Commit
•
d4e67c9
1
Parent(s):
3b4b27c
Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +2905 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,2905 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 43.10069742838263,
|
| 3 |
+
"best_model_checkpoint": "./whisper-distil-v3/checkpoint-8000",
|
| 4 |
+
"epoch": 0.43838018521562827,
|
| 5 |
+
"eval_steps": 1000,
|
| 6 |
+
"global_step": 8000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.0010959504630390707,
|
| 13 |
+
"grad_norm": 5.807405471801758,
|
| 14 |
+
"learning_rate": 3.6e-06,
|
| 15 |
+
"loss": 8.6037,
|
| 16 |
+
"step": 20
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.0021919009260781414,
|
| 20 |
+
"grad_norm": 5.726860523223877,
|
| 21 |
+
"learning_rate": 7.6e-06,
|
| 22 |
+
"loss": 8.4874,
|
| 23 |
+
"step": 40
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.003287851389117212,
|
| 27 |
+
"grad_norm": 7.615314960479736,
|
| 28 |
+
"learning_rate": 1.16e-05,
|
| 29 |
+
"loss": 8.0934,
|
| 30 |
+
"step": 60
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.004383801852156283,
|
| 34 |
+
"grad_norm": 7.089465618133545,
|
| 35 |
+
"learning_rate": 1.56e-05,
|
| 36 |
+
"loss": 7.4227,
|
| 37 |
+
"step": 80
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.005479752315195353,
|
| 41 |
+
"grad_norm": 5.158086776733398,
|
| 42 |
+
"learning_rate": 1.9600000000000002e-05,
|
| 43 |
+
"loss": 6.3418,
|
| 44 |
+
"step": 100
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.006575702778234424,
|
| 48 |
+
"grad_norm": 3.288583278656006,
|
| 49 |
+
"learning_rate": 2.3400000000000003e-05,
|
| 50 |
+
"loss": 5.2747,
|
| 51 |
+
"step": 120
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.007671653241273495,
|
| 55 |
+
"grad_norm": 3.1715681552886963,
|
| 56 |
+
"learning_rate": 2.7400000000000002e-05,
|
| 57 |
+
"loss": 4.3075,
|
| 58 |
+
"step": 140
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.008767603704312565,
|
| 62 |
+
"grad_norm": 3.033198833465576,
|
| 63 |
+
"learning_rate": 3.1400000000000004e-05,
|
| 64 |
+
"loss": 3.631,
|
| 65 |
+
"step": 160
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.009863554167351636,
|
| 69 |
+
"grad_norm": 3.027251720428467,
|
| 70 |
+
"learning_rate": 3.54e-05,
|
| 71 |
+
"loss": 3.2186,
|
| 72 |
+
"step": 180
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.010959504630390707,
|
| 76 |
+
"grad_norm": 2.9063901901245117,
|
| 77 |
+
"learning_rate": 3.94e-05,
|
| 78 |
+
"loss": 2.9226,
|
| 79 |
+
"step": 200
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.012055455093429777,
|
| 83 |
+
"grad_norm": 3.1165690422058105,
|
| 84 |
+
"learning_rate": 4.3400000000000005e-05,
|
| 85 |
+
"loss": 2.8402,
|
| 86 |
+
"step": 220
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.013151405556468848,
|
| 90 |
+
"grad_norm": 2.7977383136749268,
|
| 91 |
+
"learning_rate": 4.74e-05,
|
| 92 |
+
"loss": 2.613,
|
| 93 |
+
"step": 240
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.014247356019507919,
|
| 97 |
+
"grad_norm": 3.7818286418914795,
|
| 98 |
+
"learning_rate": 5.14e-05,
|
| 99 |
+
"loss": 2.377,
|
| 100 |
+
"step": 260
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.01534330648254699,
|
| 104 |
+
"grad_norm": 3.2088804244995117,
|
| 105 |
+
"learning_rate": 5.5400000000000005e-05,
|
| 106 |
+
"loss": 2.3204,
|
| 107 |
+
"step": 280
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.01643925694558606,
|
| 111 |
+
"grad_norm": 3.2518157958984375,
|
| 112 |
+
"learning_rate": 5.94e-05,
|
| 113 |
+
"loss": 2.1812,
|
| 114 |
+
"step": 300
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.01753520740862513,
|
| 118 |
+
"grad_norm": 3.725226640701294,
|
| 119 |
+
"learning_rate": 6.340000000000001e-05,
|
| 120 |
+
"loss": 2.158,
|
| 121 |
+
"step": 320
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.0186311578716642,
|
| 125 |
+
"grad_norm": 3.5929486751556396,
|
| 126 |
+
"learning_rate": 6.740000000000001e-05,
|
| 127 |
+
"loss": 2.1241,
|
| 128 |
+
"step": 340
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.019727108334703272,
|
| 132 |
+
"grad_norm": 4.1317572593688965,
|
| 133 |
+
"learning_rate": 7.14e-05,
|
| 134 |
+
"loss": 2.2284,
|
| 135 |
+
"step": 360
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.020823058797742343,
|
| 139 |
+
"grad_norm": 3.4276161193847656,
|
| 140 |
+
"learning_rate": 7.54e-05,
|
| 141 |
+
"loss": 1.9655,
|
| 142 |
+
"step": 380
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.021919009260781414,
|
| 146 |
+
"grad_norm": 3.9775540828704834,
|
| 147 |
+
"learning_rate": 7.94e-05,
|
| 148 |
+
"loss": 1.9407,
|
| 149 |
+
"step": 400
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.023014959723820484,
|
| 153 |
+
"grad_norm": 3.67511248588562,
|
| 154 |
+
"learning_rate": 8.34e-05,
|
| 155 |
+
"loss": 1.9312,
|
| 156 |
+
"step": 420
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.024110910186859555,
|
| 160 |
+
"grad_norm": 4.781565189361572,
|
| 161 |
+
"learning_rate": 8.740000000000001e-05,
|
| 162 |
+
"loss": 1.9218,
|
| 163 |
+
"step": 440
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.025206860649898626,
|
| 167 |
+
"grad_norm": 5.2797698974609375,
|
| 168 |
+
"learning_rate": 9.140000000000001e-05,
|
| 169 |
+
"loss": 1.8729,
|
| 170 |
+
"step": 460
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.026302811112937696,
|
| 174 |
+
"grad_norm": 6.1737284660339355,
|
| 175 |
+
"learning_rate": 9.54e-05,
|
| 176 |
+
"loss": 1.6848,
|
| 177 |
+
"step": 480
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.027398761575976767,
|
| 181 |
+
"grad_norm": 4.926702976226807,
|
| 182 |
+
"learning_rate": 9.94e-05,
|
| 183 |
+
"loss": 1.8866,
|
| 184 |
+
"step": 500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.028494712039015838,
|
| 188 |
+
"grad_norm": 4.043098449707031,
|
| 189 |
+
"learning_rate": 0.0001,
|
| 190 |
+
"loss": 1.749,
|
| 191 |
+
"step": 520
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.02959066250205491,
|
| 195 |
+
"grad_norm": 4.022521495819092,
|
| 196 |
+
"learning_rate": 0.0001,
|
| 197 |
+
"loss": 1.7654,
|
| 198 |
+
"step": 540
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.03068661296509398,
|
| 202 |
+
"grad_norm": 3.1964547634124756,
|
| 203 |
+
"learning_rate": 0.0001,
|
| 204 |
+
"loss": 1.7496,
|
| 205 |
+
"step": 560
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.03178256342813305,
|
| 209 |
+
"grad_norm": 3.5182583332061768,
|
| 210 |
+
"learning_rate": 0.0001,
|
| 211 |
+
"loss": 1.7312,
|
| 212 |
+
"step": 580
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.03287851389117212,
|
| 216 |
+
"grad_norm": 3.529665231704712,
|
| 217 |
+
"learning_rate": 0.0001,
|
| 218 |
+
"loss": 1.6307,
|
| 219 |
+
"step": 600
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.03397446435421119,
|
| 223 |
+
"grad_norm": 3.329401969909668,
|
| 224 |
+
"learning_rate": 0.0001,
|
| 225 |
+
"loss": 1.7613,
|
| 226 |
+
"step": 620
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.03507041481725026,
|
| 230 |
+
"grad_norm": 3.4481399059295654,
|
| 231 |
+
"learning_rate": 0.0001,
|
| 232 |
+
"loss": 1.6204,
|
| 233 |
+
"step": 640
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.03616636528028933,
|
| 237 |
+
"grad_norm": 3.3551902770996094,
|
| 238 |
+
"learning_rate": 0.0001,
|
| 239 |
+
"loss": 1.5846,
|
| 240 |
+
"step": 660
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.0372623157433284,
|
| 244 |
+
"grad_norm": 3.591031074523926,
|
| 245 |
+
"learning_rate": 0.0001,
|
| 246 |
+
"loss": 1.6077,
|
| 247 |
+
"step": 680
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.038358266206367474,
|
| 251 |
+
"grad_norm": 3.8630764484405518,
|
| 252 |
+
"learning_rate": 0.0001,
|
| 253 |
+
"loss": 1.5275,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.039454216669406544,
|
| 258 |
+
"grad_norm": 3.77461838722229,
|
| 259 |
+
"learning_rate": 0.0001,
|
| 260 |
+
"loss": 1.5386,
|
| 261 |
+
"step": 720
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.040550167132445615,
|
| 265 |
+
"grad_norm": 2.9158153533935547,
|
| 266 |
+
"learning_rate": 0.0001,
|
| 267 |
+
"loss": 1.5536,
|
| 268 |
+
"step": 740
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.041646117595484686,
|
| 272 |
+
"grad_norm": 3.761077642440796,
|
| 273 |
+
"learning_rate": 0.0001,
|
| 274 |
+
"loss": 1.5607,
|
| 275 |
+
"step": 760
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.042742068058523756,
|
| 279 |
+
"grad_norm": 3.5758230686187744,
|
| 280 |
+
"learning_rate": 0.0001,
|
| 281 |
+
"loss": 1.5145,
|
| 282 |
+
"step": 780
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.04383801852156283,
|
| 286 |
+
"grad_norm": 3.01175856590271,
|
| 287 |
+
"learning_rate": 0.0001,
|
| 288 |
+
"loss": 1.5639,
|
| 289 |
+
"step": 800
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.0449339689846019,
|
| 293 |
+
"grad_norm": 3.8395230770111084,
|
| 294 |
+
"learning_rate": 0.0001,
|
| 295 |
+
"loss": 1.6478,
|
| 296 |
+
"step": 820
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.04602991944764097,
|
| 300 |
+
"grad_norm": 2.9240541458129883,
|
| 301 |
+
"learning_rate": 0.0001,
|
| 302 |
+
"loss": 1.5303,
|
| 303 |
+
"step": 840
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.04712586991068004,
|
| 307 |
+
"grad_norm": 3.603835344314575,
|
| 308 |
+
"learning_rate": 0.0001,
|
| 309 |
+
"loss": 1.4436,
|
| 310 |
+
"step": 860
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.04822182037371911,
|
| 314 |
+
"grad_norm": 3.1701183319091797,
|
| 315 |
+
"learning_rate": 0.0001,
|
| 316 |
+
"loss": 1.5622,
|
| 317 |
+
"step": 880
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.04931777083675818,
|
| 321 |
+
"grad_norm": 4.054835796356201,
|
| 322 |
+
"learning_rate": 0.0001,
|
| 323 |
+
"loss": 1.4354,
|
| 324 |
+
"step": 900
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.05041372129979725,
|
| 328 |
+
"grad_norm": 2.9651615619659424,
|
| 329 |
+
"learning_rate": 0.0001,
|
| 330 |
+
"loss": 1.4676,
|
| 331 |
+
"step": 920
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.05150967176283632,
|
| 335 |
+
"grad_norm": 3.2480218410491943,
|
| 336 |
+
"learning_rate": 0.0001,
|
| 337 |
+
"loss": 1.3769,
|
| 338 |
+
"step": 940
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.05260562222587539,
|
| 342 |
+
"grad_norm": 3.494356155395508,
|
| 343 |
+
"learning_rate": 0.0001,
|
| 344 |
+
"loss": 1.4928,
|
| 345 |
+
"step": 960
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.05370157268891446,
|
| 349 |
+
"grad_norm": 3.394205331802368,
|
| 350 |
+
"learning_rate": 0.0001,
|
| 351 |
+
"loss": 1.4045,
|
| 352 |
+
"step": 980
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.054797523151953534,
|
| 356 |
+
"grad_norm": 3.333587646484375,
|
| 357 |
+
"learning_rate": 0.0001,
|
| 358 |
+
"loss": 1.4953,
|
| 359 |
+
"step": 1000
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 0.054797523151953534,
|
| 363 |
+
"eval_loss": 1.4313914775848389,
|
| 364 |
+
"eval_runtime": 30911.9498,
