Training in progress, step 864, 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 83945296
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:bcdb9cb208ebb75e860a377cc3bf001ca45cff167681ab6b6dbdf44c76471274
|
3 |
size 83945296
|
last-checkpoint/optimizer.pt
CHANGED
@@ -1,3 +1,3 @@
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
-
oid sha256:
|
3 |
size 43123028
|
|
|
1 |
version https://git-lfs.github.com/spec/v1
|
2 |
+
oid sha256:568a18c89520fe1f6fd6e6493501eb7298f38eb3b112826e2a05c1183470dec9
|
3 |
size 43123028
|
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:d09d7adcedf3287181f669217e74088a48babf4de561c51e83cb3dd6c8b9b5c2
|
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 sha256:09aa92b7b6de13a857c8971fde89595b685a9abe0c0f4e2242f281b0e90222f5
|
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": 288,
|
6 |
-
"global_step":
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
@@ -4055,6 +4055,2030 @@
|
|
4055 |
"eval_samples_per_second": 17.705,
|
4056 |
"eval_steps_per_second": 8.871,
|
4057 |
"step": 576
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
4058 |
}
|
4059 |
],
|
4060 |
"logging_steps": 1,
|
@@ -4074,7 +6098,7 @@
|
|
4074 |
"attributes": {}
|
4075 |
}
|
4076 |
},
|
4077 |
-
"total_flos":
|
4078 |
"train_batch_size": 2,
|
4079 |
"trial_name": null,
|
4080 |
"trial_params": null
|
|
|
1 |
{
|
2 |
"best_metric": null,
|
3 |
"best_model_checkpoint": null,
|
4 |
+
"epoch": 0.75,
|
5 |
"eval_steps": 288,
|
6 |
+
"global_step": 864,
|
7 |
"is_hyper_param_search": false,
|
8 |
"is_local_process_zero": true,
|
9 |
"is_world_process_zero": true,
|
|
|
4055 |
"eval_samples_per_second": 17.705,
|
4056 |
"eval_steps_per_second": 8.871,
|
4057 |
"step": 576
|
4058 |
+
},
|
4059 |
+
{
|
4060 |
+
"epoch": 0.5008680555555556,
|
4061 |
+
"grad_norm": 15.022905349731445,
|
4062 |
+
"learning_rate": 0.00010110036050943203,
|
4063 |
+
"loss": 1.8069,
|
4064 |
+
"step": 577
|
4065 |
+
},
|
4066 |
+
{
|
4067 |
+
"epoch": 0.5017361111111112,
|
4068 |
+
"grad_norm": 14.939796447753906,
|
4069 |
+
"learning_rate": 0.00010082527766851151,
|
4070 |
+
"loss": 1.8282,
|
4071 |
+
"step": 578
|
4072 |
+
},
|
4073 |
+
{
|
4074 |
+
"epoch": 0.5026041666666666,
|
4075 |
+
"grad_norm": 15.488813400268555,
|
4076 |
+
"learning_rate": 0.00010055018858208886,
|
4077 |
+
"loss": 1.8347,
|
4078 |
+
"step": 579
|
4079 |
+
},
|
4080 |
+
{
|
4081 |
+
"epoch": 0.5034722222222222,
|
4082 |
+
"grad_norm": 14.688309669494629,
|
4083 |
+
"learning_rate": 0.00010027509533197197,
|
4084 |
+
"loss": 1.7819,
|
4085 |
+
"step": 580
|
4086 |
+
},
|
4087 |
+
{
|
4088 |
+
"epoch": 0.5043402777777778,
|
4089 |
+
"grad_norm": 12.491915702819824,
|
4090 |
+
"learning_rate": 0.0001,
|
4091 |
+
"loss": 1.9216,
|
4092 |
+
"step": 581
|
4093 |
+
},
|
4094 |
+
{
|
4095 |
+
"epoch": 0.5052083333333334,
|
4096 |
+
"grad_norm": 15.511882781982422,
|
4097 |
+
"learning_rate": 9.972490466802805e-05,
|
4098 |
+
"loss": 1.9545,
|
4099 |
+
"step": 582
|
4100 |
+
},
|
4101 |
+
{
|
4102 |
+
"epoch": 0.5060763888888888,
|
4103 |
+
"grad_norm": 16.1308650970459,
|
4104 |
+
"learning_rate": 9.944981141791115e-05,
|
4105 |
+
"loss": 1.7643,
|
4106 |
+
"step": 583
|
4107 |
+
},
|
4108 |
+
{
|
4109 |
+
"epoch": 0.5069444444444444,
|
4110 |
+
"grad_norm": 16.254255294799805,
|
4111 |
+
"learning_rate": 9.917472233148851e-05,
|
4112 |
+
"loss": 1.875,
|
4113 |
+
"step": 584
|
4114 |
+
},
|
4115 |
+
{
|
4116 |
+
"epoch": 0.5078125,
|
4117 |
+
"grad_norm": 13.440431594848633,
|
4118 |
+
"learning_rate": 9.889963949056799e-05,
|
4119 |
+
"loss": 1.7714,
|
4120 |
+
"step": 585
|
4121 |
+
},
|
4122 |
+
{
|
4123 |
+
"epoch": 0.5086805555555556,
|
4124 |
+
"grad_norm": 14.32579231262207,
|
4125 |
+
"learning_rate": 9.862456497691003e-05,
|
4126 |
+
"loss": 2.0208,
|
4127 |
+
"step": 586
|
4128 |
+
},
|
4129 |
+
{
|
4130 |
+
"epoch": 0.5095486111111112,
|
4131 |
+
"grad_norm": 16.822589874267578,
|
4132 |
+
"learning_rate": 9.834950087221216e-05,
|
4133 |
+
"loss": 1.9641,
|
4134 |
+
"step": 587
|
4135 |
+
},
|
4136 |
+
{
|
4137 |
+
"epoch": 0.5104166666666666,
|
4138 |
+
"grad_norm": 20.17698860168457,
|
4139 |
+
"learning_rate": 9.807444925809306e-05,
|
4140 |
+
"loss": 1.7131,
|
4141 |
+
"step": 588
|
4142 |
+
},
|
4143 |
+
{
|
4144 |
+
"epoch": 0.5112847222222222,
|
4145 |
+
"grad_norm": 14.65821647644043,
|
4146 |
+
"learning_rate": 9.77994122160769e-05,
|
4147 |
+
"loss": 2.01,
|
4148 |
+
"step": 589
|
4149 |
+
},
|
4150 |
+
{
|
4151 |
+
"epoch": 0.5121527777777778,
|
4152 |
+
"grad_norm": 14.805367469787598,
|
4153 |
+
"learning_rate": 9.752439182757764e-05,
|
4154 |
+
"loss": 2.0834,
|
4155 |
+
"step": 590
|
4156 |
+
},
|
4157 |
+
{
|
4158 |
+
"epoch": 0.5130208333333334,
|
4159 |
+
"grad_norm": 16.649839401245117,
|
4160 |
+
"learning_rate": 9.724939017388313e-05,
|
4161 |
+
"loss": 1.9286,
|
4162 |
+
"step": 591
|
4163 |
+
},
|
4164 |
+
{
|
4165 |
+
"epoch": 0.5138888888888888,
|
4166 |
+
"grad_norm": 16.793479919433594,
|
4167 |
+
"learning_rate": 9.697440933613946e-05,
|
4168 |
+
"loss": 1.9679,
|
4169 |
+
"step": 592
|
4170 |
+
},
|
4171 |
+
{
|
4172 |
+
"epoch": 0.5147569444444444,
|
4173 |
+
"grad_norm": 18.385112762451172,
|
4174 |
+
"learning_rate": 9.669945139533524e-05,
|
4175 |
+
"loss": 2.1764,
|
4176 |
+
"step": 593
|
4177 |
+
},
|
4178 |
+
{
|
4179 |
+
"epoch": 0.515625,
|
4180 |
+
"grad_norm": 16.447195053100586,
|
4181 |
+
"learning_rate": 9.642451843228568e-05,
|
4182 |
+
"loss": 2.2571,
|
4183 |
+
"step": 594
|
4184 |
+
},
|
4185 |
+
{
|
4186 |
+
"epoch": 0.5164930555555556,
|
4187 |
+
"grad_norm": 15.659928321838379,
|
4188 |
+
"learning_rate": 9.614961252761713e-05,
|
4189 |
+
"loss": 2.1288,
|
4190 |
+
"step": 595
|
4191 |
+
},
|
4192 |
+
{
|
4193 |
+
"epoch": 0.5173611111111112,
|
4194 |
+
"grad_norm": 16.766855239868164,
|
4195 |
+
"learning_rate": 9.587473576175103e-05,
|
4196 |
+
"loss": 2.0009,
|
4197 |
+
"step": 596
|
4198 |
+
},
|
4199 |
+
{
|
4200 |
+
"epoch": 0.5182291666666666,
|
4201 |
+
"grad_norm": 21.079097747802734,
|
4202 |
+
"learning_rate": 9.559989021488838e-05,
|
4203 |
+
"loss": 1.7899,
|
4204 |
+
"step": 597
|
4205 |
+
},
|
4206 |
+
{
|
4207 |
+
"epoch": 0.5190972222222222,
|
4208 |
+
"grad_norm": 20.679733276367188,
|
4209 |
+
"learning_rate": 9.532507796699388e-05,
|
4210 |
+
"loss": 1.9093,
|
4211 |
+
"step": 598
|
4212 |
+
},
|
4213 |
+
{
|
4214 |
+
"epoch": 0.5199652777777778,
|
4215 |
+
"grad_norm": 20.416423797607422,
|
4216 |
+
"learning_rate": 9.505030109778023e-05,
|
4217 |
+
"loss": 2.0236,
|
4218 |
+
"step": 599
|
4219 |
+
},
|
4220 |
+
{
|
4221 |
+
"epoch": 0.5208333333333334,
|
4222 |
+
"grad_norm": 30.040443420410156,
|
4223 |
+
"learning_rate": 9.47755616866925e-05,
|
4224 |
+
"loss": 1.7996,
|
4225 |
+
"step": 600
|
4226 |
+
},
|
4227 |
+
{
|
4228 |
+
"epoch": 0.5217013888888888,
|
4229 |
+
"grad_norm": 9.230992317199707,
|
4230 |
+
"learning_rate": 9.450086181289213e-05,
|
4231 |
+
"loss": 2.108,
|
4232 |
+
"step": 601
|
4233 |
+
},
|
4234 |
+
{
|
4235 |
+
"epoch": 0.5225694444444444,
|
4236 |
+
"grad_norm": 11.033724784851074,
|
4237 |
+
"learning_rate": 9.422620355524144e-05,
|
4238 |
+
"loss": 1.9422,
|
4239 |
+
"step": 602
|
4240 |
+
},
|
4241 |
+
{
|
4242 |
+
"epoch": 0.5234375,
|
4243 |
+
"grad_norm": 11.256922721862793,
|
4244 |
+
"learning_rate": 9.395158899228777e-05,
|
4245 |
+
"loss": 1.9754,
|
4246 |
+
"step": 603
|
4247 |
+
},
|
4248 |
+
{
|
4249 |
+
"epoch": 0.5243055555555556,
|
4250 |
+
"grad_norm": 14.465068817138672,
|
4251 |
+
"learning_rate": 9.367702020224782e-05,
|
4252 |
+
"loss": 1.8877,
|
4253 |
+
"step": 604
|
4254 |
+
},
|
4255 |
+
{
|
4256 |
+
"epoch": 0.5251736111111112,
|
4257 |
+
"grad_norm": 10.427090644836426,
|
4258 |
+
"learning_rate": 9.340249926299192e-05,
|
4259 |
+
"loss": 1.8378,
|
4260 |
+
"step": 605
|
4261 |
+
},
|
4262 |
+
{
|
4263 |
