Upload trainer_state.json
Browse files- trainer_state.json +1798 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1798 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 99.4,
|
| 5 |
+
"global_step": 9940,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 1.0,
|
| 12 |
+
"learning_rate": 0.00029699999999999996,
|
| 13 |
+
"loss": 2.3572,
|
| 14 |
+
"step": 100
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 1.0,
|
| 18 |
+
"eval_loss": 2.0208146572113037,
|
| 19 |
+
"eval_rouge1": 0.38044177163260395,
|
| 20 |
+
"eval_rouge2": 0.22166452709217227,
|
| 21 |
+
"eval_rougeL": 0.32812293401962783,
|
| 22 |
+
"eval_rougeLsum": 0.3362999547235502,
|
| 23 |
+
"eval_runtime": 101.9509,
|
| 24 |
+
"eval_samples_per_second": 2.619,
|
| 25 |
+
"eval_steps_per_second": 0.657,
|
| 26 |
+
"step": 100
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 2.0,
|
| 30 |
+
"learning_rate": 0.000294,
|
| 31 |
+
"loss": 1.8221,
|
| 32 |
+
"step": 200
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 2.0,
|
| 36 |
+
"eval_loss": 1.9186301231384277,
|
| 37 |
+
"eval_rouge1": 0.37852474610193687,
|
| 38 |
+
"eval_rouge2": 0.22935359793844923,
|
| 39 |
+
"eval_rougeL": 0.33206971678305885,
|
| 40 |
+
"eval_rougeLsum": 0.34036536115022115,
|
| 41 |
+
"eval_runtime": 88.4707,
|
| 42 |
+
"eval_samples_per_second": 3.018,
|
| 43 |
+
"eval_steps_per_second": 0.757,
|
| 44 |
+
"step": 200
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 3.0,
|
| 48 |
+
"learning_rate": 0.00029099999999999997,
|
| 49 |
+
"loss": 1.499,
|
| 50 |
+
"step": 300
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 3.0,
|
| 54 |
+
"eval_loss": 1.8877462148666382,
|
| 55 |
+
"eval_rouge1": 0.38418268500674113,
|
| 56 |
+
"eval_rouge2": 0.23776340434064813,
|
| 57 |
+
"eval_rougeL": 0.3334841887210931,
|
| 58 |
+
"eval_rougeLsum": 0.3427682236088454,
|
| 59 |
+
"eval_runtime": 84.7141,
|
| 60 |
+
"eval_samples_per_second": 3.152,
|
| 61 |
+
"eval_steps_per_second": 0.791,
|
| 62 |
+
"step": 300
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 4.0,
|
| 66 |
+
"learning_rate": 0.00028799999999999995,
|
| 67 |
+
"loss": 1.2417,
|
| 68 |
+
"step": 400
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 4.0,
|
| 72 |
+
"eval_loss": 1.9007868766784668,
|
| 73 |
+
"eval_rouge1": 0.38459815678190157,
|
| 74 |
+
"eval_rouge2": 0.2389841562885784,
|
| 75 |
+
"eval_rougeL": 0.3358997192110219,
|
| 76 |
+
"eval_rougeLsum": 0.3459186341182623,
|
| 77 |
+
"eval_runtime": 87.1466,
|
| 78 |
+
"eval_samples_per_second": 3.064,
|
| 79 |
+
"eval_steps_per_second": 0.769,
|
| 80 |
+
"step": 400
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 5.0,
|
| 84 |
+
"learning_rate": 0.000285,
|
| 85 |
+
"loss": 1.0291,
|
| 86 |
+
"step": 500
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 5.0,
|
| 90 |
+
"eval_loss": 1.9464083909988403,
|
| 91 |
+
"eval_rouge1": 0.3771057538216719,
|
| 92 |
+
"eval_rouge2": 0.23598990815679244,
|
| 93 |
+
"eval_rougeL": 0.33157674303979967,
|
| 94 |
+
"eval_rougeLsum": 0.341814656302837,
|
| 95 |
+
"eval_runtime": 84.8724,
|
| 96 |
+
"eval_samples_per_second": 3.146,
|
| 97 |
+
"eval_steps_per_second": 0.789,
|
| 98 |
+
"step": 500
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 6.0,
|
| 102 |
+
"learning_rate": 0.00028199999999999997,
|
| 103 |
+
"loss": 0.8654,
|
| 104 |
+
"step": 600
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 6.0,
|
| 108 |
+
"eval_loss": 2.028379201889038,
|
| 109 |
+
"eval_rouge1": 0.3765336053931574,
|
| 110 |
+
"eval_rouge2": 0.2346491097022591,
|
| 111 |
+
"eval_rougeL": 0.33228774930217964,
|
| 112 |
+
"eval_rougeLsum": 0.3417265321962749,
|
| 113 |
+
"eval_runtime": 81.6724,
|
| 114 |
+
"eval_samples_per_second": 3.269,
|
| 115 |
+
"eval_steps_per_second": 0.82,
|
| 116 |
+
"step": 600
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 7.0,
|
| 120 |
+
"learning_rate": 0.000279,
|
| 121 |
+
"loss": 0.7288,
|
| 122 |
+
"step": 700
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 7.0,
|
| 126 |
+
"eval_loss": 2.0554444789886475,
|
| 127 |
+
"eval_rouge1": 0.3800786333023515,
|
| 128 |
+
"eval_rouge2": 0.2409070123236104,
|
| 129 |
+
"eval_rougeL": 0.33682162100303614,
|
| 130 |
+
"eval_rougeLsum": 0.34734833881743354,
|
| 131 |
+
"eval_runtime": 73.6883,
|
| 132 |
+
"eval_samples_per_second": 3.623,
|
| 133 |
+
"eval_steps_per_second": 0.909,
|
| 134 |
+
"step": 700
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 8.0,
|
| 138 |
+
"learning_rate": 0.000276,
|
| 139 |
+
"loss": 0.6131,
|
| 140 |
+
"step": 800
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"epoch": 8.0,
|
| 144 |
+
"eval_loss": 2.166259288787842,
|
| 145 |
+
"eval_rouge1": 0.38689909609996354,
|
| 146 |
+
"eval_rouge2": 0.24839504760949166,
|
| 147 |
+
"eval_rougeL": 0.3423017815734907,
|
| 148 |
+
"eval_rougeLsum": 0.35145300780053884,
|
| 149 |
+
"eval_runtime": 84.303,
|
| 150 |
+
"eval_samples_per_second": 3.167,
|
| 151 |
+
"eval_steps_per_second": 0.795,
|
| 152 |
+
"step": 800
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 9.0,
|
| 156 |
+
"learning_rate": 0.00027299999999999997,
|
| 157 |
+
"loss": 0.5199,
|
| 158 |
+
"step": 900
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 9.0,
|
| 162 |
+
"eval_loss": 2.285539388656616,
|
| 163 |
+
"eval_rouge1": 0.3789430717182254,
|
| 164 |
+
"eval_rouge2": 0.24104702341168543,
|
| 165 |
+
"eval_rougeL": 0.33609196315407464,
|
| 166 |
+
"eval_rougeLsum": 0.34419861929575424,
|
| 167 |
+
"eval_runtime": 87.3936,
|
| 168 |
+
"eval_samples_per_second": 3.055,
|
| 169 |
+
"eval_steps_per_second": 0.767,
|
| 170 |
+
"step": 900
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 10.0,
|
| 174 |
+
"learning_rate": 0.00027,
|
| 175 |
+
"loss": 0.4377,
|
| 176 |
+
"step": 1000
|
| 177 |
+
},
|
| 178 |
+
{
|
| 179 |
+
"epoch": 10.0,
|
| 180 |
+
"eval_loss": 2.3414063453674316,
|
| 181 |
+
"eval_rouge1": 0.3804299105262684,
|
| 182 |
+
"eval_rouge2": 0.24134979578604437,
|
| 183 |
+
"eval_rougeL": 0.33465308007661587,
|
| 184 |
+
"eval_rougeLsum": 0.34441539625221235,
|
| 185 |
+
"eval_runtime": 74.2399,
|
| 186 |
+
"eval_samples_per_second": 3.596,
|
| 187 |
+
"eval_steps_per_second": 0.902,
|
| 188 |
+
"step": 1000
|
| 189 |
+
},
|
| 190 |
+
{
|
| 191 |
+
"epoch": 11.0,
|
| 192 |
+
"learning_rate": 0.000267,
|
| 193 |
+
"loss": 0.3774,
|
| 194 |
+
"step": 1100
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 11.0,
|
| 198 |
+
"eval_loss": 2.5906524658203125,
|
| 199 |
+
"eval_rouge1": 0.37488053520037123,
|
| 200 |
+
"eval_rouge2": 0.2325364461265239,
|
| 201 |
+
"eval_rougeL": 0.3284555547843171,
|
| 202 |
+
"eval_rougeLsum": 0.33842206829201854,
|
| 203 |
+
"eval_runtime": 87.4814,
|
| 204 |
+
"eval_samples_per_second": 3.052,
|
| 205 |
+
"eval_steps_per_second": 0.766,
|
| 206 |
+
"step": 1100
|
| 207 |
+
},
|
| 208 |
+
{
|
| 209 |
+
"epoch": 12.0,
|
| 210 |
+
"learning_rate": 0.00026399999999999997,
|
| 211 |
+
"loss": 0.3245,
|
| 212 |
+
"step": 1200
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 12.0,
|
| 216 |
+
"eval_loss": 2.575906276702881,
|
| 217 |
+
"eval_rouge1": 0.3809104895798368,
|
| 218 |
+
"eval_rouge2": 0.2430694084434509,
|
| 219 |
+
"eval_rougeL": 0.339586891707485,
|
| 220 |
+
"eval_rougeLsum": 0.3479448496850858,
|
| 221 |
+
"eval_runtime": 86.2075,
|
| 222 |
+
"eval_samples_per_second": 3.097,
|
| 223 |
+
"eval_steps_per_second": 0.777,
|
| 224 |
+
"step": 1200
|
| 225 |
+
},
|
| 226 |
+
{
|
| 227 |
+
"epoch": 13.0,
|
| 228 |
+
"learning_rate": 0.000261,
|
| 229 |
+
"loss": 0.2838,
|
| 230 |
+
"step": 1300
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 13.0,
|
| 234 |
