Commit
•
f58e998
1
Parent(s):
248b17c
Training in progress, step 55200, checkpoint
Browse files
last-checkpoint/model.safetensors
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 1262098580
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:c283b71e3dd9ebf1ca4e2578783ea552c8077814dd41a8c2a79a0b83ef9b71d8
|
| 3 |
size 1262098580
|
last-checkpoint/optimizer.pt
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 2490741942
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:098c6c3619389083be8e848683650ab63514cfc3e83e5466048ee7c2a9b129a4
|
| 3 |
size 2490741942
|
last-checkpoint/rng_state.pth
CHANGED
|
@@ -1,3 +1,3 @@
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
-
oid sha256:
|
| 3 |
size 14308
|
|
|
|
| 1 |
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:86cce5d4de4aedd127c08994339b8ee2f3e17d47e64c458546a64dca50dcb07e
|
| 3 |
size 14308
|
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 sha256:7a625e965fb376fc9b1c14331a8da015b0c8455084266d6cb80772f10f489d1c
|
| 3 |
size 1064
|
last-checkpoint/trainer_state.json
CHANGED
|
@@ -1,9 +1,9 @@
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.2541787922382355,
|
| 3 |
"best_model_checkpoint": "test-wav2vec2-large-xls-r-300m-nepali-openslr/checkpoint-34400",
|
| 4 |
-
"epoch":
|
| 5 |
"eval_steps": 800,
|
| 6 |
-
"global_step":
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
@@ -2867,6 +2867,1631 @@
|
|
| 2867 |
"eval_steps_per_second": 5.356,
|
| 2868 |
"eval_wer": 0.29589356196251665,
|
| 2869 |
"step": 35200
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 2870 |
}
|
| 2871 |
],
|
| 2872 |
"logging_steps": 100,
|
|
@@ -2886,7 +4511,7 @@
|
|
| 2886 |
"attributes": {}
|
| 2887 |
}
|
| 2888 |
},
|
| 2889 |
-
"total_flos": 1.
|
| 2890 |
"train_batch_size": 16,
|
| 2891 |
"trial_name": null,
|
| 2892 |
"trial_params": null
|
|
|
|
| 1 |
{
|
| 2 |
"best_metric": 0.2541787922382355,
|
| 3 |
"best_model_checkpoint": "test-wav2vec2-large-xls-r-300m-nepali-openslr/checkpoint-34400",
|
| 4 |
+
"epoch": 16.224834680382074,
|
| 5 |
"eval_steps": 800,
|
| 6 |
+
"global_step": 55200,
|
| 7 |
"is_hyper_param_search": false,
|
| 8 |
"is_local_process_zero": true,
|
| 9 |
"is_world_process_zero": true,
|
|
|
|
| 2867 |
"eval_steps_per_second": 5.356,
|
| 2868 |
"eval_wer": 0.29589356196251665,
|
| 2869 |
"step": 35200
|
| 2870 |
+
},
|
| 2871 |
+
{
|
| 2872 |
+
"epoch": 10.376193975018369,
|
| 2873 |
+
"grad_norm": 1.7660410404205322,
|
| 2874 |
+
"learning_rate": 0.00019725384009452537,
|
| 2875 |
+
"loss": 0.2465,
|
| 2876 |
+
"step": 35300
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 10.40558412931668,
|
| 2880 |
+
"grad_norm": 0.5960471034049988,
|
| 2881 |
+
"learning_rate": 0.00019695844820795586,
|
| 2882 |
+
"loss": 0.2459,
|
| 2883 |
+
"step": 35400
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 10.43497428361499,
|
| 2887 |
+
"grad_norm": 0.6564743518829346,
|
| 2888 |
+
"learning_rate": 0.00019666601024025205,
|
| 2889 |
+
"loss": 0.2527,
|
| 2890 |
+
"step": 35500
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 10.464364437913298,
|
| 2894 |
+
"grad_norm": 0.4739319384098053,
|
| 2895 |
+
"learning_rate": 0.00019637061835368254,
|
| 2896 |
+
"loss": 0.2511,
|
| 2897 |
+
"step": 35600
|
| 2898 |
+
},
|
| 2899 |
+
{
|
| 2900 |
+
"epoch": 10.493754592211609,
|
| 2901 |
+
"grad_norm": 2.3556315898895264,
|
| 2902 |
+
"learning_rate": 0.00019607522646711303,
|
| 2903 |
+
"loss": 0.2523,
|
| 2904 |
+
"step": 35700
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 10.523144746509919,
|
| 2908 |
+
"grad_norm": 1.1853305101394653,
|
| 2909 |
+
"learning_rate": 0.0001957798345805435,
|
| 2910 |
+
"loss": 0.2549,
|
| 2911 |
+
"step": 35800
|
| 2912 |
+
},
|
| 2913 |
+
{
|
| 2914 |
+
"epoch": 10.55253490080823,
|
| 2915 |
+
"grad_norm": 1.668135643005371,
|
| 2916 |
+
"learning_rate": 0.00019548444269397398,
|
| 2917 |
+
"loss": 0.256,
|
| 2918 |
+
"step": 35900
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 10.58192505510654,
|
| 2922 |
+
"grad_norm": 2.2829437255859375,
|
| 2923 |
+
"learning_rate": 0.00019518905080740447,
|
| 2924 |
+
"loss": 0.2482,
|
| 2925 |
+
"step": 36000
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 10.58192505510654,
|
| 2929 |
+
"eval_loss": 0.2800774574279785,
|
| 2930 |
+
"eval_runtime": 338.1888,
|
| 2931 |
+
"eval_samples_per_second": 40.241,
|
| 2932 |
+
"eval_steps_per_second": 5.033,
|
| 2933 |
+
"eval_wer": 0.3015657265337062,
|
| 2934 |
+
"step": 36000
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 10.61131520940485,
|
| 2938 |
+
"grad_norm": 1.2068873643875122,
|
| 2939 |
+
"learning_rate": 0.00019489365892083496,
|
| 2940 |
+
"loss": 0.2584,
|
| 2941 |
+
"step": 36100
|
| 2942 |
+
},
|
| 2943 |
+
{
|
| 2944 |
+
"epoch": 10.640705363703159,
|
| 2945 |
+
"grad_norm": 6.462821006774902,
|
| 2946 |
+
"learning_rate": 0.00019459826703426545,
|
| 2947 |
+
"loss": 0.2597,
|
| 2948 |
+
"step": 36200
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 10.67009551800147,
|
| 2952 |
+
"grad_norm": 0.7631235122680664,
|
| 2953 |
+
"learning_rate": 0.0001943028751476959,
|
| 2954 |
+
"loss": 0.25,
|
| 2955 |
+
"step": 36300
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 10.69948567229978,
|
| 2959 |
+
"grad_norm": 2.4132533073425293,
|
| 2960 |
+
"learning_rate": 0.0001940074832611264,
|
| 2961 |
+
"loss": 0.2409,
|
| 2962 |
+
"step": 36400
|
| 2963 |
+
},
|
| 2964 |
+
{
|
| 2965 |
+
"epoch": 10.72887582659809,
|
| 2966 |
+
"grad_norm": 0.5647678375244141,
|
| 2967 |
+
"learning_rate": 0.0001937120913745569,
|
| 2968 |
+
"loss": 0.2538,
|
| 2969 |
+
"step": 36500
|
| 2970 |
+
},
|
| 2971 |
+
{
|
| 2972 |
+
"epoch": 10.7582659808964,
|
| 2973 |
+
"grad_norm": 0.5179501175880432,
|
| 2974 |
+
"learning_rate": 0.00019341669948798738,
|
| 2975 |
+
"loss": 0.2507,
|
| 2976 |
+
"step": 36600
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 10.787656135194709,
|
| 2980 |
+
"grad_norm": 0.6502715945243835,
|
| 2981 |
+
"learning_rate": 0.00019312130760141787,
|
| 2982 |
+
"loss": 0.2479,
|
| 2983 |
+
"step": 36700
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 10.81704628949302,
|
| 2987 |
+
"grad_norm": 1.30823814868927,
|
| 2988 |
+
"learning_rate": 0.00019282591571484833,
|
| 2989 |
+
"loss": 0.254,
|
| 2990 |
+
"step": 36800
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 10.81704628949302,
|
| 2994 |
+
"eval_loss": 0.3001057505607605,
|
| 2995 |
+
"eval_runtime": 311.6753,
|
| 2996 |
+
"eval_samples_per_second": 43.664,
|
| 2997 |
+
"eval_steps_per_second": 5.461,
|
| 2998 |
+
"eval_wer": 0.3196277805368952,
|
| 2999 |
+
"step": 36800
|
| 3000 |
+
},
|
| 3001 |
+
{
|
| 3002 |
+
"epoch": 10.84643644379133,
|
| 3003 |
+
"grad_norm": 0.7160775065422058,
|
| 3004 |
+
"learning_rate": 0.00019253052382827885,
|
| 3005 |
+
"loss": 0.2566,
|
| 3006 |
+
"step": 36900
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 10.87582659808964,
|
| 3010 |
+
"grad_norm": 1.0508613586425781,
|
| 3011 |
+
"learning_rate": 0.0001922351319417093,
|
| 3012 |
+
