Training in progress, step 1197, 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 125248064
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:668a0889c090e215207d74b332e056b7ca4f6b162bd86ee71bd67cc47290b8bd
|
3 |
size 125248064
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 64220436
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:f6fc8ccd5ba1739971410ba99e9a9f580c06ccddc135d2ee3d386e83d30f3b77
|
3 |
size 64220436
|
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 sha256:731ce61144e7f56e88aacf2b872738f4fe3d94e583f7cc379e0a681bdfdee45e
|
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 sha256:810af60324f4f8b9381e2cfbd8f738b2d5b90fe5950a608079a90f9cdf320a0e
|
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,
|
@@ -5593,6 +5593,2799 @@
|
|
5593 |
"learning_rate": 5.0246824315994423e-05,
|
5594 |
"loss": 1.5175,
|
5595 |
"step": 798
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
5596 |
}
|
5597 |
],
|
5598 |
"logging_steps": 1,
|
@@ -5612,7 +8405,7 @@
|
|
5612 |
"attributes": {}
|
5613 |
}
|
5614 |
},
|
5615 |
-
"total_flos": 1.
|
5616 |
"train_batch_size": 4,
|
5617 |
"trial_name": null,
|
5618 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.2047553882996921,
|
5 |
"eval_steps": 500,
|
6 |
+
"global_step": 1197,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
5593 |
"learning_rate": 5.0246824315994423e-05,
|
5594 |
"loss": 1.5175,
|
5595 |
"step": 798
|
5596 |
+
},
|
5597 |
+
{
|
5598 |
+
"epoch": 0.13667464933287718,
|
5599 |
+
"grad_norm": 3.0145645141601562,
|
5600 |
+
"learning_rate": 5.014809497455023e-05,
|
5601 |
+
"loss": 1.7339,
|
5602 |
+
"step": 799
|
5603 |
+
},
|
5604 |
+
{
|
5605 |
+
"epoch": 0.13684570646595964,
|
5606 |
+
"grad_norm": 3.4925601482391357,
|
5607 |
+
"learning_rate": 5.004936505567575e-05,
|
5608 |
+
"loss": 1.3301,
|
5609 |
+
"step": 800
|
5610 |
+
},
|
5611 |
+
{
|
5612 |
+
"epoch": 0.13701676359904208,
|
5613 |
+
"grad_norm": 0.493186891078949,
|
5614 |
+
"learning_rate": 4.995063494432426e-05,
|
5615 |
+
"loss": 1.1961,
|
5616 |
+
"step": 801
|
5617 |
+
},
|
5618 |
+
{
|
5619 |
+
"epoch": 0.13718782073212452,
|
5620 |
+
"grad_norm": 0.46137088537216187,
|
5621 |
+
"learning_rate": 4.9851905025449765e-05,
|
5622 |
+
"loss": 1.0953,
|
5623 |
+
"step": 802
|
5624 |
+
},
|
5625 |
+
{
|
5626 |
+
"epoch": 0.13735887786520698,
|
5627 |
+
"grad_norm": 0.4970177114009857,
|
5628 |
+
"learning_rate": 4.975317568400559e-05,
|
5629 |
+
"loss": 1.1871,
|
5630 |
+
"step": 803
|
5631 |
+
},
|
5632 |
+
{
|
5633 |
+
"epoch": 0.13752993499828942,
|
5634 |
+
"grad_norm": 0.5300405621528625,
|
5635 |
+
"learning_rate": 4.965444730494272e-05,
|
5636 |
+
"loss": 1.209,
|
5637 |
+
"step": 804
|
5638 |
+
},
|
5639 |
+
{
|
5640 |
+
"epoch": 0.13770099213137188,
|
5641 |
+
"grad_norm": 0.5243121981620789,
|
5642 |
+
"learning_rate": 4.955572027320847e-05,
|
5643 |
+
"loss": 1.2081,
|
5644 |
+
"step": 805
|
5645 |
+
},
|
5646 |
+
{
|
5647 |
+
"epoch": 0.13787204926445432,
|
5648 |
+
"grad_norm": 0.5058876276016235,
|
5649 |
+
"learning_rate": 4.945699497374483e-05,
|
5650 |
+
"loss": 1.1047,
|
5651 |
+
"step": 806
|
5652 |
+
},
|
5653 |
+
{
|
5654 |
+
"epoch": 0.13804310639753678,
|
5655 |
+
"grad_norm": 0.5102903246879578,
|
5656 |
+
"learning_rate": 4.9358271791487093e-05,
|
5657 |
+
"loss": 1.1757,
|
5658 |
+
"step": 807
|
5659 |
+
},
|
5660 |
+
{
|
5661 |
+
"epoch": 0.13821416353061922,
|
5662 |
+
"grad_norm": 0.5319022536277771,
|
5663 |
+
"learning_rate": 4.925955111136226e-05,
|
5664 |
+
"loss": 1.3227,
|
5665 |
+
"step": 808
|
5666 |
+
},
|
5667 |
+
{
|
5668 |
+
"epoch": 0.13838522066370168,
|
5669 |
+
"grad_norm": 0.5021986961364746,
|
5670 |
+
"learning_rate": 4.916083331828761e-05,
|
5671 |
+
"loss": 1.2571,
|
5672 |
+
"step": 809
|
5673 |
+
},
|
5674 |
+
{
|
5675 |
+
"epoch": 0.13855627779678412,
|
5676 |
+
"grad_norm": 0.5275747776031494,
|
5677 |
+
"learning_rate": 4.90621187971691e-05,
|
5678 |
+
"loss": 1.2374,
|
5679 |
+
"step": 810
|
5680 |
+
},
|
5681 |
+
{
|
5682 |
+
"epoch": 0.13872733492986658,
|
5683 |
+
"grad_norm": 0.5663225054740906,
|
5684 |
+
"learning_rate": 4.8963407932900044e-05,
|
5685 |
+
"loss": 1.3147,
|
5686 |
+
"step": 811
|
5687 |
+
},
|
5688 |
+
{
|
5689 |
+
"epoch": 0.13889839206294902,
|
5690 |
+
"grad_norm": 0.5300834774971008,
|
5691 |
+
"learning_rate": 4.886470111035936e-05,
|
5692 |
+
"loss": 1.2838,
|
5693 |
+
"step": 812
|
5694 |
+
},
|
5695 |
+
{
|
5696 |
+
"epoch": 0.13906944919603148,
|
5697 |
+
"grad_norm": 0.5162017941474915,
|
5698 |
+
"learning_rate": 4.8765998714410297e-05,
|
5699 |
+
"loss": 1.3455,
|
5700 |
+
"step": 813
|
5701 |
+
},
|
5702 |
+
{
|
5703 |
+
"epoch": 0.13924050632911392,
|
5704 |
+
"grad_norm": 0.5711933374404907,
|
5705 |
+
"learning_rate": 4.866730112989885e-05,
|
5706 |
+
"loss": 1.4489,
|
5707 |
+
"step": 814
|
5708 |
+
},
|
5709 |
+
{
|
5710 |
+
"epoch": 0.13941156346219638,
|
5711 |
+
"grad_norm": 0.5531579256057739,
|
5712 |
+
"learning_rate": 4.856860874165218e-05,
|
5713 |
+
"loss": 1.2977,
|
5714 |
+
"step": 815
|
5715 |
+
},
|
5716 |
+
{
|
5717 |
+
"epoch": 0.13958262059527882,
|
5718 |
+
"grad_norm": 0.5703489780426025,
|
5719 |
+
"learning_rate": 4.846992193447728e-05,
|
5720 |
+
"loss": 1.3127,
|
5721 |
+
"step": 816
|
5722 |
+
},
|
5723 |
+
{
|
5724 |
+
"epoch": 0.13975367772836128,
|
5725 |
+
"grad_norm": 0.6078313589096069,
|
5726 |
+
"learning_rate": 4.8371241093159286e-05,
|
5727 |
+
"loss": 1.2557,
|
5728 |
+
"step": 817
|
5729 |
+
},
|
5730 |
+
{
|
5731 |
+
"epoch": 0.13992473486144372,
|
5732 |
+
"grad_norm": 0.5923083424568176,
|
5733 |
+
"learning_rate": 4.827256660246017e-05,
|
5734 |
+
"loss": 1.3205,
|
5735 |
+
"step": 818
|
5736 |
+
},
|
5737 |
+
{
|
5738 |
+
"epoch": 0.14009579199452618,
|
5739 |
+
"grad_norm": 0.5954465866088867,
|
5740 |
+
"learning_rate": 4.817389884711705e-05,
|
5741 |
+
"loss": 1.1735,
|
5742 |
+
"step": 819
|
5743 |
+
},
|
5744 |
+
{
|
5745 |
+
"epoch": 0.14026684912760862,
|
5746 |
+
"grad_norm": 0.5864128470420837,
|
5747 |
+
"learning_rate": 4.807523821184089e-05,
|
5748 |
+
"loss": 1.2,
|
5749 |
+
"step": 820
|
5750 |
+
},
|
5751 |
+
{
|
5752 |
+
"epoch": 0.14043790626069108,
|
5753 |
+
"grad_norm": 0.5825478434562683,
|
5754 |
+
"learning_rate": 4.797658508131476e-05,
|
5755 |
+
"loss": 1.2568,
|
5756 |
+
"step": 821
|
5757 |
+
},
|
5758 |
+
{
|
5759 |
+
"epoch": 0.14060896339377352,
|
5760 |
+
"grad_norm": 0.5934092998504639,
|
5761 |
+
"learning_rate": 4.78779398401926e-05,
|
5762 |
+
"loss": 1.3924,
|
5763 |
+
"step": 822
|
5764 |
+
},
|
5765 |
+
{
|
5766 |
+
"epoch": 0.14078002052685598,
|
5767 |
+
"grad_norm": 0.5498810410499573,
|
5768 |
+
"learning_rate": 4.777930287309747e-05,
|
5769 |
+
"loss": 1.0919,
|
5770 |
+
"step": 823
|
5771 |
+
},
|
5772 |
+
{
|
5773 |
+
"epoch": 0.14095107765993842,
|
5774 |
+
"grad_norm": 0.6112752556800842,
|
5775 |
+
"learning_rate": 4.768067456462028e-05,
|
5776 |
+
"loss": 1.1125,
|
5777 |
+
"step": 824
|
5778 |
+
},
|
5779 |
+
{
|
5780 |
+
"epoch": 0.14112213479302088,
|
5781 |
+
"grad_norm": 0.6238945722579956,
|
5782 |
+
"learning_rate": 4.758205529931808e-05,
|
5783 |
+
"loss": 1.0413,
|
5784 |
+
"step": 825
|
5785 |
+
},
|
5786 |
+
{
|
5787 |
+
"epoch": 0.14129319192610332,
|
5788 |
+
"grad_norm": 0.6571114659309387,
|
5789 |
+
"learning_rate": 4.7483445461712736e-05,
|
5790 |
+
"loss": 1.1468,
|
5791 |
+
"step": 826
|
5792 |
+
},
|
5793 |
+
{
|
5794 |
+
"epoch": 0.14146424905918578,
|
5795 |
+
"grad_norm": 0.7267754077911377,
|
5796 |
+
"learning_rate": 4.7384845436289314e-05,
|
5797 |
+
"loss": 1.1779,
|
5798 |
+
"step": 827
|
5799 |
+
},
|
5800 |
+
{
|
5801 |
+
"epoch": 0.14163530619226822,
|
5802 |
+
"grad_norm": 0.6937477588653564,
|
5803 |
+
"learning_rate": 4.728625560749466e-05,
|
5804 |
+
"loss": 1.1945,
|
5805 |
+
"step": 828
|
5806 |
+
},
|
5807 |
+
{
|
5808 |
+
"epoch": 0.14180636332535065,
|
5809 |
+
"grad_norm": 0.722042977809906,
|
5810 |
+
"learning_rate": 4.718767635973578e-05,
|
5811 |
+
"loss": 1.0714,
|
5812 |
+
"step": 829
|
5813 |
+
},
|
5814 |
+
{
|
5815 |
+
"epoch": 0.14197742045843312,
|
5816 |
+
"grad_norm": 0.7702199220657349,
|
5817 |
+
"learning_rate": 4.708910807737854e-05,
|
5818 |
+
"loss": 1.2561,
|
5819 |
+
"step": 830
|
5820 |
+
},
|
5821 |
+
{
|
5822 |
+
"epoch": 0.14214847759151555,
|
5823 |
+
"grad_norm": 0.8672346472740173,
|
5824 |
+
"learning_rate": 4.699055114474594e-05,
|
5825 |
+
"loss": 1.154,
|
5826 |
+
"step": 831
|
5827 |
+
},
|
5828 |
+
{
|
5829 |
+
"epoch": 0.14231953472459802,
|
5830 |
+
"grad_norm": 0.9063764214515686,
|
5831 |
+
"learning_rate": 4.689200594611681e-05,
|
5832 |
+
"loss": 1.2949,
|
5833 |
+
"step": 832
|
5834 |
+
},
|
5835 |
+
{
|
5836 |
+
"epoch": 0.14249059185768045,
|
5837 |
+
"grad_norm": 0.8779078125953674,
|
5838 |
+
"learning_rate": 4.67934728657242e-05,
|
5839 |
+
"loss": 1.2612,
|
5840 |
+
"step": 833
|
5841 |
+
},
|
5842 |
+
{
|
5843 |
+
"epoch": 0.14266164899076292,
|
5844 |
+
"grad_norm": 0.910697877407074,
|
5845 |
+
"learning_rate": 4.669495228775388e-05,
|
5846 |
+
"loss": 1.1209,
|
5847 |
+
"step": 834
|
5848 |
+
},
|
5849 |
+
{
|
5850 |
+
"epoch": 0.14283270612384535,
|
5851 |
+
"grad_norm": 0.8607608079910278,
|
5852 |
+
"learning_rate": 4.659644459634293e-05,
|
5853 |
+
"loss": 0.8503,
|
5854 |
+
"step": 835
|
5855 |
+
},
|
5856 |
+
{
|
5857 |
+
"epoch": 0.14300376325692782,
|
5858 |
+
"grad_norm": 0.9338657259941101,
|
5859 |
+
"learning_rate": 4.6497950175578116e-05,
|
5860 |
+
"loss": 1.1526,
|
5861 |
+
"step": 836
|
5862 |
+
},
|
5863 |
+
{
|
5864 |
+
"epoch": 0.14317482039001025,
|
5865 |
+
"grad_norm": 1.069631814956665,
|
5866 |
+
"learning_rate": 4.639946940949453e-05,
|
5867 |
+
"loss": 1.272,
|
5868 |
+
"step": 837
|
5869 |
+
},
|
5870 |
+
{
|
5871 |
+
"epoch": 0.14334587752309272,
|
5872 |
+
"grad_norm": 1.040444016456604,
|
5873 |
+
"learning_rate": 4.6301002682073973e-05,
|
5874 |
+
"loss": 0.973,
|
5875 |
+
"step": 838
|
5876 |
+
},
|
5877 |
+
{
|
5878 |
+
"epoch": 0.14351693465617515,
|
5879 |
+
"grad_norm": 1.1585819721221924,
|
5880 |
+
"learning_rate": 4.620255037724356e-05,
|
5881 |
+
"loss": 1.1979,
|
5882 |
+
"step": 839
|
5883 |
+
},
|
5884 |
+
{
|
5885 |
+
"epoch": 0.14368799178925762,
|
5886 |
+
"grad_norm": 1.2856534719467163,
|
5887 |
+
"learning_rate": 4.610411287887409e-05,
|
5888 |
+
"loss": 1.3253,
|
5889 |
+
"step": 840
|
5890 |
+
},
|
5891 |
+
{
|
5892 |
+
"epoch": 0.14385904892234005,
|
5893 |
+
"grad_norm": 1.4903264045715332,
|
5894 |
+
"learning_rate": 4.600569057077874e-05,
|
5895 |
+
"loss": 1.0979,
|
5896 |
+
"step": 841
|
5897 |
+
},
|
5898 |
+
{
|
5899 |
+
"epoch": 0.14403010605542252,
|
5900 |
+
"grad_norm": 1.4957740306854248,
|
5901 |
+
"learning_rate": 4.590728383671134e-05,
|
5902 |
+
"loss": 1.2658,
|
5903 |
+
"step": 842
|
5904 |
+
},
|
5905 |
+
{
|
5906 |
+
"epoch": 0.14420116318850495,
