Commit 路
b36d201
1
Parent(s): 8dc016a
Upload trainer_state.json
Browse files- trainer_state.json +4406 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,4406 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 0.14961867034435272,
|
| 3 |
+
"best_model_checkpoint": "/home/ubuntu/ML/source_code/fine-tuned-legalroberta/qa/checkpoint-364860",
|
| 4 |
+
"epoch": 2.0,
|
| 5 |
+
"global_step": 364860,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.0,
|
| 12 |
+
"learning_rate": 1.9981728151437082e-05,
|
| 13 |
+
"loss": 0.3901,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"learning_rate": 1.9963456302874163e-05,
|
| 19 |
+
"loss": 0.1441,
|
| 20 |
+
"step": 1000
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.01,
|
| 24 |
+
"learning_rate": 1.9945184454311244e-05,
|
| 25 |
+
"loss": 0.107,
|
| 26 |
+
"step": 1500
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.01,
|
| 30 |
+
"learning_rate": 1.9926912605748324e-05,
|
| 31 |
+
"loss": 0.1122,
|
| 32 |
+
"step": 2000
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.01,
|
| 36 |
+
"learning_rate": 1.9908640757185405e-05,
|
| 37 |
+
"loss": 0.1177,
|
| 38 |
+
"step": 2500
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.02,
|
| 42 |
+
"learning_rate": 1.9890368908622486e-05,
|
| 43 |
+
"loss": 0.1039,
|
| 44 |
+
"step": 3000
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.02,
|
| 48 |
+
"learning_rate": 1.9872097060059567e-05,
|
| 49 |
+
"loss": 0.0989,
|
| 50 |
+
"step": 3500
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.02,
|
| 54 |
+
"learning_rate": 1.985382521149665e-05,
|
| 55 |
+
"loss": 0.1081,
|
| 56 |
+
"step": 4000
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.02,
|
| 60 |
+
"learning_rate": 1.983555336293373e-05,
|
| 61 |
+
"loss": 0.0978,
|
| 62 |
+
"step": 4500
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.03,
|
| 66 |
+
"learning_rate": 1.981728151437081e-05,
|
| 67 |
+
"loss": 0.0979,
|
| 68 |
+
"step": 5000
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.03,
|
| 72 |
+
"learning_rate": 1.9799009665807893e-05,
|
| 73 |
+
"loss": 0.1083,
|
| 74 |
+
"step": 5500
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.03,
|
| 78 |
+
"learning_rate": 1.9780737817244973e-05,
|
| 79 |
+
"loss": 0.0935,
|
| 80 |
+
"step": 6000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.04,
|
| 84 |
+
"learning_rate": 1.9762465968682054e-05,
|
| 85 |
+
"loss": 0.091,
|
| 86 |
+
"step": 6500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.04,
|
| 90 |
+
"learning_rate": 1.9744194120119135e-05,
|
| 91 |
+
"loss": 0.1068,
|
| 92 |
+
"step": 7000
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.04,
|
| 96 |
+
"learning_rate": 1.9725922271556216e-05,
|
| 97 |
+
"loss": 0.1264,
|
| 98 |
+
"step": 7500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.04,
|
| 102 |
+
"learning_rate": 1.9707650422993296e-05,
|
| 103 |
+
"loss": 0.1381,
|
| 104 |
+
"step": 8000
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.05,
|
| 108 |
+
"learning_rate": 1.9689378574430377e-05,
|
| 109 |
+
"loss": 0.0971,
|
| 110 |
+
"step": 8500
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.05,
|
| 114 |
+
"learning_rate": 1.9671106725867458e-05,
|
| 115 |
+
"loss": 0.1426,
|
| 116 |
+
"step": 9000
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.05,
|
| 120 |
+
"learning_rate": 1.965283487730454e-05,
|
| 121 |
+
"loss": 0.1167,
|
| 122 |
+
"step": 9500
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.05,
|
| 126 |
+
"learning_rate": 1.963456302874162e-05,
|
| 127 |
+
"loss": 0.1031,
|
| 128 |
+
"step": 10000
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.06,
|
| 132 |
+
"learning_rate": 1.96162911801787e-05,
|
| 133 |
+
"loss": 0.1154,
|
| 134 |
+
"step": 10500
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.06,
|
| 138 |
+
"learning_rate": 1.959801933161578e-05,
|
| 139 |
+
"loss": 0.1116,
|
| 140 |
+
"step": 11000
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 0.06,
|
| 144 |
+
"learning_rate": 1.957974748305286e-05,
|
| 145 |
+
"loss": 0.0929,
|
| 146 |
+
"step": 11500
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.07,
|
| 150 |
+
"learning_rate": 1.9561475634489945e-05,
|
| 151 |
+
"loss": 0.0812,
|
| 152 |
+
"step": 12000
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.07,
|
| 156 |
+
"learning_rate": 1.9543203785927023e-05,
|
| 157 |
+
"loss": 0.1107,
|
| 158 |
+
"step": 12500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.07,
|
| 162 |
+
"learning_rate": 1.9524931937364103e-05,
|
| 163 |
+
"loss": 0.0934,
|
| 164 |
+
"step": 13000
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.07,
|
| 168 |
+
"learning_rate": 1.9506660088801187e-05,
|
| 169 |
+
"loss": 0.0963,
|
| 170 |
+
"step": 13500
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.08,
|
| 174 |
+
"learning_rate": 1.9488388240238268e-05,
|
| 175 |
+
"loss": 0.1121,
|
| 176 |
+
"step": 14000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 0.08,
|
| 180 |
+
"learning_rate": 1.9470116391675345e-05,
|
| 181 |
+
"loss": 0.1094,
|
| 182 |
+
"step": 14500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.08,
|
| 186 |
+
"learning_rate": 1.945184454311243e-05,
|
| 187 |
+
"loss": 0.0955,
|
| 188 |
+
"step": 15000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 0.08,
|
| 192 |
+
"learning_rate": 1.943357269454951e-05,
|
| 193 |
+
"loss": 0.1061,
|
| 194 |
+
"step": 15500
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.09,
|
| 198 |
+
"learning_rate": 1.941530084598659e-05,
|
| 199 |
+
"loss": 0.1049,
|
| 200 |
+
"step": 16000
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.09,
|
| 204 |
+
"learning_rate": 1.939702899742367e-05,
|
| 205 |
+
"loss": 0.1018,
|
| 206 |
+
"step": 16500
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 0.09,
|
| 210 |
+
"learning_rate": 1.9378757148860752e-05,
|
| 211 |
+
"loss": 0.1117,
|
| 212 |
+
"step": 17000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.1,
|
| 216 |
+
"learning_rate": 1.9360485300297833e-05,
|
| 217 |
+
"loss": 0.0995,
|
| 218 |
+
"step": 17500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.1,
|
| 222 |
+
"learning_rate": 1.9342213451734914e-05,
|
| 223 |
+
"loss": 0.1019,
|
| 224 |
+
"step": 18000
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 0.1,
|
| 228 |
+
"learning_rate": 1.9323941603171994e-05,
|
| 229 |
+
"loss": 0.0892,
|
| 230 |
+
"step": 18500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.1,
|
| 234 |
+
"learning_rate": 1.9305669754609075e-05,
|
| 235 |
+
"loss": 0.0982,
|
| 236 |
+
"step": 19000
|
| 237 |
+
},
|
| 238 |
+
{
|
| 239 |
+
"epoch": 0.11,
|
| 240 |
+
"learning_rate": 1.9287397906046156e-05,
|
| 241 |
+
"loss": 0.0943,
|
| 242 |
+
"step": 19500
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.11,
|
| 246 |
+
"learning_rate": 1.9269126057483236e-05,
|
| 247 |
+
"loss": 0.0885,
|
| 248 |
+
"step": 20000
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 0.11,
|
| 252 |
+
"learning_rate": 1.9250854208920317e-05,
|
| 253 |
+
"loss": 0.0956,
|
| 254 |
+
"step": 20500
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.12,
|
| 258 |
+
"learning_rate": 1.9232582360357398e-05,
|
| 259 |
+
"loss": 0.1101,
|
| 260 |
+
"step": 21000
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.12,
|
| 264 |
+
"learning_rate": 1.9214310511794482e-05,
|
| 265 |
+
"loss": 0.099,
|
| 266 |
+
"step": 21500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.12,
|
| 270 |
+
"learning_rate": 1.919603866323156e-05,
|
| 271 |
+
"loss": 0.0919,
|
| 272 |
+
"step": 22000
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.12,
|
| 276 |
+
"learning_rate": 1.917776681466864e-05,
|
| 277 |
+
"loss": 0.0944,
|
| 278 |
+
"step": 22500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.13,
|
| 282 |
+
"learning_rate": 1.9159494966105724e-05,
|
| 283 |
+
"loss": 0.1062,
|
| 284 |
+
"step": 23000
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.13,
|
| 288 |
+
"learning_rate": 1.9141223117542805e-05,
|
| 289 |
+
"loss": 0.0833,
|
| 290 |
+
"step": 23500
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.13,
|
| 294 |
+
"learning_rate": 1.9122951268979882e-05,
|
| 295 |
+
"loss": 0.1055,
|
| 296 |
+
"step": 24000
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.13,
|
| 300 |
+
"learning_rate": 1.9104679420416966e-05,
|
| 301 |
+
"loss": 0.0978,
|
| 302 |
+
"step": 24500
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.14,
|
| 306 |
+
"learning_rate": 1.9086407571854047e-05,
|
| 307 |
+
"loss": 0.0844,
|
| 308 |
+
"step": 25000
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.14,
|
| 312 |
+
"learning_rate": 1.9068135723291127e-05,
|
| 313 |
+
"loss": 0.092,
|
| 314 |
+
"step": 25500
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.14,
|
| 318 |
+
"learning_rate": 1.9049863874728208e-05,
|
| 319 |
+
"loss": 0.1131,
|
| 320 |
+
"step": 26000
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.15,
|
| 324 |
+
"learning_rate": 1.903159202616529e-05,
|
| 325 |
+
"loss": 0.0858,
|
| 326 |
+
"step": 26500
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.15,
|
| 330 |
+
"learning_rate": 1.901332017760237e-05,
|
| 331 |
+
"loss": 0.093,
|
| 332 |
+
"step": 27000
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.15,
|
| 336 |
+
"learning_rate": 1.899504832903945e-05,
|
| 337 |
+
"loss": 0.086,
|
| 338 |
+
"step": 27500
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.15,
|
| 342 |
+
"learning_rate": 1.897677648047653e-05,
|
| 343 |
+
"loss": 0.0785,
|
| 344 |
+
"step": 28000
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.16,
|
| 348 |
+
"learning_rate": 1.895850463191361e-05,
|
| 349 |
+
"loss": 0.0949,
|
| 350 |
+
"step": 28500
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.16,
|
| 354 |
+
"learning_rate": 1.8940232783350692e-05,
|
| 355 |
+
"loss": 0.1032,
|
| 356 |
+
"step": 29000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.16,
|
| 360 |
+
"learning_rate": 1.8921960934787773e-05,
|
| 361 |
+
"loss": 0.0887,
|
| 362 |
+
"step": 29500
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.16,
|
| 366 |
+
"learning_rate": 1.8903689086224854e-05,
|
| 367 |
+
"loss": 0.102,
|
| 368 |
+
"step": 30000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.17,
|
| 372 |
+
"learning_rate": 1.8885417237661934e-05,
|
| 373 |
+
"loss": 0.099,
|
| 374 |
+
"step": 30500
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.17,
|
| 378 |
+
"learning_rate": 1.886714538909902e-05,
|
| 379 |
+
"loss": 0.0886,
|
| 380 |
+
"step": 31000
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.17,
|
| 384 |
+
"learning_rate": 1.8848873540536096e-05,
|
| 385 |
+
"loss": 0.0955,
|
| 386 |
+
"step": 31500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.18,
|
| 390 |
+
"learning_rate": 1.8830601691973177e-05,
|
| 391 |
+
"loss": 0.0786,
|
| 392 |
+
"step": 32000
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 0.18,
|
| 396 |
+
"learning_rate": 1.881232984341026e-05,
|
| 397 |
+
"loss": 0.109,
|
| 398 |
+
"step": 32500
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.18,
|
| 402 |
+
"learning_rate": 1.879405799484734e-05,
|
| 403 |
+
"loss": 0.0798,
|
| 404 |
+
"step": 33000
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 0.18,
|
| 408 |
+
"learning_rate": 1.8775786146284422e-05,
|
| 409 |
+
"loss": 0.075,
|
| 410 |
+
"step": 33500
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 0.19,
|
| 414 |
+
"learning_rate": 1.8757514297721503e-05,
|
| 415 |
+
"loss": 0.1025,
|
| 416 |
+
"step": 34000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.19,
|
| 420 |
+
"learning_rate": 1.8739242449158583e-05,
|
| 421 |
+
"loss": 0.0832,
|
| 422 |
+
"step": 34500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.19,
|
| 426 |
+
"learning_rate": 1.8720970600595664e-05,
|
| 427 |
+
"loss": 0.0942,
|
| 428 |
+
"step": 35000
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.19,
|
| 432 |
+
"learning_rate": 1.8702698752032745e-05,
|
| 433 |
+
"loss": 0.0905,
|
| 434 |
+
"step": 35500
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.2,
|
| 438 |
+
"learning_rate": 1.8684426903469826e-05,
|
| 439 |
+
"loss": 0.0967,
|
| 440 |
+
"step": 36000
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.2,
|
| 444 |
+
"learning_rate": 1.8666155054906906e-05,
|
| 445 |
+
"loss": 0.1022,
|
| 446 |
+
"step": 36500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 0.2,
|
| 450 |
+
"learning_rate": 1.8647883206343987e-05,
|
| 451 |
+
"loss": 0.091,
|
| 452 |
+
"step": 37000
|
| 453 |
+
},
|
| 454 |
+
{
|
| 455 |
+
"epoch": 0.21,
|
| 456 |
+
"learning_rate": 1.8629611357781068e-05,
|
| 457 |
+
"loss": 0.1029,
|
| 458 |
+
"step": 37500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 0.21,
|
| 462 |
+
"learning_rate": 1.861133950921815e-05,
|
| 463 |
+
"loss": 0.0867,
|
| 464 |
+
"step": 38000
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.21,
|
| 468 |
+
"learning_rate": 1.8593067660655232e-05,
|
| 469 |
+
"loss": 0.1072,
|
| 470 |
+
"step": 38500
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 0.21,
|
| 474 |
+
"learning_rate": 1.857479581209231e-05,
|
| 475 |
+
"loss": 0.1039,
|
| 476 |
+
"step": 39000
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 0.22,
|
| 480 |
+
"learning_rate": 1.855652396352939e-05,
|
| 481 |
+
"loss": 0.0988,
|
| 482 |
+
"step": 39500
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.22,
|
| 486 |
+
"learning_rate": 1.8538252114966475e-05,
|
| 487 |
+
"loss": 0.0769,
|
| 488 |
+
"step": 40000
|
| 489 |
+
},
|
| 490 |
+
{
|
| 491 |
+
"epoch": 0.22,
|
| 492 |
+
"learning_rate": 1.8519980266403555e-05,
|
| 493 |
+
"loss": 0.0818,
|
| 494 |
+
"step": 40500
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.22,
|
| 498 |
+
"learning_rate": 1.8501708417840633e-05,
|
| 499 |
+
"loss": 0.1,
|
| 500 |
+
"step": 41000
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 0.23,
|
| 504 |
+
"learning_rate": 1.8483436569277717e-05,
|
| 505 |
+
"loss": 0.1093,
|
| 506 |
+
"step": 41500
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 0.23,
|
| 510 |
+
"learning_rate": 1.8465164720714797e-05,
|
| 511 |
+
"loss": 0.0806,
|
| 512 |
+
"step": 42000
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 0.23,
|
| 516 |
+
"learning_rate": 1.8446892872151878e-05,
|
| 517 |
+
"loss": 0.0877,
|
| 518 |
+
"step": 42500
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.24,
|
| 522 |
+
"learning_rate": 1.842862102358896e-05,
|
| 523 |
+
"loss": 0.1024,
|
| 524 |
+
"step": 43000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.24,
|
| 528 |
+
"learning_rate": 1.841034917502604e-05,
|
| 529 |
+
"loss": 0.0899,
|
| 530 |
+
"step": 43500
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.24,
|
| 534 |
+
"learning_rate": 1.839207732646312e-05,
|
| 535 |
+
"loss": 0.0768,
|
| 536 |
+
"step": 44000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.24,
|
| 540 |
+
"learning_rate": 1.83738054779002e-05,
|
| 541 |
+
"loss": 0.0896,
|
| 542 |
+
"step": 44500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.25,
|
| 546 |
+
"learning_rate": 1.835553362933728e-05,
|
| 547 |
+
"loss": 0.0875,
|
| 548 |
+
"step": 45000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.25,
|
| 552 |
+
"learning_rate": 1.8337261780774362e-05,
|
| 553 |
+
"loss": 0.0855,
|
| 554 |
+
"step": 45500
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.25,
|
| 558 |
+
"learning_rate": 1.8318989932211443e-05,
|
| 559 |
+
"loss": 0.1109,
|
| 560 |
+
"step": 46000
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.25,
