Upload trainer_state.json
Browse files- trainer_state.json +3891 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,3891 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 10.695329087048833,
|
| 3 |
+
"best_model_checkpoint": "./model_out/checkpoint-10000",
|
| 4 |
+
"epoch": 5.5514433752775725,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 15000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01,
|
| 13 |
+
"learning_rate": 2.4000000000000003e-06,
|
| 14 |
+
"loss": 1.2092,
|
| 15 |
+
"step": 25
|
| 16 |
+
},
|
| 17 |
+
{
|
| 18 |
+
"epoch": 0.02,
|
| 19 |
+
"learning_rate": 4.9000000000000005e-06,
|
| 20 |
+
"loss": 0.8939,
|
| 21 |
+
"step": 50
|
| 22 |
+
},
|
| 23 |
+
{
|
| 24 |
+
"epoch": 0.03,
|
| 25 |
+
"learning_rate": 7.4e-06,
|
| 26 |
+
"loss": 0.7337,
|
| 27 |
+
"step": 75
|
| 28 |
+
},
|
| 29 |
+
{
|
| 30 |
+
"epoch": 0.04,
|
| 31 |
+
"learning_rate": 9.900000000000002e-06,
|
| 32 |
+
"loss": 0.6496,
|
| 33 |
+
"step": 100
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.05,
|
| 37 |
+
"learning_rate": 1.24e-05,
|
| 38 |
+
"loss": 0.5998,
|
| 39 |
+
"step": 125
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.06,
|
| 43 |
+
"learning_rate": 1.49e-05,
|
| 44 |
+
"loss": 0.5262,
|
| 45 |
+
"step": 150
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.06,
|
| 49 |
+
"learning_rate": 1.74e-05,
|
| 50 |
+
"loss": 0.501,
|
| 51 |
+
"step": 175
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.07,
|
| 55 |
+
"learning_rate": 1.9900000000000003e-05,
|
| 56 |
+
"loss": 0.4801,
|
| 57 |
+
"step": 200
|
| 58 |
+
},
|
| 59 |
+
{
|
| 60 |
+
"epoch": 0.08,
|
| 61 |
+
"learning_rate": 2.2400000000000002e-05,
|
| 62 |
+
"loss": 0.4495,
|
| 63 |
+
"step": 225
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.09,
|
| 67 |
+
"learning_rate": 2.4900000000000002e-05,
|
| 68 |
+
"loss": 0.4372,
|
| 69 |
+
"step": 250
|
| 70 |
+
},
|
| 71 |
+
{
|
| 72 |
+
"epoch": 0.1,
|
| 73 |
+
"learning_rate": 2.7300000000000003e-05,
|
| 74 |
+
"loss": 0.425,
|
| 75 |
+
"step": 275
|
| 76 |
+
},
|
| 77 |
+
{
|
| 78 |
+
"epoch": 0.11,
|
| 79 |
+
"learning_rate": 2.98e-05,
|
| 80 |
+
"loss": 0.4097,
|
| 81 |
+
"step": 300
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.12,
|
| 85 |
+
"learning_rate": 3.2300000000000006e-05,
|
| 86 |
+
"loss": 0.3997,
|
| 87 |
+
"step": 325
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.13,
|
| 91 |
+
"learning_rate": 3.48e-05,
|
| 92 |
+
"loss": 0.3828,
|
| 93 |
+
"step": 350
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.14,
|
| 97 |
+
"learning_rate": 3.73e-05,
|
| 98 |
+
"loss": 0.3817,
|
| 99 |
+
"step": 375
|
| 100 |
+
},
|
| 101 |
+
{
|
| 102 |
+
"epoch": 0.15,
|
| 103 |
+
"learning_rate": 3.9800000000000005e-05,
|
| 104 |
+
"loss": 0.3696,
|
| 105 |
+
"step": 400
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.16,
|
| 109 |
+
"learning_rate": 4.23e-05,
|
| 110 |
+
"loss": 0.3808,
|
| 111 |
+
"step": 425
|
| 112 |
+
},
|
| 113 |
+
{
|
| 114 |
+
"epoch": 0.17,
|
| 115 |
+
"learning_rate": 4.4800000000000005e-05,
|
| 116 |
+
"loss": 0.3764,
|
| 117 |
+
"step": 450
|
| 118 |
+
},
|
| 119 |
+
{
|
| 120 |
+
"epoch": 0.18,
|
| 121 |
+
"learning_rate": 4.73e-05,
|
| 122 |
+
"loss": 0.3712,
|
| 123 |
+
"step": 475
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.19,
|
| 127 |
+
"learning_rate": 4.9800000000000004e-05,
|
| 128 |
+
"loss": 0.3416,
|
| 129 |
+
"step": 500
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.19,
|
| 133 |
+
"eval_cer": 13.553988474370641,
|
| 134 |
+
"eval_loss": 0.2625325620174408,
|
| 135 |
+
"eval_runtime": 367.6787,
|
| 136 |
+
"eval_samples_per_second": 7.02,
|
| 137 |
+
"eval_steps_per_second": 0.878,
|
| 138 |
+
"step": 500
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.19,
|
| 142 |
+
"learning_rate": 4.992068965517242e-05,
|
| 143 |
+
"loss": 0.3682,
|
| 144 |
+
"step": 525
|
| 145 |
+
},
|
| 146 |
+
{
|
| 147 |
+
"epoch": 0.2,
|
| 148 |
+
"learning_rate": 4.9834482758620695e-05,
|
| 149 |
+
"loss": 0.3716,
|
| 150 |
+
"step": 550
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.21,
|
| 154 |
+
"learning_rate": 4.9748275862068966e-05,
|
| 155 |
+
"loss": 0.3581,
|
| 156 |
+
"step": 575
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.22,
|
| 160 |
+
"learning_rate": 4.966206896551724e-05,
|
| 161 |
+
"loss": 0.3542,
|
| 162 |
+
"step": 600
|
| 163 |
+
},
|
| 164 |
+
{
|
| 165 |
+
"epoch": 0.23,
|
| 166 |
+
"learning_rate": 4.9575862068965514e-05,
|
| 167 |
+
"loss": 0.347,
|
| 168 |
+
"step": 625
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.24,
|
| 172 |
+
"learning_rate": 4.94896551724138e-05,
|
| 173 |
+
"loss": 0.3603,
|
| 174 |
+
"step": 650
|
| 175 |
+
},
|
| 176 |
+
{
|
| 177 |
+
"epoch": 0.25,
|
| 178 |
+
"learning_rate": 4.940344827586207e-05,
|
| 179 |
+
"loss": 0.3568,
|
| 180 |
+
"step": 675
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.26,
|
| 184 |
+
"learning_rate": 4.931724137931035e-05,
|
| 185 |
+
"loss": 0.3408,
|
| 186 |
+
"step": 700
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.27,
|
| 190 |
+
"learning_rate": 4.9231034482758625e-05,
|
| 191 |
+
"loss": 0.346,
|
| 192 |
+
"step": 725
|
| 193 |
+
},
|
| 194 |
+
{
|
| 195 |
+
"epoch": 0.28,
|
| 196 |
+
"learning_rate": 4.9144827586206896e-05,
|
| 197 |
+
"loss": 0.3382,
|
| 198 |
+
"step": 750
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.29,
|
| 202 |
+
"learning_rate": 4.9058620689655174e-05,
|
| 203 |
+
"loss": 0.3422,
|
| 204 |
+
"step": 775
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.3,
|
| 208 |
+
"learning_rate": 4.897241379310345e-05,
|
| 209 |
+
"loss": 0.3423,
|
| 210 |
+
"step": 800
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"epoch": 0.31,
|
| 214 |
+
"learning_rate": 4.888620689655173e-05,
|
| 215 |
+
"loss": 0.3245,
|
| 216 |
+
"step": 825
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.31,
|
| 220 |
+
"learning_rate": 4.88e-05,
|
| 221 |
+
"loss": 0.3378,
|
| 222 |
+
"step": 850
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.32,
|
| 226 |
+
"learning_rate": 4.871379310344828e-05,
|
| 227 |
+
"loss": 0.3388,
|
| 228 |
+
"step": 875
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.33,
|
| 232 |
+
"learning_rate": 4.8627586206896556e-05,
|
| 233 |
+
"loss": 0.3189,
|
| 234 |
+
"step": 900
|
| 235 |
+
},
|
| 236 |
+
{
|
| 237 |
+
"epoch": 0.34,
|
| 238 |
+
"learning_rate": 4.8541379310344834e-05,
|
| 239 |
+
"loss": 0.3264,
|
| 240 |
+
"step": 925
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.35,
|
| 244 |
+
"learning_rate": 4.8455172413793105e-05,
|
| 245 |
+
"loss": 0.3231,
|
| 246 |
+
"step": 950
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.36,
|
| 250 |
+
"learning_rate": 4.836896551724138e-05,
|
| 251 |
+
"loss": 0.3046,
|
| 252 |
+
"step": 975
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.37,
|
| 256 |
+
"learning_rate": 4.828275862068965e-05,
|
| 257 |
+
"loss": 0.3115,
|
| 258 |
+
"step": 1000
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.37,
|
| 262 |
+
"eval_cer": 12.473460721868365,
|
| 263 |
+
"eval_loss": 0.2618062496185303,
|
| 264 |
+
"eval_runtime": 363.8751,
|
| 265 |
+
"eval_samples_per_second": 7.093,
|
| 266 |
+
"eval_steps_per_second": 0.888,
|
| 267 |
+
"step": 1000
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.38,
|
| 271 |
+
"learning_rate": 4.819655172413793e-05,
|
| 272 |
+
"loss": 0.3251,
|
| 273 |
+
"step": 1025
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.39,
|
| 277 |
+
"learning_rate": 4.811034482758621e-05,
|
| 278 |
+
"loss": 0.3158,
|
| 279 |
+
"step": 1050
|
| 280 |
+
},
|
| 281 |
+
{
|
| 282 |
+
"epoch": 0.4,
|
| 283 |
+
"learning_rate": 4.8024137931034486e-05,
|
| 284 |
+
"loss": 0.3093,
|
| 285 |
+
"step": 1075
|
| 286 |
+
},
|
| 287 |
+
{
|
| 288 |
+
"epoch": 0.41,
|
| 289 |
+
"learning_rate": 4.7937931034482764e-05,
|
| 290 |
+
"loss": 0.3238,
|
| 291 |
+
"step": 1100
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.42,
|
| 295 |
+
"learning_rate": 4.7851724137931035e-05,
|
| 296 |
+
"loss": 0.3128,
|
| 297 |
+
"step": 1125
|
| 298 |
+
},
|
| 299 |
+
{
|
| 300 |
+
"epoch": 0.43,
|
| 301 |
+
"learning_rate": 4.776551724137931e-05,
|
| 302 |
+
"loss": 0.3012,
|
| 303 |
+
"step": 1150
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.43,
|
| 307 |
+
"learning_rate": 4.767931034482759e-05,
|
| 308 |
+
"loss": 0.3103,
|
| 309 |
+
"step": 1175
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.44,
|
| 313 |
+
"learning_rate": 4.759310344827587e-05,
|
| 314 |
+
"loss": 0.314,
|
| 315 |
+
"step": 1200
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.45,
|
| 319 |
+
"learning_rate": 4.750689655172414e-05,
|
| 320 |
+
"loss": 0.325,
|
| 321 |
+
"step": 1225
|
| 322 |
+
},
|
| 323 |
+
{
|
| 324 |
+
"epoch": 0.46,
|
| 325 |
+
"learning_rate": 4.742068965517242e-05,
|
| 326 |
+
"loss": 0.2923,
|
| 327 |
+
"step": 1250
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.47,
|
| 331 |
+
"learning_rate": 4.733448275862069e-05,
|
| 332 |
+
"loss": 0.3047,
|
| 333 |
+
"step": 1275
|
| 334 |
+
},
|
| 335 |
+
{
|
| 336 |
+
"epoch": 0.48,
|
| 337 |
+
"learning_rate": 4.7251724137931034e-05,
|
| 338 |
+
"loss": 0.3007,
|
| 339 |
+
"step": 1300
|
| 340 |
+
},
|
| 341 |
+
{
|
| 342 |
+
"epoch": 0.49,
|
| 343 |
+
"learning_rate": 4.716551724137931e-05,
|
| 344 |
+
"loss": 0.3233,
|
| 345 |
+
"step": 1325
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.5,
|
| 349 |
+
"learning_rate": 4.707931034482759e-05,
|
| 350 |
+
"loss": 0.2983,
|
| 351 |
+
"step": 1350
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.51,
|
| 355 |
+
"learning_rate": 4.699310344827587e-05,
|
| 356 |
+
"loss": 0.2961,
|
| 357 |
+
"step": 1375
|
| 358 |
+
},
|
| 359 |
+
{
|
| 360 |
+
"epoch": 0.52,
|
| 361 |
+
"learning_rate": 4.690689655172414e-05,
|
| 362 |
+
"loss": 0.2929,
|
| 363 |
+
"step": 1400
|
| 364 |
+
},
|
| 365 |
+
{
|
| 366 |
+
"epoch": 0.53,
|
| 367 |
+
"learning_rate": 4.6820689655172416e-05,
|
| 368 |
+
"loss": 0.2839,
|
| 369 |
+
"step": 1425
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.54,
|
| 373 |
+
"learning_rate": 4.673448275862069e-05,
|
| 374 |
+
"loss": 0.2973,
|
| 375 |
+
"step": 1450
|
| 376 |
+
},
|
| 377 |
+
{
|
| 378 |
+
"epoch": 0.55,
|
| 379 |
+
"learning_rate": 4.664827586206897e-05,
|
| 380 |
+
"loss": 0.2896,
|
| 381 |
+
"step": 1475
|
| 382 |
+
},
|
| 383 |
+
{
|
| 384 |
+
"epoch": 0.56,
|
| 385 |
+
"learning_rate": 4.656206896551724e-05,
|
| 386 |
+
"loss": 0.289,
|
| 387 |
+
"step": 1500
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 0.56,
|
| 391 |
+
"eval_cer": 11.809978768577494,
|
| 392 |
+
"eval_loss": 0.2488066703081131,
|
| 393 |
+
"eval_runtime": 362.9261,
|
| 394 |
+
"eval_samples_per_second": 7.112,
|
| 395 |
+
"eval_steps_per_second": 0.89,
|
| 396 |
+
"step": 1500
|
| 397 |
+
},
|
| 398 |
+
{
|
| 399 |
+
"epoch": 0.56,
|
| 400 |
+
"learning_rate": 4.647586206896552e-05,
|
| 401 |
+
"loss": 0.2872,
|
| 402 |
+
"step": 1525
|
| 403 |
+
},
|
| 404 |
+
{
|
| 405 |
+
"epoch": 0.57,
|
| 406 |
+
"learning_rate": 4.638965517241379e-05,
|
| 407 |
+
"loss": 0.2965,
|
| 408 |
+
"step": 1550
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 0.58,
|
| 412 |
+
"learning_rate": 4.630344827586207e-05,
|
| 413 |
+
"loss": 0.3036,
|
| 414 |
+
"step": 1575
|
| 415 |
+
},
|
| 416 |
+
{
|
| 417 |
+
"epoch": 0.59,
|
| 418 |
+
"learning_rate": 4.621724137931035e-05,
|
| 419 |
+
"loss": 0.2891,
|
| 420 |
+
"step": 1600
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.6,
|
| 424 |
+
"learning_rate": 4.6131034482758624e-05,
|
| 425 |
+
"loss": 0.2896,
|
| 426 |
+
"step": 1625
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.61,
|
| 430 |
+
"learning_rate": 4.60448275862069e-05,
|
| 431 |
+
"loss": 0.2992,
|
| 432 |
+
"step": 1650
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 0.62,
|
| 436 |
+
"learning_rate": 4.595862068965517e-05,
|
| 437 |
+
"loss": 0.2865,
|
| 438 |
+
"step": 1675
|
| 439 |
+
},
|
| 440 |
+
{
|
| 441 |
+
"epoch": 0.63,
|
| 442 |
+
"learning_rate": 4.587241379310345e-05,
|
| 443 |
+
"loss": 0.3007,
|
| 444 |
+
"step": 1700
|
| 445 |
+
},
|
| 446 |
+
{
|
| 447 |
+
"epoch": 0.64,
|
| 448 |
+
"learning_rate": 4.578620689655173e-05,
|
| 449 |
+
"loss": 0.3108,
|
| 450 |
+
"step": 1725
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.65,
|
| 454 |
+
"learning_rate": 4.5700000000000006e-05,
|
| 455 |
+
"loss": 0.2861,
|
| 456 |
+
"step": 1750
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 0.66,
|
| 460 |
+
"learning_rate": 4.561379310344828e-05,
|
| 461 |
+
"loss": 0.2996,
|
| 462 |
+
"step": 1775
|
| 463 |
+
},
|
| 464 |
+
{
|
| 465 |
+
"epoch": 0.67,
|
| 466 |
+
"learning_rate": 4.5527586206896555e-05,
|
| 467 |
+
"loss": 0.295,
|
| 468 |
+
"step": 1800
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.68,
|
| 472 |
+
"learning_rate": 4.5441379310344826e-05,
|
| 473 |
+
"loss": 0.2848,
|
| 474 |
+
"step": 1825
|
| 475 |
+
},
|
| 476 |
+
{
|
| 477 |
+
"epoch": 0.68,
|
| 478 |
+
"learning_rate": 4.5355172413793104e-05,
|
| 479 |
+
"loss": 0.2997,
|
| 480 |
+
"step": 1850
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"epoch": 0.69,
|
| 484 |
+
"learning_rate": 4.526896551724138e-05,
|
| 485 |
+
"loss": 0.2828,
|
| 486 |
+
"step": 1875
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.7,
|
| 490 |
+
"learning_rate": 4.518620689655173e-05,
|
| 491 |
+
"loss": 0.2889,
|
| 492 |
+
"step": 1900
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 0.71,
|
| 496 |
+
"learning_rate": 4.5100000000000005e-05,
|
| 497 |
+
"loss": 0.2893,
|
| 498 |
+
"step": 1925
|
| 499 |
+
},
|
| 500 |
+
{
|
| 501 |
+
"epoch": 0.72,
|
| 502 |
+
"learning_rate": 4.5013793103448276e-05,
|
| 503 |
+
"loss": 0.276,
