Commit
•
c47c2d0
1
Parent(s):
0d29f0d
Upload trainer_state.json
Browse files- trainer_state.json +1681 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1681 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.4481608271598816,
|
| 3 |
+
"best_model_checkpoint": "./distilbert-marian-training-1/checkpoint-123440",
|
| 4 |
+
"epoch": 10.0,
|
| 5 |
+
"global_step": 123440,
|
| 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": 8.068697342838627e-07,
|
| 13 |
+
"loss": 0.5266,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.08,
|
| 18 |
+
"learning_rate": 1.6169799092676605e-06,
|
| 19 |
+
"loss": 0.497,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.12,
|
| 24 |
+
"learning_rate": 2.4270900842514583e-06,
|
| 25 |
+
"loss": 0.5398,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.16,
|
| 30 |
+
"learning_rate": 3.2372002592352565e-06,
|
| 31 |
+
"loss": 0.5439,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.2,
|
| 36 |
+
"learning_rate": 4.047310434219054e-06,
|
| 37 |
+
"loss": 0.5357,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.24,
|
| 42 |
+
"learning_rate": 4.857420609202852e-06,
|
| 43 |
+
"loss": 0.5693,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.28,
|
| 48 |
+
"learning_rate": 5.66753078418665e-06,
|
| 49 |
+
"loss": 0.5311,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.32,
|
| 54 |
+
"learning_rate": 6.477640959170448e-06,
|
| 55 |
+
"loss": 0.5433,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.36,
|
| 60 |
+
"learning_rate": 7.286130913804278e-06,
|
| 61 |
+
"loss": 0.5403,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.41,
|
| 66 |
+
"learning_rate": 8.094620868438109e-06,
|
| 67 |
+
"loss": 0.5478,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.45,
|
| 72 |
+
"learning_rate": 8.904731043421905e-06,
|
| 73 |
+
"loss": 0.5556,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.49,
|
| 78 |
+
"learning_rate": 9.714841218405703e-06,
|
| 79 |
+
"loss": 0.5835,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.53,
|
| 84 |
+
"learning_rate": 9.972370979295291e-06,
|
| 85 |
+
"loss": 0.5578,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.57,
|
| 90 |
+
"learning_rate": 9.929733601664564e-06,
|
| 91 |
+
"loss": 0.5389,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.61,
|
| 96 |
+
"learning_rate": 9.887181498789099e-06,
|
| 97 |
+
"loss": 0.5645,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.65,
|
| 102 |
+
"learning_rate": 9.844544121158372e-06,
|
| 103 |
+
"loss": 0.5672,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.69,
|
| 108 |
+
"learning_rate": 9.801906743527647e-06,
|
| 109 |
+
"loss": 0.559,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.73,
|
| 114 |
+
"learning_rate": 9.759354640652181e-06,
|
| 115 |
+
"loss": 0.5927,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.77,
|
| 120 |
+
"learning_rate": 9.716717263021456e-06,
|
| 121 |
+
"loss": 0.5241,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.81,
|
| 126 |
+
"learning_rate": 9.67407988539073e-06,
|
| 127 |
+
"loss": 0.5458,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.85,
|
| 132 |
+
"learning_rate": 9.631442507760004e-06,
|
| 133 |
+
"loss": 0.5902,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.89,
|
| 138 |
+
"learning_rate": 9.588805130129278e-06,
|
| 139 |
+
"loss": 0.5425,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.93,
|
| 144 |
+
"learning_rate": 9.546167752498551e-06,
|
| 145 |
+
"loss": 0.5225,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.97,
|
| 150 |
+
"learning_rate": 9.503615649623087e-06,
|
| 151 |
+
"loss": 0.5301,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 1.0,
|
| 156 |
+
"eval_BLEU": 39.52568484782476,
|
| 157 |
+
"eval_BLEU-Bigram-Precision": 44.566237100676446,
|
| 158 |
+
"eval_BLEU-Trigram-Precision": 34.507782919646615,
|
| 159 |
+
"eval_BLEU-Unigram-Precision": 63.26198231735691,
|
| 160 |
+
"eval_ROUGE-2": 33.35721035298087,
|
| 161 |
+
"eval_ROUGE-L": 47.29001112126434,
|
| 162 |
+
"eval_Sacre-Bigram-Precision": 42.86996462699963,
|
| 163 |
+
"eval_Sacre-Trigram-Precision": 34.29733393583371,
|
| 164 |
+
"eval_Sacre-Unigram-Precision": 61.48775894538606,
|
| 165 |
+
"eval_SacreBLEU": 38.73636411085535,
|
| 166 |
+
"eval_loss": 0.7005472183227539,
|
| 167 |
+
"eval_runtime": 848.4996,
|
| 168 |
+
"eval_samples_per_second": 1.458,
|
| 169 |
+
"eval_steps_per_second": 1.458,
|
| 170 |
+
"step": 12344
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 1.01,
|
| 174 |
+
"learning_rate": 9.46097827199236e-06,
|
| 175 |
+
"loss": 0.5475,
|
| 176 |
+
"step": 12500
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 1.05,
|
| 180 |
+
"learning_rate": 9.418340894361634e-06,
|
| 181 |
+
"loss": 0.5107,
|
| 182 |
+
"step": 13000
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 1.09,
|
| 186 |
+
"learning_rate": 9.375703516730907e-06,
|
| 187 |
+
"loss": 0.5036,
|
| 188 |
+
"step": 13500
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 1.13,
|
| 192 |
+
"learning_rate": 9.333066139100182e-06,
|
| 193 |
+
"loss": 0.4865,
|
| 194 |
+
"step": 14000
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 1.17,
|
| 198 |
+
"learning_rate": 9.290428761469455e-06,
|
| 199 |
+
"loss": 0.5065,
|
| 200 |
+
"step": 14500
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 1.22,
|
| 204 |
+
"learning_rate": 9.247791383838728e-06,
|
| 205 |
+
"loss": 0.5159,
|
| 206 |
+
"step": 15000
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 1.26,
|
| 210 |
+
"learning_rate": 9.205154006208003e-06,
|
| 211 |
+
"loss": 0.5221,
|
| 212 |
+
"step": 15500
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 1.3,
|
| 216 |
+
"learning_rate": 9.16260190333254e-06,
|
| 217 |
+
"loss": 0.4597,
|
| 218 |
+
"step": 16000
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 1.34,
|
| 222 |
+
"learning_rate": 9.119964525701812e-06,
|
| 223 |
+
"loss": 0.5024,
|
| 224 |
+
"step": 16500
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 1.38,
|
| 228 |
+
"learning_rate": 9.077327148071086e-06,
|
| 229 |
+
"loss": 0.4852,
|
| 230 |
+
"step": 17000
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 1.42,
|
| 234 |
+
"learning_rate": 9.034689770440359e-06,
