End of training
Browse files- trainer_state.json +2614 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,2614 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": 32500,
|
| 3 |
+
"best_metric": 0.8929190695762744,
|
| 4 |
+
"best_model_checkpoint": "./lang-ner-xlmr/checkpoint-32500",
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 2500,
|
| 7 |
+
"global_step": 34550,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.005788879562360705,
|
| 14 |
+
"grad_norm": 2.855613946914673,
|
| 15 |
+
"learning_rate": 4.9856729377713464e-05,
|
| 16 |
+
"loss": 2.9271701049804686,
|
| 17 |
+
"step": 100
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.01157775912472141,
|
| 21 |
+
"grad_norm": 1.3880442380905151,
|
| 22 |
+
"learning_rate": 4.9712011577424025e-05,
|
| 23 |
+
"loss": 0.34735004425048827,
|
| 24 |
+
"step": 200
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.017366638687082114,
|
| 28 |
+
"grad_norm": 2.103525400161743,
|
| 29 |
+
"learning_rate": 4.9567293777134585e-05,
|
| 30 |
+
"loss": 0.14239531517028808,
|
| 31 |
+
"step": 300
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.02315551824944282,
|
| 35 |
+
"grad_norm": 1.3836640119552612,
|
| 36 |
+
"learning_rate": 4.942257597684515e-05,
|
| 37 |
+
"loss": 0.10217083930969238,
|
| 38 |
+
"step": 400
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.028944397811803524,
|
| 42 |
+
"grad_norm": 2.793973445892334,
|
| 43 |
+
"learning_rate": 4.927785817655572e-05,
|
| 44 |
+
"loss": 0.08668763160705567,
|
| 45 |
+
"step": 500
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.03473327737416423,
|
| 49 |
+
"grad_norm": 0.9237315654754639,
|
| 50 |
+
"learning_rate": 4.913314037626629e-05,
|
| 51 |
+
"loss": 0.07951077461242675,
|
| 52 |
+
"step": 600
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.04052215693652494,
|
| 56 |
+
"grad_norm": 1.0332331657409668,
|
| 57 |
+
"learning_rate": 4.898842257597685e-05,
|
| 58 |
+
"loss": 0.07264325618743897,
|
| 59 |
+
"step": 700
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.04631103649888564,
|
| 63 |
+
"grad_norm": 0.7395448088645935,
|
| 64 |
+
"learning_rate": 4.884370477568741e-05,
|
| 65 |
+
"loss": 0.0672088623046875,
|
| 66 |
+
"step": 800
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.052099916061246344,
|
| 70 |
+
"grad_norm": 0.4162349998950958,
|
| 71 |
+
"learning_rate": 4.8698986975397976e-05,
|
| 72 |
+
"loss": 0.06357911109924316,
|
| 73 |
+
"step": 900
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.05788879562360705,
|
| 77 |
+
"grad_norm": 1.134838342666626,
|
| 78 |
+
"learning_rate": 4.855426917510854e-05,
|
| 79 |
+
"loss": 0.06361278057098389,
|
| 80 |
+
"step": 1000
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.06367767518596776,
|
| 84 |
+
"grad_norm": 0.9246254563331604,
|
| 85 |
+
"learning_rate": 4.8409551374819104e-05,
|
| 86 |
+
"loss": 0.06069922924041748,
|
| 87 |
+
"step": 1100
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.06946655474832845,
|
| 91 |
+
"grad_norm": 0.4614603519439697,
|
| 92 |
+
"learning_rate": 4.826483357452967e-05,
|
| 93 |
+
"loss": 0.0631745719909668,
|
| 94 |
+
"step": 1200
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.07525543431068916,
|
| 98 |
+
"grad_norm": 1.585123062133789,
|
| 99 |
+
"learning_rate": 4.812011577424023e-05,
|
| 100 |
+
"loss": 0.05860543251037598,
|
| 101 |
+
"step": 1300
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.08104431387304988,
|
| 105 |
+
"grad_norm": 0.521318793296814,
|
| 106 |
+
"learning_rate": 4.79753979739508e-05,
|
| 107 |
+
"loss": 0.056866631507873536,
|
| 108 |
+
"step": 1400
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.08683319343541057,
|
| 112 |
+
"grad_norm": 0.9092019200325012,
|
| 113 |
+
"learning_rate": 4.7830680173661366e-05,
|
| 114 |
+
"loss": 0.06285685062408447,
|
| 115 |
+
"step": 1500
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.09262207299777128,
|
| 119 |
+
"grad_norm": 0.612201452255249,
|
| 120 |
+
"learning_rate": 4.7685962373371927e-05,
|
| 121 |
+
"loss": 0.054396772384643556,
|
| 122 |
+
"step": 1600
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.09841095256013199,
|
| 126 |
+
"grad_norm": 0.6812730431556702,
|
| 127 |
+
"learning_rate": 4.7541244573082494e-05,
|
| 128 |
+
"loss": 0.05651024341583252,
|
| 129 |
+
"step": 1700
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.10419983212249269,
|
| 133 |
+
"grad_norm": 0.8203144073486328,
|
| 134 |
+
"learning_rate": 4.7396526772793054e-05,
|
| 135 |
+
"loss": 0.061346669197082516,
|
| 136 |
+
"step": 1800
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.1099887116848534,
|
| 140 |
+
"grad_norm": 1.9262521266937256,
|
| 141 |
+
"learning_rate": 4.725180897250362e-05,
|
| 142 |
+
"loss": 0.05345334053039551,
|
| 143 |
+
"step": 1900
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.1157775912472141,
|
| 147 |
+
"grad_norm": 1.366341471672058,
|
| 148 |
+
"learning_rate": 4.710709117221418e-05,
|
| 149 |
+
"loss": 0.049096508026123045,
|
| 150 |
+
"step": 2000
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.1215664708095748,
|
| 154 |
+
"grad_norm": 1.9058470726013184,
|
| 155 |
+
"learning_rate": 4.696237337192475e-05,
|
| 156 |
+
"loss": 0.04964198589324951,
|
| 157 |
+
"step": 2100
|
| 158 |
+
},
|
| 159 |
+
{
|
| 160 |
+
"epoch": 0.12735535037193552,
|
| 161 |
+
"grad_norm": 0.20036761462688446,
|
| 162 |
+
"learning_rate": 4.681765557163532e-05,
|
| 163 |
+
"loss": 0.048866801261901856,
|
| 164 |
+
"step": 2200
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.13314422993429623,
|
| 168 |
+
"grad_norm": 0.6949800252914429,
|
| 169 |
+
"learning_rate": 4.667293777134588e-05,
|
| 170 |
+
"loss": 0.049001736640930174,
|
| 171 |
+
"step": 2300
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.1389331094966569,
|
| 175 |
+
"grad_norm": 1.4212485551834106,
|
| 176 |
+
"learning_rate": 4.6528219971056445e-05,
|
| 177 |
+
"loss": 0.05099475383758545,
|
| 178 |
+
"step": 2400
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.14472198905901762,
|
| 182 |
+
"grad_norm": 0.5998656749725342,
|
| 183 |
+
"learning_rate": 4.6383502170767005e-05,
|
| 184 |
+
"loss": 0.046523327827453616,
|
| 185 |
+
"step": 2500
|
| 186 |
+
},
|
| 187 |
+
{
|
| 188 |
+
"epoch": 0.14472198905901762,
|
| 189 |
+
"eval_accuracy": 0.9827913999033898,
|
| 190 |
+
"eval_f1": 0.8260160577581378,
|
| 191 |
+
"eval_loss": 0.08191228657960892,
|
| 192 |
+
"eval_precision": 0.794459210305773,
|
| 193 |
+
"eval_recall": 0.8601835528362543,
|
| 194 |
+
"eval_runtime": 41.7733,
|
| 195 |
+
"eval_samples_per_second": 299.234,
|
| 196 |
+
"eval_steps_per_second": 8.331,
|
| 197 |
+
"step": 2500
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.15051086862137833,
|
| 201 |
+
"grad_norm": 0.26346391439437866,
|
| 202 |
+
"learning_rate": 4.623878437047757e-05,
|
| 203 |
+
"loss": 0.04394114971160889,
|
| 204 |
+
"step": 2600
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.15629974818373904,
|
| 208 |
+
"grad_norm": 0.6683024168014526,
|
| 209 |
+
"learning_rate": 4.609406657018814e-05,
|
| 210 |
+
"loss": 0.04297324180603027,
|
| 211 |
+
"step": 2700
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.16208862774609975,
|
| 215 |
+
"grad_norm": 0.792713463306427,
|
| 216 |
+
"learning_rate": 4.59493487698987e-05,
|
| 217 |
+
"loss": 0.044585137367248534,
|
| 218 |
+
"step": 2800
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.16787750730846046,
|
| 222 |
+
"grad_norm": 0.8176506161689758,
|
| 223 |
+
"learning_rate": 4.580463096960926e-05,
|
| 224 |
+
"loss": 0.04621150970458984,
|
| 225 |
+
"step": 2900
|
| 226 |
+
},
|
| 227 |
+
{
|
| 228 |
+
"epoch": 0.17366638687082114,
|
| 229 |
+
"grad_norm": 0.21292151510715485,
|
| 230 |
+
"learning_rate": 4.565991316931983e-05,
|
| 231 |
+
"loss": 0.04335964679718018,
|
| 232 |
+
"step": 3000
|
| 233 |
+
},
|
| 234 |
+
{
|
| 235 |
+
"epoch": 0.17945526643318185,
|
| 236 |
+
"grad_norm": 0.5183402299880981,
|
| 237 |
+
"learning_rate": 4.5515195369030395e-05,
|
| 238 |
+
"loss": 0.045092535018920896,
|
| 239 |
+
"step": 3100
|
| 240 |
+
},
|
| 241 |
+
{
|
| 242 |
+
"epoch": 0.18524414599554256,
|
| 243 |
+
"grad_norm": 0.8505319356918335,
|
| 244 |
+
"learning_rate": 4.537047756874096e-05,
|
| 245 |
+
"loss": 0.03996892213821411,
|
| 246 |
+
"step": 3200
|
| 247 |
+
},
|
| 248 |
+
{
|
| 249 |
+
"epoch": 0.19103302555790327,
|
| 250 |
+
"grad_norm": 2.187335252761841,
|
| 251 |
+
"learning_rate": 4.522575976845152e-05,
|
| 252 |
+
"loss": 0.048550710678100586,
|
| 253 |
+
"step": 3300
|
| 254 |
+
},
|
| 255 |
+
{
|
| 256 |
+
"epoch": 0.19682190512026398,
|
| 257 |
+
"grad_norm": 0.5401766896247864,
|
| 258 |
+
"learning_rate": 4.5081041968162084e-05,
|
| 259 |
+
"loss": 0.03830349683761597,
|
| 260 |
+
"step": 3400
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"epoch": 0.20261078468262467,
|
| 264 |
+
"grad_norm": 3.9655873775482178,
|
| 265 |
+
"learning_rate": 4.493632416787265e-05,
|
| 266 |
+
"loss": 0.04094114303588867,
|
| 267 |
+
"step": 3500
|
| 268 |
+
},
|
| 269 |
+
{
|
| 270 |
+
"epoch": 0.20839966424498538,
|
| 271 |
+
"grad_norm": 2.4231624603271484,
|
| 272 |
+
"learning_rate": 4.479160636758322e-05,
|
| 273 |
+
"loss": 0.050821685791015626,
|
| 274 |
+
"step": 3600
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.2141885438073461,
|
| 278 |
+
"grad_norm": 0.8014847040176392,
|
| 279 |
+
"learning_rate": 4.464688856729378e-05,
|
| 280 |
+
"loss": 0.04339597225189209,
|
| 281 |
+
"step": 3700
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.2199774233697068,
|
| 285 |
+
"grad_norm": 0.6645370125770569,
|
| 286 |
+
"learning_rate": 4.450217076700434e-05,
|
| 287 |
+
"loss": 0.04568838119506836,
|
| 288 |
+
"step": 3800
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.2257663029320675,
|
| 292 |
+
"grad_norm": 0.7347345948219299,
|
| 293 |
+
"learning_rate": 4.435745296671491e-05,
|
| 294 |
+
"loss": 0.038632638454437256,
|
| 295 |
+
"step": 3900
|
| 296 |
+
},
|
| 297 |
+
{
|
| 298 |
+
"epoch": 0.2315551824944282,
|
| 299 |
+
"grad_norm": 0.5599935054779053,
|
| 300 |
+
"learning_rate": 4.4212735166425474e-05,
|
| 301 |
+
"loss": 0.048587980270385744,
|
| 302 |
+
"step": 4000
|
| 303 |
+
},
|
| 304 |
+
{
|
| 305 |
+
"epoch": 0.2373440620567889,
|
| 306 |
+
"grad_norm": 0.6407144665718079,
|
| 307 |
+
"learning_rate": 4.406801736613604e-05,
|
| 308 |
+
"loss": 0.04263918876647949,
|
| 309 |
+
"step": 4100
|
| 310 |
+
},
|
| 311 |
+
{
|
| 312 |
+
"epoch": 0.2431329416191496,
|
| 313 |
+
"grad_norm": 0.6048207879066467,
|
| 314 |
+
"learning_rate": 4.39232995658466e-05,
|
| 315 |
+
"loss": 0.039798662662506104,
|
| 316 |
+
"step": 4200
|
| 317 |
+
},
|
| 318 |
+
{
|
| 319 |
+
"epoch": 0.24892182118151032,
|
| 320 |
+
"grad_norm": 1.7978732585906982,
|
| 321 |
+
"learning_rate": 4.377858176555716e-05,
|
| 322 |
+
"loss": 0.04390395164489746,
|
| 323 |
+
"step": 4300
|
| 324 |
+
},
|
| 325 |
+
{
|
| 326 |
+
"epoch": 0.25471070074387103,
|
| 327 |
+
"grad_norm": 0.5556352734565735,
|
| 328 |
+
"learning_rate": 4.363386396526773e-05,
|
| 329 |
+
"loss": 0.03865460634231568,
|
| 330 |
+
"step": 4400
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.2604995803062317,
|
| 334 |
+
"grad_norm": 1.2710750102996826,
|
| 335 |