+
"epoch": 0.5260416666666666,
|
4264 |
+
"grad_norm": 11.574509620666504,
|
4265 |
+
"learning_rate": 9.312802825202817e-05,
|
4266 |
+
"loss": 1.8754,
|
4267 |
+
"step": 606
|
4268 |
+
},
|
4269 |
+
{
|
4270 |
+
"epoch": 0.5269097222222222,
|
4271 |
+
"grad_norm": 11.767155647277832,
|
4272 |
+
"learning_rate": 9.285360924648695e-05,
|
4273 |
+
"loss": 1.8248,
|
4274 |
+
"step": 607
|
4275 |
+
},
|
4276 |
+
{
|
4277 |
+
"epoch": 0.5277777777777778,
|
4278 |
+
"grad_norm": 11.879499435424805,
|
4279 |
+
"learning_rate": 9.257924432310497e-05,
|
4280 |
+
"loss": 1.8081,
|
4281 |
+
"step": 608
|
4282 |
+
},
|
4283 |
+
{
|
4284 |
+
"epoch": 0.5286458333333334,
|
4285 |
+
"grad_norm": 13.207613945007324,
|
4286 |
+
"learning_rate": 9.230493555820971e-05,
|
4287 |
+
"loss": 2.1181,
|
4288 |
+
"step": 609
|
4289 |
+
},
|
4290 |
+
{
|
4291 |
+
"epoch": 0.5295138888888888,
|
4292 |
+
"grad_norm": 11.93727970123291,
|
4293 |
+
"learning_rate": 9.203068502770371e-05,
|
4294 |
+
"loss": 2.1128,
|
4295 |
+
"step": 610
|
4296 |
+
},
|
4297 |
+
{
|
4298 |
+
"epoch": 0.5303819444444444,
|
4299 |
+
"grad_norm": 11.551005363464355,
|
4300 |
+
"learning_rate": 9.17564948070487e-05,
|
4301 |
+
"loss": 1.9466,
|
4302 |
+
"step": 611
|
4303 |
+
},
|
4304 |
+
{
|
4305 |
+
"epoch": 0.53125,
|
4306 |
+
"grad_norm": 12.332575798034668,
|
4307 |
+
"learning_rate": 9.148236697125008e-05,
|
4308 |
+
"loss": 1.809,
|
4309 |
+
"step": 612
|
4310 |
+
},
|
4311 |
+
{
|
4312 |
+
"epoch": 0.5321180555555556,
|
4313 |
+
"grad_norm": 12.674099922180176,
|
4314 |
+
"learning_rate": 9.120830359484105e-05,
|
4315 |
+
"loss": 1.8998,
|
4316 |
+
"step": 613
|
4317 |
+
},
|
4318 |
+
{
|
4319 |
+
"epoch": 0.5329861111111112,
|
4320 |
+
"grad_norm": 13.731175422668457,
|
4321 |
+
"learning_rate": 9.093430675186708e-05,
|
4322 |
+
"loss": 1.8718,
|
4323 |
+
"step": 614
|
4324 |
+
},
|
4325 |
+
{
|
4326 |
+
"epoch": 0.5338541666666666,
|
4327 |
+
"grad_norm": 14.158843040466309,
|
4328 |
+
"learning_rate": 9.066037851587015e-05,
|
4329 |
+
"loss": 1.834,
|
4330 |
+
"step": 615
|
4331 |
+
},
|
4332 |
+
{
|
4333 |
+
"epoch": 0.5347222222222222,
|
4334 |
+
"grad_norm": 14.000466346740723,
|
4335 |
+
"learning_rate": 9.038652095987294e-05,
|
4336 |
+
"loss": 1.7906,
|
4337 |
+
"step": 616
|
4338 |
+
},
|
4339 |
+
{
|
4340 |
+
"epoch": 0.5355902777777778,
|
4341 |
+
"grad_norm": 13.29843807220459,
|
4342 |
+
"learning_rate": 9.01127361563633e-05,
|
4343 |
+
"loss": 1.7885,
|
4344 |
+
"step": 617
|
4345 |
+
},
|
4346 |
+
{
|
4347 |
+
"epoch": 0.5364583333333334,
|
4348 |
+
"grad_norm": 13.127632141113281,
|
4349 |
+
"learning_rate": 8.983902617727852e-05,
|
4350 |
+
"loss": 2.044,
|
4351 |
+
"step": 618
|
4352 |
+
},
|
4353 |
+
{
|
4354 |
+
"epoch": 0.5373263888888888,
|
4355 |
+
"grad_norm": 13.030488014221191,
|
4356 |
+
"learning_rate": 8.956539309398957e-05,
|
4357 |
+
"loss": 1.7404,
|
4358 |
+
"step": 619
|
4359 |
+
},
|
4360 |
+
{
|
4361 |
+
"epoch": 0.5381944444444444,
|
4362 |
+
"grad_norm": 15.269593238830566,
|
4363 |
+
"learning_rate": 8.92918389772856e-05,
|
4364 |
+
"loss": 1.917,
|
4365 |
+
"step": 620
|
4366 |
+
},
|
4367 |
+
{
|
4368 |
+
"epoch": 0.5390625,
|
4369 |
+
"grad_norm": 13.017799377441406,
|
4370 |
+
"learning_rate": 8.901836589735807e-05,
|
4371 |
+
"loss": 2.0645,
|
4372 |
+
"step": 621
|
4373 |
+
},
|
4374 |
+
{
|
4375 |
+
"epoch": 0.5399305555555556,
|
4376 |
+
"grad_norm": 12.069060325622559,
|
4377 |
+
"learning_rate": 8.87449759237852e-05,
|
4378 |
+
"loss": 1.8962,
|
4379 |
+
"step": 622
|
4380 |
+
},
|
4381 |
+
{
|
4382 |
+
"epoch": 0.5407986111111112,
|
4383 |
+
"grad_norm": 15.789977073669434,
|
4384 |
+
"learning_rate": 8.847167112551627e-05,
|
4385 |
+
"loss": 1.7837,
|
4386 |
+
"step": 623
|
4387 |
+
},
|
4388 |
+
{
|
4389 |
+
"epoch": 0.5416666666666666,
|
4390 |
+
"grad_norm": 13.995849609375,
|
4391 |
+
"learning_rate": 8.819845357085598e-05,
|
4392 |
+
"loss": 1.8621,
|
4393 |
+
"step": 624
|
4394 |
+
},
|
4395 |
+
{
|
4396 |
+
"epoch": 0.5425347222222222,
|
4397 |
+
"grad_norm": 13.478865623474121,
|
4398 |
+
"learning_rate": 8.792532532744885e-05,
|
4399 |
+
"loss": 1.9676,
|
4400 |
+
"step": 625
|
4401 |
+
},
|
4402 |
+
{
|
4403 |
+
"epoch": 0.5434027777777778,
|
4404 |
+
"grad_norm": 15.19690990447998,
|
4405 |
+
"learning_rate": 8.765228846226341e-05,
|
4406 |
+
"loss": 1.6361,
|
4407 |
+
"step": 626
|
4408 |
+
},
|
4409 |
+
{
|
4410 |
+
"epoch": 0.5442708333333334,
|
4411 |
+
"grad_norm": 14.974311828613281,
|
4412 |
+
"learning_rate": 8.737934504157675e-05,
|
4413 |
+
"loss": 1.9447,
|
4414 |
+
"step": 627
|
4415 |
+
},
|
4416 |
+
{
|
4417 |
+
"epoch": 0.5451388888888888,
|
4418 |
+
"grad_norm": 15.628581047058105,
|
4419 |
+
"learning_rate": 8.710649713095872e-05,
|
4420 |
+
"loss": 1.8223,
|
4421 |
+
"step": 628
|
4422 |
+
},
|
4423 |
+
{
|
4424 |
+
"epoch": 0.5460069444444444,
|
4425 |
+
"grad_norm": 13.494054794311523,
|
4426 |
+
"learning_rate": 8.683374679525644e-05,
|
4427 |
+
"loss": 1.8148,
|
4428 |
+
"step": 629
|
4429 |
+
},
|
4430 |
+
{
|
4431 |
+
"epoch": 0.546875,
|
4432 |
+
"grad_norm": 14.555102348327637,
|
4433 |
+
"learning_rate": 8.656109609857854e-05,
|
4434 |
+
"loss": 1.8495,
|
4435 |
+
"step": 630
|
4436 |
+
},
|
4437 |
+
{
|
4438 |
+
"epoch": 0.5477430555555556,
|
4439 |
+
"grad_norm": 14.314188957214355,
|
4440 |
+
"learning_rate": 8.628854710427968e-05,
|
4441 |
+
"loss": 1.7517,
|
4442 |
+
"step": 631
|
4443 |
+
},
|
4444 |
+
{
|
4445 |
+
"epoch": 0.5486111111111112,
|
4446 |
+
"grad_norm": 13.424491882324219,
|
4447 |
+
"learning_rate": 8.601610187494484e-05,
|
4448 |
+
"loss": 1.8415,
|
4449 |
+
"step": 632
|
4450 |
+
},
|
4451 |
+
{
|
4452 |
+
"epoch": 0.5494791666666666,
|
4453 |
+
"grad_norm": 19.24957847595215,
|
4454 |
+
"learning_rate": 8.574376247237367e-05,
|
4455 |
+
"loss": 1.6742,
|
4456 |
+
"step": 633
|
4457 |
+
},
|
4458 |
+
{
|
4459 |
+
"epoch": 0.5503472222222222,
|
4460 |
+
"grad_norm": 14.664512634277344,
|
4461 |
+
"learning_rate": 8.547153095756503e-05,
|
4462 |
+
"loss": 2.1591,
|
4463 |
+
"step": 634
|
4464 |
+
},
|
4465 |
+
{
|
4466 |
+
"epoch": 0.5512152777777778,
|
4467 |
+
"grad_norm": 17.534164428710938,
|
4468 |
+
"learning_rate": 8.519940939070121e-05,
|
4469 |
+
"loss": 1.8432,
|
4470 |
+
"step": 635
|
4471 |
+
},
|
4472 |
+
{
|
4473 |
+
"epoch": 0.5520833333333334,
|
4474 |
+
"grad_norm": 15.532064437866211,
|
4475 |
+
"learning_rate": 8.49273998311326e-05,
|
4476 |
+
"loss": 2.0388,
|
4477 |
+
"step": 636
|
4478 |
+
},
|
4479 |
+
{
|
4480 |
+
"epoch": 0.5529513888888888,
|
4481 |
+
"grad_norm": 15.507984161376953,
|
4482 |
+
"learning_rate": 8.46555043373618e-05,
|
4483 |
+
"loss": 2.1336,
|
4484 |
+
"step": 637
|
4485 |
+
},
|
4486 |
+
{
|
4487 |
+
"epoch": 0.5538194444444444,
|
4488 |
+
"grad_norm": 17.11254119873047,
|
4489 |
+
"learning_rate": 8.438372496702826e-05,
|
4490 |
+
"loss": 2.1056,
|
4491 |
+
"step": 638
|
4492 |
+
},
|
4493 |
+
{
|
4494 |
+
"epoch": 0.5546875,
|
4495 |
+
"grad_norm": 16.4410400390625,
|
4496 |
+
"learning_rate": 8.41120637768926e-05,
|
4497 |
+
"loss": 2.143,
|
4498 |
+
"step": 639
|
4499 |
+
},
|
4500 |
+
{
|
4501 |
+
"epoch": 0.5555555555555556,
|
4502 |
+
"grad_norm": 14.77703857421875,
|
4503 |
+
"learning_rate": 8.384052282282107e-05,
|
4504 |
+
"loss": 1.9428,
|
4505 |
+
"step": 640
|
4506 |
+
},
|
4507 |
+
{
|
4508 |
+
"epoch": 0.5564236111111112,
|
4509 |
+
"grad_norm": 17.501911163330078,
|
4510 |
+
"learning_rate": 8.356910415977008e-05,
|
4511 |
+
"loss": 1.8171,
|
4512 |
+
"step": 641
|
4513 |
+
},
|
4514 |
+
{
|
4515 |
+
"epoch": 0.5572916666666666,
|
4516 |
+
"grad_norm": 16.587329864501953,
|
4517 |
+
"learning_rate": 8.329780984177053e-05,
|
4518 |
+
"loss": 2.1078,
|
4519 |
+
"step": 642
|
4520 |
+
},
|
4521 |
+
{
|
4522 |
+
"epoch": 0.5581597222222222,
|
4523 |
+
"grad_norm": 15.023578643798828,