|
| 365 |
+
"eval_samples_per_second": 2.099,
|
| 366 |
+
"eval_steps_per_second": 0.066,
|
| 367 |
+
"eval_wer": 70.88587442180551,
|
| 368 |
+
"step": 1000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.055893473614992605,
|
| 372 |
+
"grad_norm": 3.2317609786987305,
|
| 373 |
+
"learning_rate": 0.0001,
|
| 374 |
+
"loss": 1.4376,
|
| 375 |
+
"step": 1020
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.056989424078031675,
|
| 379 |
+
"grad_norm": 2.9077706336975098,
|
| 380 |
+
"learning_rate": 0.0001,
|
| 381 |
+
"loss": 1.4398,
|
| 382 |
+
"step": 1040
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.058085374541070746,
|
| 386 |
+
"grad_norm": 3.0054707527160645,
|
| 387 |
+
"learning_rate": 0.0001,
|
| 388 |
+
"loss": 1.4326,
|
| 389 |
+
"step": 1060
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 0.05918132500410982,
|
| 393 |
+
"grad_norm": 3.7243480682373047,
|
| 394 |
+
"learning_rate": 0.0001,
|
| 395 |
+
"loss": 1.4915,
|
| 396 |
+
"step": 1080
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.06027727546714889,
|
| 400 |
+
"grad_norm": 2.9608402252197266,
|
| 401 |
+
"learning_rate": 0.0001,
|
| 402 |
+
"loss": 1.4595,
|
| 403 |
+
"step": 1100
|
| 404 |
+
},
|
| 405 |
+
{
|
| 406 |
+
"epoch": 0.06137322593018796,
|
| 407 |
+
"grad_norm": 2.652829885482788,
|
| 408 |
+
"learning_rate": 0.0001,
|
| 409 |
+
"loss": 1.4378,
|
| 410 |
+
"step": 1120
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.06246917639322703,
|
| 414 |
+
"grad_norm": 2.9017295837402344,
|
| 415 |
+
"learning_rate": 0.0001,
|
| 416 |
+
"loss": 1.4257,
|
| 417 |
+
"step": 1140
|
| 418 |
+
},
|
| 419 |
+
{
|
| 420 |
+
"epoch": 0.0635651268562661,
|
| 421 |
+
"grad_norm": 3.0610859394073486,
|
| 422 |
+
"learning_rate": 0.0001,
|
| 423 |
+
"loss": 1.348,
|
| 424 |
+
"step": 1160
|
| 425 |
+
},
|
| 426 |
+
{
|
| 427 |
+
"epoch": 0.06466107731930516,
|
| 428 |
+
"grad_norm": 2.701765775680542,
|
| 429 |
+
"learning_rate": 0.0001,
|
| 430 |
+
"loss": 1.3853,
|
| 431 |
+
"step": 1180
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.06575702778234424,
|
| 435 |
+
"grad_norm": 3.83376145362854,
|
| 436 |
+
"learning_rate": 0.0001,
|
| 437 |
+
"loss": 1.4708,
|
| 438 |
+
"step": 1200
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.0668529782453833,
|
| 442 |
+
"grad_norm": 3.178449869155884,
|
| 443 |
+
"learning_rate": 0.0001,
|
| 444 |
+
"loss": 1.362,
|
| 445 |
+
"step": 1220
|
| 446 |
+
},
|
| 447 |
+
{
|
| 448 |
+
"epoch": 0.06794892870842238,
|
| 449 |
+
"grad_norm": 3.796205997467041,
|
| 450 |
+
"learning_rate": 0.0001,
|
| 451 |
+
"loss": 1.4331,
|
| 452 |
+
"step": 1240
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.06904487917146145,
|
| 456 |
+
"grad_norm": 2.8163928985595703,
|
| 457 |
+
"learning_rate": 0.0001,
|
| 458 |
+
"loss": 1.2835,
|
| 459 |
+
"step": 1260
|
| 460 |
+
},
|
| 461 |
+
{
|
| 462 |
+
"epoch": 0.07014082963450052,
|
| 463 |
+
"grad_norm": 2.698793888092041,
|
| 464 |
+
"learning_rate": 0.0001,
|
| 465 |
+
"loss": 1.3444,
|
| 466 |
+
"step": 1280
|
| 467 |
+
},
|
| 468 |
+
{
|
| 469 |
+
"epoch": 0.07123678009753959,
|
| 470 |
+
"grad_norm": 2.584484815597534,
|
| 471 |
+
"learning_rate": 0.0001,
|
| 472 |
+
"loss": 1.2145,
|
| 473 |
+
"step": 1300
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.07233273056057866,
|
| 477 |
+
"grad_norm": 2.696967363357544,
|
| 478 |
+
"learning_rate": 0.0001,
|
| 479 |
+
"loss": 1.2855,
|
| 480 |
+
"step": 1320
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.07342868102361773,
|
| 484 |
+
"grad_norm": 3.382924795150757,
|
| 485 |
+
"learning_rate": 0.0001,
|
| 486 |
+
"loss": 1.2164,
|
| 487 |
+
"step": 1340
|
| 488 |
+
},
|
| 489 |
+
{
|
| 490 |
+
"epoch": 0.0745246314866568,
|
| 491 |
+
"grad_norm": 2.8127260208129883,
|
| 492 |
+
"learning_rate": 0.0001,
|
| 493 |
+
"loss": 1.2873,
|
| 494 |
+
"step": 1360
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.07562058194969587,
|
| 498 |
+
"grad_norm": 2.631011724472046,
|
| 499 |
+
"learning_rate": 0.0001,
|
| 500 |
+
"loss": 1.3759,
|
| 501 |
+
"step": 1380
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.07671653241273495,
|
| 505 |
+
"grad_norm": 2.913276433944702,
|
| 506 |
+
"learning_rate": 0.0001,
|
| 507 |
+
"loss": 1.2688,
|
| 508 |
+
"step": 1400
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.07781248287577401,
|
| 512 |
+
"grad_norm": 2.811455488204956,
|
| 513 |
+
"learning_rate": 0.0001,
|
| 514 |
+
"loss": 1.2179,
|
| 515 |
+
"step": 1420
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.07890843333881309,
|
| 519 |
+
"grad_norm": 2.8242247104644775,
|
| 520 |
+
"learning_rate": 0.0001,
|
| 521 |
+
"loss": 1.142,
|
| 522 |
+
"step": 1440
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 0.08000438380185215,
|
| 526 |
+
"grad_norm": 3.1733341217041016,
|
| 527 |
+
"learning_rate": 0.0001,
|
| 528 |
+
"loss": 1.2934,
|
| 529 |
+
"step": 1460
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 0.08110033426489123,
|
| 533 |
+
"grad_norm": 2.491945743560791,
|
| 534 |
+
"learning_rate": 0.0001,
|
| 535 |
+
"loss": 1.3274,
|
| 536 |
+
"step": 1480
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.0821962847279303,
|
| 540 |
+
"grad_norm": 2.717165470123291,
|
| 541 |
+
"learning_rate": 0.0001,
|
| 542 |
+
"loss": 1.2484,
|
| 543 |
+
"step": 1500
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.08329223519096937,
|
| 547 |
+
"grad_norm": 2.3187918663024902,
|
| 548 |
+
"learning_rate": 0.0001,
|
| 549 |
+
"loss": 1.2038,
|
| 550 |
+
"step": 1520
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 0.08438818565400844,
|
| 554 |
+
"grad_norm": 2.9296529293060303,
|
| 555 |
+
"learning_rate": 0.0001,
|
| 556 |
+
"loss": 1.1962,
|
| 557 |
+
"step": 1540
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 0.08548413611704751,
|
| 561 |
+
"grad_norm": 2.5763330459594727,
|
| 562 |
+
"learning_rate": 0.0001,
|
| 563 |
+
"loss": 1.2122,
|
| 564 |
+
"step": 1560
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 0.08658008658008658,
|
| 568 |
+
"grad_norm": 3.4159390926361084,
|
| 569 |
+
"learning_rate": 0.0001,
|
| 570 |
+
"loss": 1.2302,
|
| 571 |
+
"step": 1580
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.08767603704312565,
|
| 575 |
+
"grad_norm": 2.893261432647705,
|
| 576 |
+
"learning_rate": 0.0001,
|
| 577 |
+
"loss": 1.2106,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.08877198750616472,
|
| 582 |
+
"grad_norm": 2.1891727447509766,
|
| 583 |
+
"learning_rate": 0.0001,
|
| 584 |
+
"loss": 1.2282,
|
| 585 |
+
"step": 1620
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.0898679379692038,
|
| 589 |
+
"grad_norm": 2.4100029468536377,
|
| 590 |
+
"learning_rate": 0.0001,
|
| 591 |
+
"loss": 1.2039,
|
| 592 |
+
"step": 1640
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 0.09096388843224286,
|
| 596 |
+
"grad_norm": 2.5420494079589844,
|
| 597 |
+
"learning_rate": 0.0001,
|
| 598 |
+
"loss": 1.2201,
|
| 599 |
+
"step": 1660
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.09205983889528194,
|
| 603 |
+
"grad_norm": 3.1885313987731934,
|
| 604 |
+
"learning_rate": 0.0001,
|
| 605 |
+
"loss": 1.2446,
|
| 606 |
+
"step": 1680
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 0.093155789358321,
|
| 610 |
+
"grad_norm": 3.120586633682251,
|
| 611 |
+
"learning_rate": 0.0001,
|
| 612 |
+
"loss": 1.2308,
|
| 613 |
+
"step": 1700
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 0.09425173982136008,
|
| 617 |
+
"grad_norm": 2.4548628330230713,
|
| 618 |
+
"learning_rate": 0.0001,
|
| 619 |
+
"loss": 1.1777,
|
| 620 |
+
"step": 1720
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.09534769028439914,
|
| 624 |
+
"grad_norm": 3.101803779602051,
|
| 625 |
+
"learning_rate": 0.0001,
|
| 626 |
+
"loss": 1.2123,
|
| 627 |
+
"step": 1740
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.09644364074743822,
|
| 631 |
+
"grad_norm": 2.536121368408203,
|
| 632 |
+
"learning_rate": 0.0001,
|
| 633 |
+
"loss": 1.1914,
|
| 634 |
+
"step": 1760
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.09753959121047728,
|
| 638 |
+
"grad_norm": 2.3796801567077637,
|
| 639 |
+
"learning_rate": 0.0001,
|
| 640 |
+
"loss": 1.1848,
|
| 641 |
+
"step": 1780
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.09863554167351636,
|
| 645 |
+
"grad_norm": 2.67964243888855,
|
| 646 |
+
"learning_rate": 0.0001,
|
| 647 |
+
"loss": 1.1973,
|
| 648 |
+
"step": 1800
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.09973149213655542,
|
| 652 |
+
"grad_norm": 3.160212755203247,
|
| 653 |
+
"learning_rate": 0.0001,
|
| 654 |
+
"loss": 1.2472,
|
| 655 |
+
"step": 1820
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.1008274425995945,
|
| 659 |
+
"grad_norm": 2.7035927772521973,
|
| 660 |
+
"learning_rate": 0.0001,
|
| 661 |
+
"loss": 1.1844,
|
| 662 |
+
"step": 1840
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.10192339306263357,
|
| 666 |
+
"grad_norm": 2.7725090980529785,
|
| 667 |
+
"learning_rate": 0.0001,
|
| 668 |
+
"loss": 1.1262,
|
| 669 |
+
"step": 1860
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.10301934352567264,
|
| 673 |
+
"grad_norm": 2.2705016136169434,
|
| 674 |
+
"learning_rate": 0.0001,
|
| 675 |
+
"loss": 1.182,
|
| 676 |
+
"step": 1880
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.10411529398871171,
|
| 680 |
+
"grad_norm": 3.0717403888702393,
|
| 681 |
+
"learning_rate": 0.0001,
|
| 682 |
+
"loss": 1.137,
|
| 683 |
+
"step": 1900
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 0.10521124445175078,
|
| 687 |
+
"grad_norm": 2.9270904064178467,
|
| 688 |
+
"learning_rate": 0.0001,
|
| 689 |
+
"loss": 1.2556,
|
| 690 |
+
"step": 1920
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.10630719491478985,
|
| 694 |
+
"grad_norm": 2.4564895629882812,
|
| 695 |
+
"learning_rate": 0.0001,
|
| 696 |
+
"loss": 1.1812,
|
| 697 |
+
"step": 1940
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.10740314537782893,
|
| 701 |
+
"grad_norm": 2.983851909637451,
|
| 702 |
+
"learning_rate": 0.0001,
|
| 703 |
+
"loss": 1.1445,
|
| 704 |
+
"step": 1960
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.10849909584086799,
|
| 708 |
+
"grad_norm": 2.772733688354492,
|
| 709 |
+
"learning_rate": 0.0001,
|
| 710 |
+
"loss": 1.1968,
|
| 711 |
+
"step": 1980
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.10959504630390707,
|
| 715 |
+
"grad_norm": 2.9768126010894775,
|
| 716 |
+
"learning_rate": 0.0001,
|
| 717 |
+
"loss": 1.0942,
|
| 718 |
+
"step": 2000
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.10959504630390707,
|
| 722 |
+
"eval_loss": 1.1446514129638672,
|
| 723 |
+
"eval_runtime": 30634.8587,
|
| 724 |
+
"eval_samples_per_second": 2.118,
|
| 725 |
+
"eval_steps_per_second": 0.066,
|
| 726 |
+
"eval_wer": 61.28519240053001,
|
| 727 |
+
"step": 2000
|
| 728 |
+
},
|
| 729 |
+
{
|
| 730 |
+
"epoch": 0.11069099676694613,
|
| 731 |
+
"grad_norm": 2.806312322616577,
|
| 732 |
+
"learning_rate": 0.0001,
|
| 733 |
+
"loss": 1.1924,
