Upload trainer_state.json
Browse files- trainer_state.json +1951 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,1951 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": 4.054176686782522,
|
| 3 |
+
"best_model_checkpoint": "./raid/liyan/whisper/whisper-medium-mls_sjt/checkpoint-7500",
|
| 4 |
+
"epoch": 2.5737817433081673,
|
| 5 |
+
"global_step": 7500,
|
| 6 |
+
"is_hyper_param_search": false,
|
| 7 |
+
"is_local_process_zero": true,
|
| 8 |
+
"is_world_process_zero": true,
|
| 9 |
+
"log_history": [
|
| 10 |
+
{
|
| 11 |
+
"epoch": 0.01,
|
| 12 |
+
"learning_rate": 5.000000000000001e-07,
|
| 13 |
+
"loss": 0.4039,
|
| 14 |
+
"step": 25
|
| 15 |
+
},
|
| 16 |
+
{
|
| 17 |
+
"epoch": 0.02,
|
| 18 |
+
"learning_rate": 9.800000000000001e-07,
|
| 19 |
+
"loss": 0.3669,
|
| 20 |
+
"step": 50
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"epoch": 0.03,
|
| 24 |
+
"learning_rate": 1.46e-06,
|
| 25 |
+
"loss": 0.3318,
|
| 26 |
+
"step": 75
|
| 27 |
+
},
|
| 28 |
+
{
|
| 29 |
+
"epoch": 0.03,
|
| 30 |
+
"learning_rate": 1.9600000000000003e-06,
|
| 31 |
+
"loss": 0.3159,
|
| 32 |
+
"step": 100
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.04,
|
| 36 |
+
"learning_rate": 2.46e-06,
|
| 37 |
+
"loss": 0.296,
|
| 38 |
+
"step": 125
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.05,
|
| 42 |
+
"learning_rate": 2.96e-06,
|
| 43 |
+
"loss": 0.3047,
|
| 44 |
+
"step": 150
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.06,
|
| 48 |
+
"learning_rate": 3.46e-06,
|
| 49 |
+
"loss": 0.2821,
|
| 50 |
+
"step": 175
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"epoch": 0.07,
|
| 54 |
+
"learning_rate": 3.96e-06,
|
| 55 |
+
"loss": 0.2893,
|
| 56 |
+
"step": 200
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.08,
|
| 60 |
+
"learning_rate": 4.4600000000000005e-06,
|
| 61 |
+
"loss": 0.2679,
|
| 62 |
+
"step": 225
|
| 63 |
+
},
|
| 64 |
+
{
|
| 65 |
+
"epoch": 0.09,
|
| 66 |
+
"learning_rate": 4.960000000000001e-06,
|
| 67 |
+
"loss": 0.2591,
|
| 68 |
+
"step": 250
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.09,
|
| 72 |
+
"learning_rate": 5.460000000000001e-06,
|
| 73 |
+
"loss": 0.2453,
|
| 74 |
+
"step": 275
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.1,
|
| 78 |
+
"learning_rate": 5.9600000000000005e-06,
|
| 79 |
+
"loss": 0.2507,
|
| 80 |
+
"step": 300
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.11,
|
| 84 |
+
"learning_rate": 6.460000000000001e-06,
|
| 85 |
+
"loss": 0.2435,
|
| 86 |
+
"step": 325
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.12,
|
| 90 |
+
"learning_rate": 6.96e-06,
|
| 91 |
+
"loss": 0.2305,
|
| 92 |
+
"step": 350
|
| 93 |
+
},
|
| 94 |
+
{
|
| 95 |
+
"epoch": 0.13,
|
| 96 |
+
"learning_rate": 7.4600000000000006e-06,
|
| 97 |
+
"loss": 0.2224,
|
| 98 |
+
"step": 375
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.14,
|
| 102 |
+
"learning_rate": 7.960000000000002e-06,
|
| 103 |
+
"loss": 0.2443,
|
| 104 |
+
"step": 400
|
| 105 |
+
},
|
| 106 |
+
{
|
| 107 |
+
"epoch": 0.15,
|
| 108 |
+
"learning_rate": 8.46e-06,
|
| 109 |
+
"loss": 0.2443,
|
| 110 |
+
"step": 425
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.15,
|
| 114 |
+
"learning_rate": 8.96e-06,
|
| 115 |
+
"loss": 0.2292,
|
| 116 |
+
"step": 450
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.16,
|
| 120 |
+
"learning_rate": 9.460000000000001e-06,
|
| 121 |
+
"loss": 0.2445,
|
| 122 |
+
"step": 475
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.17,
|
| 126 |
+
"learning_rate": 9.960000000000001e-06,
|
| 127 |
+
"loss": 0.2463,
|
| 128 |
+
"step": 500
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.17,
|
| 132 |
+
"eval_loss": 0.14288775622844696,
|
| 133 |
+
"eval_runtime": 13972.8923,
|
| 134 |
+
"eval_samples_per_second": 1.601,
|
| 135 |
+
"eval_steps_per_second": 0.067,
|
| 136 |
+
"eval_wer": 5.199872952493307,
|
| 137 |
+
"step": 500
|
| 138 |
+
},
|
| 139 |
+
{
|
| 140 |
+
"epoch": 0.18,
|
| 141 |
+
"learning_rate": 9.99220338983051e-06,
|
| 142 |
+
"loss": 0.2158,
|
| 143 |
+
"step": 525
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.19,
|
| 147 |
+
"learning_rate": 9.983728813559322e-06,
|
| 148 |
+
"loss": 0.2302,
|
| 149 |
+
"step": 550
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.2,
|
| 153 |
+
"learning_rate": 9.975254237288137e-06,
|
| 154 |
+
"loss": 0.2223,
|
| 155 |
+
"step": 575
|
| 156 |
+
},
|
| 157 |
+
{
|
| 158 |
+
"epoch": 0.21,
|
| 159 |
+
"learning_rate": 9.96677966101695e-06,
|
| 160 |
+
"loss": 0.216,
|
| 161 |
+
"step": 600
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.21,
|
| 165 |
+
"learning_rate": 9.958305084745764e-06,
|
| 166 |
+
"loss": 0.2111,
|
| 167 |
+
"step": 625
|
| 168 |
+
},
|
| 169 |
+
{
|
| 170 |
+
"epoch": 0.22,
|
| 171 |
+
"learning_rate": 9.949830508474577e-06,
|
| 172 |
+
"loss": 0.2208,
|
| 173 |
+
"step": 650
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.23,
|
| 177 |
+
"learning_rate": 9.94135593220339e-06,
|
| 178 |
+
"loss": 0.2254,
|
| 179 |
+
"step": 675
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.24,
|
| 183 |
+
"learning_rate": 9.932881355932205e-06,
|
| 184 |
+
"loss": 0.2008,
|
| 185 |
+
"step": 700
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.25,
|
| 189 |
+
"learning_rate": 9.924406779661018e-06,
|
| 190 |
+
"loss": 0.2284,
|
| 191 |
+
"step": 725
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.26,
|
| 195 |
+
"learning_rate": 9.915932203389831e-06,
|
| 196 |
+
"loss": 0.2193,
|
| 197 |
+
"step": 750
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.27,
|
| 201 |
+
"learning_rate": 9.907457627118646e-06,
|
| 202 |
+
"loss": 0.2049,
|
| 203 |
+
"step": 775
|
| 204 |
+
},
|
| 205 |
+
{
|
| 206 |
+
"epoch": 0.27,
|
| 207 |
+
"learning_rate": 9.898983050847459e-06,
|
| 208 |
+
"loss": 0.2048,
|
| 209 |
+
"step": 800
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.28,
|
| 213 |
+
"learning_rate": 9.890508474576272e-06,
|
| 214 |
+
"loss": 0.2272,
|
| 215 |
+
"step": 825
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.29,
|
| 219 |
+
"learning_rate": 9.882033898305086e-06,
|
| 220 |
+
"loss": 0.2096,
|
| 221 |
+
"step": 850
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.3,
|
| 225 |
+
"learning_rate": 9.8735593220339e-06,
|
| 226 |
+
"loss": 0.2078,
|
| 227 |
+
"step": 875
|
| 228 |
+
},
|
| 229 |
+
{
|
| 230 |
+
"epoch": 0.31,
|
| 231 |
+
"learning_rate": 9.865084745762712e-06,
|
| 232 |
+
"loss": 0.206,
|
| 233 |
+
"step": 900
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.32,
|
| 237 |
+
"learning_rate": 9.856610169491525e-06,
|
| 238 |
+
"loss": 0.211,
|
| 239 |
+
"step": 925
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.33,
|
| 243 |
+
"learning_rate": 9.84813559322034e-06,
|
| 244 |
+
"loss": 0.2239,
|
| 245 |
+
"step": 950
|
| 246 |
+
},
|
| 247 |
+