+
"learning_rate": 4.34891461649783e-05,
|
| 336 |
+
"loss": 0.041819486618041996,
|
| 337 |
+
"step": 4500
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.26628845986859245,
|
| 341 |
+
"grad_norm": 0.5540153384208679,
|
| 342 |
+
"learning_rate": 4.334442836468886e-05,
|
| 343 |
+
"loss": 0.03990748643875122,
|
| 344 |
+
"step": 4600
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.27207733943095314,
|
| 348 |
+
"grad_norm": 0.24696502089500427,
|
| 349 |
+
"learning_rate": 4.319971056439942e-05,
|
| 350 |
+
"loss": 0.039545681476593014,
|
| 351 |
+
"step": 4700
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.2778662189933138,
|
| 355 |
+
"grad_norm": 0.2856484353542328,
|
| 356 |
+
"learning_rate": 4.3054992764109985e-05,
|
| 357 |
+
"loss": 0.04045855045318603,
|
| 358 |
+
"step": 4800
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 0.28365509855567456,
|
| 362 |
+
"grad_norm": 0.4371595084667206,
|
| 363 |
+
"learning_rate": 4.291027496382055e-05,
|
| 364 |
+
"loss": 0.04244920253753662,
|
| 365 |
+
"step": 4900
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.28944397811803524,
|
| 369 |
+
"grad_norm": 0.42544451355934143,
|
| 370 |
+
"learning_rate": 4.276555716353112e-05,
|
| 371 |
+
"loss": 0.04395281791687012,
|
| 372 |
+
"step": 5000
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.28944397811803524,
|
| 376 |
+
"eval_accuracy": 0.9842557123690646,
|
| 377 |
+
"eval_f1": 0.8330216670406662,
|
| 378 |
+
"eval_loss": 0.07031189650297165,
|
| 379 |
+
"eval_precision": 0.8023149704625138,
|
| 380 |
+
"eval_recall": 0.8661723530021778,
|
| 381 |
+
"eval_runtime": 32.2916,
|
| 382 |
+
"eval_samples_per_second": 387.098,
|
| 383 |
+
"eval_steps_per_second": 10.777,
|
| 384 |
+
"step": 5000
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.295232857680396,
|
| 388 |
+
"grad_norm": 0.46878165006637573,
|
| 389 |
+
"learning_rate": 4.262083936324168e-05,
|
| 390 |
+
"loss": 0.041322021484375,
|
| 391 |
+
"step": 5100
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.30102173724275666,
|
| 395 |
+
"grad_norm": 0.6579214930534363,
|
| 396 |
+
"learning_rate": 4.247612156295224e-05,
|
| 397 |
+
"loss": 0.03830409288406372,
|
| 398 |
+
"step": 5200
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.30681061680511734,
|
| 402 |
+
"grad_norm": 0.42139366269111633,
|
| 403 |
+
"learning_rate": 4.233140376266281e-05,
|
| 404 |
+
"loss": 0.03726105690002442,
|
| 405 |
+
"step": 5300
|
| 406 |
+
},
|
| 407 |
+
{
|
| 408 |
+
"epoch": 0.3125994963674781,
|
| 409 |
+
"grad_norm": 0.2871117889881134,
|
| 410 |
+
"learning_rate": 4.2186685962373376e-05,
|
| 411 |
+
"loss": 0.04068236351013184,
|
| 412 |
+
"step": 5400
|
| 413 |
+
},
|
| 414 |
+
{
|
| 415 |
+
"epoch": 0.31838837592983876,
|
| 416 |
+
"grad_norm": 0.5657113790512085,
|
| 417 |
+
"learning_rate": 4.2041968162083936e-05,
|
| 418 |
+
"loss": 0.041853628158569335,
|
| 419 |
+
"step": 5500
|
| 420 |
+
},
|
| 421 |
+
{
|
| 422 |
+
"epoch": 0.3241772554921995,
|
| 423 |
+
"grad_norm": 1.3855081796646118,
|
| 424 |
+
"learning_rate": 4.1897250361794504e-05,
|
| 425 |
+
"loss": 0.039269649982452394,
|
| 426 |
+
"step": 5600
|
| 427 |
+
},
|
| 428 |
+
{
|
| 429 |
+
"epoch": 0.3299661350545602,
|
| 430 |
+
"grad_norm": 0.8465105295181274,
|
| 431 |
+
"learning_rate": 4.1752532561505064e-05,
|
| 432 |
+
"loss": 0.03952852487564087,
|
| 433 |
+
"step": 5700
|
| 434 |
+
},
|
| 435 |
+
{
|
| 436 |
+
"epoch": 0.3357550146169209,
|
| 437 |
+
"grad_norm": 0.3637921214103699,
|
| 438 |
+
"learning_rate": 4.160781476121563e-05,
|
| 439 |
+
"loss": 0.03640928745269775,
|
| 440 |
+
"step": 5800
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"epoch": 0.3415438941792816,
|
| 444 |
+
"grad_norm": 1.1824228763580322,
|
| 445 |
+
"learning_rate": 4.14630969609262e-05,
|
| 446 |
+
"loss": 0.03920279264450073,
|
| 447 |
+
"step": 5900
|
| 448 |
+
},
|
| 449 |
+
{
|
| 450 |
+
"epoch": 0.3473327737416423,
|
| 451 |
+
"grad_norm": 0.18278339505195618,
|
| 452 |
+
"learning_rate": 4.131837916063676e-05,
|
| 453 |
+
"loss": 0.03853697776794433,
|
| 454 |
+
"step": 6000
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 0.353121653304003,
|
| 458 |
+
"grad_norm": 0.6014581322669983,
|
| 459 |
+
"learning_rate": 4.1173661360347326e-05,
|
| 460 |
+
"loss": 0.03705440759658814,
|
| 461 |
+
"step": 6100
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 0.3589105328663637,
|
| 465 |
+
"grad_norm": 0.46531063318252563,
|
| 466 |
+
"learning_rate": 4.102894356005789e-05,
|
| 467 |
+
"loss": 0.03999743700027466,
|
| 468 |
+
"step": 6200
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 0.36469941242872445,
|
| 472 |
+
"grad_norm": 0.1958639770746231,
|
| 473 |
+
"learning_rate": 4.0884225759768454e-05,
|
| 474 |
+
"loss": 0.03664716958999634,
|
| 475 |
+
"step": 6300
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 0.37048829199108513,
|
| 479 |
+
"grad_norm": 0.8862726092338562,
|
| 480 |
+
"learning_rate": 4.0739507959479015e-05,
|
| 481 |
+
"loss": 0.0379261326789856,
|
| 482 |
+
"step": 6400
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 0.3762771715534458,
|
| 486 |
+
"grad_norm": 0.9592189192771912,
|
| 487 |
+
"learning_rate": 4.059479015918958e-05,
|
| 488 |
+
"loss": 0.03712235450744629,
|
| 489 |
+
"step": 6500
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 0.38206605111580655,
|
| 493 |
+
"grad_norm": 0.6839034557342529,
|
| 494 |
+
"learning_rate": 4.045007235890015e-05,
|
| 495 |
+
"loss": 0.03996901988983154,
|
| 496 |
+
"step": 6600
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 0.38785493067816723,
|
| 500 |
+
"grad_norm": 0.4922831952571869,
|
| 501 |
+
"learning_rate": 4.030535455861071e-05,
|
| 502 |
+
"loss": 0.03477381706237793,
|
| 503 |
+
"step": 6700
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 0.39364381024052797,
|
| 507 |
+
"grad_norm": 0.7703688740730286,
|
| 508 |
+
"learning_rate": 4.016063675832128e-05,
|
| 509 |
+
"loss": 0.03694391012191772,
|
| 510 |
+
"step": 6800
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 0.39943268980288865,
|
| 514 |
+
"grad_norm": 0.9563616514205933,
|
| 515 |
+
"learning_rate": 4.001591895803184e-05,
|
| 516 |
+
"loss": 0.038102331161499026,
|
| 517 |
+
"step": 6900
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 0.40522156936524933,
|
| 521 |
+
"grad_norm": 0.5397209525108337,
|
| 522 |
+
"learning_rate": 3.9871201157742405e-05,
|
| 523 |
+
"loss": 0.03435644626617432,
|
| 524 |
+
"step": 7000
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 0.4110104489276101,
|
| 528 |
+
"grad_norm": 0.33264386653900146,
|
| 529 |
+
"learning_rate": 3.972648335745297e-05,
|
| 530 |
+
"loss": 0.034106507301330566,
|
| 531 |
+
"step": 7100
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 0.41679932848997076,
|
| 535 |
+
"grad_norm": 0.4908270239830017,
|
| 536 |
+
"learning_rate": 3.958176555716353e-05,
|
| 537 |
+
"loss": 0.03293986797332764,
|
| 538 |
+
"step": 7200
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 0.4225882080523315,
|
| 542 |
+
"grad_norm": 1.4851772785186768,
|
| 543 |
+
"learning_rate": 3.9437047756874093e-05,
|
| 544 |
+
"loss": 0.041512365341186526,
|
| 545 |
+
"step": 7300
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 0.4283770876146922,
|
| 549 |
+
"grad_norm": 1.109512209892273,
|
| 550 |
+
"learning_rate": 3.929232995658466e-05,
|
| 551 |
+
"loss": 0.03690081834793091,
|
| 552 |
+
"step": 7400
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 0.43416596717705286,
|
| 556 |
+
"grad_norm": 1.2585190534591675,
|
| 557 |
+
"learning_rate": 3.914761215629523e-05,
|
| 558 |
+
"loss": 0.035063812732696535,
|
| 559 |
+
"step": 7500
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 0.43416596717705286,
|
| 563 |
+
"eval_accuracy": 0.9860376945037634,
|
| 564 |
+
"eval_f1": 0.8609184178154288,
|
| 565 |
+
"eval_loss": 0.06113845109939575,
|
| 566 |
+
"eval_precision": 0.8426553461929942,
|
| 567 |
+
"eval_recall": 0.8799906668049362,
|
| 568 |
+
"eval_runtime": 32.811,
|
| 569 |
+
"eval_samples_per_second": 380.97,
|
| 570 |
+
"eval_steps_per_second": 10.606,
|
| 571 |
+
"step": 7500
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 0.4399548467394136,
|
| 575 |
+
"grad_norm": 0.5866221189498901,
|
| 576 |
+
"learning_rate": 3.9002894356005795e-05,
|
| 577 |
+
"loss": 0.03833510398864746,
|
| 578 |
+
"step": 7600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 0.4457437263017743,
|
| 582 |
+
"grad_norm": 0.3515453636646271,
|
| 583 |
+
"learning_rate": 3.885817655571635e-05,
|
| 584 |
+
"loss": 0.03402453660964966,
|
| 585 |
+
"step": 7700
|
| 586 |
+
},
|
| 587 |
+
{
|
| 588 |
+
"epoch": 0.451532605864135,
|
| 589 |
+
"grad_norm": 0.5338568687438965,
|
| 590 |
+
"learning_rate": 3.8713458755426916e-05,
|
| 591 |
+
"loss": 0.03593117237091065,
|
| 592 |
+
"step": 7800
|
| 593 |
+
},
|
| 594 |
+
{
|
| 595 |
+
"epoch": 0.4573214854264957,
|
| 596 |
+
"grad_norm": 0.42428648471832275,
|
| 597 |
+
"learning_rate": 3.8568740955137484e-05,
|
| 598 |
+
"loss": 0.036249592304229736,
|
| 599 |
+
"step": 7900
|
| 600 |
+
},
|
| 601 |
+
{
|
| 602 |
+
"epoch": 0.4631103649888564,
|
| 603 |
+
"grad_norm": 0.5833432078361511,
|
| 604 |
+
"learning_rate": 3.842402315484805e-05,
|
| 605 |
+
"loss": 0.033896231651306154,
|
| 606 |
+
"step": 8000
|
| 607 |
+
},
|
| 608 |
+
{
|
| 609 |
+
"epoch": 0.4688992445512171,
|
| 610 |
+
"grad_norm": 0.24067725241184235,
|
| 611 |
+
"learning_rate": 3.827930535455861e-05,
|
| 612 |
+
"loss": 0.03889668464660644,
|
| 613 |
+
"step": 8100
|
| 614 |
+
},
|
| 615 |
+
{
|
| 616 |
+
"epoch": 0.4746881241135778,
|
| 617 |
+
"grad_norm": 1.6670092344284058,
|
| 618 |
+
"learning_rate": 3.813458755426917e-05,
|
| 619 |
+
"loss": 0.036442911624908446,
|
| 620 |
+
"step": 8200
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"epoch": 0.48047700367593854,
|
| 624 |
+
"grad_norm": 0.6004341244697571,
|
| 625 |
+
"learning_rate": 3.798986975397974e-05,
|
| 626 |
+
"loss": 0.037013137340545656,
|
| 627 |
+
"step": 8300
|
| 628 |
+
},
|
| 629 |
+
{
|
| 630 |
+
"epoch": 0.4862658832382992,
|
| 631 |
+
"grad_norm": 0.3944297432899475,
|
| 632 |
+
"learning_rate": 3.784515195369031e-05,
|
| 633 |
+
"loss": 0.036278882026672364,
|
| 634 |
+
"step": 8400
|
| 635 |
+
},
|
| 636 |
+
{
|
| 637 |
+
"epoch": 0.4920547628006599,
|
| 638 |
+
"grad_norm": 0.6234197020530701,
|
| 639 |
+
"learning_rate": 3.7700434153400874e-05,
|
| 640 |
+
"loss": 0.038663127422332765,
|
| 641 |
+
"step": 8500
|
| 642 |
+
},
|
| 643 |
+
{
|
| 644 |
+
"epoch": 0.49784364236302064,
|
| 645 |
+
"grad_norm": 0.403590589761734,
|
| 646 |
+
"learning_rate": 3.7555716353111435e-05,
|
| 647 |
+
"loss": 0.037761948108673095,
|
| 648 |
+
"step": 8600
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 0.5036325219253813,
|
| 652 |
+
"grad_norm": 0.44773924350738525,
|
| 653 |
+
"learning_rate": 3.7410998552821995e-05,
|
| 654 |
+
"loss": 0.03594711542129517,
|
| 655 |
+
"step": 8700
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 0.5094214014877421,
|
| 659 |
+
"grad_norm": 0.8247023820877075,
|
| 660 |
+
"learning_rate": 3.726628075253256e-05,
|
| 661 |
+
"loss": 0.03601913452148438,
|
| 662 |
+
"step": 8800
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 0.5152102810501028,