|
| 734 |
+
"step": 2020
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.11178694722998521,
|
| 738 |
+
"grad_norm": 2.639443874359131,
|
| 739 |
+
"learning_rate": 0.0001,
|
| 740 |
+
"loss": 1.0572,
|
| 741 |
+
"step": 2040
|
| 742 |
+
},
|
| 743 |
+
{
|
| 744 |
+
"epoch": 0.11288289769302427,
|
| 745 |
+
"grad_norm": 2.2005367279052734,
|
| 746 |
+
"learning_rate": 0.0001,
|
| 747 |
+
"loss": 1.1337,
|
| 748 |
+
"step": 2060
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 0.11397884815606335,
|
| 752 |
+
"grad_norm": 2.4102020263671875,
|
| 753 |
+
"learning_rate": 0.0001,
|
| 754 |
+
"loss": 1.1297,
|
| 755 |
+
"step": 2080
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 0.11507479861910241,
|
| 759 |
+
"grad_norm": 3.410691976547241,
|
| 760 |
+
"learning_rate": 0.0001,
|
| 761 |
+
"loss": 1.1354,
|
| 762 |
+
"step": 2100
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.11617074908214149,
|
| 766 |
+
"grad_norm": 2.1337172985076904,
|
| 767 |
+
"learning_rate": 0.0001,
|
| 768 |
+
"loss": 1.1725,
|
| 769 |
+
"step": 2120
|
| 770 |
+
},
|
| 771 |
+
{
|
| 772 |
+
"epoch": 0.11726669954518056,
|
| 773 |
+
"grad_norm": 2.627319097518921,
|
| 774 |
+
"learning_rate": 0.0001,
|
| 775 |
+
"loss": 1.1006,
|
| 776 |
+
"step": 2140
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.11836265000821963,
|
| 780 |
+
"grad_norm": 2.6450726985931396,
|
| 781 |
+
"learning_rate": 0.0001,
|
| 782 |
+
"loss": 1.0985,
|
| 783 |
+
"step": 2160
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 0.1194586004712587,
|
| 787 |
+
"grad_norm": 2.3205084800720215,
|
| 788 |
+
"learning_rate": 0.0001,
|
| 789 |
+
"loss": 1.1634,
|
| 790 |
+
"step": 2180
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 0.12055455093429777,
|
| 794 |
+
"grad_norm": 2.51177978515625,
|
| 795 |
+
"learning_rate": 0.0001,
|
| 796 |
+
"loss": 1.1697,
|
| 797 |
+
"step": 2200
|
| 798 |
+
},
|
| 799 |
+
{
|
| 800 |
+
"epoch": 0.12165050139733684,
|
| 801 |
+
"grad_norm": 2.6632323265075684,
|
| 802 |
+
"learning_rate": 0.0001,
|
| 803 |
+
"loss": 1.071,
|
| 804 |
+
"step": 2220
|
| 805 |
+
},
|
| 806 |
+
{
|
| 807 |
+
"epoch": 0.12274645186037592,
|
| 808 |
+
"grad_norm": 2.8322274684906006,
|
| 809 |
+
"learning_rate": 0.0001,
|
| 810 |
+
"loss": 1.0983,
|
| 811 |
+
"step": 2240
|
| 812 |
+
},
|
| 813 |
+
{
|
| 814 |
+
"epoch": 0.12384240232341498,
|
| 815 |
+
"grad_norm": 2.547708749771118,
|
| 816 |
+
"learning_rate": 0.0001,
|
| 817 |
+
"loss": 1.0629,
|
| 818 |
+
"step": 2260
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.12493835278645406,
|
| 822 |
+
"grad_norm": 2.6638150215148926,
|
| 823 |
+
"learning_rate": 0.0001,
|
| 824 |
+
"loss": 1.1985,
|
| 825 |
+
"step": 2280
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 0.12603430324949313,
|
| 829 |
+
"grad_norm": 2.980463743209839,
|
| 830 |
+
"learning_rate": 0.0001,
|
| 831 |
+
"loss": 1.1885,
|
| 832 |
+
"step": 2300
|
| 833 |
+
},
|
| 834 |
+
{
|
| 835 |
+
"epoch": 0.1271302537125322,
|
| 836 |
+
"grad_norm": 1.9924368858337402,
|
| 837 |
+
"learning_rate": 0.0001,
|
| 838 |
+
"loss": 1.0971,
|
| 839 |
+
"step": 2320
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 0.12822620417557126,
|
| 843 |
+
"grad_norm": 2.2847180366516113,
|
| 844 |
+
"learning_rate": 0.0001,
|
| 845 |
+
"loss": 1.1149,
|
| 846 |
+
"step": 2340
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 0.12932215463861033,
|
| 850 |
+
"grad_norm": 2.4860479831695557,
|
| 851 |
+
"learning_rate": 0.0001,
|
| 852 |
+
"loss": 1.0927,
|
| 853 |
+
"step": 2360
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 0.13041810510164942,
|
| 857 |
+
"grad_norm": 2.3988494873046875,
|
| 858 |
+
"learning_rate": 0.0001,
|
| 859 |
+
"loss": 1.1918,
|
| 860 |
+
"step": 2380
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.13151405556468848,
|
| 864 |
+
"grad_norm": 2.5361902713775635,
|
| 865 |
+
"learning_rate": 0.0001,
|
| 866 |
+
"loss": 1.0603,
|
| 867 |
+
"step": 2400
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 0.13261000602772754,
|
| 871 |
+
"grad_norm": 2.4060215950012207,
|
| 872 |
+
"learning_rate": 0.0001,
|
| 873 |
+
"loss": 1.056,
|
| 874 |
+
"step": 2420
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 0.1337059564907666,
|
| 878 |
+
"grad_norm": 2.4094231128692627,
|
| 879 |
+
"learning_rate": 0.0001,
|
| 880 |
+
"loss": 1.0787,
|
| 881 |
+
"step": 2440
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.1348019069538057,
|
| 885 |
+
"grad_norm": 2.5207912921905518,
|
| 886 |
+
"learning_rate": 0.0001,
|
| 887 |
+
"loss": 1.0901,
|
| 888 |
+
"step": 2460
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 0.13589785741684476,
|
| 892 |
+
"grad_norm": 2.1340293884277344,
|
| 893 |
+
"learning_rate": 0.0001,
|
| 894 |
+
"loss": 1.1691,
|
| 895 |
+
"step": 2480
|
| 896 |
+
},
|
| 897 |
+
{
|
| 898 |
+
"epoch": 0.13699380787988383,
|
| 899 |
+
"grad_norm": 2.312554359436035,
|
| 900 |
+
"learning_rate": 0.0001,
|
| 901 |
+
"loss": 0.9791,
|
| 902 |
+
"step": 2500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.1380897583429229,
|
| 906 |
+
"grad_norm": 2.2881298065185547,
|
| 907 |
+
"learning_rate": 0.0001,
|
| 908 |
+
"loss": 0.9998,
|
| 909 |
+
"step": 2520
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.13918570880596198,
|
| 913 |
+
"grad_norm": 2.2146573066711426,
|
| 914 |
+
"learning_rate": 0.0001,
|
| 915 |
+
"loss": 1.094,
|
| 916 |
+
"step": 2540
|
| 917 |
+
},
|
| 918 |
+
{
|
| 919 |
+
"epoch": 0.14028165926900105,
|
| 920 |
+
"grad_norm": 2.3992650508880615,
|
| 921 |
+
"learning_rate": 0.0001,
|
| 922 |
+
"loss": 1.0667,
|
| 923 |
+
"step": 2560
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 0.1413776097320401,
|
| 927 |
+
"grad_norm": 2.7630209922790527,
|
| 928 |
+
"learning_rate": 0.0001,
|
| 929 |
+
"loss": 1.1541,
|
| 930 |
+
"step": 2580
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 0.14247356019507917,
|
| 934 |
+
"grad_norm": 2.9216675758361816,
|
| 935 |
+
"learning_rate": 0.0001,
|
| 936 |
+
"loss": 1.0463,
|
| 937 |
+
"step": 2600
|
| 938 |
+
},
|
| 939 |
+
{
|
| 940 |
+
"epoch": 0.14356951065811827,
|
| 941 |
+
"grad_norm": 2.366373062133789,
|
| 942 |
+
"learning_rate": 0.0001,
|
| 943 |
+
"loss": 1.0557,
|
| 944 |
+
"step": 2620
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.14466546112115733,
|
| 948 |
+
"grad_norm": 2.7161865234375,
|
| 949 |
+
"learning_rate": 0.0001,
|
| 950 |
+
"loss": 1.1066,
|
| 951 |
+
"step": 2640
|
| 952 |
+
},
|
| 953 |
+
{
|
| 954 |
+
"epoch": 0.1457614115841964,
|
| 955 |
+
"grad_norm": 2.046992778778076,
|
| 956 |
+
"learning_rate": 0.0001,
|
| 957 |
+
"loss": 0.9786,
|
| 958 |
+
"step": 2660
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 0.14685736204723546,
|
| 962 |
+
"grad_norm": 2.6320793628692627,
|
| 963 |
+
"learning_rate": 0.0001,
|
| 964 |
+
"loss": 0.9564,
|
| 965 |
+
"step": 2680
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 0.14795331251027455,
|
| 969 |
+
"grad_norm": 2.485445737838745,
|
| 970 |
+
"learning_rate": 0.0001,
|
| 971 |
+
"loss": 1.0283,
|
| 972 |
+
"step": 2700
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 0.1490492629733136,
|
| 976 |
+
"grad_norm": 2.267420768737793,
|
| 977 |
+
"learning_rate": 0.0001,
|
| 978 |
+
"loss": 1.0092,
|
| 979 |
+
"step": 2720
|
| 980 |
+
},
|
| 981 |
+
{
|
| 982 |
+
"epoch": 0.15014521343635268,
|
| 983 |
+
"grad_norm": 2.618067502975464,
|
| 984 |
+
"learning_rate": 0.0001,
|
| 985 |
+
"loss": 1.0369,
|
| 986 |
+
"step": 2740
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.15124116389939174,
|
| 990 |
+
"grad_norm": 2.502471685409546,
|
| 991 |
+
"learning_rate": 0.0001,
|
| 992 |
+
"loss": 0.9982,
|
| 993 |
+
"step": 2760
|
| 994 |
+
},
|
| 995 |
+
{
|
| 996 |
+
"epoch": 0.15233711436243083,
|
| 997 |
+
"grad_norm": 2.936964273452759,
|
| 998 |
+
"learning_rate": 0.0001,
|
| 999 |
+
"loss": 1.1122,
|
| 1000 |
+
"step": 2780
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 0.1534330648254699,
|
| 1004 |
+
"grad_norm": 2.5342159271240234,
|
| 1005 |
+
"learning_rate": 0.0001,
|
| 1006 |
+
"loss": 1.0409,
|
| 1007 |
+
"step": 2800
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 0.15452901528850896,
|
| 1011 |
+
"grad_norm": 2.88598895072937,
|
| 1012 |
+
"learning_rate": 0.0001,
|
| 1013 |
+
"loss": 1.0259,
|
| 1014 |
+
"step": 2820
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 0.15562496575154802,
|
| 1018 |
+
"grad_norm": 2.6327946186065674,
|
| 1019 |
+
"learning_rate": 0.0001,
|
| 1020 |
+
"loss": 0.9829,
|
| 1021 |
+
"step": 2840
|
| 1022 |
+
},
|
| 1023 |
+
{
|
| 1024 |
+
"epoch": 0.1567209162145871,
|
| 1025 |
+
"grad_norm": 2.4873671531677246,
|
| 1026 |
+
"learning_rate": 0.0001,
|
| 1027 |
+
"loss": 1.0472,
|
| 1028 |
+
"step": 2860
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.15781686667762618,
|
| 1032 |
+
"grad_norm": 2.1543166637420654,
|
| 1033 |
+
"learning_rate": 0.0001,
|
| 1034 |
+
"loss": 1.0157,
|
| 1035 |
+
"step": 2880
|
| 1036 |
+
},
|
| 1037 |
+
{
|
| 1038 |
+
"epoch": 0.15891281714066524,
|
| 1039 |
+
"grad_norm": 1.9687381982803345,
|
| 1040 |
+
"learning_rate": 0.0001,
|
| 1041 |
+
"loss": 1.0465,
|
| 1042 |
+
"step": 2900
|
| 1043 |
+
},
|
| 1044 |
+
{
|
| 1045 |
+
"epoch": 0.1600087676037043,
|
| 1046 |
+
"grad_norm": 2.868544816970825,
|
| 1047 |
+
"learning_rate": 0.0001,
|
| 1048 |
+
"loss": 0.9835,
|
| 1049 |
+
"step": 2920
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 0.1611047180667434,
|
| 1053 |
+
"grad_norm": 2.3211984634399414,
|
| 1054 |
+
"learning_rate": 0.0001,
|
| 1055 |
+
"loss": 1.1204,
|
| 1056 |
+
"step": 2940
|
| 1057 |
+
},
|
| 1058 |
+
{
|
| 1059 |
+
"epoch": 0.16220066852978246,
|
| 1060 |
+
"grad_norm": 2.631458282470703,
|
| 1061 |
+
"learning_rate": 0.0001,
|
| 1062 |
+
"loss": 1.0175,
|
| 1063 |
+
"step": 2960
|
| 1064 |
+
},
|
| 1065 |
+
{
|
| 1066 |
+
"epoch": 0.16329661899282152,
|
| 1067 |
+
"grad_norm": 2.7994022369384766,
|
| 1068 |
+
"learning_rate": 0.0001,
|
| 1069 |
+
"loss": 1.0828,
|
| 1070 |
+
"step": 2980
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.1643925694558606,
|
| 1074 |
+
"grad_norm": 2.051626443862915,
|
| 1075 |
+
"learning_rate": 0.0001,
|
| 1076 |
+
"loss": 0.97,
|
| 1077 |
+
"step": 3000
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 0.1643925694558606,
|
| 1081 |
+
"eval_loss": 1.0072325468063354,
|
| 1082 |
+
"eval_runtime": 30710.9249,
|
| 1083 |
+
"eval_samples_per_second": 2.113,
|
| 1084 |
+
"eval_steps_per_second": 0.066,
|
| 1085 |
+
"eval_wer": 55.08434535201816,
|
| 1086 |
+
"step": 3000
|
| 1087 |
+
},
|
| 1088 |
+
{
|
| 1089 |
+
"epoch": 0.16548851991889968,
|
| 1090 |
+
"grad_norm": 2.6088364124298096,
|
| 1091 |
+
"learning_rate": 0.0001,
|
| 1092 |
+
"loss": 0.9803,
|
| 1093 |
+
"step": 3020
|
| 1094 |
+
},
|
| 1095 |
+
{
|
| 1096 |
+
"epoch": 0.16658447038193874,
|
| 1097 |
+
"grad_norm": 2.234034299850464,
|
| 1098 |
+
"learning_rate": 0.0001,
|
| 1099 |
+
"loss": 1.0757,
|
| 1100 |
+
"step": 3040