{
|
| 248 |
+
"epoch": 0.33,
|
| 249 |
+
"learning_rate": 9.839661016949153e-06,
|
| 250 |
+
"loss": 0.2099,
|
| 251 |
+
"step": 975
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.34,
|
| 255 |
+
"learning_rate": 9.831186440677966e-06,
|
| 256 |
+
"loss": 0.208,
|
| 257 |
+
"step": 1000
|
| 258 |
+
},
|
| 259 |
+
{
|
| 260 |
+
"epoch": 0.34,
|
| 261 |
+
"eval_loss": 0.1417330801486969,
|
| 262 |
+
"eval_runtime": 14048.5832,
|
| 263 |
+
"eval_samples_per_second": 1.592,
|
| 264 |
+
"eval_steps_per_second": 0.066,
|
| 265 |
+
"eval_wer": 5.172194745678116,
|
| 266 |
+
"step": 1000
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.35,
|
| 270 |
+
"learning_rate": 9.82271186440678e-06,
|
| 271 |
+
"loss": 0.2234,
|
| 272 |
+
"step": 1025
|
| 273 |
+
},
|
| 274 |
+
{
|
| 275 |
+
"epoch": 0.36,
|
| 276 |
+
"learning_rate": 9.814237288135594e-06,
|
| 277 |
+
"loss": 0.1974,
|
| 278 |
+
"step": 1050
|
| 279 |
+
},
|
| 280 |
+
{
|
| 281 |
+
"epoch": 0.37,
|
| 282 |
+
"learning_rate": 9.805762711864407e-06,
|
| 283 |
+
"loss": 0.2175,
|
| 284 |
+
"step": 1075
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.38,
|
| 288 |
+
"learning_rate": 9.797288135593221e-06,
|
| 289 |
+
"loss": 0.207,
|
| 290 |
+
"step": 1100
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.39,
|
| 294 |
+
"learning_rate": 9.788813559322034e-06,
|
| 295 |
+
"loss": 0.1922,
|
| 296 |
+
"step": 1125
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.39,
|
| 300 |
+
"learning_rate": 9.780338983050847e-06,
|
| 301 |
+
"loss": 0.2128,
|
| 302 |
+
"step": 1150
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.4,
|
| 306 |
+
"learning_rate": 9.771864406779662e-06,
|
| 307 |
+
"loss": 0.2038,
|
| 308 |
+
"step": 1175
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.41,
|
| 312 |
+
"learning_rate": 9.763389830508475e-06,
|
| 313 |
+
"loss": 0.1996,
|
| 314 |
+
"step": 1200
|
| 315 |
+
},
|
| 316 |
+
{
|
| 317 |
+
"epoch": 0.42,
|
| 318 |
+
"learning_rate": 9.75491525423729e-06,
|
| 319 |
+
"loss": 0.1982,
|
| 320 |
+
"step": 1225
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.43,
|
| 324 |
+
"learning_rate": 9.746440677966103e-06,
|
| 325 |
+
"loss": 0.2026,
|
| 326 |
+
"step": 1250
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.44,
|
| 330 |
+
"learning_rate": 9.737966101694916e-06,
|
| 331 |
+
"loss": 0.2056,
|
| 332 |
+
"step": 1275
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.45,
|
| 336 |
+
"learning_rate": 9.72949152542373e-06,
|
| 337 |
+
"loss": 0.2038,
|
| 338 |
+
"step": 1300
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.45,
|
| 342 |
+
"learning_rate": 9.721016949152543e-06,
|
| 343 |
+
"loss": 0.1906,
|
| 344 |
+
"step": 1325
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.46,
|
| 348 |
+
"learning_rate": 9.712542372881358e-06,
|
| 349 |
+
"loss": 0.1953,
|
| 350 |
+
"step": 1350
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"epoch": 0.47,
|
| 354 |
+
"learning_rate": 9.70406779661017e-06,
|
| 355 |
+
"loss": 0.1952,
|
| 356 |
+
"step": 1375
|
| 357 |
+
},
|
| 358 |
+
{
|
| 359 |
+
"epoch": 0.48,
|
| 360 |
+
"learning_rate": 9.695593220338984e-06,
|
| 361 |
+
"loss": 0.1988,
|
| 362 |
+
"step": 1400
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.49,
|
| 366 |
+
"learning_rate": 9.687118644067798e-06,
|
| 367 |
+
"loss": 0.1956,
|
| 368 |
+
"step": 1425
|
| 369 |
+
},
|
| 370 |
+
{
|
| 371 |
+
"epoch": 0.5,
|
| 372 |
+
"learning_rate": 9.678644067796611e-06,
|
| 373 |
+
"loss": 0.19,
|
| 374 |
+
"step": 1450
|
| 375 |
+
},
|
| 376 |
+
{
|
| 377 |
+
"epoch": 0.51,
|
| 378 |
+
"learning_rate": 9.670169491525424e-06,
|
| 379 |
+
"loss": 0.2183,
|
| 380 |
+
"step": 1475
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 0.51,
|
| 384 |
+
"learning_rate": 9.661694915254239e-06,
|
| 385 |
+
"loss": 0.1744,
|
| 386 |
+
"step": 1500
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.51,
|
| 390 |
+
"eval_loss": 0.1370771825313568,
|
| 391 |
+
"eval_runtime": 14137.0887,
|
| 392 |
+
"eval_samples_per_second": 1.582,
|
| 393 |
+
"eval_steps_per_second": 0.066,
|
| 394 |
+
"eval_wer": 4.992967013022369,
|
| 395 |
+
"step": 1500
|
| 396 |
+
},
|
| 397 |
+
{
|
| 398 |
+
"epoch": 0.52,
|
| 399 |
+
"learning_rate": 9.653220338983052e-06,
|
| 400 |
+
"loss": 0.2261,
|
| 401 |
+
"step": 1525
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 0.53,
|
| 405 |
+
"learning_rate": 9.644745762711865e-06,
|
| 406 |
+
"loss": 0.2001,
|
| 407 |
+
"step": 1550
|
| 408 |
+
},
|
| 409 |
+
{
|
| 410 |
+
"epoch": 0.54,
|
| 411 |
+
"learning_rate": 9.636271186440678e-06,
|
| 412 |
+
"loss": 0.1984,
|
| 413 |
+
"step": 1575
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.55,
|
| 417 |
+
"learning_rate": 9.627796610169493e-06,
|
| 418 |
+
"loss": 0.2003,
|
| 419 |
+
"step": 1600
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.56,
|
| 423 |
+
"learning_rate": 9.619322033898306e-06,
|
| 424 |
+
"loss": 0.1771,
|
| 425 |
+
"step": 1625
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 0.57,
|
| 429 |
+
"learning_rate": 9.610847457627119e-06,
|
| 430 |
+
"loss": 0.1979,
|
| 431 |
+
"step": 1650
|
| 432 |
+
},
|
| 433 |
+
{
|
| 434 |
+
"epoch": 0.57,
|
| 435 |
+
"learning_rate": 9.602372881355932e-06,
|
| 436 |
+
"loss": 0.2013,
|
| 437 |
+
"step": 1675
|
| 438 |
+
},
|
| 439 |
+
{
|
| 440 |
+
"epoch": 0.58,
|
| 441 |
+
"learning_rate": 9.593898305084746e-06,
|
| 442 |
+
"loss": 0.1899,
|
| 443 |
+
"step": 1700
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 0.59,
|
| 447 |
+
"learning_rate": 9.58542372881356e-06,
|
| 448 |
+
"loss": 0.1782,
|
| 449 |
+
"step": 1725
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 0.6,
|
| 453 |
+
"learning_rate": 9.576949152542374e-06,
|
| 454 |
+
"loss": 0.1689,
|
| 455 |
+
"step": 1750
|
| 456 |
+
},
|
| 457 |
+
{
|
| 458 |
+
"epoch": 0.61,
|
| 459 |
+
"learning_rate": 9.568474576271187e-06,
|
| 460 |
+
"loss": 0.1985,
|
| 461 |
+
"step": 1775
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.62,
|
| 465 |
+
"learning_rate": 9.56e-06,
|
| 466 |
+
"loss": 0.1788,
|
| 467 |
+
"step": 1800
|
| 468 |
+
},
|
| 469 |
+
{
|
| 470 |
+
"epoch": 0.63,
|
| 471 |
+
"learning_rate": 9.551525423728815e-06,
|
| 472 |
+
"loss": 0.186,
|
| 473 |
+
"step": 1825
|
| 474 |
+
},
|
| 475 |
+
{
|
| 476 |
+
"epoch": 0.63,
|
| 477 |
+
"learning_rate": 9.543050847457628e-06,
|
| 478 |
+
"loss": 0.1873,
|
| 479 |
+
"step": 1850
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.64,
|
| 483 |
+
"learning_rate": 9.53457627118644e-06,
|
| 484 |
+
"loss": 0.1944,
|
| 485 |
+
"step": 1875
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 0.65,
|
| 489 |
+
"learning_rate": 9.526101694915255e-06,
|
| 490 |
+
"loss": 0.1954,
|
| 491 |
+
"step": 1900
|
| 492 |
+
},
|
| 493 |
+
{
|
| 494 |
+
"epoch": 0.66,
|
| 495 |
+
