Upload trainer_state.json
Browse files- trainer_state.json +3161 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,3161 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 0.9983917797277898,
|
| 5 |
+
"global_step": 92500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 4.9730164383857356e-05,
|
| 13 |
+
"loss": 2.1298,
|
| 14 |
+
"step": 500
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.01,
|
| 18 |
+
"eval_bleu": 6.9284,
|
| 19 |
+
"eval_gen_len": 23.9635,
|
| 20 |
+
"eval_loss": 4.2180352210998535,
|
| 21 |
+
"eval_meteor": 0.2087,
|
| 22 |
+
"eval_runtime": 55.8494,
|
| 23 |
+
"eval_samples_per_second": 9.311,
|
| 24 |
+
"eval_steps_per_second": 1.164,
|
| 25 |
+
"step": 500
|
| 26 |
+
},
|
| 27 |
+
{
|
| 28 |
+
"epoch": 0.01,
|
| 29 |
+
"learning_rate": 4.946032876771471e-05,
|
| 30 |
+
"loss": 2.0441,
|
| 31 |
+
"step": 1000
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.01,
|
| 35 |
+
"eval_bleu": 6.89,
|
| 36 |
+
"eval_gen_len": 35.1442,
|
| 37 |
+
"eval_loss": 4.191596508026123,
|
| 38 |
+
"eval_meteor": 0.2133,
|
| 39 |
+
"eval_runtime": 113.3213,
|
| 40 |
+
"eval_samples_per_second": 4.589,
|
| 41 |
+
"eval_steps_per_second": 0.574,
|
| 42 |
+
"step": 1000
|
| 43 |
+
},
|
| 44 |
+
{
|
| 45 |
+
"epoch": 0.02,
|
| 46 |
+
"learning_rate": 4.9190493151572064e-05,
|
| 47 |
+
"loss": 2.0547,
|
| 48 |
+
"step": 1500
|
| 49 |
+
},
|
| 50 |
+
{
|
| 51 |
+
"epoch": 0.02,
|
| 52 |
+
"eval_bleu": 6.5842,
|
| 53 |
+
"eval_gen_len": 26.2904,
|
| 54 |
+
"eval_loss": 4.191189289093018,
|
| 55 |
+
"eval_meteor": 0.214,
|
| 56 |
+
"eval_runtime": 68.7752,
|
| 57 |
+
"eval_samples_per_second": 7.561,
|
| 58 |
+
"eval_steps_per_second": 0.945,
|
| 59 |
+
"step": 1500
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.02,
|
| 63 |
+
"learning_rate": 4.892065753542942e-05,
|
| 64 |
+
"loss": 1.9988,
|
| 65 |
+
"step": 2000
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.02,
|
| 69 |
+
"eval_bleu": 6.3595,
|
| 70 |
+
"eval_gen_len": 43.4173,
|
| 71 |
+
"eval_loss": 4.144649982452393,
|
| 72 |
+
"eval_meteor": 0.206,
|
| 73 |
+
"eval_runtime": 168.802,
|
| 74 |
+
"eval_samples_per_second": 3.081,
|
| 75 |
+
"eval_steps_per_second": 0.385,
|
| 76 |
+
"step": 2000
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.03,
|
| 80 |
+
"learning_rate": 4.865082191928677e-05,
|
| 81 |
+
"loss": 1.9714,
|
| 82 |
+
"step": 2500
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.03,
|
| 86 |
+
"eval_bleu": 6.0327,
|
| 87 |
+
"eval_gen_len": 35.4712,
|
| 88 |
+
"eval_loss": 4.147974491119385,
|
| 89 |
+
"eval_meteor": 0.2087,
|
| 90 |
+
"eval_runtime": 124.6806,
|
| 91 |
+
"eval_samples_per_second": 4.171,
|
| 92 |
+
"eval_steps_per_second": 0.521,
|
| 93 |
+
"step": 2500
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.03,
|
| 97 |
+
"learning_rate": 4.8380986303144125e-05,
|
| 98 |
+
"loss": 2.0021,
|
| 99 |
+
"step": 3000
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.03,
|
| 103 |
+
"eval_bleu": 6.626,
|
| 104 |
+
"eval_gen_len": 28.4385,
|
| 105 |
+
"eval_loss": 4.138648509979248,
|
| 106 |
+
"eval_meteor": 0.2004,
|
| 107 |
+
"eval_runtime": 79.9371,
|
| 108 |
+
"eval_samples_per_second": 6.505,
|
| 109 |
+
"eval_steps_per_second": 0.813,
|
| 110 |
+
"step": 3000
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.04,
|
| 114 |
+
"learning_rate": 4.8111150687001485e-05,
|
| 115 |
+
"loss": 1.9541,
|
| 116 |
+
"step": 3500
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.04,
|
| 120 |
+
"eval_bleu": 7.3794,
|
| 121 |
+
"eval_gen_len": 32.2192,
|
| 122 |
+
"eval_loss": 4.102533340454102,
|
| 123 |
+
"eval_meteor": 0.2092,
|
| 124 |
+
"eval_runtime": 104.2846,
|
| 125 |
+
"eval_samples_per_second": 4.986,
|
| 126 |
+
"eval_steps_per_second": 0.623,
|
| 127 |
+
"step": 3500
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.04,
|
| 131 |
+
"learning_rate": 4.784131507085883e-05,
|
| 132 |
+
"loss": 1.9843,
|
| 133 |
+
"step": 4000
|
| 134 |
+
},
|
| 135 |
+
{
|
| 136 |
+
"epoch": 0.04,
|
| 137 |
+
"eval_bleu": 6.8478,
|
| 138 |
+
"eval_gen_len": 28.3692,
|
| 139 |
+
"eval_loss": 4.0901265144348145,
|
| 140 |
+
"eval_meteor": 0.2026,
|
| 141 |
+
"eval_runtime": 84.5232,
|
| 142 |
+
"eval_samples_per_second": 6.152,
|
| 143 |
+
"eval_steps_per_second": 0.769,
|
| 144 |
+
"step": 4000
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.05,
|
| 148 |
+
"learning_rate": 4.7571479454716186e-05,
|
| 149 |
+
"loss": 2.0089,
|
| 150 |
+
"step": 4500
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.05,
|
| 154 |
+
"eval_bleu": 7.3969,
|
| 155 |
+
"eval_gen_len": 32.6115,
|
| 156 |
+
"eval_loss": 4.115621089935303,
|
| 157 |
+
"eval_meteor": 0.2061,
|
| 158 |
+
"eval_runtime": 108.942,
|
| 159 |
+
"eval_samples_per_second": 4.773,
|
| 160 |
+
"eval_steps_per_second": 0.597,
|
| 161 |
+
"step": 4500
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.05,
|
| 165 |
+
"learning_rate": 4.730164383857355e-05,
|
| 166 |
+
"loss": 1.8734,
|
| 167 |
+
"step": 5000
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.05,
|
| 171 |
+
"eval_bleu": 6.4957,
|
| 172 |
+
"eval_gen_len": 53.5423,
|
| 173 |
+
"eval_loss": 4.048145771026611,
|
| 174 |
+
"eval_meteor": 0.2047,
|
| 175 |
+
"eval_runtime": 248.2193,
|
| 176 |
+
"eval_samples_per_second": 2.095,
|
| 177 |
+
"eval_steps_per_second": 0.262,
|
| 178 |
+
"step": 5000
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.06,
|
| 182 |
+
"learning_rate": 4.7031808222430894e-05,
|
| 183 |
+
"loss": 1.959,
|
| 184 |
+
"step": 5500
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.06,
|
| 188 |
+
"eval_bleu": 6.7135,
|
| 189 |
+
"eval_gen_len": 28.0327,
|
| 190 |
+
"eval_loss": 4.0486321449279785,
|
| 191 |
+
"eval_meteor": 0.1982,
|
| 192 |
+
"eval_runtime": 78.1961,
|
| 193 |
+
"eval_samples_per_second": 6.65,
|
| 194 |
+
"eval_steps_per_second": 0.831,
|
| 195 |
+
"step": 5500
|
| 196 |
+
},
|
| 197 |
+
{
|
| 198 |
+
"epoch": 0.06,
|
| 199 |
+
"learning_rate": 4.6761972606288254e-05,
|
| 200 |
+
"loss": 1.9442,
|
| 201 |
+
"step": 6000
|
| 202 |
+
},
|
| 203 |
+
{
|
| 204 |
+
"epoch": 0.06,
|
| 205 |
+
"eval_bleu": 7.8174,
|
| 206 |
+
"eval_gen_len": 30.275,
|
| 207 |
+
"eval_loss": 4.008225440979004,
|
| 208 |
+
"eval_meteor": 0.2121,
|
| 209 |
+
"eval_runtime": 88.4227,
|
| 210 |
+
"eval_samples_per_second": 5.881,
|
| 211 |
+
"eval_steps_per_second": 0.735,
|
| 212 |
+
"step": 6000
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.07,
|
| 216 |
+
"learning_rate": 4.649213699014561e-05,
|
| 217 |
+
"loss": 1.8855,
|
| 218 |
+
"step": 6500
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.07,
|
| 222 |
+
"eval_bleu": 7.0941,
|
| 223 |
+
"eval_gen_len": 30.8077,
|
| 224 |
+
"eval_loss": 4.006565570831299,
|
| 225 |
+
"eval_meteor": 0.2117,
|
| 226 |
+
"eval_runtime": 90.6701,
|
| 227 |
+
"eval_samples_per_second": 5.735,
|
| 228 |
+
"eval_steps_per_second": 0.717,
|
| 229 |
+
"step": 6500
|
| 230 |
+
},
|
| 231 |
+
{
|
| 232 |
+
"epoch": 0.08,
|
| 233 |
+
"learning_rate": 4.6222301374002955e-05,
|
| 234 |
+
"loss": 1.873,
|
| 235 |
+
"step": 7000
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.08,
|
| 239 |
+
"eval_bleu": 7.0348,
|
| 240 |
+
"eval_gen_len": 34.7673,
|
| 241 |
+
"eval_loss": 4.008674621582031,
|
| 242 |
+
"eval_meteor": 0.2117,
|
| 243 |
+
"eval_runtime": 140.0997,
|
| 244 |
+
"eval_samples_per_second": 3.712,
|
| 245 |
+
"eval_steps_per_second": 0.464,
|
| 246 |
+
"step": 7000
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.08,
|
| 250 |
+
"learning_rate": 4.5952465757860316e-05,
|
| 251 |
+
"loss": 1.9126,
|
| 252 |
+
"step": 7500
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.08,
|
| 256 |
+
"eval_bleu": 7.2617,
|
| 257 |
+
"eval_gen_len": 31.5173,
|
| 258 |
+
"eval_loss": 3.955263614654541,
|
| 259 |
+
"eval_meteor": 0.2188,
|
| 260 |
+
"eval_runtime": 135.15,
|
| 261 |
+
"eval_samples_per_second": 3.848,
|
| 262 |
+
"eval_steps_per_second": 0.481,
|
| 263 |
+
"step": 7500
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.09,
|
| 267 |
+
"learning_rate": 4.568263014171767e-05,
|
| 268 |
+
"loss": 1.8181,
|
| 269 |
+
"step": 8000
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.09,
|
| 273 |
+
"eval_bleu": 7.0348,
|
| 274 |
+
"eval_gen_len": 29.8442,
|
| 275 |
+
"eval_loss": 3.9835426807403564,
|
| 276 |
+
"eval_meteor": 0.211,
|
| 277 |
+
"eval_runtime": 112.2816,
|
| 278 |
+
"eval_samples_per_second": 4.631,
|
| 279 |
+
"eval_steps_per_second": 0.579,
|
| 280 |
+
"step": 8000
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.09,
|
| 284 |
+
"learning_rate": 4.541279452557502e-05,
|
| 285 |
+
"loss": 1.8553,
|
| 286 |
+
"step": 8500
|
| 287 |
+
},
|
| 288 |
+
{
|
| 289 |
+
"epoch": 0.09,
|
| 290 |
+
"eval_bleu": 6.9695,
|
| 291 |
+
"eval_gen_len": 31.0673,
|
| 292 |
+
"eval_loss": 3.934365749359131,
|
| 293 |
+
"eval_meteor": 0.2167,
|
| 294 |
+
"eval_runtime": 126.5544,
|
| 295 |
+
"eval_samples_per_second": 4.109,
|
| 296 |
+
"eval_steps_per_second": 0.514,
|
| 297 |
+
"step": 8500
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.1,
|
| 301 |
+
"learning_rate": 4.514295890943238e-05,
|
| 302 |
+
"loss": 1.8856,
|
| 303 |
+
"step": 9000
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.1,
|
| 307 |
+
"eval_bleu": 7.8161,
|
| 308 |
+
"eval_gen_len": 25.7885,
|
| 309 |
+
"eval_loss": 3.9472157955169678,
|
| 310 |
+
"eval_meteor": 0.2202,
|
| 311 |
+
"eval_runtime": 90.2192,
|
| 312 |
+
"eval_samples_per_second": 5.764,
|
| 313 |
+
"eval_steps_per_second": 0.72,
|
| 314 |
+
"step": 9000
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.1,
|
| 318 |
+
"learning_rate": 4.487312329328973e-05,
|
| 319 |
+
"loss": 1.8375,
|
| 320 |
+
"step": 9500
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.1,
|
| 324 |
+
"eval_bleu": 7.3226,
|
| 325 |
+
"eval_gen_len": 31.6615,
|
| 326 |
+
"eval_loss": 3.9212045669555664,
|
| 327 |
+
"eval_meteor": 0.2131,
|
| 328 |
+
"eval_runtime": 122.1859,
|
| 329 |
+
"eval_samples_per_second": 4.256,
|
| 330 |
+
"eval_steps_per_second": 0.532,
|
| 331 |
+
"step": 9500
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.11,
|
| 335 |
+
"learning_rate": 4.4603287677147084e-05,
|
| 336 |
+
"loss": 1.8245,
|
| 337 |
+
"step": 10000
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.11,
|
| 341 |
+
"eval_bleu": 7.5927,
|
| 342 |
+
"eval_gen_len": 25.6827,
|
| 343 |
+
"eval_loss": 3.9012672901153564,
|
| 344 |
+
"eval_meteor": 0.2162,
|
| 345 |
+
"eval_runtime": 87.2113,
|
| 346 |
+
"eval_samples_per_second": 5.963,
|
| 347 |
+
"eval_steps_per_second": 0.745,
|
| 348 |
+
"step": 10000
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.11,
|
| 352 |
+
"learning_rate": 4.433345206100444e-05,
|
| 353 |
+
"loss": 1.7563,
|
| 354 |
+
"step": 10500
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.11,
|
| 358 |
+
"eval_bleu": 7.6493,
|
| 359 |
+
"eval_gen_len": 24.7904,
|
| 360 |
+
"eval_loss": 3.9165987968444824,
|
| 361 |
+
"eval_meteor": 0.2225,
|
| 362 |
+
"eval_runtime": 71.5438,
|
| 363 |
+
"eval_samples_per_second": 7.268,
|
| 364 |
+
"eval_steps_per_second": 0.909,
|
| 365 |
+
"step": 10500
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.12,
|
| 369 |
+
"learning_rate": 4.406361644486179e-05,
|
| 370 |
+
"loss": 1.7739,
|
| 371 |
+
"step": 11000
|
| 372 |
+
},
|
| 373 |
+
{
|
| 374 |
+
"epoch": 0.12,
|
| 375 |
+
"eval_bleu": 7.0986,
|
| 376 |
+
"eval_gen_len": 32.6269,
|
| 377 |
+
"eval_loss": 3.892530679702759,
|
| 378 |
+
"eval_meteor": 0.2162,
|
| 379 |
+
"eval_runtime": 135.1518,
|
| 380 |
+
"eval_samples_per_second": 3.848,
|
| 381 |
+
"eval_steps_per_second": 0.481,
|
| 382 |
+
"step": 11000
|
| 383 |
+
},
|
| 384 |
+
{
|
| 385 |
+
"epoch": 0.12,
|
| 386 |
+
"learning_rate": 4.3793780828719146e-05,
|
| 387 |
+
"loss": 1.8389,
|
| 388 |
+
"step": 11500
|
| 389 |
+
},
|
| 390 |
+
{
|
| 391 |
+
"epoch": 0.12,
|
| 392 |
+
"eval_bleu": 7.355,
|
| 393 |
+
"eval_gen_len": 29.1173,
|
| 394 |
+
"eval_loss": 3.891714572906494,
|
| 395 |
+
"eval_meteor": 0.222,
|
| 396 |
+
"eval_runtime": 103.9412,
|
| 397 |
+
"eval_samples_per_second": 5.003,
|
| 398 |
+
"eval_steps_per_second": 0.625,
|
| 399 |
+
"step": 11500
|
| 400 |
+
},
|
| 401 |
+
{
|
| 402 |
+
"epoch": 0.13,
|
| 403 |
+
"learning_rate": 4.35239452125765e-05,
|
| 404 |
+
"loss": 1.8359,
|
| 405 |
+
"step": 12000
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.13,
|
| 409 |
+
"eval_bleu": 7.7276,
|
