Training in progress, step 652, 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 5752
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:a4dc3ffbc70d72e8a6f8cd9c6b32117c67f4c69e2f655b5f9efc0e29231d04cb
|
3 |
size 5752
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 15814
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:b07138d8a35daad2023f83ed7649f70360d21a19247f3c171cc20bc72059ecd7
|
3 |
size 15814
|
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:f3fdef36d1efa4aaf4a4dce89dd2084ac7e63b444a9b3a9fa2fb0726eae0c3be
|
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 sha256:6d4177d19f0947a00edc20f3dcb0c32c4727e92e98a9515fe706eb662b354864
|
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": 326,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -2305,6 +2305,2296 @@
|
|
2305 |
"eval_samples_per_second": 166.07,
|
2306 |
"eval_steps_per_second": 83.035,
|
2307 |
"step": 326
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
2308 |
}
|
2309 |
],
|
2310 |
"logging_steps": 1,
|
@@ -2324,7 +4614,7 @@
|
|
2324 |
"attributes": {}
|
2325 |
}
|
2326 |
},
|
2327 |
-
"total_flos":
|
2328 |
"train_batch_size": 2,
|
2329 |
"trial_name": null,
|
2330 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.013124389826584941,
|
5 |
"eval_steps": 326,
|
6 |
+
"global_step": 652,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
2305 |
"eval_samples_per_second": 166.07,
|
2306 |
"eval_steps_per_second": 83.035,
|
2307 |
"step": 326
|
2308 |
+
},
|
2309 |
+
{
|
2310 |
+
"epoch": 0.006582324345541834,
|
2311 |
+
"grad_norm": 7.556305354228243e-05,
|
2312 |
+
"learning_rate": 0.00017177626661356884,
|
2313 |
+
"loss": 46.0,
|
2314 |
+
"step": 327
|
2315 |
+
},
|
2316 |
+
{
|
2317 |
+
"epoch": 0.006602453777791197,
|
2318 |
+
"grad_norm": 7.451939745806158e-05,
|
2319 |
+
"learning_rate": 0.0001716068783702729,
|
2320 |
+
"loss": 46.0,
|
2321 |
+
"step": 328
|
2322 |
+
},
|
2323 |
+
{
|
2324 |
+
"epoch": 0.006622583210040561,
|
2325 |
+
"grad_norm": 9.252296149497852e-05,
|
2326 |
+
"learning_rate": 0.00017143706740266733,
|
2327 |
+
"loss": 46.0,
|
2328 |
+
"step": 329
|
2329 |
+
},
|
2330 |
+
{
|
2331 |
+
"epoch": 0.006642712642289925,
|
2332 |
+
"grad_norm": 7.886350795160979e-05,
|
2333 |
+
"learning_rate": 0.00017126683471321494,
|
2334 |
+
"loss": 46.0,
|
2335 |
+
"step": 330
|
2336 |
+
},
|
2337 |
+
{
|
2338 |
+
"epoch": 0.0066628420745392875,
|
2339 |
+
"grad_norm": 7.060460484353825e-05,
|
2340 |
+
"learning_rate": 0.00017109618130686793,
|
2341 |
+
"loss": 46.0,
|
2342 |
+
"step": 331
|
2343 |
+
},
|
2344 |
+
{
|
2345 |
+
"epoch": 0.006682971506788651,
|
2346 |
+
"grad_norm": 9.136456355918199e-05,
|
2347 |
+
"learning_rate": 0.00017092510819106228,
|
2348 |
+
"loss": 46.0,
|
2349 |
+
"step": 332
|
2350 |
+
},
|
2351 |
+
{
|
2352 |
+
"epoch": 0.006703100939038014,
|
2353 |
+
"grad_norm": 0.00014541424752678722,
|
2354 |
+
"learning_rate": 0.00017075361637571164,
|
2355 |
+
"loss": 46.0,
|
2356 |
+
"step": 333
|
2357 |
+
},
|
2358 |
+
{
|
2359 |
+
"epoch": 0.006723230371287378,
|
2360 |
+
"grad_norm": 5.236260767560452e-05,
|
2361 |
+
"learning_rate": 0.00017058170687320144,
|
2362 |
+
"loss": 46.0,
|
2363 |
+
"step": 334
|
2364 |
+
},
|
2365 |
+
{
|
2366 |
+
"epoch": 0.0067433598035367415,
|
2367 |
+
"grad_norm": 3.249632572988048e-05,
|
2368 |
+
"learning_rate": 0.00017040938069838284,
|
2369 |
+
"loss": 46.0,
|
2370 |
+
"step": 335
|
2371 |
+
},
|
2372 |
+
{
|
2373 |
+
"epoch": 0.006763489235786104,
|
2374 |
+
"grad_norm": 0.00012241276272106916,
|
2375 |
+
"learning_rate": 0.00017023663886856681,
|
2376 |
+
"loss": 46.0,
|
2377 |
+
"step": 336
|
2378 |
+
},
|
2379 |
+
{
|
2380 |
+
"epoch": 0.006783618668035468,
|
2381 |
+
"grad_norm": 7.044512312859297e-05,
|
2382 |
+
"learning_rate": 0.0001700634824035182,
|
2383 |
+
"loss": 46.0,
|
2384 |
+
"step": 337
|
2385 |
+
},
|
2386 |
+
{
|
2387 |
+
"epoch": 0.006803748100284832,
|
2388 |
+
"grad_norm": 3.4949163818964735e-05,
|
2389 |
+
"learning_rate": 0.00016988991232544943,
|
2390 |
+
"loss": 46.0,
|
2391 |
+
"step": 338
|
2392 |
+
},
|
2393 |
+
{
|
2394 |
+
"epoch": 0.006823877532534195,
|
2395 |
+
"grad_norm": 5.8312176406616345e-05,
|
2396 |
+
"learning_rate": 0.00016971592965901472,
|
2397 |
+
"loss": 46.0,
|
2398 |
+
"step": 339
|
2399 |
+
},
|
2400 |
+
{
|
2401 |
+
"epoch": 0.006844006964783558,
|
2402 |
+
"grad_norm": 6.132174894446507e-05,
|
2403 |
+
"learning_rate": 0.00016954153543130405,
|
2404 |
+
"loss": 46.0,
|
2405 |
+
"step": 340
|
2406 |
+
},
|
2407 |
+
{
|
2408 |
+
"epoch": 0.006864136397032922,
|
2409 |
+
"grad_norm": 7.451802957803011e-05,
|
2410 |
+
"learning_rate": 0.00016936673067183695,
|
2411 |
+
"loss": 46.0,
|
2412 |
+
"step": 341
|
2413 |
+
},
|
2414 |
+
{
|
2415 |
+
"epoch": 0.006884265829282285,
|
2416 |
+
"grad_norm": 4.202104901196435e-05,
|
2417 |
+
"learning_rate": 0.00016919151641255642,
|
2418 |
+
"loss": 46.0,
|
2419 |
+
"step": 342
|
2420 |
+
},
|
2421 |
+
{
|
2422 |
+
"epoch": 0.006904395261531649,
|
2423 |
+
"grad_norm": 5.8602174249244854e-05,
|
2424 |
+
"learning_rate": 0.00016901589368782303,
|
2425 |
+
"loss": 46.0,
|
2426 |
+
"step": 343
|
2427 |
+
},
|
2428 |
+
{
|
2429 |
+
"epoch": 0.006924524693781012,
|
2430 |
+
"grad_norm": 6.12097283010371e-05,
|
2431 |
+
"learning_rate": 0.00016883986353440856,
|
2432 |
+
"loss": 46.0,
|
2433 |
+
"step": 344
|
2434 |
+
},
|
2435 |
+
{
|
2436 |
+
"epoch": 0.006944654126030375,
|
2437 |
+
"grad_norm": 7.694535452174023e-05,
|
2438 |
+
"learning_rate": 0.0001686634269914901,
|
2439 |
+
"loss": 46.0,
|
2440 |
+
"step": 345
|
2441 |
+
},
|
2442 |
+
{
|
2443 |
+
"epoch": 0.006964783558279739,
|
2444 |
+
"grad_norm": 4.796484427060932e-05,
|
2445 |
+
"learning_rate": 0.00016848658510064377,
|
2446 |
+
"loss": 46.0,
|
2447 |
+
"step": 346
|
2448 |
+
},
|
2449 |
+
{
|
2450 |
+
"epoch": 0.0069849129905291026,
|
2451 |
+
"grad_norm": 3.562847996363416e-05,
|
2452 |
+
"learning_rate": 0.00016830933890583865,
|
2453 |
+
"loss": 46.0,
|
2454 |
+
"step": 347
|
2455 |
+
},
|
2456 |
+
{
|
2457 |
+
"epoch": 0.007005042422778465,
|
2458 |
+
"grad_norm": 5.831445741932839e-05,
|
2459 |
+
"learning_rate": 0.00016813168945343062,
|
2460 |
+
"loss": 46.0,
|
2461 |
+
"step": 348
|
2462 |
+
},
|
2463 |
+
{
|
2464 |
+
"epoch": 0.007025171855027829,
|
2465 |
+
"grad_norm": 5.701630288967863e-05,
|
2466 |
+
"learning_rate": 0.000167953637792156,
|
2467 |
+
"loss": 46.0,
|
2468 |
+
"step": 349
|
2469 |
+
},
|
2470 |
+
{
|
2471 |
+
"epoch": 0.007045301287277192,
|
2472 |
+
"grad_norm": 6.599428888875991e-05,
|
2473 |
+
"learning_rate": 0.00016777518497312576,
|
2474 |
+
"loss": 46.0,
|
2475 |
+
"step": 350
|
2476 |
+
},
|
2477 |
+
{
|
2478 |
+
"epoch": 0.007065430719526556,
|
2479 |
+
"grad_norm": 7.63636635383591e-05,
|
2480 |
+
"learning_rate": 0.00016759633204981885,
|
2481 |
+
"loss": 46.0,
|
2482 |
+
"step": 351
|
2483 |
+
},
|
2484 |
+
{
|
2485 |
+
"epoch": 0.007085560151775919,
|
2486 |
+
"grad_norm": 5.485086148837581e-05,
|
2487 |
+
"learning_rate": 0.00016741708007807625,
|
2488 |
+
"loss": 46.0,
|
2489 |
+
"step": 352
|
2490 |
+
},
|
2491 |
+
{
|
2492 |
+
"epoch": 0.007105689584025282,
|
2493 |
+
"grad_norm": 8.1733385741245e-05,
|
2494 |
+
"learning_rate": 0.0001672374301160948,
|
2495 |
+
"loss": 46.0,
|
2496 |
+
"step": 353
|
2497 |
+
},
|
2498 |
+
{
|
2499 |
+
"epoch": 0.007125819016274646,
|
2500 |
+
"grad_norm": 7.260946586029604e-05,
|
2501 |
+
"learning_rate": 0.00016705738322442067,
|
2502 |
+
"loss": 46.0,
|
2503 |
+
"step": 354
|
2504 |
+
},
|
2505 |
+
{
|
2506 |
+
"epoch": 0.00714594844852401,
|
2507 |
+
"grad_norm": 0.00012448117195162922,
|
2508 |
+
"learning_rate": 0.0001668769404659434,
|
2509 |
+
"loss": 46.0,
|
2510 |
+
"step": 355
|
2511 |
+
},
|
2512 |
+
{
|
2513 |
+
"epoch": 0.0071660778807733725,
|
2514 |
+
"grad_norm": 0.00018412985082250088,
|
2515 |
+
"learning_rate": 0.00016669610290588938,
|
2516 |
+
"loss": 46.0,
|
2517 |
+
"step": 356
|
2518 |
+
},
|
2519 |
+
{
|
2520 |
+
"epoch": 0.007186207313022736,
|
2521 |
+
"grad_norm": 0.00010343602480133995,
|
2522 |
+
"learning_rate": 0.00016651487161181575,
|
2523 |
+
"loss": 46.0,
|
2524 |
+
"step": 357
|
2525 |
+
},
|
2526 |
+
{
|
2527 |
+
"epoch": 0.0072063367452721,
|
2528 |
+
"grad_norm": 7.606980216223747e-05,
|
2529 |
+
"learning_rate": 0.00016633324765360404,
|
2530 |
+
"loss": 46.0,
|
2531 |
+
"step": 358
|
2532 |
+
},
|
2533 |
+
{
|
2534 |
+
"epoch": 0.007226466177521463,
|
2535 |
+
"grad_norm": 4.61212002846878e-05,
|
2536 |
+
"learning_rate": 0.00016615123210345374,
|
2537 |
+
"loss": 46.0,
|
2538 |
+
"step": 359
|
2539 |
+
},
|
2540 |
+
{
|
2541 |
+
"epoch": 0.0072465956097708265,
|
2542 |
+
"grad_norm": 3.344099968671799e-05,
|
2543 |
+
"learning_rate": 0.00016596882603587613,
|
2544 |
+
"loss": 46.0,
|
2545 |
+
"step": 360
|
2546 |
+
},
|
2547 |
+
{
|
2548 |
+
"epoch": 0.00726672504202019,
|
2549 |
+
"grad_norm": 6.0584614402614534e-05,
|
2550 |
+
"learning_rate": 0.00016578603052768787,
|
2551 |
+
"loss": 46.0,
|
2552 |
+
"step": 361
|
2553 |
+
},
|