+
"eval_loss": 2.713919162750244,
|
| 235 |
+
"eval_rouge1": 0.3817118270434364,
|
| 236 |
+
"eval_rouge2": 0.24584042005918305,
|
| 237 |
+
"eval_rougeL": 0.3378501156668257,
|
| 238 |
+
"eval_rougeLsum": 0.34886771837118585,
|
| 239 |
+
"eval_runtime": 88.4266,
|
| 240 |
+
"eval_samples_per_second": 3.019,
|
| 241 |
+
"eval_steps_per_second": 0.758,
|
| 242 |
+
"step": 1300
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 14.0,
|
| 246 |
+
"learning_rate": 0.000258,
|
| 247 |
+
"loss": 0.2468,
|
| 248 |
+
"step": 1400
|
| 249 |
+
},
|
| 250 |
+
{
|
| 251 |
+
"epoch": 14.0,
|
| 252 |
+
"eval_loss": 2.8145313262939453,
|
| 253 |
+
"eval_rouge1": 0.379472522792231,
|
| 254 |
+
"eval_rouge2": 0.24368264109880694,
|
| 255 |
+
"eval_rougeL": 0.3402990243711923,
|
| 256 |
+
"eval_rougeLsum": 0.34789779861372483,
|
| 257 |
+
"eval_runtime": 87.9487,
|
| 258 |
+
"eval_samples_per_second": 3.036,
|
| 259 |
+
"eval_steps_per_second": 0.762,
|
| 260 |
+
"step": 1400
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 15.0,
|
| 264 |
+
"learning_rate": 0.00025499999999999996,
|
| 265 |
+
"loss": 0.2176,
|
| 266 |
+
"step": 1500
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 15.0,
|
| 270 |
+
"eval_loss": 2.9211342334747314,
|
| 271 |
+
"eval_rouge1": 0.38563833059926744,
|
| 272 |
+
"eval_rouge2": 0.25107254535227563,
|
| 273 |
+
"eval_rougeL": 0.3454046659035601,
|
| 274 |
+
"eval_rougeLsum": 0.3550077794346308,
|
| 275 |
+
"eval_runtime": 87.7555,
|
| 276 |
+
"eval_samples_per_second": 3.043,
|
| 277 |
+
"eval_steps_per_second": 0.763,
|
| 278 |
+
"step": 1500
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 16.0,
|
| 282 |
+
"learning_rate": 0.00025199999999999995,
|
| 283 |
+
"loss": 0.1971,
|
| 284 |
+
"step": 1600
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 16.0,
|
| 288 |
+
"eval_loss": 3.035965919494629,
|
| 289 |
+
"eval_rouge1": 0.3748339777411154,
|
| 290 |
+
"eval_rouge2": 0.24127754538743285,
|
| 291 |
+
"eval_rougeL": 0.33399204978041663,
|
| 292 |
+
"eval_rougeLsum": 0.3436267443506058,
|
| 293 |
+
"eval_runtime": 88.2167,
|
| 294 |
+
"eval_samples_per_second": 3.027,
|
| 295 |
+
"eval_steps_per_second": 0.759,
|
| 296 |
+
"step": 1600
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 17.0,
|
| 300 |
+
"learning_rate": 0.000249,
|
| 301 |
+
"loss": 0.1762,
|
| 302 |
+
"step": 1700
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 17.0,
|
| 306 |
+
"eval_loss": 3.1633944511413574,
|
| 307 |
+
"eval_rouge1": 0.3756943708883763,
|
| 308 |
+
"eval_rouge2": 0.2398374976205459,
|
| 309 |
+
"eval_rougeL": 0.3344204501634195,
|
| 310 |
+
"eval_rougeLsum": 0.34415067068109806,
|
| 311 |
+
"eval_runtime": 88.3785,
|
| 312 |
+
"eval_samples_per_second": 3.021,
|
| 313 |
+
"eval_steps_per_second": 0.758,
|
| 314 |
+
"step": 1700
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 18.0,
|
| 318 |
+
"learning_rate": 0.00024599999999999996,
|
| 319 |
+
"loss": 0.1586,
|
| 320 |
+
"step": 1800
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 18.0,
|
| 324 |
+
"eval_loss": 3.361802339553833,
|
| 325 |
+
"eval_rouge1": 0.3771231957637745,
|
| 326 |
+
"eval_rouge2": 0.24367019161252174,
|
| 327 |
+
"eval_rougeL": 0.33712070645495773,
|
| 328 |
+
"eval_rougeLsum": 0.3467573160545896,
|
| 329 |
+
"eval_runtime": 88.6648,
|
| 330 |
+
"eval_samples_per_second": 3.011,
|
| 331 |
+
"eval_steps_per_second": 0.756,
|
| 332 |
+
"step": 1800
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 19.0,
|
| 336 |
+
"learning_rate": 0.000243,
|
| 337 |
+
"loss": 0.1431,
|
| 338 |
+
"step": 1900
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 19.0,
|
| 342 |
+
"eval_loss": 3.339010000228882,
|
| 343 |
+
"eval_rouge1": 0.38323576739072185,
|
| 344 |
+
"eval_rouge2": 0.24826737117741765,
|
| 345 |
+
"eval_rougeL": 0.3410750002762108,
|
| 346 |
+
"eval_rougeLsum": 0.3503168292106818,
|
| 347 |
+
"eval_runtime": 88.2341,
|
| 348 |
+
"eval_samples_per_second": 3.026,
|
| 349 |
+
"eval_steps_per_second": 0.759,
|
| 350 |
+
"step": 1900
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 20.0,
|
| 354 |
+
"learning_rate": 0.00023999999999999998,
|
| 355 |
+
"loss": 0.1369,
|
| 356 |
+
"step": 2000
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 20.0,
|
| 360 |
+
"eval_loss": 3.3738720417022705,
|
| 361 |
+
"eval_rouge1": 0.3803519989007255,
|
| 362 |
+
"eval_rouge2": 0.24324683470771802,
|
| 363 |
+
"eval_rougeL": 0.3413320322258584,
|
| 364 |
+
"eval_rougeLsum": 0.3497926224765554,
|
| 365 |
+
"eval_runtime": 87.7456,
|
| 366 |
+
"eval_samples_per_second": 3.043,
|
| 367 |
+
"eval_steps_per_second": 0.764,
|
| 368 |
+
"step": 2000
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 21.0,
|
| 372 |
+
"learning_rate": 0.000237,
|
| 373 |
+
"loss": 0.1191,
|
| 374 |
+
"step": 2100
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 21.0,
|
| 378 |
+
"eval_loss": 3.4585511684417725,
|
| 379 |
+
"eval_rouge1": 0.3856774466701469,
|
| 380 |
+
"eval_rouge2": 0.24573309119665773,
|
| 381 |
+
"eval_rougeL": 0.340491594204651,
|
| 382 |
+
"eval_rougeLsum": 0.35036530615920203,
|
| 383 |
+
"eval_runtime": 84.6104,
|
| 384 |
+
"eval_samples_per_second": 3.156,
|
| 385 |
+
"eval_steps_per_second": 0.792,
|
| 386 |
+
"step": 2100
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 22.0,
|
| 390 |
+
"learning_rate": 0.000234,
|
| 391 |
+
"loss": 0.1123,
|
| 392 |
+
"step": 2200
|
| 393 |
+
},
|
| 394 |
+
{
|
| 395 |
+
"epoch": 22.0,
|
| 396 |
+
"eval_loss": 3.4547746181488037,
|
| 397 |
+
"eval_rouge1": 0.379890674444173,
|
| 398 |
+
"eval_rouge2": 0.23951241543587243,
|
| 399 |
+
"eval_rougeL": 0.3381674920029052,
|
| 400 |
+
"eval_rougeLsum": 0.34605025320246763,
|
| 401 |
+
"eval_runtime": 86.4576,
|
| 402 |
+
"eval_samples_per_second": 3.088,
|
| 403 |
+
"eval_steps_per_second": 0.775,
|
| 404 |
+
"step": 2200
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 23.0,
|
| 408 |
+
"learning_rate": 0.00023099999999999998,
|
| 409 |
+
"loss": 0.1006,
|
| 410 |
+
"step": 2300
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"epoch": 23.0,
|
| 414 |
+
"eval_loss": 3.563164234161377,
|
| 415 |
+
"eval_rouge1": 0.383309277836589,
|
| 416 |
+
"eval_rouge2": 0.24671832804779853,
|
| 417 |
+
"eval_rougeL": 0.3418864003163312,
|
| 418 |
+
"eval_rougeLsum": 0.3510389909721277,
|
| 419 |
+
"eval_runtime": 86.7629,
|
| 420 |
+
"eval_samples_per_second": 3.077,
|
| 421 |
+
"eval_steps_per_second": 0.772,
|
| 422 |
+
"step": 2300
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 24.0,
|
| 426 |
+
"learning_rate": 0.00022799999999999999,
|
| 427 |
+
"loss": 0.0946,
|
| 428 |
+
"step": 2400
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 24.0,
|
| 432 |
+
"eval_loss": 3.452104091644287,
|
| 433 |
+
"eval_rouge1": 0.38778417244680397,
|
| 434 |
+
"eval_rouge2": 0.24864493001126428,
|
| 435 |
+
"eval_rougeL": 0.34531411329508677,
|
| 436 |
+
"eval_rougeLsum": 0.3544382672740283,
|
| 437 |
+
"eval_runtime": 81.1305,
|
| 438 |
+
"eval_samples_per_second": 3.291,
|
| 439 |
+
"eval_steps_per_second": 0.826,
|
| 440 |
+
"step": 2400
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 25.0,
|
| 444 |
+
"learning_rate": 0.000225,
|
| 445 |
+
"loss": 0.0852,
|
| 446 |
+
"step": 2500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 25.0,
|
| 450 |
+
"eval_loss": 3.722707509994507,
|
| 451 |
+
"eval_rouge1": 0.38544627790159214,
|
| 452 |
+
"eval_rouge2": 0.24835200716378503,
|
| 453 |
+
"eval_rougeL": 0.34171087371739595,
|
| 454 |
+
"eval_rougeLsum": 0.35119063257011257,
|
| 455 |
+
"eval_runtime": 88.2083,
|
| 456 |
+
"eval_samples_per_second": 3.027,
|
| 457 |
+
