Commit
•
e4d435e
1
Parent(s):
54dd7f3
Upload trainer_state.json with huggingface_hub
Browse files- trainer_state.json +2974 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,2974 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.04516952857375145,
|
| 3 |
+
"best_model_checkpoint": "./models/results_short_jokes_47/checkpoint-3400",
|
| 4 |
+
"epoch": 0.6262086748319367,
|
| 5 |
+
"global_step": 3400,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 8.000000000000001e-07,
|
| 13 |
+
"loss": 0.691,
|
| 14 |
+
"step": 8
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.0,
|
| 18 |
+
"learning_rate": 1.6000000000000001e-06,
|
| 19 |
+
"loss": 0.6954,
|
| 20 |
+
"step": 16
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.0,
|
| 24 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 25 |
+
"loss": 0.6964,
|
| 26 |
+
"step": 24
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.01,
|
| 30 |
+
"learning_rate": 3.2000000000000003e-06,
|
| 31 |
+
"loss": 0.6855,
|
| 32 |
+
"step": 32
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.01,
|
| 36 |
+
"learning_rate": 4.000000000000001e-06,
|
| 37 |
+
"loss": 0.6916,
|
| 38 |
+
"step": 40
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.01,
|
| 42 |
+
"learning_rate": 4.800000000000001e-06,
|
| 43 |
+
"loss": 0.6845,
|
| 44 |
+
"step": 48
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.01,
|
| 48 |
+
"learning_rate": 5.600000000000001e-06,
|
| 49 |
+
"loss": 0.6708,
|
| 50 |
+
"step": 56
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.01,
|
| 54 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 55 |
+
"loss": 0.6411,
|
| 56 |
+
"step": 64
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.01,
|
| 60 |
+
"learning_rate": 7.2e-06,
|
| 61 |
+
"loss": 0.5579,
|
| 62 |
+
"step": 72
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.01,
|
| 66 |
+
"learning_rate": 7.9e-06,
|
| 67 |
+
"loss": 0.3916,
|
| 68 |
+
"step": 80
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.02,
|
| 72 |
+
"learning_rate": 8.7e-06,
|
| 73 |
+
"loss": 0.262,
|
| 74 |
+
"step": 88
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.02,
|
| 78 |
+
"learning_rate": 9.5e-06,
|
| 79 |
+
"loss": 0.2353,
|
| 80 |
+
"step": 96
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.02,
|
| 84 |
+
"eval_accuracy": 0.9376661947024899,
|
| 85 |
+
"eval_f1": 0.9405800441123219,
|
| 86 |
+
"eval_loss": 0.1874532550573349,
|
| 87 |
+
"eval_precision": 0.9014102280972963,
|
| 88 |
+
"eval_recall": 0.9833086691674984,
|
| 89 |
+
"eval_runtime": 312.5596,
|
| 90 |
+
"eval_samples_per_second": 185.289,
|
| 91 |
+
"eval_steps_per_second": 23.164,
|
| 92 |
+
"step": 100
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.02,
|
| 96 |
+
"learning_rate": 1.03e-05,
|
| 97 |
+
"loss": 0.193,
|
| 98 |
+
"step": 104
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.02,
|
| 102 |
+
"learning_rate": 1.11e-05,
|
| 103 |
+
"loss": 0.1671,
|
| 104 |
+
"step": 112
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.02,
|
| 108 |
+
"learning_rate": 1.19e-05,
|
| 109 |
+
"loss": 0.1242,
|
| 110 |
+
"step": 120
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.02,
|
| 114 |
+
"learning_rate": 1.2600000000000001e-05,
|
| 115 |
+
"loss": 0.1497,
|
| 116 |
+
"step": 128
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.03,
|
| 120 |
+
"learning_rate": 1.3400000000000002e-05,
|
| 121 |
+
"loss": 0.1173,
|
| 122 |
+
"step": 136
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.03,
|
| 126 |
+
"learning_rate": 1.42e-05,
|
| 127 |
+
"loss": 0.1598,
|
| 128 |
+
"step": 144
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.03,
|
| 132 |
+
"learning_rate": 1.5e-05,
|
| 133 |
+
"loss": 0.1312,
|
| 134 |
+
"step": 152
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.03,
|
| 138 |
+
"learning_rate": 1.58e-05,
|
| 139 |
+
"loss": 0.1374,
|
| 140 |
+
"step": 160
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.03,
|
| 144 |
+
"learning_rate": 1.66e-05,
|
| 145 |
+
"loss": 0.1055,
|
| 146 |
+
"step": 168
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.03,
|
| 150 |
+
"learning_rate": 1.74e-05,
|
| 151 |
+
"loss": 0.1114,
|
| 152 |
+
"step": 176
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.03,
|
| 156 |
+
"learning_rate": 1.8200000000000002e-05,
|
| 157 |
+
"loss": 0.1404,
|
| 158 |
+
"step": 184
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.04,
|
| 162 |
+
"learning_rate": 1.9e-05,
|
| 163 |
+
"loss": 0.1289,
|
| 164 |
+
"step": 192
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.04,
|
| 168 |
+
"learning_rate": 1.9800000000000004e-05,
|
| 169 |
+
"loss": 0.136,
|
| 170 |
+
"step": 200
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.04,
|
| 174 |
+
"eval_accuracy": 0.962185309251649,
|
| 175 |
+
"eval_f1": 0.9613947257086447,
|
| 176 |
+
"eval_loss": 0.11203108727931976,
|
| 177 |
+
"eval_precision": 0.9854721549636803,
|
| 178 |
+
"eval_recall": 0.9384657741680146,
|
| 179 |
+
"eval_runtime": 315.0228,
|
| 180 |
+
"eval_samples_per_second": 183.841,
|
| 181 |
+
"eval_steps_per_second": 22.982,
|
| 182 |
+
"step": 200
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.04,
|
| 186 |
+
"learning_rate": 2.06e-05,
|
| 187 |
+
"loss": 0.1112,
|
| 188 |
+
"step": 208
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.04,
|
| 192 |
+
"learning_rate": 2.1400000000000002e-05,
|
| 193 |
+
"loss": 0.1447,
|
| 194 |
+
"step": 216
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.04,
|
| 198 |
+
"learning_rate": 2.22e-05,
|
| 199 |
+
"loss": 0.143,
|
| 200 |
+
"step": 224
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.04,
|
| 204 |
+
"learning_rate": 2.3000000000000003e-05,
|
| 205 |
+
"loss": 0.1175,
|
| 206 |
+
"step": 232
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.04,
|
| 210 |
+
"learning_rate": 2.38e-05,
|
| 211 |
+
"loss": 0.1208,
|
| 212 |
+
"step": 240
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.05,
|
| 216 |
+
"learning_rate": 2.46e-05,
|
| 217 |
+
"loss": 0.1563,
|
| 218 |
+
"step": 248
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.05,
|
| 222 |
+
"learning_rate": 2.54e-05,
|
| 223 |
+
"loss": 0.0701,
|
| 224 |
+
"step": 256
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.05,
|
| 228 |
+
"learning_rate": 2.6200000000000003e-05,
|
| 229 |
+
"loss": 0.0881,
|
| 230 |
+
"step": 264
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.05,
|
| 234 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 235 |
+
"loss": 0.1091,
|
| 236 |
+
"step": 272
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.05,
|
| 240 |
+
"learning_rate": 2.7800000000000005e-05,
|
| 241 |
+
"loss": 0.1512,
|
| 242 |
+
"step": 280
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.05,
|
| 246 |
+
"learning_rate": 2.86e-05,
|
| 247 |
+
"loss": 0.0802,
|
| 248 |
+
"step": 288
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.05,
|
| 252 |
+
"learning_rate": 2.94e-05,
|
| 253 |
+
"loss": 0.103,
|
| 254 |
+
"step": 296
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.06,
|
| 258 |
+
"eval_accuracy": 0.9501674897261456,
|
| 259 |
+
"eval_f1": 0.9481028591979861,
|
| 260 |
+
"eval_loss": 0.13948623836040497,
|
| 261 |
+
"eval_precision": 0.992806839151885,
|
| 262 |
+
"eval_recall": 0.9072512647554806,
|
| 263 |
+
"eval_runtime": 304.8107,
|
| 264 |
+
"eval_samples_per_second": 190.0,
|
| 265 |
+
"eval_steps_per_second": 23.752,
|
| 266 |
+
"step": 300
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.06,
|
| 270 |
+
"learning_rate": 3.02e-05,
|
| 271 |
+
"loss": 0.0864,
|
| 272 |
+
"step": 304
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.06,
|
| 276 |
+
"learning_rate": 3.1e-05,
|
| 277 |
+
"loss": 0.0991,
|
| 278 |
+
"step": 312
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.06,
|
| 282 |
+
"learning_rate": 3.18e-05,
|
| 283 |
+
"loss": 0.0577,
|
| 284 |
+
"step": 320
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.06,
|
| 288 |
+
"learning_rate": 3.26e-05,
|
| 289 |
+
"loss": 0.1446,
|
| 290 |
+
"step": 328
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.06,
|
| 294 |
+
"learning_rate": 3.3400000000000005e-05,
|
| 295 |
+
"loss": 0.1009,
|
| 296 |
+
"step": 336
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.06,
|
| 300 |
+
"learning_rate": 3.4200000000000005e-05,
|
| 301 |
+
"loss": 0.1059,
|
| 302 |
+
"step": 344
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.06,
|
| 306 |
+
"learning_rate": 3.5e-05,
|
| 307 |
+
"loss": 0.1156,
|
| 308 |
+
"step": 352
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.07,
|
| 312 |
+
"learning_rate": 3.58e-05,
|
| 313 |
+
"loss": 0.0714,
|
| 314 |
+
"step": 360
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.07,
|
| 318 |
+
"learning_rate": 3.66e-05,
|
| 319 |
+
"loss": 0.0862,
|
| 320 |
+
"step": 368
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.07,
|
| 324 |
+
"learning_rate": 3.74e-05,
|
| 325 |
+
"loss": 0.1634,
|
| 326 |
+
"step": 376
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.07,
|
| 330 |
+
"learning_rate": 3.82e-05,
|
| 331 |
+
"loss": 0.0953,
|
| 332 |
+
"step": 384
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.07,
|
| 336 |
+
"learning_rate": 3.9000000000000006e-05,
|
| 337 |
+
"loss": 0.0822,
|
| 338 |
+
"step": 392
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.07,
|
| 342 |
+
"learning_rate": 3.9800000000000005e-05,
|
| 343 |
+
"loss": 0.0768,
|
| 344 |
+
"step": 400
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.07,
|
| 348 |
+
"eval_accuracy": 0.973028973995925,
|
| 349 |
+
"eval_f1": 0.9733447098976109,
|
| 350 |
+
"eval_loss": 0.08835102617740631,
|
| 351 |
+
"eval_precision": 0.9653386589039705,
|
| 352 |
+
"eval_recall": 0.9814846680662147,
|
| 353 |
+
"eval_runtime": 320.6331,
|
| 354 |
+
"eval_samples_per_second": 180.624,
|
| 355 |
+
"eval_steps_per_second": 22.58,
|
| 356 |
+
"step": 400
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.08,
|
| 360 |
+
"learning_rate": 4.0600000000000004e-05,
|
| 361 |
+
"loss": 0.1154,
|
| 362 |
+
"step": 408
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.08,
|
| 366 |
+
"learning_rate": 4.14e-05,
|
| 367 |
+
"loss": 0.1243,
|
| 368 |
+
"step": 416
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.08,
|
| 372 |
+
"learning_rate": 4.22e-05,
|
| 373 |
+
"loss": 0.0727,
|
| 374 |
+
"step": 424
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.08,
|
| 378 |
+
"learning_rate": 4.3e-05,
|
| 379 |
+
"loss": 0.0955,
|
| 380 |
+
"step": 432
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.08,
|
| 384 |
+
"learning_rate": 4.38e-05,
|
| 385 |
+
"loss": 0.0822,
|
| 386 |
+
"step": 440
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.08,
|
| 390 |
+
"learning_rate": 4.46e-05,
|
| 391 |
+
"loss": 0.1043,
|
| 392 |
+
