Commit
•
12dee61
1
Parent(s):
e2decb2
Upload trainer_state.json
Browse files- trainer_state.json +3495 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,3495 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 1.0465439558029175,
|
| 3 |
+
"best_model_checkpoint": "output/bertlawbr/checkpoint-227500",
|
| 4 |
+
"epoch": 20.0,
|
| 5 |
+
"global_step": 228880,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.04,
|
| 12 |
+
"learning_rate": 5e-06,
|
| 13 |
+
"loss": 9.0906,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.09,
|
| 18 |
+
"learning_rate": 1e-05,
|
| 19 |
+
"loss": 7.3178,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.13,
|
| 24 |
+
"learning_rate": 1.5e-05,
|
| 25 |
+
"loss": 6.5353,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.17,
|
| 30 |
+
"learning_rate": 2e-05,
|
| 31 |
+
"loss": 6.3111,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.22,
|
| 36 |
+
"learning_rate": 2.5e-05,
|
| 37 |
+
"loss": 6.1291,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.22,
|
| 42 |
+
"eval_loss": 5.988766670227051,
|
| 43 |
+
"eval_runtime": 683.0879,
|
| 44 |
+
"eval_samples_per_second": 112.864,
|
| 45 |
+
"eval_steps_per_second": 7.055,
|
| 46 |
+
"step": 2500
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.26,
|
| 50 |
+
"learning_rate": 3e-05,
|
| 51 |
+
"loss": 5.9734,
|
| 52 |
+
"step": 3000
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.31,
|
| 56 |
+
"learning_rate": 3.5e-05,
|
| 57 |
+
"loss": 5.797,
|
| 58 |
+
"step": 3500
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.35,
|
| 62 |
+
"learning_rate": 4e-05,
|
| 63 |
+
"loss": 5.5955,
|
| 64 |
+
"step": 4000
|
| 65 |
+
},
|
| 66 |
+
{
|
| 67 |
+
"epoch": 0.39,
|
| 68 |
+
"learning_rate": 4.5e-05,
|
| 69 |
+
"loss": 5.2733,
|
| 70 |
+
"step": 4500
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.44,
|
| 74 |
+
"learning_rate": 5e-05,
|
| 75 |
+
"loss": 4.8604,
|
| 76 |
+
"step": 5000
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.44,
|
| 80 |
+
"eval_loss": 4.484074115753174,
|
| 81 |
+
"eval_runtime": 681.6714,
|
| 82 |
+
"eval_samples_per_second": 113.098,
|
| 83 |
+
"eval_steps_per_second": 7.069,
|
| 84 |
+
"step": 5000
|
| 85 |
+
},
|
| 86 |
+
{
|
| 87 |
+
"epoch": 0.48,
|
| 88 |
+
"learning_rate": 5.500000000000001e-05,
|
| 89 |
+
"loss": 4.4273,
|
| 90 |
+
"step": 5500
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.52,
|
| 94 |
+
"learning_rate": 6e-05,
|
| 95 |
+
"loss": 4.049,
|
| 96 |
+
"step": 6000
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.57,
|
| 100 |
+
"learning_rate": 6.500000000000001e-05,
|
| 101 |
+
"loss": 3.7574,
|
| 102 |
+
"step": 6500
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.61,
|
| 106 |
+
"learning_rate": 7e-05,
|
| 107 |
+
"loss": 3.5134,
|
| 108 |
+
"step": 7000
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.66,
|
| 112 |
+
"learning_rate": 7.500000000000001e-05,
|
| 113 |
+
"loss": 3.3321,
|
| 114 |
+
"step": 7500
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.66,
|
| 118 |
+
"eval_loss": 3.118971586227417,
|
| 119 |
+
"eval_runtime": 682.3171,
|
| 120 |
+
"eval_samples_per_second": 112.991,
|
| 121 |
+
"eval_steps_per_second": 7.063,
|
| 122 |
+
"step": 7500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.7,
|
| 126 |
+
"learning_rate": 8e-05,
|
| 127 |
+
"loss": 3.1651,
|
| 128 |
+
"step": 8000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.74,
|
| 132 |
+
"learning_rate": 8.5e-05,
|
| 133 |
+
"loss": 3.0254,
|
| 134 |
+
"step": 8500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.79,
|
| 138 |
+
"learning_rate": 8.999000000000001e-05,
|
| 139 |
+
"loss": 2.9285,
|
| 140 |
+
"step": 9000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.83,
|
| 144 |
+
"learning_rate": 9.499e-05,
|
| 145 |
+
"loss": 2.8345,
|
| 146 |
+
"step": 9500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.87,
|
| 150 |
+
"learning_rate": 9.999000000000001e-05,
|
| 151 |
+
"loss": 2.7579,
|
| 152 |
+
"step": 10000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.87,
|
| 156 |
+
"eval_loss": 2.608943223953247,
|
| 157 |
+
"eval_runtime": 683.6111,
|
| 158 |
+
"eval_samples_per_second": 112.778,
|
| 159 |
+
"eval_steps_per_second": 7.049,
|
| 160 |
+
"step": 10000
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"epoch": 0.92,
|
| 164 |
+
"learning_rate": 9.89432443879712e-05,
|
| 165 |
+
"loss": 2.6686,
|
| 166 |
+
"step": 10500
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.96,
|
| 170 |
+
"learning_rate": 9.788437102922491e-05,
|
| 171 |
+
"loss": 2.597,
|
| 172 |
+
"step": 11000
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 1.0,
|
| 176 |
+
"learning_rate": 9.682549767047862e-05,
|
| 177 |
+
"loss": 2.5214,
|
| 178 |
+
"step": 11500
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 1.05,
|
| 182 |
+
"learning_rate": 9.576662431173232e-05,
|
| 183 |
+
"loss": 2.4561,
|
| 184 |
+
"step": 12000
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 1.09,
|
| 188 |
+
"learning_rate": 9.470775095298603e-05,
|
| 189 |
+
"loss": 2.4135,
|
| 190 |
+
"step": 12500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 1.09,
|
| 194 |
+
"eval_loss": 2.3029026985168457,
|
| 195 |
+
"eval_runtime": 682.8117,
|
| 196 |
+
"eval_samples_per_second": 112.91,
|
| 197 |
+
"eval_steps_per_second": 7.058,
|
| 198 |
+
"step": 12500
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 1.14,
|
| 202 |
+
"learning_rate": 9.365099534095722e-05,
|
| 203 |
+
"loss": 2.373,
|
| 204 |
+
"step": 13000
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 1.18,
|
| 208 |
+
"learning_rate": 9.259212198221094e-05,
|
| 209 |
+
"loss": 2.3231,
|
| 210 |
+
"step": 13500
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 1.22,
|
| 214 |
+
"learning_rate": 9.153324862346464e-05,
|
| 215 |
+
"loss": 2.2866,
|
| 216 |
+
"step": 14000
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 1.27,
|
| 220 |
+
"learning_rate": 9.047437526471835e-05,
|
| 221 |
+
"loss": 2.2534,
|
| 222 |
+
"step": 14500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 1.31,
|
| 226 |
+
"learning_rate": 8.941550190597205e-05,
|
| 227 |
+
"loss": 2.2136,
|
| 228 |
+
"step": 15000
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 1.31,
|
| 232 |
+
"eval_loss": 2.124361753463745,
|
| 233 |
+
"eval_runtime": 681.8811,
|
| 234 |
+
"eval_samples_per_second": 113.064,
|
| 235 |
+
"eval_steps_per_second": 7.067,
|
| 236 |
+
"step": 15000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.35,
|
| 240 |
+
"learning_rate": 8.835874629394325e-05,
|
| 241 |
+
"loss": 2.1843,
|
| 242 |
+
"step": 15500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.4,
|
| 246 |
+
"learning_rate": 8.729987293519695e-05,
|
| 247 |
+
"loss": 2.1551,
|
| 248 |
+
"step": 16000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.44,
|
| 252 |
+
"learning_rate": 8.624099957645066e-05,
|
| 253 |
+
"loss": 2.1253,
|
| 254 |
+
"step": 16500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.49,
|
| 258 |
+
"learning_rate": 8.518212621770438e-05,
|
| 259 |
+
"loss": 2.1124,
|
| 260 |
+
"step": 17000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 1.53,
|
| 264 |
+
"learning_rate": 8.412537060567557e-05,
|
| 265 |
+
"loss": 2.0735,
|
| 266 |
+
"step": 17500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.53,
|
| 270 |
+
"eval_loss": 1.9931037425994873,
|
| 271 |
+
"eval_runtime": 682.7879,
|
| 272 |
+
"eval_samples_per_second": 112.914,
|
| 273 |
+
"eval_steps_per_second": 7.058,
|
| 274 |
+
"step": 17500
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 1.57,
|
| 278 |
+
"learning_rate": 8.306649724692928e-05,
|
| 279 |
+
"loss": 2.0535,
|
| 280 |
+
"step": 18000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 1.62,
|
| 284 |
+
"learning_rate": 8.200762388818297e-05,
|
| 285 |
+
"loss": 2.0385,
|
| 286 |
+
"step": 18500
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 1.66,
|
| 290 |
+
"learning_rate": 8.095086827615418e-05,
|
| 291 |
+
"loss": 2.0165,
|
| 292 |
+
"step": 19000
|
| 293 |
+
},
|
| 294 |
+
{
|
| 295 |
+
"epoch": 1.7,
|
| 296 |
+
"learning_rate": 7.989199491740788e-05,
|
| 297 |
+
"loss": 1.9965,
|
| 298 |
+
"step": 19500
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 1.75,
|
| 302 |
+
"learning_rate": 7.883312155866158e-05,
|
| 303 |
+
"loss": 1.9684,
|
| 304 |
+
"step": 20000
|
| 305 |
+
},
|
| 306 |
+
{
|
| 307 |
+
"epoch": 1.75,
|
| 308 |
+
"eval_loss": 1.887757420539856,
|
| 309 |
+
"eval_runtime": 683.0468,
|
| 310 |
+
"eval_samples_per_second": 112.871,
|
| 311 |
+
"eval_steps_per_second": 7.055,
|
| 312 |
+
"step": 20000
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 1.79,
|
| 316 |
+
"learning_rate": 7.77742481999153e-05,
|
| 317 |
+
"loss": 1.9452,
|
| 318 |
+
"step": 20500
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 1.84,
|
| 322 |
+
"learning_rate": 7.6715374841169e-05,
|
| 323 |
+
"loss": 1.9475,
|
| 324 |
+
"step": 21000
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 1.88,
|
| 328 |
+
"learning_rate": 7.565650148242271e-05,
|
| 329 |
+
"loss": 1.9304,
|
| 330 |
+
"step": 21500
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 1.92,
|
| 334 |
+
"learning_rate": 7.459762812367641e-05,
|
| 335 |
+
"loss": 1.9201,
|
| 336 |
+
"step": 22000
|
| 337 |
+
},
|
| 338 |
+
{
|
| 339 |
+
"epoch": 1.97,
|
| 340 |
+
"learning_rate": 7.353875476493012e-05,
|
| 341 |
+
"loss": 1.891,
|
| 342 |
+
"step": 22500
|
| 343 |
+
},
|
| 344 |
+
{
|
| 345 |
+
"epoch": 1.97,
|
| 346 |
+
"eval_loss": 1.8076796531677246,
|
| 347 |
+
"eval_runtime": 683.1394,
|
| 348 |
+
"eval_samples_per_second": 112.855,
|
| 349 |
+
"eval_steps_per_second": 7.054,
|
| 350 |
+
"step": 22500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.01,
|
| 354 |
+
"learning_rate": 7.248199915290131e-05,
|
| 355 |
+
"loss": 1.8747,
|
| 356 |
+
"step": 23000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.05,
|
| 360 |
+
"learning_rate": 7.142312579415502e-05,
|
| 361 |
+
"loss": 1.8568,
|
| 362 |
+
"step": 23500
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.1,
|
| 366 |
+
"learning_rate": 7.036425243540872e-05,
|
| 367 |
+
"loss": 1.832,
|
| 368 |
+
"step": 24000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.14,
|
| 372 |
+
"learning_rate": 6.930749682337993e-05,
|
| 373 |
+
"loss": 1.8274,
|
| 374 |
+
"step": 24500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.18,
|
| 378 |
+
"learning_rate": 6.824862346463363e-05,
|
| 379 |
+
"loss": 1.8215,
|
| 380 |
+
"step": 25000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.18,
|
| 384 |
+
"eval_loss": 1.748734951019287,
|
| 385 |
+
"eval_runtime": 675.7524,
|
| 386 |
+
"eval_samples_per_second": 114.089,
|
| 387 |
+
"eval_steps_per_second": 7.131,
|
| 388 |
+
"step": 25000
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 2.23,
|
| 392 |
+
"learning_rate": 6.718975010588734e-05,
|
| 393 |
+
"loss": 1.8079,
|
| 394 |
+
"step": 25500
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 2.27,
|
| 398 |
+
"learning_rate": 6.613087674714105e-05,
|
| 399 |
+
"loss": 1.7946,
|
| 400 |
+
"step": 26000
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"epoch": 2.32,
|
| 404 |
+
"learning_rate": 6.507200338839475e-05,
|
| 405 |
+
"loss": 1.7862,
|
| 406 |
+
"step": 26500
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 2.36,
|
| 410 |
+
"learning_rate": 6.401313002964846e-05,
|
| 411 |
+
"loss": 1.7719,
|
| 412 |
+
"step": 27000
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 2.4,
|
| 416 |
+
"learning_rate": 6.295425667090216e-05,
|
| 417 |
+
"loss": 1.7577,
|
| 418 |
+
"step": 27500
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 2.4,
|
| 422 |
+
"eval_loss": 1.6874935626983643,
|
| 423 |
+
"eval_runtime": 682.5902,
|
| 424 |
+
"eval_samples_per_second": 112.946,
|
| 425 |
+
"eval_steps_per_second": 7.06,
|
| 426 |
+
"step": 27500
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 2.45,
|
| 430 |
+
"learning_rate": 6.189538331215588e-05,
|
| 431 |
+
"loss": 1.7455,
|
| 432 |
+
"step": 28000
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 2.49,
|
| 436 |
+
"learning_rate": 6.0838627700127074e-05,
|
| 437 |
+
"loss": 1.7409,
|
| 438 |
+
"step": 28500
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 2.53,
|
| 442 |
+
"learning_rate": 5.977975434138078e-05,
|
| 443 |
+
"loss": 1.7286,
|
| 444 |
+
"step": 29000
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 2.58,
|
| 448 |
+
"learning_rate": 5.8720880982634486e-05,
|
| 449 |
+
"loss": 1.7205,
|
| 450 |
+
"step": 29500
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 2.62,
|
| 454 |
+
"learning_rate": 5.766200762388818e-05,
|
| 455 |
+
"loss": 1.7113,
|
| 456 |
+
"step": 30000
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 2.62,
