Commit
•
81e9b06
1
Parent(s):
8e56be0
Upload trainer_state.json
Browse files- trainer_state.json +1486 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1486 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.0,
|
| 3 |
+
"best_model_checkpoint": "./models/checkpoint-602",
|
| 4 |
+
"epoch": 98.0,
|
| 5 |
+
"global_step": 58996,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"learning_rate": 5.99e-05,
|
| 13 |
+
"loss": 9.7183,
|
| 14 |
+
"step": 602
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.0,
|
| 18 |
+
"eval_loss": 3.9950010776519775,
|
| 19 |
+
"eval_runtime": 377.3539,
|
| 20 |
+
"eval_samples_per_second": 12.24,
|
| 21 |
+
"eval_steps_per_second": 0.154,
|
| 22 |
+
"eval_wer": 1.0,
|
| 23 |
+
"step": 602
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 2.0,
|
| 27 |
+
"learning_rate": 9.966047297297298e-05,
|
| 28 |
+
"loss": 2.4521,
|
| 29 |
+
"step": 1204
|
| 30 |
+
},
|
| 31 |
+
{
|
| 32 |
+
"epoch": 2.0,
|
| 33 |
+
"eval_loss": 0.7947585582733154,
|
| 34 |
+
"eval_runtime": 335.0691,
|
| 35 |
+
"eval_samples_per_second": 13.785,
|
| 36 |
+
"eval_steps_per_second": 0.173,
|
| 37 |
+
"eval_wer": 0.7436259622854722,
|
| 38 |
+
"step": 1204
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 3.0,
|
| 42 |
+
"learning_rate": 9.864358108108108e-05,
|
| 43 |
+
"loss": 0.7245,
|
| 44 |
+
"step": 1806
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 3.0,
|
| 48 |
+
"eval_loss": 0.5130816698074341,
|
| 49 |
+
"eval_runtime": 367.2509,
|
| 50 |
+
"eval_samples_per_second": 12.577,
|
| 51 |
+
"eval_steps_per_second": 0.158,
|
| 52 |
+
"eval_wer": 0.6208418673635143,
|
| 53 |
+
"step": 1806
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 4.0,
|
| 57 |
+
"learning_rate": 9.762668918918919e-05,
|
| 58 |
+
"loss": 0.5488,
|
| 59 |
+
"step": 2408
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 4.0,
|
| 63 |
+
"eval_loss": 0.4323376417160034,
|
| 64 |
+
"eval_runtime": 384.1184,
|
| 65 |
+
"eval_samples_per_second": 12.025,
|
| 66 |
+
"eval_steps_per_second": 0.151,
|
| 67 |
+
"eval_wer": 0.5652941591920333,
|
| 68 |
+
"step": 2408
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 5.0,
|
| 72 |
+
"learning_rate": 9.66097972972973e-05,
|
| 73 |
+
"loss": 0.4613,
|
| 74 |
+
"step": 3010
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 5.0,
|
| 78 |
+
"eval_loss": 0.3905767500400543,
|
| 79 |
+
"eval_runtime": 369.1827,
|
| 80 |
+
"eval_samples_per_second": 12.511,
|
| 81 |
+
"eval_steps_per_second": 0.157,
|
| 82 |
+
"eval_wer": 0.5313934599901123,
|
| 83 |
+
"step": 3010
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 6.0,
|
| 87 |
+
"learning_rate": 9.559290540540541e-05,
|
| 88 |
+
"loss": 0.4138,
|
| 89 |
+
"step": 3612
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 6.0,
|
| 93 |
+
"eval_loss": 0.36640074849128723,
|
| 94 |
+
"eval_runtime": 327.6142,
|
| 95 |
+
"eval_samples_per_second": 14.099,
|
| 96 |
+
"eval_steps_per_second": 0.177,
|
| 97 |
+
"eval_wer": 0.5138074722791157,
|
| 98 |
+
"step": 3612
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 7.0,
|
| 102 |
+
"learning_rate": 9.457601351351352e-05,
|
| 103 |
+
"loss": 0.3768,
|
| 104 |
+
"step": 4214
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 7.0,
|
| 108 |
+
"eval_loss": 0.34569260478019714,
|
| 109 |
+
"eval_runtime": 320.036,
|
| 110 |
+
"eval_samples_per_second": 14.433,
|
| 111 |
+
"eval_steps_per_second": 0.181,
|
| 112 |
+
"eval_wer": 0.49265484850625046,
|
| 113 |
+
"step": 4214
|
| 114 |
+
},
|
| 115 |
+
{
|
| 116 |
+
"epoch": 8.0,
|
| 117 |
+
"learning_rate": 9.355912162162163e-05,
|
| 118 |
+
"loss": 0.3493,
|
| 119 |
+
"step": 4816
|
| 120 |
+
},
|
| 121 |
+
{
|
| 122 |
+
"epoch": 8.0,
|
| 123 |
+
"eval_loss": 0.3470996916294098,
|
| 124 |
+
"eval_runtime": 322.8379,
|
| 125 |
+
"eval_samples_per_second": 14.307,
|
| 126 |
+
"eval_steps_per_second": 0.18,
|
| 127 |
+
"eval_wer": 0.48280245780069214,
|
| 128 |
+
"step": 4816
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 9.0,
|
| 132 |
+
"learning_rate": 9.254222972972973e-05,
|
| 133 |
+
"loss": 0.3278,
|
| 134 |
+
"step": 5418
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 9.0,
|
| 138 |
+
"eval_loss": 0.329658180475235,
|
| 139 |
+
"eval_runtime": 351.53,
|
| 140 |
+
"eval_samples_per_second": 13.14,
|
| 141 |
+
"eval_steps_per_second": 0.165,
|
| 142 |
+
"eval_wer": 0.4710431527650258,
|
| 143 |
+
"step": 5418
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 10.0,
|
| 147 |
+
"learning_rate": 9.152533783783784e-05,
|
| 148 |
+
"loss": 0.3042,
|
| 149 |
+
"step": 6020
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 10.0,
|
| 153 |
+
"eval_loss": 0.3340738117694855,
|
| 154 |
+
"eval_runtime": 307.6844,
|
| 155 |
+
"eval_samples_per_second": 15.012,
|
| 156 |
+
"eval_steps_per_second": 0.189,
|
| 157 |
+
"eval_wer": 0.4614732678861502,
|
| 158 |
+
"step": 6020
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 11.0,
|
| 162 |
+
"learning_rate": 9.050844594594595e-05,
|
| 163 |
+
"loss": 0.2902,
|
| 164 |
+
"step": 6622
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 11.0,
|
| 168 |
+
"eval_loss": 0.31241220235824585,
|
| 169 |
+
"eval_runtime": 313.7253,
|
| 170 |
+
"eval_samples_per_second": 14.723,
|
| 171 |
+
"eval_steps_per_second": 0.185,
|
| 172 |
+
"eval_wer": 0.4491842644254538,
|
| 173 |
+
"step": 6622
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 12.0,
|
| 177 |
+
"learning_rate": 8.949155405405406e-05,
|
| 178 |
+
"loss": 0.2748,
|
| 179 |
+
"step": 7224
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 12.0,
|
| 183 |
+
"eval_loss": 0.3046228885650635,
|
| 184 |
+
"eval_runtime": 310.6627,
|
| 185 |
+
"eval_samples_per_second": 14.868,
|
| 186 |
+
"eval_steps_per_second": 0.187,
|
| 187 |
+
"eval_wer": 0.4424747510417402,
|
| 188 |
+
"step": 7224