|
| 564 |
+
"learning_rate": 1.8300718083648524e-05,
|
| 565 |
+
"loss": 0.0956,
|
| 566 |
+
"step": 46500
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.26,
|
| 570 |
+
"learning_rate": 1.8282446235085604e-05,
|
| 571 |
+
"loss": 0.0936,
|
| 572 |
+
"step": 47000
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.26,
|
| 576 |
+
"learning_rate": 1.8264174386522685e-05,
|
| 577 |
+
"loss": 0.0734,
|
| 578 |
+
"step": 47500
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.26,
|
| 582 |
+
"learning_rate": 1.824590253795977e-05,
|
| 583 |
+
"loss": 0.0805,
|
| 584 |
+
"step": 48000
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.27,
|
| 588 |
+
"learning_rate": 1.8227630689396846e-05,
|
| 589 |
+
"loss": 0.106,
|
| 590 |
+
"step": 48500
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.27,
|
| 594 |
+
"learning_rate": 1.8209358840833927e-05,
|
| 595 |
+
"loss": 0.0693,
|
| 596 |
+
"step": 49000
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.27,
|
| 600 |
+
"learning_rate": 1.819108699227101e-05,
|
| 601 |
+
"loss": 0.0893,
|
| 602 |
+
"step": 49500
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.27,
|
| 606 |
+
"learning_rate": 1.8172815143708092e-05,
|
| 607 |
+
"loss": 0.096,
|
| 608 |
+
"step": 50000
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.28,
|
| 612 |
+
"learning_rate": 1.815454329514517e-05,
|
| 613 |
+
"loss": 0.0879,
|
| 614 |
+
"step": 50500
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.28,
|
| 618 |
+
"learning_rate": 1.8136271446582253e-05,
|
| 619 |
+
"loss": 0.0934,
|
| 620 |
+
"step": 51000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.28,
|
| 624 |
+
"learning_rate": 1.8117999598019334e-05,
|
| 625 |
+
"loss": 0.0797,
|
| 626 |
+
"step": 51500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.29,
|
| 630 |
+
"learning_rate": 1.8099727749456415e-05,
|
| 631 |
+
"loss": 0.0825,
|
| 632 |
+
"step": 52000
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.29,
|
| 636 |
+
"learning_rate": 1.8081455900893495e-05,
|
| 637 |
+
"loss": 0.0926,
|
| 638 |
+
"step": 52500
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.29,
|
| 642 |
+
"learning_rate": 1.8063184052330576e-05,
|
| 643 |
+
"loss": 0.0796,
|
| 644 |
+
"step": 53000
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 0.29,
|
| 648 |
+
"learning_rate": 1.8044912203767657e-05,
|
| 649 |
+
"loss": 0.0604,
|
| 650 |
+
"step": 53500
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 0.3,
|
| 654 |
+
"learning_rate": 1.8026640355204737e-05,
|
| 655 |
+
"loss": 0.0844,
|
| 656 |
+
"step": 54000
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 0.3,
|
| 660 |
+
"learning_rate": 1.8008368506641818e-05,
|
| 661 |
+
"loss": 0.0701,
|
| 662 |
+
"step": 54500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.3,
|
| 666 |
+
"learning_rate": 1.79900966580789e-05,
|
| 667 |
+
"loss": 0.0764,
|
| 668 |
+
"step": 55000
|
| 669 |
+
},
|
| 670 |
+
{
|
| 671 |
+
"epoch": 0.3,
|
| 672 |
+
"learning_rate": 1.797182480951598e-05,
|
| 673 |
+
"loss": 0.0856,
|
| 674 |
+
"step": 55500
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 0.31,
|
| 678 |
+
"learning_rate": 1.795355296095306e-05,
|
| 679 |
+
"loss": 0.0933,
|
| 680 |
+
"step": 56000
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 0.31,
|
| 684 |
+
"learning_rate": 1.793528111239014e-05,
|
| 685 |
+
"loss": 0.0891,
|
| 686 |
+
"step": 56500
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 0.31,
|
| 690 |
+
"learning_rate": 1.791700926382722e-05,
|
| 691 |
+
"loss": 0.0736,
|
| 692 |
+
"step": 57000
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 0.32,
|
| 696 |
+
"learning_rate": 1.7898737415264306e-05,
|
| 697 |
+
"loss": 0.0875,
|
| 698 |
+
"step": 57500
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 0.32,
|
| 702 |
+
"learning_rate": 1.7880465566701383e-05,
|
| 703 |
+
"loss": 0.0802,
|
| 704 |
+
"step": 58000
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.32,
|
| 708 |
+
"learning_rate": 1.7862193718138464e-05,
|
| 709 |
+
"loss": 0.0759,
|
| 710 |
+
"step": 58500
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 0.32,
|
| 714 |
+
"learning_rate": 1.7843921869575548e-05,
|
| 715 |
+
"loss": 0.0957,
|
| 716 |
+
"step": 59000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 0.33,
|
| 720 |
+
"learning_rate": 1.782565002101263e-05,
|
| 721 |
+
"loss": 0.0853,
|
| 722 |
+
"step": 59500
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.33,
|
| 726 |
+
"learning_rate": 1.7807378172449706e-05,
|
| 727 |
+
"loss": 0.0829,
|
| 728 |
+
"step": 60000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.33,
|
| 732 |
+
"learning_rate": 1.778910632388679e-05,
|
| 733 |
+
"loss": 0.0798,
|
| 734 |
+
"step": 60500
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 0.33,
|
| 738 |
+
"learning_rate": 1.777083447532387e-05,
|
| 739 |
+
"loss": 0.0881,
|
| 740 |
+
"step": 61000
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"epoch": 0.34,
|
| 744 |
+
"learning_rate": 1.775256262676095e-05,
|
| 745 |
+
"loss": 0.0842,
|
| 746 |
+
"step": 61500
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.34,
|
| 750 |
+
"learning_rate": 1.7734290778198032e-05,
|
| 751 |
+
"loss": 0.0722,
|
| 752 |
+
"step": 62000
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 0.34,
|
| 756 |
+
"learning_rate": 1.7716018929635113e-05,
|
| 757 |
+
"loss": 0.081,
|
| 758 |
+
"step": 62500
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.35,
|
| 762 |
+
"learning_rate": 1.7697747081072193e-05,
|
| 763 |
+
"loss": 0.0696,
|
| 764 |
+
"step": 63000
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 0.35,
|
| 768 |
+
"learning_rate": 1.7679475232509274e-05,
|
| 769 |
+
"loss": 0.098,
|
| 770 |
+
"step": 63500
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 0.35,
|
| 774 |
+
"learning_rate": 1.7661203383946355e-05,
|
| 775 |
+
"loss": 0.067,
|
| 776 |
+
"step": 64000
|
| 777 |
+
},
|
| 778 |
+
{
|
| 779 |
+
"epoch": 0.35,
|
| 780 |
+
"learning_rate": 1.7642931535383436e-05,
|
| 781 |
+
"loss": 0.0674,
|
| 782 |
+
"step": 64500
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 0.36,
|
| 786 |
+
"learning_rate": 1.7624659686820516e-05,
|
| 787 |
+
"loss": 0.0803,
|
| 788 |
+
"step": 65000
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 0.36,
|
| 792 |
+
"learning_rate": 1.7606387838257597e-05,
|
| 793 |
+
"loss": 0.0715,
|
| 794 |
+
"step": 65500
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 0.36,
|
| 798 |
+
"learning_rate": 1.7588115989694678e-05,
|
| 799 |
+
"loss": 0.0857,
|
| 800 |
+
"step": 66000
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.36,
|
| 804 |
+
"learning_rate": 1.756984414113176e-05,
|
| 805 |
+
"loss": 0.0849,
|
| 806 |
+
"step": 66500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 0.37,
|
| 810 |
+
"learning_rate": 1.7551572292568842e-05,
|
| 811 |
+
"loss": 0.08,
|
| 812 |
+
"step": 67000
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 0.37,
|
| 816 |
+
"learning_rate": 1.753330044400592e-05,
|
| 817 |
+
"loss": 0.0651,
|
| 818 |
+
"step": 67500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 0.37,
|
| 822 |
+
"learning_rate": 1.7515028595443004e-05,
|
| 823 |
+
"loss": 0.0664,
|
| 824 |
+
"step": 68000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.38,
|
| 828 |
+
"learning_rate": 1.7496756746880085e-05,
|
| 829 |
+
"loss": 0.0555,
|
| 830 |
+
"step": 68500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.38,
|
| 834 |
+
"learning_rate": 1.7478484898317165e-05,
|
| 835 |
+
"loss": 0.0775,
|
| 836 |
+
"step": 69000
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 0.38,
|
| 840 |
+
"learning_rate": 1.7460213049754246e-05,
|
| 841 |
+
"loss": 0.0731,
|
| 842 |
+
"step": 69500
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.38,
|
| 846 |
+
"learning_rate": 1.7441941201191327e-05,
|
| 847 |
+
"loss": 0.0723,
|
| 848 |
+
"step": 70000
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 0.39,
|
| 852 |
+
"learning_rate": 1.7423669352628407e-05,
|
| 853 |
+
"loss": 0.0674,
|
| 854 |
+
"step": 70500
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 0.39,
|
| 858 |
+
"learning_rate": 1.7405397504065488e-05,
|
| 859 |
+
"loss": 0.0636,
|
| 860 |
+
"step": 71000
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 0.39,
|
| 864 |
+
"learning_rate": 1.738712565550257e-05,
|
| 865 |
+
"loss": 0.0855,
|
| 866 |
+
"step": 71500
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 0.39,
|
| 870 |
+
"learning_rate": 1.736885380693965e-05,
|
| 871 |
+
"loss": 0.0696,
|
| 872 |
+
"step": 72000
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 0.4,
|
| 876 |
+
"learning_rate": 1.735058195837673e-05,
|
| 877 |
+
"loss": 0.0647,
|
| 878 |
+
"step": 72500
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 0.4,
|
| 882 |
+
"learning_rate": 1.733231010981381e-05,
|
| 883 |
+
"loss": 0.0649,
|
| 884 |
+
"step": 73000
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.4,
|
| 888 |
+
"learning_rate": 1.731403826125089e-05,
|
| 889 |
+
"loss": 0.077,
|
| 890 |
+
"step": 73500
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 0.41,
|
| 894 |
+
"learning_rate": 1.7295766412687972e-05,
|
| 895 |
+
"loss": 0.0663,
|
| 896 |
+
"step": 74000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 0.41,
|
| 900 |
+
"learning_rate": 1.7277494564125056e-05,
|
| 901 |
+
"loss": 0.0675,
|
| 902 |
+
"step": 74500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 0.41,
|
| 906 |
+
"learning_rate": 1.7259222715562134e-05,
|
| 907 |
+
"loss": 0.0708,
|
| 908 |
+
"step": 75000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 0.41,
|
| 912 |
+
"learning_rate": 1.7240950866999214e-05,
|
| 913 |
+
"loss": 0.0805,
|
| 914 |
+
"step": 75500
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 0.42,
|
| 918 |
+
"learning_rate": 1.72226790184363e-05,
|
| 919 |
+
"loss": 0.0776,
|
| 920 |
+
"step": 76000
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 0.42,
|
| 924 |
+
"learning_rate": 1.720440716987338e-05,
|
| 925 |
+
"loss": 0.0604,
|
| 926 |
+
"step": 76500
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.42,
|
| 930 |
+
"learning_rate": 1.7186135321310456e-05,
|
| 931 |
+
"loss": 0.0612,
|
| 932 |
+
"step": 77000
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 0.42,
|
| 936 |
+
"learning_rate": 1.716786347274754e-05,
|
| 937 |
+
"loss": 0.0546,
|
| 938 |
+
"step": 77500
|
| 939 |
+
},
|
| 940 |
+
{
|
| 941 |
+
"epoch": 0.43,
|
| 942 |
+
"learning_rate": 1.714959162418462e-05,
|
| 943 |
+
"loss": 0.0753,
|
| 944 |
+
"step": 78000
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 0.43,
|
| 948 |
+
"learning_rate": 1.7131319775621702e-05,
|
| 949 |
+
"loss": 0.0688,
|
| 950 |
+
"step": 78500
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 0.43,
|
| 954 |
+
"learning_rate": 1.7113047927058783e-05,
|
| 955 |
+
"loss": 0.0603,
|
| 956 |
+
"step": 79000
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 0.44,
|
| 960 |
+
"learning_rate": 1.7094776078495863e-05,
|
| 961 |
+
"loss": 0.0679,
|
| 962 |
+
"step": 79500
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 0.44,
|
| 966 |
+
"learning_rate": 1.7076504229932944e-05,
|
| 967 |
+
"loss": 0.0715,
|
| 968 |
+
"step": 80000
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 0.44,
|
| 972 |
+
"learning_rate": 1.7058232381370025e-05,
|
| 973 |
+
"loss": 0.0655,
|
| 974 |
+
"step": 80500
|
| 975 |
+
},
|
| 976 |
+
{
|
| 977 |
+
"epoch": 0.44,
|
| 978 |
+
"learning_rate": 1.7039960532807105e-05,
|
| 979 |
+
"loss": 0.0744,
|
| 980 |
+
"step": 81000
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.45,
|
| 984 |
+
"learning_rate": 1.7021688684244186e-05,
|
| 985 |
+
"loss": 0.0636,
|
| 986 |
+
"step": 81500
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 0.45,
|
| 990 |
+
"learning_rate": 1.7003416835681267e-05,
|
| 991 |
+
"loss": 0.0838,
|
| 992 |
+
"step": 82000
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 0.45,
|
| 996 |
+
"learning_rate": 1.6985144987118347e-05,
|
| 997 |
+
"loss": 0.064,
|
| 998 |
+
"step": 82500
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 0.45,
|
| 1002 |
+
"learning_rate": 1.6966873138555428e-05,
|
| 1003 |
+
"loss": 0.0632,
|
| 1004 |
+
"step": 83000
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 0.46,
|
| 1008 |
+
"learning_rate": 1.694860128999251e-05,
|
| 1009 |
+
"loss": 0.069,
|
| 1010 |
+
"step": 83500
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 0.46,
|
| 1014 |
+
"learning_rate": 1.6930329441429593e-05,
|
| 1015 |
+
"loss": 0.0617,
|
| 1016 |
+
"step": 84000
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 0.46,
|
| 1020 |
+
"learning_rate": 1.691205759286667e-05,
|
| 1021 |
+
"loss": 0.0716,
|
| 1022 |
+
"step": 84500
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.47,
|
| 1026 |
+
"learning_rate": 1.689378574430375e-05,
|
| 1027 |
+
"loss": 0.075,
|
| 1028 |
+
"step": 85000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.47,
|
| 1032 |
+
"learning_rate": 1.6875513895740835e-05,
|
| 1033 |
+
"loss": 0.0729,
|
| 1034 |
+
"step": 85500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.47,
|
| 1038 |
+
"learning_rate": 1.6857242047177916e-05,
|
| 1039 |
+
"loss": 0.0669,
|
| 1040 |
+
"step": 86000
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 0.47,
|
| 1044 |
+
"learning_rate": 1.6838970198614993e-05,
|
| 1045 |
+
"loss": 0.0664,
|
| 1046 |
+
"step": 86500
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 0.48,
|
| 1050 |
+
"learning_rate": 1.6820698350052077e-05,
|
| 1051 |
+
"loss": 0.077,
|
| 1052 |
+
"step": 87000
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 0.48,
|
| 1056 |
+
"learning_rate": 1.6802426501489158e-05,
|
| 1057 |
+
"loss": 0.0781,
|
| 1058 |
+
"step": 87500
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 0.48,
|
| 1062 |
+
"learning_rate": 1.678415465292624e-05,
|
| 1063 |
+
"loss": 0.0666,
|
| 1064 |
+
"step": 88000
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 0.49,
|
| 1068 |
+
"learning_rate": 1.676588280436332e-05,
|
| 1069 |
+
"loss": 0.0657,
|
| 1070 |
+
"step": 88500
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 0.49,
|
| 1074 |
+
"learning_rate": 1.67476109558004e-05,
|
| 1075 |
+
"loss": 0.0612,
|
| 1076 |
+
"step": 89000
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 0.49,
|
| 1080 |
+
"learning_rate": 1.672933910723748e-05,
|
| 1081 |
+
"loss": 0.0694,
|
| 1082 |
+
"step": 89500
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 0.49,
|
| 1086 |
+
"learning_rate": 1.671106725867456e-05,
|
| 1087 |
+
"loss": 0.0671,
|
| 1088 |
+
"step": 90000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 0.5,
|
| 1092 |
+
"learning_rate": 1.6692795410111642e-05,
|
| 1093 |
+
"loss": 0.0729,
|
| 1094 |
+
"step": 90500
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 0.5,
|
| 1098 |
+
"learning_rate": 1.6674523561548723e-05,
|
| 1099 |
+
"loss": 0.0723,
|
| 1100 |
+
"step": 91000
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 0.5,
|
| 1104 |
+
"learning_rate": 1.6656251712985803e-05,
|
| 1105 |
+
"loss": 0.0777,
|
| 1106 |
+
"step": 91500
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 0.5,
|
| 1110 |
+
"learning_rate": 1.6637979864422884e-05,
|
| 1111 |
+
"loss": 0.0798,
|
| 1112 |
+
"step": 92000
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 0.51,
|
| 1116 |
+
"learning_rate": 1.6619708015859965e-05,
|
| 1117 |
+
"loss": 0.0548,
|
| 1118 |
+
"step": 92500
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 0.51,
|
| 1122 |
+
"learning_rate": 1.6601436167297046e-05,