|
5907 |
+
"grad_norm": 1.8818494081497192,
|
5908 |
+
"learning_rate": 4.58088930603651e-05,
|
5909 |
+
"loss": 1.7231,
|
5910 |
+
"step": 843
|
5911 |
+
},
|
5912 |
+
{
|
5913 |
+
"epoch": 0.14437222032158742,
|
5914 |
+
"grad_norm": 1.8076220750808716,
|
5915 |
+
"learning_rate": 4.5710518625370935e-05,
|
5916 |
+
"loss": 1.1955,
|
5917 |
+
"step": 844
|
5918 |
+
},
|
5919 |
+
{
|
5920 |
+
"epoch": 0.14454327745466986,
|
5921 |
+
"grad_norm": 1.798600196838379,
|
5922 |
+
"learning_rate": 4.5612160915296106e-05,
|
5923 |
+
"loss": 1.3661,
|
5924 |
+
"step": 845
|
5925 |
+
},
|
5926 |
+
{
|
5927 |
+
"epoch": 0.14471433458775232,
|
5928 |
+
"grad_norm": 2.155062198638916,
|
5929 |
+
"learning_rate": 4.55138203136426e-05,
|
5930 |
+
"loss": 1.508,
|
5931 |
+
"step": 846
|
5932 |
+
},
|
5933 |
+
{
|
5934 |
+
"epoch": 0.14488539172083476,
|
5935 |
+
"grad_norm": 2.1003410816192627,
|
5936 |
+
"learning_rate": 4.5415497203845754e-05,
|
5937 |
+
"loss": 1.526,
|
5938 |
+
"step": 847
|
5939 |
+
},
|
5940 |
+
{
|
5941 |
+
"epoch": 0.14505644885391722,
|
5942 |
+
"grad_norm": 2.4393603801727295,
|
5943 |
+
"learning_rate": 4.531719196927265e-05,
|
5944 |
+
"loss": 1.6606,
|
5945 |
+
"step": 848
|
5946 |
+
},
|
5947 |
+
{
|
5948 |
+
"epoch": 0.14522750598699966,
|
5949 |
+
"grad_norm": 2.322680950164795,
|
5950 |
+
"learning_rate": 4.5218904993220737e-05,
|
5951 |
+
"loss": 1.2985,
|
5952 |
+
"step": 849
|
5953 |
+
},
|
5954 |
+
{
|
5955 |
+
"epoch": 0.14539856312008212,
|
5956 |
+
"grad_norm": 4.077210426330566,
|
5957 |
+
"learning_rate": 4.51206366589162e-05,
|
5958 |
+
"loss": 1.7742,
|
5959 |
+
"step": 850
|
5960 |
+
},
|
5961 |
+
{
|
5962 |
+
"epoch": 0.14556962025316456,
|
5963 |
+
"grad_norm": 0.48009493947029114,
|
5964 |
+
"learning_rate": 4.502238734951262e-05,
|
5965 |
+
"loss": 1.1631,
|
5966 |
+
"step": 851
|
5967 |
+
},
|
5968 |
+
{
|
5969 |
+
"epoch": 0.14574067738624702,
|
5970 |
+
"grad_norm": 0.5012683868408203,
|
5971 |
+
"learning_rate": 4.492415744808931e-05,
|
5972 |
+
"loss": 1.0785,
|
5973 |
+
"step": 852
|
5974 |
+
},
|
5975 |
+
{
|
5976 |
+
"epoch": 0.14591173451932946,
|
5977 |
+
"grad_norm": 0.4767197072505951,
|
5978 |
+
"learning_rate": 4.4825947337649995e-05,
|
5979 |
+
"loss": 1.1628,
|
5980 |
+
"step": 853
|
5981 |
+
},
|
5982 |
+
{
|
5983 |
+
"epoch": 0.1460827916524119,
|
5984 |
+
"grad_norm": 0.5286329984664917,
|
5985 |
+
"learning_rate": 4.4727757401121164e-05,
|
5986 |
+
"loss": 1.2211,
|
5987 |
+
"step": 854
|
5988 |
+
},
|
5989 |
+
{
|
5990 |
+
"epoch": 0.14625384878549436,
|
5991 |
+
"grad_norm": 0.5109902024269104,
|
5992 |
+
"learning_rate": 4.462958802135069e-05,
|
5993 |
+
"loss": 1.2054,
|
5994 |
+
"step": 855
|
5995 |
+
},
|
5996 |
+
{
|
5997 |
+
"epoch": 0.1464249059185768,
|
5998 |
+
"grad_norm": 0.5533120632171631,
|
5999 |
+
"learning_rate": 4.4531439581106295e-05,
|
6000 |
+
"loss": 1.3572,
|
6001 |
+
"step": 856
|
6002 |
+
},
|
6003 |
+
{
|
6004 |
+
"epoch": 0.14659596305165926,
|
6005 |
+
"grad_norm": 0.5145012736320496,
|
6006 |
+
"learning_rate": 4.4433312463074e-05,
|
6007 |
+
"loss": 1.2788,
|
6008 |
+
"step": 857
|
6009 |
+
},
|
6010 |
+
{
|
6011 |
+
"epoch": 0.1467670201847417,
|
6012 |
+
"grad_norm": 0.5198351740837097,
|
6013 |
+
"learning_rate": 4.4335207049856794e-05,
|
6014 |
+
"loss": 1.2127,
|
6015 |
+
"step": 858
|
6016 |
+
},
|
6017 |
+
{
|
6018 |
+
"epoch": 0.14693807731782416,
|
6019 |
+
"grad_norm": 0.5251686573028564,
|
6020 |
+
"learning_rate": 4.423712372397293e-05,
|
6021 |
+
"loss": 1.2763,
|
6022 |
+
"step": 859
|
6023 |
+
},
|
6024 |
+
{
|
6025 |
+
"epoch": 0.1471091344509066,
|
6026 |
+
"grad_norm": 0.5199826955795288,
|
6027 |
+
"learning_rate": 4.4139062867854604e-05,
|
6028 |
+
"loss": 1.3722,
|
6029 |
+
"step": 860
|
6030 |
+
},
|
6031 |
+
{
|
6032 |
+
"epoch": 0.14728019158398906,
|
6033 |
+
"grad_norm": 0.5486602783203125,
|
6034 |
+
"learning_rate": 4.404102486384637e-05,
|
6035 |
+
"loss": 1.3776,
|
6036 |
+
"step": 861
|
6037 |
+
},
|
6038 |
+
{
|
6039 |
+
"epoch": 0.1474512487170715,
|
6040 |
+
"grad_norm": 0.5736585259437561,
|
6041 |
+
"learning_rate": 4.3943010094203735e-05,
|
6042 |
+
"loss": 1.4151,
|
6043 |
+
"step": 862
|
6044 |
+
},
|
6045 |
+
{
|
6046 |
+
"epoch": 0.14762230585015396,
|
6047 |
+
"grad_norm": 0.5580637454986572,
|
6048 |
+
"learning_rate": 4.384501894109152e-05,
|
6049 |
+
"loss": 1.3646,
|
6050 |
+
"step": 863
|
6051 |
+
},
|
6052 |
+
{
|
6053 |
+
"epoch": 0.1477933629832364,
|
6054 |
+
"grad_norm": 0.5629317760467529,
|
6055 |
+
"learning_rate": 4.374705178658257e-05,
|
6056 |
+
"loss": 1.4959,
|
6057 |
+
"step": 864
|
6058 |
+
},
|
6059 |
+
{
|
6060 |
+
"epoch": 0.14796442011631886,
|
6061 |
+
"grad_norm": 0.5788300037384033,
|
6062 |
+
"learning_rate": 4.364910901265606e-05,
|
6063 |
+
"loss": 1.504,
|
6064 |
+
"step": 865
|
6065 |
+
},
|
6066 |
+
{
|
6067 |
+
"epoch": 0.1481354772494013,
|
6068 |
+
"grad_norm": 0.555449366569519,
|
6069 |
+
"learning_rate": 4.35511910011962e-05,
|
6070 |
+
"loss": 1.5292,
|
6071 |
+
"step": 866
|
6072 |
+
},
|
6073 |
+
{
|
6074 |
+
"epoch": 0.14830653438248376,
|
6075 |
+
"grad_norm": 0.5640177130699158,
|
6076 |
+
"learning_rate": 4.3453298133990543e-05,
|
6077 |
+
"loss": 1.2963,
|
6078 |
+
"step": 867
|
6079 |
+
},
|
6080 |
+
{
|
6081 |
+
"epoch": 0.1484775915155662,
|
6082 |
+
"grad_norm": 0.612964928150177,
|
6083 |
+
"learning_rate": 4.33554307927287e-05,
|
6084 |
+
"loss": 1.2783,
|
6085 |
+
"step": 868
|
6086 |
+
},
|
6087 |
+
{
|
6088 |
+
"epoch": 0.14864864864864866,
|
6089 |
+
"grad_norm": 0.5650842785835266,
|
6090 |
+
"learning_rate": 4.325758935900069e-05,
|
6091 |
+
"loss": 1.3257,
|
6092 |
+
"step": 869
|
6093 |
+
},
|
6094 |
+
{
|
6095 |
+
"epoch": 0.1488197057817311,
|
6096 |
+
"grad_norm": 0.58852618932724,
|
6097 |
+
"learning_rate": 4.315977421429559e-05,
|
6098 |
+
"loss": 1.3433,
|
6099 |
+
"step": 870
|
6100 |
+
},
|
6101 |
+
{
|
6102 |
+
"epoch": 0.14899076291481356,
|
6103 |
+
"grad_norm": 0.5934543609619141,
|
6104 |
+
"learning_rate": 4.3061985739999866e-05,
|
6105 |
+
"loss": 1.1679,
|
6106 |
+
"step": 871
|
6107 |
+
},
|
6108 |
+
{
|
6109 |
+
"epoch": 0.149161820047896,
|
6110 |
+
"grad_norm": 0.6740942001342773,
|
6111 |
+
"learning_rate": 4.2964224317396097e-05,
|
6112 |
+
"loss": 1.1713,
|
6113 |
+
"step": 872
|
6114 |
+
},
|
6115 |
+
{
|
6116 |
+
"epoch": 0.14933287718097846,
|
6117 |
+
"grad_norm": 0.638868510723114,
|
6118 |
+
"learning_rate": 4.286649032766131e-05,
|
6119 |
+
"loss": 1.2192,
|
6120 |
+
"step": 873
|
6121 |
+
},
|
6122 |
+
{
|
6123 |
+
"epoch": 0.1495039343140609,
|
6124 |
+
"grad_norm": 0.6754473447799683,
|
6125 |
+
"learning_rate": 4.276878415186564e-05,
|
6126 |
+
"loss": 1.1301,
|
6127 |
+
"step": 874
|
6128 |
+
},
|
6129 |
+
{
|
6130 |
+
"epoch": 0.14967499144714336,
|
6131 |
+
"grad_norm": 0.6947224140167236,
|
6132 |
+
"learning_rate": 4.267110617097073e-05,
|
6133 |
+
"loss": 1.1836,
|
6134 |
+
"step": 875
|
6135 |
+
},
|
6136 |
+
{
|
6137 |
+
"epoch": 0.1498460485802258,
|
6138 |
+
"grad_norm": 0.6807590126991272,
|
6139 |
+
"learning_rate": 4.2573456765828266e-05,
|
6140 |
+
"loss": 1.0204,
|
6141 |
+
"step": 876
|
6142 |
+
},
|
6143 |
+
{
|
6144 |
+
"epoch": 0.15001710571330826,
|
6145 |
+
"grad_norm": 0.7608630657196045,
|
6146 |
+
"learning_rate": 4.24758363171786e-05,
|
6147 |
+
"loss": 0.9777,
|
6148 |
+
"step": 877
|
6149 |
+
},
|
6150 |
+
{
|
6151 |
+
"epoch": 0.1501881628463907,
|
6152 |
+
"grad_norm": 0.7378666996955872,
|
6153 |
+
"learning_rate": 4.237824520564907e-05,
|
6154 |
+
"loss": 1.2896,
|
6155 |
+
"step": 878
|
6156 |
+
},
|
6157 |
+
{
|
6158 |
+
"epoch": 0.15035921997947316,
|
6159 |
+
"grad_norm": 0.8934102058410645,
|
6160 |
+
"learning_rate": 4.228068381175273e-05,
|
6161 |
+
"loss": 1.172,
|
6162 |
+
"step": 879
|
6163 |
+
},
|
6164 |
+
{
|
6165 |
+
"epoch": 0.1505302771125556,
|
6166 |
+
"grad_norm": 0.8139393329620361,
|
6167 |
+
"learning_rate": 4.218315251588669e-05,
|
6168 |
+
"loss": 1.1671,
|
6169 |
+
"step": 880
|
6170 |
+
},
|
6171 |
+
{
|
6172 |
+
"epoch": 0.15070133424563803,
|
6173 |
+
"grad_norm": 0.877079427242279,
|
6174 |
+
"learning_rate": 4.208565169833077e-05,
|
6175 |
+
"loss": 1.2129,
|
6176 |
+
"step": 881
|
6177 |
+
},
|
6178 |
+
{
|
6179 |
+
"epoch": 0.1508723913787205,
|
6180 |
+
"grad_norm": 0.8891003727912903,
|
6181 |
+
"learning_rate": 4.198818173924588e-05,
|
6182 |
+
"loss": 1.1182,
|
6183 |
+
"step": 882
|
6184 |
+
},
|
6185 |
+
{
|
6186 |
+
"epoch": 0.15104344851180293,
|
6187 |
+
"grad_norm": 1.0355708599090576,
|
6188 |
+
"learning_rate": 4.1890743018672685e-05,
|
6189 |
+
"loss": 1.1476,
|
6190 |
+
"step": 883
|
6191 |
+
},
|
6192 |
+
{
|
6193 |
+
"epoch": 0.1512145056448854,
|
6194 |
+
"grad_norm": 1.0708569288253784,
|
6195 |
+
"learning_rate": 4.179333591652997e-05,
|
6196 |
+
"loss": 1.2621,
|
6197 |
+
"step": 884
|
6198 |
+
},
|
6199 |
+
{
|
6200 |
+
"epoch": 0.15138556277796783,
|
6201 |
+
"grad_norm": 1.0150701999664307,
|
6202 |
+
"learning_rate": 4.169596081261332e-05,
|
6203 |
+
"loss": 1.2615,
|
6204 |
+
"step": 885
|
6205 |
+
},
|
6206 |
+
{
|
6207 |
+
"epoch": 0.1515566199110503,
|
6208 |
+
"grad_norm": 0.9939287900924683,
|
6209 |
+
"learning_rate": 4.15986180865935e-05,
|
6210 |
+
"loss": 1.22,
|
6211 |
+
"step": 886
|
6212 |
+
},
|
6213 |
+
{
|
6214 |
+
"epoch": 0.15172767704413273,
|
6215 |
+
"grad_norm": 1.0266711711883545,
|
6216 |
+
"learning_rate": 4.150130811801507e-05,
|
6217 |
+
"loss": 1.4713,
|
6218 |
+
"step": 887
|
6219 |
+
},
|
6220 |
+
{
|
6221 |
+
"epoch": 0.1518987341772152,
|
6222 |
+
"grad_norm": 1.098502516746521,
|
6223 |
+
"learning_rate": 4.140403128629482e-05,
|
6224 |
+
"loss": 1.213,
|
6225 |
+
"step": 888
|
6226 |
+
},
|
6227 |
+
{
|
6228 |
+
"epoch": 0.15206979131029763,
|
6229 |
+
"grad_norm": 1.242407202720642,
|
6230 |
+
"learning_rate": 4.13067879707204e-05,
|
6231 |
+
"loss": 1.0548,
|
6232 |
+
"step": 889
|
6233 |
+
},
|
6234 |
+
{
|
6235 |
+
"epoch": 0.1522408484433801,
|
6236 |
+
"grad_norm": 1.4239497184753418,
|
6237 |
+
"learning_rate": 4.1209578550448703e-05,
|
6238 |
+
"loss": 1.3371,
|
6239 |
+
"step": 890
|
6240 |
+
},
|
6241 |
+
{
|
6242 |
+
"epoch": 0.15241190557646253,
|
6243 |
+
"grad_norm": 1.5052733421325684,
|
6244 |
+
"learning_rate": 4.111240340450453e-05,
|
6245 |
+
"loss": 1.5574,
|
6246 |
+
"step": 891
|
6247 |
+
},
|
6248 |
+
{
|
6249 |
+
"epoch": 0.152582962709545,
|
6250 |
+
"grad_norm": 1.519946575164795,
|
6251 |
+
"learning_rate": 4.1015262911779003e-05,
|
6252 |
+
"loss": 1.2863,
|
6253 |
+
"step": 892
|
6254 |
+
},
|
6255 |
+
{
|
6256 |
+
"epoch": 0.15275401984262743,
|
6257 |
+
"grad_norm": 1.4786194562911987,
|
6258 |
+
"learning_rate": 4.0918157451028185e-05,
|
6259 |
+
"loss": 1.3222,
|
6260 |
+
"step": 893
|
6261 |
+
},
|
6262 |
+
{
|
6263 |
+
"epoch": 0.1529250769757099,
|
6264 |
+