|
| 504 |
+
"step": 1950
|
| 505 |
+
},
|
| 506 |
+
{
|
| 507 |
+
"epoch": 0.73,
|
| 508 |
+
"learning_rate": 4.4927586206896554e-05,
|
| 509 |
+
"loss": 0.2784,
|
| 510 |
+
"step": 1975
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 0.74,
|
| 514 |
+
"learning_rate": 4.4841379310344825e-05,
|
| 515 |
+
"loss": 0.2826,
|
| 516 |
+
"step": 2000
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.74,
|
| 520 |
+
"eval_cer": 13.197603882317258,
|
| 521 |
+
"eval_loss": 0.2393285036087036,
|
| 522 |
+
"eval_runtime": 368.8043,
|
| 523 |
+
"eval_samples_per_second": 6.998,
|
| 524 |
+
"eval_steps_per_second": 0.876,
|
| 525 |
+
"step": 2000
|
| 526 |
+
},
|
| 527 |
+
{
|
| 528 |
+
"epoch": 0.75,
|
| 529 |
+
"learning_rate": 4.475517241379311e-05,
|
| 530 |
+
"loss": 0.2707,
|
| 531 |
+
"step": 2025
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.76,
|
| 535 |
+
"learning_rate": 4.466896551724138e-05,
|
| 536 |
+
"loss": 0.2824,
|
| 537 |
+
"step": 2050
|
| 538 |
+
},
|
| 539 |
+
{
|
| 540 |
+
"epoch": 0.77,
|
| 541 |
+
"learning_rate": 4.458275862068966e-05,
|
| 542 |
+
"loss": 0.296,
|
| 543 |
+
"step": 2075
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 0.78,
|
| 547 |
+
"learning_rate": 4.4496551724137936e-05,
|
| 548 |
+
"loss": 0.2786,
|
| 549 |
+
"step": 2100
|
| 550 |
+
},
|
| 551 |
+
{
|
| 552 |
+
"epoch": 0.79,
|
| 553 |
+
"learning_rate": 4.441034482758621e-05,
|
| 554 |
+
"loss": 0.2802,
|
| 555 |
+
"step": 2125
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 0.8,
|
| 559 |
+
"learning_rate": 4.4324137931034485e-05,
|
| 560 |
+
"loss": 0.2791,
|
| 561 |
+
"step": 2150
|
| 562 |
+
},
|
| 563 |
+
{
|
| 564 |
+
"epoch": 0.8,
|
| 565 |
+
"learning_rate": 4.423793103448276e-05,
|
| 566 |
+
"loss": 0.2953,
|
| 567 |
+
"step": 2175
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 0.81,
|
| 571 |
+
"learning_rate": 4.415172413793104e-05,
|
| 572 |
+
"loss": 0.2807,
|
| 573 |
+
"step": 2200
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 0.82,
|
| 577 |
+
"learning_rate": 4.406551724137931e-05,
|
| 578 |
+
"loss": 0.2776,
|
| 579 |
+
"step": 2225
|
| 580 |
+
},
|
| 581 |
+
{
|
| 582 |
+
"epoch": 0.83,
|
| 583 |
+
"learning_rate": 4.397931034482759e-05,
|
| 584 |
+
"loss": 0.2798,
|
| 585 |
+
"step": 2250
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.84,
|
| 589 |
+
"learning_rate": 4.389310344827586e-05,
|
| 590 |
+
"loss": 0.2804,
|
| 591 |
+
"step": 2275
|
| 592 |
+
},
|
| 593 |
+
{
|
| 594 |
+
"epoch": 0.85,
|
| 595 |
+
"learning_rate": 4.3806896551724144e-05,
|
| 596 |
+
"loss": 0.2892,
|
| 597 |
+
"step": 2300
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 0.86,
|
| 601 |
+
"learning_rate": 4.3720689655172415e-05,
|
| 602 |
+
"loss": 0.2773,
|
| 603 |
+
"step": 2325
|
| 604 |
+
},
|
| 605 |
+
{
|
| 606 |
+
"epoch": 0.87,
|
| 607 |
+
"learning_rate": 4.363448275862069e-05,
|
| 608 |
+
"loss": 0.2733,
|
| 609 |
+
"step": 2350
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 0.88,
|
| 613 |
+
"learning_rate": 4.3548275862068964e-05,
|
| 614 |
+
"loss": 0.2805,
|
| 615 |
+
"step": 2375
|
| 616 |
+
},
|
| 617 |
+
{
|
| 618 |
+
"epoch": 0.89,
|
| 619 |
+
"learning_rate": 4.346206896551724e-05,
|
| 620 |
+
"loss": 0.274,
|
| 621 |
+
"step": 2400
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 0.9,
|
| 625 |
+
"learning_rate": 4.337586206896552e-05,
|
| 626 |
+
"loss": 0.2711,
|
| 627 |
+
"step": 2425
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.91,
|
| 631 |
+
"learning_rate": 4.32896551724138e-05,
|
| 632 |
+
"loss": 0.2684,
|
| 633 |
+
"step": 2450
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 0.92,
|
| 637 |
+
"learning_rate": 4.320344827586207e-05,
|
| 638 |
+
"loss": 0.2731,
|
| 639 |
+
"step": 2475
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 0.93,
|
| 643 |
+
"learning_rate": 4.3117241379310346e-05,
|
| 644 |
+
"loss": 0.2689,
|
| 645 |
+
"step": 2500
|
| 646 |
+
},
|
| 647 |
+
{
|
| 648 |
+
"epoch": 0.93,
|
| 649 |
+
"eval_cer": 13.10282074613285,
|
| 650 |
+
"eval_loss": 0.24490514397621155,
|
| 651 |
+
"eval_runtime": 366.5991,
|
| 652 |
+
"eval_samples_per_second": 7.04,
|
| 653 |
+
"eval_steps_per_second": 0.881,
|
| 654 |
+
"step": 2500
|
| 655 |
+
},
|
| 656 |
+
{
|
| 657 |
+
"epoch": 0.93,
|
| 658 |
+
"learning_rate": 4.3031034482758624e-05,
|
| 659 |
+
"loss": 0.2702,
|
| 660 |
+
"step": 2525
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 0.94,
|
| 664 |
+
"learning_rate": 4.29448275862069e-05,
|
| 665 |
+
"loss": 0.2623,
|
| 666 |
+
"step": 2550
|
| 667 |
+
},
|
| 668 |
+
{
|
| 669 |
+
"epoch": 0.95,
|
| 670 |
+
"learning_rate": 4.285862068965518e-05,
|
| 671 |
+
"loss": 0.2749,
|
| 672 |
+
"step": 2575
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 0.96,
|
| 676 |
+
"learning_rate": 4.277241379310345e-05,
|
| 677 |
+
"loss": 0.2658,
|
| 678 |
+
"step": 2600
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 0.97,
|
| 682 |
+
"learning_rate": 4.268620689655173e-05,
|
| 683 |
+
"loss": 0.2637,
|
| 684 |
+
"step": 2625
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 0.98,
|
| 688 |
+
"learning_rate": 4.26e-05,
|
| 689 |
+
"loss": 0.2823,
|
| 690 |
+
"step": 2650
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.99,
|
| 694 |
+
"learning_rate": 4.2513793103448276e-05,
|
| 695 |
+
"loss": 0.2641,
|
| 696 |
+
"step": 2675
|
| 697 |
+
},
|
| 698 |
+
{
|
| 699 |
+
"epoch": 1.0,
|
| 700 |
+
"learning_rate": 4.2427586206896554e-05,
|
| 701 |
+
"loss": 0.2721,
|
| 702 |
+
"step": 2700
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 1.01,
|
| 706 |
+
"learning_rate": 4.234137931034483e-05,
|
| 707 |
+
"loss": 0.1882,
|
| 708 |
+
"step": 2725
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.02,
|
| 712 |
+
"learning_rate": 4.22551724137931e-05,
|
| 713 |
+
"loss": 0.1776,
|
| 714 |
+
"step": 2750
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.03,
|
| 718 |
+
"learning_rate": 4.216896551724138e-05,
|
| 719 |
+
"loss": 0.1768,
|
| 720 |
+
"step": 2775
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"epoch": 1.04,
|
| 724 |
+
"learning_rate": 4.208275862068966e-05,
|
| 725 |
+
"loss": 0.1714,
|
| 726 |
+
"step": 2800
|
| 727 |
+
},
|
| 728 |
+
{
|
| 729 |
+
"epoch": 1.05,
|
| 730 |
+
"learning_rate": 4.1996551724137936e-05,
|
| 731 |
+
"loss": 0.1771,
|
| 732 |
+
"step": 2825
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 1.05,
|
| 736 |
+
"learning_rate": 4.191034482758621e-05,
|
| 737 |
+
"loss": 0.193,
|
| 738 |
+
"step": 2850
|
| 739 |
+
},
|
| 740 |
+
{
|
| 741 |
+
"epoch": 1.06,
|
| 742 |
+
"learning_rate": 4.1824137931034485e-05,
|
| 743 |
+
"loss": 0.1788,
|
| 744 |
+
"step": 2875
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.07,
|
| 748 |
+
"learning_rate": 4.1737931034482756e-05,
|
| 749 |
+
"loss": 0.1784,
|
| 750 |
+
"step": 2900
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.08,
|
| 754 |
+
"learning_rate": 4.1651724137931034e-05,
|
| 755 |
+
"loss": 0.1879,
|
| 756 |
+
"step": 2925
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 1.09,
|
| 760 |
+
"learning_rate": 4.156551724137932e-05,
|
| 761 |
+
"loss": 0.1825,
|
| 762 |
+
"step": 2950
|
| 763 |
+
},
|
| 764 |
+
{
|
| 765 |
+
"epoch": 1.1,
|
| 766 |
+
"learning_rate": 4.147931034482759e-05,
|
| 767 |
+
"loss": 0.1854,
|
| 768 |
+
"step": 2975
|
| 769 |
+
},
|
| 770 |
+
{
|
| 771 |
+
"epoch": 1.11,
|
| 772 |
+
"learning_rate": 4.139310344827587e-05,
|
| 773 |
+
"loss": 0.1882,
|
| 774 |
+
"step": 3000
|
| 775 |
+
},
|
| 776 |
+
{
|
| 777 |
+
"epoch": 1.11,
|
| 778 |
+
"eval_cer": 11.67728237791932,
|
| 779 |
+
"eval_loss": 0.25736209750175476,
|
| 780 |
+
"eval_runtime": 361.0573,
|
| 781 |
+
"eval_samples_per_second": 7.148,
|
| 782 |
+
"eval_steps_per_second": 0.895,
|
| 783 |
+
"step": 3000
|
| 784 |
+
},
|
| 785 |
+
{
|
| 786 |
+
"epoch": 1.12,
|
| 787 |
+
"learning_rate": 4.130689655172414e-05,
|
| 788 |
+
"loss": 0.1827,
|
| 789 |
+
"step": 3025
|
| 790 |
+
},
|
| 791 |
+
{
|
| 792 |
+
"epoch": 1.13,
|
| 793 |
+
"learning_rate": 4.1220689655172415e-05,
|
| 794 |
+
"loss": 0.1889,
|
| 795 |
+
"step": 3050
|
| 796 |
+
},
|
| 797 |
+
{
|
| 798 |
+
"epoch": 1.14,
|
| 799 |
+
"learning_rate": 4.113448275862069e-05,
|
| 800 |
+
"loss": 0.1938,
|
| 801 |
+
"step": 3075
|
| 802 |
+
},
|
| 803 |
+
{
|
| 804 |
+
"epoch": 1.15,
|
| 805 |
+
"learning_rate": 4.104827586206897e-05,
|
| 806 |
+
"loss": 0.182,
|
| 807 |
+
"step": 3100
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 1.16,
|
| 811 |
+
"learning_rate": 4.096206896551724e-05,
|
| 812 |
+
"loss": 0.1884,
|
| 813 |
+
"step": 3125
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 1.17,
|
| 817 |
+
"learning_rate": 4.087586206896552e-05,
|
| 818 |
+
"loss": 0.182,
|
| 819 |
+
"step": 3150
|
| 820 |
+
},
|
| 821 |
+
{
|
| 822 |
+
"epoch": 1.18,
|
| 823 |
+
"learning_rate": 4.078965517241379e-05,
|
| 824 |
+
"loss": 0.1795,
|
| 825 |
+
"step": 3175
|
| 826 |
+
},
|
| 827 |
+
{
|
| 828 |
+
"epoch": 1.18,
|
| 829 |
+
"learning_rate": 4.070344827586207e-05,
|
| 830 |
+
"loss": 0.176,
|
| 831 |
+
"step": 3200
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.19,
|
| 835 |
+
"learning_rate": 4.0617241379310346e-05,
|
| 836 |
+
"loss": 0.1834,
|
| 837 |
+
"step": 3225
|
| 838 |
+
},
|
| 839 |
+
{
|
| 840 |
+
"epoch": 1.2,
|
| 841 |
+
"learning_rate": 4.0531034482758624e-05,
|
| 842 |
+
"loss": 0.1886,
|
| 843 |
+
"step": 3250
|
| 844 |
+
},
|
| 845 |
+
{
|
| 846 |
+
"epoch": 1.21,
|
| 847 |
+
"learning_rate": 4.0444827586206895e-05,
|
| 848 |
+
"loss": 0.1908,
|
| 849 |
+
"step": 3275
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 1.22,
|
| 853 |
+
"learning_rate": 4.035862068965517e-05,
|
| 854 |
+
"loss": 0.1813,
|
| 855 |
+
"step": 3300
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 1.23,
|
| 859 |
+
"learning_rate": 4.027241379310345e-05,
|
| 860 |
+
"loss": 0.1875,
|
| 861 |
+
"step": 3325
|
| 862 |
+
},
|
| 863 |
+
{
|
| 864 |
+
"epoch": 1.24,
|
| 865 |
+
"learning_rate": 4.018620689655173e-05,
|
| 866 |
+
"loss": 0.1925,
|
| 867 |
+
"step": 3350
|
| 868 |
+
},
|
| 869 |
+
{
|
| 870 |
+
"epoch": 1.25,
|
| 871 |
+
"learning_rate": 4.0100000000000006e-05,
|
| 872 |
+
"loss": 0.1855,
|
| 873 |
+
"step": 3375
|
| 874 |
+
},
|
| 875 |
+
{
|
| 876 |
+
"epoch": 1.26,
|
| 877 |
+
"learning_rate": 4.001379310344828e-05,
|
| 878 |
+
"loss": 0.194,
|
| 879 |
+
"step": 3400
|
| 880 |
+
},
|
| 881 |
+
{
|
| 882 |
+
"epoch": 1.27,
|
| 883 |
+
"learning_rate": 3.9927586206896554e-05,
|
| 884 |
+
"loss": 0.1827,
|
| 885 |
+
"step": 3425
|
| 886 |
+
},
|
| 887 |
+
{
|
| 888 |
+
"epoch": 1.28,
|
| 889 |
+
"learning_rate": 3.9841379310344825e-05,
|
| 890 |
+
"loss": 0.1806,
|
| 891 |
+
"step": 3450
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 1.29,
|
| 895 |
+
"learning_rate": 3.975517241379311e-05,
|
| 896 |
+
"loss": 0.1783,
|
| 897 |
+
"step": 3475
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.3,
|
| 901 |
+
"learning_rate": 3.966896551724138e-05,
|
| 902 |
+
"loss": 0.1889,
|
| 903 |
+
"step": 3500
|
| 904 |
+
},
|
| 905 |
+
{
|
| 906 |
+
"epoch": 1.3,
|
| 907 |
+
"eval_cer": 11.434637549287231,
|
| 908 |
+
"eval_loss": 0.24476666748523712,
|
| 909 |
+
"eval_runtime": 361.1188,
|
| 910 |
+
"eval_samples_per_second": 7.147,
|
| 911 |
+
"eval_steps_per_second": 0.894,
|
| 912 |
+
"step": 3500
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.3,
|
| 916 |
+
"learning_rate": 3.958275862068966e-05,
|
| 917 |
+
"loss": 0.185,
|
| 918 |
+
"step": 3525
|
| 919 |
+
},
|
| 920 |
+
{
|
| 921 |
+
"epoch": 1.31,
|
| 922 |
+
"learning_rate": 3.949655172413793e-05,
|
| 923 |
+
"loss": 0.1833,
|
| 924 |
+
"step": 3550
|
| 925 |
+
},
|
| 926 |
+
{
|
| 927 |
+
"epoch": 1.32,
|
| 928 |
+
"learning_rate": 3.941034482758621e-05,
|
| 929 |
+
"loss": 0.1749,
|
| 930 |
+
"step": 3575
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"epoch": 1.33,
|
| 934 |
+
"learning_rate": 3.9324137931034485e-05,
|
| 935 |
+
"loss": 0.1906,
|
| 936 |
+
"step": 3600
|
| 937 |
+
},
|
| 938 |
+
{
|
| 939 |
+
"epoch": 1.34,
|
| 940 |
+
"learning_rate": 3.923793103448276e-05,
|
| 941 |
+
"loss": 0.1934,
|
| 942 |
+
"step": 3625
|
| 943 |
+
},
|
| 944 |
+
{
|
| 945 |
+
"epoch": 1.35,
|
| 946 |
+
"learning_rate": 3.9151724137931034e-05,
|
| 947 |
+
"loss": 0.1782,
|
| 948 |
+
"step": 3650
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.36,
|
| 952 |
+
"learning_rate": 3.906551724137931e-05,
|
| 953 |
+
"loss": 0.19,
|
| 954 |
+
"step": 3675
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 1.37,
|
| 958 |
+
"learning_rate": 3.897931034482758e-05,
|
| 959 |
+
"loss": 0.1943,
|
| 960 |
+
"step": 3700
|
| 961 |
+
},
|
| 962 |
+
{
|
| 963 |
+
"epoch": 1.38,
|
| 964 |
+
"learning_rate": 3.889310344827587e-05,
|
| 965 |
+
"loss": 0.1838,
|
| 966 |
+
"step": 3725
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 1.39,
|
| 970 |
+
"learning_rate": 3.8806896551724145e-05,
|
| 971 |
+
"loss": 0.2248,
|
| 972 |
+
"step": 3750
|
| 973 |
+
},
|
| 974 |
+
{
|
| 975 |
+
"epoch": 1.4,
|
| 976 |
+
"learning_rate": 3.8720689655172416e-05,
|
| 977 |
+
"loss": 0.1853,
|
| 978 |
+
"step": 3775
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.41,
|
| 982 |
+
"learning_rate": 3.863448275862069e-05,
|
| 983 |
+
"loss": 0.1867,
|
| 984 |
+
"step": 3800
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 1.42,
|
| 988 |
+
"learning_rate": 3.8548275862068964e-05,
|
| 989 |
+
"loss": 0.1801,
|
| 990 |
+
"step": 3825
|
| 991 |
+
},
|
| 992 |
+
{
|
| 993 |
+
"epoch": 1.42,
|
| 994 |
+
"learning_rate": 3.846206896551724e-05,
|
| 995 |
+
"loss": 0.182,
|
| 996 |
+
"step": 3850
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 1.43,
|
| 1000 |
+
