Training in progress, step 1398, 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 399632
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ace5f6fd569aa002f8e50b5e644ae4e1fc8f22b5a7eedcc53d713f250181f6d9
|
| 3 |
size 399632
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 531194
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c2872b59dff828ab0a9e8185cf6ebb37029409ed74f970689f62c84051969ee2
|
| 3 |
size 531194
|
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 sha256:8ccc937e836621898cc4102e58fe43f0f73b53f62c9ae401e376b29e0e79e97f
|
| 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 sha256:51033279aeda792f7ca86d40f672281cde5a848dbd42f1d736b1995bea420a0b
|
| 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": 350,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -7389,6 +7389,2442 @@
|
|
| 7389 |
"eval_samples_per_second": 212.202,
|
| 7390 |
"eval_steps_per_second": 106.187,
|
| 7391 |
"step": 1050
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 7392 |
}
|
| 7393 |
],
|
| 7394 |
"logging_steps": 1,
|
|
@@ -7403,12 +9839,12 @@
|
|
| 7403 |
"should_evaluate": false,
|
| 7404 |
"should_log": false,
|
| 7405 |
"should_save": true,
|
| 7406 |
-
"should_training_stop":
|
| 7407 |
},
|
| 7408 |
"attributes": {}
|
| 7409 |
}
|
| 7410 |
},
|
| 7411 |
-
"total_flos":
|
| 7412 |
"train_batch_size": 2,
|
| 7413 |
"trial_name": null,
|
| 7414 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": null,
|
| 3 |
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.476117496807152,
|
| 5 |
"eval_steps": 350,
|
| 6 |
+
"global_step": 1398,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 7389 |
"eval_samples_per_second": 212.202,
|
| 7390 |
"eval_steps_per_second": 106.187,
|
| 7391 |
"step": 1050
|
| 7392 |
+
},
|
| 7393 |
+
{
|
| 7394 |
+
"epoch": 0.35793954874414646,
|
| 7395 |
+
"grad_norm": 9016.091796875,
|
| 7396 |
+
"learning_rate": 2.9289321881345254e-05,
|
| 7397 |
+
"loss": 17.3543,
|
| 7398 |
+
"step": 1051
|
| 7399 |
+
},
|
| 7400 |
+
{
|
| 7401 |
+
"epoch": 0.3582801191996594,
|
| 7402 |
+
"grad_norm": 9640.099609375,
|
| 7403 |
+
"learning_rate": 2.9129456926492548e-05,
|
| 7404 |
+
"loss": 17.0311,
|
| 7405 |
+
"step": 1052
|
| 7406 |
+
},
|
| 7407 |
+
{
|
| 7408 |
+
"epoch": 0.3586206896551724,
|
| 7409 |
+
"grad_norm": 4575.740234375,
|
| 7410 |
+
"learning_rate": 2.8969955038315277e-05,
|
| 7411 |
+
"loss": 16.3338,
|
| 7412 |
+
"step": 1053
|
| 7413 |
+
},
|
| 7414 |
+
{
|
| 7415 |
+
"epoch": 0.3589612601106854,
|
| 7416 |
+
"grad_norm": 5429.5908203125,
|
| 7417 |
+
"learning_rate": 2.8810817033934656e-05,
|
| 7418 |
+
"loss": 16.5567,
|
| 7419 |
+
"step": 1054
|
| 7420 |
+
},
|
| 7421 |
+
{
|
| 7422 |
+
"epoch": 0.35930183056619835,
|
| 7423 |
+
"grad_norm": 7361.03125,
|
| 7424 |
+
"learning_rate": 2.8652043728607625e-05,
|
| 7425 |
+
"loss": 15.3858,
|
| 7426 |
+
"step": 1055
|
| 7427 |
+
},
|
| 7428 |
+
{
|
| 7429 |
+
"epoch": 0.35964240102171136,
|
| 7430 |
+
"grad_norm": 8986.16015625,
|
| 7431 |
+
"learning_rate": 2.8493635935722928e-05,
|
| 7432 |
+
"loss": 18.7631,
|
| 7433 |
+
"step": 1056
|
| 7434 |
+
},
|
| 7435 |
+
{
|
| 7436 |
+
"epoch": 0.35998297147722436,
|
| 7437 |
+
"grad_norm": 14620.3515625,
|
| 7438 |
+
"learning_rate": 2.8335594466796656e-05,
|
| 7439 |
+
"loss": 17.9584,
|
| 7440 |
+
"step": 1057
|
| 7441 |
+
},
|
| 7442 |
+
{
|
| 7443 |
+
"epoch": 0.36032354193273736,
|
| 7444 |
+
"grad_norm": 12298.29296875,
|
| 7445 |
+
"learning_rate": 2.8177920131468273e-05,
|
| 7446 |
+
"loss": 16.7883,
|
| 7447 |
+
"step": 1058
|
| 7448 |
+
},
|
| 7449 |
+
{
|
| 7450 |
+
"epoch": 0.3606641123882503,
|
| 7451 |
+
"grad_norm": 6544.71435546875,
|
| 7452 |
+
"learning_rate": 2.8020613737496547e-05,
|
| 7453 |
+
"loss": 15.5192,
|
| 7454 |
+
"step": 1059
|
| 7455 |
+
},
|
| 7456 |
+
{
|
| 7457 |
+
"epoch": 0.3610046828437633,
|
| 7458 |
+
"grad_norm": 43706.015625,
|
| 7459 |
+
"learning_rate": 2.7863676090755176e-05,
|
| 7460 |
+
"loss": 13.725,
|
| 7461 |
+
"step": 1060
|
| 7462 |
+
},
|
| 7463 |
+
{
|
| 7464 |
+
"epoch": 0.3613452532992763,
|
| 7465 |
+
"grad_norm": 10494.02734375,
|
| 7466 |
+
"learning_rate": 2.770710799522879e-05,
|
| 7467 |
+
"loss": 18.6528,
|
| 7468 |
+
"step": 1061
|
| 7469 |
+
},
|
| 7470 |
+
{
|
| 7471 |
+
"epoch": 0.36168582375478925,
|
| 7472 |
+
"grad_norm": 9141.2724609375,
|
| 7473 |
+
"learning_rate": 2.7550910253008933e-05,
|
| 7474 |
+
"loss": 16.2579,
|
| 7475 |
+
"step": 1062
|
| 7476 |
+
},
|
| 7477 |
+
{
|
| 7478 |
+
"epoch": 0.36202639421030225,
|
| 7479 |
+
"grad_norm": 19279.275390625,
|
| 7480 |
+
"learning_rate": 2.739508366428969e-05,
|
| 7481 |
+
"loss": 29.6578,
|
| 7482 |
+
"step": 1063
|
| 7483 |
+
},
|
| 7484 |
+
{
|
| 7485 |
+
"epoch": 0.36236696466581525,
|
| 7486 |
+
"grad_norm": 6543.4921875,
|
| 7487 |
+
"learning_rate": 2.723962902736389e-05,
|
| 7488 |
+
"loss": 16.8491,
|
| 7489 |
+
"step": 1064
|
| 7490 |
+
},
|
| 7491 |
+
{
|
| 7492 |
+
"epoch": 0.36270753512132825,
|
| 7493 |
+
"grad_norm": 212370.234375,
|
| 7494 |
+
"learning_rate": 2.7084547138618778e-05,
|
| 7495 |
+
"loss": 23.5439,
|
| 7496 |
+
"step": 1065
|
| 7497 |
+
},
|
| 7498 |
+
{
|
| 7499 |
+
"epoch": 0.3630481055768412,
|
| 7500 |
+
"grad_norm": 184131.515625,
|
| 7501 |
+
"learning_rate": 2.6929838792532037e-05,
|
| 7502 |
+
"loss": 41.623,
|
| 7503 |
+
"step": 1066
|
| 7504 |
+
},
|
| 7505 |
+
{
|
| 7506 |
+
"epoch": 0.3633886760323542,
|
| 7507 |
+
"grad_norm": 21050.416015625,
|
| 7508 |
+
"learning_rate": 2.6775504781667725e-05,
|
| 7509 |
+
"loss": 17.4598,
|
| 7510 |
+
"step": 1067
|
| 7511 |
+
},
|
| 7512 |
+
{
|
| 7513 |
+
"epoch": 0.3637292464878672,
|
| 7514 |
+
"grad_norm": 6398.333984375,
|
| 7515 |
+
"learning_rate": 2.6621545896672174e-05,
|
| 7516 |
+
"loss": 16.7309,
|
| 7517 |
+
"step": 1068
|
| 7518 |
+
},
|
| 7519 |
+
{
|
| 7520 |
+
"epoch": 0.36406981694338014,
|
| 7521 |
+
"grad_norm": 6304.9189453125,
|
| 7522 |
+
"learning_rate": 2.6467962926270017e-05,
|
| 7523 |
+
"loss": 15.6597,
|
| 7524 |
+
"step": 1069
|
| 7525 |
+
},
|
| 7526 |
+
{
|
| 7527 |
+
"epoch": 0.36441038739889314,
|
| 7528 |
+
"grad_norm": 7631.1796875,
|
| 7529 |
+
"learning_rate": 2.6314756657260054e-05,
|
| 7530 |
+
"loss": 16.0431,
|
| 7531 |
+
"step": 1070
|
| 7532 |
+
},
|
| 7533 |
+
{
|
| 7534 |
+
"epoch": 0.36475095785440614,
|
| 7535 |
+
"grad_norm": 7347.66064453125,
|
| 7536 |
+
"learning_rate": 2.6161927874511216e-05,
|
| 7537 |
+
"loss": 15.47,
|
| 7538 |
+
"step": 1071
|
| 7539 |
+
},
|
| 7540 |
+
{
|
| 7541 |
+
"epoch": 0.3650915283099191,
|
| 7542 |
+
"grad_norm": 7265.25390625,
|
| 7543 |
+
"learning_rate": 2.6009477360958712e-05,
|
| 7544 |
+
"loss": 15.6487,
|
| 7545 |
+
"step": 1072
|
| 7546 |
+
},
|
| 7547 |
+
{
|
| 7548 |
+
"epoch": 0.3654320987654321,
|
| 7549 |
+
"grad_norm": 6069.59716796875,
|
| 7550 |
+
"learning_rate": 2.585740589759976e-05,
|
| 7551 |
+
"loss": 17.3132,
|
| 7552 |
+
"step": 1073
|
| 7553 |
+
},
|
| 7554 |
+
{
|
| 7555 |
+
"epoch": 0.3657726692209451,
|
| 7556 |
+
"grad_norm": 4936.7646484375,
|
| 7557 |
+
"learning_rate": 2.5705714263489776e-05,
|
| 7558 |
+
"loss": 15.3776,
|
| 7559 |
+
"step": 1074
|
| 7560 |
+
},
|
| 7561 |
+
{
|
| 7562 |
+
"epoch": 0.3661132396764581,
|
| 7563 |
+
"grad_norm": 6368.90625,
|
| 7564 |
+
"learning_rate": 2.555440323573839e-05,
|
| 7565 |
+
"loss": 19.6503,
|
| 7566 |
+
"step": 1075
|
| 7567 |
+
},
|
| 7568 |
+
{
|
| 7569 |
+
"epoch": 0.36645381013197104,
|
| 7570 |
+
"grad_norm": 82272.46875,
|
| 7571 |
+
"learning_rate": 2.540347358950529e-05,
|
| 7572 |
+
"loss": 20.1391,
|
| 7573 |
+
"step": 1076
|
| 7574 |
+
},
|
| 7575 |
+
{
|
| 7576 |
+
"epoch": 0.36679438058748404,
|
| 7577 |
+
"grad_norm": 133955.78125,
|
| 7578 |
+
"learning_rate": 2.5252926097996445e-05,
|
| 7579 |
+
"loss": 34.3992,
|
| 7580 |
+
"step": 1077
|
| 7581 |
+
},
|
| 7582 |
+
{
|
| 7583 |
+
"epoch": 0.36713495104299704,
|
| 7584 |
+
"grad_norm": 22073.888671875,
|
| 7585 |
+
"learning_rate": 2.5102761532460008e-05,
|
| 7586 |
+
"loss": 18.7588,
|
| 7587 |
+
"step": 1078
|
| 7588 |
+
},
|
| 7589 |
+
{
|
| 7590 |
+
"epoch": 0.36747552149851,
|
| 7591 |
+
"grad_norm": 13817.8994140625,
|
| 7592 |
+
"learning_rate": 2.4952980662182425e-05,
|
| 7593 |
+
"loss": 18.0995,
|
| 7594 |
+
"step": 1079
|
| 7595 |
+
},
|
| 7596 |
+
{
|
| 7597 |
+
"epoch": 0.367816091954023,
|
| 7598 |
+
"grad_norm": 5851.48095703125,
|
| 7599 |
+
"learning_rate": 2.4803584254484568e-05,
|
| 7600 |
+
"loss": 15.224,
|
| 7601 |
+
"step": 1080
|
| 7602 |
+
},
|
| 7603 |
+
{
|
| 7604 |
+
"epoch": 0.368156662409536,
|
| 7605 |
+
"grad_norm": 4619.5224609375,
|
| 7606 |
+
"learning_rate": 2.4654573074717602e-05,
|
| 7607 |
+
"loss": 18.7544,
|
| 7608 |
+
"step": 1081
|
| 7609 |
+
},
|
| 7610 |
+
{
|
| 7611 |
+
"epoch": 0.36849723286504893,
|
| 7612 |
+
"grad_norm": 38233.484375,
|
| 7613 |
+
"learning_rate": 2.4505947886259318e-05,
|
| 7614 |
+
"loss": 17.1859,
|
| 7615 |
+
"step": 1082
|
| 7616 |
+
},
|
| 7617 |
+
{
|
| 7618 |
+
"epoch": 0.36883780332056193,
|
| 7619 |
+
"grad_norm": 13702.75390625,
|
| 7620 |
+
"learning_rate": 2.435770945050997e-05,
|
| 7621 |
+
"loss": 17.4571,
|
| 7622 |
+
"step": 1083
|
| 7623 |
+
},
|
| 7624 |
+
{
|
| 7625 |
+
"epoch": 0.36917837377607493,
|
| 7626 |
+
"grad_norm": 19916.818359375,
|
| 7627 |
+
"learning_rate": 2.420985852688854e-05,
|
| 7628 |
+
"loss": 25.3175,
|
| 7629 |
+
"step": 1084
|
| 7630 |
+
},
|
| 7631 |
+
{
|
| 7632 |
+
"epoch": 0.36951894423158793,
|
| 7633 |
+
"grad_norm": 17139.83203125,
|
| 7634 |
+
"learning_rate": 2.4062395872828846e-05,
|
| 7635 |
+
"loss": 16.3809,
|
| 7636 |
+
"step": 1085
|
| 7637 |
+
},
|
| 7638 |
+
{
|
| 7639 |
+
"epoch": 0.3698595146871009,
|
| 7640 |
+
"grad_norm": 32343.576171875,
|
| 7641 |
+
"learning_rate": 2.3915322243775562e-05,
|
| 7642 |
+
"loss": 22.5745,
|
| 7643 |
+
"step": 1086
|
| 7644 |
+
},
|
| 7645 |
+
{
|
| 7646 |
+
"epoch": 0.3702000851426139,
|
| 7647 |
+
"grad_norm": 89483.1640625,
|
| 7648 |
+
"learning_rate": 2.3768638393180407e-05,
|
| 7649 |
+
"loss": 23.2377,
|
| 7650 |
