Training in progress, step 1537, 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 125040
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:063f306e45e8b39d154dbc7bfc68fde278c659f3c2883d2758099ee159ff24b5
|
3 |
size 125040
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 162868
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:25575660cf81bc26c2811396f3d2d48bbf015d309d37ebe2067b66a905a05767
|
3 |
size 162868
|
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 sha256:2c5a44562b89c59a7cfb08b5955ea23594890c2c99652b9f1f1d8d78a0b9bded
|
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 sha256:0ee8dee9cedcd1fe2ff7ecd82756d62d8c60cb522ecedc8de6c7ca59c472201d
|
3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
@@ -1,9 +1,9 @@
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
-
"epoch": 0.
|
5 |
"eval_steps": 385,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -8124,6 +8124,2680 @@
|
|
8124 |
"eval_samples_per_second": 380.835,
|
8125 |
"eval_steps_per_second": 190.462,
|
8126 |
"step": 1155
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
8127 |
}
|
8128 |
],
|
8129 |
"logging_steps": 1,
|
@@ -8138,12 +10812,12 @@
|
|
8138 |
"should_evaluate": false,
|
8139 |
"should_log": false,
|
8140 |
"should_save": true,
|
8141 |
-
"should_training_stop":
|
8142 |
},
|
8143 |
"attributes": {}
|
8144 |
}
|
8145 |
},
|
8146 |
-
"total_flos":
|
8147 |
"train_batch_size": 2,
|
8148 |
"trial_name": null,
|
8149 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.15244991073199762,
|
5 |
"eval_steps": 385,
|
6 |
+
"global_step": 1537,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
8124 |
"eval_samples_per_second": 380.835,
|
8125 |
"eval_steps_per_second": 190.462,
|
8126 |
"step": 1155
|
8127 |
+
},
|
8128 |
+
{
|
8129 |
+
"epoch": 0.11465978972426107,
|
8130 |
+
"grad_norm": 1.9147896766662598,
|
8131 |
+
"learning_rate": 2.9180297975576364e-05,
|
8132 |
+
"loss": 43.2374,
|
8133 |
+
"step": 1156
|
8134 |
+
},
|
8135 |
+
{
|
8136 |
+
"epoch": 0.1147589763935727,
|
8137 |
+
"grad_norm": 1.8915377855300903,
|
8138 |
+
"learning_rate": 2.9035195103899825e-05,
|
8139 |
+
"loss": 43.0964,
|
8140 |
+
"step": 1157
|
8141 |
+
},
|
8142 |
+
{
|
8143 |
+
"epoch": 0.11485816306288435,
|
8144 |
+
"grad_norm": 2.041377305984497,
|
8145 |
+
"learning_rate": 2.889039260774018e-05,
|
8146 |
+
"loss": 43.2494,
|
8147 |
+
"step": 1158
|
8148 |
+
},
|
8149 |
+
{
|
8150 |
+
"epoch": 0.114957349732196,
|
8151 |
+
"grad_norm": 2.3205907344818115,
|
8152 |
+
"learning_rate": 2.8745891100008683e-05,
|
8153 |
+
"loss": 43.4031,
|
8154 |
+
"step": 1159
|
8155 |
+
},
|
8156 |
+
{
|
8157 |
+
"epoch": 0.11505653640150763,
|
8158 |
+
"grad_norm": 2.3421196937561035,
|
8159 |
+
"learning_rate": 2.8601691192342493e-05,
|
8160 |
+
"loss": 43.2697,
|
8161 |
+
"step": 1160
|
8162 |
+
},
|
8163 |
+
{
|
8164 |
+
"epoch": 0.11515572307081928,
|
8165 |
+
"grad_norm": 2.095224380493164,
|
8166 |
+
"learning_rate": 2.845779349510227e-05,
|
8167 |
+
"loss": 43.666,
|
8168 |
+
"step": 1161
|
8169 |
+
},
|
8170 |
+
{
|
8171 |
+
"epoch": 0.11525490974013093,
|
8172 |
+
"grad_norm": 2.259864568710327,
|
8173 |
+
"learning_rate": 2.83141986173694e-05,
|
8174 |
+
"loss": 42.9196,
|
8175 |
+
"step": 1162
|
8176 |
+
},
|
8177 |
+
{
|
8178 |
+
"epoch": 0.11535409640944257,
|
8179 |
+
"grad_norm": 2.165755033493042,
|
8180 |
+
"learning_rate": 2.817090716694363e-05,
|
8181 |
+
"loss": 43.5717,
|
8182 |
+
"step": 1163
|
8183 |
+
},
|
8184 |
+
{
|
8185 |
+
"epoch": 0.11545328307875422,
|
8186 |
+
"grad_norm": 2.1785712242126465,
|
8187 |
+
"learning_rate": 2.802791975034024e-05,
|
8188 |
+
"loss": 43.2673,
|
8189 |
+
"step": 1164
|
8190 |
+
},
|
8191 |
+
{
|
8192 |
+
"epoch": 0.11555246974806586,
|
8193 |
+
"grad_norm": 1.9809575080871582,
|
8194 |
+
"learning_rate": 2.788523697278773e-05,
|
8195 |
+
"loss": 43.2572,
|
8196 |
+
"step": 1165
|
8197 |
+
},
|
8198 |
+
{
|
8199 |
+
"epoch": 0.1156516564173775,
|
8200 |
+
"grad_norm": 1.791969656944275,
|
8201 |
+
"learning_rate": 2.7742859438225055e-05,
|
8202 |
+
"loss": 43.4034,
|
8203 |
+
"step": 1166
|
8204 |
+
},
|
8205 |
+
{
|
8206 |
+
"epoch": 0.11575084308668915,
|
8207 |
+
"grad_norm": 2.295616388320923,
|
8208 |
+
"learning_rate": 2.7600787749299263e-05,
|
8209 |
+
"loss": 43.4226,
|
8210 |
+
"step": 1167
|
8211 |
+
},
|
8212 |
+
{
|
8213 |
+
"epoch": 0.1158500297560008,
|
8214 |
+
"grad_norm": 1.8220045566558838,
|
8215 |
+
"learning_rate": 2.7459022507362686e-05,
|
8216 |
+
"loss": 43.349,
|
8217 |
+
"step": 1168
|
8218 |
+
},
|
8219 |
+
{
|
8220 |
+
"epoch": 0.11594921642531243,
|
8221 |
+
"grad_norm": 1.9346619844436646,
|
8222 |
+
"learning_rate": 2.731756431247072e-05,
|
8223 |
+
"loss": 43.2268,
|
8224 |
+
"step": 1169
|
8225 |
+
},
|
8226 |
+
{
|
8227 |
+
"epoch": 0.11604840309462408,
|
8228 |
+
"grad_norm": 1.8838238716125488,
|
8229 |
+
"learning_rate": 2.717641376337895e-05,
|
8230 |
+
"loss": 43.0441,
|
8231 |
+
"step": 1170
|
8232 |
+
},
|
8233 |
+
{
|
8234 |
+
"epoch": 0.11614758976393573,
|
8235 |
+
"grad_norm": 2.1250696182250977,
|
8236 |
+
"learning_rate": 2.7035571457540865e-05,
|
8237 |
+
"loss": 43.1962,
|
8238 |
+
"step": 1171
|
8239 |
+
},
|
8240 |
+
{
|
8241 |
+
"epoch": 0.11624677643324736,
|
8242 |
+
"grad_norm": 1.7829358577728271,
|
8243 |
+
"learning_rate": 2.6895037991105286e-05,
|
8244 |
+
"loss": 43.4272,
|
8245 |
+
"step": 1172
|
8246 |
+
},
|
8247 |
+
{
|
8248 |
+
"epoch": 0.11634596310255901,
|
8249 |
+
"grad_norm": 1.9922783374786377,
|
8250 |
+
"learning_rate": 2.675481395891365e-05,
|
8251 |
+
"loss": 43.112,
|
8252 |
+
"step": 1173
|
8253 |
+
},
|
8254 |
+
{
|
8255 |
+
"epoch": 0.11644514977187066,
|
8256 |
+
"grad_norm": 2.087114095687866,
|
8257 |
+
"learning_rate": 2.6614899954497795e-05,
|
8258 |
+
"loss": 43.0554,
|
8259 |
+
"step": 1174
|
8260 |
+
},
|
8261 |
+
{
|
8262 |
+
"epoch": 0.11654433644118231,
|
8263 |
+
"grad_norm": 2.154203414916992,
|
8264 |
+
"learning_rate": 2.6475296570077158e-05,
|
8265 |
+
"loss": 43.0469,
|
8266 |
+
"step": 1175
|
8267 |
+
},
|
8268 |
+
{
|
8269 |
+
"epoch": 0.11664352311049395,
|
8270 |
+
"grad_norm": 2.1721832752227783,
|
8271 |
+
"learning_rate": 2.6336004396556534e-05,
|
8272 |
+
"loss": 43.5545,
|
8273 |
+
"step": 1176
|
8274 |
+
},
|
8275 |
+
{
|
8276 |
+
"epoch": 0.1167427097798056,
|
8277 |
+
"grad_norm": 1.7544825077056885,
|
8278 |
+
"learning_rate": 2.619702402352332e-05,
|
8279 |
+
"loss": 43.5061,
|
8280 |
+
"step": 1177
|
8281 |
+
},
|
8282 |
+
{
|
8283 |
+
"epoch": 0.11684189644911724,
|
8284 |
+
"grad_norm": 2.276806592941284,
|
8285 |
+
"learning_rate": 2.6058356039245246e-05,
|
8286 |
+
"loss": 43.3631,
|
8287 |
+
"step": 1178
|
8288 |
+
},
|
8289 |
+
{
|
8290 |
+
"epoch": 0.11694108311842888,
|
8291 |
+
"grad_norm": 1.8149032592773438,
|
8292 |
+
"learning_rate": 2.5920001030667684e-05,
|
8293 |
+
"loss": 43.539,
|
8294 |
+
"step": 1179
|
8295 |
+
},
|
8296 |
+
{
|
8297 |
+
"epoch": 0.11704026978774053,
|
8298 |
+
"grad_norm": 2.0274498462677,
|
8299 |
+
"learning_rate": 2.5781959583411374e-05,
|
8300 |
+
"loss": 43.03,
|
8301 |
+
"step": 1180
|
8302 |
+
},
|
8303 |
+
{
|
8304 |
+
"epoch": 0.11713945645705218,
|
8305 |
+
"grad_norm": 2.6130144596099854,
|
8306 |
+
"learning_rate": 2.564423228176971e-05,
|
8307 |
+
"loss": 42.7279,
|
8308 |
+
"step": 1181
|
8309 |
+
},
|
8310 |
+
{
|
8311 |
+
"epoch": 0.11723864312636381,
|
8312 |
+
"grad_norm": 1.7635635137557983,
|
8313 |
+
"learning_rate": 2.5506819708706507e-05,
|
8314 |
+
"loss": 43.1984,
|
8315 |
+
"step": 1182
|
8316 |
+
},
|
8317 |
+
{
|
8318 |
+
"epoch": 0.11733782979567546,
|
8319 |
+
"grad_norm": 2.270341157913208,
|
8320 |
+
"learning_rate": 2.5369722445853304e-05,
|
8321 |
+
"loss": 43.2886,
|
8322 |
+
"step": 1183
|
8323 |
+
},
|
8324 |
+
{
|
8325 |
+
"epoch": 0.11743701646498711,
|
8326 |
+
"grad_norm": 2.2385268211364746,
|
8327 |
+
"learning_rate": 2.523294107350711e-05,
|
8328 |
+
"loss": 43.0718,
|
8329 |
+
"step": 1184
|
8330 |
+
},
|
8331 |
+
{
|
8332 |
+
"epoch": 0.11753620313429874,
|
8333 |
+
"grad_norm": 2.09205961227417,
|
8334 |
+
"learning_rate": 2.5096476170627825e-05,
|
8335 |
+
"loss": 43.4038,
|
8336 |
+
"step": 1185
|
8337 |
+
},
|
8338 |
+
{
|
8339 |
+
"epoch": 0.1176353898036104,
|
8340 |
+
"grad_norm": 1.9170998334884644,
|
8341 |
+
"learning_rate": 2.4960328314835745e-05,
|
8342 |
+
"loss": 43.4337,
|
8343 |
+
"step": 1186
|
8344 |
+
},
|
8345 |
+
{
|
8346 |
+
"epoch": 0.11773457647292204,
|
8347 |
+
"grad_norm": 2.0192651748657227,
|
8348 |
+
"learning_rate": 2.482449808240931e-05,
|
8349 |
+
"loss": 43.3694,
|
8350 |
+
"step": 1187
|
8351 |
+
},
|
8352 |
+
{
|
8353 |
+
"epoch": 0.11783376314223368,
|
8354 |
+
"grad_norm": 2.1703100204467773,
|
8355 |
+
"learning_rate": 2.4688986048282425e-05,
|
8356 |
+
"loss": 43.1931,
|
8357 |
+
"step": 1188
|
8358 |
+
},
|
8359 |
+
{
|
8360 |
+
"epoch": 0.11793294981154533,
|
8361 |
+
"grad_norm": 2.178591728210449,
|
8362 |
+
"learning_rate": 2.4553792786042262e-05,
|
8363 |
+
"loss": 43.0468,
|
8364 |
+
"step": 1189
|
8365 |
+
},
|
8366 |
+
{
|
8367 |
+
"epoch": 0.11803213648085698,
|
8368 |
+
"grad_norm": 1.8173599243164062,
|
8369 |
+
"learning_rate": 2.4418918867926577e-05,
|
8370 |
+
"loss": 43.1974,
|
8371 |
+
"step": 1190
|
8372 |
+
},
|
8373 |
+
{
|
8374 |
+
"epoch": 0.11813132315016862,
|
8375 |
+
"grad_norm": 2.2965517044067383,
|
8376 |
+
"learning_rate": 2.4284364864821563e-05,
|
8377 |
+
"loss": 43.2255,
|
8378 |
+
"step": 1191
|
8379 |
+
},
|
8380 |
+
{
|
8381 |
+
"epoch": 0.11823050981948026,
|
8382 |
+
"grad_norm": 2.666412115097046,
|
8383 |
+
"learning_rate": 2.4150131346259197e-05,
|
8384 |
+
"loss": 43.412,
|
8385 |
+
"step": 1192
|
8386 |
+
},
|
8387 |
+
{
|
8388 |
+
"epoch": 0.11832969648879191,
|
8389 |
+
"grad_norm": 2.0623011589050293,
|
8390 |
+
"learning_rate": 2.4016218880414998e-05,
|
8391 |
+
"loss": 42.8318,
|
8392 |
+
"step": 1193
|
8393 |
+
},
|
8394 |
+
{
|
8395 |
+
"epoch": 0.11842888315810356,
|
8396 |
+
"grad_norm": 2.0412418842315674,
|
8397 |
+
"learning_rate": 2.388262803410547e-05,
|
8398 |
+
"loss": 43.3501,
|
8399 |
+
"step": 1194
|
8400 |
+
},
|
8401 |
+
{
|
8402 |
+
"epoch": 0.11852806982741519,
|
8403 |
+
"grad_norm": 1.9982753992080688,
|
8404 |
+
"learning_rate": 2.3749359372785883e-05,
|
8405 |
+
"loss": 43.4671,
|
8406 |
+
"step": 1195
|
8407 |
+
},
|
8408 |
+
{
|
8409 |
+
"epoch": 0.11862725649672684,
|
8410 |
+
"grad_norm": 1.9406991004943848,
|
8411 |
+
"learning_rate": 2.3616413460547702e-05,
|
8412 |
+
"loss": 43.2969,
|
8413 |
+
"step": 1196
|
8414 |
+
},
|
8415 |
+
{
|
8416 |
+
"epoch": 0.11872644316603849,