"learning_rate": 3.837586206896552e-05,
|
| 1001 |
+
"loss": 0.1843,
|
| 1002 |
+
"step": 3875
|
| 1003 |
+
},
|
| 1004 |
+
{
|
| 1005 |
+
"epoch": 1.44,
|
| 1006 |
+
"learning_rate": 3.82896551724138e-05,
|
| 1007 |
+
"loss": 0.184,
|
| 1008 |
+
"step": 3900
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 1.45,
|
| 1012 |
+
"learning_rate": 3.820344827586207e-05,
|
| 1013 |
+
"loss": 0.1808,
|
| 1014 |
+
"step": 3925
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.46,
|
| 1018 |
+
"learning_rate": 3.8117241379310346e-05,
|
| 1019 |
+
"loss": 0.188,
|
| 1020 |
+
"step": 3950
|
| 1021 |
+
},
|
| 1022 |
+
{
|
| 1023 |
+
"epoch": 1.47,
|
| 1024 |
+
"learning_rate": 3.8031034482758624e-05,
|
| 1025 |
+
"loss": 0.1691,
|
| 1026 |
+
"step": 3975
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 1.48,
|
| 1030 |
+
"learning_rate": 3.79448275862069e-05,
|
| 1031 |
+
"loss": 0.1923,
|
| 1032 |
+
"step": 4000
|
| 1033 |
+
},
|
| 1034 |
+
{
|
| 1035 |
+
"epoch": 1.48,
|
| 1036 |
+
"eval_cer": 11.8895966029724,
|
| 1037 |
+
"eval_loss": 0.24924282729625702,
|
| 1038 |
+
"eval_runtime": 360.2098,
|
| 1039 |
+
"eval_samples_per_second": 7.165,
|
| 1040 |
+
"eval_steps_per_second": 0.897,
|
| 1041 |
+
"step": 4000
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 1.49,
|
| 1045 |
+
"learning_rate": 3.785862068965517e-05,
|
| 1046 |
+
"loss": 0.175,
|
| 1047 |
+
"step": 4025
|
| 1048 |
+
},
|
| 1049 |
+
{
|
| 1050 |
+
"epoch": 1.5,
|
| 1051 |
+
"learning_rate": 3.777241379310345e-05,
|
| 1052 |
+
"loss": 0.1814,
|
| 1053 |
+
"step": 4050
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 1.51,
|
| 1057 |
+
"learning_rate": 3.768620689655172e-05,
|
| 1058 |
+
"loss": 0.1808,
|
| 1059 |
+
"step": 4075
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 1.52,
|
| 1063 |
+
"learning_rate": 3.76e-05,
|
| 1064 |
+
"loss": 0.1875,
|
| 1065 |
+
"step": 4100
|
| 1066 |
+
},
|
| 1067 |
+
{
|
| 1068 |
+
"epoch": 1.53,
|
| 1069 |
+
"learning_rate": 3.7513793103448284e-05,
|
| 1070 |
+
"loss": 0.1831,
|
| 1071 |
+
"step": 4125
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 1.54,
|
| 1075 |
+
"learning_rate": 3.7427586206896555e-05,
|
| 1076 |
+
"loss": 0.184,
|
| 1077 |
+
"step": 4150
|
| 1078 |
+
},
|
| 1079 |
+
{
|
| 1080 |
+
"epoch": 1.55,
|
| 1081 |
+
"learning_rate": 3.734137931034483e-05,
|
| 1082 |
+
"loss": 0.1774,
|
| 1083 |
+
"step": 4175
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 1.55,
|
| 1087 |
+
"learning_rate": 3.72551724137931e-05,
|
| 1088 |
+
"loss": 0.1716,
|
| 1089 |
+
"step": 4200
|
| 1090 |
+
},
|
| 1091 |
+
{
|
| 1092 |
+
"epoch": 1.56,
|
| 1093 |
+
"learning_rate": 3.716896551724138e-05,
|
| 1094 |
+
"loss": 0.1845,
|
| 1095 |
+
"step": 4225
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 1.57,
|
| 1099 |
+
"learning_rate": 3.708275862068966e-05,
|
| 1100 |
+
"loss": 0.1854,
|
| 1101 |
+
"step": 4250
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 1.58,
|
| 1105 |
+
"learning_rate": 3.6996551724137937e-05,
|
| 1106 |
+
"loss": 0.1824,
|
| 1107 |
+
"step": 4275
|
| 1108 |
+
},
|
| 1109 |
+
{
|
| 1110 |
+
"epoch": 1.59,
|
| 1111 |
+
"learning_rate": 3.691034482758621e-05,
|
| 1112 |
+
"loss": 0.1796,
|
| 1113 |
+
"step": 4300
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 1.6,
|
| 1117 |
+
"learning_rate": 3.6824137931034485e-05,
|
| 1118 |
+
"loss": 0.1882,
|
| 1119 |
+
"step": 4325
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 1.61,
|
| 1123 |
+
"learning_rate": 3.6737931034482756e-05,
|
| 1124 |
+
"loss": 0.1866,
|
| 1125 |
+
"step": 4350
|
| 1126 |
+
},
|
| 1127 |
+
{
|
| 1128 |
+
"epoch": 1.62,
|
| 1129 |
+
"learning_rate": 3.6651724137931034e-05,
|
| 1130 |
+
"loss": 0.1881,
|
| 1131 |
+
"step": 4375
|
| 1132 |
+
},
|
| 1133 |
+
{
|
| 1134 |
+
"epoch": 1.63,
|
| 1135 |
+
"learning_rate": 3.656551724137931e-05,
|
| 1136 |
+
"loss": 0.1822,
|
| 1137 |
+
"step": 4400
|
| 1138 |
+
},
|
| 1139 |
+
{
|
| 1140 |
+
"epoch": 1.64,
|
| 1141 |
+
"learning_rate": 3.647931034482759e-05,
|
| 1142 |
+
"loss": 0.1803,
|
| 1143 |
+
"step": 4425
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 1.65,
|
| 1147 |
+
"learning_rate": 3.639310344827586e-05,
|
| 1148 |
+
"loss": 0.1851,
|
| 1149 |
+
"step": 4450
|
| 1150 |
+
},
|
| 1151 |
+
{
|
| 1152 |
+
"epoch": 1.66,
|
| 1153 |
+
"learning_rate": 3.630689655172414e-05,
|
| 1154 |
+
"loss": 0.1825,
|
| 1155 |
+
"step": 4475
|
| 1156 |
+
},
|
| 1157 |
+
{
|
| 1158 |
+
"epoch": 1.67,
|
| 1159 |
+
"learning_rate": 3.6220689655172416e-05,
|
| 1160 |
+
"loss": 0.1886,
|
| 1161 |
+
"step": 4500
|
| 1162 |
+
},
|
| 1163 |
+
{
|
| 1164 |
+
"epoch": 1.67,
|
| 1165 |
+
"eval_cer": 11.75310888686685,
|
| 1166 |
+
"eval_loss": 0.23957248032093048,
|
| 1167 |
+
"eval_runtime": 361.2575,
|
| 1168 |
+
"eval_samples_per_second": 7.144,
|
| 1169 |
+
"eval_steps_per_second": 0.894,
|
| 1170 |
+
"step": 4500
|
| 1171 |
+
},
|
| 1172 |
+
{
|
| 1173 |
+
"epoch": 1.67,
|
| 1174 |
+
"learning_rate": 3.6134482758620694e-05,
|
| 1175 |
+
"loss": 0.1944,
|
| 1176 |
+
"step": 4525
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 1.68,
|
| 1180 |
+
"learning_rate": 3.604827586206897e-05,
|
| 1181 |
+
"loss": 0.1896,
|
| 1182 |
+
"step": 4550
|
| 1183 |
+
},
|
| 1184 |
+
{
|
| 1185 |
+
"epoch": 1.69,
|
| 1186 |
+
"learning_rate": 3.596206896551724e-05,
|
| 1187 |
+
"loss": 0.1819,
|
| 1188 |
+
"step": 4575
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 1.7,
|
| 1192 |
+
"learning_rate": 3.587586206896552e-05,
|
| 1193 |
+
"loss": 0.1892,
|
| 1194 |
+
"step": 4600
|
| 1195 |
+
},
|
| 1196 |
+
{
|
| 1197 |
+
"epoch": 1.71,
|
| 1198 |
+
"learning_rate": 3.578965517241379e-05,
|
| 1199 |
+
"loss": 0.1914,
|
| 1200 |
+
"step": 4625
|
| 1201 |
+
},
|
| 1202 |
+
{
|
| 1203 |
+
"epoch": 1.72,
|
| 1204 |
+
"learning_rate": 3.5703448275862076e-05,
|
| 1205 |
+
"loss": 0.181,
|
| 1206 |
+
"step": 4650
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 1.73,
|
| 1210 |
+
"learning_rate": 3.5617241379310346e-05,
|
| 1211 |
+
"loss": 0.1799,
|
| 1212 |
+
"step": 4675
|
| 1213 |
+
},
|
| 1214 |
+
{
|
| 1215 |
+
"epoch": 1.74,
|
| 1216 |
+
"learning_rate": 3.5531034482758624e-05,
|
| 1217 |
+
"loss": 0.1784,
|
| 1218 |
+
"step": 4700
|
| 1219 |
+
},
|
| 1220 |
+
{
|
| 1221 |
+
"epoch": 1.75,
|
| 1222 |
+
"learning_rate": 3.5444827586206895e-05,
|
| 1223 |
+
"loss": 0.1812,
|
| 1224 |
+
"step": 4725
|
| 1225 |
+
},
|
| 1226 |
+
{
|
| 1227 |
+
"epoch": 1.76,
|
| 1228 |
+
"learning_rate": 3.535862068965517e-05,
|
| 1229 |
+
"loss": 0.1782,
|
| 1230 |
+
"step": 4750
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 1.77,
|
| 1234 |
+
"learning_rate": 3.527241379310345e-05,
|
| 1235 |
+
"loss": 0.1783,
|
| 1236 |
+
"step": 4775
|
| 1237 |
+
},
|
| 1238 |
+
{
|
| 1239 |
+
"epoch": 1.78,
|
| 1240 |
+
"learning_rate": 3.518620689655173e-05,
|
| 1241 |
+
"loss": 0.2002,
|
| 1242 |
+
"step": 4800
|
| 1243 |
+
},
|
| 1244 |
+
{
|
| 1245 |
+
"epoch": 1.79,
|
| 1246 |
+
"learning_rate": 3.51e-05,
|
| 1247 |
+
"loss": 0.1948,
|
| 1248 |
+
"step": 4825
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 1.79,
|
| 1252 |
+
"learning_rate": 3.501379310344828e-05,
|
| 1253 |
+
"loss": 0.1791,
|
| 1254 |
+
"step": 4850
|
| 1255 |
+
},
|
| 1256 |
+
{
|
| 1257 |
+
"epoch": 1.8,
|
| 1258 |
+
"learning_rate": 3.492758620689655e-05,
|
| 1259 |
+
"loss": 0.2018,
|
| 1260 |
+
"step": 4875
|
| 1261 |
+
},
|
| 1262 |
+
{
|
| 1263 |
+
"epoch": 1.81,
|
| 1264 |
+
"learning_rate": 3.484137931034483e-05,
|
| 1265 |
+
"loss": 0.1828,
|
| 1266 |
+
"step": 4900
|
| 1267 |
+
},
|
| 1268 |
+
{
|
| 1269 |
+
"epoch": 1.82,
|
| 1270 |
+
"learning_rate": 3.475517241379311e-05,
|
| 1271 |
+
"loss": 0.1834,
|
| 1272 |
+
"step": 4925
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 1.83,
|
| 1276 |
+
"learning_rate": 3.466896551724138e-05,
|
| 1277 |
+
"loss": 0.1821,
|
| 1278 |
+
"step": 4950
|
| 1279 |
+
},
|
| 1280 |
+
{
|
| 1281 |
+
"epoch": 1.84,
|
| 1282 |
+
"learning_rate": 3.458275862068966e-05,
|
| 1283 |
+
"loss": 0.1864,
|
| 1284 |
+
"step": 4975
|
| 1285 |
+
},
|
| 1286 |
+
{
|
| 1287 |
+
"epoch": 1.85,
|
| 1288 |
+
"learning_rate": 3.449655172413793e-05,
|
| 1289 |
+
"loss": 0.1899,
|
| 1290 |
+
"step": 5000
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 1.85,
|
| 1294 |
+
"eval_cer": 12.048832271762208,
|
| 1295 |
+
"eval_loss": 0.23497651517391205,
|
| 1296 |
+
"eval_runtime": 363.6465,
|
| 1297 |
+
"eval_samples_per_second": 7.098,
|
| 1298 |
+
"eval_steps_per_second": 0.888,
|
| 1299 |
+
"step": 5000
|
| 1300 |
+
},
|
| 1301 |
+
{
|
| 1302 |
+
"epoch": 1.86,
|
| 1303 |
+
"learning_rate": 3.441034482758621e-05,
|
| 1304 |
+
"loss": 0.1805,
|
| 1305 |
+
"step": 5025
|
| 1306 |
+
},
|
| 1307 |
+
{
|
| 1308 |
+
"epoch": 1.87,
|
| 1309 |
+
"learning_rate": 3.4324137931034485e-05,
|
| 1310 |
+
"loss": 0.1826,
|
| 1311 |
+
"step": 5050
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 1.88,
|
| 1315 |
+
"learning_rate": 3.423793103448276e-05,
|
| 1316 |
+
"loss": 0.1863,
|
| 1317 |
+
"step": 5075
|
| 1318 |
+
},
|
| 1319 |
+
{
|
| 1320 |
+
"epoch": 1.89,
|
| 1321 |
+
"learning_rate": 3.4151724137931034e-05,
|
| 1322 |
+
"loss": 0.1754,
|
| 1323 |
+
"step": 5100
|
| 1324 |
+
},
|
| 1325 |
+
{
|
| 1326 |
+
"epoch": 1.9,
|
| 1327 |
+
"learning_rate": 3.406551724137931e-05,
|
| 1328 |
+
"loss": 0.1801,
|
| 1329 |
+
"step": 5125
|
| 1330 |
+
},
|
| 1331 |
+
{
|
| 1332 |
+
"epoch": 1.91,
|
| 1333 |
+
"learning_rate": 3.397931034482758e-05,
|
| 1334 |
+
"loss": 0.1846,
|
| 1335 |
+
"step": 5150
|
| 1336 |
+
},
|
| 1337 |
+
{
|
| 1338 |
+
"epoch": 1.92,
|
| 1339 |
+
"learning_rate": 3.389310344827587e-05,
|
| 1340 |
+
"loss": 0.1717,
|
| 1341 |
+
"step": 5175
|
| 1342 |
+
},
|
| 1343 |
+
{
|
| 1344 |
+
"epoch": 1.92,
|
| 1345 |
+
"learning_rate": 3.380689655172414e-05,
|
| 1346 |
+
"loss": 0.1754,
|
| 1347 |
+
"step": 5200
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.93,
|
| 1351 |
+
"learning_rate": 3.3720689655172416e-05,
|
| 1352 |
+
"loss": 0.1819,
|
| 1353 |
+
"step": 5225
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 1.94,
|
| 1357 |
+
"learning_rate": 3.363448275862069e-05,
|
| 1358 |
+
"loss": 0.1792,
|
| 1359 |
+
"step": 5250
|
| 1360 |
+
},
|
| 1361 |
+
{
|
| 1362 |
+
"epoch": 1.95,
|
| 1363 |
+
"learning_rate": 3.3548275862068965e-05,
|
| 1364 |
+
"loss": 0.1837,
|
| 1365 |
+
"step": 5275
|
| 1366 |
+
},
|
| 1367 |
+
{
|
| 1368 |
+
"epoch": 1.96,
|
| 1369 |
+
"learning_rate": 3.346206896551724e-05,
|
| 1370 |
+
"loss": 0.188,
|
| 1371 |
+
"step": 5300
|
| 1372 |
+
},
|
| 1373 |
+
{
|
| 1374 |
+
"epoch": 1.97,
|
| 1375 |
+
"learning_rate": 3.337586206896552e-05,
|
| 1376 |
+
"loss": 0.1875,
|
| 1377 |
+
"step": 5325
|
| 1378 |
+
},
|
| 1379 |
+
{
|
| 1380 |
+
"epoch": 1.98,
|
| 1381 |
+
"learning_rate": 3.32896551724138e-05,
|
| 1382 |
+
"loss": 0.1909,
|
| 1383 |
+
"step": 5350
|
| 1384 |
+
},
|
| 1385 |
+
{
|
| 1386 |
+
"epoch": 1.99,
|
| 1387 |
+
"learning_rate": 3.320344827586207e-05,
|
| 1388 |
+
"loss": 0.1714,
|
| 1389 |
+
"step": 5375
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 2.0,
|
| 1393 |
+
"learning_rate": 3.311724137931035e-05,
|
| 1394 |
+
"loss": 0.1921,
|
| 1395 |
+
"step": 5400
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 2.01,
|
| 1399 |
+
"learning_rate": 3.3031034482758624e-05,
|
| 1400 |
+
"loss": 0.1109,
|
| 1401 |
+
"step": 5425
|
| 1402 |
+
},
|
| 1403 |
+
{
|
| 1404 |
+
"epoch": 2.02,
|
| 1405 |
+
"learning_rate": 3.29448275862069e-05,
|
| 1406 |
+
"loss": 0.0946,
|
| 1407 |
+
"step": 5450
|
| 1408 |
+
},
|
| 1409 |
+
{
|
| 1410 |
+
"epoch": 2.03,
|
| 1411 |
+
"learning_rate": 3.285862068965517e-05,
|
| 1412 |
+
"loss": 0.0967,
|
| 1413 |
+
"step": 5475
|
| 1414 |
+
},
|
| 1415 |
+
{
|
| 1416 |
+
"epoch": 2.04,
|
| 1417 |
+
"learning_rate": 3.277241379310345e-05,
|
| 1418 |
+
"loss": 0.1009,
|
| 1419 |
+
"step": 5500
|
| 1420 |
+
},
|
| 1421 |
+
{
|
| 1422 |
+
"epoch": 2.04,
|
| 1423 |
+
"eval_cer": 11.404306945708221,
|
| 1424 |
+
"eval_loss": 0.24928008019924164,
|
| 1425 |
+
"eval_runtime": 359.254,
|
| 1426 |
+
"eval_samples_per_second": 7.184,
|
| 1427 |
+
"eval_steps_per_second": 0.899,
|
| 1428 |
+
"step": 5500
|
| 1429 |
+
},
|
| 1430 |
+
{
|
| 1431 |
+
"epoch": 2.04,
|
| 1432 |
+
"learning_rate": 3.268620689655172e-05,
|
| 1433 |
+
"loss": 0.0954,
|
| 1434 |
+
"step": 5525
|
| 1435 |
+
},
|
| 1436 |
+
{
|
| 1437 |
+
"epoch": 2.05,
|
| 1438 |
+
"learning_rate": 3.26e-05,
|
| 1439 |
+
"loss": 0.0999,
|
| 1440 |
+
"step": 5550
|
| 1441 |
+
},
|
| 1442 |
+
{
|
| 1443 |
+
"epoch": 2.06,
|
| 1444 |
+
"learning_rate": 3.251379310344828e-05,
|
| 1445 |
+
"loss": 0.0946,
|
| 1446 |
+
"step": 5575
|
| 1447 |
+
},
|
| 1448 |
+
{
|
| 1449 |
+
"epoch": 2.07,
|
| 1450 |
+
"learning_rate": 3.2427586206896555e-05,
|
| 1451 |
+
"loss": 0.0949,
|
| 1452 |
+
"step": 5600
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 2.08,
|
| 1456 |
+
"learning_rate": 3.2341379310344826e-05,
|
| 1457 |
+
"loss": 0.1032,
|
| 1458 |
+
"step": 5625
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 2.09,
|
| 1462 |
+
"learning_rate": 3.2255172413793104e-05,
|
| 1463 |
+
"loss": 0.1002,
|
| 1464 |
+
"step": 5650
|
| 1465 |
+
},
|
| 1466 |
+
{
|
| 1467 |
+
"epoch": 2.1,
|
| 1468 |
+
"learning_rate": 3.216896551724138e-05,
|
| 1469 |
+
"loss": 0.0973,
|
| 1470 |
+
"step": 5675
|
| 1471 |
+
},
|
| 1472 |
+
{
|
| 1473 |
+
"epoch": 2.11,
|
| 1474 |
+
"learning_rate": 3.208275862068966e-05,
|
| 1475 |
+
"loss": 0.0984,
|
| 1476 |
+
"step": 5700
|
| 1477 |
+
},
|
| 1478 |
+
{
|
| 1479 |
+
"epoch": 2.12,
|
| 1480 |
+
"learning_rate": 3.199655172413794e-05,
|
| 1481 |
+
"loss": 0.1031,
|
| 1482 |
+
"step": 5725
|
| 1483 |
+
},
|
| 1484 |
+
{
