Training in progress, step 18500, 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 6832600
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:68dce92e38b5669ed6af11f3c9b53539b4b596623a39b20a0447a548beedcdd3
|
| 3 |
size 6832600
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 3612474
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:4be4e12d392e4c555e25b8c9bb85f734284e812f224aa91cc5771977c0329d01
|
| 3 |
size 3612474
|
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 sha256:17134db01c21afe75f26941e4b935cb7f5490e9c25fded9cd19de49137ddf1a9
|
| 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 sha256:5d04c4fae5a8a22ac933ab7ace077380dd0ddd468b3feecba88e276032e86917
|
| 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": 500,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -126007,6 +126007,3506 @@
|
|
| 126007 |
"learning_rate": 2.631024493020806e-05,
|
| 126008 |
"loss": 1.8519,
|
| 126009 |
"step": 18000
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 126010 |
}
|
| 126011 |
],
|
| 126012 |
"logging_steps": 1,
|
|
@@ -126026,7 +129526,7 @@
|
|
| 126026 |
"attributes": {}
|
| 126027 |
}
|
| 126028 |
},
|
| 126029 |
-
"total_flos": 3.
|
| 126030 |
"train_batch_size": 1,
|
| 126031 |
"trial_name": null,
|
| 126032 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.4872267579668159,
|
| 5 |
"eval_steps": 500,
|
| 6 |
+
"global_step": 18500,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 126007 |
"learning_rate": 2.631024493020806e-05,
|
| 126008 |
"loss": 1.8519,
|
| 126009 |
"step": 18000
|
| 126010 |
+
},
|
| 126011 |
+
{
|
| 126012 |
+
"epoch": 0.47408480379246776,
|
| 126013 |
+
"grad_norm": 3.261470317840576,
|
| 126014 |
+
"learning_rate": 2.6308928101132473e-05,
|
| 126015 |
+
"loss": 1.7502,
|
| 126016 |
+
"step": 18001
|
| 126017 |
+
},
|
| 126018 |
+
{
|
| 126019 |
+
"epoch": 0.47411114037397944,
|
| 126020 |
+
"grad_norm": 3.209693431854248,
|
| 126021 |
+
"learning_rate": 2.6307611272056888e-05,
|
| 126022 |
+
"loss": 0.7119,
|
| 126023 |
+
"step": 18002
|
| 126024 |
+
},
|
| 126025 |
+
{
|
| 126026 |
+
"epoch": 0.4741374769554912,
|
| 126027 |
+
"grad_norm": 1.4962162971496582,
|
| 126028 |
+
"learning_rate": 2.63062944429813e-05,
|
| 126029 |
+
"loss": 1.766,
|
| 126030 |
+
"step": 18003
|
| 126031 |
+
},
|
| 126032 |
+
{
|
| 126033 |
+
"epoch": 0.4741638135370029,
|
| 126034 |
+
"grad_norm": 1.7053147554397583,
|
| 126035 |
+
"learning_rate": 2.630497761390572e-05,
|
| 126036 |
+
"loss": 1.629,
|
| 126037 |
+
"step": 18004
|
| 126038 |
+
},
|
| 126039 |
+
{
|
| 126040 |
+
"epoch": 0.4741901501185146,
|
| 126041 |
+
"grad_norm": 3.461153268814087,
|
| 126042 |
+
"learning_rate": 2.630366078483013e-05,
|
| 126043 |
+
"loss": 1.2987,
|
| 126044 |
+
"step": 18005
|
| 126045 |
+
},
|
| 126046 |
+
{
|
| 126047 |
+
"epoch": 0.47421648670002636,
|
| 126048 |
+
"grad_norm": 5.332024574279785,
|
| 126049 |
+
"learning_rate": 2.6302343955754544e-05,
|
| 126050 |
+
"loss": 1.4677,
|
| 126051 |
+
"step": 18006
|
| 126052 |
+
},
|
| 126053 |
+
{
|
| 126054 |
+
"epoch": 0.47424282328153805,
|
| 126055 |
+
"grad_norm": 3.26020884513855,
|
| 126056 |
+
"learning_rate": 2.6301027126678956e-05,
|
| 126057 |
+
"loss": 2.0604,
|
| 126058 |
+
"step": 18007
|
| 126059 |
+
},
|
| 126060 |
+
{
|
| 126061 |
+
"epoch": 0.4742691598630498,
|
| 126062 |
+
"grad_norm": 1.7316187620162964,
|
| 126063 |
+
"learning_rate": 2.6299710297603375e-05,
|
| 126064 |
+
"loss": 1.8331,
|
| 126065 |
+
"step": 18008
|
| 126066 |
+
},
|
| 126067 |
+
{
|
| 126068 |
+
"epoch": 0.4742954964445615,
|
| 126069 |
+
"grad_norm": 3.970562219619751,
|
| 126070 |
+
"learning_rate": 2.6298393468527787e-05,
|
| 126071 |
+
"loss": 1.4967,
|
| 126072 |
+
"step": 18009
|
| 126073 |
+
},
|
| 126074 |
+
{
|
| 126075 |
+
"epoch": 0.4743218330260732,
|
| 126076 |
+
"grad_norm": 1.3757078647613525,
|
| 126077 |
+
"learning_rate": 2.62970766394522e-05,
|
| 126078 |
+
"loss": 2.077,
|
| 126079 |
+
"step": 18010
|
| 126080 |
+
},
|
| 126081 |
+
{
|
| 126082 |
+
"epoch": 0.4743481696075849,
|
| 126083 |
+
"grad_norm": 2.3905718326568604,
|
| 126084 |
+
"learning_rate": 2.6295759810376615e-05,
|
| 126085 |
+
"loss": 1.4518,
|
| 126086 |
+
"step": 18011
|
| 126087 |
+
},
|
| 126088 |
+
{
|
| 126089 |
+
"epoch": 0.47437450618909666,
|
| 126090 |
+
"grad_norm": 4.833202362060547,
|
| 126091 |
+
"learning_rate": 2.6294442981301027e-05,
|
| 126092 |
+
"loss": 2.1774,
|
| 126093 |
+
"step": 18012
|
| 126094 |
+
},
|
| 126095 |
+
{
|
| 126096 |
+
"epoch": 0.4744008427706084,
|
| 126097 |
+
"grad_norm": 2.995392322540283,
|
| 126098 |
+
"learning_rate": 2.6293126152225446e-05,
|
| 126099 |
+
"loss": 1.8886,
|
| 126100 |
+
"step": 18013
|
| 126101 |
+
},
|
| 126102 |
+
{
|
| 126103 |
+
"epoch": 0.4744271793521201,
|
| 126104 |
+
"grad_norm": 1.6298213005065918,
|
| 126105 |
+
"learning_rate": 2.629180932314986e-05,
|
| 126106 |
+
"loss": 1.6117,
|
| 126107 |
+
"step": 18014
|
| 126108 |
+
},
|
| 126109 |
+
{
|
| 126110 |
+
"epoch": 0.47445351593363183,
|
| 126111 |
+
"grad_norm": 2.83526611328125,
|
| 126112 |
+
"learning_rate": 2.629049249407427e-05,
|
| 126113 |
+
"loss": 0.5354,
|
| 126114 |
+
"step": 18015
|
| 126115 |
+
},
|
| 126116 |
+
{
|
| 126117 |
+
"epoch": 0.4744798525151435,
|
| 126118 |
+
"grad_norm": 2.0443403720855713,
|
| 126119 |
+
"learning_rate": 2.6289175664998683e-05,
|
| 126120 |
+
"loss": 1.8038,
|
| 126121 |
+
"step": 18016
|
| 126122 |
+
},
|
| 126123 |
+
{
|
| 126124 |
+
"epoch": 0.47450618909665526,
|
| 126125 |
+
"grad_norm": 2.0520477294921875,
|
| 126126 |
+
"learning_rate": 2.6287858835923095e-05,
|
| 126127 |
+
"loss": 0.98,
|
| 126128 |
+
"step": 18017
|
| 126129 |
+
},
|
| 126130 |
+
{
|
| 126131 |
+
"epoch": 0.47453252567816695,
|
| 126132 |
+
"grad_norm": 1.6275315284729004,
|
| 126133 |
+
"learning_rate": 2.6286542006847514e-05,
|
| 126134 |
+
"loss": 1.947,
|
| 126135 |
+
"step": 18018
|
| 126136 |
+
},
|
| 126137 |
+
{
|
| 126138 |
+
"epoch": 0.4745588622596787,
|
| 126139 |
+
"grad_norm": 1.6136394739151,
|
| 126140 |
+
"learning_rate": 2.6285225177771926e-05,
|
| 126141 |
+
"loss": 1.8218,
|
| 126142 |
+
"step": 18019
|
| 126143 |
+
},
|
| 126144 |
+
{
|
| 126145 |
+
"epoch": 0.47458519884119044,
|
| 126146 |
+
"grad_norm": 3.588017225265503,
|
| 126147 |
+
"learning_rate": 2.628390834869634e-05,
|
| 126148 |
+
"loss": 1.897,
|
| 126149 |
+
"step": 18020
|
| 126150 |
+
},
|
| 126151 |
+
{
|
| 126152 |
+
"epoch": 0.4746115354227021,
|
| 126153 |
+
"grad_norm": 1.9499776363372803,
|
| 126154 |
+
"learning_rate": 2.6282591519620754e-05,
|
| 126155 |
+
"loss": 2.0482,
|
| 126156 |
+
"step": 18021
|
| 126157 |
+
},
|
| 126158 |
+
{
|
| 126159 |
+
"epoch": 0.47463787200421387,
|
| 126160 |
+
"grad_norm": 1.7800002098083496,
|
| 126161 |
+
"learning_rate": 2.628127469054517e-05,
|
| 126162 |
+
"loss": 1.8747,
|
| 126163 |
+
"step": 18022
|
| 126164 |
+
},
|
| 126165 |
+
{
|
| 126166 |
+
"epoch": 0.47466420858572556,
|
| 126167 |
+
"grad_norm": 4.833249568939209,
|
| 126168 |
+
"learning_rate": 2.6279957861469585e-05,
|
| 126169 |
+
"loss": 1.8213,
|
| 126170 |
+
"step": 18023
|
| 126171 |
+
},
|
| 126172 |
+
{
|
| 126173 |
+
"epoch": 0.4746905451672373,
|
| 126174 |
+
"grad_norm": 2.453986883163452,
|
| 126175 |
+
"learning_rate": 2.6278641032393997e-05,
|
| 126176 |
+
"loss": 1.0012,
|
| 126177 |
+
"step": 18024
|
| 126178 |
+
},
|
| 126179 |
+
{
|
| 126180 |
+
"epoch": 0.474716881748749,
|
| 126181 |
+
"grad_norm": 2.8022005558013916,
|
| 126182 |
+
"learning_rate": 2.627732420331841e-05,
|
| 126183 |
+
"loss": 0.8829,
|
| 126184 |
+
"step": 18025
|
| 126185 |
+
},
|
| 126186 |
+
{
|
| 126187 |
+
"epoch": 0.47474321833026073,
|
| 126188 |
+
"grad_norm": 1.5881845951080322,
|
| 126189 |
+
"learning_rate": 2.6276007374242822e-05,
|
| 126190 |
+
"loss": 1.8127,
|
| 126191 |
+
"step": 18026
|
| 126192 |
+
},
|
| 126193 |
+
{
|
| 126194 |
+
"epoch": 0.4747695549117725,
|
| 126195 |
+
"grad_norm": 2.413527011871338,
|
| 126196 |
+
"learning_rate": 2.627469054516724e-05,
|
| 126197 |
+
"loss": 2.3022,
|
| 126198 |
+
"step": 18027
|
| 126199 |
+
},
|
| 126200 |
+
{
|
| 126201 |
+
"epoch": 0.47479589149328416,
|
| 126202 |
+
"grad_norm": 1.7777903079986572,
|
| 126203 |
+
"learning_rate": 2.6273373716091653e-05,
|
| 126204 |
+
"loss": 0.6391,
|
| 126205 |
+
"step": 18028
|
| 126206 |
+
},
|
| 126207 |
+
{
|
| 126208 |
+
"epoch": 0.4748222280747959,
|
| 126209 |
+
"grad_norm": 2.0796475410461426,
|
| 126210 |
+
"learning_rate": 2.6272056887016065e-05,
|
| 126211 |
+
"loss": 1.2444,
|
| 126212 |
+
"step": 18029
|
| 126213 |
+
},
|
| 126214 |
+
{
|
| 126215 |
+
"epoch": 0.4748485646563076,
|
| 126216 |
+
"grad_norm": 1.6656315326690674,
|
| 126217 |
+
"learning_rate": 2.627074005794048e-05,
|
| 126218 |
+
"loss": 1.5637,
|
| 126219 |
+
"step": 18030
|
| 126220 |
+
},
|
| 126221 |
+
{
|
| 126222 |
+
"epoch": 0.47487490123781934,
|
| 126223 |
+
"grad_norm": 1.4762217998504639,
|
| 126224 |
+
"learning_rate": 2.6269423228864893e-05,
|
| 126225 |
+
"loss": 1.8589,
|
| 126226 |
+
"step": 18031
|
| 126227 |
+
},
|
| 126228 |
+
{
|
| 126229 |
+
"epoch": 0.474901237819331,
|
| 126230 |
+
"grad_norm": 1.888584852218628,
|
| 126231 |
+
"learning_rate": 2.6268106399789312e-05,
|
| 126232 |
+
"loss": 2.2992,
|
| 126233 |
+
"step": 18032
|
| 126234 |
+
},
|
| 126235 |
+
{
|
| 126236 |
+
"epoch": 0.47492757440084277,
|
| 126237 |
+
"grad_norm": 3.5453085899353027,
|
| 126238 |
+
"learning_rate": 2.6266789570713724e-05,
|
| 126239 |
+
"loss": 1.7032,
|
| 126240 |
+
"step": 18033
|
| 126241 |
+
},
|
| 126242 |
+
{
|
| 126243 |
+
"epoch": 0.4749539109823545,
|
| 126244 |
+
"grad_norm": 2.0390994548797607,
|
| 126245 |
+
"learning_rate": 2.6265472741638136e-05,
|
| 126246 |
+
"loss": 1.3441,
|
| 126247 |
+
"step": 18034
|
| 126248 |
+
},
|
| 126249 |
+
{
|
| 126250 |
+
"epoch": 0.4749802475638662,
|
| 126251 |
+
"grad_norm": 1.6482908725738525,
|
| 126252 |
+
"learning_rate": 2.626415591256255e-05,
|
| 126253 |
+
"loss": 0.3402,
|
| 126254 |
+
"step": 18035
|
| 126255 |
+
},
|
| 126256 |
+
{
|
| 126257 |
+
"epoch": 0.47500658414537794,
|
| 126258 |
+
"grad_norm": 3.6160387992858887,
|
| 126259 |
+
"learning_rate": 2.626283908348696e-05,
|
| 126260 |
+
"loss": 0.8216,
|
| 126261 |
+
"step": 18036
|
| 126262 |
+
},
|
| 126263 |
+
{
|
| 126264 |
+
"epoch": 0.47503292072688963,
|
| 126265 |
+
"grad_norm": 1.7986429929733276,
|
| 126266 |
+
"learning_rate": 2.626152225441138e-05,
|
| 126267 |
+
"loss": 1.7053,
|
| 126268 |
+
"step": 18037
|
| 126269 |
+
},
|
| 126270 |
+
{
|
| 126271 |
+
"epoch": 0.4750592573084014,
|
| 126272 |
+
"grad_norm": 1.9232220649719238,
|
| 126273 |
+
"learning_rate": 2.6260205425335792e-05,
|
| 126274 |
+
"loss": 2.0929,
|
| 126275 |
+
"step": 18038
|
| 126276 |
+
},
|
| 126277 |
+
{
|
| 126278 |
+
"epoch": 0.4750855938899131,
|
| 126279 |
+
"grad_norm": 2.988654851913452,
|
| 126280 |
+
"learning_rate": 2.6258888596260208e-05,
|
| 126281 |
+
"loss": 1.2714,
|
| 126282 |
+
"step": 18039
|
| 126283 |
+
},
|
| 126284 |
+
{
|
| 126285 |
+
"epoch": 0.4751119304714248,
|
| 126286 |
+
"grad_norm": 1.9279531240463257,
|
| 126287 |
+
"learning_rate": 2.625757176718462e-05,
|
| 126288 |
+
"loss": 1.3988,
|
| 126289 |
+
"step": 18040
|
| 126290 |
+
},
|
| 126291 |
+
{
|
| 126292 |
+
"epoch": 0.47513826705293655,
|
| 126293 |
+
"grad_norm": 3.3459930419921875,
|
| 126294 |
+
"learning_rate": 2.625625493810904e-05,
|
| 126295 |
+
"loss": 1.2646,
|
| 126296 |
+
"step": 18041
|
| 126297 |
+
},
|
| 126298 |
+
{
|
| 126299 |
+
"epoch": 0.47516460363444823,
|
| 126300 |
+
"grad_norm": 4.728421688079834,
|
| 126301 |
+
"learning_rate": 2.625493810903345e-05,
|
| 126302 |
+
"loss": 0.8438,
|
| 126303 |
+
"step": 18042
|
| 126304 |
+
},
|
| 126305 |
+
{
|
| 126306 |
+
"epoch": 0.47519094021596,
|
| 126307 |
+
"grad_norm": 2.0103824138641357,
|
| 126308 |
+
"learning_rate": 2.6253621279957863e-05,
|
| 126309 |
+
"loss": 1.8723,
|
| 126310 |
+
"step": 18043
|
| 126311 |
+
},
|
| 126312 |
+
{
|
| 126313 |
+
"epoch": 0.47521727679747167,
|
| 126314 |
+
"grad_norm": 2.583697557449341,
|
| 126315 |
+
"learning_rate": 2.6252304450882275e-05,
|
| 126316 |
+
"loss": 2.0797,
|
| 126317 |
+
"step": 18044
|
| 126318 |
+
},
|
| 126319 |
+
{
|
| 126320 |
+
"epoch": 0.4752436133789834,
|
| 126321 |
+
"grad_norm": 6.7954559326171875,
|
| 126322 |
+
"learning_rate": 2.6250987621806688e-05,
|
| 126323 |
+
"loss": 2.6605,
|
| 126324 |
+
"step": 18045
|
| 126325 |
+
},
|
| 126326 |
+
{
|
| 126327 |
+
"epoch": 0.47526994996049515,
|
| 126328 |
+
"grad_norm": 3.5881361961364746,
|
| 126329 |
+
"learning_rate": 2.6249670792731107e-05,
|
| 126330 |
+
"loss": 2.1259,
|
| 126331 |
+
"step": 18046
|
| 126332 |
+
},
|
| 126333 |
+
{
|
| 126334 |
+
"epoch": 0.47529628654200684,
|
| 126335 |
+
"grad_norm": 2.464256525039673,
|
| 126336 |
+
"learning_rate": 2.624835396365552e-05,
|
| 126337 |
+
"loss": 0.9051,
|
| 126338 |
+
"step": 18047
|
| 126339 |
+
},
|
| 126340 |
+
{
|
| 126341 |
+
"epoch": 0.4753226231235186,
|
| 126342 |
+
"grad_norm": 7.696624755859375,
|
| 126343 |
+
"learning_rate": 2.624703713457993e-05,
|
| 126344 |
+
"loss": 2.0595,
|
| 126345 |
+
"step": 18048
|
| 126346 |
+
},
|
| 126347 |
+
{
|
| 126348 |
+
"epoch": 0.47534895970503027,
|
| 126349 |
+
"grad_norm": 5.125574588775635,
|
| 126350 |
+
"learning_rate": 2.6245720305504347e-05,
|
| 126351 |
+
"loss": 0.432,
|
| 126352 |
+
"step": 18049
|
| 126353 |
+
},
|
| 126354 |
+
{
|
| 126355 |
+
"epoch": 0.475375296286542,
|
| 126356 |
+
"grad_norm": 2.2241811752319336,
|
| 126357 |
+
"learning_rate": 2.624440347642876e-05,
|
| 126358 |
+
"loss": 1.2333,
|
| 126359 |
+
"step": 18050
|
| 126360 |
+
},
|
| 126361 |
+
{
|
| 126362 |
+
"epoch": 0.4754016328680537,
|
| 126363 |
+
"grad_norm": 2.0132031440734863,
|
| 126364 |
+
"learning_rate": 2.6243086647353178e-05,
|
| 126365 |
+
"loss": 1.7055,
|
| 126366 |
+
"step": 18051
|
| 126367 |
+
},
|
| 126368 |
+
{
|
| 126369 |
+
"epoch": 0.47542796944956545,
|
| 126370 |
+
"grad_norm": 1.9278985261917114,
|
| 126371 |
+
"learning_rate": 2.624176981827759e-05,
|
| 126372 |
+
"loss": 1.8635,
|
| 126373 |
+
"step": 18052
|
| 126374 |
+
},
|
| 126375 |
+
{
|
| 126376 |
+
"epoch": 0.4754543060310772,
|
| 126377 |
+
"grad_norm": 2.030756711959839,
|
| 126378 |
+
"learning_rate": 2.6240452989202002e-05,
|
| 126379 |
+
"loss": 0.4427,
|
| 126380 |
+
"step": 18053
|
| 126381 |
+
},
|
| 126382 |
+
{
|
| 126383 |
+
"epoch": 0.4754806426125889,
|
| 126384 |
+
"grad_norm": 1.6091283559799194,
|
| 126385 |
+
"learning_rate": 2.6239136160126415e-05,
|
| 126386 |
+
"loss": 1.5705,
|
| 126387 |
+
"step": 18054
|
| 126388 |
+
},
|
| 126389 |
+
{
|
| 126390 |
+
"epoch": 0.4755069791941006,
|
| 126391 |
+
"grad_norm": 2.5038156509399414,
|
| 126392 |
+
"learning_rate": 2.6237819331050834e-05,
|
| 126393 |
+
"loss": 0.2297,
|
| 126394 |
+
"step": 18055
|
| 126395 |
+
},
|
| 126396 |
+
{
|
| 126397 |
+
"epoch": 0.4755333157756123,
|
| 126398 |
+
"grad_norm": 2.3296658992767334,
|
| 126399 |
+
"learning_rate": 2.6236502501975246e-05,
|
| 126400 |
+
"loss": 1.9659,
|
| 126401 |
+
"step": 18056
|
| 126402 |
+
},
|
| 126403 |
+
{
|
| 126404 |
+
"epoch": 0.47555965235712405,
|
| 126405 |
+
"grad_norm": 2.38259220123291,
|
| 126406 |
+
"learning_rate": 2.6235185672899658e-05,
|
| 126407 |
+
"loss": 1.9652,
|
| 126408 |
+
"step": 18057
|
| 126409 |
