| 6612 |
+
"loss": 10.338,
|
| 6613 |
+
"step": 940
|
| 6614 |
+
},
|
| 6615 |
+
{
|
| 6616 |
+
"epoch": 0.15387131060420245,
|
| 6617 |
+
"grad_norm": 0.0864914059638977,
|
| 6618 |
+
"learning_rate": 7.233009656343304e-05,
|
| 6619 |
+
"loss": 10.3417,
|
| 6620 |
+
"step": 941
|
| 6621 |
+
},
|
| 6622 |
+
{
|
| 6623 |
+
"epoch": 0.15403482953151829,
|
| 6624 |
+
"grad_norm": 0.06251879036426544,
|
| 6625 |
+
"learning_rate": 7.213907959586014e-05,
|
| 6626 |
+
"loss": 10.3404,
|
| 6627 |
+
"step": 942
|
| 6628 |
+
},
|
| 6629 |
+
{
|
| 6630 |
+
"epoch": 0.15419834845883412,
|
| 6631 |
+
"grad_norm": 0.12224212288856506,
|
| 6632 |
+
"learning_rate": 7.194817277737071e-05,
|
| 6633 |
+
"loss": 10.3551,
|
| 6634 |
+
"step": 943
|
| 6635 |
+
},
|
| 6636 |
+
{
|
| 6637 |
+
"epoch": 0.15436186738614996,
|
| 6638 |
+
"grad_norm": 0.08379900455474854,
|
| 6639 |
+
"learning_rate": 7.17573768627213e-05,
|
| 6640 |
+
"loss": 10.3439,
|
| 6641 |
+
"step": 944
|
| 6642 |
+
},
|
| 6643 |
+
{
|
| 6644 |
+
"epoch": 0.1545253863134658,
|
| 6645 |
+
"grad_norm": 0.06541034579277039,
|
| 6646 |
+
"learning_rate": 7.156669260622996e-05,
|
| 6647 |
+
"loss": 10.3493,
|
| 6648 |
+
"step": 945
|
| 6649 |
+
},
|
| 6650 |
+
{
|
| 6651 |
+
"epoch": 0.15468890524078163,
|
| 6652 |
+
"grad_norm": 0.08822879940271378,
|
| 6653 |
+
"learning_rate": 7.137612076177329e-05,
|
| 6654 |
+
"loss": 10.3484,
|
| 6655 |
+
"step": 946
|
| 6656 |
+
},
|
| 6657 |
+
{
|
| 6658 |
+
"epoch": 0.15485242416809747,
|
| 6659 |
+
"grad_norm": 0.07290088385343552,
|
| 6660 |
+
"learning_rate": 7.118566208278346e-05,
|
| 6661 |
+
"loss": 10.3446,
|
| 6662 |
+
"step": 947
|
| 6663 |
+
},
|
| 6664 |
+
{
|
| 6665 |
+
"epoch": 0.1550159430954133,
|
| 6666 |
+
"grad_norm": 0.062057286500930786,
|
| 6667 |
+
"learning_rate": 7.099531732224524e-05,
|
| 6668 |
+
"loss": 10.3438,
|
| 6669 |
+
"step": 948
|
| 6670 |
+
},
|
| 6671 |
+
{
|
| 6672 |
+
"epoch": 0.15517946202272914,
|
| 6673 |
+
"grad_norm": 0.1268312633037567,
|
| 6674 |
+
"learning_rate": 7.080508723269308e-05,
|
| 6675 |
+
"loss": 10.3384,
|
| 6676 |
+
"step": 949
|
| 6677 |
+
},
|
| 6678 |
+
{
|
| 6679 |
+
"epoch": 0.15534298095004498,
|
| 6680 |
+
"grad_norm": 0.05009286105632782,
|
| 6681 |
+
"learning_rate": 7.061497256620793e-05,
|
| 6682 |
+
"loss": 10.3511,
|
| 6683 |
+
"step": 950
|
| 6684 |
+
},
|
| 6685 |
+
{
|
| 6686 |
+
"epoch": 0.15550649987736082,
|
| 6687 |
+
"grad_norm": 0.0904565081000328,
|
| 6688 |
+
"learning_rate": 7.042497407441457e-05,
|
| 6689 |
+
"loss": 10.3454,
|
| 6690 |
+
"step": 951
|
| 6691 |
+
},
|
| 6692 |
+
{
|
| 6693 |
+
"epoch": 0.15567001880467665,
|
| 6694 |
+
"grad_norm": 0.07426796853542328,
|
| 6695 |
+
"learning_rate": 7.023509250847838e-05,
|
| 6696 |
+
"loss": 10.3345,
|
| 6697 |
+
"step": 952
|
| 6698 |
+
},
|
| 6699 |
+
{
|
| 6700 |
+
"epoch": 0.1558335377319925,
|
| 6701 |
+
"grad_norm": 0.09245802462100983,
|
| 6702 |
+
"learning_rate": 7.004532861910251e-05,
|
| 6703 |
+
"loss": 10.3428,
|
| 6704 |
+
"step": 953
|
| 6705 |
+
},
|
| 6706 |
+
{
|
| 6707 |
+
"epoch": 0.15599705665930833,
|
| 6708 |
+
"grad_norm": 0.07755796611309052,
|
| 6709 |
+
"learning_rate": 6.985568315652488e-05,
|
| 6710 |
+
"loss": 10.3531,
|
| 6711 |
+
"step": 954
|
| 6712 |
+
},
|
| 6713 |
+
{
|
| 6714 |
+
"epoch": 0.15616057558662416,
|
| 6715 |
+
"grad_norm": 0.08339790254831314,
|
| 6716 |
+
"learning_rate": 6.966615687051516e-05,
|
| 6717 |
+
"loss": 10.3439,
|
| 6718 |
+
"step": 955
|
| 6719 |
+
},
|
| 6720 |
+
{
|
| 6721 |
+
"epoch": 0.15632409451394,
|
| 6722 |
+
"grad_norm": 0.06234053149819374,
|
| 6723 |
+
"learning_rate": 6.947675051037192e-05,
|
| 6724 |
+
"loss": 10.3468,
|
| 6725 |
+
"step": 956
|
| 6726 |
+
},
|
| 6727 |
+
{
|
| 6728 |
+
"epoch": 0.15648761344125584,
|
| 6729 |
+
"grad_norm": 0.06956285983324051,
|
| 6730 |
+
"learning_rate": 6.928746482491952e-05,
|
| 6731 |
+
"loss": 10.3473,
|
| 6732 |
+
"step": 957
|
| 6733 |
+
},
|
| 6734 |
+
{
|
| 6735 |
+
"epoch": 0.15665113236857167,
|
| 6736 |
+
"grad_norm": 0.06687292456626892,
|
| 6737 |
+
"learning_rate": 6.909830056250527e-05,
|
| 6738 |
+
"loss": 10.345,
|
| 6739 |
+
"step": 958
|
| 6740 |
+
},
|
| 6741 |
+
{
|
| 6742 |
+
"epoch": 0.1568146512958875,
|
| 6743 |
+
"grad_norm": 0.10145143419504166,
|
| 6744 |
+
"learning_rate": 6.890925847099645e-05,
|
| 6745 |
+
"loss": 10.3386,
|
| 6746 |
+
"step": 959
|
| 6747 |
+
},
|
| 6748 |
+
{
|
| 6749 |
+
"epoch": 0.15697817022320334,
|
| 6750 |
+
"grad_norm": 0.07209116965532303,
|
| 6751 |
+
"learning_rate": 6.87203392977773e-05,
|
| 6752 |
+
"loss": 10.3356,
|
| 6753 |
+
"step": 960
|
| 6754 |
+
},
|
| 6755 |
+
{
|
| 6756 |
+
"epoch": 0.15714168915051918,
|
| 6757 |
+
"grad_norm": 0.05291639640927315,
|
| 6758 |
+
"learning_rate": 6.85315437897461e-05,
|
| 6759 |
+
"loss": 10.3493,
|
| 6760 |
+
"step": 961
|
| 6761 |
+
},
|
| 6762 |
+
{
|
| 6763 |
+
"epoch": 0.15730520807783502,
|
| 6764 |
+
"grad_norm": 0.06835281848907471,
|
| 6765 |
+
"learning_rate": 6.834287269331226e-05,
|
| 6766 |
+
"loss": 10.3414,
|
| 6767 |
+
"step": 962
|
| 6768 |
+
},
|
| 6769 |
+
{
|
| 6770 |
+
"epoch": 0.15746872700515085,
|
| 6771 |
+
"grad_norm": 0.11120449751615524,
|
| 6772 |
+
"learning_rate": 6.815432675439322e-05,
|
| 6773 |
+
"loss": 10.3339,
|
| 6774 |
+
"step": 963
|
| 6775 |
+
},
|
| 6776 |
+
{
|
| 6777 |
+
"epoch": 0.1576322459324667,
|
| 6778 |
+
"grad_norm": 0.11330325901508331,
|
| 6779 |
+
"learning_rate": 6.796590671841177e-05,
|
| 6780 |
+
"loss": 10.3388,
|
| 6781 |
+
"step": 964
|
| 6782 |
+
},
|
| 6783 |
+
{
|
| 6784 |
+
"epoch": 0.15779576485978253,
|
| 6785 |
+
"grad_norm": 0.08197707682847977,
|
| 6786 |
+
"learning_rate": 6.777761333029275e-05,
|
| 6787 |
+
"loss": 10.3441,
|
| 6788 |
+
"step": 965
|
| 6789 |
+
},
|
| 6790 |
+
{
|
| 6791 |
+
"epoch": 0.15795928378709836,
|
| 6792 |
+
"grad_norm": 0.12177684158086777,
|
| 6793 |
+
"learning_rate": 6.758944733446041e-05,
|
| 6794 |
+
"loss": 10.3344,
|
| 6795 |
+
"step": 966
|
| 6796 |
+
},
|
| 6797 |
+
{
|
| 6798 |
+
"epoch": 0.1581228027144142,
|
| 6799 |
+
"grad_norm": 0.08609756827354431,
|
| 6800 |
+
"learning_rate": 6.740140947483541e-05,
|
| 6801 |
+
"loss": 10.3436,
|
| 6802 |
+
"step": 967
|
| 6803 |
+
},
|
| 6804 |
+
{
|
| 6805 |
+
"epoch": 0.15828632164173004,
|
| 6806 |
+
"grad_norm": 0.06757651269435883,
|
| 6807 |
+
"learning_rate": 6.721350049483162e-05,
|
| 6808 |
+
"loss": 10.3526,
|
| 6809 |
+
"step": 968
|
| 6810 |
+
},
|
| 6811 |
+
{
|
| 6812 |
+
"epoch": 0.15844984056904587,
|
| 6813 |
+
"grad_norm": 0.10005096346139908,
|
| 6814 |
+
"learning_rate": 6.702572113735358e-05,
|
| 6815 |
+
"loss": 10.3518,
|
| 6816 |
+
"step": 969
|
| 6817 |
+
},
|
| 6818 |
+
{
|
| 6819 |
+
"epoch": 0.1586133594963617,
|
| 6820 |
+
"grad_norm": 0.07108530402183533,
|
| 6821 |
+
"learning_rate": 6.683807214479323e-05,
|
| 6822 |
+
"loss": 10.3425,
|
| 6823 |
+
"step": 970
|
| 6824 |
+
},
|
| 6825 |
+
{
|
| 6826 |
+
"epoch": 0.15877687842367755,
|
| 6827 |
+
"grad_norm": 0.07789983600378036,
|
| 6828 |
+
"learning_rate": 6.665055425902716e-05,
|
| 6829 |
+
"loss": 10.3454,
|
| 6830 |
+
"step": 971
|
| 6831 |
+
},
|
| 6832 |
+
{
|
| 6833 |
+
"epoch": 0.15894039735099338,
|
| 6834 |
+
"grad_norm": 0.08780791610479355,
|
| 6835 |
+
"learning_rate": 6.646316822141368e-05,
|
| 6836 |
+
"loss": 10.3375,
|
| 6837 |
+