|
8417 |
+
"grad_norm": 1.9013112783432007,
|
8418 |
+
"learning_rate": 2.3483790860116316e-05,
|
8419 |
+
"loss": 43.5889,
|
8420 |
+
"step": 1197
|
8421 |
+
},
|
8422 |
+
{
|
8423 |
+
"epoch": 0.11882562983535012,
|
8424 |
+
"grad_norm": 1.7011783123016357,
|
8425 |
+
"learning_rate": 2.3351492132848664e-05,
|
8426 |
+
"loss": 43.3593,
|
8427 |
+
"step": 1198
|
8428 |
+
},
|
8429 |
+
{
|
8430 |
+
"epoch": 0.11892481650466177,
|
8431 |
+
"grad_norm": 1.7616310119628906,
|
8432 |
+
"learning_rate": 2.3219517838730686e-05,
|
8433 |
+
"loss": 43.3636,
|
8434 |
+
"step": 1199
|
8435 |
+
},
|
8436 |
+
{
|
8437 |
+
"epoch": 0.11902400317397342,
|
8438 |
+
"grad_norm": 1.9877979755401611,
|
8439 |
+
"learning_rate": 2.3087868536375234e-05,
|
8440 |
+
"loss": 43.4894,
|
8441 |
+
"step": 1200
|
8442 |
+
},
|
8443 |
+
{
|
8444 |
+
"epoch": 0.11912318984328506,
|
8445 |
+
"grad_norm": 2.02473521232605,
|
8446 |
+
"learning_rate": 2.2956544783019418e-05,
|
8447 |
+
"loss": 43.1826,
|
8448 |
+
"step": 1201
|
8449 |
+
},
|
8450 |
+
{
|
8451 |
+
"epoch": 0.1192223765125967,
|
8452 |
+
"grad_norm": 1.9803928136825562,
|
8453 |
+
"learning_rate": 2.2825547134522495e-05,
|
8454 |
+
"loss": 43.4213,
|
8455 |
+
"step": 1202
|
8456 |
+
},
|
8457 |
+
{
|
8458 |
+
"epoch": 0.11932156318190835,
|
8459 |
+
"grad_norm": 2.13665771484375,
|
8460 |
+
"learning_rate": 2.2694876145363308e-05,
|
8461 |
+
"loss": 43.2816,
|
8462 |
+
"step": 1203
|
8463 |
+
},
|
8464 |
+
{
|
8465 |
+
"epoch": 0.11942074985121999,
|
8466 |
+
"grad_norm": 2.1694679260253906,
|
8467 |
+
"learning_rate": 2.2564532368638146e-05,
|
8468 |
+
"loss": 43.4692,
|
8469 |
+
"step": 1204
|
8470 |
+
},
|
8471 |
+
{
|
8472 |
+
"epoch": 0.11951993652053164,
|
8473 |
+
"grad_norm": 2.1723930835723877,
|
8474 |
+
"learning_rate": 2.2434516356058177e-05,
|
8475 |
+
"loss": 43.3492,
|
8476 |
+
"step": 1205
|
8477 |
+
},
|
8478 |
+
{
|
8479 |
+
"epoch": 0.11961912318984329,
|
8480 |
+
"grad_norm": 2.2894623279571533,
|
8481 |
+
"learning_rate": 2.230482865794733e-05,
|
8482 |
+
"loss": 43.2695,
|
8483 |
+
"step": 1206
|
8484 |
+
},
|
8485 |
+
{
|
8486 |
+
"epoch": 0.11971830985915492,
|
8487 |
+
"grad_norm": 1.7843515872955322,
|
8488 |
+
"learning_rate": 2.2175469823239768e-05,
|
8489 |
+
"loss": 43.1204,
|
8490 |
+
"step": 1207
|
8491 |
+
},
|
8492 |
+
{
|
8493 |
+
"epoch": 0.11981749652846657,
|
8494 |
+
"grad_norm": 1.960874080657959,
|
8495 |
+
"learning_rate": 2.2046440399477762e-05,
|
8496 |
+
"loss": 43.4489,
|
8497 |
+
"step": 1208
|
8498 |
+
},
|
8499 |
+
{
|
8500 |
+
"epoch": 0.11991668319777822,
|
8501 |
+
"grad_norm": 1.8730344772338867,
|
8502 |
+
"learning_rate": 2.1917740932809173e-05,
|
8503 |
+
"loss": 42.9379,
|
8504 |
+
"step": 1209
|
8505 |
+
},
|
8506 |
+
{
|
8507 |
+
"epoch": 0.12001586986708987,
|
8508 |
+
"grad_norm": 2.100027322769165,
|
8509 |
+
"learning_rate": 2.1789371967985338e-05,
|
8510 |
+
"loss": 43.5234,
|
8511 |
+
"step": 1210
|
8512 |
+
},
|
8513 |
+
{
|
8514 |
+
"epoch": 0.1201150565364015,
|
8515 |
+
"grad_norm": 2.344916820526123,
|
8516 |
+
"learning_rate": 2.1661334048358573e-05,
|
8517 |
+
"loss": 43.3028,
|
8518 |
+
"step": 1211
|
8519 |
+
},
|
8520 |
+
{
|
8521 |
+
"epoch": 0.12021424320571315,
|
8522 |
+
"grad_norm": 1.93478262424469,
|
8523 |
+
"learning_rate": 2.1533627715880023e-05,
|
8524 |
+
"loss": 42.8419,
|
8525 |
+
"step": 1212
|
8526 |
+
},
|
8527 |
+
{
|
8528 |
+
"epoch": 0.1203134298750248,
|
8529 |
+
"grad_norm": 2.0599870681762695,
|
8530 |
+
"learning_rate": 2.140625351109733e-05,
|
8531 |
+
"loss": 43.5079,
|
8532 |
+
"step": 1213
|
8533 |
+
},
|
8534 |
+
{
|
8535 |
+
"epoch": 0.12041261654433644,
|
8536 |
+
"grad_norm": 2.1910853385925293,
|
8537 |
+
"learning_rate": 2.1279211973152234e-05,
|
8538 |
+
"loss": 43.2715,
|
8539 |
+
"step": 1214
|
8540 |
+
},
|
8541 |
+
{
|
8542 |
+
"epoch": 0.12051180321364809,
|
8543 |
+
"grad_norm": 1.9811651706695557,
|
8544 |
+
"learning_rate": 2.115250363977851e-05,
|
8545 |
+
"loss": 43.4015,
|
8546 |
+
"step": 1215
|
8547 |
+
},
|
8548 |
+
{
|
8549 |
+
"epoch": 0.12061098988295973,
|
8550 |
+
"grad_norm": 1.7748483419418335,
|
8551 |
+
"learning_rate": 2.1026129047299436e-05,
|
8552 |
+
"loss": 43.3905,
|
8553 |
+
"step": 1216
|
8554 |
+
},
|
8555 |
+
{
|
8556 |
+
"epoch": 0.12071017655227137,
|
8557 |
+
"grad_norm": 1.954972267150879,
|
8558 |
+
"learning_rate": 2.0900088730625755e-05,
|
8559 |
+
"loss": 43.2451,
|
8560 |
+
"step": 1217
|
8561 |
+
},
|
8562 |
+
{
|
8563 |
+
"epoch": 0.12080936322158302,
|
8564 |
+
"grad_norm": 2.3269190788269043,
|
8565 |
+
"learning_rate": 2.0774383223253225e-05,
|
8566 |
+
"loss": 43.2175,
|
8567 |
+
"step": 1218
|
8568 |
+
},
|
8569 |
+
{
|
8570 |
+
"epoch": 0.12090854989089467,
|
8571 |
+
"grad_norm": 2.1407523155212402,
|
8572 |
+
"learning_rate": 2.0649013057260546e-05,
|
8573 |
+
"loss": 43.0747,
|
8574 |
+
"step": 1219
|
8575 |
+
},
|
8576 |
+
{
|
8577 |
+
"epoch": 0.1210077365602063,
|
8578 |
+
"grad_norm": 1.73319673538208,
|
8579 |
+
"learning_rate": 2.0523978763306873e-05,
|
8580 |
+
"loss": 43.4118,
|
8581 |
+
"step": 1220
|
8582 |
+
},
|
8583 |
+
{
|
8584 |
+
"epoch": 0.12110692322951795,
|
8585 |
+
"grad_norm": 1.7930755615234375,
|
8586 |
+
"learning_rate": 2.0399280870629823e-05,
|
8587 |
+
"loss": 43.1181,
|
8588 |
+
"step": 1221
|
8589 |
+
},
|
8590 |
+
{
|
8591 |
+
"epoch": 0.1212061098988296,
|
8592 |
+
"grad_norm": 2.0965116024017334,
|
8593 |
+
"learning_rate": 2.0274919907043033e-05,
|
8594 |
+
"loss": 43.3303,
|
8595 |
+
"step": 1222
|
8596 |
+
},
|
8597 |
+
{
|
8598 |
+
"epoch": 0.12130529656814124,
|
8599 |
+
"grad_norm": 1.9727377891540527,
|
8600 |
+
"learning_rate": 2.0150896398934073e-05,
|
8601 |
+
"loss": 43.2962,
|
8602 |
+
"step": 1223
|
8603 |
+
},
|
8604 |
+
{
|
8605 |
+
"epoch": 0.12140448323745288,
|
8606 |
+
"grad_norm": 2.1021900177001953,
|
8607 |
+
"learning_rate": 2.0027210871262068e-05,
|
8608 |
+
"loss": 43.2473,
|
8609 |
+
"step": 1224
|
8610 |
+
},
|
8611 |
+
{
|
8612 |
+
"epoch": 0.12150366990676453,
|
8613 |
+
"grad_norm": 2.155071258544922,
|
8614 |
+
"learning_rate": 1.9903863847555648e-05,
|
8615 |
+
"loss": 43.3419,
|
8616 |
+
"step": 1225
|
8617 |
+
},
|
8618 |
+
{
|
8619 |
+
"epoch": 0.12160285657607618,
|
8620 |
+
"grad_norm": 2.007256507873535,
|
8621 |
+
"learning_rate": 1.9780855849910605e-05,
|
8622 |
+
"loss": 43.5286,
|
8623 |
+
"step": 1226
|
8624 |
+
},
|
8625 |
+
{
|
8626 |
+
"epoch": 0.12170204324538782,
|
8627 |
+
"grad_norm": 2.027193546295166,
|
8628 |
+
"learning_rate": 1.9658187398987692e-05,
|
8629 |
+
"loss": 43.5892,
|
8630 |
+
"step": 1227
|
8631 |
+
},
|
8632 |
+
{
|
8633 |
+
"epoch": 0.12180122991469947,
|
8634 |
+
"grad_norm": 1.863158941268921,
|
8635 |
+
"learning_rate": 1.9535859014010526e-05,
|
8636 |
+
"loss": 43.0008,
|
8637 |
+
"step": 1228
|
8638 |
+
},
|
8639 |
+
{
|
8640 |
+
"epoch": 0.12190041658401111,
|
8641 |
+
"grad_norm": 1.9494130611419678,
|
8642 |
+
"learning_rate": 1.9413871212763212e-05,
|
8643 |
+
"loss": 43.3241,
|
8644 |
+
"step": 1229
|
8645 |
+
},
|
8646 |
+
{
|
8647 |
+
"epoch": 0.12199960325332275,
|
8648 |
+
"grad_norm": 1.9798649549484253,
|
8649 |
+
"learning_rate": 1.9292224511588365e-05,
|
8650 |
+
"loss": 43.1907,
|
8651 |
+
"step": 1230
|
8652 |
+
},
|
8653 |
+
{
|
8654 |
+
"epoch": 0.1220987899226344,
|
8655 |
+
"grad_norm": 2.2833685874938965,
|
8656 |
+
"learning_rate": 1.917091942538469e-05,
|
8657 |
+
"loss": 43.007,
|
8658 |
+
"step": 1231
|
8659 |
+
},
|
8660 |
+
{
|
8661 |
+
"epoch": 0.12219797659194605,
|
8662 |
+
"grad_norm": 2.0100409984588623,
|
8663 |
+
"learning_rate": 1.9049956467605055e-05,
|
8664 |
+
"loss": 43.1298,
|
8665 |
+
"step": 1232
|
8666 |
+
},
|
8667 |
+
{
|
8668 |
+
"epoch": 0.12229716326125768,
|
8669 |
+
"grad_norm": 2.5046565532684326,
|
8670 |
+
"learning_rate": 1.8929336150254064e-05,
|
8671 |
+
"loss": 43.3805,
|
8672 |
+
"step": 1233
|
8673 |
+
},
|
8674 |
+
{
|
8675 |
+
"epoch": 0.12239634993056933,
|
8676 |
+
"grad_norm": 1.741270661354065,
|
8677 |
+
"learning_rate": 1.880905898388612e-05,
|
8678 |
+
"loss": 43.0911,
|
8679 |
+
"step": 1234
|
8680 |
+
},
|
8681 |
+
{
|
8682 |
+
"epoch": 0.12249553659988098,
|
8683 |
+
"grad_norm": 2.1863491535186768,
|
8684 |
+
"learning_rate": 1.8689125477603063e-05,
|
8685 |
+
"loss": 43.0767,
|
8686 |
+
"step": 1235
|
8687 |
+
},
|
8688 |
+
{
|
8689 |
+
"epoch": 0.12259472326919262,
|
8690 |
+
"grad_norm": 2.0383381843566895,
|
8691 |
+
"learning_rate": 1.8569536139052213e-05,
|
8692 |
+
"loss": 43.4303,
|
8693 |
+
"step": 1236
|
8694 |
+
},
|
8695 |
+
{
|
8696 |
+
"epoch": 0.12269390993850426,
|
8697 |
+
"grad_norm": 2.046947956085205,
|
8698 |
+
"learning_rate": 1.8450291474423998e-05,
|
8699 |
+
"loss": 43.3978,
|
8700 |
+
"step": 1237
|
8701 |
+
},
|
8702 |
+
{
|
8703 |
+
"epoch": 0.12279309660781591,
|
8704 |
+
"grad_norm": 2.215596914291382,
|
8705 |
+
"learning_rate": 1.8331391988450052e-05,
|
8706 |
+
"loss": 43.1113,
|
8707 |
+
"step": 1238
|
8708 |
+
},
|
8709 |
+
{
|
8710 |
+
"epoch": 0.12289228327712755,
|
8711 |
+
"grad_norm": 1.731846570968628,
|
8712 |
+
"learning_rate": 1.8212838184400838e-05,
|
8713 |
+
"loss": 43.5589,
|
8714 |
+
"step": 1239
|
8715 |
+
},
|
8716 |
+
{
|
8717 |
+
"epoch": 0.1229914699464392,
|
8718 |
+
"grad_norm": 2.037714719772339,
|
8719 |
+
"learning_rate": 1.8094630564083736e-05,
|
8720 |
+
"loss": 43.4503,
|
8721 |
+
"step": 1240
|
8722 |
+
},
|
8723 |
+
{
|
8724 |
+
"epoch": 0.12309065661575085,
|
8725 |
+
"grad_norm": 2.0696065425872803,
|
8726 |
+
"learning_rate": 1.7976769627840806e-05,
|
8727 |
+
"loss": 42.9741,
|
8728 |
+
"step": 1241
|
8729 |
+
},
|
8730 |
+
{
|
8731 |
+
"epoch": 0.12318984328506248,
|
8732 |
+
"grad_norm": 2.007958173751831,
|
8733 |
+
"learning_rate": 1.785925587454661e-05,
|
8734 |
+
"loss": 43.2596,
|
8735 |
+
"step": 1242
|
8736 |
+
},
|
8737 |
+
{
|
8738 |
+
"epoch": 0.12328902995437413,
|
8739 |
+
"grad_norm": 2.0569231510162354,
|
8740 |
+
"learning_rate": 1.7742089801606276e-05,
|
8741 |
+
"loss": 43.4014,
|
8742 |
+
"step": 1243
|
8743 |
+
},
|
8744 |
+
{
|
8745 |
+
"epoch": 0.12338821662368578,
|
8746 |
+
"grad_norm": 2.107531785964966,
|
8747 |
+
"learning_rate": 1.762527190495319e-05,
|
8748 |
+
"loss": 43.0654,
|
8749 |
+
"step": 1244
|
8750 |
+
},
|
8751 |
+
{
|
8752 |
+
"epoch": 0.12348740329299743,
|
8753 |
+
"grad_norm": 1.771912693977356,
|
8754 |
+
"learning_rate": 1.750880267904711e-05,
|
8755 |
+
"loss": 43.5022,
|
8756 |
+
"step": 1245
|
8757 |
+
},
|
8758 |
+
{
|
8759 |
+
"epoch": 0.12358658996230906,
|
8760 |
+
"grad_norm": 2.4979348182678223,
|
8761 |
+