"loss": 0.2643,
|
| 3013 |
+
"step": 37000
|
| 3014 |
+
},
|
| 3015 |
+
{
|
| 3016 |
+
"epoch": 10.90521675238795,
|
| 3017 |
+
"grad_norm": 1.081013798713684,
|
| 3018 |
+
"learning_rate": 0.0001919397400551398,
|
| 3019 |
+
"loss": 0.252,
|
| 3020 |
+
"step": 37100
|
| 3021 |
+
},
|
| 3022 |
+
{
|
| 3023 |
+
"epoch": 10.93460690668626,
|
| 3024 |
+
"grad_norm": 3.057793140411377,
|
| 3025 |
+
"learning_rate": 0.0001916443481685703,
|
| 3026 |
+
"loss": 0.2509,
|
| 3027 |
+
"step": 37200
|
| 3028 |
+
},
|
| 3029 |
+
{
|
| 3030 |
+
"epoch": 10.96399706098457,
|
| 3031 |
+
"grad_norm": 0.9743184447288513,
|
| 3032 |
+
"learning_rate": 0.00019134895628200075,
|
| 3033 |
+
"loss": 0.2457,
|
| 3034 |
+
"step": 37300
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 10.99338721528288,
|
| 3038 |
+
"grad_norm": 0.9493071436882019,
|
| 3039 |
+
"learning_rate": 0.00019105356439543127,
|
| 3040 |
+
"loss": 0.252,
|
| 3041 |
+
"step": 37400
|
| 3042 |
+
},
|
| 3043 |
+
{
|
| 3044 |
+
"epoch": 11.02277736958119,
|
| 3045 |
+
"grad_norm": 0.5813459753990173,
|
| 3046 |
+
"learning_rate": 0.00019075817250886173,
|
| 3047 |
+
"loss": 0.2411,
|
| 3048 |
+
"step": 37500
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 11.0521675238795,
|
| 3052 |
+
"grad_norm": 0.5560538172721863,
|
| 3053 |
+
"learning_rate": 0.00019046278062229222,
|
| 3054 |
+
"loss": 0.2458,
|
| 3055 |
+
"step": 37600
|
| 3056 |
+
},
|
| 3057 |
+
{
|
| 3058 |
+
"epoch": 11.0521675238795,
|
| 3059 |
+
"eval_loss": 0.2737277150154114,
|
| 3060 |
+
"eval_runtime": 314.8681,
|
| 3061 |
+
"eval_samples_per_second": 43.221,
|
| 3062 |
+
"eval_steps_per_second": 5.405,
|
| 3063 |
+
"eval_wer": 0.2969652612593774,
|
| 3064 |
+
"step": 37600
|
| 3065 |
+
},
|
| 3066 |
+
{
|
| 3067 |
+
"epoch": 11.081557678177811,
|
| 3068 |
+
"grad_norm": 0.6797336935997009,
|
| 3069 |
+
"learning_rate": 0.00019017034265458838,
|
| 3070 |
+
"loss": 0.2493,
|
| 3071 |
+
"step": 37700
|
| 3072 |
+
},
|
| 3073 |
+
{
|
| 3074 |
+
"epoch": 11.110947832476121,
|
| 3075 |
+
"grad_norm": 1.0297170877456665,
|
| 3076 |
+
"learning_rate": 0.0001898749507680189,
|
| 3077 |
+
"loss": 0.2344,
|
| 3078 |
+
"step": 37800
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 11.14033798677443,
|
| 3082 |
+
"grad_norm": 5.9290995597839355,
|
| 3083 |
+
"learning_rate": 0.00018957955888144936,
|
| 3084 |
+
"loss": 0.2438,
|
| 3085 |
+
"step": 37900
|
| 3086 |
+
},
|
| 3087 |
+
{
|
| 3088 |
+
"epoch": 11.16972814107274,
|
| 3089 |
+
"grad_norm": 0.9283130168914795,
|
| 3090 |
+
"learning_rate": 0.00018928416699487988,
|
| 3091 |
+
"loss": 0.2366,
|
| 3092 |
+
"step": 38000
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 11.19911829537105,
|
| 3096 |
+
"grad_norm": 0.508132815361023,
|
| 3097 |
+
"learning_rate": 0.00018898877510831034,
|
| 3098 |
+
"loss": 0.2376,
|
| 3099 |
+
"step": 38100
|
| 3100 |
+
},
|
| 3101 |
+
{
|
| 3102 |
+
"epoch": 11.228508449669361,
|
| 3103 |
+
"grad_norm": 1.174662470817566,
|
| 3104 |
+
"learning_rate": 0.0001886933832217408,
|
| 3105 |
+
"loss": 0.2589,
|
| 3106 |
+
"step": 38200
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 11.257898603967671,
|
| 3110 |
+
"grad_norm": 0.5180801153182983,
|
| 3111 |
+
"learning_rate": 0.00018839799133517132,
|
| 3112 |
+
"loss": 0.2449,
|
| 3113 |
+
"step": 38300
|
| 3114 |
+
},
|
| 3115 |
+
{
|
| 3116 |
+
"epoch": 11.28728875826598,
|
| 3117 |
+
"grad_norm": 0.9165621399879456,
|
| 3118 |
+
"learning_rate": 0.00018810259944860178,
|
| 3119 |
+
"loss": 0.239,
|
| 3120 |
+
"step": 38400
|
| 3121 |
+
},
|
| 3122 |
+
{
|
| 3123 |
+
"epoch": 11.28728875826598,
|
| 3124 |
+
"eval_loss": 0.28334641456604004,
|
| 3125 |
+
"eval_runtime": 347.4905,
|
| 3126 |
+
"eval_samples_per_second": 39.164,
|
| 3127 |
+
"eval_steps_per_second": 4.898,
|
| 3128 |
+
"eval_wer": 0.3057218286849466,
|
| 3129 |
+
"step": 38400
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 11.31667891256429,
|
| 3133 |
+
"grad_norm": 22.9261474609375,
|
| 3134 |
+
"learning_rate": 0.0001878072075620323,
|
| 3135 |
+
"loss": 0.2488,
|
| 3136 |
+
"step": 38500
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 11.3460690668626,
|
| 3140 |
+
"grad_norm": 0.8717837929725647,
|
| 3141 |
+
"learning_rate": 0.00018751181567546276,
|
| 3142 |
+
"loss": 0.2465,
|
| 3143 |
+
"step": 38600
|
| 3144 |
+
},
|
| 3145 |
+
{
|
| 3146 |
+
"epoch": 11.375459221160911,
|
| 3147 |
+
"grad_norm": 0.9947065711021423,
|
| 3148 |
+
"learning_rate": 0.00018721642378889322,
|
| 3149 |
+
"loss": 0.2434,
|
| 3150 |
+
"step": 38700
|
| 3151 |
+
},
|
| 3152 |
+
{
|
| 3153 |
+
"epoch": 11.404849375459222,
|
| 3154 |
+
"grad_norm": 0.5525564551353455,
|
| 3155 |
+
"learning_rate": 0.00018692103190232374,
|
| 3156 |
+
"loss": 0.2459,
|
| 3157 |
+
"step": 38800
|
| 3158 |
+
},
|
| 3159 |
+
{
|
| 3160 |
+
"epoch": 11.434239529757532,
|
| 3161 |
+
"grad_norm": 0.5672886371612549,
|
| 3162 |
+
"learning_rate": 0.0001866256400157542,
|
| 3163 |
+
"loss": 0.2445,
|
| 3164 |
+
"step": 38900
|
| 3165 |
+
},
|
| 3166 |
+
{
|
| 3167 |
+
"epoch": 11.46362968405584,
|
| 3168 |
+
"grad_norm": 0.8696480393409729,
|
| 3169 |
+
"learning_rate": 0.00018633024812918472,
|
| 3170 |
+
"loss": 0.2395,
|
| 3171 |
+
"step": 39000
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 11.49301983835415,
|
| 3175 |
+
"grad_norm": 0.6398095488548279,
|
| 3176 |
+
"learning_rate": 0.00018603485624261518,
|
| 3177 |
+
"loss": 0.2405,
|
| 3178 |
+
"step": 39100
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 11.522409992652461,
|
| 3182 |
+
"grad_norm": 0.6440668106079102,
|
| 3183 |
+
"learning_rate": 0.0001857394643560457,
|
| 3184 |
+
"loss": 0.2372,
|
| 3185 |
+
"step": 39200
|
| 3186 |
+
},
|
| 3187 |
+
{
|
| 3188 |
+
"epoch": 11.522409992652461,
|
| 3189 |
+
"eval_loss": 0.2920873165130615,
|
| 3190 |
+
"eval_runtime": 349.3517,
|
| 3191 |
+
"eval_samples_per_second": 38.955,
|
| 3192 |
+
"eval_steps_per_second": 4.872,
|
| 3193 |
+
"eval_wer": 0.3065844159238832,
|
| 3194 |
+
"step": 39200
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 11.551800146950772,
|
| 3198 |
+
"grad_norm": 1.4424521923065186,
|
| 3199 |
+
"learning_rate": 0.00018544407246947616,
|
| 3200 |
+
"loss": 0.2379,
|
| 3201 |
+
"step": 39300
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 11.581190301249082,
|
| 3205 |
+
"grad_norm": 0.8933115601539612,
|
| 3206 |
+
"learning_rate": 0.00018514868058290662,
|
| 3207 |
+
"loss": 0.2548,
|
| 3208 |
+
"step": 39400
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 11.610580455547392,
|
| 3212 |
+
"grad_norm": 4.223176956176758,
|
| 3213 |
+
"learning_rate": 0.00018485328869633714,
|
| 3214 |
+
"loss": 0.2414,
|
| 3215 |
+
"step": 39500
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 11.639970609845701,
|
| 3219 |
+
"grad_norm": 1.3100603818893433,
|
| 3220 |
+
"learning_rate": 0.0001845578968097676,
|
| 3221 |
+
"loss": 0.238,
|
| 3222 |
+
"step": 39600
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 11.669360764144011,
|
| 3226 |
+