| 410 |
+
"eval_gen_len": 26.5288,
|
| 411 |
+
"eval_loss": 3.8721096515655518,
|
| 412 |
+
"eval_meteor": 0.2224,
|
| 413 |
+
"eval_runtime": 92.6122,
|
| 414 |
+
"eval_samples_per_second": 5.615,
|
| 415 |
+
"eval_steps_per_second": 0.702,
|
| 416 |
+
"step": 12000
|
| 417 |
+
},
|
| 418 |
+
{
|
| 419 |
+
"epoch": 0.13,
|
| 420 |
+
"learning_rate": 4.325410959643385e-05,
|
| 421 |
+
"loss": 1.774,
|
| 422 |
+
"step": 12500
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 0.13,
|
| 426 |
+
"eval_bleu": 7.3227,
|
| 427 |
+
"eval_gen_len": 25.8019,
|
| 428 |
+
"eval_loss": 3.852199077606201,
|
| 429 |
+
"eval_meteor": 0.2256,
|
| 430 |
+
"eval_runtime": 80.7019,
|
| 431 |
+
"eval_samples_per_second": 6.443,
|
| 432 |
+
"eval_steps_per_second": 0.805,
|
| 433 |
+
"step": 12500
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.14,
|
| 437 |
+
"learning_rate": 4.298427398029121e-05,
|
| 438 |
+
"loss": 1.8502,
|
| 439 |
+
"step": 13000
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 0.14,
|
| 443 |
+
"eval_bleu": 8.0288,
|
| 444 |
+
"eval_gen_len": 25.7327,
|
| 445 |
+
"eval_loss": 3.838671922683716,
|
| 446 |
+
"eval_meteor": 0.2306,
|
| 447 |
+
"eval_runtime": 81.2754,
|
| 448 |
+
"eval_samples_per_second": 6.398,
|
| 449 |
+
"eval_steps_per_second": 0.8,
|
| 450 |
+
"step": 13000
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.15,
|
| 454 |
+
"learning_rate": 4.271443836414856e-05,
|
| 455 |
+
"loss": 1.7748,
|
| 456 |
+
"step": 13500
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 0.15,
|
| 460 |
+
"eval_bleu": 7.8972,
|
| 461 |
+
"eval_gen_len": 25.8135,
|
| 462 |
+
"eval_loss": 3.847898006439209,
|
| 463 |
+
"eval_meteor": 0.2289,
|
| 464 |
+
"eval_runtime": 77.3621,
|
| 465 |
+
"eval_samples_per_second": 6.722,
|
| 466 |
+
"eval_steps_per_second": 0.84,
|
| 467 |
+
"step": 13500
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.15,
|
| 471 |
+
"learning_rate": 4.244460274800592e-05,
|
| 472 |
+
"loss": 1.8212,
|
| 473 |
+
"step": 14000
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.15,
|
| 477 |
+
"eval_bleu": 7.617,
|
| 478 |
+
"eval_gen_len": 31.2462,
|
| 479 |
+
"eval_loss": 3.831773281097412,
|
| 480 |
+
"eval_meteor": 0.2249,
|
| 481 |
+
"eval_runtime": 133.705,
|
| 482 |
+
"eval_samples_per_second": 3.889,
|
| 483 |
+
"eval_steps_per_second": 0.486,
|
| 484 |
+
"step": 14000
|
| 485 |
+
},
|
| 486 |
+
{
|
| 487 |
+
"epoch": 0.16,
|
| 488 |
+
"learning_rate": 4.217476713186327e-05,
|
| 489 |
+
"loss": 1.7943,
|
| 490 |
+
"step": 14500
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"epoch": 0.16,
|
| 494 |
+
"eval_bleu": 6.6778,
|
| 495 |
+
"eval_gen_len": 27.7673,
|
| 496 |
+
"eval_loss": 3.8318052291870117,
|
| 497 |
+
"eval_meteor": 0.2197,
|
| 498 |
+
"eval_runtime": 95.8818,
|
| 499 |
+
"eval_samples_per_second": 5.423,
|
| 500 |
+
"eval_steps_per_second": 0.678,
|
| 501 |
+
"step": 14500
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.16,
|
| 505 |
+
"learning_rate": 4.190493151572062e-05,
|
| 506 |
+
"loss": 1.7869,
|
| 507 |
+
"step": 15000
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 0.16,
|
| 511 |
+
"eval_bleu": 7.5995,
|
| 512 |
+
"eval_gen_len": 25.6712,
|
| 513 |
+
"eval_loss": 3.793452262878418,
|
| 514 |
+
"eval_meteor": 0.2261,
|
| 515 |
+
"eval_runtime": 84.1529,
|
| 516 |
+
"eval_samples_per_second": 6.179,
|
| 517 |
+
"eval_steps_per_second": 0.772,
|
| 518 |
+
"step": 15000
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 0.17,
|
| 522 |
+
"learning_rate": 4.163509589957798e-05,
|
| 523 |
+
"loss": 1.7802,
|
| 524 |
+
"step": 15500
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.17,
|
| 528 |
+
"eval_bleu": 6.9484,
|
| 529 |
+
"eval_gen_len": 34.8865,
|
| 530 |
+
"eval_loss": 3.788137674331665,
|
| 531 |
+
"eval_meteor": 0.2214,
|
| 532 |
+
"eval_runtime": 159.4169,
|
| 533 |
+
"eval_samples_per_second": 3.262,
|
| 534 |
+
"eval_steps_per_second": 0.408,
|
| 535 |
+
"step": 15500
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 0.17,
|
| 539 |
+
"learning_rate": 4.136526028343533e-05,
|
| 540 |
+
"loss": 1.6783,
|
| 541 |
+
"step": 16000
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 0.17,
|
| 545 |
+
"eval_bleu": 7.1202,
|
| 546 |
+
"eval_gen_len": 26.2692,
|
| 547 |
+
"eval_loss": 3.7846884727478027,
|
| 548 |
+
"eval_meteor": 0.2171,
|
| 549 |
+
"eval_runtime": 88.5814,
|
| 550 |
+
"eval_samples_per_second": 5.87,
|
| 551 |
+
"eval_steps_per_second": 0.734,
|
| 552 |
+
"step": 16000
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.18,
|
| 556 |
+
"learning_rate": 4.109542466729269e-05,
|
| 557 |
+
"loss": 1.6773,
|
| 558 |
+
"step": 16500
|
| 559 |
+
},
|
| 560 |
+
{
|
| 561 |
+
"epoch": 0.18,
|
| 562 |
+
"eval_bleu": 8.093,
|
| 563 |
+
"eval_gen_len": 26.9346,
|
| 564 |
+
"eval_loss": 3.772686243057251,
|
| 565 |
+
"eval_meteor": 0.222,
|
| 566 |
+
"eval_runtime": 96.6464,
|
| 567 |
+
"eval_samples_per_second": 5.38,
|
| 568 |
+
"eval_steps_per_second": 0.673,
|
| 569 |
+
"step": 16500
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 0.18,
|
| 573 |
+
"learning_rate": 4.0825589051150044e-05,
|
| 574 |
+
"loss": 1.6435,
|
| 575 |
+
"step": 17000
|
| 576 |
+
},
|
| 577 |
+
{
|
| 578 |
+
"epoch": 0.18,
|
| 579 |
+
"eval_bleu": 7.5486,
|
| 580 |
+
"eval_gen_len": 26.6231,
|
| 581 |
+
"eval_loss": 3.785611152648926,
|
| 582 |
+
"eval_meteor": 0.2136,
|
| 583 |
+
"eval_runtime": 89.9499,
|
| 584 |
+
"eval_samples_per_second": 5.781,
|
| 585 |
+
"eval_steps_per_second": 0.723,
|
| 586 |
+
"step": 17000
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 0.19,
|
| 590 |
+
"learning_rate": 4.055575343500739e-05,
|
| 591 |
+
"loss": 1.6892,
|
| 592 |
+
"step": 17500
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 0.19,
|
| 596 |
+
"eval_bleu": 7.2576,
|
| 597 |
+
"eval_gen_len": 31.4712,
|
| 598 |
+
"eval_loss": 3.778566837310791,
|
| 599 |
+
"eval_meteor": 0.2088,
|
| 600 |
+
"eval_runtime": 124.5211,
|
| 601 |
+
"eval_samples_per_second": 4.176,
|
| 602 |
+
"eval_steps_per_second": 0.522,
|
| 603 |
+
"step": 17500
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.19,
|
| 607 |
+
"learning_rate": 4.028591781886475e-05,
|
| 608 |
+
"loss": 1.7355,
|
| 609 |
+
"step": 18000
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.19,
|
| 613 |
+
"eval_bleu": 7.8495,
|
| 614 |
+
"eval_gen_len": 27.7846,
|
| 615 |
+
"eval_loss": 3.739633798599243,
|
| 616 |
+
"eval_meteor": 0.2249,
|
| 617 |
+
"eval_runtime": 103.8056,
|
| 618 |
+
"eval_samples_per_second": 5.009,
|
| 619 |
+
"eval_steps_per_second": 0.626,
|
| 620 |
+
"step": 18000
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.2,
|
| 624 |
+
"learning_rate": 4.0016082202722105e-05,
|
| 625 |
+
"loss": 1.708,
|
| 626 |
+
"step": 18500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.2,
|
| 630 |
+
"eval_bleu": 8.0278,
|
| 631 |
+
"eval_gen_len": 23.0538,
|
| 632 |
+
"eval_loss": 3.7668354511260986,
|
| 633 |
+
"eval_meteor": 0.2207,
|
| 634 |
+
"eval_runtime": 70.0581,
|
| 635 |
+
"eval_samples_per_second": 7.422,
|
| 636 |
+
"eval_steps_per_second": 0.928,
|
| 637 |
+
"step": 18500
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 0.21,
|
| 641 |
+
"learning_rate": 3.974624658657945e-05,
|
| 642 |
+
"loss": 1.6596,
|
| 643 |
+
"step": 19000
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 0.21,
|
| 647 |
+
"eval_bleu": 7.5685,
|
| 648 |
+
"eval_gen_len": 28.1115,
|
| 649 |
+
"eval_loss": 3.7208786010742188,
|
| 650 |
+
"eval_meteor": 0.2287,
|
| 651 |
+
"eval_runtime": 102.7576,
|
| 652 |
+
"eval_samples_per_second": 5.06,
|
| 653 |
+
"eval_steps_per_second": 0.633,
|
| 654 |
+
"step": 19000
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.21,
|
| 658 |
+
"learning_rate": 3.947641097043681e-05,
|
| 659 |
+
"loss": 1.7048,
|
| 660 |
+
"step": 19500
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.21,
|
| 664 |
+
"eval_bleu": 8.1065,
|
| 665 |
+
"eval_gen_len": 23.6308,
|
| 666 |
+
"eval_loss": 3.7291922569274902,
|
| 667 |
+
"eval_meteor": 0.2308,
|
| 668 |
+
"eval_runtime": 60.5983,
|
| 669 |
+
"eval_samples_per_second": 8.581,
|
| 670 |
+
"eval_steps_per_second": 1.073,
|
| 671 |
+
"step": 19500
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.22,
|
| 675 |
+
"learning_rate": 3.9206575354294166e-05,
|
| 676 |
+
"loss": 1.671,
|
| 677 |
+
"step": 20000
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 0.22,
|
| 681 |
+
"eval_bleu": 8.2527,
|
| 682 |
+
"eval_gen_len": 28.6962,
|
| 683 |
+
"eval_loss": 3.6946775913238525,
|
| 684 |
+
"eval_meteor": 0.2352,
|
| 685 |
+
"eval_runtime": 109.8594,
|
| 686 |
+
"eval_samples_per_second": 4.733,
|
| 687 |
+
"eval_steps_per_second": 0.592,
|
| 688 |
+
"step": 20000
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 0.22,
|
| 692 |
+
"learning_rate": 3.893673973815152e-05,
|
| 693 |
+
"loss": 1.7021,
|
| 694 |
+
"step": 20500
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 0.22,
|
| 698 |
+
"eval_bleu": 7.9211,
|
| 699 |
+
"eval_gen_len": 25.9538,
|
| 700 |
+
"eval_loss": 3.711056709289551,
|
| 701 |
+
"eval_meteor": 0.2288,
|
| 702 |
+
"eval_runtime": 99.5819,
|
| 703 |
+
"eval_samples_per_second": 5.222,
|
| 704 |
+
"eval_steps_per_second": 0.653,
|
| 705 |
+
"step": 20500
|
| 706 |
+
},
|
| 707 |
+
{
|
| 708 |
+
"epoch": 0.23,
|
| 709 |
+
"learning_rate": 3.8666904122008874e-05,
|
| 710 |
+
"loss": 1.6431,
|
| 711 |
+
"step": 21000
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.23,
|
| 715 |
+
"eval_bleu": 8.1937,
|
| 716 |
+
"eval_gen_len": 27.2942,
|
| 717 |
+
"eval_loss": 3.7131733894348145,
|
| 718 |
+
"eval_meteor": 0.2303,
|
| 719 |
+
"eval_runtime": 84.3649,
|
| 720 |
+
"eval_samples_per_second": 6.164,
|
| 721 |
+
"eval_steps_per_second": 0.77,
|
| 722 |
+
"step": 21000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 0.23,
|
| 726 |
+
"learning_rate": 3.839706850586623e-05,
|
| 727 |
+
"loss": 1.7551,
|
| 728 |
+
"step": 21500
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 0.23,
|
| 732 |
+
"eval_bleu": 8.0326,
|
| 733 |
+
"eval_gen_len": 32.4615,
|
| 734 |
+
"eval_loss": 3.7119548320770264,
|
| 735 |
+
"eval_meteor": 0.2299,
|
| 736 |
+
"eval_runtime": 130.1926,
|
| 737 |
+
"eval_samples_per_second": 3.994,
|
| 738 |
+
"eval_steps_per_second": 0.499,
|
| 739 |
+
"step": 21500
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 0.24,
|
| 743 |
+
"learning_rate": 3.812723288972358e-05,
|
| 744 |
+
"loss": 1.672,
|
| 745 |
+
"step": 22000
|
| 746 |
+
},
|
| 747 |
+
{
|
| 748 |
+
"epoch": 0.24,
|
| 749 |
+
"eval_bleu": 8.1471,
|
| 750 |
+
"eval_gen_len": 26.1,
|
| 751 |
+
"eval_loss": 3.708371639251709,
|
| 752 |
+
"eval_meteor": 0.2274,
|
| 753 |
+
"eval_runtime": 101.2958,
|
| 754 |
+
"eval_samples_per_second": 5.133,
|
| 755 |
+
"eval_steps_per_second": 0.642,
|
| 756 |
+
"step": 22000
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 0.24,
|
| 760 |
+
"learning_rate": 3.7857397273580935e-05,
|
| 761 |
+
"loss": 1.6464,
|
| 762 |
+
"step": 22500
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 0.24,
|
| 766 |
+
"eval_bleu": 7.974,
|
| 767 |
+
"eval_gen_len": 26.8558,
|
| 768 |
+
"eval_loss": 3.677321195602417,
|
| 769 |
+
"eval_meteor": 0.2287,
|
| 770 |
+
"eval_runtime": 101.1748,
|
| 771 |
+
"eval_samples_per_second": 5.14,
|
| 772 |
+
"eval_steps_per_second": 0.642,
|
| 773 |
+
"step": 22500
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 0.25,
|
| 777 |
+
"learning_rate": 3.758756165743829e-05,
|
| 778 |
+
"loss": 1.6484,
|
| 779 |
+
"step": 23000
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.25,
|
| 783 |
+
"eval_bleu": 7.609,
|
| 784 |
+
"eval_gen_len": 28.8173,
|
| 785 |
+
"eval_loss": 3.6681010723114014,
|
| 786 |
+
"eval_meteor": 0.2303,
|
| 787 |
+
"eval_runtime": 110.9868,
|
| 788 |
+
"eval_samples_per_second": 4.685,
|
| 789 |
+
"eval_steps_per_second": 0.586,
|
| 790 |
+
"step": 23000
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"epoch": 0.25,
|
| 794 |
+
"learning_rate": 3.731772604129564e-05,
|
| 795 |
+
"loss": 1.6568,
|
| 796 |
+
"step": 23500
|
| 797 |
+
},
|
| 798 |
+
{
|
| 799 |
+
"epoch": 0.25,
|
| 800 |
+
"eval_bleu": 7.1469,
|
| 801 |
+
"eval_gen_len": 27.2288,
|
| 802 |
+
"eval_loss": 3.692307949066162,
|
| 803 |
+
"eval_meteor": 0.2255,
|
| 804 |
+
"eval_runtime": 94.7667,
|
| 805 |
+
"eval_samples_per_second": 5.487,
|
| 806 |
+
"eval_steps_per_second": 0.686,
|
| 807 |
+
"step": 23500