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 13.0,
|
| 192 |
+
"learning_rate": 8.847466216216217e-05,
|
| 193 |
+
"loss": 0.263,
|
| 194 |
+
"step": 7826
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 13.0,
|
| 198 |
+
"eval_loss": 0.3139520585536957,
|
| 199 |
+
"eval_runtime": 306.886,
|
| 200 |
+
"eval_samples_per_second": 15.051,
|
| 201 |
+
"eval_steps_per_second": 0.189,
|
| 202 |
+
"eval_wer": 0.4403206441132848,
|
| 203 |
+
"step": 7826
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 14.0,
|
| 207 |
+
"learning_rate": 8.745945945945946e-05,
|
| 208 |
+
"loss": 0.2506,
|
| 209 |
+
"step": 8428
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 14.0,
|
| 213 |
+
"eval_loss": 0.3264484107494354,
|
| 214 |
+
"eval_runtime": 314.9752,
|
| 215 |
+
"eval_samples_per_second": 14.665,
|
| 216 |
+
"eval_steps_per_second": 0.184,
|
| 217 |
+
"eval_wer": 0.4372837064764461,
|
| 218 |
+
"step": 8428
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 15.0,
|
| 222 |
+
"learning_rate": 8.644256756756757e-05,
|
| 223 |
+
"loss": 0.2395,
|
| 224 |
+
"step": 9030
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 15.0,
|
| 228 |
+
"eval_loss": 0.3298631012439728,
|
| 229 |
+
"eval_runtime": 313.3642,
|
| 230 |
+
"eval_samples_per_second": 14.74,
|
| 231 |
+
"eval_steps_per_second": 0.185,
|
| 232 |
+
"eval_wer": 0.4310332650610919,
|
| 233 |
+
"step": 9030
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 16.0,
|
| 237 |
+
"learning_rate": 8.542567567567568e-05,
|
| 238 |
+
"loss": 0.2289,
|
| 239 |
+
"step": 9632
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 16.0,
|
| 243 |
+
"eval_loss": 0.32563555240631104,
|
| 244 |
+
"eval_runtime": 316.6154,
|
| 245 |
+
"eval_samples_per_second": 14.589,
|
| 246 |
+
"eval_steps_per_second": 0.183,
|
| 247 |
+
"eval_wer": 0.4247828236457377,
|
| 248 |
+
"step": 9632
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 17.0,
|
| 252 |
+
"learning_rate": 8.440878378378379e-05,
|
| 253 |
+
"loss": 0.2229,
|
| 254 |
+
"step": 10234
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 17.0,
|
| 258 |
+
"eval_loss": 0.30847427248954773,
|
| 259 |
+
"eval_runtime": 314.5985,
|
| 260 |
+
"eval_samples_per_second": 14.682,
|
| 261 |
+
"eval_steps_per_second": 0.184,
|
| 262 |
+
"eval_wer": 0.4183205028603715,
|
| 263 |
+
"step": 10234
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 18.0,
|
| 267 |
+
"learning_rate": 8.339358108108109e-05,
|
| 268 |
+
"loss": 0.2123,
|
| 269 |
+
"step": 10836
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 18.0,
|
| 273 |
+
"eval_loss": 0.32277214527130127,
|
| 274 |
+
"eval_runtime": 319.0938,
|
| 275 |
+
"eval_samples_per_second": 14.475,
|
| 276 |
+
"eval_steps_per_second": 0.182,
|
| 277 |
+
"eval_wer": 0.41549544459354476,
|
| 278 |
+
"step": 10836
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 19.0,
|
| 282 |
+
"learning_rate": 8.23766891891892e-05,
|
| 283 |
+
"loss": 0.2059,
|
| 284 |
+
"step": 11438
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 19.0,
|
| 288 |
+
"eval_loss": 0.32871213555336,
|
| 289 |
+
"eval_runtime": 322.1095,
|
| 290 |
+
"eval_samples_per_second": 14.34,
|
| 291 |
+
"eval_steps_per_second": 0.18,
|
| 292 |
+
"eval_wer": 0.4104103397132566,
|
| 293 |
+
"step": 11438
|
| 294 |
+
},
|
| 295 |
+
{
|
| 296 |
+
"epoch": 20.0,
|
| 297 |
+
"learning_rate": 8.135979729729731e-05,
|
| 298 |
+
"loss": 0.2011,
|
| 299 |
+
"step": 12040
|
| 300 |
+
},
|
| 301 |
+
{
|
| 302 |
+
"epoch": 20.0,
|
| 303 |
+
"eval_loss": 0.3290630877017975,
|
| 304 |
+
"eval_runtime": 327.3838,
|
| 305 |
+
"eval_samples_per_second": 14.109,
|
| 306 |
+
"eval_steps_per_second": 0.177,
|
| 307 |
+
"eval_wer": 0.4098100148315559,
|
| 308 |
+
"step": 12040
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 21.0,
|
| 312 |
+
"learning_rate": 8.03429054054054e-05,
|
| 313 |
+
"loss": 0.1948,
|
| 314 |
+
"step": 12642
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 21.0,
|
| 318 |
+
"eval_loss": 0.3254837393760681,
|
| 319 |
+
"eval_runtime": 343.6765,
|
| 320 |
+
"eval_samples_per_second": 13.44,
|
| 321 |
+
"eval_steps_per_second": 0.169,
|
| 322 |
+
"eval_wer": 0.4006638886927043,
|
| 323 |
+
"step": 12642
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 22.0,
|
| 327 |
+
"learning_rate": 7.932770270270272e-05,
|
| 328 |
+
"loss": 0.1884,
|
| 329 |
+
"step": 13244
|
| 330 |
+
},
|
| 331 |
+
{
|
| 332 |
+
"epoch": 22.0,
|
| 333 |
+
"eval_loss": 0.3118044137954712,
|
| 334 |
+
"eval_runtime": 335.2687,
|
| 335 |
+
"eval_samples_per_second": 13.777,
|
| 336 |
+
"eval_steps_per_second": 0.173,
|
| 337 |
+
"eval_wer": 0.39858040822091956,
|
| 338 |
+
"step": 13244
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 23.0,
|
| 342 |
+
"learning_rate": 7.831081081081081e-05,
|
| 343 |
+
"loss": 0.1806,
|
| 344 |
+
"step": 13846
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 23.0,
|
| 348 |
+
"eval_loss": 0.3262838125228882,
|
| 349 |
+
"eval_runtime": 390.3862,
|
| 350 |
+
"eval_samples_per_second": 11.832,
|
| 351 |
+
"eval_steps_per_second": 0.149,
|
| 352 |
+
"eval_wer": 0.3974150716858535,
|
| 353 |
+
"step": 13846
|
| 354 |
+
},
|
| 355 |
+
{
|
| 356 |
+
"epoch": 24.0,
|
| 357 |
+
"learning_rate": 7.729391891891892e-05,
|
| 358 |
+
"loss": 0.1782,
|
| 359 |
+
"step": 14448
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 24.0,
|
| 363 |
+
"eval_loss": 0.3207845687866211,
|
| 364 |
+
"eval_runtime": 360.7865,
|
| 365 |
+
"eval_samples_per_second": 12.803,
|
| 366 |
+
"eval_steps_per_second": 0.161,
|
| 367 |
+
"eval_wer": 0.39123525672717,
|
| 368 |
+
"step": 14448
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 25.0,
|
| 372 |
+
"learning_rate": 7.627702702702702e-05,
|
| 373 |
+
"loss": 0.1729,
|
| 374 |
+
"step": 15050
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 25.0,
|
| 378 |
+
