Model save
Browse files- trainer_state.json +3530 -0
trainer_state.json
ADDED
|
@@ -0,0 +1,3530 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_metric": null,
|
| 3 |
+
"best_model_checkpoint": null,
|
| 4 |
+
"epoch": 9.99000999000999,
|
| 5 |
+
"eval_steps": 500,
|
| 6 |
+
"global_step": 5000,
|
| 7 |
+
"is_hyper_param_search": false,
|
| 8 |
+
"is_local_process_zero": true,
|
| 9 |
+
"is_world_process_zero": true,
|
| 10 |
+
"log_history": [
|
| 11 |
+
{
|
| 12 |
+
"epoch": 0.01998001998001998,
|
| 13 |
+
"grad_norm": 1.9170171022415161,
|
| 14 |
+
"learning_rate": 0.0001999980260856137,
|
| 15 |
+
"loss": 1.4846,
|
| 16 |
+
"step": 10
|
| 17 |
+
},
|
| 18 |
+
{
|
| 19 |
+
"epoch": 0.03996003996003996,
|
| 20 |
+
"grad_norm": 0.5531741976737976,
|
| 21 |
+
"learning_rate": 0.00019999210442038162,
|
| 22 |
+
"loss": 1.0709,
|
| 23 |
+
"step": 20
|
| 24 |
+
},
|
| 25 |
+
{
|
| 26 |
+
"epoch": 0.059940059940059943,
|
| 27 |
+
"grad_norm": 0.39242061972618103,
|
| 28 |
+
"learning_rate": 0.0001999822352380809,
|
| 29 |
+
"loss": 0.9892,
|
| 30 |
+
"step": 30
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"epoch": 0.07992007992007992,
|
| 34 |
+
"grad_norm": 0.46874135732650757,
|
| 35 |
+
"learning_rate": 0.00019996841892833,
|
| 36 |
+
"loss": 0.9705,
|
| 37 |
+
"step": 40
|
| 38 |
+
},
|
| 39 |
+
{
|
| 40 |
+
"epoch": 0.0999000999000999,
|
| 41 |
+
"grad_norm": 0.4039924740791321,
|
| 42 |
+
"learning_rate": 0.00019995065603657316,
|
| 43 |
+
"loss": 0.953,
|
| 44 |
+
"step": 50
|
| 45 |
+
},
|
| 46 |
+
{
|
| 47 |
+
"epoch": 0.11988011988011989,
|
| 48 |
+
"grad_norm": 0.37654027342796326,
|
| 49 |
+
"learning_rate": 0.00019992894726405893,
|
| 50 |
+
"loss": 0.9138,
|
| 51 |
+
"step": 60
|
| 52 |
+
},
|
| 53 |
+
{
|
| 54 |
+
"epoch": 0.13986013986013987,
|
| 55 |
+
"grad_norm": 0.41351592540740967,
|
| 56 |
+
"learning_rate": 0.0001999032934678125,
|
| 57 |
+
"loss": 0.9121,
|
| 58 |
+
"step": 70
|
| 59 |
+
},
|
| 60 |
+
{
|
| 61 |
+
"epoch": 0.15984015984015984,
|
| 62 |
+
"grad_norm": 0.42355260252952576,
|
| 63 |
+
"learning_rate": 0.00019987369566060176,
|
| 64 |
+
"loss": 0.8971,
|
| 65 |
+
"step": 80
|
| 66 |
+
},
|
| 67 |
+
{
|
| 68 |
+
"epoch": 0.1798201798201798,
|
| 69 |
+
"grad_norm": 0.40265560150146484,
|
| 70 |
+
"learning_rate": 0.00019984015501089752,
|
| 71 |
+
"loss": 0.892,
|
| 72 |
+
"step": 90
|
| 73 |
+
},
|
| 74 |
+
{
|
| 75 |
+
"epoch": 0.1998001998001998,
|
| 76 |
+
"grad_norm": 0.36668843030929565,
|
| 77 |
+
"learning_rate": 0.00019980267284282717,
|
| 78 |
+
"loss": 0.8907,
|
| 79 |
+
"step": 100
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.21978021978021978,
|
| 83 |
+
"grad_norm": 0.3516446352005005,
|
| 84 |
+
"learning_rate": 0.00019976125063612252,
|
| 85 |
+
"loss": 0.888,
|
| 86 |
+
"step": 110
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.23976023976023977,
|
| 90 |
+
"grad_norm": 0.3761754631996155,
|
| 91 |
+
"learning_rate": 0.0001997158900260614,
|
| 92 |
+
"loss": 0.8883,
|
| 93 |
+
"step": 120
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.2597402597402597,
|
| 97 |
+
"grad_norm": 0.3486793041229248,
|
| 98 |
+
"learning_rate": 0.00019966659280340297,
|
| 99 |
+
"loss": 0.8709,
|
| 100 |
+
"step": 130
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"epoch": 0.27972027972027974,
|
| 104 |
+
"grad_norm": 0.39413630962371826,
|
| 105 |
+
"learning_rate": 0.00019961336091431727,
|
| 106 |
+
"loss": 0.8544,
|
| 107 |
+
"step": 140
|
| 108 |
+
},
|
| 109 |
+
{
|
| 110 |
+
"epoch": 0.2997002997002997,
|
| 111 |
+
"grad_norm": 0.3653990924358368,
|
| 112 |
+
"learning_rate": 0.00019955619646030802,
|
| 113 |
+
"loss": 0.8647,
|
| 114 |
+
"step": 150
|
| 115 |
+
},
|
| 116 |
+
{
|
| 117 |
+
"epoch": 0.3196803196803197,
|
| 118 |
+
"grad_norm": 0.4523209035396576,
|
| 119 |
+
"learning_rate": 0.00019949510169813003,
|
| 120 |
+
"loss": 0.8698,
|
| 121 |
+
"step": 160
|
| 122 |
+
},
|
| 123 |
+
{
|
| 124 |
+
"epoch": 0.33966033966033965,
|
| 125 |
+
"grad_norm": 0.3841874897480011,
|
| 126 |
+
"learning_rate": 0.0001994300790396999,
|
| 127 |
+
"loss": 0.8513,
|
| 128 |
+
"step": 170
|
| 129 |
+
},
|
| 130 |
+
{
|
| 131 |
+
"epoch": 0.3596403596403596,
|
| 132 |
+
"grad_norm": 0.3849908709526062,
|
| 133 |
+
"learning_rate": 0.00019936113105200085,
|
| 134 |
+
"loss": 0.8553,
|
| 135 |
+
"step": 180
|
| 136 |
+
},
|
| 137 |
+
{
|
| 138 |
+
"epoch": 0.37962037962037964,
|
| 139 |
+
"grad_norm": 0.3563358783721924,
|
| 140 |
+
"learning_rate": 0.00019928826045698136,
|
| 141 |
+
"loss": 0.8615,
|
| 142 |
+
"step": 190
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.3996003996003996,
|
| 146 |
+
"grad_norm": 0.3968392610549927,
|
| 147 |
+
"learning_rate": 0.0001992114701314478,
|
| 148 |
+
"loss": 0.8502,
|
| 149 |
+
"step": 200
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.4195804195804196,
|
| 153 |
+
"grad_norm": 0.366230845451355,
|
| 154 |
+
"learning_rate": 0.00019913076310695068,
|
| 155 |
+
"loss": 0.8368,
|
| 156 |
+
"step": 210
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.43956043956043955,
|
| 160 |
+
"grad_norm": 0.38813525438308716,
|
| 161 |
+
"learning_rate": 0.00019904614256966512,
|
| 162 |
+
"loss": 0.862,
|
| 163 |
+
"step": 220
|
| 164 |
+
},
|
| 165 |
+
{
|
| 166 |
+
"epoch": 0.4595404595404595,
|
| 167 |
+
"grad_norm": 0.35268592834472656,
|
| 168 |
+
"learning_rate": 0.0001989576118602651,
|
| 169 |
+
"loss": 0.8468,
|
| 170 |
+
"step": 230
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"epoch": 0.47952047952047955,
|
| 174 |
+
"grad_norm": 0.342580109834671,
|
| 175 |
+
"learning_rate": 0.0001988651744737914,
|
| 176 |
+
"loss": 0.8575,
|
| 177 |
+
"step": 240
|
| 178 |
+
},
|
| 179 |
+
{
|
| 180 |
+
"epoch": 0.4995004995004995,
|
| 181 |
+
"grad_norm": 0.37153083086013794,
|
| 182 |
+
"learning_rate": 0.00019876883405951377,
|
| 183 |
+
"loss": 0.8374,
|
| 184 |
+
"step": 250
|
| 185 |
+
},
|
| 186 |
+
{
|
| 187 |
+
"epoch": 0.5194805194805194,
|
| 188 |
+
"grad_norm": 0.3486216366291046,
|
| 189 |
+
"learning_rate": 0.0001986685944207868,
|
| 190 |
+
"loss": 0.8333,
|
| 191 |
+
"step": 260
|
| 192 |
+
},
|
| 193 |
+
{
|
| 194 |
+
"epoch": 0.5394605394605395,
|
| 195 |
+
"grad_norm": 0.3562557101249695,
|
| 196 |
+
"learning_rate": 0.00019856445951489982,
|
| 197 |
+
"loss": 0.8238,
|
| 198 |
+
"step": 270
|
| 199 |
+
},
|
| 200 |
+
{
|
| 201 |
+
"epoch": 0.5594405594405595,
|
| 202 |
+
"grad_norm": 0.3600502610206604,
|
| 203 |
+
"learning_rate": 0.00019845643345292054,
|
| 204 |
+
"loss": 0.8331,
|
| 205 |
+
"step": 280
|
| 206 |
+
},
|
| 207 |
+
{
|
| 208 |
+
"epoch": 0.5794205794205795,
|
| 209 |
+
"grad_norm": 0.3475654423236847,
|
| 210 |
+
"learning_rate": 0.00019834452049953297,
|
| 211 |
+
"loss": 0.8093,
|
| 212 |
+
"step": 290
|
| 213 |
+
},
|
| 214 |
+
{
|
| 215 |
+
"epoch": 0.5994005994005994,
|
| 216 |
+
"grad_norm": 0.358980655670166,
|
| 217 |
+
"learning_rate": 0.0001982287250728689,
|
| 218 |
+
"loss": 0.8302,
|
| 219 |
+
"step": 300
|
| 220 |
+
},
|
| 221 |
+
{
|
| 222 |
+
"epoch": 0.6193806193806194,
|
| 223 |
+
"grad_norm": 0.3721815347671509,
|
| 224 |
+
"learning_rate": 0.0001981090517443334,
|
| 225 |
+
"loss": 0.8175,
|
| 226 |
+
"step": 310
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.6393606393606394,
|
| 230 |
+
"grad_norm": 0.35128098726272583,
|
| 231 |
+
"learning_rate": 0.0001979855052384247,
|
| 232 |
+
"loss": 0.8193,
|
| 233 |
+
"step": 320
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.6593406593406593,
|
| 237 |
+
"grad_norm": 0.3471618592739105,
|
| 238 |
+
"learning_rate": 0.00019785809043254722,
|
| 239 |
+
"loss": 0.8232,
|
| 240 |
+
"step": 330
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.6793206793206793,
|
| 244 |
+
"grad_norm": 0.35060420632362366,
|
| 245 |
+
"learning_rate": 0.00019772681235681936,
|
| 246 |
+
"loss": 0.8194,
|
| 247 |
+
"step": 340
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.6993006993006993,
|
| 251 |
+
"grad_norm": 0.3695327341556549,
|
| 252 |
+
"learning_rate": 0.00019759167619387476,
|
| 253 |
+
"loss": 0.806,
|
| 254 |
+
"step": 350
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.7192807192807192,
|
| 258 |
+
"grad_norm": 0.35857513546943665,
|
| 259 |
+
"learning_rate": 0.00019745268727865774,
|
| 260 |
+
"loss": 0.8019,
|
| 261 |
+
"step": 360
|
| 262 |
+
},
|
| 263 |
+
{
|
| 264 |
+
"epoch": 0.7392607392607392,
|
| 265 |
+
"grad_norm": 0.3612421154975891,
|
| 266 |
+
"learning_rate": 0.00019730985109821266,
|
| 267 |
+
"loss": 0.8061,
|
| 268 |
+
"step": 370
|
| 269 |
+
},
|
| 270 |
+
{
|
| 271 |
+
"epoch": 0.7592407592407593,
|
| 272 |
+
"grad_norm": 0.34007078409194946,
|
| 273 |
+
"learning_rate": 0.0001971631732914674,
|
| 274 |
+
"loss": 0.7919,
|
| 275 |
+
"step": 380
|
| 276 |
+
},
|
| 277 |
+
{
|
| 278 |
+
"epoch": 0.7792207792207793,
|
| 279 |
+
"grad_norm": 0.3594492971897125,
|
| 280 |
+
"learning_rate": 0.0001970126596490106,
|
| 281 |
+
"loss": 0.7821,
|
| 282 |
+
"step": 390
|
| 283 |
+
},
|
| 284 |
+
{
|
| 285 |
+
"epoch": 0.7992007992007992,
|
| 286 |
+
"grad_norm": 0.37426885962486267,
|
| 287 |
+
"learning_rate": 0.0001968583161128631,
|
| 288 |
+
"loss": 0.8054,
|
| 289 |
+
"step": 400
|
| 290 |
+
},
|
| 291 |
+
{
|
| 292 |
+
"epoch": 0.8191808191808192,
|
| 293 |
+
"grad_norm": 0.3551250398159027,
|
| 294 |
+
"learning_rate": 0.00019670014877624353,
|
| 295 |
+
"loss": 0.7954,
|
| 296 |
+
"step": 410
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.8391608391608392,
|
| 300 |
+
"grad_norm": 0.35951119661331177,
|
| 301 |
+
"learning_rate": 0.0001965381638833274,
|
| 302 |
+
"loss": 0.7966,
|
| 303 |
+
"step": 420
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.8591408591408591,
|
| 307 |
+
"grad_norm": 0.36964887380599976,
|
| 308 |
+
"learning_rate": 0.000196372367829001,
|
| 309 |
+
"loss": 0.7888,
|
| 310 |
+
"step": 430
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.8791208791208791,
|
| 314 |
+
"grad_norm": 0.36829873919487,
|
| 315 |
+
"learning_rate": 0.0001962027671586086,
|
| 316 |
+
"loss": 0.7902,
|
| 317 |
+
"step": 440
|
| 318 |
+
},
|
| 319 |
+
{
|
| 320 |
+
"epoch": 0.8991008991008991,
|
| 321 |
+
"grad_norm": 0.34358304738998413,
|
| 322 |
+
"learning_rate": 0.0001960293685676943,
|
| 323 |
+
"loss": 0.7733,
|
| 324 |
+
"step": 450
|
| 325 |
+
},
|
| 326 |
+
{
|
| 327 |
+
"epoch": 0.919080919080919,
|
| 328 |
+
"grad_norm": 0.37369629740715027,
|
| 329 |
+
"learning_rate": 0.0001958521789017376,
|
| 330 |
+
"loss": 0.796,
|
| 331 |
+
"step": 460
|
| 332 |
+
},
|
| 333 |
+
{
|
| 334 |
+
"epoch": 0.939060939060939,
|
| 335 |
+
"grad_norm": 0.40985429286956787,
|
| 336 |
+
"learning_rate": 0.00019567120515588308,
|
| 337 |
+
"loss": 0.7931,
|
| 338 |
+
"step": 470
|
| 339 |
+
},
|
| 340 |
+
{
|
| 341 |
+
"epoch": 0.9590409590409591,
|
| 342 |
+
"grad_norm": 0.34838569164276123,
|
| 343 |
+
"learning_rate": 0.00019548645447466431,
|
| 344 |
+
"loss": 0.7682,
|
| 345 |
+
"step": 480
|
| 346 |
+
},
|
| 347 |
+
{
|
| 348 |
+
"epoch": 0.9790209790209791,
|
| 349 |
+
"grad_norm": 0.36467525362968445,
|
| 350 |
+
"learning_rate": 0.00019529793415172192,
|
| 351 |
+
"loss": 0.7781,
|
| 352 |
+
"step": 490
|
| 353 |
+
},
|
| 354 |
+
{
|
| 355 |
+
"epoch": 0.999000999000999,
|
| 356 |
+
"grad_norm": 0.37112316489219666,
|
| 357 |
+
"learning_rate": 0.00019510565162951537,
|
| 358 |
+
"loss": 0.7773,
|
| 359 |
+
"step": 500
|
| 360 |
+
},
|
| 361 |
+
{
|
| 362 |
+
"epoch": 1.018981018981019,
|
| 363 |
+
"grad_norm": 0.3998737931251526,
|
| 364 |
+
"learning_rate": 0.00019490961449902946,
|
| 365 |
+
"loss": 0.7324,
|
| 366 |
+
"step": 510
|
| 367 |
+
},
|
| 368 |
+
{
|
| 369 |
+
"epoch": 1.0389610389610389,
|
| 370 |
+
"grad_norm": 0.3966336250305176,
|
| 371 |
+
"learning_rate": 0.00019470983049947444,
|
| 372 |
+
"loss": 0.7395,
|
| 373 |
+
"step": 520
|
| 374 |
+
},
|
| 375 |
+
{
|
| 376 |
+
"epoch": 1.058941058941059,
|
| 377 |
+
"grad_norm": 0.39721325039863586,
|
| 378 |
+
"learning_rate": 0.00019450630751798048,
|
| 379 |
+
"loss": 0.7302,
|
| 380 |
+
"step": 530
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"epoch": 1.078921078921079,
|
| 384 |
+
"grad_norm": 0.38532692193984985,
|
| 385 |
+
"learning_rate": 0.00019429905358928646,
|
| 386 |
+
"loss": 0.7177,
|
| 387 |
+
"step": 540
|
| 388 |
+
},
|
| 389 |
+
{
|
| 390 |
+
"epoch": 1.098901098901099,
|
| 391 |
+
"grad_norm": 0.3948540985584259,
|
| 392 |
+
"learning_rate": 0.00019408807689542257,
|
| 393 |
+
"loss": 0.7382,
|
| 394 |
+
"step": 550
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 1.118881118881119,
|
| 398 |
+
"grad_norm": 0.399676650762558,
|
| 399 |
+
"learning_rate": 0.00019387338576538744,
|
| 400 |
+
"loss": 0.7286,
|
| 401 |
+
"step": 560
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.138861138861139,
|
| 405 |
+
"grad_norm": 0.4208274781703949,
|
| 406 |
+
"learning_rate": 0.00019365498867481923,
|
| 407 |
+
"loss": 0.7251,
|
| 408 |
+
"step": 570
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.158841158841159,
|
| 412 |
+
"grad_norm": 0.4160782992839813,
|
| 413 |
+
"learning_rate": 0.00019343289424566122,
|
| 414 |
+
"loss": 0.7138,
|
| 415 |
+
"step": 580
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 1.1788211788211789,
|
| 419 |
+
"grad_norm": 0.4297160804271698,
|
| 420 |
+
"learning_rate": 0.0001932071112458211,
|
| 421 |
+
"loss": 0.7296,
|
| 422 |
+
"step": 590
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.1988011988011988,
|
| 426 |
+
"grad_norm": 0.4196039140224457,
|
| 427 |
+
"learning_rate": 0.00019297764858882514,
|
| 428 |
+
"loss": 0.7091,
|
| 429 |
+
"step": 600
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.2187812187812188,
|
| 433 |
+
"grad_norm": 0.406012624502182,
|
| 434 |
+
"learning_rate": 0.00019274451533346615,
|
| 435 |
+
"loss": 0.7021,
|
| 436 |
+
"step": 610
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 1.2387612387612388,
|
| 440 |
+
"grad_norm": 0.41200658679008484,
|
| 441 |
+
"learning_rate": 0.0001925077206834458,
|
| 442 |
+
"loss": 0.7238,
|
| 443 |
+
"step": 620
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.2587412587412588,
|
| 447 |
+
"grad_norm": 0.4819345772266388,