"step": 972
|
| 6838 |
+
},
|
| 6839 |
+
{
|
| 6840 |
+
"epoch": 0.15910391627830922,
|
| 6841 |
+
"grad_norm": 0.04945690557360649,
|
| 6842 |
+
"learning_rate": 6.627591477278969e-05,
|
| 6843 |
+
"loss": 10.3404,
|
| 6844 |
+
"step": 973
|
| 6845 |
+
},
|
| 6846 |
+
{
|
| 6847 |
+
"epoch": 0.15926743520562506,
|
| 6848 |
+
"grad_norm": 0.09763111919164658,
|
| 6849 |
+
"learning_rate": 6.608879465346804e-05,
|
| 6850 |
+
"loss": 10.3646,
|
| 6851 |
+
"step": 974
|
| 6852 |
+
},
|
| 6853 |
+
{
|
| 6854 |
+
"epoch": 0.1594309541329409,
|
| 6855 |
+
"grad_norm": 0.09061010926961899,
|
| 6856 |
+
"learning_rate": 6.59018086032344e-05,
|
| 6857 |
+
"loss": 10.3494,
|
| 6858 |
+
"step": 975
|
| 6859 |
+
},
|
| 6860 |
+
{
|
| 6861 |
+
"epoch": 0.15959447306025673,
|
| 6862 |
+
"grad_norm": 0.13620643317699432,
|
| 6863 |
+
"learning_rate": 6.571495736134434e-05,
|
| 6864 |
+
"loss": 10.3502,
|
| 6865 |
+
"step": 976
|
| 6866 |
+
},
|
| 6867 |
+
{
|
| 6868 |
+
"epoch": 0.15975799198757257,
|
| 6869 |
+
"grad_norm": 0.10544710606336594,
|
| 6870 |
+
"learning_rate": 6.552824166652059e-05,
|
| 6871 |
+
"loss": 10.3366,
|
| 6872 |
+
"step": 977
|
| 6873 |
+
},
|
| 6874 |
+
{
|
| 6875 |
+
"epoch": 0.1599215109148884,
|
| 6876 |
+
"grad_norm": 0.062366314232349396,
|
| 6877 |
+
"learning_rate": 6.534166225694984e-05,
|
| 6878 |
+
"loss": 10.3492,
|
| 6879 |
+
"step": 978
|
| 6880 |
+
},
|
| 6881 |
+
{
|
| 6882 |
+
"epoch": 0.16008502984220424,
|
| 6883 |
+
"grad_norm": 0.07919897139072418,
|
| 6884 |
+
"learning_rate": 6.515521987028009e-05,
|
| 6885 |
+
"loss": 10.3515,
|
| 6886 |
+
"step": 979
|
| 6887 |
+
},
|
| 6888 |
+
{
|
| 6889 |
+
"epoch": 0.16024854876952008,
|
| 6890 |
+
"grad_norm": 0.08079025149345398,
|
| 6891 |
+
"learning_rate": 6.496891524361757e-05,
|
| 6892 |
+
"loss": 10.3453,
|
| 6893 |
+
"step": 980
|
| 6894 |
+
},
|
| 6895 |
+
{
|
| 6896 |
+
"epoch": 0.1604120676968359,
|
| 6897 |
+
"grad_norm": 0.1006539836525917,
|
| 6898 |
+
"learning_rate": 6.478274911352385e-05,
|
| 6899 |
+
"loss": 10.3425,
|
| 6900 |
+
"step": 981
|
| 6901 |
+
},
|
| 6902 |
+
{
|
| 6903 |
+
"epoch": 0.16057558662415175,
|
| 6904 |
+
"grad_norm": 0.06720111519098282,
|
| 6905 |
+
"learning_rate": 6.4596722216013e-05,
|
| 6906 |
+
"loss": 10.3407,
|
| 6907 |
+
"step": 982
|
| 6908 |
+
},
|
| 6909 |
+
{
|
| 6910 |
+
"epoch": 0.1607391055514676,
|
| 6911 |
+
"grad_norm": 0.07172605395317078,
|
| 6912 |
+
"learning_rate": 6.441083528654856e-05,
|
| 6913 |
+
"loss": 10.3534,
|
| 6914 |
+
"step": 983
|
| 6915 |
+
},
|
| 6916 |
+
{
|
| 6917 |
+
"epoch": 0.16090262447878342,
|
| 6918 |
+
"grad_norm": 0.06686363369226456,
|
| 6919 |
+
"learning_rate": 6.422508906004078e-05,
|
| 6920 |
+
"loss": 10.3495,
|
| 6921 |
+
"step": 984
|
| 6922 |
+
},
|
| 6923 |
+
{
|
| 6924 |
+
"epoch": 0.16106614340609926,
|
| 6925 |
+
"grad_norm": 0.08130093663930893,
|
| 6926 |
+
"learning_rate": 6.403948427084356e-05,
|
| 6927 |
+
"loss": 10.3437,
|
| 6928 |
+
"step": 985
|
| 6929 |
+
},
|
| 6930 |
+
{
|
| 6931 |
+
"epoch": 0.1612296623334151,
|
| 6932 |
+
"grad_norm": 0.0772842988371849,
|
| 6933 |
+
"learning_rate": 6.38540216527517e-05,
|
| 6934 |
+
"loss": 10.342,
|
| 6935 |
+
"step": 986
|
| 6936 |
+
},
|
| 6937 |
+
{
|
| 6938 |
+
"epoch": 0.16139318126073093,
|
| 6939 |
+
"grad_norm": 0.06907609850168228,
|
| 6940 |
+
"learning_rate": 6.366870193899784e-05,
|
| 6941 |
+
"loss": 10.3424,
|
| 6942 |
+
"step": 987
|
| 6943 |
+
},
|
| 6944 |
+
{
|
| 6945 |
+
"epoch": 0.16155670018804677,
|
| 6946 |
+
"grad_norm": 0.07291463762521744,
|
| 6947 |
+
"learning_rate": 6.348352586224972e-05,
|
| 6948 |
+
"loss": 10.3475,
|
| 6949 |
+
"step": 988
|
| 6950 |
+
},
|
| 6951 |
+
{
|
| 6952 |
+
"epoch": 0.1617202191153626,
|
| 6953 |
+
"grad_norm": 0.09994920343160629,
|
| 6954 |
+
"learning_rate": 6.329849415460717e-05,
|
| 6955 |
+
"loss": 10.3437,
|
| 6956 |
+
"step": 989
|
| 6957 |
+
},
|
| 6958 |
+
{
|
| 6959 |
+
"epoch": 0.16188373804267844,
|
| 6960 |
+
"grad_norm": 0.07714927941560745,
|
| 6961 |
+
"learning_rate": 6.311360754759923e-05,
|
| 6962 |
+
"loss": 10.3489,
|
| 6963 |
+
"step": 990
|
| 6964 |
+
},
|
| 6965 |
+
{
|
| 6966 |
+
"epoch": 0.16204725696999428,
|
| 6967 |
+
"grad_norm": 0.08187807351350784,
|
| 6968 |
+
"learning_rate": 6.292886677218134e-05,
|
| 6969 |
+
"loss": 10.3495,
|
| 6970 |
+
"step": 991
|
| 6971 |
+
},
|
| 6972 |
+
{
|
| 6973 |
+
"epoch": 0.16221077589731012,
|
| 6974 |
+
"grad_norm": 0.23332437872886658,
|
| 6975 |
+
"learning_rate": 6.27442725587323e-05,
|
| 6976 |
+
"loss": 10.3674,
|
| 6977 |
+
"step": 992
|
| 6978 |
+
},
|
| 6979 |
+
{
|
| 6980 |
+
"epoch": 0.16237429482462595,
|
| 6981 |
+
"grad_norm": 0.09756692498922348,
|
| 6982 |
+
"learning_rate": 6.255982563705159e-05,
|
| 6983 |
+
"loss": 10.3367,
|
| 6984 |
+
"step": 993
|
| 6985 |
+
},
|
| 6986 |
+
{
|
| 6987 |
+
"epoch": 0.1625378137519418,
|
| 6988 |
+
"grad_norm": 0.0997442975640297,
|
| 6989 |
+
"learning_rate": 6.237552673635624e-05,
|
| 6990 |
+
"loss": 10.3521,
|
| 6991 |
+
"step": 994
|
| 6992 |
+
},
|
| 6993 |
+
{
|
| 6994 |
+
"epoch": 0.16270133267925763,
|
| 6995 |
+
"grad_norm": 0.05881795287132263,
|
| 6996 |
+
"learning_rate": 6.219137658527818e-05,
|
| 6997 |
+
"loss": 10.3434,
|
| 6998 |
+
"step": 995
|
| 6999 |
+
},
|
| 7000 |
+
{
|
| 7001 |
+
"epoch": 0.16286485160657346,
|
| 7002 |
+
"grad_norm": 0.08922906965017319,
|
| 7003 |
+
"learning_rate": 6.20073759118612e-05,
|
| 7004 |
+
"loss": 10.3469,
|
| 7005 |
+
"step": 996
|
| 7006 |
+
},
|
| 7007 |
+
{
|
| 7008 |
+
"epoch": 0.1630283705338893,
|
| 7009 |
+
"grad_norm": 0.09267131984233856,
|
| 7010 |
+
"learning_rate": 6.182352544355814e-05,
|
| 7011 |
+
"loss": 10.3435,
|
| 7012 |
+
"step": 997
|
| 7013 |
+
},
|
| 7014 |
+
{
|
| 7015 |
+
"epoch": 0.16319188946120514,
|
| 7016 |
+
"grad_norm": 0.09405411779880524,
|
| 7017 |
+
"learning_rate": 6.163982590722797e-05,
|
| 7018 |
+
"loss": 10.3454,
|
| 7019 |
+
"step": 998
|
| 7020 |
+
},
|
| 7021 |
+
{
|
| 7022 |
+
"epoch": 0.16335540838852097,
|
| 7023 |
+
"grad_norm": 0.10673796385526657,
|
| 7024 |
+
"learning_rate": 6.1456278029133e-05,
|
| 7025 |
+
"loss": 10.3344,
|
| 7026 |
+
"step": 999
|
| 7027 |
+
},
|
| 7028 |
+
{
|
| 7029 |
+
"epoch": 0.1635189273158368,
|
| 7030 |
+
"grad_norm": 0.09203071892261505,
|
| 7031 |
+
"learning_rate": 6.127288253493591e-05,
|
| 7032 |
+
"loss": 10.3426,
|
| 7033 |
+
"step": 1000
|
| 7034 |
+
},
|
| 7035 |
+
{
|
| 7036 |
+
"epoch": 0.16368244624315265,
|
| 7037 |
+
"grad_norm": 0.06706003844738007,
|
| 7038 |
+
"learning_rate": 6.108964014969692e-05,
|
| 7039 |
+
"loss": 10.3421,
|
| 7040 |
+
"step": 1001
|
| 7041 |
+
},
|
| 7042 |
+
{
|
| 7043 |
+
"epoch": 0.16384596517046848,
|
| 7044 |
+
"grad_norm": 0.07696403563022614,
|
| 7045 |
+
"learning_rate": 6.0906551597871e-05,
|
| 7046 |
+
"loss": 10.3428,
|
| 7047 |
+
"step": 1002
|
| 7048 |
+
},
|
| 7049 |
+
{
|
| 7050 |
+
"epoch": 0.16400948409778432,
|
| 7051 |
+
"grad_norm": 0.13336971402168274,
|
| 7052 |
+
"learning_rate": 6.07236176033048e-05,
|
| 7053 |
+
"loss": 10.3566,
|
| 7054 |
+
"step": 1003
|
| 7055 |
+
},
|
| 7056 |
+
{
|
| 7057 |
+
"epoch": 0.16417300302510016,
|
| 7058 |
+
"grad_norm": 0.08602435886859894,
|
| 7059 |
+
"learning_rate": 6.054083888923407e-05,
|
| 7060 |
+
"loss": 10.3445,
|
| 7061 |
+
"step": 1004
|
| 7062 |
+
},
|
| 7063 |
+
{
|
| 7064 |