"grad_norm": 0.5544066429138184,
|
| 3227 |
+
"learning_rate": 0.00018426250492319812,
|
| 3228 |
+
"loss": 0.2522,
|
| 3229 |
+
"step": 39700
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 11.698750918442322,
|
| 3233 |
+
"grad_norm": 0.6405923366546631,
|
| 3234 |
+
"learning_rate": 0.00018396711303662858,
|
| 3235 |
+
"loss": 0.2385,
|
| 3236 |
+
"step": 39800
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 11.728141072740632,
|
| 3240 |
+
"grad_norm": 0.6519741415977478,
|
| 3241 |
+
"learning_rate": 0.00018367172115005905,
|
| 3242 |
+
"loss": 0.2415,
|
| 3243 |
+
"step": 39900
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 11.757531227038942,
|
| 3247 |
+
"grad_norm": 1.0055747032165527,
|
| 3248 |
+
"learning_rate": 0.00018337632926348956,
|
| 3249 |
+
"loss": 0.2456,
|
| 3250 |
+
"step": 40000
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 11.757531227038942,
|
| 3254 |
+
"eval_loss": 0.2584936022758484,
|
| 3255 |
+
"eval_runtime": 312.8843,
|
| 3256 |
+
"eval_samples_per_second": 43.495,
|
| 3257 |
+
"eval_steps_per_second": 5.44,
|
| 3258 |
+
"eval_wer": 0.29210340591264344,
|
| 3259 |
+
"step": 40000
|
| 3260 |
+
},
|
| 3261 |
+
{
|
| 3262 |
+
"epoch": 11.786921381337251,
|
| 3263 |
+
"grad_norm": 1.399174451828003,
|
| 3264 |
+
"learning_rate": 0.00018308093737692002,
|
| 3265 |
+
"loss": 0.2482,
|
| 3266 |
+
"step": 40100
|
| 3267 |
+
},
|
| 3268 |
+
{
|
| 3269 |
+
"epoch": 11.816311535635561,
|
| 3270 |
+
"grad_norm": 0.5650673508644104,
|
| 3271 |
+
"learning_rate": 0.0001827884994092162,
|
| 3272 |
+
"loss": 0.247,
|
| 3273 |
+
"step": 40200
|
| 3274 |
+
},
|
| 3275 |
+
{
|
| 3276 |
+
"epoch": 11.845701689933872,
|
| 3277 |
+
"grad_norm": 1.0477852821350098,
|
| 3278 |
+
"learning_rate": 0.0001824931075226467,
|
| 3279 |
+
"loss": 0.2327,
|
| 3280 |
+
"step": 40300
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 11.875091844232182,
|
| 3284 |
+
"grad_norm": 0.4462454617023468,
|
| 3285 |
+
"learning_rate": 0.0001821977156360772,
|
| 3286 |
+
"loss": 0.2315,
|
| 3287 |
+
"step": 40400
|
| 3288 |
+
},
|
| 3289 |
+
{
|
| 3290 |
+
"epoch": 11.904481998530493,
|
| 3291 |
+
"grad_norm": 0.7627229690551758,
|
| 3292 |
+
"learning_rate": 0.00018190232374950765,
|
| 3293 |
+
"loss": 0.2311,
|
| 3294 |
+
"step": 40500
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 11.933872152828803,
|
| 3298 |
+
"grad_norm": 1.4027870893478394,
|
| 3299 |
+
"learning_rate": 0.00018160693186293814,
|
| 3300 |
+
"loss": 0.2309,
|
| 3301 |
+
"step": 40600
|
| 3302 |
+
},
|
| 3303 |
+
{
|
| 3304 |
+
"epoch": 11.963262307127113,
|
| 3305 |
+
"grad_norm": 1.4544415473937988,
|
| 3306 |
+
"learning_rate": 0.00018131153997636863,
|
| 3307 |
+
"loss": 0.2496,
|
| 3308 |
+
"step": 40700
|
| 3309 |
+
},
|
| 3310 |
+
{
|
| 3311 |
+
"epoch": 11.992652461425422,
|
| 3312 |
+
"grad_norm": 0.7475928664207458,
|
| 3313 |
+
"learning_rate": 0.00018101614808979912,
|
| 3314 |
+
"loss": 0.2395,
|
| 3315 |
+
"step": 40800
|
| 3316 |
+
},
|
| 3317 |
+
{
|
| 3318 |
+
"epoch": 11.992652461425422,
|
| 3319 |
+
"eval_loss": 0.2983209788799286,
|
| 3320 |
+
"eval_runtime": 318.565,
|
| 3321 |
+
"eval_samples_per_second": 42.72,
|
| 3322 |
+
"eval_steps_per_second": 5.343,
|
| 3323 |
+
"eval_wer": 0.30773453224246544,
|
| 3324 |
+
"step": 40800
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 12.022042615723732,
|
| 3328 |
+
"grad_norm": 2.2734031677246094,
|
| 3329 |
+
"learning_rate": 0.0001807207562032296,
|
| 3330 |
+
"loss": 0.2325,
|
| 3331 |
+
"step": 40900
|
| 3332 |
+
},
|
| 3333 |
+
{
|
| 3334 |
+
"epoch": 12.051432770022043,
|
| 3335 |
+
"grad_norm": 1.2131147384643555,
|
| 3336 |
+
"learning_rate": 0.00018042536431666007,
|
| 3337 |
+
"loss": 0.2216,
|
| 3338 |
+
"step": 41000
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 12.080822924320353,
|
| 3342 |
+
"grad_norm": 1.4542584419250488,
|
| 3343 |
+
"learning_rate": 0.00018012997243009056,
|
| 3344 |
+
"loss": 0.2115,
|
| 3345 |
+
"step": 41100
|
| 3346 |
+
},
|
| 3347 |
+
{
|
| 3348 |
+
"epoch": 12.110213078618663,
|
| 3349 |
+
"grad_norm": 0.8182443380355835,
|
| 3350 |
+
"learning_rate": 0.00017983458054352105,
|
| 3351 |
+
"loss": 0.2145,
|
| 3352 |
+
"step": 41200
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 12.139603232916972,
|
| 3356 |
+
"grad_norm": 0.5297923684120178,
|
| 3357 |
+
"learning_rate": 0.00017953918865695154,
|
| 3358 |
+
"loss": 0.2266,
|
| 3359 |
+
"step": 41300
|
| 3360 |
+
},
|
| 3361 |
+
{
|
| 3362 |
+
"epoch": 12.168993387215282,
|
| 3363 |
+
"grad_norm": 0.7970225811004639,
|
| 3364 |
+
"learning_rate": 0.00017924379677038203,
|
| 3365 |
+
"loss": 0.2303,
|
| 3366 |
+
"step": 41400
|
| 3367 |
+
},
|
| 3368 |
+
{
|
| 3369 |
+
"epoch": 12.198383541513593,
|
| 3370 |
+
"grad_norm": 0.8761687278747559,
|
| 3371 |
+
"learning_rate": 0.00017894840488381252,
|
| 3372 |
+
"loss": 0.2285,
|
| 3373 |
+
"step": 41500
|
| 3374 |
+
},
|
| 3375 |
+
{
|
| 3376 |
+
"epoch": 12.227773695811903,
|
| 3377 |
+
"grad_norm": 0.47353067994117737,
|
| 3378 |
+
"learning_rate": 0.00017865301299724298,
|
| 3379 |
+
"loss": 0.2226,
|
| 3380 |
+
"step": 41600
|
| 3381 |
+
},
|
| 3382 |
+
{
|
| 3383 |
+
"epoch": 12.227773695811903,
|
| 3384 |
+
"eval_loss": 0.269490122795105,
|
| 3385 |
+
"eval_runtime": 321.6439,
|
| 3386 |
+
"eval_samples_per_second": 42.311,
|
| 3387 |
+
"eval_steps_per_second": 5.292,
|
| 3388 |
+
"eval_wer": 0.2917113208040359,
|
| 3389 |
+
"step": 41600
|
| 3390 |
+
},
|
| 3391 |
+
{
|
| 3392 |
+
"epoch": 12.257163850110214,
|
| 3393 |
+
"grad_norm": 0.5443155169487,
|
| 3394 |
+
"learning_rate": 0.00017835762111067347,
|
| 3395 |
+
"loss": 0.2171,
|
| 3396 |
+
"step": 41700
|
| 3397 |
+
},
|
| 3398 |
+
{
|
| 3399 |
+
"epoch": 12.286554004408524,
|
| 3400 |
+
"grad_norm": 0.6112808585166931,
|
| 3401 |
+
"learning_rate": 0.00017806222922410396,
|
| 3402 |
+
"loss": 0.2292,
|
| 3403 |
+
"step": 41800
|
| 3404 |
+
},
|
| 3405 |
+
{
|
| 3406 |
+
"epoch": 12.315944158706833,
|
| 3407 |
+
"grad_norm": 0.6121761202812195,
|
| 3408 |
+
"learning_rate": 0.00017776683733753445,
|
| 3409 |
+
"loss": 0.2314,
|
| 3410 |
+
"step": 41900
|
| 3411 |
+
},
|
| 3412 |
+
{
|
| 3413 |
+
"epoch": 12.345334313005143,
|
| 3414 |
+
"grad_norm": 0.8672870993614197,
|
| 3415 |
+
"learning_rate": 0.00017747144545096494,
|
| 3416 |
+
"loss": 0.2177,
|
| 3417 |
+
"step": 42000
|
| 3418 |
+
},
|
| 3419 |
+
{
|
| 3420 |
+
"epoch": 12.374724467303453,
|
| 3421 |
+
"grad_norm": 1.1109092235565186,
|
| 3422 |
+
"learning_rate": 0.0001771760535643954,
|
| 3423 |
+
"loss": 0.2242,
|
| 3424 |
+
"step": 42100
|
| 3425 |
+
},
|
| 3426 |
+
{
|
| 3427 |
+
"epoch": 12.404114621601764,
|
| 3428 |
+
"grad_norm": 0.6793526411056519,
|
| 3429 |
+
"learning_rate": 0.0001768806616778259,
|
| 3430 |
+
"loss": 0.2323,
|
| 3431 |
+
"step": 42200
|
| 3432 |
+
},
|
| 3433 |
+
{
|
| 3434 |
+
"epoch": 12.433504775900074,
|
| 3435 |
+
"grad_norm": 0.6174443364143372,
|
| 3436 |
+
"learning_rate": 0.00017658526979125638,
|
| 3437 |
+
"loss": 0.244,
|
| 3438 |
+
"step": 42300