|
| 1123 |
+
"loss": 0.0646,
|
| 1124 |
+
"step": 93000
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 0.51,
|
| 1128 |
+
"learning_rate": 1.658316431873413e-05,
|
| 1129 |
+
"loss": 0.0639,
|
| 1130 |
+
"step": 93500
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.52,
|
| 1134 |
+
"learning_rate": 1.6564892470171207e-05,
|
| 1135 |
+
"loss": 0.0529,
|
| 1136 |
+
"step": 94000
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.52,
|
| 1140 |
+
"learning_rate": 1.6546620621608288e-05,
|
| 1141 |
+
"loss": 0.0577,
|
| 1142 |
+
"step": 94500
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 0.52,
|
| 1146 |
+
"learning_rate": 1.6528348773045372e-05,
|
| 1147 |
+
"loss": 0.0632,
|
| 1148 |
+
"step": 95000
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 0.52,
|
| 1152 |
+
"learning_rate": 1.6510076924482452e-05,
|
| 1153 |
+
"loss": 0.0463,
|
| 1154 |
+
"step": 95500
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 0.53,
|
| 1158 |
+
"learning_rate": 1.649180507591953e-05,
|
| 1159 |
+
"loss": 0.0661,
|
| 1160 |
+
"step": 96000
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.53,
|
| 1164 |
+
"learning_rate": 1.6473533227356614e-05,
|
| 1165 |
+
"loss": 0.0742,
|
| 1166 |
+
"step": 96500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 0.53,
|
| 1170 |
+
"learning_rate": 1.6455261378793695e-05,
|
| 1171 |
+
"loss": 0.0549,
|
| 1172 |
+
"step": 97000
|
| 1173 |
+
},
|
| 1174 |
+
{
|
| 1175 |
+
"epoch": 0.53,
|
| 1176 |
+
"learning_rate": 1.6436989530230775e-05,
|
| 1177 |
+
"loss": 0.064,
|
| 1178 |
+
"step": 97500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 0.54,
|
| 1182 |
+
"learning_rate": 1.6418717681667856e-05,
|
| 1183 |
+
"loss": 0.0597,
|
| 1184 |
+
"step": 98000
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 0.54,
|
| 1188 |
+
"learning_rate": 1.6400445833104937e-05,
|
| 1189 |
+
"loss": 0.0795,
|
| 1190 |
+
"step": 98500
|
| 1191 |
+
},
|
| 1192 |
+
{
|
| 1193 |
+
"epoch": 0.54,
|
| 1194 |
+
"learning_rate": 1.6382173984542017e-05,
|
| 1195 |
+
"loss": 0.0658,
|
| 1196 |
+
"step": 99000
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 0.55,
|
| 1200 |
+
"learning_rate": 1.6363902135979098e-05,
|
| 1201 |
+
"loss": 0.0574,
|
| 1202 |
+
"step": 99500
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 0.55,
|
| 1206 |
+
"learning_rate": 1.634563028741618e-05,
|
| 1207 |
+
"loss": 0.0641,
|
| 1208 |
+
"step": 100000
|
| 1209 |
+
},
|
| 1210 |
+
{
|
| 1211 |
+
"epoch": 0.55,
|
| 1212 |
+
"learning_rate": 1.632735843885326e-05,
|
| 1213 |
+
"loss": 0.0594,
|
| 1214 |
+
"step": 100500
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 0.55,
|
| 1218 |
+
"learning_rate": 1.630908659029034e-05,
|
| 1219 |
+
"loss": 0.0572,
|
| 1220 |
+
"step": 101000
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 0.56,
|
| 1224 |
+
"learning_rate": 1.629081474172742e-05,
|
| 1225 |
+
"loss": 0.067,
|
| 1226 |
+
"step": 101500
|
| 1227 |
+
},
|
| 1228 |
+
{
|
| 1229 |
+
"epoch": 0.56,
|
| 1230 |
+
"learning_rate": 1.62725428931645e-05,
|
| 1231 |
+
"loss": 0.0617,
|
| 1232 |
+
"step": 102000
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 0.56,
|
| 1236 |
+
"learning_rate": 1.6254271044601582e-05,
|
| 1237 |
+
"loss": 0.0689,
|
| 1238 |
+
"step": 102500
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 0.56,
|
| 1242 |
+
"learning_rate": 1.6235999196038666e-05,
|
| 1243 |
+
"loss": 0.0676,
|
| 1244 |
+
"step": 103000
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.57,
|
| 1248 |
+
"learning_rate": 1.6217727347475744e-05,
|
| 1249 |
+
"loss": 0.067,
|
| 1250 |
+
"step": 103500
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 0.57,
|
| 1254 |
+
"learning_rate": 1.6199455498912828e-05,
|
| 1255 |
+
"loss": 0.0527,
|
| 1256 |
+
"step": 104000
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 0.57,
|
| 1260 |
+
"learning_rate": 1.618118365034991e-05,
|
| 1261 |
+
"loss": 0.0581,
|
| 1262 |
+
"step": 104500
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 0.58,
|
| 1266 |
+
"learning_rate": 1.616291180178699e-05,
|
| 1267 |
+
"loss": 0.0647,
|
| 1268 |
+
"step": 105000
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 0.58,
|
| 1272 |
+
"learning_rate": 1.614463995322407e-05,
|
| 1273 |
+
"loss": 0.0826,
|
| 1274 |
+
"step": 105500
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 0.58,
|
| 1278 |
+
"learning_rate": 1.612636810466115e-05,
|
| 1279 |
+
"loss": 0.0654,
|
| 1280 |
+
"step": 106000
|
| 1281 |
+
},
|
| 1282 |
+
{
|
| 1283 |
+
"epoch": 0.58,
|
| 1284 |
+
"learning_rate": 1.610809625609823e-05,
|
| 1285 |
+
"loss": 0.066,
|
| 1286 |
+
"step": 106500
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 0.59,
|
| 1290 |
+
"learning_rate": 1.6089824407535312e-05,
|
| 1291 |
+
"loss": 0.056,
|
| 1292 |
+
"step": 107000
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 0.59,
|
| 1296 |
+
"learning_rate": 1.6071552558972393e-05,
|
| 1297 |
+
"loss": 0.0715,
|
| 1298 |
+
"step": 107500
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 0.59,
|
| 1302 |
+
"learning_rate": 1.6053280710409473e-05,
|
| 1303 |
+
"loss": 0.0664,
|
| 1304 |
+
"step": 108000
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 0.59,
|
| 1308 |
+
"learning_rate": 1.6035008861846554e-05,
|
| 1309 |
+
"loss": 0.072,
|
| 1310 |
+
"step": 108500
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 0.6,
|
| 1314 |
+
"learning_rate": 1.6016737013283635e-05,
|
| 1315 |
+
"loss": 0.065,
|
| 1316 |
+
"step": 109000
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 0.6,
|
| 1320 |
+
"learning_rate": 1.5998465164720715e-05,
|
| 1321 |
+
"loss": 0.0648,
|
| 1322 |
+
"step": 109500
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 0.6,
|
| 1326 |
+
"learning_rate": 1.5980193316157796e-05,
|
| 1327 |
+
"loss": 0.0874,
|
| 1328 |
+
"step": 110000
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 0.61,
|
| 1332 |
+
"learning_rate": 1.596192146759488e-05,
|
| 1333 |
+
"loss": 0.0721,
|
| 1334 |
+
"step": 110500
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.61,
|
| 1338 |
+
"learning_rate": 1.5943649619031958e-05,
|
| 1339 |
+
"loss": 0.0536,
|
| 1340 |
+
"step": 111000
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.61,
|
| 1344 |
+
"learning_rate": 1.5925377770469038e-05,
|
| 1345 |
+
"loss": 0.0591,
|
| 1346 |
+
"step": 111500
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 0.61,
|
| 1350 |
+
"learning_rate": 1.5907105921906122e-05,
|
| 1351 |
+
"loss": 0.0627,
|
| 1352 |
+
"step": 112000
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 0.62,
|
| 1356 |
+
"learning_rate": 1.5888834073343203e-05,
|
| 1357 |
+
"loss": 0.0583,
|
| 1358 |
+
"step": 112500
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 0.62,
|
| 1362 |
+
"learning_rate": 1.587056222478028e-05,
|
| 1363 |
+
"loss": 0.0569,
|
| 1364 |
+
"step": 113000
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 0.62,
|
| 1368 |
+
"learning_rate": 1.5852290376217364e-05,
|
| 1369 |
+
"loss": 0.0597,
|
| 1370 |
+
"step": 113500
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 0.62,
|
| 1374 |
+
"learning_rate": 1.5834018527654445e-05,
|
| 1375 |
+
"loss": 0.0672,
|
| 1376 |
+
"step": 114000
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 0.63,
|
| 1380 |
+
"learning_rate": 1.5815746679091526e-05,
|
| 1381 |
+
"loss": 0.0565,
|
| 1382 |
+
"step": 114500
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 0.63,
|
| 1386 |
+
"learning_rate": 1.5797474830528606e-05,
|
| 1387 |
+
"loss": 0.053,
|
| 1388 |
+
"step": 115000
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 0.63,
|
| 1392 |
+
"learning_rate": 1.5779202981965687e-05,
|
| 1393 |
+
"loss": 0.0535,
|
| 1394 |
+
"step": 115500
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 0.64,
|
| 1398 |
+
"learning_rate": 1.5760931133402768e-05,
|
| 1399 |
+
"loss": 0.0645,
|
| 1400 |
+
"step": 116000
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 0.64,
|
| 1404 |
+
"learning_rate": 1.574265928483985e-05,
|
| 1405 |
+
"loss": 0.059,
|
| 1406 |
+
"step": 116500
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 0.64,
|
| 1410 |
+
"learning_rate": 1.572438743627693e-05,
|
| 1411 |
+
"loss": 0.0703,
|
| 1412 |
+
"step": 117000
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 0.64,
|
| 1416 |
+
"learning_rate": 1.570611558771401e-05,
|
| 1417 |
+
"loss": 0.0471,
|
| 1418 |
+
"step": 117500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 0.65,
|
| 1422 |
+
"learning_rate": 1.568784373915109e-05,
|
| 1423 |
+
"loss": 0.044,
|
| 1424 |
+
"step": 118000
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 0.65,
|
| 1428 |
+
"learning_rate": 1.566957189058817e-05,
|
| 1429 |
+
"loss": 0.0639,
|
| 1430 |
+
"step": 118500
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 0.65,
|
| 1434 |
+
"learning_rate": 1.5651300042025252e-05,
|
| 1435 |
+
"loss": 0.0551,
|
| 1436 |
+
"step": 119000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 0.66,
|
| 1440 |
+
"learning_rate": 1.5633028193462333e-05,
|
| 1441 |
+
"loss": 0.0573,
|
| 1442 |
+
"step": 119500
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.66,
|
| 1446 |
+
"learning_rate": 1.5614756344899417e-05,
|
| 1447 |
+
"loss": 0.0453,
|
| 1448 |
+
"step": 120000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 0.66,
|
| 1452 |
+
"learning_rate": 1.5596484496336494e-05,
|
| 1453 |
+
"loss": 0.063,
|
| 1454 |
+
"step": 120500
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 0.66,
|
| 1458 |
+
"learning_rate": 1.5578212647773575e-05,
|
| 1459 |
+
"loss": 0.0611,
|
| 1460 |
+
"step": 121000
|
| 1461 |
+
},
|
| 1462 |
+
{
|
| 1463 |
+
"epoch": 0.67,
|
| 1464 |
+
"learning_rate": 1.555994079921066e-05,
|
| 1465 |
+
"loss": 0.0519,
|
| 1466 |
+
"step": 121500
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 0.67,
|
| 1470 |
+
"learning_rate": 1.554166895064774e-05,
|
| 1471 |
+
"loss": 0.0636,
|
| 1472 |
+
"step": 122000
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 0.67,
|
| 1476 |
+
"learning_rate": 1.5523397102084817e-05,
|
| 1477 |
+
"loss": 0.0803,
|
| 1478 |
+
"step": 122500
|
| 1479 |
+
},
|
| 1480 |
+
{
|
| 1481 |
+
"epoch": 0.67,
|
| 1482 |
+
"learning_rate": 1.55051252535219e-05,
|
| 1483 |
+
"loss": 0.0626,
|
| 1484 |
+
"step": 123000
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 0.68,
|
| 1488 |
+
"learning_rate": 1.5486853404958982e-05,
|
| 1489 |
+
"loss": 0.0476,
|
| 1490 |
+
"step": 123500
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 0.68,
|
| 1494 |
+
"learning_rate": 1.5468581556396062e-05,
|
| 1495 |
+
"loss": 0.0617,
|
| 1496 |
+
"step": 124000
|
| 1497 |
+
},
|
| 1498 |
+
{
|
| 1499 |
+
"epoch": 0.68,
|
| 1500 |
+
"learning_rate": 1.5450309707833143e-05,
|
| 1501 |
+
"loss": 0.0524,
|
| 1502 |
+
"step": 124500
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 0.69,
|
| 1506 |
+
"learning_rate": 1.5432037859270224e-05,
|
| 1507 |
+
"loss": 0.0498,
|
| 1508 |
+
"step": 125000
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 0.69,
|
| 1512 |
+
"learning_rate": 1.5413766010707305e-05,
|
| 1513 |
+
"loss": 0.0535,
|
| 1514 |
+
"step": 125500
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 0.69,
|
| 1518 |
+
"learning_rate": 1.5395494162144385e-05,
|
| 1519 |
+
"loss": 0.0497,
|
| 1520 |
+
"step": 126000
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 0.69,
|
| 1524 |
+
"learning_rate": 1.5377222313581466e-05,
|
| 1525 |
+
"loss": 0.0592,
|
| 1526 |
+
"step": 126500
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 0.7,
|
| 1530 |
+
"learning_rate": 1.5358950465018547e-05,
|
| 1531 |
+
"loss": 0.0443,
|
| 1532 |
+
"step": 127000
|
| 1533 |
+
},
|
| 1534 |
+
{
|
| 1535 |
+
"epoch": 0.7,
|
| 1536 |
+
"learning_rate": 1.5340678616455627e-05,
|
| 1537 |
+
"loss": 0.0547,
|
| 1538 |
+
"step": 127500
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 0.7,
|
| 1542 |
+
"learning_rate": 1.5322406767892708e-05,
|
| 1543 |
+
"loss": 0.06,
|
| 1544 |
+
"step": 128000
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.7,
|
| 1548 |
+
"learning_rate": 1.530413491932979e-05,
|
| 1549 |
+
"loss": 0.0482,
|
| 1550 |
+
"step": 128500
|
| 1551 |
+
},
|
| 1552 |
+
{
|
| 1553 |
+
"epoch": 0.71,
|
| 1554 |
+
"learning_rate": 1.528586307076687e-05,
|
| 1555 |
+
"loss": 0.0559,
|
| 1556 |
+
"step": 129000
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 0.71,
|
| 1560 |
+
"learning_rate": 1.5267591222203954e-05,
|
| 1561 |
+
"loss": 0.0533,
|
| 1562 |
+
"step": 129500
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 0.71,
|
| 1566 |
+
"learning_rate": 1.5249319373641031e-05,
|
| 1567 |
+
"loss": 0.0581,
|
| 1568 |
+
"step": 130000
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 0.72,
|
| 1572 |
+
"learning_rate": 1.5231047525078113e-05,
|
| 1573 |
+
"loss": 0.0615,
|
| 1574 |
+
"step": 130500
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 0.72,
|
| 1578 |
+
"learning_rate": 1.5212775676515194e-05,
|
| 1579 |
+
"loss": 0.0518,
|
| 1580 |
+
"step": 131000
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 0.72,
|
| 1584 |
+
"learning_rate": 1.5194503827952276e-05,
|
| 1585 |
+
"loss": 0.0436,
|
| 1586 |
+
"step": 131500
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 0.72,
|
| 1590 |
+
"learning_rate": 1.5176231979389355e-05,
|
| 1591 |
+
"loss": 0.0558,
|
| 1592 |
+
"step": 132000
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 0.73,
|
| 1596 |
+
"learning_rate": 1.5157960130826436e-05,
|
| 1597 |
+
"loss": 0.0472,
|
| 1598 |
+
"step": 132500
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 0.73,
|
| 1602 |
+
"learning_rate": 1.5139688282263518e-05,
|
| 1603 |
+
"loss": 0.0503,
|
| 1604 |
+
"step": 133000
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 0.73,
|
| 1608 |
+
"learning_rate": 1.5121416433700599e-05,
|
| 1609 |
+
"loss": 0.0482,
|
| 1610 |
+
"step": 133500
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 0.73,
|
| 1614 |
+
"learning_rate": 1.510314458513768e-05,
|
| 1615 |
+
"loss": 0.0584,
|
| 1616 |
+
"step": 134000
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 0.74,
|
| 1620 |
+
"learning_rate": 1.508487273657476e-05,
|
| 1621 |
+
"loss": 0.0632,
|
| 1622 |
+
"step": 134500
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 0.74,
|
| 1626 |
+
"learning_rate": 1.5066600888011841e-05,
|
| 1627 |
+
"loss": 0.0571,
|
| 1628 |
+
"step": 135000
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 0.74,
|
| 1632 |
+
"learning_rate": 1.5048329039448922e-05,
|
| 1633 |
+
"loss": 0.0524,
|
| 1634 |
+
"step": 135500
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 0.75,
|
| 1638 |
+
"learning_rate": 1.5030057190886003e-05,
|
| 1639 |
+
"loss": 0.0447,
|
| 1640 |
+
"step": 136000
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.75,
|
| 1644 |
+
"learning_rate": 1.5011785342323083e-05,
|
| 1645 |
+
"loss": 0.0541,
|
| 1646 |
+
"step": 136500
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 0.75,
|
| 1650 |
+
"learning_rate": 1.4993513493760166e-05,
|
| 1651 |
+
"loss": 0.0487,
|
| 1652 |
+
"step": 137000
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 0.75,
|
| 1656 |
+
"learning_rate": 1.4975241645197245e-05,
|
| 1657 |
+
"loss": 0.063,
|
| 1658 |
+
"step": 137500
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 0.76,
|
| 1662 |
+
"learning_rate": 1.4956969796634327e-05,
|
| 1663 |
+
"loss": 0.0576,
|
| 1664 |
+
"step": 138000
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 0.76,
|
| 1668 |
+
"learning_rate": 1.4938697948071408e-05,
|