|
| 666 |
+
"grad_norm": 1.1590162515640259,
|
| 667 |
+
"learning_rate": 3.712156295224313e-05,
|
| 668 |
+
"loss": 0.03685137271881103,
|
| 669 |
+
"step": 8900
|
| 670 |
+
},
|
| 671 |
+
{
|
| 672 |
+
"epoch": 0.5209991606124634,
|
| 673 |
+
"grad_norm": 0.6263375878334045,
|
| 674 |
+
"learning_rate": 3.697684515195369e-05,
|
| 675 |
+
"loss": 0.036658039093017576,
|
| 676 |
+
"step": 9000
|
| 677 |
+
},
|
| 678 |
+
{
|
| 679 |
+
"epoch": 0.5267880401748242,
|
| 680 |
+
"grad_norm": 0.5682932138442993,
|
| 681 |
+
"learning_rate": 3.683212735166426e-05,
|
| 682 |
+
"loss": 0.03221390724182129,
|
| 683 |
+
"step": 9100
|
| 684 |
+
},
|
| 685 |
+
{
|
| 686 |
+
"epoch": 0.5325769197371849,
|
| 687 |
+
"grad_norm": 0.4283082187175751,
|
| 688 |
+
"learning_rate": 3.668740955137482e-05,
|
| 689 |
+
"loss": 0.03447196245193482,
|
| 690 |
+
"step": 9200
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"epoch": 0.5383657992995455,
|
| 694 |
+
"grad_norm": 0.513534426689148,
|
| 695 |
+
"learning_rate": 3.6542691751085385e-05,
|
| 696 |
+
"loss": 0.03253072500228882,
|
| 697 |
+
"step": 9300
|
| 698 |
+
},
|
| 699 |
+
{
|
| 700 |
+
"epoch": 0.5441546788619063,
|
| 701 |
+
"grad_norm": 0.1790156066417694,
|
| 702 |
+
"learning_rate": 3.639797395079595e-05,
|
| 703 |
+
"loss": 0.03559961318969727,
|
| 704 |
+
"step": 9400
|
| 705 |
+
},
|
| 706 |
+
{
|
| 707 |
+
"epoch": 0.549943558424267,
|
| 708 |
+
"grad_norm": 1.3116710186004639,
|
| 709 |
+
"learning_rate": 3.625325615050651e-05,
|
| 710 |
+
"loss": 0.0339055609703064,
|
| 711 |
+
"step": 9500
|
| 712 |
+
},
|
| 713 |
+
{
|
| 714 |
+
"epoch": 0.5557324379866276,
|
| 715 |
+
"grad_norm": 0.7609701752662659,
|
| 716 |
+
"learning_rate": 3.610853835021708e-05,
|
| 717 |
+
"loss": 0.033302297592163084,
|
| 718 |
+
"step": 9600
|
| 719 |
+
},
|
| 720 |
+
{
|
| 721 |
+
"epoch": 0.5615213175489884,
|
| 722 |
+
"grad_norm": 0.6684858798980713,
|
| 723 |
+
"learning_rate": 3.596382054992764e-05,
|
| 724 |
+
"loss": 0.03894152641296387,
|
| 725 |
+
"step": 9700
|
| 726 |
+
},
|
| 727 |
+
{
|
| 728 |
+
"epoch": 0.5673101971113491,
|
| 729 |
+
"grad_norm": 0.25678926706314087,
|
| 730 |
+
"learning_rate": 3.581910274963821e-05,
|
| 731 |
+
"loss": 0.033086774349212644,
|
| 732 |
+
"step": 9800
|
| 733 |
+
},
|
| 734 |
+
{
|
| 735 |
+
"epoch": 0.5730990766737099,
|
| 736 |
+
"grad_norm": 1.2110711336135864,
|
| 737 |
+
"learning_rate": 3.567438494934877e-05,
|
| 738 |
+
"loss": 0.037588162422180174,
|
| 739 |
+
"step": 9900
|
| 740 |
+
},
|
| 741 |
+
{
|
| 742 |
+
"epoch": 0.5788879562360705,
|
| 743 |
+
"grad_norm": 0.7344034910202026,
|
| 744 |
+
"learning_rate": 3.5529667149059336e-05,
|
| 745 |
+
"loss": 0.03140888452529907,
|
| 746 |
+
"step": 10000
|
| 747 |
+
},
|
| 748 |
+
{
|
| 749 |
+
"epoch": 0.5788879562360705,
|
| 750 |
+
"eval_accuracy": 0.9871526392423338,
|
| 751 |
+
"eval_f1": 0.8693991367126322,
|
| 752 |
+
"eval_loss": 0.0593421570956707,
|
| 753 |
+
"eval_precision": 0.8542474665332166,
|
| 754 |
+
"eval_recall": 0.8850979985481696,
|
| 755 |
+
"eval_runtime": 32.4321,
|
| 756 |
+
"eval_samples_per_second": 385.42,
|
| 757 |
+
"eval_steps_per_second": 10.73,
|
| 758 |
+
"step": 10000
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 0.5846768357984312,
|
| 762 |
+
"grad_norm": 1.8443788290023804,
|
| 763 |
+
"learning_rate": 3.5384949348769903e-05,
|
| 764 |
+
"loss": 0.034446282386779783,
|
| 765 |
+
"step": 10100
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 0.590465715360792,
|
| 769 |
+
"grad_norm": 3.829798698425293,
|
| 770 |
+
"learning_rate": 3.5240231548480464e-05,
|
| 771 |
+
"loss": 0.031233868598937987,
|
| 772 |
+
"step": 10200
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 0.5962545949231526,
|
| 776 |
+
"grad_norm": 0.22466252744197845,
|
| 777 |
+
"learning_rate": 3.5095513748191025e-05,
|
| 778 |
+
"loss": 0.03304744958877563,
|
| 779 |
+
"step": 10300
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 0.6020434744855133,
|
| 783 |
+
"grad_norm": 0.2421451061964035,
|
| 784 |
+
"learning_rate": 3.495079594790159e-05,
|
| 785 |
+
"loss": 0.032493886947631834,
|
| 786 |
+
"step": 10400
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 0.6078323540478741,
|
| 790 |
+
"grad_norm": 6.648340225219727,
|
| 791 |
+
"learning_rate": 3.480607814761216e-05,
|
| 792 |
+
"loss": 0.034116878509521484,
|
| 793 |
+
"step": 10500
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 0.6136212336102347,
|
| 797 |
+
"grad_norm": 0.5613166093826294,
|
| 798 |
+
"learning_rate": 3.4661360347322726e-05,
|
| 799 |
+
"loss": 0.03445070266723633,
|
| 800 |
+
"step": 10600
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 0.6194101131725954,
|
| 804 |
+
"grad_norm": 0.24381904304027557,
|
| 805 |
+
"learning_rate": 3.451664254703329e-05,
|
| 806 |
+
"loss": 0.03339603900909424,
|
| 807 |
+
"step": 10700
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 0.6251989927349562,
|
| 811 |
+
"grad_norm": 0.5657049417495728,
|
| 812 |
+
"learning_rate": 3.437192474674385e-05,
|
| 813 |
+
"loss": 0.03580770492553711,
|
| 814 |
+
"step": 10800
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 0.6309878722973169,
|
| 818 |
+
"grad_norm": 0.16176201403141022,
|
| 819 |
+
"learning_rate": 3.4227206946454415e-05,
|
| 820 |
+
"loss": 0.03211415767669678,
|
| 821 |
+
"step": 10900
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 0.6367767518596775,
|
| 825 |
+
"grad_norm": 0.9659017324447632,
|
| 826 |
+
"learning_rate": 3.408248914616498e-05,
|
| 827 |
+
"loss": 0.032091012001037596,
|
| 828 |
+
"step": 11000
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 0.6425656314220383,
|
| 832 |
+
"grad_norm": 0.37151074409484863,
|
| 833 |
+
"learning_rate": 3.393777134587555e-05,
|
| 834 |
+
"loss": 0.030981764793395997,
|
| 835 |
+
"step": 11100
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 0.648354510984399,
|
| 839 |
+
"grad_norm": 0.24677230417728424,
|
| 840 |
+
"learning_rate": 3.37930535455861e-05,
|
| 841 |
+
"loss": 0.031599972248077396,
|
| 842 |
+
"step": 11200
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 0.6541433905467596,
|
| 846 |
+
"grad_norm": 0.5559372901916504,
|
| 847 |
+
"learning_rate": 3.364833574529667e-05,
|
| 848 |
+
"loss": 0.030684573650360106,
|
| 849 |
+
"step": 11300
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 0.6599322701091204,
|
| 853 |
+
"grad_norm": 0.309211790561676,
|
| 854 |
+
"learning_rate": 3.350361794500724e-05,
|
| 855 |
+
"loss": 0.03191032409667969,
|
| 856 |
+
"step": 11400
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 0.6657211496714811,
|
| 860 |
+
"grad_norm": 0.1990319788455963,
|
| 861 |
+
"learning_rate": 3.3358900144717805e-05,
|
| 862 |
+
"loss": 0.030000259876251222,
|
| 863 |
+
"step": 11500
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 0.6715100292338418,
|
| 867 |
+
"grad_norm": 0.6448608636856079,
|
| 868 |
+
"learning_rate": 3.3214182344428366e-05,
|
| 869 |
+
"loss": 0.03020137071609497,
|
| 870 |
+
"step": 11600
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 0.6772989087962025,
|
| 874 |
+
"grad_norm": 0.4042809009552002,
|
| 875 |
+
"learning_rate": 3.3069464544138926e-05,
|
| 876 |
+
"loss": 0.0288107705116272,
|
| 877 |
+
"step": 11700
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 0.6830877883585632,
|
| 881 |
+
"grad_norm": 0.48739156126976013,
|
| 882 |
+
"learning_rate": 3.2924746743849493e-05,
|
| 883 |
+
"loss": 0.02975822925567627,
|
| 884 |
+
"step": 11800
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 0.688876667920924,
|
| 888 |
+
"grad_norm": 0.3511544466018677,
|
| 889 |
+
"learning_rate": 3.278002894356006e-05,
|
| 890 |
+
"loss": 0.030011208057403566,
|
| 891 |
+
"step": 11900
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 0.6946655474832846,
|
| 895 |
+
"grad_norm": 0.20981945097446442,
|
| 896 |
+
"learning_rate": 3.263531114327063e-05,
|
| 897 |
+
"loss": 0.030392658710479737,
|
| 898 |
+
"step": 12000
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 0.7004544270456453,
|
| 902 |
+
"grad_norm": 1.725892186164856,
|
| 903 |
+
"learning_rate": 3.249059334298119e-05,
|
| 904 |
+
"loss": 0.0313503360748291,
|
| 905 |
+
"step": 12100
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 0.706243306608006,
|
| 909 |
+
"grad_norm": 0.16606950759887695,
|
| 910 |
+
"learning_rate": 3.234587554269175e-05,
|
| 911 |
+
"loss": 0.03735621213912964,
|
| 912 |
+
"step": 12200
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 0.7120321861703667,
|
| 916 |
+
"grad_norm": 0.4616575539112091,
|
| 917 |
+
"learning_rate": 3.2201157742402316e-05,
|
| 918 |
+
"loss": 0.03038395643234253,
|
| 919 |
+
"step": 12300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 0.7178210657327274,
|
| 923 |
+
"grad_norm": 0.40673601627349854,
|
| 924 |
+
"learning_rate": 3.2056439942112884e-05,
|
| 925 |
+
"loss": 0.035007147789001464,
|
| 926 |
+
"step": 12400
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 0.7236099452950882,
|
| 930 |
+
"grad_norm": 0.27561140060424805,
|
| 931 |
+
"learning_rate": 3.1911722141823444e-05,
|
| 932 |
+
"loss": 0.032888281345367434,
|
| 933 |
+
"step": 12500
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 0.7236099452950882,
|
| 937 |
+
"eval_accuracy": 0.9868453058965968,
|
| 938 |
+
"eval_f1": 0.8583013393139975,
|
| 939 |
+
"eval_loss": 0.056312501430511475,
|
| 940 |
+
"eval_precision": 0.8394012243178269,
|
| 941 |
+
"eval_recall": 0.8780721767084932,
|
| 942 |
+
"eval_runtime": 32.461,
|
| 943 |
+
"eval_samples_per_second": 385.078,
|
| 944 |
+
"eval_steps_per_second": 10.721,
|
| 945 |
+
"step": 12500
|
| 946 |
+
},
|
| 947 |
+
{
|
| 948 |
+
"epoch": 0.7293988248574489,
|
| 949 |
+
"grad_norm": 0.7054789662361145,
|
| 950 |
+
"learning_rate": 3.176700434153401e-05,
|
| 951 |
+
"loss": 0.031692807674407956,
|
| 952 |
+
"step": 12600
|
| 953 |
+
},
|
| 954 |
+
{
|
| 955 |
+
"epoch": 0.7351877044198095,
|
| 956 |
+
"grad_norm": 0.31711819767951965,
|
| 957 |
+
"learning_rate": 3.162228654124457e-05,
|
| 958 |
+
"loss": 0.03108912706375122,
|
| 959 |
+
"step": 12700
|
| 960 |
+
},
|
| 961 |
+
{
|
| 962 |
+
"epoch": 0.7409765839821703,
|
| 963 |
+
"grad_norm": 0.24462510645389557,
|
| 964 |
+
"learning_rate": 3.147756874095514e-05,
|
| 965 |
+
"loss": 0.03314091444015503,
|
| 966 |
+
"step": 12800
|
| 967 |
+
},
|
| 968 |
+
{
|
| 969 |
+
"epoch": 0.746765463544531,
|
| 970 |
+
"grad_norm": 0.37591752409935,
|
| 971 |
+
"learning_rate": 3.133285094066571e-05,
|
| 972 |
+
"loss": 0.031636784076690676,
|
| 973 |
+
"step": 12900
|
| 974 |
+
},
|
| 975 |
+
{
|
| 976 |
+
"epoch": 0.7525543431068916,
|
| 977 |
+
"grad_norm": 0.16671085357666016,
|
| 978 |
+
"learning_rate": 3.118813314037627e-05,
|
| 979 |
+
"loss": 0.027464551925659178,
|
| 980 |
+
"step": 13000
|
| 981 |
+
},
|
| 982 |
+
{
|
| 983 |
+
"epoch": 0.7583432226692524,
|
| 984 |
+
"grad_norm": 1.5406149625778198,
|
| 985 |
+
"learning_rate": 3.1043415340086835e-05,
|
| 986 |
+
"loss": 0.03511166572570801,
|
| 987 |
+
"step": 13100
|
| 988 |
+
},
|
| 989 |
+
{
|
| 990 |
+
"epoch": 0.7641321022316131,
|
| 991 |
+
"grad_norm": 0.1813717931509018,
|
| 992 |
+
"learning_rate": 3.0898697539797395e-05,
|
| 993 |
+
"loss": 0.0282421875,
|
| 994 |
+
"step": 13200
|
| 995 |
+