|
4524 |
+
"learning_rate": 8.302664192191225e-05,
|
4525 |
+
"loss": 2.2078,
|
4526 |
+
"step": 643
|
4527 |
+
},
|
4528 |
+
{
|
4529 |
+
"epoch": 0.5590277777777778,
|
4530 |
+
"grad_norm": 15.555996894836426,
|
4531 |
+
"learning_rate": 8.27556024523286e-05,
|
4532 |
+
"loss": 1.9277,
|
4533 |
+
"step": 644
|
4534 |
+
},
|
4535 |
+
{
|
4536 |
+
"epoch": 0.5598958333333334,
|
4537 |
+
"grad_norm": 17.908096313476562,
|
4538 |
+
"learning_rate": 8.248469348418076e-05,
|
4539 |
+
"loss": 2.0114,
|
4540 |
+
"step": 645
|
4541 |
+
},
|
4542 |
+
{
|
4543 |
+
"epoch": 0.5607638888888888,
|
4544 |
+
"grad_norm": 18.860708236694336,
|
4545 |
+
"learning_rate": 8.221391706764245e-05,
|
4546 |
+
"loss": 2.079,
|
4547 |
+
"step": 646
|
4548 |
+
},
|
4549 |
+
{
|
4550 |
+
"epoch": 0.5616319444444444,
|
4551 |
+
"grad_norm": 20.25312614440918,
|
4552 |
+
"learning_rate": 8.194327525188419e-05,
|
4553 |
+
"loss": 1.9205,
|
4554 |
+
"step": 647
|
4555 |
+
},
|
4556 |
+
{
|
4557 |
+
"epoch": 0.5625,
|
4558 |
+
"grad_norm": 26.201187133789062,
|
4559 |
+
"learning_rate": 8.167277008505783e-05,
|
4560 |
+
"loss": 1.9173,
|
4561 |
+
"step": 648
|
4562 |
+
},
|
4563 |
+
{
|
4564 |
+
"epoch": 0.5633680555555556,
|
4565 |
+
"grad_norm": 18.467588424682617,
|
4566 |
+
"learning_rate": 8.140240361428119e-05,
|
4567 |
+
"loss": 2.0093,
|
4568 |
+
"step": 649
|
4569 |
+
},
|
4570 |
+
{
|
4571 |
+
"epoch": 0.5642361111111112,
|
4572 |
+
"grad_norm": 28.01093101501465,
|
4573 |
+
"learning_rate": 8.113217788562235e-05,
|
4574 |
+
"loss": 1.8234,
|
4575 |
+
"step": 650
|
4576 |
+
},
|
4577 |
+
{
|
4578 |
+
"epoch": 0.5651041666666666,
|
4579 |
+
"grad_norm": 9.87932014465332,
|
4580 |
+
"learning_rate": 8.086209494408443e-05,
|
4581 |
+
"loss": 1.6589,
|
4582 |
+
"step": 651
|
4583 |
+
},
|
4584 |
+
{
|
4585 |
+
"epoch": 0.5659722222222222,
|
4586 |
+
"grad_norm": 11.396955490112305,
|
4587 |
+
"learning_rate": 8.059215683358991e-05,
|
4588 |
+
"loss": 2.0744,
|
4589 |
+
"step": 652
|
4590 |
+
},
|
4591 |
+
{
|
4592 |
+
"epoch": 0.5668402777777778,
|
4593 |
+
"grad_norm": 10.34410285949707,
|
4594 |
+
"learning_rate": 8.032236559696517e-05,
|
4595 |
+
"loss": 1.6944,
|
4596 |
+
"step": 653
|
4597 |
+
},
|
4598 |
+
{
|
4599 |
+
"epoch": 0.5677083333333334,
|
4600 |
+
"grad_norm": 12.291624069213867,
|
4601 |
+
"learning_rate": 8.005272327592517e-05,
|
4602 |
+
"loss": 1.7822,
|
4603 |
+
"step": 654
|
4604 |
+
},
|
4605 |
+
{
|
4606 |
+
"epoch": 0.5685763888888888,
|
4607 |
+
"grad_norm": 12.043850898742676,
|
4608 |
+
"learning_rate": 7.978323191105781e-05,
|
4609 |
+
"loss": 1.8903,
|
4610 |
+
"step": 655
|
4611 |
+
},
|
4612 |
+
{
|
4613 |
+
"epoch": 0.5694444444444444,
|
4614 |
+
"grad_norm": 10.367390632629395,
|
4615 |
+
"learning_rate": 7.951389354180871e-05,
|
4616 |
+
"loss": 1.8261,
|
4617 |
+
"step": 656
|
4618 |
+
},
|
4619 |
+
{
|
4620 |
+
"epoch": 0.5703125,
|
4621 |
+
"grad_norm": 10.694307327270508,
|
4622 |
+
"learning_rate": 7.924471020646559e-05,
|
4623 |
+
"loss": 1.8233,
|
4624 |
+
"step": 657
|
4625 |
+
},
|
4626 |
+
{
|
4627 |
+
"epoch": 0.5711805555555556,
|
4628 |
+
"grad_norm": 12.436077117919922,
|
4629 |
+
"learning_rate": 7.897568394214292e-05,
|
4630 |
+
"loss": 1.9725,
|
4631 |
+
"step": 658
|
4632 |
+
},
|
4633 |
+
{
|
4634 |
+
"epoch": 0.5720486111111112,
|
4635 |
+
"grad_norm": 11.22568416595459,
|
4636 |
+
"learning_rate": 7.87068167847665e-05,
|
4637 |
+
"loss": 1.7682,
|
4638 |
+
"step": 659
|
4639 |
+
},
|
4640 |
+
{
|
4641 |
+
"epoch": 0.5729166666666666,
|
4642 |
+
"grad_norm": 11.406827926635742,
|
4643 |
+
"learning_rate": 7.8438110769058e-05,
|
4644 |
+
"loss": 2.1154,
|
4645 |
+
"step": 660
|
4646 |
+
},
|
4647 |
+
{
|
4648 |
+
"epoch": 0.5737847222222222,
|
4649 |
+
"grad_norm": 13.207112312316895,
|
4650 |
+
"learning_rate": 7.81695679285197e-05,
|
4651 |
+
"loss": 1.8807,
|
4652 |
+
"step": 661
|
4653 |
+
},
|
4654 |
+
{
|
4655 |
+
"epoch": 0.5746527777777778,
|
4656 |
+
"grad_norm": 12.020392417907715,
|
4657 |
+
"learning_rate": 7.790119029541893e-05,
|
4658 |
+
"loss": 1.7712,
|
4659 |
+
"step": 662
|
4660 |
+
},
|
4661 |
+
{
|
4662 |
+
"epoch": 0.5755208333333334,
|
4663 |
+
"grad_norm": 12.263861656188965,
|
4664 |
+
"learning_rate": 7.76329799007728e-05,
|
4665 |
+
"loss": 1.7847,
|
4666 |
+
"step": 663
|
4667 |
+
},
|
4668 |
+
{
|
4669 |
+
"epoch": 0.5763888888888888,
|
4670 |
+
"grad_norm": 12.855914115905762,
|
4671 |
+
"learning_rate": 7.736493877433285e-05,
|
4672 |
+
"loss": 1.8194,
|
4673 |
+
"step": 664
|
4674 |
+
},
|
4675 |
+
{
|
4676 |
+
"epoch": 0.5772569444444444,
|
4677 |
+
"grad_norm": 13.804559707641602,
|
4678 |
+
"learning_rate": 7.709706894456949e-05,
|
4679 |
+
"loss": 1.9374,
|
4680 |
+
"step": 665
|
4681 |
+
},
|
4682 |
+
{
|
4683 |
+
"epoch": 0.578125,
|
4684 |
+
"grad_norm": 12.961804389953613,
|
4685 |
+
"learning_rate": 7.682937243865701e-05,
|
4686 |
+
"loss": 2.0661,
|
4687 |
+
"step": 666
|
4688 |
+
},
|
4689 |
+
{
|
4690 |
+
"epoch": 0.5789930555555556,
|
4691 |
+
"grad_norm": 13.783415794372559,
|
4692 |
+
"learning_rate": 7.656185128245786e-05,
|
4693 |
+
"loss": 1.8875,
|
4694 |
+
"step": 667
|
4695 |
+
},
|
4696 |
+
{
|
4697 |
+
"epoch": 0.5798611111111112,
|
4698 |
+
"grad_norm": 14.353681564331055,
|
4699 |
+
"learning_rate": 7.62945075005075e-05,
|
4700 |
+
"loss": 2.1549,
|
4701 |
+
"step": 668
|
4702 |
+
},
|
4703 |
+
{
|
4704 |
+
"epoch": 0.5807291666666666,
|
4705 |
+
"grad_norm": 15.880180358886719,
|
4706 |
+
"learning_rate": 7.602734311599918e-05,
|
4707 |
+
"loss": 1.5437,
|
4708 |
+
"step": 669
|
4709 |
+
},
|
4710 |
+
{
|
4711 |
+
"epoch": 0.5815972222222222,
|
4712 |
+
"grad_norm": 17.51071548461914,
|
4713 |
+
"learning_rate": 7.576036015076837e-05,
|
4714 |
+
"loss": 1.9075,
|
4715 |
+
"step": 670
|
4716 |
+
},
|
4717 |
+
{
|
4718 |
+
"epoch": 0.5824652777777778,
|
4719 |
+
"grad_norm": 12.360156059265137,
|
4720 |
+
"learning_rate": 7.549356062527773e-05,
|
4721 |
+
"loss": 1.8808,
|
4722 |
+
"step": 671
|
4723 |
+
},
|
4724 |
+
{
|
4725 |
+
"epoch": 0.5833333333333334,
|
4726 |
+
"grad_norm": 14.916133880615234,
|
4727 |
+
"learning_rate": 7.52269465586016e-05,
|
4728 |
+
"loss": 1.772,
|
4729 |
+
"step": 672
|
4730 |
+
},
|
4731 |
+
{
|
4732 |
+
"epoch": 0.5842013888888888,
|
4733 |
+
"grad_norm": 14.223649024963379,
|
4734 |
+
"learning_rate": 7.496051996841084e-05,
|
4735 |
+
"loss": 1.959,
|
4736 |
+
"step": 673
|
4737 |
+
},
|
4738 |
+
{
|
4739 |
+
"epoch": 0.5850694444444444,
|
4740 |
+
"grad_norm": 15.88371753692627,
|
4741 |
+
"learning_rate": 7.469428287095757e-05,
|
4742 |
+
"loss": 1.7439,
|
4743 |
+
"step": 674
|
4744 |
+
},
|
4745 |
+
{
|
4746 |
+
"epoch": 0.5859375,
|
4747 |
+
"grad_norm": 13.079479217529297,
|
4748 |
+
"learning_rate": 7.442823728105979e-05,
|
4749 |
+
"loss": 1.8794,
|
4750 |
+
"step": 675
|
4751 |
+
},
|
4752 |
+
{
|
4753 |
+
"epoch": 0.5868055555555556,
|
4754 |
+
"grad_norm": 17.964012145996094,
|
4755 |
+
"learning_rate": 7.416238521208634e-05,
|
4756 |
+
"loss": 1.5014,
|
4757 |
+
"step": 676
|
4758 |
+
},
|
4759 |
+
{
|
4760 |
+
"epoch": 0.5876736111111112,
|
4761 |
+
"grad_norm": 14.494117736816406,
|
4762 |
+
"learning_rate": 7.389672867594142e-05,
|
4763 |
+
"loss": 2.2386,
|
4764 |
+
"step": 677
|
4765 |
+
},
|
4766 |
+
{
|
4767 |
+
"epoch": 0.5885416666666666,
|
4768 |
+
"grad_norm": 16.09971809387207,
|
4769 |
+
"learning_rate": 7.363126968304955e-05,
|
4770 |
+
"loss": 1.8013,
|
4771 |
+
"step": 678
|
4772 |
+
},
|
4773 |
+
{
|
4774 |
+
"epoch": 0.5894097222222222,
|
4775 |
+
"grad_norm": 13.50377368927002,
|
4776 |
+
"learning_rate": 7.336601024234028e-05,
|
4777 |
+
"loss": 1.7716,
|
4778 |
+
"step": 679
|
4779 |
+
},
|
4780 |
+
{
|
4781 |
+
"epoch": 0.5902777777777778,
|
4782 |
+
"grad_norm": 17.585285186767578,
|
4783 |
+
"learning_rate": 7.310095236123295e-05,
|
4784 |
+