2554 |
+
{
|
2555 |
+
"epoch": 0.007286854474269553,
|
2556 |
+
"grad_norm": 0.00010081299842568114,
|
2557 |
+
"learning_rate": 0.00016560284665800463,
|
2558 |
+
"loss": 46.0,
|
2559 |
+
"step": 362
|
2560 |
+
},
|
2561 |
+
{
|
2562 |
+
"epoch": 0.007306983906518917,
|
2563 |
+
"grad_norm": 0.00018100201850757003,
|
2564 |
+
"learning_rate": 0.00016541927550823475,
|
2565 |
+
"loss": 46.0,
|
2566 |
+
"step": 363
|
2567 |
+
},
|
2568 |
+
{
|
2569 |
+
"epoch": 0.0073271133387682805,
|
2570 |
+
"grad_norm": 6.115916767157614e-05,
|
2571 |
+
"learning_rate": 0.00016523531816207285,
|
2572 |
+
"loss": 46.0,
|
2573 |
+
"step": 364
|
2574 |
+
},
|
2575 |
+
{
|
2576 |
+
"epoch": 0.007347242771017643,
|
2577 |
+
"grad_norm": 0.0001454094162909314,
|
2578 |
+
"learning_rate": 0.00016505097570549334,
|
2579 |
+
"loss": 46.0,
|
2580 |
+
"step": 365
|
2581 |
+
},
|
2582 |
+
{
|
2583 |
+
"epoch": 0.007367372203267007,
|
2584 |
+
"grad_norm": 7.579627708764747e-05,
|
2585 |
+
"learning_rate": 0.00016486624922674423,
|
2586 |
+
"loss": 46.0,
|
2587 |
+
"step": 366
|
2588 |
+
},
|
2589 |
+
{
|
2590 |
+
"epoch": 0.00738750163551637,
|
2591 |
+
"grad_norm": 5.6147979194065556e-05,
|
2592 |
+
"learning_rate": 0.0001646811398163405,
|
2593 |
+
"loss": 46.0,
|
2594 |
+
"step": 367
|
2595 |
+
},
|
2596 |
+
{
|
2597 |
+
"epoch": 0.007407631067765734,
|
2598 |
+
"grad_norm": 2.967609543702565e-05,
|
2599 |
+
"learning_rate": 0.00016449564856705763,
|
2600 |
+
"loss": 46.0,
|
2601 |
+
"step": 368
|
2602 |
+
},
|
2603 |
+
{
|
2604 |
+
"epoch": 0.007427760500015097,
|
2605 |
+
"grad_norm": 4.6177185140550137e-05,
|
2606 |
+
"learning_rate": 0.00016430977657392543,
|
2607 |
+
"loss": 46.0,
|
2608 |
+
"step": 369
|
2609 |
+
},
|
2610 |
+
{
|
2611 |
+
"epoch": 0.00744788993226446,
|
2612 |
+
"grad_norm": 6.124021456344053e-05,
|
2613 |
+
"learning_rate": 0.00016412352493422132,
|
2614 |
+
"loss": 46.0,
|
2615 |
+
"step": 370
|
2616 |
+
},
|
2617 |
+
{
|
2618 |
+
"epoch": 0.007468019364513824,
|
2619 |
+
"grad_norm": 6.325580761767924e-05,
|
2620 |
+
"learning_rate": 0.00016393689474746383,
|
2621 |
+
"loss": 46.0,
|
2622 |
+
"step": 371
|
2623 |
+
},
|
2624 |
+
{
|
2625 |
+
"epoch": 0.007488148796763188,
|
2626 |
+
"grad_norm": 4.645885928766802e-05,
|
2627 |
+
"learning_rate": 0.00016374988711540634,
|
2628 |
+
"loss": 46.0,
|
2629 |
+
"step": 372
|
2630 |
+
},
|
2631 |
+
{
|
2632 |
+
"epoch": 0.00750827822901255,
|
2633 |
+
"grad_norm": 6.079759623389691e-05,
|
2634 |
+
"learning_rate": 0.00016356250314203044,
|
2635 |
+
"loss": 46.0,
|
2636 |
+
"step": 373
|
2637 |
+
},
|
2638 |
+
{
|
2639 |
+
"epoch": 0.007528407661261914,
|
2640 |
+
"grad_norm": 6.65177867631428e-05,
|
2641 |
+
"learning_rate": 0.00016337474393353932,
|
2642 |
+
"loss": 46.0,
|
2643 |
+
"step": 374
|
2644 |
+
},
|
2645 |
+
{
|
2646 |
+
"epoch": 0.007548537093511278,
|
2647 |
+
"grad_norm": 6.509361992357299e-05,
|
2648 |
+
"learning_rate": 0.00016318661059835135,
|
2649 |
+
"loss": 46.0,
|
2650 |
+
"step": 375
|
2651 |
+
},
|
2652 |
+
{
|
2653 |
+
"epoch": 0.007568666525760641,
|
2654 |
+
"grad_norm": 6.14354939898476e-05,
|
2655 |
+
"learning_rate": 0.0001629981042470936,
|
2656 |
+
"loss": 46.0,
|
2657 |
+
"step": 376
|
2658 |
+
},
|
2659 |
+
{
|
2660 |
+
"epoch": 0.007588795958010004,
|
2661 |
+
"grad_norm": 0.00011426959827076644,
|
2662 |
+
"learning_rate": 0.00016280922599259517,
|
2663 |
+
"loss": 46.0,
|
2664 |
+
"step": 377
|
2665 |
+
},
|
2666 |
+
{
|
2667 |
+
"epoch": 0.007608925390259368,
|
2668 |
+
"grad_norm": 5.5513559345854446e-05,
|
2669 |
+
"learning_rate": 0.00016261997694988064,
|
2670 |
+
"loss": 46.0,
|
2671 |
+
"step": 378
|
2672 |
+
},
|
2673 |
+
{
|
2674 |
+
"epoch": 0.007629054822508731,
|
2675 |
+
"grad_norm": 5.053329368820414e-05,
|
2676 |
+
"learning_rate": 0.00016243035823616347,
|
2677 |
+
"loss": 46.0,
|
2678 |
+
"step": 379
|
2679 |
+
},
|
2680 |
+
{
|
2681 |
+
"epoch": 0.007649184254758095,
|
2682 |
+
"grad_norm": 4.8892205086303875e-05,
|
2683 |
+
"learning_rate": 0.0001622403709708395,
|
2684 |
+
"loss": 46.0,
|
2685 |
+
"step": 380
|
2686 |
+
},
|
2687 |
+
{
|
2688 |
+
"epoch": 0.007669313687007458,
|
2689 |
+
"grad_norm": 6.474481051554903e-05,
|
2690 |
+
"learning_rate": 0.00016205001627548019,
|
2691 |
+
"loss": 46.0,
|
2692 |
+
"step": 381
|
2693 |
+
},
|
2694 |
+
{
|
2695 |
+
"epoch": 0.007689443119256821,
|
2696 |
+
"grad_norm": 6.014668906573206e-05,
|
2697 |
+
"learning_rate": 0.0001618592952738263,
|
2698 |
+
"loss": 46.0,
|
2699 |
+
"step": 382
|
2700 |
+
},
|
2701 |
+
{
|
2702 |
+
"epoch": 0.007709572551506185,
|
2703 |
+
"grad_norm": 7.018409087322652e-05,
|
2704 |
+
"learning_rate": 0.00016166820909178074,
|
2705 |
+
"loss": 46.0,
|
2706 |
+
"step": 383
|
2707 |
+
},
|
2708 |
+
{
|
2709 |
+
"epoch": 0.007729701983755548,
|
2710 |
+
"grad_norm": 0.00011436323984526098,
|
2711 |
+
"learning_rate": 0.00016147675885740242,
|
2712 |
+
"loss": 46.0,
|
2713 |
+
"step": 384
|
2714 |
+
},
|
2715 |
+
{
|
2716 |
+
"epoch": 0.0077498314160049115,
|
2717 |
+
"grad_norm": 7.127400749595836e-05,
|
2718 |
+
"learning_rate": 0.00016128494570089944,
|
2719 |
+
"loss": 46.0,
|
2720 |
+
"step": 385
|
2721 |
+
},
|
2722 |
+
{
|
2723 |
+
"epoch": 0.007769960848254275,
|
2724 |
+
"grad_norm": 8.55454636621289e-05,
|
2725 |
+
"learning_rate": 0.0001610927707546222,
|
2726 |
+
"loss": 46.0,
|
2727 |
+
"step": 386
|
2728 |
+
},
|
2729 |
+
{
|
2730 |
+
"epoch": 0.007790090280503638,
|
2731 |
+
"grad_norm": 6.892836972838268e-05,
|
2732 |
+
"learning_rate": 0.00016090023515305703,
|
2733 |
+
"loss": 46.0,
|
2734 |
+
"step": 387
|
2735 |
+
},
|
2736 |
+
{
|
2737 |
+
"epoch": 0.007810219712753002,
|
2738 |
+
"grad_norm": 0.0001008848994388245,
|
2739 |
+
"learning_rate": 0.0001607073400328193,
|
2740 |
+
"loss": 46.0,
|
2741 |
+
"step": 388
|
2742 |
+
},
|
2743 |
+
{
|
2744 |
+
"epoch": 0.007830349145002365,
|
2745 |
+
"grad_norm": 5.973876977805048e-05,
|
2746 |
+
"learning_rate": 0.00016051408653264675,
|
2747 |
+
"loss": 46.0,
|
2748 |
+
"step": 389
|
2749 |
+
},
|
2750 |
+
{
|
2751 |
+
"epoch": 0.007850478577251728,
|
2752 |
+
"grad_norm": 9.112283441936597e-05,
|
2753 |
+
"learning_rate": 0.00016032047579339287,
|
2754 |
+
"loss": 46.0,
|
2755 |
+
"step": 390
|
2756 |
+
},
|
2757 |
+
{
|
2758 |
+
"epoch": 0.007870608009501092,
|
2759 |
+
"grad_norm": 9.579762991052121e-05,
|
2760 |
+
"learning_rate": 0.00016012650895801995,
|
2761 |
+
"loss": 46.0,
|
2762 |
+
"step": 391
|
2763 |
+
},
|
2764 |
+
{
|
2765 |
+
"epoch": 0.007890737441750456,
|
2766 |
+
"grad_norm": 6.637965998379514e-05,
|
2767 |
+
"learning_rate": 0.00015993218717159254,
|
2768 |
+
"loss": 46.0,
|
2769 |
+
"step": 392
|
2770 |
+
},
|
2771 |
+
{
|
2772 |
+
"epoch": 0.00791086687399982,
|
2773 |
+
"grad_norm": 3.44514446624089e-05,
|
2774 |
+
"learning_rate": 0.00015973751158127058,
|
2775 |
+
"loss": 46.0,
|
2776 |
+
"step": 393
|
2777 |
+
},
|
2778 |
+
{
|
2779 |
+
"epoch": 0.007930996306249181,
|
2780 |
+
"grad_norm": 3.686073250719346e-05,
|
2781 |
+
"learning_rate": 0.00015954248333630266,
|
2782 |
+
"loss": 46.0,
|
2783 |
+
"step": 394
|
2784 |
+
},
|
2785 |
+
{
|
2786 |
+
"epoch": 0.007951125738498545,
|
2787 |
+
"grad_norm": 5.8047575294040143e-05,
|
2788 |
+
"learning_rate": 0.0001593471035880193,
|
2789 |
+
"loss": 46.0,
|
2790 |
+
"step": 395
|
2791 |
+
},
|
2792 |
+
{
|
2793 |
+
"epoch": 0.007971255170747909,
|
2794 |
+
"grad_norm": 0.0001278682757401839,
|
2795 |
+
"learning_rate": 0.00015915137348982596,
|
2796 |
+
"loss": 46.0,
|
2797 |
+
"step": 396
|
2798 |
+
},
|
2799 |
+
{
|
2800 |
+
"epoch": 0.007991384602997273,
|
2801 |
+
"grad_norm": 5.004106424166821e-05,
|
2802 |
+
"learning_rate": 0.00015895529419719643,
|
2803 |
+
"loss": 46.0,
|
2804 |
+
"step": 397
|
2805 |
+
},
|
2806 |
+
{
|
2807 |
+
"epoch": 0.008011514035246636,
|
2808 |
+
"grad_norm": 0.00018534505215939134,
|
2809 |
+
"learning_rate": 0.00015875886686766597,
|
2810 |
+
"loss": 46.0,
|
2811 |
+
"step": 398
|
2812 |
+
},
|
2813 |
+
{
|
2814 |
+
"epoch": 0.008031643467496,
|
2815 |
+
"grad_norm": 7.75425141910091e-05,
|
2816 |
+
"learning_rate": 0.00015856209266082436,
|
2817 |
+
"loss": 46.0,
|
2818 |
+
"step": 399
|
2819 |
+
},
|
2820 |
+
{
|
2821 |
+
"epoch": 0.008051772899745362,
|
2822 |
+
"grad_norm": 0.00016450489056296647,
|
2823 |
+
"learning_rate": 0.0001583649727383092,
|
2824 |
+
"loss": 46.0,
|
2825 |
+
"step": 400
|
2826 |
+
},
|
2827 |
+
{
|
2828 |
+
"epoch": 0.008071902331994726,
|
2829 |
+
"grad_norm": 0.00011991400242550299,
|
2830 |
+
"learning_rate": 0.00015816750826379896,
|
2831 |
+
"loss": 46.0,
|
2832 |
+
"step": 401
|
2833 |
+
},
|
2834 |
+
{
|
2835 |
+
"epoch": 0.00809203176424409,
|
2836 |
+
"grad_norm": 7.743191963527352e-05,
|
2837 |
+
"learning_rate": 0.00015796970040300612,
|
2838 |
+
"loss": 46.0,
|
2839 |
+
"step": 402
|
2840 |
+
},
|
2841 |
+
{
|
2842 |
+
"epoch": 0.008112161196493453,
|
2843 |
+
"grad_norm": 5.860636883880943e-05,
|
2844 |
+
"learning_rate": 0.0001577715503236704,
|
2845 |
+
"loss": 46.0,
|
2846 |
+
"step": 403