+
"epoch": 0.164336521952416,
|
| 7065 |
+
"grad_norm": 0.055001333355903625,
|
| 7066 |
+
"learning_rate": 6.03582161782806e-05,
|
| 7067 |
+
"loss": 10.3419,
|
| 7068 |
+
"step": 1005
|
| 7069 |
+
},
|
| 7070 |
+
{
|
| 7071 |
+
"epoch": 0.16450004087973183,
|
| 7072 |
+
"grad_norm": 0.14026737213134766,
|
| 7073 |
+
"learning_rate": 6.0175750192449346e-05,
|
| 7074 |
+
"loss": 10.3403,
|
| 7075 |
+
"step": 1006
|
| 7076 |
+
},
|
| 7077 |
+
{
|
| 7078 |
+
"epoch": 0.16466355980704767,
|
| 7079 |
+
"grad_norm": 0.09036275744438171,
|
| 7080 |
+
"learning_rate": 5.999344165312576e-05,
|
| 7081 |
+
"loss": 10.3398,
|
| 7082 |
+
"step": 1007
|
| 7083 |
+
},
|
| 7084 |
+
{
|
| 7085 |
+
"epoch": 0.1648270787343635,
|
| 7086 |
+
"grad_norm": 0.08993495255708694,
|
| 7087 |
+
"learning_rate": 5.981129128107272e-05,
|
| 7088 |
+
"loss": 10.347,
|
| 7089 |
+
"step": 1008
|
| 7090 |
+
},
|
| 7091 |
+
{
|
| 7092 |
+
"epoch": 0.16499059766167934,
|
| 7093 |
+
"grad_norm": 0.10535188764333725,
|
| 7094 |
+
"learning_rate": 5.962929979642784e-05,
|
| 7095 |
+
"loss": 10.35,
|
| 7096 |
+
"step": 1009
|
| 7097 |
+
},
|
| 7098 |
+
{
|
| 7099 |
+
"epoch": 0.16515411658899518,
|
| 7100 |
+
"grad_norm": 0.0683152824640274,
|
| 7101 |
+
"learning_rate": 5.9447467918700614e-05,
|
| 7102 |
+
"loss": 10.3507,
|
| 7103 |
+
"step": 1010
|
| 7104 |
+
},
|
| 7105 |
+
{
|
| 7106 |
+
"epoch": 0.165317635516311,
|
| 7107 |
+
"grad_norm": 0.0865989699959755,
|
| 7108 |
+
"learning_rate": 5.92657963667694e-05,
|
| 7109 |
+
"loss": 10.3399,
|
| 7110 |
+
"step": 1011
|
| 7111 |
+
},
|
| 7112 |
+
{
|
| 7113 |
+
"epoch": 0.16548115444362685,
|
| 7114 |
+
"grad_norm": 0.060239121317863464,
|
| 7115 |
+
"learning_rate": 5.908428585887883e-05,
|
| 7116 |
+
"loss": 10.3467,
|
| 7117 |
+
"step": 1012
|
| 7118 |
+
},
|
| 7119 |
+
{
|
| 7120 |
+
"epoch": 0.16564467337094269,
|
| 7121 |
+
"grad_norm": 0.07495436817407608,
|
| 7122 |
+
"learning_rate": 5.8902937112636736e-05,
|
| 7123 |
+
"loss": 10.3408,
|
| 7124 |
+
"step": 1013
|
| 7125 |
+
},
|
| 7126 |
+
{
|
| 7127 |
+
"epoch": 0.16580819229825852,
|
| 7128 |
+
"grad_norm": 0.09576012194156647,
|
| 7129 |
+
"learning_rate": 5.8721750845011534e-05,
|
| 7130 |
+
"loss": 10.3501,
|
| 7131 |
+
"step": 1014
|
| 7132 |
+
},
|
| 7133 |
+
{
|
| 7134 |
+
"epoch": 0.16597171122557436,
|
| 7135 |
+
"grad_norm": 0.06893251091241837,
|
| 7136 |
+
"learning_rate": 5.854072777232914e-05,
|
| 7137 |
+
"loss": 10.3433,
|
| 7138 |
+
"step": 1015
|
| 7139 |
+
},
|
| 7140 |
+
{
|
| 7141 |
+
"epoch": 0.1661352301528902,
|
| 7142 |
+
"grad_norm": 0.10778357833623886,
|
| 7143 |
+
"learning_rate": 5.835986861027038e-05,
|
| 7144 |
+
"loss": 10.3496,
|
| 7145 |
+
"step": 1016
|
| 7146 |
+
},
|
| 7147 |
+
{
|
| 7148 |
+
"epoch": 0.16629874908020603,
|
| 7149 |
+
"grad_norm": 0.08453772962093353,
|
| 7150 |
+
"learning_rate": 5.817917407386802e-05,
|
| 7151 |
+
"loss": 10.347,
|
| 7152 |
+
"step": 1017
|
| 7153 |
+
},
|
| 7154 |
+
{
|
| 7155 |
+
"epoch": 0.16646226800752187,
|
| 7156 |
+
"grad_norm": 0.0690169557929039,
|
| 7157 |
+
"learning_rate": 5.799864487750395e-05,
|
| 7158 |
+
"loss": 10.3495,
|
| 7159 |
+
"step": 1018
|
| 7160 |
+
},
|
| 7161 |
+
{
|
| 7162 |
+
"epoch": 0.1666257869348377,
|
| 7163 |
+
"grad_norm": 0.07818324863910675,
|
| 7164 |
+
"learning_rate": 5.7818281734906435e-05,
|
| 7165 |
+
"loss": 10.3496,
|
| 7166 |
+
"step": 1019
|
| 7167 |
+
},
|
| 7168 |
+
{
|
| 7169 |
+
"epoch": 0.16678930586215354,
|
| 7170 |
+
"grad_norm": 0.10107786953449249,
|
| 7171 |
+
"learning_rate": 5.763808535914723e-05,
|
| 7172 |
+
"loss": 10.3466,
|
| 7173 |
+
"step": 1020
|
| 7174 |
+
},
|
| 7175 |
+
{
|
| 7176 |
+
"epoch": 0.16695282478946938,
|
| 7177 |
+
"grad_norm": 0.06788674741983414,
|
| 7178 |
+
"learning_rate": 5.745805646263867e-05,
|
| 7179 |
+
"loss": 10.3417,
|
| 7180 |
+
"step": 1021
|
| 7181 |
+
},
|
| 7182 |
+
{
|
| 7183 |
+
"epoch": 0.16711634371678522,
|
| 7184 |
+
"grad_norm": 0.07959935069084167,
|
| 7185 |
+
"learning_rate": 5.72781957571311e-05,
|
| 7186 |
+
"loss": 10.3524,
|
| 7187 |
+
"step": 1022
|
| 7188 |
+
},
|
| 7189 |
+
{
|
| 7190 |
+
"epoch": 0.16727986264410105,
|
| 7191 |
+
"grad_norm": 0.08590371906757355,
|
| 7192 |
+
"learning_rate": 5.709850395370984e-05,
|
| 7193 |
+
"loss": 10.3463,
|
| 7194 |
+
"step": 1023
|
| 7195 |
+
},
|
| 7196 |
+
{
|
| 7197 |
+
"epoch": 0.1674433815714169,
|
| 7198 |
+
"grad_norm": 0.13714461028575897,
|
| 7199 |
+
"learning_rate": 5.691898176279245e-05,
|
| 7200 |
+
"loss": 10.3373,
|
| 7201 |
+
"step": 1024
|
| 7202 |
+
},
|
| 7203 |
+
{
|
| 7204 |
+
"epoch": 0.16760690049873272,
|
| 7205 |
+
"grad_norm": 0.11352202296257019,
|
| 7206 |
+
"learning_rate": 5.673962989412599e-05,
|
| 7207 |
+
"loss": 10.3455,
|
| 7208 |
+
"step": 1025
|
| 7209 |
+
},
|
| 7210 |
+
{
|
| 7211 |
+
"epoch": 0.16777041942604856,
|
| 7212 |
+
"grad_norm": 0.07253246009349823,
|
| 7213 |
+
"learning_rate": 5.6560449056784014e-05,
|
| 7214 |
+
"loss": 10.3389,
|
| 7215 |
+
"step": 1026
|
| 7216 |
+
},
|
| 7217 |
+
{
|
| 7218 |
+
"epoch": 0.1679339383533644,
|
| 7219 |
+
"grad_norm": 0.1057049110531807,
|
| 7220 |
+
"learning_rate": 5.6381439959164e-05,
|
| 7221 |
+
"loss": 10.34,
|
| 7222 |
+
"step": 1027
|
| 7223 |
+
},
|
| 7224 |
+
{
|
| 7225 |
+
"epoch": 0.16809745728068023,
|
| 7226 |
+
"grad_norm": 0.09184250235557556,
|
| 7227 |
+
"learning_rate": 5.620260330898442e-05,
|
| 7228 |
+
"loss": 10.3349,
|
| 7229 |
+
"step": 1028
|
| 7230 |
+
},
|
| 7231 |
+
{
|
| 7232 |
+
"epoch": 0.16826097620799607,
|
| 7233 |
+
"grad_norm": 0.04531189426779747,
|
| 7234 |
+
"learning_rate": 5.6023939813281975e-05,
|
| 7235 |
+
"loss": 10.3507,
|
| 7236 |
+
"step": 1029
|
| 7237 |
+
},
|
| 7238 |
+
{
|
| 7239 |
+
"epoch": 0.1684244951353119,
|
| 7240 |
+
"grad_norm": 0.06983935087919235,
|
| 7241 |
+
"learning_rate": 5.584545017840885e-05,
|
| 7242 |
+
"loss": 10.3397,
|
| 7243 |
+
"step": 1030
|
| 7244 |
+
},
|
| 7245 |
+
{
|
| 7246 |
+
"epoch": 0.16858801406262774,
|
| 7247 |
+
"grad_norm": 0.051930516958236694,
|
| 7248 |
+
"learning_rate": 5.566713511002969e-05,
|
| 7249 |
+
"loss": 10.3486,
|
| 7250 |
+
"step": 1031
|
| 7251 |
+
},
|
| 7252 |
+
{
|
| 7253 |
+
"epoch": 0.16875153298994358,
|
| 7254 |
+
"grad_norm": 0.06173890084028244,
|
| 7255 |
+
"learning_rate": 5.548899531311915e-05,
|
| 7256 |
+
"loss": 10.3401,
|
| 7257 |
+
"step": 1032
|
| 7258 |
+
},
|
| 7259 |
+
{
|
| 7260 |
+
"epoch": 0.16891505191725942,
|
| 7261 |
+
"grad_norm": 0.052987031638622284,
|
| 7262 |
+
"learning_rate": 5.531103149195891e-05,
|
| 7263 |
+
"loss": 10.3464,
|
| 7264 |
+
"step": 1033
|
| 7265 |
+
},
|
| 7266 |
+
{
|
| 7267 |
+
"epoch": 0.16907857084457525,
|
| 7268 |
+
"grad_norm": 0.10105844587087631,
|
| 7269 |
+
"learning_rate": 5.5133244350134895e-05,
|
| 7270 |
+
"loss": 10.3427,
|
| 7271 |
+
"step": 1034
|
| 7272 |
+
},
|
| 7273 |
+
{
|
| 7274 |
+
"epoch": 0.1692420897718911,
|
| 7275 |
+
"grad_norm": 0.06929253041744232,
|
| 7276 |
+
"learning_rate": 5.4955634590534545e-05,
|
| 7277 |
+
"loss": 10.3486,
|
| 7278 |
+
"step": 1035
|
| 7279 |
+
},
|
| 7280 |
+
{
|
| 7281 |
+
"epoch": 0.16940560869920693,
|
| 7282 |
+
"grad_norm": 0.10541786253452301,
|
| 7283 |
+
"learning_rate": 5.4778202915344015e-05,
|
| 7284 |
+
"loss": 10.3443,
|
| 7285 |
+
"step": 1036
|
| 7286 |
+
},
|
| 7287 |
+
{
|
| 7288 |
+