"learning_rate": 9.517627118644068e-06,
|
| 496 |
+
"loss": 0.1948,
|
| 497 |
+
"step": 1925
|
| 498 |
+
},
|
| 499 |
+
{
|
| 500 |
+
"epoch": 0.67,
|
| 501 |
+
"learning_rate": 9.509152542372883e-06,
|
| 502 |
+
"loss": 0.1785,
|
| 503 |
+
"step": 1950
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 0.68,
|
| 507 |
+
"learning_rate": 9.500677966101696e-06,
|
| 508 |
+
"loss": 0.1795,
|
| 509 |
+
"step": 1975
|
| 510 |
+
},
|
| 511 |
+
{
|
| 512 |
+
"epoch": 0.69,
|
| 513 |
+
"learning_rate": 9.492203389830509e-06,
|
| 514 |
+
"loss": 0.1821,
|
| 515 |
+
"step": 2000
|
| 516 |
+
},
|
| 517 |
+
{
|
| 518 |
+
"epoch": 0.69,
|
| 519 |
+
"eval_loss": 0.13245970010757446,
|
| 520 |
+
"eval_runtime": 13962.1433,
|
| 521 |
+
"eval_samples_per_second": 1.602,
|
| 522 |
+
"eval_steps_per_second": 0.067,
|
| 523 |
+
"eval_wer": 4.777893733835473,
|
| 524 |
+
"step": 2000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.69,
|
| 528 |
+
"learning_rate": 9.483728813559324e-06,
|
| 529 |
+
"loss": 0.1895,
|
| 530 |
+
"step": 2025
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.7,
|
| 534 |
+
"learning_rate": 9.475254237288137e-06,
|
| 535 |
+
"loss": 0.1693,
|
| 536 |
+
"step": 2050
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 0.71,
|
| 540 |
+
"learning_rate": 9.466779661016951e-06,
|
| 541 |
+
"loss": 0.1791,
|
| 542 |
+
"step": 2075
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 0.72,
|
| 546 |
+
"learning_rate": 9.458305084745764e-06,
|
| 547 |
+
"loss": 0.1889,
|
| 548 |
+
"step": 2100
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 0.73,
|
| 552 |
+
"learning_rate": 9.449830508474577e-06,
|
| 553 |
+
"loss": 0.17,
|
| 554 |
+
"step": 2125
|
| 555 |
+
},
|
| 556 |
+
{
|
| 557 |
+
"epoch": 0.74,
|
| 558 |
+
"learning_rate": 9.44135593220339e-06,
|
| 559 |
+
"loss": 0.2053,
|
| 560 |
+
"step": 2150
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 0.75,
|
| 564 |
+
"learning_rate": 9.432881355932205e-06,
|
| 565 |
+
"loss": 0.1754,
|
| 566 |
+
"step": 2175
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 0.75,
|
| 570 |
+
"learning_rate": 9.424406779661018e-06,
|
| 571 |
+
"loss": 0.1791,
|
| 572 |
+
"step": 2200
|
| 573 |
+
},
|
| 574 |
+
{
|
| 575 |
+
"epoch": 0.76,
|
| 576 |
+
"learning_rate": 9.41593220338983e-06,
|
| 577 |
+
"loss": 0.1901,
|
| 578 |
+
"step": 2225
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.77,
|
| 582 |
+
"learning_rate": 9.407457627118645e-06,
|
| 583 |
+
"loss": 0.1682,
|
| 584 |
+
"step": 2250
|
| 585 |
+
},
|
| 586 |
+
{
|
| 587 |
+
"epoch": 0.78,
|
| 588 |
+
"learning_rate": 9.398983050847458e-06,
|
| 589 |
+
"loss": 0.176,
|
| 590 |
+
"step": 2275
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 0.79,
|
| 594 |
+
"learning_rate": 9.390508474576271e-06,
|
| 595 |
+
"loss": 0.1846,
|
| 596 |
+
"step": 2300
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 0.8,
|
| 600 |
+
"learning_rate": 9.382033898305084e-06,
|
| 601 |
+
"loss": 0.1789,
|
| 602 |
+
"step": 2325
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 0.81,
|
| 606 |
+
"learning_rate": 9.373559322033899e-06,
|
| 607 |
+
"loss": 0.1645,
|
| 608 |
+
"step": 2350
|
| 609 |
+
},
|
| 610 |
+
{
|
| 611 |
+
"epoch": 0.82,
|
| 612 |
+
"learning_rate": 9.365084745762712e-06,
|
| 613 |
+
"loss": 0.1772,
|
| 614 |
+
"step": 2375
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 0.82,
|
| 618 |
+
"learning_rate": 9.356610169491525e-06,
|
| 619 |
+
"loss": 0.1652,
|
| 620 |
+
"step": 2400
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.83,
|
| 624 |
+
"learning_rate": 9.34813559322034e-06,
|
| 625 |
+
"loss": 0.1752,
|
| 626 |
+
"step": 2425
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 0.84,
|
| 630 |
+
"learning_rate": 9.339661016949153e-06,
|
| 631 |
+
"loss": 0.1864,
|
| 632 |
+
"step": 2450
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 0.85,
|
| 636 |
+
"learning_rate": 9.331186440677967e-06,
|
| 637 |
+
"loss": 0.17,
|
| 638 |
+
"step": 2475
|
| 639 |
+
},
|
| 640 |
+
{
|
| 641 |
+
"epoch": 0.86,
|
| 642 |
+
"learning_rate": 9.32271186440678e-06,
|
| 643 |
+
"loss": 0.1705,
|
| 644 |
+
"step": 2500
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 0.86,
|
| 648 |
+
"eval_loss": 0.13045215606689453,
|
| 649 |
+
"eval_runtime": 14199.68,
|
| 650 |
+
"eval_samples_per_second": 1.575,
|
| 651 |
+
"eval_steps_per_second": 0.066,
|
| 652 |
+
"eval_wer": 4.807840646127319,
|
| 653 |
+
"step": 2500
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 0.87,
|
| 657 |
+
"learning_rate": 9.314237288135593e-06,
|
| 658 |
+
"loss": 0.1801,
|
| 659 |
+
"step": 2525
|
| 660 |
+
},
|
| 661 |
+
{
|
| 662 |
+
"epoch": 0.88,
|
| 663 |
+
"learning_rate": 9.305762711864408e-06,
|
| 664 |
+
"loss": 0.1708,
|
| 665 |
+
"step": 2550
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 0.88,
|
| 669 |
+
"learning_rate": 9.297288135593221e-06,
|
| 670 |
+
"loss": 0.1831,
|
| 671 |
+
"step": 2575
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 0.89,
|
| 675 |
+
"learning_rate": 9.288813559322036e-06,
|
| 676 |
+
"loss": 0.1803,
|
| 677 |
+
"step": 2600
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 0.9,
|
| 681 |
+
"learning_rate": 9.280338983050849e-06,
|
| 682 |
+
"loss": 0.1787,
|
| 683 |
+
"step": 2625
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 0.91,
|
| 687 |
+
"learning_rate": 9.271864406779662e-06,
|
| 688 |
+
"loss": 0.1561,
|
| 689 |
+
"step": 2650
|
| 690 |
+
},
|
| 691 |
+
{
|
| 692 |
+
"epoch": 0.92,
|
| 693 |
+
"learning_rate": 9.263389830508476e-06,
|
| 694 |
+
"loss": 0.1632,
|
| 695 |
+
"step": 2675
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 0.93,
|
| 699 |
+
"learning_rate": 9.25491525423729e-06,
|
| 700 |
+
"loss": 0.1867,
|
| 701 |
+
"step": 2700
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 0.94,
|
| 705 |
+
"learning_rate": 9.246440677966102e-06,
|
| 706 |
+
"loss": 0.1767,
|
| 707 |
+
"step": 2725
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 0.94,
|
| 711 |
+
"learning_rate": 9.237966101694917e-06,
|
| 712 |
+
"loss": 0.1662,
|
| 713 |
+
"step": 2750
|
| 714 |
+
},
|
| 715 |
+
{
|
| 716 |
+
"epoch": 0.95,
|
| 717 |
+
"learning_rate": 9.22949152542373e-06,
|
| 718 |
+
"loss": 0.179,
|
| 719 |
+
"step": 2775
|
| 720 |
+
},
|
| 721 |
+
{
|
| 722 |
+
"epoch": 0.96,
|
| 723 |
+
"learning_rate": 9.221016949152543e-06,
|
| 724 |
+
"loss": 0.1696,
|
| 725 |
+
"step": 2800
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 0.97,
|
| 729 |
+
"learning_rate": 9.212542372881358e-06,
|
| 730 |
+
"loss": 0.1697,
|
| 731 |
+
"step": 2825
|
| 732 |
+
},
|
| 733 |
+
{
|
| 734 |
+
"epoch": 0.98,
|
| 735 |
+
"learning_rate": 9.20406779661017e-06,
|
| 736 |
+
"loss": 0.1776,
|
| 737 |
+
"step": 2850
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 0.99,
|
| 741 |
+