+
},
|
| 126410 |
+
{
|
| 126411 |
+
"epoch": 0.47558598893863574,
|
| 126412 |
+
"grad_norm": 1.7944610118865967,
|
| 126413 |
+
"learning_rate": 2.6233868843824074e-05,
|
| 126414 |
+
"loss": 0.7851,
|
| 126415 |
+
"step": 18058
|
| 126416 |
+
},
|
| 126417 |
+
{
|
| 126418 |
+
"epoch": 0.4756123255201475,
|
| 126419 |
+
"grad_norm": 1.4154953956604004,
|
| 126420 |
+
"learning_rate": 2.6232552014748486e-05,
|
| 126421 |
+
"loss": 1.5766,
|
| 126422 |
+
"step": 18059
|
| 126423 |
+
},
|
| 126424 |
+
{
|
| 126425 |
+
"epoch": 0.4756386621016592,
|
| 126426 |
+
"grad_norm": 1.9560688734054565,
|
| 126427 |
+
"learning_rate": 2.6231235185672905e-05,
|
| 126428 |
+
"loss": 1.9532,
|
| 126429 |
+
"step": 18060
|
| 126430 |
+
},
|
| 126431 |
+
{
|
| 126432 |
+
"epoch": 0.4756649986831709,
|
| 126433 |
+
"grad_norm": 1.9043681621551514,
|
| 126434 |
+
"learning_rate": 2.6229918356597317e-05,
|
| 126435 |
+
"loss": 1.5029,
|
| 126436 |
+
"step": 18061
|
| 126437 |
+
},
|
| 126438 |
+
{
|
| 126439 |
+
"epoch": 0.47569133526468266,
|
| 126440 |
+
"grad_norm": 2.039125680923462,
|
| 126441 |
+
"learning_rate": 2.622860152752173e-05,
|
| 126442 |
+
"loss": 1.3208,
|
| 126443 |
+
"step": 18062
|
| 126444 |
+
},
|
| 126445 |
+
{
|
| 126446 |
+
"epoch": 0.47571767184619435,
|
| 126447 |
+
"grad_norm": 1.4919660091400146,
|
| 126448 |
+
"learning_rate": 2.622728469844614e-05,
|
| 126449 |
+
"loss": 1.4258,
|
| 126450 |
+
"step": 18063
|
| 126451 |
+
},
|
| 126452 |
+
{
|
| 126453 |
+
"epoch": 0.4757440084277061,
|
| 126454 |
+
"grad_norm": 2.184338331222534,
|
| 126455 |
+
"learning_rate": 2.6225967869370554e-05,
|
| 126456 |
+
"loss": 1.5314,
|
| 126457 |
+
"step": 18064
|
| 126458 |
+
},
|
| 126459 |
+
{
|
| 126460 |
+
"epoch": 0.4757703450092178,
|
| 126461 |
+
"grad_norm": 3.4698526859283447,
|
| 126462 |
+
"learning_rate": 2.6224651040294973e-05,
|
| 126463 |
+
"loss": 1.8335,
|
| 126464 |
+
"step": 18065
|
| 126465 |
+
},
|
| 126466 |
+
{
|
| 126467 |
+
"epoch": 0.4757966815907295,
|
| 126468 |
+
"grad_norm": 2.0690839290618896,
|
| 126469 |
+
"learning_rate": 2.6223334211219385e-05,
|
| 126470 |
+
"loss": 1.3201,
|
| 126471 |
+
"step": 18066
|
| 126472 |
+
},
|
| 126473 |
+
{
|
| 126474 |
+
"epoch": 0.47582301817224126,
|
| 126475 |
+
"grad_norm": 1.9289405345916748,
|
| 126476 |
+
"learning_rate": 2.62220173821438e-05,
|
| 126477 |
+
"loss": 1.8343,
|
| 126478 |
+
"step": 18067
|
| 126479 |
+
},
|
| 126480 |
+
{
|
| 126481 |
+
"epoch": 0.47584935475375295,
|
| 126482 |
+
"grad_norm": 1.8096168041229248,
|
| 126483 |
+
"learning_rate": 2.6220700553068213e-05,
|
| 126484 |
+
"loss": 0.6406,
|
| 126485 |
+
"step": 18068
|
| 126486 |
+
},
|
| 126487 |
+
{
|
| 126488 |
+
"epoch": 0.4758756913352647,
|
| 126489 |
+
"grad_norm": 1.51333749294281,
|
| 126490 |
+
"learning_rate": 2.6219383723992625e-05,
|
| 126491 |
+
"loss": 2.2432,
|
| 126492 |
+
"step": 18069
|
| 126493 |
+
},
|
| 126494 |
+
{
|
| 126495 |
+
"epoch": 0.4759020279167764,
|
| 126496 |
+
"grad_norm": 1.675201416015625,
|
| 126497 |
+
"learning_rate": 2.6218066894917044e-05,
|
| 126498 |
+
"loss": 1.9982,
|
| 126499 |
+
"step": 18070
|
| 126500 |
+
},
|
| 126501 |
+
{
|
| 126502 |
+
"epoch": 0.4759283644982881,
|
| 126503 |
+
"grad_norm": 3.723294734954834,
|
| 126504 |
+
"learning_rate": 2.6216750065841456e-05,
|
| 126505 |
+
"loss": 1.6848,
|
| 126506 |
+
"step": 18071
|
| 126507 |
+
},
|
| 126508 |
+
{
|
| 126509 |
+
"epoch": 0.47595470107979987,
|
| 126510 |
+
"grad_norm": 1.902472734451294,
|
| 126511 |
+
"learning_rate": 2.6215433236765868e-05,
|
| 126512 |
+
"loss": 1.6854,
|
| 126513 |
+
"step": 18072
|
| 126514 |
+
},
|
| 126515 |
+
{
|
| 126516 |
+
"epoch": 0.47598103766131156,
|
| 126517 |
+
"grad_norm": 2.0708296298980713,
|
| 126518 |
+
"learning_rate": 2.621411640769028e-05,
|
| 126519 |
+
"loss": 1.6196,
|
| 126520 |
+
"step": 18073
|
| 126521 |
+
},
|
| 126522 |
+
{
|
| 126523 |
+
"epoch": 0.4760073742428233,
|
| 126524 |
+
"grad_norm": 2.0596871376037598,
|
| 126525 |
+
"learning_rate": 2.62127995786147e-05,
|
| 126526 |
+
"loss": 0.9181,
|
| 126527 |
+
"step": 18074
|
| 126528 |
+
},
|
| 126529 |
+
{
|
| 126530 |
+
"epoch": 0.476033710824335,
|
| 126531 |
+
"grad_norm": 1.582932949066162,
|
| 126532 |
+
"learning_rate": 2.621148274953911e-05,
|
| 126533 |
+
"loss": 1.9876,
|
| 126534 |
+
"step": 18075
|
| 126535 |
+
},
|
| 126536 |
+
{
|
| 126537 |
+
"epoch": 0.47606004740584673,
|
| 126538 |
+
"grad_norm": 2.512171983718872,
|
| 126539 |
+
"learning_rate": 2.6210165920463524e-05,
|
| 126540 |
+
"loss": 2.3788,
|
| 126541 |
+
"step": 18076
|
| 126542 |
+
},
|
| 126543 |
+
{
|
| 126544 |
+
"epoch": 0.4760863839873584,
|
| 126545 |
+
"grad_norm": 2.1394431591033936,
|
| 126546 |
+
"learning_rate": 2.620884909138794e-05,
|
| 126547 |
+
"loss": 0.6411,
|
| 126548 |
+
"step": 18077
|
| 126549 |
+
},
|
| 126550 |
+
{
|
| 126551 |
+
"epoch": 0.47611272056887016,
|
| 126552 |
+
"grad_norm": 1.6871570348739624,
|
| 126553 |
+
"learning_rate": 2.620753226231235e-05,
|
| 126554 |
+
"loss": 1.7547,
|
| 126555 |
+
"step": 18078
|
| 126556 |
+
},
|
| 126557 |
+
{
|
| 126558 |
+
"epoch": 0.4761390571503819,
|
| 126559 |
+
"grad_norm": 3.2810633182525635,
|
| 126560 |
+
"learning_rate": 2.620621543323677e-05,
|
| 126561 |
+
"loss": 1.3224,
|
| 126562 |
+
"step": 18079
|
| 126563 |
+
},
|
| 126564 |
+
{
|
| 126565 |
+
"epoch": 0.4761653937318936,
|
| 126566 |
+
"grad_norm": 2.3216912746429443,
|
| 126567 |
+
"learning_rate": 2.6204898604161183e-05,
|
| 126568 |
+
"loss": 0.2762,
|
| 126569 |
+
"step": 18080
|
| 126570 |
+
},
|
| 126571 |
+
{
|
| 126572 |
+
"epoch": 0.47619173031340534,
|
| 126573 |
+
"grad_norm": 2.0798046588897705,
|
| 126574 |
+
"learning_rate": 2.6203581775085595e-05,
|
| 126575 |
+
"loss": 1.8598,
|
| 126576 |
+
"step": 18081
|
| 126577 |
+
},
|
| 126578 |
+
{
|
| 126579 |
+
"epoch": 0.476218066894917,
|
| 126580 |
+
"grad_norm": 2.1319127082824707,
|
| 126581 |
+
"learning_rate": 2.6202264946010007e-05,
|
| 126582 |
+
"loss": 1.3377,
|
| 126583 |
+
"step": 18082
|
| 126584 |
+
},
|
| 126585 |
+
{
|
| 126586 |
+
"epoch": 0.47624440347642877,
|
| 126587 |
+
"grad_norm": 2.078587055206299,
|
| 126588 |
+
"learning_rate": 2.620094811693442e-05,
|
| 126589 |
+
"loss": 1.2751,
|
| 126590 |
+
"step": 18083
|
| 126591 |
+
},
|
| 126592 |
+
{
|
| 126593 |
+
"epoch": 0.47627074005794046,
|
| 126594 |
+
"grad_norm": 2.120863914489746,
|
| 126595 |
+
"learning_rate": 2.619963128785884e-05,
|
| 126596 |
+
"loss": 0.5739,
|
| 126597 |
+
"step": 18084
|
| 126598 |
+
},
|
| 126599 |
+
{
|
| 126600 |
+
"epoch": 0.4762970766394522,
|
| 126601 |
+
"grad_norm": 1.5160518884658813,
|
| 126602 |
+
"learning_rate": 2.619831445878325e-05,
|
| 126603 |
+
"loss": 1.5942,
|
| 126604 |
+
"step": 18085
|
| 126605 |
+
},
|
| 126606 |
+
{
|
| 126607 |
+
"epoch": 0.47632341322096394,
|
| 126608 |
+
"grad_norm": 2.545600175857544,
|
| 126609 |
+
"learning_rate": 2.6196997629707666e-05,
|
| 126610 |
+
"loss": 1.5817,
|
| 126611 |
+
"step": 18086
|
| 126612 |
+
},
|
| 126613 |
+
{
|
| 126614 |
+
"epoch": 0.47634974980247563,
|
| 126615 |
+
"grad_norm": 2.5309865474700928,
|
| 126616 |
+
"learning_rate": 2.619568080063208e-05,
|
| 126617 |
+
"loss": 1.8868,
|
| 126618 |
+
"step": 18087
|
| 126619 |
+
},
|
| 126620 |
+
{
|
| 126621 |
+
"epoch": 0.4763760863839874,
|
| 126622 |
+
"grad_norm": 3.169647216796875,
|
| 126623 |
+
"learning_rate": 2.6194363971556497e-05,
|
| 126624 |
+
"loss": 1.0168,
|
| 126625 |
+
"step": 18088
|
| 126626 |
+
},
|
| 126627 |
+
{
|
| 126628 |
+
"epoch": 0.47640242296549906,
|
| 126629 |
+
"grad_norm": 1.8885867595672607,
|
| 126630 |
+
"learning_rate": 2.619304714248091e-05,
|
| 126631 |
+
"loss": 1.8344,
|
| 126632 |
+
"step": 18089
|
| 126633 |
+
},
|
| 126634 |
+
{
|
| 126635 |
+
"epoch": 0.4764287595470108,
|
| 126636 |
+
"grad_norm": 1.8588078022003174,
|
| 126637 |
+
"learning_rate": 2.6191730313405322e-05,
|
| 126638 |
+
"loss": 1.9987,
|
| 126639 |
+
"step": 18090
|
| 126640 |
+
},
|
| 126641 |
+
{
|
| 126642 |
+
"epoch": 0.4764550961285225,
|
| 126643 |
+
"grad_norm": 2.69921875,
|
| 126644 |
+
"learning_rate": 2.6190413484329734e-05,
|
| 126645 |
+
"loss": 1.5042,
|
| 126646 |
+
"step": 18091
|
| 126647 |
+
},
|
| 126648 |
+
{
|
| 126649 |
+
"epoch": 0.47648143271003424,
|
| 126650 |
+
"grad_norm": 1.9179502725601196,
|
| 126651 |
+
"learning_rate": 2.6189096655254146e-05,
|
| 126652 |
+
"loss": 2.1202,
|
| 126653 |
+
"step": 18092
|
| 126654 |
+
},
|
| 126655 |
+
{
|
| 126656 |
+
"epoch": 0.476507769291546,
|
| 126657 |
+
"grad_norm": 1.971161127090454,
|
| 126658 |
+
"learning_rate": 2.6187779826178565e-05,
|
| 126659 |
+
"loss": 0.6615,
|
| 126660 |
+
"step": 18093
|
| 126661 |
+
},
|
| 126662 |
+
{
|
| 126663 |
+
"epoch": 0.47653410587305767,
|
| 126664 |
+
"grad_norm": 4.208078384399414,
|
| 126665 |
+
"learning_rate": 2.6186462997102977e-05,
|
| 126666 |
+
"loss": 1.2328,
|
| 126667 |
+
"step": 18094
|
| 126668 |
+
},
|
| 126669 |
+
{
|
| 126670 |
+
"epoch": 0.4765604424545694,
|
| 126671 |
+
"grad_norm": 2.099757432937622,
|
| 126672 |
+
"learning_rate": 2.618514616802739e-05,
|
| 126673 |
+
"loss": 2.2761,
|
| 126674 |
+
"step": 18095
|
| 126675 |
+
},
|
| 126676 |
+
{
|
| 126677 |
+
"epoch": 0.4765867790360811,
|
| 126678 |
+
"grad_norm": 1.5804299116134644,
|
| 126679 |
+
"learning_rate": 2.6183829338951805e-05,
|
| 126680 |
+
"loss": 1.5915,
|
| 126681 |
+
"step": 18096
|
| 126682 |
+
},
|
| 126683 |
+
{
|
| 126684 |
+
"epoch": 0.47661311561759284,
|
| 126685 |
+
"grad_norm": 2.426229953765869,
|
| 126686 |
+
"learning_rate": 2.6182512509876217e-05,
|
| 126687 |
+
"loss": 0.1391,
|
| 126688 |
+
"step": 18097
|
| 126689 |
+
},
|
| 126690 |
+
{
|
| 126691 |
+
"epoch": 0.47663945219910453,
|
| 126692 |
+
"grad_norm": 3.338651418685913,
|
| 126693 |
+
"learning_rate": 2.6181195680800636e-05,
|
| 126694 |
+
"loss": 0.873,
|
| 126695 |
+
"step": 18098
|
| 126696 |
+
},
|
| 126697 |
+
{
|
| 126698 |
+
"epoch": 0.4766657887806163,
|
| 126699 |
+
"grad_norm": 3.4374072551727295,
|
| 126700 |
+
"learning_rate": 2.617987885172505e-05,
|
| 126701 |
+
"loss": 1.6158,
|
| 126702 |
+
"step": 18099
|
| 126703 |
+
},
|
| 126704 |
+
{
|
| 126705 |
+
"epoch": 0.476692125362128,
|
| 126706 |
+
"grad_norm": 1.865473985671997,
|
| 126707 |
+
"learning_rate": 2.617856202264946e-05,
|
| 126708 |
+
"loss": 2.1476,
|
| 126709 |
+
"step": 18100
|
| 126710 |
+
},
|
| 126711 |
+
{
|
| 126712 |
+
"epoch": 0.4767184619436397,
|
| 126713 |
+
"grad_norm": 5.584752559661865,
|
| 126714 |
+
"learning_rate": 2.6177245193573873e-05,
|
| 126715 |
+
"loss": 0.7848,
|
| 126716 |
+
"step": 18101
|
| 126717 |
+
},
|
| 126718 |
+
{
|
| 126719 |
+
"epoch": 0.47674479852515145,
|
| 126720 |
+
"grad_norm": 2.1740102767944336,
|
| 126721 |
+
"learning_rate": 2.6175928364498285e-05,
|
| 126722 |
+
"loss": 1.6618,
|
| 126723 |
+
"step": 18102
|
| 126724 |
+
},
|
| 126725 |
+
{
|
| 126726 |
+
"epoch": 0.47677113510666314,
|
| 126727 |
+
"grad_norm": 1.813247561454773,
|
| 126728 |
+
"learning_rate": 2.6174611535422704e-05,
|
| 126729 |
+
"loss": 2.3947,
|
| 126730 |
+
"step": 18103
|
| 126731 |
+
},
|
| 126732 |
+
{
|
| 126733 |
+
"epoch": 0.4767974716881749,
|
| 126734 |
+
"grad_norm": 2.389674663543701,
|
| 126735 |
+
"learning_rate": 2.6173294706347116e-05,
|
| 126736 |
+
"loss": 0.7996,
|
| 126737 |
+
"step": 18104
|
| 126738 |
+
},
|
| 126739 |
+
{
|
| 126740 |
+
"epoch": 0.4768238082696866,
|
| 126741 |
+
"grad_norm": 2.842863082885742,
|
| 126742 |
+
"learning_rate": 2.6171977877271532e-05,
|
| 126743 |
+
"loss": 1.9356,
|
| 126744 |
+
"step": 18105
|
| 126745 |
+
},
|
| 126746 |
+
{
|
| 126747 |
+
"epoch": 0.4768501448511983,
|
| 126748 |
+
"grad_norm": 2.1585025787353516,
|
| 126749 |
+
"learning_rate": 2.6170661048195944e-05,
|
| 126750 |
+
"loss": 2.4486,
|
| 126751 |
+
"step": 18106
|
| 126752 |
+
},
|
| 126753 |
+
{
|
| 126754 |
+
"epoch": 0.47687648143271005,
|
| 126755 |
+
"grad_norm": 2.1451637744903564,
|
| 126756 |
+
"learning_rate": 2.6169344219120363e-05,
|
| 126757 |
+
"loss": 2.1151,
|
| 126758 |
+
"step": 18107
|
| 126759 |
+
},
|
| 126760 |
+
{
|
| 126761 |
+
"epoch": 0.47690281801422174,
|
| 126762 |
+
"grad_norm": 5.163798809051514,
|
| 126763 |
+
"learning_rate": 2.6168027390044775e-05,
|
| 126764 |
+
"loss": 0.9405,
|
| 126765 |
+
"step": 18108
|
| 126766 |
+
},
|
| 126767 |
+
{
|
| 126768 |
+
"epoch": 0.4769291545957335,
|
| 126769 |
+
"grad_norm": 1.7317043542861938,
|
| 126770 |
+
"learning_rate": 2.6166710560969188e-05,
|
| 126771 |
+
"loss": 2.1485,
|
| 126772 |
+
"step": 18109
|
| 126773 |
+
},
|
| 126774 |
+
{
|
| 126775 |
+
"epoch": 0.47695549117724517,
|
| 126776 |
+
"grad_norm": 1.8907835483551025,
|
| 126777 |
+
"learning_rate": 2.61653937318936e-05,
|
| 126778 |
+
"loss": 1.6107,
|
| 126779 |
+
"step": 18110
|
| 126780 |
+
},
|
| 126781 |
+
{
|
| 126782 |
+
"epoch": 0.4769818277587569,
|
| 126783 |
+
"grad_norm": 3.3649308681488037,
|
| 126784 |
+
"learning_rate": 2.6164076902818012e-05,
|
| 126785 |
+
"loss": 1.0804,
|
| 126786 |
+
"step": 18111
|
| 126787 |
+
},
|
| 126788 |
+
{
|
| 126789 |
+
"epoch": 0.47700816434026866,
|
| 126790 |
+
"grad_norm": 4.079046726226807,
|
| 126791 |
+
"learning_rate": 2.616276007374243e-05,
|
| 126792 |
+
"loss": 1.4524,
|
| 126793 |
+
"step": 18112
|
| 126794 |
+
},
|
| 126795 |
+
{
|
| 126796 |
+
"epoch": 0.47703450092178035,
|
| 126797 |
+
"grad_norm": 1.972220540046692,
|
| 126798 |
+
"learning_rate": 2.6161443244666843e-05,
|
| 126799 |
+
"loss": 2.1815,
|
| 126800 |
+
"step": 18113
|
| 126801 |
+
},
|
| 126802 |
+
{
|
| 126803 |
+
"epoch": 0.4770608375032921,
|
| 126804 |
+
"grad_norm": 1.8180415630340576,
|
| 126805 |
+
"learning_rate": 2.616012641559126e-05,
|
| 126806 |
+
"loss": 2.4495,
|
| 126807 |
+
"step": 18114
|
| 126808 |
+
},
|
| 126809 |
+
{
|
| 126810 |
+
"epoch": 0.4770871740848038,
|
| 126811 |
+
"grad_norm": 1.9662320613861084,
|
| 126812 |
+
"learning_rate": 2.615880958651567e-05,
|
| 126813 |
+
"loss": 1.647,
|
| 126814 |
+
"step": 18115
|
| 126815 |
+
},
|
| 126816 |
+
{
|
| 126817 |
+
"epoch": 0.4771135106663155,
|
| 126818 |
+
"grad_norm": 2.095525026321411,
|
| 126819 |
+
"learning_rate": 2.6157492757440083e-05,
|
| 126820 |
+
"loss": 1.5167,
|
| 126821 |
+
"step": 18116
|
| 126822 |
+
},
|
| 126823 |
+
{
|
| 126824 |
+
"epoch": 0.4771398472478272,
|
| 126825 |
+
"grad_norm": 2.066540479660034,
|
| 126826 |
+
"learning_rate": 2.6156175928364502e-05,
|
| 126827 |
+
"loss": 1.2558,
|
| 126828 |
+
"step": 18117
|
| 126829 |
+
},
|
| 126830 |
+
{
|
| 126831 |
+
"epoch": 0.47716618382933895,
|
| 126832 |
+
"grad_norm": 3.7202439308166504,
|
| 126833 |
+
"learning_rate": 2.6154859099288915e-05,
|
| 126834 |
+
"loss": 1.4575,
|
| 126835 |
+
"step": 18118
|
| 126836 |
+
},
|
| 126837 |
+
{
|
| 126838 |
+
"epoch": 0.4771925204108507,
|
| 126839 |
+
"grad_norm": 1.8115872144699097,
|
| 126840 |
+
"learning_rate": 2.6153542270213327e-05,
|
| 126841 |
+
"loss": 2.5709,
|
| 126842 |
+
"step": 18119
|
| 126843 |
+
},
|
| 126844 |
+
{
|
| 126845 |
+
"epoch": 0.4772188569923624,
|
| 126846 |
+
"grad_norm": 2.043302536010742,
|
| 126847 |
+
"learning_rate": 2.615222544113774e-05,
|
| 126848 |
+
"loss": 1.1423,
|
| 126849 |
+
"step": 18120
|
| 126850 |
+
},
|
| 126851 |
+
{
|
| 126852 |
+
"epoch": 0.4772451935738741,
|
| 126853 |
+
"grad_norm": 2.0807416439056396,