|
| 1485 |
+
"epoch": 2.13,
|
| 1486 |
+
"learning_rate": 3.191034482758621e-05,
|
| 1487 |
+
"loss": 0.0968,
|
| 1488 |
+
"step": 5750
|
| 1489 |
+
},
|
| 1490 |
+
{
|
| 1491 |
+
"epoch": 2.14,
|
| 1492 |
+
"learning_rate": 3.1824137931034486e-05,
|
| 1493 |
+
"loss": 0.0983,
|
| 1494 |
+
"step": 5775
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 2.15,
|
| 1498 |
+
"learning_rate": 3.173793103448276e-05,
|
| 1499 |
+
"loss": 0.1088,
|
| 1500 |
+
"step": 5800
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 2.16,
|
| 1504 |
+
"learning_rate": 3.165172413793104e-05,
|
| 1505 |
+
"loss": 0.0998,
|
| 1506 |
+
"step": 5825
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 2.17,
|
| 1510 |
+
"learning_rate": 3.156551724137931e-05,
|
| 1511 |
+
"loss": 0.0991,
|
| 1512 |
+
"step": 5850
|
| 1513 |
+
},
|
| 1514 |
+
{
|
| 1515 |
+
"epoch": 2.17,
|
| 1516 |
+
"learning_rate": 3.147931034482759e-05,
|
| 1517 |
+
"loss": 0.103,
|
| 1518 |
+
"step": 5875
|
| 1519 |
+
},
|
| 1520 |
+
{
|
| 1521 |
+
"epoch": 2.18,
|
| 1522 |
+
"learning_rate": 3.139310344827586e-05,
|
| 1523 |
+
"loss": 0.1046,
|
| 1524 |
+
"step": 5900
|
| 1525 |
+
},
|
| 1526 |
+
{
|
| 1527 |
+
"epoch": 2.19,
|
| 1528 |
+
"learning_rate": 3.130689655172414e-05,
|
| 1529 |
+
"loss": 0.099,
|
| 1530 |
+
"step": 5925
|
| 1531 |
+
},
|
| 1532 |
+
{
|
| 1533 |
+
"epoch": 2.2,
|
| 1534 |
+
"learning_rate": 3.1220689655172416e-05,
|
| 1535 |
+
"loss": 0.1002,
|
| 1536 |
+
"step": 5950
|
| 1537 |
+
},
|
| 1538 |
+
{
|
| 1539 |
+
"epoch": 2.21,
|
| 1540 |
+
"learning_rate": 3.1134482758620694e-05,
|
| 1541 |
+
"loss": 0.1028,
|
| 1542 |
+
"step": 5975
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 2.22,
|
| 1546 |
+
"learning_rate": 3.1048275862068965e-05,
|
| 1547 |
+
"loss": 0.1003,
|
| 1548 |
+
"step": 6000
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 2.22,
|
| 1552 |
+
"eval_cer": 11.373976342129207,
|
| 1553 |
+
"eval_loss": 0.2572503387928009,
|
| 1554 |
+
"eval_runtime": 359.7119,
|
| 1555 |
+
"eval_samples_per_second": 7.175,
|
| 1556 |
+
"eval_steps_per_second": 0.898,
|
| 1557 |
+
"step": 6000
|
| 1558 |
+
},
|
| 1559 |
+
{
|
| 1560 |
+
"epoch": 2.23,
|
| 1561 |
+
"learning_rate": 3.096206896551724e-05,
|
| 1562 |
+
"loss": 0.0954,
|
| 1563 |
+
"step": 6025
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 2.24,
|
| 1567 |
+
"learning_rate": 3.0875862068965514e-05,
|
| 1568 |
+
"loss": 0.0981,
|
| 1569 |
+
"step": 6050
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 2.25,
|
| 1573 |
+
"learning_rate": 3.07896551724138e-05,
|
| 1574 |
+
"loss": 0.1006,
|
| 1575 |
+
"step": 6075
|
| 1576 |
+
},
|
| 1577 |
+
{
|
| 1578 |
+
"epoch": 2.26,
|
| 1579 |
+
"learning_rate": 3.070344827586207e-05,
|
| 1580 |
+
"loss": 0.1005,
|
| 1581 |
+
"step": 6100
|
| 1582 |
+
},
|
| 1583 |
+
{
|
| 1584 |
+
"epoch": 2.27,
|
| 1585 |
+
"learning_rate": 3.061724137931035e-05,
|
| 1586 |
+
"loss": 0.1007,
|
| 1587 |
+
"step": 6125
|
| 1588 |
+
},
|
| 1589 |
+
{
|
| 1590 |
+
"epoch": 2.28,
|
| 1591 |
+
"learning_rate": 3.0531034482758625e-05,
|
| 1592 |
+
"loss": 0.1005,
|
| 1593 |
+
"step": 6150
|
| 1594 |
+
},
|
| 1595 |
+
{
|
| 1596 |
+
"epoch": 2.29,
|
| 1597 |
+
"learning_rate": 3.04448275862069e-05,
|
| 1598 |
+
"loss": 0.0968,
|
| 1599 |
+
"step": 6175
|
| 1600 |
+
},
|
| 1601 |
+
{
|
| 1602 |
+
"epoch": 2.29,
|
| 1603 |
+
"learning_rate": 3.0358620689655177e-05,
|
| 1604 |
+
"loss": 0.1034,
|
| 1605 |
+
"step": 6200
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 2.3,
|
| 1609 |
+
"learning_rate": 3.0272413793103448e-05,
|
| 1610 |
+
"loss": 0.0982,
|
| 1611 |
+
"step": 6225
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 2.31,
|
| 1615 |
+
"learning_rate": 3.018620689655173e-05,
|
| 1616 |
+
"loss": 0.1033,
|
| 1617 |
+
"step": 6250
|
| 1618 |
+
},
|
| 1619 |
+
{
|
| 1620 |
+
"epoch": 2.32,
|
| 1621 |
+
"learning_rate": 3.01e-05,
|
| 1622 |
+
"loss": 0.1031,
|
| 1623 |
+
"step": 6275
|
| 1624 |
+
},
|
| 1625 |
+
{
|
| 1626 |
+
"epoch": 2.33,
|
| 1627 |
+
"learning_rate": 3.0013793103448278e-05,
|
| 1628 |
+
"loss": 0.0985,
|
| 1629 |
+
"step": 6300
|
| 1630 |
+
},
|
| 1631 |
+
{
|
| 1632 |
+
"epoch": 2.34,
|
| 1633 |
+
"learning_rate": 2.9931034482758624e-05,
|
| 1634 |
+
"loss": 0.1405,
|
| 1635 |
+
"step": 6325
|
| 1636 |
+
},
|
| 1637 |
+
{
|
| 1638 |
+
"epoch": 2.35,
|
| 1639 |
+
"learning_rate": 2.9844827586206898e-05,
|
| 1640 |
+
"loss": 0.1059,
|
| 1641 |
+
"step": 6350
|
| 1642 |
+
},
|
| 1643 |
+
{
|
| 1644 |
+
"epoch": 2.36,
|
| 1645 |
+
"learning_rate": 2.9758620689655176e-05,
|
| 1646 |
+
"loss": 0.098,
|
| 1647 |
+
"step": 6375
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 2.37,
|
| 1651 |
+
"learning_rate": 2.967241379310345e-05,
|
| 1652 |
+
"loss": 0.1029,
|
| 1653 |
+
"step": 6400
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 2.38,
|
| 1657 |
+
"learning_rate": 2.9586206896551728e-05,
|
| 1658 |
+
"loss": 0.1082,
|
| 1659 |
+
"step": 6425
|
| 1660 |
+
},
|
| 1661 |
+
{
|
| 1662 |
+
"epoch": 2.39,
|
| 1663 |
+
"learning_rate": 2.95e-05,
|
| 1664 |
+
"loss": 0.1049,
|
| 1665 |
+
"step": 6450
|
| 1666 |
+
},
|
| 1667 |
+
{
|
| 1668 |
+
"epoch": 2.4,
|
| 1669 |
+
"learning_rate": 2.941379310344828e-05,
|
| 1670 |
+
"loss": 0.0996,
|
| 1671 |
+
"step": 6475
|
| 1672 |
+
},
|
| 1673 |
+
{
|
| 1674 |
+
"epoch": 2.41,
|
| 1675 |
+
"learning_rate": 2.932758620689655e-05,
|
| 1676 |
+
"loss": 0.1031,
|
| 1677 |
+
"step": 6500
|
| 1678 |
+
},
|
| 1679 |
+
{
|
| 1680 |
+
"epoch": 2.41,
|
| 1681 |
+
"eval_cer": 12.954959053685169,
|
| 1682 |
+
"eval_loss": 0.26603972911834717,
|
| 1683 |
+
"eval_runtime": 369.9518,
|
| 1684 |
+
"eval_samples_per_second": 6.977,
|
| 1685 |
+
"eval_steps_per_second": 0.873,
|
| 1686 |
+
"step": 6500
|
| 1687 |
+
},
|
| 1688 |
+
{
|
| 1689 |
+
"epoch": 2.41,
|
| 1690 |
+
"learning_rate": 2.924137931034483e-05,
|
| 1691 |
+
"loss": 0.1062,
|
| 1692 |
+
"step": 6525
|
| 1693 |
+
},
|
| 1694 |
+
{
|
| 1695 |
+
"epoch": 2.42,
|
| 1696 |
+
"learning_rate": 2.9155172413793103e-05,
|
| 1697 |
+
"loss": 0.1057,
|
| 1698 |
+
"step": 6550
|
| 1699 |
+
},
|
| 1700 |
+
{
|
| 1701 |
+
"epoch": 2.43,
|
| 1702 |
+
"learning_rate": 2.906896551724138e-05,
|
| 1703 |
+
"loss": 0.1063,
|
| 1704 |
+
"step": 6575
|
| 1705 |
+
},
|
| 1706 |
+
{
|
| 1707 |
+
"epoch": 2.44,
|
| 1708 |
+
"learning_rate": 2.8982758620689655e-05,
|
| 1709 |
+
"loss": 0.1064,
|
| 1710 |
+
"step": 6600
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 2.45,
|
| 1714 |
+
"learning_rate": 2.8896551724137933e-05,
|
| 1715 |
+
"loss": 0.1026,
|
| 1716 |
+
"step": 6625
|
| 1717 |
+
},
|
| 1718 |
+
{
|
| 1719 |
+
"epoch": 2.46,
|
| 1720 |
+
"learning_rate": 2.881034482758621e-05,
|
| 1721 |
+
"loss": 0.1068,
|
| 1722 |
+
"step": 6650
|
| 1723 |
+
},
|
| 1724 |
+
{
|
| 1725 |
+
"epoch": 2.47,
|
| 1726 |
+
"learning_rate": 2.8724137931034485e-05,
|
| 1727 |
+
"loss": 0.1008,
|
| 1728 |
+
"step": 6675
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 2.48,
|
| 1732 |
+
"learning_rate": 2.8637931034482763e-05,
|
| 1733 |
+
"loss": 0.1118,
|
| 1734 |
+
"step": 6700
|
| 1735 |
+
},
|
| 1736 |
+
{
|
| 1737 |
+
"epoch": 2.49,
|
| 1738 |
+
"learning_rate": 2.8551724137931034e-05,
|
| 1739 |
+
"loss": 0.1017,
|
| 1740 |
+
"step": 6725
|
| 1741 |
+
},
|
| 1742 |
+
{
|
| 1743 |
+
"epoch": 2.5,
|
| 1744 |
+
"learning_rate": 2.8465517241379315e-05,
|
| 1745 |
+
"loss": 0.1059,
|
| 1746 |
+
"step": 6750
|
| 1747 |
+
},
|
| 1748 |
+
{
|
| 1749 |
+
"epoch": 2.51,
|
| 1750 |
+
"learning_rate": 2.8379310344827586e-05,
|
| 1751 |
+
"loss": 0.1071,
|
| 1752 |
+
"step": 6775
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 2.52,
|
| 1756 |
+
"learning_rate": 2.8293103448275863e-05,
|
| 1757 |
+
"loss": 0.1057,
|
| 1758 |
+
"step": 6800
|
| 1759 |
+
},
|
| 1760 |
+
{
|
| 1761 |
+
"epoch": 2.53,
|
| 1762 |
+
"learning_rate": 2.8206896551724138e-05,
|
| 1763 |
+
"loss": 0.1067,
|
| 1764 |
+
"step": 6825
|
| 1765 |
+
},
|
| 1766 |
+
{
|
| 1767 |
+
"epoch": 2.54,
|
| 1768 |
+
"learning_rate": 2.8120689655172415e-05,
|
| 1769 |
+
"loss": 0.1068,
|
| 1770 |
+
"step": 6850
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 2.54,
|
| 1774 |
+
"learning_rate": 2.803448275862069e-05,
|
| 1775 |
+
"loss": 0.1175,
|
| 1776 |
+
"step": 6875
|
| 1777 |
+
},
|
| 1778 |
+
{
|
| 1779 |
+
"epoch": 2.55,
|
| 1780 |
+
"learning_rate": 2.7948275862068968e-05,
|
| 1781 |
+
"loss": 0.0992,
|
| 1782 |
+
"step": 6900
|
| 1783 |
+
},
|
| 1784 |
+
{
|
| 1785 |
+
"epoch": 2.56,
|
| 1786 |
+
"learning_rate": 2.7862068965517242e-05,
|
| 1787 |
+
"loss": 0.1085,
|
| 1788 |
+
"step": 6925
|
| 1789 |
+
},
|
| 1790 |
+
{
|
| 1791 |
+
"epoch": 2.57,
|
| 1792 |
+
"learning_rate": 2.777586206896552e-05,
|
| 1793 |
+
"loss": 0.1066,
|
| 1794 |
+
"step": 6950
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 2.58,
|
| 1798 |
+
"learning_rate": 2.768965517241379e-05,
|
| 1799 |
+
"loss": 0.1013,
|
| 1800 |
+
"step": 6975
|
| 1801 |
+
},
|
| 1802 |
+
{
|
| 1803 |
+
"epoch": 2.59,
|
| 1804 |
+
"learning_rate": 2.7603448275862072e-05,
|
| 1805 |
+
"loss": 0.1048,
|
| 1806 |
+
"step": 7000
|
| 1807 |
+
},
|
| 1808 |
+
{
|
| 1809 |
+
"epoch": 2.59,
|
| 1810 |
+
"eval_cer": 12.026084319077949,
|
| 1811 |
+
"eval_loss": 0.25693368911743164,
|
| 1812 |
+
"eval_runtime": 363.146,
|
| 1813 |
+
"eval_samples_per_second": 7.107,
|
| 1814 |
+
"eval_steps_per_second": 0.889,
|
| 1815 |
+
"step": 7000
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 2.6,
|
| 1819 |
+
"learning_rate": 2.7517241379310343e-05,
|
| 1820 |
+
"loss": 0.1017,
|
| 1821 |
+
"step": 7025
|
| 1822 |
+
},
|
| 1823 |
+
{
|
| 1824 |
+
"epoch": 2.61,
|
| 1825 |
+
"learning_rate": 2.743103448275862e-05,
|
| 1826 |
+
"loss": 0.1024,
|
| 1827 |
+
"step": 7050
|
| 1828 |
+
},
|
| 1829 |
+
{
|
| 1830 |
+
"epoch": 2.62,
|
| 1831 |
+
"learning_rate": 2.73448275862069e-05,
|
| 1832 |
+
"loss": 0.108,
|
| 1833 |
+
"step": 7075
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 2.63,
|
| 1837 |
+
"learning_rate": 2.7258620689655173e-05,
|
| 1838 |
+
"loss": 0.0979,
|
| 1839 |
+
"step": 7100
|
| 1840 |
+
},
|
| 1841 |
+
{
|
| 1842 |
+
"epoch": 2.64,
|
| 1843 |
+
"learning_rate": 2.717241379310345e-05,
|
| 1844 |
+
"loss": 0.1076,
|
| 1845 |
+
"step": 7125
|
| 1846 |
+
},
|
| 1847 |
+
{
|
| 1848 |
+
"epoch": 2.65,
|
| 1849 |
+
"learning_rate": 2.7086206896551725e-05,
|
| 1850 |
+
"loss": 0.1024,
|
| 1851 |
+
"step": 7150
|
| 1852 |
+
},
|
| 1853 |
+
{
|
| 1854 |
+
"epoch": 2.66,
|
| 1855 |
+
"learning_rate": 2.7000000000000002e-05,
|
| 1856 |
+
"loss": 0.1037,
|
| 1857 |
+
"step": 7175
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 2.66,
|
| 1861 |
+
"learning_rate": 2.6913793103448277e-05,
|
| 1862 |
+
"loss": 0.1063,
|
| 1863 |
+
"step": 7200
|
| 1864 |
+
},
|
| 1865 |
+
{
|
| 1866 |
+
"epoch": 2.67,
|
| 1867 |
+
"learning_rate": 2.6827586206896554e-05,
|
| 1868 |
+
"loss": 0.1006,
|
| 1869 |
+
"step": 7225
|
| 1870 |
+
},
|
| 1871 |
+
{
|
| 1872 |
+
"epoch": 2.68,
|
| 1873 |
+
"learning_rate": 2.674137931034483e-05,
|
| 1874 |
+
"loss": 0.1093,
|
| 1875 |
+
"step": 7250
|
| 1876 |
+
},
|
| 1877 |
+
{
|
| 1878 |
+
"epoch": 2.69,
|
| 1879 |
+
"learning_rate": 2.6655172413793107e-05,
|
| 1880 |
+
"loss": 0.1057,
|
| 1881 |
+
"step": 7275
|
| 1882 |
+
},
|
| 1883 |
+
{
|
| 1884 |
+
"epoch": 2.7,
|
| 1885 |
+
"learning_rate": 2.6568965517241378e-05,
|
| 1886 |
+
"loss": 0.1015,
|
| 1887 |
+
"step": 7300
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 2.71,
|
| 1891 |
+
"learning_rate": 2.648275862068966e-05,
|
| 1892 |
+
"loss": 0.1057,
|
| 1893 |
+
"step": 7325
|
| 1894 |
+
},
|
| 1895 |
+
{
|
| 1896 |
+
"epoch": 2.72,
|
| 1897 |
+
"learning_rate": 2.639655172413793e-05,
|
| 1898 |
+
"loss": 0.1067,
|
| 1899 |
+
"step": 7350
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 2.73,
|
| 1903 |
+
"learning_rate": 2.6310344827586207e-05,
|
| 1904 |
+
"loss": 0.104,
|
| 1905 |
+
"step": 7375
|
| 1906 |
+
},
|
| 1907 |
+
{
|
| 1908 |
+
"epoch": 2.74,
|
| 1909 |
+
"learning_rate": 2.6224137931034482e-05,
|
| 1910 |
+
"loss": 0.1035,
|
| 1911 |
+
"step": 7400
|
| 1912 |
+
},
|
| 1913 |
+
{
|
| 1914 |
+
"epoch": 2.75,
|
| 1915 |
+
"learning_rate": 2.613793103448276e-05,
|
| 1916 |
+
"loss": 0.1054,
|
| 1917 |
+
"step": 7425
|
| 1918 |
+
},
|
| 1919 |
+
{
|
| 1920 |
+
"epoch": 2.76,
|
| 1921 |
+
"learning_rate": 2.6051724137931037e-05,
|
| 1922 |
+
"loss": 0.1083,
|
| 1923 |
+
"step": 7450
|
| 1924 |
+
},
|
| 1925 |
+
{
|
| 1926 |
+
"epoch": 2.77,
|
| 1927 |
+
"learning_rate": 2.596551724137931e-05,
|
| 1928 |
+
"loss": 0.1081,
|
| 1929 |
+
"step": 7475
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 2.78,
|
| 1933 |
+
"learning_rate": 2.587931034482759e-05,
|
| 1934 |
+
"loss": 0.1056,
|
| 1935 |
+
"step": 7500
|
| 1936 |
+
},
|
| 1937 |
+
{
|
| 1938 |
+
"epoch": 2.78,
|
| 1939 |
+
"eval_cer": 11.631786472550804,
|
| 1940 |
+
"eval_loss": 0.26435503363609314,
|
| 1941 |
+
"eval_runtime": 358.991,
|
| 1942 |
+
"eval_samples_per_second": 7.19,
|
| 1943 |
+
"eval_steps_per_second": 0.9,
|
| 1944 |
+
"step": 7500
|
| 1945 |
+
},
|
| 1946 |
+
{
|
| 1947 |
+
"epoch": 2.78,
|
| 1948 |
+
"learning_rate": 2.5793103448275864e-05,
|
| 1949 |
+
"loss": 0.103,
|
| 1950 |
+
"step": 7525
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 2.79,
|
| 1954 |
+
"learning_rate": 2.570689655172414e-05,
|
| 1955 |
+
"loss": 0.1068,
|
| 1956 |
+
"step": 7550
|
| 1957 |
+
},
|
| 1958 |
+
{
|
| 1959 |
+
"epoch": 2.8,
|
| 1960 |
+
"learning_rate": 2.5620689655172416e-05,
|
| 1961 |
+
"loss": 0.1,
|
| 1962 |
+
"step": 7575
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 2.81,
|
| 1966 |
+
"learning_rate": 2.5534482758620693e-05,