|
2847 |
+
},
|
2848 |
+
{
|
2849 |
+
"epoch": 0.008132290628742817,
|
2850 |
+
"grad_norm": 6.259889778448269e-05,
|
2851 |
+
"learning_rate": 0.00015757305919555164,
|
2852 |
+
"loss": 46.0,
|
2853 |
+
"step": 404
|
2854 |
+
},
|
2855 |
+
{
|
2856 |
+
"epoch": 0.00815242006099218,
|
2857 |
+
"grad_norm": 0.0001673314836807549,
|
2858 |
+
"learning_rate": 0.00015737422819042313,
|
2859 |
+
"loss": 46.0,
|
2860 |
+
"step": 405
|
2861 |
+
},
|
2862 |
+
{
|
2863 |
+
"epoch": 0.008172549493241543,
|
2864 |
+
"grad_norm": 0.00013351505913306028,
|
2865 |
+
"learning_rate": 0.00015717505848206455,
|
2866 |
+
"loss": 46.0,
|
2867 |
+
"step": 406
|
2868 |
+
},
|
2869 |
+
{
|
2870 |
+
"epoch": 0.008192678925490906,
|
2871 |
+
"grad_norm": 5.856342249899171e-05,
|
2872 |
+
"learning_rate": 0.00015697555124625508,
|
2873 |
+
"loss": 46.0,
|
2874 |
+
"step": 407
|
2875 |
+
},
|
2876 |
+
{
|
2877 |
+
"epoch": 0.00821280835774027,
|
2878 |
+
"grad_norm": 8.595505642006174e-05,
|
2879 |
+
"learning_rate": 0.00015677570766076652,
|
2880 |
+
"loss": 46.0,
|
2881 |
+
"step": 408
|
2882 |
+
},
|
2883 |
+
{
|
2884 |
+
"epoch": 0.008232937789989634,
|
2885 |
+
"grad_norm": 8.992596121970564e-05,
|
2886 |
+
"learning_rate": 0.0001565755289053562,
|
2887 |
+
"loss": 46.0,
|
2888 |
+
"step": 409
|
2889 |
+
},
|
2890 |
+
{
|
2891 |
+
"epoch": 0.008253067222238997,
|
2892 |
+
"grad_norm": 0.00011127449397463351,
|
2893 |
+
"learning_rate": 0.00015637501616176005,
|
2894 |
+
"loss": 46.0,
|
2895 |
+
"step": 410
|
2896 |
+
},
|
2897 |
+
{
|
2898 |
+
"epoch": 0.00827319665448836,
|
2899 |
+
"grad_norm": 7.977043424034491e-05,
|
2900 |
+
"learning_rate": 0.00015617417061368586,
|
2901 |
+
"loss": 46.0,
|
2902 |
+
"step": 411
|
2903 |
+
},
|
2904 |
+
{
|
2905 |
+
"epoch": 0.008293326086737723,
|
2906 |
+
"grad_norm": 7.366786303464323e-05,
|
2907 |
+
"learning_rate": 0.0001559729934468059,
|
2908 |
+
"loss": 46.0,
|
2909 |
+
"step": 412
|
2910 |
+
},
|
2911 |
+
{
|
2912 |
+
"epoch": 0.008313455518987087,
|
2913 |
+
"grad_norm": 6.224372191354632e-05,
|
2914 |
+
"learning_rate": 0.0001557714858487502,
|
2915 |
+
"loss": 46.0,
|
2916 |
+
"step": 413
|
2917 |
+
},
|
2918 |
+
{
|
2919 |
+
"epoch": 0.00833358495123645,
|
2920 |
+
"grad_norm": 0.00012246175901964307,
|
2921 |
+
"learning_rate": 0.00015556964900909952,
|
2922 |
+
"loss": 46.0,
|
2923 |
+
"step": 414
|
2924 |
+
},
|
2925 |
+
{
|
2926 |
+
"epoch": 0.008353714383485814,
|
2927 |
+
"grad_norm": 9.144249634118751e-05,
|
2928 |
+
"learning_rate": 0.00015536748411937814,
|
2929 |
+
"loss": 46.0,
|
2930 |
+
"step": 415
|
2931 |
+
},
|
2932 |
+
{
|
2933 |
+
"epoch": 0.008373843815735178,
|
2934 |
+
"grad_norm": 4.544908370007761e-05,
|
2935 |
+
"learning_rate": 0.00015516499237304703,
|
2936 |
+
"loss": 46.0,
|
2937 |
+
"step": 416
|
2938 |
+
},
|
2939 |
+
{
|
2940 |
+
"epoch": 0.00839397324798454,
|
2941 |
+
"grad_norm": 7.032406574580818e-05,
|
2942 |
+
"learning_rate": 0.00015496217496549673,
|
2943 |
+
"loss": 46.0,
|
2944 |
+
"step": 417
|
2945 |
+
},
|
2946 |
+
{
|
2947 |
+
"epoch": 0.008414102680233904,
|
2948 |
+
"grad_norm": 0.00028849008958786726,
|
2949 |
+
"learning_rate": 0.00015475903309404023,
|
2950 |
+
"loss": 46.0,
|
2951 |
+
"step": 418
|
2952 |
+
},
|
2953 |
+
{
|
2954 |
+
"epoch": 0.008434232112483267,
|
2955 |
+
"grad_norm": 7.347687642322853e-05,
|
2956 |
+
"learning_rate": 0.00015455556795790603,
|
2957 |
+
"loss": 46.0,
|
2958 |
+
"step": 419
|
2959 |
+
},
|
2960 |
+
{
|
2961 |
+
"epoch": 0.008454361544732631,
|
2962 |
+
"grad_norm": 0.0001396266306983307,
|
2963 |
+
"learning_rate": 0.000154351780758231,
|
2964 |
+
"loss": 46.0,
|
2965 |
+
"step": 420
|
2966 |
+
},
|
2967 |
+
{
|
2968 |
+
"epoch": 0.008474490976981995,
|
2969 |
+
"grad_norm": 7.422738417517394e-05,
|
2970 |
+
"learning_rate": 0.00015414767269805317,
|
2971 |
+
"loss": 46.0,
|
2972 |
+
"step": 421
|
2973 |
+
},
|
2974 |
+
{
|
2975 |
+
"epoch": 0.008494620409231358,
|
2976 |
+
"grad_norm": 0.00010193362686550245,
|
2977 |
+
"learning_rate": 0.00015394324498230487,
|
2978 |
+
"loss": 46.0,
|
2979 |
+
"step": 422
|
2980 |
+
},
|
2981 |
+
{
|
2982 |
+
"epoch": 0.00851474984148072,
|
2983 |
+
"grad_norm": 3.995158476755023e-05,
|
2984 |
+
"learning_rate": 0.00015373849881780542,
|
2985 |
+
"loss": 46.0,
|
2986 |
+
"step": 423
|
2987 |
+
},
|
2988 |
+
{
|
2989 |
+
"epoch": 0.008534879273730084,
|
2990 |
+
"grad_norm": 0.00010771408415166661,
|
2991 |
+
"learning_rate": 0.00015353343541325406,
|
2992 |
+
"loss": 46.0,
|
2993 |
+
"step": 424
|
2994 |
+
},
|
2995 |
+
{
|
2996 |
+
"epoch": 0.008555008705979448,
|
2997 |
+
"grad_norm": 9.801267879083753e-05,
|
2998 |
+
"learning_rate": 0.00015332805597922285,
|
2999 |
+
"loss": 46.0,
|
3000 |
+
"step": 425
|
3001 |
+
},
|
3002 |
+
{
|
3003 |
+
"epoch": 0.008575138138228812,
|
3004 |
+
"grad_norm": 5.4972933867247775e-05,
|
3005 |
+
"learning_rate": 0.00015312236172814955,
|
3006 |
+
"loss": 46.0,
|
3007 |
+
"step": 426
|
3008 |
+
},
|
3009 |
+
{
|
3010 |
+
"epoch": 0.008595267570478175,
|
3011 |
+
"grad_norm": 0.00011087340681115165,
|
3012 |
+
"learning_rate": 0.0001529163538743303,
|
3013 |
+
"loss": 46.0,
|
3014 |
+
"step": 427
|
3015 |
+
},
|
3016 |
+
{
|
3017 |
+
"epoch": 0.008615397002727537,
|
3018 |
+
"grad_norm": 9.032541129272431e-05,
|
3019 |
+
"learning_rate": 0.00015271003363391268,
|
3020 |
+
"loss": 46.0,
|
3021 |
+
"step": 428
|
3022 |
+
},
|
3023 |
+
{
|
3024 |
+
"epoch": 0.008635526434976901,
|
3025 |
+
"grad_norm": 9.607595711713657e-05,
|
3026 |
+
"learning_rate": 0.00015250340222488826,
|
3027 |
+
"loss": 46.0,
|
3028 |
+
"step": 429
|
3029 |
+
},
|
3030 |
+
{
|
3031 |
+
"epoch": 0.008655655867226265,
|
3032 |
+
"grad_norm": 0.00015448669728357345,
|
3033 |
+
"learning_rate": 0.00015229646086708574,
|
3034 |
+
"loss": 46.0,
|
3035 |
+
"step": 430
|
3036 |
+
},
|
3037 |
+
{
|
3038 |
+
"epoch": 0.008675785299475628,
|
3039 |
+
"grad_norm": 4.9507445510244e-05,
|
3040 |
+
"learning_rate": 0.0001520892107821635,
|
3041 |
+
"loss": 46.0,
|
3042 |
+
"step": 431
|
3043 |
+
},
|
3044 |
+
{
|
3045 |
+
"epoch": 0.008695914731724992,
|
3046 |
+
"grad_norm": 0.0001763905311236158,
|
3047 |
+
"learning_rate": 0.0001518816531936024,
|
3048 |
+
"loss": 46.0,
|
3049 |
+
"step": 432
|
3050 |
+
},
|
3051 |
+
{
|
3052 |
+
"epoch": 0.008716044163974356,
|
3053 |
+
"grad_norm": 0.0001318747381446883,
|
3054 |
+
"learning_rate": 0.0001516737893266987,
|
3055 |
+
"loss": 46.0,
|
3056 |
+
"step": 433
|
3057 |
+
},
|
3058 |
+
{
|
3059 |
+
"epoch": 0.008736173596223718,
|
3060 |
+
"grad_norm": 0.00010619282693369314,
|
3061 |
+
"learning_rate": 0.00015146562040855676,
|
3062 |
+
"loss": 46.0,
|
3063 |
+
"step": 434
|
3064 |
+
},
|
3065 |
+
{
|
3066 |
+
"epoch": 0.008756303028473082,
|
3067 |
+
"grad_norm": 8.555618114769459e-05,
|
3068 |
+
"learning_rate": 0.00015125714766808167,
|
3069 |
+
"loss": 46.0,
|
3070 |
+
"step": 435
|
3071 |
+
},
|
3072 |
+
{
|
3073 |
+
"epoch": 0.008776432460722445,
|
3074 |
+
"grad_norm": 0.00016039168986026198,
|
3075 |
+
"learning_rate": 0.00015104837233597223,
|
3076 |
+
"loss": 46.0,
|
3077 |
+
"step": 436
|
3078 |
+
},
|
3079 |
+
{
|
3080 |
+
"epoch": 0.008796561892971809,
|
3081 |
+
"grad_norm": 0.00014670997916255146,
|
3082 |
+
"learning_rate": 0.00015083929564471343,
|
3083 |
+
"loss": 46.0,
|
3084 |
+
"step": 437
|
3085 |
+
},
|
3086 |
+
{
|
3087 |
+
"epoch": 0.008816691325221173,
|
3088 |
+
"grad_norm": 6.258589564822614e-05,
|
3089 |
+
"learning_rate": 0.00015062991882856946,
|
3090 |
+
"loss": 46.0,
|
3091 |
+
"step": 438
|
3092 |
+
},
|
3093 |
+
{
|
3094 |
+
"epoch": 0.008836820757470536,
|
3095 |
+
"grad_norm": 7.728593482170254e-05,
|
3096 |
+
"learning_rate": 0.00015042024312357616,
|
3097 |
+
"loss": 46.0,
|
3098 |
+
"step": 439
|
3099 |
+
},
|
3100 |
+
{
|
3101 |
+
"epoch": 0.008856950189719898,
|
3102 |
+
"grad_norm": 9.296549978898838e-05,
|
3103 |
+
"learning_rate": 0.00015021026976753385,
|
3104 |
+
"loss": 46.0,
|
3105 |
+
"step": 440
|
3106 |
+
},
|
3107 |
+
{
|
3108 |
+
"epoch": 0.008877079621969262,
|
3109 |
+
"grad_norm": 9.728507575346157e-05,
|
3110 |
+
"learning_rate": 0.00015000000000000001,
|
3111 |
+
"loss": 46.0,
|
3112 |
+
"step": 441
|
3113 |
+
},
|
3114 |
+
{
|
3115 |
+
"epoch": 0.008897209054218626,
|
3116 |
+
"grad_norm": 0.00014637406275141984,
|
3117 |
+
"learning_rate": 0.000149789435062282,
|
3118 |
+
"loss": 46.0,
|
3119 |
+
"step": 442
|
3120 |
+
},
|
3121 |
+
{
|
3122 |
+
"epoch": 0.00891733848646799,
|
3123 |
+
"grad_norm": 5.648445221595466e-05,
|
3124 |
+
"learning_rate": 0.00014957857619742957,
|
3125 |
+
"loss": 46.0,
|
3126 |
+
"step": 443
|
3127 |
+
},
|
3128 |
+
{
|
3129 |
+
"epoch": 0.008937467918717353,
|
3130 |
+
"grad_norm": 0.00013621490506920964,
|
3131 |
+
"learning_rate": 0.0001493674246502278,
|
3132 |
+
"loss": 46.0,
|
3133 |
+
"step": 444
|
3134 |
+
},
|
3135 |
+
{
|
3136 |
+
"epoch": 0.008957597350966715,
|
3137 |
+
"grad_norm": 6.250360456760973e-05,
|
3138 |
+
"learning_rate": 0.00014915598166718945,
|
3139 |
+
"loss": 46.0,
|