|
| 448 |
+
"learning_rate": 0.0001922672739870115,
|
| 449 |
+
"loss": 0.7275,
|
| 450 |
+
"step": 630
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.2787212787212787,
|
| 454 |
+
"grad_norm": 0.40825748443603516,
|
| 455 |
+
"learning_rate": 0.00019202318473658705,
|
| 456 |
+
"loss": 0.7183,
|
| 457 |
+
"step": 640
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 1.2987012987012987,
|
| 461 |
+
"grad_norm": 0.41940203309059143,
|
| 462 |
+
"learning_rate": 0.00019177546256839812,
|
| 463 |
+
"loss": 0.7149,
|
| 464 |
+
"step": 650
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 1.3186813186813187,
|
| 468 |
+
"grad_norm": 0.40075168013572693,
|
| 469 |
+
"learning_rate": 0.00019152411726209176,
|
| 470 |
+
"loss": 0.722,
|
| 471 |
+
"step": 660
|
| 472 |
+
},
|
| 473 |
+
{
|
| 474 |
+
"epoch": 1.3386613386613386,
|
| 475 |
+
"grad_norm": 0.4254063665866852,
|
| 476 |
+
"learning_rate": 0.0001912691587403503,
|
| 477 |
+
"loss": 0.7254,
|
| 478 |
+
"step": 670
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.3586413586413586,
|
| 482 |
+
"grad_norm": 0.39732539653778076,
|
| 483 |
+
"learning_rate": 0.00019101059706849957,
|
| 484 |
+
"loss": 0.7115,
|
| 485 |
+
"step": 680
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.3786213786213786,
|
| 489 |
+
"grad_norm": 0.3889389932155609,
|
| 490 |
+
"learning_rate": 0.0001907484424541117,
|
| 491 |
+
"loss": 0.7031,
|
| 492 |
+
"step": 690
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.3986013986013985,
|
| 496 |
+
"grad_norm": 0.3994196355342865,
|
| 497 |
+
"learning_rate": 0.00019048270524660196,
|
| 498 |
+
"loss": 0.7095,
|
| 499 |
+
"step": 700
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.4185814185814185,
|
| 503 |
+
"grad_norm": 0.4238826036453247,
|
| 504 |
+
"learning_rate": 0.00019021339593682028,
|
| 505 |
+
"loss": 0.7156,
|
| 506 |
+
"step": 710
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.4385614385614387,
|
| 510 |
+
"grad_norm": 0.4787987172603607,
|
| 511 |
+
"learning_rate": 0.0001899405251566371,
|
| 512 |
+
"loss": 0.7142,
|
| 513 |
+
"step": 720
|
| 514 |
+
},
|
| 515 |
+
{
|
| 516 |
+
"epoch": 1.4585414585414584,
|
| 517 |
+
"grad_norm": 0.4219954013824463,
|
| 518 |
+
"learning_rate": 0.00018966410367852362,
|
| 519 |
+
"loss": 0.7267,
|
| 520 |
+
"step": 730
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"epoch": 1.4785214785214786,
|
| 524 |
+
"grad_norm": 0.4154765009880066,
|
| 525 |
+
"learning_rate": 0.0001893841424151264,
|
| 526 |
+
"loss": 0.721,
|
| 527 |
+
"step": 740
|
| 528 |
+
},
|
| 529 |
+
{
|
| 530 |
+
"epoch": 1.4985014985014984,
|
| 531 |
+
"grad_norm": 0.44605547189712524,
|
| 532 |
+
"learning_rate": 0.0001891006524188368,
|
| 533 |
+
"loss": 0.7266,
|
| 534 |
+
"step": 750
|
| 535 |
+
},
|
| 536 |
+
{
|
| 537 |
+
"epoch": 1.5184815184815186,
|
| 538 |
+
"grad_norm": 0.4613310992717743,
|
| 539 |
+
"learning_rate": 0.00018881364488135448,
|
| 540 |
+
"loss": 0.7253,
|
| 541 |
+
"step": 760
|
| 542 |
+
},
|
| 543 |
+
{
|
| 544 |
+
"epoch": 1.5384615384615383,
|
| 545 |
+
"grad_norm": 0.41615426540374756,
|
| 546 |
+
"learning_rate": 0.00018852313113324552,
|
| 547 |
+
"loss": 0.69,
|
| 548 |
+
"step": 770
|
| 549 |
+
},
|
| 550 |
+
{
|
| 551 |
+
"epoch": 1.5584415584415585,
|
| 552 |
+
"grad_norm": 0.4512516260147095,
|
| 553 |
+
"learning_rate": 0.00018822912264349534,
|
| 554 |
+
"loss": 0.7124,
|
| 555 |
+
"step": 780
|
| 556 |
+
},
|
| 557 |
+
{
|
| 558 |
+
"epoch": 1.5784215784215783,
|
| 559 |
+
"grad_norm": 0.464336633682251,
|
| 560 |
+
"learning_rate": 0.00018793163101905563,
|
| 561 |
+
"loss": 0.7067,
|
| 562 |
+
"step": 790
|
| 563 |
+
},
|
| 564 |
+
{
|
| 565 |
+
"epoch": 1.5984015984015985,
|
| 566 |
+
"grad_norm": 0.4427087604999542,
|
| 567 |
+
"learning_rate": 0.00018763066800438636,
|
| 568 |
+
"loss": 0.7097,
|
| 569 |
+
"step": 800
|
| 570 |
+
},
|
| 571 |
+
{
|
| 572 |
+
"epoch": 1.6183816183816184,
|
| 573 |
+
"grad_norm": 0.43341028690338135,
|
| 574 |
+
"learning_rate": 0.00018732624548099204,
|
| 575 |
+
"loss": 0.7068,
|
| 576 |
+
"step": 810
|
| 577 |
+
},
|
| 578 |
+
{
|
| 579 |
+
"epoch": 1.6383616383616384,
|
| 580 |
+
"grad_norm": 0.4100460112094879,
|
| 581 |
+
"learning_rate": 0.0001870183754669526,
|
| 582 |
+
"loss": 0.705,
|
| 583 |
+
"step": 820
|
| 584 |
+
},
|
| 585 |
+
{
|
| 586 |
+
"epoch": 1.6583416583416584,
|
| 587 |
+
"grad_norm": 0.43942147493362427,
|
| 588 |
+
"learning_rate": 0.000186707070116449,
|
| 589 |
+
"loss": 0.7043,
|
| 590 |
+
"step": 830
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"epoch": 1.6783216783216783,
|
| 594 |
+
"grad_norm": 0.430095911026001,
|
| 595 |
+
"learning_rate": 0.00018639234171928353,
|
| 596 |
+
"loss": 0.6989,
|
| 597 |
+
"step": 840
|
| 598 |
+
},
|
| 599 |
+
{
|
| 600 |
+
"epoch": 1.6983016983016983,
|
| 601 |
+
"grad_norm": 0.40418198704719543,
|
| 602 |
+
"learning_rate": 0.0001860742027003944,
|
| 603 |
+
"loss": 0.6933,
|
| 604 |
+
"step": 850
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.7182817182817183,
|
| 608 |
+
"grad_norm": 0.40910184383392334,
|
| 609 |
+
"learning_rate": 0.00018575266561936523,
|
| 610 |
+
"loss": 0.6848,
|
| 611 |
+
"step": 860
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.7382617382617382,
|
| 615 |
+
"grad_norm": 0.4620640277862549,
|
| 616 |
+
"learning_rate": 0.0001854277431699295,
|
| 617 |
+
"loss": 0.6943,
|
| 618 |
+
"step": 870
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.7582417582417582,
|
| 622 |
+
"grad_norm": 0.4648028314113617,
|
| 623 |
+
"learning_rate": 0.00018509944817946922,
|
| 624 |
+
"loss": 0.6993,
|
| 625 |
+
"step": 880
|
| 626 |
+
},
|
| 627 |
+
{
|
| 628 |
+
"epoch": 1.7782217782217782,
|
| 629 |
+
"grad_norm": 0.43752139806747437,
|
| 630 |
+
"learning_rate": 0.00018476779360850832,
|
| 631 |
+
"loss": 0.6827,
|
| 632 |
+
"step": 890
|
| 633 |
+
},
|
| 634 |
+
{
|
| 635 |
+
"epoch": 1.7982017982017982,
|
| 636 |
+
"grad_norm": 0.4481639862060547,
|
| 637 |
+
"learning_rate": 0.00018443279255020152,
|
| 638 |
+
"loss": 0.6978,
|
| 639 |
+
"step": 900
|
| 640 |
+
},
|
| 641 |
+
{
|
| 642 |
+
"epoch": 1.8181818181818183,
|
| 643 |
+
"grad_norm": 0.4678110182285309,
|
| 644 |
+
"learning_rate": 0.00018409445822981693,
|
| 645 |
+
"loss": 0.6848,
|
| 646 |
+
"step": 910
|
| 647 |
+
},
|
| 648 |
+
{
|
| 649 |
+
"epoch": 1.838161838161838,
|
| 650 |
+
"grad_norm": 0.433933824300766,
|
| 651 |
+
"learning_rate": 0.0001837528040042142,
|
| 652 |
+
"loss": 0.658,
|
| 653 |
+
"step": 920
|
| 654 |
+
},
|
| 655 |
+
{
|
| 656 |
+
"epoch": 1.8581418581418583,
|
| 657 |
+
"grad_norm": 0.4601323902606964,
|
| 658 |
+
"learning_rate": 0.00018340784336131713,
|
| 659 |
+
"loss": 0.6912,
|
| 660 |
+
"step": 930
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"epoch": 1.878121878121878,
|
| 664 |
+
"grad_norm": 0.4591493308544159,
|
| 665 |
+
"learning_rate": 0.00018305958991958127,
|
| 666 |
+
"loss": 0.697,
|
| 667 |
+
"step": 940
|
| 668 |
+
},
|
| 669 |
+
{
|
| 670 |
+
"epoch": 1.8981018981018982,
|
| 671 |
+
"grad_norm": 0.445711225271225,
|
| 672 |
+
"learning_rate": 0.00018270805742745617,
|
| 673 |
+
"loss": 0.6922,
|
| 674 |
+
"step": 950
|
| 675 |
+
},
|
| 676 |
+
{
|
| 677 |
+
"epoch": 1.918081918081918,
|
| 678 |
+
"grad_norm": 0.43125954270362854,
|
| 679 |
+
"learning_rate": 0.00018235325976284275,
|
| 680 |
+
"loss": 0.6742,
|
| 681 |
+
"step": 960
|
| 682 |
+
},
|
| 683 |
+
{
|
| 684 |
+
"epoch": 1.9380619380619382,
|
| 685 |
+
"grad_norm": 0.4716484248638153,
|
| 686 |
+
"learning_rate": 0.00018199521093254523,
|
| 687 |
+
"loss": 0.6796,
|
| 688 |
+
"step": 970
|
| 689 |
+
},
|
| 690 |
+
{
|
| 691 |
+
"epoch": 1.958041958041958,
|
| 692 |
+
"grad_norm": 0.4613405764102936,
|
| 693 |
+
"learning_rate": 0.00018163392507171842,
|
| 694 |
+
"loss": 0.6832,
|
| 695 |
+
"step": 980
|
| 696 |
+
},
|
| 697 |
+
{
|
| 698 |
+
"epoch": 1.978021978021978,
|
| 699 |
+
"grad_norm": 0.48080363869667053,
|
| 700 |
+
"learning_rate": 0.0001812694164433094,
|
| 701 |
+
"loss": 0.6807,
|
| 702 |
+
"step": 990
|
| 703 |
+
},
|
| 704 |
+
{
|
| 705 |
+
"epoch": 1.9980019980019978,
|
| 706 |
+
"grad_norm": 0.47017648816108704,
|
| 707 |
+
"learning_rate": 0.00018090169943749476,
|
| 708 |
+
"loss": 0.6785,
|
| 709 |
+
"step": 1000
|
| 710 |
+
},
|
| 711 |
+
{
|
| 712 |
+
"epoch": 2.017982017982018,
|
| 713 |
+
"grad_norm": 0.516197919845581,
|
| 714 |
+
"learning_rate": 0.0001805307885711122,
|
| 715 |
+
"loss": 0.6019,
|
| 716 |
+
"step": 1010
|
| 717 |
+
},
|
| 718 |
+
{
|
| 719 |
+
"epoch": 2.037962037962038,
|
| 720 |
+
"grad_norm": 0.5556052923202515,
|
| 721 |
+
"learning_rate": 0.00018015669848708767,
|
| 722 |
+
"loss": 0.5906,
|
| 723 |
+
"step": 1020
|
| 724 |
+
},
|
| 725 |
+
{
|
| 726 |
+
"epoch": 2.057942057942058,
|
| 727 |
+
"grad_norm": 0.5169907808303833,
|
| 728 |
+
"learning_rate": 0.0001797794439538571,
|
| 729 |
+
"loss": 0.6076,
|
| 730 |
+
"step": 1030
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 2.0779220779220777,
|
| 734 |
+
"grad_norm": 0.5560281276702881,
|
| 735 |
+
"learning_rate": 0.00017939903986478355,
|
| 736 |
+
"loss": 0.582,
|
| 737 |
+
"step": 1040
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 2.097902097902098,
|
| 741 |
+
"grad_norm": 0.521091878414154,
|
| 742 |
+
"learning_rate": 0.00017901550123756906,
|
| 743 |
+
"loss": 0.5929,
|
| 744 |
+
"step": 1050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 2.117882117882118,
|
| 748 |
+
"grad_norm": 0.5990195870399475,
|
| 749 |
+
"learning_rate": 0.00017862884321366188,
|
| 750 |
+
"loss": 0.5863,
|
| 751 |
+
"step": 1060
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 2.137862137862138,
|
| 755 |
+
"grad_norm": 0.5285313725471497,
|
| 756 |
+
"learning_rate": 0.0001782390810576588,
|
| 757 |
+
"loss": 0.5845,
|
| 758 |
+
"step": 1070
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 2.157842157842158,
|
| 762 |
+
"grad_norm": 0.5402159690856934,
|
| 763 |
+
"learning_rate": 0.00017784623015670238,
|
| 764 |
+
"loss": 0.5926,
|
| 765 |
+
"step": 1080
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 2.177822177822178,
|
| 769 |
+
"grad_norm": 0.5576025247573853,
|
| 770 |
+
"learning_rate": 0.00017745030601987337,
|
| 771 |
+
"loss": 0.5964,
|
| 772 |
+
"step": 1090
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 2.197802197802198,
|
| 776 |
+
"grad_norm": 0.5605506896972656,
|
| 777 |
+
"learning_rate": 0.00017705132427757895,
|
| 778 |
+
"loss": 0.5877,
|
| 779 |
+
"step": 1100
|
| 780 |
+
},
|
| 781 |
+
{
|
| 782 |
+
"epoch": 2.2177822177822177,
|
| 783 |
+
"grad_norm": 0.5754747986793518,
|
| 784 |
+
"learning_rate": 0.00017664930068093498,
|
| 785 |
+
"loss": 0.6002,
|
| 786 |
+
"step": 1110
|
| 787 |
+
},
|
| 788 |
+
{
|
| 789 |
+
"epoch": 2.237762237762238,
|
| 790 |
+
"grad_norm": 0.5654470324516296,
|
| 791 |
+
"learning_rate": 0.0001762442511011448,
|
| 792 |
+
"loss": 0.5922,
|
| 793 |
+
"step": 1120
|
| 794 |
+
},
|
| 795 |
+
{
|
| 796 |
+
"epoch": 2.2577422577422577,
|
| 797 |
+
"grad_norm": 0.5414491891860962,
|
| 798 |
+
"learning_rate": 0.0001758361915288722,
|
| 799 |
+
"loss": 0.5917,
|
| 800 |
+
"step": 1130
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"epoch": 2.277722277722278,
|
| 804 |
+
"grad_norm": 0.5563125014305115,
|
| 805 |
+
"learning_rate": 0.00017542513807361037,
|
| 806 |
+
"loss": 0.5867,
|
| 807 |
+
"step": 1140
|
| 808 |
+
},
|
| 809 |
+
{
|
| 810 |
+
"epoch": 2.2977022977022976,
|
| 811 |
+
"grad_norm": 0.5236257314682007,
|
| 812 |
+
"learning_rate": 0.00017501110696304596,
|
| 813 |
+
"loss": 0.5888,
|
| 814 |
+
"step": 1150
|
| 815 |
+
},
|
| 816 |
+
{
|
| 817 |
+
"epoch": 2.317682317682318,
|
| 818 |
+
"grad_norm": 0.614734411239624,
|
| 819 |
+
"learning_rate": 0.00017459411454241822,
|
| 820 |
+
"loss": 0.6001,
|
| 821 |
+
"step": 1160
|
| 822 |
+
},
|
| 823 |
+
{
|
| 824 |
+
"epoch": 2.3376623376623376,
|
| 825 |
+
"grad_norm": 0.605421781539917,
|
| 826 |
+
"learning_rate": 0.00017417417727387394,
|
| 827 |
+
"loss": 0.5968,
|
| 828 |
+
"step": 1170
|
| 829 |
+
},
|
| 830 |
+
{
|
| 831 |
+
"epoch": 2.3576423576423577,
|
| 832 |
+
"grad_norm": 0.5595569014549255,
|
| 833 |
+
"learning_rate": 0.0001737513117358174,
|
| 834 |
+
"loss": 0.5924,
|
| 835 |
+
"step": 1180
|
| 836 |
+
},
|
| 837 |
+
{
|
| 838 |
+
"epoch": 2.3776223776223775,
|
| 839 |
+
"grad_norm": 0.5283003449440002,
|
| 840 |
+
"learning_rate": 0.00017332553462225602,
|
| 841 |
+
"loss": 0.5952,
|
| 842 |
+
"step": 1190
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.3976023976023977,
|
| 846 |
+
"grad_norm": 0.5287072658538818,
|
| 847 |
+
"learning_rate": 0.00017289686274214118,
|
| 848 |
+
"loss": 0.5763,
|
| 849 |
+
"step": 1200
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 2.4175824175824174,
|
| 853 |
+
"grad_norm": 0.5907203555107117,
|
| 854 |
+
"learning_rate": 0.0001724653130187047,
|
| 855 |
+
"loss": 0.5993,
|
| 856 |
+
"step": 1210
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 2.4375624375624376,
|
| 860 |
+
"grad_norm": 0.5622738003730774,
|
| 861 |
+
"learning_rate": 0.0001720309024887907,
|
| 862 |
+
"loss": 0.6001,
|
| 863 |
+
"step": 1220
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 2.4575424575424574,
|
| 867 |
+
"grad_norm": 0.5795326232910156,
|
| 868 |
+
"learning_rate": 0.00017159364830218312,
|
| 869 |
+
"loss": 0.5857,
|
| 870 |
+
"step": 1230
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 2.4775224775224776,
|
| 874 |
+
"grad_norm": 0.5654671788215637,
|
| 875 |
+
"learning_rate": 0.00017115356772092857,
|
| 876 |
+
"loss": 0.5809,
|
| 877 |
+
"step": 1240
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 2.4975024975024973,
|
| 881 |
+
"grad_norm": 0.5641043186187744,
|
| 882 |
+
"learning_rate": 0.00017071067811865476,
|
| 883 |
+
"loss": 0.5824,
|
| 884 |
+
"step": 1250
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.5174825174825175,
|
| 888 |
+
"grad_norm": 0.5851653218269348,
|
| 889 |
+
"learning_rate": 0.00017026499697988493,
|
| 890 |
+
"loss": 0.59,
|
| 891 |
+
"step": 1260
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.5374625374625372,
|
| 895 |
+
"grad_norm": 0.570210337638855,
