Training in progress, step 939, 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 80013120
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:3604846a19ac3b77115d585ead1aac5a9e15883670144e52dc5ed0bd5a4e3df3
|
3 |
size 80013120
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 41120084
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:ea98c7a9e0c7f22c65f26850f6d629d23542fca4f3fe6044e1636d5082b49ea8
|
3 |
size 41120084
|
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 sha256:05a47dc53eae4b2a045d7c32619d68939422882e33af538a7f3cf93707b17ea2
|
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 sha256:32b655efac7593192052f720e800b384367a7b71e88224566b3b344d49dfcb74
|
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": 313,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4405,6 +4405,2205 @@
|
|
4405 |
"eval_samples_per_second": 13.066,
|
4406 |
"eval_steps_per_second": 6.545,
|
4407 |
"step": 626
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4408 |
}
|
4409 |
],
|
4410 |
"logging_steps": 1,
|
@@ -4424,7 +6623,7 @@
|
|
4424 |
"attributes": {}
|
4425 |
}
|
4426 |
},
|
4427 |
-
"total_flos":
|
4428 |
"train_batch_size": 2,
|
4429 |
"trial_name": null,
|
4430 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.750599520383693,
|
5 |
"eval_steps": 313,
|
6 |
+
"global_step": 939,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4405 |
"eval_samples_per_second": 13.066,
|
4406 |
"eval_steps_per_second": 6.545,
|
4407 |
"step": 626
|
4408 |
+
},
|
4409 |
+
{
|
4410 |
+
"epoch": 0.5011990407673861,
|
4411 |
+
"grad_norm": 0.4395071268081665,
|
4412 |
+
"learning_rate": 0.00010088601373262486,
|
4413 |
+
"loss": 0.9102,
|
4414 |
+
"step": 627
|
4415 |
+
},
|
4416 |
+
{
|
4417 |
+
"epoch": 0.5019984012789768,
|
4418 |
+
"grad_norm": 0.4352056384086609,
|
4419 |
+
"learning_rate": 0.00010063287100762985,
|
4420 |
+
"loss": 1.109,
|
4421 |
+
"step": 628
|
4422 |
+
},
|
4423 |
+
{
|
4424 |
+
"epoch": 0.5027977617905676,
|
4425 |
+
"grad_norm": 0.46828141808509827,
|
4426 |
+
"learning_rate": 0.00010037972422688511,
|
4427 |
+
"loss": 1.0963,
|
4428 |
+
"step": 629
|
4429 |
+
},
|
4430 |
+
{
|
4431 |
+
"epoch": 0.5035971223021583,
|
4432 |
+
"grad_norm": 0.45365041494369507,
|
4433 |
+
"learning_rate": 0.0001001265750126801,
|
4434 |
+
"loss": 0.9817,
|
4435 |
+
"step": 630
|
4436 |
+
},
|
4437 |
+
{
|
4438 |
+
"epoch": 0.504396482813749,
|
4439 |
+
"grad_norm": 0.5557889342308044,
|
4440 |
+
"learning_rate": 9.987342498731992e-05,
|
4441 |
+
"loss": 1.0607,
|
4442 |
+
"step": 631
|
4443 |
+
},
|
4444 |
+
{
|
4445 |
+
"epoch": 0.5051958433253397,
|
4446 |
+
"grad_norm": 0.4783964157104492,
|
4447 |
+
"learning_rate": 9.962027577311491e-05,
|
4448 |
+
"loss": 1.0659,
|
4449 |
+
"step": 632
|
4450 |
+
},
|
4451 |
+
{
|
4452 |
+
"epoch": 0.5059952038369304,
|
4453 |
+
"grad_norm": 0.5517037510871887,
|
4454 |
+
"learning_rate": 9.936712899237017e-05,
|
4455 |
+
"loss": 1.0411,
|
4456 |
+
"step": 633
|
4457 |
+
},
|
4458 |
+
{
|
4459 |
+
"epoch": 0.5067945643485212,
|
4460 |
+
"grad_norm": 0.5102303624153137,
|
4461 |
+
"learning_rate": 9.911398626737517e-05,
|
4462 |
+
"loss": 0.7951,
|
4463 |
+
"step": 634
|
4464 |
+
},
|
4465 |
+
{
|
4466 |
+
"epoch": 0.5075939248601119,
|
4467 |
+
"grad_norm": 0.547484278678894,
|
4468 |
+
"learning_rate": 9.886084922039344e-05,
|
4469 |
+
"loss": 0.9594,
|
4470 |
+
"step": 635
|
4471 |
+
},
|
4472 |
+
{
|
4473 |
+
"epoch": 0.5083932853717026,
|
4474 |
+
"grad_norm": 0.719420850276947,
|
4475 |
+
"learning_rate": 9.860771947365214e-05,
|
4476 |
+
"loss": 1.0052,
|
4477 |
+
"step": 636
|
4478 |
+
},
|
4479 |
+
{
|
4480 |
+
"epoch": 0.5091926458832934,
|
4481 |
+
"grad_norm": 0.6604675650596619,
|
4482 |
+
"learning_rate": 9.835459864933156e-05,
|
4483 |
+
"loss": 1.0055,
|
4484 |
+
"step": 637
|
4485 |
+
},
|
4486 |
+
{
|
4487 |
+
"epoch": 0.5099920063948841,
|
4488 |
+
"grad_norm": 0.6773166656494141,
|
4489 |
+
"learning_rate": 9.810148836955492e-05,
|
4490 |
+
"loss": 1.1806,
|
4491 |
+
"step": 638
|
4492 |
+
},
|
4493 |
+
{
|
4494 |
+
"epoch": 0.5107913669064749,
|
4495 |
+
"grad_norm": 0.877698540687561,
|
4496 |
+
"learning_rate": 9.784839025637778e-05,
|
4497 |
+
"loss": 1.1224,
|
4498 |
+
"step": 639
|
4499 |
+
},
|
4500 |
+
{
|
4501 |
+
"epoch": 0.5115907274180655,
|
4502 |
+
"grad_norm": 0.7317598462104797,
|
4503 |
+
"learning_rate": 9.759530593177777e-05,
|
4504 |
+
"loss": 1.0127,
|
4505 |
+
"step": 640
|
4506 |
+
},
|
4507 |
+
{
|
4508 |
+
"epoch": 0.5123900879296562,
|
4509 |
+
"grad_norm": 0.8157115578651428,
|
4510 |
+
"learning_rate": 9.734223701764415e-05,
|
4511 |
+
"loss": 1.2444,
|
4512 |
+
"step": 641
|
4513 |
+
},
|
4514 |
+
{
|
4515 |
+
"epoch": 0.513189448441247,
|
4516 |
+
"grad_norm": 0.9267081022262573,
|
4517 |
+
"learning_rate": 9.708918513576743e-05,
|
4518 |
+
"loss": 1.1454,
|
4519 |
+
"step": 642
|
4520 |
+
},
|
4521 |
+
{
|
4522 |
+
"epoch": 0.5139888089528377,
|
4523 |
+
"grad_norm": 0.762115478515625,
|
4524 |
+
"learning_rate": 9.683615190782892e-05,
|
4525 |
+
"loss": 1.0811,
|
4526 |
+
"step": 643
|
4527 |
+
},
|
4528 |
+
{
|
4529 |
+
"epoch": 0.5147881694644284,
|
4530 |
+
"grad_norm": 0.9379759430885315,
|
4531 |
+
"learning_rate": 9.658313895539045e-05,
|
4532 |
+
"loss": 1.0864,
|
4533 |
+
"step": 644
|
4534 |
+
},
|
4535 |
+
{
|
4536 |
+
"epoch": 0.5155875299760192,
|
4537 |
+
"grad_norm": 1.0492953062057495,
|
4538 |
+
"learning_rate": 9.63301478998839e-05,
|
4539 |
+
"loss": 1.1515,
|
4540 |
+
"step": 645
|
4541 |
+
},
|
4542 |
+
{
|
4543 |
+
"epoch": 0.5163868904876099,
|
4544 |
+
"grad_norm": 1.03159499168396,
|
4545 |
+
"learning_rate": 9.607718036260083e-05,
|
4546 |
+
"loss": 1.2317,
|
4547 |
+
"step": 646
|
4548 |
+
},
|
4549 |
+
{
|
4550 |
+
"epoch": 0.5171862509992007,
|
4551 |
+
"grad_norm": 1.086949348449707,
|
4552 |
+
"learning_rate": 9.582423796468201e-05,
|
4553 |
+
"loss": 1.1982,
|
4554 |
+
"step": 647
|
4555 |
+
},
|
4556 |
+
{
|
4557 |
+
"epoch": 0.5179856115107914,
|
4558 |
+
"grad_norm": 1.140920877456665,
|
4559 |
+
"learning_rate": 9.557132232710724e-05,
|
4560 |
+
"loss": 1.0614,
|
4561 |
+
"step": 648
|
4562 |
+
},
|
4563 |
+
{
|
4564 |
+
"epoch": 0.518784972022382,
|
4565 |
+
"grad_norm": 1.3680213689804077,
|
4566 |
+
"learning_rate": 9.531843507068465e-05,
|
4567 |
+
"loss": 1.3409,
|
4568 |
+
"step": 649
|
4569 |
+
},
|
4570 |
+
{
|
4571 |
+
"epoch": 0.5195843325339728,
|
4572 |
+
"grad_norm": 1.9634144306182861,
|
4573 |
+
"learning_rate": 9.506557781604066e-05,
|
4574 |
+
"loss": 1.0384,
|
4575 |
+
"step": 650
|
4576 |
+
},
|
4577 |
+
{
|
4578 |
+
"epoch": 0.5203836930455635,
|
4579 |
+
"grad_norm": 0.35742971301078796,
|
4580 |
+
"learning_rate": 9.481275218360932e-05,
|
4581 |
+
"loss": 0.9856,
|
4582 |
+
"step": 651
|
4583 |
+
},
|
4584 |
+
{
|
4585 |
+
"epoch": 0.5211830535571543,
|
4586 |
+
"grad_norm": 0.39657095074653625,
|
4587 |
+
"learning_rate": 9.455995979362206e-05,
|
4588 |
+
"loss": 0.9763,
|
4589 |
+
"step": 652
|
4590 |
+
},
|
4591 |
+
{
|
4592 |
+
"epoch": 0.521982414068745,
|
4593 |
+
"grad_norm": 0.3931129276752472,
|
4594 |
+
"learning_rate": 9.43072022660973e-05,
|
4595 |
+
"loss": 0.9867,
|
4596 |
+
"step": 653
|
4597 |
+
},
|
4598 |
+
{
|
4599 |
+
"epoch": 0.5227817745803357,
|
4600 |
+
"grad_norm": 0.35188305377960205,
|
4601 |
+
"learning_rate": 9.405448122083002e-05,
|
4602 |
+
"loss": 1.0704,
|
4603 |
+
"step": 654
|
4604 |
+
},
|
4605 |
+
{
|
4606 |
+
"epoch": 0.5235811350919265,
|
4607 |
+
"grad_norm": 0.35903307795524597,
|
4608 |
+
"learning_rate": 9.380179827738136e-05,
|
4609 |
+
"loss": 1.144,
|
4610 |
+
"step": 655
|
4611 |
+
},
|
4612 |
+
{
|
4613 |
+
"epoch": 0.5243804956035172,
|
4614 |
+
"grad_norm": 0.36907997727394104,
|
4615 |
+
"learning_rate": 9.354915505506839e-05,
|
4616 |
+
"loss": 1.0573,
|
4617 |
+
"step": 656
|
4618 |
+
},
|
4619 |
+
{
|
4620 |
+
"epoch": 0.5251798561151079,
|
4621 |
+
"grad_norm": 0.3568347096443176,
|
4622 |
+
"learning_rate": 9.329655317295354e-05,
|
4623 |
+
"loss": 0.8698,
|
4624 |
+
"step": 657
|
4625 |
+
},
|
4626 |
+
{
|
4627 |
+
"epoch": 0.5259792166266987,
|
4628 |
+
"grad_norm": 0.368610143661499,
|
4629 |
+
"learning_rate": 9.304399424983432e-05,
|
4630 |
+
"loss": 1.0712,
|
4631 |
+
"step": 658
|
4632 |
+
},
|
4633 |
+
{
|
4634 |
+
"epoch": 0.5267785771382894,
|
4635 |
+
"grad_norm": 0.4084762930870056,
|
4636 |
+
"learning_rate": 9.279147990423296e-05,
|
4637 |
+
"loss": 1.1245,
|
4638 |
+
"step": 659
|
4639 |
+
},
|
4640 |
+
{
|
4641 |
+
"epoch": 0.5275779376498801,