|
| 460 |
+
"eval_loss": 1.6443524360656738,
|
| 461 |
+
"eval_runtime": 681.9031,
|
| 462 |
+
"eval_samples_per_second": 113.06,
|
| 463 |
+
"eval_steps_per_second": 7.067,
|
| 464 |
+
"step": 30000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.67,
|
| 468 |
+
"learning_rate": 5.660525201185939e-05,
|
| 469 |
+
"loss": 1.7067,
|
| 470 |
+
"step": 30500
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.71,
|
| 474 |
+
"learning_rate": 5.5546378653113096e-05,
|
| 475 |
+
"loss": 1.6999,
|
| 476 |
+
"step": 31000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.75,
|
| 480 |
+
"learning_rate": 5.448750529436679e-05,
|
| 481 |
+
"loss": 1.6787,
|
| 482 |
+
"step": 31500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 2.8,
|
| 486 |
+
"learning_rate": 5.34286319356205e-05,
|
| 487 |
+
"loss": 1.6862,
|
| 488 |
+
"step": 32000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 2.84,
|
| 492 |
+
"learning_rate": 5.2371876323591706e-05,
|
| 493 |
+
"loss": 1.6776,
|
| 494 |
+
"step": 32500
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 2.84,
|
| 498 |
+
"eval_loss": 1.6035559177398682,
|
| 499 |
+
"eval_runtime": 677.3156,
|
| 500 |
+
"eval_samples_per_second": 113.826,
|
| 501 |
+
"eval_steps_per_second": 7.115,
|
| 502 |
+
"step": 32500
|
| 503 |
+
},
|
| 504 |
+
{
|
| 505 |
+
"epoch": 2.88,
|
| 506 |
+
"learning_rate": 5.1313002964845405e-05,
|
| 507 |
+
"loss": 1.6667,
|
| 508 |
+
"step": 33000
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 2.93,
|
| 512 |
+
"learning_rate": 5.025412960609911e-05,
|
| 513 |
+
"loss": 1.6636,
|
| 514 |
+
"step": 33500
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 2.97,
|
| 518 |
+
"learning_rate": 4.919737399407031e-05,
|
| 519 |
+
"loss": 1.6574,
|
| 520 |
+
"step": 34000
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 3.01,
|
| 524 |
+
"learning_rate": 4.813850063532402e-05,
|
| 525 |
+
"loss": 1.6315,
|
| 526 |
+
"step": 34500
|
| 527 |
+
},
|
| 528 |
+
{
|
| 529 |
+
"epoch": 3.06,
|
| 530 |
+
"learning_rate": 4.707962727657773e-05,
|
| 531 |
+
"loss": 1.6203,
|
| 532 |
+
"step": 35000
|
| 533 |
+
},
|
| 534 |
+
{
|
| 535 |
+
"epoch": 3.06,
|
| 536 |
+
"eval_loss": 1.5608155727386475,
|
| 537 |
+
"eval_runtime": 683.2194,
|
| 538 |
+
"eval_samples_per_second": 112.842,
|
| 539 |
+
"eval_steps_per_second": 7.053,
|
| 540 |
+
"step": 35000
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"epoch": 3.1,
|
| 544 |
+
"learning_rate": 4.602075391783143e-05,
|
| 545 |
+
"loss": 1.6021,
|
| 546 |
+
"step": 35500
|
| 547 |
+
},
|
| 548 |
+
{
|
| 549 |
+
"epoch": 3.15,
|
| 550 |
+
"learning_rate": 4.496188055908513e-05,
|
| 551 |
+
"loss": 1.609,
|
| 552 |
+
"step": 36000
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 3.19,
|
| 556 |
+
"learning_rate": 4.390300720033884e-05,
|
| 557 |
+
"loss": 1.6007,
|
| 558 |
+
"step": 36500
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 3.23,
|
| 562 |
+
"learning_rate": 4.284413384159255e-05,
|
| 563 |
+
"loss": 1.6025,
|
| 564 |
+
"step": 37000
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 3.28,
|
| 568 |
+
"learning_rate": 4.178526048284626e-05,
|
| 569 |
+
"loss": 1.6018,
|
| 570 |
+
"step": 37500
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"epoch": 3.28,
|
| 574 |
+
"eval_loss": 1.529254674911499,
|
| 575 |
+
"eval_runtime": 682.7102,
|
| 576 |
+
"eval_samples_per_second": 112.926,
|
| 577 |
+
"eval_steps_per_second": 7.059,
|
| 578 |
+
"step": 37500
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 3.32,
|
| 582 |
+
"learning_rate": 4.072638712409996e-05,
|
| 583 |
+
"loss": 1.5893,
|
| 584 |
+
"step": 38000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 3.36,
|
| 588 |
+
"learning_rate": 3.966963151207116e-05,
|
| 589 |
+
"loss": 1.5829,
|
| 590 |
+
"step": 38500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 3.41,
|
| 594 |
+
"learning_rate": 3.861075815332487e-05,
|
| 595 |
+
"loss": 1.5757,
|
| 596 |
+
"step": 39000
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 3.45,
|
| 600 |
+
"learning_rate": 3.755188479457857e-05,
|
| 601 |
+
"loss": 1.5741,
|
| 602 |
+
"step": 39500
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 3.5,
|
| 606 |
+
"learning_rate": 3.649301143583227e-05,
|
| 607 |
+
"loss": 1.5602,
|
| 608 |
+
"step": 40000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 3.5,
|
| 612 |
+
"eval_loss": 1.5043796300888062,
|
| 613 |
+
"eval_runtime": 682.2924,
|
| 614 |
+
"eval_samples_per_second": 112.996,
|
| 615 |
+
"eval_steps_per_second": 7.063,
|
| 616 |
+
"step": 40000
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 3.54,
|
| 620 |
+
"learning_rate": 3.543625582380348e-05,
|
| 621 |
+
"loss": 1.5609,
|
| 622 |
+
"step": 40500
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 3.58,
|
| 626 |
+
"learning_rate": 3.4377382465057176e-05,
|
| 627 |
+
"loss": 1.5567,
|
| 628 |
+
"step": 41000
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 3.63,
|
| 632 |
+
"learning_rate": 3.331850910631088e-05,
|
| 633 |
+
"loss": 1.5423,
|
| 634 |
+
"step": 41500
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 3.67,
|
| 638 |
+
"learning_rate": 3.2259635747564595e-05,
|
| 639 |
+
"loss": 1.5451,
|
| 640 |
+
"step": 42000
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 3.71,
|
| 644 |
+
"learning_rate": 3.1202880135535786e-05,
|
| 645 |
+
"loss": 1.5429,
|
| 646 |
+
"step": 42500
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 3.71,
|
| 650 |
+
"eval_loss": 1.475334644317627,
|
| 651 |
+
"eval_runtime": 682.6287,
|
| 652 |
+
"eval_samples_per_second": 112.94,
|
| 653 |
+
"eval_steps_per_second": 7.059,
|
| 654 |
+
"step": 42500
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 3.76,
|
| 658 |
+
"learning_rate": 3.0144006776789495e-05,
|
| 659 |
+
"loss": 1.5331,
|
| 660 |
+
"step": 43000
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 3.8,
|
| 664 |
+
"learning_rate": 2.9085133418043205e-05,
|
| 665 |
+
"loss": 1.5283,
|
| 666 |
+
"step": 43500
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 3.84,
|
| 670 |
+
"learning_rate": 2.802626005929691e-05,
|
| 671 |
+
"loss": 1.528,
|
| 672 |
+
"step": 44000
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 3.89,
|
| 676 |
+
"learning_rate": 2.696950444726811e-05,
|
| 677 |
+
"loss": 1.5216,
|
| 678 |
+
"step": 44500
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 3.93,
|
| 682 |
+
"learning_rate": 2.5910631088521814e-05,
|
| 683 |
+
"loss": 1.5148,
|
| 684 |
+
"step": 45000
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 3.93,
|
| 688 |
+
"eval_loss": 1.4471635818481445,
|
| 689 |
+
"eval_runtime": 682.5313,
|
| 690 |
+
"eval_samples_per_second": 112.956,
|
| 691 |
+
"eval_steps_per_second": 7.06,
|
| 692 |
+
"step": 45000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 3.98,
|
| 696 |
+
"learning_rate": 2.485175772977552e-05,
|
| 697 |
+
"loss": 1.5166,
|
| 698 |
+
"step": 45500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 4.02,
|
| 702 |
+
"learning_rate": 2.3795002117746718e-05,
|
| 703 |
+
"loss": 1.49,
|
| 704 |
+
"step": 46000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 4.06,
|
| 708 |
+
"learning_rate": 2.2736128759000424e-05,
|
| 709 |
+
"loss": 1.4958,
|
| 710 |
+
"step": 46500
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 4.11,
|
| 714 |
+
"learning_rate": 2.167725540025413e-05,
|
| 715 |
+
"loss": 1.4896,
|
| 716 |
+
"step": 47000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 4.15,
|
| 720 |
+
"learning_rate": 2.0618382041507836e-05,
|
| 721 |
+
"loss": 1.4786,
|
| 722 |
+
"step": 47500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 4.15,
|
| 726 |
+
"eval_loss": 1.430184006690979,
|
| 727 |
+
"eval_runtime": 682.3288,
|
| 728 |
+
"eval_samples_per_second": 112.99,
|
| 729 |
+
"eval_steps_per_second": 7.063,
|
| 730 |
+
"step": 47500
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 4.19,
|
| 734 |
+
"learning_rate": 1.9559508682761542e-05,
|
| 735 |
+
"loss": 1.4665,
|
| 736 |
+
"step": 48000
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 4.24,
|
| 740 |
+
"learning_rate": 1.850063532401525e-05,
|
| 741 |
+
"loss": 1.4743,
|
| 742 |
+
"step": 48500
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 4.28,
|
| 746 |
+
"learning_rate": 1.7441761965268955e-05,
|
| 747 |
+
"loss": 1.4731,
|
| 748 |
+
"step": 49000
|
| 749 |
+
},
|
| 750 |
+
{
|
| 751 |
+
"epoch": 4.33,
|
| 752 |
+
"learning_rate": 1.638288860652266e-05,
|
| 753 |
+
"loss": 1.4792,
|
| 754 |
+
"step": 49500
|
| 755 |
+
},
|
| 756 |
+
{
|
| 757 |
+
"epoch": 4.37,
|
| 758 |
+
"learning_rate": 1.5326132994493858e-05,
|
| 759 |
+
"loss": 1.4653,
|
| 760 |
+
"step": 50000
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"epoch": 4.37,
|
| 764 |
+
"eval_loss": 1.4128450155258179,
|
| 765 |
+
"eval_runtime": 683.1234,
|
| 766 |
+
"eval_samples_per_second": 112.858,
|
| 767 |
+
"eval_steps_per_second": 7.054,
|
| 768 |
+
"step": 50000
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 4.41,
|
| 772 |
+
"learning_rate": 1.4267259635747566e-05,
|
| 773 |
+
"loss": 1.4658,
|
| 774 |
+
"step": 50500
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 4.46,
|
| 778 |
+
"learning_rate": 1.3208386277001272e-05,
|
| 779 |
+
"loss": 1.4619,
|
| 780 |
+
"step": 51000
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 4.5,
|
| 784 |
+
"learning_rate": 1.215163066497247e-05,
|
| 785 |
+
"loss": 1.456,
|
| 786 |
+
"step": 51500
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 4.54,
|
| 790 |
+
"learning_rate": 1.1092757306226175e-05,
|
| 791 |
+
"loss": 1.4599,
|
| 792 |
+
"step": 52000
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 4.59,
|
| 796 |
+
"learning_rate": 1.0033883947479882e-05,
|
| 797 |
+
"loss": 1.4496,
|
| 798 |
+
"step": 52500
|
| 799 |
+
},
|
| 800 |
+
{
|
| 801 |
+
"epoch": 4.59,
|
| 802 |
+
"eval_loss": 1.3991329669952393,
|
| 803 |
+
"eval_runtime": 682.7978,
|
| 804 |
+
"eval_samples_per_second": 112.912,
|
| 805 |
+
"eval_steps_per_second": 7.058,
|
| 806 |
+
"step": 52500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 4.63,
|
| 810 |
+
"learning_rate": 8.975010588733588e-06,
|
| 811 |
+
"loss": 1.4476,
|
| 812 |
+
"step": 53000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 4.67,
|
| 816 |
+
"learning_rate": 7.916137229987294e-06,
|
| 817 |
+
"loss": 1.4422,
|
| 818 |
+
"step": 53500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 4.72,
|
| 822 |
+
"learning_rate": 6.859381617958492e-06,
|
| 823 |
+
"loss": 1.4399,
|
| 824 |
+
"step": 54000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 4.76,
|
| 828 |
+
"learning_rate": 5.800508259212199e-06,
|
| 829 |
+
"loss": 1.4356,
|
| 830 |
+
"step": 54500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 4.81,
|
| 834 |
+
"learning_rate": 4.741634900465904e-06,
|
| 835 |
+
"loss": 1.4445,
|
| 836 |
+
"step": 55000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 4.81,
|
| 840 |
+
"eval_loss": 1.3943445682525635,
|
| 841 |
+
"eval_runtime": 679.9865,
|
| 842 |
+
"eval_samples_per_second": 113.379,
|
| 843 |
+
"eval_steps_per_second": 7.087,
|
| 844 |
+
"step": 55000
|
| 845 |
+
},
|
| 846 |
+
{
|
| 847 |
+
"epoch": 4.85,
|
| 848 |
+
"learning_rate": 5.645250861738798e-05,
|
| 849 |
+
"loss": 1.4914,
|
| 850 |
+
"step": 55500
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"epoch": 4.89,
|
| 854 |
+
"learning_rate": 5.5973764841057066e-05,
|
| 855 |
+
"loss": 1.5131,
|
| 856 |
+
"step": 56000
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 4.94,
|
| 860 |
+
"learning_rate": 5.549502106472616e-05,
|
| 861 |
+
"loss": 1.5147,
|
| 862 |
+
"step": 56500
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 4.98,
|
| 866 |
+
"learning_rate": 5.5017234775947915e-05,
|
| 867 |
+
"loss": 1.5136,
|
| 868 |
+
"step": 57000
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 5.02,
|
| 872 |
+
"learning_rate": 5.4539448487169664e-05,
|
| 873 |
+
"loss": 1.5114,
|
| 874 |
+
"step": 57500
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 5.02,
|
| 878 |
+
"eval_loss": 1.455129861831665,
|
| 879 |
+
"eval_runtime": 799.4542,
|
| 880 |
+
"eval_samples_per_second": 96.436,
|
| 881 |
+
"eval_steps_per_second": 6.028,
|
| 882 |
+
"step": 57500
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 5.07,
|
| 886 |
+
"learning_rate": 5.4060704710838764e-05,
|
| 887 |
+
"loss": 1.5048,
|
| 888 |
+
"step": 58000
|
| 889 |
+
},
|
| 890 |
+
{
|
| 891 |
+
"epoch": 5.11,
|
| 892 |
+
"learning_rate": 5.3581960934507856e-05,
|
| 893 |
+
"loss": 1.496,
|
| 894 |
+
"step": 58500
|
| 895 |
+
},
|
| 896 |
+
{
|
| 897 |
+
"epoch": 5.16,
|
| 898 |
+
"learning_rate": 5.310321715817694e-05,
|
| 899 |
+
"loss": 1.4956,
|
| 900 |
+
"step": 59000