"grad_norm": 1.6151409149169922,
|
6265 |
+
"learning_rate": 4.0821087400871447e-05,
|
6266 |
+
"loss": 1.2022,
|
6267 |
+
"step": 894
|
6268 |
+
},
|
6269 |
+
{
|
6270 |
+
"epoch": 0.15309613410879233,
|
6271 |
+
"grad_norm": 1.8268282413482666,
|
6272 |
+
"learning_rate": 4.072405313979021e-05,
|
6273 |
+
"loss": 1.4788,
|
6274 |
+
"step": 895
|
6275 |
+
},
|
6276 |
+
{
|
6277 |
+
"epoch": 0.1532671912418748,
|
6278 |
+
"grad_norm": 1.672358512878418,
|
6279 |
+
"learning_rate": 4.062705504612624e-05,
|
6280 |
+
"loss": 1.3023,
|
6281 |
+
"step": 896
|
6282 |
+
},
|
6283 |
+
{
|
6284 |
+
"epoch": 0.15343824837495723,
|
6285 |
+
"grad_norm": 2.530867338180542,
|
6286 |
+
"learning_rate": 4.0530093498080365e-05,
|
6287 |
+
"loss": 1.4235,
|
6288 |
+
"step": 897
|
6289 |
+
},
|
6290 |
+
{
|
6291 |
+
"epoch": 0.1536093055080397,
|
6292 |
+
"grad_norm": 2.0230932235717773,
|
6293 |
+
"learning_rate": 4.043316887371091e-05,
|
6294 |
+
"loss": 1.3824,
|
6295 |
+
"step": 898
|
6296 |
+
},
|
6297 |
+
{
|
6298 |
+
"epoch": 0.15378036264112213,
|
6299 |
+
"grad_norm": 1.991603136062622,
|
6300 |
+
"learning_rate": 4.033628155093217e-05,
|
6301 |
+
"loss": 1.0061,
|
6302 |
+
"step": 899
|
6303 |
+
},
|
6304 |
+
{
|
6305 |
+
"epoch": 0.1539514197742046,
|
6306 |
+
"grad_norm": 2.928663969039917,
|
6307 |
+
"learning_rate": 4.023943190751309e-05,
|
6308 |
+
"loss": 1.2392,
|
6309 |
+
"step": 900
|
6310 |
+
},
|
6311 |
+
{
|
6312 |
+
"epoch": 0.15412247690728703,
|
6313 |
+
"grad_norm": 0.6104402542114258,
|
6314 |
+
"learning_rate": 4.014262032107561e-05,
|
6315 |
+
"loss": 1.137,
|
6316 |
+
"step": 901
|
6317 |
+
},
|
6318 |
+
{
|
6319 |
+
"epoch": 0.1542935340403695,
|
6320 |
+
"grad_norm": 0.6759240627288818,
|
6321 |
+
"learning_rate": 4.004584716909337e-05,
|
6322 |
+
"loss": 1.2021,
|
6323 |
+
"step": 902
|
6324 |
+
},
|
6325 |
+
{
|
6326 |
+
"epoch": 0.15446459117345193,
|
6327 |
+
"grad_norm": 0.6067174077033997,
|
6328 |
+
"learning_rate": 3.994911282889008e-05,
|
6329 |
+
"loss": 1.2497,
|
6330 |
+
"step": 903
|
6331 |
+
},
|
6332 |
+
{
|
6333 |
+
"epoch": 0.1546356483065344,
|
6334 |
+
"grad_norm": 0.5458953380584717,
|
6335 |
+
"learning_rate": 3.985241767763818e-05,
|
6336 |
+
"loss": 1.1707,
|
6337 |
+
"step": 904
|
6338 |
+
},
|
6339 |
+
{
|
6340 |
+
"epoch": 0.15480670543961683,
|
6341 |
+
"grad_norm": 0.5344793796539307,
|
6342 |
+
"learning_rate": 3.975576209235726e-05,
|
6343 |
+
"loss": 1.2682,
|
6344 |
+
"step": 905
|
6345 |
+
},
|
6346 |
+
{
|
6347 |
+
"epoch": 0.15497776257269927,
|
6348 |
+
"grad_norm": 0.5440839529037476,
|
6349 |
+
"learning_rate": 3.965914644991268e-05,
|
6350 |
+
"loss": 1.2877,
|
6351 |
+
"step": 906
|
6352 |
+
},
|
6353 |
+
{
|
6354 |
+
"epoch": 0.15514881970578173,
|
6355 |
+
"grad_norm": 0.5137394070625305,
|
6356 |
+
"learning_rate": 3.9562571127014014e-05,
|
6357 |
+
"loss": 1.1467,
|
6358 |
+
"step": 907
|
6359 |
+
},
|
6360 |
+
{
|
6361 |
+
"epoch": 0.15531987683886417,
|
6362 |
+
"grad_norm": 0.5427122116088867,
|
6363 |
+
"learning_rate": 3.94660365002137e-05,
|
6364 |
+
"loss": 1.3611,
|
6365 |
+
"step": 908
|
6366 |
+
},
|
6367 |
+
{
|
6368 |
+
"epoch": 0.15549093397194663,
|
6369 |
+
"grad_norm": 0.5587217807769775,
|
6370 |
+
"learning_rate": 3.936954294590541e-05,
|
6371 |
+
"loss": 1.3282,
|
6372 |
+
"step": 909
|
6373 |
+
},
|
6374 |
+
{
|
6375 |
+
"epoch": 0.15566199110502907,
|
6376 |
+
"grad_norm": 0.623623788356781,
|
6377 |
+
"learning_rate": 3.9273090840322747e-05,
|
6378 |
+
"loss": 1.3397,
|
6379 |
+
"step": 910
|
6380 |
+
},
|
6381 |
+
{
|
6382 |
+
"epoch": 0.15583304823811153,
|
6383 |
+
"grad_norm": 0.5721731781959534,
|
6384 |
+
"learning_rate": 3.917668055953766e-05,
|
6385 |
+
"loss": 1.3586,
|
6386 |
+
"step": 911
|
6387 |
+
},
|
6388 |
+
{
|
6389 |
+
"epoch": 0.15600410537119397,
|
6390 |
+
"grad_norm": 0.5948416590690613,
|
6391 |
+
"learning_rate": 3.9080312479459055e-05,
|
6392 |
+
"loss": 1.3822,
|
6393 |
+
"step": 912
|
6394 |
+
},
|
6395 |
+
{
|
6396 |
+
"epoch": 0.15617516250427643,
|
6397 |
+
"grad_norm": 0.5462729334831238,
|
6398 |
+
"learning_rate": 3.898398697583125e-05,
|
6399 |
+
"loss": 1.3299,
|
6400 |
+
"step": 913
|
6401 |
+
},
|
6402 |
+
{
|
6403 |
+
"epoch": 0.15634621963735887,
|
6404 |
+
"grad_norm": 0.6059056520462036,
|
6405 |
+
"learning_rate": 3.8887704424232624e-05,
|
6406 |
+
"loss": 1.4446,
|
6407 |
+
"step": 914
|
6408 |
+
},
|
6409 |
+
{
|
6410 |
+
"epoch": 0.15651727677044133,
|
6411 |
+
"grad_norm": 0.5938013195991516,
|
6412 |
+
"learning_rate": 3.879146520007399e-05,
|
6413 |
+
"loss": 1.3418,
|
6414 |
+
"step": 915
|
6415 |
+
},
|
6416 |
+
{
|
6417 |
+
"epoch": 0.15668833390352377,
|
6418 |
+
"grad_norm": 0.5638803839683533,
|
6419 |
+
"learning_rate": 3.86952696785973e-05,
|
6420 |
+
"loss": 1.2861,
|
6421 |
+
"step": 916
|
6422 |
+
},
|
6423 |
+
{
|
6424 |
+
"epoch": 0.15685939103660623,
|
6425 |
+
"grad_norm": 0.6107637286186218,
|
6426 |
+
"learning_rate": 3.859911823487407e-05,
|
6427 |
+
"loss": 1.3343,
|
6428 |
+
"step": 917
|
6429 |
+
},
|
6430 |
+
{
|
6431 |
+
"epoch": 0.15703044816968867,
|
6432 |
+
"grad_norm": 0.5767939686775208,
|
6433 |
+
"learning_rate": 3.8503011243804e-05,
|
6434 |
+
"loss": 1.3328,
|
6435 |
+
"step": 918
|
6436 |
+
},
|
6437 |
+
{
|
6438 |
+
"epoch": 0.15720150530277113,
|
6439 |
+
"grad_norm": 0.6362277269363403,
|
6440 |
+
"learning_rate": 3.840694908011341e-05,
|
6441 |
+
"loss": 1.4285,
|
6442 |
+
"step": 919
|
6443 |
+
},
|
6444 |
+
{
|
6445 |
+
"epoch": 0.15737256243585357,
|
6446 |
+
"grad_norm": 0.6359758377075195,
|
6447 |
+
"learning_rate": 3.831093211835384e-05,
|
6448 |
+
"loss": 1.2114,
|
6449 |
+
"step": 920
|
6450 |
+
},
|
6451 |
+
{
|
6452 |
+
"epoch": 0.15754361956893603,
|
6453 |
+
"grad_norm": 0.6466010212898254,
|
6454 |
+
"learning_rate": 3.821496073290063e-05,
|
6455 |
+
"loss": 1.2787,
|
6456 |
+
"step": 921
|
6457 |
+
},
|
6458 |
+
{
|
6459 |
+
"epoch": 0.15771467670201847,
|
6460 |
+
"grad_norm": 0.7036415338516235,
|
6461 |
+
"learning_rate": 3.8119035297951376e-05,
|
6462 |
+
"loss": 1.2148,
|
6463 |
+
"step": 922
|
6464 |
+
},
|
6465 |
+
{
|
6466 |
+
"epoch": 0.15788573383510093,
|
6467 |
+
"grad_norm": 0.6389268636703491,
|
6468 |
+
"learning_rate": 3.8023156187524555e-05,
|
6469 |
+
"loss": 1.0703,
|
6470 |
+
"step": 923
|
6471 |
+
},
|
6472 |
+
{
|
6473 |
+
"epoch": 0.15805679096818337,
|
6474 |
+
"grad_norm": 0.7053115963935852,
|
6475 |
+
"learning_rate": 3.7927323775457946e-05,
|
6476 |
+
"loss": 1.302,
|
6477 |
+
"step": 924
|
6478 |
+
},
|
6479 |
+
{
|
6480 |
+
"epoch": 0.15822784810126583,
|
6481 |
+
"grad_norm": 0.7031298279762268,
|
6482 |
+
"learning_rate": 3.783153843540734e-05,
|
6483 |
+
"loss": 1.0863,
|
6484 |
+
"step": 925
|
6485 |
+
},
|
6486 |
+
{
|
6487 |
+
"epoch": 0.15839890523434827,
|
6488 |
+
"grad_norm": 0.8524893522262573,
|
6489 |
+
"learning_rate": 3.77358005408449e-05,
|
6490 |
+
"loss": 1.255,
|
6491 |
+
"step": 926
|
6492 |
+
},
|
6493 |
+
{
|
6494 |
+
"epoch": 0.15856996236743073,
|
6495 |
+
"grad_norm": 0.7578400373458862,
|
6496 |
+
"learning_rate": 3.764011046505788e-05,
|
6497 |
+
"loss": 1.114,
|
6498 |
+
"step": 927
|
6499 |
+
},
|
6500 |
+
{
|
6501 |
+
"epoch": 0.15874101950051317,
|
6502 |
+
"grad_norm": 0.7838466763496399,
|
6503 |
+
"learning_rate": 3.754446858114703e-05,
|
6504 |
+
"loss": 1.1101,
|
6505 |
+
"step": 928
|
6506 |
+
},
|
6507 |
+
{
|
6508 |
+
"epoch": 0.15891207663359563,
|
6509 |
+
"grad_norm": 0.8556280732154846,
|
6510 |
+
"learning_rate": 3.744887526202523e-05,
|
6511 |
+
"loss": 1.1158,
|
6512 |
+
"step": 929
|
6513 |
+
},
|
6514 |
+
{
|
6515 |
+
"epoch": 0.15908313376667807,
|
6516 |
+
"grad_norm": 0.8539929986000061,
|
6517 |
+
"learning_rate": 3.735333088041596e-05,
|
6518 |
+
"loss": 1.1193,
|
6519 |
+
"step": 930
|
6520 |
+
},
|
6521 |
+
{
|
6522 |
+
"epoch": 0.15925419089976053,
|
6523 |
+
"grad_norm": 0.8921190500259399,
|
6524 |
+
"learning_rate": 3.7257835808851965e-05,
|
6525 |
+
"loss": 0.8338,
|
6526 |
+
"step": 931
|
6527 |
+
},
|
6528 |
+
{
|
6529 |
+
"epoch": 0.15942524803284297,
|
6530 |
+
"grad_norm": 0.9892418384552002,
|
6531 |
+
"learning_rate": 3.716239041967363e-05,
|
6532 |
+
"loss": 1.2079,
|
6533 |
+
"step": 932
|
6534 |
+
},
|
6535 |
+
{
|
6536 |
+
"epoch": 0.1595963051659254,
|
6537 |
+
"grad_norm": 0.9015114903450012,
|
6538 |
+
"learning_rate": 3.706699508502771e-05,
|
6539 |
+
"loss": 1.1499,
|
6540 |
+
"step": 933
|
6541 |
+
},
|
6542 |
+
{
|
6543 |
+
"epoch": 0.15976736229900787,
|
6544 |
+
"grad_norm": 1.1171592473983765,
|
6545 |
+
"learning_rate": 3.6971650176865745e-05,
|
6546 |
+
"loss": 1.4387,
|
6547 |
+
"step": 934
|
6548 |
+
},
|
6549 |
+
{
|
6550 |
+
"epoch": 0.1599384194320903,
|
6551 |
+
"grad_norm": 1.0954560041427612,
|
6552 |
+
"learning_rate": 3.6876356066942706e-05,
|
6553 |
+
"loss": 1.2449,
|
6554 |
+
"step": 935
|
6555 |
+
},
|
6556 |
+
{
|
6557 |
+
"epoch": 0.16010947656517277,
|
6558 |
+
"grad_norm": 1.084873080253601,
|
6559 |
+
"learning_rate": 3.678111312681543e-05,
|
6560 |
+
"loss": 1.2496,
|
6561 |
+
"step": 936
|
6562 |
+
},
|
6563 |
+
{
|
6564 |
+
"epoch": 0.1602805336982552,
|
6565 |
+
"grad_norm": 1.2658636569976807,
|
6566 |
+
"learning_rate": 3.668592172784132e-05,
|
6567 |
+
"loss": 1.3727,
|
6568 |
+
"step": 937
|
6569 |
+
},
|
6570 |
+
{
|
6571 |
+
"epoch": 0.16045159083133767,
|
6572 |
+
"grad_norm": 1.3302921056747437,
|
6573 |
+
"learning_rate": 3.659078224117673e-05,
|
6574 |
+
"loss": 1.2088,
|
6575 |
+
"step": 938
|
6576 |
+
},
|
6577 |
+
{
|
6578 |
+
"epoch": 0.1606226479644201,
|
6579 |
+
"grad_norm": 1.3747141361236572,
|
6580 |
+
"learning_rate": 3.649569503777569e-05,
|
6581 |
+
"loss": 1.1221,
|
6582 |
+
"step": 939
|
6583 |
+
},
|
6584 |
+
{
|
6585 |
+
"epoch": 0.16079370509750257,
|
6586 |
+
"grad_norm": 1.4150470495224,
|
6587 |
+
"learning_rate": 3.6400660488388324e-05,
|
6588 |
+
"loss": 1.2708,
|
6589 |
+
"step": 940
|
6590 |
+
},
|
6591 |
+
{
|
6592 |
+
"epoch": 0.160964762230585,
|
6593 |
+
"grad_norm": 1.542067050933838,
|
6594 |
+
"learning_rate": 3.6305678963559446e-05,
|
6595 |
+
"loss": 1.5618,
|
6596 |
+
"step": 941
|
6597 |
+
},
|
6598 |
+
{
|
6599 |
+
"epoch": 0.16113581936366747,
|
6600 |
+
"grad_norm": 2.0603606700897217,
|
6601 |
+
"learning_rate": 3.621075083362719e-05,
|
6602 |
+
"loss": 1.1855,
|
6603 |
+
"step": 942
|
6604 |
+
},
|
6605 |
+
{
|
6606 |
+
"epoch": 0.1613068764967499,
|
6607 |
+
"grad_norm": 1.7317147254943848,
|
6608 |
+
"learning_rate": 3.6115876468721414e-05,
|
6609 |
+
"loss": 1.571,
|
6610 |
+
"step": 943
|
6611 |
+
},
|
6612 |
+
{
|
6613 |
+
"epoch": 0.16147793362983237,
|
6614 |
+
"grad_norm": 1.9348877668380737,
|
6615 |
+
"learning_rate": 3.602105623876242e-05,
|
6616 |
+
"loss": 1.5018,
|
6617 |
+
"step": 944
|
6618 |
+
},
|
6619 |
+
{
|
6620 |
+
"epoch": 0.1616489907629148,
|
6621 |
+
