Training in progress, step 1538, checkpoint
Browse files
last-checkpoint/adapter_model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 27024
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:2eeb37550f3717557101813f5b68445432705ab7b2a9c243835f654ee91469b3
|
| 3 |
size 27024
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 64038
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a6a97b190add553d71b895ee7fac0be372f566d39707da665d6732520f1703d6
|
| 3 |
size 64038
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14244
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:6a2696529baf6716ca3275e46bd6352c90d1b8ad699f59fa8c81259de910bd4d
|
| 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:ee4c376ac9817d05eaa03fee64830b713dc4d1abe5488024f05f717c9099682e
|
| 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": 385,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -8124,6 +8124,2687 @@
|
|
| 8124 |
"eval_samples_per_second": 281.621,
|
| 8125 |
"eval_steps_per_second": 140.952,
|
| 8126 |
"step": 1155
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 8127 |
}
|
| 8128 |
],
|
| 8129 |
"logging_steps": 1,
|
|
@@ -8138,12 +10819,12 @@
|
|
| 8138 |
"should_evaluate": false,
|
| 8139 |
"should_log": false,
|
| 8140 |
"should_save": true,
|
| 8141 |
-
"should_training_stop":
|
| 8142 |
},
|
| 8143 |
"attributes": {}
|
| 8144 |
}
|
| 8145 |
},
|
| 8146 |
-
"total_flos":
|
| 8147 |
"train_batch_size": 2,
|
| 8148 |
"trial_name": null,
|
| 8149 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.6513499205929063,
|
| 5 |
"eval_steps": 385,
|
| 6 |
+
"global_step": 1538,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 8124 |
"eval_samples_per_second": 281.621,
|
| 8125 |
"eval_steps_per_second": 140.952,
|
| 8126 |
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.489571201694018,
|
| 8130 |
+
"grad_norm": 0.04014997556805611,
|
| 8131 |
+
"learning_rate": 2.9289321881345254e-05,
|
| 8132 |
+
"loss": 10.3332,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.4899947061937533,
|
| 8137 |
+
"grad_norm": 0.03895261138677597,
|
| 8138 |
+
"learning_rate": 2.9144089142794696e-05,
|
| 8139 |
+
"loss": 10.332,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.4904182106934886,
|
| 8144 |
+
"grad_norm": 0.037048932164907455,
|
| 8145 |
+
"learning_rate": 2.8999155926411202e-05,
|
| 8146 |
+
"loss": 10.3293,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.49084171519322395,
|
| 8151 |
+
"grad_norm": 0.04298751428723335,
|
| 8152 |
+
"learning_rate": 2.8854522844856614e-05,
|
| 8153 |
+
"loss": 10.33,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.4912652196929592,
|
| 8158 |
+
"grad_norm": 0.034429386258125305,
|
| 8159 |
+
"learning_rate": 2.871019050952395e-05,
|
| 8160 |
+
"loss": 10.3253,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.49168872419269455,
|
| 8165 |
+
"grad_norm": 0.03166850283741951,
|
| 8166 |
+
"learning_rate": 2.856615953053502e-05,
|
| 8167 |
+
"loss": 10.3286,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.4921122286924299,
|
| 8172 |
+
"grad_norm": 0.04494647681713104,
|
| 8173 |
+
"learning_rate": 2.842243051673773e-05,
|
| 8174 |
+
"loss": 10.3272,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.49253573319216515,
|
| 8179 |
+
"grad_norm": 0.0398864820599556,
|
| 8180 |
+
"learning_rate": 2.827900407570343e-05,
|
| 8181 |
+
"loss": 10.3259,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.4929592376919005,
|
| 8186 |
+
"grad_norm": 0.04810144007205963,
|
| 8187 |
+
"learning_rate": 2.8135880813724557e-05,
|
| 8188 |
+
"loss": 10.3253,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.4933827421916358,
|
| 8193 |
+
"grad_norm": 0.03601117059588432,
|
| 8194 |
+
"learning_rate": 2.799306133581182e-05,
|
| 8195 |
+
"loss": 10.3254,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.4938062466913711,
|
| 8200 |
+
"grad_norm": 0.044376593083143234,
|
| 8201 |
+
"learning_rate": 2.7850546245691867e-05,
|
| 8202 |
+
"loss": 10.3261,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.4942297511911064,
|
| 8207 |
+
"grad_norm": 0.04961050674319267,
|
| 8208 |
+
"learning_rate": 2.7708336145804625e-05,
|
| 8209 |
+
"loss": 10.3291,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.49465325569084173,
|
| 8214 |
+
"grad_norm": 0.029532672837376595,
|
| 8215 |
+
"learning_rate": 2.7566431637300738e-05,
|
| 8216 |
+
"loss": 10.3294,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.495076760190577,
|
| 8221 |
+
"grad_norm": 0.0377267524600029,
|
| 8222 |
+
"learning_rate": 2.742483332003909e-05,
|
| 8223 |
+
"loss": 10.3304,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.49550026469031233,
|
| 8228 |
+
"grad_norm": 0.04420830309391022,
|
| 8229 |
+
"learning_rate": 2.7283541792584168e-05,
|
| 8230 |
+
"loss": 10.3231,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.49592376919004766,
|
| 8235 |
+
"grad_norm": 0.04091082885861397,
|
| 8236 |
+
"learning_rate": 2.714255765220365e-05,
|
| 8237 |
+
"loss": 10.3283,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.49634727368978293,
|
| 8242 |
+
"grad_norm": 0.041371431201696396,
|
| 8243 |
+
"learning_rate": 2.7001881494865844e-05,
|
| 8244 |
+
"loss": 10.3289,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.49677077818951826,
|
| 8249 |
+
"grad_norm": 0.035521551966667175,
|
| 8250 |
+
"learning_rate": 2.686151391523708e-05,
|
| 8251 |
+
"loss": 10.3283,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.4971942826892536,
|
| 8256 |
+
"grad_norm": 0.04131960868835449,
|
| 8257 |
+
"learning_rate": 2.6721455506679326e-05,
|
| 8258 |
+
"loss": 10.3269,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.49761778718898886,
|
| 8263 |
+
"grad_norm": 0.032809097319841385,
|
| 8264 |
+
"learning_rate": 2.6581706861247647e-05,
|
| 8265 |
+
"loss": 10.3295,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.4980412916887242,
|
| 8270 |
+
"grad_norm": 0.04025952145457268,
|
| 8271 |
+
"learning_rate": 2.644226856968757e-05,
|
| 8272 |
+
"loss": 10.3295,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.4984647961884595,
|
| 8277 |
+
"grad_norm": 0.034779418259859085,
|
| 8278 |
+
"learning_rate": 2.630314122143278e-05,
|
| 8279 |
+
"loss": 10.3328,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.4988883006881948,
|
| 8284 |
+
"grad_norm": 0.030629297718405724,
|
| 8285 |
+
"learning_rate": 2.616432540460255e-05,
|
| 8286 |
+
"loss": 10.3285,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.4993118051879301,
|
| 8291 |
+
"grad_norm": 0.041318267583847046,
|
| 8292 |
+
"learning_rate": 2.60258217059992e-05,
|
| 8293 |
+
"loss": 10.3231,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.49973530968766544,
|
| 8298 |
+
"grad_norm": 0.03383251279592514,
|
| 8299 |
+
"learning_rate": 2.5887630711105702e-05,
|
| 8300 |
+
"loss": 10.328,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.5001588141874007,
|
| 8305 |
+
"grad_norm": 0.03106045164167881,
|
| 8306 |
+
"learning_rate": 2.5749753004083122e-05,
|
| 8307 |
+
"loss": 10.324,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.5005823186871361,
|
| 8312 |
+
"grad_norm": 0.0408039465546608,
|
| 8313 |
+
"learning_rate": 2.561218916776823e-05,
|
| 8314 |
+
"loss": 10.3301,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.5010058231868714,
|
| 8319 |
+
"grad_norm": 0.039002787321805954,
|
| 8320 |
+
"learning_rate": 2.5474939783671016e-05,
|
| 8321 |
+
"loss": 10.3252,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.5014293276866066,
|
| 8326 |
+
"grad_norm": 0.034892577677965164,
|
| 8327 |
+
"learning_rate": 2.5338005431972143e-05,
|
| 8328 |
+
"loss": 10.3327,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.501852832186342,
|
| 8333 |
+
"grad_norm": 0.03931496664881706,
|
| 8334 |
+
"learning_rate": 2.520138669152069e-05,
|
| 8335 |
+
"loss": 10.3263,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.5022763366860773,
|
| 8340 |
+
"grad_norm": 0.03569362685084343,
|
| 8341 |
+
"learning_rate": 2.506508413983144e-05,
|
| 8342 |
+
"loss": 10.3325,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.5026998411858126,
|
| 8347 |
+
"grad_norm": 0.03220602497458458,
|
| 8348 |
+
"learning_rate": 2.4929098353082692e-05,
|
| 8349 |
+
"loss": 10.3284,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.503123345685548,
|
| 8354 |
+
"grad_norm": 0.05201113596558571,
|
| 8355 |
+
"learning_rate": 2.4793429906113673e-05,
|
| 8356 |
+
"loss": 10.3219,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.5035468501852832,
|
| 8361 |
+
"grad_norm": 0.03895171731710434,
|
| 8362 |
+
"learning_rate": 2.465807937242216e-05,
|
| 8363 |
+
"loss": 10.3234,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.5039703546850185,
|
| 8368 |
+
"grad_norm": 0.03379373997449875,
|
| 8369 |
+
"learning_rate": 2.4523047324162085e-05,
|
| 8370 |
+
"loss": 10.326,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.5043938591847539,
|
| 8375 |
+
"grad_norm": 0.0329679436981678,
|
| 8376 |
+
"learning_rate": 2.4388334332140973e-05,
|
| 8377 |
+
"loss": 10.3301,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.5048173636844892,
|
| 8382 |
+
"grad_norm": 0.0346069410443306,
|
| 8383 |
+
"learning_rate": 2.425394096581772e-05,
|
| 8384 |
+
"loss": 10.3243,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.5052408681842244,
|
| 8389 |
+
"grad_norm": 0.03509392589330673,
|
| 8390 |
+
"learning_rate": 2.411986779330012e-05,
|
| 8391 |
+
"loss": 10.3267,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.5056643726839598,
|
| 8396 |
+
"grad_norm": 0.034833189100027084,
|
| 8397 |
+
"learning_rate": 2.3986115381342346e-05,
|
| 8398 |
+
"loss": 10.3246,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.5060878771836951,
|
| 8403 |
+
"grad_norm": 0.046047694981098175,
|
| 8404 |
+
"learning_rate": 2.3852684295342753e-05,
|
| 8405 |
+
"loss": 10.3244,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.5065113816834304,
|
| 8410 |
+
"grad_norm": 0.03512802720069885,
|
| 8411 |
+
"learning_rate": 2.3719575099341295e-05,
|
| 8412 |
+
"loss": 10.3318,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.5069348861831657,
|
| 8417 |
+
"grad_norm": 0.037637531757354736,
|
| 8418 |
+
"learning_rate": 2.358678835601731e-05,
|
| 8419 |
+
"loss": 10.3257,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.507358390682901,
|
| 8424 |
+
"grad_norm": 0.037445951253175735,
|
| 8425 |
+
"learning_rate": 2.345432462668702e-05,
|
| 8426 |
+
"loss": 10.3311,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.5077818951826363,