"eval_steps_per_second": 0.76,
|
| 458 |
+
"step": 2500
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 26.0,
|
| 462 |
+
"learning_rate": 0.00022199999999999998,
|
| 463 |
+
"loss": 0.0819,
|
| 464 |
+
"step": 2600
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 26.0,
|
| 468 |
+
"eval_loss": 3.700070858001709,
|
| 469 |
+
"eval_rouge1": 0.3821803340000923,
|
| 470 |
+
"eval_rouge2": 0.24725796483857881,
|
| 471 |
+
"eval_rougeL": 0.34338516151002396,
|
| 472 |
+
"eval_rougeLsum": 0.35257138073899147,
|
| 473 |
+
"eval_runtime": 83.8321,
|
| 474 |
+
"eval_samples_per_second": 3.185,
|
| 475 |
+
"eval_steps_per_second": 0.799,
|
| 476 |
+
"step": 2600
|
| 477 |
+
},
|
| 478 |
+
{
|
| 479 |
+
"epoch": 27.0,
|
| 480 |
+
"learning_rate": 0.00021899999999999998,
|
| 481 |
+
"loss": 0.076,
|
| 482 |
+
"step": 2700
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 27.0,
|
| 486 |
+
"eval_loss": 3.705507278442383,
|
| 487 |
+
"eval_rouge1": 0.38289988903470296,
|
| 488 |
+
"eval_rouge2": 0.2435924255504497,
|
| 489 |
+
"eval_rougeL": 0.34095899111986594,
|
| 490 |
+
"eval_rougeLsum": 0.35027734448953907,
|
| 491 |
+
"eval_runtime": 84.7019,
|
| 492 |
+
"eval_samples_per_second": 3.152,
|
| 493 |
+
"eval_steps_per_second": 0.791,
|
| 494 |
+
"step": 2700
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 28.0,
|
| 498 |
+
"learning_rate": 0.00021599999999999996,
|
| 499 |
+
"loss": 0.0737,
|
| 500 |
+
"step": 2800
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 28.0,
|
| 504 |
+
"eval_loss": 3.8778345584869385,
|
| 505 |
+
"eval_rouge1": 0.3863238066090623,
|
| 506 |
+
"eval_rouge2": 0.24626416138837423,
|
| 507 |
+
"eval_rougeL": 0.33940239308122,
|
| 508 |
+
"eval_rougeLsum": 0.35008151797367537,
|
| 509 |
+
"eval_runtime": 88.4609,
|
| 510 |
+
"eval_samples_per_second": 3.018,
|
| 511 |
+
"eval_steps_per_second": 0.757,
|
| 512 |
+
"step": 2800
|
| 513 |
+
},
|
| 514 |
+
{
|
| 515 |
+
"epoch": 29.0,
|
| 516 |
+
"learning_rate": 0.00021299999999999997,
|
| 517 |
+
"loss": 0.071,
|
| 518 |
+
"step": 2900
|
| 519 |
+
},
|
| 520 |
+
{
|
| 521 |
+
"epoch": 29.0,
|
| 522 |
+
"eval_loss": 3.7274887561798096,
|
| 523 |
+
"eval_rouge1": 0.38452724844736197,
|
| 524 |
+
"eval_rouge2": 0.24655003737302858,
|
| 525 |
+
"eval_rougeL": 0.34600558721773045,
|
| 526 |
+
"eval_rougeLsum": 0.3548096862359722,
|
| 527 |
+
"eval_runtime": 87.9179,
|
| 528 |
+
"eval_samples_per_second": 3.037,
|
| 529 |
+
"eval_steps_per_second": 0.762,
|
| 530 |
+
"step": 2900
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 30.0,
|
| 534 |
+
"learning_rate": 0.00020999999999999998,
|
| 535 |
+
"loss": 0.0651,
|
| 536 |
+
"step": 3000
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 30.0,
|
| 540 |
+
"eval_loss": 3.9305973052978516,
|
| 541 |
+
"eval_rouge1": 0.38263859087914487,
|
| 542 |
+
"eval_rouge2": 0.24388219812962653,
|
| 543 |
+
"eval_rougeL": 0.33932726653150014,
|
| 544 |
+
"eval_rougeLsum": 0.350347980074978,
|
| 545 |
+
"eval_runtime": 86.4392,
|
| 546 |
+
"eval_samples_per_second": 3.089,
|
| 547 |
+
"eval_steps_per_second": 0.775,
|
| 548 |
+
"step": 3000
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 31.0,
|
| 552 |
+
"learning_rate": 0.00020699999999999996,
|
| 553 |
+
"loss": 0.059,
|
| 554 |
+
"step": 3100
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 31.0,
|
| 558 |
+
"eval_loss": 3.8749561309814453,
|
| 559 |
+
"eval_rouge1": 0.3838656202146266,
|
| 560 |
+
"eval_rouge2": 0.24797544855564618,
|
| 561 |
+
"eval_rougeL": 0.3455085604915118,
|
| 562 |
+
"eval_rougeLsum": 0.35485772003639693,
|
| 563 |
+
"eval_runtime": 80.6768,
|
| 564 |
+
"eval_samples_per_second": 3.31,
|
| 565 |
+
"eval_steps_per_second": 0.83,
|
| 566 |
+
"step": 3100
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 32.0,
|
| 570 |
+
"learning_rate": 0.000204,
|
| 571 |
+
"loss": 0.0575,
|
| 572 |
+
"step": 3200
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 32.0,
|
| 576 |
+
"eval_loss": 4.035264492034912,
|
| 577 |
+
"eval_rouge1": 0.3836411899648502,
|
| 578 |
+
"eval_rouge2": 0.25059245453298623,
|
| 579 |
+
"eval_rougeL": 0.3411469702531179,
|
| 580 |
+
"eval_rougeLsum": 0.3521664869788945,
|
| 581 |
+
"eval_runtime": 88.2658,
|
| 582 |
+
"eval_samples_per_second": 3.025,
|
| 583 |
+
"eval_steps_per_second": 0.759,
|
| 584 |
+
"step": 3200
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 33.0,
|
| 588 |
+
"learning_rate": 0.000201,
|
| 589 |
+
"loss": 0.0497,
|
| 590 |
+
"step": 3300
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 33.0,
|
| 594 |
+
"eval_loss": 4.112032890319824,
|
| 595 |
+
"eval_rouge1": 0.38502854106397544,
|
| 596 |
+
"eval_rouge2": 0.24923225337454086,
|
| 597 |
+
"eval_rougeL": 0.341407070704573,
|
| 598 |
+
"eval_rougeLsum": 0.35156091483395113,
|
| 599 |
+
"eval_runtime": 84.7807,
|
| 600 |
+
"eval_samples_per_second": 3.149,
|
| 601 |
+
"eval_steps_per_second": 0.79,
|
| 602 |
+
"step": 3300
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 34.0,
|
| 606 |
+
"learning_rate": 0.000198,
|
| 607 |
+
"loss": 0.0501,
|
| 608 |
+
"step": 3400
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 34.0,
|
| 612 |
+
"eval_loss": 3.95038104057312,
|
| 613 |
+
"eval_rouge1": 0.38531511762507886,
|
| 614 |
+
"eval_rouge2": 0.25053367172713387,
|
| 615 |
+
"eval_rougeL": 0.341979636574758,
|
| 616 |
+
"eval_rougeLsum": 0.3530873088985891,
|
| 617 |
+
"eval_runtime": 85.2338,
|
| 618 |
+
"eval_samples_per_second": 3.133,
|
| 619 |
+
"eval_steps_per_second": 0.786,
|
| 620 |
+
"step": 3400
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 35.0,
|
| 624 |
+
"learning_rate": 0.000195,
|
| 625 |
+
"loss": 0.047,
|
| 626 |
+
"step": 3500
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 35.0,
|
| 630 |
+
"eval_loss": 4.134089469909668,
|
| 631 |
+
"eval_rouge1": 0.38128230648241745,
|
| 632 |
+
"eval_rouge2": 0.2492424888975831,
|
| 633 |
+
"eval_rougeL": 0.34103290544972104,
|
| 634 |
+
"eval_rougeLsum": 0.3516319313437394,
|
| 635 |
+
"eval_runtime": 88.3744,
|
| 636 |
+
"eval_samples_per_second": 3.021,
|
| 637 |
+
"eval_steps_per_second": 0.758,
|
| 638 |
+
"step": 3500
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 36.0,
|
| 642 |
+
"learning_rate": 0.00019199999999999998,
|
| 643 |
+
"loss": 0.0453,
|
| 644 |
+
"step": 3600
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 36.0,
|
| 648 |
+
"eval_loss": 4.057723045349121,
|
| 649 |
+
"eval_rouge1": 0.3846704975418777,
|
| 650 |
+
"eval_rouge2": 0.2541194854411035,
|
| 651 |
+
"eval_rougeL": 0.3458972635607298,
|
| 652 |
+
"eval_rougeLsum": 0.3547659584174555,
|
| 653 |
+
"eval_runtime": 85.1914,
|
| 654 |
+
"eval_samples_per_second": 3.134,
|
| 655 |
+
"eval_steps_per_second": 0.786,
|
| 656 |
+
"step": 3600
|
| 657 |
+
},
|
| 658 |
+
{
|
| 659 |
+
"epoch": 37.0,
|
| 660 |
+
"learning_rate": 0.00018899999999999999,
|
| 661 |
+
"loss": 0.0462,
|
| 662 |
+
"step": 3700
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 37.0,
|
| 666 |
+
"eval_loss": 4.16575288772583,
|
| 667 |
+
"eval_rouge1": 0.38354434412263094,
|
| 668 |
+
"eval_rouge2": 0.2510921708756821,
|
| 669 |
+
"eval_rougeL": 0.34240824035427875,
|
| 670 |
+
"eval_rougeLsum": 0.35173537331908256,
|
| 671 |
+
"eval_runtime": 87.6215,
|
| 672 |
+
"eval_samples_per_second": 3.047,
|
| 673 |
+
"eval_steps_per_second": 0.765,
|
| 674 |
+
"step": 3700
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 38.0,
|
| 678 |
+
"learning_rate": 0.000186,
|
| 679 |
+
"loss": 0.0424,
|
| 680 |
+
"step": 3800
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"epoch": 38.0,
|
| 684 |
+
"eval_loss": 4.315454483032227,
|
| 685 |
+