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 0.1676804208449778,
|
| 1104 |
+
"grad_norm": 2.3472328186035156,
|
| 1105 |
+
"learning_rate": 0.0001,
|
| 1106 |
+
"loss": 0.9408,
|
| 1107 |
+
"step": 3060
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 0.16877637130801687,
|
| 1111 |
+
"grad_norm": 2.5871200561523438,
|
| 1112 |
+
"learning_rate": 0.0001,
|
| 1113 |
+
"loss": 0.9269,
|
| 1114 |
+
"step": 3080
|
| 1115 |
+
},
|
| 1116 |
+
{
|
| 1117 |
+
"epoch": 0.16987232177105596,
|
| 1118 |
+
"grad_norm": 2.0150465965270996,
|
| 1119 |
+
"learning_rate": 0.0001,
|
| 1120 |
+
"loss": 1.0547,
|
| 1121 |
+
"step": 3100
|
| 1122 |
+
},
|
| 1123 |
+
{
|
| 1124 |
+
"epoch": 0.17096827223409503,
|
| 1125 |
+
"grad_norm": 2.5823395252227783,
|
| 1126 |
+
"learning_rate": 0.0001,
|
| 1127 |
+
"loss": 1.0559,
|
| 1128 |
+
"step": 3120
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 0.1720642226971341,
|
| 1132 |
+
"grad_norm": 2.8252885341644287,
|
| 1133 |
+
"learning_rate": 0.0001,
|
| 1134 |
+
"loss": 1.1219,
|
| 1135 |
+
"step": 3140
|
| 1136 |
+
},
|
| 1137 |
+
{
|
| 1138 |
+
"epoch": 0.17316017316017315,
|
| 1139 |
+
"grad_norm": 2.1086535453796387,
|
| 1140 |
+
"learning_rate": 0.0001,
|
| 1141 |
+
"loss": 1.0089,
|
| 1142 |
+
"step": 3160
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 0.17425612362321224,
|
| 1146 |
+
"grad_norm": 2.2288014888763428,
|
| 1147 |
+
"learning_rate": 0.0001,
|
| 1148 |
+
"loss": 1.136,
|
| 1149 |
+
"step": 3180
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 0.1753520740862513,
|
| 1153 |
+
"grad_norm": 2.6622703075408936,
|
| 1154 |
+
"learning_rate": 0.0001,
|
| 1155 |
+
"loss": 1.0395,
|
| 1156 |
+
"step": 3200
|
| 1157 |
+
},
|
| 1158 |
+
{
|
| 1159 |
+
"epoch": 0.17644802454929037,
|
| 1160 |
+
"grad_norm": 1.9478541612625122,
|
| 1161 |
+
"learning_rate": 0.0001,
|
| 1162 |
+
"loss": 1.0658,
|
| 1163 |
+
"step": 3220
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 0.17754397501232944,
|
| 1167 |
+
"grad_norm": 2.55828857421875,
|
| 1168 |
+
"learning_rate": 0.0001,
|
| 1169 |
+
"loss": 0.9904,
|
| 1170 |
+
"step": 3240
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.1786399254753685,
|
| 1174 |
+
"grad_norm": 2.533651828765869,
|
| 1175 |
+
"learning_rate": 0.0001,
|
| 1176 |
+
"loss": 0.9733,
|
| 1177 |
+
"step": 3260
|
| 1178 |
+
},
|
| 1179 |
+
{
|
| 1180 |
+
"epoch": 0.1797358759384076,
|
| 1181 |
+
"grad_norm": 1.8745101690292358,
|
| 1182 |
+
"learning_rate": 0.0001,
|
| 1183 |
+
"loss": 0.9903,
|
| 1184 |
+
"step": 3280
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 0.18083182640144665,
|
| 1188 |
+
"grad_norm": 1.8459206819534302,
|
| 1189 |
+
"learning_rate": 0.0001,
|
| 1190 |
+
"loss": 0.9095,
|
| 1191 |
+
"step": 3300
|
| 1192 |
+
},
|
| 1193 |
+
{
|
| 1194 |
+
"epoch": 0.18192777686448572,
|
| 1195 |
+
"grad_norm": 2.6654012203216553,
|
| 1196 |
+
"learning_rate": 0.0001,
|
| 1197 |
+
"loss": 0.9854,
|
| 1198 |
+
"step": 3320
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 0.18302372732752478,
|
| 1202 |
+
"grad_norm": 2.6444480419158936,
|
| 1203 |
+
"learning_rate": 0.0001,
|
| 1204 |
+
"loss": 0.8857,
|
| 1205 |
+
"step": 3340
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 0.18411967779056387,
|
| 1209 |
+
"grad_norm": 2.190462827682495,
|
| 1210 |
+
"learning_rate": 0.0001,
|
| 1211 |
+
"loss": 0.9375,
|
| 1212 |
+
"step": 3360
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 0.18521562825360294,
|
| 1216 |
+
"grad_norm": 2.8208882808685303,
|
| 1217 |
+
"learning_rate": 0.0001,
|
| 1218 |
+
"loss": 0.9646,
|
| 1219 |
+
"step": 3380
|
| 1220 |
+
},
|
| 1221 |
+
{
|
| 1222 |
+
"epoch": 0.186311578716642,
|
| 1223 |
+
"grad_norm": 2.4978795051574707,
|
| 1224 |
+
"learning_rate": 0.0001,
|
| 1225 |
+
"loss": 0.9724,
|
| 1226 |
+
"step": 3400
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.18740752917968106,
|
| 1230 |
+
"grad_norm": 2.4202938079833984,
|
| 1231 |
+
"learning_rate": 0.0001,
|
| 1232 |
+
"loss": 0.9659,
|
| 1233 |
+
"step": 3420
|
| 1234 |
+
},
|
| 1235 |
+
{
|
| 1236 |
+
"epoch": 0.18850347964272016,
|
| 1237 |
+
"grad_norm": 1.9026118516921997,
|
| 1238 |
+
"learning_rate": 0.0001,
|
| 1239 |
+
"loss": 1.0321,
|
| 1240 |
+
"step": 3440
|
| 1241 |
+
},
|
| 1242 |
+
{
|
| 1243 |
+
"epoch": 0.18959943010575922,
|
| 1244 |
+
"grad_norm": 2.6031651496887207,
|
| 1245 |
+
"learning_rate": 0.0001,
|
| 1246 |
+
"loss": 0.9622,
|
| 1247 |
+
"step": 3460
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 0.19069538056879828,
|
| 1251 |
+
"grad_norm": 1.962509274482727,
|
| 1252 |
+
"learning_rate": 0.0001,
|
| 1253 |
+
"loss": 1.0262,
|
| 1254 |
+
"step": 3480
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 0.19179133103183735,
|
| 1258 |
+
"grad_norm": 2.794633626937866,
|
| 1259 |
+
"learning_rate": 0.0001,
|
| 1260 |
+
"loss": 1.0626,
|
| 1261 |
+
"step": 3500
|
| 1262 |
+
},
|
| 1263 |
+
{
|
| 1264 |
+
"epoch": 0.19288728149487644,
|
| 1265 |
+
"grad_norm": 2.4276185035705566,
|
| 1266 |
+
"learning_rate": 0.0001,
|
| 1267 |
+
"loss": 0.9961,
|
| 1268 |
+
"step": 3520
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 0.1939832319579155,
|
| 1272 |
+
"grad_norm": 2.0747737884521484,
|
| 1273 |
+
"learning_rate": 0.0001,
|
| 1274 |
+
"loss": 0.8945,
|
| 1275 |
+
"step": 3540
|
| 1276 |
+
},
|
| 1277 |
+
{
|
| 1278 |
+
"epoch": 0.19507918242095457,
|
| 1279 |
+
"grad_norm": 1.9151681661605835,
|
| 1280 |
+
"learning_rate": 0.0001,
|
| 1281 |
+
"loss": 1.0664,
|
| 1282 |
+
"step": 3560
|
| 1283 |
+
},
|
| 1284 |
+
{
|
| 1285 |
+
"epoch": 0.19617513288399363,
|
| 1286 |
+
"grad_norm": 2.11547589302063,
|
| 1287 |
+
"learning_rate": 0.0001,
|
| 1288 |
+
"loss": 0.9865,
|
| 1289 |
+
"step": 3580
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 0.19727108334703272,
|
| 1293 |
+
"grad_norm": 2.359848737716675,
|
| 1294 |
+
"learning_rate": 0.0001,
|
| 1295 |
+
"loss": 0.95,
|
| 1296 |
+
"step": 3600
|
| 1297 |
+
},
|
| 1298 |
+
{
|
| 1299 |
+
"epoch": 0.19836703381007179,
|
| 1300 |
+
"grad_norm": 1.9854378700256348,
|
| 1301 |
+
"learning_rate": 0.0001,
|
| 1302 |
+
"loss": 0.9992,
|
| 1303 |
+
"step": 3620
|
| 1304 |
+
},
|
| 1305 |
+
{
|
| 1306 |
+
"epoch": 0.19946298427311085,
|
| 1307 |
+
"grad_norm": 2.476423978805542,
|
| 1308 |
+
"learning_rate": 0.0001,
|
| 1309 |
+
"loss": 0.9097,
|
| 1310 |
+
"step": 3640
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 0.2005589347361499,
|
| 1314 |
+
"grad_norm": 2.420011281967163,
|
| 1315 |
+
"learning_rate": 0.0001,
|
| 1316 |
+
"loss": 1.0167,
|
| 1317 |
+
"step": 3660
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 0.201654885199189,
|
| 1321 |
+
"grad_norm": 2.12312388420105,
|
| 1322 |
+
"learning_rate": 0.0001,
|
| 1323 |
+
"loss": 0.9298,
|
| 1324 |
+
"step": 3680
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 0.20275083566222807,
|
| 1328 |
+
"grad_norm": 1.9679986238479614,
|
| 1329 |
+
"learning_rate": 0.0001,
|
| 1330 |
+
"loss": 1.0064,
|
| 1331 |
+
"step": 3700
|
| 1332 |
+
},
|
| 1333 |
+
{
|
| 1334 |
+
"epoch": 0.20384678612526713,
|
| 1335 |
+
"grad_norm": 2.608135461807251,
|
| 1336 |
+
"learning_rate": 0.0001,
|
| 1337 |
+
"loss": 0.9396,
|
| 1338 |
+
"step": 3720
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 0.2049427365883062,
|
| 1342 |
+
"grad_norm": 2.542102098464966,
|
| 1343 |
+
"learning_rate": 0.0001,
|
| 1344 |
+
"loss": 1.0868,
|
| 1345 |
+
"step": 3740
|
| 1346 |
+
},
|
| 1347 |
+
{
|
| 1348 |
+
"epoch": 0.2060386870513453,
|
| 1349 |
+
"grad_norm": 2.5252091884613037,
|
| 1350 |
+
"learning_rate": 0.0001,
|
| 1351 |
+
"loss": 1.0417,
|
| 1352 |
+
"step": 3760
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.20713463751438435,
|
| 1356 |
+
"grad_norm": 1.98774254322052,
|
| 1357 |
+
"learning_rate": 0.0001,
|
| 1358 |
+
"loss": 0.9949,
|
| 1359 |
+
"step": 3780
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 0.20823058797742341,
|
| 1363 |
+
"grad_norm": 1.9502965211868286,
|
| 1364 |
+
"learning_rate": 0.0001,
|
| 1365 |
+
"loss": 0.9862,
|
| 1366 |
+
"step": 3800
|
| 1367 |
+
},
|
| 1368 |
+
{
|
| 1369 |
+
"epoch": 0.20932653844046248,
|
| 1370 |
+
"grad_norm": 2.2537944316864014,
|
| 1371 |
+
"learning_rate": 0.0001,
|
| 1372 |
+
"loss": 0.9087,
|
| 1373 |
+
"step": 3820
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 0.21042248890350157,
|
| 1377 |
+
"grad_norm": 2.2866523265838623,
|
| 1378 |
+
"learning_rate": 0.0001,
|
| 1379 |
+
"loss": 1.0128,
|
| 1380 |
+
"step": 3840
|
| 1381 |
+
},
|
| 1382 |
+
{
|
| 1383 |
+
"epoch": 0.21151843936654063,
|
| 1384 |
+
"grad_norm": 2.2907001972198486,
|
| 1385 |
+
"learning_rate": 0.0001,
|
| 1386 |
+
"loss": 0.9654,
|
| 1387 |
+
"step": 3860
|
| 1388 |
+
},
|
| 1389 |
+
{
|
| 1390 |
+
"epoch": 0.2126143898295797,
|
| 1391 |
+
"grad_norm": 2.5648560523986816,
|
| 1392 |
+
"learning_rate": 0.0001,
|
| 1393 |
+
"loss": 1.0269,
|
| 1394 |
+
"step": 3880
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 0.21371034029261876,
|
| 1398 |
+
"grad_norm": 2.198974847793579,
|
| 1399 |
+
"learning_rate": 0.0001,
|
| 1400 |
+
"loss": 0.9823,
|
| 1401 |
+
"step": 3900
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 0.21480629075565785,
|
| 1405 |
+
"grad_norm": 2.1045591831207275,
|
| 1406 |
+
"learning_rate": 0.0001,
|
| 1407 |
+
"loss": 0.9139,
|
| 1408 |
+
"step": 3920
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 0.21590224121869692,
|
| 1412 |
+
"grad_norm": 2.1462857723236084,
|
| 1413 |
+
"learning_rate": 0.0001,
|
| 1414 |
+
"loss": 0.9406,
|
| 1415 |
+
"step": 3940
|
| 1416 |
+
},
|
| 1417 |
+
{
|
| 1418 |
+
"epoch": 0.21699819168173598,
|
| 1419 |
+
"grad_norm": 2.3216285705566406,
|
| 1420 |
+
"learning_rate": 0.0001,
|
| 1421 |
+
"loss": 0.8597,
|
| 1422 |
+
"step": 3960
|
| 1423 |
+
},
|
| 1424 |
+
{
|
| 1425 |
+
"epoch": 0.21809414214477504,
|
| 1426 |
+
"grad_norm": 1.867150068283081,
|
| 1427 |
+
"learning_rate": 0.0001,
|
| 1428 |
+
"loss": 0.9776,
|
| 1429 |
+
"step": 3980
|
| 1430 |
+
},
|
| 1431 |
+
{
|
| 1432 |
+
"epoch": 0.21919009260781414,
|
| 1433 |
+
"grad_norm": 2.3432791233062744,
|
| 1434 |
+
"learning_rate": 0.0001,
|
| 1435 |
+
"loss": 0.9546,
|
| 1436 |
+
"step": 4000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 0.21919009260781414,
|
| 1440 |
+
"eval_loss": 0.9323587417602539,
|
| 1441 |
+
"eval_runtime": 30935.2713,
|
| 1442 |
+
"eval_samples_per_second": 2.098,
|
| 1443 |
+
"eval_steps_per_second": 0.066,
|
| 1444 |
+
"eval_wer": 63.836951720973865,
|
| 1445 |
+
"step": 4000
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 0.2202860430708532,
|
| 1449 |
+
"grad_norm": 1.9426536560058594,
|
| 1450 |
+
"learning_rate": 0.0001,
|
| 1451 |
+
"loss": 0.9291,
|
| 1452 |
+
"step": 4020
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 0.22138199353389226,
|
| 1456 |
+
"grad_norm": 2.693723201751709,
|
| 1457 |
+
"learning_rate": 0.0001,
|
| 1458 |
+
"loss": 0.9072,
|
| 1459 |
+
"step": 4040
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.22247794399693133,