"learning_rate": 1.7392682616871837e-05,
|
8762 |
+
"loss": 43.4391,
|
8763 |
+
"step": 1246
|
8764 |
+
},
|
8765 |
+
{
|
8766 |
+
"epoch": 0.12368577663162071,
|
8767 |
+
"grad_norm": 2.4610869884490967,
|
8768 |
+
"learning_rate": 1.7276912209933392e-05,
|
8769 |
+
"loss": 43.3589,
|
8770 |
+
"step": 1247
|
8771 |
+
},
|
8772 |
+
{
|
8773 |
+
"epoch": 0.12378496330093236,
|
8774 |
+
"grad_norm": 1.9890514612197876,
|
8775 |
+
"learning_rate": 1.7161491948257657e-05,
|
8776 |
+
"loss": 43.2026,
|
8777 |
+
"step": 1248
|
8778 |
+
},
|
8779 |
+
{
|
8780 |
+
"epoch": 0.123884149970244,
|
8781 |
+
"grad_norm": 2.3719217777252197,
|
8782 |
+
"learning_rate": 1.7046422320388556e-05,
|
8783 |
+
"loss": 43.1041,
|
8784 |
+
"step": 1249
|
8785 |
+
},
|
8786 |
+
{
|
8787 |
+
"epoch": 0.12398333663955564,
|
8788 |
+
"grad_norm": 2.3359856605529785,
|
8789 |
+
"learning_rate": 1.693170381338578e-05,
|
8790 |
+
"loss": 43.5385,
|
8791 |
+
"step": 1250
|
8792 |
+
},
|
8793 |
+
{
|
8794 |
+
"epoch": 0.12408252330886729,
|
8795 |
+
"grad_norm": 1.738102912902832,
|
8796 |
+
"learning_rate": 1.6817336912822912e-05,
|
8797 |
+
"loss": 43.344,
|
8798 |
+
"step": 1251
|
8799 |
+
},
|
8800 |
+
{
|
8801 |
+
"epoch": 0.12418170997817893,
|
8802 |
+
"grad_norm": 2.1407241821289062,
|
8803 |
+
"learning_rate": 1.6703322102785168e-05,
|
8804 |
+
"loss": 43.3036,
|
8805 |
+
"step": 1252
|
8806 |
+
},
|
8807 |
+
{
|
8808 |
+
"epoch": 0.12428089664749058,
|
8809 |
+
"grad_norm": 1.9617172479629517,
|
8810 |
+
"learning_rate": 1.6589659865867534e-05,
|
8811 |
+
"loss": 43.4202,
|
8812 |
+
"step": 1253
|
8813 |
+
},
|
8814 |
+
{
|
8815 |
+
"epoch": 0.12438008331680223,
|
8816 |
+
"grad_norm": 2.001938819885254,
|
8817 |
+
"learning_rate": 1.647635068317265e-05,
|
8818 |
+
"loss": 43.2289,
|
8819 |
+
"step": 1254
|
8820 |
+
},
|
8821 |
+
{
|
8822 |
+
"epoch": 0.12447926998611386,
|
8823 |
+
"grad_norm": 2.245469331741333,
|
8824 |
+
"learning_rate": 1.6363395034308703e-05,
|
8825 |
+
"loss": 43.215,
|
8826 |
+
"step": 1255
|
8827 |
+
},
|
8828 |
+
{
|
8829 |
+
"epoch": 0.12457845665542551,
|
8830 |
+
"grad_norm": 2.2097630500793457,
|
8831 |
+
"learning_rate": 1.6250793397387555e-05,
|
8832 |
+
"loss": 43.4554,
|
8833 |
+
"step": 1256
|
8834 |
+
},
|
8835 |
+
{
|
8836 |
+
"epoch": 0.12467764332473716,
|
8837 |
+
"grad_norm": 2.113950729370117,
|
8838 |
+
"learning_rate": 1.6138546249022513e-05,
|
8839 |
+
"loss": 43.2097,
|
8840 |
+
"step": 1257
|
8841 |
+
},
|
8842 |
+
{
|
8843 |
+
"epoch": 0.1247768299940488,
|
8844 |
+
"grad_norm": 2.116868019104004,
|
8845 |
+
"learning_rate": 1.6026654064326553e-05,
|
8846 |
+
"loss": 43.3248,
|
8847 |
+
"step": 1258
|
8848 |
+
},
|
8849 |
+
{
|
8850 |
+
"epoch": 0.12487601666336044,
|
8851 |
+
"grad_norm": 2.185568332672119,
|
8852 |
+
"learning_rate": 1.5915117316910055e-05,
|
8853 |
+
"loss": 43.0621,
|
8854 |
+
"step": 1259
|
8855 |
+
},
|
8856 |
+
{
|
8857 |
+
"epoch": 0.12497520333267209,
|
8858 |
+
"grad_norm": 2.0478570461273193,
|
8859 |
+
"learning_rate": 1.5803936478879057e-05,
|
8860 |
+
"loss": 43.1099,
|
8861 |
+
"step": 1260
|
8862 |
+
},
|
8863 |
+
{
|
8864 |
+
"epoch": 0.12507439000198373,
|
8865 |
+
"grad_norm": 2.1610467433929443,
|
8866 |
+
"learning_rate": 1.5693112020833013e-05,
|
8867 |
+
"loss": 43.2475,
|
8868 |
+
"step": 1261
|
8869 |
+
},
|
8870 |
+
{
|
8871 |
+
"epoch": 0.1251735766712954,
|
8872 |
+
"grad_norm": 1.7375211715698242,
|
8873 |
+
"learning_rate": 1.558264441186299e-05,
|
8874 |
+
"loss": 43.4413,
|
8875 |
+
"step": 1262
|
8876 |
+
},
|
8877 |
+
{
|
8878 |
+
"epoch": 0.12527276334060702,
|
8879 |
+
"grad_norm": 1.689758539199829,
|
8880 |
+
"learning_rate": 1.547253411954953e-05,
|
8881 |
+
"loss": 43.1902,
|
8882 |
+
"step": 1263
|
8883 |
+
},
|
8884 |
+
{
|
8885 |
+
"epoch": 0.12537195000991866,
|
8886 |
+
"grad_norm": 2.010601043701172,
|
8887 |
+
"learning_rate": 1.5362781609960852e-05,
|
8888 |
+
"loss": 43.3252,
|
8889 |
+
"step": 1264
|
8890 |
+
},
|
8891 |
+
{
|
8892 |
+
"epoch": 0.12547113667923032,
|
8893 |
+
"grad_norm": 2.1295485496520996,
|
8894 |
+
"learning_rate": 1.5253387347650661e-05,
|
8895 |
+
"loss": 43.4037,
|
8896 |
+
"step": 1265
|
8897 |
+
},
|
8898 |
+
{
|
8899 |
+
"epoch": 0.12557032334854196,
|
8900 |
+
"grad_norm": 2.2790913581848145,
|
8901 |
+
"learning_rate": 1.5144351795656387e-05,
|
8902 |
+
"loss": 43.5434,
|
8903 |
+
"step": 1266
|
8904 |
+
},
|
8905 |
+
{
|
8906 |
+
"epoch": 0.1256695100178536,
|
8907 |
+
"grad_norm": 2.4958717823028564,
|
8908 |
+
"learning_rate": 1.5035675415497063e-05,
|
8909 |
+
"loss": 43.0798,
|
8910 |
+
"step": 1267
|
8911 |
+
},
|
8912 |
+
{
|
8913 |
+
"epoch": 0.12576869668716525,
|
8914 |
+
"grad_norm": 2.2541685104370117,
|
8915 |
+
"learning_rate": 1.4927358667171476e-05,
|
8916 |
+
"loss": 43.1865,
|
8917 |
+
"step": 1268
|
8918 |
+
},
|
8919 |
+
{
|
8920 |
+
"epoch": 0.1258678833564769,
|
8921 |
+
"grad_norm": 1.9234634637832642,
|
8922 |
+
"learning_rate": 1.4819402009156214e-05,
|
8923 |
+
"loss": 43.5065,
|
8924 |
+
"step": 1269
|
8925 |
+
},
|
8926 |
+
{
|
8927 |
+
"epoch": 0.12596707002578852,
|
8928 |
+
"grad_norm": 2.2710721492767334,
|
8929 |
+
"learning_rate": 1.471180589840363e-05,
|
8930 |
+
"loss": 43.2936,
|
8931 |
+
"step": 1270
|
8932 |
+
},
|
8933 |
+
{
|
8934 |
+
"epoch": 0.1260662566951002,
|
8935 |
+
"grad_norm": 2.3189542293548584,
|
8936 |
+
"learning_rate": 1.4604570790340023e-05,
|
8937 |
+
"loss": 43.1946,
|
8938 |
+
"step": 1271
|
8939 |
+
},
|
8940 |
+
{
|
8941 |
+
"epoch": 0.12616544336441182,
|
8942 |
+
"grad_norm": 2.0585691928863525,
|
8943 |
+
"learning_rate": 1.449769713886363e-05,
|
8944 |
+
"loss": 43.1392,
|
8945 |
+
"step": 1272
|
8946 |
+
},
|
8947 |
+
{
|
8948 |
+
"epoch": 0.12626463003372346,
|
8949 |
+
"grad_norm": 2.1090476512908936,
|
8950 |
+
"learning_rate": 1.4391185396342789e-05,
|
8951 |
+
"loss": 43.2654,
|
8952 |
+
"step": 1273
|
8953 |
+
},
|
8954 |
+
{
|
8955 |
+
"epoch": 0.12636381670303512,
|
8956 |
+
"grad_norm": 1.9602532386779785,
|
8957 |
+
"learning_rate": 1.4285036013613872e-05,
|
8958 |
+
"loss": 42.953,
|
8959 |
+
"step": 1274
|
8960 |
+
},
|
8961 |
+
{
|
8962 |
+
"epoch": 0.12646300337234676,
|
8963 |
+
"grad_norm": 2.2727339267730713,
|
8964 |
+
"learning_rate": 1.4179249439979613e-05,
|
8965 |
+
"loss": 43.0101,
|
8966 |
+
"step": 1275
|
8967 |
+
},
|
8968 |
+
{
|
8969 |
+
"epoch": 0.1265621900416584,
|
8970 |
+
"grad_norm": 2.0381500720977783,
|
8971 |
+
"learning_rate": 1.4073826123206946e-05,
|
8972 |
+
"loss": 43.1364,
|
8973 |
+
"step": 1276
|
8974 |
+
},
|
8975 |
+
{
|
8976 |
+
"epoch": 0.12666137671097005,
|
8977 |
+
"grad_norm": 2.0481743812561035,
|
8978 |
+
"learning_rate": 1.3968766509525322e-05,
|
8979 |
+
"loss": 43.5437,
|
8980 |
+
"step": 1277
|
8981 |
+
},
|
8982 |
+
{
|
8983 |
+
"epoch": 0.1267605633802817,
|
8984 |
+
"grad_norm": 2.1864757537841797,
|
8985 |
+
"learning_rate": 1.386407104362467e-05,
|
8986 |
+
"loss": 43.0872,
|
8987 |
+
"step": 1278
|
8988 |
+
},
|
8989 |
+
{
|
8990 |
+
"epoch": 0.12685975004959332,
|
8991 |
+
"grad_norm": 2.3352155685424805,
|
8992 |
+
"learning_rate": 1.375974016865359e-05,
|
8993 |
+
"loss": 43.2959,
|
8994 |
+
"step": 1279
|
8995 |
+
},
|
8996 |
+
{
|
8997 |
+
"epoch": 0.12695893671890499,
|
8998 |
+
"grad_norm": 2.0664021968841553,
|
8999 |
+
"learning_rate": 1.3655774326217507e-05,
|
9000 |
+
"loss": 43.2873,
|
9001 |
+
"step": 1280
|
9002 |
+
},
|
9003 |
+
{
|
9004 |
+
"epoch": 0.12705812338821662,
|
9005 |
+
"grad_norm": 2.0694098472595215,
|
9006 |
+
"learning_rate": 1.3552173956376668e-05,
|
9007 |
+
"loss": 43.305,
|
9008 |
+
"step": 1281
|
9009 |
+
},
|
9010 |
+
{
|
9011 |
+
"epoch": 0.12715731005752826,
|
9012 |
+
"grad_norm": 2.1970510482788086,
|
9013 |
+
"learning_rate": 1.3448939497644509e-05,
|
9014 |
+
"loss": 43.1985,
|
9015 |
+
"step": 1282
|
9016 |
+
},
|
9017 |
+
{
|
9018 |
+
"epoch": 0.12725649672683992,
|
9019 |
+
"grad_norm": 1.886986494064331,
|
9020 |
+
"learning_rate": 1.3346071386985515e-05,
|
9021 |
+
"loss": 43.5169,
|
9022 |
+
"step": 1283
|
9023 |
+
},
|
9024 |
+
{
|
9025 |
+
"epoch": 0.12735568339615155,
|
9026 |
+
"grad_norm": 1.7438634634017944,
|
9027 |
+
"learning_rate": 1.3243570059813626e-05,
|
9028 |
+
"loss": 43.2525,
|
9029 |
+
"step": 1284
|
9030 |
+
},
|
9031 |
+
{
|
9032 |
+
"epoch": 0.1274548700654632,
|
9033 |
+
"grad_norm": 2.0268194675445557,
|
9034 |
+
"learning_rate": 1.3141435949990188e-05,
|
9035 |
+
"loss": 43.4215,
|
9036 |
+
"step": 1285
|
9037 |
+
},
|
9038 |
+
{
|
9039 |
+
"epoch": 0.12755405673477485,
|
9040 |
+
"grad_norm": 2.101165533065796,
|
9041 |
+
"learning_rate": 1.3039669489822292e-05,
|
9042 |
+
"loss": 42.7054,
|
9043 |
+
"step": 1286
|
9044 |
+
},
|
9045 |
+
{
|
9046 |
+
"epoch": 0.12765324340408649,
|
9047 |
+
"grad_norm": 1.8700491189956665,
|
9048 |
+
"learning_rate": 1.2938271110060807e-05,
|
9049 |
+
"loss": 43.267,
|
9050 |
+
"step": 1287
|
9051 |
+
},
|
9052 |
+
{
|
9053 |
+
"epoch": 0.12775243007339815,
|
9054 |
+
"grad_norm": 1.8919130563735962,
|
9055 |
+
"learning_rate": 1.2837241239898667e-05,
|
9056 |
+
"loss": 43.4118,
|
9057 |
+
"step": 1288
|
9058 |
+
},
|
9059 |
+
{
|
9060 |
+
"epoch": 0.12785161674270978,
|
9061 |
+
"grad_norm": 2.1673426628112793,
|
9062 |
+
"learning_rate": 1.2736580306968938e-05,
|
9063 |
+
"loss": 43.2691,
|
9064 |
+
"step": 1289
|
9065 |
+
},
|
9066 |
+
{
|
9067 |
+
"epoch": 0.12795080341202142,
|
9068 |
+
"grad_norm": 1.9596093893051147,
|
9069 |
+
"learning_rate": 1.2636288737343138e-05,
|
9070 |
+
"loss": 43.1042,
|
9071 |
+
"step": 1290
|
9072 |
+
},
|
9073 |
+
{
|
9074 |
+
"epoch": 0.12804999008133308,
|
9075 |
+
"grad_norm": 1.9993942975997925,
|
9076 |
+
"learning_rate": 1.253636695552931e-05,
|
9077 |
+
"loss": 43.4206,
|
9078 |
+
"step": 1291
|
9079 |
+
},
|
9080 |
+
{
|
9081 |
+
"epoch": 0.12814917675064472,
|
9082 |
+
"grad_norm": 1.9622743129730225,
|
9083 |
+
"learning_rate": 1.2436815384470301e-05,
|
9084 |
+
"loss": 43.5773,
|
9085 |
+
"step": 1292
|
9086 |
+
},
|
9087 |
+
{
|
9088 |
+
"epoch": 0.12824836341995635,
|
9089 |
+
"grad_norm": 2.12934947013855,
|
9090 |
+
"learning_rate": 1.233763444554198e-05,
|
9091 |
+
"loss": 42.9882,
|
9092 |
+
"step": 1293
|
9093 |
+
},
|
9094 |
+
{
|
9095 |
+
"epoch": 0.12834755008926801,
|
9096 |
+
"grad_norm": 1.997297763824463,
|
9097 |
+
"learning_rate": 1.2238824558551365e-05,
|
9098 |
+
"loss": 43.2069,
|
9099 |
+
"step": 1294
|
9100 |
+
},
|
9101 |
+
{
|
9102 |
+
"epoch": 0.12844673675857965,
|
9103 |
+
"grad_norm": 1.7098406553268433,
|
9104 |
+
"learning_rate": 1.2140386141734972e-05,
|
9105 |
+
"loss": 43.3409,
|
9106 |
+
"step": 1295