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.26,
|
| 811 |
+
"learning_rate": 3.7047890425152996e-05,
|
| 812 |
+
"loss": 1.6389,
|
| 813 |
+
"step": 24000
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 0.26,
|
| 817 |
+
"eval_bleu": 7.7818,
|
| 818 |
+
"eval_gen_len": 24.6692,
|
| 819 |
+
"eval_loss": 3.660385847091675,
|
| 820 |
+
"eval_meteor": 0.2249,
|
| 821 |
+
"eval_runtime": 81.5305,
|
| 822 |
+
"eval_samples_per_second": 6.378,
|
| 823 |
+
"eval_steps_per_second": 0.797,
|
| 824 |
+
"step": 24000
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 0.26,
|
| 828 |
+
"learning_rate": 3.677805480901036e-05,
|
| 829 |
+
"loss": 1.6639,
|
| 830 |
+
"step": 24500
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 0.26,
|
| 834 |
+
"eval_bleu": 8.0474,
|
| 835 |
+
"eval_gen_len": 23.9308,
|
| 836 |
+
"eval_loss": 3.659407377243042,
|
| 837 |
+
"eval_meteor": 0.2284,
|
| 838 |
+
"eval_runtime": 73.4805,
|
| 839 |
+
"eval_samples_per_second": 7.077,
|
| 840 |
+
"eval_steps_per_second": 0.885,
|
| 841 |
+
"step": 24500
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 0.27,
|
| 845 |
+
"learning_rate": 3.6508219192867704e-05,
|
| 846 |
+
"loss": 1.6965,
|
| 847 |
+
"step": 25000
|
| 848 |
+
},
|
| 849 |
+
{
|
| 850 |
+
"epoch": 0.27,
|
| 851 |
+
"eval_bleu": 7.8415,
|
| 852 |
+
"eval_gen_len": 29.1077,
|
| 853 |
+
"eval_loss": 3.663590431213379,
|
| 854 |
+
"eval_meteor": 0.2242,
|
| 855 |
+
"eval_runtime": 99.3013,
|
| 856 |
+
"eval_samples_per_second": 5.237,
|
| 857 |
+
"eval_steps_per_second": 0.655,
|
| 858 |
+
"step": 25000
|
| 859 |
+
},
|
| 860 |
+
{
|
| 861 |
+
"epoch": 0.28,
|
| 862 |
+
"learning_rate": 3.623838357672506e-05,
|
| 863 |
+
"loss": 1.6891,
|
| 864 |
+
"step": 25500
|
| 865 |
+
},
|
| 866 |
+
{
|
| 867 |
+
"epoch": 0.28,
|
| 868 |
+
"eval_bleu": 7.6988,
|
| 869 |
+
"eval_gen_len": 26.9077,
|
| 870 |
+
"eval_loss": 3.633793354034424,
|
| 871 |
+
"eval_meteor": 0.2278,
|
| 872 |
+
"eval_runtime": 87.8219,
|
| 873 |
+
"eval_samples_per_second": 5.921,
|
| 874 |
+
"eval_steps_per_second": 0.74,
|
| 875 |
+
"step": 25500
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 0.28,
|
| 879 |
+
"learning_rate": 3.596854796058242e-05,
|
| 880 |
+
"loss": 1.6518,
|
| 881 |
+
"step": 26000
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 0.28,
|
| 885 |
+
"eval_bleu": 8.0628,
|
| 886 |
+
"eval_gen_len": 24.2231,
|
| 887 |
+
"eval_loss": 3.6404640674591064,
|
| 888 |
+
"eval_meteor": 0.2272,
|
| 889 |
+
"eval_runtime": 79.4014,
|
| 890 |
+
"eval_samples_per_second": 6.549,
|
| 891 |
+
"eval_steps_per_second": 0.819,
|
| 892 |
+
"step": 26000
|
| 893 |
+
},
|
| 894 |
+
{
|
| 895 |
+
"epoch": 0.29,
|
| 896 |
+
"learning_rate": 3.5698712344439765e-05,
|
| 897 |
+
"loss": 1.5915,
|
| 898 |
+
"step": 26500
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.29,
|
| 902 |
+
"eval_bleu": 8.4956,
|
| 903 |
+
"eval_gen_len": 25.925,
|
| 904 |
+
"eval_loss": 3.6168606281280518,
|
| 905 |
+
"eval_meteor": 0.2327,
|
| 906 |
+
"eval_runtime": 84.1477,
|
| 907 |
+
"eval_samples_per_second": 6.18,
|
| 908 |
+
"eval_steps_per_second": 0.772,
|
| 909 |
+
"step": 26500
|
| 910 |
+
},
|
| 911 |
+
{
|
| 912 |
+
"epoch": 0.29,
|
| 913 |
+
"learning_rate": 3.5428876728297126e-05,
|
| 914 |
+
"loss": 1.6756,
|
| 915 |
+
"step": 27000
|
| 916 |
+
},
|
| 917 |
+
{
|
| 918 |
+
"epoch": 0.29,
|
| 919 |
+
"eval_bleu": 7.8645,
|
| 920 |
+
"eval_gen_len": 25.7115,
|
| 921 |
+
"eval_loss": 3.6374764442443848,
|
| 922 |
+
"eval_meteor": 0.2293,
|
| 923 |
+
"eval_runtime": 68.4092,
|
| 924 |
+
"eval_samples_per_second": 7.601,
|
| 925 |
+
"eval_steps_per_second": 0.95,
|
| 926 |
+
"step": 27000
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.3,
|
| 930 |
+
"learning_rate": 3.515904111215448e-05,
|
| 931 |
+
"loss": 1.6085,
|
| 932 |
+
"step": 27500
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 0.3,
|
| 936 |
+
"eval_bleu": 7.5502,
|
| 937 |
+
"eval_gen_len": 27.7,
|
| 938 |
+
"eval_loss": 3.6482863426208496,
|
| 939 |
+
"eval_meteor": 0.2193,
|
| 940 |
+
"eval_runtime": 107.6058,
|
| 941 |
+
"eval_samples_per_second": 4.832,
|
| 942 |
+
"eval_steps_per_second": 0.604,
|
| 943 |
+
"step": 27500
|
| 944 |
+
},
|
| 945 |
+
{
|
| 946 |
+
"epoch": 0.3,
|
| 947 |
+
"learning_rate": 3.4889205496011826e-05,
|
| 948 |
+
"loss": 1.5557,
|
| 949 |
+
"step": 28000
|
| 950 |
+
},
|
| 951 |
+
{
|
| 952 |
+
"epoch": 0.3,
|
| 953 |
+
"eval_bleu": 7.4827,
|
| 954 |
+
"eval_gen_len": 37.5385,
|
| 955 |
+
"eval_loss": 3.6158738136291504,
|
| 956 |
+
"eval_meteor": 0.2182,
|
| 957 |
+
"eval_runtime": 151.1373,
|
| 958 |
+
"eval_samples_per_second": 3.441,
|
| 959 |
+
"eval_steps_per_second": 0.43,
|
| 960 |
+
"step": 28000
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 0.31,
|
| 964 |
+
"learning_rate": 3.461936987986919e-05,
|
| 965 |
+
"loss": 1.5766,
|
| 966 |
+
"step": 28500
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 0.31,
|
| 970 |
+
"eval_bleu": 7.9417,
|
| 971 |
+
"eval_gen_len": 28.1673,
|
| 972 |
+
"eval_loss": 3.6332404613494873,
|
| 973 |
+
"eval_meteor": 0.2271,
|
| 974 |
+
"eval_runtime": 96.6869,
|
| 975 |
+
"eval_samples_per_second": 5.378,
|
| 976 |
+
"eval_steps_per_second": 0.672,
|
| 977 |
+
"step": 28500
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 0.31,
|
| 981 |
+
"learning_rate": 3.434953426372654e-05,
|
| 982 |
+
"loss": 1.6282,
|
| 983 |
+
"step": 29000
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 0.31,
|
| 987 |
+
"eval_bleu": 7.6697,
|
| 988 |
+
"eval_gen_len": 27.3019,
|
| 989 |
+
"eval_loss": 3.6174378395080566,
|
| 990 |
+
"eval_meteor": 0.2208,
|
| 991 |
+
"eval_runtime": 89.3468,
|
| 992 |
+
"eval_samples_per_second": 5.82,
|
| 993 |
+
"eval_steps_per_second": 0.728,
|
| 994 |
+
"step": 29000
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.32,
|
| 998 |
+
"learning_rate": 3.407969864758389e-05,
|
| 999 |
+
"loss": 1.5901,
|
| 1000 |
+
"step": 29500
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 0.32,
|
| 1004 |
+
"eval_bleu": 8.211,
|
| 1005 |
+
"eval_gen_len": 27.4885,
|
| 1006 |
+
"eval_loss": 3.6009418964385986,
|
| 1007 |
+
"eval_meteor": 0.2295,
|
| 1008 |
+
"eval_runtime": 86.4935,
|
| 1009 |
+
"eval_samples_per_second": 6.012,
|
| 1010 |
+
"eval_steps_per_second": 0.752,
|
| 1011 |
+
"step": 29500
|
| 1012 |
+
},
|
| 1013 |
+
{
|
| 1014 |
+
"epoch": 0.32,
|
| 1015 |
+
"learning_rate": 3.380986303144125e-05,
|
| 1016 |
+
"loss": 1.618,
|
| 1017 |
+
"step": 30000
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 0.32,
|
| 1021 |
+
"eval_bleu": 7.8473,
|
| 1022 |
+
"eval_gen_len": 26.3385,
|
| 1023 |
+
"eval_loss": 3.597890853881836,
|
| 1024 |
+
"eval_meteor": 0.2276,
|
| 1025 |
+
"eval_runtime": 67.1228,
|
| 1026 |
+
"eval_samples_per_second": 7.747,
|
| 1027 |
+
"eval_steps_per_second": 0.968,
|
| 1028 |
+
"step": 30000
|
| 1029 |
+
},
|
| 1030 |
+
{
|
| 1031 |
+
"epoch": 0.33,
|
| 1032 |
+
"learning_rate": 3.35400274152986e-05,
|
| 1033 |
+
"loss": 1.5292,
|
| 1034 |
+
"step": 30500
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 0.33,
|
| 1038 |
+
"eval_bleu": 7.6579,
|
| 1039 |
+
"eval_gen_len": 28.0481,
|
| 1040 |
+
"eval_loss": 3.578139543533325,
|
| 1041 |
+
"eval_meteor": 0.2348,
|
| 1042 |
+
"eval_runtime": 88.2928,
|
| 1043 |
+
"eval_samples_per_second": 5.889,
|
| 1044 |
+
"eval_steps_per_second": 0.736,
|
| 1045 |
+
"step": 30500
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 0.33,
|
| 1049 |
+
"learning_rate": 3.3270191799155956e-05,
|
| 1050 |
+
"loss": 1.539,
|
| 1051 |
+
"step": 31000
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 0.33,
|
| 1055 |
+
"eval_bleu": 7.9101,
|
| 1056 |
+
"eval_gen_len": 25.1115,
|
| 1057 |
+
"eval_loss": 3.56876540184021,
|
| 1058 |
+
"eval_meteor": 0.2294,
|
| 1059 |
+
"eval_runtime": 78.3276,
|
| 1060 |
+
"eval_samples_per_second": 6.639,
|
| 1061 |
+
"eval_steps_per_second": 0.83,
|
| 1062 |
+
"step": 31000
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 0.34,
|
| 1066 |
+
"learning_rate": 3.300035618301331e-05,
|
| 1067 |
+
"loss": 1.6394,
|
| 1068 |
+
"step": 31500
|
| 1069 |
+
},
|
| 1070 |
+
{
|
| 1071 |
+
"epoch": 0.34,
|
| 1072 |
+
"eval_bleu": 8.1847,
|
| 1073 |
+
"eval_gen_len": 24.5731,
|
| 1074 |
+
"eval_loss": 3.5614802837371826,
|
| 1075 |
+
"eval_meteor": 0.2322,
|
| 1076 |
+
"eval_runtime": 70.0131,
|
| 1077 |
+
"eval_samples_per_second": 7.427,
|
| 1078 |
+
"eval_steps_per_second": 0.928,
|
| 1079 |
+
"step": 31500
|
| 1080 |
+
},
|
| 1081 |
+
{
|
| 1082 |
+
"epoch": 0.35,
|
| 1083 |
+
"learning_rate": 3.273052056687066e-05,
|
| 1084 |
+
"loss": 1.5546,
|
| 1085 |
+
"step": 32000
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.35,
|
| 1089 |
+
"eval_bleu": 7.9997,
|
| 1090 |
+
"eval_gen_len": 25.1596,
|
| 1091 |
+
"eval_loss": 3.56968355178833,
|
| 1092 |
+
"eval_meteor": 0.2339,
|
| 1093 |
+
"eval_runtime": 68.1359,
|
| 1094 |
+
"eval_samples_per_second": 7.632,
|
| 1095 |
+
"eval_steps_per_second": 0.954,
|
| 1096 |
+
"step": 32000
|
| 1097 |
+
},
|
| 1098 |
+
{
|
| 1099 |
+
"epoch": 0.35,
|
| 1100 |
+
"learning_rate": 3.246068495072802e-05,
|
| 1101 |
+
"loss": 1.5173,
|
| 1102 |
+
"step": 32500
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 0.35,
|
| 1106 |
+
"eval_bleu": 8.1149,
|
| 1107 |
+
"eval_gen_len": 27.0596,
|
| 1108 |
+
"eval_loss": 3.5782711505889893,
|
| 1109 |
+
"eval_meteor": 0.229,
|
| 1110 |
+
"eval_runtime": 92.9196,
|
| 1111 |
+
"eval_samples_per_second": 5.596,
|
| 1112 |
+
"eval_steps_per_second": 0.7,
|
| 1113 |
+
"step": 32500
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.36,
|
| 1117 |
+
"learning_rate": 3.219084933458537e-05,
|
| 1118 |
+
"loss": 1.6158,
|
| 1119 |
+
"step": 33000
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 0.36,
|
| 1123 |
+
"eval_bleu": 8.6261,
|
| 1124 |
+
"eval_gen_len": 25.5346,
|
| 1125 |
+
"eval_loss": 3.5298867225646973,
|
| 1126 |
+
"eval_meteor": 0.2419,
|
| 1127 |
+
"eval_runtime": 64.5727,
|
| 1128 |
+
"eval_samples_per_second": 8.053,
|
| 1129 |
+
"eval_steps_per_second": 1.007,
|
| 1130 |
+
"step": 33000
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 0.36,
|
| 1134 |
+
"learning_rate": 3.1921013718442724e-05,
|
| 1135 |
+
"loss": 1.5266,
|
| 1136 |
+
"step": 33500
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 0.36,
|
| 1140 |
+
"eval_bleu": 8.7619,
|
| 1141 |
+
"eval_gen_len": 31.4077,
|
| 1142 |
+
"eval_loss": 3.522501230239868,
|
| 1143 |
+
"eval_meteor": 0.2319,
|
| 1144 |
+
"eval_runtime": 122.5254,
|
| 1145 |
+
"eval_samples_per_second": 4.244,
|
| 1146 |
+
"eval_steps_per_second": 0.531,
|
| 1147 |
+
"step": 33500
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 0.37,
|
| 1151 |
+
"learning_rate": 3.1651178102300085e-05,
|
| 1152 |
+
"loss": 1.577,
|
| 1153 |
+
"step": 34000
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 0.37,
|
| 1157 |
+
"eval_bleu": 8.611,
|
| 1158 |
+
"eval_gen_len": 26.15,
|
| 1159 |
+
"eval_loss": 3.542168140411377,
|
| 1160 |
+
"eval_meteor": 0.2346,
|
| 1161 |
+
"eval_runtime": 72.838,
|
| 1162 |
+
"eval_samples_per_second": 7.139,
|
| 1163 |
+
"eval_steps_per_second": 0.892,
|
| 1164 |
+
"step": 34000
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 0.37,
|
| 1168 |
+
"learning_rate": 3.138134248615743e-05,
|
| 1169 |
+
"loss": 1.5787,
|
| 1170 |
+
"step": 34500
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 0.37,
|
| 1174 |
+
"eval_bleu": 8.0924,
|
| 1175 |
+
"eval_gen_len": 29.2692,
|
| 1176 |
+
"eval_loss": 3.5313644409179688,
|
| 1177 |
+
"eval_meteor": 0.2301,
|
| 1178 |
+
"eval_runtime": 104.0102,
|
| 1179 |
+
"eval_samples_per_second": 5.0,
|
| 1180 |
+
"eval_steps_per_second": 0.625,
|
| 1181 |
+
"step": 34500
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 0.38,
|
| 1185 |
+
"learning_rate": 3.111150687001479e-05,
|
| 1186 |
+
"loss": 1.5635,
|
| 1187 |
+
"step": 35000
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 0.38,
|
| 1191 |
+
"eval_bleu": 8.2094,
|
| 1192 |
+
"eval_gen_len": 25.4942,
|
| 1193 |
+
"eval_loss": 3.5328898429870605,
|
| 1194 |
+
"eval_meteor": 0.2314,
|
| 1195 |
+
"eval_runtime": 62.8928,
|
| 1196 |
+
"eval_samples_per_second": 8.268,
|
| 1197 |
+
"eval_steps_per_second": 1.034,
|
| 1198 |
+
"step": 35000
|
| 1199 |
+
},
|
| 1200 |
+
{
|
| 1201 |
+
"epoch": 0.38,
|
| 1202 |
+
"learning_rate": 3.0841671253872146e-05,
|
| 1203 |
+
"loss": 1.4967,
|
| 1204 |
+
"step": 35500
|
| 1205 |
+
},
|