+
"step": 1087
|
| 7651 |
+
},
|
| 7652 |
+
{
|
| 7653 |
+
"epoch": 0.3705406555981269,
|
| 7654 |
+
"grad_norm": 13901.6123046875,
|
| 7655 |
+
"learning_rate": 2.362234507249832e-05,
|
| 7656 |
+
"loss": 16.3438,
|
| 7657 |
+
"step": 1088
|
| 7658 |
+
},
|
| 7659 |
+
{
|
| 7660 |
+
"epoch": 0.3708812260536398,
|
| 7661 |
+
"grad_norm": 7761.7646484375,
|
| 7662 |
+
"learning_rate": 2.3476443031183503e-05,
|
| 7663 |
+
"loss": 16.1746,
|
| 7664 |
+
"step": 1089
|
| 7665 |
+
},
|
| 7666 |
+
{
|
| 7667 |
+
"epoch": 0.3712217965091528,
|
| 7668 |
+
"grad_norm": 272404.53125,
|
| 7669 |
+
"learning_rate": 2.3330933016685754e-05,
|
| 7670 |
+
"loss": 17.872,
|
| 7671 |
+
"step": 1090
|
| 7672 |
+
},
|
| 7673 |
+
{
|
| 7674 |
+
"epoch": 0.3715623669646658,
|
| 7675 |
+
"grad_norm": 22655.146484375,
|
| 7676 |
+
"learning_rate": 2.318581577444646e-05,
|
| 7677 |
+
"loss": 18.6592,
|
| 7678 |
+
"step": 1091
|
| 7679 |
+
},
|
| 7680 |
+
{
|
| 7681 |
+
"epoch": 0.3719029374201788,
|
| 7682 |
+
"grad_norm": 10099.595703125,
|
| 7683 |
+
"learning_rate": 2.304109204789484e-05,
|
| 7684 |
+
"loss": 17.5878,
|
| 7685 |
+
"step": 1092
|
| 7686 |
+
},
|
| 7687 |
+
{
|
| 7688 |
+
"epoch": 0.37224350787569177,
|
| 7689 |
+
"grad_norm": 8442.248046875,
|
| 7690 |
+
"learning_rate": 2.289676257844423e-05,
|
| 7691 |
+
"loss": 18.1415,
|
| 7692 |
+
"step": 1093
|
| 7693 |
+
},
|
| 7694 |
+
{
|
| 7695 |
+
"epoch": 0.37258407833120477,
|
| 7696 |
+
"grad_norm": 5178.3564453125,
|
| 7697 |
+
"learning_rate": 2.275282810548811e-05,
|
| 7698 |
+
"loss": 15.1448,
|
| 7699 |
+
"step": 1094
|
| 7700 |
+
},
|
| 7701 |
+
{
|
| 7702 |
+
"epoch": 0.37292464878671777,
|
| 7703 |
+
"grad_norm": 9764.1787109375,
|
| 7704 |
+
"learning_rate": 2.2609289366396502e-05,
|
| 7705 |
+
"loss": 16.3465,
|
| 7706 |
+
"step": 1095
|
| 7707 |
+
},
|
| 7708 |
+
{
|
| 7709 |
+
"epoch": 0.3732652192422307,
|
| 7710 |
+
"grad_norm": 19003.330078125,
|
| 7711 |
+
"learning_rate": 2.2466147096512035e-05,
|
| 7712 |
+
"loss": 24.4724,
|
| 7713 |
+
"step": 1096
|
| 7714 |
+
},
|
| 7715 |
+
{
|
| 7716 |
+
"epoch": 0.3736057896977437,
|
| 7717 |
+
"grad_norm": 9890.1943359375,
|
| 7718 |
+
"learning_rate": 2.2323402029146244e-05,
|
| 7719 |
+
"loss": 16.9699,
|
| 7720 |
+
"step": 1097
|
| 7721 |
+
},
|
| 7722 |
+
{
|
| 7723 |
+
"epoch": 0.3739463601532567,
|
| 7724 |
+
"grad_norm": 7291.38330078125,
|
| 7725 |
+
"learning_rate": 2.2181054895575847e-05,
|
| 7726 |
+
"loss": 15.6684,
|
| 7727 |
+
"step": 1098
|
| 7728 |
+
},
|
| 7729 |
+
{
|
| 7730 |
+
"epoch": 0.37428693060876966,
|
| 7731 |
+
"grad_norm": 11182.4765625,
|
| 7732 |
+
"learning_rate": 2.2039106425038924e-05,
|
| 7733 |
+
"loss": 15.8272,
|
| 7734 |
+
"step": 1099
|
| 7735 |
+
},
|
| 7736 |
+
{
|
| 7737 |
+
"epoch": 0.37462750106428266,
|
| 7738 |
+
"grad_norm": 16134.654296875,
|
| 7739 |
+
"learning_rate": 2.189755734473129e-05,
|
| 7740 |
+
"loss": 14.9627,
|
| 7741 |
+
"step": 1100
|
| 7742 |
+
},
|
| 7743 |
+
{
|
| 7744 |
+
"epoch": 0.37496807151979566,
|
| 7745 |
+
"grad_norm": 8052.66796875,
|
| 7746 |
+
"learning_rate": 2.175640837980265e-05,
|
| 7747 |
+
"loss": 15.3499,
|
| 7748 |
+
"step": 1101
|
| 7749 |
+
},
|
| 7750 |
+
{
|
| 7751 |
+
"epoch": 0.37530864197530867,
|
| 7752 |
+
"grad_norm": 20322.666015625,
|
| 7753 |
+
"learning_rate": 2.161566025335289e-05,
|
| 7754 |
+
"loss": 16.7828,
|
| 7755 |
+
"step": 1102
|
| 7756 |
+
},
|
| 7757 |
+
{
|
| 7758 |
+
"epoch": 0.3756492124308216,
|
| 7759 |
+
"grad_norm": 4153.91455078125,
|
| 7760 |
+
"learning_rate": 2.1475313686428544e-05,
|
| 7761 |
+
"loss": 15.6488,
|
| 7762 |
+
"step": 1103
|
| 7763 |
+
},
|
| 7764 |
+
{
|
| 7765 |
+
"epoch": 0.3759897828863346,
|
| 7766 |
+
"grad_norm": 7169.74365234375,
|
| 7767 |
+
"learning_rate": 2.133536939801888e-05,
|
| 7768 |
+
"loss": 15.2577,
|
| 7769 |
+
"step": 1104
|
| 7770 |
+
},
|
| 7771 |
+
{
|
| 7772 |
+
"epoch": 0.3763303533418476,
|
| 7773 |
+
"grad_norm": 8697.2861328125,
|
| 7774 |
+
"learning_rate": 2.1195828105052283e-05,
|
| 7775 |
+
"loss": 15.7897,
|
| 7776 |
+
"step": 1105
|
| 7777 |
+
},
|
| 7778 |
+
{
|
| 7779 |
+
"epoch": 0.37667092379736056,
|
| 7780 |
+
"grad_norm": 15461.00390625,
|
| 7781 |
+
"learning_rate": 2.105669052239274e-05,
|
| 7782 |
+
"loss": 19.3535,
|
| 7783 |
+
"step": 1106
|
| 7784 |
+
},
|
| 7785 |
+
{
|
| 7786 |
+
"epoch": 0.37701149425287356,
|
| 7787 |
+
"grad_norm": 10803.2568359375,
|
| 7788 |
+
"learning_rate": 2.091795736283593e-05,
|
| 7789 |
+
"loss": 17.862,
|
| 7790 |
+
"step": 1107
|
| 7791 |
+
},
|
| 7792 |
+
{
|
| 7793 |
+
"epoch": 0.37735206470838656,
|
| 7794 |
+
"grad_norm": 11586.3193359375,
|
| 7795 |
+
"learning_rate": 2.0779629337105722e-05,
|
| 7796 |
+
"loss": 16.2983,
|
| 7797 |
+
"step": 1108
|
| 7798 |
+
},
|
| 7799 |
+
{
|
| 7800 |
+
"epoch": 0.37769263516389956,
|
| 7801 |
+
"grad_norm": 11368.2099609375,
|
| 7802 |
+
"learning_rate": 2.064170715385052e-05,
|
| 7803 |
+
"loss": 16.6631,
|
| 7804 |
+
"step": 1109
|
| 7805 |
+
},
|
| 7806 |
+
{
|
| 7807 |
+
"epoch": 0.3780332056194125,
|
| 7808 |
+
"grad_norm": 8291.4638671875,
|
| 7809 |
+
"learning_rate": 2.050419151963957e-05,
|
| 7810 |
+
"loss": 16.1702,
|
| 7811 |
+
"step": 1110
|
| 7812 |
+
},
|
| 7813 |
+
{
|
| 7814 |
+
"epoch": 0.3783737760749255,
|
| 7815 |
+
"grad_norm": 6780.28564453125,
|
| 7816 |
+
"learning_rate": 2.0367083138959476e-05,
|
| 7817 |
+
"loss": 18.0787,
|
| 7818 |
+
"step": 1111
|
| 7819 |
+
},
|
| 7820 |
+
{
|
| 7821 |
+
"epoch": 0.3787143465304385,
|
| 7822 |
+
"grad_norm": 9689.455078125,
|
| 7823 |
+
"learning_rate": 2.0230382714210384e-05,
|
| 7824 |
+
"loss": 14.1774,
|
| 7825 |
+
"step": 1112
|
| 7826 |
+
},
|
| 7827 |
+
{
|
| 7828 |
+
"epoch": 0.37905491698595145,
|
| 7829 |
+
"grad_norm": 8263.84765625,
|
| 7830 |
+
"learning_rate": 2.0094090945702616e-05,
|
| 7831 |
+
"loss": 18.9439,
|
| 7832 |
+
"step": 1113
|
| 7833 |
+
},
|
| 7834 |
+
{
|
| 7835 |
+
"epoch": 0.37939548744146445,
|
| 7836 |
+
"grad_norm": 581882.3125,
|
| 7837 |
+
"learning_rate": 1.9958208531652877e-05,
|
| 7838 |
+
"loss": 26.3807,
|
| 7839 |
+
"step": 1114
|
| 7840 |
+
},
|
| 7841 |
+
{
|
| 7842 |
+
"epoch": 0.37973605789697745,
|
| 7843 |
+
"grad_norm": 43517.40625,
|
| 7844 |
+
"learning_rate": 1.9822736168180778e-05,
|
| 7845 |
+
"loss": 22.9196,
|
| 7846 |
+
"step": 1115
|
| 7847 |
+
},
|
| 7848 |
+
{
|
| 7849 |
+
"epoch": 0.3800766283524904,
|
| 7850 |
+
"grad_norm": 185219.921875,
|
| 7851 |
+
"learning_rate": 1.9687674549305335e-05,
|
| 7852 |
+
"loss": 21.8024,
|
| 7853 |
+
"step": 1116
|
| 7854 |
+
},
|
| 7855 |
+
{
|
| 7856 |
+
"epoch": 0.3804171988080034,
|
| 7857 |
+
"grad_norm": 8173.0595703125,
|
| 7858 |
+
"learning_rate": 1.9553024366941242e-05,
|
| 7859 |
+
"loss": 17.0993,
|
| 7860 |
+
"step": 1117
|
| 7861 |
+
},
|
| 7862 |
+
{
|
| 7863 |
+
"epoch": 0.3807577692635164,
|
| 7864 |
+
"grad_norm": 4755.5341796875,
|
| 7865 |
+
"learning_rate": 1.9418786310895464e-05,
|
| 7866 |
+
"loss": 14.3767,
|
| 7867 |
+
"step": 1118
|
| 7868 |
+
},
|
| 7869 |
+
{
|
| 7870 |
+
"epoch": 0.3810983397190294,
|
| 7871 |
+
"grad_norm": 3945.259765625,
|
| 7872 |
+
"learning_rate": 1.9284961068863673e-05,
|
| 7873 |
+
"loss": 14.9059,
|
| 7874 |
+
"step": 1119
|
| 7875 |
+
},
|
| 7876 |
+
{
|
| 7877 |
+
"epoch": 0.38143891017454234,
|
| 7878 |
+
"grad_norm": 8374.2470703125,
|
| 7879 |
+
"learning_rate": 1.9151549326426656e-05,
|
| 7880 |
+
"loss": 15.8377,
|
| 7881 |
+
"step": 1120
|
| 7882 |
+
},
|
| 7883 |
+
{
|
| 7884 |
+
"epoch": 0.38177948063005535,
|
| 7885 |
+
"grad_norm": 7485.6328125,
|
| 7886 |
+
"learning_rate": 1.9018551767046966e-05,
|
| 7887 |
+
"loss": 17.28,
|
| 7888 |
+
"step": 1121
|
| 7889 |
+
},
|
| 7890 |
+
{
|
| 7891 |
+
"epoch": 0.38212005108556835,
|
| 7892 |
+
"grad_norm": 50922.8203125,
|
| 7893 |
+
"learning_rate": 1.8885969072065225e-05,
|
| 7894 |
+
"loss": 15.6675,
|
| 7895 |
+
"step": 1122
|
| 7896 |
+
},
|
| 7897 |
+
{
|
| 7898 |
+
"epoch": 0.3824606215410813,
|
| 7899 |
+
"grad_norm": 5945.76611328125,
|
| 7900 |
+
"learning_rate": 1.8753801920696712e-05,
|
| 7901 |
+
"loss": 14.9502,
|
| 7902 |
+
"step": 1123
|
| 7903 |
+
},
|
| 7904 |
+
{
|
| 7905 |
+
"epoch": 0.3828011919965943,
|
| 7906 |
+
"grad_norm": 200846.65625,
|
| 7907 |
+
"learning_rate": 1.8622050990027995e-05,
|
| 7908 |
+
"loss": 22.2888,
|
| 7909 |
+
"step": 1124
|
| 7910 |
+
},
|
| 7911 |
+
{
|
| 7912 |
+
"epoch": 0.3831417624521073,
|
| 7913 |
+
"grad_norm": 5579.0,
|
| 7914 |
+
"learning_rate": 1.8490716955013232e-05,
|
| 7915 |
+
"loss": 15.7866,
|
| 7916 |
+
"step": 1125
|
| 7917 |
+
},
|
| 7918 |
+
{
|
| 7919 |
+
"epoch": 0.38348233290762024,
|
| 7920 |
+
"grad_norm": 5315.63525390625,
|
| 7921 |
+
"learning_rate": 1.8359800488470978e-05,
|
| 7922 |
+
"loss": 15.4055,
|
| 7923 |
+
"step": 1126
|
| 7924 |
+
},
|
| 7925 |
+
{
|
| 7926 |
+
"epoch": 0.38382290336313324,
|
| 7927 |
+
"grad_norm": 9051.3017578125,
|
| 7928 |
+
"learning_rate": 1.8229302261080495e-05,
|
| 7929 |
+
"loss": 19.5171,
|
| 7930 |
+
"step": 1127
|
| 7931 |
+
},
|
| 7932 |
+
{
|
| 7933 |
+
"epoch": 0.38416347381864624,
|
| 7934 |
+
"grad_norm": 8009.796875,
|
| 7935 |
+
"learning_rate": 1.809922294137847e-05,
|
| 7936 |
+
"loss": 21.9852,
|
| 7937 |
+
"step": 1128
|
| 7938 |
+
},
|
| 7939 |
+
{
|
| 7940 |
+
"epoch": 0.38450404427415924,
|
| 7941 |
+
"grad_norm": 13589.388671875,
|
| 7942 |
+
"learning_rate": 1.7969563195755535e-05,
|
| 7943 |
+
"loss": 20.9339,
|
| 7944 |
+
"step": 1129
|
| 7945 |
+
},
|
| 7946 |
+
{
|
| 7947 |
+
"epoch": 0.3848446147296722,
|
| 7948 |
+
"grad_norm": 9078.146484375,
|
| 7949 |
+
"learning_rate": 1.784032368845283e-05,
|
| 7950 |
+
"loss": 14.0946,
|
| 7951 |
+
"step": 1130
|
| 7952 |
+
},
|
| 7953 |
+
{
|
| 7954 |
+
"epoch": 0.3851851851851852,
|
| 7955 |
+
"grad_norm": 8657.4013671875,
|
| 7956 |
+
"learning_rate": 1.7711505081558734e-05,
|
| 7957 |
+
"loss": 17.6492,
|
| 7958 |
+
"step": 1131
|
| 7959 |
+
},
|
| 7960 |
+
{
|
| 7961 |
+
"epoch": 0.3855257556406982,
|
| 7962 |
+
"grad_norm": 6441.10302734375,
|
| 7963 |