"epoch": 0.16956912762652276,
|
| 7289 |
+
"grad_norm": 0.07011062651872635,
|
| 7290 |
+
"learning_rate": 5.4600950026045326e-05,
|
| 7291 |
+
"loss": 10.3456,
|
| 7292 |
+
"step": 1037
|
| 7293 |
+
},
|
| 7294 |
+
{
|
| 7295 |
+
"epoch": 0.1697326465538386,
|
| 7296 |
+
"grad_norm": 0.06883953511714935,
|
| 7297 |
+
"learning_rate": 5.442387662341377e-05,
|
| 7298 |
+
"loss": 10.339,
|
| 7299 |
+
"step": 1038
|
| 7300 |
+
},
|
| 7301 |
+
{
|
| 7302 |
+
"epoch": 0.16989616548115444,
|
| 7303 |
+
"grad_norm": 0.1161513701081276,
|
| 7304 |
+
"learning_rate": 5.4246983407514975e-05,
|
| 7305 |
+
"loss": 10.3403,
|
| 7306 |
+
"step": 1039
|
| 7307 |
+
},
|
| 7308 |
+
{
|
| 7309 |
+
"epoch": 0.17005968440847027,
|
| 7310 |
+
"grad_norm": 0.06931749731302261,
|
| 7311 |
+
"learning_rate": 5.407027107770219e-05,
|
| 7312 |
+
"loss": 10.3525,
|
| 7313 |
+
"step": 1040
|
| 7314 |
+
},
|
| 7315 |
+
{
|
| 7316 |
+
"epoch": 0.1702232033357861,
|
| 7317 |
+
"grad_norm": 0.08152681589126587,
|
| 7318 |
+
"learning_rate": 5.389374033261361e-05,
|
| 7319 |
+
"loss": 10.3428,
|
| 7320 |
+
"step": 1041
|
| 7321 |
+
},
|
| 7322 |
+
{
|
| 7323 |
+
"epoch": 0.17038672226310195,
|
| 7324 |
+
"grad_norm": 0.0889812782406807,
|
| 7325 |
+
"learning_rate": 5.371739187016935e-05,
|
| 7326 |
+
"loss": 10.357,
|
| 7327 |
+
"step": 1042
|
| 7328 |
+
},
|
| 7329 |
+
{
|
| 7330 |
+
"epoch": 0.17055024119041778,
|
| 7331 |
+
"grad_norm": 0.056913699954748154,
|
| 7332 |
+
"learning_rate": 5.354122638756903e-05,
|
| 7333 |
+
"loss": 10.3499,
|
| 7334 |
+
"step": 1043
|
| 7335 |
+
},
|
| 7336 |
+
{
|
| 7337 |
+
"epoch": 0.17071376011773362,
|
| 7338 |
+
"grad_norm": 0.13452935218811035,
|
| 7339 |
+
"learning_rate": 5.33652445812888e-05,
|
| 7340 |
+
"loss": 10.3359,
|
| 7341 |
+
"step": 1044
|
| 7342 |
+
},
|
| 7343 |
+
{
|
| 7344 |
+
"epoch": 0.17087727904504946,
|
| 7345 |
+
"grad_norm": 0.065815769135952,
|
| 7346 |
+
"learning_rate": 5.318944714707861e-05,
|
| 7347 |
+
"loss": 10.3395,
|
| 7348 |
+
"step": 1045
|
| 7349 |
+
},
|
| 7350 |
+
{
|
| 7351 |
+
"epoch": 0.1710407979723653,
|
| 7352 |
+
"grad_norm": 0.11757277697324753,
|
| 7353 |
+
"learning_rate": 5.3013834779959556e-05,
|
| 7354 |
+
"loss": 10.3444,
|
| 7355 |
+
"step": 1046
|
| 7356 |
+
},
|
| 7357 |
+
{
|
| 7358 |
+
"epoch": 0.17120431689968113,
|
| 7359 |
+
"grad_norm": 0.055451128631830215,
|
| 7360 |
+
"learning_rate": 5.2838408174221024e-05,
|
| 7361 |
+
"loss": 10.341,
|
| 7362 |
+
"step": 1047
|
| 7363 |
+
},
|
| 7364 |
+
{
|
| 7365 |
+
"epoch": 0.17136783582699697,
|
| 7366 |
+
"grad_norm": 0.06440325826406479,
|
| 7367 |
+
"learning_rate": 5.26631680234179e-05,
|
| 7368 |
+
"loss": 10.3393,
|
| 7369 |
+
"step": 1048
|
| 7370 |
+
},
|
| 7371 |
+
{
|
| 7372 |
+
"epoch": 0.1715313547543128,
|
| 7373 |
+
"grad_norm": 0.11301703751087189,
|
| 7374 |
+
"learning_rate": 5.248811502036806e-05,
|
| 7375 |
+
"loss": 10.329,
|
| 7376 |
+
"step": 1049
|
| 7377 |
+
},
|
| 7378 |
+
{
|
| 7379 |
+
"epoch": 0.17169487368162864,
|
| 7380 |
+
"grad_norm": 0.06889182329177856,
|
| 7381 |
+
"learning_rate": 5.2313249857149414e-05,
|
| 7382 |
+
"loss": 10.3471,
|
| 7383 |
+
"step": 1050
|
| 7384 |
+
},
|
| 7385 |
+
{
|
| 7386 |
+
"epoch": 0.17185839260894448,
|
| 7387 |
+
"grad_norm": 0.08504924178123474,
|
| 7388 |
+
"learning_rate": 5.2138573225097266e-05,
|
| 7389 |
+
"loss": 10.3369,
|
| 7390 |
+
"step": 1051
|
| 7391 |
+
},
|
| 7392 |
+
{
|
| 7393 |
+
"epoch": 0.1720219115362603,
|
| 7394 |
+
"grad_norm": 0.06605865061283112,
|
| 7395 |
+
"learning_rate": 5.196408581480152e-05,
|
| 7396 |
+
"loss": 10.3555,
|
| 7397 |
+
"step": 1052
|
| 7398 |
+
},
|
| 7399 |
+
{
|
| 7400 |
+
"epoch": 0.17218543046357615,
|
| 7401 |
+
"grad_norm": 0.09931064397096634,
|
| 7402 |
+
"learning_rate": 5.1789788316104074e-05,
|
| 7403 |
+
"loss": 10.3295,
|
| 7404 |
+
"step": 1053
|
| 7405 |
+
},
|
| 7406 |
+
{
|
| 7407 |
+
"epoch": 0.172348949390892,
|
| 7408 |
+
"grad_norm": 0.07454589754343033,
|
| 7409 |
+
"learning_rate": 5.161568141809583e-05,
|
| 7410 |
+
"loss": 10.3474,
|
| 7411 |
+
"step": 1054
|
| 7412 |
+
},
|
| 7413 |
+
{
|
| 7414 |
+
"epoch": 0.17251246831820782,
|
| 7415 |
+
"grad_norm": 0.08683858811855316,
|
| 7416 |
+
"learning_rate": 5.144176580911431e-05,
|
| 7417 |
+
"loss": 10.3542,
|
| 7418 |
+
"step": 1055
|
| 7419 |
+
},
|
| 7420 |
+
{
|
| 7421 |
+
"epoch": 0.17267598724552366,
|
| 7422 |
+
"grad_norm": 0.07568492740392685,
|
| 7423 |
+
"learning_rate": 5.126804217674068e-05,
|
| 7424 |
+
"loss": 10.3421,
|
| 7425 |
+
"step": 1056
|
| 7426 |
+
},
|
| 7427 |
+
{
|
| 7428 |
+
"epoch": 0.1728395061728395,
|
| 7429 |
+
"grad_norm": 0.06950981914997101,
|
| 7430 |
+
"learning_rate": 5.109451120779718e-05,
|
| 7431 |
+
"loss": 10.3512,
|
| 7432 |
+
"step": 1057
|
| 7433 |
+
},
|
| 7434 |
+
{
|
| 7435 |
+
"epoch": 0.17300302510015533,
|
| 7436 |
+
"grad_norm": 0.06341303884983063,
|
| 7437 |
+
"learning_rate": 5.092117358834434e-05,
|
| 7438 |
+
"loss": 10.3443,
|
| 7439 |
+
"step": 1058
|
| 7440 |
+
},
|
| 7441 |
+
{
|
| 7442 |
+
"epoch": 0.17316654402747117,
|
| 7443 |
+
"grad_norm": 0.06856871396303177,
|
| 7444 |
+
"learning_rate": 5.074803000367818e-05,
|
| 7445 |
+
"loss": 10.3561,
|
| 7446 |
+
"step": 1059
|
| 7447 |
+
},
|
| 7448 |
+
{
|
| 7449 |
+
"epoch": 0.173330062954787,
|
| 7450 |
+
"grad_norm": 0.08887424319982529,
|
| 7451 |
+
"learning_rate": 5.0575081138327715e-05,
|
| 7452 |
+
"loss": 10.3539,
|
| 7453 |
+
"step": 1060
|
| 7454 |
+
},
|
| 7455 |
+
{
|
| 7456 |
+
"epoch": 0.17349358188210284,
|
| 7457 |
+
"grad_norm": 0.1078650951385498,
|
| 7458 |
+
"learning_rate": 5.040232767605209e-05,
|
| 7459 |
+
"loss": 10.3435,
|
| 7460 |
+
"step": 1061
|
| 7461 |
+
},
|
| 7462 |
+
{
|
| 7463 |
+
"epoch": 0.17365710080941868,
|
| 7464 |
+
"grad_norm": 0.09324576705694199,
|
| 7465 |
+
"learning_rate": 5.022977029983789e-05,
|
| 7466 |
+
"loss": 10.3434,
|
| 7467 |
+
"step": 1062
|
| 7468 |
+
},
|
| 7469 |
+
{
|
| 7470 |
+
"epoch": 0.17382061973673452,
|
| 7471 |
+
"grad_norm": 0.08254747837781906,
|
| 7472 |
+
"learning_rate": 5.005740969189655e-05,
|
| 7473 |
+
"loss": 10.3404,
|
| 7474 |
+
"step": 1063
|
| 7475 |
+
},
|
| 7476 |
+
{
|
| 7477 |
+
"epoch": 0.17398413866405035,
|
| 7478 |
+
"grad_norm": 0.07239022850990295,
|
| 7479 |
+
"learning_rate": 4.9885246533661433e-05,
|
| 7480 |
+
"loss": 10.3408,
|
| 7481 |
+
"step": 1064
|
| 7482 |
+
},
|
| 7483 |
+
{
|
| 7484 |
+
"epoch": 0.1741476575913662,
|
| 7485 |
+
"grad_norm": 0.058068279176950455,
|
| 7486 |
+
"learning_rate": 4.97132815057854e-05,
|
| 7487 |
+
"loss": 10.3401,
|
| 7488 |
+
"step": 1065
|
| 7489 |
+
},
|
| 7490 |
+
{
|
| 7491 |
+
"epoch": 0.17431117651868203,
|
| 7492 |
+
"grad_norm": 0.055977217853069305,
|
| 7493 |
+
"learning_rate": 4.954151528813795e-05,
|
| 7494 |
+
"loss": 10.3477,
|
| 7495 |
+
"step": 1066
|
| 7496 |
+
},
|
| 7497 |
+
{
|
| 7498 |
+
"epoch": 0.17447469544599786,
|
| 7499 |
+
"grad_norm": 0.09501728415489197,
|
| 7500 |
+
"learning_rate": 4.9369948559802605e-05,
|
| 7501 |
+
"loss": 10.3508,
|
| 7502 |
+
"step": 1067
|
| 7503 |
+
},
|
| 7504 |
+
{
|
| 7505 |
+
"epoch": 0.1746382143733137,
|
| 7506 |
+
"grad_norm": 0.07803209125995636,
|
| 7507 |
+
"learning_rate": 4.91985819990742e-05,
|
| 7508 |
+
"loss": 10.3376,
|
| 7509 |
+
"step": 1068
|
| 7510 |
+
},
|
| 7511 |
+
{
|
| 7512 |
+