|
| 235 |
+
"loss": 0.4887,
|
| 236 |
+
"step": 17500
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 1.46,
|
| 240 |
+
"learning_rate": 8.992052392809634e-06,
|
| 241 |
+
"loss": 0.4796,
|
| 242 |
+
"step": 18000
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 1.5,
|
| 246 |
+
"learning_rate": 8.949415015178907e-06,
|
| 247 |
+
"loss": 0.4754,
|
| 248 |
+
"step": 18500
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 1.54,
|
| 252 |
+
"learning_rate": 8.90677763754818e-06,
|
| 253 |
+
"loss": 0.5062,
|
| 254 |
+
"step": 19000
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 1.58,
|
| 258 |
+
"learning_rate": 8.864140259917455e-06,
|
| 259 |
+
"loss": 0.5204,
|
| 260 |
+
"step": 19500
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 1.62,
|
| 264 |
+
"learning_rate": 8.821502882286729e-06,
|
| 265 |
+
"loss": 0.4799,
|
| 266 |
+
"step": 20000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 1.66,
|
| 270 |
+
"learning_rate": 8.778950779411265e-06,
|
| 271 |
+
"loss": 0.4666,
|
| 272 |
+
"step": 20500
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 1.7,
|
| 276 |
+
"learning_rate": 8.736313401780538e-06,
|
| 277 |
+
"loss": 0.4484,
|
| 278 |
+
"step": 21000
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 1.74,
|
| 282 |
+
"learning_rate": 8.693676024149813e-06,
|
| 283 |
+
"loss": 0.4721,
|
| 284 |
+
"step": 21500
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 1.78,
|
| 288 |
+
"learning_rate": 8.651038646519086e-06,
|
| 289 |
+
"loss": 0.4867,
|
| 290 |
+
"step": 22000
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 1.82,
|
| 294 |
+
"learning_rate": 8.60840126888836e-06,
|
| 295 |
+
"loss": 0.47,
|
| 296 |
+
"step": 22500
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 1.86,
|
| 300 |
+
"learning_rate": 8.565763891257632e-06,
|
| 301 |
+
"loss": 0.4848,
|
| 302 |
+
"step": 23000
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 1.9,
|
| 306 |
+
"learning_rate": 8.523126513626907e-06,
|
| 307 |
+
"loss": 0.4429,
|
| 308 |
+
"step": 23500
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 1.94,
|
| 312 |
+
"learning_rate": 8.480574410751442e-06,
|
| 313 |
+
"loss": 0.4501,
|
| 314 |
+
"step": 24000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 1.98,
|
| 318 |
+
"learning_rate": 8.437937033120715e-06,
|
| 319 |
+
"loss": 0.4363,
|
| 320 |
+
"step": 24500
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 2.0,
|
| 324 |
+
"eval_BLEU": 44.07579241900836,
|
| 325 |
+
"eval_BLEU-Bigram-Precision": 48.638266436838215,
|
| 326 |
+
"eval_BLEU-Trigram-Precision": 39.453534717851895,
|
| 327 |
+
"eval_BLEU-Unigram-Precision": 65.81561600594188,
|
| 328 |
+
"eval_ROUGE-2": 37.93490065934162,
|
| 329 |
+
"eval_ROUGE-L": 51.37484166047172,
|
| 330 |
+
"eval_Sacre-Bigram-Precision": 47.097215642928575,
|
| 331 |
+
"eval_Sacre-Trigram-Precision": 39.10595653642608,
|
| 332 |
+
"eval_Sacre-Unigram-Precision": 64.20241154378336,
|
| 333 |
+
"eval_SacreBLEU": 43.28181370216754,
|
| 334 |
+
"eval_loss": 0.6306515336036682,
|
| 335 |
+
"eval_runtime": 969.349,
|
| 336 |
+
"eval_samples_per_second": 1.276,
|
| 337 |
+
"eval_steps_per_second": 1.276,
|
| 338 |
+
"step": 24688
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 2.03,
|
| 342 |
+
"learning_rate": 8.395299655489988e-06,
|
| 343 |
+
"loss": 0.4153,
|
| 344 |
+
"step": 25000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 2.07,
|
| 348 |
+
"learning_rate": 8.352662277859263e-06,
|
| 349 |
+
"loss": 0.4241,
|
| 350 |
+
"step": 25500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 2.11,
|
| 354 |
+
"learning_rate": 8.310024900228536e-06,
|
| 355 |
+
"loss": 0.401,
|
| 356 |
+
"step": 26000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 2.15,
|
| 360 |
+
"learning_rate": 8.267387522597811e-06,
|
| 361 |
+
"loss": 0.4351,
|
| 362 |
+
"step": 26500
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 2.19,
|
| 366 |
+
"learning_rate": 8.224750144967085e-06,
|
| 367 |
+
"loss": 0.4075,
|
| 368 |
+
"step": 27000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 2.23,
|
| 372 |
+
"learning_rate": 8.18211276733636e-06,
|
| 373 |
+
"loss": 0.4106,
|
| 374 |
+
"step": 27500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 2.27,
|
| 378 |
+
"learning_rate": 8.139560664460894e-06,
|
| 379 |
+
"loss": 0.4175,
|
| 380 |
+
"step": 28000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 2.31,
|
| 384 |
+
"learning_rate": 8.096923286830167e-06,
|
| 385 |
+
"loss": 0.3897,
|
| 386 |
+
"step": 28500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 2.35,
|
| 390 |
+
"learning_rate": 8.05428590919944e-06,
|
| 391 |
+
"loss": 0.4152,
|
| 392 |
+
"step": 29000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 2.39,
|
| 396 |
+
"learning_rate": 8.011648531568715e-06,
|
| 397 |
+
"loss": 0.3917,
|
| 398 |
+
"step": 29500
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 2.43,
|
| 402 |
+
"learning_rate": 7.96909642869325e-06,
|
| 403 |
+
"loss": 0.4081,
|
| 404 |
+
"step": 30000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 2.47,
|
| 408 |
+
"learning_rate": 7.926459051062525e-06,
|
| 409 |
+
"loss": 0.3964,
|
| 410 |
+
"step": 30500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 2.51,
|
| 414 |
+
"learning_rate": 7.883821673431798e-06,
|
| 415 |
+
"loss": 0.3833,
|
| 416 |
+
"step": 31000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 2.55,
|
| 420 |
+
"learning_rate": 7.841184295801073e-06,
|
| 421 |
+
"loss": 0.3884,
|
| 422 |
+
"step": 31500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 2.59,
|
| 426 |
+
"learning_rate": 7.798546918170346e-06,
|
| 427 |
+
"loss": 0.4027,
|
| 428 |
+
"step": 32000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 2.63,
|
| 432 |
+
"learning_rate": 7.75590954053962e-06,
|
| 433 |
+
"loss": 0.4272,
|
| 434 |
+
"step": 32500
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 2.67,
|
| 438 |
+
"learning_rate": 7.713272162908894e-06,
|
| 439 |
+
"loss": 0.3868,
|
| 440 |
+
"step": 33000
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 2.71,
|
| 444 |
+
"learning_rate": 7.670634785278167e-06,
|
| 445 |
+
"loss": 0.3848,
|
| 446 |
+
"step": 33500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 2.75,