3140 |
+
"step": 445
|
3141 |
+
},
|
3142 |
+
{
|
3143 |
+
"epoch": 0.008977726783216079,
|
3144 |
+
"grad_norm": 0.00017833786841947585,
|
3145 |
+
"learning_rate": 0.00014894424849654783,
|
3146 |
+
"loss": 46.0,
|
3147 |
+
"step": 446
|
3148 |
+
},
|
3149 |
+
{
|
3150 |
+
"epoch": 0.008997856215465443,
|
3151 |
+
"grad_norm": 6.154891161713749e-05,
|
3152 |
+
"learning_rate": 0.00014873222638824937,
|
3153 |
+
"loss": 46.0,
|
3154 |
+
"step": 447
|
3155 |
+
},
|
3156 |
+
{
|
3157 |
+
"epoch": 0.009017985647714806,
|
3158 |
+
"grad_norm": 0.0001515242620371282,
|
3159 |
+
"learning_rate": 0.0001485199165939461,
|
3160 |
+
"loss": 46.0,
|
3161 |
+
"step": 448
|
3162 |
+
},
|
3163 |
+
{
|
3164 |
+
"epoch": 0.00903811507996417,
|
3165 |
+
"grad_norm": 9.545722423354164e-05,
|
3166 |
+
"learning_rate": 0.00014830732036698845,
|
3167 |
+
"loss": 46.0,
|
3168 |
+
"step": 449
|
3169 |
+
},
|
3170 |
+
{
|
3171 |
+
"epoch": 0.009058244512213534,
|
3172 |
+
"grad_norm": 5.560795034398325e-05,
|
3173 |
+
"learning_rate": 0.0001480944389624178,
|
3174 |
+
"loss": 46.0,
|
3175 |
+
"step": 450
|
3176 |
+
},
|
3177 |
+
{
|
3178 |
+
"epoch": 0.009078373944462896,
|
3179 |
+
"grad_norm": 0.000126198137877509,
|
3180 |
+
"learning_rate": 0.00014788127363695897,
|
3181 |
+
"loss": 46.0,
|
3182 |
+
"step": 451
|
3183 |
+
},
|
3184 |
+
{
|
3185 |
+
"epoch": 0.00909850337671226,
|
3186 |
+
"grad_norm": 9.512303222436458e-05,
|
3187 |
+
"learning_rate": 0.00014766782564901298,
|
3188 |
+
"loss": 46.0,
|
3189 |
+
"step": 452
|
3190 |
+
},
|
3191 |
+
{
|
3192 |
+
"epoch": 0.009118632808961623,
|
3193 |
+
"grad_norm": 0.0001474691671319306,
|
3194 |
+
"learning_rate": 0.00014745409625864942,
|
3195 |
+
"loss": 46.0,
|
3196 |
+
"step": 453
|
3197 |
+
},
|
3198 |
+
{
|
3199 |
+
"epoch": 0.009138762241210987,
|
3200 |
+
"grad_norm": 7.026526873232797e-05,
|
3201 |
+
"learning_rate": 0.0001472400867275992,
|
3202 |
+
"loss": 46.0,
|
3203 |
+
"step": 454
|
3204 |
+
},
|
3205 |
+
{
|
3206 |
+
"epoch": 0.00915889167346035,
|
3207 |
+
"grad_norm": 0.00010350546654080972,
|
3208 |
+
"learning_rate": 0.00014702579831924698,
|
3209 |
+
"loss": 46.0,
|
3210 |
+
"step": 455
|
3211 |
+
},
|
3212 |
+
{
|
3213 |
+
"epoch": 0.009179021105709714,
|
3214 |
+
"grad_norm": 6.921013118699193e-05,
|
3215 |
+
"learning_rate": 0.00014681123229862367,
|
3216 |
+
"loss": 46.0,
|
3217 |
+
"step": 456
|
3218 |
+
},
|
3219 |
+
{
|
3220 |
+
"epoch": 0.009199150537959076,
|
3221 |
+
"grad_norm": 0.00011726860975613818,
|
3222 |
+
"learning_rate": 0.0001465963899323992,
|
3223 |
+
"loss": 46.0,
|
3224 |
+
"step": 457
|
3225 |
+
},
|
3226 |
+
{
|
3227 |
+
"epoch": 0.00921927997020844,
|
3228 |
+
"grad_norm": 8.906117000151426e-05,
|
3229 |
+
"learning_rate": 0.00014638127248887473,
|
3230 |
+
"loss": 46.0,
|
3231 |
+
"step": 458
|
3232 |
+
},
|
3233 |
+
{
|
3234 |
+
"epoch": 0.009239409402457804,
|
3235 |
+
"grad_norm": 0.00013452931307256222,
|
3236 |
+
"learning_rate": 0.00014616588123797535,
|
3237 |
+
"loss": 46.0,
|
3238 |
+
"step": 459
|
3239 |
+
},
|
3240 |
+
{
|
3241 |
+
"epoch": 0.009259538834707167,
|
3242 |
+
"grad_norm": 9.775407670531422e-05,
|
3243 |
+
"learning_rate": 0.0001459502174512426,
|
3244 |
+
"loss": 46.0,
|
3245 |
+
"step": 460
|
3246 |
+
},
|
3247 |
+
{
|
3248 |
+
"epoch": 0.009279668266956531,
|
3249 |
+
"grad_norm": 8.385746332351118e-05,
|
3250 |
+
"learning_rate": 0.0001457342824018269,
|
3251 |
+
"loss": 46.0,
|
3252 |
+
"step": 461
|
3253 |
+
},
|
3254 |
+
{
|
3255 |
+
"epoch": 0.009299797699205893,
|
3256 |
+
"grad_norm": 5.9415571740828454e-05,
|
3257 |
+
"learning_rate": 0.00014551807736447995,
|
3258 |
+
"loss": 46.0,
|
3259 |
+
"step": 462
|
3260 |
+
},
|
3261 |
+
{
|
3262 |
+
"epoch": 0.009319927131455257,
|
3263 |
+
"grad_norm": 0.00011705618089763448,
|
3264 |
+
"learning_rate": 0.0001453016036155474,
|
3265 |
+
"loss": 46.0,
|
3266 |
+
"step": 463
|
3267 |
+
},
|
3268 |
+
{
|
3269 |
+
"epoch": 0.00934005656370462,
|
3270 |
+
"grad_norm": 8.00532943685539e-05,
|
3271 |
+
"learning_rate": 0.00014508486243296122,
|
3272 |
+
"loss": 46.0,
|
3273 |
+
"step": 464
|
3274 |
+
},
|
3275 |
+
{
|
3276 |
+
"epoch": 0.009360185995953984,
|
3277 |
+
"grad_norm": 0.00010752366506494582,
|
3278 |
+
"learning_rate": 0.00014486785509623202,
|
3279 |
+
"loss": 46.0,
|
3280 |
+
"step": 465
|
3281 |
+
},
|
3282 |
+
{
|
3283 |
+
"epoch": 0.009380315428203348,
|
3284 |
+
"grad_norm": 0.00011672089749481529,
|
3285 |
+
"learning_rate": 0.00014465058288644174,
|
3286 |
+
"loss": 46.0,
|
3287 |
+
"step": 466
|
3288 |
+
},
|
3289 |
+
{
|
3290 |
+
"epoch": 0.009400444860452712,
|
3291 |
+
"grad_norm": 0.0001032515792758204,
|
3292 |
+
"learning_rate": 0.00014443304708623597,
|
3293 |
+
"loss": 46.0,
|
3294 |
+
"step": 467
|
3295 |
+
},
|
3296 |
+
{
|
3297 |
+
"epoch": 0.009420574292702074,
|
3298 |
+
"grad_norm": 7.637974340468645e-05,
|
3299 |
+
"learning_rate": 0.00014421524897981637,
|
3300 |
+
"loss": 46.0,
|
3301 |
+
"step": 468
|
3302 |
+
},
|
3303 |
+
{
|
3304 |
+
"epoch": 0.009440703724951437,
|
3305 |
+
"grad_norm": 0.00010200442193308845,
|
3306 |
+
"learning_rate": 0.00014399718985293297,
|
3307 |
+
"loss": 46.0,
|
3308 |
+
"step": 469
|
3309 |
+
},
|
3310 |
+
{
|
3311 |
+
"epoch": 0.009460833157200801,
|
3312 |
+
"grad_norm": 0.00010882026253966615,
|
3313 |
+
"learning_rate": 0.00014377887099287698,
|
3314 |
+
"loss": 46.0,
|
3315 |
+
"step": 470
|
3316 |
+
},
|
3317 |
+
{
|
3318 |
+
"epoch": 0.009480962589450165,
|
3319 |
+
"grad_norm": 0.00022409467783290893,
|
3320 |
+
"learning_rate": 0.00014356029368847264,
|
3321 |
+
"loss": 46.0,
|
3322 |
+
"step": 471
|
3323 |
+
},
|
3324 |
+
{
|
3325 |
+
"epoch": 0.009501092021699529,
|
3326 |
+
"grad_norm": 0.00014206068590283394,
|
3327 |
+
"learning_rate": 0.0001433414592300701,
|
3328 |
+
"loss": 46.0,
|
3329 |
+
"step": 472
|
3330 |
+
},
|
3331 |
+
{
|
3332 |
+
"epoch": 0.009521221453948892,
|
3333 |
+
"grad_norm": 5.281836274662055e-05,
|
3334 |
+
"learning_rate": 0.00014312236890953744,
|
3335 |
+
"loss": 46.0,
|
3336 |
+
"step": 473
|
3337 |
+
},
|
3338 |
+
{
|
3339 |
+
"epoch": 0.009541350886198254,
|
3340 |
+
"grad_norm": 0.00011217795690754429,
|
3341 |
+
"learning_rate": 0.00014290302402025334,
|
3342 |
+
"loss": 46.0,
|
3343 |
+
"step": 474
|
3344 |
+
},
|
3345 |
+
{
|
3346 |
+
"epoch": 0.009561480318447618,
|
3347 |
+
"grad_norm": 0.00011971918138442561,
|
3348 |
+
"learning_rate": 0.00014268342585709913,
|
3349 |
+
"loss": 46.0,
|
3350 |
+
"step": 475
|
3351 |
+
},
|
3352 |
+
{
|
3353 |
+
"epoch": 0.009581609750696982,
|
3354 |
+
"grad_norm": 0.0002392362366663292,
|
3355 |
+
"learning_rate": 0.00014246357571645152,
|
3356 |
+
"loss": 46.0,
|
3357 |
+
"step": 476
|
3358 |
+
},
|
3359 |
+
{
|
3360 |
+
"epoch": 0.009601739182946345,
|
3361 |
+
"grad_norm": 7.42652773624286e-05,
|
3362 |
+
"learning_rate": 0.00014224347489617456,
|
3363 |
+
"loss": 46.0,
|
3364 |
+
"step": 477
|
3365 |
+
},
|
3366 |
+
{
|
3367 |
+
"epoch": 0.009621868615195709,
|
3368 |
+
"grad_norm": 0.00019455078290775418,
|
3369 |
+
"learning_rate": 0.00014202312469561228,
|
3370 |
+
"loss": 46.0,
|
3371 |
+
"step": 478
|
3372 |
+
},
|
3373 |
+
{
|
3374 |
+
"epoch": 0.009641998047445071,
|
3375 |
+
"grad_norm": 0.00014721274783369154,
|
3376 |
+
"learning_rate": 0.00014180252641558084,
|
3377 |
+
"loss": 46.0,
|
3378 |
+
"step": 479
|
3379 |
+
},
|
3380 |
+
{
|
3381 |
+
"epoch": 0.009662127479694435,
|
3382 |
+
"grad_norm": 0.00016650428005959839,
|
3383 |
+
"learning_rate": 0.00014158168135836095,
|
3384 |
+
"loss": 46.0,
|
3385 |
+
"step": 480
|
3386 |
+
},
|
3387 |
+
{
|
3388 |
+
"epoch": 0.009682256911943798,
|
3389 |
+
"grad_norm": 0.0002557269181124866,
|
3390 |
+
"learning_rate": 0.00014136059082769017,
|
3391 |
+
"loss": 46.0,
|
3392 |
+
"step": 481
|
3393 |
+
},
|
3394 |
+
{
|
3395 |
+
"epoch": 0.009702386344193162,
|
3396 |
+
"grad_norm": 0.00013280926214065403,
|
3397 |
+
"learning_rate": 0.00014113925612875512,
|
3398 |
+
"loss": 46.0,
|
3399 |
+
"step": 482
|
3400 |
+
},
|
3401 |
+
{
|
3402 |
+
"epoch": 0.009722515776442526,
|
3403 |
+
"grad_norm": 0.00014411240408662707,
|
3404 |
+
"learning_rate": 0.00014091767856818388,
|
3405 |
+
"loss": 46.0,
|
3406 |
+
"step": 483
|
3407 |
+
},
|
3408 |
+
{
|
3409 |
+
"epoch": 0.00974264520869189,
|
3410 |
+
"grad_norm": 0.00010560146620264277,
|
3411 |
+
"learning_rate": 0.00014069585945403822,
|
3412 |
+
"loss": 46.0,
|
3413 |
+
"step": 484
|
3414 |
+
},
|
3415 |
+
{
|
3416 |
+
"epoch": 0.009762774640941252,
|
3417 |
+
"grad_norm": 0.00022722291760146618,
|
3418 |
+
"learning_rate": 0.00014047380009580594,
|
3419 |
+
"loss": 46.0,
|
3420 |
+
"step": 485
|
3421 |
+
},
|
3422 |
+
{
|
3423 |
+
"epoch": 0.009782904073190615,
|
3424 |
+
"grad_norm": 0.00011408658610889688,
|
3425 |
+
"learning_rate": 0.00014025150180439308,
|
3426 |
+
"loss": 46.0,
|
3427 |
+
"step": 486
|
3428 |
+
},
|
3429 |
+
{
|
3430 |
+
"epoch": 0.009803033505439979,
|
3431 |
+
"grad_norm": 5.5666998378001153e-05,
|
3432 |
+
"learning_rate": 0.00014002896589211618,