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"epoch": 5.2,
|
| 904 |
+
"learning_rate": 5.262447338184604e-05,
|
| 905 |
+
"loss": 1.4965,
|
| 906 |
+
"step": 59500
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 5.24,
|
| 910 |
+
"learning_rate": 5.214668709306779e-05,
|
| 911 |
+
"loss": 1.5054,
|
| 912 |
+
"step": 60000
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 5.24,
|
| 916 |
+
"eval_loss": 1.4524710178375244,
|
| 917 |
+
"eval_runtime": 799.9515,
|
| 918 |
+
"eval_samples_per_second": 96.376,
|
| 919 |
+
"eval_steps_per_second": 6.024,
|
| 920 |
+
"step": 60000
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 5.29,
|
| 924 |
+
"learning_rate": 5.166794331673689e-05,
|
| 925 |
+
"loss": 1.4931,
|
| 926 |
+
"step": 60500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 5.33,
|
| 930 |
+
"learning_rate": 5.1189199540405975e-05,
|
| 931 |
+
"loss": 1.4855,
|
| 932 |
+
"step": 61000
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 5.37,
|
| 936 |
+
"learning_rate": 5.0710455764075074e-05,
|
| 937 |
+
"loss": 1.4925,
|
| 938 |
+
"step": 61500
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 5.42,
|
| 942 |
+
"learning_rate": 5.023171198774416e-05,
|
| 943 |
+
"loss": 1.4789,
|
| 944 |
+
"step": 62000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 5.46,
|
| 948 |
+
"learning_rate": 4.975296821141325e-05,
|
| 949 |
+
"loss": 1.4817,
|
| 950 |
+
"step": 62500
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 5.46,
|
| 954 |
+
"eval_loss": 1.4259248971939087,
|
| 955 |
+
"eval_runtime": 800.5391,
|
| 956 |
+
"eval_samples_per_second": 96.305,
|
| 957 |
+
"eval_steps_per_second": 6.02,
|
| 958 |
+
"step": 62500
|
| 959 |
+
},
|
| 960 |
+
{
|
| 961 |
+
"epoch": 5.51,
|
| 962 |
+
"learning_rate": 4.9274224435082345e-05,
|
| 963 |
+
"loss": 1.4705,
|
| 964 |
+
"step": 63000
|
| 965 |
+
},
|
| 966 |
+
{
|
| 967 |
+
"epoch": 5.55,
|
| 968 |
+
"learning_rate": 4.8795480658751444e-05,
|
| 969 |
+
"loss": 1.4764,
|
| 970 |
+
"step": 63500
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 5.59,
|
| 974 |
+
"learning_rate": 4.8317694369973194e-05,
|
| 975 |
+
"loss": 1.4703,
|
| 976 |
+
"step": 64000
|
| 977 |
+
},
|
| 978 |
+
{
|
| 979 |
+
"epoch": 5.64,
|
| 980 |
+
"learning_rate": 4.7838950593642286e-05,
|
| 981 |
+
"loss": 1.4647,
|
| 982 |
+
"step": 64500
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 5.68,
|
| 986 |
+
"learning_rate": 4.736020681731138e-05,
|
| 987 |
+
"loss": 1.48,
|
| 988 |
+
"step": 65000
|
| 989 |
+
},
|
| 990 |
+
{
|
| 991 |
+
"epoch": 5.68,
|
| 992 |
+
"eval_loss": 1.4077460765838623,
|
| 993 |
+
"eval_runtime": 800.4986,
|
| 994 |
+
"eval_samples_per_second": 96.31,
|
| 995 |
+
"eval_steps_per_second": 6.02,
|
| 996 |
+
"step": 65000
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 5.72,
|
| 1000 |
+
"learning_rate": 4.688146304098047e-05,
|
| 1001 |
+
"loss": 1.4646,
|
| 1002 |
+
"step": 65500
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 5.77,
|
| 1006 |
+
"learning_rate": 4.640271926464956e-05,
|
| 1007 |
+
"loss": 1.4588,
|
| 1008 |
+
"step": 66000
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 5.81,
|
| 1012 |
+
"learning_rate": 4.5923975488318656e-05,
|
| 1013 |
+
"loss": 1.4652,
|
| 1014 |
+
"step": 66500
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 5.85,
|
| 1018 |
+
"learning_rate": 4.5446189199540405e-05,
|
| 1019 |
+
"loss": 1.4505,
|
| 1020 |
+
"step": 67000
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 5.9,
|
| 1024 |
+
"learning_rate": 4.49674454232095e-05,
|
| 1025 |
+
"loss": 1.4526,
|
| 1026 |
+
"step": 67500
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 5.9,
|
| 1030 |
+
"eval_loss": 1.3912079334259033,
|
| 1031 |
+
"eval_runtime": 800.7475,
|
| 1032 |
+
"eval_samples_per_second": 96.28,
|
| 1033 |
+
"eval_steps_per_second": 6.018,
|
| 1034 |
+
"step": 67500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 5.94,
|
| 1038 |
+
"learning_rate": 4.44887016468786e-05,
|
| 1039 |
+
"loss": 1.4444,
|
| 1040 |
+
"step": 68000
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 5.99,
|
| 1044 |
+
"learning_rate": 4.400995787054768e-05,
|
| 1045 |
+
"loss": 1.4416,
|
| 1046 |
+
"step": 68500
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 6.03,
|
| 1050 |
+
"learning_rate": 4.353217158176944e-05,
|
| 1051 |
+
"loss": 1.4435,
|
| 1052 |
+
"step": 69000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 6.07,
|
| 1056 |
+
"learning_rate": 4.305342780543853e-05,
|
| 1057 |
+
"loss": 1.4261,
|
| 1058 |
+
"step": 69500
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 6.12,
|
| 1062 |
+
"learning_rate": 4.2574684029107624e-05,
|
| 1063 |
+
"loss": 1.4272,
|
| 1064 |
+
"step": 70000
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 6.12,
|
| 1068 |
+
"eval_loss": 1.3726316690444946,
|
| 1069 |
+
"eval_runtime": 801.0967,
|
| 1070 |
+
"eval_samples_per_second": 96.238,
|
| 1071 |
+
"eval_steps_per_second": 6.016,
|
| 1072 |
+
"step": 70000
|
| 1073 |
+
},
|
| 1074 |
+
{
|
| 1075 |
+
"epoch": 6.16,
|
| 1076 |
+
"learning_rate": 4.2095940252776716e-05,
|
| 1077 |
+
"loss": 1.4291,
|
| 1078 |
+
"step": 70500
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 6.2,
|
| 1082 |
+
"learning_rate": 4.161719647644581e-05,
|
| 1083 |
+
"loss": 1.4204,
|
| 1084 |
+
"step": 71000
|
| 1085 |
+
},
|
| 1086 |
+
{
|
| 1087 |
+
"epoch": 6.25,
|
| 1088 |
+
"learning_rate": 4.11384527001149e-05,
|
| 1089 |
+
"loss": 1.4143,
|
| 1090 |
+
"step": 71500
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 6.29,
|
| 1094 |
+
"learning_rate": 4.065970892378399e-05,
|
| 1095 |
+
"loss": 1.4219,
|
| 1096 |
+
"step": 72000
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 6.34,
|
| 1100 |
+
"learning_rate": 4.0180965147453086e-05,
|
| 1101 |
+
"loss": 1.4078,
|
| 1102 |
+
"step": 72500
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 6.34,
|
| 1106 |
+
"eval_loss": 1.3595777750015259,
|
| 1107 |
+
"eval_runtime": 801.4171,
|
| 1108 |
+
"eval_samples_per_second": 96.2,
|
| 1109 |
+
"eval_steps_per_second": 6.013,
|
| 1110 |
+
"step": 72500
|
| 1111 |
+
},
|
| 1112 |
+
{
|
| 1113 |
+
"epoch": 6.38,
|
| 1114 |
+
"learning_rate": 3.970317885867484e-05,
|
| 1115 |
+
"loss": 1.3986,
|
| 1116 |
+
"step": 73000
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 6.42,
|
| 1120 |
+
"learning_rate": 3.922443508234393e-05,
|
| 1121 |
+
"loss": 1.3945,
|
| 1122 |
+
"step": 73500
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 6.47,
|
| 1126 |
+
"learning_rate": 3.874569130601302e-05,
|
| 1127 |
+
"loss": 1.404,
|
| 1128 |
+
"step": 74000
|
| 1129 |
+
},
|
| 1130 |
+
{
|
| 1131 |
+
"epoch": 6.51,
|
| 1132 |
+
"learning_rate": 3.826694752968212e-05,
|
| 1133 |
+
"loss": 1.4029,
|
| 1134 |
+
"step": 74500
|
| 1135 |
+
},
|
| 1136 |
+
{
|
| 1137 |
+
"epoch": 6.55,
|
| 1138 |
+
"learning_rate": 3.778916124090387e-05,
|
| 1139 |
+
"loss": 1.399,
|
| 1140 |
+
"step": 75000
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 6.55,
|
| 1144 |
+
"eval_loss": 1.3450202941894531,
|
| 1145 |
+
"eval_runtime": 801.3866,
|
| 1146 |
+
"eval_samples_per_second": 96.203,
|
| 1147 |
+
"eval_steps_per_second": 6.013,
|
| 1148 |
+
"step": 75000
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 6.6,
|
| 1152 |
+
"learning_rate": 3.731041746457296e-05,
|
| 1153 |
+
"loss": 1.3884,
|
| 1154 |
+
"step": 75500
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 6.64,
|
| 1158 |
+
"learning_rate": 3.683263117579472e-05,
|
| 1159 |
+
"loss": 1.382,
|
| 1160 |
+
"step": 76000
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 6.68,
|
| 1164 |
+
"learning_rate": 3.635388739946381e-05,
|
| 1165 |
+
"loss": 1.3888,
|
| 1166 |
+
"step": 76500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 6.73,
|
| 1170 |
+
"learning_rate": 3.58751436231329e-05,
|
| 1171 |
+
"loss": 1.3847,
|
| 1172 |
+
"step": 77000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 6.77,
|
| 1176 |
+
"learning_rate": 3.5396399846801995e-05,
|
| 1177 |
+
"loss": 1.386,
|
| 1178 |
+
"step": 77500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 6.77,
|
| 1182 |
+
"eval_loss": 1.3327932357788086,
|
| 1183 |
+
"eval_runtime": 800.4263,
|
| 1184 |
+
"eval_samples_per_second": 96.319,
|
| 1185 |
+
"eval_steps_per_second": 6.021,
|
| 1186 |
+
"step": 77500
|
| 1187 |
+
},
|
| 1188 |
+
{
|
| 1189 |
+
"epoch": 6.82,
|
| 1190 |
+
"learning_rate": 3.491765607047108e-05,
|
| 1191 |
+
"loss": 1.3941,
|
| 1192 |
+
"step": 78000
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 6.86,
|
| 1196 |
+
"learning_rate": 3.443891229414018e-05,
|
| 1197 |
+
"loss": 1.3698,
|
| 1198 |
+
"step": 78500
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 6.9,
|
| 1202 |
+
"learning_rate": 3.396016851780927e-05,
|
| 1203 |
+
"loss": 1.3712,
|
| 1204 |
+
"step": 79000
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 6.95,
|
| 1208 |
+
"learning_rate": 3.3481424741478364e-05,
|
| 1209 |
+
"loss": 1.3751,
|
| 1210 |
+
"step": 79500
|
| 1211 |
+
},
|
| 1212 |
+
{
|
| 1213 |
+
"epoch": 6.99,
|
| 1214 |
+
"learning_rate": 3.300268096514745e-05,
|
| 1215 |
+
"loss": 1.3704,
|
| 1216 |
+
"step": 80000
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 6.99,
|
| 1220 |
+
"eval_loss": 1.319200038909912,
|
| 1221 |
+
"eval_runtime": 800.9368,
|
| 1222 |
+
"eval_samples_per_second": 96.257,
|
| 1223 |
+
"eval_steps_per_second": 6.017,
|
| 1224 |
+
"step": 80000
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 7.03,
|
| 1228 |
+
"learning_rate": 3.252489467636921e-05,
|
| 1229 |
+
"loss": 1.3614,
|
| 1230 |
+
"step": 80500
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 7.08,
|
| 1234 |
+
"learning_rate": 3.20461509000383e-05,
|
| 1235 |
+
"loss": 1.3624,
|
| 1236 |
+
"step": 81000
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 7.12,
|
| 1240 |
+
"learning_rate": 3.156740712370739e-05,
|
| 1241 |
+
"loss": 1.3534,
|
| 1242 |
+
"step": 81500
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 7.17,
|
| 1246 |
+
"learning_rate": 3.1088663347376484e-05,
|
| 1247 |
+
"loss": 1.3537,
|
| 1248 |
+
"step": 82000
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 7.21,
|
| 1252 |
+
"learning_rate": 3.061087705859824e-05,
|
| 1253 |
+
"loss": 1.3538,
|
| 1254 |
+
"step": 82500
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 7.21,
|
| 1258 |
+
"eval_loss": 1.3130738735198975,
|
| 1259 |
+
"eval_runtime": 801.3578,
|
| 1260 |
+
"eval_samples_per_second": 96.207,
|
| 1261 |
+
"eval_steps_per_second": 6.014,
|
| 1262 |
+
"step": 82500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 7.25,
|
| 1266 |
+
"learning_rate": 3.0132133282267332e-05,
|
| 1267 |
+
"loss": 1.343,
|
| 1268 |
+
"step": 83000
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 7.3,
|
| 1272 |
+
"learning_rate": 2.9653389505936425e-05,
|
| 1273 |
+
"loss": 1.3452,
|
| 1274 |
+
"step": 83500
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 7.34,
|
| 1278 |
+
"learning_rate": 2.9174645729605514e-05,
|
| 1279 |
+
"loss": 1.3414,
|
| 1280 |
+
"step": 84000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 7.38,
|
| 1284 |
+
"learning_rate": 2.8695901953274606e-05,
|
| 1285 |
+
"loss": 1.3449,
|
| 1286 |
+
"step": 84500
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 7.43,
|
| 1290 |
+
"learning_rate": 2.8217158176943702e-05,
|
| 1291 |
+
"loss": 1.3468,
|
| 1292 |
+
"step": 85000
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 7.43,
|
| 1296 |
+
"eval_loss": 1.2915565967559814,
|
| 1297 |
+
"eval_runtime": 801.1997,
|
| 1298 |
+
"eval_samples_per_second": 96.226,
|
| 1299 |
+
"eval_steps_per_second": 6.015,
|
| 1300 |
+
"step": 85000
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 7.47,
|
| 1304 |
+
"learning_rate": 2.7739371888165455e-05,
|
| 1305 |
+
"loss": 1.3406,
|
| 1306 |
+
"step": 85500
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 7.51,
|
| 1310 |
+
"learning_rate": 2.7260628111834547e-05,
|
| 1311 |
+
"loss": 1.337,
|
| 1312 |
+
"step": 86000
|
| 1313 |
+
},
|
| 1314 |
+
{
|
| 1315 |
+
"epoch": 7.56,
|
| 1316 |
+
"learning_rate": 2.6781884335503636e-05,
|
| 1317 |
+
"loss": 1.3374,
|
| 1318 |
+
"step": 86500
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 7.6,
|
| 1322 |
+
"learning_rate": 2.6303140559172735e-05,
|
| 1323 |
+
"loss": 1.3232,
|
| 1324 |
+
"step": 87000
|
| 1325 |
+
},
|
| 1326 |
+
{
|
| 1327 |
+
"epoch": 7.65,
|
| 1328 |
+
"learning_rate": 2.5824396782841825e-05,
|
| 1329 |
+
"loss": 1.323,
|
| 1330 |
+
"step": 87500
|
| 1331 |
+
},
|
| 1332 |
+
{
|
| 1333 |
+
"epoch": 7.65,
|
| 1334 |
+
"eval_loss": 1.2871359586715698,
|
| 1335 |
+
"eval_runtime": 801.3386,
|
| 1336 |
+
"eval_samples_per_second": 96.209,
|
| 1337 |
+
"eval_steps_per_second": 6.014,
|
| 1338 |
+
"step": 87500
|
| 1339 |