| 1206 |
+
{
|
| 1207 |
+
"epoch": 0.38,
|
| 1208 |
+
"eval_bleu": 7.9224,
|
| 1209 |
+
"eval_gen_len": 28.1365,
|
| 1210 |
+
"eval_loss": 3.5347652435302734,
|
| 1211 |
+
"eval_meteor": 0.2298,
|
| 1212 |
+
"eval_runtime": 95.4251,
|
| 1213 |
+
"eval_samples_per_second": 5.449,
|
| 1214 |
+
"eval_steps_per_second": 0.681,
|
| 1215 |
+
"step": 35500
|
| 1216 |
+
},
|
| 1217 |
+
{
|
| 1218 |
+
"epoch": 0.39,
|
| 1219 |
+
"learning_rate": 3.057183563772949e-05,
|
| 1220 |
+
"loss": 1.5375,
|
| 1221 |
+
"step": 36000
|
| 1222 |
+
},
|
| 1223 |
+
{
|
| 1224 |
+
"epoch": 0.39,
|
| 1225 |
+
"eval_bleu": 8.062,
|
| 1226 |
+
"eval_gen_len": 25.7346,
|
| 1227 |
+
"eval_loss": 3.5164051055908203,
|
| 1228 |
+
"eval_meteor": 0.2375,
|
| 1229 |
+
"eval_runtime": 72.4588,
|
| 1230 |
+
"eval_samples_per_second": 7.176,
|
| 1231 |
+
"eval_steps_per_second": 0.897,
|
| 1232 |
+
"step": 36000
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 0.39,
|
| 1236 |
+
"learning_rate": 3.030200002158685e-05,
|
| 1237 |
+
"loss": 1.5928,
|
| 1238 |
+
"step": 36500
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 0.39,
|
| 1242 |
+
"eval_bleu": 8.4845,
|
| 1243 |
+
"eval_gen_len": 25.8462,
|
| 1244 |
+
"eval_loss": 3.5118257999420166,
|
| 1245 |
+
"eval_meteor": 0.2439,
|
| 1246 |
+
"eval_runtime": 69.4765,
|
| 1247 |
+
"eval_samples_per_second": 7.485,
|
| 1248 |
+
"eval_steps_per_second": 0.936,
|
| 1249 |
+
"step": 36500
|
| 1250 |
+
},
|
| 1251 |
+
{
|
| 1252 |
+
"epoch": 0.4,
|
| 1253 |
+
"learning_rate": 3.0032164405444208e-05,
|
| 1254 |
+
"loss": 1.582,
|
| 1255 |
+
"step": 37000
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 0.4,
|
| 1259 |
+
"eval_bleu": 8.2526,
|
| 1260 |
+
"eval_gen_len": 25.9327,
|
| 1261 |
+
"eval_loss": 3.505610227584839,
|
| 1262 |
+
"eval_meteor": 0.2395,
|
| 1263 |
+
"eval_runtime": 60.8773,
|
| 1264 |
+
"eval_samples_per_second": 8.542,
|
| 1265 |
+
"eval_steps_per_second": 1.068,
|
| 1266 |
+
"step": 37000
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 0.4,
|
| 1270 |
+
"learning_rate": 2.9762328789301558e-05,
|
| 1271 |
+
"loss": 1.4958,
|
| 1272 |
+
"step": 37500
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 0.4,
|
| 1276 |
+
"eval_bleu": 8.3315,
|
| 1277 |
+
"eval_gen_len": 28.5365,
|
| 1278 |
+
"eval_loss": 3.517829418182373,
|
| 1279 |
+
"eval_meteor": 0.2271,
|
| 1280 |
+
"eval_runtime": 94.0897,
|
| 1281 |
+
"eval_samples_per_second": 5.527,
|
| 1282 |
+
"eval_steps_per_second": 0.691,
|
| 1283 |
+
"step": 37500
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 0.41,
|
| 1287 |
+
"learning_rate": 2.9492493173158915e-05,
|
| 1288 |
+
"loss": 1.5524,
|
| 1289 |
+
"step": 38000
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 0.41,
|
| 1293 |
+
"eval_bleu": 8.765,
|
| 1294 |
+
"eval_gen_len": 26.2865,
|
| 1295 |
+
"eval_loss": 3.506021738052368,
|
| 1296 |
+
"eval_meteor": 0.2364,
|
| 1297 |
+
"eval_runtime": 79.4441,
|
| 1298 |
+
"eval_samples_per_second": 6.545,
|
| 1299 |
+
"eval_steps_per_second": 0.818,
|
| 1300 |
+
"step": 38000
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 0.42,
|
| 1304 |
+
"learning_rate": 2.922265755701627e-05,
|
| 1305 |
+
"loss": 1.4689,
|
| 1306 |
+
"step": 38500
|
| 1307 |
+
},
|
| 1308 |
+
{
|
| 1309 |
+
"epoch": 0.42,
|
| 1310 |
+
"eval_bleu": 8.5295,
|
| 1311 |
+
"eval_gen_len": 29.5769,
|
| 1312 |
+
"eval_loss": 3.5012190341949463,
|
| 1313 |
+
"eval_meteor": 0.2327,
|
| 1314 |
+
"eval_runtime": 103.4617,
|
| 1315 |
+
"eval_samples_per_second": 5.026,
|
| 1316 |
+
"eval_steps_per_second": 0.628,
|
| 1317 |
+
"step": 38500
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 0.42,
|
| 1321 |
+
"learning_rate": 2.895282194087362e-05,
|
| 1322 |
+
"loss": 1.5345,
|
| 1323 |
+
"step": 39000
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 0.42,
|
| 1327 |
+
"eval_bleu": 8.7423,
|
| 1328 |
+
"eval_gen_len": 25.3269,
|
| 1329 |
+
"eval_loss": 3.498300313949585,
|
| 1330 |
+
"eval_meteor": 0.2342,
|
| 1331 |
+
"eval_runtime": 73.6072,
|
| 1332 |
+
"eval_samples_per_second": 7.065,
|
| 1333 |
+
"eval_steps_per_second": 0.883,
|
| 1334 |
+
"step": 39000
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 0.43,
|
| 1338 |
+
"learning_rate": 2.8682986324730976e-05,
|
| 1339 |
+
"loss": 1.5282,
|
| 1340 |
+
"step": 39500
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 0.43,
|
| 1344 |
+
"eval_bleu": 8.9296,
|
| 1345 |
+
"eval_gen_len": 25.1173,
|
| 1346 |
+
"eval_loss": 3.4799299240112305,
|
| 1347 |
+
"eval_meteor": 0.2425,
|
| 1348 |
+
"eval_runtime": 71.1898,
|
| 1349 |
+
"eval_samples_per_second": 7.304,
|
| 1350 |
+
"eval_steps_per_second": 0.913,
|
| 1351 |
+
"step": 39500
|
| 1352 |
+
},
|
| 1353 |
+
{
|
| 1354 |
+
"epoch": 0.43,
|
| 1355 |
+
"learning_rate": 2.8413150708588333e-05,
|
| 1356 |
+
"loss": 1.5021,
|
| 1357 |
+
"step": 40000
|
| 1358 |
+
},
|
| 1359 |
+
{
|
| 1360 |
+
"epoch": 0.43,
|
| 1361 |
+
"eval_bleu": 7.9284,
|
| 1362 |
+
"eval_gen_len": 26.2962,
|
| 1363 |
+
"eval_loss": 3.4839093685150146,
|
| 1364 |
+
"eval_meteor": 0.236,
|
| 1365 |
+
"eval_runtime": 77.544,
|
| 1366 |
+
"eval_samples_per_second": 6.706,
|
| 1367 |
+
"eval_steps_per_second": 0.838,
|
| 1368 |
+
"step": 40000
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 0.44,
|
| 1372 |
+
"learning_rate": 2.8143315092445684e-05,
|
| 1373 |
+
"loss": 1.5072,
|
| 1374 |
+
"step": 40500
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 0.44,
|
| 1378 |
+
"eval_bleu": 8.7855,
|
| 1379 |
+
"eval_gen_len": 25.425,
|
| 1380 |
+
"eval_loss": 3.4775073528289795,
|
| 1381 |
+
"eval_meteor": 0.238,
|
| 1382 |
+
"eval_runtime": 84.0571,
|
| 1383 |
+
"eval_samples_per_second": 6.186,
|
| 1384 |
+
"eval_steps_per_second": 0.773,
|
| 1385 |
+
"step": 40500
|
| 1386 |
+
},
|
| 1387 |
+
{
|
| 1388 |
+
"epoch": 0.44,
|
| 1389 |
+
"learning_rate": 2.7873479476303038e-05,
|
| 1390 |
+
"loss": 1.5348,
|
| 1391 |
+
"step": 41000
|
| 1392 |
+
},
|
| 1393 |
+
{
|
| 1394 |
+
"epoch": 0.44,
|
| 1395 |
+
"eval_bleu": 8.3592,
|
| 1396 |
+
"eval_gen_len": 25.4673,
|
| 1397 |
+
"eval_loss": 3.4518544673919678,
|
| 1398 |
+
"eval_meteor": 0.24,
|
| 1399 |
+
"eval_runtime": 81.9583,
|
| 1400 |
+
"eval_samples_per_second": 6.345,
|
| 1401 |
+
"eval_steps_per_second": 0.793,
|
| 1402 |
+
"step": 41000
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 0.45,
|
| 1406 |
+
"learning_rate": 2.7603643860160395e-05,
|
| 1407 |
+
"loss": 1.5236,
|
| 1408 |
+
"step": 41500
|
| 1409 |
+
},
|
| 1410 |
+
{
|
| 1411 |
+
"epoch": 0.45,
|
| 1412 |
+
"eval_bleu": 8.589,
|
| 1413 |
+
"eval_gen_len": 28.1846,
|
| 1414 |
+
"eval_loss": 3.466686248779297,
|
| 1415 |
+
"eval_meteor": 0.2352,
|
| 1416 |
+
"eval_runtime": 91.2864,
|
| 1417 |
+
"eval_samples_per_second": 5.696,
|
| 1418 |
+
"eval_steps_per_second": 0.712,
|
| 1419 |
+
"step": 41500
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 0.45,
|
| 1423 |
+
"learning_rate": 2.7333808244017745e-05,
|
| 1424 |
+
"loss": 1.4695,
|
| 1425 |
+
"step": 42000
|
| 1426 |
+
},
|
| 1427 |
+
{
|
| 1428 |
+
"epoch": 0.45,
|
| 1429 |
+
"eval_bleu": 8.5467,
|
| 1430 |
+
"eval_gen_len": 26.9385,
|
| 1431 |
+
"eval_loss": 3.4435806274414062,
|
| 1432 |
+
"eval_meteor": 0.2396,
|
| 1433 |
+
"eval_runtime": 89.9168,
|
| 1434 |
+
"eval_samples_per_second": 5.783,
|
| 1435 |
+
"eval_steps_per_second": 0.723,
|
| 1436 |
+
"step": 42000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 0.46,
|
| 1440 |
+
"learning_rate": 2.70639726278751e-05,
|
| 1441 |
+
"loss": 1.4733,
|
| 1442 |
+
"step": 42500
|
| 1443 |
+
},
|
| 1444 |
+
{
|
| 1445 |
+
"epoch": 0.46,
|
| 1446 |
+
"eval_bleu": 8.3771,
|
| 1447 |
+
"eval_gen_len": 26.6096,
|
| 1448 |
+
"eval_loss": 3.455449342727661,
|
| 1449 |
+
"eval_meteor": 0.2347,
|
| 1450 |
+
"eval_runtime": 85.0784,
|
| 1451 |
+
"eval_samples_per_second": 6.112,
|
| 1452 |
+
"eval_steps_per_second": 0.764,
|
| 1453 |
+
"step": 42500
|
| 1454 |
+
},
|
| 1455 |
+
{
|
| 1456 |
+
"epoch": 0.46,
|
| 1457 |
+
"learning_rate": 2.6794137011732456e-05,
|
| 1458 |
+
"loss": 1.5398,
|
| 1459 |
+
"step": 43000
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 0.46,
|
| 1463 |
+
"eval_bleu": 9.2126,
|
| 1464 |
+
"eval_gen_len": 24.6538,
|
| 1465 |
+
"eval_loss": 3.436018943786621,
|
| 1466 |
+
"eval_meteor": 0.2434,
|
| 1467 |
+
"eval_runtime": 81.8122,
|
| 1468 |
+
"eval_samples_per_second": 6.356,
|
| 1469 |
+
"eval_steps_per_second": 0.795,
|
| 1470 |
+
"step": 43000
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 0.47,
|
| 1474 |
+
"learning_rate": 2.6524301395589806e-05,
|
| 1475 |
+
"loss": 1.5596,
|
| 1476 |
+
"step": 43500
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 0.47,
|
| 1480 |
+
"eval_bleu": 9.2745,
|
| 1481 |
+
"eval_gen_len": 26.1,
|
| 1482 |
+
"eval_loss": 3.425743579864502,
|
| 1483 |
+
"eval_meteor": 0.2521,
|
| 1484 |
+
"eval_runtime": 70.71,
|
| 1485 |
+
"eval_samples_per_second": 7.354,
|
| 1486 |
+
"eval_steps_per_second": 0.919,
|
| 1487 |
+
"step": 43500
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 0.47,
|
| 1491 |
+
"learning_rate": 2.6254465779447164e-05,
|
| 1492 |
+
"loss": 1.505,
|
| 1493 |
+
"step": 44000
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 0.47,
|
| 1497 |
+
"eval_bleu": 8.8478,
|
| 1498 |
+
"eval_gen_len": 24.9731,
|
| 1499 |
+
"eval_loss": 3.428138017654419,
|
| 1500 |
+
"eval_meteor": 0.2421,
|
| 1501 |
+
"eval_runtime": 81.3515,
|
| 1502 |
+
"eval_samples_per_second": 6.392,
|
| 1503 |
+
"eval_steps_per_second": 0.799,
|
| 1504 |
+
"step": 44000
|
| 1505 |
+
},
|
| 1506 |
+
{
|
| 1507 |
+
"epoch": 0.48,
|
| 1508 |
+
"learning_rate": 2.5984630163304517e-05,
|
| 1509 |
+
"loss": 1.48,
|
| 1510 |
+
"step": 44500
|
| 1511 |
+
},
|
| 1512 |
+
{
|
| 1513 |
+
"epoch": 0.48,
|
| 1514 |
+
"eval_bleu": 9.237,
|
| 1515 |
+
"eval_gen_len": 24.2058,
|
| 1516 |
+
"eval_loss": 3.465346336364746,
|
| 1517 |
+
"eval_meteor": 0.2392,
|
| 1518 |
+
"eval_runtime": 78.2893,
|
| 1519 |
+
"eval_samples_per_second": 6.642,
|
| 1520 |
+
"eval_steps_per_second": 0.83,
|
| 1521 |
+
"step": 44500
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 0.49,
|
| 1525 |
+
"learning_rate": 2.5714794547161868e-05,
|
| 1526 |
+
"loss": 1.5167,
|
| 1527 |
+
"step": 45000
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 0.49,
|
| 1531 |
+
"eval_bleu": 8.8344,
|
| 1532 |
+
"eval_gen_len": 25.8192,
|
| 1533 |
+
"eval_loss": 3.440796375274658,
|
| 1534 |
+
"eval_meteor": 0.2355,
|
| 1535 |
+
"eval_runtime": 73.3566,
|
| 1536 |
+
"eval_samples_per_second": 7.089,
|
| 1537 |
+
"eval_steps_per_second": 0.886,
|
| 1538 |
+
"step": 45000
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 0.49,
|
| 1542 |
+
"learning_rate": 2.5444958931019225e-05,
|
| 1543 |
+
"loss": 1.4691,
|
| 1544 |
+
"step": 45500
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 0.49,
|
| 1548 |
+
"eval_bleu": 9.0831,
|
| 1549 |
+
"eval_gen_len": 25.8577,
|
| 1550 |
+
"eval_loss": 3.424842596054077,
|
| 1551 |
+
"eval_meteor": 0.2449,
|
| 1552 |
+
"eval_runtime": 83.8868,
|
| 1553 |
+
"eval_samples_per_second": 6.199,
|
| 1554 |
+
"eval_steps_per_second": 0.775,
|
| 1555 |
+
"step": 45500
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 0.5,
|
| 1559 |
+
"learning_rate": 2.5175123314876582e-05,
|
| 1560 |
+
"loss": 1.4734,
|
| 1561 |
+
"step": 46000
|
| 1562 |
+
},
|
| 1563 |
+
{
|
| 1564 |
+
"epoch": 0.5,
|
| 1565 |
+
"eval_bleu": 8.6622,
|
| 1566 |
+
"eval_gen_len": 26.0385,
|
| 1567 |
+
"eval_loss": 3.406123399734497,
|
| 1568 |
+
"eval_meteor": 0.2472,
|
| 1569 |
+
"eval_runtime": 87.066,
|
| 1570 |
+
"eval_samples_per_second": 5.972,
|
| 1571 |
+
"eval_steps_per_second": 0.747,
|
| 1572 |
+
"step": 46000
|
| 1573 |
+
},
|
| 1574 |
+
{
|
| 1575 |
+
"epoch": 0.5,
|
| 1576 |
+
"learning_rate": 2.4905287698733932e-05,
|
| 1577 |
+
"loss": 1.4354,
|
| 1578 |
+
"step": 46500
|
| 1579 |
+
},
|
| 1580 |
+
{
|
| 1581 |
+
"epoch": 0.5,
|
| 1582 |
+
"eval_bleu": 9.3409,
|
| 1583 |
+
"eval_gen_len": 26.0077,
|
| 1584 |
+
"eval_loss": 3.4149692058563232,
|
| 1585 |
+
"eval_meteor": 0.2462,
|
| 1586 |
+
"eval_runtime": 75.3366,
|
| 1587 |
+
"eval_samples_per_second": 6.902,
|
| 1588 |
+
"eval_steps_per_second": 0.863,
|
| 1589 |
+
"step": 46500
|
| 1590 |
+
},
|
| 1591 |
+
{
|
| 1592 |
+
"epoch": 0.51,
|
| 1593 |
+
"learning_rate": 2.4635452082591286e-05,
|
| 1594 |
+
"loss": 1.4841,
|
| 1595 |
+
"step": 47000
|
| 1596 |
+
},
|
| 1597 |
+
{
|