|
3433 |
+
"loss": 46.0,
|
3434 |
+
"step": 487
|
3435 |
+
},
|
3436 |
+
{
|
3437 |
+
"epoch": 0.009823162937689343,
|
3438 |
+
"grad_norm": 6.966435466893017e-05,
|
3439 |
+
"learning_rate": 0.00013980619367269455,
|
3440 |
+
"loss": 46.0,
|
3441 |
+
"step": 488
|
3442 |
+
},
|
3443 |
+
{
|
3444 |
+
"epoch": 0.009843292369938706,
|
3445 |
+
"grad_norm": 0.00026750058168545365,
|
3446 |
+
"learning_rate": 0.00013958318646124259,
|
3447 |
+
"loss": 46.0,
|
3448 |
+
"step": 489
|
3449 |
+
},
|
3450 |
+
{
|
3451 |
+
"epoch": 0.00986342180218807,
|
3452 |
+
"grad_norm": 7.481938519049436e-05,
|
3453 |
+
"learning_rate": 0.0001393599455742618,
|
3454 |
+
"loss": 46.0,
|
3455 |
+
"step": 490
|
3456 |
+
},
|
3457 |
+
{
|
3458 |
+
"epoch": 0.009883551234437432,
|
3459 |
+
"grad_norm": 8.790163701632991e-05,
|
3460 |
+
"learning_rate": 0.00013913647232963332,
|
3461 |
+
"loss": 46.0,
|
3462 |
+
"step": 491
|
3463 |
+
},
|
3464 |
+
{
|
3465 |
+
"epoch": 0.009903680666686796,
|
3466 |
+
"grad_norm": 0.00020705680071841925,
|
3467 |
+
"learning_rate": 0.00013891276804660991,
|
3468 |
+
"loss": 46.0,
|
3469 |
+
"step": 492
|
3470 |
+
},
|
3471 |
+
{
|
3472 |
+
"epoch": 0.00992381009893616,
|
3473 |
+
"grad_norm": 8.553229417884722e-05,
|
3474 |
+
"learning_rate": 0.00013868883404580823,
|
3475 |
+
"loss": 46.0,
|
3476 |
+
"step": 493
|
3477 |
+
},
|
3478 |
+
{
|
3479 |
+
"epoch": 0.009943939531185523,
|
3480 |
+
"grad_norm": 7.284261664608493e-05,
|
3481 |
+
"learning_rate": 0.00013846467164920116,
|
3482 |
+
"loss": 46.0,
|
3483 |
+
"step": 494
|
3484 |
+
},
|
3485 |
+
{
|
3486 |
+
"epoch": 0.009964068963434887,
|
3487 |
+
"grad_norm": 0.00010190778266405687,
|
3488 |
+
"learning_rate": 0.00013824028218010977,
|
3489 |
+
"loss": 46.0,
|
3490 |
+
"step": 495
|
3491 |
+
},
|
3492 |
+
{
|
3493 |
+
"epoch": 0.009984198395684249,
|
3494 |
+
"grad_norm": 0.00012867138138972223,
|
3495 |
+
"learning_rate": 0.00013801566696319562,
|
3496 |
+
"loss": 46.0,
|
3497 |
+
"step": 496
|
3498 |
+
},
|
3499 |
+
{
|
3500 |
+
"epoch": 0.010004327827933613,
|
3501 |
+
"grad_norm": 0.00010176874639000744,
|
3502 |
+
"learning_rate": 0.0001377908273244531,
|
3503 |
+
"loss": 46.0,
|
3504 |
+
"step": 497
|
3505 |
+
},
|
3506 |
+
{
|
3507 |
+
"epoch": 0.010024457260182976,
|
3508 |
+
"grad_norm": 0.00021709667635150254,
|
3509 |
+
"learning_rate": 0.0001375657645912014,
|
3510 |
+
"loss": 46.0,
|
3511 |
+
"step": 498
|
3512 |
+
},
|
3513 |
+
{
|
3514 |
+
"epoch": 0.01004458669243234,
|
3515 |
+
"grad_norm": 0.0002606755297165364,
|
3516 |
+
"learning_rate": 0.0001373404800920765,
|
3517 |
+
"loss": 46.0,
|
3518 |
+
"step": 499
|
3519 |
+
},
|
3520 |
+
{
|
3521 |
+
"epoch": 0.010064716124681704,
|
3522 |
+
"grad_norm": 0.00011558020196389407,
|
3523 |
+
"learning_rate": 0.00013711497515702398,
|
3524 |
+
"loss": 46.0,
|
3525 |
+
"step": 500
|
3526 |
+
},
|
3527 |
+
{
|
3528 |
+
"epoch": 0.010084845556931068,
|
3529 |
+
"grad_norm": 0.0002265808725496754,
|
3530 |
+
"learning_rate": 0.0001368892511172903,
|
3531 |
+
"loss": 46.0,
|
3532 |
+
"step": 501
|
3533 |
+
},
|
3534 |
+
{
|
3535 |
+
"epoch": 0.01010497498918043,
|
3536 |
+
"grad_norm": 0.00011339668708387762,
|
3537 |
+
"learning_rate": 0.0001366633093054157,
|
3538 |
+
"loss": 46.0,
|
3539 |
+
"step": 502
|
3540 |
+
},
|
3541 |
+
{
|
3542 |
+
"epoch": 0.010125104421429793,
|
3543 |
+
"grad_norm": 0.00012689345749095082,
|
3544 |
+
"learning_rate": 0.00013643715105522589,
|
3545 |
+
"loss": 46.0,
|
3546 |
+
"step": 503
|
3547 |
+
},
|
3548 |
+
{
|
3549 |
+
"epoch": 0.010145233853679157,
|
3550 |
+
"grad_norm": 0.0001654831285122782,
|
3551 |
+
"learning_rate": 0.0001362107777018243,
|
3552 |
+
"loss": 46.0,
|
3553 |
+
"step": 504
|
3554 |
+
},
|
3555 |
+
{
|
3556 |
+
"epoch": 0.01016536328592852,
|
3557 |
+
"grad_norm": 0.00022711421479471028,
|
3558 |
+
"learning_rate": 0.0001359841905815842,
|
3559 |
+
"loss": 46.0,
|
3560 |
+
"step": 505
|
3561 |
+
},
|
3562 |
+
{
|
3563 |
+
"epoch": 0.010185492718177884,
|
3564 |
+
"grad_norm": 0.00016627443255856633,
|
3565 |
+
"learning_rate": 0.00013575739103214088,
|
3566 |
+
"loss": 46.0,
|
3567 |
+
"step": 506
|
3568 |
+
},
|
3569 |
+
{
|
3570 |
+
"epoch": 0.010205622150427248,
|
3571 |
+
"grad_norm": 0.00012627684918697923,
|
3572 |
+
"learning_rate": 0.0001355303803923836,
|
3573 |
+
"loss": 46.0,
|
3574 |
+
"step": 507
|
3575 |
+
},
|
3576 |
+
{
|
3577 |
+
"epoch": 0.01022575158267661,
|
3578 |
+
"grad_norm": 9.957009751815349e-05,
|
3579 |
+
"learning_rate": 0.00013530316000244782,
|
3580 |
+
"loss": 46.0,
|
3581 |
+
"step": 508
|
3582 |
+
},
|
3583 |
+
{
|
3584 |
+
"epoch": 0.010245881014925974,
|
3585 |
+
"grad_norm": 0.00010302881128154695,
|
3586 |
+
"learning_rate": 0.0001350757312037072,
|
3587 |
+
"loss": 46.0,
|
3588 |
+
"step": 509
|
3589 |
+
},
|
3590 |
+
{
|
3591 |
+
"epoch": 0.010266010447175337,
|
3592 |
+
"grad_norm": 7.863016071496531e-05,
|
3593 |
+
"learning_rate": 0.00013484809533876582,
|
3594 |
+
"loss": 46.0,
|
3595 |
+
"step": 510
|
3596 |
+
},
|
3597 |
+
{
|
3598 |
+
"epoch": 0.010286139879424701,
|
3599 |
+
"grad_norm": 0.00017418930656276643,
|
3600 |
+
"learning_rate": 0.00013462025375145,
|
3601 |
+
"loss": 46.0,
|
3602 |
+
"step": 511
|
3603 |
+
},
|
3604 |
+
{
|
3605 |
+
"epoch": 0.010306269311674065,
|
3606 |
+
"grad_norm": 0.0002265576331410557,
|
3607 |
+
"learning_rate": 0.00013439220778680067,
|
3608 |
+
"loss": 46.0,
|
3609 |
+
"step": 512
|
3610 |
+
},
|
3611 |
+
{
|
3612 |
+
"epoch": 0.010326398743923427,
|
3613 |
+
"grad_norm": 9.675358887761831e-05,
|
3614 |
+
"learning_rate": 0.00013416395879106515,
|
3615 |
+
"loss": 46.0,
|
3616 |
+
"step": 513
|
3617 |
+
},
|
3618 |
+
{
|
3619 |
+
"epoch": 0.01034652817617279,
|
3620 |
+
"grad_norm": 0.00034641881939023733,
|
3621 |
+
"learning_rate": 0.00013393550811168948,
|
3622 |
+
"loss": 46.0,
|
3623 |
+
"step": 514
|
3624 |
+
},
|
3625 |
+
{
|
3626 |
+
"epoch": 0.010366657608422154,
|
3627 |
+
"grad_norm": 0.00022542629449162632,
|
3628 |
+
"learning_rate": 0.00013370685709731015,
|
3629 |
+
"loss": 46.0,
|
3630 |
+
"step": 515
|
3631 |
+
},
|
3632 |
+
{
|
3633 |
+
"epoch": 0.010386787040671518,
|
3634 |
+
"grad_norm": 8.584894385421649e-05,
|
3635 |
+
"learning_rate": 0.00013347800709774652,
|
3636 |
+
"loss": 46.0,
|
3637 |
+
"step": 516
|
3638 |
+
},
|
3639 |
+
{
|
3640 |
+
"epoch": 0.010406916472920882,
|
3641 |
+
"grad_norm": 0.0001710738433757797,
|
3642 |
+
"learning_rate": 0.0001332489594639924,
|
3643 |
+
"loss": 46.0,
|
3644 |
+
"step": 517
|
3645 |
+
},
|
3646 |
+
{
|
3647 |
+
"epoch": 0.010427045905170245,
|
3648 |
+
"grad_norm": 0.00018284583347849548,
|
3649 |
+
"learning_rate": 0.00013301971554820853,
|
3650 |
+
"loss": 46.0,
|
3651 |
+
"step": 518
|
3652 |
+
},
|
3653 |
+
{
|
3654 |
+
"epoch": 0.010447175337419607,
|
3655 |
+
"grad_norm": 6.281906826188788e-05,
|
3656 |
+
"learning_rate": 0.00013279027670371426,
|
3657 |
+
"loss": 46.0,
|
3658 |
+
"step": 519
|
3659 |
+
},
|
3660 |
+
{
|
3661 |
+
"epoch": 0.010467304769668971,
|
3662 |
+
"grad_norm": 0.00013591159950010478,
|
3663 |
+
"learning_rate": 0.00013256064428497966,
|
3664 |
+
"loss": 46.0,
|
3665 |
+
"step": 520
|
3666 |
+
},
|
3667 |
+
{
|
3668 |
+
"epoch": 0.010487434201918335,
|
3669 |
+
"grad_norm": 0.00013636577932629734,
|
3670 |
+
"learning_rate": 0.00013233081964761766,
|
3671 |
+
"loss": 46.0,
|
3672 |
+
"step": 521
|
3673 |
+
},
|
3674 |
+
{
|
3675 |
+
"epoch": 0.010507563634167699,
|
3676 |
+
"grad_norm": 0.00017571232456248254,
|
3677 |
+
"learning_rate": 0.0001321008041483758,
|
3678 |
+
"loss": 46.0,
|
3679 |
+
"step": 522
|
3680 |
+
},
|
3681 |
+
{
|
3682 |
+
"epoch": 0.010527693066417062,
|
3683 |
+
"grad_norm": 6.699233199469745e-05,
|
3684 |
+
"learning_rate": 0.0001318705991451285,
|
3685 |
+
"loss": 46.0,
|
3686 |
+
"step": 523
|
3687 |
+
},
|
3688 |
+
{
|
3689 |
+
"epoch": 0.010547822498666426,
|
3690 |
+
"grad_norm": 0.00018198716861661524,
|
3691 |
+
"learning_rate": 0.00013164020599686882,
|
3692 |
+
"loss": 46.0,
|
3693 |
+
"step": 524
|
3694 |
+
},
|
3695 |
+
{
|
3696 |
+
"epoch": 0.010567951930915788,
|
3697 |
+
"grad_norm": 0.000168314523762092,
|
3698 |
+
"learning_rate": 0.00013140962606370048,
|
3699 |
+
"loss": 46.0,
|
3700 |
+
"step": 525
|
3701 |
+
},
|
3702 |
+
{
|
3703 |
+
"epoch": 0.010588081363165152,
|
3704 |
+
"grad_norm": 0.00014153076335787773,
|
3705 |
+
"learning_rate": 0.0001311788607068299,
|
3706 |
+
"loss": 46.0,
|
3707 |
+
"step": 526
|
3708 |
+
},
|
3709 |
+
{
|
3710 |
+
"epoch": 0.010608210795414515,
|
3711 |
+
"grad_norm": 0.00012689942377619445,
|
3712 |
+
"learning_rate": 0.00013094791128855814,
|
3713 |
+
"loss": 46.0,
|
3714 |
+
"step": 527
|
3715 |
+
},
|
3716 |
+
{
|
3717 |
+
"epoch": 0.010628340227663879,
|
3718 |
+
"grad_norm": 9.956786379916593e-05,
|
3719 |
+
"learning_rate": 0.0001307167791722729,
|
3720 |
+
"loss": 46.0,
|
3721 |
+
"step": 528
|
3722 |
+
},
|
3723 |
+
{
|
3724 |
+
"epoch": 0.010648469659913243,
|
3725 |
+
"grad_norm": 8.192278619389981e-05,
|
3726 |
+