+
"learning_rate": 1.758310803500527e-05,
|
| 7964 |
+
"loss": 16.6765,
|
| 7965 |
+
"step": 1132
|
| 7966 |
+
},
|
| 7967 |
+
{
|
| 7968 |
+
"epoch": 0.38586632609621113,
|
| 7969 |
+
"grad_norm": 9048.892578125,
|
| 7970 |
+
"learning_rate": 1.7455133206564832e-05,
|
| 7971 |
+
"loss": 17.9983,
|
| 7972 |
+
"step": 1133
|
| 7973 |
+
},
|
| 7974 |
+
{
|
| 7975 |
+
"epoch": 0.38620689655172413,
|
| 7976 |
+
"grad_norm": 19367.732421875,
|
| 7977 |
+
"learning_rate": 1.73275812518469e-05,
|
| 7978 |
+
"loss": 21.241,
|
| 7979 |
+
"step": 1134
|
| 7980 |
+
},
|
| 7981 |
+
{
|
| 7982 |
+
"epoch": 0.38654746700723713,
|
| 7983 |
+
"grad_norm": 4897.97802734375,
|
| 7984 |
+
"learning_rate": 1.7200452824294498e-05,
|
| 7985 |
+
"loss": 16.6184,
|
| 7986 |
+
"step": 1135
|
| 7987 |
+
},
|
| 7988 |
+
{
|
| 7989 |
+
"epoch": 0.38688803746275013,
|
| 7990 |
+
"grad_norm": 11805.7529296875,
|
| 7991 |
+
"learning_rate": 1.707374857518094e-05,
|
| 7992 |
+
"loss": 16.1727,
|
| 7993 |
+
"step": 1136
|
| 7994 |
+
},
|
| 7995 |
+
{
|
| 7996 |
+
"epoch": 0.3872286079182631,
|
| 7997 |
+
"grad_norm": 10455.5537109375,
|
| 7998 |
+
"learning_rate": 1.6947469153606577e-05,
|
| 7999 |
+
"loss": 13.7626,
|
| 8000 |
+
"step": 1137
|
| 8001 |
+
},
|
| 8002 |
+
{
|
| 8003 |
+
"epoch": 0.3875691783737761,
|
| 8004 |
+
"grad_norm": 125463.8828125,
|
| 8005 |
+
"learning_rate": 1.6821615206495312e-05,
|
| 8006 |
+
"loss": 21.2509,
|
| 8007 |
+
"step": 1138
|
| 8008 |
+
},
|
| 8009 |
+
{
|
| 8010 |
+
"epoch": 0.3879097488292891,
|
| 8011 |
+
"grad_norm": 7911.271484375,
|
| 8012 |
+
"learning_rate": 1.6696187378591376e-05,
|
| 8013 |
+
"loss": 15.4732,
|
| 8014 |
+
"step": 1139
|
| 8015 |
+
},
|
| 8016 |
+
{
|
| 8017 |
+
"epoch": 0.388250319284802,
|
| 8018 |
+
"grad_norm": 8106.412109375,
|
| 8019 |
+
"learning_rate": 1.657118631245601e-05,
|
| 8020 |
+
"loss": 15.4888,
|
| 8021 |
+
"step": 1140
|
| 8022 |
+
},
|
| 8023 |
+
{
|
| 8024 |
+
"epoch": 0.388590889740315,
|
| 8025 |
+
"grad_norm": 12514.365234375,
|
| 8026 |
+
"learning_rate": 1.6446612648464164e-05,
|
| 8027 |
+
"loss": 16.7501,
|
| 8028 |
+
"step": 1141
|
| 8029 |
+
},
|
| 8030 |
+
{
|
| 8031 |
+
"epoch": 0.388931460195828,
|
| 8032 |
+
"grad_norm": 6361.5712890625,
|
| 8033 |
+
"learning_rate": 1.632246702480128e-05,
|
| 8034 |
+
"loss": 15.536,
|
| 8035 |
+
"step": 1142
|
| 8036 |
+
},
|
| 8037 |
+
{
|
| 8038 |
+
"epoch": 0.389272030651341,
|
| 8039 |
+
"grad_norm": 5804.33740234375,
|
| 8040 |
+
"learning_rate": 1.619875007745989e-05,
|
| 8041 |
+
"loss": 16.9502,
|
| 8042 |
+
"step": 1143
|
| 8043 |
+
},
|
| 8044 |
+
{
|
| 8045 |
+
"epoch": 0.389612601106854,
|
| 8046 |
+
"grad_norm": 16940.232421875,
|
| 8047 |
+
"learning_rate": 1.607546244023651e-05,
|
| 8048 |
+
"loss": 19.4423,
|
| 8049 |
+
"step": 1144
|
| 8050 |
+
},
|
| 8051 |
+
{
|
| 8052 |
+
"epoch": 0.389953171562367,
|
| 8053 |
+
"grad_norm": 18570.146484375,
|
| 8054 |
+
"learning_rate": 1.5952604744728272e-05,
|
| 8055 |
+
"loss": 14.8142,
|
| 8056 |
+
"step": 1145
|
| 8057 |
+
},
|
| 8058 |
+
{
|
| 8059 |
+
"epoch": 0.39029374201788,
|
| 8060 |
+
"grad_norm": 10272.63671875,
|
| 8061 |
+
"learning_rate": 1.5830177620329712e-05,
|
| 8062 |
+
"loss": 18.5908,
|
| 8063 |
+
"step": 1146
|
| 8064 |
+
},
|
| 8065 |
+
{
|
| 8066 |
+
"epoch": 0.3906343124733929,
|
| 8067 |
+
"grad_norm": 38148.55859375,
|
| 8068 |
+
"learning_rate": 1.570818169422966e-05,
|
| 8069 |
+
"loss": 20.6591,
|
| 8070 |
+
"step": 1147
|
| 8071 |
+
},
|
| 8072 |
+
{
|
| 8073 |
+
"epoch": 0.3909748829289059,
|
| 8074 |
+
"grad_norm": 5822.15283203125,
|
| 8075 |
+
"learning_rate": 1.558661759140786e-05,
|
| 8076 |
+
"loss": 14.0913,
|
| 8077 |
+
"step": 1148
|
| 8078 |
+
},
|
| 8079 |
+
{
|
| 8080 |
+
"epoch": 0.3913154533844189,
|
| 8081 |
+
"grad_norm": 14074.4423828125,
|
| 8082 |
+
"learning_rate": 1.5465485934631853e-05,
|
| 8083 |
+
"loss": 21.9162,
|
| 8084 |
+
"step": 1149
|
| 8085 |
+
},
|
| 8086 |
+
{
|
| 8087 |
+
"epoch": 0.39165602383993187,
|
| 8088 |
+
"grad_norm": 6325.41357421875,
|
| 8089 |
+
"learning_rate": 1.5344787344453805e-05,
|
| 8090 |
+
"loss": 16.2195,
|
| 8091 |
+
"step": 1150
|
| 8092 |
+
},
|
| 8093 |
+
{
|
| 8094 |
+
"epoch": 0.39199659429544487,
|
| 8095 |
+
"grad_norm": 127774.578125,
|
| 8096 |
+
"learning_rate": 1.5224522439207246e-05,
|
| 8097 |
+
"loss": 27.4325,
|
| 8098 |
+
"step": 1151
|
| 8099 |
+
},
|
| 8100 |
+
{
|
| 8101 |
+
"epoch": 0.39233716475095787,
|
| 8102 |
+
"grad_norm": 12522.1953125,
|
| 8103 |
+
"learning_rate": 1.5104691835004048e-05,
|
| 8104 |
+
"loss": 19.8276,
|
| 8105 |
+
"step": 1152
|
| 8106 |
+
},
|
| 8107 |
+
{
|
| 8108 |
+
"epoch": 0.3926777352064708,
|
| 8109 |
+
"grad_norm": 7569.482421875,
|
| 8110 |
+
"learning_rate": 1.498529614573111e-05,
|
| 8111 |
+
"loss": 15.6422,
|
| 8112 |
+
"step": 1153
|
| 8113 |
+
},
|
| 8114 |
+
{
|
| 8115 |
+
"epoch": 0.3930183056619838,
|
| 8116 |
+
"grad_norm": 4825.58056640625,
|
| 8117 |
+
"learning_rate": 1.4866335983047264e-05,
|
| 8118 |
+
"loss": 16.9562,
|
| 8119 |
+
"step": 1154
|
| 8120 |
+
},
|
| 8121 |
+
{
|
| 8122 |
+
"epoch": 0.3933588761174968,
|
| 8123 |
+
"grad_norm": 7882.283203125,
|
| 8124 |
+
"learning_rate": 1.4747811956380242e-05,
|
| 8125 |
+
"loss": 16.0085,
|
| 8126 |
+
"step": 1155
|
| 8127 |
+
},
|
| 8128 |
+
{
|
| 8129 |
+
"epoch": 0.3936994465730098,
|
| 8130 |
+
"grad_norm": 49450.83984375,
|
| 8131 |
+
"learning_rate": 1.4629724672923384e-05,
|
| 8132 |
+
"loss": 20.0522,
|
| 8133 |
+
"step": 1156
|
| 8134 |
+
},
|
| 8135 |
+
{
|
| 8136 |
+
"epoch": 0.39404001702852276,
|
| 8137 |
+
"grad_norm": 41817.09765625,
|
| 8138 |
+
"learning_rate": 1.4512074737632686e-05,
|
| 8139 |
+
"loss": 15.0558,
|
| 8140 |
+
"step": 1157
|
| 8141 |
+
},
|
| 8142 |
+
{
|
| 8143 |
+
"epoch": 0.39438058748403576,
|
| 8144 |
+
"grad_norm": 17985.326171875,
|
| 8145 |
+
"learning_rate": 1.439486275322357e-05,
|
| 8146 |
+
"loss": 17.9776,
|
| 8147 |
+
"step": 1158
|
| 8148 |
+
},
|
| 8149 |
+
{
|
| 8150 |
+
"epoch": 0.39472115793954876,
|
| 8151 |
+
"grad_norm": 32287.5625,
|
| 8152 |
+
"learning_rate": 1.4278089320167876e-05,
|
| 8153 |
+
"loss": 18.4001,
|
| 8154 |
+
"step": 1159
|
| 8155 |
+
},
|
| 8156 |
+
{
|
| 8157 |
+
"epoch": 0.3950617283950617,
|
| 8158 |
+
"grad_norm": 6474.7412109375,
|
| 8159 |
+
"learning_rate": 1.4161755036690771e-05,
|
| 8160 |
+
"loss": 16.0384,
|
| 8161 |
+
"step": 1160
|
| 8162 |
+
},
|
| 8163 |
+
{
|
| 8164 |
+
"epoch": 0.3954022988505747,
|
| 8165 |
+
"grad_norm": 21569.9921875,
|
| 8166 |
+
"learning_rate": 1.4045860498767671e-05,
|
| 8167 |
+
"loss": 22.5467,
|
| 8168 |
+
"step": 1161
|
| 8169 |
+
},
|
| 8170 |
+
{
|
| 8171 |
+
"epoch": 0.3957428693060877,
|
| 8172 |
+
"grad_norm": 14132.955078125,
|
| 8173 |
+
"learning_rate": 1.3930406300121179e-05,
|
| 8174 |
+
"loss": 18.0639,
|
| 8175 |
+
"step": 1162
|
| 8176 |
+
},
|
| 8177 |
+
{
|
| 8178 |
+
"epoch": 0.3960834397616007,
|
| 8179 |
+
"grad_norm": 54925.08984375,
|
| 8180 |
+
"learning_rate": 1.3815393032218115e-05,
|
| 8181 |
+
"loss": 20.547,
|
| 8182 |
+
"step": 1163
|
| 8183 |
+
},
|
| 8184 |
+
{
|
| 8185 |
+
"epoch": 0.39642401021711365,
|
| 8186 |
+
"grad_norm": 10113.197265625,
|
| 8187 |
+
"learning_rate": 1.3700821284266351e-05,
|
| 8188 |
+
"loss": 17.8488,
|
| 8189 |
+
"step": 1164
|
| 8190 |
+
},
|
| 8191 |
+
{
|
| 8192 |
+
"epoch": 0.39676458067262665,
|
| 8193 |
+
"grad_norm": 14740.6005859375,
|
| 8194 |
+
"learning_rate": 1.3586691643211957e-05,
|
| 8195 |
+
"loss": 17.4523,
|
| 8196 |
+
"step": 1165
|
| 8197 |
+
},
|
| 8198 |
+
{
|
| 8199 |
+
"epoch": 0.39710515112813966,
|
| 8200 |
+
"grad_norm": 25044.357421875,
|
| 8201 |
+
"learning_rate": 1.3473004693736036e-05,
|
| 8202 |
+
"loss": 14.8799,
|
| 8203 |
+
"step": 1166
|
| 8204 |
+
},
|
| 8205 |
+
{
|
| 8206 |
+
"epoch": 0.3974457215836526,
|
| 8207 |
+
"grad_norm": 11815.2529296875,
|
| 8208 |
+
"learning_rate": 1.3359761018251826e-05,
|
| 8209 |
+
"loss": 16.286,
|
| 8210 |
+
"step": 1167
|
| 8211 |
+
},
|
| 8212 |
+
{
|
| 8213 |
+
"epoch": 0.3977862920391656,
|
| 8214 |
+
"grad_norm": 5622.22900390625,
|
| 8215 |
+
"learning_rate": 1.324696119690173e-05,
|
| 8216 |
+
"loss": 16.7942,
|
| 8217 |
+
"step": 1168
|
| 8218 |
+
},
|
| 8219 |
+
{
|
| 8220 |
+
"epoch": 0.3981268624946786,
|
| 8221 |
+
"grad_norm": 31308.080078125,
|
| 8222 |
+
"learning_rate": 1.3134605807554246e-05,
|
| 8223 |
+
"loss": 17.2891,
|
| 8224 |
+
"step": 1169
|
| 8225 |
+
},
|
| 8226 |
+
{
|
| 8227 |
+
"epoch": 0.39846743295019155,
|
| 8228 |
+
"grad_norm": 17020.232421875,
|
| 8229 |
+
"learning_rate": 1.302269542580109e-05,
|
| 8230 |
+
"loss": 16.6771,
|
| 8231 |
+
"step": 1170
|
| 8232 |
+
},
|
| 8233 |
+
{
|
| 8234 |
+
"epoch": 0.39880800340570455,
|
| 8235 |
+
"grad_norm": 5603.55908203125,
|
| 8236 |
+
"learning_rate": 1.291123062495424e-05,
|
| 8237 |
+
"loss": 15.5519,
|
| 8238 |
+
"step": 1171
|
| 8239 |
+
},
|
| 8240 |
+
{
|
| 8241 |
+
"epoch": 0.39914857386121755,
|
| 8242 |
+
"grad_norm": 15514.7919921875,
|
| 8243 |
+
"learning_rate": 1.2800211976042941e-05,
|
| 8244 |
+
"loss": 18.2958,
|
| 8245 |
+
"step": 1172
|
| 8246 |
+
},
|
| 8247 |
+
{
|
| 8248 |
+
"epoch": 0.39948914431673055,
|
| 8249 |
+
"grad_norm": 14003.712890625,
|
| 8250 |
+
"learning_rate": 1.268964004781089e-05,
|
| 8251 |
+
"loss": 15.9076,
|
| 8252 |
+
"step": 1173
|
| 8253 |
+
},
|
| 8254 |
+
{
|
| 8255 |
+
"epoch": 0.3998297147722435,
|
| 8256 |
+
"grad_norm": 12478.578125,
|
| 8257 |
+
"learning_rate": 1.2579515406713193e-05,
|
| 8258 |
+
"loss": 16.1259,
|
| 8259 |
+
"step": 1174
|
| 8260 |
+
},
|
| 8261 |
+
{
|
| 8262 |
+
"epoch": 0.4001702852277565,
|
| 8263 |
+
"grad_norm": 4842.283203125,
|
| 8264 |
+
"learning_rate": 1.246983861691352e-05,
|
| 8265 |
+
"loss": 15.9744,
|
| 8266 |
+
"step": 1175
|
| 8267 |
+
},
|
| 8268 |
+
{
|
| 8269 |
+
"epoch": 0.4005108556832695,
|
| 8270 |
+
"grad_norm": 8658.658203125,
|
| 8271 |
+
"learning_rate": 1.236061024028129e-05,
|
| 8272 |
+
"loss": 18.019,
|
| 8273 |
+
"step": 1176
|
| 8274 |
+
},
|
| 8275 |
+
{
|
| 8276 |