"step": 448
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.08,
|
| 396 |
+
"learning_rate": 4.5400000000000006e-05,
|
| 397 |
+
"loss": 0.1473,
|
| 398 |
+
"step": 456
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.09,
|
| 402 |
+
"learning_rate": 4.6200000000000005e-05,
|
| 403 |
+
"loss": 0.0848,
|
| 404 |
+
"step": 464
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.09,
|
| 408 |
+
"learning_rate": 4.7e-05,
|
| 409 |
+
"loss": 0.139,
|
| 410 |
+
"step": 472
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.09,
|
| 414 |
+
"learning_rate": 4.78e-05,
|
| 415 |
+
"loss": 0.103,
|
| 416 |
+
"step": 480
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.09,
|
| 420 |
+
"learning_rate": 4.86e-05,
|
| 421 |
+
"loss": 0.1075,
|
| 422 |
+
"step": 488
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.09,
|
| 426 |
+
"learning_rate": 4.94e-05,
|
| 427 |
+
"loss": 0.1285,
|
| 428 |
+
"step": 496
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.09,
|
| 432 |
+
"eval_accuracy": 0.9650861622405636,
|
| 433 |
+
"eval_f1": 0.9643335920412051,
|
| 434 |
+
"eval_loss": 0.10335744917392731,
|
| 435 |
+
"eval_precision": 0.9891442011941378,
|
| 436 |
+
"eval_recall": 0.9407371717658396,
|
| 437 |
+
"eval_runtime": 317.7591,
|
| 438 |
+
"eval_samples_per_second": 182.258,
|
| 439 |
+
"eval_steps_per_second": 22.785,
|
| 440 |
+
"step": 500
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.09,
|
| 444 |
+
"learning_rate": 4.997971190910936e-05,
|
| 445 |
+
"loss": 0.085,
|
| 446 |
+
"step": 504
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.09,
|
| 450 |
+
"learning_rate": 4.989855954554677e-05,
|
| 451 |
+
"loss": 0.1381,
|
| 452 |
+
"step": 512
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.1,
|
| 456 |
+
"learning_rate": 4.9817407181984176e-05,
|
| 457 |
+
"loss": 0.1202,
|
| 458 |
+
"step": 520
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.1,
|
| 462 |
+
"learning_rate": 4.973625481842159e-05,
|
| 463 |
+
"loss": 0.0778,
|
| 464 |
+
"step": 528
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.1,
|
| 468 |
+
"learning_rate": 4.9655102454859e-05,
|
| 469 |
+
"loss": 0.0872,
|
| 470 |
+
"step": 536
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.1,
|
| 474 |
+
"learning_rate": 4.957395009129641e-05,
|
| 475 |
+
"loss": 0.1364,
|
| 476 |
+
"step": 544
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.1,
|
| 480 |
+
"learning_rate": 4.9492797727733825e-05,
|
| 481 |
+
"loss": 0.1305,
|
| 482 |
+
"step": 552
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.1,
|
| 486 |
+
"learning_rate": 4.9411645364171234e-05,
|
| 487 |
+
"loss": 0.0677,
|
| 488 |
+
"step": 560
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.1,
|
| 492 |
+
"learning_rate": 4.933049300060865e-05,
|
| 493 |
+
"loss": 0.1187,
|
| 494 |
+
"step": 568
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.11,
|
| 498 |
+
"learning_rate": 4.924934063704605e-05,
|
| 499 |
+
"loss": 0.1605,
|
| 500 |
+
"step": 576
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.11,
|
| 504 |
+
"learning_rate": 4.916818827348347e-05,
|
| 505 |
+
"loss": 0.1732,
|
| 506 |
+
"step": 584
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.11,
|
| 510 |
+
"learning_rate": 4.908703590992088e-05,
|
| 511 |
+
"loss": 0.104,
|
| 512 |
+
"step": 592
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.11,
|
| 516 |
+
"learning_rate": 4.900588354635829e-05,
|
| 517 |
+
"loss": 0.0872,
|
| 518 |
+
"step": 600
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.11,
|
| 522 |
+
"eval_accuracy": 0.973305245709155,
|
| 523 |
+
"eval_f1": 0.9731382701462973,
|
| 524 |
+
"eval_loss": 0.081203892827034,
|
| 525 |
+
"eval_precision": 0.982699933326315,
|
| 526 |
+
"eval_recall": 0.9637608837801562,
|
| 527 |
+
"eval_runtime": 323.1202,
|
| 528 |
+
"eval_samples_per_second": 179.234,
|
| 529 |
+
"eval_steps_per_second": 22.407,
|
| 530 |
+
"step": 600
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.11,
|
| 534 |
+
"learning_rate": 4.89247311827957e-05,
|
| 535 |
+
"loss": 0.0921,
|
| 536 |
+
"step": 608
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.11,
|
| 540 |
+
"learning_rate": 4.884357881923311e-05,
|
| 541 |
+
"loss": 0.069,
|
| 542 |
+
"step": 616
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.11,
|
| 546 |
+
"learning_rate": 4.8762426455670526e-05,
|
| 547 |
+
"loss": 0.0517,
|
| 548 |
+
"step": 624
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.12,
|
| 552 |
+
"learning_rate": 4.8681274092107935e-05,
|
| 553 |
+
"loss": 0.1178,
|
| 554 |
+
"step": 632
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.12,
|
| 558 |
+
"learning_rate": 4.8600121728545343e-05,
|
| 559 |
+
"loss": 0.154,
|
| 560 |
+
"step": 640
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.12,
|
| 564 |
+
"learning_rate": 4.851896936498276e-05,
|
| 565 |
+
"loss": 0.0999,
|
| 566 |
+
"step": 648
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.12,
|
| 570 |
+
"learning_rate": 4.843781700142017e-05,
|
| 571 |
+
"loss": 0.0949,
|
| 572 |
+
"step": 656
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.12,
|
| 576 |
+
"learning_rate": 4.835666463785758e-05,
|
| 577 |
+
"loss": 0.0821,
|
| 578 |
+
"step": 664
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.12,
|
| 582 |
+
"learning_rate": 4.827551227429499e-05,
|
| 583 |
+
"loss": 0.0947,
|
| 584 |
+
"step": 672
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.13,
|
| 588 |
+
"learning_rate": 4.81943599107324e-05,
|
| 589 |
+
"loss": 0.1451,
|
| 590 |
+
"step": 680
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.13,
|
| 594 |
+
"learning_rate": 4.811320754716982e-05,
|
| 595 |
+
"loss": 0.084,
|
| 596 |
+
"step": 688
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.13,
|
| 600 |
+
"learning_rate": 4.8032055183607226e-05,
|
| 601 |
+
"loss": 0.0659,
|
| 602 |
+
"step": 696
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.13,
|
| 606 |
+
"eval_accuracy": 0.9683668888351694,
|
| 607 |
+
"eval_f1": 0.9678054266835373,
|
| 608 |
+
"eval_loss": 0.13082686066627502,
|
| 609 |
+
"eval_precision": 0.9888318310769563,
|
| 610 |
+
"eval_recall": 0.9476546099046701,
|
| 611 |
+
"eval_runtime": 319.8968,
|
| 612 |
+
"eval_samples_per_second": 181.04,
|
| 613 |
+
"eval_steps_per_second": 22.632,
|
| 614 |
+
"step": 700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.13,
|
| 618 |
+
"learning_rate": 4.7950902820044635e-05,
|
| 619 |
+
"loss": 0.1008,
|
| 620 |
+
"step": 704
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.13,
|
| 624 |
+
"learning_rate": 4.786975045648205e-05,
|
| 625 |
+
"loss": 0.1163,
|
| 626 |
+
"step": 712
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.13,
|
| 630 |
+
"learning_rate": 4.778859809291946e-05,
|
| 631 |
+
"loss": 0.0985,
|
| 632 |
+
"step": 720
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.13,
|
| 636 |
+
"learning_rate": 4.770744572935687e-05,
|
| 637 |
+
"loss": 0.1044,
|
| 638 |
+
"step": 728
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.14,
|
| 642 |
+
"learning_rate": 4.7626293365794284e-05,
|
| 643 |
+
"loss": 0.0954,
|
| 644 |
+
"step": 736
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 0.14,
|
| 648 |
+
"learning_rate": 4.754514100223169e-05,
|
| 649 |
+
"loss": 0.084,
|
| 650 |
+
"step": 744
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.14,
|
| 654 |
+
"learning_rate": 4.74639886386691e-05,
|
| 655 |
+
"loss": 0.1018,
|
| 656 |
+
"step": 752
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 0.14,
|
| 660 |
+
"learning_rate": 4.738283627510651e-05,
|
| 661 |
+
"loss": 0.0698,
|
| 662 |
+
"step": 760
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.14,
|
| 666 |
+
"learning_rate": 4.730168391154393e-05,
|
| 667 |
+
"loss": 0.1247,
|
| 668 |
+
"step": 768
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.14,
|
| 672 |
+
"learning_rate": 4.722053154798134e-05,
|
| 673 |
+
"loss": 0.0551,
|
| 674 |
+
"step": 776
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.14,
|
| 678 |
+
"learning_rate": 4.7139379184418745e-05,
|
| 679 |
+
"loss": 0.0788,
|
| 680 |
+
"step": 784
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 0.15,
|
| 684 |
+
"learning_rate": 4.705822682085616e-05,
|
| 685 |
+
"loss": 0.1147,
|
| 686 |
+
"step": 792
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.15,
|
| 690 |
+
"learning_rate": 4.697707445729357e-05,
|
| 691 |
+
"loss": 0.0704,
|
| 692 |
+
"step": 800
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.15,
|
| 696 |
+
"eval_accuracy": 0.9718720861967746,
|
| 697 |
+
"eval_f1": 0.9722992160797185,
|
| 698 |
+
"eval_loss": 0.08833611011505127,
|
| 699 |
+
"eval_precision": 0.9609747899159664,
|
| 700 |
+
"eval_recall": 0.9838937261245139,
|
| 701 |
+
"eval_runtime": 320.3064,
|
| 702 |
+
"eval_samples_per_second": 180.808,
|
| 703 |
+
"eval_steps_per_second": 22.603,
|
| 704 |
+
"step": 800
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.15,
|
| 708 |
+
"learning_rate": 4.6895922093730985e-05,
|
| 709 |
+
"loss": 0.1034,
|
| 710 |
+
"step": 808
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.15,
|
| 714 |
+
"learning_rate": 4.6814769730168394e-05,
|
| 715 |
+
"loss": 0.1137,
|
| 716 |
+
"step": 816
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.15,
|
| 720 |
+
"learning_rate": 4.67336173666058e-05,
|
| 721 |
+
"loss": 0.0647,
|
| 722 |
+
"step": 824
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.15,
|
| 726 |
+
"learning_rate": 4.665246500304322e-05,
|
| 727 |
+
"loss": 0.1009,
|
| 728 |
+
"step": 832
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.15,
|
| 732 |
+
"learning_rate": 4.657131263948063e-05,
|
| 733 |
+
"loss": 0.0659,
|
| 734 |
+
"step": 840
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.16,
|
| 738 |
+
"learning_rate": 4.6490160275918036e-05,
|
| 739 |
+
"loss": 0.0619,
|
| 740 |
+
"step": 848
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.16,
|
| 744 |
+
"learning_rate": 4.640900791235545e-05,
|
| 745 |
+
"loss": 0.0762,
|
| 746 |
+
"step": 856
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.16,
|
| 750 |
+
"learning_rate": 4.632785554879286e-05,
|
| 751 |
+
"loss": 0.0348,
|
| 752 |
+
"step": 864
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.16,
|
| 756 |
+
"learning_rate": 4.624670318523027e-05,
|
| 757 |
+
"loss": 0.0976,
|
| 758 |
+
"step": 872
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.16,
|
| 762 |
+
"learning_rate": 4.6165550821667686e-05,
|
| 763 |
+
"loss": 0.0727,
|
| 764 |
+
"step": 880
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 0.16,
|
| 768 |
+
"learning_rate": 4.6084398458105094e-05,
|
| 769 |
+
"loss": 0.0572,
|
| 770 |
+
"step": 888
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.17,
|
| 774 |
+