"grad_norm": 1.688513159751892,
|
6622 |
+
"learning_rate": 3.592629051345936e-05,
|
6623 |
+
"loss": 0.9605,
|
6624 |
+
"step": 945
|
6625 |
+
},
|
6626 |
+
{
|
6627 |
+
"epoch": 0.16182004789599727,
|
6628 |
+
"grad_norm": 2.312366247177124,
|
6629 |
+
"learning_rate": 3.5831579662308967e-05,
|
6630 |
+
"loss": 1.244,
|
6631 |
+
"step": 946
|
6632 |
+
},
|
6633 |
+
{
|
6634 |
+
"epoch": 0.1619911050290797,
|
6635 |
+
"grad_norm": 2.249746322631836,
|
6636 |
+
"learning_rate": 3.57369240545939e-05,
|
6637 |
+
"loss": 1.2131,
|
6638 |
+
"step": 947
|
6639 |
+
},
|
6640 |
+
{
|
6641 |
+
"epoch": 0.16216216216216217,
|
6642 |
+
"grad_norm": 2.077162265777588,
|
6643 |
+
"learning_rate": 3.564232405938153e-05,
|
6644 |
+
"loss": 1.2328,
|
6645 |
+
"step": 948
|
6646 |
+
},
|
6647 |
+
{
|
6648 |
+
"epoch": 0.1623332192952446,
|
6649 |
+
"grad_norm": 2.7018041610717773,
|
6650 |
+
"learning_rate": 3.5547780045522306e-05,
|
6651 |
+
"loss": 1.7778,
|
6652 |
+
"step": 949
|
6653 |
+
},
|
6654 |
+
{
|
6655 |
+
"epoch": 0.16250427642832707,
|
6656 |
+
"grad_norm": 3.0157792568206787,
|
6657 |
+
"learning_rate": 3.5453292381648475e-05,
|
6658 |
+
"loss": 1.3563,
|
6659 |
+
"step": 950
|
6660 |
+
},
|
6661 |
+
{
|
6662 |
+
"epoch": 0.1626753335614095,
|
6663 |
+
"grad_norm": 0.47604596614837646,
|
6664 |
+
"learning_rate": 3.5358861436172485e-05,
|
6665 |
+
"loss": 1.1677,
|
6666 |
+
"step": 951
|
6667 |
+
},
|
6668 |
+
{
|
6669 |
+
"epoch": 0.16284639069449197,
|
6670 |
+
"grad_norm": 0.5650733113288879,
|
6671 |
+
"learning_rate": 3.526448757728573e-05,
|
6672 |
+
"loss": 1.1382,
|
6673 |
+
"step": 952
|
6674 |
+
},
|
6675 |
+
{
|
6676 |
+
"epoch": 0.1630174478275744,
|
6677 |
+
"grad_norm": 0.49029967188835144,
|
6678 |
+
"learning_rate": 3.517017117295697e-05,
|
6679 |
+
"loss": 1.1003,
|
6680 |
+
"step": 953
|
6681 |
+
},
|
6682 |
+
{
|
6683 |
+
"epoch": 0.16318850496065687,
|
6684 |
+
"grad_norm": 0.502017617225647,
|
6685 |
+
"learning_rate": 3.507591259093096e-05,
|
6686 |
+
"loss": 1.1292,
|
6687 |
+
"step": 954
|
6688 |
+
},
|
6689 |
+
{
|
6690 |
+
"epoch": 0.1633595620937393,
|
6691 |
+
"grad_norm": 0.5313692092895508,
|
6692 |
+
"learning_rate": 3.498171219872696e-05,
|
6693 |
+
"loss": 1.226,
|
6694 |
+
"step": 955
|
6695 |
+
},
|
6696 |
+
{
|
6697 |
+
"epoch": 0.16353061922682177,
|
6698 |
+
"grad_norm": 0.5255832672119141,
|
6699 |
+
"learning_rate": 3.488757036363744e-05,
|
6700 |
+
"loss": 1.1977,
|
6701 |
+
"step": 956
|
6702 |
+
},
|
6703 |
+
{
|
6704 |
+
"epoch": 0.1637016763599042,
|
6705 |
+
"grad_norm": 0.5410541296005249,
|
6706 |
+
"learning_rate": 3.4793487452726435e-05,
|
6707 |
+
"loss": 1.3308,
|
6708 |
+
"step": 957
|
6709 |
+
},
|
6710 |
+
{
|
6711 |
+
"epoch": 0.16387273349298664,
|
6712 |
+
"grad_norm": 0.5462161302566528,
|
6713 |
+
"learning_rate": 3.469946383282833e-05,
|
6714 |
+
"loss": 1.3385,
|
6715 |
+
"step": 958
|
6716 |
+
},
|
6717 |
+
{
|
6718 |
+
"epoch": 0.1640437906260691,
|
6719 |
+
"grad_norm": 0.5103150606155396,
|
6720 |
+
"learning_rate": 3.4605499870546273e-05,
|
6721 |
+
"loss": 1.1753,
|
6722 |
+
"step": 959
|
6723 |
+
},
|
6724 |
+
{
|
6725 |
+
"epoch": 0.16421484775915154,
|
6726 |
+
"grad_norm": 0.5624702572822571,
|
6727 |
+
"learning_rate": 3.451159593225085e-05,
|
6728 |
+
"loss": 1.3285,
|
6729 |
+
"step": 960
|
6730 |
+
},
|
6731 |
+
{
|
6732 |
+
"epoch": 0.164385904892234,
|
6733 |
+
"grad_norm": 0.5326713919639587,
|
6734 |
+
"learning_rate": 3.441775238407857e-05,
|
6735 |
+
"loss": 1.2662,
|
6736 |
+
"step": 961
|
6737 |
+
},
|
6738 |
+
{
|
6739 |
+
"epoch": 0.16455696202531644,
|
6740 |
+
"grad_norm": 0.5674173831939697,
|
6741 |
+
"learning_rate": 3.432396959193046e-05,
|
6742 |
+
"loss": 1.4137,
|
6743 |
+
"step": 962
|
6744 |
+
},
|
6745 |
+
{
|
6746 |
+
"epoch": 0.1647280191583989,
|
6747 |
+
"grad_norm": 0.5856799483299255,
|
6748 |
+
"learning_rate": 3.4230247921470716e-05,
|
6749 |
+
"loss": 1.3947,
|
6750 |
+
"step": 963
|
6751 |
+
},
|
6752 |
+
{
|
6753 |
+
"epoch": 0.16489907629148134,
|
6754 |
+
"grad_norm": 0.6064618229866028,
|
6755 |
+
"learning_rate": 3.4136587738125183e-05,
|
6756 |
+
"loss": 1.3143,
|
6757 |
+
"step": 964
|
6758 |
+
},
|
6759 |
+
{
|
6760 |
+
"epoch": 0.1650701334245638,
|
6761 |
+
"grad_norm": 0.6087377071380615,
|
6762 |
+
"learning_rate": 3.4042989407079985e-05,
|
6763 |
+
"loss": 1.4938,
|
6764 |
+
"step": 965
|
6765 |
+
},
|
6766 |
+
{
|
6767 |
+
"epoch": 0.16524119055764624,
|
6768 |
+
"grad_norm": 0.5864953994750977,
|
6769 |
+
"learning_rate": 3.394945329328002e-05,
|
6770 |
+
"loss": 1.2633,
|
6771 |
+
"step": 966
|
6772 |
+
},
|
6773 |
+
{
|
6774 |
+
"epoch": 0.1654122476907287,
|
6775 |
+
"grad_norm": 0.5981158018112183,
|
6776 |
+
"learning_rate": 3.38559797614277e-05,
|
6777 |
+
"loss": 1.3156,
|
6778 |
+
"step": 967
|
6779 |
+
},
|
6780 |
+
{
|
6781 |
+
"epoch": 0.16558330482381114,
|
6782 |
+
"grad_norm": 0.6343181133270264,
|
6783 |
+
"learning_rate": 3.376256917598132e-05,
|
6784 |
+
"loss": 1.3255,
|
6785 |
+
"step": 968
|
6786 |
+
},
|
6787 |
+
{
|
6788 |
+
"epoch": 0.1657543619568936,
|
6789 |
+
"grad_norm": 0.5806681513786316,
|
6790 |
+
"learning_rate": 3.366922190115382e-05,
|
6791 |
+
"loss": 1.0793,
|
6792 |
+
"step": 969
|
6793 |
+
},
|
6794 |
+
{
|
6795 |
+
"epoch": 0.16592541908997605,
|
6796 |
+
"grad_norm": 0.6168162822723389,
|
6797 |
+
"learning_rate": 3.3575938300911255e-05,
|
6798 |
+
"loss": 1.1826,
|
6799 |
+
"step": 970
|
6800 |
+
},
|
6801 |
+
{
|
6802 |
+
"epoch": 0.1660964762230585,
|
6803 |
+
"grad_norm": 0.6655823588371277,
|
6804 |
+
"learning_rate": 3.348271873897143e-05,
|
6805 |
+
"loss": 1.1851,
|
6806 |
+
"step": 971
|
6807 |
+
},
|
6808 |
+
{
|
6809 |
+
"epoch": 0.16626753335614095,
|
6810 |
+
"grad_norm": 0.6643422245979309,
|
6811 |
+
"learning_rate": 3.338956357880241e-05,
|
6812 |
+
"loss": 1.2795,
|
6813 |
+
"step": 972
|
6814 |
+
},
|
6815 |
+
{
|
6816 |
+
"epoch": 0.1664385904892234,
|
6817 |
+
"grad_norm": 0.665899932384491,
|
6818 |
+
"learning_rate": 3.329647318362123e-05,
|
6819 |
+
"loss": 1.1815,
|
6820 |
+
"step": 973
|
6821 |
+
},
|
6822 |
+
{
|
6823 |
+
"epoch": 0.16660964762230585,
|
6824 |
+
"grad_norm": 0.6848543882369995,
|
6825 |
+
"learning_rate": 3.3203447916392325e-05,
|
6826 |
+
"loss": 1.3136,
|
6827 |
+
"step": 974
|
6828 |
+
},
|
6829 |
+
{
|
6830 |
+
"epoch": 0.1667807047553883,
|
6831 |
+
"grad_norm": 0.6881781220436096,
|
6832 |
+
"learning_rate": 3.311048813982627e-05,
|
6833 |
+
"loss": 1.1044,
|
6834 |
+
"step": 975
|
6835 |
+
},
|
6836 |
+
{
|
6837 |
+
"epoch": 0.16695176188847075,
|
6838 |
+
"grad_norm": 0.7546820640563965,
|
6839 |
+
"learning_rate": 3.301759421637821e-05,
|
6840 |
+
"loss": 1.2274,
|
6841 |
+
"step": 976
|
6842 |
+
},
|
6843 |
+
{
|
6844 |
+
"epoch": 0.1671228190215532,
|
6845 |
+
"grad_norm": 0.8200444579124451,
|
6846 |
+
"learning_rate": 3.2924766508246617e-05,
|
6847 |
+
"loss": 1.2202,
|
6848 |
+
"step": 977
|
6849 |
+
},
|
6850 |
+
{
|
6851 |
+
"epoch": 0.16729387615463565,
|
6852 |
+
"grad_norm": 0.8180068135261536,
|
6853 |
+
"learning_rate": 3.283200537737168e-05,
|
6854 |
+
"loss": 1.097,
|
6855 |
+
"step": 978
|
6856 |
+
},
|
6857 |
+
{
|
6858 |
+
"epoch": 0.1674649332877181,
|
6859 |
+
"grad_norm": 0.8673319816589355,
|
6860 |
+
"learning_rate": 3.273931118543408e-05,
|
6861 |
+
"loss": 1.1548,
|
6862 |
+
"step": 979
|
6863 |
+
},
|
6864 |
+
{
|
6865 |
+
"epoch": 0.16763599042080055,
|
6866 |
+
"grad_norm": 0.8381339311599731,
|
6867 |
+
"learning_rate": 3.2646684293853466e-05,
|
6868 |
+
"loss": 1.0368,
|
6869 |
+
"step": 980
|
6870 |
+
},
|
6871 |
+
{
|
6872 |
+
"epoch": 0.167807047553883,
|
6873 |
+
"grad_norm": 0.9022047519683838,
|
6874 |
+
"learning_rate": 3.255412506378709e-05,
|
6875 |
+
"loss": 1.3728,
|
6876 |
+
"step": 981
|
6877 |
+
},
|
6878 |
+
{
|
6879 |
+
"epoch": 0.16797810468696545,
|
6880 |
+
"grad_norm": 0.8893918991088867,
|
6881 |
+
"learning_rate": 3.2461633856128405e-05,
|
6882 |
+
"loss": 1.1926,
|
6883 |
+
"step": 982
|
6884 |
+
},
|
6885 |
+
{
|
6886 |
+
"epoch": 0.16814916182004788,
|
6887 |
+
"grad_norm": 1.0304150581359863,
|
6888 |
+
"learning_rate": 3.2369211031505565e-05,
|
6889 |
+
"loss": 1.2228,
|
6890 |
+
"step": 983
|
6891 |
+
},
|
6892 |
+
{
|
6893 |
+
"epoch": 0.16832021895313035,
|
6894 |
+
"grad_norm": 1.1268855333328247,
|
6895 |
+
"learning_rate": 3.22768569502802e-05,
|
6896 |
+
"loss": 1.5402,
|
6897 |
+
"step": 984
|
6898 |
+
},
|
6899 |
+
{
|
6900 |
+
"epoch": 0.16849127608621278,
|
6901 |
+
"grad_norm": 1.158115029335022,
|
6902 |
+
"learning_rate": 3.218457197254583e-05,
|
6903 |
+
"loss": 1.4032,
|
6904 |
+
"step": 985
|
6905 |
+
},
|
6906 |
+
{
|
6907 |
+
"epoch": 0.16866233321929525,
|
6908 |
+
"grad_norm": 1.0024998188018799,
|
6909 |
+
"learning_rate": 3.209235645812657e-05,
|
6910 |
+
"loss": 1.2276,
|
6911 |
+
"step": 986
|
6912 |
+
},
|
6913 |
+
{
|
6914 |
+
"epoch": 0.16883339035237768,
|
6915 |
+
"grad_norm": 1.1588295698165894,
|
6916 |
+
"learning_rate": 3.200021076657568e-05,
|
6917 |
+
"loss": 1.2341,
|
6918 |
+
"step": 987
|
6919 |
+
},
|
6920 |
+
{
|
6921 |
+
"epoch": 0.16900444748546015,
|
6922 |
+
"grad_norm": 1.2362090349197388,
|
6923 |
+
"learning_rate": 3.190813525717421e-05,
|
6924 |
+
"loss": 1.0726,
|
6925 |
+
"step": 988
|
6926 |
+
},
|
6927 |
+
{
|
6928 |
+
"epoch": 0.16917550461854258,
|
6929 |
+
"grad_norm": 1.3645862340927124,
|
6930 |
+
"learning_rate": 3.18161302889295e-05,
|
6931 |
+
"loss": 1.2231,
|
6932 |
+
"step": 989
|
6933 |
+
},
|
6934 |
+
{
|
6935 |
+
"epoch": 0.16934656175162505,
|
6936 |
+
"grad_norm": 1.3612865209579468,
|
6937 |
+
"learning_rate": 3.172419622057392e-05,
|
6938 |
+
"loss": 1.4351,
|
6939 |
+
"step": 990
|
6940 |
+
},
|
6941 |
+
{
|
6942 |
+
"epoch": 0.16951761888470748,
|
6943 |
+
"grad_norm": 1.5641461610794067,
|
6944 |
+
"learning_rate": 3.163233341056333e-05,
|
6945 |
+
"loss": 1.345,
|
6946 |
+
"step": 991
|
6947 |
+
},
|
6948 |
+
{
|
6949 |
+
"epoch": 0.16968867601778995,
|
6950 |
+
"grad_norm": 1.4898855686187744,
|
6951 |
+
"learning_rate": 3.154054221707582e-05,
|
6952 |
+
"loss": 0.929,
|
6953 |
+
"step": 992
|
6954 |
+
},
|
6955 |
+
{
|
6956 |
+
"epoch": 0.16985973315087238,
|
6957 |
+
"grad_norm": 1.880551815032959,
|
6958 |
+
"learning_rate": 3.144882299801016e-05,
|
6959 |
+
"loss": 1.2601,
|
6960 |
+
"step": 993
|
6961 |
+
},
|
6962 |
+
{
|
6963 |
+
"epoch": 0.17003079028395485,
|
6964 |
+
"grad_norm": 2.0110533237457275,
|
6965 |
+
"learning_rate": 3.135717611098458e-05,
|
6966 |
+
"loss": 1.1735,
|
6967 |
+
"step": 994
|
6968 |
+
},
|
6969 |
+
{
|
6970 |
+
"epoch": 0.17020184741703728,
|
6971 |
+
"grad_norm": 2.06731915473938,
|
6972 |
+
"learning_rate": 3.1265601913335194e-05,
|
6973 |
+
"loss": 1.3196,
|
6974 |
+
"step": 995
|
6975 |
+
},
|
6976 |
+
{
|
6977 |
+
"epoch": 0.17037290455011975,
|
6978 |
+