"eval_loss": 0.3174630105495453,
|
| 379 |
+
"eval_runtime": 356.0372,
|
| 380 |
+
"eval_samples_per_second": 12.973,
|
| 381 |
+
"eval_steps_per_second": 0.163,
|
| 382 |
+
"eval_wer": 0.3948372060173741,
|
| 383 |
+
"step": 15050
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 26.0,
|
| 387 |
+
"learning_rate": 7.526013513513513e-05,
|
| 388 |
+
"loss": 0.1687,
|
| 389 |
+
"step": 15652
|
| 390 |
+
},
|
| 391 |
+
{
|
| 392 |
+
"epoch": 26.0,
|
| 393 |
+
"eval_loss": 0.3447326123714447,
|
| 394 |
+
"eval_runtime": 369.502,
|
| 395 |
+
"eval_samples_per_second": 12.501,
|
| 396 |
+
"eval_steps_per_second": 0.157,
|
| 397 |
+
"eval_wer": 0.38544388728017515,
|
| 398 |
+
"step": 15652
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 27.0,
|
| 402 |
+
"learning_rate": 7.424324324324324e-05,
|
| 403 |
+
"loss": 0.167,
|
| 404 |
+
"step": 16254
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 27.0,
|
| 408 |
+
"eval_loss": 0.3525393605232239,
|
| 409 |
+
"eval_runtime": 355.9444,
|
| 410 |
+
"eval_samples_per_second": 12.977,
|
| 411 |
+
"eval_steps_per_second": 0.163,
|
| 412 |
+
"eval_wer": 0.3901052334204393,
|
| 413 |
+
"step": 16254
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 28.0,
|
| 417 |
+
"learning_rate": 7.322635135135135e-05,
|
| 418 |
+
"loss": 0.1619,
|
| 419 |
+
"step": 16856
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 28.0,
|
| 423 |
+
"eval_loss": 0.3377859592437744,
|
| 424 |
+
"eval_runtime": 331.4224,
|
| 425 |
+
"eval_samples_per_second": 13.937,
|
| 426 |
+
"eval_steps_per_second": 0.175,
|
| 427 |
+
"eval_wer": 0.3853732608235045,
|
| 428 |
+
"step": 16856
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 29.0,
|
| 432 |
+
"learning_rate": 7.220945945945946e-05,
|
| 433 |
+
"loss": 0.1549,
|
| 434 |
+
"step": 17458
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 29.0,
|
| 438 |
+
"eval_loss": 0.35701897740364075,
|
| 439 |
+
"eval_runtime": 327.8021,
|
| 440 |
+
"eval_samples_per_second": 14.091,
|
| 441 |
+
"eval_steps_per_second": 0.177,
|
| 442 |
+
"eval_wer": 0.3870682957836005,
|
| 443 |
+
"step": 17458
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 30.0,
|
| 447 |
+
"learning_rate": 7.119425675675676e-05,
|
| 448 |
+
"loss": 0.1543,
|
| 449 |
+
"step": 18060
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 30.0,
|
| 453 |
+
"eval_loss": 0.3298207223415375,
|
| 454 |
+
"eval_runtime": 329.7188,
|
| 455 |
+
"eval_samples_per_second": 14.009,
|
| 456 |
+
"eval_steps_per_second": 0.176,
|
| 457 |
+
"eval_wer": 0.38085316759658167,
|
| 458 |
+
"step": 18060
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 31.0,
|
| 462 |
+
"learning_rate": 7.017736486486487e-05,
|
| 463 |
+
"loss": 0.15,
|
| 464 |
+
"step": 18662
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 31.0,
|
| 468 |
+
"eval_loss": 0.34915244579315186,
|
| 469 |
+
"eval_runtime": 362.9462,
|
| 470 |
+
"eval_samples_per_second": 12.726,
|
| 471 |
+
"eval_steps_per_second": 0.16,
|
| 472 |
+
"eval_wer": 0.3779221696447489,
|
| 473 |
+
"step": 18662
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 32.0,
|
| 477 |
+
"learning_rate": 6.916047297297298e-05,
|
| 478 |
+
"loss": 0.1459,
|
| 479 |
+
"step": 19264
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 32.0,
|
| 483 |
+
"eval_loss": 0.3343390226364136,
|
| 484 |
+
"eval_runtime": 363.1934,
|
| 485 |
+
"eval_samples_per_second": 12.718,
|
| 486 |
+
"eval_steps_per_second": 0.16,
|
| 487 |
+
"eval_wer": 0.37530899074793417,
|
| 488 |
+
"step": 19264
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 33.0,
|
| 492 |
+
"learning_rate": 6.814527027027028e-05,
|
| 493 |
+
"loss": 0.1439,
|
| 494 |
+
"step": 19866
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 33.0,
|
| 498 |
+
"eval_loss": 0.3707464039325714,
|
| 499 |
+
"eval_runtime": 319.1722,
|
| 500 |
+
"eval_samples_per_second": 14.472,
|
| 501 |
+
"eval_steps_per_second": 0.182,
|
| 502 |
+
"eval_wer": 0.3783106151564376,
|
| 503 |
+
"step": 19866
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 34.0,
|
| 507 |
+
"learning_rate": 6.713006756756756e-05,
|
| 508 |
+
"loss": 0.1386,
|
| 509 |
+
"step": 20468
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 34.0,
|
| 513 |
+
"eval_loss": 0.3442770838737488,
|
| 514 |
+
"eval_runtime": 326.9396,
|
| 515 |
+
"eval_samples_per_second": 14.128,
|
| 516 |
+
"eval_steps_per_second": 0.177,
|
| 517 |
+
"eval_wer": 0.3738964616145208,
|
| 518 |
+
"step": 20468
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 35.0,
|
| 522 |
+
"learning_rate": 6.611317567567568e-05,
|
| 523 |
+
"loss": 0.1374,
|
| 524 |
+
"step": 21070
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 35.0,
|
| 528 |
+
"eval_loss": 0.36009541153907776,
|
| 529 |
+
"eval_runtime": 311.0687,
|
| 530 |
+
"eval_samples_per_second": 14.849,
|
| 531 |
+
"eval_steps_per_second": 0.186,
|
| 532 |
+
"eval_wer": 0.37350801610283213,
|
| 533 |
+
"step": 21070
|
| 534 |
+
},
|
| 535 |
+
{
|
| 536 |
+
"epoch": 36.0,
|
| 537 |
+
"learning_rate": 6.509628378378379e-05,
|
| 538 |
+
"loss": 0.1334,
|
| 539 |
+
"step": 21672
|
| 540 |
+
},
|
| 541 |
+
{
|
| 542 |
+
"epoch": 36.0,
|
| 543 |
+
"eval_loss": 0.37101128697395325,
|
| 544 |
+
"eval_runtime": 289.3281,
|
| 545 |
+
"eval_samples_per_second": 15.965,
|
| 546 |
+
"eval_steps_per_second": 0.2,
|
| 547 |
+
"eval_wer": 0.37138922240271205,
|
| 548 |
+
"step": 21672
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 37.0,
|
| 552 |
+
"learning_rate": 6.40793918918919e-05,
|
| 553 |
+
"loss": 0.13,
|
| 554 |
+
"step": 22274
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 37.0,
|
| 558 |
+
"eval_loss": 0.3545927107334137,
|
| 559 |
+
"eval_runtime": 291.5095,
|
| 560 |
+
"eval_samples_per_second": 15.845,
|
| 561 |
+
"eval_steps_per_second": 0.199,
|
| 562 |
+
"eval_wer": 0.3684935376792146,
|
| 563 |
+