| 1669 |
+
"loss": 0.0618,
|
| 1670 |
+
"step": 138500
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 0.76,
|
| 1674 |
+
"learning_rate": 1.4920426099508489e-05,
|
| 1675 |
+
"loss": 0.0439,
|
| 1676 |
+
"step": 139000
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 0.76,
|
| 1680 |
+
"learning_rate": 1.490215425094557e-05,
|
| 1681 |
+
"loss": 0.0455,
|
| 1682 |
+
"step": 139500
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 0.77,
|
| 1686 |
+
"learning_rate": 1.488388240238265e-05,
|
| 1687 |
+
"loss": 0.0581,
|
| 1688 |
+
"step": 140000
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 0.77,
|
| 1692 |
+
"learning_rate": 1.4865610553819732e-05,
|
| 1693 |
+
"loss": 0.0544,
|
| 1694 |
+
"step": 140500
|
| 1695 |
+
},
|
| 1696 |
+
{
|
| 1697 |
+
"epoch": 0.77,
|
| 1698 |
+
"learning_rate": 1.4847338705256813e-05,
|
| 1699 |
+
"loss": 0.0521,
|
| 1700 |
+
"step": 141000
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 0.78,
|
| 1704 |
+
"learning_rate": 1.4829066856693892e-05,
|
| 1705 |
+
"loss": 0.0454,
|
| 1706 |
+
"step": 141500
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 0.78,
|
| 1710 |
+
"learning_rate": 1.4810795008130974e-05,
|
| 1711 |
+
"loss": 0.0518,
|
| 1712 |
+
"step": 142000
|
| 1713 |
+
},
|
| 1714 |
+
{
|
| 1715 |
+
"epoch": 0.78,
|
| 1716 |
+
"learning_rate": 1.4792523159568055e-05,
|
| 1717 |
+
"loss": 0.0579,
|
| 1718 |
+
"step": 142500
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 0.78,
|
| 1722 |
+
"learning_rate": 1.4774251311005138e-05,
|
| 1723 |
+
"loss": 0.0357,
|
| 1724 |
+
"step": 143000
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 0.79,
|
| 1728 |
+
"learning_rate": 1.4755979462442217e-05,
|
| 1729 |
+
"loss": 0.0494,
|
| 1730 |
+
"step": 143500
|
| 1731 |
+
},
|
| 1732 |
+
{
|
| 1733 |
+
"epoch": 0.79,
|
| 1734 |
+
"learning_rate": 1.4737707613879297e-05,
|
| 1735 |
+
"loss": 0.0459,
|
| 1736 |
+
"step": 144000
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 0.79,
|
| 1740 |
+
"learning_rate": 1.471943576531638e-05,
|
| 1741 |
+
"loss": 0.0581,
|
| 1742 |
+
"step": 144500
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 0.79,
|
| 1746 |
+
"learning_rate": 1.4701163916753459e-05,
|
| 1747 |
+
"loss": 0.0546,
|
| 1748 |
+
"step": 145000
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 0.8,
|
| 1752 |
+
"learning_rate": 1.468289206819054e-05,
|
| 1753 |
+
"loss": 0.0551,
|
| 1754 |
+
"step": 145500
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 0.8,
|
| 1758 |
+
"learning_rate": 1.4664620219627622e-05,
|
| 1759 |
+
"loss": 0.0476,
|
| 1760 |
+
"step": 146000
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 0.8,
|
| 1764 |
+
"learning_rate": 1.4646348371064702e-05,
|
| 1765 |
+
"loss": 0.0533,
|
| 1766 |
+
"step": 146500
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 0.81,
|
| 1770 |
+
"learning_rate": 1.4628076522501781e-05,
|
| 1771 |
+
"loss": 0.0561,
|
| 1772 |
+
"step": 147000
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 0.81,
|
| 1776 |
+
"learning_rate": 1.4609804673938864e-05,
|
| 1777 |
+
"loss": 0.0504,
|
| 1778 |
+
"step": 147500
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 0.81,
|
| 1782 |
+
"learning_rate": 1.4591532825375944e-05,
|
| 1783 |
+
"loss": 0.0398,
|
| 1784 |
+
"step": 148000
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 0.81,
|
| 1788 |
+
"learning_rate": 1.4573260976813027e-05,
|
| 1789 |
+
"loss": 0.054,
|
| 1790 |
+
"step": 148500
|
| 1791 |
+
},
|
| 1792 |
+
{
|
| 1793 |
+
"epoch": 0.82,
|
| 1794 |
+
"learning_rate": 1.4554989128250106e-05,
|
| 1795 |
+
"loss": 0.047,
|
| 1796 |
+
"step": 149000
|
| 1797 |
+
},
|
| 1798 |
+
{
|
| 1799 |
+
"epoch": 0.82,
|
| 1800 |
+
"learning_rate": 1.4536717279687187e-05,
|
| 1801 |
+
"loss": 0.047,
|
| 1802 |
+
"step": 149500
|
| 1803 |
+
},
|
| 1804 |
+
{
|
| 1805 |
+
"epoch": 0.82,
|
| 1806 |
+
"learning_rate": 1.4518445431124269e-05,
|
| 1807 |
+
"loss": 0.0489,
|
| 1808 |
+
"step": 150000
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 0.82,
|
| 1812 |
+
"learning_rate": 1.450017358256135e-05,
|
| 1813 |
+
"loss": 0.0551,
|
| 1814 |
+
"step": 150500
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 0.83,
|
| 1818 |
+
"learning_rate": 1.4481901733998429e-05,
|
| 1819 |
+
"loss": 0.043,
|
| 1820 |
+
"step": 151000
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 0.83,
|
| 1824 |
+
"learning_rate": 1.4463629885435511e-05,
|
| 1825 |
+
"loss": 0.0451,
|
| 1826 |
+
"step": 151500
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 0.83,
|
| 1830 |
+
"learning_rate": 1.4445358036872592e-05,
|
| 1831 |
+
"loss": 0.057,
|
| 1832 |
+
"step": 152000
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 0.84,
|
| 1836 |
+
"learning_rate": 1.4427086188309674e-05,
|
| 1837 |
+
"loss": 0.0554,
|
| 1838 |
+
"step": 152500
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 0.84,
|
| 1842 |
+
"learning_rate": 1.4408814339746753e-05,
|
| 1843 |
+
"loss": 0.0469,
|
| 1844 |
+
"step": 153000
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.84,
|
| 1848 |
+
"learning_rate": 1.4390542491183834e-05,
|
| 1849 |
+
"loss": 0.0504,
|
| 1850 |
+
"step": 153500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.84,
|
| 1854 |
+
"learning_rate": 1.4372270642620916e-05,
|
| 1855 |
+
"loss": 0.0567,
|
| 1856 |
+
"step": 154000
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 0.85,
|
| 1860 |
+
"learning_rate": 1.4353998794057995e-05,
|
| 1861 |
+
"loss": 0.0634,
|
| 1862 |
+
"step": 154500
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 0.85,
|
| 1866 |
+
"learning_rate": 1.4335726945495076e-05,
|
| 1867 |
+
"loss": 0.0471,
|
| 1868 |
+
"step": 155000
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 0.85,
|
| 1872 |
+
"learning_rate": 1.4317455096932158e-05,
|
| 1873 |
+
"loss": 0.0391,
|
| 1874 |
+
"step": 155500
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 0.86,
|
| 1878 |
+
"learning_rate": 1.4299183248369239e-05,
|
| 1879 |
+
"loss": 0.0481,
|
| 1880 |
+
"step": 156000
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 0.86,
|
| 1884 |
+
"learning_rate": 1.4280911399806318e-05,
|
| 1885 |
+
"loss": 0.0554,
|
| 1886 |
+
"step": 156500
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 0.86,
|
| 1890 |
+
"learning_rate": 1.42626395512434e-05,
|
| 1891 |
+
"loss": 0.0501,
|
| 1892 |
+
"step": 157000
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 0.86,
|
| 1896 |
+
"learning_rate": 1.4244367702680481e-05,
|
| 1897 |
+
"loss": 0.0456,
|
| 1898 |
+
"step": 157500
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 0.87,
|
| 1902 |
+
"learning_rate": 1.4226095854117564e-05,
|
| 1903 |
+
"loss": 0.0331,
|
| 1904 |
+
"step": 158000
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 0.87,
|
| 1908 |
+
"learning_rate": 1.4207824005554643e-05,
|
| 1909 |
+
"loss": 0.0514,
|
| 1910 |
+
"step": 158500
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 0.87,
|
| 1914 |
+
"learning_rate": 1.4189552156991723e-05,
|
| 1915 |
+
"loss": 0.0434,
|
| 1916 |
+
"step": 159000
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 0.87,
|
| 1920 |
+
"learning_rate": 1.4171280308428806e-05,
|
| 1921 |
+
"loss": 0.0519,
|
| 1922 |
+
"step": 159500
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 0.88,
|
| 1926 |
+
"learning_rate": 1.4153008459865886e-05,
|
| 1927 |
+
"loss": 0.0434,
|
| 1928 |
+
"step": 160000
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 0.88,
|
| 1932 |
+
"learning_rate": 1.4134736611302965e-05,
|
| 1933 |
+
"loss": 0.0385,
|
| 1934 |
+
"step": 160500
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 0.88,
|
| 1938 |
+
"learning_rate": 1.4116464762740048e-05,
|
| 1939 |
+
"loss": 0.058,
|
| 1940 |
+
"step": 161000
|
| 1941 |
+
},
|
| 1942 |
+
{
|
| 1943 |
+
"epoch": 0.89,
|
| 1944 |
+
"learning_rate": 1.4098192914177128e-05,
|
| 1945 |
+
"loss": 0.0476,
|
| 1946 |
+
"step": 161500
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.89,
|
| 1950 |
+
"learning_rate": 1.4079921065614207e-05,
|
| 1951 |
+
"loss": 0.0462,
|
| 1952 |
+
"step": 162000
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 0.89,
|
| 1956 |
+
"learning_rate": 1.406164921705129e-05,
|
| 1957 |
+
"loss": 0.0447,
|
| 1958 |
+
"step": 162500
|
| 1959 |
+
},
|
| 1960 |
+
{
|
| 1961 |
+
"epoch": 0.89,
|
| 1962 |
+
"learning_rate": 1.404337736848837e-05,
|
| 1963 |
+
"loss": 0.048,
|
| 1964 |
+
"step": 163000
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 0.9,
|
| 1968 |
+
"learning_rate": 1.4025105519925453e-05,
|
| 1969 |
+
"loss": 0.0442,
|
| 1970 |
+
"step": 163500
|
| 1971 |
+
},
|
| 1972 |
+
{
|
| 1973 |
+
"epoch": 0.9,
|
| 1974 |
+
"learning_rate": 1.4006833671362532e-05,
|
| 1975 |
+
"loss": 0.0555,
|
| 1976 |
+
"step": 164000
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 0.9,
|
| 1980 |
+
"learning_rate": 1.3988561822799613e-05,
|
| 1981 |
+
"loss": 0.0434,
|
| 1982 |
+
"step": 164500
|
| 1983 |
+
},
|
| 1984 |
+
{
|
| 1985 |
+
"epoch": 0.9,
|
| 1986 |
+
"learning_rate": 1.3970289974236695e-05,
|
| 1987 |
+
"loss": 0.0577,
|
| 1988 |
+
"step": 165000
|
| 1989 |
+
},
|
| 1990 |
+
{
|
| 1991 |
+
"epoch": 0.91,
|
| 1992 |
+
"learning_rate": 1.3952018125673776e-05,
|
| 1993 |
+
"loss": 0.0522,
|
| 1994 |
+
"step": 165500
|
| 1995 |
+
},
|
| 1996 |
+
{
|
| 1997 |
+
"epoch": 0.91,
|
| 1998 |
+
"learning_rate": 1.3933746277110855e-05,
|
| 1999 |
+
"loss": 0.0412,
|
| 2000 |
+
"step": 166000
|
| 2001 |
+
},
|
| 2002 |
+
{
|
| 2003 |
+
"epoch": 0.91,
|
| 2004 |
+
"learning_rate": 1.3915474428547937e-05,
|
| 2005 |
+
"loss": 0.0601,
|
| 2006 |
+
"step": 166500
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 0.92,
|
| 2010 |
+
"learning_rate": 1.3897202579985018e-05,
|
| 2011 |
+
"loss": 0.0553,
|
| 2012 |
+
"step": 167000
|
| 2013 |
+
},
|
| 2014 |
+
{
|
| 2015 |
+
"epoch": 0.92,
|
| 2016 |
+
"learning_rate": 1.38789307314221e-05,
|
| 2017 |
+
"loss": 0.0522,
|
| 2018 |
+
"step": 167500
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 0.92,
|
| 2022 |
+
"learning_rate": 1.386065888285918e-05,
|
| 2023 |
+
"loss": 0.0505,
|
| 2024 |
+
"step": 168000
|
| 2025 |
+
},
|
| 2026 |
+
{
|
| 2027 |
+
"epoch": 0.92,
|
| 2028 |
+
"learning_rate": 1.384238703429626e-05,
|
| 2029 |
+
"loss": 0.0497,
|
| 2030 |
+
"step": 168500
|
| 2031 |
+
},
|
| 2032 |
+
{
|
| 2033 |
+
"epoch": 0.93,
|
| 2034 |
+
"learning_rate": 1.3824115185733342e-05,
|
| 2035 |
+
"loss": 0.0378,
|
| 2036 |
+
"step": 169000
|
| 2037 |
+
},
|
| 2038 |
+
{
|
| 2039 |
+
"epoch": 0.93,
|
| 2040 |
+
"learning_rate": 1.3805843337170423e-05,
|
| 2041 |
+
"loss": 0.0399,
|
| 2042 |
+
"step": 169500
|
| 2043 |
+
},
|
| 2044 |
+
{
|
| 2045 |
+
"epoch": 0.93,
|
| 2046 |
+
"learning_rate": 1.3787571488607504e-05,
|
| 2047 |
+
"loss": 0.0485,
|
| 2048 |
+
"step": 170000
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 0.93,
|
| 2052 |
+
"learning_rate": 1.3769299640044584e-05,
|
| 2053 |
+
"loss": 0.0416,
|
| 2054 |
+
"step": 170500
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.94,
|
| 2058 |
+
"learning_rate": 1.3751027791481665e-05,
|
| 2059 |
+
"loss": 0.0486,
|
| 2060 |
+
"step": 171000
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 0.94,
|
| 2064 |
+
"learning_rate": 1.3732755942918746e-05,
|
| 2065 |
+
"loss": 0.0498,
|
| 2066 |
+
"step": 171500
|
| 2067 |
+
},
|
| 2068 |
+
{
|
| 2069 |
+
"epoch": 0.94,
|
| 2070 |
+
"learning_rate": 1.3714484094355827e-05,
|
| 2071 |
+
"loss": 0.0515,
|
| 2072 |
+
"step": 172000
|
| 2073 |
+
},
|
| 2074 |
+
{
|
| 2075 |
+
"epoch": 0.95,
|
| 2076 |
+
"learning_rate": 1.3696212245792909e-05,
|
| 2077 |
+
"loss": 0.0403,
|
| 2078 |
+
"step": 172500
|
| 2079 |
+
},
|
| 2080 |
+
{
|
| 2081 |
+
"epoch": 0.95,
|
| 2082 |
+
"learning_rate": 1.367794039722999e-05,
|
| 2083 |
+
"loss": 0.0461,
|
| 2084 |
+
"step": 173000
|
| 2085 |
+
},
|
| 2086 |
+
{
|
| 2087 |
+
"epoch": 0.95,
|
| 2088 |
+
"learning_rate": 1.3659668548667069e-05,
|
| 2089 |
+
"loss": 0.0557,
|
| 2090 |
+
"step": 173500
|
| 2091 |
+
},
|
| 2092 |
+
{
|
| 2093 |
+
"epoch": 0.95,
|
| 2094 |
+
"learning_rate": 1.3641396700104151e-05,
|
| 2095 |
+
"loss": 0.0602,
|
| 2096 |
+
"step": 174000
|
| 2097 |
+
},
|
| 2098 |
+
{
|
| 2099 |
+
"epoch": 0.96,
|
| 2100 |
+
"learning_rate": 1.3623124851541232e-05,
|
| 2101 |
+
"loss": 0.0416,
|
| 2102 |
+
"step": 174500
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 0.96,
|
| 2106 |
+
"learning_rate": 1.3604853002978312e-05,
|
| 2107 |
+
"loss": 0.0472,
|
| 2108 |
+
"step": 175000
|
| 2109 |
+
},
|
| 2110 |
+
{
|
| 2111 |
+
"epoch": 0.96,
|
| 2112 |
+
"learning_rate": 1.3586581154415393e-05,
|
| 2113 |
+
"loss": 0.0475,
|
| 2114 |
+
"step": 175500
|
| 2115 |
+
},
|
| 2116 |
+
{
|
| 2117 |
+
"epoch": 0.96,
|
| 2118 |
+
"learning_rate": 1.3568309305852474e-05,
|
| 2119 |
+
"loss": 0.0478,
|
| 2120 |
+
"step": 176000
|
| 2121 |
+
},
|
| 2122 |
+
{
|
| 2123 |
+
"epoch": 0.97,
|
| 2124 |
+
"learning_rate": 1.3550037457289556e-05,
|
| 2125 |
+
"loss": 0.0448,
|
| 2126 |
+
"step": 176500
|
| 2127 |
+
},
|
| 2128 |
+
{
|
| 2129 |
+
"epoch": 0.97,
|
| 2130 |
+
"learning_rate": 1.3531765608726637e-05,
|
| 2131 |
+
"loss": 0.0444,
|
| 2132 |
+
"step": 177000
|
| 2133 |
+
},
|
| 2134 |
+
{
|
| 2135 |
+
"epoch": 0.97,
|
| 2136 |
+
"learning_rate": 1.3513493760163716e-05,
|
| 2137 |
+
"loss": 0.0459,
|
| 2138 |
+
"step": 177500
|
| 2139 |
+
},
|
| 2140 |
+
{
|
| 2141 |
+
"epoch": 0.98,
|
| 2142 |
+
"learning_rate": 1.3495221911600798e-05,
|
| 2143 |
+
"loss": 0.0381,
|
| 2144 |
+
"step": 178000
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 0.98,
|
| 2148 |
+
"learning_rate": 1.3476950063037879e-05,
|
| 2149 |
+
"loss": 0.0402,
|
| 2150 |
+
"step": 178500
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 0.98,
|
| 2154 |
+
"learning_rate": 1.3458678214474961e-05,
|
| 2155 |
+
"loss": 0.0338,
|
| 2156 |
+
"step": 179000
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 0.98,
|
| 2160 |
+
"learning_rate": 1.344040636591204e-05,
|
| 2161 |
+
"loss": 0.0527,
|
| 2162 |
+
"step": 179500
|
| 2163 |
+
},
|
| 2164 |
+
{
|
| 2165 |
+
"epoch": 0.99,
|
| 2166 |
+
"learning_rate": 1.3422134517349121e-05,
|
| 2167 |
+
"loss": 0.0381,
|
| 2168 |
+
"step": 180000
|
| 2169 |
+
},
|
| 2170 |
+
{
|
| 2171 |
+
"epoch": 0.99,
|
| 2172 |
+
"learning_rate": 1.3403862668786203e-05,
|
| 2173 |
+
"loss": 0.0473,
|
| 2174 |
+
"step": 180500
|
| 2175 |
+
},
|
| 2176 |
+
{
|
| 2177 |
+
"epoch": 0.99,
|
| 2178 |
+
"learning_rate": 1.3385590820223282e-05,
|
| 2179 |
+
"loss": 0.0475,
|
| 2180 |
+
"step": 181000
|
| 2181 |
+
},
|
| 2182 |
+
{
|
| 2183 |
+
"epoch": 0.99,
|
| 2184 |
+
"learning_rate": 1.3367318971660363e-05,
|
| 2185 |
+
"loss": 0.0482,
|
| 2186 |
+
"step": 181500
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 1.0,
|
| 2190 |
+
"learning_rate": 1.3349047123097446e-05,
|
| 2191 |
+
"loss": 0.0432,
|
| 2192 |
+
"step": 182000
|
| 2193 |
+
},
|
| 2194 |
+
{
|
| 2195 |
+
"epoch": 1.0,
|
| 2196 |
+
"eval_loss": 0.15653175115585327,
|
| 2197 |
+
"eval_runtime": 1676.2688,
|
| 2198 |
+
"eval_samples_per_second": 138.216,
|
| 2199 |
+
"eval_steps_per_second": 13.822,
|
| 2200 |
+
"step": 182430
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 1.0,
|
| 2204 |
+
"learning_rate": 1.3330775274534526e-05,
|
| 2205 |
+
"loss": 0.0411,
|
| 2206 |
+
"step": 182500
|
| 2207 |
+
},
|
| 2208 |
+
{
|
| 2209 |
+
"epoch": 1.0,
|
| 2210 |
+
"learning_rate": 1.3312503425971605e-05,
|
| 2211 |
+
"loss": 0.0428,
|
| 2212 |
+
"step": 183000
|
| 2213 |
+
},
|