|
| 1967 |
+
"loss": 0.101,
|
| 1968 |
+
"step": 7600
|
| 1969 |
+
},
|
| 1970 |
+
{
|
| 1971 |
+
"epoch": 2.82,
|
| 1972 |
+
"learning_rate": 2.5448275862068964e-05,
|
| 1973 |
+
"loss": 0.1086,
|
| 1974 |
+
"step": 7625
|
| 1975 |
+
},
|
| 1976 |
+
{
|
| 1977 |
+
"epoch": 2.83,
|
| 1978 |
+
"learning_rate": 2.5362068965517246e-05,
|
| 1979 |
+
"loss": 0.1039,
|
| 1980 |
+
"step": 7650
|
| 1981 |
+
},
|
| 1982 |
+
{
|
| 1983 |
+
"epoch": 2.84,
|
| 1984 |
+
"learning_rate": 2.5275862068965516e-05,
|
| 1985 |
+
"loss": 0.1007,
|
| 1986 |
+
"step": 7675
|
| 1987 |
+
},
|
| 1988 |
+
{
|
| 1989 |
+
"epoch": 2.85,
|
| 1990 |
+
"learning_rate": 2.5189655172413794e-05,
|
| 1991 |
+
"loss": 0.1029,
|
| 1992 |
+
"step": 7700
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 2.86,
|
| 1996 |
+
"learning_rate": 2.510344827586207e-05,
|
| 1997 |
+
"loss": 0.0989,
|
| 1998 |
+
"step": 7725
|
| 1999 |
+
},
|
| 2000 |
+
{
|
| 2001 |
+
"epoch": 2.87,
|
| 2002 |
+
"learning_rate": 2.5017241379310346e-05,
|
| 2003 |
+
"loss": 0.1042,
|
| 2004 |
+
"step": 7750
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 2.88,
|
| 2008 |
+
"learning_rate": 2.493103448275862e-05,
|
| 2009 |
+
"loss": 0.1018,
|
| 2010 |
+
"step": 7775
|
| 2011 |
+
},
|
| 2012 |
+
{
|
| 2013 |
+
"epoch": 2.89,
|
| 2014 |
+
"learning_rate": 2.48448275862069e-05,
|
| 2015 |
+
"loss": 0.1017,
|
| 2016 |
+
"step": 7800
|
| 2017 |
+
},
|
| 2018 |
+
{
|
| 2019 |
+
"epoch": 2.9,
|
| 2020 |
+
"learning_rate": 2.4758620689655173e-05,
|
| 2021 |
+
"loss": 0.1028,
|
| 2022 |
+
"step": 7825
|
| 2023 |
+
},
|
| 2024 |
+
{
|
| 2025 |
+
"epoch": 2.91,
|
| 2026 |
+
"learning_rate": 2.467241379310345e-05,
|
| 2027 |
+
"loss": 0.0994,
|
| 2028 |
+
"step": 7850
|
| 2029 |
+
},
|
| 2030 |
+
{
|
| 2031 |
+
"epoch": 2.91,
|
| 2032 |
+
"learning_rate": 2.4586206896551725e-05,
|
| 2033 |
+
"loss": 0.1049,
|
| 2034 |
+
"step": 7875
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 2.92,
|
| 2038 |
+
"learning_rate": 2.45e-05,
|
| 2039 |
+
"loss": 0.1005,
|
| 2040 |
+
"step": 7900
|
| 2041 |
+
},
|
| 2042 |
+
{
|
| 2043 |
+
"epoch": 2.93,
|
| 2044 |
+
"learning_rate": 2.4413793103448277e-05,
|
| 2045 |
+
"loss": 0.1027,
|
| 2046 |
+
"step": 7925
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 2.94,
|
| 2050 |
+
"learning_rate": 2.432758620689655e-05,
|
| 2051 |
+
"loss": 0.1096,
|
| 2052 |
+
"step": 7950
|
| 2053 |
+
},
|
| 2054 |
+
{
|
| 2055 |
+
"epoch": 2.95,
|
| 2056 |
+
"learning_rate": 2.424137931034483e-05,
|
| 2057 |
+
"loss": 0.0999,
|
| 2058 |
+
"step": 7975
|
| 2059 |
+
},
|
| 2060 |
+
{
|
| 2061 |
+
"epoch": 2.96,
|
| 2062 |
+
"learning_rate": 2.4155172413793103e-05,
|
| 2063 |
+
"loss": 0.1021,
|
| 2064 |
+
"step": 8000
|
| 2065 |
+
},
|
| 2066 |
+
{
|
| 2067 |
+
"epoch": 2.96,
|
| 2068 |
+
"eval_cer": 11.506672732787383,
|
| 2069 |
+
"eval_loss": 0.2621135413646698,
|
| 2070 |
+
"eval_runtime": 358.5537,
|
| 2071 |
+
"eval_samples_per_second": 7.198,
|
| 2072 |
+
"eval_steps_per_second": 0.901,
|
| 2073 |
+
"step": 8000
|
| 2074 |
+
},
|
| 2075 |
+
{
|
| 2076 |
+
"epoch": 2.97,
|
| 2077 |
+
"learning_rate": 2.4068965517241378e-05,
|
| 2078 |
+
"loss": 0.0984,
|
| 2079 |
+
"step": 8025
|
| 2080 |
+
},
|
| 2081 |
+
{
|
| 2082 |
+
"epoch": 2.98,
|
| 2083 |
+
"learning_rate": 2.398275862068966e-05,
|
| 2084 |
+
"loss": 0.0976,
|
| 2085 |
+
"step": 8050
|
| 2086 |
+
},
|
| 2087 |
+
{
|
| 2088 |
+
"epoch": 2.99,
|
| 2089 |
+
"learning_rate": 2.3896551724137933e-05,
|
| 2090 |
+
"loss": 0.1034,
|
| 2091 |
+
"step": 8075
|
| 2092 |
+
},
|
| 2093 |
+
{
|
| 2094 |
+
"epoch": 3.0,
|
| 2095 |
+
"learning_rate": 2.3810344827586208e-05,
|
| 2096 |
+
"loss": 0.1011,
|
| 2097 |
+
"step": 8100
|
| 2098 |
+
},
|
| 2099 |
+
{
|
| 2100 |
+
"epoch": 3.01,
|
| 2101 |
+
"learning_rate": 2.3724137931034485e-05,
|
| 2102 |
+
"loss": 0.0651,
|
| 2103 |
+
"step": 8125
|
| 2104 |
+
},
|
| 2105 |
+
{
|
| 2106 |
+
"epoch": 3.02,
|
| 2107 |
+
"learning_rate": 2.363793103448276e-05,
|
| 2108 |
+
"loss": 0.0473,
|
| 2109 |
+
"step": 8150
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 3.03,
|
| 2113 |
+
"learning_rate": 2.3551724137931037e-05,
|
| 2114 |
+
"loss": 0.0448,
|
| 2115 |
+
"step": 8175
|
| 2116 |
+
},
|
| 2117 |
+
{
|
| 2118 |
+
"epoch": 3.03,
|
| 2119 |
+
"learning_rate": 2.3465517241379312e-05,
|
| 2120 |
+
"loss": 0.0473,
|
| 2121 |
+
"step": 8200
|
| 2122 |
+
},
|
| 2123 |
+
{
|
| 2124 |
+
"epoch": 3.04,
|
| 2125 |
+
"learning_rate": 2.3379310344827586e-05,
|
| 2126 |
+
"loss": 0.0457,
|
| 2127 |
+
"step": 8225
|
| 2128 |
+
},
|
| 2129 |
+
{
|
| 2130 |
+
"epoch": 3.05,
|
| 2131 |
+
"learning_rate": 2.3293103448275864e-05,
|
| 2132 |
+
"loss": 0.0451,
|
| 2133 |
+
"step": 8250
|
| 2134 |
+
},
|
| 2135 |
+
{
|
| 2136 |
+
"epoch": 3.06,
|
| 2137 |
+
"learning_rate": 2.3206896551724138e-05,
|
| 2138 |
+
"loss": 0.0453,
|
| 2139 |
+
"step": 8275
|
| 2140 |
+
},
|
| 2141 |
+
{
|
| 2142 |
+
"epoch": 3.07,
|
| 2143 |
+
"learning_rate": 2.3120689655172416e-05,
|
| 2144 |
+
"loss": 0.0502,
|
| 2145 |
+
"step": 8300
|
| 2146 |
+
},
|
| 2147 |
+
{
|
| 2148 |
+
"epoch": 3.08,
|
| 2149 |
+
"learning_rate": 2.303448275862069e-05,
|
| 2150 |
+
"loss": 0.0471,
|
| 2151 |
+
"step": 8325
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 3.09,
|
| 2155 |
+
"learning_rate": 2.2948275862068965e-05,
|
| 2156 |
+
"loss": 0.0476,
|
| 2157 |
+
"step": 8350
|
| 2158 |
+
},
|
| 2159 |
+
{
|
| 2160 |
+
"epoch": 3.1,
|
| 2161 |
+
"learning_rate": 2.2862068965517242e-05,
|
| 2162 |
+
"loss": 0.0473,
|
| 2163 |
+
"step": 8375
|
| 2164 |
+
},
|
| 2165 |
+
{
|
| 2166 |
+
"epoch": 3.11,
|
| 2167 |
+
"learning_rate": 2.2775862068965517e-05,
|
| 2168 |
+
"loss": 0.0476,
|
| 2169 |
+
"step": 8400
|
| 2170 |
+
},
|
| 2171 |
+
{
|
| 2172 |
+
"epoch": 3.12,
|
| 2173 |
+
"learning_rate": 2.2689655172413794e-05,
|
| 2174 |
+
"loss": 0.0466,
|
| 2175 |
+
"step": 8425
|
| 2176 |
+
},
|
| 2177 |
+
{
|
| 2178 |
+
"epoch": 3.13,
|
| 2179 |
+
"learning_rate": 2.2603448275862072e-05,
|
| 2180 |
+
"loss": 0.0478,
|
| 2181 |
+
"step": 8450
|
| 2182 |
+
},
|
| 2183 |
+
{
|
| 2184 |
+
"epoch": 3.14,
|
| 2185 |
+
"learning_rate": 2.2517241379310347e-05,
|
| 2186 |
+
"loss": 0.0464,
|
| 2187 |
+
"step": 8475
|
| 2188 |
+
},
|
| 2189 |
+
{
|
| 2190 |
+
"epoch": 3.15,
|
| 2191 |
+
"learning_rate": 2.2431034482758624e-05,
|
| 2192 |
+
"loss": 0.0468,
|
| 2193 |
+
"step": 8500
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 3.15,
|
| 2197 |
+
"eval_cer": 11.332271762208068,
|
| 2198 |
+
"eval_loss": 0.2711898386478424,
|
| 2199 |
+
"eval_runtime": 359.6452,
|
| 2200 |
+
"eval_samples_per_second": 7.177,
|
| 2201 |
+
"eval_steps_per_second": 0.898,
|
| 2202 |
+
"step": 8500
|
| 2203 |
+
},
|
| 2204 |
+
{
|
| 2205 |
+
"epoch": 3.16,
|
| 2206 |
+
"learning_rate": 2.23448275862069e-05,
|
| 2207 |
+
"loss": 0.0468,
|
| 2208 |
+
"step": 8525
|
| 2209 |
+
},
|
| 2210 |
+
{
|
| 2211 |
+
"epoch": 3.16,
|
| 2212 |
+
"learning_rate": 2.2258620689655173e-05,
|
| 2213 |
+
"loss": 0.0526,
|
| 2214 |
+
"step": 8550
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 3.17,
|
| 2218 |
+
"learning_rate": 2.217241379310345e-05,
|
| 2219 |
+
"loss": 0.0465,
|
| 2220 |
+
"step": 8575
|
| 2221 |
+
},
|
| 2222 |
+
{
|
| 2223 |
+
"epoch": 3.18,
|
| 2224 |
+
"learning_rate": 2.2086206896551725e-05,
|
| 2225 |
+
"loss": 0.0499,
|
| 2226 |
+
"step": 8600
|
| 2227 |
+
},
|
| 2228 |
+
{
|
| 2229 |
+
"epoch": 3.19,
|
| 2230 |
+
"learning_rate": 2.2000000000000003e-05,
|
| 2231 |
+
"loss": 0.0466,
|
| 2232 |
+
"step": 8625
|
| 2233 |
+
},
|
| 2234 |
+
{
|
| 2235 |
+
"epoch": 3.2,
|
| 2236 |
+
"learning_rate": 2.1913793103448277e-05,
|
| 2237 |
+
"loss": 0.0451,
|
| 2238 |
+
"step": 8650
|
| 2239 |
+
},
|
| 2240 |
+
{
|
| 2241 |
+
"epoch": 3.21,
|
| 2242 |
+
"learning_rate": 2.182758620689655e-05,
|
| 2243 |
+
"loss": 0.047,
|
| 2244 |
+
"step": 8675
|
| 2245 |
+
},
|
| 2246 |
+
{
|
| 2247 |
+
"epoch": 3.22,
|
| 2248 |
+
"learning_rate": 2.174137931034483e-05,
|
| 2249 |
+
"loss": 0.0482,
|
| 2250 |
+
"step": 8700
|
| 2251 |
+
},
|
| 2252 |
+
{
|
| 2253 |
+
"epoch": 3.23,
|
| 2254 |
+
"learning_rate": 2.1655172413793104e-05,
|
| 2255 |
+
"loss": 0.0513,
|
| 2256 |
+
"step": 8725
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 3.24,
|
| 2260 |
+
"learning_rate": 2.1568965517241378e-05,
|
| 2261 |
+
"loss": 0.0483,
|
| 2262 |
+
"step": 8750
|
| 2263 |
+
},
|
| 2264 |
+
{
|
| 2265 |
+
"epoch": 3.25,
|
| 2266 |
+
"learning_rate": 2.1482758620689656e-05,
|
| 2267 |
+
"loss": 0.0491,
|
| 2268 |
+
"step": 8775
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 3.26,
|
| 2272 |
+
"learning_rate": 2.139655172413793e-05,
|
| 2273 |
+
"loss": 0.0483,
|
| 2274 |
+
"step": 8800
|
| 2275 |
+
},
|
| 2276 |
+
{
|
| 2277 |
+
"epoch": 3.27,
|
| 2278 |
+
"learning_rate": 2.1310344827586208e-05,
|
| 2279 |
+
"loss": 0.0504,
|
| 2280 |
+
"step": 8825
|
| 2281 |
+
},
|
| 2282 |
+
{
|
| 2283 |
+
"epoch": 3.28,
|
| 2284 |
+
"learning_rate": 2.1224137931034486e-05,
|
| 2285 |
+
"loss": 0.0514,
|
| 2286 |
+
"step": 8850
|
| 2287 |
+
},
|
| 2288 |
+
{
|
| 2289 |
+
"epoch": 3.28,
|
| 2290 |
+
"learning_rate": 2.113793103448276e-05,
|
| 2291 |
+
"loss": 0.0484,
|
| 2292 |
+
"step": 8875
|
| 2293 |
+
},
|
| 2294 |
+
{
|
| 2295 |
+
"epoch": 3.29,
|
| 2296 |
+
"learning_rate": 2.1051724137931038e-05,
|
| 2297 |
+
"loss": 0.0474,
|
| 2298 |
+
"step": 8900
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 3.3,
|
| 2302 |
+
"learning_rate": 2.0965517241379312e-05,
|
| 2303 |
+
"loss": 0.0462,
|
| 2304 |
+
"step": 8925
|
| 2305 |
+
},
|
| 2306 |
+
{
|
| 2307 |
+
"epoch": 3.31,
|
| 2308 |
+
"learning_rate": 2.0879310344827586e-05,
|
| 2309 |
+
"loss": 0.0501,
|
| 2310 |
+
"step": 8950
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 3.32,
|
| 2314 |
+
"learning_rate": 2.0793103448275864e-05,
|
| 2315 |
+
"loss": 0.0487,
|
| 2316 |
+
"step": 8975
|
| 2317 |
+
},
|
| 2318 |
+
{
|
| 2319 |
+
"epoch": 3.33,
|
| 2320 |
+
"learning_rate": 2.070689655172414e-05,
|
| 2321 |
+
"loss": 0.0499,
|
| 2322 |
+
"step": 9000
|
| 2323 |
+
},
|
| 2324 |
+
{
|
| 2325 |
+
"epoch": 3.33,
|
| 2326 |
+
"eval_cer": 11.555959963603275,
|
| 2327 |
+
"eval_loss": 0.28167569637298584,
|
| 2328 |
+
"eval_runtime": 359.0627,
|
| 2329 |
+
"eval_samples_per_second": 7.188,
|
| 2330 |
+
"eval_steps_per_second": 0.9,
|
| 2331 |
+
"step": 9000
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 3.34,
|
| 2335 |
+
"learning_rate": 2.0620689655172416e-05,
|
| 2336 |
+
"loss": 0.0496,
|
| 2337 |
+
"step": 9025
|
| 2338 |
+
},
|
| 2339 |
+
{
|
| 2340 |
+
"epoch": 3.35,
|
| 2341 |
+
"learning_rate": 2.053448275862069e-05,
|
| 2342 |
+
"loss": 0.0471,
|
| 2343 |
+
"step": 9050
|
| 2344 |
+
},
|
| 2345 |
+
{
|
| 2346 |
+
"epoch": 3.36,
|
| 2347 |
+
"learning_rate": 2.0448275862068965e-05,
|
| 2348 |
+
"loss": 0.0469,
|
| 2349 |
+
"step": 9075
|
| 2350 |
+
},
|
| 2351 |
+
{
|
| 2352 |
+
"epoch": 3.37,
|
| 2353 |
+
"learning_rate": 2.0362068965517243e-05,
|
| 2354 |
+
"loss": 0.0511,
|
| 2355 |
+
"step": 9100
|
| 2356 |
+
},
|
| 2357 |
+
{
|
| 2358 |
+
"epoch": 3.38,
|
| 2359 |
+
"learning_rate": 2.0275862068965517e-05,
|
| 2360 |
+
"loss": 0.0476,
|
| 2361 |
+
"step": 9125
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 3.39,
|
| 2365 |
+
"learning_rate": 2.0189655172413795e-05,
|
| 2366 |
+
"loss": 0.0483,
|
| 2367 |
+
"step": 9150
|
| 2368 |
+
},
|
| 2369 |
+
{
|
| 2370 |
+
"epoch": 3.4,
|
| 2371 |
+
"learning_rate": 2.010344827586207e-05,
|
| 2372 |
+
"loss": 0.0484,
|
| 2373 |
+
"step": 9175
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 3.4,
|
| 2377 |
+
"learning_rate": 2.0017241379310343e-05,
|
| 2378 |
+
"loss": 0.0524,
|
| 2379 |
+
"step": 9200
|
| 2380 |
+
},
|
| 2381 |
+
{
|
| 2382 |
+
"epoch": 3.41,
|
| 2383 |
+
"learning_rate": 1.993103448275862e-05,
|
| 2384 |
+
"loss": 0.0469,
|
| 2385 |
+
"step": 9225
|
| 2386 |
+
},
|
| 2387 |
+
{
|
| 2388 |
+
"epoch": 3.42,
|
| 2389 |
+
"learning_rate": 1.98448275862069e-05,
|
| 2390 |
+
"loss": 0.0454,
|
| 2391 |
+
"step": 9250
|
| 2392 |
+
},
|
| 2393 |
+
{
|
| 2394 |
+
"epoch": 3.43,
|
| 2395 |
+
"learning_rate": 1.9758620689655173e-05,
|
| 2396 |
+
"loss": 0.0489,
|
| 2397 |
+
"step": 9275
|
| 2398 |
+
},
|
| 2399 |
+
{
|
| 2400 |
+
"epoch": 3.44,
|
| 2401 |
+
"learning_rate": 1.967241379310345e-05,
|
| 2402 |
+
"loss": 0.0482,
|
| 2403 |
+
"step": 9300
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 3.45,
|
| 2407 |
+
"learning_rate": 1.9586206896551725e-05,
|
| 2408 |
+
"loss": 0.0487,
|
| 2409 |
+
"step": 9325
|
| 2410 |
+
},
|
| 2411 |
+
{
|
| 2412 |
+
"epoch": 3.46,
|
| 2413 |
+
"learning_rate": 1.9500000000000003e-05,
|
| 2414 |
+
"loss": 0.048,
|
| 2415 |
+
"step": 9350
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 3.47,
|
| 2419 |
+
"learning_rate": 1.9413793103448277e-05,
|
| 2420 |
+
"loss": 0.0477,
|
| 2421 |
+
"step": 9375
|
| 2422 |
+
},
|
| 2423 |
+
{
|
| 2424 |
+
"epoch": 3.48,
|
| 2425 |
+
"learning_rate": 1.9327586206896552e-05,
|
| 2426 |
+
"loss": 0.0511,
|
| 2427 |
+
"step": 9400
|
| 2428 |
+
},
|
| 2429 |
+
{
|
| 2430 |
+
"epoch": 3.49,
|
| 2431 |
+
"learning_rate": 1.924137931034483e-05,
|
| 2432 |
+
"loss": 0.0468,
|
| 2433 |
+
"step": 9425
|
| 2434 |
+
},
|
| 2435 |
+
{
|
| 2436 |
+
"epoch": 3.5,
|
| 2437 |
+
"learning_rate": 1.9155172413793104e-05,
|
| 2438 |
+
"loss": 0.0502,
|
| 2439 |
+
"step": 9450
|
| 2440 |
+
},
|
| 2441 |
+
{
|
| 2442 |
+
"epoch": 3.51,
|
| 2443 |
+
"learning_rate": 1.906896551724138e-05,
|
| 2444 |
+
"loss": 0.0476,
|
| 2445 |
+
"step": 9475
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 3.52,
|
| 2449 |
+
"learning_rate": 1.8982758620689656e-05,
|
| 2450 |
+
"loss": 0.0489,
|