"eval_rouge1": 0.38423210902707555,
|
| 686 |
+
"eval_rouge2": 0.24894255969353363,
|
| 687 |
+
"eval_rougeL": 0.342343883008779,
|
| 688 |
+
"eval_rougeLsum": 0.3524699587337098,
|
| 689 |
+
"eval_runtime": 87.9865,
|
| 690 |
+
"eval_samples_per_second": 3.035,
|
| 691 |
+
"eval_steps_per_second": 0.761,
|
| 692 |
+
"step": 3800
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 39.0,
|
| 696 |
+
"learning_rate": 0.00018299999999999998,
|
| 697 |
+
"loss": 0.0412,
|
| 698 |
+
"step": 3900
|
| 699 |
+
},
|
| 700 |
+
{
|
| 701 |
+
"epoch": 39.0,
|
| 702 |
+
"eval_loss": 4.139455318450928,
|
| 703 |
+
"eval_rouge1": 0.3888396904741606,
|
| 704 |
+
"eval_rouge2": 0.25152424393630335,
|
| 705 |
+
"eval_rougeL": 0.3452689238846711,
|
| 706 |
+
"eval_rougeLsum": 0.3565387881051132,
|
| 707 |
+
"eval_runtime": 88.7657,
|
| 708 |
+
"eval_samples_per_second": 3.008,
|
| 709 |
+
"eval_steps_per_second": 0.755,
|
| 710 |
+
"step": 3900
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"epoch": 40.0,
|
| 714 |
+
"learning_rate": 0.00017999999999999998,
|
| 715 |
+
"loss": 0.0405,
|
| 716 |
+
"step": 4000
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 40.0,
|
| 720 |
+
"eval_loss": 4.292513847351074,
|
| 721 |
+
"eval_rouge1": 0.3849173670043722,
|
| 722 |
+
"eval_rouge2": 0.2516518779916327,
|
| 723 |
+
"eval_rougeL": 0.34160286102286824,
|
| 724 |
+
"eval_rougeLsum": 0.35354307416998204,
|
| 725 |
+
"eval_runtime": 87.0473,
|
| 726 |
+
"eval_samples_per_second": 3.067,
|
| 727 |
+
"eval_steps_per_second": 0.77,
|
| 728 |
+
"step": 4000
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 41.0,
|
| 732 |
+
"learning_rate": 0.00017699999999999997,
|
| 733 |
+
"loss": 0.0337,
|
| 734 |
+
"step": 4100
|
| 735 |
+
},
|
| 736 |
+
{
|
| 737 |
+
"epoch": 41.0,
|
| 738 |
+
"eval_loss": 4.333091735839844,
|
| 739 |
+
"eval_rouge1": 0.38820912557719256,
|
| 740 |
+
"eval_rouge2": 0.2534094086585099,
|
| 741 |
+
"eval_rougeL": 0.3455980647738194,
|
| 742 |
+
"eval_rougeLsum": 0.3547951760533251,
|
| 743 |
+
"eval_runtime": 88.6931,
|
| 744 |
+
"eval_samples_per_second": 3.01,
|
| 745 |
+
"eval_steps_per_second": 0.755,
|
| 746 |
+
"step": 4100
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 42.0,
|
| 750 |
+
"learning_rate": 0.00017399999999999997,
|
| 751 |
+
"loss": 0.034,
|
| 752 |
+
"step": 4200
|
| 753 |
+
},
|
| 754 |
+
{
|
| 755 |
+
"epoch": 42.0,
|
| 756 |
+
"eval_loss": 4.243014335632324,
|
| 757 |
+
"eval_rouge1": 0.3856953885948385,
|
| 758 |
+
"eval_rouge2": 0.2513949176068939,
|
| 759 |
+
"eval_rougeL": 0.34297989861385025,
|
| 760 |
+
"eval_rougeLsum": 0.352859061364724,
|
| 761 |
+
"eval_runtime": 88.2977,
|
| 762 |
+
"eval_samples_per_second": 3.024,
|
| 763 |
+
"eval_steps_per_second": 0.759,
|
| 764 |
+
"step": 4200
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 43.0,
|
| 768 |
+
"learning_rate": 0.00017099999999999998,
|
| 769 |
+
"loss": 0.0352,
|
| 770 |
+
"step": 4300
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 43.0,
|
| 774 |
+
"eval_loss": 4.183932304382324,
|
| 775 |
+
"eval_rouge1": 0.3806969958693784,
|
| 776 |
+
"eval_rouge2": 0.2473825570846937,
|
| 777 |
+
"eval_rougeL": 0.33932671729837177,
|
| 778 |
+
"eval_rougeLsum": 0.34894151771308557,
|
| 779 |
+
"eval_runtime": 80.31,
|
| 780 |
+
"eval_samples_per_second": 3.325,
|
| 781 |
+
"eval_steps_per_second": 0.834,
|
| 782 |
+
"step": 4300
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 44.0,
|
| 786 |
+
"learning_rate": 0.000168,
|
| 787 |
+
"loss": 0.0324,
|
| 788 |
+
"step": 4400
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 44.0,
|
| 792 |
+
"eval_loss": 4.373414039611816,
|
| 793 |
+
"eval_rouge1": 0.3786215226813438,
|
| 794 |
+
"eval_rouge2": 0.2465198631844377,
|
| 795 |
+
"eval_rougeL": 0.33922712749886663,
|
| 796 |
+
"eval_rougeLsum": 0.34917815226425697,
|
| 797 |
+
"eval_runtime": 77.6446,
|
| 798 |
+
"eval_samples_per_second": 3.439,
|
| 799 |
+
"eval_steps_per_second": 0.863,
|
| 800 |
+
"step": 4400
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 45.0,
|
| 804 |
+
"learning_rate": 0.000165,
|
| 805 |
+
"loss": 0.0286,
|
| 806 |
+
"step": 4500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 45.0,
|
| 810 |
+
"eval_loss": 4.281849384307861,
|
| 811 |
+
"eval_rouge1": 0.38349120769596134,
|
| 812 |
+
"eval_rouge2": 0.2492446761527376,
|
| 813 |
+
"eval_rougeL": 0.34069084890798684,
|
| 814 |
+
"eval_rougeLsum": 0.35114847864272203,
|
| 815 |
+
"eval_runtime": 88.42,
|
| 816 |
+
"eval_samples_per_second": 3.02,
|
| 817 |
+
"eval_steps_per_second": 0.758,
|
| 818 |
+
"step": 4500
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 46.0,
|
| 822 |
+
"learning_rate": 0.000162,
|
| 823 |
+
"loss": 0.0282,
|
| 824 |
+
"step": 4600
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 46.0,
|
| 828 |
+
"eval_loss": 4.2815632820129395,
|
| 829 |
+
"eval_rouge1": 0.3826976100476265,
|
| 830 |
+
"eval_rouge2": 0.2509544107400279,
|
| 831 |
+
"eval_rougeL": 0.34278980646197255,
|
| 832 |
+
"eval_rougeLsum": 0.3538927647811483,
|
| 833 |
+
"eval_runtime": 85.7947,
|
| 834 |
+
"eval_samples_per_second": 3.112,
|
| 835 |
+
"eval_steps_per_second": 0.781,
|
| 836 |
+
"step": 4600
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 47.0,
|
| 840 |
+
"learning_rate": 0.000159,
|
| 841 |
+
"loss": 0.028,
|
| 842 |
+
"step": 4700
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 47.0,
|
| 846 |
+
"eval_loss": 4.38587760925293,
|
| 847 |
+
"eval_rouge1": 0.3849127298046784,
|
| 848 |
+
"eval_rouge2": 0.24789381337766286,
|
| 849 |
+
"eval_rougeL": 0.34234361939069524,
|
| 850 |
+
"eval_rougeLsum": 0.35325038564459454,
|
| 851 |
+
"eval_runtime": 76.1287,
|
| 852 |
+
"eval_samples_per_second": 3.507,
|
| 853 |
+
"eval_steps_per_second": 0.88,
|
| 854 |
+
"step": 4700
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 48.0,
|
| 858 |
+
"learning_rate": 0.000156,
|
| 859 |
+
"loss": 0.0273,
|
| 860 |
+
"step": 4800
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 48.0,
|
| 864 |
+
"eval_loss": 4.371572017669678,
|
| 865 |
+
"eval_rouge1": 0.38145021792566747,
|
| 866 |
+
"eval_rouge2": 0.2455942994201823,
|
| 867 |
+
"eval_rougeL": 0.33902654428346374,
|
| 868 |
+
"eval_rougeLsum": 0.3506993238959008,
|
| 869 |
+
"eval_runtime": 85.6829,
|
| 870 |
+
"eval_samples_per_second": 3.116,
|
| 871 |
+
"eval_steps_per_second": 0.782,
|
| 872 |
+
"step": 4800
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 49.0,
|
| 876 |
+
"learning_rate": 0.00015299999999999998,
|
| 877 |
+
"loss": 0.0242,
|
| 878 |
+
"step": 4900
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 49.0,
|
| 882 |
+
"eval_loss": 4.31757116317749,
|
| 883 |
+
"eval_rouge1": 0.3823041816880143,
|
| 884 |
+
"eval_rouge2": 0.2482259591949097,
|
| 885 |
+
"eval_rougeL": 0.3423574705681002,
|
| 886 |
+
"eval_rougeLsum": 0.3529132375278098,
|
| 887 |
+
"eval_runtime": 85.4774,
|
| 888 |
+
"eval_samples_per_second": 3.124,
|
| 889 |
+
"eval_steps_per_second": 0.784,
|
| 890 |
+
"step": 4900
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 50.0,
|
| 894 |
+
"learning_rate": 0.00015,
|
| 895 |
+
"loss": 0.0245,
|
| 896 |
+
"step": 5000
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 50.0,
|
| 900 |
+
"eval_loss": 4.483373641967773,
|
| 901 |
+
"eval_rouge1": 0.38129727161097815,
|
| 902 |
+
"eval_rouge2": 0.24693927554416154,
|
| 903 |
+
"eval_rougeL": 0.3404625026445964,
|
| 904 |
+
"eval_rougeLsum": 0.3512449096988902,
|
| 905 |
+
"eval_runtime": 86.2171,
|
| 906 |
+
"eval_samples_per_second": 3.097,
|
| 907 |
+
"eval_steps_per_second": 0.777,
|
| 908 |
+
"step": 5000
|
| 909 |
+
},
|
| 910 |
+
{
|
| 911 |