|
| 3439 |
+
},
|
| 3440 |
+
{
|
| 3441 |
+
"epoch": 12.462894930198383,
|
| 3442 |
+
"grad_norm": 0.585985541343689,
|
| 3443 |
+
"learning_rate": 0.00017628987790468687,
|
| 3444 |
+
"loss": 0.2218,
|
| 3445 |
+
"step": 42400
|
| 3446 |
+
},
|
| 3447 |
+
{
|
| 3448 |
+
"epoch": 12.462894930198383,
|
| 3449 |
+
"eval_loss": 0.2780846059322357,
|
| 3450 |
+
"eval_runtime": 323.894,
|
| 3451 |
+
"eval_samples_per_second": 42.017,
|
| 3452 |
+
"eval_steps_per_second": 5.255,
|
| 3453 |
+
"eval_wer": 0.3029249549102125,
|
| 3454 |
+
"step": 42400
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 12.492285084496693,
|
| 3458 |
+
"grad_norm": 0.5204631686210632,
|
| 3459 |
+
"learning_rate": 0.00017599448601811736,
|
| 3460 |
+
"loss": 0.2247,
|
| 3461 |
+
"step": 42500
|
| 3462 |
+
},
|
| 3463 |
+
{
|
| 3464 |
+
"epoch": 12.521675238795003,
|
| 3465 |
+
"grad_norm": 0.690335214138031,
|
| 3466 |
+
"learning_rate": 0.00017569909413154783,
|
| 3467 |
+
"loss": 0.2381,
|
| 3468 |
+
"step": 42600
|
| 3469 |
+
},
|
| 3470 |
+
{
|
| 3471 |
+
"epoch": 12.551065393093314,
|
| 3472 |
+
"grad_norm": 0.7468460202217102,
|
| 3473 |
+
"learning_rate": 0.00017540370224497834,
|
| 3474 |
+
"loss": 0.2241,
|
| 3475 |
+
"step": 42700
|
| 3476 |
+
},
|
| 3477 |
+
{
|
| 3478 |
+
"epoch": 12.580455547391624,
|
| 3479 |
+
"grad_norm": 0.6808618307113647,
|
| 3480 |
+
"learning_rate": 0.0001751083103584088,
|
| 3481 |
+
"loss": 0.2319,
|
| 3482 |
+
"step": 42800
|
| 3483 |
+
},
|
| 3484 |
+
{
|
| 3485 |
+
"epoch": 12.609845701689935,
|
| 3486 |
+
"grad_norm": 1.9951025247573853,
|
| 3487 |
+
"learning_rate": 0.0001748129184718393,
|
| 3488 |
+
"loss": 0.229,
|
| 3489 |
+
"step": 42900
|
| 3490 |
+
},
|
| 3491 |
+
{
|
| 3492 |
+
"epoch": 12.639235855988243,
|
| 3493 |
+
"grad_norm": 1.8629333972930908,
|
| 3494 |
+
"learning_rate": 0.00017451752658526978,
|
| 3495 |
+
"loss": 0.2311,
|
| 3496 |
+
"step": 43000
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 12.668626010286554,
|
| 3500 |
+
"grad_norm": 0.5588394403457642,
|
| 3501 |
+
"learning_rate": 0.00017422213469870025,
|
| 3502 |
+
"loss": 0.2334,
|
| 3503 |
+
"step": 43100
|
| 3504 |
+
},
|
| 3505 |
+
{
|
| 3506 |
+
"epoch": 12.698016164584864,
|
| 3507 |
+
"grad_norm": 0.49918389320373535,
|
| 3508 |
+
"learning_rate": 0.00017392674281213076,
|
| 3509 |
+
"loss": 0.2347,
|
| 3510 |
+
"step": 43200
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 12.698016164584864,
|
| 3514 |
+
"eval_loss": 0.2730830907821655,
|
| 3515 |
+
"eval_runtime": 316.9319,
|
| 3516 |
+
"eval_samples_per_second": 42.94,
|
| 3517 |
+
"eval_steps_per_second": 5.37,
|
| 3518 |
+
"eval_wer": 0.2982722116214026,
|
| 3519 |
+
"step": 43200
|
| 3520 |
+
},
|
| 3521 |
+
{
|
| 3522 |
+
"epoch": 12.727406318883174,
|
| 3523 |
+
"grad_norm": 0.8213378190994263,
|
| 3524 |
+
"learning_rate": 0.00017363135092556123,
|
| 3525 |
+
"loss": 0.2223,
|
| 3526 |
+
"step": 43300
|
| 3527 |
+
},
|
| 3528 |
+
{
|
| 3529 |
+
"epoch": 12.756796473181485,
|
| 3530 |
+
"grad_norm": 1.098509669303894,
|
| 3531 |
+
"learning_rate": 0.0001733359590389917,
|
| 3532 |
+
"loss": 0.2341,
|
| 3533 |
+
"step": 43400
|
| 3534 |
+
},
|
| 3535 |
+
{
|
| 3536 |
+
"epoch": 12.786186627479795,
|
| 3537 |
+
"grad_norm": 0.5582231879234314,
|
| 3538 |
+
"learning_rate": 0.0001730405671524222,
|
| 3539 |
+
"loss": 0.2229,
|
| 3540 |
+
"step": 43500
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 12.815576781778104,
|
| 3544 |
+
"grad_norm": 0.7057173252105713,
|
| 3545 |
+
"learning_rate": 0.00017274517526585267,
|
| 3546 |
+
"loss": 0.2277,
|
| 3547 |
+
"step": 43600
|
| 3548 |
+
},
|
| 3549 |
+
{
|
| 3550 |
+
"epoch": 12.844966936076414,
|
| 3551 |
+
"grad_norm": 0.9320444464683533,
|
| 3552 |
+
"learning_rate": 0.00017244978337928318,
|
| 3553 |
+
"loss": 0.2314,
|
| 3554 |
+
"step": 43700
|
| 3555 |
+
},
|
| 3556 |
+
{
|
| 3557 |
+
"epoch": 12.874357090374724,
|
| 3558 |
+
"grad_norm": 0.48274803161621094,
|
| 3559 |
+
"learning_rate": 0.00017215439149271365,
|
| 3560 |
+
"loss": 0.2327,
|
| 3561 |
+
"step": 43800
|
| 3562 |
+
},
|
| 3563 |
+
{
|
| 3564 |
+
"epoch": 12.903747244673035,
|
| 3565 |
+
"grad_norm": 0.5930495858192444,
|
| 3566 |
+
"learning_rate": 0.0001718589996061441,
|
| 3567 |
+
"loss": 0.2324,
|
| 3568 |
+
"step": 43900
|
| 3569 |
+
},
|
| 3570 |
+
{
|
| 3571 |
+
"epoch": 12.933137398971345,
|
| 3572 |
+
"grad_norm": 1.6704206466674805,
|
| 3573 |
+
"learning_rate": 0.00017156360771957463,
|
| 3574 |
+
"loss": 0.2279,
|
| 3575 |
+
"step": 44000
|
| 3576 |
+
},
|
| 3577 |
+
{
|
| 3578 |
+
"epoch": 12.933137398971345,
|
| 3579 |
+
"eval_loss": 0.2571496367454529,
|
| 3580 |
+
"eval_runtime": 317.9463,
|
| 3581 |
+
"eval_samples_per_second": 42.803,
|
| 3582 |
+
"eval_steps_per_second": 5.353,
|
| 3583 |
+
"eval_wer": 0.2777269519303657,
|
| 3584 |
+
"step": 44000
|
| 3585 |
+
},
|
| 3586 |
+
{
|
| 3587 |
+
"epoch": 12.962527553269656,
|
| 3588 |
+
"grad_norm": 0.8804912567138672,
|
| 3589 |
+
"learning_rate": 0.0001712682158330051,
|
| 3590 |
+
"loss": 0.2316,
|
| 3591 |
+
"step": 44100
|
| 3592 |
+
},
|
| 3593 |
+
{
|
| 3594 |
+
"epoch": 12.991917707567964,
|
| 3595 |
+
"grad_norm": 0.7364503741264343,
|
| 3596 |
+
"learning_rate": 0.0001709728239464356,
|
| 3597 |
+
"loss": 0.2372,
|
| 3598 |
+
"step": 44200
|
| 3599 |
+
},
|
| 3600 |
+
{
|
| 3601 |
+
"epoch": 13.021307861866275,
|
| 3602 |
+
"grad_norm": 1.2249096632003784,
|
| 3603 |
+
"learning_rate": 0.0001706803859787318,
|
| 3604 |
+
"loss": 0.2258,
|
| 3605 |
+
"step": 44300
|
| 3606 |
+
},
|
| 3607 |
+
{
|
| 3608 |
+
"epoch": 13.050698016164585,
|
| 3609 |
+
"grad_norm": 0.6756930351257324,
|
| 3610 |
+
"learning_rate": 0.00017038499409216225,
|
| 3611 |
+
"loss": 0.2175,
|
| 3612 |
+
"step": 44400
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 13.080088170462895,
|
| 3616 |
+
"grad_norm": 0.839859664440155,
|
| 3617 |
+
"learning_rate": 0.00017008960220559272,
|
| 3618 |
+
"loss": 0.2153,
|
| 3619 |
+
"step": 44500
|
| 3620 |
+
},
|
| 3621 |
+
{
|
| 3622 |
+
"epoch": 13.109478324761206,
|
| 3623 |
+
"grad_norm": 0.5334445238113403,
|
| 3624 |
+
"learning_rate": 0.00016979421031902323,
|
| 3625 |
+
"loss": 0.2064,
|
| 3626 |
+
"step": 44600
|
| 3627 |
+
},
|
| 3628 |
+
{
|
| 3629 |
+
"epoch": 13.138868479059514,
|
| 3630 |
+
"grad_norm": 0.47394588589668274,
|
| 3631 |
+
"learning_rate": 0.0001694988184324537,
|
| 3632 |
+
"loss": 0.2057,
|
| 3633 |
+
"step": 44700
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 13.168258633357825,
|
| 3637 |
+
"grad_norm": 0.7335759997367859,
|
| 3638 |
+
"learning_rate": 0.0001692034265458842,
|
| 3639 |
+
"loss": 0.223,
|
| 3640 |
+
"step": 44800
|
| 3641 |
+
},
|
| 3642 |
+
{
|
| 3643 |
+
"epoch": 13.168258633357825,
|
| 3644 |
+
"eval_loss": 0.278480589389801,
|
| 3645 |
+
"eval_runtime": 319.2984,
|
| 3646 |
+
"eval_samples_per_second": 42.622,
|
| 3647 |
+
"eval_steps_per_second": 5.33,
|
| 3648 |
+
"eval_wer": 0.2841048696970489,
|
| 3649 |
+
"step": 44800
|
| 3650 |
+
},
|
| 3651 |
+
{
|