|
4642 |
+
"grad_norm": 0.35369983315467834,
|
4643 |
+
"learning_rate": 9.253901175438607e-05,
|
4644 |
+
"loss": 1.0057,
|
4645 |
+
"step": 660
|
4646 |
+
},
|
4647 |
+
{
|
4648 |
+
"epoch": 0.5283772981614708,
|
4649 |
+
"grad_norm": 0.38250893354415894,
|
4650 |
+
"learning_rate": 9.228659141823406e-05,
|
4651 |
+
"loss": 1.1446,
|
4652 |
+
"step": 661
|
4653 |
+
},
|
4654 |
+
{
|
4655 |
+
"epoch": 0.5291766586730615,
|
4656 |
+
"grad_norm": 0.3296395540237427,
|
4657 |
+
"learning_rate": 9.203422051341109e-05,
|
4658 |
+
"loss": 1.1662,
|
4659 |
+
"step": 662
|
4660 |
+
},
|
4661 |
+
{
|
4662 |
+
"epoch": 0.5299760191846523,
|
4663 |
+
"grad_norm": 0.31364706158638,
|
4664 |
+
"learning_rate": 9.178190065723447e-05,
|
4665 |
+
"loss": 1.0356,
|
4666 |
+
"step": 663
|
4667 |
+
},
|
4668 |
+
{
|
4669 |
+
"epoch": 0.530775379696243,
|
4670 |
+
"grad_norm": 0.3382437527179718,
|
4671 |
+
"learning_rate": 9.152963346669433e-05,
|
4672 |
+
"loss": 1.1452,
|
4673 |
+
"step": 664
|
4674 |
+
},
|
4675 |
+
{
|
4676 |
+
"epoch": 0.5315747402078337,
|
4677 |
+
"grad_norm": 0.3270893692970276,
|
4678 |
+
"learning_rate": 9.127742055844334e-05,
|
4679 |
+
"loss": 0.9644,
|
4680 |
+
"step": 665
|
4681 |
+
},
|
4682 |
+
{
|
4683 |
+
"epoch": 0.5323741007194245,
|
4684 |
+
"grad_norm": 0.38368576765060425,
|
4685 |
+
"learning_rate": 9.102526354878631e-05,
|
4686 |
+
"loss": 1.0437,
|
4687 |
+
"step": 666
|
4688 |
+
},
|
4689 |
+
{
|
4690 |
+
"epoch": 0.5331734612310152,
|
4691 |
+
"grad_norm": 0.4607539772987366,
|
4692 |
+
"learning_rate": 9.077316405366981e-05,
|
4693 |
+
"loss": 1.1424,
|
4694 |
+
"step": 667
|
4695 |
+
},
|
4696 |
+
{
|
4697 |
+
"epoch": 0.533972821742606,
|
4698 |
+
"grad_norm": 0.3516378700733185,
|
4699 |
+
"learning_rate": 9.052112368867181e-05,
|
4700 |
+
"loss": 0.9903,
|
4701 |
+
"step": 668
|
4702 |
+
},
|
4703 |
+
{
|
4704 |
+
"epoch": 0.5347721822541966,
|
4705 |
+
"grad_norm": 0.4142468571662903,
|
4706 |
+
"learning_rate": 9.026914406899134e-05,
|
4707 |
+
"loss": 1.0158,
|
4708 |
+
"step": 669
|
4709 |
+
},
|
4710 |
+
{
|
4711 |
+
"epoch": 0.5355715427657873,
|
4712 |
+
"grad_norm": 0.5051367282867432,
|
4713 |
+
"learning_rate": 9.00172268094382e-05,
|
4714 |
+
"loss": 1.0143,
|
4715 |
+
"step": 670
|
4716 |
+
},
|
4717 |
+
{
|
4718 |
+
"epoch": 0.5363709032773781,
|
4719 |
+
"grad_norm": 0.39056387543678284,
|
4720 |
+
"learning_rate": 8.97653735244225e-05,
|
4721 |
+
"loss": 1.1368,
|
4722 |
+
"step": 671
|
4723 |
+
},
|
4724 |
+
{
|
4725 |
+
"epoch": 0.5371702637889688,
|
4726 |
+
"grad_norm": 0.4168921411037445,
|
4727 |
+
"learning_rate": 8.951358582794438e-05,
|
4728 |
+
"loss": 1.1265,
|
4729 |
+
"step": 672
|
4730 |
+
},
|
4731 |
+
{
|
4732 |
+
"epoch": 0.5379696243005595,
|
4733 |
+
"grad_norm": 0.43402183055877686,
|
4734 |
+
"learning_rate": 8.92618653335837e-05,
|
4735 |
+
"loss": 1.1295,
|
4736 |
+
"step": 673
|
4737 |
+
},
|
4738 |
+
{
|
4739 |
+
"epoch": 0.5387689848121503,
|
4740 |
+
"grad_norm": 0.43355968594551086,
|
4741 |
+
"learning_rate": 8.901021365448958e-05,
|
4742 |
+
"loss": 1.0647,
|
4743 |
+
"step": 674
|
4744 |
+
},
|
4745 |
+
{
|
4746 |
+
"epoch": 0.539568345323741,
|
4747 |
+
"grad_norm": 0.4178990125656128,
|
4748 |
+
"learning_rate": 8.87586324033702e-05,
|
4749 |
+
"loss": 1.0942,
|
4750 |
+
"step": 675
|
4751 |
+
},
|
4752 |
+
{
|
4753 |
+
"epoch": 0.5403677058353318,
|
4754 |
+
"grad_norm": 0.4249877333641052,
|
4755 |
+
"learning_rate": 8.850712319248236e-05,
|
4756 |
+
"loss": 0.9605,
|
4757 |
+
"step": 676
|
4758 |
+
},
|
4759 |
+
{
|
4760 |
+
"epoch": 0.5411670663469225,
|
4761 |
+
"grad_norm": 0.46403768658638,
|
4762 |
+
"learning_rate": 8.825568763362127e-05,
|
4763 |
+
"loss": 1.1758,
|
4764 |
+
"step": 677
|
4765 |
+
},
|
4766 |
+
{
|
4767 |
+
"epoch": 0.5419664268585132,
|
4768 |
+
"grad_norm": 0.5160415768623352,
|
4769 |
+
"learning_rate": 8.800432733811001e-05,
|
4770 |
+
"loss": 1.2294,
|
4771 |
+
"step": 678
|
4772 |
+
},
|
4773 |
+
{
|
4774 |
+
"epoch": 0.542765787370104,
|
4775 |
+
"grad_norm": 0.3881043791770935,
|
4776 |
+
"learning_rate": 8.775304391678942e-05,
|
4777 |
+
"loss": 1.078,
|
4778 |
+
"step": 679
|
4779 |
+
},
|
4780 |
+
{
|
4781 |
+
"epoch": 0.5435651478816946,
|
4782 |
+
"grad_norm": 0.5268367528915405,
|
4783 |
+
"learning_rate": 8.750183898000774e-05,
|
4784 |
+
"loss": 0.92,
|
4785 |
+
"step": 680
|
4786 |
+
},
|
4787 |
+
{
|
4788 |
+
"epoch": 0.5443645083932853,
|
4789 |
+
"grad_norm": 0.45545732975006104,
|
4790 |
+
"learning_rate": 8.725071413761012e-05,
|
4791 |
+
"loss": 0.9236,
|
4792 |
+
"step": 681
|
4793 |
+
},
|
4794 |
+
{
|
4795 |
+
"epoch": 0.5451638689048761,
|
4796 |
+
"grad_norm": 0.5602855682373047,
|
4797 |
+
"learning_rate": 8.699967099892852e-05,
|
4798 |
+
"loss": 1.2108,
|
4799 |
+
"step": 682
|
4800 |
+
},
|
4801 |
+
{
|
4802 |
+
"epoch": 0.5459632294164668,
|
4803 |
+
"grad_norm": 0.5163658261299133,
|
4804 |
+
"learning_rate": 8.67487111727713e-05,
|
4805 |
+
"loss": 0.9604,
|
4806 |
+
"step": 683
|
4807 |
+
},
|
4808 |
+
{
|
4809 |
+
"epoch": 0.5467625899280576,
|
4810 |
+
"grad_norm": 0.5549715757369995,
|
4811 |
+
"learning_rate": 8.64978362674128e-05,
|
4812 |
+
"loss": 0.9453,
|
4813 |
+
"step": 684
|
4814 |
+
},
|
4815 |
+
{
|
4816 |
+
"epoch": 0.5475619504396483,
|
4817 |
+
"grad_norm": 0.5914304256439209,
|
4818 |
+
"learning_rate": 8.624704789058331e-05,
|
4819 |
+
"loss": 0.9931,
|
4820 |
+
"step": 685
|
4821 |
+
},
|
4822 |
+
{
|
4823 |
+
"epoch": 0.548361310951239,
|
4824 |
+
"grad_norm": 0.6247863173484802,
|
4825 |
+
"learning_rate": 8.599634764945845e-05,
|
4826 |
+
"loss": 1.0431,
|
4827 |
+
"step": 686
|
4828 |
+
},
|
4829 |
+
{
|
4830 |
+
"epoch": 0.5491606714628298,
|
4831 |
+
"grad_norm": 0.7612833380699158,
|
4832 |
+
"learning_rate": 8.574573715064913e-05,
|
4833 |
+
"loss": 0.9153,
|
4834 |
+
"step": 687
|
4835 |
+
},
|
4836 |
+
{
|
4837 |
+
"epoch": 0.5499600319744204,
|
4838 |
+
"grad_norm": 0.6234817504882812,
|
4839 |
+
"learning_rate": 8.549521800019113e-05,
|
4840 |
+
"loss": 1.0812,
|
4841 |
+
"step": 688
|
4842 |
+
},
|
4843 |
+
{
|
4844 |
+
"epoch": 0.5507593924860112,
|
4845 |
+
"grad_norm": 0.6706773638725281,
|
4846 |
+
"learning_rate": 8.524479180353478e-05,
|
4847 |
+
"loss": 0.8517,
|
4848 |
+
"step": 689
|
4849 |
+
},
|
4850 |
+
{
|
4851 |
+
"epoch": 0.5515587529976019,
|
4852 |
+
"grad_norm": 0.761870801448822,
|
4853 |
+
"learning_rate": 8.499446016553474e-05,
|
4854 |
+
"loss": 1.1793,
|
4855 |
+
"step": 690
|
4856 |
+
},
|
4857 |
+
{
|
4858 |
+
"epoch": 0.5523581135091926,
|
4859 |
+
"grad_norm": 0.7513264417648315,
|
4860 |
+
"learning_rate": 8.474422469043973e-05,
|
4861 |
+
"loss": 1.3344,
|
4862 |
+
"step": 691
|
4863 |
+
},
|
4864 |
+
{
|
4865 |
+
"epoch": 0.5531574740207834,
|
4866 |
+
"grad_norm": 0.7695756554603577,
|
4867 |
+
"learning_rate": 8.449408698188213e-05,
|
4868 |
+
"loss": 0.9859,
|
4869 |
+
"step": 692
|
4870 |
+
},
|
4871 |
+
{
|
4872 |
+
"epoch": 0.5539568345323741,
|
4873 |
+
"grad_norm": 0.8410113453865051,
|
4874 |
+
"learning_rate": 8.424404864286784e-05,
|
4875 |
+
"loss": 0.9965,
|
4876 |
+
"step": 693
|
4877 |
+
},
|
4878 |
+
{
|
4879 |
+
"epoch": 0.5547561950439648,
|
4880 |
+
"grad_norm": 0.9044365286827087,
|
4881 |
+
"learning_rate": 8.399411127576597e-05,
|
4882 |
+
"loss": 1.0025,
|
4883 |
+
"step": 694
|
4884 |
+
},
|
4885 |
+
{
|
4886 |
+
"epoch": 0.5555555555555556,
|
4887 |
+
"grad_norm": 0.9618779420852661,
|
4888 |
+
"learning_rate": 8.374427648229846e-05,
|
4889 |
+
"loss": 1.1298,
|
4890 |
+
"step": 695
|
4891 |
+
},
|
4892 |
+
{
|
4893 |
+
"epoch": 0.5563549160671463,
|
4894 |
+
"grad_norm": 1.178658127784729,
|
4895 |
+
"learning_rate": 8.349454586353e-05,
|
4896 |
+
"loss": 1.0035,
|
4897 |
+
"step": 696
|
4898 |
+
},
|
4899 |
+
{
|
4900 |
+
"epoch": 0.5571542765787371,
|
4901 |
+
"grad_norm": 1.1606285572052002,
|
4902 |
+
"learning_rate": 8.324492101985768e-05,
|
4903 |
+
"loss": 1.1512,
|
4904 |
+
"step": 697
|
4905 |
+
},
|
4906 |
+
{
|
4907 |
+
"epoch": 0.5579536370903277,
|
4908 |
+
"grad_norm": 1.2002030611038208,
|
4909 |
+
"learning_rate": 8.299540355100065e-05,
|
4910 |
+
"loss": 1.2104,
|
4911 |
+
"step": 698
|
4912 |
+
},
|
4913 |
+
{
|
4914 |
+
"epoch": 0.5587529976019184,
|
4915 |
+
"grad_norm": 1.7446719408035278,
|
4916 |
+
"learning_rate": 8.274599505599e-05,
|
4917 |
+
"loss": 1.4774,
|
4918 |
+
"step": 699
|
4919 |
+
},
|
4920 |
+
{
|
4921 |
+
"epoch": 0.5595523581135092,
|
4922 |
+
"grad_norm": 1.540687084197998,
|
4923 |
+
"learning_rate": 8.249669713315848e-05,