"loss": 2.0324,
|
4785 |
+
"step": 680
|
4786 |
+
},
|
4787 |
+
{
|
4788 |
+
"epoch": 0.5911458333333334,
|
4789 |
+
"grad_norm": 15.79974365234375,
|
4790 |
+
"learning_rate": 7.283609804562168e-05,
|
4791 |
+
"loss": 2.0764,
|
4792 |
+
"step": 681
|
4793 |
+
},
|
4794 |
+
{
|
4795 |
+
"epoch": 0.5920138888888888,
|
4796 |
+
"grad_norm": 15.37767505645752,
|
4797 |
+
"learning_rate": 7.257144929985988e-05,
|
4798 |
+
"loss": 1.6588,
|
4799 |
+
"step": 682
|
4800 |
+
},
|
4801 |
+
{
|
4802 |
+
"epoch": 0.5928819444444444,
|
4803 |
+
"grad_norm": 16.093576431274414,
|
4804 |
+
"learning_rate": 7.230700812674538e-05,
|
4805 |
+
"loss": 2.0325,
|
4806 |
+
"step": 683
|
4807 |
+
},
|
4808 |
+
{
|
4809 |
+
"epoch": 0.59375,
|
4810 |
+
"grad_norm": 18.15459442138672,
|
4811 |
+
"learning_rate": 7.204277652750512e-05,
|
4812 |
+
"loss": 1.7897,
|
4813 |
+
"step": 684
|
4814 |
+
},
|
4815 |
+
{
|
4816 |
+
"epoch": 0.5946180555555556,
|
4817 |
+
"grad_norm": 12.621081352233887,
|
4818 |
+
"learning_rate": 7.177875650177998e-05,
|
4819 |
+
"loss": 1.7661,
|
4820 |
+
"step": 685
|
4821 |
+
},
|
4822 |
+
{
|
4823 |
+
"epoch": 0.5954861111111112,
|
4824 |
+
"grad_norm": 14.055307388305664,
|
4825 |
+
"learning_rate": 7.15149500476098e-05,
|
4826 |
+
"loss": 1.9542,
|
4827 |
+
"step": 686
|
4828 |
+
},
|
4829 |
+
{
|
4830 |
+
"epoch": 0.5963541666666666,
|
4831 |
+
"grad_norm": 18.60757064819336,
|
4832 |
+
"learning_rate": 7.12513591614181e-05,
|
4833 |
+
"loss": 1.9329,
|
4834 |
+
"step": 687
|
4835 |
+
},
|
4836 |
+
{
|
4837 |
+
"epoch": 0.5972222222222222,
|
4838 |
+
"grad_norm": 15.511977195739746,
|
4839 |
+
"learning_rate": 7.098798583799701e-05,
|
4840 |
+
"loss": 1.9413,
|
4841 |
+
"step": 688
|
4842 |
+
},
|
4843 |
+
{
|
4844 |
+
"epoch": 0.5980902777777778,
|
4845 |
+
"grad_norm": 17.695068359375,
|
4846 |
+
"learning_rate": 7.072483207049232e-05,
|
4847 |
+
"loss": 1.9205,
|
4848 |
+
"step": 689
|
4849 |
+
},
|
4850 |
+
{
|
4851 |
+
"epoch": 0.5989583333333334,
|
4852 |
+
"grad_norm": 39.13572692871094,
|
4853 |
+
"learning_rate": 7.046189985038807e-05,
|
4854 |
+
"loss": 1.7923,
|
4855 |
+
"step": 690
|
4856 |
+
},
|
4857 |
+
{
|
4858 |
+
"epoch": 0.5998263888888888,
|
4859 |
+
"grad_norm": 13.981321334838867,
|
4860 |
+
"learning_rate": 7.01991911674919e-05,
|
4861 |
+
"loss": 1.999,
|
4862 |
+
"step": 691
|
4863 |
+
},
|
4864 |
+
{
|
4865 |
+
"epoch": 0.6006944444444444,
|
4866 |
+
"grad_norm": 16.080123901367188,
|
4867 |
+
"learning_rate": 6.99367080099196e-05,
|
4868 |
+
"loss": 1.8519,
|
4869 |
+
"step": 692
|
4870 |
+
},
|
4871 |
+
{
|
4872 |
+
"epoch": 0.6015625,
|
4873 |
+
"grad_norm": 16.789819717407227,
|
4874 |
+
"learning_rate": 6.967445236408037e-05,
|
4875 |
+
"loss": 2.07,
|
4876 |
+
"step": 693
|
4877 |
+
},
|
4878 |
+
{
|
4879 |
+
"epoch": 0.6024305555555556,
|
4880 |
+
"grad_norm": 18.676746368408203,
|
4881 |
+
"learning_rate": 6.941242621466157e-05,
|
4882 |
+
"loss": 1.8944,
|
4883 |
+
"step": 694
|
4884 |
+
},
|
4885 |
+
{
|
4886 |
+
"epoch": 0.6032986111111112,
|
4887 |
+
"grad_norm": 18.478113174438477,
|
4888 |
+
"learning_rate": 6.915063154461379e-05,
|
4889 |
+
"loss": 2.1065,
|
4890 |
+
"step": 695
|
4891 |
+
},
|
4892 |
+
{
|
4893 |
+
"epoch": 0.6041666666666666,
|
4894 |
+
"grad_norm": 15.811859130859375,
|
4895 |
+
"learning_rate": 6.888907033513592e-05,
|
4896 |
+
"loss": 2.0354,
|
4897 |
+
"step": 696
|
4898 |
+
},
|
4899 |
+
{
|
4900 |
+
"epoch": 0.6050347222222222,
|
4901 |
+
"grad_norm": 21.135038375854492,
|
4902 |
+
"learning_rate": 6.862774456565999e-05,
|
4903 |
+
"loss": 1.9646,
|
4904 |
+
"step": 697
|
4905 |
+
},
|
4906 |
+
{
|
4907 |
+
"epoch": 0.6059027777777778,
|
4908 |
+
"grad_norm": 27.442007064819336,
|
4909 |
+
"learning_rate": 6.836665621383635e-05,
|
4910 |
+
"loss": 1.7368,
|
4911 |
+
"step": 698
|
4912 |
+
},
|
4913 |
+
{
|
4914 |
+
"epoch": 0.6067708333333334,
|
4915 |
+
"grad_norm": 21.151498794555664,
|
4916 |
+
"learning_rate": 6.810580725551853e-05,
|
4917 |
+
"loss": 1.8765,
|
4918 |
+
"step": 699
|
4919 |
+
},
|
4920 |
+
{
|
4921 |
+
"epoch": 0.6076388888888888,
|
4922 |
+
"grad_norm": 29.93486213684082,
|
4923 |
+
"learning_rate": 6.784519966474845e-05,
|
4924 |
+
"loss": 2.1535,
|
4925 |
+
"step": 700
|
4926 |
+
},
|
4927 |
+
{
|
4928 |
+
"epoch": 0.6085069444444444,
|
4929 |
+
"grad_norm": 9.31676959991455,
|
4930 |
+
"learning_rate": 6.758483541374148e-05,
|
4931 |
+
"loss": 1.8469,
|
4932 |
+
"step": 701
|
4933 |
+
},
|
4934 |
+
{
|
4935 |
+
"epoch": 0.609375,
|
4936 |
+
"grad_norm": 11.149514198303223,
|
4937 |
+
"learning_rate": 6.732471647287133e-05,
|
4938 |
+
"loss": 1.9865,
|
4939 |
+
"step": 702
|
4940 |
+
},
|
4941 |
+
{
|
4942 |
+
"epoch": 0.6102430555555556,
|
4943 |
+
"grad_norm": 10.093915939331055,
|
4944 |
+
"learning_rate": 6.706484481065539e-05,
|
4945 |
+
"loss": 1.7877,
|
4946 |
+
"step": 703
|
4947 |
+
},
|
4948 |
+
{
|
4949 |
+
"epoch": 0.6111111111111112,
|
4950 |
+
"grad_norm": 10.389727592468262,
|
4951 |
+
"learning_rate": 6.680522239373953e-05,
|
4952 |
+
"loss": 1.7289,
|
4953 |
+
"step": 704
|
4954 |
+
},
|
4955 |
+
{
|
4956 |
+
"epoch": 0.6119791666666666,
|
4957 |
+
"grad_norm": 10.033374786376953,
|
4958 |
+
"learning_rate": 6.654585118688356e-05,
|
4959 |
+
"loss": 1.9985,
|
4960 |
+
"step": 705
|
4961 |
+
},
|
4962 |
+
{
|
4963 |
+
"epoch": 0.6128472222222222,
|
4964 |
+
"grad_norm": 10.770398139953613,
|
4965 |
+
"learning_rate": 6.628673315294616e-05,
|
4966 |
+
"loss": 1.9511,
|
4967 |
+
"step": 706
|
4968 |
+
},
|
4969 |
+
{
|
4970 |
+
"epoch": 0.6137152777777778,
|
4971 |
+
"grad_norm": 10.429977416992188,
|
4972 |
+
"learning_rate": 6.602787025286998e-05,
|
4973 |
+
"loss": 2.0326,
|
4974 |
+
"step": 707
|
4975 |
+
},
|
4976 |
+
{
|
4977 |
+
"epoch": 0.6145833333333334,
|
4978 |
+
"grad_norm": 10.858638763427734,
|
4979 |
+
"learning_rate": 6.576926444566698e-05,
|
4980 |
+
"loss": 2.0785,
|
4981 |
+
"step": 708
|
4982 |
+
},
|
4983 |
+
{
|
4984 |
+
"epoch": 0.6154513888888888,
|
4985 |
+
"grad_norm": 12.585272789001465,
|
4986 |
+
"learning_rate": 6.551091768840338e-05,
|
4987 |
+
"loss": 2.1141,
|
4988 |
+
"step": 709
|
4989 |
+
},
|
4990 |
+
{
|
4991 |
+
"epoch": 0.6163194444444444,
|
4992 |
+
"grad_norm": 12.708643913269043,
|
4993 |
+
"learning_rate": 6.525283193618508e-05,
|
4994 |
+
"loss": 1.6813,
|
4995 |
+
"step": 710
|
4996 |
+
},
|
4997 |
+
{
|
4998 |
+
"epoch": 0.6171875,
|
4999 |
+
"grad_norm": 11.747328758239746,
|
5000 |
+
"learning_rate": 6.499500914214276e-05,
|
5001 |
+
"loss": 1.9528,
|
5002 |
+
"step": 711
|
5003 |
+
},
|
5004 |
+
{
|
5005 |
+
"epoch": 0.6180555555555556,
|
5006 |
+
"grad_norm": 15.570380210876465,
|
5007 |
+
"learning_rate": 6.473745125741701e-05,
|
5008 |
+
"loss": 1.7305,
|
5009 |
+
"step": 712
|
5010 |
+
},
|
5011 |
+
{
|
5012 |
+
"epoch": 0.6189236111111112,
|
5013 |
+
"grad_norm": 12.348931312561035,
|
5014 |
+
"learning_rate": 6.448016023114372e-05,
|
5015 |
+
"loss": 1.7636,
|
5016 |
+
"step": 713
|
5017 |
+
},
|
5018 |
+
{
|
5019 |
+
"epoch": 0.6197916666666666,
|
5020 |
+
"grad_norm": 12.029678344726562,
|
5021 |
+
"learning_rate": 6.422313801043922e-05,
|
5022 |
+
"loss": 1.7962,
|
5023 |
+
"step": 714
|
5024 |
+
},
|
5025 |
+
{
|
5026 |
+
"epoch": 0.6206597222222222,
|
5027 |
+
"grad_norm": 13.390205383300781,
|
5028 |
+
"learning_rate": 6.396638654038557e-05,
|
5029 |
+
"loss": 1.8049,
|
5030 |
+
"step": 715
|
5031 |
+
},
|
5032 |
+
{
|
5033 |
+
"epoch": 0.6215277777777778,
|
5034 |
+
"grad_norm": 12.542319297790527,
|
5035 |
+
"learning_rate": 6.370990776401597e-05,
|
5036 |
+
"loss": 1.69,
|
5037 |
+
"step": 716
|
5038 |
+
},
|
5039 |
+
{
|
5040 |
+
"epoch": 0.6223958333333334,
|
5041 |
+
"grad_norm": 14.50907039642334,
|
5042 |
+
"learning_rate": 6.34537036222998e-05,
|
5043 |
+
"loss": 1.7879,
|
5044 |
+
"step": 717
|
5045 |
+
},
|
5046 |
+
{
|
5047 |
+