},
|
| 996 |
+
{
|
| 997 |
+
"epoch": 0.7699209817939737,
|
| 998 |
+
"grad_norm": 0.25257691740989685,
|
| 999 |
+
"learning_rate": 3.075397973950796e-05,
|
| 1000 |
+
"loss": 0.029794423580169677,
|
| 1001 |
+
"step": 13300
|
| 1002 |
+
},
|
| 1003 |
+
{
|
| 1004 |
+
"epoch": 0.7757098613563345,
|
| 1005 |
+
"grad_norm": 0.534304141998291,
|
| 1006 |
+
"learning_rate": 3.060926193921852e-05,
|
| 1007 |
+
"loss": 0.030250344276428223,
|
| 1008 |
+
"step": 13400
|
| 1009 |
+
},
|
| 1010 |
+
{
|
| 1011 |
+
"epoch": 0.7814987409186952,
|
| 1012 |
+
"grad_norm": 0.8739162087440491,
|
| 1013 |
+
"learning_rate": 3.046454413892909e-05,
|
| 1014 |
+
"loss": 0.031027588844299316,
|
| 1015 |
+
"step": 13500
|
| 1016 |
+
},
|
| 1017 |
+
{
|
| 1018 |
+
"epoch": 0.7872876204810559,
|
| 1019 |
+
"grad_norm": 0.540971577167511,
|
| 1020 |
+
"learning_rate": 3.0319826338639658e-05,
|
| 1021 |
+
"loss": 0.03038400173187256,
|
| 1022 |
+
"step": 13600
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 0.7930765000434166,
|
| 1026 |
+
"grad_norm": 0.7075596451759338,
|
| 1027 |
+
"learning_rate": 3.0175108538350215e-05,
|
| 1028 |
+
"loss": 0.03126862049102783,
|
| 1029 |
+
"step": 13700
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 0.7988653796057773,
|
| 1033 |
+
"grad_norm": 0.22685867547988892,
|
| 1034 |
+
"learning_rate": 3.0030390738060782e-05,
|
| 1035 |
+
"loss": 0.03233437061309814,
|
| 1036 |
+
"step": 13800
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 0.804654259168138,
|
| 1040 |
+
"grad_norm": 3.8470654487609863,
|
| 1041 |
+
"learning_rate": 2.9885672937771346e-05,
|
| 1042 |
+
"loss": 0.03189040184020996,
|
| 1043 |
+
"step": 13900
|
| 1044 |
+
},
|
| 1045 |
+
{
|
| 1046 |
+
"epoch": 0.8104431387304987,
|
| 1047 |
+
"grad_norm": 0.16255338490009308,
|
| 1048 |
+
"learning_rate": 2.9740955137481913e-05,
|
| 1049 |
+
"loss": 0.030559961795806886,
|
| 1050 |
+
"step": 14000
|
| 1051 |
+
},
|
| 1052 |
+
{
|
| 1053 |
+
"epoch": 0.8162320182928594,
|
| 1054 |
+
"grad_norm": 0.4145297110080719,
|
| 1055 |
+
"learning_rate": 2.9596237337192477e-05,
|
| 1056 |
+
"loss": 0.027420687675476074,
|
| 1057 |
+
"step": 14100
|
| 1058 |
+
},
|
| 1059 |
+
{
|
| 1060 |
+
"epoch": 0.8220208978552201,
|
| 1061 |
+
"grad_norm": 0.23214294016361237,
|
| 1062 |
+
"learning_rate": 2.9451519536903038e-05,
|
| 1063 |
+
"loss": 0.02794215202331543,
|
| 1064 |
+
"step": 14200
|
| 1065 |
+
},
|
| 1066 |
+
{
|
| 1067 |
+
"epoch": 0.8278097774175808,
|
| 1068 |
+
"grad_norm": 0.18062900006771088,
|
| 1069 |
+
"learning_rate": 2.9306801736613605e-05,
|
| 1070 |
+
"loss": 0.028336293697357177,
|
| 1071 |
+
"step": 14300
|
| 1072 |
+
},
|
| 1073 |
+
{
|
| 1074 |
+
"epoch": 0.8335986569799415,
|
| 1075 |
+
"grad_norm": 0.7959395051002502,
|
| 1076 |
+
"learning_rate": 2.916208393632417e-05,
|
| 1077 |
+
"loss": 0.028267955780029295,
|
| 1078 |
+
"step": 14400
|
| 1079 |
+
},
|
| 1080 |
+
{
|
| 1081 |
+
"epoch": 0.8393875365423022,
|
| 1082 |
+
"grad_norm": 0.5956525802612305,
|
| 1083 |
+
"learning_rate": 2.9017366136034736e-05,
|
| 1084 |
+
"loss": 0.028834567070007325,
|
| 1085 |
+
"step": 14500
|
| 1086 |
+
},
|
| 1087 |
+
{
|
| 1088 |
+
"epoch": 0.845176416104663,
|
| 1089 |
+
"grad_norm": 0.3449164628982544,
|
| 1090 |
+
"learning_rate": 2.88726483357453e-05,
|
| 1091 |
+
"loss": 0.03145886182785034,
|
| 1092 |
+
"step": 14600
|
| 1093 |
+
},
|
| 1094 |
+
{
|
| 1095 |
+
"epoch": 0.8509652956670236,
|
| 1096 |
+
"grad_norm": 0.16741275787353516,
|
| 1097 |
+
"learning_rate": 2.872793053545586e-05,
|
| 1098 |
+
"loss": 0.026835625171661378,
|
| 1099 |
+
"step": 14700
|
| 1100 |
+
},
|
| 1101 |
+
{
|
| 1102 |
+
"epoch": 0.8567541752293844,
|
| 1103 |
+
"grad_norm": 0.46983611583709717,
|
| 1104 |
+
"learning_rate": 2.8583212735166425e-05,
|
| 1105 |
+
"loss": 0.03117095708847046,
|
| 1106 |
+
"step": 14800
|
| 1107 |
+
},
|
| 1108 |
+
{
|
| 1109 |
+
"epoch": 0.8625430547917451,
|
| 1110 |
+
"grad_norm": 0.22663377225399017,
|
| 1111 |
+
"learning_rate": 2.8438494934876992e-05,
|
| 1112 |
+
"loss": 0.028009552955627442,
|
| 1113 |
+
"step": 14900
|
| 1114 |
+
},
|
| 1115 |
+
{
|
| 1116 |
+
"epoch": 0.8683319343541057,
|
| 1117 |
+
"grad_norm": 0.23122131824493408,
|
| 1118 |
+
"learning_rate": 2.8293777134587556e-05,
|
| 1119 |
+
"loss": 0.028059725761413575,
|
| 1120 |
+
"step": 15000
|
| 1121 |
+
},
|
| 1122 |
+
{
|
| 1123 |
+
"epoch": 0.8683319343541057,
|
| 1124 |
+
"eval_accuracy": 0.9886279771195148,
|
| 1125 |
+
"eval_f1": 0.8722127148731833,
|
| 1126 |
+
"eval_loss": 0.04878264293074608,
|
| 1127 |
+
"eval_precision": 0.8595016360432922,
|
| 1128 |
+
"eval_recall": 0.8853054028829203,
|
| 1129 |
+
"eval_runtime": 32.7448,
|
| 1130 |
+
"eval_samples_per_second": 381.74,
|
| 1131 |
+
"eval_steps_per_second": 10.628,
|
| 1132 |
+
"step": 15000
|
| 1133 |
+
},
|
| 1134 |
+
{
|
| 1135 |
+
"epoch": 0.8741208139164665,
|
| 1136 |
+
"grad_norm": 0.6293914318084717,
|
| 1137 |
+
"learning_rate": 2.8149059334298123e-05,
|
| 1138 |
+
"loss": 0.030007996559143067,
|
| 1139 |
+
"step": 15100
|
| 1140 |
+
},
|
| 1141 |
+
{
|
| 1142 |
+
"epoch": 0.8799096934788272,
|
| 1143 |
+
"grad_norm": 0.9233108758926392,
|
| 1144 |
+
"learning_rate": 2.8004341534008684e-05,
|
| 1145 |
+
"loss": 0.030494704246520996,
|
| 1146 |
+
"step": 15200
|
| 1147 |
+
},
|
| 1148 |
+
{
|
| 1149 |
+
"epoch": 0.8856985730411879,
|
| 1150 |
+
"grad_norm": 0.28081145882606506,
|
| 1151 |
+
"learning_rate": 2.7859623733719247e-05,
|
| 1152 |
+
"loss": 0.026717009544372557,
|
| 1153 |
+
"step": 15300
|
| 1154 |
+
},
|
| 1155 |
+
{
|
| 1156 |
+
"epoch": 0.8914874526035486,
|
| 1157 |
+
"grad_norm": 0.6658106446266174,
|
| 1158 |
+
"learning_rate": 2.7714905933429815e-05,
|
| 1159 |
+
"loss": 0.031118297576904298,
|
| 1160 |
+
"step": 15400
|
| 1161 |
+
},
|
| 1162 |
+
{
|
| 1163 |
+
"epoch": 0.8972763321659093,
|
| 1164 |
+
"grad_norm": 0.940838098526001,
|
| 1165 |
+
"learning_rate": 2.757018813314038e-05,
|
| 1166 |
+
"loss": 0.03049607515335083,
|
| 1167 |
+
"step": 15500
|
| 1168 |
+
},
|
| 1169 |
+
{
|
| 1170 |
+
"epoch": 0.90306521172827,
|
| 1171 |
+
"grad_norm": 0.12419965863227844,
|
| 1172 |
+
"learning_rate": 2.7425470332850946e-05,
|
| 1173 |
+
"loss": 0.026502671241760253,
|
| 1174 |
+
"step": 15600
|
| 1175 |
+
},
|
| 1176 |
+
{
|
| 1177 |
+
"epoch": 0.9088540912906307,
|
| 1178 |
+
"grad_norm": 1.1096726655960083,
|
| 1179 |
+
"learning_rate": 2.7280752532561503e-05,
|
| 1180 |
+
"loss": 0.02529672861099243,
|
| 1181 |
+
"step": 15700
|
| 1182 |
+
},
|
| 1183 |
+
{
|
| 1184 |
+
"epoch": 0.9146429708529914,
|
| 1185 |
+
"grad_norm": 0.3621964752674103,
|
| 1186 |
+
"learning_rate": 2.713603473227207e-05,
|
| 1187 |
+
"loss": 0.02636525869369507,
|
| 1188 |
+
"step": 15800
|
| 1189 |
+
},
|
| 1190 |
+
{
|
| 1191 |
+
"epoch": 0.9204318504153521,
|
| 1192 |
+
"grad_norm": 0.21582575142383575,
|
| 1193 |
+
"learning_rate": 2.6991316931982634e-05,
|
| 1194 |
+
"loss": 0.03223026037216187,
|
| 1195 |
+
"step": 15900
|
| 1196 |
+
},
|
| 1197 |
+
{
|
| 1198 |
+
"epoch": 0.9262207299777128,
|
| 1199 |
+
"grad_norm": 0.25821372866630554,
|
| 1200 |
+
"learning_rate": 2.68465991316932e-05,
|
| 1201 |
+
"loss": 0.026826431751251222,
|
| 1202 |
+
"step": 16000
|
| 1203 |
+
},
|
| 1204 |
+
{
|
| 1205 |
+
"epoch": 0.9320096095400735,
|
| 1206 |
+
"grad_norm": 0.399935245513916,
|
| 1207 |
+
"learning_rate": 2.6701881331403766e-05,
|
| 1208 |
+
"loss": 0.028431081771850587,
|
| 1209 |
+
"step": 16100
|
| 1210 |
+
},
|
| 1211 |
+
{
|
| 1212 |
+
"epoch": 0.9377984891024342,
|
| 1213 |
+
"grad_norm": 0.5942548513412476,
|
| 1214 |
+
"learning_rate": 2.6557163531114326e-05,
|
| 1215 |
+
"loss": 0.027340149879455565,
|
| 1216 |
+
"step": 16200
|
| 1217 |
+
},
|
| 1218 |
+
{
|
| 1219 |
+
"epoch": 0.943587368664795,
|
| 1220 |
+
"grad_norm": 0.3555188775062561,
|
| 1221 |
+
"learning_rate": 2.6412445730824893e-05,
|
| 1222 |
+
"loss": 0.02627647399902344,
|
| 1223 |
+
"step": 16300
|
| 1224 |
+
},
|
| 1225 |
+
{
|
| 1226 |
+
"epoch": 0.9493762482271556,
|
| 1227 |
+
"grad_norm": 0.14022284746170044,
|
| 1228 |
+
"learning_rate": 2.6267727930535457e-05,
|
| 1229 |
+
"loss": 0.027896411418914795,
|
| 1230 |
+
"step": 16400
|
| 1231 |
+
},
|
| 1232 |
+
{
|
| 1233 |
+
"epoch": 0.9551651277895163,
|
| 1234 |
+
"grad_norm": 0.5751340389251709,
|
| 1235 |
+
"learning_rate": 2.612301013024602e-05,
|
| 1236 |
+
"loss": 0.02570687770843506,
|
| 1237 |
+
"step": 16500
|
| 1238 |
+
},
|
| 1239 |
+
{
|
| 1240 |
+
"epoch": 0.9609540073518771,
|
| 1241 |
+
"grad_norm": 0.44881781935691833,
|
| 1242 |
+
"learning_rate": 2.597829232995659e-05,
|
| 1243 |
+
"loss": 0.030154764652252197,
|
| 1244 |
+
"step": 16600
|
| 1245 |
+
},
|
| 1246 |
+
{
|
| 1247 |
+
"epoch": 0.9667428869142377,
|
| 1248 |
+
"grad_norm": 0.1963057667016983,
|
| 1249 |
+
"learning_rate": 2.583357452966715e-05,
|
| 1250 |
+
"loss": 0.031277601718902585,
|
| 1251 |
+
"step": 16700
|
| 1252 |
+
},
|
| 1253 |
+
{
|
| 1254 |
+
"epoch": 0.9725317664765984,
|
| 1255 |
+
"grad_norm": 0.4031456410884857,
|
| 1256 |
+
"learning_rate": 2.5688856729377713e-05,
|
| 1257 |
+
"loss": 0.030245101451873778,
|
| 1258 |
+
"step": 16800
|
| 1259 |
+
},
|
| 1260 |
+
{
|
| 1261 |
+
"epoch": 0.9783206460389592,
|
| 1262 |
+
"grad_norm": 0.29256272315979004,
|
| 1263 |
+
"learning_rate": 2.554413892908828e-05,
|
| 1264 |
+
"loss": 0.031824679374694825,
|
| 1265 |
+
"step": 16900
|
| 1266 |
+
},
|
| 1267 |
+
{
|
| 1268 |
+
"epoch": 0.9841095256013198,
|
| 1269 |
+
"grad_norm": 0.42400193214416504,
|
| 1270 |
+
"learning_rate": 2.5399421128798844e-05,
|
| 1271 |
+
"loss": 0.028290491104125976,
|
| 1272 |
+
"step": 17000
|
| 1273 |
+
},
|
| 1274 |
+
{
|
| 1275 |
+
"epoch": 0.9898984051636806,
|
| 1276 |
+
"grad_norm": 1.173027753829956,
|
| 1277 |
+
"learning_rate": 2.525470332850941e-05,
|
| 1278 |
+
"loss": 0.02786562204360962,
|
| 1279 |
+
"step": 17100
|
| 1280 |
+
},
|
| 1281 |
+
{
|
| 1282 |
+
"epoch": 0.9956872847260413,
|
| 1283 |
+
"grad_norm": 0.13319961726665497,
|
| 1284 |
+
"learning_rate": 2.510998552821997e-05,
|
| 1285 |
+
"loss": 0.030247974395751952,
|
| 1286 |
+
"step": 17200
|
| 1287 |
+
},
|
| 1288 |
+
{
|
| 1289 |
+
"epoch": 1.0014472198905902,
|
| 1290 |
+
"grad_norm": 0.8692908883094788,
|
| 1291 |
+
"learning_rate": 2.4965267727930536e-05,
|
| 1292 |
+
"loss": 0.029876837730407713,
|
| 1293 |
+
"step": 17300
|
| 1294 |
+
},
|
| 1295 |
+
{
|
| 1296 |
+
"epoch": 1.0072360994529508,
|
| 1297 |
+
"grad_norm": 0.132135272026062,
|
| 1298 |
+
"learning_rate": 2.48205499276411e-05,
|
| 1299 |
+
"loss": 0.028613386154174806,
|
| 1300 |
+
"step": 17400
|
| 1301 |
+
},
|
| 1302 |
+
{
|
| 1303 |
+
"epoch": 1.0130249790153116,
|
| 1304 |
+
"grad_norm": 0.43961101770401,
|
| 1305 |
+
"learning_rate": 2.4675832127351664e-05,
|
| 1306 |
+
"loss": 0.02742033004760742,
|
| 1307 |
+
"step": 17500
|
| 1308 |
+
},
|
| 1309 |
+
{
|
| 1310 |
+
"epoch": 1.0130249790153116,
|
| 1311 |
+
"eval_accuracy": 0.9893659906022149,
|
| 1312 |
+
"eval_f1": 0.8761479591836735,
|
| 1313 |
+
"eval_loss": 0.04772739112377167,
|
| 1314 |
+
"eval_precision": 0.8623330320377623,
|
| 1315 |
+
"eval_recall": 0.8904127346261537,
|
| 1316 |
+
"eval_runtime": 32.383,
|
| 1317 |
+
"eval_samples_per_second": 386.005,
|
| 1318 |
+