|
4924 |
+
"loss": 1.1088,
|
4925 |
+
"step": 700
|
4926 |
+
},
|
4927 |
+
{
|
4928 |
+
"epoch": 0.5603517186250999,
|
4929 |
+
"grad_norm": 0.32139134407043457,
|
4930 |
+
"learning_rate": 8.224751138013023e-05,
|
4931 |
+
"loss": 0.8488,
|
4932 |
+
"step": 701
|
4933 |
+
},
|
4934 |
+
{
|
4935 |
+
"epoch": 0.5611510791366906,
|
4936 |
+
"grad_norm": 0.3686629831790924,
|
4937 |
+
"learning_rate": 8.199843939381049e-05,
|
4938 |
+
"loss": 1.0761,
|
4939 |
+
"step": 702
|
4940 |
+
},
|
4941 |
+
{
|
4942 |
+
"epoch": 0.5619504396482814,
|
4943 |
+
"grad_norm": 0.4355558454990387,
|
4944 |
+
"learning_rate": 8.174948277037548e-05,
|
4945 |
+
"loss": 1.1466,
|
4946 |
+
"step": 703
|
4947 |
+
},
|
4948 |
+
{
|
4949 |
+
"epoch": 0.5627498001598721,
|
4950 |
+
"grad_norm": 0.3635938763618469,
|
4951 |
+
"learning_rate": 8.150064310526217e-05,
|
4952 |
+
"loss": 1.0464,
|
4953 |
+
"step": 704
|
4954 |
+
},
|
4955 |
+
{
|
4956 |
+
"epoch": 0.5635491606714629,
|
4957 |
+
"grad_norm": 0.34856486320495605,
|
4958 |
+
"learning_rate": 8.125192199315787e-05,
|
4959 |
+
"loss": 0.9698,
|
4960 |
+
"step": 705
|
4961 |
+
},
|
4962 |
+
{
|
4963 |
+
"epoch": 0.5643485211830536,
|
4964 |
+
"grad_norm": 0.39139944314956665,
|
4965 |
+
"learning_rate": 8.100332102799025e-05,
|
4966 |
+
"loss": 1.081,
|
4967 |
+
"step": 706
|
4968 |
+
},
|
4969 |
+
{
|
4970 |
+
"epoch": 0.5651478816946442,
|
4971 |
+
"grad_norm": 0.4321231544017792,
|
4972 |
+
"learning_rate": 8.075484180291701e-05,
|
4973 |
+
"loss": 1.0198,
|
4974 |
+
"step": 707
|
4975 |
+
},
|
4976 |
+
{
|
4977 |
+
"epoch": 0.565947242206235,
|
4978 |
+
"grad_norm": 0.3629261255264282,
|
4979 |
+
"learning_rate": 8.050648591031562e-05,
|
4980 |
+
"loss": 1.0162,
|
4981 |
+
"step": 708
|
4982 |
+
},
|
4983 |
+
{
|
4984 |
+
"epoch": 0.5667466027178257,
|
4985 |
+
"grad_norm": 0.39623329043388367,
|
4986 |
+
"learning_rate": 8.025825494177323e-05,
|
4987 |
+
"loss": 1.0521,
|
4988 |
+
"step": 709
|
4989 |
+
},
|
4990 |
+
{
|
4991 |
+
"epoch": 0.5675459632294164,
|
4992 |
+
"grad_norm": 0.39008572697639465,
|
4993 |
+
"learning_rate": 8.00101504880764e-05,
|
4994 |
+
"loss": 1.2911,
|
4995 |
+
"step": 710
|
4996 |
+
},
|
4997 |
+
{
|
4998 |
+
"epoch": 0.5683453237410072,
|
4999 |
+
"grad_norm": 0.3346228003501892,
|
5000 |
+
"learning_rate": 7.976217413920093e-05,
|
5001 |
+
"loss": 1.0933,
|
5002 |
+
"step": 711
|
5003 |
+
},
|
5004 |
+
{
|
5005 |
+
"epoch": 0.5691446842525979,
|
5006 |
+
"grad_norm": 0.34785088896751404,
|
5007 |
+
"learning_rate": 7.95143274843016e-05,
|
5008 |
+
"loss": 1.1413,
|
5009 |
+
"step": 712
|
5010 |
+
},
|
5011 |
+
{
|
5012 |
+
"epoch": 0.5699440447641887,
|
5013 |
+
"grad_norm": 0.313931405544281,
|
5014 |
+
"learning_rate": 7.926661211170213e-05,
|
5015 |
+
"loss": 1.1282,
|
5016 |
+
"step": 713
|
5017 |
+
},
|
5018 |
+
{
|
5019 |
+
"epoch": 0.5707434052757794,
|
5020 |
+
"grad_norm": 0.39661192893981934,
|
5021 |
+
"learning_rate": 7.901902960888491e-05,
|
5022 |
+
"loss": 1.241,
|
5023 |
+
"step": 714
|
5024 |
+
},
|
5025 |
+
{
|
5026 |
+
"epoch": 0.5715427657873701,
|
5027 |
+
"grad_norm": 0.38687777519226074,
|
5028 |
+
"learning_rate": 7.877158156248074e-05,
|
5029 |
+
"loss": 1.2047,
|
5030 |
+
"step": 715
|
5031 |
+
},
|
5032 |
+
{
|
5033 |
+
"epoch": 0.5723421262989609,
|
5034 |
+
"grad_norm": 0.33725664019584656,
|
5035 |
+
"learning_rate": 7.852426955825887e-05,
|
5036 |
+
"loss": 1.0205,
|
5037 |
+
"step": 716
|
5038 |
+
},
|
5039 |
+
{
|
5040 |
+
"epoch": 0.5731414868105515,
|
5041 |
+
"grad_norm": 0.3965463936328888,
|
5042 |
+
"learning_rate": 7.827709518111673e-05,
|
5043 |
+
"loss": 1.017,
|
5044 |
+
"step": 717
|
5045 |
+
},
|
5046 |
+
{
|
5047 |
+
"epoch": 0.5739408473221422,
|
5048 |
+
"grad_norm": 0.5649027228355408,
|
5049 |
+
"learning_rate": 7.803006001506961e-05,
|
5050 |
+
"loss": 1.2147,
|
5051 |
+
"step": 718
|
5052 |
+
},
|
5053 |
+
{
|
5054 |
+
"epoch": 0.574740207833733,
|
5055 |
+
"grad_norm": 0.35998645424842834,
|
5056 |
+
"learning_rate": 7.778316564324085e-05,
|
5057 |
+
"loss": 1.0301,
|
5058 |
+
"step": 719
|
5059 |
+
},
|
5060 |
+
{
|
5061 |
+
"epoch": 0.5755395683453237,
|
5062 |
+
"grad_norm": 0.4070900082588196,
|
5063 |
+
"learning_rate": 7.753641364785138e-05,
|
5064 |
+
"loss": 1.1825,
|
5065 |
+
"step": 720
|
5066 |
+
},
|
5067 |
+
{
|
5068 |
+
"epoch": 0.5763389288569145,
|
5069 |
+
"grad_norm": 0.4043642580509186,
|
5070 |
+
"learning_rate": 7.728980561020979e-05,
|
5071 |
+
"loss": 1.2504,
|
5072 |
+
"step": 721
|
5073 |
+
},
|
5074 |
+
{
|
5075 |
+
"epoch": 0.5771382893685052,
|
5076 |
+
"grad_norm": 0.4256274402141571,
|
5077 |
+
"learning_rate": 7.704334311070215e-05,
|
5078 |
+
"loss": 1.0733,
|
5079 |
+
"step": 722
|
5080 |
+
},
|
5081 |
+
{
|
5082 |
+
"epoch": 0.5779376498800959,
|
5083 |
+
"grad_norm": 0.38983389735221863,
|
5084 |
+
"learning_rate": 7.679702772878172e-05,
|
5085 |
+
"loss": 0.9375,
|
5086 |
+
"step": 723
|
5087 |
+
},
|
5088 |
+
{
|
5089 |
+
"epoch": 0.5787370103916867,
|
5090 |
+
"grad_norm": 0.449326753616333,
|
5091 |
+
"learning_rate": 7.655086104295904e-05,
|
5092 |
+
"loss": 1.1876,
|
5093 |
+
"step": 724
|
5094 |
+
},
|
5095 |
+
{
|
5096 |
+
"epoch": 0.5795363709032774,
|
5097 |
+
"grad_norm": 0.42246901988983154,
|
5098 |
+
"learning_rate": 7.630484463079177e-05,
|
5099 |
+
"loss": 1.0497,
|
5100 |
+
"step": 725
|
5101 |
+
},
|
5102 |
+
{
|
5103 |
+
"epoch": 0.580335731414868,
|
5104 |
+
"grad_norm": 0.3945838510990143,
|
5105 |
+
"learning_rate": 7.605898006887447e-05,
|
5106 |
+
"loss": 1.0072,
|
5107 |
+
"step": 726
|
5108 |
+
},
|
5109 |
+
{
|
5110 |
+
"epoch": 0.5811350919264588,
|
5111 |
+
"grad_norm": 0.4597523510456085,
|
5112 |
+
"learning_rate": 7.581326893282858e-05,
|
5113 |
+
"loss": 1.034,
|
5114 |
+
"step": 727
|
5115 |
+
},
|
5116 |
+
{
|
5117 |
+
"epoch": 0.5819344524380495,
|
5118 |
+
"grad_norm": 0.4484693706035614,
|
5119 |
+
"learning_rate": 7.556771279729236e-05,
|
5120 |
+
"loss": 1.025,
|
5121 |
+
"step": 728
|
5122 |
+
},
|
5123 |
+
{
|
5124 |
+
"epoch": 0.5827338129496403,
|
5125 |
+
"grad_norm": 0.47144320607185364,
|
5126 |
+
"learning_rate": 7.532231323591067e-05,
|
5127 |
+
"loss": 1.0669,
|
5128 |
+
"step": 729
|
5129 |
+
},
|
5130 |
+
{
|
5131 |
+
"epoch": 0.583533173461231,
|
5132 |
+
"grad_norm": 0.44177764654159546,
|
5133 |
+
"learning_rate": 7.507707182132506e-05,
|
5134 |
+
"loss": 1.0858,
|
5135 |
+
"step": 730
|
5136 |
+
},
|
5137 |
+
{
|
5138 |
+
"epoch": 0.5843325339728217,
|
5139 |
+
"grad_norm": 0.5104259848594666,
|
5140 |
+
"learning_rate": 7.483199012516353e-05,
|
5141 |
+
"loss": 1.2095,
|
5142 |
+
"step": 731
|
5143 |
+
},
|
5144 |
+
{
|
5145 |
+
"epoch": 0.5851318944844125,
|
5146 |
+
"grad_norm": 0.548528254032135,
|
5147 |
+
"learning_rate": 7.458706971803055e-05,
|
5148 |
+
"loss": 0.9455,
|
5149 |
+
"step": 732
|
5150 |
+
},
|
5151 |
+
{
|
5152 |
+
"epoch": 0.5859312549960032,
|
5153 |
+
"grad_norm": 0.5504273176193237,
|
5154 |
+
"learning_rate": 7.434231216949696e-05,
|
5155 |
+
"loss": 0.8916,
|
5156 |
+
"step": 733
|
5157 |
+
},
|
5158 |
+
{
|
5159 |
+
"epoch": 0.586730615507594,
|
5160 |
+
"grad_norm": 0.6220442652702332,
|
5161 |
+
"learning_rate": 7.409771904808993e-05,
|
5162 |
+
"loss": 1.0541,
|
5163 |
+
"step": 734
|
5164 |
+
},
|
5165 |
+
{
|
5166 |
+
"epoch": 0.5875299760191847,
|
5167 |
+
"grad_norm": 0.8554894924163818,
|
5168 |
+
"learning_rate": 7.38532919212829e-05,
|
5169 |
+
"loss": 1.0596,
|
5170 |
+
"step": 735
|
5171 |
+
},
|
5172 |
+
{
|
5173 |
+
"epoch": 0.5883293365307753,
|
5174 |
+
"grad_norm": 0.6036530137062073,
|
5175 |
+
"learning_rate": 7.360903235548553e-05,
|
5176 |
+
"loss": 1.0417,
|
5177 |
+
"step": 736
|
5178 |
+
},
|
5179 |
+
{
|
5180 |
+
"epoch": 0.5891286970423661,
|
5181 |
+
"grad_norm": 0.5838643312454224,
|
5182 |
+
"learning_rate": 7.336494191603364e-05,
|
5183 |
+
"loss": 0.9401,
|
5184 |
+
"step": 737
|
5185 |
+
},
|
5186 |
+
{
|
5187 |
+
"epoch": 0.5899280575539568,
|
5188 |
+
"grad_norm": 0.7749225497245789,
|
5189 |
+
"learning_rate": 7.312102216717929e-05,
|
5190 |
+
"loss": 1.0759,
|
5191 |
+
"step": 738
|
5192 |
+
},
|
5193 |
+
{
|
5194 |
+
"epoch": 0.5907274180655475,
|
5195 |
+
"grad_norm": 0.7013788223266602,
|
5196 |
+
"learning_rate": 7.287727467208056e-05,
|
5197 |
+
"loss": 0.9142,
|
5198 |
+
"step": 739
|
5199 |
+
},
|
5200 |
+
{
|
5201 |
+
"epoch": 0.5915267785771383,
|
5202 |
+
"grad_norm": 0.7764909863471985,
|
5203 |
+
"learning_rate": 7.263370099279172e-05,
|
5204 |
+
"loss": 1.1373,
|
5205 |
+
"step": 740