| 1598 |
+
"epoch": 0.51,
|
| 1599 |
+
"eval_bleu": 8.3645,
|
| 1600 |
+
"eval_gen_len": 27.2692,
|
| 1601 |
+
"eval_loss": 3.410163640975952,
|
| 1602 |
+
"eval_meteor": 0.2377,
|
| 1603 |
+
"eval_runtime": 88.8645,
|
| 1604 |
+
"eval_samples_per_second": 5.852,
|
| 1605 |
+
"eval_steps_per_second": 0.731,
|
| 1606 |
+
"step": 47000
|
| 1607 |
+
},
|
| 1608 |
+
{
|
| 1609 |
+
"epoch": 0.51,
|
| 1610 |
+
"learning_rate": 2.436561646644864e-05,
|
| 1611 |
+
"loss": 1.4163,
|
| 1612 |
+
"step": 47500
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 0.51,
|
| 1616 |
+
"eval_bleu": 8.7482,
|
| 1617 |
+
"eval_gen_len": 25.1442,
|
| 1618 |
+
"eval_loss": 3.4322900772094727,
|
| 1619 |
+
"eval_meteor": 0.2329,
|
| 1620 |
+
"eval_runtime": 80.4665,
|
| 1621 |
+
"eval_samples_per_second": 6.462,
|
| 1622 |
+
"eval_steps_per_second": 0.808,
|
| 1623 |
+
"step": 47500
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 0.52,
|
| 1627 |
+
"learning_rate": 2.4095780850305994e-05,
|
| 1628 |
+
"loss": 1.4859,
|
| 1629 |
+
"step": 48000
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 0.52,
|
| 1633 |
+
"eval_bleu": 9.0356,
|
| 1634 |
+
"eval_gen_len": 27.9096,
|
| 1635 |
+
"eval_loss": 3.409453868865967,
|
| 1636 |
+
"eval_meteor": 0.238,
|
| 1637 |
+
"eval_runtime": 97.3658,
|
| 1638 |
+
"eval_samples_per_second": 5.341,
|
| 1639 |
+
"eval_steps_per_second": 0.668,
|
| 1640 |
+
"step": 48000
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 0.52,
|
| 1644 |
+
"learning_rate": 2.382594523416335e-05,
|
| 1645 |
+
"loss": 1.5179,
|
| 1646 |
+
"step": 48500
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 0.52,
|
| 1650 |
+
"eval_bleu": 8.9689,
|
| 1651 |
+
"eval_gen_len": 25.9173,
|
| 1652 |
+
"eval_loss": 3.4041757583618164,
|
| 1653 |
+
"eval_meteor": 0.2426,
|
| 1654 |
+
"eval_runtime": 66.1422,
|
| 1655 |
+
"eval_samples_per_second": 7.862,
|
| 1656 |
+
"eval_steps_per_second": 0.983,
|
| 1657 |
+
"step": 48500
|
| 1658 |
+
},
|
| 1659 |
+
{
|
| 1660 |
+
"epoch": 0.53,
|
| 1661 |
+
"learning_rate": 2.35561096180207e-05,
|
| 1662 |
+
"loss": 1.4451,
|
| 1663 |
+
"step": 49000
|
| 1664 |
+
},
|
| 1665 |
+
{
|
| 1666 |
+
"epoch": 0.53,
|
| 1667 |
+
"eval_bleu": 9.0897,
|
| 1668 |
+
"eval_gen_len": 24.6654,
|
| 1669 |
+
"eval_loss": 3.383331537246704,
|
| 1670 |
+
"eval_meteor": 0.244,
|
| 1671 |
+
"eval_runtime": 60.7417,
|
| 1672 |
+
"eval_samples_per_second": 8.561,
|
| 1673 |
+
"eval_steps_per_second": 1.07,
|
| 1674 |
+
"step": 49000
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 0.53,
|
| 1678 |
+
"learning_rate": 2.3286274001878055e-05,
|
| 1679 |
+
"loss": 1.4377,
|
| 1680 |
+
"step": 49500
|
| 1681 |
+
},
|
| 1682 |
+
{
|
| 1683 |
+
"epoch": 0.53,
|
| 1684 |
+
"eval_bleu": 9.0223,
|
| 1685 |
+
"eval_gen_len": 25.3788,
|
| 1686 |
+
"eval_loss": 3.3942527770996094,
|
| 1687 |
+
"eval_meteor": 0.2481,
|
| 1688 |
+
"eval_runtime": 71.1329,
|
| 1689 |
+
"eval_samples_per_second": 7.31,
|
| 1690 |
+
"eval_steps_per_second": 0.914,
|
| 1691 |
+
"step": 49500
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 0.54,
|
| 1695 |
+
"learning_rate": 2.3016438385735412e-05,
|
| 1696 |
+
"loss": 1.4162,
|
| 1697 |
+
"step": 50000
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 0.54,
|
| 1701 |
+
"eval_bleu": 9.5501,
|
| 1702 |
+
"eval_gen_len": 25.05,
|
| 1703 |
+
"eval_loss": 3.3857383728027344,
|
| 1704 |
+
"eval_meteor": 0.2468,
|
| 1705 |
+
"eval_runtime": 77.3856,
|
| 1706 |
+
"eval_samples_per_second": 6.72,
|
| 1707 |
+
"eval_steps_per_second": 0.84,
|
| 1708 |
+
"step": 50000
|
| 1709 |
+
},
|
| 1710 |
+
{
|
| 1711 |
+
"epoch": 0.55,
|
| 1712 |
+
"learning_rate": 2.2746602769592766e-05,
|
| 1713 |
+
"loss": 1.5209,
|
| 1714 |
+
"step": 50500
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 0.55,
|
| 1718 |
+
"eval_bleu": 8.9994,
|
| 1719 |
+
"eval_gen_len": 25.1058,
|
| 1720 |
+
"eval_loss": 3.3768198490142822,
|
| 1721 |
+
"eval_meteor": 0.2424,
|
| 1722 |
+
"eval_runtime": 72.911,
|
| 1723 |
+
"eval_samples_per_second": 7.132,
|
| 1724 |
+
"eval_steps_per_second": 0.891,
|
| 1725 |
+
"step": 50500
|
| 1726 |
+
},
|
| 1727 |
+
{
|
| 1728 |
+
"epoch": 0.55,
|
| 1729 |
+
"learning_rate": 2.247676715345012e-05,
|
| 1730 |
+
"loss": 1.426,
|
| 1731 |
+
"step": 51000
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 0.55,
|
| 1735 |
+
"eval_bleu": 9.2556,
|
| 1736 |
+
"eval_gen_len": 25.3865,
|
| 1737 |
+
"eval_loss": 3.3841092586517334,
|
| 1738 |
+
"eval_meteor": 0.2443,
|
| 1739 |
+
"eval_runtime": 62.6291,
|
| 1740 |
+
"eval_samples_per_second": 8.303,
|
| 1741 |
+
"eval_steps_per_second": 1.038,
|
| 1742 |
+
"step": 51000
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 0.56,
|
| 1746 |
+
"learning_rate": 2.2206931537307473e-05,
|
| 1747 |
+
"loss": 1.4453,
|
| 1748 |
+
"step": 51500
|
| 1749 |
+
},
|
| 1750 |
+
{
|
| 1751 |
+
"epoch": 0.56,
|
| 1752 |
+
"eval_bleu": 9.583,
|
| 1753 |
+
"eval_gen_len": 26.3269,
|
| 1754 |
+
"eval_loss": 3.3895211219787598,
|
| 1755 |
+
"eval_meteor": 0.2448,
|
| 1756 |
+
"eval_runtime": 85.3314,
|
| 1757 |
+
"eval_samples_per_second": 6.094,
|
| 1758 |
+
"eval_steps_per_second": 0.762,
|
| 1759 |
+
"step": 51500
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 0.56,
|
| 1763 |
+
"learning_rate": 2.1937095921164827e-05,
|
| 1764 |
+
"loss": 1.4162,
|
| 1765 |
+
"step": 52000
|
| 1766 |
+
},
|
| 1767 |
+
{
|
| 1768 |
+
"epoch": 0.56,
|
| 1769 |
+
"eval_bleu": 9.0803,
|
| 1770 |
+
"eval_gen_len": 25.5231,
|
| 1771 |
+
"eval_loss": 3.385866165161133,
|
| 1772 |
+
"eval_meteor": 0.2413,
|
| 1773 |
+
"eval_runtime": 75.8979,
|
| 1774 |
+
"eval_samples_per_second": 6.851,
|
| 1775 |
+
"eval_steps_per_second": 0.856,
|
| 1776 |
+
"step": 52000
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 0.57,
|
| 1780 |
+
"learning_rate": 2.1667260305022184e-05,
|
| 1781 |
+
"loss": 1.4107,
|
| 1782 |
+
"step": 52500
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 0.57,
|
| 1786 |
+
"eval_bleu": 8.9249,
|
| 1787 |
+
"eval_gen_len": 24.8615,
|
| 1788 |
+
"eval_loss": 3.3849904537200928,
|
| 1789 |
+
"eval_meteor": 0.241,
|
| 1790 |
+
"eval_runtime": 72.4747,
|
| 1791 |
+
"eval_samples_per_second": 7.175,
|
| 1792 |
+
"eval_steps_per_second": 0.897,
|
| 1793 |
+
"step": 52500
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 0.57,
|
| 1797 |
+
"learning_rate": 2.1397424688879535e-05,
|
| 1798 |
+
"loss": 1.4474,
|
| 1799 |
+
"step": 53000
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 0.57,
|
| 1803 |
+
"eval_bleu": 9.5048,
|
| 1804 |
+
"eval_gen_len": 23.975,
|
| 1805 |
+
"eval_loss": 3.3705201148986816,
|
| 1806 |
+
"eval_meteor": 0.2474,
|
| 1807 |
+
"eval_runtime": 76.3998,
|
| 1808 |
+
"eval_samples_per_second": 6.806,
|
| 1809 |
+
"eval_steps_per_second": 0.851,
|
| 1810 |
+
"step": 53000
|
| 1811 |
+
},
|
| 1812 |
+
{
|
| 1813 |
+
"epoch": 0.58,
|
| 1814 |
+
"learning_rate": 2.1127589072736888e-05,
|
| 1815 |
+
"loss": 1.4336,
|
| 1816 |
+
"step": 53500
|
| 1817 |
+
},
|
| 1818 |
+
{
|
| 1819 |
+
"epoch": 0.58,
|
| 1820 |
+
"eval_bleu": 9.7945,
|
| 1821 |
+
"eval_gen_len": 24.3885,
|
| 1822 |
+
"eval_loss": 3.349461078643799,
|
| 1823 |
+
"eval_meteor": 0.2549,
|
| 1824 |
+
"eval_runtime": 69.7201,
|
| 1825 |
+
"eval_samples_per_second": 7.458,
|
| 1826 |
+
"eval_steps_per_second": 0.932,
|
| 1827 |
+
"step": 53500
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 0.58,
|
| 1831 |
+
"learning_rate": 2.0857753456594245e-05,
|
| 1832 |
+
"loss": 1.4261,
|
| 1833 |
+
"step": 54000
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 0.58,
|
| 1837 |
+
"eval_bleu": 9.5374,
|
| 1838 |
+
"eval_gen_len": 24.2692,
|
| 1839 |
+
"eval_loss": 3.369290351867676,
|
| 1840 |
+
"eval_meteor": 0.2444,
|
| 1841 |
+
"eval_runtime": 67.4768,
|
| 1842 |
+
"eval_samples_per_second": 7.706,
|
| 1843 |
+
"eval_steps_per_second": 0.963,
|
| 1844 |
+
"step": 54000
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 0.59,
|
| 1848 |
+
"learning_rate": 2.05879178404516e-05,
|
| 1849 |
+
"loss": 1.409,
|
| 1850 |
+
"step": 54500
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 0.59,
|
| 1854 |
+
"eval_bleu": 9.3034,
|
| 1855 |
+
"eval_gen_len": 24.625,
|
| 1856 |
+
"eval_loss": 3.3803343772888184,
|
| 1857 |
+
"eval_meteor": 0.2436,
|
| 1858 |
+
"eval_runtime": 68.3273,
|
| 1859 |
+
"eval_samples_per_second": 7.61,
|
| 1860 |
+
"eval_steps_per_second": 0.951,
|
| 1861 |
+
"step": 54500
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 0.59,
|
| 1865 |
+
"learning_rate": 2.0318082224308953e-05,
|
| 1866 |
+
"loss": 1.4364,
|
| 1867 |
+
"step": 55000
|
| 1868 |
+
},
|
| 1869 |
+
{
|
| 1870 |
+
"epoch": 0.59,
|
| 1871 |
+
"eval_bleu": 9.6554,
|
| 1872 |
+
"eval_gen_len": 25.5654,
|
| 1873 |
+
"eval_loss": 3.371992349624634,
|
| 1874 |
+
"eval_meteor": 0.2457,
|
| 1875 |
+
"eval_runtime": 82.9163,
|
| 1876 |
+
"eval_samples_per_second": 6.271,
|
| 1877 |
+
"eval_steps_per_second": 0.784,
|
| 1878 |
+
"step": 55000
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 0.6,
|
| 1882 |
+
"learning_rate": 2.0048246608166307e-05,
|
| 1883 |
+
"loss": 1.4184,
|
| 1884 |
+
"step": 55500
|
| 1885 |
+
},
|
| 1886 |
+
{
|
| 1887 |
+
"epoch": 0.6,
|
| 1888 |
+
"eval_bleu": 9.4698,
|
| 1889 |
+
"eval_gen_len": 27.6,
|
| 1890 |
+
"eval_loss": 3.3737027645111084,
|
| 1891 |
+
"eval_meteor": 0.2414,
|
| 1892 |
+
"eval_runtime": 99.5873,
|
| 1893 |
+
"eval_samples_per_second": 5.222,
|
| 1894 |
+
"eval_steps_per_second": 0.653,
|
| 1895 |
+
"step": 55500
|
| 1896 |
+
},
|
| 1897 |
+
{
|
| 1898 |
+
"epoch": 0.6,
|
| 1899 |
+
"learning_rate": 1.977841099202366e-05,
|
| 1900 |
+
"loss": 1.4417,
|
| 1901 |
+
"step": 56000
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 0.6,
|
| 1905 |
+
"eval_bleu": 9.3662,
|
| 1906 |
+
"eval_gen_len": 25.1481,
|
| 1907 |
+
"eval_loss": 3.3708438873291016,
|
| 1908 |
+
"eval_meteor": 0.2446,
|
| 1909 |
+
"eval_runtime": 87.8071,
|
| 1910 |
+
"eval_samples_per_second": 5.922,
|
| 1911 |
+
"eval_steps_per_second": 0.74,
|
| 1912 |
+
"step": 56000
|
| 1913 |
+
},
|
| 1914 |
+
{
|
| 1915 |
+
"epoch": 0.61,
|
| 1916 |
+
"learning_rate": 1.9508575375881014e-05,
|
| 1917 |
+
"loss": 1.3421,
|
| 1918 |
+
"step": 56500
|
| 1919 |
+
},
|
| 1920 |
+
{
|
| 1921 |
+
"epoch": 0.61,
|
| 1922 |
+
"eval_bleu": 9.3448,
|
| 1923 |
+
"eval_gen_len": 24.8404,
|
| 1924 |
+
"eval_loss": 3.3396267890930176,
|
| 1925 |
+
"eval_meteor": 0.2478,
|
| 1926 |
+
"eval_runtime": 73.0359,
|
| 1927 |
+
"eval_samples_per_second": 7.12,
|
| 1928 |
+
"eval_steps_per_second": 0.89,
|
| 1929 |
+
"step": 56500
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 0.62,
|
| 1933 |
+
"learning_rate": 1.9238739759738368e-05,
|
| 1934 |
+
"loss": 1.396,
|
| 1935 |
+
"step": 57000
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 0.62,
|
| 1939 |
+
"eval_bleu": 9.0765,
|
| 1940 |
+
"eval_gen_len": 24.7731,
|
| 1941 |
+
"eval_loss": 3.3500242233276367,
|
| 1942 |
+
"eval_meteor": 0.2413,
|
| 1943 |
+
"eval_runtime": 71.4703,
|
| 1944 |
+
"eval_samples_per_second": 7.276,
|
| 1945 |
+
"eval_steps_per_second": 0.909,
|
| 1946 |
+
"step": 57000
|
| 1947 |
+
},
|
| 1948 |
+
{
|
| 1949 |
+
"epoch": 0.62,
|
| 1950 |
+
"learning_rate": 1.8968904143595722e-05,
|
| 1951 |
+
"loss": 1.4152,
|
| 1952 |
+
"step": 57500
|
| 1953 |
+
},
|
| 1954 |
+
{
|
| 1955 |
+
"epoch": 0.62,
|
| 1956 |
+
"eval_bleu": 9.4934,
|
| 1957 |
+
"eval_gen_len": 24.7885,
|
| 1958 |
+
"eval_loss": 3.335568904876709,
|
| 1959 |
+
"eval_meteor": 0.2496,
|
| 1960 |
+
"eval_runtime": 67.659,
|
| 1961 |
+
"eval_samples_per_second": 7.686,
|
| 1962 |
+
"eval_steps_per_second": 0.961,
|
| 1963 |
+
"step": 57500
|
| 1964 |
+
},
|
| 1965 |
+
{
|
| 1966 |
+
"epoch": 0.63,
|
| 1967 |
+
"learning_rate": 1.8699068527453075e-05,
|
| 1968 |
+
"loss": 1.4245,
|
| 1969 |
+
"step": 58000
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 0.63,
|
| 1973 |
+
"eval_bleu": 9.7477,
|
| 1974 |
+
"eval_gen_len": 24.0096,
|
| 1975 |
+
"eval_loss": 3.344538688659668,
|
| 1976 |
+
"eval_meteor": 0.2494,
|
| 1977 |
+
"eval_runtime": 69.4763,
|
| 1978 |
+
"eval_samples_per_second": 7.485,
|
| 1979 |
+
"eval_steps_per_second": 0.936,
|
| 1980 |
+
"step": 58000
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 0.63,
|
| 1984 |
+
"learning_rate": 1.8429232911310433e-05,
|
| 1985 |
+
"loss": 1.4661,
|
| 1986 |
+
"step": 58500
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 0.63,