"step": 22274
|
| 564 |
+
},
|
| 565 |
+
{
|
| 566 |
+
"epoch": 38.0,
|
| 567 |
+
"learning_rate": 6.306250000000001e-05,
|
| 568 |
+
"loss": 0.1266,
|
| 569 |
+
"step": 22876
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 38.0,
|
| 573 |
+
"eval_loss": 0.37049075961112976,
|
| 574 |
+
"eval_runtime": 328.1134,
|
| 575 |
+
"eval_samples_per_second": 14.077,
|
| 576 |
+
"eval_steps_per_second": 0.177,
|
| 577 |
+
"eval_wer": 0.367787273112508,
|
| 578 |
+
"step": 22876
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 39.0,
|
| 582 |
+
"learning_rate": 6.20456081081081e-05,
|
| 583 |
+
"loss": 0.1253,
|
| 584 |
+
"step": 23478
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 39.0,
|
| 588 |
+
"eval_loss": 0.3741365373134613,
|
| 589 |
+
"eval_runtime": 476.2923,
|
| 590 |
+
"eval_samples_per_second": 9.698,
|
| 591 |
+
"eval_steps_per_second": 0.122,
|
| 592 |
+
"eval_wer": 0.37011794618264,
|
| 593 |
+
"step": 23478
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 40.0,
|
| 597 |
+
"learning_rate": 6.103040540540541e-05,
|
| 598 |
+
"loss": 0.1221,
|
| 599 |
+
"step": 24080
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 40.0,
|
| 603 |
+
"eval_loss": 0.3850318491458893,
|
| 604 |
+
"eval_runtime": 331.7813,
|
| 605 |
+
"eval_samples_per_second": 13.922,
|
| 606 |
+
"eval_steps_per_second": 0.175,
|
| 607 |
+
"eval_wer": 0.37220142665442474,
|
| 608 |
+
"step": 24080
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 41.0,
|
| 612 |
+
"learning_rate": 6.001351351351352e-05,
|
| 613 |
+
"loss": 0.1205,
|
| 614 |
+
"step": 24682
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 41.0,
|
| 618 |
+
"eval_loss": 0.3865145146846771,
|
| 619 |
+
"eval_runtime": 328.7451,
|
| 620 |
+
"eval_samples_per_second": 14.05,
|
| 621 |
+
"eval_steps_per_second": 0.176,
|
| 622 |
+
"eval_wer": 0.3651034677590225,
|
| 623 |
+
"step": 24682
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 42.0,
|
| 627 |
+
"learning_rate": 5.899662162162163e-05,
|
| 628 |
+
"loss": 0.1179,
|
| 629 |
+
"step": 25284
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 42.0,
|
| 633 |
+
"eval_loss": 0.3818467855453491,
|
| 634 |
+
"eval_runtime": 325.7829,
|
| 635 |
+
"eval_samples_per_second": 14.178,
|
| 636 |
+
"eval_steps_per_second": 0.178,
|
| 637 |
+
"eval_wer": 0.36644537043576525,
|
| 638 |
+
"step": 25284
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 43.0,
|
| 642 |
+
"learning_rate": 5.7981418918918915e-05,
|
| 643 |
+
"loss": 0.1185,
|
| 644 |
+
"step": 25886
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 43.0,
|
| 648 |
+
"eval_loss": 0.3811143636703491,
|
| 649 |
+
"eval_runtime": 326.5779,
|
| 650 |
+
"eval_samples_per_second": 14.144,
|
| 651 |
+
"eval_steps_per_second": 0.178,
|
| 652 |
+
"eval_wer": 0.36644537043576525,
|
| 653 |
+
"step": 25886
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 44.0,
|
| 657 |
+
"learning_rate": 5.6964527027027026e-05,
|
| 658 |
+
"loss": 0.114,
|
| 659 |
+
"step": 26488
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 44.0,
|
| 663 |
+
"eval_loss": 0.3834444284439087,
|
| 664 |
+
"eval_runtime": 335.6248,
|
| 665 |
+
"eval_samples_per_second": 13.762,
|
| 666 |
+
"eval_steps_per_second": 0.173,
|
| 667 |
+
"eval_wer": 0.3639028179956212,
|
| 668 |
+
"step": 26488
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 45.0,
|
| 672 |
+
"learning_rate": 5.594763513513514e-05,
|
| 673 |
+
"loss": 0.1147,
|
| 674 |
+
"step": 27090
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 45.0,
|
| 678 |
+
"eval_loss": 0.41083571314811707,
|
| 679 |
+
"eval_runtime": 333.8024,
|
| 680 |
+
"eval_samples_per_second": 13.838,
|
| 681 |
+
"eval_steps_per_second": 0.174,
|
| 682 |
+
"eval_wer": 0.3716364150010594,
|
| 683 |
+
"step": 27090
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 46.0,
|
| 687 |
+
"learning_rate": 5.493074324324324e-05,
|
| 688 |
+
"loss": 0.1115,
|
| 689 |
+
"step": 27692
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 46.0,
|
| 693 |
+
"eval_loss": 0.39661291241645813,
|
| 694 |
+
"eval_runtime": 339.3131,
|
| 695 |
+
"eval_samples_per_second": 13.613,
|
| 696 |
+
"eval_steps_per_second": 0.171,
|
| 697 |
+
"eval_wer": 0.36227840949219575,
|
| 698 |
+
"step": 27692
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 47.0,
|
| 702 |
+
"learning_rate": 5.3915540540540545e-05,
|
| 703 |
+
"loss": 0.109,
|
| 704 |
+
"step": 28294
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 47.0,
|
| 708 |
+
"eval_loss": 0.38050925731658936,
|
| 709 |
+
"eval_runtime": 397.3812,
|
| 710 |
+
"eval_samples_per_second": 11.624,
|
| 711 |
+
"eval_steps_per_second": 0.146,
|
| 712 |
+
"eval_wer": 0.358817713115333,
|
| 713 |
+
"step": 28294
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 48.0,
|
| 717 |
+
"learning_rate": 5.289864864864865e-05,
|
| 718 |
+
"loss": 0.1072,
|
| 719 |
+
"step": 28896
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 48.0,
|
| 723 |
+
"eval_loss": 0.3901868164539337,
|
| 724 |
+
"eval_runtime": 304.2685,
|
| 725 |
+
"eval_samples_per_second": 15.181,
|
| 726 |
+
"eval_steps_per_second": 0.191,
|
| 727 |
+
"eval_wer": 0.35631047390352427,
|
| 728 |
+
"step": 28896
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 49.0,
|
| 732 |
+
"learning_rate": 5.188175675675676e-05,
|
| 733 |
+
"loss": 0.1066,
|
| 734 |
+
"step": 29498
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 49.0,
|
| 738 |
+
"eval_loss": 0.41520482301712036,
|
| 739 |
+
"eval_runtime": 344.9108,
|
| 740 |
+
"eval_samples_per_second": 13.392,
|
| 741 |
+
"eval_steps_per_second": 0.168,
|
| 742 |
+
"eval_wer": 0.36079525390211176,
|
| 743 |
+
"step": 29498
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 50.0,
|
| 747 |
+
"learning_rate": 5.086486486486487e-05,
|
| 748 |
+
"loss": 0.103,
|
| 749 |
+
"step": 30100