|
5206 |
+
},
|
5207 |
+
{
|
5208 |
+
"epoch": 0.592326139088729,
|
5209 |
+
"grad_norm": 0.6547873616218567,
|
5210 |
+
"learning_rate": 7.239030269025311e-05,
|
5211 |
+
"loss": 1.1794,
|
5212 |
+
"step": 741
|
5213 |
+
},
|
5214 |
+
{
|
5215 |
+
"epoch": 0.5931254996003198,
|
5216 |
+
"grad_norm": 0.7785171866416931,
|
5217 |
+
"learning_rate": 7.214708132428121e-05,
|
5218 |
+
"loss": 1.0173,
|
5219 |
+
"step": 742
|
5220 |
+
},
|
5221 |
+
{
|
5222 |
+
"epoch": 0.5939248601119105,
|
5223 |
+
"grad_norm": 0.96076500415802,
|
5224 |
+
"learning_rate": 7.190403845355853e-05,
|
5225 |
+
"loss": 1.2693,
|
5226 |
+
"step": 743
|
5227 |
+
},
|
5228 |
+
{
|
5229 |
+
"epoch": 0.5947242206235012,
|
5230 |
+
"grad_norm": 0.8062298893928528,
|
5231 |
+
"learning_rate": 7.166117563562375e-05,
|
5232 |
+
"loss": 1.1167,
|
5233 |
+
"step": 744
|
5234 |
+
},
|
5235 |
+
{
|
5236 |
+
"epoch": 0.595523581135092,
|
5237 |
+
"grad_norm": 1.0789361000061035,
|
5238 |
+
"learning_rate": 7.141849442686168e-05,
|
5239 |
+
"loss": 1.3157,
|
5240 |
+
"step": 745
|
5241 |
+
},
|
5242 |
+
{
|
5243 |
+
"epoch": 0.5963229416466826,
|
5244 |
+
"grad_norm": 0.9564920663833618,
|
5245 |
+
"learning_rate": 7.117599638249326e-05,
|
5246 |
+
"loss": 1.0115,
|
5247 |
+
"step": 746
|
5248 |
+
},
|
5249 |
+
{
|
5250 |
+
"epoch": 0.5971223021582733,
|
5251 |
+
"grad_norm": 1.0859782695770264,
|
5252 |
+
"learning_rate": 7.093368305656563e-05,
|
5253 |
+
"loss": 1.1539,
|
5254 |
+
"step": 747
|
5255 |
+
},
|
5256 |
+
{
|
5257 |
+
"epoch": 0.5979216626698641,
|
5258 |
+
"grad_norm": 1.1587315797805786,
|
5259 |
+
"learning_rate": 7.069155600194223e-05,
|
5260 |
+
"loss": 1.2031,
|
5261 |
+
"step": 748
|
5262 |
+
},
|
5263 |
+
{
|
5264 |
+
"epoch": 0.5987210231814548,
|
5265 |
+
"grad_norm": 1.4126029014587402,
|
5266 |
+
"learning_rate": 7.044961677029265e-05,
|
5267 |
+
"loss": 1.0768,
|
5268 |
+
"step": 749
|
5269 |
+
},
|
5270 |
+
{
|
5271 |
+
"epoch": 0.5995203836930456,
|
5272 |
+
"grad_norm": 1.9330767393112183,
|
5273 |
+
"learning_rate": 7.020786691208292e-05,
|
5274 |
+
"loss": 1.2553,
|
5275 |
+
"step": 750
|
5276 |
+
},
|
5277 |
+
{
|
5278 |
+
"epoch": 0.6003197442046363,
|
5279 |
+
"grad_norm": 0.33225733041763306,
|
5280 |
+
"learning_rate": 6.996630797656547e-05,
|
5281 |
+
"loss": 1.0322,
|
5282 |
+
"step": 751
|
5283 |
+
},
|
5284 |
+
{
|
5285 |
+
"epoch": 0.601119104716227,
|
5286 |
+
"grad_norm": 0.33560457825660706,
|
5287 |
+
"learning_rate": 6.972494151176915e-05,
|
5288 |
+
"loss": 1.2175,
|
5289 |
+
"step": 752
|
5290 |
+
},
|
5291 |
+
{
|
5292 |
+
"epoch": 0.6019184652278178,
|
5293 |
+
"grad_norm": 0.34604617953300476,
|
5294 |
+
"learning_rate": 6.948376906448942e-05,
|
5295 |
+
"loss": 1.0292,
|
5296 |
+
"step": 753
|
5297 |
+
},
|
5298 |
+
{
|
5299 |
+
"epoch": 0.6027178257394085,
|
5300 |
+
"grad_norm": 0.3931783139705658,
|
5301 |
+
"learning_rate": 6.924279218027828e-05,
|
5302 |
+
"loss": 1.1754,
|
5303 |
+
"step": 754
|
5304 |
+
},
|
5305 |
+
{
|
5306 |
+
"epoch": 0.6035171862509991,
|
5307 |
+
"grad_norm": 0.3545131981372833,
|
5308 |
+
"learning_rate": 6.900201240343458e-05,
|
5309 |
+
"loss": 1.0322,
|
5310 |
+
"step": 755
|
5311 |
+
},
|
5312 |
+
{
|
5313 |
+
"epoch": 0.60431654676259,
|
5314 |
+
"grad_norm": 0.3037162721157074,
|
5315 |
+
"learning_rate": 6.876143127699398e-05,
|
5316 |
+
"loss": 0.9488,
|
5317 |
+
"step": 756
|
5318 |
+
},
|
5319 |
+
{
|
5320 |
+
"epoch": 0.6051159072741806,
|
5321 |
+
"grad_norm": 0.41264012455940247,
|
5322 |
+
"learning_rate": 6.852105034271902e-05,
|
5323 |
+
"loss": 1.0289,
|
5324 |
+
"step": 757
|
5325 |
+
},
|
5326 |
+
{
|
5327 |
+
"epoch": 0.6059152677857714,
|
5328 |
+
"grad_norm": 0.4278363287448883,
|
5329 |
+
"learning_rate": 6.82808711410894e-05,
|
5330 |
+
"loss": 1.1346,
|
5331 |
+
"step": 758
|
5332 |
+
},
|
5333 |
+
{
|
5334 |
+
"epoch": 0.6067146282973621,
|
5335 |
+
"grad_norm": 0.3742501437664032,
|
5336 |
+
"learning_rate": 6.804089521129202e-05,
|
5337 |
+
"loss": 1.0822,
|
5338 |
+
"step": 759
|
5339 |
+
},
|
5340 |
+
{
|
5341 |
+
"epoch": 0.6075139888089528,
|
5342 |
+
"grad_norm": 0.3738841414451599,
|
5343 |
+
"learning_rate": 6.780112409121099e-05,
|
5344 |
+
"loss": 1.1363,
|
5345 |
+
"step": 760
|
5346 |
+
},
|
5347 |
+
{
|
5348 |
+
"epoch": 0.6083133493205436,
|
5349 |
+
"grad_norm": 0.32057246565818787,
|
5350 |
+
"learning_rate": 6.756155931741802e-05,
|
5351 |
+
"loss": 1.1435,
|
5352 |
+
"step": 761
|
5353 |
+
},
|
5354 |
+
{
|
5355 |
+
"epoch": 0.6091127098321343,
|
5356 |
+
"grad_norm": 0.3369678556919098,
|
5357 |
+
"learning_rate": 6.732220242516242e-05,
|
5358 |
+
"loss": 1.0887,
|
5359 |
+
"step": 762
|
5360 |
+
},
|
5361 |
+
{
|
5362 |
+
"epoch": 0.609912070343725,
|
5363 |
+
"grad_norm": 0.34893402457237244,
|
5364 |
+
"learning_rate": 6.708305494836131e-05,
|
5365 |
+
"loss": 1.1949,
|
5366 |
+
"step": 763
|
5367 |
+
},
|
5368 |
+
{
|
5369 |
+
"epoch": 0.6107114308553158,
|
5370 |
+
"grad_norm": 0.36019623279571533,
|
5371 |
+
"learning_rate": 6.68441184195897e-05,
|
5372 |
+
"loss": 1.1954,
|
5373 |
+
"step": 764
|
5374 |
+
},
|
5375 |
+
{
|
5376 |
+
"epoch": 0.6115107913669064,
|
5377 |
+
"grad_norm": 0.3615095913410187,
|
5378 |
+
"learning_rate": 6.660539437007081e-05,
|
5379 |
+
"loss": 1.1763,
|
5380 |
+
"step": 765
|
5381 |
+
},
|
5382 |
+
{
|
5383 |
+
"epoch": 0.6123101518784972,
|
5384 |
+
"grad_norm": 0.40820664167404175,
|
5385 |
+
"learning_rate": 6.636688432966618e-05,
|
5386 |
+
"loss": 1.0974,
|
5387 |
+
"step": 766
|
5388 |
+
},
|
5389 |
+
{
|
5390 |
+
"epoch": 0.6131095123900879,
|
5391 |
+
"grad_norm": 0.3878170847892761,
|
5392 |
+
"learning_rate": 6.612858982686584e-05,
|
5393 |
+
"loss": 1.2213,
|
5394 |
+
"step": 767
|
5395 |
+
},
|
5396 |
+
{
|
5397 |
+
"epoch": 0.6139088729016786,
|
5398 |
+
"grad_norm": 0.4106099009513855,
|
5399 |
+
"learning_rate": 6.589051238877858e-05,
|
5400 |
+
"loss": 1.0086,
|
5401 |
+
"step": 768
|
5402 |
+
},
|
5403 |
+
{
|
5404 |
+
"epoch": 0.6147082334132694,
|
5405 |
+
"grad_norm": 0.37214210629463196,
|
5406 |
+
"learning_rate": 6.565265354112214e-05,
|
5407 |
+
"loss": 0.9487,
|
5408 |
+
"step": 769
|
5409 |
+
},
|
5410 |
+
{
|
5411 |
+
"epoch": 0.6155075939248601,
|
5412 |
+
"grad_norm": 0.41581258177757263,
|
5413 |
+
"learning_rate": 6.541501480821338e-05,
|
5414 |
+
"loss": 0.9857,
|
5415 |
+
"step": 770
|
5416 |
+
},
|
5417 |
+
{
|
5418 |
+
"epoch": 0.6163069544364509,
|
5419 |
+
"grad_norm": 0.3811487853527069,
|
5420 |
+
"learning_rate": 6.51775977129586e-05,
|
5421 |
+
"loss": 1.0552,
|
5422 |
+
"step": 771
|
5423 |
+
},
|
5424 |
+
{
|
5425 |
+
"epoch": 0.6171063149480416,
|
5426 |
+
"grad_norm": 0.4689180552959442,
|
5427 |
+
"learning_rate": 6.494040377684372e-05,
|
5428 |
+
"loss": 1.0779,
|
5429 |
+
"step": 772
|
5430 |
+
},
|
5431 |
+
{
|
5432 |
+
"epoch": 0.6179056754596323,
|
5433 |
+
"grad_norm": 0.3837207853794098,
|
5434 |
+
"learning_rate": 6.470343451992452e-05,
|
5435 |
+
"loss": 1.0286,
|
5436 |
+
"step": 773
|
5437 |
+
},
|
5438 |
+
{
|
5439 |
+
"epoch": 0.6187050359712231,
|
5440 |
+
"grad_norm": 0.4035284221172333,
|
5441 |
+
"learning_rate": 6.446669146081698e-05,
|
5442 |
+
"loss": 1.0707,
|
5443 |
+
"step": 774
|
5444 |
+
},
|
5445 |
+
{
|
5446 |
+
"epoch": 0.6195043964828137,
|
5447 |
+
"grad_norm": 0.40140479803085327,
|
5448 |
+
"learning_rate": 6.423017611668745e-05,
|
5449 |
+
"loss": 0.9489,
|
5450 |
+
"step": 775
|
5451 |
+
},
|
5452 |
+
{
|
5453 |
+
"epoch": 0.6203037569944044,
|
5454 |
+
"grad_norm": 0.39459696412086487,
|
5455 |
+
"learning_rate": 6.3993890003243e-05,
|
5456 |
+
"loss": 0.9949,
|
5457 |
+
"step": 776
|
5458 |
+
},
|
5459 |
+
{
|
5460 |
+
"epoch": 0.6211031175059952,
|
5461 |
+
"grad_norm": 0.4292733073234558,
|
5462 |
+
"learning_rate": 6.375783463472165e-05,
|
5463 |
+
"loss": 1.0554,
|
5464 |
+
"step": 777
|
5465 |
+
},
|
5466 |
+
{
|
5467 |
+
"epoch": 0.6219024780175859,
|
5468 |
+
"grad_norm": 0.4641924202442169,
|
5469 |
+
"learning_rate": 6.352201152388268e-05,
|
5470 |
+
"loss": 0.9926,
|
5471 |
+
"step": 778
|
5472 |
+
},
|
5473 |
+
{
|
5474 |
+
"epoch": 0.6227018385291767,
|
5475 |
+
"grad_norm": 0.527754545211792,
|
5476 |
+
"learning_rate": 6.328642218199702e-05,
|
5477 |
+
"loss": 0.9438,
|
5478 |
+
"step": 779
|
5479 |
+
},
|
5480 |
+
{
|
5481 |
+
"epoch": 0.6235011990407674,
|
5482 |
+
"grad_norm": 0.4878404140472412,
|
5483 |
+
"learning_rate": 6.30510681188374e-05,
|
5484 |
+
"loss": 0.9029,
|
5485 |
+
"step": 780
|
5486 |
+
},
|
5487 |
+
{
|
5488 |
+
"epoch": 0.6243005595523581,