|
| 1990 |
+
"eval_bleu": 9.2664,
|
| 1991 |
+
"eval_gen_len": 24.1404,
|
| 1992 |
+
"eval_loss": 3.3510961532592773,
|
| 1993 |
+
"eval_meteor": 0.2416,
|
| 1994 |
+
"eval_runtime": 71.3645,
|
| 1995 |
+
"eval_samples_per_second": 7.287,
|
| 1996 |
+
"eval_steps_per_second": 0.911,
|
| 1997 |
+
"step": 58500
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 0.64,
|
| 2001 |
+
"learning_rate": 1.8159397295167786e-05,
|
| 2002 |
+
"loss": 1.4219,
|
| 2003 |
+
"step": 59000
|
| 2004 |
+
},
|
| 2005 |
+
{
|
| 2006 |
+
"epoch": 0.64,
|
| 2007 |
+
"eval_bleu": 9.5738,
|
| 2008 |
+
"eval_gen_len": 24.6769,
|
| 2009 |
+
"eval_loss": 3.3354907035827637,
|
| 2010 |
+
"eval_meteor": 0.2436,
|
| 2011 |
+
"eval_runtime": 76.9335,
|
| 2012 |
+
"eval_samples_per_second": 6.759,
|
| 2013 |
+
"eval_steps_per_second": 0.845,
|
| 2014 |
+
"step": 59000
|
| 2015 |
+
},
|
| 2016 |
+
{
|
| 2017 |
+
"epoch": 0.64,
|
| 2018 |
+
"learning_rate": 1.7889561679025137e-05,
|
| 2019 |
+
"loss": 1.4051,
|
| 2020 |
+
"step": 59500
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"epoch": 0.64,
|
| 2024 |
+
"eval_bleu": 9.4506,
|
| 2025 |
+
"eval_gen_len": 24.2615,
|
| 2026 |
+
"eval_loss": 3.3152685165405273,
|
| 2027 |
+
"eval_meteor": 0.2495,
|
| 2028 |
+
"eval_runtime": 68.6589,
|
| 2029 |
+
"eval_samples_per_second": 7.574,
|
| 2030 |
+
"eval_steps_per_second": 0.947,
|
| 2031 |
+
"step": 59500
|
| 2032 |
+
},
|
| 2033 |
+
{
|
| 2034 |
+
"epoch": 0.65,
|
| 2035 |
+
"learning_rate": 1.7619726062882494e-05,
|
| 2036 |
+
"loss": 1.4743,
|
| 2037 |
+
"step": 60000
|
| 2038 |
+
},
|
| 2039 |
+
{
|
| 2040 |
+
"epoch": 0.65,
|
| 2041 |
+
"eval_bleu": 9.6423,
|
| 2042 |
+
"eval_gen_len": 24.4865,
|
| 2043 |
+
"eval_loss": 3.3143715858459473,
|
| 2044 |
+
"eval_meteor": 0.2463,
|
| 2045 |
+
"eval_runtime": 71.0116,
|
| 2046 |
+
"eval_samples_per_second": 7.323,
|
| 2047 |
+
"eval_steps_per_second": 0.915,
|
| 2048 |
+
"step": 60000
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 0.65,
|
| 2052 |
+
"learning_rate": 1.7349890446739848e-05,
|
| 2053 |
+
"loss": 1.3961,
|
| 2054 |
+
"step": 60500
|
| 2055 |
+
},
|
| 2056 |
+
{
|
| 2057 |
+
"epoch": 0.65,
|
| 2058 |
+
"eval_bleu": 9.7681,
|
| 2059 |
+
"eval_gen_len": 24.4615,
|
| 2060 |
+
"eval_loss": 3.3041951656341553,
|
| 2061 |
+
"eval_meteor": 0.2526,
|
| 2062 |
+
"eval_runtime": 66.5982,
|
| 2063 |
+
"eval_samples_per_second": 7.808,
|
| 2064 |
+
"eval_steps_per_second": 0.976,
|
| 2065 |
+
"step": 60500
|
| 2066 |
+
},
|
| 2067 |
+
{
|
| 2068 |
+
"epoch": 0.66,
|
| 2069 |
+
"learning_rate": 1.70800548305972e-05,
|
| 2070 |
+
"loss": 1.447,
|
| 2071 |
+
"step": 61000
|
| 2072 |
+
},
|
| 2073 |
+
{
|
| 2074 |
+
"epoch": 0.66,
|
| 2075 |
+
"eval_bleu": 9.6435,
|
| 2076 |
+
"eval_gen_len": 24.6365,
|
| 2077 |
+
"eval_loss": 3.3025214672088623,
|
| 2078 |
+
"eval_meteor": 0.2507,
|
| 2079 |
+
"eval_runtime": 73.4548,
|
| 2080 |
+
"eval_samples_per_second": 7.079,
|
| 2081 |
+
"eval_steps_per_second": 0.885,
|
| 2082 |
+
"step": 61000
|
| 2083 |
+
},
|
| 2084 |
+
{
|
| 2085 |
+
"epoch": 0.66,
|
| 2086 |
+
"learning_rate": 1.6810219214454555e-05,
|
| 2087 |
+
"loss": 1.4102,
|
| 2088 |
+
"step": 61500
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 0.66,
|
| 2092 |
+
"eval_bleu": 9.6897,
|
| 2093 |
+
"eval_gen_len": 24.8731,
|
| 2094 |
+
"eval_loss": 3.3014609813690186,
|
| 2095 |
+
"eval_meteor": 0.25,
|
| 2096 |
+
"eval_runtime": 77.7893,
|
| 2097 |
+
"eval_samples_per_second": 6.685,
|
| 2098 |
+
"eval_steps_per_second": 0.836,
|
| 2099 |
+
"step": 61500
|
| 2100 |
+
},
|
| 2101 |
+
{
|
| 2102 |
+
"epoch": 0.67,
|
| 2103 |
+
"learning_rate": 1.654038359831191e-05,
|
| 2104 |
+
"loss": 1.3739,
|
| 2105 |
+
"step": 62000
|
| 2106 |
+
},
|
| 2107 |
+
{
|
| 2108 |
+
"epoch": 0.67,
|
| 2109 |
+
"eval_bleu": 9.2607,
|
| 2110 |
+
"eval_gen_len": 24.7885,
|
| 2111 |
+
"eval_loss": 3.313209056854248,
|
| 2112 |
+
"eval_meteor": 0.2447,
|
| 2113 |
+
"eval_runtime": 70.1449,
|
| 2114 |
+
"eval_samples_per_second": 7.413,
|
| 2115 |
+
"eval_steps_per_second": 0.927,
|
| 2116 |
+
"step": 62000
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 0.67,
|
| 2120 |
+
"learning_rate": 1.6270547982169263e-05,
|
| 2121 |
+
"loss": 1.3794,
|
| 2122 |
+
"step": 62500
|
| 2123 |
+
},
|
| 2124 |
+
{
|
| 2125 |
+
"epoch": 0.67,
|
| 2126 |
+
"eval_bleu": 9.6355,
|
| 2127 |
+
"eval_gen_len": 24.4365,
|
| 2128 |
+
"eval_loss": 3.3016440868377686,
|
| 2129 |
+
"eval_meteor": 0.2488,
|
| 2130 |
+
"eval_runtime": 72.723,
|
| 2131 |
+
"eval_samples_per_second": 7.15,
|
| 2132 |
+
"eval_steps_per_second": 0.894,
|
| 2133 |
+
"step": 62500
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 0.68,
|
| 2137 |
+
"learning_rate": 1.600071236602662e-05,
|
| 2138 |
+
"loss": 1.3819,
|
| 2139 |
+
"step": 63000
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 0.68,
|
| 2143 |
+
"eval_bleu": 9.9125,
|
| 2144 |
+
"eval_gen_len": 24.5038,
|
| 2145 |
+
"eval_loss": 3.298640489578247,
|
| 2146 |
+
"eval_meteor": 0.248,
|
| 2147 |
+
"eval_runtime": 71.6443,
|
| 2148 |
+
"eval_samples_per_second": 7.258,
|
| 2149 |
+
"eval_steps_per_second": 0.907,
|
| 2150 |
+
"step": 63000
|
| 2151 |
+
},
|
| 2152 |
+
{
|
| 2153 |
+
"epoch": 0.69,
|
| 2154 |
+
"learning_rate": 1.573087674988397e-05,
|
| 2155 |
+
"loss": 1.3598,
|
| 2156 |
+
"step": 63500
|
| 2157 |
+
},
|
| 2158 |
+
{
|
| 2159 |
+
"epoch": 0.69,
|
| 2160 |
+
"eval_bleu": 10.2032,
|
| 2161 |
+
"eval_gen_len": 24.6558,
|
| 2162 |
+
"eval_loss": 3.2921295166015625,
|
| 2163 |
+
"eval_meteor": 0.2534,
|
| 2164 |
+
"eval_runtime": 74.7575,
|
| 2165 |
+
"eval_samples_per_second": 6.956,
|
| 2166 |
+
"eval_steps_per_second": 0.869,
|
| 2167 |
+
"step": 63500
|
| 2168 |
+
},
|
| 2169 |
+
{
|
| 2170 |
+
"epoch": 0.69,
|
| 2171 |
+
"learning_rate": 1.5461041133741324e-05,
|
| 2172 |
+
"loss": 1.3846,
|
| 2173 |
+
"step": 64000
|
| 2174 |
+
},
|
| 2175 |
+
{
|
| 2176 |
+
"epoch": 0.69,
|
| 2177 |
+
"eval_bleu": 9.9342,
|
| 2178 |
+
"eval_gen_len": 23.9923,
|
| 2179 |
+
"eval_loss": 3.2913033962249756,
|
| 2180 |
+
"eval_meteor": 0.2468,
|
| 2181 |
+
"eval_runtime": 68.6789,
|
| 2182 |
+
"eval_samples_per_second": 7.571,
|
| 2183 |
+
"eval_steps_per_second": 0.946,
|
| 2184 |
+
"step": 64000
|
| 2185 |
+
},
|
| 2186 |
+
{
|
| 2187 |
+
"epoch": 0.7,
|
| 2188 |
+
"learning_rate": 1.5191205517598681e-05,
|
| 2189 |
+
"loss": 1.4024,
|
| 2190 |
+
"step": 64500
|
| 2191 |
+
},
|
| 2192 |
+
{
|
| 2193 |
+
"epoch": 0.7,
|
| 2194 |
+
"eval_bleu": 9.445,
|
| 2195 |
+
"eval_gen_len": 25.3865,
|
| 2196 |
+
"eval_loss": 3.2889387607574463,
|
| 2197 |
+
"eval_meteor": 0.2426,
|
| 2198 |
+
"eval_runtime": 78.1546,
|
| 2199 |
+
"eval_samples_per_second": 6.653,
|
| 2200 |
+
"eval_steps_per_second": 0.832,
|
| 2201 |
+
"step": 64500
|
| 2202 |
+
},
|
| 2203 |
+
{
|
| 2204 |
+
"epoch": 0.7,
|
| 2205 |
+
"learning_rate": 1.4921369901456033e-05,
|
| 2206 |
+
"loss": 1.3775,
|
| 2207 |
+
"step": 65000
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 0.7,
|
| 2211 |
+
"eval_bleu": 9.711,
|
| 2212 |
+
"eval_gen_len": 25.5769,
|
| 2213 |
+
"eval_loss": 3.2912492752075195,
|
| 2214 |
+
"eval_meteor": 0.2459,
|
| 2215 |
+
"eval_runtime": 102.461,
|
| 2216 |
+
"eval_samples_per_second": 5.075,
|
| 2217 |
+
"eval_steps_per_second": 0.634,
|
| 2218 |
+
"step": 65000
|
| 2219 |
+
},
|
| 2220 |
+
{
|
| 2221 |
+
"epoch": 0.71,
|
| 2222 |
+
"learning_rate": 1.4651534285313387e-05,
|
| 2223 |
+
"loss": 1.3192,
|
| 2224 |
+
"step": 65500
|
| 2225 |
+
},
|
| 2226 |
+
{
|
| 2227 |
+
"epoch": 0.71,
|
| 2228 |
+
"eval_bleu": 9.5794,
|
| 2229 |
+
"eval_gen_len": 25.2038,
|
| 2230 |
+
"eval_loss": 3.288285970687866,
|
| 2231 |
+
"eval_meteor": 0.2438,
|
| 2232 |
+
"eval_runtime": 81.3849,
|
| 2233 |
+
"eval_samples_per_second": 6.389,
|
| 2234 |
+
"eval_steps_per_second": 0.799,
|
| 2235 |
+
"step": 65500
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 0.71,
|
| 2239 |
+
"learning_rate": 1.4381698669170742e-05,
|
| 2240 |
+
"loss": 1.3426,
|
| 2241 |
+
"step": 66000
|
| 2242 |
+
},
|
| 2243 |
+
{
|
| 2244 |
+
"epoch": 0.71,
|
| 2245 |
+
"eval_bleu": 9.9763,
|
| 2246 |
+
"eval_gen_len": 25.1942,
|
| 2247 |
+
"eval_loss": 3.282227039337158,
|
| 2248 |
+
"eval_meteor": 0.2473,
|
| 2249 |
+
"eval_runtime": 81.2601,
|
| 2250 |
+
"eval_samples_per_second": 6.399,
|
| 2251 |
+
"eval_steps_per_second": 0.8,
|
| 2252 |
+
"step": 66000
|
| 2253 |
+
},
|
| 2254 |
+
{
|
| 2255 |
+
"epoch": 0.72,
|
| 2256 |
+
"learning_rate": 1.4111863053028096e-05,
|
| 2257 |
+
"loss": 1.3669,
|
| 2258 |
+
"step": 66500
|
| 2259 |
+
},
|
| 2260 |
+
{
|
| 2261 |
+
"epoch": 0.72,
|
| 2262 |
+
"eval_bleu": 9.4662,
|
| 2263 |
+
"eval_gen_len": 23.9865,
|
| 2264 |
+
"eval_loss": 3.2904016971588135,
|
| 2265 |
+
"eval_meteor": 0.2483,
|
| 2266 |
+
"eval_runtime": 66.1145,
|
| 2267 |
+
"eval_samples_per_second": 7.865,
|
| 2268 |
+
"eval_steps_per_second": 0.983,
|
| 2269 |
+
"step": 66500
|
| 2270 |
+
},
|
| 2271 |
+
{
|
| 2272 |
+
"epoch": 0.72,
|
| 2273 |
+
"learning_rate": 1.384202743688545e-05,
|
| 2274 |
+
"loss": 1.3624,
|
| 2275 |
+
"step": 67000
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 0.72,
|
| 2279 |
+
"eval_bleu": 9.5602,
|
| 2280 |
+
"eval_gen_len": 24.1096,
|
| 2281 |
+
"eval_loss": 3.299604654312134,
|
| 2282 |
+
"eval_meteor": 0.2452,
|
| 2283 |
+
"eval_runtime": 72.3193,
|
| 2284 |
+
"eval_samples_per_second": 7.19,
|
| 2285 |
+
"eval_steps_per_second": 0.899,
|
| 2286 |
+
"step": 67000
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 0.73,
|
| 2290 |
+
"learning_rate": 1.3572191820742805e-05,
|
| 2291 |
+
"loss": 1.3435,
|
| 2292 |
+
"step": 67500
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 0.73,
|
| 2296 |
+
"eval_bleu": 9.756,
|
| 2297 |
+
"eval_gen_len": 25.4038,
|
| 2298 |
+
"eval_loss": 3.2783830165863037,
|
| 2299 |
+
"eval_meteor": 0.2482,
|
| 2300 |
+
"eval_runtime": 64.4382,
|
| 2301 |
+
"eval_samples_per_second": 8.07,
|
| 2302 |
+
"eval_steps_per_second": 1.009,
|
| 2303 |
+
"step": 67500
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 0.73,
|
| 2307 |
+
"learning_rate": 1.3302356204600157e-05,
|
| 2308 |
+
"loss": 1.3668,
|
| 2309 |
+
"step": 68000
|
| 2310 |
+
},
|
| 2311 |
+
{
|
| 2312 |
+
"epoch": 0.73,
|
| 2313 |
+
"eval_bleu": 9.9278,
|
| 2314 |
+
"eval_gen_len": 24.3115,
|
| 2315 |
+
"eval_loss": 3.2847084999084473,
|
| 2316 |
+
"eval_meteor": 0.246,
|
| 2317 |
+
"eval_runtime": 74.6646,
|
| 2318 |
+
"eval_samples_per_second": 6.964,
|
| 2319 |
+
"eval_steps_per_second": 0.871,
|
| 2320 |
+
"step": 68000
|
| 2321 |
+
},
|
| 2322 |
+
{
|
| 2323 |
+
"epoch": 0.74,
|
| 2324 |
+
"learning_rate": 1.3032520588457511e-05,
|
| 2325 |
+
"loss": 1.4025,
|
| 2326 |
+
"step": 68500
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 0.74,
|
| 2330 |
+
"eval_bleu": 9.8915,
|
| 2331 |
+
"eval_gen_len": 24.4904,
|
| 2332 |
+
"eval_loss": 3.2775745391845703,
|
| 2333 |
+
"eval_meteor": 0.2502,
|
| 2334 |
+
"eval_runtime": 76.4639,
|
| 2335 |
+
"eval_samples_per_second": 6.801,
|
| 2336 |
+
"eval_steps_per_second": 0.85,
|
| 2337 |
+
"step": 68500
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 0.74,
|
| 2341 |
+
"learning_rate": 1.2762684972314867e-05,
|
| 2342 |
+
"loss": 1.334,
|
| 2343 |
+
"step": 69000
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 0.74,
|
| 2347 |
+
"eval_bleu": 10.1572,
|
| 2348 |
+
"eval_gen_len": 25.5462,
|
| 2349 |
+
"eval_loss": 3.2732129096984863,
|
| 2350 |
+
"eval_meteor": 0.2489,
|
| 2351 |
+
"eval_runtime": 88.0016,
|
| 2352 |
+
"eval_samples_per_second": 5.909,
|
| 2353 |
+
"eval_steps_per_second": 0.739,
|
| 2354 |
+
"step": 69000
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 0.75,
|
| 2358 |
+
"learning_rate": 1.249284935617222e-05,
|
| 2359 |
+
"loss": 1.407,
|
| 2360 |
+
"step": 69500
|
| 2361 |
+
},
|
| 2362 |
+
{
|
| 2363 |
+
"epoch": 0.75,
|
| 2364 |
+
"eval_bleu": 9.9716,
|
| 2365 |
+
"eval_gen_len": 24.8692,
|
| 2366 |
+
"eval_loss": 3.2843334674835205,
|
| 2367 |
+
"eval_meteor": 0.2503,
|
| 2368 |
+
"eval_runtime": 76.3651,
|
| 2369 |
+
"eval_samples_per_second": 6.809,
|
| 2370 |
+
"eval_steps_per_second": 0.851,
|
| 2371 |
+
"step": 69500
|
| 2372 |
+
},
|
| 2373 |
+
{
|
| 2374 |
+
"epoch": 0.76,
|
| 2375 |
+
"learning_rate": 1.2223013740029574e-05,
|
| 2376 |
+
"loss": 1.449,
|
| 2377 |
+
"step": 70000
|
| 2378 |
+
},
|
| 2379 |
+
{
|
| 2380 |