|
| 896 |
+
"learning_rate": 0.00016981654189934727,
|
| 897 |
+
"loss": 0.5761,
|
| 898 |
+
"step": 1270
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 2.5574425574425574,
|
| 902 |
+
"grad_norm": 0.5725647807121277,
|
| 903 |
+
"learning_rate": 0.0001693653305812805,
|
| 904 |
+
"loss": 0.589,
|
| 905 |
+
"step": 1280
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 2.5774225774225776,
|
| 909 |
+
"grad_norm": 0.5896579623222351,
|
| 910 |
+
"learning_rate": 0.00016891138083873487,
|
| 911 |
+
"loss": 0.5852,
|
| 912 |
+
"step": 1290
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 2.5974025974025974,
|
| 916 |
+
"grad_norm": 0.5988901853561401,
|
| 917 |
+
"learning_rate": 0.00016845471059286887,
|
| 918 |
+
"loss": 0.5723,
|
| 919 |
+
"step": 1300
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 2.617382617382617,
|
| 923 |
+
"grad_norm": 0.5854650735855103,
|
| 924 |
+
"learning_rate": 0.00016799533787224192,
|
| 925 |
+
"loss": 0.5845,
|
| 926 |
+
"step": 1310
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 2.6373626373626373,
|
| 930 |
+
"grad_norm": 0.5547802448272705,
|
| 931 |
+
"learning_rate": 0.00016753328081210245,
|
| 932 |
+
"loss": 0.5909,
|
| 933 |
+
"step": 1320
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 2.6573426573426575,
|
| 937 |
+
"grad_norm": 0.5562127232551575,
|
| 938 |
+
"learning_rate": 0.000167068557653672,
|
| 939 |
+
"loss": 0.5799,
|
| 940 |
+
"step": 1330
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 2.6773226773226773,
|
| 944 |
+
"grad_norm": 0.5999246835708618,
|
| 945 |
+
"learning_rate": 0.00016660118674342517,
|
| 946 |
+
"loss": 0.5757,
|
| 947 |
+
"step": 1340
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 2.6973026973026974,
|
| 951 |
+
"grad_norm": 0.5909945368766785,
|
| 952 |
+
"learning_rate": 0.00016613118653236518,
|
| 953 |
+
"loss": 0.5674,
|
| 954 |
+
"step": 1350
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 2.717282717282717,
|
| 958 |
+
"grad_norm": 0.6357455849647522,
|
| 959 |
+
"learning_rate": 0.00016565857557529566,
|
| 960 |
+
"loss": 0.5821,
|
| 961 |
+
"step": 1360
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 2.7372627372627374,
|
| 965 |
+
"grad_norm": 0.6019343733787537,
|
| 966 |
+
"learning_rate": 0.0001651833725300879,
|
| 967 |
+
"loss": 0.5783,
|
| 968 |
+
"step": 1370
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 2.757242757242757,
|
| 972 |
+
"grad_norm": 0.6180288791656494,
|
| 973 |
+
"learning_rate": 0.00016470559615694446,
|
| 974 |
+
"loss": 0.6056,
|
| 975 |
+
"step": 1380
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 2.7772227772227773,
|
| 979 |
+
"grad_norm": 0.6171667575836182,
|
| 980 |
+
"learning_rate": 0.00016422526531765846,
|
| 981 |
+
"loss": 0.5799,
|
| 982 |
+
"step": 1390
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 2.797202797202797,
|
| 986 |
+
"grad_norm": 0.5991246700286865,
|
| 987 |
+
"learning_rate": 0.000163742398974869,
|
| 988 |
+
"loss": 0.5668,
|
| 989 |
+
"step": 1400
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 2.8171828171828173,
|
| 993 |
+
"grad_norm": 0.6568031907081604,
|
| 994 |
+
"learning_rate": 0.00016325701619131246,
|
| 995 |
+
"loss": 0.5662,
|
| 996 |
+
"step": 1410
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 2.837162837162837,
|
| 1000 |
+
"grad_norm": 0.6639891266822815,
|
| 1001 |
+
"learning_rate": 0.00016276913612907007,
|
| 1002 |
+
"loss": 0.5797,
|
| 1003 |
+
"step": 1420
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 2.857142857142857,
|
| 1007 |
+
"grad_norm": 0.5978193879127502,
|
| 1008 |
+
"learning_rate": 0.00016227877804881127,
|
| 1009 |
+
"loss": 0.5613,
|
| 1010 |
+
"step": 1430
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 2.8771228771228774,
|
| 1014 |
+
"grad_norm": 0.576871395111084,
|
| 1015 |
+
"learning_rate": 0.00016178596130903344,
|
| 1016 |
+
"loss": 0.5796,
|
| 1017 |
+
"step": 1440
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.897102897102897,
|
| 1021 |
+
"grad_norm": 0.5936170220375061,
|
| 1022 |
+
"learning_rate": 0.00016129070536529766,
|
| 1023 |
+
"loss": 0.5791,
|
| 1024 |
+
"step": 1450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 2.917082917082917,
|
| 1028 |
+
"grad_norm": 0.6093722581863403,
|
| 1029 |
+
"learning_rate": 0.00016079302976946055,
|
| 1030 |
+
"loss": 0.5836,
|
| 1031 |
+
"step": 1460
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 2.937062937062937,
|
| 1035 |
+
"grad_norm": 0.5815151929855347,
|
| 1036 |
+
"learning_rate": 0.00016029295416890248,
|
| 1037 |
+
"loss": 0.5644,
|
| 1038 |
+
"step": 1470
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 2.9570429570429573,
|
| 1042 |
+
"grad_norm": 0.621591329574585,
|
| 1043 |
+
"learning_rate": 0.0001597904983057519,
|
| 1044 |
+
"loss": 0.5779,
|
| 1045 |
+
"step": 1480
|
| 1046 |
+
},
|
| 1047 |
+
{
|
| 1048 |
+
"epoch": 2.977022977022977,
|
| 1049 |
+
"grad_norm": 0.5824622511863708,
|
| 1050 |
+
"learning_rate": 0.00015928568201610595,
|
| 1051 |
+
"loss": 0.5659,
|
| 1052 |
+
"step": 1490
|
| 1053 |
+
},
|
| 1054 |
+
{
|
| 1055 |
+
"epoch": 2.9970029970029968,
|
| 1056 |
+
"grad_norm": 0.6264435052871704,
|
| 1057 |
+
"learning_rate": 0.00015877852522924732,
|
| 1058 |
+
"loss": 0.5823,
|
| 1059 |
+
"step": 1500
|
| 1060 |
+
},
|
| 1061 |
+
{
|
| 1062 |
+
"epoch": 3.016983016983017,
|
| 1063 |
+
"grad_norm": 0.7021110653877258,
|
| 1064 |
+
"learning_rate": 0.00015826904796685762,
|
| 1065 |
+
"loss": 0.4732,
|
| 1066 |
+
"step": 1510
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 3.036963036963037,
|
| 1070 |
+
"grad_norm": 0.7195537686347961,
|
| 1071 |
+
"learning_rate": 0.00015775727034222675,
|
| 1072 |
+
"loss": 0.4484,
|
| 1073 |
+
"step": 1520
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 3.056943056943057,
|
| 1077 |
+
"grad_norm": 0.7159614562988281,
|
| 1078 |
+
"learning_rate": 0.0001572432125594591,
|
| 1079 |
+
"loss": 0.4533,
|
| 1080 |
+
"step": 1530
|
| 1081 |
+
},
|
| 1082 |
+
{
|
| 1083 |
+
"epoch": 3.076923076923077,
|
| 1084 |
+
"grad_norm": 0.686655580997467,
|
| 1085 |
+
"learning_rate": 0.00015672689491267567,
|
| 1086 |
+
"loss": 0.4588,
|
| 1087 |
+
"step": 1540
|
| 1088 |
+
},
|
| 1089 |
+
{
|
| 1090 |
+
"epoch": 3.096903096903097,
|
| 1091 |
+
"grad_norm": 0.6840978264808655,
|
| 1092 |
+
"learning_rate": 0.00015620833778521307,
|
| 1093 |
+
"loss": 0.4632,
|
| 1094 |
+
"step": 1550
|
| 1095 |
+
},
|
| 1096 |
+
{
|
| 1097 |
+
"epoch": 3.116883116883117,
|
| 1098 |
+
"grad_norm": 0.6888960003852844,
|
| 1099 |
+
"learning_rate": 0.00015568756164881882,
|
| 1100 |
+
"loss": 0.463,
|
| 1101 |
+
"step": 1560
|
| 1102 |
+
},
|
| 1103 |
+
{
|
| 1104 |
+
"epoch": 3.136863136863137,
|
| 1105 |
+
"grad_norm": 0.6887105107307434,
|
| 1106 |
+
"learning_rate": 0.00015516458706284303,
|
| 1107 |
+
"loss": 0.4683,
|
| 1108 |
+
"step": 1570
|
| 1109 |
+
},
|
| 1110 |
+
{
|
| 1111 |
+
"epoch": 3.156843156843157,
|
| 1112 |
+
"grad_norm": 0.6880657076835632,
|
| 1113 |
+
"learning_rate": 0.00015463943467342693,
|
| 1114 |
+
"loss": 0.4703,
|
| 1115 |
+
"step": 1580
|
| 1116 |
+
},
|
| 1117 |
+
{
|
| 1118 |
+
"epoch": 3.1768231768231767,
|
| 1119 |
+
"grad_norm": 0.667488157749176,
|
| 1120 |
+
"learning_rate": 0.00015411212521268758,
|
| 1121 |
+
"loss": 0.4681,
|
| 1122 |
+
"step": 1590
|
| 1123 |
+
},
|
| 1124 |
+
{
|
| 1125 |
+
"epoch": 3.196803196803197,
|
| 1126 |
+
"grad_norm": 0.7201547026634216,
|
| 1127 |
+
"learning_rate": 0.00015358267949789966,
|
| 1128 |
+
"loss": 0.4708,
|
| 1129 |
+
"step": 1600
|
| 1130 |
+
},
|
| 1131 |
+
{
|
| 1132 |
+
"epoch": 3.2167832167832167,
|
| 1133 |
+
"grad_norm": 0.7887006998062134,
|
| 1134 |
+
"learning_rate": 0.0001530511184306734,
|
| 1135 |
+
"loss": 0.4692,
|
| 1136 |
+
"step": 1610
|
| 1137 |
+
},
|
| 1138 |
+
{
|
| 1139 |
+
"epoch": 3.236763236763237,
|
| 1140 |
+
"grad_norm": 0.6850538849830627,
|
| 1141 |
+
"learning_rate": 0.0001525174629961296,
|
| 1142 |
+
"loss": 0.4652,
|
| 1143 |
+
"step": 1620
|
| 1144 |
+
},
|
| 1145 |
+
{
|
| 1146 |
+
"epoch": 3.2567432567432566,
|
| 1147 |
+
"grad_norm": 0.7573882937431335,
|
| 1148 |
+
"learning_rate": 0.00015198173426207094,
|
| 1149 |
+
"loss": 0.4618,
|
| 1150 |
+
"step": 1630
|
| 1151 |
+
},
|
| 1152 |
+
{
|
| 1153 |
+
"epoch": 3.276723276723277,
|
| 1154 |
+
"grad_norm": 0.7027117609977722,
|
| 1155 |
+
"learning_rate": 0.00015144395337815064,
|
| 1156 |
+
"loss": 0.4665,
|
| 1157 |
+
"step": 1640
|
| 1158 |
+
},
|
| 1159 |
+
{
|
| 1160 |
+
"epoch": 3.2967032967032965,
|
| 1161 |
+
"grad_norm": 0.6847530007362366,
|
| 1162 |
+
"learning_rate": 0.00015090414157503714,
|
| 1163 |
+
"loss": 0.4669,
|
| 1164 |
+
"step": 1650
|
| 1165 |
+
},
|
| 1166 |
+
{
|
| 1167 |
+
"epoch": 3.3166833166833167,
|
| 1168 |
+
"grad_norm": 0.7099263072013855,
|
| 1169 |
+
"learning_rate": 0.0001503623201635761,
|
| 1170 |
+
"loss": 0.4666,
|
| 1171 |
+
"step": 1660
|
| 1172 |
+
},
|
| 1173 |
+
{
|
| 1174 |
+
"epoch": 3.3366633366633365,
|
| 1175 |
+
"grad_norm": 0.6803727149963379,
|
| 1176 |
+
"learning_rate": 0.0001498185105339491,
|
| 1177 |
+
"loss": 0.4674,
|
| 1178 |
+
"step": 1670
|
| 1179 |
+
},
|
| 1180 |
+
{
|
| 1181 |
+
"epoch": 3.3566433566433567,
|
| 1182 |
+
"grad_norm": 0.7080752849578857,
|
| 1183 |
+
"learning_rate": 0.00014927273415482915,
|
| 1184 |
+
"loss": 0.4694,
|
| 1185 |
+
"step": 1680
|
| 1186 |
+
},
|
| 1187 |
+
{
|
| 1188 |
+
"epoch": 3.3766233766233764,
|
| 1189 |
+
"grad_norm": 0.7016042470932007,
|
| 1190 |
+
"learning_rate": 0.00014872501257253323,
|
| 1191 |
+
"loss": 0.4716,
|
| 1192 |
+
"step": 1690
|
| 1193 |
+
},
|
| 1194 |
+
{
|
| 1195 |
+
"epoch": 3.3966033966033966,
|
| 1196 |
+
"grad_norm": 0.6896219849586487,
|
| 1197 |
+
"learning_rate": 0.00014817536741017152,
|
| 1198 |
+
"loss": 0.4706,
|
| 1199 |
+
"step": 1700
|
| 1200 |
+
},
|
| 1201 |
+
{
|
| 1202 |
+
"epoch": 3.416583416583417,
|
| 1203 |
+
"grad_norm": 0.7319151163101196,
|
| 1204 |
+
"learning_rate": 0.0001476238203667939,
|
| 1205 |
+
"loss": 0.4657,
|
| 1206 |
+
"step": 1710
|
| 1207 |
+
},
|
| 1208 |
+
{
|
| 1209 |
+
"epoch": 3.4365634365634365,
|
| 1210 |
+
"grad_norm": 0.7796220779418945,
|
| 1211 |
+
"learning_rate": 0.0001470703932165333,
|
| 1212 |
+
"loss": 0.4762,
|
| 1213 |
+
"step": 1720
|
| 1214 |
+
},
|
| 1215 |
+
{
|
| 1216 |
+
"epoch": 3.4565434565434563,
|
| 1217 |
+
"grad_norm": 0.6749796271324158,
|
| 1218 |
+
"learning_rate": 0.00014651510780774583,
|
| 1219 |
+
"loss": 0.4602,
|
| 1220 |
+
"step": 1730
|
| 1221 |
+
},
|
| 1222 |
+
{
|
| 1223 |
+
"epoch": 3.4765234765234765,
|
| 1224 |
+
"grad_norm": 0.6736605167388916,
|
| 1225 |
+
"learning_rate": 0.00014595798606214882,
|
| 1226 |
+
"loss": 0.4751,
|
| 1227 |
+
"step": 1740
|
| 1228 |
+
},
|
| 1229 |
+
{
|
| 1230 |
+
"epoch": 3.4965034965034967,
|
| 1231 |
+
"grad_norm": 0.7386316657066345,
|
| 1232 |
+
"learning_rate": 0.00014539904997395468,
|
| 1233 |
+
"loss": 0.4658,
|
| 1234 |
+
"step": 1750
|
| 1235 |
+
},
|
| 1236 |
+
{
|
| 1237 |
+
"epoch": 3.5164835164835164,
|
| 1238 |
+
"grad_norm": 0.7023107409477234,
|
| 1239 |
+
"learning_rate": 0.00014483832160900326,
|
| 1240 |
+
"loss": 0.4678,
|
| 1241 |
+
"step": 1760
|
| 1242 |
+
},
|
| 1243 |
+
{
|
| 1244 |
+
"epoch": 3.5364635364635366,
|
| 1245 |
+
"grad_norm": 0.6938359141349792,
|
| 1246 |
+
"learning_rate": 0.0001442758231038902,
|
| 1247 |
+
"loss": 0.4619,
|
| 1248 |
+
"step": 1770
|
| 1249 |
+
},
|
| 1250 |
+
{
|
| 1251 |
+
"epoch": 3.5564435564435564,
|
| 1252 |
+
"grad_norm": 0.7815272212028503,
|
| 1253 |
+
"learning_rate": 0.0001437115766650933,
|
| 1254 |
+
"loss": 0.4744,
|
| 1255 |
+
"step": 1780
|
| 1256 |
+
},
|
| 1257 |
+
{
|
| 1258 |
+
"epoch": 3.5764235764235766,
|
| 1259 |
+
"grad_norm": 0.7307267189025879,
|
| 1260 |
+
"learning_rate": 0.0001431456045680959,
|
| 1261 |
+
"loss": 0.4767,
|
| 1262 |
+
"step": 1790
|
| 1263 |
+
},
|
| 1264 |
+
{
|
| 1265 |
+
"epoch": 3.5964035964035963,
|
| 1266 |
+
"grad_norm": 0.6948580741882324,
|
| 1267 |
+
"learning_rate": 0.00014257792915650728,
|
| 1268 |
+
"loss": 0.4644,
|
| 1269 |
+
"step": 1800
|
| 1270 |
+
},
|
| 1271 |
+
{
|
| 1272 |
+
"epoch": 3.6163836163836165,
|
| 1273 |
+
"grad_norm": 0.691348671913147,
|
| 1274 |
+
"learning_rate": 0.00014200857284118066,
|
| 1275 |
+
"loss": 0.4609,
|
| 1276 |
+
"step": 1810
|
| 1277 |
+
},
|
| 1278 |
+
{
|
| 1279 |
+
"epoch": 3.6363636363636362,
|
| 1280 |
+
"grad_norm": 0.7828198671340942,
|
| 1281 |
+
"learning_rate": 0.00014143755809932845,
|
| 1282 |
+
"loss": 0.4506,
|
| 1283 |
+
"step": 1820
|
| 1284 |
+
},
|
| 1285 |
+
{
|
| 1286 |
+
"epoch": 3.6563436563436564,
|
| 1287 |
+
"grad_norm": 0.73238205909729,
|
| 1288 |
+
"learning_rate": 0.00014086490747363493,
|
| 1289 |
+
"loss": 0.4599,
|
| 1290 |
+
"step": 1830
|
| 1291 |
+
},
|
| 1292 |
+
{
|
| 1293 |
+
"epoch": 3.676323676323676,
|
| 1294 |
+
"grad_norm": 0.7216520309448242,
|
| 1295 |
+
"learning_rate": 0.00014029064357136628,
|
| 1296 |
+
"loss": 0.4582,
|
| 1297 |
+
"step": 1840
|
| 1298 |
+
},
|
| 1299 |
+
{
|
| 1300 |
+
"epoch": 3.6963036963036964,
|
| 1301 |
+
"grad_norm": 0.7676394581794739,
|
| 1302 |
+
"learning_rate": 0.00013971478906347806,
|
| 1303 |
+
"loss": 0.4494,
|
| 1304 |
+
"step": 1850
|
| 1305 |
+
},
|
| 1306 |
+
{
|
| 1307 |
+
"epoch": 3.716283716283716,
|
| 1308 |
+
"grad_norm": 0.7596750259399414,
|
| 1309 |
+
"learning_rate": 0.00013913736668372026,
|
| 1310 |
+
"loss": 0.4704,
|
| 1311 |
+
"step": 1860
|
| 1312 |
+
},
|
| 1313 |
+
{
|
| 1314 |
+
"epoch": 3.7362637362637363,
|
| 1315 |
+
"grad_norm": 0.7686085104942322,
|
| 1316 |
+
"learning_rate": 0.00013855839922773968,
|
| 1317 |
+
"loss": 0.4603,
|
| 1318 |
+
"step": 1870
|
| 1319 |
+
},
|
| 1320 |
+
{
|
| 1321 |
+
"epoch": 3.756243756243756,
|
| 1322 |
+
"grad_norm": 0.6850613951683044,
|
| 1323 |
+
"learning_rate": 0.00013797790955218014,
|
| 1324 |
+
"loss": 0.4503,
|
| 1325 |
+
"step": 1880
|
| 1326 |
+
},
|
| 1327 |
+
{
|
| 1328 |
+
"epoch": 3.7762237762237763,
|
| 1329 |
+
"grad_norm": 0.721778392791748,
|
| 1330 |
+
"learning_rate": 0.00013739592057378003,
|
| 1331 |
+
"loss": 0.4713,
|
| 1332 |
+
"step": 1890
|
| 1333 |
+
},
|
| 1334 |