+
},
|
| 1340 |
+
{
|
| 1341 |
+
"epoch": 7.69,
|
| 1342 |
+
"learning_rate": 2.5345653006510917e-05,
|
| 1343 |
+
"loss": 1.3165,
|
| 1344 |
+
"step": 88000
|
| 1345 |
+
},
|
| 1346 |
+
{
|
| 1347 |
+
"epoch": 7.73,
|
| 1348 |
+
"learning_rate": 2.4866909230180006e-05,
|
| 1349 |
+
"loss": 1.3311,
|
| 1350 |
+
"step": 88500
|
| 1351 |
+
},
|
| 1352 |
+
{
|
| 1353 |
+
"epoch": 7.78,
|
| 1354 |
+
"learning_rate": 2.4389122941401762e-05,
|
| 1355 |
+
"loss": 1.3187,
|
| 1356 |
+
"step": 89000
|
| 1357 |
+
},
|
| 1358 |
+
{
|
| 1359 |
+
"epoch": 7.82,
|
| 1360 |
+
"learning_rate": 2.3910379165070858e-05,
|
| 1361 |
+
"loss": 1.3243,
|
| 1362 |
+
"step": 89500
|
| 1363 |
+
},
|
| 1364 |
+
{
|
| 1365 |
+
"epoch": 7.86,
|
| 1366 |
+
"learning_rate": 2.3431635388739947e-05,
|
| 1367 |
+
"loss": 1.322,
|
| 1368 |
+
"step": 90000
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 7.86,
|
| 1372 |
+
"eval_loss": 1.262225866317749,
|
| 1373 |
+
"eval_runtime": 799.8871,
|
| 1374 |
+
"eval_samples_per_second": 96.384,
|
| 1375 |
+
"eval_steps_per_second": 6.025,
|
| 1376 |
+
"step": 90000
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 7.91,
|
| 1380 |
+
"learning_rate": 2.295289161240904e-05,
|
| 1381 |
+
"loss": 1.3195,
|
| 1382 |
+
"step": 90500
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 7.95,
|
| 1386 |
+
"learning_rate": 2.2475105323630792e-05,
|
| 1387 |
+
"loss": 1.3097,
|
| 1388 |
+
"step": 91000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 8.0,
|
| 1392 |
+
"learning_rate": 2.1996361547299888e-05,
|
| 1393 |
+
"loss": 1.3154,
|
| 1394 |
+
"step": 91500
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 8.04,
|
| 1398 |
+
"learning_rate": 2.1517617770968977e-05,
|
| 1399 |
+
"loss": 1.2994,
|
| 1400 |
+
"step": 92000
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 8.08,
|
| 1404 |
+
"learning_rate": 2.103887399463807e-05,
|
| 1405 |
+
"loss": 1.2956,
|
| 1406 |
+
"step": 92500
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 8.08,
|
| 1410 |
+
"eval_loss": 1.2623521089553833,
|
| 1411 |
+
"eval_runtime": 801.376,
|
| 1412 |
+
"eval_samples_per_second": 96.205,
|
| 1413 |
+
"eval_steps_per_second": 6.013,
|
| 1414 |
+
"step": 92500
|
| 1415 |
+
},
|
| 1416 |
+
{
|
| 1417 |
+
"epoch": 8.13,
|
| 1418 |
+
"learning_rate": 2.0560130218307162e-05,
|
| 1419 |
+
"loss": 1.2956,
|
| 1420 |
+
"step": 93000
|
| 1421 |
+
},
|
| 1422 |
+
{
|
| 1423 |
+
"epoch": 8.17,
|
| 1424 |
+
"learning_rate": 2.0081386441976255e-05,
|
| 1425 |
+
"loss": 1.2945,
|
| 1426 |
+
"step": 93500
|
| 1427 |
+
},
|
| 1428 |
+
{
|
| 1429 |
+
"epoch": 8.21,
|
| 1430 |
+
"learning_rate": 1.960360015319801e-05,
|
| 1431 |
+
"loss": 1.2971,
|
| 1432 |
+
"step": 94000
|
| 1433 |
+
},
|
| 1434 |
+
{
|
| 1435 |
+
"epoch": 8.26,
|
| 1436 |
+
"learning_rate": 1.91248563768671e-05,
|
| 1437 |
+
"loss": 1.2941,
|
| 1438 |
+
"step": 94500
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 8.3,
|
| 1442 |
+
"learning_rate": 1.8646112600536196e-05,
|
| 1443 |
+
"loss": 1.2869,
|
| 1444 |
+
"step": 95000
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 8.3,
|
| 1448 |
+
"eval_loss": 1.2546786069869995,
|
| 1449 |
+
"eval_runtime": 800.9015,
|
| 1450 |
+
"eval_samples_per_second": 96.262,
|
| 1451 |
+
"eval_steps_per_second": 6.017,
|
| 1452 |
+
"step": 95000
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 8.34,
|
| 1456 |
+
"learning_rate": 1.8167368824205285e-05,
|
| 1457 |
+
"loss": 1.2868,
|
| 1458 |
+
"step": 95500
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 8.39,
|
| 1462 |
+
"learning_rate": 1.768958253542704e-05,
|
| 1463 |
+
"loss": 1.2939,
|
| 1464 |
+
"step": 96000
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 8.43,
|
| 1468 |
+
"learning_rate": 1.7210838759096133e-05,
|
| 1469 |
+
"loss": 1.2876,
|
| 1470 |
+
"step": 96500
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 8.48,
|
| 1474 |
+
"learning_rate": 1.6732094982765226e-05,
|
| 1475 |
+
"loss": 1.2843,
|
| 1476 |
+
"step": 97000
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 8.52,
|
| 1480 |
+
"learning_rate": 1.625430869398698e-05,
|
| 1481 |
+
"loss": 1.2763,
|
| 1482 |
+
"step": 97500
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 8.52,
|
| 1486 |
+
"eval_loss": 1.2404190301895142,
|
| 1487 |
+
"eval_runtime": 800.8453,
|
| 1488 |
+
"eval_samples_per_second": 96.268,
|
| 1489 |
+
"eval_steps_per_second": 6.017,
|
| 1490 |
+
"step": 97500
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 8.56,
|
| 1494 |
+
"learning_rate": 1.577556491765607e-05,
|
| 1495 |
+
"loss": 1.2845,
|
| 1496 |
+
"step": 98000
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 8.61,
|
| 1500 |
+
"learning_rate": 1.5296821141325163e-05,
|
| 1501 |
+
"loss": 1.2795,
|
| 1502 |
+
"step": 98500
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 8.65,
|
| 1506 |
+
"learning_rate": 1.4818077364994256e-05,
|
| 1507 |
+
"loss": 1.275,
|
| 1508 |
+
"step": 99000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 8.69,
|
| 1512 |
+
"learning_rate": 1.4339333588663348e-05,
|
| 1513 |
+
"loss": 1.2763,
|
| 1514 |
+
"step": 99500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 8.74,
|
| 1518 |
+
"learning_rate": 1.3860589812332439e-05,
|
| 1519 |
+
"loss": 1.275,
|
| 1520 |
+
"step": 100000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 8.74,
|
| 1524 |
+
"eval_loss": 1.2305477857589722,
|
| 1525 |
+
"eval_runtime": 799.9521,
|
| 1526 |
+
"eval_samples_per_second": 96.376,
|
| 1527 |
+
"eval_steps_per_second": 6.024,
|
| 1528 |
+
"step": 100000
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 8.78,
|
| 1532 |
+
"learning_rate": 1.3381846036001533e-05,
|
| 1533 |
+
"loss": 1.2724,
|
| 1534 |
+
"step": 100500
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 8.83,
|
| 1538 |
+
"learning_rate": 1.2903102259670624e-05,
|
| 1539 |
+
"loss": 1.2779,
|
| 1540 |
+
"step": 101000
|
| 1541 |
+
},
|
| 1542 |
+
{
|
| 1543 |
+
"epoch": 8.87,
|
| 1544 |
+
"learning_rate": 1.2424358483339716e-05,
|
| 1545 |
+
"loss": 1.2625,
|
| 1546 |
+
"step": 101500
|
| 1547 |
+
},
|
| 1548 |
+
{
|
| 1549 |
+
"epoch": 8.91,
|
| 1550 |
+
"learning_rate": 1.1946572194561471e-05,
|
| 1551 |
+
"loss": 1.2647,
|
| 1552 |
+
"step": 102000
|
| 1553 |
+
},
|
| 1554 |
+
{
|
| 1555 |
+
"epoch": 8.96,
|
| 1556 |
+
"learning_rate": 1.1467828418230563e-05,
|
| 1557 |
+
"loss": 1.2709,
|
| 1558 |
+
"step": 102500
|
| 1559 |
+
},
|
| 1560 |
+
{
|
| 1561 |
+
"epoch": 8.96,
|
| 1562 |
+
"eval_loss": 1.2300862073898315,
|
| 1563 |
+
"eval_runtime": 800.2626,
|
| 1564 |
+
"eval_samples_per_second": 96.338,
|
| 1565 |
+
"eval_steps_per_second": 6.022,
|
| 1566 |
+
"step": 102500
|
| 1567 |
+
},
|
| 1568 |
+
{
|
| 1569 |
+
"epoch": 9.0,
|
| 1570 |
+
"learning_rate": 1.0989084641899656e-05,
|
| 1571 |
+
"loss": 1.2647,
|
| 1572 |
+
"step": 103000
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 9.04,
|
| 1576 |
+
"learning_rate": 1.0510340865568748e-05,
|
| 1577 |
+
"loss": 1.2462,
|
| 1578 |
+
"step": 103500
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 9.09,
|
| 1582 |
+
"learning_rate": 1.003159708923784e-05,
|
| 1583 |
+
"loss": 1.2586,
|
| 1584 |
+
"step": 104000
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 9.13,
|
| 1588 |
+
"learning_rate": 9.552853312906933e-06,
|
| 1589 |
+
"loss": 1.2455,
|
| 1590 |
+
"step": 104500
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 9.18,
|
| 1594 |
+
"learning_rate": 9.074109536576026e-06,
|
| 1595 |
+
"loss": 1.2514,
|
| 1596 |
+
"step": 105000
|
| 1597 |
+
},
|
| 1598 |
+
{
|
| 1599 |
+
"epoch": 9.18,
|
| 1600 |
+
"eval_loss": 1.2178542613983154,
|
| 1601 |
+
"eval_runtime": 829.9046,
|
| 1602 |
+
"eval_samples_per_second": 92.897,
|
| 1603 |
+
"eval_steps_per_second": 5.807,
|
| 1604 |
+
"step": 105000
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 9.22,
|
| 1608 |
+
"learning_rate": 8.595365760245118e-06,
|
| 1609 |
+
"loss": 1.2526,
|
| 1610 |
+
"step": 105500
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 9.26,
|
| 1614 |
+
"learning_rate": 8.11757947146687e-06,
|
| 1615 |
+
"loss": 1.2523,
|
| 1616 |
+
"step": 106000
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 9.31,
|
| 1620 |
+
"learning_rate": 7.638835695135963e-06,
|
| 1621 |
+
"loss": 1.2504,
|
| 1622 |
+
"step": 106500
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 9.35,
|
| 1626 |
+
"learning_rate": 7.1600919188050565e-06,
|
| 1627 |
+
"loss": 1.2516,
|
| 1628 |
+
"step": 107000
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 9.39,
|
| 1632 |
+
"learning_rate": 6.681348142474147e-06,
|
| 1633 |
+
"loss": 1.2563,
|
| 1634 |
+
"step": 107500
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 9.39,
|
| 1638 |
+
"eval_loss": 1.2134195566177368,
|
| 1639 |
+
"eval_runtime": 802.5122,
|
| 1640 |
+
"eval_samples_per_second": 96.068,
|
| 1641 |
+
"eval_steps_per_second": 6.005,
|
| 1642 |
+
"step": 107500
|
| 1643 |
+
},
|
| 1644 |
+
{
|
| 1645 |
+
"epoch": 9.44,
|
| 1646 |
+
"learning_rate": 6.204519341248564e-06,
|
| 1647 |
+
"loss": 1.2407,
|
| 1648 |
+
"step": 108000
|
| 1649 |
+
},
|
| 1650 |
+
{
|
| 1651 |
+
"epoch": 9.48,
|
| 1652 |
+
"learning_rate": 5.725775564917656e-06,
|
| 1653 |
+
"loss": 1.243,
|
| 1654 |
+
"step": 108500
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 9.52,
|
| 1658 |
+
"learning_rate": 5.247031788586749e-06,
|
| 1659 |
+
"loss": 1.2449,
|
| 1660 |
+
"step": 109000
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 9.57,
|
| 1664 |
+
"learning_rate": 4.768288012255841e-06,
|
| 1665 |
+
"loss": 1.2442,
|
| 1666 |
+
"step": 109500
|
| 1667 |
+
},
|
| 1668 |
+
{
|
| 1669 |
+
"epoch": 9.61,
|
| 1670 |
+
"learning_rate": 4.290501723477595e-06,
|
| 1671 |
+
"loss": 1.2487,
|
| 1672 |
+
"step": 110000
|
| 1673 |
+
},
|
| 1674 |
+
{
|
| 1675 |
+
"epoch": 9.61,
|
| 1676 |
+
"eval_loss": 1.211136817932129,
|
| 1677 |
+
"eval_runtime": 800.5992,
|
| 1678 |
+
"eval_samples_per_second": 96.298,
|
| 1679 |
+
"eval_steps_per_second": 6.019,
|
| 1680 |
+
"step": 110000
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 9.66,
|
| 1684 |
+
"learning_rate": 3.811757947146687e-06,
|
| 1685 |
+
"loss": 1.2438,
|
| 1686 |
+
"step": 110500
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 9.7,
|
| 1690 |
+
"learning_rate": 3.33301417081578e-06,
|
| 1691 |
+
"loss": 1.2399,
|
| 1692 |
+
"step": 111000
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 9.74,
|
| 1696 |
+
"learning_rate": 2.8542703944848716e-06,
|
| 1697 |
+
"loss": 1.236,
|
| 1698 |
+
"step": 111500
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 9.79,
|
| 1702 |
+
"learning_rate": 2.375526618153964e-06,
|
| 1703 |
+
"loss": 1.2381,
|
| 1704 |
+
"step": 112000
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 9.83,
|
| 1708 |
+
"learning_rate": 1.8977403293757183e-06,
|
| 1709 |
+
"loss": 1.2337,
|
| 1710 |
+
"step": 112500
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 9.83,
|
| 1714 |
+
"eval_loss": 1.2040736675262451,
|
| 1715 |
+
"eval_runtime": 801.2606,
|
| 1716 |
+
"eval_samples_per_second": 96.218,
|
| 1717 |
+
"eval_steps_per_second": 6.014,
|
| 1718 |
+
"step": 112500
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 9.87,
|
| 1722 |
+
"learning_rate": 5.296144005847954e-05,
|
| 1723 |
+
"loss": 1.2919,
|
| 1724 |
+
"step": 113000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 9.92,
|
| 1728 |
+
"learning_rate": 5.273300438596491e-05,
|
| 1729 |
+
"loss": 1.3121,
|
| 1730 |
+
"step": 113500
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 9.96,
|
| 1734 |
+
"learning_rate": 5.25045687134503e-05,
|
| 1735 |
+
"loss": 1.3188,
|
| 1736 |
+
"step": 114000
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 10.01,
|
| 1740 |
+
"learning_rate": 5.2276589912280705e-05,
|
| 1741 |
+
"loss": 1.3266,
|
| 1742 |
+
"step": 114500
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 10.05,
|
| 1746 |
+
"learning_rate": 5.204815423976609e-05,
|
| 1747 |
+
"loss": 1.3215,
|
| 1748 |
+
"step": 115000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 10.05,
|
| 1752 |
+
"eval_loss": 1.2878538370132446,
|
| 1753 |
+
"eval_runtime": 801.5103,
|
| 1754 |
+
"eval_samples_per_second": 96.188,
|
| 1755 |
+
"eval_steps_per_second": 6.012,
|
| 1756 |
+
"step": 115000
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 10.09,
|
| 1760 |
+
"learning_rate": 5.1819718567251463e-05,
|
| 1761 |
+
"loss": 1.3241,
|
| 1762 |
+
"step": 115500
|
| 1763 |
+
},
|
| 1764 |
+
{
|
| 1765 |
+
"epoch": 10.14,
|
| 1766 |
+
"learning_rate": 5.1591282894736846e-05,
|
| 1767 |
+
"loss": 1.3216,
|
| 1768 |
+
"step": 116000
|
| 1769 |
+
},
|
| 1770 |
+
{
|
| 1771 |
+
"epoch": 10.18,
|
| 1772 |
+
"learning_rate": 5.136284722222222e-05,
|
| 1773 |
+