"epoch": 0.17480173330062954,
|
| 7513 |
+
"grad_norm": 0.08067754656076431,
|
| 7514 |
+
"learning_rate": 4.902741628345612e-05,
|
| 7515 |
+
"loss": 10.3497,
|
| 7516 |
+
"step": 1069
|
| 7517 |
+
},
|
| 7518 |
+
{
|
| 7519 |
+
"epoch": 0.17496525222794537,
|
| 7520 |
+
"grad_norm": 0.07680683583021164,
|
| 7521 |
+
"learning_rate": 4.885645208965779e-05,
|
| 7522 |
+
"loss": 10.3394,
|
| 7523 |
+
"step": 1070
|
| 7524 |
+
},
|
| 7525 |
+
{
|
| 7526 |
+
"epoch": 0.1751287711552612,
|
| 7527 |
+
"grad_norm": 0.12665748596191406,
|
| 7528 |
+
"learning_rate": 4.8685690093591896e-05,
|
| 7529 |
+
"loss": 10.3586,
|
| 7530 |
+
"step": 1071
|
| 7531 |
+
},
|
| 7532 |
+
{
|
| 7533 |
+
"epoch": 0.17529229008257705,
|
| 7534 |
+
"grad_norm": 0.0813855454325676,
|
| 7535 |
+
"learning_rate": 4.85151309703717e-05,
|
| 7536 |
+
"loss": 10.3522,
|
| 7537 |
+
"step": 1072
|
| 7538 |
+
},
|
| 7539 |
+
{
|
| 7540 |
+
"epoch": 0.17545580900989288,
|
| 7541 |
+
"grad_norm": 0.0832417830824852,
|
| 7542 |
+
"learning_rate": 4.8344775394308484e-05,
|
| 7543 |
+
"loss": 10.3594,
|
| 7544 |
+
"step": 1073
|
| 7545 |
+
},
|
| 7546 |
+
{
|
| 7547 |
+
"epoch": 0.17561932793720872,
|
| 7548 |
+
"grad_norm": 0.07607519626617432,
|
| 7549 |
+
"learning_rate": 4.8174624038908645e-05,
|
| 7550 |
+
"loss": 10.3441,
|
| 7551 |
+
"step": 1074
|
| 7552 |
+
},
|
| 7553 |
+
{
|
| 7554 |
+
"epoch": 0.17578284686452456,
|
| 7555 |
+
"grad_norm": 0.04467737302184105,
|
| 7556 |
+
"learning_rate": 4.80046775768713e-05,
|
| 7557 |
+
"loss": 10.3389,
|
| 7558 |
+
"step": 1075
|
| 7559 |
+
},
|
| 7560 |
+
{
|
| 7561 |
+
"epoch": 0.1759463657918404,
|
| 7562 |
+
"grad_norm": 0.07424350082874298,
|
| 7563 |
+
"learning_rate": 4.783493668008551e-05,
|
| 7564 |
+
"loss": 10.349,
|
| 7565 |
+
"step": 1076
|
| 7566 |
+
},
|
| 7567 |
+
{
|
| 7568 |
+
"epoch": 0.17610988471915623,
|
| 7569 |
+
"grad_norm": 0.10817383229732513,
|
| 7570 |
+
"learning_rate": 4.766540201962758e-05,
|
| 7571 |
+
"loss": 10.3473,
|
| 7572 |
+
"step": 1077
|
| 7573 |
+
},
|
| 7574 |
+
{
|
| 7575 |
+
"epoch": 0.17627340364647207,
|
| 7576 |
+
"grad_norm": 0.09051129221916199,
|
| 7577 |
+
"learning_rate": 4.7496074265758494e-05,
|
| 7578 |
+
"loss": 10.3451,
|
| 7579 |
+
"step": 1078
|
| 7580 |
+
},
|
| 7581 |
+
{
|
| 7582 |
+
"epoch": 0.1764369225737879,
|
| 7583 |
+
"grad_norm": 0.08685331046581268,
|
| 7584 |
+
"learning_rate": 4.732695408792125e-05,
|
| 7585 |
+
"loss": 10.346,
|
| 7586 |
+
"step": 1079
|
| 7587 |
+
},
|
| 7588 |
+
{
|
| 7589 |
+
"epoch": 0.17660044150110377,
|
| 7590 |
+
"grad_norm": 0.09308838844299316,
|
| 7591 |
+
"learning_rate": 4.715804215473809e-05,
|
| 7592 |
+
"loss": 10.3536,
|
| 7593 |
+
"step": 1080
|
| 7594 |
+
},
|
| 7595 |
+
{
|
| 7596 |
+
"epoch": 0.1767639604284196,
|
| 7597 |
+
"grad_norm": 0.08783134818077087,
|
| 7598 |
+
"learning_rate": 4.698933913400798e-05,
|
| 7599 |
+
"loss": 10.3449,
|
| 7600 |
+
"step": 1081
|
| 7601 |
+
},
|
| 7602 |
+
{
|
| 7603 |
+
"epoch": 0.17692747935573544,
|
| 7604 |
+
"grad_norm": 0.055539608001708984,
|
| 7605 |
+
"learning_rate": 4.682084569270402e-05,
|
| 7606 |
+
"loss": 10.3374,
|
| 7607 |
+
"step": 1082
|
| 7608 |
+
},
|
| 7609 |
+
{
|
| 7610 |
+
"epoch": 0.17709099828305128,
|
| 7611 |
+
"grad_norm": 0.0705205500125885,
|
| 7612 |
+
"learning_rate": 4.6652562496970667e-05,
|
| 7613 |
+
"loss": 10.3497,
|
| 7614 |
+
"step": 1083
|
| 7615 |
+
},
|
| 7616 |
+
{
|
| 7617 |
+
"epoch": 0.1772545172103671,
|
| 7618 |
+
"grad_norm": 0.05567767843604088,
|
| 7619 |
+
"learning_rate": 4.648449021212118e-05,
|
| 7620 |
+
"loss": 10.3539,
|
| 7621 |
+
"step": 1084
|
| 7622 |
+
},
|
| 7623 |
+
{
|
| 7624 |
+
"epoch": 0.17741803613768295,
|
| 7625 |
+
"grad_norm": 0.05686875432729721,
|
| 7626 |
+
"learning_rate": 4.6316629502635025e-05,
|
| 7627 |
+
"loss": 10.342,
|
| 7628 |
+
"step": 1085
|
| 7629 |
+
},
|
| 7630 |
+
{
|
| 7631 |
+
"epoch": 0.1775815550649988,
|
| 7632 |
+
"grad_norm": 0.08918684720993042,
|
| 7633 |
+
"learning_rate": 4.614898103215507e-05,
|
| 7634 |
+
"loss": 10.3434,
|
| 7635 |
+
"step": 1086
|
| 7636 |
+
},
|
| 7637 |
+
{
|
| 7638 |
+
"epoch": 0.17774507399231462,
|
| 7639 |
+
"grad_norm": 0.061265699565410614,
|
| 7640 |
+
"learning_rate": 4.59815454634852e-05,
|
| 7641 |
+
"loss": 10.3424,
|
| 7642 |
+
"step": 1087
|
| 7643 |
+
},
|
| 7644 |
+
{
|
| 7645 |
+
"epoch": 0.17790859291963046,
|
| 7646 |
+
"grad_norm": 0.06800343096256256,
|
| 7647 |
+
"learning_rate": 4.5814323458587563e-05,
|
| 7648 |
+
"loss": 10.3337,
|
| 7649 |
+
"step": 1088
|
| 7650 |
+
},
|
| 7651 |
+
{
|
| 7652 |
+
"epoch": 0.1780721118469463,
|
| 7653 |
+
"grad_norm": 0.11221577972173691,
|
| 7654 |
+
"learning_rate": 4.564731567857995e-05,
|
| 7655 |
+
"loss": 10.3525,
|
| 7656 |
+
"step": 1089
|
| 7657 |
+
},
|
| 7658 |
+
{
|
| 7659 |
+
"epoch": 0.17823563077426213,
|
| 7660 |
+
"grad_norm": 0.07759296149015427,
|
| 7661 |
+
"learning_rate": 4.548052278373327e-05,
|
| 7662 |
+
"loss": 10.343,
|
| 7663 |
+
"step": 1090
|
| 7664 |
+
},
|
| 7665 |
+
{
|
| 7666 |
+
"epoch": 0.17839914970157797,
|
| 7667 |
+
"grad_norm": 0.130192831158638,
|
| 7668 |
+
"learning_rate": 4.5313945433468755e-05,
|
| 7669 |
+
"loss": 10.3446,
|
| 7670 |
+
"step": 1091
|
| 7671 |
+
},
|
| 7672 |
+
{
|
| 7673 |
+
"epoch": 0.1785626686288938,
|
| 7674 |
+
"grad_norm": 0.07616015523672104,
|
| 7675 |
+
"learning_rate": 4.514758428635557e-05,
|
| 7676 |
+
"loss": 10.3374,
|
| 7677 |
+
"step": 1092
|
| 7678 |
+
},
|
| 7679 |
+
{
|
| 7680 |
+
"epoch": 0.17872618755620964,
|
| 7681 |
+
"grad_norm": 0.08629491925239563,
|
| 7682 |
+
"learning_rate": 4.498144000010811e-05,
|
| 7683 |
+
"loss": 10.3442,
|
| 7684 |
+
"step": 1093
|
| 7685 |
+
},
|
| 7686 |
+
{
|
| 7687 |
+
"epoch": 0.17888970648352548,
|
| 7688 |
+
"grad_norm": 0.06363014131784439,
|
| 7689 |
+
"learning_rate": 4.481551323158338e-05,
|
| 7690 |
+
"loss": 10.3468,
|
| 7691 |
+
"step": 1094
|
| 7692 |
+
},
|
| 7693 |
+
{
|
| 7694 |
+
"epoch": 0.17905322541084132,
|
| 7695 |
+
"grad_norm": 0.15810951590538025,
|
| 7696 |
+
"learning_rate": 4.4649804636778456e-05,
|
| 7697 |
+
"loss": 10.3404,
|
| 7698 |
+
"step": 1095
|
| 7699 |
+
},
|
| 7700 |
+
{
|
| 7701 |
+
"epoch": 0.17921674433815715,
|
| 7702 |
+
"grad_norm": 0.08180614560842514,
|
| 7703 |
+
"learning_rate": 4.448431487082776e-05,
|
| 7704 |
+
"loss": 10.3474,
|
| 7705 |
+
"step": 1096
|
| 7706 |
+
},
|
| 7707 |
+
{
|
| 7708 |
+
"epoch": 0.179380263265473,
|
| 7709 |
+
"grad_norm": 0.04673139005899429,
|
| 7710 |
+
"learning_rate": 4.431904458800066e-05,
|
| 7711 |
+
"loss": 10.3369,
|
| 7712 |
+
"step": 1097
|
| 7713 |
+
},
|
| 7714 |
+
{
|
| 7715 |
+
"epoch": 0.17954378219278883,
|
| 7716 |
+
"grad_norm": 0.06742433458566666,
|
| 7717 |
+
"learning_rate": 4.4153994441698776e-05,
|
| 7718 |
+
"loss": 10.3495,
|
| 7719 |
+
"step": 1098
|
| 7720 |
+
},
|
| 7721 |
+
{
|
| 7722 |
+
"epoch": 0.17970730112010466,
|
| 7723 |
+
"grad_norm": 0.06503660976886749,
|
| 7724 |
+
"learning_rate": 4.3989165084453386e-05,
|
| 7725 |
+
"loss": 10.3467,
|
| 7726 |
+
"step": 1099
|
| 7727 |
+
},
|
| 7728 |
+
{
|
| 7729 |
+
"epoch": 0.1798708200474205,
|
| 7730 |
+
"grad_norm": 0.051366645842790604,
|
| 7731 |
+
"learning_rate": 4.382455716792291e-05,
|
| 7732 |
+
"loss": 10.3386,
|
| 7733 |
+
"step": 1100
|
| 7734 |
+
},
|
| 7735 |
+
{
|
| 7736 |
+
"epoch": 0.18003433897473634,