+
"epoch": 51.0,
|
| 912 |
+
"learning_rate": 0.000147,
|
| 913 |
+
"loss": 0.0233,
|
| 914 |
+
"step": 5100
|
| 915 |
+
},
|
| 916 |
+
{
|
| 917 |
+
"epoch": 51.0,
|
| 918 |
+
"eval_loss": 4.312454700469971,
|
| 919 |
+
"eval_rouge1": 0.3858793124229507,
|
| 920 |
+
"eval_rouge2": 0.2506880394304446,
|
| 921 |
+
"eval_rougeL": 0.3432986381645482,
|
| 922 |
+
"eval_rougeLsum": 0.35437517778600014,
|
| 923 |
+
"eval_runtime": 85.0765,
|
| 924 |
+
"eval_samples_per_second": 3.138,
|
| 925 |
+
"eval_steps_per_second": 0.788,
|
| 926 |
+
"step": 5100
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 52.0,
|
| 930 |
+
"learning_rate": 0.00014399999999999998,
|
| 931 |
+
"loss": 0.0226,
|
| 932 |
+
"step": 5200
|
| 933 |
+
},
|
| 934 |
+
{
|
| 935 |
+
"epoch": 52.0,
|
| 936 |
+
"eval_loss": 4.3415327072143555,
|
| 937 |
+
"eval_rouge1": 0.3857470971280717,
|
| 938 |
+
"eval_rouge2": 0.24913722198218521,
|
| 939 |
+
"eval_rougeL": 0.3436699140258663,
|
| 940 |
+
"eval_rougeLsum": 0.3530913028701871,
|
| 941 |
+
"eval_runtime": 88.4089,
|
| 942 |
+
"eval_samples_per_second": 3.02,
|
| 943 |
+
"eval_steps_per_second": 0.758,
|
| 944 |
+
"step": 5200
|
| 945 |
+
},
|
| 946 |
+
{
|
| 947 |
+
"epoch": 53.0,
|
| 948 |
+
"learning_rate": 0.00014099999999999998,
|
| 949 |
+
"loss": 0.0229,
|
| 950 |
+
"step": 5300
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"epoch": 53.0,
|
| 954 |
+
"eval_loss": 4.483485698699951,
|
| 955 |
+
"eval_rouge1": 0.38253789740936506,
|
| 956 |
+
"eval_rouge2": 0.2479324630750173,
|
| 957 |
+
"eval_rougeL": 0.34339596051490895,
|
| 958 |
+
"eval_rougeLsum": 0.3517531826655991,
|
| 959 |
+
"eval_runtime": 88.363,
|
| 960 |
+
"eval_samples_per_second": 3.022,
|
| 961 |
+
"eval_steps_per_second": 0.758,
|
| 962 |
+
"step": 5300
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 54.0,
|
| 966 |
+
"learning_rate": 0.000138,
|
| 967 |
+
"loss": 0.0205,
|
| 968 |
+
"step": 5400
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 54.0,
|
| 972 |
+
"eval_loss": 4.5731024742126465,
|
| 973 |
+
"eval_rouge1": 0.3844011329346221,
|
| 974 |
+
"eval_rouge2": 0.24959249662582936,
|
| 975 |
+
"eval_rougeL": 0.3437539050438877,
|
| 976 |
+
"eval_rougeLsum": 0.3524112747853908,
|
| 977 |
+
"eval_runtime": 87.6537,
|
| 978 |
+
"eval_samples_per_second": 3.046,
|
| 979 |
+
"eval_steps_per_second": 0.764,
|
| 980 |
+
"step": 5400
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 55.0,
|
| 984 |
+
"learning_rate": 0.000135,
|
| 985 |
+
"loss": 0.0194,
|
| 986 |
+
"step": 5500
|
| 987 |
+
},
|
| 988 |
+
{
|
| 989 |
+
"epoch": 55.0,
|
| 990 |
+
"eval_loss": 4.562352657318115,
|
| 991 |
+
"eval_rouge1": 0.37834930186879323,
|
| 992 |
+
"eval_rouge2": 0.24392421472693115,
|
| 993 |
+
"eval_rougeL": 0.33801495271361487,
|
| 994 |
+
"eval_rougeLsum": 0.34801929824525824,
|
| 995 |
+
"eval_runtime": 74.8854,
|
| 996 |
+
"eval_samples_per_second": 3.565,
|
| 997 |
+
"eval_steps_per_second": 0.895,
|
| 998 |
+
"step": 5500
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 56.0,
|
| 1002 |
+
"learning_rate": 0.00013199999999999998,
|
| 1003 |
+
"loss": 0.0195,
|
| 1004 |
+
"step": 5600
|
| 1005 |
+
},
|
| 1006 |
+
{
|
| 1007 |
+
"epoch": 56.0,
|
| 1008 |
+
"eval_loss": 4.631711483001709,
|
| 1009 |
+
"eval_rouge1": 0.38562125740872527,
|
| 1010 |
+
"eval_rouge2": 0.25396377473858134,
|
| 1011 |
+
"eval_rougeL": 0.34647449965497257,
|
| 1012 |
+
"eval_rougeLsum": 0.3559086726171456,
|
| 1013 |
+
"eval_runtime": 80.6509,
|
| 1014 |
+
"eval_samples_per_second": 3.311,
|
| 1015 |
+
"eval_steps_per_second": 0.831,
|
| 1016 |
+
"step": 5600
|
| 1017 |
+
},
|
| 1018 |
+
{
|
| 1019 |
+
"epoch": 57.0,
|
| 1020 |
+
"learning_rate": 0.000129,
|
| 1021 |
+
"loss": 0.0187,
|
| 1022 |
+
"step": 5700
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 57.0,
|
| 1026 |
+
"eval_loss": 4.58750581741333,
|
| 1027 |
+
"eval_rouge1": 0.3810792743714226,
|
| 1028 |
+
"eval_rouge2": 0.24677881916117747,
|
| 1029 |
+
"eval_rougeL": 0.3413798314429182,
|
| 1030 |
+
"eval_rougeLsum": 0.3512899517237632,
|
| 1031 |
+
"eval_runtime": 87.4748,
|
| 1032 |
+
"eval_samples_per_second": 3.052,
|
| 1033 |
+
"eval_steps_per_second": 0.766,
|
| 1034 |
+
"step": 5700
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 58.0,
|
| 1038 |
+
"learning_rate": 0.00012599999999999997,
|
| 1039 |
+
"loss": 0.0184,
|
| 1040 |
+
"step": 5800
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 58.0,
|
| 1044 |
+
"eval_loss": 4.566098213195801,
|
| 1045 |
+
"eval_rouge1": 0.3816478906137458,
|
| 1046 |
+
"eval_rouge2": 0.24638333085125486,
|
| 1047 |
+
"eval_rougeL": 0.34077211167545884,
|
| 1048 |
+
"eval_rougeLsum": 0.3515136659487619,
|
| 1049 |
+
"eval_runtime": 88.8228,
|
| 1050 |
+
"eval_samples_per_second": 3.006,
|
| 1051 |
+
"eval_steps_per_second": 0.754,
|
| 1052 |
+
"step": 5800
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 59.0,
|
| 1056 |
+
"learning_rate": 0.00012299999999999998,
|
| 1057 |
+
"loss": 0.0181,
|
| 1058 |
+
"step": 5900
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 59.0,
|
| 1062 |
+
"eval_loss": 4.4710798263549805,
|
| 1063 |
+
"eval_rouge1": 0.38077163264424585,
|
| 1064 |
+
"eval_rouge2": 0.24804038660323457,
|
| 1065 |
+
"eval_rougeL": 0.3439666979964766,
|
| 1066 |
+
"eval_rougeLsum": 0.3531348236604766,
|
| 1067 |
+
"eval_runtime": 85.4002,
|
| 1068 |
+
"eval_samples_per_second": 3.126,
|
| 1069 |
+
"eval_steps_per_second": 0.785,
|
| 1070 |
+
"step": 5900
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 60.0,
|
| 1074 |
+
"learning_rate": 0.00011999999999999999,
|
| 1075 |
+
"loss": 0.0191,
|
| 1076 |
+
"step": 6000
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 60.0,
|
| 1080 |
+
"eval_loss": 4.539062023162842,
|
| 1081 |
+
"eval_rouge1": 0.38696808219881823,
|
| 1082 |
+
"eval_rouge2": 0.25247148225241317,
|
| 1083 |
+
"eval_rougeL": 0.34520470532094527,
|
| 1084 |
+
"eval_rougeLsum": 0.3549651652722977,
|
| 1085 |
+
"eval_runtime": 85.8906,
|
| 1086 |
+
"eval_samples_per_second": 3.109,
|
| 1087 |
+
"eval_steps_per_second": 0.78,
|
| 1088 |
+
"step": 6000
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 61.0,
|
| 1092 |
+
"learning_rate": 0.000117,
|
| 1093 |
+
"loss": 0.0159,
|
| 1094 |
+
"step": 6100
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 61.0,
|
| 1098 |
+
"eval_loss": 4.586240768432617,
|
| 1099 |
+
"eval_rouge1": 0.3859647235684582,
|
| 1100 |
+
"eval_rouge2": 0.25007880823210193,
|
| 1101 |
+
"eval_rougeL": 0.3429738967595123,
|
| 1102 |
+
"eval_rougeLsum": 0.3537661741374731,
|
| 1103 |
+
"eval_runtime": 85.7948,
|
| 1104 |
+
"eval_samples_per_second": 3.112,
|
| 1105 |
+
"eval_steps_per_second": 0.781,
|
| 1106 |
+
"step": 6100
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 62.0,
|
| 1110 |
+
"learning_rate": 0.00011399999999999999,
|
| 1111 |
+
"loss": 0.015,
|
| 1112 |
+
"step": 6200
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 62.0,
|
| 1116 |
+
"eval_loss": 4.743378639221191,
|
| 1117 |
+
"eval_rouge1": 0.3846367169172429,
|
| 1118 |
+
"eval_rouge2": 0.24839404159206457,
|
| 1119 |
+
"eval_rougeL": 0.34243506876400887,
|
| 1120 |
+
"eval_rougeLsum": 0.3538437492007503,
|
| 1121 |
+
"eval_runtime": 85.518,
|
| 1122 |
+
"eval_samples_per_second": 3.122,
|
| 1123 |
+
"eval_steps_per_second": 0.783,
|
| 1124 |
+
"step": 6200
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 63.0,
|
| 1128 |
+
"learning_rate": 0.00011099999999999999,
|
| 1129 |
+
"loss": 0.0149,
|
| 1130 |
+
"step": 6300
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 63.0,
|
| 1134 |
+
"eval_loss": 4.629330635070801,
|
| 1135 |
+