+
"epoch": 0.40085142613878244,
|
| 8277 |
+
"grad_norm": 5850.47021484375,
|
| 8278 |
+
"learning_rate": 1.2251830836388622e-05,
|
| 8279 |
+
"loss": 16.8717,
|
| 8280 |
+
"step": 1177
|
| 8281 |
+
},
|
| 8282 |
+
{
|
| 8283 |
+
"epoch": 0.40119199659429544,
|
| 8284 |
+
"grad_norm": 13316.6796875,
|
| 8285 |
+
"learning_rate": 1.214350096250767e-05,
|
| 8286 |
+
"loss": 14.8869,
|
| 8287 |
+
"step": 1178
|
| 8288 |
+
},
|
| 8289 |
+
{
|
| 8290 |
+
"epoch": 0.40153256704980844,
|
| 8291 |
+
"grad_norm": 729673.375,
|
| 8292 |
+
"learning_rate": 1.2035621173607581e-05,
|
| 8293 |
+
"loss": 16.5223,
|
| 8294 |
+
"step": 1179
|
| 8295 |
+
},
|
| 8296 |
+
{
|
| 8297 |
+
"epoch": 0.4018731375053214,
|
| 8298 |
+
"grad_norm": 7339.2041015625,
|
| 8299 |
+
"learning_rate": 1.192819202235178e-05,
|
| 8300 |
+
"loss": 17.0888,
|
| 8301 |
+
"step": 1180
|
| 8302 |
+
},
|
| 8303 |
+
{
|
| 8304 |
+
"epoch": 0.4022137079608344,
|
| 8305 |
+
"grad_norm": 24987.361328125,
|
| 8306 |
+
"learning_rate": 1.1821214059095088e-05,
|
| 8307 |
+
"loss": 16.3947,
|
| 8308 |
+
"step": 1181
|
| 8309 |
+
},
|
| 8310 |
+
{
|
| 8311 |
+
"epoch": 0.4025542784163474,
|
| 8312 |
+
"grad_norm": 20888.8515625,
|
| 8313 |
+
"learning_rate": 1.1714687831880865e-05,
|
| 8314 |
+
"loss": 18.692,
|
| 8315 |
+
"step": 1182
|
| 8316 |
+
},
|
| 8317 |
+
{
|
| 8318 |
+
"epoch": 0.4028948488718604,
|
| 8319 |
+
"grad_norm": 5645.3427734375,
|
| 8320 |
+
"learning_rate": 1.1608613886438346e-05,
|
| 8321 |
+
"loss": 15.4461,
|
| 8322 |
+
"step": 1183
|
| 8323 |
+
},
|
| 8324 |
+
{
|
| 8325 |
+
"epoch": 0.40323541932737333,
|
| 8326 |
+
"grad_norm": 8248.0419921875,
|
| 8327 |
+
"learning_rate": 1.1502992766179666e-05,
|
| 8328 |
+
"loss": 17.0269,
|
| 8329 |
+
"step": 1184
|
| 8330 |
+
},
|
| 8331 |
+
{
|
| 8332 |
+
"epoch": 0.40357598978288634,
|
| 8333 |
+
"grad_norm": 5726.205078125,
|
| 8334 |
+
"learning_rate": 1.139782501219715e-05,
|
| 8335 |
+
"loss": 16.3764,
|
| 8336 |
+
"step": 1185
|
| 8337 |
+
},
|
| 8338 |
+
{
|
| 8339 |
+
"epoch": 0.40391656023839934,
|
| 8340 |
+
"grad_norm": 13888.7041015625,
|
| 8341 |
+
"learning_rate": 1.1293111163260639e-05,
|
| 8342 |
+
"loss": 15.9403,
|
| 8343 |
+
"step": 1186
|
| 8344 |
+
},
|
| 8345 |
+
{
|
| 8346 |
+
"epoch": 0.4042571306939123,
|
| 8347 |
+
"grad_norm": 29682.453125,
|
| 8348 |
+
"learning_rate": 1.118885175581451e-05,
|
| 8349 |
+
"loss": 16.3516,
|
| 8350 |
+
"step": 1187
|
| 8351 |
+
},
|
| 8352 |
+
{
|
| 8353 |
+
"epoch": 0.4045977011494253,
|
| 8354 |
+
"grad_norm": 8840.107421875,
|
| 8355 |
+
"learning_rate": 1.1085047323975173e-05,
|
| 8356 |
+
"loss": 17.1984,
|
| 8357 |
+
"step": 1188
|
| 8358 |
+
},
|
| 8359 |
+
{
|
| 8360 |
+
"epoch": 0.4049382716049383,
|
| 8361 |
+
"grad_norm": 7931.0126953125,
|
| 8362 |
+
"learning_rate": 1.0981698399528151e-05,
|
| 8363 |
+
"loss": 14.5578,
|
| 8364 |
+
"step": 1189
|
| 8365 |
+
},
|
| 8366 |
+
{
|
| 8367 |
+
"epoch": 0.4052788420604513,
|
| 8368 |
+
"grad_norm": 9766.60546875,
|
| 8369 |
+
"learning_rate": 1.0878805511925438e-05,
|
| 8370 |
+
"loss": 15.2207,
|
| 8371 |
+
"step": 1190
|
| 8372 |
+
},
|
| 8373 |
+
{
|
| 8374 |
+
"epoch": 0.40561941251596423,
|
| 8375 |
+
"grad_norm": 14362.5185546875,
|
| 8376 |
+
"learning_rate": 1.0776369188282775e-05,
|
| 8377 |
+
"loss": 16.0247,
|
| 8378 |
+
"step": 1191
|
| 8379 |
+
},
|
| 8380 |
+
{
|
| 8381 |
+
"epoch": 0.40595998297147723,
|
| 8382 |
+
"grad_norm": 8455.6484375,
|
| 8383 |
+
"learning_rate": 1.0674389953376928e-05,
|
| 8384 |
+
"loss": 19.4151,
|
| 8385 |
+
"step": 1192
|
| 8386 |
+
},
|
| 8387 |
+
{
|
| 8388 |
+
"epoch": 0.40630055342699023,
|
| 8389 |
+
"grad_norm": 5301.83740234375,
|
| 8390 |
+
"learning_rate": 1.0572868329643027e-05,
|
| 8391 |
+
"loss": 15.8368,
|
| 8392 |
+
"step": 1193
|
| 8393 |
+
},
|
| 8394 |
+
{
|
| 8395 |
+
"epoch": 0.4066411238825032,
|
| 8396 |
+
"grad_norm": 5115.203125,
|
| 8397 |
+
"learning_rate": 1.0471804837171916e-05,
|
| 8398 |
+
"loss": 15.3727,
|
| 8399 |
+
"step": 1194
|
| 8400 |
+
},
|
| 8401 |
+
{
|
| 8402 |
+
"epoch": 0.4069816943380162,
|
| 8403 |
+
"grad_norm": 6488.130859375,
|
| 8404 |
+
"learning_rate": 1.0371199993707392e-05,
|
| 8405 |
+
"loss": 15.679,
|
| 8406 |
+
"step": 1195
|
| 8407 |
+
},
|
| 8408 |
+
{
|
| 8409 |
+
"epoch": 0.4073222647935292,
|
| 8410 |
+
"grad_norm": 6294.16650390625,
|
| 8411 |
+
"learning_rate": 1.027105431464368e-05,
|
| 8412 |
+
"loss": 15.6614,
|
| 8413 |
+
"step": 1196
|
| 8414 |
+
},
|
| 8415 |
+
{
|
| 8416 |
+
"epoch": 0.4076628352490421,
|
| 8417 |
+
"grad_norm": 20521.2421875,
|
| 8418 |
+
"learning_rate": 1.0171368313022677e-05,
|
| 8419 |
+
"loss": 17.586,
|
| 8420 |
+
"step": 1197
|
| 8421 |
+
},
|
| 8422 |
+
{
|
| 8423 |
+
"epoch": 0.4080034057045551,
|
| 8424 |
+
"grad_norm": 14800.046875,
|
| 8425 |
+
"learning_rate": 1.0072142499531344e-05,
|
| 8426 |
+
"loss": 17.7802,
|
| 8427 |
+
"step": 1198
|
| 8428 |
+
},
|
| 8429 |
+
{
|
| 8430 |
+
"epoch": 0.4083439761600681,
|
| 8431 |
+
"grad_norm": 7743.5634765625,
|
| 8432 |
+
"learning_rate": 9.973377382499227e-06,
|
| 8433 |
+
"loss": 14.6301,
|
| 8434 |
+
"step": 1199
|
| 8435 |
+
},
|
| 8436 |
+
{
|
| 8437 |
+
"epoch": 0.4086845466155811,
|
| 8438 |
+
"grad_norm": 6069.34375,
|
| 8439 |
+
"learning_rate": 9.875073467895634e-06,
|
| 8440 |
+
"loss": 16.4805,
|
| 8441 |
+
"step": 1200
|
| 8442 |
+
},
|
| 8443 |
+
{
|
| 8444 |
+
"epoch": 0.40902511707109407,
|
| 8445 |
+
"grad_norm": 25095.98046875,
|
| 8446 |
+
"learning_rate": 9.777231259327212e-06,
|
| 8447 |
+
"loss": 21.9673,
|
| 8448 |
+
"step": 1201
|
| 8449 |
+
},
|
| 8450 |
+
{
|
| 8451 |
+
"epoch": 0.40936568752660707,
|
| 8452 |
+
"grad_norm": 35090.16015625,
|
| 8453 |
+
"learning_rate": 9.679851258035277e-06,
|
| 8454 |
+
"loss": 22.9736,
|
| 8455 |
+
"step": 1202
|
| 8456 |
+
},
|
| 8457 |
+
{
|
| 8458 |
+
"epoch": 0.40970625798212007,
|
| 8459 |
+
"grad_norm": 6459.52587890625,
|
| 8460 |
+
"learning_rate": 9.582933962893293e-06,
|
| 8461 |
+
"loss": 14.4086,
|
| 8462 |
+
"step": 1203
|
| 8463 |
+
},
|
| 8464 |
+
{
|
| 8465 |
+
"epoch": 0.410046828437633,
|
| 8466 |
+
"grad_norm": 7384.51025390625,
|
| 8467 |
+
"learning_rate": 9.48647987040433e-06,
|
| 8468 |
+
"loss": 18.5129,
|
| 8469 |
+
"step": 1204
|
| 8470 |
+
},
|
| 8471 |
+
{
|
| 8472 |
+
"epoch": 0.410387398893146,
|
| 8473 |
+
"grad_norm": 8684.4658203125,
|
| 8474 |
+
"learning_rate": 9.390489474698439e-06,
|
| 8475 |
+
"loss": 17.6151,
|
| 8476 |
+
"step": 1205
|
| 8477 |
+
},
|
| 8478 |
+
{
|
| 8479 |
+
"epoch": 0.410727969348659,
|
| 8480 |
+
"grad_norm": 5376.8583984375,
|
| 8481 |
+
"learning_rate": 9.294963267530176e-06,
|
| 8482 |
+
"loss": 15.7366,
|
| 8483 |
+
"step": 1206
|
| 8484 |
+
},
|
| 8485 |
+
{
|
| 8486 |
+
"epoch": 0.41106853980417196,
|
| 8487 |
+
"grad_norm": 12402.8740234375,
|
| 8488 |
+
"learning_rate": 9.19990173827615e-06,
|
| 8489 |
+
"loss": 19.216,
|
| 8490 |
+
"step": 1207
|
| 8491 |
+
},
|
| 8492 |
+
{
|
| 8493 |
+
"epoch": 0.41140911025968496,
|
| 8494 |
+
"grad_norm": 535757.375,
|
| 8495 |
+
"learning_rate": 9.105305373932338e-06,
|
| 8496 |
+
"loss": 25.0988,
|
| 8497 |
+
"step": 1208
|
| 8498 |
+
},
|
| 8499 |
+
{
|
| 8500 |
+
"epoch": 0.41174968071519796,
|
| 8501 |
+
"grad_norm": 7353.357421875,
|
| 8502 |
+
"learning_rate": 9.01117465911181e-06,
|
| 8503 |
+
"loss": 16.4689,
|
| 8504 |
+
"step": 1209
|
| 8505 |
+
},
|
| 8506 |
+
{
|
| 8507 |
+
"epoch": 0.41209025117071096,
|
| 8508 |
+
"grad_norm": 34880.5234375,
|
| 8509 |
+
"learning_rate": 8.917510076042057e-06,
|
| 8510 |
+
"loss": 17.3109,
|
| 8511 |
+
"step": 1210
|
| 8512 |
+
},
|
| 8513 |
+
{
|
| 8514 |
+
"epoch": 0.4124308216262239,
|
| 8515 |
+
"grad_norm": 57832.23046875,
|
| 8516 |
+
"learning_rate": 8.824312104562615e-06,
|
| 8517 |
+
"loss": 18.9605,
|
| 8518 |
+
"step": 1211
|
| 8519 |
+
},
|
| 8520 |
+
{
|
| 8521 |
+
"epoch": 0.4127713920817369,
|
| 8522 |
+
"grad_norm": 8867.4931640625,
|
| 8523 |
+
"learning_rate": 8.731581222122587e-06,
|
| 8524 |
+
"loss": 16.0094,
|
| 8525 |
+
"step": 1212
|
| 8526 |
+
},
|
| 8527 |
+
{
|
| 8528 |
+
"epoch": 0.4131119625372499,
|
| 8529 |
+
"grad_norm": 4194.0400390625,
|
| 8530 |
+
"learning_rate": 8.639317903778189e-06,
|
| 8531 |
+
"loss": 16.842,
|
| 8532 |
+
"step": 1213
|
| 8533 |
+
},
|
| 8534 |
+
{
|
| 8535 |
+
"epoch": 0.41345253299276286,
|
| 8536 |
+
"grad_norm": 36969.97265625,
|
| 8537 |
+
"learning_rate": 8.547522622190385e-06,
|
| 8538 |
+
"loss": 28.8818,
|
| 8539 |
+
"step": 1214
|
| 8540 |
+
},
|
| 8541 |
+
{
|
| 8542 |
+
"epoch": 0.41379310344827586,
|
| 8543 |
+
"grad_norm": 6608.91650390625,
|
| 8544 |
+
"learning_rate": 8.45619584762235e-06,
|
| 8545 |
+
"loss": 19.3621,
|
| 8546 |
+
"step": 1215
|
| 8547 |
+
},
|
| 8548 |
+
{
|
| 8549 |
+
"epoch": 0.41413367390378886,
|
| 8550 |
+
"grad_norm": 51639.66796875,
|
| 8551 |
+
"learning_rate": 8.365338047937121e-06,
|
| 8552 |
+
"loss": 25.1496,
|
| 8553 |
+
"step": 1216
|
| 8554 |
+
},
|
| 8555 |
+
{
|
| 8556 |
+
"epoch": 0.41447424435930186,
|
| 8557 |
+
"grad_norm": 11334.556640625,
|
| 8558 |
+
"learning_rate": 8.274949688595224e-06,
|
| 8559 |
+
"loss": 16.7963,
|
| 8560 |
+
"step": 1217
|
| 8561 |
+
},
|
| 8562 |
+
{
|
| 8563 |
+
"epoch": 0.4148148148148148,
|
| 8564 |
+
"grad_norm": 147909.71875,
|
| 8565 |
+
"learning_rate": 8.185031232652251e-06,
|
| 8566 |
+
"loss": 15.2766,
|
| 8567 |
+
"step": 1218
|
| 8568 |
+
},
|
| 8569 |
+
{
|
| 8570 |
+
"epoch": 0.4151553852703278,
|
| 8571 |
+
"grad_norm": 11644.173828125,
|
| 8572 |
+
"learning_rate": 8.095583140756468e-06,
|
| 8573 |
+
"loss": 16.4577,
|
| 8574 |
+
"step": 1219
|
| 8575 |
+
},
|
| 8576 |
+
{
|
| 8577 |
+
"epoch": 0.4154959557258408,
|
| 8578 |
+
"grad_norm": 4928.1103515625,
|
| 8579 |
+
"learning_rate": 8.006605871146577e-06,
|
| 8580 |
+
"loss": 15.5102,
|
| 8581 |
+
"step": 1220
|
| 8582 |
+
},
|
| 8583 |
+
{
|
| 8584 |
+
"epoch": 0.41583652618135375,
|
| 8585 |
+
"grad_norm": 29271.455078125,
|
| 8586 |
+
"learning_rate": 7.918099879649144e-06,
|
| 8587 |
+
"loss": 19.6712,
|
| 8588 |
+