"learning_rate": 0.00013048546572244036,
|
3727 |
+
"loss": 46.0,
|
3728 |
+
"step": 529
|
3729 |
+
},
|
3730 |
+
{
|
3731 |
+
"epoch": 0.010668599092162605,
|
3732 |
+
"grad_norm": 0.00021154977730475366,
|
3733 |
+
"learning_rate": 0.0001302539723045971,
|
3734 |
+
"loss": 46.0,
|
3735 |
+
"step": 530
|
3736 |
+
},
|
3737 |
+
{
|
3738 |
+
"epoch": 0.010688728524411968,
|
3739 |
+
"grad_norm": 0.00013896448945160955,
|
3740 |
+
"learning_rate": 0.00013002230028534234,
|
3741 |
+
"loss": 46.0,
|
3742 |
+
"step": 531
|
3743 |
+
},
|
3744 |
+
{
|
3745 |
+
"epoch": 0.010708857956661332,
|
3746 |
+
"grad_norm": 0.0001234956580447033,
|
3747 |
+
"learning_rate": 0.00012979045103232945,
|
3748 |
+
"loss": 46.0,
|
3749 |
+
"step": 532
|
3750 |
+
},
|
3751 |
+
{
|
3752 |
+
"epoch": 0.010728987388910696,
|
3753 |
+
"grad_norm": 0.0001562229444971308,
|
3754 |
+
"learning_rate": 0.00012955842591425818,
|
3755 |
+
"loss": 46.0,
|
3756 |
+
"step": 533
|
3757 |
+
},
|
3758 |
+
{
|
3759 |
+
"epoch": 0.01074911682116006,
|
3760 |
+
"grad_norm": 0.0003348892496433109,
|
3761 |
+
"learning_rate": 0.00012932622630086648,
|
3762 |
+
"loss": 46.0,
|
3763 |
+
"step": 534
|
3764 |
+
},
|
3765 |
+
{
|
3766 |
+
"epoch": 0.010769246253409423,
|
3767 |
+
"grad_norm": 8.87279020389542e-05,
|
3768 |
+
"learning_rate": 0.0001290938535629224,
|
3769 |
+
"loss": 46.0,
|
3770 |
+
"step": 535
|
3771 |
+
},
|
3772 |
+
{
|
3773 |
+
"epoch": 0.010789375685658785,
|
3774 |
+
"grad_norm": 8.013709157239646e-05,
|
3775 |
+
"learning_rate": 0.00012886130907221603,
|
3776 |
+
"loss": 46.0,
|
3777 |
+
"step": 536
|
3778 |
+
},
|
3779 |
+
{
|
3780 |
+
"epoch": 0.010809505117908149,
|
3781 |
+
"grad_norm": 0.00012092996621504426,
|
3782 |
+
"learning_rate": 0.00012862859420155134,
|
3783 |
+
"loss": 46.0,
|
3784 |
+
"step": 537
|
3785 |
+
},
|
3786 |
+
{
|
3787 |
+
"epoch": 0.010829634550157513,
|
3788 |
+
"grad_norm": 9.32335969991982e-05,
|
3789 |
+
"learning_rate": 0.00012839571032473814,
|
3790 |
+
"loss": 46.0,
|
3791 |
+
"step": 538
|
3792 |
+
},
|
3793 |
+
{
|
3794 |
+
"epoch": 0.010849763982406876,
|
3795 |
+
"grad_norm": 0.00016681026318110526,
|
3796 |
+
"learning_rate": 0.00012816265881658405,
|
3797 |
+
"loss": 46.0,
|
3798 |
+
"step": 539
|
3799 |
+
},
|
3800 |
+
{
|
3801 |
+
"epoch": 0.01086989341465624,
|
3802 |
+
"grad_norm": 0.00014514310169033706,
|
3803 |
+
"learning_rate": 0.00012792944105288612,
|
3804 |
+
"loss": 46.0,
|
3805 |
+
"step": 540
|
3806 |
+
},
|
3807 |
+
{
|
3808 |
+
"epoch": 0.010890022846905604,
|
3809 |
+
"grad_norm": 0.00012650150165427476,
|
3810 |
+
"learning_rate": 0.000127696058410423,
|
3811 |
+
"loss": 46.0,
|
3812 |
+
"step": 541
|
3813 |
+
},
|
3814 |
+
{
|
3815 |
+
"epoch": 0.010910152279154966,
|
3816 |
+
"grad_norm": 0.00029365395312197506,
|
3817 |
+
"learning_rate": 0.00012746251226694662,
|
3818 |
+
"loss": 46.0,
|
3819 |
+
"step": 542
|
3820 |
+
},
|
3821 |
+
{
|
3822 |
+
"epoch": 0.01093028171140433,
|
3823 |
+
"grad_norm": 0.00012223645171616226,
|
3824 |
+
"learning_rate": 0.00012722880400117413,
|
3825 |
+
"loss": 46.0,
|
3826 |
+
"step": 543
|
3827 |
+
},
|
3828 |
+
{
|
3829 |
+
"epoch": 0.010950411143653693,
|
3830 |
+
"grad_norm": 0.0004023347864858806,
|
3831 |
+
"learning_rate": 0.00012699493499277983,
|
3832 |
+
"loss": 46.0,
|
3833 |
+
"step": 544
|
3834 |
+
},
|
3835 |
+
{
|
3836 |
+
"epoch": 0.010970540575903057,
|
3837 |
+
"grad_norm": 0.00011891735630342737,
|
3838 |
+
"learning_rate": 0.00012676090662238682,
|
3839 |
+
"loss": 46.0,
|
3840 |
+
"step": 545
|
3841 |
+
},
|
3842 |
+
{
|
3843 |
+
"epoch": 0.01099067000815242,
|
3844 |
+
"grad_norm": 0.00016821158351376653,
|
3845 |
+
"learning_rate": 0.00012652672027155904,
|
3846 |
+
"loss": 46.0,
|
3847 |
+
"step": 546
|
3848 |
+
},
|
3849 |
+
{
|
3850 |
+
"epoch": 0.011010799440401783,
|
3851 |
+
"grad_norm": 7.64864671509713e-05,
|
3852 |
+
"learning_rate": 0.00012629237732279314,
|
3853 |
+
"loss": 46.0,
|
3854 |
+
"step": 547
|
3855 |
+
},
|
3856 |
+
{
|
3857 |
+
"epoch": 0.011030928872651146,
|
3858 |
+
"grad_norm": 0.00015364130376838148,
|
3859 |
+
"learning_rate": 0.0001260578791595101,
|
3860 |
+
"loss": 46.0,
|
3861 |
+
"step": 548
|
3862 |
+
},
|
3863 |
+
{
|
3864 |
+
"epoch": 0.01105105830490051,
|
3865 |
+
"grad_norm": 0.0001618131500435993,
|
3866 |
+
"learning_rate": 0.00012582322716604718,
|
3867 |
+
"loss": 46.0,
|
3868 |
+
"step": 549
|
3869 |
+
},
|
3870 |
+
{
|
3871 |
+
"epoch": 0.011071187737149874,
|
3872 |
+
"grad_norm": 0.0001929528807522729,
|
3873 |
+
"learning_rate": 0.0001255884227276499,
|
3874 |
+
"loss": 46.0,
|
3875 |
+
"step": 550
|
3876 |
+
},
|
3877 |
+
{
|
3878 |
+
"epoch": 0.011091317169399238,
|
3879 |
+
"grad_norm": 0.00011355496098985896,
|
3880 |
+
"learning_rate": 0.0001253534672304636,
|
3881 |
+
"loss": 46.0,
|
3882 |
+
"step": 551
|
3883 |
+
},
|
3884 |
+
{
|
3885 |
+
"epoch": 0.011111446601648601,
|
3886 |
+
"grad_norm": 0.00011806943803094327,
|
3887 |
+
"learning_rate": 0.00012511836206152545,
|
3888 |
+
"loss": 46.0,
|
3889 |
+
"step": 552
|
3890 |
+
},
|
3891 |
+
{
|
3892 |
+
"epoch": 0.011131576033897963,
|
3893 |
+
"grad_norm": 0.00016547176346648484,
|
3894 |
+
"learning_rate": 0.00012488310860875622,
|
3895 |
+
"loss": 46.0,
|
3896 |
+
"step": 553
|
3897 |
+
},
|
3898 |
+
{
|
3899 |
+
"epoch": 0.011151705466147327,
|
3900 |
+
"grad_norm": 0.00012969633098691702,
|
3901 |
+
"learning_rate": 0.0001246477082609519,
|
3902 |
+
"loss": 46.0,
|
3903 |
+
"step": 554
|
3904 |
+
},
|
3905 |
+
{
|
3906 |
+
"epoch": 0.01117183489839669,
|
3907 |
+
"grad_norm": 9.5839895948302e-05,
|
3908 |
+
"learning_rate": 0.00012441216240777585,
|
3909 |
+
"loss": 46.0,
|
3910 |
+
"step": 555
|
3911 |
+
},
|
3912 |
+
{
|
3913 |
+
"epoch": 0.011191964330646054,
|
3914 |
+
"grad_norm": 0.00026142006390728056,
|
3915 |
+
"learning_rate": 0.0001241764724397503,
|
3916 |
+
"loss": 46.0,
|
3917 |
+
"step": 556
|
3918 |
+
},
|
3919 |
+
{
|
3920 |
+
"epoch": 0.011212093762895418,
|
3921 |
+
"grad_norm": 5.916105510550551e-05,
|
3922 |
+
"learning_rate": 0.00012394063974824828,
|
3923 |
+
"loss": 46.0,
|
3924 |
+
"step": 557
|
3925 |
+
},
|
3926 |
+
{
|
3927 |
+
"epoch": 0.011232223195144782,
|
3928 |
+
"grad_norm": 0.00010269715858157724,
|
3929 |
+
"learning_rate": 0.00012370466572548538,
|
3930 |
+
"loss": 46.0,
|
3931 |
+
"step": 558
|
3932 |
+
},
|
3933 |
+
{
|
3934 |
+
"epoch": 0.011252352627394144,
|
3935 |
+
"grad_norm": 0.00017692089022602886,
|
3936 |
+
"learning_rate": 0.0001234685517645115,
|
3937 |
+
"loss": 46.0,
|
3938 |
+
"step": 559
|
3939 |
+
},
|
3940 |
+
{
|
3941 |
+
"epoch": 0.011272482059643507,
|
3942 |
+
"grad_norm": 0.00013197977386880666,
|
3943 |
+
"learning_rate": 0.00012323229925920273,
|
3944 |
+
"loss": 46.0,
|
3945 |
+
"step": 560
|
3946 |
+
},
|
3947 |
+
{
|
3948 |
+
"epoch": 0.011292611491892871,
|
3949 |
+
"grad_norm": 0.00019733706722036004,
|
3950 |
+
"learning_rate": 0.00012299590960425288,
|
3951 |
+
"loss": 46.0,
|
3952 |
+
"step": 561
|
3953 |
+
},
|
3954 |
+
{
|
3955 |
+
"epoch": 0.011312740924142235,
|
3956 |
+
"grad_norm": 0.0001352672406937927,
|
3957 |
+
"learning_rate": 0.00012275938419516552,
|
3958 |
+
"loss": 46.0,
|
3959 |
+
"step": 562
|
3960 |
+
},
|
3961 |
+
{
|
3962 |
+
"epoch": 0.011332870356391599,
|
3963 |
+
"grad_norm": 0.00016101048095151782,
|
3964 |
+
"learning_rate": 0.0001225227244282457,
|
3965 |
+
"loss": 46.0,
|
3966 |
+
"step": 563
|
3967 |
+
},
|
3968 |
+
{
|
3969 |
+
"epoch": 0.01135299978864096,
|
3970 |
+
"grad_norm": 0.0002455017529428005,
|
3971 |
+
"learning_rate": 0.00012228593170059151,
|
3972 |
+
"loss": 46.0,
|
3973 |
+
"step": 564
|
3974 |
+
},
|
3975 |
+
{
|
3976 |
+
"epoch": 0.011373129220890324,
|
3977 |
+
"grad_norm": 0.00036364022525958717,
|
3978 |
+
"learning_rate": 0.000122049007410086,
|
3979 |
+
"loss": 46.0,
|
3980 |
+
"step": 565
|
3981 |
+
},
|
3982 |
+
{
|
3983 |
+
"epoch": 0.011393258653139688,
|
3984 |
+
"grad_norm": 0.00016836596478242427,
|
3985 |
+
"learning_rate": 0.00012181195295538895,
|
3986 |
+
"loss": 46.0,
|
3987 |
+
"step": 566
|
3988 |
+
},
|
3989 |
+
{
|
3990 |
+
"epoch": 0.011413388085389052,
|
3991 |
+
"grad_norm": 0.00024714317987672985,
|
3992 |
+
"learning_rate": 0.00012157476973592842,
|
3993 |
+
"loss": 46.0,
|
3994 |
+
"step": 567
|
3995 |
+
},
|
3996 |
+
{
|
3997 |
+
"epoch": 0.011433517517638415,
|
3998 |
+
"grad_norm": 0.0002926415763795376,
|
3999 |
+
"learning_rate": 0.00012133745915189278,
|
4000 |
+
"loss": 46.0,
|
4001 |
+
"step": 568
|
4002 |
+
},
|
4003 |
+
{
|
4004 |
+
"epoch": 0.01145364694988778,
|
4005 |
+
"grad_norm": 0.00012063339818269014,
|
4006 |
+
"learning_rate": 0.00012110002260422218,
|
4007 |
+
"loss": 46.0,
|
4008 |
+
"step": 569
|
4009 |
+
},
|
4010 |
+
{
|
4011 |
+
"epoch": 0.011473776382137141,
|
4012 |
+
"grad_norm": 0.00014082356938160956,
|
4013 |
+
"learning_rate": 0.00012086246149460038,
|
4014 |
+
"loss": 46.0,
|
4015 |
+
"step": 570
|
4016 |
+
},
|
4017 |
+
{
|
4018 |
+
"epoch": 0.011493905814386505,
|
4019 |
+
"grad_norm": 0.0002749360864982009,
|
4020 |