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 50.0,
|
| 753 |
+
"eval_loss": 0.3830413520336151,
|
| 754 |
+
"eval_runtime": 326.4997,
|
| 755 |
+
"eval_samples_per_second": 14.147,
|
| 756 |
+
"eval_steps_per_second": 0.178,
|
| 757 |
+
"eval_wer": 0.3593120983120277,
|
| 758 |
+
"step": 30100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 51.0,
|
| 762 |
+
"learning_rate": 4.9847972972972975e-05,
|
| 763 |
+
"loss": 0.1014,
|
| 764 |
+
"step": 30702
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 51.0,
|
| 768 |
+
"eval_loss": 0.40889114141464233,
|
| 769 |
+
"eval_runtime": 440.9092,
|
| 770 |
+
"eval_samples_per_second": 10.476,
|
| 771 |
+
"eval_steps_per_second": 0.132,
|
| 772 |
+
"eval_wer": 0.35578077547849424,
|
| 773 |
+
"step": 30702
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 52.0,
|
| 777 |
+
"learning_rate": 4.883108108108108e-05,
|
| 778 |
+
"loss": 0.1012,
|
| 779 |
+
"step": 31304
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 52.0,
|
| 783 |
+
"eval_loss": 0.3804776668548584,
|
| 784 |
+
"eval_runtime": 399.7783,
|
| 785 |
+
"eval_samples_per_second": 11.554,
|
| 786 |
+
"eval_steps_per_second": 0.145,
|
| 787 |
+
"eval_wer": 0.3551098241401229,
|
| 788 |
+
"step": 31304
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 53.0,
|
| 792 |
+
"learning_rate": 4.781418918918919e-05,
|
| 793 |
+
"loss": 0.0988,
|
| 794 |
+
"step": 31906
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 53.0,
|
| 798 |
+
"eval_loss": 0.38413405418395996,
|
| 799 |
+
"eval_runtime": 402.2767,
|
| 800 |
+
"eval_samples_per_second": 11.482,
|
| 801 |
+
"eval_steps_per_second": 0.144,
|
| 802 |
+
"eval_wer": 0.35320290981001484,
|
| 803 |
+
"step": 31906
|
| 804 |
+
},
|
| 805 |
+
{
|
| 806 |
+
"epoch": 54.0,
|
| 807 |
+
"learning_rate": 4.679898648648649e-05,
|
| 808 |
+
"loss": 0.0984,
|
| 809 |
+
"step": 32508
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 54.0,
|
| 813 |
+
"eval_loss": 0.4035479724407196,
|
| 814 |
+
"eval_runtime": 382.7739,
|
| 815 |
+
"eval_samples_per_second": 12.067,
|
| 816 |
+
"eval_steps_per_second": 0.152,
|
| 817 |
+
"eval_wer": 0.3548273183134402,
|
| 818 |
+
"step": 32508
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 55.0,
|
| 822 |
+
"learning_rate": 4.5783783783783785e-05,
|
| 823 |
+
"loss": 0.0961,
|
| 824 |
+
"step": 33110
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 55.0,
|
| 828 |
+
"eval_loss": 0.3776606619358063,
|
| 829 |
+
"eval_runtime": 439.3383,
|
| 830 |
+
"eval_samples_per_second": 10.514,
|
| 831 |
+
"eval_steps_per_second": 0.132,
|
| 832 |
+
"eval_wer": 0.3542269934317395,
|
| 833 |
+
"step": 33110
|
| 834 |
+
},
|
| 835 |
+
{
|
| 836 |
+
"epoch": 56.0,
|
| 837 |
+
"learning_rate": 4.4766891891891895e-05,
|
| 838 |
+
"loss": 0.0949,
|
| 839 |
+
"step": 33712
|
| 840 |
+
},
|
| 841 |
+
{
|
| 842 |
+
"epoch": 56.0,
|
| 843 |
+
"eval_loss": 0.41516759991645813,
|
| 844 |
+
"eval_runtime": 422.6534,
|
| 845 |
+
"eval_samples_per_second": 10.929,
|
| 846 |
+
"eval_steps_per_second": 0.137,
|
| 847 |
+
"eval_wer": 0.35461543894342823,
|
| 848 |
+
"step": 33712
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 57.0,
|
| 852 |
+
"learning_rate": 4.375e-05,
|
| 853 |
+
"loss": 0.0935,
|
| 854 |
+
"step": 34314
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 57.0,
|
| 858 |
+
"eval_loss": 0.3894650340080261,
|
| 859 |
+
"eval_runtime": 423.1466,
|
| 860 |
+
"eval_samples_per_second": 10.916,
|
| 861 |
+
"eval_steps_per_second": 0.137,
|
| 862 |
+
"eval_wer": 0.347729359418038,
|
| 863 |
+
"step": 34314
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 58.0,
|
| 867 |
+
"learning_rate": 4.273310810810811e-05,
|
| 868 |
+
"loss": 0.0925,
|
| 869 |
+
"step": 34916
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 58.0,
|
| 873 |
+
"eval_loss": 0.4071078896522522,
|
| 874 |
+
"eval_runtime": 435.5872,
|
| 875 |
+
"eval_samples_per_second": 10.604,
|
| 876 |
+
"eval_steps_per_second": 0.133,
|
| 877 |
+
"eval_wer": 0.35087223673988277,
|
| 878 |
+
"step": 34916
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 59.0,
|
| 882 |
+
"learning_rate": 4.171621621621622e-05,
|
| 883 |
+
"loss": 0.0922,
|
| 884 |
+
"step": 35518
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 59.0,
|
| 888 |
+
"eval_loss": 0.4226244390010834,
|
| 889 |
+
"eval_runtime": 403.3952,
|
| 890 |
+
"eval_samples_per_second": 11.45,
|
| 891 |
+
"eval_steps_per_second": 0.144,
|
| 892 |
+
"eval_wer": 0.35309697012500885,
|
| 893 |
+
"step": 35518
|
| 894 |
+
},
|
| 895 |
+
{
|
| 896 |
+
"epoch": 60.0,
|
| 897 |
+
"learning_rate": 4.070101351351351e-05,
|
| 898 |
+
"loss": 0.0895,
|
| 899 |
+
"step": 36120
|
| 900 |
+
},
|
| 901 |
+
{
|
| 902 |
+
"epoch": 60.0,
|
| 903 |
+
"eval_loss": 0.43153172731399536,
|
| 904 |
+
"eval_runtime": 398.9709,
|
| 905 |
+
"eval_samples_per_second": 11.577,
|
| 906 |
+
"eval_steps_per_second": 0.145,
|
| 907 |
+
"eval_wer": 0.35094286319655343,
|
| 908 |
+
"step": 36120
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 61.0,
|
| 912 |
+
"learning_rate": 3.968412162162162e-05,
|
| 913 |
+
"loss": 0.0897,
|
| 914 |
+
"step": 36722
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 61.0,
|
| 918 |
+
"eval_loss": 0.4538232386112213,
|
| 919 |
+
"eval_runtime": 383.2903,
|
| 920 |
+
"eval_samples_per_second": 12.051,
|
| 921 |
+
"eval_steps_per_second": 0.151,
|
| 922 |
+
"eval_wer": 0.3565929797302069,
|
| 923 |
+
"step": 36722
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 62.0,
|
| 927 |
+
"learning_rate": 3.8667229729729734e-05,
|
| 928 |
+
"loss": 0.0873,
|
| 929 |
+
"step": 37324
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 62.0,
|
| 933 |
+
"eval_loss": 0.421456903219223,
|
| 934 |
+
"eval_runtime": 398.9508,
|
| 935 |
+
"eval_samples_per_second": 11.578,
|
| 936 |
+
"eval_steps_per_second": 0.145,
|