|
| 1463 |
+
"grad_norm": 2.237900972366333,
|
| 1464 |
+
"learning_rate": 0.0001,
|
| 1465 |
+
"loss": 0.8571,
|
| 1466 |
+
"step": 4060
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.22357389445997042,
|
| 1470 |
+
"grad_norm": 2.739129066467285,
|
| 1471 |
+
"learning_rate": 0.0001,
|
| 1472 |
+
"loss": 0.9132,
|
| 1473 |
+
"step": 4080
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 0.22466984492300948,
|
| 1477 |
+
"grad_norm": 1.886438012123108,
|
| 1478 |
+
"learning_rate": 0.0001,
|
| 1479 |
+
"loss": 0.9646,
|
| 1480 |
+
"step": 4100
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 0.22576579538604855,
|
| 1484 |
+
"grad_norm": 2.3505897521972656,
|
| 1485 |
+
"learning_rate": 0.0001,
|
| 1486 |
+
"loss": 1.0479,
|
| 1487 |
+
"step": 4120
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.2268617458490876,
|
| 1491 |
+
"grad_norm": 2.4302868843078613,
|
| 1492 |
+
"learning_rate": 0.0001,
|
| 1493 |
+
"loss": 0.9956,
|
| 1494 |
+
"step": 4140
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 0.2279576963121267,
|
| 1498 |
+
"grad_norm": 2.2747528553009033,
|
| 1499 |
+
"learning_rate": 0.0001,
|
| 1500 |
+
"loss": 0.9621,
|
| 1501 |
+
"step": 4160
|
| 1502 |
+
},
|
| 1503 |
+
{
|
| 1504 |
+
"epoch": 0.22905364677516576,
|
| 1505 |
+
"grad_norm": 2.312248945236206,
|
| 1506 |
+
"learning_rate": 0.0001,
|
| 1507 |
+
"loss": 0.9292,
|
| 1508 |
+
"step": 4180
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.23014959723820483,
|
| 1512 |
+
"grad_norm": 2.0439066886901855,
|
| 1513 |
+
"learning_rate": 0.0001,
|
| 1514 |
+
"loss": 0.8804,
|
| 1515 |
+
"step": 4200
|
| 1516 |
+
},
|
| 1517 |
+
{
|
| 1518 |
+
"epoch": 0.2312455477012439,
|
| 1519 |
+
"grad_norm": 2.615898609161377,
|
| 1520 |
+
"learning_rate": 0.0001,
|
| 1521 |
+
"loss": 0.9302,
|
| 1522 |
+
"step": 4220
|
| 1523 |
+
},
|
| 1524 |
+
{
|
| 1525 |
+
"epoch": 0.23234149816428298,
|
| 1526 |
+
"grad_norm": 2.306796073913574,
|
| 1527 |
+
"learning_rate": 0.0001,
|
| 1528 |
+
"loss": 1.0401,
|
| 1529 |
+
"step": 4240
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 0.23343744862732205,
|
| 1533 |
+
"grad_norm": 2.4527432918548584,
|
| 1534 |
+
"learning_rate": 0.0001,
|
| 1535 |
+
"loss": 0.9195,
|
| 1536 |
+
"step": 4260
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 0.2345333990903611,
|
| 1540 |
+
"grad_norm": 1.8589290380477905,
|
| 1541 |
+
"learning_rate": 0.0001,
|
| 1542 |
+
"loss": 0.9284,
|
| 1543 |
+
"step": 4280
|
| 1544 |
+
},
|
| 1545 |
+
{
|
| 1546 |
+
"epoch": 0.23562934955340017,
|
| 1547 |
+
"grad_norm": 1.8492025136947632,
|
| 1548 |
+
"learning_rate": 0.0001,
|
| 1549 |
+
"loss": 0.8898,
|
| 1550 |
+
"step": 4300
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.23672530001643927,
|
| 1554 |
+
"grad_norm": 2.574871063232422,
|
| 1555 |
+
"learning_rate": 0.0001,
|
| 1556 |
+
"loss": 1.0026,
|
| 1557 |
+
"step": 4320
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 0.23782125047947833,
|
| 1561 |
+
"grad_norm": 2.2600936889648438,
|
| 1562 |
+
"learning_rate": 0.0001,
|
| 1563 |
+
"loss": 1.0738,
|
| 1564 |
+
"step": 4340
|
| 1565 |
+
},
|
| 1566 |
+
{
|
| 1567 |
+
"epoch": 0.2389172009425174,
|
| 1568 |
+
"grad_norm": 2.35066556930542,
|
| 1569 |
+
"learning_rate": 0.0001,
|
| 1570 |
+
"loss": 0.8573,
|
| 1571 |
+
"step": 4360
|
| 1572 |
+
},
|
| 1573 |
+
{
|
| 1574 |
+
"epoch": 0.24001315140555646,
|
| 1575 |
+
"grad_norm": 2.165745496749878,
|
| 1576 |
+
"learning_rate": 0.0001,
|
| 1577 |
+
"loss": 0.8989,
|
| 1578 |
+
"step": 4380
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.24110910186859555,
|
| 1582 |
+
"grad_norm": 2.1494085788726807,
|
| 1583 |
+
"learning_rate": 0.0001,
|
| 1584 |
+
"loss": 0.8292,
|
| 1585 |
+
"step": 4400
|
| 1586 |
+
},
|
| 1587 |
+
{
|
| 1588 |
+
"epoch": 0.2422050523316346,
|
| 1589 |
+
"grad_norm": 2.185359239578247,
|
| 1590 |
+
"learning_rate": 0.0001,
|
| 1591 |
+
"loss": 0.8954,
|
| 1592 |
+
"step": 4420
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.24330100279467368,
|
| 1596 |
+
"grad_norm": 2.193904161453247,
|
| 1597 |
+
"learning_rate": 0.0001,
|
| 1598 |
+
"loss": 0.8944,
|
| 1599 |
+
"step": 4440
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 0.24439695325771274,
|
| 1603 |
+
"grad_norm": 2.1101438999176025,
|
| 1604 |
+
"learning_rate": 0.0001,
|
| 1605 |
+
"loss": 0.9059,
|
| 1606 |
+
"step": 4460
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.24549290372075183,
|
| 1610 |
+
"grad_norm": 2.026642084121704,
|
| 1611 |
+
"learning_rate": 0.0001,
|
| 1612 |
+
"loss": 0.8978,
|
| 1613 |
+
"step": 4480
|
| 1614 |
+
},
|
| 1615 |
+
{
|
| 1616 |
+
"epoch": 0.2465888541837909,
|
| 1617 |
+
"grad_norm": 2.0481228828430176,
|
| 1618 |
+
"learning_rate": 0.0001,
|
| 1619 |
+
"loss": 0.8835,
|
| 1620 |
+
"step": 4500
|
| 1621 |
+
},
|
| 1622 |
+
{
|
| 1623 |
+
"epoch": 0.24768480464682996,
|
| 1624 |
+
"grad_norm": 2.201350688934326,
|
| 1625 |
+
"learning_rate": 0.0001,
|
| 1626 |
+
"loss": 0.9519,
|
| 1627 |
+
"step": 4520
|
| 1628 |
+
},
|
| 1629 |
+
{
|
| 1630 |
+
"epoch": 0.24878075510986902,
|
| 1631 |
+
"grad_norm": 1.852100133895874,
|
| 1632 |
+
"learning_rate": 0.0001,
|
| 1633 |
+
"loss": 0.8458,
|
| 1634 |
+
"step": 4540
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.24987670557290811,
|
| 1638 |
+
"grad_norm": 2.1303794384002686,
|
| 1639 |
+
"learning_rate": 0.0001,
|
| 1640 |
+
"loss": 0.9092,
|
| 1641 |
+
"step": 4560
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 0.25097265603594715,
|
| 1645 |
+
"grad_norm": 2.2715415954589844,
|
| 1646 |
+
"learning_rate": 0.0001,
|
| 1647 |
+
"loss": 0.8931,
|
| 1648 |
+
"step": 4580
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 0.25206860649898627,
|
| 1652 |
+
"grad_norm": 2.091785192489624,
|
| 1653 |
+
"learning_rate": 0.0001,
|
| 1654 |
+
"loss": 0.8645,
|
| 1655 |
+
"step": 4600
|
| 1656 |
+
},
|
| 1657 |
+
{
|
| 1658 |
+
"epoch": 0.25316455696202533,
|
| 1659 |
+
"grad_norm": 2.108103036880493,
|
| 1660 |
+
"learning_rate": 0.0001,
|
| 1661 |
+
"loss": 0.8387,
|
| 1662 |
+
"step": 4620
|
| 1663 |
+
},
|
| 1664 |
+
{
|
| 1665 |
+
"epoch": 0.2542605074250644,
|
| 1666 |
+
"grad_norm": 2.083848237991333,
|
| 1667 |
+
"learning_rate": 0.0001,
|
| 1668 |
+
"loss": 0.8315,
|
| 1669 |
+
"step": 4640
|
| 1670 |
+
},
|
| 1671 |
+
{
|
| 1672 |
+
"epoch": 0.25535645788810346,
|
| 1673 |
+
"grad_norm": 1.570475459098816,
|
| 1674 |
+
"learning_rate": 0.0001,
|
| 1675 |
+
"loss": 0.9355,
|
| 1676 |
+
"step": 4660
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.2564524083511425,
|
| 1680 |
+
"grad_norm": 1.90199875831604,
|
| 1681 |
+
"learning_rate": 0.0001,
|
| 1682 |
+
"loss": 0.8308,
|
| 1683 |
+
"step": 4680
|
| 1684 |
+
},
|
| 1685 |
+
{
|
| 1686 |
+
"epoch": 0.2575483588141816,
|
| 1687 |
+
"grad_norm": 2.1952812671661377,
|
| 1688 |
+
"learning_rate": 0.0001,
|
| 1689 |
+
"loss": 0.8618,
|
| 1690 |
+
"step": 4700
|
| 1691 |
+
},
|
| 1692 |
+
{
|
| 1693 |
+
"epoch": 0.25864430927722065,
|
| 1694 |
+
"grad_norm": 2.0530431270599365,
|
| 1695 |
+
"learning_rate": 0.0001,
|
| 1696 |
+
"loss": 0.7951,
|
| 1697 |
+
"step": 4720
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.2597402597402597,
|
| 1701 |
+
"grad_norm": 2.202252149581909,
|
| 1702 |
+
"learning_rate": 0.0001,
|
| 1703 |
+
"loss": 0.8858,
|
| 1704 |
+
"step": 4740
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 0.26083621020329883,
|
| 1708 |
+
"grad_norm": 1.9541796445846558,
|
| 1709 |
+
"learning_rate": 0.0001,
|
| 1710 |
+
"loss": 0.8466,
|
| 1711 |
+
"step": 4760
|
| 1712 |
+
},
|
| 1713 |
+
{
|
| 1714 |
+
"epoch": 0.2619321606663379,
|
| 1715 |
+
"grad_norm": 1.9440534114837646,
|
| 1716 |
+
"learning_rate": 0.0001,
|
| 1717 |
+
"loss": 0.8488,
|
| 1718 |
+
"step": 4780
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.26302811112937696,
|
| 1722 |
+
"grad_norm": 2.569821834564209,
|
| 1723 |
+
"learning_rate": 0.0001,
|
| 1724 |
+
"loss": 0.963,
|
| 1725 |
+
"step": 4800
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.264124061592416,
|
| 1729 |
+
"grad_norm": 1.8896031379699707,
|
| 1730 |
+
"learning_rate": 0.0001,
|
| 1731 |
+
"loss": 0.837,
|
| 1732 |
+
"step": 4820
|
| 1733 |
+
},
|
| 1734 |
+
{
|
| 1735 |
+
"epoch": 0.2652200120554551,
|
| 1736 |
+
"grad_norm": 1.9390859603881836,
|
| 1737 |
+
"learning_rate": 0.0001,
|
| 1738 |
+
"loss": 0.8855,
|
| 1739 |
+
"step": 4840
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 0.26631596251849415,
|
| 1743 |
+
"grad_norm": 2.2261974811553955,
|
| 1744 |
+
"learning_rate": 0.0001,
|
| 1745 |
+
"loss": 0.8901,
|
| 1746 |
+
"step": 4860
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 0.2674119129815332,
|
| 1750 |
+
"grad_norm": 2.0486056804656982,
|
| 1751 |
+
"learning_rate": 0.0001,
|
| 1752 |
+
"loss": 0.8073,
|
| 1753 |
+
"step": 4880
|
| 1754 |
+
},
|
| 1755 |
+
{
|
| 1756 |
+
"epoch": 0.2685078634445723,
|
| 1757 |
+
"grad_norm": 2.292015314102173,
|
| 1758 |
+
"learning_rate": 0.0001,
|
| 1759 |
+
"loss": 0.9492,
|
| 1760 |
+
"step": 4900
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.2696038139076114,
|
| 1764 |
+
"grad_norm": 2.0762240886688232,
|
| 1765 |
+
"learning_rate": 0.0001,
|
| 1766 |
+
"loss": 0.8528,
|
| 1767 |
+
"step": 4920
|
| 1768 |
+
},
|
| 1769 |
+
{
|
| 1770 |
+
"epoch": 0.27069976437065046,
|
| 1771 |
+
"grad_norm": 1.870642066001892,
|
| 1772 |
+
"learning_rate": 0.0001,
|
| 1773 |
+
"loss": 0.9482,
|
| 1774 |
+
"step": 4940
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 0.27179571483368953,
|
| 1778 |
+
"grad_norm": 2.436768054962158,
|
| 1779 |
+
"learning_rate": 0.0001,
|
| 1780 |
+
"loss": 0.9299,
|
| 1781 |
+
"step": 4960
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 0.2728916652967286,
|
| 1785 |
+
"grad_norm": 2.505880832672119,
|
| 1786 |
+
"learning_rate": 0.0001,
|
| 1787 |
+
"loss": 0.9259,
|
| 1788 |
+
"step": 4980
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 0.27398761575976766,
|
| 1792 |
+
"grad_norm": 1.717252492904663,
|
| 1793 |
+
"learning_rate": 0.0001,
|
| 1794 |
+
"loss": 0.8134,
|
| 1795 |
+
"step": 5000
|
| 1796 |
+
},
|
| 1797 |
+
{
|
| 1798 |
+
"epoch": 0.27398761575976766,
|
| 1799 |
+
"eval_loss": 0.8726964592933655,
|
| 1800 |
+
"eval_runtime": 30710.3822,
|
| 1801 |
+
"eval_samples_per_second": 2.113,
|
| 1802 |
+
"eval_steps_per_second": 0.066,
|
| 1803 |
+
"eval_wer": 52.213316533880224,
|
| 1804 |
+
"step": 5000
|
| 1805 |
+
},
|
| 1806 |
+
{
|
| 1807 |
+
"epoch": 0.2750835662228067,
|
| 1808 |
+
"grad_norm": 2.28765869140625,
|
| 1809 |
+
"learning_rate": 0.0001,
|
| 1810 |
+
"loss": 1.0229,
|
| 1811 |
+
"step": 5020
|
| 1812 |
+
},
|
| 1813 |
+
{
|
| 1814 |
+
"epoch": 0.2761795166858458,
|
| 1815 |
+
"grad_norm": 2.2264580726623535,
|
| 1816 |
+
"learning_rate": 0.0001,
|
| 1817 |
+
"loss": 0.8291,
|
| 1818 |
+
"step": 5040
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 0.27727546714888485,
|
| 1822 |
+
"grad_norm": 1.9387757778167725,
|
| 1823 |
+
"learning_rate": 0.0001,