|
5489 |
+
"grad_norm": 0.5755017995834351,
|
5490 |
+
"learning_rate": 6.281595084266883e-05,
|
5491 |
+
"loss": 0.971,
|
5492 |
+
"step": 781
|
5493 |
+
},
|
5494 |
+
{
|
5495 |
+
"epoch": 0.6250999200639489,
|
5496 |
+
"grad_norm": 0.5024919509887695,
|
5497 |
+
"learning_rate": 6.258107186023885e-05,
|
5498 |
+
"loss": 1.0335,
|
5499 |
+
"step": 782
|
5500 |
+
},
|
5501 |
+
{
|
5502 |
+
"epoch": 0.6258992805755396,
|
5503 |
+
"grad_norm": 0.5928468108177185,
|
5504 |
+
"learning_rate": 6.23464326767679e-05,
|
5505 |
+
"loss": 1.0721,
|
5506 |
+
"step": 783
|
5507 |
+
},
|
5508 |
+
{
|
5509 |
+
"epoch": 0.6266986410871302,
|
5510 |
+
"grad_norm": 0.6729336380958557,
|
5511 |
+
"learning_rate": 6.211203479593968e-05,
|
5512 |
+
"loss": 0.9853,
|
5513 |
+
"step": 784
|
5514 |
+
},
|
5515 |
+
{
|
5516 |
+
"epoch": 0.627498001598721,
|
5517 |
+
"grad_norm": 0.6404271721839905,
|
5518 |
+
"learning_rate": 6.18778797198915e-05,
|
5519 |
+
"loss": 0.9947,
|
5520 |
+
"step": 785
|
5521 |
+
},
|
5522 |
+
{
|
5523 |
+
"epoch": 0.6282973621103117,
|
5524 |
+
"grad_norm": 0.7536145448684692,
|
5525 |
+
"learning_rate": 6.164396894920462e-05,
|
5526 |
+
"loss": 1.2851,
|
5527 |
+
"step": 786
|
5528 |
+
},
|
5529 |
+
{
|
5530 |
+
"epoch": 0.6290967226219025,
|
5531 |
+
"grad_norm": 0.669144332408905,
|
5532 |
+
"learning_rate": 6.141030398289473e-05,
|
5533 |
+
"loss": 1.1976,
|
5534 |
+
"step": 787
|
5535 |
+
},
|
5536 |
+
{
|
5537 |
+
"epoch": 0.6298960831334932,
|
5538 |
+
"grad_norm": 0.6938853859901428,
|
5539 |
+
"learning_rate": 6.117688631840224e-05,
|
5540 |
+
"loss": 1.0081,
|
5541 |
+
"step": 788
|
5542 |
+
},
|
5543 |
+
{
|
5544 |
+
"epoch": 0.6306954436450839,
|
5545 |
+
"grad_norm": 0.7957220673561096,
|
5546 |
+
"learning_rate": 6.0943717451582724e-05,
|
5547 |
+
"loss": 1.0587,
|
5548 |
+
"step": 789
|
5549 |
+
},
|
5550 |
+
{
|
5551 |
+
"epoch": 0.6314948041566747,
|
5552 |
+
"grad_norm": 0.7660002112388611,
|
5553 |
+
"learning_rate": 6.071079887669742e-05,
|
5554 |
+
"loss": 1.0468,
|
5555 |
+
"step": 790
|
5556 |
+
},
|
5557 |
+
{
|
5558 |
+
"epoch": 0.6322941646682654,
|
5559 |
+
"grad_norm": 1.0254074335098267,
|
5560 |
+
"learning_rate": 6.0478132086403426e-05,
|
5561 |
+
"loss": 0.9698,
|
5562 |
+
"step": 791
|
5563 |
+
},
|
5564 |
+
{
|
5565 |
+
"epoch": 0.6330935251798561,
|
5566 |
+
"grad_norm": 0.902427613735199,
|
5567 |
+
"learning_rate": 6.024571857174443e-05,
|
5568 |
+
"loss": 1.0807,
|
5569 |
+
"step": 792
|
5570 |
+
},
|
5571 |
+
{
|
5572 |
+
"epoch": 0.6338928856914469,
|
5573 |
+
"grad_norm": 0.872407078742981,
|
5574 |
+
"learning_rate": 6.001355982214092e-05,
|
5575 |
+
"loss": 1.0333,
|
5576 |
+
"step": 793
|
5577 |
+
},
|
5578 |
+
{
|
5579 |
+
"epoch": 0.6346922462030375,
|
5580 |
+
"grad_norm": 0.7746816277503967,
|
5581 |
+
"learning_rate": 5.978165732538073e-05,
|
5582 |
+
"loss": 1.0817,
|
5583 |
+
"step": 794
|
5584 |
+
},
|
5585 |
+
{
|
5586 |
+
"epoch": 0.6354916067146283,
|
5587 |
+
"grad_norm": 0.9597741365432739,
|
5588 |
+
"learning_rate": 5.95500125676095e-05,
|
5589 |
+
"loss": 1.0546,
|
5590 |
+
"step": 795
|
5591 |
+
},
|
5592 |
+
{
|
5593 |
+
"epoch": 0.636290967226219,
|
5594 |
+
"grad_norm": 0.8365216255187988,
|
5595 |
+
"learning_rate": 5.9318627033321204e-05,
|
5596 |
+
"loss": 0.9174,
|
5597 |
+
"step": 796
|
5598 |
+
},
|
5599 |
+
{
|
5600 |
+
"epoch": 0.6370903277378097,
|
5601 |
+
"grad_norm": 1.751240849494934,
|
5602 |
+
"learning_rate": 5.9087502205348534e-05,
|
5603 |
+
"loss": 1.2519,
|
5604 |
+
"step": 797
|
5605 |
+
},
|
5606 |
+
{
|
5607 |
+
"epoch": 0.6378896882494005,
|
5608 |
+
"grad_norm": 1.1197962760925293,
|
5609 |
+
"learning_rate": 5.885663956485341e-05,
|
5610 |
+
"loss": 1.084,
|
5611 |
+
"step": 798
|
5612 |
+
},
|
5613 |
+
{
|
5614 |
+
"epoch": 0.6386890487609912,
|
5615 |
+
"grad_norm": 1.6539058685302734,
|
5616 |
+
"learning_rate": 5.86260405913176e-05,
|
5617 |
+
"loss": 1.1596,
|
5618 |
+
"step": 799
|
5619 |
+
},
|
5620 |
+
{
|
5621 |
+
"epoch": 0.6394884092725819,
|
5622 |
+
"grad_norm": 1.8245570659637451,
|
5623 |
+
"learning_rate": 5.839570676253316e-05,
|
5624 |
+
"loss": 1.0606,
|
5625 |
+
"step": 800
|
5626 |
+
},
|
5627 |
+
{
|
5628 |
+
"epoch": 0.6402877697841727,
|
5629 |
+
"grad_norm": 0.3631491959095001,
|
5630 |
+
"learning_rate": 5.81656395545929e-05,
|
5631 |
+
"loss": 1.0136,
|
5632 |
+
"step": 801
|
5633 |
+
},
|
5634 |
+
{
|
5635 |
+
"epoch": 0.6410871302957634,
|
5636 |
+
"grad_norm": 0.34658652544021606,
|
5637 |
+
"learning_rate": 5.7935840441881054e-05,
|
5638 |
+
"loss": 0.9605,
|
5639 |
+
"step": 802
|
5640 |
+
},
|
5641 |
+
{
|
5642 |
+
"epoch": 0.6418864908073542,
|
5643 |
+
"grad_norm": 0.3638405203819275,
|
5644 |
+
"learning_rate": 5.770631089706373e-05,
|
5645 |
+
"loss": 0.8942,
|
5646 |
+
"step": 803
|
5647 |
+
},
|
5648 |
+
{
|
5649 |
+
"epoch": 0.6426858513189448,
|
5650 |
+
"grad_norm": 0.42472997307777405,
|
5651 |
+
"learning_rate": 5.747705239107958e-05,
|
5652 |
+
"loss": 1.0283,
|
5653 |
+
"step": 804
|
5654 |
+
},
|
5655 |
+
{
|
5656 |
+
"epoch": 0.6434852118305355,
|
5657 |
+
"grad_norm": 0.4047344923019409,
|
5658 |
+
"learning_rate": 5.724806639313025e-05,
|
5659 |
+
"loss": 1.0082,
|
5660 |
+
"step": 805
|
5661 |
+
},
|
5662 |
+
{
|
5663 |
+
"epoch": 0.6442845723421263,
|
5664 |
+
"grad_norm": 0.39204633235931396,
|
5665 |
+
"learning_rate": 5.701935437067096e-05,
|
5666 |
+
"loss": 1.0789,
|
5667 |
+
"step": 806
|
5668 |
+
},
|
5669 |
+
{
|
5670 |
+
"epoch": 0.645083932853717,
|
5671 |
+
"grad_norm": 0.35815802216529846,
|
5672 |
+
"learning_rate": 5.6790917789401285e-05,
|
5673 |
+
"loss": 1.1287,
|
5674 |
+
"step": 807
|
5675 |
+
},
|
5676 |
+
{
|
5677 |
+
"epoch": 0.6458832933653078,
|
5678 |
+
"grad_norm": 0.34175869822502136,
|
5679 |
+
"learning_rate": 5.656275811325557e-05,
|
5680 |
+
"loss": 1.0832,
|
5681 |
+
"step": 808
|
5682 |
+
},
|
5683 |
+
{
|
5684 |
+
"epoch": 0.6466826538768985,
|
5685 |
+
"grad_norm": 0.3721088469028473,
|
5686 |
+
"learning_rate": 5.633487680439361e-05,
|
5687 |
+
"loss": 1.0313,
|
5688 |
+
"step": 809
|
5689 |
+
},
|
5690 |
+
{
|
5691 |
+
"epoch": 0.6474820143884892,
|
5692 |
+
"grad_norm": 0.399013489484787,
|
5693 |
+
"learning_rate": 5.610727532319131e-05,
|
5694 |
+
"loss": 1.0316,
|
5695 |
+
"step": 810
|
5696 |
+
},
|
5697 |
+
{
|
5698 |
+
"epoch": 0.64828137490008,
|
5699 |
+
"grad_norm": 0.3578621447086334,
|
5700 |
+
"learning_rate": 5.587995512823133e-05,
|
5701 |
+
"loss": 1.1265,
|
5702 |
+
"step": 811
|
5703 |
+
},
|
5704 |
+
{
|
5705 |
+
"epoch": 0.6490807354116707,
|
5706 |
+
"grad_norm": 0.3983486294746399,
|
5707 |
+
"learning_rate": 5.565291767629359e-05,
|
5708 |
+
"loss": 1.0861,
|
5709 |
+
"step": 812
|
5710 |
+
},
|
5711 |
+
{
|
5712 |
+
"epoch": 0.6498800959232613,
|
5713 |
+
"grad_norm": 0.3946421444416046,
|
5714 |
+
"learning_rate": 5.542616442234618e-05,
|
5715 |
+
"loss": 1.0956,
|
5716 |
+
"step": 813
|
5717 |
+
},
|
5718 |
+
{
|
5719 |
+
"epoch": 0.6506794564348521,
|
5720 |
+
"grad_norm": 0.35159727931022644,
|
5721 |
+
"learning_rate": 5.5199696819535854e-05,
|
5722 |
+
"loss": 1.2644,
|
5723 |
+
"step": 814
|
5724 |
+
},
|
5725 |
+
{
|
5726 |
+
"epoch": 0.6514788169464428,
|
5727 |
+
"grad_norm": 0.36194556951522827,
|
5728 |
+
"learning_rate": 5.4973516319178774e-05,
|
5729 |
+
"loss": 1.1822,
|
5730 |
+
"step": 815
|
5731 |
+
},
|
5732 |
+
{
|
5733 |
+
"epoch": 0.6522781774580336,
|
5734 |
+
"grad_norm": 0.3506593108177185,
|
5735 |
+
"learning_rate": 5.474762437075123e-05,
|
5736 |
+
"loss": 1.119,
|
5737 |
+
"step": 816
|
5738 |
+
},
|
5739 |
+
{
|
5740 |
+
"epoch": 0.6530775379696243,
|
5741 |
+
"grad_norm": 0.4893072545528412,
|
5742 |
+
"learning_rate": 5.4522022421880295e-05,
|
5743 |
+
"loss": 1.0987,
|
5744 |
+
"step": 817
|
5745 |
+
},
|
5746 |
+
{
|
5747 |
+
"epoch": 0.653876898481215,
|
5748 |
+
"grad_norm": 0.3882545530796051,
|
5749 |
+
"learning_rate": 5.4296711918334644e-05,
|
5750 |
+
"loss": 1.1101,
|
5751 |
+
"step": 818
|
5752 |
+
},
|
5753 |
+
{
|
5754 |
+
"epoch": 0.6546762589928058,
|
5755 |
+
"grad_norm": 0.4101043939590454,
|
5756 |
+
"learning_rate": 5.407169430401511e-05,
|
5757 |
+
"loss": 1.2256,
|
5758 |
+
"step": 819
|
5759 |
+
},
|
5760 |
+
{
|
5761 |
+
"epoch": 0.6554756195043965,
|
5762 |
+
"grad_norm": 0.4425513744354248,
|
5763 |
+
"learning_rate": 5.384697102094565e-05,
|
5764 |
+
"loss": 1.1876,
|
5765 |
+
"step": 820
|
5766 |
+
},
|
5767 |
+
{
|
5768 |
+
"epoch": 0.6562749800159872,
|
5769 |
+
"grad_norm": 0.40541303157806396,
|
5770 |
+