"grad_norm": 2.0801901817321777,
|
6979 |
+
"learning_rate": 3.117410076211479e-05,
|
6980 |
+
"loss": 1.3804,
|
6981 |
+
"step": 996
|
6982 |
+
},
|
6983 |
+
{
|
6984 |
+
"epoch": 0.17054396168320218,
|
6985 |
+
"grad_norm": 2.814969062805176,
|
6986 |
+
"learning_rate": 3.1082673014091246e-05,
|
6987 |
+
"loss": 1.3725,
|
6988 |
+
"step": 997
|
6989 |
+
},
|
6990 |
+
{
|
6991 |
+
"epoch": 0.17071501881628465,
|
6992 |
+
"grad_norm": 2.79948091506958,
|
6993 |
+
"learning_rate": 3.0991319025746327e-05,
|
6994 |
+
"loss": 1.7117,
|
6995 |
+
"step": 998
|
6996 |
+
},
|
6997 |
+
{
|
6998 |
+
"epoch": 0.17088607594936708,
|
6999 |
+
"grad_norm": 3.2036147117614746,
|
7000 |
+
"learning_rate": 3.090003915327411e-05,
|
7001 |
+
"loss": 1.5203,
|
7002 |
+
"step": 999
|
7003 |
+
},
|
7004 |
+
{
|
7005 |
+
"epoch": 0.17105713308244955,
|
7006 |
+
"grad_norm": 4.76022481918335,
|
7007 |
+
"learning_rate": 3.080883375257979e-05,
|
7008 |
+
"loss": 1.7787,
|
7009 |
+
"step": 1000
|
7010 |
+
},
|
7011 |
+
{
|
7012 |
+
"epoch": 0.17122819021553198,
|
7013 |
+
"grad_norm": 0.49586012959480286,
|
7014 |
+
"learning_rate": 3.0717703179278115e-05,
|
7015 |
+
"loss": 1.1062,
|
7016 |
+
"step": 1001
|
7017 |
+
},
|
7018 |
+
{
|
7019 |
+
"epoch": 0.17139924734861445,
|
7020 |
+
"grad_norm": 0.5908662676811218,
|
7021 |
+
"learning_rate": 3.0626647788692134e-05,
|
7022 |
+
"loss": 1.1972,
|
7023 |
+
"step": 1002
|
7024 |
+
},
|
7025 |
+
{
|
7026 |
+
"epoch": 0.17157030448169688,
|
7027 |
+
"grad_norm": 0.5462906360626221,
|
7028 |
+
"learning_rate": 3.053566793585173e-05,
|
7029 |
+
"loss": 1.3135,
|
7030 |
+
"step": 1003
|
7031 |
+
},
|
7032 |
+
{
|
7033 |
+
"epoch": 0.17174136161477935,
|
7034 |
+
"grad_norm": 0.5159565806388855,
|
7035 |
+
"learning_rate": 3.0444763975492208e-05,
|
7036 |
+
"loss": 1.1127,
|
7037 |
+
"step": 1004
|
7038 |
+
},
|
7039 |
+
{
|
7040 |
+
"epoch": 0.17191241874786178,
|
7041 |
+
"grad_norm": 0.5438991785049438,
|
7042 |
+
"learning_rate": 3.0353936262053064e-05,
|
7043 |
+
"loss": 1.2553,
|
7044 |
+
"step": 1005
|
7045 |
+
},
|
7046 |
+
{
|
7047 |
+
"epoch": 0.17208347588094425,
|
7048 |
+
"grad_norm": 0.5255780816078186,
|
7049 |
+
"learning_rate": 3.026318514967642e-05,
|
7050 |
+
"loss": 1.2302,
|
7051 |
+
"step": 1006
|
7052 |
+
},
|
7053 |
+
{
|
7054 |
+
"epoch": 0.17225453301402668,
|
7055 |
+
"grad_norm": 0.5119370818138123,
|
7056 |
+
"learning_rate": 3.0172510992205806e-05,
|
7057 |
+
"loss": 1.3538,
|
7058 |
+
"step": 1007
|
7059 |
+
},
|
7060 |
+
{
|
7061 |
+
"epoch": 0.17242559014710915,
|
7062 |
+
"grad_norm": 0.5363578796386719,
|
7063 |
+
"learning_rate": 3.008191414318461e-05,
|
7064 |
+
"loss": 1.2825,
|
7065 |
+
"step": 1008
|
7066 |
+
},
|
7067 |
+
{
|
7068 |
+
"epoch": 0.17259664728019158,
|
7069 |
+
"grad_norm": 0.547842264175415,
|
7070 |
+
"learning_rate": 2.9991394955854857e-05,
|
7071 |
+
"loss": 1.1889,
|
7072 |
+
"step": 1009
|
7073 |
+
},
|
7074 |
+
{
|
7075 |
+
"epoch": 0.17276770441327402,
|
7076 |
+
"grad_norm": 0.5618794560432434,
|
7077 |
+
"learning_rate": 2.990095378315572e-05,
|
7078 |
+
"loss": 1.3579,
|
7079 |
+
"step": 1010
|
7080 |
+
},
|
7081 |
+
{
|
7082 |
+
"epoch": 0.17293876154635648,
|
7083 |
+
"grad_norm": 0.5350480079650879,
|
7084 |
+
"learning_rate": 2.981059097772224e-05,
|
7085 |
+
"loss": 1.1268,
|
7086 |
+
"step": 1011
|
7087 |
+
},
|
7088 |
+
{
|
7089 |
+
"epoch": 0.17310981867943892,
|
7090 |
+
"grad_norm": 0.5529703497886658,
|
7091 |
+
"learning_rate": 2.9720306891883844e-05,
|
7092 |
+
"loss": 1.3048,
|
7093 |
+
"step": 1012
|
7094 |
+
},
|
7095 |
+
{
|
7096 |
+
"epoch": 0.17328087581252138,
|
7097 |
+
"grad_norm": 0.5668827295303345,
|
7098 |
+
"learning_rate": 2.9630101877663087e-05,
|
7099 |
+
"loss": 1.2466,
|
7100 |
+
"step": 1013
|
7101 |
+
},
|
7102 |
+
{
|
7103 |
+
"epoch": 0.17345193294560382,
|
7104 |
+
"grad_norm": 0.5508885383605957,
|
7105 |
+
"learning_rate": 2.953997628677415e-05,
|
7106 |
+
"loss": 1.3011,
|
7107 |
+
"step": 1014
|
7108 |
+
},
|
7109 |
+
{
|
7110 |
+
"epoch": 0.17362299007868628,
|
7111 |
+
"grad_norm": 0.553832471370697,
|
7112 |
+
"learning_rate": 2.9449930470621605e-05,
|
7113 |
+
"loss": 1.3579,
|
7114 |
+
"step": 1015
|
7115 |
+
},
|
7116 |
+
{
|
7117 |
+
"epoch": 0.17379404721176872,
|
7118 |
+
"grad_norm": 0.5975161790847778,
|
7119 |
+
"learning_rate": 2.935996478029892e-05,
|
7120 |
+
"loss": 1.4405,
|
7121 |
+
"step": 1016
|
7122 |
+
},
|
7123 |
+
{
|
7124 |
+
"epoch": 0.17396510434485118,
|
7125 |
+
"grad_norm": 0.6038892865180969,
|
7126 |
+
"learning_rate": 2.9270079566587195e-05,
|
7127 |
+
"loss": 1.3255,
|
7128 |
+
"step": 1017
|
7129 |
+
},
|
7130 |
+
{
|
7131 |
+
"epoch": 0.17413616147793362,
|
7132 |
+
"grad_norm": 0.5833334922790527,
|
7133 |
+
"learning_rate": 2.918027517995372e-05,
|
7134 |
+
"loss": 1.2959,
|
7135 |
+
"step": 1018
|
7136 |
+
},
|
7137 |
+
{
|
7138 |
+
"epoch": 0.17430721861101608,
|
7139 |
+
"grad_norm": 0.6514463424682617,
|
7140 |
+
"learning_rate": 2.9090551970550668e-05,
|
7141 |
+
"loss": 1.3835,
|
7142 |
+
"step": 1019
|
7143 |
+
},
|
7144 |
+
{
|
7145 |
+
"epoch": 0.17447827574409852,
|
7146 |
+
"grad_norm": 0.605550229549408,
|
7147 |
+
"learning_rate": 2.9000910288213612e-05,
|
7148 |
+
"loss": 1.1497,
|
7149 |
+
"step": 1020
|
7150 |
+
},
|
7151 |
+
{
|
7152 |
+
"epoch": 0.17464933287718098,
|
7153 |
+
"grad_norm": 0.5976765155792236,
|
7154 |
+
"learning_rate": 2.8911350482460374e-05,
|
7155 |
+
"loss": 1.2176,
|
7156 |
+
"step": 1021
|
7157 |
+
},
|
7158 |
+
{
|
7159 |
+
"epoch": 0.17482039001026342,
|
7160 |
+
"grad_norm": 0.6232640743255615,
|
7161 |
+
"learning_rate": 2.8821872902489437e-05,
|
7162 |
+
"loss": 1.1811,
|
7163 |
+
"step": 1022
|
7164 |
+
},
|
7165 |
+
{
|
7166 |
+
"epoch": 0.17499144714334589,
|
7167 |
+
"grad_norm": 0.5897780060768127,
|
7168 |
+
"learning_rate": 2.873247789717872e-05,
|
7169 |
+
"loss": 0.9642,
|
7170 |
+
"step": 1023
|
7171 |
+
},
|
7172 |
+
{
|
7173 |
+
"epoch": 0.17516250427642832,
|
7174 |
+
"grad_norm": 0.651821494102478,
|
7175 |
+
"learning_rate": 2.8643165815084194e-05,
|
7176 |
+
"loss": 1.2541,
|
7177 |
+
"step": 1024
|
7178 |
+
},
|
7179 |
+
{
|
7180 |
+
"epoch": 0.17533356140951079,
|
7181 |
+
"grad_norm": 0.6755815744400024,
|
7182 |
+
"learning_rate": 2.8553937004438425e-05,
|
7183 |
+
"loss": 1.2069,
|
7184 |
+
"step": 1025
|
7185 |
+
},
|
7186 |
+
{
|
7187 |
+
"epoch": 0.17550461854259322,
|
7188 |
+
"grad_norm": 0.6916977167129517,
|
7189 |
+
"learning_rate": 2.8464791813149418e-05,
|
7190 |
+
"loss": 0.9846,
|
7191 |
+
"step": 1026
|
7192 |
+
},
|
7193 |
+
{
|
7194 |
+
"epoch": 0.17567567567567569,
|
7195 |
+
"grad_norm": 0.7493119239807129,
|
7196 |
+
"learning_rate": 2.8375730588799055e-05,
|
7197 |
+
"loss": 1.1846,
|
7198 |
+
"step": 1027
|
7199 |
+
},
|
7200 |
+
{
|
7201 |
+
"epoch": 0.17584673280875812,
|
7202 |
+
"grad_norm": 0.7904576063156128,
|
7203 |
+
"learning_rate": 2.828675367864187e-05,
|
7204 |
+
"loss": 1.02,
|
7205 |
+
"step": 1028
|
7206 |
+
},
|
7207 |
+
{
|
7208 |
+
"epoch": 0.17601778994184059,
|
7209 |
+
"grad_norm": 0.8208119869232178,
|
7210 |
+
"learning_rate": 2.8197861429603628e-05,
|
7211 |
+
"loss": 1.3182,
|
7212 |
+
"step": 1029
|
7213 |
+
},
|
7214 |
+
{
|
7215 |
+
"epoch": 0.17618884707492302,
|
7216 |
+
"grad_norm": 0.8201320171356201,
|
7217 |
+
"learning_rate": 2.8109054188280004e-05,
|
7218 |
+
"loss": 1.0203,
|
7219 |
+
"step": 1030
|
7220 |
+
},
|
7221 |
+
{
|
7222 |
+
"epoch": 0.17635990420800549,
|
7223 |
+
"grad_norm": 0.8551913499832153,
|
7224 |
+
"learning_rate": 2.8020332300935225e-05,
|
7225 |
+
"loss": 1.0894,
|
7226 |
+
"step": 1031
|
7227 |
+
},
|
7228 |
+
{
|
7229 |
+
"epoch": 0.17653096134108792,
|
7230 |
+
"grad_norm": 0.9669012427330017,
|
7231 |
+
"learning_rate": 2.7931696113500726e-05,
|
7232 |
+
"loss": 1.1606,
|
7233 |
+
"step": 1032
|
7234 |
+
},
|
7235 |
+
{
|
7236 |
+
"epoch": 0.17670201847417039,
|
7237 |
+
"grad_norm": 0.9525704383850098,
|
7238 |
+
"learning_rate": 2.7843145971573747e-05,
|
7239 |
+
"loss": 1.1445,
|
7240 |
+
"step": 1033
|
7241 |
+
},
|
7242 |
+
{
|
7243 |
+
"epoch": 0.17687307560725282,
|
7244 |
+
"grad_norm": 1.0303407907485962,
|
7245 |
+
"learning_rate": 2.7754682220416138e-05,
|
7246 |
+
"loss": 1.4024,
|
7247 |
+
"step": 1034
|
7248 |
+
},
|
7249 |
+
{
|
7250 |
+
"epoch": 0.17704413274033526,
|
7251 |
+
"grad_norm": 1.0042873620986938,
|
7252 |
+
"learning_rate": 2.766630520495277e-05,
|
7253 |
+
"loss": 1.1223,
|
7254 |
+
"step": 1035
|
7255 |
+
},
|
7256 |
+
{
|
7257 |
+
"epoch": 0.17721518987341772,
|
7258 |
+
"grad_norm": 1.1444025039672852,
|
7259 |
+
"learning_rate": 2.7578015269770453e-05,
|
7260 |
+
"loss": 1.4585,
|
7261 |
+
"step": 1036
|
7262 |
+
},
|
7263 |
+
{
|
7264 |
+
"epoch": 0.17738624700650016,
|
7265 |
+
"grad_norm": 1.1342707872390747,
|
7266 |
+
"learning_rate": 2.748981275911633e-05,
|
7267 |
+
"loss": 1.0676,
|
7268 |
+
"step": 1037
|
7269 |
+
},
|
7270 |
+
{
|
7271 |
+
"epoch": 0.17755730413958262,
|
7272 |
+
"grad_norm": 1.1423594951629639,
|
7273 |
+
"learning_rate": 2.7401698016896827e-05,
|
7274 |
+
"loss": 0.9903,
|
7275 |
+
"step": 1038
|
7276 |
+
},
|
7277 |
+
{
|
7278 |
+
"epoch": 0.17772836127266506,
|
7279 |
+
"grad_norm": 1.6184778213500977,
|
7280 |
+
"learning_rate": 2.7313671386676042e-05,
|
7281 |
+
"loss": 1.4268,
|
7282 |
+
"step": 1039
|
7283 |
+
},
|
7284 |
+
{
|
7285 |
+
"epoch": 0.17789941840574752,
|
7286 |
+
"grad_norm": 1.3052678108215332,
|
7287 |
+
"learning_rate": 2.7225733211674563e-05,
|
7288 |
+
"loss": 1.2684,
|
7289 |
+
"step": 1040
|
7290 |
+
},
|
7291 |
+
{
|
7292 |
+
"epoch": 0.17807047553882996,
|
7293 |
+
"grad_norm": 1.6033530235290527,
|
7294 |
+
"learning_rate": 2.7137883834768073e-05,
|
7295 |
+
"loss": 1.2586,
|
7296 |
+
"step": 1041
|
7297 |
+
},
|
7298 |
+
{
|
7299 |
+
"epoch": 0.17824153267191242,
|
7300 |
+
"grad_norm": 1.5226807594299316,
|
7301 |
+
"learning_rate": 2.7050123598486043e-05,
|
7302 |
+
"loss": 1.2644,
|
7303 |
+
"step": 1042
|
7304 |
+
},
|
7305 |
+
{
|
7306 |
+
"epoch": 0.17841258980499486,
|
7307 |
+
"grad_norm": 1.6148399114608765,
|
7308 |
+
"learning_rate": 2.696245284501036e-05,
|
7309 |
+
"loss": 1.1198,
|
7310 |
+
"step": 1043
|
7311 |
+
},
|
7312 |
+
{
|
7313 |
+
"epoch": 0.17858364693807732,
|
7314 |
+
"grad_norm": 1.7775901556015015,
|
7315 |
+
"learning_rate": 2.6874871916174027e-05,
|
7316 |
+
"loss": 1.3457,
|
7317 |
+
"step": 1044
|
7318 |
+
},
|
7319 |
+
{
|
7320 |
+
"epoch": 0.17875470407115976,
|
7321 |
+
"grad_norm": 2.0315780639648438,
|
7322 |
+
"learning_rate": 2.67873811534598e-05,
|
7323 |
+
"loss": 1.2369,
|
7324 |
+
"step": 1045
|
7325 |
+
},
|
7326 |
+
{
|
7327 |
+
"epoch": 0.17892576120424222,
|
7328 |
+
"grad_norm": 2.3106753826141357,
|
7329 |
+
"learning_rate": 2.66999808979989e-05,
|
7330 |
+
"loss": 1.7191,
|
7331 |
+
"step": 1046
|
7332 |
+