|
| 8431 |
+
"grad_norm": 0.04620397090911865,
|
| 8432 |
+
"learning_rate": 2.3322184471301222e-05,
|
| 8433 |
+
"loss": 10.3295,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.5082053996823717,
|
| 8438 |
+
"grad_norm": 0.04575134068727493,
|
| 8439 |
+
"learning_rate": 2.3190368448442935e-05,
|
| 8440 |
+
"loss": 10.3291,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.5086289041821069,
|
| 8445 |
+
"grad_norm": 0.0280012134462595,
|
| 8446 |
+
"learning_rate": 2.3058877115324885e-05,
|
| 8447 |
+
"loss": 10.328,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.5090524086818422,
|
| 8452 |
+
"grad_norm": 0.030293826013803482,
|
| 8453 |
+
"learning_rate": 2.292771102778739e-05,
|
| 8454 |
+
"loss": 10.3269,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.5094759131815776,
|
| 8459 |
+
"grad_norm": 0.03328132629394531,
|
| 8460 |
+
"learning_rate": 2.279687074029584e-05,
|
| 8461 |
+
"loss": 10.3284,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.5098994176813129,
|
| 8466 |
+
"grad_norm": 0.029259130358695984,
|
| 8467 |
+
"learning_rate": 2.266635680593845e-05,
|
| 8468 |
+
"loss": 10.3271,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.5103229221810481,
|
| 8473 |
+
"grad_norm": 0.04118121415376663,
|
| 8474 |
+
"learning_rate": 2.2536169776423767e-05,
|
| 8475 |
+
"loss": 10.3312,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.5107464266807835,
|
| 8480 |
+
"grad_norm": 0.049123574048280716,
|
| 8481 |
+
"learning_rate": 2.2406310202078585e-05,
|
| 8482 |
+
"loss": 10.3216,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.5111699311805188,
|
| 8487 |
+
"grad_norm": 0.03439442068338394,
|
| 8488 |
+
"learning_rate": 2.227677863184533e-05,
|
| 8489 |
+
"loss": 10.3315,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.5115934356802541,
|
| 8494 |
+
"grad_norm": 0.03493933379650116,
|
| 8495 |
+
"learning_rate": 2.2147575613280015e-05,
|
| 8496 |
+
"loss": 10.328,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.5120169401799894,
|
| 8501 |
+
"grad_norm": 0.03778016194701195,
|
| 8502 |
+
"learning_rate": 2.2018701692549736e-05,
|
| 8503 |
+
"loss": 10.3271,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.5124404446797247,
|
| 8508 |
+
"grad_norm": 0.04461103677749634,
|
| 8509 |
+
"learning_rate": 2.1890157414430444e-05,
|
| 8510 |
+
"loss": 10.3271,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.51286394917946,
|
| 8515 |
+
"grad_norm": 0.0383584089577198,
|
| 8516 |
+
"learning_rate": 2.1761943322304656e-05,
|
| 8517 |
+
"loss": 10.3342,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.5132874536791954,
|
| 8522 |
+
"grad_norm": 0.034361306577920914,
|
| 8523 |
+
"learning_rate": 2.1634059958159036e-05,
|
| 8524 |
+
"loss": 10.3231,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.5137109581789306,
|
| 8529 |
+
"grad_norm": 0.03422437235713005,
|
| 8530 |
+
"learning_rate": 2.150650786258227e-05,
|
| 8531 |
+
"loss": 10.3293,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.5141344626786659,
|
| 8536 |
+
"grad_norm": 0.028060471639037132,
|
| 8537 |
+
"learning_rate": 2.137928757476272e-05,
|
| 8538 |
+
"loss": 10.3275,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.5145579671784013,
|
| 8543 |
+
"grad_norm": 0.03801143169403076,
|
| 8544 |
+
"learning_rate": 2.125239963248604e-05,
|
| 8545 |
+
"loss": 10.3257,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.5149814716781366,
|
| 8550 |
+
"grad_norm": 0.031281728297472,
|
| 8551 |
+
"learning_rate": 2.11258445721331e-05,
|
| 8552 |
+
"loss": 10.3275,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.5154049761778718,
|
| 8557 |
+
"grad_norm": 0.037522025406360626,
|
| 8558 |
+
"learning_rate": 2.099962292867751e-05,
|
| 8559 |
+
"loss": 10.3324,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.5158284806776072,
|
| 8564 |
+
"grad_norm": 0.03667249158024788,
|
| 8565 |
+
"learning_rate": 2.0873735235683535e-05,
|
| 8566 |
+
"loss": 10.3256,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.5162519851773425,
|
| 8571 |
+
"grad_norm": 0.03992689400911331,
|
| 8572 |
+
"learning_rate": 2.074818202530373e-05,
|
| 8573 |
+
"loss": 10.3233,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.5166754896770778,
|
| 8578 |
+
"grad_norm": 0.03932832181453705,
|
| 8579 |
+
"learning_rate": 2.0622963828276742e-05,
|
| 8580 |
+
"loss": 10.3329,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.5170989941768132,
|
| 8585 |
+
"grad_norm": 0.03914317116141319,
|
| 8586 |
+
"learning_rate": 2.0498081173925067e-05,
|
| 8587 |
+
"loss": 10.329,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.5175224986765484,
|
| 8592 |
+
"grad_norm": 0.034946508705616,
|
| 8593 |
+
"learning_rate": 2.037353459015272e-05,
|
| 8594 |
+
"loss": 10.3257,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.5179460031762837,
|
| 8599 |
+
"grad_norm": 0.042370010167360306,
|
| 8600 |
+
"learning_rate": 2.0249324603443143e-05,
|
| 8601 |
+
"loss": 10.328,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.5183695076760191,
|
| 8606 |
+
"grad_norm": 0.037683628499507904,
|
| 8607 |
+
"learning_rate": 2.0125451738856904e-05,
|
| 8608 |
+
"loss": 10.3281,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.5187930121757544,
|
| 8613 |
+
"grad_norm": 0.03259679675102234,
|
| 8614 |
+
"learning_rate": 2.0001916520029494e-05,
|
| 8615 |
+
"loss": 10.3293,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.5192165166754896,
|
| 8620 |
+
"grad_norm": 0.03760567307472229,
|
| 8621 |
+
"learning_rate": 1.9878719469169105e-05,
|
| 8622 |
+
"loss": 10.3208,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.519640021175225,
|
| 8627 |
+
"grad_norm": 0.03157936409115791,
|
| 8628 |
+
"learning_rate": 1.975586110705435e-05,
|
| 8629 |
+
"loss": 10.3262,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.5200635256749603,
|
| 8634 |
+
"grad_norm": 0.04091958329081535,
|
| 8635 |
+
"learning_rate": 1.9633341953032245e-05,
|
| 8636 |
+
"loss": 10.3222,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.5204870301746956,
|
| 8641 |
+
"grad_norm": 0.03274989873170853,
|
| 8642 |
+
"learning_rate": 1.951116252501588e-05,
|
| 8643 |
+
"loss": 10.33,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.5209105346744309,
|
| 8648 |
+
"grad_norm": 0.03612561151385307,
|
| 8649 |
+
"learning_rate": 1.9389323339482203e-05,
|
| 8650 |
+
"loss": 10.3293,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.5213340391741662,
|
| 8655 |
+
"grad_norm": 0.043458499014377594,
|
| 8656 |
+
"learning_rate": 1.9267824911469977e-05,
|
| 8657 |
+
"loss": 10.3279,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.5217575436739015,
|
| 8662 |
+
"grad_norm": 0.03678362816572189,
|
| 8663 |
+
"learning_rate": 1.9146667754577408e-05,
|
| 8664 |
+
"loss": 10.3308,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.5221810481736369,
|
| 8669 |
+
"grad_norm": 0.03507538139820099,
|
| 8670 |
+
"learning_rate": 1.902585238096014e-05,
|
| 8671 |
+
"loss": 10.3288,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.5226045526733721,
|
| 8676 |
+
"grad_norm": 0.03404875099658966,
|
| 8677 |
+
"learning_rate": 1.890537930132903e-05,
|
| 8678 |
+
"loss": 10.3279,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.5230280571731075,
|
| 8683 |
+
"grad_norm": 0.03995542228221893,
|
| 8684 |
+
"learning_rate": 1.8785249024948026e-05,
|
| 8685 |
+
"loss": 10.3274,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.5234515616728428,
|
| 8690 |
+
"grad_norm": 0.028712738305330276,
|
| 8691 |
+
"learning_rate": 1.8665462059631867e-05,
|
| 8692 |
+
"loss": 10.3291,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.5238750661725781,
|
| 8697 |
+
"grad_norm": 0.038584861904382706,
|
| 8698 |
+
"learning_rate": 1.8546018911744168e-05,
|
| 8699 |
+
"loss": 10.3245,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.5242985706723134,
|
| 8704 |
+
"grad_norm": 0.04092830419540405,
|
| 8705 |
+
"learning_rate": 1.8426920086195065e-05,
|
| 8706 |
+
"loss": 10.3305,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.5247220751720487,
|
| 8711 |
+
"grad_norm": 0.028606008738279343,
|
| 8712 |
+
"learning_rate": 1.8308166086439237e-05,
|
| 8713 |
+
"loss": 10.3277,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.525145579671784,
|
| 8718 |
+
"grad_norm": 0.03429469093680382,
|
| 8719 |
+
"learning_rate": 1.8189757414473684e-05,
|
| 8720 |
+
"loss": 10.3263,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.5255690841715194,
|
| 8725 |
+
"grad_norm": 0.03285999968647957,
|
| 8726 |
+
"learning_rate": 1.807169457083565e-05,
|
| 8727 |
+
"loss": 10.3263,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.5259925886712546,
|
| 8732 |
+
"grad_norm": 0.030071139335632324,
|
| 8733 |
+
"learning_rate": 1.7953978054600527e-05,
|
| 8734 |
+
"loss": 10.3262,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.5264160931709899,
|
| 8739 |
+
"grad_norm": 0.03396671265363693,
|
| 8740 |
+
"learning_rate": 1.7836608363379604e-05,
|
| 8741 |
+
"loss": 10.3267,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.5268395976707253,
|
| 8746 |
+
"grad_norm": 0.03475523740053177,
|
| 8747 |
+
"learning_rate": 1.7719585993318178e-05,
|
| 8748 |
+
"loss": 10.3264,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.5272631021704606,
|
| 8753 |
+
"grad_norm": 0.04322926327586174,
|
| 8754 |
+
"learning_rate": 1.7602911439093305e-05,
|
| 8755 |
+
"loss": 10.3281,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.5276866066701958,
|
| 8760 |
+
"grad_norm": 0.03885841369628906,
|
| 8761 |
+
"learning_rate": 1.7486585193911786e-05,
|
| 8762 |
+
"loss": 10.3248,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.5281101111699312,
|
| 8767 |
+
"grad_norm": 0.03156493604183197,
|
| 8768 |
+
"learning_rate": 1.7370607749507985e-05,
|
| 8769 |
+
"loss": 10.3278,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.5285336156696665,
|
| 8774 |
+
"grad_norm": 0.043746136128902435,
|
| 8775 |
+
"learning_rate": 1.7254979596141886e-05,
|
| 8776 |
+
"loss": 10.3254,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.5289571201694018,
|
| 8781 |
+
"grad_norm": 0.03123902529478073,
|
| 8782 |
+
"learning_rate": 1.713970122259688e-05,
|
| 8783 |
+
"loss": 10.3256,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.5293806246691372,
|
| 8788 |
+
"grad_norm": 0.04213595762848854,
|
| 8789 |
+
"learning_rate": 1.702477311617784e-05,
|
| 8790 |
+
"loss": 10.3245,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.5298041291688724,
|
| 8795 |
+