+
{
|
| 1335 |
+
"epoch": 3.7962037962037964,
|
| 1336 |
+
"grad_norm": 0.7122541069984436,
|
| 1337 |
+
"learning_rate": 0.00013681245526846783,
|
| 1338 |
+
"loss": 0.4664,
|
| 1339 |
+
"step": 1900
|
| 1340 |
+
},
|
| 1341 |
+
{
|
| 1342 |
+
"epoch": 3.816183816183816,
|
| 1343 |
+
"grad_norm": 0.7361748218536377,
|
| 1344 |
+
"learning_rate": 0.00013622753667045457,
|
| 1345 |
+
"loss": 0.4571,
|
| 1346 |
+
"step": 1910
|
| 1347 |
+
},
|
| 1348 |
+
{
|
| 1349 |
+
"epoch": 3.836163836163836,
|
| 1350 |
+
"grad_norm": 0.8220844864845276,
|
| 1351 |
+
"learning_rate": 0.00013564118787132506,
|
| 1352 |
+
"loss": 0.4521,
|
| 1353 |
+
"step": 1920
|
| 1354 |
+
},
|
| 1355 |
+
{
|
| 1356 |
+
"epoch": 3.856143856143856,
|
| 1357 |
+
"grad_norm": 0.7139246463775635,
|
| 1358 |
+
"learning_rate": 0.0001350534320191259,
|
| 1359 |
+
"loss": 0.4491,
|
| 1360 |
+
"step": 1930
|
| 1361 |
+
},
|
| 1362 |
+
{
|
| 1363 |
+
"epoch": 3.8761238761238763,
|
| 1364 |
+
"grad_norm": 0.7244653701782227,
|
| 1365 |
+
"learning_rate": 0.0001344642923174517,
|
| 1366 |
+
"loss": 0.4552,
|
| 1367 |
+
"step": 1940
|
| 1368 |
+
},
|
| 1369 |
+
{
|
| 1370 |
+
"epoch": 3.896103896103896,
|
| 1371 |
+
"grad_norm": 0.7056713700294495,
|
| 1372 |
+
"learning_rate": 0.00013387379202452917,
|
| 1373 |
+
"loss": 0.4548,
|
| 1374 |
+
"step": 1950
|
| 1375 |
+
},
|
| 1376 |
+
{
|
| 1377 |
+
"epoch": 3.916083916083916,
|
| 1378 |
+
"grad_norm": 0.7653645277023315,
|
| 1379 |
+
"learning_rate": 0.00013328195445229868,
|
| 1380 |
+
"loss": 0.4492,
|
| 1381 |
+
"step": 1960
|
| 1382 |
+
},
|
| 1383 |
+
{
|
| 1384 |
+
"epoch": 3.936063936063936,
|
| 1385 |
+
"grad_norm": 0.6818165183067322,
|
| 1386 |
+
"learning_rate": 0.00013268880296549425,
|
| 1387 |
+
"loss": 0.4463,
|
| 1388 |
+
"step": 1970
|
| 1389 |
+
},
|
| 1390 |
+
{
|
| 1391 |
+
"epoch": 3.956043956043956,
|
| 1392 |
+
"grad_norm": 0.687439501285553,
|
| 1393 |
+
"learning_rate": 0.00013209436098072095,
|
| 1394 |
+
"loss": 0.457,
|
| 1395 |
+
"step": 1980
|
| 1396 |
+
},
|
| 1397 |
+
{
|
| 1398 |
+
"epoch": 3.976023976023976,
|
| 1399 |
+
"grad_norm": 0.7704656720161438,
|
| 1400 |
+
"learning_rate": 0.0001314986519655305,
|
| 1401 |
+
"loss": 0.4522,
|
| 1402 |
+
"step": 1990
|
| 1403 |
+
},
|
| 1404 |
+
{
|
| 1405 |
+
"epoch": 3.996003996003996,
|
| 1406 |
+
"grad_norm": 0.7227702736854553,
|
| 1407 |
+
"learning_rate": 0.00013090169943749476,
|
| 1408 |
+
"loss": 0.4454,
|
| 1409 |
+
"step": 2000
|
| 1410 |
+
},
|
| 1411 |
+
{
|
| 1412 |
+
"epoch": 4.015984015984016,
|
| 1413 |
+
"grad_norm": 0.8689281344413757,
|
| 1414 |
+
"learning_rate": 0.00013030352696327742,
|
| 1415 |
+
"loss": 0.3645,
|
| 1416 |
+
"step": 2010
|
| 1417 |
+
},
|
| 1418 |
+
{
|
| 1419 |
+
"epoch": 4.035964035964036,
|
| 1420 |
+
"grad_norm": 0.7620906829833984,
|
| 1421 |
+
"learning_rate": 0.0001297041581577035,
|
| 1422 |
+
"loss": 0.3478,
|
| 1423 |
+
"step": 2020
|
| 1424 |
+
},
|
| 1425 |
+
{
|
| 1426 |
+
"epoch": 4.055944055944056,
|
| 1427 |
+
"grad_norm": 0.768671989440918,
|
| 1428 |
+
"learning_rate": 0.00012910361668282719,
|
| 1429 |
+
"loss": 0.3595,
|
| 1430 |
+
"step": 2030
|
| 1431 |
+
},
|
| 1432 |
+
{
|
| 1433 |
+
"epoch": 4.075924075924076,
|
| 1434 |
+
"grad_norm": 0.7327402234077454,
|
| 1435 |
+
"learning_rate": 0.0001285019262469976,
|
| 1436 |
+
"loss": 0.3471,
|
| 1437 |
+
"step": 2040
|
| 1438 |
+
},
|
| 1439 |
+
{
|
| 1440 |
+
"epoch": 4.095904095904096,
|
| 1441 |
+
"grad_norm": 0.6913720965385437,
|
| 1442 |
+
"learning_rate": 0.00012789911060392294,
|
| 1443 |
+
"loss": 0.3501,
|
| 1444 |
+
"step": 2050
|
| 1445 |
+
},
|
| 1446 |
+
{
|
| 1447 |
+
"epoch": 4.115884115884116,
|
| 1448 |
+
"grad_norm": 0.7310584783554077,
|
| 1449 |
+
"learning_rate": 0.00012729519355173254,
|
| 1450 |
+
"loss": 0.3509,
|
| 1451 |
+
"step": 2060
|
| 1452 |
+
},
|
| 1453 |
+
{
|
| 1454 |
+
"epoch": 4.135864135864136,
|
| 1455 |
+
"grad_norm": 0.7578213214874268,
|
| 1456 |
+
"learning_rate": 0.00012669019893203759,
|
| 1457 |
+
"loss": 0.3506,
|
| 1458 |
+
"step": 2070
|
| 1459 |
+
},
|
| 1460 |
+
{
|
| 1461 |
+
"epoch": 4.1558441558441555,
|
| 1462 |
+
"grad_norm": 0.7301665544509888,
|
| 1463 |
+
"learning_rate": 0.00012608415062898972,
|
| 1464 |
+
"loss": 0.3536,
|
| 1465 |
+
"step": 2080
|
| 1466 |
+
},
|
| 1467 |
+
{
|
| 1468 |
+
"epoch": 4.175824175824176,
|
| 1469 |
+
"grad_norm": 0.8198577165603638,
|
| 1470 |
+
"learning_rate": 0.00012547707256833823,
|
| 1471 |
+
"loss": 0.3578,
|
| 1472 |
+
"step": 2090
|
| 1473 |
+
},
|
| 1474 |
+
{
|
| 1475 |
+
"epoch": 4.195804195804196,
|
| 1476 |
+
"grad_norm": 0.7331268787384033,
|
| 1477 |
+
"learning_rate": 0.0001248689887164855,
|
| 1478 |
+
"loss": 0.3508,
|
| 1479 |
+
"step": 2100
|
| 1480 |
+
},
|
| 1481 |
+
{
|
| 1482 |
+
"epoch": 4.215784215784216,
|
| 1483 |
+
"grad_norm": 0.7666186094284058,
|
| 1484 |
+
"learning_rate": 0.00012425992307954075,
|
| 1485 |
+
"loss": 0.3468,
|
| 1486 |
+
"step": 2110
|
| 1487 |
+
},
|
| 1488 |
+
{
|
| 1489 |
+
"epoch": 4.235764235764236,
|
| 1490 |
+
"grad_norm": 0.7020666599273682,
|
| 1491 |
+
"learning_rate": 0.00012364989970237248,
|
| 1492 |
+
"loss": 0.3586,
|
| 1493 |
+
"step": 2120
|
| 1494 |
+
},
|
| 1495 |
+
{
|
| 1496 |
+
"epoch": 4.255744255744256,
|
| 1497 |
+
"grad_norm": 0.7276338338851929,
|
| 1498 |
+
"learning_rate": 0.00012303894266765908,
|
| 1499 |
+
"loss": 0.3672,
|
| 1500 |
+
"step": 2130
|
| 1501 |
+
},
|
| 1502 |
+
{
|
| 1503 |
+
"epoch": 4.275724275724276,
|
| 1504 |
+
"grad_norm": 0.6978778839111328,
|
| 1505 |
+
"learning_rate": 0.00012242707609493814,
|
| 1506 |
+
"loss": 0.3576,
|
| 1507 |
+
"step": 2140
|
| 1508 |
+
},
|
| 1509 |
+
{
|
| 1510 |
+
"epoch": 4.2957042957042955,
|
| 1511 |
+
"grad_norm": 0.822030246257782,
|
| 1512 |
+
"learning_rate": 0.00012181432413965428,
|
| 1513 |
+
"loss": 0.3618,
|
| 1514 |
+
"step": 2150
|
| 1515 |
+
},
|
| 1516 |
+
{
|
| 1517 |
+
"epoch": 4.315684315684316,
|
| 1518 |
+
"grad_norm": 0.744611918926239,
|
| 1519 |
+
"learning_rate": 0.00012120071099220549,
|
| 1520 |
+
"loss": 0.3578,
|
| 1521 |
+
"step": 2160
|
| 1522 |
+
},
|
| 1523 |
+
{
|
| 1524 |
+
"epoch": 4.335664335664336,
|
| 1525 |
+
"grad_norm": 0.7712835669517517,
|
| 1526 |
+
"learning_rate": 0.00012058626087698814,
|
| 1527 |
+
"loss": 0.3632,
|
| 1528 |
+
"step": 2170
|
| 1529 |
+
},
|
| 1530 |
+
{
|
| 1531 |
+
"epoch": 4.355644355644356,
|
| 1532 |
+
"grad_norm": 0.7824398279190063,
|
| 1533 |
+
"learning_rate": 0.00011997099805144069,
|
| 1534 |
+
"loss": 0.36,
|
| 1535 |
+
"step": 2180
|
| 1536 |
+
},
|
| 1537 |
+
{
|
| 1538 |
+
"epoch": 4.375624375624375,
|
| 1539 |
+
"grad_norm": 0.8473492860794067,
|
| 1540 |
+
"learning_rate": 0.00011935494680508606,
|
| 1541 |
+
"loss": 0.3645,
|
| 1542 |
+
"step": 2190
|
| 1543 |
+
},
|
| 1544 |
+
{
|
| 1545 |
+
"epoch": 4.395604395604396,
|
| 1546 |
+
"grad_norm": 0.7394326329231262,
|
| 1547 |
+
"learning_rate": 0.00011873813145857249,
|
| 1548 |
+
"loss": 0.3604,
|
| 1549 |
+
"step": 2200
|
| 1550 |
+
},
|
| 1551 |
+
{
|
| 1552 |
+
"epoch": 4.415584415584416,
|
| 1553 |
+
"grad_norm": 0.763633131980896,
|
| 1554 |
+
"learning_rate": 0.00011812057636271374,
|
| 1555 |
+
"loss": 0.3634,
|
| 1556 |
+
"step": 2210
|
| 1557 |
+
},
|
| 1558 |
+
{
|
| 1559 |
+
"epoch": 4.4355644355644355,
|
| 1560 |
+
"grad_norm": 0.7612594962120056,
|
| 1561 |
+
"learning_rate": 0.00011750230589752762,
|
| 1562 |
+
"loss": 0.355,
|
| 1563 |
+
"step": 2220
|
| 1564 |
+
},
|
| 1565 |
+
{
|
| 1566 |
+
"epoch": 4.455544455544455,
|
| 1567 |
+
"grad_norm": 0.7789061665534973,
|
| 1568 |
+
"learning_rate": 0.00011688334447127338,
|
| 1569 |
+
"loss": 0.3629,
|
| 1570 |
+
"step": 2230
|
| 1571 |
+
},
|
| 1572 |
+
{
|
| 1573 |
+
"epoch": 4.475524475524476,
|
| 1574 |
+
"grad_norm": 0.7422770261764526,
|
| 1575 |
+
"learning_rate": 0.00011626371651948838,
|
| 1576 |
+
"loss": 0.361,
|
| 1577 |
+
"step": 2240
|
| 1578 |
+
},
|
| 1579 |
+
{
|
| 1580 |
+
"epoch": 4.495504495504496,
|
| 1581 |
+
"grad_norm": 0.7636354565620422,
|
| 1582 |
+
"learning_rate": 0.0001156434465040231,
|
| 1583 |
+
"loss": 0.3593,
|
| 1584 |
+
"step": 2250
|
| 1585 |
+
},
|
| 1586 |
+
{
|
| 1587 |
+
"epoch": 4.515484515484515,
|
| 1588 |
+
"grad_norm": 0.7884863615036011,
|
| 1589 |
+
"learning_rate": 0.00011502255891207572,
|
| 1590 |
+
"loss": 0.3587,
|
| 1591 |
+
"step": 2260
|
| 1592 |
+
},
|
| 1593 |
+
{
|
| 1594 |
+
"epoch": 4.535464535464535,
|
| 1595 |
+
"grad_norm": 0.7233232855796814,
|
| 1596 |
+
"learning_rate": 0.00011440107825522521,
|
| 1597 |
+
"loss": 0.3577,
|
| 1598 |
+
"step": 2270
|
| 1599 |
+
},
|
| 1600 |
+
{
|
| 1601 |
+
"epoch": 4.555444555444556,
|
| 1602 |
+
"grad_norm": 0.8420186638832092,
|
| 1603 |
+
"learning_rate": 0.0001137790290684638,
|
| 1604 |
+
"loss": 0.3686,
|
| 1605 |
+
"step": 2280
|
| 1606 |
+
},
|
| 1607 |
+
{
|
| 1608 |
+
"epoch": 4.5754245754245755,
|
| 1609 |
+
"grad_norm": 0.7679941654205322,
|
| 1610 |
+
"learning_rate": 0.00011315643590922827,
|
| 1611 |
+
"loss": 0.3539,
|
| 1612 |
+
"step": 2290
|
| 1613 |
+
},
|
| 1614 |
+
{
|
| 1615 |
+
"epoch": 4.595404595404595,
|
| 1616 |
+
"grad_norm": 0.826885461807251,
|
| 1617 |
+
"learning_rate": 0.00011253332335643043,
|
| 1618 |
+
"loss": 0.3627,
|
| 1619 |
+
"step": 2300
|
| 1620 |
+
},
|
| 1621 |
+
{
|
| 1622 |
+
"epoch": 4.615384615384615,
|
| 1623 |
+
"grad_norm": 0.7590234875679016,
|
| 1624 |
+
"learning_rate": 0.00011190971600948699,
|
| 1625 |
+
"loss": 0.3613,
|
| 1626 |
+
"step": 2310
|
| 1627 |
+
},
|
| 1628 |
+
{
|
| 1629 |
+
"epoch": 4.635364635364636,
|
| 1630 |
+
"grad_norm": 0.7376580238342285,
|
| 1631 |
+
"learning_rate": 0.00011128563848734816,
|
| 1632 |
+
"loss": 0.3694,
|
| 1633 |
+
"step": 2320
|
| 1634 |
+
},
|
| 1635 |
+
{
|
| 1636 |
+
"epoch": 4.655344655344655,
|
| 1637 |
+
"grad_norm": 0.7795658111572266,
|
| 1638 |
+
"learning_rate": 0.000110661115427526,
|
| 1639 |
+
"loss": 0.3598,
|
| 1640 |
+
"step": 2330
|
| 1641 |
+
},
|
| 1642 |
+
{
|
| 1643 |
+
"epoch": 4.675324675324675,
|
| 1644 |
+
"grad_norm": 0.7736489176750183,
|
| 1645 |
+
"learning_rate": 0.00011003617148512149,
|
| 1646 |
+
"loss": 0.3598,
|
| 1647 |
+
"step": 2340
|
| 1648 |
+
},
|
| 1649 |
+
{
|
| 1650 |
+
"epoch": 4.695304695304696,
|
| 1651 |
+
"grad_norm": 0.757072925567627,
|
| 1652 |
+
"learning_rate": 0.00010941083133185146,
|
| 1653 |
+
"loss": 0.366,
|
| 1654 |
+
"step": 2350
|
| 1655 |
+
},
|
| 1656 |
+
{
|
| 1657 |
+
"epoch": 4.7152847152847155,
|
| 1658 |
+
"grad_norm": 0.8167831301689148,
|
| 1659 |
+
"learning_rate": 0.00010878511965507434,
|
| 1660 |
+
"loss": 0.3633,
|
| 1661 |
+
"step": 2360
|
| 1662 |
+
},
|
| 1663 |
+
{
|
| 1664 |
+
"epoch": 4.735264735264735,
|
| 1665 |
+
"grad_norm": 0.8083499670028687,
|
| 1666 |
+
"learning_rate": 0.00010815906115681578,
|
| 1667 |
+
"loss": 0.3562,
|
| 1668 |
+
"step": 2370
|
| 1669 |
+
},
|
| 1670 |
+
{
|
| 1671 |
+
"epoch": 4.755244755244755,
|
| 1672 |
+
"grad_norm": 0.7758758068084717,
|
| 1673 |
+
"learning_rate": 0.00010753268055279329,
|
| 1674 |
+
"loss": 0.3614,
|
| 1675 |
+
"step": 2380
|
| 1676 |
+
},
|
| 1677 |
+
{
|
| 1678 |
+
"epoch": 4.775224775224775,
|
| 1679 |
+
"grad_norm": 0.8572462797164917,
|
| 1680 |
+
"learning_rate": 0.00010690600257144061,
|
| 1681 |
+
"loss": 0.3652,
|
| 1682 |
+
"step": 2390
|
| 1683 |
+
},
|
| 1684 |
+
{
|
| 1685 |
+
"epoch": 4.795204795204795,
|
| 1686 |
+
"grad_norm": 0.8319938778877258,
|
| 1687 |
+
"learning_rate": 0.00010627905195293135,
|
| 1688 |
+
"loss": 0.3622,
|
| 1689 |
+
"step": 2400
|
| 1690 |
+
},
|
| 1691 |
+
{
|
| 1692 |
+
"epoch": 4.815184815184815,
|
| 1693 |
+
"grad_norm": 0.8004459142684937,
|
| 1694 |
+
"learning_rate": 0.00010565185344820247,
|
| 1695 |
+
"loss": 0.3604,
|
| 1696 |
+
"step": 2410
|
| 1697 |
+
},
|
| 1698 |
+
{
|
| 1699 |
+
"epoch": 4.835164835164835,
|
| 1700 |
+
"grad_norm": 0.790908694267273,
|
| 1701 |
+
"learning_rate": 0.00010502443181797697,
|
| 1702 |
+
"loss": 0.3587,
|
| 1703 |
+
"step": 2420
|
| 1704 |
+
},
|
| 1705 |
+
{
|
| 1706 |
+
"epoch": 4.8551448551448555,
|
| 1707 |
+
"grad_norm": 0.7726609110832214,
|
| 1708 |
+
"learning_rate": 0.0001043968118317865,
|
| 1709 |
+
"loss": 0.364,
|
| 1710 |
+
"step": 2430
|
| 1711 |
+
},
|
| 1712 |
+
{
|
| 1713 |
+
"epoch": 4.875124875124875,
|
| 1714 |
+
"grad_norm": 0.7808167338371277,
|
| 1715 |
+
"learning_rate": 0.00010376901826699348,
|
| 1716 |
+
"loss": 0.3637,
|
| 1717 |
+
"step": 2440
|
| 1718 |
+
},
|
| 1719 |
+
{
|
| 1720 |
+
"epoch": 4.895104895104895,
|
| 1721 |
+
"grad_norm": 0.8596636652946472,
|
| 1722 |
+
"learning_rate": 0.00010314107590781284,
|
| 1723 |
+
"loss": 0.3536,
|
| 1724 |
+
"step": 2450
|
| 1725 |
+
},
|
| 1726 |
+
{
|
| 1727 |
+
"epoch": 4.915084915084915,
|
| 1728 |
+
"grad_norm": 0.8091081380844116,
|
| 1729 |
+
"learning_rate": 0.00010251300954433376,
|
| 1730 |
+
"loss": 0.3522,
|
| 1731 |
+
"step": 2460
|
| 1732 |
+
},
|
| 1733 |
+
{
|
| 1734 |
+
"epoch": 4.935064935064935,
|
| 1735 |
+
"grad_norm": 0.8672420978546143,
|
| 1736 |
+
"learning_rate": 0.00010188484397154084,
|
| 1737 |
+
"loss": 0.3643,
|
| 1738 |
+
"step": 2470
|
| 1739 |
+
},
|
| 1740 |
+
{
|
| 1741 |
+
"epoch": 4.955044955044955,
|
| 1742 |
+
"grad_norm": 0.7860444188117981,
|
| 1743 |
+
"learning_rate": 0.00010125660398833528,
|
| 1744 |
+
"loss": 0.3493,
|
| 1745 |
+
"step": 2480
|
| 1746 |
+
},
|
| 1747 |
+
{
|
| 1748 |
+
"epoch": 4.975024975024975,
|
| 1749 |
+
"grad_norm": 0.7510725259780884,
|
| 1750 |
+
"learning_rate": 0.00010062831439655591,
|
| 1751 |
+
"loss": 0.3497,
|
| 1752 |
+
"step": 2490
|
| 1753 |
+
},
|
| 1754 |
+
{
|
| 1755 |
+
"epoch": 4.995004995004995,
|
| 1756 |
+
"grad_norm": 0.7850112915039062,
|
| 1757 |
+
"learning_rate": 0.0001,
|
| 1758 |
+
"loss": 0.361,
|
| 1759 |
+
"step": 2500
|
| 1760 |
+
},
|
| 1761 |
+
{
|
| 1762 |
+
"epoch": 5.014985014985015,
|
| 1763 |
+
"grad_norm": 0.9001740217208862,
|
| 1764 |
+
"learning_rate": 9.937168560344412e-05,
|
| 1765 |
+
"loss": 0.2983,
|
| 1766 |
+
"step": 2510
|
| 1767 |
+
},
|
| 1768 |
+
{
|
| 1769 |
+
"epoch": 5.034965034965035,
|
| 1770 |
+
"grad_norm": 0.683803141117096,
|
| 1771 |
+
"learning_rate": 9.874339601166473e-05,
|
| 1772 |