},
|
7333 |
+
{
|
7334 |
+
"epoch": 0.17909681833732466,
|
7335 |
+
"grad_norm": 2.3446497917175293,
|
7336 |
+
"learning_rate": 2.661267149056963e-05,
|
7337 |
+
"loss": 1.4838,
|
7338 |
+
"step": 1047
|
7339 |
+
},
|
7340 |
+
{
|
7341 |
+
"epoch": 0.17926787547040712,
|
7342 |
+
"grad_norm": 2.2386043071746826,
|
7343 |
+
"learning_rate": 2.6525453271596083e-05,
|
7344 |
+
"loss": 1.1141,
|
7345 |
+
"step": 1048
|
7346 |
+
},
|
7347 |
+
{
|
7348 |
+
"epoch": 0.17943893260348956,
|
7349 |
+
"grad_norm": 3.4601492881774902,
|
7350 |
+
"learning_rate": 2.64383265811468e-05,
|
7351 |
+
"loss": 1.2839,
|
7352 |
+
"step": 1049
|
7353 |
+
},
|
7354 |
+
{
|
7355 |
+
"epoch": 0.17960998973657202,
|
7356 |
+
"grad_norm": 2.9660983085632324,
|
7357 |
+
"learning_rate": 2.6351291758933428e-05,
|
7358 |
+
"loss": 1.1077,
|
7359 |
+
"step": 1050
|
7360 |
+
},
|
7361 |
+
{
|
7362 |
+
"epoch": 0.17978104686965446,
|
7363 |
+
"grad_norm": 0.557556688785553,
|
7364 |
+
"learning_rate": 2.6264349144309487e-05,
|
7365 |
+
"loss": 1.1663,
|
7366 |
+
"step": 1051
|
7367 |
+
},
|
7368 |
+
{
|
7369 |
+
"epoch": 0.17995210400273692,
|
7370 |
+
"grad_norm": 0.5130797028541565,
|
7371 |
+
"learning_rate": 2.617749907626885e-05,
|
7372 |
+
"loss": 1.2021,
|
7373 |
+
"step": 1052
|
7374 |
+
},
|
7375 |
+
{
|
7376 |
+
"epoch": 0.18012316113581936,
|
7377 |
+
"grad_norm": 0.5419288873672485,
|
7378 |
+
"learning_rate": 2.6090741893444704e-05,
|
7379 |
+
"loss": 1.2136,
|
7380 |
+
"step": 1053
|
7381 |
+
},
|
7382 |
+
{
|
7383 |
+
"epoch": 0.18029421826890182,
|
7384 |
+
"grad_norm": 0.5429167747497559,
|
7385 |
+
"learning_rate": 2.6004077934107895e-05,
|
7386 |
+
"loss": 1.2599,
|
7387 |
+
"step": 1054
|
7388 |
+
},
|
7389 |
+
{
|
7390 |
+
"epoch": 0.18046527540198426,
|
7391 |
+
"grad_norm": 0.5111317038536072,
|
7392 |
+
"learning_rate": 2.5917507536165956e-05,
|
7393 |
+
"loss": 1.1689,
|
7394 |
+
"step": 1055
|
7395 |
+
},
|
7396 |
+
{
|
7397 |
+
"epoch": 0.18063633253506672,
|
7398 |
+
"grad_norm": 0.5068162679672241,
|
7399 |
+
"learning_rate": 2.583103103716147e-05,
|
7400 |
+
"loss": 1.1609,
|
7401 |
+
"step": 1056
|
7402 |
+
},
|
7403 |
+
{
|
7404 |
+
"epoch": 0.18080738966814916,
|
7405 |
+
"grad_norm": 0.5176076889038086,
|
7406 |
+
"learning_rate": 2.574464877427103e-05,
|
7407 |
+
"loss": 1.2752,
|
7408 |
+
"step": 1057
|
7409 |
+
},
|
7410 |
+
{
|
7411 |
+
"epoch": 0.18097844680123162,
|
7412 |
+
"grad_norm": 0.5316131711006165,
|
7413 |
+
"learning_rate": 2.565836108430371e-05,
|
7414 |
+
"loss": 1.1965,
|
7415 |
+
"step": 1058
|
7416 |
+
},
|
7417 |
+
{
|
7418 |
+
"epoch": 0.18114950393431406,
|
7419 |
+
"grad_norm": 0.5631779432296753,
|
7420 |
+
"learning_rate": 2.5572168303699877e-05,
|
7421 |
+
"loss": 1.3192,
|
7422 |
+
"step": 1059
|
7423 |
+
},
|
7424 |
+
{
|
7425 |
+
"epoch": 0.18132056106739652,
|
7426 |
+
"grad_norm": 0.5244829058647156,
|
7427 |
+
"learning_rate": 2.5486070768529837e-05,
|
7428 |
+
"loss": 1.2116,
|
7429 |
+
"step": 1060
|
7430 |
+
},
|
7431 |
+
{
|
7432 |
+
"epoch": 0.18149161820047896,
|
7433 |
+
"grad_norm": 0.540724515914917,
|
7434 |
+
"learning_rate": 2.5400068814492506e-05,
|
7435 |
+
"loss": 1.2249,
|
7436 |
+
"step": 1061
|
7437 |
+
},
|
7438 |
+
{
|
7439 |
+
"epoch": 0.1816626753335614,
|
7440 |
+
"grad_norm": 0.5655743479728699,
|
7441 |
+
"learning_rate": 2.531416277691417e-05,
|
7442 |
+
"loss": 1.4045,
|
7443 |
+
"step": 1062
|
7444 |
+
},
|
7445 |
+
{
|
7446 |
+
"epoch": 0.18183373246664386,
|
7447 |
+
"grad_norm": 0.577039361000061,
|
7448 |
+
"learning_rate": 2.5228352990747085e-05,
|
7449 |
+
"loss": 1.2942,
|
7450 |
+
"step": 1063
|
7451 |
+
},
|
7452 |
+
{
|
7453 |
+
"epoch": 0.1820047895997263,
|
7454 |
+
"grad_norm": 0.590980589389801,
|
7455 |
+
"learning_rate": 2.5142639790568222e-05,
|
7456 |
+
"loss": 1.4284,
|
7457 |
+
"step": 1064
|
7458 |
+
},
|
7459 |
+
{
|
7460 |
+
"epoch": 0.18217584673280876,
|
7461 |
+
"grad_norm": 0.5911884903907776,
|
7462 |
+
"learning_rate": 2.5057023510578037e-05,
|
7463 |
+
"loss": 1.3287,
|
7464 |
+
"step": 1065
|
7465 |
+
},
|
7466 |
+
{
|
7467 |
+
"epoch": 0.1823469038658912,
|
7468 |
+
"grad_norm": 0.6275861859321594,
|
7469 |
+
"learning_rate": 2.4971504484598958e-05,
|
7470 |
+
"loss": 1.4182,
|
7471 |
+
"step": 1066
|
7472 |
+
},
|
7473 |
+
{
|
7474 |
+
"epoch": 0.18251796099897366,
|
7475 |
+
"grad_norm": 0.6274673342704773,
|
7476 |
+
"learning_rate": 2.488608304607431e-05,
|
7477 |
+
"loss": 1.3304,
|
7478 |
+
"step": 1067
|
7479 |
+
},
|
7480 |
+
{
|
7481 |
+
"epoch": 0.1826890181320561,
|
7482 |
+
"grad_norm": 0.578771710395813,
|
7483 |
+
"learning_rate": 2.4800759528066885e-05,
|
7484 |
+
"loss": 1.1852,
|
7485 |
+
"step": 1068
|
7486 |
+
},
|
7487 |
+
{
|
7488 |
+
"epoch": 0.18286007526513856,
|
7489 |
+
"grad_norm": 0.6625096201896667,
|
7490 |
+
"learning_rate": 2.4715534263257668e-05,
|
7491 |
+
"loss": 1.2603,
|
7492 |
+
"step": 1069
|
7493 |
+
},
|
7494 |
+
{
|
7495 |
+
"epoch": 0.183031132398221,
|
7496 |
+
"grad_norm": 0.6263986229896545,
|
7497 |
+
"learning_rate": 2.4630407583944615e-05,
|
7498 |
+
"loss": 1.2788,
|
7499 |
+
"step": 1070
|
7500 |
+
},
|
7501 |
+
{
|
7502 |
+
"epoch": 0.18320218953130346,
|
7503 |
+
"grad_norm": 0.6890533566474915,
|
7504 |
+
"learning_rate": 2.4545379822041175e-05,
|
7505 |
+
"loss": 1.348,
|
7506 |
+
"step": 1071
|
7507 |
+
},
|
7508 |
+
{
|
7509 |
+
"epoch": 0.1833732466643859,
|
7510 |
+
"grad_norm": 0.6721847653388977,
|
7511 |
+
"learning_rate": 2.4460451309075248e-05,
|
7512 |
+
"loss": 1.3921,
|
7513 |
+
"step": 1072
|
7514 |
+
},
|
7515 |
+
{
|
7516 |
+
"epoch": 0.18354430379746836,
|
7517 |
+
"grad_norm": 0.6540228128433228,
|
7518 |
+
"learning_rate": 2.4375622376187618e-05,
|
7519 |
+
"loss": 1.1462,
|
7520 |
+
"step": 1073
|
7521 |
+
},
|
7522 |
+
{
|
7523 |
+
"epoch": 0.1837153609305508,
|
7524 |
+
"grad_norm": 0.7116681337356567,
|
7525 |
+
"learning_rate": 2.4290893354130922e-05,
|
7526 |
+
"loss": 1.2309,
|
7527 |
+
"step": 1074
|
7528 |
+
},
|
7529 |
+
{
|
7530 |
+
"epoch": 0.18388641806363326,
|
7531 |
+
"grad_norm": 0.6965073943138123,
|
7532 |
+
"learning_rate": 2.4206264573268172e-05,
|
7533 |
+
"loss": 1.051,
|
7534 |
+
"step": 1075
|
7535 |
+
},
|
7536 |
+
{
|
7537 |
+
"epoch": 0.1840574751967157,
|
7538 |
+
"grad_norm": 0.7276071310043335,
|
7539 |
+
"learning_rate": 2.4121736363571557e-05,
|
7540 |
+
"loss": 1.0871,
|
7541 |
+
"step": 1076
|
7542 |
+
},
|
7543 |
+
{
|
7544 |
+
"epoch": 0.18422853232979816,
|
7545 |
+
"grad_norm": 0.7418241500854492,
|
7546 |
+
"learning_rate": 2.4037309054621115e-05,
|
7547 |
+
"loss": 1.0666,
|
7548 |
+
"step": 1077
|
7549 |
+
},
|
7550 |
+
{
|
7551 |
+
"epoch": 0.1843995894628806,
|
7552 |
+
"grad_norm": 0.7333283424377441,
|
7553 |
+
"learning_rate": 2.3952982975603496e-05,
|
7554 |
+
"loss": 1.1881,
|
7555 |
+
"step": 1078
|
7556 |
+
},
|
7557 |
+
{
|
7558 |
+
"epoch": 0.18457064659596306,
|
7559 |
+
"grad_norm": 0.7455992102622986,
|
7560 |
+
"learning_rate": 2.3868758455310614e-05,
|
7561 |
+
"loss": 1.0302,
|
7562 |
+
"step": 1079
|
7563 |
+
},
|
7564 |
+
{
|
7565 |
+
"epoch": 0.1847417037290455,
|
7566 |
+
"grad_norm": 0.8666993379592896,
|
7567 |
+
"learning_rate": 2.3784635822138424e-05,
|
7568 |
+
"loss": 1.2001,
|
7569 |
+
"step": 1080
|
7570 |
+
},
|
7571 |
+
{
|
7572 |
+
"epoch": 0.18491276086212796,
|
7573 |
+
"grad_norm": 0.8161441683769226,
|
7574 |
+
"learning_rate": 2.3700615404085598e-05,
|
7575 |
+
"loss": 1.2939,
|
7576 |
+
"step": 1081
|
7577 |
+
},
|
7578 |
+
{
|
7579 |
+
"epoch": 0.1850838179952104,
|
7580 |
+
"grad_norm": 0.9533507823944092,
|
7581 |
+
"learning_rate": 2.361669752875233e-05,
|
7582 |
+
"loss": 1.2736,
|
7583 |
+
"step": 1082
|
7584 |
+
},
|
7585 |
+
{
|
7586 |
+
"epoch": 0.18525487512829286,
|
7587 |
+
"grad_norm": 0.8938537836074829,
|
7588 |
+
"learning_rate": 2.3532882523338873e-05,
|
7589 |
+
"loss": 0.9843,
|
7590 |
+
"step": 1083
|
7591 |
+
},
|
7592 |
+
{
|
7593 |
+
"epoch": 0.1854259322613753,
|
7594 |
+
"grad_norm": 0.8930996060371399,
|
7595 |
+
"learning_rate": 2.3449170714644524e-05,
|
7596 |
+
"loss": 1.2126,
|
7597 |
+
"step": 1084
|
7598 |
+
},
|
7599 |
+
{
|
7600 |
+
"epoch": 0.18559698939445776,
|
7601 |
+
"grad_norm": 0.9868687391281128,
|
7602 |
+
"learning_rate": 2.336556242906608e-05,
|
7603 |
+
"loss": 1.2214,
|
7604 |
+
"step": 1085
|
7605 |
+
},
|
7606 |
+
{
|
7607 |
+
"epoch": 0.1857680465275402,
|
7608 |
+
"grad_norm": 0.9706618785858154,
|
7609 |
+
"learning_rate": 2.3282057992596797e-05,
|
7610 |
+
"loss": 0.9344,
|
7611 |
+
"step": 1086
|
7612 |
+
},
|
7613 |
+
{
|
7614 |
+
"epoch": 0.18593910366062263,
|
7615 |
+
"grad_norm": 0.9862615466117859,
|
7616 |
+
"learning_rate": 2.3198657730824985e-05,
|
7617 |
+
"loss": 1.0938,
|
7618 |
+
"step": 1087
|
7619 |
+
},
|
7620 |
+
{
|
7621 |
+
"epoch": 0.1861101607937051,
|
7622 |
+
"grad_norm": 1.3694723844528198,
|
7623 |
+
"learning_rate": 2.3115361968932714e-05,
|
7624 |
+
"loss": 1.2747,
|
7625 |
+
"step": 1088
|
7626 |
+
},
|
7627 |
+
{
|
7628 |
+
"epoch": 0.18628121792678753,
|
7629 |
+
"grad_norm": 1.3156152963638306,
|
7630 |
+
"learning_rate": 2.3032171031694723e-05,
|
7631 |
+
"loss": 1.4315,
|
7632 |
+
"step": 1089
|
7633 |
+
},
|
7634 |
+
{
|
7635 |
+
"epoch": 0.18645227505987,
|
7636 |
+
"grad_norm": 1.4331876039505005,
|
7637 |
+
"learning_rate": 2.2949085243476886e-05,
|
7638 |
+
"loss": 1.3056,
|
7639 |
+
"step": 1090
|
7640 |
+
},
|
7641 |
+
{
|
7642 |
+
"epoch": 0.18662333219295243,
|
7643 |
+
"grad_norm": 1.6338574886322021,
|
7644 |
+
"learning_rate": 2.2866104928235243e-05,
|
7645 |
+
"loss": 1.3037,
|
7646 |
+
"step": 1091
|
7647 |
+
},
|
7648 |
+
{
|
7649 |
+
"epoch": 0.1867943893260349,
|
7650 |
+
"grad_norm": 1.6983578205108643,
|
7651 |
+
"learning_rate": 2.2783230409514488e-05,
|
7652 |
+
"loss": 1.2716,
|
7653 |
+
"step": 1092
|
7654 |
+
},
|
7655 |
+
{
|
7656 |
+
"epoch": 0.18696544645911733,
|
7657 |
+
"grad_norm": 1.7741411924362183,
|
7658 |
+
"learning_rate": 2.270046201044685e-05,
|
7659 |
+
"loss": 1.2031,
|
7660 |
+
"step": 1093
|
7661 |
+
},
|
7662 |
+
{
|
7663 |
+
"epoch": 0.1871365035921998,
|
7664 |
+
"grad_norm": 1.93906831741333,
|
7665 |
+
"learning_rate": 2.2617800053750782e-05,
|
7666 |
+
"loss": 1.2603,
|
7667 |
+
"step": 1094
|
7668 |
+
},
|
7669 |
+
{
|
7670 |
+
"epoch": 0.18730756072528223,
|
7671 |
+
"grad_norm": 2.0896623134613037,
|
7672 |
+
"learning_rate": 2.2535244861729705e-05,
|
7673 |
+
"loss": 1.6169,
|
7674 |
+
"step": 1095
|
7675 |
+
},
|
7676 |
+
{
|
7677 |
+
"epoch": 0.1874786178583647,
|
7678 |
+
"grad_norm": 2.2424557209014893,
|
7679 |
+
"learning_rate": 2.2452796756270777e-05,
|
7680 |
+
"loss": 1.3498,
|
7681 |
+
"step": 1096
|
7682 |
+
},
|
7683 |
+
{
|
7684 |
+
"epoch": 0.18764967499144714,
|
7685 |
+
"grad_norm": 2.4930217266082764,
|
7686 |
+
"learning_rate": 2.2370456058843604e-05,