|
| 126854 |
+
"learning_rate": 2.6150908612062158e-05,
|
| 126855 |
+
"loss": 1.6837,
|
| 126856 |
+
"step": 18121
|
| 126857 |
+
},
|
| 126858 |
+
{
|
| 126859 |
+
"epoch": 0.4772715301553858,
|
| 126860 |
+
"grad_norm": 3.1593334674835205,
|
| 126861 |
+
"learning_rate": 2.614959178298657e-05,
|
| 126862 |
+
"loss": 1.8308,
|
| 126863 |
+
"step": 18122
|
| 126864 |
+
},
|
| 126865 |
+
{
|
| 126866 |
+
"epoch": 0.47729786673689756,
|
| 126867 |
+
"grad_norm": 3.4992871284484863,
|
| 126868 |
+
"learning_rate": 2.6148274953910982e-05,
|
| 126869 |
+
"loss": 2.1369,
|
| 126870 |
+
"step": 18123
|
| 126871 |
+
},
|
| 126872 |
+
{
|
| 126873 |
+
"epoch": 0.47732420331840925,
|
| 126874 |
+
"grad_norm": 1.5538591146469116,
|
| 126875 |
+
"learning_rate": 2.6146958124835398e-05,
|
| 126876 |
+
"loss": 2.5318,
|
| 126877 |
+
"step": 18124
|
| 126878 |
+
},
|
| 126879 |
+
{
|
| 126880 |
+
"epoch": 0.477350539899921,
|
| 126881 |
+
"grad_norm": 1.703216552734375,
|
| 126882 |
+
"learning_rate": 2.614564129575981e-05,
|
| 126883 |
+
"loss": 2.0921,
|
| 126884 |
+
"step": 18125
|
| 126885 |
+
},
|
| 126886 |
+
{
|
| 126887 |
+
"epoch": 0.47737687648143273,
|
| 126888 |
+
"grad_norm": 2.0972647666931152,
|
| 126889 |
+
"learning_rate": 2.614432446668423e-05,
|
| 126890 |
+
"loss": 2.5599,
|
| 126891 |
+
"step": 18126
|
| 126892 |
+
},
|
| 126893 |
+
{
|
| 126894 |
+
"epoch": 0.4774032130629444,
|
| 126895 |
+
"grad_norm": 4.844968795776367,
|
| 126896 |
+
"learning_rate": 2.614300763760864e-05,
|
| 126897 |
+
"loss": 1.8993,
|
| 126898 |
+
"step": 18127
|
| 126899 |
+
},
|
| 126900 |
+
{
|
| 126901 |
+
"epoch": 0.47742954964445616,
|
| 126902 |
+
"grad_norm": 1.9064234495162964,
|
| 126903 |
+
"learning_rate": 2.6141690808533054e-05,
|
| 126904 |
+
"loss": 1.6282,
|
| 126905 |
+
"step": 18128
|
| 126906 |
+
},
|
| 126907 |
+
{
|
| 126908 |
+
"epoch": 0.47745588622596785,
|
| 126909 |
+
"grad_norm": 3.336031198501587,
|
| 126910 |
+
"learning_rate": 2.6140373979457466e-05,
|
| 126911 |
+
"loss": 1.6488,
|
| 126912 |
+
"step": 18129
|
| 126913 |
+
},
|
| 126914 |
+
{
|
| 126915 |
+
"epoch": 0.4774822228074796,
|
| 126916 |
+
"grad_norm": 1.554504632949829,
|
| 126917 |
+
"learning_rate": 2.6139057150381878e-05,
|
| 126918 |
+
"loss": 0.3676,
|
| 126919 |
+
"step": 18130
|
| 126920 |
+
},
|
| 126921 |
+
{
|
| 126922 |
+
"epoch": 0.4775085593889913,
|
| 126923 |
+
"grad_norm": 1.4876564741134644,
|
| 126924 |
+
"learning_rate": 2.6137740321306297e-05,
|
| 126925 |
+
"loss": 1.8622,
|
| 126926 |
+
"step": 18131
|
| 126927 |
+
},
|
| 126928 |
+
{
|
| 126929 |
+
"epoch": 0.477534895970503,
|
| 126930 |
+
"grad_norm": 3.5113332271575928,
|
| 126931 |
+
"learning_rate": 2.613642349223071e-05,
|
| 126932 |
+
"loss": 1.1884,
|
| 126933 |
+
"step": 18132
|
| 126934 |
+
},
|
| 126935 |
+
{
|
| 126936 |
+
"epoch": 0.47756123255201477,
|
| 126937 |
+
"grad_norm": 2.460200786590576,
|
| 126938 |
+
"learning_rate": 2.6135106663155125e-05,
|
| 126939 |
+
"loss": 0.6754,
|
| 126940 |
+
"step": 18133
|
| 126941 |
+
},
|
| 126942 |
+
{
|
| 126943 |
+
"epoch": 0.47758756913352646,
|
| 126944 |
+
"grad_norm": 2.1989099979400635,
|
| 126945 |
+
"learning_rate": 2.6133789834079537e-05,
|
| 126946 |
+
"loss": 2.3714,
|
| 126947 |
+
"step": 18134
|
| 126948 |
+
},
|
| 126949 |
+
{
|
| 126950 |
+
"epoch": 0.4776139057150382,
|
| 126951 |
+
"grad_norm": 2.667492389678955,
|
| 126952 |
+
"learning_rate": 2.613247300500395e-05,
|
| 126953 |
+
"loss": 0.6263,
|
| 126954 |
+
"step": 18135
|
| 126955 |
+
},
|
| 126956 |
+
{
|
| 126957 |
+
"epoch": 0.4776402422965499,
|
| 126958 |
+
"grad_norm": 2.299617052078247,
|
| 126959 |
+
"learning_rate": 2.6131156175928368e-05,
|
| 126960 |
+
"loss": 2.8219,
|
| 126961 |
+
"step": 18136
|
| 126962 |
+
},
|
| 126963 |
+
{
|
| 126964 |
+
"epoch": 0.47766657887806163,
|
| 126965 |
+
"grad_norm": 1.612807035446167,
|
| 126966 |
+
"learning_rate": 2.612983934685278e-05,
|
| 126967 |
+
"loss": 0.2344,
|
| 126968 |
+
"step": 18137
|
| 126969 |
+
},
|
| 126970 |
+
{
|
| 126971 |
+
"epoch": 0.4776929154595734,
|
| 126972 |
+
"grad_norm": 1.7575221061706543,
|
| 126973 |
+
"learning_rate": 2.6128522517777193e-05,
|
| 126974 |
+
"loss": 1.6634,
|
| 126975 |
+
"step": 18138
|
| 126976 |
+
},
|
| 126977 |
+
{
|
| 126978 |
+
"epoch": 0.47771925204108506,
|
| 126979 |
+
"grad_norm": 4.1696457862854,
|
| 126980 |
+
"learning_rate": 2.6127205688701605e-05,
|
| 126981 |
+
"loss": 0.5157,
|
| 126982 |
+
"step": 18139
|
| 126983 |
+
},
|
| 126984 |
+
{
|
| 126985 |
+
"epoch": 0.4777455886225968,
|
| 126986 |
+
"grad_norm": 1.7531259059906006,
|
| 126987 |
+
"learning_rate": 2.6125888859626024e-05,
|
| 126988 |
+
"loss": 2.1103,
|
| 126989 |
+
"step": 18140
|
| 126990 |
+
},
|
| 126991 |
+
{
|
| 126992 |
+
"epoch": 0.4777719252041085,
|
| 126993 |
+
"grad_norm": 1.9951260089874268,
|
| 126994 |
+
"learning_rate": 2.6124572030550436e-05,
|
| 126995 |
+
"loss": 1.7099,
|
| 126996 |
+
"step": 18141
|
| 126997 |
+
},
|
| 126998 |
+
{
|
| 126999 |
+
"epoch": 0.47779826178562024,
|
| 127000 |
+
"grad_norm": 2.5088744163513184,
|
| 127001 |
+
"learning_rate": 2.612325520147485e-05,
|
| 127002 |
+
"loss": 1.277,
|
| 127003 |
+
"step": 18142
|
| 127004 |
+
},
|
| 127005 |
+
{
|
| 127006 |
+
"epoch": 0.4778245983671319,
|
| 127007 |
+
"grad_norm": 1.8694902658462524,
|
| 127008 |
+
"learning_rate": 2.6121938372399264e-05,
|
| 127009 |
+
"loss": 1.7016,
|
| 127010 |
+
"step": 18143
|
| 127011 |
+
},
|
| 127012 |
+
{
|
| 127013 |
+
"epoch": 0.47785093494864367,
|
| 127014 |
+
"grad_norm": 2.1027448177337646,
|
| 127015 |
+
"learning_rate": 2.6120621543323676e-05,
|
| 127016 |
+
"loss": 2.3466,
|
| 127017 |
+
"step": 18144
|
| 127018 |
+
},
|
| 127019 |
+
{
|
| 127020 |
+
"epoch": 0.4778772715301554,
|
| 127021 |
+
"grad_norm": 5.901076793670654,
|
| 127022 |
+
"learning_rate": 2.6119304714248095e-05,
|
| 127023 |
+
"loss": 1.5969,
|
| 127024 |
+
"step": 18145
|
| 127025 |
+
},
|
| 127026 |
+
{
|
| 127027 |
+
"epoch": 0.4779036081116671,
|
| 127028 |
+
"grad_norm": 1.9883469343185425,
|
| 127029 |
+
"learning_rate": 2.6117987885172507e-05,
|
| 127030 |
+
"loss": 1.7854,
|
| 127031 |
+
"step": 18146
|
| 127032 |
+
},
|
| 127033 |
+
{
|
| 127034 |
+
"epoch": 0.47792994469317884,
|
| 127035 |
+
"grad_norm": 2.4258086681365967,
|
| 127036 |
+
"learning_rate": 2.611667105609692e-05,
|
| 127037 |
+
"loss": 2.0893,
|
| 127038 |
+
"step": 18147
|
| 127039 |
+
},
|
| 127040 |
+
{
|
| 127041 |
+
"epoch": 0.47795628127469053,
|
| 127042 |
+
"grad_norm": 1.5321277379989624,
|
| 127043 |
+
"learning_rate": 2.611535422702133e-05,
|
| 127044 |
+
"loss": 2.0222,
|
| 127045 |
+
"step": 18148
|
| 127046 |
+
},
|
| 127047 |
+
{
|
| 127048 |
+
"epoch": 0.4779826178562023,
|
| 127049 |
+
"grad_norm": 2.2107131481170654,
|
| 127050 |
+
"learning_rate": 2.6114037397945744e-05,
|
| 127051 |
+
"loss": 0.276,
|
| 127052 |
+
"step": 18149
|
| 127053 |
+
},
|
| 127054 |
+
{
|
| 127055 |
+
"epoch": 0.47800895443771396,
|
| 127056 |
+
"grad_norm": 2.2959635257720947,
|
| 127057 |
+
"learning_rate": 2.6112720568870163e-05,
|
| 127058 |
+
"loss": 2.1733,
|
| 127059 |
+
"step": 18150
|
| 127060 |
+
},
|
| 127061 |
+
{
|
| 127062 |
+
"epoch": 0.4780352910192257,
|
| 127063 |
+
"grad_norm": 2.7273948192596436,
|
| 127064 |
+
"learning_rate": 2.6111403739794575e-05,
|
| 127065 |
+
"loss": 2.3528,
|
| 127066 |
+
"step": 18151
|
| 127067 |
+
},
|
| 127068 |
+
{
|
| 127069 |
+
"epoch": 0.47806162760073745,
|
| 127070 |
+
"grad_norm": 1.675957441329956,
|
| 127071 |
+
"learning_rate": 2.611008691071899e-05,
|
| 127072 |
+
"loss": 1.7183,
|
| 127073 |
+
"step": 18152
|
| 127074 |
+
},
|
| 127075 |
+
{
|
| 127076 |
+
"epoch": 0.47808796418224914,
|
| 127077 |
+
"grad_norm": 2.2190258502960205,
|
| 127078 |
+
"learning_rate": 2.6108770081643403e-05,
|
| 127079 |
+
"loss": 1.826,
|
| 127080 |
+
"step": 18153
|
| 127081 |
+
},
|
| 127082 |
+
{
|
| 127083 |
+
"epoch": 0.4781143007637609,
|
| 127084 |
+
"grad_norm": 3.512843370437622,
|
| 127085 |
+
"learning_rate": 2.6107453252567822e-05,
|
| 127086 |
+
"loss": 2.0663,
|
| 127087 |
+
"step": 18154
|
| 127088 |
+
},
|
| 127089 |
+
{
|
| 127090 |
+
"epoch": 0.47814063734527257,
|
| 127091 |
+
"grad_norm": 1.5189660787582397,
|
| 127092 |
+
"learning_rate": 2.6106136423492234e-05,
|
| 127093 |
+
"loss": 2.2426,
|
| 127094 |
+
"step": 18155
|
| 127095 |
+
},
|
| 127096 |
+
{
|
| 127097 |
+
"epoch": 0.4781669739267843,
|
| 127098 |
+
"grad_norm": 5.404087543487549,
|
| 127099 |
+
"learning_rate": 2.6104819594416646e-05,
|
| 127100 |
+
"loss": 2.1597,
|
| 127101 |
+
"step": 18156
|
| 127102 |
+
},
|
| 127103 |
+
{
|
| 127104 |
+
"epoch": 0.478193310508296,
|
| 127105 |
+
"grad_norm": 2.591707229614258,
|
| 127106 |
+
"learning_rate": 2.610350276534106e-05,
|
| 127107 |
+
"loss": 2.4166,
|
| 127108 |
+
"step": 18157
|
| 127109 |
+
},
|
| 127110 |
+
{
|
| 127111 |
+
"epoch": 0.47821964708980774,
|
| 127112 |
+
"grad_norm": 4.782519817352295,
|
| 127113 |
+
"learning_rate": 2.610218593626547e-05,
|
| 127114 |
+
"loss": 1.4625,
|
| 127115 |
+
"step": 18158
|
| 127116 |
+
},
|
| 127117 |
+
{
|
| 127118 |
+
"epoch": 0.4782459836713195,
|
| 127119 |
+
"grad_norm": 2.67582106590271,
|
| 127120 |
+
"learning_rate": 2.610086910718989e-05,
|
| 127121 |
+
"loss": 1.7045,
|
| 127122 |
+
"step": 18159
|
| 127123 |
+
},
|
| 127124 |
+
{
|
| 127125 |
+
"epoch": 0.4782723202528312,
|
| 127126 |
+
"grad_norm": 1.6655147075653076,
|
| 127127 |
+
"learning_rate": 2.6099552278114302e-05,
|
| 127128 |
+
"loss": 1.421,
|
| 127129 |
+
"step": 18160
|
| 127130 |
+
},
|
| 127131 |
+
{
|
| 127132 |
+
"epoch": 0.4782986568343429,
|
| 127133 |
+
"grad_norm": 3.701991081237793,
|
| 127134 |
+
"learning_rate": 2.6098235449038717e-05,
|
| 127135 |
+
"loss": 1.9044,
|
| 127136 |
+
"step": 18161
|
| 127137 |
+
},
|
| 127138 |
+
{
|
| 127139 |
+
"epoch": 0.4783249934158546,
|
| 127140 |
+
"grad_norm": 1.59872567653656,
|
| 127141 |
+
"learning_rate": 2.609691861996313e-05,
|
| 127142 |
+
"loss": 2.3976,
|
| 127143 |
+
"step": 18162
|
| 127144 |
+
},
|
| 127145 |
+
{
|
| 127146 |
+
"epoch": 0.47835132999736635,
|
| 127147 |
+
"grad_norm": 1.5172570943832397,
|
| 127148 |
+
"learning_rate": 2.6095601790887542e-05,
|
| 127149 |
+
"loss": 1.5677,
|
| 127150 |
+
"step": 18163
|
| 127151 |
+
},
|
| 127152 |
+
{
|
| 127153 |
+
"epoch": 0.47837766657887804,
|
| 127154 |
+
"grad_norm": 2.137535333633423,
|
| 127155 |
+
"learning_rate": 2.609428496181196e-05,
|
| 127156 |
+
"loss": 1.8379,
|
| 127157 |
+
"step": 18164
|
| 127158 |
+
},
|
| 127159 |
+
{
|
| 127160 |
+
"epoch": 0.4784040031603898,
|
| 127161 |
+
"grad_norm": 5.482143402099609,
|
| 127162 |
+
"learning_rate": 2.6092968132736373e-05,
|
| 127163 |
+
"loss": 1.0263,
|
| 127164 |
+
"step": 18165
|
| 127165 |
+
},
|
| 127166 |
+
{
|
| 127167 |
+
"epoch": 0.4784303397419015,
|
| 127168 |
+
"grad_norm": 1.9915279150009155,
|
| 127169 |
+
"learning_rate": 2.6091651303660785e-05,
|
| 127170 |
+
"loss": 1.9052,
|
| 127171 |
+
"step": 18166
|
| 127172 |
+
},
|
| 127173 |
+
{
|
| 127174 |
+
"epoch": 0.4784566763234132,
|
| 127175 |
+
"grad_norm": 2.3375186920166016,
|
| 127176 |
+
"learning_rate": 2.6090334474585197e-05,
|
| 127177 |
+
"loss": 0.7127,
|
| 127178 |
+
"step": 18167
|
| 127179 |
+
},
|
| 127180 |
+
{
|
| 127181 |
+
"epoch": 0.47848301290492495,
|
| 127182 |
+
"grad_norm": 1.8854601383209229,
|
| 127183 |
+
"learning_rate": 2.608901764550961e-05,
|
| 127184 |
+
"loss": 2.0186,
|
| 127185 |
+
"step": 18168
|
| 127186 |
+
},
|
| 127187 |
+
{
|
| 127188 |
+
"epoch": 0.47850934948643664,
|
| 127189 |
+
"grad_norm": 1.6205345392227173,
|
| 127190 |
+
"learning_rate": 2.608770081643403e-05,
|
| 127191 |
+
"loss": 2.5091,
|
| 127192 |
+
"step": 18169
|
| 127193 |
+
},
|
| 127194 |
+
{
|
| 127195 |
+
"epoch": 0.4785356860679484,
|
| 127196 |
+
"grad_norm": 1.597391128540039,
|
| 127197 |
+
"learning_rate": 2.608638398735844e-05,
|
| 127198 |
+
"loss": 1.7359,
|
| 127199 |
+
"step": 18170
|
| 127200 |
+
},
|
| 127201 |
+
{
|
| 127202 |
+
"epoch": 0.4785620226494601,
|
| 127203 |
+
"grad_norm": 3.3700568675994873,
|
| 127204 |
+
"learning_rate": 2.6085067158282856e-05,
|
| 127205 |
+
"loss": 1.5659,
|
| 127206 |
+
"step": 18171
|
| 127207 |
+
},
|
| 127208 |
+
{
|
| 127209 |
+
"epoch": 0.4785883592309718,
|
| 127210 |
+
"grad_norm": 1.645609974861145,
|
| 127211 |
+
"learning_rate": 2.608375032920727e-05,
|
| 127212 |
+
"loss": 1.4701,
|
| 127213 |
+
"step": 18172
|
| 127214 |
+
},
|
| 127215 |
+
{
|
| 127216 |
+
"epoch": 0.47861469581248356,
|
| 127217 |
+
"grad_norm": 2.141824245452881,
|
| 127218 |
+
"learning_rate": 2.6082433500131688e-05,
|
| 127219 |
+
"loss": 2.4564,
|
| 127220 |
+
"step": 18173
|
| 127221 |
+
},
|
| 127222 |
+
{
|
| 127223 |
+
"epoch": 0.47864103239399525,
|
| 127224 |
+
"grad_norm": 1.818982481956482,
|
| 127225 |
+
"learning_rate": 2.60811166710561e-05,
|
| 127226 |
+
"loss": 1.02,
|
| 127227 |
+
"step": 18174
|
| 127228 |
+
},
|
| 127229 |
+
{
|
| 127230 |
+
"epoch": 0.478667368975507,
|
| 127231 |
+
"grad_norm": 1.7090140581130981,
|
| 127232 |
+
"learning_rate": 2.6079799841980512e-05,
|
| 127233 |
+
"loss": 1.7205,
|
| 127234 |
+
"step": 18175
|
| 127235 |
+
},
|
| 127236 |
+
{
|
| 127237 |
+
"epoch": 0.4786937055570187,
|
| 127238 |
+
"grad_norm": 2.0209977626800537,
|
| 127239 |
+
"learning_rate": 2.6078483012904924e-05,
|
| 127240 |
+
"loss": 1.3472,
|
| 127241 |
+
"step": 18176
|
| 127242 |
+
},
|
| 127243 |
+
{
|
| 127244 |
+
"epoch": 0.4787200421385304,
|
| 127245 |
+
"grad_norm": 1.909293532371521,
|
| 127246 |
+
"learning_rate": 2.6077166183829337e-05,
|
| 127247 |
+
"loss": 1.2932,
|
| 127248 |
+
"step": 18177
|
| 127249 |
+
},
|
| 127250 |
+
{
|
| 127251 |
+
"epoch": 0.47874637872004216,
|
| 127252 |
+
"grad_norm": 2.426377773284912,
|
| 127253 |
+
"learning_rate": 2.6075849354753756e-05,
|
| 127254 |
+
"loss": 1.3127,
|
| 127255 |
+
"step": 18178
|
| 127256 |
+
},
|
| 127257 |
+
{
|
| 127258 |
+
"epoch": 0.47877271530155385,
|
| 127259 |
+
"grad_norm": 5.718318939208984,
|
| 127260 |
+
"learning_rate": 2.6074532525678168e-05,
|
| 127261 |
+
"loss": 2.0245,
|
| 127262 |
+
"step": 18179
|
| 127263 |
+
},
|
| 127264 |
+
{
|
| 127265 |
+
"epoch": 0.4787990518830656,
|
| 127266 |
+
"grad_norm": 3.2824246883392334,
|
| 127267 |
+
"learning_rate": 2.6073215696602583e-05,
|
| 127268 |
+
"loss": 1.6523,
|
| 127269 |
+
"step": 18180
|
| 127270 |
+
},
|
| 127271 |
+
{
|
| 127272 |
+
"epoch": 0.4788253884645773,
|
| 127273 |
+
"grad_norm": 2.88808274269104,
|
| 127274 |
+
"learning_rate": 2.6071898867526996e-05,
|
| 127275 |
+
"loss": 1.4564,
|
| 127276 |
+
"step": 18181
|
| 127277 |
+
},
|
| 127278 |
+
{
|
| 127279 |
+
"epoch": 0.478851725046089,
|
| 127280 |
+
"grad_norm": 1.67848539352417,
|
| 127281 |
+
"learning_rate": 2.6070582038451408e-05,
|
| 127282 |
+
"loss": 1.81,
|
| 127283 |
+
"step": 18182
|
| 127284 |
+
},
|
| 127285 |
+
{
|
| 127286 |
+
"epoch": 0.4788780616276007,
|
| 127287 |
+
"grad_norm": 6.433162689208984,
|
| 127288 |
+
"learning_rate": 2.6069265209375827e-05,
|
| 127289 |
+
"loss": 1.011,
|
| 127290 |
+
"step": 18183
|
| 127291 |
+
},
|
| 127292 |
+
{
|
| 127293 |
+
"epoch": 0.47890439820911246,
|
| 127294 |
+
"grad_norm": 2.4767961502075195,
|
| 127295 |
+
"learning_rate": 2.606794838030024e-05,
|
| 127296 |
+
"loss": 1.3045,
|
| 127297 |
+
"step": 18184