+
"loss": 0.2805,
|
| 1773 |
+
"step": 2520
|
| 1774 |
+
},
|
| 1775 |
+
{
|
| 1776 |
+
"epoch": 5.054945054945055,
|
| 1777 |
+
"grad_norm": 0.7267177700996399,
|
| 1778 |
+
"learning_rate": 9.81151560284592e-05,
|
| 1779 |
+
"loss": 0.2751,
|
| 1780 |
+
"step": 2530
|
| 1781 |
+
},
|
| 1782 |
+
{
|
| 1783 |
+
"epoch": 5.0749250749250745,
|
| 1784 |
+
"grad_norm": 0.7268999814987183,
|
| 1785 |
+
"learning_rate": 9.748699045566626e-05,
|
| 1786 |
+
"loss": 0.2805,
|
| 1787 |
+
"step": 2540
|
| 1788 |
+
},
|
| 1789 |
+
{
|
| 1790 |
+
"epoch": 5.094905094905095,
|
| 1791 |
+
"grad_norm": 0.6958262324333191,
|
| 1792 |
+
"learning_rate": 9.685892409218717e-05,
|
| 1793 |
+
"loss": 0.2808,
|
| 1794 |
+
"step": 2550
|
| 1795 |
+
},
|
| 1796 |
+
{
|
| 1797 |
+
"epoch": 5.114885114885115,
|
| 1798 |
+
"grad_norm": 0.7481863498687744,
|
| 1799 |
+
"learning_rate": 9.623098173300654e-05,
|
| 1800 |
+
"loss": 0.2808,
|
| 1801 |
+
"step": 2560
|
| 1802 |
+
},
|
| 1803 |
+
{
|
| 1804 |
+
"epoch": 5.134865134865135,
|
| 1805 |
+
"grad_norm": 0.6923096179962158,
|
| 1806 |
+
"learning_rate": 9.560318816821353e-05,
|
| 1807 |
+
"loss": 0.2802,
|
| 1808 |
+
"step": 2570
|
| 1809 |
+
},
|
| 1810 |
+
{
|
| 1811 |
+
"epoch": 5.154845154845155,
|
| 1812 |
+
"grad_norm": 0.8236074447631836,
|
| 1813 |
+
"learning_rate": 9.497556818202306e-05,
|
| 1814 |
+
"loss": 0.2845,
|
| 1815 |
+
"step": 2580
|
| 1816 |
+
},
|
| 1817 |
+
{
|
| 1818 |
+
"epoch": 5.174825174825175,
|
| 1819 |
+
"grad_norm": 0.7225534915924072,
|
| 1820 |
+
"learning_rate": 9.434814655179755e-05,
|
| 1821 |
+
"loss": 0.2802,
|
| 1822 |
+
"step": 2590
|
| 1823 |
+
},
|
| 1824 |
+
{
|
| 1825 |
+
"epoch": 5.194805194805195,
|
| 1826 |
+
"grad_norm": 0.7639855146408081,
|
| 1827 |
+
"learning_rate": 9.372094804706867e-05,
|
| 1828 |
+
"loss": 0.2846,
|
| 1829 |
+
"step": 2600
|
| 1830 |
+
},
|
| 1831 |
+
{
|
| 1832 |
+
"epoch": 5.2147852147852145,
|
| 1833 |
+
"grad_norm": 0.7572929859161377,
|
| 1834 |
+
"learning_rate": 9.309399742855942e-05,
|
| 1835 |
+
"loss": 0.2826,
|
| 1836 |
+
"step": 2610
|
| 1837 |
+
},
|
| 1838 |
+
{
|
| 1839 |
+
"epoch": 5.234765234765235,
|
| 1840 |
+
"grad_norm": 0.8045923709869385,
|
| 1841 |
+
"learning_rate": 9.246731944720675e-05,
|
| 1842 |
+
"loss": 0.2862,
|
| 1843 |
+
"step": 2620
|
| 1844 |
+
},
|
| 1845 |
+
{
|
| 1846 |
+
"epoch": 5.254745254745255,
|
| 1847 |
+
"grad_norm": 0.7385067939758301,
|
| 1848 |
+
"learning_rate": 9.184093884318425e-05,
|
| 1849 |
+
"loss": 0.2886,
|
| 1850 |
+
"step": 2630
|
| 1851 |
+
},
|
| 1852 |
+
{
|
| 1853 |
+
"epoch": 5.274725274725275,
|
| 1854 |
+
"grad_norm": 0.7742624282836914,
|
| 1855 |
+
"learning_rate": 9.121488034492569e-05,
|
| 1856 |
+
"loss": 0.2857,
|
| 1857 |
+
"step": 2640
|
| 1858 |
+
},
|
| 1859 |
+
{
|
| 1860 |
+
"epoch": 5.294705294705294,
|
| 1861 |
+
"grad_norm": 0.73873370885849,
|
| 1862 |
+
"learning_rate": 9.058916866814858e-05,
|
| 1863 |
+
"loss": 0.2874,
|
| 1864 |
+
"step": 2650
|
| 1865 |
+
},
|
| 1866 |
+
{
|
| 1867 |
+
"epoch": 5.314685314685315,
|
| 1868 |
+
"grad_norm": 0.8087053298950195,
|
| 1869 |
+
"learning_rate": 8.99638285148785e-05,
|
| 1870 |
+
"loss": 0.2814,
|
| 1871 |
+
"step": 2660
|
| 1872 |
+
},
|
| 1873 |
+
{
|
| 1874 |
+
"epoch": 5.334665334665335,
|
| 1875 |
+
"grad_norm": 0.7873129844665527,
|
| 1876 |
+
"learning_rate": 8.933888457247402e-05,
|
| 1877 |
+
"loss": 0.2827,
|
| 1878 |
+
"step": 2670
|
| 1879 |
+
},
|
| 1880 |
+
{
|
| 1881 |
+
"epoch": 5.3546453546453545,
|
| 1882 |
+
"grad_norm": 0.776678204536438,
|
| 1883 |
+
"learning_rate": 8.871436151265184e-05,
|
| 1884 |
+
"loss": 0.2861,
|
| 1885 |
+
"step": 2680
|
| 1886 |
+
},
|
| 1887 |
+
{
|
| 1888 |
+
"epoch": 5.374625374625374,
|
| 1889 |
+
"grad_norm": 0.7478957772254944,
|
| 1890 |
+
"learning_rate": 8.809028399051302e-05,
|
| 1891 |
+
"loss": 0.2841,
|
| 1892 |
+
"step": 2690
|
| 1893 |
+
},
|
| 1894 |
+
{
|
| 1895 |
+
"epoch": 5.394605394605395,
|
| 1896 |
+
"grad_norm": 0.7491159439086914,
|
| 1897 |
+
"learning_rate": 8.746667664356956e-05,
|
| 1898 |
+
"loss": 0.2781,
|
| 1899 |
+
"step": 2700
|
| 1900 |
+
},
|
| 1901 |
+
{
|
| 1902 |
+
"epoch": 5.414585414585415,
|
| 1903 |
+
"grad_norm": 0.7022270560264587,
|
| 1904 |
+
"learning_rate": 8.684356409077176e-05,
|
| 1905 |
+
"loss": 0.2831,
|
| 1906 |
+
"step": 2710
|
| 1907 |
+
},
|
| 1908 |
+
{
|
| 1909 |
+
"epoch": 5.434565434565434,
|
| 1910 |
+
"grad_norm": 0.714643120765686,
|
| 1911 |
+
"learning_rate": 8.62209709315362e-05,
|
| 1912 |
+
"loss": 0.2816,
|
| 1913 |
+
"step": 2720
|
| 1914 |
+
},
|
| 1915 |
+
{
|
| 1916 |
+
"epoch": 5.454545454545454,
|
| 1917 |
+
"grad_norm": 0.7695267796516418,
|
| 1918 |
+
"learning_rate": 8.559892174477479e-05,
|
| 1919 |
+
"loss": 0.2845,
|
| 1920 |
+
"step": 2730
|
| 1921 |
+
},
|
| 1922 |
+
{
|
| 1923 |
+
"epoch": 5.474525474525475,
|
| 1924 |
+
"grad_norm": 0.7670512795448303,
|
| 1925 |
+
"learning_rate": 8.497744108792429e-05,
|
| 1926 |
+
"loss": 0.284,
|
| 1927 |
+
"step": 2740
|
| 1928 |
+
},
|
| 1929 |
+
{
|
| 1930 |
+
"epoch": 5.4945054945054945,
|
| 1931 |
+
"grad_norm": 0.7777095437049866,
|
| 1932 |
+
"learning_rate": 8.435655349597689e-05,
|
| 1933 |
+
"loss": 0.2849,
|
| 1934 |
+
"step": 2750
|
| 1935 |
+
},
|
| 1936 |
+
{
|
| 1937 |
+
"epoch": 5.514485514485514,
|
| 1938 |
+
"grad_norm": 0.7117462158203125,
|
| 1939 |
+
"learning_rate": 8.373628348051165e-05,
|
| 1940 |
+
"loss": 0.2892,
|
| 1941 |
+
"step": 2760
|
| 1942 |
+
},
|
| 1943 |
+
{
|
| 1944 |
+
"epoch": 5.534465534465534,
|
| 1945 |
+
"grad_norm": 0.7786485552787781,
|
| 1946 |
+
"learning_rate": 8.311665552872662e-05,
|
| 1947 |
+
"loss": 0.2867,
|
| 1948 |
+
"step": 2770
|
| 1949 |
+
},
|
| 1950 |
+
{
|
| 1951 |
+
"epoch": 5.554445554445555,
|
| 1952 |
+
"grad_norm": 0.7926625609397888,
|
| 1953 |
+
"learning_rate": 8.249769410247239e-05,
|
| 1954 |
+
"loss": 0.2862,
|
| 1955 |
+
"step": 2780
|
| 1956 |
+
},
|
| 1957 |
+
{
|
| 1958 |
+
"epoch": 5.574425574425574,
|
| 1959 |
+
"grad_norm": 0.7426894307136536,
|
| 1960 |
+
"learning_rate": 8.187942363728625e-05,
|
| 1961 |
+
"loss": 0.288,
|
| 1962 |
+
"step": 2790
|
| 1963 |
+
},
|
| 1964 |
+
{
|
| 1965 |
+
"epoch": 5.594405594405594,
|
| 1966 |
+
"grad_norm": 0.7075335383415222,
|
| 1967 |
+
"learning_rate": 8.126186854142752e-05,
|
| 1968 |
+
"loss": 0.2847,
|
| 1969 |
+
"step": 2800
|
| 1970 |
+
},
|
| 1971 |
+
{
|
| 1972 |
+
"epoch": 5.614385614385615,
|
| 1973 |
+
"grad_norm": 0.7743814587593079,
|
| 1974 |
+
"learning_rate": 8.064505319491398e-05,
|
| 1975 |
+
"loss": 0.2912,
|
| 1976 |
+
"step": 2810
|
| 1977 |
+
},
|
| 1978 |
+
{
|
| 1979 |
+
"epoch": 5.6343656343656345,
|
| 1980 |
+
"grad_norm": 0.7679479122161865,
|
| 1981 |
+
"learning_rate": 8.002900194855932e-05,
|
| 1982 |
+
"loss": 0.2944,
|
| 1983 |
+
"step": 2820
|
| 1984 |
+
},
|
| 1985 |
+
{
|
| 1986 |
+
"epoch": 5.654345654345654,
|
| 1987 |
+
"grad_norm": 0.8007961511611938,
|
| 1988 |
+
"learning_rate": 7.941373912301189e-05,
|
| 1989 |
+
"loss": 0.2934,
|
| 1990 |
+
"step": 2830
|
| 1991 |
+
},
|
| 1992 |
+
{
|
| 1993 |
+
"epoch": 5.674325674325674,
|
| 1994 |
+
"grad_norm": 0.8405194878578186,
|
| 1995 |
+
"learning_rate": 7.879928900779456e-05,
|
| 1996 |
+
"loss": 0.2848,
|
| 1997 |
+
"step": 2840
|
| 1998 |
+
},
|
| 1999 |
+
{
|
| 2000 |
+
"epoch": 5.694305694305695,
|
| 2001 |
+
"grad_norm": 0.7828160524368286,
|
| 2002 |
+
"learning_rate": 7.818567586034577e-05,
|
| 2003 |
+
"loss": 0.2932,
|
| 2004 |
+
"step": 2850
|
| 2005 |
+
},
|
| 2006 |
+
{
|
| 2007 |
+
"epoch": 5.714285714285714,
|
| 2008 |
+
"grad_norm": 0.7869848608970642,
|
| 2009 |
+
"learning_rate": 7.75729239050619e-05,
|
| 2010 |
+
"loss": 0.2851,
|
| 2011 |
+
"step": 2860
|
| 2012 |
+
},
|
| 2013 |
+
{
|
| 2014 |
+
"epoch": 5.734265734265734,
|
| 2015 |
+
"grad_norm": 0.7781445980072021,
|
| 2016 |
+
"learning_rate": 7.696105733234098e-05,
|
| 2017 |
+
"loss": 0.2849,
|
| 2018 |
+
"step": 2870
|
| 2019 |
+
},
|
| 2020 |
+
{
|
| 2021 |
+
"epoch": 5.754245754245754,
|
| 2022 |
+
"grad_norm": 0.8406656980514526,
|
| 2023 |
+
"learning_rate": 7.635010029762756e-05,
|
| 2024 |
+
"loss": 0.2854,
|
| 2025 |
+
"step": 2880
|
| 2026 |
+
},
|
| 2027 |
+
{
|
| 2028 |
+
"epoch": 5.7742257742257745,
|
| 2029 |
+
"grad_norm": 0.7491788864135742,
|
| 2030 |
+
"learning_rate": 7.574007692045928e-05,
|
| 2031 |
+
"loss": 0.288,
|
| 2032 |
+
"step": 2890
|
| 2033 |
+
},
|
| 2034 |
+
{
|
| 2035 |
+
"epoch": 5.794205794205794,
|
| 2036 |
+
"grad_norm": 0.7962749004364014,
|
| 2037 |
+
"learning_rate": 7.513101128351454e-05,
|
| 2038 |
+
"loss": 0.2888,
|
| 2039 |
+
"step": 2900
|
| 2040 |
+
},
|
| 2041 |
+
{
|
| 2042 |
+
"epoch": 5.814185814185814,
|
| 2043 |
+
"grad_norm": 0.7898345589637756,
|
| 2044 |
+
"learning_rate": 7.45229274316618e-05,
|
| 2045 |
+
"loss": 0.2875,
|
| 2046 |
+
"step": 2910
|
| 2047 |
+
},
|
| 2048 |
+
{
|
| 2049 |
+
"epoch": 5.834165834165834,
|
| 2050 |
+
"grad_norm": 0.7886426448822021,
|
| 2051 |
+
"learning_rate": 7.391584937101033e-05,
|
| 2052 |
+
"loss": 0.2947,
|
| 2053 |
+
"step": 2920
|
| 2054 |
+
},
|
| 2055 |
+
{
|
| 2056 |
+
"epoch": 5.854145854145854,
|
| 2057 |
+
"grad_norm": 0.7488512396812439,
|
| 2058 |
+
"learning_rate": 7.330980106796246e-05,
|
| 2059 |
+
"loss": 0.2846,
|
| 2060 |
+
"step": 2930
|
| 2061 |
+
},
|
| 2062 |
+
{
|
| 2063 |
+
"epoch": 5.874125874125874,
|
| 2064 |
+
"grad_norm": 0.7348522543907166,
|
| 2065 |
+
"learning_rate": 7.270480644826749e-05,
|
| 2066 |
+
"loss": 0.2883,
|
| 2067 |
+
"step": 2940
|
| 2068 |
+
},
|
| 2069 |
+
{
|
| 2070 |
+
"epoch": 5.894105894105894,
|
| 2071 |
+
"grad_norm": 0.7618998885154724,
|
| 2072 |
+
"learning_rate": 7.210088939607708e-05,
|
| 2073 |
+
"loss": 0.2899,
|
| 2074 |
+
"step": 2950
|
| 2075 |
+
},
|
| 2076 |
+
{
|
| 2077 |
+
"epoch": 5.9140859140859146,
|
| 2078 |
+
"grad_norm": 0.78291255235672,
|
| 2079 |
+
"learning_rate": 7.149807375300239e-05,
|
| 2080 |
+
"loss": 0.2865,
|
| 2081 |
+
"step": 2960
|
| 2082 |
+
},
|
| 2083 |
+
{
|
| 2084 |
+
"epoch": 5.934065934065934,
|
| 2085 |
+
"grad_norm": 0.7446394562721252,
|
| 2086 |
+
"learning_rate": 7.089638331717284e-05,
|
| 2087 |
+
"loss": 0.2846,
|
| 2088 |
+
"step": 2970
|
| 2089 |
+
},
|
| 2090 |
+
{
|
| 2091 |
+
"epoch": 5.954045954045954,
|
| 2092 |
+
"grad_norm": 0.767301619052887,
|
| 2093 |
+
"learning_rate": 7.029584184229653e-05,
|
| 2094 |
+
"loss": 0.2887,
|
| 2095 |
+
"step": 2980
|
| 2096 |
+
},
|
| 2097 |
+
{
|
| 2098 |
+
"epoch": 5.974025974025974,
|
| 2099 |
+
"grad_norm": 0.7523135542869568,
|
| 2100 |
+
"learning_rate": 6.969647303672262e-05,
|
| 2101 |
+
"loss": 0.2873,
|
| 2102 |
+
"step": 2990
|
| 2103 |
+
},
|
| 2104 |
+
{
|
| 2105 |
+
"epoch": 5.9940059940059935,
|
| 2106 |
+
"grad_norm": 0.7532919049263,
|
| 2107 |
+
"learning_rate": 6.909830056250527e-05,
|
| 2108 |
+
"loss": 0.2882,
|
| 2109 |
+
"step": 3000
|
| 2110 |
+
},
|
| 2111 |
+
{
|
| 2112 |
+
"epoch": 6.013986013986014,
|
| 2113 |
+
"grad_norm": 0.6552711129188538,
|
| 2114 |
+
"learning_rate": 6.850134803446954e-05,
|
| 2115 |
+
"loss": 0.2488,
|
| 2116 |
+
"step": 3010
|
| 2117 |
+
},
|
| 2118 |
+
{
|
| 2119 |
+
"epoch": 6.033966033966034,
|
| 2120 |
+
"grad_norm": 0.6565443873405457,
|
| 2121 |
+
"learning_rate": 6.790563901927907e-05,
|
| 2122 |
+
"loss": 0.2345,
|
| 2123 |
+
"step": 3020
|
| 2124 |
+
},
|
| 2125 |
+
{
|
| 2126 |
+
"epoch": 6.053946053946054,
|
| 2127 |
+
"grad_norm": 0.6884881854057312,
|
| 2128 |
+
"learning_rate": 6.731119703450577e-05,
|
| 2129 |
+
"loss": 0.233,
|
| 2130 |
+
"step": 3030
|
| 2131 |
+
},
|
| 2132 |
+
{
|
| 2133 |
+
"epoch": 6.073926073926074,
|
| 2134 |
+
"grad_norm": 0.6287186741828918,
|
| 2135 |
+
"learning_rate": 6.671804554770135e-05,
|
| 2136 |
+
"loss": 0.2356,
|
| 2137 |
+
"step": 3040
|
| 2138 |
+
},
|
| 2139 |
+
{
|
| 2140 |
+
"epoch": 6.093906093906094,
|
| 2141 |
+
"grad_norm": 0.754036545753479,
|
| 2142 |
+
"learning_rate": 6.612620797547087e-05,
|
| 2143 |
+
"loss": 0.2352,
|
| 2144 |
+
"step": 3050
|
| 2145 |
+
},
|
| 2146 |
+
{
|
| 2147 |
+
"epoch": 6.113886113886114,
|
| 2148 |
+
"grad_norm": 0.6492979526519775,
|
| 2149 |
+
"learning_rate": 6.55357076825483e-05,
|
| 2150 |
+
"loss": 0.2329,
|
| 2151 |
+
"step": 3060
|
| 2152 |
+
},
|
| 2153 |
+
{
|
| 2154 |
+
"epoch": 6.1338661338661336,
|
| 2155 |
+
"grad_norm": 0.6303039789199829,
|
| 2156 |
+
"learning_rate": 6.494656798087412e-05,
|
| 2157 |
+
"loss": 0.2339,
|
| 2158 |
+
"step": 3070
|
| 2159 |
+
},
|
| 2160 |
+
{
|
| 2161 |
+
"epoch": 6.153846153846154,
|
| 2162 |
+
"grad_norm": 0.6423007845878601,
|
| 2163 |
+
"learning_rate": 6.435881212867493e-05,
|
| 2164 |
+
"loss": 0.2377,
|
| 2165 |
+
"step": 3080
|
| 2166 |
+
},
|
| 2167 |
+
{
|
| 2168 |
+
"epoch": 6.173826173826174,
|
| 2169 |
+
"grad_norm": 0.6716975569725037,
|
| 2170 |
+
"learning_rate": 6.377246332954544e-05,
|
| 2171 |
+
"loss": 0.2365,
|
| 2172 |
+
"step": 3090
|
| 2173 |
+
},
|
| 2174 |
+
{
|
| 2175 |
+
"epoch": 6.193806193806194,
|
| 2176 |
+
"grad_norm": 0.6927747130393982,
|
| 2177 |
+
"learning_rate": 6.318754473153221e-05,
|
| 2178 |
+
"loss": 0.2346,
|
| 2179 |
+
"step": 3100
|
| 2180 |
+
},
|
| 2181 |
+
{
|
| 2182 |
+
"epoch": 6.213786213786213,
|
| 2183 |
+
"grad_norm": 0.6551555395126343,
|
| 2184 |
+
"learning_rate": 6.260407942621998e-05,
|
| 2185 |
+
"loss": 0.235,
|
| 2186 |
+
"step": 3110
|
| 2187 |
+
},
|
| 2188 |
+
{
|
| 2189 |
+
"epoch": 6.233766233766234,
|
| 2190 |
+
"grad_norm": 0.7131916284561157,
|
| 2191 |
+
"learning_rate": 6.20220904478199e-05,
|
| 2192 |
+
"loss": 0.2401,
|
| 2193 |
+
"step": 3120
|
| 2194 |
+
},
|
| 2195 |
+
{
|
| 2196 |
+
"epoch": 6.253746253746254,
|
| 2197 |
+
"grad_norm": 0.7002174258232117,
|
| 2198 |
+
"learning_rate": 6.144160077226036e-05,
|
| 2199 |
+
"loss": 0.2398,
|
| 2200 |
+
"step": 3130
|
| 2201 |
+
},
|
| 2202 |
+
{
|
| 2203 |
+
"epoch": 6.273726273726274,
|
| 2204 |
+
"grad_norm": 0.7129354476928711,
|
| 2205 |
+
"learning_rate": 6.086263331627976e-05,
|
| 2206 |
+
"loss": 0.2401,
|
| 2207 |
+
"step": 3140
|
| 2208 |
+
},
|
| 2209 |
+
{
|
| 2210 |
+
"epoch": 6.293706293706293,
|
| 2211 |
+
"grad_norm": 0.6942778825759888,
|
| 2212 |
+