|
| 450 |
+
"learning_rate": 7.628082682402702e-06,
|
| 451 |
+
"loss": 0.3919,
|
| 452 |
+
"step": 34000
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 2.79,
|
| 456 |
+
"learning_rate": 7.585445304771976e-06,
|
| 457 |
+
"loss": 0.3803,
|
| 458 |
+
"step": 34500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 2.84,
|
| 462 |
+
"learning_rate": 7.54280792714125e-06,
|
| 463 |
+
"loss": 0.3986,
|
| 464 |
+
"step": 35000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 2.88,
|
| 468 |
+
"learning_rate": 7.500170549510524e-06,
|
| 469 |
+
"loss": 0.3722,
|
| 470 |
+
"step": 35500
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 2.92,
|
| 474 |
+
"learning_rate": 7.4576184466350585e-06,
|
| 475 |
+
"loss": 0.3795,
|
| 476 |
+
"step": 36000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 2.96,
|
| 480 |
+
"learning_rate": 7.4149810690043326e-06,
|
| 481 |
+
"loss": 0.3768,
|
| 482 |
+
"step": 36500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 3.0,
|
| 486 |
+
"learning_rate": 7.372343691373607e-06,
|
| 487 |
+
"loss": 0.3983,
|
| 488 |
+
"step": 37000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 3.0,
|
| 492 |
+
"eval_BLEU": 47.64065251911936,
|
| 493 |
+
"eval_BLEU-Bigram-Precision": 52.275310044962694,
|
| 494 |
+
"eval_BLEU-Trigram-Precision": 43.41648247552889,
|
| 495 |
+
"eval_BLEU-Unigram-Precision": 68.46246973365618,
|
| 496 |
+
"eval_ROUGE-2": 41.91844760109249,
|
| 497 |
+
"eval_ROUGE-L": 54.326234865452385,
|
| 498 |
+
"eval_Sacre-Bigram-Precision": 50.88183888565224,
|
| 499 |
+
"eval_Sacre-Trigram-Precision": 43.238494672409885,
|
| 500 |
+
"eval_Sacre-Unigram-Precision": 67.02455512476389,
|
| 501 |
+
"eval_SacreBLEU": 46.828668155215276,
|
| 502 |
+
"eval_loss": 0.5844058990478516,
|
| 503 |
+
"eval_runtime": 950.5601,
|
| 504 |
+
"eval_samples_per_second": 1.301,
|
| 505 |
+
"eval_steps_per_second": 1.301,
|
| 506 |
+
"step": 37032
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 3.04,
|
| 510 |
+
"learning_rate": 7.329706313742881e-06,
|
| 511 |
+
"loss": 0.3116,
|
| 512 |
+
"step": 37500
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 3.08,
|
| 516 |
+
"learning_rate": 7.287154210867415e-06,
|
| 517 |
+
"loss": 0.3262,
|
| 518 |
+
"step": 38000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 3.12,
|
| 522 |
+
"learning_rate": 7.244516833236689e-06,
|
| 523 |
+
"loss": 0.3429,
|
| 524 |
+
"step": 38500
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 3.16,
|
| 528 |
+
"learning_rate": 7.201879455605963e-06,
|
| 529 |
+
"loss": 0.3415,
|
| 530 |
+
"step": 39000
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 3.2,
|
| 534 |
+
"learning_rate": 7.159242077975237e-06,
|
| 535 |
+
"loss": 0.3407,
|
| 536 |
+
"step": 39500
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 3.24,
|
| 540 |
+
"learning_rate": 7.116689975099772e-06,
|
| 541 |
+
"loss": 0.3288,
|
| 542 |
+
"step": 40000
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 3.28,
|
| 546 |
+
"learning_rate": 7.074052597469046e-06,
|
| 547 |
+
"loss": 0.3499,
|
| 548 |
+
"step": 40500
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 3.32,
|
| 552 |
+
"learning_rate": 7.03141521983832e-06,
|
| 553 |
+
"loss": 0.3396,
|
| 554 |
+
"step": 41000
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 3.36,
|
| 558 |
+
"learning_rate": 6.988863116962855e-06,
|
| 559 |
+
"loss": 0.3395,
|
| 560 |
+
"step": 41500
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 3.4,
|
| 564 |
+
"learning_rate": 6.946225739332128e-06,
|
| 565 |
+
"loss": 0.3462,
|
| 566 |
+
"step": 42000
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 3.44,
|
| 570 |
+
"learning_rate": 6.9035883617014025e-06,
|
| 571 |
+
"loss": 0.3362,
|
| 572 |
+
"step": 42500
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 3.48,
|
| 576 |
+
"learning_rate": 6.8609509840706765e-06,
|
| 577 |
+
"loss": 0.3284,
|
| 578 |
+
"step": 43000
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 3.52,
|
| 582 |
+
"learning_rate": 6.818313606439951e-06,
|
| 583 |
+
"loss": 0.3446,
|
| 584 |
+
"step": 43500
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 3.56,
|
| 588 |
+
"learning_rate": 6.775676228809224e-06,
|
| 589 |
+
"loss": 0.36,
|
| 590 |
+
"step": 44000
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 3.6,
|
| 594 |
+
"learning_rate": 6.733038851178497e-06,
|
| 595 |
+
"loss": 0.3368,
|
| 596 |
+
"step": 44500
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 3.65,
|
| 600 |
+
"learning_rate": 6.690401473547772e-06,
|
| 601 |
+
"loss": 0.3362,
|
| 602 |
+
"step": 45000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 3.69,
|
| 606 |
+
"learning_rate": 6.647764095917045e-06,
|
| 607 |
+
"loss": 0.338,
|
| 608 |
+
"step": 45500
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 3.73,
|
| 612 |
+
"learning_rate": 6.6052119930415805e-06,
|
| 613 |
+
"loss": 0.3206,
|
| 614 |
+
"step": 46000
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 3.77,
|
| 618 |
+
"learning_rate": 6.562659890166116e-06,
|
| 619 |
+
"loss": 0.339,
|
| 620 |
+
"step": 46500
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 3.81,
|
| 624 |
+
"learning_rate": 6.52002251253539e-06,
|
| 625 |
+
"loss": 0.3226,
|
| 626 |
+
"step": 47000
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 3.85,
|
| 630 |
+
"learning_rate": 6.477385134904664e-06,
|
| 631 |
+
"loss": 0.3386,
|
| 632 |
+
"step": 47500
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 3.89,
|
| 636 |
+
"learning_rate": 6.434747757273937e-06,
|
| 637 |
+
"loss": 0.3406,
|
| 638 |
+
"step": 48000
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 3.93,
|
| 642 |
+
"learning_rate": 6.39211037964321e-06,
|
| 643 |
+
"loss": 0.3231,
|
| 644 |
+
"step": 48500
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 3.97,
|
| 648 |
+
"learning_rate": 6.349473002012485e-06,
|
| 649 |
+
"loss": 0.3418,
|
| 650 |
+
"step": 49000
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 4.0,
|
| 654 |
+
"eval_BLEU": 51.32592897573737,
|
| 655 |
+
"eval_BLEU-Bigram-Precision": 55.485615933120236,
|
| 656 |
+
"eval_BLEU-Trigram-Precision": 47.36897219749725,
|
| 657 |
+
"eval_BLEU-Unigram-Precision": 70.36899684776562,