"learning_rate": 4.600324609454251e-05,
|
| 775 |
+
"loss": 0.0582,
|
| 776 |
+
"step": 896
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.17,
|
| 780 |
+
"eval_accuracy": 0.973028973995925,
|
| 781 |
+
"eval_f1": 0.973034561336878,
|
| 782 |
+
"eval_loss": 0.08047471195459366,
|
| 783 |
+
"eval_precision": 0.9762028473449028,
|
| 784 |
+
"eval_recall": 0.9698867742712599,
|
| 785 |
+
"eval_runtime": 320.2553,
|
| 786 |
+
"eval_samples_per_second": 180.837,
|
| 787 |
+
"eval_steps_per_second": 22.607,
|
| 788 |
+
"step": 900
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 0.17,
|
| 792 |
+
"learning_rate": 4.592209373097991e-05,
|
| 793 |
+
"loss": 0.098,
|
| 794 |
+
"step": 904
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.17,
|
| 798 |
+
"learning_rate": 4.584094136741733e-05,
|
| 799 |
+
"loss": 0.1056,
|
| 800 |
+
"step": 912
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.17,
|
| 804 |
+
"learning_rate": 4.5759789003854744e-05,
|
| 805 |
+
"loss": 0.0511,
|
| 806 |
+
"step": 920
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 0.17,
|
| 810 |
+
"learning_rate": 4.567863664029215e-05,
|
| 811 |
+
"loss": 0.092,
|
| 812 |
+
"step": 928
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.17,
|
| 816 |
+
"learning_rate": 4.559748427672956e-05,
|
| 817 |
+
"loss": 0.0727,
|
| 818 |
+
"step": 936
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.17,
|
| 822 |
+
"learning_rate": 4.551633191316697e-05,
|
| 823 |
+
"loss": 0.0495,
|
| 824 |
+
"step": 944
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.18,
|
| 828 |
+
"learning_rate": 4.5435179549604386e-05,
|
| 829 |
+
"loss": 0.0912,
|
| 830 |
+
"step": 952
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.18,
|
| 834 |
+
"learning_rate": 4.5354027186041795e-05,
|
| 835 |
+
"loss": 0.0717,
|
| 836 |
+
"step": 960
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.18,
|
| 840 |
+
"learning_rate": 4.5272874822479204e-05,
|
| 841 |
+
"loss": 0.0395,
|
| 842 |
+
"step": 968
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.18,
|
| 846 |
+
"learning_rate": 4.519172245891662e-05,
|
| 847 |
+
"loss": 0.1236,
|
| 848 |
+
"step": 976
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.18,
|
| 852 |
+
"learning_rate": 4.511057009535403e-05,
|
| 853 |
+
"loss": 0.1675,
|
| 854 |
+
"step": 984
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.18,
|
| 858 |
+
"learning_rate": 4.502941773179144e-05,
|
| 859 |
+
"loss": 0.0863,
|
| 860 |
+
"step": 992
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.18,
|
| 864 |
+
"learning_rate": 4.494826536822885e-05,
|
| 865 |
+
"loss": 0.0867,
|
| 866 |
+
"step": 1000
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.18,
|
| 870 |
+
"eval_accuracy": 0.9773457195151432,
|
| 871 |
+
"eval_f1": 0.977261698440208,
|
| 872 |
+
"eval_loss": 0.06294739246368408,
|
| 873 |
+
"eval_precision": 0.984324267709388,
|
| 874 |
+
"eval_recall": 0.9702997556526827,
|
| 875 |
+
"eval_runtime": 318.1602,
|
| 876 |
+
"eval_samples_per_second": 182.028,
|
| 877 |
+
"eval_steps_per_second": 22.756,
|
| 878 |
+
"step": 1000
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.19,
|
| 882 |
+
"learning_rate": 4.486711300466626e-05,
|
| 883 |
+
"loss": 0.1165,
|
| 884 |
+
"step": 1008
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.19,
|
| 888 |
+
"learning_rate": 4.478596064110368e-05,
|
| 889 |
+
"loss": 0.083,
|
| 890 |
+
"step": 1016
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.19,
|
| 894 |
+
"learning_rate": 4.470480827754109e-05,
|
| 895 |
+
"loss": 0.083,
|
| 896 |
+
"step": 1024
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.19,
|
| 900 |
+
"learning_rate": 4.4623655913978496e-05,
|
| 901 |
+
"loss": 0.0775,
|
| 902 |
+
"step": 1032
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.19,
|
| 906 |
+
"learning_rate": 4.454250355041591e-05,
|
| 907 |
+
"loss": 0.0958,
|
| 908 |
+
"step": 1040
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 0.19,
|
| 912 |
+
"learning_rate": 4.446135118685332e-05,
|
| 913 |
+
"loss": 0.0814,
|
| 914 |
+
"step": 1048
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.19,
|
| 918 |
+
"learning_rate": 4.438019882329073e-05,
|
| 919 |
+
"loss": 0.0633,
|
| 920 |
+
"step": 1056
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.2,
|
| 924 |
+
"learning_rate": 4.4299046459728145e-05,
|
| 925 |
+
"loss": 0.0876,
|
| 926 |
+
"step": 1064
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.2,
|
| 930 |
+
"learning_rate": 4.4217894096165554e-05,
|
| 931 |
+
"loss": 0.0853,
|
| 932 |
+
"step": 1072
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 0.2,
|
| 936 |
+
"learning_rate": 4.413674173260296e-05,
|
| 937 |
+
"loss": 0.0678,
|
| 938 |
+
"step": 1080
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.2,
|
| 942 |
+
"learning_rate": 4.405558936904037e-05,
|
| 943 |
+
"loss": 0.0541,
|
| 944 |
+
"step": 1088
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.2,
|
| 948 |
+
"learning_rate": 4.397443700547779e-05,
|
| 949 |
+
"loss": 0.1059,
|
| 950 |
+
"step": 1096
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 0.2,
|
| 954 |
+
"eval_accuracy": 0.9705252615947785,
|
| 955 |
+
"eval_f1": 0.9711601817905354,
|
| 956 |
+
"eval_loss": 0.08772066235542297,
|
| 957 |
+
"eval_precision": 0.9538364529403955,
|
| 958 |
+
"eval_recall": 0.9891248236225351,
|
| 959 |
+
"eval_runtime": 320.9793,
|
| 960 |
+
"eval_samples_per_second": 180.429,
|
| 961 |
+
"eval_steps_per_second": 22.556,
|
| 962 |
+
"step": 1100
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.2,
|
| 966 |
+
"learning_rate": 4.38932846419152e-05,
|
| 967 |
+
"loss": 0.0416,
|
| 968 |
+
"step": 1104
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.2,
|
| 972 |
+
"learning_rate": 4.3812132278352605e-05,
|
| 973 |
+
"loss": 0.108,
|
| 974 |
+
"step": 1112
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.21,
|
| 978 |
+
"learning_rate": 4.373097991479002e-05,
|
| 979 |
+
"loss": 0.0706,
|
| 980 |
+
"step": 1120
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.21,
|
| 984 |
+
"learning_rate": 4.364982755122743e-05,
|
| 985 |
+
"loss": 0.1139,
|
| 986 |
+
"step": 1128
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.21,
|
| 990 |
+
"learning_rate": 4.3568675187664845e-05,
|
| 991 |
+
"loss": 0.0468,
|
| 992 |
+
"step": 1136
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 0.21,
|
| 996 |
+
"learning_rate": 4.3487522824102254e-05,
|
| 997 |
+
"loss": 0.113,
|
| 998 |
+
"step": 1144
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 0.21,
|
| 1002 |
+
"learning_rate": 4.340637046053966e-05,
|
| 1003 |
+
"loss": 0.0934,
|
| 1004 |
+
"step": 1152
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.21,
|
| 1008 |
+
"learning_rate": 4.332521809697708e-05,
|
| 1009 |
+
"loss": 0.1363,
|
| 1010 |
+
"step": 1160
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.22,
|
| 1014 |
+
"learning_rate": 4.324406573341449e-05,
|
| 1015 |
+
"loss": 0.0816,
|
| 1016 |
+
"step": 1168
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 0.22,
|
| 1020 |
+
"learning_rate": 4.31629133698519e-05,
|
| 1021 |
+
"loss": 0.0846,
|
| 1022 |
+
"step": 1176
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.22,
|
| 1026 |
+
"learning_rate": 4.308176100628931e-05,
|
| 1027 |
+
"loss": 0.0652,
|
| 1028 |
+
"step": 1184
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.22,
|
| 1032 |
+
"learning_rate": 4.300060864272672e-05,
|
| 1033 |
+
"loss": 0.1003,
|
| 1034 |
+
"step": 1192
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.22,
|
| 1038 |
+
"learning_rate": 4.291945627916413e-05,
|
| 1039 |
+
"loss": 0.1021,
|
| 1040 |
+
"step": 1200
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.22,
|
| 1044 |
+
"eval_accuracy": 0.9757571571640709,
|
| 1045 |
+
"eval_f1": 0.9760752504941722,
|
| 1046 |
+
"eval_loss": 0.0945325568318367,
|
| 1047 |
+
"eval_precision": 0.966685793364161,
|
| 1048 |
+
"eval_recall": 0.9856488969955605,
|
| 1049 |
+
"eval_runtime": 324.5414,
|
| 1050 |
+
"eval_samples_per_second": 178.449,
|
| 1051 |
+
"eval_steps_per_second": 22.308,
|
| 1052 |
+
"step": 1200
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.22,
|
| 1056 |
+
"learning_rate": 4.2838303915601546e-05,
|
| 1057 |
+
"loss": 0.0968,
|
| 1058 |
+
"step": 1208
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 0.22,
|
| 1062 |
+
"learning_rate": 4.2757151552038955e-05,
|
| 1063 |
+
"loss": 0.076,
|
| 1064 |
+
"step": 1216
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 0.23,
|
| 1068 |
+
"learning_rate": 4.267599918847637e-05,
|
| 1069 |
+
"loss": 0.0614,
|
| 1070 |
+
"step": 1224
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.23,
|
| 1074 |
+
"learning_rate": 4.259484682491377e-05,
|
| 1075 |
+
"loss": 0.0889,
|
| 1076 |
+
"step": 1232
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 0.23,
|
| 1080 |
+
"learning_rate": 4.251369446135119e-05,
|
| 1081 |
+
"loss": 0.0951,
|
| 1082 |
+
"step": 1240
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.23,
|
| 1086 |
+
"learning_rate": 4.2432542097788604e-05,
|
| 1087 |
+
"loss": 0.0406,
|
| 1088 |
+
"step": 1248
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.23,
|
| 1092 |
+
"learning_rate": 4.235138973422601e-05,
|
| 1093 |
+
"loss": 0.0786,
|
| 1094 |
+
"step": 1256
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.23,
|
| 1098 |
+
"learning_rate": 4.227023737066342e-05,
|
| 1099 |
+
"loss": 0.104,
|
| 1100 |
+
"step": 1264
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 0.23,
|
| 1104 |
+
"learning_rate": 4.218908500710083e-05,
|
| 1105 |
+
"loss": 0.1195,
|
| 1106 |
+
"step": 1272
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 0.24,
|
| 1110 |
+
"learning_rate": 4.210793264353825e-05,
|
| 1111 |
+
"loss": 0.0644,
|
| 1112 |
+
"step": 1280
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 0.24,
|
| 1116 |
+
"learning_rate": 4.2026780279975656e-05,
|
| 1117 |
+
"loss": 0.0485,
|
| 1118 |
+
"step": 1288
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 0.24,
|
| 1122 |
+
"learning_rate": 4.1945627916413064e-05,
|
| 1123 |
+
"loss": 0.0737,
|
| 1124 |
+
"step": 1296
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 0.24,
|
| 1128 |
+
"eval_accuracy": 0.9794522913285216,
|
| 1129 |
+
"eval_f1": 0.9795680093403386,
|
| 1130 |
+
"eval_loss": 0.062313616275787354,
|
| 1131 |
+
"eval_precision": 0.9774199074867226,
|
| 1132 |
+
"eval_recall": 0.9817255738720446,
|
| 1133 |
+
"eval_runtime": 318.2777,
|
| 1134 |
+
"eval_samples_per_second": 181.961,
|
| 1135 |
+
"eval_steps_per_second": 22.747,
|
| 1136 |
+
"step": 1300
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.24,
|
| 1140 |
+
"learning_rate": 4.186447555285048e-05,
|
| 1141 |
+
"loss": 0.1107,
|
| 1142 |
+
"step": 1304