| 2451 |
+
"step": 9500
|
| 2452 |
+
},
|
| 2453 |
+
{
|
| 2454 |
+
"epoch": 3.52,
|
| 2455 |
+
"eval_cer": 11.548377312708523,
|
| 2456 |
+
"eval_loss": 0.28109079599380493,
|
| 2457 |
+
"eval_runtime": 359.545,
|
| 2458 |
+
"eval_samples_per_second": 7.179,
|
| 2459 |
+
"eval_steps_per_second": 0.898,
|
| 2460 |
+
"step": 9500
|
| 2461 |
+
},
|
| 2462 |
+
{
|
| 2463 |
+
"epoch": 3.53,
|
| 2464 |
+
"learning_rate": 1.889655172413793e-05,
|
| 2465 |
+
"loss": 0.0463,
|
| 2466 |
+
"step": 9525
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 3.53,
|
| 2470 |
+
"learning_rate": 1.8810344827586208e-05,
|
| 2471 |
+
"loss": 0.0457,
|
| 2472 |
+
"step": 9550
|
| 2473 |
+
},
|
| 2474 |
+
{
|
| 2475 |
+
"epoch": 3.54,
|
| 2476 |
+
"learning_rate": 1.8724137931034482e-05,
|
| 2477 |
+
"loss": 0.0463,
|
| 2478 |
+
"step": 9575
|
| 2479 |
+
},
|
| 2480 |
+
{
|
| 2481 |
+
"epoch": 3.55,
|
| 2482 |
+
"learning_rate": 1.863793103448276e-05,
|
| 2483 |
+
"loss": 0.0526,
|
| 2484 |
+
"step": 9600
|
| 2485 |
+
},
|
| 2486 |
+
{
|
| 2487 |
+
"epoch": 3.56,
|
| 2488 |
+
"learning_rate": 1.8551724137931034e-05,
|
| 2489 |
+
"loss": 0.0499,
|
| 2490 |
+
"step": 9625
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 3.57,
|
| 2494 |
+
"learning_rate": 1.8465517241379312e-05,
|
| 2495 |
+
"loss": 0.0488,
|
| 2496 |
+
"step": 9650
|
| 2497 |
+
},
|
| 2498 |
+
{
|
| 2499 |
+
"epoch": 3.58,
|
| 2500 |
+
"learning_rate": 1.837931034482759e-05,
|
| 2501 |
+
"loss": 0.0491,
|
| 2502 |
+
"step": 9675
|
| 2503 |
+
},
|
| 2504 |
+
{
|
| 2505 |
+
"epoch": 3.59,
|
| 2506 |
+
"learning_rate": 1.8293103448275864e-05,
|
| 2507 |
+
"loss": 0.046,
|
| 2508 |
+
"step": 9700
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 3.6,
|
| 2512 |
+
"learning_rate": 1.820689655172414e-05,
|
| 2513 |
+
"loss": 0.0475,
|
| 2514 |
+
"step": 9725
|
| 2515 |
+
},
|
| 2516 |
+
{
|
| 2517 |
+
"epoch": 3.61,
|
| 2518 |
+
"learning_rate": 1.8120689655172416e-05,
|
| 2519 |
+
"loss": 0.0512,
|
| 2520 |
+
"step": 9750
|
| 2521 |
+
},
|
| 2522 |
+
{
|
| 2523 |
+
"epoch": 3.62,
|
| 2524 |
+
"learning_rate": 1.803448275862069e-05,
|
| 2525 |
+
"loss": 0.0478,
|
| 2526 |
+
"step": 9775
|
| 2527 |
+
},
|
| 2528 |
+
{
|
| 2529 |
+
"epoch": 3.63,
|
| 2530 |
+
"learning_rate": 1.7948275862068965e-05,
|
| 2531 |
+
"loss": 0.0602,
|
| 2532 |
+
"step": 9800
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 3.64,
|
| 2536 |
+
"learning_rate": 1.7862068965517243e-05,
|
| 2537 |
+
"loss": 0.0482,
|
| 2538 |
+
"step": 9825
|
| 2539 |
+
},
|
| 2540 |
+
{
|
| 2541 |
+
"epoch": 3.65,
|
| 2542 |
+
"learning_rate": 1.7775862068965517e-05,
|
| 2543 |
+
"loss": 0.0463,
|
| 2544 |
+
"step": 9850
|
| 2545 |
+
},
|
| 2546 |
+
{
|
| 2547 |
+
"epoch": 3.65,
|
| 2548 |
+
"learning_rate": 1.7689655172413795e-05,
|
| 2549 |
+
"loss": 0.0513,
|
| 2550 |
+
"step": 9875
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 3.66,
|
| 2554 |
+
"learning_rate": 1.760344827586207e-05,
|
| 2555 |
+
"loss": 0.0495,
|
| 2556 |
+
"step": 9900
|
| 2557 |
+
},
|
| 2558 |
+
{
|
| 2559 |
+
"epoch": 3.67,
|
| 2560 |
+
"learning_rate": 1.7517241379310344e-05,
|
| 2561 |
+
"loss": 0.0547,
|
| 2562 |
+
"step": 9925
|
| 2563 |
+
},
|
| 2564 |
+
{
|
| 2565 |
+
"epoch": 3.68,
|
| 2566 |
+
"learning_rate": 1.743103448275862e-05,
|
| 2567 |
+
"loss": 0.0485,
|
| 2568 |
+
"step": 9950
|
| 2569 |
+
},
|
| 2570 |
+
{
|
| 2571 |
+
"epoch": 3.69,
|
| 2572 |
+
"learning_rate": 1.7344827586206896e-05,
|
| 2573 |
+
"loss": 0.0498,
|
| 2574 |
+
"step": 9975
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 3.7,
|
| 2578 |
+
"learning_rate": 1.7258620689655173e-05,
|
| 2579 |
+
"loss": 0.0479,
|
| 2580 |
+
"step": 10000
|
| 2581 |
+
},
|
| 2582 |
+
{
|
| 2583 |
+
"epoch": 3.7,
|
| 2584 |
+
"eval_cer": 10.695329087048833,
|
| 2585 |
+
"eval_loss": 0.27692216634750366,
|
| 2586 |
+
"eval_runtime": 359.4994,
|
| 2587 |
+
"eval_samples_per_second": 7.179,
|
| 2588 |
+
"eval_steps_per_second": 0.898,
|
| 2589 |
+
"step": 10000
|
| 2590 |
+
},
|
| 2591 |
+
{
|
| 2592 |
+
"epoch": 3.71,
|
| 2593 |
+
"learning_rate": 1.7172413793103448e-05,
|
| 2594 |
+
"loss": 0.0456,
|
| 2595 |
+
"step": 10025
|
| 2596 |
+
},
|
| 2597 |
+
{
|
| 2598 |
+
"epoch": 3.72,
|
| 2599 |
+
"learning_rate": 1.7086206896551722e-05,
|
| 2600 |
+
"loss": 0.0512,
|
| 2601 |
+
"step": 10050
|
| 2602 |
+
},
|
| 2603 |
+
{
|
| 2604 |
+
"epoch": 3.73,
|
| 2605 |
+
"learning_rate": 1.7000000000000003e-05,
|
| 2606 |
+
"loss": 0.0487,
|
| 2607 |
+
"step": 10075
|
| 2608 |
+
},
|
| 2609 |
+
{
|
| 2610 |
+
"epoch": 3.74,
|
| 2611 |
+
"learning_rate": 1.6913793103448278e-05,
|
| 2612 |
+
"loss": 0.0463,
|
| 2613 |
+
"step": 10100
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 3.75,
|
| 2617 |
+
"learning_rate": 1.6827586206896552e-05,
|
| 2618 |
+
"loss": 0.0625,
|
| 2619 |
+
"step": 10125
|
| 2620 |
+
},
|
| 2621 |
+
{
|
| 2622 |
+
"epoch": 3.76,
|
| 2623 |
+
"learning_rate": 1.674137931034483e-05,
|
| 2624 |
+
"loss": 0.0474,
|
| 2625 |
+
"step": 10150
|
| 2626 |
+
},
|
| 2627 |
+
{
|
| 2628 |
+
"epoch": 3.77,
|
| 2629 |
+
"learning_rate": 1.6655172413793104e-05,
|
| 2630 |
+
"loss": 0.0476,
|
| 2631 |
+
"step": 10175
|
| 2632 |
+
},
|
| 2633 |
+
{
|
| 2634 |
+
"epoch": 3.77,
|
| 2635 |
+
"learning_rate": 1.6568965517241382e-05,
|
| 2636 |
+
"loss": 0.0502,
|
| 2637 |
+
"step": 10200
|
| 2638 |
+
},
|
| 2639 |
+
{
|
| 2640 |
+
"epoch": 3.78,
|
| 2641 |
+
"learning_rate": 1.6482758620689656e-05,
|
| 2642 |
+
"loss": 0.0522,
|
| 2643 |
+
"step": 10225
|
| 2644 |
+
},
|
| 2645 |
+
{
|
| 2646 |
+
"epoch": 3.79,
|
| 2647 |
+
"learning_rate": 1.639655172413793e-05,
|
| 2648 |
+
"loss": 0.0458,
|
| 2649 |
+
"step": 10250
|
| 2650 |
+
},
|
| 2651 |
+
{
|
| 2652 |
+
"epoch": 3.8,
|
| 2653 |
+
"learning_rate": 1.6310344827586208e-05,
|
| 2654 |
+
"loss": 0.0482,
|
| 2655 |
+
"step": 10275
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 3.81,
|
| 2659 |
+
"learning_rate": 1.6224137931034483e-05,
|
| 2660 |
+
"loss": 0.0466,
|
| 2661 |
+
"step": 10300
|
| 2662 |
+
},
|
| 2663 |
+
{
|
| 2664 |
+
"epoch": 3.82,
|
| 2665 |
+
"learning_rate": 1.613793103448276e-05,
|
| 2666 |
+
"loss": 0.0466,
|
| 2667 |
+
"step": 10325
|
| 2668 |
+
},
|
| 2669 |
+
{
|
| 2670 |
+
"epoch": 3.83,
|
| 2671 |
+
"learning_rate": 1.6051724137931035e-05,
|
| 2672 |
+
"loss": 0.0451,
|
| 2673 |
+
"step": 10350
|
| 2674 |
+
},
|
| 2675 |
+
{
|
| 2676 |
+
"epoch": 3.84,
|
| 2677 |
+
"learning_rate": 1.596551724137931e-05,
|
| 2678 |
+
"loss": 0.0492,
|
| 2679 |
+
"step": 10375
|
| 2680 |
+
},
|
| 2681 |
+
{
|
| 2682 |
+
"epoch": 3.85,
|
| 2683 |
+
"learning_rate": 1.5879310344827587e-05,
|
| 2684 |
+
"loss": 0.045,
|
| 2685 |
+
"step": 10400
|
| 2686 |
+
},
|
| 2687 |
+
{
|
| 2688 |
+
"epoch": 3.86,
|
| 2689 |
+
"learning_rate": 1.579310344827586e-05,
|
| 2690 |
+
"loss": 0.0457,
|
| 2691 |
+
"step": 10425
|
| 2692 |
+
},
|
| 2693 |
+
{
|
| 2694 |
+
"epoch": 3.87,
|
| 2695 |
+
"learning_rate": 1.570689655172414e-05,
|
| 2696 |
+
"loss": 0.0471,
|
| 2697 |
+
"step": 10450
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 3.88,
|
| 2701 |
+
"learning_rate": 1.5620689655172417e-05,
|
| 2702 |
+
"loss": 0.0473,
|
| 2703 |
+
"step": 10475
|
| 2704 |
+
},
|
| 2705 |
+
{
|
| 2706 |
+
"epoch": 3.89,
|
| 2707 |
+
"learning_rate": 1.553448275862069e-05,
|
| 2708 |
+
"loss": 0.049,
|
| 2709 |
+
"step": 10500
|
| 2710 |
+
},
|
| 2711 |
+
{
|
| 2712 |
+
"epoch": 3.89,
|
| 2713 |
+
"eval_cer": 11.51046405823476,
|
| 2714 |
+
"eval_loss": 0.28438711166381836,
|
| 2715 |
+
"eval_runtime": 359.2212,
|
| 2716 |
+
"eval_samples_per_second": 7.185,
|
| 2717 |
+
"eval_steps_per_second": 0.899,
|
| 2718 |
+
"step": 10500
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 3.9,
|
| 2722 |
+
"learning_rate": 1.544827586206897e-05,
|
| 2723 |
+
"loss": 0.0458,
|
| 2724 |
+
"step": 10525
|
| 2725 |
+
},
|
| 2726 |
+
{
|
| 2727 |
+
"epoch": 3.9,
|
| 2728 |
+
"learning_rate": 1.5362068965517243e-05,
|
| 2729 |
+
"loss": 0.0456,
|
| 2730 |
+
"step": 10550
|
| 2731 |
+
},
|
| 2732 |
+
{
|
| 2733 |
+
"epoch": 3.91,
|
| 2734 |
+
"learning_rate": 1.5275862068965517e-05,
|
| 2735 |
+
"loss": 0.0474,
|
| 2736 |
+
"step": 10575
|
| 2737 |
+
},
|
| 2738 |
+
{
|
| 2739 |
+
"epoch": 3.92,
|
| 2740 |
+
"learning_rate": 1.5189655172413793e-05,
|
| 2741 |
+
"loss": 0.0447,
|
| 2742 |
+
"step": 10600
|
| 2743 |
+
},
|
| 2744 |
+
{
|
| 2745 |
+
"epoch": 3.93,
|
| 2746 |
+
"learning_rate": 1.510344827586207e-05,
|
| 2747 |
+
"loss": 0.046,
|
| 2748 |
+
"step": 10625
|
| 2749 |
+
},
|
| 2750 |
+
{
|
| 2751 |
+
"epoch": 3.94,
|
| 2752 |
+
"learning_rate": 1.5017241379310346e-05,
|
| 2753 |
+
"loss": 0.0442,
|
| 2754 |
+
"step": 10650
|
| 2755 |
+
},
|
| 2756 |
+
{
|
| 2757 |
+
"epoch": 3.95,
|
| 2758 |
+
"learning_rate": 1.4931034482758622e-05,
|
| 2759 |
+
"loss": 0.0457,
|
| 2760 |
+
"step": 10675
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 3.96,
|
| 2764 |
+
"learning_rate": 1.4844827586206898e-05,
|
| 2765 |
+
"loss": 0.0482,
|
| 2766 |
+
"step": 10700
|
| 2767 |
+
},
|
| 2768 |
+
{
|
| 2769 |
+
"epoch": 3.97,
|
| 2770 |
+
"learning_rate": 1.4758620689655172e-05,
|
| 2771 |
+
"loss": 0.0476,
|
| 2772 |
+
"step": 10725
|
| 2773 |
+
},
|
| 2774 |
+
{
|
| 2775 |
+
"epoch": 3.98,
|
| 2776 |
+
"learning_rate": 1.4672413793103448e-05,
|
| 2777 |
+
"loss": 0.0449,
|
| 2778 |
+
"step": 10750
|
| 2779 |
+
},
|
| 2780 |
+
{
|
| 2781 |
+
"epoch": 3.99,
|
| 2782 |
+
"learning_rate": 1.4586206896551724e-05,
|
| 2783 |
+
"loss": 0.0478,
|
| 2784 |
+
"step": 10775
|
| 2785 |
+
},
|
| 2786 |
+
{
|
| 2787 |
+
"epoch": 4.0,
|
| 2788 |
+
"learning_rate": 1.45e-05,
|
| 2789 |
+
"loss": 0.0517,
|
| 2790 |
+
"step": 10800
|
| 2791 |
+
},
|
| 2792 |
+
{
|
| 2793 |
+
"epoch": 4.01,
|
| 2794 |
+
"learning_rate": 1.4413793103448276e-05,
|
| 2795 |
+
"loss": 0.0285,
|
| 2796 |
+
"step": 10825
|
| 2797 |
+
},
|
| 2798 |
+
{
|
| 2799 |
+
"epoch": 4.02,
|
| 2800 |
+
"learning_rate": 1.432758620689655e-05,
|
| 2801 |
+
"loss": 0.0197,
|
| 2802 |
+
"step": 10850
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 4.02,
|
| 2806 |
+
"learning_rate": 1.424137931034483e-05,
|
| 2807 |
+
"loss": 0.0183,
|
| 2808 |
+
"step": 10875
|
| 2809 |
+
},
|
| 2810 |
+
{
|
| 2811 |
+
"epoch": 4.03,
|
| 2812 |
+
"learning_rate": 1.4155172413793106e-05,
|
| 2813 |
+
"loss": 0.0186,
|
| 2814 |
+
"step": 10900
|
| 2815 |
+
},
|
| 2816 |
+
{
|
| 2817 |
+
"epoch": 4.04,
|
| 2818 |
+
"learning_rate": 1.406896551724138e-05,
|
| 2819 |
+
"loss": 0.0203,
|
| 2820 |
+
"step": 10925
|
| 2821 |
+
},
|
| 2822 |
+
{
|
| 2823 |
+
"epoch": 4.05,
|
| 2824 |
+
"learning_rate": 1.3982758620689656e-05,
|
| 2825 |
+
"loss": 0.0185,
|
| 2826 |
+
"step": 10950
|
| 2827 |
+
},
|
| 2828 |
+
{
|
| 2829 |
+
"epoch": 4.06,
|
| 2830 |
+
"learning_rate": 1.3896551724137932e-05,
|
| 2831 |
+
"loss": 0.0173,
|
| 2832 |
+
"step": 10975
|
| 2833 |
+
},
|
| 2834 |
+
{
|
| 2835 |
+
"epoch": 4.07,
|
| 2836 |
+
"learning_rate": 1.3810344827586208e-05,
|
| 2837 |
+
"loss": 0.0193,
|
| 2838 |
+
"step": 11000
|
| 2839 |
+
},
|
| 2840 |
+
{
|
| 2841 |
+
"epoch": 4.07,
|
| 2842 |
+
"eval_cer": 11.116166211707613,
|
| 2843 |
+
"eval_loss": 0.29933854937553406,
|
| 2844 |
+
"eval_runtime": 359.5628,
|
| 2845 |
+
"eval_samples_per_second": 7.178,
|
| 2846 |
+
"eval_steps_per_second": 0.898,
|
| 2847 |
+
"step": 11000
|
| 2848 |
+
},
|
| 2849 |
+
{
|
| 2850 |
+
"epoch": 4.08,
|
| 2851 |
+
"learning_rate": 1.3724137931034484e-05,
|
| 2852 |
+
"loss": 0.0176,
|
| 2853 |
+
"step": 11025
|
| 2854 |
+
},
|
| 2855 |
+
{
|
| 2856 |
+
"epoch": 4.09,
|
| 2857 |
+
"learning_rate": 1.3637931034482759e-05,
|
| 2858 |
+
"loss": 0.0171,
|
| 2859 |
+
"step": 11050
|
| 2860 |
+
},
|
| 2861 |
+
{
|
| 2862 |
+
"epoch": 4.1,
|
| 2863 |
+
"learning_rate": 1.3551724137931035e-05,
|
| 2864 |
+
"loss": 0.0173,
|
| 2865 |
+
"step": 11075
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 4.11,
|
| 2869 |
+
"learning_rate": 1.3465517241379311e-05,
|
| 2870 |
+
"loss": 0.0185,
|
| 2871 |
+
"step": 11100
|
| 2872 |
+
},
|
| 2873 |
+
{
|
| 2874 |
+
"epoch": 4.12,
|
| 2875 |
+
"learning_rate": 1.3379310344827587e-05,
|
| 2876 |
+
"loss": 0.0161,
|
| 2877 |
+
"step": 11125
|
| 2878 |
+
},
|
| 2879 |
+
{
|
| 2880 |
+
"epoch": 4.13,
|
| 2881 |
+
"learning_rate": 1.3293103448275861e-05,
|
| 2882 |
+
"loss": 0.0186,
|
| 2883 |
+
"step": 11150
|
| 2884 |
+
},
|
| 2885 |
+
{
|
| 2886 |
+
"epoch": 4.14,
|
| 2887 |
+
"learning_rate": 1.3206896551724137e-05,
|
| 2888 |
+
"loss": 0.0191,
|
| 2889 |
+
"step": 11175
|
| 2890 |
+
},
|
| 2891 |
+
{
|
| 2892 |
+
"epoch": 4.15,
|
| 2893 |
+
"learning_rate": 1.3120689655172413e-05,
|
| 2894 |
+
"loss": 0.0177,
|
| 2895 |
+
"step": 11200
|
| 2896 |
+
},
|
| 2897 |
+
{
|
| 2898 |
+
"epoch": 4.15,
|
| 2899 |
+
"learning_rate": 1.303448275862069e-05,
|
| 2900 |
+
"loss": 0.0184,
|
| 2901 |
+
"step": 11225
|
| 2902 |
+
},
|
| 2903 |
+
{
|
| 2904 |
+
"epoch": 4.16,
|
| 2905 |
+
"learning_rate": 1.2948275862068966e-05,
|
| 2906 |
+
"loss": 0.0195,
|
| 2907 |
+
"step": 11250
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 4.17,
|
| 2911 |
+
"learning_rate": 1.2862068965517243e-05,
|
| 2912 |
+
"loss": 0.019,
|
| 2913 |
+
"step": 11275
|
| 2914 |
+
},
|
| 2915 |
+
{
|
| 2916 |
+
"epoch": 4.18,
|
| 2917 |
+
"learning_rate": 1.277586206896552e-05,
|
| 2918 |
+
"loss": 0.0183,
|
| 2919 |
+
"step": 11300
|
| 2920 |
+
},
|
| 2921 |
+
{
|
| 2922 |
+
"epoch": 4.19,
|
| 2923 |
+
"learning_rate": 1.2689655172413795e-05,
|
| 2924 |
+
"loss": 0.0188,
|
| 2925 |
+
"step": 11325
|
| 2926 |
+
},
|
| 2927 |
+
{
|
| 2928 |
+
"epoch": 4.2,
|
| 2929 |
+
"learning_rate": 1.260344827586207e-05,
|
| 2930 |
+
"loss": 0.0209,
|