|
| 658 |
+
"eval_ROUGE-2": 45.802241938436794,
|
| 659 |
+
"eval_ROUGE-L": 57.842922925143036,
|
| 660 |
+
"eval_Sacre-Bigram-Precision": 54.30648474951272,
|
| 661 |
+
"eval_Sacre-Trigram-Precision": 47.28686538569899,
|
| 662 |
+
"eval_Sacre-Unigram-Precision": 69.06033630069238,
|
| 663 |
+
"eval_SacreBLEU": 50.59726409455209,
|
| 664 |
+
"eval_loss": 0.5526171326637268,
|
| 665 |
+
"eval_runtime": 916.9598,
|
| 666 |
+
"eval_samples_per_second": 1.349,
|
| 667 |
+
"eval_steps_per_second": 1.349,
|
| 668 |
+
"step": 49376
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 4.01,
|
| 672 |
+
"learning_rate": 6.3069208991370205e-06,
|
| 673 |
+
"loss": 0.326,
|
| 674 |
+
"step": 49500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 4.05,
|
| 678 |
+
"learning_rate": 6.264283521506294e-06,
|
| 679 |
+
"loss": 0.2931,
|
| 680 |
+
"step": 50000
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 4.09,
|
| 684 |
+
"learning_rate": 6.221646143875567e-06,
|
| 685 |
+
"loss": 0.2846,
|
| 686 |
+
"step": 50500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 4.13,
|
| 690 |
+
"learning_rate": 6.179008766244842e-06,
|
| 691 |
+
"loss": 0.2896,
|
| 692 |
+
"step": 51000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 4.17,
|
| 696 |
+
"learning_rate": 6.136371388614115e-06,
|
| 697 |
+
"loss": 0.2961,
|
| 698 |
+
"step": 51500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 4.21,
|
| 702 |
+
"learning_rate": 6.0937340109833884e-06,
|
| 703 |
+
"loss": 0.2679,
|
| 704 |
+
"step": 52000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 4.25,
|
| 708 |
+
"learning_rate": 6.051096633352663e-06,
|
| 709 |
+
"loss": 0.3005,
|
| 710 |
+
"step": 52500
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 4.29,
|
| 714 |
+
"learning_rate": 6.0085445304771986e-06,
|
| 715 |
+
"loss": 0.2921,
|
| 716 |
+
"step": 53000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 4.33,
|
| 720 |
+
"learning_rate": 5.965907152846472e-06,
|
| 721 |
+
"loss": 0.2871,
|
| 722 |
+
"step": 53500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 4.37,
|
| 726 |
+
"learning_rate": 5.923269775215745e-06,
|
| 727 |
+
"loss": 0.2853,
|
| 728 |
+
"step": 54000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 4.42,
|
| 732 |
+
"learning_rate": 5.88063239758502e-06,
|
| 733 |
+
"loss": 0.2759,
|
| 734 |
+
"step": 54500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 4.46,
|
| 738 |
+
"learning_rate": 5.837995019954293e-06,
|
| 739 |
+
"loss": 0.2793,
|
| 740 |
+
"step": 55000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 4.5,
|
| 744 |
+
"learning_rate": 5.7953576423235665e-06,
|
| 745 |
+
"loss": 0.2924,
|
| 746 |
+
"step": 55500
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 4.54,
|
| 750 |
+
"learning_rate": 5.7527202646928406e-06,
|
| 751 |
+
"loss": 0.3162,
|
| 752 |
+
"step": 56000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 4.58,
|
| 756 |
+
"learning_rate": 5.710168161817377e-06,
|
| 757 |
+
"loss": 0.2815,
|
| 758 |
+
"step": 56500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 4.62,
|
| 762 |
+
"learning_rate": 5.66753078418665e-06,
|
| 763 |
+
"loss": 0.308,
|
| 764 |
+
"step": 57000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 4.66,
|
| 768 |
+
"learning_rate": 5.624893406555923e-06,
|
| 769 |
+
"loss": 0.2984,
|
| 770 |
+
"step": 57500
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 4.7,
|
| 774 |
+
"learning_rate": 5.582256028925197e-06,
|
| 775 |
+
"loss": 0.2928,
|
| 776 |
+
"step": 58000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 4.74,
|
| 780 |
+
"learning_rate": 5.539618651294471e-06,
|
| 781 |
+
"loss": 0.2903,
|
| 782 |
+
"step": 58500
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 4.78,
|
| 786 |
+
"learning_rate": 5.496981273663745e-06,
|
| 787 |
+
"loss": 0.2971,
|
| 788 |
+
"step": 59000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 4.82,
|
| 792 |
+
"learning_rate": 5.454343896033019e-06,
|
| 793 |
+
"loss": 0.2815,
|
| 794 |
+
"step": 59500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 4.86,
|
| 798 |
+
"learning_rate": 5.411706518402292e-06,
|
| 799 |
+
"loss": 0.3028,
|
| 800 |
+
"step": 60000
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 4.9,
|
| 804 |
+
"learning_rate": 5.369154415526828e-06,
|
| 805 |
+
"loss": 0.2931,
|
| 806 |
+
"step": 60500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 4.94,
|
| 810 |
+
"learning_rate": 5.326517037896102e-06,
|
| 811 |
+
"loss": 0.2859,
|
| 812 |
+
"step": 61000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 4.98,
|
| 816 |
+
"learning_rate": 5.283879660265375e-06,
|
| 817 |
+
"loss": 0.2838,
|
| 818 |
+
"step": 61500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 5.0,
|
| 822 |
+
"eval_BLEU": 53.29609495182536,
|
| 823 |
+
"eval_BLEU-Bigram-Precision": 57.167285518858705,
|
| 824 |
+
"eval_BLEU-Trigram-Precision": 49.0354115750611,
|
| 825 |
+
"eval_BLEU-Unigram-Precision": 71.76226675881134,
|
| 826 |
+
"eval_ROUGE-2": 48.21696209970032,
|
| 827 |
+
"eval_ROUGE-L": 59.61148640581645,
|
| 828 |
+
"eval_Sacre-Bigram-Precision": 56.00606885005755,
|
| 829 |
+
"eval_Sacre-Trigram-Precision": 48.951166303070984,
|
| 830 |
+
"eval_Sacre-Unigram-Precision": 70.51250552798388,
|
| 831 |
+
"eval_SacreBLEU": 52.60559213285124,
|
| 832 |
+
"eval_loss": 0.5165597796440125,
|
| 833 |
+
"eval_runtime": 915.3947,
|
| 834 |
+
"eval_samples_per_second": 1.351,
|
| 835 |
+
"eval_steps_per_second": 1.351,
|
| 836 |
+
"step": 61720
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 5.02,
|
| 840 |
+
"learning_rate": 5.2412422826346485e-06,
|
| 841 |
+
"loss": 0.2663,
|
| 842 |
+
"step": 62000
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 5.06,
|
| 846 |
+
"learning_rate": 5.1986901797591846e-06,
|
| 847 |
+
"loss": 0.25,
|
| 848 |
+
"step": 62500
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 5.1,
|
| 852 |
+
"learning_rate": 5.156052802128459e-06,
|
| 853 |
+
"loss": 0.2474,
|
| 854 |
+
"step": 63000
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 5.14,
|
| 858 |
+
"learning_rate": 5.113415424497732e-06,
|
| 859 |
+
"loss": 0.2469,
|
| 860 |
+
"step": 63500
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 5.18,