| 3652 |
+
"epoch": 13.197648787656135,
|
| 3653 |
+
"grad_norm": 0.6104578971862793,
|
| 3654 |
+
"learning_rate": 0.00016890803465931467,
|
| 3655 |
+
"loss": 0.2079,
|
| 3656 |
+
"step": 44900
|
| 3657 |
+
},
|
| 3658 |
+
{
|
| 3659 |
+
"epoch": 13.227038941954445,
|
| 3660 |
+
"grad_norm": 0.9121543169021606,
|
| 3661 |
+
"learning_rate": 0.00016861264277274516,
|
| 3662 |
+
"loss": 0.2206,
|
| 3663 |
+
"step": 45000
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 13.256429096252756,
|
| 3667 |
+
"grad_norm": 1.9033684730529785,
|
| 3668 |
+
"learning_rate": 0.00016831725088617565,
|
| 3669 |
+
"loss": 0.2155,
|
| 3670 |
+
"step": 45100
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 13.285819250551066,
|
| 3674 |
+
"grad_norm": 0.5979294776916504,
|
| 3675 |
+
"learning_rate": 0.00016802185899960612,
|
| 3676 |
+
"loss": 0.2048,
|
| 3677 |
+
"step": 45200
|
| 3678 |
+
},
|
| 3679 |
+
{
|
| 3680 |
+
"epoch": 13.315209404849375,
|
| 3681 |
+
"grad_norm": 0.6869368553161621,
|
| 3682 |
+
"learning_rate": 0.00016772646711303663,
|
| 3683 |
+
"loss": 0.2139,
|
| 3684 |
+
"step": 45300
|
| 3685 |
+
},
|
| 3686 |
+
{
|
| 3687 |
+
"epoch": 13.344599559147685,
|
| 3688 |
+
"grad_norm": 0.6324857473373413,
|
| 3689 |
+
"learning_rate": 0.0001674310752264671,
|
| 3690 |
+
"loss": 0.2134,
|
| 3691 |
+
"step": 45400
|
| 3692 |
+
},
|
| 3693 |
+
{
|
| 3694 |
+
"epoch": 13.373989713445996,
|
| 3695 |
+
"grad_norm": 0.5607621669769287,
|
| 3696 |
+
"learning_rate": 0.00016713863725876328,
|
| 3697 |
+
"loss": 0.2187,
|
| 3698 |
+
"step": 45500
|
| 3699 |
+
},
|
| 3700 |
+
{
|
| 3701 |
+
"epoch": 13.403379867744306,
|
| 3702 |
+
"grad_norm": 0.5023543238639832,
|
| 3703 |
+
"learning_rate": 0.00016684324537219377,
|
| 3704 |
+
"loss": 0.1996,
|
| 3705 |
+
"step": 45600
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 13.403379867744306,
|
| 3709 |
+
"eval_loss": 0.2802923917770386,
|
| 3710 |
+
"eval_runtime": 314.2514,
|
| 3711 |
+
"eval_samples_per_second": 43.306,
|
| 3712 |
+
"eval_steps_per_second": 5.416,
|
| 3713 |
+
"eval_wer": 0.28700629950074497,
|
| 3714 |
+
"step": 45600
|
| 3715 |
+
},
|
| 3716 |
+
{
|
| 3717 |
+
"epoch": 13.432770022042616,
|
| 3718 |
+
"grad_norm": 1.5691343545913696,
|
| 3719 |
+
"learning_rate": 0.00016654785348562426,
|
| 3720 |
+
"loss": 0.2,
|
| 3721 |
+
"step": 45700
|
| 3722 |
+
},
|
| 3723 |
+
{
|
| 3724 |
+
"epoch": 13.462160176340927,
|
| 3725 |
+
"grad_norm": 0.6861874461174011,
|
| 3726 |
+
"learning_rate": 0.00016625246159905472,
|
| 3727 |
+
"loss": 0.2129,
|
| 3728 |
+
"step": 45800
|
| 3729 |
+
},
|
| 3730 |
+
{
|
| 3731 |
+
"epoch": 13.491550330639235,
|
| 3732 |
+
"grad_norm": 0.4962269365787506,
|
| 3733 |
+
"learning_rate": 0.0001659570697124852,
|
| 3734 |
+
"loss": 0.2114,
|
| 3735 |
+
"step": 45900
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 13.520940484937546,
|
| 3739 |
+
"grad_norm": 20.76888084411621,
|
| 3740 |
+
"learning_rate": 0.0001656616778259157,
|
| 3741 |
+
"loss": 0.2126,
|
| 3742 |
+
"step": 46000
|
| 3743 |
+
},
|
| 3744 |
+
{
|
| 3745 |
+
"epoch": 13.550330639235856,
|
| 3746 |
+
"grad_norm": 0.7075533270835876,
|
| 3747 |
+
"learning_rate": 0.0001653662859393462,
|
| 3748 |
+
"loss": 0.2272,
|
| 3749 |
+
"step": 46100
|
| 3750 |
+
},
|
| 3751 |
+
{
|
| 3752 |
+
"epoch": 13.579720793534166,
|
| 3753 |
+
"grad_norm": 0.5811964869499207,
|
| 3754 |
+
"learning_rate": 0.00016507089405277668,
|
| 3755 |
+
"loss": 0.2152,
|
| 3756 |
+
"step": 46200
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 13.609110947832477,
|
| 3760 |
+
"grad_norm": 0.7360579371452332,
|
| 3761 |
+
"learning_rate": 0.00016477550216620714,
|
| 3762 |
+
"loss": 0.2146,
|
| 3763 |
+
"step": 46300
|
| 3764 |
+
},
|
| 3765 |
+
{
|
| 3766 |
+
"epoch": 13.638501102130785,
|
| 3767 |
+
"grad_norm": 0.7185118794441223,
|
| 3768 |
+
"learning_rate": 0.00016448011027963763,
|
| 3769 |
+
"loss": 0.2112,
|
| 3770 |
+
"step": 46400
|
| 3771 |
+
},
|
| 3772 |
+
{
|
| 3773 |
+
"epoch": 13.638501102130785,
|
| 3774 |
+
"eval_loss": 0.2654140889644623,
|
| 3775 |
+
"eval_runtime": 311.7275,
|
| 3776 |
+
"eval_samples_per_second": 43.657,
|
| 3777 |
+
"eval_steps_per_second": 5.46,
|
| 3778 |
+
"eval_wer": 0.28823483284104867,
|
| 3779 |
+
"step": 46400
|
| 3780 |
+
},
|
| 3781 |
+
{
|
| 3782 |
+
"epoch": 13.667891256429096,
|
| 3783 |
+
"grad_norm": 0.9151061773300171,
|
| 3784 |
+
"learning_rate": 0.00016418471839306812,
|
| 3785 |
+
"loss": 0.2202,
|
| 3786 |
+
"step": 46500
|
| 3787 |
+
},
|
| 3788 |
+
{
|
| 3789 |
+
"epoch": 13.697281410727406,
|
| 3790 |
+
"grad_norm": 0.5849810838699341,
|
| 3791 |
+
"learning_rate": 0.0001638893265064986,
|
| 3792 |
+
"loss": 0.2155,
|
| 3793 |
+
"step": 46600
|
| 3794 |
+
},
|
| 3795 |
+
{
|
| 3796 |
+
"epoch": 13.726671565025717,
|
| 3797 |
+
"grad_norm": 0.865153431892395,
|
| 3798 |
+
"learning_rate": 0.0001635939346199291,
|
| 3799 |
+
"loss": 0.2222,
|
| 3800 |
+
"step": 46700
|
| 3801 |
+
},
|
| 3802 |
+
{
|
| 3803 |
+
"epoch": 13.756061719324027,
|
| 3804 |
+
"grad_norm": 1.4333291053771973,
|
| 3805 |
+
"learning_rate": 0.00016329854273335956,
|
| 3806 |
+
"loss": 0.2148,
|
| 3807 |
+
"step": 46800
|
| 3808 |
+
},
|
| 3809 |
+
{
|
| 3810 |
+
"epoch": 13.785451873622337,
|
| 3811 |
+
"grad_norm": 0.46400517225265503,
|
| 3812 |
+
"learning_rate": 0.00016300315084679005,
|
| 3813 |
+
"loss": 0.2105,
|
| 3814 |
+
"step": 46900
|
| 3815 |
+
},
|
| 3816 |
+
{
|
| 3817 |
+
"epoch": 13.814842027920646,
|
| 3818 |
+
"grad_norm": 0.6644917130470276,
|
| 3819 |
+
"learning_rate": 0.00016271071287908624,
|
| 3820 |
+
"loss": 0.2159,
|
| 3821 |
+
"step": 47000
|
| 3822 |
+
},
|
| 3823 |
+
{
|
| 3824 |
+
"epoch": 13.844232182218956,
|
| 3825 |
+
"grad_norm": 0.9118102192878723,
|
| 3826 |
+
"learning_rate": 0.00016241532099251673,
|
| 3827 |
+
"loss": 0.225,
|
| 3828 |
+
"step": 47100
|
| 3829 |
+
},
|
| 3830 |
+
{
|
| 3831 |
+
"epoch": 13.873622336517267,
|
| 3832 |
+
"grad_norm": 0.968024730682373,
|
| 3833 |
+
"learning_rate": 0.00016211992910594722,
|
| 3834 |
+
"loss": 0.2033,
|
| 3835 |
+
"step": 47200
|
| 3836 |
+
},
|
| 3837 |
+
{
|
| 3838 |
+
"epoch": 13.873622336517267,
|
| 3839 |
+
"eval_loss": 0.26217180490493774,
|
| 3840 |
+
"eval_runtime": 311.3948,
|
| 3841 |
+
"eval_samples_per_second": 43.703,
|
| 3842 |
+
"eval_steps_per_second": 5.466,
|
| 3843 |
+
"eval_wer": 0.27966123846616303,
|
| 3844 |
+
"step": 47200
|
| 3845 |
+
},
|
| 3846 |
+
{
|
| 3847 |
+
"epoch": 13.903012490815577,
|
| 3848 |
+
"grad_norm": 0.7965731620788574,
|
| 3849 |
+
"learning_rate": 0.00016182453721937768,
|
| 3850 |
+
"loss": 0.2233,
|
| 3851 |
+
"step": 47300
|
| 3852 |
+
},
|
| 3853 |
+
{
|
| 3854 |
+
"epoch": 13.932402645113887,
|
| 3855 |
+
"grad_norm": 0.31087547540664673,
|
| 3856 |
+
"learning_rate": 0.00016152914533280817,
|
| 3857 |
+
"loss": 0.2151,
|
| 3858 |
+
"step": 47400
|
| 3859 |
+
},
|
| 3860 |
+
{
|
| 3861 |
+
"epoch": 13.961792799412198,
|
| 3862 |
+
"grad_norm": 0.8611846566200256,
|
| 3863 |
+
"learning_rate": 0.00016123375344623866,