"epoch": 0.6232638888888888,
|
5048 |
+
"grad_norm": 14.686508178710938,
|
5049 |
+
"learning_rate": 6.319777605412816e-05,
|
5050 |
+
"loss": 1.685,
|
5051 |
+
"step": 718
|
5052 |
+
},
|
5053 |
+
{
|
5054 |
+
"epoch": 0.6241319444444444,
|
5055 |
+
"grad_norm": 12.963170051574707,
|
5056 |
+
"learning_rate": 6.294212699629902e-05,
|
5057 |
+
"loss": 1.8536,
|
5058 |
+
"step": 719
|
5059 |
+
},
|
5060 |
+
{
|
5061 |
+
"epoch": 0.625,
|
5062 |
+
"grad_norm": 10.97075366973877,
|
5063 |
+
"learning_rate": 6.268675838350273e-05,
|
5064 |
+
"loss": 1.7408,
|
5065 |
+
"step": 720
|
5066 |
+
},
|
5067 |
+
{
|
5068 |
+
"epoch": 0.6258680555555556,
|
5069 |
+
"grad_norm": 12.989684104919434,
|
5070 |
+
"learning_rate": 6.243167214830733e-05,
|
5071 |
+
"loss": 1.9632,
|
5072 |
+
"step": 721
|
5073 |
+
},
|
5074 |
+
{
|
5075 |
+
"epoch": 0.6267361111111112,
|
5076 |
+
"grad_norm": 16.108964920043945,
|
5077 |
+
"learning_rate": 6.217687022114377e-05,
|
5078 |
+
"loss": 1.6076,
|
5079 |
+
"step": 722
|
5080 |
+
},
|
5081 |
+
{
|
5082 |
+
"epoch": 0.6276041666666666,
|
5083 |
+
"grad_norm": 13.303749084472656,
|
5084 |
+
"learning_rate": 6.192235453029155e-05,
|
5085 |
+
"loss": 1.7805,
|
5086 |
+
"step": 723
|
5087 |
+
},
|
5088 |
+
{
|
5089 |
+
"epoch": 0.6284722222222222,
|
5090 |
+
"grad_norm": 17.04582405090332,
|
5091 |
+
"learning_rate": 6.166812700186389e-05,
|
5092 |
+
"loss": 1.9505,
|
5093 |
+
"step": 724
|
5094 |
+
},
|
5095 |
+
{
|
5096 |
+
"epoch": 0.6293402777777778,
|
5097 |
+
"grad_norm": 14.280027389526367,
|
5098 |
+
"learning_rate": 6.141418955979339e-05,
|
5099 |
+
"loss": 1.7364,
|
5100 |
+
"step": 725
|
5101 |
+
},
|
5102 |
+
{
|
5103 |
+
"epoch": 0.6302083333333334,
|
5104 |
+
"grad_norm": 14.48975658416748,
|
5105 |
+
"learning_rate": 6.116054412581722e-05,
|
5106 |
+
"loss": 2.0984,
|
5107 |
+
"step": 726
|
5108 |
+
},
|
5109 |
+
{
|
5110 |
+
"epoch": 0.6310763888888888,
|
5111 |
+
"grad_norm": 12.331202507019043,
|
5112 |
+
"learning_rate": 6.090719261946279e-05,
|
5113 |
+
"loss": 1.7384,
|
5114 |
+
"step": 727
|
5115 |
+
},
|
5116 |
+
{
|
5117 |
+
"epoch": 0.6319444444444444,
|
5118 |
+
"grad_norm": 20.513586044311523,
|
5119 |
+
"learning_rate": 6.065413695803316e-05,
|
5120 |
+
"loss": 1.4269,
|
5121 |
+
"step": 728
|
5122 |
+
},
|
5123 |
+
{
|
5124 |
+
"epoch": 0.6328125,
|
5125 |
+
"grad_norm": 14.325490951538086,
|
5126 |
+
"learning_rate": 6.0401379056592375e-05,
|
5127 |
+
"loss": 1.878,
|
5128 |
+
"step": 729
|
5129 |
+
},
|
5130 |
+
{
|
5131 |
+
"epoch": 0.6336805555555556,
|
5132 |
+
"grad_norm": 13.367268562316895,
|
5133 |
+
"learning_rate": 6.014892082795127e-05,
|
5134 |
+
"loss": 2.1823,
|
5135 |
+
"step": 730
|
5136 |
+
},
|
5137 |
+
{
|
5138 |
+
"epoch": 0.6345486111111112,
|
5139 |
+
"grad_norm": 17.7767276763916,
|
5140 |
+
"learning_rate": 5.9896764182652654e-05,
|
5141 |
+
"loss": 1.9079,
|
5142 |
+
"step": 731
|
5143 |
+
},
|
5144 |
+
{
|
5145 |
+
"epoch": 0.6354166666666666,
|
5146 |
+
"grad_norm": 17.653276443481445,
|
5147 |
+
"learning_rate": 5.9644911028957196e-05,
|
5148 |
+
"loss": 1.9413,
|
5149 |
+
"step": 732
|
5150 |
+
},
|
5151 |
+
{
|
5152 |
+
"epoch": 0.6362847222222222,
|
5153 |
+
"grad_norm": 15.516522407531738,
|
5154 |
+
"learning_rate": 5.939336327282874e-05,
|
5155 |
+
"loss": 1.9241,
|
5156 |
+
"step": 733
|
5157 |
+
},
|
5158 |
+
{
|
5159 |
+
"epoch": 0.6371527777777778,
|
5160 |
+
"grad_norm": 14.670297622680664,
|
5161 |
+
"learning_rate": 5.914212281791989e-05,
|
5162 |
+
"loss": 2.0385,
|
5163 |
+
"step": 734
|
5164 |
+
},
|
5165 |
+
{
|
5166 |
+
"epoch": 0.6380208333333334,
|
5167 |
+
"grad_norm": 14.384023666381836,
|
5168 |
+
"learning_rate": 5.889119156555778e-05,
|
5169 |
+
"loss": 1.9389,
|
5170 |
+
"step": 735
|
5171 |
+
},
|
5172 |
+
{
|
5173 |
+
"epoch": 0.6388888888888888,
|
5174 |
+
"grad_norm": 15.61893367767334,
|
5175 |
+
"learning_rate": 5.8640571414729486e-05,
|
5176 |
+
"loss": 1.9736,
|
5177 |
+
"step": 736
|
5178 |
+
},
|
5179 |
+
{
|
5180 |
+
"epoch": 0.6397569444444444,
|
5181 |
+
"grad_norm": 15.846280097961426,
|
5182 |
+
"learning_rate": 5.839026426206782e-05,
|
5183 |
+
"loss": 1.6474,
|
5184 |
+
"step": 737
|
5185 |
+
},
|
5186 |
+
{
|
5187 |
+
"epoch": 0.640625,
|
5188 |
+
"grad_norm": 15.824593544006348,
|
5189 |
+
"learning_rate": 5.814027200183685e-05,
|
5190 |
+
"loss": 1.8774,
|
5191 |
+
"step": 738
|
5192 |
+
},
|
5193 |
+
{
|
5194 |
+
"epoch": 0.6414930555555556,
|
5195 |
+
"grad_norm": 16.56097984313965,
|
5196 |
+
"learning_rate": 5.7890596525917593e-05,
|
5197 |
+
"loss": 1.9392,
|
5198 |
+
"step": 739
|
5199 |
+
},
|
5200 |
+
{
|
5201 |
+
"epoch": 0.6423611111111112,
|
5202 |
+
"grad_norm": 19.102895736694336,
|
5203 |
+
"learning_rate": 5.7641239723793814e-05,
|
5204 |
+
"loss": 2.1913,
|
5205 |
+
"step": 740
|
5206 |
+
},
|
5207 |
+
{
|
5208 |
+
"epoch": 0.6432291666666666,
|
5209 |
+
"grad_norm": 18.151823043823242,
|
5210 |
+
"learning_rate": 5.739220348253753e-05,
|
5211 |
+
"loss": 1.9148,
|
5212 |
+
"step": 741
|
5213 |
+
},
|
5214 |
+
{
|
5215 |
+
"epoch": 0.6440972222222222,
|
5216 |
+
"grad_norm": 16.493192672729492,
|
5217 |
+
"learning_rate": 5.714348968679483e-05,
|
5218 |
+
"loss": 1.9077,
|
5219 |
+
"step": 742
|
5220 |
+
},
|
5221 |
+
{
|
5222 |
+
"epoch": 0.6449652777777778,
|
5223 |
+
"grad_norm": 18.61972999572754,
|
5224 |
+
"learning_rate": 5.689510021877173e-05,
|
5225 |
+
"loss": 2.1327,
|
5226 |
+
"step": 743
|
5227 |
+
},
|
5228 |
+
{
|
5229 |
+
"epoch": 0.6458333333333334,
|
5230 |
+
"grad_norm": 17.50888442993164,
|
5231 |
+
"learning_rate": 5.664703695821965e-05,
|
5232 |
+
"loss": 2.013,
|
5233 |
+
"step": 744
|
5234 |
+
},
|
5235 |
+
{
|
5236 |
+
"epoch": 0.6467013888888888,
|
5237 |
+
"grad_norm": 16.999921798706055,
|
5238 |
+
"learning_rate": 5.639930178242152e-05,
|
5239 |
+
"loss": 2.0242,
|
5240 |
+
"step": 745
|
5241 |
+
},
|
5242 |
+
{
|
5243 |
+
"epoch": 0.6475694444444444,
|
5244 |
+
"grad_norm": 15.626893997192383,
|
5245 |
+
"learning_rate": 5.6151896566177275e-05,
|
5246 |
+
"loss": 2.1479,
|
5247 |
+
"step": 746
|
5248 |
+
},
|
5249 |
+
{
|
5250 |
+
"epoch": 0.6484375,
|
5251 |
+
"grad_norm": 23.982009887695312,
|
5252 |
+
"learning_rate": 5.590482318178982e-05,
|
5253 |
+
"loss": 1.8645,
|
5254 |
+
"step": 747
|
5255 |
+
},
|
5256 |
+
{
|
5257 |
+
"epoch": 0.6493055555555556,
|
5258 |
+
"grad_norm": 20.309349060058594,
|
5259 |
+
"learning_rate": 5.56580834990509e-05,
|
5260 |
+
"loss": 1.7871,
|
5261 |
+
"step": 748
|
5262 |
+
},
|
5263 |
+
{
|
5264 |
+
"epoch": 0.6501736111111112,
|
5265 |
+
"grad_norm": 20.062475204467773,
|
5266 |
+
"learning_rate": 5.54116793852268e-05,
|
5267 |
+
"loss": 1.9625,
|
5268 |
+
"step": 749
|
5269 |
+
},
|
5270 |
+
{
|
5271 |
+
"epoch": 0.6510416666666666,
|
5272 |
+
"grad_norm": 22.787874221801758,
|
5273 |
+
"learning_rate": 5.5165612705044386e-05,
|
5274 |
+
"loss": 1.9159,
|
5275 |
+
"step": 750
|
5276 |
+
},
|
5277 |
+
{
|
5278 |
+
"epoch": 0.6519097222222222,
|
5279 |
+
"grad_norm": 8.314199447631836,
|
5280 |
+
"learning_rate": 5.491988532067687e-05,
|
5281 |
+
"loss": 2.0528,
|
5282 |
+
"step": 751
|
5283 |
+
},
|
5284 |
+
{
|
5285 |
+
"epoch": 0.6527777777777778,
|
5286 |
+
"grad_norm": 8.974541664123535,
|
5287 |
+
"learning_rate": 5.467449909172966e-05,
|
5288 |
+
"loss": 1.9216,
|
5289 |
+
"step": 752
|
5290 |
+
},
|
5291 |
+
{
|
5292 |
+
"epoch": 0.6536458333333334,
|
5293 |
+
"grad_norm": 9.058680534362793,
|
5294 |
+
"learning_rate": 5.442945587522659e-05,
|
5295 |
+
"loss": 1.8281,
|
5296 |
+
"step": 753
|
5297 |
+
},
|
5298 |
+
{
|
5299 |
+
"epoch": 0.6545138888888888,
|
5300 |
+
"grad_norm": 22.46303367614746,
|
5301 |
+
"learning_rate": 5.418475752559544e-05,
|
5302 |
+
"loss": 1.8423,
|
5303 |
+
"step": 754
|
5304 |
+
},
|
5305 |
+
{
|
5306 |
+
"epoch": 0.6553819444444444,
|
5307 |
+
"grad_norm": 9.852974891662598,