|
| 864 |
+
"learning_rate": 5.070778046867005e-06,
|
| 865 |
+
"loss": 0.277,
|
| 866 |
+
"step": 64000
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 5.23,
|
| 870 |
+
"learning_rate": 5.02814066923628e-06,
|
| 871 |
+
"loss": 0.2478,
|
| 872 |
+
"step": 64500
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 5.27,
|
| 876 |
+
"learning_rate": 4.985588566360815e-06,
|
| 877 |
+
"loss": 0.2593,
|
| 878 |
+
"step": 65000
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 5.31,
|
| 882 |
+
"learning_rate": 4.9429511887300885e-06,
|
| 883 |
+
"loss": 0.2565,
|
| 884 |
+
"step": 65500
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 5.35,
|
| 888 |
+
"learning_rate": 4.900313811099363e-06,
|
| 889 |
+
"loss": 0.2678,
|
| 890 |
+
"step": 66000
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 5.39,
|
| 894 |
+
"learning_rate": 4.857676433468637e-06,
|
| 895 |
+
"loss": 0.2434,
|
| 896 |
+
"step": 66500
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 5.43,
|
| 900 |
+
"learning_rate": 4.815124330593172e-06,
|
| 901 |
+
"loss": 0.2776,
|
| 902 |
+
"step": 67000
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 5.47,
|
| 906 |
+
"learning_rate": 4.772486952962445e-06,
|
| 907 |
+
"loss": 0.2678,
|
| 908 |
+
"step": 67500
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 5.51,
|
| 912 |
+
"learning_rate": 4.729849575331719e-06,
|
| 913 |
+
"loss": 0.2784,
|
| 914 |
+
"step": 68000
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 5.55,
|
| 918 |
+
"learning_rate": 4.687212197700993e-06,
|
| 919 |
+
"loss": 0.2404,
|
| 920 |
+
"step": 68500
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 5.59,
|
| 924 |
+
"learning_rate": 4.6445748200702666e-06,
|
| 925 |
+
"loss": 0.26,
|
| 926 |
+
"step": 69000
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 5.63,
|
| 930 |
+
"learning_rate": 4.601937442439541e-06,
|
| 931 |
+
"loss": 0.242,
|
| 932 |
+
"step": 69500
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 5.67,
|
| 936 |
+
"learning_rate": 4.559300064808814e-06,
|
| 937 |
+
"loss": 0.2636,
|
| 938 |
+
"step": 70000
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 5.71,
|
| 942 |
+
"learning_rate": 4.516662687178088e-06,
|
| 943 |
+
"loss": 0.2703,
|
| 944 |
+
"step": 70500
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 5.75,
|
| 948 |
+
"learning_rate": 4.474110584302623e-06,
|
| 949 |
+
"loss": 0.2457,
|
| 950 |
+
"step": 71000
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 5.79,
|
| 954 |
+
"learning_rate": 4.431473206671897e-06,
|
| 955 |
+
"loss": 0.2492,
|
| 956 |
+
"step": 71500
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 5.83,
|
| 960 |
+
"learning_rate": 4.3888358290411705e-06,
|
| 961 |
+
"loss": 0.2677,
|
| 962 |
+
"step": 72000
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 5.87,
|
| 966 |
+
"learning_rate": 4.346198451410445e-06,
|
| 967 |
+
"loss": 0.2557,
|
| 968 |
+
"step": 72500
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 5.91,
|
| 972 |
+
"learning_rate": 4.303561073779719e-06,
|
| 973 |
+
"loss": 0.262,
|
| 974 |
+
"step": 73000
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 5.95,
|
| 978 |
+
"learning_rate": 4.261008970904254e-06,
|
| 979 |
+
"loss": 0.247,
|
| 980 |
+
"step": 73500
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 5.99,
|
| 984 |
+
"learning_rate": 4.218371593273527e-06,
|
| 985 |
+
"loss": 0.2461,
|
| 986 |
+
"step": 74000
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 6.0,
|
| 990 |
+
"eval_BLEU": 56.19871507978884,
|
| 991 |
+
"eval_BLEU-Bigram-Precision": 58.651824957003626,
|
| 992 |
+
"eval_BLEU-Trigram-Precision": 51.109418634171114,
|
| 993 |
+
"eval_BLEU-Unigram-Precision": 72.28111326627273,
|
| 994 |
+
"eval_ROUGE-2": 50.26575052573452,
|
| 995 |
+
"eval_ROUGE-L": 61.216031502511356,
|
| 996 |
+
"eval_Sacre-Bigram-Precision": 57.60808534531162,
|
| 997 |
+
"eval_Sacre-Trigram-Precision": 51.07878391631252,
|
| 998 |
+
"eval_Sacre-Unigram-Precision": 71.06299212598425,
|
| 999 |
+
"eval_SacreBLEU": 55.641908183294326,
|
| 1000 |
+
"eval_loss": 0.49508827924728394,
|
| 1001 |
+
"eval_runtime": 1114.4052,
|
| 1002 |
+
"eval_samples_per_second": 1.11,
|
| 1003 |
+
"eval_steps_per_second": 1.11,
|
| 1004 |
+
"step": 74064
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 6.04,
|
| 1008 |
+
"learning_rate": 4.175734215642801e-06,
|
| 1009 |
+
"loss": 0.2288,
|
| 1010 |
+
"step": 74500
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 6.08,
|
| 1014 |
+
"learning_rate": 4.133096838012075e-06,
|
| 1015 |
+
"loss": 0.2143,
|
| 1016 |
+
"step": 75000
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 6.12,
|
| 1020 |
+
"learning_rate": 4.0904594603813494e-06,
|
| 1021 |
+
"loss": 0.2089,
|
| 1022 |
+
"step": 75500
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 6.16,
|
| 1026 |
+
"learning_rate": 4.047907357505884e-06,
|
| 1027 |
+
"loss": 0.2401,
|
| 1028 |
+
"step": 76000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 6.2,
|
| 1032 |
+
"learning_rate": 4.00535525463042e-06,
|
| 1033 |
+
"loss": 0.2384,
|
| 1034 |
+
"step": 76500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 6.24,
|
| 1038 |
+
"learning_rate": 3.962717876999693e-06,
|
| 1039 |
+
"loss": 0.2268,
|
| 1040 |
+
"step": 77000
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 6.28,
|
| 1044 |
+
"learning_rate": 3.920080499368967e-06,
|
| 1045 |
+
"loss": 0.2197,
|
| 1046 |
+
"step": 77500
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 6.32,
|
| 1050 |
+
"learning_rate": 3.87744312173824e-06,
|
| 1051 |
+
"loss": 0.2373,
|
| 1052 |
+
"step": 78000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 6.36,
|
| 1056 |
+
"learning_rate": 3.8348910188627765e-06,
|
| 1057 |
+
"loss": 0.2261,
|
| 1058 |
+
"step": 78500
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 6.4,
|
| 1062 |
+
"learning_rate": 3.7922536412320497e-06,
|
| 1063 |
+
"loss": 0.2299,
|
| 1064 |
+
"step": 79000
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 6.44,
|
| 1068 |
+
"learning_rate": 3.749616263601324e-06,
|
| 1069 |
+
"loss": 0.2267,
|
| 1070 |
+
"step": 79500
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 6.48,
|
| 1074 |
+