"grad_norm": 0.0351736918091774,
|
| 8796 |
+
"learning_rate": 1.6910195762708926e-05,
|
| 8797 |
+
"loss": 10.3243,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.5302276336686077,
|
| 8802 |
+
"grad_norm": 0.035408440977334976,
|
| 8803 |
+
"learning_rate": 1.6795969646531685e-05,
|
| 8804 |
+
"loss": 10.3265,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.5306511381683431,
|
| 8809 |
+
"grad_norm": 0.036797985434532166,
|
| 8810 |
+
"learning_rate": 1.668209525050277e-05,
|
| 8811 |
+
"loss": 10.329,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.5310746426680784,
|
| 8816 |
+
"grad_norm": 0.04717336595058441,
|
| 8817 |
+
"learning_rate": 1.6568573055992187e-05,
|
| 8818 |
+
"loss": 10.3258,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.5314981471678136,
|
| 8823 |
+
"grad_norm": 0.038243409246206284,
|
| 8824 |
+
"learning_rate": 1.645540354288102e-05,
|
| 8825 |
+
"loss": 10.3287,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.531921651667549,
|
| 8830 |
+
"grad_norm": 0.04253392294049263,
|
| 8831 |
+
"learning_rate": 1.6342587189559576e-05,
|
| 8832 |
+
"loss": 10.3258,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.5323451561672843,
|
| 8837 |
+
"grad_norm": 0.032199427485466,
|
| 8838 |
+
"learning_rate": 1.6230124472925177e-05,
|
| 8839 |
+
"loss": 10.3263,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.5327686606670196,
|
| 8844 |
+
"grad_norm": 0.04099227860569954,
|
| 8845 |
+
"learning_rate": 1.6118015868380388e-05,
|
| 8846 |
+
"loss": 10.3288,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.5331921651667549,
|
| 8851 |
+
"grad_norm": 0.045609183609485626,
|
| 8852 |
+
"learning_rate": 1.6006261849830727e-05,
|
| 8853 |
+
"loss": 10.3271,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.5336156696664902,
|
| 8858 |
+
"grad_norm": 0.0431225411593914,
|
| 8859 |
+
"learning_rate": 1.5894862889682905e-05,
|
| 8860 |
+
"loss": 10.3208,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.5340391741662255,
|
| 8865 |
+
"grad_norm": 0.043088626116514206,
|
| 8866 |
+
"learning_rate": 1.578381945884271e-05,
|
| 8867 |
+
"loss": 10.3252,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.5344626786659609,
|
| 8872 |
+
"grad_norm": 0.035403814166784286,
|
| 8873 |
+
"learning_rate": 1.5673132026713046e-05,
|
| 8874 |
+
"loss": 10.3289,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.5348861831656961,
|
| 8879 |
+
"grad_norm": 0.049196433275938034,
|
| 8880 |
+
"learning_rate": 1.5562801061191868e-05,
|
| 8881 |
+
"loss": 10.3255,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.5353096876654314,
|
| 8886 |
+
"grad_norm": 0.035701338201761246,
|
| 8887 |
+
"learning_rate": 1.5452827028670357e-05,
|
| 8888 |
+
"loss": 10.3224,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.5357331921651668,
|
| 8893 |
+
"grad_norm": 0.03386475145816803,
|
| 8894 |
+
"learning_rate": 1.534321039403085e-05,
|
| 8895 |
+
"loss": 10.3294,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.5361566966649021,
|
| 8900 |
+
"grad_norm": 0.043618202209472656,
|
| 8901 |
+
"learning_rate": 1.523395162064486e-05,
|
| 8902 |
+
"loss": 10.3255,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.5365802011646373,
|
| 8907 |
+
"grad_norm": 0.03436095267534256,
|
| 8908 |
+
"learning_rate": 1.5125051170371197e-05,
|
| 8909 |
+
"loss": 10.3231,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.5370037056643727,
|
| 8914 |
+
"grad_norm": 0.04222296178340912,
|
| 8915 |
+
"learning_rate": 1.50165095035539e-05,
|
| 8916 |
+
"loss": 10.3309,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.537427210164108,
|
| 8921 |
+
"grad_norm": 0.026362288743257523,
|
| 8922 |
+
"learning_rate": 1.4908327079020423e-05,
|
| 8923 |
+
"loss": 10.3293,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.5378507146638433,
|
| 8928 |
+
"grad_norm": 0.043024007230997086,
|
| 8929 |
+
"learning_rate": 1.480050435407957e-05,
|
| 8930 |
+
"loss": 10.3268,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.5382742191635786,
|
| 8935 |
+
"grad_norm": 0.036363277584314346,
|
| 8936 |
+
"learning_rate": 1.4693041784519678e-05,
|
| 8937 |
+
"loss": 10.3279,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.5386977236633139,
|
| 8942 |
+
"grad_norm": 0.04140538349747658,
|
| 8943 |
+
"learning_rate": 1.458593982460662e-05,
|
| 8944 |
+
"loss": 10.3274,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.5391212281630492,
|
| 8949 |
+
"grad_norm": 0.0490729995071888,
|
| 8950 |
+
"learning_rate": 1.4479198927081906e-05,
|
| 8951 |
+
"loss": 10.3273,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.5395447326627846,
|
| 8956 |
+
"grad_norm": 0.02927100658416748,
|
| 8957 |
+
"learning_rate": 1.4372819543160709e-05,
|
| 8958 |
+
"loss": 10.3247,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.5399682371625198,
|
| 8963 |
+
"grad_norm": 0.03926154226064682,
|
| 8964 |
+
"learning_rate": 1.4266802122530066e-05,
|
| 8965 |
+
"loss": 10.3242,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.5403917416622551,
|
| 8970 |
+
"grad_norm": 0.042115967720746994,
|
| 8971 |
+
"learning_rate": 1.4161147113346918e-05,
|
| 8972 |
+
"loss": 10.3312,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.5408152461619905,
|
| 8977 |
+
"grad_norm": 0.03355708345770836,
|
| 8978 |
+
"learning_rate": 1.4055854962236204e-05,
|
| 8979 |
+
"loss": 10.3278,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.5412387506617258,
|
| 8984 |
+
"grad_norm": 0.0461774580180645,
|
| 8985 |
+
"learning_rate": 1.395092611428902e-05,
|
| 8986 |
+
"loss": 10.3289,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.541662255161461,
|
| 8991 |
+
"grad_norm": 0.048963211476802826,
|
| 8992 |
+
"learning_rate": 1.384636101306066e-05,
|
| 8993 |
+
"loss": 10.3287,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.5420857596611964,
|
| 8998 |
+
"grad_norm": 0.042128488421440125,
|
| 8999 |
+
"learning_rate": 1.374216010056879e-05,
|
| 9000 |
+
"loss": 10.3321,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.5425092641609317,
|
| 9005 |
+
"grad_norm": 0.08573438227176666,
|
| 9006 |
+
"learning_rate": 1.363832381729162e-05,
|
| 9007 |
+
"loss": 10.3282,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.542932768660667,
|
| 9012 |
+
"grad_norm": 0.043098084628582,
|
| 9013 |
+
"learning_rate": 1.353485260216596e-05,
|
| 9014 |
+
"loss": 10.3251,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.5433562731604024,
|
| 9019 |
+
"grad_norm": 0.041679129004478455,
|
| 9020 |
+
"learning_rate": 1.3431746892585462e-05,
|
| 9021 |
+
"loss": 10.3307,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.5437797776601376,
|
| 9026 |
+
"grad_norm": 0.038137055933475494,
|
| 9027 |
+
"learning_rate": 1.3329007124398608e-05,
|
| 9028 |
+
"loss": 10.3268,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.5442032821598729,
|
| 9033 |
+
"grad_norm": 0.0336623378098011,
|
| 9034 |
+
"learning_rate": 1.3226633731907068e-05,
|
| 9035 |
+
"loss": 10.3257,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.5446267866596083,
|
| 9040 |
+
"grad_norm": 0.04125112295150757,
|
| 9041 |
+
"learning_rate": 1.3124627147863733e-05,
|
| 9042 |
+
"loss": 10.3224,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.5450502911593436,
|
| 9047 |
+
"grad_norm": 0.03276829421520233,
|
| 9048 |
+
"learning_rate": 1.302298780347092e-05,
|
| 9049 |
+
"loss": 10.3258,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.5454737956590788,
|
| 9054 |
+
"grad_norm": 0.03559970483183861,
|
| 9055 |
+
"learning_rate": 1.2921716128378581e-05,
|
| 9056 |
+
"loss": 10.3287,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.5458973001588142,
|
| 9061 |
+
"grad_norm": 0.05309802666306496,
|
| 9062 |
+
"learning_rate": 1.2820812550682359e-05,
|
| 9063 |
+
"loss": 10.3329,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.5463208046585495,
|
| 9068 |
+
"grad_norm": 0.045987412333488464,
|
| 9069 |
+
"learning_rate": 1.2720277496922029e-05,
|
| 9070 |
+
"loss": 10.3269,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.5467443091582848,
|
| 9075 |
+
"grad_norm": 0.0374799519777298,
|
| 9076 |
+
"learning_rate": 1.262011139207937e-05,
|
| 9077 |
+
"loss": 10.328,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.5471678136580201,
|
| 9082 |
+
"grad_norm": 0.041651055216789246,
|
| 9083 |
+
"learning_rate": 1.2520314659576681e-05,
|
| 9084 |
+
"loss": 10.3261,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.5475913181577554,
|
| 9089 |
+
"grad_norm": 0.03640596568584442,
|
| 9090 |
+
"learning_rate": 1.2420887721274798e-05,
|
| 9091 |
+
"loss": 10.3312,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.5480148226574907,
|
| 9096 |
+
"grad_norm": 0.03759313374757767,
|
| 9097 |
+
"learning_rate": 1.2321830997471328e-05,
|
| 9098 |
+
"loss": 10.3232,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.5484383271572261,
|
| 9103 |
+
"grad_norm": 0.030599461868405342,
|
| 9104 |
+
"learning_rate": 1.2223144906898953e-05,
|
| 9105 |
+
"loss": 10.3276,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.5488618316569613,
|
| 9110 |
+
"grad_norm": 0.04197867587208748,
|
| 9111 |
+
"learning_rate": 1.2124829866723608e-05,
|
| 9112 |
+
"loss": 10.3315,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.5492853361566966,
|
| 9117 |
+
"grad_norm": 0.03175166994333267,
|
| 9118 |
+
"learning_rate": 1.2026886292542695e-05,
|
| 9119 |
+
"loss": 10.323,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.549708840656432,
|
| 9124 |
+
"grad_norm": 0.044329218566417694,
|
| 9125 |
+
"learning_rate": 1.1929314598383423e-05,
|
| 9126 |
+
"loss": 10.3214,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.5501323451561673,
|
| 9131 |
+
"grad_norm": 0.03441780060529709,
|
| 9132 |
+
"learning_rate": 1.1832115196700877e-05,
|
| 9133 |
+
"loss": 10.3258,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.5505558496559025,
|
| 9138 |
+
"grad_norm": 0.038743555545806885,
|
| 9139 |
+
"learning_rate": 1.1735288498376495e-05,
|
| 9140 |
+
"loss": 10.3215,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.5509793541556379,
|
| 9145 |
+
"grad_norm": 0.031063832342624664,
|
| 9146 |
+
"learning_rate": 1.1638834912716157e-05,
|
| 9147 |
+
"loss": 10.3242,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.5514028586553732,
|
| 9152 |
+
"grad_norm": 0.03892136737704277,
|
| 9153 |
+
"learning_rate": 1.1542754847448546e-05,
|
| 9154 |
+
"loss": 10.3263,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.5518263631551085,
|
| 9159 |
+
"grad_norm": 0.03525496646761894,