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 0.24,
|
| 1146 |
+
"learning_rate": 4.178332318928789e-05,
|
| 1147 |
+
"loss": 0.0381,
|
| 1148 |
+
"step": 1312
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 0.24,
|
| 1152 |
+
"learning_rate": 4.17021708257253e-05,
|
| 1153 |
+
"loss": 0.0901,
|
| 1154 |
+
"step": 1320
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.24,
|
| 1158 |
+
"learning_rate": 4.1621018462162714e-05,
|
| 1159 |
+
"loss": 0.1092,
|
| 1160 |
+
"step": 1328
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.25,
|
| 1164 |
+
"learning_rate": 4.153986609860012e-05,
|
| 1165 |
+
"loss": 0.0992,
|
| 1166 |
+
"step": 1336
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.25,
|
| 1170 |
+
"learning_rate": 4.145871373503754e-05,
|
| 1171 |
+
"loss": 0.0763,
|
| 1172 |
+
"step": 1344
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.25,
|
| 1176 |
+
"learning_rate": 4.137756137147495e-05,
|
| 1177 |
+
"loss": 0.0871,
|
| 1178 |
+
"step": 1352
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.25,
|
| 1182 |
+
"learning_rate": 4.1296409007912356e-05,
|
| 1183 |
+
"loss": 0.0894,
|
| 1184 |
+
"step": 1360
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 0.25,
|
| 1188 |
+
"learning_rate": 4.121525664434977e-05,
|
| 1189 |
+
"loss": 0.0837,
|
| 1190 |
+
"step": 1368
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 0.25,
|
| 1194 |
+
"learning_rate": 4.113410428078718e-05,
|
| 1195 |
+
"loss": 0.0786,
|
| 1196 |
+
"step": 1376
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 0.25,
|
| 1200 |
+
"learning_rate": 4.105295191722459e-05,
|
| 1201 |
+
"loss": 0.0617,
|
| 1202 |
+
"step": 1384
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 0.26,
|
| 1206 |
+
"learning_rate": 4.0971799553662005e-05,
|
| 1207 |
+
"loss": 0.1143,
|
| 1208 |
+
"step": 1392
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.26,
|
| 1212 |
+
"learning_rate": 4.0890647190099414e-05,
|
| 1213 |
+
"loss": 0.0505,
|
| 1214 |
+
"step": 1400
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.26,
|
| 1218 |
+
"eval_accuracy": 0.980471043271057,
|
| 1219 |
+
"eval_f1": 0.9804956283304879,
|
| 1220 |
+
"eval_loss": 0.060013771057128906,
|
| 1221 |
+
"eval_precision": 0.9826477704804701,
|
| 1222 |
+
"eval_recall": 0.9783528925904257,
|
| 1223 |
+
"eval_runtime": 321.185,
|
| 1224 |
+
"eval_samples_per_second": 180.314,
|
| 1225 |
+
"eval_steps_per_second": 22.542,
|
| 1226 |
+
"step": 1400
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.26,
|
| 1230 |
+
"learning_rate": 4.080949482653682e-05,
|
| 1231 |
+
"loss": 0.0471,
|
| 1232 |
+
"step": 1408
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 0.26,
|
| 1236 |
+
"learning_rate": 4.072834246297423e-05,
|
| 1237 |
+
"loss": 0.0651,
|
| 1238 |
+
"step": 1416
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 0.26,
|
| 1242 |
+
"learning_rate": 4.064719009941165e-05,
|
| 1243 |
+
"loss": 0.0796,
|
| 1244 |
+
"step": 1424
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.26,
|
| 1248 |
+
"learning_rate": 4.0566037735849064e-05,
|
| 1249 |
+
"loss": 0.0486,
|
| 1250 |
+
"step": 1432
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.27,
|
| 1254 |
+
"learning_rate": 4.0484885372286466e-05,
|
| 1255 |
+
"loss": 0.1081,
|
| 1256 |
+
"step": 1440
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.27,
|
| 1260 |
+
"learning_rate": 4.040373300872388e-05,
|
| 1261 |
+
"loss": 0.0397,
|
| 1262 |
+
"step": 1448
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.27,
|
| 1266 |
+
"learning_rate": 4.032258064516129e-05,
|
| 1267 |
+
"loss": 0.0586,
|
| 1268 |
+
"step": 1456
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 0.27,
|
| 1272 |
+
"learning_rate": 4.0241428281598706e-05,
|
| 1273 |
+
"loss": 0.0511,
|
| 1274 |
+
"step": 1464
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.27,
|
| 1278 |
+
"learning_rate": 4.0160275918036115e-05,
|
| 1279 |
+
"loss": 0.0273,
|
| 1280 |
+
"step": 1472
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 0.27,
|
| 1284 |
+
"learning_rate": 4.0079123554473524e-05,
|
| 1285 |
+
"loss": 0.09,
|
| 1286 |
+
"step": 1480
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 0.27,
|
| 1290 |
+
"learning_rate": 3.999797119091094e-05,
|
| 1291 |
+
"loss": 0.1026,
|
| 1292 |
+
"step": 1488
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 0.28,
|
| 1296 |
+
"learning_rate": 3.991681882734835e-05,
|
| 1297 |
+
"loss": 0.1159,
|
| 1298 |
+
"step": 1496
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.28,
|
| 1302 |
+
"eval_accuracy": 0.979538626238906,
|
| 1303 |
+
"eval_f1": 0.9796374258956955,
|
| 1304 |
+
"eval_loss": 0.059322111308574677,
|
| 1305 |
+
"eval_precision": 0.9782757910632164,
|
| 1306 |
+
"eval_recall": 0.9810028564545549,
|
| 1307 |
+
"eval_runtime": 320.1061,
|
| 1308 |
+
"eval_samples_per_second": 180.921,
|
| 1309 |
+
"eval_steps_per_second": 22.618,
|
| 1310 |
+
"step": 1500
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 0.28,
|
| 1314 |
+
"learning_rate": 3.983566646378576e-05,
|
| 1315 |
+
"loss": 0.055,
|
| 1316 |
+
"step": 1504
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 0.28,
|
| 1320 |
+
"learning_rate": 3.975451410022317e-05,
|
| 1321 |
+
"loss": 0.0556,
|
| 1322 |
+
"step": 1512
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.28,
|
| 1326 |
+
"learning_rate": 3.967336173666058e-05,
|
| 1327 |
+
"loss": 0.0569,
|
| 1328 |
+
"step": 1520
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.28,
|
| 1332 |
+
"learning_rate": 3.959220937309799e-05,
|
| 1333 |
+
"loss": 0.0622,
|
| 1334 |
+
"step": 1528
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.28,
|
| 1338 |
+
"learning_rate": 3.9511057009535407e-05,
|
| 1339 |
+
"loss": 0.0546,
|
| 1340 |
+
"step": 1536
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.28,
|
| 1344 |
+
"learning_rate": 3.9429904645972815e-05,
|
| 1345 |
+
"loss": 0.08,
|
| 1346 |
+
"step": 1544
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.29,
|
| 1350 |
+
"learning_rate": 3.934875228241023e-05,
|
| 1351 |
+
"loss": 0.0657,
|
| 1352 |
+
"step": 1552
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.29,
|
| 1356 |
+
"learning_rate": 3.926759991884763e-05,
|
| 1357 |
+
"loss": 0.0414,
|
| 1358 |
+
"step": 1560
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.29,
|
| 1362 |
+
"learning_rate": 3.918644755528505e-05,
|
| 1363 |
+
"loss": 0.0747,
|
| 1364 |
+
"step": 1568
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 0.29,
|
| 1368 |
+
"learning_rate": 3.9105295191722465e-05,
|
| 1369 |
+
"loss": 0.0617,
|
| 1370 |
+
"step": 1576
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.29,
|
| 1374 |
+
"learning_rate": 3.9024142828159874e-05,
|
| 1375 |
+
"loss": 0.0318,
|
| 1376 |
+
"step": 1584
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 0.29,
|
| 1380 |
+
"learning_rate": 3.894299046459728e-05,
|
| 1381 |
+
"loss": 0.1025,
|
| 1382 |
+
"step": 1592
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.29,
|
| 1386 |
+
"learning_rate": 3.886183810103469e-05,
|
| 1387 |
+
"loss": 0.1075,
|
| 1388 |
+
"step": 1600
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.29,
|
| 1392 |
+
"eval_accuracy": 0.9777946610491418,
|
| 1393 |
+
"eval_f1": 0.9776805859280085,
|
| 1394 |
+
"eval_loss": 0.07334825396537781,
|
| 1395 |
+
"eval_precision": 0.9861699520324919,
|
| 1396 |
+
"eval_recall": 0.969336132429363,
|
| 1397 |
+
"eval_runtime": 318.519,
|
| 1398 |
+
"eval_samples_per_second": 181.823,
|
| 1399 |
+
"eval_steps_per_second": 22.73,
|
| 1400 |
+
"step": 1600
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 0.3,
|
| 1404 |
+
"learning_rate": 3.878068573747211e-05,
|
| 1405 |
+
"loss": 0.0735,
|
| 1406 |
+
"step": 1608
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 0.3,
|
| 1410 |
+
"learning_rate": 3.8699533373909516e-05,
|
| 1411 |
+
"loss": 0.1103,
|
| 1412 |
+
"step": 1616
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.3,
|
| 1416 |
+
"learning_rate": 3.8618381010346925e-05,
|
| 1417 |
+
"loss": 0.0638,
|
| 1418 |
+
"step": 1624
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.3,
|
| 1422 |
+
"learning_rate": 3.853722864678434e-05,
|
| 1423 |
+
"loss": 0.0608,
|
| 1424 |
+
"step": 1632
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.3,
|
| 1428 |
+
"learning_rate": 3.845607628322175e-05,
|
| 1429 |
+
"loss": 0.0609,
|
| 1430 |
+
"step": 1640
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.3,
|
| 1434 |
+
"learning_rate": 3.837492391965916e-05,
|
| 1435 |
+
"loss": 0.0606,
|
| 1436 |
+
"step": 1648
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 0.31,
|
| 1440 |
+
"learning_rate": 3.8293771556096574e-05,
|
| 1441 |
+
"loss": 0.0834,
|
| 1442 |
+
"step": 1656
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.31,
|
| 1446 |
+
"learning_rate": 3.821261919253398e-05,
|
| 1447 |
+
"loss": 0.0889,
|
| 1448 |
+
"step": 1664
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 0.31,
|
| 1452 |
+
"learning_rate": 3.81314668289714e-05,
|
| 1453 |
+
"loss": 0.0464,
|
| 1454 |
+
"step": 1672
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 0.31,
|
| 1458 |
+
"learning_rate": 3.805031446540881e-05,
|
| 1459 |
+
"loss": 0.0473,
|
| 1460 |
+
"step": 1680
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 0.31,
|
| 1464 |
+
"learning_rate": 3.796916210184622e-05,
|
| 1465 |
+
"loss": 0.0484,
|
| 1466 |
+
"step": 1688
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.31,
|
| 1470 |
+
"learning_rate": 3.788800973828363e-05,
|
| 1471 |
+
"loss": 0.0938,
|
| 1472 |
+
"step": 1696
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.31,
|
| 1476 |
+
"eval_accuracy": 0.9813343923749007,
|
| 1477 |
+
"eval_f1": 0.9813185863648146,
|
| 1478 |
+
"eval_loss": 0.06325095146894455,
|
| 1479 |
+
"eval_precision": 0.9855595667870036,
|
| 1480 |
+
"eval_recall": 0.9771139484461575,
|
| 1481 |
+
"eval_runtime": 318.5283,
|
| 1482 |
+
"eval_samples_per_second": 181.817,
|
| 1483 |
+
"eval_steps_per_second": 22.73,
|
| 1484 |
+
"step": 1700
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 0.31,
|
| 1488 |
+
"learning_rate": 3.780685737472104e-05,
|
| 1489 |
+
"loss": 0.0528,
|
| 1490 |
+
"step": 1704
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 0.32,
|
| 1494 |
+
"learning_rate": 3.772570501115845e-05,
|
| 1495 |
+
"loss": 0.0541,
|
| 1496 |
+
"step": 1712
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 0.32,
|
| 1500 |
+
"learning_rate": 3.7644552647595866e-05,
|
| 1501 |
+
"loss": 0.0737,
|
| 1502 |
+
"step": 1720
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.32,
|
| 1506 |
+
"learning_rate": 3.7563400284033275e-05,
|
| 1507 |
+
"loss": 0.0593,
|
| 1508 |
+
"step": 1728
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.32,
|
| 1512 |
+