"learning_rate": 9.195593220338984e-06,
|
| 742 |
+
"loss": 0.1624,
|
| 743 |
+
"step": 2875
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.0,
|
| 747 |
+
"learning_rate": 9.187118644067796e-06,
|
| 748 |
+
"loss": 0.1629,
|
| 749 |
+
"step": 2900
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.0,
|
| 753 |
+
"learning_rate": 9.178644067796611e-06,
|
| 754 |
+
"loss": 0.1614,
|
| 755 |
+
"step": 2925
|
| 756 |
+
},
|
| 757 |
+
{
|
| 758 |
+
"epoch": 1.01,
|
| 759 |
+
"learning_rate": 9.170169491525424e-06,
|
| 760 |
+
"loss": 0.1087,
|
| 761 |
+
"step": 2950
|
| 762 |
+
},
|
| 763 |
+
{
|
| 764 |
+
"epoch": 1.02,
|
| 765 |
+
"learning_rate": 9.161694915254237e-06,
|
| 766 |
+
"loss": 0.1122,
|
| 767 |
+
"step": 2975
|
| 768 |
+
},
|
| 769 |
+
{
|
| 770 |
+
"epoch": 1.03,
|
| 771 |
+
"learning_rate": 9.153220338983052e-06,
|
| 772 |
+
"loss": 0.1069,
|
| 773 |
+
"step": 3000
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.03,
|
| 777 |
+
"eval_loss": 0.1263168603181839,
|
| 778 |
+
"eval_runtime": 13889.4582,
|
| 779 |
+
"eval_samples_per_second": 1.61,
|
| 780 |
+
"eval_steps_per_second": 0.067,
|
| 781 |
+
"eval_wer": 4.3962974726620985,
|
| 782 |
+
"step": 3000
|
| 783 |
+
},
|
| 784 |
+
{
|
| 785 |
+
"epoch": 1.04,
|
| 786 |
+
"learning_rate": 9.144745762711865e-06,
|
| 787 |
+
"loss": 0.1051,
|
| 788 |
+
"step": 3025
|
| 789 |
+
},
|
| 790 |
+
{
|
| 791 |
+
"epoch": 1.05,
|
| 792 |
+
"learning_rate": 9.136271186440678e-06,
|
| 793 |
+
"loss": 0.103,
|
| 794 |
+
"step": 3050
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.06,
|
| 798 |
+
"learning_rate": 9.127796610169492e-06,
|
| 799 |
+
"loss": 0.1039,
|
| 800 |
+
"step": 3075
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 1.06,
|
| 804 |
+
"learning_rate": 9.119322033898305e-06,
|
| 805 |
+
"loss": 0.1113,
|
| 806 |
+
"step": 3100
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 1.07,
|
| 810 |
+
"learning_rate": 9.110847457627118e-06,
|
| 811 |
+
"loss": 0.1092,
|
| 812 |
+
"step": 3125
|
| 813 |
+
},
|
| 814 |
+
{
|
| 815 |
+
"epoch": 1.08,
|
| 816 |
+
"learning_rate": 9.102372881355933e-06,
|
| 817 |
+
"loss": 0.1272,
|
| 818 |
+
"step": 3150
|
| 819 |
+
},
|
| 820 |
+
{
|
| 821 |
+
"epoch": 1.09,
|
| 822 |
+
"learning_rate": 9.093898305084746e-06,
|
| 823 |
+
"loss": 0.1079,
|
| 824 |
+
"step": 3175
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.1,
|
| 828 |
+
"learning_rate": 9.08542372881356e-06,
|
| 829 |
+
"loss": 0.1137,
|
| 830 |
+
"step": 3200
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.11,
|
| 834 |
+
"learning_rate": 9.076949152542374e-06,
|
| 835 |
+
"loss": 0.1097,
|
| 836 |
+
"step": 3225
|
| 837 |
+
},
|
| 838 |
+
{
|
| 839 |
+
"epoch": 1.12,
|
| 840 |
+
"learning_rate": 9.068474576271187e-06,
|
| 841 |
+
"loss": 0.1127,
|
| 842 |
+
"step": 3250
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 1.12,
|
| 846 |
+
"learning_rate": 9.060000000000001e-06,
|
| 847 |
+
"loss": 0.1139,
|
| 848 |
+
"step": 3275
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 1.13,
|
| 852 |
+
"learning_rate": 9.051525423728814e-06,
|
| 853 |
+
"loss": 0.1059,
|
| 854 |
+
"step": 3300
|
| 855 |
+
},
|
| 856 |
+
{
|
| 857 |
+
"epoch": 1.14,
|
| 858 |
+
"learning_rate": 9.043050847457629e-06,
|
| 859 |
+
"loss": 0.1103,
|
| 860 |
+
"step": 3325
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.15,
|
| 864 |
+
"learning_rate": 9.034576271186442e-06,
|
| 865 |
+
"loss": 0.1238,
|
| 866 |
+
"step": 3350
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.16,
|
| 870 |
+
"learning_rate": 9.026101694915255e-06,
|
| 871 |
+
"loss": 0.1008,
|
| 872 |
+
"step": 3375
|
| 873 |
+
},
|
| 874 |
+
{
|
| 875 |
+
"epoch": 1.17,
|
| 876 |
+
"learning_rate": 9.01762711864407e-06,
|
| 877 |
+
"loss": 0.1077,
|
| 878 |
+
"step": 3400
|
| 879 |
+
},
|
| 880 |
+
{
|
| 881 |
+
"epoch": 1.18,
|
| 882 |
+
"learning_rate": 9.009152542372883e-06,
|
| 883 |
+
"loss": 0.1086,
|
| 884 |
+
"step": 3425
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 1.18,
|
| 888 |
+
"learning_rate": 9.000677966101696e-06,
|
| 889 |
+
"loss": 0.1126,
|
| 890 |
+
"step": 3450
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.19,
|
| 894 |
+
"learning_rate": 8.99220338983051e-06,
|
| 895 |
+
"loss": 0.1151,
|
| 896 |
+
"step": 3475
|
| 897 |
+
},
|
| 898 |
+
{
|
| 899 |
+
"epoch": 1.2,
|
| 900 |
+
"learning_rate": 8.983728813559323e-06,
|
| 901 |
+
"loss": 0.1098,
|
| 902 |
+
"step": 3500
|
| 903 |
+
},
|
| 904 |
+
{
|
| 905 |
+
"epoch": 1.2,
|
| 906 |
+
"eval_loss": 0.12618736922740936,
|
| 907 |
+
"eval_runtime": 14301.962,
|
| 908 |
+
"eval_samples_per_second": 1.564,
|
| 909 |
+
"eval_steps_per_second": 0.065,
|
| 910 |
+
"eval_wer": 4.478878352012342,
|
| 911 |
+
"step": 3500
|
| 912 |
+
},
|
| 913 |
+
{
|
| 914 |
+
"epoch": 1.21,
|
| 915 |
+
"learning_rate": 8.975254237288136e-06,
|
| 916 |
+
"loss": 0.1181,
|
| 917 |
+
"step": 3525
|
| 918 |
+
},
|
| 919 |
+
{
|
| 920 |
+
"epoch": 1.22,
|
| 921 |
+
"learning_rate": 8.96677966101695e-06,
|
| 922 |
+
"loss": 0.1102,
|
| 923 |
+
"step": 3550
|
| 924 |
+
},
|
| 925 |
+
{
|
| 926 |
+
"epoch": 1.23,
|
| 927 |
+
"learning_rate": 8.958305084745764e-06,
|
| 928 |
+
"loss": 0.1046,
|
| 929 |
+
"step": 3575
|
| 930 |
+
},
|
| 931 |
+
{
|
| 932 |
+
"epoch": 1.24,
|
| 933 |
+
"learning_rate": 8.949830508474577e-06,
|
| 934 |
+
"loss": 0.1146,
|
| 935 |
+
"step": 3600
|
| 936 |
+
},
|
| 937 |
+
{
|
| 938 |
+
"epoch": 1.24,
|
| 939 |
+
"learning_rate": 8.94135593220339e-06,
|
| 940 |
+
"loss": 0.1192,
|
| 941 |
+
"step": 3625
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.25,
|
| 945 |
+
"learning_rate": 8.932881355932203e-06,
|
| 946 |
+
"loss": 0.1087,
|
| 947 |
+
"step": 3650
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 1.26,
|
| 951 |
+
"learning_rate": 8.924406779661017e-06,
|
| 952 |
+
"loss": 0.1143,
|
| 953 |
+
"step": 3675
|
| 954 |
+
},
|
| 955 |
+
{
|
| 956 |
+
"epoch": 1.27,
|
| 957 |
+
"learning_rate": 8.91593220338983e-06,
|
| 958 |
+
"loss": 0.1039,
|
| 959 |
+
"step": 3700
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 1.28,
|
| 963 |
+
"learning_rate": 8.907457627118645e-06,
|
| 964 |
+
"loss": 0.1194,
|
| 965 |
+
"step": 3725
|
| 966 |
+
},
|
| 967 |
+
{
|
| 968 |
+
"epoch": 1.29,
|
| 969 |
+
"learning_rate": 8.898983050847458e-06,
|
| 970 |
+
"loss": 0.1128,
|
| 971 |
+
"step": 3750
|
| 972 |
+
},
|
| 973 |
+
{
|
| 974 |
+
"epoch": 1.3,
|
| 975 |
+
"learning_rate": 8.890508474576271e-06,
|
| 976 |
+
"loss": 0.108,
|
| 977 |
+
"step": 3775
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 1.3,
|
| 981 |
+
"learning_rate": 8.882033898305086e-06,
|
| 982 |
+
"loss": 0.1049,
|
| 983 |
+
"step": 3800
|
| 984 |
+
},
|
| 985 |
+
{
|
| 986 |
+
"epoch": 1.31,
|
| 987 |
+
"learning_rate": 8.873559322033899e-06,
|
| 988 |
+
"loss": 0.1253,
|
| 989 |