"learning_rate": 5.362254350926403e-05,
|
5771 |
+
"loss": 0.9728,
|
5772 |
+
"step": 821
|
5773 |
+
},
|
5774 |
+
{
|
5775 |
+
"epoch": 0.657074340527578,
|
5776 |
+
"grad_norm": 0.39156901836395264,
|
5777 |
+
"learning_rate": 5.3398413207212396e-05,
|
5778 |
+
"loss": 1.0248,
|
5779 |
+
"step": 822
|
5780 |
+
},
|
5781 |
+
{
|
5782 |
+
"epoch": 0.6578737010391686,
|
5783 |
+
"grad_norm": 0.3938673734664917,
|
5784 |
+
"learning_rate": 5.317458155112842e-05,
|
5785 |
+
"loss": 1.1646,
|
5786 |
+
"step": 823
|
5787 |
+
},
|
5788 |
+
{
|
5789 |
+
"epoch": 0.6586730615507594,
|
5790 |
+
"grad_norm": 0.4352782666683197,
|
5791 |
+
"learning_rate": 5.2951049975435786e-05,
|
5792 |
+
"loss": 1.159,
|
5793 |
+
"step": 824
|
5794 |
+
},
|
5795 |
+
{
|
5796 |
+
"epoch": 0.6594724220623501,
|
5797 |
+
"grad_norm": 0.44448021054267883,
|
5798 |
+
"learning_rate": 5.272781991263516e-05,
|
5799 |
+
"loss": 1.0434,
|
5800 |
+
"step": 825
|
5801 |
+
},
|
5802 |
+
{
|
5803 |
+
"epoch": 0.6602717825739408,
|
5804 |
+
"grad_norm": 0.415777325630188,
|
5805 |
+
"learning_rate": 5.2504892793295e-05,
|
5806 |
+
"loss": 0.9778,
|
5807 |
+
"step": 826
|
5808 |
+
},
|
5809 |
+
{
|
5810 |
+
"epoch": 0.6610711430855316,
|
5811 |
+
"grad_norm": 0.4741330146789551,
|
5812 |
+
"learning_rate": 5.228227004604225e-05,
|
5813 |
+
"loss": 1.0315,
|
5814 |
+
"step": 827
|
5815 |
+
},
|
5816 |
+
{
|
5817 |
+
"epoch": 0.6618705035971223,
|
5818 |
+
"grad_norm": 0.501546323299408,
|
5819 |
+
"learning_rate": 5.205995309755335e-05,
|
5820 |
+
"loss": 1.1244,
|
5821 |
+
"step": 828
|
5822 |
+
},
|
5823 |
+
{
|
5824 |
+
"epoch": 0.662669864108713,
|
5825 |
+
"grad_norm": 0.5114132165908813,
|
5826 |
+
"learning_rate": 5.183794337254504e-05,
|
5827 |
+
"loss": 1.0154,
|
5828 |
+
"step": 829
|
5829 |
+
},
|
5830 |
+
{
|
5831 |
+
"epoch": 0.6634692246203038,
|
5832 |
+
"grad_norm": 0.4979310631752014,
|
5833 |
+
"learning_rate": 5.1616242293765226e-05,
|
5834 |
+
"loss": 0.999,
|
5835 |
+
"step": 830
|
5836 |
+
},
|
5837 |
+
{
|
5838 |
+
"epoch": 0.6642685851318945,
|
5839 |
+
"grad_norm": 0.5462687611579895,
|
5840 |
+
"learning_rate": 5.139485128198381e-05,
|
5841 |
+
"loss": 0.997,
|
5842 |
+
"step": 831
|
5843 |
+
},
|
5844 |
+
{
|
5845 |
+
"epoch": 0.6650679456434853,
|
5846 |
+
"grad_norm": 0.5672511458396912,
|
5847 |
+
"learning_rate": 5.1173771755983724e-05,
|
5848 |
+
"loss": 1.0864,
|
5849 |
+
"step": 832
|
5850 |
+
},
|
5851 |
+
{
|
5852 |
+
"epoch": 0.6658673061550759,
|
5853 |
+
"grad_norm": 0.5668870210647583,
|
5854 |
+
"learning_rate": 5.095300513255157e-05,
|
5855 |
+
"loss": 1.1027,
|
5856 |
+
"step": 833
|
5857 |
+
},
|
5858 |
+
{
|
5859 |
+
"epoch": 0.6666666666666666,
|
5860 |
+
"grad_norm": 0.6862683296203613,
|
5861 |
+
"learning_rate": 5.0732552826468896e-05,
|
5862 |
+
"loss": 0.8694,
|
5863 |
+
"step": 834
|
5864 |
+
},
|
5865 |
+
{
|
5866 |
+
"epoch": 0.6674660271782574,
|
5867 |
+
"grad_norm": 0.7277308106422424,
|
5868 |
+
"learning_rate": 5.051241625050287e-05,
|
5869 |
+
"loss": 1.2427,
|
5870 |
+
"step": 835
|
5871 |
+
},
|
5872 |
+
{
|
5873 |
+
"epoch": 0.6682653876898481,
|
5874 |
+
"grad_norm": 0.610167384147644,
|
5875 |
+
"learning_rate": 5.029259681539732e-05,
|
5876 |
+
"loss": 1.1544,
|
5877 |
+
"step": 836
|
5878 |
+
},
|
5879 |
+
{
|
5880 |
+
"epoch": 0.6690647482014388,
|
5881 |
+
"grad_norm": 0.7170546054840088,
|
5882 |
+
"learning_rate": 5.007309592986367e-05,
|
5883 |
+
"loss": 0.8898,
|
5884 |
+
"step": 837
|
5885 |
+
},
|
5886 |
+
{
|
5887 |
+
"epoch": 0.6698641087130296,
|
5888 |
+
"grad_norm": 0.6002946496009827,
|
5889 |
+
"learning_rate": 4.985391500057192e-05,
|
5890 |
+
"loss": 1.0542,
|
5891 |
+
"step": 838
|
5892 |
+
},
|
5893 |
+
{
|
5894 |
+
"epoch": 0.6706634692246203,
|
5895 |
+
"grad_norm": 0.7680448889732361,
|
5896 |
+
"learning_rate": 4.963505543214167e-05,
|
5897 |
+
"loss": 0.8763,
|
5898 |
+
"step": 839
|
5899 |
+
},
|
5900 |
+
{
|
5901 |
+
"epoch": 0.6714628297362111,
|
5902 |
+
"grad_norm": 0.7870104312896729,
|
5903 |
+
"learning_rate": 4.9416518627133e-05,
|
5904 |
+
"loss": 0.9651,
|
5905 |
+
"step": 840
|
5906 |
+
},
|
5907 |
+
{
|
5908 |
+
"epoch": 0.6722621902478018,
|
5909 |
+
"grad_norm": 0.8171029686927795,
|
5910 |
+
"learning_rate": 4.919830598603763e-05,
|
5911 |
+
"loss": 1.1376,
|
5912 |
+
"step": 841
|
5913 |
+
},
|
5914 |
+
{
|
5915 |
+
"epoch": 0.6730615507593924,
|
5916 |
+
"grad_norm": 1.0324592590332031,
|
5917 |
+
"learning_rate": 4.8980418907269863e-05,
|
5918 |
+
"loss": 0.9915,
|
5919 |
+
"step": 842
|
5920 |
+
},
|
5921 |
+
{
|
5922 |
+
"epoch": 0.6738609112709832,
|
5923 |
+
"grad_norm": 0.8155781626701355,
|
5924 |
+
"learning_rate": 4.876285878715764e-05,
|
5925 |
+
"loss": 1.1595,
|
5926 |
+
"step": 843
|
5927 |
+
},
|
5928 |
+
{
|
5929 |
+
"epoch": 0.6746602717825739,
|
5930 |
+
"grad_norm": 0.9791491627693176,
|
5931 |
+
"learning_rate": 4.8545627019933584e-05,
|
5932 |
+
"loss": 1.2055,
|
5933 |
+
"step": 844
|
5934 |
+
},
|
5935 |
+
{
|
5936 |
+
"epoch": 0.6754596322941646,
|
5937 |
+
"grad_norm": 1.083111047744751,
|
5938 |
+
"learning_rate": 4.8328724997726095e-05,
|
5939 |
+
"loss": 1.1738,
|
5940 |
+
"step": 845
|
5941 |
+
},
|
5942 |
+
{
|
5943 |
+
"epoch": 0.6762589928057554,
|
5944 |
+
"grad_norm": 0.8534079194068909,
|
5945 |
+
"learning_rate": 4.811215411055031e-05,
|
5946 |
+
"loss": 1.1604,
|
5947 |
+
"step": 846
|
5948 |
+
},
|
5949 |
+
{
|
5950 |
+
"epoch": 0.6770583533173461,
|
5951 |
+
"grad_norm": 0.9267480969429016,
|
5952 |
+
"learning_rate": 4.78959157462994e-05,
|
5953 |
+
"loss": 1.0576,
|
5954 |
+
"step": 847
|
5955 |
+
},
|
5956 |
+
{
|
5957 |
+
"epoch": 0.6778577138289369,
|
5958 |
+
"grad_norm": 1.196876049041748,
|
5959 |
+
"learning_rate": 4.768001129073553e-05,
|
5960 |
+
"loss": 1.3485,
|
5961 |
+
"step": 848
|
5962 |
+
},
|
5963 |
+
{
|
5964 |
+
"epoch": 0.6786570743405276,
|
5965 |
+
"grad_norm": 1.2235941886901855,
|
5966 |
+
"learning_rate": 4.7464442127481e-05,
|
5967 |
+
"loss": 1.204,
|
5968 |
+
"step": 849
|
5969 |
+
},
|
5970 |
+
{
|
5971 |
+
"epoch": 0.6794564348521183,
|
5972 |
+
"grad_norm": 1.7451354265213013,
|
5973 |
+
"learning_rate": 4.724920963800939e-05,
|
5974 |
+
"loss": 1.3127,
|
5975 |
+
"step": 850
|
5976 |
+
},
|
5977 |
+
{
|
5978 |
+
"epoch": 0.6802557953637091,
|
5979 |
+
"grad_norm": 0.3947245180606842,
|
5980 |
+
"learning_rate": 4.703431520163675e-05,
|
5981 |
+
"loss": 1.0509,
|
5982 |
+
"step": 851
|
5983 |
+
},
|
5984 |
+
{
|
5985 |
+
"epoch": 0.6810551558752997,
|
5986 |
+
"grad_norm": 0.3554235100746155,
|
5987 |
+
"learning_rate": 4.6819760195512686e-05,
|
5988 |
+
"loss": 0.8586,
|
5989 |
+
"step": 852
|
5990 |
+
},
|
5991 |
+
{
|
5992 |
+
"epoch": 0.6818545163868905,
|
5993 |
+
"grad_norm": 0.39175039529800415,
|
5994 |
+
"learning_rate": 4.660554599461151e-05,
|
5995 |
+
"loss": 1.0003,
|
5996 |
+
"step": 853
|
5997 |
+
},
|
5998 |
+
{
|
5999 |
+
"epoch": 0.6826538768984812,
|
6000 |
+
"grad_norm": 0.38634535670280457,
|
6001 |
+
"learning_rate": 4.639167397172358e-05,
|
6002 |
+
"loss": 1.0581,
|
6003 |
+
"step": 854
|
6004 |
+
},
|
6005 |
+
{
|
6006 |
+
"epoch": 0.6834532374100719,
|
6007 |
+
"grad_norm": 0.41433098912239075,
|
6008 |
+
"learning_rate": 4.6178145497446404e-05,
|
6009 |
+
"loss": 1.0889,
|
6010 |
+
"step": 855
|
6011 |
+
},
|
6012 |
+
{
|
6013 |
+
"epoch": 0.6842525979216627,
|
6014 |
+
"grad_norm": 0.3703967034816742,
|
6015 |
+
"learning_rate": 4.596496194017576e-05,
|
6016 |
+
"loss": 1.1005,
|
6017 |
+
"step": 856
|
6018 |
+
},
|
6019 |
+
{
|
6020 |
+
"epoch": 0.6850519584332534,
|
6021 |
+
"grad_norm": 0.36882877349853516,
|
6022 |
+
"learning_rate": 4.575212466609713e-05,
|
6023 |
+
"loss": 0.9307,
|
6024 |
+
"step": 857
|
6025 |
+
},
|
6026 |
+
{
|
6027 |
+
"epoch": 0.6858513189448441,
|
6028 |
+
"grad_norm": 0.3600471019744873,
|
6029 |
+
"learning_rate": 4.553963503917681e-05,
|
6030 |
+
"loss": 0.893,
|
6031 |
+
"step": 858
|
6032 |
+
},
|
6033 |
+
{
|
6034 |
+
"epoch": 0.6866506794564349,
|
6035 |
+
"grad_norm": 0.33053648471832275,
|
6036 |
+
"learning_rate": 4.53274944211532e-05,
|
6037 |
+
"loss": 1.1211,
|
6038 |
+
"step": 859
|
6039 |
+
},
|
6040 |
+
{
|
6041 |
+
"epoch": 0.6874500399680256,
|
6042 |
+
"grad_norm": 0.37433984875679016,
|
6043 |
+
"learning_rate": 4.5115704171528105e-05,
|
6044 |
+
"loss": 1.0067,
|
6045 |
+
"step": 860
|
6046 |
+
},
|
6047 |
+
{
|
6048 |
+
"epoch": 0.6882494004796164,
|
6049 |
+
"grad_norm": 0.40663594007492065,
|
6050 |
+
"learning_rate": 4.490426564755792e-05,
|
6051 |
+
"loss": 1.0298,