|
| 9160 |
+
"learning_rate": 1.1447048708723396e-05,
|
| 9161 |
+
"loss": 10.3229,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.5522498676548439,
|
| 9166 |
+
"grad_norm": 0.03807550668716431,
|
| 9167 |
+
"learning_rate": 1.13517169011098e-05,
|
| 9168 |
+
"loss": 10.3304,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.5526733721545791,
|
| 9173 |
+
"grad_norm": 0.028498059138655663,
|
| 9174 |
+
"learning_rate": 1.125675982759442e-05,
|
| 9175 |
+
"loss": 10.3264,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.5530968766543145,
|
| 9180 |
+
"grad_norm": 0.05457071214914322,
|
| 9181 |
+
"learning_rate": 1.1162177889579905e-05,
|
| 9182 |
+
"loss": 10.3308,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.5535203811540498,
|
| 9187 |
+
"grad_norm": 0.03866454213857651,
|
| 9188 |
+
"learning_rate": 1.10679714868831e-05,
|
| 9189 |
+
"loss": 10.3257,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.553943885653785,
|
| 9194 |
+
"grad_norm": 0.03566194698214531,
|
| 9195 |
+
"learning_rate": 1.0974141017733386e-05,
|
| 9196 |
+
"loss": 10.328,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.5543673901535204,
|
| 9201 |
+
"grad_norm": 0.03313031047582626,
|
| 9202 |
+
"learning_rate": 1.0880686878771018e-05,
|
| 9203 |
+
"loss": 10.3269,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.5547908946532557,
|
| 9208 |
+
"grad_norm": 0.03430391848087311,
|
| 9209 |
+
"learning_rate": 1.0787609465045389e-05,
|
| 9210 |
+
"loss": 10.3222,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.555214399152991,
|
| 9215 |
+
"grad_norm": 0.03857209533452988,
|
| 9216 |
+
"learning_rate": 1.0694909170013446e-05,
|
| 9217 |
+
"loss": 10.3285,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.5556379036527264,
|
| 9222 |
+
"grad_norm": 0.038265954703092575,
|
| 9223 |
+
"learning_rate": 1.0602586385537928e-05,
|
| 9224 |
+
"loss": 10.3251,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.5560614081524616,
|
| 9229 |
+
"grad_norm": 0.03672931715846062,
|
| 9230 |
+
"learning_rate": 1.0510641501885798e-05,
|
| 9231 |
+
"loss": 10.3275,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.5564849126521969,
|
| 9236 |
+
"grad_norm": 0.03286921605467796,
|
| 9237 |
+
"learning_rate": 1.0419074907726579e-05,
|
| 9238 |
+
"loss": 10.3257,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.5569084171519323,
|
| 9243 |
+
"grad_norm": 0.03168274834752083,
|
| 9244 |
+
"learning_rate": 1.0327886990130597e-05,
|
| 9245 |
+
"loss": 10.3272,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.5573319216516676,
|
| 9250 |
+
"grad_norm": 0.03598780184984207,
|
| 9251 |
+
"learning_rate": 1.0237078134567535e-05,
|
| 9252 |
+
"loss": 10.3252,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.5577554261514028,
|
| 9257 |
+
"grad_norm": 0.03541802614927292,
|
| 9258 |
+
"learning_rate": 1.0146648724904662e-05,
|
| 9259 |
+
"loss": 10.3282,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.5581789306511382,
|
| 9264 |
+
"grad_norm": 0.03859534487128258,
|
| 9265 |
+
"learning_rate": 1.0056599143405243e-05,
|
| 9266 |
+
"loss": 10.3251,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.5586024351508735,
|
| 9271 |
+
"grad_norm": 0.03148134425282478,
|
| 9272 |
+
"learning_rate": 9.966929770726973e-06,
|
| 9273 |
+
"loss": 10.3294,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.5590259396506088,
|
| 9278 |
+
"grad_norm": 0.03948875889182091,
|
| 9279 |
+
"learning_rate": 9.877640985920267e-06,
|
| 9280 |
+
"loss": 10.3244,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.5594494441503441,
|
| 9285 |
+
"grad_norm": 0.033598676323890686,
|
| 9286 |
+
"learning_rate": 9.788733166426789e-06,
|
| 9287 |
+
"loss": 10.3235,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.5598729486500794,
|
| 9292 |
+
"grad_norm": 0.03082294762134552,
|
| 9293 |
+
"learning_rate": 9.700206688077706e-06,
|
| 9294 |
+
"loss": 10.3309,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.5602964531498147,
|
| 9299 |
+
"grad_norm": 0.0461772195994854,
|
| 9300 |
+
"learning_rate": 9.612061925092253e-06,
|
| 9301 |
+
"loss": 10.3286,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.5607199576495501,
|
| 9306 |
+
"grad_norm": 0.043604668229818344,
|
| 9307 |
+
"learning_rate": 9.524299250076053e-06,
|
| 9308 |
+
"loss": 10.3249,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.5611434621492853,
|
| 9313 |
+
"grad_norm": 0.04446884244680405,
|
| 9314 |
+
"learning_rate": 9.436919034019531e-06,
|
| 9315 |
+
"loss": 10.3259,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.5615669666490206,
|
| 9320 |
+
"grad_norm": 0.03795298561453819,
|
| 9321 |
+
"learning_rate": 9.349921646296423e-06,
|
| 9322 |
+
"loss": 10.325,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.561990471148756,
|
| 9327 |
+
"grad_norm": 0.03730008751153946,
|
| 9328 |
+
"learning_rate": 9.263307454662174e-06,
|
| 9329 |
+
"loss": 10.3252,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.5624139756484913,
|
| 9334 |
+
"grad_norm": 0.047635212540626526,
|
| 9335 |
+
"learning_rate": 9.17707682525235e-06,
|
| 9336 |
+
"loss": 10.3216,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.5628374801482265,
|
| 9341 |
+
"grad_norm": 0.038714662194252014,
|
| 9342 |
+
"learning_rate": 9.091230122581184e-06,
|
| 9343 |
+
"loss": 10.3299,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.5632609846479619,
|
| 9348 |
+
"grad_norm": 0.03284423053264618,
|
| 9349 |
+
"learning_rate": 9.00576770953987e-06,
|
| 9350 |
+
"loss": 10.326,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.5636844891476972,
|
| 9355 |
+
"grad_norm": 0.033383384346961975,
|
| 9356 |
+
"learning_rate": 8.920689947395232e-06,
|
| 9357 |
+
"loss": 10.3238,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.5641079936474325,
|
| 9362 |
+
"grad_norm": 0.03223501518368721,
|
| 9363 |
+
"learning_rate": 8.83599719578807e-06,
|
| 9364 |
+
"loss": 10.3232,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.5645314981471679,
|
| 9369 |
+
"grad_norm": 0.04168533906340599,
|
| 9370 |
+
"learning_rate": 8.751689812731612e-06,
|
| 9371 |
+
"loss": 10.3292,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.5649550026469031,
|
| 9376 |
+
"grad_norm": 0.033274777233600616,
|
| 9377 |
+
"learning_rate": 8.667768154610123e-06,
|
| 9378 |
+
"loss": 10.3264,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.5653785071466384,
|
| 9383 |
+
"grad_norm": 0.043340083211660385,
|
| 9384 |
+
"learning_rate": 8.584232576177343e-06,
|
| 9385 |
+
"loss": 10.3238,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.5658020116463738,
|
| 9390 |
+
"grad_norm": 0.04212178289890289,
|
| 9391 |
+
"learning_rate": 8.501083430554868e-06,
|
| 9392 |
+
"loss": 10.3229,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.566225516146109,
|
| 9397 |
+
"grad_norm": 0.03570551052689552,
|
| 9398 |
+
"learning_rate": 8.418321069230894e-06,
|
| 9399 |
+
"loss": 10.3289,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.5666490206458443,
|
| 9404 |
+
"grad_norm": 0.043488387018442154,
|
| 9405 |
+
"learning_rate": 8.335945842058523e-06,
|
| 9406 |
+
"loss": 10.3308,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.5670725251455797,
|
| 9411 |
+
"grad_norm": 0.03983880206942558,
|
| 9412 |
+
"learning_rate": 8.253958097254388e-06,
|
| 9413 |
+
"loss": 10.3296,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.567496029645315,
|
| 9418 |
+
"grad_norm": 0.05708296224474907,
|
| 9419 |
+
"learning_rate": 8.172358181397178e-06,
|
| 9420 |
+
"loss": 10.3222,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.5679195341450503,
|
| 9425 |
+
"grad_norm": 0.041659433394670486,
|
| 9426 |
+
"learning_rate": 8.091146439426089e-06,
|
| 9427 |
+
"loss": 10.3297,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.5683430386447856,
|
| 9432 |
+
"grad_norm": 0.03511781617999077,
|
| 9433 |
+
"learning_rate": 8.010323214639492e-06,
|
| 9434 |
+
"loss": 10.3291,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.5687665431445209,
|
| 9439 |
+
"grad_norm": 0.038728345185518265,
|
| 9440 |
+
"learning_rate": 7.929888848693378e-06,
|
| 9441 |
+
"loss": 10.3299,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.5691900476442562,
|
| 9446 |
+
"grad_norm": 0.037921249866485596,
|
| 9447 |
+
"learning_rate": 7.849843681599977e-06,
|
| 9448 |
+
"loss": 10.3258,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.5696135521439916,
|
| 9453 |
+
"grad_norm": 0.03447211533784866,
|
| 9454 |
+
"learning_rate": 7.77018805172629e-06,
|
| 9455 |
+
"loss": 10.3263,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.5700370566437268,
|
| 9460 |
+
"grad_norm": 0.03284087032079697,
|
| 9461 |
+
"learning_rate": 7.690922295792646e-06,
|
| 9462 |
+
"loss": 10.3278,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.5704605611434621,
|
| 9467 |
+
"grad_norm": 0.03763739392161369,
|
| 9468 |
+
"learning_rate": 7.612046748871327e-06,
|
| 9469 |
+
"loss": 10.3237,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.5708840656431975,
|
| 9474 |
+
"grad_norm": 0.035049282014369965,
|
| 9475 |
+
"learning_rate": 7.53356174438512e-06,
|
| 9476 |
+
"loss": 10.3244,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.5713075701429328,
|
| 9481 |
+
"grad_norm": 0.042994335293769836,
|
| 9482 |
+
"learning_rate": 7.455467614105893e-06,
|
| 9483 |
+
"loss": 10.3283,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.571731074642668,
|
| 9488 |
+
"grad_norm": 0.036672353744506836,
|
| 9489 |
+
"learning_rate": 7.377764688153243e-06,
|
| 9490 |
+
"loss": 10.3236,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.5721545791424034,
|
| 9495 |
+
"grad_norm": 0.035631466656923294,
|
| 9496 |
+
"learning_rate": 7.300453294993004e-06,
|
| 9497 |
+
"loss": 10.329,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.5725780836421387,
|
| 9502 |
+
"grad_norm": 0.03919555991888046,
|
| 9503 |
+
"learning_rate": 7.223533761435985e-06,
|
| 9504 |
+
"loss": 10.327,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.573001588141874,
|
| 9509 |
+
"grad_norm": 0.02915988676249981,
|
| 9510 |
+
"learning_rate": 7.147006412636536e-06,
|
| 9511 |
+
"loss": 10.33,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.5734250926416093,
|
| 9516 |
+
"grad_norm": 0.055577103048563004,
|
| 9517 |
+
"learning_rate": 7.070871572091076e-06,
|
| 9518 |
+
"loss": 10.3237,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.5738485971413446,
|
| 9523 |
+
"grad_norm": 0.04068098962306976,
|
| 9524 |
+
"learning_rate": 6.995129561636904e-06,