"learning_rate": 6.0285210936521955e-05,
|
| 2213 |
+
"loss": 0.2391,
|
| 2214 |
+
"step": 3150
|
| 2215 |
+
},
|
| 2216 |
+
{
|
| 2217 |
+
"epoch": 6.313686313686314,
|
| 2218 |
+
"grad_norm": 0.7181575298309326,
|
| 2219 |
+
"learning_rate": 5.9709356428633746e-05,
|
| 2220 |
+
"loss": 0.2434,
|
| 2221 |
+
"step": 3160
|
| 2222 |
+
},
|
| 2223 |
+
{
|
| 2224 |
+
"epoch": 6.333666333666334,
|
| 2225 |
+
"grad_norm": 0.720330536365509,
|
| 2226 |
+
"learning_rate": 5.913509252636511e-05,
|
| 2227 |
+
"loss": 0.2352,
|
| 2228 |
+
"step": 3170
|
| 2229 |
+
},
|
| 2230 |
+
{
|
| 2231 |
+
"epoch": 6.353646353646353,
|
| 2232 |
+
"grad_norm": 0.6518005728721619,
|
| 2233 |
+
"learning_rate": 5.856244190067159e-05,
|
| 2234 |
+
"loss": 0.2377,
|
| 2235 |
+
"step": 3180
|
| 2236 |
+
},
|
| 2237 |
+
{
|
| 2238 |
+
"epoch": 6.373626373626374,
|
| 2239 |
+
"grad_norm": 0.6705808639526367,
|
| 2240 |
+
"learning_rate": 5.799142715881938e-05,
|
| 2241 |
+
"loss": 0.2416,
|
| 2242 |
+
"step": 3190
|
| 2243 |
+
},
|
| 2244 |
+
{
|
| 2245 |
+
"epoch": 6.393606393606394,
|
| 2246 |
+
"grad_norm": 0.7210578322410583,
|
| 2247 |
+
"learning_rate": 5.7422070843492734e-05,
|
| 2248 |
+
"loss": 0.2406,
|
| 2249 |
+
"step": 3200
|
| 2250 |
+
},
|
| 2251 |
+
{
|
| 2252 |
+
"epoch": 6.413586413586414,
|
| 2253 |
+
"grad_norm": 0.6428204774856567,
|
| 2254 |
+
"learning_rate": 5.6854395431904094e-05,
|
| 2255 |
+
"loss": 0.2397,
|
| 2256 |
+
"step": 3210
|
| 2257 |
+
},
|
| 2258 |
+
{
|
| 2259 |
+
"epoch": 6.433566433566433,
|
| 2260 |
+
"grad_norm": 0.697733461856842,
|
| 2261 |
+
"learning_rate": 5.6288423334906735e-05,
|
| 2262 |
+
"loss": 0.2425,
|
| 2263 |
+
"step": 3220
|
| 2264 |
+
},
|
| 2265 |
+
{
|
| 2266 |
+
"epoch": 6.453546453546454,
|
| 2267 |
+
"grad_norm": 0.7867773175239563,
|
| 2268 |
+
"learning_rate": 5.572417689610987e-05,
|
| 2269 |
+
"loss": 0.2401,
|
| 2270 |
+
"step": 3230
|
| 2271 |
+
},
|
| 2272 |
+
{
|
| 2273 |
+
"epoch": 6.473526473526474,
|
| 2274 |
+
"grad_norm": 0.6750375032424927,
|
| 2275 |
+
"learning_rate": 5.5161678390996796e-05,
|
| 2276 |
+
"loss": 0.2396,
|
| 2277 |
+
"step": 3240
|
| 2278 |
+
},
|
| 2279 |
+
{
|
| 2280 |
+
"epoch": 6.4935064935064934,
|
| 2281 |
+
"grad_norm": 0.677237868309021,
|
| 2282 |
+
"learning_rate": 5.4600950026045326e-05,
|
| 2283 |
+
"loss": 0.2434,
|
| 2284 |
+
"step": 3250
|
| 2285 |
+
},
|
| 2286 |
+
{
|
| 2287 |
+
"epoch": 6.513486513486513,
|
| 2288 |
+
"grad_norm": 0.6781632304191589,
|
| 2289 |
+
"learning_rate": 5.404201393785122e-05,
|
| 2290 |
+
"loss": 0.2454,
|
| 2291 |
+
"step": 3260
|
| 2292 |
+
},
|
| 2293 |
+
{
|
| 2294 |
+
"epoch": 6.533466533466534,
|
| 2295 |
+
"grad_norm": 0.7506418824195862,
|
| 2296 |
+
"learning_rate": 5.348489219225416e-05,
|
| 2297 |
+
"loss": 0.2397,
|
| 2298 |
+
"step": 3270
|
| 2299 |
+
},
|
| 2300 |
+
{
|
| 2301 |
+
"epoch": 6.553446553446554,
|
| 2302 |
+
"grad_norm": 0.7256707549095154,
|
| 2303 |
+
"learning_rate": 5.292960678346675e-05,
|
| 2304 |
+
"loss": 0.2403,
|
| 2305 |
+
"step": 3280
|
| 2306 |
+
},
|
| 2307 |
+
{
|
| 2308 |
+
"epoch": 6.573426573426573,
|
| 2309 |
+
"grad_norm": 0.664169430732727,
|
| 2310 |
+
"learning_rate": 5.237617963320608e-05,
|
| 2311 |
+
"loss": 0.2392,
|
| 2312 |
+
"step": 3290
|
| 2313 |
+
},
|
| 2314 |
+
{
|
| 2315 |
+
"epoch": 6.593406593406593,
|
| 2316 |
+
"grad_norm": 0.7900999188423157,
|
| 2317 |
+
"learning_rate": 5.182463258982846e-05,
|
| 2318 |
+
"loss": 0.2426,
|
| 2319 |
+
"step": 3300
|
| 2320 |
+
},
|
| 2321 |
+
{
|
| 2322 |
+
"epoch": 6.613386613386614,
|
| 2323 |
+
"grad_norm": 0.7012047171592712,
|
| 2324 |
+
"learning_rate": 5.127498742746675e-05,
|
| 2325 |
+
"loss": 0.2429,
|
| 2326 |
+
"step": 3310
|
| 2327 |
+
},
|
| 2328 |
+
{
|
| 2329 |
+
"epoch": 6.6333666333666335,
|
| 2330 |
+
"grad_norm": 0.752498984336853,
|
| 2331 |
+
"learning_rate": 5.072726584517086e-05,
|
| 2332 |
+
"loss": 0.2425,
|
| 2333 |
+
"step": 3320
|
| 2334 |
+
},
|
| 2335 |
+
{
|
| 2336 |
+
"epoch": 6.653346653346653,
|
| 2337 |
+
"grad_norm": 0.7256404161453247,
|
| 2338 |
+
"learning_rate": 5.018148946605092e-05,
|
| 2339 |
+
"loss": 0.2381,
|
| 2340 |
+
"step": 3330
|
| 2341 |
+
},
|
| 2342 |
+
{
|
| 2343 |
+
"epoch": 6.673326673326673,
|
| 2344 |
+
"grad_norm": 0.6938993334770203,
|
| 2345 |
+
"learning_rate": 4.9637679836423924e-05,
|
| 2346 |
+
"loss": 0.2428,
|
| 2347 |
+
"step": 3340
|
| 2348 |
+
},
|
| 2349 |
+
{
|
| 2350 |
+
"epoch": 6.693306693306694,
|
| 2351 |
+
"grad_norm": 0.7288166284561157,
|
| 2352 |
+
"learning_rate": 4.909585842496287e-05,
|
| 2353 |
+
"loss": 0.2409,
|
| 2354 |
+
"step": 3350
|
| 2355 |
+
},
|
| 2356 |
+
{
|
| 2357 |
+
"epoch": 6.713286713286713,
|
| 2358 |
+
"grad_norm": 0.7148503661155701,
|
| 2359 |
+
"learning_rate": 4.8556046621849346e-05,
|
| 2360 |
+
"loss": 0.2402,
|
| 2361 |
+
"step": 3360
|
| 2362 |
+
},
|
| 2363 |
+
{
|
| 2364 |
+
"epoch": 6.733266733266733,
|
| 2365 |
+
"grad_norm": 0.7477458715438843,
|
| 2366 |
+
"learning_rate": 4.8018265737929044e-05,
|
| 2367 |
+
"loss": 0.2394,
|
| 2368 |
+
"step": 3370
|
| 2369 |
+
},
|
| 2370 |
+
{
|
| 2371 |
+
"epoch": 6.753246753246753,
|
| 2372 |
+
"grad_norm": 0.7404049634933472,
|
| 2373 |
+
"learning_rate": 4.748253700387042e-05,
|
| 2374 |
+
"loss": 0.2422,
|
| 2375 |
+
"step": 3380
|
| 2376 |
+
},
|
| 2377 |
+
{
|
| 2378 |
+
"epoch": 6.7732267732267735,
|
| 2379 |
+
"grad_norm": 0.6715726852416992,
|
| 2380 |
+
"learning_rate": 4.694888156932658e-05,
|
| 2381 |
+
"loss": 0.2405,
|
| 2382 |
+
"step": 3390
|
| 2383 |
+
},
|
| 2384 |
+
{
|
| 2385 |
+
"epoch": 6.793206793206793,
|
| 2386 |
+
"grad_norm": 0.6998412609100342,
|
| 2387 |
+
"learning_rate": 4.6417320502100316e-05,
|
| 2388 |
+
"loss": 0.2405,
|
| 2389 |
+
"step": 3400
|
| 2390 |
+
},
|
| 2391 |
+
{
|
| 2392 |
+
"epoch": 6.813186813186813,
|
| 2393 |
+
"grad_norm": 0.7061425447463989,
|
| 2394 |
+
"learning_rate": 4.588787478731242e-05,
|
| 2395 |
+
"loss": 0.2368,
|
| 2396 |
+
"step": 3410
|
| 2397 |
+
},
|
| 2398 |
+
{
|
| 2399 |
+
"epoch": 6.833166833166834,
|
| 2400 |
+
"grad_norm": 0.7432896494865417,
|
| 2401 |
+
"learning_rate": 4.5360565326573104e-05,
|
| 2402 |
+
"loss": 0.2399,
|
| 2403 |
+
"step": 3420
|
| 2404 |
+
},
|
| 2405 |
+
{
|
| 2406 |
+
"epoch": 6.853146853146853,
|
| 2407 |
+
"grad_norm": 0.7876798510551453,
|
| 2408 |
+
"learning_rate": 4.483541293715698e-05,
|
| 2409 |
+
"loss": 0.2395,
|
| 2410 |
+
"step": 3430
|
| 2411 |
+
},
|
| 2412 |
+
{
|
| 2413 |
+
"epoch": 6.873126873126873,
|
| 2414 |
+
"grad_norm": 0.7446125149726868,
|
| 2415 |
+
"learning_rate": 4.431243835118124e-05,
|
| 2416 |
+
"loss": 0.241,
|
| 2417 |
+
"step": 3440
|
| 2418 |
+
},
|
| 2419 |
+
{
|
| 2420 |
+
"epoch": 6.893106893106893,
|
| 2421 |
+
"grad_norm": 0.6832261085510254,
|
| 2422 |
+
"learning_rate": 4.379166221478697e-05,
|
| 2423 |
+
"loss": 0.2396,
|
| 2424 |
+
"step": 3450
|
| 2425 |
+
},
|
| 2426 |
+
{
|
| 2427 |
+
"epoch": 6.913086913086913,
|
| 2428 |
+
"grad_norm": 0.7039461135864258,
|
| 2429 |
+
"learning_rate": 4.327310508732437e-05,
|
| 2430 |
+
"loss": 0.2408,
|
| 2431 |
+
"step": 3460
|
| 2432 |
+
},
|
| 2433 |
+
{
|
| 2434 |
+
"epoch": 6.933066933066933,
|
| 2435 |
+
"grad_norm": 0.7428474426269531,
|
| 2436 |
+
"learning_rate": 4.2756787440540936e-05,
|
| 2437 |
+
"loss": 0.2407,
|
| 2438 |
+
"step": 3470
|
| 2439 |
+
},
|
| 2440 |
+
{
|
| 2441 |
+
"epoch": 6.953046953046953,
|
| 2442 |
+
"grad_norm": 0.7313565015792847,
|
| 2443 |
+
"learning_rate": 4.224272965777326e-05,
|
| 2444 |
+
"loss": 0.2406,
|
| 2445 |
+
"step": 3480
|
| 2446 |
+
},
|
| 2447 |
+
{
|
| 2448 |
+
"epoch": 6.973026973026973,
|
| 2449 |
+
"grad_norm": 0.7175894975662231,
|
| 2450 |
+
"learning_rate": 4.173095203314241e-05,
|
| 2451 |
+
"loss": 0.2409,
|
| 2452 |
+
"step": 3490
|
| 2453 |
+
},
|
| 2454 |
+
{
|
| 2455 |
+
"epoch": 6.993006993006993,
|
| 2456 |
+
"grad_norm": 0.6897133588790894,
|
| 2457 |
+
"learning_rate": 4.12214747707527e-05,
|
| 2458 |
+
"loss": 0.2389,
|
| 2459 |
+
"step": 3500
|
| 2460 |
+
},
|
| 2461 |
+
{
|
| 2462 |
+
"epoch": 7.012987012987013,
|
| 2463 |
+
"grad_norm": 0.5959777235984802,
|
| 2464 |
+
"learning_rate": 4.071431798389408e-05,
|
| 2465 |
+
"loss": 0.2184,
|
| 2466 |
+
"step": 3510
|
| 2467 |
+
},
|
| 2468 |
+
{
|
| 2469 |
+
"epoch": 7.032967032967033,
|
| 2470 |
+
"grad_norm": 0.7147582173347473,
|
| 2471 |
+
"learning_rate": 4.020950169424815e-05,
|
| 2472 |
+
"loss": 0.2087,
|
| 2473 |
+
"step": 3520
|
| 2474 |
+
},
|
| 2475 |
+
{
|
| 2476 |
+
"epoch": 7.052947052947053,
|
| 2477 |
+
"grad_norm": 0.6122413873672485,
|
| 2478 |
+
"learning_rate": 3.9707045831097555e-05,
|
| 2479 |
+
"loss": 0.2106,
|
| 2480 |
+
"step": 3530
|
| 2481 |
+
},
|
| 2482 |
+
{
|
| 2483 |
+
"epoch": 7.072927072927073,
|
| 2484 |
+
"grad_norm": 0.633969783782959,
|
| 2485 |
+
"learning_rate": 3.920697023053949e-05,
|
| 2486 |
+
"loss": 0.2099,
|
| 2487 |
+
"step": 3540
|
| 2488 |
+
},
|
| 2489 |
+
{
|
| 2490 |
+
"epoch": 7.092907092907093,
|
| 2491 |
+
"grad_norm": 0.6842843890190125,
|
| 2492 |
+
"learning_rate": 3.8709294634702376e-05,
|
| 2493 |
+
"loss": 0.2104,
|
| 2494 |
+
"step": 3550
|
| 2495 |
+
},
|
| 2496 |
+
{
|
| 2497 |
+
"epoch": 7.112887112887113,
|
| 2498 |
+
"grad_norm": 0.5708280205726624,
|
| 2499 |
+
"learning_rate": 3.821403869096658e-05,
|
| 2500 |
+
"loss": 0.2125,
|
| 2501 |
+
"step": 3560
|
| 2502 |
+
},
|
| 2503 |
+
{
|
| 2504 |
+
"epoch": 7.1328671328671325,
|
| 2505 |
+
"grad_norm": 0.6579930782318115,
|
| 2506 |
+
"learning_rate": 3.7721221951188765e-05,
|
| 2507 |
+
"loss": 0.2107,
|
| 2508 |
+
"step": 3570
|
| 2509 |
+
},
|
| 2510 |
+
{
|
| 2511 |
+
"epoch": 7.152847152847153,
|
| 2512 |
+
"grad_norm": 0.5980693101882935,
|
| 2513 |
+
"learning_rate": 3.7230863870929964e-05,
|
| 2514 |
+
"loss": 0.2085,
|
| 2515 |
+
"step": 3580
|
| 2516 |
+
},
|
| 2517 |
+
{
|
| 2518 |
+
"epoch": 7.172827172827173,
|
| 2519 |
+
"grad_norm": 0.5968551635742188,
|
| 2520 |
+
"learning_rate": 3.674298380868756e-05,
|
| 2521 |
+
"loss": 0.209,
|
| 2522 |
+
"step": 3590
|
| 2523 |
+
},
|
| 2524 |
+
{
|
| 2525 |
+
"epoch": 7.192807192807193,
|
| 2526 |
+
"grad_norm": 0.6218951940536499,
|
| 2527 |
+
"learning_rate": 3.6257601025131026e-05,
|
| 2528 |
+
"loss": 0.2095,
|
| 2529 |
+
"step": 3600
|
| 2530 |
+
},
|
| 2531 |
+
{
|
| 2532 |
+
"epoch": 7.212787212787212,
|
| 2533 |
+
"grad_norm": 0.6248393058776855,
|
| 2534 |
+
"learning_rate": 3.577473468234156e-05,
|
| 2535 |
+
"loss": 0.2155,
|
| 2536 |
+
"step": 3610
|
| 2537 |
+
},
|
| 2538 |
+
{
|
| 2539 |
+
"epoch": 7.232767232767233,
|
| 2540 |
+
"grad_norm": 0.6496105194091797,
|
| 2541 |
+
"learning_rate": 3.52944038430556e-05,
|
| 2542 |
+
"loss": 0.2139,
|
| 2543 |
+
"step": 3620
|
| 2544 |
+
},
|
| 2545 |
+
{
|
| 2546 |
+
"epoch": 7.252747252747253,
|
| 2547 |
+
"grad_norm": 0.6064103841781616,
|
| 2548 |
+
"learning_rate": 3.481662746991214e-05,
|
| 2549 |
+
"loss": 0.2081,
|
| 2550 |
+
"step": 3630
|
| 2551 |
+
},
|
| 2552 |
+
{
|
| 2553 |
+
"epoch": 7.2727272727272725,
|
| 2554 |
+
"grad_norm": 0.6504641771316528,
|
| 2555 |
+
"learning_rate": 3.4341424424704375e-05,
|
| 2556 |
+
"loss": 0.2111,
|
| 2557 |
+
"step": 3640
|
| 2558 |
+
},
|
| 2559 |
+
{
|
| 2560 |
+
"epoch": 7.292707292707293,
|
| 2561 |
+
"grad_norm": 0.6580168604850769,
|
| 2562 |
+
"learning_rate": 3.386881346763483e-05,
|
| 2563 |
+
"loss": 0.2123,
|
| 2564 |
+
"step": 3650
|
| 2565 |
+
},
|
| 2566 |
+
{
|
| 2567 |
+
"epoch": 7.312687312687313,
|
| 2568 |
+
"grad_norm": 0.5861549973487854,
|
| 2569 |
+
"learning_rate": 3.339881325657484e-05,
|
| 2570 |
+
"loss": 0.2084,
|
| 2571 |
+
"step": 3660
|
| 2572 |
+
},
|
| 2573 |
+
{
|
| 2574 |
+
"epoch": 7.332667332667333,
|
| 2575 |
+
"grad_norm": 0.6313382387161255,
|
| 2576 |
+
"learning_rate": 3.2931442346328004e-05,
|
| 2577 |
+
"loss": 0.2078,
|
| 2578 |
+
"step": 3670
|
| 2579 |
+
},
|
| 2580 |
+
{
|
| 2581 |
+
"epoch": 7.352647352647352,
|
| 2582 |
+
"grad_norm": 0.646842896938324,
|
| 2583 |
+
"learning_rate": 3.246671918789755e-05,
|
| 2584 |
+
"loss": 0.2135,
|
| 2585 |
+
"step": 3680
|
| 2586 |
+
},
|
| 2587 |
+
{
|
| 2588 |
+
"epoch": 7.372627372627373,
|
| 2589 |
+
"grad_norm": 0.6964268088340759,
|
| 2590 |
+
"learning_rate": 3.200466212775808e-05,
|
| 2591 |
+
"loss": 0.2126,
|
| 2592 |
+
"step": 3690
|
| 2593 |
+
},
|
| 2594 |
+
{
|
| 2595 |
+
"epoch": 7.392607392607393,
|
| 2596 |
+
"grad_norm": 0.6139673590660095,
|
| 2597 |
+
"learning_rate": 3.154528940713113e-05,
|
| 2598 |
+
"loss": 0.215,
|
| 2599 |
+
"step": 3700
|
| 2600 |
+
},
|
| 2601 |
+
{
|
| 2602 |
+
"epoch": 7.4125874125874125,
|
| 2603 |
+
"grad_norm": 0.6455628871917725,
|
| 2604 |
+
"learning_rate": 3.108861916126518e-05,
|
| 2605 |
+
"loss": 0.2114,
|
| 2606 |
+
"step": 3710
|
| 2607 |
+
},
|
| 2608 |
+
{
|
| 2609 |
+
"epoch": 7.432567432567432,
|
| 2610 |
+
"grad_norm": 0.6227108240127563,
|
| 2611 |
+
"learning_rate": 3.063466941871952e-05,
|
| 2612 |
+
"loss": 0.2114,
|
| 2613 |
+
"step": 3720
|
| 2614 |
+
},
|
| 2615 |
+
{
|
| 2616 |
+
"epoch": 7.452547452547453,
|
| 2617 |
+
"grad_norm": 0.5858675837516785,
|
| 2618 |
+
"learning_rate": 3.018345810065275e-05,
|
| 2619 |
+
"loss": 0.2107,
|
| 2620 |
+
"step": 3730
|
| 2621 |
+
},
|
| 2622 |
+
{
|
| 2623 |
+
"epoch": 7.472527472527473,
|
| 2624 |
+
"grad_norm": 0.6218124628067017,
|
| 2625 |
+
"learning_rate": 2.9735003020115092e-05,
|
| 2626 |
+
"loss": 0.2115,
|
| 2627 |
+
"step": 3740
|
| 2628 |
+
},
|
| 2629 |
+
{
|
| 2630 |
+
"epoch": 7.492507492507492,
|
| 2631 |
+
"grad_norm": 0.6510396003723145,
|
| 2632 |
+
"learning_rate": 2.9289321881345254e-05,
|
| 2633 |
+
"loss": 0.2124,
|
| 2634 |
+
"step": 3750
|
| 2635 |
+
},
|
| 2636 |
+
{
|
| 2637 |
+
"epoch": 7.512487512487512,
|
| 2638 |
+
"grad_norm": 0.6465820074081421,
|
| 2639 |
+
"learning_rate": 2.8846432279071467e-05,
|
| 2640 |
+
"loss": 0.2132,
|
| 2641 |
+
"step": 3760
|
| 2642 |
+
},
|
| 2643 |
+
{
|
| 2644 |
+
"epoch": 7.532467532467533,
|
| 2645 |
+
"grad_norm": 0.7002317905426025,
|
| 2646 |
+
"learning_rate": 2.840635169781688e-05,
|
| 2647 |
+
"loss": 0.2129,
|
| 2648 |
+
"step": 3770
|
| 2649 |
+
},
|
| 2650 |
+
{
|
| 2651 |
+
"epoch": 7.5524475524475525,
|
| 2652 |