"eval_steps_per_second": 10.746,
|
| 1319 |
+
"step": 17500
|
| 1320 |
+
},
|
| 1321 |
+
{
|
| 1322 |
+
"epoch": 1.0188138585776723,
|
| 1323 |
+
"grad_norm": 1.3907451629638672,
|
| 1324 |
+
"learning_rate": 2.453111432706223e-05,
|
| 1325 |
+
"loss": 0.026595606803894042,
|
| 1326 |
+
"step": 17600
|
| 1327 |
+
},
|
| 1328 |
+
{
|
| 1329 |
+
"epoch": 1.0246027381400329,
|
| 1330 |
+
"grad_norm": 0.15164747834205627,
|
| 1331 |
+
"learning_rate": 2.4386396526772795e-05,
|
| 1332 |
+
"loss": 0.024490804672241212,
|
| 1333 |
+
"step": 17700
|
| 1334 |
+
},
|
| 1335 |
+
{
|
| 1336 |
+
"epoch": 1.0303916177023937,
|
| 1337 |
+
"grad_norm": 0.44126424193382263,
|
| 1338 |
+
"learning_rate": 2.424167872648336e-05,
|
| 1339 |
+
"loss": 0.023133702278137207,
|
| 1340 |
+
"step": 17800
|
| 1341 |
+
},
|
| 1342 |
+
{
|
| 1343 |
+
"epoch": 1.0361804972647544,
|
| 1344 |
+
"grad_norm": 0.14820177853107452,
|
| 1345 |
+
"learning_rate": 2.4096960926193923e-05,
|
| 1346 |
+
"loss": 0.02218240737915039,
|
| 1347 |
+
"step": 17900
|
| 1348 |
+
},
|
| 1349 |
+
{
|
| 1350 |
+
"epoch": 1.0419693768271152,
|
| 1351 |
+
"grad_norm": 0.5905808210372925,
|
| 1352 |
+
"learning_rate": 2.3952243125904487e-05,
|
| 1353 |
+
"loss": 0.020209569931030274,
|
| 1354 |
+
"step": 18000
|
| 1355 |
+
},
|
| 1356 |
+
{
|
| 1357 |
+
"epoch": 1.0477582563894758,
|
| 1358 |
+
"grad_norm": 0.17811790108680725,
|
| 1359 |
+
"learning_rate": 2.380752532561505e-05,
|
| 1360 |
+
"loss": 0.023463377952575682,
|
| 1361 |
+
"step": 18100
|
| 1362 |
+
},
|
| 1363 |
+
{
|
| 1364 |
+
"epoch": 1.0535471359518365,
|
| 1365 |
+
"grad_norm": 0.18444955348968506,
|
| 1366 |
+
"learning_rate": 2.3662807525325618e-05,
|
| 1367 |
+
"loss": 0.02178111791610718,
|
| 1368 |
+
"step": 18200
|
| 1369 |
+
},
|
| 1370 |
+
{
|
| 1371 |
+
"epoch": 1.0593360155141973,
|
| 1372 |
+
"grad_norm": 0.3301508128643036,
|
| 1373 |
+
"learning_rate": 2.351808972503618e-05,
|
| 1374 |
+
"loss": 0.02543956518173218,
|
| 1375 |
+
"step": 18300
|
| 1376 |
+
},
|
| 1377 |
+
{
|
| 1378 |
+
"epoch": 1.065124895076558,
|
| 1379 |
+
"grad_norm": 0.1861371099948883,
|
| 1380 |
+
"learning_rate": 2.3373371924746746e-05,
|
| 1381 |
+
"loss": 0.024148540496826174,
|
| 1382 |
+
"step": 18400
|
| 1383 |
+
},
|
| 1384 |
+
{
|
| 1385 |
+
"epoch": 1.0709137746389186,
|
| 1386 |
+
"grad_norm": 0.3149067163467407,
|
| 1387 |
+
"learning_rate": 2.3228654124457306e-05,
|
| 1388 |
+
"loss": 0.023476483821868895,
|
| 1389 |
+
"step": 18500
|
| 1390 |
+
},
|
| 1391 |
+
{
|
| 1392 |
+
"epoch": 1.0767026542012794,
|
| 1393 |
+
"grad_norm": 0.17988641560077667,
|
| 1394 |
+
"learning_rate": 2.3083936324167874e-05,
|
| 1395 |
+
"loss": 0.02149003028869629,
|
| 1396 |
+
"step": 18600
|
| 1397 |
+
},
|
| 1398 |
+
{
|
| 1399 |
+
"epoch": 1.08249153376364,
|
| 1400 |
+
"grad_norm": 0.1312856823205948,
|
| 1401 |
+
"learning_rate": 2.2939218523878438e-05,
|
| 1402 |
+
"loss": 0.023462765216827393,
|
| 1403 |
+
"step": 18700
|
| 1404 |
+
},
|
| 1405 |
+
{
|
| 1406 |
+
"epoch": 1.0882804133260007,
|
| 1407 |
+
"grad_norm": 0.20052380859851837,
|
| 1408 |
+
"learning_rate": 2.2794500723589e-05,
|
| 1409 |
+
"loss": 0.025409071445465087,
|
| 1410 |
+
"step": 18800
|
| 1411 |
+
},
|
| 1412 |
+
{
|
| 1413 |
+
"epoch": 1.0940692928883615,
|
| 1414 |
+
"grad_norm": 0.2507052421569824,
|
| 1415 |
+
"learning_rate": 2.264978292329957e-05,
|
| 1416 |
+
"loss": 0.026879096031188966,
|
| 1417 |
+
"step": 18900
|
| 1418 |
+
},
|
| 1419 |
+
{
|
| 1420 |
+
"epoch": 1.0998581724507221,
|
| 1421 |
+
"grad_norm": 0.15323978662490845,
|
| 1422 |
+
"learning_rate": 2.250506512301013e-05,
|
| 1423 |
+
"loss": 0.023210296630859373,
|
| 1424 |
+
"step": 19000
|
| 1425 |
+
},
|
| 1426 |
+
{
|
| 1427 |
+
"epoch": 1.1056470520130828,
|
| 1428 |
+
"grad_norm": 0.1345296949148178,
|
| 1429 |
+
"learning_rate": 2.2360347322720697e-05,
|
| 1430 |
+
"loss": 0.020816826820373537,
|
| 1431 |
+
"step": 19100
|
| 1432 |
+
},
|
| 1433 |
+
{
|
| 1434 |
+
"epoch": 1.1114359315754436,
|
| 1435 |
+
"grad_norm": 0.2258956879377365,
|
| 1436 |
+
"learning_rate": 2.221562952243126e-05,
|
| 1437 |
+
"loss": 0.023039839267730712,
|
| 1438 |
+
"step": 19200
|
| 1439 |
+
},
|
| 1440 |
+
{
|
| 1441 |
+
"epoch": 1.1172248111378043,
|
| 1442 |
+
"grad_norm": 0.17817442119121552,
|
| 1443 |
+
"learning_rate": 2.2070911722141824e-05,
|
| 1444 |
+
"loss": 0.02491870164871216,
|
| 1445 |
+
"step": 19300
|
| 1446 |
+
},
|
| 1447 |
+
{
|
| 1448 |
+
"epoch": 1.1230136907001649,
|
| 1449 |
+
"grad_norm": 0.23916760087013245,
|
| 1450 |
+
"learning_rate": 2.192619392185239e-05,
|
| 1451 |
+
"loss": 0.02417280673980713,
|
| 1452 |
+
"step": 19400
|
| 1453 |
+
},
|
| 1454 |
+
{
|
| 1455 |
+
"epoch": 1.1288025702625257,
|
| 1456 |
+
"grad_norm": 0.5979987978935242,
|
| 1457 |
+
"learning_rate": 2.1781476121562952e-05,
|
| 1458 |
+
"loss": 0.025201258659362794,
|
| 1459 |
+
"step": 19500
|
| 1460 |
+
},
|
| 1461 |
+
{
|
| 1462 |
+
"epoch": 1.1345914498248864,
|
| 1463 |
+
"grad_norm": 0.7608785033226013,
|
| 1464 |
+
"learning_rate": 2.1636758321273516e-05,
|
| 1465 |
+
"loss": 0.02450251579284668,
|
| 1466 |
+
"step": 19600
|
| 1467 |
+
},
|
| 1468 |
+
{
|
| 1469 |
+
"epoch": 1.1403803293872472,
|
| 1470 |
+
"grad_norm": 1.2524911165237427,
|
| 1471 |
+
"learning_rate": 2.1492040520984084e-05,
|
| 1472 |
+
"loss": 0.02275413990020752,
|
| 1473 |
+
"step": 19700
|
| 1474 |
+
},
|
| 1475 |
+
{
|
| 1476 |
+
"epoch": 1.1461692089496078,
|
| 1477 |
+
"grad_norm": 0.4412006139755249,
|
| 1478 |
+
"learning_rate": 2.1347322720694647e-05,
|
| 1479 |
+
"loss": 0.023493762016296386,
|
| 1480 |
+
"step": 19800
|
| 1481 |
+
},
|
| 1482 |
+
{
|
| 1483 |
+
"epoch": 1.1519580885119685,
|
| 1484 |
+
"grad_norm": 0.1437130719423294,
|
| 1485 |
+
"learning_rate": 2.120260492040521e-05,
|
| 1486 |
+
"loss": 0.024696297645568847,
|
| 1487 |
+
"step": 19900
|
| 1488 |
+
},
|
| 1489 |
+
{
|
| 1490 |
+
"epoch": 1.1577469680743293,
|
| 1491 |
+
"grad_norm": 0.6046096086502075,
|
| 1492 |
+
"learning_rate": 2.1057887120115775e-05,
|
| 1493 |
+
"loss": 0.023616561889648436,
|
| 1494 |
+
"step": 20000
|
| 1495 |
+
},
|
| 1496 |
+
{
|
| 1497 |
+
"epoch": 1.1577469680743293,
|
| 1498 |
+
"eval_accuracy": 0.9893949323074188,
|
| 1499 |
+
"eval_f1": 0.8802002835647409,
|
| 1500 |
+
"eval_loss": 0.04831967130303383,
|
| 1501 |
+
"eval_precision": 0.8675126520129919,
|
| 1502 |
+
"eval_recall": 0.8932645442289744,
|
| 1503 |
+
"eval_runtime": 32.3669,
|
| 1504 |
+
"eval_samples_per_second": 386.197,
|
| 1505 |
+
"eval_steps_per_second": 10.752,
|
| 1506 |
+
"step": 20000
|
| 1507 |
+
},
|
| 1508 |
+
{
|
| 1509 |
+
"epoch": 1.16353584763669,
|
| 1510 |
+
"grad_norm": 0.3236995339393616,
|
| 1511 |
+
"learning_rate": 2.091316931982634e-05,
|
| 1512 |
+
"loss": 0.024802815914154053,
|
| 1513 |
+
"step": 20100
|
| 1514 |
+
},
|
| 1515 |
+
{
|
| 1516 |
+
"epoch": 1.1693247271990506,
|
| 1517 |
+
"grad_norm": 1.4749183654785156,
|
| 1518 |
+
"learning_rate": 2.0768451519536907e-05,
|
| 1519 |
+
"loss": 0.0224468994140625,
|
| 1520 |
+
"step": 20200
|
| 1521 |
+
},
|
| 1522 |
+
{
|
| 1523 |
+
"epoch": 1.1751136067614114,
|
| 1524 |
+
"grad_norm": 0.7911006212234497,
|
| 1525 |
+
"learning_rate": 2.0623733719247467e-05,
|
| 1526 |
+
"loss": 0.021509246826171877,
|
| 1527 |
+
"step": 20300
|
| 1528 |
+
},
|
| 1529 |
+
{
|
| 1530 |
+
"epoch": 1.180902486323772,
|
| 1531 |
+
"grad_norm": 0.23929397761821747,
|
| 1532 |
+
"learning_rate": 2.0479015918958034e-05,
|
| 1533 |
+
"loss": 0.02354356527328491,
|
| 1534 |
+
"step": 20400
|
| 1535 |
+
},
|
| 1536 |
+
{
|
| 1537 |
+
"epoch": 1.1866913658861327,
|
| 1538 |
+
"grad_norm": 0.28904756903648376,
|
| 1539 |
+
"learning_rate": 2.0334298118668595e-05,
|
| 1540 |
+
"loss": 0.021830506324768066,
|
| 1541 |
+
"step": 20500
|
| 1542 |
+
},
|
| 1543 |
+
{
|
| 1544 |
+
"epoch": 1.1924802454484935,
|
| 1545 |
+
"grad_norm": 0.24184764921665192,
|
| 1546 |
+
"learning_rate": 2.0189580318379162e-05,
|
| 1547 |
+
"loss": 0.022603347301483154,
|
| 1548 |
+
"step": 20600
|
| 1549 |
+
},
|
| 1550 |
+
{
|
| 1551 |
+
"epoch": 1.1982691250108541,
|
| 1552 |
+
"grad_norm": 0.9602212905883789,
|
| 1553 |
+
"learning_rate": 2.0044862518089726e-05,
|
| 1554 |
+
"loss": 0.024089906215667724,
|
| 1555 |
+
"step": 20700
|
| 1556 |
+
},
|
| 1557 |
+
{
|
| 1558 |
+
"epoch": 1.2040580045732148,
|
| 1559 |
+
"grad_norm": 0.18168984353542328,
|
| 1560 |
+
"learning_rate": 1.990014471780029e-05,
|
| 1561 |
+
"loss": 0.02312016725540161,
|
| 1562 |
+
"step": 20800
|
| 1563 |
+
},
|
| 1564 |
+
{
|
| 1565 |
+
"epoch": 1.2098468841355756,
|
| 1566 |
+
"grad_norm": 0.5313889384269714,
|
| 1567 |
+
"learning_rate": 1.9755426917510854e-05,
|
| 1568 |
+
"loss": 0.022191174030303955,
|
| 1569 |
+
"step": 20900
|
| 1570 |
+
},
|
| 1571 |
+
{
|
| 1572 |
+
"epoch": 1.2156357636979362,
|
| 1573 |
+
"grad_norm": 1.4984052181243896,
|
| 1574 |
+
"learning_rate": 1.9610709117221418e-05,
|
| 1575 |
+
"loss": 0.020048980712890626,
|
| 1576 |
+
"step": 21000
|
| 1577 |
+
},
|
| 1578 |
+
{
|
| 1579 |
+
"epoch": 1.2214246432602969,
|
| 1580 |
+
"grad_norm": 0.9525922536849976,
|
| 1581 |
+
"learning_rate": 1.9465991316931985e-05,
|
| 1582 |
+
"loss": 0.023054752349853515,
|
| 1583 |
+
"step": 21100
|
| 1584 |
+
},
|
| 1585 |
+
{
|
| 1586 |
+
"epoch": 1.2272135228226577,
|
| 1587 |
+
"grad_norm": 0.20787988603115082,
|
| 1588 |
+
"learning_rate": 1.932127351664255e-05,
|
| 1589 |
+
"loss": 0.02273550271987915,
|
| 1590 |
+
"step": 21200
|
| 1591 |
+
},
|
| 1592 |
+
{
|
| 1593 |
+
"epoch": 1.2330024023850183,
|
| 1594 |
+
"grad_norm": 0.19591836631298065,
|
| 1595 |
+
"learning_rate": 1.9176555716353113e-05,
|
| 1596 |
+
"loss": 0.02390467405319214,
|
| 1597 |
+
"step": 21300
|
| 1598 |
+
},
|
| 1599 |
+
{
|
| 1600 |
+
"epoch": 1.2387912819473792,
|
| 1601 |
+
"grad_norm": 0.1931363046169281,
|
| 1602 |
+
"learning_rate": 1.9031837916063677e-05,
|
| 1603 |
+
"loss": 0.023615493774414062,
|
| 1604 |
+
"step": 21400
|
| 1605 |
+
},
|
| 1606 |
+
{
|
| 1607 |
+
"epoch": 1.2445801615097398,
|
| 1608 |
+
"grad_norm": 0.16858945786952972,
|
| 1609 |
+
"learning_rate": 1.888712011577424e-05,
|
| 1610 |
+
"loss": 0.02515049457550049,
|
| 1611 |
+
"step": 21500
|
| 1612 |
+
},
|
| 1613 |
+
{
|
| 1614 |
+
"epoch": 1.2503690410721005,
|
| 1615 |
+
"grad_norm": 0.1835613250732422,
|
| 1616 |
+
"learning_rate": 1.8742402315484805e-05,
|
| 1617 |
+
"loss": 0.02049088478088379,
|
| 1618 |
+
"step": 21600
|
| 1619 |
+
},
|
| 1620 |
+
{
|
| 1621 |
+
"epoch": 1.256157920634461,
|
| 1622 |
+
"grad_norm": 1.2275546789169312,
|
| 1623 |
+
"learning_rate": 1.8597684515195372e-05,
|
| 1624 |
+
"loss": 0.022141406536102293,
|
| 1625 |
+
"step": 21700
|
| 1626 |
+
},
|
| 1627 |
+
{
|
| 1628 |
+
"epoch": 1.261946800196822,
|
| 1629 |
+
"grad_norm": 1.089022159576416,
|
| 1630 |
+
"learning_rate": 1.8452966714905933e-05,
|
| 1631 |
+
"loss": 0.022219130992889403,
|
| 1632 |
+
"step": 21800
|
| 1633 |
+
},
|
| 1634 |
+
{
|
| 1635 |
+
"epoch": 1.2677356797591826,
|
| 1636 |
+
"grad_norm": 0.32157450914382935,
|
| 1637 |
+
"learning_rate": 1.83082489146165e-05,
|
| 1638 |
+
"loss": 0.025665369033813477,
|
| 1639 |
+
"step": 21900
|
| 1640 |
+
},
|
| 1641 |
+
{
|
| 1642 |
+
"epoch": 1.2735245593215434,
|
| 1643 |
+