| 2214 |
+
{
|
| 2215 |
+
"epoch": 1.01,
|
| 2216 |
+
"learning_rate": 1.3294231577408688e-05,
|
| 2217 |
+
"loss": 0.0454,
|
| 2218 |
+
"step": 183500
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 1.01,
|
| 2222 |
+
"learning_rate": 1.3275959728845768e-05,
|
| 2223 |
+
"loss": 0.0453,
|
| 2224 |
+
"step": 184000
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 1.01,
|
| 2228 |
+
"learning_rate": 1.325768788028285e-05,
|
| 2229 |
+
"loss": 0.0451,
|
| 2230 |
+
"step": 184500
|
| 2231 |
+
},
|
| 2232 |
+
{
|
| 2233 |
+
"epoch": 1.01,
|
| 2234 |
+
"learning_rate": 1.323941603171993e-05,
|
| 2235 |
+
"loss": 0.0426,
|
| 2236 |
+
"step": 185000
|
| 2237 |
+
},
|
| 2238 |
+
{
|
| 2239 |
+
"epoch": 1.02,
|
| 2240 |
+
"learning_rate": 1.322114418315701e-05,
|
| 2241 |
+
"loss": 0.0386,
|
| 2242 |
+
"step": 185500
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 1.02,
|
| 2246 |
+
"learning_rate": 1.3202872334594093e-05,
|
| 2247 |
+
"loss": 0.0422,
|
| 2248 |
+
"step": 186000
|
| 2249 |
+
},
|
| 2250 |
+
{
|
| 2251 |
+
"epoch": 1.02,
|
| 2252 |
+
"learning_rate": 1.3184600486031174e-05,
|
| 2253 |
+
"loss": 0.0491,
|
| 2254 |
+
"step": 186500
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 1.03,
|
| 2258 |
+
"learning_rate": 1.3166328637468253e-05,
|
| 2259 |
+
"loss": 0.0493,
|
| 2260 |
+
"step": 187000
|
| 2261 |
+
},
|
| 2262 |
+
{
|
| 2263 |
+
"epoch": 1.03,
|
| 2264 |
+
"learning_rate": 1.3148056788905335e-05,
|
| 2265 |
+
"loss": 0.0388,
|
| 2266 |
+
"step": 187500
|
| 2267 |
+
},
|
| 2268 |
+
{
|
| 2269 |
+
"epoch": 1.03,
|
| 2270 |
+
"learning_rate": 1.3129784940342416e-05,
|
| 2271 |
+
"loss": 0.0429,
|
| 2272 |
+
"step": 188000
|
| 2273 |
+
},
|
| 2274 |
+
{
|
| 2275 |
+
"epoch": 1.03,
|
| 2276 |
+
"learning_rate": 1.3111513091779498e-05,
|
| 2277 |
+
"loss": 0.0501,
|
| 2278 |
+
"step": 188500
|
| 2279 |
+
},
|
| 2280 |
+
{
|
| 2281 |
+
"epoch": 1.04,
|
| 2282 |
+
"learning_rate": 1.3093241243216577e-05,
|
| 2283 |
+
"loss": 0.0423,
|
| 2284 |
+
"step": 189000
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 1.04,
|
| 2288 |
+
"learning_rate": 1.3074969394653658e-05,
|
| 2289 |
+
"loss": 0.0458,
|
| 2290 |
+
"step": 189500
|
| 2291 |
+
},
|
| 2292 |
+
{
|
| 2293 |
+
"epoch": 1.04,
|
| 2294 |
+
"learning_rate": 1.305669754609074e-05,
|
| 2295 |
+
"loss": 0.0411,
|
| 2296 |
+
"step": 190000
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 1.04,
|
| 2300 |
+
"learning_rate": 1.303842569752782e-05,
|
| 2301 |
+
"loss": 0.0411,
|
| 2302 |
+
"step": 190500
|
| 2303 |
+
},
|
| 2304 |
+
{
|
| 2305 |
+
"epoch": 1.05,
|
| 2306 |
+
"learning_rate": 1.30201538489649e-05,
|
| 2307 |
+
"loss": 0.037,
|
| 2308 |
+
"step": 191000
|
| 2309 |
+
},
|
| 2310 |
+
{
|
| 2311 |
+
"epoch": 1.05,
|
| 2312 |
+
"learning_rate": 1.3001882000401982e-05,
|
| 2313 |
+
"loss": 0.039,
|
| 2314 |
+
"step": 191500
|
| 2315 |
+
},
|
| 2316 |
+
{
|
| 2317 |
+
"epoch": 1.05,
|
| 2318 |
+
"learning_rate": 1.2983610151839063e-05,
|
| 2319 |
+
"loss": 0.0468,
|
| 2320 |
+
"step": 192000
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 1.06,
|
| 2324 |
+
"learning_rate": 1.2965338303276142e-05,
|
| 2325 |
+
"loss": 0.0404,
|
| 2326 |
+
"step": 192500
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 1.06,
|
| 2330 |
+
"learning_rate": 1.2947066454713224e-05,
|
| 2331 |
+
"loss": 0.0394,
|
| 2332 |
+
"step": 193000
|
| 2333 |
+
},
|
| 2334 |
+
{
|
| 2335 |
+
"epoch": 1.06,
|
| 2336 |
+
"learning_rate": 1.2928794606150305e-05,
|
| 2337 |
+
"loss": 0.0454,
|
| 2338 |
+
"step": 193500
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 1.06,
|
| 2342 |
+
"learning_rate": 1.2910522757587387e-05,
|
| 2343 |
+
"loss": 0.0402,
|
| 2344 |
+
"step": 194000
|
| 2345 |
+
},
|
| 2346 |
+
{
|
| 2347 |
+
"epoch": 1.07,
|
| 2348 |
+
"learning_rate": 1.2892250909024466e-05,
|
| 2349 |
+
"loss": 0.0507,
|
| 2350 |
+
"step": 194500
|
| 2351 |
+
},
|
| 2352 |
+
{
|
| 2353 |
+
"epoch": 1.07,
|
| 2354 |
+
"learning_rate": 1.2873979060461547e-05,
|
| 2355 |
+
"loss": 0.0429,
|
| 2356 |
+
"step": 195000
|
| 2357 |
+
},
|
| 2358 |
+
{
|
| 2359 |
+
"epoch": 1.07,
|
| 2360 |
+
"learning_rate": 1.285570721189863e-05,
|
| 2361 |
+
"loss": 0.0452,
|
| 2362 |
+
"step": 195500
|
| 2363 |
+
},
|
| 2364 |
+
{
|
| 2365 |
+
"epoch": 1.07,
|
| 2366 |
+
"learning_rate": 1.283743536333571e-05,
|
| 2367 |
+
"loss": 0.042,
|
| 2368 |
+
"step": 196000
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 1.08,
|
| 2372 |
+
"learning_rate": 1.281916351477279e-05,
|
| 2373 |
+
"loss": 0.0482,
|
| 2374 |
+
"step": 196500
|
| 2375 |
+
},
|
| 2376 |
+
{
|
| 2377 |
+
"epoch": 1.08,
|
| 2378 |
+
"learning_rate": 1.2800891666209872e-05,
|
| 2379 |
+
"loss": 0.0457,
|
| 2380 |
+
"step": 197000
|
| 2381 |
+
},
|
| 2382 |
+
{
|
| 2383 |
+
"epoch": 1.08,
|
| 2384 |
+
"learning_rate": 1.2782619817646952e-05,
|
| 2385 |
+
"loss": 0.0383,
|
| 2386 |
+
"step": 197500
|
| 2387 |
+
},
|
| 2388 |
+
{
|
| 2389 |
+
"epoch": 1.09,
|
| 2390 |
+
"learning_rate": 1.2764347969084031e-05,
|
| 2391 |
+
"loss": 0.0384,
|
| 2392 |
+
"step": 198000
|
| 2393 |
+
},
|
| 2394 |
+
{
|
| 2395 |
+
"epoch": 1.09,
|
| 2396 |
+
"learning_rate": 1.2746076120521114e-05,
|
| 2397 |
+
"loss": 0.0558,
|
| 2398 |
+
"step": 198500
|
| 2399 |
+
},
|
| 2400 |
+
{
|
| 2401 |
+
"epoch": 1.09,
|
| 2402 |
+
"learning_rate": 1.2727804271958194e-05,
|
| 2403 |
+
"loss": 0.0362,
|
| 2404 |
+
"step": 199000
|
| 2405 |
+
},
|
| 2406 |
+
{
|
| 2407 |
+
"epoch": 1.09,
|
| 2408 |
+
"learning_rate": 1.2709532423395277e-05,
|
| 2409 |
+
"loss": 0.044,
|
| 2410 |
+
"step": 199500
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 1.1,
|
| 2414 |
+
"learning_rate": 1.2691260574832356e-05,
|
| 2415 |
+
"loss": 0.0475,
|
| 2416 |
+
"step": 200000
|
| 2417 |
+
},
|
| 2418 |
+
{
|
| 2419 |
+
"epoch": 1.1,
|
| 2420 |
+
"learning_rate": 1.2672988726269437e-05,
|
| 2421 |
+
"loss": 0.0404,
|
| 2422 |
+
"step": 200500
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 1.1,
|
| 2426 |
+
"learning_rate": 1.2654716877706519e-05,
|
| 2427 |
+
"loss": 0.0384,
|
| 2428 |
+
"step": 201000
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 1.1,
|
| 2432 |
+
"learning_rate": 1.26364450291436e-05,
|
| 2433 |
+
"loss": 0.0528,
|
| 2434 |
+
"step": 201500
|
| 2435 |
+
},
|
| 2436 |
+
{
|
| 2437 |
+
"epoch": 1.11,
|
| 2438 |
+
"learning_rate": 1.261817318058068e-05,
|
| 2439 |
+
"loss": 0.0572,
|
| 2440 |
+
"step": 202000
|
| 2441 |
+
},
|
| 2442 |
+
{
|
| 2443 |
+
"epoch": 1.11,
|
| 2444 |
+
"learning_rate": 1.2599901332017761e-05,
|
| 2445 |
+
"loss": 0.0365,
|
| 2446 |
+
"step": 202500
|
| 2447 |
+
},
|
| 2448 |
+
{
|
| 2449 |
+
"epoch": 1.11,
|
| 2450 |
+
"learning_rate": 1.2581629483454842e-05,
|
| 2451 |
+
"loss": 0.0354,
|
| 2452 |
+
"step": 203000
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 1.12,
|
| 2456 |
+
"learning_rate": 1.2563357634891924e-05,
|
| 2457 |
+
"loss": 0.0391,
|
| 2458 |
+
"step": 203500
|
| 2459 |
+
},
|
| 2460 |
+
{
|
| 2461 |
+
"epoch": 1.12,
|
| 2462 |
+
"learning_rate": 1.2545085786329003e-05,
|
| 2463 |
+
"loss": 0.0394,
|
| 2464 |
+
"step": 204000
|
| 2465 |
+
},
|
| 2466 |
+
{
|
| 2467 |
+
"epoch": 1.12,
|
| 2468 |
+
"learning_rate": 1.2526813937766086e-05,
|
| 2469 |
+
"loss": 0.044,
|
| 2470 |
+
"step": 204500
|
| 2471 |
+
},
|
| 2472 |
+
{
|
| 2473 |
+
"epoch": 1.12,
|
| 2474 |
+
"learning_rate": 1.2508542089203166e-05,
|
| 2475 |
+
"loss": 0.0436,
|
| 2476 |
+
"step": 205000
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 1.13,
|
| 2480 |
+
"learning_rate": 1.2490270240640247e-05,
|
| 2481 |
+
"loss": 0.0294,
|
| 2482 |
+
"step": 205500
|
| 2483 |
+
},
|
| 2484 |
+
{
|
| 2485 |
+
"epoch": 1.13,
|
| 2486 |
+
"learning_rate": 1.2471998392077328e-05,
|
| 2487 |
+
"loss": 0.0422,
|
| 2488 |
+
"step": 206000
|
| 2489 |
+
},
|
| 2490 |
+
{
|
| 2491 |
+
"epoch": 1.13,
|
| 2492 |
+
"learning_rate": 1.2453726543514408e-05,
|
| 2493 |
+
"loss": 0.0428,
|
| 2494 |
+
"step": 206500
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 1.13,
|
| 2498 |
+
"learning_rate": 1.2435454694951489e-05,
|
| 2499 |
+
"loss": 0.0414,
|
| 2500 |
+
"step": 207000
|
| 2501 |
+
},
|
| 2502 |
+
{
|
| 2503 |
+
"epoch": 1.14,
|
| 2504 |
+
"learning_rate": 1.241718284638857e-05,
|
| 2505 |
+
"loss": 0.0411,
|
| 2506 |
+
"step": 207500
|
| 2507 |
+
},
|
| 2508 |
+
{
|
| 2509 |
+
"epoch": 1.14,
|
| 2510 |
+
"learning_rate": 1.239891099782565e-05,
|
| 2511 |
+
"loss": 0.0261,
|
| 2512 |
+
"step": 208000
|
| 2513 |
+
},
|
| 2514 |
+
{
|
| 2515 |
+
"epoch": 1.14,
|
| 2516 |
+
"learning_rate": 1.2380639149262733e-05,
|
| 2517 |
+
"loss": 0.0327,
|
| 2518 |
+
"step": 208500
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 1.15,
|
| 2522 |
+
"learning_rate": 1.2362367300699814e-05,
|
| 2523 |
+
"loss": 0.0373,
|
| 2524 |
+
"step": 209000
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 1.15,
|
| 2528 |
+
"learning_rate": 1.2344095452136893e-05,
|
| 2529 |
+
"loss": 0.0467,
|
| 2530 |
+
"step": 209500
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 1.15,
|
| 2534 |
+
"learning_rate": 1.2325823603573975e-05,
|
| 2535 |
+
"loss": 0.0375,
|
| 2536 |
+
"step": 210000
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 1.15,
|
| 2540 |
+
"learning_rate": 1.2307551755011056e-05,
|
| 2541 |
+
"loss": 0.042,
|
| 2542 |
+
"step": 210500
|
| 2543 |
+
},
|
| 2544 |
+
{
|
| 2545 |
+
"epoch": 1.16,
|
| 2546 |
+
"learning_rate": 1.2289279906448138e-05,
|
| 2547 |
+
"loss": 0.0365,
|
| 2548 |
+
"step": 211000
|
| 2549 |
+
},
|
| 2550 |
+
{
|
| 2551 |
+
"epoch": 1.16,
|
| 2552 |
+
"learning_rate": 1.2271008057885217e-05,
|
| 2553 |
+
"loss": 0.0464,
|
| 2554 |
+
"step": 211500
|
| 2555 |
+
},
|
| 2556 |
+
{
|
| 2557 |
+
"epoch": 1.16,
|
| 2558 |
+
"learning_rate": 1.2252736209322298e-05,
|
| 2559 |
+
"loss": 0.0398,
|
| 2560 |
+
"step": 212000
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 1.16,
|
| 2564 |
+
"learning_rate": 1.223446436075938e-05,
|
| 2565 |
+
"loss": 0.0391,
|
| 2566 |
+
"step": 212500
|
| 2567 |
+
},
|
| 2568 |
+
{
|
| 2569 |
+
"epoch": 1.17,
|
| 2570 |
+
"learning_rate": 1.221619251219646e-05,
|
| 2571 |
+
"loss": 0.0413,
|
| 2572 |
+
"step": 213000
|
| 2573 |
+
},
|
| 2574 |
+
{
|
| 2575 |
+
"epoch": 1.17,
|
| 2576 |
+
"learning_rate": 1.219792066363354e-05,
|
| 2577 |
+
"loss": 0.0358,
|
| 2578 |
+
"step": 213500
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 1.17,
|
| 2582 |
+
"learning_rate": 1.2179648815070622e-05,
|
| 2583 |
+
"loss": 0.043,
|
| 2584 |
+
"step": 214000
|
| 2585 |
+
},
|
| 2586 |
+
{
|
| 2587 |
+
"epoch": 1.18,
|
| 2588 |
+
"learning_rate": 1.2161376966507703e-05,
|
| 2589 |
+
"loss": 0.0413,
|
| 2590 |
+
"step": 214500
|
| 2591 |
+
},
|
| 2592 |
+
{
|
| 2593 |
+
"epoch": 1.18,
|
| 2594 |
+
"learning_rate": 1.2143105117944785e-05,
|
| 2595 |
+
"loss": 0.0513,
|
| 2596 |
+
"step": 215000
|
| 2597 |
+
},
|
| 2598 |
+
{
|
| 2599 |
+
"epoch": 1.18,
|
| 2600 |
+
"learning_rate": 1.2124833269381864e-05,
|
| 2601 |
+
"loss": 0.0421,
|
| 2602 |
+
"step": 215500
|
| 2603 |
+
},
|
| 2604 |
+
{
|
| 2605 |
+
"epoch": 1.18,
|
| 2606 |
+
"learning_rate": 1.2106561420818945e-05,
|
| 2607 |
+
"loss": 0.0474,
|
| 2608 |
+
"step": 216000
|
| 2609 |
+
},
|
| 2610 |
+
{
|
| 2611 |
+
"epoch": 1.19,
|
| 2612 |
+
"learning_rate": 1.2088289572256027e-05,
|
| 2613 |
+
"loss": 0.0445,
|
| 2614 |
+
"step": 216500
|
| 2615 |
+
},
|
| 2616 |
+
{
|
| 2617 |
+
"epoch": 1.19,
|
| 2618 |
+
"learning_rate": 1.2070017723693106e-05,
|
| 2619 |
+
"loss": 0.0396,
|
| 2620 |
+
"step": 217000
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 1.19,
|
| 2624 |
+
"learning_rate": 1.2051745875130187e-05,
|
| 2625 |
+
"loss": 0.038,
|
| 2626 |
+
"step": 217500
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 1.19,
|
| 2630 |
+
"learning_rate": 1.203347402656727e-05,
|
| 2631 |
+
"loss": 0.0269,
|
| 2632 |
+
"step": 218000
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"epoch": 1.2,
|
| 2636 |
+
"learning_rate": 1.201520217800435e-05,
|
| 2637 |
+
"loss": 0.0422,
|
| 2638 |
+
"step": 218500
|
| 2639 |
+
},
|
| 2640 |
+
{
|
| 2641 |
+
"epoch": 1.2,
|
| 2642 |
+
"learning_rate": 1.199693032944143e-05,
|
| 2643 |
+
"loss": 0.0522,
|
| 2644 |
+
"step": 219000
|
| 2645 |
+
},
|
| 2646 |
+
{
|
| 2647 |
+
"epoch": 1.2,
|
| 2648 |
+
"learning_rate": 1.1978658480878512e-05,
|
| 2649 |
+
"loss": 0.035,
|
| 2650 |
+
"step": 219500
|
| 2651 |
+
},
|
| 2652 |
+
{
|
| 2653 |
+
"epoch": 1.21,
|
| 2654 |
+
"learning_rate": 1.1960386632315592e-05,
|
| 2655 |
+
"loss": 0.0388,
|
| 2656 |
+
"step": 220000
|
| 2657 |
+
},
|
| 2658 |
+
{
|
| 2659 |
+
"epoch": 1.21,
|
| 2660 |
+
"learning_rate": 1.1942114783752675e-05,
|
| 2661 |
+
"loss": 0.033,
|
| 2662 |
+
"step": 220500
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 1.21,
|
| 2666 |
+
"learning_rate": 1.1923842935189754e-05,
|
| 2667 |
+
"loss": 0.0459,
|
| 2668 |
+
"step": 221000
|
| 2669 |
+
},
|
| 2670 |
+
{
|
| 2671 |
+
"epoch": 1.21,
|
| 2672 |
+
"learning_rate": 1.1905571086626834e-05,
|
| 2673 |
+
"loss": 0.0368,
|
| 2674 |
+
"step": 221500
|
| 2675 |
+
},
|
| 2676 |
+
{
|
| 2677 |
+
"epoch": 1.22,
|
| 2678 |
+
"learning_rate": 1.1887299238063917e-05,
|
| 2679 |
+
"loss": 0.0415,
|
| 2680 |
+
"step": 222000
|
| 2681 |
+
},
|
| 2682 |
+
{
|
| 2683 |
+
"epoch": 1.22,
|
| 2684 |
+
"learning_rate": 1.1869027389500997e-05,
|
| 2685 |
+
"loss": 0.0373,
|
| 2686 |
+
"step": 222500
|
| 2687 |
+
},
|
| 2688 |
+
{
|
| 2689 |
+
"epoch": 1.22,
|
| 2690 |
+
"learning_rate": 1.1850755540938076e-05,
|
| 2691 |
+
"loss": 0.0308,
|
| 2692 |
+
"step": 223000
|
| 2693 |
+
},
|
| 2694 |
+
{
|
| 2695 |
+
"epoch": 1.23,
|
| 2696 |
+
"learning_rate": 1.1832483692375159e-05,
|
| 2697 |
+
"loss": 0.0378,
|
| 2698 |
+
"step": 223500
|
| 2699 |
+
},
|
| 2700 |
+
{
|
| 2701 |
+
"epoch": 1.23,
|
| 2702 |
+
"learning_rate": 1.181421184381224e-05,
|
| 2703 |
+
"loss": 0.0424,
|
| 2704 |
+
"step": 224000
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 1.23,
|
| 2708 |
+
"learning_rate": 1.1795939995249322e-05,
|
| 2709 |
+
"loss": 0.0516,
|
| 2710 |
+
"step": 224500
|
| 2711 |
+
},
|
| 2712 |
+
{
|
| 2713 |
+
"epoch": 1.23,
|
| 2714 |
+
"learning_rate": 1.1777668146686401e-05,
|
| 2715 |
+
"loss": 0.0349,
|
| 2716 |
+
"step": 225000
|
| 2717 |
+
},
|
| 2718 |
+
{
|
| 2719 |
+
"epoch": 1.24,
|
| 2720 |
+
"learning_rate": 1.1759396298123482e-05,
|
| 2721 |
+
"loss": 0.0281,
|
| 2722 |
+
"step": 225500
|
| 2723 |
+
},
|
| 2724 |
+
{
|
| 2725 |
+
"epoch": 1.24,
|
| 2726 |
+
"learning_rate": 1.1741124449560564e-05,
|
| 2727 |
+
"loss": 0.0424,
|
| 2728 |
+
"step": 226000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 1.24,
|
| 2732 |
+
"learning_rate": 1.1722852600997643e-05,
|
| 2733 |
+
"loss": 0.033,
|
| 2734 |
+
"step": 226500
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"epoch": 1.24,
|
| 2738 |
+
"learning_rate": 1.1704580752434724e-05,
|
| 2739 |
+
"loss": 0.0482,
|
| 2740 |
+
"step": 227000
|
| 2741 |
+
},
|
| 2742 |
+
{
|
| 2743 |
+
"epoch": 1.25,
|
| 2744 |
+
"learning_rate": 1.1686308903871806e-05,
|
| 2745 |
+
"loss": 0.0361,
|
| 2746 |
+
"step": 227500
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 1.25,
|
| 2750 |
+
"learning_rate": 1.1668037055308887e-05,
|
| 2751 |
+
"loss": 0.0392,
|
| 2752 |
+
"step": 228000
|
| 2753 |
+
},
|
| 2754 |
+
{
|
| 2755 |
+
"epoch": 1.25,
|
| 2756 |
+
"learning_rate": 1.1649765206745966e-05,
|
| 2757 |
+
"loss": 0.0357,
|
| 2758 |
+
"step": 228500
|
| 2759 |
+
},
|
| 2760 |
+
{
|