|
| 9525 |
+
"loss": 10.3236,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.5742721016410799,
|
| 9530 |
+
"grad_norm": 0.03457869589328766,
|
| 9531 |
+
"learning_rate": 6.919780701450685e-06,
|
| 9532 |
+
"loss": 10.3235,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.5746956061408153,
|
| 9537 |
+
"grad_norm": 0.03804881125688553,
|
| 9538 |
+
"learning_rate": 6.844825310047209e-06,
|
| 9539 |
+
"loss": 10.3236,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.5751191106405505,
|
| 9544 |
+
"grad_norm": 0.03735931217670441,
|
| 9545 |
+
"learning_rate": 6.770263704277957e-06,
|
| 9546 |
+
"loss": 10.3262,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.5755426151402858,
|
| 9551 |
+
"grad_norm": 0.038536667823791504,
|
| 9552 |
+
"learning_rate": 6.696096199329816e-06,
|
| 9553 |
+
"loss": 10.3246,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.5759661196400212,
|
| 9558 |
+
"grad_norm": 0.04373512417078018,
|
| 9559 |
+
"learning_rate": 6.6223231087237495e-06,
|
| 9560 |
+
"loss": 10.3259,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.5763896241397565,
|
| 9565 |
+
"grad_norm": 0.044846098870038986,
|
| 9566 |
+
"learning_rate": 6.54894474431339e-06,
|
| 9567 |
+
"loss": 10.334,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.5768131286394917,
|
| 9572 |
+
"grad_norm": 0.04429624229669571,
|
| 9573 |
+
"learning_rate": 6.4759614162838375e-06,
|
| 9574 |
+
"loss": 10.3257,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.5772366331392271,
|
| 9579 |
+
"grad_norm": 0.031095974147319794,
|
| 9580 |
+
"learning_rate": 6.403373433150306e-06,
|
| 9581 |
+
"loss": 10.326,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.5776601376389624,
|
| 9586 |
+
"grad_norm": 0.04691673815250397,
|
| 9587 |
+
"learning_rate": 6.331181101756734e-06,
|
| 9588 |
+
"loss": 10.3281,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.5780836421386977,
|
| 9593 |
+
"grad_norm": 0.0471339076757431,
|
| 9594 |
+
"learning_rate": 6.259384727274631e-06,
|
| 9595 |
+
"loss": 10.3215,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.5785071466384331,
|
| 9600 |
+
"grad_norm": 0.04510781541466713,
|
| 9601 |
+
"learning_rate": 6.187984613201703e-06,
|
| 9602 |
+
"loss": 10.33,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.5789306511381683,
|
| 9607 |
+
"grad_norm": 0.048788487911224365,
|
| 9608 |
+
"learning_rate": 6.116981061360539e-06,
|
| 9609 |
+
"loss": 10.3255,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.5793541556379036,
|
| 9614 |
+
"grad_norm": 0.04672354832291603,
|
| 9615 |
+
"learning_rate": 6.0463743718974455e-06,
|
| 9616 |
+
"loss": 10.3316,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.579777660137639,
|
| 9621 |
+
"grad_norm": 0.036004580557346344,
|
| 9622 |
+
"learning_rate": 5.976164843281074e-06,
|
| 9623 |
+
"loss": 10.3302,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.5802011646373743,
|
| 9628 |
+
"grad_norm": 0.036676954478025436,
|
| 9629 |
+
"learning_rate": 5.906352772301193e-06,
|
| 9630 |
+
"loss": 10.3305,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.5806246691371095,
|
| 9635 |
+
"grad_norm": 0.049092430621385574,
|
| 9636 |
+
"learning_rate": 5.836938454067475e-06,
|
| 9637 |
+
"loss": 10.3271,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.5810481736368449,
|
| 9642 |
+
"grad_norm": 0.039018791168928146,
|
| 9643 |
+
"learning_rate": 5.7679221820081456e-06,
|
| 9644 |
+
"loss": 10.33,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.5814716781365802,
|
| 9649 |
+
"grad_norm": 0.03440770506858826,
|
| 9650 |
+
"learning_rate": 5.699304247868875e-06,
|
| 9651 |
+
"loss": 10.3317,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.5818951826363155,
|
| 9656 |
+
"grad_norm": 0.03338101506233215,
|
| 9657 |
+
"learning_rate": 5.6310849417114734e-06,
|
| 9658 |
+
"loss": 10.3281,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.5823186871360508,
|
| 9663 |
+
"grad_norm": 0.04007452726364136,
|
| 9664 |
+
"learning_rate": 5.5632645519125946e-06,
|
| 9665 |
+
"loss": 10.3276,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.5827421916357861,
|
| 9670 |
+
"grad_norm": 0.03317529335618019,
|
| 9671 |
+
"learning_rate": 5.4958433651627e-06,
|
| 9672 |
+
"loss": 10.3267,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.5831656961355215,
|
| 9677 |
+
"grad_norm": 0.04026415944099426,
|
| 9678 |
+
"learning_rate": 5.428821666464656e-06,
|
| 9679 |
+
"loss": 10.3285,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.5835892006352568,
|
| 9684 |
+
"grad_norm": 0.03652326017618179,
|
| 9685 |
+
"learning_rate": 5.3621997391326565e-06,
|
| 9686 |
+
"loss": 10.3305,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.584012705134992,
|
| 9691 |
+
"grad_norm": 0.03854870796203613,
|
| 9692 |
+
"learning_rate": 5.295977864790969e-06,
|
| 9693 |
+
"loss": 10.3284,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.5844362096347274,
|
| 9698 |
+
"grad_norm": 0.03762730956077576,
|
| 9699 |
+
"learning_rate": 5.230156323372759e-06,
|
| 9700 |
+
"loss": 10.3254,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.5848597141344627,
|
| 9705 |
+
"grad_norm": 0.0510566271841526,
|
| 9706 |
+
"learning_rate": 5.164735393118914e-06,
|
| 9707 |
+
"loss": 10.3204,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.585283218634198,
|
| 9712 |
+
"grad_norm": 0.035746194422245026,
|
| 9713 |
+
"learning_rate": 5.099715350576817e-06,
|
| 9714 |
+
"loss": 10.3279,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.5857067231339333,
|
| 9719 |
+
"grad_norm": 0.03165269270539284,
|
| 9720 |
+
"learning_rate": 5.035096470599265e-06,
|
| 9721 |
+
"loss": 10.3288,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.5861302276336686,
|
| 9726 |
+
"grad_norm": 0.03464503213763237,
|
| 9727 |
+
"learning_rate": 4.970879026343256e-06,
|
| 9728 |
+
"loss": 10.3276,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.5865537321334039,
|
| 9733 |
+
"grad_norm": 0.03952055796980858,
|
| 9734 |
+
"learning_rate": 4.907063289268798e-06,
|
| 9735 |
+
"loss": 10.3311,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.5869772366331393,
|
| 9740 |
+
"grad_norm": 0.0374918058514595,
|
| 9741 |
+
"learning_rate": 4.8436495291378615e-06,
|
| 9742 |
+
"loss": 10.3268,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.5874007411328745,
|
| 9747 |
+
"grad_norm": 0.03804197534918785,
|
| 9748 |
+
"learning_rate": 4.780638014013106e-06,
|
| 9749 |
+
"loss": 10.3239,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.5878242456326098,
|
| 9754 |
+
"grad_norm": 0.03120071440935135,
|
| 9755 |
+
"learning_rate": 4.718029010256897e-06,
|
| 9756 |
+
"loss": 10.3236,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.5882477501323452,
|
| 9761 |
+
"grad_norm": 0.03194151446223259,
|
| 9762 |
+
"learning_rate": 4.655822782530051e-06,
|
| 9763 |
+
"loss": 10.3297,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.5886712546320805,
|
| 9768 |
+
"grad_norm": 0.032020777463912964,
|
| 9769 |
+
"learning_rate": 4.594019593790799e-06,
|
| 9770 |
+
"loss": 10.3258,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.5890947591318157,
|
| 9775 |
+
"grad_norm": 0.04302985966205597,
|
| 9776 |
+
"learning_rate": 4.5326197052936285e-06,
|
| 9777 |
+
"loss": 10.3243,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.5895182636315511,
|
| 9782 |
+
"grad_norm": 0.031170273199677467,
|
| 9783 |
+
"learning_rate": 4.471623376588196e-06,
|
| 9784 |
+
"loss": 10.3258,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.5899417681312864,
|
| 9789 |
+
"grad_norm": 0.035049792379140854,
|
| 9790 |
+
"learning_rate": 4.4110308655182396e-06,
|
| 9791 |
+
"loss": 10.3266,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.5903652726310217,
|
| 9796 |
+
"grad_norm": 0.03425823897123337,
|
| 9797 |
+
"learning_rate": 4.350842428220469e-06,
|
| 9798 |
+
"loss": 10.3244,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.5907887771307571,
|
| 9803 |
+
"grad_norm": 0.03348357602953911,
|
| 9804 |
+
"learning_rate": 4.29105831912352e-06,
|
| 9805 |
+
"loss": 10.3325,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.5912122816304923,
|
| 9810 |
+
"grad_norm": 0.03666883707046509,
|
| 9811 |
+
"learning_rate": 4.231678790946791e-06,
|
| 9812 |
+
"loss": 10.3311,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.5916357861302276,
|
| 9817 |
+
"grad_norm": 0.03748328611254692,
|
| 9818 |
+
"learning_rate": 4.172704094699509e-06,
|
| 9819 |
+
"loss": 10.325,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.592059290629963,
|
| 9824 |
+
"grad_norm": 0.036303918808698654,
|
| 9825 |
+
"learning_rate": 4.1141344796795436e-06,
|
| 9826 |
+
"loss": 10.3309,
|
| 9827 |
+
"step": 1398
|
| 9828 |
+
},
|
| 9829 |
+
{
|
| 9830 |
+
"epoch": 0.5924827951296983,
|
| 9831 |
+
"grad_norm": 0.03511890396475792,
|
| 9832 |
+
"learning_rate": 4.0559701934724135e-06,
|
| 9833 |
+
"loss": 10.3282,
|
| 9834 |
+
"step": 1399
|
| 9835 |
+
},
|
| 9836 |
+
{
|
| 9837 |
+
"epoch": 0.5929062996294335,
|
| 9838 |
+
"grad_norm": 0.04446948692202568,
|
| 9839 |
+
"learning_rate": 3.998211481950254e-06,
|
| 9840 |
+
"loss": 10.3252,
|
| 9841 |
+
"step": 1400
|
| 9842 |
+
},
|
| 9843 |
+
{
|
| 9844 |
+
"epoch": 0.5933298041291689,
|
| 9845 |
+
"grad_norm": 0.03524168208241463,
|
| 9846 |
+
"learning_rate": 3.940858589270735e-06,
|
| 9847 |
+
"loss": 10.3276,
|
| 9848 |
+
"step": 1401
|
| 9849 |
+
},
|
| 9850 |
+
{
|
| 9851 |
+
"epoch": 0.5937533086289042,
|
| 9852 |
+
"grad_norm": 0.040072131901979446,
|
| 9853 |
+
"learning_rate": 3.883911757876058e-06,
|
| 9854 |
+
"loss": 10.3264,
|
| 9855 |
+
"step": 1402
|
| 9856 |
+
},
|
| 9857 |
+
{
|
| 9858 |
+
"epoch": 0.5941768131286395,
|
| 9859 |
+
"grad_norm": 0.04590757191181183,
|
| 9860 |
+
"learning_rate": 3.827371228491916e-06,
|
| 9861 |
+
"loss": 10.3287,
|
| 9862 |
+
"step": 1403
|
| 9863 |
+
},
|
| 9864 |
+
{
|
| 9865 |
+
"epoch": 0.5946003176283748,
|
| 9866 |
+
"grad_norm": 0.0415809340775013,
|
| 9867 |
+
"learning_rate": 3.7712372401264684e-06,
|
| 9868 |
+
"loss": 10.328,
|
| 9869 |
+
"step": 1404
|
| 9870 |
+
},
|
| 9871 |
+
{
|
| 9872 |
+
"epoch": 0.5950238221281101,
|
| 9873 |
+
"grad_norm": 0.026884714141488075,
|
| 9874 |
+
"learning_rate": 3.715510030069391e-06,
|
| 9875 |
+
"loss": 10.327,
|
| 9876 |
+
"step": 1405
|
| 9877 |
+
},
|
| 9878 |
+
{
|
| 9879 |
+
"epoch": 0.5954473266278454,
|
| 9880 |
+
"grad_norm": 0.036991361528635025,
|
| 9881 |
+
"learning_rate": 3.6601898338907703e-06,
|
| 9882 |
+