|
6052 |
+
"step": 861
|
6053 |
+
},
|
6054 |
+
{
|
6055 |
+
"epoch": 0.689048760991207,
|
6056 |
+
"grad_norm": 0.49298176169395447,
|
6057 |
+
"learning_rate": 4.4693180204245074e-05,
|
6058 |
+
"loss": 1.0433,
|
6059 |
+
"step": 862
|
6060 |
+
},
|
6061 |
+
{
|
6062 |
+
"epoch": 0.6898481215027977,
|
6063 |
+
"grad_norm": 0.3631519377231598,
|
6064 |
+
"learning_rate": 4.448244919432929e-05,
|
6065 |
+
"loss": 1.1514,
|
6066 |
+
"step": 863
|
6067 |
+
},
|
6068 |
+
{
|
6069 |
+
"epoch": 0.6906474820143885,
|
6070 |
+
"grad_norm": 0.38605815172195435,
|
6071 |
+
"learning_rate": 4.427207396827886e-05,
|
6072 |
+
"loss": 1.1713,
|
6073 |
+
"step": 864
|
6074 |
+
},
|
6075 |
+
{
|
6076 |
+
"epoch": 0.6914468425259792,
|
6077 |
+
"grad_norm": 0.341621071100235,
|
6078 |
+
"learning_rate": 4.406205587428206e-05,
|
6079 |
+
"loss": 1.1695,
|
6080 |
+
"step": 865
|
6081 |
+
},
|
6082 |
+
{
|
6083 |
+
"epoch": 0.6922462030375699,
|
6084 |
+
"grad_norm": 0.3484615981578827,
|
6085 |
+
"learning_rate": 4.3852396258238536e-05,
|
6086 |
+
"loss": 1.0565,
|
6087 |
+
"step": 866
|
6088 |
+
},
|
6089 |
+
{
|
6090 |
+
"epoch": 0.6930455635491607,
|
6091 |
+
"grad_norm": 0.4451755881309509,
|
6092 |
+
"learning_rate": 4.3643096463750496e-05,
|
6093 |
+
"loss": 1.1425,
|
6094 |
+
"step": 867
|
6095 |
+
},
|
6096 |
+
{
|
6097 |
+
"epoch": 0.6938449240607514,
|
6098 |
+
"grad_norm": 0.4082989990711212,
|
6099 |
+
"learning_rate": 4.3434157832114373e-05,
|
6100 |
+
"loss": 0.9311,
|
6101 |
+
"step": 868
|
6102 |
+
},
|
6103 |
+
{
|
6104 |
+
"epoch": 0.6946442845723422,
|
6105 |
+
"grad_norm": 0.3935573101043701,
|
6106 |
+
"learning_rate": 4.3225581702312025e-05,
|
6107 |
+
"loss": 1.1656,
|
6108 |
+
"step": 869
|
6109 |
+
},
|
6110 |
+
{
|
6111 |
+
"epoch": 0.6954436450839329,
|
6112 |
+
"grad_norm": 0.44610002636909485,
|
6113 |
+
"learning_rate": 4.301736941100223e-05,
|
6114 |
+
"loss": 1.2297,
|
6115 |
+
"step": 870
|
6116 |
+
},
|
6117 |
+
{
|
6118 |
+
"epoch": 0.6962430055955235,
|
6119 |
+
"grad_norm": 0.44538775086402893,
|
6120 |
+
"learning_rate": 4.280952229251212e-05,
|
6121 |
+
"loss": 1.0755,
|
6122 |
+
"step": 871
|
6123 |
+
},
|
6124 |
+
{
|
6125 |
+
"epoch": 0.6970423661071143,
|
6126 |
+
"grad_norm": 0.4480975568294525,
|
6127 |
+
"learning_rate": 4.26020416788286e-05,
|
6128 |
+
"loss": 1.2264,
|
6129 |
+
"step": 872
|
6130 |
+
},
|
6131 |
+
{
|
6132 |
+
"epoch": 0.697841726618705,
|
6133 |
+
"grad_norm": 0.5919811129570007,
|
6134 |
+
"learning_rate": 4.239492889958988e-05,
|
6135 |
+
"loss": 1.118,
|
6136 |
+
"step": 873
|
6137 |
+
},
|
6138 |
+
{
|
6139 |
+
"epoch": 0.6986410871302957,
|
6140 |
+
"grad_norm": 0.4208342730998993,
|
6141 |
+
"learning_rate": 4.21881852820768e-05,
|
6142 |
+
"loss": 0.9921,
|
6143 |
+
"step": 874
|
6144 |
+
},
|
6145 |
+
{
|
6146 |
+
"epoch": 0.6994404476418865,
|
6147 |
+
"grad_norm": 0.4386043846607208,
|
6148 |
+
"learning_rate": 4.1981812151204513e-05,
|
6149 |
+
"loss": 0.9987,
|
6150 |
+
"step": 875
|
6151 |
+
},
|
6152 |
+
{
|
6153 |
+
"epoch": 0.7002398081534772,
|
6154 |
+
"grad_norm": 0.5912065505981445,
|
6155 |
+
"learning_rate": 4.177581082951393e-05,
|
6156 |
+
"loss": 1.1662,
|
6157 |
+
"step": 876
|
6158 |
+
},
|
6159 |
+
{
|
6160 |
+
"epoch": 0.701039168665068,
|
6161 |
+
"grad_norm": 0.5079301595687866,
|
6162 |
+
"learning_rate": 4.1570182637163155e-05,
|
6163 |
+
"loss": 1.0607,
|
6164 |
+
"step": 877
|
6165 |
+
},
|
6166 |
+
{
|
6167 |
+
"epoch": 0.7018385291766587,
|
6168 |
+
"grad_norm": 0.5225832462310791,
|
6169 |
+
"learning_rate": 4.136492889191914e-05,
|
6170 |
+
"loss": 1.0134,
|
6171 |
+
"step": 878
|
6172 |
+
},
|
6173 |
+
{
|
6174 |
+
"epoch": 0.7026378896882494,
|
6175 |
+
"grad_norm": 0.49494823813438416,
|
6176 |
+
"learning_rate": 4.1160050909149196e-05,
|
6177 |
+
"loss": 0.9564,
|
6178 |
+
"step": 879
|
6179 |
+
},
|
6180 |
+
{
|
6181 |
+
"epoch": 0.7034372501998402,
|
6182 |
+
"grad_norm": 0.5825321674346924,
|
6183 |
+
"learning_rate": 4.095555000181257e-05,
|
6184 |
+
"loss": 1.1002,
|
6185 |
+
"step": 880
|
6186 |
+
},
|
6187 |
+
{
|
6188 |
+
"epoch": 0.7042366107114308,
|
6189 |
+
"grad_norm": 0.5384085774421692,
|
6190 |
+
"learning_rate": 4.075142748045194e-05,
|
6191 |
+
"loss": 0.8761,
|
6192 |
+
"step": 881
|
6193 |
+
},
|
6194 |
+
{
|
6195 |
+
"epoch": 0.7050359712230215,
|
6196 |
+
"grad_norm": 0.5739712119102478,
|
6197 |
+
"learning_rate": 4.05476846531852e-05,
|
6198 |
+
"loss": 1.1312,
|
6199 |
+
"step": 882
|
6200 |
+
},
|
6201 |
+
{
|
6202 |
+
"epoch": 0.7058353317346123,
|
6203 |
+
"grad_norm": 0.5879523754119873,
|
6204 |
+
"learning_rate": 4.034432282569693e-05,
|
6205 |
+
"loss": 1.1856,
|
6206 |
+
"step": 883
|
6207 |
+
},
|
6208 |
+
{
|
6209 |
+
"epoch": 0.706634692246203,
|
6210 |
+
"grad_norm": 0.5788174867630005,
|
6211 |
+
"learning_rate": 4.0141343301230116e-05,
|
6212 |
+
"loss": 1.0818,
|
6213 |
+
"step": 884
|
6214 |
+
},
|
6215 |
+
{
|
6216 |
+
"epoch": 0.7074340527577938,
|
6217 |
+
"grad_norm": 0.6709664463996887,
|
6218 |
+
"learning_rate": 3.99387473805777e-05,
|
6219 |
+
"loss": 1.1062,
|
6220 |
+
"step": 885
|
6221 |
+
},
|
6222 |
+
{
|
6223 |
+
"epoch": 0.7082334132693845,
|
6224 |
+
"grad_norm": 0.7911237478256226,
|
6225 |
+
"learning_rate": 3.973653636207437e-05,
|
6226 |
+
"loss": 1.1432,
|
6227 |
+
"step": 886
|
6228 |
+
},
|
6229 |
+
{
|
6230 |
+
"epoch": 0.7090327737809752,
|
6231 |
+
"grad_norm": 0.6831583380699158,
|
6232 |
+
"learning_rate": 3.953471154158808e-05,
|
6233 |
+
"loss": 0.8802,
|
6234 |
+
"step": 887
|
6235 |
+
},
|
6236 |
+
{
|
6237 |
+
"epoch": 0.709832134292566,
|
6238 |
+
"grad_norm": 0.7837221026420593,
|
6239 |
+
"learning_rate": 3.9333274212511887e-05,
|
6240 |
+
"loss": 0.8944,
|
6241 |
+
"step": 888
|
6242 |
+
},
|
6243 |
+
{
|
6244 |
+
"epoch": 0.7106314948041567,
|
6245 |
+
"grad_norm": 0.6791643500328064,
|
6246 |
+
"learning_rate": 3.913222566575564e-05,
|
6247 |
+
"loss": 0.8216,
|
6248 |
+
"step": 889
|
6249 |
+
},
|
6250 |
+
{
|
6251 |
+
"epoch": 0.7114308553157475,
|
6252 |
+
"grad_norm": 0.8798190951347351,
|
6253 |
+
"learning_rate": 3.893156718973767e-05,
|
6254 |
+
"loss": 1.3466,
|
6255 |
+
"step": 890
|
6256 |
+
},
|
6257 |
+
{
|
6258 |
+
"epoch": 0.7122302158273381,
|
6259 |
+
"grad_norm": 0.8324375152587891,
|
6260 |
+
"learning_rate": 3.8731300070376464e-05,
|
6261 |
+
"loss": 1.074,
|
6262 |
+
"step": 891
|
6263 |
+
},
|
6264 |
+
{
|
6265 |
+
"epoch": 0.7130295763389288,
|
6266 |
+
"grad_norm": 1.132907748222351,
|
6267 |
+
"learning_rate": 3.8531425591082606e-05,
|
6268 |
+
"loss": 1.3694,
|
6269 |
+
"step": 892
|
6270 |
+
},
|
6271 |
+
{
|
6272 |
+
"epoch": 0.7138289368505196,
|
6273 |
+
"grad_norm": 1.0507283210754395,
|
6274 |
+
"learning_rate": 3.8331945032750374e-05,
|
6275 |
+
"loss": 1.2547,
|
6276 |
+
"step": 893
|
6277 |
+
},
|
6278 |
+
{
|
6279 |
+
"epoch": 0.7146282973621103,
|
6280 |
+
"grad_norm": 0.9351432919502258,
|
6281 |
+
"learning_rate": 3.813285967374969e-05,
|
6282 |
+
"loss": 1.0931,
|
6283 |
+
"step": 894
|
6284 |
+
},
|
6285 |
+
{
|
6286 |
+
"epoch": 0.715427657873701,
|
6287 |
+
"grad_norm": 1.1382817029953003,
|
6288 |
+
"learning_rate": 3.7934170789917724e-05,
|
6289 |
+
"loss": 1.2737,
|
6290 |
+
"step": 895
|
6291 |
+
},
|
6292 |
+
{
|
6293 |
+
"epoch": 0.7162270183852918,
|
6294 |
+
"grad_norm": 0.9496870636940002,
|
6295 |
+
"learning_rate": 3.773587965455093e-05,
|
6296 |
+
"loss": 1.1127,
|
6297 |
+
"step": 896
|
6298 |
+
},
|
6299 |
+
{
|
6300 |
+
"epoch": 0.7170263788968825,
|
6301 |
+
"grad_norm": 1.0737744569778442,
|
6302 |
+
"learning_rate": 3.753798753839677e-05,
|
6303 |
+
"loss": 1.2043,
|
6304 |
+
"step": 897
|
6305 |
+
},
|
6306 |
+
{
|
6307 |
+
"epoch": 0.7178257394084733,
|
6308 |
+
"grad_norm": 1.1727651357650757,
|
6309 |
+
"learning_rate": 3.7340495709645585e-05,
|
6310 |
+
"loss": 1.2318,
|
6311 |
+
"step": 898
|
6312 |
+
},
|
6313 |
+
{
|
6314 |
+
"epoch": 0.718625099920064,
|
6315 |
+
"grad_norm": 1.544258952140808,
|
6316 |
+
"learning_rate": 3.7143405433922485e-05,
|
6317 |
+
"loss": 1.3744,
|
6318 |
+
"step": 899
|
6319 |
+
},
|
6320 |
+
{
|
6321 |
+
"epoch": 0.7194244604316546,
|
6322 |
+
"grad_norm": 1.887011170387268,
|
6323 |
+
"learning_rate": 3.694671797427924e-05,
|
6324 |
+
"loss": 0.7498,
|
6325 |
+
"step": 900
|
6326 |
+
},
|
6327 |
+
{
|
6328 |
+
"epoch": 0.7202238209432454,
|
6329 |
+
"grad_norm": 0.359557181596756,
|
6330 |
+
"learning_rate": 3.675043459118619e-05,
|
6331 |
+
"loss": 0.9036,
|
6332 |
+
"step": 901
|
6333 |
+
},
|
6334 |
+
{
|
6335 |