| 2761 |
+
"epoch": 1.26,
|
| 2762 |
+
"learning_rate": 1.1631493358183048e-05,
|
| 2763 |
+
"loss": 0.0428,
|
| 2764 |
+
"step": 229000
|
| 2765 |
+
},
|
| 2766 |
+
{
|
| 2767 |
+
"epoch": 1.26,
|
| 2768 |
+
"learning_rate": 1.1613221509620129e-05,
|
| 2769 |
+
"loss": 0.0337,
|
| 2770 |
+
"step": 229500
|
| 2771 |
+
},
|
| 2772 |
+
{
|
| 2773 |
+
"epoch": 1.26,
|
| 2774 |
+
"learning_rate": 1.1594949661057211e-05,
|
| 2775 |
+
"loss": 0.0395,
|
| 2776 |
+
"step": 230000
|
| 2777 |
+
},
|
| 2778 |
+
{
|
| 2779 |
+
"epoch": 1.26,
|
| 2780 |
+
"learning_rate": 1.157667781249429e-05,
|
| 2781 |
+
"loss": 0.0376,
|
| 2782 |
+
"step": 230500
|
| 2783 |
+
},
|
| 2784 |
+
{
|
| 2785 |
+
"epoch": 1.27,
|
| 2786 |
+
"learning_rate": 1.1558405963931371e-05,
|
| 2787 |
+
"loss": 0.0342,
|
| 2788 |
+
"step": 231000
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 1.27,
|
| 2792 |
+
"learning_rate": 1.1540134115368453e-05,
|
| 2793 |
+
"loss": 0.0442,
|
| 2794 |
+
"step": 231500
|
| 2795 |
+
},
|
| 2796 |
+
{
|
| 2797 |
+
"epoch": 1.27,
|
| 2798 |
+
"learning_rate": 1.1521862266805534e-05,
|
| 2799 |
+
"loss": 0.0445,
|
| 2800 |
+
"step": 232000
|
| 2801 |
+
},
|
| 2802 |
+
{
|
| 2803 |
+
"epoch": 1.27,
|
| 2804 |
+
"learning_rate": 1.1503590418242613e-05,
|
| 2805 |
+
"loss": 0.0396,
|
| 2806 |
+
"step": 232500
|
| 2807 |
+
},
|
| 2808 |
+
{
|
| 2809 |
+
"epoch": 1.28,
|
| 2810 |
+
"learning_rate": 1.1485318569679696e-05,
|
| 2811 |
+
"loss": 0.039,
|
| 2812 |
+
"step": 233000
|
| 2813 |
+
},
|
| 2814 |
+
{
|
| 2815 |
+
"epoch": 1.28,
|
| 2816 |
+
"learning_rate": 1.1467046721116776e-05,
|
| 2817 |
+
"loss": 0.0394,
|
| 2818 |
+
"step": 233500
|
| 2819 |
+
},
|
| 2820 |
+
{
|
| 2821 |
+
"epoch": 1.28,
|
| 2822 |
+
"learning_rate": 1.1448774872553857e-05,
|
| 2823 |
+
"loss": 0.0299,
|
| 2824 |
+
"step": 234000
|
| 2825 |
+
},
|
| 2826 |
+
{
|
| 2827 |
+
"epoch": 1.29,
|
| 2828 |
+
"learning_rate": 1.1430503023990938e-05,
|
| 2829 |
+
"loss": 0.0251,
|
| 2830 |
+
"step": 234500
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 1.29,
|
| 2834 |
+
"learning_rate": 1.1412231175428018e-05,
|
| 2835 |
+
"loss": 0.0359,
|
| 2836 |
+
"step": 235000
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"epoch": 1.29,
|
| 2840 |
+
"learning_rate": 1.13939593268651e-05,
|
| 2841 |
+
"loss": 0.0364,
|
| 2842 |
+
"step": 235500
|
| 2843 |
+
},
|
| 2844 |
+
{
|
| 2845 |
+
"epoch": 1.29,
|
| 2846 |
+
"learning_rate": 1.137568747830218e-05,
|
| 2847 |
+
"loss": 0.0334,
|
| 2848 |
+
"step": 236000
|
| 2849 |
+
},
|
| 2850 |
+
{
|
| 2851 |
+
"epoch": 1.3,
|
| 2852 |
+
"learning_rate": 1.135741562973926e-05,
|
| 2853 |
+
"loss": 0.0369,
|
| 2854 |
+
"step": 236500
|
| 2855 |
+
},
|
| 2856 |
+
{
|
| 2857 |
+
"epoch": 1.3,
|
| 2858 |
+
"learning_rate": 1.1339143781176343e-05,
|
| 2859 |
+
"loss": 0.0326,
|
| 2860 |
+
"step": 237000
|
| 2861 |
+
},
|
| 2862 |
+
{
|
| 2863 |
+
"epoch": 1.3,
|
| 2864 |
+
"learning_rate": 1.1320871932613424e-05,
|
| 2865 |
+
"loss": 0.0429,
|
| 2866 |
+
"step": 237500
|
| 2867 |
+
},
|
| 2868 |
+
{
|
| 2869 |
+
"epoch": 1.3,
|
| 2870 |
+
"learning_rate": 1.1302600084050504e-05,
|
| 2871 |
+
"loss": 0.0433,
|
| 2872 |
+
"step": 238000
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 1.31,
|
| 2876 |
+
"learning_rate": 1.1284328235487585e-05,
|
| 2877 |
+
"loss": 0.0383,
|
| 2878 |
+
"step": 238500
|
| 2879 |
+
},
|
| 2880 |
+
{
|
| 2881 |
+
"epoch": 1.31,
|
| 2882 |
+
"learning_rate": 1.1266056386924666e-05,
|
| 2883 |
+
"loss": 0.0335,
|
| 2884 |
+
"step": 239000
|
| 2885 |
+
},
|
| 2886 |
+
{
|
| 2887 |
+
"epoch": 1.31,
|
| 2888 |
+
"learning_rate": 1.1247784538361748e-05,
|
| 2889 |
+
"loss": 0.0388,
|
| 2890 |
+
"step": 239500
|
| 2891 |
+
},
|
| 2892 |
+
{
|
| 2893 |
+
"epoch": 1.32,
|
| 2894 |
+
"learning_rate": 1.1229512689798827e-05,
|
| 2895 |
+
"loss": 0.0349,
|
| 2896 |
+
"step": 240000
|
| 2897 |
+
},
|
| 2898 |
+
{
|
| 2899 |
+
"epoch": 1.32,
|
| 2900 |
+
"learning_rate": 1.121124084123591e-05,
|
| 2901 |
+
"loss": 0.0311,
|
| 2902 |
+
"step": 240500
|
| 2903 |
+
},
|
| 2904 |
+
{
|
| 2905 |
+
"epoch": 1.32,
|
| 2906 |
+
"learning_rate": 1.119296899267299e-05,
|
| 2907 |
+
"loss": 0.0453,
|
| 2908 |
+
"step": 241000
|
| 2909 |
+
},
|
| 2910 |
+
{
|
| 2911 |
+
"epoch": 1.32,
|
| 2912 |
+
"learning_rate": 1.117469714411007e-05,
|
| 2913 |
+
"loss": 0.0392,
|
| 2914 |
+
"step": 241500
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 1.33,
|
| 2918 |
+
"learning_rate": 1.1156425295547152e-05,
|
| 2919 |
+
"loss": 0.0429,
|
| 2920 |
+
"step": 242000
|
| 2921 |
+
},
|
| 2922 |
+
{
|
| 2923 |
+
"epoch": 1.33,
|
| 2924 |
+
"learning_rate": 1.1138153446984232e-05,
|
| 2925 |
+
"loss": 0.0441,
|
| 2926 |
+
"step": 242500
|
| 2927 |
+
},
|
| 2928 |
+
{
|
| 2929 |
+
"epoch": 1.33,
|
| 2930 |
+
"learning_rate": 1.1119881598421315e-05,
|
| 2931 |
+
"loss": 0.0365,
|
| 2932 |
+
"step": 243000
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 1.33,
|
| 2936 |
+
"learning_rate": 1.1101609749858394e-05,
|
| 2937 |
+
"loss": 0.0374,
|
| 2938 |
+
"step": 243500
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"epoch": 1.34,
|
| 2942 |
+
"learning_rate": 1.1083337901295474e-05,
|
| 2943 |
+
"loss": 0.0434,
|
| 2944 |
+
"step": 244000
|
| 2945 |
+
},
|
| 2946 |
+
{
|
| 2947 |
+
"epoch": 1.34,
|
| 2948 |
+
"learning_rate": 1.1065066052732557e-05,
|
| 2949 |
+
"loss": 0.0411,
|
| 2950 |
+
"step": 244500
|
| 2951 |
+
},
|
| 2952 |
+
{
|
| 2953 |
+
"epoch": 1.34,
|
| 2954 |
+
"learning_rate": 1.1046794204169637e-05,
|
| 2955 |
+
"loss": 0.0409,
|
| 2956 |
+
"step": 245000
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 1.35,
|
| 2960 |
+
"learning_rate": 1.1028522355606716e-05,
|
| 2961 |
+
"loss": 0.0373,
|
| 2962 |
+
"step": 245500
|
| 2963 |
+
},
|
| 2964 |
+
{
|
| 2965 |
+
"epoch": 1.35,
|
| 2966 |
+
"learning_rate": 1.1010250507043799e-05,
|
| 2967 |
+
"loss": 0.0427,
|
| 2968 |
+
"step": 246000
|
| 2969 |
+
},
|
| 2970 |
+
{
|
| 2971 |
+
"epoch": 1.35,
|
| 2972 |
+
"learning_rate": 1.099197865848088e-05,
|
| 2973 |
+
"loss": 0.0351,
|
| 2974 |
+
"step": 246500
|
| 2975 |
+
},
|
| 2976 |
+
{
|
| 2977 |
+
"epoch": 1.35,
|
| 2978 |
+
"learning_rate": 1.0973706809917962e-05,
|
| 2979 |
+
"loss": 0.0285,
|
| 2980 |
+
"step": 247000
|
| 2981 |
+
},
|
| 2982 |
+
{
|
| 2983 |
+
"epoch": 1.36,
|
| 2984 |
+
"learning_rate": 1.0955434961355041e-05,
|
| 2985 |
+
"loss": 0.0364,
|
| 2986 |
+
"step": 247500
|
| 2987 |
+
},
|
| 2988 |
+
{
|
| 2989 |
+
"epoch": 1.36,
|
| 2990 |
+
"learning_rate": 1.0937163112792122e-05,
|
| 2991 |
+
"loss": 0.0448,
|
| 2992 |
+
"step": 248000
|
| 2993 |
+
},
|
| 2994 |
+
{
|
| 2995 |
+
"epoch": 1.36,
|
| 2996 |
+
"learning_rate": 1.0918891264229204e-05,
|
| 2997 |
+
"loss": 0.0333,
|
| 2998 |
+
"step": 248500
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 1.36,
|
| 3002 |
+
"learning_rate": 1.0900619415666285e-05,
|
| 3003 |
+
"loss": 0.0427,
|
| 3004 |
+
"step": 249000
|
| 3005 |
+
},
|
| 3006 |
+
{
|
| 3007 |
+
"epoch": 1.37,
|
| 3008 |
+
"learning_rate": 1.0882347567103364e-05,
|
| 3009 |
+
"loss": 0.0393,
|
| 3010 |
+
"step": 249500
|
| 3011 |
+
},
|
| 3012 |
+
{
|
| 3013 |
+
"epoch": 1.37,
|
| 3014 |
+
"learning_rate": 1.0864075718540446e-05,
|
| 3015 |
+
"loss": 0.0413,
|
| 3016 |
+
"step": 250000
|
| 3017 |
+
},
|
| 3018 |
+
{
|
| 3019 |
+
"epoch": 1.37,
|
| 3020 |
+
"learning_rate": 1.0845803869977527e-05,
|
| 3021 |
+
"loss": 0.0322,
|
| 3022 |
+
"step": 250500
|
| 3023 |
+
},
|
| 3024 |
+
{
|
| 3025 |
+
"epoch": 1.38,
|
| 3026 |
+
"learning_rate": 1.082753202141461e-05,
|
| 3027 |
+
"loss": 0.0341,
|
| 3028 |
+
"step": 251000
|
| 3029 |
+
},
|
| 3030 |
+
{
|
| 3031 |
+
"epoch": 1.38,
|
| 3032 |
+
"learning_rate": 1.0809260172851688e-05,
|
| 3033 |
+
"loss": 0.0363,
|
| 3034 |
+
"step": 251500
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 1.38,
|
| 3038 |
+
"learning_rate": 1.0790988324288769e-05,
|
| 3039 |
+
"loss": 0.0316,
|
| 3040 |
+
"step": 252000
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 1.38,
|
| 3044 |
+
"learning_rate": 1.0772716475725851e-05,
|
| 3045 |
+
"loss": 0.0393,
|
| 3046 |
+
"step": 252500
|
| 3047 |
+
},
|
| 3048 |
+
{
|
| 3049 |
+
"epoch": 1.39,
|
| 3050 |
+
"learning_rate": 1.075444462716293e-05,
|
| 3051 |
+
"loss": 0.0328,
|
| 3052 |
+
"step": 253000
|
| 3053 |
+
},
|
| 3054 |
+
{
|
| 3055 |
+
"epoch": 1.39,
|
| 3056 |
+
"learning_rate": 1.0736172778600011e-05,
|
| 3057 |
+
"loss": 0.0354,
|
| 3058 |
+
"step": 253500
|
| 3059 |
+
},
|
| 3060 |
+
{
|
| 3061 |
+
"epoch": 1.39,
|
| 3062 |
+
"learning_rate": 1.0717900930037093e-05,
|
| 3063 |
+
"loss": 0.03,
|
| 3064 |
+
"step": 254000
|
| 3065 |
+
},
|
| 3066 |
+
{
|
| 3067 |
+
"epoch": 1.4,
|
| 3068 |
+
"learning_rate": 1.0699629081474174e-05,
|
| 3069 |
+
"loss": 0.0423,
|
| 3070 |
+
"step": 254500
|
| 3071 |
+
},
|
| 3072 |
+
{
|
| 3073 |
+
"epoch": 1.4,
|
| 3074 |
+
"learning_rate": 1.0681357232911253e-05,
|
| 3075 |
+
"loss": 0.04,
|
| 3076 |
+
"step": 255000
|
| 3077 |
+
},
|
| 3078 |
+
{
|
| 3079 |
+
"epoch": 1.4,
|
| 3080 |
+
"learning_rate": 1.0663085384348335e-05,
|
| 3081 |
+
"loss": 0.029,
|
| 3082 |
+
"step": 255500
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 1.4,
|
| 3086 |
+
"learning_rate": 1.0644813535785416e-05,
|
| 3087 |
+
"loss": 0.039,
|
| 3088 |
+
"step": 256000
|
| 3089 |
+
},
|
| 3090 |
+
{
|
| 3091 |
+
"epoch": 1.41,
|
| 3092 |
+
"learning_rate": 1.0626541687222499e-05,
|
| 3093 |
+
"loss": 0.0335,
|
| 3094 |
+
"step": 256500
|
| 3095 |
+
},
|
| 3096 |
+
{
|
| 3097 |
+
"epoch": 1.41,
|
| 3098 |
+
"learning_rate": 1.0608269838659578e-05,
|
| 3099 |
+
"loss": 0.0317,
|
| 3100 |
+
"step": 257000
|
| 3101 |
+
},
|
| 3102 |
+
{
|
| 3103 |
+
"epoch": 1.41,
|
| 3104 |
+
"learning_rate": 1.0589997990096658e-05,
|
| 3105 |
+
"loss": 0.0515,
|
| 3106 |
+
"step": 257500
|
| 3107 |
+
},
|
| 3108 |
+
{
|
| 3109 |
+
"epoch": 1.41,
|
| 3110 |
+
"learning_rate": 1.057172614153374e-05,
|
| 3111 |
+
"loss": 0.0376,
|
| 3112 |
+
"step": 258000
|
| 3113 |
+
},
|
| 3114 |
+
{
|
| 3115 |
+
"epoch": 1.42,
|
| 3116 |
+
"learning_rate": 1.0553454292970821e-05,
|
| 3117 |
+
"loss": 0.0413,
|
| 3118 |
+
"step": 258500
|
| 3119 |
+
},
|
| 3120 |
+
{
|
| 3121 |
+
"epoch": 1.42,
|
| 3122 |
+
"learning_rate": 1.05351824444079e-05,
|
| 3123 |
+
"loss": 0.0363,
|
| 3124 |
+
"step": 259000
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 1.42,
|
| 3128 |
+
"learning_rate": 1.0516910595844983e-05,
|
| 3129 |
+
"loss": 0.0336,
|
| 3130 |
+
"step": 259500
|
| 3131 |
+
},
|
| 3132 |
+
{
|
| 3133 |
+
"epoch": 1.43,
|
| 3134 |
+
"learning_rate": 1.0498638747282063e-05,
|
| 3135 |
+
"loss": 0.0405,
|
| 3136 |
+
"step": 260000
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 1.43,
|
| 3140 |
+
"learning_rate": 1.0480366898719146e-05,
|
| 3141 |
+
"loss": 0.042,
|
| 3142 |
+
"step": 260500
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 1.43,
|
| 3146 |
+
"learning_rate": 1.0462095050156225e-05,
|
| 3147 |
+
"loss": 0.0366,
|
| 3148 |
+
"step": 261000
|
| 3149 |
+
},
|
| 3150 |
+
{
|
| 3151 |
+
"epoch": 1.43,
|
| 3152 |
+
"learning_rate": 1.0443823201593306e-05,
|
| 3153 |
+
"loss": 0.041,
|
| 3154 |
+
"step": 261500
|
| 3155 |
+
},
|
| 3156 |
+
{
|
| 3157 |
+
"epoch": 1.44,
|
| 3158 |
+
"learning_rate": 1.0425551353030388e-05,
|
| 3159 |
+
"loss": 0.0277,
|
| 3160 |
+
"step": 262000
|
| 3161 |
+
},
|
| 3162 |
+
{
|
| 3163 |
+
"epoch": 1.44,
|
| 3164 |
+
"learning_rate": 1.0407279504467467e-05,
|
| 3165 |
+
"loss": 0.0321,
|
| 3166 |
+
"step": 262500
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 1.44,
|
| 3170 |
+
"learning_rate": 1.0389007655904548e-05,
|
| 3171 |
+
"loss": 0.0412,
|
| 3172 |
+
"step": 263000
|
| 3173 |
+
},
|
| 3174 |
+
{
|
| 3175 |
+
"epoch": 1.44,
|
| 3176 |
+
"learning_rate": 1.037073580734163e-05,
|
| 3177 |
+
"loss": 0.0411,
|
| 3178 |
+
"step": 263500
|
| 3179 |
+
},
|
| 3180 |
+
{
|
| 3181 |
+
"epoch": 1.45,
|
| 3182 |
+
"learning_rate": 1.035246395877871e-05,
|
| 3183 |
+
"loss": 0.0294,
|
| 3184 |
+
"step": 264000
|
| 3185 |
+
},
|
| 3186 |
+
{
|
| 3187 |
+
"epoch": 1.45,
|
| 3188 |
+
"learning_rate": 1.033419211021579e-05,
|
| 3189 |
+
"loss": 0.0361,
|
| 3190 |
+
"step": 264500
|
| 3191 |
+
},
|
| 3192 |
+
{
|
| 3193 |
+
"epoch": 1.45,
|
| 3194 |
+
"learning_rate": 1.0315920261652872e-05,
|
| 3195 |
+
"loss": 0.0506,
|
| 3196 |
+
"step": 265000
|
| 3197 |
+
},
|
| 3198 |
+
{
|
| 3199 |
+
"epoch": 1.46,
|
| 3200 |
+
"learning_rate": 1.0297648413089953e-05,
|
| 3201 |
+
"loss": 0.038,
|
| 3202 |
+
"step": 265500
|
| 3203 |
+
},
|
| 3204 |
+
{
|
| 3205 |
+
"epoch": 1.46,
|
| 3206 |
+
"learning_rate": 1.0279376564527035e-05,
|
| 3207 |
+
"loss": 0.0323,
|
| 3208 |
+
"step": 266000
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 1.46,
|
| 3212 |
+
"learning_rate": 1.0261104715964114e-05,
|
| 3213 |
+
"loss": 0.0346,
|
| 3214 |
+
"step": 266500
|
| 3215 |
+
},
|
| 3216 |
+
{
|
| 3217 |
+
"epoch": 1.46,
|
| 3218 |
+
"learning_rate": 1.0242832867401195e-05,
|
| 3219 |
+
"loss": 0.0336,
|
| 3220 |
+
"step": 267000
|
| 3221 |
+
},
|
| 3222 |
+
{
|
| 3223 |
+
"epoch": 1.47,
|
| 3224 |
+
"learning_rate": 1.0224561018838277e-05,
|
| 3225 |
+
"loss": 0.0387,
|
| 3226 |
+
"step": 267500
|
| 3227 |
+
},
|
| 3228 |
+
{
|
| 3229 |
+
"epoch": 1.47,
|
| 3230 |
+
"learning_rate": 1.0206289170275358e-05,
|
| 3231 |
+
"loss": 0.0362,
|
| 3232 |
+
"step": 268000
|
| 3233 |
+
},
|
| 3234 |
+
{
|
| 3235 |
+
"epoch": 1.47,
|
| 3236 |
+
"learning_rate": 1.0188017321712437e-05,
|
| 3237 |
+
"loss": 0.0304,
|
| 3238 |
+
"step": 268500
|
| 3239 |
+
},
|
| 3240 |
+
{
|
| 3241 |
+
"epoch": 1.47,
|
| 3242 |
+
"learning_rate": 1.016974547314952e-05,
|
| 3243 |
+
"loss": 0.038,
|
| 3244 |
+
"step": 269000
|
| 3245 |
+
},
|
| 3246 |
+
{
|
| 3247 |
+
"epoch": 1.48,
|
| 3248 |
+
"learning_rate": 1.01514736245866e-05,
|
| 3249 |
+
"loss": 0.036,
|
| 3250 |
+
"step": 269500
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 1.48,
|
| 3254 |
+
"learning_rate": 1.013320177602368e-05,
|
| 3255 |
+
"loss": 0.0385,
|
| 3256 |
+
"step": 270000
|
| 3257 |
+
},
|
| 3258 |
+
{
|
| 3259 |
+
"epoch": 1.48,
|
| 3260 |
+
"learning_rate": 1.0114929927460762e-05,
|
| 3261 |
+
"loss": 0.033,
|
| 3262 |
+
"step": 270500
|
| 3263 |
+
},
|
| 3264 |
+
{
|
| 3265 |
+
"epoch": 1.49,
|
| 3266 |
+
"learning_rate": 1.0096658078897842e-05,
|
| 3267 |
+
"loss": 0.0283,
|
| 3268 |
+
"step": 271000
|
| 3269 |
+
},
|
| 3270 |
+
{
|
| 3271 |
+
"epoch": 1.49,
|
| 3272 |
+
"learning_rate": 1.0078386230334925e-05,
|
| 3273 |
+
"loss": 0.0372,
|
| 3274 |
+
"step": 271500
|
| 3275 |
+
},
|
| 3276 |
+
{
|
| 3277 |
+
"epoch": 1.49,
|
| 3278 |
+
"learning_rate": 1.0060114381772004e-05,
|
| 3279 |
+
"loss": 0.0387,
|
| 3280 |
+
"step": 272000
|
| 3281 |
+
},
|
| 3282 |
+
{
|
| 3283 |
+
"epoch": 1.49,
|
| 3284 |
+
"learning_rate": 1.0041842533209086e-05,
|
| 3285 |
+
"loss": 0.0378,
|
| 3286 |
+
"step": 272500
|
| 3287 |
+
},
|
| 3288 |
+
{
|
| 3289 |
+
"epoch": 1.5,
|
| 3290 |
+
"learning_rate": 1.0023570684646167e-05,
|
| 3291 |
+
"loss": 0.0308,
|
| 3292 |
+
"step": 273000
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 1.5,
|
| 3296 |
+
"learning_rate": 1.0005298836083247e-05,
|
| 3297 |
+
"loss": 0.0363,
|
| 3298 |
+
"step": 273500
|
| 3299 |
+
},
|
| 3300 |
+
{
|
| 3301 |
+
"epoch": 1.5,
|
| 3302 |
+
"learning_rate": 9.987026987520328e-06,
|
| 3303 |
+
"loss": 0.0289,
|
| 3304 |
+
"step": 274000
|
| 3305 |
+
},
|
| 3306 |
+
{
|
| 3307 |
+
"epoch": 1.5,
|