"loss": 1.3297,
|
| 1774 |
+
"step": 116500
|
| 1775 |
+
},
|
| 1776 |
+
{
|
| 1777 |
+
"epoch": 10.22,
|
| 1778 |
+
"learning_rate": 5.1134411549707604e-05,
|
| 1779 |
+
"loss": 1.332,
|
| 1780 |
+
"step": 117000
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 10.27,
|
| 1784 |
+
"learning_rate": 5.090597587719298e-05,
|
| 1785 |
+
"loss": 1.3364,
|
| 1786 |
+
"step": 117500
|
| 1787 |
+
},
|
| 1788 |
+
{
|
| 1789 |
+
"epoch": 10.27,
|
| 1790 |
+
"eval_loss": 1.2850462198257446,
|
| 1791 |
+
"eval_runtime": 799.6277,
|
| 1792 |
+
"eval_samples_per_second": 96.415,
|
| 1793 |
+
"eval_steps_per_second": 6.027,
|
| 1794 |
+
"step": 117500
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 10.31,
|
| 1798 |
+
"learning_rate": 5.06779970760234e-05,
|
| 1799 |
+
"loss": 1.3357,
|
| 1800 |
+
"step": 118000
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 10.35,
|
| 1804 |
+
"learning_rate": 5.044956140350877e-05,
|
| 1805 |
+
"loss": 1.3299,
|
| 1806 |
+
"step": 118500
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 10.4,
|
| 1810 |
+
"learning_rate": 5.022112573099416e-05,
|
| 1811 |
+
"loss": 1.3297,
|
| 1812 |
+
"step": 119000
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 10.44,
|
| 1816 |
+
"learning_rate": 4.9992690058479533e-05,
|
| 1817 |
+
"loss": 1.3227,
|
| 1818 |
+
"step": 119500
|
| 1819 |
+
},
|
| 1820 |
+
{
|
| 1821 |
+
"epoch": 10.49,
|
| 1822 |
+
"learning_rate": 4.9764254385964916e-05,
|
| 1823 |
+
"loss": 1.3286,
|
| 1824 |
+
"step": 120000
|
| 1825 |
+
},
|
| 1826 |
+
{
|
| 1827 |
+
"epoch": 10.49,
|
| 1828 |
+
"eval_loss": 1.2779029607772827,
|
| 1829 |
+
"eval_runtime": 798.5622,
|
| 1830 |
+
"eval_samples_per_second": 96.544,
|
| 1831 |
+
"eval_steps_per_second": 6.035,
|
| 1832 |
+
"step": 120000
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 10.53,
|
| 1836 |
+
"learning_rate": 4.953627558479532e-05,
|
| 1837 |
+
"loss": 1.3185,
|
| 1838 |
+
"step": 120500
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 10.57,
|
| 1842 |
+
"learning_rate": 4.9307839912280704e-05,
|
| 1843 |
+
"loss": 1.3202,
|
| 1844 |
+
"step": 121000
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 10.62,
|
| 1848 |
+
"learning_rate": 4.9079861111111116e-05,
|
| 1849 |
+
"loss": 1.3174,
|
| 1850 |
+
"step": 121500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 10.66,
|
| 1854 |
+
"learning_rate": 4.885142543859649e-05,
|
| 1855 |
+
"loss": 1.3146,
|
| 1856 |
+
"step": 122000
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 10.7,
|
| 1860 |
+
"learning_rate": 4.8622989766081875e-05,
|
| 1861 |
+
"loss": 1.3202,
|
| 1862 |
+
"step": 122500
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 10.7,
|
| 1866 |
+
"eval_loss": 1.2730140686035156,
|
| 1867 |
+
"eval_runtime": 800.6627,
|
| 1868 |
+
"eval_samples_per_second": 96.29,
|
| 1869 |
+
"eval_steps_per_second": 6.019,
|
| 1870 |
+
"step": 122500
|
| 1871 |
+
},
|
| 1872 |
+
{
|
| 1873 |
+
"epoch": 10.75,
|
| 1874 |
+
"learning_rate": 4.839455409356725e-05,
|
| 1875 |
+
"loss": 1.3214,
|
| 1876 |
+
"step": 123000
|
| 1877 |
+
},
|
| 1878 |
+
{
|
| 1879 |
+
"epoch": 10.79,
|
| 1880 |
+
"learning_rate": 4.816611842105263e-05,
|
| 1881 |
+
"loss": 1.3105,
|
| 1882 |
+
"step": 123500
|
| 1883 |
+
},
|
| 1884 |
+
{
|
| 1885 |
+
"epoch": 10.84,
|
| 1886 |
+
"learning_rate": 4.7937682748538016e-05,
|
| 1887 |
+
"loss": 1.3203,
|
| 1888 |
+
"step": 124000
|
| 1889 |
+
},
|
| 1890 |
+
{
|
| 1891 |
+
"epoch": 10.88,
|
| 1892 |
+
"learning_rate": 4.770924707602339e-05,
|
| 1893 |
+
"loss": 1.3061,
|
| 1894 |
+
"step": 124500
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 10.92,
|
| 1898 |
+
"learning_rate": 4.7480811403508774e-05,
|
| 1899 |
+
"loss": 1.3181,
|
| 1900 |
+
"step": 125000
|
| 1901 |
+
},
|
| 1902 |
+
{
|
| 1903 |
+
"epoch": 10.92,
|
| 1904 |
+
"eval_loss": 1.2651313543319702,
|
| 1905 |
+
"eval_runtime": 800.4049,
|
| 1906 |
+
"eval_samples_per_second": 96.321,
|
| 1907 |
+
"eval_steps_per_second": 6.021,
|
| 1908 |
+
"step": 125000
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 10.97,
|
| 1912 |
+
"learning_rate": 4.725237573099415e-05,
|
| 1913 |
+
"loss": 1.3121,
|
| 1914 |
+
"step": 125500
|
| 1915 |
+
},
|
| 1916 |
+
{
|
| 1917 |
+
"epoch": 11.01,
|
| 1918 |
+
"learning_rate": 4.702439692982457e-05,
|
| 1919 |
+
"loss": 1.2988,
|
| 1920 |
+
"step": 126000
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 11.05,
|
| 1924 |
+
"learning_rate": 4.6795961257309945e-05,
|
| 1925 |
+
"loss": 1.2894,
|
| 1926 |
+
"step": 126500
|
| 1927 |
+
},
|
| 1928 |
+
{
|
| 1929 |
+
"epoch": 11.1,
|
| 1930 |
+
"learning_rate": 4.656752558479533e-05,
|
| 1931 |
+
"loss": 1.2976,
|
| 1932 |
+
"step": 127000
|
| 1933 |
+
},
|
| 1934 |
+
{
|
| 1935 |
+
"epoch": 11.14,
|
| 1936 |
+
"learning_rate": 4.63390899122807e-05,
|
| 1937 |
+
"loss": 1.2952,
|
| 1938 |
+
"step": 127500
|
| 1939 |
+
},
|
| 1940 |
+
{
|
| 1941 |
+
"epoch": 11.14,
|
| 1942 |
+
"eval_loss": 1.2544220685958862,
|
| 1943 |
+
"eval_runtime": 798.8455,
|
| 1944 |
+
"eval_samples_per_second": 96.509,
|
| 1945 |
+
"eval_steps_per_second": 6.032,
|
| 1946 |
+
"step": 127500
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 11.18,
|
| 1950 |
+
"learning_rate": 4.6111111111111115e-05,
|
| 1951 |
+
"loss": 1.2929,
|
| 1952 |
+
"step": 128000
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 11.23,
|
| 1956 |
+
"learning_rate": 4.58826754385965e-05,
|
| 1957 |
+
"loss": 1.2877,
|
| 1958 |
+
"step": 128500
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 11.27,
|
| 1962 |
+
"learning_rate": 4.5654239766081874e-05,
|
| 1963 |
+
"loss": 1.2868,
|
| 1964 |
+
"step": 129000
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 11.32,
|
| 1968 |
+
"learning_rate": 4.5425804093567256e-05,
|
| 1969 |
+
"loss": 1.2839,
|
| 1970 |
+
"step": 129500
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 11.36,
|
| 1974 |
+
"learning_rate": 4.519736842105263e-05,
|
| 1975 |
+
"loss": 1.2889,
|
| 1976 |
+
"step": 130000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 11.36,
|
| 1980 |
+
"eval_loss": 1.2505890130996704,
|
| 1981 |
+
"eval_runtime": 798.9045,
|
| 1982 |
+
"eval_samples_per_second": 96.502,
|
| 1983 |
+
"eval_steps_per_second": 6.032,
|
| 1984 |
+
"step": 130000
|
| 1985 |
+
},
|
| 1986 |
+
{
|
| 1987 |
+
"epoch": 11.4,
|
| 1988 |
+
"learning_rate": 4.4969389619883044e-05,
|
| 1989 |
+
"loss": 1.2844,
|
| 1990 |
+
"step": 130500
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 11.45,
|
| 1994 |
+
"learning_rate": 4.474095394736843e-05,
|
| 1995 |
+
"loss": 1.286,
|
| 1996 |
+
"step": 131000
|
| 1997 |
+
},
|
| 1998 |
+
{
|
| 1999 |
+
"epoch": 11.49,
|
| 2000 |
+
"learning_rate": 4.45125182748538e-05,
|
| 2001 |
+
"loss": 1.2997,
|
| 2002 |
+
"step": 131500
|
| 2003 |
+
},
|
| 2004 |
+
{
|
| 2005 |
+
"epoch": 11.53,
|
| 2006 |
+
"learning_rate": 4.4284082602339185e-05,
|
| 2007 |
+
"loss": 1.2813,
|
| 2008 |
+
"step": 132000
|
| 2009 |
+
},
|
| 2010 |
+
{
|
| 2011 |
+
"epoch": 11.58,
|
| 2012 |
+
"learning_rate": 4.405610380116959e-05,
|
| 2013 |
+
"loss": 1.2747,
|
| 2014 |
+
"step": 132500
|
| 2015 |
+
},
|
| 2016 |
+
{
|
| 2017 |
+
"epoch": 11.58,
|
| 2018 |
+
"eval_loss": 1.2338725328445435,
|
| 2019 |
+
"eval_runtime": 799.247,
|
| 2020 |
+
"eval_samples_per_second": 96.461,
|
| 2021 |
+
"eval_steps_per_second": 6.029,
|
| 2022 |
+
"step": 132500
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 11.62,
|
| 2026 |
+
"learning_rate": 4.3827668128654973e-05,
|
| 2027 |
+
"loss": 1.2803,
|
| 2028 |
+
"step": 133000
|
| 2029 |
+
},
|
| 2030 |
+
{
|
| 2031 |
+
"epoch": 11.67,
|
| 2032 |
+
"learning_rate": 4.3599232456140356e-05,
|
| 2033 |
+
"loss": 1.2869,
|
| 2034 |
+
"step": 133500
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 11.71,
|
| 2038 |
+
"learning_rate": 4.337125365497076e-05,
|
| 2039 |
+
"loss": 1.2837,
|
| 2040 |
+
"step": 134000
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 11.75,
|
| 2044 |
+
"learning_rate": 4.3142817982456144e-05,
|
| 2045 |
+
"loss": 1.2826,
|
| 2046 |
+
"step": 134500
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 11.8,
|
| 2050 |
+
"learning_rate": 4.291438230994152e-05,
|
| 2051 |
+
"loss": 1.2729,
|
| 2052 |
+
"step": 135000
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 11.8,
|
| 2056 |
+
"eval_loss": 1.2277261018753052,
|
| 2057 |
+
"eval_runtime": 794.6626,
|
| 2058 |
+
"eval_samples_per_second": 97.017,
|
| 2059 |
+
"eval_steps_per_second": 6.064,
|
| 2060 |
+
"step": 135000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 11.84,
|
| 2064 |
+
"learning_rate": 4.26859466374269e-05,
|
| 2065 |
+
"loss": 1.2705,
|
| 2066 |
+
"step": 135500
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 11.88,
|
| 2070 |
+
"learning_rate": 4.2457510964912285e-05,
|
| 2071 |
+
"loss": 1.2719,
|
| 2072 |
+
"step": 136000
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 11.93,
|
| 2076 |
+
"learning_rate": 4.222907529239766e-05,
|
| 2077 |
+
"loss": 1.2672,
|
| 2078 |
+
"step": 136500
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 11.97,
|
| 2082 |
+
"learning_rate": 4.200063961988304e-05,
|
| 2083 |
+
"loss": 1.2699,
|
| 2084 |
+
"step": 137000
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 12.02,
|
| 2088 |
+
"learning_rate": 4.177220394736842e-05,
|
| 2089 |
+
"loss": 1.2699,
|
| 2090 |
+
"step": 137500
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 12.02,
|
| 2094 |
+
"eval_loss": 1.220065951347351,
|
| 2095 |
+
"eval_runtime": 798.5913,
|
| 2096 |
+
"eval_samples_per_second": 96.54,
|
| 2097 |
+
"eval_steps_per_second": 6.034,
|
| 2098 |
+
"step": 137500
|
| 2099 |
+
},
|
| 2100 |
+
{
|
| 2101 |
+
"epoch": 12.06,
|
| 2102 |
+
"learning_rate": 4.154422514619883e-05,
|
| 2103 |
+
"loss": 1.2554,
|
| 2104 |
+
"step": 138000
|
| 2105 |
+
},
|
| 2106 |
+
{
|
| 2107 |
+
"epoch": 12.1,
|
| 2108 |
+
"learning_rate": 4.1315789473684214e-05,
|
| 2109 |
+
"loss": 1.2609,
|
| 2110 |
+
"step": 138500
|
| 2111 |
+
},
|
| 2112 |
+
{
|
| 2113 |
+
"epoch": 12.15,
|
| 2114 |
+
"learning_rate": 4.108735380116959e-05,
|
| 2115 |
+
"loss": 1.2639,
|
| 2116 |
+
"step": 139000
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 12.19,
|
| 2120 |
+
"learning_rate": 4.085891812865497e-05,
|
| 2121 |
+
"loss": 1.2491,
|
| 2122 |
+
"step": 139500
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 12.23,
|
| 2126 |
+
"learning_rate": 4.063048245614035e-05,
|
| 2127 |
+
"loss": 1.2508,
|
| 2128 |
+
"step": 140000
|
| 2129 |
+
},
|
| 2130 |
+
{
|
| 2131 |
+
"epoch": 12.23,
|
| 2132 |
+
"eval_loss": 1.2163188457489014,
|
| 2133 |
+
"eval_runtime": 791.6231,
|
| 2134 |
+
"eval_samples_per_second": 97.39,
|
| 2135 |
+
"eval_steps_per_second": 6.087,
|
| 2136 |
+
"step": 140000
|
| 2137 |
+
},
|
| 2138 |
+
{
|
| 2139 |
+
"epoch": 12.28,
|
| 2140 |
+
"learning_rate": 4.040250365497076e-05,
|
| 2141 |
+
"loss": 1.2521,
|
| 2142 |
+
"step": 140500
|
| 2143 |
+
},
|
| 2144 |
+
{
|
| 2145 |
+
"epoch": 12.32,
|
| 2146 |
+
"learning_rate": 4.017406798245614e-05,
|
| 2147 |
+
"loss": 1.256,
|
| 2148 |
+
"step": 141000
|
| 2149 |
+
},
|
| 2150 |
+
{
|
| 2151 |
+
"epoch": 12.36,
|
| 2152 |
+
"learning_rate": 3.994563230994152e-05,
|
| 2153 |
+
"loss": 1.245,
|
| 2154 |
+
"step": 141500
|
| 2155 |
+
},
|
| 2156 |
+
{
|
| 2157 |
+
"epoch": 12.41,
|
| 2158 |
+
"learning_rate": 3.97171966374269e-05,
|
| 2159 |
+
"loss": 1.2516,
|
| 2160 |
+
"step": 142000
|
| 2161 |
+
},
|
| 2162 |
+
{
|
| 2163 |
+
"epoch": 12.45,
|
| 2164 |
+
"learning_rate": 3.9489217836257314e-05,
|
| 2165 |
+
"loss": 1.2438,
|
| 2166 |
+
"step": 142500
|
| 2167 |
+
},
|
| 2168 |
+
{
|
| 2169 |
+
"epoch": 12.45,
|
| 2170 |
+
"eval_loss": 1.2091211080551147,
|
| 2171 |
+
"eval_runtime": 798.8782,
|
| 2172 |
+
"eval_samples_per_second": 96.505,
|
| 2173 |
+
"eval_steps_per_second": 6.032,
|
| 2174 |
+
"step": 142500
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 12.5,
|
| 2178 |
+
"learning_rate": 3.9260782163742696e-05,
|
| 2179 |
+
"loss": 1.2505,
|
| 2180 |
+
"step": 143000
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 12.54,
|
| 2184 |
+
"learning_rate": 3.903234649122807e-05,
|
| 2185 |
+
"loss": 1.2397,
|
| 2186 |
+
"step": 143500
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 12.58,
|
| 2190 |
+
"learning_rate": 3.8803910818713455e-05,
|
| 2191 |
+
"loss": 1.248,
|
| 2192 |
+
"step": 144000
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 12.63,
|
| 2196 |
+
"learning_rate": 3.857547514619883e-05,
|
| 2197 |
+
"loss": 1.2403,
|
| 2198 |
+
"step": 144500
|
| 2199 |
+
},
|
| 2200 |
+
{
|
| 2201 |
+
"epoch": 12.67,
|
| 2202 |
+
"learning_rate": 3.834703947368421e-05,
|
| 2203 |
+
"loss": 1.2445,
|
| 2204 |
+
"step": 145000
|
| 2205 |
+
},
|