|
9107 |
+
},
|
9108 |
+
{
|
9109 |
+
"epoch": 0.12854592342789128,
|
9110 |
+
"grad_norm": 1.924278736114502,
|
9111 |
+
"learning_rate": 1.2042319611756948e-05,
|
9112 |
+
"loss": 43.202,
|
9113 |
+
"step": 1296
|
9114 |
+
},
|
9115 |
+
{
|
9116 |
+
"epoch": 0.12864511009720295,
|
9117 |
+
"grad_norm": 2.1477646827697754,
|
9118 |
+
"learning_rate": 1.1944625383707374e-05,
|
9119 |
+
"loss": 43.1039,
|
9120 |
+
"step": 1297
|
9121 |
+
},
|
9122 |
+
{
|
9123 |
+
"epoch": 0.12874429676651458,
|
9124 |
+
"grad_norm": 1.9721094369888306,
|
9125 |
+
"learning_rate": 1.184730387110039e-05,
|
9126 |
+
"loss": 43.5037,
|
9127 |
+
"step": 1298
|
9128 |
+
},
|
9129 |
+
{
|
9130 |
+
"epoch": 0.12884348343582622,
|
9131 |
+
"grad_norm": 1.9413458108901978,
|
9132 |
+
"learning_rate": 1.1750355485872644e-05,
|
9133 |
+
"loss": 43.7498,
|
9134 |
+
"step": 1299
|
9135 |
+
},
|
9136 |
+
{
|
9137 |
+
"epoch": 0.12894267010513788,
|
9138 |
+
"grad_norm": 2.565575122833252,
|
9139 |
+
"learning_rate": 1.1653780638381328e-05,
|
9140 |
+
"loss": 43.1948,
|
9141 |
+
"step": 1300
|
9142 |
+
},
|
9143 |
+
{
|
9144 |
+
"epoch": 0.12904185677444951,
|
9145 |
+
"grad_norm": 2.2411508560180664,
|
9146 |
+
"learning_rate": 1.1557579737402623e-05,
|
9147 |
+
"loss": 43.4395,
|
9148 |
+
"step": 1301
|
9149 |
+
},
|
9150 |
+
{
|
9151 |
+
"epoch": 0.12914104344376115,
|
9152 |
+
"grad_norm": 1.8507673740386963,
|
9153 |
+
"learning_rate": 1.146175319012983e-05,
|
9154 |
+
"loss": 43.1794,
|
9155 |
+
"step": 1302
|
9156 |
+
},
|
9157 |
+
{
|
9158 |
+
"epoch": 0.1292402301130728,
|
9159 |
+
"grad_norm": 2.061469078063965,
|
9160 |
+
"learning_rate": 1.1366301402171775e-05,
|
9161 |
+
"loss": 43.0605,
|
9162 |
+
"step": 1303
|
9163 |
+
},
|
9164 |
+
{
|
9165 |
+
"epoch": 0.12933941678238445,
|
9166 |
+
"grad_norm": 2.121185302734375,
|
9167 |
+
"learning_rate": 1.127122477755097e-05,
|
9168 |
+
"loss": 43.431,
|
9169 |
+
"step": 1304
|
9170 |
+
},
|
9171 |
+
{
|
9172 |
+
"epoch": 0.12943860345169608,
|
9173 |
+
"grad_norm": 2.0242440700531006,
|
9174 |
+
"learning_rate": 1.117652371870197e-05,
|
9175 |
+
"loss": 43.4726,
|
9176 |
+
"step": 1305
|
9177 |
+
},
|
9178 |
+
{
|
9179 |
+
"epoch": 0.12953779012100775,
|
9180 |
+
"grad_norm": 2.2344465255737305,
|
9181 |
+
"learning_rate": 1.1082198626469686e-05,
|
9182 |
+
"loss": 42.8992,
|
9183 |
+
"step": 1306
|
9184 |
+
},
|
9185 |
+
{
|
9186 |
+
"epoch": 0.12963697679031938,
|
9187 |
+
"grad_norm": 1.7026211023330688,
|
9188 |
+
"learning_rate": 1.0988249900107605e-05,
|
9189 |
+
"loss": 43.401,
|
9190 |
+
"step": 1307
|
9191 |
+
},
|
9192 |
+
{
|
9193 |
+
"epoch": 0.12973616345963102,
|
9194 |
+
"grad_norm": 2.0010180473327637,
|
9195 |
+
"learning_rate": 1.0894677937276232e-05,
|
9196 |
+
"loss": 43.3382,
|
9197 |
+
"step": 1308
|
9198 |
+
},
|
9199 |
+
{
|
9200 |
+
"epoch": 0.12983535012894268,
|
9201 |
+
"grad_norm": 1.8324205875396729,
|
9202 |
+
"learning_rate": 1.0801483134041268e-05,
|
9203 |
+
"loss": 43.4343,
|
9204 |
+
"step": 1309
|
9205 |
+
},
|
9206 |
+
{
|
9207 |
+
"epoch": 0.1299345367982543,
|
9208 |
+
"grad_norm": 2.0147500038146973,
|
9209 |
+
"learning_rate": 1.0708665884872037e-05,
|
9210 |
+
"loss": 43.0836,
|
9211 |
+
"step": 1310
|
9212 |
+
},
|
9213 |
+
{
|
9214 |
+
"epoch": 0.13003372346756595,
|
9215 |
+
"grad_norm": 2.060643434524536,
|
9216 |
+
"learning_rate": 1.0616226582639732e-05,
|
9217 |
+
"loss": 43.3897,
|
9218 |
+
"step": 1311
|
9219 |
+
},
|
9220 |
+
{
|
9221 |
+
"epoch": 0.1301329101368776,
|
9222 |
+
"grad_norm": 1.9988598823547363,
|
9223 |
+
"learning_rate": 1.0524165618615845e-05,
|
9224 |
+
"loss": 43.5015,
|
9225 |
+
"step": 1312
|
9226 |
+
},
|
9227 |
+
{
|
9228 |
+
"epoch": 0.13023209680618925,
|
9229 |
+
"grad_norm": 1.6039632558822632,
|
9230 |
+
"learning_rate": 1.0432483382470393e-05,
|
9231 |
+
"loss": 43.352,
|
9232 |
+
"step": 1313
|
9233 |
+
},
|
9234 |
+
{
|
9235 |
+
"epoch": 0.13033128347550088,
|
9236 |
+
"grad_norm": 2.301841974258423,
|
9237 |
+
"learning_rate": 1.0341180262270422e-05,
|
9238 |
+
"loss": 43.1568,
|
9239 |
+
"step": 1314
|
9240 |
+
},
|
9241 |
+
{
|
9242 |
+
"epoch": 0.13043047014481254,
|
9243 |
+
"grad_norm": 2.549867868423462,
|
9244 |
+
"learning_rate": 1.0250256644478195e-05,
|
9245 |
+
"loss": 43.1025,
|
9246 |
+
"step": 1315
|
9247 |
+
},
|
9248 |
+
{
|
9249 |
+
"epoch": 0.13052965681412418,
|
9250 |
+
"grad_norm": 1.9916534423828125,
|
9251 |
+
"learning_rate": 1.0159712913949715e-05,
|
9252 |
+
"loss": 43.4582,
|
9253 |
+
"step": 1316
|
9254 |
+
},
|
9255 |
+
{
|
9256 |
+
"epoch": 0.1306288434834358,
|
9257 |
+
"grad_norm": 2.260643720626831,
|
9258 |
+
"learning_rate": 1.0069549453932969e-05,
|
9259 |
+
"loss": 43.3585,
|
9260 |
+
"step": 1317
|
9261 |
+
},
|
9262 |
+
{
|
9263 |
+
"epoch": 0.13072803015274748,
|
9264 |
+
"grad_norm": 2.251774549484253,
|
9265 |
+
"learning_rate": 9.979766646066368e-06,
|
9266 |
+
"loss": 43.3543,
|
9267 |
+
"step": 1318
|
9268 |
+
},
|
9269 |
+
{
|
9270 |
+
"epoch": 0.1308272168220591,
|
9271 |
+
"grad_norm": 2.2947351932525635,
|
9272 |
+
"learning_rate": 9.890364870377156e-06,
|
9273 |
+
"loss": 43.4266,
|
9274 |
+
"step": 1319
|
9275 |
+
},
|
9276 |
+
{
|
9277 |
+
"epoch": 0.13092640349137075,
|
9278 |
+
"grad_norm": 2.2526769638061523,
|
9279 |
+
"learning_rate": 9.801344505279698e-06,
|
9280 |
+
"loss": 43.4395,
|
9281 |
+
"step": 1320
|
9282 |
+
},
|
9283 |
+
{
|
9284 |
+
"epoch": 0.1310255901606824,
|
9285 |
+
"grad_norm": 2.026726007461548,
|
9286 |
+
"learning_rate": 9.71270592757404e-06,
|
9287 |
+
"loss": 43.4505,
|
9288 |
+
"step": 1321
|
9289 |
+
},
|
9290 |
+
{
|
9291 |
+
"epoch": 0.13112477682999404,
|
9292 |
+
"grad_norm": 2.089557647705078,
|
9293 |
+
"learning_rate": 9.6244495124441e-06,
|
9294 |
+
"loss": 43.5353,
|
9295 |
+
"step": 1322
|
9296 |
+
},
|
9297 |
+
{
|
9298 |
+
"epoch": 0.1312239634993057,
|
9299 |
+
"grad_norm": 1.9276314973831177,
|
9300 |
+
"learning_rate": 9.536575633456369e-06,
|
9301 |
+
"loss": 43.4943,
|
9302 |
+
"step": 1323
|
9303 |
+
},
|
9304 |
+
{
|
9305 |
+
"epoch": 0.13132315016861734,
|
9306 |
+
"grad_norm": 1.8411586284637451,
|
9307 |
+
"learning_rate": 9.449084662557982e-06,
|
9308 |
+
"loss": 43.3444,
|
9309 |
+
"step": 1324
|
9310 |
+
},
|
9311 |
+
{
|
9312 |
+
"epoch": 0.13142233683792898,
|
9313 |
+
"grad_norm": 2.3631200790405273,
|
9314 |
+
"learning_rate": 9.36197697007547e-06,
|
9315 |
+
"loss": 42.7248,
|
9316 |
+
"step": 1325
|
9317 |
+
},
|
9318 |
+
{
|
9319 |
+
"epoch": 0.13152152350724064,
|
9320 |
+
"grad_norm": 2.6807281970977783,
|
9321 |
+
"learning_rate": 9.27525292471294e-06,
|
9322 |
+
"loss": 43.3432,
|
9323 |
+
"step": 1326
|
9324 |
+
},
|
9325 |
+
{
|
9326 |
+
"epoch": 0.13162071017655227,
|
9327 |
+
"grad_norm": 1.9830889701843262,
|
9328 |
+
"learning_rate": 9.188912893550695e-06,
|
9329 |
+
"loss": 43.2004,
|
9330 |
+
"step": 1327
|
9331 |
+
},
|
9332 |
+
{
|
9333 |
+
"epoch": 0.1317198968458639,
|
9334 |
+
"grad_norm": 2.103409767150879,
|
9335 |
+
"learning_rate": 9.102957242043552e-06,
|
9336 |
+
"loss": 43.3022,
|
9337 |
+
"step": 1328
|
9338 |
+
},
|
9339 |
+
{
|
9340 |
+
"epoch": 0.13181908351517557,
|
9341 |
+
"grad_norm": 2.049328327178955,
|
9342 |
+
"learning_rate": 9.017386334019373e-06,
|
9343 |
+
"loss": 43.3698,
|
9344 |
+
"step": 1329
|
9345 |
+
},
|
9346 |
+
{
|
9347 |
+
"epoch": 0.1319182701844872,
|
9348 |
+
"grad_norm": 2.0538601875305176,
|
9349 |
+
"learning_rate": 8.932200531677537e-06,
|
9350 |
+
"loss": 43.3331,
|
9351 |
+
"step": 1330
|
9352 |
+
},
|
9353 |
+
{
|
9354 |
+
"epoch": 0.13201745685379884,
|
9355 |
+
"grad_norm": 1.9742823839187622,
|
9356 |
+
"learning_rate": 8.847400195587274e-06,
|
9357 |
+
"loss": 43.3083,
|
9358 |
+
"step": 1331
|
9359 |
+
},
|
9360 |
+
{
|
9361 |
+
"epoch": 0.1321166435231105,
|
9362 |
+
"grad_norm": 2.0585734844207764,
|
9363 |
+
"learning_rate": 8.762985684686365e-06,
|
9364 |
+
"loss": 43.3448,
|
9365 |
+
"step": 1332
|
9366 |
+
},
|
9367 |
+
{
|
9368 |
+
"epoch": 0.13221583019242214,
|
9369 |
+
"grad_norm": 2.1064953804016113,
|
9370 |
+
"learning_rate": 8.678957356279371e-06,
|
9371 |
+
"loss": 43.6047,
|
9372 |
+
"step": 1333
|
9373 |
+
},
|
9374 |
+
{
|
9375 |
+
"epoch": 0.13231501686173378,
|
9376 |
+
"grad_norm": 2.0612661838531494,
|
9377 |
+
"learning_rate": 8.595315566036333e-06,
|
9378 |
+
"loss": 43.451,
|
9379 |
+
"step": 1334
|
9380 |
+
},
|
9381 |
+
{
|
9382 |
+
"epoch": 0.13241420353104544,
|
9383 |
+
"grad_norm": 1.9173448085784912,
|
9384 |
+
"learning_rate": 8.512060667991117e-06,
|
9385 |
+
"loss": 43.3475,
|
9386 |
+
"step": 1335
|
9387 |
+
},
|
9388 |
+
{
|
9389 |
+
"epoch": 0.13251339020035707,
|
9390 |
+
"grad_norm": 1.8247857093811035,
|
9391 |
+
"learning_rate": 8.429193014540015e-06,
|
9392 |
+
"loss": 43.4886,
|
9393 |
+
"step": 1336
|
9394 |
+
},
|
9395 |
+
{
|
9396 |
+
"epoch": 0.1326125768696687,
|
9397 |
+
"grad_norm": 1.8412190675735474,
|
9398 |
+
"learning_rate": 8.346712956440195e-06,
|
9399 |
+
"loss": 43.2855,
|
9400 |
+
"step": 1337
|
9401 |
+
},
|
9402 |
+
{
|
9403 |
+
"epoch": 0.13271176353898037,
|
9404 |
+
"grad_norm": 1.8325763940811157,
|
9405 |
+
"learning_rate": 8.264620842808247e-06,
|
9406 |
+
"loss": 43.215,
|
9407 |
+
"step": 1338
|
9408 |
+
},
|
9409 |
+
{
|
9410 |
+
"epoch": 0.132810950208292,
|
9411 |
+
"grad_norm": 2.0761842727661133,
|
9412 |
+
"learning_rate": 8.182917021118663e-06,
|
9413 |
+
"loss": 43.1977,
|
9414 |
+
"step": 1339
|
9415 |
+
},
|
9416 |
+
{
|
9417 |
+
"epoch": 0.13291013687760364,
|
9418 |
+
"grad_norm": 2.371229410171509,
|
9419 |
+
"learning_rate": 8.101601837202456e-06,
|
9420 |
+
"loss": 43.4271,
|
9421 |
+
"step": 1340
|
9422 |
+
},
|
9423 |
+
{
|
9424 |
+
"epoch": 0.1330093235469153,
|
9425 |
+
"grad_norm": 2.003568410873413,
|
9426 |
+
"learning_rate": 8.020675635245567e-06,
|
9427 |
+
"loss": 43.716,
|
9428 |
+
"step": 1341
|
9429 |
+
},
|
9430 |
+
{
|
9431 |
+
"epoch": 0.13310851021622694,
|
9432 |
+
"grad_norm": 2.2959647178649902,
|
9433 |
+
"learning_rate": 7.940138757787507e-06,
|
9434 |
+
"loss": 42.9468,
|
9435 |
+
"step": 1342
|
9436 |
+
},
|
9437 |
+
{
|
9438 |
+
"epoch": 0.13320769688553857,
|
9439 |
+
"grad_norm": 2.332428455352783,
|
9440 |
+
"learning_rate": 7.859991545719902e-06,
|
9441 |
+
"loss": 43.3107,
|
9442 |
+
"step": 1343
|
9443 |
+
},
|
9444 |
+
{
|
9445 |
+
"epoch": 0.13330688355485024,
|
9446 |
+
"grad_norm": 2.177833080291748,
|
9447 |
+
"learning_rate": 7.780234338284974e-06,
|
9448 |
+
"loss": 43.4292,
|
9449 |
+
"step": 1344
|
9450 |
+
},
|
9451 |
+
{
|
9452 |
+
"epoch": 0.13340607022416187,
|
9453 |
+
"grad_norm": 2.025319814682007,