+
"grad_norm": 0.647723913192749,
|
| 2653 |
+
"learning_rate": 2.7969097511209308e-05,
|
| 2654 |
+
"loss": 0.2136,
|
| 2655 |
+
"step": 3780
|
| 2656 |
+
},
|
| 2657 |
+
{
|
| 2658 |
+
"epoch": 7.572427572427572,
|
| 2659 |
+
"grad_norm": 0.5907153487205505,
|
| 2660 |
+
"learning_rate": 2.753468698129533e-05,
|
| 2661 |
+
"loss": 0.2115,
|
| 2662 |
+
"step": 3790
|
| 2663 |
+
},
|
| 2664 |
+
{
|
| 2665 |
+
"epoch": 7.592407592407593,
|
| 2666 |
+
"grad_norm": 0.6074231863021851,
|
| 2667 |
+
"learning_rate": 2.7103137257858868e-05,
|
| 2668 |
+
"loss": 0.2128,
|
| 2669 |
+
"step": 3800
|
| 2670 |
+
},
|
| 2671 |
+
{
|
| 2672 |
+
"epoch": 7.612387612387613,
|
| 2673 |
+
"grad_norm": 0.6356890797615051,
|
| 2674 |
+
"learning_rate": 2.6674465377744017e-05,
|
| 2675 |
+
"loss": 0.2108,
|
| 2676 |
+
"step": 3810
|
| 2677 |
+
},
|
| 2678 |
+
{
|
| 2679 |
+
"epoch": 7.632367632367632,
|
| 2680 |
+
"grad_norm": 0.6739248633384705,
|
| 2681 |
+
"learning_rate": 2.624868826418262e-05,
|
| 2682 |
+
"loss": 0.2129,
|
| 2683 |
+
"step": 3820
|
| 2684 |
+
},
|
| 2685 |
+
{
|
| 2686 |
+
"epoch": 7.652347652347652,
|
| 2687 |
+
"grad_norm": 0.6241906881332397,
|
| 2688 |
+
"learning_rate": 2.582582272612609e-05,
|
| 2689 |
+
"loss": 0.211,
|
| 2690 |
+
"step": 3830
|
| 2691 |
+
},
|
| 2692 |
+
{
|
| 2693 |
+
"epoch": 7.672327672327672,
|
| 2694 |
+
"grad_norm": 0.6532058715820312,
|
| 2695 |
+
"learning_rate": 2.540588545758179e-05,
|
| 2696 |
+
"loss": 0.2137,
|
| 2697 |
+
"step": 3840
|
| 2698 |
+
},
|
| 2699 |
+
{
|
| 2700 |
+
"epoch": 7.6923076923076925,
|
| 2701 |
+
"grad_norm": 0.7098828554153442,
|
| 2702 |
+
"learning_rate": 2.4988893036954043e-05,
|
| 2703 |
+
"loss": 0.2105,
|
| 2704 |
+
"step": 3850
|
| 2705 |
+
},
|
| 2706 |
+
{
|
| 2707 |
+
"epoch": 7.712287712287712,
|
| 2708 |
+
"grad_norm": 0.6868453025817871,
|
| 2709 |
+
"learning_rate": 2.4574861926389615e-05,
|
| 2710 |
+
"loss": 0.214,
|
| 2711 |
+
"step": 3860
|
| 2712 |
+
},
|
| 2713 |
+
{
|
| 2714 |
+
"epoch": 7.732267732267732,
|
| 2715 |
+
"grad_norm": 0.6777834296226501,
|
| 2716 |
+
"learning_rate": 2.4163808471127812e-05,
|
| 2717 |
+
"loss": 0.2125,
|
| 2718 |
+
"step": 3870
|
| 2719 |
+
},
|
| 2720 |
+
{
|
| 2721 |
+
"epoch": 7.752247752247753,
|
| 2722 |
+
"grad_norm": 0.6967138648033142,
|
| 2723 |
+
"learning_rate": 2.37557488988552e-05,
|
| 2724 |
+
"loss": 0.2118,
|
| 2725 |
+
"step": 3880
|
| 2726 |
+
},
|
| 2727 |
+
{
|
| 2728 |
+
"epoch": 7.772227772227772,
|
| 2729 |
+
"grad_norm": 0.6641217470169067,
|
| 2730 |
+
"learning_rate": 2.3350699319065026e-05,
|
| 2731 |
+
"loss": 0.2134,
|
| 2732 |
+
"step": 3890
|
| 2733 |
+
},
|
| 2734 |
+
{
|
| 2735 |
+
"epoch": 7.792207792207792,
|
| 2736 |
+
"grad_norm": 0.6727011799812317,
|
| 2737 |
+
"learning_rate": 2.2948675722421086e-05,
|
| 2738 |
+
"loss": 0.217,
|
| 2739 |
+
"step": 3900
|
| 2740 |
+
},
|
| 2741 |
+
{
|
| 2742 |
+
"epoch": 7.812187812187812,
|
| 2743 |
+
"grad_norm": 0.6331846117973328,
|
| 2744 |
+
"learning_rate": 2.254969398012663e-05,
|
| 2745 |
+
"loss": 0.2127,
|
| 2746 |
+
"step": 3910
|
| 2747 |
+
},
|
| 2748 |
+
{
|
| 2749 |
+
"epoch": 7.8321678321678325,
|
| 2750 |
+
"grad_norm": 0.6486308574676514,
|
| 2751 |
+
"learning_rate": 2.2153769843297667e-05,
|
| 2752 |
+
"loss": 0.2096,
|
| 2753 |
+
"step": 3920
|
| 2754 |
+
},
|
| 2755 |
+
{
|
| 2756 |
+
"epoch": 7.852147852147852,
|
| 2757 |
+
"grad_norm": 0.6658995151519775,
|
| 2758 |
+
"learning_rate": 2.1760918942341192e-05,
|
| 2759 |
+
"loss": 0.211,
|
| 2760 |
+
"step": 3930
|
| 2761 |
+
},
|
| 2762 |
+
{
|
| 2763 |
+
"epoch": 7.872127872127872,
|
| 2764 |
+
"grad_norm": 0.687493085861206,
|
| 2765 |
+
"learning_rate": 2.137115678633811e-05,
|
| 2766 |
+
"loss": 0.2163,
|
| 2767 |
+
"step": 3940
|
| 2768 |
+
},
|
| 2769 |
+
{
|
| 2770 |
+
"epoch": 7.892107892107892,
|
| 2771 |
+
"grad_norm": 0.6267641186714172,
|
| 2772 |
+
"learning_rate": 2.098449876243096e-05,
|
| 2773 |
+
"loss": 0.2142,
|
| 2774 |
+
"step": 3950
|
| 2775 |
+
},
|
| 2776 |
+
{
|
| 2777 |
+
"epoch": 7.912087912087912,
|
| 2778 |
+
"grad_norm": 0.6141098141670227,
|
| 2779 |
+
"learning_rate": 2.0600960135216462e-05,
|
| 2780 |
+
"loss": 0.2134,
|
| 2781 |
+
"step": 3960
|
| 2782 |
+
},
|
| 2783 |
+
{
|
| 2784 |
+
"epoch": 7.932067932067932,
|
| 2785 |
+
"grad_norm": 0.6436827182769775,
|
| 2786 |
+
"learning_rate": 2.0220556046142893e-05,
|
| 2787 |
+
"loss": 0.214,
|
| 2788 |
+
"step": 3970
|
| 2789 |
+
},
|
| 2790 |
+
{
|
| 2791 |
+
"epoch": 7.952047952047952,
|
| 2792 |
+
"grad_norm": 0.6543010473251343,
|
| 2793 |
+
"learning_rate": 1.9843301512912327e-05,
|
| 2794 |
+
"loss": 0.2126,
|
| 2795 |
+
"step": 3980
|
| 2796 |
+
},
|
| 2797 |
+
{
|
| 2798 |
+
"epoch": 7.972027972027972,
|
| 2799 |
+
"grad_norm": 0.6083731651306152,
|
| 2800 |
+
"learning_rate": 1.946921142888781e-05,
|
| 2801 |
+
"loss": 0.2135,
|
| 2802 |
+
"step": 3990
|
| 2803 |
+
},
|
| 2804 |
+
{
|
| 2805 |
+
"epoch": 7.992007992007992,
|
| 2806 |
+
"grad_norm": 0.6408571600914001,
|
| 2807 |
+
"learning_rate": 1.9098300562505266e-05,
|
| 2808 |
+
"loss": 0.2123,
|
| 2809 |
+
"step": 4000
|
| 2810 |
+
},
|
| 2811 |
+
{
|
| 2812 |
+
"epoch": 8.011988011988011,
|
| 2813 |
+
"grad_norm": 0.556982159614563,
|
| 2814 |
+
"learning_rate": 1.8730583556690605e-05,
|
| 2815 |
+
"loss": 0.2042,
|
| 2816 |
+
"step": 4010
|
| 2817 |
+
},
|
| 2818 |
+
{
|
| 2819 |
+
"epoch": 8.031968031968033,
|
| 2820 |
+
"grad_norm": 0.5726343393325806,
|
| 2821 |
+
"learning_rate": 1.8366074928281607e-05,
|
| 2822 |
+
"loss": 0.1941,
|
| 2823 |
+
"step": 4020
|
| 2824 |
+
},
|
| 2825 |
+
{
|
| 2826 |
+
"epoch": 8.051948051948052,
|
| 2827 |
+
"grad_norm": 0.5825814604759216,
|
| 2828 |
+
"learning_rate": 1.8004789067454764e-05,
|
| 2829 |
+
"loss": 0.1976,
|
| 2830 |
+
"step": 4030
|
| 2831 |
+
},
|
| 2832 |
+
{
|
| 2833 |
+
"epoch": 8.071928071928072,
|
| 2834 |
+
"grad_norm": 0.569325864315033,
|
| 2835 |
+
"learning_rate": 1.7646740237157256e-05,
|
| 2836 |
+
"loss": 0.196,
|
| 2837 |
+
"step": 4040
|
| 2838 |
+
},
|
| 2839 |
+
{
|
| 2840 |
+
"epoch": 8.091908091908092,
|
| 2841 |
+
"grad_norm": 0.5917354226112366,
|
| 2842 |
+
"learning_rate": 1.7291942572543807e-05,
|
| 2843 |
+
"loss": 0.195,
|
| 2844 |
+
"step": 4050
|
| 2845 |
+
},
|
| 2846 |
+
{
|
| 2847 |
+
"epoch": 8.111888111888112,
|
| 2848 |
+
"grad_norm": 0.5817933678627014,
|
| 2849 |
+
"learning_rate": 1.6940410080418723e-05,
|
| 2850 |
+
"loss": 0.1971,
|
| 2851 |
+
"step": 4060
|
| 2852 |
+
},
|
| 2853 |
+
{
|
| 2854 |
+
"epoch": 8.131868131868131,
|
| 2855 |
+
"grad_norm": 0.6475218534469604,
|
| 2856 |
+
"learning_rate": 1.6592156638682886e-05,
|
| 2857 |
+
"loss": 0.197,
|
| 2858 |
+
"step": 4070
|
| 2859 |
+
},
|
| 2860 |
+
{
|
| 2861 |
+
"epoch": 8.151848151848151,
|
| 2862 |
+
"grad_norm": 0.6248770356178284,
|
| 2863 |
+
"learning_rate": 1.6247195995785837e-05,
|
| 2864 |
+
"loss": 0.1971,
|
| 2865 |
+
"step": 4080
|
| 2866 |
+
},
|
| 2867 |
+
{
|
| 2868 |
+
"epoch": 8.171828171828173,
|
| 2869 |
+
"grad_norm": 0.5749895572662354,
|
| 2870 |
+
"learning_rate": 1.5905541770183096e-05,
|
| 2871 |
+
"loss": 0.1964,
|
| 2872 |
+
"step": 4090
|
| 2873 |
+
},
|
| 2874 |
+
{
|
| 2875 |
+
"epoch": 8.191808191808192,
|
| 2876 |
+
"grad_norm": 0.6148300766944885,
|
| 2877 |
+
"learning_rate": 1.5567207449798515e-05,
|
| 2878 |
+
"loss": 0.1966,
|
| 2879 |
+
"step": 4100
|
| 2880 |
+
},
|
| 2881 |
+
{
|
| 2882 |
+
"epoch": 8.211788211788212,
|
| 2883 |
+
"grad_norm": 0.6778724789619446,
|
| 2884 |
+
"learning_rate": 1.5232206391491699e-05,
|
| 2885 |
+
"loss": 0.1955,
|
| 2886 |
+
"step": 4110
|
| 2887 |
+
},
|
| 2888 |
+
{
|
| 2889 |
+
"epoch": 8.231768231768232,
|
| 2890 |
+
"grad_norm": 0.5883269906044006,
|
| 2891 |
+
"learning_rate": 1.4900551820530828e-05,
|
| 2892 |
+
"loss": 0.1919,
|
| 2893 |
+
"step": 4120
|
| 2894 |
+
},
|
| 2895 |
+
{
|
| 2896 |
+
"epoch": 8.251748251748252,
|
| 2897 |
+
"grad_norm": 0.567950963973999,
|
| 2898 |
+
"learning_rate": 1.4572256830070497e-05,
|
| 2899 |
+
"loss": 0.1966,
|
| 2900 |
+
"step": 4130
|
| 2901 |
+
},
|
| 2902 |
+
{
|
| 2903 |
+
"epoch": 8.271728271728271,
|
| 2904 |
+
"grad_norm": 0.5733300447463989,
|
| 2905 |
+
"learning_rate": 1.4247334380634792e-05,
|
| 2906 |
+
"loss": 0.1964,
|
| 2907 |
+
"step": 4140
|
| 2908 |
+
},
|
| 2909 |
+
{
|
| 2910 |
+
"epoch": 8.291708291708291,
|
| 2911 |
+
"grad_norm": 0.638990044593811,
|
| 2912 |
+
"learning_rate": 1.3925797299605647e-05,
|
| 2913 |
+
"loss": 0.1944,
|
| 2914 |
+
"step": 4150
|
| 2915 |
+
},
|
| 2916 |
+
{
|
| 2917 |
+
"epoch": 8.311688311688311,
|
| 2918 |
+
"grad_norm": 0.6272343397140503,
|
| 2919 |
+
"learning_rate": 1.3607658280716473e-05,
|
| 2920 |
+
"loss": 0.1951,
|
| 2921 |
+
"step": 4160
|
| 2922 |
+
},
|
| 2923 |
+
{
|
| 2924 |
+
"epoch": 8.331668331668332,
|
| 2925 |
+
"grad_norm": 0.5631300210952759,
|
| 2926 |
+
"learning_rate": 1.3292929883550998e-05,
|
| 2927 |
+
"loss": 0.1983,
|
| 2928 |
+
"step": 4170
|
| 2929 |
+
},
|
| 2930 |
+
{
|
| 2931 |
+
"epoch": 8.351648351648352,
|
| 2932 |
+
"grad_norm": 0.6056917309761047,
|
| 2933 |
+
"learning_rate": 1.2981624533047432e-05,
|
| 2934 |
+
"loss": 0.1976,
|
| 2935 |
+
"step": 4180
|
| 2936 |
+
},
|
| 2937 |
+
{
|
| 2938 |
+
"epoch": 8.371628371628372,
|
| 2939 |
+
"grad_norm": 0.6021771430969238,
|
| 2940 |
+
"learning_rate": 1.2673754519008008e-05,
|
| 2941 |
+
"loss": 0.1968,
|
| 2942 |
+
"step": 4190
|
| 2943 |
+
},
|
| 2944 |
+
{
|
| 2945 |
+
"epoch": 8.391608391608392,
|
| 2946 |
+
"grad_norm": 0.5835386514663696,
|
| 2947 |
+
"learning_rate": 1.2369331995613665e-05,
|
| 2948 |
+
"loss": 0.1977,
|
| 2949 |
+
"step": 4200
|
| 2950 |
+
},
|
| 2951 |
+
{
|
| 2952 |
+
"epoch": 8.411588411588411,
|
| 2953 |
+
"grad_norm": 0.5700567960739136,
|
| 2954 |
+
"learning_rate": 1.206836898094439e-05,
|
| 2955 |
+
"loss": 0.1992,
|
| 2956 |
+
"step": 4210
|
| 2957 |
+
},
|
| 2958 |
+
{
|
| 2959 |
+
"epoch": 8.431568431568431,
|
| 2960 |
+
"grad_norm": 0.6391722559928894,
|
| 2961 |
+
"learning_rate": 1.1770877356504683e-05,
|
| 2962 |
+
"loss": 0.1977,
|
| 2963 |
+
"step": 4220
|
| 2964 |
+
},
|
| 2965 |
+
{
|
| 2966 |
+
"epoch": 8.451548451548451,
|
| 2967 |
+
"grad_norm": 0.5633198022842407,
|
| 2968 |
+
"learning_rate": 1.1476868866754486e-05,
|
| 2969 |
+
"loss": 0.1975,
|
| 2970 |
+
"step": 4230
|
| 2971 |
+
},
|
| 2972 |
+
{
|
| 2973 |
+
"epoch": 8.471528471528472,
|
| 2974 |
+
"grad_norm": 0.6308007836341858,
|
| 2975 |
+
"learning_rate": 1.1186355118645554e-05,
|
| 2976 |
+
"loss": 0.2002,
|
| 2977 |
+
"step": 4240
|
| 2978 |
+
},
|
| 2979 |
+
{
|
| 2980 |
+
"epoch": 8.491508491508492,
|
| 2981 |
+
"grad_norm": 0.6147842407226562,
|
| 2982 |
+
"learning_rate": 1.0899347581163221e-05,
|
| 2983 |
+
"loss": 0.199,
|
| 2984 |
+
"step": 4250
|
| 2985 |
+
},
|
| 2986 |
+
{
|
| 2987 |
+
"epoch": 8.511488511488512,
|
| 2988 |
+
"grad_norm": 0.6099655628204346,
|
| 2989 |
+
"learning_rate": 1.0615857584873623e-05,
|
| 2990 |
+
"loss": 0.1971,
|
| 2991 |
+
"step": 4260
|
| 2992 |
+
},
|
| 2993 |
+
{
|
| 2994 |
+
"epoch": 8.531468531468532,
|
| 2995 |
+
"grad_norm": 0.6306450366973877,
|
| 2996 |
+
"learning_rate": 1.0335896321476413e-05,
|
| 2997 |
+
"loss": 0.1971,
|
| 2998 |
+
"step": 4270
|
| 2999 |
+
},
|
| 3000 |
+
{
|
| 3001 |
+
"epoch": 8.551448551448551,
|
| 3002 |
+
"grad_norm": 0.5740554928779602,
|
| 3003 |
+
"learning_rate": 1.0059474843362892e-05,
|
| 3004 |
+
"loss": 0.1964,
|
| 3005 |
+
"step": 4280
|
| 3006 |
+
},
|
| 3007 |
+
{
|
| 3008 |
+
"epoch": 8.571428571428571,
|
| 3009 |
+
"grad_norm": 0.566005289554596,
|
| 3010 |
+
"learning_rate": 9.786604063179728e-06,
|
| 3011 |
+
"loss": 0.197,
|
| 3012 |
+
"step": 4290
|
| 3013 |
+
},
|
| 3014 |
+
{
|
| 3015 |
+
"epoch": 8.591408591408591,
|
| 3016 |
+
"grad_norm": 0.6008467674255371,
|
| 3017 |
+
"learning_rate": 9.517294753398064e-06,
|
| 3018 |
+
"loss": 0.1969,
|
| 3019 |
+
"step": 4300
|
| 3020 |
+
},
|
| 3021 |
+
{
|
| 3022 |
+
"epoch": 8.61138861138861,
|
| 3023 |
+
"grad_norm": 0.5880402326583862,
|
| 3024 |
+
"learning_rate": 9.251557545888312e-06,
|
| 3025 |
+
"loss": 0.1944,
|
| 3026 |
+
"step": 4310
|
| 3027 |
+
},
|
| 3028 |
+
{
|
| 3029 |
+
"epoch": 8.631368631368632,
|
| 3030 |
+
"grad_norm": 0.6250616908073425,
|
| 3031 |
+
"learning_rate": 8.989402931500434e-06,
|
| 3032 |
+
"loss": 0.1978,
|
| 3033 |
+
"step": 4320
|
| 3034 |
+
},
|
| 3035 |
+
{
|
| 3036 |
+
"epoch": 8.651348651348652,
|
| 3037 |
+
"grad_norm": 0.554460883140564,
|
| 3038 |
+
"learning_rate": 8.730841259649725e-06,
|
| 3039 |
+
"loss": 0.1998,
|
| 3040 |
+
"step": 4330
|
| 3041 |
+
},
|
| 3042 |
+
{
|
| 3043 |
+
"epoch": 8.671328671328672,
|
| 3044 |
+
"grad_norm": 0.5680242776870728,
|
| 3045 |
+
"learning_rate": 8.475882737908248e-06,
|
| 3046 |
+
"loss": 0.2,
|
| 3047 |
+
"step": 4340
|
| 3048 |
+
},
|
| 3049 |
+
{
|
| 3050 |
+
"epoch": 8.691308691308691,
|
| 3051 |
+
"grad_norm": 0.5889159440994263,
|
| 3052 |
+
"learning_rate": 8.224537431601886e-06,
|
| 3053 |
+
"loss": 0.1985,
|
| 3054 |
+
"step": 4350
|
| 3055 |
+
},
|
| 3056 |
+
{
|
| 3057 |
+
"epoch": 8.711288711288711,
|
| 3058 |
+
"grad_norm": 0.6051207780838013,
|
| 3059 |
+
"learning_rate": 7.976815263412963e-06,
|
| 3060 |
+
"loss": 0.1944,
|
| 3061 |
+
"step": 4360
|
| 3062 |
+
},
|
| 3063 |
+
{
|
| 3064 |
+
"epoch": 8.731268731268731,
|
| 3065 |
+
"grad_norm": 0.6148102283477783,
|
| 3066 |
+
"learning_rate": 7.73272601298851e-06,
|
| 3067 |
+
"loss": 0.1952,
|
| 3068 |
+
"step": 4370
|
| 3069 |
+
},
|
| 3070 |
+
{
|
| 3071 |
+
"epoch": 8.75124875124875,
|
| 3072 |
+
"grad_norm": 0.6123753786087036,
|
| 3073 |
+
"learning_rate": 7.492279316554207e-06,
|
| 3074 |
+
"loss": 0.1955,
|
| 3075 |
+
"step": 4380
|
| 3076 |
+
},
|
| 3077 |
+
{
|
| 3078 |
+
"epoch": 8.77122877122877,
|
| 3079 |
+
"grad_norm": 0.5911871790885925,
|
| 3080 |
+
"learning_rate": 7.255484666533874e-06,
|
| 3081 |
+
"loss": 0.1987,
|
| 3082 |
+
"step": 4390
|
| 3083 |
+
},
|
| 3084 |
+
{
|
| 3085 |
+
"epoch": 8.791208791208792,
|
| 3086 |
+
"grad_norm": 0.5861064195632935,
|
| 3087 |
+
"learning_rate": 7.022351411174866e-06,
|
| 3088 |
+
"loss": 0.1972,
|
| 3089 |
+
"step": 4400
|
| 3090 |
+
},
|