"step": 1221
|
| 8589 |
+
},
|
| 8590 |
+
{
|
| 8591 |
+
"epoch": 0.41617709663686675,
|
| 8592 |
+
"grad_norm": 6887.5703125,
|
| 8593 |
+
"learning_rate": 7.830065619676518e-06,
|
| 8594 |
+
"loss": 16.2319,
|
| 8595 |
+
"step": 1222
|
| 8596 |
+
},
|
| 8597 |
+
{
|
| 8598 |
+
"epoch": 0.41651766709237975,
|
| 8599 |
+
"grad_norm": 85129.625,
|
| 8600 |
+
"learning_rate": 7.742503542224334e-06,
|
| 8601 |
+
"loss": 20.2058,
|
| 8602 |
+
"step": 1223
|
| 8603 |
+
},
|
| 8604 |
+
{
|
| 8605 |
+
"epoch": 0.4168582375478927,
|
| 8606 |
+
"grad_norm": 4843.767578125,
|
| 8607 |
+
"learning_rate": 7.65541409586924e-06,
|
| 8608 |
+
"loss": 15.402,
|
| 8609 |
+
"step": 1224
|
| 8610 |
+
},
|
| 8611 |
+
{
|
| 8612 |
+
"epoch": 0.4171988080034057,
|
| 8613 |
+
"grad_norm": 10344.7509765625,
|
| 8614 |
+
"learning_rate": 7.568797726766686e-06,
|
| 8615 |
+
"loss": 17.7642,
|
| 8616 |
+
"step": 1225
|
| 8617 |
+
},
|
| 8618 |
+
{
|
| 8619 |
+
"epoch": 0.4175393784589187,
|
| 8620 |
+
"grad_norm": 52196.25390625,
|
| 8621 |
+
"learning_rate": 7.482654878648465e-06,
|
| 8622 |
+
"loss": 21.7869,
|
| 8623 |
+
"step": 1226
|
| 8624 |
+
},
|
| 8625 |
+
{
|
| 8626 |
+
"epoch": 0.4178799489144317,
|
| 8627 |
+
"grad_norm": 4337.2626953125,
|
| 8628 |
+
"learning_rate": 7.396985992820648e-06,
|
| 8629 |
+
"loss": 18.0511,
|
| 8630 |
+
"step": 1227
|
| 8631 |
+
},
|
| 8632 |
+
{
|
| 8633 |
+
"epoch": 0.41822051936994464,
|
| 8634 |
+
"grad_norm": 185742.5,
|
| 8635 |
+
"learning_rate": 7.311791508161159e-06,
|
| 8636 |
+
"loss": 24.8165,
|
| 8637 |
+
"step": 1228
|
| 8638 |
+
},
|
| 8639 |
+
{
|
| 8640 |
+
"epoch": 0.41856108982545764,
|
| 8641 |
+
"grad_norm": 7755.84130859375,
|
| 8642 |
+
"learning_rate": 7.227071861117562e-06,
|
| 8643 |
+
"loss": 16.6333,
|
| 8644 |
+
"step": 1229
|
| 8645 |
+
},
|
| 8646 |
+
{
|
| 8647 |
+
"epoch": 0.41890166028097064,
|
| 8648 |
+
"grad_norm": 10233.14453125,
|
| 8649 |
+
"learning_rate": 7.14282748570495e-06,
|
| 8650 |
+
"loss": 14.4851,
|
| 8651 |
+
"step": 1230
|
| 8652 |
+
},
|
| 8653 |
+
{
|
| 8654 |
+
"epoch": 0.4192422307364836,
|
| 8655 |
+
"grad_norm": 33464.37109375,
|
| 8656 |
+
"learning_rate": 7.059058813503483e-06,
|
| 8657 |
+
"loss": 23.639,
|
| 8658 |
+
"step": 1231
|
| 8659 |
+
},
|
| 8660 |
+
{
|
| 8661 |
+
"epoch": 0.4195828011919966,
|
| 8662 |
+
"grad_norm": 4944.98876953125,
|
| 8663 |
+
"learning_rate": 6.975766273656425e-06,
|
| 8664 |
+
"loss": 16.1417,
|
| 8665 |
+
"step": 1232
|
| 8666 |
+
},
|
| 8667 |
+
{
|
| 8668 |
+
"epoch": 0.4199233716475096,
|
| 8669 |
+
"grad_norm": 4892.0244140625,
|
| 8670 |
+
"learning_rate": 6.892950292867784e-06,
|
| 8671 |
+
"loss": 18.6007,
|
| 8672 |
+
"step": 1233
|
| 8673 |
+
},
|
| 8674 |
+
{
|
| 8675 |
+
"epoch": 0.42026394210302254,
|
| 8676 |
+
"grad_norm": 19357.00390625,
|
| 8677 |
+
"learning_rate": 6.810611295400171e-06,
|
| 8678 |
+
"loss": 17.0264,
|
| 8679 |
+
"step": 1234
|
| 8680 |
+
},
|
| 8681 |
+
{
|
| 8682 |
+
"epoch": 0.42060451255853554,
|
| 8683 |
+
"grad_norm": 6050.00634765625,
|
| 8684 |
+
"learning_rate": 6.728749703072679e-06,
|
| 8685 |
+
"loss": 14.3341,
|
| 8686 |
+
"step": 1235
|
| 8687 |
+
},
|
| 8688 |
+
{
|
| 8689 |
+
"epoch": 0.42094508301404854,
|
| 8690 |
+
"grad_norm": 7513.56787109375,
|
| 8691 |
+
"learning_rate": 6.647365935258642e-06,
|
| 8692 |
+
"loss": 16.7578,
|
| 8693 |
+
"step": 1236
|
| 8694 |
+
},
|
| 8695 |
+
{
|
| 8696 |
+
"epoch": 0.42128565346956154,
|
| 8697 |
+
"grad_norm": 8825.5263671875,
|
| 8698 |
+
"learning_rate": 6.56646040888349e-06,
|
| 8699 |
+
"loss": 16.8743,
|
| 8700 |
+
"step": 1237
|
| 8701 |
+
},
|
| 8702 |
+
{
|
| 8703 |
+
"epoch": 0.4216262239250745,
|
| 8704 |
+
"grad_norm": 12023.978515625,
|
| 8705 |
+
"learning_rate": 6.48603353842272e-06,
|
| 8706 |
+
"loss": 16.0098,
|
| 8707 |
+
"step": 1238
|
| 8708 |
+
},
|
| 8709 |
+
{
|
| 8710 |
+
"epoch": 0.4219667943805875,
|
| 8711 |
+
"grad_norm": 445020.15625,
|
| 8712 |
+
"learning_rate": 6.406085735899625e-06,
|
| 8713 |
+
"loss": 21.9515,
|
| 8714 |
+
"step": 1239
|
| 8715 |
+
},
|
| 8716 |
+
{
|
| 8717 |
+
"epoch": 0.4223073648361005,
|
| 8718 |
+
"grad_norm": 387298.3125,
|
| 8719 |
+
"learning_rate": 6.326617410883295e-06,
|
| 8720 |
+
"loss": 27.4016,
|
| 8721 |
+
"step": 1240
|
| 8722 |
+
},
|
| 8723 |
+
{
|
| 8724 |
+
"epoch": 0.42264793529161343,
|
| 8725 |
+
"grad_norm": 9483.25,
|
| 8726 |
+
"learning_rate": 6.247628970486463e-06,
|
| 8727 |
+
"loss": 15.6205,
|
| 8728 |
+
"step": 1241
|
| 8729 |
+
},
|
| 8730 |
+
{
|
| 8731 |
+
"epoch": 0.42298850574712643,
|
| 8732 |
+
"grad_norm": 49060.64453125,
|
| 8733 |
+
"learning_rate": 6.169120819363405e-06,
|
| 8734 |
+
"loss": 16.1669,
|
| 8735 |
+
"step": 1242
|
| 8736 |
+
},
|
| 8737 |
+
{
|
| 8738 |
+
"epoch": 0.42332907620263943,
|
| 8739 |
+
"grad_norm": 9993.9287109375,
|
| 8740 |
+
"learning_rate": 6.091093359707977e-06,
|
| 8741 |
+
"loss": 16.4029,
|
| 8742 |
+
"step": 1243
|
| 8743 |
+
},
|
| 8744 |
+
{
|
| 8745 |
+
"epoch": 0.42366964665815243,
|
| 8746 |
+
"grad_norm": 9434.439453125,
|
| 8747 |
+
"learning_rate": 6.013546991251373e-06,
|
| 8748 |
+
"loss": 16.5022,
|
| 8749 |
+
"step": 1244
|
| 8750 |
+
},
|
| 8751 |
+
{
|
| 8752 |
+
"epoch": 0.4240102171136654,
|
| 8753 |
+
"grad_norm": 12037.0615234375,
|
| 8754 |
+
"learning_rate": 5.936482111260278e-06,
|
| 8755 |
+
"loss": 14.414,
|
| 8756 |
+
"step": 1245
|
| 8757 |
+
},
|
| 8758 |
+
{
|
| 8759 |
+
"epoch": 0.4243507875691784,
|
| 8760 |
+
"grad_norm": 10470.322265625,
|
| 8761 |
+
"learning_rate": 5.859899114534661e-06,
|
| 8762 |
+
"loss": 15.7482,
|
| 8763 |
+
"step": 1246
|
| 8764 |
+
},
|
| 8765 |
+
{
|
| 8766 |
+
"epoch": 0.4246913580246914,
|
| 8767 |
+
"grad_norm": 13876.06640625,
|
| 8768 |
+
"learning_rate": 5.783798393405826e-06,
|
| 8769 |
+
"loss": 16.1122,
|
| 8770 |
+
"step": 1247
|
| 8771 |
+
},
|
| 8772 |
+
{
|
| 8773 |
+
"epoch": 0.4250319284802043,
|
| 8774 |
+
"grad_norm": 4980.587890625,
|
| 8775 |
+
"learning_rate": 5.708180337734448e-06,
|
| 8776 |
+
"loss": 15.3704,
|
| 8777 |
+
"step": 1248
|
| 8778 |
+
},
|
| 8779 |
+
{
|
| 8780 |
+
"epoch": 0.4253724989357173,
|
| 8781 |
+
"grad_norm": 6813.8017578125,
|
| 8782 |
+
"learning_rate": 5.633045334908493e-06,
|
| 8783 |
+
"loss": 17.365,
|
| 8784 |
+
"step": 1249
|
| 8785 |
+
},
|
| 8786 |
+
{
|
| 8787 |
+
"epoch": 0.4257130693912303,
|
| 8788 |
+
"grad_norm": 10425.455078125,
|
| 8789 |
+
"learning_rate": 5.5583937698412856e-06,
|
| 8790 |
+
"loss": 16.8553,
|
| 8791 |
+
"step": 1250
|
| 8792 |
+
},
|
| 8793 |
+
{
|
| 8794 |
+
"epoch": 0.42605363984674327,
|
| 8795 |
+
"grad_norm": 13005.51171875,
|
| 8796 |
+
"learning_rate": 5.4842260249694964e-06,
|
| 8797 |
+
"loss": 16.6651,
|
| 8798 |
+
"step": 1251
|
| 8799 |
+
},
|
| 8800 |
+
{
|
| 8801 |
+
"epoch": 0.42639421030225627,
|
| 8802 |
+
"grad_norm": 26289.990234375,
|
| 8803 |
+
"learning_rate": 5.410542480251202e-06,
|
| 8804 |
+
"loss": 20.0705,
|
| 8805 |
+
"step": 1252
|
| 8806 |
+
},
|
| 8807 |
+
{
|
| 8808 |
+
"epoch": 0.42673478075776927,
|
| 8809 |
+
"grad_norm": 5403.8447265625,
|
| 8810 |
+
"learning_rate": 5.337343513164006e-06,
|
| 8811 |
+
"loss": 14.9475,
|
| 8812 |
+
"step": 1253
|
| 8813 |
+
},
|
| 8814 |
+
{
|
| 8815 |
+
"epoch": 0.4270753512132823,
|
| 8816 |
+
"grad_norm": 9579.4716796875,
|
| 8817 |
+
"learning_rate": 5.264629498702967e-06,
|
| 8818 |
+
"loss": 17.8629,
|
| 8819 |
+
"step": 1254
|
| 8820 |
+
},
|
| 8821 |
+
{
|
| 8822 |
+
"epoch": 0.4274159216687952,
|
| 8823 |
+
"grad_norm": 6556.83447265625,
|
| 8824 |
+
"learning_rate": 5.192400809378783e-06,
|
| 8825 |
+
"loss": 15.712,
|
| 8826 |
+
"step": 1255
|
| 8827 |
+
},
|
| 8828 |
+
{
|
| 8829 |
+
"epoch": 0.4277564921243082,
|
| 8830 |
+
"grad_norm": 4352.24560546875,
|
| 8831 |
+
"learning_rate": 5.120657815215879e-06,
|
| 8832 |
+
"loss": 15.5358,
|
| 8833 |
+
"step": 1256
|
| 8834 |
+
},
|
| 8835 |
+
{
|
| 8836 |
+
"epoch": 0.4280970625798212,
|
| 8837 |
+
"grad_norm": 1119215.625,
|
| 8838 |
+
"learning_rate": 5.0494008837504214e-06,
|
| 8839 |
+
"loss": 30.8416,
|
| 8840 |
+
"step": 1257
|
| 8841 |
+
},
|
| 8842 |
+
{
|
| 8843 |
+
"epoch": 0.42843763303533416,
|
| 8844 |
+
"grad_norm": 8932.01171875,
|
| 8845 |
+
"learning_rate": 4.978630380028582e-06,
|
| 8846 |
+
"loss": 16.3537,
|
| 8847 |
+
"step": 1258
|
| 8848 |
+
},
|
| 8849 |
+
{
|
| 8850 |
+
"epoch": 0.42877820349084717,
|
| 8851 |
+
"grad_norm": 708067.375,
|
| 8852 |
+
"learning_rate": 4.908346666604502e-06,
|
| 8853 |
+
"loss": 20.7258,
|
| 8854 |
+
"step": 1259
|
| 8855 |
+
},
|
| 8856 |
+
{
|
| 8857 |
+
"epoch": 0.42911877394636017,
|
| 8858 |
+
"grad_norm": 90272.8984375,
|
| 8859 |
+
"learning_rate": 4.8385501035385746e-06,
|
| 8860 |
+
"loss": 19.9443,
|
| 8861 |
+
"step": 1260
|
| 8862 |
+
},
|
| 8863 |
+
{
|
| 8864 |
+
"epoch": 0.4294593444018731,
|
| 8865 |
+
"grad_norm": 7611.046875,
|
| 8866 |
+
"learning_rate": 4.769241048395512e-06,
|
| 8867 |
+
"loss": 16.7847,
|
| 8868 |
+
"step": 1261
|
| 8869 |
+
},
|
| 8870 |
+
{
|
| 8871 |
+
"epoch": 0.4297999148573861,
|
| 8872 |
+
"grad_norm": 4970.75634765625,
|
| 8873 |
+
"learning_rate": 4.700419856242555e-06,
|
| 8874 |
+
"loss": 14.9592,
|
| 8875 |
+
"step": 1262
|
| 8876 |
+
},
|
| 8877 |
+
{
|
| 8878 |
+
"epoch": 0.4301404853128991,
|
| 8879 |
+
"grad_norm": 100432.1796875,
|
| 8880 |
+
"learning_rate": 4.632086879647635e-06,
|
| 8881 |
+
"loss": 15.552,
|
| 8882 |
+
"step": 1263
|
| 8883 |
+
},
|
| 8884 |
+
{
|
| 8885 |
+
"epoch": 0.4304810557684121,
|
| 8886 |
+
"grad_norm": 66502.8046875,
|
| 8887 |
+
"learning_rate": 4.564242468677615e-06,
|
| 8888 |
+
"loss": 19.3739,
|
| 8889 |
+
"step": 1264
|
| 8890 |
+
},
|
| 8891 |
+
{
|
| 8892 |
+
"epoch": 0.43082162622392506,
|
| 8893 |
+
"grad_norm": 6716.9658203125,
|
| 8894 |
+
"learning_rate": 4.496886970896396e-06,
|
| 8895 |
+
"loss": 15.4825,
|
| 8896 |
+
"step": 1265
|
| 8897 |
+
},
|
| 8898 |
+
{
|
| 8899 |
+
"epoch": 0.43116219667943806,
|
| 8900 |
+
"grad_norm": 4253.85400390625,
|
| 8901 |
+
"learning_rate": 4.430020731363271e-06,
|
| 8902 |