"grad_norm": 0.28916555643081665,
|
| 1644 |
+
"learning_rate": 1.816353111432706e-05,
|
| 1645 |
+
"loss": 0.022141153812408446,
|
| 1646 |
+
"step": 22000
|
| 1647 |
+
},
|
| 1648 |
+
{
|
| 1649 |
+
"epoch": 1.279313438883904,
|
| 1650 |
+
"grad_norm": 0.4547838866710663,
|
| 1651 |
+
"learning_rate": 1.8018813314037628e-05,
|
| 1652 |
+
"loss": 0.02140897512435913,
|
| 1653 |
+
"step": 22100
|
| 1654 |
+
},
|
| 1655 |
+
{
|
| 1656 |
+
"epoch": 1.2851023184462647,
|
| 1657 |
+
"grad_norm": 0.2740997076034546,
|
| 1658 |
+
"learning_rate": 1.787409551374819e-05,
|
| 1659 |
+
"loss": 0.01996962308883667,
|
| 1660 |
+
"step": 22200
|
| 1661 |
+
},
|
| 1662 |
+
{
|
| 1663 |
+
"epoch": 1.2908911980086255,
|
| 1664 |
+
"grad_norm": 0.17777739465236664,
|
| 1665 |
+
"learning_rate": 1.7729377713458756e-05,
|
| 1666 |
+
"loss": 0.020210413932800292,
|
| 1667 |
+
"step": 22300
|
| 1668 |
+
},
|
| 1669 |
+
{
|
| 1670 |
+
"epoch": 1.2966800775709861,
|
| 1671 |
+
"grad_norm": 0.2631765604019165,
|
| 1672 |
+
"learning_rate": 1.7584659913169323e-05,
|
| 1673 |
+
"loss": 0.02249443531036377,
|
| 1674 |
+
"step": 22400
|
| 1675 |
+
},
|
| 1676 |
+
{
|
| 1677 |
+
"epoch": 1.3024689571333468,
|
| 1678 |
+
"grad_norm": 2.136570453643799,
|
| 1679 |
+
"learning_rate": 1.7439942112879883e-05,
|
| 1680 |
+
"loss": 0.02353119134902954,
|
| 1681 |
+
"step": 22500
|
| 1682 |
+
},
|
| 1683 |
+
{
|
| 1684 |
+
"epoch": 1.3024689571333468,
|
| 1685 |
+
"eval_accuracy": 0.9900812641736834,
|
| 1686 |
+
"eval_f1": 0.8825427722569409,
|
| 1687 |
+
"eval_loss": 0.046070199459791183,
|
| 1688 |
+
"eval_precision": 0.8720012146978439,
|
| 1689 |
+
"eval_recall": 0.8933423208545058,
|
| 1690 |
+
"eval_runtime": 32.9824,
|
| 1691 |
+
"eval_samples_per_second": 378.99,
|
| 1692 |
+
"eval_steps_per_second": 10.551,
|
| 1693 |
+
"step": 22500
|
| 1694 |
+
},
|
| 1695 |
+
{
|
| 1696 |
+
"epoch": 1.3082578366957076,
|
| 1697 |
+
"grad_norm": 0.16276246309280396,
|
| 1698 |
+
"learning_rate": 1.729522431259045e-05,
|
| 1699 |
+
"loss": 0.021528096199035646,
|
| 1700 |
+
"step": 22600
|
| 1701 |
+
},
|
| 1702 |
+
{
|
| 1703 |
+
"epoch": 1.3140467162580682,
|
| 1704 |
+
"grad_norm": 0.1779128611087799,
|
| 1705 |
+
"learning_rate": 1.7150506512301015e-05,
|
| 1706 |
+
"loss": 0.021381580829620363,
|
| 1707 |
+
"step": 22700
|
| 1708 |
+
},
|
| 1709 |
+
{
|
| 1710 |
+
"epoch": 1.3198355958204289,
|
| 1711 |
+
"grad_norm": 0.12842315435409546,
|
| 1712 |
+
"learning_rate": 1.700578871201158e-05,
|
| 1713 |
+
"loss": 0.022963883876800536,
|
| 1714 |
+
"step": 22800
|
| 1715 |
+
},
|
| 1716 |
+
{
|
| 1717 |
+
"epoch": 1.3256244753827897,
|
| 1718 |
+
"grad_norm": 0.38589125871658325,
|
| 1719 |
+
"learning_rate": 1.6861070911722142e-05,
|
| 1720 |
+
"loss": 0.025010223388671874,
|
| 1721 |
+
"step": 22900
|
| 1722 |
+
},
|
| 1723 |
+
{
|
| 1724 |
+
"epoch": 1.3314133549451503,
|
| 1725 |
+
"grad_norm": 0.17700180411338806,
|
| 1726 |
+
"learning_rate": 1.6716353111432706e-05,
|
| 1727 |
+
"loss": 0.025014562606811522,
|
| 1728 |
+
"step": 23000
|
| 1729 |
+
},
|
| 1730 |
+
{
|
| 1731 |
+
"epoch": 1.3372022345075112,
|
| 1732 |
+
"grad_norm": 0.1977829933166504,
|
| 1733 |
+
"learning_rate": 1.657163531114327e-05,
|
| 1734 |
+
"loss": 0.019888348579406738,
|
| 1735 |
+
"step": 23100
|
| 1736 |
+
},
|
| 1737 |
+
{
|
| 1738 |
+
"epoch": 1.3429911140698718,
|
| 1739 |
+
"grad_norm": 0.21066619455814362,
|
| 1740 |
+
"learning_rate": 1.6426917510853838e-05,
|
| 1741 |
+
"loss": 0.021409926414489747,
|
| 1742 |
+
"step": 23200
|
| 1743 |
+
},
|
| 1744 |
+
{
|
| 1745 |
+
"epoch": 1.3487799936322324,
|
| 1746 |
+
"grad_norm": 0.1378607600927353,
|
| 1747 |
+
"learning_rate": 1.6282199710564398e-05,
|
| 1748 |
+
"loss": 0.02372417688369751,
|
| 1749 |
+
"step": 23300
|
| 1750 |
+
},
|
| 1751 |
+
{
|
| 1752 |
+
"epoch": 1.354568873194593,
|
| 1753 |
+
"grad_norm": 0.3123644292354584,
|
| 1754 |
+
"learning_rate": 1.6137481910274965e-05,
|
| 1755 |
+
"loss": 0.021678264141082763,
|
| 1756 |
+
"step": 23400
|
| 1757 |
+
},
|
| 1758 |
+
{
|
| 1759 |
+
"epoch": 1.360357752756954,
|
| 1760 |
+
"grad_norm": 0.19485028088092804,
|
| 1761 |
+
"learning_rate": 1.599276410998553e-05,
|
| 1762 |
+
"loss": 0.024895586967468263,
|
| 1763 |
+
"step": 23500
|
| 1764 |
+
},
|
| 1765 |
+
{
|
| 1766 |
+
"epoch": 1.3661466323193145,
|
| 1767 |
+
"grad_norm": 0.2177608162164688,
|
| 1768 |
+
"learning_rate": 1.5848046309696093e-05,
|
| 1769 |
+
"loss": 0.020630009174346924,
|
| 1770 |
+
"step": 23600
|
| 1771 |
+
},
|
| 1772 |
+
{
|
| 1773 |
+
"epoch": 1.3719355118816754,
|
| 1774 |
+
"grad_norm": 1.222109317779541,
|
| 1775 |
+
"learning_rate": 1.570332850940666e-05,
|
| 1776 |
+
"loss": 0.022696774005889892,
|
| 1777 |
+
"step": 23700
|
| 1778 |
+
},
|
| 1779 |
+
{
|
| 1780 |
+
"epoch": 1.377724391444036,
|
| 1781 |
+
"grad_norm": 0.5391749143600464,
|
| 1782 |
+
"learning_rate": 1.555861070911722e-05,
|
| 1783 |
+
"loss": 0.019349946975708007,
|
| 1784 |
+
"step": 23800
|
| 1785 |
+
},
|
| 1786 |
+
{
|
| 1787 |
+
"epoch": 1.3835132710063966,
|
| 1788 |
+
"grad_norm": 0.8667725920677185,
|
| 1789 |
+
"learning_rate": 1.541389290882779e-05,
|
| 1790 |
+
"loss": 0.021944737434387206,
|
| 1791 |
+
"step": 23900
|
| 1792 |
+
},
|
| 1793 |
+
{
|
| 1794 |
+
"epoch": 1.3893021505687575,
|
| 1795 |
+
"grad_norm": 0.2695491909980774,
|
| 1796 |
+
"learning_rate": 1.526917510853835e-05,
|
| 1797 |
+
"loss": 0.023090031147003174,
|
| 1798 |
+
"step": 24000
|
| 1799 |
+
},
|
| 1800 |
+
{
|
| 1801 |
+
"epoch": 1.3950910301311181,
|
| 1802 |
+
"grad_norm": 0.470904678106308,
|
| 1803 |
+
"learning_rate": 1.5124457308248916e-05,
|
| 1804 |
+
"loss": 0.023335537910461425,
|
| 1805 |
+
"step": 24100
|
| 1806 |
+
},
|
| 1807 |
+
{
|
| 1808 |
+
"epoch": 1.4008799096934788,
|
| 1809 |
+
"grad_norm": 0.12368060648441315,
|
| 1810 |
+
"learning_rate": 1.4979739507959482e-05,
|
| 1811 |
+
"loss": 0.020203118324279786,
|
| 1812 |
+
"step": 24200
|
| 1813 |
+
},
|
| 1814 |
+
{
|
| 1815 |
+
"epoch": 1.4066687892558396,
|
| 1816 |
+
"grad_norm": 0.8601803779602051,
|
| 1817 |
+
"learning_rate": 1.4835021707670044e-05,
|
| 1818 |
+
"loss": 0.024733288288116453,
|
| 1819 |
+
"step": 24300
|
| 1820 |
+
},
|
| 1821 |
+
{
|
| 1822 |
+
"epoch": 1.4124576688182002,
|
| 1823 |
+
"grad_norm": 0.12211176753044128,
|
| 1824 |
+
"learning_rate": 1.469030390738061e-05,
|
| 1825 |
+
"loss": 0.0239523983001709,
|
| 1826 |
+
"step": 24400
|
| 1827 |
+
},
|
| 1828 |
+
{
|
| 1829 |
+
"epoch": 1.4182465483805609,
|
| 1830 |
+
"grad_norm": 0.15338654816150665,
|
| 1831 |
+
"learning_rate": 1.4545586107091172e-05,
|
| 1832 |
+
"loss": 0.02348433494567871,
|
| 1833 |
+
"step": 24500
|
| 1834 |
+
},
|
| 1835 |
+
{
|
| 1836 |
+
"epoch": 1.4240354279429217,
|
| 1837 |
+
"grad_norm": 0.14509518444538116,
|
| 1838 |
+
"learning_rate": 1.4400868306801738e-05,
|
| 1839 |
+
"loss": 0.024761919975280763,
|
| 1840 |
+
"step": 24600
|
| 1841 |
+
},
|
| 1842 |
+
{
|
| 1843 |
+
"epoch": 1.4298243075052823,
|
| 1844 |
+
"grad_norm": 0.3267483115196228,
|
| 1845 |
+
"learning_rate": 1.4256150506512303e-05,
|
| 1846 |
+
"loss": 0.02035109281539917,
|
| 1847 |
+
"step": 24700
|
| 1848 |
+
},
|
| 1849 |
+
{
|
| 1850 |
+
"epoch": 1.4356131870676432,
|
| 1851 |
+
"grad_norm": 0.7865790128707886,
|
| 1852 |
+
"learning_rate": 1.4111432706222865e-05,
|
| 1853 |
+
"loss": 0.020213403701782227,
|
| 1854 |
+
"step": 24800
|
| 1855 |
+
},
|
| 1856 |
+
{
|
| 1857 |
+
"epoch": 1.4414020666300038,
|
| 1858 |
+
"grad_norm": 0.6819100975990295,
|
| 1859 |
+
"learning_rate": 1.3966714905933431e-05,
|
| 1860 |
+
"loss": 0.02440610408782959,
|
| 1861 |
+
"step": 24900
|
| 1862 |
+
},
|
| 1863 |
+
{
|
| 1864 |
+
"epoch": 1.4471909461923644,
|
| 1865 |
+
"grad_norm": 0.22851134836673737,
|
| 1866 |
+
"learning_rate": 1.3821997105643993e-05,
|
| 1867 |
+
"loss": 0.019472746849060057,
|
| 1868 |
+
"step": 25000
|
| 1869 |
+
},
|
| 1870 |
+
{
|
| 1871 |
+
"epoch": 1.4471909461923644,
|
| 1872 |
+
"eval_accuracy": 0.9902693852575088,
|
| 1873 |
+
"eval_f1": 0.8853341536568484,
|
| 1874 |
+
"eval_loss": 0.04390441253781319,
|
| 1875 |
+
"eval_precision": 0.8755260355929625,
|
| 1876 |
+
"eval_recall": 0.8953645131183242,
|
| 1877 |
+
"eval_runtime": 32.2884,
|
| 1878 |
+
"eval_samples_per_second": 387.136,
|
| 1879 |
+
"eval_steps_per_second": 10.778,
|
| 1880 |
+
"step": 25000
|
| 1881 |
+
},
|
| 1882 |
+
{
|
| 1883 |
+
"epoch": 1.452979825754725,
|
| 1884 |
+
"grad_norm": 0.3433781564235687,
|
| 1885 |
+
"learning_rate": 1.3677279305354559e-05,
|
| 1886 |
+
"loss": 0.021682107448577882,
|
| 1887 |
+
"step": 25100
|
| 1888 |
+
},
|
| 1889 |
+
{
|
| 1890 |
+
"epoch": 1.458768705317086,
|
| 1891 |
+
"grad_norm": 0.2263897955417633,
|
| 1892 |
+
"learning_rate": 1.3532561505065124e-05,
|
| 1893 |
+
"loss": 0.020724060535430907,
|
| 1894 |
+
"step": 25200
|
| 1895 |
+
},
|
| 1896 |
+
{
|
| 1897 |
+
"epoch": 1.4645575848794465,
|
| 1898 |
+
"grad_norm": 0.5709189772605896,
|
| 1899 |
+
"learning_rate": 1.3387843704775688e-05,
|
| 1900 |
+
"loss": 0.023897509574890136,
|
| 1901 |
+
"step": 25300
|
| 1902 |
+
},
|
| 1903 |
+
{
|
| 1904 |
+
"epoch": 1.4703464644418074,
|
| 1905 |
+
"grad_norm": 0.13530676066875458,
|
| 1906 |
+
"learning_rate": 1.3243125904486254e-05,
|
| 1907 |
+
"loss": 0.022753942012786865,
|
| 1908 |
+
"step": 25400
|
| 1909 |
+
},
|
| 1910 |
+
{
|
| 1911 |
+
"epoch": 1.476135344004168,
|
| 1912 |
+
"grad_norm": 0.42096275091171265,
|
| 1913 |
+
"learning_rate": 1.3098408104196816e-05,
|
| 1914 |
+
"loss": 0.021938772201538087,
|
| 1915 |
+
"step": 25500
|
| 1916 |
+
},
|
| 1917 |
+
{
|
| 1918 |
+
"epoch": 1.4819242235665286,
|
| 1919 |
+
"grad_norm": 2.0374867916107178,
|
| 1920 |
+
"learning_rate": 1.2953690303907382e-05,
|
| 1921 |
+
"loss": 0.023144633769989015,
|
| 1922 |
+
"step": 25600
|
| 1923 |
+
},
|
| 1924 |
+
{
|
| 1925 |
+
"epoch": 1.4877131031288893,
|
| 1926 |
+
"grad_norm": 0.22001959383487701,
|
| 1927 |
+
"learning_rate": 1.2808972503617947e-05,
|
| 1928 |
+
"loss": 0.021838405132293702,
|
| 1929 |
+
"step": 25700
|
| 1930 |
+
},
|
| 1931 |
+
{
|
| 1932 |
+
"epoch": 1.4935019826912501,
|
| 1933 |
+
"grad_norm": 0.2641585171222687,
|
| 1934 |
+
"learning_rate": 1.266425470332851e-05,
|
| 1935 |
+
"loss": 0.021520249843597412,
|
| 1936 |
+
"step": 25800
|
| 1937 |
+
},
|
| 1938 |
+
{
|
| 1939 |
+
"epoch": 1.4992908622536107,
|
| 1940 |
+
"grad_norm": 0.191110759973526,
|
| 1941 |
+
"learning_rate": 1.2519536903039075e-05,
|
| 1942 |
+
"loss": 0.01960566997528076,
|
| 1943 |
+
"step": 25900
|
| 1944 |
+
},
|
| 1945 |
+
{
|
| 1946 |
+
"epoch": 1.5050797418159716,
|
| 1947 |
+
"grad_norm": 1.1695928573608398,
|
| 1948 |
+
"learning_rate": 1.2374819102749639e-05,
|
| 1949 |
+
"loss": 0.022222683429718018,
|
| 1950 |
+
"step": 26000
|
| 1951 |
+
},
|
| 1952 |
+
{
|
| 1953 |
+
"epoch": 1.5108686213783322,
|
| 1954 |
+
"grad_norm": 0.19455936551094055,
|
| 1955 |
+
"learning_rate": 1.2230101302460203e-05,
|
| 1956 |
+
"loss": 0.020654516220092775,
|
| 1957 |
+
"step": 26100
|
| 1958 |
+
},
|
| 1959 |
+
{
|
| 1960 |
+
"epoch": 1.5166575009406928,
|
| 1961 |
+
"grad_norm": 0.15471012890338898,
|
| 1962 |
+
"learning_rate": 1.2085383502170767e-05,
|
| 1963 |
+
"loss": 0.022886896133422853,
|
| 1964 |
+
"step": 26200