| 3091 |
+
{
|
| 3092 |
+
"epoch": 8.811188811188812,
|
| 3093 |
+
"grad_norm": 0.6565813422203064,
|
| 3094 |
+
"learning_rate": 6.7928887541789055e-06,
|
| 3095 |
+
"loss": 0.1966,
|
| 3096 |
+
"step": 4410
|
| 3097 |
+
},
|
| 3098 |
+
{
|
| 3099 |
+
"epoch": 8.831168831168831,
|
| 3100 |
+
"grad_norm": 0.6338573694229126,
|
| 3101 |
+
"learning_rate": 6.5671057543387985e-06,
|
| 3102 |
+
"loss": 0.1987,
|
| 3103 |
+
"step": 4420
|
| 3104 |
+
},
|
| 3105 |
+
{
|
| 3106 |
+
"epoch": 8.851148851148851,
|
| 3107 |
+
"grad_norm": 0.5672295093536377,
|
| 3108 |
+
"learning_rate": 6.345011325180772e-06,
|
| 3109 |
+
"loss": 0.198,
|
| 3110 |
+
"step": 4430
|
| 3111 |
+
},
|
| 3112 |
+
{
|
| 3113 |
+
"epoch": 8.871128871128871,
|
| 3114 |
+
"grad_norm": 0.6036155223846436,
|
| 3115 |
+
"learning_rate": 6.126614234612593e-06,
|
| 3116 |
+
"loss": 0.199,
|
| 3117 |
+
"step": 4440
|
| 3118 |
+
},
|
| 3119 |
+
{
|
| 3120 |
+
"epoch": 8.89110889110889,
|
| 3121 |
+
"grad_norm": 0.5816395878791809,
|
| 3122 |
+
"learning_rate": 5.911923104577455e-06,
|
| 3123 |
+
"loss": 0.1985,
|
| 3124 |
+
"step": 4450
|
| 3125 |
+
},
|
| 3126 |
+
{
|
| 3127 |
+
"epoch": 8.91108891108891,
|
| 3128 |
+
"grad_norm": 0.5562584400177002,
|
| 3129 |
+
"learning_rate": 5.700946410713548e-06,
|
| 3130 |
+
"loss": 0.1964,
|
| 3131 |
+
"step": 4460
|
| 3132 |
+
},
|
| 3133 |
+
{
|
| 3134 |
+
"epoch": 8.931068931068932,
|
| 3135 |
+
"grad_norm": 0.6179762482643127,
|
| 3136 |
+
"learning_rate": 5.49369248201953e-06,
|
| 3137 |
+
"loss": 0.1948,
|
| 3138 |
+
"step": 4470
|
| 3139 |
+
},
|
| 3140 |
+
{
|
| 3141 |
+
"epoch": 8.951048951048952,
|
| 3142 |
+
"grad_norm": 0.5566456317901611,
|
| 3143 |
+
"learning_rate": 5.290169500525577e-06,
|
| 3144 |
+
"loss": 0.1958,
|
| 3145 |
+
"step": 4480
|
| 3146 |
+
},
|
| 3147 |
+
{
|
| 3148 |
+
"epoch": 8.971028971028971,
|
| 3149 |
+
"grad_norm": 0.6196462512016296,
|
| 3150 |
+
"learning_rate": 5.0903855009705514e-06,
|
| 3151 |
+
"loss": 0.1978,
|
| 3152 |
+
"step": 4490
|
| 3153 |
+
},
|
| 3154 |
+
{
|
| 3155 |
+
"epoch": 8.991008991008991,
|
| 3156 |
+
"grad_norm": 0.5933112502098083,
|
| 3157 |
+
"learning_rate": 4.8943483704846475e-06,
|
| 3158 |
+
"loss": 0.1962,
|
| 3159 |
+
"step": 4500
|
| 3160 |
+
},
|
| 3161 |
+
{
|
| 3162 |
+
"epoch": 9.010989010989011,
|
| 3163 |
+
"grad_norm": 0.5680419206619263,
|
| 3164 |
+
"learning_rate": 4.702065848278126e-06,
|
| 3165 |
+
"loss": 0.1948,
|
| 3166 |
+
"step": 4510
|
| 3167 |
+
},
|
| 3168 |
+
{
|
| 3169 |
+
"epoch": 9.03096903096903,
|
| 3170 |
+
"grad_norm": 0.5447672605514526,
|
| 3171 |
+
"learning_rate": 4.513545525335705e-06,
|
| 3172 |
+
"loss": 0.1894,
|
| 3173 |
+
"step": 4520
|
| 3174 |
+
},
|
| 3175 |
+
{
|
| 3176 |
+
"epoch": 9.05094905094905,
|
| 3177 |
+
"grad_norm": 0.5605758428573608,
|
| 3178 |
+
"learning_rate": 4.328794844116946e-06,
|
| 3179 |
+
"loss": 0.1903,
|
| 3180 |
+
"step": 4530
|
| 3181 |
+
},
|
| 3182 |
+
{
|
| 3183 |
+
"epoch": 9.07092907092907,
|
| 3184 |
+
"grad_norm": 0.5727641582489014,
|
| 3185 |
+
"learning_rate": 4.147821098262405e-06,
|
| 3186 |
+
"loss": 0.1899,
|
| 3187 |
+
"step": 4540
|
| 3188 |
+
},
|
| 3189 |
+
{
|
| 3190 |
+
"epoch": 9.090909090909092,
|
| 3191 |
+
"grad_norm": 0.5076532363891602,
|
| 3192 |
+
"learning_rate": 3.970631432305694e-06,
|
| 3193 |
+
"loss": 0.1872,
|
| 3194 |
+
"step": 4550
|
| 3195 |
+
},
|
| 3196 |
+
{
|
| 3197 |
+
"epoch": 9.110889110889111,
|
| 3198 |
+
"grad_norm": 0.5827686190605164,
|
| 3199 |
+
"learning_rate": 3.797232841391407e-06,
|
| 3200 |
+
"loss": 0.1871,
|
| 3201 |
+
"step": 4560
|
| 3202 |
+
},
|
| 3203 |
+
{
|
| 3204 |
+
"epoch": 9.130869130869131,
|
| 3205 |
+
"grad_norm": 0.5457426905632019,
|
| 3206 |
+
"learning_rate": 3.627632170999029e-06,
|
| 3207 |
+
"loss": 0.1903,
|
| 3208 |
+
"step": 4570
|
| 3209 |
+
},
|
| 3210 |
+
{
|
| 3211 |
+
"epoch": 9.150849150849151,
|
| 3212 |
+
"grad_norm": 0.5931391716003418,
|
| 3213 |
+
"learning_rate": 3.461836116672612e-06,
|
| 3214 |
+
"loss": 0.1935,
|
| 3215 |
+
"step": 4580
|
| 3216 |
+
},
|
| 3217 |
+
{
|
| 3218 |
+
"epoch": 9.17082917082917,
|
| 3219 |
+
"grad_norm": 0.5335982441902161,
|
| 3220 |
+
"learning_rate": 3.2998512237565005e-06,
|
| 3221 |
+
"loss": 0.188,
|
| 3222 |
+
"step": 4590
|
| 3223 |
+
},
|
| 3224 |
+
{
|
| 3225 |
+
"epoch": 9.19080919080919,
|
| 3226 |
+
"grad_norm": 0.5809586048126221,
|
| 3227 |
+
"learning_rate": 3.1416838871368924e-06,
|
| 3228 |
+
"loss": 0.1882,
|
| 3229 |
+
"step": 4600
|
| 3230 |
+
},
|
| 3231 |
+
{
|
| 3232 |
+
"epoch": 9.21078921078921,
|
| 3233 |
+
"grad_norm": 0.5997488498687744,
|
| 3234 |
+
"learning_rate": 2.9873403509894203e-06,
|
| 3235 |
+
"loss": 0.189,
|
| 3236 |
+
"step": 4610
|
| 3237 |
+
},
|
| 3238 |
+
{
|
| 3239 |
+
"epoch": 9.23076923076923,
|
| 3240 |
+
"grad_norm": 0.5423487424850464,
|
| 3241 |
+
"learning_rate": 2.836826708532603e-06,
|
| 3242 |
+
"loss": 0.1916,
|
| 3243 |
+
"step": 4620
|
| 3244 |
+
},
|
| 3245 |
+
{
|
| 3246 |
+
"epoch": 9.250749250749251,
|
| 3247 |
+
"grad_norm": 0.5920736193656921,
|
| 3248 |
+
"learning_rate": 2.690148901787337e-06,
|
| 3249 |
+
"loss": 0.1914,
|
| 3250 |
+
"step": 4630
|
| 3251 |
+
},
|
| 3252 |
+
{
|
| 3253 |
+
"epoch": 9.270729270729271,
|
| 3254 |
+
"grad_norm": 0.5774621367454529,
|
| 3255 |
+
"learning_rate": 2.5473127213422763e-06,
|
| 3256 |
+
"loss": 0.1901,
|
| 3257 |
+
"step": 4640
|
| 3258 |
+
},
|
| 3259 |
+
{
|
| 3260 |
+
"epoch": 9.290709290709291,
|
| 3261 |
+
"grad_norm": 0.6183256506919861,
|
| 3262 |
+
"learning_rate": 2.4083238061252567e-06,
|
| 3263 |
+
"loss": 0.1918,
|
| 3264 |
+
"step": 4650
|
| 3265 |
+
},
|
| 3266 |
+
{
|
| 3267 |
+
"epoch": 9.31068931068931,
|
| 3268 |
+
"grad_norm": 0.5502414107322693,
|
| 3269 |
+
"learning_rate": 2.273187643180652e-06,
|
| 3270 |
+
"loss": 0.1888,
|
| 3271 |
+
"step": 4660
|
| 3272 |
+
},
|
| 3273 |
+
{
|
| 3274 |
+
"epoch": 9.33066933066933,
|
| 3275 |
+
"grad_norm": 0.5888564586639404,
|
| 3276 |
+
"learning_rate": 2.141909567452793e-06,
|
| 3277 |
+
"loss": 0.189,
|
| 3278 |
+
"step": 4670
|
| 3279 |
+
},
|
| 3280 |
+
{
|
| 3281 |
+
"epoch": 9.35064935064935,
|
| 3282 |
+
"grad_norm": 0.582281231880188,
|
| 3283 |
+
"learning_rate": 2.014494761575314e-06,
|
| 3284 |
+
"loss": 0.188,
|
| 3285 |
+
"step": 4680
|
| 3286 |
+
},
|
| 3287 |
+
{
|
| 3288 |
+
"epoch": 9.37062937062937,
|
| 3289 |
+
"grad_norm": 0.549766480922699,
|
| 3290 |
+
"learning_rate": 1.8909482556666024e-06,
|
| 3291 |
+
"loss": 0.1911,
|
| 3292 |
+
"step": 4690
|
| 3293 |
+
},
|
| 3294 |
+
{
|
| 3295 |
+
"epoch": 9.390609390609391,
|
| 3296 |
+
"grad_norm": 0.6442523002624512,
|
| 3297 |
+
"learning_rate": 1.771274927131139e-06,
|
| 3298 |
+
"loss": 0.1913,
|
| 3299 |
+
"step": 4700
|
| 3300 |
+
},
|
| 3301 |
+
{
|
| 3302 |
+
"epoch": 9.410589410589411,
|
| 3303 |
+
"grad_norm": 0.5612021684646606,
|
| 3304 |
+
"learning_rate": 1.6554795004670388e-06,
|
| 3305 |
+
"loss": 0.1926,
|
| 3306 |
+
"step": 4710
|
| 3307 |
+
},
|
| 3308 |
+
{
|
| 3309 |
+
"epoch": 9.430569430569431,
|
| 3310 |
+
"grad_norm": 0.6060473918914795,
|
| 3311 |
+
"learning_rate": 1.543566547079467e-06,
|
| 3312 |
+
"loss": 0.19,
|
| 3313 |
+
"step": 4720
|
| 3314 |
+
},
|
| 3315 |
+
{
|
| 3316 |
+
"epoch": 9.45054945054945,
|
| 3317 |
+
"grad_norm": 0.5958064794540405,
|
| 3318 |
+
"learning_rate": 1.4355404851001952e-06,
|
| 3319 |
+
"loss": 0.1885,
|
| 3320 |
+
"step": 4730
|
| 3321 |
+
},
|
| 3322 |
+
{
|
| 3323 |
+
"epoch": 9.47052947052947,
|
| 3324 |
+
"grad_norm": 0.536431610584259,
|
| 3325 |
+
"learning_rate": 1.3314055792131964e-06,
|
| 3326 |
+
"loss": 0.1891,
|
| 3327 |
+
"step": 4740
|
| 3328 |
+
},
|
| 3329 |
+
{
|
| 3330 |
+
"epoch": 9.49050949050949,
|
| 3331 |
+
"grad_norm": 0.5971366763114929,
|
| 3332 |
+
"learning_rate": 1.231165940486234e-06,
|
| 3333 |
+
"loss": 0.1889,
|
| 3334 |
+
"step": 4750
|
| 3335 |
+
},
|
| 3336 |
+
{
|
| 3337 |
+
"epoch": 9.51048951048951,
|
| 3338 |
+
"grad_norm": 0.5461220145225525,
|
| 3339 |
+
"learning_rate": 1.134825526208605e-06,
|
| 3340 |
+
"loss": 0.1874,
|
| 3341 |
+
"step": 4760
|
| 3342 |
+
},
|
| 3343 |
+
{
|
| 3344 |
+
"epoch": 9.53046953046953,
|
| 3345 |
+
"grad_norm": 0.570928156375885,
|
| 3346 |
+
"learning_rate": 1.0423881397349068e-06,
|
| 3347 |
+
"loss": 0.1884,
|
| 3348 |
+
"step": 4770
|
| 3349 |
+
},
|
| 3350 |
+
{
|
| 3351 |
+
"epoch": 9.550449550449551,
|
| 3352 |
+
"grad_norm": 0.5855159759521484,
|
| 3353 |
+
"learning_rate": 9.538574303348813e-07,
|
| 3354 |
+
"loss": 0.1895,
|
| 3355 |
+
"step": 4780
|
| 3356 |
+
},
|
| 3357 |
+
{
|
| 3358 |
+
"epoch": 9.570429570429571,
|
| 3359 |
+
"grad_norm": 0.5505802631378174,
|
| 3360 |
+
"learning_rate": 8.692368930493521e-07,
|
| 3361 |
+
"loss": 0.1904,
|
| 3362 |
+
"step": 4790
|
| 3363 |
+
},
|
| 3364 |
+
{
|
| 3365 |
+
"epoch": 9.59040959040959,
|
| 3366 |
+
"grad_norm": 0.5663396716117859,
|
| 3367 |
+
"learning_rate": 7.885298685522235e-07,
|
| 3368 |
+
"loss": 0.1909,
|
| 3369 |
+
"step": 4800
|
| 3370 |
+
},
|
| 3371 |
+
{
|
| 3372 |
+
"epoch": 9.61038961038961,
|
| 3373 |
+
"grad_norm": 0.6069871783256531,
|
| 3374 |
+
"learning_rate": 7.117395430186414e-07,
|
| 3375 |
+
"loss": 0.1895,
|
| 3376 |
+
"step": 4810
|
| 3377 |
+
},
|
| 3378 |
+
{
|
| 3379 |
+
"epoch": 9.63036963036963,
|
| 3380 |
+
"grad_norm": 0.5576395988464355,
|
| 3381 |
+
"learning_rate": 6.388689479991605e-07,
|
| 3382 |
+
"loss": 0.1906,
|
| 3383 |
+
"step": 4820
|
| 3384 |
+
},
|
| 3385 |
+
{
|
| 3386 |
+
"epoch": 9.65034965034965,
|
| 3387 |
+
"grad_norm": 0.5069971084594727,
|
| 3388 |
+
"learning_rate": 5.699209603001076e-07,
|
| 3389 |
+
"loss": 0.1889,
|
| 3390 |
+
"step": 4830
|
| 3391 |
+
},
|
| 3392 |
+
{
|
| 3393 |
+
"epoch": 9.67032967032967,
|
| 3394 |
+
"grad_norm": 0.5770872235298157,
|
| 3395 |
+
"learning_rate": 5.048983018699827e-07,
|
| 3396 |
+
"loss": 0.1907,
|
| 3397 |
+
"step": 4840
|
| 3398 |
+
},
|
| 3399 |
+
{
|
| 3400 |
+
"epoch": 9.69030969030969,
|
| 3401 |
+
"grad_norm": 0.6914857029914856,
|
| 3402 |
+
"learning_rate": 4.438035396920004e-07,
|
| 3403 |
+
"loss": 0.1939,
|
| 3404 |
+
"step": 4850
|
| 3405 |
+
},
|
| 3406 |
+
{
|
| 3407 |
+
"epoch": 9.710289710289711,
|
| 3408 |
+
"grad_norm": 0.5999007821083069,
|
| 3409 |
+
"learning_rate": 3.866390856827495e-07,
|
| 3410 |
+
"loss": 0.1924,
|
| 3411 |
+
"step": 4860
|
| 3412 |
+
},
|
| 3413 |
+
{
|
| 3414 |
+
"epoch": 9.73026973026973,
|
| 3415 |
+
"grad_norm": 0.569180965423584,
|
| 3416 |
+
"learning_rate": 3.3340719659701313e-07,
|
| 3417 |
+
"loss": 0.1887,
|
| 3418 |
+
"step": 4870
|
| 3419 |
+
},
|
| 3420 |
+
{
|
| 3421 |
+
"epoch": 9.75024975024975,
|
| 3422 |
+
"grad_norm": 0.5442143082618713,
|
| 3423 |
+
"learning_rate": 2.841099739386066e-07,
|
| 3424 |
+
"loss": 0.1897,
|
| 3425 |
+
"step": 4880
|
| 3426 |
+
},
|
| 3427 |
+
{
|
| 3428 |
+
"epoch": 9.77022977022977,
|
| 3429 |
+
"grad_norm": 0.5622804164886475,
|
| 3430 |
+
"learning_rate": 2.387493638774774e-07,
|
| 3431 |
+
"loss": 0.1898,
|
| 3432 |
+
"step": 4890
|
| 3433 |
+
},
|
| 3434 |
+
{
|
| 3435 |
+
"epoch": 9.79020979020979,
|
| 3436 |
+
"grad_norm": 0.6558981537818909,
|
| 3437 |
+
"learning_rate": 1.973271571728441e-07,
|
| 3438 |
+
"loss": 0.1917,
|
| 3439 |
+
"step": 4900
|
| 3440 |
+
},
|
| 3441 |
+
{
|
| 3442 |
+
"epoch": 9.81018981018981,
|
| 3443 |
+
"grad_norm": 0.5756235122680664,
|
| 3444 |
+
"learning_rate": 1.598449891024978e-07,
|
| 3445 |
+
"loss": 0.192,
|
| 3446 |
+
"step": 4910
|
| 3447 |
+
},
|
| 3448 |
+
{
|
| 3449 |
+
"epoch": 9.83016983016983,
|
| 3450 |
+
"grad_norm": 0.5818027257919312,
|
| 3451 |
+
"learning_rate": 1.2630433939825327e-07,
|
| 3452 |
+
"loss": 0.1899,
|
| 3453 |
+
"step": 4920
|
| 3454 |
+
},
|
| 3455 |
+
{
|
| 3456 |
+
"epoch": 9.850149850149851,
|
| 3457 |
+
"grad_norm": 0.5986452698707581,
|
| 3458 |
+
"learning_rate": 9.670653218752934e-08,
|
| 3459 |
+
"loss": 0.1918,
|
| 3460 |
+
"step": 4930
|
| 3461 |
+
},
|
| 3462 |
+
{
|
| 3463 |
+
"epoch": 9.87012987012987,
|
| 3464 |
+
"grad_norm": 0.5438185334205627,
|
| 3465 |
+
"learning_rate": 7.105273594107953e-08,
|
| 3466 |
+
"loss": 0.1905,
|
| 3467 |
+
"step": 4940
|
| 3468 |
+
},
|
| 3469 |
+
{
|
| 3470 |
+
"epoch": 9.89010989010989,
|
| 3471 |
+
"grad_norm": 0.5430960059165955,
|
| 3472 |
+
"learning_rate": 4.934396342684e-08,
|
| 3473 |
+
"loss": 0.1913,
|
| 3474 |
+
"step": 4950
|
| 3475 |
+
},
|
| 3476 |
+
{
|
| 3477 |
+
"epoch": 9.91008991008991,
|
| 3478 |
+
"grad_norm": 0.5492510199546814,
|
| 3479 |
+
"learning_rate": 3.1581071670006015e-08,
|
| 3480 |
+
"loss": 0.1904,
|
| 3481 |
+
"step": 4960
|
| 3482 |
+
},
|
| 3483 |
+
{
|
| 3484 |
+
"epoch": 9.93006993006993,
|
| 3485 |
+
"grad_norm": 0.5370259881019592,
|
| 3486 |
+
"learning_rate": 1.7764761919103477e-08,
|
| 3487 |
+
"loss": 0.1901,
|
| 3488 |
+
"step": 4970
|
| 3489 |
+
},
|
| 3490 |
+
{
|
| 3491 |
+
"epoch": 9.95004995004995,
|
| 3492 |
+
"grad_norm": 0.5463282465934753,
|
| 3493 |
+
"learning_rate": 7.895579618388827e-09,
|
| 3494 |
+
"loss": 0.191,
|
| 3495 |
+
"step": 4980
|
| 3496 |
+
},
|
| 3497 |
+
{
|
| 3498 |
+
"epoch": 9.97002997002997,
|
| 3499 |
+
"grad_norm": 0.5733128190040588,
|
| 3500 |
+
"learning_rate": 1.973914386288467e-09,
|
| 3501 |
+
"loss": 0.1885,
|
| 3502 |
+
"step": 4990
|
| 3503 |
+
},
|
| 3504 |
+
{
|
| 3505 |
+
"epoch": 9.99000999000999,
|
| 3506 |
+
"grad_norm": 0.5241893529891968,
|
| 3507 |
+
"learning_rate": 0.0,
|
| 3508 |
+
"loss": 0.1916,
|
| 3509 |
+
"step": 5000
|
| 3510 |
+
},
|
| 3511 |
+
{
|
| 3512 |
+
"epoch": 9.99000999000999,
|
| 3513 |
+
"step": 5000,
|
| 3514 |
+
"total_flos": 7.37720834306605e+17,
|
| 3515 |
+
"train_loss": 0.4096551623106003,
|
| 3516 |
+
"train_runtime": 80947.5605,
|
| 3517 |
+
"train_samples_per_second": 0.742,
|
| 3518 |
+
"train_steps_per_second": 0.062
|
| 3519 |
+
}
|
| 3520 |
+
],
|
| 3521 |
+
"logging_steps": 10,
|
| 3522 |
+
"max_steps": 5000,
|
| 3523 |
+
"num_input_tokens_seen": 0,
|
| 3524 |
+
"num_train_epochs": 10,
|
| 3525 |
+
"save_steps": 500,
|
| 3526 |
+
"total_flos": 7.37720834306605e+17,
|
| 3527 |
+
"train_batch_size": 3,
|
| 3528 |
+
"trial_name": null,
|
| 3529 |
+
"trial_params": null
|
| 3530 |
+
}
|