|
9454 |
+
"learning_rate": 7.700867473074224e-06,
|
9455 |
+
"loss": 43.2634,
|
9456 |
+
"step": 1345
|
9457 |
+
},
|
9458 |
+
{
|
9459 |
+
"epoch": 0.1335052568934735,
|
9460 |
+
"grad_norm": 1.993777871131897,
|
9461 |
+
"learning_rate": 7.621891286026872e-06,
|
9462 |
+
"loss": 43.3163,
|
9463 |
+
"step": 1346
|
9464 |
+
},
|
9465 |
+
{
|
9466 |
+
"epoch": 0.13360444356278517,
|
9467 |
+
"grad_norm": 1.8726677894592285,
|
9468 |
+
"learning_rate": 7.543306111428539e-06,
|
9469 |
+
"loss": 43.4174,
|
9470 |
+
"step": 1347
|
9471 |
+
},
|
9472 |
+
{
|
9473 |
+
"epoch": 0.1337036302320968,
|
9474 |
+
"grad_norm": 2.088778495788574,
|
9475 |
+
"learning_rate": 7.46511228190977e-06,
|
9476 |
+
"loss": 43.2981,
|
9477 |
+
"step": 1348
|
9478 |
+
},
|
9479 |
+
{
|
9480 |
+
"epoch": 0.13380281690140844,
|
9481 |
+
"grad_norm": 2.2077391147613525,
|
9482 |
+
"learning_rate": 7.38731012844468e-06,
|
9483 |
+
"loss": 43.5584,
|
9484 |
+
"step": 1349
|
9485 |
+
},
|
9486 |
+
{
|
9487 |
+
"epoch": 0.1339020035707201,
|
9488 |
+
"grad_norm": 2.162419319152832,
|
9489 |
+
"learning_rate": 7.309899980349455e-06,
|
9490 |
+
"loss": 43.2591,
|
9491 |
+
"step": 1350
|
9492 |
+
},
|
9493 |
+
{
|
9494 |
+
"epoch": 0.13400119024003174,
|
9495 |
+
"grad_norm": 2.255364418029785,
|
9496 |
+
"learning_rate": 7.232882165281141e-06,
|
9497 |
+
"loss": 42.9439,
|
9498 |
+
"step": 1351
|
9499 |
+
},
|
9500 |
+
{
|
9501 |
+
"epoch": 0.13410037690934337,
|
9502 |
+
"grad_norm": 1.9275546073913574,
|
9503 |
+
"learning_rate": 7.156257009236e-06,
|
9504 |
+
"loss": 43.1532,
|
9505 |
+
"step": 1352
|
9506 |
+
},
|
9507 |
+
{
|
9508 |
+
"epoch": 0.13419956357865503,
|
9509 |
+
"grad_norm": 1.801643967628479,
|
9510 |
+
"learning_rate": 7.0800248365483825e-06,
|
9511 |
+
"loss": 43.3247,
|
9512 |
+
"step": 1353
|
9513 |
+
},
|
9514 |
+
{
|
9515 |
+
"epoch": 0.13429875024796667,
|
9516 |
+
"grad_norm": 2.316209554672241,
|
9517 |
+
"learning_rate": 7.004185969889187e-06,
|
9518 |
+
"loss": 43.4831,
|
9519 |
+
"step": 1354
|
9520 |
+
},
|
9521 |
+
{
|
9522 |
+
"epoch": 0.1343979369172783,
|
9523 |
+
"grad_norm": 2.057708263397217,
|
9524 |
+
"learning_rate": 6.928740730264549e-06,
|
9525 |
+
"loss": 42.9086,
|
9526 |
+
"step": 1355
|
9527 |
+
},
|
9528 |
+
{
|
9529 |
+
"epoch": 0.13449712358658997,
|
9530 |
+
"grad_norm": 2.5561258792877197,
|
9531 |
+
"learning_rate": 6.853689437014521e-06,
|
9532 |
+
"loss": 43.3944,
|
9533 |
+
"step": 1356
|
9534 |
+
},
|
9535 |
+
{
|
9536 |
+
"epoch": 0.1345963102559016,
|
9537 |
+
"grad_norm": 1.672929048538208,
|
9538 |
+
"learning_rate": 6.7790324078116364e-06,
|
9539 |
+
"loss": 43.1944,
|
9540 |
+
"step": 1357
|
9541 |
+
},
|
9542 |
+
{
|
9543 |
+
"epoch": 0.13469549692521327,
|
9544 |
+
"grad_norm": 1.909250259399414,
|
9545 |
+
"learning_rate": 6.704769958659652e-06,
|
9546 |
+
"loss": 43.4062,
|
9547 |
+
"step": 1358
|
9548 |
+
},
|
9549 |
+
{
|
9550 |
+
"epoch": 0.1347946835945249,
|
9551 |
+
"grad_norm": 1.9488860368728638,
|
9552 |
+
"learning_rate": 6.6309024038921475e-06,
|
9553 |
+
"loss": 43.4522,
|
9554 |
+
"step": 1359
|
9555 |
+
},
|
9556 |
+
{
|
9557 |
+
"epoch": 0.13489387026383654,
|
9558 |
+
"grad_norm": 2.196932315826416,
|
9559 |
+
"learning_rate": 6.557430056171221e-06,
|
9560 |
+
"loss": 43.2967,
|
9561 |
+
"step": 1360
|
9562 |
+
},
|
9563 |
+
{
|
9564 |
+
"epoch": 0.1349930569331482,
|
9565 |
+
"grad_norm": 2.2892370223999023,
|
9566 |
+
"learning_rate": 6.484353226486162e-06,
|
9567 |
+
"loss": 43.4358,
|
9568 |
+
"step": 1361
|
9569 |
+
},
|
9570 |
+
{
|
9571 |
+
"epoch": 0.13509224360245983,
|
9572 |
+
"grad_norm": 2.0446407794952393,
|
9573 |
+
"learning_rate": 6.411672224152154e-06,
|
9574 |
+
"loss": 43.493,
|
9575 |
+
"step": 1362
|
9576 |
+
},
|
9577 |
+
{
|
9578 |
+
"epoch": 0.13519143027177147,
|
9579 |
+
"grad_norm": 1.935500979423523,
|
9580 |
+
"learning_rate": 6.339387356808912e-06,
|
9581 |
+
"loss": 43.3435,
|
9582 |
+
"step": 1363
|
9583 |
+
},
|
9584 |
+
{
|
9585 |
+
"epoch": 0.13529061694108313,
|
9586 |
+
"grad_norm": 1.5912455320358276,
|
9587 |
+
"learning_rate": 6.267498930419435e-06,
|
9588 |
+
"loss": 43.3201,
|
9589 |
+
"step": 1364
|
9590 |
+
},
|
9591 |
+
{
|
9592 |
+
"epoch": 0.13538980361039477,
|
9593 |
+
"grad_norm": 1.9594824314117432,
|
9594 |
+
"learning_rate": 6.196007249268687e-06,
|
9595 |
+
"loss": 43.4821,
|
9596 |
+
"step": 1365
|
9597 |
+
},
|
9598 |
+
{
|
9599 |
+
"epoch": 0.1354889902797064,
|
9600 |
+
"grad_norm": 2.056896924972534,
|
9601 |
+
"learning_rate": 6.124912615962341e-06,
|
9602 |
+
"loss": 43.3942,
|
9603 |
+
"step": 1366
|
9604 |
+
},
|
9605 |
+
{
|
9606 |
+
"epoch": 0.13558817694901806,
|
9607 |
+
"grad_norm": 1.7575795650482178,
|
9608 |
+
"learning_rate": 6.054215331425428e-06,
|
9609 |
+
"loss": 43.166,
|
9610 |
+
"step": 1367
|
9611 |
+
},
|
9612 |
+
{
|
9613 |
+
"epoch": 0.1356873636183297,
|
9614 |
+
"grad_norm": 2.0878756046295166,
|
9615 |
+
"learning_rate": 5.983915694901088e-06,
|
9616 |
+
"loss": 43.4634,
|
9617 |
+
"step": 1368
|
9618 |
+
},
|
9619 |
+
{
|
9620 |
+
"epoch": 0.13578655028764133,
|
9621 |
+
"grad_norm": 2.114713668823242,
|
9622 |
+
"learning_rate": 5.9140140039494084e-06,
|
9623 |
+
"loss": 43.2898,
|
9624 |
+
"step": 1369
|
9625 |
+
},
|
9626 |
+
{
|
9627 |
+
"epoch": 0.135885736956953,
|
9628 |
+
"grad_norm": 2.129774332046509,
|
9629 |
+
"learning_rate": 5.844510554445981e-06,
|
9630 |
+
"loss": 43.4494,
|
9631 |
+
"step": 1370
|
9632 |
+
},
|
9633 |
+
{
|
9634 |
+
"epoch": 0.13598492362626463,
|
9635 |
+
"grad_norm": 2.3913118839263916,
|
9636 |
+
"learning_rate": 5.775405640580822e-06,
|
9637 |
+
"loss": 43.3137,
|
9638 |
+
"step": 1371
|
9639 |
+
},
|
9640 |
+
{
|
9641 |
+
"epoch": 0.13608411029557627,
|
9642 |
+
"grad_norm": 1.7819161415100098,
|
9643 |
+
"learning_rate": 5.706699554856964e-06,
|
9644 |
+
"loss": 43.3058,
|
9645 |
+
"step": 1372
|
9646 |
+
},
|
9647 |
+
{
|
9648 |
+
"epoch": 0.13618329696488793,
|
9649 |
+
"grad_norm": 1.9535526037216187,
|
9650 |
+
"learning_rate": 5.6383925880894135e-06,
|
9651 |
+
"loss": 43.3178,
|
9652 |
+
"step": 1373
|
9653 |
+
},
|
9654 |
+
{
|
9655 |
+
"epoch": 0.13628248363419956,
|
9656 |
+
"grad_norm": 2.282865285873413,
|
9657 |
+
"learning_rate": 5.5704850294036934e-06,
|
9658 |
+
"loss": 43.0879,
|
9659 |
+
"step": 1374
|
9660 |
+
},
|
9661 |
+
{
|
9662 |
+
"epoch": 0.1363816703035112,
|
9663 |
+
"grad_norm": 1.8366732597351074,
|
9664 |
+
"learning_rate": 5.502977166234857e-06,
|
9665 |
+
"loss": 43.3269,
|
9666 |
+
"step": 1375
|
9667 |
+
},
|
9668 |
+
{
|
9669 |
+
"epoch": 0.13648085697282286,
|
9670 |
+
"grad_norm": 2.4548354148864746,
|
9671 |
+
"learning_rate": 5.435869284326023e-06,
|
9672 |
+
"loss": 43.1059,
|
9673 |
+
"step": 1376
|
9674 |
+
},
|
9675 |
+
{
|
9676 |
+
"epoch": 0.1365800436421345,
|
9677 |
+
"grad_norm": 2.022681474685669,
|
9678 |
+
"learning_rate": 5.3691616677274185e-06,
|
9679 |
+
"loss": 43.1864,
|
9680 |
+
"step": 1377
|
9681 |
+
},
|
9682 |
+
{
|
9683 |
+
"epoch": 0.13667923031144613,
|
9684 |
+
"grad_norm": 2.470262289047241,
|
9685 |
+
"learning_rate": 5.302854598794937e-06,
|
9686 |
+
"loss": 43.3405,
|
9687 |
+
"step": 1378
|
9688 |
+
},
|
9689 |
+
{
|
9690 |
+
"epoch": 0.1367784169807578,
|
9691 |
+
"grad_norm": 2.0381100177764893,
|
9692 |
+
"learning_rate": 5.2369483581891535e-06,
|
9693 |
+
"loss": 43.3638,
|
9694 |
+
"step": 1379
|
9695 |
+
},
|
9696 |
+
{
|
9697 |
+
"epoch": 0.13687760365006943,
|
9698 |
+
"grad_norm": 2.0829873085021973,
|
9699 |
+
"learning_rate": 5.171443224873995e-06,
|
9700 |
+
"loss": 43.4733,
|
9701 |
+
"step": 1380
|
9702 |
+
},
|
9703 |
+
{
|
9704 |
+
"epoch": 0.13697679031938106,
|
9705 |
+
"grad_norm": 2.2366280555725098,
|
9706 |
+
"learning_rate": 5.106339476115596e-06,
|
9707 |
+
"loss": 43.3411,
|
9708 |
+
"step": 1381
|
9709 |
+
},
|
9710 |
+
{
|
9711 |
+
"epoch": 0.13707597698869273,
|
9712 |
+
"grad_norm": 2.3511526584625244,
|
9713 |
+
"learning_rate": 5.0416373874811684e-06,
|
9714 |
+
"loss": 42.9051,
|
9715 |
+
"step": 1382
|
9716 |
+
},
|
9717 |
+
{
|
9718 |
+
"epoch": 0.13717516365800436,
|
9719 |
+
"grad_norm": 2.483473062515259,
|
9720 |
+
"learning_rate": 4.977337232837775e-06,
|
9721 |
+
"loss": 43.109,
|
9722 |
+
"step": 1383
|
9723 |
+
},
|
9724 |
+
{
|
9725 |
+
"epoch": 0.137274350327316,
|
9726 |
+
"grad_norm": 2.010577440261841,
|
9727 |
+
"learning_rate": 4.913439284351207e-06,
|
9728 |
+
"loss": 43.135,
|
9729 |
+
"step": 1384
|
9730 |
+
},
|
9731 |
+
{
|
9732 |
+
"epoch": 0.13737353699662766,
|
9733 |
+
"grad_norm": 2.0378925800323486,
|
9734 |
+
"learning_rate": 4.849943812484814e-06,
|
9735 |
+
"loss": 42.8639,
|
9736 |
+
"step": 1385
|
9737 |
+
},
|
9738 |
+
{
|
9739 |
+
"epoch": 0.1374727236659393,
|
9740 |
+
"grad_norm": 2.2532873153686523,
|
9741 |
+
"learning_rate": 4.7868510859983895e-06,
|
9742 |
+
"loss": 43.4281,
|
9743 |
+
"step": 1386
|
9744 |
+
},
|
9745 |
+
{
|
9746 |
+
"epoch": 0.13757191033525093,
|
9747 |
+
"grad_norm": 1.898999571800232,
|
9748 |
+
"learning_rate": 4.724161371946978e-06,
|
9749 |
+
"loss": 43.4366,
|
9750 |
+
"step": 1387
|
9751 |
+
},
|
9752 |
+
{
|
9753 |
+
"epoch": 0.1376710970045626,
|
9754 |
+
"grad_norm": 1.8824591636657715,
|
9755 |
+
"learning_rate": 4.661874935679822e-06,
|
9756 |
+
"loss": 43.4965,
|
9757 |
+
"step": 1388
|
9758 |
+
},
|
9759 |
+
{
|
9760 |
+
"epoch": 0.13777028367387423,
|
9761 |
+
"grad_norm": 2.134467124938965,
|
9762 |
+
"learning_rate": 4.599992040839118e-06,
|
9763 |
+
"loss": 43.329,
|
9764 |
+
"step": 1389
|
9765 |
+
},
|
9766 |
+
{
|
9767 |
+
"epoch": 0.13786947034318586,
|
9768 |
+
"grad_norm": 2.0056257247924805,
|
9769 |
+
"learning_rate": 4.538512949359075e-06,
|
9770 |
+
"loss": 43.0327,
|
9771 |
+
"step": 1390
|
9772 |
+
},
|
9773 |
+
{
|
9774 |
+
"epoch": 0.13796865701249753,
|
9775 |
+
"grad_norm": 2.1297647953033447,
|
9776 |
+
"learning_rate": 4.4774379214646225e-06,
|
9777 |
+
"loss": 43.2565,
|
9778 |
+
"step": 1391
|
9779 |
+
},
|
9780 |
+
{
|
9781 |
+
"epoch": 0.13806784368180916,
|
9782 |
+
"grad_norm": 1.80899977684021,
|
9783 |
+
"learning_rate": 4.416767215670448e-06,
|
9784 |
+
"loss": 43.0096,
|
9785 |
+
"step": 1392
|
9786 |
+
},
|
9787 |
+
{
|
9788 |
+
"epoch": 0.13816703035112082,
|
9789 |
+
"grad_norm": 2.1779046058654785,
|
9790 |
+
"learning_rate": 4.356501088779841e-06,
|
9791 |
+
"loss": 43.1146,
|
9792 |
+
"step": 1393
|
9793 |
+
},
|
9794 |
+
{
|
9795 |
+
"epoch": 0.13826621702043246,
|
9796 |
+
"grad_norm": 2.2874755859375,
|
9797 |
+
"learning_rate": 4.296639795883572e-06,
|
9798 |
+
"loss": 43.1322,
|
9799 |
+
"step": 1394
|
9800 |
+
},
|