| 2206 |
+
{
|
| 2207 |
+
"epoch": 12.67,
|
| 2208 |
+
"eval_loss": 1.200273036956787,
|
| 2209 |
+
"eval_runtime": 795.7818,
|
| 2210 |
+
"eval_samples_per_second": 96.881,
|
| 2211 |
+
"eval_steps_per_second": 6.056,
|
| 2212 |
+
"step": 145000
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 12.71,
|
| 2216 |
+
"learning_rate": 3.8118603801169596e-05,
|
| 2217 |
+
"loss": 1.246,
|
| 2218 |
+
"step": 145500
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 12.76,
|
| 2222 |
+
"learning_rate": 3.789016812865497e-05,
|
| 2223 |
+
"loss": 1.2457,
|
| 2224 |
+
"step": 146000
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 12.8,
|
| 2228 |
+
"learning_rate": 3.7661732456140354e-05,
|
| 2229 |
+
"loss": 1.2365,
|
| 2230 |
+
"step": 146500
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"epoch": 12.85,
|
| 2234 |
+
"learning_rate": 3.743375365497076e-05,
|
| 2235 |
+
"loss": 1.2353,
|
| 2236 |
+
"step": 147000
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 12.89,
|
| 2240 |
+
"learning_rate": 3.720531798245614e-05,
|
| 2241 |
+
"loss": 1.2314,
|
| 2242 |
+
"step": 147500
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 12.89,
|
| 2246 |
+
"eval_loss": 1.195684790611267,
|
| 2247 |
+
"eval_runtime": 798.8193,
|
| 2248 |
+
"eval_samples_per_second": 96.512,
|
| 2249 |
+
"eval_steps_per_second": 6.033,
|
| 2250 |
+
"step": 147500
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 12.93,
|
| 2254 |
+
"learning_rate": 3.6976882309941525e-05,
|
| 2255 |
+
"loss": 1.2283,
|
| 2256 |
+
"step": 148000
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 12.98,
|
| 2260 |
+
"learning_rate": 3.67484466374269e-05,
|
| 2261 |
+
"loss": 1.2423,
|
| 2262 |
+
"step": 148500
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 13.02,
|
| 2266 |
+
"learning_rate": 3.652001096491228e-05,
|
| 2267 |
+
"loss": 1.2213,
|
| 2268 |
+
"step": 149000
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 13.06,
|
| 2272 |
+
"learning_rate": 3.629203216374269e-05,
|
| 2273 |
+
"loss": 1.2237,
|
| 2274 |
+
"step": 149500
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 13.11,
|
| 2278 |
+
"learning_rate": 3.606359649122807e-05,
|
| 2279 |
+
"loss": 1.2188,
|
| 2280 |
+
"step": 150000
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 13.11,
|
| 2284 |
+
"eval_loss": 1.1842824220657349,
|
| 2285 |
+
"eval_runtime": 798.645,
|
| 2286 |
+
"eval_samples_per_second": 96.534,
|
| 2287 |
+
"eval_steps_per_second": 6.034,
|
| 2288 |
+
"step": 150000
|
| 2289 |
+
},
|
| 2290 |
+
{
|
| 2291 |
+
"epoch": 13.15,
|
| 2292 |
+
"learning_rate": 3.5835160818713454e-05,
|
| 2293 |
+
"loss": 1.2147,
|
| 2294 |
+
"step": 150500
|
| 2295 |
+
},
|
| 2296 |
+
{
|
| 2297 |
+
"epoch": 13.19,
|
| 2298 |
+
"learning_rate": 3.560672514619883e-05,
|
| 2299 |
+
"loss": 1.2252,
|
| 2300 |
+
"step": 151000
|
| 2301 |
+
},
|
| 2302 |
+
{
|
| 2303 |
+
"epoch": 13.24,
|
| 2304 |
+
"learning_rate": 3.537874634502924e-05,
|
| 2305 |
+
"loss": 1.2265,
|
| 2306 |
+
"step": 151500
|
| 2307 |
+
},
|
| 2308 |
+
{
|
| 2309 |
+
"epoch": 13.28,
|
| 2310 |
+
"learning_rate": 3.515031067251462e-05,
|
| 2311 |
+
"loss": 1.2158,
|
| 2312 |
+
"step": 152000
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 13.33,
|
| 2316 |
+
"learning_rate": 3.492187500000001e-05,
|
| 2317 |
+
"loss": 1.2071,
|
| 2318 |
+
"step": 152500
|
| 2319 |
+
},
|
| 2320 |
+
{
|
| 2321 |
+
"epoch": 13.33,
|
| 2322 |
+
"eval_loss": 1.1805495023727417,
|
| 2323 |
+
"eval_runtime": 799.0741,
|
| 2324 |
+
"eval_samples_per_second": 96.482,
|
| 2325 |
+
"eval_steps_per_second": 6.031,
|
| 2326 |
+
"step": 152500
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 13.37,
|
| 2330 |
+
"learning_rate": 3.469343932748538e-05,
|
| 2331 |
+
"loss": 1.2199,
|
| 2332 |
+
"step": 153000
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 13.41,
|
| 2336 |
+
"learning_rate": 3.4465003654970765e-05,
|
| 2337 |
+
"loss": 1.2059,
|
| 2338 |
+
"step": 153500
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 13.46,
|
| 2342 |
+
"learning_rate": 3.42374817251462e-05,
|
| 2343 |
+
"loss": 1.2106,
|
| 2344 |
+
"step": 154000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 13.5,
|
| 2348 |
+
"learning_rate": 3.400904605263158e-05,
|
| 2349 |
+
"loss": 1.2129,
|
| 2350 |
+
"step": 154500
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 13.54,
|
| 2354 |
+
"learning_rate": 3.378061038011696e-05,
|
| 2355 |
+
"loss": 1.2123,
|
| 2356 |
+
"step": 155000
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 13.54,
|
| 2360 |
+
"eval_loss": 1.1766316890716553,
|
| 2361 |
+
"eval_runtime": 795.4577,
|
| 2362 |
+
"eval_samples_per_second": 96.92,
|
| 2363 |
+
"eval_steps_per_second": 6.058,
|
| 2364 |
+
"step": 155000
|
| 2365 |
+
},
|
| 2366 |
+
{
|
| 2367 |
+
"epoch": 13.59,
|
| 2368 |
+
"learning_rate": 3.355217470760234e-05,
|
| 2369 |
+
"loss": 1.2058,
|
| 2370 |
+
"step": 155500
|
| 2371 |
+
},
|
| 2372 |
+
{
|
| 2373 |
+
"epoch": 13.63,
|
| 2374 |
+
"learning_rate": 3.332373903508772e-05,
|
| 2375 |
+
"loss": 1.204,
|
| 2376 |
+
"step": 156000
|
| 2377 |
+
},
|
| 2378 |
+
{
|
| 2379 |
+
"epoch": 13.68,
|
| 2380 |
+
"learning_rate": 3.30953033625731e-05,
|
| 2381 |
+
"loss": 1.1967,
|
| 2382 |
+
"step": 156500
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 13.72,
|
| 2386 |
+
"learning_rate": 3.286686769005848e-05,
|
| 2387 |
+
"loss": 1.2056,
|
| 2388 |
+
"step": 157000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 13.76,
|
| 2392 |
+
"learning_rate": 3.2638432017543865e-05,
|
| 2393 |
+
"loss": 1.2016,
|
| 2394 |
+
"step": 157500
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 13.76,
|
| 2398 |
+
"eval_loss": 1.166089653968811,
|
| 2399 |
+
"eval_runtime": 799.396,
|
| 2400 |
+
"eval_samples_per_second": 96.443,
|
| 2401 |
+
"eval_steps_per_second": 6.028,
|
| 2402 |
+
"step": 157500
|
| 2403 |
+
},
|
| 2404 |
+
{
|
| 2405 |
+
"epoch": 13.81,
|
| 2406 |
+
"learning_rate": 3.241045321637427e-05,
|
| 2407 |
+
"loss": 1.2101,
|
| 2408 |
+
"step": 158000
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 13.85,
|
| 2412 |
+
"learning_rate": 3.2182017543859646e-05,
|
| 2413 |
+
"loss": 1.2096,
|
| 2414 |
+
"step": 158500
|
| 2415 |
+
},
|
| 2416 |
+
{
|
| 2417 |
+
"epoch": 13.89,
|
| 2418 |
+
"learning_rate": 3.195358187134503e-05,
|
| 2419 |
+
"loss": 1.1916,
|
| 2420 |
+
"step": 159000
|
| 2421 |
+
},
|
| 2422 |
+
{
|
| 2423 |
+
"epoch": 13.94,
|
| 2424 |
+
"learning_rate": 3.172514619883041e-05,
|
| 2425 |
+
"loss": 1.2013,
|
| 2426 |
+
"step": 159500
|
| 2427 |
+
},
|
| 2428 |
+
{
|
| 2429 |
+
"epoch": 13.98,
|
| 2430 |
+
"learning_rate": 3.1496710526315794e-05,
|
| 2431 |
+
"loss": 1.2079,
|
| 2432 |
+
"step": 160000
|
| 2433 |
+
},
|
| 2434 |
+
{
|
| 2435 |
+
"epoch": 13.98,
|
| 2436 |
+
"eval_loss": 1.1625038385391235,
|
| 2437 |
+
"eval_runtime": 801.5699,
|
| 2438 |
+
"eval_samples_per_second": 96.181,
|
| 2439 |
+
"eval_steps_per_second": 6.012,
|
| 2440 |
+
"step": 160000
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 14.02,
|
| 2444 |
+
"learning_rate": 3.12687317251462e-05,
|
| 2445 |
+
"loss": 1.1861,
|
| 2446 |
+
"step": 160500
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 14.07,
|
| 2450 |
+
"learning_rate": 3.104029605263158e-05,
|
| 2451 |
+
"loss": 1.1892,
|
| 2452 |
+
"step": 161000
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 14.11,
|
| 2456 |
+
"learning_rate": 3.081186038011696e-05,
|
| 2457 |
+
"loss": 1.1907,
|
| 2458 |
+
"step": 161500
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 14.16,
|
| 2462 |
+
"learning_rate": 3.058342470760234e-05,
|
| 2463 |
+
"loss": 1.1865,
|
| 2464 |
+
"step": 162000
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 14.2,
|
| 2468 |
+
"learning_rate": 3.035544590643275e-05,
|
| 2469 |
+
"loss": 1.1884,
|
| 2470 |
+
"step": 162500
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 14.2,
|
| 2474 |
+
"eval_loss": 1.1524909734725952,
|
| 2475 |
+
"eval_runtime": 798.9251,
|
| 2476 |
+
"eval_samples_per_second": 96.5,
|
| 2477 |
+
"eval_steps_per_second": 6.032,
|
| 2478 |
+
"step": 162500
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 14.24,
|
| 2482 |
+
"learning_rate": 3.012701023391813e-05,
|
| 2483 |
+
"loss": 1.1843,
|
| 2484 |
+
"step": 163000
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 14.29,
|
| 2488 |
+
"learning_rate": 2.9898574561403508e-05,
|
| 2489 |
+
"loss": 1.1913,
|
| 2490 |
+
"step": 163500
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 14.33,
|
| 2494 |
+
"learning_rate": 2.9670138888888887e-05,
|
| 2495 |
+
"loss": 1.1804,
|
| 2496 |
+
"step": 164000
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 14.37,
|
| 2500 |
+
"learning_rate": 2.9441703216374273e-05,
|
| 2501 |
+
"loss": 1.1778,
|
| 2502 |
+
"step": 164500
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 14.42,
|
| 2506 |
+
"learning_rate": 2.921372441520468e-05,
|
| 2507 |
+
"loss": 1.177,
|
| 2508 |
+
"step": 165000
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 14.42,
|
| 2512 |
+
"eval_loss": 1.14187490940094,
|
| 2513 |
+
"eval_runtime": 799.1692,
|
| 2514 |
+
"eval_samples_per_second": 96.47,
|
| 2515 |
+
"eval_steps_per_second": 6.03,
|
| 2516 |
+
"step": 165000
|
| 2517 |
+
},
|
| 2518 |
+
{
|
| 2519 |
+
"epoch": 14.46,
|
| 2520 |
+
"learning_rate": 2.8985288742690058e-05,
|
| 2521 |
+
"loss": 1.1769,
|
| 2522 |
+
"step": 165500
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 14.51,
|
| 2526 |
+
"learning_rate": 2.8756853070175437e-05,
|
| 2527 |
+
"loss": 1.1869,
|
| 2528 |
+
"step": 166000
|
| 2529 |
+
},
|
| 2530 |
+
{
|
| 2531 |
+
"epoch": 14.55,
|
| 2532 |
+
"learning_rate": 2.8528417397660823e-05,
|
| 2533 |
+
"loss": 1.1713,
|
| 2534 |
+
"step": 166500
|
| 2535 |
+
},
|
| 2536 |
+
{
|
| 2537 |
+
"epoch": 14.59,
|
| 2538 |
+
"learning_rate": 2.830043859649123e-05,
|
| 2539 |
+
"loss": 1.1749,
|
| 2540 |
+
"step": 167000
|
| 2541 |
+
},
|
| 2542 |
+
{
|
| 2543 |
+
"epoch": 14.64,
|
| 2544 |
+
"learning_rate": 2.807200292397661e-05,
|
| 2545 |
+
"loss": 1.1793,
|
| 2546 |
+
"step": 167500
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 14.64,
|
| 2550 |
+
"eval_loss": 1.1453900337219238,
|
| 2551 |
+
"eval_runtime": 795.1979,
|
| 2552 |
+
"eval_samples_per_second": 96.952,
|
| 2553 |
+
"eval_steps_per_second": 6.06,
|
| 2554 |
+
"step": 167500
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 14.68,
|
| 2558 |
+
"learning_rate": 2.784356725146199e-05,
|
| 2559 |
+
"loss": 1.1743,
|
| 2560 |
+
"step": 168000
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 14.72,
|
| 2564 |
+
"learning_rate": 2.7615131578947366e-05,
|
| 2565 |
+
"loss": 1.1727,
|
| 2566 |
+
"step": 168500
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 14.77,
|
| 2570 |
+
"learning_rate": 2.738715277777778e-05,
|
| 2571 |
+
"loss": 1.1743,
|
| 2572 |
+
"step": 169000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 14.81,
|
| 2576 |
+
"learning_rate": 2.715871710526316e-05,
|
| 2577 |
+
"loss": 1.1702,
|
| 2578 |
+
"step": 169500
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 14.85,
|
| 2582 |
+
"learning_rate": 2.693028143274854e-05,
|
| 2583 |
+
"loss": 1.173,
|
| 2584 |
+
"step": 170000
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 14.85,
|
| 2588 |
+
"eval_loss": 1.1378823518753052,
|
| 2589 |
+
"eval_runtime": 799.1257,
|
| 2590 |
+
"eval_samples_per_second": 96.475,
|
| 2591 |
+
"eval_steps_per_second": 6.03,
|
| 2592 |
+
"step": 170000
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 14.9,
|
| 2596 |
+
"learning_rate": 2.670184576023392e-05,
|
| 2597 |
+
"loss": 1.1762,
|
| 2598 |
+
"step": 170500
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 14.94,
|
| 2602 |
+
"learning_rate": 2.6473410087719298e-05,
|
| 2603 |
+
"loss": 1.1578,
|
| 2604 |
+
"step": 171000
|
| 2605 |
+
},
|
| 2606 |
+
{
|
| 2607 |
+
"epoch": 14.99,
|
| 2608 |
+
"learning_rate": 2.624497441520468e-05,
|
| 2609 |
+
"loss": 1.1615,
|
| 2610 |
+
"step": 171500
|
| 2611 |
+
},
|
| 2612 |
+
{
|
| 2613 |
+
"epoch": 15.03,
|
| 2614 |
+
"learning_rate": 2.6017452485380116e-05,
|
| 2615 |
+
"loss": 1.1529,
|
| 2616 |
+
"step": 172000
|
| 2617 |
+
},
|
| 2618 |
+
{
|
| 2619 |
+
"epoch": 15.07,
|
| 2620 |
+
"learning_rate": 2.5789016812865495e-05,
|
| 2621 |
+
"loss": 1.1502,
|
| 2622 |
+
"step": 172500
|
| 2623 |
+
},
|
| 2624 |
+
{
|
| 2625 |
+
"epoch": 15.07,
|
| 2626 |
+
"eval_loss": 1.1370670795440674,
|
| 2627 |
+
"eval_runtime": 798.9647,
|
| 2628 |
+
"eval_samples_per_second": 96.495,
|
| 2629 |
+
"eval_steps_per_second": 6.032,
|
| 2630 |
+
"step": 172500
|
| 2631 |
+
},
|
| 2632 |
+
{
|
| 2633 |
+
"epoch": 15.12,
|
| 2634 |
+
"learning_rate": 2.556058114035088e-05,
|
| 2635 |
+
"loss": 1.1504,
|