|
| 127298 |
+
},
|
| 127299 |
+
{
|
| 127300 |
+
"epoch": 0.4789307347906242,
|
| 127301 |
+
"grad_norm": 5.748504638671875,
|
| 127302 |
+
"learning_rate": 2.606663155122465e-05,
|
| 127303 |
+
"loss": 1.2878,
|
| 127304 |
+
"step": 18185
|
| 127305 |
+
},
|
| 127306 |
+
{
|
| 127307 |
+
"epoch": 0.4789570713721359,
|
| 127308 |
+
"grad_norm": 1.5263091325759888,
|
| 127309 |
+
"learning_rate": 2.6065314722149063e-05,
|
| 127310 |
+
"loss": 1.87,
|
| 127311 |
+
"step": 18186
|
| 127312 |
+
},
|
| 127313 |
+
{
|
| 127314 |
+
"epoch": 0.47898340795364763,
|
| 127315 |
+
"grad_norm": 1.9353070259094238,
|
| 127316 |
+
"learning_rate": 2.6063997893073482e-05,
|
| 127317 |
+
"loss": 2.8552,
|
| 127318 |
+
"step": 18187
|
| 127319 |
+
},
|
| 127320 |
+
{
|
| 127321 |
+
"epoch": 0.4790097445351593,
|
| 127322 |
+
"grad_norm": 1.4300810098648071,
|
| 127323 |
+
"learning_rate": 2.6062681063997895e-05,
|
| 127324 |
+
"loss": 2.3514,
|
| 127325 |
+
"step": 18188
|
| 127326 |
+
},
|
| 127327 |
+
{
|
| 127328 |
+
"epoch": 0.47903608111667106,
|
| 127329 |
+
"grad_norm": 1.5277667045593262,
|
| 127330 |
+
"learning_rate": 2.606136423492231e-05,
|
| 127331 |
+
"loss": 1.7833,
|
| 127332 |
+
"step": 18189
|
| 127333 |
+
},
|
| 127334 |
+
{
|
| 127335 |
+
"epoch": 0.47906241769818275,
|
| 127336 |
+
"grad_norm": 1.8952159881591797,
|
| 127337 |
+
"learning_rate": 2.6060047405846722e-05,
|
| 127338 |
+
"loss": 1.7228,
|
| 127339 |
+
"step": 18190
|
| 127340 |
+
},
|
| 127341 |
+
{
|
| 127342 |
+
"epoch": 0.4790887542796945,
|
| 127343 |
+
"grad_norm": 1.6774052381515503,
|
| 127344 |
+
"learning_rate": 2.6058730576771135e-05,
|
| 127345 |
+
"loss": 1.7671,
|
| 127346 |
+
"step": 18191
|
| 127347 |
+
},
|
| 127348 |
+
{
|
| 127349 |
+
"epoch": 0.47911509086120624,
|
| 127350 |
+
"grad_norm": 2.0082924365997314,
|
| 127351 |
+
"learning_rate": 2.6057413747695554e-05,
|
| 127352 |
+
"loss": 1.6773,
|
| 127353 |
+
"step": 18192
|
| 127354 |
+
},
|
| 127355 |
+
{
|
| 127356 |
+
"epoch": 0.4791414274427179,
|
| 127357 |
+
"grad_norm": 4.2807297706604,
|
| 127358 |
+
"learning_rate": 2.6056096918619966e-05,
|
| 127359 |
+
"loss": 0.799,
|
| 127360 |
+
"step": 18193
|
| 127361 |
+
},
|
| 127362 |
+
{
|
| 127363 |
+
"epoch": 0.47916776402422967,
|
| 127364 |
+
"grad_norm": 2.550044059753418,
|
| 127365 |
+
"learning_rate": 2.6054780089544378e-05,
|
| 127366 |
+
"loss": 2.0308,
|
| 127367 |
+
"step": 18194
|
| 127368 |
+
},
|
| 127369 |
+
{
|
| 127370 |
+
"epoch": 0.47919410060574136,
|
| 127371 |
+
"grad_norm": 1.816925287246704,
|
| 127372 |
+
"learning_rate": 2.605346326046879e-05,
|
| 127373 |
+
"loss": 1.781,
|
| 127374 |
+
"step": 18195
|
| 127375 |
+
},
|
| 127376 |
+
{
|
| 127377 |
+
"epoch": 0.4792204371872531,
|
| 127378 |
+
"grad_norm": 1.9360864162445068,
|
| 127379 |
+
"learning_rate": 2.6052146431393202e-05,
|
| 127380 |
+
"loss": 2.1195,
|
| 127381 |
+
"step": 18196
|
| 127382 |
+
},
|
| 127383 |
+
{
|
| 127384 |
+
"epoch": 0.4792467737687648,
|
| 127385 |
+
"grad_norm": 2.2189080715179443,
|
| 127386 |
+
"learning_rate": 2.605082960231762e-05,
|
| 127387 |
+
"loss": 1.9188,
|
| 127388 |
+
"step": 18197
|
| 127389 |
+
},
|
| 127390 |
+
{
|
| 127391 |
+
"epoch": 0.47927311035027653,
|
| 127392 |
+
"grad_norm": 1.4427847862243652,
|
| 127393 |
+
"learning_rate": 2.6049512773242034e-05,
|
| 127394 |
+
"loss": 1.5472,
|
| 127395 |
+
"step": 18198
|
| 127396 |
+
},
|
| 127397 |
+
{
|
| 127398 |
+
"epoch": 0.4792994469317883,
|
| 127399 |
+
"grad_norm": 2.8791027069091797,
|
| 127400 |
+
"learning_rate": 2.604819594416645e-05,
|
| 127401 |
+
"loss": 1.3314,
|
| 127402 |
+
"step": 18199
|
| 127403 |
+
},
|
| 127404 |
+
{
|
| 127405 |
+
"epoch": 0.47932578351329996,
|
| 127406 |
+
"grad_norm": 5.540414333343506,
|
| 127407 |
+
"learning_rate": 2.604687911509086e-05,
|
| 127408 |
+
"loss": 2.2421,
|
| 127409 |
+
"step": 18200
|
| 127410 |
+
},
|
| 127411 |
+
{
|
| 127412 |
+
"epoch": 0.4793521200948117,
|
| 127413 |
+
"grad_norm": 5.057022571563721,
|
| 127414 |
+
"learning_rate": 2.6045562286015274e-05,
|
| 127415 |
+
"loss": 0.7574,
|
| 127416 |
+
"step": 18201
|
| 127417 |
+
},
|
| 127418 |
+
{
|
| 127419 |
+
"epoch": 0.4793784566763234,
|
| 127420 |
+
"grad_norm": 2.3301544189453125,
|
| 127421 |
+
"learning_rate": 2.6044245456939693e-05,
|
| 127422 |
+
"loss": 1.0735,
|
| 127423 |
+
"step": 18202
|
| 127424 |
+
},
|
| 127425 |
+
{
|
| 127426 |
+
"epoch": 0.47940479325783514,
|
| 127427 |
+
"grad_norm": 1.5644989013671875,
|
| 127428 |
+
"learning_rate": 2.6042928627864105e-05,
|
| 127429 |
+
"loss": 1.4331,
|
| 127430 |
+
"step": 18203
|
| 127431 |
+
},
|
| 127432 |
+
{
|
| 127433 |
+
"epoch": 0.4794311298393468,
|
| 127434 |
+
"grad_norm": 3.384436845779419,
|
| 127435 |
+
"learning_rate": 2.6041611798788517e-05,
|
| 127436 |
+
"loss": 0.5878,
|
| 127437 |
+
"step": 18204
|
| 127438 |
+
},
|
| 127439 |
+
{
|
| 127440 |
+
"epoch": 0.47945746642085857,
|
| 127441 |
+
"grad_norm": 4.111256122589111,
|
| 127442 |
+
"learning_rate": 2.604029496971293e-05,
|
| 127443 |
+
"loss": 2.0897,
|
| 127444 |
+
"step": 18205
|
| 127445 |
+
},
|
| 127446 |
+
{
|
| 127447 |
+
"epoch": 0.4794838030023703,
|
| 127448 |
+
"grad_norm": 2.0982885360717773,
|
| 127449 |
+
"learning_rate": 2.6038978140637348e-05,
|
| 127450 |
+
"loss": 1.9615,
|
| 127451 |
+
"step": 18206
|
| 127452 |
+
},
|
| 127453 |
+
{
|
| 127454 |
+
"epoch": 0.479510139583882,
|
| 127455 |
+
"grad_norm": 3.7931666374206543,
|
| 127456 |
+
"learning_rate": 2.603766131156176e-05,
|
| 127457 |
+
"loss": 0.831,
|
| 127458 |
+
"step": 18207
|
| 127459 |
+
},
|
| 127460 |
+
{
|
| 127461 |
+
"epoch": 0.47953647616539374,
|
| 127462 |
+
"grad_norm": 4.855052947998047,
|
| 127463 |
+
"learning_rate": 2.6036344482486176e-05,
|
| 127464 |
+
"loss": 0.9936,
|
| 127465 |
+
"step": 18208
|
| 127466 |
+
},
|
| 127467 |
+
{
|
| 127468 |
+
"epoch": 0.47956281274690543,
|
| 127469 |
+
"grad_norm": 1.6595323085784912,
|
| 127470 |
+
"learning_rate": 2.6035027653410588e-05,
|
| 127471 |
+
"loss": 1.7072,
|
| 127472 |
+
"step": 18209
|
| 127473 |
+
},
|
| 127474 |
+
{
|
| 127475 |
+
"epoch": 0.4795891493284172,
|
| 127476 |
+
"grad_norm": 1.57021963596344,
|
| 127477 |
+
"learning_rate": 2.6033710824335e-05,
|
| 127478 |
+
"loss": 1.811,
|
| 127479 |
+
"step": 18210
|
| 127480 |
+
},
|
| 127481 |
+
{
|
| 127482 |
+
"epoch": 0.4796154859099289,
|
| 127483 |
+
"grad_norm": 1.3585107326507568,
|
| 127484 |
+
"learning_rate": 2.603239399525942e-05,
|
| 127485 |
+
"loss": 1.5773,
|
| 127486 |
+
"step": 18211
|
| 127487 |
+
},
|
| 127488 |
+
{
|
| 127489 |
+
"epoch": 0.4796418224914406,
|
| 127490 |
+
"grad_norm": 1.8488690853118896,
|
| 127491 |
+
"learning_rate": 2.603107716618383e-05,
|
| 127492 |
+
"loss": 1.7188,
|
| 127493 |
+
"step": 18212
|
| 127494 |
+
},
|
| 127495 |
+
{
|
| 127496 |
+
"epoch": 0.47966815907295235,
|
| 127497 |
+
"grad_norm": 3.2584662437438965,
|
| 127498 |
+
"learning_rate": 2.6029760337108244e-05,
|
| 127499 |
+
"loss": 1.1955,
|
| 127500 |
+
"step": 18213
|
| 127501 |
+
},
|
| 127502 |
+
{
|
| 127503 |
+
"epoch": 0.47969449565446404,
|
| 127504 |
+
"grad_norm": 1.6268631219863892,
|
| 127505 |
+
"learning_rate": 2.6028443508032656e-05,
|
| 127506 |
+
"loss": 1.4492,
|
| 127507 |
+
"step": 18214
|
| 127508 |
+
},
|
| 127509 |
+
{
|
| 127510 |
+
"epoch": 0.4797208322359758,
|
| 127511 |
+
"grad_norm": 1.6208415031433105,
|
| 127512 |
+
"learning_rate": 2.602712667895707e-05,
|
| 127513 |
+
"loss": 1.8537,
|
| 127514 |
+
"step": 18215
|
| 127515 |
+
},
|
| 127516 |
+
{
|
| 127517 |
+
"epoch": 0.47974716881748747,
|
| 127518 |
+
"grad_norm": 1.8901538848876953,
|
| 127519 |
+
"learning_rate": 2.6025809849881487e-05,
|
| 127520 |
+
"loss": 2.321,
|
| 127521 |
+
"step": 18216
|
| 127522 |
+
},
|
| 127523 |
+
{
|
| 127524 |
+
"epoch": 0.4797735053989992,
|
| 127525 |
+
"grad_norm": 1.631158471107483,
|
| 127526 |
+
"learning_rate": 2.6024493020805903e-05,
|
| 127527 |
+
"loss": 0.445,
|
| 127528 |
+
"step": 18217
|
| 127529 |
+
},
|
| 127530 |
+
{
|
| 127531 |
+
"epoch": 0.47979984198051095,
|
| 127532 |
+
"grad_norm": 2.484971523284912,
|
| 127533 |
+
"learning_rate": 2.6023176191730315e-05,
|
| 127534 |
+
"loss": 2.1254,
|
| 127535 |
+
"step": 18218
|
| 127536 |
+
},
|
| 127537 |
+
{
|
| 127538 |
+
"epoch": 0.47982617856202264,
|
| 127539 |
+
"grad_norm": 1.7783170938491821,
|
| 127540 |
+
"learning_rate": 2.6021859362654727e-05,
|
| 127541 |
+
"loss": 1.9419,
|
| 127542 |
+
"step": 18219
|
| 127543 |
+
},
|
| 127544 |
+
{
|
| 127545 |
+
"epoch": 0.4798525151435344,
|
| 127546 |
+
"grad_norm": 2.196690082550049,
|
| 127547 |
+
"learning_rate": 2.6020542533579146e-05,
|
| 127548 |
+
"loss": 2.1618,
|
| 127549 |
+
"step": 18220
|
| 127550 |
+
},
|
| 127551 |
+
{
|
| 127552 |
+
"epoch": 0.4798788517250461,
|
| 127553 |
+
"grad_norm": 1.8459336757659912,
|
| 127554 |
+
"learning_rate": 2.601922570450356e-05,
|
| 127555 |
+
"loss": 1.5345,
|
| 127556 |
+
"step": 18221
|
| 127557 |
+
},
|
| 127558 |
+
{
|
| 127559 |
+
"epoch": 0.4799051883065578,
|
| 127560 |
+
"grad_norm": 2.015507936477661,
|
| 127561 |
+
"learning_rate": 2.601790887542797e-05,
|
| 127562 |
+
"loss": 1.4419,
|
| 127563 |
+
"step": 18222
|
| 127564 |
+
},
|
| 127565 |
+
{
|
| 127566 |
+
"epoch": 0.4799315248880695,
|
| 127567 |
+
"grad_norm": 1.6010900735855103,
|
| 127568 |
+
"learning_rate": 2.6016592046352383e-05,
|
| 127569 |
+
"loss": 1.935,
|
| 127570 |
+
"step": 18223
|
| 127571 |
+
},
|
| 127572 |
+
{
|
| 127573 |
+
"epoch": 0.47995786146958125,
|
| 127574 |
+
"grad_norm": 1.7097499370574951,
|
| 127575 |
+
"learning_rate": 2.6015275217276795e-05,
|
| 127576 |
+
"loss": 1.0837,
|
| 127577 |
+
"step": 18224
|
| 127578 |
+
},
|
| 127579 |
+
{
|
| 127580 |
+
"epoch": 0.479984198051093,
|
| 127581 |
+
"grad_norm": 3.9856553077697754,
|
| 127582 |
+
"learning_rate": 2.6013958388201214e-05,
|
| 127583 |
+
"loss": 2.4888,
|
| 127584 |
+
"step": 18225
|
| 127585 |
+
},
|
| 127586 |
+
{
|
| 127587 |
+
"epoch": 0.4800105346326047,
|
| 127588 |
+
"grad_norm": 3.3391363620758057,
|
| 127589 |
+
"learning_rate": 2.6012641559125626e-05,
|
| 127590 |
+
"loss": 1.6933,
|
| 127591 |
+
"step": 18226
|
| 127592 |
+
},
|
| 127593 |
+
{
|
| 127594 |
+
"epoch": 0.4800368712141164,
|
| 127595 |
+
"grad_norm": 2.165415048599243,
|
| 127596 |
+
"learning_rate": 2.6011324730050042e-05,
|
| 127597 |
+
"loss": 2.046,
|
| 127598 |
+
"step": 18227
|
| 127599 |
+
},
|
| 127600 |
+
{
|
| 127601 |
+
"epoch": 0.4800632077956281,
|
| 127602 |
+
"grad_norm": 3.911407232284546,
|
| 127603 |
+
"learning_rate": 2.6010007900974454e-05,
|
| 127604 |
+
"loss": 0.6511,
|
| 127605 |
+
"step": 18228
|
| 127606 |
+
},
|
| 127607 |
+
{
|
| 127608 |
+
"epoch": 0.48008954437713985,
|
| 127609 |
+
"grad_norm": 1.8012775182724,
|
| 127610 |
+
"learning_rate": 2.6008691071898866e-05,
|
| 127611 |
+
"loss": 1.3854,
|
| 127612 |
+
"step": 18229
|
| 127613 |
+
},
|
| 127614 |
+
{
|
| 127615 |
+
"epoch": 0.48011588095865154,
|
| 127616 |
+
"grad_norm": 2.567828416824341,
|
| 127617 |
+
"learning_rate": 2.6007374242823285e-05,
|
| 127618 |
+
"loss": 0.7743,
|
| 127619 |
+
"step": 18230
|
| 127620 |
+
},
|
| 127621 |
+
{
|
| 127622 |
+
"epoch": 0.4801422175401633,
|
| 127623 |
+
"grad_norm": 2.0181589126586914,
|
| 127624 |
+
"learning_rate": 2.6006057413747697e-05,
|
| 127625 |
+
"loss": 2.3429,
|
| 127626 |
+
"step": 18231
|
| 127627 |
+
},
|
| 127628 |
+
{
|
| 127629 |
+
"epoch": 0.48016855412167503,
|
| 127630 |
+
"grad_norm": 1.8341394662857056,
|
| 127631 |
+
"learning_rate": 2.600474058467211e-05,
|
| 127632 |
+
"loss": 1.6094,
|
| 127633 |
+
"step": 18232
|
| 127634 |
+
},
|
| 127635 |
+
{
|
| 127636 |
+
"epoch": 0.4801948907031867,
|
| 127637 |
+
"grad_norm": 3.8895981311798096,
|
| 127638 |
+
"learning_rate": 2.6003423755596522e-05,
|
| 127639 |
+
"loss": 1.0459,
|
| 127640 |
+
"step": 18233
|
| 127641 |
+
},
|
| 127642 |
+
{
|
| 127643 |
+
"epoch": 0.48022122728469846,
|
| 127644 |
+
"grad_norm": 2.5247411727905273,
|
| 127645 |
+
"learning_rate": 2.600210692652094e-05,
|
| 127646 |
+
"loss": 1.2766,
|
| 127647 |
+
"step": 18234
|
| 127648 |
+
},
|
| 127649 |
+
{
|
| 127650 |
+
"epoch": 0.48024756386621015,
|
| 127651 |
+
"grad_norm": 1.8207776546478271,
|
| 127652 |
+
"learning_rate": 2.6000790097445353e-05,
|
| 127653 |
+
"loss": 2.0494,
|
| 127654 |
+
"step": 18235
|
| 127655 |
+
},
|
| 127656 |
+
{
|
| 127657 |
+
"epoch": 0.4802739004477219,
|
| 127658 |
+
"grad_norm": 2.1421167850494385,
|
| 127659 |
+
"learning_rate": 2.599947326836977e-05,
|
| 127660 |
+
"loss": 2.0554,
|
| 127661 |
+
"step": 18236
|
| 127662 |
+
},
|
| 127663 |
+
{
|
| 127664 |
+
"epoch": 0.4803002370292336,
|
| 127665 |
+
"grad_norm": 1.9610754251480103,
|
| 127666 |
+
"learning_rate": 2.599815643929418e-05,
|
| 127667 |
+
"loss": 1.6792,
|
| 127668 |
+
"step": 18237
|
| 127669 |
+
},
|
| 127670 |
+
{
|
| 127671 |
+
"epoch": 0.4803265736107453,
|
| 127672 |
+
"grad_norm": 2.152301073074341,
|
| 127673 |
+
"learning_rate": 2.5996839610218593e-05,
|
| 127674 |
+
"loss": 1.2206,
|
| 127675 |
+
"step": 18238
|
| 127676 |
+
},
|
| 127677 |
+
{
|
| 127678 |
+
"epoch": 0.48035291019225707,
|
| 127679 |
+
"grad_norm": 1.690805435180664,
|
| 127680 |
+
"learning_rate": 2.5995522781143012e-05,
|
| 127681 |
+
"loss": 2.1232,
|
| 127682 |
+
"step": 18239
|
| 127683 |
+
},
|
| 127684 |
+
{
|
| 127685 |
+
"epoch": 0.48037924677376875,
|
| 127686 |
+
"grad_norm": 1.7222952842712402,
|
| 127687 |
+
"learning_rate": 2.5994205952067424e-05,
|
| 127688 |
+
"loss": 0.591,
|
| 127689 |
+
"step": 18240
|
| 127690 |
+
},
|
| 127691 |
+
{
|
| 127692 |
+
"epoch": 0.4804055833552805,
|
| 127693 |
+
"grad_norm": 1.9552398920059204,
|
| 127694 |
+
"learning_rate": 2.5992889122991837e-05,
|
| 127695 |
+
"loss": 1.0758,
|
| 127696 |
+
"step": 18241
|
| 127697 |
+
},
|
| 127698 |
+
{
|
| 127699 |
+
"epoch": 0.4804319199367922,
|
| 127700 |
+
"grad_norm": 2.508448839187622,
|
| 127701 |
+
"learning_rate": 2.599157229391625e-05,
|
| 127702 |
+
"loss": 1.3292,
|
| 127703 |
+
"step": 18242
|
| 127704 |
+
},
|
| 127705 |
+
{
|
| 127706 |
+
"epoch": 0.48045825651830393,
|
| 127707 |
+
"grad_norm": 2.046084403991699,
|
| 127708 |
+
"learning_rate": 2.5990255464840664e-05,
|
| 127709 |
+
"loss": 1.4425,
|
| 127710 |
+
"step": 18243
|
| 127711 |
+
},
|
| 127712 |
+
{
|
| 127713 |
+
"epoch": 0.48048459309981567,
|
| 127714 |
+
"grad_norm": 1.7621982097625732,
|
| 127715 |
+
"learning_rate": 2.598893863576508e-05,
|
| 127716 |
+
"loss": 1.9327,
|
| 127717 |
+
"step": 18244
|
| 127718 |
+
},
|
| 127719 |
+
{
|
| 127720 |
+
"epoch": 0.48051092968132736,
|
| 127721 |
+
"grad_norm": 3.080350399017334,
|
| 127722 |
+
"learning_rate": 2.5987621806689496e-05,
|
| 127723 |
+
"loss": 0.7348,
|
| 127724 |
+
"step": 18245
|
| 127725 |
+
},
|
| 127726 |
+
{
|
| 127727 |
+
"epoch": 0.4805372662628391,
|
| 127728 |
+
"grad_norm": 2.543154001235962,
|
| 127729 |
+
"learning_rate": 2.5986304977613908e-05,
|
| 127730 |
+
"loss": 1.0225,
|
| 127731 |
+
"step": 18246
|
| 127732 |
+
},
|
| 127733 |
+
{
|
| 127734 |
+
"epoch": 0.4805636028443508,
|
| 127735 |
+
"grad_norm": 2.1923797130584717,
|
| 127736 |
+
"learning_rate": 2.598498814853832e-05,
|
| 127737 |
+
"loss": 2.7072,
|
| 127738 |
+
"step": 18247
|
| 127739 |
+
},
|
| 127740 |
+
{
|
| 127741 |
+
"epoch": 0.48058993942586253,
|
| 127742 |
+