+
"step": 3825
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 1.32,
|
| 993 |
+
"learning_rate": 8.865084745762712e-06,
|
| 994 |
+
"loss": 0.1165,
|
| 995 |
+
"step": 3850
|
| 996 |
+
},
|
| 997 |
+
{
|
| 998 |
+
"epoch": 1.33,
|
| 999 |
+
"learning_rate": 8.856610169491526e-06,
|
| 1000 |
+
"loss": 0.1189,
|
| 1001 |
+
"step": 3875
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 1.34,
|
| 1005 |
+
"learning_rate": 8.84813559322034e-06,
|
| 1006 |
+
"loss": 0.1115,
|
| 1007 |
+
"step": 3900
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.35,
|
| 1011 |
+
"learning_rate": 8.839661016949154e-06,
|
| 1012 |
+
"loss": 0.1214,
|
| 1013 |
+
"step": 3925
|
| 1014 |
+
},
|
| 1015 |
+
{
|
| 1016 |
+
"epoch": 1.36,
|
| 1017 |
+
"learning_rate": 8.831186440677967e-06,
|
| 1018 |
+
"loss": 0.101,
|
| 1019 |
+
"step": 3950
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 1.36,
|
| 1023 |
+
"learning_rate": 8.82271186440678e-06,
|
| 1024 |
+
"loss": 0.1263,
|
| 1025 |
+
"step": 3975
|
| 1026 |
+
},
|
| 1027 |
+
{
|
| 1028 |
+
"epoch": 1.37,
|
| 1029 |
+
"learning_rate": 8.814237288135595e-06,
|
| 1030 |
+
"loss": 0.1059,
|
| 1031 |
+
"step": 4000
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 1.37,
|
| 1035 |
+
"eval_loss": 0.12528666853904724,
|
| 1036 |
+
"eval_runtime": 14038.6098,
|
| 1037 |
+
"eval_samples_per_second": 1.593,
|
| 1038 |
+
"eval_steps_per_second": 0.066,
|
| 1039 |
+
"eval_wer": 4.441217841099868,
|
| 1040 |
+
"step": 4000
|
| 1041 |
+
},
|
| 1042 |
+
{
|
| 1043 |
+
"epoch": 1.38,
|
| 1044 |
+
"learning_rate": 8.805762711864408e-06,
|
| 1045 |
+
"loss": 0.1241,
|
| 1046 |
+
"step": 4025
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 1.39,
|
| 1050 |
+
"learning_rate": 8.797288135593222e-06,
|
| 1051 |
+
"loss": 0.1159,
|
| 1052 |
+
"step": 4050
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 1.4,
|
| 1056 |
+
"learning_rate": 8.788813559322035e-06,
|
| 1057 |
+
"loss": 0.1161,
|
| 1058 |
+
"step": 4075
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.41,
|
| 1062 |
+
"learning_rate": 8.780338983050848e-06,
|
| 1063 |
+
"loss": 0.1064,
|
| 1064 |
+
"step": 4100
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 1.42,
|
| 1068 |
+
"learning_rate": 8.771864406779661e-06,
|
| 1069 |
+
"loss": 0.1106,
|
| 1070 |
+
"step": 4125
|
| 1071 |
+
},
|
| 1072 |
+
{
|
| 1073 |
+
"epoch": 1.42,
|
| 1074 |
+
"learning_rate": 8.763389830508476e-06,
|
| 1075 |
+
"loss": 0.1048,
|
| 1076 |
+
"step": 4150
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 1.43,
|
| 1080 |
+
"learning_rate": 8.754915254237289e-06,
|
| 1081 |
+
"loss": 0.1122,
|
| 1082 |
+
"step": 4175
|
| 1083 |
+
},
|
| 1084 |
+
{
|
| 1085 |
+
"epoch": 1.44,
|
| 1086 |
+
"learning_rate": 8.746440677966102e-06,
|
| 1087 |
+
"loss": 0.1262,
|
| 1088 |
+
"step": 4200
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 1.45,
|
| 1092 |
+
"learning_rate": 8.737966101694917e-06,
|
| 1093 |
+
"loss": 0.1277,
|
| 1094 |
+
"step": 4225
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 1.46,
|
| 1098 |
+
"learning_rate": 8.72949152542373e-06,
|
| 1099 |
+
"loss": 0.112,
|
| 1100 |
+
"step": 4250
|
| 1101 |
+
},
|
| 1102 |
+
{
|
| 1103 |
+
"epoch": 1.47,
|
| 1104 |
+
"learning_rate": 8.721016949152543e-06,
|
| 1105 |
+
"loss": 0.1011,
|
| 1106 |
+
"step": 4275
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 1.48,
|
| 1110 |
+
"learning_rate": 8.712542372881356e-06,
|
| 1111 |
+
"loss": 0.106,
|
| 1112 |
+
"step": 4300
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 1.48,
|
| 1116 |
+
"learning_rate": 8.70406779661017e-06,
|
| 1117 |
+
"loss": 0.1206,
|
| 1118 |
+
"step": 4325
|
| 1119 |
+
},
|
| 1120 |
+
{
|
| 1121 |
+
"epoch": 1.49,
|
| 1122 |
+
"learning_rate": 8.695593220338983e-06,
|
| 1123 |
+
"loss": 0.1097,
|
| 1124 |
+
"step": 4350
|
| 1125 |
+
},
|
| 1126 |
+
{
|
| 1127 |
+
"epoch": 1.5,
|
| 1128 |
+
"learning_rate": 8.687118644067796e-06,
|
| 1129 |
+
"loss": 0.1212,
|
| 1130 |
+
"step": 4375
|
| 1131 |
+
},
|
| 1132 |
+
{
|
| 1133 |
+
"epoch": 1.51,
|
| 1134 |
+
"learning_rate": 8.678644067796611e-06,
|
| 1135 |
+
"loss": 0.1068,
|
| 1136 |
+
"step": 4400
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 1.52,
|
| 1140 |
+
"learning_rate": 8.670169491525424e-06,
|
| 1141 |
+
"loss": 0.1096,
|
| 1142 |
+
"step": 4425
|
| 1143 |
+
},
|
| 1144 |
+
{
|
| 1145 |
+
"epoch": 1.53,
|
| 1146 |
+
"learning_rate": 8.661694915254239e-06,
|
| 1147 |
+
"loss": 0.1062,
|
| 1148 |
+
"step": 4450
|
| 1149 |
+
},
|
| 1150 |
+
{
|
| 1151 |
+
"epoch": 1.54,
|
| 1152 |
+
"learning_rate": 8.653220338983051e-06,
|
| 1153 |
+
"loss": 0.1,
|
| 1154 |
+
"step": 4475
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 1.54,
|
| 1158 |
+
"learning_rate": 8.644745762711864e-06,
|
| 1159 |
+
"loss": 0.1088,
|
| 1160 |
+
"step": 4500
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 1.54,
|
| 1164 |
+
"eval_loss": 0.12188256531953812,
|
| 1165 |
+
"eval_runtime": 13978.7976,
|
| 1166 |
+
"eval_samples_per_second": 1.6,
|
| 1167 |
+
"eval_steps_per_second": 0.067,
|
| 1168 |
+
"eval_wer": 4.305549253595898,
|
| 1169 |
+
"step": 4500
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 1.55,
|
| 1173 |
+
"learning_rate": 8.636271186440679e-06,
|
| 1174 |
+
"loss": 0.122,
|
| 1175 |
+
"step": 4525
|
| 1176 |
+
},
|
| 1177 |
+
{
|
| 1178 |
+
"epoch": 1.56,
|
| 1179 |
+
"learning_rate": 8.627796610169492e-06,
|
| 1180 |
+
"loss": 0.1094,
|
| 1181 |
+
"step": 4550
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 1.57,
|
| 1185 |
+
"learning_rate": 8.619322033898305e-06,
|
| 1186 |
+
"loss": 0.1163,
|
| 1187 |
+
"step": 4575
|
| 1188 |
+
},
|
| 1189 |
+
{
|
| 1190 |
+
"epoch": 1.58,
|
| 1191 |
+
"learning_rate": 8.61084745762712e-06,
|
| 1192 |
+
"loss": 0.127,
|
| 1193 |
+
"step": 4600
|
| 1194 |
+
},
|
| 1195 |
+
{
|
| 1196 |
+
"epoch": 1.59,
|
| 1197 |
+
"learning_rate": 8.602372881355933e-06,
|
| 1198 |
+
"loss": 0.1122,
|
| 1199 |
+
"step": 4625
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 1.6,
|
| 1203 |
+
"learning_rate": 8.593898305084747e-06,
|
| 1204 |
+
"loss": 0.1056,
|
| 1205 |
+
"step": 4650
|
| 1206 |
+
},
|
| 1207 |
+
{
|
| 1208 |
+
"epoch": 1.6,
|
| 1209 |
+
"learning_rate": 8.58542372881356e-06,
|
| 1210 |
+
"loss": 0.1051,
|
| 1211 |
+
"step": 4675
|
| 1212 |
+
},
|
| 1213 |
+
{
|
| 1214 |
+
"epoch": 1.61,
|
| 1215 |
+
"learning_rate": 8.576949152542373e-06,
|
| 1216 |
+
"loss": 0.1087,
|
| 1217 |
+
"step": 4700
|
| 1218 |
+
},
|
| 1219 |
+
{
|
| 1220 |
+
"epoch": 1.62,
|
| 1221 |
+
"learning_rate": 8.568474576271188e-06,
|
| 1222 |
+
"loss": 0.1155,
|
| 1223 |
+
"step": 4725
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 1.63,
|
| 1227 |
+
"learning_rate": 8.560000000000001e-06,
|
| 1228 |
+
"loss": 0.114,
|
| 1229 |
+
"step": 4750
|
| 1230 |