"learning_rate": 3.7069788859705975e-06,
|
| 1075 |
+
"loss": 0.2196,
|
| 1076 |
+
"step": 80000
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 6.52,
|
| 1080 |
+
"learning_rate": 3.6643415083398716e-06,
|
| 1081 |
+
"loss": 0.2192,
|
| 1082 |
+
"step": 80500
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 6.56,
|
| 1086 |
+
"learning_rate": 3.621704130709145e-06,
|
| 1087 |
+
"loss": 0.2365,
|
| 1088 |
+
"step": 81000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 6.6,
|
| 1092 |
+
"learning_rate": 3.579066753078419e-06,
|
| 1093 |
+
"loss": 0.2187,
|
| 1094 |
+
"step": 81500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 6.64,
|
| 1098 |
+
"learning_rate": 3.5364293754476926e-06,
|
| 1099 |
+
"loss": 0.242,
|
| 1100 |
+
"step": 82000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 6.68,
|
| 1104 |
+
"learning_rate": 3.493877272572228e-06,
|
| 1105 |
+
"loss": 0.2483,
|
| 1106 |
+
"step": 82500
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 6.72,
|
| 1110 |
+
"learning_rate": 3.4512398949415014e-06,
|
| 1111 |
+
"loss": 0.2399,
|
| 1112 |
+
"step": 83000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 6.76,
|
| 1116 |
+
"learning_rate": 3.4086025173107755e-06,
|
| 1117 |
+
"loss": 0.2409,
|
| 1118 |
+
"step": 83500
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 6.8,
|
| 1122 |
+
"learning_rate": 3.365965139680049e-06,
|
| 1123 |
+
"loss": 0.242,
|
| 1124 |
+
"step": 84000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 6.85,
|
| 1128 |
+
"learning_rate": 3.323413036804585e-06,
|
| 1129 |
+
"loss": 0.2484,
|
| 1130 |
+
"step": 84500
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 6.89,
|
| 1134 |
+
"learning_rate": 3.280775659173858e-06,
|
| 1135 |
+
"loss": 0.2562,
|
| 1136 |
+
"step": 85000
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 6.93,
|
| 1140 |
+
"learning_rate": 3.238138281543132e-06,
|
| 1141 |
+
"loss": 0.2283,
|
| 1142 |
+
"step": 85500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 6.97,
|
| 1146 |
+
"learning_rate": 3.1955009039124062e-06,
|
| 1147 |
+
"loss": 0.236,
|
| 1148 |
+
"step": 86000
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 7.0,
|
| 1152 |
+
"eval_BLEU": 57.78119869959703,
|
| 1153 |
+
"eval_BLEU-Bigram-Precision": 61.05482573076737,
|
| 1154 |
+
"eval_BLEU-Trigram-Precision": 53.63036303630363,
|
| 1155 |
+
"eval_BLEU-Unigram-Precision": 74.23397054788255,
|
| 1156 |
+
"eval_ROUGE-2": 52.1883256479329,
|
| 1157 |
+
"eval_ROUGE-L": 63.12941898321425,
|
| 1158 |
+
"eval_Sacre-Bigram-Precision": 60.065305276251685,
|
| 1159 |
+
"eval_Sacre-Trigram-Precision": 53.55263886581381,
|
| 1160 |
+
"eval_Sacre-Unigram-Precision": 73.0894744532658,
|
| 1161 |
+
"eval_SacreBLEU": 57.19500795208099,
|
| 1162 |
+
"eval_loss": 0.4768661558628082,
|
| 1163 |
+
"eval_runtime": 1205.3601,
|
| 1164 |
+
"eval_samples_per_second": 1.026,
|
| 1165 |
+
"eval_steps_per_second": 1.026,
|
| 1166 |
+
"step": 86408
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 7.01,
|
| 1170 |
+
"learning_rate": 3.1529488010369415e-06,
|
| 1171 |
+
"loss": 0.2257,
|
| 1172 |
+
"step": 86500
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 7.05,
|
| 1176 |
+
"learning_rate": 3.1103114234062147e-06,
|
| 1177 |
+
"loss": 0.2245,
|
| 1178 |
+
"step": 87000
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 7.09,
|
| 1182 |
+
"learning_rate": 3.0676740457754888e-06,
|
| 1183 |
+
"loss": 0.215,
|
| 1184 |
+
"step": 87500
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 7.13,
|
| 1188 |
+
"learning_rate": 3.025036668144763e-06,
|
| 1189 |
+
"loss": 0.2102,
|
| 1190 |
+
"step": 88000
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 7.17,
|
| 1194 |
+
"learning_rate": 2.982484565269298e-06,
|
| 1195 |
+
"loss": 0.211,
|
| 1196 |
+
"step": 88500
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 7.21,
|
| 1200 |
+
"learning_rate": 2.9398471876385713e-06,
|
| 1201 |
+
"loss": 0.236,
|
| 1202 |
+
"step": 89000
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 7.25,
|
| 1206 |
+
"learning_rate": 2.8972098100078454e-06,
|
| 1207 |
+
"loss": 0.2069,
|
| 1208 |
+
"step": 89500
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 7.29,
|
| 1212 |
+
"learning_rate": 2.8545724323771195e-06,
|
| 1213 |
+
"loss": 0.2169,
|
| 1214 |
+
"step": 90000
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 7.33,
|
| 1218 |
+
"learning_rate": 2.8120203295016547e-06,
|
| 1219 |
+
"loss": 0.209,
|
| 1220 |
+
"step": 90500
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 7.37,
|
| 1224 |
+
"learning_rate": 2.769382951870928e-06,
|
| 1225 |
+
"loss": 0.2195,
|
| 1226 |
+
"step": 91000
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 7.41,
|
| 1230 |
+
"learning_rate": 2.726745574240202e-06,
|
| 1231 |
+
"loss": 0.2074,
|
| 1232 |
+
"step": 91500
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 7.45,
|
| 1236 |
+
"learning_rate": 2.6841934713647373e-06,
|
| 1237 |
+
"loss": 0.2112,
|
| 1238 |
+
"step": 92000
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 7.49,
|
| 1242 |
+
"learning_rate": 2.6415560937340113e-06,
|
| 1243 |
+
"loss": 0.2145,
|
| 1244 |
+
"step": 92500
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 7.53,
|
| 1248 |
+
"learning_rate": 2.5989187161032846e-06,
|
| 1249 |
+
"loss": 0.213,
|
| 1250 |
+
"step": 93000
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 7.57,
|
| 1254 |
+
"learning_rate": 2.5562813384725587e-06,
|
| 1255 |
+
"loss": 0.1966,
|
| 1256 |
+
"step": 93500
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 7.62,
|
| 1260 |
+
"learning_rate": 2.5136439608418328e-06,
|
| 1261 |
+
"loss": 0.2097,
|
| 1262 |
+
"step": 94000
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 7.66,
|
| 1266 |
+
"learning_rate": 2.4710065832111064e-06,
|
| 1267 |
+
"loss": 0.2167,
|
| 1268 |
+
"step": 94500
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 7.7,
|
| 1272 |
+
"learning_rate": 2.42836920558038e-06,
|
| 1273 |
+
"loss": 0.2144,
|
| 1274 |
+
"step": 95000
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 7.74,
|
| 1278 |
+
"learning_rate": 2.3858171027049153e-06,
|
| 1279 |
+
"loss": 0.2277,
|
| 1280 |
+
"step": 95500