9801 |
+
{
|
9802 |
+
"epoch": 0.1383654036897441,
|
9803 |
+
"grad_norm": 1.7321752309799194,
|
9804 |
+
"learning_rate": 4.237183590358928e-06,
|
9805 |
+
"loss": 43.2184,
|
9806 |
+
"step": 1395
|
9807 |
+
},
|
9808 |
+
{
|
9809 |
+
"epoch": 0.13846459035905576,
|
9810 |
+
"grad_norm": 2.2388689517974854,
|
9811 |
+
"learning_rate": 4.178132723868477e-06,
|
9812 |
+
"loss": 43.6094,
|
9813 |
+
"step": 1396
|
9814 |
+
},
|
9815 |
+
{
|
9816 |
+
"epoch": 0.1385637770283674,
|
9817 |
+
"grad_norm": 1.992551565170288,
|
9818 |
+
"learning_rate": 4.119487446359193e-06,
|
9819 |
+
"loss": 43.1707,
|
9820 |
+
"step": 1397
|
9821 |
+
},
|
9822 |
+
{
|
9823 |
+
"epoch": 0.13866296369767903,
|
9824 |
+
"grad_norm": 2.0371286869049072,
|
9825 |
+
"learning_rate": 4.061248006061191e-06,
|
9826 |
+
"loss": 43.2199,
|
9827 |
+
"step": 1398
|
9828 |
+
},
|
9829 |
+
{
|
9830 |
+
"epoch": 0.1387621503669907,
|
9831 |
+
"grad_norm": 2.181845188140869,
|
9832 |
+
"learning_rate": 4.003414649486892e-06,
|
9833 |
+
"loss": 43.2273,
|
9834 |
+
"step": 1399
|
9835 |
+
},
|
9836 |
+
{
|
9837 |
+
"epoch": 0.13886133703630232,
|
9838 |
+
"grad_norm": 1.8432098627090454,
|
9839 |
+
"learning_rate": 3.94598762142977e-06,
|
9840 |
+
"loss": 43.3462,
|
9841 |
+
"step": 1400
|
9842 |
+
},
|
9843 |
+
{
|
9844 |
+
"epoch": 0.13896052370561396,
|
9845 |
+
"grad_norm": 1.937048077583313,
|
9846 |
+
"learning_rate": 3.888967164963498e-06,
|
9847 |
+
"loss": 43.1906,
|
9848 |
+
"step": 1401
|
9849 |
+
},
|
9850 |
+
{
|
9851 |
+
"epoch": 0.13905971037492562,
|
9852 |
+
"grad_norm": 1.9433784484863281,
|
9853 |
+
"learning_rate": 3.832353521440768e-06,
|
9854 |
+
"loss": 43.3311,
|
9855 |
+
"step": 1402
|
9856 |
+
},
|
9857 |
+
{
|
9858 |
+
"epoch": 0.13915889704423726,
|
9859 |
+
"grad_norm": 2.5950987339019775,
|
9860 |
+
"learning_rate": 3.7761469304923967e-06,
|
9861 |
+
"loss": 43.1959,
|
9862 |
+
"step": 1403
|
9863 |
+
},
|
9864 |
+
{
|
9865 |
+
"epoch": 0.1392580837135489,
|
9866 |
+
"grad_norm": 2.0187675952911377,
|
9867 |
+
"learning_rate": 3.720347630026222e-06,
|
9868 |
+
"loss": 43.417,
|
9869 |
+
"step": 1404
|
9870 |
+
},
|
9871 |
+
{
|
9872 |
+
"epoch": 0.13935727038286055,
|
9873 |
+
"grad_norm": 2.4105563163757324,
|
9874 |
+
"learning_rate": 3.6649558562261375e-06,
|
9875 |
+
"loss": 43.4502,
|
9876 |
+
"step": 1405
|
9877 |
+
},
|
9878 |
+
{
|
9879 |
+
"epoch": 0.1394564570521722,
|
9880 |
+
"grad_norm": 2.395488739013672,
|
9881 |
+
"learning_rate": 3.6099718435510633e-06,
|
9882 |
+
"loss": 43.0515,
|
9883 |
+
"step": 1406
|
9884 |
+
},
|
9885 |
+
{
|
9886 |
+
"epoch": 0.13955564372148382,
|
9887 |
+
"grad_norm": 1.9171357154846191,
|
9888 |
+
"learning_rate": 3.555395824734009e-06,
|
9889 |
+
"loss": 43.3508,
|
9890 |
+
"step": 1407
|
9891 |
+
},
|
9892 |
+
{
|
9893 |
+
"epoch": 0.1396548303907955,
|
9894 |
+
"grad_norm": 1.94326913356781,
|
9895 |
+
"learning_rate": 3.501228030781034e-06,
|
9896 |
+
"loss": 43.2267,
|
9897 |
+
"step": 1408
|
9898 |
+
},
|
9899 |
+
{
|
9900 |
+
"epoch": 0.13975401706010712,
|
9901 |
+
"grad_norm": 2.010206699371338,
|
9902 |
+
"learning_rate": 3.447468690970268e-06,
|
9903 |
+
"loss": 43.3639,
|
9904 |
+
"step": 1409
|
9905 |
+
},
|
9906 |
+
{
|
9907 |
+
"epoch": 0.13985320372941876,
|
9908 |
+
"grad_norm": 2.325780153274536,
|
9909 |
+
"learning_rate": 3.3941180328510123e-06,
|
9910 |
+
"loss": 43.2116,
|
9911 |
+
"step": 1410
|
9912 |
+
},
|
9913 |
+
{
|
9914 |
+
"epoch": 0.13995239039873042,
|
9915 |
+
"grad_norm": 2.193591356277466,
|
9916 |
+
"learning_rate": 3.341176282242653e-06,
|
9917 |
+
"loss": 43.3946,
|
9918 |
+
"step": 1411
|
9919 |
+
},
|
9920 |
+
{
|
9921 |
+
"epoch": 0.14005157706804205,
|
9922 |
+
"grad_norm": 2.0868961811065674,
|
9923 |
+
"learning_rate": 3.288643663233848e-06,
|
9924 |
+
"loss": 43.1527,
|
9925 |
+
"step": 1412
|
9926 |
+
},
|
9927 |
+
{
|
9928 |
+
"epoch": 0.1401507637373537,
|
9929 |
+
"grad_norm": 1.874427080154419,
|
9930 |
+
"learning_rate": 3.236520398181464e-06,
|
9931 |
+
"loss": 43.3895,
|
9932 |
+
"step": 1413
|
9933 |
+
},
|
9934 |
+
{
|
9935 |
+
"epoch": 0.14024995040666535,
|
9936 |
+
"grad_norm": 1.8140789270401,
|
9937 |
+
"learning_rate": 3.184806707709698e-06,
|
9938 |
+
"loss": 43.1514,
|
9939 |
+
"step": 1414
|
9940 |
+
},
|
9941 |
+
{
|
9942 |
+
"epoch": 0.140349137075977,
|
9943 |
+
"grad_norm": 1.7428981065750122,
|
9944 |
+
"learning_rate": 3.133502810709099e-06,
|
9945 |
+
"loss": 43.4509,
|
9946 |
+
"step": 1415
|
9947 |
+
},
|
9948 |
+
{
|
9949 |
+
"epoch": 0.14044832374528862,
|
9950 |
+
"grad_norm": 2.1699752807617188,
|
9951 |
+
"learning_rate": 3.082608924335717e-06,
|
9952 |
+
"loss": 43.1771,
|
9953 |
+
"step": 1416
|
9954 |
+
},
|
9955 |
+
{
|
9956 |
+
"epoch": 0.14054751041460029,
|
9957 |
+
"grad_norm": 2.0626633167266846,
|
9958 |
+
"learning_rate": 3.0321252640100885e-06,
|
9959 |
+
"loss": 43.5371,
|
9960 |
+
"step": 1417
|
9961 |
+
},
|
9962 |
+
{
|
9963 |
+
"epoch": 0.14064669708391192,
|
9964 |
+
"grad_norm": 2.053748607635498,
|
9965 |
+
"learning_rate": 2.9820520434163836e-06,
|
9966 |
+
"loss": 43.4088,
|
9967 |
+
"step": 1418
|
9968 |
+
},
|
9969 |
+
{
|
9970 |
+
"epoch": 0.14074588375322356,
|
9971 |
+
"grad_norm": 2.031611442565918,
|
9972 |
+
"learning_rate": 2.932389474501496e-06,
|
9973 |
+
"loss": 43.6168,
|
9974 |
+
"step": 1419
|
9975 |
+
},
|
9976 |
+
{
|
9977 |
+
"epoch": 0.14084507042253522,
|
9978 |
+
"grad_norm": 2.2878761291503906,
|
9979 |
+
"learning_rate": 2.88313776747412e-06,
|
9980 |
+
"loss": 43.3679,
|
9981 |
+
"step": 1420
|
9982 |
+
},
|
9983 |
+
{
|
9984 |
+
"epoch": 0.14094425709184685,
|
9985 |
+
"grad_norm": 1.8133399486541748,
|
9986 |
+
"learning_rate": 2.834297130803909e-06,
|
9987 |
+
"loss": 43.2068,
|
9988 |
+
"step": 1421
|
9989 |
+
},
|
9990 |
+
{
|
9991 |
+
"epoch": 0.1410434437611585,
|
9992 |
+
"grad_norm": 2.0135884284973145,
|
9993 |
+
"learning_rate": 2.7858677712205296e-06,
|
9994 |
+
"loss": 43.1477,
|
9995 |
+
"step": 1422
|
9996 |
+
},
|
9997 |
+
{
|
9998 |
+
"epoch": 0.14114263043047015,
|
9999 |
+
"grad_norm": 1.9554595947265625,
|
10000 |
+
"learning_rate": 2.7378498937128404e-06,
|
10001 |
+
"loss": 43.3352,
|
10002 |
+
"step": 1423
|
10003 |
+
},
|
10004 |
+
{
|
10005 |
+
"epoch": 0.14124181709978179,
|
10006 |
+
"grad_norm": 1.9702895879745483,
|
10007 |
+
"learning_rate": 2.6902437015279837e-06,
|
10008 |
+
"loss": 43.4386,
|
10009 |
+
"step": 1424
|
10010 |
+
},
|
10011 |
+
{
|
10012 |
+
"epoch": 0.14134100376909342,
|
10013 |
+
"grad_norm": 2.0473744869232178,
|
10014 |
+
"learning_rate": 2.6430493961705937e-06,
|
10015 |
+
"loss": 43.3568,
|
10016 |
+
"step": 1425
|
10017 |
+
},
|
10018 |
+
{
|
10019 |
+
"epoch": 0.14144019043840508,
|
10020 |
+
"grad_norm": 2.1810948848724365,
|
10021 |
+
"learning_rate": 2.5962671774018234e-06,
|
10022 |
+
"loss": 42.9775,
|
10023 |
+
"step": 1426
|
10024 |
+
},
|
10025 |
+
{
|
10026 |
+
"epoch": 0.14153937710771672,
|
10027 |
+
"grad_norm": 1.9952651262283325,
|
10028 |
+
"learning_rate": 2.549897243238619e-06,
|
10029 |
+
"loss": 43.1992,
|
10030 |
+
"step": 1427
|
10031 |
+
},
|
10032 |
+
{
|
10033 |
+
"epoch": 0.14163856377702838,
|
10034 |
+
"grad_norm": 2.113675832748413,
|
10035 |
+
"learning_rate": 2.503939789952825e-06,
|
10036 |
+
"loss": 43.426,
|
10037 |
+
"step": 1428
|
10038 |
+
},
|
10039 |
+
{
|
10040 |
+
"epoch": 0.14173775044634002,
|
10041 |
+
"grad_norm": 1.946372628211975,
|
10042 |
+
"learning_rate": 2.458395012070369e-06,
|
10043 |
+
"loss": 43.1353,
|
10044 |
+
"step": 1429
|
10045 |
+
},
|
10046 |
+
{
|
10047 |
+
"epoch": 0.14183693711565165,
|
10048 |
+
"grad_norm": 2.023200750350952,
|
10049 |
+
"learning_rate": 2.4132631023704e-06,
|
10050 |
+
"loss": 43.2011,
|
10051 |
+
"step": 1430
|
10052 |
+
},
|
10053 |
+
{
|
10054 |
+
"epoch": 0.14193612378496331,
|
10055 |
+
"grad_norm": 2.0508205890655518,
|
10056 |
+
"learning_rate": 2.3685442518845522e-06,
|
10057 |
+
"loss": 43.2194,
|
10058 |
+
"step": 1431
|
10059 |
+
},
|
10060 |
+
{
|
10061 |
+
"epoch": 0.14203531045427495,
|
10062 |
+
"grad_norm": 2.032804489135742,
|
10063 |
+
"learning_rate": 2.3242386498960266e-06,
|
10064 |
+
"loss": 43.2455,
|
10065 |
+
"step": 1432
|
10066 |
+
},
|
10067 |
+
{
|
10068 |
+
"epoch": 0.14213449712358658,
|
10069 |
+
"grad_norm": 2.1437833309173584,
|
10070 |
+
"learning_rate": 2.2803464839388998e-06,
|
10071 |
+
"loss": 43.2066,
|
10072 |
+
"step": 1433
|
10073 |
+
},
|
10074 |
+
{
|
10075 |
+
"epoch": 0.14223368379289825,
|
10076 |
+
"grad_norm": 2.121809482574463,
|
10077 |
+
"learning_rate": 2.2368679397972824e-06,
|
10078 |
+
"loss": 43.0669,
|
10079 |
+
"step": 1434
|
10080 |
+
},
|
10081 |
+
{
|
10082 |
+
"epoch": 0.14233287046220988,
|
10083 |
+
"grad_norm": 1.9407131671905518,
|
10084 |
+
"learning_rate": 2.1938032015044964e-06,
|
10085 |
+
"loss": 43.597,
|
10086 |
+
"step": 1435
|
10087 |
+
},
|
10088 |
+
{
|
10089 |
+
"epoch": 0.14243205713152152,
|
10090 |
+
"grad_norm": 1.904272198677063,
|
10091 |
+
"learning_rate": 2.151152451342364e-06,
|
10092 |
+
"loss": 43.0514,
|
10093 |
+
"step": 1436
|
10094 |
+
},
|
10095 |
+
{
|
10096 |
+
"epoch": 0.14253124380083318,
|
10097 |
+
"grad_norm": 2.0880041122436523,
|
10098 |
+
"learning_rate": 2.1089158698403665e-06,
|
10099 |
+
"loss": 43.1989,
|
10100 |
+
"step": 1437
|
10101 |
+
},
|
10102 |
+
{
|
10103 |
+
"epoch": 0.14263043047014481,
|
10104 |
+
"grad_norm": 1.813359260559082,
|
10105 |
+
"learning_rate": 2.067093635774975e-06,
|
10106 |
+
"loss": 43.3236,
|
10107 |
+
"step": 1438
|
10108 |
+
},
|
10109 |
+
{
|
10110 |
+
"epoch": 0.14272961713945645,
|
10111 |
+
"grad_norm": 2.1034488677978516,
|
10112 |
+
"learning_rate": 2.0256859261687856e-06,
|
10113 |
+
"loss": 43.2322,
|
10114 |
+
"step": 1439
|
10115 |
+
},
|
10116 |
+
{
|
10117 |
+
"epoch": 0.1428288038087681,
|
10118 |
+
"grad_norm": 1.979280948638916,
|
10119 |
+
"learning_rate": 1.984692916289832e-06,
|
10120 |
+
"loss": 43.4587,
|
10121 |
+
"step": 1440
|
10122 |
+
},
|
10123 |
+
{
|
10124 |
+
"epoch": 0.14292799047807975,
|
10125 |
+
"grad_norm": 2.622344732284546,
|
10126 |
+
"learning_rate": 1.9441147796508407e-06,
|
10127 |
+
"loss": 43.1863,
|
10128 |
+
"step": 1441
|
10129 |
+
},
|
10130 |
+
{
|
10131 |
+
"epoch": 0.14302717714739138,
|
10132 |
+
"grad_norm": 2.316920518875122,
|
10133 |
+
"learning_rate": 1.9039516880084873e-06,
|
10134 |
+
"loss": 43.2889,
|
10135 |
+
"step": 1442
|
10136 |
+
},
|
10137 |
+
{
|
10138 |
+
"epoch": 0.14312636381670304,
|
10139 |
+
"grad_norm": 1.8629003763198853,
|
10140 |
+
"learning_rate": 1.8642038113626635e-06,
|
10141 |
+
"loss": 43.2849,
|