| 2636 |
+
"step": 173000
|
| 2637 |
+
},
|
| 2638 |
+
{
|
| 2639 |
+
"epoch": 15.16,
|
| 2640 |
+
"learning_rate": 2.533214546783626e-05,
|
| 2641 |
+
"loss": 1.1534,
|
| 2642 |
+
"step": 173500
|
| 2643 |
+
},
|
| 2644 |
+
{
|
| 2645 |
+
"epoch": 15.2,
|
| 2646 |
+
"learning_rate": 2.510370979532164e-05,
|
| 2647 |
+
"loss": 1.1597,
|
| 2648 |
+
"step": 174000
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 15.25,
|
| 2652 |
+
"learning_rate": 2.4875730994152045e-05,
|
| 2653 |
+
"loss": 1.1549,
|
| 2654 |
+
"step": 174500
|
| 2655 |
+
},
|
| 2656 |
+
{
|
| 2657 |
+
"epoch": 15.29,
|
| 2658 |
+
"learning_rate": 2.4647295321637428e-05,
|
| 2659 |
+
"loss": 1.1504,
|
| 2660 |
+
"step": 175000
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 15.29,
|
| 2664 |
+
"eval_loss": 1.1295256614685059,
|
| 2665 |
+
"eval_runtime": 798.4053,
|
| 2666 |
+
"eval_samples_per_second": 96.562,
|
| 2667 |
+
"eval_steps_per_second": 6.036,
|
| 2668 |
+
"step": 175000
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"epoch": 15.34,
|
| 2672 |
+
"learning_rate": 2.4418859649122807e-05,
|
| 2673 |
+
"loss": 1.1583,
|
| 2674 |
+
"step": 175500
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 15.38,
|
| 2678 |
+
"learning_rate": 2.4190423976608186e-05,
|
| 2679 |
+
"loss": 1.153,
|
| 2680 |
+
"step": 176000
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 15.42,
|
| 2684 |
+
"learning_rate": 2.3962445175438598e-05,
|
| 2685 |
+
"loss": 1.1572,
|
| 2686 |
+
"step": 176500
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 15.47,
|
| 2690 |
+
"learning_rate": 2.3734009502923977e-05,
|
| 2691 |
+
"loss": 1.1526,
|
| 2692 |
+
"step": 177000
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 15.51,
|
| 2696 |
+
"learning_rate": 2.3505573830409357e-05,
|
| 2697 |
+
"loss": 1.146,
|
| 2698 |
+
"step": 177500
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 15.51,
|
| 2702 |
+
"eval_loss": 1.1202627420425415,
|
| 2703 |
+
"eval_runtime": 793.634,
|
| 2704 |
+
"eval_samples_per_second": 97.143,
|
| 2705 |
+
"eval_steps_per_second": 6.072,
|
| 2706 |
+
"step": 177500
|
| 2707 |
+
},
|
| 2708 |
+
{
|
| 2709 |
+
"epoch": 15.55,
|
| 2710 |
+
"learning_rate": 2.3277138157894736e-05,
|
| 2711 |
+
"loss": 1.1595,
|
| 2712 |
+
"step": 178000
|
| 2713 |
+
},
|
| 2714 |
+
{
|
| 2715 |
+
"epoch": 15.6,
|
| 2716 |
+
"learning_rate": 2.304870248538012e-05,
|
| 2717 |
+
"loss": 1.1496,
|
| 2718 |
+
"step": 178500
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 15.64,
|
| 2722 |
+
"learning_rate": 2.2820266812865498e-05,
|
| 2723 |
+
"loss": 1.1466,
|
| 2724 |
+
"step": 179000
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 15.69,
|
| 2728 |
+
"learning_rate": 2.2591831140350877e-05,
|
| 2729 |
+
"loss": 1.1483,
|
| 2730 |
+
"step": 179500
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 15.73,
|
| 2734 |
+
"learning_rate": 2.236339546783626e-05,
|
| 2735 |
+
"loss": 1.1487,
|
| 2736 |
+
"step": 180000
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 15.73,
|
| 2740 |
+
"eval_loss": 1.1136623620986938,
|
| 2741 |
+
"eval_runtime": 798.4421,
|
| 2742 |
+
"eval_samples_per_second": 96.558,
|
| 2743 |
+
"eval_steps_per_second": 6.036,
|
| 2744 |
+
"step": 180000
|
| 2745 |
+
},
|
| 2746 |
+
{
|
| 2747 |
+
"epoch": 15.77,
|
| 2748 |
+
"learning_rate": 2.2135416666666668e-05,
|
| 2749 |
+
"loss": 1.1383,
|
| 2750 |
+
"step": 180500
|
| 2751 |
+
},
|
| 2752 |
+
{
|
| 2753 |
+
"epoch": 15.82,
|
| 2754 |
+
"learning_rate": 2.1906980994152047e-05,
|
| 2755 |
+
"loss": 1.1463,
|
| 2756 |
+
"step": 181000
|
| 2757 |
+
},
|
| 2758 |
+
{
|
| 2759 |
+
"epoch": 15.86,
|
| 2760 |
+
"learning_rate": 2.1678545321637427e-05,
|
| 2761 |
+
"loss": 1.1351,
|
| 2762 |
+
"step": 181500
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 15.9,
|
| 2766 |
+
"learning_rate": 2.1450109649122806e-05,
|
| 2767 |
+
"loss": 1.1432,
|
| 2768 |
+
"step": 182000
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 15.95,
|
| 2772 |
+
"learning_rate": 2.122167397660819e-05,
|
| 2773 |
+
"loss": 1.1329,
|
| 2774 |
+
"step": 182500
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 15.95,
|
| 2778 |
+
"eval_loss": 1.1195679903030396,
|
| 2779 |
+
"eval_runtime": 798.9095,
|
| 2780 |
+
"eval_samples_per_second": 96.502,
|
| 2781 |
+
"eval_steps_per_second": 6.032,
|
| 2782 |
+
"step": 182500
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 15.99,
|
| 2786 |
+
"learning_rate": 2.0993695175438597e-05,
|
| 2787 |
+
"loss": 1.1394,
|
| 2788 |
+
"step": 183000
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 16.03,
|
| 2792 |
+
"learning_rate": 2.0765716374269006e-05,
|
| 2793 |
+
"loss": 1.1325,
|
| 2794 |
+
"step": 183500
|
| 2795 |
+
},
|
| 2796 |
+
{
|
| 2797 |
+
"epoch": 16.08,
|
| 2798 |
+
"learning_rate": 2.0537280701754385e-05,
|
| 2799 |
+
"loss": 1.1326,
|
| 2800 |
+
"step": 184000
|
| 2801 |
+
},
|
| 2802 |
+
{
|
| 2803 |
+
"epoch": 16.12,
|
| 2804 |
+
"learning_rate": 2.0308845029239768e-05,
|
| 2805 |
+
"loss": 1.1353,
|
| 2806 |
+
"step": 184500
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 16.17,
|
| 2810 |
+
"learning_rate": 2.0080409356725147e-05,
|
| 2811 |
+
"loss": 1.1259,
|
| 2812 |
+
"step": 185000
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 16.17,
|
| 2816 |
+
"eval_loss": 1.1075031757354736,
|
| 2817 |
+
"eval_runtime": 793.7166,
|
| 2818 |
+
"eval_samples_per_second": 97.133,
|
| 2819 |
+
"eval_steps_per_second": 6.071,
|
| 2820 |
+
"step": 185000
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 16.21,
|
| 2824 |
+
"learning_rate": 1.985197368421053e-05,
|
| 2825 |
+
"loss": 1.129,
|
| 2826 |
+
"step": 185500
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 16.25,
|
| 2830 |
+
"learning_rate": 1.962353801169591e-05,
|
| 2831 |
+
"loss": 1.1308,
|
| 2832 |
+
"step": 186000
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 16.3,
|
| 2836 |
+
"learning_rate": 1.9395102339181288e-05,
|
| 2837 |
+
"loss": 1.1335,
|
| 2838 |
+
"step": 186500
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 16.34,
|
| 2842 |
+
"learning_rate": 1.9166666666666667e-05,
|
| 2843 |
+
"loss": 1.1374,
|
| 2844 |
+
"step": 187000
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 16.38,
|
| 2848 |
+
"learning_rate": 1.8938687865497076e-05,
|
| 2849 |
+
"loss": 1.1287,
|
| 2850 |
+
"step": 187500
|
| 2851 |
+
},
|
| 2852 |
+
{
|
| 2853 |
+
"epoch": 16.38,
|
| 2854 |
+
"eval_loss": 1.1037468910217285,
|
| 2855 |
+
"eval_runtime": 802.495,
|
| 2856 |
+
"eval_samples_per_second": 96.07,
|
| 2857 |
+
"eval_steps_per_second": 6.005,
|
| 2858 |
+
"step": 187500
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 16.43,
|
| 2862 |
+
"learning_rate": 1.871025219298246e-05,
|
| 2863 |
+
"loss": 1.1144,
|
| 2864 |
+
"step": 188000
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 16.47,
|
| 2868 |
+
"learning_rate": 1.8481816520467838e-05,
|
| 2869 |
+
"loss": 1.1244,
|
| 2870 |
+
"step": 188500
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 16.52,
|
| 2874 |
+
"learning_rate": 1.8253380847953217e-05,
|
| 2875 |
+
"loss": 1.1198,
|
| 2876 |
+
"step": 189000
|
| 2877 |
+
},
|
| 2878 |
+
{
|
| 2879 |
+
"epoch": 16.56,
|
| 2880 |
+
"learning_rate": 1.80249451754386e-05,
|
| 2881 |
+
"loss": 1.118,
|
| 2882 |
+
"step": 189500
|
| 2883 |
+
},
|
| 2884 |
+
{
|
| 2885 |
+
"epoch": 16.6,
|
| 2886 |
+
"learning_rate": 1.7796966374269005e-05,
|
| 2887 |
+
"loss": 1.126,
|
| 2888 |
+
"step": 190000
|
| 2889 |
+
},
|
| 2890 |
+
{
|
| 2891 |
+
"epoch": 16.6,
|
| 2892 |
+
"eval_loss": 1.1041918992996216,
|
| 2893 |
+
"eval_runtime": 798.7101,
|
| 2894 |
+
"eval_samples_per_second": 96.526,
|
| 2895 |
+
"eval_steps_per_second": 6.033,
|
| 2896 |
+
"step": 190000
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"epoch": 16.65,
|
| 2900 |
+
"learning_rate": 1.7568987573099414e-05,
|
| 2901 |
+
"loss": 1.115,
|
| 2902 |
+
"step": 190500
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 16.69,
|
| 2906 |
+
"learning_rate": 1.7340551900584793e-05,
|
| 2907 |
+
"loss": 1.1211,
|
| 2908 |
+
"step": 191000
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 16.73,
|
| 2912 |
+
"learning_rate": 1.7112116228070176e-05,
|
| 2913 |
+
"loss": 1.1191,
|
| 2914 |
+
"step": 191500
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 16.78,
|
| 2918 |
+
"learning_rate": 1.6883680555555555e-05,
|
| 2919 |
+
"loss": 1.1095,
|
| 2920 |
+
"step": 192000
|
| 2921 |
+
},
|
| 2922 |
+
{
|
| 2923 |
+
"epoch": 16.82,
|
| 2924 |
+
"learning_rate": 1.6655244883040934e-05,
|
| 2925 |
+
"loss": 1.1199,
|
| 2926 |
+
"step": 192500
|
| 2927 |
+
},
|
| 2928 |
+
{
|
| 2929 |
+
"epoch": 16.82,
|
| 2930 |
+
"eval_loss": 1.0953432321548462,
|
| 2931 |
+
"eval_runtime": 793.9545,
|
| 2932 |
+
"eval_samples_per_second": 97.104,
|
| 2933 |
+
"eval_steps_per_second": 6.07,
|
| 2934 |
+
"step": 192500
|
| 2935 |
+
},
|
| 2936 |
+
{
|
| 2937 |
+
"epoch": 16.86,
|
| 2938 |
+
"learning_rate": 1.6426809210526317e-05,
|
| 2939 |
+
"loss": 1.1304,
|
| 2940 |
+
"step": 193000
|
| 2941 |
+
},
|
| 2942 |
+
{
|
| 2943 |
+
"epoch": 16.91,
|
| 2944 |
+
"learning_rate": 1.6198373538011696e-05,
|
| 2945 |
+
"loss": 1.1208,
|
| 2946 |
+
"step": 193500
|
| 2947 |
+
},
|
| 2948 |
+
{
|
| 2949 |
+
"epoch": 16.95,
|
| 2950 |
+
"learning_rate": 1.5969937865497075e-05,
|
| 2951 |
+
"loss": 1.1123,
|
| 2952 |
+
"step": 194000
|
| 2953 |
+
},
|
| 2954 |
+
{
|
| 2955 |
+
"epoch": 17.0,
|
| 2956 |
+
"learning_rate": 1.5741502192982458e-05,
|
| 2957 |
+
"loss": 1.1146,
|
| 2958 |
+
"step": 194500
|
| 2959 |
+
},
|
| 2960 |
+
{
|
| 2961 |
+
"epoch": 17.04,
|
| 2962 |
+
"learning_rate": 1.5513523391812867e-05,
|
| 2963 |
+
"loss": 1.1072,
|
| 2964 |
+
"step": 195000
|
| 2965 |
+
},
|
| 2966 |
+
{
|
| 2967 |
+
"epoch": 17.04,
|
| 2968 |
+
"eval_loss": 1.0885491371154785,
|
| 2969 |
+
"eval_runtime": 799.287,
|
| 2970 |
+
"eval_samples_per_second": 96.456,
|
| 2971 |
+
"eval_steps_per_second": 6.029,
|
| 2972 |
+
"step": 195000
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 17.08,
|
| 2976 |
+
"learning_rate": 1.5285544590643276e-05,
|
| 2977 |
+
"loss": 1.1109,
|
| 2978 |
+
"step": 195500
|
| 2979 |
+
},
|
| 2980 |
+
{
|
| 2981 |
+
"epoch": 17.13,
|
| 2982 |
+
"learning_rate": 1.5057108918128656e-05,
|
| 2983 |
+
"loss": 1.1012,
|
| 2984 |
+
"step": 196000
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 17.17,
|
| 2988 |
+
"learning_rate": 1.4828673245614036e-05,
|
| 2989 |
+
"loss": 1.112,
|
| 2990 |
+
"step": 196500
|
| 2991 |
+
},
|
| 2992 |
+
{
|
| 2993 |
+
"epoch": 17.21,
|
| 2994 |
+
"learning_rate": 1.4600237573099415e-05,
|
| 2995 |
+
"loss": 1.1014,
|
| 2996 |
+
"step": 197000
|
| 2997 |
+
},
|
| 2998 |
+
{
|
| 2999 |
+
"epoch": 17.26,
|
| 3000 |
+
"learning_rate": 1.4371801900584797e-05,
|
| 3001 |
+
"loss": 1.1043,
|
| 3002 |
+
"step": 197500
|
| 3003 |
+
},
|
| 3004 |
+
{
|
| 3005 |
+
"epoch": 17.26,
|
| 3006 |
+
"eval_loss": 1.0877279043197632,
|
| 3007 |
+
"eval_runtime": 799.1548,
|
| 3008 |
+
"eval_samples_per_second": 96.472,
|
| 3009 |
+
"eval_steps_per_second": 6.03,
|
| 3010 |
+
"step": 197500
|
| 3011 |
+
},
|
| 3012 |
+
{
|
| 3013 |
+
"epoch": 17.3,
|
| 3014 |
+
"learning_rate": 1.4143366228070177e-05,
|
| 3015 |
+
"loss": 1.0985,
|
| 3016 |
+
"step": 198000
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 17.35,
|
| 3020 |
+
"learning_rate": 1.3914930555555556e-05,
|
| 3021 |
+
"loss": 1.1058,
|
| 3022 |
+
"step": 198500
|
| 3023 |
+
},
|
| 3024 |
+
{
|
| 3025 |
+
"epoch": 17.39,
|
| 3026 |
+
"learning_rate": 1.3686494883040937e-05,
|
| 3027 |
+
"loss": 1.0983,
|
| 3028 |
+
"step": 199000
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 17.43,
|
| 3032 |
+
"learning_rate": 1.3458516081871344e-05,
|
| 3033 |
+
"loss": 1.1015,
|
| 3034 |
+
"step": 199500
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 17.48,
|
| 3038 |
+
"learning_rate": 1.3230080409356726e-05,
|
| 3039 |
+
"loss": 1.1007,
|
| 3040 |
+
"step": 200000
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 17.48,
|
| 3044 |
+
"eval_loss": 1.0834912061691284,
|
| 3045 |
+
"eval_runtime": 795.2591,
|
| 3046 |
+
"eval_samples_per_second": 96.945,
|
| 3047 |
+
"eval_steps_per_second": 6.06,
|
| 3048 |
+
"step": 200000
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 17.52,
|
| 3052 |
+
"learning_rate": 1.3001644736842106e-05,
|
| 3053 |
+
"loss": 1.1048,
|
| 3054 |
+
"step": 200500
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 17.56,
|
| 3058 |
+
"learning_rate": 1.2773209064327485e-05,
|
| 3059 |
+
"loss": 1.1018,
|
| 3060 |
+
"step": 201000
|
| 3061 |
+
},
|
| 3062 |
+
{
|
| 3063 |
+
"epoch": 17.61,
|
| 3064 |
+
"learning_rate": 1.2544773391812867e-05,
|
| 3065 |
+
"loss": 1.1071,
|
| 3066 |
+