+
"epoch": 0.7210231814548361,
|
6336 |
+
"grad_norm": 0.3796878159046173,
|
6337 |
+
"learning_rate": 3.6554556542524044e-05,
|
6338 |
+
"loss": 1.0923,
|
6339 |
+
"step": 902
|
6340 |
+
},
|
6341 |
+
{
|
6342 |
+
"epoch": 0.7218225419664268,
|
6343 |
+
"grad_norm": 0.39099475741386414,
|
6344 |
+
"learning_rate": 3.635908508357606e-05,
|
6345 |
+
"loss": 0.987,
|
6346 |
+
"step": 903
|
6347 |
+
},
|
6348 |
+
{
|
6349 |
+
"epoch": 0.7226219024780176,
|
6350 |
+
"grad_norm": 0.4106023907661438,
|
6351 |
+
"learning_rate": 3.6164021467019814e-05,
|
6352 |
+
"loss": 1.1539,
|
6353 |
+
"step": 904
|
6354 |
+
},
|
6355 |
+
{
|
6356 |
+
"epoch": 0.7234212629896083,
|
6357 |
+
"grad_norm": 0.391400545835495,
|
6358 |
+
"learning_rate": 3.5969366942919234e-05,
|
6359 |
+
"loss": 1.0677,
|
6360 |
+
"step": 905
|
6361 |
+
},
|
6362 |
+
{
|
6363 |
+
"epoch": 0.7242206235011991,
|
6364 |
+
"grad_norm": 0.35718274116516113,
|
6365 |
+
"learning_rate": 3.5775122758716585e-05,
|
6366 |
+
"loss": 1.1138,
|
6367 |
+
"step": 906
|
6368 |
+
},
|
6369 |
+
{
|
6370 |
+
"epoch": 0.7250199840127898,
|
6371 |
+
"grad_norm": 0.3467647135257721,
|
6372 |
+
"learning_rate": 3.558129015922448e-05,
|
6373 |
+
"loss": 1.097,
|
6374 |
+
"step": 907
|
6375 |
+
},
|
6376 |
+
{
|
6377 |
+
"epoch": 0.7258193445243805,
|
6378 |
+
"grad_norm": 0.37146949768066406,
|
6379 |
+
"learning_rate": 3.5387870386617826e-05,
|
6380 |
+
"loss": 1.0741,
|
6381 |
+
"step": 908
|
6382 |
+
},
|
6383 |
+
{
|
6384 |
+
"epoch": 0.7266187050359713,
|
6385 |
+
"grad_norm": 0.42259925603866577,
|
6386 |
+
"learning_rate": 3.519486468042602e-05,
|
6387 |
+
"loss": 1.042,
|
6388 |
+
"step": 909
|
6389 |
+
},
|
6390 |
+
{
|
6391 |
+
"epoch": 0.7274180655475619,
|
6392 |
+
"grad_norm": 0.3869469463825226,
|
6393 |
+
"learning_rate": 3.5002274277524874e-05,
|
6394 |
+
"loss": 0.9691,
|
6395 |
+
"step": 910
|
6396 |
+
},
|
6397 |
+
{
|
6398 |
+
"epoch": 0.7282174260591526,
|
6399 |
+
"grad_norm": 0.40922337770462036,
|
6400 |
+
"learning_rate": 3.4810100412128747e-05,
|
6401 |
+
"loss": 1.2811,
|
6402 |
+
"step": 911
|
6403 |
+
},
|
6404 |
+
{
|
6405 |
+
"epoch": 0.7290167865707434,
|
6406 |
+
"grad_norm": 0.4367547035217285,
|
6407 |
+
"learning_rate": 3.4618344315782604e-05,
|
6408 |
+
"loss": 1.1267,
|
6409 |
+
"step": 912
|
6410 |
+
},
|
6411 |
+
{
|
6412 |
+
"epoch": 0.7298161470823341,
|
6413 |
+
"grad_norm": 0.40415459871292114,
|
6414 |
+
"learning_rate": 3.442700721735417e-05,
|
6415 |
+
"loss": 1.118,
|
6416 |
+
"step": 913
|
6417 |
+
},
|
6418 |
+
{
|
6419 |
+
"epoch": 0.7306155075939249,
|
6420 |
+
"grad_norm": 0.37041985988616943,
|
6421 |
+
"learning_rate": 3.4236090343026004e-05,
|
6422 |
+
"loss": 0.9999,
|
6423 |
+
"step": 914
|
6424 |
+
},
|
6425 |
+
{
|
6426 |
+
"epoch": 0.7314148681055156,
|
6427 |
+
"grad_norm": 0.3673413395881653,
|
6428 |
+
"learning_rate": 3.4045594916287616e-05,
|
6429 |
+
"loss": 1.0145,
|
6430 |
+
"step": 915
|
6431 |
+
},
|
6432 |
+
{
|
6433 |
+
"epoch": 0.7322142286171063,
|
6434 |
+
"grad_norm": 0.44406071305274963,
|
6435 |
+
"learning_rate": 3.3855522157927756e-05,
|
6436 |
+
"loss": 1.0797,
|
6437 |
+
"step": 916
|
6438 |
+
},
|
6439 |
+
{
|
6440 |
+
"epoch": 0.7330135891286971,
|
6441 |
+
"grad_norm": 0.4258328080177307,
|
6442 |
+
"learning_rate": 3.3665873286026436e-05,
|
6443 |
+
"loss": 0.9991,
|
6444 |
+
"step": 917
|
6445 |
+
},
|
6446 |
+
{
|
6447 |
+
"epoch": 0.7338129496402878,
|
6448 |
+
"grad_norm": 0.381394624710083,
|
6449 |
+
"learning_rate": 3.3476649515947255e-05,
|
6450 |
+
"loss": 0.8769,
|
6451 |
+
"step": 918
|
6452 |
+
},
|
6453 |
+
{
|
6454 |
+
"epoch": 0.7346123101518784,
|
6455 |
+
"grad_norm": 0.4326583445072174,
|
6456 |
+
"learning_rate": 3.3287852060329514e-05,
|
6457 |
+
"loss": 1.1577,
|
6458 |
+
"step": 919
|
6459 |
+
},
|
6460 |
+
{
|
6461 |
+
"epoch": 0.7354116706634692,
|
6462 |
+
"grad_norm": 0.40116122364997864,
|
6463 |
+
"learning_rate": 3.309948212908052e-05,
|
6464 |
+
"loss": 0.9067,
|
6465 |
+
"step": 920
|
6466 |
+
},
|
6467 |
+
{
|
6468 |
+
"epoch": 0.7362110311750599,
|
6469 |
+
"grad_norm": 0.418855220079422,
|
6470 |
+
"learning_rate": 3.291154092936768e-05,
|
6471 |
+
"loss": 0.9349,
|
6472 |
+
"step": 921
|
6473 |
+
},
|
6474 |
+
{
|
6475 |
+
"epoch": 0.7370103916866507,
|
6476 |
+
"grad_norm": 0.3745797276496887,
|
6477 |
+
"learning_rate": 3.2724029665611e-05,
|
6478 |
+
"loss": 0.9903,
|
6479 |
+
"step": 922
|
6480 |
+
},
|
6481 |
+
{
|
6482 |
+
"epoch": 0.7378097521982414,
|
6483 |
+
"grad_norm": 0.44138798117637634,
|
6484 |
+
"learning_rate": 3.2536949539475193e-05,
|
6485 |
+
"loss": 1.0919,
|
6486 |
+
"step": 923
|
6487 |
+
},
|
6488 |
+
{
|
6489 |
+
"epoch": 0.7386091127098321,
|
6490 |
+
"grad_norm": 0.4620357155799866,
|
6491 |
+
"learning_rate": 3.235030174986209e-05,
|
6492 |
+
"loss": 1.3411,
|
6493 |
+
"step": 924
|
6494 |
+
},
|
6495 |
+
{
|
6496 |
+
"epoch": 0.7394084732214229,
|
6497 |
+
"grad_norm": 0.4606659412384033,
|
6498 |
+
"learning_rate": 3.216408749290276e-05,
|
6499 |
+
"loss": 1.0656,
|
6500 |
+
"step": 925
|
6501 |
+
},
|
6502 |
+
{
|
6503 |
+
"epoch": 0.7402078337330136,
|
6504 |
+
"grad_norm": 0.5463187098503113,
|
6505 |
+
"learning_rate": 3.197830796195011e-05,
|
6506 |
+
"loss": 1.1517,
|
6507 |
+
"step": 926
|
6508 |
+
},
|
6509 |
+
{
|
6510 |
+
"epoch": 0.7410071942446043,
|
6511 |
+
"grad_norm": 0.511232852935791,
|
6512 |
+
"learning_rate": 3.179296434757105e-05,
|
6513 |
+
"loss": 1.0564,
|
6514 |
+
"step": 927
|
6515 |
+
},
|
6516 |
+
{
|
6517 |
+
"epoch": 0.7418065547561951,
|
6518 |
+
"grad_norm": 0.5773586630821228,
|
6519 |
+
"learning_rate": 3.160805783753897e-05,
|
6520 |
+
"loss": 0.764,
|
6521 |
+
"step": 928
|
6522 |
+
},
|
6523 |
+
{
|
6524 |
+
"epoch": 0.7426059152677857,
|
6525 |
+
"grad_norm": 0.5721275806427002,
|
6526 |
+
"learning_rate": 3.1423589616825986e-05,
|
6527 |
+
"loss": 0.8128,
|
6528 |
+
"step": 929
|
6529 |
+
},
|
6530 |
+
{
|
6531 |
+
"epoch": 0.7434052757793765,
|
6532 |
+
"grad_norm": 0.6489303112030029,
|
6533 |
+
"learning_rate": 3.123956086759553e-05,
|
6534 |
+
"loss": 1.2348,
|
6535 |
+
"step": 930
|
6536 |
+
},
|
6537 |
+
{
|
6538 |
+
"epoch": 0.7442046362909672,
|
6539 |
+
"grad_norm": 0.6034092903137207,
|
6540 |
+
"learning_rate": 3.105597276919465e-05,
|
6541 |
+
"loss": 1.1123,
|
6542 |
+
"step": 931
|
6543 |
+
},
|
6544 |
+
{
|
6545 |
+
"epoch": 0.7450039968025579,
|
6546 |
+
"grad_norm": 0.6793666481971741,
|
6547 |
+
"learning_rate": 3.087282649814648e-05,
|
6548 |
+
"loss": 1.1772,
|
6549 |
+
"step": 932
|
6550 |
+
},
|
6551 |
+
{
|
6552 |
+
"epoch": 0.7458033573141487,
|
6553 |
+
"grad_norm": 0.6495110392570496,
|
6554 |
+
"learning_rate": 3.069012322814272e-05,
|
6555 |
+
"loss": 1.0015,
|
6556 |
+
"step": 933
|
6557 |
+
},
|
6558 |
+
{
|
6559 |
+
"epoch": 0.7466027178257394,
|
6560 |
+
"grad_norm": 0.6330723762512207,
|
6561 |
+
"learning_rate": 3.05078641300361e-05,
|
6562 |
+
"loss": 1.0119,
|
6563 |
+
"step": 934
|
6564 |
+
},
|
6565 |
+
{
|
6566 |
+
"epoch": 0.7474020783373302,
|
6567 |
+
"grad_norm": 0.8546098470687866,
|
6568 |
+
"learning_rate": 3.0326050371832892e-05,
|
6569 |
+
"loss": 0.9274,
|
6570 |
+
"step": 935
|
6571 |
+
},
|
6572 |
+
{
|
6573 |
+
"epoch": 0.7482014388489209,
|
6574 |
+
"grad_norm": 0.637100100517273,
|
6575 |
+
"learning_rate": 3.0144683118685335e-05,
|
6576 |
+
"loss": 0.9028,
|
6577 |
+
"step": 936
|
6578 |
+
},
|
6579 |
+
{
|
6580 |
+
"epoch": 0.7490007993605116,
|
6581 |
+
"grad_norm": 0.7471299767494202,
|
6582 |
+
"learning_rate": 2.9963763532884326e-05,
|
6583 |
+
"loss": 1.0216,
|
6584 |
+
"step": 937
|
6585 |
+
},
|
6586 |
+
{
|
6587 |
+
"epoch": 0.7498001598721024,
|
6588 |
+
"grad_norm": 0.6812106370925903,
|
6589 |
+
"learning_rate": 2.9783292773851857e-05,
|
6590 |
+
"loss": 1.013,
|
6591 |
+
"step": 938
|
6592 |
+
},
|
6593 |
+
{
|
6594 |
+
"epoch": 0.750599520383693,
|
6595 |
+
"grad_norm": 0.8330495357513428,
|
6596 |
+
"learning_rate": 2.960327199813362e-05,
|
6597 |
+
"loss": 1.1521,
|
6598 |
+
"step": 939
|
6599 |
+
},
|
6600 |
+
{
|
6601 |
+
"epoch": 0.750599520383693,
|
6602 |
+
"eval_loss": 1.0665621757507324,
|
6603 |
+
"eval_runtime": 40.4579,
|
6604 |
+
"eval_samples_per_second": 13.026,
|
6605 |
+
"eval_steps_per_second": 6.525,
|
6606 |
+
"step": 939
|
6607 |
}
|
6608 |
],
|
6609 |
"logging_steps": 1,
|
|
|
6623 |
"attributes": {}
|
6624 |
}
|
6625 |
},
|
6626 |
+
"total_flos": 4.0964084154866074e+17,
|
6627 |
"train_batch_size": 2,
|
6628 |
"trial_name": null,
|
6629 |
"trial_params": null
|