"learning_rate": 3.7482247920470684e-05,
|
| 1513 |
+
"loss": 0.0866,
|
| 1514 |
+
"step": 1736
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.32,
|
| 1518 |
+
"learning_rate": 3.740109555690809e-05,
|
| 1519 |
+
"loss": 0.0564,
|
| 1520 |
+
"step": 1744
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 0.32,
|
| 1524 |
+
"learning_rate": 3.731994319334551e-05,
|
| 1525 |
+
"loss": 0.0599,
|
| 1526 |
+
"step": 1752
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 0.32,
|
| 1530 |
+
"learning_rate": 3.7238790829782924e-05,
|
| 1531 |
+
"loss": 0.0726,
|
| 1532 |
+
"step": 1760
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 0.33,
|
| 1536 |
+
"learning_rate": 3.7157638466220326e-05,
|
| 1537 |
+
"loss": 0.0675,
|
| 1538 |
+
"step": 1768
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 0.33,
|
| 1542 |
+
"learning_rate": 3.707648610265774e-05,
|
| 1543 |
+
"loss": 0.0783,
|
| 1544 |
+
"step": 1776
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.33,
|
| 1548 |
+
"learning_rate": 3.699533373909515e-05,
|
| 1549 |
+
"loss": 0.065,
|
| 1550 |
+
"step": 1784
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.33,
|
| 1554 |
+
"learning_rate": 3.6914181375532566e-05,
|
| 1555 |
+
"loss": 0.0599,
|
| 1556 |
+
"step": 1792
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.33,
|
| 1560 |
+
"learning_rate": 3.6833029011969975e-05,
|
| 1561 |
+
"loss": 0.0708,
|
| 1562 |
+
"step": 1800
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.33,
|
| 1566 |
+
"eval_accuracy": 0.9794695583105985,
|
| 1567 |
+
"eval_f1": 0.9794158890639337,
|
| 1568 |
+
"eval_loss": 0.06451611965894699,
|
| 1569 |
+
"eval_precision": 0.9854037483452937,
|
| 1570 |
+
"eval_recall": 0.9735003613587088,
|
| 1571 |
+
"eval_runtime": 315.4864,
|
| 1572 |
+
"eval_samples_per_second": 183.57,
|
| 1573 |
+
"eval_steps_per_second": 22.949,
|
| 1574 |
+
"step": 1800
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 0.33,
|
| 1578 |
+
"learning_rate": 3.6751876648407384e-05,
|
| 1579 |
+
"loss": 0.0811,
|
| 1580 |
+
"step": 1808
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 0.33,
|
| 1584 |
+
"learning_rate": 3.66707242848448e-05,
|
| 1585 |
+
"loss": 0.0723,
|
| 1586 |
+
"step": 1816
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.34,
|
| 1590 |
+
"learning_rate": 3.658957192128221e-05,
|
| 1591 |
+
"loss": 0.0748,
|
| 1592 |
+
"step": 1824
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.34,
|
| 1596 |
+
"learning_rate": 3.650841955771962e-05,
|
| 1597 |
+
"loss": 0.0869,
|
| 1598 |
+
"step": 1832
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.34,
|
| 1602 |
+
"learning_rate": 3.6427267194157034e-05,
|
| 1603 |
+
"loss": 0.0578,
|
| 1604 |
+
"step": 1840
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 0.34,
|
| 1608 |
+
"learning_rate": 3.634611483059444e-05,
|
| 1609 |
+
"loss": 0.0813,
|
| 1610 |
+
"step": 1848
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.34,
|
| 1614 |
+
"learning_rate": 3.626496246703185e-05,
|
| 1615 |
+
"loss": 0.0748,
|
| 1616 |
+
"step": 1856
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.34,
|
| 1620 |
+
"learning_rate": 3.618381010346927e-05,
|
| 1621 |
+
"loss": 0.0686,
|
| 1622 |
+
"step": 1864
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 0.34,
|
| 1626 |
+
"learning_rate": 3.6102657739906676e-05,
|
| 1627 |
+
"loss": 0.0697,
|
| 1628 |
+
"step": 1872
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.35,
|
| 1632 |
+
"learning_rate": 3.602150537634409e-05,
|
| 1633 |
+
"loss": 0.0728,
|
| 1634 |
+
"step": 1880
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.35,
|
| 1638 |
+
"learning_rate": 3.5940353012781494e-05,
|
| 1639 |
+
"loss": 0.079,
|
| 1640 |
+
"step": 1888
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.35,
|
| 1644 |
+
"learning_rate": 3.585920064921891e-05,
|
| 1645 |
+
"loss": 0.0719,
|
| 1646 |
+
"step": 1896
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 0.35,
|
| 1650 |
+
"eval_accuracy": 0.9799012328625203,
|
| 1651 |
+
"eval_f1": 0.9800301948943178,
|
| 1652 |
+
"eval_loss": 0.0590737946331501,
|
| 1653 |
+
"eval_precision": 0.9771133385789059,
|
| 1654 |
+
"eval_recall": 0.9829645180163128,
|
| 1655 |
+
"eval_runtime": 317.0745,
|
| 1656 |
+
"eval_samples_per_second": 182.651,
|
| 1657 |
+
"eval_steps_per_second": 22.834,
|
| 1658 |
+
"step": 1900
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.35,
|
| 1662 |
+
"learning_rate": 3.5778048285656325e-05,
|
| 1663 |
+
"loss": 0.0502,
|
| 1664 |
+
"step": 1904
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 0.35,
|
| 1668 |
+
"learning_rate": 3.5696895922093734e-05,
|
| 1669 |
+
"loss": 0.0592,
|
| 1670 |
+
"step": 1912
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 0.35,
|
| 1674 |
+
"learning_rate": 3.561574355853114e-05,
|
| 1675 |
+
"loss": 0.0849,
|
| 1676 |
+
"step": 1920
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.36,
|
| 1680 |
+
"learning_rate": 3.553459119496855e-05,
|
| 1681 |
+
"loss": 0.1225,
|
| 1682 |
+
"step": 1928
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.36,
|
| 1686 |
+
"learning_rate": 3.545343883140597e-05,
|
| 1687 |
+
"loss": 0.0855,
|
| 1688 |
+
"step": 1936
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 0.36,
|
| 1692 |
+
"learning_rate": 3.5372286467843377e-05,
|
| 1693 |
+
"loss": 0.0652,
|
| 1694 |
+
"step": 1944
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 0.36,
|
| 1698 |
+
"learning_rate": 3.5291134104280785e-05,
|
| 1699 |
+
"loss": 0.0585,
|
| 1700 |
+
"step": 1952
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 0.36,
|
| 1704 |
+
"learning_rate": 3.52099817407182e-05,
|
| 1705 |
+
"loss": 0.0544,
|
| 1706 |
+
"step": 1960
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.36,
|
| 1710 |
+
"learning_rate": 3.512882937715561e-05,
|
| 1711 |
+
"loss": 0.0836,
|
| 1712 |
+
"step": 1968
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 0.36,
|
| 1716 |
+
"learning_rate": 3.504767701359302e-05,
|
| 1717 |
+
"loss": 0.0505,
|
| 1718 |
+
"step": 1976
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.37,
|
| 1722 |
+
"learning_rate": 3.4966524650030435e-05,
|
| 1723 |
+
"loss": 0.0435,
|
| 1724 |
+
"step": 1984
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.37,
|
| 1728 |
+
"learning_rate": 3.4885372286467844e-05,
|
| 1729 |
+
"loss": 0.078,
|
| 1730 |
+
"step": 1992
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 0.37,
|
| 1734 |
+
"learning_rate": 3.480421992290526e-05,
|
| 1735 |
+
"loss": 0.0365,
|
| 1736 |
+
"step": 2000
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 0.37,
|
| 1740 |
+
"eval_accuracy": 0.9795731602030597,
|
| 1741 |
+
"eval_f1": 0.9794998873620184,
|
| 1742 |
+
"eval_loss": 0.08517105132341385,
|
| 1743 |
+
"eval_precision": 0.986457242582897,
|
| 1744 |
+
"eval_recall": 0.9726399834807448,
|
| 1745 |
+
"eval_runtime": 319.0106,
|
| 1746 |
+
"eval_samples_per_second": 181.543,
|
| 1747 |
+
"eval_steps_per_second": 22.695,
|
| 1748 |
+
"step": 2000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 0.37,
|
| 1752 |
+
"learning_rate": 3.472306755934267e-05,
|
| 1753 |
+
"loss": 0.0755,
|
| 1754 |
+
"step": 2008
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.37,
|
| 1758 |
+
"learning_rate": 3.464191519578008e-05,
|
| 1759 |
+
"loss": 0.0446,
|
| 1760 |
+
"step": 2016
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.37,
|
| 1764 |
+
"learning_rate": 3.456076283221749e-05,
|
| 1765 |
+
"loss": 0.045,
|
| 1766 |
+
"step": 2024
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.37,
|
| 1770 |
+
"learning_rate": 3.44796104686549e-05,
|
| 1771 |
+
"loss": 0.0446,
|
| 1772 |
+
"step": 2032
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 0.38,
|
| 1776 |
+
"learning_rate": 3.439845810509231e-05,
|
| 1777 |
+
"loss": 0.0601,
|
| 1778 |
+
"step": 2040
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 0.38,
|
| 1782 |
+
"learning_rate": 3.4317305741529726e-05,
|
| 1783 |
+
"loss": 0.0784,
|
| 1784 |
+
"step": 2048
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 0.38,
|
| 1788 |
+
"learning_rate": 3.4236153377967135e-05,
|
| 1789 |
+
"loss": 0.0601,
|
| 1790 |
+
"step": 2056
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 0.38,
|
| 1794 |
+
"learning_rate": 3.4155001014404544e-05,
|
| 1795 |
+
"loss": 0.0628,
|
| 1796 |
+
"step": 2064
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 0.38,
|
| 1800 |
+
"learning_rate": 3.407384865084195e-05,
|
| 1801 |
+
"loss": 0.0805,
|
| 1802 |
+
"step": 2072
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.38,
|
| 1806 |
+
"learning_rate": 3.399269628727937e-05,
|
| 1807 |
+
"loss": 0.0429,
|
| 1808 |
+
"step": 2080
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.38,
|
| 1812 |
+
"learning_rate": 3.3911543923716785e-05,
|
| 1813 |
+
"loss": 0.0589,
|
| 1814 |
+
"step": 2088
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 0.39,
|
| 1818 |
+
"learning_rate": 3.383039156015419e-05,
|
| 1819 |
+
"loss": 0.0564,
|
| 1820 |
+
"step": 2096
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 0.39,
|
| 1824 |
+
"eval_accuracy": 0.9813171253928239,
|
| 1825 |
+
"eval_f1": 0.9813274427915645,
|
| 1826 |
+
"eval_loss": 0.05837487801909447,
|
| 1827 |
+
"eval_precision": 0.9841808300737305,
|
| 1828 |
+
"eval_recall": 0.9784905530508999,
|
| 1829 |
+
"eval_runtime": 319.719,
|
| 1830 |
+
"eval_samples_per_second": 181.14,
|
| 1831 |
+
"eval_steps_per_second": 22.645,
|
| 1832 |
+
"step": 2100
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 0.39,
|
| 1836 |
+
"learning_rate": 3.37492391965916e-05,
|
| 1837 |
+
"loss": 0.0755,
|
| 1838 |
+
"step": 2104
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 0.39,
|
| 1842 |
+
"learning_rate": 3.366808683302901e-05,
|
| 1843 |
+
"loss": 0.0578,
|
| 1844 |
+
"step": 2112
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.39,
|
| 1848 |
+
"learning_rate": 3.358693446946643e-05,
|
| 1849 |
+
"loss": 0.076,
|
| 1850 |
+
"step": 2120
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.39,
|
| 1854 |
+
"learning_rate": 3.3505782105903836e-05,
|
| 1855 |
+
"loss": 0.0767,
|
| 1856 |
+
"step": 2128
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 0.39,
|
| 1860 |
+
"learning_rate": 3.3424629742341245e-05,
|
| 1861 |
+
"loss": 0.0393,
|
| 1862 |
+
"step": 2136
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 0.39,
|
| 1866 |
+
"learning_rate": 3.334347737877866e-05,
|
| 1867 |
+
"loss": 0.057,
|
| 1868 |
+
"step": 2144
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 0.4,
|
| 1872 |
+
"learning_rate": 3.326232501521607e-05,
|
| 1873 |
+
"loss": 0.0671,
|
| 1874 |
+
"step": 2152
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.4,
|
| 1878 |
+
"learning_rate": 3.318117265165348e-05,
|
| 1879 |
+