+
"epoch": 0.76,
|
| 2381 |
+
"eval_bleu": 9.7516,
|
| 2382 |
+
"eval_gen_len": 24.6115,
|
| 2383 |
+
"eval_loss": 3.26999831199646,
|
| 2384 |
+
"eval_meteor": 0.2446,
|
| 2385 |
+
"eval_runtime": 57.4151,
|
| 2386 |
+
"eval_samples_per_second": 9.057,
|
| 2387 |
+
"eval_steps_per_second": 1.132,
|
| 2388 |
+
"step": 70000
|
| 2389 |
+
},
|
| 2390 |
+
{
|
| 2391 |
+
"epoch": 0.76,
|
| 2392 |
+
"learning_rate": 1.1953178123886928e-05,
|
| 2393 |
+
"loss": 1.3357,
|
| 2394 |
+
"step": 70500
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 0.76,
|
| 2398 |
+
"eval_bleu": 9.7942,
|
| 2399 |
+
"eval_gen_len": 25.0538,
|
| 2400 |
+
"eval_loss": 3.2684452533721924,
|
| 2401 |
+
"eval_meteor": 0.2478,
|
| 2402 |
+
"eval_runtime": 69.1367,
|
| 2403 |
+
"eval_samples_per_second": 7.521,
|
| 2404 |
+
"eval_steps_per_second": 0.94,
|
| 2405 |
+
"step": 70500
|
| 2406 |
+
},
|
| 2407 |
+
{
|
| 2408 |
+
"epoch": 0.77,
|
| 2409 |
+
"learning_rate": 1.1683342507744283e-05,
|
| 2410 |
+
"loss": 1.3437,
|
| 2411 |
+
"step": 71000
|
| 2412 |
+
},
|
| 2413 |
+
{
|
| 2414 |
+
"epoch": 0.77,
|
| 2415 |
+
"eval_bleu": 9.9427,
|
| 2416 |
+
"eval_gen_len": 24.1538,
|
| 2417 |
+
"eval_loss": 3.2602343559265137,
|
| 2418 |
+
"eval_meteor": 0.2486,
|
| 2419 |
+
"eval_runtime": 72.8541,
|
| 2420 |
+
"eval_samples_per_second": 7.138,
|
| 2421 |
+
"eval_steps_per_second": 0.892,
|
| 2422 |
+
"step": 71000
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 0.77,
|
| 2426 |
+
"learning_rate": 1.1413506891601637e-05,
|
| 2427 |
+
"loss": 1.3518,
|
| 2428 |
+
"step": 71500
|
| 2429 |
+
},
|
| 2430 |
+
{
|
| 2431 |
+
"epoch": 0.77,
|
| 2432 |
+
"eval_bleu": 9.8195,
|
| 2433 |
+
"eval_gen_len": 25.35,
|
| 2434 |
+
"eval_loss": 3.2659668922424316,
|
| 2435 |
+
"eval_meteor": 0.2466,
|
| 2436 |
+
"eval_runtime": 68.8561,
|
| 2437 |
+
"eval_samples_per_second": 7.552,
|
| 2438 |
+
"eval_steps_per_second": 0.944,
|
| 2439 |
+
"step": 71500
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 0.78,
|
| 2443 |
+
"learning_rate": 1.1143671275458991e-05,
|
| 2444 |
+
"loss": 1.3762,
|
| 2445 |
+
"step": 72000
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 0.78,
|
| 2449 |
+
"eval_bleu": 10.0553,
|
| 2450 |
+
"eval_gen_len": 24.8635,
|
| 2451 |
+
"eval_loss": 3.257364511489868,
|
| 2452 |
+
"eval_meteor": 0.2516,
|
| 2453 |
+
"eval_runtime": 81.5604,
|
| 2454 |
+
"eval_samples_per_second": 6.376,
|
| 2455 |
+
"eval_steps_per_second": 0.797,
|
| 2456 |
+
"step": 72000
|
| 2457 |
+
},
|
| 2458 |
+
{
|
| 2459 |
+
"epoch": 0.78,
|
| 2460 |
+
"learning_rate": 1.0873835659316345e-05,
|
| 2461 |
+
"loss": 1.3404,
|
| 2462 |
+
"step": 72500
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 0.78,
|
| 2466 |
+
"eval_bleu": 9.9627,
|
| 2467 |
+
"eval_gen_len": 23.9154,
|
| 2468 |
+
"eval_loss": 3.2508692741394043,
|
| 2469 |
+
"eval_meteor": 0.2509,
|
| 2470 |
+
"eval_runtime": 55.132,
|
| 2471 |
+
"eval_samples_per_second": 9.432,
|
| 2472 |
+
"eval_steps_per_second": 1.179,
|
| 2473 |
+
"step": 72500
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 0.79,
|
| 2477 |
+
"learning_rate": 1.06040000431737e-05,
|
| 2478 |
+
"loss": 1.3548,
|
| 2479 |
+
"step": 73000
|
| 2480 |
+
},
|
| 2481 |
+
{
|
| 2482 |
+
"epoch": 0.79,
|
| 2483 |
+
"eval_bleu": 9.8402,
|
| 2484 |
+
"eval_gen_len": 24.7827,
|
| 2485 |
+
"eval_loss": 3.260327100753784,
|
| 2486 |
+
"eval_meteor": 0.2499,
|
| 2487 |
+
"eval_runtime": 75.4748,
|
| 2488 |
+
"eval_samples_per_second": 6.89,
|
| 2489 |
+
"eval_steps_per_second": 0.861,
|
| 2490 |
+
"step": 73000
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 0.79,
|
| 2494 |
+
"learning_rate": 1.0334164427031052e-05,
|
| 2495 |
+
"loss": 1.3642,
|
| 2496 |
+
"step": 73500
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 0.79,
|
| 2500 |
+
"eval_bleu": 9.952,
|
| 2501 |
+
"eval_gen_len": 24.5327,
|
| 2502 |
+
"eval_loss": 3.2416298389434814,
|
| 2503 |
+
"eval_meteor": 0.2534,
|
| 2504 |
+
"eval_runtime": 74.6757,
|
| 2505 |
+
"eval_samples_per_second": 6.963,
|
| 2506 |
+
"eval_steps_per_second": 0.87,
|
| 2507 |
+
"step": 73500
|
| 2508 |
+
},
|
| 2509 |
+
{
|
| 2510 |
+
"epoch": 0.8,
|
| 2511 |
+
"learning_rate": 1.0064328810888408e-05,
|
| 2512 |
+
"loss": 1.3284,
|
| 2513 |
+
"step": 74000
|
| 2514 |
+
},
|
| 2515 |
+
{
|
| 2516 |
+
"epoch": 0.8,
|
| 2517 |
+
"eval_bleu": 9.8073,
|
| 2518 |
+
"eval_gen_len": 25.0212,
|
| 2519 |
+
"eval_loss": 3.2586073875427246,
|
| 2520 |
+
"eval_meteor": 0.2499,
|
| 2521 |
+
"eval_runtime": 76.8596,
|
| 2522 |
+
"eval_samples_per_second": 6.766,
|
| 2523 |
+
"eval_steps_per_second": 0.846,
|
| 2524 |
+
"step": 74000
|
| 2525 |
+
},
|
| 2526 |
+
{
|
| 2527 |
+
"epoch": 0.8,
|
| 2528 |
+
"learning_rate": 9.794493194745761e-06,
|
| 2529 |
+
"loss": 1.3697,
|
| 2530 |
+
"step": 74500
|
| 2531 |
+
},
|
| 2532 |
+
{
|
| 2533 |
+
"epoch": 0.8,
|
| 2534 |
+
"eval_bleu": 10.0465,
|
| 2535 |
+
"eval_gen_len": 24.95,
|
| 2536 |
+
"eval_loss": 3.2522428035736084,
|
| 2537 |
+
"eval_meteor": 0.249,
|
| 2538 |
+
"eval_runtime": 86.1073,
|
| 2539 |
+
"eval_samples_per_second": 6.039,
|
| 2540 |
+
"eval_steps_per_second": 0.755,
|
| 2541 |
+
"step": 74500
|
| 2542 |
+
},
|
| 2543 |
+
{
|
| 2544 |
+
"epoch": 0.81,
|
| 2545 |
+
"learning_rate": 9.524657578603115e-06,
|
| 2546 |
+
"loss": 1.3718,
|
| 2547 |
+
"step": 75000
|
| 2548 |
+
},
|
| 2549 |
+
{
|
| 2550 |
+
"epoch": 0.81,
|
| 2551 |
+
"eval_bleu": 9.8922,
|
| 2552 |
+
"eval_gen_len": 25.0846,
|
| 2553 |
+
"eval_loss": 3.246650457382202,
|
| 2554 |
+
"eval_meteor": 0.2499,
|
| 2555 |
+
"eval_runtime": 78.8353,
|
| 2556 |
+
"eval_samples_per_second": 6.596,
|
| 2557 |
+
"eval_steps_per_second": 0.825,
|
| 2558 |
+
"step": 75000
|
| 2559 |
+
},
|
| 2560 |
+
{
|
| 2561 |
+
"epoch": 0.81,
|
| 2562 |
+
"learning_rate": 9.254821962460469e-06,
|
| 2563 |
+
"loss": 1.3333,
|
| 2564 |
+
"step": 75500
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 0.81,
|
| 2568 |
+
"eval_bleu": 9.5919,
|
| 2569 |
+
"eval_gen_len": 24.5365,
|
| 2570 |
+
"eval_loss": 3.244452953338623,
|
| 2571 |
+
"eval_meteor": 0.2509,
|
| 2572 |
+
"eval_runtime": 69.8377,
|
| 2573 |
+
"eval_samples_per_second": 7.446,
|
| 2574 |
+
"eval_steps_per_second": 0.931,
|
| 2575 |
+
"step": 75500
|
| 2576 |
+
},
|
| 2577 |
+
{
|
| 2578 |
+
"epoch": 0.82,
|
| 2579 |
+
"learning_rate": 8.984986346317824e-06,
|
| 2580 |
+
"loss": 1.4192,
|
| 2581 |
+
"step": 76000
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 0.82,
|
| 2585 |
+
"eval_bleu": 10.1256,
|
| 2586 |
+
"eval_gen_len": 24.8904,
|
| 2587 |
+
"eval_loss": 3.230175733566284,
|
| 2588 |
+
"eval_meteor": 0.2524,
|
| 2589 |
+
"eval_runtime": 66.1729,
|
| 2590 |
+
"eval_samples_per_second": 7.858,
|
| 2591 |
+
"eval_steps_per_second": 0.982,
|
| 2592 |
+
"step": 76000
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 0.83,
|
| 2596 |
+
"learning_rate": 8.715150730175178e-06,
|
| 2597 |
+
"loss": 1.3068,
|
| 2598 |
+
"step": 76500
|
| 2599 |
+
},
|
| 2600 |
+
{
|
| 2601 |
+
"epoch": 0.83,
|
| 2602 |
+
"eval_bleu": 10.1132,
|
| 2603 |
+
"eval_gen_len": 24.5346,
|
| 2604 |
+
"eval_loss": 3.241743326187134,
|
| 2605 |
+
"eval_meteor": 0.2533,
|
| 2606 |
+
"eval_runtime": 80.1867,
|
| 2607 |
+
"eval_samples_per_second": 6.485,
|
| 2608 |
+
"eval_steps_per_second": 0.811,
|
| 2609 |
+
"step": 76500
|
| 2610 |
+
},
|
| 2611 |
+
{
|
| 2612 |
+
"epoch": 0.83,
|
| 2613 |
+
"learning_rate": 8.445315114032532e-06,
|
| 2614 |
+
"loss": 1.4282,
|
| 2615 |
+
"step": 77000
|
| 2616 |
+
},
|
| 2617 |
+
{
|
| 2618 |
+
"epoch": 0.83,
|
| 2619 |
+
"eval_bleu": 9.95,
|
| 2620 |
+
"eval_gen_len": 24.3462,
|
| 2621 |
+
"eval_loss": 3.242541551589966,
|
| 2622 |
+
"eval_meteor": 0.2509,
|
| 2623 |
+
"eval_runtime": 61.0641,
|
| 2624 |
+
"eval_samples_per_second": 8.516,
|
| 2625 |
+
"eval_steps_per_second": 1.064,
|
| 2626 |
+
"step": 77000
|
| 2627 |
+
},
|
| 2628 |
+
{
|
| 2629 |
+
"epoch": 0.84,
|
| 2630 |
+
"learning_rate": 8.175479497889886e-06,
|
| 2631 |
+
"loss": 1.4073,
|
| 2632 |
+
"step": 77500
|
| 2633 |
+
},
|
| 2634 |
+
{
|
| 2635 |
+
"epoch": 0.84,
|
| 2636 |
+
"eval_bleu": 10.0739,
|
| 2637 |
+
"eval_gen_len": 25.3173,
|
| 2638 |
+
"eval_loss": 3.234609842300415,
|
| 2639 |
+
"eval_meteor": 0.2523,
|
| 2640 |
+
"eval_runtime": 76.5129,
|
| 2641 |
+
"eval_samples_per_second": 6.796,
|
| 2642 |
+
"eval_steps_per_second": 0.85,
|
| 2643 |
+
"step": 77500
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 0.84,
|
| 2647 |
+
"learning_rate": 7.905643881747241e-06,
|
| 2648 |
+
"loss": 1.3717,
|
| 2649 |
+
"step": 78000
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 0.84,
|
| 2653 |
+
"eval_bleu": 10.2478,
|
| 2654 |
+
"eval_gen_len": 24.7577,
|
| 2655 |
+
"eval_loss": 3.2297909259796143,
|
| 2656 |
+
"eval_meteor": 0.2543,
|
| 2657 |
+
"eval_runtime": 79.2758,
|
| 2658 |
+
"eval_samples_per_second": 6.559,
|
| 2659 |
+
"eval_steps_per_second": 0.82,
|
| 2660 |
+
"step": 78000
|
| 2661 |
+
},
|
| 2662 |
+
{
|
| 2663 |
+
"epoch": 0.85,
|
| 2664 |
+
"learning_rate": 7.635808265604593e-06,
|
| 2665 |
+
"loss": 1.3212,
|
| 2666 |
+
"step": 78500
|
| 2667 |
+
},
|
| 2668 |
+
{
|
| 2669 |
+
"epoch": 0.85,
|
| 2670 |
+
"eval_bleu": 9.9788,
|
| 2671 |
+
"eval_gen_len": 24.3962,
|
| 2672 |
+
"eval_loss": 3.2253217697143555,
|
| 2673 |
+
"eval_meteor": 0.2558,
|
| 2674 |
+
"eval_runtime": 74.6105,
|
| 2675 |
+
"eval_samples_per_second": 6.97,
|
| 2676 |
+
"eval_steps_per_second": 0.871,
|
| 2677 |
+
"step": 78500
|
| 2678 |
+
},
|
| 2679 |
+
{
|
| 2680 |
+
"epoch": 0.85,
|
| 2681 |
+
"learning_rate": 7.3659726494619485e-06,
|
| 2682 |
+
"loss": 1.3623,
|
| 2683 |
+
"step": 79000
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 0.85,
|
| 2687 |
+
"eval_bleu": 10.1533,
|
| 2688 |
+
"eval_gen_len": 24.2692,
|
| 2689 |
+
"eval_loss": 3.225541830062866,
|
| 2690 |
+
"eval_meteor": 0.2516,
|
| 2691 |
+
"eval_runtime": 76.3252,
|
| 2692 |
+
"eval_samples_per_second": 6.813,
|
| 2693 |
+
"eval_steps_per_second": 0.852,
|
| 2694 |
+
"step": 79000
|
| 2695 |
+
},
|
| 2696 |
+
{
|
| 2697 |
+
"epoch": 0.86,
|
| 2698 |
+
"learning_rate": 7.096137033319303e-06,
|
| 2699 |
+
"loss": 1.2651,
|
| 2700 |
+
"step": 79500
|
| 2701 |
+
},
|
| 2702 |
+
{
|
| 2703 |
+
"epoch": 0.86,
|
| 2704 |
+
"eval_bleu": 10.1154,
|
| 2705 |
+
"eval_gen_len": 24.7058,
|
| 2706 |
+
"eval_loss": 3.2230679988861084,
|
| 2707 |
+
"eval_meteor": 0.2535,
|
| 2708 |
+
"eval_runtime": 78.0816,
|
| 2709 |
+
"eval_samples_per_second": 6.66,
|
| 2710 |
+
"eval_steps_per_second": 0.832,
|
| 2711 |
+
"step": 79500
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 0.86,
|
| 2715 |
+
"learning_rate": 6.826301417176656e-06,
|
| 2716 |
+
"loss": 1.3287,
|
| 2717 |
+
"step": 80000
|
| 2718 |
+
},
|
| 2719 |
+
{
|
| 2720 |
+
"epoch": 0.86,
|
| 2721 |
+
"eval_bleu": 10.0863,
|
| 2722 |
+
"eval_gen_len": 24.5962,
|
| 2723 |
+
"eval_loss": 3.2264089584350586,
|
| 2724 |
+
"eval_meteor": 0.2544,
|
| 2725 |
+
"eval_runtime": 100.662,
|
| 2726 |
+
"eval_samples_per_second": 5.166,
|
| 2727 |
+
"eval_steps_per_second": 0.646,
|
| 2728 |
+
"step": 80000
|
| 2729 |
+
},
|
| 2730 |
+
{
|
| 2731 |
+
"epoch": 0.87,
|
| 2732 |
+
"learning_rate": 6.556465801034011e-06,
|
| 2733 |
+
"loss": 1.33,
|
| 2734 |
+
"step": 80500
|
| 2735 |
+
},
|
| 2736 |
+
{
|
| 2737 |
+
"epoch": 0.87,
|
| 2738 |
+
"eval_bleu": 10.1522,
|
| 2739 |
+
"eval_gen_len": 25.4885,
|
| 2740 |
+
"eval_loss": 3.221473217010498,
|
| 2741 |
+
"eval_meteor": 0.2513,
|
| 2742 |
+
"eval_runtime": 70.9756,
|
| 2743 |
+
"eval_samples_per_second": 7.326,
|
| 2744 |
+
"eval_steps_per_second": 0.916,
|
| 2745 |
+
"step": 80500
|
| 2746 |
+
},
|
| 2747 |
+
{
|
| 2748 |
+
"epoch": 0.87,
|
| 2749 |
+
"learning_rate": 6.286630184891365e-06,
|
| 2750 |
+
"loss": 1.2862,
|
| 2751 |
+
"step": 81000
|
| 2752 |
+
},
|
| 2753 |
+
{
|
| 2754 |
+
"epoch": 0.87,
|
| 2755 |
+
"eval_bleu": 10.1425,
|
| 2756 |
+
"eval_gen_len": 25.1442,
|
| 2757 |
+
"eval_loss": 3.226158618927002,
|
| 2758 |
+
"eval_meteor": 0.2538,
|
| 2759 |
+
"eval_runtime": 74.1814,
|
| 2760 |
+
"eval_samples_per_second": 7.01,
|
| 2761 |
+
"eval_steps_per_second": 0.876,
|
| 2762 |
+
"step": 81000
|
| 2763 |
+
},
|
| 2764 |
+
{
|
| 2765 |
+
"epoch": 0.88,
|
| 2766 |
+
"learning_rate": 6.016794568748719e-06,
|
| 2767 |
+
"loss": 1.3738,
|
| 2768 |
+
"step": 81500
|
| 2769 |
+
},
|
| 2770 |
+
{
|
| 2771 |
+
"epoch": 0.88,