+
"learning_rate": 0.00012062477722544656,
|
4021 |
+
"loss": 46.0,
|
4022 |
+
"step": 571
|
4023 |
+
},
|
4024 |
+
{
|
4025 |
+
"epoch": 0.011514035246635869,
|
4026 |
+
"grad_norm": 0.00013284625310916454,
|
4027 |
+
"learning_rate": 0.00012038697119990687,
|
4028 |
+
"loss": 46.0,
|
4029 |
+
"step": 572
|
4030 |
+
},
|
4031 |
+
{
|
4032 |
+
"epoch": 0.011534164678885232,
|
4033 |
+
"grad_norm": 8.731486741453409e-05,
|
4034 |
+
"learning_rate": 0.00012014904482184633,
|
4035 |
+
"loss": 46.0,
|
4036 |
+
"step": 573
|
4037 |
+
},
|
4038 |
+
{
|
4039 |
+
"epoch": 0.011554294111134596,
|
4040 |
+
"grad_norm": 7.719992572674528e-05,
|
4041 |
+
"learning_rate": 0.00011991099949584032,
|
4042 |
+
"loss": 46.0,
|
4043 |
+
"step": 574
|
4044 |
+
},
|
4045 |
+
{
|
4046 |
+
"epoch": 0.01157442354338396,
|
4047 |
+
"grad_norm": 0.00010548291902523488,
|
4048 |
+
"learning_rate": 0.00011967283662716653,
|
4049 |
+
"loss": 46.0,
|
4050 |
+
"step": 575
|
4051 |
+
},
|
4052 |
+
{
|
4053 |
+
"epoch": 0.011594552975633322,
|
4054 |
+
"grad_norm": 0.0002630742092151195,
|
4055 |
+
"learning_rate": 0.00011943455762179654,
|
4056 |
+
"loss": 46.0,
|
4057 |
+
"step": 576
|
4058 |
+
},
|
4059 |
+
{
|
4060 |
+
"epoch": 0.011614682407882685,
|
4061 |
+
"grad_norm": 0.00016908656107261777,
|
4062 |
+
"learning_rate": 0.00011919616388638748,
|
4063 |
+
"loss": 46.0,
|
4064 |
+
"step": 577
|
4065 |
+
},
|
4066 |
+
{
|
4067 |
+
"epoch": 0.011634811840132049,
|
4068 |
+
"grad_norm": 0.00010908886906690896,
|
4069 |
+
"learning_rate": 0.0001189576568282738,
|
4070 |
+
"loss": 46.0,
|
4071 |
+
"step": 578
|
4072 |
+
},
|
4073 |
+
{
|
4074 |
+
"epoch": 0.011654941272381413,
|
4075 |
+
"grad_norm": 8.737723692320287e-05,
|
4076 |
+
"learning_rate": 0.00011871903785545897,
|
4077 |
+
"loss": 46.0,
|
4078 |
+
"step": 579
|
4079 |
+
},
|
4080 |
+
{
|
4081 |
+
"epoch": 0.011675070704630777,
|
4082 |
+
"grad_norm": 0.00011381749209249392,
|
4083 |
+
"learning_rate": 0.00011848030837660709,
|
4084 |
+
"loss": 46.0,
|
4085 |
+
"step": 580
|
4086 |
+
},
|
4087 |
+
{
|
4088 |
+
"epoch": 0.011695200136880139,
|
4089 |
+
"grad_norm": 0.00023508115555159748,
|
4090 |
+
"learning_rate": 0.00011824146980103467,
|
4091 |
+
"loss": 46.0,
|
4092 |
+
"step": 581
|
4093 |
+
},
|
4094 |
+
{
|
4095 |
+
"epoch": 0.011715329569129502,
|
4096 |
+
"grad_norm": 0.0002138228010153398,
|
4097 |
+
"learning_rate": 0.00011800252353870224,
|
4098 |
+
"loss": 46.0,
|
4099 |
+
"step": 582
|
4100 |
+
},
|
4101 |
+
{
|
4102 |
+
"epoch": 0.011735459001378866,
|
4103 |
+
"grad_norm": 0.0002100839774357155,
|
4104 |
+
"learning_rate": 0.00011776347100020602,
|
4105 |
+
"loss": 46.0,
|
4106 |
+
"step": 583
|
4107 |
+
},
|
4108 |
+
{
|
4109 |
+
"epoch": 0.01175558843362823,
|
4110 |
+
"grad_norm": 0.00025784672470763326,
|
4111 |
+
"learning_rate": 0.00011752431359676968,
|
4112 |
+
"loss": 46.0,
|
4113 |
+
"step": 584
|
4114 |
+
},
|
4115 |
+
{
|
4116 |
+
"epoch": 0.011775717865877593,
|
4117 |
+
"grad_norm": 9.872866212390363e-05,
|
4118 |
+
"learning_rate": 0.00011728505274023584,
|
4119 |
+
"loss": 46.0,
|
4120 |
+
"step": 585
|
4121 |
+
},
|
4122 |
+
{
|
4123 |
+
"epoch": 0.011795847298126957,
|
4124 |
+
"grad_norm": 8.426361455349252e-05,
|
4125 |
+
"learning_rate": 0.00011704568984305802,
|
4126 |
+
"loss": 46.0,
|
4127 |
+
"step": 586
|
4128 |
+
},
|
4129 |
+
{
|
4130 |
+
"epoch": 0.011815976730376319,
|
4131 |
+
"grad_norm": 0.0003703351248987019,
|
4132 |
+
"learning_rate": 0.00011680622631829197,
|
4133 |
+
"loss": 46.0,
|
4134 |
+
"step": 587
|
4135 |
+
},
|
4136 |
+
{
|
4137 |
+
"epoch": 0.011836106162625683,
|
4138 |
+
"grad_norm": 0.00012765347491949797,
|
4139 |
+
"learning_rate": 0.00011656666357958751,
|
4140 |
+
"loss": 46.0,
|
4141 |
+
"step": 588
|
4142 |
+
},
|
4143 |
+
{
|
4144 |
+
"epoch": 0.011856235594875046,
|
4145 |
+
"grad_norm": 0.00010474120790604502,
|
4146 |
+
"learning_rate": 0.00011632700304118032,
|
4147 |
+
"loss": 46.0,
|
4148 |
+
"step": 589
|
4149 |
+
},
|
4150 |
+
{
|
4151 |
+
"epoch": 0.01187636502712441,
|
4152 |
+
"grad_norm": 0.00026702586910687387,
|
4153 |
+
"learning_rate": 0.0001160872461178832,
|
4154 |
+
"loss": 46.0,
|
4155 |
+
"step": 590
|
4156 |
+
},
|
4157 |
+
{
|
4158 |
+
"epoch": 0.011896494459373774,
|
4159 |
+
"grad_norm": 0.00018060579895973206,
|
4160 |
+
"learning_rate": 0.00011584739422507804,
|
4161 |
+
"loss": 46.0,
|
4162 |
+
"step": 591
|
4163 |
+
},
|
4164 |
+
{
|
4165 |
+
"epoch": 0.011916623891623138,
|
4166 |
+
"grad_norm": 0.0004400400212034583,
|
4167 |
+
"learning_rate": 0.00011560744877870748,
|
4168 |
+
"loss": 46.0,
|
4169 |
+
"step": 592
|
4170 |
+
},
|
4171 |
+
{
|
4172 |
+
"epoch": 0.0119367533238725,
|
4173 |
+
"grad_norm": 0.00011154530511703342,
|
4174 |
+
"learning_rate": 0.00011536741119526628,
|
4175 |
+
"loss": 46.0,
|
4176 |
+
"step": 593
|
4177 |
+
},
|
4178 |
+
{
|
4179 |
+
"epoch": 0.011956882756121863,
|
4180 |
+
"grad_norm": 9.707292338134721e-05,
|
4181 |
+
"learning_rate": 0.00011512728289179323,
|
4182 |
+
"loss": 46.0,
|
4183 |
+
"step": 594
|
4184 |
+
},
|
4185 |
+
{
|
4186 |
+
"epoch": 0.011977012188371227,
|
4187 |
+
"grad_norm": 0.00012266647536307573,
|
4188 |
+
"learning_rate": 0.00011488706528586261,
|
4189 |
+
"loss": 46.0,
|
4190 |
+
"step": 595
|
4191 |
+
},
|
4192 |
+
{
|
4193 |
+
"epoch": 0.01199714162062059,
|
4194 |
+
"grad_norm": 6.54510804452002e-05,
|
4195 |
+
"learning_rate": 0.00011464675979557593,
|
4196 |
+
"loss": 46.0,
|
4197 |
+
"step": 596
|
4198 |
+
},
|
4199 |
+
{
|
4200 |
+
"epoch": 0.012017271052869954,
|
4201 |
+
"grad_norm": 0.00019303473527543247,
|
4202 |
+
"learning_rate": 0.00011440636783955356,
|
4203 |
+
"loss": 46.0,
|
4204 |
+
"step": 597
|
4205 |
+
},
|
4206 |
+
{
|
4207 |
+
"epoch": 0.012037400485119316,
|
4208 |
+
"grad_norm": 0.00021508029021788388,
|
4209 |
+
"learning_rate": 0.00011416589083692619,
|
4210 |
+
"loss": 46.0,
|
4211 |
+
"step": 598
|
4212 |
+
},
|
4213 |
+
{
|
4214 |
+
"epoch": 0.01205752991736868,
|
4215 |
+
"grad_norm": 0.00014299601025413722,
|
4216 |
+
"learning_rate": 0.00011392533020732666,
|
4217 |
+
"loss": 46.0,
|
4218 |
+
"step": 599
|
4219 |
+
},
|
4220 |
+
{
|
4221 |
+
"epoch": 0.012077659349618044,
|
4222 |
+
"grad_norm": 0.00017883341934066266,
|
4223 |
+
"learning_rate": 0.00011368468737088148,
|
4224 |
+
"loss": 46.0,
|
4225 |
+
"step": 600
|
4226 |
+
},
|
4227 |
+
{
|
4228 |
+
"epoch": 0.012097788781867408,
|
4229 |
+
"grad_norm": 0.0002254635328426957,
|
4230 |
+
"learning_rate": 0.00011344396374820244,
|
4231 |
+
"loss": 46.0,
|
4232 |
+
"step": 601
|
4233 |
+
},
|
4234 |
+
{
|
4235 |
+
"epoch": 0.012117918214116771,
|
4236 |
+
"grad_norm": 0.0002371317968936637,
|
4237 |
+
"learning_rate": 0.0001132031607603783,
|
4238 |
+
"loss": 46.0,
|
4239 |
+
"step": 602
|
4240 |
+
},
|
4241 |
+
{
|
4242 |
+
"epoch": 0.012138047646366135,
|
4243 |
+
"grad_norm": 0.00012665463145822287,
|
4244 |
+
"learning_rate": 0.0001129622798289663,
|
4245 |
+
"loss": 46.0,
|
4246 |
+
"step": 603
|
4247 |
+
},
|
4248 |
+
{
|
4249 |
+
"epoch": 0.012158177078615497,
|
4250 |
+
"grad_norm": 0.00019802236056420952,
|
4251 |
+
"learning_rate": 0.00011272132237598376,
|
4252 |
+
"loss": 46.0,
|
4253 |
+
"step": 604
|
4254 |
+
},
|
4255 |
+
{
|
4256 |
+
"epoch": 0.01217830651086486,
|
4257 |
+
"grad_norm": 0.00011026608990505338,
|
4258 |
+
"learning_rate": 0.00011248028982389989,
|
4259 |
+
"loss": 46.0,
|
4260 |
+
"step": 605
|
4261 |
+
},
|
4262 |
+
{
|
4263 |
+
"epoch": 0.012198435943114224,
|
4264 |
+
"grad_norm": 9.928762301569805e-05,
|
4265 |
+
"learning_rate": 0.00011223918359562708,
|
4266 |
+
"loss": 46.0,
|
4267 |
+
"step": 606
|
4268 |
+
},
|
4269 |
+
{
|
4270 |
+
"epoch": 0.012218565375363588,
|
4271 |
+
"grad_norm": 0.00023735944705549628,
|
4272 |
+
"learning_rate": 0.00011199800511451273,
|
4273 |
+
"loss": 46.0,
|
4274 |
+
"step": 607
|
4275 |
+
},
|
4276 |
+
{
|
4277 |
+
"epoch": 0.012238694807612952,
|
4278 |
+
"grad_norm": 0.00011722726048901677,
|
4279 |
+
"learning_rate": 0.0001117567558043308,
|
4280 |
+
"loss": 46.0,
|
4281 |
+
"step": 608
|
4282 |
+
},
|
4283 |
+
{
|
4284 |
+
"epoch": 0.012258824239862316,
|
4285 |
+
"grad_norm": 0.0003526516375131905,
|
4286 |
+
"learning_rate": 0.00011151543708927335,
|
4287 |
+
"loss": 46.0,
|
4288 |
+
"step": 609
|
4289 |
+
},
|
4290 |
+
{
|
4291 |
+
"epoch": 0.012278953672111678,
|
4292 |
+
"grad_norm": 0.00022850584355182946,
|
4293 |
+
"learning_rate": 0.00011127405039394216,
|
4294 |
+
"loss": 46.0,
|
4295 |
+
"step": 610
|
4296 |
+
},
|
4297 |
+
{
|
4298 |
+
"epoch": 0.012299083104361041,
|
4299 |
+
"grad_norm": 0.0005510961636900902,
|
4300 |
+
"learning_rate": 0.00011103259714334034,
|
4301 |
+
"loss": 46.0,
|
4302 |
+
"step": 611
|
4303 |
+
},
|
4304 |
+
{
|
4305 |
+
"epoch": 0.012319212536610405,
|
4306 |
+
"grad_norm": 0.00013431145634967834,
|
4307 |
+
"learning_rate": 0.00011079107876286387,
|
4308 |
+
"loss": 46.0,
|
4309 |
+
"step": 612
|
4310 |
+
},
|
4311 |
+
{
|
4312 |
+
"epoch": 0.012339341968859769,
|
4313 |
+