"grad_norm": 1.423647403717041,
|
| 127743 |
+
"learning_rate": 2.5983671319462732e-05,
|
| 127744 |
+
"loss": 1.8857,
|
| 127745 |
+
"step": 18248
|
| 127746 |
+
},
|
| 127747 |
+
{
|
| 127748 |
+
"epoch": 0.4806162760073742,
|
| 127749 |
+
"grad_norm": 2.073836326599121,
|
| 127750 |
+
"learning_rate": 2.598235449038715e-05,
|
| 127751 |
+
"loss": 1.231,
|
| 127752 |
+
"step": 18249
|
| 127753 |
+
},
|
| 127754 |
+
{
|
| 127755 |
+
"epoch": 0.48064261258888596,
|
| 127756 |
+
"grad_norm": 2.9497969150543213,
|
| 127757 |
+
"learning_rate": 2.5981037661311563e-05,
|
| 127758 |
+
"loss": 1.6494,
|
| 127759 |
+
"step": 18250
|
| 127760 |
+
},
|
| 127761 |
+
{
|
| 127762 |
+
"epoch": 0.4806689491703977,
|
| 127763 |
+
"grad_norm": 1.5961450338363647,
|
| 127764 |
+
"learning_rate": 2.5979720832235976e-05,
|
| 127765 |
+
"loss": 2.1089,
|
| 127766 |
+
"step": 18251
|
| 127767 |
+
},
|
| 127768 |
+
{
|
| 127769 |
+
"epoch": 0.4806952857519094,
|
| 127770 |
+
"grad_norm": 1.5951591730117798,
|
| 127771 |
+
"learning_rate": 2.5978404003160388e-05,
|
| 127772 |
+
"loss": 1.7174,
|
| 127773 |
+
"step": 18252
|
| 127774 |
+
},
|
| 127775 |
+
{
|
| 127776 |
+
"epoch": 0.48072162233342114,
|
| 127777 |
+
"grad_norm": 2.378539800643921,
|
| 127778 |
+
"learning_rate": 2.5977087174084807e-05,
|
| 127779 |
+
"loss": 1.7267,
|
| 127780 |
+
"step": 18253
|
| 127781 |
+
},
|
| 127782 |
+
{
|
| 127783 |
+
"epoch": 0.4807479589149328,
|
| 127784 |
+
"grad_norm": 2.2286441326141357,
|
| 127785 |
+
"learning_rate": 2.597577034500922e-05,
|
| 127786 |
+
"loss": 1.2339,
|
| 127787 |
+
"step": 18254
|
| 127788 |
+
},
|
| 127789 |
+
{
|
| 127790 |
+
"epoch": 0.48077429549644457,
|
| 127791 |
+
"grad_norm": 2.03003191947937,
|
| 127792 |
+
"learning_rate": 2.5974453515933635e-05,
|
| 127793 |
+
"loss": 2.0078,
|
| 127794 |
+
"step": 18255
|
| 127795 |
+
},
|
| 127796 |
+
{
|
| 127797 |
+
"epoch": 0.48080063207795626,
|
| 127798 |
+
"grad_norm": 1.9827549457550049,
|
| 127799 |
+
"learning_rate": 2.5973136686858047e-05,
|
| 127800 |
+
"loss": 1.9373,
|
| 127801 |
+
"step": 18256
|
| 127802 |
+
},
|
| 127803 |
+
{
|
| 127804 |
+
"epoch": 0.480826968659468,
|
| 127805 |
+
"grad_norm": 3.3949034214019775,
|
| 127806 |
+
"learning_rate": 2.597181985778246e-05,
|
| 127807 |
+
"loss": 2.0879,
|
| 127808 |
+
"step": 18257
|
| 127809 |
+
},
|
| 127810 |
+
{
|
| 127811 |
+
"epoch": 0.48085330524097974,
|
| 127812 |
+
"grad_norm": 1.7559983730316162,
|
| 127813 |
+
"learning_rate": 2.5970503028706878e-05,
|
| 127814 |
+
"loss": 1.0735,
|
| 127815 |
+
"step": 18258
|
| 127816 |
+
},
|
| 127817 |
+
{
|
| 127818 |
+
"epoch": 0.48087964182249143,
|
| 127819 |
+
"grad_norm": 2.3310892581939697,
|
| 127820 |
+
"learning_rate": 2.596918619963129e-05,
|
| 127821 |
+
"loss": 0.3768,
|
| 127822 |
+
"step": 18259
|
| 127823 |
+
},
|
| 127824 |
+
{
|
| 127825 |
+
"epoch": 0.4809059784040032,
|
| 127826 |
+
"grad_norm": 1.615135908126831,
|
| 127827 |
+
"learning_rate": 2.5967869370555702e-05,
|
| 127828 |
+
"loss": 2.1364,
|
| 127829 |
+
"step": 18260
|
| 127830 |
+
},
|
| 127831 |
+
{
|
| 127832 |
+
"epoch": 0.48093231498551486,
|
| 127833 |
+
"grad_norm": 3.7796401977539062,
|
| 127834 |
+
"learning_rate": 2.5966552541480115e-05,
|
| 127835 |
+
"loss": 1.1534,
|
| 127836 |
+
"step": 18261
|
| 127837 |
+
},
|
| 127838 |
+
{
|
| 127839 |
+
"epoch": 0.4809586515670266,
|
| 127840 |
+
"grad_norm": 2.0392417907714844,
|
| 127841 |
+
"learning_rate": 2.596523571240453e-05,
|
| 127842 |
+
"loss": 1.9045,
|
| 127843 |
+
"step": 18262
|
| 127844 |
+
},
|
| 127845 |
+
{
|
| 127846 |
+
"epoch": 0.4809849881485383,
|
| 127847 |
+
"grad_norm": 6.02974271774292,
|
| 127848 |
+
"learning_rate": 2.5963918883328946e-05,
|
| 127849 |
+
"loss": 0.8431,
|
| 127850 |
+
"step": 18263
|
| 127851 |
+
},
|
| 127852 |
+
{
|
| 127853 |
+
"epoch": 0.48101132473005004,
|
| 127854 |
+
"grad_norm": 1.9574013948440552,
|
| 127855 |
+
"learning_rate": 2.596260205425336e-05,
|
| 127856 |
+
"loss": 2.3259,
|
| 127857 |
+
"step": 18264
|
| 127858 |
+
},
|
| 127859 |
+
{
|
| 127860 |
+
"epoch": 0.4810376613115618,
|
| 127861 |
+
"grad_norm": 4.27133321762085,
|
| 127862 |
+
"learning_rate": 2.5961285225177774e-05,
|
| 127863 |
+
"loss": 0.8533,
|
| 127864 |
+
"step": 18265
|
| 127865 |
+
},
|
| 127866 |
+
{
|
| 127867 |
+
"epoch": 0.48106399789307347,
|
| 127868 |
+
"grad_norm": 2.9611685276031494,
|
| 127869 |
+
"learning_rate": 2.5959968396102186e-05,
|
| 127870 |
+
"loss": 1.4665,
|
| 127871 |
+
"step": 18266
|
| 127872 |
+
},
|
| 127873 |
+
{
|
| 127874 |
+
"epoch": 0.4810903344745852,
|
| 127875 |
+
"grad_norm": 2.1282074451446533,
|
| 127876 |
+
"learning_rate": 2.5958651567026605e-05,
|
| 127877 |
+
"loss": 0.51,
|
| 127878 |
+
"step": 18267
|
| 127879 |
+
},
|
| 127880 |
+
{
|
| 127881 |
+
"epoch": 0.4811166710560969,
|
| 127882 |
+
"grad_norm": 3.044217109680176,
|
| 127883 |
+
"learning_rate": 2.5957334737951017e-05,
|
| 127884 |
+
"loss": 2.7244,
|
| 127885 |
+
"step": 18268
|
| 127886 |
+
},
|
| 127887 |
+
{
|
| 127888 |
+
"epoch": 0.48114300763760864,
|
| 127889 |
+
"grad_norm": 1.874723196029663,
|
| 127890 |
+
"learning_rate": 2.595601790887543e-05,
|
| 127891 |
+
"loss": 1.4811,
|
| 127892 |
+
"step": 18269
|
| 127893 |
+
},
|
| 127894 |
+
{
|
| 127895 |
+
"epoch": 0.48116934421912033,
|
| 127896 |
+
"grad_norm": 1.7602438926696777,
|
| 127897 |
+
"learning_rate": 2.595470107979984e-05,
|
| 127898 |
+
"loss": 1.9281,
|
| 127899 |
+
"step": 18270
|
| 127900 |
+
},
|
| 127901 |
+
{
|
| 127902 |
+
"epoch": 0.4811956808006321,
|
| 127903 |
+
"grad_norm": 2.15421462059021,
|
| 127904 |
+
"learning_rate": 2.5953384250724254e-05,
|
| 127905 |
+
"loss": 2.1792,
|
| 127906 |
+
"step": 18271
|
| 127907 |
+
},
|
| 127908 |
+
{
|
| 127909 |
+
"epoch": 0.4812220173821438,
|
| 127910 |
+
"grad_norm": 3.2429089546203613,
|
| 127911 |
+
"learning_rate": 2.5952067421648673e-05,
|
| 127912 |
+
"loss": 2.1389,
|
| 127913 |
+
"step": 18272
|
| 127914 |
+
},
|
| 127915 |
+
{
|
| 127916 |
+
"epoch": 0.4812483539636555,
|
| 127917 |
+
"grad_norm": 1.878605604171753,
|
| 127918 |
+
"learning_rate": 2.5950750592573085e-05,
|
| 127919 |
+
"loss": 1.7448,
|
| 127920 |
+
"step": 18273
|
| 127921 |
+
},
|
| 127922 |
+
{
|
| 127923 |
+
"epoch": 0.48127469054516725,
|
| 127924 |
+
"grad_norm": 2.79583477973938,
|
| 127925 |
+
"learning_rate": 2.59494337634975e-05,
|
| 127926 |
+
"loss": 1.6102,
|
| 127927 |
+
"step": 18274
|
| 127928 |
+
},
|
| 127929 |
+
{
|
| 127930 |
+
"epoch": 0.48130102712667894,
|
| 127931 |
+
"grad_norm": 2.2260453701019287,
|
| 127932 |
+
"learning_rate": 2.5948116934421913e-05,
|
| 127933 |
+
"loss": 1.8317,
|
| 127934 |
+
"step": 18275
|
| 127935 |
+
},
|
| 127936 |
+
{
|
| 127937 |
+
"epoch": 0.4813273637081907,
|
| 127938 |
+
"grad_norm": 2.351231813430786,
|
| 127939 |
+
"learning_rate": 2.5946800105346325e-05,
|
| 127940 |
+
"loss": 1.7722,
|
| 127941 |
+
"step": 18276
|
| 127942 |
+
},
|
| 127943 |
+
{
|
| 127944 |
+
"epoch": 0.4813537002897024,
|
| 127945 |
+
"grad_norm": 2.5532896518707275,
|
| 127946 |
+
"learning_rate": 2.5945483276270744e-05,
|
| 127947 |
+
"loss": 2.4244,
|
| 127948 |
+
"step": 18277
|
| 127949 |
+
},
|
| 127950 |
+
{
|
| 127951 |
+
"epoch": 0.4813800368712141,
|
| 127952 |
+
"grad_norm": 4.415773391723633,
|
| 127953 |
+
"learning_rate": 2.5944166447195156e-05,
|
| 127954 |
+
"loss": 1.5416,
|
| 127955 |
+
"step": 18278
|
| 127956 |
+
},
|
| 127957 |
+
{
|
| 127958 |
+
"epoch": 0.48140637345272586,
|
| 127959 |
+
"grad_norm": 2.400348663330078,
|
| 127960 |
+
"learning_rate": 2.5942849618119568e-05,
|
| 127961 |
+
"loss": 1.9609,
|
| 127962 |
+
"step": 18279
|
| 127963 |
+
},
|
| 127964 |
+
{
|
| 127965 |
+
"epoch": 0.48143271003423754,
|
| 127966 |
+
"grad_norm": 2.8336586952209473,
|
| 127967 |
+
"learning_rate": 2.594153278904398e-05,
|
| 127968 |
+
"loss": 0.6991,
|
| 127969 |
+
"step": 18280
|
| 127970 |
+
},
|
| 127971 |
+
{
|
| 127972 |
+
"epoch": 0.4814590466157493,
|
| 127973 |
+
"grad_norm": 1.6775230169296265,
|
| 127974 |
+
"learning_rate": 2.5940215959968396e-05,
|
| 127975 |
+
"loss": 1.4768,
|
| 127976 |
+
"step": 18281
|
| 127977 |
+
},
|
| 127978 |
+
{
|
| 127979 |
+
"epoch": 0.481485383197261,
|
| 127980 |
+
"grad_norm": 1.9266937971115112,
|
| 127981 |
+
"learning_rate": 2.593889913089281e-05,
|
| 127982 |
+
"loss": 1.7842,
|
| 127983 |
+
"step": 18282
|
| 127984 |
+
},
|
| 127985 |
+
{
|
| 127986 |
+
"epoch": 0.4815117197787727,
|
| 127987 |
+
"grad_norm": 2.667290210723877,
|
| 127988 |
+
"learning_rate": 2.5937582301817227e-05,
|
| 127989 |
+
"loss": 1.1571,
|
| 127990 |
+
"step": 18283
|
| 127991 |
+
},
|
| 127992 |
+
{
|
| 127993 |
+
"epoch": 0.48153805636028446,
|
| 127994 |
+
"grad_norm": 1.645289421081543,
|
| 127995 |
+
"learning_rate": 2.593626547274164e-05,
|
| 127996 |
+
"loss": 1.5977,
|
| 127997 |
+
"step": 18284
|
| 127998 |
+
},
|
| 127999 |
+
{
|
| 128000 |
+
"epoch": 0.48156439294179615,
|
| 128001 |
+
"grad_norm": 3.0950942039489746,
|
| 128002 |
+
"learning_rate": 2.593494864366605e-05,
|
| 128003 |
+
"loss": 2.1776,
|
| 128004 |
+
"step": 18285
|
| 128005 |
+
},
|
| 128006 |
+
{
|
| 128007 |
+
"epoch": 0.4815907295233079,
|
| 128008 |
+
"grad_norm": 2.822568893432617,
|
| 128009 |
+
"learning_rate": 2.593363181459047e-05,
|
| 128010 |
+
"loss": 1.7076,
|
| 128011 |
+
"step": 18286
|
| 128012 |
+
},
|
| 128013 |
+
{
|
| 128014 |
+
"epoch": 0.4816170661048196,
|
| 128015 |
+
"grad_norm": 1.8029379844665527,
|
| 128016 |
+
"learning_rate": 2.5932314985514883e-05,
|
| 128017 |
+
"loss": 1.6493,
|
| 128018 |
+
"step": 18287
|
| 128019 |
+
},
|
| 128020 |
+
{
|
| 128021 |
+
"epoch": 0.4816434026863313,
|
| 128022 |
+
"grad_norm": 1.8903818130493164,
|
| 128023 |
+
"learning_rate": 2.5930998156439295e-05,
|
| 128024 |
+
"loss": 1.4257,
|
| 128025 |
+
"step": 18288
|
| 128026 |
+
},
|
| 128027 |
+
{
|
| 128028 |
+
"epoch": 0.481669739267843,
|
| 128029 |
+
"grad_norm": 1.7043659687042236,
|
| 128030 |
+
"learning_rate": 2.5929681327363707e-05,
|
| 128031 |
+
"loss": 1.8823,
|
| 128032 |
+
"step": 18289
|
| 128033 |
+
},
|
| 128034 |
+
{
|
| 128035 |
+
"epoch": 0.48169607584935475,
|
| 128036 |
+
"grad_norm": 2.1602940559387207,
|
| 128037 |
+
"learning_rate": 2.5928364498288123e-05,
|
| 128038 |
+
"loss": 1.569,
|
| 128039 |
+
"step": 18290
|
| 128040 |
+
},
|
| 128041 |
+
{
|
| 128042 |
+
"epoch": 0.4817224124308665,
|
| 128043 |
+
"grad_norm": 1.5170824527740479,
|
| 128044 |
+
"learning_rate": 2.592704766921254e-05,
|
| 128045 |
+
"loss": 2.5523,
|
| 128046 |
+
"step": 18291
|
| 128047 |
+
},
|
| 128048 |
+
{
|
| 128049 |
+
"epoch": 0.4817487490123782,
|
| 128050 |
+
"grad_norm": 1.7479232549667358,
|
| 128051 |
+
"learning_rate": 2.5925730840136954e-05,
|
| 128052 |
+
"loss": 2.0913,
|
| 128053 |
+
"step": 18292
|
| 128054 |
+
},
|
| 128055 |
+
{
|
| 128056 |
+
"epoch": 0.48177508559388993,
|
| 128057 |
+
"grad_norm": 2.471151351928711,
|
| 128058 |
+
"learning_rate": 2.5924414011061366e-05,
|
| 128059 |
+
"loss": 1.2299,
|
| 128060 |
+
"step": 18293
|
| 128061 |
+
},
|
| 128062 |
+
{
|
| 128063 |
+
"epoch": 0.4818014221754016,
|
| 128064 |
+
"grad_norm": 2.0613322257995605,
|
| 128065 |
+
"learning_rate": 2.592309718198578e-05,
|
| 128066 |
+
"loss": 0.9024,
|
| 128067 |
+
"step": 18294
|
| 128068 |
+
},
|
| 128069 |
+
{
|
| 128070 |
+
"epoch": 0.48182775875691336,
|
| 128071 |
+
"grad_norm": 2.1863367557525635,
|
| 128072 |
+
"learning_rate": 2.592178035291019e-05,
|
| 128073 |
+
"loss": 1.3717,
|
| 128074 |
+
"step": 18295
|
| 128075 |
+
},
|
| 128076 |
+
{
|
| 128077 |
+
"epoch": 0.48185409533842505,
|
| 128078 |
+
"grad_norm": 1.8875648975372314,
|
| 128079 |
+
"learning_rate": 2.592046352383461e-05,
|
| 128080 |
+
"loss": 1.5818,
|
| 128081 |
+
"step": 18296
|
| 128082 |
+
},
|
| 128083 |
+
{
|
| 128084 |
+
"epoch": 0.4818804319199368,
|
| 128085 |
+
"grad_norm": 3.3052406311035156,
|
| 128086 |
+
"learning_rate": 2.5919146694759022e-05,
|
| 128087 |
+
"loss": 1.8997,
|
| 128088 |
+
"step": 18297
|
| 128089 |
+
},
|
| 128090 |
+
{
|
| 128091 |
+
"epoch": 0.48190676850144853,
|
| 128092 |
+
"grad_norm": 2.7625679969787598,
|
| 128093 |
+
"learning_rate": 2.5917829865683434e-05,
|
| 128094 |
+
"loss": 1.493,
|
| 128095 |
+
"step": 18298
|
| 128096 |
+
},
|
| 128097 |
+
{
|
| 128098 |
+
"epoch": 0.4819331050829602,
|
| 128099 |
+
"grad_norm": 2.534208297729492,
|
| 128100 |
+
"learning_rate": 2.5916513036607846e-05,
|
| 128101 |
+
"loss": 1.599,
|
| 128102 |
+
"step": 18299
|
| 128103 |
+
},
|
| 128104 |
+
{
|
| 128105 |
+
"epoch": 0.48195944166447197,
|
| 128106 |
+
"grad_norm": 4.247109413146973,
|
| 128107 |
+
"learning_rate": 2.5915196207532265e-05,
|
| 128108 |
+
"loss": 1.4251,
|
| 128109 |
+
"step": 18300
|
| 128110 |
+
},
|
| 128111 |
+
{
|
| 128112 |
+
"epoch": 0.48198577824598365,
|
| 128113 |
+
"grad_norm": 2.1316025257110596,
|
| 128114 |
+
"learning_rate": 2.5913879378456678e-05,
|
| 128115 |
+
"loss": 2.5669,
|
| 128116 |
+
"step": 18301
|
| 128117 |
+
},
|
| 128118 |
+
{
|
| 128119 |
+
"epoch": 0.4820121148274954,
|
| 128120 |
+
"grad_norm": 3.536114454269409,
|
| 128121 |
+
"learning_rate": 2.5912562549381093e-05,
|
| 128122 |
+
"loss": 0.5308,
|
| 128123 |
+
"step": 18302
|
| 128124 |
+
},
|
| 128125 |
+
{
|
| 128126 |
+
"epoch": 0.4820384514090071,
|
| 128127 |
+
"grad_norm": 2.2415454387664795,
|
| 128128 |
+
"learning_rate": 2.5911245720305505e-05,
|
| 128129 |
+
"loss": 1.7294,
|
| 128130 |
+
"step": 18303
|
| 128131 |
+
},
|
| 128132 |
+
{
|
| 128133 |
+
"epoch": 0.48206478799051883,
|
| 128134 |
+
"grad_norm": 1.5645477771759033,
|
| 128135 |
+
"learning_rate": 2.5909928891229918e-05,
|
| 128136 |
+
"loss": 1.7197,
|
| 128137 |
+
"step": 18304
|
| 128138 |
+
},
|
| 128139 |
+
{
|
| 128140 |
+
"epoch": 0.48209112457203057,
|
| 128141 |
+
"grad_norm": 1.7059056758880615,
|
| 128142 |
+
"learning_rate": 2.5908612062154337e-05,
|
| 128143 |
+
"loss": 0.9997,
|
| 128144 |
+
"step": 18305
|
| 128145 |
+
},
|
| 128146 |
+
{
|
| 128147 |
+
"epoch": 0.48211746115354226,
|
| 128148 |
+
"grad_norm": 3.156730890274048,
|
| 128149 |
+
"learning_rate": 2.590729523307875e-05,
|
| 128150 |
+
"loss": 0.336,
|
| 128151 |
+
"step": 18306
|
| 128152 |
+
},
|
| 128153 |
+
{
|
| 128154 |
+
"epoch": 0.482143797735054,
|
| 128155 |
+
"grad_norm": 1.3372992277145386,
|
| 128156 |
+
"learning_rate": 2.590597840400316e-05,
|
| 128157 |
+
"loss": 1.411,
|
| 128158 |
+
"step": 18307
|
| 128159 |
+
},
|
| 128160 |
+
{
|
| 128161 |
+
"epoch": 0.4821701343165657,
|
| 128162 |
+
"grad_norm": 2.8801941871643066,
|
| 128163 |
+
"learning_rate": 2.5904661574927573e-05,
|
| 128164 |
+
"loss": 1.6704,
|
| 128165 |
+
"step": 18308
|
| 128166 |
+
},
|
| 128167 |
+
{
|
| 128168 |
+
"epoch": 0.48219647089807743,
|
| 128169 |
+
"grad_norm": 1.8397315740585327,
|
| 128170 |
+
"learning_rate": 2.590334474585199e-05,
|
| 128171 |
+
"loss": 2.1636,
|
| 128172 |
+
"step": 18309
|
| 128173 |
+
},
|
| 128174 |
+
{
|
| 128175 |
+
"epoch": 0.4822228074795891,
|
| 128176 |
+
"grad_norm": 2.7086033821105957,
|
| 128177 |
+
"learning_rate": 2.5902027916776404e-05,
|
| 128178 |
+
"loss": 0.6801,
|
| 128179 |
+
"step": 18310
|
| 128180 |
+
},
|
| 128181 |
+
{
|
| 128182 |
+
"epoch": 0.48224914406110087,
|
| 128183 |
+
"grad_norm": 4.560272693634033,
|
| 128184 |
+
"learning_rate": 2.590071108770082e-05,
|
| 128185 |
+
"loss": 0.6993,
|