"eval_rouge1": 0.38030212626045157,
|
| 1136 |
+
"eval_rouge2": 0.24751450469686587,
|
| 1137 |
+
"eval_rougeL": 0.33809073704758,
|
| 1138 |
+
"eval_rougeLsum": 0.3483896449835656,
|
| 1139 |
+
"eval_runtime": 80.0249,
|
| 1140 |
+
"eval_samples_per_second": 3.336,
|
| 1141 |
+
"eval_steps_per_second": 0.837,
|
| 1142 |
+
"step": 6300
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 64.0,
|
| 1146 |
+
"learning_rate": 0.00010799999999999998,
|
| 1147 |
+
"loss": 0.0126,
|
| 1148 |
+
"step": 6400
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 64.0,
|
| 1152 |
+
"eval_loss": 4.710384368896484,
|
| 1153 |
+
"eval_rouge1": 0.3842642647245845,
|
| 1154 |
+
"eval_rouge2": 0.2512060270485085,
|
| 1155 |
+
"eval_rougeL": 0.34379327783470054,
|
| 1156 |
+
"eval_rougeLsum": 0.3545941692511255,
|
| 1157 |
+
"eval_runtime": 85.176,
|
| 1158 |
+
"eval_samples_per_second": 3.135,
|
| 1159 |
+
"eval_steps_per_second": 0.787,
|
| 1160 |
+
"step": 6400
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 65.0,
|
| 1164 |
+
"learning_rate": 0.00010499999999999999,
|
| 1165 |
+
"loss": 0.0138,
|
| 1166 |
+
"step": 6500
|
| 1167 |
+
},
|
| 1168 |
+
{
|
| 1169 |
+
"epoch": 65.0,
|
| 1170 |
+
"eval_loss": 4.689241886138916,
|
| 1171 |
+
"eval_rouge1": 0.38298024943411063,
|
| 1172 |
+
"eval_rouge2": 0.25101113690854776,
|
| 1173 |
+
"eval_rougeL": 0.3456091975587352,
|
| 1174 |
+
"eval_rougeLsum": 0.3550951514101951,
|
| 1175 |
+
"eval_runtime": 88.4731,
|
| 1176 |
+
"eval_samples_per_second": 3.018,
|
| 1177 |
+
"eval_steps_per_second": 0.757,
|
| 1178 |
+
"step": 6500
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 66.0,
|
| 1182 |
+
"learning_rate": 0.000102,
|
| 1183 |
+
"loss": 0.0145,
|
| 1184 |
+
"step": 6600
|
| 1185 |
+
},
|
| 1186 |
+
{
|
| 1187 |
+
"epoch": 66.0,
|
| 1188 |
+
"eval_loss": 4.557338237762451,
|
| 1189 |
+
"eval_rouge1": 0.380241659345879,
|
| 1190 |
+
"eval_rouge2": 0.24617583106160457,
|
| 1191 |
+
"eval_rougeL": 0.3415380685661965,
|
| 1192 |
+
"eval_rougeLsum": 0.3528959521802908,
|
| 1193 |
+
"eval_runtime": 88.0004,
|
| 1194 |
+
"eval_samples_per_second": 3.034,
|
| 1195 |
+
"eval_steps_per_second": 0.761,
|
| 1196 |
+
"step": 6600
|
| 1197 |
+
},
|
| 1198 |
+
{
|
| 1199 |
+
"epoch": 67.0,
|
| 1200 |
+
"learning_rate": 9.9e-05,
|
| 1201 |
+
"loss": 0.014,
|
| 1202 |
+
"step": 6700
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 67.0,
|
| 1206 |
+
"eval_loss": 4.767906188964844,
|
| 1207 |
+
"eval_rouge1": 0.38450247499658086,
|
| 1208 |
+
"eval_rouge2": 0.24867049264532454,
|
| 1209 |
+
"eval_rougeL": 0.3433152946814418,
|
| 1210 |
+
"eval_rougeLsum": 0.35393881134127675,
|
| 1211 |
+
"eval_runtime": 86.9796,
|
| 1212 |
+
"eval_samples_per_second": 3.07,
|
| 1213 |
+
"eval_steps_per_second": 0.77,
|
| 1214 |
+
"step": 6700
|
| 1215 |
+
},
|
| 1216 |
+
{
|
| 1217 |
+
"epoch": 68.0,
|
| 1218 |
+
"learning_rate": 9.599999999999999e-05,
|
| 1219 |
+
"loss": 0.0115,
|
| 1220 |
+
"step": 6800
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 68.0,
|
| 1224 |
+
"eval_loss": 4.784436225891113,
|
| 1225 |
+
"eval_rouge1": 0.38354762861806485,
|
| 1226 |
+
"eval_rouge2": 0.24990989075753453,
|
| 1227 |
+
"eval_rougeL": 0.342385020020958,
|
| 1228 |
+
"eval_rougeLsum": 0.35342424612693624,
|
| 1229 |
+
"eval_runtime": 84.5851,
|
| 1230 |
+
"eval_samples_per_second": 3.157,
|
| 1231 |
+
"eval_steps_per_second": 0.792,
|
| 1232 |
+
"step": 6800
|
| 1233 |
+
},
|
| 1234 |
+
{
|
| 1235 |
+
"epoch": 69.0,
|
| 1236 |
+
"learning_rate": 9.3e-05,
|
| 1237 |
+
"loss": 0.012,
|
| 1238 |
+
"step": 6900
|
| 1239 |
+
},
|
| 1240 |
+
{
|
| 1241 |
+
"epoch": 69.0,
|
| 1242 |
+
"eval_loss": 4.792604923248291,
|
| 1243 |
+
"eval_rouge1": 0.3781881725835631,
|
| 1244 |
+
"eval_rouge2": 0.24531643415669946,
|
| 1245 |
+
"eval_rougeL": 0.3374030505076982,
|
| 1246 |
+
"eval_rougeLsum": 0.34856794790036694,
|
| 1247 |
+
"eval_runtime": 83.9778,
|
| 1248 |
+
"eval_samples_per_second": 3.179,
|
| 1249 |
+
"eval_steps_per_second": 0.798,
|
| 1250 |
+
"step": 6900
|
| 1251 |
+
},
|
| 1252 |
+
{
|
| 1253 |
+
"epoch": 70.0,
|
| 1254 |
+
"learning_rate": 8.999999999999999e-05,
|
| 1255 |
+
"loss": 0.0112,
|
| 1256 |
+
"step": 7000
|
| 1257 |
+
},
|
| 1258 |
+
{
|
| 1259 |
+
"epoch": 70.0,
|
| 1260 |
+
"eval_loss": 4.755610942840576,
|
| 1261 |
+
"eval_rouge1": 0.3817160833408457,
|
| 1262 |
+
"eval_rouge2": 0.24591914048878963,
|
| 1263 |
+
"eval_rougeL": 0.3399489131887798,
|
| 1264 |
+
"eval_rougeLsum": 0.34979297011876465,
|
| 1265 |
+
"eval_runtime": 84.4798,
|
| 1266 |
+
"eval_samples_per_second": 3.161,
|
| 1267 |
+
"eval_steps_per_second": 0.793,
|
| 1268 |
+
"step": 7000
|
| 1269 |
+
},
|
| 1270 |
+
{
|
| 1271 |
+
"epoch": 71.0,
|
| 1272 |
+
"learning_rate": 8.699999999999999e-05,
|
| 1273 |
+
"loss": 0.0119,
|
| 1274 |
+
"step": 7100
|
| 1275 |
+
},
|
| 1276 |
+
{
|
| 1277 |
+
"epoch": 71.0,
|
| 1278 |
+
"eval_loss": 4.798295974731445,
|
| 1279 |
+
"eval_rouge1": 0.3851093231346726,
|
| 1280 |
+
"eval_rouge2": 0.2537220090344301,
|
| 1281 |
+
"eval_rougeL": 0.345936300548305,
|
| 1282 |
+
"eval_rougeLsum": 0.35628312786353933,
|
| 1283 |
+
"eval_runtime": 85.2857,
|
| 1284 |
+
"eval_samples_per_second": 3.131,
|
| 1285 |
+
"eval_steps_per_second": 0.786,
|
| 1286 |
+
"step": 7100
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 72.0,
|
| 1290 |
+
"learning_rate": 8.4e-05,
|
| 1291 |
+
"loss": 0.0114,
|
| 1292 |
+
"step": 7200
|
| 1293 |
+
},
|
| 1294 |
+
{
|
| 1295 |
+
"epoch": 72.0,
|
| 1296 |
+
"eval_loss": 4.856568813323975,
|
| 1297 |
+
"eval_rouge1": 0.38245157631697796,
|
| 1298 |
+
"eval_rouge2": 0.2494936430554049,
|
| 1299 |
+
"eval_rougeL": 0.34129479377535843,
|
| 1300 |
+
"eval_rougeLsum": 0.35318946273145274,
|
| 1301 |
+
"eval_runtime": 86.7372,
|
| 1302 |
+
"eval_samples_per_second": 3.078,
|
| 1303 |
+
"eval_steps_per_second": 0.772,
|
| 1304 |
+
"step": 7200
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 73.0,
|
| 1308 |
+
"learning_rate": 8.1e-05,
|
| 1309 |
+
"loss": 0.0098,
|
| 1310 |
+
"step": 7300
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 73.0,
|
| 1314 |
+
"eval_loss": 4.907617092132568,
|
| 1315 |
+
"eval_rouge1": 0.38646161684013636,
|
| 1316 |
+
"eval_rouge2": 0.24944761609509186,
|
| 1317 |
+
"eval_rougeL": 0.3455163948277883,
|
| 1318 |
+
"eval_rougeLsum": 0.35471516616705556,
|
| 1319 |
+
"eval_runtime": 85.0789,
|
| 1320 |
+
"eval_samples_per_second": 3.138,
|
| 1321 |
+
"eval_steps_per_second": 0.788,
|
| 1322 |
+
"step": 7300
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 74.0,
|
| 1326 |
+
"learning_rate": 7.8e-05,
|
| 1327 |
+
"loss": 0.0108,
|
| 1328 |
+
"step": 7400
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 74.0,
|
| 1332 |
+
"eval_loss": 4.841740608215332,
|
| 1333 |
+
"eval_rouge1": 0.38215421507562997,
|
| 1334 |
+
"eval_rouge2": 0.244320679535119,
|
| 1335 |
+
"eval_rougeL": 0.33902569869492843,
|
| 1336 |
+
"eval_rougeLsum": 0.34863305654053733,
|
| 1337 |
+
"eval_runtime": 83.1089,
|
| 1338 |
+
"eval_samples_per_second": 3.213,
|
| 1339 |
+
"eval_steps_per_second": 0.806,
|
| 1340 |
+
"step": 7400
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 75.0,
|
| 1344 |
+
"learning_rate": 7.5e-05,
|
| 1345 |
+
"loss": 0.0098,
|
| 1346 |
+
"step": 7500
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 75.0,
|
| 1350 |
+
"eval_loss": 4.904059886932373,
|
| 1351 |
+
"eval_rouge1": 0.3843792931439769,
|
| 1352 |
+
"eval_rouge2": 0.25004411752753497,
|
| 1353 |
+
"eval_rougeL": 0.3433196711815563,
|
| 1354 |
+
"eval_rougeLsum": 0.3532736703238627,
|
| 1355 |
+
"eval_runtime": 76.8263,
|
| 1356 |
+