|
5308 |
+
"learning_rate": 5.394040589465431e-05,
|
5309 |
+
"loss": 1.7332,
|
5310 |
+
"step": 755
|
5311 |
+
},
|
5312 |
+
{
|
5313 |
+
"epoch": 0.65625,
|
5314 |
+
"grad_norm": 11.416216850280762,
|
5315 |
+
"learning_rate": 5.369640283159729e-05,
|
5316 |
+
"loss": 1.8474,
|
5317 |
+
"step": 756
|
5318 |
+
},
|
5319 |
+
{
|
5320 |
+
"epoch": 0.6571180555555556,
|
5321 |
+
"grad_norm": 11.507524490356445,
|
5322 |
+
"learning_rate": 5.3452750182980625e-05,
|
5323 |
+
"loss": 1.9673,
|
5324 |
+
"step": 757
|
5325 |
+
},
|
5326 |
+
{
|
5327 |
+
"epoch": 0.6579861111111112,
|
5328 |
+
"grad_norm": 10.947381019592285,
|
5329 |
+
"learning_rate": 5.320944979270868e-05,
|
5330 |
+
"loss": 1.8135,
|
5331 |
+
"step": 758
|
5332 |
+
},
|
5333 |
+
{
|
5334 |
+
"epoch": 0.6588541666666666,
|
5335 |
+
"grad_norm": 10.233997344970703,
|
5336 |
+
"learning_rate": 5.296650350202008e-05,
|
5337 |
+
"loss": 2.0056,
|
5338 |
+
"step": 759
|
5339 |
+
},
|
5340 |
+
{
|
5341 |
+
"epoch": 0.6597222222222222,
|
5342 |
+
"grad_norm": 10.532794952392578,
|
5343 |
+
"learning_rate": 5.2723913149473737e-05,
|
5344 |
+
"loss": 1.831,
|
5345 |
+
"step": 760
|
5346 |
+
},
|
5347 |
+
{
|
5348 |
+
"epoch": 0.6605902777777778,
|
5349 |
+
"grad_norm": 12.286763191223145,
|
5350 |
+
"learning_rate": 5.248168057093478e-05,
|
5351 |
+
"loss": 1.9529,
|
5352 |
+
"step": 761
|
5353 |
+
},
|
5354 |
+
{
|
5355 |
+
"epoch": 0.6614583333333334,
|
5356 |
+
"grad_norm": 10.542374610900879,
|
5357 |
+
"learning_rate": 5.223980759956089e-05,
|
5358 |
+
"loss": 1.7063,
|
5359 |
+
"step": 762
|
5360 |
+
},
|
5361 |
+
{
|
5362 |
+
"epoch": 0.6623263888888888,
|
5363 |
+
"grad_norm": 12.384016036987305,
|
5364 |
+
"learning_rate": 5.199829606578823e-05,
|
5365 |
+
"loss": 1.8829,
|
5366 |
+
"step": 763
|
5367 |
+
},
|
5368 |
+
{
|
5369 |
+
"epoch": 0.6631944444444444,
|
5370 |
+
"grad_norm": 13.105708122253418,
|
5371 |
+
"learning_rate": 5.175714779731782e-05,
|
5372 |
+
"loss": 1.7577,
|
5373 |
+
"step": 764
|
5374 |
+
},
|
5375 |
+
{
|
5376 |
+
"epoch": 0.6640625,
|
5377 |
+
"grad_norm": 14.688854217529297,
|
5378 |
+
"learning_rate": 5.151636461910154e-05,
|
5379 |
+
"loss": 1.961,
|
5380 |
+
"step": 765
|
5381 |
+
},
|
5382 |
+
{
|
5383 |
+
"epoch": 0.6649305555555556,
|
5384 |
+
"grad_norm": 14.315075874328613,
|
5385 |
+
"learning_rate": 5.127594835332831e-05,
|
5386 |
+
"loss": 1.8002,
|
5387 |
+
"step": 766
|
5388 |
+
},
|
5389 |
+
{
|
5390 |
+
"epoch": 0.6657986111111112,
|
5391 |
+
"grad_norm": 13.88724136352539,
|
5392 |
+
"learning_rate": 5.1035900819410346e-05,
|
5393 |
+
"loss": 1.7724,
|
5394 |
+
"step": 767
|
5395 |
+
},
|
5396 |
+
{
|
5397 |
+
"epoch": 0.6666666666666666,
|
5398 |
+
"grad_norm": 13.456408500671387,
|
5399 |
+
"learning_rate": 5.079622383396938e-05,
|
5400 |
+
"loss": 1.8523,
|
5401 |
+
"step": 768
|
5402 |
+
},
|
5403 |
+
{
|
5404 |
+
"epoch": 0.6675347222222222,
|
5405 |
+
"grad_norm": 14.679458618164062,
|
5406 |
+
"learning_rate": 5.055691921082299e-05,
|
5407 |
+
"loss": 1.8991,
|
5408 |
+
"step": 769
|
5409 |
+
},
|
5410 |
+
{
|
5411 |
+
"epoch": 0.6684027777777778,
|
5412 |
+
"grad_norm": 12.704297065734863,
|
5413 |
+
"learning_rate": 5.0317988760970804e-05,
|
5414 |
+
"loss": 1.8636,
|
5415 |
+
"step": 770
|
5416 |
+
},
|
5417 |
+
{
|
5418 |
+
"epoch": 0.6692708333333334,
|
5419 |
+
"grad_norm": 12.467072486877441,
|
5420 |
+
"learning_rate": 5.007943429258073e-05,
|
5421 |
+
"loss": 1.8262,
|
5422 |
+
"step": 771
|
5423 |
+
},
|
5424 |
+
{
|
5425 |
+
"epoch": 0.6701388888888888,
|
5426 |
+
"grad_norm": 16.064525604248047,
|
5427 |
+
"learning_rate": 4.9841257610975364e-05,
|
5428 |
+
"loss": 1.6718,
|
5429 |
+
"step": 772
|
5430 |
+
},
|
5431 |
+
{
|
5432 |
+
"epoch": 0.6710069444444444,
|
5433 |
+
"grad_norm": 14.80754566192627,
|
5434 |
+
"learning_rate": 4.9603460518618274e-05,
|
5435 |
+
"loss": 1.8804,
|
5436 |
+
"step": 773
|
5437 |
+
},
|
5438 |
+
{
|
5439 |
+
"epoch": 0.671875,
|
5440 |
+
"grad_norm": 13.693035125732422,
|
5441 |
+
"learning_rate": 4.936604481510045e-05,
|
5442 |
+
"loss": 1.9249,
|
5443 |
+
"step": 774
|
5444 |
+
},
|
5445 |
+
{
|
5446 |
+
"epoch": 0.6727430555555556,
|
5447 |
+
"grad_norm": 14.119590759277344,
|
5448 |
+
"learning_rate": 4.912901229712665e-05,
|
5449 |
+
"loss": 1.8726,
|
5450 |
+
"step": 775
|
5451 |
+
},
|
5452 |
+
{
|
5453 |
+
"epoch": 0.6736111111111112,
|
5454 |
+
"grad_norm": 11.714713096618652,
|
5455 |
+
"learning_rate": 4.889236475850169e-05,
|
5456 |
+
"loss": 1.9406,
|
5457 |
+
"step": 776
|
5458 |
+
},
|
5459 |
+
{
|
5460 |
+
"epoch": 0.6744791666666666,
|
5461 |
+
"grad_norm": 13.115860939025879,
|
5462 |
+
"learning_rate": 4.865610399011696e-05,
|
5463 |
+
"loss": 1.9581,
|
5464 |
+
"step": 777
|
5465 |
+
},
|
5466 |
+
{
|
5467 |
+
"epoch": 0.6753472222222222,
|
5468 |
+
"grad_norm": 14.199614524841309,
|
5469 |
+
"learning_rate": 4.8420231779936885e-05,
|
5470 |
+
"loss": 1.964,
|
5471 |
+
"step": 778
|
5472 |
+
},
|
5473 |
+
{
|
5474 |
+
"epoch": 0.6762152777777778,
|
5475 |
+
"grad_norm": 14.712546348571777,
|
5476 |
+
"learning_rate": 4.818474991298541e-05,
|
5477 |
+
"loss": 2.0005,
|
5478 |
+
"step": 779
|
5479 |
+
},
|
5480 |
+
{
|
5481 |
+
"epoch": 0.6770833333333334,
|
5482 |
+
"grad_norm": 14.282362937927246,
|
5483 |
+
"learning_rate": 4.7949660171332475e-05,
|
5484 |
+
"loss": 1.8183,
|
5485 |
+
"step": 780
|
5486 |
+
},
|
5487 |
+
{
|
5488 |
+
"epoch": 0.6779513888888888,
|
5489 |
+
"grad_norm": 20.02650260925293,
|
5490 |
+
"learning_rate": 4.771496433408043e-05,
|
5491 |
+
"loss": 2.0001,
|
5492 |
+
"step": 781
|
5493 |
+
},
|
5494 |
+
{
|
5495 |
+
"epoch": 0.6788194444444444,
|
5496 |
+
"grad_norm": 16.86726188659668,
|
5497 |
+
"learning_rate": 4.748066417735068e-05,
|
5498 |
+
"loss": 1.8494,
|
5499 |
+
"step": 782
|
5500 |
+
},
|
5501 |
+
{
|
5502 |
+
"epoch": 0.6796875,
|
5503 |
+
"grad_norm": 16.731599807739258,
|
5504 |
+
"learning_rate": 4.724676147427018e-05,
|
5505 |
+
"loss": 2.0999,
|
5506 |
+
"step": 783
|
5507 |
+
},
|
5508 |
+
{
|
5509 |
+
"epoch": 0.6805555555555556,
|
5510 |
+
"grad_norm": 12.958636283874512,
|
5511 |
+
"learning_rate": 4.701325799495817e-05,
|
5512 |
+
"loss": 1.9338,
|
5513 |
+
"step": 784
|
5514 |
+
},
|
5515 |
+
{
|
5516 |
+
"epoch": 0.6814236111111112,
|
5517 |
+
"grad_norm": 16.30995750427246,
|
5518 |
+
"learning_rate": 4.678015550651258e-05,
|
5519 |
+
"loss": 1.934,
|
5520 |
+
"step": 785
|
5521 |
+
},
|
5522 |
+
{
|
5523 |
+
"epoch": 0.6822916666666666,
|
5524 |
+
"grad_norm": 15.418025970458984,
|
5525 |
+
"learning_rate": 4.654745577299674e-05,
|
5526 |
+
"loss": 1.6663,
|
5527 |
+
"step": 786
|
5528 |
+
},
|
5529 |
+
{
|
5530 |
+
"epoch": 0.6831597222222222,
|
5531 |
+
"grad_norm": 13.731332778930664,
|
5532 |
+
"learning_rate": 4.631516055542601e-05,
|
5533 |
+
"loss": 1.8579,
|
5534 |
+
"step": 787
|
5535 |
+
},
|
5536 |
+
{
|
5537 |
+
"epoch": 0.6840277777777778,
|
5538 |
+
"grad_norm": 12.98118782043457,
|
5539 |
+
"learning_rate": 4.6083271611754476e-05,
|
5540 |
+
"loss": 2.1614,
|
5541 |
+
"step": 788
|
5542 |
+
},
|
5543 |
+
{
|
5544 |
+
"epoch": 0.6848958333333334,
|
5545 |
+
"grad_norm": 12.849787712097168,
|
5546 |
+
"learning_rate": 4.585179069686171e-05,
|
5547 |
+
"loss": 1.8389,
|
5548 |
+
"step": 789
|
5549 |
+
},
|
5550 |
+
{
|
5551 |
+
"epoch": 0.6857638888888888,
|
5552 |
+
"grad_norm": 15.817699432373047,
|
5553 |
+
"learning_rate": 4.5620719562539314e-05,
|
5554 |
+
"loss": 1.9226,
|
5555 |
+
"step": 790
|
5556 |
+
},
|
5557 |
+
{
|
5558 |
+
"epoch": 0.6866319444444444,
|
5559 |
+
"grad_norm": 14.717578887939453,
|
5560 |
+
"learning_rate": 4.5390059957477904e-05,
|
5561 |
+
"loss": 1.8674,
|
5562 |
+
"step": 791
|
5563 |
+
},
|
5564 |
+
{
|
5565 |
+
"epoch": 0.6875,
|
5566 |
+
"grad_norm": 17.49150276184082,
|
5567 |
+
"learning_rate": 4.515981362725361e-05,