|
| 1824 |
+
"loss": 0.821,
|
| 1825 |
+
"step": 5060
|
| 1826 |
+
},
|
| 1827 |
+
{
|
| 1828 |
+
"epoch": 0.27837141761192397,
|
| 1829 |
+
"grad_norm": 2.8628933429718018,
|
| 1830 |
+
"learning_rate": 0.0001,
|
| 1831 |
+
"loss": 0.9521,
|
| 1832 |
+
"step": 5080
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 0.27946736807496303,
|
| 1836 |
+
"grad_norm": 2.2691447734832764,
|
| 1837 |
+
"learning_rate": 0.0001,
|
| 1838 |
+
"loss": 0.8182,
|
| 1839 |
+
"step": 5100
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 0.2805633185380021,
|
| 1843 |
+
"grad_norm": 1.9515260457992554,
|
| 1844 |
+
"learning_rate": 0.0001,
|
| 1845 |
+
"loss": 0.9342,
|
| 1846 |
+
"step": 5120
|
| 1847 |
+
},
|
| 1848 |
+
{
|
| 1849 |
+
"epoch": 0.28165926900104116,
|
| 1850 |
+
"grad_norm": 2.1714837551116943,
|
| 1851 |
+
"learning_rate": 0.0001,
|
| 1852 |
+
"loss": 0.9663,
|
| 1853 |
+
"step": 5140
|
| 1854 |
+
},
|
| 1855 |
+
{
|
| 1856 |
+
"epoch": 0.2827552194640802,
|
| 1857 |
+
"grad_norm": 2.0159664154052734,
|
| 1858 |
+
"learning_rate": 0.0001,
|
| 1859 |
+
"loss": 0.8294,
|
| 1860 |
+
"step": 5160
|
| 1861 |
+
},
|
| 1862 |
+
{
|
| 1863 |
+
"epoch": 0.2838511699271193,
|
| 1864 |
+
"grad_norm": 2.024634599685669,
|
| 1865 |
+
"learning_rate": 0.0001,
|
| 1866 |
+
"loss": 0.896,
|
| 1867 |
+
"step": 5180
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 0.28494712039015835,
|
| 1871 |
+
"grad_norm": 2.0035595893859863,
|
| 1872 |
+
"learning_rate": 0.0001,
|
| 1873 |
+
"loss": 0.8446,
|
| 1874 |
+
"step": 5200
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.2860430708531974,
|
| 1878 |
+
"grad_norm": 2.4142866134643555,
|
| 1879 |
+
"learning_rate": 0.0001,
|
| 1880 |
+
"loss": 0.8835,
|
| 1881 |
+
"step": 5220
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 0.28713902131623653,
|
| 1885 |
+
"grad_norm": 2.070338010787964,
|
| 1886 |
+
"learning_rate": 0.0001,
|
| 1887 |
+
"loss": 0.8687,
|
| 1888 |
+
"step": 5240
|
| 1889 |
+
},
|
| 1890 |
+
{
|
| 1891 |
+
"epoch": 0.2882349717792756,
|
| 1892 |
+
"grad_norm": 1.9818578958511353,
|
| 1893 |
+
"learning_rate": 0.0001,
|
| 1894 |
+
"loss": 0.8296,
|
| 1895 |
+
"step": 5260
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 0.28933092224231466,
|
| 1899 |
+
"grad_norm": 1.8923412561416626,
|
| 1900 |
+
"learning_rate": 0.0001,
|
| 1901 |
+
"loss": 0.8999,
|
| 1902 |
+
"step": 5280
|
| 1903 |
+
},
|
| 1904 |
+
{
|
| 1905 |
+
"epoch": 0.2904268727053537,
|
| 1906 |
+
"grad_norm": 2.200206995010376,
|
| 1907 |
+
"learning_rate": 0.0001,
|
| 1908 |
+
"loss": 0.8662,
|
| 1909 |
+
"step": 5300
|
| 1910 |
+
},
|
| 1911 |
+
{
|
| 1912 |
+
"epoch": 0.2915228231683928,
|
| 1913 |
+
"grad_norm": 1.982446551322937,
|
| 1914 |
+
"learning_rate": 0.0001,
|
| 1915 |
+
"loss": 0.8301,
|
| 1916 |
+
"step": 5320
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 0.29261877363143185,
|
| 1920 |
+
"grad_norm": 1.934844732284546,
|
| 1921 |
+
"learning_rate": 0.0001,
|
| 1922 |
+
"loss": 0.8219,
|
| 1923 |
+
"step": 5340
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 0.2937147240944709,
|
| 1927 |
+
"grad_norm": 2.2790510654449463,
|
| 1928 |
+
"learning_rate": 0.0001,
|
| 1929 |
+
"loss": 0.8666,
|
| 1930 |
+
"step": 5360
|
| 1931 |
+
},
|
| 1932 |
+
{
|
| 1933 |
+
"epoch": 0.29481067455751,
|
| 1934 |
+
"grad_norm": 1.771672248840332,
|
| 1935 |
+
"learning_rate": 0.0001,
|
| 1936 |
+
"loss": 0.843,
|
| 1937 |
+
"step": 5380
|
| 1938 |
+
},
|
| 1939 |
+
{
|
| 1940 |
+
"epoch": 0.2959066250205491,
|
| 1941 |
+
"grad_norm": 2.3459877967834473,
|
| 1942 |
+
"learning_rate": 0.0001,
|
| 1943 |
+
"loss": 0.8516,
|
| 1944 |
+
"step": 5400
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 0.29700257548358816,
|
| 1948 |
+
"grad_norm": 2.156458854675293,
|
| 1949 |
+
"learning_rate": 0.0001,
|
| 1950 |
+
"loss": 0.8425,
|
| 1951 |
+
"step": 5420
|
| 1952 |
+
},
|
| 1953 |
+
{
|
| 1954 |
+
"epoch": 0.2980985259466272,
|
| 1955 |
+
"grad_norm": 1.9492950439453125,
|
| 1956 |
+
"learning_rate": 0.0001,
|
| 1957 |
+
"loss": 0.8445,
|
| 1958 |
+
"step": 5440
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 0.2991944764096663,
|
| 1962 |
+
"grad_norm": 2.1061997413635254,
|
| 1963 |
+
"learning_rate": 0.0001,
|
| 1964 |
+
"loss": 0.8858,
|
| 1965 |
+
"step": 5460
|
| 1966 |
+
},
|
| 1967 |
+
{
|
| 1968 |
+
"epoch": 0.30029042687270535,
|
| 1969 |
+
"grad_norm": 2.3567299842834473,
|
| 1970 |
+
"learning_rate": 0.0001,
|
| 1971 |
+
"loss": 0.8376,
|
| 1972 |
+
"step": 5480
|
| 1973 |
+
},
|
| 1974 |
+
{
|
| 1975 |
+
"epoch": 0.3013863773357444,
|
| 1976 |
+
"grad_norm": 2.1302335262298584,
|
| 1977 |
+
"learning_rate": 0.0001,
|
| 1978 |
+
"loss": 0.8272,
|
| 1979 |
+
"step": 5500
|
| 1980 |
+
},
|
| 1981 |
+
{
|
| 1982 |
+
"epoch": 0.3024823277987835,
|
| 1983 |
+
"grad_norm": 2.2098424434661865,
|
| 1984 |
+
"learning_rate": 0.0001,
|
| 1985 |
+
"loss": 0.8742,
|
| 1986 |
+
"step": 5520
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 0.30357827826182254,
|
| 1990 |
+
"grad_norm": 1.7558562755584717,
|
| 1991 |
+
"learning_rate": 0.0001,
|
| 1992 |
+
"loss": 0.8863,
|
| 1993 |
+
"step": 5540
|
| 1994 |
+
},
|
| 1995 |
+
{
|
| 1996 |
+
"epoch": 0.30467422872486166,
|
| 1997 |
+
"grad_norm": 1.8461397886276245,
|
| 1998 |
+
"learning_rate": 0.0001,
|
| 1999 |
+
"loss": 0.8792,
|
| 2000 |
+
"step": 5560
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 0.3057701791879007,
|
| 2004 |
+
"grad_norm": 2.0006344318389893,
|
| 2005 |
+
"learning_rate": 0.0001,
|
| 2006 |
+
"loss": 0.8263,
|
| 2007 |
+
"step": 5580
|
| 2008 |
+
},
|
| 2009 |
+
{
|
| 2010 |
+
"epoch": 0.3068661296509398,
|
| 2011 |
+
"grad_norm": 1.6772565841674805,
|
| 2012 |
+
"learning_rate": 0.0001,
|
| 2013 |
+
"loss": 0.789,
|
| 2014 |
+
"step": 5600
|
| 2015 |
+
},
|
| 2016 |
+
{
|
| 2017 |
+
"epoch": 0.30796208011397885,
|
| 2018 |
+
"grad_norm": 1.9263228178024292,
|
| 2019 |
+
"learning_rate": 0.0001,
|
| 2020 |
+
"loss": 0.842,
|
| 2021 |
+
"step": 5620
|
| 2022 |
+
},
|
| 2023 |
+
{
|
| 2024 |
+
"epoch": 0.3090580305770179,
|
| 2025 |
+
"grad_norm": 1.8888592720031738,
|
| 2026 |
+
"learning_rate": 0.0001,
|
| 2027 |
+
"loss": 0.8475,
|
| 2028 |
+
"step": 5640
|
| 2029 |
+
},
|
| 2030 |
+
{
|
| 2031 |
+
"epoch": 0.310153981040057,
|
| 2032 |
+
"grad_norm": 2.2354602813720703,
|
| 2033 |
+
"learning_rate": 0.0001,
|
| 2034 |
+
"loss": 1.0036,
|
| 2035 |
+
"step": 5660
|
| 2036 |
+
},
|
| 2037 |
+
{
|
| 2038 |
+
"epoch": 0.31124993150309604,
|
| 2039 |
+
"grad_norm": 1.9634332656860352,
|
| 2040 |
+
"learning_rate": 0.0001,
|
| 2041 |
+
"loss": 0.8517,
|
| 2042 |
+
"step": 5680
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.3123458819661351,
|
| 2046 |
+
"grad_norm": 2.348825216293335,
|
| 2047 |
+
"learning_rate": 0.0001,
|
| 2048 |
+
"loss": 0.8731,
|
| 2049 |
+
"step": 5700
|
| 2050 |
+
},
|
| 2051 |
+
{
|
| 2052 |
+
"epoch": 0.3134418324291742,
|
| 2053 |
+
"grad_norm": 2.487741708755493,
|
| 2054 |
+
"learning_rate": 0.0001,
|
| 2055 |
+
"loss": 0.8556,
|
| 2056 |
+
"step": 5720
|
| 2057 |
+
},
|
| 2058 |
+
{
|
| 2059 |
+
"epoch": 0.3145377828922133,
|
| 2060 |
+
"grad_norm": 1.999516248703003,
|
| 2061 |
+
"learning_rate": 0.0001,
|
| 2062 |
+
"loss": 0.7969,
|
| 2063 |
+
"step": 5740
|
| 2064 |
+
},
|
| 2065 |
+
{
|
| 2066 |
+
"epoch": 0.31563373335525235,
|
| 2067 |
+
"grad_norm": 1.9654616117477417,
|
| 2068 |
+
"learning_rate": 0.0001,
|
| 2069 |
+
"loss": 0.7843,
|
| 2070 |
+
"step": 5760
|
| 2071 |
+
},
|
| 2072 |
+
{
|
| 2073 |
+
"epoch": 0.3167296838182914,
|
| 2074 |
+
"grad_norm": 2.1070950031280518,
|
| 2075 |
+
"learning_rate": 0.0001,
|
| 2076 |
+
"loss": 0.8399,
|
| 2077 |
+
"step": 5780
|
| 2078 |
+
},
|
| 2079 |
+
{
|
| 2080 |
+
"epoch": 0.3178256342813305,
|
| 2081 |
+
"grad_norm": 2.257129192352295,
|
| 2082 |
+
"learning_rate": 0.0001,
|
| 2083 |
+
"loss": 0.8224,
|
| 2084 |
+
"step": 5800
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 0.31892158474436955,
|
| 2088 |
+
"grad_norm": 1.8256118297576904,
|
| 2089 |
+
"learning_rate": 0.0001,
|
| 2090 |
+
"loss": 0.794,
|
| 2091 |
+
"step": 5820
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 0.3200175352074086,
|
| 2095 |
+
"grad_norm": 1.8899625539779663,
|
| 2096 |
+
"learning_rate": 0.0001,
|
| 2097 |
+
"loss": 0.8614,
|
| 2098 |
+
"step": 5840
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 0.3211134856704477,
|
| 2102 |
+
"grad_norm": 2.221484661102295,
|
| 2103 |
+
"learning_rate": 0.0001,
|
| 2104 |
+
"loss": 0.765,
|
| 2105 |
+
"step": 5860
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 0.3222094361334868,
|
| 2109 |
+
"grad_norm": 1.796877384185791,
|
| 2110 |
+
"learning_rate": 0.0001,
|
| 2111 |
+
"loss": 0.8359,
|
| 2112 |
+
"step": 5880
|
| 2113 |
+
},
|
| 2114 |
+
{
|
| 2115 |
+
"epoch": 0.32330538659652586,
|
| 2116 |
+
"grad_norm": 1.7495447397232056,
|
| 2117 |
+
"learning_rate": 0.0001,
|
| 2118 |
+
"loss": 0.8688,
|
| 2119 |
+
"step": 5900
|
| 2120 |
+
},
|
| 2121 |
+
{
|
| 2122 |
+
"epoch": 0.3244013370595649,
|
| 2123 |
+
"grad_norm": 2.136664628982544,
|
| 2124 |
+
"learning_rate": 0.0001,
|
| 2125 |
+
"loss": 0.9163,
|
| 2126 |
+
"step": 5920
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 0.325497287522604,
|
| 2130 |
+
"grad_norm": 1.8508238792419434,
|
| 2131 |
+
"learning_rate": 0.0001,
|
| 2132 |
+
"loss": 0.7975,
|
| 2133 |
+
"step": 5940
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 0.32659323798564305,
|
| 2137 |
+
"grad_norm": 2.144523859024048,
|
| 2138 |
+
"learning_rate": 0.0001,
|
| 2139 |
+
"loss": 0.7749,
|
| 2140 |
+
"step": 5960
|
| 2141 |
+
},
|
| 2142 |
+
{
|
| 2143 |
+
"epoch": 0.3276891884486821,
|
| 2144 |
+
"grad_norm": 2.208815336227417,
|
| 2145 |
+
"learning_rate": 0.0001,
|
| 2146 |
+
"loss": 0.8148,
|
| 2147 |
+
"step": 5980
|
| 2148 |
+
},
|
| 2149 |
+
{
|
| 2150 |
+
"epoch": 0.3287851389117212,
|
| 2151 |
+
"grad_norm": 2.0617401599884033,
|
| 2152 |
+
"learning_rate": 0.0001,
|
| 2153 |
+
"loss": 0.8884,
|
| 2154 |
+
"step": 6000
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 0.3287851389117212,
|
| 2158 |
+
"eval_loss": 0.8316722512245178,
|
| 2159 |
+
"eval_runtime": 30850.8589,
|
| 2160 |
+
"eval_samples_per_second": 2.103,
|
| 2161 |
+
"eval_steps_per_second": 0.066,
|
| 2162 |
+
"eval_wer": 45.9960352377659,
|
| 2163 |
+
"step": 6000
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 0.32988108937476024,
|
| 2167 |
+
"grad_norm": 2.0406434535980225,
|
| 2168 |
+
"learning_rate": 0.0001,
|
| 2169 |
+
"loss": 0.8504,
|
| 2170 |
+
"step": 6020
|
| 2171 |
+
},
|
| 2172 |
+
{
|
| 2173 |
+
"epoch": 0.33097703983779936,
|
| 2174 |
+
"grad_norm": 2.1899139881134033,
|
| 2175 |
+
"learning_rate": 0.0001,
|
| 2176 |
+
"loss": 0.7782,
|
| 2177 |
+
"step": 6040
|
| 2178 |
+
},
|
| 2179 |
+
{
|
| 2180 |
+
"epoch": 0.3320729903008384,
|
| 2181 |
+
"grad_norm": 2.650421380996704,
|
| 2182 |
+
"learning_rate": 0.0001,
|
| 2183 |
+
"loss": 0.7823,
|
| 2184 |
+
"step": 6060