+
},
|
| 1231 |
+
{
|
| 1232 |
+
"epoch": 1.64,
|
| 1233 |
+
"learning_rate": 8.551525423728814e-06,
|
| 1234 |
+
"loss": 0.1286,
|
| 1235 |
+
"step": 4775
|
| 1236 |
+
},
|
| 1237 |
+
{
|
| 1238 |
+
"epoch": 1.65,
|
| 1239 |
+
"learning_rate": 8.543050847457629e-06,
|
| 1240 |
+
"loss": 0.1052,
|
| 1241 |
+
"step": 4800
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 1.66,
|
| 1245 |
+
"learning_rate": 8.534576271186442e-06,
|
| 1246 |
+
"loss": 0.1156,
|
| 1247 |
+
"step": 4825
|
| 1248 |
+
},
|
| 1249 |
+
{
|
| 1250 |
+
"epoch": 1.66,
|
| 1251 |
+
"learning_rate": 8.526101694915255e-06,
|
| 1252 |
+
"loss": 0.0998,
|
| 1253 |
+
"step": 4850
|
| 1254 |
+
},
|
| 1255 |
+
{
|
| 1256 |
+
"epoch": 1.67,
|
| 1257 |
+
"learning_rate": 8.517627118644068e-06,
|
| 1258 |
+
"loss": 0.1147,
|
| 1259 |
+
"step": 4875
|
| 1260 |
+
},
|
| 1261 |
+
{
|
| 1262 |
+
"epoch": 1.68,
|
| 1263 |
+
"learning_rate": 8.509152542372882e-06,
|
| 1264 |
+
"loss": 0.1007,
|
| 1265 |
+
"step": 4900
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 1.69,
|
| 1269 |
+
"learning_rate": 8.500677966101695e-06,
|
| 1270 |
+
"loss": 0.1102,
|
| 1271 |
+
"step": 4925
|
| 1272 |
+
},
|
| 1273 |
+
{
|
| 1274 |
+
"epoch": 1.7,
|
| 1275 |
+
"learning_rate": 8.492203389830508e-06,
|
| 1276 |
+
"loss": 0.1113,
|
| 1277 |
+
"step": 4950
|
| 1278 |
+
},
|
| 1279 |
+
{
|
| 1280 |
+
"epoch": 1.71,
|
| 1281 |
+
"learning_rate": 8.483728813559323e-06,
|
| 1282 |
+
"loss": 0.1118,
|
| 1283 |
+
"step": 4975
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 1.72,
|
| 1287 |
+
"learning_rate": 8.475254237288136e-06,
|
| 1288 |
+
"loss": 0.1074,
|
| 1289 |
+
"step": 5000
|
| 1290 |
+
},
|
| 1291 |
+
{
|
| 1292 |
+
"epoch": 1.72,
|
| 1293 |
+
"eval_loss": 0.12055900692939758,
|
| 1294 |
+
"eval_runtime": 13950.039,
|
| 1295 |
+
"eval_samples_per_second": 1.603,
|
| 1296 |
+
"eval_steps_per_second": 0.067,
|
| 1297 |
+
"eval_wer": 4.194382685239803,
|
| 1298 |
+
"step": 5000
|
| 1299 |
+
},
|
| 1300 |
+
{
|
| 1301 |
+
"epoch": 1.72,
|
| 1302 |
+
"learning_rate": 8.466779661016949e-06,
|
| 1303 |
+
"loss": 0.1024,
|
| 1304 |
+
"step": 5025
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 1.73,
|
| 1308 |
+
"learning_rate": 8.458305084745764e-06,
|
| 1309 |
+
"loss": 0.1078,
|
| 1310 |
+
"step": 5050
|
| 1311 |
+
},
|
| 1312 |
+
{
|
| 1313 |
+
"epoch": 1.74,
|
| 1314 |
+
"learning_rate": 8.449830508474577e-06,
|
| 1315 |
+
"loss": 0.1198,
|
| 1316 |
+
"step": 5075
|
| 1317 |
+
},
|
| 1318 |
+
{
|
| 1319 |
+
"epoch": 1.75,
|
| 1320 |
+
"learning_rate": 8.44135593220339e-06,
|
| 1321 |
+
"loss": 0.1052,
|
| 1322 |
+
"step": 5100
|
| 1323 |
+
},
|
| 1324 |
+
{
|
| 1325 |
+
"epoch": 1.76,
|
| 1326 |
+
"learning_rate": 8.432881355932204e-06,
|
| 1327 |
+
"loss": 0.1104,
|
| 1328 |
+
"step": 5125
|
| 1329 |
+
},
|
| 1330 |
+
{
|
| 1331 |
+
"epoch": 1.77,
|
| 1332 |
+
"learning_rate": 8.424406779661017e-06,
|
| 1333 |
+
"loss": 0.111,
|
| 1334 |
+
"step": 5150
|
| 1335 |
+
},
|
| 1336 |
+
{
|
| 1337 |
+
"epoch": 1.78,
|
| 1338 |
+
"learning_rate": 8.415932203389832e-06,
|
| 1339 |
+
"loss": 0.114,
|
| 1340 |
+
"step": 5175
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 1.78,
|
| 1344 |
+
"learning_rate": 8.407457627118645e-06,
|
| 1345 |
+
"loss": 0.1074,
|
| 1346 |
+
"step": 5200
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 1.79,
|
| 1350 |
+
"learning_rate": 8.398983050847458e-06,
|
| 1351 |
+
"loss": 0.1117,
|
| 1352 |
+
"step": 5225
|
| 1353 |
+
},
|
| 1354 |
+
{
|
| 1355 |
+
"epoch": 1.8,
|
| 1356 |
+
"learning_rate": 8.390508474576273e-06,
|
| 1357 |
+
"loss": 0.1059,
|
| 1358 |
+
"step": 5250
|
| 1359 |
+
},
|
| 1360 |
+
{
|
| 1361 |
+
"epoch": 1.81,
|
| 1362 |
+
"learning_rate": 8.382033898305085e-06,
|
| 1363 |
+
"loss": 0.126,
|
| 1364 |
+
"step": 5275
|
| 1365 |
+
},
|
| 1366 |
+
{
|
| 1367 |
+
"epoch": 1.82,
|
| 1368 |
+
"learning_rate": 8.3735593220339e-06,
|
| 1369 |
+
"loss": 0.1037,
|
| 1370 |
+
"step": 5300
|
| 1371 |
+
},
|
| 1372 |
+
{
|
| 1373 |
+
"epoch": 1.83,
|
| 1374 |
+
"learning_rate": 8.365084745762713e-06,
|
| 1375 |
+
"loss": 0.0988,
|
| 1376 |
+
"step": 5325
|
| 1377 |
+
},
|
| 1378 |
+
{
|
| 1379 |
+
"epoch": 1.84,
|
| 1380 |
+
"learning_rate": 8.356610169491526e-06,
|
| 1381 |
+
"loss": 0.1053,
|
| 1382 |
+
"step": 5350
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 1.84,
|
| 1386 |
+
"learning_rate": 8.34813559322034e-06,
|
| 1387 |
+
"loss": 0.1128,
|
| 1388 |
+
"step": 5375
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 1.85,
|
| 1392 |
+
"learning_rate": 8.339661016949154e-06,
|
| 1393 |
+
"loss": 0.1132,
|
| 1394 |
+
"step": 5400
|
| 1395 |
+
},
|
| 1396 |
+
{
|
| 1397 |
+
"epoch": 1.86,
|
| 1398 |
+
"learning_rate": 8.331186440677967e-06,
|
| 1399 |
+
"loss": 0.1083,
|
| 1400 |
+
"step": 5425
|
| 1401 |
+
},
|
| 1402 |
+
{
|
| 1403 |
+
"epoch": 1.87,
|
| 1404 |
+
"learning_rate": 8.32271186440678e-06,
|
| 1405 |
+
"loss": 0.128,
|
| 1406 |
+
"step": 5450
|
| 1407 |
+
},
|
| 1408 |
+
{
|
| 1409 |
+
"epoch": 1.88,
|
| 1410 |
+
"learning_rate": 8.314237288135594e-06,
|
| 1411 |
+
"loss": 0.1044,
|
| 1412 |
+
"step": 5475
|
| 1413 |
+
},
|
| 1414 |
+
{
|
| 1415 |
+
"epoch": 1.89,
|
| 1416 |
+
"learning_rate": 8.305762711864407e-06,
|
| 1417 |
+
"loss": 0.1037,
|
| 1418 |
+
"step": 5500
|
| 1419 |
+
},
|
| 1420 |
+
{
|
| 1421 |
+
"epoch": 1.89,
|
| 1422 |
+
"eval_loss": 0.1189337745308876,
|
| 1423 |
+
"eval_runtime": 14041.5407,
|
| 1424 |
+
"eval_samples_per_second": 1.593,
|
| 1425 |
+
"eval_steps_per_second": 0.066,
|
| 1426 |
+
"eval_wer": 4.197558872907119,
|
| 1427 |
+
"step": 5500
|
| 1428 |
+
},
|
| 1429 |
+
{
|
| 1430 |
+
"epoch": 1.9,
|
| 1431 |
+
"learning_rate": 8.29728813559322e-06,
|
| 1432 |
+
"loss": 0.1185,
|
| 1433 |
+
"step": 5525
|
| 1434 |
+
},
|
| 1435 |
+
{
|
| 1436 |
+
"epoch": 1.9,
|
| 1437 |
+
"learning_rate": 8.288813559322035e-06,
|
| 1438 |
+
"loss": 0.1117,
|
| 1439 |
+
"step": 5550
|
| 1440 |
+
},
|
| 1441 |
+
{
|
| 1442 |
+
"epoch": 1.91,
|
| 1443 |
+
"learning_rate": 8.280338983050848e-06,
|
| 1444 |
+
"loss": 0.1185,
|
| 1445 |
+
"step": 5575
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 1.92,
|
| 1449 |
+
"learning_rate": 8.271864406779661e-06,
|
| 1450 |
+
"loss": 0.1155,
|
| 1451 |
+
"step": 5600
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 1.93,
|
| 1455 |
+
"learning_rate": 8.263389830508474e-06,
|
| 1456 |
+
"loss": 0.1203,
|
| 1457 |
+
"step": 5625
|
| 1458 |
+
},
|
| 1459 |
+
{
|
| 1460 |
+
"epoch": 1.94,
|
| 1461 |
+
"learning_rate": 8.254915254237289e-06,
|
| 1462 |
+
"loss": 0.1065,
|
| 1463 |
+
"step": 5650
|
| 1464 |
+
},
|
| 1465 |
+
{
|
| 1466 |
+
"epoch": 1.95,
|
| 1467 |
+
"learning_rate": 8.246440677966102e-06,
|
| 1468 |
+
"loss": 0.1063,
|
| 1469 |
+
"step": 5675
|
| 1470 |
+
},
|
| 1471 |