"eval_samples_per_second": 3.475,
|
| 1357 |
+
"eval_steps_per_second": 0.872,
|
| 1358 |
+
"step": 7500
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 76.0,
|
| 1362 |
+
"learning_rate": 7.199999999999999e-05,
|
| 1363 |
+
"loss": 0.0107,
|
| 1364 |
+
"step": 7600
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 76.0,
|
| 1368 |
+
"eval_loss": 4.855184555053711,
|
| 1369 |
+
"eval_rouge1": 0.3829128308316423,
|
| 1370 |
+
"eval_rouge2": 0.2462514914494317,
|
| 1371 |
+
"eval_rougeL": 0.34099347275539427,
|
| 1372 |
+
"eval_rougeLsum": 0.35125207376644263,
|
| 1373 |
+
"eval_runtime": 88.1794,
|
| 1374 |
+
"eval_samples_per_second": 3.028,
|
| 1375 |
+
"eval_steps_per_second": 0.76,
|
| 1376 |
+
"step": 7600
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 77.0,
|
| 1380 |
+
"learning_rate": 6.9e-05,
|
| 1381 |
+
"loss": 0.0087,
|
| 1382 |
+
"step": 7700
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 77.0,
|
| 1386 |
+
"eval_loss": 4.914191246032715,
|
| 1387 |
+
"eval_rouge1": 0.3858568840845371,
|
| 1388 |
+
"eval_rouge2": 0.2502511174813851,
|
| 1389 |
+
"eval_rougeL": 0.3440629332400241,
|
| 1390 |
+
"eval_rougeLsum": 0.3544972438925257,
|
| 1391 |
+
"eval_runtime": 76.5236,
|
| 1392 |
+
"eval_samples_per_second": 3.489,
|
| 1393 |
+
"eval_steps_per_second": 0.876,
|
| 1394 |
+
"step": 7700
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 78.0,
|
| 1398 |
+
"learning_rate": 6.599999999999999e-05,
|
| 1399 |
+
"loss": 0.0083,
|
| 1400 |
+
"step": 7800
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 78.0,
|
| 1404 |
+
"eval_loss": 4.938214302062988,
|
| 1405 |
+
"eval_rouge1": 0.38030330309287785,
|
| 1406 |
+
"eval_rouge2": 0.24533353097305155,
|
| 1407 |
+
"eval_rougeL": 0.33927399233566136,
|
| 1408 |
+
"eval_rougeLsum": 0.34936840630758315,
|
| 1409 |
+
"eval_runtime": 86.2001,
|
| 1410 |
+
"eval_samples_per_second": 3.097,
|
| 1411 |
+
"eval_steps_per_second": 0.777,
|
| 1412 |
+
"step": 7800
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 79.0,
|
| 1416 |
+
"learning_rate": 6.299999999999999e-05,
|
| 1417 |
+
"loss": 0.0092,
|
| 1418 |
+
"step": 7900
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 79.0,
|
| 1422 |
+
"eval_loss": 4.884538173675537,
|
| 1423 |
+
"eval_rouge1": 0.38434467710443954,
|
| 1424 |
+
"eval_rouge2": 0.2497101085906894,
|
| 1425 |
+
"eval_rougeL": 0.3426642731168459,
|
| 1426 |
+
"eval_rougeLsum": 0.35354404579979554,
|
| 1427 |
+
"eval_runtime": 83.0616,
|
| 1428 |
+
"eval_samples_per_second": 3.214,
|
| 1429 |
+
"eval_steps_per_second": 0.807,
|
| 1430 |
+
"step": 7900
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 80.0,
|
| 1434 |
+
"learning_rate": 5.9999999999999995e-05,
|
| 1435 |
+
"loss": 0.0072,
|
| 1436 |
+
"step": 8000
|
| 1437 |
+
},
|
| 1438 |
+
{
|
| 1439 |
+
"epoch": 80.0,
|
| 1440 |
+
"eval_loss": 4.894115924835205,
|
| 1441 |
+
"eval_rouge1": 0.38544645384117493,
|
| 1442 |
+
"eval_rouge2": 0.2508322384005437,
|
| 1443 |
+
"eval_rougeL": 0.3447843356379286,
|
| 1444 |
+
"eval_rougeLsum": 0.3554394657131237,
|
| 1445 |
+
"eval_runtime": 83.5657,
|
| 1446 |
+
"eval_samples_per_second": 3.195,
|
| 1447 |
+
"eval_steps_per_second": 0.802,
|
| 1448 |
+
"step": 8000
|
| 1449 |
+
},
|
| 1450 |
+
{
|
| 1451 |
+
"epoch": 81.0,
|
| 1452 |
+
"learning_rate": 5.6999999999999996e-05,
|
| 1453 |
+
"loss": 0.0077,
|
| 1454 |
+
"step": 8100
|
| 1455 |
+
},
|
| 1456 |
+
{
|
| 1457 |
+
"epoch": 81.0,
|
| 1458 |
+
"eval_loss": 4.973346710205078,
|
| 1459 |
+
"eval_rouge1": 0.387446314618786,
|
| 1460 |
+
"eval_rouge2": 0.2529628384702761,
|
| 1461 |
+
"eval_rougeL": 0.3469758611038962,
|
| 1462 |
+
"eval_rougeLsum": 0.3572774036661956,
|
| 1463 |
+
"eval_runtime": 87.8578,
|
| 1464 |
+
"eval_samples_per_second": 3.039,
|
| 1465 |
+
"eval_steps_per_second": 0.763,
|
| 1466 |
+
"step": 8100
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 82.0,
|
| 1470 |
+
"learning_rate": 5.399999999999999e-05,
|
| 1471 |
+
"loss": 0.008,
|
| 1472 |
+
"step": 8200
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 82.0,
|
| 1476 |
+
"eval_loss": 4.913274765014648,
|
| 1477 |
+
"eval_rouge1": 0.3829560613774901,
|
| 1478 |
+
"eval_rouge2": 0.25172789870975754,
|
| 1479 |
+
"eval_rougeL": 0.34496794149173104,
|
| 1480 |
+
"eval_rougeLsum": 0.35508158648356575,
|
| 1481 |
+
"eval_runtime": 85.6734,
|
| 1482 |
+
"eval_samples_per_second": 3.116,
|
| 1483 |
+
"eval_steps_per_second": 0.782,
|
| 1484 |
+
"step": 8200
|
| 1485 |
+
},
|
| 1486 |
+
{
|
| 1487 |
+
"epoch": 83.0,
|
| 1488 |
+
"learning_rate": 5.1e-05,
|
| 1489 |
+
"loss": 0.0075,
|
| 1490 |
+
"step": 8300
|
| 1491 |
+
},
|
| 1492 |
+
{
|
| 1493 |
+
"epoch": 83.0,
|
| 1494 |
+
"eval_loss": 4.925784587860107,
|
| 1495 |
+
"eval_rouge1": 0.385854429603814,
|
| 1496 |
+
"eval_rouge2": 0.2531988664068715,
|
| 1497 |
+
"eval_rougeL": 0.346050946742826,
|
| 1498 |
+
"eval_rougeLsum": 0.35696920925711373,
|
| 1499 |
+
"eval_runtime": 75.2297,
|
| 1500 |
+
"eval_samples_per_second": 3.549,
|
| 1501 |
+
"eval_steps_per_second": 0.891,
|
| 1502 |
+
"step": 8300
|
| 1503 |
+
},
|
| 1504 |
+
{
|
| 1505 |
+
"epoch": 84.0,
|
| 1506 |
+
"learning_rate": 4.7999999999999994e-05,
|
| 1507 |
+
"loss": 0.0076,
|
| 1508 |
+
"step": 8400
|
| 1509 |
+
},
|
| 1510 |
+
{
|
| 1511 |
+
"epoch": 84.0,
|
| 1512 |
+
"eval_loss": 4.87917423248291,
|
| 1513 |
+
"eval_rouge1": 0.38643959051361326,
|
| 1514 |
+
"eval_rouge2": 0.24825036815478088,
|
| 1515 |
+
"eval_rougeL": 0.34543928091296466,
|
| 1516 |
+
"eval_rougeLsum": 0.3560830652803086,
|
| 1517 |
+
"eval_runtime": 82.6422,
|
| 1518 |
+
"eval_samples_per_second": 3.231,
|
| 1519 |
+
"eval_steps_per_second": 0.811,
|
| 1520 |
+
"step": 8400
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 85.0,
|
| 1524 |
+
"learning_rate": 4.4999999999999996e-05,
|
| 1525 |
+
"loss": 0.0073,
|
| 1526 |
+
"step": 8500
|
| 1527 |
+
},
|
| 1528 |
+
{
|
| 1529 |
+
"epoch": 85.0,
|
| 1530 |
+
"eval_loss": 4.9377264976501465,
|
| 1531 |
+
"eval_rouge1": 0.38212230866847685,
|
| 1532 |
+
"eval_rouge2": 0.24573634944784611,
|
| 1533 |
+
"eval_rougeL": 0.34138760450808503,
|
| 1534 |
+
"eval_rougeLsum": 0.3523812948696121,
|
| 1535 |
+
"eval_runtime": 85.716,
|
| 1536 |
+
"eval_samples_per_second": 3.115,
|
| 1537 |
+
"eval_steps_per_second": 0.782,
|
| 1538 |
+
"step": 8500
|
| 1539 |
+
},
|
| 1540 |
+
{
|
| 1541 |
+
"epoch": 86.0,
|
| 1542 |
+
"learning_rate": 4.2e-05,
|
| 1543 |
+
"loss": 0.0062,
|
| 1544 |
+
"step": 8600
|
| 1545 |
+
},
|
| 1546 |
+
{
|
| 1547 |
+
"epoch": 86.0,
|
| 1548 |
+
"eval_loss": 5.010465621948242,
|
| 1549 |
+
"eval_rouge1": 0.38248533103068416,
|
| 1550 |
+
"eval_rouge2": 0.24842178766177614,
|
| 1551 |
+
"eval_rougeL": 0.3425108161941025,
|
| 1552 |
+
"eval_rougeLsum": 0.35344680511965376,
|
| 1553 |
+
"eval_runtime": 85.9789,
|
| 1554 |
+
"eval_samples_per_second": 3.105,
|
| 1555 |
+
"eval_steps_per_second": 0.779,
|
| 1556 |
+
"step": 8600
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 87.0,
|
| 1560 |
+
"learning_rate": 3.9e-05,
|
| 1561 |
+
"loss": 0.0067,
|
| 1562 |
+
"step": 8700
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 87.0,
|
| 1566 |
+
"eval_loss": 5.000132083892822,
|
| 1567 |
+
"eval_rouge1": 0.3844108716094221,
|
| 1568 |
+
"eval_rouge2": 0.25183492085762804,
|
| 1569 |
+
"eval_rougeL": 0.34541276301829477,
|
| 1570 |
+
"eval_rougeLsum": 0.35582368167109873,
|
| 1571 |
+
"eval_runtime": 85.4159,
|
| 1572 |
+
"eval_samples_per_second": 3.126,
|
| 1573 |
+
"eval_steps_per_second": 0.784,
|
| 1574 |
+
"step": 8700
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 88.0,
|
| 1578 |
+