"loss": 10.3271,
|
| 9883 |
+
"step": 1406
|
| 9884 |
+
},
|
| 9885 |
+
{
|
| 9886 |
+
"epoch": 0.5958708311275808,
|
| 9887 |
+
"grad_norm": 0.03274872526526451,
|
| 9888 |
+
"learning_rate": 3.60527688544019e-06,
|
| 9889 |
+
"loss": 10.3297,
|
| 9890 |
+
"step": 1407
|
| 9891 |
+
},
|
| 9892 |
+
{
|
| 9893 |
+
"epoch": 0.596294335627316,
|
| 9894 |
+
"grad_norm": 0.030266571789979935,
|
| 9895 |
+
"learning_rate": 3.5507714168456997e-06,
|
| 9896 |
+
"loss": 10.3318,
|
| 9897 |
+
"step": 1408
|
| 9898 |
+
},
|
| 9899 |
+
{
|
| 9900 |
+
"epoch": 0.5967178401270513,
|
| 9901 |
+
"grad_norm": 0.03449970856308937,
|
| 9902 |
+
"learning_rate": 3.4966736585128834e-06,
|
| 9903 |
+
"loss": 10.3276,
|
| 9904 |
+
"step": 1409
|
| 9905 |
+
},
|
| 9906 |
+
{
|
| 9907 |
+
"epoch": 0.5971413446267867,
|
| 9908 |
+
"grad_norm": 0.03458021953701973,
|
| 9909 |
+
"learning_rate": 3.4429838391238255e-06,
|
| 9910 |
+
"loss": 10.3259,
|
| 9911 |
+
"step": 1410
|
| 9912 |
+
},
|
| 9913 |
+
{
|
| 9914 |
+
"epoch": 0.597564849126522,
|
| 9915 |
+
"grad_norm": 0.035373665392398834,
|
| 9916 |
+
"learning_rate": 3.3897021856361786e-06,
|
| 9917 |
+
"loss": 10.3287,
|
| 9918 |
+
"step": 1411
|
| 9919 |
+
},
|
| 9920 |
+
{
|
| 9921 |
+
"epoch": 0.5979883536262572,
|
| 9922 |
+
"grad_norm": 0.030641932040452957,
|
| 9923 |
+
"learning_rate": 3.3368289232822093e-06,
|
| 9924 |
+
"loss": 10.3236,
|
| 9925 |
+
"step": 1412
|
| 9926 |
+
},
|
| 9927 |
+
{
|
| 9928 |
+
"epoch": 0.5984118581259926,
|
| 9929 |
+
"grad_norm": 0.03205382078886032,
|
| 9930 |
+
"learning_rate": 3.284364275567775e-06,
|
| 9931 |
+
"loss": 10.3238,
|
| 9932 |
+
"step": 1413
|
| 9933 |
+
},
|
| 9934 |
+
{
|
| 9935 |
+
"epoch": 0.5988353626257279,
|
| 9936 |
+
"grad_norm": 0.036099858582019806,
|
| 9937 |
+
"learning_rate": 3.232308464271505e-06,
|
| 9938 |
+
"loss": 10.3276,
|
| 9939 |
+
"step": 1414
|
| 9940 |
+
},
|
| 9941 |
+
{
|
| 9942 |
+
"epoch": 0.5992588671254632,
|
| 9943 |
+
"grad_norm": 0.03898213803768158,
|
| 9944 |
+
"learning_rate": 3.1806617094437773e-06,
|
| 9945 |
+
"loss": 10.3286,
|
| 9946 |
+
"step": 1415
|
| 9947 |
+
},
|
| 9948 |
+
{
|
| 9949 |
+
"epoch": 0.5996823716251986,
|
| 9950 |
+
"grad_norm": 0.042888887226581573,
|
| 9951 |
+
"learning_rate": 3.1294242294057976e-06,
|
| 9952 |
+
"loss": 10.3261,
|
| 9953 |
+
"step": 1416
|
| 9954 |
+
},
|
| 9955 |
+
{
|
| 9956 |
+
"epoch": 0.6001058761249338,
|
| 9957 |
+
"grad_norm": 0.04815137758851051,
|
| 9958 |
+
"learning_rate": 3.0785962407486767e-06,
|
| 9959 |
+
"loss": 10.3275,
|
| 9960 |
+
"step": 1417
|
| 9961 |
+
},
|
| 9962 |
+
{
|
| 9963 |
+
"epoch": 0.6005293806246691,
|
| 9964 |
+
"grad_norm": 0.0393349751830101,
|
| 9965 |
+
"learning_rate": 3.028177958332512e-06,
|
| 9966 |
+
"loss": 10.329,
|
| 9967 |
+
"step": 1418
|
| 9968 |
+
},
|
| 9969 |
+
{
|
| 9970 |
+
"epoch": 0.6009528851244045,
|
| 9971 |
+
"grad_norm": 0.03242544084787369,
|
| 9972 |
+
"learning_rate": 2.9781695952855404e-06,
|
| 9973 |
+
"loss": 10.3332,
|
| 9974 |
+
"step": 1419
|
| 9975 |
+
},
|
| 9976 |
+
{
|
| 9977 |
+
"epoch": 0.6013763896241398,
|
| 9978 |
+
"grad_norm": 0.04108821600675583,
|
| 9979 |
+
"learning_rate": 2.928571363003152e-06,
|
| 9980 |
+
"loss": 10.3272,
|
| 9981 |
+
"step": 1420
|
| 9982 |
+
},
|
| 9983 |
+
{
|
| 9984 |
+
"epoch": 0.601799894123875,
|
| 9985 |
+
"grad_norm": 0.05158897116780281,
|
| 9986 |
+
"learning_rate": 2.8793834711470235e-06,
|
| 9987 |
+
"loss": 10.3246,
|
| 9988 |
+
"step": 1421
|
| 9989 |
+
},
|
| 9990 |
+
{
|
| 9991 |
+
"epoch": 0.6022233986236104,
|
| 9992 |
+
"grad_norm": 0.0375213548541069,
|
| 9993 |
+
"learning_rate": 2.8306061276442752e-06,
|
| 9994 |
+
"loss": 10.3241,
|
| 9995 |
+
"step": 1422
|
| 9996 |
+
},
|
| 9997 |
+
{
|
| 9998 |
+
"epoch": 0.6026469031233457,
|
| 9999 |
+
"grad_norm": 0.04019343852996826,
|
| 10000 |
+
"learning_rate": 2.782239538686504e-06,
|
| 10001 |
+
"loss": 10.3258,
|
| 10002 |
+
"step": 1423
|
| 10003 |
+
},
|
| 10004 |
+
{
|
| 10005 |
+
"epoch": 0.603070407623081,
|
| 10006 |
+
"grad_norm": 0.028071366250514984,
|
| 10007 |
+
"learning_rate": 2.7342839087290183e-06,
|
| 10008 |
+
"loss": 10.3286,
|
| 10009 |
+
"step": 1424
|
| 10010 |
+
},
|
| 10011 |
+
{
|
| 10012 |
+
"epoch": 0.6034939121228163,
|
| 10013 |
+
"grad_norm": 0.03758038207888603,
|
| 10014 |
+
"learning_rate": 2.6867394404898938e-06,
|
| 10015 |
+
"loss": 10.324,
|
| 10016 |
+
"step": 1425
|
| 10017 |
+
},
|
| 10018 |
+
{
|
| 10019 |
+
"epoch": 0.6039174166225516,
|
| 10020 |
+
"grad_norm": 0.03263649344444275,
|
| 10021 |
+
"learning_rate": 2.6396063349491627e-06,
|
| 10022 |
+
"loss": 10.3266,
|
| 10023 |
+
"step": 1426
|
| 10024 |
+
},
|
| 10025 |
+
{
|
| 10026 |
+
"epoch": 0.6043409211222869,
|
| 10027 |
+
"grad_norm": 0.04192793369293213,
|
| 10028 |
+
"learning_rate": 2.592884791347905e-06,
|
| 10029 |
+
"loss": 10.3265,
|
| 10030 |
+
"step": 1427
|
| 10031 |
+
},
|
| 10032 |
+
{
|
| 10033 |
+
"epoch": 0.6047644256220223,
|
| 10034 |
+
"grad_norm": 0.03950565308332443,
|
| 10035 |
+
"learning_rate": 2.5465750071874793e-06,
|
| 10036 |
+
"loss": 10.3257,
|
| 10037 |
+
"step": 1428
|
| 10038 |
+
},
|
| 10039 |
+
{
|
| 10040 |
+
"epoch": 0.6051879301217575,
|
| 10041 |
+
"grad_norm": 0.03863437846302986,
|
| 10042 |
+
"learning_rate": 2.500677178228639e-06,
|
| 10043 |
+
"loss": 10.3259,
|
| 10044 |
+
"step": 1429
|
| 10045 |
+
},
|
| 10046 |
+
{
|
| 10047 |
+
"epoch": 0.6056114346214928,
|
| 10048 |
+
"grad_norm": 0.030046673491597176,
|
| 10049 |
+
"learning_rate": 2.455191498490739e-06,
|
| 10050 |
+
"loss": 10.3291,
|
| 10051 |
+
"step": 1430
|
| 10052 |
+
},
|
| 10053 |
+
{
|
| 10054 |
+
"epoch": 0.6060349391212282,
|
| 10055 |
+
"grad_norm": 0.03240976110100746,
|
| 10056 |
+
"learning_rate": 2.4101181602508626e-06,
|
| 10057 |
+
"loss": 10.3295,
|
| 10058 |
+
"step": 1431
|
| 10059 |
+
},
|
| 10060 |
+
{
|
| 10061 |
+
"epoch": 0.6064584436209635,
|
| 10062 |
+
"grad_norm": 0.036038123071193695,
|
| 10063 |
+
"learning_rate": 2.365457354043088e-06,
|
| 10064 |
+
"loss": 10.3299,
|
| 10065 |
+
"step": 1432
|
| 10066 |
+
},
|
| 10067 |
+
{
|
| 10068 |
+
"epoch": 0.6068819481206987,
|
| 10069 |
+
"grad_norm": 0.03231705725193024,
|
| 10070 |
+
"learning_rate": 2.3212092686576093e-06,
|
| 10071 |
+
"loss": 10.3237,
|
| 10072 |
+
"step": 1433
|
| 10073 |
+
},
|
| 10074 |
+
{
|
| 10075 |
+
"epoch": 0.6073054526204341,
|
| 10076 |
+
"grad_norm": 0.03693152591586113,
|
| 10077 |
+
"learning_rate": 2.27737409113995e-06,
|
| 10078 |
+
"loss": 10.3234,
|
| 10079 |
+
"step": 1434
|
| 10080 |
+
},
|
| 10081 |
+
{
|
| 10082 |
+
"epoch": 0.6077289571201694,
|
| 10083 |
+
"grad_norm": 0.02976830117404461,
|
| 10084 |
+
"learning_rate": 2.2339520067902185e-06,
|
| 10085 |
+
"loss": 10.3287,
|
| 10086 |
+
"step": 1435
|
| 10087 |
+
},
|
| 10088 |
+
{
|
| 10089 |
+
"epoch": 0.6081524616199047,
|
| 10090 |
+
"grad_norm": 0.034811802208423615,
|
| 10091 |
+
"learning_rate": 2.19094319916231e-06,
|
| 10092 |
+
"loss": 10.3291,
|
| 10093 |
+
"step": 1436
|
| 10094 |
+
},
|
| 10095 |
+
{
|
| 10096 |
+
"epoch": 0.60857596611964,
|
| 10097 |
+
"grad_norm": 0.03957490250468254,
|
| 10098 |
+
"learning_rate": 2.1483478500630705e-06,
|
| 10099 |
+
"loss": 10.3293,
|
| 10100 |
+
"step": 1437
|
| 10101 |
+
},
|
| 10102 |
+
{
|
| 10103 |
+
"epoch": 0.6089994706193753,
|
| 10104 |
+
"grad_norm": 0.046876002103090286,
|
| 10105 |
+
"learning_rate": 2.1061661395516018e-06,
|
| 10106 |
+
"loss": 10.3283,
|
| 10107 |
+
"step": 1438
|
| 10108 |
+
},
|
| 10109 |
+
{
|
| 10110 |
+
"epoch": 0.6094229751191106,
|
| 10111 |
+
"grad_norm": 0.03128580003976822,
|
| 10112 |
+
"learning_rate": 2.06439824593847e-06,
|
| 10113 |
+
"loss": 10.3279,
|
| 10114 |
+
"step": 1439
|
| 10115 |
+
},
|
| 10116 |
+
{
|
| 10117 |
+
"epoch": 0.609846479618846,
|
| 10118 |
+
"grad_norm": 0.04396365210413933,
|
| 10119 |
+
"learning_rate": 2.0230443457849414e-06,
|
| 10120 |
+
"loss": 10.3248,
|
| 10121 |
+
"step": 1440
|
| 10122 |
+
},
|
| 10123 |
+
{
|
| 10124 |
+
"epoch": 0.6102699841185812,
|
| 10125 |
+
"grad_norm": 0.0318458117544651,
|
| 10126 |
+
"learning_rate": 1.982104613902258e-06,
|
| 10127 |
+
"loss": 10.3241,
|
| 10128 |
+
"step": 1441
|
| 10129 |
+
},
|
| 10130 |
+
{
|
| 10131 |
+
"epoch": 0.6106934886183165,
|
| 10132 |
+
"grad_norm": 0.03930509835481644,
|
| 10133 |
+
"learning_rate": 1.9415792233508977e-06,
|
| 10134 |
+
"loss": 10.3288,
|
| 10135 |
+
"step": 1442
|
| 10136 |
+
},
|
| 10137 |
+
{
|
| 10138 |
+
"epoch": 0.6111169931180519,
|
| 10139 |
+
"grad_norm": 0.04166901111602783,
|
| 10140 |
+
"learning_rate": 1.9014683454398386e-06,
|
| 10141 |
+
"loss": 10.3262,
|
| 10142 |
+
"step": 1443
|
| 10143 |
+
},
|
| 10144 |
+
{
|
| 10145 |
+
"epoch": 0.6115404976177872,
|
| 10146 |
+
"grad_norm": 0.0358477458357811,
|
| 10147 |
+
"learning_rate": 1.8617721497257823e-06,
|
| 10148 |
+
"loss": 10.3265,
|
| 10149 |
+
"step": 1444
|
| 10150 |
+
},
|
| 10151 |
+
{
|
| 10152 |
+
"epoch": 0.6119640021175226,
|
| 10153 |
+
"grad_norm": 0.06854399293661118,
|
| 10154 |
+
"learning_rate": 1.8224908040125554e-06,
|
| 10155 |
+
"loss": 10.3329,
|
| 10156 |
+
"step": 1445
|
| 10157 |
+
},
|
| 10158 |
+
{
|
| 10159 |
+
"epoch": 0.6123875066172578,
|
| 10160 |
+
"grad_norm": 0.03350813686847687,
|
| 10161 |
+
"learning_rate": 1.783624474350276e-06,
|
| 10162 |
+
"loss": 10.3262,
|
| 10163 |
+
"step": 1446
|
| 10164 |
+
},
|
| 10165 |
+
{
|
| 10166 |
+
"epoch": 0.6128110111169931,
|
| 10167 |
+
"grad_norm": 0.04633032903075218,
|
| 10168 |
+
"learning_rate": 1.7451733250347323e-06,
|
| 10169 |
+
"loss": 10.3289,
|
| 10170 |
+
"step": 1447
|
| 10171 |
+
},
|
| 10172 |
+
{
|
| 10173 |
+
"epoch": 0.6132345156167285,
|
| 10174 |
+
"grad_norm": 0.043577518314123154,
|
| 10175 |
+
"learning_rate": 1.7071375186066607e-06,
|
| 10176 |
+
"loss": 10.3283,
|
| 10177 |
+
"step": 1448
|
| 10178 |
+
},
|
| 10179 |
+
{
|
| 10180 |
+
"epoch": 0.6136580201164638,
|
| 10181 |
+
"grad_norm": 0.04235590621829033,
|
| 10182 |
+
"learning_rate": 1.6695172158510576e-06,
|
| 10183 |
+
"loss": 10.3256,
|
| 10184 |
+
"step": 1449
|
| 10185 |
+
},
|
| 10186 |
+
{
|
| 10187 |
+
"epoch": 0.614081524616199,
|
| 10188 |
+
"grad_norm": 0.033999886363744736,
|
| 10189 |
+
"learning_rate": 1.6323125757964797e-06,
|
| 10190 |
+
"loss": 10.3266,
|
| 10191 |
+
"step": 1450
|
| 10192 |
+
},
|
| 10193 |
+
{
|
| 10194 |
+
"epoch": 0.6145050291159344,
|
| 10195 |
+
"grad_norm": 0.05530522018671036,
|
| 10196 |
+
"learning_rate": 1.5955237557144343e-06,
|