"step": 201500
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 17.65,
|
| 3070 |
+
"learning_rate": 1.2316794590643276e-05,
|
| 3071 |
+
"loss": 1.1002,
|
| 3072 |
+
"step": 202000
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 17.69,
|
| 3076 |
+
"learning_rate": 1.2088358918128655e-05,
|
| 3077 |
+
"loss": 1.0879,
|
| 3078 |
+
"step": 202500
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 17.69,
|
| 3082 |
+
"eval_loss": 1.0818740129470825,
|
| 3083 |
+
"eval_runtime": 801.1636,
|
| 3084 |
+
"eval_samples_per_second": 96.23,
|
| 3085 |
+
"eval_steps_per_second": 6.015,
|
| 3086 |
+
"step": 202500
|
| 3087 |
+
},
|
| 3088 |
+
{
|
| 3089 |
+
"epoch": 17.74,
|
| 3090 |
+
"learning_rate": 1.1859923245614035e-05,
|
| 3091 |
+
"loss": 1.097,
|
| 3092 |
+
"step": 203000
|
| 3093 |
+
},
|
| 3094 |
+
{
|
| 3095 |
+
"epoch": 17.78,
|
| 3096 |
+
"learning_rate": 1.1631487573099415e-05,
|
| 3097 |
+
"loss": 1.0993,
|
| 3098 |
+
"step": 203500
|
| 3099 |
+
},
|
| 3100 |
+
{
|
| 3101 |
+
"epoch": 17.83,
|
| 3102 |
+
"learning_rate": 1.1403051900584796e-05,
|
| 3103 |
+
"loss": 1.0991,
|
| 3104 |
+
"step": 204000
|
| 3105 |
+
},
|
| 3106 |
+
{
|
| 3107 |
+
"epoch": 17.87,
|
| 3108 |
+
"learning_rate": 1.1174616228070176e-05,
|
| 3109 |
+
"loss": 1.0953,
|
| 3110 |
+
"step": 204500
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 17.91,
|
| 3114 |
+
"learning_rate": 1.0946637426900586e-05,
|
| 3115 |
+
"loss": 1.1,
|
| 3116 |
+
"step": 205000
|
| 3117 |
+
},
|
| 3118 |
+
{
|
| 3119 |
+
"epoch": 17.91,
|
| 3120 |
+
"eval_loss": 1.0743597745895386,
|
| 3121 |
+
"eval_runtime": 798.4761,
|
| 3122 |
+
"eval_samples_per_second": 96.554,
|
| 3123 |
+
"eval_steps_per_second": 6.035,
|
| 3124 |
+
"step": 205000
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 17.96,
|
| 3128 |
+
"learning_rate": 1.0718201754385965e-05,
|
| 3129 |
+
"loss": 1.0917,
|
| 3130 |
+
"step": 205500
|
| 3131 |
+
},
|
| 3132 |
+
{
|
| 3133 |
+
"epoch": 18.0,
|
| 3134 |
+
"learning_rate": 1.0489766081871345e-05,
|
| 3135 |
+
"loss": 1.0929,
|
| 3136 |
+
"step": 206000
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 18.04,
|
| 3140 |
+
"learning_rate": 1.0261330409356725e-05,
|
| 3141 |
+
"loss": 1.089,
|
| 3142 |
+
"step": 206500
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 18.09,
|
| 3146 |
+
"learning_rate": 1.0033351608187134e-05,
|
| 3147 |
+
"loss": 1.0919,
|
| 3148 |
+
"step": 207000
|
| 3149 |
+
},
|
| 3150 |
+
{
|
| 3151 |
+
"epoch": 18.13,
|
| 3152 |
+
"learning_rate": 9.804915935672515e-06,
|
| 3153 |
+
"loss": 1.0863,
|
| 3154 |
+
"step": 207500
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 18.13,
|
| 3158 |
+
"eval_loss": 1.0773507356643677,
|
| 3159 |
+
"eval_runtime": 798.9691,
|
| 3160 |
+
"eval_samples_per_second": 96.494,
|
| 3161 |
+
"eval_steps_per_second": 6.032,
|
| 3162 |
+
"step": 207500
|
| 3163 |
+
},
|
| 3164 |
+
{
|
| 3165 |
+
"epoch": 18.18,
|
| 3166 |
+
"learning_rate": 9.576480263157896e-06,
|
| 3167 |
+
"loss": 1.0895,
|
| 3168 |
+
"step": 208000
|
| 3169 |
+
},
|
| 3170 |
+
{
|
| 3171 |
+
"epoch": 18.22,
|
| 3172 |
+
"learning_rate": 9.348044590643275e-06,
|
| 3173 |
+
"loss": 1.0782,
|
| 3174 |
+
"step": 208500
|
| 3175 |
+
},
|
| 3176 |
+
{
|
| 3177 |
+
"epoch": 18.26,
|
| 3178 |
+
"learning_rate": 9.120065789473684e-06,
|
| 3179 |
+
"loss": 1.0847,
|
| 3180 |
+
"step": 209000
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 18.31,
|
| 3184 |
+
"learning_rate": 8.891630116959063e-06,
|
| 3185 |
+
"loss": 1.0819,
|
| 3186 |
+
"step": 209500
|
| 3187 |
+
},
|
| 3188 |
+
{
|
| 3189 |
+
"epoch": 18.35,
|
| 3190 |
+
"learning_rate": 8.663194444444444e-06,
|
| 3191 |
+
"loss": 1.087,
|
| 3192 |
+
"step": 210000
|
| 3193 |
+
},
|
| 3194 |
+
{
|
| 3195 |
+
"epoch": 18.35,
|
| 3196 |
+
"eval_loss": 1.0759390592575073,
|
| 3197 |
+
"eval_runtime": 799.2137,
|
| 3198 |
+
"eval_samples_per_second": 96.465,
|
| 3199 |
+
"eval_steps_per_second": 6.03,
|
| 3200 |
+
"step": 210000
|
| 3201 |
+
},
|
| 3202 |
+
{
|
| 3203 |
+
"epoch": 18.39,
|
| 3204 |
+
"learning_rate": 8.434758771929825e-06,
|
| 3205 |
+
"loss": 1.0857,
|
| 3206 |
+
"step": 210500
|
| 3207 |
+
},
|
| 3208 |
+
{
|
| 3209 |
+
"epoch": 18.44,
|
| 3210 |
+
"learning_rate": 8.206779970760234e-06,
|
| 3211 |
+
"loss": 1.0831,
|
| 3212 |
+
"step": 211000
|
| 3213 |
+
},
|
| 3214 |
+
{
|
| 3215 |
+
"epoch": 18.48,
|
| 3216 |
+
"learning_rate": 7.978344298245615e-06,
|
| 3217 |
+
"loss": 1.0752,
|
| 3218 |
+
"step": 211500
|
| 3219 |
+
},
|
| 3220 |
+
{
|
| 3221 |
+
"epoch": 18.52,
|
| 3222 |
+
"learning_rate": 7.749908625730996e-06,
|
| 3223 |
+
"loss": 1.0865,
|
| 3224 |
+
"step": 212000
|
| 3225 |
+
},
|
| 3226 |
+
{
|
| 3227 |
+
"epoch": 18.57,
|
| 3228 |
+
"learning_rate": 7.521472953216374e-06,
|
| 3229 |
+
"loss": 1.0755,
|
| 3230 |
+
"step": 212500
|
| 3231 |
+
},
|
| 3232 |
+
{
|
| 3233 |
+
"epoch": 18.57,
|
| 3234 |
+
"eval_loss": 1.061818242073059,
|
| 3235 |
+
"eval_runtime": 795.7294,
|
| 3236 |
+
"eval_samples_per_second": 96.887,
|
| 3237 |
+
"eval_steps_per_second": 6.056,
|
| 3238 |
+
"step": 212500
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 18.61,
|
| 3242 |
+
"learning_rate": 7.293037280701755e-06,
|
| 3243 |
+
"loss": 1.0806,
|
| 3244 |
+
"step": 213000
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 18.66,
|
| 3248 |
+
"learning_rate": 7.064601608187135e-06,
|
| 3249 |
+
"loss": 1.0826,
|
| 3250 |
+
"step": 213500
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 18.7,
|
| 3254 |
+
"learning_rate": 6.836165935672514e-06,
|
| 3255 |
+
"loss": 1.0854,
|
| 3256 |
+
"step": 214000
|
| 3257 |
+
},
|
| 3258 |
+
{
|
| 3259 |
+
"epoch": 18.74,
|
| 3260 |
+
"learning_rate": 6.607730263157895e-06,
|
| 3261 |
+
"loss": 1.0783,
|
| 3262 |
+
"step": 214500
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 18.79,
|
| 3266 |
+
"learning_rate": 6.379294590643276e-06,
|
| 3267 |
+
"loss": 1.0832,
|
| 3268 |
+
"step": 215000
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 18.79,
|
| 3272 |
+
"eval_loss": 1.0627787113189697,
|
| 3273 |
+
"eval_runtime": 799.1419,
|
| 3274 |
+
"eval_samples_per_second": 96.473,
|
| 3275 |
+
"eval_steps_per_second": 6.03,
|
| 3276 |
+
"step": 215000
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 18.83,
|
| 3280 |
+
"learning_rate": 6.151315789473684e-06,
|
| 3281 |
+
"loss": 1.0786,
|
| 3282 |
+
"step": 215500
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 18.87,
|
| 3286 |
+
"learning_rate": 5.922880116959065e-06,
|
| 3287 |
+
"loss": 1.0719,
|
| 3288 |
+
"step": 216000
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 18.92,
|
| 3292 |
+
"learning_rate": 5.694444444444445e-06,
|
| 3293 |
+
"loss": 1.0768,
|
| 3294 |
+
"step": 216500
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 18.96,
|
| 3298 |
+
"learning_rate": 5.466008771929825e-06,
|
| 3299 |
+
"loss": 1.075,
|
| 3300 |
+
"step": 217000
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 19.01,
|
| 3304 |
+
"learning_rate": 5.237573099415205e-06,
|
| 3305 |
+
"loss": 1.0771,
|
| 3306 |
+
"step": 217500
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 19.01,
|
| 3310 |
+
"eval_loss": 1.061143398284912,
|
| 3311 |
+
"eval_runtime": 799.1588,
|
| 3312 |
+
"eval_samples_per_second": 96.471,
|
| 3313 |
+
"eval_steps_per_second": 6.03,
|
| 3314 |
+
"step": 217500
|
| 3315 |
+
},
|
| 3316 |
+
{
|
| 3317 |
+
"epoch": 19.05,
|
| 3318 |
+
"learning_rate": 5.009137426900585e-06,
|
| 3319 |
+
"loss": 1.0815,
|
| 3320 |
+
"step": 218000
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 19.09,
|
| 3324 |
+
"learning_rate": 4.781158625730994e-06,
|
| 3325 |
+
"loss": 1.0656,
|
| 3326 |
+
"step": 218500
|
| 3327 |
+
},
|
| 3328 |
+
{
|
| 3329 |
+
"epoch": 19.14,
|
| 3330 |
+
"learning_rate": 4.552722953216375e-06,
|
| 3331 |
+
"loss": 1.0718,
|
| 3332 |
+
"step": 219000
|
| 3333 |
+
},
|
| 3334 |
+
{
|
| 3335 |
+
"epoch": 19.18,
|
| 3336 |
+
"learning_rate": 4.324287280701755e-06,
|
| 3337 |
+
"loss": 1.0697,
|
| 3338 |
+
"step": 219500
|
| 3339 |
+
},
|
| 3340 |
+
{
|
| 3341 |
+
"epoch": 19.22,
|
| 3342 |
+
"learning_rate": 4.095851608187135e-06,
|
| 3343 |
+
"loss": 1.0703,
|
| 3344 |
+
"step": 220000
|
| 3345 |
+
},
|
| 3346 |
+
{
|
| 3347 |
+
"epoch": 19.22,
|
| 3348 |
+
"eval_loss": 1.0554795265197754,
|
| 3349 |
+
"eval_runtime": 793.6761,
|
| 3350 |
+
"eval_samples_per_second": 97.138,
|
| 3351 |
+
"eval_steps_per_second": 6.072,
|
| 3352 |
+
"step": 220000
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 19.27,
|
| 3356 |
+
"learning_rate": 3.867415935672515e-06,
|
| 3357 |
+
"loss": 1.0681,
|
| 3358 |
+
"step": 220500
|
| 3359 |
+
},
|
| 3360 |
+
{
|
| 3361 |
+
"epoch": 19.31,
|
| 3362 |
+
"learning_rate": 3.638980263157895e-06,
|
| 3363 |
+
"loss": 1.0706,
|
| 3364 |
+
"step": 221000
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 19.36,
|
| 3368 |
+
"learning_rate": 3.4110014619883042e-06,
|
| 3369 |
+
"loss": 1.0671,
|
| 3370 |
+
"step": 221500
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 19.4,
|
| 3374 |
+
"learning_rate": 3.1825657894736842e-06,
|
| 3375 |
+
"loss": 1.063,
|
| 3376 |
+
"step": 222000
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 19.44,
|
| 3380 |
+
"learning_rate": 2.9541301169590643e-06,
|
| 3381 |
+
"loss": 1.069,
|
| 3382 |
+
"step": 222500
|
| 3383 |
+
},
|
| 3384 |
+
{
|
| 3385 |
+
"epoch": 19.44,
|
| 3386 |
+
"eval_loss": 1.0551636219024658,
|
| 3387 |
+
"eval_runtime": 799.5142,
|
| 3388 |
+
"eval_samples_per_second": 96.429,
|
| 3389 |
+
"eval_steps_per_second": 6.027,
|
| 3390 |
+
"step": 222500
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 19.49,
|
| 3394 |
+
"learning_rate": 2.7256944444444447e-06,
|
| 3395 |
+
"loss": 1.0662,
|
| 3396 |
+
"step": 223000
|
| 3397 |
+
},
|
| 3398 |
+
{
|
| 3399 |
+
"epoch": 19.53,
|
| 3400 |
+
"learning_rate": 2.4972587719298248e-06,
|
| 3401 |
+
"loss": 1.0646,
|
| 3402 |
+
"step": 223500
|
| 3403 |
+
},
|
| 3404 |
+
{
|
| 3405 |
+
"epoch": 19.57,
|
| 3406 |
+
"learning_rate": 2.2688230994152048e-06,
|
| 3407 |
+
"loss": 1.0651,
|
| 3408 |
+
"step": 224000
|
| 3409 |
+
},
|
| 3410 |
+
{
|
| 3411 |
+
"epoch": 19.62,
|
| 3412 |
+
"learning_rate": 2.040844298245614e-06,
|
| 3413 |
+
"loss": 1.0729,
|
| 3414 |
+
"step": 224500
|
| 3415 |
+
},
|
| 3416 |
+
{
|
| 3417 |
+
"epoch": 19.66,
|
| 3418 |
+
"learning_rate": 1.8124086257309944e-06,
|
| 3419 |
+
"loss": 1.0706,
|
| 3420 |
+
"step": 225000
|
| 3421 |
+
},
|
| 3422 |
+
{
|
| 3423 |
+
"epoch": 19.66,
|
| 3424 |
+
"eval_loss": 1.0508904457092285,
|
| 3425 |
+
"eval_runtime": 799.3295,
|
| 3426 |
+
"eval_samples_per_second": 96.451,
|
| 3427 |
+
"eval_steps_per_second": 6.029,
|
| 3428 |
+
"step": 225000
|
| 3429 |
+
},
|
| 3430 |
+
{
|
| 3431 |
+
"epoch": 19.7,
|
| 3432 |
+
"learning_rate": 1.5839729532163744e-06,
|
| 3433 |
+
"loss": 1.0639,
|
| 3434 |
+
"step": 225500
|
| 3435 |
+
},
|
| 3436 |
+
{
|
| 3437 |
+
"epoch": 19.75,
|
| 3438 |
+
"learning_rate": 1.3555372807017544e-06,
|
| 3439 |
+
"loss": 1.0681,
|
| 3440 |
+
"step": 226000
|
| 3441 |
+
},
|
| 3442 |
+
{
|
| 3443 |
+
"epoch": 19.79,
|
| 3444 |
+
"learning_rate": 1.1271016081871344e-06,
|
| 3445 |
+
"loss": 1.0707,
|
| 3446 |
+
"step": 226500
|
| 3447 |
+
},
|
| 3448 |
+
{
|
| 3449 |
+
"epoch": 19.84,
|
| 3450 |
+
"learning_rate": 8.991228070175439e-07,
|
| 3451 |
+
"loss": 1.0674,
|
| 3452 |
+
"step": 227000
|
| 3453 |
+
},
|
| 3454 |
+
{
|
| 3455 |
+
"epoch": 19.88,
|
| 3456 |
+
"learning_rate": 6.70687134502924e-07,
|
| 3457 |
+
"loss": 1.0633,
|
| 3458 |
+
"step": 227500
|
| 3459 |
+
},
|
| 3460 |
+
{
|
| 3461 |
+
"epoch": 19.88,
|
| 3462 |
+
"eval_loss": 1.0465439558029175,
|
| 3463 |
+
"eval_runtime": 799.9551,
|
| 3464 |
+
"eval_samples_per_second": 96.375,
|
| 3465 |
+
"eval_steps_per_second": 6.024,
|
| 3466 |
+
"step": 227500
|
| 3467 |
+
},
|
| 3468 |
+
{
|
| 3469 |
+
"epoch": 19.92,
|
| 3470 |
+
"learning_rate": 4.4225146198830405e-07,
|
| 3471 |
+
"loss": 1.06,
|
| 3472 |
+
"step": 228000
|
| 3473 |
+
},
|
| 3474 |
+
{
|
| 3475 |
+
"epoch": 19.97,
|
| 3476 |
+
"learning_rate": 2.1381578947368424e-07,
|
| 3477 |
+
"loss": 1.0695,
|
| 3478 |
+
"step": 228500
|
| 3479 |
+
},
|
| 3480 |
+
{
|
| 3481 |
+
"epoch": 20.0,
|
| 3482 |
+
"step": 228880,
|
| 3483 |
+
"total_flos": 2.8703856730616376e+18,
|
| 3484 |
+
"train_loss": 0.5992164485148723,
|
| 3485 |
+
"train_runtime": 340057.0464,
|
| 3486 |
+
"train_samples_per_second": 86.151,
|
| 3487 |
+
"train_steps_per_second": 0.673
|
| 3488 |
+
}
|
| 3489 |
+
],
|
| 3490 |
+
"max_steps": 228880,
|
| 3491 |
+
"num_train_epochs": 20,
|
| 3492 |
+
"total_flos": 2.8703856730616376e+18,
|
| 3493 |
+
"trial_name": null,
|
| 3494 |
+
"trial_params": null
|
| 3495 |
+
}
|