| 937 |
+
"eval_wer": 0.34794123878805,
|
| 938 |
+
"step": 37324
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 63.0,
|
| 942 |
+
"learning_rate": 3.765202702702703e-05,
|
| 943 |
+
"loss": 0.0871,
|
| 944 |
+
"step": 37926
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 63.0,
|
| 948 |
+
"eval_loss": 0.42824164032936096,
|
| 949 |
+
"eval_runtime": 403.2616,
|
| 950 |
+
"eval_samples_per_second": 11.454,
|
| 951 |
+
"eval_steps_per_second": 0.144,
|
| 952 |
+
"eval_wer": 0.3506956705982061,
|
| 953 |
+
"step": 37926
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 64.0,
|
| 957 |
+
"learning_rate": 3.663682432432433e-05,
|
| 958 |
+
"loss": 0.0862,
|
| 959 |
+
"step": 38528
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 64.0,
|
| 963 |
+
"eval_loss": 0.4110530912876129,
|
| 964 |
+
"eval_runtime": 413.0672,
|
| 965 |
+
"eval_samples_per_second": 11.182,
|
| 966 |
+
"eval_steps_per_second": 0.14,
|
| 967 |
+
"eval_wer": 0.3477999858747087,
|
| 968 |
+
"step": 38528
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 65.0,
|
| 972 |
+
"learning_rate": 3.561993243243243e-05,
|
| 973 |
+
"loss": 0.0847,
|
| 974 |
+
"step": 39130
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 65.0,
|
| 978 |
+
"eval_loss": 0.4184618890285492,
|
| 979 |
+
"eval_runtime": 401.9246,
|
| 980 |
+
"eval_samples_per_second": 11.492,
|
| 981 |
+
"eval_steps_per_second": 0.144,
|
| 982 |
+
"eval_wer": 0.3505544176848647,
|
| 983 |
+
"step": 39130
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 66.0,
|
| 987 |
+
"learning_rate": 3.460304054054054e-05,
|
| 988 |
+
"loss": 0.0843,
|
| 989 |
+
"step": 39732
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 66.0,
|
| 993 |
+
"eval_loss": 0.41128143668174744,
|
| 994 |
+
"eval_runtime": 410.7955,
|
| 995 |
+
"eval_samples_per_second": 11.244,
|
| 996 |
+
"eval_steps_per_second": 0.141,
|
| 997 |
+
"eval_wer": 0.3463168302846246,
|
| 998 |
+
"step": 39732
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 67.0,
|
| 1002 |
+
"learning_rate": 3.3586148648648654e-05,
|
| 1003 |
+
"loss": 0.0829,
|
| 1004 |
+
"step": 40334
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 67.0,
|
| 1008 |
+
"eval_loss": 0.4144081771373749,
|
| 1009 |
+
"eval_runtime": 396.7416,
|
| 1010 |
+
"eval_samples_per_second": 11.642,
|
| 1011 |
+
"eval_steps_per_second": 0.146,
|
| 1012 |
+
"eval_wer": 0.3457871318595946,
|
| 1013 |
+
"step": 40334
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 68.0,
|
| 1017 |
+
"learning_rate": 3.256925675675676e-05,
|
| 1018 |
+
"loss": 0.0831,
|
| 1019 |
+
"step": 40936
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 68.0,
|
| 1023 |
+
"eval_loss": 0.4325215220451355,
|
| 1024 |
+
"eval_runtime": 401.3277,
|
| 1025 |
+
"eval_samples_per_second": 11.509,
|
| 1026 |
+
"eval_steps_per_second": 0.145,
|
| 1027 |
+
"eval_wer": 0.34493961437954657,
|
| 1028 |
+
"step": 40936
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 69.0,
|
| 1032 |
+
"learning_rate": 3.155236486486486e-05,
|
| 1033 |
+
"loss": 0.082,
|
| 1034 |
+
"step": 41538
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 69.0,
|
| 1038 |
+
"eval_loss": 0.4353320300579071,
|
| 1039 |
+
"eval_runtime": 266.5733,
|
| 1040 |
+
"eval_samples_per_second": 17.327,
|
| 1041 |
+
"eval_steps_per_second": 0.218,
|
| 1042 |
+
"eval_wer": 0.34246768839607317,
|
| 1043 |
+
"step": 41538
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 70.0,
|
| 1047 |
+
"learning_rate": 3.053547297297297e-05,
|
| 1048 |
+
"loss": 0.0802,
|
| 1049 |
+
"step": 42140
|
| 1050 |
+
},
|
| 1051 |
+
{
|
| 1052 |
+
"epoch": 70.0,
|
| 1053 |
+
"eval_loss": 0.39984482526779175,
|
| 1054 |
+
"eval_runtime": 366.598,
|
| 1055 |
+
"eval_samples_per_second": 12.6,
|
| 1056 |
+
"eval_steps_per_second": 0.158,
|
| 1057 |
+
"eval_wer": 0.34059608729430046,
|
| 1058 |
+
"step": 42140
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 71.0,
|
| 1062 |
+
"learning_rate": 2.951858108108108e-05,
|
| 1063 |
+
"loss": 0.0792,
|
| 1064 |
+
"step": 42742
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 71.0,
|
| 1068 |
+
"eval_loss": 0.4112567901611328,
|
| 1069 |
+
"eval_runtime": 290.4304,
|
| 1070 |
+
"eval_samples_per_second": 15.904,
|
| 1071 |
+
"eval_steps_per_second": 0.2,
|
| 1072 |
+
"eval_wer": 0.3403488946959531,
|
| 1073 |
+
"step": 42742
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 72.0,
|
| 1077 |
+
"learning_rate": 2.850337837837838e-05,
|
| 1078 |
+
"loss": 0.0761,
|
| 1079 |
+
"step": 43344
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 72.0,
|
| 1083 |
+
"eval_loss": 0.422593891620636,
|
| 1084 |
+
"eval_runtime": 332.1718,
|
| 1085 |
+
"eval_samples_per_second": 13.905,
|
| 1086 |
+
"eval_steps_per_second": 0.175,
|
| 1087 |
+
"eval_wer": 0.3413376650893425,
|
| 1088 |
+
"step": 43344
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 73.0,
|
| 1092 |
+
"learning_rate": 2.7488175675675675e-05,
|
| 1093 |
+
"loss": 0.078,
|
| 1094 |
+
"step": 43946
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 73.0,
|
| 1098 |
+
"eval_loss": 0.420549213886261,
|
| 1099 |
+
"eval_runtime": 348.4845,
|
| 1100 |
+
"eval_samples_per_second": 13.255,
|
| 1101 |
+
"eval_steps_per_second": 0.166,
|
| 1102 |
+
"eval_wer": 0.3414436047743485,
|
| 1103 |
+
"step": 43946
|
| 1104 |
+
},
|
| 1105 |
+
{
|
| 1106 |
+
"epoch": 74.0,
|
| 1107 |
+
"learning_rate": 2.6471283783783786e-05,
|
| 1108 |
+
"loss": 0.0771,
|
| 1109 |
+
"step": 44548
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 74.0,
|
| 1113 |
+
"eval_loss": 0.4529767334461212,
|
| 1114 |
+
"eval_runtime": 337.122,
|
| 1115 |
+
"eval_samples_per_second": 13.701,
|
| 1116 |
+
"eval_steps_per_second": 0.172,
|
| 1117 |
+
"eval_wer": 0.34349177201779785,
|
| 1118 |
+
"step": 44548
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 75.0,
|