+
"loss": 15.5192,
|
| 8903 |
+
"step": 1266
|
| 8904 |
+
},
|
| 8905 |
+
{
|
| 8906 |
+
"epoch": 0.43150276713495106,
|
| 8907 |
+
"grad_norm": 7634.0341796875,
|
| 8908 |
+
"learning_rate": 4.3636440926310144e-06,
|
| 8909 |
+
"loss": 15.7577,
|
| 8910 |
+
"step": 1267
|
| 8911 |
+
},
|
| 8912 |
+
{
|
| 8913 |
+
"epoch": 0.431843337590464,
|
| 8914 |
+
"grad_norm": 70074.7109375,
|
| 8915 |
+
"learning_rate": 4.2977573947442175e-06,
|
| 8916 |
+
"loss": 20.0215,
|
| 8917 |
+
"step": 1268
|
| 8918 |
+
},
|
| 8919 |
+
{
|
| 8920 |
+
"epoch": 0.432183908045977,
|
| 8921 |
+
"grad_norm": 27130.666015625,
|
| 8922 |
+
"learning_rate": 4.232360975237571e-06,
|
| 8923 |
+
"loss": 21.458,
|
| 8924 |
+
"step": 1269
|
| 8925 |
+
},
|
| 8926 |
+
{
|
| 8927 |
+
"epoch": 0.43252447850149,
|
| 8928 |
+
"grad_norm": 179363.8125,
|
| 8929 |
+
"learning_rate": 4.167455169134027e-06,
|
| 8930 |
+
"loss": 15.0974,
|
| 8931 |
+
"step": 1270
|
| 8932 |
+
},
|
| 8933 |
+
{
|
| 8934 |
+
"epoch": 0.432865048957003,
|
| 8935 |
+
"grad_norm": 7130.95654296875,
|
| 8936 |
+
"learning_rate": 4.103040308943195e-06,
|
| 8937 |
+
"loss": 17.9218,
|
| 8938 |
+
"step": 1271
|
| 8939 |
+
},
|
| 8940 |
+
{
|
| 8941 |
+
"epoch": 0.43320561941251595,
|
| 8942 |
+
"grad_norm": 20771.81640625,
|
| 8943 |
+
"learning_rate": 4.039116724659564e-06,
|
| 8944 |
+
"loss": 16.0444,
|
| 8945 |
+
"step": 1272
|
| 8946 |
+
},
|
| 8947 |
+
{
|
| 8948 |
+
"epoch": 0.43354618986802895,
|
| 8949 |
+
"grad_norm": 8334.24609375,
|
| 8950 |
+
"learning_rate": 3.975684743760832e-06,
|
| 8951 |
+
"loss": 16.1791,
|
| 8952 |
+
"step": 1273
|
| 8953 |
+
},
|
| 8954 |
+
{
|
| 8955 |
+
"epoch": 0.43388676032354195,
|
| 8956 |
+
"grad_norm": 35193.2265625,
|
| 8957 |
+
"learning_rate": 3.91274469120626e-06,
|
| 8958 |
+
"loss": 27.447,
|
| 8959 |
+
"step": 1274
|
| 8960 |
+
},
|
| 8961 |
+
{
|
| 8962 |
+
"epoch": 0.4342273307790549,
|
| 8963 |
+
"grad_norm": 9046.802734375,
|
| 8964 |
+
"learning_rate": 3.850296889434968e-06,
|
| 8965 |
+
"loss": 16.1523,
|
| 8966 |
+
"step": 1275
|
| 8967 |
+
},
|
| 8968 |
+
{
|
| 8969 |
+
"epoch": 0.4345679012345679,
|
| 8970 |
+
"grad_norm": 7756.3515625,
|
| 8971 |
+
"learning_rate": 3.788341658364314e-06,
|
| 8972 |
+
"loss": 16.2914,
|
| 8973 |
+
"step": 1276
|
| 8974 |
+
},
|
| 8975 |
+
{
|
| 8976 |
+
"epoch": 0.4349084716900809,
|
| 8977 |
+
"grad_norm": 9144.1083984375,
|
| 8978 |
+
"learning_rate": 3.726879315388199e-06,
|
| 8979 |
+
"loss": 18.0792,
|
| 8980 |
+
"step": 1277
|
| 8981 |
+
},
|
| 8982 |
+
{
|
| 8983 |
+
"epoch": 0.43524904214559385,
|
| 8984 |
+
"grad_norm": 8346.263671875,
|
| 8985 |
+
"learning_rate": 3.665910175375498e-06,
|
| 8986 |
+
"loss": 18.9423,
|
| 8987 |
+
"step": 1278
|
| 8988 |
+
},
|
| 8989 |
+
{
|
| 8990 |
+
"epoch": 0.43558961260110685,
|
| 8991 |
+
"grad_norm": 4618.94970703125,
|
| 8992 |
+
"learning_rate": 3.6054345506684627e-06,
|
| 8993 |
+
"loss": 16.439,
|
| 8994 |
+
"step": 1279
|
| 8995 |
+
},
|
| 8996 |
+
{
|
| 8997 |
+
"epoch": 0.43593018305661985,
|
| 8998 |
+
"grad_norm": 7234.6689453125,
|
| 8999 |
+
"learning_rate": 3.5454527510810352e-06,
|
| 9000 |
+
"loss": 17.5187,
|
| 9001 |
+
"step": 1280
|
| 9002 |
+
},
|
| 9003 |
+
{
|
| 9004 |
+
"epoch": 0.43627075351213285,
|
| 9005 |
+
"grad_norm": 8177.525390625,
|
| 9006 |
+
"learning_rate": 3.485965083897347e-06,
|
| 9007 |
+
"loss": 15.7604,
|
| 9008 |
+
"step": 1281
|
| 9009 |
+
},
|
| 9010 |
+
{
|
| 9011 |
+
"epoch": 0.4366113239676458,
|
| 9012 |
+
"grad_norm": 7070.98486328125,
|
| 9013 |
+
"learning_rate": 3.426971853870109e-06,
|
| 9014 |
+
"loss": 15.3893,
|
| 9015 |
+
"step": 1282
|
| 9016 |
+
},
|
| 9017 |
+
{
|
| 9018 |
+
"epoch": 0.4369518944231588,
|
| 9019 |
+
"grad_norm": 5454.4384765625,
|
| 9020 |
+
"learning_rate": 3.3684733632190157e-06,
|
| 9021 |
+
"loss": 16.2147,
|
| 9022 |
+
"step": 1283
|
| 9023 |
+
},
|
| 9024 |
+
{
|
| 9025 |
+
"epoch": 0.4372924648786718,
|
| 9026 |
+
"grad_norm": 6743.98388671875,
|
| 9027 |
+
"learning_rate": 3.310469911629288e-06,
|
| 9028 |
+
"loss": 15.3899,
|
| 9029 |
+
"step": 1284
|
| 9030 |
+
},
|
| 9031 |
+
{
|
| 9032 |
+
"epoch": 0.43763303533418474,
|
| 9033 |
+
"grad_norm": 6508.32373046875,
|
| 9034 |
+
"learning_rate": 3.252961796250054e-06,
|
| 9035 |
+
"loss": 15.603,
|
| 9036 |
+
"step": 1285
|
| 9037 |
+
},
|
| 9038 |
+
{
|
| 9039 |
+
"epoch": 0.43797360578969774,
|
| 9040 |
+
"grad_norm": 7391.115234375,
|
| 9041 |
+
"learning_rate": 3.1959493116928476e-06,
|
| 9042 |
+
"loss": 17.4832,
|
| 9043 |
+
"step": 1286
|
| 9044 |
+
},
|
| 9045 |
+
{
|
| 9046 |
+
"epoch": 0.43831417624521074,
|
| 9047 |
+
"grad_norm": 12391.041015625,
|
| 9048 |
+
"learning_rate": 3.1394327500301357e-06,
|
| 9049 |
+
"loss": 21.1157,
|
| 9050 |
+
"step": 1287
|
| 9051 |
+
},
|
| 9052 |
+
{
|
| 9053 |
+
"epoch": 0.4386547467007237,
|
| 9054 |
+
"grad_norm": 5282.49462890625,
|
| 9055 |
+
"learning_rate": 3.0834124007937614e-06,
|
| 9056 |
+
"loss": 13.754,
|
| 9057 |
+
"step": 1288
|
| 9058 |
+
},
|
| 9059 |
+
{
|
| 9060 |
+
"epoch": 0.4389953171562367,
|
| 9061 |
+
"grad_norm": 8414.6005859375,
|
| 9062 |
+
"learning_rate": 3.0278885509735234e-06,
|
| 9063 |
+
"loss": 18.947,
|
| 9064 |
+
"step": 1289
|
| 9065 |
+
},
|
| 9066 |
+
{
|
| 9067 |
+
"epoch": 0.4393358876117497,
|
| 9068 |
+
"grad_norm": 7058.46875,
|
| 9069 |
+
"learning_rate": 2.9728614850156653e-06,
|
| 9070 |
+
"loss": 16.05,
|
| 9071 |
+
"step": 1290
|
| 9072 |
+
},
|
| 9073 |
+
{
|
| 9074 |
+
"epoch": 0.4396764580672627,
|
| 9075 |
+
"grad_norm": 7870.88037109375,
|
| 9076 |
+
"learning_rate": 2.9183314848214127e-06,
|
| 9077 |
+
"loss": 15.7352,
|
| 9078 |
+
"step": 1291
|
| 9079 |
+
},
|
| 9080 |
+
{
|
| 9081 |
+
"epoch": 0.44001702852277563,
|
| 9082 |
+
"grad_norm": 17012.94921875,
|
| 9083 |
+
"learning_rate": 2.864298829745571e-06,
|
| 9084 |
+
"loss": 15.5253,
|
| 9085 |
+
"step": 1292
|
| 9086 |
+
},
|
| 9087 |
+
{
|
| 9088 |
+
"epoch": 0.44035759897828863,
|
| 9089 |
+
"grad_norm": 6111.1611328125,
|
| 9090 |
+
"learning_rate": 2.8107637965950506e-06,
|
| 9091 |
+
"loss": 15.5991,
|
| 9092 |
+
"step": 1293
|
| 9093 |
+
},
|
| 9094 |
+
{
|
| 9095 |
+
"epoch": 0.44069816943380163,
|
| 9096 |
+
"grad_norm": 605149.3125,
|
| 9097 |
+
"learning_rate": 2.7577266596274576e-06,
|
| 9098 |
+
"loss": 18.6256,
|
| 9099 |
+
"step": 1294
|
| 9100 |
+
},
|
| 9101 |
+
{
|
| 9102 |
+
"epoch": 0.4410387398893146,
|
| 9103 |
+
"grad_norm": 11858.4453125,
|
| 9104 |
+
"learning_rate": 2.7051876905497375e-06,
|
| 9105 |
+
"loss": 14.8108,
|
| 9106 |
+
"step": 1295
|
| 9107 |
+
},
|
| 9108 |
+
{
|
| 9109 |
+
"epoch": 0.4413793103448276,
|
| 9110 |
+
"grad_norm": 818897.9375,
|
| 9111 |
+
"learning_rate": 2.6531471585167e-06,
|
| 9112 |
+
"loss": 16.1163,
|
| 9113 |
+
"step": 1296
|
| 9114 |
+
},
|
| 9115 |
+
{
|
| 9116 |
+
"epoch": 0.4417198808003406,
|
| 9117 |
+
"grad_norm": 7953.72802734375,
|
| 9118 |
+
"learning_rate": 2.6016053301297196e-06,
|
| 9119 |
+
"loss": 15.9181,
|
| 9120 |
+
"step": 1297
|
| 9121 |
+
},
|
| 9122 |
+
{
|
| 9123 |
+
"epoch": 0.4420604512558536,
|
| 9124 |
+
"grad_norm": 10290.4580078125,
|
| 9125 |
+
"learning_rate": 2.5505624694353024e-06,
|
| 9126 |
+
"loss": 16.1613,
|
| 9127 |
+
"step": 1298
|
| 9128 |
+
},
|
| 9129 |
+
{
|
| 9130 |
+
"epoch": 0.4424010217113665,
|
| 9131 |
+
"grad_norm": 15566.787109375,
|
| 9132 |
+
"learning_rate": 2.5000188379237786e-06,
|
| 9133 |
+
"loss": 20.0587,
|
| 9134 |
+
"step": 1299
|
| 9135 |
+
},
|
| 9136 |
+
{
|
| 9137 |
+
"epoch": 0.4427415921668795,
|
| 9138 |
+
"grad_norm": 8086.72998046875,
|
| 9139 |
+
"learning_rate": 2.4499746945279566e-06,
|
| 9140 |
+
"loss": 17.9596,
|
| 9141 |
+
"step": 1300
|
| 9142 |
+
},
|
| 9143 |
+
{
|
| 9144 |
+
"epoch": 0.44308216262239253,
|
| 9145 |
+
"grad_norm": 14851.4423828125,
|
| 9146 |
+
"learning_rate": 2.4004302956217804e-06,
|
| 9147 |
+
"loss": 18.9166,
|
| 9148 |
+
"step": 1301
|
| 9149 |
+
},
|
| 9150 |
+
{
|
| 9151 |
+
"epoch": 0.4434227330779055,
|
| 9152 |
+
"grad_norm": 9827.1484375,
|
| 9153 |
+
"learning_rate": 2.3513858950190204e-06,
|
| 9154 |
+
"loss": 16.1631,
|
| 9155 |
+
"step": 1302
|
| 9156 |
+
},
|
| 9157 |
+
{
|
| 9158 |
+
"epoch": 0.4437633035334185,
|
| 9159 |
+
"grad_norm": 36845.34375,
|
| 9160 |
+
"learning_rate": 2.302841743971995e-06,
|
| 9161 |
+
"loss": 31.0316,
|
| 9162 |
+
"step": 1303
|
| 9163 |
+
},
|
| 9164 |
+
{
|
| 9165 |
+
"epoch": 0.4441038739889315,
|
| 9166 |
+
"grad_norm": 18422.634765625,
|
| 9167 |
+
"learning_rate": 2.2547980911702404e-06,
|
| 9168 |
+
"loss": 20.1273,
|
| 9169 |
+
"step": 1304
|
| 9170 |
+
},
|
| 9171 |
+
{
|
| 9172 |
+
"epoch": 0.4444444444444444,
|
| 9173 |
+
"grad_norm": 4094.016357421875,
|
| 9174 |
+
"learning_rate": 2.2072551827392983e-06,
|
| 9175 |
+
"loss": 14.6568,
|
| 9176 |
+
"step": 1305
|
| 9177 |
+
},
|
| 9178 |
+
{
|
| 9179 |
+
"epoch": 0.4447850148999574,
|
| 9180 |
+
"grad_norm": 65638.265625,
|
| 9181 |
+
"learning_rate": 2.1602132622393746e-06,
|
| 9182 |
+
"loss": 28.6189,
|
| 9183 |
+
"step": 1306
|
| 9184 |
+
},
|
| 9185 |
+
{
|
| 9186 |
+
"epoch": 0.4451255853554704,
|
| 9187 |
+
"grad_norm": 6780.50927734375,
|
| 9188 |
+
"learning_rate": 2.1136725706641712e-06,
|
| 9189 |
+
"loss": 16.332,
|
| 9190 |
+
"step": 1307
|
| 9191 |
+
},
|
| 9192 |
+
{
|
| 9193 |
+
"epoch": 0.4454661558109834,
|
| 9194 |
+
"grad_norm": 7344.103515625,
|
| 9195 |
+
"learning_rate": 2.0676333464396126e-06,
|
| 9196 |
+
"loss": 18.0674,
|
| 9197 |
+
"step": 1308
|
| 9198 |
+
},
|
| 9199 |
+
{
|
| 9200 |
+
"epoch": 0.44580672626649637,
|
| 9201 |
+
"grad_norm": 8943.794921875,
|
| 9202 |
+
"learning_rate": 2.0220958254225984e-06,
|
| 9203 |
+
"loss": 16.6553,
|
| 9204 |
+
"step": 1309
|
| 9205 |
+
},
|
| 9206 |
+
{
|
| 9207 |
+
"epoch": 0.44614729672200937,
|
| 9208 |
+
"grad_norm": 7794.298828125,
|
| 9209 |
+
"learning_rate": 1.977060240899864e-06,
|
| 9210 |
+
"loss": 18.4036,
|
| 9211 |
+
"step": 1310
|
| 9212 |
+
},
|
| 9213 |
+
{
|
| 9214 |
+
"epoch": 0.44648786717752237,
|
| 9215 |
+
"grad_norm": 6309.8984375,
|