"loss": 0.0338,
|
| 1880 |
+
"step": 2160
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 0.4,
|
| 1884 |
+
"learning_rate": 3.3100020288090894e-05,
|
| 1885 |
+
"loss": 0.0637,
|
| 1886 |
+
"step": 2168
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.4,
|
| 1890 |
+
"learning_rate": 3.30188679245283e-05,
|
| 1891 |
+
"loss": 0.0585,
|
| 1892 |
+
"step": 2176
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.4,
|
| 1896 |
+
"learning_rate": 3.293771556096571e-05,
|
| 1897 |
+
"loss": 0.0699,
|
| 1898 |
+
"step": 2184
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.4,
|
| 1902 |
+
"learning_rate": 3.285656319740313e-05,
|
| 1903 |
+
"loss": 0.0357,
|
| 1904 |
+
"step": 2192
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.41,
|
| 1908 |
+
"learning_rate": 3.2775410833840536e-05,
|
| 1909 |
+
"loss": 0.0951,
|
| 1910 |
+
"step": 2200
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 0.41,
|
| 1914 |
+
"eval_accuracy": 0.9820250716579756,
|
| 1915 |
+
"eval_f1": 0.9820328276290582,
|
| 1916 |
+
"eval_loss": 0.05581849440932274,
|
| 1917 |
+
"eval_precision": 0.9850079634374351,
|
| 1918 |
+
"eval_recall": 0.9790756100079154,
|
| 1919 |
+
"eval_runtime": 315.2538,
|
| 1920 |
+
"eval_samples_per_second": 183.706,
|
| 1921 |
+
"eval_steps_per_second": 22.966,
|
| 1922 |
+
"step": 2200
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 0.41,
|
| 1926 |
+
"learning_rate": 3.269425847027795e-05,
|
| 1927 |
+
"loss": 0.0712,
|
| 1928 |
+
"step": 2208
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.41,
|
| 1932 |
+
"learning_rate": 3.2613106106715354e-05,
|
| 1933 |
+
"loss": 0.0666,
|
| 1934 |
+
"step": 2216
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.41,
|
| 1938 |
+
"learning_rate": 3.253195374315277e-05,
|
| 1939 |
+
"loss": 0.0806,
|
| 1940 |
+
"step": 2224
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 0.41,
|
| 1944 |
+
"learning_rate": 3.2450801379590186e-05,
|
| 1945 |
+
"loss": 0.0511,
|
| 1946 |
+
"step": 2232
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.41,
|
| 1950 |
+
"learning_rate": 3.2369649016027595e-05,
|
| 1951 |
+
"loss": 0.0523,
|
| 1952 |
+
"step": 2240
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 0.41,
|
| 1956 |
+
"learning_rate": 3.2288496652465004e-05,
|
| 1957 |
+
"loss": 0.0738,
|
| 1958 |
+
"step": 2248
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 0.42,
|
| 1962 |
+
"learning_rate": 3.220734428890241e-05,
|
| 1963 |
+
"loss": 0.0555,
|
| 1964 |
+
"step": 2256
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 0.42,
|
| 1968 |
+
"learning_rate": 3.212619192533983e-05,
|
| 1969 |
+
"loss": 0.0723,
|
| 1970 |
+
"step": 2264
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.42,
|
| 1974 |
+
"learning_rate": 3.204503956177724e-05,
|
| 1975 |
+
"loss": 0.0603,
|
| 1976 |
+
"step": 2272
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.42,
|
| 1980 |
+
"learning_rate": 3.1963887198214646e-05,
|
| 1981 |
+
"loss": 0.0477,
|
| 1982 |
+
"step": 2280
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 0.42,
|
| 1986 |
+
"learning_rate": 3.188273483465206e-05,
|
| 1987 |
+
"loss": 0.0724,
|
| 1988 |
+
"step": 2288
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 0.42,
|
| 1992 |
+
"learning_rate": 3.180158247108947e-05,
|
| 1993 |
+
"loss": 0.0909,
|
| 1994 |
+
"step": 2296
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.42,
|
| 1998 |
+
"eval_accuracy": 0.9805573781814414,
|
| 1999 |
+
"eval_f1": 0.9807237990892596,
|
| 2000 |
+
"eval_loss": 0.05231529101729393,
|
| 2001 |
+
"eval_precision": 0.9757127771911299,
|
| 2002 |
+
"eval_recall": 0.9857865574560347,
|
| 2003 |
+
"eval_runtime": 314.1871,
|
| 2004 |
+
"eval_samples_per_second": 184.33,
|
| 2005 |
+
"eval_steps_per_second": 23.044,
|
| 2006 |
+
"step": 2300
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 0.42,
|
| 2010 |
+
"learning_rate": 3.172043010752688e-05,
|
| 2011 |
+
"loss": 0.0451,
|
| 2012 |
+
"step": 2304
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.43,
|
| 2016 |
+
"learning_rate": 3.1639277743964295e-05,
|
| 2017 |
+
"loss": 0.0539,
|
| 2018 |
+
"step": 2312
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.43,
|
| 2022 |
+
"learning_rate": 3.1558125380401704e-05,
|
| 2023 |
+
"loss": 0.0669,
|
| 2024 |
+
"step": 2320
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 0.43,
|
| 2028 |
+
"learning_rate": 3.147697301683912e-05,
|
| 2029 |
+
"loss": 0.0439,
|
| 2030 |
+
"step": 2328
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 0.43,
|
| 2034 |
+
"learning_rate": 3.139582065327653e-05,
|
| 2035 |
+
"loss": 0.0849,
|
| 2036 |
+
"step": 2336
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 0.43,
|
| 2040 |
+
"learning_rate": 3.131466828971394e-05,
|
| 2041 |
+
"loss": 0.0798,
|
| 2042 |
+
"step": 2344
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.43,
|
| 2046 |
+
"learning_rate": 3.123351592615135e-05,
|
| 2047 |
+
"loss": 0.0699,
|
| 2048 |
+
"step": 2352
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 0.43,
|
| 2052 |
+
"learning_rate": 3.115236356258876e-05,
|
| 2053 |
+
"loss": 0.0707,
|
| 2054 |
+
"step": 2360
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.44,
|
| 2058 |
+
"learning_rate": 3.107121119902617e-05,
|
| 2059 |
+
"loss": 0.0432,
|
| 2060 |
+
"step": 2368
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 0.44,
|
| 2064 |
+
"learning_rate": 3.099005883546359e-05,
|
| 2065 |
+
"loss": 0.074,
|
| 2066 |
+
"step": 2376
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 0.44,
|
| 2070 |
+
"learning_rate": 3.0908906471900996e-05,
|
| 2071 |
+
"loss": 0.091,
|
| 2072 |
+
"step": 2384
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 0.44,
|
| 2076 |
+
"learning_rate": 3.0827754108338405e-05,
|
| 2077 |
+
"loss": 0.0482,
|
| 2078 |
+
"step": 2392
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 0.44,
|
| 2082 |
+
"learning_rate": 3.0746601744775814e-05,
|
| 2083 |
+
"loss": 0.0673,
|
| 2084 |
+
"step": 2400
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 0.44,
|
| 2088 |
+
"eval_accuracy": 0.9821804744966675,
|
| 2089 |
+
"eval_f1": 0.9821866261608035,
|
| 2090 |
+
"eval_loss": 0.05170663446187973,
|
| 2091 |
+
"eval_precision": 0.9852477750458842,
|
| 2092 |
+
"eval_recall": 0.9791444402381526,
|
| 2093 |
+
"eval_runtime": 314.3647,
|
| 2094 |
+
"eval_samples_per_second": 184.226,
|
| 2095 |
+
"eval_steps_per_second": 23.031,
|
| 2096 |
+
"step": 2400
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 0.44,
|
| 2100 |
+
"learning_rate": 3.066544938121323e-05,
|
| 2101 |
+
"loss": 0.0329,
|
| 2102 |
+
"step": 2408
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.44,
|
| 2106 |
+
"learning_rate": 3.0584297017650645e-05,
|
| 2107 |
+
"loss": 0.0723,
|
| 2108 |
+
"step": 2416
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 0.45,
|
| 2112 |
+
"learning_rate": 3.050314465408805e-05,
|
| 2113 |
+
"loss": 0.0806,
|
| 2114 |
+
"step": 2424
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 0.45,
|
| 2118 |
+
"learning_rate": 3.0421992290525463e-05,
|
| 2119 |
+
"loss": 0.0854,
|
| 2120 |
+
"step": 2432
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 0.45,
|
| 2124 |
+
"learning_rate": 3.0340839926962872e-05,
|
| 2125 |
+
"loss": 0.09,
|
| 2126 |
+
"step": 2440
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 0.45,
|
| 2130 |
+
"learning_rate": 3.0259687563400284e-05,
|
| 2131 |
+
"loss": 0.0537,
|
| 2132 |
+
"step": 2448
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 0.45,
|
| 2136 |
+
"learning_rate": 3.01785351998377e-05,
|
| 2137 |
+
"loss": 0.0342,
|
| 2138 |
+
"step": 2456
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.45,
|
| 2142 |
+
"learning_rate": 3.0097382836275105e-05,
|
| 2143 |
+
"loss": 0.0874,
|
| 2144 |
+
"step": 2464
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 0.46,
|
| 2148 |
+
"learning_rate": 3.001623047271252e-05,
|
| 2149 |
+
"loss": 0.0435,
|
| 2150 |
+
"step": 2472
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 0.46,
|
| 2154 |
+
"learning_rate": 2.9935078109149927e-05,
|
| 2155 |
+
"loss": 0.08,
|
| 2156 |
+
"step": 2480
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 0.46,
|
| 2160 |
+
"learning_rate": 2.9853925745587342e-05,
|
| 2161 |
+
"loss": 0.0496,
|
| 2162 |
+
"step": 2488
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.46,
|
| 2166 |
+
"learning_rate": 2.9772773382024755e-05,
|
| 2167 |
+
"loss": 0.0338,
|
| 2168 |
+
"step": 2496
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 0.46,
|
| 2172 |
+
"eval_accuracy": 0.9790551507407536,
|
| 2173 |
+
"eval_f1": 0.9788532278028625,
|
| 2174 |
+
"eval_loss": 0.07079575955867767,
|
| 2175 |
+
"eval_precision": 0.9918739400791408,
|
| 2176 |
+
"eval_recall": 0.9661699418384555,
|
| 2177 |
+
"eval_runtime": 321.7416,
|
| 2178 |
+
"eval_samples_per_second": 180.002,
|
| 2179 |
+
"eval_steps_per_second": 22.503,
|
| 2180 |
+
"step": 2500
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.46,
|
| 2184 |
+
"learning_rate": 2.9691621018462163e-05,
|
| 2185 |
+
"loss": 0.1099,
|
| 2186 |
+
"step": 2504
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 0.46,
|
| 2190 |
+
"learning_rate": 2.9610468654899576e-05,
|
| 2191 |
+
"loss": 0.0828,
|
| 2192 |
+
"step": 2512
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 0.46,
|
| 2196 |
+
"learning_rate": 2.9529316291336988e-05,
|
| 2197 |
+
"loss": 0.0722,
|
| 2198 |
+
"step": 2520
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 0.47,
|
| 2202 |
+
"learning_rate": 2.9448163927774397e-05,
|
| 2203 |
+
"loss": 0.0693,
|
| 2204 |
+
"step": 2528
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 0.47,
|
| 2208 |
+
"learning_rate": 2.936701156421181e-05,
|
| 2209 |
+
"loss": 0.0606,
|
| 2210 |
+
"step": 2536
|
| 2211 |
+
},
|
| 2212 |
+
{
|
| 2213 |
+
"epoch": 0.47,
|
| 2214 |
+
"learning_rate": 2.9285859200649218e-05,
|
| 2215 |
+
"loss": 0.0794,
|
| 2216 |
+
"step": 2544
|
| 2217 |
+
},
|
| 2218 |
+
{
|
| 2219 |
+
"epoch": 0.47,
|
| 2220 |
+
"learning_rate": 2.920470683708663e-05,
|
| 2221 |
+
"loss": 0.0794,
|
| 2222 |
+
"step": 2552
|
| 2223 |
+
},
|
| 2224 |
+
{
|
| 2225 |
+
"epoch": 0.47,
|
| 2226 |
+
"learning_rate": 2.9123554473524046e-05,
|
| 2227 |
+
"loss": 0.0353,
|
| 2228 |
+
"step": 2560
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 0.47,
|
| 2232 |
+
"learning_rate": 2.9042402109961452e-05,
|
| 2233 |
+
"loss": 0.087,
|
| 2234 |
+
"step": 2568
|
| 2235 |
+
},
|
| 2236 |
+
{
|
| 2237 |
+
"epoch": 0.47,
|
| 2238 |
+
"learning_rate": 2.8961249746398867e-05,
|
| 2239 |
+
"loss": 0.0708,
|
| 2240 |
+
"step": 2576
|
| 2241 |
+
},
|
| 2242 |
+
{
|
| 2243 |
+
"epoch": 0.48,
|
| 2244 |
+
"learning_rate": 2.8880097382836273e-05,
|
| 2245 |
+
"loss": 0.102,
|
| 2246 |