| 10197 |
+
"loss": 10.3309,
|
| 10198 |
+
"step": 1451
|
| 10199 |
+
},
|
| 10200 |
+
{
|
| 10201 |
+
"epoch": 0.6149285336156697,
|
| 10202 |
+
"grad_norm": 0.028483441099524498,
|
| 10203 |
+
"learning_rate": 1.5591509111186341e-06,
|
| 10204 |
+
"loss": 10.328,
|
| 10205 |
+
"step": 1452
|
| 10206 |
+
},
|
| 10207 |
+
{
|
| 10208 |
+
"epoch": 0.615352038115405,
|
| 10209 |
+
"grad_norm": 0.03823675215244293,
|
| 10210 |
+
"learning_rate": 1.5231941957644213e-06,
|
| 10211 |
+
"loss": 10.3247,
|
| 10212 |
+
"step": 1453
|
| 10213 |
+
},
|
| 10214 |
+
{
|
| 10215 |
+
"epoch": 0.6157755426151403,
|
| 10216 |
+
"grad_norm": 0.05358748883008957,
|
| 10217 |
+
"learning_rate": 1.4876537616480334e-06,
|
| 10218 |
+
"loss": 10.3325,
|
| 10219 |
+
"step": 1454
|
| 10220 |
+
},
|
| 10221 |
+
{
|
| 10222 |
+
"epoch": 0.6161990471148756,
|
| 10223 |
+
"grad_norm": 0.09875933080911636,
|
| 10224 |
+
"learning_rate": 1.4525297590060272e-06,
|
| 10225 |
+
"loss": 10.3403,
|
| 10226 |
+
"step": 1455
|
| 10227 |
+
},
|
| 10228 |
+
{
|
| 10229 |
+
"epoch": 0.6166225516146109,
|
| 10230 |
+
"grad_norm": 0.03407939523458481,
|
| 10231 |
+
"learning_rate": 1.4178223363146225e-06,
|
| 10232 |
+
"loss": 10.3211,
|
| 10233 |
+
"step": 1456
|
| 10234 |
+
},
|
| 10235 |
+
{
|
| 10236 |
+
"epoch": 0.6170460561143463,
|
| 10237 |
+
"grad_norm": 0.035402923822402954,
|
| 10238 |
+
"learning_rate": 1.3835316402890707e-06,
|
| 10239 |
+
"loss": 10.3278,
|
| 10240 |
+
"step": 1457
|
| 10241 |
+
},
|
| 10242 |
+
{
|
| 10243 |
+
"epoch": 0.6174695606140815,
|
| 10244 |
+
"grad_norm": 0.04178466647863388,
|
| 10245 |
+
"learning_rate": 1.349657815883032e-06,
|
| 10246 |
+
"loss": 10.3271,
|
| 10247 |
+
"step": 1458
|
| 10248 |
+
},
|
| 10249 |
+
{
|
| 10250 |
+
"epoch": 0.6178930651138168,
|
| 10251 |
+
"grad_norm": 0.045721668750047684,
|
| 10252 |
+
"learning_rate": 1.3162010062879538e-06,
|
| 10253 |
+
"loss": 10.324,
|
| 10254 |
+
"step": 1459
|
| 10255 |
+
},
|
| 10256 |
+
{
|
| 10257 |
+
"epoch": 0.6183165696135522,
|
| 10258 |
+
"grad_norm": 0.03632737323641777,
|
| 10259 |
+
"learning_rate": 1.2831613529325048e-06,
|
| 10260 |
+
"loss": 10.3263,
|
| 10261 |
+
"step": 1460
|
| 10262 |
+
},
|
| 10263 |
+
{
|
| 10264 |
+
"epoch": 0.6187400741132875,
|
| 10265 |
+
"grad_norm": 0.0301133394241333,
|
| 10266 |
+
"learning_rate": 1.2505389954819203e-06,
|
| 10267 |
+
"loss": 10.3259,
|
| 10268 |
+
"step": 1461
|
| 10269 |
+
},
|
| 10270 |
+
{
|
| 10271 |
+
"epoch": 0.6191635786130227,
|
| 10272 |
+
"grad_norm": 0.0355866476893425,
|
| 10273 |
+
"learning_rate": 1.2183340718374681e-06,
|
| 10274 |
+
"loss": 10.3341,
|
| 10275 |
+
"step": 1462
|
| 10276 |
+
},
|
| 10277 |
+
{
|
| 10278 |
+
"epoch": 0.6195870831127581,
|
| 10279 |
+
"grad_norm": 0.036043938249349594,
|
| 10280 |
+
"learning_rate": 1.186546718135817e-06,
|
| 10281 |
+
"loss": 10.3318,
|
| 10282 |
+
"step": 1463
|
| 10283 |
+
},
|
| 10284 |
+
{
|
| 10285 |
+
"epoch": 0.6200105876124934,
|
| 10286 |
+
"grad_norm": 0.0343465656042099,
|
| 10287 |
+
"learning_rate": 1.1551770687485141e-06,
|
| 10288 |
+
"loss": 10.3306,
|
| 10289 |
+
"step": 1464
|
| 10290 |
+
},
|
| 10291 |
+
{
|
| 10292 |
+
"epoch": 0.6204340921122287,
|
| 10293 |
+
"grad_norm": 0.04068030044436455,
|
| 10294 |
+
"learning_rate": 1.1242252562813527e-06,
|
| 10295 |
+
"loss": 10.3256,
|
| 10296 |
+
"step": 1465
|
| 10297 |
+
},
|
| 10298 |
+
{
|
| 10299 |
+
"epoch": 0.620857596611964,
|
| 10300 |
+
"grad_norm": 0.035130009055137634,
|
| 10301 |
+
"learning_rate": 1.0936914115738717e-06,
|
| 10302 |
+
"loss": 10.3253,
|
| 10303 |
+
"step": 1466
|
| 10304 |
+
},
|
| 10305 |
+
{
|
| 10306 |
+
"epoch": 0.6212811011116993,
|
| 10307 |
+
"grad_norm": 0.035075314342975616,
|
| 10308 |
+
"learning_rate": 1.063575663698757e-06,
|
| 10309 |
+
"loss": 10.3274,
|
| 10310 |
+
"step": 1467
|
| 10311 |
+
},
|
| 10312 |
+
{
|
| 10313 |
+
"epoch": 0.6217046056114346,
|
| 10314 |
+
"grad_norm": 0.035737186670303345,
|
| 10315 |
+
"learning_rate": 1.0338781399613307e-06,
|
| 10316 |
+
"loss": 10.3272,
|
| 10317 |
+
"step": 1468
|
| 10318 |
+
},
|
| 10319 |
+
{
|
| 10320 |
+
"epoch": 0.62212811011117,
|
| 10321 |
+
"grad_norm": 0.033577874302864075,
|
| 10322 |
+
"learning_rate": 1.004598965898995e-06,
|
| 10323 |
+
"loss": 10.3275,
|
| 10324 |
+
"step": 1469
|
| 10325 |
+
},
|
| 10326 |
+
{
|
| 10327 |
+
"epoch": 0.6225516146109052,
|
| 10328 |
+
"grad_norm": 0.05437571182847023,
|
| 10329 |
+
"learning_rate": 9.75738265280679e-07,
|
| 10330 |
+
"loss": 10.327,
|
| 10331 |
+
"step": 1470
|
| 10332 |
+
},
|
| 10333 |
+
{
|
| 10334 |
+
"epoch": 0.6229751191106405,
|
| 10335 |
+
"grad_norm": 0.03711257502436638,
|
| 10336 |
+
"learning_rate": 9.472961601063813e-07,
|
| 10337 |
+
"loss": 10.3246,
|
| 10338 |
+
"step": 1471
|
| 10339 |
+
},
|
| 10340 |
+
{
|
| 10341 |
+
"epoch": 0.6233986236103759,
|
| 10342 |
+
"grad_norm": 0.044933151453733444,
|
| 10343 |
+
"learning_rate": 9.192727706065829e-07,
|
| 10344 |
+
"loss": 10.3289,
|
| 10345 |
+
"step": 1472
|
| 10346 |
+
},
|
| 10347 |
+
{
|
| 10348 |
+
"epoch": 0.6238221281101112,
|
| 10349 |
+
"grad_norm": 0.03690815716981888,
|
| 10350 |
+
"learning_rate": 8.91668215241781e-07,
|
| 10351 |
+
"loss": 10.3351,
|
| 10352 |
+
"step": 1473
|
| 10353 |
+
},
|
| 10354 |
+
{
|
| 10355 |
+
"epoch": 0.6242456326098464,
|
| 10356 |
+
"grad_norm": 0.04347551241517067,
|
| 10357 |
+
"learning_rate": 8.644826107019888e-07,
|
| 10358 |
+
"loss": 10.3303,
|
| 10359 |
+
"step": 1474
|
| 10360 |
+
},
|
| 10361 |
+
{
|
| 10362 |
+
"epoch": 0.6246691371095818,
|
| 10363 |
+
"grad_norm": 0.0514223575592041,
|
| 10364 |
+
"learning_rate": 8.377160719062028e-07,
|
| 10365 |
+
"loss": 10.3238,
|
| 10366 |
+
"step": 1475
|
| 10367 |
+
},
|
| 10368 |
+
{
|
| 10369 |
+
"epoch": 0.6250926416093171,
|
| 10370 |
+
"grad_norm": 0.03661144897341728,
|
| 10371 |
+
"learning_rate": 8.113687120019586e-07,
|
| 10372 |
+
"loss": 10.3241,
|
| 10373 |
+
"step": 1476
|
| 10374 |
+
},
|
| 10375 |
+
{
|
| 10376 |
+
"epoch": 0.6255161461090524,
|
| 10377 |
+
"grad_norm": 0.03711770474910736,
|
| 10378 |
+
"learning_rate": 7.85440642364843e-07,
|
| 10379 |
+
"loss": 10.3275,
|
| 10380 |
+
"step": 1477
|
| 10381 |
+
},
|
| 10382 |
+
{
|
| 10383 |
+
"epoch": 0.6259396506087878,
|
| 10384 |
+
"grad_norm": 0.04083530232310295,
|
| 10385 |
+
"learning_rate": 7.599319725980048e-07,
|
| 10386 |
+
"loss": 10.3207,
|
| 10387 |
+
"step": 1478
|
| 10388 |
+
},
|
| 10389 |
+
{
|
| 10390 |
+
"epoch": 0.626363155108523,
|
| 10391 |
+
"grad_norm": 0.03891532123088837,
|
| 10392 |
+
"learning_rate": 7.348428105317329e-07,
|
| 10393 |
+
"loss": 10.328,
|
| 10394 |
+
"step": 1479
|
| 10395 |
+
},
|
| 10396 |
+
{
|
| 10397 |
+
"epoch": 0.6267866596082583,
|
| 10398 |
+
"grad_norm": 0.03179390728473663,
|
| 10399 |
+
"learning_rate": 7.101732622229463e-07,
|
| 10400 |
+
"loss": 10.3237,
|
| 10401 |
+
"step": 1480
|
| 10402 |
+
},
|
| 10403 |
+
{
|
| 10404 |
+
"epoch": 0.6272101641079937,
|
| 10405 |
+
"grad_norm": 0.034473542124032974,
|
| 10406 |
+
"learning_rate": 6.859234319547825e-07,
|
| 10407 |
+
"loss": 10.327,
|
| 10408 |
+
"step": 1481
|
| 10409 |
+
},
|
| 10410 |
+
{
|
| 10411 |
+
"epoch": 0.627633668607729,
|
| 10412 |
+
"grad_norm": 0.0452442467212677,
|
| 10413 |
+
"learning_rate": 6.620934222361319e-07,
|
| 10414 |
+
"loss": 10.3252,
|
| 10415 |
+
"step": 1482
|
| 10416 |
+
},
|
| 10417 |
+
{
|
| 10418 |
+
"epoch": 0.6280571731074642,
|
| 10419 |
+
"grad_norm": 0.036436449736356735,
|
| 10420 |
+
"learning_rate": 6.386833338012488e-07,
|
| 10421 |
+
"loss": 10.3265,
|
| 10422 |
+
"step": 1483
|
| 10423 |
+
},
|
| 10424 |
+
{
|
| 10425 |
+
"epoch": 0.6284806776071996,
|
| 10426 |
+
"grad_norm": 0.034642040729522705,
|
| 10427 |
+
"learning_rate": 6.15693265609274e-07,
|
| 10428 |
+
"loss": 10.3268,
|
| 10429 |
+
"step": 1484
|
| 10430 |
+
},
|
| 10431 |
+
{
|
| 10432 |
+
"epoch": 0.6289041821069349,
|
| 10433 |
+
"grad_norm": 0.047027528285980225,
|
| 10434 |
+
"learning_rate": 5.931233148438354e-07,
|
| 10435 |
+
"loss": 10.3272,
|
| 10436 |
+
"step": 1485
|
| 10437 |
+
},
|
| 10438 |
+
{
|
| 10439 |
+
"epoch": 0.6293276866066702,
|
| 10440 |
+
"grad_norm": 0.04386279359459877,
|
| 10441 |
+
"learning_rate": 5.709735769126478e-07,
|
| 10442 |
+
"loss": 10.3263,
|
| 10443 |
+
"step": 1486
|
| 10444 |
+
},
|
| 10445 |
+
{
|
| 10446 |
+
"epoch": 0.6297511911064055,
|
| 10447 |
+
"grad_norm": 0.041581086814403534,
|
| 10448 |
+
"learning_rate": 5.492441454471142e-07,
|
| 10449 |
+
"loss": 10.3251,
|
| 10450 |
+
"step": 1487
|
| 10451 |
+
},
|
| 10452 |
+
{
|
| 10453 |
+
"epoch": 0.6301746956061408,
|
| 10454 |
+
"grad_norm": 0.040245793759822845,
|
| 10455 |
+
"learning_rate": 5.279351123019027e-07,
|
| 10456 |
+
"loss": 10.326,
|
| 10457 |
+
"step": 1488
|
| 10458 |
+
},
|
| 10459 |
+
{
|
| 10460 |
+
"epoch": 0.6305982001058761,
|
| 10461 |
+
"grad_norm": 0.04215911403298378,
|
| 10462 |
+
"learning_rate": 5.07046567554581e-07,
|
| 10463 |
+
"loss": 10.3307,
|
| 10464 |
+
"step": 1489
|
| 10465 |
+
},
|
| 10466 |
+
{
|
| 10467 |
+
"epoch": 0.6310217046056115,
|
| 10468 |
+
"grad_norm": 0.03327825665473938,
|
| 10469 |
+
"learning_rate": 4.865785995052052e-07,
|
| 10470 |
+
"loss": 10.3239,
|
| 10471 |
+
"step": 1490
|
| 10472 |
+
},
|
| 10473 |
+
{
|
| 10474 |
+
"epoch": 0.6314452091053467,
|
| 10475 |
+
"grad_norm": 0.03097568452358246,
|
| 10476 |
+
"learning_rate": 4.665312946760092e-07,
|
| 10477 |
+
"loss": 10.326,
|
| 10478 |
+
"step": 1491
|
| 10479 |
+
},
|
| 10480 |
+
{
|
| 10481 |
+
"epoch": 0.631868713605082,
|
| 10482 |
+
"grad_norm": 0.03316866606473923,
|
| 10483 |
+
"learning_rate": 4.469047378109603e-07,
|
| 10484 |
+
"loss": 10.3247,
|
| 10485 |
+
"step": 1492
|
| 10486 |
+
},
|
| 10487 |
+
{
|
| 10488 |
+
"epoch": 0.6322922181048174,
|
| 10489 |
+
"grad_norm": 0.03606574982404709,
|
| 10490 |
+
"learning_rate": 4.276990118754598e-07,
|
| 10491 |
+
"loss": 10.3255,
|
| 10492 |
+
"step": 1493
|
| 10493 |
+
},
|
| 10494 |
+
{
|
| 10495 |
+
"epoch": 0.6327157226045527,
|
| 10496 |
+
"grad_norm": 0.031570881605148315,
|
| 10497 |
+
"learning_rate": 4.089141980559763e-07,
|
| 10498 |
+
"loss": 10.328,
|
| 10499 |
+
"step": 1494
|
| 10500 |
+
},
|
| 10501 |
+
{
|
| 10502 |
+
"epoch": 0.6331392271042879,
|
| 10503 |
+
"grad_norm": 0.03704217076301575,
|
| 10504 |
+
"learning_rate": 3.9055037575969067e-07,
|
| 10505 |
+
"loss": 10.3322,
|
| 10506 |
+
"step": 1495
|
| 10507 |
+
},
|
| 10508 |
+
{
|
| 10509 |
+
"epoch": 0.6335627316040233,