| 1122 |
+
"learning_rate": 2.5454391891891894e-05,
|
| 1123 |
+
"loss": 0.0753,
|
| 1124 |
+
"step": 45150
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 75.0,
|
| 1128 |
+
"eval_loss": 0.42464756965637207,
|
| 1129 |
+
"eval_runtime": 276.2812,
|
| 1130 |
+
"eval_samples_per_second": 16.718,
|
| 1131 |
+
"eval_steps_per_second": 0.21,
|
| 1132 |
+
"eval_wer": 0.33936012430256374,
|
| 1133 |
+
"step": 45150
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 76.0,
|
| 1137 |
+
"learning_rate": 2.44375e-05,
|
| 1138 |
+
"loss": 0.0747,
|
| 1139 |
+
"step": 45752
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 76.0,
|
| 1143 |
+
"eval_loss": 0.4271426200866699,
|
| 1144 |
+
"eval_runtime": 305.7967,
|
| 1145 |
+
"eval_samples_per_second": 15.105,
|
| 1146 |
+
"eval_steps_per_second": 0.19,
|
| 1147 |
+
"eval_wer": 0.3410904724909951,
|
| 1148 |
+
"step": 45752
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 77.0,
|
| 1152 |
+
"learning_rate": 2.34222972972973e-05,
|
| 1153 |
+
"loss": 0.0744,
|
| 1154 |
+
"step": 46354
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 77.0,
|
| 1158 |
+
"eval_loss": 0.4557636082172394,
|
| 1159 |
+
"eval_runtime": 291.9496,
|
| 1160 |
+
"eval_samples_per_second": 15.821,
|
| 1161 |
+
"eval_steps_per_second": 0.199,
|
| 1162 |
+
"eval_wer": 0.34193798997104313,
|
| 1163 |
+
"step": 46354
|
| 1164 |
+
},
|
| 1165 |
+
{
|
| 1166 |
+
"epoch": 78.0,
|
| 1167 |
+
"learning_rate": 2.2405405405405406e-05,
|
| 1168 |
+
"loss": 0.0735,
|
| 1169 |
+
"step": 46956
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 78.0,
|
| 1173 |
+
"eval_loss": 0.44490668177604675,
|
| 1174 |
+
"eval_runtime": 315.5036,
|
| 1175 |
+
"eval_samples_per_second": 14.64,
|
| 1176 |
+
"eval_steps_per_second": 0.184,
|
| 1177 |
+
"eval_wer": 0.34151423123101915,
|
| 1178 |
+
"step": 46956
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 79.0,
|
| 1182 |
+
"learning_rate": 2.1388513513513513e-05,
|
| 1183 |
+
"loss": 0.073,
|
| 1184 |
+
"step": 47558
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 79.0,
|
| 1188 |
+
"eval_loss": 0.431576669216156,
|
| 1189 |
+
"eval_runtime": 341.3029,
|
| 1190 |
+
"eval_samples_per_second": 13.533,
|
| 1191 |
+
"eval_steps_per_second": 0.17,
|
| 1192 |
+
"eval_wer": 0.33844198036584505,
|
| 1193 |
+
"step": 47558
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 80.0,
|
| 1197 |
+
"learning_rate": 2.0371621621621624e-05,
|
| 1198 |
+
"loss": 0.0735,
|
| 1199 |
+
"step": 48160
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 80.0,
|
| 1203 |
+
"eval_loss": 0.4496508538722992,
|
| 1204 |
+
"eval_runtime": 314.7178,
|
| 1205 |
+
"eval_samples_per_second": 14.677,
|
| 1206 |
+
"eval_steps_per_second": 0.184,
|
| 1207 |
+
"eval_wer": 0.33833604068083906,
|
| 1208 |
+
"step": 48160
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 81.0,
|
| 1212 |
+
"learning_rate": 1.935641891891892e-05,
|
| 1213 |
+
"loss": 0.0721,
|
| 1214 |
+
"step": 48762
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 81.0,
|
| 1218 |
+
"eval_loss": 0.43915238976478577,
|
| 1219 |
+
"eval_runtime": 285.1559,
|
| 1220 |
+
"eval_samples_per_second": 16.198,
|
| 1221 |
+
"eval_steps_per_second": 0.203,
|
| 1222 |
+
"eval_wer": 0.33667631894907835,
|
| 1223 |
+
"step": 48762
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 82.0,
|
| 1227 |
+
"learning_rate": 1.8341216216216216e-05,
|
| 1228 |
+
"loss": 0.0705,
|
| 1229 |
+
"step": 49364
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 82.0,
|
| 1233 |
+
"eval_loss": 0.42993664741516113,
|
| 1234 |
+
"eval_runtime": 312.9699,
|
| 1235 |
+
"eval_samples_per_second": 14.759,
|
| 1236 |
+
"eval_steps_per_second": 0.185,
|
| 1237 |
+
"eval_wer": 0.33678225863408434,
|
| 1238 |
+
"step": 49364
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 83.0,
|
| 1242 |
+
"learning_rate": 1.7324324324324326e-05,
|
| 1243 |
+
"loss": 0.0706,
|
| 1244 |
+
"step": 49966
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 83.0,
|
| 1248 |
+
"eval_loss": 0.442228227853775,
|
| 1249 |
+
"eval_runtime": 282.6844,
|
| 1250 |
+
"eval_samples_per_second": 16.34,
|
| 1251 |
+
"eval_steps_per_second": 0.205,
|
| 1252 |
+
"eval_wer": 0.33738258351578504,
|
| 1253 |
+
"step": 49966
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 84.0,
|
| 1257 |
+
"learning_rate": 1.6307432432432434e-05,
|
| 1258 |
+
"loss": 0.0707,
|
| 1259 |
+
"step": 50568
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 84.0,
|
| 1263 |
+
"eval_loss": 0.43407562375068665,
|
| 1264 |
+
"eval_runtime": 348.7351,
|
| 1265 |
+
"eval_samples_per_second": 13.245,
|
| 1266 |
+
"eval_steps_per_second": 0.166,
|
| 1267 |
+
"eval_wer": 0.33586411469736566,
|
| 1268 |
+
"step": 50568
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 85.0,
|
| 1272 |
+
"learning_rate": 1.529054054054054e-05,
|
| 1273 |
+
"loss": 0.0692,
|
| 1274 |
+
"step": 51170
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 85.0,
|
| 1278 |
+
"eval_loss": 0.44134435057640076,
|
| 1279 |
+
"eval_runtime": 326.8142,
|
| 1280 |
+
"eval_samples_per_second": 14.133,
|
| 1281 |
+
"eval_steps_per_second": 0.177,
|
| 1282 |
+
"eval_wer": 0.33568754855568894,
|
| 1283 |
+
"step": 51170
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 86.0,
|
| 1287 |
+
"learning_rate": 1.4273648648648649e-05,
|
| 1288 |
+
"loss": 0.069,
|
| 1289 |
+
"step": 51772
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 86.0,
|
| 1293 |
+
"eval_loss": 0.4566088318824768,
|
| 1294 |
+
"eval_runtime": 315.2023,
|
| 1295 |
+
"eval_samples_per_second": 14.654,
|
| 1296 |
+
"eval_steps_per_second": 0.184,
|
| 1297 |
+
"eval_wer": 0.337912281940815,
|
| 1298 |
+
"step": 51772
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 87.0,
|
| 1302 |
+
"learning_rate": 1.3256756756756758e-05,
|
| 1303 |
+
"loss": 0.0673,
|
| 1304 |