| 9216 |
+
"learning_rate": 1.932526823586722e-06,
|
| 9217 |
+
"loss": 14.4397,
|
| 9218 |
+
"step": 1311
|
| 9219 |
+
},
|
| 9220 |
+
{
|
| 9221 |
+
"epoch": 0.4468284376330353,
|
| 9222 |
+
"grad_norm": 59588.3359375,
|
| 9223 |
+
"learning_rate": 1.8884958016259113e-06,
|
| 9224 |
+
"loss": 21.6283,
|
| 9225 |
+
"step": 1312
|
| 9226 |
+
},
|
| 9227 |
+
{
|
| 9228 |
+
"epoch": 0.4471690080885483,
|
| 9229 |
+
"grad_norm": 99062.6953125,
|
| 9230 |
+
"learning_rate": 1.844967400586428e-06,
|
| 9231 |
+
"loss": 35.7518,
|
| 9232 |
+
"step": 1313
|
| 9233 |
+
},
|
| 9234 |
+
{
|
| 9235 |
+
"epoch": 0.4475095785440613,
|
| 9236 |
+
"grad_norm": 11542.525390625,
|
| 9237 |
+
"learning_rate": 1.8019418434623404e-06,
|
| 9238 |
+
"loss": 14.3759,
|
| 9239 |
+
"step": 1314
|
| 9240 |
+
},
|
| 9241 |
+
{
|
| 9242 |
+
"epoch": 0.44785014899957426,
|
| 9243 |
+
"grad_norm": 8130.87939453125,
|
| 9244 |
+
"learning_rate": 1.7594193506716983e-06,
|
| 9245 |
+
"loss": 14.8368,
|
| 9246 |
+
"step": 1315
|
| 9247 |
+
},
|
| 9248 |
+
{
|
| 9249 |
+
"epoch": 0.44819071945508726,
|
| 9250 |
+
"grad_norm": 7084.0517578125,
|
| 9251 |
+
"learning_rate": 1.7174001400553586e-06,
|
| 9252 |
+
"loss": 16.4896,
|
| 9253 |
+
"step": 1316
|
| 9254 |
+
},
|
| 9255 |
+
{
|
| 9256 |
+
"epoch": 0.44853128991060026,
|
| 9257 |
+
"grad_norm": 15228.7529296875,
|
| 9258 |
+
"learning_rate": 1.6758844268758843e-06,
|
| 9259 |
+
"loss": 29.1965,
|
| 9260 |
+
"step": 1317
|
| 9261 |
+
},
|
| 9262 |
+
{
|
| 9263 |
+
"epoch": 0.44887186036611326,
|
| 9264 |
+
"grad_norm": 14657.353515625,
|
| 9265 |
+
"learning_rate": 1.634872423816458e-06,
|
| 9266 |
+
"loss": 15.9394,
|
| 9267 |
+
"step": 1318
|
| 9268 |
+
},
|
| 9269 |
+
{
|
| 9270 |
+
"epoch": 0.4492124308216262,
|
| 9271 |
+
"grad_norm": 6427.775390625,
|
| 9272 |
+
"learning_rate": 1.5943643409797594e-06,
|
| 9273 |
+
"loss": 16.98,
|
| 9274 |
+
"step": 1319
|
| 9275 |
+
},
|
| 9276 |
+
{
|
| 9277 |
+
"epoch": 0.4495530012771392,
|
| 9278 |
+
"grad_norm": 7228.78857421875,
|
| 9279 |
+
"learning_rate": 1.5543603858869215e-06,
|
| 9280 |
+
"loss": 17.1118,
|
| 9281 |
+
"step": 1320
|
| 9282 |
+
},
|
| 9283 |
+
{
|
| 9284 |
+
"epoch": 0.4498935717326522,
|
| 9285 |
+
"grad_norm": 17582.966796875,
|
| 9286 |
+
"learning_rate": 1.5148607634764446e-06,
|
| 9287 |
+
"loss": 19.4363,
|
| 9288 |
+
"step": 1321
|
| 9289 |
+
},
|
| 9290 |
+
{
|
| 9291 |
+
"epoch": 0.45023414218816515,
|
| 9292 |
+
"grad_norm": 333400.9375,
|
| 9293 |
+
"learning_rate": 1.475865676103161e-06,
|
| 9294 |
+
"loss": 16.0002,
|
| 9295 |
+
"step": 1322
|
| 9296 |
+
},
|
| 9297 |
+
{
|
| 9298 |
+
"epoch": 0.45057471264367815,
|
| 9299 |
+
"grad_norm": 13839.4765625,
|
| 9300 |
+
"learning_rate": 1.4373753235371823e-06,
|
| 9301 |
+
"loss": 14.8031,
|
| 9302 |
+
"step": 1323
|
| 9303 |
+
},
|
| 9304 |
+
{
|
| 9305 |
+
"epoch": 0.45091528309919116,
|
| 9306 |
+
"grad_norm": 7787.62890625,
|
| 9307 |
+
"learning_rate": 1.3993899029628997e-06,
|
| 9308 |
+
"loss": 18.8672,
|
| 9309 |
+
"step": 1324
|
| 9310 |
+
},
|
| 9311 |
+
{
|
| 9312 |
+
"epoch": 0.45125585355470416,
|
| 9313 |
+
"grad_norm": 9136.2880859375,
|
| 9314 |
+
"learning_rate": 1.3619096089779293e-06,
|
| 9315 |
+
"loss": 15.5514,
|
| 9316 |
+
"step": 1325
|
| 9317 |
+
},
|
| 9318 |
+
{
|
| 9319 |
+
"epoch": 0.4515964240102171,
|
| 9320 |
+
"grad_norm": 8548.505859375,
|
| 9321 |
+
"learning_rate": 1.3249346335922007e-06,
|
| 9322 |
+
"loss": 18.1674,
|
| 9323 |
+
"step": 1326
|
| 9324 |
+
},
|
| 9325 |
+
{
|
| 9326 |
+
"epoch": 0.4519369944657301,
|
| 9327 |
+
"grad_norm": 6381.31689453125,
|
| 9328 |
+
"learning_rate": 1.2884651662268709e-06,
|
| 9329 |
+
"loss": 15.7302,
|
| 9330 |
+
"step": 1327
|
| 9331 |
+
},
|
| 9332 |
+
{
|
| 9333 |
+
"epoch": 0.4522775649212431,
|
| 9334 |
+
"grad_norm": 7535.04736328125,
|
| 9335 |
+
"learning_rate": 1.2525013937134122e-06,
|
| 9336 |
+
"loss": 16.0855,
|
| 9337 |
+
"step": 1328
|
| 9338 |
+
},
|
| 9339 |
+
{
|
| 9340 |
+
"epoch": 0.45261813537675605,
|
| 9341 |
+
"grad_norm": 4663.029296875,
|
| 9342 |
+
"learning_rate": 1.2170435002926694e-06,
|
| 9343 |
+
"loss": 14.7933,
|
| 9344 |
+
"step": 1329
|
| 9345 |
+
},
|
| 9346 |
+
{
|
| 9347 |
+
"epoch": 0.45295870583226905,
|
| 9348 |
+
"grad_norm": 7590.89453125,
|
| 9349 |
+
"learning_rate": 1.1820916676138382e-06,
|
| 9350 |
+
"loss": 18.698,
|
| 9351 |
+
"step": 1330
|
| 9352 |
+
},
|
| 9353 |
+
{
|
| 9354 |
+
"epoch": 0.45329927628778205,
|
| 9355 |
+
"grad_norm": 24820.05078125,
|
| 9356 |
+
"learning_rate": 1.147646074733655e-06,
|
| 9357 |
+
"loss": 20.8962,
|
| 9358 |
+
"step": 1331
|
| 9359 |
+
},
|
| 9360 |
+
{
|
| 9361 |
+
"epoch": 0.453639846743295,
|
| 9362 |
+
"grad_norm": 4408.1025390625,
|
| 9363 |
+
"learning_rate": 1.1137068981153632e-06,
|
| 9364 |
+
"loss": 14.9484,
|
| 9365 |
+
"step": 1332
|
| 9366 |
+
},
|
| 9367 |
+
{
|
| 9368 |
+
"epoch": 0.453980417198808,
|
| 9369 |
+
"grad_norm": 38103.05078125,
|
| 9370 |
+
"learning_rate": 1.0802743116278714e-06,
|
| 9371 |
+
"loss": 16.4813,
|
| 9372 |
+
"step": 1333
|
| 9373 |
+
},
|
| 9374 |
+
{
|
| 9375 |
+
"epoch": 0.454320987654321,
|
| 9376 |
+
"grad_norm": 8795.52734375,
|
| 9377 |
+
"learning_rate": 1.0473484865448525e-06,
|
| 9378 |
+
"loss": 15.2594,
|
| 9379 |
+
"step": 1334
|
| 9380 |
+
},
|
| 9381 |
+
{
|
| 9382 |
+
"epoch": 0.454661558109834,
|
| 9383 |
+
"grad_norm": 6825.61962890625,
|
| 9384 |
+
"learning_rate": 1.014929591543845e-06,
|
| 9385 |
+
"loss": 15.9325,
|
| 9386 |
+
"step": 1335
|
| 9387 |
+
},
|
| 9388 |
+
{
|
| 9389 |
+
"epoch": 0.45500212856534694,
|
| 9390 |
+
"grad_norm": 10258.9130859375,
|
| 9391 |
+
"learning_rate": 9.830177927054428e-07,
|
| 9392 |
+
"loss": 18.3511,
|
| 9393 |
+
"step": 1336
|
| 9394 |
+
},
|
| 9395 |
+
{
|
| 9396 |
+
"epoch": 0.45534269902085994,
|
| 9397 |
+
"grad_norm": 10838.5400390625,
|
| 9398 |
+
"learning_rate": 9.516132535123846e-07,
|
| 9399 |
+
"loss": 15.3048,
|
| 9400 |
+
"step": 1337
|
| 9401 |
+
},
|
| 9402 |
+
{
|
| 9403 |
+
"epoch": 0.45568326947637294,
|
| 9404 |
+
"grad_norm": 9795.0498046875,
|
| 9405 |
+
"learning_rate": 9.207161348487315e-07,
|
| 9406 |
+
"loss": 15.5718,
|
| 9407 |
+
"step": 1338
|
| 9408 |
+
},
|
| 9409 |
+
{
|
| 9410 |
+
"epoch": 0.4560238399318859,
|
| 9411 |
+
"grad_norm": 7020.3154296875,
|
| 9412 |
+
"learning_rate": 8.903265949990691e-07,
|
| 9413 |
+
"loss": 15.6469,
|
| 9414 |
+
"step": 1339
|
| 9415 |
+
},
|
| 9416 |
+
{
|
| 9417 |
+
"epoch": 0.4563644103873989,
|
| 9418 |
+
"grad_norm": 22910.3203125,
|
| 9419 |
+
"learning_rate": 8.604447896476852e-07,
|
| 9420 |
+
"loss": 16.3105,
|
| 9421 |
+
"step": 1340
|
| 9422 |
+
},
|
| 9423 |
+
{
|
| 9424 |
+
"epoch": 0.4567049808429119,
|
| 9425 |
+
"grad_norm": 7943.2607421875,
|
| 9426 |
+
"learning_rate": 8.310708718777371e-07,
|
| 9427 |
+
"loss": 20.3918,
|
| 9428 |
+
"step": 1341
|
| 9429 |
+
},
|
| 9430 |
+
{
|
| 9431 |
+
"epoch": 0.45704555129842483,
|
| 9432 |
+
"grad_norm": 12415.36328125,
|
| 9433 |
+
"learning_rate": 8.022049921705299e-07,
|
| 9434 |
+
"loss": 15.288,
|
| 9435 |
+
"step": 1342
|
| 9436 |
+
},
|
| 9437 |
+
{
|
| 9438 |
+
"epoch": 0.45738612175393784,
|
| 9439 |
+
"grad_norm": 5995.4189453125,
|
| 9440 |
+
"learning_rate": 7.73847298404684e-07,
|
| 9441 |
+
"loss": 16.8328,
|
| 9442 |
+
"step": 1343
|
| 9443 |
+
},
|
| 9444 |
+
{
|
| 9445 |
+
"epoch": 0.45772669220945084,
|
| 9446 |
+
"grad_norm": 6708.55615234375,
|
| 9447 |
+
"learning_rate": 7.459979358554248e-07,
|
| 9448 |
+
"loss": 17.2592,
|
| 9449 |
+
"step": 1344
|
| 9450 |
+
},
|
| 9451 |
+
{
|
| 9452 |
+
"epoch": 0.45806726266496384,
|
| 9453 |
+
"grad_norm": 136200.859375,
|
| 9454 |
+
"learning_rate": 7.186570471937937e-07,
|
| 9455 |
+
"loss": 28.2067,
|
| 9456 |
+
"step": 1345
|
| 9457 |
+
},
|
| 9458 |
+
{
|
| 9459 |
+
"epoch": 0.4584078331204768,
|
| 9460 |
+
"grad_norm": 6124.50341796875,
|
| 9461 |
+
"learning_rate": 6.918247724859939e-07,
|
| 9462 |
+
"loss": 16.6268,
|
| 9463 |
+
"step": 1346
|
| 9464 |
+
},
|
| 9465 |
+
{
|
| 9466 |
+
"epoch": 0.4587484035759898,
|
| 9467 |
+
"grad_norm": 5020.45556640625,
|
| 9468 |
+
"learning_rate": 6.655012491925683e-07,
|
| 9469 |
+
"loss": 16.8696,
|
| 9470 |
+
"step": 1347
|
| 9471 |
+
},
|
| 9472 |
+
{
|
| 9473 |
+
"epoch": 0.4590889740315028,
|
| 9474 |
+
"grad_norm": 15443.595703125,
|
| 9475 |
+
"learning_rate": 6.396866121677559e-07,
|
| 9476 |
+
"loss": 22.0466,
|
| 9477 |
+
"step": 1348
|
| 9478 |
+
},
|
| 9479 |
+
{
|
| 9480 |
+
"epoch": 0.45942954448701573,
|
| 9481 |
+
"grad_norm": 6612.12939453125,
|
| 9482 |
+
"learning_rate": 6.143809936588363e-07,
|
| 9483 |
+
"loss": 16.9246,
|
| 9484 |
+
"step": 1349
|
| 9485 |
+
},
|
| 9486 |
+
{
|
| 9487 |
+
"epoch": 0.45977011494252873,
|
| 9488 |
+
"grad_norm": 7099.29248046875,
|
| 9489 |
+
"learning_rate": 5.895845233053643e-07,
|
| 9490 |
+
"loss": 17.0276,
|
| 9491 |
+
"step": 1350
|
| 9492 |
+
},
|
| 9493 |
+
{
|
| 9494 |
+
"epoch": 0.46011068539804173,
|
| 9495 |
+
"grad_norm": 12668.201171875,
|
| 9496 |
+
"learning_rate": 5.652973281385588e-07,
|
| 9497 |
+
"loss": 18.9227,
|
| 9498 |
+
"step": 1351
|
| 9499 |
+
},
|
| 9500 |
+
{
|
| 9501 |
+
"epoch": 0.46045125585355473,
|
| 9502 |
+
"grad_norm": 7605.4775390625,
|
| 9503 |
+
"learning_rate": 5.415195325806699e-07,
|
| 9504 |
+
"loss": 16.7068,
|
| 9505 |
+
"step": 1352
|
| 9506 |
+
},
|
| 9507 |
+
{
|
| 9508 |
+
"epoch": 0.4607918263090677,
|
| 9509 |
+
"grad_norm": 304335.15625,
|
| 9510 |
+
"learning_rate": 5.182512584443022e-07,
|
| 9511 |
+
"loss": 22.2657,
|
| 9512 |
+
"step": 1353
|
| 9513 |
+
},
|
| 9514 |
+
{
|
| 9515 |
+
"epoch": 0.4611323967645807,
|
| 9516 |
+
"grad_norm": 71510.953125,
|
| 9517 |
+
"learning_rate": 4.954926249317815e-07,
|
| 9518 |
+
"loss": 20.3036,
|
| 9519 |
+
"step": 1354
|
| 9520 |
+
},
|
| 9521 |
+
{
|
| 9522 |
+
"epoch": 0.4614729672200937,
|
| 9523 |
+
"grad_norm": 4975.6982421875,
|
| 9524 |
+
"learning_rate": 4.732437486345886e-07,
|
| 9525 |
+
"loss": 18.521,
|
| 9526 |
+
"step": 1355
|
| 9527 |
+
},
|
| 9528 |
+
{
|
| 9529 |