"grad_norm": 0.0001544607657706365,
|
4314 |
+
"learning_rate": 0.0001105494966782933,
|
4315 |
+
"loss": 46.0,
|
4316 |
+
"step": 613
|
4317 |
+
},
|
4318 |
+
{
|
4319 |
+
"epoch": 0.012359471401109132,
|
4320 |
+
"grad_norm": 8.791110303718597e-05,
|
4321 |
+
"learning_rate": 0.0001103078523157852,
|
4322 |
+
"loss": 46.0,
|
4323 |
+
"step": 614
|
4324 |
+
},
|
4325 |
+
{
|
4326 |
+
"epoch": 0.012379600833358494,
|
4327 |
+
"grad_norm": 0.00023833720479160547,
|
4328 |
+
"learning_rate": 0.00011006614710186372,
|
4329 |
+
"loss": 46.0,
|
4330 |
+
"step": 615
|
4331 |
+
},
|
4332 |
+
{
|
4333 |
+
"epoch": 0.012399730265607858,
|
4334 |
+
"grad_norm": 0.00016108158160932362,
|
4335 |
+
"learning_rate": 0.00010982438246341238,
|
4336 |
+
"loss": 46.0,
|
4337 |
+
"step": 616
|
4338 |
+
},
|
4339 |
+
{
|
4340 |
+
"epoch": 0.012419859697857222,
|
4341 |
+
"grad_norm": 0.00022613555483985692,
|
4342 |
+
"learning_rate": 0.00010958255982766538,
|
4343 |
+
"loss": 46.0,
|
4344 |
+
"step": 617
|
4345 |
+
},
|
4346 |
+
{
|
4347 |
+
"epoch": 0.012439989130106585,
|
4348 |
+
"grad_norm": 0.00022462922788690776,
|
4349 |
+
"learning_rate": 0.00010934068062219945,
|
4350 |
+
"loss": 46.0,
|
4351 |
+
"step": 618
|
4352 |
+
},
|
4353 |
+
{
|
4354 |
+
"epoch": 0.01246011856235595,
|
4355 |
+
"grad_norm": 0.00031467695953324437,
|
4356 |
+
"learning_rate": 0.0001090987462749251,
|
4357 |
+
"loss": 46.0,
|
4358 |
+
"step": 619
|
4359 |
+
},
|
4360 |
+
{
|
4361 |
+
"epoch": 0.012480247994605313,
|
4362 |
+
"grad_norm": 0.0002814480976667255,
|
4363 |
+
"learning_rate": 0.00010885675821407844,
|
4364 |
+
"loss": 46.0,
|
4365 |
+
"step": 620
|
4366 |
+
},
|
4367 |
+
{
|
4368 |
+
"epoch": 0.012500377426854675,
|
4369 |
+
"grad_norm": 0.0001427562237950042,
|
4370 |
+
"learning_rate": 0.00010861471786821275,
|
4371 |
+
"loss": 46.0,
|
4372 |
+
"step": 621
|
4373 |
+
},
|
4374 |
+
{
|
4375 |
+
"epoch": 0.012520506859104039,
|
4376 |
+
"grad_norm": 0.00010559640941210091,
|
4377 |
+
"learning_rate": 0.00010837262666618983,
|
4378 |
+
"loss": 46.0,
|
4379 |
+
"step": 622
|
4380 |
+
},
|
4381 |
+
{
|
4382 |
+
"epoch": 0.012540636291353402,
|
4383 |
+
"grad_norm": 0.00022459396859630942,
|
4384 |
+
"learning_rate": 0.00010813048603717182,
|
4385 |
+
"loss": 46.0,
|
4386 |
+
"step": 623
|
4387 |
+
},
|
4388 |
+
{
|
4389 |
+
"epoch": 0.012560765723602766,
|
4390 |
+
"grad_norm": 0.0003838833072222769,
|
4391 |
+
"learning_rate": 0.0001078882974106126,
|
4392 |
+
"loss": 46.0,
|
4393 |
+
"step": 624
|
4394 |
+
},
|
4395 |
+
{
|
4396 |
+
"epoch": 0.01258089515585213,
|
4397 |
+
"grad_norm": 0.00013236599625088274,
|
4398 |
+
"learning_rate": 0.00010764606221624933,
|
4399 |
+
"loss": 46.0,
|
4400 |
+
"step": 625
|
4401 |
+
},
|
4402 |
+
{
|
4403 |
+
"epoch": 0.012601024588101493,
|
4404 |
+
"grad_norm": 0.0002815214393194765,
|
4405 |
+
"learning_rate": 0.00010740378188409426,
|
4406 |
+
"loss": 46.0,
|
4407 |
+
"step": 626
|
4408 |
+
},
|
4409 |
+
{
|
4410 |
+
"epoch": 0.012621154020350855,
|
4411 |
+
"grad_norm": 9.716377826407552e-05,
|
4412 |
+
"learning_rate": 0.00010716145784442593,
|
4413 |
+
"loss": 46.0,
|
4414 |
+
"step": 627
|
4415 |
+
},
|
4416 |
+
{
|
4417 |
+
"epoch": 0.01264128345260022,
|
4418 |
+
"grad_norm": 0.00029817328322678804,
|
4419 |
+
"learning_rate": 0.00010691909152778094,
|
4420 |
+
"loss": 46.0,
|
4421 |
+
"step": 628
|
4422 |
+
},
|
4423 |
+
{
|
4424 |
+
"epoch": 0.012661412884849583,
|
4425 |
+
"grad_norm": 0.00011512849596329033,
|
4426 |
+
"learning_rate": 0.00010667668436494558,
|
4427 |
+
"loss": 46.0,
|
4428 |
+
"step": 629
|
4429 |
+
},
|
4430 |
+
{
|
4431 |
+
"epoch": 0.012681542317098947,
|
4432 |
+
"grad_norm": 0.00021800924150738865,
|
4433 |
+
"learning_rate": 0.00010643423778694712,
|
4434 |
+
"loss": 46.0,
|
4435 |
+
"step": 630
|
4436 |
+
},
|
4437 |
+
{
|
4438 |
+
"epoch": 0.01270167174934831,
|
4439 |
+
"grad_norm": 0.00012208014959469438,
|
4440 |
+
"learning_rate": 0.0001061917532250456,
|
4441 |
+
"loss": 46.0,
|
4442 |
+
"step": 631
|
4443 |
+
},
|
4444 |
+
{
|
4445 |
+
"epoch": 0.012721801181597672,
|
4446 |
+
"grad_norm": 0.00015742589312139899,
|
4447 |
+
"learning_rate": 0.00010594923211072532,
|
4448 |
+
"loss": 46.0,
|
4449 |
+
"step": 632
|
4450 |
+
},
|
4451 |
+
{
|
4452 |
+
"epoch": 0.012741930613847036,
|
4453 |
+
"grad_norm": 0.00029806559905409813,
|
4454 |
+
"learning_rate": 0.00010570667587568626,
|
4455 |
+
"loss": 46.0,
|
4456 |
+
"step": 633
|
4457 |
+
},
|
4458 |
+
{
|
4459 |
+
"epoch": 0.0127620600460964,
|
4460 |
+
"grad_norm": 0.00018120172899216413,
|
4461 |
+
"learning_rate": 0.00010546408595183578,
|
4462 |
+
"loss": 46.0,
|
4463 |
+
"step": 634
|
4464 |
+
},
|
4465 |
+
{
|
4466 |
+
"epoch": 0.012782189478345763,
|
4467 |
+
"grad_norm": 0.00016689879703335464,
|
4468 |
+
"learning_rate": 0.00010522146377128021,
|
4469 |
+
"loss": 46.0,
|
4470 |
+
"step": 635
|
4471 |
+
},
|
4472 |
+
{
|
4473 |
+
"epoch": 0.012802318910595127,
|
4474 |
+
"grad_norm": 0.000355700176442042,
|
4475 |
+
"learning_rate": 0.00010497881076631615,
|
4476 |
+
"loss": 46.0,
|
4477 |
+
"step": 636
|
4478 |
+
},
|
4479 |
+
{
|
4480 |
+
"epoch": 0.01282244834284449,
|
4481 |
+
"grad_norm": 0.0001786172651918605,
|
4482 |
+
"learning_rate": 0.00010473612836942226,
|
4483 |
+
"loss": 46.0,
|
4484 |
+
"step": 637
|
4485 |
+
},
|
4486 |
+
{
|
4487 |
+
"epoch": 0.012842577775093853,
|
4488 |
+
"grad_norm": 0.0003312894550617784,
|
4489 |
+
"learning_rate": 0.00010449341801325073,
|
4490 |
+
"loss": 46.0,
|
4491 |
+
"step": 638
|
4492 |
+
},
|
4493 |
+
{
|
4494 |
+
"epoch": 0.012862707207343217,
|
4495 |
+
"grad_norm": 0.00023194189998321235,
|
4496 |
+
"learning_rate": 0.00010425068113061873,
|
4497 |
+
"loss": 46.0,
|
4498 |
+
"step": 639
|
4499 |
+
},
|
4500 |
+
{
|
4501 |
+
"epoch": 0.01288283663959258,
|
4502 |
+
"grad_norm": 0.0004088008718099445,
|
4503 |
+
"learning_rate": 0.00010400791915450009,
|
4504 |
+
"loss": 46.0,
|
4505 |
+
"step": 640
|
4506 |
+
},
|
4507 |
+
{
|
4508 |
+
"epoch": 0.012902966071841944,
|
4509 |
+
"grad_norm": 0.0001682123402133584,
|
4510 |
+
"learning_rate": 0.00010376513351801673,
|
4511 |
+
"loss": 46.0,
|
4512 |
+
"step": 641
|
4513 |
+
},
|
4514 |
+
{
|
4515 |
+
"epoch": 0.012923095504091308,
|
4516 |
+
"grad_norm": 0.00013815666898153722,
|
4517 |
+
"learning_rate": 0.00010352232565443032,
|
4518 |
+
"loss": 46.0,
|
4519 |
+
"step": 642
|
4520 |
+
},
|
4521 |
+
{
|
4522 |
+
"epoch": 0.012943224936340671,
|
4523 |
+
"grad_norm": 0.0002450251195114106,
|
4524 |
+
"learning_rate": 0.00010327949699713366,
|
4525 |
+
"loss": 46.0,
|
4526 |
+
"step": 643
|
4527 |
+
},
|
4528 |
+
{
|
4529 |
+
"epoch": 0.012963354368590033,
|
4530 |
+
"grad_norm": 0.00019522267393767834,
|
4531 |
+
"learning_rate": 0.00010303664897964232,
|
4532 |
+
"loss": 46.0,
|
4533 |
+
"step": 644
|
4534 |
+
},
|
4535 |
+
{
|
4536 |
+
"epoch": 0.012983483800839397,
|
4537 |
+
"grad_norm": 0.00013197500084061176,
|
4538 |
+
"learning_rate": 0.00010279378303558624,
|
4539 |
+
"loss": 46.0,
|
4540 |
+
"step": 645
|
4541 |
+
},
|
4542 |
+
{
|
4543 |
+
"epoch": 0.01300361323308876,
|
4544 |
+
"grad_norm": 0.00034008765942417085,
|
4545 |
+
"learning_rate": 0.00010255090059870107,
|
4546 |
+
"loss": 46.0,
|
4547 |
+
"step": 646
|
4548 |
+
},
|
4549 |
+
{
|
4550 |
+
"epoch": 0.013023742665338124,
|
4551 |
+
"grad_norm": 0.00030950226937420666,
|
4552 |
+
"learning_rate": 0.00010230800310281992,
|
4553 |
+
"loss": 46.0,
|
4554 |
+
"step": 647
|
4555 |
+
},
|
4556 |
+
{
|
4557 |
+
"epoch": 0.013043872097587488,
|
4558 |
+
"grad_norm": 0.0002216809953097254,
|
4559 |
+
"learning_rate": 0.00010206509198186476,
|
4560 |
+
"loss": 46.0,
|
4561 |
+
"step": 648
|
4562 |
+
},
|
4563 |
+
{
|
4564 |
+
"epoch": 0.01306400152983685,
|
4565 |
+
"grad_norm": 0.0005021773395128548,
|
4566 |
+
"learning_rate": 0.00010182216866983796,
|
4567 |
+
"loss": 46.0,
|
4568 |
+
"step": 649
|
4569 |
+
},
|
4570 |
+
{
|
4571 |
+
"epoch": 0.013084130962086214,
|
4572 |
+
"grad_norm": 0.0003279381198808551,
|
4573 |
+
"learning_rate": 0.00010157923460081394,
|
4574 |
+
"loss": 46.0,
|
4575 |
+
"step": 650
|
4576 |
+
},
|
4577 |
+
{
|
4578 |
+
"epoch": 0.013104260394335578,
|
4579 |
+
"grad_norm": 0.000169211023603566,
|
4580 |
+
"learning_rate": 0.00010133629120893055,
|
4581 |
+
"loss": 46.0,
|
4582 |
+
"step": 651
|
4583 |
+
},
|
4584 |
+
{
|
4585 |
+
"epoch": 0.013124389826584941,
|
4586 |
+
"grad_norm": 0.0004348910879343748,
|
4587 |
+
"learning_rate": 0.00010109333992838072,
|
4588 |
+
"loss": 46.0,
|
4589 |
+
"step": 652
|
4590 |
+
},
|
4591 |
+
{
|
4592 |
+
"epoch": 0.013124389826584941,
|
4593 |
+
"eval_loss": 11.5,
|
4594 |
+
"eval_runtime": 126.4432,
|
4595 |
+
"eval_samples_per_second": 165.434,
|
4596 |
+
"eval_steps_per_second": 82.717,
|
4597 |
+
"step": 652
|
4598 |
}
|
4599 |
],
|
4600 |
"logging_steps": 1,
|
|
|
4614 |
"attributes": {}
|
4615 |
}
|
4616 |
},
|
4617 |
+
"total_flos": 15091580436480.0,
|
4618 |
"train_batch_size": 2,
|
4619 |
"trial_name": null,
|
4620 |
"trial_params": null
|