10142 |
+
"step": 1443
|
10143 |
+
},
|
10144 |
+
{
|
10145 |
+
"epoch": 0.14322555048601468,
|
10146 |
+
"grad_norm": 1.8128821849822998,
|
10147 |
+
"learning_rate": 1.8248713179557786e-06,
|
10148 |
+
"loss": 43.5935,
|
10149 |
+
"step": 1444
|
10150 |
+
},
|
10151 |
+
{
|
10152 |
+
"epoch": 0.14332473715532631,
|
10153 |
+
"grad_norm": 2.1895546913146973,
|
10154 |
+
"learning_rate": 1.785954374272003e-06,
|
10155 |
+
"loss": 43.1179,
|
10156 |
+
"step": 1445
|
10157 |
+
},
|
10158 |
+
{
|
10159 |
+
"epoch": 0.14342392382463798,
|
10160 |
+
"grad_norm": 1.8723267316818237,
|
10161 |
+
"learning_rate": 1.7474531450366371e-06,
|
10162 |
+
"loss": 43.4893,
|
10163 |
+
"step": 1446
|
10164 |
+
},
|
10165 |
+
{
|
10166 |
+
"epoch": 0.1435231104939496,
|
10167 |
+
"grad_norm": 2.146305799484253,
|
10168 |
+
"learning_rate": 1.7093677932153218e-06,
|
10169 |
+
"loss": 43.4065,
|
10170 |
+
"step": 1447
|
10171 |
+
},
|
10172 |
+
{
|
10173 |
+
"epoch": 0.14362229716326125,
|
10174 |
+
"grad_norm": 2.2784523963928223,
|
10175 |
+
"learning_rate": 1.6716984800134282e-06,
|
10176 |
+
"loss": 43.2687,
|
10177 |
+
"step": 1448
|
10178 |
+
},
|
10179 |
+
{
|
10180 |
+
"epoch": 0.1437214838325729,
|
10181 |
+
"grad_norm": 2.066173791885376,
|
10182 |
+
"learning_rate": 1.6344453648753588e-06,
|
10183 |
+
"loss": 43.2307,
|
10184 |
+
"step": 1449
|
10185 |
+
},
|
10186 |
+
{
|
10187 |
+
"epoch": 0.14382067050188455,
|
10188 |
+
"grad_norm": 2.0641512870788574,
|
10189 |
+
"learning_rate": 1.5976086054838025e-06,
|
10190 |
+
"loss": 43.3932,
|
10191 |
+
"step": 1450
|
10192 |
+
},
|
10193 |
+
{
|
10194 |
+
"epoch": 0.14391985717119618,
|
10195 |
+
"grad_norm": 1.8807092905044556,
|
10196 |
+
"learning_rate": 1.5611883577591912e-06,
|
10197 |
+
"loss": 43.2905,
|
10198 |
+
"step": 1451
|
10199 |
+
},
|
10200 |
+
{
|
10201 |
+
"epoch": 0.14401904384050784,
|
10202 |
+
"grad_norm": 2.4049503803253174,
|
10203 |
+
"learning_rate": 1.5251847758589121e-06,
|
10204 |
+
"loss": 43.0874,
|
10205 |
+
"step": 1452
|
10206 |
+
},
|
10207 |
+
{
|
10208 |
+
"epoch": 0.14411823050981948,
|
10209 |
+
"grad_norm": 2.178377389907837,
|
10210 |
+
"learning_rate": 1.4895980121767627e-06,
|
10211 |
+
"loss": 43.3295,
|
10212 |
+
"step": 1453
|
10213 |
+
},
|
10214 |
+
{
|
10215 |
+
"epoch": 0.1442174171791311,
|
10216 |
+
"grad_norm": 2.1683950424194336,
|
10217 |
+
"learning_rate": 1.4544282173422074e-06,
|
10218 |
+
"loss": 43.3708,
|
10219 |
+
"step": 1454
|
10220 |
+
},
|
10221 |
+
{
|
10222 |
+
"epoch": 0.14431660384844278,
|
10223 |
+
"grad_norm": 2.144456624984741,
|
10224 |
+
"learning_rate": 1.4196755402198448e-06,
|
10225 |
+
"loss": 43.3268,
|
10226 |
+
"step": 1455
|
10227 |
+
},
|
10228 |
+
{
|
10229 |
+
"epoch": 0.1444157905177544,
|
10230 |
+
"grad_norm": 2.1770358085632324,
|
10231 |
+
"learning_rate": 1.3853401279086854e-06,
|
10232 |
+
"loss": 43.3516,
|
10233 |
+
"step": 1456
|
10234 |
+
},
|
10235 |
+
{
|
10236 |
+
"epoch": 0.14451497718706605,
|
10237 |
+
"grad_norm": 2.203105926513672,
|
10238 |
+
"learning_rate": 1.3514221257415748e-06,
|
10239 |
+
"loss": 43.2459,
|
10240 |
+
"step": 1457
|
10241 |
+
},
|
10242 |
+
{
|
10243 |
+
"epoch": 0.1446141638563777,
|
10244 |
+
"grad_norm": 1.9053125381469727,
|
10245 |
+
"learning_rate": 1.3179216772845727e-06,
|
10246 |
+
"loss": 43.2546,
|
10247 |
+
"step": 1458
|
10248 |
+
},
|
10249 |
+
{
|
10250 |
+
"epoch": 0.14471335052568934,
|
10251 |
+
"grad_norm": 2.0264782905578613,
|
10252 |
+
"learning_rate": 1.2848389243363512e-06,
|
10253 |
+
"loss": 43.2631,
|
10254 |
+
"step": 1459
|
10255 |
+
},
|
10256 |
+
{
|
10257 |
+
"epoch": 0.14481253719500098,
|
10258 |
+
"grad_norm": 1.9565973281860352,
|
10259 |
+
"learning_rate": 1.2521740069275644e-06,
|
10260 |
+
"loss": 43.4172,
|
10261 |
+
"step": 1460
|
10262 |
+
},
|
10263 |
+
{
|
10264 |
+
"epoch": 0.14491172386431264,
|
10265 |
+
"grad_norm": 1.9662400484085083,
|
10266 |
+
"learning_rate": 1.219927063320303e-06,
|
10267 |
+
"loss": 43.2851,
|
10268 |
+
"step": 1461
|
10269 |
+
},
|
10270 |
+
{
|
10271 |
+
"epoch": 0.14501091053362428,
|
10272 |
+
"grad_norm": 1.6376630067825317,
|
10273 |
+
"learning_rate": 1.1880982300074838e-06,
|
10274 |
+
"loss": 43.4667,
|
10275 |
+
"step": 1462
|
10276 |
+
},
|
10277 |
+
{
|
10278 |
+
"epoch": 0.14511009720293594,
|
10279 |
+
"grad_norm": 2.073845863342285,
|
10280 |
+
"learning_rate": 1.1566876417122285e-06,
|
10281 |
+
"loss": 43.1126,
|
10282 |
+
"step": 1463
|
10283 |
+
},
|
10284 |
+
{
|
10285 |
+
"epoch": 0.14520928387224757,
|
10286 |
+
"grad_norm": 2.2647864818573,
|
10287 |
+
"learning_rate": 1.1256954313873858e-06,
|
10288 |
+
"loss": 43.2001,
|
10289 |
+
"step": 1464
|
10290 |
+
},
|
10291 |
+
{
|
10292 |
+
"epoch": 0.1453084705415592,
|
10293 |
+
"grad_norm": 2.2635481357574463,
|
10294 |
+
"learning_rate": 1.0951217302148986e-06,
|
10295 |
+
"loss": 43.2655,
|
10296 |
+
"step": 1465
|
10297 |
+
},
|
10298 |
+
{
|
10299 |
+
"epoch": 0.14540765721087087,
|
10300 |
+
"grad_norm": 2.0819239616394043,
|
10301 |
+
"learning_rate": 1.0649666676052828e-06,
|
10302 |
+
"loss": 43.3746,
|
10303 |
+
"step": 1466
|
10304 |
+
},
|
10305 |
+
{
|
10306 |
+
"epoch": 0.1455068438801825,
|
10307 |
+
"grad_norm": 1.84273362159729,
|
10308 |
+
"learning_rate": 1.0352303711970379e-06,
|
10309 |
+
"loss": 43.547,
|
10310 |
+
"step": 1467
|
10311 |
+
},
|
10312 |
+
{
|
10313 |
+
"epoch": 0.14560603054949414,
|
10314 |
+
"grad_norm": 1.9994412660598755,
|
10315 |
+
"learning_rate": 1.0059129668561707e-06,
|
10316 |
+
"loss": 43.311,
|
10317 |
+
"step": 1468
|
10318 |
+
},
|
10319 |
+
{
|
10320 |
+
"epoch": 0.1457052172188058,
|
10321 |
+
"grad_norm": 1.9508692026138306,
|
10322 |
+
"learning_rate": 9.770145786755946e-07,
|
10323 |
+
"loss": 43.2921,
|
10324 |
+
"step": 1469
|
10325 |
+
},
|
10326 |
+
{
|
10327 |
+
"epoch": 0.14580440388811744,
|
10328 |
+
"grad_norm": 2.2198736667633057,
|
10329 |
+
"learning_rate": 9.485353289746645e-07,
|
10330 |
+
"loss": 43.4834,
|
10331 |
+
"step": 1470
|
10332 |
+
},
|
10333 |
+
{
|
10334 |
+
"epoch": 0.14590359055742907,
|
10335 |
+
"grad_norm": 1.7677615880966187,
|
10336 |
+
"learning_rate": 9.204753382986097e-07,
|
10337 |
+
"loss": 43.3114,
|
10338 |
+
"step": 1471
|
10339 |
+
},
|
10340 |
+
{
|
10341 |
+
"epoch": 0.14600277722674074,
|
10342 |
+
"grad_norm": 2.2032389640808105,
|
10343 |
+
"learning_rate": 8.92834725418068e-07,
|
10344 |
+
"loss": 43.2516,
|
10345 |
+
"step": 1472
|
10346 |
+
},
|
10347 |
+
{
|
10348 |
+
"epoch": 0.14610196389605237,
|
10349 |
+
"grad_norm": 1.672240972518921,
|
10350 |
+
"learning_rate": 8.656136073285303e-07,
|
10351 |
+
"loss": 43.4628,
|
10352 |
+
"step": 1473
|
10353 |
+
},
|
10354 |
+
{
|
10355 |
+
"epoch": 0.146201150565364,
|
10356 |
+
"grad_norm": 2.144817352294922,
|
10357 |
+
"learning_rate": 8.388120992499083e-07,
|
10358 |
+
"loss": 43.451,
|
10359 |
+
"step": 1474
|
10360 |
+
},
|
10361 |
+
{
|
10362 |
+
"epoch": 0.14630033723467567,
|
10363 |
+
"grad_norm": 2.099653482437134,
|
10364 |
+
"learning_rate": 8.124303146259782e-07,
|
10365 |
+
"loss": 43.2155,
|
10366 |
+
"step": 1475
|
10367 |
+
},
|
10368 |
+
{
|
10369 |
+
"epoch": 0.1463995239039873,
|
10370 |
+
"grad_norm": 2.6512537002563477,
|
10371 |
+
"learning_rate": 7.864683651239824e-07,
|
10372 |
+
"loss": 42.983,
|
10373 |
+
"step": 1476
|
10374 |
+
},
|
10375 |
+
{
|
10376 |
+
"epoch": 0.14649871057329894,
|
10377 |
+
"grad_norm": 2.3182458877563477,
|
10378 |
+
"learning_rate": 7.609263606340622e-07,
|
10379 |
+
"loss": 43.3022,
|
10380 |
+
"step": 1477
|
10381 |
+
},
|
10382 |
+
{
|
10383 |
+
"epoch": 0.1465978972426106,
|
10384 |
+
"grad_norm": 2.0440011024475098,
|
10385 |
+
"learning_rate": 7.358044092688699e-07,
|
10386 |
+
"loss": 43.3016,
|
10387 |
+
"step": 1478
|
10388 |
+
},
|
10389 |
+
{
|
10390 |
+
"epoch": 0.14669708391192224,
|
10391 |
+
"grad_norm": 2.181492805480957,
|
10392 |
+
"learning_rate": 7.111026173630797e-07,
|
10393 |
+
"loss": 43.1403,
|
10394 |
+
"step": 1479
|
10395 |
+
},
|
10396 |
+
{
|
10397 |
+
"epoch": 0.14679627058123387,
|
10398 |
+
"grad_norm": 1.6637905836105347,
|
10399 |
+
"learning_rate": 6.868210894729332e-07,
|
10400 |
+
"loss": 43.4882,
|
10401 |
+
"step": 1480
|
10402 |
+
},
|
10403 |
+
{
|
10404 |
+
"epoch": 0.14689545725054554,
|
10405 |
+
"grad_norm": 2.085326671600342,
|
10406 |
+
"learning_rate": 6.629599283758059e-07,
|
10407 |
+
"loss": 43.2251,
|
10408 |
+
"step": 1481
|
10409 |
+
},
|
10410 |
+
{
|
10411 |
+
"epoch": 0.14699464391985717,
|
10412 |
+
"grad_norm": 2.2099697589874268,
|
10413 |
+
"learning_rate": 6.395192350697743e-07,
|
10414 |
+
"loss": 43.2126,
|
10415 |
+
"step": 1482
|
10416 |
+
},
|
10417 |
+
{
|
10418 |
+
"epoch": 0.1470938305891688,
|
10419 |
+
"grad_norm": 2.3335022926330566,
|
10420 |
+
"learning_rate": 6.164991087731831e-07,
|
10421 |
+
"loss": 43.3794,
|
10422 |
+
"step": 1483
|
10423 |
+
},
|
10424 |
+
{
|
10425 |
+
"epoch": 0.14719301725848047,
|
10426 |
+
"grad_norm": 1.7655500173568726,
|
10427 |
+
"learning_rate": 5.938996469242231e-07,
|
10428 |
+
"loss": 43.4807,
|
10429 |
+
"step": 1484
|
10430 |
+
},
|
10431 |
+
{
|
10432 |
+
"epoch": 0.1472922039277921,
|
10433 |
+
"grad_norm": 2.487640380859375,
|
10434 |
+
"learning_rate": 5.717209451805205e-07,
|
10435 |
+
"loss": 42.9848,
|
10436 |
+
"step": 1485
|
10437 |
+
},
|
10438 |
+
{
|
10439 |
+
"epoch": 0.14739139059710374,
|
10440 |
+
"grad_norm": 1.804437279701233,
|
10441 |
+
"learning_rate": 5.499630974187375e-07,
|
10442 |
+
"loss": 43.4311,
|
10443 |
+
"step": 1486
|
10444 |
+
},
|
10445 |
+
{
|
10446 |
+
"epoch": 0.1474905772664154,
|
10447 |
+
"grad_norm": 2.1954946517944336,
|
10448 |
+
"learning_rate": 5.286261957341831e-07,
|
10449 |
+
"loss": 42.9627,
|
10450 |
+
"step": 1487
|
10451 |
+
},
|
10452 |
+
{
|
10453 |
+
"epoch": 0.14758976393572704,
|
10454 |
+
"grad_norm": 2.0663084983825684,
|
10455 |
+
"learning_rate": 5.077103304403807e-07,
|
10456 |
+
"loss": 43.292,
|
10457 |
+
"step": 1488
|
10458 |
+
},
|
10459 |
+
{
|
10460 |
+
"epoch": 0.14768895060503867,
|
10461 |
+
"grad_norm": 2.280759811401367,
|
10462 |
+
"learning_rate": 4.872155900687347e-07,
|
10463 |
+
"loss": 43.044,
|
10464 |
+
"step": 1489
|
10465 |
+
},
|
10466 |
+
{
|
10467 |
+
"epoch": 0.14778813727435033,
|
10468 |
+
"grad_norm": 2.1535468101501465,
|
10469 |
+
"learning_rate": 4.671420613681643e-07,
|
10470 |
+
"loss": 43.2001,
|
10471 |
+
"step": 1490
|
10472 |
+
},
|
10473 |
+
{
|
10474 |
+
"epoch": 0.14788732394366197,
|
10475 |
+
"grad_norm": 2.0268213748931885,
|
10476 |
+
"learning_rate": 4.4748982930465923e-07,
|
10477 |