| 2931 |
+
"step": 11350
|
| 2932 |
+
},
|
| 2933 |
+
{
|
| 2934 |
+
"epoch": 4.21,
|
| 2935 |
+
"learning_rate": 1.2517241379310346e-05,
|
| 2936 |
+
"loss": 0.0171,
|
| 2937 |
+
"step": 11375
|
| 2938 |
+
},
|
| 2939 |
+
{
|
| 2940 |
+
"epoch": 4.22,
|
| 2941 |
+
"learning_rate": 1.2431034482758622e-05,
|
| 2942 |
+
"loss": 0.0163,
|
| 2943 |
+
"step": 11400
|
| 2944 |
+
},
|
| 2945 |
+
{
|
| 2946 |
+
"epoch": 4.23,
|
| 2947 |
+
"learning_rate": 1.2344827586206898e-05,
|
| 2948 |
+
"loss": 0.0181,
|
| 2949 |
+
"step": 11425
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 4.24,
|
| 2953 |
+
"learning_rate": 1.2258620689655174e-05,
|
| 2954 |
+
"loss": 0.0169,
|
| 2955 |
+
"step": 11450
|
| 2956 |
+
},
|
| 2957 |
+
{
|
| 2958 |
+
"epoch": 4.25,
|
| 2959 |
+
"learning_rate": 1.2172413793103448e-05,
|
| 2960 |
+
"loss": 0.0197,
|
| 2961 |
+
"step": 11475
|
| 2962 |
+
},
|
| 2963 |
+
{
|
| 2964 |
+
"epoch": 4.26,
|
| 2965 |
+
"learning_rate": 1.2086206896551724e-05,
|
| 2966 |
+
"loss": 0.0163,
|
| 2967 |
+
"step": 11500
|
| 2968 |
+
},
|
| 2969 |
+
{
|
| 2970 |
+
"epoch": 4.26,
|
| 2971 |
+
"eval_cer": 11.010009099181074,
|
| 2972 |
+
"eval_loss": 0.3046729266643524,
|
| 2973 |
+
"eval_runtime": 358.1502,
|
| 2974 |
+
"eval_samples_per_second": 7.206,
|
| 2975 |
+
"eval_steps_per_second": 0.902,
|
| 2976 |
+
"step": 11500
|
| 2977 |
+
},
|
| 2978 |
+
{
|
| 2979 |
+
"epoch": 4.27,
|
| 2980 |
+
"learning_rate": 1.2e-05,
|
| 2981 |
+
"loss": 0.0174,
|
| 2982 |
+
"step": 11525
|
| 2983 |
+
},
|
| 2984 |
+
{
|
| 2985 |
+
"epoch": 4.27,
|
| 2986 |
+
"learning_rate": 1.1913793103448276e-05,
|
| 2987 |
+
"loss": 0.0179,
|
| 2988 |
+
"step": 11550
|
| 2989 |
+
},
|
| 2990 |
+
{
|
| 2991 |
+
"epoch": 4.28,
|
| 2992 |
+
"learning_rate": 1.1827586206896552e-05,
|
| 2993 |
+
"loss": 0.0189,
|
| 2994 |
+
"step": 11575
|
| 2995 |
+
},
|
| 2996 |
+
{
|
| 2997 |
+
"epoch": 4.29,
|
| 2998 |
+
"learning_rate": 1.1741379310344828e-05,
|
| 2999 |
+
"loss": 0.0202,
|
| 3000 |
+
"step": 11600
|
| 3001 |
+
},
|
| 3002 |
+
{
|
| 3003 |
+
"epoch": 4.3,
|
| 3004 |
+
"learning_rate": 1.1655172413793104e-05,
|
| 3005 |
+
"loss": 0.017,
|
| 3006 |
+
"step": 11625
|
| 3007 |
+
},
|
| 3008 |
+
{
|
| 3009 |
+
"epoch": 4.31,
|
| 3010 |
+
"learning_rate": 1.156896551724138e-05,
|
| 3011 |
+
"loss": 0.0186,
|
| 3012 |
+
"step": 11650
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 4.32,
|
| 3016 |
+
"learning_rate": 1.1482758620689655e-05,
|
| 3017 |
+
"loss": 0.0193,
|
| 3018 |
+
"step": 11675
|
| 3019 |
+
},
|
| 3020 |
+
{
|
| 3021 |
+
"epoch": 4.33,
|
| 3022 |
+
"learning_rate": 1.1396551724137931e-05,
|
| 3023 |
+
"loss": 0.0173,
|
| 3024 |
+
"step": 11700
|
| 3025 |
+
},
|
| 3026 |
+
{
|
| 3027 |
+
"epoch": 4.34,
|
| 3028 |
+
"learning_rate": 1.1310344827586207e-05,
|
| 3029 |
+
"loss": 0.018,
|
| 3030 |
+
"step": 11725
|
| 3031 |
+
},
|
| 3032 |
+
{
|
| 3033 |
+
"epoch": 4.35,
|
| 3034 |
+
"learning_rate": 1.1224137931034483e-05,
|
| 3035 |
+
"loss": 0.0166,
|
| 3036 |
+
"step": 11750
|
| 3037 |
+
},
|
| 3038 |
+
{
|
| 3039 |
+
"epoch": 4.36,
|
| 3040 |
+
"learning_rate": 1.1137931034482759e-05,
|
| 3041 |
+
"loss": 0.0176,
|
| 3042 |
+
"step": 11775
|
| 3043 |
+
},
|
| 3044 |
+
{
|
| 3045 |
+
"epoch": 4.37,
|
| 3046 |
+
"learning_rate": 1.1051724137931035e-05,
|
| 3047 |
+
"loss": 0.0195,
|
| 3048 |
+
"step": 11800
|
| 3049 |
+
},
|
| 3050 |
+
{
|
| 3051 |
+
"epoch": 4.38,
|
| 3052 |
+
"learning_rate": 1.0965517241379311e-05,
|
| 3053 |
+
"loss": 0.0179,
|
| 3054 |
+
"step": 11825
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 4.39,
|
| 3058 |
+
"learning_rate": 1.0879310344827587e-05,
|
| 3059 |
+
"loss": 0.0174,
|
| 3060 |
+
"step": 11850
|
| 3061 |
+
},
|
| 3062 |
+
{
|
| 3063 |
+
"epoch": 4.39,
|
| 3064 |
+
"learning_rate": 1.0793103448275863e-05,
|
| 3065 |
+
"loss": 0.0183,
|
| 3066 |
+
"step": 11875
|
| 3067 |
+
},
|
| 3068 |
+
{
|
| 3069 |
+
"epoch": 4.4,
|
| 3070 |
+
"learning_rate": 1.0706896551724138e-05,
|
| 3071 |
+
"loss": 0.0178,
|
| 3072 |
+
"step": 11900
|
| 3073 |
+
},
|
| 3074 |
+
{
|
| 3075 |
+
"epoch": 4.41,
|
| 3076 |
+
"learning_rate": 1.0620689655172414e-05,
|
| 3077 |
+
"loss": 0.0177,
|
| 3078 |
+
"step": 11925
|
| 3079 |
+
},
|
| 3080 |
+
{
|
| 3081 |
+
"epoch": 4.42,
|
| 3082 |
+
"learning_rate": 1.053448275862069e-05,
|
| 3083 |
+
"loss": 0.0186,
|
| 3084 |
+
"step": 11950
|
| 3085 |
+
},
|
| 3086 |
+
{
|
| 3087 |
+
"epoch": 4.43,
|
| 3088 |
+
"learning_rate": 1.0448275862068967e-05,
|
| 3089 |
+
"loss": 0.0174,
|
| 3090 |
+
"step": 11975
|
| 3091 |
+
},
|
| 3092 |
+
{
|
| 3093 |
+
"epoch": 4.44,
|
| 3094 |
+
"learning_rate": 1.0362068965517242e-05,
|
| 3095 |
+
"loss": 0.0175,
|
| 3096 |
+
"step": 12000
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 4.44,
|
| 3100 |
+
"eval_cer": 10.911434637549288,
|
| 3101 |
+
"eval_loss": 0.30570268630981445,
|
| 3102 |
+
"eval_runtime": 358.4295,
|
| 3103 |
+
"eval_samples_per_second": 7.201,
|
| 3104 |
+
"eval_steps_per_second": 0.901,
|
| 3105 |
+
"step": 12000
|
| 3106 |
+
},
|
| 3107 |
+
{
|
| 3108 |
+
"epoch": 4.45,
|
| 3109 |
+
"learning_rate": 1.0275862068965518e-05,
|
| 3110 |
+
"loss": 0.019,
|
| 3111 |
+
"step": 12025
|
| 3112 |
+
},
|
| 3113 |
+
{
|
| 3114 |
+
"epoch": 4.46,
|
| 3115 |
+
"learning_rate": 1.0189655172413794e-05,
|
| 3116 |
+
"loss": 0.0179,
|
| 3117 |
+
"step": 12050
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 4.47,
|
| 3121 |
+
"learning_rate": 1.010344827586207e-05,
|
| 3122 |
+
"loss": 0.0192,
|
| 3123 |
+
"step": 12075
|
| 3124 |
+
},
|
| 3125 |
+
{
|
| 3126 |
+
"epoch": 4.48,
|
| 3127 |
+
"learning_rate": 1.0017241379310344e-05,
|
| 3128 |
+
"loss": 0.0167,
|
| 3129 |
+
"step": 12100
|
| 3130 |
+
},
|
| 3131 |
+
{
|
| 3132 |
+
"epoch": 4.49,
|
| 3133 |
+
"learning_rate": 9.93103448275862e-06,
|
| 3134 |
+
"loss": 0.0178,
|
| 3135 |
+
"step": 12125
|
| 3136 |
+
},
|
| 3137 |
+
{
|
| 3138 |
+
"epoch": 4.5,
|
| 3139 |
+
"learning_rate": 9.844827586206896e-06,
|
| 3140 |
+
"loss": 0.0209,
|
| 3141 |
+
"step": 12150
|
| 3142 |
+
},
|
| 3143 |
+
{
|
| 3144 |
+
"epoch": 4.51,
|
| 3145 |
+
"learning_rate": 9.758620689655174e-06,
|
| 3146 |
+
"loss": 0.019,
|
| 3147 |
+
"step": 12175
|
| 3148 |
+
},
|
| 3149 |
+
{
|
| 3150 |
+
"epoch": 4.52,
|
| 3151 |
+
"learning_rate": 9.672413793103448e-06,
|
| 3152 |
+
"loss": 0.0169,
|
| 3153 |
+
"step": 12200
|
| 3154 |
+
},
|
| 3155 |
+
{
|
| 3156 |
+
"epoch": 4.52,
|
| 3157 |
+
"learning_rate": 9.586206896551724e-06,
|
| 3158 |
+
"loss": 0.0175,
|
| 3159 |
+
"step": 12225
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 4.53,
|
| 3163 |
+
"learning_rate": 9.5e-06,
|
| 3164 |
+
"loss": 0.0166,
|
| 3165 |
+
"step": 12250
|
| 3166 |
+
},
|
| 3167 |
+
{
|
| 3168 |
+
"epoch": 4.54,
|
| 3169 |
+
"learning_rate": 9.413793103448277e-06,
|
| 3170 |
+
"loss": 0.0184,
|
| 3171 |
+
"step": 12275
|
| 3172 |
+
},
|
| 3173 |
+
{
|
| 3174 |
+
"epoch": 4.55,
|
| 3175 |
+
"learning_rate": 9.327586206896553e-06,
|
| 3176 |
+
"loss": 0.0256,
|
| 3177 |
+
"step": 12300
|
| 3178 |
+
},
|
| 3179 |
+
{
|
| 3180 |
+
"epoch": 4.56,
|
| 3181 |
+
"learning_rate": 9.241379310344827e-06,
|
| 3182 |
+
"loss": 0.0177,
|
| 3183 |
+
"step": 12325
|
| 3184 |
+
},
|
| 3185 |
+
{
|
| 3186 |
+
"epoch": 4.57,
|
| 3187 |
+
"learning_rate": 9.155172413793103e-06,
|
| 3188 |
+
"loss": 0.018,
|
| 3189 |
+
"step": 12350
|
| 3190 |
+
},
|
| 3191 |
+
{
|
| 3192 |
+
"epoch": 4.58,
|
| 3193 |
+
"learning_rate": 9.06896551724138e-06,
|
| 3194 |
+
"loss": 0.0187,
|
| 3195 |
+
"step": 12375
|
| 3196 |
+
},
|
| 3197 |
+
{
|
| 3198 |
+
"epoch": 4.59,
|
| 3199 |
+
"learning_rate": 8.982758620689657e-06,
|
| 3200 |
+
"loss": 0.0175,
|
| 3201 |
+
"step": 12400
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 4.6,
|
| 3205 |
+
"learning_rate": 8.896551724137931e-06,
|
| 3206 |
+
"loss": 0.0204,
|
| 3207 |
+
"step": 12425
|
| 3208 |
+
},
|
| 3209 |
+
{
|
| 3210 |
+
"epoch": 4.61,
|
| 3211 |
+
"learning_rate": 8.810344827586207e-06,
|
| 3212 |
+
"loss": 0.0177,
|
| 3213 |
+
"step": 12450
|
| 3214 |
+
},
|
| 3215 |
+
{
|
| 3216 |
+
"epoch": 4.62,
|
| 3217 |
+
"learning_rate": 8.724137931034483e-06,
|
| 3218 |
+
"loss": 0.0174,
|
| 3219 |
+
"step": 12475
|
| 3220 |
+
},
|
| 3221 |
+
{
|
| 3222 |
+
"epoch": 4.63,
|
| 3223 |
+
"learning_rate": 8.63793103448276e-06,
|
| 3224 |
+
"loss": 0.0186,
|
| 3225 |
+
"step": 12500
|
| 3226 |
+
},
|
| 3227 |
+
{
|
| 3228 |
+
"epoch": 4.63,
|
| 3229 |
+
"eval_cer": 11.127540188049743,
|
| 3230 |
+
"eval_loss": 0.30721473693847656,
|
| 3231 |
+
"eval_runtime": 359.2661,
|
| 3232 |
+
"eval_samples_per_second": 7.184,
|
| 3233 |
+
"eval_steps_per_second": 0.899,
|
| 3234 |
+
"step": 12500
|
| 3235 |
+
},
|
| 3236 |
+
{
|
| 3237 |
+
"epoch": 4.64,
|
| 3238 |
+
"learning_rate": 8.555172413793104e-06,
|
| 3239 |
+
"loss": 0.0166,
|
| 3240 |
+
"step": 12525
|
| 3241 |
+
},
|
| 3242 |
+
{
|
| 3243 |
+
"epoch": 4.64,
|
| 3244 |
+
"learning_rate": 8.46896551724138e-06,
|
| 3245 |
+
"loss": 0.0196,
|
| 3246 |
+
"step": 12550
|
| 3247 |
+
},
|
| 3248 |
+
{
|
| 3249 |
+
"epoch": 4.65,
|
| 3250 |
+
"learning_rate": 8.382758620689656e-06,
|
| 3251 |
+
"loss": 0.0187,
|
| 3252 |
+
"step": 12575
|
| 3253 |
+
},
|
| 3254 |
+
{
|
| 3255 |
+
"epoch": 4.66,
|
| 3256 |
+
"learning_rate": 8.296551724137932e-06,
|
| 3257 |
+
"loss": 0.0179,
|
| 3258 |
+
"step": 12600
|
| 3259 |
+
},
|
| 3260 |
+
{
|
| 3261 |
+
"epoch": 4.67,
|
| 3262 |
+
"learning_rate": 8.210344827586206e-06,
|
| 3263 |
+
"loss": 0.016,
|
| 3264 |
+
"step": 12625
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 4.68,
|
| 3268 |
+
"learning_rate": 8.124137931034482e-06,
|
| 3269 |
+
"loss": 0.022,
|
| 3270 |
+
"step": 12650
|
| 3271 |
+
},
|
| 3272 |
+
{
|
| 3273 |
+
"epoch": 4.69,
|
| 3274 |
+
"learning_rate": 8.03793103448276e-06,
|
| 3275 |
+
"loss": 0.0156,
|
| 3276 |
+
"step": 12675
|
| 3277 |
+
},
|
| 3278 |
+
{
|
| 3279 |
+
"epoch": 4.7,
|
| 3280 |
+
"learning_rate": 7.951724137931036e-06,
|
| 3281 |
+
"loss": 0.0172,
|
| 3282 |
+
"step": 12700
|
| 3283 |
+
},
|
| 3284 |
+
{
|
| 3285 |
+
"epoch": 4.71,
|
| 3286 |
+
"learning_rate": 7.86551724137931e-06,
|
| 3287 |
+
"loss": 0.0188,
|
| 3288 |
+
"step": 12725
|
| 3289 |
+
},
|
| 3290 |
+
{
|
| 3291 |
+
"epoch": 4.72,
|
| 3292 |
+
"learning_rate": 7.779310344827586e-06,
|
| 3293 |
+
"loss": 0.0208,
|
| 3294 |
+
"step": 12750
|
| 3295 |
+
},
|
| 3296 |
+
{
|
| 3297 |
+
"epoch": 4.73,
|
| 3298 |
+
"learning_rate": 7.693103448275862e-06,
|
| 3299 |
+
"loss": 0.0176,
|
| 3300 |
+
"step": 12775
|
| 3301 |
+
},
|
| 3302 |
+
{
|
| 3303 |
+
"epoch": 4.74,
|
| 3304 |
+
"learning_rate": 7.6068965517241385e-06,
|
| 3305 |
+
"loss": 0.018,
|
| 3306 |
+
"step": 12800
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 4.75,
|
| 3310 |
+
"learning_rate": 7.520689655172414e-06,
|
| 3311 |
+
"loss": 0.0168,
|
| 3312 |
+
"step": 12825
|
| 3313 |
+
},
|
| 3314 |
+
{
|
| 3315 |
+
"epoch": 4.76,
|
| 3316 |
+
"learning_rate": 7.43448275862069e-06,
|
| 3317 |
+
"loss": 0.0159,
|
| 3318 |
+
"step": 12850
|
| 3319 |
+
},
|
| 3320 |
+
{
|
| 3321 |
+
"epoch": 4.76,
|
| 3322 |
+
"learning_rate": 7.348275862068967e-06,
|
| 3323 |
+
"loss": 0.0166,
|
| 3324 |
+
"step": 12875
|
| 3325 |
+
},
|
| 3326 |
+
{
|
| 3327 |
+
"epoch": 4.77,
|
| 3328 |
+
"learning_rate": 7.262068965517242e-06,
|
| 3329 |
+
"loss": 0.0182,
|
| 3330 |
+
"step": 12900
|
| 3331 |
+
},
|
| 3332 |
+
{
|
| 3333 |
+
"epoch": 4.78,
|
| 3334 |
+
"learning_rate": 7.175862068965518e-06,
|
| 3335 |
+
"loss": 0.0166,
|
| 3336 |
+
"step": 12925
|
| 3337 |
+
},
|
| 3338 |
+
{
|
| 3339 |
+
"epoch": 4.79,
|
| 3340 |
+
"learning_rate": 7.089655172413794e-06,
|
| 3341 |
+
"loss": 0.0168,
|
| 3342 |
+
"step": 12950
|
| 3343 |
+
},
|
| 3344 |
+
{
|
| 3345 |
+
"epoch": 4.8,
|
| 3346 |
+
"learning_rate": 7.003448275862069e-06,
|
| 3347 |
+
"loss": 0.0174,
|
| 3348 |
+
"step": 12975
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 4.81,
|
| 3352 |
+
"learning_rate": 6.917241379310345e-06,
|
| 3353 |
+
"loss": 0.0176,
|
| 3354 |
+
"step": 13000
|
| 3355 |
+
},
|
| 3356 |
+
{
|
| 3357 |
+
"epoch": 4.81,
|
| 3358 |
+
"eval_cer": 11.229905975128904,
|
| 3359 |
+
"eval_loss": 0.30905863642692566,
|
| 3360 |
+
"eval_runtime": 359.315,
|
| 3361 |
+
"eval_samples_per_second": 7.183,
|
| 3362 |
+
"eval_steps_per_second": 0.899,
|
| 3363 |
+
"step": 13000
|
| 3364 |
+
},
|
| 3365 |
+
{
|
| 3366 |
+
"epoch": 4.82,
|
| 3367 |
+
"learning_rate": 6.83103448275862e-06,
|
| 3368 |
+
"loss": 0.0175,
|
| 3369 |
+
"step": 13025
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 4.83,
|
| 3373 |
+
"learning_rate": 6.744827586206896e-06,
|
| 3374 |
+
"loss": 0.017,
|
| 3375 |
+
"step": 13050
|
| 3376 |
+
},
|
| 3377 |
+
{
|
| 3378 |
+
"epoch": 4.84,
|
| 3379 |
+
"learning_rate": 6.658620689655173e-06,
|
| 3380 |
+
"loss": 0.0158,
|
| 3381 |
+
"step": 13075
|
| 3382 |
+
},
|
| 3383 |
+
{
|
| 3384 |
+
"epoch": 4.85,
|
| 3385 |
+
"learning_rate": 6.572413793103449e-06,
|
| 3386 |
+
"loss": 0.0182,
|
| 3387 |
+
"step": 13100
|
| 3388 |
+
},
|
| 3389 |
+
{
|
| 3390 |
+
"epoch": 4.86,
|
| 3391 |
+
"learning_rate": 6.4862068965517245e-06,
|
| 3392 |
+
"loss": 0.0175,
|
| 3393 |
+
"step": 13125
|
| 3394 |
+
},
|
| 3395 |
+
{
|
| 3396 |
+
"epoch": 4.87,
|
| 3397 |
+
"learning_rate": 6.4000000000000006e-06,
|
| 3398 |
+
"loss": 0.0191,
|
| 3399 |
+
"step": 13150
|
| 3400 |
+
},
|
| 3401 |
+
{
|
| 3402 |
+
"epoch": 4.88,
|
| 3403 |
+
"learning_rate": 6.313793103448276e-06,
|
| 3404 |
+
"loss": 0.0153,
|
| 3405 |
+
"step": 13175
|
| 3406 |
+
},
|
| 3407 |
+
{
|
| 3408 |
+
"epoch": 4.89,
|
| 3409 |
+
"learning_rate": 6.227586206896552e-06,
|
| 3410 |
+
"loss": 0.0197,
|
| 3411 |
+
"step": 13200