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 7.78,
|
| 1284 |
+
"learning_rate": 2.343179725074189e-06,
|
| 1285 |
+
"loss": 0.2347,
|
| 1286 |
+
"step": 96000
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 7.82,
|
| 1290 |
+
"learning_rate": 2.300542347443463e-06,
|
| 1291 |
+
"loss": 0.2235,
|
| 1292 |
+
"step": 96500
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 7.86,
|
| 1296 |
+
"learning_rate": 2.2579049698127367e-06,
|
| 1297 |
+
"loss": 0.2197,
|
| 1298 |
+
"step": 97000
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 7.9,
|
| 1302 |
+
"learning_rate": 2.2152675921820104e-06,
|
| 1303 |
+
"loss": 0.2219,
|
| 1304 |
+
"step": 97500
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 7.94,
|
| 1308 |
+
"learning_rate": 2.1726302145512845e-06,
|
| 1309 |
+
"loss": 0.21,
|
| 1310 |
+
"step": 98000
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 7.98,
|
| 1314 |
+
"learning_rate": 2.129992836920558e-06,
|
| 1315 |
+
"loss": 0.2014,
|
| 1316 |
+
"step": 98500
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 8.0,
|
| 1320 |
+
"eval_BLEU": 59.598575339554586,
|
| 1321 |
+
"eval_BLEU-Bigram-Precision": 63.173579559333625,
|
| 1322 |
+
"eval_BLEU-Trigram-Precision": 56.19508137056102,
|
| 1323 |
+
"eval_BLEU-Unigram-Precision": 75.73942688657515,
|
| 1324 |
+
"eval_ROUGE-2": 54.010901558700496,
|
| 1325 |
+
"eval_ROUGE-L": 64.55563198822112,
|
| 1326 |
+
"eval_Sacre-Bigram-Precision": 62.145801805562805,
|
| 1327 |
+
"eval_Sacre-Trigram-Precision": 56.11089418196017,
|
| 1328 |
+
"eval_Sacre-Unigram-Precision": 74.65196078431373,
|
| 1329 |
+
"eval_SacreBLEU": 59.05906731573519,
|
| 1330 |
+
"eval_loss": 0.4607221782207489,
|
| 1331 |
+
"eval_runtime": 1200.6288,
|
| 1332 |
+
"eval_samples_per_second": 1.03,
|
| 1333 |
+
"eval_steps_per_second": 1.03,
|
| 1334 |
+
"step": 98752
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 8.02,
|
| 1338 |
+
"learning_rate": 2.0873554592898322e-06,
|
| 1339 |
+
"loss": 0.2166,
|
| 1340 |
+
"step": 99000
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 8.06,
|
| 1344 |
+
"learning_rate": 2.0448033564143674e-06,
|
| 1345 |
+
"loss": 0.1819,
|
| 1346 |
+
"step": 99500
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 8.1,
|
| 1350 |
+
"learning_rate": 2.002165978783641e-06,
|
| 1351 |
+
"loss": 0.1905,
|
| 1352 |
+
"step": 100000
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 8.14,
|
| 1356 |
+
"learning_rate": 1.9595286011529148e-06,
|
| 1357 |
+
"loss": 0.2006,
|
| 1358 |
+
"step": 100500
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 8.18,
|
| 1362 |
+
"learning_rate": 1.916891223522189e-06,
|
| 1363 |
+
"loss": 0.2078,
|
| 1364 |
+
"step": 101000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 8.22,
|
| 1368 |
+
"learning_rate": 1.8742538458914625e-06,
|
| 1369 |
+
"loss": 0.2051,
|
| 1370 |
+
"step": 101500
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 8.26,
|
| 1374 |
+
"learning_rate": 1.8317017430159977e-06,
|
| 1375 |
+
"loss": 0.207,
|
| 1376 |
+
"step": 102000
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 8.3,
|
| 1380 |
+
"learning_rate": 1.7890643653852716e-06,
|
| 1381 |
+
"loss": 0.1757,
|
| 1382 |
+
"step": 102500
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 8.34,
|
| 1386 |
+
"learning_rate": 1.7464269877545453e-06,
|
| 1387 |
+
"loss": 0.2117,
|
| 1388 |
+
"step": 103000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 8.38,
|
| 1392 |
+
"learning_rate": 1.7037896101238192e-06,
|
| 1393 |
+
"loss": 0.2073,
|
| 1394 |
+
"step": 103500
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 8.43,
|
| 1398 |
+
"learning_rate": 1.6611522324930928e-06,
|
| 1399 |
+
"loss": 0.2056,
|
| 1400 |
+
"step": 104000
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 8.47,
|
| 1404 |
+
"learning_rate": 1.6186001296176283e-06,
|
| 1405 |
+
"loss": 0.1992,
|
| 1406 |
+
"step": 104500
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 8.51,
|
| 1410 |
+
"learning_rate": 1.575962751986902e-06,
|
| 1411 |
+
"loss": 0.2169,
|
| 1412 |
+
"step": 105000
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 8.55,
|
| 1416 |
+
"learning_rate": 1.5333253743561758e-06,
|
| 1417 |
+
"loss": 0.2039,
|
| 1418 |
+
"step": 105500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 8.59,
|
| 1422 |
+
"learning_rate": 1.4906879967254495e-06,
|
| 1423 |
+
"loss": 0.1897,
|
| 1424 |
+
"step": 106000
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 8.63,
|
| 1428 |
+
"learning_rate": 1.4480506190947233e-06,
|
| 1429 |
+
"loss": 0.2109,
|
| 1430 |
+
"step": 106500
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 8.67,
|
| 1434 |
+
"learning_rate": 1.4054985162192586e-06,
|
| 1435 |
+
"loss": 0.2002,
|
| 1436 |
+
"step": 107000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 8.71,
|
| 1440 |
+
"learning_rate": 1.3628611385885324e-06,
|
| 1441 |
+
"loss": 0.2071,
|
| 1442 |
+
"step": 107500
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 8.75,
|
| 1446 |
+
"learning_rate": 1.3203090357130676e-06,
|
| 1447 |
+
"loss": 0.2247,
|
| 1448 |
+
"step": 108000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 8.79,
|
| 1452 |
+
"learning_rate": 1.2776716580823415e-06,
|
| 1453 |
+
"loss": 0.2092,
|
| 1454 |
+
"step": 108500
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 8.83,
|
| 1458 |
+
"learning_rate": 1.2350342804516152e-06,
|
| 1459 |
+
"loss": 0.2184,
|
| 1460 |
+
"step": 109000
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 8.87,
|
| 1464 |
+
"learning_rate": 1.1924821775761504e-06,
|
| 1465 |
+
"loss": 0.2012,
|
| 1466 |
+
"step": 109500
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 8.91,
|
| 1470 |
+
"learning_rate": 1.1498447999454243e-06,
|
| 1471 |
+
"loss": 0.2024,
|
| 1472 |
+
"step": 110000
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 8.95,
|
| 1476 |
+
"learning_rate": 1.1072074223146981e-06,
|
| 1477 |
+
"loss": 0.2178,
|
| 1478 |
+
"step": 110500
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 8.99,
|
| 1482 |
+
"learning_rate": 1.0645700446839718e-06,
|
| 1483 |
+
"loss": 0.2087,
|
| 1484 |
+
"step": 111000
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 9.0,
|
| 1488 |
+
"eval_BLEU": 59.228700736361574,
|
| 1489 |