+
{
|
| 1472 |
+
"epoch": 1.96,
|
| 1473 |
+
"learning_rate": 8.237966101694916e-06,
|
| 1474 |
+
"loss": 0.113,
|
| 1475 |
+
"step": 5700
|
| 1476 |
+
},
|
| 1477 |
+
{
|
| 1478 |
+
"epoch": 1.96,
|
| 1479 |
+
"learning_rate": 8.22949152542373e-06,
|
| 1480 |
+
"loss": 0.1073,
|
| 1481 |
+
"step": 5725
|
| 1482 |
+
},
|
| 1483 |
+
{
|
| 1484 |
+
"epoch": 1.97,
|
| 1485 |
+
"learning_rate": 8.221016949152542e-06,
|
| 1486 |
+
"loss": 0.1045,
|
| 1487 |
+
"step": 5750
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 1.98,
|
| 1491 |
+
"learning_rate": 8.212542372881357e-06,
|
| 1492 |
+
"loss": 0.1053,
|
| 1493 |
+
"step": 5775
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 1.99,
|
| 1497 |
+
"learning_rate": 8.20406779661017e-06,
|
| 1498 |
+
"loss": 0.0961,
|
| 1499 |
+
"step": 5800
|
| 1500 |
+
},
|
| 1501 |
+
{
|
| 1502 |
+
"epoch": 2.0,
|
| 1503 |
+
"learning_rate": 8.195593220338983e-06,
|
| 1504 |
+
"loss": 0.1045,
|
| 1505 |
+
"step": 5825
|
| 1506 |
+
},
|
| 1507 |
+
{
|
| 1508 |
+
"epoch": 2.01,
|
| 1509 |
+
"learning_rate": 8.187118644067798e-06,
|
| 1510 |
+
"loss": 0.0678,
|
| 1511 |
+
"step": 5850
|
| 1512 |
+
},
|
| 1513 |
+
{
|
| 1514 |
+
"epoch": 2.02,
|
| 1515 |
+
"learning_rate": 8.17864406779661e-06,
|
| 1516 |
+
"loss": 0.0583,
|
| 1517 |
+
"step": 5875
|
| 1518 |
+
},
|
| 1519 |
+
{
|
| 1520 |
+
"epoch": 2.02,
|
| 1521 |
+
"learning_rate": 8.170169491525425e-06,
|
| 1522 |
+
"loss": 0.0615,
|
| 1523 |
+
"step": 5900
|
| 1524 |
+
},
|
| 1525 |
+
{
|
| 1526 |
+
"epoch": 2.03,
|
| 1527 |
+
"learning_rate": 8.161694915254238e-06,
|
| 1528 |
+
"loss": 0.0624,
|
| 1529 |
+
"step": 5925
|
| 1530 |
+
},
|
| 1531 |
+
{
|
| 1532 |
+
"epoch": 2.04,
|
| 1533 |
+
"learning_rate": 8.153220338983051e-06,
|
| 1534 |
+
"loss": 0.0561,
|
| 1535 |
+
"step": 5950
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 2.05,
|
| 1539 |
+
"learning_rate": 8.144745762711866e-06,
|
| 1540 |
+
"loss": 0.0633,
|
| 1541 |
+
"step": 5975
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 2.06,
|
| 1545 |
+
"learning_rate": 8.136271186440679e-06,
|
| 1546 |
+
"loss": 0.0575,
|
| 1547 |
+
"step": 6000
|
| 1548 |
+
},
|
| 1549 |
+
{
|
| 1550 |
+
"epoch": 2.06,
|
| 1551 |
+
"eval_loss": 0.12541243433952332,
|
| 1552 |
+
"eval_runtime": 13932.2659,
|
| 1553 |
+
"eval_samples_per_second": 1.605,
|
| 1554 |
+
"eval_steps_per_second": 0.067,
|
| 1555 |
+
"eval_wer": 4.1412949770860745,
|
| 1556 |
+
"step": 6000
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 2.07,
|
| 1560 |
+
"learning_rate": 8.127796610169494e-06,
|
| 1561 |
+
"loss": 0.0622,
|
| 1562 |
+
"step": 6025
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 2.08,
|
| 1566 |
+
"learning_rate": 8.119322033898307e-06,
|
| 1567 |
+
"loss": 0.0632,
|
| 1568 |
+
"step": 6050
|
| 1569 |
+
},
|
| 1570 |
+
{
|
| 1571 |
+
"epoch": 2.08,
|
| 1572 |
+
"learning_rate": 8.11084745762712e-06,
|
| 1573 |
+
"loss": 0.0542,
|
| 1574 |
+
"step": 6075
|
| 1575 |
+
},
|
| 1576 |
+
{
|
| 1577 |
+
"epoch": 2.09,
|
| 1578 |
+
"learning_rate": 8.102372881355932e-06,
|
| 1579 |
+
"loss": 0.0644,
|
| 1580 |
+
"step": 6100
|
| 1581 |
+
},
|
| 1582 |
+
{
|
| 1583 |
+
"epoch": 2.1,
|
| 1584 |
+
"learning_rate": 8.093898305084747e-06,
|
| 1585 |
+
"loss": 0.0671,
|
| 1586 |
+
"step": 6125
|
| 1587 |
+
},
|
| 1588 |
+
{
|
| 1589 |
+
"epoch": 2.11,
|
| 1590 |
+
"learning_rate": 8.08542372881356e-06,
|
| 1591 |
+
"loss": 0.0532,
|
| 1592 |
+
"step": 6150
|
| 1593 |
+
},
|
| 1594 |
+
{
|
| 1595 |
+
"epoch": 2.12,
|
| 1596 |
+
"learning_rate": 8.076949152542373e-06,
|
| 1597 |
+
"loss": 0.064,
|
| 1598 |
+
"step": 6175
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 2.13,
|
| 1602 |
+
"learning_rate": 8.068474576271186e-06,
|
| 1603 |
+
"loss": 0.0624,
|
| 1604 |
+
"step": 6200
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 2.14,
|
| 1608 |
+
"learning_rate": 8.06e-06,
|
| 1609 |
+
"loss": 0.0722,
|
| 1610 |
+
"step": 6225
|
| 1611 |
+
},
|
| 1612 |
+
{
|
| 1613 |
+
"epoch": 2.14,
|
| 1614 |
+
"learning_rate": 8.051525423728814e-06,
|
| 1615 |
+
"loss": 0.0674,
|
| 1616 |
+
"step": 6250
|
| 1617 |
+
},
|
| 1618 |
+
{
|
| 1619 |
+
"epoch": 2.15,
|
| 1620 |
+
"learning_rate": 8.043050847457627e-06,
|
| 1621 |
+
"loss": 0.0573,
|
| 1622 |
+
"step": 6275
|
| 1623 |
+
},
|
| 1624 |
+
{
|
| 1625 |
+
"epoch": 2.16,
|
| 1626 |
+
"learning_rate": 8.034576271186441e-06,
|
| 1627 |
+
"loss": 0.0618,
|
| 1628 |
+
"step": 6300
|
| 1629 |
+
},
|
| 1630 |
+
{
|
| 1631 |
+
"epoch": 2.17,
|
| 1632 |
+
"learning_rate": 8.026101694915254e-06,
|
| 1633 |
+
"loss": 0.0647,
|
| 1634 |
+
"step": 6325
|
| 1635 |
+
},
|
| 1636 |
+
{
|
| 1637 |
+
"epoch": 2.18,
|
| 1638 |
+
"learning_rate": 8.017627118644067e-06,
|
| 1639 |
+
"loss": 0.0667,
|
| 1640 |
+
"step": 6350
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 2.19,
|
| 1644 |
+
"learning_rate": 8.009152542372882e-06,
|
| 1645 |
+
"loss": 0.0569,
|
| 1646 |
+
"step": 6375
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 2.2,
|
| 1650 |
+
"learning_rate": 8.000677966101695e-06,
|
| 1651 |
+
"loss": 0.0755,
|
| 1652 |
+
"step": 6400
|
| 1653 |
+
},
|
| 1654 |
+
{
|
| 1655 |
+
"epoch": 2.2,
|
| 1656 |
+
"learning_rate": 7.99220338983051e-06,
|
| 1657 |
+
"loss": 0.0617,
|
| 1658 |
+
"step": 6425
|
| 1659 |
+
},
|
| 1660 |
+
{
|
| 1661 |
+
"epoch": 2.21,
|
| 1662 |
+
"learning_rate": 7.983728813559323e-06,
|
| 1663 |
+
"loss": 0.0592,
|
| 1664 |
+
"step": 6450
|
| 1665 |
+
},
|
| 1666 |
+
{
|
| 1667 |
+
"epoch": 2.22,
|
| 1668 |
+
"learning_rate": 7.975254237288136e-06,
|
| 1669 |
+
"loss": 0.0578,
|
| 1670 |
+
"step": 6475
|
| 1671 |
+
},
|
| 1672 |
+
{
|
| 1673 |
+
"epoch": 2.23,
|
| 1674 |
+
"learning_rate": 7.96677966101695e-06,
|
| 1675 |
+
"loss": 0.0623,
|
| 1676 |
+
"step": 6500
|
| 1677 |
+
},
|
| 1678 |
+
{
|
| 1679 |
+
"epoch": 2.23,
|
| 1680 |
+
"eval_loss": 0.12663590908050537,
|
| 1681 |
+
"eval_runtime": 14047.6478,
|
| 1682 |
+
"eval_samples_per_second": 1.592,
|
| 1683 |
+
"eval_steps_per_second": 0.066,
|
| 1684 |
+
"eval_wer": 4.139933753800082,
|
| 1685 |
+
"step": 6500
|
| 1686 |
+
},
|
| 1687 |
+
{
|
| 1688 |
+
"epoch": 2.24,
|
| 1689 |
+
"learning_rate": 7.958305084745763e-06,
|
| 1690 |
+
"loss": 0.0718,
|
| 1691 |
+
"step": 6525
|
| 1692 |
+
},
|
| 1693 |
+
{
|
| 1694 |
+
"epoch": 2.25,
|
| 1695 |
+
"learning_rate": 7.949830508474576e-06,
|
| 1696 |
+
"loss": 0.0643,
|
| 1697 |
+
"step": 6550
|
| 1698 |
+
},
|
| 1699 |
+
{
|
| 1700 |
+
"epoch": 2.26,
|
| 1701 |
+
"learning_rate": 7.941355932203391e-06,
|
| 1702 |
+
"loss": 0.0648,
|
| 1703 |
+
"step": 6575
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 2.26,
|
| 1707 |
+
"learning_rate": 7.932881355932204e-06,
|
| 1708 |
+
"loss": 0.0632,
|
| 1709 |
+
"step": 6600
|
| 1710 |
+
},
|
| 1711 |
+
{
|
| 1712 |
+
"epoch": 2.27,