| 3308 |
+
"learning_rate": 9.968755138957409e-06,
|
| 3309 |
+
"loss": 0.0329,
|
| 3310 |
+
"step": 274500
|
| 3311 |
+
},
|
| 3312 |
+
{
|
| 3313 |
+
"epoch": 1.51,
|
| 3314 |
+
"learning_rate": 9.95048329039449e-06,
|
| 3315 |
+
"loss": 0.0385,
|
| 3316 |
+
"step": 275000
|
| 3317 |
+
},
|
| 3318 |
+
{
|
| 3319 |
+
"epoch": 1.51,
|
| 3320 |
+
"learning_rate": 9.93221144183157e-06,
|
| 3321 |
+
"loss": 0.0415,
|
| 3322 |
+
"step": 275500
|
| 3323 |
+
},
|
| 3324 |
+
{
|
| 3325 |
+
"epoch": 1.51,
|
| 3326 |
+
"learning_rate": 9.913939593268653e-06,
|
| 3327 |
+
"loss": 0.0353,
|
| 3328 |
+
"step": 276000
|
| 3329 |
+
},
|
| 3330 |
+
{
|
| 3331 |
+
"epoch": 1.52,
|
| 3332 |
+
"learning_rate": 9.895667744705733e-06,
|
| 3333 |
+
"loss": 0.0308,
|
| 3334 |
+
"step": 276500
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 1.52,
|
| 3338 |
+
"learning_rate": 9.877395896142814e-06,
|
| 3339 |
+
"loss": 0.0356,
|
| 3340 |
+
"step": 277000
|
| 3341 |
+
},
|
| 3342 |
+
{
|
| 3343 |
+
"epoch": 1.52,
|
| 3344 |
+
"learning_rate": 9.859124047579895e-06,
|
| 3345 |
+
"loss": 0.0348,
|
| 3346 |
+
"step": 277500
|
| 3347 |
+
},
|
| 3348 |
+
{
|
| 3349 |
+
"epoch": 1.52,
|
| 3350 |
+
"learning_rate": 9.840852199016975e-06,
|
| 3351 |
+
"loss": 0.0439,
|
| 3352 |
+
"step": 278000
|
| 3353 |
+
},
|
| 3354 |
+
{
|
| 3355 |
+
"epoch": 1.53,
|
| 3356 |
+
"learning_rate": 9.822580350454056e-06,
|
| 3357 |
+
"loss": 0.0329,
|
| 3358 |
+
"step": 278500
|
| 3359 |
+
},
|
| 3360 |
+
{
|
| 3361 |
+
"epoch": 1.53,
|
| 3362 |
+
"learning_rate": 9.804308501891137e-06,
|
| 3363 |
+
"loss": 0.0386,
|
| 3364 |
+
"step": 279000
|
| 3365 |
+
},
|
| 3366 |
+
{
|
| 3367 |
+
"epoch": 1.53,
|
| 3368 |
+
"learning_rate": 9.786036653328217e-06,
|
| 3369 |
+
"loss": 0.0445,
|
| 3370 |
+
"step": 279500
|
| 3371 |
+
},
|
| 3372 |
+
{
|
| 3373 |
+
"epoch": 1.53,
|
| 3374 |
+
"learning_rate": 9.767764804765298e-06,
|
| 3375 |
+
"loss": 0.0385,
|
| 3376 |
+
"step": 280000
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 1.54,
|
| 3380 |
+
"learning_rate": 9.74949295620238e-06,
|
| 3381 |
+
"loss": 0.0308,
|
| 3382 |
+
"step": 280500
|
| 3383 |
+
},
|
| 3384 |
+
{
|
| 3385 |
+
"epoch": 1.54,
|
| 3386 |
+
"learning_rate": 9.73122110763946e-06,
|
| 3387 |
+
"loss": 0.0322,
|
| 3388 |
+
"step": 281000
|
| 3389 |
+
},
|
| 3390 |
+
{
|
| 3391 |
+
"epoch": 1.54,
|
| 3392 |
+
"learning_rate": 9.712949259076542e-06,
|
| 3393 |
+
"loss": 0.0424,
|
| 3394 |
+
"step": 281500
|
| 3395 |
+
},
|
| 3396 |
+
{
|
| 3397 |
+
"epoch": 1.55,
|
| 3398 |
+
"learning_rate": 9.694677410513623e-06,
|
| 3399 |
+
"loss": 0.0274,
|
| 3400 |
+
"step": 282000
|
| 3401 |
+
},
|
| 3402 |
+
{
|
| 3403 |
+
"epoch": 1.55,
|
| 3404 |
+
"learning_rate": 9.676405561950703e-06,
|
| 3405 |
+
"loss": 0.0471,
|
| 3406 |
+
"step": 282500
|
| 3407 |
+
},
|
| 3408 |
+
{
|
| 3409 |
+
"epoch": 1.55,
|
| 3410 |
+
"learning_rate": 9.658133713387784e-06,
|
| 3411 |
+
"loss": 0.0425,
|
| 3412 |
+
"step": 283000
|
| 3413 |
+
},
|
| 3414 |
+
{
|
| 3415 |
+
"epoch": 1.55,
|
| 3416 |
+
"learning_rate": 9.639861864824865e-06,
|
| 3417 |
+
"loss": 0.0353,
|
| 3418 |
+
"step": 283500
|
| 3419 |
+
},
|
| 3420 |
+
{
|
| 3421 |
+
"epoch": 1.56,
|
| 3422 |
+
"learning_rate": 9.621590016261945e-06,
|
| 3423 |
+
"loss": 0.0364,
|
| 3424 |
+
"step": 284000
|
| 3425 |
+
},
|
| 3426 |
+
{
|
| 3427 |
+
"epoch": 1.56,
|
| 3428 |
+
"learning_rate": 9.603318167699028e-06,
|
| 3429 |
+
"loss": 0.0258,
|
| 3430 |
+
"step": 284500
|
| 3431 |
+
},
|
| 3432 |
+
{
|
| 3433 |
+
"epoch": 1.56,
|
| 3434 |
+
"learning_rate": 9.585046319136107e-06,
|
| 3435 |
+
"loss": 0.0458,
|
| 3436 |
+
"step": 285000
|
| 3437 |
+
},
|
| 3438 |
+
{
|
| 3439 |
+
"epoch": 1.56,
|
| 3440 |
+
"learning_rate": 9.56677447057319e-06,
|
| 3441 |
+
"loss": 0.0319,
|
| 3442 |
+
"step": 285500
|
| 3443 |
+
},
|
| 3444 |
+
{
|
| 3445 |
+
"epoch": 1.57,
|
| 3446 |
+
"learning_rate": 9.54850262201027e-06,
|
| 3447 |
+
"loss": 0.045,
|
| 3448 |
+
"step": 286000
|
| 3449 |
+
},
|
| 3450 |
+
{
|
| 3451 |
+
"epoch": 1.57,
|
| 3452 |
+
"learning_rate": 9.53023077344735e-06,
|
| 3453 |
+
"loss": 0.0317,
|
| 3454 |
+
"step": 286500
|
| 3455 |
+
},
|
| 3456 |
+
{
|
| 3457 |
+
"epoch": 1.57,
|
| 3458 |
+
"learning_rate": 9.511958924884431e-06,
|
| 3459 |
+
"loss": 0.0307,
|
| 3460 |
+
"step": 287000
|
| 3461 |
+
},
|
| 3462 |
+
{
|
| 3463 |
+
"epoch": 1.58,
|
| 3464 |
+
"learning_rate": 9.493687076321512e-06,
|
| 3465 |
+
"loss": 0.0373,
|
| 3466 |
+
"step": 287500
|
| 3467 |
+
},
|
| 3468 |
+
{
|
| 3469 |
+
"epoch": 1.58,
|
| 3470 |
+
"learning_rate": 9.475415227758593e-06,
|
| 3471 |
+
"loss": 0.0294,
|
| 3472 |
+
"step": 288000
|
| 3473 |
+
},
|
| 3474 |
+
{
|
| 3475 |
+
"epoch": 1.58,
|
| 3476 |
+
"learning_rate": 9.457143379195673e-06,
|
| 3477 |
+
"loss": 0.038,
|
| 3478 |
+
"step": 288500
|
| 3479 |
+
},
|
| 3480 |
+
{
|
| 3481 |
+
"epoch": 1.58,
|
| 3482 |
+
"learning_rate": 9.438871530632754e-06,
|
| 3483 |
+
"loss": 0.0328,
|
| 3484 |
+
"step": 289000
|
| 3485 |
+
},
|
| 3486 |
+
{
|
| 3487 |
+
"epoch": 1.59,
|
| 3488 |
+
"learning_rate": 9.420599682069835e-06,
|
| 3489 |
+
"loss": 0.0416,
|
| 3490 |
+
"step": 289500
|
| 3491 |
+
},
|
| 3492 |
+
{
|
| 3493 |
+
"epoch": 1.59,
|
| 3494 |
+
"learning_rate": 9.402327833506917e-06,
|
| 3495 |
+
"loss": 0.033,
|
| 3496 |
+
"step": 290000
|
| 3497 |
+
},
|
| 3498 |
+
{
|
| 3499 |
+
"epoch": 1.59,
|
| 3500 |
+
"learning_rate": 9.384055984943998e-06,
|
| 3501 |
+
"loss": 0.0253,
|
| 3502 |
+
"step": 290500
|
| 3503 |
+
},
|
| 3504 |
+
{
|
| 3505 |
+
"epoch": 1.6,
|
| 3506 |
+
"learning_rate": 9.365784136381079e-06,
|
| 3507 |
+
"loss": 0.0316,
|
| 3508 |
+
"step": 291000
|
| 3509 |
+
},
|
| 3510 |
+
{
|
| 3511 |
+
"epoch": 1.6,
|
| 3512 |
+
"learning_rate": 9.34751228781816e-06,
|
| 3513 |
+
"loss": 0.0327,
|
| 3514 |
+
"step": 291500
|
| 3515 |
+
},
|
| 3516 |
+
{
|
| 3517 |
+
"epoch": 1.6,
|
| 3518 |
+
"learning_rate": 9.32924043925524e-06,
|
| 3519 |
+
"loss": 0.0287,
|
| 3520 |
+
"step": 292000
|
| 3521 |
+
},
|
| 3522 |
+
{
|
| 3523 |
+
"epoch": 1.6,
|
| 3524 |
+
"learning_rate": 9.31096859069232e-06,
|
| 3525 |
+
"loss": 0.0327,
|
| 3526 |
+
"step": 292500
|
| 3527 |
+
},
|
| 3528 |
+
{
|
| 3529 |
+
"epoch": 1.61,
|
| 3530 |
+
"learning_rate": 9.292696742129403e-06,
|
| 3531 |
+
"loss": 0.0341,
|
| 3532 |
+
"step": 293000
|
| 3533 |
+
},
|
| 3534 |
+
{
|
| 3535 |
+
"epoch": 1.61,
|
| 3536 |
+
"learning_rate": 9.274424893566482e-06,
|
| 3537 |
+
"loss": 0.0399,
|
| 3538 |
+
"step": 293500
|
| 3539 |
+
},
|
| 3540 |
+
{
|
| 3541 |
+
"epoch": 1.61,
|
| 3542 |
+
"learning_rate": 9.256153045003565e-06,
|
| 3543 |
+
"loss": 0.021,
|
| 3544 |
+
"step": 294000
|
| 3545 |
+
},
|
| 3546 |
+
{
|
| 3547 |
+
"epoch": 1.61,
|
| 3548 |
+
"learning_rate": 9.237881196440645e-06,
|
| 3549 |
+
"loss": 0.0511,
|
| 3550 |
+
"step": 294500
|
| 3551 |
+
},
|
| 3552 |
+
{
|
| 3553 |
+
"epoch": 1.62,
|
| 3554 |
+
"learning_rate": 9.219609347877726e-06,
|
| 3555 |
+
"loss": 0.0241,
|
| 3556 |
+
"step": 295000
|
| 3557 |
+
},
|
| 3558 |
+
{
|
| 3559 |
+
"epoch": 1.62,
|
| 3560 |
+
"learning_rate": 9.201337499314807e-06,
|
| 3561 |
+
"loss": 0.0296,
|
| 3562 |
+
"step": 295500
|
| 3563 |
+
},
|
| 3564 |
+
{
|
| 3565 |
+
"epoch": 1.62,
|
| 3566 |
+
"learning_rate": 9.183065650751887e-06,
|
| 3567 |
+
"loss": 0.037,
|
| 3568 |
+
"step": 296000
|
| 3569 |
+
},
|
| 3570 |
+
{
|
| 3571 |
+
"epoch": 1.63,
|
| 3572 |
+
"learning_rate": 9.164793802188968e-06,
|
| 3573 |
+
"loss": 0.0362,
|
| 3574 |
+
"step": 296500
|
| 3575 |
+
},
|
| 3576 |
+
{
|
| 3577 |
+
"epoch": 1.63,
|
| 3578 |
+
"learning_rate": 9.146521953626049e-06,
|
| 3579 |
+
"loss": 0.0349,
|
| 3580 |
+
"step": 297000
|
| 3581 |
+
},
|
| 3582 |
+
{
|
| 3583 |
+
"epoch": 1.63,
|
| 3584 |
+
"learning_rate": 9.12825010506313e-06,
|
| 3585 |
+
"loss": 0.0367,
|
| 3586 |
+
"step": 297500
|
| 3587 |
+
},
|
| 3588 |
+
{
|
| 3589 |
+
"epoch": 1.63,
|
| 3590 |
+
"learning_rate": 9.10997825650021e-06,
|
| 3591 |
+
"loss": 0.0343,
|
| 3592 |
+
"step": 298000
|
| 3593 |
+
},
|
| 3594 |
+
{
|
| 3595 |
+
"epoch": 1.64,
|
| 3596 |
+
"learning_rate": 9.091706407937293e-06,
|
| 3597 |
+
"loss": 0.0296,
|
| 3598 |
+
"step": 298500
|
| 3599 |
+
},
|
| 3600 |
+
{
|
| 3601 |
+
"epoch": 1.64,
|
| 3602 |
+
"learning_rate": 9.073434559374372e-06,
|
| 3603 |
+
"loss": 0.0305,
|
| 3604 |
+
"step": 299000
|
| 3605 |
+
},
|
| 3606 |
+
{
|
| 3607 |
+
"epoch": 1.64,
|
| 3608 |
+
"learning_rate": 9.055162710811454e-06,
|
| 3609 |
+
"loss": 0.0389,
|
| 3610 |
+
"step": 299500
|
| 3611 |
+
},
|
| 3612 |
+
{
|
| 3613 |
+
"epoch": 1.64,
|
| 3614 |
+
"learning_rate": 9.036890862248535e-06,
|
| 3615 |
+
"loss": 0.0324,
|
| 3616 |
+
"step": 300000
|
| 3617 |
+
},
|
| 3618 |
+
{
|
| 3619 |
+
"epoch": 1.65,
|
| 3620 |
+
"learning_rate": 9.018619013685615e-06,
|
| 3621 |
+
"loss": 0.0243,
|
| 3622 |
+
"step": 300500
|
| 3623 |
+
},
|
| 3624 |
+
{
|
| 3625 |
+
"epoch": 1.65,
|
| 3626 |
+
"learning_rate": 9.000347165122696e-06,
|
| 3627 |
+
"loss": 0.0287,
|
| 3628 |
+
"step": 301000
|
| 3629 |
+
},
|
| 3630 |
+
{
|
| 3631 |
+
"epoch": 1.65,
|
| 3632 |
+
"learning_rate": 8.982075316559777e-06,
|
| 3633 |
+
"loss": 0.0306,
|
| 3634 |
+
"step": 301500
|
| 3635 |
+
},
|
| 3636 |
+
{
|
| 3637 |
+
"epoch": 1.66,
|
| 3638 |
+
"learning_rate": 8.963803467996857e-06,
|
| 3639 |
+
"loss": 0.0323,
|
| 3640 |
+
"step": 302000
|
| 3641 |
+
},
|
| 3642 |
+
{
|
| 3643 |
+
"epoch": 1.66,
|
| 3644 |
+
"learning_rate": 8.94553161943394e-06,
|
| 3645 |
+
"loss": 0.0309,
|
| 3646 |
+
"step": 302500
|
| 3647 |
+
},
|
| 3648 |
+
{
|
| 3649 |
+
"epoch": 1.66,
|
| 3650 |
+
"learning_rate": 8.927259770871019e-06,
|
| 3651 |
+
"loss": 0.031,
|
| 3652 |
+
"step": 303000
|
| 3653 |
+
},
|
| 3654 |
+
{
|
| 3655 |
+
"epoch": 1.66,
|
| 3656 |
+
"learning_rate": 8.908987922308101e-06,
|
| 3657 |
+
"loss": 0.0367,
|
| 3658 |
+
"step": 303500
|
| 3659 |
+
},
|
| 3660 |
+
{
|
| 3661 |
+
"epoch": 1.67,
|
| 3662 |
+
"learning_rate": 8.890716073745182e-06,
|
| 3663 |
+
"loss": 0.0331,
|
| 3664 |
+
"step": 304000
|
| 3665 |
+
},
|
| 3666 |
+
{
|
| 3667 |
+
"epoch": 1.67,
|
| 3668 |
+
"learning_rate": 8.872444225182263e-06,
|
| 3669 |
+
"loss": 0.0323,
|
| 3670 |
+
"step": 304500
|
| 3671 |
+
},
|
| 3672 |
+
{
|
| 3673 |
+
"epoch": 1.67,
|
| 3674 |
+
"learning_rate": 8.854172376619343e-06,
|
| 3675 |
+
"loss": 0.0362,
|
| 3676 |
+
"step": 305000
|
| 3677 |
+
},
|
| 3678 |
+
{
|
| 3679 |
+
"epoch": 1.67,
|
| 3680 |
+
"learning_rate": 8.835900528056424e-06,
|
| 3681 |
+
"loss": 0.0269,
|
| 3682 |
+
"step": 305500
|
| 3683 |
+
},
|
| 3684 |
+
{
|
| 3685 |
+
"epoch": 1.68,
|
| 3686 |
+
"learning_rate": 8.817628679493505e-06,
|
| 3687 |
+
"loss": 0.0395,
|
| 3688 |
+
"step": 306000
|
| 3689 |
+
},
|
| 3690 |
+
{
|
| 3691 |
+
"epoch": 1.68,
|
| 3692 |
+
"learning_rate": 8.799356830930585e-06,
|
| 3693 |
+
"loss": 0.0424,
|
| 3694 |
+
"step": 306500
|
| 3695 |
+
},
|
| 3696 |
+
{
|
| 3697 |
+
"epoch": 1.68,
|
| 3698 |
+
"learning_rate": 8.781084982367666e-06,
|
| 3699 |
+
"loss": 0.0296,
|
| 3700 |
+
"step": 307000
|
| 3701 |
+
},
|
| 3702 |
+
{
|
| 3703 |
+
"epoch": 1.69,
|
| 3704 |
+
"learning_rate": 8.762813133804747e-06,
|
| 3705 |
+
"loss": 0.0375,
|
| 3706 |
+
"step": 307500
|
| 3707 |
+
},
|
| 3708 |
+
{
|
| 3709 |
+
"epoch": 1.69,
|
| 3710 |
+
"learning_rate": 8.74454128524183e-06,
|
| 3711 |
+
"loss": 0.0321,
|
| 3712 |
+
"step": 308000
|
| 3713 |
+
},
|
| 3714 |
+
{
|
| 3715 |
+
"epoch": 1.69,
|
| 3716 |
+
"learning_rate": 8.72626943667891e-06,
|
| 3717 |
+
"loss": 0.0321,
|
| 3718 |
+
"step": 308500
|
| 3719 |
+
},
|
| 3720 |
+
{
|
| 3721 |
+
"epoch": 1.69,
|
| 3722 |
+
"learning_rate": 8.70799758811599e-06,
|
| 3723 |
+
"loss": 0.0331,
|
| 3724 |
+
"step": 309000
|
| 3725 |
+
},
|
| 3726 |
+
{
|
| 3727 |
+
"epoch": 1.7,
|
| 3728 |
+
"learning_rate": 8.689725739553071e-06,
|
| 3729 |
+
"loss": 0.0259,
|
| 3730 |
+
"step": 309500
|
| 3731 |
+
},
|
| 3732 |
+
{
|
| 3733 |
+
"epoch": 1.7,
|
| 3734 |
+
"learning_rate": 8.671453890990152e-06,
|
| 3735 |
+
"loss": 0.0258,
|
| 3736 |
+
"step": 310000
|
| 3737 |
+
},
|
| 3738 |
+
{
|
| 3739 |
+
"epoch": 1.7,
|
| 3740 |
+
"learning_rate": 8.653182042427233e-06,
|
| 3741 |
+
"loss": 0.035,
|
| 3742 |
+
"step": 310500
|
| 3743 |
+
},
|
| 3744 |
+
{
|
| 3745 |
+
"epoch": 1.7,
|
| 3746 |
+
"learning_rate": 8.634910193864315e-06,
|
| 3747 |
+
"loss": 0.027,
|
| 3748 |
+
"step": 311000
|
| 3749 |
+
},
|
| 3750 |
+
{
|
| 3751 |
+
"epoch": 1.71,
|
| 3752 |
+
"learning_rate": 8.616638345301394e-06,
|
| 3753 |
+
"loss": 0.0329,
|
| 3754 |
+
"step": 311500
|
| 3755 |
+
},
|
| 3756 |
+
{
|
| 3757 |
+
"epoch": 1.71,
|
| 3758 |
+
"learning_rate": 8.598366496738476e-06,
|
| 3759 |
+
"loss": 0.021,
|
| 3760 |
+
"step": 312000
|
| 3761 |
+
},
|
| 3762 |
+
{
|
| 3763 |
+
"epoch": 1.71,
|
| 3764 |
+
"learning_rate": 8.580094648175557e-06,
|
| 3765 |
+
"loss": 0.0329,
|
| 3766 |
+
"step": 312500
|
| 3767 |
+
},
|
| 3768 |
+
{
|
| 3769 |
+
"epoch": 1.72,
|
| 3770 |
+
"learning_rate": 8.561822799612638e-06,
|
| 3771 |
+
"loss": 0.0496,
|
| 3772 |
+
"step": 313000
|
| 3773 |
+
},
|
| 3774 |
+
{
|
| 3775 |
+
"epoch": 1.72,
|
| 3776 |
+
"learning_rate": 8.543550951049719e-06,
|
| 3777 |
+
"loss": 0.0434,
|
| 3778 |
+
"step": 313500
|
| 3779 |
+
},
|
| 3780 |
+
{
|
| 3781 |
+
"epoch": 1.72,
|
| 3782 |
+
"learning_rate": 8.5252791024868e-06,
|
| 3783 |
+
"loss": 0.0299,
|
| 3784 |
+
"step": 314000
|
| 3785 |
+
},
|
| 3786 |
+
{
|
| 3787 |
+
"epoch": 1.72,
|
| 3788 |
+
"learning_rate": 8.50700725392388e-06,
|
| 3789 |
+
"loss": 0.0358,
|
| 3790 |
+
"step": 314500
|
| 3791 |
+
},
|
| 3792 |
+
{
|
| 3793 |
+
"epoch": 1.73,
|
| 3794 |
+
"learning_rate": 8.48873540536096e-06,
|
| 3795 |
+
"loss": 0.0337,
|
| 3796 |
+
"step": 315000
|
| 3797 |
+
},
|
| 3798 |
+
{
|
| 3799 |
+
"epoch": 1.73,
|
| 3800 |
+
"learning_rate": 8.470463556798041e-06,
|
| 3801 |
+
"loss": 0.0348,
|
| 3802 |
+
"step": 315500
|
| 3803 |
+
},
|
| 3804 |
+
{
|
| 3805 |
+
"epoch": 1.73,
|
| 3806 |
+
"learning_rate": 8.452191708235122e-06,
|
| 3807 |
+
"loss": 0.0372,
|
| 3808 |
+
"step": 316000
|
| 3809 |
+
},
|
| 3810 |
+
{
|
| 3811 |
+
"epoch": 1.73,
|
| 3812 |
+
"learning_rate": 8.433919859672204e-06,
|
| 3813 |
+
"loss": 0.0284,
|
| 3814 |
+
"step": 316500
|
| 3815 |
+
},
|
| 3816 |
+
{
|
| 3817 |
+
"epoch": 1.74,
|
| 3818 |
+
"learning_rate": 8.415648011109283e-06,
|
| 3819 |
+
"loss": 0.0297,
|
| 3820 |
+
"step": 317000
|
| 3821 |
+
},
|
| 3822 |
+
{
|
| 3823 |
+
"epoch": 1.74,
|
| 3824 |
+
"learning_rate": 8.397376162546366e-06,
|
| 3825 |
+
"loss": 0.0294,
|
| 3826 |
+
"step": 317500
|
| 3827 |
+
},
|
| 3828 |
+
{
|
| 3829 |
+
"epoch": 1.74,
|
| 3830 |
+
"learning_rate": 8.379104313983447e-06,
|
| 3831 |
+
"loss": 0.0302,
|
| 3832 |
+
"step": 318000
|
| 3833 |
+
},
|
| 3834 |
+
{
|
| 3835 |
+
"epoch": 1.75,
|
| 3836 |
+
"learning_rate": 8.360832465420527e-06,
|
| 3837 |
+
"loss": 0.0305,
|
| 3838 |
+
"step": 318500
|
| 3839 |
+
},
|
| 3840 |
+
{
|
| 3841 |
+
"epoch": 1.75,
|
| 3842 |
+
"learning_rate": 8.342560616857608e-06,
|
| 3843 |
+
"loss": 0.0222,
|
| 3844 |
+
"step": 319000
|
| 3845 |
+
},
|
| 3846 |
+
{
|
| 3847 |
+
"epoch": 1.75,
|
| 3848 |
+
"learning_rate": 8.324288768294689e-06,
|
| 3849 |
+
"loss": 0.0289,
|
| 3850 |
+
"step": 319500
|
| 3851 |
+
},
|
| 3852 |
+
{
|
| 3853 |
+
"epoch": 1.75,
|
| 3854 |
+
"learning_rate": 8.30601691973177e-06,
|
| 3855 |
+
"loss": 0.0253,
|
| 3856 |
+
"step": 320000
|
| 3857 |
+
},
|
| 3858 |
+
{
|