|
5568 |
+
"loss": 1.8261,
|
5569 |
+
"step": 792
|
5570 |
+
},
|
5571 |
+
{
|
5572 |
+
"epoch": 0.6883680555555556,
|
5573 |
+
"grad_norm": 19.413713455200195,
|
5574 |
+
"learning_rate": 4.4929982314315024e-05,
|
5575 |
+
"loss": 1.7202,
|
5576 |
+
"step": 793
|
5577 |
+
},
|
5578 |
+
{
|
5579 |
+
"epoch": 0.6892361111111112,
|
5580 |
+
"grad_norm": 19.134525299072266,
|
5581 |
+
"learning_rate": 4.470056775797008e-05,
|
5582 |
+
"loss": 1.9247,
|
5583 |
+
"step": 794
|
5584 |
+
},
|
5585 |
+
{
|
5586 |
+
"epoch": 0.6901041666666666,
|
5587 |
+
"grad_norm": 16.60801124572754,
|
5588 |
+
"learning_rate": 4.447157169437267e-05,
|
5589 |
+
"loss": 2.0034,
|
5590 |
+
"step": 795
|
5591 |
+
},
|
5592 |
+
{
|
5593 |
+
"epoch": 0.6909722222222222,
|
5594 |
+
"grad_norm": 16.824369430541992,
|
5595 |
+
"learning_rate": 4.424299585650977e-05,
|
5596 |
+
"loss": 1.9362,
|
5597 |
+
"step": 796
|
5598 |
+
},
|
5599 |
+
{
|
5600 |
+
"epoch": 0.6918402777777778,
|
5601 |
+
"grad_norm": 17.961692810058594,
|
5602 |
+
"learning_rate": 4.401484197418807e-05,
|
5603 |
+
"loss": 1.8643,
|
5604 |
+
"step": 797
|
5605 |
+
},
|
5606 |
+
{
|
5607 |
+
"epoch": 0.6927083333333334,
|
5608 |
+
"grad_norm": 21.15089988708496,
|
5609 |
+
"learning_rate": 4.3787111774021014e-05,
|
5610 |
+
"loss": 1.9506,
|
5611 |
+
"step": 798
|
5612 |
+
},
|
5613 |
+
{
|
5614 |
+
"epoch": 0.6935763888888888,
|
5615 |
+
"grad_norm": 24.73053550720215,
|
5616 |
+
"learning_rate": 4.355980697941584e-05,
|
5617 |
+
"loss": 2.0877,
|
5618 |
+
"step": 799
|
5619 |
+
},
|
5620 |
+
{
|
5621 |
+
"epoch": 0.6944444444444444,
|
5622 |
+
"grad_norm": 34.46087646484375,
|
5623 |
+
"learning_rate": 4.3332929310560265e-05,
|
5624 |
+
"loss": 1.8522,
|
5625 |
+
"step": 800
|
5626 |
+
},
|
5627 |
+
{
|
5628 |
+
"epoch": 0.6953125,
|
5629 |
+
"grad_norm": 8.951723098754883,
|
5630 |
+
"learning_rate": 4.310648048440975e-05,
|
5631 |
+
"loss": 1.9765,
|
5632 |
+
"step": 801
|
5633 |
+
},
|
5634 |
+
{
|
5635 |
+
"epoch": 0.6961805555555556,
|
5636 |
+
"grad_norm": 8.993708610534668,
|
5637 |
+
"learning_rate": 4.2880462214674287e-05,
|
5638 |
+
"loss": 1.8379,
|
5639 |
+
"step": 802
|
5640 |
+
},
|
5641 |
+
{
|
5642 |
+
"epoch": 0.6970486111111112,
|
5643 |
+
"grad_norm": 9.19201946258545,
|
5644 |
+
"learning_rate": 4.2654876211805527e-05,
|
5645 |
+
"loss": 1.7827,
|
5646 |
+
"step": 803
|
5647 |
+
},
|
5648 |
+
{
|
5649 |
+
"epoch": 0.6979166666666666,
|
5650 |
+
"grad_norm": 9.752361297607422,
|
5651 |
+
"learning_rate": 4.2429724182983935e-05,
|
5652 |
+
"loss": 1.7326,
|
5653 |
+
"step": 804
|
5654 |
+
},
|
5655 |
+
{
|
5656 |
+
"epoch": 0.6987847222222222,
|
5657 |
+
"grad_norm": 11.15304183959961,
|
5658 |
+
"learning_rate": 4.220500783210559e-05,
|
5659 |
+
"loss": 2.0252,
|
5660 |
+
"step": 805
|
5661 |
+
},
|
5662 |
+
{
|
5663 |
+
"epoch": 0.6996527777777778,
|
5664 |
+
"grad_norm": 11.194067001342773,
|
5665 |
+
"learning_rate": 4.198072885976966e-05,
|
5666 |
+
"loss": 1.9484,
|
5667 |
+
"step": 806
|
5668 |
+
},
|
5669 |
+
{
|
5670 |
+
"epoch": 0.7005208333333334,
|
5671 |
+
"grad_norm": 9.688783645629883,
|
5672 |
+
"learning_rate": 4.1756888963265194e-05,
|
5673 |
+
"loss": 1.9918,
|
5674 |
+
"step": 807
|
5675 |
+
},
|
5676 |
+
{
|
5677 |
+
"epoch": 0.7013888888888888,
|
5678 |
+
"grad_norm": 11.591280937194824,
|
5679 |
+
"learning_rate": 4.153348983655844e-05,
|
5680 |
+
"loss": 1.806,
|
5681 |
+
"step": 808
|
5682 |
+
},
|
5683 |
+
{
|
5684 |
+
"epoch": 0.7022569444444444,
|
5685 |
+
"grad_norm": 11.173543930053711,
|
5686 |
+
"learning_rate": 4.1310533170280084e-05,
|
5687 |
+
"loss": 1.8658,
|
5688 |
+
"step": 809
|
5689 |
+
},
|
5690 |
+
{
|
5691 |
+
"epoch": 0.703125,
|
5692 |
+
"grad_norm": 14.811725616455078,
|
5693 |
+
"learning_rate": 4.108802065171229e-05,
|
5694 |
+
"loss": 1.8853,
|
5695 |
+
"step": 810
|
5696 |
+
},
|
5697 |
+
{
|
5698 |
+
"epoch": 0.7039930555555556,
|
5699 |
+
"grad_norm": 15.026256561279297,
|
5700 |
+
"learning_rate": 4.086595396477611e-05,
|
5701 |
+
"loss": 2.0679,
|
5702 |
+
"step": 811
|
5703 |
+
},
|
5704 |
+
{
|
5705 |
+
"epoch": 0.7048611111111112,
|
5706 |
+
"grad_norm": 10.38699722290039,
|
5707 |
+
"learning_rate": 4.064433479001857e-05,
|
5708 |
+
"loss": 1.9456,
|
5709 |
+
"step": 812
|
5710 |
+
},
|
5711 |
+
{
|
5712 |
+
"epoch": 0.7057291666666666,
|
5713 |
+
"grad_norm": 13.468440055847168,
|
5714 |
+
"learning_rate": 4.042316480460001e-05,
|
5715 |
+
"loss": 1.9493,
|
5716 |
+
"step": 813
|
5717 |
+
},
|
5718 |
+
{
|
5719 |
+
"epoch": 0.7065972222222222,
|
5720 |
+
"grad_norm": 15.555776596069336,
|
5721 |
+
"learning_rate": 4.020244568228155e-05,
|
5722 |
+
"loss": 1.8852,
|
5723 |
+
"step": 814
|
5724 |
+
},
|
5725 |
+
{
|
5726 |
+
"epoch": 0.7074652777777778,
|
5727 |
+
"grad_norm": 11.897953987121582,
|
5728 |
+
"learning_rate": 3.998217909341213e-05,
|
5729 |
+
"loss": 2.0327,
|
5730 |
+
"step": 815
|
5731 |
+
},
|
5732 |
+
{
|
5733 |
+
"epoch": 0.7083333333333334,
|
5734 |
+
"grad_norm": 12.633040428161621,
|
5735 |
+
"learning_rate": 3.9762366704916165e-05,
|
5736 |
+
"loss": 1.709,
|
5737 |
+
"step": 816
|
5738 |
+
},
|
5739 |
+
{
|
5740 |
+
"epoch": 0.7092013888888888,
|
5741 |
+
"grad_norm": 12.071878433227539,
|
5742 |
+
"learning_rate": 3.954301018028068e-05,
|
5743 |
+
"loss": 1.7543,
|
5744 |
+
"step": 817
|
5745 |
+
},
|
5746 |
+
{
|
5747 |
+
"epoch": 0.7100694444444444,
|
5748 |
+
"grad_norm": 12.684844017028809,
|
5749 |
+
"learning_rate": 3.932411117954286e-05,
|
5750 |
+
"loss": 1.7642,
|
5751 |
+
"step": 818
|
5752 |
+
},
|
5753 |
+
{
|
5754 |
+
"epoch": 0.7109375,
|
5755 |
+
"grad_norm": 11.684479713439941,
|
5756 |
+
"learning_rate": 3.9105671359277517e-05,
|
5757 |
+
"loss": 2.0108,
|
5758 |
+
"step": 819
|
5759 |
+
},
|
5760 |
+
{
|
5761 |
+
"epoch": 0.7118055555555556,
|
5762 |
+
"grad_norm": 22.061206817626953,
|
5763 |
+
"learning_rate": 3.8887692372584405e-05,
|
5764 |
+
"loss": 1.6975,
|
5765 |
+
"step": 820
|
5766 |
+
},
|
5767 |
+
{
|
5768 |
+
"epoch": 0.7126736111111112,
|
5769 |
+
"grad_norm": 13.567614555358887,
|
5770 |
+
"learning_rate": 3.867017586907591e-05,
|
5771 |
+
"loss": 1.9444,
|
5772 |
+
"step": 821
|
5773 |
+
},
|
5774 |
+
{
|
5775 |
+
"epoch": 0.7135416666666666,
|
5776 |
+
"grad_norm": 13.34210205078125,
|
5777 |
+
"learning_rate": 3.845312349486437e-05,
|
5778 |
+
"loss": 1.619,
|
5779 |
+
"step": 822
|
5780 |
+
},
|
5781 |
+
{
|
5782 |
+
"epoch": 0.7144097222222222,
|
5783 |
+
"grad_norm": 12.922858238220215,
|
5784 |
+
"learning_rate": 3.82365368925497e-05,
|
5785 |
+
"loss": 1.9284,
|
5786 |
+
"step": 823
|
5787 |
+
},
|
5788 |
+
{
|
5789 |
+
"epoch": 0.7152777777777778,
|
5790 |
+
"grad_norm": 12.862595558166504,
|
5791 |
+
"learning_rate": 3.8020417701207056e-05,
|
5792 |
+
"loss": 1.9332,
|
5793 |
+
"step": 824
|
5794 |
+
},
|
5795 |
+
{
|
5796 |
+
"epoch": 0.7161458333333334,
|
5797 |
+
"grad_norm": 14.289531707763672,
|
5798 |
+
"learning_rate": 3.780476755637421e-05,
|
5799 |
+
"loss": 1.7273,
|
5800 |
+
"step": 825
|
5801 |
+
},
|
5802 |
+
{
|
5803 |
+
"epoch": 0.7170138888888888,
|
5804 |
+
"grad_norm": 12.516698837280273,
|
5805 |
+
"learning_rate": 3.758958809003945e-05,
|
5806 |
+
"loss": 1.9387,
|
5807 |
+
"step": 826
|
5808 |
+
},
|
5809 |
+
{
|
5810 |
+
"epoch": 0.7178819444444444,
|
5811 |
+
"grad_norm": 14.04725456237793,
|
5812 |
+
"learning_rate": 3.7374880930628986e-05,
|
5813 |
+
"loss": 1.824,
|
5814 |
+
"step": 827
|
5815 |
+
},
|
5816 |
+
{
|
5817 |
+
"epoch": 0.71875,
|
5818 |
+
"grad_norm": 13.029735565185547,
|
5819 |
+
"learning_rate": 3.71606477029947e-05,
|
5820 |
+
"loss": 2.0721,
|
5821 |
+
"step": 828
|
5822 |
+
},
|
5823 |
+
{
|
5824 |
+
"epoch": 0.7196180555555556,
|
5825 |
+
"grad_norm": 16.414342880249023,
|
5826 |
+
"learning_rate": 3.694689002840198e-05,
|
5827 |
+
"loss": 2.0021,
|
5828 |
+
"step": 829
|
5829 |