+
"step": 2584
|
| 2247 |
+
},
|
| 2248 |
+
{
|
| 2249 |
+
"epoch": 0.48,
|
| 2250 |
+
"learning_rate": 2.879894501927369e-05,
|
| 2251 |
+
"loss": 0.0967,
|
| 2252 |
+
"step": 2592
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 0.48,
|
| 2256 |
+
"learning_rate": 2.87177926557111e-05,
|
| 2257 |
+
"loss": 0.0528,
|
| 2258 |
+
"step": 2600
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 0.48,
|
| 2262 |
+
"eval_accuracy": 0.9754981524329178,
|
| 2263 |
+
"eval_f1": 0.9751056999000018,
|
| 2264 |
+
"eval_loss": 0.06617435812950134,
|
| 2265 |
+
"eval_precision": 0.994524763813341,
|
| 2266 |
+
"eval_recall": 0.956430464259903,
|
| 2267 |
+
"eval_runtime": 317.2016,
|
| 2268 |
+
"eval_samples_per_second": 182.578,
|
| 2269 |
+
"eval_steps_per_second": 22.825,
|
| 2270 |
+
"step": 2600
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 0.48,
|
| 2274 |
+
"learning_rate": 2.863664029214851e-05,
|
| 2275 |
+
"loss": 0.0414,
|
| 2276 |
+
"step": 2608
|
| 2277 |
+
},
|
| 2278 |
+
{
|
| 2279 |
+
"epoch": 0.48,
|
| 2280 |
+
"learning_rate": 2.8555487928585922e-05,
|
| 2281 |
+
"loss": 0.0709,
|
| 2282 |
+
"step": 2616
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 0.48,
|
| 2286 |
+
"learning_rate": 2.847433556502333e-05,
|
| 2287 |
+
"loss": 0.0676,
|
| 2288 |
+
"step": 2624
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 0.48,
|
| 2292 |
+
"learning_rate": 2.8393183201460743e-05,
|
| 2293 |
+
"loss": 0.0444,
|
| 2294 |
+
"step": 2632
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 0.49,
|
| 2298 |
+
"learning_rate": 2.8312030837898156e-05,
|
| 2299 |
+
"loss": 0.074,
|
| 2300 |
+
"step": 2640
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 0.49,
|
| 2304 |
+
"learning_rate": 2.8230878474335565e-05,
|
| 2305 |
+
"loss": 0.0723,
|
| 2306 |
+
"step": 2648
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 0.49,
|
| 2310 |
+
"learning_rate": 2.8149726110772977e-05,
|
| 2311 |
+
"loss": 0.0471,
|
| 2312 |
+
"step": 2656
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 0.49,
|
| 2316 |
+
"learning_rate": 2.8068573747210393e-05,
|
| 2317 |
+
"loss": 0.0475,
|
| 2318 |
+
"step": 2664
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 0.49,
|
| 2322 |
+
"learning_rate": 2.7987421383647798e-05,
|
| 2323 |
+
"loss": 0.0507,
|
| 2324 |
+
"step": 2672
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 0.49,
|
| 2328 |
+
"learning_rate": 2.7906269020085214e-05,
|
| 2329 |
+
"loss": 0.0465,
|
| 2330 |
+
"step": 2680
|
| 2331 |
+
},
|
| 2332 |
+
{
|
| 2333 |
+
"epoch": 0.5,
|
| 2334 |
+
"learning_rate": 2.782511665652262e-05,
|
| 2335 |
+
"loss": 0.0358,
|
| 2336 |
+
"step": 2688
|
| 2337 |
+
},
|
| 2338 |
+
{
|
| 2339 |
+
"epoch": 0.5,
|
| 2340 |
+
"learning_rate": 2.7743964292960035e-05,
|
| 2341 |
+
"loss": 0.0662,
|
| 2342 |
+
"step": 2696
|
| 2343 |
+
},
|
| 2344 |
+
{
|
| 2345 |
+
"epoch": 0.5,
|
| 2346 |
+
"eval_accuracy": 0.9828020858514349,
|
| 2347 |
+
"eval_f1": 0.9828204774389403,
|
| 2348 |
+
"eval_loss": 0.05437169969081879,
|
| 2349 |
+
"eval_precision": 0.9851654621529099,
|
| 2350 |
+
"eval_recall": 0.9804866297277764,
|
| 2351 |
+
"eval_runtime": 318.7613,
|
| 2352 |
+
"eval_samples_per_second": 181.685,
|
| 2353 |
+
"eval_steps_per_second": 22.713,
|
| 2354 |
+
"step": 2700
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.5,
|
| 2358 |
+
"learning_rate": 2.7662811929397447e-05,
|
| 2359 |
+
"loss": 0.0568,
|
| 2360 |
+
"step": 2704
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 0.5,
|
| 2364 |
+
"learning_rate": 2.7581659565834856e-05,
|
| 2365 |
+
"loss": 0.0558,
|
| 2366 |
+
"step": 2712
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 0.5,
|
| 2370 |
+
"learning_rate": 2.750050720227227e-05,
|
| 2371 |
+
"loss": 0.0535,
|
| 2372 |
+
"step": 2720
|
| 2373 |
+
},
|
| 2374 |
+
{
|
| 2375 |
+
"epoch": 0.5,
|
| 2376 |
+
"learning_rate": 2.7419354838709678e-05,
|
| 2377 |
+
"loss": 0.0641,
|
| 2378 |
+
"step": 2728
|
| 2379 |
+
},
|
| 2380 |
+
{
|
| 2381 |
+
"epoch": 0.5,
|
| 2382 |
+
"learning_rate": 2.733820247514709e-05,
|
| 2383 |
+
"loss": 0.0529,
|
| 2384 |
+
"step": 2736
|
| 2385 |
+
},
|
| 2386 |
+
{
|
| 2387 |
+
"epoch": 0.51,
|
| 2388 |
+
"learning_rate": 2.7257050111584502e-05,
|
| 2389 |
+
"loss": 0.0754,
|
| 2390 |
+
"step": 2744
|
| 2391 |
+
},
|
| 2392 |
+
{
|
| 2393 |
+
"epoch": 0.51,
|
| 2394 |
+
"learning_rate": 2.717589774802191e-05,
|
| 2395 |
+
"loss": 0.0746,
|
| 2396 |
+
"step": 2752
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 0.51,
|
| 2400 |
+
"learning_rate": 2.7094745384459323e-05,
|
| 2401 |
+
"loss": 0.0752,
|
| 2402 |
+
"step": 2760
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 0.51,
|
| 2406 |
+
"learning_rate": 2.7013593020896732e-05,
|
| 2407 |
+
"loss": 0.0733,
|
| 2408 |
+
"step": 2768
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 0.51,
|
| 2412 |
+
"learning_rate": 2.6932440657334145e-05,
|
| 2413 |
+
"loss": 0.0418,
|
| 2414 |
+
"step": 2776
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 0.51,
|
| 2418 |
+
"learning_rate": 2.685128829377156e-05,
|
| 2419 |
+
"loss": 0.0575,
|
| 2420 |
+
"step": 2784
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 0.51,
|
| 2424 |
+
"learning_rate": 2.6770135930208966e-05,
|
| 2425 |
+
"loss": 0.0463,
|
| 2426 |
+
"step": 2792
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 0.52,
|
| 2430 |
+
"learning_rate": 2.668898356664638e-05,
|
| 2431 |
+
"loss": 0.0606,
|
| 2432 |
+
"step": 2800
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 0.52,
|
| 2436 |
+
"eval_accuracy": 0.9828711537797423,
|
| 2437 |
+
"eval_f1": 0.9828776581054958,
|
| 2438 |
+
"eval_loss": 0.052921637892723083,
|
| 2439 |
+
"eval_precision": 0.9859067142214065,
|
| 2440 |
+
"eval_recall": 0.9798671576556424,
|
| 2441 |
+
"eval_runtime": 324.5523,
|
| 2442 |
+
"eval_samples_per_second": 178.443,
|
| 2443 |
+
"eval_steps_per_second": 22.308,
|
| 2444 |
+
"step": 2800
|
| 2445 |
+
},
|
| 2446 |
+
{
|
| 2447 |
+
"epoch": 0.52,
|
| 2448 |
+
"learning_rate": 2.6607831203083787e-05,
|
| 2449 |
+
"loss": 0.0665,
|
| 2450 |
+
"step": 2808
|
| 2451 |
+
},
|
| 2452 |
+
{
|
| 2453 |
+
"epoch": 0.52,
|
| 2454 |
+
"learning_rate": 2.6526678839521203e-05,
|
| 2455 |
+
"loss": 0.0533,
|
| 2456 |
+
"step": 2816
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 0.52,
|
| 2460 |
+
"learning_rate": 2.6445526475958615e-05,
|
| 2461 |
+
"loss": 0.0706,
|
| 2462 |
+
"step": 2824
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 0.52,
|
| 2466 |
+
"learning_rate": 2.6364374112396024e-05,
|
| 2467 |
+
"loss": 0.0729,
|
| 2468 |
+
"step": 2832
|
| 2469 |
+
},
|
| 2470 |
+
{
|
| 2471 |
+
"epoch": 0.52,
|
| 2472 |
+
"learning_rate": 2.6283221748833436e-05,
|
| 2473 |
+
"loss": 0.0498,
|
| 2474 |
+
"step": 2840
|
| 2475 |
+
},
|
| 2476 |
+
{
|
| 2477 |
+
"epoch": 0.52,
|
| 2478 |
+
"learning_rate": 2.620206938527085e-05,
|
| 2479 |
+
"loss": 0.0438,
|
| 2480 |
+
"step": 2848
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 0.53,
|
| 2484 |
+
"learning_rate": 2.6120917021708257e-05,
|
| 2485 |
+
"loss": 0.0599,
|
| 2486 |
+
"step": 2856
|
| 2487 |
+
},
|
| 2488 |
+
{
|
| 2489 |
+
"epoch": 0.53,
|
| 2490 |
+
"learning_rate": 2.603976465814567e-05,
|
| 2491 |
+
"loss": 0.0413,
|
| 2492 |
+
"step": 2864
|
| 2493 |
+
},
|
| 2494 |
+
{
|
| 2495 |
+
"epoch": 0.53,
|
| 2496 |
+
"learning_rate": 2.595861229458308e-05,
|
| 2497 |
+
"loss": 0.0323,
|
| 2498 |
+
"step": 2872
|
| 2499 |
+
},
|
| 2500 |
+
{
|
| 2501 |
+
"epoch": 0.53,
|
| 2502 |
+
"learning_rate": 2.587745993102049e-05,
|
| 2503 |
+
"loss": 0.0706,
|
| 2504 |
+
"step": 2880
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 0.53,
|
| 2508 |
+
"learning_rate": 2.5796307567457907e-05,
|
| 2509 |
+
"loss": 0.116,
|
| 2510 |
+
"step": 2888
|
| 2511 |
+
},
|
| 2512 |
+
{
|
| 2513 |
+
"epoch": 0.53,
|
| 2514 |
+
"learning_rate": 2.5715155203895312e-05,
|
| 2515 |
+
"loss": 0.0585,
|
| 2516 |
+
"step": 2896
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 0.53,
|
| 2520 |
+
"eval_accuracy": 0.9815415961598232,
|
| 2521 |
+
"eval_f1": 0.9814451599465398,
|
| 2522 |
+
"eval_loss": 0.0516064278781414,
|
| 2523 |
+
"eval_precision": 0.99005462949993,
|
| 2524 |
+
"eval_recall": 0.9729841346319303,
|
| 2525 |
+
"eval_runtime": 325.834,
|
| 2526 |
+
"eval_samples_per_second": 177.741,
|
| 2527 |
+
"eval_steps_per_second": 22.22,
|
| 2528 |
+
"step": 2900
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 0.53,
|
| 2532 |
+
"learning_rate": 2.5634002840332728e-05,
|
| 2533 |
+
"loss": 0.0475,
|
| 2534 |
+
"step": 2904
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 0.54,
|
| 2538 |
+
"learning_rate": 2.5552850476770133e-05,
|
| 2539 |
+
"loss": 0.0612,
|
| 2540 |
+
"step": 2912
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 0.54,
|
| 2544 |
+
"learning_rate": 2.547169811320755e-05,
|
| 2545 |
+
"loss": 0.0558,
|
| 2546 |
+
"step": 2920
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 0.54,
|
| 2550 |
+
"learning_rate": 2.539054574964496e-05,
|
| 2551 |
+
"loss": 0.0572,
|
| 2552 |
+
"step": 2928
|
| 2553 |
+
},
|
| 2554 |
+
{
|
| 2555 |
+
"epoch": 0.54,
|
| 2556 |
+
"learning_rate": 2.530939338608237e-05,
|
| 2557 |
+
"loss": 0.0581,
|
| 2558 |
+
"step": 2936
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 0.54,
|
| 2562 |
+
"learning_rate": 2.5228241022519783e-05,
|
| 2563 |
+
"loss": 0.0487,
|
| 2564 |
+
"step": 2944
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.54,
|
| 2568 |
+
"learning_rate": 2.514708865895719e-05,
|
| 2569 |
+
"loss": 0.0681,
|
| 2570 |
+
"step": 2952
|
| 2571 |
+
},
|
| 2572 |
+
{
|
| 2573 |
+
"epoch": 0.55,
|
| 2574 |
+
"learning_rate": 2.5065936295394604e-05,
|
| 2575 |
+
"loss": 0.0658,
|
| 2576 |
+
"step": 2960
|
| 2577 |
+
},
|
| 2578 |
+
{
|
| 2579 |
+
"epoch": 0.55,
|
| 2580 |
+
"learning_rate": 2.4984783931832016e-05,
|
| 2581 |
+
"loss": 0.0769,
|
| 2582 |
+
"step": 2968
|
| 2583 |
+
},
|
| 2584 |
+
{
|
| 2585 |
+
"epoch": 0.55,
|
| 2586 |
+
"learning_rate": 2.490363156826943e-05,
|
| 2587 |
+
"loss": 0.0736,
|
| 2588 |
+
"step": 2976
|
| 2589 |
+
},
|
| 2590 |
+
{
|
| 2591 |
+
"epoch": 0.55,
|
| 2592 |
+
"learning_rate": 2.4822479204706837e-05,
|
| 2593 |
+
"loss": 0.0446,
|
| 2594 |
+
"step": 2984
|
| 2595 |
+
},
|
| 2596 |
+
{
|
| 2597 |
+
"epoch": 0.55,
|
| 2598 |
+
"learning_rate": 2.474132684114425e-05,
|
| 2599 |
+
"loss": 0.0472,
|
| 2600 |
+
"step": 2992
|
| 2601 |
+
},
|
| 2602 |
+
{
|
| 2603 |
+
"epoch": 0.55,
|
| 2604 |
+
"learning_rate": 2.466017447758166e-05,
|
| 2605 |
+
"loss": 0.058,
|
| 2606 |
+
"step": 3000
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 0.55,
|
| 2610 |
+
"eval_accuracy": 0.9817142659805919,
|
| 2611 |
+
"eval_f1": 0.981694352733747,
|
| 2612 |
+