| 128186 |
+
"step": 18311
|
| 128187 |
+
},
|
| 128188 |
+
{
|
| 128189 |
+
"epoch": 0.4822754806426126,
|
| 128190 |
+
"grad_norm": 1.794438362121582,
|
| 128191 |
+
"learning_rate": 2.5899394258625232e-05,
|
| 128192 |
+
"loss": 2.3285,
|
| 128193 |
+
"step": 18312
|
| 128194 |
+
},
|
| 128195 |
+
{
|
| 128196 |
+
"epoch": 0.4823018172241243,
|
| 128197 |
+
"grad_norm": 1.6634902954101562,
|
| 128198 |
+
"learning_rate": 2.5898077429549644e-05,
|
| 128199 |
+
"loss": 1.7613,
|
| 128200 |
+
"step": 18313
|
| 128201 |
+
},
|
| 128202 |
+
{
|
| 128203 |
+
"epoch": 0.48232815380563604,
|
| 128204 |
+
"grad_norm": 2.5338780879974365,
|
| 128205 |
+
"learning_rate": 2.5896760600474057e-05,
|
| 128206 |
+
"loss": 0.7794,
|
| 128207 |
+
"step": 18314
|
| 128208 |
+
},
|
| 128209 |
+
{
|
| 128210 |
+
"epoch": 0.4823544903871477,
|
| 128211 |
+
"grad_norm": 1.5597007274627686,
|
| 128212 |
+
"learning_rate": 2.5895443771398476e-05,
|
| 128213 |
+
"loss": 1.9812,
|
| 128214 |
+
"step": 18315
|
| 128215 |
+
},
|
| 128216 |
+
{
|
| 128217 |
+
"epoch": 0.48238082696865947,
|
| 128218 |
+
"grad_norm": 1.5586351156234741,
|
| 128219 |
+
"learning_rate": 2.5894126942322888e-05,
|
| 128220 |
+
"loss": 2.5375,
|
| 128221 |
+
"step": 18316
|
| 128222 |
+
},
|
| 128223 |
+
{
|
| 128224 |
+
"epoch": 0.4824071635501712,
|
| 128225 |
+
"grad_norm": 2.1696550846099854,
|
| 128226 |
+
"learning_rate": 2.58928101132473e-05,
|
| 128227 |
+
"loss": 1.6872,
|
| 128228 |
+
"step": 18317
|
| 128229 |
+
},
|
| 128230 |
+
{
|
| 128231 |
+
"epoch": 0.4824335001316829,
|
| 128232 |
+
"grad_norm": 2.1278188228607178,
|
| 128233 |
+
"learning_rate": 2.5891493284171716e-05,
|
| 128234 |
+
"loss": 1.9642,
|
| 128235 |
+
"step": 18318
|
| 128236 |
+
},
|
| 128237 |
+
{
|
| 128238 |
+
"epoch": 0.48245983671319465,
|
| 128239 |
+
"grad_norm": 1.742224097251892,
|
| 128240 |
+
"learning_rate": 2.589017645509613e-05,
|
| 128241 |
+
"loss": 0.7188,
|
| 128242 |
+
"step": 18319
|
| 128243 |
+
},
|
| 128244 |
+
{
|
| 128245 |
+
"epoch": 0.48248617329470633,
|
| 128246 |
+
"grad_norm": 4.575339317321777,
|
| 128247 |
+
"learning_rate": 2.5888859626020547e-05,
|
| 128248 |
+
"loss": 1.0558,
|
| 128249 |
+
"step": 18320
|
| 128250 |
+
},
|
| 128251 |
+
{
|
| 128252 |
+
"epoch": 0.4825125098762181,
|
| 128253 |
+
"grad_norm": 5.727512359619141,
|
| 128254 |
+
"learning_rate": 2.588754279694496e-05,
|
| 128255 |
+
"loss": 1.3297,
|
| 128256 |
+
"step": 18321
|
| 128257 |
+
},
|
| 128258 |
+
{
|
| 128259 |
+
"epoch": 0.48253884645772976,
|
| 128260 |
+
"grad_norm": 1.7224305868148804,
|
| 128261 |
+
"learning_rate": 2.588622596786937e-05,
|
| 128262 |
+
"loss": 1.606,
|
| 128263 |
+
"step": 18322
|
| 128264 |
+
},
|
| 128265 |
+
{
|
| 128266 |
+
"epoch": 0.4825651830392415,
|
| 128267 |
+
"grad_norm": 1.7859320640563965,
|
| 128268 |
+
"learning_rate": 2.5884909138793783e-05,
|
| 128269 |
+
"loss": 1.5833,
|
| 128270 |
+
"step": 18323
|
| 128271 |
+
},
|
| 128272 |
+
{
|
| 128273 |
+
"epoch": 0.48259151962075325,
|
| 128274 |
+
"grad_norm": 1.3457127809524536,
|
| 128275 |
+
"learning_rate": 2.5883592309718202e-05,
|
| 128276 |
+
"loss": 1.9783,
|
| 128277 |
+
"step": 18324
|
| 128278 |
+
},
|
| 128279 |
+
{
|
| 128280 |
+
"epoch": 0.48261785620226494,
|
| 128281 |
+
"grad_norm": 1.807705044746399,
|
| 128282 |
+
"learning_rate": 2.5882275480642615e-05,
|
| 128283 |
+
"loss": 1.6852,
|
| 128284 |
+
"step": 18325
|
| 128285 |
+
},
|
| 128286 |
+
{
|
| 128287 |
+
"epoch": 0.4826441927837767,
|
| 128288 |
+
"grad_norm": 3.553330421447754,
|
| 128289 |
+
"learning_rate": 2.5880958651567027e-05,
|
| 128290 |
+
"loss": 1.1183,
|
| 128291 |
+
"step": 18326
|
| 128292 |
+
},
|
| 128293 |
+
{
|
| 128294 |
+
"epoch": 0.48267052936528837,
|
| 128295 |
+
"grad_norm": 3.3810200691223145,
|
| 128296 |
+
"learning_rate": 2.587964182249144e-05,
|
| 128297 |
+
"loss": 2.8145,
|
| 128298 |
+
"step": 18327
|
| 128299 |
+
},
|
| 128300 |
+
{
|
| 128301 |
+
"epoch": 0.4826968659468001,
|
| 128302 |
+
"grad_norm": 2.4965574741363525,
|
| 128303 |
+
"learning_rate": 2.5878324993415855e-05,
|
| 128304 |
+
"loss": 1.7972,
|
| 128305 |
+
"step": 18328
|
| 128306 |
+
},
|
| 128307 |
+
{
|
| 128308 |
+
"epoch": 0.4827232025283118,
|
| 128309 |
+
"grad_norm": 2.483815908432007,
|
| 128310 |
+
"learning_rate": 2.587700816434027e-05,
|
| 128311 |
+
"loss": 0.8307,
|
| 128312 |
+
"step": 18329
|
| 128313 |
+
},
|
| 128314 |
+
{
|
| 128315 |
+
"epoch": 0.48274953910982354,
|
| 128316 |
+
"grad_norm": 2.7984135150909424,
|
| 128317 |
+
"learning_rate": 2.5875691335264686e-05,
|
| 128318 |
+
"loss": 2.2607,
|
| 128319 |
+
"step": 18330
|
| 128320 |
+
},
|
| 128321 |
+
{
|
| 128322 |
+
"epoch": 0.4827758756913353,
|
| 128323 |
+
"grad_norm": 1.6728575229644775,
|
| 128324 |
+
"learning_rate": 2.5874374506189098e-05,
|
| 128325 |
+
"loss": 1.8287,
|
| 128326 |
+
"step": 18331
|
| 128327 |
+
},
|
| 128328 |
+
{
|
| 128329 |
+
"epoch": 0.482802212272847,
|
| 128330 |
+
"grad_norm": 2.076826810836792,
|
| 128331 |
+
"learning_rate": 2.587305767711351e-05,
|
| 128332 |
+
"loss": 1.2793,
|
| 128333 |
+
"step": 18332
|
| 128334 |
+
},
|
| 128335 |
+
{
|
| 128336 |
+
"epoch": 0.4828285488543587,
|
| 128337 |
+
"grad_norm": 2.7556533813476562,
|
| 128338 |
+
"learning_rate": 2.587174084803793e-05,
|
| 128339 |
+
"loss": 2.1633,
|
| 128340 |
+
"step": 18333
|
| 128341 |
+
},
|
| 128342 |
+
{
|
| 128343 |
+
"epoch": 0.4828548854358704,
|
| 128344 |
+
"grad_norm": 2.11775803565979,
|
| 128345 |
+
"learning_rate": 2.587042401896234e-05,
|
| 128346 |
+
"loss": 1.8455,
|
| 128347 |
+
"step": 18334
|
| 128348 |
+
},
|
| 128349 |
+
{
|
| 128350 |
+
"epoch": 0.48288122201738215,
|
| 128351 |
+
"grad_norm": 3.899864435195923,
|
| 128352 |
+
"learning_rate": 2.5869107189886754e-05,
|
| 128353 |
+
"loss": 1.2825,
|
| 128354 |
+
"step": 18335
|
| 128355 |
+
},
|
| 128356 |
+
{
|
| 128357 |
+
"epoch": 0.48290755859889384,
|
| 128358 |
+
"grad_norm": 2.0632147789001465,
|
| 128359 |
+
"learning_rate": 2.5867790360811166e-05,
|
| 128360 |
+
"loss": 2.3838,
|
| 128361 |
+
"step": 18336
|
| 128362 |
+
},
|
| 128363 |
+
{
|
| 128364 |
+
"epoch": 0.4829338951804056,
|
| 128365 |
+
"grad_norm": 1.4783732891082764,
|
| 128366 |
+
"learning_rate": 2.586647353173558e-05,
|
| 128367 |
+
"loss": 0.6448,
|
| 128368 |
+
"step": 18337
|
| 128369 |
+
},
|
| 128370 |
+
{
|
| 128371 |
+
"epoch": 0.4829602317619173,
|
| 128372 |
+
"grad_norm": 3.414386510848999,
|
| 128373 |
+
"learning_rate": 2.5865156702659997e-05,
|
| 128374 |
+
"loss": 1.6303,
|
| 128375 |
+
"step": 18338
|
| 128376 |
+
},
|
| 128377 |
+
{
|
| 128378 |
+
"epoch": 0.482986568343429,
|
| 128379 |
+
"grad_norm": 2.6229095458984375,
|
| 128380 |
+
"learning_rate": 2.5863839873584413e-05,
|
| 128381 |
+
"loss": 1.1264,
|
| 128382 |
+
"step": 18339
|
| 128383 |
+
},
|
| 128384 |
+
{
|
| 128385 |
+
"epoch": 0.48301290492494076,
|
| 128386 |
+
"grad_norm": 1.6566879749298096,
|
| 128387 |
+
"learning_rate": 2.5862523044508825e-05,
|
| 128388 |
+
"loss": 1.7613,
|
| 128389 |
+
"step": 18340
|
| 128390 |
+
},
|
| 128391 |
+
{
|
| 128392 |
+
"epoch": 0.48303924150645244,
|
| 128393 |
+
"grad_norm": 2.410510778427124,
|
| 128394 |
+
"learning_rate": 2.5861206215433237e-05,
|
| 128395 |
+
"loss": 0.9519,
|
| 128396 |
+
"step": 18341
|
| 128397 |
+
},
|
| 128398 |
+
{
|
| 128399 |
+
"epoch": 0.4830655780879642,
|
| 128400 |
+
"grad_norm": 1.8161232471466064,
|
| 128401 |
+
"learning_rate": 2.585988938635765e-05,
|
| 128402 |
+
"loss": 1.752,
|
| 128403 |
+
"step": 18342
|
| 128404 |
+
},
|
| 128405 |
+
{
|
| 128406 |
+
"epoch": 0.4830919146694759,
|
| 128407 |
+
"grad_norm": 3.2136988639831543,
|
| 128408 |
+
"learning_rate": 2.5858572557282068e-05,
|
| 128409 |
+
"loss": 1.9701,
|
| 128410 |
+
"step": 18343
|
| 128411 |
+
},
|
| 128412 |
+
{
|
| 128413 |
+
"epoch": 0.4831182512509876,
|
| 128414 |
+
"grad_norm": 1.500807523727417,
|
| 128415 |
+
"learning_rate": 2.585725572820648e-05,
|
| 128416 |
+
"loss": 1.8475,
|
| 128417 |
+
"step": 18344
|
| 128418 |
+
},
|
| 128419 |
+
{
|
| 128420 |
+
"epoch": 0.48314458783249936,
|
| 128421 |
+
"grad_norm": 2.430117607116699,
|
| 128422 |
+
"learning_rate": 2.5855938899130893e-05,
|
| 128423 |
+
"loss": 2.1502,
|
| 128424 |
+
"step": 18345
|
| 128425 |
+
},
|
| 128426 |
+
{
|
| 128427 |
+
"epoch": 0.48317092441401105,
|
| 128428 |
+
"grad_norm": 1.8485294580459595,
|
| 128429 |
+
"learning_rate": 2.5854622070055305e-05,
|
| 128430 |
+
"loss": 2.0114,
|
| 128431 |
+
"step": 18346
|
| 128432 |
+
},
|
| 128433 |
+
{
|
| 128434 |
+
"epoch": 0.4831972609955228,
|
| 128435 |
+
"grad_norm": 3.348314046859741,
|
| 128436 |
+
"learning_rate": 2.585330524097972e-05,
|
| 128437 |
+
"loss": 1.9875,
|
| 128438 |
+
"step": 18347
|
| 128439 |
+
},
|
| 128440 |
+
{
|
| 128441 |
+
"epoch": 0.4832235975770345,
|
| 128442 |
+
"grad_norm": 1.8929311037063599,
|
| 128443 |
+
"learning_rate": 2.585198841190414e-05,
|
| 128444 |
+
"loss": 1.7272,
|
| 128445 |
+
"step": 18348
|
| 128446 |
+
},
|
| 128447 |
+
{
|
| 128448 |
+
"epoch": 0.4832499341585462,
|
| 128449 |
+
"grad_norm": 1.764998197555542,
|
| 128450 |
+
"learning_rate": 2.585067158282855e-05,
|
| 128451 |
+
"loss": 1.7428,
|
| 128452 |
+
"step": 18349
|
| 128453 |
+
},
|
| 128454 |
+
{
|
| 128455 |
+
"epoch": 0.48327627074005797,
|
| 128456 |
+
"grad_norm": 4.091507911682129,
|
| 128457 |
+
"learning_rate": 2.5849354753752964e-05,
|
| 128458 |
+
"loss": 2.518,
|
| 128459 |
+
"step": 18350
|
| 128460 |
+
},
|
| 128461 |
+
{
|
| 128462 |
+
"epoch": 0.48330260732156965,
|
| 128463 |
+
"grad_norm": 3.2935714721679688,
|
| 128464 |
+
"learning_rate": 2.5848037924677376e-05,
|
| 128465 |
+
"loss": 1.1943,
|
| 128466 |
+
"step": 18351
|
| 128467 |
+
},
|
| 128468 |
+
{
|
| 128469 |
+
"epoch": 0.4833289439030814,
|
| 128470 |
+
"grad_norm": 2.3311710357666016,
|
| 128471 |
+
"learning_rate": 2.5846721095601795e-05,
|
| 128472 |
+
"loss": 0.9638,
|
| 128473 |
+
"step": 18352
|
| 128474 |
+
},
|
| 128475 |
+
{
|
| 128476 |
+
"epoch": 0.4833552804845931,
|
| 128477 |
+
"grad_norm": 1.7031148672103882,
|
| 128478 |
+
"learning_rate": 2.5845404266526207e-05,
|
| 128479 |
+
"loss": 2.1818,
|
| 128480 |
+
"step": 18353
|
| 128481 |
+
},
|
| 128482 |
+
{
|
| 128483 |
+
"epoch": 0.48338161706610483,
|
| 128484 |
+
"grad_norm": 3.7465553283691406,
|
| 128485 |
+
"learning_rate": 2.584408743745062e-05,
|
| 128486 |
+
"loss": 1.2991,
|
| 128487 |
+
"step": 18354
|
| 128488 |
+
},
|
| 128489 |
+
{
|
| 128490 |
+
"epoch": 0.4834079536476165,
|
| 128491 |
+
"grad_norm": 3.8707756996154785,
|
| 128492 |
+
"learning_rate": 2.5842770608375032e-05,
|
| 128493 |
+
"loss": 2.0623,
|
| 128494 |
+
"step": 18355
|
| 128495 |
+
},
|
| 128496 |
+
{
|
| 128497 |
+
"epoch": 0.48343429022912826,
|
| 128498 |
+
"grad_norm": 1.9105716943740845,
|
| 128499 |
+
"learning_rate": 2.5841453779299447e-05,
|
| 128500 |
+
"loss": 1.3937,
|
| 128501 |
+
"step": 18356
|
| 128502 |
+
},
|
| 128503 |
+
{
|
| 128504 |
+
"epoch": 0.48346062681064,
|
| 128505 |
+
"grad_norm": 2.868964672088623,
|
| 128506 |
+
"learning_rate": 2.5840136950223863e-05,
|
| 128507 |
+
"loss": 1.844,
|
| 128508 |
+
"step": 18357
|
| 128509 |
+
},
|
| 128510 |
+
{
|
| 128511 |
+
"epoch": 0.4834869633921517,
|
| 128512 |
+
"grad_norm": 2.0943338871002197,
|
| 128513 |
+
"learning_rate": 2.583882012114828e-05,
|
| 128514 |
+
"loss": 1.582,
|
| 128515 |
+
"step": 18358
|
| 128516 |
+
},
|
| 128517 |
+
{
|
| 128518 |
+
"epoch": 0.48351329997366344,
|
| 128519 |
+
"grad_norm": 3.172908067703247,
|
| 128520 |
+
"learning_rate": 2.583750329207269e-05,
|
| 128521 |
+
"loss": 1.6974,
|
| 128522 |
+
"step": 18359
|
| 128523 |
+
},
|
| 128524 |
+
{
|
| 128525 |
+
"epoch": 0.4835396365551751,
|
| 128526 |
+
"grad_norm": 2.3387062549591064,
|
| 128527 |
+
"learning_rate": 2.5836186462997103e-05,
|
| 128528 |
+
"loss": 1.1011,
|
| 128529 |
+
"step": 18360
|
| 128530 |
+
},
|
| 128531 |
+
{
|
| 128532 |
+
"epoch": 0.48356597313668687,
|
| 128533 |
+
"grad_norm": 3.5115859508514404,
|
| 128534 |
+
"learning_rate": 2.5834869633921515e-05,
|
| 128535 |
+
"loss": 1.3581,
|
| 128536 |
+
"step": 18361
|
| 128537 |
+
},
|
| 128538 |
+
{
|
| 128539 |
+
"epoch": 0.48359230971819855,
|
| 128540 |
+
"grad_norm": 2.8621480464935303,
|
| 128541 |
+
"learning_rate": 2.5833552804845934e-05,
|
| 128542 |
+
"loss": 1.1026,
|
| 128543 |
+
"step": 18362
|
| 128544 |
+
},
|
| 128545 |
+
{
|
| 128546 |
+
"epoch": 0.4836186462997103,
|
| 128547 |
+
"grad_norm": 2.5172183513641357,
|
| 128548 |
+
"learning_rate": 2.5832235975770346e-05,
|
| 128549 |
+
"loss": 1.8828,
|
| 128550 |
+
"step": 18363
|
| 128551 |
+
},
|
| 128552 |
+
{
|
| 128553 |
+
"epoch": 0.48364498288122204,
|
| 128554 |
+
"grad_norm": 2.953972816467285,
|
| 128555 |
+
"learning_rate": 2.583091914669476e-05,
|
| 128556 |
+
"loss": 0.8589,
|
| 128557 |
+
"step": 18364
|
| 128558 |
+
},
|
| 128559 |
+
{
|
| 128560 |
+
"epoch": 0.48367131946273373,
|
| 128561 |
+
"grad_norm": 1.8448625802993774,
|
| 128562 |
+
"learning_rate": 2.5829602317619174e-05,
|
| 128563 |
+
"loss": 1.9318,
|
| 128564 |
+
"step": 18365
|
| 128565 |
+
},
|
| 128566 |
+
{
|
| 128567 |
+
"epoch": 0.48369765604424547,
|
| 128568 |
+
"grad_norm": 1.8998172283172607,
|
| 128569 |
+
"learning_rate": 2.582828548854359e-05,
|
| 128570 |
+
"loss": 1.8291,
|
| 128571 |
+
"step": 18366
|
| 128572 |
+
},
|
| 128573 |
+
{
|
| 128574 |
+
"epoch": 0.48372399262575716,
|
| 128575 |
+
"grad_norm": 1.906227469444275,
|
| 128576 |
+
"learning_rate": 2.5826968659468005e-05,
|
| 128577 |
+
"loss": 1.3863,
|
| 128578 |
+
"step": 18367
|
| 128579 |
+
},
|
| 128580 |
+
{
|
| 128581 |
+
"epoch": 0.4837503292072689,
|
| 128582 |
+
"grad_norm": 1.814586877822876,
|
| 128583 |
+
"learning_rate": 2.5825651830392418e-05,
|
| 128584 |
+
"loss": 1.9128,
|
| 128585 |
+
"step": 18368
|
| 128586 |
+
},
|
| 128587 |
+
{
|
| 128588 |
+
"epoch": 0.4837766657887806,
|
| 128589 |
+
"grad_norm": 2.9641284942626953,
|
| 128590 |
+
"learning_rate": 2.582433500131683e-05,
|
| 128591 |
+
"loss": 1.5397,
|
| 128592 |
+
"step": 18369
|
| 128593 |
+
},
|
| 128594 |
+
{
|
| 128595 |
+
"epoch": 0.48380300237029233,
|
| 128596 |
+
"grad_norm": 4.049939155578613,
|
| 128597 |
+
"learning_rate": 2.5823018172241242e-05,
|
| 128598 |
+
"loss": 1.5262,
|
| 128599 |
+
"step": 18370
|
| 128600 |
+
},
|
| 128601 |
+
{
|
| 128602 |
+
"epoch": 0.4838293389518041,
|
| 128603 |
+
"grad_norm": 1.870164155960083,
|
| 128604 |
+
"learning_rate": 2.582170134316566e-05,
|
| 128605 |
+
"loss": 0.9732,
|
| 128606 |
+
"step": 18371
|
| 128607 |
+
},
|
| 128608 |
+
{
|
| 128609 |
+
"epoch": 0.48385567553331577,
|
| 128610 |
+
"grad_norm": 3.756333351135254,
|
| 128611 |
+
"learning_rate": 2.5820384514090073e-05,
|
| 128612 |
+
"loss": 1.1043,
|
| 128613 |
+
"step": 18372
|
| 128614 |
+
},
|
| 128615 |
+
{
|
| 128616 |
+
"epoch": 0.4838820121148275,
|
| 128617 |
+
"grad_norm": 2.0509378910064697,
|
| 128618 |
+
"learning_rate": 2.5819067685014485e-05,
|
| 128619 |
+
"loss": 1.9517,
|
| 128620 |
+
"step": 18373
|
| 128621 |
+
},
|
| 128622 |
+
{
|
| 128623 |
+
"epoch": 0.4839083486963392,
|
| 128624 |
+
"grad_norm": 5.011322975158691,
|
| 128625 |
+
"learning_rate": 2.5817750855938898e-05,
|
| 128626 |
+
"loss": 1.9828,
|
| 128627 |
+
"step": 18374
|
| 128628 |
+
},
|
| 128629 |
+
{
|
| 128630 |
+
"epoch": 0.48393468527785094,