|
| 3864 |
+
"loss": 0.2207,
|
| 3865 |
+
"step": 47500
|
| 3866 |
+
},
|
| 3867 |
+
{
|
| 3868 |
+
"epoch": 13.991182953710506,
|
| 3869 |
+
"grad_norm": 0.7616714239120483,
|
| 3870 |
+
"learning_rate": 0.00016093836155966915,
|
| 3871 |
+
"loss": 0.2136,
|
| 3872 |
+
"step": 47600
|
| 3873 |
+
},
|
| 3874 |
+
{
|
| 3875 |
+
"epoch": 14.020573108008817,
|
| 3876 |
+
"grad_norm": 0.4521709084510803,
|
| 3877 |
+
"learning_rate": 0.00016064296967309964,
|
| 3878 |
+
"loss": 0.2039,
|
| 3879 |
+
"step": 47700
|
| 3880 |
+
},
|
| 3881 |
+
{
|
| 3882 |
+
"epoch": 14.049963262307127,
|
| 3883 |
+
"grad_norm": 1.4151129722595215,
|
| 3884 |
+
"learning_rate": 0.0001603475777865301,
|
| 3885 |
+
"loss": 0.2081,
|
| 3886 |
+
"step": 47800
|
| 3887 |
+
},
|
| 3888 |
+
{
|
| 3889 |
+
"epoch": 14.079353416605437,
|
| 3890 |
+
"grad_norm": 0.7611672878265381,
|
| 3891 |
+
"learning_rate": 0.00016005218589996062,
|
| 3892 |
+
"loss": 0.2114,
|
| 3893 |
+
"step": 47900
|
| 3894 |
+
},
|
| 3895 |
+
{
|
| 3896 |
+
"epoch": 14.108743570903748,
|
| 3897 |
+
"grad_norm": 0.7990052700042725,
|
| 3898 |
+
"learning_rate": 0.00015975679401339108,
|
| 3899 |
+
"loss": 0.2016,
|
| 3900 |
+
"step": 48000
|
| 3901 |
+
},
|
| 3902 |
+
{
|
| 3903 |
+
"epoch": 14.108743570903748,
|
| 3904 |
+
"eval_loss": 0.2599799633026123,
|
| 3905 |
+
"eval_runtime": 313.6246,
|
| 3906 |
+
"eval_samples_per_second": 43.393,
|
| 3907 |
+
"eval_steps_per_second": 5.427,
|
| 3908 |
+
"eval_wer": 0.2701466398306192,
|
| 3909 |
+
"step": 48000
|
| 3910 |
+
},
|
| 3911 |
+
{
|
| 3912 |
+
"epoch": 14.138133725202056,
|
| 3913 |
+
"grad_norm": 0.8380939364433289,
|
| 3914 |
+
"learning_rate": 0.00015946140212682157,
|
| 3915 |
+
"loss": 0.2056,
|
| 3916 |
+
"step": 48100
|
| 3917 |
+
},
|
| 3918 |
+
{
|
| 3919 |
+
"epoch": 14.167523879500367,
|
| 3920 |
+
"grad_norm": 1.3163111209869385,
|
| 3921 |
+
"learning_rate": 0.00015916601024025206,
|
| 3922 |
+
"loss": 0.1944,
|
| 3923 |
+
"step": 48200
|
| 3924 |
+
},
|
| 3925 |
+
{
|
| 3926 |
+
"epoch": 14.196914033798677,
|
| 3927 |
+
"grad_norm": 0.8919910788536072,
|
| 3928 |
+
"learning_rate": 0.00015887061835368252,
|
| 3929 |
+
"loss": 0.198,
|
| 3930 |
+
"step": 48300
|
| 3931 |
+
},
|
| 3932 |
+
{
|
| 3933 |
+
"epoch": 14.226304188096988,
|
| 3934 |
+
"grad_norm": 0.5834262371063232,
|
| 3935 |
+
"learning_rate": 0.00015857522646711304,
|
| 3936 |
+
"loss": 0.1975,
|
| 3937 |
+
"step": 48400
|
| 3938 |
+
},
|
| 3939 |
+
{
|
| 3940 |
+
"epoch": 14.255694342395298,
|
| 3941 |
+
"grad_norm": 0.6582635641098022,
|
| 3942 |
+
"learning_rate": 0.0001582798345805435,
|
| 3943 |
+
"loss": 0.2035,
|
| 3944 |
+
"step": 48500
|
| 3945 |
+
},
|
| 3946 |
+
{
|
| 3947 |
+
"epoch": 14.285084496693608,
|
| 3948 |
+
"grad_norm": 4.873237609863281,
|
| 3949 |
+
"learning_rate": 0.000157984442693974,
|
| 3950 |
+
"loss": 0.2074,
|
| 3951 |
+
"step": 48600
|
| 3952 |
+
},
|
| 3953 |
+
{
|
| 3954 |
+
"epoch": 14.314474650991917,
|
| 3955 |
+
"grad_norm": 0.701461672782898,
|
| 3956 |
+
"learning_rate": 0.00015768905080740448,
|
| 3957 |
+
"loss": 0.2036,
|
| 3958 |
+
"step": 48700
|
| 3959 |
+
},
|
| 3960 |
+
{
|
| 3961 |
+
"epoch": 14.343864805290227,
|
| 3962 |
+
"grad_norm": 2.8444039821624756,
|
| 3963 |
+
"learning_rate": 0.00015739365892083494,
|
| 3964 |
+
"loss": 0.2053,
|
| 3965 |
+
"step": 48800
|
| 3966 |
+
},
|
| 3967 |
+
{
|
| 3968 |
+
"epoch": 14.343864805290227,
|
| 3969 |
+
"eval_loss": 0.27362382411956787,
|
| 3970 |
+
"eval_runtime": 315.1753,
|
| 3971 |
+
"eval_samples_per_second": 43.179,
|
| 3972 |
+
"eval_steps_per_second": 5.4,
|
| 3973 |
+
"eval_wer": 0.2920772669054029,
|
| 3974 |
+
"step": 48800
|
| 3975 |
+
},
|
| 3976 |
+
{
|
| 3977 |
+
"epoch": 14.373254959588538,
|
| 3978 |
+
"grad_norm": 0.7985801696777344,
|
| 3979 |
+
"learning_rate": 0.00015709826703426546,
|
| 3980 |
+
"loss": 0.2046,
|
| 3981 |
+
"step": 48900
|
| 3982 |
+
},
|
| 3983 |
+
{
|
| 3984 |
+
"epoch": 14.402645113886848,
|
| 3985 |
+
"grad_norm": 1.104037880897522,
|
| 3986 |
+
"learning_rate": 0.00015680287514769592,
|
| 3987 |
+
"loss": 0.2015,
|
| 3988 |
+
"step": 49000
|
| 3989 |
+
},
|
| 3990 |
+
{
|
| 3991 |
+
"epoch": 14.432035268185158,
|
| 3992 |
+
"grad_norm": 1.5909720659255981,
|
| 3993 |
+
"learning_rate": 0.00015650748326112639,
|
| 3994 |
+
"loss": 0.2047,
|
| 3995 |
+
"step": 49100
|
| 3996 |
+
},
|
| 3997 |
+
{
|
| 3998 |
+
"epoch": 14.461425422483469,
|
| 3999 |
+
"grad_norm": 1.6235400438308716,
|
| 4000 |
+
"learning_rate": 0.0001562120913745569,
|
| 4001 |
+
"loss": 0.2022,
|
| 4002 |
+
"step": 49200
|
| 4003 |
+
},
|
| 4004 |
+
{
|
| 4005 |
+
"epoch": 14.490815576781777,
|
| 4006 |
+
"grad_norm": 1.2231897115707397,
|
| 4007 |
+
"learning_rate": 0.00015591669948798736,
|
| 4008 |
+
"loss": 0.2043,
|
| 4009 |
+
"step": 49300
|
| 4010 |
+
},
|
| 4011 |
+
{
|
| 4012 |
+
"epoch": 14.520205731080088,
|
| 4013 |
+
"grad_norm": 0.4401768445968628,
|
| 4014 |
+
"learning_rate": 0.00015562130760141788,
|
| 4015 |
+
"loss": 0.1955,
|
| 4016 |
+
"step": 49400
|
| 4017 |
+
},
|
| 4018 |
+
{
|
| 4019 |
+
"epoch": 14.549595885378398,
|
| 4020 |
+
"grad_norm": 0.49192237854003906,
|
| 4021 |
+
"learning_rate": 0.00015532591571484834,
|
| 4022 |
+
"loss": 0.2038,
|
| 4023 |
+
"step": 49500
|
| 4024 |
+
},
|
| 4025 |
+
{
|
| 4026 |
+
"epoch": 14.578986039676709,
|
| 4027 |
+
"grad_norm": 1.6232017278671265,
|
| 4028 |
+
"learning_rate": 0.00015503052382827886,
|
| 4029 |
+
"loss": 0.1904,
|
| 4030 |
+
"step": 49600
|
| 4031 |
+
},
|
| 4032 |
+
{
|
| 4033 |
+
"epoch": 14.578986039676709,
|
| 4034 |
+
"eval_loss": 0.2584870159626007,
|
| 4035 |
+
"eval_runtime": 313.9077,
|
| 4036 |
+
"eval_samples_per_second": 43.354,
|
| 4037 |
+
"eval_steps_per_second": 5.422,
|
| 4038 |
+
"eval_wer": 0.26975455472201165,
|
| 4039 |
+
"step": 49600
|
| 4040 |
+
},
|
| 4041 |
+
{
|
| 4042 |
+
"epoch": 14.608376193975019,
|
| 4043 |
+
"grad_norm": 0.7949817180633545,
|
| 4044 |
+
"learning_rate": 0.00015473513194170932,
|
| 4045 |
+
"loss": 0.2022,
|
| 4046 |
+
"step": 49700
|
| 4047 |
+
},
|
| 4048 |
+
{
|
| 4049 |
+
"epoch": 14.637766348273328,
|
| 4050 |
+
"grad_norm": 0.5467865467071533,
|
| 4051 |
+
"learning_rate": 0.00015443974005513979,
|
| 4052 |
+
"loss": 0.1979,
|
| 4053 |
+
"step": 49800
|
| 4054 |
+
},
|
| 4055 |
+
{
|
| 4056 |
+
"epoch": 14.667156502571638,
|
| 4057 |
+
"grad_norm": 0.5495031476020813,
|
| 4058 |
+
"learning_rate": 0.0001541443481685703,
|
| 4059 |
+
"loss": 0.2042,
|
| 4060 |
+
"step": 49900
|
| 4061 |
+
},
|
| 4062 |
+
{
|
| 4063 |
+
"epoch": 14.696546656869948,
|
| 4064 |
+
"grad_norm": 0.7616658806800842,
|
| 4065 |
+
"learning_rate": 0.00015384895628200076,
|
| 4066 |
+
"loss": 0.1995,
|
| 4067 |
+
"step": 50000
|
| 4068 |
+
},
|
| 4069 |
+
{
|
| 4070 |
+
"epoch": 14.725936811168259,
|
| 4071 |
+
"grad_norm": 0.4436214864253998,
|
| 4072 |
+
"learning_rate": 0.00015355356439543128,
|
| 4073 |
+
"loss": 0.1901,
|
| 4074 |
+
"step": 50100
|
| 4075 |
+
},
|
| 4076 |
+
{
|
| 4077 |
+
"epoch": 14.755326965466569,