"eval_loss": 0.061614930629730225,
|
| 2613 |
+
"eval_precision": 0.9861776759047024,
|
| 2614 |
+
"eval_recall": 0.9772516089066318,
|
| 2615 |
+
"eval_runtime": 317.9761,
|
| 2616 |
+
"eval_samples_per_second": 182.133,
|
| 2617 |
+
"eval_steps_per_second": 22.769,
|
| 2618 |
+
"step": 3000
|
| 2619 |
+
},
|
| 2620 |
+
{
|
| 2621 |
+
"epoch": 0.55,
|
| 2622 |
+
"learning_rate": 2.457902211401907e-05,
|
| 2623 |
+
"loss": 0.0432,
|
| 2624 |
+
"step": 3008
|
| 2625 |
+
},
|
| 2626 |
+
{
|
| 2627 |
+
"epoch": 0.56,
|
| 2628 |
+
"learning_rate": 2.4497869750456483e-05,
|
| 2629 |
+
"loss": 0.0392,
|
| 2630 |
+
"step": 3016
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 0.56,
|
| 2634 |
+
"learning_rate": 2.4416717386893896e-05,
|
| 2635 |
+
"loss": 0.0581,
|
| 2636 |
+
"step": 3024
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 0.56,
|
| 2640 |
+
"learning_rate": 2.4335565023331305e-05,
|
| 2641 |
+
"loss": 0.0642,
|
| 2642 |
+
"step": 3032
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 0.56,
|
| 2646 |
+
"learning_rate": 2.4254412659768717e-05,
|
| 2647 |
+
"loss": 0.039,
|
| 2648 |
+
"step": 3040
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 0.56,
|
| 2652 |
+
"learning_rate": 2.417326029620613e-05,
|
| 2653 |
+
"loss": 0.0361,
|
| 2654 |
+
"step": 3048
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 0.56,
|
| 2658 |
+
"learning_rate": 2.409210793264354e-05,
|
| 2659 |
+
"loss": 0.0591,
|
| 2660 |
+
"step": 3056
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 0.56,
|
| 2664 |
+
"learning_rate": 2.401095556908095e-05,
|
| 2665 |
+
"loss": 0.0819,
|
| 2666 |
+
"step": 3064
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 0.57,
|
| 2670 |
+
"learning_rate": 2.3929803205518363e-05,
|
| 2671 |
+
"loss": 0.1084,
|
| 2672 |
+
"step": 3072
|
| 2673 |
+
},
|
| 2674 |
+
{
|
| 2675 |
+
"epoch": 0.57,
|
| 2676 |
+
"learning_rate": 2.384865084195577e-05,
|
| 2677 |
+
"loss": 0.0425,
|
| 2678 |
+
"step": 3080
|
| 2679 |
+
},
|
| 2680 |
+
{
|
| 2681 |
+
"epoch": 0.57,
|
| 2682 |
+
"learning_rate": 2.3767498478393184e-05,
|
| 2683 |
+
"loss": 0.0711,
|
| 2684 |
+
"step": 3088
|
| 2685 |
+
},
|
| 2686 |
+
{
|
| 2687 |
+
"epoch": 0.57,
|
| 2688 |
+
"learning_rate": 2.3686346114830596e-05,
|
| 2689 |
+
"loss": 0.0449,
|
| 2690 |
+
"step": 3096
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 0.57,
|
| 2694 |
+
"eval_accuracy": 0.9834064302241254,
|
| 2695 |
+
"eval_f1": 0.9835052608091176,
|
| 2696 |
+
"eval_loss": 0.04805440455675125,
|
| 2697 |
+
"eval_precision": 0.9810299958909738,
|
| 2698 |
+
"eval_recall": 0.985993048146746,
|
| 2699 |
+
"eval_runtime": 326.0918,
|
| 2700 |
+
"eval_samples_per_second": 177.6,
|
| 2701 |
+
"eval_steps_per_second": 22.202,
|
| 2702 |
+
"step": 3100
|
| 2703 |
+
},
|
| 2704 |
+
{
|
| 2705 |
+
"epoch": 0.57,
|
| 2706 |
+
"learning_rate": 2.3605193751268005e-05,
|
| 2707 |
+
"loss": 0.0558,
|
| 2708 |
+
"step": 3104
|
| 2709 |
+
},
|
| 2710 |
+
{
|
| 2711 |
+
"epoch": 0.57,
|
| 2712 |
+
"learning_rate": 2.3524041387705417e-05,
|
| 2713 |
+
"loss": 0.0629,
|
| 2714 |
+
"step": 3112
|
| 2715 |
+
},
|
| 2716 |
+
{
|
| 2717 |
+
"epoch": 0.57,
|
| 2718 |
+
"learning_rate": 2.344288902414283e-05,
|
| 2719 |
+
"loss": 0.0392,
|
| 2720 |
+
"step": 3120
|
| 2721 |
+
},
|
| 2722 |
+
{
|
| 2723 |
+
"epoch": 0.58,
|
| 2724 |
+
"learning_rate": 2.3361736660580242e-05,
|
| 2725 |
+
"loss": 0.0564,
|
| 2726 |
+
"step": 3128
|
| 2727 |
+
},
|
| 2728 |
+
{
|
| 2729 |
+
"epoch": 0.58,
|
| 2730 |
+
"learning_rate": 2.328058429701765e-05,
|
| 2731 |
+
"loss": 0.0418,
|
| 2732 |
+
"step": 3136
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 0.58,
|
| 2736 |
+
"learning_rate": 2.3199431933455063e-05,
|
| 2737 |
+
"loss": 0.059,
|
| 2738 |
+
"step": 3144
|
| 2739 |
+
},
|
| 2740 |
+
{
|
| 2741 |
+
"epoch": 0.58,
|
| 2742 |
+
"learning_rate": 2.3118279569892472e-05,
|
| 2743 |
+
"loss": 0.0511,
|
| 2744 |
+
"step": 3152
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 0.58,
|
| 2748 |
+
"learning_rate": 2.3037127206329888e-05,
|
| 2749 |
+
"loss": 0.0383,
|
| 2750 |
+
"step": 3160
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 0.58,
|
| 2754 |
+
"learning_rate": 2.2955974842767297e-05,
|
| 2755 |
+
"loss": 0.0831,
|
| 2756 |
+
"step": 3168
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 0.58,
|
| 2760 |
+
"learning_rate": 2.287482247920471e-05,
|
| 2761 |
+
"loss": 0.0407,
|
| 2762 |
+
"step": 3176
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 0.59,
|
| 2766 |
+
"learning_rate": 2.2793670115642118e-05,
|
| 2767 |
+
"loss": 0.0354,
|
| 2768 |
+
"step": 3184
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.59,
|
| 2772 |
+
"learning_rate": 2.271251775207953e-05,
|
| 2773 |
+
"loss": 0.0524,
|
| 2774 |
+
"step": 3192
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 0.59,
|
| 2778 |
+
"learning_rate": 2.2631365388516943e-05,
|
| 2779 |
+
"loss": 0.0257,
|
| 2780 |
+
"step": 3200
|
| 2781 |
+
},
|
| 2782 |
+
{
|
| 2783 |
+
"epoch": 0.59,
|
| 2784 |
+
"eval_accuracy": 0.9817315329626688,
|
| 2785 |
+
"eval_f1": 0.9816478751084129,
|
| 2786 |
+
"eval_loss": 0.06091897934675217,
|
| 2787 |
+
"eval_precision": 0.9896128423040604,
|
| 2788 |
+
"eval_recall": 0.9738100973947758,
|
| 2789 |
+
"eval_runtime": 313.118,
|
| 2790 |
+
"eval_samples_per_second": 184.959,
|
| 2791 |
+
"eval_steps_per_second": 23.122,
|
| 2792 |
+
"step": 3200
|
| 2793 |
+
},
|
| 2794 |
+
{
|
| 2795 |
+
"epoch": 0.59,
|
| 2796 |
+
"learning_rate": 2.255021302495435e-05,
|
| 2797 |
+
"loss": 0.0473,
|
| 2798 |
+
"step": 3208
|
| 2799 |
+
},
|
| 2800 |
+
{
|
| 2801 |
+
"epoch": 0.59,
|
| 2802 |
+
"learning_rate": 2.2469060661391764e-05,
|
| 2803 |
+
"loss": 0.0779,
|
| 2804 |
+
"step": 3216
|
| 2805 |
+
},
|
| 2806 |
+
{
|
| 2807 |
+
"epoch": 0.59,
|
| 2808 |
+
"learning_rate": 2.2387908297829173e-05,
|
| 2809 |
+
"loss": 0.0436,
|
| 2810 |
+
"step": 3224
|
| 2811 |
+
},
|
| 2812 |
+
{
|
| 2813 |
+
"epoch": 0.6,
|
| 2814 |
+
"learning_rate": 2.230675593426659e-05,
|
| 2815 |
+
"loss": 0.0541,
|
| 2816 |
+
"step": 3232
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 0.6,
|
| 2820 |
+
"learning_rate": 2.2225603570703997e-05,
|
| 2821 |
+
"loss": 0.0576,
|
| 2822 |
+
"step": 3240
|
| 2823 |
+
},
|
| 2824 |
+
{
|
| 2825 |
+
"epoch": 0.6,
|
| 2826 |
+
"learning_rate": 2.214445120714141e-05,
|
| 2827 |
+
"loss": 0.0649,
|
| 2828 |
+
"step": 3248
|
| 2829 |
+
},
|
| 2830 |
+
{
|
| 2831 |
+
"epoch": 0.6,
|
| 2832 |
+
"learning_rate": 2.206329884357882e-05,
|
| 2833 |
+
"loss": 0.0404,
|
| 2834 |
+
"step": 3256
|
| 2835 |
+
},
|
| 2836 |
+
{
|
| 2837 |
+
"epoch": 0.6,
|
| 2838 |
+
"learning_rate": 2.198214648001623e-05,
|
| 2839 |
+
"loss": 0.0351,
|
| 2840 |
+
"step": 3264
|
| 2841 |
+
},
|
| 2842 |
+
{
|
| 2843 |
+
"epoch": 0.6,
|
| 2844 |
+
"learning_rate": 2.1900994116453643e-05,
|
| 2845 |
+
"loss": 0.0364,
|
| 2846 |
+
"step": 3272
|
| 2847 |
+
},
|
| 2848 |
+
{
|
| 2849 |
+
"epoch": 0.6,
|
| 2850 |
+
"learning_rate": 2.1819841752891055e-05,
|
| 2851 |
+
"loss": 0.0483,
|
| 2852 |
+
"step": 3280
|
| 2853 |
+
},
|
| 2854 |
+
{
|
| 2855 |
+
"epoch": 0.61,
|
| 2856 |
+
"learning_rate": 2.1738689389328464e-05,
|
| 2857 |
+
"loss": 0.0562,
|
| 2858 |
+
"step": 3288
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 0.61,
|
| 2862 |
+
"learning_rate": 2.1657537025765877e-05,
|
| 2863 |
+
"loss": 0.0627,
|
| 2864 |
+
"step": 3296
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 0.61,
|
| 2868 |
+
"eval_accuracy": 0.9826121490485893,
|
| 2869 |
+
"eval_f1": 0.9827441438045136,
|
| 2870 |
+
"eval_loss": 0.04961591958999634,
|
| 2871 |
+
"eval_precision": 0.9786689419795221,
|
| 2872 |
+
"eval_recall": 0.98685342602471,
|
| 2873 |
+
"eval_runtime": 320.6595,
|
| 2874 |
+
"eval_samples_per_second": 180.609,
|
| 2875 |
+
"eval_steps_per_second": 22.578,
|
| 2876 |
+
"step": 3300
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 0.61,
|
| 2880 |
+
"learning_rate": 2.157638466220329e-05,
|
| 2881 |
+
"loss": 0.0339,
|
| 2882 |
+
"step": 3304
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 0.61,
|
| 2886 |
+
"learning_rate": 2.1495232298640698e-05,
|
| 2887 |
+
"loss": 0.0618,
|
| 2888 |
+
"step": 3312
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 0.61,
|
| 2892 |
+
"learning_rate": 2.141407993507811e-05,
|
| 2893 |
+
"loss": 0.074,
|
| 2894 |
+
"step": 3320
|
| 2895 |
+
},
|
| 2896 |
+
{
|
| 2897 |
+
"epoch": 0.61,
|
| 2898 |
+
"learning_rate": 2.133292757151552e-05,
|
| 2899 |
+
"loss": 0.0781,
|
| 2900 |
+
"step": 3328
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 0.61,
|
| 2904 |
+
"learning_rate": 2.125177520795293e-05,
|
| 2905 |
+
"loss": 0.059,
|
| 2906 |
+
"step": 3336
|
| 2907 |
+
},
|
| 2908 |
+
{
|
| 2909 |
+
"epoch": 0.62,
|
| 2910 |
+
"learning_rate": 2.1170622844390344e-05,
|
| 2911 |
+
"loss": 0.0411,
|
| 2912 |
+
"step": 3344
|
| 2913 |
+
},
|
| 2914 |
+
{
|
| 2915 |
+
"epoch": 0.62,
|
| 2916 |
+
"learning_rate": 2.1089470480827756e-05,
|
| 2917 |
+
"loss": 0.0604,
|
| 2918 |
+
"step": 3352
|
| 2919 |
+
},
|
| 2920 |
+
{
|
| 2921 |
+
"epoch": 0.62,
|
| 2922 |
+
"learning_rate": 2.1008318117265165e-05,
|
| 2923 |
+
"loss": 0.0602,
|
| 2924 |
+
"step": 3360
|
| 2925 |
+
},
|
| 2926 |
+
{
|
| 2927 |
+
"epoch": 0.62,
|
| 2928 |
+
"learning_rate": 2.0927165753702577e-05,
|
| 2929 |
+
"loss": 0.035,
|
| 2930 |
+
"step": 3368
|
| 2931 |
+
},
|
| 2932 |
+
{
|
| 2933 |
+
"epoch": 0.62,
|
| 2934 |
+
"learning_rate": 2.084601339013999e-05,
|
| 2935 |
+
"loss": 0.0545,
|
| 2936 |
+
"step": 3376
|
| 2937 |
+
},
|
| 2938 |
+
{
|
| 2939 |
+
"epoch": 0.62,
|
| 2940 |
+
"learning_rate": 2.0764861026577402e-05,
|
| 2941 |
+
"loss": 0.05,
|
| 2942 |
+
"step": 3384
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 0.62,
|
| 2946 |
+
"learning_rate": 2.068370866301481e-05,
|
| 2947 |
+
"loss": 0.0676,
|
| 2948 |
+
"step": 3392
|
| 2949 |
+
},
|
| 2950 |
+
{
|
| 2951 |
+
"epoch": 0.63,
|
| 2952 |
+
"learning_rate": 2.0602556299452223e-05,
|
| 2953 |
+
"loss": 0.0501,
|
| 2954 |
+
"step": 3400
|
| 2955 |
+
},
|
| 2956 |
+
{
|
| 2957 |
+
"epoch": 0.63,
|
| 2958 |
+
"eval_accuracy": 0.9837863038298166,
|
| 2959 |
+
"eval_f1": 0.9838000103514312,
|
| 2960 |
+
"eval_loss": 0.04516952857375145,
|
| 2961 |
+
"eval_precision": 0.9863696118452916,
|
| 2962 |
+
"eval_recall": 0.9812437622603848,
|
| 2963 |
+
"eval_runtime": 311.4407,
|
| 2964 |
+
"eval_samples_per_second": 185.955,
|
| 2965 |
+
"eval_steps_per_second": 23.247,
|
| 2966 |
+
"step": 3400
|
| 2967 |
+
}
|
| 2968 |
+
],
|
| 2969 |
+
"max_steps": 5429,
|
| 2970 |
+
"num_train_epochs": 1,
|
| 2971 |
+
"total_flos": 5.7252965646336e+16,
|
| 2972 |
+
"trial_name": null,
|
| 2973 |
+
"trial_params": null
|
| 2974 |
+
}
|