+
},
|
5830 |
+
{
|
5831 |
+
"epoch": 0.7204861111111112,
|
5832 |
+
"grad_norm": 15.33515739440918,
|
5833 |
+
"learning_rate": 3.673360952451722e-05,
|
5834 |
+
"loss": 2.2091,
|
5835 |
+
"step": 830
|
5836 |
+
},
|
5837 |
+
{
|
5838 |
+
"epoch": 0.7213541666666666,
|
5839 |
+
"grad_norm": 13.48131275177002,
|
5840 |
+
"learning_rate": 3.652080780539584e-05,
|
5841 |
+
"loss": 1.8521,
|
5842 |
+
"step": 831
|
5843 |
+
},
|
5844 |
+
{
|
5845 |
+
"epoch": 0.7222222222222222,
|
5846 |
+
"grad_norm": 13.361886024475098,
|
5847 |
+
"learning_rate": 3.6308486481469805e-05,
|
5848 |
+
"loss": 2.0773,
|
5849 |
+
"step": 832
|
5850 |
+
},
|
5851 |
+
{
|
5852 |
+
"epoch": 0.7230902777777778,
|
5853 |
+
"grad_norm": 11.877111434936523,
|
5854 |
+
"learning_rate": 3.609664715953559e-05,
|
5855 |
+
"loss": 2.1115,
|
5856 |
+
"step": 833
|
5857 |
+
},
|
5858 |
+
{
|
5859 |
+
"epoch": 0.7239583333333334,
|
5860 |
+
"grad_norm": 12.948386192321777,
|
5861 |
+
"learning_rate": 3.5885291442742084e-05,
|
5862 |
+
"loss": 1.9481,
|
5863 |
+
"step": 834
|
5864 |
+
},
|
5865 |
+
{
|
5866 |
+
"epoch": 0.7248263888888888,
|
5867 |
+
"grad_norm": 16.159982681274414,
|
5868 |
+
"learning_rate": 3.567442093057825e-05,
|
5869 |
+
"loss": 1.9765,
|
5870 |
+
"step": 835
|
5871 |
+
},
|
5872 |
+
{
|
5873 |
+
"epoch": 0.7256944444444444,
|
5874 |
+
"grad_norm": 14.263628005981445,
|
5875 |
+
"learning_rate": 3.546403721886126e-05,
|
5876 |
+
"loss": 2.0117,
|
5877 |
+
"step": 836
|
5878 |
+
},
|
5879 |
+
{
|
5880 |
+
"epoch": 0.7265625,
|
5881 |
+
"grad_norm": 16.48017120361328,
|
5882 |
+
"learning_rate": 3.5254141899724194e-05,
|
5883 |
+
"loss": 1.8066,
|
5884 |
+
"step": 837
|
5885 |
+
},
|
5886 |
+
{
|
5887 |
+
"epoch": 0.7274305555555556,
|
5888 |
+
"grad_norm": 13.726114273071289,
|
5889 |
+
"learning_rate": 3.50447365616041e-05,
|
5890 |
+
"loss": 2.1094,
|
5891 |
+
"step": 838
|
5892 |
+
},
|
5893 |
+
{
|
5894 |
+
"epoch": 0.7282986111111112,
|
5895 |
+
"grad_norm": 14.276023864746094,
|
5896 |
+
"learning_rate": 3.483582278923007e-05,
|
5897 |
+
"loss": 1.8664,
|
5898 |
+
"step": 839
|
5899 |
+
},
|
5900 |
+
{
|
5901 |
+
"epoch": 0.7291666666666666,
|
5902 |
+
"grad_norm": 18.12653923034668,
|
5903 |
+
"learning_rate": 3.462740216361101e-05,
|
5904 |
+
"loss": 1.8786,
|
5905 |
+
"step": 840
|
5906 |
+
},
|
5907 |
+
{
|
5908 |
+
"epoch": 0.7300347222222222,
|
5909 |
+
"grad_norm": 21.137271881103516,
|
5910 |
+
"learning_rate": 3.4419476262023956e-05,
|
5911 |
+
"loss": 1.8441,
|
5912 |
+
"step": 841
|
5913 |
+
},
|
5914 |
+
{
|
5915 |
+
"epoch": 0.7309027777777778,
|
5916 |
+
"grad_norm": 17.655916213989258,
|
5917 |
+
"learning_rate": 3.421204665800186e-05,
|
5918 |
+
"loss": 1.8011,
|
5919 |
+
"step": 842
|
5920 |
+
},
|
5921 |
+
{
|
5922 |
+
"epoch": 0.7317708333333334,
|
5923 |
+
"grad_norm": 18.322494506835938,
|
5924 |
+
"learning_rate": 3.400511492132186e-05,
|
5925 |
+
"loss": 1.8712,
|
5926 |
+
"step": 843
|
5927 |
+
},
|
5928 |
+
{
|
5929 |
+
"epoch": 0.7326388888888888,
|
5930 |
+
"grad_norm": 17.434764862060547,
|
5931 |
+
"learning_rate": 3.3798682617993405e-05,
|
5932 |
+
"loss": 2.154,
|
5933 |
+
"step": 844
|
5934 |
+
},
|
5935 |
+
{
|
5936 |
+
"epoch": 0.7335069444444444,
|
5937 |
+
"grad_norm": 19.5579891204834,
|
5938 |
+
"learning_rate": 3.359275131024625e-05,
|
5939 |
+
"loss": 1.9943,
|
5940 |
+
"step": 845
|
5941 |
+
},
|
5942 |
+
{
|
5943 |
+
"epoch": 0.734375,
|
5944 |
+
"grad_norm": 17.29096794128418,
|
5945 |
+
"learning_rate": 3.338732255651887e-05,
|
5946 |
+
"loss": 1.9581,
|
5947 |
+
"step": 846
|
5948 |
+
},
|
5949 |
+
{
|
5950 |
+
"epoch": 0.7352430555555556,
|
5951 |
+
"grad_norm": 18.496713638305664,
|
5952 |
+
"learning_rate": 3.3182397911446436e-05,
|
5953 |
+
"loss": 2.2199,
|
5954 |
+
"step": 847
|
5955 |
+
},
|
5956 |
+
{
|
5957 |
+
"epoch": 0.7361111111111112,
|
5958 |
+
"grad_norm": 18.556528091430664,
|
5959 |
+
"learning_rate": 3.297797892584912e-05,
|
5960 |
+
"loss": 1.9854,
|
5961 |
+
"step": 848
|
5962 |
+
},
|
5963 |
+
{
|
5964 |
+
"epoch": 0.7369791666666666,
|
5965 |
+
"grad_norm": 25.773399353027344,
|
5966 |
+
"learning_rate": 3.277406714672049e-05,
|
5967 |
+
"loss": 1.9984,
|
5968 |
+
"step": 849
|
5969 |
+
},
|
5970 |
+
{
|
5971 |
+
"epoch": 0.7378472222222222,
|
5972 |
+
"grad_norm": 25.727313995361328,
|
5973 |
+
"learning_rate": 3.25706641172156e-05,
|
5974 |
+
"loss": 2.1087,
|
5975 |
+
"step": 850
|
5976 |
+
},
|
5977 |
+
{
|
5978 |
+
"epoch": 0.7387152777777778,
|
5979 |
+
"grad_norm": 7.924264907836914,
|
5980 |
+
"learning_rate": 3.236777137663951e-05,
|
5981 |
+
"loss": 1.934,
|
5982 |
+
"step": 851
|
5983 |
+
},
|
5984 |
+
{
|
5985 |
+
"epoch": 0.7395833333333334,
|
5986 |
+
"grad_norm": 9.656967163085938,
|
5987 |
+
"learning_rate": 3.216539046043544e-05,
|
5988 |
+
"loss": 1.9493,
|
5989 |
+
"step": 852
|
5990 |
+
},
|
5991 |
+
{
|
5992 |
+
"epoch": 0.7404513888888888,
|
5993 |
+
"grad_norm": 11.180512428283691,
|
5994 |
+
"learning_rate": 3.1963522900173294e-05,
|
5995 |
+
"loss": 1.757,
|
5996 |
+
"step": 853
|
5997 |
+
},
|
5998 |
+
{
|
5999 |
+
"epoch": 0.7413194444444444,
|
6000 |
+
"grad_norm": 10.119717597961426,
|
6001 |
+
"learning_rate": 3.176217022353798e-05,
|
6002 |
+
"loss": 1.8852,
|
6003 |
+
"step": 854
|
6004 |
+
},
|
6005 |
+
{
|
6006 |
+
"epoch": 0.7421875,
|
6007 |
+
"grad_norm": 9.676684379577637,
|
6008 |
+
"learning_rate": 3.156133395431795e-05,
|
6009 |
+
"loss": 1.9624,
|
6010 |
+
"step": 855
|
6011 |
+
},
|
6012 |
+
{
|
6013 |
+
"epoch": 0.7430555555555556,
|
6014 |
+
"grad_norm": 11.20287799835205,
|
6015 |
+
"learning_rate": 3.136101561239363e-05,
|
6016 |
+
"loss": 1.9457,
|
6017 |
+
"step": 856
|
6018 |
+
},
|
6019 |
+
{
|
6020 |
+
"epoch": 0.7439236111111112,
|
6021 |
+
"grad_norm": 11.181528091430664,
|
6022 |
+
"learning_rate": 3.116121671372582e-05,
|
6023 |
+
"loss": 1.7557,
|
6024 |
+
"step": 857
|
6025 |
+
},
|
6026 |
+
{
|
6027 |
+
"epoch": 0.7447916666666666,
|
6028 |
+
"grad_norm": 10.499610900878906,
|
6029 |
+
"learning_rate": 3.09619387703443e-05,
|
6030 |
+
"loss": 1.9088,
|
6031 |
+
"step": 858
|
6032 |
+
},
|
6033 |
+
{
|
6034 |
+
"epoch": 0.7456597222222222,
|
6035 |
+
"grad_norm": 11.516838073730469,
|
6036 |
+
"learning_rate": 3.076318329033643e-05,
|
6037 |
+
"loss": 1.8707,
|
6038 |
+
"step": 859
|
6039 |
+
},
|
6040 |
+
{
|
6041 |
+
"epoch": 0.7465277777777778,
|
6042 |
+
"grad_norm": 10.864989280700684,
|
6043 |
+
"learning_rate": 3.056495177783567e-05,
|
6044 |
+
"loss": 1.8237,
|
6045 |
+
"step": 860
|
6046 |
+
},
|
6047 |
+
{
|
6048 |
+
"epoch": 0.7473958333333334,
|
6049 |
+
"grad_norm": 11.022418022155762,
|
6050 |
+
"learning_rate": 3.0367245733010298e-05,
|
6051 |
+
"loss": 1.8337,
|
6052 |
+
"step": 861
|
6053 |
+
},
|
6054 |
+
{
|
6055 |
+
"epoch": 0.7482638888888888,
|
6056 |
+
"grad_norm": 11.150386810302734,
|
6057 |
+
"learning_rate": 3.0170066652051866e-05,
|
6058 |
+
"loss": 1.7956,
|
6059 |
+
"step": 862
|
6060 |
+
},
|
6061 |
+
{
|
6062 |
+
"epoch": 0.7491319444444444,
|
6063 |
+
"grad_norm": 13.350703239440918,
|
6064 |
+
"learning_rate": 2.9973416027164025e-05,
|
6065 |
+
"loss": 2.0799,
|
6066 |
+
"step": 863
|
6067 |
+
},
|
6068 |
+
{
|
6069 |
+
"epoch": 0.75,
|
6070 |
+
"grad_norm": 11.85857105255127,
|
6071 |
+
"learning_rate": 2.9777295346551194e-05,
|
6072 |
+
"loss": 1.8546,
|
6073 |
+
"step": 864
|
6074 |
+
},
|
6075 |
+
{
|
6076 |
+
"epoch": 0.75,
|
6077 |
+
"eval_loss": 1.9340541362762451,
|
6078 |
+
"eval_runtime": 27.3912,
|
6079 |
+
"eval_samples_per_second": 17.706,
|
6080 |
+
"eval_steps_per_second": 8.871,
|
6081 |
+
"step": 864
|
6082 |
}
|
6083 |
],
|
6084 |
"logging_steps": 1,
|
|
|
6098 |
"attributes": {}
|
6099 |
}
|
6100 |
},
|
6101 |
+
"total_flos": 3.0286133901184205e+17,
|
6102 |
"train_batch_size": 2,
|
6103 |
"trial_name": null,
|
6104 |
"trial_params": null
|