|
| 128631 |
+
"grad_norm": 2.274775743484497,
|
| 128632 |
+
"learning_rate": 2.5816434026863313e-05,
|
| 128633 |
+
"loss": 2.181,
|
| 128634 |
+
"step": 18375
|
| 128635 |
+
},
|
| 128636 |
+
{
|
| 128637 |
+
"epoch": 0.48396102185936263,
|
| 128638 |
+
"grad_norm": 3.4376394748687744,
|
| 128639 |
+
"learning_rate": 2.581511719778773e-05,
|
| 128640 |
+
"loss": 2.1538,
|
| 128641 |
+
"step": 18376
|
| 128642 |
+
},
|
| 128643 |
+
{
|
| 128644 |
+
"epoch": 0.48398735844087437,
|
| 128645 |
+
"grad_norm": 1.577967643737793,
|
| 128646 |
+
"learning_rate": 2.5813800368712144e-05,
|
| 128647 |
+
"loss": 2.4193,
|
| 128648 |
+
"step": 18377
|
| 128649 |
+
},
|
| 128650 |
+
{
|
| 128651 |
+
"epoch": 0.4840136950223861,
|
| 128652 |
+
"grad_norm": 2.75225830078125,
|
| 128653 |
+
"learning_rate": 2.5812483539636557e-05,
|
| 128654 |
+
"loss": 1.4493,
|
| 128655 |
+
"step": 18378
|
| 128656 |
+
},
|
| 128657 |
+
{
|
| 128658 |
+
"epoch": 0.4840400316038978,
|
| 128659 |
+
"grad_norm": 1.5984172821044922,
|
| 128660 |
+
"learning_rate": 2.581116671056097e-05,
|
| 128661 |
+
"loss": 1.8152,
|
| 128662 |
+
"step": 18379
|
| 128663 |
+
},
|
| 128664 |
+
{
|
| 128665 |
+
"epoch": 0.48406636818540955,
|
| 128666 |
+
"grad_norm": 2.0202548503875732,
|
| 128667 |
+
"learning_rate": 2.580984988148538e-05,
|
| 128668 |
+
"loss": 1.726,
|
| 128669 |
+
"step": 18380
|
| 128670 |
+
},
|
| 128671 |
+
{
|
| 128672 |
+
"epoch": 0.48409270476692123,
|
| 128673 |
+
"grad_norm": 1.554030418395996,
|
| 128674 |
+
"learning_rate": 2.58085330524098e-05,
|
| 128675 |
+
"loss": 1.9233,
|
| 128676 |
+
"step": 18381
|
| 128677 |
+
},
|
| 128678 |
+
{
|
| 128679 |
+
"epoch": 0.484119041348433,
|
| 128680 |
+
"grad_norm": 1.8082032203674316,
|
| 128681 |
+
"learning_rate": 2.5807216223334212e-05,
|
| 128682 |
+
"loss": 1.015,
|
| 128683 |
+
"step": 18382
|
| 128684 |
+
},
|
| 128685 |
+
{
|
| 128686 |
+
"epoch": 0.4841453779299447,
|
| 128687 |
+
"grad_norm": 1.779860258102417,
|
| 128688 |
+
"learning_rate": 2.5805899394258624e-05,
|
| 128689 |
+
"loss": 0.5784,
|
| 128690 |
+
"step": 18383
|
| 128691 |
+
},
|
| 128692 |
+
{
|
| 128693 |
+
"epoch": 0.4841717145114564,
|
| 128694 |
+
"grad_norm": 2.2123851776123047,
|
| 128695 |
+
"learning_rate": 2.580458256518304e-05,
|
| 128696 |
+
"loss": 1.8172,
|
| 128697 |
+
"step": 18384
|
| 128698 |
+
},
|
| 128699 |
+
{
|
| 128700 |
+
"epoch": 0.48419805109296815,
|
| 128701 |
+
"grad_norm": 1.8945305347442627,
|
| 128702 |
+
"learning_rate": 2.5803265736107456e-05,
|
| 128703 |
+
"loss": 1.8525,
|
| 128704 |
+
"step": 18385
|
| 128705 |
+
},
|
| 128706 |
+
{
|
| 128707 |
+
"epoch": 0.48422438767447984,
|
| 128708 |
+
"grad_norm": 2.622844934463501,
|
| 128709 |
+
"learning_rate": 2.580194890703187e-05,
|
| 128710 |
+
"loss": 0.7779,
|
| 128711 |
+
"step": 18386
|
| 128712 |
+
},
|
| 128713 |
+
{
|
| 128714 |
+
"epoch": 0.4842507242559916,
|
| 128715 |
+
"grad_norm": 2.0889954566955566,
|
| 128716 |
+
"learning_rate": 2.5800632077956283e-05,
|
| 128717 |
+
"loss": 1.5266,
|
| 128718 |
+
"step": 18387
|
| 128719 |
+
},
|
| 128720 |
+
{
|
| 128721 |
+
"epoch": 0.48427706083750327,
|
| 128722 |
+
"grad_norm": 2.3298048973083496,
|
| 128723 |
+
"learning_rate": 2.5799315248880696e-05,
|
| 128724 |
+
"loss": 2.2032,
|
| 128725 |
+
"step": 18388
|
| 128726 |
+
},
|
| 128727 |
+
{
|
| 128728 |
+
"epoch": 0.484303397419015,
|
| 128729 |
+
"grad_norm": 1.5841577053070068,
|
| 128730 |
+
"learning_rate": 2.5797998419805108e-05,
|
| 128731 |
+
"loss": 1.8769,
|
| 128732 |
+
"step": 18389
|
| 128733 |
+
},
|
| 128734 |
+
{
|
| 128735 |
+
"epoch": 0.48432973400052676,
|
| 128736 |
+
"grad_norm": 3.13581919670105,
|
| 128737 |
+
"learning_rate": 2.5796681590729527e-05,
|
| 128738 |
+
"loss": 1.656,
|
| 128739 |
+
"step": 18390
|
| 128740 |
+
},
|
| 128741 |
+
{
|
| 128742 |
+
"epoch": 0.48435607058203844,
|
| 128743 |
+
"grad_norm": 2.65350341796875,
|
| 128744 |
+
"learning_rate": 2.579536476165394e-05,
|
| 128745 |
+
"loss": 0.3908,
|
| 128746 |
+
"step": 18391
|
| 128747 |
+
},
|
| 128748 |
+
{
|
| 128749 |
+
"epoch": 0.4843824071635502,
|
| 128750 |
+
"grad_norm": 3.9427220821380615,
|
| 128751 |
+
"learning_rate": 2.579404793257835e-05,
|
| 128752 |
+
"loss": 0.5817,
|
| 128753 |
+
"step": 18392
|
| 128754 |
+
},
|
| 128755 |
+
{
|
| 128756 |
+
"epoch": 0.4844087437450619,
|
| 128757 |
+
"grad_norm": 2.438901424407959,
|
| 128758 |
+
"learning_rate": 2.5792731103502767e-05,
|
| 128759 |
+
"loss": 2.3633,
|
| 128760 |
+
"step": 18393
|
| 128761 |
+
},
|
| 128762 |
+
{
|
| 128763 |
+
"epoch": 0.4844350803265736,
|
| 128764 |
+
"grad_norm": 2.4363667964935303,
|
| 128765 |
+
"learning_rate": 2.579141427442718e-05,
|
| 128766 |
+
"loss": 2.085,
|
| 128767 |
+
"step": 18394
|
| 128768 |
+
},
|
| 128769 |
+
{
|
| 128770 |
+
"epoch": 0.4844614169080853,
|
| 128771 |
+
"grad_norm": 1.9361634254455566,
|
| 128772 |
+
"learning_rate": 2.5790097445351598e-05,
|
| 128773 |
+
"loss": 2.1063,
|
| 128774 |
+
"step": 18395
|
| 128775 |
+
},
|
| 128776 |
+
{
|
| 128777 |
+
"epoch": 0.48448775348959705,
|
| 128778 |
+
"grad_norm": 2.282162666320801,
|
| 128779 |
+
"learning_rate": 2.578878061627601e-05,
|
| 128780 |
+
"loss": 0.898,
|
| 128781 |
+
"step": 18396
|
| 128782 |
+
},
|
| 128783 |
+
{
|
| 128784 |
+
"epoch": 0.4845140900711088,
|
| 128785 |
+
"grad_norm": 1.8887823820114136,
|
| 128786 |
+
"learning_rate": 2.5787463787200422e-05,
|
| 128787 |
+
"loss": 1.9128,
|
| 128788 |
+
"step": 18397
|
| 128789 |
+
},
|
| 128790 |
+
{
|
| 128791 |
+
"epoch": 0.4845404266526205,
|
| 128792 |
+
"grad_norm": 2.7833309173583984,
|
| 128793 |
+
"learning_rate": 2.5786146958124835e-05,
|
| 128794 |
+
"loss": 1.4563,
|
| 128795 |
+
"step": 18398
|
| 128796 |
+
},
|
| 128797 |
+
{
|
| 128798 |
+
"epoch": 0.4845667632341322,
|
| 128799 |
+
"grad_norm": 1.7120144367218018,
|
| 128800 |
+
"learning_rate": 2.5784830129049254e-05,
|
| 128801 |
+
"loss": 1.7172,
|
| 128802 |
+
"step": 18399
|
| 128803 |
+
},
|
| 128804 |
+
{
|
| 128805 |
+
"epoch": 0.4845930998156439,
|
| 128806 |
+
"grad_norm": 2.7394728660583496,
|
| 128807 |
+
"learning_rate": 2.5783513299973666e-05,
|
| 128808 |
+
"loss": 1.1691,
|
| 128809 |
+
"step": 18400
|
| 128810 |
+
},
|
| 128811 |
+
{
|
| 128812 |
+
"epoch": 0.48461943639715566,
|
| 128813 |
+
"grad_norm": 2.6390151977539062,
|
| 128814 |
+
"learning_rate": 2.5782196470898078e-05,
|
| 128815 |
+
"loss": 0.4177,
|
| 128816 |
+
"step": 18401
|
| 128817 |
+
},
|
| 128818 |
+
{
|
| 128819 |
+
"epoch": 0.48464577297866734,
|
| 128820 |
+
"grad_norm": 1.8348826169967651,
|
| 128821 |
+
"learning_rate": 2.578087964182249e-05,
|
| 128822 |
+
"loss": 3.2067,
|
| 128823 |
+
"step": 18402
|
| 128824 |
+
},
|
| 128825 |
+
{
|
| 128826 |
+
"epoch": 0.4846721095601791,
|
| 128827 |
+
"grad_norm": 2.5606539249420166,
|
| 128828 |
+
"learning_rate": 2.5779562812746906e-05,
|
| 128829 |
+
"loss": 2.3002,
|
| 128830 |
+
"step": 18403
|
| 128831 |
+
},
|
| 128832 |
+
{
|
| 128833 |
+
"epoch": 0.48469844614169083,
|
| 128834 |
+
"grad_norm": 2.3585174083709717,
|
| 128835 |
+
"learning_rate": 2.577824598367132e-05,
|
| 128836 |
+
"loss": 1.7013,
|
| 128837 |
+
"step": 18404
|
| 128838 |
+
},
|
| 128839 |
+
{
|
| 128840 |
+
"epoch": 0.4847247827232025,
|
| 128841 |
+
"grad_norm": 1.8755167722702026,
|
| 128842 |
+
"learning_rate": 2.5776929154595737e-05,
|
| 128843 |
+
"loss": 1.9228,
|
| 128844 |
+
"step": 18405
|
| 128845 |
+
},
|
| 128846 |
+
{
|
| 128847 |
+
"epoch": 0.48475111930471426,
|
| 128848 |
+
"grad_norm": 1.3895341157913208,
|
| 128849 |
+
"learning_rate": 2.577561232552015e-05,
|
| 128850 |
+
"loss": 1.8269,
|
| 128851 |
+
"step": 18406
|
| 128852 |
+
},
|
| 128853 |
+
{
|
| 128854 |
+
"epoch": 0.48477745588622595,
|
| 128855 |
+
"grad_norm": 1.6985033750534058,
|
| 128856 |
+
"learning_rate": 2.577429549644456e-05,
|
| 128857 |
+
"loss": 1.8735,
|
| 128858 |
+
"step": 18407
|
| 128859 |
+
},
|
| 128860 |
+
{
|
| 128861 |
+
"epoch": 0.4848037924677377,
|
| 128862 |
+
"grad_norm": 3.8967151641845703,
|
| 128863 |
+
"learning_rate": 2.5772978667368974e-05,
|
| 128864 |
+
"loss": 0.7406,
|
| 128865 |
+
"step": 18408
|
| 128866 |
+
},
|
| 128867 |
+
{
|
| 128868 |
+
"epoch": 0.4848301290492494,
|
| 128869 |
+
"grad_norm": 4.177510738372803,
|
| 128870 |
+
"learning_rate": 2.5771661838293393e-05,
|
| 128871 |
+
"loss": 1.4695,
|
| 128872 |
+
"step": 18409
|
| 128873 |
+
},
|
| 128874 |
+
{
|
| 128875 |
+
"epoch": 0.4848564656307611,
|
| 128876 |
+
"grad_norm": 1.816548228263855,
|
| 128877 |
+
"learning_rate": 2.5770345009217805e-05,
|
| 128878 |
+
"loss": 1.8405,
|
| 128879 |
+
"step": 18410
|
| 128880 |
+
},
|
| 128881 |
+
{
|
| 128882 |
+
"epoch": 0.48488280221227287,
|
| 128883 |
+
"grad_norm": 2.2118592262268066,
|
| 128884 |
+
"learning_rate": 2.5769028180142217e-05,
|
| 128885 |
+
"loss": 1.6375,
|
| 128886 |
+
"step": 18411
|
| 128887 |
+
},
|
| 128888 |
+
{
|
| 128889 |
+
"epoch": 0.48490913879378456,
|
| 128890 |
+
"grad_norm": 1.968703269958496,
|
| 128891 |
+
"learning_rate": 2.5767711351066633e-05,
|
| 128892 |
+
"loss": 1.5512,
|
| 128893 |
+
"step": 18412
|
| 128894 |
+
},
|
| 128895 |
+
{
|
| 128896 |
+
"epoch": 0.4849354753752963,
|
| 128897 |
+
"grad_norm": 2.9055683612823486,
|
| 128898 |
+
"learning_rate": 2.5766394521991045e-05,
|
| 128899 |
+
"loss": 1.5991,
|
| 128900 |
+
"step": 18413
|
| 128901 |
+
},
|
| 128902 |
+
{
|
| 128903 |
+
"epoch": 0.484961811956808,
|
| 128904 |
+
"grad_norm": 2.03057861328125,
|
| 128905 |
+
"learning_rate": 2.5765077692915464e-05,
|
| 128906 |
+
"loss": 0.9919,
|
| 128907 |
+
"step": 18414
|
| 128908 |
+
},
|
| 128909 |
+
{
|
| 128910 |
+
"epoch": 0.48498814853831973,
|
| 128911 |
+
"grad_norm": 1.6863213777542114,
|
| 128912 |
+
"learning_rate": 2.5763760863839876e-05,
|
| 128913 |
+
"loss": 1.9363,
|
| 128914 |
+
"step": 18415
|
| 128915 |
+
},
|
| 128916 |
+
{
|
| 128917 |
+
"epoch": 0.4850144851198315,
|
| 128918 |
+
"grad_norm": 2.704174041748047,
|
| 128919 |
+
"learning_rate": 2.5762444034764288e-05,
|
| 128920 |
+
"loss": 1.5539,
|
| 128921 |
+
"step": 18416
|
| 128922 |
+
},
|
| 128923 |
+
{
|
| 128924 |
+
"epoch": 0.48504082170134316,
|
| 128925 |
+
"grad_norm": 6.385892391204834,
|
| 128926 |
+
"learning_rate": 2.57611272056887e-05,
|
| 128927 |
+
"loss": 1.2474,
|
| 128928 |
+
"step": 18417
|
| 128929 |
+
},
|
| 128930 |
+
{
|
| 128931 |
+
"epoch": 0.4850671582828549,
|
| 128932 |
+
"grad_norm": 1.6319364309310913,
|
| 128933 |
+
"learning_rate": 2.575981037661312e-05,
|
| 128934 |
+
"loss": 2.1538,
|
| 128935 |
+
"step": 18418
|
| 128936 |
+
},
|
| 128937 |
+
{
|
| 128938 |
+
"epoch": 0.4850934948643666,
|
| 128939 |
+
"grad_norm": 2.219048261642456,
|
| 128940 |
+
"learning_rate": 2.5758493547537532e-05,
|
| 128941 |
+
"loss": 1.7055,
|
| 128942 |
+
"step": 18419
|
| 128943 |
+
},
|
| 128944 |
+
{
|
| 128945 |
+
"epoch": 0.48511983144587834,
|
| 128946 |
+
"grad_norm": 1.4454690217971802,
|
| 128947 |
+
"learning_rate": 2.5757176718461944e-05,
|
| 128948 |
+
"loss": 2.0454,
|
| 128949 |
+
"step": 18420
|
| 128950 |
+
},
|
| 128951 |
+
{
|
| 128952 |
+
"epoch": 0.48514616802739,
|
| 128953 |
+
"grad_norm": 1.5924925804138184,
|
| 128954 |
+
"learning_rate": 2.575585988938636e-05,
|
| 128955 |
+
"loss": 1.7619,
|
| 128956 |
+
"step": 18421
|
| 128957 |
+
},
|
| 128958 |
+
{
|
| 128959 |
+
"epoch": 0.48517250460890177,
|
| 128960 |
+
"grad_norm": 1.780044436454773,
|
| 128961 |
+
"learning_rate": 2.5754543060310772e-05,
|
| 128962 |
+
"loss": 1.9857,
|
| 128963 |
+
"step": 18422
|
| 128964 |
+
},
|
| 128965 |
+
{
|
| 128966 |
+
"epoch": 0.4851988411904135,
|
| 128967 |
+
"grad_norm": 2.2101314067840576,
|
| 128968 |
+
"learning_rate": 2.575322623123519e-05,
|
| 128969 |
+
"loss": 1.9862,
|
| 128970 |
+
"step": 18423
|
| 128971 |
+
},
|
| 128972 |
+
{
|
| 128973 |
+
"epoch": 0.4852251777719252,
|
| 128974 |
+
"grad_norm": 2.3404316902160645,
|
| 128975 |
+
"learning_rate": 2.5751909402159603e-05,
|
| 128976 |
+
"loss": 1.8558,
|
| 128977 |
+
"step": 18424
|
| 128978 |
+
},
|
| 128979 |
+
{
|
| 128980 |
+
"epoch": 0.48525151435343694,
|
| 128981 |
+
"grad_norm": 2.050753355026245,
|
| 128982 |
+
"learning_rate": 2.5750592573084015e-05,
|
| 128983 |
+
"loss": 1.8381,
|
| 128984 |
+
"step": 18425
|
| 128985 |
+
},
|
| 128986 |
+
{
|
| 128987 |
+
"epoch": 0.48527785093494863,
|
| 128988 |
+
"grad_norm": 3.0578525066375732,
|
| 128989 |
+
"learning_rate": 2.5749275744008427e-05,
|
| 128990 |
+
"loss": 0.5958,
|
| 128991 |
+
"step": 18426
|
| 128992 |
+
},
|
| 128993 |
+
{
|
| 128994 |
+
"epoch": 0.4853041875164604,
|
| 128995 |
+
"grad_norm": 1.8695067167282104,
|
| 128996 |
+
"learning_rate": 2.574795891493284e-05,
|
| 128997 |
+
"loss": 1.622,
|
| 128998 |
+
"step": 18427
|
| 128999 |
+
},
|
| 129000 |
+
{
|
| 129001 |
+
"epoch": 0.48533052409797206,
|
| 129002 |
+
"grad_norm": 2.1240391731262207,
|
| 129003 |
+
"learning_rate": 2.574664208585726e-05,
|
| 129004 |
+
"loss": 1.2508,
|
| 129005 |
+
"step": 18428
|
| 129006 |
+
},
|
| 129007 |
+
{
|
| 129008 |
+
"epoch": 0.4853568606794838,
|
| 129009 |
+
"grad_norm": 2.0256903171539307,
|
| 129010 |
+
"learning_rate": 2.574532525678167e-05,
|
| 129011 |
+
"loss": 1.4946,
|
| 129012 |
+
"step": 18429
|
| 129013 |
+
},
|
| 129014 |
+
{
|
| 129015 |
+
"epoch": 0.48538319726099555,
|
| 129016 |
+
"grad_norm": 2.7355430126190186,
|
| 129017 |
+
"learning_rate": 2.5744008427706083e-05,
|
| 129018 |
+
"loss": 2.0422,
|
| 129019 |
+
"step": 18430
|
| 129020 |
+
},
|
| 129021 |
+
{
|
| 129022 |
+
"epoch": 0.48540953384250723,
|
| 129023 |
+
"grad_norm": 1.7283477783203125,
|
| 129024 |
+
"learning_rate": 2.57426915986305e-05,
|
| 129025 |
+
"loss": 1.4768,
|
| 129026 |
+
"step": 18431
|
| 129027 |
+
},
|
| 129028 |
+
{
|
| 129029 |
+
"epoch": 0.485435870424019,
|
| 129030 |
+
"grad_norm": 3.126765727996826,
|
| 129031 |
+
"learning_rate": 2.5741374769554914e-05,
|
| 129032 |
+
"loss": 1.2515,
|
| 129033 |
+
"step": 18432
|
| 129034 |
+
},
|
| 129035 |
+
{
|
| 129036 |
+
"epoch": 0.48546220700553067,
|
| 129037 |
+
"grad_norm": 3.3315203189849854,
|
| 129038 |
+
"learning_rate": 2.574005794047933e-05,
|
| 129039 |
+
"loss": 2.1508,
|
| 129040 |
+
"step": 18433
|
| 129041 |
+
},
|
| 129042 |
+
{
|
| 129043 |
+
"epoch": 0.4854885435870424,
|
| 129044 |
+
"grad_norm": 1.8240033388137817,
|
| 129045 |
+
"learning_rate": 2.5738741111403742e-05,
|
| 129046 |
+
"loss": 1.3215,
|
| 129047 |
+
"step": 18434
|
| 129048 |
+
},
|
| 129049 |
+
{
|
| 129050 |
+
"epoch": 0.4855148801685541,
|
| 129051 |
+
"grad_norm": 1.2926433086395264,
|
| 129052 |
+
"learning_rate": 2.5737424282328154e-05,
|
| 129053 |
+
"loss": 1.5452,
|
| 129054 |
+
"step": 18435
|
| 129055 |
+
},
|
| 129056 |
+
{
|
| 129057 |
+
"epoch": 0.48554121675006584,
|
| 129058 |
+
"grad_norm": 3.809260845184326,
|
| 129059 |
+
"learning_rate": 2.5736107453252566e-05,
|
| 129060 |
+
"loss": 1.4537,
|
| 129061 |
+
"step": 18436
|
| 129062 |
+
},
|
| 129063 |
+
{
|
| 129064 |
+
"epoch": 0.4855675533315776,
|
| 129065 |
+
"grad_norm": 2.7715842723846436,
|
| 129066 |
+
"learning_rate": 2.5734790624176985e-05,
|
| 129067 |
+
"loss": 0.8447,
|
| 129068 |
+
"step": 18437
|
| 129069 |
+
},
|
| 129070 |
+
{
|
| 129071 |
+
"epoch": 0.48559388991308927,
|
| 129072 |
+
"grad_norm": 5.419256687164307,
|
| 129073 |
+
"learning_rate": 2.5733473795101398e-05,
|