|
| 7737 |
+
"grad_norm": 0.10040915012359619,
|
| 7738 |
+
"learning_rate": 4.366017134289027e-05,
|
| 7739 |
+
"loss": 10.3482,
|
| 7740 |
+
"step": 1101
|
| 7741 |
+
},
|
| 7742 |
+
{
|
| 7743 |
+
"epoch": 0.18019785790205217,
|
| 7744 |
+
"grad_norm": 0.0676763504743576,
|
| 7745 |
+
"learning_rate": 4.349600825926028e-05,
|
| 7746 |
+
"loss": 10.3443,
|
| 7747 |
+
"step": 1102
|
| 7748 |
+
},
|
| 7749 |
+
{
|
| 7750 |
+
"epoch": 0.180361376829368,
|
| 7751 |
+
"grad_norm": 0.07206389307975769,
|
| 7752 |
+
"learning_rate": 4.333206856605725e-05,
|
| 7753 |
+
"loss": 10.3534,
|
| 7754 |
+
"step": 1103
|
| 7755 |
+
},
|
| 7756 |
+
{
|
| 7757 |
+
"epoch": 0.18052489575668385,
|
| 7758 |
+
"grad_norm": 0.08064892888069153,
|
| 7759 |
+
"learning_rate": 4.316835291142223e-05,
|
| 7760 |
+
"loss": 10.3424,
|
| 7761 |
+
"step": 1104
|
| 7762 |
+
},
|
| 7763 |
+
{
|
| 7764 |
+
"epoch": 0.18068841468399968,
|
| 7765 |
+
"grad_norm": 0.06373365223407745,
|
| 7766 |
+
"learning_rate": 4.300486194261057e-05,
|
| 7767 |
+
"loss": 10.3411,
|
| 7768 |
+
"step": 1105
|
| 7769 |
+
},
|
| 7770 |
+
{
|
| 7771 |
+
"epoch": 0.18085193361131552,
|
| 7772 |
+
"grad_norm": 0.09212401509284973,
|
| 7773 |
+
"learning_rate": 4.2841596305989354e-05,
|
| 7774 |
+
"loss": 10.3521,
|
| 7775 |
+
"step": 1106
|
| 7776 |
+
},
|
| 7777 |
+
{
|
| 7778 |
+
"epoch": 0.18101545253863136,
|
| 7779 |
+
"grad_norm": 0.0825531855225563,
|
| 7780 |
+
"learning_rate": 4.267855664703464e-05,
|
| 7781 |
+
"loss": 10.3436,
|
| 7782 |
+
"step": 1107
|
| 7783 |
+
},
|
| 7784 |
+
{
|
| 7785 |
+
"epoch": 0.1811789714659472,
|
| 7786 |
+
"grad_norm": 0.08634812384843826,
|
| 7787 |
+
"learning_rate": 4.251574361032925e-05,
|
| 7788 |
+
"loss": 10.3498,
|
| 7789 |
+
"step": 1108
|
| 7790 |
+
},
|
| 7791 |
+
{
|
| 7792 |
+
"epoch": 0.18134249039326303,
|
| 7793 |
+
"grad_norm": 0.05569440871477127,
|
| 7794 |
+
"learning_rate": 4.2353157839559984e-05,
|
| 7795 |
+
"loss": 10.3461,
|
| 7796 |
+
"step": 1109
|
| 7797 |
+
},
|
| 7798 |
+
{
|
| 7799 |
+
"epoch": 0.18150600932057886,
|
| 7800 |
+
"grad_norm": 0.0835985317826271,
|
| 7801 |
+
"learning_rate": 4.219079997751515e-05,
|
| 7802 |
+
"loss": 10.3511,
|
| 7803 |
+
"step": 1110
|
| 7804 |
+
},
|
| 7805 |
+
{
|
| 7806 |
+
"epoch": 0.1816695282478947,
|
| 7807 |
+
"grad_norm": 0.047303736209869385,
|
| 7808 |
+
"learning_rate": 4.2028670666082015e-05,
|
| 7809 |
+
"loss": 10.3453,
|
| 7810 |
+
"step": 1111
|
| 7811 |
+
},
|
| 7812 |
+
{
|
| 7813 |
+
"epoch": 0.18183304717521054,
|
| 7814 |
+
"grad_norm": 0.0699780285358429,
|
| 7815 |
+
"learning_rate": 4.1866770546244204e-05,
|
| 7816 |
+
"loss": 10.3424,
|
| 7817 |
+
"step": 1112
|
| 7818 |
+
},
|
| 7819 |
+
{
|
| 7820 |
+
"epoch": 0.18199656610252637,
|
| 7821 |
+
"grad_norm": 0.11049481481313705,
|
| 7822 |
+
"learning_rate": 4.170510025807932e-05,
|
| 7823 |
+
"loss": 10.3401,
|
| 7824 |
+
"step": 1113
|
| 7825 |
+
},
|
| 7826 |
+
{
|
| 7827 |
+
"epoch": 0.1821600850298422,
|
| 7828 |
+
"grad_norm": 0.09954044222831726,
|
| 7829 |
+
"learning_rate": 4.154366044075623e-05,
|
| 7830 |
+
"loss": 10.3389,
|
| 7831 |
+
"step": 1114
|
| 7832 |
+
},
|
| 7833 |
+
{
|
| 7834 |
+
"epoch": 0.18232360395715805,
|
| 7835 |
+
"grad_norm": 0.06754378229379654,
|
| 7836 |
+
"learning_rate": 4.1382451732532665e-05,
|
| 7837 |
+
"loss": 10.3331,
|
| 7838 |
+
"step": 1115
|
| 7839 |
+
},
|
| 7840 |
+
{
|
| 7841 |
+
"epoch": 0.18248712288447388,
|
| 7842 |
+
"grad_norm": 0.06585376709699631,
|
| 7843 |
+
"learning_rate": 4.12214747707527e-05,
|
| 7844 |
+
"loss": 10.3447,
|
| 7845 |
+
"step": 1116
|
| 7846 |
+
},
|
| 7847 |
+
{
|
| 7848 |
+
"epoch": 0.18265064181178972,
|
| 7849 |
+
"grad_norm": 0.076618492603302,
|
| 7850 |
+
"learning_rate": 4.1060730191844154e-05,
|
| 7851 |
+
"loss": 10.3459,
|
| 7852 |
+
"step": 1117
|
| 7853 |
+
},
|
| 7854 |
+
{
|
| 7855 |
+
"epoch": 0.18281416073910556,
|
| 7856 |
+
"grad_norm": 0.08645208925008774,
|
| 7857 |
+
"learning_rate": 4.090021863131608e-05,
|
| 7858 |
+
"loss": 10.3457,
|
| 7859 |
+
"step": 1118
|
| 7860 |
+
},
|
| 7861 |
+
{
|
| 7862 |
+
"epoch": 0.1829776796664214,
|
| 7863 |
+
"grad_norm": 0.059928491711616516,
|
| 7864 |
+
"learning_rate": 4.073994072375631e-05,
|
| 7865 |
+
"loss": 10.3568,
|
| 7866 |
+
"step": 1119
|
| 7867 |
+
},
|
| 7868 |
+
{
|
| 7869 |
+
"epoch": 0.18314119859373723,
|
| 7870 |
+
"grad_norm": 0.07743990421295166,
|
| 7871 |
+
"learning_rate": 4.0579897102828966e-05,
|
| 7872 |
+
"loss": 10.3403,
|
| 7873 |
+
"step": 1120
|
| 7874 |
+
},
|
| 7875 |
+
{
|
| 7876 |
+
"epoch": 0.18330471752105307,
|
| 7877 |
+
"grad_norm": 0.10486611723899841,
|
| 7878 |
+
"learning_rate": 4.042008840127185e-05,
|
| 7879 |
+
"loss": 10.346,
|
| 7880 |
+
"step": 1121
|
| 7881 |
+
},
|
| 7882 |
+
{
|
| 7883 |
+
"epoch": 0.1834682364483689,
|
| 7884 |
+
"grad_norm": 0.06754382699728012,
|
| 7885 |
+
"learning_rate": 4.0260515250894025e-05,
|
| 7886 |
+
"loss": 10.3418,
|
| 7887 |
+
"step": 1122
|
| 7888 |
+
},
|
| 7889 |
+
{
|
| 7890 |
+
"epoch": 0.18363175537568474,
|
| 7891 |
+
"grad_norm": 0.06686645746231079,
|
| 7892 |
+
"learning_rate": 4.010117828257335e-05,
|
| 7893 |
+
"loss": 10.3515,
|
| 7894 |
+
"step": 1123
|
| 7895 |
+
},
|
| 7896 |
+
{
|
| 7897 |
+
"epoch": 0.18379527430300058,
|
| 7898 |
+
"grad_norm": 0.10611660033464432,
|
| 7899 |
+
"learning_rate": 3.9942078126253776e-05,
|
| 7900 |
+
"loss": 10.3385,
|
| 7901 |
+
"step": 1124
|
| 7902 |
+
},
|
| 7903 |
+
{
|
| 7904 |
+
"epoch": 0.18395879323031641,
|
| 7905 |
+
"grad_norm": 0.07262091338634491,
|
| 7906 |
+
"learning_rate": 3.9783215410943174e-05,
|
| 7907 |
+
"loss": 10.3425,
|
| 7908 |
+
"step": 1125
|
| 7909 |
+
},
|
| 7910 |
+
{
|
| 7911 |
+
"epoch": 0.18412231215763225,
|
| 7912 |
+
"grad_norm": 0.1426948755979538,
|
| 7913 |
+
"learning_rate": 3.962459076471059e-05,
|
| 7914 |
+
"loss": 10.331,
|
| 7915 |
+
"step": 1126
|
| 7916 |
+
},
|
| 7917 |
+
{
|
| 7918 |
+
"epoch": 0.1842858310849481,
|
| 7919 |
+
"grad_norm": 0.10427655279636383,
|
| 7920 |
+
"learning_rate": 3.946620481468391e-05,
|
| 7921 |
+
"loss": 10.3488,
|
| 7922 |
+
"step": 1127
|
| 7923 |
+
},
|
| 7924 |
+
{
|
| 7925 |
+
"epoch": 0.18444935001226392,
|
| 7926 |
+
"grad_norm": 0.09149676561355591,
|
| 7927 |
+
"learning_rate": 3.93080581870473e-05,
|
| 7928 |
+
"loss": 10.34,
|
| 7929 |
+
"step": 1128
|
| 7930 |
+
},
|
| 7931 |
+
{
|
| 7932 |
+
"epoch": 0.18461286893957976,
|
| 7933 |
+
"grad_norm": 0.09051238745450974,
|
| 7934 |
+
"learning_rate": 3.9150151507038736e-05,
|
| 7935 |
+
"loss": 10.3439,
|
| 7936 |
+
"step": 1129
|
| 7937 |
+
},
|
| 7938 |
+
{
|
| 7939 |
+
"epoch": 0.1847763878668956,
|
| 7940 |
+
"grad_norm": 0.06384994834661484,
|
| 7941 |
+
"learning_rate": 3.899248539894757e-05,
|
| 7942 |
+
"loss": 10.3479,
|
| 7943 |
+
"step": 1130
|
| 7944 |
+
},
|
| 7945 |
+
{
|
| 7946 |
+
"epoch": 0.18493990679421143,
|
| 7947 |
+
"grad_norm": 0.07397767901420593,
|
| 7948 |
+
"learning_rate": 3.883506048611206e-05,
|
| 7949 |
+
"loss": 10.3283,
|
| 7950 |
+
"step": 1131
|
| 7951 |
+
},
|
| 7952 |
+
{
|
| 7953 |
+
"epoch": 0.18510342572152727,
|
| 7954 |
+
"grad_norm": 0.07018252462148666,
|
| 7955 |
+
"learning_rate": 3.867787739091691e-05,
|
| 7956 |
+
"loss": 10.3504,
|
| 7957 |
+
"step": 1132
|
| 7958 |
+
},
|
| 7959 |
+
{
|
| 7960 |
+
"epoch": 0.1852669446488431,