|
| 1713 |
+
"learning_rate": 7.924406779661019e-06,
|
| 1714 |
+
"loss": 0.0686,
|
| 1715 |
+
"step": 6625
|
| 1716 |
+
},
|
| 1717 |
+
{
|
| 1718 |
+
"epoch": 2.28,
|
| 1719 |
+
"learning_rate": 7.915932203389832e-06,
|
| 1720 |
+
"loss": 0.0675,
|
| 1721 |
+
"step": 6650
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 2.29,
|
| 1725 |
+
"learning_rate": 7.907457627118645e-06,
|
| 1726 |
+
"loss": 0.0624,
|
| 1727 |
+
"step": 6675
|
| 1728 |
+
},
|
| 1729 |
+
{
|
| 1730 |
+
"epoch": 2.3,
|
| 1731 |
+
"learning_rate": 7.89898305084746e-06,
|
| 1732 |
+
"loss": 0.0553,
|
| 1733 |
+
"step": 6700
|
| 1734 |
+
},
|
| 1735 |
+
{
|
| 1736 |
+
"epoch": 2.31,
|
| 1737 |
+
"learning_rate": 7.890508474576272e-06,
|
| 1738 |
+
"loss": 0.0643,
|
| 1739 |
+
"step": 6725
|
| 1740 |
+
},
|
| 1741 |
+
{
|
| 1742 |
+
"epoch": 2.32,
|
| 1743 |
+
"learning_rate": 7.882033898305085e-06,
|
| 1744 |
+
"loss": 0.0573,
|
| 1745 |
+
"step": 6750
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 2.32,
|
| 1749 |
+
"learning_rate": 7.8735593220339e-06,
|
| 1750 |
+
"loss": 0.0698,
|
| 1751 |
+
"step": 6775
|
| 1752 |
+
},
|
| 1753 |
+
{
|
| 1754 |
+
"epoch": 2.33,
|
| 1755 |
+
"learning_rate": 7.865084745762713e-06,
|
| 1756 |
+
"loss": 0.0659,
|
| 1757 |
+
"step": 6800
|
| 1758 |
+
},
|
| 1759 |
+
{
|
| 1760 |
+
"epoch": 2.34,
|
| 1761 |
+
"learning_rate": 7.856610169491526e-06,
|
| 1762 |
+
"loss": 0.0654,
|
| 1763 |
+
"step": 6825
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 2.35,
|
| 1767 |
+
"learning_rate": 7.848135593220339e-06,
|
| 1768 |
+
"loss": 0.064,
|
| 1769 |
+
"step": 6850
|
| 1770 |
+
},
|
| 1771 |
+
{
|
| 1772 |
+
"epoch": 2.36,
|
| 1773 |
+
"learning_rate": 7.839661016949153e-06,
|
| 1774 |
+
"loss": 0.0615,
|
| 1775 |
+
"step": 6875
|
| 1776 |
+
},
|
| 1777 |
+
{
|
| 1778 |
+
"epoch": 2.37,
|
| 1779 |
+
"learning_rate": 7.831186440677966e-06,
|
| 1780 |
+
"loss": 0.0579,
|
| 1781 |
+
"step": 6900
|
| 1782 |
+
},
|
| 1783 |
+
{
|
| 1784 |
+
"epoch": 2.38,
|
| 1785 |
+
"learning_rate": 7.82271186440678e-06,
|
| 1786 |
+
"loss": 0.0675,
|
| 1787 |
+
"step": 6925
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 2.39,
|
| 1791 |
+
"learning_rate": 7.814237288135594e-06,
|
| 1792 |
+
"loss": 0.0682,
|
| 1793 |
+
"step": 6950
|
| 1794 |
+
},
|
| 1795 |
+
{
|
| 1796 |
+
"epoch": 2.39,
|
| 1797 |
+
"learning_rate": 7.805762711864407e-06,
|
| 1798 |
+
"loss": 0.0682,
|
| 1799 |
+
"step": 6975
|
| 1800 |
+
},
|
| 1801 |
+
{
|
| 1802 |
+
"epoch": 2.4,
|
| 1803 |
+
"learning_rate": 7.79728813559322e-06,
|
| 1804 |
+
"loss": 0.0745,
|
| 1805 |
+
"step": 7000
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 2.4,
|
| 1809 |
+
"eval_loss": 0.12451925128698349,
|
| 1810 |
+
"eval_runtime": 13911.3227,
|
| 1811 |
+
"eval_samples_per_second": 1.608,
|
| 1812 |
+
"eval_steps_per_second": 0.067,
|
| 1813 |
+
"eval_wer": 4.105449430554926,
|
| 1814 |
+
"step": 7000
|
| 1815 |
+
},
|
| 1816 |
+
{
|
| 1817 |
+
"epoch": 2.41,
|
| 1818 |
+
"learning_rate": 7.788813559322035e-06,
|
| 1819 |
+
"loss": 0.0574,
|
| 1820 |
+
"step": 7025
|
| 1821 |
+
},
|
| 1822 |
+
{
|
| 1823 |
+
"epoch": 2.42,
|
| 1824 |
+
"learning_rate": 7.780338983050848e-06,
|
| 1825 |
+
"loss": 0.0597,
|
| 1826 |
+
"step": 7050
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 2.43,
|
| 1830 |
+
"learning_rate": 7.77186440677966e-06,
|
| 1831 |
+
"loss": 0.0613,
|
| 1832 |
+
"step": 7075
|
| 1833 |
+
},
|
| 1834 |
+
{
|
| 1835 |
+
"epoch": 2.44,
|
| 1836 |
+
"learning_rate": 7.763389830508475e-06,
|
| 1837 |
+
"loss": 0.0712,
|
| 1838 |
+
"step": 7100
|
| 1839 |
+
},
|
| 1840 |
+
{
|
| 1841 |
+
"epoch": 2.45,
|
| 1842 |
+
"learning_rate": 7.754915254237288e-06,
|
| 1843 |
+
"loss": 0.0622,
|
| 1844 |
+
"step": 7125
|
| 1845 |
+
},
|
| 1846 |
+
{
|
| 1847 |
+
"epoch": 2.45,
|
| 1848 |
+
"learning_rate": 7.746440677966103e-06,
|
| 1849 |
+
"loss": 0.0631,
|
| 1850 |
+
"step": 7150
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 2.46,
|
| 1854 |
+
"learning_rate": 7.737966101694916e-06,
|
| 1855 |
+
"loss": 0.0696,
|
| 1856 |
+
"step": 7175
|
| 1857 |
+
},
|
| 1858 |
+
{
|
| 1859 |
+
"epoch": 2.47,
|
| 1860 |
+
"learning_rate": 7.729491525423729e-06,
|
| 1861 |
+
"loss": 0.0657,
|
| 1862 |
+
"step": 7200
|
| 1863 |
+
},
|
| 1864 |
+
{
|
| 1865 |
+
"epoch": 2.48,
|
| 1866 |
+
"learning_rate": 7.721016949152544e-06,
|
| 1867 |
+
"loss": 0.0593,
|
| 1868 |
+
"step": 7225
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 2.49,
|
| 1872 |
+
"learning_rate": 7.712542372881357e-06,
|
| 1873 |
+
"loss": 0.0644,
|
| 1874 |
+
"step": 7250
|
| 1875 |
+
},
|
| 1876 |
+
{
|
| 1877 |
+
"epoch": 2.5,
|
| 1878 |
+
"learning_rate": 7.70406779661017e-06,
|
| 1879 |
+
"loss": 0.0602,
|
| 1880 |
+
"step": 7275
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 2.51,
|
| 1884 |
+
"learning_rate": 7.695593220338984e-06,
|
| 1885 |
+
"loss": 0.0662,
|
| 1886 |
+
"step": 7300
|
| 1887 |
+
},
|
| 1888 |
+
{
|
| 1889 |
+
"epoch": 2.51,
|
| 1890 |
+
"learning_rate": 7.687118644067797e-06,
|
| 1891 |
+
"loss": 0.0654,
|
| 1892 |
+
"step": 7325
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 2.52,
|
| 1896 |
+
"learning_rate": 7.678983050847458e-06,
|
| 1897 |
+
"loss": 0.0699,
|
| 1898 |
+
"step": 7350
|
| 1899 |
+
},
|
| 1900 |
+
{
|
| 1901 |
+
"epoch": 2.53,
|
| 1902 |
+
"learning_rate": 7.670508474576271e-06,
|
| 1903 |
+
"loss": 0.0657,
|
| 1904 |
+
"step": 7375
|
| 1905 |
+
},
|
| 1906 |
+
{
|
| 1907 |
+
"epoch": 2.54,
|
| 1908 |
+
"learning_rate": 7.662033898305086e-06,
|
| 1909 |
+
"loss": 0.0668,
|
| 1910 |
+
"step": 7400
|
| 1911 |
+
},
|
| 1912 |
+
{
|
| 1913 |
+
"epoch": 2.55,
|
| 1914 |
+
"learning_rate": 7.653559322033899e-06,
|
| 1915 |
+
"loss": 0.0598,
|
| 1916 |
+
"step": 7425
|
| 1917 |
+
},
|
| 1918 |
+
{
|
| 1919 |
+
"epoch": 2.56,
|
| 1920 |
+
"learning_rate": 7.645084745762712e-06,
|
| 1921 |
+
"loss": 0.062,
|
| 1922 |
+
"step": 7450
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 2.57,
|
| 1926 |
+
"learning_rate": 7.636610169491526e-06,
|
| 1927 |
+
"loss": 0.0657,
|
| 1928 |
+
"step": 7475
|
| 1929 |
+
},
|
| 1930 |
+
{
|
| 1931 |
+
"epoch": 2.57,
|
| 1932 |
+
"learning_rate": 7.628135593220339e-06,
|
| 1933 |
+
"loss": 0.0686,
|
| 1934 |
+
"step": 7500
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 2.57,
|
| 1938 |
+
"eval_loss": 0.12511542439460754,
|
| 1939 |
+
"eval_runtime": 13875.855,
|
| 1940 |
+
"eval_samples_per_second": 1.612,
|
| 1941 |
+
"eval_steps_per_second": 0.067,
|
| 1942 |
+
"eval_wer": 4.054176686782522,
|
| 1943 |
+
"step": 7500
|
| 1944 |
+
}
|
| 1945 |
+
],
|
| 1946 |
+
"max_steps": 30000,
|
| 1947 |
+
"num_train_epochs": 11,
|
| 1948 |
+
"total_flos": 3.6740551944830976e+20,
|
| 1949 |
+
"trial_name": null,
|
| 1950 |
+
"trial_params": null
|
| 1951 |
+
}
|