|
| 1965 |
+
},
|
| 1966 |
+
{
|
| 1967 |
+
"epoch": 1.5224463805030535,
|
| 1968 |
+
"grad_norm": 1.4628580808639526,
|
| 1969 |
+
"learning_rate": 1.1940665701881331e-05,
|
| 1970 |
+
"loss": 0.02138796329498291,
|
| 1971 |
+
"step": 26300
|
| 1972 |
+
},
|
| 1973 |
+
{
|
| 1974 |
+
"epoch": 1.5282352600654143,
|
| 1975 |
+
"grad_norm": 0.49336832761764526,
|
| 1976 |
+
"learning_rate": 1.1795947901591896e-05,
|
| 1977 |
+
"loss": 0.021676077842712402,
|
| 1978 |
+
"step": 26400
|
| 1979 |
+
},
|
| 1980 |
+
{
|
| 1981 |
+
"epoch": 1.5340241396277752,
|
| 1982 |
+
"grad_norm": 0.18711985647678375,
|
| 1983 |
+
"learning_rate": 1.1651230101302462e-05,
|
| 1984 |
+
"loss": 0.022171633243560793,
|
| 1985 |
+
"step": 26500
|
| 1986 |
+
},
|
| 1987 |
+
{
|
| 1988 |
+
"epoch": 1.5398130191901358,
|
| 1989 |
+
"grad_norm": 0.197096049785614,
|
| 1990 |
+
"learning_rate": 1.1506512301013026e-05,
|
| 1991 |
+
"loss": 0.021774094104766845,
|
| 1992 |
+
"step": 26600
|
| 1993 |
+
},
|
| 1994 |
+
{
|
| 1995 |
+
"epoch": 1.5456018987524964,
|
| 1996 |
+
"grad_norm": 0.17335055768489838,
|
| 1997 |
+
"learning_rate": 1.136179450072359e-05,
|
| 1998 |
+
"loss": 0.019333064556121826,
|
| 1999 |
+
"step": 26700
|
| 2000 |
+
},
|
| 2001 |
+
{
|
| 2002 |
+
"epoch": 1.551390778314857,
|
| 2003 |
+
"grad_norm": 0.9639333486557007,
|
| 2004 |
+
"learning_rate": 1.1217076700434154e-05,
|
| 2005 |
+
"loss": 0.021504545211791994,
|
| 2006 |
+
"step": 26800
|
| 2007 |
+
},
|
| 2008 |
+
{
|
| 2009 |
+
"epoch": 1.557179657877218,
|
| 2010 |
+
"grad_norm": 0.7893505692481995,
|
| 2011 |
+
"learning_rate": 1.1072358900144718e-05,
|
| 2012 |
+
"loss": 0.01952719211578369,
|
| 2013 |
+
"step": 26900
|
| 2014 |
+
},
|
| 2015 |
+
{
|
| 2016 |
+
"epoch": 1.5629685374395785,
|
| 2017 |
+
"grad_norm": 0.11000892519950867,
|
| 2018 |
+
"learning_rate": 1.0927641099855283e-05,
|
| 2019 |
+
"loss": 0.023864905834197998,
|
| 2020 |
+
"step": 27000
|
| 2021 |
+
},
|
| 2022 |
+
{
|
| 2023 |
+
"epoch": 1.5687574170019394,
|
| 2024 |
+
"grad_norm": 1.0035220384597778,
|
| 2025 |
+
"learning_rate": 1.0782923299565847e-05,
|
| 2026 |
+
"loss": 0.022396252155303956,
|
| 2027 |
+
"step": 27100
|
| 2028 |
+
},
|
| 2029 |
+
{
|
| 2030 |
+
"epoch": 1.5745462965643,
|
| 2031 |
+
"grad_norm": 0.18122516572475433,
|
| 2032 |
+
"learning_rate": 1.0638205499276411e-05,
|
| 2033 |
+
"loss": 0.02149169921875,
|
| 2034 |
+
"step": 27200
|
| 2035 |
+
},
|
| 2036 |
+
{
|
| 2037 |
+
"epoch": 1.5803351761266606,
|
| 2038 |
+
"grad_norm": 0.4648451805114746,
|
| 2039 |
+
"learning_rate": 1.0493487698986975e-05,
|
| 2040 |
+
"loss": 0.02050316572189331,
|
| 2041 |
+
"step": 27300
|
| 2042 |
+
},
|
| 2043 |
+
{
|
| 2044 |
+
"epoch": 1.5861240556890213,
|
| 2045 |
+
"grad_norm": 0.7127793431282043,
|
| 2046 |
+
"learning_rate": 1.0348769898697539e-05,
|
| 2047 |
+
"loss": 0.021997590065002442,
|
| 2048 |
+
"step": 27400
|
| 2049 |
+
},
|
| 2050 |
+
{
|
| 2051 |
+
"epoch": 1.591912935251382,
|
| 2052 |
+
"grad_norm": 0.41411492228507996,
|
| 2053 |
+
"learning_rate": 1.0204052098408105e-05,
|
| 2054 |
+
"loss": 0.02217501640319824,
|
| 2055 |
+
"step": 27500
|
| 2056 |
+
},
|
| 2057 |
+
{
|
| 2058 |
+
"epoch": 1.591912935251382,
|
| 2059 |
+
"eval_accuracy": 0.9900791969090259,
|
| 2060 |
+
"eval_f1": 0.886322562452736,
|
| 2061 |
+
"eval_loss": 0.04419637843966484,
|
| 2062 |
+
"eval_precision": 0.8764925089360408,
|
| 2063 |
+
"eval_recall": 0.8963756092502333,
|
| 2064 |
+
"eval_runtime": 32.2485,
|
| 2065 |
+
"eval_samples_per_second": 387.615,
|
| 2066 |
+
"eval_steps_per_second": 10.791,
|
| 2067 |
+
"step": 27500
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 1.597701814813743,
|
| 2071 |
+
"grad_norm": 0.7603561878204346,
|
| 2072 |
+
"learning_rate": 1.005933429811867e-05,
|
| 2073 |
+
"loss": 0.019264650344848634,
|
| 2074 |
+
"step": 27600
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 1.6034906943761036,
|
| 2078 |
+
"grad_norm": 0.44727957248687744,
|
| 2079 |
+
"learning_rate": 9.914616497829234e-06,
|
| 2080 |
+
"loss": 0.024952189922332765,
|
| 2081 |
+
"step": 27700
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 1.6092795739384642,
|
| 2085 |
+
"grad_norm": 0.3509667217731476,
|
| 2086 |
+
"learning_rate": 9.769898697539798e-06,
|
| 2087 |
+
"loss": 0.02343012571334839,
|
| 2088 |
+
"step": 27800
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 1.6150684535008248,
|
| 2092 |
+
"grad_norm": 0.1554228961467743,
|
| 2093 |
+
"learning_rate": 9.625180897250362e-06,
|
| 2094 |
+
"loss": 0.022485718727111817,
|
| 2095 |
+
"step": 27900
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 1.6208573330631855,
|
| 2099 |
+
"grad_norm": 0.48960787057876587,
|
| 2100 |
+
"learning_rate": 9.480463096960928e-06,
|
| 2101 |
+
"loss": 0.022733292579650878,
|
| 2102 |
+
"step": 28000
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 1.6266462126255463,
|
| 2106 |
+
"grad_norm": 0.4882723093032837,
|
| 2107 |
+
"learning_rate": 9.335745296671492e-06,
|
| 2108 |
+
"loss": 0.02101508617401123,
|
| 2109 |
+
"step": 28100
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 1.6324350921879072,
|
| 2113 |
+
"grad_norm": 0.3346649408340454,
|
| 2114 |
+
"learning_rate": 9.191027496382055e-06,
|
| 2115 |
+
"loss": 0.02136120557785034,
|
| 2116 |
+
"step": 28200
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 1.6382239717502678,
|
| 2120 |
+
"grad_norm": 0.43525007367134094,
|
| 2121 |
+
"learning_rate": 9.04630969609262e-06,
|
| 2122 |
+
"loss": 0.019716554880142213,
|
| 2123 |
+
"step": 28300
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 1.6440128513126284,
|
| 2127 |
+
"grad_norm": 1.4369014501571655,
|
| 2128 |
+
"learning_rate": 8.901591895803183e-06,
|
| 2129 |
+
"loss": 0.021362924575805665,
|
| 2130 |
+
"step": 28400
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 1.649801730874989,
|
| 2134 |
+
"grad_norm": 0.36247485876083374,
|
| 2135 |
+
"learning_rate": 8.756874095513747e-06,
|
| 2136 |
+
"loss": 0.022539958953857422,
|
| 2137 |
+
"step": 28500
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 1.65559061043735,
|
| 2141 |
+
"grad_norm": 0.1396828442811966,
|
| 2142 |
+
"learning_rate": 8.612156295224313e-06,
|
| 2143 |
+
"loss": 0.01977320432662964,
|
| 2144 |
+
"step": 28600
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 1.6613794899997105,
|
| 2148 |
+
"grad_norm": 0.09776929765939713,
|
| 2149 |
+
"learning_rate": 8.467438494934877e-06,
|
| 2150 |
+
"loss": 0.017783541679382325,
|
| 2151 |
+
"step": 28700
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 1.6671683695620714,
|
| 2155 |
+
"grad_norm": 0.20341573655605316,
|
| 2156 |
+
"learning_rate": 8.322720694645442e-06,
|
| 2157 |
+
"loss": 0.01956522822380066,
|
| 2158 |
+
"step": 28800
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 1.672957249124432,
|
| 2162 |
+
"grad_norm": 0.24861733615398407,
|
| 2163 |
+
"learning_rate": 8.178002894356006e-06,
|
| 2164 |
+
"loss": 0.02077195405960083,
|
| 2165 |
+
"step": 28900
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 1.6787461286867926,
|
| 2169 |
+
"grad_norm": 0.14596840739250183,
|
| 2170 |
+
"learning_rate": 8.03328509406657e-06,
|
| 2171 |
+
"loss": 0.019236416816711427,
|
| 2172 |
+
"step": 29000
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 1.6845350082491533,
|
| 2176 |
+
"grad_norm": 0.1839938908815384,
|
| 2177 |
+
"learning_rate": 7.888567293777136e-06,
|
| 2178 |
+
"loss": 0.023349857330322264,
|
| 2179 |
+
"step": 29100
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 1.690323887811514,
|
| 2183 |
+
"grad_norm": 0.5248807668685913,
|
| 2184 |
+
"learning_rate": 7.7438494934877e-06,
|
| 2185 |
+
"loss": 0.020743234157562254,
|
| 2186 |
+
"step": 29200
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 1.696112767373875,
|
| 2190 |
+
"grad_norm": 0.23738259077072144,
|
| 2191 |
+
"learning_rate": 7.599131693198264e-06,
|
| 2192 |
+
"loss": 0.02022059917449951,
|
| 2193 |
+
"step": 29300
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 1.7019016469362356,
|
| 2197 |
+
"grad_norm": 0.6698647737503052,
|
| 2198 |
+
"learning_rate": 7.4544138929088275e-06,
|
| 2199 |
+
"loss": 0.01865126848220825,
|
| 2200 |
+
"step": 29400
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 1.7076905264985962,
|
| 2204 |
+
"grad_norm": 1.6622151136398315,
|
| 2205 |
+
"learning_rate": 7.309696092619392e-06,
|
| 2206 |
+
"loss": 0.020002634525299073,
|
| 2207 |
+
"step": 29500
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 1.7134794060609568,
|
| 2211 |
+
"grad_norm": 0.3179205358028412,
|
| 2212 |
+
"learning_rate": 7.164978292329957e-06,
|
| 2213 |
+
"loss": 0.021126785278320313,
|
| 2214 |
+
"step": 29600
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 1.7192682856233175,
|
| 2218 |
+
"grad_norm": 0.10927774757146835,
|
| 2219 |
+
"learning_rate": 7.020260492040522e-06,
|
| 2220 |
+
"loss": 0.018858327865600585,
|
| 2221 |
+
"step": 29700
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 1.7250571651856783,
|
| 2225 |
+
"grad_norm": 0.186599463224411,
|
| 2226 |
+
"learning_rate": 6.875542691751086e-06,
|
| 2227 |
+
"loss": 0.01988905429840088,
|
| 2228 |
+
"step": 29800
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 1.7308460447480392,
|
| 2232 |
+
"grad_norm": 0.22831496596336365,
|
| 2233 |
+
"learning_rate": 6.73082489146165e-06,
|
| 2234 |
+
"loss": 0.020202479362487792,
|
| 2235 |
+
"step": 29900
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 1.7366349243103998,
|
| 2239 |
+
"grad_norm": 0.3853333294391632,
|
| 2240 |
+
"learning_rate": 6.586107091172214e-06,
|
| 2241 |
+
"loss": 0.01944092631340027,
|
| 2242 |
+
"step": 30000
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 1.7366349243103998,
|
| 2246 |
+
"eval_accuracy": 0.9902431999051815,
|
| 2247 |
+
"eval_f1": 0.8897238929427909,
|
| 2248 |
+
"eval_loss": 0.043809473514556885,
|
| 2249 |
+
"eval_precision": 0.8803197563760944,
|
| 2250 |
+
"eval_recall": 0.8993311210204293,
|
| 2251 |
+
"eval_runtime": 32.3718,
|
| 2252 |
+
"eval_samples_per_second": 386.139,
|
| 2253 |
+
"eval_steps_per_second": 10.75,
|
| 2254 |
+
"step": 30000
|
| 2255 |
+
},
|
| 2256 |
+
{
|
| 2257 |
+
"epoch": 1.7424238038727604,
|
| 2258 |
+
"grad_norm": 0.5895107984542847,
|
| 2259 |
+
"learning_rate": 6.441389290882779e-06,
|
| 2260 |
+
"loss": 0.020335075855255128,
|
| 2261 |
+
"step": 30100
|
| 2262 |
+
},
|
| 2263 |
+
{
|
| 2264 |
+
"epoch": 1.748212683435121,
|
| 2265 |
+
"grad_norm": 0.10662174969911575,
|
| 2266 |
+
"learning_rate": 6.296671490593344e-06,
|
| 2267 |
+
"loss": 0.020258052349090575,
|
| 2268 |
+
"step": 30200
|
| 2269 |
+
},
|
| 2270 |
+
{
|
| 2271 |
+
"epoch": 1.7540015629974819,
|
| 2272 |
+
"grad_norm": 3.988093376159668,
|
| 2273 |
+
"learning_rate": 6.151953690303908e-06,
|
| 2274 |
+
"loss": 0.0205714750289917,
|
| 2275 |
+
"step": 30300
|
| 2276 |
+
},
|
| 2277 |
+
{
|
| 2278 |
+
"epoch": 1.7597904425598425,
|
| 2279 |
+
"grad_norm": 0.5093377828598022,
|
| 2280 |
+
"learning_rate": 6.007235890014472e-06,
|
| 2281 |
+
"loss": 0.020437090396881102,
|
| 2282 |
+
"step": 30400
|
| 2283 |
+
},
|
| 2284 |
+
{
|
| 2285 |
+
"epoch": 1.7655793221222034,
|
| 2286 |
+
"grad_norm": 0.29351383447647095,
|
| 2287 |
+
"learning_rate": 5.8625180897250366e-06,