|
| 2772 |
+
"eval_bleu": 10.1001,
|
| 2773 |
+
"eval_gen_len": 25.1846,
|
| 2774 |
+
"eval_loss": 3.2149925231933594,
|
| 2775 |
+
"eval_meteor": 0.255,
|
| 2776 |
+
"eval_runtime": 81.4268,
|
| 2777 |
+
"eval_samples_per_second": 6.386,
|
| 2778 |
+
"eval_steps_per_second": 0.798,
|
| 2779 |
+
"step": 81500
|
| 2780 |
+
},
|
| 2781 |
+
{
|
| 2782 |
+
"epoch": 0.89,
|
| 2783 |
+
"learning_rate": 5.746958952606073e-06,
|
| 2784 |
+
"loss": 1.3141,
|
| 2785 |
+
"step": 82000
|
| 2786 |
+
},
|
| 2787 |
+
{
|
| 2788 |
+
"epoch": 0.89,
|
| 2789 |
+
"eval_bleu": 10.2016,
|
| 2790 |
+
"eval_gen_len": 24.6769,
|
| 2791 |
+
"eval_loss": 3.217388391494751,
|
| 2792 |
+
"eval_meteor": 0.2549,
|
| 2793 |
+
"eval_runtime": 75.3151,
|
| 2794 |
+
"eval_samples_per_second": 6.904,
|
| 2795 |
+
"eval_steps_per_second": 0.863,
|
| 2796 |
+
"step": 82000
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 0.89,
|
| 2800 |
+
"learning_rate": 5.4771233364634265e-06,
|
| 2801 |
+
"loss": 1.3326,
|
| 2802 |
+
"step": 82500
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 0.89,
|
| 2806 |
+
"eval_bleu": 10.2847,
|
| 2807 |
+
"eval_gen_len": 24.2038,
|
| 2808 |
+
"eval_loss": 3.2135159969329834,
|
| 2809 |
+
"eval_meteor": 0.2529,
|
| 2810 |
+
"eval_runtime": 74.3155,
|
| 2811 |
+
"eval_samples_per_second": 6.997,
|
| 2812 |
+
"eval_steps_per_second": 0.875,
|
| 2813 |
+
"step": 82500
|
| 2814 |
+
},
|
| 2815 |
+
{
|
| 2816 |
+
"epoch": 0.9,
|
| 2817 |
+
"learning_rate": 5.207287720320781e-06,
|
| 2818 |
+
"loss": 1.3112,
|
| 2819 |
+
"step": 83000
|
| 2820 |
+
},
|
| 2821 |
+
{
|
| 2822 |
+
"epoch": 0.9,
|
| 2823 |
+
"eval_bleu": 10.1374,
|
| 2824 |
+
"eval_gen_len": 24.7308,
|
| 2825 |
+
"eval_loss": 3.2098002433776855,
|
| 2826 |
+
"eval_meteor": 0.2572,
|
| 2827 |
+
"eval_runtime": 71.2774,
|
| 2828 |
+
"eval_samples_per_second": 7.295,
|
| 2829 |
+
"eval_steps_per_second": 0.912,
|
| 2830 |
+
"step": 83000
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 0.9,
|
| 2834 |
+
"learning_rate": 4.937452104178135e-06,
|
| 2835 |
+
"loss": 1.3101,
|
| 2836 |
+
"step": 83500
|
| 2837 |
+
},
|
| 2838 |
+
{
|
| 2839 |
+
"epoch": 0.9,
|
| 2840 |
+
"eval_bleu": 10.0941,
|
| 2841 |
+
"eval_gen_len": 24.6654,
|
| 2842 |
+
"eval_loss": 3.206242084503174,
|
| 2843 |
+
"eval_meteor": 0.256,
|
| 2844 |
+
"eval_runtime": 68.2494,
|
| 2845 |
+
"eval_samples_per_second": 7.619,
|
| 2846 |
+
"eval_steps_per_second": 0.952,
|
| 2847 |
+
"step": 83500
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 0.91,
|
| 2851 |
+
"learning_rate": 4.667616488035489e-06,
|
| 2852 |
+
"loss": 1.3403,
|
| 2853 |
+
"step": 84000
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 0.91,
|
| 2857 |
+
"eval_bleu": 10.1955,
|
| 2858 |
+
"eval_gen_len": 24.8115,
|
| 2859 |
+
"eval_loss": 3.2050940990448,
|
| 2860 |
+
"eval_meteor": 0.2569,
|
| 2861 |
+
"eval_runtime": 67.097,
|
| 2862 |
+
"eval_samples_per_second": 7.75,
|
| 2863 |
+
"eval_steps_per_second": 0.969,
|
| 2864 |
+
"step": 84000
|
| 2865 |
+
},
|
| 2866 |
+
{
|
| 2867 |
+
"epoch": 0.91,
|
| 2868 |
+
"learning_rate": 4.397780871892843e-06,
|
| 2869 |
+
"loss": 1.3651,
|
| 2870 |
+
"step": 84500
|
| 2871 |
+
},
|
| 2872 |
+
{
|
| 2873 |
+
"epoch": 0.91,
|
| 2874 |
+
"eval_bleu": 10.0488,
|
| 2875 |
+
"eval_gen_len": 25.5019,
|
| 2876 |
+
"eval_loss": 3.209751605987549,
|
| 2877 |
+
"eval_meteor": 0.2581,
|
| 2878 |
+
"eval_runtime": 61.2968,
|
| 2879 |
+
"eval_samples_per_second": 8.483,
|
| 2880 |
+
"eval_steps_per_second": 1.06,
|
| 2881 |
+
"step": 84500
|
| 2882 |
+
},
|
| 2883 |
+
{
|
| 2884 |
+
"epoch": 0.92,
|
| 2885 |
+
"learning_rate": 4.127945255750197e-06,
|
| 2886 |
+
"loss": 1.3243,
|
| 2887 |
+
"step": 85000
|
| 2888 |
+
},
|
| 2889 |
+
{
|
| 2890 |
+
"epoch": 0.92,
|
| 2891 |
+
"eval_bleu": 10.1076,
|
| 2892 |
+
"eval_gen_len": 24.7385,
|
| 2893 |
+
"eval_loss": 3.202362298965454,
|
| 2894 |
+
"eval_meteor": 0.2567,
|
| 2895 |
+
"eval_runtime": 68.0409,
|
| 2896 |
+
"eval_samples_per_second": 7.642,
|
| 2897 |
+
"eval_steps_per_second": 0.955,
|
| 2898 |
+
"step": 85000
|
| 2899 |
+
},
|
| 2900 |
+
{
|
| 2901 |
+
"epoch": 0.92,
|
| 2902 |
+
"learning_rate": 3.858109639607551e-06,
|
| 2903 |
+
"loss": 1.3147,
|
| 2904 |
+
"step": 85500
|
| 2905 |
+
},
|
| 2906 |
+
{
|
| 2907 |
+
"epoch": 0.92,
|
| 2908 |
+
"eval_bleu": 10.0392,
|
| 2909 |
+
"eval_gen_len": 25.1269,
|
| 2910 |
+
"eval_loss": 3.2062337398529053,
|
| 2911 |
+
"eval_meteor": 0.2544,
|
| 2912 |
+
"eval_runtime": 71.9112,
|
| 2913 |
+
"eval_samples_per_second": 7.231,
|
| 2914 |
+
"eval_steps_per_second": 0.904,
|
| 2915 |
+
"step": 85500
|
| 2916 |
+
},
|
| 2917 |
+
{
|
| 2918 |
+
"epoch": 0.93,
|
| 2919 |
+
"learning_rate": 3.5882740234649054e-06,
|
| 2920 |
+
"loss": 1.2663,
|
| 2921 |
+
"step": 86000
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 0.93,
|
| 2925 |
+
"eval_bleu": 10.2884,
|
| 2926 |
+
"eval_gen_len": 24.925,
|
| 2927 |
+
"eval_loss": 3.2030699253082275,
|
| 2928 |
+
"eval_meteor": 0.2588,
|
| 2929 |
+
"eval_runtime": 70.7104,
|
| 2930 |
+
"eval_samples_per_second": 7.354,
|
| 2931 |
+
"eval_steps_per_second": 0.919,
|
| 2932 |
+
"step": 86000
|
| 2933 |
+
},
|
| 2934 |
+
{
|
| 2935 |
+
"epoch": 0.93,
|
| 2936 |
+
"learning_rate": 3.3184384073222596e-06,
|
| 2937 |
+
"loss": 1.3497,
|
| 2938 |
+
"step": 86500
|
| 2939 |
+
},
|
| 2940 |
+
{
|
| 2941 |
+
"epoch": 0.93,
|
| 2942 |
+
"eval_bleu": 10.1993,
|
| 2943 |
+
"eval_gen_len": 25.0269,
|
| 2944 |
+
"eval_loss": 3.208573579788208,
|
| 2945 |
+
"eval_meteor": 0.2542,
|
| 2946 |
+
"eval_runtime": 73.9279,
|
| 2947 |
+
"eval_samples_per_second": 7.034,
|
| 2948 |
+
"eval_steps_per_second": 0.879,
|
| 2949 |
+
"step": 86500
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 0.94,
|
| 2953 |
+
"learning_rate": 3.0486027911796137e-06,
|
| 2954 |
+
"loss": 1.3252,
|
| 2955 |
+
"step": 87000
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 0.94,
|
| 2959 |
+
"eval_bleu": 10.1155,
|
| 2960 |
+
"eval_gen_len": 24.7173,
|
| 2961 |
+
"eval_loss": 3.2008321285247803,
|
| 2962 |
+
"eval_meteor": 0.2575,
|
| 2963 |
+
"eval_runtime": 68.983,
|
| 2964 |
+
"eval_samples_per_second": 7.538,
|
| 2965 |
+
"eval_steps_per_second": 0.942,
|
| 2966 |
+
"step": 87000
|
| 2967 |
+
},
|
| 2968 |
+
{
|
| 2969 |
+
"epoch": 0.94,
|
| 2970 |
+
"learning_rate": 2.778767175036968e-06,
|
| 2971 |
+
"loss": 1.277,
|
| 2972 |
+
"step": 87500
|
| 2973 |
+
},
|
| 2974 |
+
{
|
| 2975 |
+
"epoch": 0.94,
|
| 2976 |
+
"eval_bleu": 10.3075,
|
| 2977 |
+
"eval_gen_len": 24.5346,
|
| 2978 |
+
"eval_loss": 3.2011780738830566,
|
| 2979 |
+
"eval_meteor": 0.2592,
|
| 2980 |
+
"eval_runtime": 67.2436,
|
| 2981 |
+
"eval_samples_per_second": 7.733,
|
| 2982 |
+
"eval_steps_per_second": 0.967,
|
| 2983 |
+
"step": 87500
|
| 2984 |
+
},
|
| 2985 |
+
{
|
| 2986 |
+
"epoch": 0.95,
|
| 2987 |
+
"learning_rate": 2.5089315588943217e-06,
|
| 2988 |
+
"loss": 1.3134,
|
| 2989 |
+
"step": 88000
|
| 2990 |
+
},
|
| 2991 |
+
{
|
| 2992 |
+
"epoch": 0.95,
|
| 2993 |
+
"eval_bleu": 10.31,
|
| 2994 |
+
"eval_gen_len": 24.4,
|
| 2995 |
+
"eval_loss": 3.2086844444274902,
|
| 2996 |
+
"eval_meteor": 0.2565,
|
| 2997 |
+
"eval_runtime": 90.4691,
|
| 2998 |
+
"eval_samples_per_second": 5.748,
|
| 2999 |
+
"eval_steps_per_second": 0.718,
|
| 3000 |
+
"step": 88000
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 0.96,
|
| 3004 |
+
"learning_rate": 2.239095942751676e-06,
|
| 3005 |
+
"loss": 1.371,
|
| 3006 |
+
"step": 88500
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 0.96,
|
| 3010 |
+
"eval_bleu": 10.3715,
|
| 3011 |
+
"eval_gen_len": 24.3385,
|
| 3012 |
+
"eval_loss": 3.206083297729492,
|
| 3013 |
+
"eval_meteor": 0.2597,
|
| 3014 |
+
"eval_runtime": 67.5985,
|
| 3015 |
+
"eval_samples_per_second": 7.692,
|
| 3016 |
+
"eval_steps_per_second": 0.962,
|
| 3017 |
+
"step": 88500
|
| 3018 |
+
},
|
| 3019 |
+
{
|
| 3020 |
+
"epoch": 0.96,
|
| 3021 |
+
"learning_rate": 1.96926032660903e-06,
|
| 3022 |
+
"loss": 1.2951,
|
| 3023 |
+
"step": 89000
|
| 3024 |
+
},
|
| 3025 |
+
{
|
| 3026 |
+
"epoch": 0.96,
|
| 3027 |
+
"eval_bleu": 10.415,
|
| 3028 |
+
"eval_gen_len": 24.4481,
|
| 3029 |
+
"eval_loss": 3.2051799297332764,
|
| 3030 |
+
"eval_meteor": 0.2597,
|
| 3031 |
+
"eval_runtime": 70.0734,
|
| 3032 |
+
"eval_samples_per_second": 7.421,
|
| 3033 |
+
"eval_steps_per_second": 0.928,
|
| 3034 |
+
"step": 89000
|
| 3035 |
+
},
|
| 3036 |
+
{
|
| 3037 |
+
"epoch": 0.97,
|
| 3038 |
+
"learning_rate": 1.6994247104663838e-06,
|
| 3039 |
+
"loss": 1.2891,
|
| 3040 |
+
"step": 89500
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 0.97,
|
| 3044 |
+
"eval_bleu": 10.3082,
|
| 3045 |
+
"eval_gen_len": 24.3154,
|
| 3046 |
+
"eval_loss": 3.2073869705200195,
|
| 3047 |
+
"eval_meteor": 0.2566,
|
| 3048 |
+
"eval_runtime": 67.5441,
|
| 3049 |
+
"eval_samples_per_second": 7.699,
|
| 3050 |
+
"eval_steps_per_second": 0.962,
|
| 3051 |
+
"step": 89500
|
| 3052 |
+
},
|
| 3053 |
+
{
|
| 3054 |
+
"epoch": 0.97,
|
| 3055 |
+
"learning_rate": 1.429589094323738e-06,
|
| 3056 |
+
"loss": 1.3057,
|
| 3057 |
+
"step": 90000
|
| 3058 |
+
},
|
| 3059 |
+
{
|
| 3060 |
+
"epoch": 0.97,
|
| 3061 |
+
"eval_bleu": 10.3117,
|
| 3062 |
+
"eval_gen_len": 24.7635,
|
| 3063 |
+
"eval_loss": 3.204622745513916,
|
| 3064 |
+
"eval_meteor": 0.2582,
|
| 3065 |
+
"eval_runtime": 67.7948,
|
| 3066 |
+
"eval_samples_per_second": 7.67,
|
| 3067 |
+
"eval_steps_per_second": 0.959,
|
| 3068 |
+
"step": 90000
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 0.98,
|
| 3072 |
+
"learning_rate": 1.1597534781810922e-06,
|
| 3073 |
+
"loss": 1.294,
|
| 3074 |
+
"step": 90500
|
| 3075 |
+
},
|
| 3076 |
+
{
|
| 3077 |
+
"epoch": 0.98,
|
| 3078 |
+
"eval_bleu": 10.4707,
|
| 3079 |
+
"eval_gen_len": 24.8135,
|
| 3080 |
+
"eval_loss": 3.204620838165283,
|
| 3081 |
+
"eval_meteor": 0.2593,
|
| 3082 |
+
"eval_runtime": 69.6992,
|
| 3083 |
+
"eval_samples_per_second": 7.461,
|
| 3084 |
+
"eval_steps_per_second": 0.933,
|
| 3085 |
+
"step": 90500
|
| 3086 |
+
},
|
| 3087 |
+
{
|
| 3088 |
+
"epoch": 0.98,
|
| 3089 |
+
"learning_rate": 8.899178620384462e-07,
|
| 3090 |
+
"loss": 1.2979,
|
| 3091 |
+
"step": 91000
|
| 3092 |
+
},
|
| 3093 |
+
{
|
| 3094 |
+
"epoch": 0.98,
|
| 3095 |
+
"eval_bleu": 10.179,
|
| 3096 |
+
"eval_gen_len": 24.8058,
|
| 3097 |
+
"eval_loss": 3.2011451721191406,
|
| 3098 |
+
"eval_meteor": 0.2588,
|
| 3099 |
+
"eval_runtime": 76.3465,
|
| 3100 |
+
"eval_samples_per_second": 6.811,
|
| 3101 |
+
"eval_steps_per_second": 0.851,
|
| 3102 |
+
"step": 91000
|
| 3103 |
+
},
|
| 3104 |
+
{
|
| 3105 |
+
"epoch": 0.99,
|
| 3106 |
+
"learning_rate": 6.200822458958003e-07,
|
| 3107 |
+
"loss": 1.3096,
|
| 3108 |
+
"step": 91500
|
| 3109 |
+
},
|
| 3110 |
+
{
|
| 3111 |
+
"epoch": 0.99,
|
| 3112 |
+
"eval_bleu": 10.407,
|
| 3113 |
+
"eval_gen_len": 24.5077,
|
| 3114 |
+
"eval_loss": 3.199586868286133,
|
| 3115 |
+
"eval_meteor": 0.2588,
|
| 3116 |
+
"eval_runtime": 68.5257,
|
| 3117 |
+
"eval_samples_per_second": 7.588,
|
| 3118 |
+
"eval_steps_per_second": 0.949,
|
| 3119 |
+
"step": 91500
|
| 3120 |
+
},
|
| 3121 |
+
{
|
| 3122 |
+
"epoch": 0.99,
|
| 3123 |
+
"learning_rate": 3.502466297531544e-07,
|
| 3124 |
+
"loss": 1.3779,
|
| 3125 |
+
"step": 92000
|
| 3126 |
+
},
|
| 3127 |
+
{
|
| 3128 |
+
"epoch": 0.99,
|
| 3129 |
+
"eval_bleu": 10.2969,
|
| 3130 |
+
"eval_gen_len": 24.6481,
|
| 3131 |
+
"eval_loss": 3.2000153064727783,
|
| 3132 |
+
"eval_meteor": 0.2567,
|
| 3133 |
+
"eval_runtime": 66.7091,
|
| 3134 |
+
"eval_samples_per_second": 7.795,
|
| 3135 |
+
"eval_steps_per_second": 0.974,
|
| 3136 |
+
"step": 92000
|
| 3137 |
+
},
|
| 3138 |
+
{
|
| 3139 |
+
"epoch": 1.0,
|
| 3140 |
+
"learning_rate": 8.041101361050848e-08,
|
| 3141 |
+
"loss": 1.3578,
|
| 3142 |
+
"step": 92500
|
| 3143 |
+
},
|
| 3144 |
+
{
|
| 3145 |
+
"epoch": 1.0,
|
| 3146 |
+
"eval_bleu": 10.295,
|
| 3147 |
+
"eval_gen_len": 24.6423,
|
| 3148 |
+
"eval_loss": 3.2001755237579346,
|
| 3149 |
+
"eval_meteor": 0.2573,
|
| 3150 |
+
"eval_runtime": 74.0674,
|
| 3151 |
+
"eval_samples_per_second": 7.021,
|
| 3152 |
+
"eval_steps_per_second": 0.878,
|
| 3153 |
+
"step": 92500
|
| 3154 |
+
}
|
| 3155 |
+
],
|
| 3156 |
+
"max_steps": 92649,
|
| 3157 |
+
"num_train_epochs": 1,
|
| 3158 |
+
"total_flos": 1.0046721126039552e+16,
|
| 3159 |
+
"trial_name": null,
|
| 3160 |
+
"trial_params": null
|
| 3161 |
+
}
|