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 4.89,
|
| 3415 |
+
"learning_rate": 6.141379310344828e-06,
|
| 3416 |
+
"loss": 0.0165,
|
| 3417 |
+
"step": 13225
|
| 3418 |
+
},
|
| 3419 |
+
{
|
| 3420 |
+
"epoch": 4.9,
|
| 3421 |
+
"learning_rate": 6.055172413793104e-06,
|
| 3422 |
+
"loss": 0.0157,
|
| 3423 |
+
"step": 13250
|
| 3424 |
+
},
|
| 3425 |
+
{
|
| 3426 |
+
"epoch": 4.91,
|
| 3427 |
+
"learning_rate": 5.96896551724138e-06,
|
| 3428 |
+
"loss": 0.0156,
|
| 3429 |
+
"step": 13275
|
| 3430 |
+
},
|
| 3431 |
+
{
|
| 3432 |
+
"epoch": 4.92,
|
| 3433 |
+
"learning_rate": 5.882758620689655e-06,
|
| 3434 |
+
"loss": 0.0175,
|
| 3435 |
+
"step": 13300
|
| 3436 |
+
},
|
| 3437 |
+
{
|
| 3438 |
+
"epoch": 4.93,
|
| 3439 |
+
"learning_rate": 5.796551724137932e-06,
|
| 3440 |
+
"loss": 0.0166,
|
| 3441 |
+
"step": 13325
|
| 3442 |
+
},
|
| 3443 |
+
{
|
| 3444 |
+
"epoch": 4.94,
|
| 3445 |
+
"learning_rate": 5.710344827586207e-06,
|
| 3446 |
+
"loss": 0.016,
|
| 3447 |
+
"step": 13350
|
| 3448 |
+
},
|
| 3449 |
+
{
|
| 3450 |
+
"epoch": 4.95,
|
| 3451 |
+
"learning_rate": 5.624137931034483e-06,
|
| 3452 |
+
"loss": 0.0146,
|
| 3453 |
+
"step": 13375
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 4.96,
|
| 3457 |
+
"learning_rate": 5.5379310344827585e-06,
|
| 3458 |
+
"loss": 0.0181,
|
| 3459 |
+
"step": 13400
|
| 3460 |
+
},
|
| 3461 |
+
{
|
| 3462 |
+
"epoch": 4.97,
|
| 3463 |
+
"learning_rate": 5.451724137931035e-06,
|
| 3464 |
+
"loss": 0.016,
|
| 3465 |
+
"step": 13425
|
| 3466 |
+
},
|
| 3467 |
+
{
|
| 3468 |
+
"epoch": 4.98,
|
| 3469 |
+
"learning_rate": 5.3655172413793106e-06,
|
| 3470 |
+
"loss": 0.0186,
|
| 3471 |
+
"step": 13450
|
| 3472 |
+
},
|
| 3473 |
+
{
|
| 3474 |
+
"epoch": 4.99,
|
| 3475 |
+
"learning_rate": 5.279310344827587e-06,
|
| 3476 |
+
"loss": 0.0157,
|
| 3477 |
+
"step": 13475
|
| 3478 |
+
},
|
| 3479 |
+
{
|
| 3480 |
+
"epoch": 5.0,
|
| 3481 |
+
"learning_rate": 5.193103448275862e-06,
|
| 3482 |
+
"loss": 0.0173,
|
| 3483 |
+
"step": 13500
|
| 3484 |
+
},
|
| 3485 |
+
{
|
| 3486 |
+
"epoch": 5.0,
|
| 3487 |
+
"eval_cer": 11.082044282681224,
|
| 3488 |
+
"eval_loss": 0.31240326166152954,
|
| 3489 |
+
"eval_runtime": 358.0319,
|
| 3490 |
+
"eval_samples_per_second": 7.209,
|
| 3491 |
+
"eval_steps_per_second": 0.902,
|
| 3492 |
+
"step": 13500
|
| 3493 |
+
},
|
| 3494 |
+
{
|
| 3495 |
+
"epoch": 5.01,
|
| 3496 |
+
"learning_rate": 5.106896551724139e-06,
|
| 3497 |
+
"loss": 0.0115,
|
| 3498 |
+
"step": 13525
|
| 3499 |
+
},
|
| 3500 |
+
{
|
| 3501 |
+
"epoch": 5.01,
|
| 3502 |
+
"learning_rate": 5.020689655172414e-06,
|
| 3503 |
+
"loss": 0.0083,
|
| 3504 |
+
"step": 13550
|
| 3505 |
+
},
|
| 3506 |
+
{
|
| 3507 |
+
"epoch": 5.02,
|
| 3508 |
+
"learning_rate": 4.93448275862069e-06,
|
| 3509 |
+
"loss": 0.0078,
|
| 3510 |
+
"step": 13575
|
| 3511 |
+
},
|
| 3512 |
+
{
|
| 3513 |
+
"epoch": 5.03,
|
| 3514 |
+
"learning_rate": 4.848275862068966e-06,
|
| 3515 |
+
"loss": 0.0081,
|
| 3516 |
+
"step": 13600
|
| 3517 |
+
},
|
| 3518 |
+
{
|
| 3519 |
+
"epoch": 5.04,
|
| 3520 |
+
"learning_rate": 4.762068965517242e-06,
|
| 3521 |
+
"loss": 0.007,
|
| 3522 |
+
"step": 13625
|
| 3523 |
+
},
|
| 3524 |
+
{
|
| 3525 |
+
"epoch": 5.05,
|
| 3526 |
+
"learning_rate": 4.675862068965517e-06,
|
| 3527 |
+
"loss": 0.008,
|
| 3528 |
+
"step": 13650
|
| 3529 |
+
},
|
| 3530 |
+
{
|
| 3531 |
+
"epoch": 5.06,
|
| 3532 |
+
"learning_rate": 4.589655172413793e-06,
|
| 3533 |
+
"loss": 0.0076,
|
| 3534 |
+
"step": 13675
|
| 3535 |
+
},
|
| 3536 |
+
{
|
| 3537 |
+
"epoch": 5.07,
|
| 3538 |
+
"learning_rate": 4.503448275862069e-06,
|
| 3539 |
+
"loss": 0.0093,
|
| 3540 |
+
"step": 13700
|
| 3541 |
+
},
|
| 3542 |
+
{
|
| 3543 |
+
"epoch": 5.08,
|
| 3544 |
+
"learning_rate": 4.417241379310345e-06,
|
| 3545 |
+
"loss": 0.0073,
|
| 3546 |
+
"step": 13725
|
| 3547 |
+
},
|
| 3548 |
+
{
|
| 3549 |
+
"epoch": 5.09,
|
| 3550 |
+
"learning_rate": 4.331034482758621e-06,
|
| 3551 |
+
"loss": 0.0072,
|
| 3552 |
+
"step": 13750
|
| 3553 |
+
},
|
| 3554 |
+
{
|
| 3555 |
+
"epoch": 5.1,
|
| 3556 |
+
"learning_rate": 4.244827586206897e-06,
|
| 3557 |
+
"loss": 0.0079,
|
| 3558 |
+
"step": 13775
|
| 3559 |
+
},
|
| 3560 |
+
{
|
| 3561 |
+
"epoch": 5.11,
|
| 3562 |
+
"learning_rate": 4.158620689655173e-06,
|
| 3563 |
+
"loss": 0.0076,
|
| 3564 |
+
"step": 13800
|
| 3565 |
+
},
|
| 3566 |
+
{
|
| 3567 |
+
"epoch": 5.12,
|
| 3568 |
+
"learning_rate": 4.072413793103449e-06,
|
| 3569 |
+
"loss": 0.0071,
|
| 3570 |
+
"step": 13825
|
| 3571 |
+
},
|
| 3572 |
+
{
|
| 3573 |
+
"epoch": 5.13,
|
| 3574 |
+
"learning_rate": 3.986206896551725e-06,
|
| 3575 |
+
"loss": 0.0071,
|
| 3576 |
+
"step": 13850
|
| 3577 |
+
},
|
| 3578 |
+
{
|
| 3579 |
+
"epoch": 5.14,
|
| 3580 |
+
"learning_rate": 3.9e-06,
|
| 3581 |
+
"loss": 0.0067,
|
| 3582 |
+
"step": 13875
|
| 3583 |
+
},
|
| 3584 |
+
{
|
| 3585 |
+
"epoch": 5.14,
|
| 3586 |
+
"learning_rate": 3.813793103448276e-06,
|
| 3587 |
+
"loss": 0.0077,
|
| 3588 |
+
"step": 13900
|
| 3589 |
+
},
|
| 3590 |
+
{
|
| 3591 |
+
"epoch": 5.15,
|
| 3592 |
+
"learning_rate": 3.7275862068965524e-06,
|
| 3593 |
+
"loss": 0.0089,
|
| 3594 |
+
"step": 13925
|
| 3595 |
+
},
|
| 3596 |
+
{
|
| 3597 |
+
"epoch": 5.16,
|
| 3598 |
+
"learning_rate": 3.641379310344828e-06,
|
| 3599 |
+
"loss": 0.0077,
|
| 3600 |
+
"step": 13950
|
| 3601 |
+
},
|
| 3602 |
+
{
|
| 3603 |
+
"epoch": 5.17,
|
| 3604 |
+
"learning_rate": 3.5551724137931037e-06,
|
| 3605 |
+
"loss": 0.008,
|
| 3606 |
+
"step": 13975
|
| 3607 |
+
},
|
| 3608 |
+
{
|
| 3609 |
+
"epoch": 5.18,
|
| 3610 |
+
"learning_rate": 3.4689655172413793e-06,
|
| 3611 |
+
"loss": 0.007,
|
| 3612 |
+
"step": 14000
|
| 3613 |
+
},
|
| 3614 |
+
{
|
| 3615 |
+
"epoch": 5.18,
|
| 3616 |
+
"eval_cer": 10.862147406733394,
|
| 3617 |
+
"eval_loss": 0.3205486238002777,
|
| 3618 |
+
"eval_runtime": 361.4728,
|
| 3619 |
+
"eval_samples_per_second": 7.14,
|
| 3620 |
+
"eval_steps_per_second": 0.894,
|
| 3621 |
+
"step": 14000
|
| 3622 |
+
},
|
| 3623 |
+
{
|
| 3624 |
+
"epoch": 5.19,
|
| 3625 |
+
"learning_rate": 3.382758620689655e-06,
|
| 3626 |
+
"loss": 0.0088,
|
| 3627 |
+
"step": 14025
|
| 3628 |
+
},
|
| 3629 |
+
{
|
| 3630 |
+
"epoch": 5.2,
|
| 3631 |
+
"learning_rate": 3.2965517241379314e-06,
|
| 3632 |
+
"loss": 0.0072,
|
| 3633 |
+
"step": 14050
|
| 3634 |
+
},
|
| 3635 |
+
{
|
| 3636 |
+
"epoch": 5.21,
|
| 3637 |
+
"learning_rate": 3.210344827586207e-06,
|
| 3638 |
+
"loss": 0.0085,
|
| 3639 |
+
"step": 14075
|
| 3640 |
+
},
|
| 3641 |
+
{
|
| 3642 |
+
"epoch": 5.22,
|
| 3643 |
+
"learning_rate": 3.1241379310344826e-06,
|
| 3644 |
+
"loss": 0.0068,
|
| 3645 |
+
"step": 14100
|
| 3646 |
+
},
|
| 3647 |
+
{
|
| 3648 |
+
"epoch": 5.23,
|
| 3649 |
+
"learning_rate": 3.0379310344827587e-06,
|
| 3650 |
+
"loss": 0.0081,
|
| 3651 |
+
"step": 14125
|
| 3652 |
+
},
|
| 3653 |
+
{
|
| 3654 |
+
"epoch": 5.24,
|
| 3655 |
+
"learning_rate": 2.9517241379310343e-06,
|
| 3656 |
+
"loss": 0.0074,
|
| 3657 |
+
"step": 14150
|
| 3658 |
+
},
|
| 3659 |
+
{
|
| 3660 |
+
"epoch": 5.25,
|
| 3661 |
+
"learning_rate": 2.8655172413793104e-06,
|
| 3662 |
+
"loss": 0.0077,
|
| 3663 |
+
"step": 14175
|
| 3664 |
+
},
|
| 3665 |
+
{
|
| 3666 |
+
"epoch": 5.26,
|
| 3667 |
+
"learning_rate": 2.779310344827586e-06,
|
| 3668 |
+
"loss": 0.0083,
|
| 3669 |
+
"step": 14200
|
| 3670 |
+
},
|
| 3671 |
+
{
|
| 3672 |
+
"epoch": 5.26,
|
| 3673 |
+
"learning_rate": 2.693103448275862e-06,
|
| 3674 |
+
"loss": 0.008,
|
| 3675 |
+
"step": 14225
|
| 3676 |
+
},
|
| 3677 |
+
{
|
| 3678 |
+
"epoch": 5.27,
|
| 3679 |
+
"learning_rate": 2.606896551724138e-06,
|
| 3680 |
+
"loss": 0.0064,
|
| 3681 |
+
"step": 14250
|
| 3682 |
+
},
|
| 3683 |
+
{
|
| 3684 |
+
"epoch": 5.28,
|
| 3685 |
+
"learning_rate": 2.520689655172414e-06,
|
| 3686 |
+
"loss": 0.0076,
|
| 3687 |
+
"step": 14275
|
| 3688 |
+
},
|
| 3689 |
+
{
|
| 3690 |
+
"epoch": 5.29,
|
| 3691 |
+
"learning_rate": 2.4344827586206897e-06,
|
| 3692 |
+
"loss": 0.0066,
|
| 3693 |
+
"step": 14300
|
| 3694 |
+
},
|
| 3695 |
+
{
|
| 3696 |
+
"epoch": 5.3,
|
| 3697 |
+
"learning_rate": 2.3482758620689658e-06,
|
| 3698 |
+
"loss": 0.008,
|
| 3699 |
+
"step": 14325
|
| 3700 |
+
},
|
| 3701 |
+
{
|
| 3702 |
+
"epoch": 5.31,
|
| 3703 |
+
"learning_rate": 2.2620689655172414e-06,
|
| 3704 |
+
"loss": 0.0072,
|
| 3705 |
+
"step": 14350
|
| 3706 |
+
},
|
| 3707 |
+
{
|
| 3708 |
+
"epoch": 5.32,
|
| 3709 |
+
"learning_rate": 2.1758620689655174e-06,
|
| 3710 |
+
"loss": 0.0078,
|
| 3711 |
+
"step": 14375
|
| 3712 |
+
},
|
| 3713 |
+
{
|
| 3714 |
+
"epoch": 5.33,
|
| 3715 |
+
"learning_rate": 2.089655172413793e-06,
|
| 3716 |
+
"loss": 0.0091,
|
| 3717 |
+
"step": 14400
|
| 3718 |
+
},
|
| 3719 |
+
{
|
| 3720 |
+
"epoch": 5.34,
|
| 3721 |
+
"learning_rate": 2.003448275862069e-06,
|
| 3722 |
+
"loss": 0.007,
|
| 3723 |
+
"step": 14425
|
| 3724 |
+
},
|
| 3725 |
+
{
|
| 3726 |
+
"epoch": 5.35,
|
| 3727 |
+
"learning_rate": 1.9172413793103447e-06,
|
| 3728 |
+
"loss": 0.0078,
|
| 3729 |
+
"step": 14450
|
| 3730 |
+
},
|
| 3731 |
+
{
|
| 3732 |
+
"epoch": 5.36,
|
| 3733 |
+
"learning_rate": 1.8310344827586208e-06,
|
| 3734 |
+
"loss": 0.0074,
|
| 3735 |
+
"step": 14475
|
| 3736 |
+
},
|
| 3737 |
+
{
|
| 3738 |
+
"epoch": 5.37,
|
| 3739 |
+
"learning_rate": 1.7448275862068964e-06,
|
| 3740 |
+
"loss": 0.0073,
|
| 3741 |
+
"step": 14500
|
| 3742 |
+
},
|
| 3743 |
+
{
|
| 3744 |
+
"epoch": 5.37,
|
| 3745 |
+
"eval_cer": 10.790112223233242,
|
| 3746 |
+
"eval_loss": 0.320094496011734,
|
| 3747 |
+
"eval_runtime": 359.2372,
|
| 3748 |
+
"eval_samples_per_second": 7.185,
|
| 3749 |
+
"eval_steps_per_second": 0.899,
|
| 3750 |
+
"step": 14500
|
| 3751 |
+
},
|
| 3752 |
+
{
|
| 3753 |
+
"epoch": 5.38,
|
| 3754 |
+
"learning_rate": 1.6586206896551724e-06,
|
| 3755 |
+
"loss": 0.0094,
|
| 3756 |
+
"step": 14525
|
| 3757 |
+
},
|
| 3758 |
+
{
|
| 3759 |
+
"epoch": 5.38,
|
| 3760 |
+
"learning_rate": 1.5724137931034483e-06,
|
| 3761 |
+
"loss": 0.0142,
|
| 3762 |
+
"step": 14550
|
| 3763 |
+
},
|
| 3764 |
+
{
|
| 3765 |
+
"epoch": 5.39,
|
| 3766 |
+
"learning_rate": 1.4862068965517243e-06,
|
| 3767 |
+
"loss": 0.0085,
|
| 3768 |
+
"step": 14575
|
| 3769 |
+
},
|
| 3770 |
+
{
|
| 3771 |
+
"epoch": 5.4,
|
| 3772 |
+
"learning_rate": 1.4034482758620691e-06,
|
| 3773 |
+
"loss": 0.0113,
|
| 3774 |
+
"step": 14600
|
| 3775 |
+
},
|
| 3776 |
+
{
|
| 3777 |
+
"epoch": 5.41,
|
| 3778 |
+
"learning_rate": 1.317241379310345e-06,
|
| 3779 |
+
"loss": 0.0085,
|
| 3780 |
+
"step": 14625
|
| 3781 |
+
},
|
| 3782 |
+
{
|
| 3783 |
+
"epoch": 5.42,
|
| 3784 |
+
"learning_rate": 1.2310344827586208e-06,
|
| 3785 |
+
"loss": 0.0076,
|
| 3786 |
+
"step": 14650
|
| 3787 |
+
},
|
| 3788 |
+
{
|
| 3789 |
+
"epoch": 5.43,
|
| 3790 |
+
"learning_rate": 1.1448275862068966e-06,
|
| 3791 |
+
"loss": 0.0074,
|
| 3792 |
+
"step": 14675
|
| 3793 |
+
},
|
| 3794 |
+
{
|
| 3795 |
+
"epoch": 5.44,
|
| 3796 |
+
"learning_rate": 1.0586206896551725e-06,
|
| 3797 |
+
"loss": 0.0079,
|
| 3798 |
+
"step": 14700
|
| 3799 |
+
},
|
| 3800 |
+
{
|
| 3801 |
+
"epoch": 5.45,
|
| 3802 |
+
"learning_rate": 9.724137931034483e-07,
|
| 3803 |
+
"loss": 0.0067,
|
| 3804 |
+
"step": 14725
|
| 3805 |
+
},
|
| 3806 |
+
{
|
| 3807 |
+
"epoch": 5.46,
|
| 3808 |
+
"learning_rate": 8.862068965517242e-07,
|
| 3809 |
+
"loss": 0.0075,
|
| 3810 |
+
"step": 14750
|
| 3811 |
+
},
|
| 3812 |
+
{
|
| 3813 |
+
"epoch": 5.47,
|
| 3814 |
+
"learning_rate": 8.000000000000001e-07,
|
| 3815 |
+
"loss": 0.0082,
|
| 3816 |
+
"step": 14775
|
| 3817 |
+
},
|
| 3818 |
+
{
|
| 3819 |
+
"epoch": 5.48,
|
| 3820 |
+
"learning_rate": 7.137931034482759e-07,
|
| 3821 |
+
"loss": 0.008,
|
| 3822 |
+
"step": 14800
|
| 3823 |
+
},
|
| 3824 |
+
{
|
| 3825 |
+
"epoch": 5.49,
|
| 3826 |
+
"learning_rate": 6.275862068965517e-07,
|
| 3827 |
+
"loss": 0.0066,
|
| 3828 |
+
"step": 14825
|
| 3829 |
+
},
|
| 3830 |
+
{
|
| 3831 |
+
"epoch": 5.5,
|
| 3832 |
+
"learning_rate": 5.413793103448277e-07,
|
| 3833 |
+
"loss": 0.0077,
|
| 3834 |
+
"step": 14850
|
| 3835 |
+
},
|
| 3836 |
+
{
|
| 3837 |
+
"epoch": 5.51,
|
| 3838 |
+
"learning_rate": 4.551724137931035e-07,
|
| 3839 |
+
"loss": 0.0075,
|
| 3840 |
+
"step": 14875
|
| 3841 |
+
},
|
| 3842 |
+
{
|
| 3843 |
+
"epoch": 5.51,
|
| 3844 |
+
"learning_rate": 3.689655172413793e-07,
|
| 3845 |
+
"loss": 0.0086,
|
| 3846 |
+
"step": 14900
|
| 3847 |
+
},
|
| 3848 |
+
{
|
| 3849 |
+
"epoch": 5.52,
|
| 3850 |
+
"learning_rate": 2.827586206896552e-07,
|
| 3851 |
+
"loss": 0.0067,
|
| 3852 |
+
"step": 14925
|
| 3853 |
+
},
|
| 3854 |
+
{
|
| 3855 |
+
"epoch": 5.53,
|
| 3856 |
+
"learning_rate": 1.9655172413793103e-07,
|
| 3857 |
+
"loss": 0.0075,
|
| 3858 |
+
"step": 14950
|
| 3859 |
+
},
|
| 3860 |
+
{
|
| 3861 |
+
"epoch": 5.54,
|
| 3862 |
+
"learning_rate": 1.1034482758620691e-07,
|
| 3863 |
+
"loss": 0.0067,
|
| 3864 |
+
"step": 14975
|
| 3865 |
+
},
|
| 3866 |
+
{
|
| 3867 |
+
"epoch": 5.55,
|
| 3868 |
+
"learning_rate": 2.413793103448276e-08,
|
| 3869 |
+
"loss": 0.0072,
|
| 3870 |
+
"step": 15000
|
| 3871 |
+
},
|
| 3872 |
+
{
|
| 3873 |
+
"epoch": 5.55,
|
| 3874 |
+
"eval_cer": 10.763572945101608,
|
| 3875 |
+
"eval_loss": 0.3218766450881958,
|
| 3876 |
+
"eval_runtime": 359.0558,
|
| 3877 |
+
"eval_samples_per_second": 7.188,
|
| 3878 |
+
"eval_steps_per_second": 0.9,
|
| 3879 |
+
"step": 15000
|
| 3880 |
+
}
|
| 3881 |
+
],
|
| 3882 |
+
"logging_steps": 25,
|
| 3883 |
+
"max_steps": 15000,
|
| 3884 |
+
"num_input_tokens_seen": 0,
|
| 3885 |
+
"num_train_epochs": 6,
|
| 3886 |
+
"save_steps": 500,
|
| 3887 |
+
"total_flos": 4.328506848669696e+20,
|
| 3888 |
+
"train_batch_size": 25,
|
| 3889 |
+
"trial_name": null,
|
| 3890 |
+
"trial_params": null
|
| 3891 |
+
}
|