| 3859 |
+
"epoch": 1.76,
|
| 3860 |
+
"learning_rate": 8.287745071168852e-06,
|
| 3861 |
+
"loss": 0.0365,
|
| 3862 |
+
"step": 320500
|
| 3863 |
+
},
|
| 3864 |
+
{
|
| 3865 |
+
"epoch": 1.76,
|
| 3866 |
+
"learning_rate": 8.26947322260593e-06,
|
| 3867 |
+
"loss": 0.0357,
|
| 3868 |
+
"step": 321000
|
| 3869 |
+
},
|
| 3870 |
+
{
|
| 3871 |
+
"epoch": 1.76,
|
| 3872 |
+
"learning_rate": 8.251201374043013e-06,
|
| 3873 |
+
"loss": 0.0365,
|
| 3874 |
+
"step": 321500
|
| 3875 |
+
},
|
| 3876 |
+
{
|
| 3877 |
+
"epoch": 1.77,
|
| 3878 |
+
"learning_rate": 8.232929525480094e-06,
|
| 3879 |
+
"loss": 0.0324,
|
| 3880 |
+
"step": 322000
|
| 3881 |
+
},
|
| 3882 |
+
{
|
| 3883 |
+
"epoch": 1.77,
|
| 3884 |
+
"learning_rate": 8.214657676917175e-06,
|
| 3885 |
+
"loss": 0.0319,
|
| 3886 |
+
"step": 322500
|
| 3887 |
+
},
|
| 3888 |
+
{
|
| 3889 |
+
"epoch": 1.77,
|
| 3890 |
+
"learning_rate": 8.196385828354255e-06,
|
| 3891 |
+
"loss": 0.0339,
|
| 3892 |
+
"step": 323000
|
| 3893 |
+
},
|
| 3894 |
+
{
|
| 3895 |
+
"epoch": 1.77,
|
| 3896 |
+
"learning_rate": 8.178113979791336e-06,
|
| 3897 |
+
"loss": 0.0271,
|
| 3898 |
+
"step": 323500
|
| 3899 |
+
},
|
| 3900 |
+
{
|
| 3901 |
+
"epoch": 1.78,
|
| 3902 |
+
"learning_rate": 8.159842131228417e-06,
|
| 3903 |
+
"loss": 0.031,
|
| 3904 |
+
"step": 324000
|
| 3905 |
+
},
|
| 3906 |
+
{
|
| 3907 |
+
"epoch": 1.78,
|
| 3908 |
+
"learning_rate": 8.141570282665497e-06,
|
| 3909 |
+
"loss": 0.0304,
|
| 3910 |
+
"step": 324500
|
| 3911 |
+
},
|
| 3912 |
+
{
|
| 3913 |
+
"epoch": 1.78,
|
| 3914 |
+
"learning_rate": 8.123298434102578e-06,
|
| 3915 |
+
"loss": 0.0335,
|
| 3916 |
+
"step": 325000
|
| 3917 |
+
},
|
| 3918 |
+
{
|
| 3919 |
+
"epoch": 1.78,
|
| 3920 |
+
"learning_rate": 8.105026585539659e-06,
|
| 3921 |
+
"loss": 0.0321,
|
| 3922 |
+
"step": 325500
|
| 3923 |
+
},
|
| 3924 |
+
{
|
| 3925 |
+
"epoch": 1.79,
|
| 3926 |
+
"learning_rate": 8.086754736976741e-06,
|
| 3927 |
+
"loss": 0.0231,
|
| 3928 |
+
"step": 326000
|
| 3929 |
+
},
|
| 3930 |
+
{
|
| 3931 |
+
"epoch": 1.79,
|
| 3932 |
+
"learning_rate": 8.068482888413822e-06,
|
| 3933 |
+
"loss": 0.0433,
|
| 3934 |
+
"step": 326500
|
| 3935 |
+
},
|
| 3936 |
+
{
|
| 3937 |
+
"epoch": 1.79,
|
| 3938 |
+
"learning_rate": 8.050211039850903e-06,
|
| 3939 |
+
"loss": 0.0387,
|
| 3940 |
+
"step": 327000
|
| 3941 |
+
},
|
| 3942 |
+
{
|
| 3943 |
+
"epoch": 1.8,
|
| 3944 |
+
"learning_rate": 8.031939191287983e-06,
|
| 3945 |
+
"loss": 0.038,
|
| 3946 |
+
"step": 327500
|
| 3947 |
+
},
|
| 3948 |
+
{
|
| 3949 |
+
"epoch": 1.8,
|
| 3950 |
+
"learning_rate": 8.013667342725064e-06,
|
| 3951 |
+
"loss": 0.0336,
|
| 3952 |
+
"step": 328000
|
| 3953 |
+
},
|
| 3954 |
+
{
|
| 3955 |
+
"epoch": 1.8,
|
| 3956 |
+
"learning_rate": 7.995395494162145e-06,
|
| 3957 |
+
"loss": 0.0404,
|
| 3958 |
+
"step": 328500
|
| 3959 |
+
},
|
| 3960 |
+
{
|
| 3961 |
+
"epoch": 1.8,
|
| 3962 |
+
"learning_rate": 7.977123645599227e-06,
|
| 3963 |
+
"loss": 0.025,
|
| 3964 |
+
"step": 329000
|
| 3965 |
+
},
|
| 3966 |
+
{
|
| 3967 |
+
"epoch": 1.81,
|
| 3968 |
+
"learning_rate": 7.958851797036306e-06,
|
| 3969 |
+
"loss": 0.0259,
|
| 3970 |
+
"step": 329500
|
| 3971 |
+
},
|
| 3972 |
+
{
|
| 3973 |
+
"epoch": 1.81,
|
| 3974 |
+
"learning_rate": 7.940579948473388e-06,
|
| 3975 |
+
"loss": 0.0327,
|
| 3976 |
+
"step": 330000
|
| 3977 |
+
},
|
| 3978 |
+
{
|
| 3979 |
+
"epoch": 1.81,
|
| 3980 |
+
"learning_rate": 7.922308099910469e-06,
|
| 3981 |
+
"loss": 0.0227,
|
| 3982 |
+
"step": 330500
|
| 3983 |
+
},
|
| 3984 |
+
{
|
| 3985 |
+
"epoch": 1.81,
|
| 3986 |
+
"learning_rate": 7.90403625134755e-06,
|
| 3987 |
+
"loss": 0.0393,
|
| 3988 |
+
"step": 331000
|
| 3989 |
+
},
|
| 3990 |
+
{
|
| 3991 |
+
"epoch": 1.82,
|
| 3992 |
+
"learning_rate": 7.88576440278463e-06,
|
| 3993 |
+
"loss": 0.0336,
|
| 3994 |
+
"step": 331500
|
| 3995 |
+
},
|
| 3996 |
+
{
|
| 3997 |
+
"epoch": 1.82,
|
| 3998 |
+
"learning_rate": 7.867492554221711e-06,
|
| 3999 |
+
"loss": 0.026,
|
| 4000 |
+
"step": 332000
|
| 4001 |
+
},
|
| 4002 |
+
{
|
| 4003 |
+
"epoch": 1.82,
|
| 4004 |
+
"learning_rate": 7.849220705658792e-06,
|
| 4005 |
+
"loss": 0.0341,
|
| 4006 |
+
"step": 332500
|
| 4007 |
+
},
|
| 4008 |
+
{
|
| 4009 |
+
"epoch": 1.83,
|
| 4010 |
+
"learning_rate": 7.830948857095873e-06,
|
| 4011 |
+
"loss": 0.0322,
|
| 4012 |
+
"step": 333000
|
| 4013 |
+
},
|
| 4014 |
+
{
|
| 4015 |
+
"epoch": 1.83,
|
| 4016 |
+
"learning_rate": 7.812677008532953e-06,
|
| 4017 |
+
"loss": 0.0307,
|
| 4018 |
+
"step": 333500
|
| 4019 |
+
},
|
| 4020 |
+
{
|
| 4021 |
+
"epoch": 1.83,
|
| 4022 |
+
"learning_rate": 7.794405159970034e-06,
|
| 4023 |
+
"loss": 0.0321,
|
| 4024 |
+
"step": 334000
|
| 4025 |
+
},
|
| 4026 |
+
{
|
| 4027 |
+
"epoch": 1.83,
|
| 4028 |
+
"learning_rate": 7.776133311407116e-06,
|
| 4029 |
+
"loss": 0.0319,
|
| 4030 |
+
"step": 334500
|
| 4031 |
+
},
|
| 4032 |
+
{
|
| 4033 |
+
"epoch": 1.84,
|
| 4034 |
+
"learning_rate": 7.757861462844195e-06,
|
| 4035 |
+
"loss": 0.0265,
|
| 4036 |
+
"step": 335000
|
| 4037 |
+
},
|
| 4038 |
+
{
|
| 4039 |
+
"epoch": 1.84,
|
| 4040 |
+
"learning_rate": 7.739589614281278e-06,
|
| 4041 |
+
"loss": 0.0357,
|
| 4042 |
+
"step": 335500
|
| 4043 |
+
},
|
| 4044 |
+
{
|
| 4045 |
+
"epoch": 1.84,
|
| 4046 |
+
"learning_rate": 7.721317765718359e-06,
|
| 4047 |
+
"loss": 0.0284,
|
| 4048 |
+
"step": 336000
|
| 4049 |
+
},
|
| 4050 |
+
{
|
| 4051 |
+
"epoch": 1.84,
|
| 4052 |
+
"learning_rate": 7.70304591715544e-06,
|
| 4053 |
+
"loss": 0.0337,
|
| 4054 |
+
"step": 336500
|
| 4055 |
+
},
|
| 4056 |
+
{
|
| 4057 |
+
"epoch": 1.85,
|
| 4058 |
+
"learning_rate": 7.68477406859252e-06,
|
| 4059 |
+
"loss": 0.0263,
|
| 4060 |
+
"step": 337000
|
| 4061 |
+
},
|
| 4062 |
+
{
|
| 4063 |
+
"epoch": 1.85,
|
| 4064 |
+
"learning_rate": 7.6665022200296e-06,
|
| 4065 |
+
"loss": 0.0292,
|
| 4066 |
+
"step": 337500
|
| 4067 |
+
},
|
| 4068 |
+
{
|
| 4069 |
+
"epoch": 1.85,
|
| 4070 |
+
"learning_rate": 7.648230371466681e-06,
|
| 4071 |
+
"loss": 0.0239,
|
| 4072 |
+
"step": 338000
|
| 4073 |
+
},
|
| 4074 |
+
{
|
| 4075 |
+
"epoch": 1.86,
|
| 4076 |
+
"learning_rate": 7.629958522903764e-06,
|
| 4077 |
+
"loss": 0.0426,
|
| 4078 |
+
"step": 338500
|
| 4079 |
+
},
|
| 4080 |
+
{
|
| 4081 |
+
"epoch": 1.86,
|
| 4082 |
+
"learning_rate": 7.6116866743408436e-06,
|
| 4083 |
+
"loss": 0.0303,
|
| 4084 |
+
"step": 339000
|
| 4085 |
+
},
|
| 4086 |
+
{
|
| 4087 |
+
"epoch": 1.86,
|
| 4088 |
+
"learning_rate": 7.593414825777925e-06,
|
| 4089 |
+
"loss": 0.0347,
|
| 4090 |
+
"step": 339500
|
| 4091 |
+
},
|
| 4092 |
+
{
|
| 4093 |
+
"epoch": 1.86,
|
| 4094 |
+
"learning_rate": 7.575142977215005e-06,
|
| 4095 |
+
"loss": 0.037,
|
| 4096 |
+
"step": 340000
|
| 4097 |
+
},
|
| 4098 |
+
{
|
| 4099 |
+
"epoch": 1.87,
|
| 4100 |
+
"learning_rate": 7.5568711286520865e-06,
|
| 4101 |
+
"loss": 0.0295,
|
| 4102 |
+
"step": 340500
|
| 4103 |
+
},
|
| 4104 |
+
{
|
| 4105 |
+
"epoch": 1.87,
|
| 4106 |
+
"learning_rate": 7.538599280089167e-06,
|
| 4107 |
+
"loss": 0.0261,
|
| 4108 |
+
"step": 341000
|
| 4109 |
+
},
|
| 4110 |
+
{
|
| 4111 |
+
"epoch": 1.87,
|
| 4112 |
+
"learning_rate": 7.520327431526249e-06,
|
| 4113 |
+
"loss": 0.0329,
|
| 4114 |
+
"step": 341500
|
| 4115 |
+
},
|
| 4116 |
+
{
|
| 4117 |
+
"epoch": 1.87,
|
| 4118 |
+
"learning_rate": 7.502055582963329e-06,
|
| 4119 |
+
"loss": 0.0392,
|
| 4120 |
+
"step": 342000
|
| 4121 |
+
},
|
| 4122 |
+
{
|
| 4123 |
+
"epoch": 1.88,
|
| 4124 |
+
"learning_rate": 7.483783734400409e-06,
|
| 4125 |
+
"loss": 0.0299,
|
| 4126 |
+
"step": 342500
|
| 4127 |
+
},
|
| 4128 |
+
{
|
| 4129 |
+
"epoch": 1.88,
|
| 4130 |
+
"learning_rate": 7.465511885837491e-06,
|
| 4131 |
+
"loss": 0.0313,
|
| 4132 |
+
"step": 343000
|
| 4133 |
+
},
|
| 4134 |
+
{
|
| 4135 |
+
"epoch": 1.88,
|
| 4136 |
+
"learning_rate": 7.4472400372745716e-06,
|
| 4137 |
+
"loss": 0.0312,
|
| 4138 |
+
"step": 343500
|
| 4139 |
+
},
|
| 4140 |
+
{
|
| 4141 |
+
"epoch": 1.89,
|
| 4142 |
+
"learning_rate": 7.428968188711653e-06,
|
| 4143 |
+
"loss": 0.0329,
|
| 4144 |
+
"step": 344000
|
| 4145 |
+
},
|
| 4146 |
+
{
|
| 4147 |
+
"epoch": 1.89,
|
| 4148 |
+
"learning_rate": 7.410696340148733e-06,
|
| 4149 |
+
"loss": 0.0227,
|
| 4150 |
+
"step": 344500
|
| 4151 |
+
},
|
| 4152 |
+
{
|
| 4153 |
+
"epoch": 1.89,
|
| 4154 |
+
"learning_rate": 7.3924244915858145e-06,
|
| 4155 |
+
"loss": 0.0264,
|
| 4156 |
+
"step": 345000
|
| 4157 |
+
},
|
| 4158 |
+
{
|
| 4159 |
+
"epoch": 1.89,
|
| 4160 |
+
"learning_rate": 7.374152643022895e-06,
|
| 4161 |
+
"loss": 0.0277,
|
| 4162 |
+
"step": 345500
|
| 4163 |
+
},
|
| 4164 |
+
{
|
| 4165 |
+
"epoch": 1.9,
|
| 4166 |
+
"learning_rate": 7.355880794459977e-06,
|
| 4167 |
+
"loss": 0.0301,
|
| 4168 |
+
"step": 346000
|
| 4169 |
+
},
|
| 4170 |
+
{
|
| 4171 |
+
"epoch": 1.9,
|
| 4172 |
+
"learning_rate": 7.337608945897057e-06,
|
| 4173 |
+
"loss": 0.035,
|
| 4174 |
+
"step": 346500
|
| 4175 |
+
},
|
| 4176 |
+
{
|
| 4177 |
+
"epoch": 1.9,
|
| 4178 |
+
"learning_rate": 7.319337097334138e-06,
|
| 4179 |
+
"loss": 0.0374,
|
| 4180 |
+
"step": 347000
|
| 4181 |
+
},
|
| 4182 |
+
{
|
| 4183 |
+
"epoch": 1.9,
|
| 4184 |
+
"learning_rate": 7.301065248771219e-06,
|
| 4185 |
+
"loss": 0.0193,
|
| 4186 |
+
"step": 347500
|
| 4187 |
+
},
|
| 4188 |
+
{
|
| 4189 |
+
"epoch": 1.91,
|
| 4190 |
+
"learning_rate": 7.2827934002083e-06,
|
| 4191 |
+
"loss": 0.0264,
|
| 4192 |
+
"step": 348000
|
| 4193 |
+
},
|
| 4194 |
+
{
|
| 4195 |
+
"epoch": 1.91,
|
| 4196 |
+
"learning_rate": 7.26452155164538e-06,
|
| 4197 |
+
"loss": 0.034,
|
| 4198 |
+
"step": 348500
|
| 4199 |
+
},
|
| 4200 |
+
{
|
| 4201 |
+
"epoch": 1.91,
|
| 4202 |
+
"learning_rate": 7.246249703082462e-06,
|
| 4203 |
+
"loss": 0.0288,
|
| 4204 |
+
"step": 349000
|
| 4205 |
+
},
|
| 4206 |
+
{
|
| 4207 |
+
"epoch": 1.92,
|
| 4208 |
+
"learning_rate": 7.2279778545195425e-06,
|
| 4209 |
+
"loss": 0.0323,
|
| 4210 |
+
"step": 349500
|
| 4211 |
+
},
|
| 4212 |
+
{
|
| 4213 |
+
"epoch": 1.92,
|
| 4214 |
+
"learning_rate": 7.209706005956624e-06,
|
| 4215 |
+
"loss": 0.0279,
|
| 4216 |
+
"step": 350000
|
| 4217 |
+
},
|
| 4218 |
+
{
|
| 4219 |
+
"epoch": 1.92,
|
| 4220 |
+
"learning_rate": 7.191434157393704e-06,
|
| 4221 |
+
"loss": 0.0285,
|
| 4222 |
+
"step": 350500
|
| 4223 |
+
},
|
| 4224 |
+
{
|
| 4225 |
+
"epoch": 1.92,
|
| 4226 |
+
"learning_rate": 7.173162308830785e-06,
|
| 4227 |
+
"loss": 0.0383,
|
| 4228 |
+
"step": 351000
|
| 4229 |
+
},
|
| 4230 |
+
{
|
| 4231 |
+
"epoch": 1.93,
|
| 4232 |
+
"learning_rate": 7.154890460267866e-06,
|
| 4233 |
+
"loss": 0.033,
|
| 4234 |
+
"step": 351500
|
| 4235 |
+
},
|
| 4236 |
+
{
|
| 4237 |
+
"epoch": 1.93,
|
| 4238 |
+
"learning_rate": 7.136618611704946e-06,
|
| 4239 |
+
"loss": 0.0347,
|
| 4240 |
+
"step": 352000
|
| 4241 |
+
},
|
| 4242 |
+
{
|
| 4243 |
+
"epoch": 1.93,
|
| 4244 |
+
"learning_rate": 7.1183467631420275e-06,
|
| 4245 |
+
"loss": 0.0314,
|
| 4246 |
+
"step": 352500
|
| 4247 |
+
},
|
| 4248 |
+
{
|
| 4249 |
+
"epoch": 1.93,
|
| 4250 |
+
"learning_rate": 7.100074914579108e-06,
|
| 4251 |
+
"loss": 0.0306,
|
| 4252 |
+
"step": 353000
|
| 4253 |
+
},
|
| 4254 |
+
{
|
| 4255 |
+
"epoch": 1.94,
|
| 4256 |
+
"learning_rate": 7.08180306601619e-06,
|
| 4257 |
+
"loss": 0.0464,
|
| 4258 |
+
"step": 353500
|
| 4259 |
+
},
|
| 4260 |
+
{
|
| 4261 |
+
"epoch": 1.94,
|
| 4262 |
+
"learning_rate": 7.06353121745327e-06,
|
| 4263 |
+
"loss": 0.0314,
|
| 4264 |
+
"step": 354000
|
| 4265 |
+
},
|
| 4266 |
+
{
|
| 4267 |
+
"epoch": 1.94,
|
| 4268 |
+
"learning_rate": 7.045259368890351e-06,
|
| 4269 |
+
"loss": 0.0206,
|
| 4270 |
+
"step": 354500
|
| 4271 |
+
},
|
| 4272 |
+
{
|
| 4273 |
+
"epoch": 1.95,
|
| 4274 |
+
"learning_rate": 7.026987520327432e-06,
|
| 4275 |
+
"loss": 0.0206,
|
| 4276 |
+
"step": 355000
|
| 4277 |
+
},
|
| 4278 |
+
{
|
| 4279 |
+
"epoch": 1.95,
|
| 4280 |
+
"learning_rate": 7.008715671764513e-06,
|
| 4281 |
+
"loss": 0.0222,
|
| 4282 |
+
"step": 355500
|
| 4283 |
+
},
|
| 4284 |
+
{
|
| 4285 |
+
"epoch": 1.95,
|
| 4286 |
+
"learning_rate": 6.990443823201593e-06,
|
| 4287 |
+
"loss": 0.0324,
|
| 4288 |
+
"step": 356000
|
| 4289 |
+
},
|
| 4290 |
+
{
|
| 4291 |
+
"epoch": 1.95,
|
| 4292 |
+
"learning_rate": 6.972171974638675e-06,
|
| 4293 |
+
"loss": 0.0431,
|
| 4294 |
+
"step": 356500
|
| 4295 |
+
},
|
| 4296 |
+
{
|
| 4297 |
+
"epoch": 1.96,
|
| 4298 |
+
"learning_rate": 6.9539001260757555e-06,
|
| 4299 |
+
"loss": 0.0363,
|
| 4300 |
+
"step": 357000
|
| 4301 |
+
},
|
| 4302 |
+
{
|
| 4303 |
+
"epoch": 1.96,
|
| 4304 |
+
"learning_rate": 6.935628277512837e-06,
|
| 4305 |
+
"loss": 0.0307,
|
| 4306 |
+
"step": 357500
|
| 4307 |
+
},
|
| 4308 |
+
{
|
| 4309 |
+
"epoch": 1.96,
|
| 4310 |
+
"learning_rate": 6.917356428949917e-06,
|
| 4311 |
+
"loss": 0.0223,
|
| 4312 |
+
"step": 358000
|
| 4313 |
+
},
|
| 4314 |
+
{
|
| 4315 |
+
"epoch": 1.97,
|
| 4316 |
+
"learning_rate": 6.8990845803869985e-06,
|
| 4317 |
+
"loss": 0.0276,
|
| 4318 |
+
"step": 358500
|
| 4319 |
+
},
|
| 4320 |
+
{
|
| 4321 |
+
"epoch": 1.97,
|
| 4322 |
+
"learning_rate": 6.880812731824079e-06,
|
| 4323 |
+
"loss": 0.0283,
|
| 4324 |
+
"step": 359000
|
| 4325 |
+
},
|
| 4326 |
+
{
|
| 4327 |
+
"epoch": 1.97,
|
| 4328 |
+
"learning_rate": 6.862540883261161e-06,
|
| 4329 |
+
"loss": 0.0323,
|
| 4330 |
+
"step": 359500
|
| 4331 |
+
},
|
| 4332 |
+
{
|
| 4333 |
+
"epoch": 1.97,
|
| 4334 |
+
"learning_rate": 6.844269034698241e-06,
|
| 4335 |
+
"loss": 0.0323,
|
| 4336 |
+
"step": 360000
|
| 4337 |
+
},
|
| 4338 |
+
{
|
| 4339 |
+
"epoch": 1.98,
|
| 4340 |
+
"learning_rate": 6.825997186135321e-06,
|
| 4341 |
+
"loss": 0.0334,
|
| 4342 |
+
"step": 360500
|
| 4343 |
+
},
|
| 4344 |
+
{
|
| 4345 |
+
"epoch": 1.98,
|
| 4346 |
+
"learning_rate": 6.807725337572403e-06,
|
| 4347 |
+
"loss": 0.0325,
|
| 4348 |
+
"step": 361000
|
| 4349 |
+
},
|
| 4350 |
+
{
|
| 4351 |
+
"epoch": 1.98,
|
| 4352 |
+
"learning_rate": 6.7894534890094835e-06,
|
| 4353 |
+
"loss": 0.0333,
|
| 4354 |
+
"step": 361500
|
| 4355 |
+
},
|
| 4356 |
+
{
|
| 4357 |
+
"epoch": 1.98,
|
| 4358 |
+
"learning_rate": 6.771181640446565e-06,
|
| 4359 |
+
"loss": 0.0287,
|
| 4360 |
+
"step": 362000
|
| 4361 |
+
},
|
| 4362 |
+
{
|
| 4363 |
+
"epoch": 1.99,
|
| 4364 |
+
"learning_rate": 6.752909791883645e-06,
|
| 4365 |
+
"loss": 0.0192,
|
| 4366 |
+
"step": 362500
|
| 4367 |
+
},
|
| 4368 |
+
{
|
| 4369 |
+
"epoch": 1.99,
|
| 4370 |
+
"learning_rate": 6.7346379433207264e-06,
|
| 4371 |
+
"loss": 0.03,
|
| 4372 |
+
"step": 363000
|
| 4373 |
+
},
|
| 4374 |
+
{
|
| 4375 |
+
"epoch": 1.99,
|
| 4376 |
+
"learning_rate": 6.716366094757807e-06,
|
| 4377 |
+
"loss": 0.0238,
|
| 4378 |
+
"step": 363500
|
| 4379 |
+
},
|
| 4380 |
+
{
|
| 4381 |
+
"epoch": 2.0,
|
| 4382 |
+
"learning_rate": 6.698094246194889e-06,
|
| 4383 |
+
"loss": 0.0262,
|
| 4384 |
+
"step": 364000
|
| 4385 |
+
},
|
| 4386 |
+
{
|
| 4387 |
+
"epoch": 2.0,
|
| 4388 |
+
"learning_rate": 6.6798223976319685e-06,
|
| 4389 |
+
"loss": 0.0262,
|
| 4390 |
+
"step": 364500
|
| 4391 |
+
},
|
| 4392 |
+
{
|
| 4393 |
+
"epoch": 2.0,
|
| 4394 |
+
"eval_loss": 0.14961867034435272,
|
| 4395 |
+
"eval_runtime": 1674.7145,
|
| 4396 |
+
"eval_samples_per_second": 138.344,
|
| 4397 |
+
"eval_steps_per_second": 13.835,
|
| 4398 |
+
"step": 364860
|
| 4399 |
+
}
|
| 4400 |
+
],
|
| 4401 |
+
"max_steps": 547290,
|
| 4402 |
+
"num_train_epochs": 3,
|
| 4403 |
+
"total_flos": 7.150255099702272e+17,
|
| 4404 |
+
"trial_name": null,
|
| 4405 |
+
"trial_params": null
|
| 4406 |
+
}
|