|
| 7961 |
+
"grad_norm": 0.08793807029724121,
|
| 7962 |
+
"learning_rate": 3.8520936734790745e-05,
|
| 7963 |
+
"loss": 10.3357,
|
| 7964 |
+
"step": 1133
|
| 7965 |
+
},
|
| 7966 |
+
{
|
| 7967 |
+
"epoch": 0.18543046357615894,
|
| 7968 |
+
"grad_norm": 0.08552634716033936,
|
| 7969 |
+
"learning_rate": 3.836423913820367e-05,
|
| 7970 |
+
"loss": 10.3413,
|
| 7971 |
+
"step": 1134
|
| 7972 |
+
},
|
| 7973 |
+
{
|
| 7974 |
+
"epoch": 0.18559398250347478,
|
| 7975 |
+
"grad_norm": 0.06452400237321854,
|
| 7976 |
+
"learning_rate": 3.820778522066494e-05,
|
| 7977 |
+
"loss": 10.3424,
|
| 7978 |
+
"step": 1135
|
| 7979 |
+
},
|
| 7980 |
+
{
|
| 7981 |
+
"epoch": 0.18575750143079062,
|
| 7982 |
+
"grad_norm": 0.11659888178110123,
|
| 7983 |
+
"learning_rate": 3.805157560072033e-05,
|
| 7984 |
+
"loss": 10.3334,
|
| 7985 |
+
"step": 1136
|
| 7986 |
+
},
|
| 7987 |
+
{
|
| 7988 |
+
"epoch": 0.18592102035810645,
|
| 7989 |
+
"grad_norm": 0.08127418905496597,
|
| 7990 |
+
"learning_rate": 3.789561089594985e-05,
|
| 7991 |
+
"loss": 10.3536,
|
| 7992 |
+
"step": 1137
|
| 7993 |
+
},
|
| 7994 |
+
{
|
| 7995 |
+
"epoch": 0.1860845392854223,
|
| 7996 |
+
"grad_norm": 0.07654475420713425,
|
| 7997 |
+
"learning_rate": 3.77398917229652e-05,
|
| 7998 |
+
"loss": 10.33,
|
| 7999 |
+
"step": 1138
|
| 8000 |
+
},
|
| 8001 |
+
{
|
| 8002 |
+
"epoch": 0.18624805821273813,
|
| 8003 |
+
"grad_norm": 0.07903090119361877,
|
| 8004 |
+
"learning_rate": 3.758441869740731e-05,
|
| 8005 |
+
"loss": 10.3509,
|
| 8006 |
+
"step": 1139
|
| 8007 |
+
},
|
| 8008 |
+
{
|
| 8009 |
+
"epoch": 0.18641157714005396,
|
| 8010 |
+
"grad_norm": 0.07720071822404861,
|
| 8011 |
+
"learning_rate": 3.7429192433944014e-05,
|
| 8012 |
+
"loss": 10.342,
|
| 8013 |
+
"step": 1140
|
| 8014 |
+
},
|
| 8015 |
+
{
|
| 8016 |
+
"epoch": 0.1865750960673698,
|
| 8017 |
+
"grad_norm": 0.08252932876348495,
|
| 8018 |
+
"learning_rate": 3.727421354626756e-05,
|
| 8019 |
+
"loss": 10.3469,
|
| 8020 |
+
"step": 1141
|
| 8021 |
+
},
|
| 8022 |
+
{
|
| 8023 |
+
"epoch": 0.18673861499468564,
|
| 8024 |
+
"grad_norm": 0.09409932792186737,
|
| 8025 |
+
"learning_rate": 3.711948264709218e-05,
|
| 8026 |
+
"loss": 10.3465,
|
| 8027 |
+
"step": 1142
|
| 8028 |
+
},
|
| 8029 |
+
{
|
| 8030 |
+
"epoch": 0.18690213392200147,
|
| 8031 |
+
"grad_norm": 0.09753899276256561,
|
| 8032 |
+
"learning_rate": 3.696500034815168e-05,
|
| 8033 |
+
"loss": 10.3357,
|
| 8034 |
+
"step": 1143
|
| 8035 |
+
},
|
| 8036 |
+
{
|
| 8037 |
+
"epoch": 0.1870656528493173,
|
| 8038 |
+
"grad_norm": 0.07511698454618454,
|
| 8039 |
+
"learning_rate": 3.681076726019704e-05,
|
| 8040 |
+
"loss": 10.3388,
|
| 8041 |
+
"step": 1144
|
| 8042 |
+
},
|
| 8043 |
+
{
|
| 8044 |
+
"epoch": 0.18722917177663315,
|
| 8045 |
+
"grad_norm": 0.06391188502311707,
|
| 8046 |
+
"learning_rate": 3.665678399299388e-05,
|
| 8047 |
+
"loss": 10.3507,
|
| 8048 |
+
"step": 1145
|
| 8049 |
+
},
|
| 8050 |
+
{
|
| 8051 |
+
"epoch": 0.18739269070394898,
|
| 8052 |
+
"grad_norm": 0.0674469992518425,
|
| 8053 |
+
"learning_rate": 3.650305115532028e-05,
|
| 8054 |
+
"loss": 10.3467,
|
| 8055 |
+
"step": 1146
|
| 8056 |
+
},
|
| 8057 |
+
{
|
| 8058 |
+
"epoch": 0.18755620963126482,
|
| 8059 |
+
"grad_norm": 0.07931932061910629,
|
| 8060 |
+
"learning_rate": 3.634956935496411e-05,
|
| 8061 |
+
"loss": 10.3452,
|
| 8062 |
+
"step": 1147
|
| 8063 |
+
},
|
| 8064 |
+
{
|
| 8065 |
+
"epoch": 0.18771972855858066,
|
| 8066 |
+
"grad_norm": 0.08464358747005463,
|
| 8067 |
+
"learning_rate": 3.619633919872083e-05,
|
| 8068 |
+
"loss": 10.35,
|
| 8069 |
+
"step": 1148
|
| 8070 |
+
},
|
| 8071 |
+
{
|
| 8072 |
+
"epoch": 0.1878832474858965,
|
| 8073 |
+
"grad_norm": 0.13273005187511444,
|
| 8074 |
+
"learning_rate": 3.604336129239103e-05,
|
| 8075 |
+
"loss": 10.3488,
|
| 8076 |
+
"step": 1149
|
| 8077 |
+
},
|
| 8078 |
+
{
|
| 8079 |
+
"epoch": 0.18804676641321233,
|
| 8080 |
+
"grad_norm": 0.07931933552026749,
|
| 8081 |
+
"learning_rate": 3.589063624077802e-05,
|
| 8082 |
+
"loss": 10.3452,
|
| 8083 |
+
"step": 1150
|
| 8084 |
+
},
|
| 8085 |
+
{
|
| 8086 |
+
"epoch": 0.18821028534052817,
|
| 8087 |
+
"grad_norm": 0.0843479111790657,
|
| 8088 |
+
"learning_rate": 3.573816464768533e-05,
|
| 8089 |
+
"loss": 10.3364,
|
| 8090 |
+
"step": 1151
|
| 8091 |
+
},
|
| 8092 |
+
{
|
| 8093 |
+
"epoch": 0.188373804267844,
|
| 8094 |
+
"grad_norm": 0.06840098649263382,
|
| 8095 |
+
"learning_rate": 3.5585947115914585e-05,
|
| 8096 |
+
"loss": 10.3425,
|
| 8097 |
+
"step": 1152
|
| 8098 |
+
},
|
| 8099 |
+
{
|
| 8100 |
+
"epoch": 0.18853732319515984,
|
| 8101 |
+
"grad_norm": 0.07091546803712845,
|
| 8102 |
+
"learning_rate": 3.543398424726287e-05,
|
| 8103 |
+
"loss": 10.3517,
|
| 8104 |
+
"step": 1153
|
| 8105 |
+
},
|
| 8106 |
+
{
|
| 8107 |
+
"epoch": 0.18870084212247568,
|
| 8108 |
+
"grad_norm": 0.050506554543972015,
|
| 8109 |
+
"learning_rate": 3.5282276642520515e-05,
|
| 8110 |
+
"loss": 10.348,
|
| 8111 |
+
"step": 1154
|
| 8112 |
+
},
|
| 8113 |
+
{
|
| 8114 |
+
"epoch": 0.1888643610497915,
|
| 8115 |
+
"grad_norm": 0.07132521271705627,
|
| 8116 |
+
"learning_rate": 3.513082490146864e-05,
|
| 8117 |
+
"loss": 10.3462,
|
| 8118 |
+
"step": 1155
|
| 8119 |
+
},
|
| 8120 |
+
{
|
| 8121 |
+
"epoch": 0.18902787997710735,
|
| 8122 |
+
"grad_norm": 0.0785883367061615,
|
| 8123 |
+
"learning_rate": 3.497962962287671e-05,
|
| 8124 |
+
"loss": 10.3456,
|
| 8125 |
+
"step": 1156
|
| 8126 |
+
},
|
| 8127 |
+
{
|
| 8128 |
+
"epoch": 0.18919139890442319,
|
| 8129 |
+
"grad_norm": 0.08366641402244568,
|
| 8130 |
+
"learning_rate": 3.482869140450038e-05,
|
| 8131 |
+
"loss": 10.3391,
|
| 8132 |
+
"step": 1157
|
| 8133 |
+
},
|
| 8134 |
+
{
|
| 8135 |
+
"epoch": 0.18935491783173902,
|
| 8136 |
+
"grad_norm": 0.06656763702630997,
|
| 8137 |
+
"learning_rate": 3.467801084307895e-05,
|
| 8138 |
+
"loss": 10.3466,
|
| 8139 |
+
"step": 1158
|
| 8140 |
+
},
|
| 8141 |
+
{
|
| 8142 |
+
"epoch": 0.18951843675905486,
|
| 8143 |
+
"grad_norm": 0.09059333801269531,
|
| 8144 |
+
"learning_rate": 3.452758853433309e-05,
|
| 8145 |
+
"loss": 10.3499,
|
| 8146 |
+
"step": 1159
|
| 8147 |
+
},
|
| 8148 |
+
{
|
| 8149 |
+
"epoch": 0.1896819556863707,
|
| 8150 |
+
"grad_norm": 0.08396489173173904,
|
| 8151 |
+
"learning_rate": 3.4377425072962465e-05,
|
| 8152 |
+
"loss": 10.3515,
|
| 8153 |
+
"step": 1160
|
| 8154 |
+
},
|
| 8155 |
+
{
|
| 8156 |
+
"epoch": 0.18984547461368653,
|
| 8157 |
+
"grad_norm": 0.08763958513736725,
|
| 8158 |
+
"learning_rate": 3.422752105264329e-05,
|
| 8159 |
+
"loss": 10.3503,
|
| 8160 |
+
"step": 1161
|
| 8161 |
+
},
|
| 8162 |
+
{
|
| 8163 |
+
"epoch": 0.19000899354100237,
|
| 8164 |
+
"grad_norm": 0.09006370604038239,
|
| 8165 |
+
"learning_rate": 3.4077877066026176e-05,
|
| 8166 |
+
"loss": 10.3468,
|
| 8167 |
+
"step": 1162
|
| 8168 |
+
},
|
| 8169 |
+
{
|
| 8170 |
+
"epoch": 0.1901725124683182,
|
| 8171 |
+
"grad_norm": 0.07651616632938385,
|
| 8172 |
+
"learning_rate": 3.3928493704733634e-05,
|
| 8173 |
+
"loss": 10.3509,
|
| 8174 |
+
"step": 1163
|
| 8175 |
+
},
|
| 8176 |
+
{
|
| 8177 |
+
"epoch": 0.19033603139563404,
|
| 8178 |
+
"grad_norm": 0.05758075416088104,
|
| 8179 |
+
"learning_rate": 3.377937155935781e-05,
|
| 8180 |
+
"loss": 10.3465,
|
| 8181 |
+
"step": 1164
|
| 8182 |
+
},
|
| 8183 |
+
{
|
| 8184 |
+
"epoch": 0.19049955032294988,
|
| 8185 |