|
| 2288 |
+
"loss": 0.01850534439086914,
|
| 2289 |
+
"step": 30500
|
| 2290 |
+
},
|
| 2291 |
+
{
|
| 2292 |
+
"epoch": 1.771368201684564,
|
| 2293 |
+
"grad_norm": 0.199212908744812,
|
| 2294 |
+
"learning_rate": 5.7178002894356005e-06,
|
| 2295 |
+
"loss": 0.02155954599380493,
|
| 2296 |
+
"step": 30600
|
| 2297 |
+
},
|
| 2298 |
+
{
|
| 2299 |
+
"epoch": 1.7771570812469246,
|
| 2300 |
+
"grad_norm": 0.346227765083313,
|
| 2301 |
+
"learning_rate": 5.573082489146165e-06,
|
| 2302 |
+
"loss": 0.019093557596206664,
|
| 2303 |
+
"step": 30700
|
| 2304 |
+
},
|
| 2305 |
+
{
|
| 2306 |
+
"epoch": 1.7829459608092852,
|
| 2307 |
+
"grad_norm": 0.28586652874946594,
|
| 2308 |
+
"learning_rate": 5.42836468885673e-06,
|
| 2309 |
+
"loss": 0.022267727851867675,
|
| 2310 |
+
"step": 30800
|
| 2311 |
+
},
|
| 2312 |
+
{
|
| 2313 |
+
"epoch": 1.788734840371646,
|
| 2314 |
+
"grad_norm": 0.16195492446422577,
|
| 2315 |
+
"learning_rate": 5.283646888567294e-06,
|
| 2316 |
+
"loss": 0.019566291570663454,
|
| 2317 |
+
"step": 30900
|
| 2318 |
+
},
|
| 2319 |
+
{
|
| 2320 |
+
"epoch": 1.7945237199340067,
|
| 2321 |
+
"grad_norm": 0.2971166968345642,
|
| 2322 |
+
"learning_rate": 5.138929088277859e-06,
|
| 2323 |
+
"loss": 0.01942929744720459,
|
| 2324 |
+
"step": 31000
|
| 2325 |
+
},
|
| 2326 |
+
{
|
| 2327 |
+
"epoch": 1.8003125994963676,
|
| 2328 |
+
"grad_norm": 0.17818105220794678,
|
| 2329 |
+
"learning_rate": 4.994211287988423e-06,
|
| 2330 |
+
"loss": 0.02038184642791748,
|
| 2331 |
+
"step": 31100
|
| 2332 |
+
},
|
| 2333 |
+
{
|
| 2334 |
+
"epoch": 1.8061014790587282,
|
| 2335 |
+
"grad_norm": 0.29902341961860657,
|
| 2336 |
+
"learning_rate": 4.849493487698987e-06,
|
| 2337 |
+
"loss": 0.019359084367752074,
|
| 2338 |
+
"step": 31200
|
| 2339 |
+
},
|
| 2340 |
+
{
|
| 2341 |
+
"epoch": 1.8118903586210888,
|
| 2342 |
+
"grad_norm": 0.25175297260284424,
|
| 2343 |
+
"learning_rate": 4.704775687409551e-06,
|
| 2344 |
+
"loss": 0.02133084535598755,
|
| 2345 |
+
"step": 31300
|
| 2346 |
+
},
|
| 2347 |
+
{
|
| 2348 |
+
"epoch": 1.8176792381834495,
|
| 2349 |
+
"grad_norm": 0.37808459997177124,
|
| 2350 |
+
"learning_rate": 4.560057887120116e-06,
|
| 2351 |
+
"loss": 0.01949509859085083,
|
| 2352 |
+
"step": 31400
|
| 2353 |
+
},
|
| 2354 |
+
{
|
| 2355 |
+
"epoch": 1.8234681177458103,
|
| 2356 |
+
"grad_norm": 0.6678466200828552,
|
| 2357 |
+
"learning_rate": 4.415340086830681e-06,
|
| 2358 |
+
"loss": 0.018650727272033693,
|
| 2359 |
+
"step": 31500
|
| 2360 |
+
},
|
| 2361 |
+
{
|
| 2362 |
+
"epoch": 1.8292569973081712,
|
| 2363 |
+
"grad_norm": 0.5661903619766235,
|
| 2364 |
+
"learning_rate": 4.270622286541245e-06,
|
| 2365 |
+
"loss": 0.020224461555480956,
|
| 2366 |
+
"step": 31600
|
| 2367 |
+
},
|
| 2368 |
+
{
|
| 2369 |
+
"epoch": 1.8350458768705318,
|
| 2370 |
+
"grad_norm": 0.295162558555603,
|
| 2371 |
+
"learning_rate": 4.125904486251809e-06,
|
| 2372 |
+
"loss": 0.0190382981300354,
|
| 2373 |
+
"step": 31700
|
| 2374 |
+
},
|
| 2375 |
+
{
|
| 2376 |
+
"epoch": 1.8408347564328924,
|
| 2377 |
+
"grad_norm": 0.5357158184051514,
|
| 2378 |
+
"learning_rate": 3.981186685962373e-06,
|
| 2379 |
+
"loss": 0.02294764518737793,
|
| 2380 |
+
"step": 31800
|
| 2381 |
+
},
|
| 2382 |
+
{
|
| 2383 |
+
"epoch": 1.846623635995253,
|
| 2384 |
+
"grad_norm": 0.2613368630409241,
|
| 2385 |
+
"learning_rate": 3.836468885672938e-06,
|
| 2386 |
+
"loss": 0.019658113718032836,
|
| 2387 |
+
"step": 31900
|
| 2388 |
+
},
|
| 2389 |
+
{
|
| 2390 |
+
"epoch": 1.8524125155576137,
|
| 2391 |
+
"grad_norm": 0.3176723122596741,
|
| 2392 |
+
"learning_rate": 3.6917510853835025e-06,
|
| 2393 |
+
"loss": 0.01859914779663086,
|
| 2394 |
+
"step": 32000
|
| 2395 |
+
},
|
| 2396 |
+
{
|
| 2397 |
+
"epoch": 1.8582013951199745,
|
| 2398 |
+
"grad_norm": 0.1181308925151825,
|
| 2399 |
+
"learning_rate": 3.547033285094067e-06,
|
| 2400 |
+
"loss": 0.02106498718261719,
|
| 2401 |
+
"step": 32100
|
| 2402 |
+
},
|
| 2403 |
+
{
|
| 2404 |
+
"epoch": 1.8639902746823354,
|
| 2405 |
+
"grad_norm": 0.6285836100578308,
|
| 2406 |
+
"learning_rate": 3.4023154848046308e-06,
|
| 2407 |
+
"loss": 0.019543524980545044,
|
| 2408 |
+
"step": 32200
|
| 2409 |
+
},
|
| 2410 |
+
{
|
| 2411 |
+
"epoch": 1.869779154244696,
|
| 2412 |
+
"grad_norm": 0.221060112118721,
|
| 2413 |
+
"learning_rate": 3.2575976845151955e-06,
|
| 2414 |
+
"loss": 0.019698797464370726,
|
| 2415 |
+
"step": 32300
|
| 2416 |
+
},
|
| 2417 |
+
{
|
| 2418 |
+
"epoch": 1.8755680338070566,
|
| 2419 |
+
"grad_norm": 0.46701422333717346,
|
| 2420 |
+
"learning_rate": 3.11287988422576e-06,
|
| 2421 |
+
"loss": 0.02131035327911377,
|
| 2422 |
+
"step": 32400
|
| 2423 |
+
},
|
| 2424 |
+
{
|
| 2425 |
+
"epoch": 1.8813569133694172,
|
| 2426 |
+
"grad_norm": 0.8717579245567322,
|
| 2427 |
+
"learning_rate": 2.968162083936324e-06,
|
| 2428 |
+
"loss": 0.0199610698223114,
|
| 2429 |
+
"step": 32500
|
| 2430 |
+
},
|
| 2431 |
+
{
|
| 2432 |
+
"epoch": 1.8813569133694172,
|
| 2433 |
+
"eval_accuracy": 0.9909171281835014,
|
| 2434 |
+
"eval_f1": 0.8929190695762744,
|
| 2435 |
+
"eval_loss": 0.04036634415388107,
|
| 2436 |
+
"eval_precision": 0.8848160875652284,
|
| 2437 |
+
"eval_recall": 0.9011718344913409,
|
| 2438 |
+
"eval_runtime": 32.8086,
|
| 2439 |
+
"eval_samples_per_second": 380.998,
|
| 2440 |
+
"eval_steps_per_second": 10.607,
|
| 2441 |
+
"step": 32500
|
| 2442 |
+
},
|
| 2443 |
+
{
|
| 2444 |
+
"epoch": 1.887145792931778,
|
| 2445 |
+
"grad_norm": 0.42211058735847473,
|
| 2446 |
+
"learning_rate": 2.8234442836468885e-06,
|
| 2447 |
+
"loss": 0.01953967571258545,
|
| 2448 |
+
"step": 32600
|
| 2449 |
+
},
|
| 2450 |
+
{
|
| 2451 |
+
"epoch": 1.8929346724941387,
|
| 2452 |
+
"grad_norm": 0.14807343482971191,
|
| 2453 |
+
"learning_rate": 2.6787264833574533e-06,
|
| 2454 |
+
"loss": 0.02133486270904541,
|
| 2455 |
+
"step": 32700
|
| 2456 |
+
},
|
| 2457 |
+
{
|
| 2458 |
+
"epoch": 1.8987235520564996,
|
| 2459 |
+
"grad_norm": 0.2488560974597931,
|
| 2460 |
+
"learning_rate": 2.5340086830680177e-06,
|
| 2461 |
+
"loss": 0.02061129570007324,
|
| 2462 |
+
"step": 32800
|
| 2463 |
+
},
|
| 2464 |
+
{
|
| 2465 |
+
"epoch": 1.9045124316188602,
|
| 2466 |
+
"grad_norm": 0.12760600447654724,
|
| 2467 |
+
"learning_rate": 2.3892908827785816e-06,
|
| 2468 |
+
"loss": 0.018405574560165405,
|
| 2469 |
+
"step": 32900
|
| 2470 |
+
},
|
| 2471 |
+
{
|
| 2472 |
+
"epoch": 1.9103013111812208,
|
| 2473 |
+
"grad_norm": 0.14860856533050537,
|
| 2474 |
+
"learning_rate": 2.2445730824891463e-06,
|
| 2475 |
+
"loss": 0.018643089532852174,
|
| 2476 |
+
"step": 33000
|
| 2477 |
+
},
|
| 2478 |
+
{
|
| 2479 |
+
"epoch": 1.9160901907435814,
|
| 2480 |
+
"grad_norm": 0.21985752880573273,
|
| 2481 |
+
"learning_rate": 2.0998552821997107e-06,
|
| 2482 |
+
"loss": 0.01826516628265381,
|
| 2483 |
+
"step": 33100
|
| 2484 |
+
},
|
| 2485 |
+
{
|
| 2486 |
+
"epoch": 1.9218790703059423,
|
| 2487 |
+
"grad_norm": 0.12793774902820587,
|
| 2488 |
+
"learning_rate": 1.955137481910275e-06,
|
| 2489 |
+
"loss": 0.018108237981796265,
|
| 2490 |
+
"step": 33200
|
| 2491 |
+
},
|
| 2492 |
+
{
|
| 2493 |
+
"epoch": 1.9276679498683031,
|
| 2494 |
+
"grad_norm": 0.6992884278297424,
|
| 2495 |
+
"learning_rate": 1.8104196816208394e-06,
|
| 2496 |
+
"loss": 0.018568735122680664,
|
| 2497 |
+
"step": 33300
|
| 2498 |
+
},
|
| 2499 |
+
{
|
| 2500 |
+
"epoch": 1.9334568294306638,
|
| 2501 |
+
"grad_norm": 0.16569721698760986,
|
| 2502 |
+
"learning_rate": 1.6657018813314037e-06,
|
| 2503 |
+
"loss": 0.020773212909698486,
|
| 2504 |
+
"step": 33400
|
| 2505 |
+
},
|
| 2506 |
+
{
|
| 2507 |
+
"epoch": 1.9392457089930244,
|
| 2508 |
+
"grad_norm": 0.2137010395526886,
|
| 2509 |
+
"learning_rate": 1.5209840810419682e-06,
|
| 2510 |
+
"loss": 0.020217933654785157,
|
| 2511 |
+
"step": 33500
|
| 2512 |
+
},
|
| 2513 |
+
{
|
| 2514 |
+
"epoch": 1.945034588555385,
|
| 2515 |
+
"grad_norm": 0.2512764036655426,
|
| 2516 |
+
"learning_rate": 1.3762662807525328e-06,
|
| 2517 |
+
"loss": 0.021186413764953612,
|
| 2518 |
+
"step": 33600
|
| 2519 |
+
},
|
| 2520 |
+
{
|
| 2521 |
+
"epoch": 1.9508234681177457,
|
| 2522 |
+
"grad_norm": 1.0330688953399658,
|
| 2523 |
+
"learning_rate": 1.231548480463097e-06,
|
| 2524 |
+
"loss": 0.02138917684555054,
|
| 2525 |
+
"step": 33700
|
| 2526 |
+
},
|
| 2527 |
+
{
|
| 2528 |
+
"epoch": 1.9566123476801065,
|
| 2529 |
+
"grad_norm": 0.3796994090080261,
|
| 2530 |
+
"learning_rate": 1.0868306801736615e-06,
|
| 2531 |
+
"loss": 0.017367198467254638,
|
| 2532 |
+
"step": 33800
|
| 2533 |
+
},
|
| 2534 |
+
{
|
| 2535 |
+
"epoch": 1.9624012272424673,
|
| 2536 |
+
"grad_norm": 0.35818561911582947,
|
| 2537 |
+
"learning_rate": 9.421128798842257e-07,
|
| 2538 |
+
"loss": 0.019348199367523192,
|
| 2539 |
+
"step": 33900
|
| 2540 |
+
},
|
| 2541 |
+
{
|
| 2542 |
+
"epoch": 1.968190106804828,
|
| 2543 |
+
"grad_norm": 0.4799649715423584,
|
| 2544 |
+
"learning_rate": 7.973950795947903e-07,
|
| 2545 |
+
"loss": 0.022617738246917724,
|
| 2546 |
+
"step": 34000
|
| 2547 |
+
},
|
| 2548 |
+
{
|
| 2549 |
+
"epoch": 1.9739789863671886,
|
| 2550 |
+
"grad_norm": 0.8583610653877258,
|
| 2551 |
+
"learning_rate": 6.526772793053546e-07,
|
| 2552 |
+
"loss": 0.019666857719421386,
|
| 2553 |
+
"step": 34100
|
| 2554 |
+
},
|
| 2555 |
+
{
|
| 2556 |
+
"epoch": 1.9797678659295492,
|
| 2557 |
+
"grad_norm": 0.6525281667709351,
|
| 2558 |
+
"learning_rate": 5.079594790159189e-07,
|
| 2559 |
+
"loss": 0.020464680194854735,
|
| 2560 |
+
"step": 34200
|
| 2561 |
+
},
|
| 2562 |
+
{
|
| 2563 |
+
"epoch": 1.98555674549191,
|
| 2564 |
+
"grad_norm": 1.2383188009262085,
|
| 2565 |
+
"learning_rate": 3.632416787264834e-07,
|
| 2566 |
+
"loss": 0.018764994144439696,
|
| 2567 |
+
"step": 34300
|
| 2568 |
+
},
|
| 2569 |
+
{
|
| 2570 |
+
"epoch": 1.9913456250542707,
|
| 2571 |
+
"grad_norm": 0.19601252675056458,
|
| 2572 |
+
"learning_rate": 2.1852387843704776e-07,
|
| 2573 |
+
"loss": 0.018834112882614135,
|
| 2574 |
+
"step": 34400
|
| 2575 |
+
},
|
| 2576 |
+
{
|
| 2577 |
+
"epoch": 1.9971345046166316,
|
| 2578 |
+
"grad_norm": 0.25431913137435913,
|
| 2579 |
+
"learning_rate": 7.380607814761216e-08,
|
| 2580 |
+
"loss": 0.018961663246154784,
|
| 2581 |
+
"step": 34500
|
| 2582 |
+
},
|
| 2583 |
+
{
|
| 2584 |
+
"epoch": 2.0,
|
| 2585 |
+
"step": 34550,
|
| 2586 |
+
"total_flos": 1.1376036548759528e+18,
|
| 2587 |
+
"train_loss": 0.03949730815487212,
|
| 2588 |
+
"train_runtime": 14944.1573,
|
| 2589 |
+
"train_samples_per_second": 332.906,
|
| 2590 |
+
"train_steps_per_second": 2.312
|
| 2591 |
+
}
|
| 2592 |
+
],
|
| 2593 |
+
"logging_steps": 100,
|
| 2594 |
+
"max_steps": 34550,
|
| 2595 |
+
"num_input_tokens_seen": 0,
|
| 2596 |
+
"num_train_epochs": 2,
|
| 2597 |
+
"save_steps": 2500,
|
| 2598 |
+
"stateful_callbacks": {
|
| 2599 |
+
"TrainerControl": {
|
| 2600 |
+
"args": {
|
| 2601 |
+
"should_epoch_stop": false,
|
| 2602 |
+
"should_evaluate": false,
|
| 2603 |
+
"should_log": false,
|
| 2604 |
+
"should_save": true,
|
| 2605 |
+
"should_training_stop": true
|
| 2606 |
+
},
|
| 2607 |
+
"attributes": {}
|
| 2608 |
+
}
|
| 2609 |
+
},
|
| 2610 |
+
"total_flos": 1.1376036548759528e+18,
|
| 2611 |
+
"train_batch_size": 72,
|
| 2612 |
+
"trial_name": null,
|
| 2613 |
+
"trial_params": null
|
| 2614 |
+
}
|