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 0.3331689407638775,
|
| 2188 |
+
"grad_norm": 2.085683584213257,
|
| 2189 |
+
"learning_rate": 0.0001,
|
| 2190 |
+
"loss": 0.754,
|
| 2191 |
+
"step": 6080
|
| 2192 |
+
},
|
| 2193 |
+
{
|
| 2194 |
+
"epoch": 0.33426489122691655,
|
| 2195 |
+
"grad_norm": 2.1783502101898193,
|
| 2196 |
+
"learning_rate": 0.0001,
|
| 2197 |
+
"loss": 0.8819,
|
| 2198 |
+
"step": 6100
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 0.3353608416899556,
|
| 2202 |
+
"grad_norm": 2.096208333969116,
|
| 2203 |
+
"learning_rate": 0.0001,
|
| 2204 |
+
"loss": 0.8702,
|
| 2205 |
+
"step": 6120
|
| 2206 |
+
},
|
| 2207 |
+
{
|
| 2208 |
+
"epoch": 0.3364567921529947,
|
| 2209 |
+
"grad_norm": 2.005629062652588,
|
| 2210 |
+
"learning_rate": 0.0001,
|
| 2211 |
+
"loss": 0.8827,
|
| 2212 |
+
"step": 6140
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 0.33755274261603374,
|
| 2216 |
+
"grad_norm": 2.1545634269714355,
|
| 2217 |
+
"learning_rate": 0.0001,
|
| 2218 |
+
"loss": 0.8496,
|
| 2219 |
+
"step": 6160
|
| 2220 |
+
},
|
| 2221 |
+
{
|
| 2222 |
+
"epoch": 0.3386486930790728,
|
| 2223 |
+
"grad_norm": 1.8190851211547852,
|
| 2224 |
+
"learning_rate": 0.0001,
|
| 2225 |
+
"loss": 0.7622,
|
| 2226 |
+
"step": 6180
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 0.3397446435421119,
|
| 2230 |
+
"grad_norm": 1.9555623531341553,
|
| 2231 |
+
"learning_rate": 0.0001,
|
| 2232 |
+
"loss": 0.8338,
|
| 2233 |
+
"step": 6200
|
| 2234 |
+
},
|
| 2235 |
+
{
|
| 2236 |
+
"epoch": 0.340840594005151,
|
| 2237 |
+
"grad_norm": 1.8530341386795044,
|
| 2238 |
+
"learning_rate": 0.0001,
|
| 2239 |
+
"loss": 0.8017,
|
| 2240 |
+
"step": 6220
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 0.34193654446819005,
|
| 2244 |
+
"grad_norm": 1.8724114894866943,
|
| 2245 |
+
"learning_rate": 0.0001,
|
| 2246 |
+
"loss": 0.848,
|
| 2247 |
+
"step": 6240
|
| 2248 |
+
},
|
| 2249 |
+
{
|
| 2250 |
+
"epoch": 0.3430324949312291,
|
| 2251 |
+
"grad_norm": 1.8598796129226685,
|
| 2252 |
+
"learning_rate": 0.0001,
|
| 2253 |
+
"loss": 0.8074,
|
| 2254 |
+
"step": 6260
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 0.3441284453942682,
|
| 2258 |
+
"grad_norm": 2.1442923545837402,
|
| 2259 |
+
"learning_rate": 0.0001,
|
| 2260 |
+
"loss": 0.8473,
|
| 2261 |
+
"step": 6280
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 0.34522439585730724,
|
| 2265 |
+
"grad_norm": 2.3083174228668213,
|
| 2266 |
+
"learning_rate": 0.0001,
|
| 2267 |
+
"loss": 0.9016,
|
| 2268 |
+
"step": 6300
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 0.3463203463203463,
|
| 2272 |
+
"grad_norm": 1.8194735050201416,
|
| 2273 |
+
"learning_rate": 0.0001,
|
| 2274 |
+
"loss": 0.8267,
|
| 2275 |
+
"step": 6320
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 0.34741629678338537,
|
| 2279 |
+
"grad_norm": 2.063523054122925,
|
| 2280 |
+
"learning_rate": 0.0001,
|
| 2281 |
+
"loss": 0.7841,
|
| 2282 |
+
"step": 6340
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 0.3485122472464245,
|
| 2286 |
+
"grad_norm": 2.17594051361084,
|
| 2287 |
+
"learning_rate": 0.0001,
|
| 2288 |
+
"loss": 0.8318,
|
| 2289 |
+
"step": 6360
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 0.34960819770946355,
|
| 2293 |
+
"grad_norm": 1.665189504623413,
|
| 2294 |
+
"learning_rate": 0.0001,
|
| 2295 |
+
"loss": 0.7983,
|
| 2296 |
+
"step": 6380
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 0.3507041481725026,
|
| 2300 |
+
"grad_norm": 2.2596445083618164,
|
| 2301 |
+
"learning_rate": 0.0001,
|
| 2302 |
+
"loss": 0.8421,
|
| 2303 |
+
"step": 6400
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 0.3518000986355417,
|
| 2307 |
+
"grad_norm": 1.7096545696258545,
|
| 2308 |
+
"learning_rate": 0.0001,
|
| 2309 |
+
"loss": 0.889,
|
| 2310 |
+
"step": 6420
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 0.35289604909858074,
|
| 2314 |
+
"grad_norm": 1.7475535869598389,
|
| 2315 |
+
"learning_rate": 0.0001,
|
| 2316 |
+
"loss": 0.8006,
|
| 2317 |
+
"step": 6440
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 0.3539919995616198,
|
| 2321 |
+
"grad_norm": 1.8176007270812988,
|
| 2322 |
+
"learning_rate": 0.0001,
|
| 2323 |
+
"loss": 0.8632,
|
| 2324 |
+
"step": 6460
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 0.35508795002465887,
|
| 2328 |
+
"grad_norm": 2.6806535720825195,
|
| 2329 |
+
"learning_rate": 0.0001,
|
| 2330 |
+
"loss": 0.8427,
|
| 2331 |
+
"step": 6480
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 0.35618390048769794,
|
| 2335 |
+
"grad_norm": 2.094172477722168,
|
| 2336 |
+
"learning_rate": 0.0001,
|
| 2337 |
+
"loss": 0.7812,
|
| 2338 |
+
"step": 6500
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 0.357279850950737,
|
| 2342 |
+
"grad_norm": 1.8341765403747559,
|
| 2343 |
+
"learning_rate": 0.0001,
|
| 2344 |
+
"loss": 0.8051,
|
| 2345 |
+
"step": 6520
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 0.3583758014137761,
|
| 2349 |
+
"grad_norm": 2.2341349124908447,
|
| 2350 |
+
"learning_rate": 0.0001,
|
| 2351 |
+
"loss": 0.8001,
|
| 2352 |
+
"step": 6540
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 0.3594717518768152,
|
| 2356 |
+
"grad_norm": 2.1017801761627197,
|
| 2357 |
+
"learning_rate": 0.0001,
|
| 2358 |
+
"loss": 0.8142,
|
| 2359 |
+
"step": 6560
|
| 2360 |
+
},
|
| 2361 |
+
{
|
| 2362 |
+
"epoch": 0.36056770233985425,
|
| 2363 |
+
"grad_norm": 1.9903994798660278,
|
| 2364 |
+
"learning_rate": 0.0001,
|
| 2365 |
+
"loss": 0.8117,
|
| 2366 |
+
"step": 6580
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 0.3616636528028933,
|
| 2370 |
+
"grad_norm": 2.273465394973755,
|
| 2371 |
+
"learning_rate": 0.0001,
|
| 2372 |
+
"loss": 0.8864,
|
| 2373 |
+
"step": 6600
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 0.3627596032659324,
|
| 2377 |
+
"grad_norm": 2.0767428874969482,
|
| 2378 |
+
"learning_rate": 0.0001,
|
| 2379 |
+
"loss": 0.7687,
|
| 2380 |
+
"step": 6620
|
| 2381 |
+
},
|
| 2382 |
+
{
|
| 2383 |
+
"epoch": 0.36385555372897144,
|
| 2384 |
+
"grad_norm": 2.559774398803711,
|
| 2385 |
+
"learning_rate": 0.0001,
|
| 2386 |
+
"loss": 0.8181,
|
| 2387 |
+
"step": 6640
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 0.3649515041920105,
|
| 2391 |
+
"grad_norm": 2.1393582820892334,
|
| 2392 |
+
"learning_rate": 0.0001,
|
| 2393 |
+
"loss": 0.7936,
|
| 2394 |
+
"step": 6660
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 0.36604745465504956,
|
| 2398 |
+
"grad_norm": 2.06675386428833,
|
| 2399 |
+
"learning_rate": 0.0001,
|
| 2400 |
+
"loss": 0.8263,
|
| 2401 |
+
"step": 6680
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 0.3671434051180887,
|
| 2405 |
+
"grad_norm": 1.7674784660339355,
|
| 2406 |
+
"learning_rate": 0.0001,
|
| 2407 |
+
"loss": 0.7818,
|
| 2408 |
+
"step": 6700
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 0.36823935558112775,
|
| 2412 |
+
"grad_norm": 1.765442132949829,
|
| 2413 |
+
"learning_rate": 0.0001,
|
| 2414 |
+
"loss": 0.8335,
|
| 2415 |
+
"step": 6720
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 0.3693353060441668,
|
| 2419 |
+
"grad_norm": 2.044288158416748,
|
| 2420 |
+
"learning_rate": 0.0001,
|
| 2421 |
+
"loss": 0.8742,
|
| 2422 |
+
"step": 6740
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 0.3704312565072059,
|
| 2426 |
+
"grad_norm": 1.9821726083755493,
|
| 2427 |
+
"learning_rate": 0.0001,
|
| 2428 |
+
"loss": 0.928,
|
| 2429 |
+
"step": 6760
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 0.37152720697024494,
|
| 2433 |
+
"grad_norm": 2.0798370838165283,
|
| 2434 |
+
"learning_rate": 0.0001,
|
| 2435 |
+
"loss": 0.7627,
|
| 2436 |
+
"step": 6780
|
| 2437 |
+
},
|
| 2438 |
+
{
|
| 2439 |
+
"epoch": 0.372623157433284,
|
| 2440 |
+
"grad_norm": 1.6817582845687866,
|
| 2441 |
+
"learning_rate": 0.0001,
|
| 2442 |
+
"loss": 0.7985,
|
| 2443 |
+
"step": 6800
|
| 2444 |
+
},
|
| 2445 |
+
{
|
| 2446 |
+
"epoch": 0.37371910789632307,
|
| 2447 |
+
"grad_norm": 1.872247576713562,
|
| 2448 |
+
"learning_rate": 0.0001,
|
| 2449 |
+
"loss": 0.8102,
|
| 2450 |
+
"step": 6820
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 0.37481505835936213,
|
| 2454 |
+
"grad_norm": 1.7761516571044922,
|
| 2455 |
+
"learning_rate": 0.0001,
|
| 2456 |
+
"loss": 0.8435,
|
| 2457 |
+
"step": 6840
|
| 2458 |
+
},
|
| 2459 |
+
{
|
| 2460 |
+
"epoch": 0.37591100882240125,
|
| 2461 |
+
"grad_norm": 1.739585518836975,
|
| 2462 |
+
"learning_rate": 0.0001,
|
| 2463 |
+
"loss": 0.8706,
|
| 2464 |
+
"step": 6860
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 0.3770069592854403,
|
| 2468 |
+
"grad_norm": 2.0503687858581543,
|
| 2469 |
+
"learning_rate": 0.0001,
|
| 2470 |
+
"loss": 0.8354,
|
| 2471 |
+
"step": 6880
|
| 2472 |
+
},
|
| 2473 |
+
{
|
| 2474 |
+
"epoch": 0.3781029097484794,
|
| 2475 |
+
"grad_norm": 2.283393621444702,
|
| 2476 |
+
"learning_rate": 0.0001,
|
| 2477 |
+
"loss": 0.7476,
|
| 2478 |
+
"step": 6900
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 0.37919886021151844,
|
| 2482 |
+
"grad_norm": 1.801018238067627,
|
| 2483 |
+
"learning_rate": 0.0001,
|
| 2484 |
+
"loss": 0.7817,
|
| 2485 |
+
"step": 6920
|
| 2486 |
+
},
|
| 2487 |
+
{
|
| 2488 |
+
"epoch": 0.3802948106745575,
|
| 2489 |
+
"grad_norm": 2.5343267917633057,
|
| 2490 |
+
"learning_rate": 0.0001,
|
| 2491 |
+
"loss": 0.7628,
|
| 2492 |
+
"step": 6940
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 0.38139076113759657,
|
| 2496 |
+
"grad_norm": 2.010507822036743,
|
| 2497 |
+
"learning_rate": 0.0001,
|
| 2498 |
+
"loss": 0.7931,
|
| 2499 |
+
"step": 6960
|
| 2500 |
+
},
|
| 2501 |
+
{
|
| 2502 |
+
"epoch": 0.38248671160063563,
|
| 2503 |
+
"grad_norm": 1.7228796482086182,
|
| 2504 |
+
"learning_rate": 0.0001,
|
| 2505 |
+
"loss": 0.7517,
|
| 2506 |
+
"step": 6980
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 0.3835826620636747,
|
| 2510 |
+
"grad_norm": 1.967822551727295,
|
| 2511 |
+
"learning_rate": 0.0001,
|
| 2512 |
+
"loss": 0.804,
|
| 2513 |
+
"step": 7000
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 0.3835826620636747,
|
| 2517 |
+
"eval_loss": 0.7978512644767761,
|
| 2518 |
+
"eval_runtime": 30977.7517,
|
| 2519 |
+
"eval_samples_per_second": 2.095,
|
| 2520 |
+
"eval_steps_per_second": 0.065,
|
| 2521 |
+
"eval_wer": 61.261910549759826,
|
| 2522 |
+
"step": 7000
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 0.3846786125267138,
|
| 2526 |
+
"grad_norm": 1.9999229907989502,
|
| 2527 |
+
"learning_rate": 0.0001,
|
| 2528 |
+
"loss": 0.7634,
|
| 2529 |
+
"step": 7020
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 0.3857745629897529,
|
| 2533 |
+
"grad_norm": 1.956128716468811,
|
| 2534 |
+
"learning_rate": 0.0001,
|
| 2535 |
+
"loss": 0.8102,
|
| 2536 |
+
"step": 7040
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 0.38687051345279194,
|
| 2540 |
+
"grad_norm": 2.0134966373443604,
|
| 2541 |
+
"learning_rate": 0.0001,
|
| 2542 |
+
"loss": 0.7957,
|
| 2543 |
+
"step": 7060
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 0.387966463915831,