|
7687 |
+
"loss": 1.4391,
|
7688 |
+
"step": 1097
|
7689 |
+
},
|
7690 |
+
{
|
7691 |
+
"epoch": 0.1878207321245296,
|
7692 |
+
"grad_norm": 2.979562759399414,
|
7693 |
+
"learning_rate": 2.2288223090499007e-05,
|
7694 |
+
"loss": 1.6536,
|
7695 |
+
"step": 1098
|
7696 |
+
},
|
7697 |
+
{
|
7698 |
+
"epoch": 0.18799178925761204,
|
7699 |
+
"grad_norm": 3.8147144317626953,
|
7700 |
+
"learning_rate": 2.2206098171867774e-05,
|
7701 |
+
"loss": 1.9091,
|
7702 |
+
"step": 1099
|
7703 |
+
},
|
7704 |
+
{
|
7705 |
+
"epoch": 0.1881628463906945,
|
7706 |
+
"grad_norm": 3.4456284046173096,
|
7707 |
+
"learning_rate": 2.2124081623159366e-05,
|
7708 |
+
"loss": 1.1332,
|
7709 |
+
"step": 1100
|
7710 |
+
},
|
7711 |
+
{
|
7712 |
+
"epoch": 0.18833390352377694,
|
7713 |
+
"grad_norm": 0.457207053899765,
|
7714 |
+
"learning_rate": 2.204217376416079e-05,
|
7715 |
+
"loss": 1.1188,
|
7716 |
+
"step": 1101
|
7717 |
+
},
|
7718 |
+
{
|
7719 |
+
"epoch": 0.1885049606568594,
|
7720 |
+
"grad_norm": 0.5947155952453613,
|
7721 |
+
"learning_rate": 2.1960374914235138e-05,
|
7722 |
+
"loss": 1.1839,
|
7723 |
+
"step": 1102
|
7724 |
+
},
|
7725 |
+
{
|
7726 |
+
"epoch": 0.18867601778994184,
|
7727 |
+
"grad_norm": 0.5215460658073425,
|
7728 |
+
"learning_rate": 2.187868539232061e-05,
|
7729 |
+
"loss": 1.1348,
|
7730 |
+
"step": 1103
|
7731 |
+
},
|
7732 |
+
{
|
7733 |
+
"epoch": 0.1888470749230243,
|
7734 |
+
"grad_norm": 0.5244162082672119,
|
7735 |
+
"learning_rate": 2.1797105516928984e-05,
|
7736 |
+
"loss": 1.1576,
|
7737 |
+
"step": 1104
|
7738 |
+
},
|
7739 |
+
{
|
7740 |
+
"epoch": 0.18901813205610674,
|
7741 |
+
"grad_norm": 0.5163097977638245,
|
7742 |
+
"learning_rate": 2.1715635606144653e-05,
|
7743 |
+
"loss": 1.1651,
|
7744 |
+
"step": 1105
|
7745 |
+
},
|
7746 |
+
{
|
7747 |
+
"epoch": 0.1891891891891892,
|
7748 |
+
"grad_norm": 0.5445159673690796,
|
7749 |
+
"learning_rate": 2.1634275977623177e-05,
|
7750 |
+
"loss": 1.0768,
|
7751 |
+
"step": 1106
|
7752 |
+
},
|
7753 |
+
{
|
7754 |
+
"epoch": 0.18936024632227164,
|
7755 |
+
"grad_norm": 0.5356796383857727,
|
7756 |
+
"learning_rate": 2.1553026948590137e-05,
|
7757 |
+
"loss": 1.1355,
|
7758 |
+
"step": 1107
|
7759 |
+
},
|
7760 |
+
{
|
7761 |
+
"epoch": 0.1895313034553541,
|
7762 |
+
"grad_norm": 0.5185807943344116,
|
7763 |
+
"learning_rate": 2.1471888835839882e-05,
|
7764 |
+
"loss": 1.2743,
|
7765 |
+
"step": 1108
|
7766 |
+
},
|
7767 |
+
{
|
7768 |
+
"epoch": 0.18970236058843654,
|
7769 |
+
"grad_norm": 0.5446439385414124,
|
7770 |
+
"learning_rate": 2.1390861955734287e-05,
|
7771 |
+
"loss": 1.2662,
|
7772 |
+
"step": 1109
|
7773 |
+
},
|
7774 |
+
{
|
7775 |
+
"epoch": 0.189873417721519,
|
7776 |
+
"grad_norm": 0.5431277751922607,
|
7777 |
+
"learning_rate": 2.1309946624201533e-05,
|
7778 |
+
"loss": 1.3458,
|
7779 |
+
"step": 1110
|
7780 |
+
},
|
7781 |
+
{
|
7782 |
+
"epoch": 0.19004447485460144,
|
7783 |
+
"grad_norm": 0.5551103353500366,
|
7784 |
+
"learning_rate": 2.1229143156734865e-05,
|
7785 |
+
"loss": 1.3616,
|
7786 |
+
"step": 1111
|
7787 |
+
},
|
7788 |
+
{
|
7789 |
+
"epoch": 0.1902155319876839,
|
7790 |
+
"grad_norm": 0.5428812503814697,
|
7791 |
+
"learning_rate": 2.1148451868391356e-05,
|
7792 |
+
"loss": 1.415,
|
7793 |
+
"step": 1112
|
7794 |
+
},
|
7795 |
+
{
|
7796 |
+
"epoch": 0.19038658912076634,
|
7797 |
+
"grad_norm": 0.5525882244110107,
|
7798 |
+
"learning_rate": 2.10678730737907e-05,
|
7799 |
+
"loss": 1.3613,
|
7800 |
+
"step": 1113
|
7801 |
+
},
|
7802 |
+
{
|
7803 |
+
"epoch": 0.19055764625384877,
|
7804 |
+
"grad_norm": 0.575508177280426,
|
7805 |
+
"learning_rate": 2.0987407087113965e-05,
|
7806 |
+
"loss": 1.2821,
|
7807 |
+
"step": 1114
|
7808 |
+
},
|
7809 |
+
{
|
7810 |
+
"epoch": 0.19072870338693124,
|
7811 |
+
"grad_norm": 0.6201297640800476,
|
7812 |
+
"learning_rate": 2.090705422210237e-05,
|
7813 |
+
"loss": 1.4153,
|
7814 |
+
"step": 1115
|
7815 |
+
},
|
7816 |
+
{
|
7817 |
+
"epoch": 0.19089976052001367,
|
7818 |
+
"grad_norm": 0.5878698229789734,
|
7819 |
+
"learning_rate": 2.082681479205607e-05,
|
7820 |
+
"loss": 1.2869,
|
7821 |
+
"step": 1116
|
7822 |
+
},
|
7823 |
+
{
|
7824 |
+
"epoch": 0.19107081765309614,
|
7825 |
+
"grad_norm": 0.6694919466972351,
|
7826 |
+
"learning_rate": 2.0746689109832918e-05,
|
7827 |
+
"loss": 1.3329,
|
7828 |
+
"step": 1117
|
7829 |
+
},
|
7830 |
+
{
|
7831 |
+
"epoch": 0.19124187478617857,
|
7832 |
+
"grad_norm": 0.6331332921981812,
|
7833 |
+
"learning_rate": 2.0666677487847325e-05,
|
7834 |
+
"loss": 1.2102,
|
7835 |
+
"step": 1118
|
7836 |
+
},
|
7837 |
+
{
|
7838 |
+
"epoch": 0.19141293191926104,
|
7839 |
+
"grad_norm": 0.6576963067054749,
|
7840 |
+
"learning_rate": 2.0586780238068848e-05,
|
7841 |
+
"loss": 1.2694,
|
7842 |
+
"step": 1119
|
7843 |
+
},
|
7844 |
+
{
|
7845 |
+
"epoch": 0.19158398905234347,
|
7846 |
+
"grad_norm": 0.6312718391418457,
|
7847 |
+
"learning_rate": 2.050699767202125e-05,
|
7848 |
+
"loss": 1.1597,
|
7849 |
+
"step": 1120
|
7850 |
+
},
|
7851 |
+
{
|
7852 |
+
"epoch": 0.19175504618542594,
|
7853 |
+
"grad_norm": 0.6914326548576355,
|
7854 |
+
"learning_rate": 2.042733010078099e-05,
|
7855 |
+
"loss": 1.1148,
|
7856 |
+
"step": 1121
|
7857 |
+
},
|
7858 |
+
{
|
7859 |
+
"epoch": 0.19192610331850837,
|
7860 |
+
"grad_norm": 0.699762225151062,
|
7861 |
+
"learning_rate": 2.034777783497628e-05,
|
7862 |
+
"loss": 1.1196,
|
7863 |
+
"step": 1122
|
7864 |
+
},
|
7865 |
+
{
|
7866 |
+
"epoch": 0.19209716045159084,
|
7867 |
+
"grad_norm": 0.7305629253387451,
|
7868 |
+
"learning_rate": 2.026834118478567e-05,
|
7869 |
+
"loss": 1.1534,
|
7870 |
+
"step": 1123
|
7871 |
+
},
|
7872 |
+
{
|
7873 |
+
"epoch": 0.19226821758467327,
|
7874 |
+
"grad_norm": 0.7772334218025208,
|
7875 |
+
"learning_rate": 2.0189020459936953e-05,
|
7876 |
+
"loss": 1.1844,
|
7877 |
+
"step": 1124
|
7878 |
+
},
|
7879 |
+
{
|
7880 |
+
"epoch": 0.19243927471775574,
|
7881 |
+
"grad_norm": 0.7272955179214478,
|
7882 |
+
"learning_rate": 2.010981596970592e-05,
|
7883 |
+
"loss": 1.1997,
|
7884 |
+
"step": 1125
|
7885 |
+
},
|
7886 |
+
{
|
7887 |
+
"epoch": 0.19261033185083817,
|
7888 |
+
"grad_norm": 0.6972033381462097,
|
7889 |
+
"learning_rate": 2.003072802291514e-05,
|
7890 |
+
"loss": 1.0817,
|
7891 |
+
"step": 1126
|
7892 |
+
},
|
7893 |
+
{
|
7894 |
+
"epoch": 0.19278138898392064,
|
7895 |
+
"grad_norm": 0.7530586123466492,
|
7896 |
+
"learning_rate": 1.995175692793279e-05,
|
7897 |
+
"loss": 1.0761,
|
7898 |
+
"step": 1127
|
7899 |
+
},
|
7900 |
+
{
|
7901 |
+
"epoch": 0.19295244611700307,
|
7902 |
+
"grad_norm": 0.8682284355163574,
|
7903 |
+
"learning_rate": 1.9872902992671433e-05,
|
7904 |
+
"loss": 1.2157,
|
7905 |
+
"step": 1128
|
7906 |
+
},
|
7907 |
+
{
|
7908 |
+
"epoch": 0.19312350325008554,
|
7909 |
+
"grad_norm": 0.8820382952690125,
|
7910 |
+
"learning_rate": 1.9794166524586798e-05,
|
7911 |
+
"loss": 1.2118,
|
7912 |
+
"step": 1129
|
7913 |
+
},
|
7914 |
+
{
|
7915 |
+
"epoch": 0.19329456038316797,
|
7916 |
+
"grad_norm": 0.8392881751060486,
|
7917 |
+
"learning_rate": 1.9715547830676673e-05,
|
7918 |
+
"loss": 1.0795,
|
7919 |
+
"step": 1130
|
7920 |
+
},
|
7921 |
+
{
|
7922 |
+
"epoch": 0.19346561751625044,
|
7923 |
+
"grad_norm": 0.9215697646141052,
|
7924 |
+
"learning_rate": 1.9637047217479538e-05,
|
7925 |
+
"loss": 1.187,
|
7926 |
+
"step": 1131
|
7927 |
+
},
|
7928 |
+
{
|
7929 |
+
"epoch": 0.19363667464933287,
|
7930 |
+
"grad_norm": 1.0534172058105469,
|
7931 |
+
"learning_rate": 1.9558664991073534e-05,
|
7932 |
+
"loss": 1.1555,
|
7933 |
+
"step": 1132
|
7934 |
+
},
|
7935 |
+
{
|
7936 |
+
"epoch": 0.19380773178241534,
|
7937 |
+
"grad_norm": 1.1032742261886597,
|
7938 |
+
"learning_rate": 1.9480401457075194e-05,
|
7939 |
+
"loss": 1.3511,
|
7940 |
+
"step": 1133
|
7941 |
+
},
|
7942 |
+
{
|
7943 |
+
"epoch": 0.19397878891549777,
|
7944 |
+
"grad_norm": 1.005715012550354,
|
7945 |
+
"learning_rate": 1.9402256920638235e-05,
|
7946 |
+
"loss": 1.136,
|
7947 |
+
"step": 1134
|
7948 |
+
},
|
7949 |
+
{
|
7950 |
+
"epoch": 0.19414984604858024,
|
7951 |
+
"grad_norm": 1.0756511688232422,
|
7952 |
+
"learning_rate": 1.9324231686452477e-05,
|
7953 |
+
"loss": 1.2249,
|
7954 |
+
"step": 1135
|
7955 |
+
},
|
7956 |
+
{
|
7957 |
+
"epoch": 0.19432090318166267,
|
7958 |
+
"grad_norm": 1.1209959983825684,
|
7959 |
+
"learning_rate": 1.9246326058742453e-05,
|
7960 |
+
"loss": 1.0826,
|
7961 |
+
"step": 1136
|
7962 |
+
},
|
7963 |
+
{
|
7964 |
+
"epoch": 0.19449196031474514,
|
7965 |
+
"grad_norm": 1.322952389717102,
|
7966 |
+
"learning_rate": 1.916854034126648e-05,
|
7967 |
+
"loss": 1.3319,
|
7968 |
+
"step": 1137
|
7969 |
+
},
|
7970 |
+
{
|
7971 |
+
"epoch": 0.19466301744782757,
|
7972 |
+
"grad_norm": 1.3385710716247559,
|
7973 |
+
"learning_rate": 1.9090874837315188e-05,
|
7974 |
+
"loss": 1.2366,
|
7975 |
+
"step": 1138
|
7976 |
+
},
|
7977 |
+
{
|
7978 |
+
"epoch": 0.19483407458091,
|
7979 |
+
"grad_norm": 1.2587987184524536,
|
7980 |
+
"learning_rate": 1.9013329849710638e-05,
|
7981 |
+
"loss": 1.3087,
|
7982 |
+
"step": 1139
|
7983 |
+
},
|
7984 |
+
{
|
7985 |
+
"epoch": 0.19500513171399247,
|
7986 |
+
"grad_norm": 1.4214056730270386,
|
7987 |
+
"learning_rate": 1.8935905680804862e-05,
|
7988 |
+
"loss": 1.3393,
|
7989 |
+
"step": 1140
|
7990 |
+
},
|
7991 |
+
{
|
7992 |
+
"epoch": 0.1951761888470749,
|
7993 |
+
"grad_norm": 1.8044555187225342,
|
7994 |
+
"learning_rate": 1.885860263247891e-05,
|
7995 |
+
"loss": 1.3181,
|
7996 |
+
"step": 1141
|
7997 |
+
},
|
7998 |
+
{
|
7999 |
+
"epoch": 0.19534724598015737,
|
8000 |
+
"grad_norm": 2.2799766063690186,
|
8001 |
+
"learning_rate": 1.878142100614153e-05,
|
8002 |
+
"loss": 1.5249,
|
8003 |
+
"step": 1142
|
8004 |
+
},
|
8005 |
+
{
|
8006 |
+
"epoch": 0.1955183031132398,
|
8007 |
+
"grad_norm": 2.19240140914917,
|
8008 |
+
"learning_rate": 1.8704361102728047e-05,
|
8009 |
+
"loss": 1.6337,
|
8010 |
+
"step": 1143
|
8011 |
+
},
|
8012 |
+
{
|
8013 |
+
"epoch": 0.19568936024632227,
|
8014 |
+
"grad_norm": 1.7697051763534546,
|
8015 |
+
"learning_rate": 1.8627423222699182e-05,
|
8016 |
+
"loss": 1.1521,
|
8017 |
+
"step": 1144
|
8018 |
+
},
|
8019 |
+
{
|
8020 |
+
"epoch": 0.1958604173794047,
|
8021 |
+
"grad_norm": 2.0701982975006104,
|
8022 |
+
"learning_rate": 1.8550607666039876e-05,
|
8023 |
+
"loss": 1.1736,
|
8024 |
+
"step": 1145
|
8025 |
+
},
|
8026 |
+
{
|
8027 |
+
"epoch": 0.19603147451248717,
|
8028 |
+
"grad_norm": 2.2206737995147705,
|
8029 |
+
"learning_rate": 1.847391473225813e-05,
|
8030 |
+
"loss": 1.1745,
|
8031 |
+
"step": 1146
|
8032 |
+
},
|
8033 |
+
{
|
8034 |
+
"epoch": 0.1962025316455696,
|
8035 |
+
"grad_norm": 2.475080728530884,
|
8036 |
+
"learning_rate": 1.8397344720383837e-05,
|
8037 |
+
"loss": 1.4157,
|
8038 |
+
"step": 1147
|
8039 |
+
},
|
8040 |
+
{
|
8041 |
+
"epoch": 0.19637358877865208,
|
8042 |
+