|
| 4078 |
+
"grad_norm": 26.983057022094727,
|
| 4079 |
+
"learning_rate": 0.00015325817250886174,
|
| 4080 |
+
"loss": 0.198,
|
| 4081 |
+
"step": 50200
|
| 4082 |
+
},
|
| 4083 |
+
{
|
| 4084 |
+
"epoch": 14.78471711976488,
|
| 4085 |
+
"grad_norm": 0.6213018894195557,
|
| 4086 |
+
"learning_rate": 0.0001529627806222922,
|
| 4087 |
+
"loss": 0.205,
|
| 4088 |
+
"step": 50300
|
| 4089 |
+
},
|
| 4090 |
+
{
|
| 4091 |
+
"epoch": 14.814107274063188,
|
| 4092 |
+
"grad_norm": 0.7121649384498596,
|
| 4093 |
+
"learning_rate": 0.00015266738873572272,
|
| 4094 |
+
"loss": 0.2048,
|
| 4095 |
+
"step": 50400
|
| 4096 |
+
},
|
| 4097 |
+
{
|
| 4098 |
+
"epoch": 14.814107274063188,
|
| 4099 |
+
"eval_loss": 0.276618093252182,
|
| 4100 |
+
"eval_runtime": 312.785,
|
| 4101 |
+
"eval_samples_per_second": 43.509,
|
| 4102 |
+
"eval_steps_per_second": 5.441,
|
| 4103 |
+
"eval_wer": 0.29165904278955485,
|
| 4104 |
+
"step": 50400
|
| 4105 |
+
},
|
| 4106 |
+
{
|
| 4107 |
+
"epoch": 14.843497428361498,
|
| 4108 |
+
"grad_norm": 0.5980726480484009,
|
| 4109 |
+
"learning_rate": 0.00015237199684915319,
|
| 4110 |
+
"loss": 0.2001,
|
| 4111 |
+
"step": 50500
|
| 4112 |
+
},
|
| 4113 |
+
{
|
| 4114 |
+
"epoch": 14.872887582659809,
|
| 4115 |
+
"grad_norm": 0.5173223614692688,
|
| 4116 |
+
"learning_rate": 0.0001520766049625837,
|
| 4117 |
+
"loss": 0.2019,
|
| 4118 |
+
"step": 50600
|
| 4119 |
+
},
|
| 4120 |
+
{
|
| 4121 |
+
"epoch": 14.90227773695812,
|
| 4122 |
+
"grad_norm": 0.5689155459403992,
|
| 4123 |
+
"learning_rate": 0.00015178121307601416,
|
| 4124 |
+
"loss": 0.2022,
|
| 4125 |
+
"step": 50700
|
| 4126 |
+
},
|
| 4127 |
+
{
|
| 4128 |
+
"epoch": 14.93166789125643,
|
| 4129 |
+
"grad_norm": 1.0008867979049683,
|
| 4130 |
+
"learning_rate": 0.00015148582118944463,
|
| 4131 |
+
"loss": 0.2136,
|
| 4132 |
+
"step": 50800
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 14.96105804555474,
|
| 4136 |
+
"grad_norm": 0.771344780921936,
|
| 4137 |
+
"learning_rate": 0.00015119042930287514,
|
| 4138 |
+
"loss": 0.2045,
|
| 4139 |
+
"step": 50900
|
| 4140 |
+
},
|
| 4141 |
+
{
|
| 4142 |
+
"epoch": 14.990448199853049,
|
| 4143 |
+
"grad_norm": 1.929983377456665,
|
| 4144 |
+
"learning_rate": 0.0001508950374163056,
|
| 4145 |
+
"loss": 0.2065,
|
| 4146 |
+
"step": 51000
|
| 4147 |
+
},
|
| 4148 |
+
{
|
| 4149 |
+
"epoch": 15.019838354151359,
|
| 4150 |
+
"grad_norm": 0.8995310664176941,
|
| 4151 |
+
"learning_rate": 0.0001506025994486018,
|
| 4152 |
+
"loss": 0.1963,
|
| 4153 |
+
"step": 51100
|
| 4154 |
+
},
|
| 4155 |
+
{
|
| 4156 |
+
"epoch": 15.04922850844967,
|
| 4157 |
+
"grad_norm": 1.213279128074646,
|
| 4158 |
+
"learning_rate": 0.00015030720756203228,
|
| 4159 |
+
"loss": 0.1946,
|
| 4160 |
+
"step": 51200
|
| 4161 |
+
},
|
| 4162 |
+
{
|
| 4163 |
+
"epoch": 15.04922850844967,
|
| 4164 |
+
"eval_loss": 0.2765796184539795,
|
| 4165 |
+
"eval_runtime": 312.0165,
|
| 4166 |
+
"eval_samples_per_second": 43.616,
|
| 4167 |
+
"eval_steps_per_second": 5.455,
|
| 4168 |
+
"eval_wer": 0.27456413205426455,
|
| 4169 |
+
"step": 51200
|
| 4170 |
+
},
|
| 4171 |
+
{
|
| 4172 |
+
"epoch": 15.07861866274798,
|
| 4173 |
+
"grad_norm": 0.6119352579116821,
|
| 4174 |
+
"learning_rate": 0.00015001181567546277,
|
| 4175 |
+
"loss": 0.1792,
|
| 4176 |
+
"step": 51300
|
| 4177 |
+
},
|
| 4178 |
+
{
|
| 4179 |
+
"epoch": 15.10800881704629,
|
| 4180 |
+
"grad_norm": 0.39601069688796997,
|
| 4181 |
+
"learning_rate": 0.00014971642378889326,
|
| 4182 |
+
"loss": 0.1953,
|
| 4183 |
+
"step": 51400
|
| 4184 |
+
},
|
| 4185 |
+
{
|
| 4186 |
+
"epoch": 15.137398971344599,
|
| 4187 |
+
"grad_norm": 0.5639697909355164,
|
| 4188 |
+
"learning_rate": 0.00014942103190232375,
|
| 4189 |
+
"loss": 0.1822,
|
| 4190 |
+
"step": 51500
|
| 4191 |
+
},
|
| 4192 |
+
{
|
| 4193 |
+
"epoch": 15.166789125642909,
|
| 4194 |
+
"grad_norm": 0.6990806460380554,
|
| 4195 |
+
"learning_rate": 0.00014912564001575424,
|
| 4196 |
+
"loss": 0.1921,
|
| 4197 |
+
"step": 51600
|
| 4198 |
+
},
|
| 4199 |
+
{
|
| 4200 |
+
"epoch": 15.19617927994122,
|
| 4201 |
+
"grad_norm": 1.0608422756195068,
|
| 4202 |
+
"learning_rate": 0.0001488302481291847,
|
| 4203 |
+
"loss": 0.1948,
|
| 4204 |
+
"step": 51700
|
| 4205 |
+
},
|
| 4206 |
+
{
|
| 4207 |
+
"epoch": 15.22556943423953,
|
| 4208 |
+
"grad_norm": 1.2730339765548706,
|
| 4209 |
+
"learning_rate": 0.0001485348562426152,
|
| 4210 |
+
"loss": 0.191,
|
| 4211 |
+
"step": 51800
|
| 4212 |
+
},
|
| 4213 |
+
{
|
| 4214 |
+
"epoch": 15.25495958853784,
|
| 4215 |
+
"grad_norm": 3.5955941677093506,
|
| 4216 |
+
"learning_rate": 0.00014823946435604568,
|
| 4217 |
+
"loss": 0.1914,
|
| 4218 |
+
"step": 51900
|
| 4219 |
+
},
|
| 4220 |
+
{
|
| 4221 |
+
"epoch": 15.28434974283615,
|
| 4222 |
+
"grad_norm": 2.9176721572875977,
|
| 4223 |
+
"learning_rate": 0.00014794407246947617,
|
| 4224 |
+
"loss": 0.185,
|
| 4225 |
+
"step": 52000
|
| 4226 |
+
},
|
| 4227 |
+
{
|
| 4228 |
+
"epoch": 15.28434974283615,
|
| 4229 |
+
"eval_loss": 0.26825693249702454,
|
| 4230 |
+
"eval_runtime": 313.865,
|
| 4231 |
+
"eval_samples_per_second": 43.359,
|
| 4232 |
+
"eval_steps_per_second": 5.423,
|
| 4233 |
+
"eval_wer": 0.2771518937710746,
|
| 4234 |
+
"step": 52000
|
| 4235 |
+
},
|
| 4236 |
+
{
|
| 4237 |
+
"epoch": 15.31373989713446,
|
| 4238 |
+
"grad_norm": 0.6137803196907043,
|
| 4239 |
+
"learning_rate": 0.00014764868058290663,
|
| 4240 |
+
"loss": 0.1922,
|
| 4241 |
+
"step": 52100
|
| 4242 |
+
},
|
| 4243 |
+
{
|
| 4244 |
+
"epoch": 15.34313005143277,
|
| 4245 |
+
"grad_norm": 0.4900952875614166,
|
| 4246 |
+
"learning_rate": 0.00014735328869633712,
|
| 4247 |
+
"loss": 0.19,
|
| 4248 |
+
"step": 52200
|
| 4249 |
+
},
|
| 4250 |
+
{
|
| 4251 |
+
"epoch": 15.37252020573108,
|
| 4252 |
+
"grad_norm": 1.1513181924819946,
|
| 4253 |
+
"learning_rate": 0.0001470578968097676,
|
| 4254 |
+
"loss": 0.1827,
|
| 4255 |
+
"step": 52300
|
| 4256 |
+
},
|
| 4257 |
+
{
|
| 4258 |
+
"epoch": 15.40191036002939,
|
| 4259 |
+
"grad_norm": 0.9658687710762024,
|
| 4260 |
+
"learning_rate": 0.0001467625049231981,
|
| 4261 |
+
"loss": 0.1996,
|
| 4262 |
+
"step": 52400
|
| 4263 |
+
},
|
| 4264 |
+
{
|
| 4265 |
+
"epoch": 15.4313005143277,
|
| 4266 |
+
"grad_norm": 0.8168590068817139,
|
| 4267 |
+
"learning_rate": 0.0001464671130366286,
|
| 4268 |
+
"loss": 0.2009,
|
| 4269 |
+
"step": 52500
|
| 4270 |
+
},
|
| 4271 |
+
{
|
| 4272 |
+
"epoch": 15.460690668626011,
|
| 4273 |
+
"grad_norm": 0.6976146101951599,
|
| 4274 |
+
"learning_rate": 0.00014617172115005906,
|
| 4275 |
+
"loss": 0.1897,
|
| 4276 |
+
"step": 52600
|
| 4277 |
+
},
|
| 4278 |
+
{
|
| 4279 |
+
"epoch": 15.49008082292432,
|
| 4280 |
+
"grad_norm": 1.0280542373657227,
|
| 4281 |
+
"learning_rate": 0.00014587632926348954,
|
| 4282 |
+
"loss": 0.1935,
|
| 4283 |
+
"step": 52700
|
| 4284 |
+
},
|
| 4285 |
+
{
|
| 4286 |
+
"epoch": 15.51947097722263,
|
| 4287 |
+
"grad_norm": 2.662313461303711,
|
| 4288 |
+
"learning_rate": 0.00014558093737692003,
|
| 4289 |
+
"loss": 0.2044,
|
| 4290 |
+