|
| 2547 |
+
"grad_norm": 2.0373167991638184,
|
| 2548 |
+
"learning_rate": 0.0001,
|
| 2549 |
+
"loss": 0.8251,
|
| 2550 |
+
"step": 7080
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 0.38906241437887007,
|
| 2554 |
+
"grad_norm": 1.7772964239120483,
|
| 2555 |
+
"learning_rate": 0.0001,
|
| 2556 |
+
"loss": 0.8128,
|
| 2557 |
+
"step": 7100
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 0.39015836484190913,
|
| 2561 |
+
"grad_norm": 1.7618379592895508,
|
| 2562 |
+
"learning_rate": 0.0001,
|
| 2563 |
+
"loss": 0.8345,
|
| 2564 |
+
"step": 7120
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.3912543153049482,
|
| 2568 |
+
"grad_norm": 2.181671380996704,
|
| 2569 |
+
"learning_rate": 0.0001,
|
| 2570 |
+
"loss": 0.8345,
|
| 2571 |
+
"step": 7140
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 0.39235026576798726,
|
| 2575 |
+
"grad_norm": 1.8794726133346558,
|
| 2576 |
+
"learning_rate": 0.0001,
|
| 2577 |
+
"loss": 0.7615,
|
| 2578 |
+
"step": 7160
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 0.3934462162310264,
|
| 2582 |
+
"grad_norm": 1.9297798871994019,
|
| 2583 |
+
"learning_rate": 0.0001,
|
| 2584 |
+
"loss": 0.7618,
|
| 2585 |
+
"step": 7180
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 0.39454216669406544,
|
| 2589 |
+
"grad_norm": 1.9441471099853516,
|
| 2590 |
+
"learning_rate": 0.0001,
|
| 2591 |
+
"loss": 0.859,
|
| 2592 |
+
"step": 7200
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 0.3956381171571045,
|
| 2596 |
+
"grad_norm": 2.2561404705047607,
|
| 2597 |
+
"learning_rate": 0.0001,
|
| 2598 |
+
"loss": 0.7877,
|
| 2599 |
+
"step": 7220
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 0.39673406762014357,
|
| 2603 |
+
"grad_norm": 1.8441416025161743,
|
| 2604 |
+
"learning_rate": 0.0001,
|
| 2605 |
+
"loss": 0.7734,
|
| 2606 |
+
"step": 7240
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 0.39783001808318263,
|
| 2610 |
+
"grad_norm": 1.686120867729187,
|
| 2611 |
+
"learning_rate": 0.0001,
|
| 2612 |
+
"loss": 0.7066,
|
| 2613 |
+
"step": 7260
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 0.3989259685462217,
|
| 2617 |
+
"grad_norm": 1.9456263780593872,
|
| 2618 |
+
"learning_rate": 0.0001,
|
| 2619 |
+
"loss": 0.7469,
|
| 2620 |
+
"step": 7280
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 0.40002191900926076,
|
| 2624 |
+
"grad_norm": 1.9112725257873535,
|
| 2625 |
+
"learning_rate": 0.0001,
|
| 2626 |
+
"loss": 0.7607,
|
| 2627 |
+
"step": 7300
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 0.4011178694722998,
|
| 2631 |
+
"grad_norm": 2.5668513774871826,
|
| 2632 |
+
"learning_rate": 0.0001,
|
| 2633 |
+
"loss": 0.7859,
|
| 2634 |
+
"step": 7320
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 0.40221381993533895,
|
| 2638 |
+
"grad_norm": 1.9502942562103271,
|
| 2639 |
+
"learning_rate": 0.0001,
|
| 2640 |
+
"loss": 0.7607,
|
| 2641 |
+
"step": 7340
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 0.403309770398378,
|
| 2645 |
+
"grad_norm": 1.6973525285720825,
|
| 2646 |
+
"learning_rate": 0.0001,
|
| 2647 |
+
"loss": 0.8313,
|
| 2648 |
+
"step": 7360
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 0.4044057208614171,
|
| 2652 |
+
"grad_norm": 2.3962297439575195,
|
| 2653 |
+
"learning_rate": 0.0001,
|
| 2654 |
+
"loss": 0.7806,
|
| 2655 |
+
"step": 7380
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 0.40550167132445614,
|
| 2659 |
+
"grad_norm": 1.887536883354187,
|
| 2660 |
+
"learning_rate": 0.0001,
|
| 2661 |
+
"loss": 0.7524,
|
| 2662 |
+
"step": 7400
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 0.4065976217874952,
|
| 2666 |
+
"grad_norm": 1.999687910079956,
|
| 2667 |
+
"learning_rate": 0.0001,
|
| 2668 |
+
"loss": 0.7349,
|
| 2669 |
+
"step": 7420
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 0.40769357225053426,
|
| 2673 |
+
"grad_norm": 1.7444576025009155,
|
| 2674 |
+
"learning_rate": 0.0001,
|
| 2675 |
+
"loss": 0.8156,
|
| 2676 |
+
"step": 7440
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 0.40878952271357333,
|
| 2680 |
+
"grad_norm": 1.7175132036209106,
|
| 2681 |
+
"learning_rate": 0.0001,
|
| 2682 |
+
"loss": 0.7419,
|
| 2683 |
+
"step": 7460
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 0.4098854731766124,
|
| 2687 |
+
"grad_norm": 2.23638653755188,
|
| 2688 |
+
"learning_rate": 0.0001,
|
| 2689 |
+
"loss": 0.666,
|
| 2690 |
+
"step": 7480
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 0.4109814236396515,
|
| 2694 |
+
"grad_norm": 2.024102210998535,
|
| 2695 |
+
"learning_rate": 0.0001,
|
| 2696 |
+
"loss": 0.7541,
|
| 2697 |
+
"step": 7500
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 0.4120773741026906,
|
| 2701 |
+
"grad_norm": 2.042541265487671,
|
| 2702 |
+
"learning_rate": 0.0001,
|
| 2703 |
+
"loss": 0.7915,
|
| 2704 |
+
"step": 7520
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 0.41317332456572964,
|
| 2708 |
+
"grad_norm": 1.9140897989273071,
|
| 2709 |
+
"learning_rate": 0.0001,
|
| 2710 |
+
"loss": 0.8712,
|
| 2711 |
+
"step": 7540
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 0.4142692750287687,
|
| 2715 |
+
"grad_norm": 1.8435416221618652,
|
| 2716 |
+
"learning_rate": 0.0001,
|
| 2717 |
+
"loss": 0.8241,
|
| 2718 |
+
"step": 7560
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 0.41536522549180777,
|
| 2722 |
+
"grad_norm": 2.027944803237915,
|
| 2723 |
+
"learning_rate": 0.0001,
|
| 2724 |
+
"loss": 0.9422,
|
| 2725 |
+
"step": 7580
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 0.41646117595484683,
|
| 2729 |
+
"grad_norm": 2.07381534576416,
|
| 2730 |
+
"learning_rate": 0.0001,
|
| 2731 |
+
"loss": 0.812,
|
| 2732 |
+
"step": 7600
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 0.4175571264178859,
|
| 2736 |
+
"grad_norm": 1.9762136936187744,
|
| 2737 |
+
"learning_rate": 0.0001,
|
| 2738 |
+
"loss": 0.7852,
|
| 2739 |
+
"step": 7620
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 0.41865307688092496,
|
| 2743 |
+
"grad_norm": 1.8222426176071167,
|
| 2744 |
+
"learning_rate": 0.0001,
|
| 2745 |
+
"loss": 0.752,
|
| 2746 |
+
"step": 7640
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 0.4197490273439641,
|
| 2750 |
+
"grad_norm": 2.0519089698791504,
|
| 2751 |
+
"learning_rate": 0.0001,
|
| 2752 |
+
"loss": 0.8031,
|
| 2753 |
+
"step": 7660
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 0.42084497780700314,
|
| 2757 |
+
"grad_norm": 1.8777110576629639,
|
| 2758 |
+
"learning_rate": 0.0001,
|
| 2759 |
+
"loss": 0.8173,
|
| 2760 |
+
"step": 7680
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 0.4219409282700422,
|
| 2764 |
+
"grad_norm": 2.323411703109741,
|
| 2765 |
+
"learning_rate": 0.0001,
|
| 2766 |
+
"loss": 0.8479,
|
| 2767 |
+
"step": 7700
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 0.42303687873308127,
|
| 2771 |
+
"grad_norm": 1.6403400897979736,
|
| 2772 |
+
"learning_rate": 0.0001,
|
| 2773 |
+
"loss": 0.7567,
|
| 2774 |
+
"step": 7720
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 0.42413282919612033,
|
| 2778 |
+
"grad_norm": 1.6627925634384155,
|
| 2779 |
+
"learning_rate": 0.0001,
|
| 2780 |
+
"loss": 0.7734,
|
| 2781 |
+
"step": 7740
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 0.4252287796591594,
|
| 2785 |
+
"grad_norm": 1.8771709203720093,
|
| 2786 |
+
"learning_rate": 0.0001,
|
| 2787 |
+
"loss": 0.7652,
|
| 2788 |
+
"step": 7760
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 0.42632473012219846,
|
| 2792 |
+
"grad_norm": 1.9806597232818604,
|
| 2793 |
+
"learning_rate": 0.0001,
|
| 2794 |
+
"loss": 0.7699,
|
| 2795 |
+
"step": 7780
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 0.4274206805852375,
|
| 2799 |
+
"grad_norm": 2.1376988887786865,
|
| 2800 |
+
"learning_rate": 0.0001,
|
| 2801 |
+
"loss": 0.7825,
|
| 2802 |
+
"step": 7800
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 0.42851663104827664,
|
| 2806 |
+
"grad_norm": 1.5566449165344238,
|
| 2807 |
+
"learning_rate": 0.0001,
|
| 2808 |
+
"loss": 0.704,
|
| 2809 |
+
"step": 7820
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 0.4296125815113157,
|
| 2813 |
+
"grad_norm": 2.1835947036743164,
|
| 2814 |
+
"learning_rate": 0.0001,
|
| 2815 |
+
"loss": 0.8101,
|
| 2816 |
+
"step": 7840
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 0.43070853197435477,
|
| 2820 |
+
"grad_norm": 2.055119037628174,
|
| 2821 |
+
"learning_rate": 0.0001,
|
| 2822 |
+
"loss": 0.703,
|
| 2823 |
+
"step": 7860
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 0.43180448243739383,
|
| 2827 |
+
"grad_norm": 1.9324967861175537,
|
| 2828 |
+
"learning_rate": 0.0001,
|
| 2829 |
+
"loss": 0.81,
|
| 2830 |
+
"step": 7880
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 0.4329004329004329,
|
| 2834 |
+
"grad_norm": 2.1087846755981445,
|
| 2835 |
+
"learning_rate": 0.0001,
|
| 2836 |
+
"loss": 0.7676,
|
| 2837 |
+
"step": 7900
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 0.43399638336347196,
|
| 2841 |
+
"grad_norm": 1.8521897792816162,
|
| 2842 |
+
"learning_rate": 0.0001,
|
| 2843 |
+
"loss": 0.7546,
|
| 2844 |
+
"step": 7920
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 0.435092333826511,
|
| 2848 |
+
"grad_norm": 2.145947217941284,
|
| 2849 |
+
"learning_rate": 0.0001,
|
| 2850 |
+
"loss": 0.7992,
|
| 2851 |
+
"step": 7940
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 0.4361882842895501,
|
| 2855 |
+
"grad_norm": 1.7739931344985962,
|
| 2856 |
+
"learning_rate": 0.0001,
|
| 2857 |
+
"loss": 0.7133,
|
| 2858 |
+
"step": 7960
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 0.4372842347525892,
|
| 2862 |
+
"grad_norm": 1.6032921075820923,
|
| 2863 |
+
"learning_rate": 0.0001,
|
| 2864 |
+
"loss": 0.8207,
|
| 2865 |
+
"step": 7980
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 0.43838018521562827,
|
| 2869 |
+
"grad_norm": 2.1895668506622314,
|
| 2870 |
+
"learning_rate": 0.0001,
|
| 2871 |
+
"loss": 0.7638,
|
| 2872 |
+
"step": 8000
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 0.43838018521562827,
|
| 2876 |
+
"eval_loss": 0.770411491394043,
|
| 2877 |
+
"eval_runtime": 30675.7059,
|
| 2878 |
+
"eval_samples_per_second": 2.115,
|
| 2879 |
+
"eval_steps_per_second": 0.066,
|
| 2880 |
+
"eval_wer": 43.10069742838263,
|
| 2881 |
+
"step": 8000
|
| 2882 |
+
}
|
| 2883 |
+
],
|
| 2884 |
+
"logging_steps": 20,
|
| 2885 |
+
"max_steps": 54747,
|
| 2886 |
+
"num_input_tokens_seen": 0,
|
| 2887 |
+
"num_train_epochs": 3,
|
| 2888 |
+
"save_steps": 1000,
|
| 2889 |
+
"stateful_callbacks": {
|
| 2890 |
+
"TrainerControl": {
|
| 2891 |
+
"args": {
|
| 2892 |
+
"should_epoch_stop": false,
|
| 2893 |
+
"should_evaluate": false,
|
| 2894 |
+
"should_log": false,
|
| 2895 |
+
"should_save": true,
|
| 2896 |
+
"should_training_stop": false
|
| 2897 |
+
},
|
| 2898 |
+
"attributes": {}
|
| 2899 |
+
}
|
| 2900 |
+
},
|
| 2901 |
+
"total_flos": 3.578723706312065e+20,
|
| 2902 |
+
"train_batch_size": 8,
|
| 2903 |
+
"trial_name": null,
|
| 2904 |
+
"trial_params": null
|
| 2905 |
+
}
|