+
"eval_BLEU-Bigram-Precision": 62.02898550724638,
|
| 1490 |
+
"eval_BLEU-Trigram-Precision": 55.13255240443896,
|
| 1491 |
+
"eval_BLEU-Unigram-Precision": 74.83247481424078,
|
| 1492 |
+
"eval_ROUGE-2": 53.99481361299642,
|
| 1493 |
+
"eval_ROUGE-L": 64.48941872388136,
|
| 1494 |
+
"eval_Sacre-Bigram-Precision": 61.05263157894737,
|
| 1495 |
+
"eval_Sacre-Trigram-Precision": 55.05952380952381,
|
| 1496 |
+
"eval_Sacre-Unigram-Precision": 73.72556628025416,
|
| 1497 |
+
"eval_SacreBLEU": 58.70218229892996,
|
| 1498 |
+
"eval_loss": 0.4529183506965637,
|
| 1499 |
+
"eval_runtime": 1120.8668,
|
| 1500 |
+
"eval_samples_per_second": 1.104,
|
| 1501 |
+
"eval_steps_per_second": 1.104,
|
| 1502 |
+
"step": 111096
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 9.03,
|
| 1506 |
+
"learning_rate": 1.0219326670532457e-06,
|
| 1507 |
+
"loss": 0.2001,
|
| 1508 |
+
"step": 111500
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 9.07,
|
| 1512 |
+
"learning_rate": 9.792952894225194e-07,
|
| 1513 |
+
"loss": 0.1909,
|
| 1514 |
+
"step": 112000
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 9.11,
|
| 1518 |
+
"learning_rate": 9.366579117917933e-07,
|
| 1519 |
+
"loss": 0.1929,
|
| 1520 |
+
"step": 112500
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 9.15,
|
| 1524 |
+
"learning_rate": 8.940205341610671e-07,
|
| 1525 |
+
"loss": 0.1847,
|
| 1526 |
+
"step": 113000
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 9.19,
|
| 1530 |
+
"learning_rate": 8.514684312856022e-07,
|
| 1531 |
+
"loss": 0.1883,
|
| 1532 |
+
"step": 113500
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 9.24,
|
| 1536 |
+
"learning_rate": 8.08831053654876e-07,
|
| 1537 |
+
"loss": 0.1995,
|
| 1538 |
+
"step": 114000
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 9.28,
|
| 1542 |
+
"learning_rate": 7.662789507794113e-07,
|
| 1543 |
+
"loss": 0.2046,
|
| 1544 |
+
"step": 114500
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 9.32,
|
| 1548 |
+
"learning_rate": 7.236415731486851e-07,
|
| 1549 |
+
"loss": 0.1945,
|
| 1550 |
+
"step": 115000
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 9.36,
|
| 1554 |
+
"learning_rate": 6.810041955179588e-07,
|
| 1555 |
+
"loss": 0.2007,
|
| 1556 |
+
"step": 115500
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 9.4,
|
| 1560 |
+
"learning_rate": 6.383668178872326e-07,
|
| 1561 |
+
"loss": 0.1966,
|
| 1562 |
+
"step": 116000
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 9.44,
|
| 1566 |
+
"learning_rate": 5.957294402565065e-07,
|
| 1567 |
+
"loss": 0.1924,
|
| 1568 |
+
"step": 116500
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 9.48,
|
| 1572 |
+
"learning_rate": 5.530920626257804e-07,
|
| 1573 |
+
"loss": 0.1974,
|
| 1574 |
+
"step": 117000
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 9.52,
|
| 1578 |
+
"learning_rate": 5.104546849950541e-07,
|
| 1579 |
+
"loss": 0.2053,
|
| 1580 |
+
"step": 117500
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 9.56,
|
| 1584 |
+
"learning_rate": 4.6781730736432786e-07,
|
| 1585 |
+
"loss": 0.2116,
|
| 1586 |
+
"step": 118000
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 9.6,
|
| 1590 |
+
"learning_rate": 4.251799297336017e-07,
|
| 1591 |
+
"loss": 0.2003,
|
| 1592 |
+
"step": 118500
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 9.64,
|
| 1596 |
+
"learning_rate": 3.8254255210287546e-07,
|
| 1597 |
+
"loss": 0.1966,
|
| 1598 |
+
"step": 119000
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 9.68,
|
| 1602 |
+
"learning_rate": 3.3990517447214934e-07,
|
| 1603 |
+
"loss": 0.1966,
|
| 1604 |
+
"step": 119500
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 9.72,
|
| 1608 |
+
"learning_rate": 2.9735307159668455e-07,
|
| 1609 |
+
"loss": 0.2016,
|
| 1610 |
+
"step": 120000
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 9.76,
|
| 1614 |
+
"learning_rate": 2.547156939659583e-07,
|
| 1615 |
+
"loss": 0.2098,
|
| 1616 |
+
"step": 120500
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 9.8,
|
| 1620 |
+
"learning_rate": 2.1207831633523215e-07,
|
| 1621 |
+
"loss": 0.2038,
|
| 1622 |
+
"step": 121000
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 9.84,
|
| 1626 |
+
"learning_rate": 1.6944093870450592e-07,
|
| 1627 |
+
"loss": 0.1956,
|
| 1628 |
+
"step": 121500
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 9.88,
|
| 1632 |
+
"learning_rate": 1.2680356107377972e-07,
|
| 1633 |
+
"loss": 0.2036,
|
| 1634 |
+
"step": 122000
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 9.92,
|
| 1638 |
+
"learning_rate": 8.416618344305353e-08,
|
| 1639 |
+
"loss": 0.22,
|
| 1640 |
+
"step": 122500
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 9.96,
|
| 1644 |
+
"learning_rate": 4.1528805812327316e-08,
|
| 1645 |
+
"loss": 0.2003,
|
| 1646 |
+
"step": 123000
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 10.0,
|
| 1650 |
+
"eval_BLEU": 60.284861769008316,
|
| 1651 |
+
"eval_BLEU-Bigram-Precision": 62.98832834957076,
|
| 1652 |
+
"eval_BLEU-Trigram-Precision": 56.25192327418197,
|
| 1653 |
+
"eval_BLEU-Unigram-Precision": 75.38118428838013,
|
| 1654 |
+
"eval_ROUGE-2": 54.858141016126524,
|
| 1655 |
+
"eval_ROUGE-L": 65.28596759853026,
|
| 1656 |
+
"eval_Sacre-Bigram-Precision": 62.07553975369712,
|
| 1657 |
+
"eval_Sacre-Trigram-Precision": 56.22145176379946,
|
| 1658 |
+
"eval_Sacre-Unigram-Precision": 74.30246076341794,
|
| 1659 |
+
"eval_SacreBLEU": 59.77673945128229,
|
| 1660 |
+
"eval_loss": 0.4481608271598816,
|
| 1661 |
+
"eval_runtime": 1123.8109,
|
| 1662 |
+
"eval_samples_per_second": 1.101,
|
| 1663 |
+
"eval_steps_per_second": 1.101,
|
| 1664 |
+
"step": 123440
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 10.0,
|
| 1668 |
+
"step": 123440,
|
| 1669 |
+
"total_flos": 3230087590379520.0,
|
| 1670 |
+
"train_loss": 0.31656123802249336,
|
| 1671 |
+
"train_runtime": 24698.2128,
|
| 1672 |
+
"train_samples_per_second": 9.995,
|
| 1673 |
+
"train_steps_per_second": 4.998
|
| 1674 |
+
}
|
| 1675 |
+
],
|
| 1676 |
+
"max_steps": 123440,
|
| 1677 |
+
"num_train_epochs": 10,
|
| 1678 |
+
"total_flos": 3230087590379520.0,
|
| 1679 |
+
"trial_name": null,
|
| 1680 |
+
"trial_params": null
|
| 1681 |
+
}
|