| 129074 |
+
"loss": 2.0753,
|
| 129075 |
+
"step": 18438
|
| 129076 |
+
},
|
| 129077 |
+
{
|
| 129078 |
+
"epoch": 0.485620226494601,
|
| 129079 |
+
"grad_norm": 1.581094741821289,
|
| 129080 |
+
"learning_rate": 2.573215696602581e-05,
|
| 129081 |
+
"loss": 1.4794,
|
| 129082 |
+
"step": 18439
|
| 129083 |
+
},
|
| 129084 |
+
{
|
| 129085 |
+
"epoch": 0.4856465630761127,
|
| 129086 |
+
"grad_norm": 2.4569525718688965,
|
| 129087 |
+
"learning_rate": 2.5730840136950225e-05,
|
| 129088 |
+
"loss": 1.0419,
|
| 129089 |
+
"step": 18440
|
| 129090 |
+
},
|
| 129091 |
+
{
|
| 129092 |
+
"epoch": 0.48567289965762445,
|
| 129093 |
+
"grad_norm": 2.8132851123809814,
|
| 129094 |
+
"learning_rate": 2.5729523307874638e-05,
|
| 129095 |
+
"loss": 1.2611,
|
| 129096 |
+
"step": 18441
|
| 129097 |
+
},
|
| 129098 |
+
{
|
| 129099 |
+
"epoch": 0.48569923623913613,
|
| 129100 |
+
"grad_norm": 3.084397315979004,
|
| 129101 |
+
"learning_rate": 2.5728206478799057e-05,
|
| 129102 |
+
"loss": 2.0227,
|
| 129103 |
+
"step": 18442
|
| 129104 |
+
},
|
| 129105 |
+
{
|
| 129106 |
+
"epoch": 0.4857255728206479,
|
| 129107 |
+
"grad_norm": 3.458346128463745,
|
| 129108 |
+
"learning_rate": 2.572688964972347e-05,
|
| 129109 |
+
"loss": 2.8425,
|
| 129110 |
+
"step": 18443
|
| 129111 |
+
},
|
| 129112 |
+
{
|
| 129113 |
+
"epoch": 0.4857519094021596,
|
| 129114 |
+
"grad_norm": 2.111849069595337,
|
| 129115 |
+
"learning_rate": 2.572557282064788e-05,
|
| 129116 |
+
"loss": 1.9156,
|
| 129117 |
+
"step": 18444
|
| 129118 |
+
},
|
| 129119 |
+
{
|
| 129120 |
+
"epoch": 0.4857782459836713,
|
| 129121 |
+
"grad_norm": 1.8623846769332886,
|
| 129122 |
+
"learning_rate": 2.5724255991572293e-05,
|
| 129123 |
+
"loss": 2.0972,
|
| 129124 |
+
"step": 18445
|
| 129125 |
+
},
|
| 129126 |
+
{
|
| 129127 |
+
"epoch": 0.48580458256518305,
|
| 129128 |
+
"grad_norm": 2.8974735736846924,
|
| 129129 |
+
"learning_rate": 2.5722939162496705e-05,
|
| 129130 |
+
"loss": 1.4019,
|
| 129131 |
+
"step": 18446
|
| 129132 |
+
},
|
| 129133 |
+
{
|
| 129134 |
+
"epoch": 0.48583091914669474,
|
| 129135 |
+
"grad_norm": 1.8519647121429443,
|
| 129136 |
+
"learning_rate": 2.5721622333421124e-05,
|
| 129137 |
+
"loss": 2.0956,
|
| 129138 |
+
"step": 18447
|
| 129139 |
+
},
|
| 129140 |
+
{
|
| 129141 |
+
"epoch": 0.4858572557282065,
|
| 129142 |
+
"grad_norm": 3.604668617248535,
|
| 129143 |
+
"learning_rate": 2.5720305504345537e-05,
|
| 129144 |
+
"loss": 0.9024,
|
| 129145 |
+
"step": 18448
|
| 129146 |
+
},
|
| 129147 |
+
{
|
| 129148 |
+
"epoch": 0.4858835923097182,
|
| 129149 |
+
"grad_norm": 2.3852856159210205,
|
| 129150 |
+
"learning_rate": 2.571898867526995e-05,
|
| 129151 |
+
"loss": 1.6787,
|
| 129152 |
+
"step": 18449
|
| 129153 |
+
},
|
| 129154 |
+
{
|
| 129155 |
+
"epoch": 0.4859099288912299,
|
| 129156 |
+
"grad_norm": 1.5685434341430664,
|
| 129157 |
+
"learning_rate": 2.5717671846194364e-05,
|
| 129158 |
+
"loss": 1.896,
|
| 129159 |
+
"step": 18450
|
| 129160 |
+
},
|
| 129161 |
+
{
|
| 129162 |
+
"epoch": 0.48593626547274166,
|
| 129163 |
+
"grad_norm": 3.2516274452209473,
|
| 129164 |
+
"learning_rate": 2.571635501711878e-05,
|
| 129165 |
+
"loss": 1.9251,
|
| 129166 |
+
"step": 18451
|
| 129167 |
+
},
|
| 129168 |
+
{
|
| 129169 |
+
"epoch": 0.48596260205425335,
|
| 129170 |
+
"grad_norm": 1.8900718688964844,
|
| 129171 |
+
"learning_rate": 2.5715038188043196e-05,
|
| 129172 |
+
"loss": 1.4476,
|
| 129173 |
+
"step": 18452
|
| 129174 |
+
},
|
| 129175 |
+
{
|
| 129176 |
+
"epoch": 0.4859889386357651,
|
| 129177 |
+
"grad_norm": 3.0719430446624756,
|
| 129178 |
+
"learning_rate": 2.5713721358967608e-05,
|
| 129179 |
+
"loss": 1.9875,
|
| 129180 |
+
"step": 18453
|
| 129181 |
+
},
|
| 129182 |
+
{
|
| 129183 |
+
"epoch": 0.4860152752172768,
|
| 129184 |
+
"grad_norm": 2.0585649013519287,
|
| 129185 |
+
"learning_rate": 2.571240452989202e-05,
|
| 129186 |
+
"loss": 0.7618,
|
| 129187 |
+
"step": 18454
|
| 129188 |
+
},
|
| 129189 |
+
{
|
| 129190 |
+
"epoch": 0.4860416117987885,
|
| 129191 |
+
"grad_norm": 1.8559067249298096,
|
| 129192 |
+
"learning_rate": 2.5711087700816432e-05,
|
| 129193 |
+
"loss": 1.5609,
|
| 129194 |
+
"step": 18455
|
| 129195 |
+
},
|
| 129196 |
+
{
|
| 129197 |
+
"epoch": 0.48606794838030026,
|
| 129198 |
+
"grad_norm": 2.169935941696167,
|
| 129199 |
+
"learning_rate": 2.570977087174085e-05,
|
| 129200 |
+
"loss": 2.1225,
|
| 129201 |
+
"step": 18456
|
| 129202 |
+
},
|
| 129203 |
+
{
|
| 129204 |
+
"epoch": 0.48609428496181195,
|
| 129205 |
+
"grad_norm": 2.295045852661133,
|
| 129206 |
+
"learning_rate": 2.5708454042665263e-05,
|
| 129207 |
+
"loss": 1.6724,
|
| 129208 |
+
"step": 18457
|
| 129209 |
+
},
|
| 129210 |
+
{
|
| 129211 |
+
"epoch": 0.4861206215433237,
|
| 129212 |
+
"grad_norm": 1.7074811458587646,
|
| 129213 |
+
"learning_rate": 2.5707137213589676e-05,
|
| 129214 |
+
"loss": 1.6424,
|
| 129215 |
+
"step": 18458
|
| 129216 |
+
},
|
| 129217 |
+
{
|
| 129218 |
+
"epoch": 0.4861469581248354,
|
| 129219 |
+
"grad_norm": 4.542028903961182,
|
| 129220 |
+
"learning_rate": 2.570582038451409e-05,
|
| 129221 |
+
"loss": 1.4893,
|
| 129222 |
+
"step": 18459
|
| 129223 |
+
},
|
| 129224 |
+
{
|
| 129225 |
+
"epoch": 0.4861732947063471,
|
| 129226 |
+
"grad_norm": 1.9196478128433228,
|
| 129227 |
+
"learning_rate": 2.5704503555438503e-05,
|
| 129228 |
+
"loss": 1.4574,
|
| 129229 |
+
"step": 18460
|
| 129230 |
+
},
|
| 129231 |
+
{
|
| 129232 |
+
"epoch": 0.4861996312878588,
|
| 129233 |
+
"grad_norm": 2.9205195903778076,
|
| 129234 |
+
"learning_rate": 2.5703186726362922e-05,
|
| 129235 |
+
"loss": 1.4391,
|
| 129236 |
+
"step": 18461
|
| 129237 |
+
},
|
| 129238 |
+
{
|
| 129239 |
+
"epoch": 0.48622596786937056,
|
| 129240 |
+
"grad_norm": 1.8574875593185425,
|
| 129241 |
+
"learning_rate": 2.5701869897287335e-05,
|
| 129242 |
+
"loss": 2.1417,
|
| 129243 |
+
"step": 18462
|
| 129244 |
+
},
|
| 129245 |
+
{
|
| 129246 |
+
"epoch": 0.4862523044508823,
|
| 129247 |
+
"grad_norm": 2.81107234954834,
|
| 129248 |
+
"learning_rate": 2.5700553068211747e-05,
|
| 129249 |
+
"loss": 2.5147,
|
| 129250 |
+
"step": 18463
|
| 129251 |
+
},
|
| 129252 |
+
{
|
| 129253 |
+
"epoch": 0.486278641032394,
|
| 129254 |
+
"grad_norm": 2.6533210277557373,
|
| 129255 |
+
"learning_rate": 2.569923623913616e-05,
|
| 129256 |
+
"loss": 1.7436,
|
| 129257 |
+
"step": 18464
|
| 129258 |
+
},
|
| 129259 |
+
{
|
| 129260 |
+
"epoch": 0.48630497761390573,
|
| 129261 |
+
"grad_norm": 4.165168762207031,
|
| 129262 |
+
"learning_rate": 2.5697919410060578e-05,
|
| 129263 |
+
"loss": 2.0054,
|
| 129264 |
+
"step": 18465
|
| 129265 |
+
},
|
| 129266 |
+
{
|
| 129267 |
+
"epoch": 0.4863313141954174,
|
| 129268 |
+
"grad_norm": 1.9256856441497803,
|
| 129269 |
+
"learning_rate": 2.569660258098499e-05,
|
| 129270 |
+
"loss": 1.9637,
|
| 129271 |
+
"step": 18466
|
| 129272 |
+
},
|
| 129273 |
+
{
|
| 129274 |
+
"epoch": 0.48635765077692916,
|
| 129275 |
+
"grad_norm": 1.7320513725280762,
|
| 129276 |
+
"learning_rate": 2.5695285751909402e-05,
|
| 129277 |
+
"loss": 1.8098,
|
| 129278 |
+
"step": 18467
|
| 129279 |
+
},
|
| 129280 |
+
{
|
| 129281 |
+
"epoch": 0.48638398735844085,
|
| 129282 |
+
"grad_norm": 2.1129705905914307,
|
| 129283 |
+
"learning_rate": 2.5693968922833818e-05,
|
| 129284 |
+
"loss": 1.1088,
|
| 129285 |
+
"step": 18468
|
| 129286 |
+
},
|
| 129287 |
+
{
|
| 129288 |
+
"epoch": 0.4864103239399526,
|
| 129289 |
+
"grad_norm": 5.155686378479004,
|
| 129290 |
+
"learning_rate": 2.569265209375823e-05,
|
| 129291 |
+
"loss": 1.3488,
|
| 129292 |
+
"step": 18469
|
| 129293 |
+
},
|
| 129294 |
+
{
|
| 129295 |
+
"epoch": 0.48643666052146434,
|
| 129296 |
+
"grad_norm": 3.6876680850982666,
|
| 129297 |
+
"learning_rate": 2.569133526468265e-05,
|
| 129298 |
+
"loss": 1.6031,
|
| 129299 |
+
"step": 18470
|
| 129300 |
+
},
|
| 129301 |
+
{
|
| 129302 |
+
"epoch": 0.486462997102976,
|
| 129303 |
+
"grad_norm": 1.8359041213989258,
|
| 129304 |
+
"learning_rate": 2.569001843560706e-05,
|
| 129305 |
+
"loss": 1.4716,
|
| 129306 |
+
"step": 18471
|
| 129307 |
+
},
|
| 129308 |
+
{
|
| 129309 |
+
"epoch": 0.48648933368448777,
|
| 129310 |
+
"grad_norm": 2.9589924812316895,
|
| 129311 |
+
"learning_rate": 2.5688701606531474e-05,
|
| 129312 |
+
"loss": 0.6421,
|
| 129313 |
+
"step": 18472
|
| 129314 |
+
},
|
| 129315 |
+
{
|
| 129316 |
+
"epoch": 0.48651567026599946,
|
| 129317 |
+
"grad_norm": 1.6885493993759155,
|
| 129318 |
+
"learning_rate": 2.5687384777455886e-05,
|
| 129319 |
+
"loss": 2.1624,
|
| 129320 |
+
"step": 18473
|
| 129321 |
+
},
|
| 129322 |
+
{
|
| 129323 |
+
"epoch": 0.4865420068475112,
|
| 129324 |
+
"grad_norm": 3.8228204250335693,
|
| 129325 |
+
"learning_rate": 2.5686067948380298e-05,
|
| 129326 |
+
"loss": 1.9418,
|
| 129327 |
+
"step": 18474
|
| 129328 |
+
},
|
| 129329 |
+
{
|
| 129330 |
+
"epoch": 0.4865683434290229,
|
| 129331 |
+
"grad_norm": 3.8486132621765137,
|
| 129332 |
+
"learning_rate": 2.5684751119304717e-05,
|
| 129333 |
+
"loss": 1.9678,
|
| 129334 |
+
"step": 18475
|
| 129335 |
+
},
|
| 129336 |
+
{
|
| 129337 |
+
"epoch": 0.48659468001053463,
|
| 129338 |
+
"grad_norm": 2.923513412475586,
|
| 129339 |
+
"learning_rate": 2.568343429022913e-05,
|
| 129340 |
+
"loss": 1.6008,
|
| 129341 |
+
"step": 18476
|
| 129342 |
+
},
|
| 129343 |
+
{
|
| 129344 |
+
"epoch": 0.4866210165920464,
|
| 129345 |
+
"grad_norm": 2.0346717834472656,
|
| 129346 |
+
"learning_rate": 2.568211746115354e-05,
|
| 129347 |
+
"loss": 1.6052,
|
| 129348 |
+
"step": 18477
|
| 129349 |
+
},
|
| 129350 |
+
{
|
| 129351 |
+
"epoch": 0.48664735317355806,
|
| 129352 |
+
"grad_norm": 4.474893569946289,
|
| 129353 |
+
"learning_rate": 2.5680800632077957e-05,
|
| 129354 |
+
"loss": 1.6256,
|
| 129355 |
+
"step": 18478
|
| 129356 |
+
},
|
| 129357 |
+
{
|
| 129358 |
+
"epoch": 0.4866736897550698,
|
| 129359 |
+
"grad_norm": 4.485257625579834,
|
| 129360 |
+
"learning_rate": 2.567948380300237e-05,
|
| 129361 |
+
"loss": 1.468,
|
| 129362 |
+
"step": 18479
|
| 129363 |
+
},
|
| 129364 |
+
{
|
| 129365 |
+
"epoch": 0.4867000263365815,
|
| 129366 |
+
"grad_norm": 2.5774567127227783,
|
| 129367 |
+
"learning_rate": 2.5678166973926788e-05,
|
| 129368 |
+
"loss": 1.6469,
|
| 129369 |
+
"step": 18480
|
| 129370 |
+
},
|
| 129371 |
+
{
|
| 129372 |
+
"epoch": 0.48672636291809324,
|
| 129373 |
+
"grad_norm": 2.047030448913574,
|
| 129374 |
+
"learning_rate": 2.56768501448512e-05,
|
| 129375 |
+
"loss": 2.1457,
|
| 129376 |
+
"step": 18481
|
| 129377 |
+
},
|
| 129378 |
+
{
|
| 129379 |
+
"epoch": 0.4867526994996049,
|
| 129380 |
+
"grad_norm": 1.8974688053131104,
|
| 129381 |
+
"learning_rate": 2.5675533315775613e-05,
|
| 129382 |
+
"loss": 2.2784,
|
| 129383 |
+
"step": 18482
|
| 129384 |
+
},
|
| 129385 |
+
{
|
| 129386 |
+
"epoch": 0.48677903608111667,
|
| 129387 |
+
"grad_norm": 4.346259593963623,
|
| 129388 |
+
"learning_rate": 2.5674216486700025e-05,
|
| 129389 |
+
"loss": 1.9088,
|
| 129390 |
+
"step": 18483
|
| 129391 |
+
},
|
| 129392 |
+
{
|
| 129393 |
+
"epoch": 0.4868053726626284,
|
| 129394 |
+
"grad_norm": 1.8092743158340454,
|
| 129395 |
+
"learning_rate": 2.5672899657624444e-05,
|
| 129396 |
+
"loss": 1.8927,
|
| 129397 |
+
"step": 18484
|
| 129398 |
+
},
|
| 129399 |
+
{
|
| 129400 |
+
"epoch": 0.4868317092441401,
|
| 129401 |
+
"grad_norm": 3.205528736114502,
|
| 129402 |
+
"learning_rate": 2.5671582828548856e-05,
|
| 129403 |
+
"loss": 1.3568,
|
| 129404 |
+
"step": 18485
|
| 129405 |
+
},
|
| 129406 |
+
{
|
| 129407 |
+
"epoch": 0.48685804582565184,
|
| 129408 |
+
"grad_norm": 1.6814615726470947,
|
| 129409 |
+
"learning_rate": 2.567026599947327e-05,
|
| 129410 |
+
"loss": 2.2728,
|
| 129411 |
+
"step": 18486
|
| 129412 |
+
},
|
| 129413 |
+
{
|
| 129414 |
+
"epoch": 0.48688438240716353,
|
| 129415 |
+
"grad_norm": 5.064326763153076,
|
| 129416 |
+
"learning_rate": 2.5668949170397684e-05,
|
| 129417 |
+
"loss": 1.3509,
|
| 129418 |
+
"step": 18487
|
| 129419 |
+
},
|
| 129420 |
+
{
|
| 129421 |
+
"epoch": 0.4869107189886753,
|
| 129422 |
+
"grad_norm": 3.4415476322174072,
|
| 129423 |
+
"learning_rate": 2.5667632341322096e-05,
|
| 129424 |
+
"loss": 2.8161,
|
| 129425 |
+
"step": 18488
|
| 129426 |
+
},
|
| 129427 |
+
{
|
| 129428 |
+
"epoch": 0.486937055570187,
|
| 129429 |
+
"grad_norm": 2.522833824157715,
|
| 129430 |
+
"learning_rate": 2.5666315512246515e-05,
|
| 129431 |
+
"loss": 1.0294,
|
| 129432 |
+
"step": 18489
|
| 129433 |
+
},
|
| 129434 |
+
{
|
| 129435 |
+
"epoch": 0.4869633921516987,
|
| 129436 |
+
"grad_norm": 2.1585633754730225,
|
| 129437 |
+
"learning_rate": 2.5664998683170927e-05,
|
| 129438 |
+
"loss": 1.3564,
|
| 129439 |
+
"step": 18490
|
| 129440 |
+
},
|
| 129441 |
+
{
|
| 129442 |
+
"epoch": 0.48698972873321045,
|
| 129443 |
+
"grad_norm": 2.394378900527954,
|
| 129444 |
+
"learning_rate": 2.566368185409534e-05,
|
| 129445 |
+
"loss": 2.4195,
|
| 129446 |
+
"step": 18491
|
| 129447 |
+
},
|
| 129448 |
+
{
|
| 129449 |
+
"epoch": 0.48701606531472214,
|
| 129450 |
+
"grad_norm": 2.5911078453063965,
|
| 129451 |
+
"learning_rate": 2.5662365025019752e-05,
|
| 129452 |
+
"loss": 1.2326,
|
| 129453 |
+
"step": 18492
|
| 129454 |
+
},
|
| 129455 |
+
{
|
| 129456 |
+
"epoch": 0.4870424018962339,
|
| 129457 |
+
"grad_norm": 1.7802774906158447,
|
| 129458 |
+
"learning_rate": 2.5661048195944164e-05,
|
| 129459 |
+
"loss": 0.6364,
|
| 129460 |
+
"step": 18493
|
| 129461 |
+
},
|
| 129462 |
+
{
|
| 129463 |
+
"epoch": 0.48706873847774557,
|
| 129464 |
+
"grad_norm": 2.638225555419922,
|
| 129465 |
+
"learning_rate": 2.5659731366868583e-05,
|
| 129466 |
+
"loss": 1.918,
|
| 129467 |
+
"step": 18494
|
| 129468 |
+
},
|
| 129469 |
+
{
|
| 129470 |
+
"epoch": 0.4870950750592573,
|
| 129471 |
+
"grad_norm": 1.846629023551941,
|
| 129472 |
+
"learning_rate": 2.5658414537792995e-05,
|
| 129473 |
+
"loss": 1.646,
|
| 129474 |
+
"step": 18495
|
| 129475 |
+
},
|
| 129476 |
+
{
|
| 129477 |
+
"epoch": 0.48712141164076905,
|
| 129478 |
+
"grad_norm": 2.197716236114502,
|
| 129479 |
+
"learning_rate": 2.565709770871741e-05,
|
| 129480 |
+
"loss": 1.7315,
|
| 129481 |
+
"step": 18496
|
| 129482 |
+
},
|
| 129483 |
+
{
|
| 129484 |
+
"epoch": 0.48714774822228074,
|
| 129485 |
+
"grad_norm": 1.91517174243927,
|
| 129486 |
+
"learning_rate": 2.5655780879641823e-05,
|
| 129487 |
+
"loss": 1.6118,
|
| 129488 |
+
"step": 18497
|
| 129489 |
+
},
|
| 129490 |
+
{
|
| 129491 |
+
"epoch": 0.4871740848037925,
|
| 129492 |
+
"grad_norm": 4.605914115905762,
|
| 129493 |
+
"learning_rate": 2.5654464050566242e-05,
|
| 129494 |
+
"loss": 1.6682,
|
| 129495 |
+
"step": 18498
|
| 129496 |
+
},
|
| 129497 |
+
{
|
| 129498 |
+
"epoch": 0.48720042138530417,
|
| 129499 |
+
"grad_norm": 2.0153045654296875,
|
| 129500 |
+
"learning_rate": 2.5653147221490654e-05,
|
| 129501 |
+
"loss": 1.3794,
|
| 129502 |
+
"step": 18499
|
| 129503 |
+
},
|
| 129504 |
+
{
|
| 129505 |
+
"epoch": 0.4872267579668159,
|
| 129506 |
+
"grad_norm": 1.849891185760498,
|
| 129507 |
+
"learning_rate": 2.5651830392415066e-05,
|
| 129508 |
+
"loss": 1.4523,
|
| 129509 |
+
"step": 18500
|
| 129510 |
}
|
| 129511 |
],
|
| 129512 |
"logging_steps": 1,
|
|
|
|
| 129526 |
"attributes": {}
|
| 129527 |
}
|
| 129528 |
},
|
| 129529 |
+
"total_flos": 3.6567889688508826e+17,
|
| 129530 |
"train_batch_size": 1,
|
| 129531 |
"trial_name": null,
|
| 129532 |
"trial_params": null
|