"step": 52800
|
| 4291 |
+
},
|
| 4292 |
+
{
|
| 4293 |
+
"epoch": 15.51947097722263,
|
| 4294 |
+
"eval_loss": 0.2819647490978241,
|
| 4295 |
+
"eval_runtime": 310.7247,
|
| 4296 |
+
"eval_samples_per_second": 43.798,
|
| 4297 |
+
"eval_steps_per_second": 5.478,
|
| 4298 |
+
"eval_wer": 0.2861698512690488,
|
| 4299 |
+
"step": 52800
|
| 4300 |
+
},
|
| 4301 |
+
{
|
| 4302 |
+
"epoch": 15.54886113152094,
|
| 4303 |
+
"grad_norm": 0.8253285884857178,
|
| 4304 |
+
"learning_rate": 0.00014528554549035052,
|
| 4305 |
+
"loss": 0.1896,
|
| 4306 |
+
"step": 52900
|
| 4307 |
+
},
|
| 4308 |
+
{
|
| 4309 |
+
"epoch": 15.57825128581925,
|
| 4310 |
+
"grad_norm": 0.95353102684021,
|
| 4311 |
+
"learning_rate": 0.00014499015360378101,
|
| 4312 |
+
"loss": 0.1935,
|
| 4313 |
+
"step": 53000
|
| 4314 |
+
},
|
| 4315 |
+
{
|
| 4316 |
+
"epoch": 15.607641440117561,
|
| 4317 |
+
"grad_norm": 3.6389894485473633,
|
| 4318 |
+
"learning_rate": 0.00014469476171721148,
|
| 4319 |
+
"loss": 0.1974,
|
| 4320 |
+
"step": 53100
|
| 4321 |
+
},
|
| 4322 |
+
{
|
| 4323 |
+
"epoch": 15.63703159441587,
|
| 4324 |
+
"grad_norm": 6.647377014160156,
|
| 4325 |
+
"learning_rate": 0.00014439936983064197,
|
| 4326 |
+
"loss": 0.1811,
|
| 4327 |
+
"step": 53200
|
| 4328 |
+
},
|
| 4329 |
+
{
|
| 4330 |
+
"epoch": 15.66642174871418,
|
| 4331 |
+
"grad_norm": 0.5306282639503479,
|
| 4332 |
+
"learning_rate": 0.00014410397794407246,
|
| 4333 |
+
"loss": 0.1985,
|
| 4334 |
+
"step": 53300
|
| 4335 |
+
},
|
| 4336 |
+
{
|
| 4337 |
+
"epoch": 15.69581190301249,
|
| 4338 |
+
"grad_norm": 0.3733627498149872,
|
| 4339 |
+
"learning_rate": 0.00014380858605750294,
|
| 4340 |
+
"loss": 0.1977,
|
| 4341 |
+
"step": 53400
|
| 4342 |
+
},
|
| 4343 |
+
{
|
| 4344 |
+
"epoch": 15.725202057310801,
|
| 4345 |
+
"grad_norm": 0.4278078079223633,
|
| 4346 |
+
"learning_rate": 0.00014351319417093343,
|
| 4347 |
+
"loss": 0.1787,
|
| 4348 |
+
"step": 53500
|
| 4349 |
+
},
|
| 4350 |
+
{
|
| 4351 |
+
"epoch": 15.754592211609111,
|
| 4352 |
+
"grad_norm": 0.9906378984451294,
|
| 4353 |
+
"learning_rate": 0.0001432178022843639,
|
| 4354 |
+
"loss": 0.1885,
|
| 4355 |
+
"step": 53600
|
| 4356 |
+
},
|
| 4357 |
+
{
|
| 4358 |
+
"epoch": 15.754592211609111,
|
| 4359 |
+
"eval_loss": 0.2929398715496063,
|
| 4360 |
+
"eval_runtime": 312.9232,
|
| 4361 |
+
"eval_samples_per_second": 43.49,
|
| 4362 |
+
"eval_steps_per_second": 5.439,
|
| 4363 |
+
"eval_wer": 0.2867971874428209,
|
| 4364 |
+
"step": 53600
|
| 4365 |
+
},
|
| 4366 |
+
{
|
| 4367 |
+
"epoch": 15.783982365907422,
|
| 4368 |
+
"grad_norm": 0.40985000133514404,
|
| 4369 |
+
"learning_rate": 0.0001429224103977944,
|
| 4370 |
+
"loss": 0.1853,
|
| 4371 |
+
"step": 53700
|
| 4372 |
+
},
|
| 4373 |
+
{
|
| 4374 |
+
"epoch": 15.81337252020573,
|
| 4375 |
+
"grad_norm": 0.4449746012687683,
|
| 4376 |
+
"learning_rate": 0.00014262701851122488,
|
| 4377 |
+
"loss": 0.1853,
|
| 4378 |
+
"step": 53800
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 15.84276267450404,
|
| 4382 |
+
"grad_norm": 1.338578701019287,
|
| 4383 |
+
"learning_rate": 0.00014233162662465537,
|
| 4384 |
+
"loss": 0.1946,
|
| 4385 |
+
"step": 53900
|
| 4386 |
+
},
|
| 4387 |
+
{
|
| 4388 |
+
"epoch": 15.872152828802351,
|
| 4389 |
+
"grad_norm": 0.6558298468589783,
|
| 4390 |
+
"learning_rate": 0.00014203623473808583,
|
| 4391 |
+
"loss": 0.1886,
|
| 4392 |
+
"step": 54000
|
| 4393 |
+
},
|
| 4394 |
+
{
|
| 4395 |
+
"epoch": 15.901542983100661,
|
| 4396 |
+
"grad_norm": 0.9642869830131531,
|
| 4397 |
+
"learning_rate": 0.00014174084285151632,
|
| 4398 |
+
"loss": 0.1953,
|
| 4399 |
+
"step": 54100
|
| 4400 |
+
},
|
| 4401 |
+
{
|
| 4402 |
+
"epoch": 15.930933137398972,
|
| 4403 |
+
"grad_norm": 6.021807670593262,
|
| 4404 |
+
"learning_rate": 0.0001414454509649468,
|
| 4405 |
+
"loss": 0.189,
|
| 4406 |
+
"step": 54200
|
| 4407 |
+
},
|
| 4408 |
+
{
|
| 4409 |
+
"epoch": 15.960323291697282,
|
| 4410 |
+
"grad_norm": 0.8446222543716431,
|
| 4411 |
+
"learning_rate": 0.0001411500590783773,
|
| 4412 |
+
"loss": 0.1907,
|
| 4413 |
+
"step": 54300
|
| 4414 |
+
},
|
| 4415 |
+
{
|
| 4416 |
+
"epoch": 15.98971344599559,
|
| 4417 |
+
"grad_norm": 0.8668615221977234,
|
| 4418 |
+
"learning_rate": 0.0001408546671918078,
|
| 4419 |
+
"loss": 0.1855,
|
| 4420 |
+
"step": 54400
|
| 4421 |
+
},
|
| 4422 |
+
{
|
| 4423 |
+
"epoch": 15.98971344599559,
|
| 4424 |
+
"eval_loss": 0.2681010067462921,
|
| 4425 |
+
"eval_runtime": 310.4575,
|
| 4426 |
+
"eval_samples_per_second": 43.835,
|
| 4427 |
+
"eval_steps_per_second": 5.482,
|
| 4428 |
+
"eval_wer": 0.2788509292417074,
|
| 4429 |
+
"step": 54400
|
| 4430 |
+
},
|
| 4431 |
+
{
|
| 4432 |
+
"epoch": 16.019103600293903,
|
| 4433 |
+
"grad_norm": 0.5163881182670593,
|
| 4434 |
+
"learning_rate": 0.00014055927530523828,
|
| 4435 |
+
"loss": 0.1858,
|
| 4436 |
+
"step": 54500
|
| 4437 |
+
},
|
| 4438 |
+
{
|
| 4439 |
+
"epoch": 16.04849375459221,
|
| 4440 |
+
"grad_norm": 1.519850254058838,
|
| 4441 |
+
"learning_rate": 0.00014026683733753443,
|
| 4442 |
+
"loss": 0.1846,
|
| 4443 |
+
"step": 54600
|
| 4444 |
+
},
|
| 4445 |
+
{
|
| 4446 |
+
"epoch": 16.07788390889052,
|
| 4447 |
+
"grad_norm": 0.6191527843475342,
|
| 4448 |
+
"learning_rate": 0.00013997144545096492,
|
| 4449 |
+
"loss": 0.1784,
|
| 4450 |
+
"step": 54700
|
| 4451 |
+
},
|
| 4452 |
+
{
|
| 4453 |
+
"epoch": 16.107274063188832,
|
| 4454 |
+
"grad_norm": 0.7046474814414978,
|
| 4455 |
+
"learning_rate": 0.00013967605356439541,
|
| 4456 |
+
"loss": 0.1785,
|
| 4457 |
+
"step": 54800
|
| 4458 |
+
},
|
| 4459 |
+
{
|
| 4460 |
+
"epoch": 16.13666421748714,
|
| 4461 |
+
"grad_norm": 0.4531545042991638,
|
| 4462 |
+
"learning_rate": 0.0001393806616778259,
|
| 4463 |
+
"loss": 0.1809,
|
| 4464 |
+
"step": 54900
|
| 4465 |
+
},
|
| 4466 |
+
{
|
| 4467 |
+
"epoch": 16.166054371785453,
|
| 4468 |
+
"grad_norm": 0.5998923182487488,
|
| 4469 |
+
"learning_rate": 0.0001390852697912564,
|
| 4470 |
+
"loss": 0.191,
|
| 4471 |
+
"step": 55000
|
| 4472 |
+
},
|
| 4473 |
+
{
|
| 4474 |
+
"epoch": 16.19544452608376,
|
| 4475 |
+
"grad_norm": 0.7027255892753601,
|
| 4476 |
+
"learning_rate": 0.00013878987790468688,
|
| 4477 |
+
"loss": 0.1909,
|
| 4478 |
+
"step": 55100
|
| 4479 |
+
},
|
| 4480 |
+
{
|
| 4481 |
+
"epoch": 16.224834680382074,
|
| 4482 |
+
"grad_norm": 1.4335699081420898,
|
| 4483 |
+
"learning_rate": 0.00013849448601811735,
|
| 4484 |
+
"loss": 0.1786,
|
| 4485 |
+
"step": 55200
|
| 4486 |
+
},
|
| 4487 |
+
{
|
| 4488 |
+
"epoch": 16.224834680382074,
|
| 4489 |
+
"eval_loss": 0.2970651388168335,
|
| 4490 |
+
"eval_runtime": 313.9408,
|
| 4491 |
+
"eval_samples_per_second": 43.349,
|
| 4492 |
+
"eval_steps_per_second": 5.421,
|
| 4493 |
+
"eval_wer": 0.2892803931306689,
|
| 4494 |
+
"step": 55200
|
| 4495 |
}
|
| 4496 |
],
|
| 4497 |
"logging_steps": 100,
|
|
|
|
| 4511 |
"attributes": {}
|
| 4512 |
}
|
| 4513 |
},
|
| 4514 |
+
"total_flos": 1.7466142485651987e+20,
|
| 4515 |
"train_batch_size": 16,
|
| 4516 |
"trial_name": null,
|
| 4517 |
"trial_params": null
|