+
"epoch": 0.4618135376756066,
|
| 9530 |
+
"grad_norm": 40805.234375,
|
| 9531 |
+
"learning_rate": 4.515047435327491e-07,
|
| 9532 |
+
"loss": 22.1671,
|
| 9533 |
+
"step": 1356
|
| 9534 |
+
},
|
| 9535 |
+
{
|
| 9536 |
+
"epoch": 0.4621541081311196,
|
| 9537 |
+
"grad_norm": 37550.95703125,
|
| 9538 |
+
"learning_rate": 4.3027572099422207e-07,
|
| 9539 |
+
"loss": 17.253,
|
| 9540 |
+
"step": 1357
|
| 9541 |
+
},
|
| 9542 |
+
{
|
| 9543 |
+
"epoch": 0.4624946785866326,
|
| 9544 |
+
"grad_norm": 6226.64990234375,
|
| 9545 |
+
"learning_rate": 4.0955678977436797e-07,
|
| 9546 |
+
"loss": 17.8961,
|
| 9547 |
+
"step": 1358
|
| 9548 |
+
},
|
| 9549 |
+
{
|
| 9550 |
+
"epoch": 0.46283524904214557,
|
| 9551 |
+
"grad_norm": 5531.28173828125,
|
| 9552 |
+
"learning_rate": 3.893480560153484e-07,
|
| 9553 |
+
"loss": 18.2577,
|
| 9554 |
+
"step": 1359
|
| 9555 |
+
},
|
| 9556 |
+
{
|
| 9557 |
+
"epoch": 0.46317581949765857,
|
| 9558 |
+
"grad_norm": 7954.1630859375,
|
| 9559 |
+
"learning_rate": 3.6964962324561593e-07,
|
| 9560 |
+
"loss": 17.3099,
|
| 9561 |
+
"step": 1360
|
| 9562 |
+
},
|
| 9563 |
+
{
|
| 9564 |
+
"epoch": 0.46351638995317157,
|
| 9565 |
+
"grad_norm": 272113.9375,
|
| 9566 |
+
"learning_rate": 3.504615923793919e-07,
|
| 9567 |
+
"loss": 29.326,
|
| 9568 |
+
"step": 1361
|
| 9569 |
+
},
|
| 9570 |
+
{
|
| 9571 |
+
"epoch": 0.46385696040868457,
|
| 9572 |
+
"grad_norm": 14789.712890625,
|
| 9573 |
+
"learning_rate": 3.317840617160894e-07,
|
| 9574 |
+
"loss": 17.4461,
|
| 9575 |
+
"step": 1362
|
| 9576 |
+
},
|
| 9577 |
+
{
|
| 9578 |
+
"epoch": 0.4641975308641975,
|
| 9579 |
+
"grad_norm": 10860.7783203125,
|
| 9580 |
+
"learning_rate": 3.136171269399024e-07,
|
| 9581 |
+
"loss": 16.5729,
|
| 9582 |
+
"step": 1363
|
| 9583 |
+
},
|
| 9584 |
+
{
|
| 9585 |
+
"epoch": 0.4645381013197105,
|
| 9586 |
+
"grad_norm": 1709174.125,
|
| 9587 |
+
"learning_rate": 2.959608811192283e-07,
|
| 9588 |
+
"loss": 27.5593,
|
| 9589 |
+
"step": 1364
|
| 9590 |
+
},
|
| 9591 |
+
{
|
| 9592 |
+
"epoch": 0.4648786717752235,
|
| 9593 |
+
"grad_norm": 7082.95849609375,
|
| 9594 |
+
"learning_rate": 2.7881541470623494e-07,
|
| 9595 |
+
"loss": 17.7374,
|
| 9596 |
+
"step": 1365
|
| 9597 |
+
},
|
| 9598 |
+
{
|
| 9599 |
+
"epoch": 0.46521924223073646,
|
| 9600 |
+
"grad_norm": 9195.6611328125,
|
| 9601 |
+
"learning_rate": 2.6218081553638364e-07,
|
| 9602 |
+
"loss": 18.8189,
|
| 9603 |
+
"step": 1366
|
| 9604 |
+
},
|
| 9605 |
+
{
|
| 9606 |
+
"epoch": 0.46555981268624946,
|
| 9607 |
+
"grad_norm": 6326.1044921875,
|
| 9608 |
+
"learning_rate": 2.4605716882801776e-07,
|
| 9609 |
+
"loss": 16.6767,
|
| 9610 |
+
"step": 1367
|
| 9611 |
+
},
|
| 9612 |
+
{
|
| 9613 |
+
"epoch": 0.46590038314176246,
|
| 9614 |
+
"grad_norm": 5304.5693359375,
|
| 9615 |
+
"learning_rate": 2.3044455718185253e-07,
|
| 9616 |
+
"loss": 14.6522,
|
| 9617 |
+
"step": 1368
|
| 9618 |
+
},
|
| 9619 |
+
{
|
| 9620 |
+
"epoch": 0.4662409535972754,
|
| 9621 |
+
"grad_norm": 6700.490234375,
|
| 9622 |
+
"learning_rate": 2.153430605806195e-07,
|
| 9623 |
+
"loss": 16.9113,
|
| 9624 |
+
"step": 1369
|
| 9625 |
+
},
|
| 9626 |
+
{
|
| 9627 |
+
"epoch": 0.4665815240527884,
|
| 9628 |
+
"grad_norm": 8465.318359375,
|
| 9629 |
+
"learning_rate": 2.0075275638862246e-07,
|
| 9630 |
+
"loss": 14.5613,
|
| 9631 |
+
"step": 1370
|
| 9632 |
+
},
|
| 9633 |
+
{
|
| 9634 |
+
"epoch": 0.4669220945083014,
|
| 9635 |
+
"grad_norm": 19922.677734375,
|
| 9636 |
+
"learning_rate": 1.8667371935133792e-07,
|
| 9637 |
+
"loss": 14.3837,
|
| 9638 |
+
"step": 1371
|
| 9639 |
+
},
|
| 9640 |
+
{
|
| 9641 |
+
"epoch": 0.4672626649638144,
|
| 9642 |
+
"grad_norm": 13750.6982421875,
|
| 9643 |
+
"learning_rate": 1.7310602159505974e-07,
|
| 9644 |
+
"loss": 17.0011,
|
| 9645 |
+
"step": 1372
|
| 9646 |
+
},
|
| 9647 |
+
{
|
| 9648 |
+
"epoch": 0.46760323541932736,
|
| 9649 |
+
"grad_norm": 42446.8125,
|
| 9650 |
+
"learning_rate": 1.6004973262651047e-07,
|
| 9651 |
+
"loss": 21.8664,
|
| 9652 |
+
"step": 1373
|
| 9653 |
+
},
|
| 9654 |
+
{
|
| 9655 |
+
"epoch": 0.46794380587484036,
|
| 9656 |
+
"grad_norm": 6920.4345703125,
|
| 9657 |
+
"learning_rate": 1.4750491933247512e-07,
|
| 9658 |
+
"loss": 16.6084,
|
| 9659 |
+
"step": 1374
|
| 9660 |
+
},
|
| 9661 |
+
{
|
| 9662 |
+
"epoch": 0.46828437633035336,
|
| 9663 |
+
"grad_norm": 5923.66259765625,
|
| 9664 |
+
"learning_rate": 1.3547164597949026e-07,
|
| 9665 |
+
"loss": 16.8637,
|
| 9666 |
+
"step": 1375
|
| 9667 |
+
},
|
| 9668 |
+
{
|
| 9669 |
+
"epoch": 0.4686249467858663,
|
| 9670 |
+
"grad_norm": 40832.45703125,
|
| 9671 |
+
"learning_rate": 1.2394997421347753e-07,
|
| 9672 |
+
"loss": 20.7282,
|
| 9673 |
+
"step": 1376
|
| 9674 |
+
},
|
| 9675 |
+
{
|
| 9676 |
+
"epoch": 0.4689655172413793,
|
| 9677 |
+
"grad_norm": 77762.2734375,
|
| 9678 |
+
"learning_rate": 1.1293996305946631e-07,
|
| 9679 |
+
"loss": 22.4176,
|
| 9680 |
+
"step": 1377
|
| 9681 |
+
},
|
| 9682 |
+
{
|
| 9683 |
+
"epoch": 0.4693060876968923,
|
| 9684 |
+
"grad_norm": 7452.88720703125,
|
| 9685 |
+
"learning_rate": 1.0244166892124928e-07,
|
| 9686 |
+
"loss": 18.7866,
|
| 9687 |
+
"step": 1378
|
| 9688 |
+
},
|
| 9689 |
+
{
|
| 9690 |
+
"epoch": 0.4696466581524053,
|
| 9691 |
+
"grad_norm": 12359.66015625,
|
| 9692 |
+
"learning_rate": 9.245514558112733e-08,
|
| 9693 |
+
"loss": 15.8441,
|
| 9694 |
+
"step": 1379
|
| 9695 |
+
},
|
| 9696 |
+
{
|
| 9697 |
+
"epoch": 0.46998722860791825,
|
| 9698 |
+
"grad_norm": 13811.025390625,
|
| 9699 |
+
"learning_rate": 8.298044419962069e-08,
|
| 9700 |
+
"loss": 19.9129,
|
| 9701 |
+
"step": 1380
|
| 9702 |
+
},
|
| 9703 |
+
{
|
| 9704 |
+
"epoch": 0.47032779906343125,
|
| 9705 |
+
"grad_norm": 14943.3193359375,
|
| 9706 |
+
"learning_rate": 7.401761331521372e-08,
|
| 9707 |
+
"loss": 24.8472,
|
| 9708 |
+
"step": 1381
|
| 9709 |
+
},
|
| 9710 |
+
{
|
| 9711 |
+
"epoch": 0.47066836951894425,
|
| 9712 |
+
"grad_norm": 131950.921875,
|
| 9713 |
+
"learning_rate": 6.556669884408839e-08,
|
| 9714 |
+
"loss": 19.8326,
|
| 9715 |
+
"step": 1382
|
| 9716 |
+
},
|
| 9717 |
+
{
|
| 9718 |
+
"epoch": 0.4710089399744572,
|
| 9719 |
+
"grad_norm": 5667.35205078125,
|
| 9720 |
+
"learning_rate": 5.7627744079902235e-08,
|
| 9721 |
+
"loss": 17.4678,
|
| 9722 |
+
"step": 1383
|
| 9723 |
+
},
|
| 9724 |
+
{
|
| 9725 |
+
"epoch": 0.4713495104299702,
|
| 9726 |
+
"grad_norm": 6688.39990234375,
|
| 9727 |
+
"learning_rate": 5.0200789693588544e-08,
|
| 9728 |
+
"loss": 18.0388,
|
| 9729 |
+
"step": 1384
|
| 9730 |
+
},
|
| 9731 |
+
{
|
| 9732 |
+
"epoch": 0.4716900808854832,
|
| 9733 |
+
"grad_norm": 285899.21875,
|
| 9734 |
+
"learning_rate": 4.32858737330899e-08,
|
| 9735 |
+
"loss": 25.0115,
|
| 9736 |
+
"step": 1385
|
| 9737 |
+
},
|
| 9738 |
+
{
|
| 9739 |
+
"epoch": 0.47203065134099614,
|
| 9740 |
+
"grad_norm": 8138.41259765625,
|
| 9741 |
+
"learning_rate": 3.6883031623224926e-08,
|
| 9742 |
+
"loss": 16.6204,
|
| 9743 |
+
"step": 1386
|
| 9744 |
+
},
|
| 9745 |
+
{
|
| 9746 |
+
"epoch": 0.47237122179650914,
|
| 9747 |
+
"grad_norm": 13012.6865234375,
|
| 9748 |
+
"learning_rate": 3.099229616547739e-08,
|
| 9749 |
+
"loss": 17.1591,
|
| 9750 |
+
"step": 1387
|
| 9751 |
+
},
|
| 9752 |
+
{
|
| 9753 |
+
"epoch": 0.47271179225202214,
|
| 9754 |
+
"grad_norm": 15416.2587890625,
|
| 9755 |
+
"learning_rate": 2.5613697537818504e-08,
|
| 9756 |
+
"loss": 15.8793,
|
| 9757 |
+
"step": 1388
|
| 9758 |
+
},
|
| 9759 |
+
{
|
| 9760 |
+
"epoch": 0.47305236270753515,
|
| 9761 |
+
"grad_norm": 7441.67529296875,
|
| 9762 |
+
"learning_rate": 2.074726329457377e-08,
|
| 9763 |
+
"loss": 17.4983,
|
| 9764 |
+
"step": 1389
|
| 9765 |
+
},
|
| 9766 |
+
{
|
| 9767 |
+
"epoch": 0.4733929331630481,
|
| 9768 |
+
"grad_norm": 12869.0673828125,
|
| 9769 |
+
"learning_rate": 1.6393018366278602e-08,
|
| 9770 |
+
"loss": 22.811,
|
| 9771 |
+
"step": 1390
|
| 9772 |
+
},
|
| 9773 |
+
{
|
| 9774 |
+
"epoch": 0.4737335036185611,
|
| 9775 |
+
"grad_norm": 9769.373046875,
|
| 9776 |
+
"learning_rate": 1.2550985059522902e-08,
|
| 9777 |
+
"loss": 15.9837,
|
| 9778 |
+
"step": 1391
|
| 9779 |
+
},
|
| 9780 |
+
{
|
| 9781 |
+
"epoch": 0.4740740740740741,
|
| 9782 |
+
"grad_norm": 4563.84619140625,
|
| 9783 |
+
"learning_rate": 9.221183056895566e-09,
|
| 9784 |
+
"loss": 15.6628,
|
| 9785 |
+
"step": 1392
|
| 9786 |
+
},
|
| 9787 |
+
{
|
| 9788 |
+
"epoch": 0.47441464452958704,
|
| 9789 |
+
"grad_norm": 256974.359375,
|
| 9790 |
+
"learning_rate": 6.4036294168068335e-09,
|
| 9791 |
+
"loss": 21.7092,
|
| 9792 |
+
"step": 1393
|
| 9793 |
+
},
|
| 9794 |
+
{
|
| 9795 |
+
"epoch": 0.47475521498510004,
|
| 9796 |
+
"grad_norm": 11716.580078125,
|
| 9797 |
+
"learning_rate": 4.0983385734660875e-09,
|
| 9798 |
+
"loss": 17.6616,
|
| 9799 |
+
"step": 1394
|
| 9800 |
+
},
|
| 9801 |
+
{
|
| 9802 |
+
"epoch": 0.47509578544061304,
|
| 9803 |
+
"grad_norm": 7115.24267578125,
|
| 9804 |
+
"learning_rate": 2.305322336781934e-09,
|
| 9805 |
+
"loss": 16.7245,
|
| 9806 |
+
"step": 1395
|
| 9807 |
+
},
|
| 9808 |
+
{
|
| 9809 |
+
"epoch": 0.475436355896126,
|
| 9810 |
+
"grad_norm": 6537.548828125,
|
| 9811 |
+
"learning_rate": 1.0245898922844888e-09,
|
| 9812 |
+
"loss": 16.2782,
|
| 9813 |
+
"step": 1396
|
| 9814 |
+
},
|
| 9815 |
+
{
|
| 9816 |
+
"epoch": 0.475776926351639,
|
| 9817 |
+
"grad_norm": 49136.9609375,
|
| 9818 |
+
"learning_rate": 2.561478011253726e-10,
|
| 9819 |
+
"loss": 15.997,
|
| 9820 |
+
"step": 1397
|
| 9821 |
+
},
|
| 9822 |
+
{
|
| 9823 |
+
"epoch": 0.476117496807152,
|
| 9824 |
+
"grad_norm": 4063.356201171875,
|
| 9825 |
+
"learning_rate": 0.0,
|
| 9826 |
+
"loss": 17.1553,
|
| 9827 |
+
"step": 1398
|
| 9828 |
}
|
| 9829 |
],
|
| 9830 |
"logging_steps": 1,
|
|
|
|
| 9839 |
"should_evaluate": false,
|
| 9840 |
"should_log": false,
|
| 9841 |
"should_save": true,
|
| 9842 |
+
"should_training_stop": true
|
| 9843 |
},
|
| 9844 |
"attributes": {}
|
| 9845 |
}
|
| 9846 |
},
|
| 9847 |
+
"total_flos": 495547462975488.0,
|
| 9848 |
"train_batch_size": 2,
|
| 9849 |
"trial_name": null,
|
| 9850 |
"trial_params": null
|