|
| 10510 |
+
"grad_norm": 0.03473872318863869,
|
| 10511 |
+
"learning_rate": 3.7260762261416284e-07,
|
| 10512 |
+
"loss": 10.3277,
|
| 10513 |
+
"step": 1496
|
| 10514 |
+
},
|
| 10515 |
+
{
|
| 10516 |
+
"epoch": 0.6339862361037586,
|
| 10517 |
+
"grad_norm": 0.0472039133310318,
|
| 10518 |
+
"learning_rate": 3.5508601446700985e-07,
|
| 10519 |
+
"loss": 10.328,
|
| 10520 |
+
"step": 1497
|
| 10521 |
+
},
|
| 10522 |
+
{
|
| 10523 |
+
"epoch": 0.6344097406034939,
|
| 10524 |
+
"grad_norm": 0.03687547892332077,
|
| 10525 |
+
"learning_rate": 3.379856253855951e-07,
|
| 10526 |
+
"loss": 10.324,
|
| 10527 |
+
"step": 1498
|
| 10528 |
+
},
|
| 10529 |
+
{
|
| 10530 |
+
"epoch": 0.6348332451032292,
|
| 10531 |
+
"grad_norm": 0.031235698610544205,
|
| 10532 |
+
"learning_rate": 3.213065276566951e-07,
|
| 10533 |
+
"loss": 10.324,
|
| 10534 |
+
"step": 1499
|
| 10535 |
+
},
|
| 10536 |
+
{
|
| 10537 |
+
"epoch": 0.6352567496029645,
|
| 10538 |
+
"grad_norm": 0.03706636279821396,
|
| 10539 |
+
"learning_rate": 3.050487917862221e-07,
|
| 10540 |
+
"loss": 10.3237,
|
| 10541 |
+
"step": 1500
|
| 10542 |
+
},
|
| 10543 |
+
{
|
| 10544 |
+
"epoch": 0.6356802541026998,
|
| 10545 |
+
"grad_norm": 0.03654515743255615,
|
| 10546 |
+
"learning_rate": 2.8921248649885767e-07,
|
| 10547 |
+
"loss": 10.3239,
|
| 10548 |
+
"step": 1501
|
| 10549 |
+
},
|
| 10550 |
+
{
|
| 10551 |
+
"epoch": 0.6361037586024352,
|
| 10552 |
+
"grad_norm": 0.05791448801755905,
|
| 10553 |
+
"learning_rate": 2.73797678737886e-07,
|
| 10554 |
+
"loss": 10.3238,
|
| 10555 |
+
"step": 1502
|
| 10556 |
+
},
|
| 10557 |
+
{
|
| 10558 |
+
"epoch": 0.6365272631021704,
|
| 10559 |
+
"grad_norm": 0.03335098177194595,
|
| 10560 |
+
"learning_rate": 2.5880443366476104e-07,
|
| 10561 |
+
"loss": 10.3257,
|
| 10562 |
+
"step": 1503
|
| 10563 |
+
},
|
| 10564 |
+
{
|
| 10565 |
+
"epoch": 0.6369507676019057,
|
| 10566 |
+
"grad_norm": 0.03681592270731926,
|
| 10567 |
+
"learning_rate": 2.442328146589512e-07,
|
| 10568 |
+
"loss": 10.3269,
|
| 10569 |
+
"step": 1504
|
| 10570 |
+
},
|
| 10571 |
+
{
|
| 10572 |
+
"epoch": 0.6373742721016411,
|
| 10573 |
+
"grad_norm": 0.034652795642614365,
|
| 10574 |
+
"learning_rate": 2.3008288331762828e-07,
|
| 10575 |
+
"loss": 10.3344,
|
| 10576 |
+
"step": 1505
|
| 10577 |
+
},
|
| 10578 |
+
{
|
| 10579 |
+
"epoch": 0.6377977766013764,
|
| 10580 |
+
"grad_norm": 0.04854245483875275,
|
| 10581 |
+
"learning_rate": 2.163546994553789e-07,
|
| 10582 |
+
"loss": 10.3299,
|
| 10583 |
+
"step": 1506
|
| 10584 |
+
},
|
| 10585 |
+
{
|
| 10586 |
+
"epoch": 0.6382212811011116,
|
| 10587 |
+
"grad_norm": 0.0356970839202404,
|
| 10588 |
+
"learning_rate": 2.0304832110400461e-07,
|
| 10589 |
+
"loss": 10.3302,
|
| 10590 |
+
"step": 1507
|
| 10591 |
+
},
|
| 10592 |
+
{
|
| 10593 |
+
"epoch": 0.638644785600847,
|
| 10594 |
+
"grad_norm": 0.033104781061410904,
|
| 10595 |
+
"learning_rate": 1.9016380451223337e-07,
|
| 10596 |
+
"loss": 10.3274,
|
| 10597 |
+
"step": 1508
|
| 10598 |
+
},
|
| 10599 |
+
{
|
| 10600 |
+
"epoch": 0.6390682901005823,
|
| 10601 |
+
"grad_norm": 0.041414547711610794,
|
| 10602 |
+
"learning_rate": 1.7770120414550839e-07,
|
| 10603 |
+
"loss": 10.3232,
|
| 10604 |
+
"step": 1509
|
| 10605 |
+
},
|
| 10606 |
+
{
|
| 10607 |
+
"epoch": 0.6394917946003176,
|
| 10608 |
+
"grad_norm": 0.03412613272666931,
|
| 10609 |
+
"learning_rate": 1.6566057268574408e-07,
|
| 10610 |
+
"loss": 10.3271,
|
| 10611 |
+
"step": 1510
|
| 10612 |
+
},
|
| 10613 |
+
{
|
| 10614 |
+
"epoch": 0.639915299100053,
|
| 10615 |
+
"grad_norm": 0.03871205449104309,
|
| 10616 |
+
"learning_rate": 1.540419610311039e-07,
|
| 10617 |
+
"loss": 10.324,
|
| 10618 |
+
"step": 1511
|
| 10619 |
+
},
|
| 10620 |
+
{
|
| 10621 |
+
"epoch": 0.6403388035997882,
|
| 10622 |
+
"grad_norm": 0.04230944812297821,
|
| 10623 |
+
"learning_rate": 1.4284541829580057e-07,
|
| 10624 |
+
"loss": 10.3321,
|
| 10625 |
+
"step": 1512
|
| 10626 |
+
},
|
| 10627 |
+
{
|
| 10628 |
+
"epoch": 0.6407623080995235,
|
| 10629 |
+
"grad_norm": 0.03660792484879494,
|
| 10630 |
+
"learning_rate": 1.3207099180986282e-07,
|
| 10631 |
+
"loss": 10.3267,
|
| 10632 |
+
"step": 1513
|
| 10633 |
+
},
|
| 10634 |
+
{
|
| 10635 |
+
"epoch": 0.6411858125992589,
|
| 10636 |
+
"grad_norm": 0.03157137706875801,
|
| 10637 |
+
"learning_rate": 1.2171872711895794e-07,
|
| 10638 |
+
"loss": 10.3337,
|
| 10639 |
+
"step": 1514
|
| 10640 |
+
},
|
| 10641 |
+
{
|
| 10642 |
+
"epoch": 0.6416093170989942,
|
| 10643 |
+
"grad_norm": 0.02765282243490219,
|
| 10644 |
+
"learning_rate": 1.1178866798419174e-07,
|
| 10645 |
+
"loss": 10.3286,
|
| 10646 |
+
"step": 1515
|
| 10647 |
+
},
|
| 10648 |
+
{
|
| 10649 |
+
"epoch": 0.6420328215987295,
|
| 10650 |
+
"grad_norm": 0.03549804165959358,
|
| 10651 |
+
"learning_rate": 1.0228085638190887e-07,
|
| 10652 |
+
"loss": 10.3284,
|
| 10653 |
+
"step": 1516
|
| 10654 |
+
},
|
| 10655 |
+
{
|
| 10656 |
+
"epoch": 0.6424563260984648,
|
| 10657 |
+
"grad_norm": 0.03579467535018921,
|
| 10658 |
+
"learning_rate": 9.319533250353729e-08,
|
| 10659 |
+
"loss": 10.3334,
|
| 10660 |
+
"step": 1517
|
| 10661 |
+
},
|
| 10662 |
+
{
|
| 10663 |
+
"epoch": 0.6428798305982001,
|
| 10664 |
+
"grad_norm": 0.04964029788970947,
|
| 10665 |
+
"learning_rate": 8.453213475543287e-08,
|
| 10666 |
+
"loss": 10.3269,
|
| 10667 |
+
"step": 1518
|
| 10668 |
+
},
|
| 10669 |
+
{
|
| 10670 |
+
"epoch": 0.6433033350979355,
|
| 10671 |
+
"grad_norm": 0.030658867210149765,
|
| 10672 |
+
"learning_rate": 7.629129975864624e-08,
|
| 10673 |
+
"loss": 10.3271,
|
| 10674 |
+
"step": 1519
|
| 10675 |
+
},
|
| 10676 |
+
{
|
| 10677 |
+
"epoch": 0.6437268395976707,
|
| 10678 |
+
"grad_norm": 0.03056299313902855,
|
| 10679 |
+
"learning_rate": 6.847286234885619e-08,
|
| 10680 |
+
"loss": 10.324,
|
| 10681 |
+
"step": 1520
|
| 10682 |
+
},
|
| 10683 |
+
{
|
| 10684 |
+
"epoch": 0.644150344097406,
|
| 10685 |
+
"grad_norm": 0.03439069911837578,
|
| 10686 |
+
"learning_rate": 6.107685557616983e-08,
|
| 10687 |
+
"loss": 10.3264,
|
| 10688 |
+
"step": 1521
|
| 10689 |
+
},
|
| 10690 |
+
{
|
| 10691 |
+
"epoch": 0.6445738485971414,
|
| 10692 |
+
"grad_norm": 0.028876757249236107,
|
| 10693 |
+
"learning_rate": 5.4103310704989305e-08,
|
| 10694 |
+
"loss": 10.3303,
|
| 10695 |
+
"step": 1522
|
| 10696 |
+
},
|
| 10697 |
+
{
|
| 10698 |
+
"epoch": 0.6449973530968767,
|
| 10699 |
+
"grad_norm": 0.04483109340071678,
|
| 10700 |
+
"learning_rate": 4.755225721388978e-08,
|
| 10701 |
+
"loss": 10.3231,
|
| 10702 |
+
"step": 1523
|
| 10703 |
+
},
|
| 10704 |
+
{
|
| 10705 |
+
"epoch": 0.6454208575966119,
|
| 10706 |
+
"grad_norm": 0.035133130848407745,
|
| 10707 |
+
"learning_rate": 4.142372279548612e-08,
|
| 10708 |
+
"loss": 10.3289,
|
| 10709 |
+
"step": 1524
|
| 10710 |
+
},
|
| 10711 |
+
{
|
| 10712 |
+
"epoch": 0.6458443620963473,
|
| 10713 |
+
"grad_norm": 0.0331132635474205,
|
| 10714 |
+
"learning_rate": 3.5717733356333e-08,
|
| 10715 |
+
"loss": 10.3289,
|
| 10716 |
+
"step": 1525
|
| 10717 |
+
},
|
| 10718 |
+
{
|
| 10719 |
+
"epoch": 0.6462678665960826,
|
| 10720 |
+
"grad_norm": 0.03210371360182762,
|
| 10721 |
+
"learning_rate": 3.0434313016780567e-08,
|
| 10722 |
+
"loss": 10.328,
|
| 10723 |
+
"step": 1526
|
| 10724 |
+
},
|
| 10725 |
+
{
|
| 10726 |
+
"epoch": 0.6466913710958179,
|
| 10727 |
+
"grad_norm": 0.030365584418177605,
|
| 10728 |
+
"learning_rate": 2.557348411093008e-08,
|
| 10729 |
+
"loss": 10.3292,
|
| 10730 |
+
"step": 1527
|
| 10731 |
+
},
|
| 10732 |
+
{
|
| 10733 |
+
"epoch": 0.6471148755955533,
|
| 10734 |
+
"grad_norm": 0.03157270327210426,
|
| 10735 |
+
"learning_rate": 2.1135267186456197e-08,
|
| 10736 |
+
"loss": 10.3302,
|
| 10737 |
+
"step": 1528
|
| 10738 |
+
},
|
| 10739 |
+
{
|
| 10740 |
+
"epoch": 0.6475383800952885,
|
| 10741 |
+
"grad_norm": 0.03890909627079964,
|
| 10742 |
+
"learning_rate": 1.7119681004607037e-08,
|
| 10743 |
+
"loss": 10.3279,
|
| 10744 |
+
"step": 1529
|
| 10745 |
+
},
|
| 10746 |
+
{
|
| 10747 |
+
"epoch": 0.6479618845950238,
|
| 10748 |
+
"grad_norm": 0.0360557846724987,
|
| 10749 |
+
"learning_rate": 1.3526742540070913e-08,
|
| 10750 |
+
"loss": 10.3284,
|
| 10751 |
+
"step": 1530
|
| 10752 |
+
},
|
| 10753 |
+
{
|
| 10754 |
+
"epoch": 0.6483853890947592,
|
| 10755 |
+
"grad_norm": 0.030892444774508476,
|
| 10756 |
+
"learning_rate": 1.0356466980909752e-08,
|
| 10757 |
+
"loss": 10.3248,
|
| 10758 |
+
"step": 1531
|
| 10759 |
+
},
|
| 10760 |
+
{
|
| 10761 |
+
"epoch": 0.6488088935944945,
|
| 10762 |
+
"grad_norm": 0.04798031598329544,
|
| 10763 |
+
"learning_rate": 7.608867728536862e-09,
|
| 10764 |
+
"loss": 10.3258,
|
| 10765 |
+
"step": 1532
|
| 10766 |
+
},
|
| 10767 |
+
{
|
| 10768 |
+
"epoch": 0.6492323980942297,
|
| 10769 |
+
"grad_norm": 0.0375826396048069,
|
| 10770 |
+
"learning_rate": 5.283956397594825e-09,
|
| 10771 |
+
"loss": 10.3237,
|
| 10772 |
+
"step": 1533
|
| 10773 |
+
},
|
| 10774 |
+
{
|
| 10775 |
+
"epoch": 0.6496559025939651,
|
| 10776 |
+
"grad_norm": 0.03595992550253868,
|
| 10777 |
+
"learning_rate": 3.3817428159443886e-09,
|
| 10778 |
+
"loss": 10.3293,
|
| 10779 |
+
"step": 1534
|
| 10780 |
+
},
|
| 10781 |
+
{
|
| 10782 |
+
"epoch": 0.6500794070937004,
|
| 10783 |
+
"grad_norm": 0.04436947777867317,
|
| 10784 |
+
"learning_rate": 1.902235024620058e-09,
|
| 10785 |
+
"loss": 10.3288,
|
| 10786 |
+
"step": 1535
|
| 10787 |
+
},
|
| 10788 |
+
{
|
| 10789 |
+
"epoch": 0.6505029115934357,
|
| 10790 |
+
"grad_norm": 0.0375913567841053,
|
| 10791 |
+
"learning_rate": 8.454392778189935e-10,
|
| 10792 |
+
"loss": 10.328,
|
| 10793 |
+
"step": 1536
|
| 10794 |
+
},
|
| 10795 |
+
{
|
| 10796 |
+
"epoch": 0.650926416093171,
|
| 10797 |
+
"grad_norm": 0.03872830793261528,
|
| 10798 |
+
"learning_rate": 2.1136004282329425e-10,
|
| 10799 |
+
"loss": 10.327,
|
| 10800 |
+
"step": 1537
|
| 10801 |
+
},
|
| 10802 |
+
{
|
| 10803 |
+
"epoch": 0.6513499205929063,
|
| 10804 |
+
"grad_norm": 0.033878691494464874,
|
| 10805 |
+
"learning_rate": 0.0,
|
| 10806 |
+
"loss": 10.3287,
|
| 10807 |
+
"step": 1538
|
| 10808 |
}
|
| 10809 |
],
|
| 10810 |
"logging_steps": 1,
|
|
|
|
| 10819 |
"should_evaluate": false,
|
| 10820 |
"should_log": false,
|
| 10821 |
"should_save": true,
|
| 10822 |
+
"should_training_stop": true
|
| 10823 |
},
|
| 10824 |
"attributes": {}
|
| 10825 |
}
|
| 10826 |
},
|
| 10827 |
+
"total_flos": 32970583080960.0,
|
| 10828 |
"train_batch_size": 2,
|
| 10829 |
"trial_name": null,
|
| 10830 |
"trial_params": null
|