"learning_rate": 3.5999999999999994e-05,
|
| 1579 |
+
"loss": 0.0064,
|
| 1580 |
+
"step": 8800
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 88.0,
|
| 1584 |
+
"eval_loss": 4.985653877258301,
|
| 1585 |
+
"eval_rouge1": 0.3842791176433939,
|
| 1586 |
+
"eval_rouge2": 0.24890479606772847,
|
| 1587 |
+
"eval_rougeL": 0.34373666785104745,
|
| 1588 |
+
"eval_rougeLsum": 0.35402492917866446,
|
| 1589 |
+
"eval_runtime": 88.491,
|
| 1590 |
+
"eval_samples_per_second": 3.017,
|
| 1591 |
+
"eval_steps_per_second": 0.757,
|
| 1592 |
+
"step": 8800
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 89.0,
|
| 1596 |
+
"learning_rate": 3.2999999999999996e-05,
|
| 1597 |
+
"loss": 0.0064,
|
| 1598 |
+
"step": 8900
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 89.0,
|
| 1602 |
+
"eval_loss": 5.027814865112305,
|
| 1603 |
+
"eval_rouge1": 0.38374734558183804,
|
| 1604 |
+
"eval_rouge2": 0.24654112382289783,
|
| 1605 |
+
"eval_rougeL": 0.3418827464509472,
|
| 1606 |
+
"eval_rougeLsum": 0.35332526206489273,
|
| 1607 |
+
"eval_runtime": 88.8366,
|
| 1608 |
+
"eval_samples_per_second": 3.006,
|
| 1609 |
+
"eval_steps_per_second": 0.754,
|
| 1610 |
+
"step": 8900
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 90.0,
|
| 1614 |
+
"learning_rate": 2.9999999999999997e-05,
|
| 1615 |
+
"loss": 0.0061,
|
| 1616 |
+
"step": 9000
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 90.0,
|
| 1620 |
+
"eval_loss": 5.058474063873291,
|
| 1621 |
+
"eval_rouge1": 0.3820717972330683,
|
| 1622 |
+
"eval_rouge2": 0.2461904812224382,
|
| 1623 |
+
"eval_rougeL": 0.340562678348688,
|
| 1624 |
+
"eval_rougeLsum": 0.35149040221377614,
|
| 1625 |
+
"eval_runtime": 87.2966,
|
| 1626 |
+
"eval_samples_per_second": 3.059,
|
| 1627 |
+
"eval_steps_per_second": 0.767,
|
| 1628 |
+
"step": 9000
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 91.0,
|
| 1632 |
+
"learning_rate": 2.6999999999999996e-05,
|
| 1633 |
+
"loss": 0.007,
|
| 1634 |
+
"step": 9100
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 91.0,
|
| 1638 |
+
"eval_loss": 5.010220050811768,
|
| 1639 |
+
"eval_rouge1": 0.3844908418040261,
|
| 1640 |
+
"eval_rouge2": 0.25006688388196396,
|
| 1641 |
+
"eval_rougeL": 0.3429817235545874,
|
| 1642 |
+
"eval_rougeLsum": 0.35434878150140914,
|
| 1643 |
+
"eval_runtime": 79.8527,
|
| 1644 |
+
"eval_samples_per_second": 3.344,
|
| 1645 |
+
"eval_steps_per_second": 0.839,
|
| 1646 |
+
"step": 9100
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 92.0,
|
| 1650 |
+
"learning_rate": 2.3999999999999997e-05,
|
| 1651 |
+
"loss": 0.0059,
|
| 1652 |
+
"step": 9200
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 92.0,
|
| 1656 |
+
"eval_loss": 5.030458450317383,
|
| 1657 |
+
"eval_rouge1": 0.38284739488530695,
|
| 1658 |
+
"eval_rouge2": 0.2494251591428921,
|
| 1659 |
+
"eval_rougeL": 0.34146682998770905,
|
| 1660 |
+
"eval_rougeLsum": 0.3524988183011109,
|
| 1661 |
+
"eval_runtime": 86.6859,
|
| 1662 |
+
"eval_samples_per_second": 3.08,
|
| 1663 |
+
"eval_steps_per_second": 0.773,
|
| 1664 |
+
"step": 9200
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 93.0,
|
| 1668 |
+
"learning_rate": 2.1e-05,
|
| 1669 |
+
"loss": 0.0066,
|
| 1670 |
+
"step": 9300
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 93.0,
|
| 1674 |
+
"eval_loss": 4.984446048736572,
|
| 1675 |
+
"eval_rouge1": 0.38152251971542894,
|
| 1676 |
+
"eval_rouge2": 0.24878451374892646,
|
| 1677 |
+
"eval_rougeL": 0.3406473614314545,
|
| 1678 |
+
"eval_rougeLsum": 0.3513187114756059,
|
| 1679 |
+
"eval_runtime": 88.5244,
|
| 1680 |
+
"eval_samples_per_second": 3.016,
|
| 1681 |
+
"eval_steps_per_second": 0.757,
|
| 1682 |
+
"step": 9300
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 94.0,
|
| 1686 |
+
"learning_rate": 1.7999999999999997e-05,
|
| 1687 |
+
"loss": 0.0058,
|
| 1688 |
+
"step": 9400
|
| 1689 |
+
},
|
| 1690 |
+
{
|
| 1691 |
+
"epoch": 94.0,
|
| 1692 |
+
"eval_loss": 4.990396022796631,
|
| 1693 |
+
"eval_rouge1": 0.3825136945109377,
|
| 1694 |
+
"eval_rouge2": 0.24843252471104438,
|
| 1695 |
+
"eval_rougeL": 0.3422545305896276,
|
| 1696 |
+
"eval_rougeLsum": 0.352581643511201,
|
| 1697 |
+
"eval_runtime": 85.262,
|
| 1698 |
+
"eval_samples_per_second": 3.132,
|
| 1699 |
+
"eval_steps_per_second": 0.786,
|
| 1700 |
+
"step": 9400
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 95.0,
|
| 1704 |
+
"learning_rate": 1.4999999999999999e-05,
|
| 1705 |
+
"loss": 0.006,
|
| 1706 |
+
"step": 9500
|
| 1707 |
+
},
|
| 1708 |
+
{
|
| 1709 |
+
"epoch": 95.0,
|
| 1710 |
+
"eval_loss": 4.97911262512207,
|
| 1711 |
+
"eval_rouge1": 0.38343787306408117,
|
| 1712 |
+
"eval_rouge2": 0.24997418070032396,
|
| 1713 |
+
"eval_rougeL": 0.3422221622019714,
|
| 1714 |
+
"eval_rougeLsum": 0.35352360810152816,
|
| 1715 |
+
"eval_runtime": 81.5407,
|
| 1716 |
+
"eval_samples_per_second": 3.274,
|
| 1717 |
+
"eval_steps_per_second": 0.822,
|
| 1718 |
+
"step": 9500
|
| 1719 |
+
},
|
| 1720 |
+
{
|
| 1721 |
+
"epoch": 96.0,
|
| 1722 |
+
"learning_rate": 1.1999999999999999e-05,
|
| 1723 |
+
"loss": 0.0049,
|
| 1724 |
+
"step": 9600
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 96.0,
|
| 1728 |
+
"eval_loss": 5.016595363616943,
|
| 1729 |
+
"eval_rouge1": 0.3837960479091025,
|
| 1730 |
+
"eval_rouge2": 0.24885786511102678,
|
| 1731 |
+
"eval_rougeL": 0.3424026645718613,
|
| 1732 |
+
"eval_rougeLsum": 0.35330084315799287,
|
| 1733 |
+
"eval_runtime": 88.6796,
|
| 1734 |
+
"eval_samples_per_second": 3.011,
|
| 1735 |
+
"eval_steps_per_second": 0.756,
|
| 1736 |
+
"step": 9600
|
| 1737 |
+
},
|
| 1738 |
+
{
|
| 1739 |
+
"epoch": 97.0,
|
| 1740 |
+
"learning_rate": 8.999999999999999e-06,
|
| 1741 |
+
"loss": 0.0054,
|
| 1742 |
+
"step": 9700
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 97.0,
|
| 1746 |
+
"eval_loss": 5.0309343338012695,
|
| 1747 |
+
"eval_rouge1": 0.383180407064846,
|
| 1748 |
+
"eval_rouge2": 0.2500666839747965,
|
| 1749 |
+
"eval_rougeL": 0.3424454114218375,
|
| 1750 |
+
"eval_rougeLsum": 0.3533197372424365,
|
| 1751 |
+
"eval_runtime": 87.6124,
|
| 1752 |
+
"eval_samples_per_second": 3.048,
|
| 1753 |
+
"eval_steps_per_second": 0.765,
|
| 1754 |
+
"step": 9700
|
| 1755 |
+
},
|
| 1756 |
+
{
|
| 1757 |
+
"epoch": 98.0,
|
| 1758 |
+
"learning_rate": 5.999999999999999e-06,
|
| 1759 |
+
"loss": 0.0058,
|
| 1760 |
+
"step": 9800
|
| 1761 |
+
},
|
| 1762 |
+
{
|
| 1763 |
+
"epoch": 98.0,
|
| 1764 |
+
"eval_loss": 5.044477462768555,
|
| 1765 |
+
"eval_rouge1": 0.3841857746723203,
|
| 1766 |
+
"eval_rouge2": 0.2492668969413123,
|
| 1767 |
+
"eval_rougeL": 0.34243325874354047,
|
| 1768 |
+
"eval_rougeLsum": 0.3533788089951856,
|
| 1769 |
+
"eval_runtime": 88.7303,
|
| 1770 |
+
"eval_samples_per_second": 3.009,
|
| 1771 |
+
"eval_steps_per_second": 0.755,
|
| 1772 |
+
"step": 9800
|
| 1773 |
+
},
|
| 1774 |
+
{
|
| 1775 |
+
"epoch": 99.0,
|
| 1776 |
+
"learning_rate": 2.9999999999999997e-06,
|
| 1777 |
+
"loss": 0.0059,
|
| 1778 |
+
"step": 9900
|
| 1779 |
+
},
|
| 1780 |
+
{
|
| 1781 |
+
"epoch": 99.0,
|
| 1782 |
+
"eval_loss": 5.052584171295166,
|
| 1783 |
+
"eval_rouge1": 0.38395960024638665,
|
| 1784 |
+
"eval_rouge2": 0.2489015750324879,
|
| 1785 |
+
"eval_rougeL": 0.3410612103718015,
|
| 1786 |
+
"eval_rougeLsum": 0.3524355955865259,
|
| 1787 |
+
"eval_runtime": 84.3598,
|
| 1788 |
+
"eval_samples_per_second": 3.165,
|
| 1789 |
+
"eval_steps_per_second": 0.794,
|
| 1790 |
+
"step": 9900
|
| 1791 |
+
}
|
| 1792 |
+
],
|
| 1793 |
+
"max_steps": 10000,
|
| 1794 |
+
"num_train_epochs": 100,
|
| 1795 |
+
"total_flos": 3683373994819584.0,
|
| 1796 |
+
"trial_name": null,
|
| 1797 |
+
"trial_params": null
|
| 1798 |
+
}
|