+
"step": 52374
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 87.0,
|
| 1308 |
+
"eval_loss": 0.4371909201145172,
|
| 1309 |
+
"eval_runtime": 288.747,
|
| 1310 |
+
"eval_samples_per_second": 15.997,
|
| 1311 |
+
"eval_steps_per_second": 0.201,
|
| 1312 |
+
"eval_wer": 0.33957200367257573,
|
| 1313 |
+
"step": 52374
|
| 1314 |
+
},
|
| 1315 |
+
{
|
| 1316 |
+
"epoch": 88.0,
|
| 1317 |
+
"learning_rate": 1.2239864864864865e-05,
|
| 1318 |
+
"loss": 0.0678,
|
| 1319 |
+
"step": 52976
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 88.0,
|
| 1323 |
+
"eval_loss": 0.438821017742157,
|
| 1324 |
+
"eval_runtime": 399.7437,
|
| 1325 |
+
"eval_samples_per_second": 11.555,
|
| 1326 |
+
"eval_steps_per_second": 0.145,
|
| 1327 |
+
"eval_wer": 0.33819478776749773,
|
| 1328 |
+
"step": 52976
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 89.0,
|
| 1332 |
+
"learning_rate": 1.1222972972972975e-05,
|
| 1333 |
+
"loss": 0.0679,
|
| 1334 |
+
"step": 53578
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 89.0,
|
| 1338 |
+
"eval_loss": 0.4385330379009247,
|
| 1339 |
+
"eval_runtime": 425.8088,
|
| 1340 |
+
"eval_samples_per_second": 10.848,
|
| 1341 |
+
"eval_steps_per_second": 0.136,
|
| 1342 |
+
"eval_wer": 0.33628787343738964,
|
| 1343 |
+
"step": 53578
|
| 1344 |
+
},
|
| 1345 |
+
{
|
| 1346 |
+
"epoch": 90.0,
|
| 1347 |
+
"learning_rate": 1.0207770270270272e-05,
|
| 1348 |
+
"loss": 0.0676,
|
| 1349 |
+
"step": 54180
|
| 1350 |
+
},
|
| 1351 |
+
{
|
| 1352 |
+
"epoch": 90.0,
|
| 1353 |
+
"eval_loss": 0.44581928849220276,
|
| 1354 |
+
"eval_runtime": 430.0001,
|
| 1355 |
+
"eval_samples_per_second": 10.742,
|
| 1356 |
+
"eval_steps_per_second": 0.135,
|
| 1357 |
+
"eval_wer": 0.3353344162723356,
|
| 1358 |
+
"step": 54180
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 91.0,
|
| 1362 |
+
"learning_rate": 9.19087837837838e-06,
|
| 1363 |
+
"loss": 0.0661,
|
| 1364 |
+
"step": 54782
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 91.0,
|
| 1368 |
+
"eval_loss": 0.44878506660461426,
|
| 1369 |
+
"eval_runtime": 414.3814,
|
| 1370 |
+
"eval_samples_per_second": 11.147,
|
| 1371 |
+
"eval_steps_per_second": 0.14,
|
| 1372 |
+
"eval_wer": 0.3352284765873296,
|
| 1373 |
+
"step": 54782
|
| 1374 |
+
},
|
| 1375 |
+
{
|
| 1376 |
+
"epoch": 92.0,
|
| 1377 |
+
"learning_rate": 8.173986486486487e-06,
|
| 1378 |
+
"loss": 0.0667,
|
| 1379 |
+
"step": 55384
|
| 1380 |
+
},
|
| 1381 |
+
{
|
| 1382 |
+
"epoch": 92.0,
|
| 1383 |
+
"eval_loss": 0.4502880275249481,
|
| 1384 |
+
"eval_runtime": 640.6777,
|
| 1385 |
+
"eval_samples_per_second": 7.21,
|
| 1386 |
+
"eval_steps_per_second": 0.091,
|
| 1387 |
+
"eval_wer": 0.33649975280740163,
|
| 1388 |
+
"step": 55384
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 93.0,
|
| 1392 |
+
"learning_rate": 7.157094594594594e-06,
|
| 1393 |
+
"loss": 0.0668,
|
| 1394 |
+
"step": 55986
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 93.0,
|
| 1398 |
+
"eval_loss": 0.4470549523830414,
|
| 1399 |
+
"eval_runtime": 464.6555,
|
| 1400 |
+
"eval_samples_per_second": 9.941,
|
| 1401 |
+
"eval_steps_per_second": 0.125,
|
| 1402 |
+
"eval_wer": 0.33371000776891024,
|
| 1403 |
+
"step": 55986
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 94.0,
|
| 1407 |
+
"learning_rate": 6.140202702702703e-06,
|
| 1408 |
+
"loss": 0.0663,
|
| 1409 |
+
"step": 56588
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 94.0,
|
| 1413 |
+
"eval_loss": 0.4470076858997345,
|
| 1414 |
+
"eval_runtime": 411.3931,
|
| 1415 |
+
"eval_samples_per_second": 11.228,
|
| 1416 |
+
"eval_steps_per_second": 0.141,
|
| 1417 |
+
"eval_wer": 0.3332862490288862,
|
| 1418 |
+
"step": 56588
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 95.0,
|
| 1422 |
+
"learning_rate": 5.123310810810811e-06,
|
| 1423 |
+
"loss": 0.0652,
|
| 1424 |
+
"step": 57190
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 95.0,
|
| 1428 |
+
"eval_loss": 0.4530947804450989,
|
| 1429 |
+
"eval_runtime": 429.7206,
|
| 1430 |
+
"eval_samples_per_second": 10.749,
|
| 1431 |
+
"eval_steps_per_second": 0.135,
|
| 1432 |
+
"eval_wer": 0.33476940461897026,
|
| 1433 |
+
"step": 57190
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 96.0,
|
| 1437 |
+
"learning_rate": 4.108108108108108e-06,
|
| 1438 |
+
"loss": 0.0648,
|
| 1439 |
+
"step": 57792
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 96.0,
|
| 1443 |
+
"eval_loss": 0.44802409410476685,
|
| 1444 |
+
"eval_runtime": 456.6508,
|
| 1445 |
+
"eval_samples_per_second": 10.115,
|
| 1446 |
+
"eval_steps_per_second": 0.127,
|
| 1447 |
+
"eval_wer": 0.33346281517056287,
|
| 1448 |
+
"step": 57792
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 97.0,
|
| 1452 |
+
"learning_rate": 3.0912162162162163e-06,
|
| 1453 |
+
"loss": 0.0638,
|
| 1454 |
+
"step": 58394
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 97.0,
|
| 1458 |
+
"eval_loss": 0.4479643702507019,
|
| 1459 |
+
"eval_runtime": 442.7361,
|
| 1460 |
+
"eval_samples_per_second": 10.433,
|
| 1461 |
+
"eval_steps_per_second": 0.131,
|
| 1462 |
+
"eval_wer": 0.33346281517056287,
|
| 1463 |
+
"step": 58394
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 98.0,
|
| 1467 |
+
"learning_rate": 2.0743243243243246e-06,
|
| 1468 |
+
"loss": 0.0637,
|
| 1469 |
+
"step": 58996
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 98.0,
|
| 1473 |
+
"eval_loss": 0.4514257311820984,
|
| 1474 |
+
"eval_runtime": 447.5804,
|
| 1475 |
+
"eval_samples_per_second": 10.32,
|
| 1476 |
+
"eval_steps_per_second": 0.13,
|
| 1477 |
+
"eval_wer": 0.3328624902888622,
|
| 1478 |
+
"step": 58996
|
| 1479 |
+
}
|
| 1480 |
+
],
|
| 1481 |
+
"max_steps": 60200,
|
| 1482 |
+
"num_train_epochs": 100,
|
| 1483 |
+
"total_flos": 1.2978795625125155e+21,
|
| 1484 |
+
"trial_name": null,
|
| 1485 |
+
"trial_params": null
|
| 1486 |
+
}
|