+
"grad_norm": 0.07791003584861755,
|
| 8186 |
+
"learning_rate": 3.363051121945809e-05,
|
| 8187 |
+
"loss": 10.3541,
|
| 8188 |
+
"step": 1165
|
| 8189 |
+
},
|
| 8190 |
+
{
|
| 8191 |
+
"epoch": 0.19066306925026572,
|
| 8192 |
+
"grad_norm": 0.06766680628061295,
|
| 8193 |
+
"learning_rate": 3.348191327355888e-05,
|
| 8194 |
+
"loss": 10.3398,
|
| 8195 |
+
"step": 1166
|
| 8196 |
+
},
|
| 8197 |
+
{
|
| 8198 |
+
"epoch": 0.19082658817758155,
|
| 8199 |
+
"grad_norm": 0.0724874958395958,
|
| 8200 |
+
"learning_rate": 3.3333578309147065e-05,
|
| 8201 |
+
"loss": 10.344,
|
| 8202 |
+
"step": 1167
|
| 8203 |
+
},
|
| 8204 |
+
{
|
| 8205 |
+
"epoch": 0.1909901071048974,
|
| 8206 |
+
"grad_norm": 0.09860213100910187,
|
| 8207 |
+
"learning_rate": 3.318550691266993e-05,
|
| 8208 |
+
"loss": 10.3421,
|
| 8209 |
+
"step": 1168
|
| 8210 |
+
},
|
| 8211 |
+
{
|
| 8212 |
+
"epoch": 0.19115362603221323,
|
| 8213 |
+
"grad_norm": 0.10376410186290741,
|
| 8214 |
+
"learning_rate": 3.3037699669532715e-05,
|
| 8215 |
+
"loss": 10.3561,
|
| 8216 |
+
"step": 1169
|
| 8217 |
+
},
|
| 8218 |
+
{
|
| 8219 |
+
"epoch": 0.19131714495952906,
|
| 8220 |
+
"grad_norm": 0.09623809158802032,
|
| 8221 |
+
"learning_rate": 3.289015716409631e-05,
|
| 8222 |
+
"loss": 10.3472,
|
| 8223 |
+
"step": 1170
|
| 8224 |
+
},
|
| 8225 |
+
{
|
| 8226 |
+
"epoch": 0.1914806638868449,
|
| 8227 |
+
"grad_norm": 0.1350593864917755,
|
| 8228 |
+
"learning_rate": 3.2742879979675004e-05,
|
| 8229 |
+
"loss": 10.3437,
|
| 8230 |
+
"step": 1171
|
| 8231 |
+
},
|
| 8232 |
+
{
|
| 8233 |
+
"epoch": 0.19164418281416074,
|
| 8234 |
+
"grad_norm": 0.06695779412984848,
|
| 8235 |
+
"learning_rate": 3.259586869853401e-05,
|
| 8236 |
+
"loss": 10.3443,
|
| 8237 |
+
"step": 1172
|
| 8238 |
+
},
|
| 8239 |
+
{
|
| 8240 |
+
"epoch": 0.19180770174147657,
|
| 8241 |
+
"grad_norm": 0.059136200696229935,
|
| 8242 |
+
"learning_rate": 3.244912390188739e-05,
|
| 8243 |
+
"loss": 10.3512,
|
| 8244 |
+
"step": 1173
|
| 8245 |
+
},
|
| 8246 |
+
{
|
| 8247 |
+
"epoch": 0.1919712206687924,
|
| 8248 |
+
"grad_norm": 0.08771318942308426,
|
| 8249 |
+
"learning_rate": 3.2302646169895636e-05,
|
| 8250 |
+
"loss": 10.3457,
|
| 8251 |
+
"step": 1174
|
| 8252 |
+
},
|
| 8253 |
+
{
|
| 8254 |
+
"epoch": 0.19213473959610824,
|
| 8255 |
+
"grad_norm": 0.06953369081020355,
|
| 8256 |
+
"learning_rate": 3.2156436081663356e-05,
|
| 8257 |
+
"loss": 10.3526,
|
| 8258 |
+
"step": 1175
|
| 8259 |
+
},
|
| 8260 |
+
{
|
| 8261 |
+
"epoch": 0.19229825852342408,
|
| 8262 |
+
"grad_norm": 0.05626130476593971,
|
| 8263 |
+
"learning_rate": 3.201049421523709e-05,
|
| 8264 |
+
"loss": 10.3424,
|
| 8265 |
+
"step": 1176
|
| 8266 |
+
},
|
| 8267 |
+
{
|
| 8268 |
+
"epoch": 0.19246177745073992,
|
| 8269 |
+
"grad_norm": 0.10361862182617188,
|
| 8270 |
+
"learning_rate": 3.186482114760282e-05,
|
| 8271 |
+
"loss": 10.3488,
|
| 8272 |
+
"step": 1177
|
| 8273 |
+
},
|
| 8274 |
+
{
|
| 8275 |
+
"epoch": 0.19262529637805575,
|
| 8276 |
+
"grad_norm": 0.11523088067770004,
|
| 8277 |
+
"learning_rate": 3.171941745468395e-05,
|
| 8278 |
+
"loss": 10.3433,
|
| 8279 |
+
"step": 1178
|
| 8280 |
+
},
|
| 8281 |
+
{
|
| 8282 |
+
"epoch": 0.1927888153053716,
|
| 8283 |
+
"grad_norm": 0.09867821633815765,
|
| 8284 |
+
"learning_rate": 3.157428371133889e-05,
|
| 8285 |
+
"loss": 10.3337,
|
| 8286 |
+
"step": 1179
|
| 8287 |
+
},
|
| 8288 |
+
{
|
| 8289 |
+
"epoch": 0.19295233423268743,
|
| 8290 |
+
"grad_norm": 0.07772684097290039,
|
| 8291 |
+
"learning_rate": 3.14294204913587e-05,
|
| 8292 |
+
"loss": 10.3495,
|
| 8293 |
+
"step": 1180
|
| 8294 |
+
},
|
| 8295 |
+
{
|
| 8296 |
+
"epoch": 0.19311585316000326,
|
| 8297 |
+
"grad_norm": 0.13682125508785248,
|
| 8298 |
+
"learning_rate": 3.1284828367465e-05,
|
| 8299 |
+
"loss": 10.3351,
|
| 8300 |
+
"step": 1181
|
| 8301 |
+
},
|
| 8302 |
+
{
|
| 8303 |
+
"epoch": 0.1932793720873191,
|
| 8304 |
+
"grad_norm": 0.08868427574634552,
|
| 8305 |
+
"learning_rate": 3.114050791130766e-05,
|
| 8306 |
+
"loss": 10.3464,
|
| 8307 |
+
"step": 1182
|
| 8308 |
+
},
|
| 8309 |
+
{
|
| 8310 |
+
"epoch": 0.19344289101463494,
|
| 8311 |
+
"grad_norm": 0.07886600494384766,
|
| 8312 |
+
"learning_rate": 3.0996459693462345e-05,
|
| 8313 |
+
"loss": 10.3503,
|
| 8314 |
+
"step": 1183
|
| 8315 |
+
},
|
| 8316 |
+
{
|
| 8317 |
+
"epoch": 0.19360640994195077,
|
| 8318 |
+
"grad_norm": 0.07837754487991333,
|
| 8319 |
+
"learning_rate": 3.085268428342858e-05,
|
| 8320 |
+
"loss": 10.343,
|
| 8321 |
+
"step": 1184
|
| 8322 |
+
},
|
| 8323 |
+
{
|
| 8324 |
+
"epoch": 0.1937699288692666,
|
| 8325 |
+
"grad_norm": 0.10123041272163391,
|
| 8326 |
+
"learning_rate": 3.0709182249627255e-05,
|
| 8327 |
+
"loss": 10.3501,
|
| 8328 |
+
"step": 1185
|
| 8329 |
+
},
|
| 8330 |
+
{
|
| 8331 |
+
"epoch": 0.19393344779658245,
|
| 8332 |
+
"grad_norm": 0.057491470128297806,
|
| 8333 |
+
"learning_rate": 3.056595415939849e-05,
|
| 8334 |
+
"loss": 10.342,
|
| 8335 |
+
"step": 1186
|
| 8336 |
+
},
|
| 8337 |
+
{
|
| 8338 |
+
"epoch": 0.19409696672389828,
|
| 8339 |
+
"grad_norm": 0.07750639319419861,
|
| 8340 |
+
"learning_rate": 3.042300057899933e-05,
|
| 8341 |
+
"loss": 10.3478,
|
| 8342 |
+
"step": 1187
|
| 8343 |
+
},
|
| 8344 |
+
{
|
| 8345 |
+
"epoch": 0.19426048565121412,
|
| 8346 |
+
"grad_norm": 0.08213985711336136,
|
| 8347 |
+
"learning_rate": 3.0280322073601585e-05,
|
| 8348 |
+
"loss": 10.3497,
|
| 8349 |
+
"step": 1188
|
| 8350 |
+
},
|
| 8351 |
+
{
|
| 8352 |
+
"epoch": 0.19442400457852996,
|
| 8353 |
+
"grad_norm": 0.05541076883673668,
|
| 8354 |
+
"learning_rate": 3.0137919207289457e-05,
|
| 8355 |
+
"loss": 10.3512,
|
| 8356 |
+
"step": 1189
|
| 8357 |
+
},
|
| 8358 |
+
{
|
| 8359 |
+
"epoch": 0.1945875235058458,
|
| 8360 |
+
"grad_norm": 0.16687621176242828,
|
| 8361 |
+
"learning_rate": 2.9995792543057478e-05,
|
| 8362 |
+
"loss": 10.3534,
|
| 8363 |
+
"step": 1190
|
| 8364 |
+
},
|
| 8365 |
+
{
|
| 8366 |
+
"epoch": 0.19475104243316163,
|
| 8367 |
+
"grad_norm": 0.0721542164683342,
|
| 8368 |
+
"learning_rate": 2.9853942642808185e-05,
|
| 8369 |
+
"loss": 10.3402,
|
| 8370 |
+
"step": 1191
|
| 8371 |
+
},
|
| 8372 |
+
{
|
| 8373 |
+
"epoch": 0.19491456136047747,
|
| 8374 |
+
"grad_norm": 0.0858822613954544,
|
| 8375 |
+
"learning_rate": 2.9712370067349916e-05,
|
| 8376 |
+
"loss": 10.3414,
|
| 8377 |
+
"step": 1192
|
| 8378 |
+
},
|
| 8379 |
+
{
|
| 8380 |
+
"epoch": 0.1950780802877933,
|
| 8381 |
+
"grad_norm": 0.06304842233657837,
|
| 8382 |
+
"learning_rate": 2.9571075376394642e-05,
|
| 8383 |
+
"loss": 10.3403,
|
| 8384 |
+
"step": 1193
|
| 8385 |
+
},
|
| 8386 |
+
{
|
| 8387 |
+
"epoch": 0.19524159921510914,
|
| 8388 |
+
"grad_norm": 0.08378353714942932,
|
| 8389 |
+
"learning_rate": 2.9430059128555585e-05,
|
| 8390 |
+
"loss": 10.3409,
|
| 8391 |
+
"step": 1194
|
| 8392 |
+
},
|
| 8393 |
+
{
|
| 8394 |
+
"epoch": 0.19524159921510914,
|
| 8395 |
+
"eval_loss": 10.343810081481934,
|
| 8396 |
+
"eval_runtime": 7.2963,
|
| 8397 |
+
"eval_samples_per_second": 352.918,
|
| 8398 |
+
"eval_steps_per_second": 176.528,
|
| 8399 |
+
"step": 1194
|
| 8400 |
}
|
| 8401 |
],
|
| 8402 |
"logging_steps": 1,
|
|
|
|
| 8416 |
"attributes": {}
|
| 8417 |
}
|
| 8418 |
},
|
| 8419 |
+
"total_flos": 15497693429760.0,
|
| 8420 |
"train_batch_size": 2,
|
| 8421 |
"trial_name": null,
|
| 8422 |
"trial_params": null
|