+
"loss": 43.3799,
|
10478 |
+
"step": 1491
|
10479 |
+
},
|
10480 |
+
{
|
10481 |
+
"epoch": 0.1479865106129736,
|
10482 |
+
"grad_norm": 1.9153803586959839,
|
10483 |
+
"learning_rate": 4.2825897706100235e-07,
|
10484 |
+
"loss": 43.1269,
|
10485 |
+
"step": 1492
|
10486 |
+
},
|
10487 |
+
{
|
10488 |
+
"epoch": 0.14808569728228527,
|
10489 |
+
"grad_norm": 2.1148533821105957,
|
10490 |
+
"learning_rate": 4.09449586036359e-07,
|
10491 |
+
"loss": 43.2859,
|
10492 |
+
"step": 1493
|
10493 |
+
},
|
10494 |
+
{
|
10495 |
+
"epoch": 0.1481848839515969,
|
10496 |
+
"grad_norm": 2.9361672401428223,
|
10497 |
+
"learning_rate": 3.9106173584601e-07,
|
10498 |
+
"loss": 43.5126,
|
10499 |
+
"step": 1494
|
10500 |
+
},
|
10501 |
+
{
|
10502 |
+
"epoch": 0.14828407062090854,
|
10503 |
+
"grad_norm": 2.002664804458618,
|
10504 |
+
"learning_rate": 3.7309550432090835e-07,
|
10505 |
+
"loss": 43.1279,
|
10506 |
+
"step": 1495
|
10507 |
+
},
|
10508 |
+
{
|
10509 |
+
"epoch": 0.1483832572902202,
|
10510 |
+
"grad_norm": 2.234067678451538,
|
10511 |
+
"learning_rate": 3.5555096750743424e-07,
|
10512 |
+
"loss": 43.5213,
|
10513 |
+
"step": 1496
|
10514 |
+
},
|
10515 |
+
{
|
10516 |
+
"epoch": 0.14848244395953183,
|
10517 |
+
"grad_norm": 2.1555521488189697,
|
10518 |
+
"learning_rate": 3.384281996670402e-07,
|
10519 |
+
"loss": 43.3817,
|
10520 |
+
"step": 1497
|
10521 |
+
},
|
10522 |
+
{
|
10523 |
+
"epoch": 0.14858163062884347,
|
10524 |
+
"grad_norm": 2.080760955810547,
|
10525 |
+
"learning_rate": 3.217272732759402e-07,
|
10526 |
+
"loss": 43.3713,
|
10527 |
+
"step": 1498
|
10528 |
+
},
|
10529 |
+
{
|
10530 |
+
"epoch": 0.14868081729815513,
|
10531 |
+
"grad_norm": 2.126169443130493,
|
10532 |
+
"learning_rate": 3.054482590247876e-07,
|
10533 |
+
"loss": 43.1336,
|
10534 |
+
"step": 1499
|
10535 |
+
},
|
10536 |
+
{
|
10537 |
+
"epoch": 0.14878000396746677,
|
10538 |
+
"grad_norm": 1.7946844100952148,
|
10539 |
+
"learning_rate": 2.8959122581840857e-07,
|
10540 |
+
"loss": 43.0998,
|
10541 |
+
"step": 1500
|
10542 |
+
},
|
10543 |
+
{
|
10544 |
+
"epoch": 0.14887919063677843,
|
10545 |
+
"grad_norm": 1.9658114910125732,
|
10546 |
+
"learning_rate": 2.741562407755138e-07,
|
10547 |
+
"loss": 43.256,
|
10548 |
+
"step": 1501
|
10549 |
+
},
|
10550 |
+
{
|
10551 |
+
"epoch": 0.14897837730609007,
|
10552 |
+
"grad_norm": 2.403459310531616,
|
10553 |
+
"learning_rate": 2.5914336922833183e-07,
|
10554 |
+
"loss": 43.117,
|
10555 |
+
"step": 1502
|
10556 |
+
},
|
10557 |
+
{
|
10558 |
+
"epoch": 0.1490775639754017,
|
10559 |
+
"grad_norm": 2.071451425552368,
|
10560 |
+
"learning_rate": 2.445526747224647e-07,
|
10561 |
+
"loss": 43.3778,
|
10562 |
+
"step": 1503
|
10563 |
+
},
|
10564 |
+
{
|
10565 |
+
"epoch": 0.14917675064471336,
|
10566 |
+
"grad_norm": 2.062016248703003,
|
10567 |
+
"learning_rate": 2.3038421901651064e-07,
|
10568 |
+
"loss": 43.6402,
|
10569 |
+
"step": 1504
|
10570 |
+
},
|
10571 |
+
{
|
10572 |
+
"epoch": 0.149275937314025,
|
10573 |
+
"grad_norm": 2.29769229888916,
|
10574 |
+
"learning_rate": 2.1663806208184201e-07,
|
10575 |
+
"loss": 43.3161,
|
10576 |
+
"step": 1505
|
10577 |
+
},
|
10578 |
+
{
|
10579 |
+
"epoch": 0.14937512398333663,
|
10580 |
+
"grad_norm": 2.2068557739257812,
|
10581 |
+
"learning_rate": 2.0331426210236093e-07,
|
10582 |
+
"loss": 43.3807,
|
10583 |
+
"step": 1506
|
10584 |
+
},
|
10585 |
+
{
|
10586 |
+
"epoch": 0.1494743106526483,
|
10587 |
+
"grad_norm": 2.1809418201446533,
|
10588 |
+
"learning_rate": 1.9041287547424403e-07,
|
10589 |
+
"loss": 43.3291,
|
10590 |
+
"step": 1507
|
10591 |
+
},
|
10592 |
+
{
|
10593 |
+
"epoch": 0.14957349732195993,
|
10594 |
+
"grad_norm": 2.355131149291992,
|
10595 |
+
"learning_rate": 1.7793395680568703e-07,
|
10596 |
+
"loss": 43.1938,
|
10597 |
+
"step": 1508
|
10598 |
+
},
|
10599 |
+
{
|
10600 |
+
"epoch": 0.14967268399127157,
|
10601 |
+
"grad_norm": 1.9455194473266602,
|
10602 |
+
"learning_rate": 1.6587755891671608e-07,
|
10603 |
+
"loss": 43.2039,
|
10604 |
+
"step": 1509
|
10605 |
+
},
|
10606 |
+
{
|
10607 |
+
"epoch": 0.14977187066058323,
|
10608 |
+
"grad_norm": 2.080108165740967,
|
10609 |
+
"learning_rate": 1.5424373283889904e-07,
|
10610 |
+
"loss": 43.2726,
|
10611 |
+
"step": 1510
|
10612 |
+
},
|
10613 |
+
{
|
10614 |
+
"epoch": 0.14987105732989486,
|
10615 |
+
"grad_norm": 2.2005887031555176,
|
10616 |
+
"learning_rate": 1.4303252781520114e-07,
|
10617 |
+
"loss": 43.4239,
|
10618 |
+
"step": 1511
|
10619 |
+
},
|
10620 |
+
{
|
10621 |
+
"epoch": 0.1499702439992065,
|
10622 |
+
"grad_norm": 2.1430304050445557,
|
10623 |
+
"learning_rate": 1.3224399129971866e-07,
|
10624 |
+
"loss": 43.2449,
|
10625 |
+
"step": 1512
|
10626 |
+
},
|
10627 |
+
{
|
10628 |
+
"epoch": 0.15006943066851816,
|
10629 |
+
"grad_norm": 1.9301306009292603,
|
10630 |
+
"learning_rate": 1.2187816895752324e-07,
|
10631 |
+
"loss": 43.3271,
|
10632 |
+
"step": 1513
|
10633 |
+
},
|
10634 |
+
{
|
10635 |
+
"epoch": 0.1501686173378298,
|
10636 |
+
"grad_norm": 2.6255922317504883,
|
10637 |
+
"learning_rate": 1.1193510466445123e-07,
|
10638 |
+
"loss": 43.4344,
|
10639 |
+
"step": 1514
|
10640 |
+
},
|
10641 |
+
{
|
10642 |
+
"epoch": 0.15026780400714143,
|
10643 |
+
"grad_norm": 2.2844913005828857,
|
10644 |
+
"learning_rate": 1.0241484050687034e-07,
|
10645 |
+
"loss": 43.3174,
|
10646 |
+
"step": 1515
|
10647 |
+
},
|
10648 |
+
{
|
10649 |
+
"epoch": 0.1503669906764531,
|
10650 |
+
"grad_norm": 2.1364171504974365,
|
10651 |
+
"learning_rate": 9.3317416781602e-08,
|
10652 |
+
"loss": 43.2297,
|
10653 |
+
"step": 1516
|
10654 |
+
},
|
10655 |
+
{
|
10656 |
+
"epoch": 0.15046617734576473,
|
10657 |
+
"grad_norm": 1.8956122398376465,
|
10658 |
+
"learning_rate": 8.46428719956549e-08,
|
10659 |
+
"loss": 43.6502,
|
10660 |
+
"step": 1517
|
10661 |
+
},
|
10662 |
+
{
|
10663 |
+
"epoch": 0.15056536401507636,
|
10664 |
+
"grad_norm": 2.3670010566711426,
|
10665 |
+
"learning_rate": 7.639124286612509e-08,
|
10666 |
+
"loss": 43.1647,
|
10667 |
+
"step": 1518
|
10668 |
+
},
|
10669 |
+
{
|
10670 |
+
"epoch": 0.15066455068438803,
|
10671 |
+
"grad_norm": 2.168613910675049,
|
10672 |
+
"learning_rate": 6.856256432000718e-08,
|
10673 |
+
"loss": 43.2205,
|
10674 |
+
"step": 1519
|
10675 |
+
},
|
10676 |
+
{
|
10677 |
+
"epoch": 0.15076373735369966,
|
10678 |
+
"grad_norm": 2.2199530601501465,
|
10679 |
+
"learning_rate": 6.115686949405008e-08,
|
10680 |
+
"loss": 43.4035,
|
10681 |
+
"step": 1520
|
10682 |
+
},
|
10683 |
+
{
|
10684 |
+
"epoch": 0.1508629240230113,
|
10685 |
+
"grad_norm": 1.9278777837753296,
|
10686 |
+
"learning_rate": 5.417418973462374e-08,
|
10687 |
+
"loss": 43.4381,
|
10688 |
+
"step": 1521
|
10689 |
+
},
|
10690 |
+
{
|
10691 |
+
"epoch": 0.15096211069232296,
|
10692 |
+
"grad_norm": 1.9266911745071411,
|
10693 |
+
"learning_rate": 4.7614554597608105e-08,
|
10694 |
+
"loss": 42.9214,
|
10695 |
+
"step": 1522
|
10696 |
+
},
|
10697 |
+
{
|
10698 |
+
"epoch": 0.1510612973616346,
|
10699 |
+
"grad_norm": 2.1948959827423096,
|
10700 |
+
"learning_rate": 4.147799184821555e-08,
|
10701 |
+
"loss": 43.3784,
|
10702 |
+
"step": 1523
|
10703 |
+
},
|
10704 |
+
{
|
10705 |
+
"epoch": 0.15116048403094623,
|
10706 |
+
"grad_norm": 2.2459158897399902,
|
10707 |
+
"learning_rate": 3.576452746092418e-08,
|
10708 |
+
"loss": 43.157,
|
10709 |
+
"step": 1524
|
10710 |
+
},
|
10711 |
+
{
|
10712 |
+
"epoch": 0.1512596707002579,
|
10713 |
+
"grad_norm": 1.832255244255066,
|
10714 |
+
"learning_rate": 3.047418561933357e-08,
|
10715 |
+
"loss": 43.1398,
|
10716 |
+
"step": 1525
|
10717 |
+
},
|
10718 |
+
{
|
10719 |
+
"epoch": 0.15135885736956953,
|
10720 |
+
"grad_norm": 2.164360761642456,
|
10721 |
+
"learning_rate": 2.5606988716075918e-08,
|
10722 |
+
"loss": 43.4648,
|
10723 |
+
"step": 1526
|
10724 |
+
},
|
10725 |
+
{
|
10726 |
+
"epoch": 0.15145804403888116,
|
10727 |
+
"grad_norm": 2.135101318359375,
|
10728 |
+
"learning_rate": 2.1162957352738323e-08,
|
10729 |
+
"loss": 43.3128,
|
10730 |
+
"step": 1527
|
10731 |
+
},
|
10732 |
+
{
|
10733 |
+
"epoch": 0.15155723070819282,
|
10734 |
+
"grad_norm": 1.6633306741714478,
|
10735 |
+
"learning_rate": 1.7142110339740668e-08,
|
10736 |
+
"loss": 43.4242,
|
10737 |
+
"step": 1528
|
10738 |
+
},
|
10739 |
+
{
|
10740 |
+
"epoch": 0.15165641737750446,
|
10741 |
+
"grad_norm": 2.3267548084259033,
|
10742 |
+
"learning_rate": 1.3544464696280124e-08,
|
10743 |
+
"loss": 43.4299,
|
10744 |
+
"step": 1529
|
10745 |
+
},
|
10746 |
+
{
|
10747 |
+
"epoch": 0.1517556040468161,
|
10748 |
+
"grad_norm": 1.9193159341812134,
|
10749 |
+
"learning_rate": 1.0370035650253407e-08,
|
10750 |
+
"loss": 43.2334,
|
10751 |
+
"step": 1530
|
10752 |
+
},
|
10753 |
+
{
|
10754 |
+
"epoch": 0.15185479071612776,
|
10755 |
+
"grad_norm": 1.8524115085601807,
|
10756 |
+
"learning_rate": 7.618836638190186e-09,
|
10757 |
+
"loss": 43.5525,
|
10758 |
+
"step": 1531
|
10759 |
+
},
|
10760 |
+
{
|
10761 |
+
"epoch": 0.1519539773854394,
|
10762 |
+
"grad_norm": 2.0548107624053955,
|
10763 |
+
"learning_rate": 5.29087930519756e-09,
|
10764 |
+
"loss": 43.15,
|
10765 |
+
"step": 1532
|
10766 |
+
},
|
10767 |
+
{
|
10768 |
+
"epoch": 0.15205316405475103,
|
10769 |
+
"grad_norm": 1.9881281852722168,
|
10770 |
+
"learning_rate": 3.3861735049267597e-09,
|
10771 |
+
"loss": 43.4443,
|
10772 |
+
"step": 1533
|
10773 |
+
},
|
10774 |
+
{
|
10775 |
+
"epoch": 0.1521523507240627,
|
10776 |
+
"grad_norm": 2.0406267642974854,
|
10777 |
+
"learning_rate": 1.904727299473219e-09,
|
10778 |
+
"loss": 43.1231,
|
10779 |
+
"step": 1534
|
10780 |
+
},
|
10781 |
+
{
|
10782 |
+
"epoch": 0.15225153739337433,
|
10783 |
+
"grad_norm": 2.137437582015991,
|
10784 |
+
"learning_rate": 8.465469594431952e-10,
|
10785 |
+
"loss": 43.3994,
|
10786 |
+
"step": 1535
|
10787 |
+
},
|
10788 |
+
{
|
10789 |
+
"epoch": 0.152350724062686,
|
10790 |
+
"grad_norm": 2.3377254009246826,
|
10791 |
+
"learning_rate": 2.116369638094362e-10,
|
10792 |
+
"loss": 43.3795,
|
10793 |
+
"step": 1536
|
10794 |
+
},
|
10795 |
+
{
|
10796 |
+
"epoch": 0.15244991073199762,
|
10797 |
+
"grad_norm": 1.9583935737609863,
|
10798 |
+
"learning_rate": 0.0,
|
10799 |
+
"loss": 43.5157,
|
10800 |
+
"step": 1537
|
10801 |
}
|
10802 |
],
|
10803 |
"logging_steps": 1,
|
|
|
10812 |
"should_evaluate": false,
|
10813 |
"should_log": false,
|
10814 |
"should_save": true,
|
10815 |
+
"should_training_stop": true
|
10816 |
},
|
10817 |
"attributes": {}
|
10818 |
}
|
10819 |
},
|
10820 |
+
"total_flos": 14932564770816.0,
|
10821 |
"train_batch_size": 2,
|
10822 |
"trial_name": null,
|
10823 |
"trial_params": null
|