"grad_norm": 2.672316074371338,
|
8043 |
+
"learning_rate": 1.8320897928967577e-05,
|
8044 |
+
"loss": 1.4707,
|
8045 |
+
"step": 1148
|
8046 |
+
},
|
8047 |
+
{
|
8048 |
+
"epoch": 0.1965446459117345,
|
8049 |
+
"grad_norm": 2.6906559467315674,
|
8050 |
+
"learning_rate": 1.8244574656079578e-05,
|
8051 |
+
"loss": 1.4775,
|
8052 |
+
"step": 1149
|
8053 |
+
},
|
8054 |
+
{
|
8055 |
+
"epoch": 0.19671570304481698,
|
8056 |
+
"grad_norm": 4.176140308380127,
|
8057 |
+
"learning_rate": 1.8168375199308325e-05,
|
8058 |
+
"loss": 2.0203,
|
8059 |
+
"step": 1150
|
8060 |
+
},
|
8061 |
+
{
|
8062 |
+
"epoch": 0.1968867601778994,
|
8063 |
+
"grad_norm": 0.5323451161384583,
|
8064 |
+
"learning_rate": 1.809229985575967e-05,
|
8065 |
+
"loss": 1.3286,
|
8066 |
+
"step": 1151
|
8067 |
+
},
|
8068 |
+
{
|
8069 |
+
"epoch": 0.19705781731098188,
|
8070 |
+
"grad_norm": 0.5143298506736755,
|
8071 |
+
"learning_rate": 1.801634892205545e-05,
|
8072 |
+
"loss": 1.2097,
|
8073 |
+
"step": 1152
|
8074 |
+
},
|
8075 |
+
{
|
8076 |
+
"epoch": 0.1972288744440643,
|
8077 |
+
"grad_norm": 0.543321967124939,
|
8078 |
+
"learning_rate": 1.794052269433244e-05,
|
8079 |
+
"loss": 1.2148,
|
8080 |
+
"step": 1153
|
8081 |
+
},
|
8082 |
+
{
|
8083 |
+
"epoch": 0.19739993157714678,
|
8084 |
+
"grad_norm": 0.5211328268051147,
|
8085 |
+
"learning_rate": 1.7864821468241243e-05,
|
8086 |
+
"loss": 1.2108,
|
8087 |
+
"step": 1154
|
8088 |
+
},
|
8089 |
+
{
|
8090 |
+
"epoch": 0.1975709887102292,
|
8091 |
+
"grad_norm": 0.5046496391296387,
|
8092 |
+
"learning_rate": 1.7789245538944972e-05,
|
8093 |
+
"loss": 1.1098,
|
8094 |
+
"step": 1155
|
8095 |
+
},
|
8096 |
+
{
|
8097 |
+
"epoch": 0.19774204584331168,
|
8098 |
+
"grad_norm": 0.5350122451782227,
|
8099 |
+
"learning_rate": 1.77137952011183e-05,
|
8100 |
+
"loss": 1.2301,
|
8101 |
+
"step": 1156
|
8102 |
+
},
|
8103 |
+
{
|
8104 |
+
"epoch": 0.1979131029763941,
|
8105 |
+
"grad_norm": 0.5448925495147705,
|
8106 |
+
"learning_rate": 1.7638470748946107e-05,
|
8107 |
+
"loss": 1.3313,
|
8108 |
+
"step": 1157
|
8109 |
+
},
|
8110 |
+
{
|
8111 |
+
"epoch": 0.19808416010947658,
|
8112 |
+
"grad_norm": 0.5379893779754639,
|
8113 |
+
"learning_rate": 1.756327247612255e-05,
|
8114 |
+
"loss": 1.2289,
|
8115 |
+
"step": 1158
|
8116 |
+
},
|
8117 |
+
{
|
8118 |
+
"epoch": 0.198255217242559,
|
8119 |
+
"grad_norm": 0.5303690433502197,
|
8120 |
+
"learning_rate": 1.748820067584973e-05,
|
8121 |
+
"loss": 1.1302,
|
8122 |
+
"step": 1159
|
8123 |
+
},
|
8124 |
+
{
|
8125 |
+
"epoch": 0.19842627437564148,
|
8126 |
+
"grad_norm": 0.5711317658424377,
|
8127 |
+
"learning_rate": 1.7413255640836652e-05,
|
8128 |
+
"loss": 1.3224,
|
8129 |
+
"step": 1160
|
8130 |
+
},
|
8131 |
+
{
|
8132 |
+
"epoch": 0.1985973315087239,
|
8133 |
+
"grad_norm": 0.5847582221031189,
|
8134 |
+
"learning_rate": 1.7338437663298044e-05,
|
8135 |
+
"loss": 1.3615,
|
8136 |
+
"step": 1161
|
8137 |
+
},
|
8138 |
+
{
|
8139 |
+
"epoch": 0.19876838864180638,
|
8140 |
+
"grad_norm": 0.5739108324050903,
|
8141 |
+
"learning_rate": 1.7263747034953242e-05,
|
8142 |
+
"loss": 1.359,
|
8143 |
+
"step": 1162
|
8144 |
+
},
|
8145 |
+
{
|
8146 |
+
"epoch": 0.1989394457748888,
|
8147 |
+
"grad_norm": 0.615225076675415,
|
8148 |
+
"learning_rate": 1.718918404702504e-05,
|
8149 |
+
"loss": 1.3037,
|
8150 |
+
"step": 1163
|
8151 |
+
},
|
8152 |
+
{
|
8153 |
+
"epoch": 0.19911050290797128,
|
8154 |
+
"grad_norm": 0.569936215877533,
|
8155 |
+
"learning_rate": 1.7114748990238543e-05,
|
8156 |
+
"loss": 1.3948,
|
8157 |
+
"step": 1164
|
8158 |
+
},
|
8159 |
+
{
|
8160 |
+
"epoch": 0.1992815600410537,
|
8161 |
+
"grad_norm": 0.620499312877655,
|
8162 |
+
"learning_rate": 1.7040442154820037e-05,
|
8163 |
+
"loss": 1.4661,
|
8164 |
+
"step": 1165
|
8165 |
+
},
|
8166 |
+
{
|
8167 |
+
"epoch": 0.19945261717413615,
|
8168 |
+
"grad_norm": 0.6095662713050842,
|
8169 |
+
"learning_rate": 1.6966263830495936e-05,
|
8170 |
+
"loss": 1.3705,
|
8171 |
+
"step": 1166
|
8172 |
+
},
|
8173 |
+
{
|
8174 |
+
"epoch": 0.1996236743072186,
|
8175 |
+
"grad_norm": 0.6132876873016357,
|
8176 |
+
"learning_rate": 1.6892214306491468e-05,
|
8177 |
+
"loss": 1.3012,
|
8178 |
+
"step": 1167
|
8179 |
+
},
|
8180 |
+
{
|
8181 |
+
"epoch": 0.19979473144030105,
|
8182 |
+
"grad_norm": 0.6166319847106934,
|
8183 |
+
"learning_rate": 1.6818293871529772e-05,
|
8184 |
+
"loss": 1.2791,
|
8185 |
+
"step": 1168
|
8186 |
+
},
|
8187 |
+
{
|
8188 |
+
"epoch": 0.1999657885733835,
|
8189 |
+
"grad_norm": 0.6385096311569214,
|
8190 |
+
"learning_rate": 1.6744502813830564e-05,
|
8191 |
+
"loss": 1.1081,
|
8192 |
+
"step": 1169
|
8193 |
+
},
|
8194 |
+
{
|
8195 |
+
"epoch": 0.20013684570646595,
|
8196 |
+
"grad_norm": 0.6585813164710999,
|
8197 |
+
"learning_rate": 1.66708414211092e-05,
|
8198 |
+
"loss": 1.3367,
|
8199 |
+
"step": 1170
|
8200 |
+
},
|
8201 |
+
{
|
8202 |
+
"epoch": 0.2003079028395484,
|
8203 |
+
"grad_norm": 0.6779057383537292,
|
8204 |
+
"learning_rate": 1.659730998057541e-05,
|
8205 |
+
"loss": 1.1752,
|
8206 |
+
"step": 1171
|
8207 |
+
},
|
8208 |
+
{
|
8209 |
+
"epoch": 0.20047895997263085,
|
8210 |
+
"grad_norm": 0.6424527764320374,
|
8211 |
+
"learning_rate": 1.6523908778932245e-05,
|
8212 |
+
"loss": 1.1954,
|
8213 |
+
"step": 1172
|
8214 |
+
},
|
8215 |
+
{
|
8216 |
+
"epoch": 0.2006500171057133,
|
8217 |
+
"grad_norm": 0.7097743153572083,
|
8218 |
+
"learning_rate": 1.6450638102374972e-05,
|
8219 |
+
"loss": 1.196,
|
8220 |
+
"step": 1173
|
8221 |
+
},
|
8222 |
+
{
|
8223 |
+
"epoch": 0.20082107423879575,
|
8224 |
+
"grad_norm": 0.6838304996490479,
|
8225 |
+
"learning_rate": 1.6377498236589856e-05,
|
8226 |
+
"loss": 1.1732,
|
8227 |
+
"step": 1174
|
8228 |
+
},
|
8229 |
+
{
|
8230 |
+
"epoch": 0.2009921313718782,
|
8231 |
+
"grad_norm": 0.774002194404602,
|
8232 |
+
"learning_rate": 1.6304489466753236e-05,
|
8233 |
+
"loss": 1.1021,
|
8234 |
+
"step": 1175
|
8235 |
+
},
|
8236 |
+
{
|
8237 |
+
"epoch": 0.20116318850496065,
|
8238 |
+
"grad_norm": 0.7578701972961426,
|
8239 |
+
"learning_rate": 1.623161207753023e-05,
|
8240 |
+
"loss": 1.114,
|
8241 |
+
"step": 1176
|
8242 |
+
},
|
8243 |
+
{
|
8244 |
+
"epoch": 0.2013342456380431,
|
8245 |
+
"grad_norm": 0.7900825142860413,
|
8246 |
+
"learning_rate": 1.6158866353073703e-05,
|
8247 |
+
"loss": 1.1419,
|
8248 |
+
"step": 1177
|
8249 |
+
},
|
8250 |
+
{
|
8251 |
+
"epoch": 0.20150530277112555,
|
8252 |
+
"grad_norm": 0.8154890537261963,
|
8253 |
+
"learning_rate": 1.6086252577023165e-05,
|
8254 |
+
"loss": 1.0797,
|
8255 |
+
"step": 1178
|
8256 |
+
},
|
8257 |
+
{
|
8258 |
+
"epoch": 0.201676359904208,
|
8259 |
+
"grad_norm": 0.8612083196640015,
|
8260 |
+
"learning_rate": 1.6013771032503642e-05,
|
8261 |
+
"loss": 1.1258,
|
8262 |
+
"step": 1179
|
8263 |
+
},
|
8264 |
+
{
|
8265 |
+
"epoch": 0.20184741703729045,
|
8266 |
+
"grad_norm": 0.9238095879554749,
|
8267 |
+
"learning_rate": 1.5941422002124596e-05,
|
8268 |
+
"loss": 1.0973,
|
8269 |
+
"step": 1180
|
8270 |
+
},
|
8271 |
+
{
|
8272 |
+
"epoch": 0.2020184741703729,
|
8273 |
+
"grad_norm": 0.8815112113952637,
|
8274 |
+
"learning_rate": 1.586920576797879e-05,
|
8275 |
+
"loss": 1.1965,
|
8276 |
+
"step": 1181
|
8277 |
+
},
|
8278 |
+
{
|
8279 |
+
"epoch": 0.20218953130345535,
|
8280 |
+
"grad_norm": 0.93575119972229,
|
8281 |
+
"learning_rate": 1.5797122611641222e-05,
|
8282 |
+
"loss": 1.2994,
|
8283 |
+
"step": 1182
|
8284 |
+
},
|
8285 |
+
{
|
8286 |
+
"epoch": 0.2023605884365378,
|
8287 |
+
"grad_norm": 0.9842674732208252,
|
8288 |
+
"learning_rate": 1.572517281416801e-05,
|
8289 |
+
"loss": 1.441,
|
8290 |
+
"step": 1183
|
8291 |
+
},
|
8292 |
+
{
|
8293 |
+
"epoch": 0.20253164556962025,
|
8294 |
+
"grad_norm": 0.9615129232406616,
|
8295 |
+
"learning_rate": 1.5653356656095276e-05,
|
8296 |
+
"loss": 1.1581,
|
8297 |
+
"step": 1184
|
8298 |
+
},
|
8299 |
+
{
|
8300 |
+
"epoch": 0.20270270270270271,
|
8301 |
+
"grad_norm": 0.9850116968154907,
|
8302 |
+
"learning_rate": 1.558167441743814e-05,
|
8303 |
+
"loss": 1.1813,
|
8304 |
+
"step": 1185
|
8305 |
+
},
|
8306 |
+
{
|
8307 |
+
"epoch": 0.20287375983578515,
|
8308 |
+
"grad_norm": 1.164113998413086,
|
8309 |
+
"learning_rate": 1.5510126377689467e-05,
|
8310 |
+
"loss": 1.3432,
|
8311 |
+
"step": 1186
|
8312 |
+
},
|
8313 |
+
{
|
8314 |
+
"epoch": 0.20304481696886761,
|
8315 |
+
"grad_norm": 1.2685987949371338,
|
8316 |
+
"learning_rate": 1.5438712815818975e-05,
|
8317 |
+
"loss": 1.1601,
|
8318 |
+
"step": 1187
|
8319 |
+
},
|
8320 |
+
{
|
8321 |
+
"epoch": 0.20321587410195005,
|
8322 |
+
"grad_norm": 1.0663849115371704,
|
8323 |
+
"learning_rate": 1.5367434010271936e-05,
|
8324 |
+
"loss": 1.0746,
|
8325 |
+
"step": 1188
|
8326 |
+
},
|
8327 |
+
{
|
8328 |
+
"epoch": 0.20338693123503251,
|
8329 |
+
"grad_norm": 1.3609206676483154,
|
8330 |
+
"learning_rate": 1.5296290238968303e-05,
|
8331 |
+
"loss": 1.1224,
|
8332 |
+
"step": 1189
|
8333 |
+
},
|
8334 |
+
{
|
8335 |
+
"epoch": 0.20355798836811495,
|
8336 |
+
"grad_norm": 1.2203283309936523,
|
8337 |
+
"learning_rate": 1.5225281779301475e-05,
|
8338 |
+
"loss": 1.138,
|
8339 |
+
"step": 1190
|
8340 |
+
},
|
8341 |
+
{
|
8342 |
+
"epoch": 0.2037290455011974,
|
8343 |
+
"grad_norm": 1.4130033254623413,
|
8344 |
+
"learning_rate": 1.5154408908137252e-05,
|
8345 |
+
"loss": 1.3855,
|
8346 |
+
"step": 1191
|
8347 |
+
},
|
8348 |
+
{
|
8349 |
+
"epoch": 0.20390010263427985,
|
8350 |
+
"grad_norm": 1.561368465423584,
|
8351 |
+
"learning_rate": 1.5083671901812796e-05,
|
8352 |
+
"loss": 1.3283,
|
8353 |
+
"step": 1192
|
8354 |
+
},
|
8355 |
+
{
|
8356 |
+
"epoch": 0.2040711597673623,
|
8357 |
+
"grad_norm": 1.6972764730453491,
|
8358 |
+
"learning_rate": 1.50130710361355e-05,
|
8359 |
+
"loss": 1.403,
|
8360 |
+
"step": 1193
|
8361 |
+
},
|
8362 |
+
{
|
8363 |
+
"epoch": 0.20424221690044475,
|
8364 |
+
"grad_norm": 1.7497429847717285,
|
8365 |
+
"learning_rate": 1.4942606586381957e-05,
|
8366 |
+
"loss": 1.432,
|
8367 |
+
"step": 1194
|
8368 |
+
},
|
8369 |
+
{
|
8370 |
+
"epoch": 0.2044132740335272,
|
8371 |
+
"grad_norm": 2.139078378677368,
|
8372 |
+
"learning_rate": 1.4872278827296854e-05,
|
8373 |
+
"loss": 1.13,
|
8374 |
+
"step": 1195
|
8375 |
+
},
|
8376 |
+
{
|
8377 |
+
"epoch": 0.20458433116660965,
|
8378 |
+
"grad_norm": 2.1031577587127686,
|
8379 |
+
"learning_rate": 1.4802088033091921e-05,
|
8380 |
+
"loss": 1.2499,
|
8381 |
+
"step": 1196
|
8382 |
+
},
|
8383 |
+
{
|
8384 |
+
"epoch": 0.2047553882996921,
|
8385 |
+
"grad_norm": 2.3110170364379883,
|
8386 |
+
"learning_rate": 1.4732034477444844e-05,
|
8387 |
+
"loss": 1.3927,
|
8388 |
+
"step": 1197
|
8389 |
}
|
8390 |
],
|
8391 |
"logging_steps": 1,
|
|
|
8405 |
"attributes": {}
|
8406 |
}
|
8407 |
},
|
8408 |
+
"total_flos": 1.5196889970376704e+18,
|
8409 |
"train_batch_size": 4,
|
8410 |
"trial_name": null,
|
8411 |
"trial_params": null
|